Datasets:

pratyushmaini

/

llm_dataset_inference

like 1

s plotting the spatial power distribution in decibel scale, i.e., $20 \log_{10} (p(\r))$. Two different scenarios are considered, namely, *nominal* and *perturbed*, to evaluate the proposed robust power distribution scheme for the ultrasonic array. In the first scenario, nominal, we assume that the array steering vectors are precisely modeled, i.e., $\atil(\r) = \mathbf{0}$. In Fig. \[fig:nominal\], the beampattern generated by the array is plotted for the nominal scenario. This figure represents how power is spatially distributed over the organ at risk in an idealistic situation. Here, the covariance matrix of the waveforms is optimized under the assumption that the steering vectors are accurately modeled by, and the performance is evaluated using exactly the same steering vectors without any perturbations. The power is noticeably concentrated in the tumor region and importantly the power in the healthy tissue is several decibels lower. ![Power distribution (transmit beampattern in dB) for the nominal scenario, i.e., using $\R_{nr}$ and $\atil(\r) \equiv 0$.[]{data-label="fig:nominal"}](nominal_05_07lines "fig:"){width="\columnwidth" height="5.5cm"}\ In the second scenario, perturbed, the idealistic assumptions are relaxed and model uncertainties and imperfections are taken into account. The second scenario represents the case where the true steering vectors are perturbed versions of the nominal steering vectors $\ahat(\r)$, i.e., the true steering vector equals $\ahat(\r) + \atil(\r)$ where $\atil(\r) \in \mathcal{E}_{\r}$. The perturbation vectors $\atil(\r)$ are unknown but deterministically bounded. In the following we illustrate the worst-case performance, i.e., using the worst steering vectors to calculate the power distribution at each point. We start by illustrating the beampattern for the non-robust covariance matrix $\R_{nr}$ under the worst steering vectors. Fig. \[fig:NonRobustWorst\] shows how steering vector errors can degrade the array performance. Notice that in the worst-case, there is a substantial power leakage that occurs in the healthy tissue surrounding the tumor compared to Fig. \[fig:nominal\]. While, in Fig. \[fig:RobustWorst\], the robust optimal covariance matrix $\

s plotting the spatial power distribution in decibel scale, i.e., $ 20 \log_{10 } (p(\r))$. Two different scenario are view, namely, * nominal * and * perturb *, to measure the proposed robust power distribution dodge for the ultrasonic array. In the first scenario, nominal, we bear that the array steering vectors are precisely model, i.e., $ \atil(\r) = \mathbf{0}$. In Fig.   \[fig: nominal\ ], the beampattern generated by the array is plotted for the nominative scenario. This figure represents how power is spatially distributed over the organ at risk in an idealistic site. Here, the covariance matrix of the waveforms is optimized under the assumption that the guidance vectors are accurately modeled by, and the operation is evaluated use exactly the like steering vectors without any perturbations. The power is perceptibly concentrated in the tumor region and importantly the power in the healthy tissue is several decibels lower. ! [ Power distribution (transmit beampattern in dB) for the nominal scenario, i.e., using $ \R_{nr}$ and $ \atil(\r) \equiv 0$.[]{data - label="fig: nominal"}](nominal_05_07lines " fig:"){width="\columnwidth " height="5.5cm"}\ In the second scenario, perturbed, the idealistic assumptions are relaxed and model uncertainties and imperfections are taken into history. The second scenario represents the case where the on-key guidance vector are perturbed versions of the nominative steering vector $ \ahat(\r)$, i.e., the true steering vector equal $ \ahat(\r) + \atil(\r)$ where $ \atil(\r) \in \mathcal{E}_{\r}$. The perturbation vectors $ \atil(\r)$ are unknown but deterministically bounded. In the following we illustrate the bad - case performance, i.e., using the worst guidance vectors to account the power distribution at each point. We start by illustrating the beampattern for the non - robust covariance matrix $ \R_{nr}$ under the bad steering vectors. Fig.   \[fig: NonRobustWorst\ ] shows how guide vector errors can degrade the array performance. Notice that in the bad - case, there be a substantial ability leakage that happen in the healthy tissue surround the tumor compared to Fig.   \[fig: nominal\ ]. While, in Fig.   \[fig: RobustWorst\ ], the full-bodied optimal covariance matrix $ \

s ppotting the spatial powev distribution iu decibxl scals, i.e., $20 \loe_{10} (p(\r))$. Two different scenarios aee cobsidered, namely, *nominau* and *pernurbed*, to evaoyate the pckposed vjbusf powzr distribution xcheme for the ultrasoniw xrxay. In the first scenario, nominal, we assume tjat the array fteegigg vsbtirs are precisely modeled, i.e., $\ztil(\r) = \kathbf{0}$. In Fig. \[gig:nominal\], the beampattern genfrated by the arraj is plottee fow the nominal scenario. Uhns figure rgpresents how power is spatially distxibuted ovet fhf organ at rmsk in an idealistlb situadion. Hete, the covariakce metriz of the waveforms is optimized under tre assumpdikn that the steerunt vecjors dre xxcufattly mkdeled by, and the psrformance us evaluated using txasnky the same ateerigg vectors without any perturbations. The kower is noticeably concentrqted in the tumor reglon and iiportantly the power in the healthy tissue is sevaral vezibtlw lowde. ![Plwer distribution (transmit beampattern in dB) skr tme nominal scenavio, i.e., using $\R_{nr}$ amd $\ayyl(\r) \equiv 0$.[]{datx-label="yjg:hominal"}](nominal_05_07lined "fig:"){wiqth="\cooumnwidth" heibht="5.5cm"}\ In the second scenario, perturbed, tke udealistic assumptnons are relcxed amd mocel uncertainties and ioperrections arf taken ihgo account. The sdcokd vcenario represents the cafe where vhe txue steefing vectows are perhurbed versions of the nlminap vteering vfctors $\ahat(\r)$, i.e., the true steeriij vector equaks $\ahdt(\r) + \atij(\r)$ whcre $\atil(\r) \in \maehcal{E}_{\r}$. The petturbatiou vecturs $\atil(\r)$ zre unkiown but detqrministicallf bounded. In vhe folloring we ullustrxge the worst-caxe performance, i.e., uwing the worst stecring bectors to calcbuqte the power dostfibttpon ae each point. Fe sgarg by iulustraunng ghe neampattern for the ton-rkbust covariance mstvix $\R_{nr}$ ubder the worst steerimg vectors. Fig. \[fig:NlnRobnstWorrt\] shpws how steering vector errors cah degrade thc array perfowmange. Njtice that in the worst-case, there is a substantiao power leakage thqt occurs in the hecluhy tissue sucroundyng the tgmor compared to Fig. \[dig:nominal\]. While, ln Fig. \[fig:RobustWorst\], tge robgst ootimal covariance matrix $\

s plotting the spatial power distribution in i.e., \log_{10} (p(\r))$. different scenarios are to the proposed robust distribution scheme for ultrasonic array. In the first scenario, we assume that the array steering vectors are precisely modeled, i.e., $\atil(\r) = In Fig. \[fig:nominal\], the beampattern generated by the array is plotted for the scenario. figure how is spatially distributed over the organ at risk in an idealistic situation. Here, the covariance matrix the waveforms is optimized under the assumption that steering vectors are accurately by, and the performance is using the same vectors any The power is concentrated in the tumor region and importantly the power in the healthy tissue is several decibels lower. distribution (transmit dB) for nominal i.e., $\R_{nr}$ and $\atil(\r) "fig:"){width="\columnwidth" height="5.5cm"}\ In the second scenario, assumptions are relaxed and model uncertainties and imperfections taken into The second scenario represents the case the true steering vectors are perturbed versions of nominal steering vectors $\ahat(\r)$, i.e., the true steering vector equals $\ahat(\r) + \atil(\r)$ where $\atil(\r) The perturbation vectors $\atil(\r)$ unknown but deterministically In following illustrate worst-case performance, using the worst steering vectors to calculate the power distribution at point. We start by illustrating the beampattern for the non-robust $\R_{nr}$ the worst steering Fig. \[fig:NonRobustWorst\] shows how vector can degrade the array that the a power that occurs in the tissue surrounding the tumor compared Fig. \[fig:nominal\]. While, in covariance matrix $\

s plotting the spatial power dIstributioN in deCibEl sCaLe, i.e., $20 \Log_{10} (p(\R))$. Two different sCEnarIos are considered, namely, *NominAl* ANd *peRTuRbed*, tO evaluaTE tHE ProPoSeD roBuST pOwer dIstRibutioN scheme for The UlTrasonic arraY. in The first scEnaRio, nominal, we AssUme thaT tHe aRRay stEerIng veCtors aRE preciSely modelEd, I.E., $\atil(\r) = \MAthbf{0}$. In fIG. \[fIg:noMinal\], the beampatteRN gENerated by the arRay is pLoTTeD FOr tHe nOminal scenArIo. ThiS Figure rEPrESENts HOw power is spatIally distriBUteD over tHe OrgAN at risK in an IdEAliStic situatiOn. HeRe, the covaRiance MAtrix of THe wavefOrms is OptImiZed uNDeR tHe aSsUMptIOn ThaT The Steering VeCtOrs arE accURATEly mOdeLed bY, and tHe performance Is eValuATed Using ExactLy thE sAme stEering VectoRs Without any pertuRbatIons. The poWer Is NotIcEably COncentRatEd iN the tumOr regioN And ImPORTaNtly the power in the hEaLTHy Tissue is SeveraL DeCiBEls lower. ![poWer DistRIButioN (traNSmIt beampaTtern iN DB) FoR the nomInAl scenArIo, i.E., usIng $\R_{nR}$ And $\aTil(\r) \eqUiv 0$.[]{data-lAbel="fIG:nominal"}](nominaL_05_07Lines "fig:"){width="\COlUMNwIDth" hEigHt="5.5cm"}\ In the seCond SCenaRio, pERtUrbED, the iDealiStIC aSSumptions are relaxed AnD model UncerTainties and imPerfectionS ARE taken inTo acCOuNT. The second scenArio rEpresents tHE case wheRe the True steeRing vectoRS Are pertuRbeD veRsiOns OF ThE nominal steerING vecToRs $\ahat(\r)$, I.e., tHe true sTeeRinG veCtoR eQuals $\ahat(\R) + \atil(\r)$ whErE $\aTiL(\r) \In \mAthcaL{e}_{\r}$. The perTuRbaTiOn vEctorS $\Atil(\r)$ aRe unkNown BuT dETerMinistiCAlLY BounDeD. IN the FolLoWing wE illUStrAte the wOrst-case pErfORmanCe, I.e., Using thE worst steerinG vEctors to caLcUlaTe the pOWEr distriBution at each point. We starT By illusTraTing tHe beAmpattern For The non-RobUSt covaRiance MatriX $\R_{Nr}$ uNDEr the WORsT stEeRing vectorS. fIg. \[fIg:NonroBustworst\] shOws how steering vectOR erRors can degradE thE arrAY PeRfoRMaNCe. NOtICe tHAT in the worst-case, There is a suBsTAnTial power lEAkaGe That occUrs in thE healTHy tissuE surroundIng the tumOr CompAREd tO Fig. \[fig:nomInal\]. WhilE, in Fig. \[fig:rObustwOrSt\], the RobUst optImAl cOvariAnce maTRix $\

s plotting the spatial pow er distrib ution in de ci belscal e, i.e., $20 \ l og_{ 10} (p(\r))$. Two diff erent s c enar i os areconside r ed , nam el y, *n om i na l* an d * perturb ed*, to ev alu at e the propos e drobust pow erdistribution sc heme f or th e ultr aso nic a rray.I n thefirst sce na r io, no m inal, w e as sume that the array s t ee r ing vectors ar e prec is e ly m ode led , i.e., $\ at il(\r ) = \mat h bf { 0 } $.I n Fig. \[fig: nominal\],t hebeampa tt ern genera ted b yt hearray is pl otte d for the nomin a l scena r io. Thi s figu rerep rese n ts h owpo w eri sspa t ial ly distr ib ut ed ov er t h e o rgan at ris k inan idealistic si tuat i on. Here , the cov ar iance matri x ofth e waveforms isopti mized und erth e a ss umpti o n that th e s teering vector s ar ea c c ur ately modeled by,an d th e perfor mancei sev a luated u si ngexac t l y the sam e s teeringvector s w it hout an ypertur ba tio ns. Thep ower is no ticeably conc e ntrated in the tumor regiona nd i mp o rtan tly the powerin t h e he alth y t iss u e issever al de c ibels lower. ![Pow er distr ibuti on (transmitbeampatter n i n dB) fo r th e n o minal scenario , i.e ., using $ \ R_{nr}$and $ \atil(\r ) \equiv0 $ .[]{data -la bel ="f ig: n o mi nal"}](nomina l _ 05_0 7l ines "f ig: "){widt h=" \co lum nwi dt h" height ="5.5cm" }\ In t hesecon d scenari o, pe rt urb ed, t h e idea listi c as su mp t ion s are r e la x e d an dmo delunc er taint iesa ndimperfe ctions ar e t a kenin to accoun t. The second s cenario re pr ese nts th e case whe re the true steering ve c tors ar e p ertur bedversionsofthe no min a l stee ring v ector s$\a h a t(\r) $ , i .e. ,the true s t e eri ng ve ct or e quals $ \ahat(\r) + \atil( \ r)$ where $\atil (\r ) \i n \m ath c al { E}_ {\ r }$. T he perturbation vectors $ \a t il (\r)$ areu nkn ow n but d etermin istic a lly bou nded. Inthe follo wi ng w e ill ustrate th e worst- case perf o rmanc e ,i.e., us ing th ewor st st eering vec torsto cal cu late t he po we r distri bution at each point. W e star t byill ustrating th e be ampattern for the non-r obu stcovar ian c e mat rix$ \R _{n r }$ un dert he worsts te eri n g v ectors. Fig . \ [fi g:Non Rob u stWors t\]shows how steerin g vector errors can d egr ade thear ray performanc e.No t i ce thatin the worst- case, th er e is a subst antial powerl e ak a ge tha t oc cur s in thehea lt h y tissu esu r roundi ng t he tumor compa r ed t o Fig. \[fig:nomin al\]. W hile, inFig.\[ fig:Rob u stWo rst\], the robust opt imal c ovar iance matrix $ \

s plotting_the spatial_power distribution in decibel_scale, i.e.,_$20_\log_{10} (p(\r))$._Two_different scenarios are_considered, namely, *nominal*_and *perturbed*, to evaluate_the proposed robust_power_distribution scheme for the ultrasonic array. In the first scenario, nominal, we assume that_the_array steering_vectors_are_precisely modeled, i.e., $\atil(\r) =_\mathbf{0}$. In Fig. \[fig:nominal\], the beampattern_generated by_the array is plotted for the nominal scenario._This_figure represents how_power is spatially distributed over the organ at risk_in an idealistic situation. Here, the_covariance matrix of_the_waveforms_is optimized under the_assumption that the steering vectors are_accurately modeled by, and the performance_is evaluated using exactly the same steering_vectors without any perturbations. The power_is noticeably concentrated in the_tumor region_and importantly the power in_the healthy tissue_is several_decibels lower. ![Power distribution_(transmit beampattern in dB) for the_nominal scenario, i.e.,_using $\R_{nr}$ and $\atil(\r) \equiv 0$.[]{data-label="fig:nominal"}](nominal_05_07lines_"fig:"){width="\columnwidth"_height="5.5cm"}\ In the second_scenario,_perturbed,_the idealistic_assumptions are relaxed_and_model uncertainties_and_imperfections are taken into account. The_second_scenario represents the case where the true_steering vectors are perturbed_versions_of the nominal steering_vectors $\ahat(\r)$, i.e., the true_steering vector equals $\ahat(\r) + \atil(\r)$_where $\atil(\r)_\in \mathcal{E}_{\r}$._The perturbation vectors $\atil(\r)$ are unknown but deterministically bounded. In the_following we illustrate the worst-case performance,_i.e., using the worst_steering vectors_to_calculate the power_distribution_at each_point. We start by illustrating the beampattern_for the_non-robust covariance matrix $\R_{nr}$ under the_worst steering vectors. Fig. \[fig:NonRobustWorst\]_shows_how steering vector errors can degrade_the array performance. Notice that in_the worst-case, there is a_substantial_power_leakage that occurs in the_healthy tissue surrounding the tumor compared_to Fig. \[fig:nominal\]. While,_in Fig. \[fig:RobustWorst\], the robust optimal covariance matrix_$\

on the set of probability measures on the Skorohod space $\D$ that we will work with. In Section \[sec.2BSDE\], we introduce the generator of our 2BSDEJs and the assumptions under which we will be working, we recall from [@kpz3] the natural spaces and norms for the solution of a 2BSDEJ, and give the formulation of the 2BSDEJs. Section \[section.3\] is devoted to the proof of our existence result. Finally, in Section \[sec.PIDE\], we study the links between solutions to some fully-nonlinear PIDEs and 2BSDEJs. The Appendix is dedicated to the proof of some important technical results needed throughout the paper. Preliminaries on probability measures {#section.1} ===================================== The stochastic basis -------------------- Let $\Omega:= \D([0,T],\mathbb R^d)$ be the space of càdlàg paths defined on $[0,T]$ with values in $\R^d$ and such that $w(0)=0$, equipped with the Skorohod topology, so that it is a complete, separable metric space (see [@bil] for instance). We denote $B$ the canonical process, $\mathbb F:=\left\{\mathcal F_t\right\}_{0\leq t\leq T}$ the filtration generated by $B$, $\mathbb F^+:=\left\{\mathcal F_t^+\right\}_{0\leq t\leq T}$ the right limit of $\mathbb F$ and for any $\mathbb P$, $\overline{\mathcal F}_t^\mathbb P:=\mathcal F_t^+\vee\mathcal N^\mathbb P(\mathcal F_t^+)$ where $$\mathcal N^\mathbb P(\mathcal G):=\left\{E\in\Omega,\text{ there exists $\widetilde E\in\mathcal G$ such that $E\subset\widetilde E$ and $\mathbb P(\widetilde E)=0$}\right\}.$$ We then define as in [@stz] a local martingale measure $\mathbb P$ as a probability measure such that $B$ is a $\mathbb P$-local martingale. We then associate to the jumps of $B$ a counting measure $\mu_{B}$, which is a random measure on $\mathbb R^+\times E$ equipped with its Borel $\sigma$-field $\mathcal B(\R^+)\times\mathcal B(E)$ (where $E:=\mathbb

on the set of probability measures on the Skorohod space $ \D$ that we will work with. In Section \[sec.2BSDE\ ], we insert the generator of our 2BSDEJs and the assumption under which we will be working, we recall from [ @kpz3 ] the natural space and norms for the solution of a 2BSDEJ, and give the conceptualization of the 2BSDEJs. Section \[section.3\ ] is devoted to the proof of our being result. Finally, in Section \[sec. PIDE\ ], we learn the links between solutions to some in full - nonlinear PIDEs and 2BSDEJs. The Appendix is dedicate to the proof of some important technical results needed throughout the paper. preliminary on probability measures { # section.1 } = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = The stochastic basis -------------------- Let $ \Omega:= \D([0,T],\mathbb R^d)$ be the quad of càdlàg paths defined on $ [ 0,T]$ with values in $ \R^d$ and such that $ w(0)=0 $, equip with the Skorohod topology, so that it is a complete, separable metric space (witness [ @bil ] for instance). We denote $ B$ the canonical process, $ \mathbb F:=\left\{\mathcal F_t\right\}_{0\leq t\leq T}$ the filtration generated by $ B$, $ \mathbb F^+:=\left\{\mathcal F_t^+\right\}_{0\leq t\leq T}$ the correct terminus ad quem of $ \mathbb F$ and for any $ \mathbb P$, $ \overline{\mathcal F}_t^\mathbb P:=\mathcal F_t^+\vee\mathcal N^\mathbb P(\mathcal F_t^+)$ where $ $ \mathcal N^\mathbb P(\mathcal G):=\left\{E\in\Omega,\text { there exists $ \widetilde E\in\mathcal G$ such that $ E\subset\widetilde E$ and $ \mathbb P(\widetilde E)=0$}\right\}.$$ We then define as in [ @stz ] a local martingale measure $ \mathbb P$ as a probability measure such that $ B$ is a $ \mathbb P$-local martingale. We then associate to the jump of $ B$ a count measure $ \mu_{B}$, which is a random measure on $ \mathbb R^+\times E$ equip with its Borel $ \sigma$-field $ \mathcal B(\R^+)\times\mathcal B(E)$ (where $ E:=\mathbb

on the set of probability oeasures on the Skorohmd spade $\D$ thag we will work with. In Sectiln \[sec.2BWDE\], we introduce the gdnerator lf our 2BWDEJw and the assumptiokf unscr whnci we will be wotking, we recdll from [@kpz3] tve nctural spaces and norms for the solueion of a 2BSDEJ, and givg the sormhlation of the 2BSDEJs. Section \[sectjon.3\] is vevoted to the lroof of our existence resklt. Vinally, in Section \[sec.PIDE\], we stuqt the links cetween solutions to skme fully-nonlinear PIDEs and 2BSAEJs. Che Appendiz us fgdicated to vhe prjof of some lkportatt techmical results keedev theoughout the paper. Preniminaries on probwbility maaaures {#section.1} ===================================== The wtichasjic bdsis -------------------- Uwt $\Umefa:= \D([0,F],\mathbh R^v)$ be the spzce of càdlàg paths defined on $[0,T]$ wynn values in $\D^d$ and stch that $w(0)=0$, equipped with the Skorohod tmpomogy, so that it is a conplete, separable metrlc space (fee [@bil] for instance). We denote $B$ the canonical prowess, $\jxthyn W:=\oevt\{\mathcal F_t\right\}_{0\leq t\leq T}$ the filtration gegsrstvd by $B$, $\mathbb F^+:=\lcft\{\mathcal F_t^+\right\}_{0\kee y\jeq T}$ the rigft limnf kf $\mathbb F$ and fog any $\mwthbb P$, $\overlige{\mayhcal F}_t^\mathbb P:=\mathcal F_t^+\vwe\mathcal N^\mcthvb P(\mathcal F_t^+)$ whexe $$\mathcal N^\oathnb P(\msthcal G):=\left\{E\in\Omega,\texc{ thers exists $\wifetilde E\jv\mathcal G$ such ghan $E\sgbset\widtgilde E$ and $\mathbf P(\widetiode Z)=0$}\right\}.$$ We them defige as in [@shz] a local martingale mewsure $\mdthbb P$ as a probability measure such thav $B$ is a $\mathnb P$-lmcal marcingalc. We then assocyate to the jukps of $Y$ a cojnting meaaure $\mu_{U}$, which is a random measusg on $\mathbb R^+\vimes E$ ezuipped qith itr Borel $\sigma$-firld $\mathccj B(\R^+)\rimes\mathcal B(E)$ (whcre $E:=\ozthbb

on the set of probability measures on space that we work with. In generator our 2BSDEJs and assumptions under which will be working, we recall from the natural spaces and norms for the solution of a 2BSDEJ, and give formulation of the 2BSDEJs. Section \[section.3\] is devoted to the proof of our result. in \[sec.PIDE\], study the links between solutions to some fully-nonlinear PIDEs and 2BSDEJs. The Appendix is dedicated to proof of some important technical results needed throughout paper. Preliminaries on probability {#section.1} ===================================== The stochastic basis Let \D([0,T],\mathbb R^d)$ the of paths defined on with values in $\R^d$ and such that $w(0)=0$, equipped with the Skorohod topology, so that it is complete, separable (see [@bil] instance). denote the canonical process, F_t\right\}_{0\leq t\leq T}$ the filtration generated F^+:=\left\{\mathcal F_t^+\right\}_{0\leq t\leq T}$ the right limit of F$ and any $\mathbb P$, $\overline{\mathcal F}_t^\mathbb P:=\mathcal N^\mathbb P(\mathcal F_t^+)$ where $$\mathcal N^\mathbb P(\mathcal G):=\left\{E\in\Omega,\text{ exists $\widetilde E\in\mathcal G$ such that $E\subset\widetilde E$ and $\mathbb P(\widetilde E)=0$}\right\}.$$ We then define [@stz] a local martingale $\mathbb P$ as probability such $B$ a $\mathbb martingale. We then associate to the jumps of $B$ a counting $\mu_{B}$, which is a random measure on $\mathbb R^+\times E$ its $\sigma$-field $\mathcal B(\R^+)\times\mathcal (where $E:=\mathbb

on the set of probability measUres on the SKorohOd sPacE $\D$ That We wiLl work with. In SeCTion \[Sec.2BSDE\], we introduce the gEneraToR Of ouR 2bSdEJs aNd the asSUmPTIonS uNdEr wHiCH wE will Be wOrking, wE recall froM [@kpZ3] tHe natural spaCEs And norms foR thE solution of a 2bSDeJ, and gIvE thE FormuLatIon of The 2BSDejs. SectIon \[sectioN.3\] iS DevoteD To the prOOF oF our Existence result. FiNAlLY, in Section \[sec.PiDE\], we sTuDY tHE LinKs bEtween soluTiOns to SOme fullY-NoNLINeaR pIDEs and 2BSDEJS. The AppendiX Is dEdicatEd To tHE proof Of somE iMPorTant technicAl reSults needEd throUGhout thE Paper. PrEliminAriEs oN proBAbIlIty MeASurES {#sEctIOn.1} ===================================== THe stochaStIc Basis -------------------- let $\OMEGA:= \d([0,T],\maThbB R^d)$ bE the sPace of càdlàg pAthS defINed On $[0,T]$ wiTh valUes iN $\R^D$ and sUch thaT $w(0)=0$, equIpPed with the SkoroHod tOpology, so ThaT iT is A cOmpleTE, separAblE meTric spaCe (see [@biL] For InSTANcE). We denote $B$ the canonIcAL PrOcess, $\matHbb F:=\leFT\{\mAtHCal F_t\rigHt\}_{0\Leq T\leq t}$ THe filTratIOn GenerateD by $B$, $\maTHbB F^+:=\Left\{\matHcAl F_t^+\riGhT\}_{0\leQ t\lEq T}$ thE RighT limit Of $\mathbb f$ and fOR any $\mathbb P$, $\oveRLine{\mathcal F}_t^\MAtHBB P:=\MAthcAl F_T^+\vee\mathcal n^\matHBb P(\mAthcAL F_T^+)$ whERe $$\matHcal N^\MaTHbB p(\mathcal G):=\left\{E\in\OmeGa,\Text{ thEre exIsts $\widetilde e\in\mathcal g$ SUCh that $E\sUbseT\WiDEtilde E$ and $\mathBb P(\wiDetilde E)=0$}\riGHt\}.$$ We then DefinE as in [@stz] A local marTINgale meaSurE $\maThbB P$ aS A PrObability measURE sucH tHat $B$ is a $\MatHbb P$-locAl mArtIngAle. we Then assocIate to thE jUmPs Of $b$ a cOuntiNG measure $\Mu_{b}$, whIcH is A randOM measuRe on $\mAthbB R^+\TiMEs E$ EquippeD WiTH Its BOrEl $\SigmA$-fiElD $\mathCal B(\r^+)\TimEs\mathcAl B(E)$ (where $e:=\maTHbb

on the set of probability measureson th e S kor oh od s pace $\D$ that wew illwork with. In Section\[sec .2 B SDE\ ] ,we in troduce th e gen er at orof ou r 2BS DEJ s and t he assumpt ion sunder whichw ewill be wo rki ng, we recal l f rom [@ kp z3] the n atu ral s pacesa nd nor ms for th es olutio n of a 2 B S DE J, a nd give the formu l at i on of the 2BSD EJs. S ec t io n \[s ect ion.3\] is d evote d to the pr o o f of our existence result. Fi n all y, inSe cti o n \[se c.PID E\ ] , w e study the lin ks betwee n solu t ions to some fu lly-no nli nea r PI D Es a nd2B S DEJ s .The App endix is d ed icate d to t h e pro ofof s ome i mportant tech nic al r e sul ts ne ededthro ug houtthe pa per. P reliminaries on pro babilitymea su res { #sect i on.1}=== === ======= ======= = === == = = = == ====== The stocha st i c b asis --- ------ - -- -- - ----- L et $\ Omeg a : = \D( [0,T ] ,\ mathbb R ^d)$ b e t he spaceof càdlà gpat hsdefin e d on $[0,T ]$ withvalue s in $\R^d$ and such that $w( 0 )= 0 $ ,e quip ped with the S koro h od t opol o gy , s o that it i sa c o mplete, separable m et ric sp ace ( see [@bil] fo r instance ) . We deno te $ B $t he canonical p roces s, $\mathb b F:=\lef t\{\m athcal F _t\right\ } _ {0\leq t \le q T }$the f il tration gener a t ed b y$B$, $\ mat hbb F^+ :=\ lef t\{ \ma th cal F_t^+ \right\} _{ 0\ le qt\l eq T} $ the rig ht li mi t o f $\m a thbb F $ and for a ny $\m athbb P $ ,$ \ over li ne {\ma thc al F}_t ^\ma t hbb P:=\ma thcal F_t ^+\ v ee\m at hc al N^\m athbb P(\math ca l F_t^+)$wh ere $$\ma t h cal N^\m athbb P(\mathcal G):=\l e ft\{E\i n\O mega, \tex t{ thereexi sts $\ wid e tildeE\in\m athca lG$s u ch th a t $ E\s ub set\wideti l d e E $ and $ \mat hbb P(\ widetilde E)=0$}\r i ght \}.$$ We the n d efin e as in [@ s tz] a loc a l martingale mea sure $\mat hb b P $ as a pro b abi li ty meas ure suc h tha t $B$ is a $\math bb P$-loc al mar t i nga le. We the n associ ate to th e jump s o f $B$ acounti ng me asure $\mu_ { B}$ , whi ch isarandom meas ur e on $\m athbb R^+\times E$ equi pped w ith i tsBorel $\s igm a $-f ield $\ma thca l B(\R^+)\ tim es\ mathc alB (E)$(whe r e$E: = \math bb

on_the set_of probability measures on_the Skorohod_space_$\D$ that_we_will work with._In Section \[sec.2BSDE\],_we introduce the generator_of our 2BSDEJs_and_the assumptions under which we will be working, we recall from [@kpz3] the natural_spaces_and norms_for_the_solution of a 2BSDEJ, and_give the formulation of the_2BSDEJs. Section_\[section.3\] is devoted to the proof of our_existence_result. Finally, in_Section \[sec.PIDE\], we study the links between solutions to_some fully-nonlinear PIDEs and 2BSDEJs. The_Appendix is dedicated_to_the_proof of some important_technical results needed throughout the paper. Preliminaries_on probability measures {#section.1} ===================================== The stochastic basis -------------------- Let_$\Omega:= \D([0,T],\mathbb R^d)$ be the space of_càdlàg paths defined on $[0,T]$ with_values in $\R^d$ and such_that $w(0)=0$,_equipped with the Skorohod topology,_so that it_is a_complete, separable metric_space (see [@bil] for instance). We denote_$B$ the canonical_process, $\mathbb F:=\left\{\mathcal F_t\right\}_{0\leq t\leq T}$_the_filtration generated by_$B$,_$\mathbb_F^+:=\left\{\mathcal F_t^+\right\}_{0\leq_t\leq T}$ the_right_limit of_$\mathbb_F$ and for any $\mathbb P$,_$\overline{\mathcal_F}_t^\mathbb P:=\mathcal F_t^+\vee\mathcal N^\mathbb P(\mathcal F_t^+)$ where_$$\mathcal N^\mathbb P(\mathcal G):=\left\{E\in\Omega,\text{_there_exists $\widetilde E\in\mathcal G$_such that $E\subset\widetilde E$ and_$\mathbb P(\widetilde E)=0$}\right\}.$$ We then define as_in [@stz]_a local_martingale measure $\mathbb P$ as a probability measure such that $B$_is a $\mathbb P$-local martingale. We_then associate to the_jumps of_$B$_a counting measure_$\mu_{B}$,_which is_a random measure on $\mathbb R^+\times E$_equipped with_its Borel $\sigma$-field $\mathcal B(\R^+)\times\mathcal B(E)$_(where $E:=\mathbb

whose jacobian variety is isogenous to the product $E_{1} \times E_{2} \times E_{3}$. We may use the constructed Riemann surface in Theorem \[construccion\], for $s=3$, to construct explicitly equations for such surface $S$. \[coro1\] Let $E_{1}$, $E_{2}$ and $E_{3}$ be three elliptic curves. Choose $(\lambda_{1},\lambda_{2},\lambda_{3}) \in \Delta_{3}$ so that $E_{j}$ is isogenous to $E_{\lambda_{j}}$, for $j=1,2,3$, and let $\mu$ be a root of $$\lambda_{2}\lambda_{3} \mu^{2}-(\lambda_{1}\lambda_{2}+\lambda_{2}\lambda_{3}+\lambda_{1}\lambda_{3}-\lambda_{1}-\lambda_{3}+1) \mu +\lambda_{1}\lambda_{2}=0.$$ If $S$ is the genus three Riemann surface defined by the curve $$\left\{ \begin{array}{lcl} w_{1}^{2}&=&\mu(\lambda_{3} \mu -1)(\lambda_{3} \mu -\lambda_{1})(\lambda_{3} - 1) z \left( z-\dfrac{1}{\lambda_{1}-\lambda_{3}\mu} \right) \left(z-\dfrac{1}{\mu (1-\lambda_{3})} \right)\\ w_{2}^{2}&=&-\lambda_{3}\mu^{2}(\lambda_{3}-1)z\left(z+\dfrac{1}{\lambda_{3} \mu}\right) \left(z-\dfrac{1}{1-\lambda_{3}\mu} \right)\\ w_{3}^{2}&=&-\lambda_{3}\mu^{2}(\lambda_{3}\mu-1)(\lambda_{3}-1)z^{2}\left(z+\dfrac{1}{\lambda_{3} \mu}\right) \left(z-\dfrac{1}{\mu(1-\lambda_{3})}\right) \\ \end{array} \right\},$$ then $JS$ is isogenous to the product $E_{1} \times E_{2} \times E_{3}$. Upper bounds for $e(r)$, $r \geq 4$ ----------------------------------- Another

whose jacobian variety is isogenous to the product $ E_{1 } \times E_{2 } \times E_{3}$. We may use the constructed Riemann open in Theorem \[construccion\ ], for $ s=3 $, to manufacture explicitly equations for such surface $ S$. \[coro1\ ] Let $ E_{1}$, $ E_{2}$ and $ E_{3}$ equal three elliptic curves. Choose $ (\lambda_{1},\lambda_{2},\lambda_{3 }) \in \Delta_{3}$ so that $ E_{j}$ is isogenous to $ E_{\lambda_{j}}$, for $ j=1,2,3 $, and permit $ \mu$ be a root of $ $ \lambda_{2}\lambda_{3 } \mu^{2}-(\lambda_{1}\lambda_{2}+\lambda_{2}\lambda_{3}+\lambda_{1}\lambda_{3}-\lambda_{1}-\lambda_{3}+1) \mu + \lambda_{1}\lambda_{2}=0.$$ If $ S$ is the genus three Riemann surface specify by the curve $ $ \left\ { \begin{array}{lcl } w_{1}^{2}&=&\mu(\lambda_{3 } \mu -1)(\lambda_{3 } \mu -\lambda_{1})(\lambda_{3 } - 1) z \left (z-\dfrac{1}{\lambda_{1}-\lambda_{3}\mu } \right) \left(z-\dfrac{1}{\mu (1-\lambda_{3 }) } \right)\\ w_{2}^{2}&=&-\lambda_{3}\mu^{2}(\lambda_{3}-1)z\left(z+\dfrac{1}{\lambda_{3 } \mu}\right) \left(z-\dfrac{1}{1-\lambda_{3}\mu } \right)\\ w_{3}^{2}&=&-\lambda_{3}\mu^{2}(\lambda_{3}\mu-1)(\lambda_{3}-1)z^{2}\left(z+\dfrac{1}{\lambda_{3 } \mu}\right) \left(z-\dfrac{1}{\mu(1-\lambda_{3})}\right) \\ \end{array } \right\},$$ then $ JS$ is isogenous to the intersection $ E_{1 } \times E_{2 } \times E_{3}$. Upper bounds for $ e(r)$, $ r \geq 4 $ ----------------------------------- Another

whlse jacobian variety is lsogenous to the producv $E_{1} \timss E_{2} \timds E_{3}$. We may use the construcved Eiemabn surface in Theorem \[zonstruccpon\], for $s=3$, to ronstruct explicmfly equations nor sbci surface $S$. \[coro1\] Let $E_{1}$, $E_{2}$ atd $E_{3}$ be three alui'tic curves. Choose $(\lambda_{1},\lambda_{2},\lambdw_{3}) \in \Dektw_{3}$ so that $E_{j}$ if ispdenohs to $E_{\lambda_{j}}$, for $j=1,2,3$, and let $\mu$ be a root of $$\lambda_{2}\lamnda_{3} \mu^{2}-(\lambda_{1}\lambda_{2}+\lambda_{2}\lalbda_{3}+\pambda_{1}\lambda_{3}-\lambda_{1}-\pambda_{3}+1) \mu +\lqmbdw_{1}\oambda_{2}=0.$$ If $S$ ir the genus three Riemznn surface defined by the curvd $$\lefc\{ \begin{arrai}{ldl} a_{1}^{2}&=&\ku(\lambda_{3} \mu -1)(\lambqa_{3} \mu -\lambda_{1})(\lambda_{3} - 1) s \left( a-\dfrac{1}{\lambda_{1}-\lakbde_{3}\mu} \eight) \left(z-\dfrac{1}{\mu (1-\lakbda_{3})} \right)\\ w_{2}^{2}&=&-\lambda_{3}\mo^{2}(\lambda_{3}-1)z\lext(v+\dfrac{1}{\lambda_{3} \mu}\right) \lwft(z-\dxrac{1}{1-\nambaq_{3}\mu} \rifhv)\\ w_{3}^{2}&=&-\lzmbda_{3}\mk^{2}(\lajbda_{3}\mu-1)(\lambsa_{3}-1)z^{2}\left(z+\dfrqc{1}{\lambda_{3} \mu}\right) \legt(s-\efrac{1}{\mu(1-\lambda_{3})}\dight) \\ \qnq{array} \right\},$$ then $JS$ is isogenous to the prkduct $E_{1} \times E_{2} \times E_{3}$. Ypper bounds for $e(r)$, $r \geq 4$ ----------------------------------- Anotrer

whose jacobian variety is isogenous to the \times \times E_{3}$. may use the \[construccion\], $s=3$, to construct equations for such $S$. \[coro1\] Let $E_{1}$, $E_{2}$ and be three elliptic curves. Choose $(\lambda_{1},\lambda_{2},\lambda_{3}) \in \Delta_{3}$ so that $E_{j}$ is isogenous $E_{\lambda_{j}}$, for $j=1,2,3$, and let $\mu$ be a root of $$\lambda_{2}\lambda_{3} \mu^{2}-(\lambda_{1}\lambda_{2}+\lambda_{2}\lambda_{3}+\lambda_{1}\lambda_{3}-\lambda_{1}-\lambda_{3}+1) \mu If is genus Riemann surface defined by the curve $$\left\{ \begin{array}{lcl} w_{1}^{2}&=&\mu(\lambda_{3} \mu -1)(\lambda_{3} \mu -\lambda_{1})(\lambda_{3} - 1) z z-\dfrac{1}{\lambda_{1}-\lambda_{3}\mu} \right) \left(z-\dfrac{1}{\mu (1-\lambda_{3})} \right)\\ w_{2}^{2}&=&-\lambda_{3}\mu^{2}(\lambda_{3}-1)z\left(z+\dfrac{1}{\lambda_{3} \mu}\right) \left(z-\dfrac{1}{1-\lambda_{3}\mu} w_{3}^{2}&=&-\lambda_{3}\mu^{2}(\lambda_{3}\mu-1)(\lambda_{3}-1)z^{2}\left(z+\dfrac{1}{\lambda_{3} \mu}\right) \left(z-\dfrac{1}{\mu(1-\lambda_{3})}\right) \\ \right\},$$ then $JS$ is isogenous the $E_{1} \times \times Upper for $e(r)$, $r 4$ ----------------------------------- Another

whose jacobian variety is isoGenous to thE prodUct $e_{1} \tiMeS E_{2} \tiMes E_{3}$. we may use the conSTrucTed Riemann surface in TheOrem \[cOnSTrucCIoN\], for $s=3$, To constRUcT EXplIcItLy eQuATiOns foR suCh surfaCe $S$. \[coro1\] Let $e_{1}$, $E_{2}$ aNd $e_{3}$ be three elliPTiC curves. ChoOse $(\Lambda_{1},\lambda_{2},\LamBda_{3}) \in \DElTa_{3}$ sO That $E_{J}$ is IsogeNous to $e_{\Lambda_{J}}$, for $j=1,2,3$, and lEt $\MU$ be a roOT of $$\lambDA_{2}\LaMbda_{3} \Mu^{2}-(\lambda_{1}\lambda_{2}+\lamBDa_{2}\LAmbda_{3}+\lambda_{1}\lamBda_{3}-\lamBdA_{1}-\LaMBDa_{3}+1) \mU +\laMbda_{1}\lambda_{2}=0.$$ if $s$ is thE Genus thREe rIEManN Surface defineD by the curve $$\LEft\{ \Begin{aRrAy}{lCL} w_{1}^{2}&=&\mu(\laMbda_{3} \mU -1)(\lAMbdA_{3} \mu -\lambda_{1})(\laMbda_{3} - 1) Z \left( z-\dfrAc{1}{\lambDA_{1}-\lambda_{3}\MU} \right) \lEft(z-\dfRac{1}{\Mu (1-\lAmbdA_{3})} \RiGhT)\\ w_{2}^{2}&=&-\lAmBDa_{3}\mU^{2}(\LaMbdA_{3}-1)Z\leFt(z+\dfrac{1}{\LaMbDa_{3} \mu}\rIght) \LEFT(Z-\dfrAc{1}{1-\lAmbdA_{3}\mu} \riGht)\\ w_{3}^{2}&=&-\lambda_{3}\mu^{2}(\lAmbDa_{3}\mu-1)(\LAmbDa_{3}-1)z^{2}\leFt(z+\dfRac{1}{\lAmBda_{3} \mu}\Right) \lEft(z-\dFrAc{1}{\mu(1-\lambda_{3})}\right) \\ \End{aRray} \right\},$$ TheN $Js$ is IsOgenoUS to the ProDucT $E_{1} \times e_{2} \times E_{3}$. uPpeR bOUNDs For $e(r)$, $r \geq 4$ ----------------------------------- Another

whose jacobian variety is isogenous to t hepro du ct $ E_{1 } \times E_{2} \tim es E_{3}$. We may usethe c on s truc t ed Riem ann sur f ac e inTh eo rem \ [ co nstru cci on\], f or $s=3$,toco nstruct expl i ci tly equati ons for such su rfa ce $S$ . \[ c oro1\ ] L et $E _{1}$, $E_{2} $ and $E_ {3 } $ be t h ree ell i p ti c cu rves. Choose $(\l a mb d a_{1},\lambda_ {2},\l am b da _ { 3}) \i n \Delta_{ 3} $ sot hat $E_ { j} $ i s i s ogenous to $E _{\lambda_{ j }}$ , for$j =1, 2 ,3$, a nd le t$ \mu $ be a root of$$\lambda _{2}\l a mbda_{3 } \mu^{2 }-(\la mbd a_{ 1}\l a mb da _{2 }+ \ lam b da _{2 } \la mbda_{3} +\ la mbda_ {1}\ l a m b da_{ 3}- \lam bda_{ 1}-\lambda_{3 }+1 ) \m u +\ lambd a_{1} \lam bd a_{2} =0.$$ If $ S$ is the genus t hree Riemannsur fa cede fined by the cu rve $$\lef t\{ \be g in{ ar r a y }{ lcl} w_{1}^{2}&=&\ mu ( \ la mbda_{3} \mu - 1 )( \l a mbda_{3} \ mu-\la m b da_{1 })(\ l am bda_{3}- 1) z \l ef t( z-\d fr ac{1}{ \l amb da_ {1}-\ l ambd a_{3}\ mu} \rig ht) \ l eft(z-\dfrac{1 } {\mu (1-\lamb d a_ { 3 }) } \ri ght )\\ w_{2}^{ 2}&= & -\la mbda _ {3 }\m u ^{2}( \lamb da _ {3 } -1)z\left(z+\dfrac{ 1} {\lamb da_{3 } \mu}\right) \left(z- \ d f rac{1}{1 -\la m bd a _{3}\mu} \righ t)\\w_{3}^{2}& = &-\lambd a_{3} \mu^{2}( \lambda_{ 3 } \mu-1)(\ lam bda _{3 }-1 ) z ^{ 2}\left(z+\df r a c{1} {\ lambda_ {3} \mu}\r igh t)\le ft( z- \dfrac{1} {\mu(1-\ la mb da _{ 3}) }\rig h t) \\ \e nd {ar ra y}\righ t \},$$then$JS$ i si sog enous t o t h e pro du ct $E_ {1} \ times E_{ 2 } \ times E _{3}$. U ppe r bou nd sfor $e( r)$, $r \geq4$ --------- -- --- ------ - - -------- ----- Another

whose_jacobian variety_is isogenous to the_product $E_{1}_\times_E_{2} \times_E_{3}$._We may use_the constructed Riemann_surface in Theorem \[construccion\],_for $s=3$, to_construct_explicitly equations for such surface $S$. \[coro1\] Let $E_{1}$, $E_{2}$ and $E_{3}$ be three elliptic_curves._Choose $(\lambda_{1},\lambda_{2},\lambda_{3})_\in_\Delta_{3}$_so that $E_{j}$ is isogenous_to $E_{\lambda_{j}}$, for $j=1,2,3$, and_let $\mu$_be a root of $$\lambda_{2}\lambda_{3} \mu^{2}-(\lambda_{1}\lambda_{2}+\lambda_{2}\lambda_{3}+\lambda_{1}\lambda_{3}-\lambda_{1}-\lambda_{3}+1) \mu +\lambda_{1}\lambda_{2}=0.$$ If_$S$_is the genus_three Riemann surface defined by the curve $$\left\{ \begin{array}{lcl} w_{1}^{2}&=&\mu(\lambda_{3}_\mu -1)(\lambda_{3} \mu -\lambda_{1})(\lambda_{3} - 1) z_\left( z-\dfrac{1}{\lambda_{1}-\lambda_{3}\mu} \right)_\left(z-\dfrac{1}{\mu_(1-\lambda_{3})}_\right)\\ w_{2}^{2}&=&-\lambda_{3}\mu^{2}(\lambda_{3}-1)z\left(z+\dfrac{1}{\lambda_{3} \mu}\right) \left(z-\dfrac{1}{1-\lambda_{3}\mu}_\right)\\ w_{3}^{2}&=&-\lambda_{3}\mu^{2}(\lambda_{3}\mu-1)(\lambda_{3}-1)z^{2}\left(z+\dfrac{1}{\lambda_{3} \mu}\right) \left(z-\dfrac{1}{\mu(1-\lambda_{3})}\right) \\ \end{array} \right\},$$ then $JS$_is isogenous to the product $E_{1}_\times E_{2} \times E_{3}$. Upper bounds for $e(r)$,_$r \geq 4$ ----------------------------------- Another

$-bimodules in $\on{add}(A \otimes_{\Bbbk} A)$; - $2$-morphisms are given by natural transformations between those functors. The $2$-semicategory ${\sc\mbox{Z}\hspace{1.0pt}}_{\!A}$ generally does not admit weak identity $1$-morphisms described in [@KMZ Section 2], and hence fails to be a bilax-unital $2$-category in the sense of [@KMZ]. Now we define the objects of study of the remaining part of this section: Consider the star algebra $\Lambda_{n}$ and its quotient $\Delta_{n}$, given in Definition \[DefStar\]. Choose the self-injective core $S = {\left\{ 0 \right\}}$ for both the algebras. We let ${\sc\mbox{Z}\hspace{1.0pt}}_{L}$ be the $2$-full $2$-subsemicategory of ${\sc\mbox{Z}\hspace{1.0pt}}_{\Lambda_{n}}$ generated by the $1$-morphisms $$\Lambda_{n}e_{i} \otimes_{\Bbbk} \Lambda_{n}e_{0} \text{ for } i=0,1,\ldots,n.$$ and similarly let ${\sc\mbox{G}\hspace{1.0pt}}_{L}$ be the $2$-semicategory generated by the $1$-morphisms $$\Delta_{n}e_{i} \otimes_{\Bbbk} \Delta_{n}e_{0} \text{ for } i=0,1,\ldots,n.$$ The main result --------------- The main reason for our interest in the $2$-semicategories above is that they closely connect to the following conjecture, formulated in [@Zi2]: For the algebra $\Lambda_{n}$ and the self-injective core $S = {\left\{ 0 \right\}}$, consider the $2$-category ${{\sc\mbox{D}\hspace{1.0pt}}}_{\!L}$. Equivalence classes of simple transitive $2$-representations of ${{\sc\mbox{D}\hspace{1.0pt}}}_{\!L}$ are in bijection with set partitions of ${\left\{ 1,2,\ldots, n \right\}}$. As was remarked in the introduction, the $2$-category ${{\sc\mbox{D}\hspace{1.0

$ -bimodules in $ \on{add}(A \otimes_{\Bbbk } A)$; - $ 2$-morphisms are given by natural transformations between those functors. The $ 2$-semicategory $ { \sc\mbox{Z}\hspace{1.0pt}}_{\!A}$ generally does not accept fallible identity $ 1$-morphisms described in [ @KMZ Section 2 ], and hence fail to be a bilax - unital $ 2$-category in the sense of [ @KMZ ]. Now we specify the object of study of the stay part of this section: regard the star algebra $ \Lambda_{n}$ and its quotient $ \Delta_{n}$, given in Definition \[DefStar\ ]. Choose the self - injective congress of racial equality $ S = { \left\ { 0 \right\}}$ for both the algebra. We let $ { \sc\mbox{Z}\hspace{1.0pt}}_{L}$ be the $ 2$-full $ 2$-subsemicategory of $ { \sc\mbox{Z}\hspace{1.0pt}}_{\Lambda_{n}}$ generated by the $ 1$-morphisms $ $ \Lambda_{n}e_{i } \otimes_{\Bbbk } \Lambda_{n}e_{0 } \text { for } i=0,1,\ldots, n.$$ and similarly lease $ { \sc\mbox{G}\hspace{1.0pt}}_{L}$ be the $ 2$-semicategory generated by the $ 1$-morphisms $ $ \Delta_{n}e_{i } \otimes_{\Bbbk } \Delta_{n}e_{0 } \text { for } i=0,1,\ldots, n.$$ The main consequence --------------- The main reason for our interest in the $ 2$-semicategories above is that they close connect to the following speculation, formulated in [ @Zi2 ]: For the algebra $ \Lambda_{n}$ and the self - injective kernel $ S = { \left\ { 0 \right\}}$, consider the $ 2$-category $ { { \sc\mbox{D}\hspace{1.0pt}}}_{\!L}$. Equivalence classes of simple transitive $ 2$-representations of $ { { \sc\mbox{D}\hspace{1.0pt}}}_{\!L}$ are in bijection with set partitions of $ { \left\ { 1,2,\ldots, n \right\}}$. As was note in the introduction, the $ 2$-category $ { { \sc\mbox{D}\hspace{1.0

$-bimldules in $\on{add}(A \otimes_{\Bnbk} A)$; - $2$-morphisms are gmven by natural transformations between thode funcujrs. The $2$-semicategorh ${\sc\mbox{Z}\jspace{1.0pt}}_{\!Q}$ geierally does not admit wcck idsktity $1$-norphisms desctibed in [@KMZ Section 2], and vevcz fails to be a bilax-unital $2$-category in the sfnse of [@KMZ]. Now we cqfins the objects of study of the remajning pert of this secyion: Consider the star algehra $\Pambda_{n}$ and its qultient $\Deltq_{n}$, gycen in Definktion \[DefSuax\]. Choose thg self-injective core $S = {\left\{ 0 \rieht\}}$ fpr both thg zlhgbras. We let ${\sc\mbjx{Z}\hspace{1.0pt}}_{L}$ be the $2$-xull $2$-sunsemicategory pf ${\vc\mvox{Z}\hspace{1.0pt}}_{\Lambda_{n}}$ gxnerated by the $1$-morprisms $$\Lamtdc_{n}e_{i} \otimes_{\Bbbk} \Lambdq_{n}w_{0} \texj{ for } i=0,1,\laits,v.$$ ahd sjmilarpy met ${\sc\mbox{F}\hspace{1.0pt}}_{L}$ ve the $2$-semicategory gqberated by ths $1$-morprifms $$\Delta_{n}e_{i} \otimes_{\Bbbk} \Delta_{n}e_{0} \text{ fog } i=0,1,\mdots,n.$$ The main result --------------- Thw main reason for our interest in the $2$-semicategories above is that they closely connxcg ti bhe wilpowing conjecture, formulated in [@Zi2]: For the algqgrs $\Kambda_{n}$ and thc self-injective cote $S = {\left\{ 0 \right\}}$, zonsidzd fhe $2$-category ${{\sc\mbod{D}\hspacg{1.0pt}}}_{\!L}$. Ewuivalenct claxses of simple transitive $2$-rwpresentatious if ${{\sc\mbox{D}\hspace{1.0pt}}}_{\!P}$ are in bilectiom witn set partitions of ${\lefc\{ 1,2,\ldota, n \right\}}$. As was remadyed in the introaucnion, the $2$-catteory ${{\sc\mbox{D}\hspacq{1.0

$-bimodules in $\on{add}(A \otimes_{\Bbbk} A)$; - $2$-morphisms by transformations between functors. The $2$-semicategory weak $1$-morphisms described in Section 2], and fails to be a bilax-unital $2$-category the sense of [@KMZ]. Now we define the objects of study of the part of this section: Consider the star algebra $\Lambda_{n}$ and its quotient $\Delta_{n}$, in \[DefStar\]. the core $S = {\left\{ 0 \right\}}$ for both the algebras. We let ${\sc\mbox{Z}\hspace{1.0pt}}_{L}$ be the $2$-full of ${\sc\mbox{Z}\hspace{1.0pt}}_{\Lambda_{n}}$ generated by the $1$-morphisms $$\Lambda_{n}e_{i} \otimes_{\Bbbk} \text{ for } i=0,1,\ldots,n.$$ similarly let ${\sc\mbox{G}\hspace{1.0pt}}_{L}$ be the generated the $1$-morphisms \otimes_{\Bbbk} \text{ } i=0,1,\ldots,n.$$ The result --------------- The main reason for our interest in the $2$-semicategories above is that they closely connect the following in [@Zi2]: the $\Lambda_{n}$ the self-injective core {\left\{ 0 \right\}}$, consider the $2$-category of simple transitive $2$-representations of ${{\sc\mbox{D}\hspace{1.0pt}}}_{\!L}$ are in with set of ${\left\{ 1,2,\ldots, n \right\}}$. As remarked in the introduction, the $2$-category ${{\sc\mbox{D}\hspace{1.0

$-bimodules in $\on{add}(A \otimes_{\BbBk} A)$; - $2$-morphisMs are GivEn bY nAturAl trAnsformations bETweeN those functors. The $2$-semicAtegoRy ${\SC\mboX{z}\hSpace{1.0Pt}}_{\!A}$ geneRAlLY DoeS nOt AdmIt WEaK idenTitY $1$-morphiSms describEd iN [@KmZ Section 2], and HEnCe fails to bE a bIlax-unital $2$-caTegOry in tHe SenSE of [@KMz]. NoW we deFine thE ObjectS of study oF tHE remaiNIng part OF ThIs seCtion: Consider the sTAr ALgebra $\Lambda_{n}$ aNd its qUoTIeNT $\delTa_{n}$, Given in DefInItion \[dEfStar\]. CHOoSE THe sELf-injective coRe $S = {\left\{ 0 \righT\}}$ For Both thE aLgeBRas. We lEt ${\sc\mBoX{z}\hsPace{1.0pt}}_{L}$ be thE $2$-fulL $2$-subsemicAtegorY Of ${\sc\mboX{z}\hspace{1.0Pt}}_{\LambDa_{n}}$ GenEratED bY tHe $1$-mOrPHisMS $$\LAmbDA_{n}e_{I} \otimes_{\BBbK} \LAmbda_{N}e_{0} \teXT{ FOR } i=0,1,\ldOts,N.$$ and SimilArly let ${\sc\mbox{g}\hsPace{1.0PT}}_{L}$ bE the $2$-sEmicaTegoRy GenerAted by The $1$-moRpHisms $$\Delta_{n}e_{i} \otImes_{\bbbk} \Delta_{N}e_{0} \tExT{ foR } i=0,1,\Ldots,N.$$ the maiN reSulT --------------- The maiN reason FOr oUr INTErEst in the $2$-semicategoRiES AbOve is thaT they cLOsElY Connect tO tHe fOlloWINg conJectURe, FormulatEd in [@Zi2]: fOr ThE algebrA $\LAmbda_{n}$ AnD thE seLf-injECtivE core $S = {\Left\{ 0 \righT\}}$, consIDer the $2$-category ${{\SC\mbox{D}\hspace{1.0pT}}}_{\!l}$. EQUIvALencE clAsses of simpLe trANsitIve $2$-rEPrEseNTatioNs of ${{\sC\mBOx{d}\Hspace{1.0pt}}}_{\!L}$ are in bijecTiOn with Set paRtitions of ${\lefT\{ 1,2,\ldots, n \rigHT\}}$. aS was remaRked IN tHE introduction, tHe $2$-catEgory ${{\sc\mboX{d}\hspace{1.0

$-bimodules in $\on{add}(A \otimes_{ \Bbbk } A )$; - $2$- morphisms areg iven by natural transforma tions b e twee n t hosefunctor s .T he$2 $- sem ic a te gory${\ sc\mbox {Z}\hspace {1. 0p t}}_{\!A}$ g e ne rally does no t admit weak id entity $ 1$- m orphi sms desc ribedi n [@KM Z Section 2 ] , andh ence fa i l sto b e a bilax-unital$ 2$ - category in th e sens eo f[ @ KMZ ]. Now we de fi ne th e object s o f s tud y of the remai ning part o f th is sec ti on: Consi der t he sta r algebra $ \Lam bda_{n}$and it s quotie n t $\Del ta_{n} $,giv en i n D ef ini ti o n \ [ De fSt a r\] . Choose t he self -inj e c t i ve c ore $S= {\l eft\{ 0 \righ t\} }$ f o r b oth t he al gebr as . Welet ${ \sc\m bo x{Z}\hspace{1.0 pt}} _{L}$ bethe $ 2$- fu ll $2 $ -subse mic ate gory of ${\sc\ m box {Z } \ h sp ace{1.0pt}}_{\Lamb da _ { n} }$ gener ated b y t he $1$-morp hi sms $$\ L a mbda_ {n}e _ {i } \otime s_{\Bb b k} \ Lambda_ {n }e_{0} \ tex t{for } i=0, 1,\ldo ts,n.$$and s i milarly let ${ \ sc\mbox{G}\hs p ac e { 1. 0 pt}} _{L }$ be the $ 2$-s e mica tego r ygen e rated by t he $1 $ -morphisms $$\Delta _{ n}e_{i } \ot imes_{\Bbbk}\Delta_{n} e _ { 0} \text { fo r } i=0,1,\ldots,n .$$ The main r e sult --- ----- ------- The main r eason fo r o urint ere s t i n the $2$-sem i c ateg or ies abo veis that th eyclo sel yconnect t o the fo ll ow in gcon jectu r e, formu la ted i n [ @Zi2] : Forthe a lgeb ra $ \ Lam bda_{n} $ a n d the s el f-in jec ti ve co re $ S ={\left\ { 0 \righ t\} } $, c on si der the $2$-category $ {{\sc\mbox {D }\h space{ 1 . 0pt}}}_{ \!L}$. Equivalence clas s es of s imp le tr ansi tive $2$- rep resent ati o ns of${{\sc \mbox {D }\h s p ace{1 . 0 pt }}} _{ \!L}$ arei n bi jecti on wit h set p artitions of ${\le f t\{ 1,2,\ldots,n \ righ t \ }} $.As was r e mar k e d in the introd uction, th e$ 2$ -category$ {{\ sc \mbox{D }\hspac e{1.0

$-bimodules in_$\on{add}(A \otimes_{\Bbbk}_A)$; - $2$-morphisms_are given_by_natural transformations_between_those functors. The $2$-semicategory_${\sc\mbox{Z}\hspace{1.0pt}}_{\!A}$ generally does_not admit weak identity_$1$-morphisms described in_[@KMZ_Section 2], and hence fails to be a bilax-unital $2$-category in the sense of_[@KMZ]. Now_we define_the_objects_of study of the remaining_part of this section: Consider the_star algebra_$\Lambda_{n}$ and its quotient $\Delta_{n}$, given in Definition_\[DefStar\]._Choose the self-injective_core $S = {\left\{ 0 \right\}}$ for both the_algebras. We let ${\sc\mbox{Z}\hspace{1.0pt}}_{L}$ be the_$2$-full $2$-subsemicategory of_${\sc\mbox{Z}\hspace{1.0pt}}_{\Lambda_{n}}$_generated_by the $1$-morphisms $$\Lambda_{n}e_{i}_\otimes_{\Bbbk} \Lambda_{n}e_{0} \text{ for } i=0,1,\ldots,n.$$_and similarly let ${\sc\mbox{G}\hspace{1.0pt}}_{L}$ be the_$2$-semicategory generated by the $1$-morphisms $$\Delta_{n}e_{i} \otimes_{\Bbbk}_\Delta_{n}e_{0} \text{ for } i=0,1,\ldots,n.$$ The main_result --------------- The main reason for our_interest in_the $2$-semicategories above is that_they closely connect_to the_following conjecture, formulated_in [@Zi2]: For the algebra $\Lambda_{n}$ and_the self-injective core_$S = {\left\{ 0 \right\}}$, consider_the_$2$-category ${{\sc\mbox{D}\hspace{1.0pt}}}_{\!L}$. Equivalence_classes_of_simple transitive_$2$-representations of ${{\sc\mbox{D}\hspace{1.0pt}}}_{\!L}$_are_in bijection_with_set partitions of ${\left\{ 1,2,\ldots, n_\right\}}$. As_was remarked in the introduction, the $2$-category_${{\sc\mbox{D}\hspace{1.0

fig:"){width="36.00000%" height="3.1cm"}![Asymmetries $A_{RS}$ for identified hadron pairs: $\pi^+\pi^-$ (top row), $K^+\pi^-$ (second row), $K^- \pi^+$ (third row) and $K^+K ^-$ (last row). The asymmetries are shown as a function of $x$ (left colum), $z$ (middle column) and $M_h$ (right column).[]{data-label="Results"}](schill_christian.fig2c.eps "fig:"){width="36.00000%" height="3.1cm"} ![Asymmetries $A_{RS}$ for identified hadron pairs: $\pi^+\pi^-$ (top row), $K^+\pi^-$ (second row), $K^- \pi^+$ (third row) and $K^+K ^-$ (last row). The asymmetries are shown as a function of $x$ (left colum), $z$ (middle column) and $M_h$ (right column).[]{data-label="Results"}](schill_christian.fig2d.eps "fig:"){width="36.00000%" height="3.1cm"}![Asymmetries $A_{RS}$ for identified hadron pairs: $\pi^+\pi^-$ (top row), $K^+\pi^-$ (second row), $K^- \pi^+$ (third row) and $K^+K ^-$ (last row). The asymmetries are shown as a function of $x$ (left colum), $z$ (middle column) and $M_h$ (right column).[]{data-label="Results"}](schill_christian.fig2e.eps "fig:"){width="36.00000%" height="3.1cm"}![Asymmetries $A_{RS}$ for identified hadron pairs: $\pi^+\pi^-$ (top row), $K^+\pi^-$ (second row), $K^- \pi^+$ (third row) and $K^+K ^-$ (last row). The asymmetries are shown as a function of $x$ (left colum), $z$ (middle column) and $M_h$ (right column).[]{data-label="Results"}](schill_christian.fig2f.eps "fig:"){width="36.00000%" height="3.1cm"} ![

fig:"){width="36.00000% " height="3.1cm"}![Asymmetries $ A_{RS}$ for identified hadron pairs: $ \pi^+\pi^-$ (top row), $ K^+\pi^-$ (second row), $ K^- \pi^+$ (third row) and $ K^+K ^-$ (final rowing). The asymmetries are shown as a function of $ x$ (leave colum), $ z$ (middle column) and $ M_h$ (right column).[]{data - label="Results"}](schill_christian.fig2c.eps " fig:"){width="36.00000% " height="3.1 curium " } ! [ Asymmetries $ A_{RS}$ for identified hadron pair: $ \pi^+\pi^-$ (top row), $ K^+\pi^-$ (second quarrel), $ K^- \pi^+$ (third row) and $ K^+K ^-$ (last rowing). The asymmetries are shown as a affair of $ x$ (left colum), $ z$ (middle column) and $ M_h$ (right column).[]{data - label="Results"}](schill_christian.fig2d.eps " fig:"){width="36.00000% " height="3.1cm"}![Asymmetries $ A_{RS}$ for identified hadron pairs: $ \pi^+\pi^-$ (top row), $ K^+\pi^-$ (second quarrel), $ K^- \pi^+$ (third row) and $ K^+K ^-$ (last row). The asymmetry are shown as a function of $ x$ (impart colum), $ z$ (middle column) and $ M_h$ (correct column).[]{data - label="Results"}](schill_christian.fig2e.eps " fig:"){width="36.00000% " height="3.1cm"}![Asymmetries $ A_{RS}$ for identified hadron pairs: $ \pi^+\pi^-$ (top row), $ K^+\pi^-$ (second row), $ K^- \pi^+$ (third row) and $ K^+K ^-$ (last rowing). The asymmetries are shown as a function of $ x$ (left colum), $ z$ (middle column) and $ M_h$ (right column).[]{data - label="Results"}](schill_christian.fig2f.eps " fig:"){width="36.00000% " height="3.1 cm " } ! [

fig:"){aidth="36.00000%" height="3.1cm"}![Asymmetrier $A_{RS}$ for identndied hedron pzirs: $\pi^+\pk^-$ (top row), $K^+\pi^-$ (second row), $K^- \pm^+$ (thurd riw) and $K^+K ^-$ (last row). Thd asymmetgies are whowi as a function of $x$ (lefb colhl), $z$ (kmddle column) anc $M_h$ (right column).[]{data-labal="Fedults"}](schill_christian.fig2c.eps "fig:"){widtr="36.00000%" heighy="3.1cl"} ![Asymmetries $A_{TS}$ fog ydenfpfled hadron pairs: $\pi^+\pi^-$ (top row), $K^+\pj^-$ (seconv row), $K^- \pi^+$ (thirc row) and $K^+K ^-$ (last row). The asylmetries are shown as a functuon jd $x$ (left coljm), $z$ (middlt eolumn) and $J_h$ (right column).[]{data-label="Results"}](rchilk_christian.dit2d.eov "fig:"){width="36.00000%" ieight="3.1bm"}![Asymmetries $A_{RS}$ for identigied hadron palrs: $\pm^+\pi^-$ (rop row), $K^+\pi^-$ (second rox), $K^- \pi^+$ (third row) and $K^+K ^-$ (last rkw). The asymmetriew qre svown as x fuvctjoi or $x$ (levt rolum), $z$ (midsle column) qnd $M_h$ (right column).[]{caeq-label="Results"}](achill_shwistian.fig2e.eps "fig:"){width="36.00000%" height="3.1cm"}![Asymmetgies $A_{RS}$ for identified hadeon pairs: $\pi^+\pi^-$ (top roa), $K^+\pi^-$ (secjnd row), $K^- \pi^+$ (third row) and $K^+K ^-$ (last row). The asymmatriea art whown qs a function of $x$ (left colum), $z$ (middle column) ags $K_h$ (right column).[]{dcta-label="Results"}](svhllk_shristian.fig2f.gps "fig:"){cjdfh="36.00000%" height="3.1cm"} ![

fig:"){width="36.00000%" height="3.1cm"}![Asymmetries $A_{RS}$ for identified hadron pairs: row), (second row), \pi^+$ (third row) The are shown as function of $x$ colum), $z$ (middle column) and $M_h$ column).[]{data-label="Results"}](schill_christian.fig2c.eps "fig:"){width="36.00000%" height="3.1cm"} ![Asymmetries $A_{RS}$ for identified hadron pairs: $\pi^+\pi^-$ (top row), $K^+\pi^-$ row), $K^- \pi^+$ (third row) and $K^+K ^-$ (last row). The asymmetries are as function $x$ colum), $z$ (middle column) and $M_h$ (right column).[]{data-label="Results"}](schill_christian.fig2d.eps "fig:"){width="36.00000%" height="3.1cm"}![Asymmetries $A_{RS}$ for identified hadron pairs: $\pi^+\pi^-$ row), $K^+\pi^-$ (second row), $K^- \pi^+$ (third row) $K^+K ^-$ (last row). asymmetries are shown as a of (left colum), (middle and (right column).[]{data-label="Results"}](schill_christian.fig2e.eps "fig:"){width="36.00000%" $A_{RS}$ for identified hadron pairs: $\pi^+\pi^-$ (top row), $K^+\pi^-$ (second row), $K^- \pi^+$ (third row) and $K^+K (last row). are shown a of (left colum), $z$ and $M_h$ (right column).[]{data-label="Results"}](schill_christian.fig2f.eps "fig:"){width="36.00000%" height="3.1cm"}

fig:"){width="36.00000%" height="3.1cm"}![AsymmetriEs $A_{RS}$ for idEntifIed HadRoN paiRs: $\pi^+\Pi^-$ (top row), $K^+\pi^-$ (secONd roW), $K^- \pi^+$ (third row) and $K^+K ^-$ (last rOw). The AsYMmetRIeS are sHown as a FUnCTIon Of $X$ (lEft CoLUm), $Z$ (middLe cOlumn) anD $M_h$ (right coLumN).[]{dAta-label="ResuLTs"}](Schill_chriStiAn.fig2c.eps "fig:"){WidTh="36.00000%" heigHt="3.1Cm"} ![ASYmmetRieS $A_{RS}$ fOr idenTIfied hAdron pairS: $\pI^+\Pi^-$ (top rOW), $K^+\pi^-$ (secOND rOw), $K^- \pI^+$ (third row) and $K^+K ^-$ (lasT RoW). the asymmetries Are shoWn AS a FUNctIon Of $x$ (left colUm), $Z$ (middLE column) ANd $m_H$ (RIghT Column).[]{data-labEl="Results"}](scHIll_ChristIaN.fiG2D.eps "fiG:"){widtH="36.00000%" hEIghT="3.1cm"}![AsymmetrIes $A_{rS}$ for idenTified HAdron paIRs: $\pi^+\pi^-$ (tOp row), $K^+\Pi^-$ (sEcoNd roW), $k^- \pI^+$ (tHirD rOW) anD $k^+K ^-$ (LasT Row). the asymmEtRiEs are ShowN AS A FuncTioN of $x$ (Left cOlum), $z$ (middle coLumN) and $m_H$ (riGht coLumn).[]{dAta-lAbEl="ResUlts"}](scHill_cHrIstian.fig2e.eps "fiG:"){widTh="36.00000%" height="3.1cM"}![AsYmMetRiEs $A_{RS}$ FOr idenTifIed Hadron pAirs: $\pi^+\pI^-$ (Top RoW), $k^+\PI^-$ (sEcond row), $K^- \pi^+$ (third roW) aND $k^+K ^-$ (Last row). THe asymMEtRiES are showN aS a fUnctION of $x$ (lEft cOLuM), $z$ (middle Column) ANd $m_h$ (Right coLuMn).[]{data-LaBel="resUlts"}](sCHill_ChristIan.fig2f.ePs "fig:"){WIdth="36.00000%" height="3.1cm"} ![

fig:"){width="36.00000%" h eight="3.1 cm"}! [As ymm et ries $A_ {RS}$ for iden t ifie d hadron pairs: $\pi^+ \pi^- $( topr ow ), $K ^+\pi^- $ ( s e con dro w), $ K ^- \pi^ +$(thirdrow) and $ K^+ K^-$ (last ro w ). The asymm etr ies are show n a s a fu nc tio n of $ x$(left colum ) , $z$(middle c ol u mn) an d $M_h$( r ig ht c olumn).[]{data-la b el = "Results"}](sc hill_c hr i st i a n.f ig2 c.eps "fig :" ){wid t h="36.0 0 00 0 % " he i ght="3.1cm"} ![Asymmetr i es$A_{RS }$ fo r ident ified h a dro n pairs: $\ pi^+ \pi^-$ (t op row ) , $K^+\ p i^-$ (s econdrow ),$K^- \p i^ +$(t h ird ro w)a nd$K^+K ^- $(l ast r ow). T h e asy mme trie s are shown as a f unc tion of$x$ ( leftcolu m) , $z$ (midd le co lu mn) and $M_h$ ( righ t column) .[] {d ata -l abel= " Result s"} ](s chill_c hristia n .fi g2 d . e ps "fig:"){width="36 .0 0 0 00 %" heigh t="3.1 c m" }! [ Asymmetr ie s $ A_{R S } $ for ide n ti fied had ron pa i rs :$\pi^+\ pi ^-$ (t op ro w), $K^+ \ pi^- $ (sec ond row) , $K^ - \pi^+$ (third row) and $K^+ K ^ - $ ( l astrow ). The asym metr i es a re s h ow n a s a fu nctio no f$ x$ (left colum), $z $(middl e col umn) and $M_h $ (right c o l u mn).[]{d ata- l ab e l="Results"}]( schil l_christia n .fig2e.e ps "f ig:"){wi dth="36.0 0 0 00%" hei ght ="3 .1c m"} ! [ As ymmetries $A_ { R S}$fo r ident ifi ed hadr onpai rs: $\ pi ^+\pi^-$(top row ), $ K^ +\ pi^ -$ (s e cond row ), $K ^- \p i^+$( thirdrow)and$K ^+ K ^- $ (last ro w ) . Th eas ymme tri es areshow n as a func tion of $ x$( left c ol um), $z $ (middle col um n) and $M_ h$ (r ight c o l umn).[]{ data-label="Results"}]( s chill_c hri stian .fig 2f.eps "f ig: "){wid th= " 36.000 00%" h eight =" 3.1 c m "} ! [

fig:"){width="36.00000%" height="3.1cm"}![Asymmetries_$A_{RS}$ for_identified hadron pairs: $\pi^+\pi^-$_(top row),_$K^+\pi^-$_(second row),_$K^-_\pi^+$ (third row)_and $K^+K ^-$_(last row). The asymmetries_are shown as_a_function of $x$ (left colum), $z$ (middle column) and $M_h$ (right column).[]{data-label="Results"}](schill_christian.fig2c.eps "fig:"){width="36.00000%" height="3.1cm"} ![Asymmetries_$A_{RS}$_for identified_hadron_pairs:_$\pi^+\pi^-$ (top row), $K^+\pi^-$ (second_row), $K^- \pi^+$ (third row)_and $K^+K_^-$ (last row). The asymmetries are shown as_a_function of $x$_(left colum), $z$ (middle column) and $M_h$ (right column).[]{data-label="Results"}](schill_christian.fig2d.eps_"fig:"){width="36.00000%" height="3.1cm"}![Asymmetries $A_{RS}$ for identified hadron_pairs: $\pi^+\pi^-$ (top_row),_$K^+\pi^-$_(second row), $K^- \pi^+$_(third row) and $K^+K ^-$ (last_row). The asymmetries are shown as_a function of $x$ (left colum), $z$_(middle column) and $M_h$ (right column).[]{data-label="Results"}](schill_christian.fig2e.eps_"fig:"){width="36.00000%" height="3.1cm"}![Asymmetries $A_{RS}$ for identified_hadron pairs:_$\pi^+\pi^-$ (top row), $K^+\pi^-$ (second_row), $K^- \pi^+$_(third row)_and $K^+K ^-$_(last row). The asymmetries are shown_as a function_of $x$ (left colum), $z$ (middle_column)_and $M_h$ (right_column).[]{data-label="Results"}](schill_christian.fig2f.eps_"fig:"){width="36.00000%"_height="3.1cm"} ![

parameter in the $d_{\rm xy}$-wave state which is stabilized when ${J_{{\rm H}}}$ is very small, ${J_{{\rm H}}}=U/12$. The $d_{\rm xy} \pm {\rm i}d_{\rm x^{2}-y^{2}}$ state is expected below [$T_{\rm c}$ ]{}and both [$a_{\rm 1g}$-Fermi surface ]{}and [$e_{\rm g}$-Fermi surface ]{}are node-less in this case. The exponential behaviors in many quantities are expected unless some accidental situation occurs as in the $p$-wave state. Our calculation does not support such an accidental situation in the $d$-wave symmetry. It should be noticed that in all of the above cases we have shown, the amplitude of order parameter is large on the [$e_{\rm g}$-Fermi surface]{}, while it is small on the [$a_{\rm 1g}$-Fermi surface]{}. This result is expected from the fact that the [$e_{\rm g}$-Fermi surface ]{}is responsible for the pairing instability as discussed in §3.3. This point will be illuminated more clearly in the next section. Reduced Models ============== We have analyzed the possibility of unconventional superconductivity in [${\rm Na_{x}Co_{}O_{2}} \cdot y{\rm H}_{2}{\rm O}$ ]{}on the basis of the three-orbital model. Because calculations for this model need much computational time, a simplified model appropriate for studying the superconductivity is highly desired for a future development in the theoretical field. In this section, we try to find an appropriate model from the comparison to the three-orbital model. We show that the two-orbital model is satisfactory for this purpose, while the single-orbital model is not. The essential origin of the results in §3 will be clarified by these trials. Failure of single-orbital Hubbard model --------------------------------------- Thus far, we have stressed some essential roles of the orbital degeneracy. They are illuminated by showing the failure of single-orbital model. Some authors have already studied single-orbital Hubbard models reproducing the LDA Fermi surface. [@rf:nisikawa; @rf:kuroki] In this paper, we try a single-orbital Hubbard model by keeping only the $\gamma$-band, [*i.e,*]{} the highest-energy eigenstates obtained in eq. (\[eq:

parameter in the $ d_{\rm xy}$-wave state which is stabilized when $ { J_{{\rm H}}}$ is very small, $ { J_{{\rm H}}}=U/12$. The $ d_{\rm xy } \pm { \rm i}d_{\rm x^{2}-y^{2}}$ department of state is expect below [ $ T_{\rm c}$ ] { } and both [ $ a_{\rm 1g}$-Fermi surface ] { } and [ $ e_{\rm g}$-Fermi surface ] { } are node - less in this lawsuit. The exponential behaviors in many quantities are expect unless some accidental position occurs as in the $ p$-wave state of matter. Our calculation does not support such an accidental situation in the $ d$-wave isotropy. It should be noticed that in all of the above cases we have shown, the amplitude of order argument is large on the [ $ e_{\rm g}$-Fermi surface ] { }, while it is humble on the [ $ a_{\rm 1g}$-Fermi surface ] { }. This result is expect from the fact that the [ $ e_{\rm g}$-Fermi surface ] { } is responsible for the pairing imbalance as discussed in § 3.3. This point will be light more clearly in the following section. Reduced Models = = = = = = = = = = = = = = We have analyzed the possibility of unconventional superconductivity in [ $ { \rm Na_{x}Co_{}O_{2 } } \cdot y{\rm H}_{2}{\rm O}$ ] { } on the basis of the three - orbital exemplar. Because calculations for this model need much computational time, a simplified model appropriate for studying the superconductivity is highly desired for a future development in the theoretical field. In this section, we try to find an appropriate model from the comparison to the three - orbital model. We show that the two - orbital model is satisfactory for this purpose, while the individual - orbital model is not. The essential lineage of the result in § 3 will be clarified by these trials. Failure of individual - orbital Hubbard model --------------------------------------- Thus far, we have stressed some essential function of the orbital degeneracy. They are illuminated by showing the failure of single - orbital model. Some generator have already studied single - orbital Hubbard models reproducing the LDA Fermi open.   [ @rf: nisikawa; @rf: kuroki ] In this paper, we try on a single - orbital Hubbard model by keeping only the $ \gamma$-band, [ * i.e, * ] { } the highest - department of energy eigenstates obtained in eq.   (\[eq:

pagameter in the $d_{\rm xy}$-wavt state which is stabilived wheh ${J_{{\rm H}}}$ ks very small, ${J_{{\rm H}}}=U/12$. The $d_{\rm xt} \pm {\em i}d_{\rm x^{2}-y^{2}}$ state is exoected bepow [$T_{\rm x}$ ]{}anv both [$a_{\rm 1g}$-Fermm surfacc ]{}and [$c_{\rm g}$-Yecmi surface ]{}are node-less hn this case. Tve erponential behaviors in many quantityes are edpected unless somt ascidsntal situation occurs as in the $p$-save stete. Our calculayion does not support such an wccidental situatiln in the $d$-qave wymmetry. It sfould be noticed that jn all of the above cases we haxe shpwn, the amkljtkge of order paraieter is larnv on tha [$e_{\rm g}$-Germi surface]{}, ehine ut is small on the [$a_{\rk 1g}$-Fermi surface]{}. Tris resuld ns expected from the daxt thdt tve [$e_{\fn g}$-Werji shrface ]{}is responsibme for the pairing instability af discussed in §3.3. This pjint will be illuminated more clearly it tge next section. Reduced Nodels ============== We have analyzef the posfibility of unconventional superconductivity in [${\rk Na_{x}Ro_{}U_{2}} \cbit y{\ro H}_{2}{\gm O}$ ]{}on the basis of the three-orbital model. Bqdaisv calculations fov this model need kufh somputational time, c sjmplified model aporopriaje for studying the superconductivity is highlt desired fog a duture development in the theuretocal gield. In this section, wz try fo find an wppropriafd model from the cokpdrison to the three-orbital model. We shoc that tfe teo-orbieal model ls sabhsfactory for this purplsa, while thf single-orbital model is not. Thx essential otighn mf the rzsults in §3 will be slarified by tkese tricls. Faiuure of sihgle-orbmtal Hubbard model --------------------------------------- Thus fas, we have strxssed somq eswentual roldr of the orbitsl degenegaey. They aee illuminated by xhodjng the failure of single-orbital koddl. Fole atdhors have anreaay ryudiea singlt-jrnitxl Hibbard models reprodgcinf the LDA Fermi sutfwce. [@rf:nisukawa; @rf:huroki] In thix paper, we try a slngle-mrbmtal Hibbwrd model by keeping only the $\famma$-band, [*i.e,*]{} the highest-qnerny eygenstates obtained in eq. (\[eq:

parameter in the $d_{\rm xy}$-wave state which when H}}}$ is small, ${J_{{\rm H}}}=U/12$. i}d_{\rm state is expected [$T_{\rm c}$ ]{}and [$a_{\rm 1g}$-Fermi surface ]{}and [$e_{\rm g}$-Fermi ]{}are node-less in this case. The exponential behaviors in many quantities are expected some accidental situation occurs as in the $p$-wave state. Our calculation does not such accidental in $d$-wave symmetry. It should be noticed that in all of the above cases we have shown, amplitude of order parameter is large on the g}$-Fermi surface]{}, while it small on the [$a_{\rm 1g}$-Fermi This is expected the that [$e_{\rm g}$-Fermi surface responsible for the pairing instability as discussed in §3.3. This point will be illuminated more clearly in next section. ============== We analyzed possibility unconventional superconductivity in \cdot y{\rm H}_{2}{\rm O}$ ]{}on the three-orbital model. Because calculations for this model need computational time, simplified model appropriate for studying the is highly desired for a future development in theoretical field. In this section, we try to find an appropriate model from the comparison three-orbital model. We show the two-orbital model satisfactory this while single-orbital model not. The essential origin of the results in §3 will be by these trials. Failure of single-orbital Hubbard model --------------------------------------- Thus have some essential roles the orbital degeneracy. They illuminated showing the failure of Some have Hubbard reproducing LDA Fermi surface. [@rf:nisikawa; In this paper, we try single-orbital Hubbard model by the highest-energy eigenstates obtained in eq. (\[eq:

parameter in the $d_{\rm xy}$-wave stAte which is StabiLizEd wHeN ${J_{{\rm h}}}$ is vEry small, ${J_{{\rm H}}}=U/12$. THE $d_{\rm Xy} \pm {\rm i}d_{\rm x^{2}-y^{2}}$ state is expEcted BeLOw [$T_{\rM C}$ ]{}aNd botH [$a_{\rm 1g}$-FeRMi SURfaCe ]{}AnD [$e_{\rM g}$-fErMi surFacE ]{}are nodE-less in thiS caSe. the exponentiAL bEhaviors in ManY quantities aRe eXpecteD uNleSS some AccIdentAl situATion ocCurs as in tHe $P$-Wave stATe. Our caLCUlAtioN does not support suCH aN Accidental situAtion iN tHE $d$-WAVe sYmmEtry. It shouLd Be notICed that IN aLL OF thE Above cases we hAve shown, the AMplItude oF oRdeR ParameTer is LaRGe oN the [$e_{\rm g}$-FerMi suRface]{}, whilE it is sMAll on thE [$A_{\rm 1g}$-FerMi surfAce]{}. thiS resULt Is ExpEcTEd fROm The FAct That the [$e_{\Rm G}$-FErmi sUrfaCE ]{}IS RespOnsIble For thE pairing instaBilIty aS DisCusseD in §3.3. ThIs poInT will Be illuMinatEd More clearly in thE nexT section. REduCeD MoDeLs ============== We hAVe analYzeD thE possibIlity of UNcoNvENTIoNal superconductiviTy IN [${\Rm na_{x}Co_{}O_{2}} \cdOt y{\rm H}_{2}{\RM O}$ ]{}On THe basis oF tHe tHree-ORBital ModeL. beCause calCulatiONs FoR this moDeL need mUcH coMpuTatioNAl tiMe, a simPlified mOdel aPPropriate for stUDying the superCOnDUCtIVity Is hIghly desireD for A FutuRe deVElOpmENt in tHe theOrETiCAl field. In this sectioN, wE try to Find aN appropriate mOdel from thE COMparison To thE ThREe-orbital model. we shoW that the twO-Orbital mOdel iS satisfaCtory for tHIS purpose, WhiLe tHe sIngLE-OrBital model is nOT. the eSsEntial oRigIn of the ResUltS in §3 WilL bE clarifieD by these TrIaLs. faIluRe of sINgle-orbiTaL HuBbArd Model --------------------------------------- tHus far, We havE strEsSeD SomE essentIAl ROLes oF tHe OrbiTal DeGenerAcy. THEy aRe illumInated by sHowINg thE fAiLure of sIngle-orbital mOdEl. Some authOrS haVe alreADY studied Single-orbital Hubbard modELs reproDucIng thE LDA fermi surfAce. [@Rf:nisiKawA; @Rf:kuroKi] In thIs papEr, We tRY A singLE-OrBitAl hubbard modEL By kEepinG oNly tHe $\gamma$-Band, [*i.e,*]{} the highest-enERgy Eigenstates obTaiNed iN EQ. (\[eQ:

parameter in the $d_{\rmxy}$-wavestate wh ich i s st abil ized when ${J_ { {\rm H}}}$ is very small,${J_{ {\ r m H} } }= U/12$ . The $ d _{ \ r m x y} \ pm{\ r mi}d_{ \rm x^{2}- y^{2}}$ st ate i s expected b e lo w [$T_{\rm c} $ ]{}and bot h [ $a_{\r m1g} $ -Ferm i s urfac e ]{}a n d [$e_ {\rm g}$- Fe r mi sur f ace ]{} a r enode -less in this cas e .T he exponential behav io r si n ma nyquantities a re ex p ected u n le s s som e accidental s ituation oc c urs as in t he$ p$-wav e sta te . Ou r calculati on d oes not s upport such an acciden tal si tua tio n in th e$d$ -w a ves ym met r y. It shou ld b e not iced t h a t in al l of theabove cases w e h aves how n, th e amp litu de of o rder p arame te r is large on t he [ $e_{\rm g }$- Fe rmi s urfac e ]{}, w hil e i t is sm all ont he[$ a _ { \r m 1g}$-Fermi surfa ce ] { }. This re sult i s e xp e cted fro mthe fac t thatthe[ $e _{\rm g} $-Ferm i s ur face ]{ }i s resp on sib lefor t h e pa iringinstabil ity a s discussed in§ 3.3. This poi n tw i ll be i llu minated mor e cl e arly int he ne x t sec tion. R ed u ced Models ======== == ==== We ha ve analyzed t he possibi l i t y of unc onve n ti o nal supercondu ctivi ty in [${\ r m Na_{x} Co_{} O_{2}} \ cdot y{\r m H}_{2}{\ rmO}$ ]{ }on t he basis of the t hree -o rbitalmod el. Bec aus e c alc ula ti ons for t his mode lne ed m uch comp u tational t ime ,a s impli f ied mo del a ppro pr ia t e f or stud y in g thesu pe rcon duc ti vityis h i ghl y desir ed for afut u re d ev el opmentin the theore ti cal field. I n t his se c t ion, wetry to find an appropri a te mode l f rom t he c omparison to the t hre e -orbit al mod el. W esho w thatt h etwo -o rbital mod e l is sati sf acto ry forthis purpose, whil e th e single-orbi tal mod e l i s n o t. The e s sen t i al origin of th e resultsin §3 will be c l ari fi ed by t hese tr ials. Failur e of sing le-orbita lHubb a r d m odel ----- -------- --------- - ----- - -- ----- --- Thus f ar, we h ave st r ess ed so me ess en tial r olesof the orb ital degeneracy. They a re ill umina ted by showi ngt hefailure o f si ngle-orbit almod el. S ome autho rs h a ve al r eadystud i ed single - or bit a l H ubbard mode l s rep roduc ing the LD A Fe rmi surface. [@rf : nisikawa; @rf: kuro k i ] I n t h is p ap er, we try a s ing le - o rbital H ub bard modelby keepi ng onlythe $\ gamma$ -band,[ * i. e ,*]{}thehig hest-ener gyei g enstate sob t ainedin e q.  (\[eq :

parameter_in the_$d_{\rm xy}$-wave state which_is stabilized_when_${J_{{\rm H}}}$_is_very small, ${J_{{\rm_H}}}=U/12$. The $d_{\rm_xy} \pm {\rm i}d_{\rm_x^{2}-y^{2}}$ state is_expected_below [$T_{\rm c}$ ]{}and both [$a_{\rm 1g}$-Fermi surface ]{}and [$e_{\rm g}$-Fermi surface ]{}are node-less_in_this case._The_exponential_behaviors in many quantities are_expected unless some accidental situation_occurs as_in the $p$-wave state. Our calculation does not_support_such an accidental_situation in the $d$-wave symmetry. It should be noticed that_in all of the above cases_we have shown,_the_amplitude_of order parameter is_large on the [$e_{\rm g}$-Fermi surface]{},_while it is small on the_[$a_{\rm 1g}$-Fermi surface]{}. This result is expected_from the fact that the [$e_{\rm_g}$-Fermi surface ]{}is responsible for_the pairing_instability as discussed in §3.3._This point will_be illuminated_more clearly in_the next section. Reduced Models ============== We have analyzed_the possibility of_unconventional superconductivity in [${\rm Na_{x}Co_{}O_{2}} \cdot_y{\rm_H}_{2}{\rm O}$ ]{}on_the_basis_of the_three-orbital model. Because_calculations_for this_model_need much computational time, a simplified_model_appropriate for studying the superconductivity is highly_desired for a future_development_in the theoretical field._In this section, we try_to find an appropriate model from_the comparison_to the_three-orbital model. We show that the two-orbital model is satisfactory for_this purpose, while the single-orbital model_is not. The essential_origin of_the_results in §3_will_be clarified_by these trials. Failure of single-orbital Hubbard model --------------------------------------- Thus_far, we_have stressed some essential roles of_the orbital degeneracy. They_are_illuminated by showing the failure of_single-orbital model. Some authors have already_studied single-orbital Hubbard models reproducing_the_LDA_Fermi surface. [@rf:nisikawa; @rf:kuroki] In this_paper, we try a single-orbital Hubbard_model by keeping_only the $\gamma$-band, [*i.e,*]{} the highest-energy eigenstates_obtained_in eq. (\[eq:

halo dark matter with this captured dark matter. Once captured, it is expected to quickly thermalise with the ordinary matter within the Earth via the kinetic mixing interactions, to a temperature $T_E \sim 5000$ K (0.4 eV). If $m_{p_d}\gg $ MeV, this is much cooler than the halo temperature and the dark protons and dark electrons can potentially combine into neutral dark atoms. The dark sphere will be largely neutral (ionised) if $T_E \ll I$ ($T_E \gg I$), where $I$ is the dark atomic binding energy given already in Eq. (\[isis\]). This motivates two limiting cases: a neutral “Moon-like” case in which the dark sphere largely absorbs the dark plasma wind, and; an ionised “Venus-like” case in which the dark sphere largely deflects the dark plasma wind by way of a current-carrying sheet at the “dark ionopause” (located where the plasma wind and “dark ionosphere” pressures equilibrate). In addition to its ionisation state, the other defining feature of the dark sphere is its effective size. Let us define a parameter, $R_{DM}$, which corresponds to the dark plasma wind stopping radius for the Moon-like case, and the dark ionopause radius for the Venus-like case. We will now attempt to estimate $R_{DM}$ in terms of the fundamental plasma parameters. If the dark sphere is Moon-like, then the (relatively stationary) dark protons accumulate at the “geometric” rate $$\begin{aligned} \frac{dN_{p_d}}{dt} \approx \pi R_{DM}^2 v_{rot} n_{p_d} \. \label{1z}\end{aligned}$$ Dark electrons will be captured at a similar rate: $dN_{e_d}/dt = Z'dN_{p_d}/dt$ given the expected approximate $U(1)'$ charge neutrality of the Earth.[^4] This represents an upper bound for the accumulation rate in the Venus-like case, though it might still be a useful estimate so long as a significant fraction ($\gtrsim 1$%) of the halo wind is stopped within the Earth. Loss rates due to mechanisms such as thermal escape and dark atmospheric stripping are difficult to evaluate. Naturally, any estimate of the total amount of dark matter captured within the Earth is uncertain. Fortunately, it

halo dark matter with this captured dark matter. Once get, it is expect to quickly thermalise with the ordinary matter within the Earth via the kinetic desegregate interactions, to a temperature $ T_E \sim 5000 $   K (0.4   eV). If $ m_{p_d}\gg $   MeV, this is much cool than the ring temperature and the dark proton and blue electrons can potentially aggregate into neutral dark atom. The blue sphere will be largely neutral (ionised) if $ T_E \ll I$ ($ T_E \gg I$), where $ I$ is the dark atomic bind energy given already in Eq.   (\[isis\ ]). This motivates two specify cases: a neutral “ Moon - alike ” case in which the dark sphere largely absorbs the dark plasma fart, and; an ionised “ Venus - like ” case in which the dark sphere largely deflects the dark plasma wind by way of a current - carrying sheet at the “ dark ionopause ” (located where the plasma wind and “ dark ionosphere ” pressures balance). In accession to its ionisation department of state, the other defining feature of the dark sector is its effective size. Let us define a parameter, $ R_{DM}$, which corresponds to the blue plasma wind stopping radius for the Moon - like case, and the dark ionopause spoke for the Venus - like case. We will now attempt to estimate $ R_{DM}$ in terms of the fundamental plasma parameter. If the dark sphere is Moon - like, then the (relatively stationary) dark protons accumulate at the “ geometric ” pace $ $ \begin{aligned } \frac{dN_{p_d}}{dt } \approx \pi R_{DM}^2 v_{rot } n_{p_d } \. \label{1z}\end{aligned}$$ Dark electrons will be captured at a similar rate: $ dN_{e_d}/dt = Z'dN_{p_d}/dt$ given the expected approximate $ U(1)'$ charge neutrality of the Earth.[^4 ] This represents an upper bind for the accumulation pace in the Venus - like case, though it might still be a useful estimate thus long as a significant fraction ($ \gtrsim 1$%) of the aura wind is intercept within the Earth. Loss rates due to mechanisms such as thermal escape and dark atmospheric denudation are difficult to evaluate. Naturally, any estimate of the full amount of dark matter captured within the Earth is uncertain. Fortunately, it

hapo dark matter with this captured dark matter. Mnce czptured, kt is expected to quickly thxrmaoise qith the ordinary mattdr within the Earrh vma the kinetic mmsing inbzractjlns, co a temperature $T_E \sim 5000$ K (0.4 aV). If $m_{p_d}\gg $ MeE, ghns is much cooler than the halo tempqrature ajd the dark prjtonx and dark electrons can potentially cojbine iito neutral darl atoms. The dark sphere wipl bf largely neutral (lonised) if $J_S \lj I$ ($T_E \gg I$), wfere $I$ is uhz dark atomjc binding energy given already in Ea. (\[isis\]). This mitigdtes two linitind cases: a nemnral “Momn-like” vase in which bhe derk wphere largely absorbv the dark plasma rind, and; dn ionised “Venus-likw” xase hn wvich rhe dadk slhere parjely deflecfs the dark plasma wind by way os a current-cardying fhqet at the “dark ionopause” (located where ths plasma wind and “dark uonosphere” pressures gquilibratq). In addition to its ionisation state, the other dafiniig fecbmre ud hhe dark sphere is its effective size. Let us qsfonv a parameter, $R_{DM}$, which corrrsoomqs to the dary plasma sind stopping radiks for jhe Moin-like cafe, amd the dark ionopause radiuw for the Veuus-oike case. We will uow attempt co estomate $R_{DM}$ in terms of the fuudamenfal plasma oarametera. If the dark sphdre iv Moon-like, then the (relatidely statmonarv) dark pfotoms acctmulate at the “naometric” rate $$\begij{aligued} \frdc{dN_{p_d}}{dt} \aoprox \pi R_{DM}^2 v_{rot} n_{p_d} \. \label{1z}\end{ekigned}$$ Dark glewtrmns will be csptured at a fimilar rate: $dU_{e_d}/dt = Z'bN_{p_d}/dt$ given the expectxd approximaee $U(1)'$ charge naotrality of tie Earth.[^4] Ehis repeesents xn upper bound for the cecumulatiin rate in the Venms-likg dase, though it nigyt still be a uxefjl qsnimete sj long as a shgniwicxmt frxction ($\gtrslm 1$%) of yhe halo wind is sto[ped within the Earth. Kofs rates due to iechanisms suvh as thermal escake and dack atmpspreric stripping are difficult fo evaluahe. Katurally, any estlmatg of the tocal amount of dark matter captured withii the Earth is uncertaib. Fortunately, it

halo dark matter with this captured dark captured, is expected quickly thermalise with Earth the kinetic mixing to a temperature \sim 5000$ K (0.4 eV). If $ MeV, this is much cooler than the halo temperature and the dark and dark electrons can potentially combine into neutral dark atoms. The dark sphere be neutral if \ll I$ ($T_E \gg I$), where $I$ is the dark atomic binding energy given already in (\[isis\]). This motivates two limiting cases: a neutral case in which the sphere largely absorbs the dark wind, an ionised case which dark sphere largely the dark plasma wind by way of a current-carrying sheet at the “dark ionopause” (located where the wind and pressures equilibrate). addition its state, the other of the dark sphere is its us define a parameter, $R_{DM}$, which corresponds to dark plasma stopping radius for the Moon-like case, the dark ionopause radius for the Venus-like case. will now attempt to estimate $R_{DM}$ in terms of the fundamental plasma parameters. If the is Moon-like, then the stationary) dark protons at “geometric” $$\begin{aligned} \approx \pi v_{rot} n_{p_d} \. \label{1z}\end{aligned}$$ Dark electrons will be captured at a rate: $dN_{e_d}/dt = Z'dN_{p_d}/dt$ given the expected approximate $U(1)'$ charge the This represents an bound for the accumulation in Venus-like case, though it be useful as significant ($\gtrsim 1$%) of the wind is stopped within the Loss rates due to and dark atmospheric stripping are difficult to evaluate. any estimate of the total amount of matter captured within the Earth is uncertain. Fortunately, it

halo dark matter with this capTured dark mAtter. oncE caPtUred, It is Expected to quicKLy thErmalise with the ordinarY mattEr WIthiN ThE EartH via the KInETIc mIxInG inTeRAcTions, To a TemperaTure $T_E \sim 5000$ K (0.4 EV). IF $m_{P_d}\gg $ MeV, this iS MuCh cooler thAn tHe halo temperAtuRe and tHe DarK ProtoNs aNd darK electROns can PotentialLy COmbine INto neutRAL dArk aToms. The dark sphere WIlL Be largely neutrAl (ioniSeD) If $t_e \Ll I$ ($t_E \gG I$), where $I$ is ThE dark ATomic biNDiNG ENerGY given already In Eq. (\[isis\]). ThiS MotIvates TwO liMIting cAses: a NeUTraL “Moon-like” caSe in Which the dArk sphERe largeLY absorbS the daRk pLasMa wiND, aNd; An iOnISed “vEnUs-lIKe” cAse in whiCh ThE dark SpheRE LARgelY deFlecTs the Dark plasma winD by Way oF A cuRrent-CarryIng sHeEt at tHe “dark IonopAuSe” (located where tHe plAsma wind aNd “dArK ioNoSpherE” PressuRes EquIlibratE). In addiTIon To ITS IoNisation state, the otHeR DEfIning feaTure of THe DaRK sphere iS iTs eFfecTIVe sizE. Let US dEfine a paRameteR, $r_{Dm}$, wHich corReSponds To The DarK plasMA winD stoppIng radiuS for tHE Moon-like case, aND the dark ionopAUsE RAdIUs foR thE Venus-like cAse. WE Will Now aTTeMpt TO estiMate $R_{dM}$ IN tERms of the fundamental PlAsma paRametErs. If the dark sPhere is MooN-LIKe, then thE (relATiVEly stationary) dArk prOtons accumULate at thE “geomEtric” ratE $$\begin{aliGNEd} \frac{dN_{P_d}}{dT} \apProX \pi r_{dm}^2 v_{Rot} n_{p_d} \. \label{1z}\eND{AligNeD}$$ Dark elEctRons wilL be CapTurEd aT a Similar raTe: $dN_{e_d}/dt = z'dn_{p_D}/dT$ gIveN the eXPected apPrOxiMaTe $U(1)'$ ChargE NeutraLity oF the eaRtH.[^4] thiS represENtS AN uppEr BoUnd fOr tHe AccumUlatIOn rAte in thE Venus-likE caSE, thoUgH iT might sTill be a useful EsTimate so loNg As a SignifICAnt fractIon ($\gtrsim 1$%) of the halo wind iS Stopped WitHin thE EarTh. Loss ratEs dUe to meChaNIsms suCh as thErmal EsCapE ANd darK ATmOspHeRic strippiNG Are DiffiCuLt to EvaluatE. Naturally, any estimATe oF the total amouNt oF darK MAtTer CApTUreD wIThiN THe Earth is uncertAin. FortunaTeLY, iT

halo dark matter with thi s captured dark ma tte r. Onc e ca ptured, it ise xpec ted to quickly thermal ise w it h the or dinar y matte r w i t hin t he Ea rt h v ia th e k ineticmixing int era ct ions, to a t e mp erature $T _E\sim 5000$ K (0 .4 eV) .If$ m_{p_ d}\ gg $MeV, t h is ismuch cool er than t h e halot e mp erat ure and the darkp ro t ons and dark e lectro ns ca n pot ent ially comb in e int o neutra l d a r k at o ms. The darksphere will belargel yneu t ral (i onise d) if$T_E \ll I$ ($T _E \gg I$ ), whe r e $I$ i s the da rk ato mic bi ndin g e ne rgy g i ven al rea d y i n Eq. (\ [i si s\]). Thi s m o tiva tes two limi ting cases: a ne utra l “M oon-l ike”case i n whi ch the dark s phere largely a bsor bs the da rkpl asm awind, and; a n i oni sed “Ve nus-lik e ” c as e i nwhich the dark sph er e la rgely de flects th ed ark plas ma wi nd b y way o f ac ur rent-car ryings he et at the “ dark i on opa use ” (lo c ated where the pla sma w i nd and “dark i o nosphere” pre s su r e se quil ibr ate). In ad diti o n to its io nis a tionstate ,t he other defining feat ur e of t he da rk sphere isits effect i v e size. L et u s d e fine a paramet er, $ R_{DM}$, w h ich corr espon ds to th e dark pl a s ma windsto ppi ngrad i u sfor the Moon- l i ke c as e, andthe dark i ono pau serad iu s for the Venus-l ik eca se . W e wil l now att em ptto es timat e $R_{D M}$ i n te rm so f t he fund a me n t al p la sm a pa ram et ers.If t h e d ark sph ere is Mo on- l ike, t he n the ( relatively st at ionary) da rk pr otonsa c cumulate at the “geometric” rat e $$\beg in{ align ed}\frac{dN_ {p_ d}}{dt } \ a pprox\pi R_ {DM}^ 2v_{ r o t} n_ { p _d } \ . \label{1z } \ end {alig ne d}$$ Dark e lectrons will be c a ptu red at a simi lar rat e : $ dN_ { e_ d }/d t= Z' d N _{p_d}/dt$ give n the expe ct e dapproximat e $U (1 )'$ cha rge neu trali t y of th e Earth.[ ^4] Thisre pres e n tsan upper b ound for the accu m ulati o nrateinthe Ve nu s-l ike c ase, t h oug h itmightst ill be a us ef ul estim ate so long as a signif icantfract ion ($\gtrsi m 1 $ %)of the ha lo w ind is sto ppe d w ithin th e Eart h. L o ss ra t es du e to mechanism s s uch a sthermal esc a p e an d dar k a t mosphe ricstripping are dif f icult to evalu ate. N atu ral l y, a ny estimate of t heto t a l amount o f dark matt er captu re d with in the Earth is unc e r ta i n. For tuna tel y, it

halo_dark matter_with this captured dark_matter. Once_captured,_it is_expected_to quickly thermalise_with the ordinary_matter within the Earth_via the kinetic_mixing_interactions, to a temperature $T_E \sim 5000$ K (0.4 eV). If $m_{p_d}\gg $ MeV, this is much_cooler_than the_halo_temperature_and the dark protons and_dark electrons can potentially combine_into neutral_dark atoms. The dark sphere will be largely_neutral_(ionised) if $T_E_\ll I$ ($T_E \gg I$), where $I$ is the_dark atomic binding energy given already_in Eq. (\[isis\]). This_motivates_two_limiting cases: a neutral_“Moon-like” case in which the dark_sphere largely absorbs the dark plasma_wind, and; an ionised “Venus-like” case in_which the dark sphere largely deflects_the dark plasma wind by_way of_a current-carrying sheet at the_“dark ionopause” (located_where the_plasma wind and_“dark ionosphere” pressures equilibrate). In addition_to its ionisation_state, the other defining feature of_the_dark sphere is_its_effective_size. Let_us define a_parameter,_$R_{DM}$, which_corresponds_to the dark plasma wind stopping_radius_for the Moon-like case, and the dark_ionopause radius for the_Venus-like_case. We will now_attempt to estimate $R_{DM}$ in_terms of the fundamental plasma parameters._If the_dark sphere_is Moon-like, then the (relatively stationary) dark protons accumulate at the_“geometric” rate $$\begin{aligned} \frac{dN_{p_d}}{dt} \approx \pi R_{DM}^2_v_{rot} n_{p_d} \. \label{1z}\end{aligned}$$ Dark_electrons will_be_captured at a_similar_rate: $dN_{e_d}/dt_= Z'dN_{p_d}/dt$ given the expected approximate $U(1)'$_charge neutrality_of the Earth.[^4] This represents an_upper bound for the_accumulation_rate in the Venus-like case, though_it might still be a useful_estimate so long as a_significant_fraction_($\gtrsim 1$%) of the halo_wind is stopped within the Earth._Loss rates due_to mechanisms such as thermal escape and_dark_atmospheric stripping are difficult to evaluate._Naturally,_any estimate of the total amount_of_dark_matter captured within the Earth_is uncertain. Fortunately, it

\leq r. \end{aligned}$$ Those functions satisfy a Poincaré type inequality of the form $$\begin{aligned} \label{Poincare} {\left\lVertu\right\rVert}_{L^2} \leq C({\left\lVert\nabla u\right\rVert}_{L^2} + 1), \end{aligned}$$ where $C$ only depends on $r$ and $\Omega$. The proof consists of two steps. First we will show that the inequality (\[InequalityM\]) implies that $M$ cannot contain arbitrarily large, constant functions and second we prove that a failure of the Poincaré inequality (\[Poincare\]) means that $M$ contains any large, constant function hence the assertion follows.  \  \ **Step 1.** Let $\xi\in\mathbb{R}$ be a constant function in $M$. We will show that there is some fixed $C>0$ depending only on $r,{\left\lVertf\right\rVert}_{H^1(\Omega)'}$ and $|\partial\Omega|$ such that $$|\xi|\leq C.$$ Using a scaled version of Young’s inequality with $\varepsilon |\Omega|^{1/2} \leq |\partial\Omega|/2$ we compute $$\begin{aligned} r \geq \int_{\partial\Omega}\xi^2\,ds - f(\xi) &\geq |\xi|^2|\partial\Omega| - {\left\lVertf\right\rVert}_{H^1(\Omega)'}{\left\lVert\xi\right\rVert}_{H^1(\Omega)} \\ &= |\xi|^2|\partial\Omega| - {\left\lVertf\right\rVert}_{H^1(\Omega)'}|\Omega|^{1/2}|\xi| \\ &\geq |\xi|^2|\partial\Omega| - C(\varepsilon){\left\lVertf\right\rVert}^2_{H^1(\Omega)'} -\varepsilon|\Omega|\cdot|\xi|^2 \\ &\geq \frac12|\partial\Omega|\,|\xi|^2 - C(\varepsilon){\left\lVertf\right\rVert}^2_{H^

\leq r. \end{aligned}$$ Those functions satisfy a Poincaré type inequality of the form $ $ \begin{aligned } \label{Poincare } { \left\lVertu\right\rVert}_{L^2 } \leq C({\left\lVert\nabla u\right\rVert}_{L^2 } + 1), \end{aligned}$$ where $ C$ merely depend on $ r$ and $ \Omega$. The proof consists of two steps. foremost we will show that the inequality (\[InequalityM\ ]) implies that $ M$ cannot contain randomly big, constant function and second we prove that a failure of the Poincaré inequality (\[Poincare\ ]) intend that $ M$ contains any large, ceaseless affair hence the assertion follows.   \   \ * * Step 1. * * Let $ \xi\in\mathbb{R}$ be a changeless function in $ M$. We will show that there is some cook $ C>0 $ depending only on $ r,{\left\lVertf\right\rVert}_{H^1(\Omega)'}$ and $ |\partial\Omega|$ such that $ $ |\xi|\leq C.$$ Using a scaled adaptation of Young ’s inequality with $ \varepsilon |\Omega|^{1/2 } \leq |\partial\Omega|/2 $ we compute $ $ \begin{aligned } r \geq \int_{\partial\Omega}\xi^2\,ds - f(\xi) & \geq |\xi|^2|\partial\Omega| - { \left\lVertf\right\rVert}_{H^1(\Omega)'}{\left\lVert\xi\right\rVert}_{H^1(\Omega) } \\ & = |\xi|^2|\partial\Omega| - { \left\lVertf\right\rVert}_{H^1(\Omega)'}|\Omega|^{1/2}|\xi| \\ & \geq |\xi|^2|\partial\Omega| - C(\varepsilon){\left\lVertf\right\rVert}^2_{H^1(\Omega)' } -\varepsilon|\Omega|\cdot|\xi|^2 \\ & \geq \frac12|\partial\Omega|\,|\xi|^2 - C(\varepsilon){\left\lVertf\right\rVert}^2_{H^

\leq r. \end{aligned}$$ Uhose functions scrisfy e Poinczré type knequality of the form $$\begin{eligbed} \lavel{Poincare} {\lefg\lVertu\rihht\rVert}_{O^2} \leq C({\lenc\lVerf\kabla n\right\rVert}_{L^2} + 1), \end{aligted}$$ where $C$ onny dzpends on $r$ and $\Omega$. The proof consifts of ywl steps. First re wplj shkw that the inequality (\[InequalityM\]) impliev that $M$ cannpt contain arbitrarily larhe, clnstant functions wnd second qe pwive that a fxilure of uhz Poincaré ihequality (\[Poincare\]) means that $M$ contcins any latyw, cltstant funcvion hvnce the assevnion fonlows.  \  \ **Syep 1.** Let $\xi\in\msthub{R}$ ve a constant functioi in $M$. We will show jhat there ia some fixed $C>0$ depebding onlf on $e,{\lewt\lBectf\dight\rGerv}_{H^1(\Omega)'}$ and $|\partial\Ometa|$ such that $$|\xi|\leq V.$$ Twing a scaled versijn of Young’s inequality with $\varepsilon |\Okegz|^{1/2} \leq |\partial\Omega|/2$ we cimpute $$\begin{aligned} r \geq \int_{\partial\Omega}\xi^2\,ds - f(\xi) &\geq |\xi|^2|\partiwm\Okena| - {\lrfh\lFgrtf\right\rVert}_{F^1(\Omega)'}{\leff\lVert\xi\right\rVert}_{J^1(\Omega)} \\ &= |\xi|^2|\partial\Onega| - {\leyt\lVertf\righc\rVert}_{N^1(\Omegs)'}|\Omega|^{1/2}|\xi| \\ &\geq |\sk|^2|\partial\Omega| - C(\varepsilog){\left\lVervf\rigkt\rVert}^2_{H^1(\Umegs)'} -\varfpsilon|\Omega|\cdot|\xi|^2 \\ &\geq \frac12|\partial\Omege|\,|\ei|^2 - E(\varepxilon){\left\lVeref\right\rVert}^2_{H^

\leq r. \end{aligned}$$ Those functions satisfy a inequality the form \label{Poincare} {\left\lVertu\right\rVert}_{L^2} \leq where only depends on and $\Omega$. The consists of two steps. First we show that the inequality (\[InequalityM\]) implies that $M$ cannot contain arbitrarily large, constant and second we prove that a failure of the Poincaré inequality (\[Poincare\]) means $M$ any constant hence the assertion follows. \ \ **Step 1.** Let $\xi\in\mathbb{R}$ be a constant function in $M$. will show that there is some fixed $C>0$ only on $r,{\left\lVertf\right\rVert}_{H^1(\Omega)'}$ and such that $$|\xi|\leq C.$$ Using scaled of Young’s with |\Omega|^{1/2} |\partial\Omega|/2$ we compute r \geq \int_{\partial\Omega}\xi^2\,ds - f(\xi) &\geq |\xi|^2|\partial\Omega| - {\left\lVertf\right\rVert}_{H^1(\Omega)'}{\left\lVert\xi\right\rVert}_{H^1(\Omega)} \\ &= |\xi|^2|\partial\Omega| - {\left\lVertf\right\rVert}_{H^1(\Omega)'}|\Omega|^{1/2}|\xi| \\ &\geq |\xi|^2|\partial\Omega| C(\varepsilon){\left\lVertf\right\rVert}^2_{H^1(\Omega)'} -\varepsilon|\Omega|\cdot|\xi|^2 \frac12|\partial\Omega|\,|\xi|^2 -

\leq r. \end{aligned}$$ Those functiOns satisfy A PoinCarÉ tyPe IneqUaliTy of the form $$\begIN{aliGned} \label{Poincare} {\left\lvertu\RiGHt\rVERt}_{l^2} \leq C({\Left\lVeRT\nABLa u\RiGhT\rVErT}_{l^2} + 1), \eNd{aliGneD}$$ where $C$ Only dependS on $R$ aNd $\Omega$. The prOOf Consists of Two Steps. First we WilL show tHaT thE InequAliTy (\[IneQualitYm\]) impliEs that $M$ caNnOT contaIN arbitrARIlY larGe, constant functioNS aND second we prove That a fAiLUrE OF thE PoIncaré ineqUaLity (\[POIncare\]) mEAnS THAt $M$ COntains any larGe, constant fUNctIon henCe The ASsertiOn folLoWS.  \  \ **StEp 1.** Let $\xi\in\maThbb{r}$ be a constAnt funCTion in $M$. wE will shOw that TheRe iS somE FiXeD $C>0$ dEpENdiNG oNly ON $r,{\lEft\lVertF\rIgHt\rVeRt}_{H^1(\OMEGA)'}$ And $|\pArtIal\OMega|$ sUch that $$|\xi|\leq C.$$ usiNg a sCAleD versIon of younG’s InequAlity wIth $\vaRePsilon |\Omega|^{1/2} \leq |\pArtiAl\Omega|/2$ we ComPuTe $$\bEgIn{aliGNed} r \geQ \inT_{\paRtial\OmEga}\xi^2\,ds - F(\Xi) &\gEq |\XI|^2|\PArTial\Omega| - {\left\lVertF\rIGHt\RVert}_{H^1(\OmEga)'}{\lefT\LVErT\Xi\right\rveRt}_{H^1(\omegA)} \\ &= |\XI|^2|\partIal\OMEgA| - {\left\lVeRtf\rigHT\rveRt}_{H^1(\OmegA)'}|\OMega|^{1/2}|\xi| \\ &\GeQ |\xi|^2|\ParTial\OMEga| - C(\VarepsIlon){\left\LVertF\Right\rVert}^2_{H^1(\OmeGA)'} -\varepsilon|\OmEGa|\CDOt|\XI|^2 \\ &\geq \FraC12|\partial\OmeGa|\,|\xi|^2 - c(\VarePsilON){\lEft\Lvertf\Right\RVERt}^2_{h^

\leq r. \end{a ligned}$$Those fu nct io ns s atis fy a Poincarét ypeinequality of the form $$\b eg i n{al i gn ed} \ label{P o in c a re} { \l eft\l Ver tu\righ t\rVert}_{ L^2 } \leq C({\ lef t\lVert\nabl a u \right \r Ver t }_{L^ 2}+ 1), \ e nd{ali gned}$$ w he r e $C$o nly dep e n ds on$r$ and $\Omega$. T h e proof consis ts oftw o s t e ps. Fi rst we wil lshowt hat the in e q u ali t y (\[Inequali tyM\]) impl i esthat $ M$ ca n not co ntain a r bit rarily larg e, c onstant f unctio n s and s e cond we prove th ata fa i lu re of t h e P o in car é in equality ( \[ Poinc are\ ] ) m eans th at $ M$ co ntains any la rge , co n sta nt fu nctio n he nc e the asser tionfo llows.  \  \ ** Step 1.** Let $\ xi \in \m athbb { R}$ be acon stant f unction in$M $ . We will show that th er e is some fi xed $C > 0$ d e pendingon lyon $ r , {\lef t\lV e rt f\right\ rVert} _ {H ^1 (\Omega )' }$ and $ |\p art ial\O m ega| $ such that $$ |\xi| \ leq C.$$ Using a scaled vers i on o fY oung ’sinequalitywith $\va reps i lo n | \ Omega |^{1/ 2} \l e q |\partial\Omega|/ 2$ we co mpute $$\begin{ali gned} r \geq \int_{\pa r tial\Ome ga}\x i^2\,ds - f( \xi ) &\g e q |\x i|^2|\p art ial \Om ega | - {\ left\ l Vertf\ri gh t\r Ve rt} _{H^1 ( \Omega )'}{\ left \l Ve r t\x i\right \ rV e r t}_{ H^ 1( \Ome ga) } \ \ &= |\xi|^ 2|\partial\Om eg a| - { \left\lVertf\right\rVer t }_{H^1( \Om ega)' }|\O mega|^{1/ 2}| \xi| \\ & \ geq |\xi|^2| \ p art ial\O me ga| - C(\ varepsilon){\ lef t\lV e r tf \ri g ht \ rVe rt } ^2_ { H ^1(\Omega)'} -\ v ar epsilon|\O m ega |\ cdot|\x i|^2 \\ &\geq \frac12|\ partial\ Omega|\,| \ xi|^2 - C(\var e psi lon){ \left\ lV ertf\r ight\ rV ert}^2_{ H^

\leq _ _ _r. __ \end{aligned}$$_Those_functions satisfy a_Poincaré type inequality_of the form $$\begin{aligned} \label{Poincare} _ __ {\left\lVertu\right\rVert}_{L^2} \leq __ ___C({\left\lVert\nabla u\right\rVert}_{L^2} + 1), _ \end{aligned}$$ where $C$_only depends_on $r$ and $\Omega$. The proof consists of two_steps._First we will_show that the inequality (\[InequalityM\]) implies that $M$ cannot_contain arbitrarily large, constant functions and_second we prove_that_a_failure of the Poincaré_inequality (\[Poincare\]) means that $M$ contains_any large, constant function hence the_assertion follows.  \  \ **Step 1.** Let $\xi\in\mathbb{R}$ be_a constant function in $M$. We_will show that there is_some fixed_$C>0$ depending only on $r,{\left\lVertf\right\rVert}_{H^1(\Omega)'}$_and $|\partial\Omega|$ such_that $$|\xi|\leq_C.$$ Using a_scaled version of Young’s inequality with_$\varepsilon |\Omega|^{1/2} \leq_|\partial\Omega|/2$ we compute $$\begin{aligned} __ ___ _ r __ __ _\geq _ _ _\int_{\partial\Omega}\xi^2\,ds _ _ _ - _ _ _ f(\xi) _ &\geq _ _ __ __|\xi|^2|\partial\Omega| _ _ _ - _ __ {\left\lVertf\right\rVert}_{H^1(\Omega)'}{\left\lVert\xi\right\rVert}_{H^1(\Omega)} _ _ \\ ___ _ &= _ _ |\xi|^2|\partial\Omega| __ __ - ___ _ {\left\lVertf\right\rVert}_{H^1(\Omega)'}|\Omega|^{1/2}|\xi| _ \\ _ _ ___ &\geq _ _ |\xi|^2|\partial\Omega| _ _- _ C(\varepsilon){\left\lVertf\right\rVert}^2_{H^1(\Omega)'} _ -\varepsilon|\Omega|\cdot|\xi|^2 _ _ \\ __ ____&\geq _ _ _ _ \frac12|\partial\Omega|\,|\xi|^2 _ _ __ - C(\varepsilon){\left\lVertf\right\rVert}^2_{H^

_1)\oplus_{comp}\Omega^k(X_2)\to\Omega^k(X_1\cup_f X_2)$$ given by setting, for every plot $p:U\to X_1\cup_f X_2$ defined on a connected $U$, $$\mathcal{L}^k(\omega_1\oplus\omega_2)(p)=\left\{\begin{array}{ll} \omega_1(p_1) & \mbox{if }p=\hat{i}_1\circ p_1\mbox{ for some plot }p_1\mbox{ of }X_1,\\ \omega_2(p_2) & \mbox{if }p=i_2\circ p_2\mbox{ for some plot }p_2\mbox{ of }X_2. \end{array}\right.$$ For any two diffeological spaces $X_1$ and $X_2$ and for every smooth map $f:X_1\supseteq Y\to X_2$ the map $\mathcal{L}^k$ is well-defined. We need to show that $\mathcal{L}^k(\omega_1\oplus\omega_2)(p)$ does not depend on the choice of the lift of $p$ to a plot $p_i$ of $X_i$, and that the assignment $p\mapsto\mathcal{L}^k(\omega_1\oplus\omega_2)(p)$ satisfies the smooth compatibility condition. The former of these claims is obvious if $p$ lifts to a plot of $X_2$; indeed, since $i_2$ is injective, such a lift is unique. Let $p_1$ and $p_1'$ be two lifts of $p$ to some plots of $X_1$. Then they are obviously $f$-equivalent. Since $\omega_1$ is $f$-invariant by assumption, we have that $\omega_1(p_1)=\omega_1(p_1')$, which implies that $\mathcal{L}^k(\omega_1\oplus\omega_2)(p)$ is well-defined. Let us now show that $\mathcal{L}^k(\omega_1\oplus\omega_2)$ satisfies a smooth compatibility condition. Let $h:U'\to U$ be an ordinary smooth map

_ 1)\oplus_{comp}\Omega^k(X_2)\to\Omega^k(X_1\cup_f X_2)$$ given by setting, for every plot $ p: U\to X_1\cup_f X_2 $ defined on a affiliated $ U$, $ $ \mathcal{L}^k(\omega_1\oplus\omega_2)(p)=\left\{\begin{array}{ll } \omega_1(p_1) & \mbox{if } p=\hat{i}_1\circ p_1\mbox { for some plot } p_1\mbox { of } X_1,\\ \omega_2(p_2) & \mbox{if } p = i_2\circ p_2\mbox { for some plot } p_2\mbox { of } X_2. \end{array}\right.$$ For any two diffeological space $ X_1 $ and $ X_2 $ and for every smooth function $ f: X_1\supseteq Y\to X_2 $ the function $ \mathcal{L}^k$ is well - defined. We need to express that $ \mathcal{L}^k(\omega_1\oplus\omega_2)(p)$ does not count on the choice of the lift of $ p$ to a plot $ p_i$ of $ X_i$, and that the assignment $ p\mapsto\mathcal{L}^k(\omega_1\oplus\omega_2)(p)$ satisfies the smooth compatibility condition. The early of these claims is obvious if $ p$ lifts to a plot of $ X_2 $; indeed, since $ i_2 $ is injective, such a airlift is alone. Let $ p_1 $ and $ p_1'$ be two lifts of $ p$ to some plots of $ X_1$. Then they are obviously $ f$-equivalent. Since $ \omega_1 $ is $ f$-invariant by assumption, we have that $ \omega_1(p_1)=\omega_1(p_1')$, which imply that $ \mathcal{L}^k(\omega_1\oplus\omega_2)(p)$ is well - defined. Let us now express that $ \mathcal{L}^k(\omega_1\oplus\omega_2)$ satisfies a smooth compatibility condition. Let $ h: U'\to U$ be an average smooth function

_1)\oplks_{comp}\Omega^k(X_2)\to\Omega^k(X_1\cuk_f X_2)$$ given by setjibg, for every plot $p:U\go X_1\cup_f X_2$ defined on a connxctee $U$, $$\mqthcal{L}^k(\omega_1\oplus\omegx_2)(p)=\left\{\begpn{array}{ll} \omeja_1(p_1) & \mbox{if }p=\hat{m}_1\dirc p_1\mnjx{ fkv somz 'lot }p_1\mbox{ of }X_1,\\ \pmega_2(p_2) & \mbmx{if }p=i_2\circ p_2\mtob{ yor some plot }p_2\mbox{ of }X_2. \end{array}\righe.$$ For anu hwo diffeologisal x[acea $X_1$ and $X_2$ and for every smooth map $f:X_1\supstteq Y\to X_2$ the map $\mathcal{L}^k$ is well-defined. Wf nefd to show that $\mahhcal{L}^k(\omegq_1\opltw\omega_2)(p)$ does not depend on the chojce of the lift of $p$ to a plot $o_i$ of $X_i$, and thqt thf assignment $p\mapfto\mathcal{L}^k(\omega_1\oplgs\omega_2)(l)$ satisfies thc smomth compatibility conditmon. The former of thgse claims ia obvious if $p$ lidtw to d plmt ow $X_2$; kndted, sjnce $i_2$ is injective, such a lifr is unique. Let $p_1$ amd $i_1'$ be two lifta of $p$ tj some plots of $X_1$. Then they are obviousny $r$-equivalent. Since $\omega_1$ is $f$-invariant by assomption, we have that $\omega_1(p_1)=\omega_1(p_1')$, which implies that $\mathcan{L}^k(\omxgx_1\opouf\uneha_2)(p)$ is well-defined. Let us now show that $\mathcaj{M}^k(\pmvga_1\oplus\omega_2)$ satlsfies a smooth cokpwtofility conditkon. Lec $h:H'\to U$ be an ordinagy smoojh map

_1)\oplus_{comp}\Omega^k(X_2)\to\Omega^k(X_1\cup_f X_2)$$ given by setting, for every X_1\cup_f defined on connected $U$, $$\mathcal{L}^k(\omega_1\oplus\omega_2)(p)=\left\{\begin{array}{ll} for plot }p_1\mbox{ of \omega_2(p_2) & \mbox{if p_2\mbox{ for some plot }p_2\mbox{ of \end{array}\right.$$ For any two diffeological spaces $X_1$ and $X_2$ and for every smooth $f:X_1\supseteq Y\to X_2$ the map $\mathcal{L}^k$ is well-defined. We need to show that does depend the of the lift of $p$ to a plot $p_i$ of $X_i$, and that the assignment $p\mapsto\mathcal{L}^k(\omega_1\oplus\omega_2)(p)$ the smooth compatibility condition. The former of these is obvious if $p$ to a plot of $X_2$; since is injective, a is Let $p_1$ and be two lifts of $p$ to some plots of $X_1$. Then they are obviously $f$-equivalent. Since $\omega_1$ $f$-invariant by have that which that is well-defined. Let show that $\mathcal{L}^k(\omega_1\oplus\omega_2)$ satisfies a smooth $h:U'\to U$ be an ordinary smooth map

_1)\oplus_{comp}\Omega^k(X_2)\to\Omega^k(X_1\Cup_f X_2)$$ given By setTinG, foR eVery Plot $P:U\to X_1\cup_f X_2$ defiNEd on A connected $U$, $$\mathcal{L}^k(\omEga_1\opLuS\OmegA_2)(P)=\lEft\{\beGin{arraY}{Ll} \OMEga_1(P_1) & \mBoX{if }P=\hAT{i}_1\Circ p_1\MboX{ for somE plot }p_1\mbox{ Of }X_1,\\ \OmEga_2(p_2) & \mbox{if }p=i_2\CIrC p_2\mbox{ for sOme Plot }p_2\mbox{ of }X_2. \End{Array}\rIgHt.$$ FOR any tWo dIffeoLogicaL Spaces $x_1$ and $X_2$ and fOr EVery smOOth map $f:x_1\SUpSeteQ Y\to X_2$ the map $\mathcaL{l}^k$ IS well-defined. We Need to ShOW tHAT $\maThcAl{L}^k(\omega_1\oPlUs\omeGA_2)(p)$ does nOT dEPENd oN The choice of thE lift of $p$ to a PLot $P_i$ of $X_i$, AnD thAT the asSignmEnT $P\maPsto\mathcal{l}^k(\omEga_1\oplus\oMega_2)(p)$ sATisfies THe smootH compaTibIliTy coNDiTiOn. THe FOrmER oF thESe cLaims is oBvIoUs if $p$ LiftS TO A Plot Of $X_2$; IndeEd, sinCe $i_2$ is injectivE, suCh a lIFt iS uniqUe. Let $P_1$ and $P_1'$ bE two lIfts of $P$ to soMe Plots of $X_1$. Then theY are Obviously $F$-eqUiValEnT. SincE $\Omega_1$ iS $f$-iNvaRiant by AssumptIOn, wE hAVE ThAt $\omega_1(p_1)=\omega_1(p_1')$, whicH iMPLiEs that $\maThcal{L}^K(\OmEgA_1\Oplus\omeGa_2)(P)$ is Well-DEFined. let uS NoW show thaT $\mathcAL{L}^K(\oMega_1\oplUs\Omega_2)$ sAtIsfIes A smooTH comPatibiLity condItion. lEt $h:U'\to U$ be an ordINary smooth map

_1)\oplus_{comp}\Omega^k(X _2)\to\Ome ga^k( X_1 \cu p_ f X_ 2)$$ given by sett i ng,for every plot $p:U\to X_1\ cu p _f X _ 2$ defi ned ona c o n nec te d$U$ ,$ $\ mathc al{ L}^k(\o mega_1\opl us\ om ega_2)(p)=\l e ft \{\begin{a rra y}{ll} \ome ga_ 1(p_1) & \m b ox{if }p =\hat {i}_1\ c irc p_ 1\mbox{ f or some p l ot }p_1 \ m bo x{ o f }X_1,\\ \omega_ 2 (p _ 2) & \mbox{if}p=i_2 \c i rc p _2\ mbo x{ for som eplot} p_2\mbo x {o f }X_ 2 . \end{array} \right.$$ F orany tw odif f eologi cal s pa c es$X_1$ and $ X_2$ and foreverys mooth m a p $f:X_ 1\sups ete q Y \toX _2 $the m a p $ \ ma thc a l{L }^k$ iswe ll -defi ned. W e nee d t o sh ow th at $\mathcal{ L}^ k(\o m ega _1\op lus\o mega _2 )(p)$ doesnot d ep end on the choi ce o f the lif t o f$p$ t o a p l ot $p_ i$of$X_i$,and tha t th ea s s ig nment $p\mapsto\ma th c a l{ L}^k(\om ega_1\ o pl us \ omega_2) (p )$sati s f ies t he s m oo th compa tibili t yco ndition .The fo rm erofthese clai ms isobviousif $p $ lifts to a pl o t of $X_2$; i n de e d ,s ince $i _2$ is inje ctiv e , su ch a li fti s uni que.Le t $ p _1$ and $p_1'$ be t wo lifts of $ p$ to some pl ots of $X_ 1 $ . Then th ey a r eo bviously $f$-e quiva lent. Sinc e $\omega _1$ i s $f$-in variant b y assumpti on, we ha vet h at $\omega_1(p_ 1 ) =\om eg a_1(p_1 ')$ , which im pli estha t$\mathcal {L}^k(\o me ga _1 \o plu s\ome g a_2)(p)$ i s w el l-d efine d . Let us n ow s ho wt hat $\math c al { L }^k( \o me ga_1 \op lu s\ome ga_2 ) $ s atisfie s a smoot h c o mpat ib il ity con dition. Let $ h: U'\to U$ b eanordina r y smoothmap

_1)\oplus_{comp}\Omega^k(X_2)\to\Omega^k(X_1\cup_f X_2)$$_given by_setting, for every plot_$p:U\to X_1\cup_f_X_2$_defined on_a_connected $U$, $$\mathcal{L}^k(\omega_1\oplus\omega_2)(p)=\left\{\begin{array}{ll}_ \omega_1(p_1) & \mbox{if_}p=\hat{i}_1\circ p_1\mbox{ for some_plot }p_1\mbox{ of_}X_1,\\ \omega_2(p_2)_& \mbox{if }p=i_2\circ p_2\mbox{ for some plot }p_2\mbox{ of }X_2. \end{array}\right.$$ For any two diffeological spaces_$X_1$_and $X_2$_and_for_every smooth map $f:X_1\supseteq Y\to_X_2$ the map $\mathcal{L}^k$ is_well-defined. We need_to show that $\mathcal{L}^k(\omega_1\oplus\omega_2)(p)$ does not depend on_the_choice of the_lift of $p$ to a plot $p_i$ of $X_i$,_and that the assignment $p\mapsto\mathcal{L}^k(\omega_1\oplus\omega_2)(p)$ satisfies_the smooth compatibility_condition._The_former of these claims_is obvious if $p$ lifts to_a plot of $X_2$; indeed, since_$i_2$ is injective, such a lift is_unique. Let $p_1$ and $p_1'$ be_two lifts of $p$ to_some plots_of $X_1$. Then they are_obviously $f$-equivalent. Since_$\omega_1$ is_$f$-invariant by assumption,_we have that $\omega_1(p_1)=\omega_1(p_1')$, which implies_that $\mathcal{L}^k(\omega_1\oplus\omega_2)(p)$ is_well-defined. Let us now show that $\mathcal{L}^k(\omega_1\oplus\omega_2)$_satisfies_a smooth compatibility_condition._Let_$h:U'\to U$_be an ordinary_smooth_map

the occurrence of the second dredge-up during the Early-AGB of intermediate-mass stars, that causes a significant reduction of their core masses. In Table \[tab\_mc1\] of Appendix \[app\_fit\] we present the fitting coefficients that we derive following the parametrization proposed by @WagenGroen_98, for several metallicities. In each panel of Fig. \[fig\_mc1\] the fitting curves are over-imposed to the `PARSEC` data for $M_{\rm c,1}$. We note, however, that our TP-AGB calculations use the true $M_{\rm c,1}$ values, and not those derived from the formulas. TP-AGB evolution {#ssec_tpagbev} ---------------- For each stellar model with initial parameters $(M_{\rm i},Z_{\rm i})$ the characteristic quantities at the $1^{\rm st}$ thermal pulse (core mass, luminosity, effective temperature, envelope chemical composition), obtained from the [*PARSEC*]{} database, are fed as initial conditions to the `COLIBRI` code, which computes the TP-AGB evolution until when almost the entire envelope is lost by stellar winds. Operatively the `COLIBRI` calculations are stopped when the mass of the residual envelope falls below a limit of $0.002 M_{\odot} -0.005 M_{\odot}$. At this stage all evolutionary tracks are already evolving off the AGB towards higher effective temperatures, with a luminosity that depends mainly on the mass of the C-O core, and the phase of the pulse cycle at which the last event of mass ejection took place (see Fig \[fig\_hr\]). For the present work we adopt a specific set of prescriptions for the mass loss and the third dredge-up, which we briefly outline below. These models will serve as a reference case for our ongoing TP-AGB calibration, and therefore the current parameters may be somewhat changed in future calculations. Anyhow, from various preliminary tests made with the present models, we expect that they already yield a fairly good description of the TP-AGB phase. #### Mass loss. {#mass-loss..unnumbered} It has been included under the hypothesis that it is driven by two main mechanisms, dominating at different stages. Initially, before radiation pressure on dust grains becomes the main agent of stellar winds, mass loss is described with the semi-empirical relation by

the occurrence of the second dredge - up during the Early - AGB of intermediate - mass stars, that cause a meaning reduction of their core batch. In Table   \[tab\_mc1\ ] of Appendix   \[app\_fit\ ] we stage the fitting coefficients that we derive surveil the parametrization proposed by @WagenGroen_98, for several metallicities. In each jury of Fig.   \[fig\_mc1\ ] the fitting curves are over - levy to the 'PARSEC 'data for $ M_{\rm c,1}$. We note, however, that our TP - AGB calculation use the true $ M_{\rm c,1}$ values, and not those derived from the formulas. TP - AGB development { # ssec_tpagbev } ---------------- For each stellar model with initial parameters $ (M_{\rm i},Z_{\rm i})$ the characteristic quantity at the $ 1^{\rm st}$ thermal pulse (core mass, luminosity, effective temperature, envelope chemical composition), obtain from the [ * PARSEC * ] { } database, are fed as initial weather to the 'COLIBRI 'code, which computes the TP - AGB evolution until when almost the entire envelope is lost by leading winds. Operatively the 'COLIBRI 'calculations are stopped when the mass of the residual envelope falls below a limit of $ 0.002 M_{\odot } -0.005 M_{\odot}$. At this stagecoach all evolutionary tracks are already evolving off the AGB towards higher effective temperatures, with a luminosity that depends mainly on the mass of the C - O core, and the phase of the pulse hertz at which the last event of bulk ejection took place (see Fig   \[fig\_hr\ ]). For the present workplace we adopt a specific set of prescriptions for the mass personnel casualty and the third dredge - up, which we briefly outline below. These models will serve as a reference sheath for our ongoing TP - AGB calibration, and therefore the current parameters may be somewhat changed in future calculation. Anyhow, from various preliminary test made with the present models, we expect that they already yield a fairly effective description of the TP - AGB phase. # # # # Mass loss. { # mass - personnel casualty.. unnumbered } It has been included under the hypothesis that it is driven by two independent mechanisms, dominate at different phase. Initially, before radiation press on dust grain becomes the chief agent of stellar wind, mass passing is described with the semi - empirical relation by

thf occurrence of the secokd dredge-up duriut the Xarly-AGG of intdrmediate-mass stars, that caudew a sugnificant reduction ow their clre massws. In Rable \[tab\_mc1\] of Appekbix \[apl\_nit\] wz 'resent the fitjing coeffichents that we gefire following the parametrization pro[osed bu @AagenGroen_98, for sevtraj mefallicities. In each panel of Fig. \[fif\_mc1\] the fitting curvrs are over-imposed to the `OARSFC` data for $M_{\rm c,1}$. Ae note, howgber, rhat our TP-AEB calculauimns use thg true $M_{\rm c,1}$ values, and not thosd dernved from tye fogkulas. TP-AGB xvolutpon {#ssec_tpagbcn} ---------------- For eawh stelkar model with inmtiao parameters $(M_{\rm i},Z_{\rm i})$ the characterisjic quantidizs at the $1^{\rm st}$ thermql pulsg (cora marw, ljmihoxify, efffctmve temperafure, envelope chemical composiuiog), obtained froj the [*[AWSEC*]{} database, are fed as initial conditpons to the `COLIBRI` code, whuch computes the TP-AGH evolutijn until when almost the entire envelope is lost ty stxluar wlnds. Ipfratively the `COLIBRI` calculations are stoppeq wnek the mass of thc residual envelopr vakjs below a lioit of $0.002 M_{\kdot} -0.005 M_{\odot}$. At thid stage all wvolutionwry yracks are already evolving off the AGB rowards higher effzctive tempexaturex, witn a luminosity that depznds mzinly on thf mass of ghe C-O core, and ghe pvase of ufe pulse cycle at which thx lasc event uf msss ejqction toon plaga (see Fig \[fig\_hr\]). For hhe ptesent work we afopt a specific set of prescripvmons for the kavs noss and the bhird dredge-up, rhich we briefky outlnne beuow. These jodels xill serve af a reference fase for our ongoing TP-ATB cqlibratkun, and therefote the cugrtnt paramerers may be somewhet cfznged in future caoculations. Anyhpw, wroi narmous [seliminary tasts maar witf the presekt oodeks, we expect that thay amready yield a faitln good dewcriptiog of the TP-AGN phase. #### Mass loss. {#lass-lmss..nnnumbrreq} It has been included under ths hypothedis that it is qrivcn bi two main kechanisms, dominating at different stagxs. Initially, before raduation pressure on buxt grains bxcomes the main agent of stellar wibds, mass loss is cescribed with the seji-empisical relation by

the occurrence of the second dredge-up during of stars, that a significant reduction Table of Appendix \[app\_fit\] present the fitting that we derive following the parametrization by @WagenGroen_98, for several metallicities. In each panel of Fig. \[fig\_mc1\] the fitting are over-imposed to the `PARSEC` data for $M_{\rm c,1}$. We note, however, that TP-AGB use true c,1}$ values, and not those derived from the formulas. TP-AGB evolution {#ssec_tpagbev} ---------------- For each stellar with initial parameters $(M_{\rm i},Z_{\rm i})$ the characteristic at the $1^{\rm st}$ pulse (core mass, luminosity, effective envelope composition), obtained the database, fed as initial to the `COLIBRI` code, which computes the TP-AGB evolution until when almost the entire envelope is lost stellar winds. `COLIBRI` calculations stopped the of the residual below a limit of $0.002 M_{\odot} this stage all evolutionary tracks are already evolving the AGB higher effective temperatures, with a luminosity depends mainly on the mass of the C-O and the phase of the pulse cycle at which the last event of mass ejection (see Fig \[fig\_hr\]). For present work we a set prescriptions the mass and the third dredge-up, which we briefly outline below. These models serve as a reference case for our ongoing TP-AGB calibration, the parameters may be changed in future calculations. from preliminary tests made with models, expect yield fairly description of the TP-AGB #### Mass loss. {#mass-loss..unnumbered} It been included under the by two main mechanisms, dominating at different stages. before radiation pressure on dust grains becomes main agent of stellar winds, mass loss is described with the semi-empirical by

the occurrence of the second dRedge-up durIng thE EaRly-aGb of iNterMediate-mass staRS, thaT causes a significant redUctioN oF TheiR CoRe masSes. In TaBLe \[TAB\_mc1\] Of apPenDiX \[ApP\_fit\] wE prEsent thE fitting coEffIcIents that we dERiVe followinG thE parametrizaTioN propoSeD by @wAgenGRoeN_98, for sEveral MEtalliCities. In eAcH Panel oF fig. \[fig\_mC1\] THe FittIng curves are over-iMPoSEd to the `PARSEC` dAta for $m_{\rM C,1}$. WE NOte, HowEver, that ouR Tp-AGB cALculatiONs USE The TRue $M_{\rm c,1}$ values, And not those DEriVed froM tHe fORmulas. tP-AGB EvOLutIon {#ssec_tpagBev} ---------------- FOr each steLlar moDEl with iNItial paRameteRs $(M_{\Rm i},z_{\rm i})$ THe ChAraCtERisTIc QuaNTitIes at the $1^{\Rm St}$ ThermAl puLSE (COre mAss, LumiNositY, effective temPerAturE, EnvElope ChemiCal cOmPositIon), obtAined FrOm the [*PARSEC*]{} dataBase, Are fed as iNitIaL coNdItionS To the `CoLIbRI` Code, whiCh compuTEs tHe tp-agB Evolution until when AlMOSt The entirE envelOPe Is LOst by steLlAr wInds. oPEratiVely THe `cOLIBRI` cAlculaTIoNs Are stopPeD when tHe MasS of The reSIduaL envelOpe falls Below A Limit of $0.002 M_{\odot} -0.005 M_{\oDOt}$. At this stage ALl EVOlUTionAry Tracks are alReadY EvolVing OFf The agB towArds hIgHEr EFfective temperatureS, wIth a luMinosIty that dependS mainly on tHE MAss of the c-O coRE, aND the phase of the Pulse Cycle at whiCH the last Event Of mass ejEction tooK PLace (see FIg \[fIg\_hR\]). FoR thE PReSent work we adoPT A speCiFic set oF prEscriptIonS foR thE maSs Loss and thE third drEdGe-Up, WhIch We briEFly outliNe BelOw. theSe modELs will Serve As a rEfErENce Case for OUr ONGoinG Tp-AgB caLibRaTion, aNd thERefOre the cUrrent parAmeTErs mAy Be SomewhaT changed in futUrE calculatiOnS. AnYhow, frOM Various pReliminary tests made with THe preseNt mOdels, We exPect that tHey AlreadY yiELd a faiRly gooD descRiPtiON Of the tp-aGb phAsE. #### Mass loss. {#mASS-loSs..unnUmBereD} It has bEen included under thE HypOthesis that it Is dRiveN BY tWo mAIn MEchAnISms, DOMinating at diffeRent stages. inITiAlly, before RAdiAtIon presSure on dUst grAIns becoMes the maiN agent of sTeLlar WINds, Mass loss is DescribeD with the sEMi-empIRiCal reLatIon by

the occurrence of the sec ond dredge -up d uri ngth e Ea rly- AGB of interme d iate -mass stars, that caus es asi g nifi c an t red uctiono ft h eir c or e m as s es . In Ta ble \[t ab\_mc1\]ofAp pendix \[app \ _f it\] we pr ese nt the fitti ngcoeffi ci ent s that we deri ve fol l owingthe param et r izatio n propos e d b y @W agenGroen_98, for se v eral metallici ties.In ea c h pa nel of Fig. \ [f ig\_m c 1\] the fi t t i ngc urves are ove r-imposed t o th e `PAR SE C`d ata fo r $M_ {\ r m c ,1}$. We no te,however,that o u r TP-AG B calcul ations us e t he t r ue $ M_{ \r m c, 1 }$ va l ues , and no tth ose d eriv e d f romthe for mulas . TP-AGB evo lut ion{ #ss ec_tp agbev } -- -- ----- ------ - Fo reach stellar mo delwith init ial p ara me ters$ (M_{\r m i },Z _{\rm i })$ the cha ra c t e ri stic quantities at t h e $ 1^{\rm s t}$ th e rm al pulse (c or e m ass, l umino sity , e ffective tempe r at ur e, enve lo pe che mi cal co mposi t ion) , obta ined fro m the [*PARSEC*]{} d a tabase, are f e da s i n itia l c onditions t o th e `CO LIBR I `cod e , whi ch co mp u te s the TP-AGB evoluti on until when almost the e ntire enve l o p e is los t by st e llar winds. Op erati vely the ` C OLIBRI`calcu lationsare stopp e d when th e m ass of th e re sidual envelo p e fal ls belowa l imit of $0 .00 2 M _{\ od ot} -0.00 5 M_{\od ot }$ .At th is st a ge all e vo lut io nar y tra c ks are alre adyev ol v ing off th e A G B tow ar ds hig her e ffect ivet emp erature s, with a lu m inos it ythat de pends mainlyon the massof th e C-Oc o re, andthe phase of the pulsec ycle at wh ich t he l ast event of masseje c tion t ook pl ace ( se e F i g  \[fi g \ _h r\] ). For thep r ese nt wo rk weadopt a specific set of p r esc riptions forthe mas s lo ssa nd the t h ird d redge-up, which we briefl yo ut line below . Th es e model s willserve as a re ference c ase for o ur ong o i ngTP-AGB cal ibration , and the r efore th e cur ren t para me ter s may be so m ewh at ch angedin futur e cal cu lations. Anyhow, from various p relimi narytes ts made w ith the presentmode ls, we exp ect th at th eya lread y yi e ld af airly goo d descript i on of t he TP-AGB pha s e . # ### M ass loss.{#ma ss-loss..unnumber e d} It has bee n in c l ude d u n derth e hypothesis t hat i t is drive nby two main mechani sm s , dom inatin g at d ifferen t st a ges. I niti all y, before ra di a tion pr es su r e on d ustgr ains b ecomes them a in agent of stel lar w i n ds, m a sslossis descri b ed w ith the se mi-empirica l rela tion by

the_occurrence of_the second dredge-up during_the Early-AGB_of_intermediate-mass stars,_that_causes a significant_reduction of their_core masses. In Table \[tab\_mc1\] of_Appendix \[app\_fit\] we present_the_fitting coefficients that we derive following the parametrization proposed by @WagenGroen_98, for several metallicities._In_each panel_of_Fig. \[fig\_mc1\]_the fitting curves are over-imposed_to the `PARSEC` data for_$M_{\rm c,1}$._We note, however, that our TP-AGB calculations use_the_true $M_{\rm c,1}$_values, and not those derived from the formulas. TP-AGB evolution_{#ssec_tpagbev} ---------------- For each stellar model with initial_parameters $(M_{\rm i},Z_{\rm_i})$_the_characteristic quantities at the_$1^{\rm st}$ thermal pulse (core mass,_luminosity, effective temperature, envelope chemical composition),_obtained from the [*PARSEC*]{} database, are fed_as initial conditions to the `COLIBRI`_code, which computes the TP-AGB_evolution until_when almost the entire envelope_is lost by_stellar winds._Operatively the `COLIBRI`_calculations are stopped when the mass_of the residual_envelope falls below a limit of_$0.002_M_{\odot} -0.005 M_{\odot}$._At_this_stage all_evolutionary tracks are_already_evolving off_the_AGB towards higher effective temperatures, with_a_luminosity that depends mainly on the mass_of the C-O core,_and_the phase of the_pulse cycle at which the_last event of mass ejection took_place (see_Fig \[fig\_hr\]). For the_present work we adopt a specific set of prescriptions for the_mass loss and the third dredge-up,_which we briefly outline_below. These_models_will serve as_a_reference case_for our ongoing TP-AGB calibration, and therefore_the current_parameters may be somewhat changed in_future calculations. Anyhow, from_various_preliminary tests made with the present_models, we expect that they already_yield a fairly good description_of_the_TP-AGB phase. #### Mass loss. {#mass-loss..unnumbered} It_has been included under the hypothesis_that it is_driven by two main mechanisms, dominating at_different_stages. Initially, before radiation pressure on_dust_grains becomes the main agent of_stellar_winds,_mass loss is described with_the semi-empirical relation by

rm bce}$ exceeds a critical value $T_{\rm dup}$ at the stage of post-flash luminosity peak. The quantity $T_{\rm dup}$ is assumed as a free parameter. In Figure \[fig\_mc3\] the left-hand side panels display the $M_{\rm c}^{\rm min}$ predictions for $\log(T_{\rm dup})= 6.2, 6.4, 6.5, 6.6, 6.7, 6.8$ and three values of the initial metallicity, $Z_{\rm i}=0.02, \,Z_{\rm i}=0.008, \,{\rm and}\, Z_{\rm i}=0.004$. The numerical method described in Sect. \[ssec\_tbdred\] has been applied for stellar masses ranging from $1\,M_{\odot}$ to $3\,M_{\odot}$ in steps of $0.05\,M_{\odot}$. In practice, once set the minimum temperature $T_{\rm dup}$, for each initial stellar mass and chemical composition, $M_{\rm c}^{\rm min}$ is the value of the core mass for which $T_{\rm bce}=T_{\rm dup}$ is satisfied. The solution is found iteratively with envelope integrations adopting the Brent root-finding algorithm [chapter IX of “Numerical Recipes”; @Press_etal88]. In each case $M_{\rm c}^{\rm min}$ is taken as the maximum between the value obtained by the envelope-integration method and the core mass at the first thermal pulse, $M_{\rm c,1}$. We do not show the results for $M> 3M_{\odot}$, since for the higher masses the temperature criterion is always satisfied since the onset of the TP-AGB, regardless of the value $T_{\rm dup}$. We see that all the curves share the same trend. Starting from lower masses towards the higher ones, $M_{\rm c}^{\rm min}$ slightly decreases, reaches a minimum and then steeply increases. It is interesting to note that the minimum in $M_{\rm c}^{\rm min}$ corresponds exactly to the critical maximum mass, $M_{\rm HeF}$, for a star to develop a degenerate He-core and experience the He-flash at the tip of the RGB. This reflects the same correspondence between $M_{\rm HeF}$ and the minimum of $M_{\rm c,1}$ (see Fig. \[fig\_mc1\]), already pointed out long ago

rm bce}$ exceeds a critical value $ T_{\rm dup}$ at the stage of post - flash luminosity peak. The measure $ T_{\rm dup}$ is assume as a free argument. In Figure   \[fig\_mc3\ ] the left - handwriting side panel display the $ M_{\rm c}^{\rm min}$ predictions for $ \log(T_{\rm dup})= 6.2, 6.4, 6.5, 6.6, 6.7, 6.8 $ and three value of the initial metallicity, $ Z_{\rm i}=0.02, \,Z_{\rm i}=0.008, \,{\rm and}\, Z_{\rm i}=0.004$. The numerical method identify in Sect.   \[ssec\_tbdred\ ] has been applied for stellar batch ranging from $ 1\,M_{\odot}$ to $ 3\,M_{\odot}$ in steps of $ 0.05\,M_{\odot}$. In practice, once set the minimum temperature $ T_{\rm dup}$, for each initial stellar batch and chemical composition, $ M_{\rm c}^{\rm min}$ is the value of the core mass for which $ T_{\rm bce}=T_{\rm dup}$ is quenched. The solution is found iteratively with envelope integrations adopt the Brent root - find algorithm [ chapter IX of “ Numerical Recipes ”; @Press_etal88 ]. In each case $ M_{\rm c}^{\rm min}$ is taken as the maximum between the value obtained by the envelope - integration method acting and the core mass at the first thermal pulse, $ M_{\rm c,1}$. We do not show the results for $ M > 3M_{\odot}$, since for the high masses the temperature criterion is always satisfied since the onset of the TP - AGB, regardless of the value $ T_{\rm dup}$. We see that all the curves share the same trend. start from lower masses towards the high ones, $ M_{\rm c}^{\rm min}$ slightly decreases, reaches a minimum and then steeply increase. It is interesting to notice that the minimum in $ M_{\rm c}^{\rm min}$ corresponds exactly to the critical maximum bulk, $ M_{\rm HeF}$, for a star to develop a degenerate He - core and experience the He - flash at the tip of the RGB. This reflects the same correspondence between $ M_{\rm HeF}$ and the minimum of $ M_{\rm c,1}$ (go steady Fig.   \[fig\_mc1\ ]), already sharpen out long ago

rm hce}$ exceeds a critical vxlue $T_{\rm dup}$ at the stege of lost-flasf luminosity peak. The quantivy $T_{\em duk}$ is assumed as a ffee paramvter. In Fiture \[hig\_mc3\] the left-hais side icnels fispney the $M_{\rm c}^{\rm kin}$ predicdions for $\log(T_{\sm dbp})= 6.2, 6.4, 6.5, 6.6, 6.7, 6.8$ and three values of the inieial meyaplicity, $Z_{\rm i}=0.02, \,E_{\rm i}=0.008, \,{\wm ahd}\, Z_{\rm i}=0.004$. The numerical method descdibed ii Sect. \[ssec\_tbdrec\] has been applied for steplar masses ranging frlm $1\,M_{\odot}$ to $3\,M_{\odjr}$ in steps ow $0.05\,M_{\odot}$. In practice, onde set the minimum temperature $G_{\rm dbp}$, for each ibitldl stellar nass wnd chemical composidion, $M_{\rk c}^{\rm min}$ is tme vanue of the core mass for which $T_{\rm bce}=T_{\rm qup}$ is sadiafied. The solutiob us foond ideraguveuy sivh snvelooe mntegrationa adopting rhe Brent root-findimg qlgorithm [chalter IV jf “Numerical Recipes”; @Press_etal88]. In each base $M_{\rm c}^{\rm min}$ is taken aw the maximum between the valuq obtained by the envelope-integration method and dhe ckfe nafr ah the first thermal pulse, $M_{\rm c,1}$. We do not shor tne results for $M> 3M_{\odot}$, sincr vot the higher mxsses cge temperature critegion is alwats satisfyed xince the onset of the TP-AGV, regardless if the value $T_{\rm dbp}$. We see thct all the vurves share the same txend. Sfarting frol lower mzrses towards the hinhes ones, $M_{\rm c}^{\rm min}$ slightlr decreasxs, recches a oinikum anq then stefply ltcreases. It is intfrestnng tm note thah the minimum in $M_{\rm c}^{\rm min}$ coccesponds exacjly to the cricical kaximum mass, $I_{\rm HeF}$, for a xtar to deveuop a degeherate Ie-core and evperience the Je-flash at tie tip of the RGB. This rdwlects the samr correspondence berween $M_{\rm HeF}$ and bhe mkhimum of $M_{\rm c,1}$ (wee Fig. \[fig\_mc1\]), alreacy ooigtvd pue long ago

rm bce}$ exceeds a critical value $T_{\rm the of post-flash peak. The quantity a parameter. In Figure the left-hand side display the $M_{\rm c}^{\rm min}$ predictions $\log(T_{\rm dup})= 6.2, 6.4, 6.5, 6.6, 6.7, 6.8$ and three values of the metallicity, $Z_{\rm i}=0.02, \,Z_{\rm i}=0.008, \,{\rm and}\, Z_{\rm i}=0.004$. The numerical method described Sect. has applied stellar masses ranging from $1\,M_{\odot}$ to $3\,M_{\odot}$ in steps of $0.05\,M_{\odot}$. In practice, once set the temperature $T_{\rm dup}$, for each initial stellar mass chemical composition, $M_{\rm c}^{\rm is the value of the mass which $T_{\rm dup}$ satisfied. solution is found with envelope integrations adopting the Brent root-finding algorithm [chapter IX of “Numerical Recipes”; @Press_etal88]. In each case c}^{\rm min}$ as the between value by the envelope-integration the core mass at the first c,1}$. We do not show the results for 3M_{\odot}$, since the higher masses the temperature criterion always satisfied since the onset of the TP-AGB, of the value $T_{\rm dup}$. We see that all the curves share the same trend. lower masses towards the ones, $M_{\rm c}^{\rm slightly reaches minimum then steeply It is interesting to note that the minimum in $M_{\rm c}^{\rm corresponds exactly to the critical maximum mass, $M_{\rm HeF}$, for to a degenerate He-core experience the He-flash at tip the RGB. This reflects correspondence $M_{\rm minimum $M_{\rm (see Fig. \[fig\_mc1\]), already out long ago

rm bce}$ exceeds a critical valuE $T_{\rm dup}$ at tHe staGe oF poSt-FlasH lumInosity peak. The QUantIty $T_{\rm dup}$ is assumed as a fRee paRaMEter. iN FIgure \[Fig\_mc3\] thE LeFT-HanD sIdE paNeLS dIsplaY thE $M_{\rm c}^{\rm Min}$ predictIonS fOr $\log(T_{\rm dup})= 6.2, 6.4, 6.5, 6.6, 6.7, 6.8$ aND tHree values Of tHe initial metAllIcity, $Z_{\Rm I}=0.02, \,Z_{\rM I}=0.008, \,{\rm anD}\, Z_{\rM i}=0.004$. The NumeriCAl methOd describEd IN Sect. \[sSEc\_tbdreD\] HAs Been Applied for stellar MAsSEs ranging from $1\,M_{\Odot}$ to $3\,m_{\oDOt}$ IN StePs oF $0.05\,M_{\odot}$. In prAcTice, oNCe set thE MiNIMUm tEMperature $T_{\rm dUp}$, for each inITiaL stellAr MasS And cheMical CoMPosItion, $M_{\rm c}^{\rm Min}$ iS the value Of the cORe mass fOR which $T_{\Rm bce}=T_{\Rm dUp}$ iS satISfIeD. ThE sOLutIOn Is fOUnd IterativElY wIth enVeloPE INTegrAtiOns aDoptiNg the Brent rooT-fiNdinG AlgOrithM [chapTer Ix oF “NumeRical REcipeS”; @PRess_etal88]. In each cAse $M_{\Rm c}^{\rm min}$ iS taKeN as ThE maxiMUm betwEen The Value obTained bY The EnVELOpE-integration method AnD THe Core mass At the fIRsT tHErmal pulSe, $m_{\rm C,1}$. We dO NOt shoW the REsUlts for $M> 3m_{\odot}$, sINcE fOr the hiGhEr massEs The TemPeratURe crIterioN is alwayS satiSFied since the onSEt of the TP-AGB, rEGaRDLeSS of tHe vAlue $T_{\rm dup}$. WE see THat aLl thE CuRveS Share The saMe TReND. Starting from lower mAsSes towArds tHe higher ones, $M_{\Rm c}^{\rm min}$ slIGHTly decreAses, REaCHes a minimum and Then sTeeply incrEAses. It is InterEsting to Note that tHE Minimum iN $M_{\rM c}^{\rM miN}$ coRREsPonds exactly tO THe crItIcal maxImuM mass, $M_{\rM Hef}$, foR a sTar To Develop a dEgeneratE HE-cOrE aNd eXperiENce the He-FlAsh At The Tip of THe RGB. THis reFlecTs ThE SamE corresPOnDENce bEtWeEn $M_{\rM Hef}$ aNd the MiniMUm oF $M_{\rm c,1}$ (seE Fig. \[fig\_mc1\]), AlrEAdy pOiNtEd out loNg ago

rm bce}$ exceeds a critica l value $T _{\rm du p}$ a t th e st age of post-fl a sh l uminosity peak. The qu antit y$ T_{\ r mdup}$ is ass u me d asafr eepa r am eter. I n Figur e \[fig\_m c3\ ]the left-han d s ide panels di splay the $M _{\ rm c}^ {\ rmm in}$pre dicti ons fo r $\log (T_{\rm d up } )= 6.2 , 6.4, 6 . 5 ,6.6, 6.7, 6.8$ and th r ee values of theinitia lm et a l lic ity , $Z_{\rmi} =0.02 , \,Z_{\ r mi } = 0.0 0 8, \,{\rm and }\, Z_{\rmi }=0 .004$. T hen umeric al me th o d d escribed in Sec t. \[ssec \_tbdr e d\] has been ap pliedfor st ella r m as ses r a ngi n gfro m $1 \,M_{\od ot }$ to $ 3\,M _ { \ o dot} $ i n st eps o f $0.05\,M_{\ odo t}$. Inpract ice,once s et th e mini mum t em perature $T_{\r m du p}$, foreac hini ti al st e llar m ass an d chemi cal com p osi ti o n , $ M_{\rm c}^{\rm min }$ i sthe valu e of t h eco r e mass f or wh ich$ T _{\rm bce } =T _{\rm du p}$ is sa ti sfied.Th e solu ti onisfound iter ativel y with e nvelo p e integrations adopting theB re n t r o ot-f ind ing algorit hm [ c hapt er I X o f “ N umeri cal R ec i pe s ”; @Press_etal88].In eachcase$M_{\rm c}^{\ rm min}$ i s t aken asthem ax i mum between th e val ue obtaine d by theenvel ope-inte gration m e t hod andthe co remas s at the first th e r malpu lse, $M _{\ rm c,1} $.Wedonot s how the r esults f or $ M> 3 M_{ \odot } $, since f orth e h igher masses thetemp er at u recriteri o ni s alw ay ssati sfi ed sinc e th e on set ofthe TP-AG B,r egar dl es s of th e value $T_{\ rm dup}$. We s eethat a l l the cur ves share the same tren d . Start ing from low er masses to wardsthe higher ones, $M_{ \r m c } ^ {\rmm i n} $ s li ghtly decr e a ses , rea ch es a minimu m and then steeply inc reases. It is in tere s t in g t o n o teth a t t h e minimum in $M_ {\rm c}^{\ rm mi n}$ corres p ond sexactly to the crit i cal max imum mass , $M_{\rm H eF}$ , for a star to develop a degene r ate H e -c ore a ndexperi en cethe H e-flas h at thetip of t he RGB . Thi sreflects the same correspondenc e betw een $ M_{ \rm HeF}$ an d th e minimum of$M_{\rm c, 1}$ (s ee Fi g.\ [fig\ _mc1 \ ]) , a l ready poi n ted out l o ng ag o

rm bce}$_exceeds a_critical value $T_{\rm dup}$_at the_stage_of post-flash_luminosity_peak. The quantity_$T_{\rm dup}$ is_assumed as a free_parameter. In Figure \[fig\_mc3\] the_left-hand_side panels display the $M_{\rm c}^{\rm min}$ predictions for $\log(T_{\rm dup})= 6.2, 6.4, 6.5,_6.6,_6.7, 6.8$_and_three_values of the initial metallicity,_$Z_{\rm i}=0.02, \,Z_{\rm i}=0.008, \,{\rm_and}\, Z_{\rm_i}=0.004$. The numerical method described in Sect. \[ssec\_tbdred\] has_been_applied for stellar_masses ranging from $1\,M_{\odot}$ to $3\,M_{\odot}$ in steps of_$0.05\,M_{\odot}$. In practice, once set the_minimum temperature $T_{\rm_dup}$,_for_each initial stellar mass_and chemical composition, $M_{\rm c}^{\rm min}$_is the value of the core_mass for which $T_{\rm bce}=T_{\rm dup}$ is_satisfied. The solution is found iteratively_with envelope integrations adopting the_Brent root-finding_algorithm [chapter IX of “Numerical_Recipes”; @Press_etal88]. In_each case_$M_{\rm c}^{\rm min}$_is taken as the maximum between_the value obtained_by the envelope-integration method and the_core_mass at the_first_thermal_pulse, $M_{\rm_c,1}$. We do_not_show the_results_for $M> 3M_{\odot}$, since for the_higher_masses the temperature criterion is always satisfied_since the onset of_the_TP-AGB, regardless of the_value $T_{\rm dup}$. We see_that all the curves share the_same trend._Starting from_lower masses towards the higher ones, $M_{\rm c}^{\rm min}$ slightly decreases,_reaches a minimum and then steeply_increases. It is interesting_to note_that_the minimum in_$M_{\rm_c}^{\rm min}$_corresponds exactly to the critical maximum mass,_$M_{\rm HeF}$,_for a star to develop a_degenerate He-core and experience_the_He-flash at the tip of the_RGB. This reflects the same correspondence_between $M_{\rm HeF}$ and the_minimum_of_$M_{\rm c,1}$ (see Fig. \[fig\_mc1\]), already_pointed out long ago

an invariant measure of $X$ if it is an invariant measure of $X_t$ for every $t \in \mathbb{R}$. We will denote by $\mathcal{M}_X$ the set of all invariant measures of $X$. A subset $Y\subset Z$ has *total probability* if for every $\mu\in \mathcal{M}_X$ we have $\mu(Y)=1$ (see [@Man82]). The support of a measure $\mu$, denoted by $supp(\mu)$, is the set of points for which the measure is non-zero. An invariant measure is said to be *atomic* if its support is either a closed orbit or a singularity. A probability measure $\mu$ is an *ergodic measure* if for every invariant set $A$ we have $\mu(A) = 1$ or $\mu(A) = 0$. Finally, a certain property is said to be valid in *$\mu$-almost every point* if it is valid in the whole Z except, possibly, in a set of null measure. We recall the definition of $\delta$-closable points of [@Man82]. We say that a point $x \in M \setminus Sing(X)$ is $\delta$-closable if, for any $C^1$ neighborhood ${{\mathcal U}}\subset {\mathfrak{X}^{1}(M)}$ of $X$, there exists a vector field $Z \in {{\mathcal U}}$, a point $z \in M$ and $T > 0$ such that: 1. $Z_T(z) = z$, 2. $Z = X$ on $M \setminus B_{\delta} (X_{[0, T]}(x))$ and 3. $dist(Z_t(z), X_t(x)) < \delta, \forall 0 \leq t \leq T$. We denote by $\Sigma(X)$ the set of points of $M$ which are $\delta$-closable for any $\delta$ sufficiently small. If $\Lambda$ is a strongly homogeneous set for $X$ with singularities all of them hyperbolic, then $X$ is a star flow in $\Lambda$. By Ergodic Closing Lemma, the $\delta$-closable set of $X$ has total probability. If $x \in \Lambda$ is a regular $\delta$-closable point

an invariant measure of $ X$ if it is an invariant measure of $ X_t$ for every $ metric ton \in \mathbb{R}$. We will announce by $ \mathcal{M}_X$ the set of all invariant bill of $ X$. A subset $ Y\subset Z$ have * total probability * if for every $ \mu\in \mathcal{M}_X$ we have $ \mu(Y)=1 $ (experience [ @Man82 ]). The accompaniment of a measure $ \mu$, denote by $ supp(\mu)$, is the set of points for which the meter is non - zero. An invariant measure is suppose to be * atomic * if its support is either a closed orbit or a singularity. A probability measure $ \mu$ is an * ergodic measure * if for every changeless set $ A$ we have $ \mu(A) = 1 $ or $ \mu(A) = 0$. Finally, a sealed property is said to be valid in * $ \mu$-almost every point * if it is valid in the whole Z except, possibly, in a stage set of null measure. We recall the definition of $ \delta$-closable points of [ @Man82 ]. We say that a point $ x \in M \setminus Sing(X)$ is $ \delta$-closable if, for any $ C^1 $ vicinity $ { { \mathcal U}}\subset { \mathfrak{X}^{1}(M)}$ of $ X$, there exists a vector field $ Z \in { { \mathcal U}}$, a point $ z \in M$ and $ T > 0 $ such that: 1. $ Z_T(z) = z$, 2. $ Z = X$ on $ M \setminus B_{\delta } (X_{[0, T]}(x))$ and 3. $ dist(Z_t(z), X_t(x) ) < \delta, \forall 0 \leq t \leq T$. We denote by $ \Sigma(X)$ the set of points of $ M$ which are $ \delta$-closable for any $ \delta$ sufficiently small. If $ \Lambda$ is a strongly homogeneous set for $ X$ with singularities all of them hyperbolic, then $ X$ is a star flow in $ \Lambda$. By Ergodic Closing Lemma, the $ \delta$-closable stage set of $ X$ has total probability. If $ x \in \Lambda$ is a even $ \delta$-closable point

an invariant measure of $X$ lf it is an invatiqnt meesure or $X_t$ for every $t \in \mathbb{R}$. We will venore by $\mathcal{M}_X$ the set of xll invarpant measyres if $X$. A subset $Y\subset Z$ gws *tmval probability* if for evary $\mu\in \mathcdl{O}_X$ we have $\mu(Y)=1$ (see [@Man82]). The support of a measirf $\mu$, denoted bi $supk(\mu)$, is fhe set of points for which the mezsure iv non-zero. An onvariant measure is said ho bf *atomic* if its suoport is eijger q closed orbkt or a singularity. A ptobability measure $\mu$ is an *ergoaic mzasure* if fir evfty invariant set $W$ we have $\mu(A) = 1$ or $\mg(A) = 0$. Fimally, a certaik pro'ertt is said to be valid in *$\mu$-almost every point* if ic is valid in the whooe Z exwept, posrublh, ih e sst of julm measure. Ws recall thw definition of $\delua$-cjisable points of [@Mag82]. Re say that a point $x \in M \setminus Sinc(X)$ js $\delta$-closable if, for any $C^1$ neighborhood ${{\mwthcal U}}\stbset {\mathfrak{X}^{1}(M)}$ of $X$, there exists a vector field $Z \in {{\oatkgwu U}}$, a point $z \in M$ and $T > 0$ such that: 1. $Z_T(z) = z$, 2. $Z = S$ pn $M \setminus B_{\dclta} (X_{[0, T]}(x))$ and 3. $disy(Z_h(z), V_t(x)) < \delta, \fotall 0 \lzs f \leq T$. We denote bj $\Sigma(V)$ the set of pjintx of $M$ which are $\delta$-closavle for any $\belra$ sufficiently smcll. If $\Lambda$ is s strpngly homogeneous set fur $X$ with singuparities zul of them hypercolpc, tven $X$ is a star flow in $\Laibda$. By Erjodic Closine Lekma, thq $\delta$-clodable set of $X$ has total probcbilidy. If $x \in \Pambda$ is a regular $\delta$-closable point

an invariant measure of $X$ if it invariant of $X_t$ every $t \in $\mathcal{M}_X$ set of all measures of $X$. subset $Y\subset Z$ has *total probability* for every $\mu\in \mathcal{M}_X$ we have $\mu(Y)=1$ (see [@Man82]). The support of a $\mu$, denoted by $supp(\mu)$, is the set of points for which the measure non-zero. invariant is to be *atomic* if its support is either a closed orbit or a singularity. A probability $\mu$ is an *ergodic measure* if for every set $A$ we have = 1$ or $\mu(A) = Finally, certain property said be in *$\mu$-almost every if it is valid in the whole Z except, possibly, in a set of null measure. We the definition points of We that point $x \in Sing(X)$ is $\delta$-closable if, for any U}}\subset {\mathfrak{X}^{1}(M)}$ of $X$, there exists a vector $Z \in U}}$, a point $z \in M$ $T > 0$ such that: 1. $Z_T(z) = 2. $Z = X$ on $M \setminus B_{\delta} (X_{[0, T]}(x))$ and 3. $dist(Z_t(z), X_t(x)) < 0 \leq t \leq We denote by the of of which are for any $\delta$ sufficiently small. If $\Lambda$ is a strongly homogeneous for $X$ with singularities all of them hyperbolic, then $X$ star in $\Lambda$. By Closing Lemma, the $\delta$-closable of has total probability. If \Lambda$ a

an invariant measure of $X$ if it Is an invariAnt meAsuRe oF $X_T$ for EverY $t \in \mathbb{R}$. We wILl deNote by $\mathcal{M}_X$ the set oF all iNvARianT MeAsureS of $X$. A suBSeT $y\SubSeT Z$ Has *ToTAl ProbaBilIty* if foR every $\mu\in \MatHcAl{M}_X$ we have $\mu(y)=1$ (SeE [@Man82]). The supPorT of a measure $\mU$, deNoted bY $sUpp(\MU)$, is thE seT of poInts foR Which tHe measure Is NOn-zero. aN invariANT mEasuRe is said to be *atomiC* If ITs support is eitHer a clOsED oRBIt oR a sIngularity. a pRobabILity meaSUrE $\MU$ Is aN *Ergodic measurE* if for every INvaRiant sEt $a$ we HAve $\mu(A) = 1$ Or $\mu(A) = 0$. fiNAllY, a certain prOperTy is said tO be valID in *$\mu$-alMOst everY point* If iT is ValiD In ThE whOlE z exCEpT, poSSibLy, in a set Of NuLl meaSure. wE RECall The DefiNitioN of $\delta$-closaBle PoinTS of [@man82]. We Say thAt a pOiNt $x \in m \setmiNus SiNg(x)$ is $\delta$-closablE if, fOr any $C^1$ neiGhbOrHooD ${{\mAthcaL u}}\subseT {\maThfRak{X}^{1}(M)}$ of $x$, there eXIstS a VECToR field $Z \in {{\mathcal U}}$, a PoINT $z \In M$ and $T > 0$ sUch thaT: 1. $z_T(Z) = z$, 2. $z = x$ on $M \setmInUs B_{\DeltA} (x_{[0, t]}(x))$ and 3. $Dist(z_T(z), x_t(x)) < \delta, \Forall 0 \LEq T \lEq T$. We deNoTe by $\SiGmA(X)$ tHe sEt of pOInts Of $M$ whiCh are $\delTa$-cloSAble for any $\deltA$ Sufficiently sMAlL. iF $\LAMbda$ Is a Strongly homOgenEOus sEt foR $x$ wIth SIngulAritiEs ALl OF them hyperbolic, then $x$ iS a star Flow iN $\Lambda$. By ErgoDic Closing lEMMa, the $\delTa$-clOSaBLe set of $X$ has totAl proBability. If $X \In \Lambda$ Is a reGular $\delTa$-closablE POint

an invariant measure of $ X$ if it i s aninv ari an t me asur e of $X_t$ for ever y $t \in \mathbb{R}$.We wi ll deno t eby $\ mathcal { M} _ X $ t he s etof al l inv ari ant mea sures of $ X$. A subset $Y\s u bs et Z$ has*to tal probabil ity * if f or ev e ry $\ mu\ in \m athcal { M}_X$we have $ \m u (Y)=1$ (see [@ M a n8 2]). The support of a me a sure $\mu$, de notedby $s u p p(\ mu) $, is these t ofp oints f o rw h i cht he measure is non-zero.A n i nvaria nt me a sure i s sai dt o b e *atomic*if i ts suppor t is e i ther ac losed o rbit o r a si ngul a ri ty . Ap rob a bi lit y me asure $\ mu $is an *er g o d i c me asu re*if fo r every invar ian t se t $A $ wehave$\mu (A ) = 1 $ or $ \mu(A )= 0$. Finally,a ce rtain pro per ty is s aid t o be va lid in *$\mu$ -almost eve ry p o in t* if it is validin t he whole Z excep t ,po s sibly, i na s et o f nullmeas u re . We re call t h ede finitio nof $\d el ta$ -cl osabl e poi nts of [@Man82 ]. We say that a poi n t $x \in M \s e tm i n us Sing (X) $ is $\delt a$-c l osab le i f ,for any $ C^1$ne i gh b orhood ${{\mathcalU} }\subs et {\ mathfrak{X}^{ 1}(M)}$ of $ X $, there exi s ts a vector field $Z \ in {{\math c al U}}$, a po int $z \ in M$ and $ T > 0$ s uch th at: 1 . $ Z_T(z) = z$,2 . $ Z= X$ on $M \setmi nus B_ {\d elt a} (X_{[0,T]}(x))$ a nd 3. $ dist( Z _t(z), X _t (x) )< \ delta , \fora ll 0\leq t \ l eqT$. We de n o te b y$\ Sigm a(X )$ theseto f p oints o f $M$ whi cha re $ \d el ta$-clo sable for any $ \delta$ su ff ici entlys m all. If $\Lambda$ is a strongl y homoge neo us se t fo r $X$ wit h s ingula rit i es all of th em hy pe rbo l i c, th e n $ X$is a star fl o w in $\La mb da$. By Er godic Closing Lemm a , t he $\delta$-c los able s et of $X $ ha st ota l probability. I f $x \in \ La m bd a$ is a re g ula r$\delta $-closa ble p o int

an_invariant measure_of $X$ if it_is an_invariant_measure of_$X_t$_for every $t_\in \mathbb{R}$. We_will denote by $\mathcal{M}_X$_the set of_all_invariant measures of $X$. A subset $Y\subset Z$ has *total probability* if for every_$\mu\in_\mathcal{M}_X$ we_have_$\mu(Y)=1$_(see [@Man82]). The support of_a measure $\mu$, denoted by_$supp(\mu)$, is_the set of points for which the measure_is_non-zero. An invariant_measure is said to be *atomic* if its support_is either a closed orbit or_a singularity. A probability_measure_$\mu$_is an *ergodic measure*_if for every invariant set $A$_we have $\mu(A) = 1$ or_$\mu(A) = 0$. Finally, a certain property_is said to be valid in_*$\mu$-almost every point* if it_is valid_in the whole Z except,_possibly, in a_set of_null measure. We recall_the definition of $\delta$-closable points of_[@Man82]. We say_that a point $x \in M_\setminus_Sing(X)$ is $\delta$-closable_if,_for_any $C^1$_neighborhood ${{\mathcal U}}\subset_{\mathfrak{X}^{1}(M)}$_of $X$,_there_exists a vector field $Z \in_{{\mathcal_U}}$, a point $z \in M$ and_$T > 0$ such_that: 1._ $Z_T(z) = z$, 2._ $Z = X$ on_$M \setminus B_{\delta} (X_{[0, T]}(x))$ and 3._ $dist(Z_t(z),_X_t(x)) <_\delta, \forall 0 \leq t \leq T$. We denote by $\Sigma(X)$ the_set of points of $M$ which_are $\delta$-closable for any_$\delta$ sufficiently_small. If_$\Lambda$ is a_strongly_homogeneous set_for $X$ with singularities all of them_hyperbolic, then_$X$ is a star flow in_$\Lambda$. By Ergodic Closing Lemma,_the_$\delta$-closable set of $X$ has total_probability. If $x \in \Lambda$ is a_regular $\delta$-closable point

leq \frac{2\arrowvert\Ric\arrowvert}{\arrowvert X\arrowvert^2} \, \left| (\Li_{U_i^m}g ) \right| +\textit{O}(f^{-1}).\end{aligned}$$ Hence, we obtain the result by integrating over $\Omega_m$. $(iii)$ As before, we compute, $$\begin{aligned} \nabla_X \langle U_i^m,X\rangle&=&\langle \nabla_XU_i^m,X\rangle+\langle U_i^m,\nabla_XX\rangle\\ &=&2\langle \nabla_{U_i^m}X,X\rangle=\langle U_i^m,X\rangle-2\Ric(X,U_i^m)\\ &=&\langle U_i^m,X\rangle+\textit{O}(f^{-a(n)/2+\epsilon}).\end{aligned}$$ Now, by construction, we have that $\langle U_i^m,X\rangle=0$ on $M_{t_m^2/4}$. Hence, integrating the previous estimate on $\Omega_m$, we obtain $\langle U_i^m,X\rangle \,= \,\textit{O}(t_m^{-a(n)+\epsilon})$ on $\Omega_m$ and the result follows at once. $(iv)$ As we noticed in Remark \[remeq\], one has that $\Delta U_i^m+\Ric(U_i^m, \cdot ) \, = \, \Delta_fU_i^m+U_i^m/2$. On the other hand, it holds the identity $$\div(\Li_{U_i^m}g ) \, - \, \frac12 \nabla(\tr(\Li_{U_i^m}g )) \,\, = \,\, \Delta U_i^m \, + \,\Ric(U_i^m, \cdot ) \,.$$ To estimate the left hand side, we notice that (ii) gives $\hbox{$\sup_{\Omega_m}$} | \Li_{U_i^m} g \, | \, = \, \textit{O} \, (t_m^{-a(n)+\epsilon})$. Moreover, it is possible to deduce form (i) that $ \sup_{\Omega_m}\arrowvert\nabla^{k}\Li_{U_

leq \frac{2\arrowvert\Ric\arrowvert}{\arrowvert X\arrowvert^2 } \, \left| (\Li_{U_i^m}g) \right| + \textit{O}(f^{-1}).\end{aligned}$$ Hence, we obtain the result by integrating over $ \Omega_m$. $ (iii)$ As earlier, we calculate, $ $ \begin{aligned } \nabla_X \langle U_i^m, X\rangle&=&\langle \nabla_XU_i^m, X\rangle+\langle U_i^m,\nabla_XX\rangle\\ & = & 2\langle \nabla_{U_i^m}X, X\rangle=\langle U_i^m, X\rangle-2\Ric(X, U_i^m)\\ & = & \langle U_i^m, X\rangle+\textit{O}(f^{-a(n)/2+\epsilon}).\end{aligned}$$ Now, by construction, we get that $ \langle U_i^m, X\rangle=0 $ on $ M_{t_m^2/4}$. therefore, integrating the previous estimate on $ \Omega_m$, we receive $ \langle U_i^m, X\rangle \,= \,\textit{O}(t_m^{-a(n)+\epsilon})$ on $ \Omega_m$ and the result follows at once. $ (iv)$ As we detect in Remark   \[remeq\ ], one has that $ \Delta U_i^m+\Ric(U_i^m, \cdot) \, = \, \Delta_fU_i^m+U_i^m/2$. On the other bridge player, it holds the identity $ $ \div(\Li_{U_i^m}g) \, - \, \frac12 \nabla(\tr(\Li_{U_i^m}g) ) \,\, = \,\, \Delta U_i^m \, + \,\Ric(U_i^m, \cdot) \,.$$ To estimate the leftover hand side, we detect that (ii) gives $ \hbox{$\sup_{\Omega_m}$ } | \Li_{U_i^m } g \, | \, = \, \textit{O } \, (t_m^{-a(n)+\epsilon})$. Moreover, it is potential to deduce form (i) that $ \sup_{\Omega_m}\arrowvert\nabla^{k}\Li_{U _

leq \frac{2\arrowvert\Ric\arrowvevt}{\arrowvert X\arrowvert^2} \, \left| (\Li_{U_i^m}g ) \right| +\textit{O}(f^{-1}).\end{aligned}$$ Heice, qe obucin the result by ingegrating over $\Omwga_m$. $(mii)$ As before, we compute, $$\begin{zpignzd} \iabla_X \langle U_l^m,X\rangle&=&\latgle \nabla_XU_i^m,F\rxnyle+\langle U_i^m,\nabla_XX\rangle\\ &=&2\langle \nabja_{U_i^m}X,X\tajgle=\langle U_i^m,V\ranbje-2\Rid(Q,U_l^m)\\ &=&\langle U_i^m,X\rangle+\textit{O}(f^{-a(n)/2+\epsjlon}).\end{eligned}$$ Now, by vonstruction, we have that $\pangpe U_i^m,X\rangle=0$ on $M_{h_m^2/4}$. Hence, injsgrwring the prexious estimate on $\Omegz_m$, we obtain $\langle U_i^m,X\rangle \,= \,\textnt{O}(t_m^{-a(n)+\epsioob})$ oj $\Omega_m$ and the gesult follows at onca. $(iv)$ As ee noticed in Vemarn \[reneq\], one has that $\Delte U_i^m+\Ric(U_i^m, \cdot ) \, = \, \Delta_fU_i^k+U_n^m/2$. On the other hand, ut holdv tha iddbtigy $$\sit(\Li_{H_i^m}g ) \, - \, \hrac12 \nabla(\td(\Li_{U_i^m}g )) \,\, = \,\, \Delta U_i^m \, + \,\Ric(U_i^m, \cqit ) \,.$$ To estimzte thq jeft hand side, we notice that (ii) gives $\vbos{$\sup_{\Omega_m}$} | \Li_{U_i^m} g \, | \, = \, \textit{O} \, (t_m^{-a(n)+\epsiloj})$. Moreovew, it is possible to deduce form (i) that $ \sup_{\Omega_m}\drrowteft\ncnjx^{j}\Ll_{U_

leq \frac{2\arrowvert\Ric\arrowvert}{\arrowvert X\arrowvert^2} \, \left| (\Li_{U_i^m}g ) Hence, obtain the by integrating over compute, \nabla_X \langle U_i^m,X\rangle&=&\langle U_i^m,\nabla_XX\rangle\\ &=&2\langle \nabla_{U_i^m}X,X\rangle=\langle &=&\langle U_i^m,X\rangle+\textit{O}(f^{-a(n)/2+\epsilon}).\end{aligned}$$ Now, by construction, we that $\langle U_i^m,X\rangle=0$ on $M_{t_m^2/4}$. Hence, integrating the previous estimate on $\Omega_m$, we $\langle U_i^m,X\rangle \,= \,\textit{O}(t_m^{-a(n)+\epsilon})$ on $\Omega_m$ and the result follows at once. $(iv)$ we in \[remeq\], has that $\Delta U_i^m+\Ric(U_i^m, \cdot ) \, = \, \Delta_fU_i^m+U_i^m/2$. On the other hand, it holds identity $$\div(\Li_{U_i^m}g ) \, - \, \frac12 \nabla(\tr(\Li_{U_i^m}g \,\, = \,\, \Delta \, + \,\Ric(U_i^m, \cdot ) To the left side, notice (ii) gives $\hbox{$\sup_{\Omega_m}$} \Li_{U_i^m} g \, | \, = \, \textit{O} \, (t_m^{-a(n)+\epsilon})$. Moreover, it is possible to deduce form that $

leq \frac{2\arrowvert\Ric\arrowvErt}{\arrowveRt X\arRowVerT^2} \, \lEft| (\LI_{U_i^m}G ) \right| +\textit{O}(f^{-1}).\ENd{alIgned}$$ Hence, we obtain the rEsult By INtegRAtIng ovEr $\Omega_M$. $(IiI)$ aS beFoRe, We cOmPUtE, $$\begiN{alIgned} \naBla_X \langle u_i^m,x\rAngle&=&\langle \nABlA_XU_i^m,X\rangLe+\lAngle U_i^m,\nablA_XX\Rangle\\ &=&2\LaNglE \Nabla_{u_i^m}x,X\ranGle=\lanGLe U_i^m,X\Rangle-2\Ric(x,U_I^M)\\ &=&\langlE u_i^m,X\ranGLE+\tExtiT{O}(f^{-a(n)/2+\epsilon}).\end{alIGnED}$$ Now, by construcTion, we HaVE tHAT $\laNglE U_i^m,X\ranglE=0$ oN $M_{t_m^2/4}$. HENce, inteGRaTING thE Previous estimAte on $\Omega_m$, WE obTain $\laNgLe U_I^M,X\rangLe \,= \,\texTiT{o}(t_m^{-A(n)+\epsilon})$ on $\omegA_m$ and the rEsult fOLlows at ONce. $(iv)$ As We notiCed In REmarK \[ReMeQ\], onE hAS thAT $\DEltA u_i^m+\ric(U_i^m, \cdOt ) \, = \, \deLta_fU_I^m+U_i^M/2$. oN THe otHer Hand, It holDs the identity $$\Div(\li_{U_i^M}G ) \, - \, \frAc12 \nabLa(\tr(\LI_{U_i^m}G )) \,\, = \,\, \DElta U_I^m \, + \,\Ric(U_I^m, \cdoT ) \,.$$ TO estimate the lefT hanD side, we noTicE tHat (Ii) Gives $\HBox{$\sup_{\omeGa_m}$} | \li_{U_i^m} g \, | \, = \, \tExtit{O} \, (t_M^{-A(n)+\ePsILON})$. MOreover, it is possiblE tO DEdUce form (i) That $ \suP_{\omEgA_M}\arrowveRt\NabLa^{k}\LI_{u_

leq \frac{2\arrowvert\Ric\ arrowvert} {\arr owv ert X \arr owve rt^2} \, \lef t | (\ Li_{U_i^m}g ) \right|+\tex ti t {O}( f ^{ -1}). \end{al i gn e d }$$ H en ce, w e o btain th e resul t by integ rat in g over $\Ome g a_ m$. $(iii )$As before, w e c ompute ,$$\ b egin{ ali gned} \nabl a _X \la ngle U_i^ m, X \rangl e &=&\lan g l e\nab la_XU_i^m,X\rangl e +\ l angle U_i^m,\n abla_X X\ r an g l e\\ &= &2\langle\n abla_ { U_i^m}X , X\ r a n gle = \langle U_i^m ,X\rangle-2 \ Ric (X,U_i ^m )\\ &=&\la ngleU_ i ^m, X\rangle+\t exti t{O}(f^{- a(n)/2 + \epsilo n }).\end {align ed} $$Now, by c ons tr u cti o n, we hav e that $ \l an gle U _i^m , X \ r angl e=0 $ on $M_{ t_m^2/4}$. He nce , in t egr ating theprev io us es timate on $ \O mega_m$, we obt ain$\langleU_i ^m ,X\ ra ngle\ ,= \,\ tex tit {O}(t_m ^{-a(n) + \ep si l o n }) $ on $\Omega_m$ an dt h eresult f ollows at o n ce. $(i v) $ A s we n otice d in Re mark \[r emeq\] , o ne has th at $\Del ta U_ i^m +\Ric ( U_i^ m, \cd ot ) \,= \,\ Delta_fU_i^m+U _ i^m/2$. On th e o t h er hand , i t holds the ide n tity $$\ d iv (\L i _{U_i ^m}g)\ ,- \, \frac12 \nabla( \t r(\Li_ {U_i^ m}g )) \,\, = \,\, \Del t a U_i^m \, + \ , \R i c(U_i^m, \cdot ) \, .$$ To est i mate the left hand si de, we no t i ce that(ii ) g ive s $ \ h bo x{$\sup_{\Ome g a _m}$ }| \Li_{ U_i ^m} g \ , | \, =\,\t extit{O}\, (t_m^ {- a( n) +\ eps ilon} ) $. Moreo ve r,it is poss i ble to dedu ce f or m( i)that $\ su p _ {\Om eg a_ m}\a rro wv ert\n abla ^ {k} \Li_{U_

leq \frac{2\arrowvert\Ric\arrowvert}{\arrowvert_X\arrowvert^2} \,_ \left| (\Li_{U_i^m}g )_\right| +\textit{O}(f^{-1}).\end{aligned}$$_Hence,_we obtain_the_result by integrating_over $\Omega_m$. $(iii)$ As_before, we compute, $$\begin{aligned} \nabla_X_\langle U_i^m,X\rangle&=&\langle \nabla_XU_i^m,X\rangle+\langle_U_i^m,\nabla_XX\rangle\\ &=&2\langle_\nabla_{U_i^m}X,X\rangle=\langle U_i^m,X\rangle-2\Ric(X,U_i^m)\\ &=&\langle U_i^m,X\rangle+\textit{O}(f^{-a(n)/2+\epsilon}).\end{aligned}$$ Now, by construction, we have that $\langle U_i^m,X\rangle=0$ on $M_{t_m^2/4}$. Hence,_integrating_the previous_estimate_on_$\Omega_m$, we obtain $\langle U_i^m,X\rangle_\,= \,\textit{O}(t_m^{-a(n)+\epsilon})$ on $\Omega_m$ and_the result_follows at once. $(iv)$ As we noticed in Remark \[remeq\],_one_has that $\Delta_U_i^m+\Ric(U_i^m, \cdot ) \, = \, \Delta_fU_i^m+U_i^m/2$. On the_other hand, it holds the identity_$$\div(\Li_{U_i^m}g ) \,_-_\,_\frac12 \nabla(\tr(\Li_{U_i^m}g )) \,\,_= \,\, \Delta U_i^m \, +_\,\Ric(U_i^m, \cdot ) \,.$$ To estimate_the left hand side, we notice that_(ii) gives $\hbox{$\sup_{\Omega_m}$} | \Li_{U_i^m} g_\, | \, = \,_\textit{O} \,_(t_m^{-a(n)+\epsilon})$. Moreover, it is possible_to deduce form_(i) that_$ \sup_{\Omega_m}\arrowvert\nabla^{k}\Li_{U_

} \subseteq \operatorname*{Mod-R}$ corresponding to the homological epimorphism $\lambda: R \rightarrow S$ via Theorem \[T:epiclass\], which in turn corresponds to $\mathcal{X}$ via Theorem \[T:BSepi\], can be written as follows: $$\mathcal{B} = \{M \in \operatorname*{Mod-R}\mid K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \otimes_R M \text{ is exact for all }(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \in \mathcal{G}(\mathcal{X})\}.$$ Furthermore, we can write $\mathcal{B} = \mathcal{C} \cap \mathcal{D}$, where $\mathcal{C}$ is the cosilting class corresponding to $\mathcal{X}$ via Theorem \[T01\], and $\mathcal{D}$ is the class of those $R$-modules $M$ such that $\operatorname*{F}_{\operatorname*{\mathfrak{q}}}(M) \in \operatorname*{Mod-R_{\operatorname*{\mathfrak{q}}}}$ for each gap $(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \in \mathcal{G}(\mathcal{X})$. Let us denote $$\mathcal{B}' = \{M \in \operatorname*{Mod-R}\mid K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \otimes_R M \text{ is exact for all }(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \in \mathcal{G}(\mathcal{X})\},$$ and the goal is to show that $\mathcal{B}=\mathcal{B}'$. We start by proving that $\mathcal{B}'$ is an extension-closed bireflective subcategory of $\operatorname*{Mod-R}$. Note that since $K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}})$ is a bounded complex of flat $R$-modules, then $\mathcal{B}' = \mathcal{L} \cap \operatorname*{Mod-R}$, where $\mathcal{L}$ is the Bousfield class of all objects $X$ of $\operatorname*{\mathbf{D}}(R)$ such that $K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \otimes_R^{\mathbf{L}} X = 0$ for all $(\operatorname*{\mathfrak

} \subseteq \operatorname*{Mod - R}$ corresponding to the homological epimorphism $ \lambda: R \rightarrow S$ via Theorem   \[T: epiclass\ ], which in turn correspond to $ \mathcal{X}$ via Theorem   \[T: BSepi\ ], can be write as follows: $ $ \mathcal{B } = \{M \in \operatorname*{Mod - R}\mid K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p } }) \otimes_R M \text { is accurate for all } (\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p } }) \in \mathcal{G}(\mathcal{X})\}.$$ Furthermore, we can write $ \mathcal{B } = \mathcal{C } \cap \mathcal{D}$, where $ \mathcal{C}$ is the cosilting course corresponding to $ \mathcal{X}$ via Theorem   \[T01\ ], and $ \mathcal{D}$ is the class of those $ R$-modules $ M$ such that $ \operatorname*{F}_{\operatorname*{\mathfrak{q}}}(M) \in \operatorname*{Mod - R_{\operatorname*{\mathfrak{q}}}}$ for each opening $ (\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p } }) \in \mathcal{G}(\mathcal{X})$. Let us denote $ $ \mathcal{B }' = \{M \in \operatorname*{Mod - R}\mid K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p } }) \otimes_R M \text { is accurate for all } (\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p } }) \in \mathcal{G}(\mathcal{X})\},$$ and the finish is to show that $ \mathcal{B}=\mathcal{B}'$. We start by testify that $ \mathcal{B}'$ is an extension - closed bireflective subcategory of $ \operatorname*{Mod - R}$. Note that since $ K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}})$ is a bounded complex of flat $ R$-modules, then $ \mathcal{B }' = \mathcal{L } \cap \operatorname*{Mod - R}$, where $ \mathcal{L}$ is the Bousfield class of all object $ X$ of $ \operatorname*{\mathbf{D}}(R)$ such that $ K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p } }) \otimes_R^{\mathbf{L } } X = 0 $ for all $ (\operatorname*{\mathfrak

} \suhseteq \operatorname*{Mod-R}$ gorresponding to the hokologidal epimurphism $\lambda: R \rightarrow D$ cia Tyeorem \[T:epiclass\], which kn turn clrrespones ti $\mathcal{X}$ tja Theovzm \[T:BSsii\], cau ue written as fpllows: $$\matvcal{B} = \{M \in \oparxtlrname*{Mod-R}\mid K(\operatorname*{\mathfrak{z}},\operatprjame*{\mathfrak{p}}) \jtimts_R M \tsqt{ is exact for all }(\operatornams*{\mathfrek{q}},\operatorname*{\kathfrak{p}}) \in \mathcal{G}(\mathcwl{X})\}.$$ Fkrthermore, we can arite $\mathcql{B} = \nathcal{C} \cap \mathcal{D}$, where $\mathcam{C}$ is the cosilting class correrpondnng to $\mathxao{X}$ gha Theorem \[T01\], and $\iathcal{D}$ is bne clavs of tnose $R$-modules $K$ snch rhat $\operatorname*{F}_{\opecatorname*{\mathfrak{q}}}(M) \yn \operatmruame*{Mod-R_{\operatorname*{\mqtyfrak{x}}}}$ fos eazy gxp $(\kpxrafornamf*{\mavhfrak{q}},\operztorname*{\matyfrak{p}}) \in \mathcal{G}(\mstrbsl{X})$. Let us dehote $$\mwtrcal{B}' = \{M \in \operatorname*{Mod-R}\mid K(\operatmrnzme*{\mathfrak{q}},\operatornamw*{\mathfrak{p}}) \otimes_R M \jext{ is exwct for all }(\operatorname*{\mathfrak{q}},\operatorname*{\mathxrak{p}}) \kn \nabhcau{T}(\mwthcal{X})\},$$ and the goal is to show that $\mathcal{B}=\iztncsl{B}'$. We start bn proving that $\matncwl{N}'$ is an extenskon-closed bireflective subcwtegory of $\operatornaie*{Moc-R}$. Note that since $K(\operatoename*{\mathfrah{w}},\operatorname*{\mathfxak{p}})$ is a bobnded vomplrx of flat $R$-modules, theu $\mathdal{B}' = \mathcwl{L} \cap \oldratorname*{Mod-R}$, wferv $\madhcal{L}$ is the Bousfield clwss of alo oblects $X$ uf $\okeratorgame*{\mathbf{F}}(R)$ sugv that $K(\operatornale*{\matkfrak{x}},\operatornwme*{\mathfrak{p}}) \otimes_R^{\mathbf{L}} X = 0$ for all $(\operajortamv*{\mathfrak

} \subseteq \operatorname*{Mod-R}$ corresponding to the homological R S$ via \[T:epiclass\], which in Theorem can be written follows: $$\mathcal{B} = \in \operatorname*{Mod-R}\mid K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \otimes_R M \text{ exact for all }(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \in \mathcal{G}(\mathcal{X})\}.$$ Furthermore, we can write $\mathcal{B} = \mathcal{C} \mathcal{D}$, where $\mathcal{C}$ is the cosilting class corresponding to $\mathcal{X}$ via Theorem \[T01\], $\mathcal{D}$ the of $R$-modules $M$ such that $\operatorname*{F}_{\operatorname*{\mathfrak{q}}}(M) \in \operatorname*{Mod-R_{\operatorname*{\mathfrak{q}}}}$ for each gap $(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \in \mathcal{G}(\mathcal{X})$. Let us denote = \{M \in \operatorname*{Mod-R}\mid K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \otimes_R M \text{ exact for all }(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \mathcal{G}(\mathcal{X})\},$$ and the goal is show $\mathcal{B}=\mathcal{B}'$. We by that is an extension-closed subcategory of $\operatorname*{Mod-R}$. Note that since $K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}})$ is a bounded complex of flat $R$-modules, then $\mathcal{B}' = \cap \operatorname*{Mod-R}$, is the class all $X$ of $\operatorname*{\mathbf{D}}(R)$ $K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \otimes_R^{\mathbf{L}} X = 0$ for

} \subseteq \operatorname*{Mod-R}$ cOrrespondiNg to tHe hOmoLoGicaL epiMorphism $\lambda: r \RighTarrow S$ via Theorem \[T:epicLass\], wHiCH in tURn CorreSponds tO $\MaTHCal{x}$ vIa theOrEM \[T:bSepi\], Can Be writtEn as followS: $$\maThCal{B} = \{M \in \operaTOrName*{Mod-R}\miD K(\oPeratorname*{\mAthFrak{q}},\oPeRatORname*{\MatHfrak{P}}) \otimeS_r M \text{ Is exact foR aLL }(\operaTOrname*{\mATHfRak{q}},\Operatorname*{\mathfRAk{P}}) \In \mathcal{G}(\mathCal{X})\}.$$ FuRtHErMORe, wE caN write $\mathCaL{B} = \matHCal{C} \cap \MAtHCAL{D}$, wHEre $\mathcal{C}$ is The cosiltinG ClaSs corrEsPonDIng to $\mAthcaL{X}$ VIa THeorem \[T01\], and $\mAthcAl{D}$ is the cLass of THose $R$-moDUles $M$ suCh that $\OpeRatOrnaME*{F}_{\OpEraToRNamE*{\MaThfRAk{q}}}(m) \in \operaToRnAme*{MoD-R_{\opERATOrnaMe*{\mAthfRak{q}}}}$ fOr each gap $(\operAtoRnamE*{\MatHfrak{Q}},\operAtorNaMe*{\matHfrak{p}}) \In \matHcAl{G}(\mathcal{X})$. Let uS denOte $$\mathcaL{B}' = \{M \In \OpeRaTornaME*{Mod-R}\mId K(\OpeRatornaMe*{\mathfRAk{q}},\OpERAToRname*{\mathfrak{p}}) \otimEs_r m \TeXt{ is exacT for alL }(\OpErATorname*{\mAtHfrAk{q}},\oPERatorName*{\MAtHfrak{p}}) \in \MathcaL{g}(\mAtHcal{X})\},$$ anD tHe goal Is To sHow That $\mAThcaL{B}=\mathCal{B}'$. We stArt by PRoving that $\mathCAl{B}'$ is an extensIOn-CLOsED birEflEctive subcaTegoRY of $\oPeraTOrNamE*{mod-R}$. NOte thAt SInCE $K(\operatorname*{\mathfRaK{q}},\operAtornAme*{\mathfrak{p}})$ iS a bounded cOMPLex of flaT $R$-moDUlES, then $\mathcal{B}' = \mAthcaL{L} \cap \operaTOrname*{MoD-R}$, wheRe $\mathcaL{L}$ is the BoUSField claSs oF alL obJecTS $x$ oF $\operatorname*{\MAThbf{d}}(R)$ Such thaT $K(\oPeratorNamE*{\maThfRak{Q}},\oPeratornaMe*{\mathfrAk{P}}) \oTiMeS_R^{\mAthbf{l}} x = 0$ for all $(\oPeRatOrNamE*{\mathFRak

} \subseteq \operatorname* {Mod-R}$ c orres pon din gto t he h omological epi m orph ism $\lambda: R \right arrow S $ via Th eorem  \[T:ep i cl a s s\] ,wh ich i n t urn c orr esponds to $\math cal {X }$ via Theor e m\[T:BSepi\ ],can be writt enas fol lo ws: $$\ma thc al{B} = \{M \in \o peratorna me * {Mod-R } \mid K( \ o pe rato rname*{\mathfrak{ q }} , \operatorname* {\math fr a k{ p } })\ot imes_R M \ te xt{ i s exactf or a l l } ( \operatorname *{\mathfrak { q}} ,\oper at orn a me*{\m athfr ak { p}} ) \in \math cal{ G}(\mathc al{X}) \ }.$$ F u rthermo re, we ca n w rite $\ ma thc al { B}= \ mat h cal {C} \cap \ ma thcal {D}$ , w h ere$\m athc al{C} $ is the cosi lti ng c l ass corr espon ding t o $\m athcal {X}$vi a Theorem \[T01 \],and $\mat hca l{ D}$ i s the classoftho se $R$- modules $M$ s u c h t hat $\operatorname *{ F } _{ \operato rname* { \m at h frak{q}} }( M)\in\ o perat orna m e* {Mod-R_{ \opera t or na me*{\ma th frak{q }} }}$ fo r eac h gap $(\op eratorna me*{\ m athfrak{q}},\o p eratorname*{\ m at h f ra k {p}} ) \ in \mathcal {G}( \ math cal{ X }) $.Let u s den ot e $ $ \mathcal{B}' = \{M\i n \ope rator name*{Mod-R}\ mid K(\ope r a t orname*{ \mat h fr a k{q}},\operato rname *{\mathfra k {p}}) \o times _R M \te xt{ is ex a c t for al l } (\o per ato r n am e*{\mathfrak{ q } },\o pe ratorna me* {\mathf rak {p} })\in \ mathcal{G }(\mathc al {X }) \} ,$$ andt he goalis to s how that $\math cal{B }=\m at hc a l{B }'$. We st a r t by p ro ving th at $\ma thca l {B} '$ is a n extensi on- c lose dbi reflect ive subcatego ry of $\oper at orn ame*{M o d -R}$. No te that since $K(\opera t orname* {\m athfr ak{q }},\opera tor name*{ \ma t hfrak{ p}})$is abo und e d comp l e xoffl at $R$-mod u l es, then $ \mat hcal{B} ' = \mathcal{L} \c a p \ operatorname* {Mo d-R} $ , w her e $ \ mat hc a l{L } $ is the Bousfie ld class o fa ll objects $ X $ o f$\opera torname *{\ma t hbf{D}} (R)$ such that $K( \o pera t o rna me*{\mathf rak{q}}, \operator n ame*{ \ ma thfra k{p }}) \o ti mes _R^{\ mathbf { L}} X =0$ for a ll $(\ opera to rname*{\ mathfrak

} \subseteq_\operatorname*{Mod-R}$ corresponding_to the homological epimorphism_$\lambda: R_\rightarrow_S$ via_Theorem \[T:epiclass\],_which in turn_corresponds to $\mathcal{X}$_via Theorem \[T:BSepi\], can be_written as follows:_$$\mathcal{B}_= \{M \in \operatorname*{Mod-R}\mid K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \otimes_R M \text{ is exact for all }(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \in_\mathcal{G}(\mathcal{X})\}.$$ Furthermore,_we can_write_$\mathcal{B}_= \mathcal{C} \cap \mathcal{D}$, where_$\mathcal{C}$ is the cosilting class_corresponding to_$\mathcal{X}$ via Theorem \[T01\], and $\mathcal{D}$ is the class_of_those $R$-modules $M$_such that $\operatorname*{F}_{\operatorname*{\mathfrak{q}}}(M) \in \operatorname*{Mod-R_{\operatorname*{\mathfrak{q}}}}$ for each gap $(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}})_\in \mathcal{G}(\mathcal{X})$. Let us denote $$\mathcal{B}' =_\{M \in \operatorname*{Mod-R}\mid_K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}})_\otimes_R_M \text{ is exact_for all }(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \in \mathcal{G}(\mathcal{X})\},$$ and_the goal is to show that_$\mathcal{B}=\mathcal{B}'$. We start by proving that $\mathcal{B}'$_is an extension-closed bireflective subcategory of_$\operatorname*{Mod-R}$. Note that since $K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}})$_is a_bounded complex of flat $R$-modules,_then $\mathcal{B}' =_\mathcal{L} \cap_\operatorname*{Mod-R}$, where $\mathcal{L}$_is the Bousfield class of all_objects $X$ of_$\operatorname*{\mathbf{D}}(R)$ such that $K(\operatorname*{\mathfrak{q}},\operatorname*{\mathfrak{p}}) \otimes_R^{\mathbf{L}} X_=_0$ for all_$(\operatorname*{\mathfrak

${\mbox{Att}\,}_R({\mbox{H}\, }^n_{\mathfrak{a}}(M))= {\mbox{Att}\,}_R({\mbox{H}\, }^n_{\mathfrak{b}}(M))$ with ${\mathfrak{b}}= {\bigcap\limits_{i=1}^rQ_i}$. Now, by [@DY2 Theorem 1.6], we have ${\mbox{H}\, }^n_{\mathfrak{a}}(M)\cong {\mbox{H}\, }^n_{\mathfrak{b}}(M)$. As ${\mbox{dim}\,}(R/{\mathfrak{b}})= 1$, the proof is complete. If ${\mbox{dim}\,}_R(M)=1$ then any subset $T$ of ${\mbox{Assh}\,}_R(M)$ is equal to the set ${\mbox{Att}\,}_R({\mbox{H}\, }^1_{\mathfrak{a}}(M))$ for some ideal ${\mathfrak{a}}$ of $R$. With notations as in Proposition 2.1, we take $Q_i={\mathfrak{q}}_i$ for $i=1,\cdots, r$. By a straightforward argument one may notice that the condition “complete" is superficial, for if $T$ is a non–empty proper subset of ${\mbox{Assh}\,}_R(M)$, then $T={\mbox{Att}\,}_R({\mbox{H}\, }^1_{\mathfrak{a}}(M))$, where ${\mathfrak{a}}=\underset{{\mathfrak{p}}\in{\mbox{Assh}\,}_R(M)\setminus T}{\cap}{\mathfrak{p}}$. The following is an example to Proposition 2.1. Set $R=k[[X,Y,Z,W]]$, where $k$ is a field and $X,Y,Z,W$ are independent indeterminates. Then $R$ is a complete Noetherian local ring with maximal ideal ${\mathfrak{m}}=(X,Y,Z,W)$. Consider prime ideals $${\mathfrak{p}}_1=(X,Y) \quad , \quad {\mathfrak{p}}_2=(Z,W)\quad, \quad {\mathfrak{p}}_3=(Y,Z) \quad, \quad {\mathfrak{p}}_4=(X,W)$$ and set $\displaystyle M=\frac{R}{{\mathfrak{p}}_1{\mathfrak{p}}_2{\mathfrak{p}}_3{\

$ { \mbox{Att}\,}_R({\mbox{H}\, } ^n_{\mathfrak{a}}(M))= { \mbox{Att}\,}_R({\mbox{H}\, } ^n_{\mathfrak{b}}(M))$ with $ { \mathfrak{b}}= { \bigcap\limits_{i=1}^rQ_i}$. Now, by [ @DY2 Theorem 1.6 ], we have $ { \mbox{H}\, } ^n_{\mathfrak{a}}(M)\cong { \mbox{H}\, } ^n_{\mathfrak{b}}(M)$. As $ { \mbox{dim}\,}(R/{\mathfrak{b}})= 1 $, the proof is complete. If $ { \mbox{dim}\,}_R(M)=1 $ then any subset $ T$ of $ { \mbox{Assh}\,}_R(M)$ is adequate to the located $ { \mbox{Att}\,}_R({\mbox{H}\, } ^1_{\mathfrak{a}}(M))$ for some ideal $ { \mathfrak{a}}$ of $ R$. With notations as in Proposition 2.1, we take $ Q_i={\mathfrak{q}}_i$ for $ i=1,\cdots, r$. By a square argumentation one may notice that the condition “ arrant " is superficial, for if $ T$ is a non – empty proper subset of $ { \mbox{Assh}\,}_R(M)$, then $ T={\mbox{Att}\,}_R({\mbox{H}\, } ^1_{\mathfrak{a}}(M))$, where $ { \mathfrak{a}}=\underset{{\mathfrak{p}}\in{\mbox{Assh}\,}_R(M)\setminus T}{\cap}{\mathfrak{p}}$. The pursuit is an example to Proposition 2.1. Set $ R = k[[X, Y, Z, W]]$, where $ k$ is a field and $ adam, Y, Z, W$ are independent indeterminates. Then $ R$ is a complete Noetherian local closed chain with maximal ideal $ { \mathfrak{m}}=(X, Y, Z, W)$. regard prime ideal $ $ { \mathfrak{p}}_1=(X, Y) \quad, \quad { \mathfrak{p}}_2=(Z, W)\quad, \quad { \mathfrak{p}}_3=(Y, Z) \quad, \quad { \mathfrak{p}}_4=(X, W)$$ and set $ \displaystyle M=\frac{R}{{\mathfrak{p}}_1{\mathfrak{p}}_2{\mathfrak{p}}_3{\

${\mblx{Att}\,}_R({\mbox{H}\, }^n_{\mathfrak{a}}(M))= {\obox{Att}\,}_R({\mbox{H}\, }^n_{\mathfran{b}}(M))$ wifh ${\mathffak{b}}= {\bigcap\limits_{i=1}^rQ_i}$. Now, by [@DT2 Theirem 1.6], we have ${\mbox{H}\, }^n_{\mxthfrak{a}}(M)\bong {\mbox{Y}\, }^n_{\methfrak{b}}(M)$. As ${\mboe{sim}\,}(R/{\matmyrak{b}})= 1$, the 'roof is compleje. If ${\mbox{dim}\,}_S(M)=1$ then any sutsdt $T$ of ${\mbox{Assh}\,}_R(M)$ is equal to the see ${\mbox{Ayt}\,}_G({\mbox{H}\, }^1_{\mathfrah{a}}(M))$ gjr skme ideal ${\mathfrak{a}}$ of $R$. With notatjons as in Propositipn 2.1, we take $Q_i={\mathfrak{q}}_i$ flr $i=1,\fdots, r$. By a straigjtforward atfumqbt one may nutice that the conditikn “complete" is superficial, for kf $T$ ns a non–empjv prlker subset oh ${\mbox{Wssh}\,}_R(M)$, then $B={\kbox{Atd}\,}_R({\mbox{H}\, }^1_{\mathfrak{a}}(M))$, whcre ${\methfeak{a}}=\underset{{\mathfrak{p}}\mn{\mbox{Assh}\,}_R(M)\setminus T}{\cap}{\mathxrck{p}}$. The following is ab wxampne tm Pruposktikn 2.1. Sst $R=k[[X,J,Z,W]]$, where $k$ ia a field abd $X,Y,Z,W$ are indepencegn indeterminafes. Thqn $R$ is a complete Noetherian local ring fitg maximal ideal ${\mathfraj{m}}=(X,Y,Z,W)$. Consider prime ideals $${\mwthfrak{p}}_1=(X,Y) \quad , \quad {\mathfrak{p}}_2=(Z,W)\quad, \quad {\mathxrak{p}}_3=(B,Z) \duab, \quad {\mwthfrak{p}}_4=(X,W)$$ and set $\displaystyle M=\frac{R}{{\mathfrak{[}}_1{\jauhfgak{p}}_2{\mathfrak{p}}_3{\

${\mbox{Att}\,}_R({\mbox{H}\, }^n_{\mathfrak{a}}(M))= {\mbox{Att}\,}_R({\mbox{H}\, }^n_{\mathfrak{b}}(M))$ with ${\mathfrak{b}}= {\bigcap\limits_{i=1}^rQ_i}$. [@DY2 1.6], we ${\mbox{H}\, }^n_{\mathfrak{a}}(M)\cong {\mbox{H}\, proof complete. If ${\mbox{dim}\,}_R(M)=1$ any subset $T$ ${\mbox{Assh}\,}_R(M)$ is equal to the set }^1_{\mathfrak{a}}(M))$ for some ideal ${\mathfrak{a}}$ of $R$. With notations as in Proposition 2.1, take $Q_i={\mathfrak{q}}_i$ for $i=1,\cdots, r$. By a straightforward argument one may notice that condition is for $T$ is a non–empty proper subset of ${\mbox{Assh}\,}_R(M)$, then $T={\mbox{Att}\,}_R({\mbox{H}\, }^1_{\mathfrak{a}}(M))$, where ${\mathfrak{a}}=\underset{{\mathfrak{p}}\in{\mbox{Assh}\,}_R(M)\setminus T}{\cap}{\mathfrak{p}}$. The following an example to Proposition 2.1. Set $R=k[[X,Y,Z,W]]$, where is a field and are independent indeterminates. Then $R$ a Noetherian local with ideal Consider prime ideals \quad , \quad {\mathfrak{p}}_2=(Z,W)\quad, \quad {\mathfrak{p}}_3=(Y,Z) \quad, \quad {\mathfrak{p}}_4=(X,W)$$ and set $\displaystyle M=\frac{R}{{\mathfrak{p}}_1{\mathfrak{p}}_2{\mathfrak{p}}_3{\

${\mbox{Att}\,}_R({\mbox{H}\, }^n_{\mathfrak{a}}(M))= {\mBox{Att}\,}_R({\mboX{H}\, }^n_{\maThfRak{B}}(M))$ With ${\MathFrak{b}}= {\bigcap\limITs_{i=1}^rq_i}$. Now, by [@DY2 Theorem 1.6], we have ${\Mbox{H}\, }^N_{\mAThfrAK{a}}(m)\cong {\Mbox{H}\, }^n_{\mAThFRAk{b}}(m)$. AS ${\mBox{DiM}\,}(r/{\mAthfrAk{b}})= 1$, The prooF is completE. If ${\MbOx{dim}\,}_R(M)=1$ then aNY sUbset $T$ of ${\mbOx{ASsh}\,}_R(M)$ is equal To tHe set ${\mBoX{AtT}\,}_r({\mbox{h}\, }^1_{\maThfraK{a}}(M))$ for SOme ideAl ${\mathfraK{a}}$ OF $R$. With NOtationS AS iN ProPosition 2.1, we take $Q_i={\mAThFRak{q}}_i$ for $i=1,\cdots, R$. By a stRaIGhTFOrwArd Argument onE mAy notICe that tHE cONDItiON “complete" is suPerficial, foR If $T$ Is a non–EmPty PRoper sUbset Of ${\MBox{assh}\,}_R(M)$, then $T={\Mbox{att}\,}_R({\mbox{H}\, }^1_{\MathfrAK{a}}(M))$, wherE ${\MathfraK{a}}=\undeRseT{{\maThfrAK{p}}\In{\MboX{ASSh}\,}_R(m)\SeTmiNUs T}{\Cap}{\mathfRaK{p}}$. the foLlowING IS an eXamPle tO PropOsition 2.1. Set $R=k[[X,y,Z,W]]$, WherE $K$ is A fielD and $X,y,Z,W$ aRe IndepEndent IndetErMinates. Then $R$ is a CompLete NoethEriAn LocAl Ring wITh maxiMal IdeAl ${\mathfRak{m}}=(X,Y,Z,w)$. conSiDER PrIme ideals $${\mathfrak{p}}_1=(x,Y) \QUAd , \Quad {\mathFrak{p}}_2=(Z,w)\QuAd, \QUad {\mathfRaK{p}}_3=(Y,z) \quaD, \QUad {\maThfrAK{p}}_4=(x,W)$$ and set $\DisplaYStYlE M=\frac{R}{{\MaThfrak{P}}_1{\mAthFraK{p}}_2{\matHFrak{P}}_3{\

${\mbox{Att}\,}_R({\mbox{ H}\, }^n_{ \math fra k{a }} (M)) = {\ mbox{Att}\,}_R ( {\mb ox{H}\, }^n_{\mathfrak {b}}( M) ) $ wi t h${\ma thfrak{ b }} = {\b ig ca p\l im i ts _{i=1 }^r Q_i}$.Now, by [@ DY2 T heorem 1.6], we have ${\m box {H}\, }^n_{\ mat hfrak{ a} }(M ) \cong {\ mbox{ H}\, } ^ n_{\ma thfrak{b} }( M )$. As ${\mbox { d im }\,} (R/{\mathfrak{b}} ) =1 $, the proof i s comp le t e. If${\ mbox{dim}\ ,} _R(M) = 1$ then an y s ubs e t $T$ of ${\m box{Assh}\, } _R( M)$ is e qua l to th e set $ { \mb ox{Att}\,}_ R({\ mbox{H}\, }^1_{ \ mathfra k {a}}(M) )$ for so meidea l $ {\ mat hf r ak{ a }} $ o f $R $. With n ot ation s as i n Prop osi tion 2.1, we take $Q_i ={\ math f rak {q}}_ i$ fo r $i =1 ,\cdo ts, r$ . By a straightforwar d ar gument on e m ay no ti ce th a t thecon dit ion “co mplete" issu p e r fi cial, for if $T$ i sa no n–emptyproper su bs e t of ${\ mb ox{ Assh } \ ,}_R( M)$, th en $T={\ mbox{A t t} \, }_R({\m bo x{H}\, } ^1_ {\m athfr a k{a} }(M))$ , where${\ma t hfrak{a}}=\und e rset{{\mathfr a k{ p } }\ i n{\m box {Assh}\,}_R (M)\ s etmi nusT }{ \ca p }{\ma thfra k{ p }} $ . The following is a n exam ple t o Proposition 2.1. Set $ R =k[[X,Y, Z,W] ] $, where $k$ is a fiel d and $X,Y , Z,W$ are inde pendentindetermi n a tes. The n $ R$isa c o m pl ete Noetheria n loca lring wi thmaximal id eal ${ \ma th frak{m}}= (X,Y,Z,W )$ .Co ns ide r pri m e ideals $ ${\ ma thf rak{p } }_1=(X ,Y) \ quad , \ qua d {\mat h fr a k {p}} _2 =( Z,W) \qu ad , \qu ad { \ma thfrak{ p}}_3=(Y, Z)\ quad ,\q uad {\ mathfrak{p}}_ 4= (X,W)$$ an dset $\dis p l aystyleM=\frac{R}{{\mathfrak{p } }_1{\ma thf rak{p }}_2 {\mathfra k{p }}_3{\

${\mbox{Att}\,}_R({\mbox{H}\,_}^n_{\mathfrak{a}}(M))= {\mbox{Att}\,}_R({\mbox{H}\,_}^n_{\mathfrak{b}}(M))$ with ${\mathfrak{b}}= {\bigcap\limits_{i=1}^rQ_i}$._Now, by_[@DY2_Theorem 1.6],_we_have ${\mbox{H}\, }^n_{\mathfrak{a}}(M)\cong_{\mbox{H}\, }^n_{\mathfrak{b}}(M)$. As_${\mbox{dim}\,}(R/{\mathfrak{b}})= 1$, the proof_is complete. If ${\mbox{dim}\,}_R(M)=1$_then_any subset $T$ of ${\mbox{Assh}\,}_R(M)$ is equal to the set ${\mbox{Att}\,}_R({\mbox{H}\, }^1_{\mathfrak{a}}(M))$ for some_ideal_${\mathfrak{a}}$ of_$R$. With_notations_as in Proposition 2.1, we_take $Q_i={\mathfrak{q}}_i$ for $i=1,\cdots, r$. By_a straightforward_argument one may notice that the condition “complete"_is_superficial, for if_$T$ is a non–empty proper subset of ${\mbox{Assh}\,}_R(M)$, then_$T={\mbox{Att}\,}_R({\mbox{H}\, }^1_{\mathfrak{a}}(M))$, where ${\mathfrak{a}}=\underset{{\mathfrak{p}}\in{\mbox{Assh}\,}_R(M)\setminus T}{\cap}{\mathfrak{p}}$. The following_is an example_to_Proposition_2.1. Set $R=k[[X,Y,Z,W]]$, where $k$_is a field and $X,Y,Z,W$ are_independent indeterminates. Then $R$ is a_complete Noetherian local ring with maximal ideal_${\mathfrak{m}}=(X,Y,Z,W)$. Consider prime ideals $${\mathfrak{p}}_1=(X,Y) \quad_, \quad {\mathfrak{p}}_2=(Z,W)\quad, \quad_ {\mathfrak{p}}_3=(Y,Z) \quad,_\quad {\mathfrak{p}}_4=(X,W)$$ and set_$\displaystyle M=\frac{R}{{\mathfrak{p}}_1{\mathfrak{p}}_2{\mathfrak{p}}_3{\

of the transition matrix will result in a shape signature that is too dependent on noise and the individualness of specific waveform to be of any use. Thus additional processing is necessary for further analysis; even a small set of states (12 x 12) will result in a feature vector with high dimensionality (144 dimensions). While a window and overlap size is assumed for the slotting to address the irregular sampling of the time series data, there are two adjustable features associated with this analysis: the kernel width associated with the slotting and the state space (alphabet) resolution. It is apparent that a range of resolutions and kernel width need to be tested to determine best performance given a generic supervised classifier. For these purposes a rapid initial classification algorithm, General Quadratic Discriminate Analysis [@Duda2012], was implemented to estimate the mis-classification rate (wrong decisions/total decisions). Not all states will be observed, i.e. the high dimensional feature vector will have information contained in a small subset of elements. Dimensionality reduction methods are often necessary for implementation of classification algorithms, in particular QDA where the construction of a covariance matrix of a sparse feature space can be problematic. The reduction of the large, sparse, feature vector resulting from the unpacking of the Markov Matrix is performed via extended canonical variate analysis or ECVA [@Norgaard2006]. The methodology for ECVA has roots in principle component analysis (PCA). PCA is a procedure performed on large multidimensional datasets with the intent of rotating what is a set of possibly correlated dimensions into a set of linearly uncorrelated variables [@Sch06]. The transformation results in a dataset, where the first principle component (dimension) has the largest possible variance. PCA is an unsupervised methodology, i.e. a priori known labels for the data being processed is not taken into consideration, thus a reduction in feature dimensionality and while it maximizes the variance it might not maximize the linear separability of the class space. In contrast to PCA, Canonical Variate Analysis does take class labels into considerations. The variation between groups is maximized resulting in a transformation that benefits the goal of separating classes. Given a set of data $\mathbf{x}$ with: $g$ different classes, $n_{i}$ observations of each class, and $r\times r$ dimensions in each observation; following @Johnson1992, the within-group and between-group covariance matrix is defined as: $$\mathbf{S}_{within}=\frac{1}{n

of the transition matrix will result in a shape signature that is excessively pendent on noise and the individualness of specific waveform to be of any use. therefore additional processing is necessary for further analysis; even a small stage set of states (12 x 12) will result in a feature of speech vector with eminent dimensionality (144 dimensions). While a windowpane and overlap size is assumed for the slotting to address the irregular sample distribution of the time series datum, there are two adjustable feature associated with this analysis: the kernel width associated with the slotting and the state quad (alphabet) resolution. It is apparent that a range of resolution and kernel width need to be tested to determine best performance give a generic supervised classifier. For these purposes a rapid initial classification algorithm, General Quadratic Discriminate Analysis [ @Duda2012 ], was implemented to estimate the mis - classification pace (incorrect decisions / total decisions). Not all states will be observed, i.e. the high dimensional feature vector will have information incorporate in a small subset of elements. Dimensionality decrease method are much necessary for implementation of classification algorithm, in particular QDA where the structure of a covariance matrix of a sparse feature space can be problematic. The decrease of the large, sparse, feature vector resulting from the unpacking of the Markov Matrix is performed via extended canonical random variable analysis or ECVA [ @Norgaard2006 ]. The methodology for ECVA has roots in principle component analysis (PCA). PCA is a procedure performed on bombastic multidimensional datasets with the intent of rotating what is a set of possibly correlated dimensions into a hardening of linearly uncorrelated variables [ @Sch06 ]. The transformation results in a dataset, where the first principle part (dimension) have the largest possible variability. PCA is an unsupervised methodology, i.e. a priori known label for the data being processed is not learn into consideration, therefore a reduction in feature dimensionality and while it maximizes the variance it might not maximize the analogue separability of the class space. In contrast to PCA, Canonical Variate Analysis does assume class label into considerations. The variation between groups is maximized resulting in a transformation that benefits the goal of separating classes. yield a set of datum $ \mathbf{x}$ with: $ g$ different classes, $ n_{i}$ observations of each course, and $ r\times r$ dimensions in each observation; following @Johnson1992, the within - group and between - group covariance matrix is defined as: $ $ \mathbf{S}_{within}=\frac{1}{n

of the transition matrix wlll result in a shape smgnaturs that ir too dependent on noise and tye ineividualness of specifkc wavefogm to be if aiy use. Thus addivjonal pvjcesalng iv necessary for further atalysis; even a soapl set of states (12 x 12) will result in a featirf vector with righ qimehsionality (144 dimensions). While a winsow and overlap size is assumed for the slottijg tl address the irrehular samplung jd the time sdries data, there are tso adjustable features associatdd wich this anaoywis: jhe kernel wmdth afsociated wibn the vlottinb and the statc spare (aophabet) resolution. It is apparent that w range ox xesolutions and kerneo qidth neeg to ve gesued tk detegmiie best perrormance gicen a generic superfifvc classifier. For trefe purposes a rapid initial classificatpon zlgorithm, General Quadrqtic Discriminate Anapysis [@Dudw2012], was implemented to estimate the mis-classificatimn rave (wrinn dezuslons/total decisions). Not all states will be obfsrfec, i.e. the high bimensional featirf fgctor will havg informatjon contained in a small fubser of elemtnts. Cimensionality reduction merhods are ofnen becessary for implzmentation oy clasxificstion algorithms, in parciculad QDA where the consffuction of a covxrisnwe matrix of a sparse feattre space can be proclemstic. Thq reductioj of bve large, sparse, fewture vactor resupting from the unpacking of the Markov Matrix iv pvrformed ria exbended canonicaj variate analisis or EEVA [@Nofgaard2006]. The methodmlogy for ESVA has roots ln principle componegt abalywis (PCA). OCA is a procecure performed on lqrge multidimensiokal dxfasets with the inuebt of rotating whxt ys a sqd of possiblf cofreusted aimensions lntu a xet of linearly uncosrelzted variables [@Sch06]. Tme transfirmation results in a dataset, where the firsv prinriple vomkonent (dimension) has the largesf possiblf vwriance. PCA ys ak unfupervised methodology, i.e. a priori known labels fir the data being processed is not taltn into consiveratijn, thus a reduction in featurw dimensionality snd while it maximizea the eariajce it might not maximize the linear separability of the class space. In cintrasv eo PCA, Canohicak Varpatz Aualysis doed take class labeks into considerations. The varietion betwaeu groups is maximized resultimg in a transfotmation that benefits the foal of xeparating classes. Given a set pf data $\mathbf{x}$ with: $g$ dlfferent xlassss, $n_{i}$ obsecvations of each claxs, and $r\timws r$ dinenslons in each ocsetvwtion; folpowiig @Lohnson1992, the within-group and betwewn-group fovariance matrie is defined xs: $$\matgbf{S}_{wijhin}=\frac{1}{n

of the transition matrix will result in signature is too on noise and to of any use. additional processing is for further analysis; even a small of states (12 x 12) will result in a feature vector with high (144 dimensions). While a window and overlap size is assumed for the slotting address irregular of time series data, there are two adjustable features associated with this analysis: the kernel width associated the slotting and the state space (alphabet) resolution. is apparent that a of resolutions and kernel width to tested to best given generic supervised classifier. these purposes a rapid initial classification algorithm, General Quadratic Discriminate Analysis [@Duda2012], was implemented to estimate the rate (wrong Not all will observed, the high dimensional will have information contained in a elements. Dimensionality reduction methods are often necessary for of classification in particular QDA where the construction a covariance matrix of a sparse feature space be problematic. The reduction of the large, sparse, feature vector resulting from the unpacking of Matrix is performed via canonical variate analysis ECVA The for has roots principle component analysis (PCA). PCA is a procedure performed on large datasets with the intent of rotating what is a set correlated into a set linearly uncorrelated variables [@Sch06]. transformation in a dataset, where principle (dimension) possible PCA an unsupervised methodology, i.e. priori known labels for the being processed is not reduction in feature dimensionality and while it maximizes variance it might not maximize the linear of the class space. In contrast to PCA, Canonical Variate Analysis does class labels The variation between groups is maximized resulting in transformation that benefits the of separating classes. Given a set of data $\mathbf{x}$ $g$ classes, $n_{i}$ of each class, $r\times r$ dimensions each observation; following within-group and covariance is

of the transition matrix will Result in a sHape sIgnAtuRe That Is toO dependent on noISe anD the individualness of spEcifiC wAVefoRM tO be of Any use. THUs ADDitIoNaL prOcESsIng is NecEssary fOr further aNalYsIs; even a small SEt Of states (12 x 12) wIll Result in a feaTurE vectoR wIth HIgh diMenSionaLity (144 diMEnsionS). While a wiNdOW and ovERlap sizE IS aSsumEd for the slotting tO AdDRess the irregulAr sampLiNG oF THe tIme Series data, ThEre arE Two adjuSTaBLE FeaTUres associateD with this anALysIs: the kErNel WIdth asSociaTeD WitH the slottinG and The state sPace (alPHabet) reSOlution. it is apParEnt That A RaNgE of ReSOluTIoNs aND keRnel widtH nEeD to be TestED TO DeteRmiNe beSt perFormance given A geNeriC SupErvisEd claSsifIeR. For tHese puRposeS a Rapid initial claSsifIcation alGorItHm, GEnEral QUAdratiC DiScrIminate analysiS [@dudA2012], wAS IMpLemented to estimate ThE MIs-ClassifiCation RAtE (wROng decisIoNs/tOtal DECisioNs). NoT AlL states wIll be oBSeRvEd, i.e. the HiGh dimeNsIonAl fEaturE VectOr will Have infoRmatiON contained in a sMAll subset of elEMeNTS. DIMensIonAlity reductIon mEThodS are OFtEn nECessaRy for ImPLeMEntation of classificAtIon algOrithMs, in particulaR QDA where tHE COnstructIon oF A cOVariance matrix Of a spArse featurE Space can Be proBlematic. the reductION of the laRge, SpaRse, FeaTURe Vector resultiNG From ThE unpackIng Of the MaRkoV MaTriX is PeRformed viA extendeD cAnOnIcAl vAriatE Analysis Or eCVa [@NOrgAard2006]. THE methoDologY for eCvA HAs rOots in pRInCIPle cOmPoNent AnaLySis (PCa). PCA IS a pRocedurE performeD on LArge MuLtIdimensIonal datasets WiTh the intenT oF roTating WHAt is a set Of possibly correlated dimENsions iNto A set oF linEarly uncoRreLated vAriABles [@ScH06]. The trAnsfoRmAtiON ResulTS In A daTaSet, where thE FIrsT prinCiPle cOmponenT (dimension) has the laRGesT possible variAncE. PCA IS An UnsUPeRVisEd MEthODOlogy, i.e. a priori kNown labels FoR ThE data being PRocEsSed is noT taken iNto coNSideratIon, thus a rEduction iN fEatuRE DimEnsionalitY and whilE it maximiZEs the VArIance It mIght noT mAxiMize tHe lineAR seParabIlity oF tHe clasS spacE. IN contrasT to PCA, Canonical Variate ANalysiS does TakE class labEls INto ConsideraTionS. The variatIon BetWeen gRouPS is maXimiZEd ResULting In a tRAnsformatIOn ThaT BEnEfits the goaL OF SepAratiNg cLAsses. GIven A set of data $\mathbf{x}$ WIth: $g$ different cLassES, $N_{i}$ oBseRVatiOnS of each class, anD $r\tImES R$ dimensiOnS in each obseRvation; fOlLOwing @johnsoN1992, the wiThin-groUP AnD BetweeN-groUp cOvariance MatRiX Is definEd As: $$\MAthbf{S}_{WithIn}=\Frac{1}{n

of the transition matrixwill resul t ina s hap esign atur e that is tood epen dent on noise and theindiv id u alne s sof sp ecificw av e f orm t obeof an y use . T hus add itional pr oce ss ing is neces s ar y for furt her analysis; e ven a sma ll se t of s tat es (1 2 x 12 ) willresult in a featur e vector w it h hi gh dimensionality (1 4 4 dimensions). While a wi n d owand overlap s iz e isa ssumedf or t h e s l otting to add ress the ir r egu lar sa mp lin g of th e tim es eri es data, th ereare two a djusta b le feat u res ass ociate d w ith thi s a na lys is : th e k ern e l w idth ass oc ia ted w itht h e slot tin g an d the state space(al phab e t)resol ution . It i s app arentthatarange of resolu tion s and ker nel w idt hneedt o be t est edto dete rmine b e stpe r f o rm ance given a gener ic s up ervisedclassi f ie r. For thes epur pose s a rap id i n it ial clas sifica t io nalgorit hm , Gene ra l Q uad ratic Disc rimina te Analy sis [ @ Duda2012], was implemented t o e s t im a te t hemis-classif icat i on r ate( wr ong decis ions/ to t al decisions). Not all s tateswillbe observed,i.e. the h i g h dimensi onal fe a ture vector wi ll ha ve informa t ion cont ained in a sm all subse t of eleme nts . D ime nsi o n al ity reduction m etho ds are of ten necess ary fo r i mpl em entationof class if ic at io n a lgori t hms, inpa rti cu lar QDAw here t he co nstr uc ti o n o f a cov a ri a n ce m at ri x of asp arsefeat u respace c an be pro ble m atic . T he redu ction of thela rge, spars e, fe aturev e ctor res ulting from the unpacki n g of th e M arkov Mat rix is pe rfo rmed v iae xtende d cano nical v ari a t e ana l y si s o rECVA [@Nor g a ard 2006] .Themethodo logy for ECVA hasr oot s in principl e c ompo n e nt an a ly s is(P C A). P CA is a procedu re perform ed on large mul t idi me nsional datase ts wi t h the i ntent ofrotatingwh at i s a s et of poss ibly cor related d i mensi o ns into aset of l ine arlyuncorr e lat ed va riable s[@Sch0 6]. T he transfo rmation results in a da taset, wher e t he firstpri n cip le compon ent(dimension ) h asthe l arg e st po ssib l evar i ance. PCA is an uns u pe rvi s e dmethodology , i .e. a pr ior i known lab els for the datab eing processed isn o t t ake n int oconsideration, th us a reducti on in feature dimensi on a lityand wh ile it maximi z e st he var ianc e i t might n otma x imize t he l i near s epar ab ilityof the clas s space. In contra st to P CA, C a non icalVa riate A n alys is does ta ke class la bels i ntoconsi deratio ns . Thevar ia tion betwe e n groupsis ma ximized r esul tin g in a tra n s forma tion t hat benefits t he go al ofsepa ratin gclas ses. Give n a set o f d a ta $\ma th bf{ x } $ with : $ g $ different cl asses , $n_{i}$ ob s erva t io n s ofeach c lass,and $r\ t ime sr$ dime nsi o n s in each observat i on; f ollo wing @Jo hn son1 99 2,th e within-groupand betw een-grou p covari anc e mat r ix isde fined as: $ $\mat hb f {S}_{wi t h in} =\frac{ 1 }{n

of_the transition_matrix will result in_a shape_signature_that is_too_dependent on noise_and the individualness_of specific waveform to_be of any_use._Thus additional processing is necessary for further analysis; even a small set of states_(12_x 12)_will_result_in a feature vector with_high dimensionality (144 dimensions). While_a window_and overlap size is assumed for the slotting_to_address the irregular_sampling of the time series data, there are two_adjustable features associated with this analysis:_the kernel width_associated_with_the slotting and the_state space (alphabet) resolution. It is_apparent that a range of resolutions_and kernel width need to be tested_to determine best performance given a_generic supervised classifier. For these_purposes a_rapid initial classification algorithm, General_Quadratic Discriminate Analysis_[@Duda2012], was_implemented to estimate_the mis-classification rate (wrong decisions/total decisions)._Not all states_will be observed, i.e. the high_dimensional_feature vector will_have_information_contained in_a small subset_of_elements. Dimensionality_reduction_methods are often necessary for implementation_of_classification algorithms, in particular QDA where the_construction of a covariance_matrix_of a sparse feature_space can be problematic. The reduction_of the large, sparse, feature vector_resulting from_the unpacking_of the Markov Matrix is performed via extended canonical variate analysis_or ECVA [@Norgaard2006]. The methodology for_ECVA has roots in_principle component_analysis_(PCA). PCA is_a_procedure performed_on large multidimensional datasets with the intent_of rotating_what is a set of possibly_correlated dimensions into a_set_of linearly uncorrelated variables [@Sch06]. The_transformation results in a dataset, where_the first principle component (dimension)_has_the_largest possible variance. PCA is_an unsupervised methodology, i.e. a priori_known labels for_the data being processed is not taken_into_consideration, thus a reduction in feature_dimensionality_and while it maximizes the variance_it_might_not maximize the linear separability_of the class space. In contrast_to PCA, Canonical Variate Analysis does take class labels_into considerations. The_variation between groups is maximized_resulting_in_a transformation that benefits the goal of separating classes. Given_a set_of data $\mathbf{x}$_with: $g$ different classes, $n_{i}$ observations of each class, and_$r\times r$ dimensions in each observation; following_@Johnson1992, the within-group and between-group covariance matrix is defined as: $$\mathbf{S}_{within}=\frac{1}{n

second generation of stars formed from material shed by stars of the first generation. Our results might suggest a different scenario. It is the cooler eMSTO objects, spectroscopically determined to be fast rotators, that dominate within 1 arcmin of the cluster core, although we have a small sample. Perhaps this is typical of less massive and/or younger clusters. Yet, it is puzzling that the rapidly rotating stars are concentrated towards the cluster center (Milone et al 2017) where it might be expected that a second stellar generation would form from the material of the first generation. This would appear to suggest that another scenario must be sought for young clusters. Acknowledgments =============== We thank the anonymous referee for thoughtful comments and advice on the manuscript. AFM and APM acknowledge support by the Australian Research Council through Discovery Early Career Researcher Awards DE160100851 and DE150101816. EWO was partially supported by NSF Grant AST1313006. This research has made use of NASA’s Astrophysics Data System Bibliographic Services. And we have used data products from 2MASS, which is a joint project of the University of Massachusetts and IPAC/Caltech, funded by NASA and the NSF. Bastian, N., Cabrera-Ziri, I., Niederhofer, F. [et al.]{}  2017,, 465, 4795 Bastian, N., Cabrera-Ziri, I., & Salaris, M. 2015,, 449, 3333 Bastian, N., & deMink, S. E. 2009,, 398, L11 Bastian, N., Niederhofer, F., Kozhurina-Platais, V. [et al.]{} 2016,, 460, L20 Bedin, L. R., Piotto, G., Anderson, J. [et al.]{} 2004, ApJ, 605, L125 Charbonnel, C. 2016, EAS Publication Series, 80-81, 177 Choi, J., Dotter, A., Conroy, C. [et al.]{}  2016,, 823, 102 Conroy, C., & Spergel, D. N. 2011,, 726, 36 Correnti, M., Goudfrooij, P., Kalirai, J. S. [et al.]{} 2014,, 793, 121 D’Antona, F., DiCriscienzo, M., Dec

second generation of stars formed from material spill by star of the first generation. Our results might hint a different scenario. It is the cooler eMSTO object, spectroscopically compulsive to be fast rotators, that dominate within 1 arcmin of the cluster core, although we have a little sample. Perhaps this is distinctive of less massive and/or younger clusters. Yet, it is puzzling that the rapidly rotating stars are concentrate towards the cluster center (Milone et al 2017) where it might be expected that a second leading generation would form from the fabric of the first genesis. This would appear to hint that another scenario must be sought for young clusters. citation = = = = = = = = = = = = = = = We thank the anonymous referee for thoughtful comments and advice on the manuscript. AFM and APM acknowledge support by the Australian Research Council through Discovery Early Career Researcher Awards DE160100851 and DE150101816. EWO was partially corroborate by NSF Grant AST1313006. This research has made use of NASA ’s Astrophysics Data System Bibliographic Services. And we have used data products from 2MASS, which is a joint project of the University of Massachusetts and IPAC / Caltech, funded by NASA and the NSF. Bastian, N., Cabrera - Ziri, I., Niederhofer, F. [ et al. ] { }   2017, , 465, 4795 Bastian, N., Cabrera - Ziri, I., & Salaris, M. 2015, , 449, 3333 Bastian, N., & deMink, S. E. 2009, , 398, L11 Bastian, N., Niederhofer, F., Kozhurina - Platais, V. [ et al. ] { }   2016, , 460, L20 Bedin, L. R., Piotto, G., Anderson, J. [ et al. ] { }   2004, ApJ, 605, L125 Charbonnel, C. 2016, EAS Publication Series, 80 - 81, 177 Choi, J., Dotter, A., Conroy, C. [ et al. ] { }   2016, , 823, 102 Conroy, C., & Spergel, D. N. 2011, , 726, 36 Correnti, M., Goudfrooij, P., Kalirai, J. S. [ et al. ] { }   2014, , 793, 121 D’Antona, F., DiCriscienzo, M., Dec

sefond generation of stars formed from majeeial sied by atars of the first generation. Our reduots mught suggest a differevt scenarpo. It is rhe rooler eMSTO objxdts, spegcrosckiicalnb determined to be fast rmtators, that dmmkncte within 1 arcmin of the cluster cowe, althpuhh we have a siall fampmv. Kerhaps this is typical of less mzssive end/or younger ckusters. Yet, it is puzzling thah the rapidly rotahing stars qre sincentrated gowards tht eluster cenjer (Milone et al 2017) where it might be erpected thaj z dgcond stellac genegation would nprm frmm the katerial of thc firvt teneration. This would appear to suggest that anodhzr scenario must be siutht fmr ymung xlurteds. Eckhowledhmeits =============== We thank the anonymius referee for thoigrngul comments and aqvyce on the manuscript. AFM and APM acknoflesge support by the Austealian Research Councll througr Discovery Early Career Researcher Awards DE160100851 and DE150101816. EXO waw iartkqlpy supported by NSF Grant AST1313006. This research hwa kace use of NASA’f Astrophysocd Cwta System Bicliogrclhjc Services. And we have ufed dqta produsts grom 2MASS, which is a joint project of tke Yniversity of Masscchusetts anb IPAC/Valtevh, funded by NASA and tke NSF. Gastian, N., Cwbrera-Zirj, I., Niederhofer, F. [et an.]{}  2017,, 465, 4795 Basukan, N., Cabrera-Ziri, I., & Salarms, M. 2015,, 449, 3333 Bastkan, M., & deMynk, S. E. 2009,, 398, P11 Basbhan, N., Niederhofer, V., Kozkurind-Platais, V. [et al.]{} 2016,, 460, L20 Bedin, L. R., Piotto, G., Anvxrson, J. [et al.]{} 2004, A[J, 605, L125 Charbjnnel, C. 2016, EAS Publisation Series, 80-81, 177 Choi, L., Dottdr, A., Conroj, C. [et al.]{}  2016,, 823, 102 Conroy, S., & Spergel, D. T. 2011,, 726, 36 Correnti, M., Goudfwooih, P., Jalirai, G. S. [et al.]{} 2014,, 793, 121 D’Amtona, F., DpCxiscienzo, M., Dec

second generation of stars formed from material stars the first Our results might is cooler eMSTO objects, determined to be rotators, that dominate within 1 arcmin the cluster core, although we have a small sample. Perhaps this is typical less massive and/or younger clusters. Yet, it is puzzling that the rapidly rotating are towards cluster (Milone et al 2017) where it might be expected that a second stellar generation would form the material of the first generation. This would to suggest that another must be sought for young Acknowledgments We thank anonymous for comments and advice the manuscript. AFM and APM acknowledge support by the Australian Research Council through Discovery Early Career Researcher DE160100851 and was partially by Grant This research has of NASA’s Astrophysics Data System Bibliographic have used data products from 2MASS, which is joint project the University of Massachusetts and IPAC/Caltech, by NASA and the NSF. Bastian, N., Cabrera-Ziri, Niederhofer, F. [et al.]{} 2017,, 465, 4795 Bastian, N., Cabrera-Ziri, I., & Salaris, M. 2015,, Bastian, N., & deMink, E. 2009,, 398, Bastian, Niederhofer, Kozhurina-Platais, [et al.]{} 460, L20 Bedin, L. R., Piotto, G., Anderson, J. [et al.]{} ApJ, 605, L125 Charbonnel, C. 2016, EAS Publication Series, 80-81, J., A., Conroy, C. al.]{} 2016,, 823, 102 C., Spergel, D. N. 2011,, Correnti, Goudfrooij, S. al.]{} 793, 121 D’Antona, F., M., Dec

second generation of stars foRmed from maTeriaL shEd bY sTars Of thE first generatiON. Our Results might suggest a diFfereNt SCenaRIo. it is tHe cooleR EMsto obJeCtS, spEcTRoScopiCalLy deterMined to be fAst RoTators, that doMInAte within 1 aRcmIn of the clustEr cOre, altHoUgh WE have A smAll saMple. PeRHaps thIs is typicAl OF less mASsive anD/OR yOungEr clusters. Yet, it is PUzZLing that the rapIdly roTaTInG STarS arE concentraTeD towaRDs the clUStER CEntER (Milone et al 2017) whEre it might bE ExpEcted tHaT a sECond stEllar GeNEraTion would foRm frOm the mateRial of THe first GEneratiOn. This WouLd aPpeaR To SuGgeSt THat ANoTheR SceNario musT bE sOught For yOUNG ClusTerS. AckNowleDgments =============== We thanK thE anoNYmoUs refEree fOr thOuGhtfuL commeNts anD aDvice on the manusCripT. AFM and APm acKnOwlEdGe supPOrt by tHe AUstRalian REsearch cOunCiL THRoUgh Discovery Early CArEER REsearcheR AwardS dE160100851 AnD dE150101816. EWO was PaRtiAlly SUPportEd by nsF grant AST1313006. this reSEaRcH has madE uSe of NAsA’S AsTroPhysiCS DatA SysteM BibliogRaphiC services. And we hAVe used data proDUcTS FrOM 2MASs, whIch is a joint ProjECt of The UNIvErsITy of MAssacHuSEtTS and IPAC/Caltech, fundEd By NASA And thE NSF. Bastian, N., CAbrera-Ziri, i., nIEderhofeR, F. [et AL.]{}  2017,, 465, 4795 BAStian, N., Cabrera-ZIri, I., & SAlaris, M. 2015,, 449, 3333 BasTIan, N., & deMiNk, S. E. 2009,, 398, L11 bastian, N., niederhofER, f., KozhuriNa-PLatAis, v. [et AL.]{} 2016,, 460, l20 BEdin, L. R., Piotto, G., aNDersOn, j. [et al.]{} 2004, Apj, 605, L125 CHarbonnEl, C. 2016, eAS pubLicAtIon Series, 80-81, 177 choi, J., DotTeR, A., coNrOy, C. [Et al.]{}  2016,, 823, 102 CONroy, C., & SpeRgEl, D. n. 2011,, 726, 36 COrrEnti, M., gOudfroOij, P., KAlirAi, j. S. [ET al.]{} 2014,, 793, 121 d’Antona, f., dicRIsciEnZo, m., Dec

second generation of star s formed f rom m ate ria lshed bystars of the f i rstgeneration. Our result s mig ht sugg e st a di fferent sc e n ari o. I t i st he cool ereMSTO o bjects, sp ect ro scopically d e te rmined tobefast rotator s,that d om ina t e wit hin 1 ar cmin o f the c luster co re , altho u gh we h a v ea sm all sample. Perha p st his is typical of le ss ma s s ive an d/or young er clus t ers. Ye t ,i t isp uzzling thatthe rapidly rot atingst ars are co ncent ra t edtowards the clu ster cent er (Mi l one eta l 2017) where it mi ghtb eex pec te d th a ta s e con d stella rge nerat ionw o u l d fo rmfrom thematerial of t hefirs t ge nerat ion.This w ouldappear to s ug gest that anoth er s cenario m ust b e s ou ght f o r youn g c lus ters. Acknowl e dgm en t s == ============= Weth a n kthe anon ymousr ef er e e for th ou ght fulc o mment s an d a dvice on the m a nu sc ript. A FM and A PM ac kno wledg e sup port b y the Au stral i an Research Co u ncil throughD is c o ve r y Ea rly Career Res earc h er A ward s D E16 0 10085 1 and D E 15 0 101816. EWO was par ti ally s uppor ted by NSF Gr ant AST131 3 0 0 6. Thisrese a rc h has made useof NA SA’s Astro p hysics D ata S ystem Bi bliograph i c Service s.And we ha v e u sed data prod u c ts f ro m 2MASS , w hich is ajoi ntpro je ct of the Univers it yof M ass achus e tts andIP AC/ Ca lte ch, f u nded b y NAS A an dth e NS F. Bas t ia n , N., C ab rera -Zi ri , I., Nie d erh ofer, F . [et al. ]{} 201 7, ,465, 47 95 Bastian, N ., Cabrera-Z ir i,I., &S a laris, M . 2015,, 449, 3333 Bast i an, N., &deMin k, S . E. 2009 ,,398, L 11B astian , N.,Niede rh ofe r , F.,K o zh uri na -Platais,V . [e t al. ]{ } 20 16,, 46 0, L20 Bedin, L. R . , P iotto, G., An der son, J .[et al . ]{}  2 0 04, A pJ, 605, L125 C harbonnel, C . 2 016, EAS P u bli ca tion Se ries, 8 0-81, 177 Cho i, J., Do tter, A., C onro y , C. [et al.]{ }  2016, , 823, 10 2 Conr o y, C.,& S pergel ,D.N. 20 11,, 7 2 6,36 Co rrenti ,M., Go udfro oi j, P., K alirai, J. S. [et al.]{ } 2014 ,, 79 3,121 D’Ant ona , F. , DiCrisc ienz o, M., Dec

second_generation of_stars formed from material_shed by_stars_of the_first_generation. Our results_might suggest a_different scenario. It is_the cooler eMSTO_objects,_spectroscopically determined to be fast rotators, that dominate within 1 arcmin of the cluster_core,_although we_have_a_small sample. Perhaps this is_typical of less massive and/or_younger clusters._Yet, it is puzzling that the rapidly rotating_stars_are concentrated towards_the cluster center (Milone et al 2017) where it_might be expected that a second_stellar generation would_form_from_the material of the_first generation. This would appear to_suggest that another scenario must be_sought for young clusters. Acknowledgments =============== We thank the anonymous_referee for thoughtful comments and advice_on the manuscript. AFM and_APM acknowledge_support by the Australian Research_Council through Discovery_Early Career_Researcher Awards DE160100851_and DE150101816. EWO was partially supported_by NSF Grant_AST1313006. This research has made use_of_NASA’s Astrophysics Data_System_Bibliographic_Services. And_we have used_data_products from_2MASS,_which is a joint project of_the_University of Massachusetts and IPAC/Caltech, funded by_NASA and the NSF. Bastian,_N.,_Cabrera-Ziri, I., Niederhofer, F._[et al.]{}  2017,, 465, 4795_Bastian, N., Cabrera-Ziri, I., & Salaris,_M. 2015,,_449, 3333_Bastian, N., & deMink, S. E. 2009,, 398, L11 Bastian, N.,_Niederhofer, F., Kozhurina-Platais, V. [et al.]{} 2016,,_460, L20 Bedin, L._R., Piotto,_G.,_Anderson, J. [et_al.]{} 2004,_ApJ, 605,_L125 Charbonnel, C. 2016, EAS Publication Series,_80-81, 177_Choi, J., Dotter, A., Conroy, C._[et al.]{}  2016,, 823,_102_Conroy, C., & Spergel, D. N._2011,, 726, 36 Correnti, M., Goudfrooij,_P., Kalirai, J. S. [et_al.]{} 2014,,_793,_121 D’Antona, F., DiCriscienzo, M.,_Dec

mpg.de/$\sim$jcg/grbgen.html [^2]: http://webast.ast.obs-mip.fr/hyperz/ [^3]: http://bat.phys.unsw.edu.au/$\sim$fsoto/hdfcat.html --- abstract: 'Pairs of pseudoscalar neutral mesons from decays of vector resonances are studied as bipartite systems in the framework of density operator. Time-dependent quantum entanglement is quantified in terms of the entanglement entropy and these dependences are demonstrated on data on correlated pairs of $\PK$ and $\PB$ mesons, as measured by the KLOE and Belle experiments. Another interesting characteristics of such bipartite systems are moments of the CP distributions. These moments are directly measurable and they appear to be very sensitive to the initial degree of entanglement of a pair.' address: 'National Centre for Nuclear Research, Warsaw, Poland' author: - Wojciech Wiślicki title: 'Entanglement, fluctuations and discrete symmetries in particle decays [^1] ' --- Introduction ============ Interferometry of neutral mesons is recognized as a powerful and sensitive tool for testing fundamentals of the quantum mechanics. In particular, pairs of pseudoscalar mesons, originating from strong decays of vector resonances, were proved to be particularly useful in many precision measurements due to their well-defined initial state’s quantum numbers and relatively high production rates. The decays $\Pphi(1020)\rightarrow \PKl\PKs$, $\Ppsi(3770)\rightarrow \PDzero\APDzero$, $\PUpsilon(10580)\rightarrow \PBd\APBd$ and $\PUpsilon(10860)\rightarrow \PBs\APBs$ were recognized long time ago as very useful for the study of the CP violation [@branco], validity of the CPT and Lorentz invariance [@cpt] or search for quantum decoherence [@frascati]. On the other hand, in case of complex initial states where the meson source does neither have well-defined quantum numbers nor symmetry properties and can be a spatially extended object, as e.g. in nuclear collisions, meson interferometry is often used to study both the spatio-temporal characteristics and the degree of coherence of the source. In any case, either the resonance with well-defined symmetry or a nuclear fireball, the meson pair can be considered as a quantum bipartite system with an

mpg.de/$\sim$jcg/grbgen.html [ ^2 ]: http://webast.ast.obs-mip.fr/hyperz/ [ ^3 ]: http://bat.phys.unsw.edu.au/$\sim$fsoto/hdfcat.html --- abstract:' Pairs of pseudoscalar neutral mesons from decays of vector plangency are study as bipartite systems in the framework of density hustler. Time - dependent quantum entanglement is quantify in terms of the entanglement entropy and these dependence are demonstrated on data on correlate pairs of $ \PK$ and $ \PB$ mesons, as quantify by the KLOE and Belle experiment. Another interesting characteristics of such bipartite systems are moments of the CP distributions. These moments are immediately measurable and they appear to be very sensitive to the initial degree of web of a pair.' address:' National Centre for Nuclear Research, Warsaw, Poland' author: - Wojciech Wiślicki title:' Entanglement, fluctuation and discrete symmetries in particle decays [ ^1 ]' --- Introduction = = = = = = = = = = = = Interferometry of neutral meson is recognized as a powerful and sensitive tool for testing fundamentals of the quantum automobile mechanic. In particular, pairs of pseudoscalar mesons, originating from strong decays of vector resonances, were proved to be particularly useful in many preciseness measurement due to their well - defined initial state ’s quantum number and relatively high production rates. The decay $ \Pphi(1020)\rightarrow \PKl\PKs$, $ \Ppsi(3770)\rightarrow \PDzero\APDzero$, $ \PUpsilon(10580)\rightarrow \PBd\APBd$ and $ \PUpsilon(10860)\rightarrow \PBs\APBs$ were acknowledge long time ago as very utilitarian for the study of the CP violation [ @branco ], validity of the CPT and Lorentz invariance [ @cpt ] or search for quantum decoherence [ @frascati ]. On the other hand, in case of complex initial states where the meson source does neither have well - defined quantum number nor isotropy property and can be a spatially extended object, as e.g. in nuclear collisions, meson interferometry is often used to study both the spatio - worldly characteristics and the academic degree of coherence of the source. In any case, either the resonance with well - defined symmetry or a nuclear powerhouse, the meson pair can be considered as a quantum bipartite arrangement with an

mpg.fe/$\sim$jcg/grbgen.html [^2]: http://wtbast.ast.obs-mip.fr/hipwrz/ [^3]: htvp://bat.phgs.unsw.edj.au/$\sim$fsoto/hdfcat.html --- abstrart: 'Pqirs if pseudoscalar neutrau mesons vrom decqys id vector rxaonances are abudieb es bipartite syxtems in tve framework ox aeusity operator. Time-dependent quantum entangkelent is quantisied yn tsgmw of the entanglement entropy and thtse dependences arr demonstrated on data on forrflated pairs of $\PK$ and $\PB$ mesins, ww measured bh the KLOE and Belle esperiments. Another interesting zharaeteristics if sufv bipartite systvms are momenbx of tve CP dostributions. Tmese komwnts are directly meavurable and they akpear to ba rery sensitive to the ibitian decree if dntznjlejent ov a pair.' addreas: 'National Centre for Nuclear Rqwearch, Warsaw, Polanq' atthor: - Wojciech Wiślicki title: 'Entanglemenu, fludtuations and discrete wymmetries in particlg decays [^1] ' --- Yntroduction ============ Interferometry of neutral mesons is racognmzdd cw a puqegful and sensitive tool for testing fundamentwms on the quantum meghanics. In particukag, lwirs of pseuduscalax mssons, originating vrom sttong dwcays of dectpr resonances, were proved ti be particujqrly useful in manv precision oeasoremenys due to their well-defnned ihitial statf’s quantuj numbers and relxtinely high production rates. The decays $\P'hi(1020)\riyhtarrow \PKl\KKs$, $\Ppsy(3770)\rightarroa \PDzcso\APDzero$, $\PUpsilon(10580)\gightcrrow \PBd\APBd$ ajd $\PUpsilon(10860)\rightarrow \PBs\APBs$ wxce recognized lmng time agj as yery useful for the study of jhe CP vijlatiun [@branco], nalidity mf the CPT wnd Lorentz itgariance [@cpt] or searsh fir qyantum adcoherence [@fraxcati]. On nht other habd, in case of compkex jnitial states cketw the meson soircd djed ieithqs have well-dafindd diantuo numbers nif sykmetry properties ang cah be a spatially ectcnded objgct, as e.g. in nuclear cpllisions, meson inuerfermmevry is ofjen used to study both the spatjo-temporap cmaracteristicf ane the degree of coherence of the source. In any case, either the resonabce with well-defineb xymmetry or a nuslear firaball, the meson pair can be considerec as a quantum bipartjte syvtem aith an

mpg.de/$\sim$jcg/grbgen.html [^2]: http://webast.ast.obs-mip.fr/hyperz/ [^3]: http://bat.phys.unsw.edu.au/$\sim$fsoto/hdfcat.html --- abstract: pseudoscalar mesons from of vector resonances in framework of density Time-dependent quantum entanglement quantified in terms of the entanglement and these dependences are demonstrated on data on correlated pairs of $\PK$ and mesons, as measured by the KLOE and Belle experiments. Another interesting characteristics of bipartite are of CP distributions. These moments are directly measurable and they appear to be very sensitive to the degree of entanglement of a pair.' address: 'National for Nuclear Research, Warsaw, author: - Wojciech Wiślicki title: fluctuations discrete symmetries particle [^1] --- Introduction ============ of neutral mesons is recognized as a powerful and sensitive tool for testing fundamentals of the quantum In particular, pseudoscalar mesons, from decays vector resonances, were be particularly useful in many precision their well-defined initial state’s quantum numbers and relatively production rates. decays $\Pphi(1020)\rightarrow \PKl\PKs$, $\Ppsi(3770)\rightarrow \PDzero\APDzero$, $\PUpsilon(10580)\rightarrow and $\PUpsilon(10860)\rightarrow \PBs\APBs$ were recognized long time ago very useful for the study of the CP violation [@branco], validity of the CPT and [@cpt] or search for decoherence [@frascati]. On other in of initial states the meson source does neither have well-defined quantum numbers nor symmetry and can be a spatially extended object, as e.g. in meson is often used study both the spatio-temporal and degree of coherence of In case, with symmetry a nuclear fireball, the pair can be considered as quantum bipartite system with

mpg.de/$\sim$jcg/grbgen.html [^2]: http://Webast.ast.oBs-mip.Fr/hYpeRz/ [^3]: Http://Bat.pHys.unsw.edu.au/$\siM$FsotO/hdfcat.html --- abstract: 'PaiRs of pSeUDoscALaR neutRal mesoNS fROM deCaYs Of vEcTOr ResonAncEs are stUdied as bipArtItE systems in thE FrAmework of dEnsIty operator. TIme-DependEnT quANtum eNtaNglemEnt is qUAntifiEd in terms Of THe entaNGlement ENTrOpy aNd these dependenceS ArE Demonstrated on Data on CoRReLATed PaiRs of $\PK$ and $\Pb$ mEsons, AS measurED bY THE KLoe and Belle expeRiments. AnotHEr iNteresTiNg cHAracteRistiCs OF suCh bipartite SystEms are momEnts of THe CP disTRibutioNs. ThesE moMenTs arE DiReCtlY mEAsuRAbLe aND thEy appear To Be Very sEnsiTIVE To thE inItiaL degrEe of entanglemEnt Of a pAIr.' aDdresS: 'NatiOnal ceNtre fOr NuclEar ReSeArch, Warsaw, PolanD' autHor: - WojcieCh WIśLicKi Title: 'eNtanglEmeNt, fLuctuatIons and DIscReTE SYmMetries in particle dEcAYS [^1] ' --- INtroductIon ============ IntERfErOMetry of nEuTraL mesONS is reCognIZeD as a poweRful anD SeNsItive toOl For tesTiNg fUndAmentALs of The quaNtum mechAnics. iN particular, paiRS of pseudoscalAR mESOnS, OrigInaTing from strOng dECays Of veCToR reSOnancEs, werE pROvED to be particularly usEfUl in maNy preCision measureMents due to THEIr well-deFineD InITial state’s quanTum nuMbers and reLAtively hIgh prOduction Rates. The dECAys $\Pphi(1020)\rIghTarRow \pKl\pkS$, $\PPsi(3770)\rightarrow \pdZero\aPdzero$, $\PUPsiLon(10580)\righTarRow \pBd\aPBD$ aNd $\PUpsiloN(10860)\rightarRoW \Pbs\aPbs$ wEre reCOgnized lOnG tiMe Ago As verY Useful For thE stuDy Of THe Cp violatIOn [@BRAnco], VaLiDity Of tHe cPT anD LorENtz InvariaNce [@cpt] or sEarCH for QuAnTum decoHerence [@frascaTi]. on the other HaNd, iN case oF COmplex inItial states where the mesoN Source dOes NeithEr haVe well-defIneD quantUm nUMbers nOr symmEtry pRoPerTIEs and CAN bE a sPaTially exteNDEd oBject, As E.g. in Nuclear Collisions, meson intERfeRometry is ofteN usEd to STUdY boTH tHE spAtIO-teMPOral characterisTics and the DeGReE of coherenCE of ThE source. in any caSe, eitHEr the reSonance wiTh well-defInEd syMMEtrY or a nucleaR firebalL, the meson PAir caN Be ConsiDerEd as a qUaNtuM bipaRtite sYSteM with An

mpg.de/$\sim$jcg/grbgen.ht ml [^2]:http: //w eba st .ast .obs -mip.fr/hyperz / [^ 3]: http://bat.phys.un sw.ed u. a u/$\ s im $fsot o/hdfca t .h t m l -- -abs tr a ct : 'Pa irs of pse udoscalarneu tr al mesons fr o mdecays ofvec tor resonanc esare st ud ied as bi par titesystem s in th e framewo rk of den s ity ope r a to r. T ime-dependent qua n tu m entanglementis qua nt i fi e d in te rms of the e ntang l ement e n tr o p y an d these depend ences are d e mon strate dond ata on corr el a ted pairs of $ \PK$ and $\PB $ meso n s, as m e asuredby the KL OEandB el le ex pe r ime n ts . A n oth er inter es ti ng ch arac t e r i stic s o f su ch bi partite syste msarem ome nts o f the CPdi strib utions . The se moments are di rect ly measur abl eand t hey a p pear t o b e v ery sen sitivet o t he i n it ial degree of enta ng l e me nt of apair.' ad dr e ss: 'Nat io nal Cen t r e for Nuc l ea r Resear ch, Wa r sa w, Poland 'author :- W ojc iechW iśli cki ti tle: 'En tangl e ment, fluctuat i ons and discr e te s ym m etri esin particle dec a ys [ ^1]' - --Intro ducti on == = ========= Interfer om etry o f neu tral mesons i s recogniz e d as a pow erfu l a n d sensitive to ol fo r testingf undament als o f the qu antum mec h a nics. In pa rti cul ar, p ai rs of pseudos c a larme sons, o rig inating fr omstr ong d ecays ofvector r es on an ce s,werep roved to b e p ar tic ularl y usefu l inmany p re c isi on meas u re m e ntsdu eto t hei rwell- defi n edinitial state’squa n tumnu mb ers and relatively h ig h producti on ra tes. T h e decays$\Pphi(1020)\rightarrow \PKl\PK s$, $\Pp si(3 770)\righ tar row \P Dze r o\APDz ero$,$\PUp si lon ( 1 0580) \ r ig hta rr ow \PBd\AP B d $ a nd $\ PU psil on(1086 0)\rightarrow \PBs \ APB s$ were recog niz ed l o n gtim e a g o a sv ery u seful for the s tudy of th eC Pviolation[ @br an co], va lidityof th e CPT an d Lorentz invarian ce [@c p t ] o r search f or quant um decohe r ence[ @f rasca ti] . On t he ot her h and, i n ca se of compl ex initi al st at es where the meson source doesneithe r hav e w ell-defin edq uan tum numbe rs n or symmetr y p rop ertie s a n d can bea s pat i allyexte n ded objec t ,ase . g. in nuclear c o lli sions , m e son in terf erometry is often used to studyboth t hespa t io-t em poral characte ris ti c s and the d egree of co herenceof the s ource. In an y case, e it h er the res ona nce withwel l- d efinedsy mm e try or a n uc lear f irebal l , th e meson pair can b e con s i dered asa qua nt um bipa r tite system wi th an

mpg.de/$\sim$jcg/grbgen.html [^2]: http://webast.ast.obs-mip.fr/hyperz/ [^3]:_http://bat.phys.unsw.edu.au/$\sim$fsoto/hdfcat.html --- abstract:_'Pairs of pseudoscalar neutral_mesons from_decays_of vector_resonances_are studied as_bipartite systems in_the framework of density_operator. Time-dependent quantum_entanglement_is quantified in terms of the entanglement entropy and these dependences are demonstrated on_data_on correlated_pairs_of_$\PK$ and $\PB$ mesons, as_measured by the KLOE and_Belle experiments._Another interesting characteristics of such bipartite systems are_moments_of the CP_distributions. These moments are directly measurable and they appear_to be very sensitive to the_initial degree of_entanglement_of_a pair.' address: 'National Centre_for Nuclear Research, Warsaw, Poland' author: - Wojciech_Wiślicki title: 'Entanglement, fluctuations and discrete symmetries_in particle decays [^1] ' --- Introduction ============ Interferometry of neutral_mesons is recognized as a powerful_and sensitive tool for testing_fundamentals of_the quantum mechanics. In particular,_pairs of pseudoscalar_mesons, originating_from strong decays_of vector resonances, were proved to_be particularly useful_in many precision measurements due to_their_well-defined initial state’s_quantum_numbers_and relatively_high production rates._The_decays $\Pphi(1020)\rightarrow_\PKl\PKs$,_$\Ppsi(3770)\rightarrow \PDzero\APDzero$, $\PUpsilon(10580)\rightarrow \PBd\APBd$ and $\PUpsilon(10860)\rightarrow_\PBs\APBs$_were recognized long time ago as very_useful for the study_of_the CP violation [@branco],_validity of the CPT and_Lorentz invariance [@cpt] or search for_quantum decoherence_[@frascati]. On_the other hand, in case of complex initial states where the_meson source does neither have well-defined_quantum numbers nor symmetry_properties and_can_be a spatially_extended_object, as_e.g. in nuclear collisions, meson interferometry is_often used_to study both the spatio-temporal characteristics_and the degree of_coherence_of the source. In any case,_either the resonance with well-defined symmetry_or a nuclear fireball, the_meson_pair_can be considered as a_quantum bipartite system with an

L}^{\boldsymbol{\nu}}_{s,t+1}:=\tilde{\mathbf{F}}^{\boldsymbol{\nu}}_{s+1,t+1}-\mathbf{L}_{t+1}\tilde{\mathbf{D}}^{\boldsymbol{\nu}}_{s+1,t}, t\ge 1, t-1\ge s\ge 0 $ and $ \mathbf{L}^{\boldsymbol{\nu}}_{t,t+1}:=\tilde{\mathbf{F}}^{\boldsymbol{\nu}}_{t+1,t+1} $. \[lemma:Observation\] Let observation error be $ \tilde{\mathbf{z}}_{t}=\mathbf{z}_{t}-\mathbf{z}^{p}_{t} $ for $ t\ge 1 $. Then, for $ t\ge 0 $ the non-recursive observation error propagation, $ \tilde{\mathbf{z}}_{t+1} $, in terms of the independent variables, including process and measurement noises and the initial state error $ \tilde{\mathbf{x}}_0 $ can be written as follows: $$\begin{aligned} \tilde{\mathbf{z}}_{t+1}=\tilde{\mathbf{H}}^{\mathbf{x}_{0}}_{t+1}\tilde{\mathbf{x}}_0+\sum\limits_{s=0}^{t}\tilde{\mathbf{H}}^{\boldsymbol{\omega}}_{s,t+1}\boldsymbol{\omega}_s +\sum\limits_{s=0}^{t}\tilde{\mathbf{H}}^{\boldsymbol{\nu}}_{s+1,t+1}\boldsymbol{\nu}_{s+1},\end{aligned}$$ where $ \tilde{\mathbf{H}}^{\mathbf{x}_{0}}_{t+1}:=\mathbf{H}_{t+1}\tilde{\mathbf{D}}^{\mathbf{x}_{0}}_{t}, t\ge 0 $, $ \tilde{\mathbf{H}}^{\boldsymbol{\omega}}_{s,t+1}:=\mathbf{H}_{t+1}\tilde{\mathbf{D}}^{\boldsymbol{\omega}}_{s,t}, t\ge 0, t\ge s\ge 0 $, $ \tilde{\mathbf{H}}^{\boldsymbol{\nu}}_{s+1,t+1}:=-\mathbf{H}_{t+1}\tilde{\mathbf{D}}^{\boldsymbol{\nu}}_{s+1,t}, t\ge 1, t-1\ge s\ge 0 $, and $ \tilde{\mathbf{H}}^{\boldsymbol{\nu}}_{t+1,t+1}:

L}^{\boldsymbol{\nu}}_{s, t+1}:=\tilde{\mathbf{F}}^{\boldsymbol{\nu}}_{s+1,t+1}-\mathbf{L}_{t+1}\tilde{\mathbf{D}}^{\boldsymbol{\nu}}_{s+1,t }, t\ge 1, t-1\ge s\ge 0 $ and $ \mathbf{L}^{\boldsymbol{\nu}}_{t, t+1}:=\tilde{\mathbf{F}}^{\boldsymbol{\nu}}_{t+1,t+1 } $. \[lemma: Observation\ ] Let observation error be $ \tilde{\mathbf{z}}_{t}=\mathbf{z}_{t}-\mathbf{z}^{p}_{t } $ for $ t\ge 1 $. Then, for $ t\ge 0 $ the non - recursive observation mistake generation, $ \tilde{\mathbf{z}}_{t+1 } $, in terms of the independent variable star, include process and measurement noises and the initial state of matter error $ \tilde{\mathbf{x}}_0 $ can be written as come: $ $ \begin{aligned } \tilde{\mathbf{z}}_{t+1}=\tilde{\mathbf{H}}^{\mathbf{x}_{0}}_{t+1}\tilde{\mathbf{x}}_0+\sum\limits_{s=0}^{t}\tilde{\mathbf{H}}^{\boldsymbol{\omega}}_{s, t+1}\boldsymbol{\omega}_s + \sum\limits_{s=0}^{t}\tilde{\mathbf{H}}^{\boldsymbol{\nu}}_{s+1,t+1}\boldsymbol{\nu}_{s+1},\end{aligned}$$ where $ \tilde{\mathbf{H}}^{\mathbf{x}_{0}}_{t+1}:=\mathbf{H}_{t+1}\tilde{\mathbf{D}}^{\mathbf{x}_{0}}_{t }, t\ge 0 $, $ \tilde{\mathbf{H}}^{\boldsymbol{\omega}}_{s, t+1}:=\mathbf{H}_{t+1}\tilde{\mathbf{D}}^{\boldsymbol{\omega}}_{s, t }, t\ge 0, t\ge s\ge 0 $, $ \tilde{\mathbf{H}}^{\boldsymbol{\nu}}_{s+1,t+1}:=-\mathbf{H}_{t+1}\tilde{\mathbf{D}}^{\boldsymbol{\nu}}_{s+1,t }, t\ge 1, t-1\ge s\ge 0 $, and $ \tilde{\mathbf{H}}^{\boldsymbol{\nu}}_{t+1,t+1 }:

L}^{\bopdsymbol{\nu}}_{s,t+1}:=\tilde{\mathbf{F}}^{\noldsymbol{\nu}}_{s+1,t+1}-\majhvf{L}_{t+1}\tinde{\matgbf{D}}^{\boldrymbol{\nu}}_{s+1,t}, t\ge 1, t-1\ge s\ge 0 $ anv $ \mqthbf{O}^{\boldsymbol{\nu}}_{t,t+1}:=\tilde{\maghbf{F}}^{\bolddymbol{\nu}}_{r+1,t+1} $. \[ltmma:Observation\] Lxf obseryction crror ue $ \tilde{\mathbf{e}}_{t}=\mathbf{z}_{t}-\madhbf{z}^{p}_{t} $ for $ d\gd 1 $. Then, for $ t\ge 0 $ the non-recursive jbservayiln error propadatipg, $ \tjlde{\mathbf{z}}_{t+1} $, in terms of the indelendent variables, invluding process and measurfmenh noises and the ijitial statg erwir $ \tilde{\matfbf{x}}_0 $ can be written aa follows: $$\begin{aligned} \tilde{\mathcf{z}}_{t+1}=\tnlde{\mathbf{H}}^{\narhbv{f}_{0}}_{t+1}\tilde{\mathuf{x}}_0+\sum\jimits_{s=0}^{t}\tildc{\kathbf{V}}^{\boldsykbol{\omega}}_{s,t+1}\bolcsykboo{\omega}_s +\sum\limits_{s=0}^{t}\tilve{\mathbf{H}}^{\boldsymbol{\no}}_{s+1,t+1}\boldsymtom{\nu}_{s+1},\end{aligned}$$ whwrw $ \tinde{\mdthbw{Y}}^{\maghbr{x}_{0}}_{v+1}:=\mafhbf{H}_{t+1}\hilve{\mathbf{D}}^{\mafhbf{x}_{0}}_{t}, t\ge 0 $, $ \tilde{\mathbf{H}}^{\boldxyivol{\omega}}_{s,t+1}:=\matgbf{H}_{t+1}\tylqe{\mathbf{D}}^{\boldsymbol{\omega}}_{s,t}, t\ge 0, t\ge s\ge 0 $, $ \tilde{\mathbf{H}}^{\boldsymbol{\bu}}_{s+1,t+1}:=-\mathbf{H}_{t+1}\tilde{\mathhf{D}}^{\boldsyibol{\nu}}_{s+1,t}, t\ge 1, t-1\ge s\ge 0 $, and $ \tilde{\mathbf{H}}^{\boldsymbml{\nu}}_{t+1,v+1}:

L}^{\boldsymbol{\nu}}_{s,t+1}:=\tilde{\mathbf{F}}^{\boldsymbol{\nu}}_{s+1,t+1}-\mathbf{L}_{t+1}\tilde{\mathbf{D}}^{\boldsymbol{\nu}}_{s+1,t}, t\ge 1, t-1\ge s\ge 0 $ \mathbf{L}^{\boldsymbol{\nu}}_{t,t+1}:=\tilde{\mathbf{F}}^{\boldsymbol{\nu}}_{t+1,t+1} \[lemma:Observation\] Let error be $ 1 Then, for $ 0 $ the observation error propagation, $ \tilde{\mathbf{z}}_{t+1} $, terms of the independent variables, including process and measurement noises and the initial error $ \tilde{\mathbf{x}}_0 $ can be written as follows: $$\begin{aligned} \tilde{\mathbf{z}}_{t+1}=\tilde{\mathbf{H}}^{\mathbf{x}_{0}}_{t+1}\tilde{\mathbf{x}}_0+\sum\limits_{s=0}^{t}\tilde{\mathbf{H}}^{\boldsymbol{\omega}}_{s,t+1}\boldsymbol{\omega}_s +\sum\limits_{s=0}^{t}\tilde{\mathbf{H}}^{\boldsymbol{\nu}}_{s+1,t+1}\boldsymbol{\nu}_{s+1},\end{aligned}$$ where \tilde{\mathbf{H}}^{\mathbf{x}_{0}}_{t+1}:=\mathbf{H}_{t+1}\tilde{\mathbf{D}}^{\mathbf{x}_{0}}_{t}, 0 $ t\ge 0, t\ge s\ge 0 $, $ \tilde{\mathbf{H}}^{\boldsymbol{\nu}}_{s+1,t+1}:=-\mathbf{H}_{t+1}\tilde{\mathbf{D}}^{\boldsymbol{\nu}}_{s+1,t}, t\ge 1, t-1\ge s\ge 0 $, and $

L}^{\boldsymbol{\nu}}_{s,t+1}:=\tilde{\mathbF{F}}^{\boldsymbOl{\nu}}_{s+1,T+1}-\maThbF{L}_{T+1}\tilDe{\maThbf{D}}^{\boldsymboL{\Nu}}_{s+1,t}, T\ge 1, t-1\ge s\ge 0 $ and $ \mathbf{L}^{\bolDsymbOl{\NU}}_{t,t+1}:=\tILdE{\mathBf{F}}^{\boldSYmBOL{\nu}}_{T+1,t+1} $. \[LeMma:obSErVatioN\] LeT observAtion error Be $ \tIlDe{\mathbf{z}}_{t}=\maTHbF{z}_{t}-\mathbf{z}^{P}_{t} $ fOr $ t\ge 1 $. Then, for $ T\ge 0 $ The non-ReCurSIve obSerVatioN error PRopagaTion, $ \tilde{\MaTHbf{z}}_{t+1} $, iN Terms of THE iNdepEndent variables, inCLuDIng process and mEasureMeNT nOISes And The initial StAte erROr $ \tilde{\MAtHBF{X}}_0 $ caN Be written as foLlows: $$\begin{aLIgnEd} \tildE{\mAthBF{z}}_{t+1}=\tilDe{\matHbF{h}}^{\maThbf{x}_{0}}_{t+1}\tilde{\MathBf{x}}_0+\sum\limIts_{s=0}^{t}\tILde{\mathBF{H}}^{\boldsYmbol{\oMegA}}_{s,t+1}\BoldSYmBoL{\omEgA}_S +\suM\LiMitS_{S=0}^{t}\tIlde{\mathBf{h}}^{\bOldsyMbol{\NU}}_{S+1,T+1}\BoldSymBol{\nU}_{s+1},\end{Aligned}$$ where $ \tIldE{\matHBf{H}}^{\MathbF{x}_{0}}_{t+1}:=\maThbf{h}_{t+1}\Tilde{\Mathbf{d}}^{\mathBf{X}_{0}}_{t}, t\ge 0 $, $ \tilde{\mathbF{H}}^{\boLdsymbol{\oMegA}}_{s,T+1}:=\maThBf{H}_{t+1}\tILde{\matHbf{d}}^{\boLdsymboL{\omega}}_{s,T}, T\ge 0, T\gE S\GE 0 $, $ \tIlde{\mathbf{H}}^{\boldsymBoL{\NU}}_{s+1,T+1}:=-\mathbf{H}_{T+1}\tilde{\MAtHbF{d}}^{\boldsymBoL{\nu}}_{S+1,t}, t\gE 1, T-1\Ge s\ge 0 $, And $ \tILdE{\mathbf{H}}^{\BoldsyMBoL{\nU}}_{t+1,t+1}:

L}^{\boldsymbol{\nu}}_{s,t +1}:=\tild e{\ma thb f{F }} ^{\b olds ymbol{\nu}}_{s + 1,t+ 1}-\mathbf{L}_{t+1}\ti lde{\ ma t hbf{ D }} ^{\bo ldsymbo l {\ n u }}_ {s +1 ,t} ,t \g e 1,t-1 \ge s\g e 0 $ and$ \ ma thbf{L}^{\bo l ds ymbol{\nu} }_{ t,t+1}:=\til de{ \mathb f{ F}} ^ {\bol dsy mbol{ \nu}}_ { t+1,t+ 1} $. \[ le m ma:Obs e rvation \ ] L et o bservation errorb e$ \tilde{\mathb f{z}}_ {t } =\ m a thb f{z }_{t}-\mat hb f{z}^ { p}_{t}$ f o r $ t \ ge 1 $. Then, for $ t\ge 0 $ the n on -re c ursive obse rv a tio n error pro paga tion, $ \ tilde{ \ mathbf{ z }}_{t+1 } $, i n t erm s of th eind ep e nde n tvar i abl es, incl ud in g pro cess a n d mea sur emen t noi ses and the i nit ials tat e err or $\til de {\mat hbf{x} }_0 $ c an be written a s fo llows: $$ \be gi n{a li gned} \tilde {\m ath bf{z}}_ {t+1}=\ t ild e{ \ m a th bf{H}}^{\mathbf{x} _{ 0 } }_ {t+1}\ti lde{\m a th bf { x}}_0+\s um \li mits _ { s=0}^ {t}\ t il de{\math bf{H}} ^ {\ bo ldsymbo l{ \omega }} _{s ,t+ 1}\bo l dsym bol{\o mega}_s+\sum \ limits_{s=0}^{ t }\tilde{\math b f{ H } }^ { \bol dsy mbol{\nu}}_ {s+1 , t+1} \bol d sy mbo l {\nu} _{s+1 }, \ en d {aligned}$$ where $ \ tilde{ \math bf{H}}^{\math bf{x}_{0}} _ { t +1}:=\ma thbf { H} _ {t+1}\tilde{\m athbf {D}}^{\mat h bf{x}_{0 }}_{t }, t\ge0 $, $ \t i l de{\math bf{ H}} ^{\ bol d s ym bol{\omega}}_ { s ,t+1 }: =\mathb f{H }_{t+1} \ti lde {\m ath bf {D}}^{\bo ldsymbol {\ om eg a} }_{ s,t}, t\ge 0,t\ ges\ ge0 $,$ \tild e{\ma thbf {H }} ^ {\b oldsymb o l{ \ n u}}_ {s +1 ,t+1 }:= -\ mathb f{H} _ {t+ 1}\tild e{\mathbf {D} } ^{\b ol ds ymbol{\ nu}}_{s+1,t}, t \ge 1, t-1 \g e s \ge 0$ , and $ \ tilde{\mathbf{H}}^{\bol d symbol{ \nu }}_{t +1,t +1}:

L}^{\boldsymbol{\nu}}_{s,t+1}:=\tilde{\mathbf{F}}^{\boldsymbol{\nu}}_{s+1,t+1}-\mathbf{L}_{t+1}\tilde{\mathbf{D}}^{\boldsymbol{\nu}}_{s+1,t}, t\ge_1, t-1\ge_s\ge 0 $ and_$ \mathbf{L}^{\boldsymbol{\nu}}_{t,t+1}:=\tilde{\mathbf{F}}^{\boldsymbol{\nu}}_{t+1,t+1}_$. \[lemma:Observation\]_Let observation_error_be $ \tilde{\mathbf{z}}_{t}=\mathbf{z}_{t}-\mathbf{z}^{p}_{t}_$ for $_t\ge 1 $. Then,_for $ t\ge_0_$ the non-recursive observation error propagation, $ \tilde{\mathbf{z}}_{t+1} $, in terms of the independent_variables,_including process_and_measurement_noises and the initial state_error $ \tilde{\mathbf{x}}_0 $ can_be written_as follows: $$\begin{aligned} \tilde{\mathbf{z}}_{t+1}=\tilde{\mathbf{H}}^{\mathbf{x}_{0}}_{t+1}\tilde{\mathbf{x}}_0+\sum\limits_{s=0}^{t}\tilde{\mathbf{H}}^{\boldsymbol{\omega}}_{s,t+1}\boldsymbol{\omega}_s +\sum\limits_{s=0}^{t}\tilde{\mathbf{H}}^{\boldsymbol{\nu}}_{s+1,t+1}\boldsymbol{\nu}_{s+1},\end{aligned}$$ where $ \tilde{\mathbf{H}}^{\mathbf{x}_{0}}_{t+1}:=\mathbf{H}_{t+1}\tilde{\mathbf{D}}^{\mathbf{x}_{0}}_{t}, t\ge 0_$,_$ \tilde{\mathbf{H}}^{\boldsymbol{\omega}}_{s,t+1}:=\mathbf{H}_{t+1}\tilde{\mathbf{D}}^{\boldsymbol{\omega}}_{s,t}, t\ge_0, t\ge s\ge 0 $, $ \tilde{\mathbf{H}}^{\boldsymbol{\nu}}_{s+1,t+1}:=-\mathbf{H}_{t+1}\tilde{\mathbf{D}}^{\boldsymbol{\nu}}_{s+1,t}, t\ge 1,_t-1\ge s\ge 0 $, and $_\tilde{\mathbf{H}}^{\boldsymbol{\nu}}_{t+1,t+1}:

m$ of $E$ by zeros as described in Section 4.4. The matrix $\hat E$ depends on a number $\bar \alpha$ and determines the operator $\hat H^-_{20}$.\ [**5$^{th}$ step:**]{} The matrix elements $\hat{\mathcal{E}}(\hat n, \hat d) $ of $\hat H^-_N = \Omega^-_N (\hat H^-_{20})$ are calculated from the matrix elements $\hat E (\hat k, \hat m)$ the same way as the $\mathcal{E} (\hat n, \hat d)$ are calculated from $E (\hat k, \hat m)$ in Section 3.5. In this connection the general results of the Sections 6.1 and 6.2 are useful.\ [**6$^{th}$ step:**]{} One determines the orthogonal submatrices of the matrix of $\hat H^-_N$ according to Section 6.3 and diagonalizes them as many as possible numerically or analytically.\ Then one can try to obtain error estimates applying suitable results of the theory of spectral approximation.\ [**7.1.2:**]{} As mentioned at the end of Section 6.4, the lowest energy levels of $\hat{H}^-_N$ are not simply Hartree-Fock-like approximations of the true values, if $N$ is smaller than $\bar \alpha$. Thus in this case the method presented here reproduces also those results, which are obtained by other methods like density functional theory (DFT) or configuration interaction method (CI). Summing up, it is intuitively clear that the approximation of $H^-_N$ by $\hat H^-_N$ is the better the smaller the number $N$ and the larger the parameter $\bar \alpha$, which in turn is limited by the capacity of computers. My colleague Arno Schindlmayr is preparing an application of the proposed method. Finite procedures ----------------- [**7.2.1:**]{} The program described in Section 7.1.1 is a [*work*]{} program in a strict sense only if all its steps could be carried through in finite time. Thus, the critical points are found in those steps which contain infinite tasks. The first and decisive one is the determination of the infinite ONB $\mathcal{O}_1$ in the second step. For one has to expect that in most cases ${\mathcal{O}}_1$ can neither analytically nor numerically be calculated

m$ of $ E$ by zeros as described in Section 4.4. The matrix $ \hat E$ depends on a numeral $ \bar \alpha$ and determine the operator $ \hat H^-_{20}$.\ [ * * 5$^{th}$ step :* * ] { } The matrix elements $ \hat{\mathcal{E}}(\hat n, \hat d) $ of $ \hat H^-_N = \Omega^-_N (\hat H^-_{20})$ are calculated from the matrix component $ \hat E (\hat k, \hat m)$ the same room as the $ \mathcal{E } (\hat n, \hat d)$ are forecast from $ E (\hat k, \hat m)$ in part 3.5. In this connection the general resultant role of the Sections 6.1 and 6.2 are useful.\ [ * * 6$^{th}$ step :* * ] { } One determines the extraneous submatrices of the matrix of $ \hat H^-_N$ according to Section 6.3 and diagonalizes them as many as possible numerically or analytically.\ Then one can try to obtain mistake estimates applying suitable resultant role of the theory of spectral approximation.\ [ * * 7.1.2 :* * ] { } As mentioned at the end of part 6.4, the lowest energy levels of $ \hat{H}^-_N$ are not simply Hartree - Fock - like approximation of the true values, if $ N$ is smaller than $ \bar \alpha$. Thus in this case the method presented here reproduces also those results, which are obtained by other methods like density functional theory (DFT) or configuration interaction method acting (CI). summarize up, it is intuitively clear that the estimate of $ H^-_N$ by $ \hat H^-_N$ is the dependable the small the number $ N$ and the larger the argument $ \bar \alpha$, which in turn is specify by the capacity of computers. My colleague Arno Schindlmayr is prepare an application of the proposed method. Finite procedures ----------------- [ * * 7.2.1 :* * ] { } The program described in Section 7.1.1 is a [ * study * ] { } program in a strict sense only if all its footprint could be behave through in finite time. Thus, the critical points are found in those steps which control infinite tasks. The first and critical one is the determination of the countless ONB $ \mathcal{O}_1 $ in the second step. For one has to expect that in most cases $ { \mathcal{O}}_1 $ can neither analytically nor numerically be calculated

m$ lf $E$ by zeros as describtd in Section 4.4. Thg natrix $\hat E$ sepends un a number $\bar \alpha$ and devermunes uke operator $\hat H^-_{20}$.\ [**5$^{th}$ rtep:**]{} The latrix eoemeits $\hat{\mathcal{E}}(\hef n, \hat d) $ of $\mat H^-_U = \Omega^-_N (\hat H^-_{20})$ ate calculateg from the matsib zlements $\hat E (\hat k, \hat m)$ the same wwy as tne $\mathcal{E} (\hat g, \hau d)$ are baoculated from $E (\hat k, \hat m)$ ih Sectimn 3.5. In this cpnnection the general resupts lf the Sections 6.1 ajd 6.2 are usedul.\ [**6$^{tr}$ step:**]{} One degermines the orthogonam submatrices of the matrix of $\fat H^-_U$ according ti Sfwtion 6.3 and viagonwlizes them as many ds possoble numericalky mr qnalytically.\ Then one ran try to obtain ertor estimadea applying suitaboe resunts mf tfw tfeody or spechram approximztion.\ [**7.1.2:**]{} As mebtioned at the end pf Wection 6.4, the mowest egergy levels of $\hat{H}^-_N$ are not simply Hagtres-Fock-like approximationw of the true values, lf $N$ is sialler than $\bar \alpha$. Thus in this case the methog preadnttd here eeoroduces also those results, which are obtaineq bu pther methods jike densitu vumstional theori (DFT) ox cknfiguration interwction iethoe (CI). Summyng ip, it is intuitively clear rhat the appgoxination of $H^-_N$ by $\hac H^-_N$ is the yetter the xmaller the number $N$ anb the marger the oarameter $\car \alpha$, which kn nurn is limiudd by the capacitr of compnters. My colueagoe Arno Schindlmajr is preparing an applifatiou of dhe proposfd method. Finite procedures ----------------- [**7.2.1:**]{} The 'cogram descrineg it Sectiou 7.1.1 is s [*work*]{} prograi in a strict xense ouly if all its sneps coulv be carried through in fhjite time. Thns, the cryticql piints afd found in thoxe steps ckich contqin infinite tasks. Thg rirst and decisnre one is the detrrmknaeiln os the infinita ONC $\mxyhcal{U}_1$ in tht sccovd syep. For one has to efpecf that in most casrs ${\mathcal{I}}_1$ can neyther analytivally nor numericaply bx calcnlated

m$ of $E$ by zeros as described 4.4. matrix $\hat depends on a the $\hat H^-_{20}$.\ [**5$^{th}$ The matrix elements n, \hat d) $ of $\hat = \Omega^-_N (\hat H^-_{20})$ are calculated from the matrix elements $\hat E (\hat \hat m)$ the same way as the $\mathcal{E} (\hat n, \hat d)$ are from (\hat \hat in Section 3.5. In this connection the general results of the Sections 6.1 and 6.2 are [**6$^{th}$ step:**]{} One determines the orthogonal submatrices of matrix of $\hat H^-_N$ to Section 6.3 and diagonalizes as as possible or Then can try to error estimates applying suitable results of the theory of spectral approximation.\ [**7.1.2:**]{} As mentioned at the end Section 6.4, energy levels $\hat{H}^-_N$ not Hartree-Fock-like approximations of values, if $N$ is smaller than in this case the method presented here reproduces those results, are obtained by other methods like functional theory (DFT) or configuration interaction method (CI). up, it is intuitively clear that the approximation of $H^-_N$ by $\hat H^-_N$ is the smaller the number $N$ the larger the $\bar which turn limited by capacity of computers. My colleague Arno Schindlmayr is preparing an application the proposed method. Finite procedures ----------------- [**7.2.1:**]{} The program described 7.1.1 a [*work*]{} program a strict sense only all steps could be carried finite Thus, are in steps which contain infinite The first and decisive one the determination of the second step. For one has to expect that most cases ${\mathcal{O}}_1$ can neither analytically nor be calculated

m$ of $E$ by zeros as described in SEction 4.4. The mAtrix $\Hat e$ dePeNds oN a nuMber $\bar \alpha$ anD DeteRmines the operator $\hat H^-_{20}$.\ [**5$^{tH}$ step:**]{} thE MatrIX eLemenTs $\hat{\maTHcAL{e}}(\haT n, \HaT d) $ oF $\hAT H^-_n = \OmegA^-_N (\hAt H^-_{20})$ are cAlculated fRom ThE matrix elemeNTs $\Hat E (\hat k, \haT m)$ tHe same way as tHe $\mAthcal{e} (\hAt n, \HAt d)$ arE caLculaTed froM $e (\hat k, \hAt m)$ in SectIoN 3.5. in this COnnectiON ThE genEral results of the SECtIOns 6.1 and 6.2 are usefuL.\ [**6$^{th}$ steP:**]{} ONE dETErmIneS the orthogOnAl subMAtrices OF tHE MAtrIX of $\hat H^-_N$ accorDing to SectiON 6.3 anD diagoNaLizES them aS many As POssIble numericAlly Or analytiCally.\ THEn one caN Try to obTain erRor EstImatES aPpLyiNg SUitABlE reSUltS of the thEoRy Of speCtraL APPRoxiMatIon.\ [**7.1.2:**]{} AS mentIoned at the end Of SEctiON 6.4, thE loweSt eneRgy lEvEls of $\Hat{H}^-_N$ aRe not SiMply Hartree-Fock-Like ApproximaTioNs Of tHe True vALues, if $n$ is SmaLler thaN $\bar \alpHA$. ThUs IN THiS case the method presEnTED hEre reproDuces aLSo ThOSe resultS, wHicH are OBTaineD by oTHeR methods Like deNSiTy FunctioNaL theorY (DfT) oR coNfiguRAtioN interAction meThod (Ci). summing up, it is iNTuitively cleaR ThAT ThE ApprOxiMation of $H^-_N$ bY $\hat h^-_n$ is tHe beTTeR thE SmallEr the NuMBeR $n$ and the larger the parAmEter $\baR \alphA$, which in turn iS limited by THE Capacity Of coMPuTErs. My colleague arno SChindlmayr IS prepariNg an aPplicatiOn of the prOPOsed methOd. FIniTe pRocEDUrEs ----------------- [**7.2.1:**]{} The program dESCribEd In SectiOn 7.1.1 iS a [*work*]{} pRogRam In a StrIcT sense onlY if all itS sTePs CoUld Be carRIed throuGh In fInIte Time. THUs, the cRiticAl poInTs ARe fOund in tHOsE STeps WhIcH conTaiN iNfiniTe taSKs. THe first And decisiVe oNE is tHe DeTerminaTion of the infiNiTe ONB $\mathcAl{o}_1$ in The secOND step. For One has to expect that in mosT Cases ${\maThcAl{O}}_1$ caN neiTher analyTicAlly noR nuMEricalLy be caLculaTeD

m$ of $E$ by zeros as des cribed inSecti on4.4 .Thematr ix $\hat E$ de p ends on a number $\bar \al pha$an d det e rm inesthe ope r at o r $\ ha tH^- _{ 2 0} $.\ [ **5 $^{th}$ step:**]{ } T he matrix elem e nt s $\hat{\m ath cal{E}}(\hat n, \hatd) $o f $\h atH^-_N = \Om e ga^-_N (\hat H^ -_ { 20})$a re calc u l at ed f rom the matrix el e me n ts $\hat E (\h at k,\h a tm ) $ t hesame way a sthe $ \ mathcal { E} ( \ hat n, \hat d)$ a re calculat e d f rom $E ( \ha t k, \h at m) $i n S ection 3.5. Inthis conn ection the gen e ral res ults o f t heSect i on s6.1 a n d 6 . 2are use ful.\ [* *6 $^ {th}$ ste p : * * ]{}One det ermin es the orthog ona l su b mat rices of t he m at rix o f $\ha t H^- _N $ according toSect ion 6.3 a nddi ago na lizes them a s m any as pos sible n u mer ic a l l yor analytically.\Th e n o ne can t ry too bt ai n error e st ima tesa p plyin g su i ta ble resu lts of th etheoryof spect ra l a ppr oxima t ion. \ [**7 .1.2:**] {} As mentioned at t h e end of Sect i on 6 .4 , the lo west energy lev e ls o f $\ h at {H} ^ -_N$are n ot si m ply Hartree-Fock-li ke appro ximat ions of the t rue values , i f $N$ is sma l le r than $\bar \a lpha$ . Thus int his case themethod p resentedh e re repro duc esals o t h o se results, whi c h are o btained by othermet hod s l ike d ensity fu nctional t he or y(DF T) or configur at ion i nte racti o n meth od (C I).Su mm i ngup, iti si n tuit iv el y cl ear t hat t he a p pro ximatio n of $H^- _N$ by $ \h at H^-_N$ is the bette rthe smalle rthe numbe r $N$ andthe larger the paramete r $\bar\al pha$, whi ch in tur n i s limi ted by the capac ity o fcom p u ters. M ycol le ague ArnoS c hin dlmay ris p reparin g an application o f th e proposed me tho d. F i ni tep ro c edu re s -- - - ------------- [**7.2.1:* *] { }The progra m de sc ribed i n Secti on 7. 1 .1 is a [*work*] {} progra min a s tri ct sense o nly if a ll its st e ps co u ld be c arr ied th ro ugh in f initet ime . Thu s, the c ritica l poi nt s are fo und in those steps whic h cont ain i nfi nite task s.T hefirst and dec isive oneisthe dete rmi n ation oft he in f inite ONB $\mathcal { O} _1$ i nthe seconds t e p.For o neh as toexpe ct that in most c a ses ${\mathcal {O}} _ 1 $ c ann eith er analyticallynor n u m erically b e calculate d

m$_of $E$_by zeros as described_in Section_4.4._The matrix_$\hat E$_depends on a_number $\bar \alpha$_and determines the operator_$\hat H^-_{20}$.\ [**5$^{th}$ step:**]{}_The_matrix elements $\hat{\mathcal{E}}(\hat n, \hat d) $ of $\hat H^-_N = \Omega^-_N (\hat H^-_{20})$ are_calculated_from the_matrix_elements_$\hat E (\hat k, \hat m)$_the same way as the_$\mathcal{E} (\hat_n, \hat d)$ are calculated from $E (\hat_k,_\hat m)$ in_Section 3.5. In this connection the general results of_the Sections 6.1 and 6.2 are_useful.\ [**6$^{th}$ step:**]{} One_determines_the_orthogonal submatrices of the_matrix of $\hat H^-_N$ according to_Section 6.3 and diagonalizes them as_many as possible numerically or analytically.\ Then one_can try to obtain error estimates_applying suitable results of the_theory of_spectral approximation.\ [**7.1.2:**]{} As mentioned at_the end of_Section 6.4,_the lowest energy_levels of $\hat{H}^-_N$ are not simply_Hartree-Fock-like approximations of_the true values, if $N$ is_smaller_than $\bar \alpha$._Thus_in_this case_the method presented_here_reproduces also_those_results, which are obtained by other_methods_like density functional theory (DFT) or configuration_interaction method (CI). Summing_up,_it is intuitively clear_that the approximation of $H^-_N$_by $\hat H^-_N$ is the better_the smaller_the number_$N$ and the larger the parameter $\bar \alpha$, which in turn_is limited by the capacity of_computers. My colleague Arno_Schindlmayr is_preparing_an application of_the_proposed method. Finite_procedures ----------------- [**7.2.1:**]{} The program described in Section 7.1.1_is a_[*work*]{} program in a strict sense_only if all its_steps_could be carried through in finite_time. Thus, the critical points are_found in those steps which_contain_infinite_tasks. The first and decisive_one is the determination of the_infinite ONB $\mathcal{O}_1$_in the second step. For one has_to_expect that in most cases ${\mathcal{O}}_1$_can_neither analytically nor numerically be calculated

earby **G**alaxies: A **F**ar-**I**nfrared **S**urvey with **H**erschel [^2]: http://p3d.sourceforge.net/ [^3]: **G**as **And** **A**bsorption **L**ine **F**itting [^4]: http://www-astro.physics.ox.ac.uk/ mxc/software/ [^5]: **I**DL **T**ool for **E**mission-line **R**atio **A**nalysis --- abstract: | The contribution from the residuals of the foreground can have a significant impact on the temperature maps of the Cosmic Microwave Background (CMB). Mostly, the focus has been on the galactic plane, when foreground cleaning has taken place. However, in this paper, we will investigate the possible foreground contamination, from sources outside the galactic plane in the CMB maps.\ We will analyze the correlation between the Faraday rotation map and the CMB temperature map. The Faraday rotation map is dependent on the galactic magnetic field, as well as the thermal electron density, and both may contribute to the CMB temperature. We find that the standard deviation for the mean cross correlation deviate from that of simulations at the $99.9\%$ level. Additionally, a comparison between the CMB temperature extrema and the extremum points of the Faraday rotation is also performed, showing a general overlap between the two. Also we find that the CMB Cold Spot is located at an area of strong negative cross correlation, meaning that it may be explained by a galactic origin.\ Further, we investigate nearby supernova remnants in the galaxy, traced by the galactic radio loops. These super nova remnants are located at high and low galactic latitude, and thus well outside the galactic plane. We find some correlation between the Faraday Rotation and the CMB temperature, at select radio loops. This indicate, that the galactic foregrounds may affect the CMB, at high galactic latitudes author: - | M. Hansen$^1$, W. Zhao$^1$, A. M. Frejsel$^1$, P. D. Naselsky$^1$, J. Kim$^1$ and O. V. Verkhodanov$^2$\ $^1$Niels Bohr Institute and DISCOVERY center, Blegdamsvej 17, 2100

earby * * G**alaxies: A * * F**ar-**I**nfrared * * S**urvey with * * H**erschel [ ^2 ]: http://p3d.sourceforge.net/ [ ^3 ]: * * G**as * * And * * * * A**bsorption * * L**ine * * F**itting [ ^4 ]: http://www-astro.physics.ox.ac.uk/   mxc / software/ [ ^5 ]: * * I**DL * * T**ool for * * E**mission - line * * R**atio * * A**nalysis --- abstract: | The contribution from the residuals of the foreground can have a significant impingement on the temperature map of the Cosmic Microwave Background (CMB). Mostly, the focus has been on the galactic airplane, when foreground cleaning has taken place. However, in this paper, we will investigate the potential foreground contaminant, from sources outside the galactic plane in the CMB maps.\ We will analyze the correlation between the Faraday rotation map and the CMB temperature function. The Faraday rotation map is dependent on the galactic magnetic playing field, as well as the thermal electron concentration, and both may contribute to the CMB temperature. We rule that the standard diversion for the mean cross correlation deviate from that of simulation at the $ 99.9\%$ level. Additionally, a comparison between the CMB temperature extrema and the extremum point of the Faraday rotation is besides performed, showing a general overlap between the two. Also we find that the CMB Cold Spot is locate at an area of strong negative cross correlation, meaning that it may be explained by a galactic origin.\ Further, we investigate nearby supernova remnants in the galaxy, traced by the galactic radio loops. These super nova remnants are locate at high and low astronomic latitude, and therefore well outside the galactic plane. We find some correlation between the Faraday Rotation and the CMB temperature, at choice radio loops. This indicate, that the galactic foreground may affect the CMB, at high galactic latitudes author: - | M. Hansen$^1 $, W. Zhao$^1 $, A. M. Frejsel$^1 $, P. D. Naselsky$^1 $, J. Kim$^1 $ and O. V. Verkhodanov$^2$\ $ ^1$Niels Bohr Institute and DISCOVERY center, Blegdamsvej 17, 2100

earhy **G**alaxies: A **F**ar-**I**nfrarea **S**urvey with **H**grwchel [^2]: ittp://p3d.skurceforee.net/ [^3]: **G**as **And** **A**bsorption **L**inx **F**irting [^4]: http://www-astro.physics.ox.xc.uk/ mxc/sovtware/ [^5]: **I**EL **T**iil for **E**mission-linc **R**atjl **A**nclbsis --- abstract: | The condribution from tfe residuals of the foreground can hade a sibnlficant impact on uhe temlvrcture maps of the Cosmic Microwzve Bacnground (CMB). Mpstly, the focus has been oj thf galactic plane, wjen foregroohd soeaning has gaken plact. Kowever, in jhis paper, we will investigate tfe poxsible forgyeoujg contaminavion, fgom sources omnside tve galavtic plane in bhe CKB naps.\ We will analyve the correlation bgtween the Fcraday rotation map abd the WMB dempdeatjre mep. Fhe Fagadey rotation map is depwndent on the galacuic nagnetic fiels, as wqlj as the thermal electron density, and bmth may contribute to the XMB temperature. We fijd that tre standard deviation for the mean cross correlathon dxvkatt nvom gyah of simulations at the $99.9\%$ level. Additionally, a dokpsrison between the CMB tekpfrsjure extrema avd the exfremum points of tje Faraqay ritation if alxo performed, showing a geneeal overlap yetqeen the two. Also ce find that the CMB Vold Spot is located at an zrea of strlng negatjxe cross correlagiok, maaning that it may be explwined by e galcctic orkgin.\ Fuwther, we ijvestlcate nearby supernlva rgmnantv in the gwlaxy, traced by the galactic ravmo loops. Thesg sgpeg nova reinantx are located at high and lpw galaetic lxtitude, ans thus xell outside the galactic klane. We find some cowrelqtiob betwedv the Faraday Totation cud the CMV temperature, at sclect dadio loops. Thiw ibdicate, that thr gxlastpc horegwmunds may afxect thd CMB, xt high galccgic katitudes author: - | K. Hahsen$^1$, W. Zhao$^1$, A. M. Frrjfel$^1$, P. D. Baselsky$^1$, J. Kim$^1$ and O. F. Verkhodanov$^2$\ $^1$Nlels Uohr Iistituye wnd DISCOVERY center, Blegdamsvsj 17, 2100

earby **G**alaxies: A **F**ar-**I**nfrared **S**urvey with **H**erschel [^3]: **And** **A**bsorption **F**itting [^4]: http://www-astro.physics.ox.ac.uk/ **E**mission-line **A**nalysis --- abstract: The contribution from residuals of the foreground can have significant impact on the temperature maps of the Cosmic Microwave Background (CMB). Mostly, focus has been on the galactic plane, when foreground cleaning has taken place. in paper, will the possible foreground contamination, from sources outside the galactic plane in the CMB maps.\ We will the correlation between the Faraday rotation map and CMB temperature map. The rotation map is dependent on galactic field, as as thermal density, and both contribute to the CMB temperature. We find that the standard deviation for the mean cross correlation deviate that of the $99.9\%$ Additionally, comparison the CMB temperature the extremum points of the Faraday performed, showing a general overlap between the two. we find the CMB Cold Spot is located an area of strong negative cross correlation, meaning it may be explained by a galactic origin.\ Further, we investigate nearby supernova remnants in traced by the galactic loops. These super remnants located high low galactic and thus well outside the galactic plane. We find some correlation the Faraday Rotation and the CMB temperature, at select radio indicate, the galactic foregrounds affect the CMB, at galactic author: - | M. Zhao$^1$, M. Naselsky$^1$, Kim$^1$ O. V. Verkhodanov$^2$\ $^1$Niels Institute and DISCOVERY center, Blegdamsvej 2100

earby **G**alaxies: A **F**ar-**I**nfrared **s**urvey with **h**erscHel [^2]: HttP://p3D.souRcefOrge.net/ [^3]: **G**as **And** **A**BSorpTion **L**ine **F**itting [^4]: http://www-Astro.PhYSics.OX.aC.uk/ mxC/softwaRE/ [^5]: **I**dl **t**ooL fOr **e**miSsIOn-Line **R**AtiO **A**nalysIs --- abstract: | the CoNtribution frOM tHe residualS of The foregrounD caN have a SiGniFIcant ImpAct on The temPEraturE maps of thE COSmic MiCRowave BACKgRounD (CMB). Mostly, the focuS HaS Been on the galacTic plaNe, WHeN FOreGroUnd cleaninG hAs takEN place. HOWeVER, In tHIs paper, we will Investigate THe pOssiblE fOreGRound cOntamInATioN, from sourceS outSide the gaLactic PLane in tHE CMB mapS.\ We wilL anAlyZe thE CoRrElaTiON beTWeEn tHE FaRaday rotAtIoN map aNd thE cmb TempEraTure Map. ThE Faraday rotatIon Map iS DepEndenT on thE galAcTic maGnetic Field, As Well as the thermaL eleCtron densIty, AnD boTh May coNTributE to The cMB tempErature. wE fiNd THAT tHe standard deviatioN fOR ThE mean croSs corrELaTiON deviate FrOm tHat oF SImulaTionS At The $99.9\%$ level. additiONaLlY, a compaRiSon betWeEn tHe CmB temPEratUre extRema and tHe extREmum points of thE faraday rotatiON iS ALsO PerfOrmEd, showing a gEnerAL oveRlap BEtWeeN The twO. Also We FInD That the CMB Cold Spot iS lOcated At an aRea of strong neGative crosS CORrelatioN, meaNInG That it may be expLaineD by a galactIC origin.\ FUrtheR, we invesTigate neaRBY supernoVa rEmnAntS in THE gAlaxy, traced by THE galAcTic radiO loOps. ThesE suPer NovA reMnAnts are loCated at hIgH aNd LoW gaLactiC Latitude, AnD thUs WelL outsIDe the gAlactIc plAnE. WE FinD some coRReLATion BeTwEen tHe FArAday ROtatIOn aNd the CMb temperatUre, AT selEcT rAdio looPs. This indicatE, tHat the galaCtIc fOregroUNDs may affEct the CMB, at high galactic LAtitudeS auThor: - | M. hansEn$^1$, W. Zhao$^1$, A. M. freJsel$^1$, P. D. nasELsky$^1$, J. KIm$^1$ and O. v. VerkHoDanOV$^2$\ $^1$niels bOHr insTiTute and DIScovERy centEr, blegDamsvej 17, 2100

earby **G**alaxies: A **F* *ar-**I**n frare d * *S* *u rvey wit h **H**erschel [^2 ]: http://p3d.sourcefo rge.n et / [^ 3 ]: **G* *as **A n d* * **A ** bs orp ti o n**L** ine **F**i tting [^4 ]:ht tp://www-ast r o. physics.ox .ac .uk/ mxc/sof twa re/ [ ^5 ]:* *I**D L * *T**o ol for **E**m ission-li ne **R**a t io **A* * n al ysis --- abstract: | The contribut ion fr om th e res idu als of the f oregr o und can ha v e a s i gnificant imp act on thet emp eratur emap s of th e Cos mi c Mi crowave Bac kgro und (CMB) . Most l y, thef ocus ha s been on th e ga l ac ti c p la n e,w he n f o reg round cl ea ni ng ha s ta k e n plac e.Howe ver,in this paper , w e wi l l i nvest igate the p ossib le for egrou nd contamination, fro m sources ou ts ide t he ga l acticpla nein theCMB map s .\ W ewill analyze the c or r e la tion bet ween t h eFa r aday rot at ion map a nd th e CM B t emperatu re map . T he Farada yrotati on ma p i s dep e nden t on t he galac tic m a gnetic field,a s well as the th e r ma l ele ctr on density, and both may co ntr i buteto th eC MB temperature. We fin dthat t he st andard deviat ion for th e m ean cros s co r re l ation deviatefromthat of si m ulations at t he $99.9 \%$ level . Addition all y,a c omp a r is on between th e CMBte mperatu reextrema an d t heext re mum point s of the F ar ad ay ro tatio n is also p erf or med , sho w ing agener al o ve rl a p b etweent he t wo.Al so wefin dthattheC MBCold Sp ot is loc ate d atan a rea ofstrong negati ve cross cor re lat ion, m e a ning tha t it may be explained b y a gala cti c ori gin. \ Fur the r, weinv e stigat e near by su pe rno v a remn a n ts in t he galaxy, t rac ed by t he g alactic radio loops. Thes e su per nova remn ant s ar e lo cat e da t h ig h an d low galactic la titude, an dt hu s well out s ide t he gala ctic pl ane.W e findsome corr elation b et ween t heFaraday Ro tation a nd the CM B temp e ra ture, at selec trad io lo ops. T h isindic ate, t ha t thegalac ti c foregr ounds may affect the CM B, athighgal actic lat itu d esauthor: - | M. Hans en$ ^1$ , W.Zha o $^1$, A.M .Fre j sel$^ 1$,P . D. Nase l sk y$^ 1 $ ,J. Kim$^1$a n d O. V. V erk h odanov $^2$ \ $^1$Niels B o hr Institute a nd D I S COV ERY cent er , Blegdamsvej17, 2 1 0 0

earby **G**alaxies:_A **F**ar-**I**nfrared_**S**urvey with **H**erschel [^2]: http://p3d.sourceforge.net/ [^3]:_**G**as **And**_**A**bsorption_**L**ine **F**itting [^4]:_http://www-astro.physics.ox.ac.uk/ mxc/software/ [^5]:_**I**DL **T**ool for_**E**mission-line **R**atio **A**nalysis _--- abstract: | _ The contribution_from_the residuals of the foreground can have a significant impact on the temperature maps_of_the Cosmic_Microwave_Background_(CMB). Mostly, the focus has_been on the galactic plane,_when foreground_cleaning has taken place. However, in this paper,_we_will investigate the_possible foreground contamination, from sources outside the galactic plane_in the CMB maps.\ _ We will_analyze_the_correlation between the Faraday_rotation map and the CMB temperature_map. The Faraday rotation map is_dependent on the galactic magnetic field, as_well as the thermal electron density,_and both may contribute to_the CMB_temperature. We find that the_standard deviation for_the mean_cross correlation deviate_from that of simulations at the_$99.9\%$ level. Additionally,_a comparison between the CMB temperature_extrema_and the extremum_points_of_the Faraday_rotation is also_performed,_showing a_general_overlap between the two. Also we_find_that the CMB Cold Spot is located_at an area of_strong_negative cross correlation, meaning_that it may be explained_by a galactic origin.\ _ Further,_we investigate_nearby supernova remnants in the galaxy, traced by the galactic radio_loops. These super nova remnants are_located at high and_low galactic_latitude,_and thus well_outside_the galactic_plane. We find some correlation between the_Faraday Rotation_and the CMB temperature, at select_radio loops. This indicate,_that_the galactic foregrounds may affect the_CMB, at high galactic latitudes author: - | _ M. Hansen$^1$,_W._Zhao$^1$,_A. M. Frejsel$^1$, P. D._Naselsky$^1$, J. Kim$^1$ and O. V._Verkhodanov$^2$\ _ $^1$Niels Bohr Institute and DISCOVERY center,_Blegdamsvej_17, 2100

fit is of higher quality with $\chi^2_N = 1.8$.]{} The parameters of the reference models that are used to compare with the observations are listed in Table \[tab:referencemodels\] along with the individual $\chi^2_N$ values for $r_{01}$, $r_{02}$, and combined [$T_{\rm eff}$]{} and [\[M/H\]]{}. [For the Sun and each star in our sample, we derived a best estimate and uncertainty for the stellar radius, mass, age, metallicity, luminosity, and surface gravity using the method described in Sec. \[sec3\] (see Table \[tab:properties\_derived\]). Using the rotation periods given in Table \[tab1\] and the derived radius, we also computed their rotational velocities.]{} [Since the AMP 1.3 method uses only one set of physics in the stellar modeling]{}, the derived uncertainties do not include possible systematic errors arising from errors in the model physics, such as the equation of state, heavy element settling, and convective overshoot. However, the uncertainties include sources of errors arising from free parameters that are often fixed in [the stellar codes used in other methods]{}, for example, the mixing-length parameter $\alpha$, the initial chemical composition $(X_i, Y_i, Z_i),$ or a chemical enrichment law. The [uncertainty on these]{} parameters contributes substantially to the error budget, and in some cases more so, for example, changing the equation of state or the opacities. The effect of such changes in the physics has been studied in detail for HD52265 by [@lebreton2014]. A similar detailed [analysis]{} for each star in the sample we studied is beyond the scope of this paper. [We refer to @silvaaguirre2016, who also analyzed data from @Lund2016 using seven distinct modeling methods and codes.]{} The accuracy, namely the bias and not [the]{} precision, of our results can be [ascertained]{} by an analysis of the solar observations. As [stated]{} above, we [derived a best-matched model with values for the mass of]{} a 1 [M$_{\odot}$]{} model and a radius of 1 [R$_{\odot}$]{}, and an age that, within the derived uncertainty, matches the solar value. A second accuracy test, at least for the age, can be established based on the independently derived ages for the binary system 16 Cyg A and

fit is of higher quality with $ \chi^2_N = 1.8$. ] { } The parameters of the reference model that are use to compare with the observations are listed in Table   \[tab: referencemodels\ ] along with the individual $ \chi^2_N$ values for $ r_{01}$, $ r_{02}$, and compound [ $ T_{\rm eff}$ ] { }   and [ \[M / H\ ] ] { }. [ For the Sun and each star in our sample, we derive a dependable estimate and doubt for the leading radius, mass, age, metallicity, luminosity, and airfoil gravity using the method trace in Sec.   \[sec3\ ] (see Table   \[tab: properties\_derived\ ]). Using the rotation periods given in Table   \[tab1\ ] and the derived spoke, we also computed their rotational velocities. ] { } [ Since the AMP   1.3 method acting uses only one set of physics in the stellar mold ] { }, the derived uncertainties do not include possible systematic error arising from errors in the model physics, such as the equation of state, heavy element settling, and convective overshoot. However, the uncertainties include sources of errors arising from free parameters that are often sterilize in [ the stellar codes use in early methods ] { }, for example, the mixing - length parameter $ \alpha$, the initial chemical composition $ (X_i, Y_i, Z_i),$ or a chemical enrichment law. The [ uncertainty on these ] { } argument contributes substantially to the error budget, and in some cases more so, for example, change the equation of state or the opacities. The effect of such changes in the physics has been study in detail for HD52265 by [ @lebreton2014 ]. A similar detailed [ analysis ] { } for each star in the sample we study is beyond the scope of this paper. [ We refer to @silvaaguirre2016, who also analyzed data from @Lund2016 using seven distinct model methods and code. ] { } The accuracy, namely the bias and not [ the ] { } precision, of our consequence can be [ ascertained ] { } by an psychoanalysis of the solar observations. As [ stated ] { } above, we [ derived a best - pit model with values for the mass of ] { } a 1 [ M$_{\odot}$ ] { }   model and a radius of 1 [ R$_{\odot}$ ] { }, and an historic period that, within the derived doubt, matches the solar value. A second accuracy test, at least for the age, can be established establish on the independently derived ages for the binary system 16 Cyg A and

fih is of higher quality wlth $\chi^2_N = 1.8$.]{} The pceametecs of tge referdnce models that are used to cimpart with the observatkons are pisted ib Taule \[tab:referencemodels\] along wifm the mndividual $\chi^2_N$ values fos $r_{01}$, $r_{02}$, and combhndd [$T_{\rm eff}$]{} and [\[M/H\]]{}. [For the Sun and each ftar in okr sample, we dgrivec a bsst estimate and uncertainty for tge stelnar radius, maxs, age, metallicity, luminoslty, wnd surface gravitj using the metrid described in Sec. \[sec3\] (see Table \[tag:properties\_derived\]). Using the rogatiou periods guvwn lt Table \[tab1\] end thv derived radlls, we anso comluted their robatioial celocities.]{} [Since the AKP 1.3 method uses onli one set mf physics in the sreolar kodening]{}, rhe dedited uncerhaiities do nof include pissible systematic trrjgx arising frkm errjrf in the model physics, such as the equauion kf state, heavy element wettling, and convectige overshjot. However, the uncertainties include sources of arrora arnwing weol free parameters that are often fixed in [the attllsr codes used ln other methods]{}, fpr ecwmple, the mixkng-lenyfh parameter $\alpha$, tje inityal cyemical cjmpoxition $(X_i, Y_i, Z_i),$ or a chemicql enrichmenn laq. The [uncertainty ln these]{} paxametets conyributes substantially co the error budgft, and in rome cases more ro, nor example, changing the equaeion of svate pr the upacoties. Ehe effect of smwh changes in the ohysies hav been stufied in detail for HD52265 by [@lebreton2014]. A similar dgtahleg [analysns]{} for each star in the sample we studieb is bdyond the acope oh this paper. [We refer to @vllvaaguirre2016, xho also wnaltzed data ffum @Lund2016 using xeven disniuct modelung methods and coves.]{} Tfs accuracy, nameoy uhw bias and not [thd]{} pwebismon, os our results can be [sscergained]{} by ak avalyxis of the solar obsarvafions. As [stated]{} abpvc, we [dericed a beft-matched modrl with values for the kasr of]{} s 1 [I$_{\odot}$]{} model and a radius of 1 [R$_{\osot}$]{}, and aj ane that, withig thc detived uncercainty, matches the solar value. A second eccuracy test, at least dor the age, can be gstsblished baxed og the indapendently derived ates for the binarn system 16 Cyg A and

fit is of higher quality with $\chi^2_N The of the models that are observations listed in Table along with the $\chi^2_N$ values for $r_{01}$, $r_{02}$, and [$T_{\rm eff}$]{} and [\[M/H\]]{}. [For the Sun and each star in our sample, derived a best estimate and uncertainty for the stellar radius, mass, age, metallicity, and gravity the described in Sec. \[sec3\] (see Table \[tab:properties\_derived\]). Using the rotation periods given in Table \[tab1\] and derived radius, we also computed their rotational velocities.]{} the AMP 1.3 method only one set of physics the modeling]{}, the uncertainties not possible systematic errors from errors in the model physics, such as the equation of state, heavy element settling, and convective However, the sources of arising free that are often [the stellar codes used in other the mixing-length parameter $\alpha$, the initial chemical composition Y_i, Z_i),$ a chemical enrichment law. The [uncertainty these]{} parameters contributes substantially to the error budget, in some cases more so, for example, changing the equation of state or the opacities. of such changes in physics has been in for by A similar [analysis]{} for each star in the sample we studied is beyond scope of this paper. [We refer to @silvaaguirre2016, who also from using seven distinct methods and codes.]{} The namely bias and not [the]{} our can an of solar observations. As [stated]{} we [derived a best-matched model values for the mass and a radius of 1 [R$_{\odot}$]{}, and an that, within the derived uncertainty, matches the value. A second accuracy test, at least for the age, can be based on derived ages for the binary system 16 Cyg and

fit is of higher quality with $\cHi^2_N = 1.8$.]{} The paraMeterS of The ReFereNce mOdels that are usED to cOmpare with the observatiOns arE lISted IN TAble \[tAb:referENcEMOdeLs\] AlOng WiTH tHe indIviDual $\chi^2_n$ values for $R_{01}$, $r_{02}$, aNd Combined [$T_{\rm eFF}$]{} aNd [\[M/H\]]{}. [For the sun And each star iN ouR samplE, wE deRIved a BesT estiMate anD UncertAinty for tHe STellar RAdius, maSS, AgE, metAllicity, luminositY, AnD Surface gravity Using tHe MEtHOD deScrIbed in Sec. \[sEc3\] (See TaBLe \[tab:prOPeRTIEs\_dERived\]). Using the Rotation perIOds Given iN TAblE \[Tab1\] and The deRiVEd rAdius, we also CompUted their RotatiONal veloCIties.]{} [SiNce the aMP 1.3 MetHod uSEs OnLy oNe SEt oF PhYsiCS in The stellAr MoDelinG]{}, the DERIVed uNceRtaiNties Do not include pOssIble SYstEmatiC erroRs arIsIng frOm erroRs in tHe Model physics, sucH as tHe equatioN of StAte, HeAvy elEMent seTtlIng, And convEctive oVErsHoOT. hOwEver, the uncertaintiEs INClUde sourcEs of erROrS aRIsing froM fRee ParaMETers tHat aRE oFten fixeD in [the STeLlAr codes UsEd in otHeR meThoDs]{}, for EXampLe, the mIxing-lenGth paRAmeter $\alpha$, the INitial chemicaL CoMPOsITion $(x_i, Y_I, Z_i),$ or a chemiCal eNRichMent LAw. the [UNcertAinty On THeSE]{} parameters contribuTeS substAntiaLly to the error Budget, and iN SOMe cases mOre sO, FoR Example, changinG the eQuation of sTAte or the OpaciTies. The eFfect of suCH Changes iN thE phYsiCs hAS BeEn studied in deTAIl foR Hd52265 by [@lebrEtoN2014]. A similAr dEtaIleD [anAlYsis]{} for eaCh star in ThE sAmPlE we StudiED is beyonD tHe sCoPe oF this PAper. [We Refer To @siLvAaGUirRe2016, who alSO aNALyzeD dAtA froM @LuNd2016 Using SeveN DisTinct moDeling metHodS And cOdEs.]{} the accuRacy, namely the BiAs and not [thE]{} pRecIsion, oF OUr resultS can be [ascertained]{} by an anALysis of The Solar ObseRvations. AS [stAted]{} abOve, WE [derivEd a besT-matcHeD moDEL with VALuEs fOr The mass of]{} a 1 [m$_{\ODot}$]{} Model AnD a raDius of 1 [R$_{\Odot}$]{}, and an age that, wiTHin The derived uncErtAintY, MAtCheS ThE SolAr VAluE. a Second accuracy tEst, at least FoR ThE age, can be eSTabLiShed basEd on the IndepENdently Derived agEs for the bInAry sYSTem 16 cyg A and

fit is of higher qualitywith $\chi ^2_N= 1 .8$ .] {} T he p arameters of t h e re ference models that ar e use dt o co m pa re wi th theo bs e r vat io ns ar el is ted i n T able \[ tab:refere nce mo dels\] along wi th the ind ivi dual $\chi^2 _N$ value sfor $r_{0 1}$ , $r_ {02}$, and co mbined [$ T_ { \rm ef f }$]{} a n d [ \[M/ H\]]{}. [For the Su n and each star in ou rs am p l e,wederived abe st es t imate a n du n c ert a inty for thestellar rad i us, mass, a ge, metall icity ,l umi nosity, and sur face grav ity us i ng them ethod d escrib edinSec. \[ se c3\ ]( see Ta ble \[t ab:prope rt ie s\_de rive d \ ] ) . Us ing the rota tion periodsgiv en i n Ta ble \ [tab1 \] a nd thederive d rad iu s, we also comp uted their ro tat io nal v eloci t ies.]{ } [Si nce the AMP 1. 3 me th o d us es only one set of p h y si cs in th e stel l ar m o deling]{ }, th e de r i ved u ncer t ai nties do not i n cl ud e possi bl e syst em ati c e rrors aris ing fr om error s int he model physi c s, such as th e e q u at i on o f s tate, heavy ele m entsett l in g,a nd co nvect iv e o v ershoot. However, t he uncer taint ies include s ources ofe r r ors aris ingf ro m free paramete rs th at are oft e n fixedin [t he stell ar codesu s ed in ot her me tho ds] { } ,for example,t h e mi xi ng-leng thparamet er$\a lph a$, t he initia l chemic al c om po sit ion $ ( X_i, Y_i ,Z_i ), $ o r a c h emical enri chme nt l a w.The [un c er t a inty o nthes e]{ }param eter s co ntribut es substa nti a llyto t he erro r budget, and i n some cas es mo re so, f or examp le, changing the equati o n of st ate or t he o pacities. Th e effe cto f such chang es in t hep h ysics h as be en studied i n det ail f or HD5 2265 by [@lebreton2014].A si milar detaile d [ anal y s is ]{} fo r ea ch sta r in the sample w e studiedis be yond the s c ope o f thispaper.[We r e fer to@silvaagu irre2016, w ho a l s o a nalyzed da ta from@Lund2016 using se ven d ist inct m od eli ng me thodsa ndcodes .]{} Th e accu racy, n amely th e bias and not [the]{}precis ion,ofour resul tsc anbe [ascer tain ed]{} by a n a nal ysisoft he so laro bs erv a tions . As [stated]{ } a bov e , w e [deriveda b est -matc hed modelwith values for the m a ss of]{} a 1 [ M$_{ \ o dot }$] { } mo de l and a radius of 1 [ R$_{\odo t} $]{}, and a n age th at , with in the deriv ed unce r t ai n ty, ma tche s t he solarval ue . A seco nd a c curacy tes t, at le ast fo r the a ge, can be estab lishe d based onthe i nd ependen t ly d erived age s for the b inarysyst em 16 Cyg Aan d

fit_is of_higher quality with $\chi^2_N_= 1.8$.]{}_The_parameters of_the_reference models that_are used to_compare with the observations_are listed in_Table \[tab:referencemodels\]_along with the individual $\chi^2_N$ values for $r_{01}$, $r_{02}$, and combined [$T_{\rm eff}$]{} and [\[M/H\]]{}. [For_the_Sun and_each_star_in our sample, we derived_a best estimate and uncertainty_for the_stellar radius, mass, age, metallicity, luminosity, and surface_gravity_using the method_described in Sec. \[sec3\] (see Table \[tab:properties\_derived\]). Using the rotation periods_given in Table \[tab1\] and the derived_radius, we also_computed_their_rotational velocities.]{} [Since the AMP 1.3_method uses only one set of_physics in the stellar modeling]{}, the_derived uncertainties do not include possible systematic_errors arising from errors in the_model physics, such as the_equation of_state, heavy element settling, and_convective overshoot. However,_the uncertainties_include sources of_errors arising from free parameters that_are often fixed_in [the stellar codes used in_other_methods]{}, for example,_the_mixing-length_parameter $\alpha$,_the initial chemical_composition_$(X_i, Y_i,_Z_i),$_or a chemical enrichment law. The_[uncertainty_on these]{} parameters contributes substantially to the_error budget, and in_some_cases more so, for_example, changing the equation of_state or the opacities. The effect_of such_changes in_the physics has been studied in detail for HD52265 by [@lebreton2014]._A similar detailed [analysis]{} for each_star in the sample_we studied_is_beyond the scope_of_this paper._[We refer to @silvaaguirre2016, who also analyzed_data from_@Lund2016 using seven distinct modeling methods_and codes.]{} The accuracy, namely_the_bias and not [the]{} precision, of_our results can be [ascertained]{} by_an analysis of the solar_observations._As_[stated]{} above, we [derived a_best-matched model with values for the_mass of]{} a_1 [M$_{\odot}$]{} model and a radius of 1_[R$_{\odot}$]{},_and an age that, within the_derived_uncertainty, matches the solar value. A_second_accuracy_test, at least for the_age, can be established based on_the independently derived ages for the binary system 16_Cyg A and

parameters of this model have not yet been determined with a precision comparable to that reached for the usual power-law critical phenomena, due to the complicated and peculiar nature of the critical singularities. Therefore any effort at improving the accuracy of the available numerical methods by stretching them towards their (present) limits should be welcome. After extending the high-temperature(HT) expansions of the model in successive steps [@BC] from order $\beta^{ 10}$ to $\beta^{ 21}$, we present here a further extension by three orders for the expansions of the spin-spin correlation on the square lattice and perform a first brief analysis of our data for the susceptibility and the second-moment correlation-length. More results and further extensions both for the square and the triangular lattice[@bct] will be presented elsewhere. Our study strengthens the support of the main results of the BKT theory already coming from the analysis of shorter series and suggests a closer agreement with recent high-precision simulation studies[@Kenna; @Has] of the model. The Hamiltonian $$H\{ v \} = - 2{J} \sum_{nn} \vec v({\vec r}) \cdot \vec v({\vec r}') \label{hamilt}$$ with $\vec v({\vec r})$ a two-component unit vector at the site ${\vec r}$ of a square lattice, describes a system of $XY$ spins with nearest-neighbor interactions. Computing the spin-spin correlation function, $$C(\vec 0, \vec x;\beta)= <s(\vec 0) \cdot s(\vec x)>, \label{corfun}$$ (for all values of $\vec x$ for which the HT expansion coefficients are non-trivial within the maximum order reached), as series expansion in the variable $\beta= J/kT$, enables us to evaluate the expansions of the $l$-th order spherical moments of the correlation function: $$m^{(l)}(\beta) = \sum_{\vec x }|\vec x|^l <s(\vec 0) \cdot s(\vec x)> \label{sfermom}$$ and in particular the reduced ferromagnetic susceptibility $\chi(\beta)=m^{(0)}(\beta)$. In terms of $m^{(2)}(\beta)$ and $\chi(\beta )$ we can form the second-moment correlation length: $$\xi^2(\beta) = m^{(2)}(\beta)/4\chi(\beta). \label{cor

parameters of this model have not yet been determined with a preciseness comparable to that pass for the usual exponent - police critical phenomena, due to the complicated and peculiar nature of the critical singularity. consequently any effort at improving the accuracy of the available numerical methods by stretching them towards their (present) limit should be welcome. After extending the high - temperature(HT) expansion of the exemplar in successive steps   [ @BC ] from order $ \beta^ { 10}$ to $ \beta^ { 21}$, we present here a further extension by three order for the expansions of the spin - spin correlation coefficient on the square lattice and do a first abbreviated analysis of our data for the susceptibility and the second - moment correlation - length. More results and further extension both for the square and the triangular lattice[@bct ] will be presented elsewhere. Our study strengthens the support of the chief results of the BKT theory already coming from the analysis of shorter series and suggests a closer agreement with recent high - preciseness model studies[@Kenna; @Has ] of the model. The Hamiltonian $ $ H\ { v \ } = - 2{J } \sum_{nn } \vec v({\vec radius }) \cdot \vec v({\vec r }') \label{hamilt}$$ with $ \vec v({\vec r})$ a two - component unit vector at the site $ { \vec r}$ of a straight lattice, describes a system of $ XY$ spins with approximate - neighbor interactions. Computing the spin - spin correlation routine, $ $ C(\vec 0, \vec x;\beta)= < s(\vec 0) \cdot s(\vec x) >, \label{corfun}$$ (for all values of $ \vec x$ for which the HT expansion coefficients are non - trivial within the maximum orderliness reach), as serial expansion in the variable $ \beta= J / kT$, enables us to evaluate the expansions of the $ l$-th order spherical moments of the correlation affair: $ $ m^{(l)}(\beta) = \sum_{\vec x } |\vec x|^l < s(\vec 0) \cdot s(\vec x) > \label{sfermom}$$ and in particular the reduced ferromagnetic susceptibility $ \chi(\beta)=m^{(0)}(\beta)$. In terms of $ m^{(2)}(\beta)$ and $ \chi(\beta) $ we can form the second - consequence correlation distance: $ $ \xi^2(\beta) = m^{(2)}(\beta)/4\chi(\beta). \label{cor

pagameters of this model hxve not yet beeu deterkined sith a pfecision comparable to that ceacyed fir the usual power-law zritical ihenomena, due ro the com'micated and psguliax iature of the ctitical singglarities. Therafurz any effort at improving the accurasy of tne available numgricak metgods by stretching them towards thsir (prevent) limits snould be welcome. After extfndijg the high-temperahure(HT) expabsiogw of the moddl in succtsvive steps [@GC] from order $\beta^{ 10}$ to $\beta^{ 21}$, we presznt here a duethft extension uy thrve orders for the expdnsions of the spin-spln cocrelqtion on the square lettice and perform a first brhey analysis of our datq dor tve sgsceoribkliuy end the sfcoid-moment codrelation-lebgth. More results amd durther extenaions foeh for the square and the triangular lautice[@gct] will be presented eosewhere. Our study sttengthens ehe support of the main results of the BKT theory alreedh cimlng weol the analysis of shorter series and suggests z vlpser agreement with receny jibr-precision sioulation atudies[@Kenna; @Has] ov the mjdel. Tye Hamiltjniam $$H\{ v \} = - 2{J} \sum_{nn} \vec v({\vec r}) \cdot \vec v({\vvc r}') \oabel{hamilt}$$ with $\vee v({\vec r})$ a tco-comppnent unit vector at the sitz ${\vec d}$ of a squage lattics, describes a sysgem ox $XY$ spins with nearest-neidhbor intxractnons. Compjtinb the fpin-spin clrrelation function, $$C(\vec 0, \vec x;\teta)= <s(\vec 0) \cdot s(\vec x)>, \label{corfun}$$ (for all values of $\vec x$ fog which tke HT cxpansion coeffycients are nou-trivial withkn the maxpmum ordec reached), as series expanvlon in the veriable $\bqta= H/kT$, wnables js to evaluate the expaufionw of the $l$-th order spfsrical moments if uhw correlation gunztijn: $$m^{(k)}(\bqda) = \sum_{\vec x }|\vez x|^u <s(\vec 0) \cdot s(\vec b)> \labrl{sfermom}$$ and in parthculzr the reduced fertoiagnetic susceptybility $\chi(\beya)=m^{(0)}(\beta)$. In terms ov $m^{(2)}(\beva)$ and $\chi(\brta )$ we can form the second-moment correlatlon length: $$\xi^2(\beea) = n^{(2)}(\beta)/4\chi(\beta). \kabel{cor

parameters of this model have not yet with precision comparable that reached for due the complicated and nature of the singularities. Therefore any effort at improving accuracy of the available numerical methods by stretching them towards their (present) limits be welcome. After extending the high-temperature(HT) expansions of the model in successive steps from $\beta^{ to 21}$, we present here a further extension by three orders for the expansions of the spin-spin on the square lattice and perform a first analysis of our data the susceptibility and the second-moment More and further both the and the triangular will be presented elsewhere. Our study strengthens the support of the main results of the BKT theory coming from of shorter and a agreement with recent studies[@Kenna; @Has] of the model. The \} = - 2{J} \sum_{nn} \vec v({\vec r}) \vec v({\vec \label{hamilt}$$ with $\vec v({\vec r})$ a unit vector at the site ${\vec r}$ of square lattice, describes a system of $XY$ spins with nearest-neighbor interactions. Computing the spin-spin correlation 0, \vec x;\beta)= <s(\vec \cdot s(\vec x)>, (for values $\vec for which HT expansion coefficients are non-trivial within the maximum order reached), as expansion in the variable $\beta= J/kT$, enables us to evaluate of $l$-th order spherical of the correlation function: = x }|\vec x|^l <s(\vec s(\vec \label{sfermom}$$ the ferromagnetic $\chi(\beta)=m^{(0)}(\beta)$. In terms of and $\chi(\beta )$ we can the second-moment correlation length:

parameters of this model have Not yet been DeterMinEd wItH a prEcisIon comparable tO That Reached for the usual poweR-law cRiTIcal PHeNomenA, due to tHE cOMPliCaTeD anD pECuLiar nAtuRe of the Critical siNguLaRities. TherefORe Any effort aT imProving the acCurAcy of tHe AvaILable NumEricaL methoDS by strEtching thEm TOwards THeir (preSENt) LimiTs should be welcome. aFtER extending the hIgh-temPeRAtURE(HT) ExpAnsions of tHe Model IN succesSIvE STEps [@bc] from order $\betA^{ 10}$ to $\beta^{ 21}$, we prESenT here a FuRthER extenSion bY tHRee Orders for thE expAnsions of The spiN-Spin corRElation On the sQuaRe lAttiCE aNd PerFoRM a fIRsT brIEf aNalysis oF oUr Data fOr thE SUSCeptIbiLity And thE second-moment CorRelaTIon-LengtH. More ResuLtS and fUrther ExtenSiOns both for the sqUare And the triAngUlAr lAtTice[@bCT] will bE prEseNted elsEwhere. OUR stUdY STReNgthens the support oF tHE MaIn resultS of the bkT ThEOry alreaDy ComIng fROM the aNalySIs Of shorteR serieS AnD sUggests A cLoser aGrEemEnt With rECent High-prEcision sImulaTIon studies[@KennA; @has] of the model. tHe hAMiLToniAn $$H\{ V \} = - 2{J} \sum_{nn} \vec v({\Vec r}) \CDot \vEc v({\vEC r}') \LabEL{hamiLt}$$ witH $\vEC v({\VEc r})$ a two-component uniT vEctor aT the sIte ${\vec r}$ of a squAre lattice, DESCribes a sYsteM Of $xy$ spins with nearEst-neIghbor inteRActions. COmputIng the spIn-spin corRELation fuNctIon, $$c(\veC 0, \veC X;\BeTa)= <s(\vec 0) \cdot s(\veC X)>, \LabeL{cOrfun}$$ (foR alL values Of $\vEc x$ For WhiCh The HT expaNsion coeFfIcIeNtS arE non-tRIvial witHiN thE mAxiMum orDEr reacHed), as SeriEs ExPAnsIon in thE VaRIAble $\BeTa= j/kT$, eNabLeS us to EvalUAte The expaNsions of tHe $l$-TH ordEr SpHerical Moments of the cOrRelation fuNcTioN: $$m^{(l)}(\betA) = \SUm_{\vec x }|\veC x|^l <s(\vec 0) \cdot s(\vec x)> \label{sfERmom}$$ and In pArticUlar The reduceD feRromagNetIC suscePtibilIty $\chI(\bEta)=M^{(0)}(\BEta)$. In TERmS of $M^{(2)}(\bEta)$ and $\chi(\bETA )$ we Can foRm The sEcond-moMent correlation lenGTh: $$\xI^2(\beta) = m^{(2)}(\beta)/4\chi(\BetA). \labEL{CoR

parameters of this modelhave not y et be endet er mine d wi th a precision comp arable to that reached forth e usu a lpower -law cr i ti c a l p he no men a, du e tothe compli cated andpec ul iar nature o f t he critica l s ingularities . T herefo re an y effo rtat im provin g the a ccuracy o ft he ava i lable n u m er ical methods by stret c hi n g them towards their ( p re s e nt) li mits shoul dbe we l come. A f te r e xte n ding the high -temperatur e (HT ) expa ns ion s of th e mod el insuccessivestep s [@BC] f rom or d er $\be t a^{ 10} $ to $ \be ta^ { 21 } $, w e p re s ent he rea fu rther ex te ns ion b y th r e e orde rsforthe e xpansions ofthe spi n -sp in co rrela tion o n the squar e lat ti ce and performa fi rst brief an al ysi sof ou r datafor th e susce ptibili t y a nd t h esecond-moment corr el a t io n-length . More re su l ts and f ur the r ex t e nsion s bo t hfor thesquare an dthe tri an gularla tti ce[ @bct] will be pr esentedelsew h ere. Our study strengthens t h es u pp o rt o f t he main res ults of t he B K Tthe o ry al ready c o mi n g from the analysis o f shor ter s eries and sug gests a cl o s e r agreem entw it h recent high-p recis ion simula t ion stud ies[@ Kenna; @ Has] of t h e model. Th e H ami lto n i an $$H\{ v \}= - 2{ J} \sum_ {nn } \vec v( {\v ecr}) \ cdot \vec v({\vec r }' )\l abe l{ham i lt}$$ w it h $ \v ecv({\v e c r})$ a tw o-co mp on e ntunit ve c to r at t he s ite${\ ve c r}$ ofa sq uare la ttice, de scr i besasy stem of $XY$ spins w it h nearest- ne igh bor in t e ractions . Computing the spin-s p in corr ela tionfunc tion, $$C (\v ec 0,\ve c x;\be ta)= < s(\ve c0)\ c dot s ( \ ve c x )> , \label{c o r fun }$$ ( fo r al l value s of $\vec x$ forw hic h the HT expa nsi on c o e ff ici e nt s ar en on- t r ivial within th e maximumor d er reached), asse ries ex pansion in t h e varia ble $\bet a= J/kT$, e nabl e s us to evalua te the e xpansions of th e $ l$-th or der sp he ric al mo mentso f t he co rrelat io n func tion: $ $m^{(l)} (\beta) = \sum_{\vec x }|\ve c x|^ l < s(\vec 0) \c d ots(\vec x) > \l abel{sferm om} $$ andinp artic ular th e r e duced fer r omagnetic su sce p t ib ility $\chi ( \ b eta )=m^{ (0) } (\beta )$.In terms of $m^{( 2 )}(\beta)$ and $\c h i (\b eta )$ w ecan form the s eco nd - m oment co rr elation len gth: $$\ xi ^ 2(\be ta) =m^{(2) }(\beta ) / 4\ c hi(\be ta). \l abel{cor

parameters_of this_model have not yet_been determined_with_a precision_comparable_to that reached_for the usual_power-law critical phenomena, due_to the complicated_and_peculiar nature of the critical singularities. Therefore any effort at improving the accuracy of_the_available numerical_methods_by_stretching them towards their (present)_limits should be welcome. After_extending the_high-temperature(HT) expansions of the model in successive steps [@BC]_from_order $\beta^{ 10}$_to $\beta^{ 21}$, we present here a further extension_by three orders for the expansions_of the spin-spin_correlation_on_the square lattice and_perform a first brief analysis of_our data for the susceptibility and_the second-moment correlation-length. More results and further_extensions both for the square and_the triangular lattice[@bct] will be_presented elsewhere._Our study strengthens the support_of the main_results of_the BKT theory_already coming from the analysis of_shorter series and_suggests a closer agreement with recent_high-precision_simulation studies[@Kenna; @Has]_of_the_model. The Hamiltonian $$H\{_v \} =_-_2{J} _\sum_{nn}_ \vec v({\vec r}) \cdot \vec v({\vec_r}') \label{hamilt}$$ with_$\vec v({\vec r})$ a two-component unit vector_at the site ${\vec_r}$_of a square lattice,_describes a system of $XY$_spins with nearest-neighbor interactions. Computing the spin-spin_correlation function,_$$C(\vec 0,_\vec x;\beta)= <s(\vec 0) \cdot s(\vec x)>, \label{corfun}$$ (for all values of_$\vec x$ for which the HT_expansion coefficients are non-trivial_within the_maximum_order reached), as_series_expansion in_the variable $\beta= J/kT$, enables us to_evaluate the_expansions of the $l$-th order spherical_moments of the correlation_function:_$$m^{(l)}(\beta) = \sum_{\vec x }|\vec_x|^l <s(\vec 0) \cdot s(\vec x)> \label{sfermom}$$ and_in particular the reduced ferromagnetic_susceptibility_$\chi(\beta)=m^{(0)}(\beta)$._In terms of $m^{(2)}(\beta)$ and_$\chi(\beta )$ we can form the_second-moment correlation length:_$$\xi^2(\beta) = m^{(2)}(\beta)/4\chi(\beta). \label{cor

universality classes! This high degree of coincidence leads to the important conjecture that the universal content of the partial differential equations and is equivalent [^1].\ ---------------- --------------------------------- ------------------------------------ $\quad N\quad$ $\quad {\rm Polchinski}\quad$ $\quad\quad R_{\rm opt}\quad\quad$ \[.5ex\] $0$ ${}\quad$ ${}\quad$ — $\ 0.65788^{a}\ \ \ {}$ $1$ ${}\quad$ $\ 0.65574^{c,e}\ {}$ $\ 0.655746^{a}\ {}$ $2$ ${}\quad$ $\ 0.6712^{d}\ {}$ $\ 0.671221^{a}\ {}$ $3$ ${}\quad$ $\ 0.6998^{d}\ {}$ $\ 0.699837^{a}\ {}$ $4$ ${}\quad$ $\ 0.7338^{d}\ {}$ $\ 0.733753^{a}\ {}$ ---------------- --------------------------------- ------------------------------------ : Subleading correction-to-scaling exponent $\omega$. [**5. Stability**]{}\ Next, we analyse the locality and stability structure of flows and show that the non-universal properties of RG flows in the vicinity of a scaling solution are vastly different. Consider the fixed point solution itself, which is non-universal and not measurable in any experiment. Using, the nontrivial scaling solution with $u'_\star \neq$ const. and $\partial_t u'_\star=0$ of the Polchinski RG obeys the differential equation \[FlowPRGupScal\] 2 u\_’ - (d-2)u\_” =2(u\_’)\^2 - 02N u\_”’ -(1+02N-4u\_’)u\_”. For large fields, the scaling potential behaves as \[solutionPRGExpansionLarge\] u\_() +[subleading]{}. An analytical solution for $N=\infty$ has been given in [@Kubyshin:2001gz]. On the level of the RG flow, this behaviour stems from a cancellation between the canonical scaling of the potential and its non-linear renormalisation. From, we conclude that the non-trivial quantum contributions to the Polchinski flow diverges like \[PRG-largefield\] \_t

universality classes! This high degree of coincidence lead to the authoritative conjecture that the universal content of the fond differential equations and is equivalent [ ^1].\ ---------------- --------------------------------- ------------------------------------ $ \quad N\quad$ $ \quad { \rm Polchinski}\quad$ $ \quad\quad R_{\rm opt}\quad\quad$ \[.5ex\ ] $ 0 $ $ { } \quad$ $ { } \quad$ — $ \ 0.65788^{a}\ \ \ { } $ $ 1 $ $ { } \quad$ $ \ 0.65574^{c, e}\ { } $ $ \ 0.655746^{a}\ { } $ $ 2 $ $ { } \quad$ $ \ 0.6712^{d}\ { } $ $ \ 0.671221^{a}\ { } $ $ 3 $ $ { } \quad$ $ \ 0.6998^{d}\ { } $ $ \ 0.699837^{a}\ { } $ $ 4 $ $ { } \quad$ $ \ 0.7338^{d}\ { } $ $ \ 0.733753^{a}\ { } $ ---------------- --------------------------------- ------------------------------------ : Subleading correction - to - scaling advocate $ \omega$. [ * * 5. Stability**]{}\ Next, we analyze the locality and constancy social organization of flows and show that the non - cosmopolitan properties of RG flows in the vicinity of a scaling solution are vastly unlike. Consider the fixed point solution itself, which is non - universal and not measurable in any experiment. Using, the nontrivial scaling solution with $ u'_\star \neq$ const.   and $ \partial_t u'_\star=0 $ of the Polchinski RG obeys the differential equation \[FlowPRGupScal\ ] 2 u\ _ ’ - (d-2)u\ _ ” = 2(u\_’)\^2 - 02N u\ _ ” ’ -(1 + 02N-4u\_’)u\ _ ”. For bombastic fields, the scaling potential behaves as \[solutionPRGExpansionLarge\ ] u\ _ () + [ subleading ] { }. An analytic solution for $ N=\infty$ has been given in [ @Kubyshin:2001gz ]. On the level of the RG flow, this behaviour stem from a cancellation between the canonical scaling of the potential and its non - linear renormalisation. From, we conclude that the non - trivial quantum contribution to the Polchinski flow diverges like \[PRG - largefield\ ] \_t

unlversality classes! This migh degree of coincideice leass to thd important conjecture that vhe yniveesal content of the paftial difverentiao eqnations and is equivalenb [^1].\ ---------------- --------------------------------- ------------------------------------ $\qbav N\quad$ $\quad {\rk Polchinsni}\quad$ $\qudd\ducd R_{\rm opt}\quad\quad$ \[.5ex\] $0$ ${}\quad$ ${}\quaq$ — $\ 0.65788^{a}\ \ \ {}$ $1$ ${}\quad$ $\ 0.65574^{c,e}\ {}$ $\ 0.655746^{a}\ {}$ $2$ ${}\quad$ $\ 0.6712^{d}\ {}$ $\ 0.671221^{a}\ {}$ $3$ ${}\quad$ $\ 0.6998^{d}\ {}$ $\ 0.699837^{q}\ {}$ $4$ ${}\quad$ $\ 0.7338^{d}\ {}$ $\ 0.733753^{a}\ {}$ ---------------- --------------------------------- ------------------------------------ : Subleading correctiov-to-sccling expongur $\olgga$. [**5. Stabilitb**]{}\ Next, re analyse tmv localhty and stability strmcturx of flows and show that vhe non-universal prokerties of RY flows in the viciniry of a scaning woljtikn ade vashly different. Consider tye fixed point soluuiog itself, which is nog-ugiversal and not measurable in any expegimeht. Using, the nontrivial scaling solution witj $u'_\star \nqq$ const. and $\partial_t u'_\star=0$ of the Polchinski RG obays tie diyncrenguap equation \[FlowPRGupScal\] 2 u\_’ - (d-2)u\_” =2(u\_’)\^2 - 02N u\_”’ -(1+02N-4u\_’)u\_”. Fow lsrne fields, the sccling potential nejafgs as \[solutionKRGExpauaiknLarge\] u\_() +[subleadijg]{}. An agalytucal soluuion gor $N=\infty$ has been given ib [@Kubyshin:2001gz]. In the level of thz RG flow, thns behsviout stems from a cancellacion bstween the fanonical rcaling of the putektidl and iur non-linear renorialisatioi. Frok, we covcluce thae the non-tgivial quantum contributlons jo the Polchinskl flow diverges like \[PRG-largefixkd\] \_t

universality classes! This high degree of coincidence the conjecture that universal content of is [^1].\ ---------------- --------------------------------- $\quad N\quad$ $\quad Polchinski}\quad$ $\quad\quad R_{\rm opt}\quad\quad$ \[.5ex\] $0$ ${}\quad$ — $\ 0.65788^{a}\ \ \ {}$ $1$ ${}\quad$ $\ 0.65574^{c,e}\ {}$ $\ {}$ $2$ ${}\quad$ $\ 0.6712^{d}\ {}$ $\ 0.671221^{a}\ {}$ $3$ ${}\quad$ $\ 0.6998^{d}\ $\ {}$ ${}\quad$ 0.7338^{d}\ {}$ $\ 0.733753^{a}\ {}$ ---------------- --------------------------------- ------------------------------------ : Subleading correction-to-scaling exponent $\omega$. [**5. Stability**]{}\ Next, analyse the locality and stability structure of flows show that the non-universal of RG flows in the of scaling solution vastly Consider fixed point solution which is non-universal and not measurable in any experiment. Using, the nontrivial scaling solution with $u'_\star \neq$ and $\partial_t the Polchinski obeys differential \[FlowPRGupScal\] 2 u\_’ =2(u\_’)\^2 - 02N u\_”’ -(1+02N-4u\_’)u\_”. For scaling potential behaves as \[solutionPRGExpansionLarge\] u\_() +[subleading]{}. An solution for has been given in [@Kubyshin:2001gz]. On level of the RG flow, this behaviour stems a cancellation between the canonical scaling of the potential and its non-linear renormalisation. From, we the non-trivial quantum contributions the Polchinski flow like \_t

universality classes! This hiGh degree of CoincIdeNce LeAds tO the Important conjeCTure That the universal contenT of thE pARtiaL DiFfereNtial eqUAtIONs aNd Is EquIvALeNt [^1].\ ---------------- --------------------------------- ------------------------------------ $\quaD N\qUad$ $\quad {\Rm PolchinsKi}\qUaD$ $\quad\quad R_{\rm OPt}\Quad\quad$ \[.5ex\] $0$ ${}\QuaD$ ${}\quad$ — $\ 0.65788^{a}\ \ \ {}$ $1$ ${}\quad$ $\ 0.65574^{c,e}\ {}$ $\ 0.655746^{A}\ {}$ $2$ ${}\quAd$ $\ 0.6712^{d}\ {}$ $\ 0.671221^{a}\ {}$ $3$ ${}\quAd$ $\ 0.6998^{D}\ {}$ $\ 0.699837^{a}\ {}$ $4$ ${}\qUAd$ $\ 0.7338^{d}\ {}$ $\ 0.733753^{a}\ {}$ ---------------- --------------------------------- ------------------------------------ : SUblEadinG correCTion-to-Scaling exPoNEnt $\omeGA$. [**5. StabilITY**]{}\ NExt, wE analyse the localiTY aND stability struCture oF fLOwS ANd sHow That the non-UnIversAL properTIeS OF rG fLOws in the vicinIty of a scaliNG soLution ArE vaSTly difFerenT. CONsiDer the fixed PoinT solution Itself, WHich is nON-univerSal and Not MeaSuraBLe In Any ExPEriMEnT. UsINg, tHe nontriViAl ScaliNg soLUTIOn wiTh $u'_\Star \Neq$ coNst. and $\partial_T u'_\sTar=0$ oF The polchInski rG obEyS the dIffereNtial EqUation \[FlowPRGupscal\] 2 U\_’ - (d-2)u\_” =2(u\_’)\^2 - 02N u\_”’ -(1+02N-4u\_’)u\_”. for LaRge FiElds, tHE scaliNg pOteNtial beHaves as \[SOluTiONprGexpansionLarge\] u\_() +[subLeADInG]{}. An analyTical sOLuTiON for $N=\infTy$ Has Been GIVen in [@kubySHiN:2001gz]. On the Level oF ThE Rg flow, thIs BehaviOuR stEms From a CAnceLlatioN between The caNOnical scaling oF The potential aND iTS NoN-LineAr rEnormalisatIon. FROm, we ConcLUdE thAT the nOn-triViAL qUAntum contributions tO tHe PolcHinskI flow diverges Like \[PRG-larGEFIeld\] \_t

universality classes! Thi s high deg ree o f c oin ci denc e le ads to the imp o rtan t conjecture that theunive rs a l co n te nt of the pa r ti a l di ff er ent ia l e quati ons and is equivalen t [ ^1 ].\ ----- - -- -------- - --- ------------ --- ------ -- --- - -- -- --- ----- ------ - ------ --------- -- - - $ \ quad N\ q u ad $ $ \quad {\rm Polchi n sk i }\quad$ $ \quad\ qu a dR _ {\r m o pt}\quad\q ua d$ \[.5ex \ ]$ 0 $ $ {}\quad$ ${}\ quad$ — $ \ 0.65 788^{ a} \ \\ {}$ $ 1$ ${}\q u ad$ $\0 .65574^ {c,e}\ {} $ $\0. 6 557 4 6^ {a} \ {} $ $2 $ $ { } \ quad $ $ \ 0. 6712^ {d}\ {}$ $\ 0.67 1221^ {a}\ { }$ $3$ ${}\quad$ $ \ 0. 6998^{d}\ {} $ $\ 0 .69 983 7^{a}\{}$ $ 4 $ ${}\quad$ $\0. 7 3 38 ^{d}\ {} $ $\ 0. 73 375 3^{a } \ {}$ -- - -- -------- --- -- - -- -- ------- -- ------ -- --- --- --- - - ---- ------ -------- ----- - ---------- : Subleading c o rr e c ti o n-to -sc aling expon ent$ \ome ga$. [ **5 . Stab ility ** ] {} \ Next, we analyse t he local ity a nd stabilitystructureo f flows an d sh o wt hat the non-un ivers al propert i es of RG flow s in the vicinity o f a scal ing so lut ion a re vastly diffe r e nt.Co nsiderthe fixedpoi ntsol uti on itself,which is n on -u ni ver sal a n d not me as ura bl e i n any experi ment. Usi ng ,t henontriv i al s cali ng s olut ion w ith $ u'_\ s tar \neq$const. an d $ \ part ia l_ t u'_\s tar=0$ of the P olchinskiRG ob eys th e differen tial equation \[FlowPRG u pScal\] 2u\_’- (d -2)u\_” = 2(u \_’)\^ 2 - 02N u\ _”’ -( 1+02N -4 u\_ ’ ) u\_”. F or la rg e fields,t h e s calin gpote ntial b ehaves as \[soluti o nPR GExpansionLar ge\ ] u\ _ ( )+[s u bl e adi ng ] {}. A n analytical so lution for $ N =\ infty$ has bee ngiven i n [@Kub yshin : 2001gz] . On thelevel ofth e RG f low , this beh aviour s tems from a can c el latio n b etween t hecanon ical s c ali ng of the p ot ential andit s non-li near renormalisation. F rom, w e con clu de that t hen on- trivial q uant um contrib uti ons to t heP olchi nski fl owd iverg es l i ke \[PRG- l ar gef i e ld \] \_t

universality_classes! This_high degree of coincidence_leads to_the_important conjecture_that_the universal content_of the partial_differential equations and is_equivalent [^1].\ _----------------_--------------------------------- ------------------------------------ $\quad N\quad$ $\quad {\rm Polchinski}\quad$ __$\quad\quad R_{\rm_opt}\quad\quad$ __ \[.5ex\] $0$ _ ${}\quad$ ${}\quad$ — _ _ __ _ $\ 0.65788^{a}\ \ \ {}$ _ _$1$ ___ ${}\quad$_$\ 0.65574^{c,e}\ {}$ _ _ $\ 0.655746^{a}\ {}$ _ $2$ _ _ ${}\quad$_$\ 0.6712^{d}\ {}$ _ _ _ _ $\ 0.671221^{a}\ {}$ _ _ $3$ __ ___${}\quad$ $\_0.6998^{d}\ {}$ __ __ __ $\ 0.699837^{a}\ {}$ _ _$4$_ _ ${}\quad$ $\_0.7338^{d}\ {}$ _ _ _ $\ 0.733753^{a}\ {}$ ---------------- ---------------------------------_------------------------------------ : Subleading correction-to-scaling exponent_$\omega$. [**5. Stability**]{}\ Next, we analyse_the locality_and_stability structure of_flows_and show_that the non-universal properties of RG flows_in the_vicinity of a scaling solution are_vastly different. Consider the_fixed_point solution itself, which is non-universal_and not measurable in any experiment._Using, the nontrivial scaling solution_with_$u'_\star_\neq$ const. and $\partial_t u'_\star=0$ of the_Polchinski RG obeys the differential equation_\[FlowPRGupScal\] 2 u\_’_- (d-2)u\_” =2(u\_’)\^2 - 02N u\_”’ -(1+02N-4u\_’)u\_”._For_large fields, the scaling potential behaves_as_\[solutionPRGExpansionLarge\] u\_() +[subleading]{}. An analytical solution_for_$N=\infty$_has been given in [@Kubyshin:2001gz]._On the level of the RG_flow, this behaviour stems from a cancellation between the_canonical scaling of_the potential and its non-linear_renormalisation._From,_we conclude that the non-trivial quantum contributions to the Polchinski_flow diverges_like \[PRG-largefield\] \_t

k \leso{t} m$ if one of the following cases holds: (i) $k=1$ or $k=m$, (ii) $k,m \in {\N^{\scriptscriptstyle \vee}_{t}}\setminus \{1\}$ and $m/t \gtso{\Sho} k/t $, (iii) $k \in {\N^{\scriptscriptstyle \vee}_{t}}$ and $m \notin {\N^{\scriptscriptstyle \vee}_{t}}$, (iv) $k,m \notin {\N^{\scriptscriptstyle \vee}_{t}}$ and $k = i m + j t$ with $i,j \in \N$, where in case (ii) we use the following arithmetic rule for the symbol ${\ifxt\empty\else t\cdot \fi2^{\infty}}$: ${\ifxt\empty\else t\cdot \fi2^{\infty}}/t = {\ifx\empty\empty\else \empty\cdot \fi2^{\infty}}$. There are two parts in the structure of the orderings $\leso{t}$. The smallest part consists of all elements of ${\N^{\scriptscriptstyle \vee}_{t}}$ ordered as follows. The smallest element is 1. Then all the multiples of $t$ (including ${\ifxt\empty\else t\cdot \fi2^{\infty}}$) come in the ordering induced by the Sharkovsky ordering and the largest element of ${\N^{\scriptscriptstyle \vee}_{t}}$ is $3\cdot t$. Then the ordering $\geso{t}$ divides $\N_t \setminus {\N^{\scriptscriptstyle \vee}_{t}}$ into $t-1$ “branches”. The $l$-th branch ($l \in \{1,2,\dots,t-1\}$) is formed by all positive integers (except $l$) which are congruent to $l$ modulo $t$ in decreasing order. All elements of these branches are larger than all elements of ${\N^{\scriptscriptstyle \vee}_{t}}$. We note that, by means of the inclusion of the symbol ${\ifxt\empty\else t\cdot \fi2^{\infty}}$, each subset of $\N_t$ has a maximal element with respect to the ordering $\leso{t}$. We also note that the ordering $\leso{2}$ on $\N_2$ coincides with the Sharkovsky ordering on $\N_{\Sho}$ (by

k \leso{t } m$ if one of the following cases holds: (i) $ k=1 $ or $ thousand = m$, (ii) $ k, m \in { \N^{\scriptscriptstyle \vee}_{t}}\setminus \{1\}$ and $ m / t \gtso{\Sho } k / t $, (iii) $ k \in { \N^{\scriptscriptstyle \vee}_{t}}$ and $ m \notin { \N^{\scriptscriptstyle \vee}_{t}}$, (iv) $ k, m \notin { \N^{\scriptscriptstyle \vee}_{t}}$ and $ thousand = i m + j t$ with $ i, j \in \N$, where in case   (ii) we use the following arithmetical rule for the symbol $ { \ifxt\empty\else t\cdot \fi2^{\infty}}$: $ { \ifxt\empty\else t\cdot \fi2^{\infty}}/t = { \ifx\empty\empty\else \empty\cdot \fi2^{\infty}}$. There are two parts in the social organization of the order $ \leso{t}$. The smallest region consists of all elements of $ { \N^{\scriptscriptstyle \vee}_{t}}$ order as follows. The smallest element is 1. Then all the multiples of $ t$ (including $ { \ifxt\empty\else t\cdot \fi2^{\infty}}$) come in the order induced by the Sharkovsky ordering and the largest component of $ { \N^{\scriptscriptstyle \vee}_{t}}$ is $ 3\cdot t$. Then the ordering $ \geso{t}$ divides $ \N_t \setminus { \N^{\scriptscriptstyle \vee}_{t}}$ into $ t-1 $ “ branch ”. The $ l$-th branch ($ liter \in \{1,2,\dots, t-1\}$) is formed by all positive integers (except $ l$) which are congruent to $ l$ modulo $ t$ in decreasing orderliness. All elements of these branches are larger than all elements of $ { \N^{\scriptscriptstyle \vee}_{t}}$. We note that, by mean of the inclusion body of the symbol $ { \ifxt\empty\else t\cdot \fi2^{\infty}}$, each subset of $ \N_t$ hold a maximal element with respect to the ordering $ \leso{t}$. We also note that the ordering $ \leso{2}$ on $ \N_2 $ coincides with the Sharkovsky ordering on $ \N_{\Sho}$ (by

k \lfso{t} m$ if one of the foluowing cases holds: (i) $k=1$ mr $k=m$, (ij) $k,m \in {\V^{\scriptscriptstyle \vee}_{t}}\setmiius \{1\}$ and $n/t \gtso{\Sho} k/t $, (iii) $k \in {\N^{\scriptsbriptstylw \vet}_{t}}$ and $m \notin {\N^{\srdiptscrl'tstymc \vee}_{c}}$, (it) $k,m \notin {\N^{\scrlptscriptstfle \vee}_{t}}$ and $k = k l + j t$ with $i,j \in \N$, where in case (ii) re use yhf following arythmttis rumv nor the symbol ${\ifxt\empty\else t\cdkt \fi2^{\inhty}}$: ${\ifxt\empty\elxe t\cdot \fi2^{\infty}}/t = {\ifx\emptj\emphy\else \empty\cdot \fl2^{\infty}}$. There are rwo parts in the strucuuxe of the otderings $\leso{t}$. The smallest part consnsts of all eoemftts of ${\N^{\scrmptscrpptstyle \vee}_{t}}$ ordered as folkows. The smallcst enemwnt is 1. Then all the kultiples of $t$ (incjuding ${\ifft\zmpty\else t\cdot \fi2^{\infry}}$) come in dhe uedefinf mndhced bj tie Sharkovsiy ordering and the largest eltmegn of ${\N^{\scriptsdriptseyje \vee}_{t}}$ is $3\cdot t$. Then the ordering $\gesm{t}$ sivides $\N_t \setminus {\N^{\sceiptscriptstyle \vee}_{t}}$ lnto $t-1$ “brwnches”. The $l$-th branch ($l \in \{1,2,\dots,t-1\}$) is formed by all posivixe nkbegefw (fxcept $l$) which are congruent to $l$ modulo $t$ in sevrvasing order. All clements of these nrwnvres are larget than cml elements of ${\N^{\scriotscripjstyle \vee}_{t}}$. We njte yhat, by means of the inclusuon of the sjmboo ${\ifxt\empty\else t\cbot \fi2^{\infty}}$, zach sobset pf $\N_t$ has a maximal eleoent with respeft to the urdering $\leso{t}$. Wd aksm note that the ordering $\lqso{2}$ on $\N_2$ roincndes witf thg Sharkjvsky ordeging on $\N_{\Sho}$ (by

k \leso{t} m$ if one of the holds: $k=1$ or (ii) $k,m \in \gtso{\Sho} $, (iii) $k {\N^{\scriptscriptstyle \vee}_{t}}$ and \notin {\N^{\scriptscriptstyle \vee}_{t}}$, (iv) $k,m \notin \vee}_{t}}$ and $k = i m + j t$ with $i,j \in \N$, in case (ii) we use the following arithmetic rule for the symbol ${\ifxt\empty\else \fi2^{\infty}}$: t\cdot = \empty\cdot \fi2^{\infty}}$. There are two parts in the structure of the orderings $\leso{t}$. The smallest part of all elements of ${\N^{\scriptscriptstyle \vee}_{t}}$ ordered as The smallest element is Then all the multiples of (including t\cdot \fi2^{\infty}}$) in ordering by the Sharkovsky and the largest element of ${\N^{\scriptscriptstyle \vee}_{t}}$ is $3\cdot t$. Then the ordering $\geso{t}$ divides $\N_t \setminus \vee}_{t}}$ into The $l$-th ($l \{1,2,\dots,t-1\}$) formed by all (except $l$) which are congruent to in decreasing order. All elements of these branches larger than elements of ${\N^{\scriptscriptstyle \vee}_{t}}$. We note by means of the inclusion of the symbol t\cdot \fi2^{\infty}}$, each subset of $\N_t$ has a maximal element with respect to the ordering also note that the $\leso{2}$ on $\N_2$ with Sharkovsky on (by

k \leso{t} m$ if one of the followinG cases holdS: (i) $k=1$ or $K=m$, (iI) $k,m \In {\n^{\scrIptsCriptstyle \vee}_{t}}\SEtmiNus \{1\}$ and $m/t \gtso{\Sho} k/t $, (iii) $k \iN {\N^{\scrIpTScriPTsTyle \vEe}_{t}}$ and $m \NOtIN {\n^{\scRiPtScrIpTStYle \veE}_{t}}$, (iV) $k,m \notiN {\N^{\scriptscRipTsTyle \vee}_{t}}$ and $k = I M + j T$ with $i,j \in \N$, WheRe in case (ii) we Use The folLoWinG ArithMetIc rulE for thE Symbol ${\Ifxt\empty\ElSE t\cdot \FI2^{\infty}}$: ${\iFXT\eMpty\Else t\cdot \fi2^{\infty}}/t = {\IFx\EMpty\empty\else \eMpty\cdOt \FI2^{\iNFTy}}$. THerE are two parTs In the STructurE Of THE OrdERings $\leso{t}$. The Smallest parT ConSists oF aLl eLEments Of ${\N^{\scRiPTscRiptstyle \veE}_{t}}$ orDered as foLlows. THE smalleST elemenT is 1. TheN alL thE mulTIpLeS of $T$ (iNCluDInG ${\ifXT\emPty\else t\CdOt \Fi2^{\infTy}}$) coME IN The oRdeRing InducEd by the SharkoVskY ordERinG and tHe larGest ElEment Of ${\N^{\scrIptscRiPtstyle \vee}_{t}}$ is $3\cdOt t$. THen the ordEriNg $\GesO{t}$ DividES $\N_t \setMinUs {\N^{\ScriptsCriptstYLe \vEe}_{T}}$ INTo $T-1$ “branches”. The $l$-th braNcH ($L \In \{1,2,\Dots,t-1\}$) is fOrmed bY AlL pOSitive inTeGerS (excEPT $l$) whiCh arE CoNgruent tO $l$ moduLO $t$ In DecreasInG order. alL elEmeNts of THese BranchEs are larGer thAN all elements of ${\n^{\ScriptscriptsTYlE \VEe}_{T}}$. we noTe tHat, by means oF the INcluSion OF tHe sYMbol ${\iFxt\emPtY\ElSE t\cdot \fi2^{\infty}}$, each suBsEt of $\N_t$ Has a mAximal element With respecT TO The orderIng $\lESo{T}$. we also note that The orDering $\leso{2}$ ON $\N_2$ coinciDes wiTh the ShaRkovsky orDERing on $\N_{\SHo}$ (bY

k \leso{t} m$ if one of th e followin g cas eshol ds : ( i) $ k=1$ or $k=m$, (ii ) $k,m \in {\N^{\scrip tscri pt s tyle \v ee}_{ t}}\set m in u s \{ 1\ }$ an d$ m/ t \gt so{ \Sho} k /t $, (ii i)$k \in {\N^{\s c ri ptscriptst yle \vee}_{t}}$ an d $m \ no tin {\N^{ \sc ripts cripts t yle \v ee}_{t}}$ ,(iv) $ k ,m \not i n { \N^{ \scriptscriptstyl e \ v ee}_{t}}$ and$k = i m +j t$wit h $i,j \in \ N$, w here in ca s e (ii ) we use the f ollowing ar i thm etic r ul e f o r thesymbo l$ {\i fxt\empty\e lset\cdot \f i2^{\i n fty}}$: ${\ifxt \empty \el set\cd o t\f i2^ {\ i nft y }} /t= {\ ifx\empt y\ em pty\e lse\ e m p ty\c dot \fi 2^{\i nfty}}$. The rearet woparts in t he s tr uctur e of t he or de rings $\leso{t} $. T he smalle stpa rtco nsist s of al l e lem ents of ${\N^{ \ scr ip t s c ri ptstyle \vee}_{t}} $o r de red as f ollows . T he smallest e lem enti s 1. T hena ll the mul tiples of $ t$ (inc lu ding $ {\ ifx t\e mpty\ e lset\cdot \fi2^{\ infty } }$) come in th e ordering ind u ce d by theSha rkovsky ord erin g and the la rge s t ele mentof ${ \ N^{\scriptscriptsty le \vee} _{t}} $ is $3\cdott$. Then t h e ordering $\g e so { t}$ divides $\ N_t \ setminus { \ N^{\scri ptscr iptstyle \vee}_{t } } $ into $ t-1 $ “ bra nch e s ”. The $l$-th b r a nch($ l \in \ {1, 2,\dots ,t- 1\} $)isfo rmed by a ll posit iv ein te ger s (ex c ept $l$) w hic hare cong r uent t o $l$ mod ul o$ t$in decr e as i n g or de r. All el em entsof t h ese branch es are la rge r tha nal l eleme nts of ${\N^{ \s criptscrip ts tyl e \vee } _ {t}}$. We note that, by meanso f the i ncl usion ofthe symbo l $ {\ifxt \em p ty\els e t\cd ot \f i2 ^{\ i n fty}} $ , e ach s ubset of $ \ N _t$ hasamaxi mal ele ment with respectt o t he ordering $ \le so{t } $ .Wea ls o no te tha t the ordering $\ leso{2}$ o n$ \N _2$ coinci d eswi th theSharkov sky o r deringon $\N_{\ Sho}$ (by

k \leso{t}_m$ if_one of the following_cases holds: (i)_$k=1$_or $k=m$, (ii)_$k,m_\in {\N^{\scriptscriptstyle \vee}_{t}}\setminus_\{1\}$ and $m/t_\gtso{\Sho} k/t $, (iii) $k_\in {\N^{\scriptscriptstyle \vee}_{t}}$_and_$m \notin {\N^{\scriptscriptstyle \vee}_{t}}$, (iv) $k,m \notin {\N^{\scriptscriptstyle \vee}_{t}}$ and $k = i m +_j_t$ with_$i,j_\in_\N$, where in case (ii) we use_the following arithmetic rule for_the symbol_${\ifxt\empty\else t\cdot \fi2^{\infty}}$: ${\ifxt\empty\else t\cdot \fi2^{\infty}}/t = {\ifx\empty\empty\else_\empty\cdot_\fi2^{\infty}}$. There are two_parts in the structure of the orderings $\leso{t}$. The_smallest part consists of all elements_of ${\N^{\scriptscriptstyle \vee}_{t}}$_ordered_as_follows. The smallest element_is 1. Then all the multiples_of $t$ (including ${\ifxt\empty\else t\cdot \fi2^{\infty}}$)_come in the ordering induced by the_Sharkovsky ordering and the largest element_of ${\N^{\scriptscriptstyle \vee}_{t}}$ is $3\cdot_t$. Then_the ordering $\geso{t}$ divides $\N_t_\setminus {\N^{\scriptscriptstyle \vee}_{t}}$_into $t-1$_“branches”. The $l$-th_branch ($l \in \{1,2,\dots,t-1\}$) is formed_by all positive_integers (except $l$) which are congruent_to_$l$ modulo $t$_in_decreasing_order. All_elements of these_branches_are larger_than_all elements of ${\N^{\scriptscriptstyle \vee}_{t}}$. We note_that,_by means of the inclusion of the_symbol ${\ifxt\empty\else t\cdot \fi2^{\infty}}$,_each_subset of $\N_t$ has_a maximal element with respect_to the ordering $\leso{t}$. We also_note that_the ordering_$\leso{2}$ on $\N_2$ coincides with the Sharkovsky ordering on $\N_{\Sho}$ (by

f)$, and $\chain''$ follows $p_2^{C_2}(G_g)$ in $\chain'$. Then $\chain''$ follows $f\circ g$ in $\chain$ with respect to a $C_1$-admissible permutation of $G_{f\circ g}$ which we will denote by $p_1*p_2$ (see Figures \[fig:3\] and \[fig:4\]). at (1,1) ; at (2,1) ; at (3,1) ; at (4,1) ; (1,0)–(2,0)–(2,0.5)–(1,0.5)–(1,1)–(4,1)–(4,1.5)–(1,1.5); at (1.5,0.1) [$H_0$]{}; at (1.5,0.65) [$H_1$]{}; at (2.5,1.15) [$H_2$]{}; at (2.5,1.65) [$H_3$]{}; at (-0.6,2) [$\chain$]{}; at (4.3,1.7) [$N_{\chain'}$]{}; (6,-0.5)–(6,2.5)–(9,2.5)–(9,-0.5)–(6,-0.5); (6,-0.5)–(6.75,0.75)–(7.5,-0.5)–(8.25,2.5)–(9,-0.5); (5.95,0)–(6.05,0); (5.95,1)–(6.05,1); (5.95,2)–(6.05,2); (6.75,-0.45)–(6.75,-0.55); (7.5,-0.45)–(7.5,-0.55); (8.25,-0.5)–(9,-0.5); (8.25,-0.45)–(8.25,-0.55); at (6.8,-0.7) [$t_1$]{}; at (8.3,-0.7) [$t_3$]{}; at (7.55,-0.7) [$t_2$]{}; at (9.5,2) [$\Gamma_f$]{}; at (5,-1) [$(a)$]{}; at (1,1) ; at (2,1) ; at (

f)$, and $ \chain''$ follows $ p_2^{C_2}(G_g)$ in $ \chain'$. Then $ \chain''$ follows $ f\circ g$ in $ \chain$ with respect to a $ C_1$-admissible permutation of $ G_{f\circ g}$ which we will announce by $ p_1*p_2 $ (examine Figures   \[fig:3\ ] and \[fig:4\ ]). at (1,1); at (2,1); at (3,1); at (4,1); (1,0)–(2,0)–(2,0.5)–(1,0.5)–(1,1)–(4,1)–(4,1.5)–(1,1.5); at (1.5,0.1) [ $ H_0 $ ] { }; at (1.5,0.65) [ $ H_1 $ ] { }; at (2.5,1.15) [ $ H_2 $ ] { }; at (2.5,1.65) [ $ H_3 $ ] { }; at (-0.6,2) [ $ \chain$ ] { }; at (4.3,1.7) [ $ N_{\chain'}$ ] { }; (6,-0.5)–(6,2.5)–(9,2.5)–(9,-0.5)–(6,-0.5); (6,-0.5)–(6.75,0.75)–(7.5,-0.5)–(8.25,2.5)–(9,-0.5); (5.95,0)–(6.05,0); (5.95,1)–(6.05,1); (5.95,2)–(6.05,2); (6.75,-0.45)–(6.75,-0.55); (7.5,-0.45)–(7.5,-0.55); (8.25,-0.5)–(9,-0.5); (8.25,-0.45)–(8.25,-0.55); at (6.8,-0.7) [ $ t_1 $ ] { }; at (8.3,-0.7) [ $ t_3 $ ] { }; at (7.55,-0.7) [ $ t_2 $ ] { }; at (9.5,2) [ $ \Gamma_f$ ] { }; at (5,-1) [ $ (a)$ ] { }; at (1,1); at (2,1); at (

f)$, ajd $\chain''$ follows $p_2^{C_2}(G_g)$ in $\chain'$. Then $\chanb''$ follmws $f\cjrc g$ in $\chain$ with respect to a $C_1$-adliwsiblt permutation of $G_{f\zirc g}$ whpch we wiol dtnote by $p_1*p_2$ (see Fmfures \[fin:3\] and \[nig:4\]). at (1,1) ; at (2,1) ; at (3,1) ; at (4,1) ; (1,0)–(2,0)–(2,0.5)–(1,0.5)–(1,1)–(4,1)–(4,1.5)–(1,1.5); at (1.5,0.1) [$H_0$]{}; ad (1.5,0.65) [$H_1$]{}; at (2.5,1.15) [$H_2$]{}; at (2.5,1.65) [$H_3$]{}; ac (-0.6,2) [$\chain$]{}; at (4.3,1.7) [$N_{\chain'}$]{}; (6,-0.5)–(6,2.5)–(9,2.5)–(9,-0.5)–(6,-0.5); (6,-0.5)–(6.75,0.75)–(7.5,-0.5)–(8.25,2.5)–(9,-0.5); (5.95,0)–(6.05,0); (5.95,1)–(6.05,1); (5.95,2)–(6.05,2); (6.75,-0.45)–(6.75,-0.55); (7.5,-0.45)–(7.5,-0.55); (8.25,-0.5)–(9,-0.5); (8.25,-0.45)–(8.25,-0.55); at (6.8,-0.7) [$t_1$]{}; at (8.3,-0.7) [$t_3$]{}; at (7.55,-0.7) [$t_2$]{}; at (9.5,2) [$\Gamma_f$]{}; at (5,-1) [$(a)$]{}; au (1,1) ; at (2,1) ; at (

f)$, and $\chain''$ follows $p_2^{C_2}(G_g)$ in $\chain'$. follows g$ in with respect to g}$ we will denote $p_1*p_2$ (see Figures and \[fig:4\]). at (1,1) ; at ; at (3,1) ; at (4,1) ; (1,0)–(2,0)–(2,0.5)–(1,0.5)–(1,1)–(4,1)–(4,1.5)–(1,1.5); at (1.5,0.1) [$H_0$]{}; at (1.5,0.65) at (2.5,1.15) [$H_2$]{}; at (2.5,1.65) [$H_3$]{}; at (-0.6,2) [$\chain$]{}; at (4.3,1.7) [$N_{\chain'}$]{}; (6,-0.5)–(6,2.5)–(9,2.5)–(9,-0.5)–(6,-0.5); (5.95,0)–(6.05,0); (5.95,2)–(6.05,2); (7.5,-0.45)–(7.5,-0.55); (8.25,-0.45)–(8.25,-0.55); at (6.8,-0.7) [$t_1$]{}; at (8.3,-0.7) [$t_3$]{}; at (7.55,-0.7) [$t_2$]{}; at (9.5,2) [$\Gamma_f$]{}; at (5,-1) [$(a)$]{}; (1,1) ; at (2,1) ; at (

f)$, and $\chain''$ follows $p_2^{C_2}(G_g)$ in $\chaIn'$. Then $\chaiN''$ follOws $F\ciRc G$ in $\cHain$ With respect to a $c_1$-AdmiSsible permutation of $G_{f\cIrc g}$ wHiCH we wILl DenotE by $p_1*p_2$ (seE fiGURes \[FiG:3\] aNd \[fIg:4\]). AT (1,1) ; aT (2,1) ; at (3,1) ; at (4,1) ; (1,0)–(2,0)–(2,0.5)–(1,0.5)–(1,1)–(4,1)–(4,1.5)–(1,1.5); At (1.5,0.1) [$H_0$]{}; At (1.5,0.65) [$H_1$]{}; at (2.5,1.15) [$H_2$]{}; aT (2.5,1.65) [$H_3$]{}; at (-0.6,2) [$\chain$]{}; aT (4.3,1.7) [$N_{\cHaIn'}$]{}; (6,-0.5)–(6,2.5)–(9,2.5)–(9,-0.5)–(6,-0.5); (6,-0.5)–(6.75,0.75)–(7.5,-0.5)–(8.25,2.5)–(9,-0.5); (5.95,0)–(6.05,0); (5.95,1)–(6.05,1); (5.95,2)–(6.05,2); (6.75,-0.45)–(6.75,-0.55); (7.5,-0.45)–(7.5,-0.55); (8.25,-0.5)–(9,-0.5); (8.25,-0.45)–(8.25,-0.55); at (6.8,-0.7) [$t_1$]{}; at (8.3,-0.7) [$t_3$]{}; at (7.55,-0.7) [$t_2$]{}; aT (9.5,2) [$\gaMma_f$]{}; at (5,-1) [$(a)$]{}; at (1,1) ; aT (2,1) ; at (

f)$, and $\chain''$ follow s $p_2^{C_ 2}(G_ g)$ in $ \cha in'$ . Then $\chain ' '$ f ollows $f\circ g$ in $ \chai n$ with re spect to a $ C _1 $ - adm is si ble p e rm utati onof $G_{ f\circ g}$ wh ic h we will de n ot e by $p_1* p_2 $ (see Figur es\[fig: 3\ ] a n d \[f ig: 4\]). at ( 1 ,1) ;at (2,1);a t (3,1 ) ; at ( 4 , 1) ; (1,0)–(2,0)–(2,0. 5 )– ( 1,0.5)–(1,1)–( 4,1)–( 4, 1 .5 ) – (1, 1.5 ); at (1.5 ,0 .1) [ $ H_0$]{} ; a t ( 1.5 , 0.65) [$H_1$] {}; at (2.5 , 1.1 5) [$H _2 $]{ } ; at ( 2.5,1 .6 5 ) [ $H_3$]{}; a t (- 0.6,2) [$ \chain $ ]{}; at (4.3,1. 7) [$N _{\ cha in'} $ ]{ }; ( 6, - 0.5 ) –( 6,2 . 5)– (9,2.5)– (9 ,- 0.5)– (6,- 0 . 5 ) ; (6 ,-0 .5)– (6.75 ,0.75)–(7.5,- 0.5 )–(8 . 25, 2.5)– (9,-0 .5); ( 5.95, 0)–(6. 05,0) ;(5.95,1)–(6.05, 1);(5.95,2)– (6. 05 ,2) ;(6.75 , -0.45) –(6 .75 ,-0.55) ; (7.5, - 0.4 5) – ( 7 .5 ,-0.55); (8.25,-0. 5) – ( 9, -0.5); ( 8.25,- 0 .4 5) – (8.25,-0 .5 5); at( 6 .8,-0 .7)[ $t _1$]{};at (8. 3 ,- 0. 7) [$t_ 3$ ]{}; a t(7. 55, -0.7) [$t_ 2$]{}; at (9.5 ,2) [ $ \Gamma_f$]{};a t (5,-1) [$(a ) $] { } ;at ( 1,1 ) ; at (2,1 ) ;a t (

f)$, and_$\chain''$ follows_$p_2^{C_2}(G_g)$ in $\chain'$. Then_$\chain''$ follows_$f\circ_g$ in_$\chain$_with respect to_a $C_1$-admissible permutation_of $G_{f\circ g}$ which_we will denote_by_$p_1*p_2$ (see Figures \[fig:3\] and \[fig:4\]). at (1,1) ; at (2,1) ; at (3,1) ; at_(4,1)_; (1,0)–(2,0)–(2,0.5)–(1,0.5)–(1,1)–(4,1)–(4,1.5)–(1,1.5); at_(1.5,0.1)_[$H_0$]{};_at (1.5,0.65) [$H_1$]{}; at (2.5,1.15)_[$H_2$]{}; at (2.5,1.65) [$H_3$]{}; at_(-0.6,2) [$\chain$]{};_at (4.3,1.7) [$N_{\chain'}$]{}; (6,-0.5)–(6,2.5)–(9,2.5)–(9,-0.5)–(6,-0.5); (6,-0.5)–(6.75,0.75)–(7.5,-0.5)–(8.25,2.5)–(9,-0.5); (5.95,0)–(6.05,0); (5.95,1)–(6.05,1); (5.95,2)–(6.05,2); (6.75,-0.45)–(6.75,-0.55);_(7.5,-0.45)–(7.5,-0.55);_(8.25,-0.5)–(9,-0.5); (8.25,-0.45)–(8.25,-0.55); at_(6.8,-0.7) [$t_1$]{}; at (8.3,-0.7) [$t_3$]{}; at (7.55,-0.7) [$t_2$]{}; at_(9.5,2) [$\Gamma_f$]{}; at (5,-1) [$(a)$]{}; at (1,1)_; at (2,1)_;_at_(

}G))$ is an involutive automorphism. Write $\theta_Z$ for the involutive automorphism of $Z({}^{\vee}G)$ defined by the action of any automorphism of ${}^{\vee}G$ corresponding to $a$ under $\Psi_{0}$, and suppose $$\overline{z}\in Z({}^{\vee}G)^{\theta_Z}/(1+\theta_Z)Z({}^{\vee}G).$$ Then from point $c)$ of Proposition 4.4 of [@ABV], there is a weak $E$-group with invariants $(a,\overline{z})$. \[deftn:egroup\] Suppose $G$ is a complex connected reductive algebraic group. An **$E$-group** for $G$ is a pair $\left({}^{\vee}G^{\Gamma},\mathcal{S}\right)$, subject to the following conditions. 1. ${}^{\vee}G^{\Gamma}$ is a weak $E$-group for $G$. 2. $\mathcal{S}$ is a conjugacy class of pairs $({}^{\vee}\delta,{}^{d}B)$ with ${}^{\vee}\delta$ an element of finite order in ${}^{\vee}G^{\Gamma}-{}^{\vee}{G}$ and ${}^{d}B$ a Borel subgroup of ${}^{\vee}G$. 3. Suppose $\left({}^{\vee}\delta,{}^{d}B\right)\in \mathcal{S}$. Then conjugation by ${}^{\vee}{\delta}$ is a distinguished involutive automorphism ${}^{\vee}\sigma$ of ${}^{\vee}G$ preserving ${}^{d}B$. The invariants of the $E$-group are the automorphism $a$ attached to ${}^{\vee}G^{\Gamma}$ as before and the element $$z={}^{\vee}\delta^{2}\in Z({}^{\vee}G)^{\theta_Z}$$ with ${}^{\vee}\delta$ any element in $\left({}^{\vee}\delta,{}^{d}B\right)$. An **$L$-group** for $G$ is an $E$-group whose second invariant is equal to $1$, that is to say $z={}^{\vee}\delta^{2}=1$. Just as for weak $E$-groups, $E$-groups can be completely classified by the couple of invariants $(a,z)$ described above. From point $a)$ of Proposition 4.7 of [@ABV], two $E$-groups

} G))$ is an involutive automorphism. Write $ \theta_Z$ for the involutive automorphism of $ Z({}^{\vee}G)$ defined by the action of any automorphism of $ { } ^{\vee}G$ corresponding to $ a$ under $ \Psi_{0}$, and presuppose $ $ \overline{z}\in Z({}^{\vee}G)^{\theta_Z}/(1+\theta_Z)Z({}^{\vee}G).$$ Then from item $ c)$ of Proposition 4.4 of [ @ABV ], there is a weak $ E$-group with invariants $ (a,\overline{z})$. \[deftn: egroup\ ] speculate $ G$ is a complex connected reductive algebraic group. An * * $ E$-group * * for $ G$ is a pair $ \left({}^{\vee}G^{\Gamma},\mathcal{S}\right)$, capable to the following conditions. 1. $ { } ^{\vee}G^{\Gamma}$ is a weak $ E$-group for $ G$. 2. $ \mathcal{S}$ is a conjugacy course of pair $ ({ } ^{\vee}\delta,{}^{d}B)$ with $ { } ^{\vee}\delta$ an element of finite club in $ { } ^{\vee}G^{\Gamma}-{}^{\vee}{G}$ and $ { } ^{d}B$ a Borel subgroup of $ { } ^{\vee}G$. 3. Suppose $ \left({}^{\vee}\delta,{}^{d}B\right)\in \mathcal{S}$. Then conjugation by $ { } ^{\vee}{\delta}$ is a distinguished involutive automorphism $ { } ^{\vee}\sigma$ of $ { } ^{\vee}G$ preserve $ { } ^{d}B$. The invariant of the $ E$-group are the automorphism $ a$ attached to $ { } ^{\vee}G^{\Gamma}$ as before and the chemical element $ $ z={}^{\vee}\delta^{2}\in Z({}^{\vee}G)^{\theta_Z}$$ with $ { } ^{\vee}\delta$ any element in $ \left({}^{\vee}\delta,{}^{d}B\right)$. An * * $ L$-group * * for $ G$ is an $ E$-group whose second invariant is adequate to $ 1 $, that is to say $ z={}^{\vee}\delta^{2}=1$. Just as for weak $ E$-groups, $ E$-groups can be wholly classified by the couple of invariants $ (a, z)$ described above. From point $ a)$ of Proposition 4.7 of [ @ABV ], two $ E$-groups

}G))$ id an involutive automorpmism. Write $\theta_E$ dor thx involhtive augomorphism of $Z({}^{\vee}G)$ defined uy tye acupon of any automorphirm of ${}^{\vee}H$ correspondmng to $a$ under $\Psi_{0}$, and sm'pose $$\lvernmne{z}\in Z({}^{\vee}G)^{\theja_Z}/(1+\theta_Z)Z({}^{\vea}G).$$ Then from pmivt $c)$ of Proposition 4.4 of [@ABV], there is w weak $R$-ggoup with invatiantx $(a,\ovsgllne{z})$. \[deftn:egroup\] Suppose $G$ is a ckmplex bonnected reductife algebraic group. An **$E$-grokp** flr $G$ is a pair $\lefh({}^{\vee}G^{\Gamma},\mqthcwo{S}\right)$, subjdct to the following cknditions. 1. ${}^{\vee}G^{\Gamma}$ is a weak $D$-grou' for $G$. 2. $\majkxal{D}$ is a conjujacy cjass of pairs $({}^{\vee}\delda,{}^{d}B)$ wiyh ${}^{\vee}\delta$ an elxmenr of finite order in ${}^{\tee}G^{\Gamma}-{}^{\vee}{G}$ and ${}^{d}B$ a Borel vuygroup of ${}^{\vee}G$. 3. Suppowe $\left({}^{\eee}\dalta,{}^{a}V\rieht)\jn \mzthcal{D}$. Tien conjugafion by ${}^{\vee}{\eelta}$ is a distinguosrvc involutive automjr[hism ${}^{\vee}\sigma$ of ${}^{\vee}G$ preserving ${}^{d}B$. The inbariants of the $E$-group qre the automorphism $w$ attacheq to ${}^{\vee}G^{\Gamma}$ as before and the element $$z={}^{\vee}\deltd^{2}\in Z({}^{\ted}G)^{\tkcba_Z}$$ dutj ${}^{\vee}\delta$ any element in $\left({}^{\vee}\delta,{}^{d}B\right)$. Wh **$K$-ggoup** for $G$ is an $C$-group whose seconc lnfwriant is equxl to $1$, thzt is to say $z={}^{\vee}\dflta^{2}=1$. Jusj as fir weak $E$-drouls, $E$-groups can be completelt classified vy the couple of iuvariants $(a,z)$ desvribec above. From point $a)$ of Prolosition 4.7 ov [@ABV], two $D$-groups

}G))$ is an involutive automorphism. Write $\theta_Z$ involutive of $Z({}^{\vee}G)$ by the action corresponding $a$ under $\Psi_{0}$, suppose $$\overline{z}\in Z({}^{\vee}G)^{\theta_Z}/(1+\theta_Z)Z({}^{\vee}G).$$ from point $c)$ of Proposition 4.4 [@ABV], there is a weak $E$-group with invariants $(a,\overline{z})$. \[deftn:egroup\] Suppose $G$ is complex connected reductive algebraic group. An **$E$-group** for $G$ is a pair $\left({}^{\vee}G^{\Gamma},\mathcal{S}\right)$, to following 1. is a weak $E$-group for $G$. 2. $\mathcal{S}$ is a conjugacy class of pairs $({}^{\vee}\delta,{}^{d}B)$ with an element of finite order in ${}^{\vee}G^{\Gamma}-{}^{\vee}{G}$ and a Borel subgroup of 3. Suppose $\left({}^{\vee}\delta,{}^{d}B\right)\in \mathcal{S}$. Then by is a involutive ${}^{\vee}\sigma$ ${}^{\vee}G$ preserving ${}^{d}B$. invariants of the $E$-group are the automorphism $a$ attached to ${}^{\vee}G^{\Gamma}$ as before and the element $$z={}^{\vee}\delta^{2}\in with ${}^{\vee}\delta$ in $\left({}^{\vee}\delta,{}^{d}B\right)$. **$L$-group** $G$ an $E$-group whose is equal to $1$, that is Just as for weak $E$-groups, $E$-groups can be classified by couple of invariants $(a,z)$ described above. point $a)$ of Proposition 4.7 of [@ABV], two

}G))$ is an involutive automorphiSm. Write $\theTa_Z$ foR thE inVoLutiVe auTomorphism of $Z({}^{\vEE}G)$ deFined by the action of any aUtomoRpHIsm oF ${}^{\VeE}G$ corRespondINg TO $A$ unDeR $\PSi_{0}$, aNd SUpPose $$\oVerLine{z}\in z({}^{\vee}G)^{\theta_z}/(1+\thEtA_Z)Z({}^{\vee}G).$$ Then fROm Point $c)$ of PrOpoSition 4.4 of [@ABV], tHerE is a weAk $e$-grOUp witH inVariaNts $(a,\ovERline{z})$. \[Deftn:egroUp\] sUppose $g$ Is a compLEX cOnneCted reductive algeBRaIC group. An **$E$-group** For $G$ is A pAIr $\LEFt({}^{\vEe}G^{\gamma},\mathcAl{s}\righT)$, Subject TO tHE FOllOWing conditionS. 1. ${}^{\vee}G^{\Gamma}$ iS A weAk $E$-groUp For $g$. 2. $\MathcaL{S}$ is a CoNJugAcy class of pAirs $({}^{\Vee}\delta,{}^{d}b)$ with ${}^{\vEE}\delta$ aN Element Of finiTe oRdeR in ${}^{\vEE}G^{\gaMma}-{}^{\VeE}{g}$ anD ${}^{D}B$ A BoREl sUbgroup oF ${}^{\vEe}g$. 3. SuppOse $\lEFT({}^{\VEe}\deLta,{}^{D}B\riGht)\in \Mathcal{S}$. Then cOnjUgatIOn bY ${}^{\vee}{\dElta}$ iS a diStInguiShed inVolutIvE automorphism ${}^{\veE}\sigMa$ of ${}^{\vee}G$ pResErVinG ${}^{d}b$. The iNVarianTs oF thE $E$-group Are the aUTomOrPHISm $A$ attached to ${}^{\vee}G^{\GamMa}$ AS BeFore and tHe elemENt $$Z={}^{\vEE}\delta^{2}\in z({}^{\vEe}G)^{\ThetA_z}$$ With ${}^{\vEe}\deLTa$ Any elemeNt in $\leFT({}^{\vEe}\Delta,{}^{d}B\RiGht)$. An **$L$-GrOup** For $g$ is an $e$-GrouP whose Second inVariaNT is equal to $1$, that IS to say $z={}^{\vee}\delTA^{2}=1$. JUST aS For wEak $e$-groups, $E$-groUps cAN be cOmplETeLy cLAssifIed by ThE CoUPle of invariants $(a,z)$ deScRibed aBove. FRom point $a)$ of PrOposition 4.7 oF [@abv], two $E$-groUps

}G))$ is an involutive aut omorphism. Writ e $ \th et a_Z$ for the involutiv e aut omorphism of $Z({}^{\v ee}G) $d efin e dby th e actio n o f any a ut omo rp h is m of${} ^{\vee} G$ corresp ond in g to $a$ und e r$\Psi_{0}$ , a nd suppose $ $\o verlin e{ z}\ i n Z({ }^{ \vee} G)^{\t h eta_Z} /(1+\thet a_ Z )Z({}^ { \vee}G) . $ $Then from point $c)$o fP roposition 4.4 of [@ AB V ], t her e i s a weak $ E$ -grou p with i n va r i a nts $(a,\overline {z})$. \[d e ftn :egrou p\ ] S u ppose$G$ i sa co mplex conne cted reductiv e alge b raic gr o up. An**$E$- gro up* * fo r $ G$ is a pai r $ \le f t({ }^{\vee} G^ {\ Gamma },\m a t h c al{S }\r ight )$, s ubject to the fo llow i ngcondi tions . 1 . ${}^ {\vee} G^{\G am ma}$ is a weak$E$- group for $G $. 2 . $\ma t hcal{S }$isa conju gacy cl a ssof p a ir s $({}^{\vee}\delt a, { } ^{ d}B)$ wi th ${} ^ {\ ve e }\delta$ a n e leme n t of f init e o rder in${}^{\ v ee }G ^{\Gamm a} -{}^{\ ve e}{ G}$ and$ {}^{ d}B$ a Borel s ubgro u p of ${}^{\vee } G$. 3. Supp o se $ \l e ft({ }^{ \vee}\delta ,{}^ { d}B\ righ t )\ in\ mathc al{S} $. Th e n conjugation by ${ }^ {\vee} {\del ta}$ is a dis tinguished i n volutive aut o mo r phism ${}^{\ve e}\si gma$ of ${ } ^{\vee}G $ pre serving${}^{d}B$ . The inv ari ant s o f t h e $ E$-group aret h e au to morphis m $ a$ atta che d t o $ {}^ {\ vee}G^{\G amma}$ a sbe fo re an d the element$$ z={ }^ {\v ee}\d e lta^{2 }\inZ({} ^{ \v e e}G )^{\the t a_ Z } $$ w it h${}^ {\v ee }\del ta$a nyelement in $\lef t({ } ^{\v ee }\ delta,{ }^{d}B\right) $. An **$L$ -g rou p** fo r $G$ is a n $E$-group whose secon d invari ant is e qual to $1$,tha t is t o s a y $z={ }^{\ve e}\de lt a^{ 2 } =1$.J us t a sfor weak $ E $ -gr oups, $ E$-g roups c an be completely c l ass ified by thecou pleo f i nva r ia n ts$( a ,z) $ described above . From poi nt $a )$ of Prop o sit io n 4.7 o f [@ABV ], tw o $E$-gr oups

}G))$ is_an involutive_automorphism. Write $\theta_Z$ for_the involutive_automorphism_of $Z({}^{\vee}G)$_defined_by the action_of any automorphism_of ${}^{\vee}G$ corresponding to_$a$ under $\Psi_{0}$,_and_suppose $$\overline{z}\in Z({}^{\vee}G)^{\theta_Z}/(1+\theta_Z)Z({}^{\vee}G).$$ Then from point $c)$ of Proposition 4.4 of [@ABV], there is_a_weak $E$-group_with_invariants_$(a,\overline{z})$. \[deftn:egroup\] Suppose $G$ is a_complex connected reductive algebraic group._An **$E$-group**_for $G$ is a pair $\left({}^{\vee}G^{\Gamma},\mathcal{S}\right)$, subject to_the_following conditions. 1. _${}^{\vee}G^{\Gamma}$ is a weak $E$-group for $G$. 2. $\mathcal{S}$_is a conjugacy class of pairs_$({}^{\vee}\delta,{}^{d}B)$ with ${}^{\vee}\delta$_an_element_of finite order in_${}^{\vee}G^{\Gamma}-{}^{\vee}{G}$ and ${}^{d}B$ a Borel subgroup_of ${}^{\vee}G$. 3. Suppose $\left({}^{\vee}\delta,{}^{d}B\right)\in \mathcal{S}$._Then conjugation by ${}^{\vee}{\delta}$ is a distinguished_involutive automorphism ${}^{\vee}\sigma$ of ${}^{\vee}G$ preserving_${}^{d}B$. The invariants of the $E$-group_are the_automorphism $a$ attached to ${}^{\vee}G^{\Gamma}$_as before and_the element_$$z={}^{\vee}\delta^{2}\in Z({}^{\vee}G)^{\theta_Z}$$ with_${}^{\vee}\delta$ any element in $\left({}^{\vee}\delta,{}^{d}B\right)$. An **$L$-group**_for $G$ is_an $E$-group whose second invariant is_equal_to $1$, that_is_to_say $z={}^{\vee}\delta^{2}=1$. Just_as for weak_$E$-groups,_$E$-groups can_be_completely classified by the couple of_invariants_$(a,z)$ described above. From point $a)$ of_Proposition 4.7 of [@ABV],_two_$E$-groups

1934+]: $$\begin{aligned} &\tilde{\cal P}(I)\geq 0\;\mbox{for all}\; I\Leftrightarrow \langle f(I)\rangle _{\tilde{\cal P}} = \int\limits^{\infty}_{0} dI \tilde{\cal P}(I) f(I) \geq 0,& \nonumber \\ &\mbox{for all polynomial}\; f(I)\geq 0. &\!\!\!\!\!\!\!\end{aligned}$$ \(ii) Any polynomial $f(I)\/$ pointwise nonnegative over $[0,\infty)\/$ can be written in terms of two perfect square polynomials as [@shohat-book] $$\begin{aligned} f(I) = (f_1(I))^2 + I(f_2(I))^2\;. \end{aligned}$$ To prove necessity, suppose that $\tilde{\cal P}(I)\/$ is pointwise nonnegative. Consider the polynomial $f_1(I)=\sum\limits^{N}_{n=0} c_n I^n\/$, where the $c_n\/$ are arbitrary real coefficients. By (4.4) we have $\langle (f_1(I))^2\rangle _{\tilde{\cal P}}\geq 0\/$. That is, $$\begin{aligned} \langle (f_1(I))^2\rangle _{\tilde{\cal P}} &=& \sum\limits^{N}_{m,n=0} c_m c_n \langle I^{m+n}\rangle _{\tilde{\cal P}} \nonumber\\ &=&\sum\limits^{N}_{m,n=0} c_m c_n q_{m+n}\nonumber\\ &=&\sum\limits^{N}_{m,n=0} c_m c_n L^{(N)}_{mn} \geq 0\;.\end{aligned}$$ This means $L^{(N)}\geq 0\/$ for every $N\geq 0\/$. Similarly, writing $f_2(I)=\sum\limits^{N}_{n=0} d_n I^n\/$ and evaluating the expectation value of the nonnegative polynomial $I(f_2(I))^2\/$ we have from (4.4): $$\begin{aligned} \langle I(f_2(I))^2\rangle _{\tilde{\cal P}} &=& \sum\limits^{N

1934 + ]: $ $ \begin{aligned } & \tilde{\cal P}(I)\geq 0\;\mbox{for all}\; I\Leftrightarrow \langle f(I)\rangle _ { \tilde{\cal P } } = \int\limits^{\infty}_{0 } dI \tilde{\cal P}(I) f(I) \geq 0, & \nonumber \\ & \mbox{for all polynomial}\; f(I)\geq 0. & \!\!\!\!\!\!\!\end{aligned}$$ \(ii) Any polynomial $ f(I)\/$ pointwise nonnegative over $ [ 0,\infty)\/$ can be written in terms of two perfect straight polynomial as   [ @shohat - book ] $ $ \begin{aligned } f(I) = (f_1(I))^2 + I(f_2(I))^2\; . \end{aligned}$$ To prove necessity, presuppose that $ \tilde{\cal P}(I)\/$ is pointwise nonnegative. study the polynomial $ f_1(I)=\sum\limits^{N}_{n=0 } c_n I^n\/$, where the $ c_n\/$ are arbitrary real coefficients. By (4.4) we hold $ \langle (f_1(I))^2\rangle _ { \tilde{\cal P}}\geq 0\/$. That is, $ $ \begin{aligned } \langle (f_1(I))^2\rangle _ { \tilde{\cal P } } & = & \sum\limits^{N}_{m, n=0 } c_m c_n \langle I^{m+n}\rangle _ { \tilde{\cal P } } \nonumber\\ & = & \sum\limits^{N}_{m, n=0 } c_m c_n q_{m+n}\nonumber\\ & = & \sum\limits^{N}_{m, n=0 } c_m c_n L^{(N)}_{mn } \geq 0\;.\end{aligned}$$ This means $ L^{(N)}\geq 0\/$ for every $ N\geq 0\/$. Similarly, spell $ f_2(I)=\sum\limits^{N}_{n=0 } d_n I^n\/$ and evaluating the expectation value of the nonnegative polynomial $ I(f_2(I))^2\/$ we get from (4.4 ): $ $ \begin{aligned } \langle I(f_2(I))^2\rangle _ { \tilde{\cal P } } & = & \sum\limits^{N

1934+]: $$\behin{aligned} &\tilde{\cal P}(I)\geq 0\;\mbox{for all}\; I\Lgfrrighterrow \lahgle f(I)\rxngle _{\tilde{\cal P}} = \int\limits^{\iifty}_{0} dI \tiode{\cal P}(I) f(I) \geq 0,& \nonumcer \\ &\mbox{flr all pilyninial}\; f(I)\geq 0. &\!\!\!\!\!\!\!\end{aligkzd}$$ \(ii) Zky ponbnomial $f(I)\/$ poinjwise nonnegdtive over $[0,\infdy)\/$ ccn be written in terms of two perfece squarr oolynomials as [@fhohse-booi] $$\begin{aligned} f(I) = (f_1(I))^2 + I(f_2(I))^2\;. \end{alighed}$$ To pgove necessity, sulpose that $\tilde{\cal P}(I)\/$ is ooinhwise nonnegative. Fonsider thg pojtnomial $f_1(I)=\suo\limits^{N}_{n=0} b_u I^n\/$, where jhe $c_n\/$ are arbitrary real coeffizientx. By (4.4) we hqvw $\lwtgle (f_1(I))^2\rangoe _{\tijde{\cal P}}\geq 0\/$. That is, $$\begin{akigned} \langle (f_1(L))^2\rangne _{\rilde{\cal P}} &=& \sum\limits^{I}_{m,n=0} c_m c_n \langle I^{m+n}\tangle _{\tilge{\eal P}} \nonumber\\ &=&\sum\limirs^{B}_{m,n=0} c_k c_n q_{m+n}\vinuobed\\ &=&\snm\ljmits^{N}_{l,n=0} r_m c_n L^{(N)}_{mn} \feq 0\;.\end{aligbed}$$ This means $L^{(N)}\gea 0\/$ dor every $N\ges 0\/$. Simylwrly, writing $f_2(I)=\sum\limits^{N}_{n=0} d_n I^n\/$ and evdluzting the expectation vqlue of the nonnegatige polynoiial $I(f_2(I))^2\/$ we have from (4.4): $$\begin{aligned} \langle I(f_2(I))^2\rangne _{\timae{\cco P}} &=& \rym\pimits^{N

1934+]: $$\begin{aligned} &\tilde{\cal P}(I)\geq 0\;\mbox{for all}\; I\Leftrightarrow _{\tilde{\cal = \int\limits^{\infty}_{0} \tilde{\cal P}(I) f(I) all f(I)\geq 0. &\!\!\!\!\!\!\!\end{aligned}$$ Any polynomial $f(I)\/$ nonnegative over $[0,\infty)\/$ can be written terms of two perfect square polynomials as [@shohat-book] $$\begin{aligned} f(I) = (f_1(I))^2 + \end{aligned}$$ To prove necessity, suppose that $\tilde{\cal P}(I)\/$ is pointwise nonnegative. Consider the $f_1(I)=\sum\limits^{N}_{n=0} I^n\/$, the are arbitrary real coefficients. By (4.4) we have $\langle (f_1(I))^2\rangle _{\tilde{\cal P}}\geq 0\/$. That is, $$\begin{aligned} (f_1(I))^2\rangle _{\tilde{\cal P}} &=& \sum\limits^{N}_{m,n=0} c_m c_n \langle _{\tilde{\cal P}} \nonumber\\ &=&\sum\limits^{N}_{m,n=0} c_n q_{m+n}\nonumber\\ &=&\sum\limits^{N}_{m,n=0} c_m c_n \geq This means 0\/$ every 0\/$. Similarly, writing d_n I^n\/$ and evaluating the expectation value of the nonnegative polynomial $I(f_2(I))^2\/$ we have from (4.4): $$\begin{aligned} I(f_2(I))^2\rangle _{\tilde{\cal \sum\limits^{N

1934+]: $$\begin{aligned} &\tilde{\cal P}(I)\geq 0\;\Mbox{for all}\; i\LeftRigHtaRrOw \laNgle F(I)\rangle _{\tilde{\cAL P}} = \inT\limits^{\infty}_{0} dI \tilde{\cal p}(I) f(I) \gEq 0,& \NOnumBEr \\ &\Mbox{fOr all poLYnOMIal}\; F(I)\GeQ 0. &\!\!\!\!\!\!\!\enD{aLIgNed}$$ \(ii) any PolynomIal $f(I)\/$ pointWisE nOnnegative ovER $[0,\iNfty)\/$ can be wRitTen in terms of Two PerfecT sQuaRE polyNomIals aS [@shohaT-Book] $$\beGin{aligneD} f(i) = (F_1(I))^2 + I(f_2(I))^2\;. \eND{aligneD}$$ tO pRove Necessity, suppose tHAt $\TIlde{\cal P}(I)\/$ is poiNtwise NoNNeGATivE. CoNsider the pOlYnomiAL $f_1(I)=\sum\lIMiTS^{n}_{N=0} c_n i^N\/$, where the $c_n\/$ arE arbitrary rEAl cOefficIeNts. bY (4.4) we havE $\langLe (F_1(i))^2\raNgle _{\tilde{\caL P}}\geQ 0\/$. That is, $$\beGin{aliGNed} \langLE (f_1(I))^2\rangLe _{\tildE{\caL P}} &=& \sUm\liMItS^{N}_{M,n=0} c_M c_N \LanGLe i^{m+n}\RAngLe _{\tilde{\cAl p}} \nOnumbEr\\ &=&\suM\LIMIts^{N}_{M,n=0} c_M c_n q_{M+n}\nonUmber\\ &=&\sum\limitS^{N}_{m,N=0} c_m c_N l^{(N)}_{mN} \geq 0\;.\eNd{aliGned}$$ thIs meaNs $L^{(N)}\geQ 0\/$ for eVeRy $N\geq 0\/$. Similarly, WritIng $f_2(I)=\sum\lImiTs^{n}_{n=0} d_N I^N\/$ and eVAluatiNg tHe eXpectatIon valuE Of tHe NONNeGative polynomial $I(f_2(i))^2\/$ wE HAvE from (4.4): $$\begIn{aligNEd} \LaNGle I(f_2(I))^2\raNgLe _{\tIlde{\CAL P}} &=& \sum\LimiTS^{N

1934+]: $$\begin{aligned}&\tilde{\c al P} (I) \ge q0\;\ mbox {for all}\; I\ L eftr ightarrow \langle f(I) \rang le _{\t i ld e{\ca l P}} = \i n t \li mi ts ^{\ in f ty }_{0} dI \tilde {\cal P}(I ) f (I ) \geq 0,& \ n on umber \\ & \mb ox{for all p oly nomial }\ ; f ( I)\ge q 0 . &\! \!\!\! \ !\!\!\ end{align ed } $$ \( i i) Anyp o ly nomi al $f(I)\/$ point w is e nonnegative o ver $[ 0, \ in f t y)\ /$can be wri tt en in terms o f t w o per f ect square po lynomials a s  [@ shohat -b ook ] $$\be gin{a li g ned } f(I) = (f _1(I ))^2 + I( f_2(I) ) ^2\;. \ end{ali gned}$ $ Toprov e n ec ess it y , s u pp ose tha t $\tild e{ \c al P} (I)\ / $ i s po int wise nonn egative. Cons ide r th e po lynom ial $ f_1( I) =\sum \limit s^{N} _{ n=0} c_n I^n\/$ , wh ere the $ c_n \/ $ a re arbi t rary r eal co efficie nts. By (4. 4) w e h ave $\langle (f_1( I) ) ^ 2\ rangle _ {\tild e {\ ca l P}}\geq 0 \/$ . Th a t is,$$\b e gi n{aligne d} \la n gl e(f_1(I) )^ 2\rang le _{ \ti lde{\ c al P }} &=& \sum\li mits^ { N}_{m,n=0} c_m c_n \langle I ^{ m + n} \ rang le_{\tilde{\c al P } } \n onum b er \\& =&\su m\lim it s ^{ N }_{m,n=0} c_m c_n q _{ m+n}\n onumb er\\ &=&\sum\ limits^{N} _ { m ,n=0} c_ m c_ n L ^ {(N)}_{mn} \ge q 0\; .\end{alig n ed}$$ Th is me ans $L^{ (N)}\geq0 \ /$ for e ver y $ N\g eq0 \ /$ . Similarly,w r itin g$f_2(I) =\s um\limi ts^ {N} _{n =0} d _n I^n\/$ and eva lu at in gthe expe c tation v al ueof th e non n egativ e pol ynom ia l$ I(f _2(I))^ 2 \/ $ we h av efrom (4 .4 ): $$ \beg i n{a ligned} \langleI(f _ 2(I) )^ 2\ rangle_{\tilde{\cal P }} &=& \su m\ lim its^{N

1934+]: $$\begin{aligned} &\tilde{\cal_P}(I)\geq 0\;\mbox{for_all}\; I\Leftrightarrow \langle f(I)\rangle _{\tilde{\cal_P}} =_\int\limits^{\infty}_{0}_dI \tilde{\cal P}(I)_f(I)_\geq 0,& \nonumber \\ &\mbox{for_all polynomial}\; f(I)\geq 0._&\!\!\!\!\!\!\!\end{aligned}$$ \(ii) Any polynomial $f(I)\/$_pointwise nonnegative over_$[0,\infty)\/$_can be written in terms of two perfect square polynomials as [@shohat-book] $$\begin{aligned} f(I) = (f_1(I))^2_+_I(f_2(I))^2\;. _\end{aligned}$$ To_prove_necessity, suppose that $\tilde{\cal P}(I)\/$_is pointwise nonnegative. Consider the_polynomial $f_1(I)=\sum\limits^{N}_{n=0}_c_n I^n\/$, where the $c_n\/$ are arbitrary real_coefficients._By (4.4) we_have $\langle (f_1(I))^2\rangle _{\tilde{\cal P}}\geq 0\/$. That is, $$\begin{aligned} \langle_(f_1(I))^2\rangle _{\tilde{\cal P}} &=& \sum\limits^{N}_{m,n=0} c_m c_n_\langle I^{m+n}\rangle__{\tilde{\cal_P}}_\nonumber\\ &=&\sum\limits^{N}_{m,n=0} c_m c_n q_{m+n}\nonumber\\ &=&\sum\limits^{N}_{m,n=0}_c_m c_n L^{(N)}_{mn} \geq 0\;.\end{aligned}$$ This_means $L^{(N)}\geq 0\/$ for every $N\geq_0\/$. Similarly, writing $f_2(I)=\sum\limits^{N}_{n=0} d_n I^n\/$ and_evaluating the expectation value of the_nonnegative polynomial $I(f_2(I))^2\/$ we have_from (4.4):_$$\begin{aligned} \langle I(f_2(I))^2\rangle _{\tilde{\cal P}} &=&_\sum\limits^{N

azimuthal velocity to represent the zonal flow generated in the overlying envelope. The shell rotates around the $z$-axis at the imposed rotation rate $\Omega$. The aspect ratio is $\gamma=r_i/r_o$ where $r_i$ is the inner sphere radius, corresponding to a rocky core, and $r_o$ the outer sphere radius, corresponding to the top of the fully conducting region. The fluid is assumed incompressible with constant density $\rho$ and constant temperature, that is, no convective motions are computed. The assumption of incompressibility is made for simplicity, although the pressure scale height at the depths of the conducting layer is roughly $8000$km [@Gui04], that is, about $1/5$ of the thickness of the layer. The effects of compressibility may well play a role in the dynamics of the conducting regions [see for instance @Evo04]. For simplicity we model the angular momentum coupling with the external zonal flow as a rigid boundary condition for the velocity at the outer boundary, rather than as a shear stress condition. The flow is driven through a boundary forcing rather than a volume forcing to avoid directly imposing bidimensionality to the velocity field. As we are interested in the bulk magnetohydrodynamical process, the exact nature of the coupling (electromagnetic or viscous, shear stress or rigid) with the upper molecular hydrogen layer is not crucial for our study. We discuss the implication of the choice of the rigid boundary condition in section \[sec:hydro\_jupiter\]. The radial profile of electrical conductivity is not well constrained in the gas giants. In particular the existence of a first order or continuous transition between the molecular and metallic hydrogen phase is still an open question, although high-pressure experiments are in favor of a continuous transition [@Nel99]. We choose to model the outer boundary as electrically insulating to simplify the coupling between the layers. The conductivity is assumed constant throughout the whole modeled conducting layer. As we do not model the molecular hydrogen layer, we assume zonal geostrophic balance within this envelope for simplicity. The amplitude of the zonal motions at the outer boundary of our model is therefore the same as the surface winds. This idealized representation of the dynamics of the molecular hydrogen layer would be altered if the magnetic forces upset the zonal geostrophic balance. Depending on the magnitude and radial profile of the electrical conductivity, the amplitude of the zonal motions might be reduced, and the zonal flow contours would tend

azimuthal velocity to represent the zonal flow generated in the overlying envelope. The carapace rotate around the $ z$-axis at the imposed rotation rate $ \Omega$. The aspect proportion is $ \gamma = r_i / r_o$ where $ r_i$ is the inner sphere radius, match to a rocky core, and $ r_o$ the out sphere radius, represent to the top of the fully conduct area. The fluid is assumed incompressible with constant concentration $ \rho$ and constant temperature, that is, no convective apparent motion are calculate. The assumption of incompressibility is made for simplicity, although the imperativeness scale height at the astuteness of the conducting layer is approximately $ 8000$km [ @Gui04 ], that is, about $ 1/5 $ of the thickness of the level. The effects of compressibility may well play a role in the dynamics of the conducting area [ see for instance @Evo04 ]. For simplicity we model the angular momentum coupling with the external zonal flow as a inflexible boundary condition for the velocity at the outer boundary, rather than as a shear stress condition. The flow is driven through a boundary storm rather than a volume force to invalidate directly imposing bidimensionality to the velocity discipline. As we are concerned in the bulk magnetohydrodynamical process, the exact nature of the coupling (electromagnetic or syrupy, shear stress or rigid) with the upper molecular hydrogen layer is not crucial for our study. We discuss the implication of the choice of the rigid boundary condition in section   \[sec: hydro\_jupiter\ ]. The radial visibility of electrical conduction is not well constrained in the gas giants. In particular the existence of a first holy order or continuous transition between the molecular and metallic hydrogen phase is still an open question, although high - pressure experiments are in favor of a continuous conversion [ @Nel99 ]. We choose to model the outer limit as electrically insulating to simplify the coupling between the layers. The conductivity is assumed changeless throughout the whole model conducting layer. As we do not model the molecular hydrogen level, we assume zonal geostrophic balance within this envelope for simplicity. The amplitude of the zonal motion at the outer limit of our model is therefore the same as the airfoil winds. This idealized representation of the dynamics of the molecular hydrogen layer would be change if the magnetic forces disturb the zonal geostrophic balance. Depending on the magnitude and radial profile of the electrical conductivity, the amplitude of the zonal motion might be reduced, and the zonal flow contours would tend

azlmuthal velocity to reprtsent the zonal flow genxrated jn the oxerlying envelope. The shell cotares aeound the $z$-axis at the imposed gotation eate $\Imega$. The eapect ratio ia $\gamke=r_i/r_o$ where $r_i$ ls the innes sphere radiuv, zoxresponding to a rocky core, and $r_o$ tre outet dphere radius, sorrtspjndihg to the top of the fully conductjng regpon. The fluid is sssumed incompressible witj cojstant density $\rho$ and constabt tqnperature, thxt is, no convective mojions are computed. The assumptiov of nncompressivioitj is made foc simpjicity, althombh the pressute scale heighb at vhe eepths of the conductmng layer is roughly $8000$km [@Gui04], tvac is, about $1/5$ of the thucjness of dhe uqyef. Tge erfects of compressigility may qell play a role in trv dynamics of the cjnqucting regions [see for instance @Evo04]. For sijplicity we model the abgular momentum coupllng with ehe external zonal flow as a rigid boundary condidion hof tkc xwllcity at the outer boundary, rather than as a fgesr stress conditlon. The flow is drovfn jhrough a bounaary forcjng rather than a golume sorcibg to avoyd dorectly imposing bidimensiobality to thv veoocity field. As we are intererted in tne bulk magnetohydrodyncmical process, thf exact nzgure of the coupuinn (enectromagnetic or viscous, fhear strxss ox rigid) dith the u[per molecklar mfdrogen layer is nlt crocial xor our stkdy. We discuss the implication of the choice pf thv rigid bjundavy condition in section \[sec:hydto\_jupiter\]. The fadial prorile of electrical conductivity ls not well ronstrainqd ib thw gas gkxnts. In particilar the vxnstence od a first order or covfinuous transitnub between the mplezulwr aid meedllic hydrogan pfasd is sgill an opek qjestoon, although high-prevsurs experiments are on favor od a contynuous transiyion [@Nel99]. We choose to mmdeu the oujer boundary as electrically inaulating ho fimplify the couilind between che layers. The conductivity is assumed cinstant throughouj the whole modeled cpkducting layxr. As re do not model the molecular hydrogen layer, wt assume zonal geostropgic banance within this envelope for simplicity. The amplitude of the zonal motions ar the outer boundady og our mjbel is eherxfure the same as nhe surface winds. This idealized represettction of the dynamics of the kouecular hydroeen layer would be altersd if thr magnetic forces upset the zonsl geostrophic balance. Dcpending in ths magnpjudx and radial profile of thz elecrrical xondmctivity, the aoplotmde of the zonel jotions might be reduced, and the zonal fpow contours wound tend

azimuthal velocity to represent the zonal flow the envelope. The rotates around the rate The aspect ratio $\gamma=r_i/r_o$ where $r_i$ the inner sphere radius, corresponding to rocky core, and $r_o$ the outer sphere radius, corresponding to the top of fully conducting region. The fluid is assumed incompressible with constant density $\rho$ and temperature, is, convective are computed. The assumption of incompressibility is made for simplicity, although the pressure scale height at depths of the conducting layer is roughly $8000$km that is, about $1/5$ the thickness of the layer. effects compressibility may play role the dynamics of conducting regions [see for instance @Evo04]. For simplicity we model the angular momentum coupling with the external flow as boundary condition the at outer boundary, rather a shear stress condition. The flow a boundary forcing rather than a volume forcing avoid directly bidimensionality to the velocity field. As are interested in the bulk magnetohydrodynamical process, the nature of the coupling (electromagnetic or viscous, shear stress or rigid) with the upper molecular is not crucial for study. We discuss implication the of rigid boundary in section \[sec:hydro\_jupiter\]. The radial profile of electrical conductivity is not constrained in the gas giants. In particular the existence of order continuous transition between molecular and metallic hydrogen is an open question, although are favor transition We to model the outer as electrically insulating to simplify coupling between the layers. throughout the whole modeled conducting layer. As we not model the molecular hydrogen layer, we zonal geostrophic balance within this envelope for simplicity. The amplitude of the motions at boundary of our model is therefore the same the surface winds. This representation of the dynamics of the molecular hydrogen layer be if the forces upset the geostrophic balance. Depending the magnitude and of the conductivity, amplitude motions might be reduced, and the flow would tend

azimuthal velocity to represEnt the zonaL flow GenEraTeD in tHe ovErlying envelopE. the sHell rotates around the $z$-aXis at ThE ImpoSEd RotatIon rate $\oMeGA$. the AsPeCt rAtIO iS $\gammA=r_i/R_o$ where $R_i$ is the innEr sPhEre radius, corREsPonding to a RocKy core, and $r_o$ tHe oUter spHeRe rADius, cOrrEsponDing to THe top oF the fully CoNDuctinG Region. THE FlUid iS assumed incompresSIbLE with constant dEnsity $\RhO$ AnD COnsTanT temperatuRe, That iS, No conveCTiVE MOtiONs are computed. the assumptiON of IncompReSsiBIlity iS made FoR SimPlicity, althOugh The pressuRe scalE Height aT The deptHs of thE coNduCtinG LaYeR is RoUGhlY $8000$Km [@gui04], THat Is, about $1/5$ oF tHe ThickNess OF THE layEr. THe efFects Of compressibiLitY may WEll Play a Role iN the DyNamicS of the ConduCtIng regions [see foR insTance @Evo04]. FOr sImPliCiTy we mODel the AngUlaR momentUm couplINg wItH THE eXternal zonal flow as A rIGId Boundary ConditIOn FoR The velocItY at The oUTEr bouNdarY, RaTher than As a sheAR sTrEss condItIon. The FlOw iS drIven tHRougH a bounDary forcIng raTHer than a volume FOrcing to avoid DIrECTlY ImpoSinG bidimensioNaliTY to tHe veLOcIty FIeld. AS we arE iNTeREsted in the bulk magneToHydrodYnamiCal process, the Exact naturE OF The couplIng (eLEcTRomagnetic or viScous, Shear stresS Or rigid) wIth thE upper moLecular hyDROgen layeR is Not CruCiaL FOr Our study. We disCUSs thE iMplicatIon Of the chOicE of The RigId Boundary cOndition In SeCtIoN \[seC:hydrO\_Jupiter\]. THe RadIaL prOfile OF electRical CondUcTiVIty Is not weLL cONStraInEd In thE gaS gIants. in paRTicUlar the Existence Of a FIrst OrDeR or contInuous transitIoN between thE mOleCular aND Metallic Hydrogen phase is still an oPEn questIon, AlthoUgh hIgh-pressuRe eXperimEntS Are in fAvor of A contInUouS TRansiTIOn [@nel99]. we Choose to moDEL thE outeR bOundAry as elEctrically insulatiNG to Simplify the coUplIng bETWeEn tHE lAYerS. THE coNDUctivity is assumEd constant ThROuGhout the whOLe mOdEled conDucting Layer. aS we do noT model the Molecular HyDrogEN LayEr, we assume Zonal geoStrophic bALance WItHin thIs eNvelopE fOr sImpliCity. ThE AmpLitudE of the ZoNal motIons aT tHe outer bOundary of our model is therEfore tHe samE as The surfacE wiNDs. THis idealiZed rEpresentatIon Of tHe dynAmiCS of thE molECuLar HYdrogEn laYEr would be ALtEreD IF tHe magnetic fORCEs uPset tHe zONal geoStroPhic balance. DependINg on the magnituDe anD RAdiAl pROfilE oF the electrical ConDuCTIvity, the AmPlitude of thE zonal moTiONs migHt be reDuced, aNd the zoNAL fLOw contOurs WouLd tend

azimuthal velocity to rep resent the zona l f low g ener ated in the overly i ng e nvelope. The shell rot atesar o undt he $z$- axis at th e imp os ed ro ta t io n rat e $ \Omega$ . The aspe ctra tio is $\gam m a= r_i/r_o$ w her e $r_i$ is t heinnersp her e radi us, corr espond i ng toa rocky c or e , and$ r_o$ th e ou tersphere radius, co r re s ponding to the top o ft he f ull y c onductingre gion. The flu i di s ass u med incompres sible withc ons tant d en sit y $\rho $ and c o nst ant tempera ture , that is , no c o nvectiv e motion s arecom put ed.T he a ssu mp t ion of in c omp ressibil it yis ma de f o r s impl ici ty,altho ugh the press ure sca l e h eight at t he d ep ths o f thecondu ct ing layer is ro ughl y $8000$k m [ @G ui0 4] , tha t is, a bou t $ 1/5$ of the th i ckn es s o fthe layer. The eff ec t s o f compre ssibil i ty m a y well p la y a rol e in th e dy n am ics of t he con d uc ti ng regi on s [see f orins tance @Evo 04]. For simp licit y we model thea ngular moment u mc o up l ingwit h the exter nalz onal flo w a s a rigid boun da r yc ondition for the ve lo city a t the outer bounda ry, rather t h an as ashea r s t ress condition . The flow is d r iven thr ougha bounda ry forcin g rather t han avol ume f or cing to avoid d irec tl y impos ing bidime nsi ona lit y t othe veloc ity fiel d. A swe ar e int e rested i nthe b ulk magn e tohydr odyna mica lpr o ces s, thee xa c t nat ur eof t heco uplin g (e l ect romagne tic or vi sco u s, s he ar stress or rigid) wi th the upper m ole cularh y drogen l ayer is not crucial for our stu dy. We d iscu ss the im pli cation of the ch oice o f the r igi d bound a r ycon di tion in se c t ion  \[se c: hydr o\_jupi ter\]. The radialp rof ile of electr ica l co n d uc tiv i ty isno t we l l constrained in the gas g ia n ts . In parti c ula rthe exi stenceof af irst or der or co ntinuoustr ansi t i onbetween th e molecu lar and m e talli c h ydrog enphaseis st ill a n open que stion , alth ou gh hig h-pre ss ure expe riments are in favor of a con tinuo ustransitio n [ @ Nel 99]. We c hoos e to model th e o uterbou n daryas e l ec tri c allyinsu l ating tos im pli f y t he coupling b e twe en th e l a yers.Theconductivity is a s sumed constant thr o u gho utt he w ho le modeled con duc ti n g layer.As we do notmodel th em olecu lar hy drogen layer, w ea ssumezona l g eostrophi c b al a nce wit hi nt his en velo pe for s implic i ty.T h e amplitude of t he zo n a l mot i ons at t he outerb ound ary of our model is t herefo re t he sa me as t he surfa cewi nds. Thisi dealizedrepre sentati on ofthe dynam icso f themole cu lar hydrogen l ay e rwo u ldbe a ltere dif t he magnet i c forces up s et thezo nal g eostro p hi c balance. D epe nding o n the magn i tude an d radi al pro file o f the e l ect ri cal con duc t i vity, the amplitud e of t he z onal mot io ns m ig htbe reduced, and t he zonal flow co n t ours w oul d ten d

azimuthal_velocity to_represent the zonal flow_generated in_the_overlying envelope._The_shell rotates around_the $z$-axis at_the imposed rotation rate_$\Omega$. The aspect_ratio_is $\gamma=r_i/r_o$ where $r_i$ is the inner sphere radius, corresponding to a rocky core,_and_$r_o$ the_outer_sphere_radius, corresponding to the top_of the fully conducting region._The fluid_is assumed incompressible with constant density $\rho$ and_constant_temperature, that is,_no convective motions are computed. The assumption of incompressibility_is made for simplicity, although the_pressure scale height_at_the_depths of the conducting_layer is roughly $8000$km [@Gui04], that_is, about $1/5$ of the thickness_of the layer. The effects of compressibility_may well play a role in_the dynamics of the conducting_regions [see_for instance @Evo04]. For simplicity we_model the angular_momentum coupling_with the external_zonal flow as a rigid boundary_condition for the_velocity at the outer boundary, rather_than_as a shear_stress_condition._The flow_is driven through_a_boundary forcing_rather_than a volume forcing to avoid_directly_imposing bidimensionality to the velocity field. As_we are interested in_the_bulk magnetohydrodynamical process, the_exact nature of the coupling_(electromagnetic or viscous, shear stress or_rigid) with_the upper_molecular hydrogen layer is not crucial for our study. We discuss_the implication of the choice of_the rigid boundary condition_in section \[sec:hydro\_jupiter\]._The_radial profile of_electrical_conductivity is_not well constrained in the gas giants._In particular_the existence of a first order_or continuous transition between_the_molecular and metallic hydrogen phase is_still an open question, although high-pressure_experiments are in favor of_a_continuous_transition [@Nel99]. We choose to_model the outer boundary as electrically_insulating to simplify_the coupling between the layers. The conductivity_is_assumed constant throughout the whole modeled_conducting_layer. As we do not model_the_molecular_hydrogen layer, we assume zonal_geostrophic balance within this envelope for_simplicity. The amplitude of the zonal motions at the_outer boundary of_our model is therefore the_same_as_the surface winds. This idealized representation of the dynamics of_the molecular_hydrogen layer would_be altered if the magnetic forces upset the zonal geostrophic_balance. Depending on the magnitude and radial_profile of the electrical conductivity, the amplitude of the zonal motions_might be reduced, and the zonal flow contours_would tend

. & Itoh, M. Structure and magnetic properties of Sr$_2$NiAO$_6$ (A = W, Te) [*Mater. Res. Bull.*]{} [**35**]{}, 449-457 (2000). Gao, H., Llobet, A., Barth, J., Winterlik, J., Felser, C., Panthöfer, M. & Tremel, W. Structure-property relations in the distorted ordered double perovskite Sr$_2$InReO$_6$. [*Phys. Rev. B*]{} [**83**]{}, 134406 (2011). Sikora, M., Kapusta, C., Borowiec, M., Oates, C. J., Prochazka, V., Rybicki, D., Zajac, D., Teresa, J. M. De., Marquina, C. & Ibarra, M. R. Evidence of unquenched Re orbital magnetic moment in AA$^{'}$FeReO$_6$ double perovskites. [**Appl. Phys. Lett.**]{} [**89**]{}, 062509 (2006). Michalik, J. M., Teresa, J. M. De., Ritter, C., Blasco, J., Serrate, D., Ibarra, M. R., Kapusta, C., Freudenberger, J. & Kozlova, N. High-field magnetization measurements in Sr$_2$CrReO$_6$ double perovskite: Evidence for orbital contribution to the magnetization. [*Eur. Phys. Lett.*]{} [**78**]{}, 17006 (2007). Jeon, B. C., Kim, C. H., Moon, S. J., Choi, W. S., Jeong, H., Lee, Y. S., Yu, J., Won, C. J., Jung, J. H., Hur, N. & Noh, T. W. Electronic structure of double perovskite A$_2$FeReO$_6$ (A = Ba and Ca): interplay between spin-orbit interaction, electron correlation, and lattice distortion. [*J. Phys.: Condens. Matter*]{} [**22**]{}, 345602 (2010). Acknowledgement =============== S.J. and P.A. thanks Council of Scientific and Industrial Research (CSIR), India for fellowship. P.A.K. thanks the Swedish Foundation for International Cooperation in

. & Itoh, M. Structure and magnetic properties of Sr$_2$NiAO$_6 $ (A = W, Te) [ * Mater. Res. Bull. * ] { } [ * * 35 * * ] { }, 449 - 457 (2000). Gao, H., Llobet, A., Barth, J., Winterlik, J., Felser, C., Panthöfer, M. & Tremel, W. Structure - property relations in the contort arrange double perovskite Sr$_2$InReO$_6$. [ * Phys. Rev. B * ] { } [ * * 83 * * ] { }, 134406 (2011). Sikora, M., Kapusta, C., Borowiec, M., Oates, C. J., Prochazka, V., Rybicki, D., Zajac, D., Teresa, J. M. De. , Marquina, C. & Ibarra, M. R. Evidence of unquenched Re orbital charismatic moment in AA$^{'}$FeReO$_6 $ bivalent perovskites. [ * * Appl. Phys. Lett. * * ] { } [ * * 89 * * ] { }, 062509 (2006). Michalik, J. M., Teresa, J. M. De. , Ritter, C., Blasco, J., Serrate, D., Ibarra, M. R., Kapusta, C., Freudenberger, J. & Kozlova, N. High - field magnetization measurement in Sr$_2$CrReO$_6 $ double perovskite: Evidence for orbital contribution to the magnetization. [ * Eur. Phys. Lett. * ] { } [ * * 78 * * ] { }, 17006 (2007). Jeon, B. C., Kim, C. H., Moon, S. J., Choi, W. S., Jeong, H., Lee, Y. S., Yu, J., Won, C. J., Jung, J. H., Hur, N. & Noh, T. W. Electronic structure of double perovskite A$ _ 2$FeReO$_6 $ (angstrom = Ba and Ca ): interplay between spin - orbit interaction, electron correlation coefficient, and lattice distortion. [ * J. Phys. : Condens. Matter * ] { } [ * * 22 * * ] { }, 345602 (2010). Acknowledgement = = = = = = = = = = = = = = = S.J. and P.A. thanks Council of Scientific and Industrial Research (CSIR), India for fellowship. P.A.K. thanks the Swedish Foundation for International Cooperation in

. & Ihoh, M. Structure and magnttic properties oy Sr$_2$NiAM$_6$ (A = W, Te) [*Matef. Res. Bull.*]{} [**35**]{}, 449-457 (2000). Gao, H., Llobet, A., Uarty, J., Wunterlik, J., Felser, C., Pavthöfer, M. & Tremel, Q. Stcucture-property cslations in tgc discocted ordered dooble perovskhte Sr$_2$InReO$_6$. [*Phfs. Rzv. B*]{} [**83**]{}, 134406 (2011). Sikora, M., Kapusta, C., Borowiec, M., Oates, V. U., Prochazka, V., Tybicly, D., Aajac, D., Teresa, J. M. De., Marquina, C. & Jbarra, K. R. Evidence pf unquenched Re orbital mwgnehic moment in AA$^{'}$FeGeO$_6$ double ksrodwkites. [**Appl. Ohys. Lett.**]{} [**89**]{}, 062509 (2006). Michalik, J. M., Teresa, J. M. De., Ritter, C., Blascu, J., Szrrate, D., Ibqrea, L. R., Kapusta, R., Freuqenberger, J. & Kozlova, N. High-gield magnetizstimn neasurements in Sr$_2$CrRxO$_6$ double perovskite: Evidence fkr orbital contrivurion jo tha maebetkzauioi. [*Ehr. Phyd. Lxtt.*]{} [**78**]{}, 17006 (2007). Jeon, G. C., Kim, C. H., Moon, S. J., Choi, W. S., Kejbg, H., Lee, Y. S., Gu, J., Wjn, C. J., Jung, J. H., Hur, N. & Noh, T. W. Electronic stducture of double perovwkite A$_2$FeReO$_6$ (A = Ba anf Ca): intewplay between spin-orbit interaction, electron corralatikv, aue latgucf distortion. [*J. Phys.: Condens. Matter*]{} [**22**]{}, 345602 (2010). Acknowleqfekekt =============== S.J. and P.A. thanhs Council pf Svyentific and Kndustxjam Research (CSIR), Infia for felliwship. P.A.H. thsnks the Swedish Foundation for Internanionql Cooperation in

. & Itoh, M. Structure and magnetic Sr$_2$NiAO$_6$ = W, [*Mater. Res. Bull.*]{} Llobet, Barth, J., Winterlik, Felser, C., Panthöfer, & Tremel, W. Structure-property relations in distorted ordered double perovskite Sr$_2$InReO$_6$. [*Phys. Rev. B*]{} [**83**]{}, 134406 (2011). Sikora, M., C., Borowiec, M., Oates, C. J., Prochazka, V., Rybicki, D., Zajac, D., Teresa, M. Marquina, & M. R. Evidence of unquenched Re orbital magnetic moment in AA$^{'}$FeReO$_6$ double perovskites. [**Appl. Phys. Lett.**]{} 062509 (2006). Michalik, J. M., Teresa, J. M. Ritter, C., Blasco, J., D., Ibarra, M. R., Kapusta, Freudenberger, & Kozlova, High-field measurements Sr$_2$CrReO$_6$ double perovskite: for orbital contribution to the magnetization. [*Eur. Phys. Lett.*]{} [**78**]{}, 17006 (2007). Jeon, B. C., Kim, C. Moon, S. W. S., H., Y. Yu, J., Won, Jung, J. H., Hur, N. & Electronic structure of double perovskite A$_2$FeReO$_6$ (A = and Ca): between spin-orbit interaction, electron correlation, and distortion. [*J. Phys.: Condens. Matter*]{} [**22**]{}, 345602 (2010). =============== S.J. and P.A. thanks Council of Scientific and Industrial Research (CSIR), India for fellowship. the Swedish Foundation for Cooperation in

. & Itoh, M. Structure and magnetic Properties Of Sr$_2$NIAO$_6$ (a = W, TE) [*MAter. res. BUll.*]{} [**35**]{}, 449-457 (2000). Gao, H., Llobet, A., bArth, j., Winterlik, J., Felser, C., PantHöfer, m. & TREmel, w. stRuctuRe-propeRTy RELatIoNs In tHe DIsTorteD orDered doUble perovsKitE SR$_2$InReO$_6$. [*Phys. ReV. b*]{} [**83**]{}, 134406 (2011). SIkora, M., KapuSta, c., Borowiec, M., OaTes, c. J., ProcHaZka, v., rybicKi, D., zajac, d., TeresA, j. M. De., MaRquina, C. & IbArRA, M. R. EviDEnce of uNQUeNcheD Re orbital magnetiC MoMEnt in AA$^{'}$FeReO$_6$ doUble peRoVSkITEs. [**APpl. phys. Lett.**]{} [**89**]{}, 062509 (2006). MiChAlik, J. m., teresa, J. m. de., rITTer, c., blasco, J., SerratE, D., Ibarra, M. R., KAPusTa, C., FreUdEnbERger, J. & KOzlovA, N. hIgh-Field magnetIzatIon measurEments IN Sr$_2$CrReo$_6$ Double pErovskIte: eviDencE FoR oRbiTaL ConTRiButIOn tO the magnEtIzAtion. [*eur. PHYS. lEtt.*]{} [**78**]{}, 17006 (2007). JEon, b. C., KiM, C. H., MoOn, S. J., Choi, W. S., JeoNg, H., lee, Y. s., yu, J., won, C. J., jung, J. h., Hur, n. & NOh, T. W. ELectroNic stRuCture of double peRovsKite A$_2$FeReo$_6$ (A = BA aNd CA): iNterpLAy betwEen SpiN-orbit iNteractIOn, eLeCTROn Correlation, and lattIcE DIsTortion. [*J. phys.: CoNDeNs. mAtter*]{} [**22**]{}, 345602 (2010). AckNoWleDgemENT =============== S.J. anD P.A. tHAnKs CounciL of SciENtIfIc and InDuStrial reSeaRch (cSIR), INDia fOr fellOwship. P.A.k. thanKS the Swedish FouNDation for InteRNaTIOnAL CooPerAtion in

. & Itoh, M. Structure and magneticprope rti esof Sr$ _2$N iAO$_6$ (A = W , Te) [*Mater. Res. Bull.*] {} [* *3 5 **]{ } ,449-4 57 (200 0 ). Gao ,H. , L lo b et , A., Ba rth, J. , Winterli k,J. , Felser, C. , P anthöfer,M.& Tremel, W. St ructur e- pro p ertyrel ation s in t h e dist orted ord er e d doub l e perov s k it e Sr $_2$InReO$_6$. [* P hy s . Rev. B*]{} [ **83** ]{ } ,1 3 440 6 ( 2011). Si ko ra, M . , Kapus t a, C . , B o rowiec, M., O ates, C. J. , Pr ochazk a, V. , Rybic ki, D ., Zaj ac, D., Ter esa, J. M. De ., Mar q uina, C . & Ibar ra, M. R. Ev iden c eof un qu e nch e dReo rbi tal magn et ic mome nt i n A A $^{' }$F eReO $_6$double perovs kit es.[ **A ppl.Phys. Let t. **]{} [**89 **]{} ,062509 (2006). Mic halik, J. M. ,Ter es a, J. M. De. , R itt er, C., Blasco , J. ,S e r ra te, D., Ibarra, M. R . , K apusta,C., Fr e ud en b erger, J .& K ozlo v a , N.High - fi eld magn etizat i on m easurem en ts inSr $_2 $Cr ReO$_ 6 $ do uble p erovskit e: Ev i dence for orbi t al contributi o nt o t h e ma gne tization. [ *Eur . Phy s. L e tt .*] { } [** 78**] {} , 1 7 006 (2007). Jeon,B. C., K im, C . H., Moon, S . J., Choi , W . S., Je ong, H. , Lee, Y. S., Y u, J. , Won, C.J ., Jung, J. H ., Hur,N. & Noh, T . W. Ele ctr oni c s tru c t ur e of double p e r ovsk it e A$_2$ FeR eO$_6$(A= B a a ndCa ): interp lay betw ee nsp in -or bit i n teractio n, el ec tro n cor r elatio n, an d la tt ic e di stortio n .[ * J. P hy s. : Co nde ns . Mat ter* ] {}[**22** ]{}, 3456 02( 2010 ). Acknowl edgement ==== == ========= S .J. and P . A . thanks Council of Scientifica nd Indu str ial R esea rch (CSIR ),Indiafor fellow ship.P.A.K .tha n k s the S we dis hFoundation f orInter na tion al Coop eration in

. &_Itoh, M._Structure and magnetic properties_of Sr$_2$NiAO$_6$_(A_= W,_Te)_[*Mater. Res. Bull.*]{}_[**35**]{}, 449-457 (2000). Gao,_H., Llobet, A., Barth,_J., Winterlik, J.,_Felser,_C., Panthöfer, M. & Tremel, W. Structure-property relations in the distorted ordered double perovskite_Sr$_2$InReO$_6$._[*Phys. Rev._B*]{}_[**83**]{},_134406 (2011). Sikora, M., Kapusta, C.,_Borowiec, M., Oates, C. J.,_Prochazka, V.,_Rybicki, D., Zajac, D., Teresa, J. M. De.,_Marquina,_C. & Ibarra,_M. R. Evidence of unquenched Re orbital magnetic moment_in AA$^{'}$FeReO$_6$ double perovskites. [**Appl. Phys._Lett.**]{} [**89**]{}, 062509_(2006). Michalik,_J._M., Teresa, J. M._De., Ritter, C., Blasco, J., Serrate,_D., Ibarra, M. R., Kapusta, C.,_Freudenberger, J. & Kozlova, N. High-field magnetization_measurements in Sr$_2$CrReO$_6$ double perovskite: Evidence_for orbital contribution to the_magnetization. [*Eur._Phys. Lett.*]{} [**78**]{}, 17006 (2007). Jeon,_B. C., Kim,_C. H.,_Moon, S. J.,_Choi, W. S., Jeong, H., Lee,_Y. S., Yu,_J., Won, C. J., Jung, J._H.,_Hur, N. &_Noh,_T._W. Electronic_structure of double_perovskite_A$_2$FeReO$_6$ (A_=_Ba and Ca): interplay between spin-orbit_interaction,_electron correlation, and lattice distortion. [*J. Phys.:_Condens. Matter*]{} [**22**]{}, 345602_(2010). Acknowledgement =============== S.J._and P.A. thanks Council_of Scientific and Industrial Research_(CSIR), India for fellowship. P.A.K. thanks_the Swedish_Foundation for_International Cooperation in

_{i=1}^{I}L\left\Vert v\left( i\right) \right\Vert M^{d-1} \\ &<&cM^{d}\end{aligned}$$ For each $x\in B_{i}$, $x\in \alpha ^{w\left( x\right) }C_{k\left( x\right) } $ for some $1\leq l\left( x\right) \leq K$ and $w\left( x\right) \in P_{k\left( x\right) }$ and $\alpha ^{v\left( i\right) }\left( x\right) $ is in $\alpha ^{u\left( x\right) }C_{l\left( x\right) }$ for some $1\leq l\left( x\right) \leq K$ for some $u\left( x\right) \in P_{l\left( x\right) } $. The map $x\mapsto \left( k\left( x\right),w\left( x\right),l\left( x\right),u\left( x\right) \right) $ is continuous. For each pair of indices $k$ and $w$ with $\alpha ^{w}C_{k}\subset B_{i}$, we select a vector $r\left( i,w,k\right) \in P_{k}$ so that $\alpha ^{r\left( i,w,k\right) }C_{k}\subset Z\setminus \theta _{2}\left( Y\right) $. We do so in such a way that if $\left( i,w\right) \neq \left( i^{\prime },w^{\prime }\right) $ then $r\left( i,w,k\right) \neq r\left( i^{\prime },w^{\prime },k\right) $. This is possible because $$\begin{aligned} \#\left\{ \left( i,k,w\right) :\alpha ^{w}C_{k}\subset B_{i}\text{ for some }% i\right\} &=&\sum_{i=1}^{I}\left\vert P_{k}\vartriangle \left( P_{k}-v\left(

_ { i=1}^{I}L\left\Vert v\left (i\right) \right\Vert M^{d-1 } \\ & < & cM^{d}\end{aligned}$$ For each $ x\in B_{i}$, $ x\in \alpha ^{w\left (x\right) } C_{k\left (x\right) } $ for some $ 1\leq l\left (x\right) \leq K$ and $ w\left (x\right) \in P_{k\left (x\right) } $ and $ \alpha ^{v\left (i\right) } \left (x\right) $ is in $ \alpha ^{u\left (x\right) } C_{l\left (x\right) } $ for some $ 1\leq l\left (x\right) \leq K$ for some $ u\left (x\right) \in P_{l\left (x\right) } $. The map $ x\mapsto \left (k\left (x\right),w\left (x\right),l\left ( x\right),u\left (x\right) \right) $ is continuous. For each pair of indices $ k$ and $ w$ with $ \alpha ^{w}C_{k}\subset B_{i}$, we select a vector $ r\left (i, w, k\right) \in P_{k}$ so that $ \alpha ^{r\left (iodine, w, k\right) } C_{k}\subset Z\setminus \theta _ { 2}\left (Y\right) $. We cause so in such a way that if $ \left (i, w\right) \neq \left (i^{\prime }, w^{\prime } \right) $ then $ r\left (i, w, k\right) \neq r\left (i^{\prime }, w^{\prime }, k\right) $. This is potential because $ $ \begin{aligned } \#\left\ { \left (i, k, w\right): \alpha ^{w}C_{k}\subset B_{i}\text { for some }% i\right\ } & = & \sum_{i=1}^{I}\left\vert P_{k}\vartriangle \left (P_{k}-v\left (

_{i=1}^{I}L\peft\Vert v\left( i\right) \rigmt\Vert M^{d-1} \\ &<&cM^{d}\end{coigned}$$ Hor eacg $x\in B_{i}$, $x\in \alpha ^{w\left( x\right) }C_{k\leht( x\eight) } $ for some $1\leq l\left( x\fight) \leq K$ and $w\oeft( z\right) \in P_{k\left( x\rlyht) }$ zkd $\al'he ^{v\left( i\right) }\keft( x\righd) $ is in $\alpha ^{u\ueyt( x\right) }C_{l\left( x\right) }$ for some $1\lez l\left( c\rlght) \leq K$ for somt $u\jeft( q\rlght) \in P_{l\left( x\right) } $. The map $x\japsto \neft( k\left( x\roght),w\left( x\right),l\left( x\righh),u\levt( x\right) \right) $ id continuouw. For wach pair of indices $k$ and $w$ with $\zlpha ^{w}C_{k}\subset B_{i}$, we select a xectox $r\left( i,w,k\tntht) \hn P_{k}$ so thet $\alpra ^{r\left( i,w,k\rlbht) }C_{k}\vubset A\setminus \thets _{2}\lxft( T\right) $. We do so in snch a way that if $\lest( i,w\righd) \ueq \left( i^{\prime },w^{\prime }\rught) $ thet $r\lddt( k,w,k\dijht) \neq r\pefv( i^{\prime },w^{\prjme },k\right) $. This is possible btcatwe $$\begin{alignsd} \#\left\{ \lqft( i,k,w\right) :\alpha ^{w}C_{k}\subset B_{i}\text{ for soje }% i\right\} &=&\sum_{i=1}^{I}\left\vert P_{k}\vartriangle \left( P_{n}-v\left(

_{i=1}^{I}L\left\Vert v\left( i\right) \right\Vert M^{d-1} \\ &<&cM^{d}\end{aligned}$$ $x\in $x\in \alpha x\right) }C_{k\left( x\right) l\left( \leq K$ and x\right) \in P_{k\left( }$ and $\alpha ^{v\left( i\right) }\left( $ is in $\alpha ^{u\left( x\right) }C_{l\left( x\right) }$ for some $1\leq l\left( \leq K$ for some $u\left( x\right) \in P_{l\left( x\right) } $. The map \left( x\right),w\left( x\right),u\left( \right) $ is continuous. For each pair of indices $k$ and $w$ with $\alpha ^{w}C_{k}\subset B_{i}$, select a vector $r\left( i,w,k\right) \in P_{k}$ so $\alpha ^{r\left( i,w,k\right) }C_{k}\subset \theta _{2}\left( Y\right) $. We so such a that $\left( \neq \left( i^{\prime }\right) $ then $r\left( i,w,k\right) \neq r\left( i^{\prime },w^{\prime },k\right) $. This is possible because $$\begin{aligned} \#\left\{ i,k,w\right) :\alpha for some i\right\} P_{k}\vartriangle P_{k}-v\left(

_{i=1}^{I}L\left\Vert v\left( i\right) \rigHt\Vert M^{d-1} \\ &<&cM^{D}\end{aLigNed}$$ foR eacH $x\in b_{i}$, $x\in \alpha ^{w\lefT( X\rigHt) }C_{k\left( x\right) } $ for some $1\lEq l\leFt( X\RighT) \LeQ K$ and $W\left( x\rIGhT) \IN P_{k\LeFt( X\riGhT) }$ AnD $\alphA ^{v\lEft( i\rigHt) }\left( x\rigHt) $ iS iN $\alpha ^{u\left( x\RIgHt) }C_{l\left( x\rIghT) }$ for some $1\leq l\LefT( x\righT) \lEq K$ FOr somE $u\lEft( x\rIght) \in p_{L\left( x\Right) } $. The mAp $X\Mapsto \LEft( k\lefT( X\RiGht),w\Left( x\right),l\left( x\rIGhT),U\left( x\right) \rigHt) $ is coNtINuOUS. FoR eaCh pair of inDiCes $k$ aND $w$ with $\aLPhA ^{W}c_{K}\suBSet B_{i}$, we select A vector $r\lefT( I,w,k\Right) \iN P_{K}$ so THat $\alpHa ^{r\leFt( I,W,k\rIght) }C_{k}\subseT Z\seTminus \theTa _{2}\left( y\Right) $. We DO so in suCh a way ThaT if $\Left( I,W\rIgHt) \nEq \LEft( I^{\PrIme },W^{\PriMe }\right) $ tHeN $r\Left( i,W,k\riGHT) \NEq r\lEft( I^{\priMe },w^{\prIme },k\right) $. This Is pOssiBLe bEcausE $$\begiN{aliGnEd} \#\lefT\{ \left( i,K,w\rigHt) :\Alpha ^{w}C_{k}\subset B_{I}\texT{ for some }% i\RigHt\} &=&\Sum_{I=1}^{I}\Left\vERt P_{k}\vaRtrIanGle \left( p_{k}-v\left(

_{i=1}^{I}L\left\Vert v\le ft( i\righ t) \r igh t\V er t M^ {d-1 } \\ &<&cM^{d} \ end{ aligned}$$ For each $ x\inB_ { i}$, $x \in \ alpha ^ { w\ l e ft( x \r igh t) }C _{k\l eft ( x\rig ht) } $ fo r s om e $1\leq l\l e ft ( x\right) \l eq K$ and $w \le ft( x\ ri ght ) \inP_{ k\lef t( x\r i ght) } $ and $\a lp h a ^{v\ l eft( i\ r i gh t) } \left( x\right) $ is in $\alpha ^{u \left( x \ ri g h t)}C_ {l\left( x \r ight) }$ fors om e $ 1\l e q l\left( x\r ight) \leqK $ f or som e$u\ l eft( x \righ t) \in P_{l\left( x\r ight) } $ . Them ap $x\m a psto \l eft( k \le ft( x\r i gh t) ,w\ le f t(x \r igh t ),l \left( x \r ig ht),u \lef t ( x \rig ht) \ri ght)$ is continuo us. Fo r ea ch pa ir of ind ic es $k $ and$w$ w it h $\alpha ^{w}C _{k} \subset B _{i }$ , w eselec t a vec tor $r \left(i,w,k\r i ght )\ i n P _{k}$ so that $\al ph a ^{ r\left(i,w,k\ r ig ht ) }C_{k}\ su bse t Z\ s e tminu s \t h et a _{2}\l eft( Y \ ri gh t) $. W edo soin su cha way that if $\ left( i, w\rig h t) \neq \left( i^{\prime },w ^ {\ p r im e }\r igh t) $ then $ r\le f t( i ,w,k \ ri ght ) \neq r\le ft ( i ^ {\prime },w^{\prime } ,k\rig ht) $ . This is pos sible beca u s e $$\begi n{al i gn e d} \#\left\{ \ left( i,k,w\rig h t) :\alp ha ^{ w}C_{k}\ subset B_ { i }\text{for so me}%i \ ri ght\} &=&\sum _ { i=1} ^{ I}\left \ve rt P_{k }\v art ria ngl e\left( P_ {k}-v\le ft (

_{i=1}^{I}L\left\Vert v\left(_i\right) \right\Vert M^{d-1}_\\ &<&cM^{d}\end{aligned}$$ For each $x\in B_{i}$,_$x\in \alpha_^{w\left(_x\right) }C_{k\left(_x\right) }_$ for some_$1\leq l\left( x\right)_\leq K$ and $w\left(_x\right) \in P_{k\left( x\right)_}$_and $\alpha ^{v\left( i\right) }\left( x\right) $ is in $\alpha ^{u\left( x\right) }C_{l\left( x\right)_}$_for some_$1\leq l\left(_x\right)_\leq K$ for some $u\left(_x\right) \in P_{l\left( x\right) } $._The map_$x\mapsto \left( k\left( x\right),w\left( x\right),l\left( x\right),u\left( x\right) \right) $_is_continuous. For each pair_of indices $k$ and $w$ with $\alpha ^{w}C_{k}\subset B_{i}$,_we select a vector $r\left( i,w,k\right)_\in P_{k}$ so_that_$\alpha ^{r\left(_i,w,k\right) }C_{k}\subset Z\setminus \theta__{2}\left( Y\right) $. We do so_in such a way that if_$\left( i,w\right) \neq \left( i^{\prime },w^{\prime }\right) $_then $r\left( i,w,k\right) \neq r\left( i^{\prime },w^{\prime_},k\right) $. This is possible_because $$\begin{aligned} \#\left\{_\left( i,k,w\right) :\alpha ^{w}C_{k}\subset B_{i}\text{_for some }% i\right\}_&=&\sum_{i=1}^{I}\left\vert P_{k}\vartriangle_\left( P_{k}-v\left(

the restriction that the shock thickness cannot exceed the dimensions of the region. Either of the above variations allows final Lorentz factors in the lab frame well above $10^9$, and the proton could reconvert to a neutron upon exiting the region. There is thus some chance that neutrons so produced could arrive intact at Earth from a distance of 10 Kpc. Given a number density of current carrying protons of $10^{18}B_{15}/R_6$ over typical dimensions of $10^6$cm, the number of available protons could be of order $10^{35}$ or more. (Note that if the burst lasts more than a dynamical timescale, the current carriers turn over, and the time integrated total can be more than at a given instant.) If a significant fraction of them are neutronized at Lorentz factors $\Gamma$ above $10^{9}$, then they propagate 10 Kpc before decaying. At a distance of 10 Kpc, a burst of $10^{44}$ erg that puts $10^{-4}$ of its energy into $10^{34}$ neutrons at $\Gamma = 10^9$ creates a fluence of $10^{-2}/km^2$. A collecting area exceeding 100 $km^2$ might have had some chance of detecting UHE neutrons from the Aug. 27, 1998 flare. Similarly, a $10^3km^2$ array might have had the chance from the April 18, 2001 flare, which emitted about $10^{43}$ ergs. The charged pions emitted typically cool until they marginally satisfy equation (\[10\]) suitably adapted to pions, i.e. until $\gamma'=3 \Gamma^3c/r_{\pi,o}\sigma\omega_{\pi,g}\sim1.3\times 10^7\Gamma^3/\sigma$, and they can produce extremely energetic neutrinos, of order $E_{\nu}=\gamma' m_{\pi}c^2/4 \sim (5\times 10^{14}\Gamma^3/\sigma)$eV. The neutrino fluence at Earth that would have resulted from the Aug. 27 flare would have been $10^{-2}$ erg/cm$^2$ if all the observed flare energy had gone into neutrinos. Several percent of this in the range 1 TeV $\le E_{\nu}\le 10^3$ TeV is detectable with 1 $km^3$ detectors. A photon fluence of $10^{-2}$ erg/

the restriction that the shock thickness cannot exceed the dimensions of the area. Either of the above variation allows final Lorentz divisor in the lab inning well above $ 10 ^ 9 $, and the proton could reconvert to a neutron upon exiting the region. There be therefore some chance that neutrons so produced could arrive intact at Earth from a distance of 10 Kpc. give a number density of current carrying proton of $ 10^{18}B_{15}/R_6 $ over typical dimensions of $ 10 ^ 6$cm, the number of available protons could be of order $ 10^{35}$ or more. (Note that if the burst last more than a dynamical timescale, the current carriers turn over, and the prison term integrated total can be more than at a given instant .) If a meaning fraction of them are neutronized at Lorentz factors $ \Gamma$ above $ 10^{9}$, then they circulate 10 Kpc before decaying. At a distance of 10 Kpc, a burst of $ 10^{44}$ erg that puts $ 10^{-4}$ of its department of energy into $ 10^{34}$ neutrons at $ \Gamma = 10 ^ 9 $ creates a fluence of $ 10^{-2}/km^2$. A collecting area exceeding 100 $ km^2 $ might have had some probability of detecting UHE neutrons from the Aug. 27, 1998 flare. Similarly, a $ 10 ^ 3km^2 $ array might have had the chance from the April 18, 2001 flare, which emitted about $ 10^{43}$ ergs. The charged pions emit typically cool until they marginally satisfy equation (\[10\ ]) appropriately adapt to pion, i.e. until $ \gamma'=3 \Gamma^3c / r_{\pi, o}\sigma\omega_{\pi, g}\sim1.3\times 10 ^ 7\Gamma^3/\sigma$, and they can produce extremely energetic neutrinos, of order $ E_{\nu}=\gamma' m_{\pi}c^2/4 \sim (5\times 10^{14}\Gamma^3/\sigma)$eV. The neutrino fluence at Earth that would have resulted from the Aug. 27 flare would have been $ 10^{-2}$ erg / cm$^2 $ if all the observed flare energy had go into neutrinos. Several percent of this in the range 1 TeV $ \le E_{\nu}\le 10 ^ 3 $ TeV is detectable with 1 $ km^3 $ detectors. A photon fluence of $ 10^{-2}$ erg/

thf restriction that the smock thickness ccbnot eeceed tge dimenrions of the region. Either of tye abive variations allows winal Lorvntz factirs mn the lab frame well above $10^9$, ahf thz 'roton could regonvert to d neutron upon ebicing the region. There is thus some crance tnah neutrons so krodubeq cohld arrive intact at Earth from a sistanct of 10 Kpc. Given a nimber density of current cwrrylng protons of $10^{18}B_{15}/R_6$ lver typicao diiwnsions of $10^6$co, the numbtr of availagle protons could be of order $10^{35}$ ur moxe. (Note thaj jf jhe burst lawts mjre than a dnmamican timesvale, the currekt cacriees turn over, and the vime integrated totaj can be koxe than at a given inwtqnt.) Ix a vignkdicxnt fcacfion ov tiem are neufronized at Lorentz factors $\Gakmw$ above $10^{9}$, then fhey pwo[agate 10 Kpc before decaying. At a distanbe or 10 Kpc, a burst of $10^{44}$ erg rhat puts $10^{-4}$ of its enetgy into $10^{34}$ geutrons at $\Gamma = 10^9$ creates a fluence of $10^{-2}/km^2$. A colnectiig arta excedeijg 100 $km^2$ might have had some chance of detectind UNE neutrons from the Aug. 27, 1998 glwrr. Similarly, a $10^3ym^2$ arrcg jight have had the chance from the Aprij 18, 2001 glare, which emitted about $10^{43}$ wrgs. The charyed pions emitted typncally cool bntil jhey msrginally satisfy equatnon (\[10\]) shitably adaoted to pjuns, i.e. until $\gamoa'=3 \Namka^3c/r_{\pi,o}\sigma\omega_{\pi,g}\sim1.3\timqs 10^7\Gamma^3/\sijma$, aud they zan kroduce extremely enernatic neutrinos, of lrder $E_{\tu}=\gamma' m_{\pl}c^2/4 \sim (5\times 10^{14}\Gamma^3/\sigma)$eV. The nenvrino fluence ad Edrth thac woulc have resultqd from the Auy. 27 flare woula have beeh $10^{-2}$ erg/ck$^2$ if all thq observed fldte energy had gone ineo nwutrunos. Sexdral percent og this in the range 1 TeV $\le E_{\nu}\le 10^3$ TeY is astectable with 1 $km^3$ detectors. A phoyon fltejcx of $10^{-2}$ arg/

the restriction that the shock thickness cannot dimensions the region. of the above in lab frame well $10^9$, and the could reconvert to a neutron upon the region. There is thus some chance that neutrons so produced could arrive at Earth from a distance of 10 Kpc. Given a number density of carrying of over dimensions of $10^6$cm, the number of available protons could be of order $10^{35}$ or more. (Note if the burst lasts more than a dynamical the current carriers turn and the time integrated total be than at given If significant fraction of are neutronized at Lorentz factors $\Gamma$ above $10^{9}$, then they propagate 10 Kpc before decaying. At a of 10 burst of erg puts of its energy neutrons at $\Gamma = 10^9$ creates $10^{-2}/km^2$. A collecting area exceeding 100 $km^2$ might had some of detecting UHE neutrons from the 27, 1998 flare. Similarly, a $10^3km^2$ array might had the chance from the April 18, 2001 flare, which emitted about $10^{43}$ ergs. The emitted typically cool until marginally satisfy equation suitably to i.e. $\gamma'=3 \Gamma^3c/r_{\pi,o}\sigma\omega_{\pi,g}\sim1.3\times and they can produce extremely energetic neutrinos, of order $E_{\nu}=\gamma' m_{\pi}c^2/4 (5\times 10^{14}\Gamma^3/\sigma)$eV. The neutrino fluence at Earth that would have the 27 flare would been $10^{-2}$ erg/cm$^2$ if the flare energy had gone Several of range TeV E_{\nu}\le 10^3$ TeV is with 1 $km^3$ detectors. A fluence of $10^{-2}$ erg/

the restriction that the shocK thickness CannoT exCeeD tHe diMensIons of the regioN. eithEr of the above variations AllowS fINal LOReNtz faCtors in THe LAB frAmE wEll AbOVe $10^9$, And thE prOton couLd reconverT to A nEutron upon exITiNg the regioN. ThEre is thus somE chAnce thAt NeuTRons sO prOduceD could ARrive iNtact at EaRtH From a dIStance oF 10 kPc. giveN a number density of CUrREnt carrying proTons of $10^{18}b_{15}/R_6$ OVeR TYpiCal Dimensions Of $10^6$Cm, the NUmber of AVaILABle PRotons could be Of order $10^{35}$ or moRE. (NoTe that If The BUrst laSts moRe THan A dynamical tImesCale, the cuRrent cARriers tURn over, aNd the tIme IntEgraTEd ToTal CaN Be mORe ThaN At a Given insTaNt.) if a siGnifICANT fraCtiOn of Them aRe neutronized At LOrenTZ faCtors $\gamma$ AbovE $10^{9}$, tHen thEy propAgate 10 kpC before decaying. at a dIstance of 10 kpc, A bUrsT oF $10^{44}$ erg tHAt puts $10^{-4}$ Of iTs eNergy inTo $10^{34}$ neutrONs aT $\GAMMA = 10^9$ cReates a fluence of $10^{-2}/km^2$. a cOLLeCting areA exceeDInG 100 $kM^2$ Might havE hAd sOme cHANce of DeteCTiNg UHE neuTrons fROm ThE Aug. 27, 1998 flaRe. similaRlY, a $10^3kM^2$ arRay miGHt haVe had tHe chance From tHE April 18, 2001 flare, whiCH emitted about $10^{43}$ ERgS. tHe CHargEd pIons emitted TypiCAlly Cool UNtIl tHEy marGinalLy SAtISfy equation (\[10\]) suitably AdApted tO pionS, i.e. until $\gamma'=3 \gamma^3c/r_{\pi,o}\SIGMa\omega_{\pI,g}\siM1.3\TiMEs 10^7\Gamma^3/\sigma$, anD they Can produce EXtremely EnergEtic neutRinos, of orDER $E_{\nu}=\gammA' m_{\pI}c^2/4 \sIm (5\tImeS 10^{14}\gAmMa^3/\sigma)$eV. The nEUTrinO fLuence aT EaRth that WouLd hAve ResUlTed from thE Aug. 27 flarE wOuLd HaVe bEen $10^{-2}$ erG/Cm$^2$ if all tHe ObsErVed Flare ENergy hAd gonE intO nEuTRinOs. SeverAL pERCent Of ThIs in The RaNge 1 Tev $\le E_{\NU}\le 10^3$ teV is deTectable wIth 1 $KM^3$ detEcToRs. A photOn fluence of $10^{-2}$ erG/

the restriction that theshock thic kness ca nno texce ed t he dimensionso f th e region. Either of t he ab ov e var i at ionsallowsf in a l Lo re nt z f ac t or s inthe lab fr ame well a bov e$10^9$, andt he proton co uld reconvert t o a neutr on up o n exi tin g the regio n . Ther e is thus s o me cha n ce that n eu tron s so produced cou l da rrive intact a t Eart hf ro m a d ist ance of 10 K pc. G iven an um b e r de n sity of curre nt carrying pro tons o f$10 ^ {18}B_ {15}/ R_ 6 $ o ver typical dim ensions o f $10^ 6 $cm, th e number of av ail abl e pr o to ns co ul d be of or d er$10^{35} $or more . (N o t e that if the burs t lasts moretha n ad yna mical time scal e, thecurren t car ri ers turn over,andthe timeint eg rat ed tota l can b e m ore than a t a giv e n i ns t a n t. ) If a significant f r a ct ion of t hem ar e n eu t ronizedat Lo rent z facto rs $ \ Ga mma$ abo ve $10 ^ {9 }$ , thenth ey pro pa gat e 1 0 Kpc befo re dec aying. A t a d i stance of 10 K p c, a burst of $1 0 ^ {4 4 }$ e rgthat puts $ 10^{ - 4}$of i t sene r gy in to $1 0^ { 34 } $ neutrons at $\Gam ma = 10^ 9$ cr eates a fluen ce of $10^ { - 2 }/km^2$. A c o ll e cting area exc eedin g 100 $km^ 2 $ mighthavehad some chance o f detectin g U HEneu tro n s f rom the Aug.2 7 , 19 98 flare. Si milarly , a $1 0^3 km^ 2$ array mi ght have h ad t he ch ancef rom theAp ril 1 8,2001f lare,which emi tt ed abo ut $10^ { 43 } $ erg s. Thecha rg ed pi onse mit ted typ ically co olu ntil t he y margi nally satisfy e quation (\ [1 0\] ) suit a b ly adapt ed to pions, i.e. until $\gamma '=3 \Gam ma^3 c/r_{\pi, o}\ sigma\ ome g a_{\pi ,g}\si m1.3\ ti mes 1 0^7\G a m ma ^3/ \s igma$, and t hey canpr oduc e extre mely energetic neu t rin os, of order$E_ {\nu } = \g amm a 'm _{\ pi } c^2 / 4 \sim (5\times10^{14}\Ga mm a ^3 /\sigma)$e V . T he neutri no flue nce a t Earththat woul d have re su lted f rom the Aug.27 flare would ha v e bee n $ 10^{- 2}$ erg/c m$ ^2$ if a ll the obs erved flare e nergyhad g on e into n eutrinos. Several perce nt ofthisinthe range 1T eV$\le E_{\ nu}\ le 10^3$ T eVisdetec tab l e wit h 1$ km ^3$ detec tors . A photo n f lue n c eof $10^{-2} $ e rg/

the_restriction that_the shock thickness cannot_exceed the_dimensions_of the_region. Either_of the above_variations allows final_Lorentz factors in the_lab frame well_above_$10^9$, and the proton could reconvert to a neutron upon exiting the region. There_is_thus some_chance_that_neutrons so produced could arrive_intact at Earth from a_distance of_10 Kpc. Given a number density of current carrying_protons_of $10^{18}B_{15}/R_6$ over_typical dimensions of $10^6$cm, the number of available protons_could be of order $10^{35}$ or_more. (Note that_if_the_burst lasts more than_a dynamical timescale, the current carriers_turn over, and the time integrated_total can be more than at a_given instant.) If a significant fraction_of them are neutronized at_Lorentz factors_$\Gamma$ above $10^{9}$, then they_propagate 10 Kpc_before decaying._At a distance_of 10 Kpc, a burst of_$10^{44}$ erg that_puts $10^{-4}$ of its energy into_$10^{34}$_neutrons at $\Gamma_=_10^9$_creates a_fluence of $10^{-2}/km^2$._A_collecting area_exceeding_100 $km^2$ might have had some_chance_of detecting UHE neutrons from the Aug._27, 1998 flare. Similarly,_a_$10^3km^2$ array might have_had the chance from the_April 18, 2001 flare, which emitted_about $10^{43}$_ergs. The charged_pions emitted typically cool until they marginally satisfy equation (\[10\]) suitably_adapted to pions, i.e. until $\gamma'=3_\Gamma^3c/r_{\pi,o}\sigma\omega_{\pi,g}\sim1.3\times 10^7\Gamma^3/\sigma$, and they can_produce extremely_energetic_neutrinos, of order_$E_{\nu}=\gamma'_m_{\pi}c^2/4 \sim_(5\times 10^{14}\Gamma^3/\sigma)$eV. The neutrino fluence at Earth that_would have_resulted from the Aug. 27 flare_would have been $10^{-2}$_erg/cm$^2$_if all the observed flare energy_had gone into neutrinos. Several percent_of this in the range_1_TeV_$\le E_{\nu}\le 10^3$ TeV is_detectable with 1 $km^3$ detectors. A photon_fluence of $10^{-2}$_erg/

representation is not unique. Normalization as well as other expectation values of product operators can be obtained from $$\begin{aligned} \nonumber \langle\psi|\bigotimes_{k=1}^N S_k|\psi\rangle &=&\operatorname{tr}\left[\prod_{k=1}^N E_{S_k}^{[k]}\right]\;,\quad\text{with}\\ E_S^{[k]}&\equiv&\sum_{i,j=1}^d \langle i|S|j\rangle \overline{A}_i^{[k]}\otimes A_j^{[k]}\label{E}\;.\end{aligned}$$ \[htbp\] Finitely correlated states -------------------------- The present work is inspired by the papers on *finitely correlated states* (FCS) which in turn generalize the findings of Affleck, Kennedy, Lieb and Tasaki (AKLT) [@AKLT]. In fact, many of the results we derive are extensions of the FCS formalism to finite and/or non-translational invariant systems. For this reason we will briefly review the work on FCS. A FCS is a translational invariant state on an infinite spin chain which is constructed from a completely positive and trace preserving map $\mathbb{E}:{\cal B}({\cal H}_A)\rightarrow {\cal B}({\cal H}_A\otimes {\cal H}_B)$ and a corresponding fixed point density operator $\Lambda=\operatorname{tr}_B[\mathbb{E}(\Lambda)]$. Here ${\cal H}_B=\mathbb{C}^d$ is the Hilbert space corresponding to one site in the chain and ${\cal H}_A=\mathbb{C}^D$ is an ancillary system. An $n$-partite reduced density matrix $\rho_n$ of the FCS is then obtained by repeated application of $\mathbb{E}$ to the ancillary system (initially in $\Lambda$) followed by tracing out the ancilla, i.e., $$\rho_n=\operatorname{tr}_A\big[\mathbb{E}^n(\Lambda)\big]\;.$$ An important instance are *purely generated* FCS where $\mathbb{E}(x)=V^\dag x V$ is given by a partial isometry $V$. The latter can be easily related to the $A$’s in the matrix product representation via $V=\sum_{i=1}^d\sum_{\alpha,\beta=

representation is not unique. Normalization as well as early anticipation values of product operator can be receive from $ $ \begin{aligned } \nonumber \langle\psi|\bigotimes_{k=1}^N S_k|\psi\rangle & = & \operatorname{tr}\left[\prod_{k=1}^N E_{S_k}^{[k]}\right]\;,\quad\text{with}\\ E_S^{[k]}&\equiv&\sum_{i, j=1}^d \langle i|S|j\rangle \overline{A}_i^{[k]}\otimes A_j^{[k]}\label{E}\;.\end{aligned}$$ \[htbp\ ] Finitely correlated states -------------------------- The present oeuvre is cheer by the papers on * finitely correlate states * (FCS) which in turn popularize the findings of Affleck, Kennedy, Lieb and Tasaki (AKLT) [ @AKLT ]. In fact, many of the results we derive are extensions of the FCS formalism to finite and/or non - translational changeless systems. For this reason we will concisely review the work on FCS. A FCS is a translational changeless state of matter on an infinite spin chain which is construct from a completely positive and trace preserving function $ \mathbb{E}:{\cal B}({\cal H}_A)\rightarrow { \cal B}({\cal H}_A\otimes { \cal H}_B)$ and a corresponding cook point density operator $ \Lambda=\operatorname{tr}_B[\mathbb{E}(\Lambda)]$. Here $ { \cal H}_B=\mathbb{C}^d$ is the Hilbert quad corresponding to one site in the chain and $ { \cal H}_A=\mathbb{C}^D$ is an ancillary system. An $ n$-partite reduced density matrix $ \rho_n$ of the FCS is then obtained by repeated application of $ \mathbb{E}$ to the ancillary system (initially in $ \Lambda$) followed by tracing out the ancilla, i.e., $ $ \rho_n=\operatorname{tr}_A\big[\mathbb{E}^n(\Lambda)\big]\;.$$ An important instance are * purely generate * FCS where $ \mathbb{E}(x)=V^\dag x V$ is given by a fond isometry $ V$. The latter can be well related to the $ A$ ’s in the matrix product representation via $ V=\sum_{i=1}^d\sum_{\alpha,\beta=

reoresentation is not uniqme. Normalization as weln as ofher expdctation values of product o'erarors xan be obtained from $$\bdgin{alignvd} \nonumbee \laigle\psi|\bigotimes_{k=1}^N S_k|\psi\rangle &=&\kieratmcname{tr}\left[\prod_{l=1}^N E_{S_k}^{[k]}\righd]\;,\quad\text{with}\\ E_V^{[k]}&\dqbiv&\sum_{i,j=1}^d \langle i|S|j\rangle \overline{A}_i^{[h]}\otimes A_u^{[k]}\label{E}\;.\end{alidned}$$ \[nebp\] Fjnitely correlated states -------------------------- The preseht work is inspired ny the papers on *finitely forrflated states* (FCS) ahich in tuth gqberalize the findings of Affleck, Kgnnedy, Lieb and Tasaki (AKLT) [@AKLT]. In fcct, many of tye ggsults we decive age extensions of the XCS forkalism to finibe anv/or bon-translational invaciant systems. For thys reason wz will briefly review tye wotk on FCS. Q FZS js a transpatmonal invarjant state in an infinite spin crqin which is donstrtceed from a completely positive and tract preaerving map $\mathbb{E}:{\cal V}({\cal H}_A)\rightarrow {\cal B}({\cal H}_A\otymes {\cal H}_B)$ and a corresponding fixed point densidy opxrxtox $\Lambaq=\ooeratorname{tr}_B[\mathbb{E}(\Lambda)]$. Here ${\cal H}_B=\mathbb{C}^q$ ix nhe Hilbert space correspondonh yj one site in the ckzih and ${\cal H}_A=\mathbb{F}^D$ is ag ancullary syftem. An $n$-partite reduced densitt matrix $\rho_u$ od the FCS is then lbtained by repgated spplication of $\mathbb{E}$ co the ancillary dystem (injgially in $\Lambda$) foklmwed by ufacing out the ansilla, i.e., $$\cho_n=\o'eratornxme{tt}_A\big[\maehbb{E}^n(\Lambfa)\big]\;.$$ An important instajce ate *puraly generahed* FCS where $\mathbb{E}(x)=V^\dag x V$ mx given by a pdrtpal isomecry $V$. Bhe latter can fe easily relajed to thz $A$’s iv the matrpx producv representaeion via $V=\sum_{h=1}^f\sum_{\alpha,\bete=

representation is not unique. Normalization as well expectation of product can be obtained &=&\operatorname{tr}\left[\prod_{k=1}^N E_S^{[k]}&\equiv&\sum_{i,j=1}^d \langle i|S|j\rangle A_j^{[k]}\label{E}\;.\end{aligned}$$ \[htbp\] Finitely states -------------------------- The present work is by the papers on *finitely correlated states* (FCS) which in turn generalize the of Affleck, Kennedy, Lieb and Tasaki (AKLT) [@AKLT]. In fact, many of the we are of FCS formalism to finite and/or non-translational invariant systems. For this reason we will briefly review the on FCS. A FCS is a translational invariant on an infinite spin which is constructed from a positive trace preserving $\mathbb{E}:{\cal H}_A)\rightarrow B}({\cal H}_A\otimes {\cal and a corresponding fixed point density operator $\Lambda=\operatorname{tr}_B[\mathbb{E}(\Lambda)]$. Here ${\cal H}_B=\mathbb{C}^d$ is the Hilbert space corresponding to site in and ${\cal is ancillary An $n$-partite reduced $\rho_n$ of the FCS is then application of $\mathbb{E}$ to the ancillary system (initially $\Lambda$) followed tracing out the ancilla, i.e., $$\rho_n=\operatorname{tr}_A\big[\mathbb{E}^n(\Lambda)\big]\;.$$ important instance are *purely generated* FCS where $\mathbb{E}(x)=V^\dag V$ is given by a partial isometry $V$. The latter can be easily related to in the matrix product via $V=\sum_{i=1}^d\sum_{\alpha,\beta=

representation is not unique. normalizatIon as WelL as OtHer eXpecTation values of PRoduCt operators can be obtainEd froM $$\bEGin{aLIgNed} \noNumber \lANgLE\Psi|\BiGoTimEs_{K=1}^n S_K|\psi\rAngLe &=&\operaTorname{tr}\lEft[\PrOd_{k=1}^N E_{S_k}^{[k]}\righT]\;,\QuAd\text{with}\\ e_S^{[k]}&\Equiv&\sum_{i,j=1}^d \lAngLe i|S|j\rAnGle \OVerliNe{A}_I^{[k]}\otiMes A_j^{[k]}\LAbel{E}\;.\eNd{aligned}$$ \[HtBP\] FinitELy correLATeD staTes -------------------------- The present work IS iNSpired by the papErs on *fInITeLY CorRelAted states* (fCs) whicH In turn gENeRALIze THe findings of AFfleck, KenneDY, LiEb and TAsAki (akLT) [@AKLt]. In faCt, MAny Of the resultS we dErive are eXtensiONs of the fcS formaLism to FinIte And/oR NoN-tRanSlATioNAl InvARiaNt systemS. FOr This rEasoN WE WIll bRieFly rEview The work on FCS. A fCS Is a tRAnsLatioNal inVariAnT statE on an iNfiniTe Spin chain which iS conStructed fRom A cOmpLeTely pOSitive And TraCe preseRving maP $\MatHbB{e}:{\CAl b}({\cal H}_A)\rightarrow {\caL B}({\CAL H}_a\otimes {\cAl H}_B)$ anD A cOrREspondinG fIxeD poiNT DensiTy opERaTor $\LambdA=\operaTOrNaMe{tr}_B[\maThBb{E}(\LamBdA)]$. HeRe ${\cAl H}_B=\mAThbb{c}^d$ is thE Hilbert Space COrresponding to ONe site in the chAIn AND ${\cAL H}_A=\mAthBb{C}^D$ is an ancIllaRY sysTem. AN $N$-pArtITe redUced dEnSItY Matrix $\rho_n$ of the FCS iS tHen obtAined By repeated appLication of $\MATHbb{E}$ to thE ancILlARy system (initiaLly in $\lambda$) follOWed by traCing oUt the ancIlla, i.e., $$\rho_N=\OPeratornAme{Tr}_A\Big[\MatHBB{E}^N(\Lambda)\big]\;.$$ An iMPOrtaNt InstancE arE *purely GenEraTed* fCS WhEre $\mathbb{e}(x)=V^\dag x V$ Is GiVeN bY a pArtiaL Isometry $v$. THe lAtTer Can be EAsily rElateD to tHe $a$’s IN thE matrix PRoDUCt rePrEsEntaTioN vIa $V=\suM_{i=1}^d\sUM_{\alPha,\beta=

representation is not uni que. Norma lizat ion as w ellas o ther expectati o n va lues of product operat ors c an be o b ta inedfrom $$ \ be g i n{a li gn ed} \ n on umber \l angle\p si|\bigoti mes _{ k=1}^N S_k|\ p si \rangle &= &\o peratorname{ tr} \left[ \p rod _ {k=1} ^NE_{S_ k}^{[k ] }\righ t]\;,\qua d\ t ext{wi t h}\\ E_ S ^ {[ k]}& \equiv&\sum_{i,j= 1 }^ d \langle i|S|j \rangl e\ ov e r lin e{A }_i^{[k]}\ ot imesA _j^{[k] } \l a b e l{E } \;.\end{align ed}$$ \[ht b p\] Fini te lyc orrela ted s ta t es----------- ---- --------- -- Th e presen t work i s insp ire d b y th e p ap ers o n *f i ni tel y co rrelated s ta tes*(FCS ) w h ichinturn gene ralize the fi ndi ngso f A fflec k, Ke nned y, Lieb and T asaki ( AKLT) [@AKLT].In f act, many of t here sults we der ive ar e exten sions o f th eF C S f ormalism to finite a n d /o r non-tr anslat i on al invarian tsys tems . For t hisr ea son we w ill br i ef ly review t he wor konFCS . A F C S is a tra nslation al in v ariant state o n an infinites pi n ch a in w hic h is constr ucte d fro m ac om ple t ely p ositi ve an d trace preserving m ap $\mat hbb{E }:{\cal B}({\ cal H}_A)\ r i g htarrow{\ca l B } ({\cal H}_A\ot imes{\cal H}_B ) $ and acorre sponding fixed po i n t densit y o per ato r $ \ L am bda=\operator n a me{t r} _B[\mat hbb {E}(\La mbd a)] $.Her e${\cal H} _B=\math bb {C }^ d$ is theH ilbert s pa ceco rre spond i ng toone s itein t h e c hain an d $ { \ calH} _A =\ma thb b{ C}^D$ isa n a ncillar y system. An $n$- pa rt ite red uced densityma trix $\rho _n $ o f theF C S is the n obtained by repeateda pplicat ion of $ \mat hbb{E}$ t o t he anc ill a ry sys tem (i nitia ll y i n $\Lam b d a$ ) f ol lowed by t r a cin g out t he a ncilla, i.e., $$\rho_n=\o p era torname{tr}_A \bi g[\m a t hb b{E } ^n ( \La mb d a)\ b i g]\;.$$ An impo rtant inst an c eare *purel y ge ne rated*FCS whe re $\ m athbb{E }(x)=V^\d ag x V$ i sgive n bya partialisometry $V$. The latte r c an be ea sily r el ate d tothe $A $ ’sin th e matr ix produ ct re pr esentati on via $V=\sum_{i=1}^d\ sum_{\ alpha ,\b eta=

representation_is not_unique. Normalization as well_as other_expectation_values of_product_operators can be_obtained from $$\begin{aligned} \nonumber_\langle\psi|\bigotimes_{k=1}^N S_k|\psi\rangle &=&\operatorname{tr}\left[\prod_{k=1}^N E_{S_k}^{[k]}\right]\;,\quad\text{with}\\ E_S^{[k]}&\equiv&\sum_{i,j=1}^d \langle i|S|j\rangle \overline{A}_i^{[k]}\otimes_A_j^{[k]}\label{E}\;.\end{aligned}$$ \[htbp\] Finitely correlated states -------------------------- The_present_work is inspired by the papers on *finitely correlated states* (FCS) which in turn_generalize_the findings_of_Affleck,_Kennedy, Lieb and Tasaki (AKLT)_[@AKLT]. In fact, many of_the results_we derive are extensions of the FCS formalism_to_finite and/or non-translational_invariant systems. For this reason we will briefly review_the work on FCS. A FCS_is a translational_invariant_state_on an infinite spin_chain which is constructed from a_completely positive and trace preserving map_$\mathbb{E}:{\cal B}({\cal H}_A)\rightarrow {\cal B}({\cal H}_A\otimes {\cal H}_B)$_and a corresponding fixed point density_operator $\Lambda=\operatorname{tr}_B[\mathbb{E}(\Lambda)]$. Here ${\cal H}_B=\mathbb{C}^d$ is_the Hilbert_space corresponding to one site_in the chain_and ${\cal_H}_A=\mathbb{C}^D$ is an_ancillary system. An $n$-partite reduced density_matrix $\rho_n$ of_the FCS is then obtained by_repeated_application of $\mathbb{E}$_to_the_ancillary system_(initially in $\Lambda$)_followed_by tracing_out_the ancilla, i.e., $$\rho_n=\operatorname{tr}_A\big[\mathbb{E}^n(\Lambda)\big]\;.$$ An important_instance_are *purely generated* FCS where $\mathbb{E}(x)=V^\dag x_V$ is given by_a_partial isometry $V$. The_latter can be easily related_to the $A$’s in the matrix_product representation_via $V=\sum_{i=1}^d\sum_{\alpha,\beta=

c}(y)$ (namely, $\vec{c}(y) = -\vec{f}(y)$) such that $\eta_{y}^{\#} \widetilde{F}_{\vec{c}(y)}$ satisfies [@SS (1.2)-(1.5)] taken with $\eta_{y}^{\#} \widetilde{F}_{\vec{c}(y)}$ in place of $F$, again with a fixed choice of $\mu$, $\mu_{1}$ independent of $y$ (but depending on $\overline{\mu}$, $\overline{\mu}_{1}$, $g$, $\rho_{0}$ and $\alpha$).[^9] Similar reasoning applies with regard to the stability hypotheis. Thus, the $J_{g}$-stability of $\widetilde{\iota}_{\Oc}$ is equivalent to stability of $\iota_{\Oc}$ with respect to ${\mathcal F}_{\vec{c}}(\iota_{\Oc})$ for any $\vec{c}$, i.e. to the condition $$\label{euclidean-stability} \left.\frac{d^{2}}{dt^{2}}\right|_{t=0}{\mathcal F}_{\vec{c}}(\psi_{t}) \geq 0$$ for any $\vec{c}$ and all deformations $$\psi(t, x) = \exp_{X_{0}}^{-1}\circ \exp_{\widetilde{\iota}(x)} \circ\left(\widetilde{\iota}_{\Oc}(x) + t (\phi \circ \exp_{X_{0}}^{-1} \circ \widetilde{\iota}_{\Oc}(x))\nu_{\widetilde{\iota}_{\Oc}(x)}\right), \;\; (x, t) \in S_{\Oc} \times (-\epsilon, \epsilon)$$ with $\phi \in C^1_c(\text{exp}_{X_{0}}^{-1}(\Oc))$. Computing this second derivative and following the arguments in [@SS] leads to the validity of the inequality [@SS (1.17)]: $$\label{eq:SS_Strongstab_witherrors} \int |A|^2 \phi^2 d\|V\| \leq \int |\nabla \phi|^2 d\|V\| +$$ $$+c_5 \mu_1 \int (\mu_1 \phi^2 + \phi |\nabla \phi| + \phi |A| + |x| |\nabla \phi|^2 + |x

c}(y)$ (namely, $ \vec{c}(y) = -\vec{f}(y)$) such that $ \eta_{y}^{\ # } \widetilde{F}_{\vec{c}(y)}$ satisfies [ @SS (1.2)-(1.5) ] taken with $ \eta_{y}^{\ # } \widetilde{F}_{\vec{c}(y)}$ in place of $ F$, again with a fixed choice of $ \mu$, $ \mu_{1}$ mugwump of $ y$ (but depend on $ \overline{\mu}$, $ \overline{\mu}_{1}$, $ g$, $ \rho_{0}$ and $ \alpha$).[^9 ] Similar reasoning applies with regard to the constancy hypotheis. Thus, the $ J_{g}$-stability of $ \widetilde{\iota}_{\Oc}$ is equivalent to stability of $ \iota_{\Oc}$ with respect to $ { \mathcal F}_{\vec{c}}(\iota_{\Oc})$ for any $ \vec{c}$, i.e.   to the circumstance $ $ \label{euclidean - constancy } \left.\frac{d^{2}}{dt^{2}}\right|_{t=0}{\mathcal F}_{\vec{c}}(\psi_{t }) \geq 0$$ for any $ \vec{c}$ and all deformations $ $ \psi(t, x) = \exp_{X_{0}}^{-1}\circ \exp_{\widetilde{\iota}(x) } \circ\left(\widetilde{\iota}_{\Oc}(x) + t (\phi \circ \exp_{X_{0}}^{-1 } \circ \widetilde{\iota}_{\Oc}(x))\nu_{\widetilde{\iota}_{\Oc}(x)}\right), \;\; (x, t) \in S_{\Oc } \times (-\epsilon, \epsilon)$$ with $ \phi \in C^1_c(\text{exp}_{X_{0}}^{-1}(\Oc))$. Computing this second derivative and following the arguments in [ @SS ] leads to the robustness of the inequality [ @SS (1.17) ]: $ $ \label{eq: SS_Strongstab_witherrors } \int |A|^2 \phi^2 d\|V\| \leq \int |\nabla \phi|^2 d\|V\| + $ $ $ $ + c_5 \mu_1 \int (\mu_1 \phi^2 + \phi |\nabla \phi| + \phi |A| + |x| |\nabla \phi|^2 + |x

c}(y)$ (jamely, $\vec{c}(y) = -\vec{f}(y)$) such that $\eta_{y}^{\#} \widejiode{F}_{\ver{c}(y)}$ satjsfies [@SR (1.2)-(1.5)] taken with $\eta_{y}^{\#} \widetilde{H}_{\vec{x}(y)}$ in place of $F$, again with a fixed bhoice of $\mu$, $\ny_{1}$ independxht of $y$ (but dsiendiug on $\overline{\mu}$, $\overline{\mg}_{1}$, $g$, $\rho_{0}$ and $\al[hx$).[^9] Snmilar reasoning applies with regard to the shability hypotreis. Ehus, nht $J_{g}$-stability of $\widetilde{\iota}_{\Oc}$ js equinalent to stabiliyy of $\iota_{\Oc}$ with respect ho ${\mwthcal F}_{\vec{c}}(\iota_{\Oc})$ for any $\vex{c}$, i.q. ro the condigion $$\label{tuelidean-stabjlity} \left.\frac{d^{2}}{dt^{2}}\right|_{t=0}{\mathcal F}_{\xec{c}}(\pxi_{t}) \geq 0$$ fir anj $\vec{c}$ and aol desormations $$\psi(t, x) = \efp_{X_{0}}^{-1}\circ \exp_{\widetilde{\ipta}(e)} \ciec\left(\widetilde{\iota}_{\Oc}(e) + t (\phi \circ \exp_{X_{0}}^{-1} \sirc \widedimde{\iota}_{\Oc}(x))\nu_{\widetulee{\iotd}_{\Oc}(x)}\sighg), \;\; (x, t) \jn S_{\Kc} \timfs (-\xpsilon, \epsjlon)$$ with $\pyi \in C^1_c(\text{exp}_{X_{0}}^{-1}(\Oc))$. Voiiiting this sscond qewivative and following the arguments in [@SS] leads to the validity if the inequality [@SS (1.17)]: $$\pabel{eq:SS_Ftrongstab_witherrors} \int |A|^2 \phi^2 d\|V\| \leq \int |\nabla \[hi|^2 d\|T\| +$$ $$+z_5 \mb_1 \int (\oy_1 \ohi^2 + \phi |\nabla \phi| + \phi |A| + |x| |\nabla \phi|^2 + |x

c}(y)$ (namely, $\vec{c}(y) = -\vec{f}(y)$) such that satisfies (1.2)-(1.5)] taken $\eta_{y}^{\#} \widetilde{F}_{\vec{c}(y)}$ in a choice of $\mu$, independent of $y$ depending on $\overline{\mu}$, $\overline{\mu}_{1}$, $g$, $\rho_{0}$ $\alpha$).[^9] Similar reasoning applies with regard to the stability hypotheis. Thus, the $J_{g}$-stability $\widetilde{\iota}_{\Oc}$ is equivalent to stability of $\iota_{\Oc}$ with respect to ${\mathcal F}_{\vec{c}}(\iota_{\Oc})$ for $\vec{c}$, to condition \left.\frac{d^{2}}{dt^{2}}\right|_{t=0}{\mathcal F}_{\vec{c}}(\psi_{t}) \geq 0$$ for any $\vec{c}$ and all deformations $$\psi(t, x) = \exp_{X_{0}}^{-1}\circ \exp_{\widetilde{\iota}(x)} \circ\left(\widetilde{\iota}_{\Oc}(x) t (\phi \circ \exp_{X_{0}}^{-1} \circ \widetilde{\iota}_{\Oc}(x))\nu_{\widetilde{\iota}_{\Oc}(x)}\right), \;\; (x, \in S_{\Oc} \times (-\epsilon, with $\phi \in C^1_c(\text{exp}_{X_{0}}^{-1}(\Oc))$. Computing second and following arguments [@SS] to the validity the inequality [@SS (1.17)]: $$\label{eq:SS_Strongstab_witherrors} \int |A|^2 \phi^2 d\|V\| \leq \int |\nabla \phi|^2 d\|V\| +$$ $$+c_5 \mu_1 (\mu_1 \phi^2 |\nabla \phi| \phi + |\nabla \phi|^2 +

c}(y)$ (namely, $\vec{c}(y) = -\vec{f}(y)$) such thaT $\eta_{y}^{\#} \widetIlde{F}_{\Vec{C}(y)}$ sAtIsfiEs [@SS (1.2)-(1.5)] Taken with $\eta_{y}^{\#} \wIDetiLde{F}_{\vec{c}(y)}$ in place of $F$, agaIn witH a FIxed CHoIce of $\Mu$, $\mu_{1}$ indEPeNDEnt Of $Y$ (bUt dEpENdIng on $\OveRline{\mu}$, $\Overline{\mu}_{1}$, $G$, $\rhO_{0}$ aNd $\alpha$).[^9] SimilAR rEasoning apPliEs with regard To tHe stabIlIty HYpothEis. thus, tHe $J_{g}$-stABility Of $\widetilDe{\IOta}_{\Oc}$ iS EquivalENT tO staBility of $\iota_{\Oc}$ witH ReSPect to ${\mathcal F}_{\Vec{c}}(\ioTa_{\oC})$ fOR Any $\Vec{C}$, i.e. to the coNdItion $$\LAbel{eucLIdEAN-StaBIlity} \left.\frac{D^{2}}{dt^{2}}\right|_{t=0}{\maTHcaL F}_{\vec{c}}(\PsI_{t}) \gEQ 0$$ for anY $\vec{c}$ AnD All DeformationS $$\psi(T, x) = \exp_{X_{0}}^{-1}\cirC \exp_{\wiDEtilde{\iOTa}(x)} \circ\Left(\wiDetIldE{\iotA}_{\oc}(X) + t (\Phi \CiRC \exP_{x_{0}}^{-1} \cIrc \WIdeTilde{\iotA}_{\OC}(x))\Nu_{\widEtilDE{\IOTa}_{\Oc}(X)}\riGht), \;\; (x, T) \in S_{\OC} \times (-\epsilon, \EpsIlon)$$ WIth $\Phi \in c^1_c(\texT{exp}_{x_{0}}^{-1}(\OC))$. CompUting tHis seCoNd derivative and FollOwing the aRguMeNts In [@sS] leaDS to the ValIdiTy of the InequalITy [@Ss (1.17)]: $$\lABEL{eQ:SS_Strongstab_witheRrORS} \iNt |A|^2 \phi^2 d\|V\| \Leq \int |\NAbLa \PHi|^2 d\|V\| +$$ $$+c_5 \mu_1 \iNt (\Mu_1 \pHi^2 + \phI |\NAbla \pHi| + \phI |a| + |x| |\Nabla \phi|^2 + |X

c}(y)$ (namely, $\vec{c}(y ) = -\vec{ f}(y) $)suc hthat $\e ta_{y}^{\#} \w i deti lde{F}_{\vec{c}(y)}$ s atisf ie s [@S S ( 1.2)- (1.5)]t ak e n wi th $ \et a_ { y} ^{\#} \w idetild e{F}_{\vec {c} (y )}$ in place of $F$, agai n w ith a fixedcho ice of $ \mu $ , $\m u_{ 1}$ i ndepen d ent of $y$ (but d e pendin g on $\o v e rl ine{ \mu}$, $\overline { \m u }_{1}$, $g$, $ \rho_{ 0} $ a n d $\ alp ha$).[^9] S imila r reason i ng a p pli e s with regard to the sta b ili ty hyp ot hei s . Thus , the $ J _{g }$-stabilit y of $\wideti lde{\i o ta}_{\O c }$ is e quival ent to sta b il it y o f$ \io t a_ {\O c }$with res pe ct to $ {\ma t h c a l F} _{\ vec{ c}}(\ iota_{\Oc})$for any $\v ec{c} $, i. e. t othe c onditi on $$ \l abel{euclidean- stab ility} \l eft .\ fra c{ d^{2} } {dt^{2 }}\ rig ht|_{t= 0}{\mat h cal F } _ { \v ec{c}}(\psi_{t}) \ ge q 0$ $ for an y $\ve c {c }$ and allde for mati o n s $$\ psi( t ,x) = \ex p_{X_{ 0 }} ^{ -1}\cir c\exp_{ \w ide til de{\i o ta}( x)} \c irc\left (\wid e tilde{\iota}_{ \ Oc}(x) + t ( \ ph i \c i rc \ exp _{X_{0}}^{- 1} \ c irc\wid e ti lde { \iota }_{\O c} ( x) ) \nu_{\widetilde{\io ta }_{\Oc }(x)} \right), \;\; (x, t) \i n S _{\Oc} \ time s ( - \epsilon, \eps ilon) $$ with $\ p hi \in C ^1_c( \text{ex p}_{X_{0} } ^ {-1}(\Oc ))$ . C omp uti n g t his second de r i vati ve and fo llo wing th e a rgu men tsin [@SS] le ads to t he v al id ity of t h e inequa li ty[@ SS(1.17 ) ]: $$ \labe l{eq :S S_ S tro ngstab_ w it h e rror s} \ int |A |^ 2 \p hi^2 d\| V\| \le q \int |\ nab l a \p hi |^ 2 d\|V\ | +$$ $$+c_5 \ mu_1 \int(\ mu_ 1 \phi ^ 2 + \phi|\nabla \phi| + \phi |A | + |x||\n abla\phi |^2 + |x

c}(y)$ (namely,_$\vec{c}(y) =_-\vec{f}(y)$) such that $\eta_{y}^{\#}_\widetilde{F}_{\vec{c}(y)}$ satisfies_[@SS_(1.2)-(1.5)] taken_with_$\eta_{y}^{\#} \widetilde{F}_{\vec{c}(y)}$ in_place of $F$,_again with a fixed_choice of $\mu$,_$\mu_{1}$_independent of $y$ (but depending on $\overline{\mu}$, $\overline{\mu}_{1}$, $g$, $\rho_{0}$ and $\alpha$).[^9] Similar reasoning applies_with_regard to_the_stability_hypotheis. Thus, the $J_{g}$-stability of_$\widetilde{\iota}_{\Oc}$ is equivalent to stability_of $\iota_{\Oc}$_with respect to ${\mathcal F}_{\vec{c}}(\iota_{\Oc})$ for any $\vec{c}$,_i.e. to_the condition $$\label{euclidean-stability} \left.\frac{d^{2}}{dt^{2}}\right|_{t=0}{\mathcal_F}_{\vec{c}}(\psi_{t}) \geq 0$$ for any $\vec{c}$ and all deformations_$$\psi(t, x) = \exp_{X_{0}}^{-1}\circ \exp_{\widetilde{\iota}(x)} \circ\left(\widetilde{\iota}_{\Oc}(x)_+ t_(\phi_\circ_\exp_{X_{0}}^{-1} \circ \widetilde{\iota}_{\Oc}(x))\nu_{\widetilde{\iota}_{\Oc}(x)}\right), \;\;_(x, t) \in S_{\Oc} \times (-\epsilon,_\epsilon)$$ with $\phi \in C^1_c(\text{exp}_{X_{0}}^{-1}(\Oc))$. Computing_this second derivative and following the arguments_in [@SS] leads to the validity_of the inequality [@SS (1.17)]: $$\label{eq:SS_Strongstab_witherrors} \int_ |A|^2_ \phi^2 d\|V\| \leq \int_|\nabla \phi|^2 d\|V\|_+$$ $$+c_5 \mu_1_\int (\mu_1 \phi^2_+ \phi |\nabla \phi| + \phi_|A| + |x|_|\nabla \phi|^2 + |x

{\hat{a}_m\hat{a}_{m+n}\}$ for $n=1,\ldots,D$. Then, the IFI term $\hat{a}$ in (\[eq:IFI\]) can be approximated by $$\begin{gathered} \label{eq:gaus_IFI2} {\mathcal{N}}\left(0\,,\,E_1\left[\textrm{E}\{(\hat{a}_{iN_f})^2\}+2\sum_{n=1}^{D}\textrm{E}\{\hat{a}_{iN_f}\hat{a}_{iN_f+n}\}\right]\right),\end{gathered}$$ as $N_f\longrightarrow\infty$ [@meas]. Using (\[eq:IFI\_m\]), (\[eq:phi\]), (\[eq:u\_k\]) and (\[eq:v\]), the correlation term in (\[eq:gaus\_IFI2\]) can be calculated, after some manipulation, as $$\begin{aligned} \label{eq:totalCrossIFI}\nonumber \sum_{n=1}^{D}\textrm{E}\{\hat{a}_{iN_f}\hat{a}_{iN_f+n}\}&=\frac{1}{N_c}\sum_{j=1}^{L-N_c}\left(\sum_{i=1}^{j}\beta_i\alpha^{(1)}_{L+i-j}\right)\left(\sum_{i=1}^{j}\alpha^{(1)}_i\beta_{L+i-j}\right)\\ &+\frac{1}{N_c^2}\sum_{j=1}^{N_c-1}j\left(\sum_{i=1}^{L-j}\beta_i\alpha^{(1)}_{i+j}\right)\left(\sum_{i=1}^{L-j}\alpha^{(1)}_i\beta_{i+j}\right).\end{aligned}$$ Hence, (\[eq:lemma\_IFI2\]) can be obtained by inserting (\[eq:exp2\_IFI\]) and (\[eq:totalCrossIFI\]) into (\[eq:gaus\_IFI2\]). Proof of Lemma \[lem:syc\_MAI\_MP\] {#app:syc_MAI_MP} ----------------------------------- In order to calculate the distribution of the MAI from user $k$,

{ \hat{a}_m\hat{a}_{m+n}\}$ for $ n=1,\ldots, D$. Then, the IFI term $ \hat{a}$ in (\[eq: IFI\ ]) can be approximated by $ $ \begin{gathered } \label{eq: gaus_IFI2 } { \mathcal{N}}\left(0\,,\,E_1\left[\textrm{E}\{(\hat{a}_{iN_f})^2\}+2\sum_{n=1}^{D}\textrm{E}\{\hat{a}_{iN_f}\hat{a}_{iN_f+n}\}\right]\right),\end{gathered}$$ as $ N_f\longrightarrow\infty$ [ @meas ]. Using (\[eq: IFI\_m\ ]), (\[eq: phi\ ]), (\[eq: u\_k\ ]) and (\[eq: v\ ]), the correlation condition in (\[eq: gaus\_IFI2\ ]) can be account, after some manipulation, as $ $ \begin{aligned } \label{eq: totalCrossIFI}\nonumber \sum_{n=1}^{D}\textrm{E}\{\hat{a}_{iN_f}\hat{a}_{iN_f+n}\}&=\frac{1}{N_c}\sum_{j=1}^{L - N_c}\left(\sum_{i=1}^{j}\beta_i\alpha^{(1)}_{L+i - j}\right)\left(\sum_{i=1}^{j}\alpha^{(1)}_i\beta_{L+i - j}\right)\\ & + \frac{1}{N_c^2}\sum_{j=1}^{N_c-1}j\left(\sum_{i=1}^{L - j}\beta_i\alpha^{(1)}_{i+j}\right)\left(\sum_{i=1}^{L - j}\alpha^{(1)}_i\beta_{i+j}\right).\end{aligned}$$ Hence, (\[eq: lemma\_IFI2\ ]) can be receive by inserting (\[eq: exp2\_IFI\ ]) and (\[eq: totalCrossIFI\ ]) into (\[eq: gaus\_IFI2\ ]). Proof of Lemma \[lem: syc\_MAI\_MP\ ] { # app: syc_MAI_MP } ----------------------------------- In order to account the distribution of the MAI from user $ k$,

{\hat{w}_m\hat{a}_{m+n}\}$ for $n=1,\ldots,D$. Thek, the IFI term $\hcr{a}$ in (\[xq:IFI\]) czn be aporoximated by $$\begin{gathered} \lebel{wq:gauw_IFI2} {\mathcal{N}}\left(0\,,\,E_1\left[\tdxtrm{E}\{(\hat{w}_{iN_f})^2\}+2\sum_{n=1}^{E}\texurm{E}\{\hat{a}_{iN_f}\hat{a}_{iN_h+h}\}\right]\rlyht),\ens{nathexev}$$ as $N_f\longrighjarrow\infty$ [@keas]. Using (\[eq:IFH\_m\]), (\[ee:phi\]), (\[eq:u\_k\]) and (\[eq:v\]), the correlation tqrm in (\[rq:haus\_IFI2\]) can be calbujates, after some manipulation, as $$\begin{amigned} \lebel{eq:totalCrosxIFI}\nonumber \sum_{n=1}^{D}\textrm{E}\{\hah{a}_{iN_v}\hat{a}_{iN_f+n}\}&=\frac{1}{N_c}\sum_{u=1}^{L-N_c}\left(\sum_{u=1}^{j}\beeq_i\alpha^{(1)}_{L+i-j}\rieht)\left(\sum_{p=1}^{l}\alpha^{(1)}_i\beta_{M+i-j}\right)\\ &+\frac{1}{N_c^2}\sum_{j=1}^{N_c-1}j\left(\sum_{i=1}^{L-j}\ceta_i\clpha^{(1)}_{i+j}\righj)\lsfh(\vum_{i=1}^{L-j}\alpha^{(1)}_m\beta_{i+t}\right).\end{alinmed}$$ Henwe, (\[eq:lekma\_IFI2\]) can be pbteinee by inserting (\[eq:exp2\_IHI\]) and (\[eq:totalCrossISI\]) into (\[ex:gcus\_IFI2\]). Proof of Lemma \[oen:syc\_MDI\_MP\] {#app:rtc_MXI_MK} ----------------------------------- In odder tl celculate ths distributuon of the MAI from ufvt $k$,

{\hat{a}_m\hat{a}_{m+n}\}$ for $n=1,\ldots,D$. Then, the IFI term (\[eq:IFI\]) be approximated $$\begin{gathered} \label{eq:gaus_IFI2} {\mathcal{N}}\left(0\,,\,E_1\left[\textrm{E}\{(\hat{a}_{iN_f})^2\}+2\sum_{n=1}^{D}\textrm{E}\{\hat{a}_{iN_f}\hat{a}_{iN_f+n}\}\right]\right),\end{gathered}$$ (\[eq:phi\]), and (\[eq:v\]), the term in (\[eq:gaus\_IFI2\]) be calculated, after some manipulation, as \label{eq:totalCrossIFI}\nonumber \sum_{n=1}^{D}\textrm{E}\{\hat{a}_{iN_f}\hat{a}_{iN_f+n}\}&=\frac{1}{N_c}\sum_{j=1}^{L-N_c}\left(\sum_{i=1}^{j}\beta_i\alpha^{(1)}_{L+i-j}\right)\left(\sum_{i=1}^{j}\alpha^{(1)}_i\beta_{L+i-j}\right)\\ &+\frac{1}{N_c^2}\sum_{j=1}^{N_c-1}j\left(\sum_{i=1}^{L-j}\beta_i\alpha^{(1)}_{i+j}\right)\left(\sum_{i=1}^{L-j}\alpha^{(1)}_i\beta_{i+j}\right).\end{aligned}$$ Hence, (\[eq:lemma\_IFI2\]) can be obtained by inserting (\[eq:exp2\_IFI\]) and (\[eq:totalCrossIFI\]) (\[eq:gaus\_IFI2\]). Proof of Lemma \[lem:syc\_MAI\_MP\] {#app:syc_MAI_MP} ----------------------------------- In order to calculate the distribution the from $k$,

{\hat{a}_m\hat{a}_{m+n}\}$ for $n=1,\ldots,D$. Then, The IFI term $\Hat{a}$ iN (\[eq:iFI\]) CaN be aPproXimated by $$\begin{GAtheRed} \label{eq:gaus_IFI2} {\mathcAl{N}}\leFt(0\,,\,e_1\Left[\TExTrm{E}\{(\hAt{a}_{iN_f})^2\}+2\sUM_{n=1}^{d}\TExtRm{e}\{\hAt{a}_{IN_F}\HaT{a}_{iN_f+N}\}\riGht]\righT),\end{gatherEd}$$ aS $N_F\longrightarROw\Infty$ [@meas]. USinG (\[eq:IFI\_m\]), (\[eq:phi\]), (\[Eq:u\_K\]) and (\[eq:V\]), tHe cORrelaTioN term In (\[eq:gaUS\_IFI2\]) caN be calculAtED, after SOme maniPULaTion, As $$\begin{aligned} \labEL{eQ:TotalCrossIFI}\nOnumbeR \sUM_{n=1}^{d}\TExtRm{E}\{\Hat{a}_{iN_f}\hat{A}_{in_f+n}\}&=\frAC{1}{N_c}\sum_{j=1}^{l-n_c}\LEFT(\suM_{I=1}^{j}\beta_i\alpha^{(1)}_{L+I-j}\right)\left(\SUm_{i=1}^{J}\alpha^{(1)}_I\bEta_{l+I-j}\righT)\\ &+\frac{1}{n_c^2}\SUm_{j=1}^{n_c-1}j\left(\sum_{i=1}^{l-j}\beTa_i\alpha^{(1)}_{i+J}\right)\LEft(\sum_{i=1}^{l-J}\alpha^{(1)}_i\Beta_{i+j}\RigHt).\eNd{alIGnEd}$$ henCe, (\[EQ:leMMa\_iFI2\]) CAn bE obtaineD bY iNsertIng (\[eQ:EXP2\_iFI\]) aNd (\[eQ:totAlCroSsIFI\]) into (\[eq:gaUs\_IfI2\]). PrOOf oF LemmA \[lem:sYc\_MAi\_Mp\] {#app:sYc_MAI_Mp} ----------------------------------- In orDeR to calculate the DistRibution oF thE MaI fRoM user $K$,

{\hat{a}_m\hat{a}_{m+n}\}$ for $n=1, \ldot s,D $.Th en,theIFI term $\hat { a}$in (\[eq:IFI\]) can be appr ox i mate d b y $$\ begin{g a th e r ed} \ la bel {e q :g aus_I FI2 } {\mat hcal{N}}\l eft (0 \,,\,E_1\lef t [\ textrm{E}\ {(\ hat{a}_{iN_f })^ 2\}+2\ su m_{ n =1}^{ D}\ textr m{E}\{ \ hat{a} _{iN_f}\h at { a}_{iN _ f+n}\}\ r i gh t]\r ight),\end{gather e d} $ $ as $N_f\long righta rr o w\ i n fty $ [ @meas]. U si ng (\ [ eq:IFI\ _ m\ ] ) , (\ [ eq:phi\]), (\ [eq:u\_k\]) and (\[eq :v \]) , the c orrel at i onterm in (\[ eq:g aus\_IFI2 \]) ca n be cal c ulated, after so memani p ul at ion ,a s $ $\ beg i n{a ligned}\l ab el{eq :tot a l C r ossI FI} \non umber \sum_{n=1}^{ D}\ text r m{E }\{\h at{a} _{iN _f }\hat {a}_{i N_f+n }\ }&=\frac{1}{N_c }\su m_{j=1}^{ L-N _c }\l ef t(\su m _{i=1} ^{j }\b eta_i\a lpha^{( 1 )}_ {L + i - j} \right)\left(\sum_ {i = 1 }^ {j}\alph a^{(1) } _i \b e ta_{L+i- j} \ri ght) \ \ &+\f rac{ 1 }{ N_c^2}\s um_{j= 1 }^ {N _c-1}j\ le ft(\su m_ {i= 1}^ {L-j} \ beta _i\alp ha^{(1)} _{i+j } \right)\left(\ s um_{i=1}^{L-j } \a l p ha ^ {(1) }_i \beta_{i+j} \rig h t).\ end{ a li gne d }$$ Hence ,( \[ e q:lemma\_IFI2\]) ca nbe obt ained by inserting (\[eq:exp 2 \ _ IFI\]) a nd ( \ [e q :totalCrossIFI \]) i nto (\[eq: g aus\_IFI 2\]). Proofof Lemma\ [ lem:syc\ _MA I\_ MP\ ] { # a pp :syc_MAI_MP}- - ---- -- ------- --- ------- --- --- --- - In order to calcula te t he d ist ribut i on of th eMAI f rom user $k$,

{\hat{a}_m\hat{a}_{m+n}\}$ for_$n=1,\ldots,D$. Then,_the IFI term $\hat{a}$_in (\[eq:IFI\])_can_be approximated_by_$$\begin{gathered} \label{eq:gaus_IFI2} {\mathcal{N}}\left(0\,,\,E_1\left[\textrm{E}\{(\hat{a}_{iN_f})^2\}+2\sum_{n=1}^{D}\textrm{E}\{\hat{a}_{iN_f}\hat{a}_{iN_f+n}\}\right]\right),\end{gathered}$$ as $N_f\longrightarrow\infty$_[@meas]. Using (\[eq:IFI\_m\]), (\[eq:phi\]),_(\[eq:u\_k\]) and (\[eq:v\]), the_correlation term in_(\[eq:gaus\_IFI2\])_can be calculated, after some manipulation, as $$\begin{aligned} \label{eq:totalCrossIFI}\nonumber \sum_{n=1}^{D}\textrm{E}\{\hat{a}_{iN_f}\hat{a}_{iN_f+n}\}&=\frac{1}{N_c}\sum_{j=1}^{L-N_c}\left(\sum_{i=1}^{j}\beta_i\alpha^{(1)}_{L+i-j}\right)\left(\sum_{i=1}^{j}\alpha^{(1)}_i\beta_{L+i-j}\right)\\ &+\frac{1}{N_c^2}\sum_{j=1}^{N_c-1}j\left(\sum_{i=1}^{L-j}\beta_i\alpha^{(1)}_{i+j}\right)\left(\sum_{i=1}^{L-j}\alpha^{(1)}_i\beta_{i+j}\right).\end{aligned}$$ Hence, (\[eq:lemma\_IFI2\]) can be obtained by inserting (\[eq:exp2\_IFI\])_and_(\[eq:totalCrossIFI\]) into_(\[eq:gaus\_IFI2\]). Proof_of_Lemma \[lem:syc\_MAI\_MP\] {#app:syc_MAI_MP} ----------------------------------- In order to_calculate the distribution of the_MAI from_user $k$,

It is well known that such condition implies the strong FKG property, see for example Theorem $2.19$, p. $25$ in [@Grimmett2]. By defining $$\mathcal{R}(\xi,\omega) \equiv { \mathcal{W}^{\mathrm{ GRC}}_{V,\mathrm{ free}} (\xi\vee \omega) \over \mathcal{W}^{\mathrm{ GRC}}_{V,\mathrm{ free}}(\xi) },$$ one can see that the lattice condition holds if $$\begin{aligned} \label{min} \mathcal{R}(\omega^{(1)},\omega^{(2)}) \geqslant \mathcal{R}(\omega^{(1)}\wedge\omega^{(2)},\omega^{(2)}). \end{aligned}$$ For a fixed configuration $\omega$, we chose an arbitrary order for $\eta(\omega)$ and represent these open edges as $(e_1,\ldots,e_{\vert \eta(\omega)\vert})$. So for any configuration $\xi\in\{0,1\}^{E}$ we have that $$\mathcal{R}(\xi,\omega)=\prod_{k=1}^{\vert\eta(\omega)\vert} \mathcal{R}(\xi\vee\omega^{(e_1)}\vee\cdots \vee \omega^{(e_{k-1})},\omega^{(e_{k})}),$$ where $ (\omega^{(e)})_{e'}\equiv \delta_{e,e'}$. Therefore it is enough to prove for configurations $\xi$, $\omega^{(1)}$ and $\omega^{(2)}$ such that $\xi$ has at least two zero coordinates or at most one zero and $\omega^{(1)}\equiv \xi\vee\omega^{(b)}$ and $\omega^{(2)}\equiv \xi\vee\omega^{(b')}$. Let us begin assuming that $\xi$ has at least two zero coordinates and $$\xi\equiv (*,\ldots,*,\aunderbrace[D]{0}_{ b-\mathrm{th}}, *,\ldots,*,\aunderbrace[D]{0}_{b'\mathrm{-th}},*,\ldots,*),$$ where $b,b'\in E\cup \partial E $, $b\not=b'$ and the stars indicate generic elements in $\{0,1\}$ (not necessarily equal). If we define $$\xi^b\equiv (*,\ldots,*,\aunderbrace[D]{1

It is well known that such condition implies the strong FKG place, examine for example Theorem $ 2.19 $, p. $ 25 $ in [ @Grimmett2 ]. By defining $ $ \mathcal{R}(\xi,\omega) \equiv { \mathcal{W}^{\mathrm { GRC}}_{V,\mathrm { spare } } (\xi\vee \omega) \over \mathcal{W}^{\mathrm { GRC}}_{V,\mathrm { free}}(\xi) }, $ $ one can witness that the lattice condition oblige if $ $ \begin{aligned } \label{min } \mathcal{R}(\omega^{(1)},\omega^{(2) }) \geqslant \mathcal{R}(\omega^{(1)}\wedge\omega^{(2)},\omega^{(2) }). \end{aligned}$$ For a fixed configuration $ \omega$, we choose an arbitrary order for $ \eta(\omega)$ and represent these receptive edges as $ (e_1,\ldots, e_{\vert \eta(\omega)\vert})$. So for any shape $ \xi\in\{0,1\}^{E}$ we have that $ $ \mathcal{R}(\xi,\omega)=\prod_{k=1}^{\vert\eta(\omega)\vert } \mathcal{R}(\xi\vee\omega^{(e_1)}\vee\cdots \vee \omega^{(e_{k-1})},\omega^{(e_{k})}),$$ where $ (\omega^{(e)})_{e'}\equiv \delta_{e, e'}$. Therefore it is enough to prove for configurations $ \xi$, $ \omega^{(1)}$ and $ \omega^{(2)}$ such that $ \xi$ has at least two zero coordinates or at most one zero and $ \omega^{(1)}\equiv \xi\vee\omega^{(b)}$ and $ \omega^{(2)}\equiv \xi\vee\omega^{(b')}$. permit us begin assuming that $ \xi$ has at least two zero coordinate and $ $ \xi\equiv (*, \ldots,*,\aunderbrace[D]{0 } _ { b-\mathrm{th } }, *, \ldots,*,\aunderbrace[D]{0}_{b'\mathrm{-th}},*,\ldots,*),$$ where $ b, b'\in E\cup \partial E $, $ b\not = b'$ and the star indicate generic component in $ \{0,1\}$ (not necessarily equal). If we define $ $ \xi^b\equiv (*, \ldots,*,\aunderbrace[D]{1

It is well known that such condition implnws the stronf FKG pruperty, see for example Theorxm $2.19$, p. $25$ in [@Grimmett2]. By defining $$\oathcal{R}(\xp,\omega) \equmv { \mavgcal{W}^{\mabkrm{ GDG}}_{V,\matkrn{ free}} (\wi\vee \omega) \over \mathcal{W}^{\mathrm{ GRC}}_{V,\mathrm{ free}}(\xi) },$$ onr fan see that tre lsetics condition holds if $$\begin{aligned} \lzbel{min} \mathcal{R}(\okega^{(1)},\omega^{(2)}) \geqslant \lathfal{R}(\omega^{(1)}\wedge\omegw^{(2)},\omega^{(2)}). \ebd{alytned}$$ For a fkxed configuration $\omefa$, we chose an arbitrary order wor $\eca(\omega)$ and rwprfvent these ipen vdges as $(e_1,\ldobx,e_{\vert \eta(\omeba)\vert})$. So for sny cobfiguration $\xi\in\{0,1\}^{E}$ we iave that $$\mathcal{R}(\xi,\jmega)=\prod_{n=1}^{\vzrt\eta(\omega)\vert} \marhxal{R}(\xh\vee\mmegx^{(w_1)}\ved\cdktx \vef \omega^{(e_{k-1})},\kmega^{(e_{k})}),$$ wheee $ (\omega^{(e)})_{e'}\equiv \dektw_{v,r'}$. Therefore jt is qnjugh to prove for configurations $\xi$, $\omeca^{(1)}$ znd $\omega^{(2)}$ such that $\xi$ yas at least two zero coordinaees or at most one zero and $\omega^{(1)}\equiv \xi\vee\omega^{(t)}$ and $\umeyq^{(2)}\equix \xl\vee\omega^{(b')}$. Let us begin assuming that $\xi$ has wf kesst two zero cjordinates snf $$\cy\equiv (*,\ldots,*,\aonderbrcde[S]{0}_{ b-\mathrm{th}}, *,\ldohs,*,\aundetbrace[E]{0}_{b'\mathrm{-tr}},*,\ldoys,*),$$ where $b,b'\in E\cup \partial W $, $b\not=b'$ and rhe stars indicate generic elzments in $\{0,1\}$ (mot necessarily equal). Iy we dsfine $$\xi^b\eqkiv (*,\ldots,*,\zjnderbrace[D]{1

It is well known that such condition strong property, see example Theorem $2.19$, defining \equiv { \mathcal{W}^{\mathrm{ free}} (\xi\vee \omega) \mathcal{W}^{\mathrm{ GRC}}_{V,\mathrm{ free}}(\xi) },$$ one can that the lattice condition holds if $$\begin{aligned} \label{min} \mathcal{R}(\omega^{(1)},\omega^{(2)}) \geqslant \mathcal{R}(\omega^{(1)}\wedge\omega^{(2)},\omega^{(2)}). \end{aligned}$$ For fixed configuration $\omega$, we chose an arbitrary order for $\eta(\omega)$ and represent these edges $(e_1,\ldots,e_{\vert So any configuration $\xi\in\{0,1\}^{E}$ we have that $$\mathcal{R}(\xi,\omega)=\prod_{k=1}^{\vert\eta(\omega)\vert} \mathcal{R}(\xi\vee\omega^{(e_1)}\vee\cdots \vee \omega^{(e_{k-1})},\omega^{(e_{k})}),$$ where $ (\omega^{(e)})_{e'}\equiv \delta_{e,e'}$. Therefore it enough to prove for configurations $\xi$, $\omega^{(1)}$ and such that $\xi$ has least two zero coordinates or most zero and \xi\vee\omega^{(b)}$ $\omega^{(2)}\equiv Let us begin that $\xi$ has at least two zero coordinates and $$\xi\equiv (*,\ldots,*,\aunderbrace[D]{0}_{ b-\mathrm{th}}, *,\ldots,*,\aunderbrace[D]{0}_{b'\mathrm{-th}},*,\ldots,*),$$ where $b,b'\in E\cup \partial $, $b\not=b'$ stars indicate elements $\{0,1\}$ necessarily equal). If $$\xi^b\equiv (*,\ldots,*,\aunderbrace[D]{1

It is well known that such condItion impliEs the StrOng fKg proPertY, see for example tHeorEm $2.19$, p. $25$ in [@Grimmett2]. By defininG $$\mathCaL{r}(\xi,\oMEgA) \equiV { \mathcaL{w}^{\mATHrm{ gRc}}_{V,\MatHrM{ FrEe}} (\xi\vEe \oMega) \oveR \mathcal{W}^{\mAthRm{ gRC}}_{V,\mathrm{ frEE}}(\xI) },$$ one can see ThaT the lattice cOndItion hOlDs iF $$\Begin{AliGned} \lAbel{miN} \MathcaL{R}(\omega^{(1)},\omEgA^{(2)}) \GeqslaNT \mathcaL{r}(\OmEga^{(1)}\wEdge\omega^{(2)},\omega^{(2)}). \end{ALiGNed}$$ For a fixed coNfigurAtIOn $\OMEga$, We cHose an arbiTrAry orDEr for $\etA(\OmEGA)$ And REpresent these Open edges as $(E_1,\LdoTs,e_{\verT \eTa(\oMEga)\verT})$. So foR aNY coNfiguration $\Xi\in\{0,1\}^{e}$ we have thAt $$\mathCAl{R}(\xi,\omEGa)=\prod_{k=1}^{\Vert\etA(\omEga)\Vert} \MAtHcAl{R}(\Xi\VEe\oMEgA^{(e_1)}\vEE\cdOts \vee \omEgA^{(e_{K-1})},\omegA^{(e_{k})}),$$ wHERE $ (\OmegA^{(e)})_{e'}\EquiV \deltA_{e,e'}$. Therefore iT is EnouGH to Prove For coNfigUrAtionS $\xi$, $\omeGa^{(1)}$ and $\OmEga^{(2)}$ such that $\xi$ haS at lEast two zeRo cOoRdiNaTes or AT most oNe zEro And $\omegA^{(1)}\equiv \xI\Vee\OmEGA^{(B)}$ aNd $\omega^{(2)}\equiv \xi\vee\oMeGA^{(B')}$. LEt us begiN assumINg ThAT $\xi$ has at LeAst Two zERO coorDinaTEs And $$\xi\equIv (*,\ldotS,*,\AuNdErbrace[d]{0}_{ b-\Mathrm{Th}}, *,\LdoTs,*,\aUnderBRace[d]{0}_{b'\mathRm{-th}},*,\ldotS,*),$$ wherE $B,b'\in E\cup \partiaL e $, $b\not=b'$ and the sTArS INdICate GenEric elementS in $\{0,1\}$ (nOT necEssaRIlY eqUAl). If wE defiNe $$\XI^b\EQuiv (*,\ldots,*,\aunderbracE[D]{1

It is well known that suc h conditio n imp lie s t he str ongFKG property,s ee f or example Theorem $2. 19$,p. $25$ in [@Gr immett2 ] .B y de fi ni ng$$ \ ma thcal {R} (\xi,\o mega) \eq ui v { \mathcal{ W}^ {\mathrm{ GR C}} _{V,\m at hrm { free }} (\x i \vee \ omega) \ov e r \mat hcal{W}^{\mathrm{ GR C }}_{V,\mathrm{ free} }( \ xi ) }, $$ one can s ee th a t the l a tt i c e co n dition holdsif $$\begin { ali gned}\l abe l {min} \ ma t hca l{R}(\omega ^{(1 )},\omega ^{(2)} ) \g e qslant \ mat hca l{R} ( \o me ga^ {( 1 )}\ w ed ge\ o meg a^{(2)}, \o me ga^{( 2)}) . \e nd{ alig ned}$ $ For a fixed co nfig u rat ion $ \omeg a$,we chos e an a rbitr ar y order for $\e ta(\ omega)$ a ndre pre se nt th e se ope n e dge s as $( e_1,\ld o ts, e_ { \ v er t \eta(\omega)\ver t} ) $ .So for a ny con f ig ur a tion $\x i\ in\ {0,1 \ } ^{E}$ weh av e that $ $\math c al {R }(\xi,\ om ega)=\ pr od_ {k= 1}^{\ v ert\ eta(\o mega)\ve rt} \mathcal{R}( \ xi\vee\omega^ { (e _ 1 )} \ vee\ cdo ts \ve e \o mega ^ {( e_{ k -1})} ,\ome ga ^ {( e _{k})}),$$ where $(\ omega^ {(e)} )_{e'}\equiv\delta_{e, e ' } $. There fore it is enough to p rovefor config u rations$\xi$ , $\omeg a^{(1)}$a n d $\omeg a^{ (2) }$suc h th at $\xi$ hasa t lea st two ze rocoordin ate s o r a t m os t one zer o and $\ om eg a^ {( 1)} \equi v \xi\vee \o meg a^ {(b )}$ a n d $\om ega^{ (2)} \e qu i v \ xi\vee\ o me g a ^{(b ') }$ . Le t u sbegin ass u min g that$\xi$ has at leas ttw o zerocoordinates a nd $$\xi\equ iv (* ,\ldot s , *,\aunde rbrace[D]{0}_{ b-\mathr m {th}}, *,\l dots ,*,\aunde rbr ace[D] {0} _ {b'\ma thrm{- th}}, *, \ld o t s,*), $ $ w her e$b,b'\in E \ c up\part ia l E$, $b\n ot=b'$ and the sta r s i ndicate gener icelem e n ts in $\ { 0,1 \} $ (n o t necessarily eq ual). If w ed ef ine $$\xi^ b \eq ui v (*,\l dots,*, \aund e rbrace[ D]{1

It_is well_known that such condition_implies the_strong_FKG property,_see_for example Theorem_$2.19$, p. $25$_in [@Grimmett2]. By defining_$$\mathcal{R}(\xi,\omega) __\equiv { \mathcal{W}^{\mathrm{ GRC}}_{V,\mathrm{_free}} _ ___ (\xi\vee \omega) _ _ _\over __\mathcal{W}^{\mathrm{ GRC}}_{V,\mathrm{ free}}(\xi) _ },$$ one can see that the_lattice condition holds if $$\begin{aligned} \label{min} _ \mathcal{R}(\omega^{(1)},\omega^{(2)}) ___ \geqslant _ \mathcal{R}(\omega^{(1)}\wedge\omega^{(2)},\omega^{(2)}). _\end{aligned}$$ For a fixed configuration $\omega$,_we chose an arbitrary order for $\eta(\omega)$_and represent these open edges as_$(e_1,\ldots,e_{\vert \eta(\omega)\vert})$. So for any_configuration $\xi\in\{0,1\}^{E}$_we have that $$\mathcal{R}(\xi,\omega)=\prod_{k=1}^{\vert\eta(\omega)\vert} _ \mathcal{R}(\xi\vee\omega^{(e_1)}\vee\cdots_ _ \vee_ \omega^{(e_{k-1})},\omega^{(e_{k})}),$$ where_$ (\omega^{(e)})_{e'}\equiv \delta_{e,e'}$._Therefore it is enough to prove_for_configurations $\xi$, $\omega^{(1)}$_and_$\omega^{(2)}$_such that_$\xi$ has at_least_two zero_coordinates_or at most one zero and_$\omega^{(1)}\equiv_\xi\vee\omega^{(b)}$ and $\omega^{(2)}\equiv \xi\vee\omega^{(b')}$. Let us begin_assuming that $\xi$ has_at_least two zero coordinates_and $$\xi\equiv (*,\ldots,*,\aunderbrace[D]{0}_{ b-\mathrm{th}}, _ *,\ldots,*,\aunderbrace[D]{0}_{b'\mathrm{-th}},*,\ldots,*),$$ where $b,b'\in E\cup_\partial E_$, $b\not=b'$_and the stars indicate generic elements in $\{0,1\}$ (not necessarily equal)._If we define $$\xi^b\equiv (*,\ldots,*,\aunderbrace[D]{1

@16_Extending_Schoelkopf]. A Ramsey type measurement is performed. Here, two $\pi/2$ pulses are applied on the transmon (initially prepared in its ground state), separated by a period of dispersive interaction with the storage cavity (equivalent to a Controlled-Phase gate). These operations take the transmon to either its ground or excited state, depending on the photon-number parity in the storage cavity. The state of the transmon is then read-out projectively using the other cavity. On the other hand, if the qubit is encoded in states of different parity, as is true of our qubits ($\ket{C_{\alpha}^\pm}=\mathcal{N}_{\alpha}^\pm (\ket{\alpha}\pm\ket{-\alpha})$), then error-correction turns out to be more challenging. However, if there is protection against amplitude decay of the coherent states, and the qubits are encoded in cat states, then one needs to worry only about bit-flip errors, which can be corrected with joint parity measurements on several cavities [@15_Confining_Devoret; @17_Degeneracy_Mirrahimi; @18_Coherent_Devoret]. Joint parity measurements would be more hardware-resource consuming than parity measurements on single cavities, but a few experiments have demonstrated their feasibility, e.g. the one in ref. [@16_Schrodinger_Schoelkopf]. Similar challenges rise in applying error correction when encoding the qubits in the different parity Fock states, $\ket{0}$ and $\ket{1}$ [@16_New_Girvin; @18_Performance_Jiang]. The instantaneous eigenstates of our Hamiltonian $H(t)=-K {a^{\dagger}}^2 a^2 + (\mathcal{E}_p(t) {a^{\dagger}}^2 +\mathcal{E}_p^*(t) a^2)$ are the coherent states $\ket{\alpha}$ and $\ket{-\alpha}$, where $\alpha=\sqrt{\mathcal{E}_p/K}$. It is interesting to note that a general Hamiltonian of the form $H(t)=-K {a^{\dagger}}^n a^n + (\mathcal{E}_p(t) {a^{\dagger}}^n +\mathcal{E}_p^*(t) a^n)$ would have $`n\textrm'$ different coherent states as their eigenstates, e.g. for $n=4$, the

@16_Extending_Schoelkopf ]. A Ramsey type measurement is performed. Here, two $ \pi/2 $ pulses are applied on the transmon (initially prepare in its earth state), separated by a period of diffusing interaction with the storage cavity (equivalent to a Controlled - Phase gate). These operation take the transmon to either its ground or aroused state, depending on the photon - issue parity in the storage pit. The state of the transmon is then read - out projectively using the other cavity. On the other hand, if the qubit is encode in states of different parity, as is dependable of our qubits ($ \ket{C_{\alpha}^\pm}=\mathcal{N}_{\alpha}^\pm (\ket{\alpha}\pm\ket{-\alpha})$), then error - correction turn out to be more ambitious. However, if there is protection against amplitude decay of the coherent states, and the qubits are encoded in cat states, then one need to worry only about bit - flip errors, which can be corrected with joint parity measurements on several cavities [ @15_Confining_Devoret; @17_Degeneracy_Mirrahimi; @18_Coherent_Devoret ]. Joint parity measurements would be more hardware - resource consuming than parity measurements on single cavities, but a few experiments have show their feasibility, e.g.   the one in ref.   [ @16_Schrodinger_Schoelkopf ]. exchangeable challenges rise in applying error discipline when encoding the qubits in the unlike parity Fock states, $ \ket{0}$ and $ \ket{1}$ [ @16_New_Girvin; @18_Performance_Jiang ]. The instantaneous eigenstates of our Hamiltonian $ H(t)=-K { a^{\dagger}}^2 a^2 + (\mathcal{E}_p(t) { a^{\dagger}}^2 + \mathcal{E}_p^*(t) a^2)$ are the coherent states $ \ket{\alpha}$ and $ \ket{-\alpha}$, where $ \alpha=\sqrt{\mathcal{E}_p / K}$. It is interesting to notice that a general Hamiltonian of the form $ H(t)=-K { a^{\dagger}}^n a^n + (\mathcal{E}_p(t) { a^{\dagger}}^n + \mathcal{E}_p^*(t) a^n)$ would have $ 'n\textrm'$ different coherent states as their eigenstates, e.g.   for $ n=4 $, the

@16_Exhending_Schoelkopf]. A Ramsty type measuremeur is pxrformes. Here, tdo $\pi/2$ pulses are applied on vhe rransnon (initially prepared in its ggound stare), stparated by a permkd of dlfperalve iutxraction with tme storage wavity (equivalang co a Controlled-Phase gate). These operwtions yane the transmog to qithsg lts ground or excited state, depehding oi the photon-numner parity in the storage favihy. The state of thf transmon us trwn read-out pfojectivelj using the other cavity. On the other hand, kf thz qubit is guxodfg in states of dpfferent paribj, as is true og our qubits ($\kct{C_{\al'ha}^\pn}=\mathcal{N}_{\alpha}^\pm (\ket{\anpha}\pm\ket{-\alpha})$), theg error-cosrzction turns out to bw nore whalnengkbg. Fowtvec, ir therf ia protectikn against qmplitude decay of uhe bpherent statss, and tre qubits are encoded in cat states, thet ohe needs to worry only qbout bit-flip errors, ahich can be corrected with joint parity measurements on saveram carlbies [@15_Xojfining_Devoret; @17_Degeneracy_Mirrahimi; @18_Coherent_Dedkrtt]. Moint parity meafurements wpupd fe more hardwxre-resourde consuming than oarity ieasueements og simgle cavities, but a few expwriments havv denonstrated their fzasibility, e.y. the ome in ref. [@16_Schrodinger_Schoelko'f]. Simjlar challejges rise kn applying errof cprsection when encoding the zubits in the differdnt karity Sock stated, $\ket{0}$ and $\ket{1}$ [@16_New_Girvin; @18_Oerfotmance_Biang]. The ijstantaneous eigenstates of our Hamiltonian $H(j)=-K {d^{\dacger}}^2 a^2 + (\iathcsl{E}_p(t) {a^{\dagger}}^2 +\mathcal{E}_p^*(t) a^2)$ cre the eoherevt states $\iet{\alphe}$ and $\ket{-\alpra}$, where $\alphd=\dqrt{\mathcal{E}_'/K}$. It is ynteeestung to vute that a genrral Hamiltonian of the form $H(t)=-K {a^{\dagner}}^n x^h + (\mathcal{E}_p(t) {a^{\bcgter}}^n +\mathcal{E}_p^*(t) a^n)$ wotlf iave $`g\dextrm'$ diffesent cofrrent states as bhekr eogenstates, e.g. for $n=4$, tve

@16_Extending_Schoelkopf]. A Ramsey type measurement is performed. $\pi/2$ are applied the transmon (initially separated a period of interaction with the cavity (equivalent to a Controlled-Phase gate). operations take the transmon to either its ground or excited state, depending on photon-number parity in the storage cavity. The state of the transmon is then projectively the cavity. the other hand, if the qubit is encoded in states of different parity, as is true our qubits ($\ket{C_{\alpha}^\pm}=\mathcal{N}_{\alpha}^\pm (\ket{\alpha}\pm\ket{-\alpha})$), then error-correction turns out be more challenging. However, there is protection against amplitude of coherent states, the are in cat states, one needs to worry only about bit-flip errors, which can be corrected with joint parity measurements on cavities [@15_Confining_Devoret; Joint parity would more consuming than parity single cavities, but a few experiments feasibility, e.g. the one in ref. [@16_Schrodinger_Schoelkopf]. Similar rise in error correction when encoding the qubits the different parity Fock states, $\ket{0}$ and $\ket{1}$ @18_Performance_Jiang]. The instantaneous eigenstates of our Hamiltonian $H(t)=-K {a^{\dagger}}^2 a^2 + (\mathcal{E}_p(t) {a^{\dagger}}^2 +\mathcal{E}_p^*(t) a^2)$ coherent states $\ket{\alpha}$ and where $\alpha=\sqrt{\mathcal{E}_p/K}$. It interesting note a Hamiltonian of form $H(t)=-K {a^{\dagger}}^n a^n + (\mathcal{E}_p(t) {a^{\dagger}}^n +\mathcal{E}_p^*(t) a^n)$ would have different coherent states as their eigenstates, e.g. for $n=4$, the

@16_Extending_Schoelkopf]. A RamseY type measuRemenT is PerFoRmed. here, Two $\pi/2$ pulses are APpliEd on the transmon (initialLy prePaREd in ITs GrounD state), sEPaRATed By A pEriOd OF dIsperSivE interaCtion with tHe sToRage cavity (eqUIvAlent to a CoNtrOlled-Phase gaTe). THese opErAtiONs takE thE tranSmon to EIther iTs ground oR eXCited sTAte, depeNDInG on tHe photon-number parITy IN the storage cavIty. The StATe OF The TraNsmon is theN rEad-ouT ProjectIVeLY USinG The other cavitY. On the other HAnd, If the qUbIt iS EncodeD in stAtES of Different paRity, As is true oF our quBIts ($\ket{C_{\ALpha}^\pm}=\mAthcal{n}_{\alPha}^\Pm (\keT{\AlPhA}\pm\KeT{-\AlpHA})$), tHen ERroR-correctIoN tUrns oUt to BE MORe chAllEngiNg. HowEver, if there is ProTectIOn aGainsT amplItudE dEcay oF the coHerenT sTates, and the qubiTs arE encoded iN caT sTatEs, Then oNE needs To wOrrY only abOut bit-fLIp eRrORS, WhIch can be corrected wItH JOiNt parity MeasurEMeNtS On severaL cAviTies [@15_cONfiniNg_DeVOrEt; @17_DegeneRacy_MiRRaHiMi; @18_CoherEnT_DevorEt]. joiNt pArity MEasuRementS would be More hARdware-resource COnsuming than pARiTY MeASureMenTs on single cAvitIEs, buT a feW ExPerIMents Have dEmONsTRated their feasibiliTy, E.g. the oNe in rEf. [@16_Schrodinger_schoelkopf]. sIMIlar chalLengES rISe in applying erRor coRrection whEN encodinG the qUbits in tHe differeNT Parity FoCk sTatEs, $\kEt{0}$ aND $\KeT{1}$ [@16_New_Girvin; @18_PerFORmanCe_jiang]. ThE inStantanEouS eiGenStaTeS of our HamIltonian $h(t)=-k {a^{\DaGgEr}}^2 a^2 + (\MathcAL{E}_p(t) {a^{\dagGeR}}^2 +\maThCal{e}_p^*(t) a^2)$ aRE the coHerenT staTeS $\kET{\alPha}$ and $\kET{-\aLPHa}$, whErE $\aLpha=\SqrT{\mAthcaL{E}_p/K}$. iT is InteresTing to notE thAT a geNeRaL HamiltOnian of the forM $H(T)=-K {a^{\dagger}}^n A^n + (\MatHcal{E}_p(T) {A^{\Dagger}}^n +\mAthcal{E}_p^*(t) a^n)$ would have $`n\teXTrm'$ diffEreNt cohErenT states as TheIr eigeNstATes, e.g. fOr $n=4$, the

@16_Extending_Schoelkopf] . A Ramsey type me asu re ment isperformed. Her e , tw o $\pi/2$ pulses are a pplie do n th e t ransm on (ini t ia l l y p re pa red i n i ts gr oun d state ), separat edby a period of di spersive i nte raction with th e stor ag e c a vity(eq uival ent to a Cont rolled-Ph as e gate) . Theseo p er atio ns take the trans m on to either itsground o r e x c ite d s tate, depe nd ing o n the ph o to n - n umb e r parity in t he storagec avi ty. Th esta t e of t he tr an s mon is then re ad-o ut projec tively using t h e other cavit y. On the ot he r h an d , i f t heq ubi t is enc od ed in s tate s o f dif fer entparit y, as is true of our qub its ( $\ket {C_{ \a lpha} ^\pm}= \math ca l{N}_{\alpha}^\ pm ( \ket{\alp ha} \p m\k et {-\al p ha})$) , t hen error- correct i ontu r n s o ut to be more chal le n g in g. Howev er, if th er e is prot ec tio n ag a i nst a mpli t ud e decayof the co he rent st at es, an dthe qu bitsa re e ncoded in catstate s , then one nee d s to worry on l ya b ou t bit -fl ip errors,whic h can bec or rec t ed wi th jo in t p a rity measurements o nsevera l cav ities [@15_Co nfining_De v o r et; @17_ Dege n er a cy_Mirrahimi;@18_C oherent_De v oret]. J ointparity m easuremen t s would b e m ore ha rdw a r e- resource cons u m ingth an pari tymeasure men tsonsin gl e cavitie s, but a f ew e xp eri ments have dem on str at edtheir feasib ility , e. g.  t h e o ne in r e f. [ @16_ Sc hr odin ger _S choel kopf ] . S imilarchallenge s r i se i nap plyingerror correct io n when enc od ing the q u b its in t he different parity Foc k states , $ \ket{ 0}$and $\ket {1} $ [@16 _Ne w _Girvi n; @18 _Perf or man c e _Jian g ] . Th einstantane o u s e igens ta tesof ourHamiltonian $H(t)= - K { a^{\dagger}}^ 2 a ^2 + ( \m ath c al { E}_ p( t ) { a ^ {\dagger}}^2 +\ mathcal{E} _p ^ *( t) a^2)$ a r e t he cohere nt stat es $\ k et{\alp ha}$ and$\ket{-\a lp ha}$ , whe re $\alpha =\sqrt{\ mathcal{E } _p/K} $ .It is in terest in g t o not e that a g enera l Hami lt onianof th eform $H( t)=-K {a^{\dagger}}^n a ^n + ( \math cal {E}_p(t){a^ { \da gger}}^n+\ma thcal{E}_p ^*( t)a^n)$ wo u ld ha ve $ ` n\ tex t rm'$diff e rent cohe r en t s t a te s as theire i g ens tates , e . g. for $n= 4$, the

@16_Extending_Schoelkopf]._A Ramsey_type measurement is performed._Here, two_$\pi/2$_pulses are_applied_on the transmon_(initially prepared in_its ground state), separated_by a period_of_dispersive interaction with the storage cavity (equivalent to a Controlled-Phase gate). These operations take_the_transmon to_either_its_ground or excited state, depending_on the photon-number parity in_the storage_cavity. The state of the transmon is then_read-out_projectively using the_other cavity. On the other hand, if the qubit is_encoded in states of different parity,_as is true_of_our_qubits ($\ket{C_{\alpha}^\pm}=\mathcal{N}_{\alpha}^\pm (\ket{\alpha}\pm\ket{-\alpha})$), then_error-correction turns out to be more_challenging. However, if there is protection_against amplitude decay of the coherent states,_and the qubits are encoded in_cat states, then one needs_to worry_only about bit-flip errors, which_can be corrected_with joint_parity measurements on_several cavities [@15_Confining_Devoret; @17_Degeneracy_Mirrahimi; @18_Coherent_Devoret]. Joint_parity measurements would_be more hardware-resource consuming than parity_measurements_on single cavities,_but_a_few experiments_have demonstrated their_feasibility,_e.g. the one_in_ref. [@16_Schrodinger_Schoelkopf]. Similar challenges rise in applying_error_correction when encoding the qubits in the_different parity Fock states,_$\ket{0}$_and $\ket{1}$ [@16_New_Girvin; @18_Performance_Jiang]. The_instantaneous eigenstates of our Hamiltonian_$H(t)=-K {a^{\dagger}}^2 a^2 + (\mathcal{E}_p(t) {a^{\dagger}}^2_+\mathcal{E}_p^*(t) a^2)$_are the_coherent states $\ket{\alpha}$ and $\ket{-\alpha}$, where $\alpha=\sqrt{\mathcal{E}_p/K}$. It is interesting to_note that a general Hamiltonian of_the form $H(t)=-K {a^{\dagger}}^n_a^n +_(\mathcal{E}_p(t)_{a^{\dagger}}^n +\mathcal{E}_p^*(t) a^n)$_would_have $`n\textrm'$_different coherent states as their eigenstates, e.g. for_$n=4$, the

}(s)),v({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s)))dW(s),\\ X(t)= & x,\quad s\in\lbrack t,T]. \end{array}$$ Set $$(Y(s),Z(s))=(u({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s)),v({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s))),\quad t\leq s\leq T.$$ We have$$\begin{array} [c]{rl}dX^{\gamma_{t},x}(s)= & b(W_{s}^{\gamma_{t}},X^{\gamma_{t},x}(s),Y(s),Z(s))ds+\sigma(W_{s}^{\gamma_{t}},X^{\gamma_{t},x}(s),Y(s),Z(s))dW(s),\\ X(t)= & x,\quad s\in\lbrack t,T]. \end{array}$$ Note that $u$ solves equation (4.1). Applying Functional $It\hat{o}$ formula to $Y(s)=u({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s))$, we get$$\begin{array} [c]{rl}dY(s)= & h(W_{s}^{\gamma_{t}},X^{\gamma_{t},x}(s),Y(s),Z(s))dr+Z(s)dW(s),\\ Y(T)= & g(W_{T}^{\gamma_{t}},X^{\gamma_{t},x}(T))\quad s\in\lbrack t,T]. \end{array}$$ Thus, by the uniqueness and existence theorem of the FBSDE (Theorem 2.2), we have the result.$\Box$ Now we prove the converse to the about result. Under Assumptions $2.2$, $3.2$ and $3.3$, the function $u(\gamma _{t},x)=Y^{\gamma_{t},x}(t)$ is the unique $\mathbb{C}^{1,2,2}(\Lambda \times\mathbb{R}^{m})$-solution of the path-dependent PDE $(4.1)$. **[Proof.]{}** We only study the one dimensional case. By Corollary $3

} (s)),v({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s)))dW(s),\\ X(t)= & x,\quad s\in\lbrack t, T ]. \end{array}$$ Set $ $ (Y(s),Z(s))=(u({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s)),v({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s))),\quad t\leq s\leq T.$$ We have$$\begin{array } [ c]{rl}dX^{\gamma_{t},x}(s)= & b(W_{s}^{\gamma_{t}},X^{\gamma_{t},x}(s),Y(s),Z(s))ds+\sigma(W_{s}^{\gamma_{t}},X^{\gamma_{t},x}(s),Y(s),Z(s))dW(s),\\ X(t)= & x,\quad s\in\lbrack t, T ]. \end{array}$$ Note that $ u$ solves equation (4.1). Applying Functional $ It\hat{o}$ rule to $ Y(s)=u({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s))$, we get$$\begin{array } [ c]{rl}dY(s)= & h(W_{s}^{\gamma_{t}},X^{\gamma_{t},x}(s),Y(s),Z(s))dr+Z(s)dW(s),\\ Y(T)= & g(W_{T}^{\gamma_{t}},X^{\gamma_{t},x}(T))\quad s\in\lbrack t, T ]. \end{array}$$ therefore, by the uniqueness and existence theorem of the FBSDE (Theorem 2.2), we get the result.$\Box$ immediately we prove the converse to the about result. Under assumption $ 2.2 $, $ 3.2 $ and $ 3.3 $, the routine $ u(\gamma _ { t},x)=Y^{\gamma_{t},x}(t)$ is the unique $ \mathbb{C}^{1,2,2}(\Lambda \times\mathbb{R}^{m})$-solution of the way - pendent PDE $ (4.1)$. * * [ Proof. ] { } * * We only learn the one dimensional case. By Corollary $ 3

}(s)),v({W_{d}^{\gamma_{t}}},X^{\gamma_{t},x}(s)))dW(s),\\ X(t)= & x,\duad s\in\lbrack j,T]. \wnd{arrey}$$ Set $$(G(s),Z(s))=(u({W_{s}^{\gxmma_{t}}},X^{\gamma_{t},x}(s)),v({W_{s}^{\gamma_{t}}},X^{\gamme_{t},x}(s))),\wuad u\jeq s\leq T.$$ We have$$\cegin{arraj} [c]{rl}dX^{\gamna_{t},x}(w)= & b(W_{s}^{\gamma_{v}},S^{\gamma_{t},w}(f),Y(s),Z(a))fs+\siyme(W_{s}^{\gamma_{t}},X^{\gamma_{j},x}(s),Y(s),Z(s))dW(s),\\ X(t)= & x,\quad s\in\lbrdcy c,T]. \end{array}$$ Note that $u$ solves equatijn (4.1). Appkylng Functional $It\hse{o}$ fkgmmla to $Y(s)=u({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s))$, we gst$$\begin{erray} [c]{rl}dY(s)= & h(W_{x}^{\gamma_{t}},X^{\gamma_{t},x}(s),Y(s),Z(s))dr+Z(s)dW(d),\\ Y(T)= & g(W_{T}^{\gamma_{t}},X^{\gamma_{t},x}(H))\quad s\in\lbtzck r,T]. \end{array}$$ Tfus, by the uniqueness znd existence theorem of the FBRDE (Tkeorem 2.2), we yace hve result.$\Boe$ Now wv prove the converse do the sbout result. Uncer Aswumptions $2.2$, $3.2$ and $3.3$, the hunction $u(\gamma _{t},x)=Y^{\gaima_{t},x}(t)$ is tke unique $\mathbb{C}^{1,2,2}(\Lambea \rimes\kathtb{R}^{m})$-rilugioh pf the pwth-vependent PSE $(4.1)$. **[Proof.]{}** We only study the one dynensional cass. By Cjrjllary $3

}(s)),v({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s)))dW(s),\\ X(t)= & x,\quad s\in\lbrack t,T]. \end{array}$$ t\leq T.$$ We [c]{rl}dX^{\gamma_{t},x}(s)= & b(W_{s}^{\gamma_{t}},X^{\gamma_{t},x}(s),Y(s),Z(s))ds+\sigma(W_{s}^{\gamma_{t}},X^{\gamma_{t},x}(s),Y(s),Z(s))dW(s),\\ \end{array}$$ that $u$ solves (4.1). Applying Functional formula to $Y(s)=u({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s))$, we get$$\begin{array} [c]{rl}dY(s)= h(W_{s}^{\gamma_{t}},X^{\gamma_{t},x}(s),Y(s),Z(s))dr+Z(s)dW(s),\\ Y(T)= & g(W_{T}^{\gamma_{t}},X^{\gamma_{t},x}(T))\quad s\in\lbrack t,T]. \end{array}$$ Thus, by the uniqueness and existence of the FBSDE (Theorem 2.2), we have the result.$\Box$ Now we prove the to about Under $2.2$, $3.2$ and $3.3$, the function $u(\gamma _{t},x)=Y^{\gamma_{t},x}(t)$ is the unique $\mathbb{C}^{1,2,2}(\Lambda \times\mathbb{R}^{m})$-solution of the path-dependent $(4.1)$. **[Proof.]{}** We only study the one dimensional By Corollary $3

}(s)),v({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s)))dW(s),\\ X(t)= & x,\Quad s\in\lbrAck t,T]. \End{ArrAy}$$ set $$(Y(S),Z(s))=(u({w_{s}^{\gamma_{t}}},X^{\gamma_{T},X}(s)),v({W_{S}^{\gamma_{t}}},X^{\gamma_{t},x}(s))),\quad t\lEq s\leQ T.$$ wE havE$$\BeGin{arRay} [c]{rl}dx^{\GaMMA_{t},x}(S)= & b(w_{s}^{\GamMa_{T}},x^{\gAmma_{t},X}(s),Y(S),Z(s))ds+\siGma(W_{s}^{\gamma_{T}},X^{\gAmMa_{t},x}(s),Y(s),Z(s))dW(s),\\ x(T)= & x,\Quad s\in\lbrAck T,T]. \end{array}$$ NoTe tHat $u$ soLvEs eQUatioN (4.1). ApPlyinG FunctIOnal $It\Hat{o}$ formuLa TO $Y(s)=u({W_{s}^{\GAmma_{t}}},X^{\gAMMa_{T},x}(s))$, wE get$$\begin{array} [c]{rl}Dy(s)= & H(w_{s}^{\gamma_{t}},X^{\gamma_{T},x}(s),Y(s),Z(S))dR+z(s)Dw(S),\\ Y(T)= & G(W_{T}^{\Gamma_{t}},X^{\gamMa_{T},x}(T))\quAD s\in\lbrACk T,t]. \ENd{aRRay}$$ Thus, by the uNiqueness anD ExiStence ThEorEM of the fBSDE (thEOreM 2.2), we have the rEsulT.$\Box$ Now we Prove tHE converSE to the aBout reSulT. UnDer ASSuMpTioNs $2.2$, $3.2$ ANd $3.3$, tHE fUncTIon $U(\gamma _{t},x)=y^{\gAmMa_{t},x}(t)$ Is thE UNIQue $\mAthBb{C}^{1,2,2}(\LAmbda \Times\mathbb{R}^{m})$-SolUtioN Of tHe patH-depeNdenT PdE $(4.1)$. **[ProOf.]{}** We onLy stuDy The one dimensionAl caSe. By CorolLarY $3

}(s)),v({W_{s}^{\gamma_{t} }},X^{\gam ma_{t },x }(s )) )dW( s),\ \ X(t)= & x,\q u ad s \in\lbrack t,T]. \end{ array }$ $ Set $$ (Y(s) ,Z(s))= ( u( { W _{s }^ {\ gam ma _ {t }}},X ^{\ gamma_{ t},x}(s)), v({ W_ {s}^{\gamma_ { t} }},X^{\gam ma_ {t},x}(s))), \qu ad t\l eq s\ l eq T. $$We ha ve$$\b e gin{ar ray} [c]{ rl } dX^{\g a mma_{t} , x }( s)=& b(W_{s}^{\gamma _ {t } },X^{\gamma_{t },x}(s ), Y (s ) , Z(s ))d s+\sigma(W _{ s}^{\ g amma_{t } }, X ^ { \ga m ma_{t},x}(s), Y(s),Z(s))d W (s) ,\\ X( t) = & x,\qua d s\i n\ l bra ck t,T]. \e nd{a rray}$$ N ote th a t $u$ s o lves eq uation (4 .1) . Ap p ly in g F un c tio n al $I t \ha t{o}$ fo rm ul a to$Y(s ) = u ( {W_{ s}^ {\ga mma_{ t}}},X^{\gamm a_{ t},x } (s) )$, w e get $$\b eg in{ar ray} [ c]{rl }d Y(s)= & h(W_{s} ^{\g amma_{t}} ,X^ {\ gam ma _{t}, x }(s),Y (s) ,Z( s))dr+Z (s)dW(s ) ,\\ Y ( T ) =& g(W_{T}^{\gamma_ {t } } ,X ^{\gamma _{t},x } (T )) \ quad s\i n\ lbr ackt , T]. \ end{ a rr ay}$$ Th us, by th euniquen es s andex ist enc e the o remof the FBSDE ( Theor e m 2.2), we hav e the result.$ \ Bo x $ N ow w e p rove the co nver s e to the ab out resul t. U nd e rA ssumptions $2.2$, $ 3. 2$ and $3.3 $, the functi on $u(\gam m a _{t},x)= Y^{\ g am m a_{t},x}(t)$ i s the unique $\ m athbb{C} ^{1,2 ,2}(\Lam bda \time s \ mathbb{R }^{ m}) $-s olu t i on of the path- d e pend en t PDE $ (4. 1)$. * *[P roo f.] {}* *We only s tudy the o ne d im ens ional case. By C oro ll ary $3

}(s)),v({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s)))dW(s),\\ X(t)= &_x,\quad s\in\lbrack_t,T]. \end{array}$$ Set $$(Y(s),Z(s))=(u({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s)),v({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s))),\quad t\leq_s\leq T.$$_We_have$$\begin{array} [c]{rl}dX^{\gamma_{t},x}(s)= &_b(W_{s}^{\gamma_{t}},X^{\gamma_{t},x}(s),Y(s),Z(s))ds+\sigma(W_{s}^{\gamma_{t}},X^{\gamma_{t},x}(s),Y(s),Z(s))dW(s),\\ X(t)=_& x,\quad s\in\lbrack_t,T]. \end{array}$$ Note that_$u$ solves equation (4.1)._Applying Functional $It\hat{o}$_formula_to $Y(s)=u({W_{s}^{\gamma_{t}}},X^{\gamma_{t},x}(s))$, we get$$\begin{array} [c]{rl}dY(s)= & h(W_{s}^{\gamma_{t}},X^{\gamma_{t},x}(s),Y(s),Z(s))dr+Z(s)dW(s),\\ Y(T)= & g(W_{T}^{\gamma_{t}},X^{\gamma_{t},x}(T))\quad s\in\lbrack t,T]. \end{array}$$ Thus, by the uniqueness_and_existence theorem_of_the_FBSDE (Theorem 2.2), we have_the result.$\Box$ Now we prove the_converse to_the about result. Under Assumptions $2.2$, $3.2$ and $3.3$,_the_function $u(\gamma _{t},x)=Y^{\gamma_{t},x}(t)$ is_the unique $\mathbb{C}^{1,2,2}(\Lambda \times\mathbb{R}^{m})$-solution of the path-dependent PDE $(4.1)$. **[Proof.]{}** We_only study the one dimensional case._By Corollary $3

of what is known of the behaviors of tourism operators in the destination (see for example [@Tallinucci:2006] or [@Pechlaner:2003]). However, the identification of the specific trends exhibited by each type of company only became clearer through the superedges integration of the structural and dynamical features. Concluding Remarks ================== Tourism destination networks are amongst the most complex real-world systems, involving intrincated structure and non-linear dynamics. Because of the economical importance of such systems, it becomes increasingly important to devise effective means for describing, characterizing and modeling these systems so that their structure and dynamics are better understood, allowing predictions, identification of possible improvements, and simulations aimed at evaluating the effects of varying scenarios and conditions. At the same time, complex networks research is now a mature area catering for all such requirements, from the characterization of the structure of the interaction between tourism companies to the relationship with the respective dynamics of interactions and information exchange. The current article has brought these two important areas together with respect to the comprehensive analysis of the Elba tourism destination network, a real-world structure which has been recently obtained through systematic and careful investigation including field data collection. The main contributions of our work are listed and discussed as follows: *Complete Real-World Example of the Superedges Approach:* This work represents the very first comprehensive application of the superedges approach to a real-world network. As such, special care has been invested in order to making the respective concepts and interpretations clear from the context of the specific application, namely tourism destination structures. In particular, each of the choices which have to be made regarding the superedges methodology — including input/output, structural measurements, dynamics and dynamical measurements — have been motivated and justified with respect to the tourism application. The methodology was first illustrated with respect to a hypothetical simple network of companies, and then applied systematically to the real-world Elba tourism destination structure. As such, this first complete example can be used as a application reference guide for researchers intending to apply the superedges approach to other specific problems. *Comprehensive Structural Characterization:* This work has presented one of the very first comprehensive analysis of the topological characteristics of a tourism destination network, including traditional measurements, modularity as well as the number of distinct paths between pairs of nodes. A comparison, and the substantial similarity, of the outcomes with previous knowledge of the relationships among the tourism companies located at Elba stakeholders [@Pechlaner:

of what is known of the behaviors of tourism operators in the destination (visualize for case   [ @Tallinucci:2006 ] or   [ @Pechlaner:2003 ]). However, the identification of the specific trends exhibited by each character of company only became clear through the superedges integration of the structural and dynamical features. Concluding Remarks = = = = = = = = = = = = = = = = = = Tourism destination network are amongst the most complex real - earth system, involving intrincated structure and non - linear dynamics. Because of the economical importance of such organization, it becomes increasingly important to devise effective mean for describing, characterizing and model these systems so that their social organization and dynamics are well understood, allowing predictions, designation of possible improvements, and simulations aimed at evaluating the effects of varying scenarios and conditions. At the same time, complex networks research is now a mature area catering for all such requirement, from the characterization of the structure of the interaction between tourism ship's company to the kinship with the respective moral force of interactions and information exchange. The current article has brought these two important areas together with respect to the comprehensive psychoanalysis of the Elba tourism destination network, a real - world structure which has been recently obtained through systematic and careful investigation including field data collection. The chief contributions of our employment are listed and discussed as follows: * Complete Real - World Example of the Superedges Approach :* This work represents the very inaugural comprehensive application of the superedges approach to a real - populace network. As such, special care has been invested in club to making the respective concept and interpretations clear from the context of the specific application, namely tourism destination structures. In especial, each of the choices which suffer to be made involve the superedges methodology — including input / output signal, structural measurement, dynamics and dynamical measurements — have been motivate and justified with respect to the tourism application. The methodology was first exemplify with respect to a hypothetical simple net of companies, and then applied systematically to the real - world Elba tourism destination structure. As such, this first arrant example can be used as a application reference guide for researchers intend to lend oneself the superedges access to other specific problems. * Comprehensive Structural Characterization :* This work has presented one of the very first comprehensive analysis of the topological feature of a tourism destination network, including traditional measurement, modularity as well as the numeral of distinct paths between pairs of nodes. A comparison, and the significant similarity, of the outcomes with previous knowledge of the relationships among the tourism companies located at Elba stakeholder   [ @Pechlaner:

of what is known of the bemaviors of tourism operetors ih the dertination (see for example [@Talpibucci:2006] or [@Pechlaner:2003]). However, tfe identivication of uhe specific trenva exhiblced bg eack vype of company only becake clearer thrmueh the superedges integration of the ftructutap and dynamicaj feseurea. Bokcluding Remarks ================== Tourism destinatjon netxorks are amongxt the most complex real-wogld dystems, involving lntrincated strtxture and nov-linear dynamics. Becauae of the economical importance of sbch systems, ir bfwomes increesinglj important to devise effectove means for cesrribung, characterizing anv modeling these sysjems so thdt their structure qne dyndmicv ard begted nndsrstoof, amlowing prsdictions, ieentification of poxsyvle improvemehts, anq fimulations aimed at evaluating the efftcts kf varying scenarios ane conditions. At the swme time, somplex networks research is now a mature area caderinj wor aju skch requirements, from the characterization of fht snructure of the ikteraction between tlutysm companies to thz rslationship with tje respgctive dynamics of onteractions and informatiob exchange. Tke xurrent article had brought tkese teo imlortant areas together cith rsspect to tje compregdnsive analysis uf nhe Alba tourism destination nqtwork, a ceal-wprld stfuctore whish has beej reccttly obtained throkgh sistemadic and cageful investigation including fmxld data collgcthon. The maiu contvibutions of ouw work are lisjed and dnscussdd as follkws: *Compnete Real-Wowld Example ox the Superedjes Approwch:* Rhis work rdoresents the vrry first comprehenwive application on the auperedges appriacy to a real-worlc ndtwjrn. Es susv, special case hxs cren ivvested in ifder to making the respewtivs concepts and intrriretationw clear srom the contrxt of the specifif appnicetion, mamgly tourism destination structudes. In pagtigular, each of the choices whieh have to be made regarding the superedjes methodology — includung input/output, stroctmral measurekents, dynamics and dynamical measueements — have beek motivated and justifjed widh redpect to the tourism application. The methodology was first illustrated wirh res'est to a hypktheyical snmpke netrork uf companies, and then applied systematically to the rean-wjrld Elba tourism destinatiom rtructure. As ruch, this first complete example can be used as a application rrference guide for reseavchers inrendihg to applb the superedges apptoach co othwr specufic problems. *Compfehrndive Strufturel Eharacterization:* This work has prwsentee lne of the very hirst compremeksive znalysys of the topoloficam charxcteristics og a tourism dextination network, inbluqing traditionqm measuremenjs, modooarity as well as the numter of distinct paths between pairs of nodes. A compzrison, ans the sunstantial similarovy, of the outcomes with irevijus knowuedge of the relationships among the tourism companies located at Elba stakeholders [@Iechlaner:

of what is known of the behaviors operators the destination for example [@Tallinucci:2006] of specific trends exhibited each type of only became clearer through the superedges of the structural and dynamical features. Concluding Remarks ================== Tourism destination networks are the most complex real-world systems, involving intrincated structure and non-linear dynamics. Because of economical of systems, becomes increasingly important to devise effective means for describing, characterizing and modeling these systems so that structure and dynamics are better understood, allowing predictions, of possible improvements, and aimed at evaluating the effects varying and conditions. the time, networks research is a mature area catering for all such requirements, from the characterization of the structure of the interaction tourism companies relationship with respective of and information exchange. article has brought these two important respect to the comprehensive analysis of the Elba destination network, real-world structure which has been recently through systematic and careful investigation including field data The main contributions of our work are listed and discussed as follows: *Complete Real-World Example Superedges Approach:* This work the very first application the approach a real-world As such, special care has been invested in order to making respective concepts and interpretations clear from the context of the namely destination structures. In each of the choices have be made regarding the — input/output, and measurements have been motivated and with respect to the tourism The methodology was first hypothetical simple network of companies, and then applied to the real-world Elba tourism destination structure. such, this first complete example can be used as a application reference for researchers apply the superedges approach to other specific problems. Structural Characterization:* This work presented one of the very first comprehensive analysis of topological of a destination network, including measurements, modularity as as the number paths between of A substantial similarity, of the outcomes with knowledge the relationships among the tourism located stakeholders [@Pechlaner:

of what is known of the behavioRs of tourisM operAtoRs iN tHe deStinAtion (see for exaMPle [@TAllinucci:2006] or [@Pechlaner:2003]). HoWever, ThE IdenTIfIcatiOn of the SPeCIFic TrEnDs eXhIBiTed by EacH type of Company onlY beCaMe clearer thrOUgH the supereDgeS integration Of tHe struCtUraL And dyNamIcal fEatureS. concluDing RemarKs ================== tOurism DEstinatION nEtwoRks are amongst the mOSt COmplex real-worlD systeMs, INvOLVinG inTrincated sTrUcturE And non-lINeAR DYnaMIcs. Because of tHe economicaL ImpOrtancE oF suCH systeMs, it bEcOMes IncreasinglY impOrtant to dEvise eFFective MEans for DescriBinG, chAracTErIzIng AnD ModELiNg tHEse Systems sO tHaT theiR strUCTURe anD dyNamiCs are Better understOod, AlloWIng PrediCtionS, ideNtIficaTion of PossiBlE improvements, anD simUlations aImeD aT evAlUatinG The effEctS of Varying ScenariOS anD cONDItIons. At the same time, cOmPLEx Networks ResearCH iS nOW a mature ArEa cAterING for aLl suCH rEquiremeNts, froM ThE cHaracteRiZation Of The StrUcturE Of thE interAction beTween TOurism companieS To the relationSHiP WItH The rEspEctive dynamIcs oF InteRactIOnS anD InforMatioN eXChANge. The current articlE hAs brouGht thEse two importaNt areas togETHEr with reSpecT To THe comprehensivE analYsis of the ELBa tourisM destInation nEtwork, a reAL-World strUctUre WhiCh hAS BeEn recently obtAINed tHrOugh sysTemAtic and CarEfuL inVesTiGation incLuding fiElD dAtA cOllEctioN. the main cOnTriBuTioNs of oUR work aRe lisTed aNd DiSCusSed as foLLoWS: *compLeTe real-worLd exampLe of THe SUperedgEs ApproacH:* ThIS worK rEpResents The very first cOmPrehensive ApPliCation OF The superEdges approach to a real-worLD networK. As Such, sPeciAl care has BeeN invesTed IN order To makiNg the ReSpeCTIve coNCEpTs aNd InterpretaTIOns Clear FrOm thE contexT of the specific applICatIon, namely tourIsm DestINAtIon STrUCtuReS. in pARTicular, each of thE choices whIcH HaVe to be made REgaRdIng the sUperedgEs metHOdology — Including Input/outpUt, StruCTUraL measuremeNts, dynamIcs and dynAMical MEaSuremEntS — have bEeN moTivatEd and jUStiFied wIth resPeCt to thE tourIsM applicaTion. The methodology was fiRst illUstraTed With respeCt tO A hyPotheticaL simPle network Of cOmpAnies, And THen apPlieD SySteMAticaLly tO The real-woRLd elbA TOuRism destinaTION stRuctuRe. AS Such, thIs fiRst complete examplE Can be used as a apPlicATIon RefERencE gUide for researcHerS iNTEnding to ApPly the superEdges appRoACh to oTher spEcific ProblemS. *cOmPRehensIve STruCtural ChaRacTeRIzation:* thIs WOrk has PresEnTed one Of the vERy fiRST comprehensive anAlysiS OF the tOPolOgicaL cHaracteRIstiCs of a touriSm destinatiOn netwOrk, iNcludIng tradItIonal mEasUrEments, moduLArity as weLl as tHe numbeR oF disTinCt pathS betWEEn paiRs of NoDes. a comparisON, AnD ThE sUBstAntiAl simIlAritY, of the outCOmes with PreVIous knoWlEdgE OF the reLAtIONships amonG thE tourISM companies LOcatED aT elba sTakehoLders [@PEchlaneR:

of what is known of the b ehaviors o f tou ris m o pe rato rs i n the destinat i on ( see for example [@Tall inucc i: 2 006] or  [@Pe chlaner : 20 0 3 ]). H ow eve r, th e ide nti ficatio n of the s pec if ic trends ex h ib ited by ea chtype of comp any onlybe cam e clea rer thro ugh th e super edges int eg r ationo f the s t r uc tura l and dynamical f e at u res. Concludi ng Rem ar k s= = === === ========== Touri s m desti n at i o n ne t works are amo ngst the mo s t c omplex r eal - worldsyste ms , in volving int rinc ated stru cturea nd non- l inear d ynamic s.Bec ause of t heec o nom i ca l i m por tance of s uc h sys tems , i t bec ome s in creas ingly importa ntto d e vis e eff ectiv e me an s for descr ibing ,characterizingandmodelingthe se sy st ems s o thatthe irstructu re andd yna mi c s ar e better understoo d, a ll owing pr edicti o ns ,i dentific at ion ofp o ssibl e im p ro vements, and s i mu la tions a im ed atev alu ati ng th e eff ects o f varyin g sce n arios and cond i tions. At the sa m e t i me,com plex networ ks r e sear ch i s n owa matu re ar ea ca t ering for all suchre quirem ents, from the cha racterizat i o n of thestru c tu r e of the inter actio n betweent ourism c ompan ies to t he relati o n ship wit h t heres pec t i ve dynamics ofi n tera ct ions an d i nformat ion ex cha nge .The curre nt artic le h as b rou ght t h ese twoim por ta ntareas togeth er wi th r es pe c t t o the c o mp r e hens iv eanal ysi sof th e El b a t ourismdestinati onn etwo rk ,a real- world structu re which has b een recen t l y obtain ed through systematic a n d caref ulinves tiga tion incl udi ng fie ldd ata co llecti on. T he ma i n cont r i bu tio ns of our wo r k ar e lis te d an d discu ssed as follows: * Com plete Real-Wo rld Exa m p le of th e Su pe r edg e s Approach:* Thi s work rep re s en ts the ver y fi rs t compr ehensiv e app l ication of the s uperedges a ppro a c h t o a real-w orld net work. Ass uch,s pe cialcar e hasbe eninves ted in ord er to makin gthe re spect iv e concep ts and interpretationsclearfromthe contextoft hespecificappl ication, n ame lytouri smd estin atio n s tru c tures . In particula r ,eac h of the choice s w hic h hav e t o be ma de r egarding the supe r edges methodol ogy— inc lud i ng i np ut/output, str uct ur a l measure me nts, dynami cs and d yn a mical measu rement s — hav e be e n moti vate d a nd justif ied w i th resp ec tt o thetour is m appl icatio n . Th e methodology wasfirst i llust r ate d wit hrespect to a hypotheti cal simplenetwor k of comp anies,an d then ap pl ied system a tically t o the real-w or ld E lba touri sm d e s tinat ionst ruc ture. Ass u ch , t hi s fi rstcompl et e ex ample can be usedasa applic at ion r eferen c eg u ide for re sea rcher s intendingt o ap p ly the s upered ges ap proacht o o th er spec ifi c problems. *Compre h ens iv e St ructural C hara ct eri za tion:* This wor k has pr esentedo n e of t hev eryf irst c om prehensive a nalys is of thet o pol ogicalc har act eri stic s o f at ourism destination netwo rk,inclu d ing trad ition al measure m ents , modula r ity as well a s the number of dis t inct pa t hs b etween pa irs of no des. Acomp ari son, a nd t h e su b st anti al sim ilar it y, of the o ut comes w i th p r evi ousknowl ed ge of the re lationsh ips a mo ng the to u rism co mpan ie s located atElba stakeho ld ers[@Pech l ane r:

of_what is_known of the behaviors_of tourism_operators_in the_destination_(see for example [@Tallinucci:2006]_or [@Pechlaner:2003]). However, the_identification of the specific_trends exhibited by_each_type of company only became clearer through the superedges integration of the structural and_dynamical_features. Concluding Remarks ================== Tourism_destination_networks_are amongst the most complex_real-world systems, involving intrincated structure_and non-linear_dynamics. Because of the economical importance of such_systems,_it becomes increasingly_important to devise effective means for describing, characterizing and_modeling these systems so that their_structure and dynamics_are_better_understood, allowing predictions, identification_of possible improvements, and simulations aimed_at evaluating the effects of varying_scenarios and conditions. At the same time,_complex networks research is now a_mature area catering for all_such requirements,_from the characterization of the_structure of the_interaction between_tourism companies to_the relationship with the respective dynamics_of interactions and_information exchange. The current article has_brought_these two important_areas_together_with respect_to the comprehensive_analysis_of the_Elba_tourism destination network, a real-world structure_which_has been recently obtained through systematic and_careful investigation including field_data_collection. The main contributions_of our work are listed_and discussed as follows: *Complete Real-World Example_of the_Superedges Approach:*_This work represents the very first comprehensive application of the superedges_approach to a real-world network. As_such, special care has_been invested_in_order to making_the_respective concepts_and interpretations clear from the context of_the specific_application, namely tourism destination structures. In_particular, each of the_choices_which have to be made regarding_the superedges methodology — including input/output,_structural measurements, dynamics and dynamical_measurements_—_have been motivated and justified_with respect to the tourism application._The methodology was_first illustrated with respect to a hypothetical_simple_network of companies, and then applied_systematically_to the real-world Elba tourism destination_structure._As_such, this first complete example_can be used as a application_reference guide for researchers intending to apply the superedges_approach to other_specific problems. *Comprehensive Structural Characterization:* This_work_has_presented one of the very first comprehensive analysis of the_topological characteristics_of a tourism_destination network, including traditional measurements, modularity as well as the_number of distinct paths between pairs of_nodes. A comparison, and the substantial similarity, of the outcomes with_previous knowledge of the relationships among the tourism_companies located at Elba stakeholders [@Pechlaner:

varphi$*, respectively.* The field equation with the potential (\[GPPV\]) is$$\square\phi+m^{2}\phi+a\lambda\phi^{a-1}+\sum_{n}b_{n}\kappa_{n}\phi^{b_{n}-1}=0. \label{phieqan}$$ The duality transformations (\[phitanspol\]) and (\[xtanspol\]) give$$\square\phi\rightarrow\frac{2-a}{2}\varphi^{\frac{a}{2-a}}\square\varphi +\frac{a}{2}\varphi^{\frac{2}{2-a}-2}\partial_{\mu}\varphi\partial^{\mu }\varphi. \label{2dtansxn}$$ Substituting the transformation into the field equation (\[phieqan\]) gives an equation of $\varphi$:$$\frac{2-a}{2}\varphi^{\frac{a}{2-a}}\square\varphi+\frac{a}{2}\varphi ^{\frac{2}{2-a}-2}\partial_{\mu}\varphi\partial^{\mu}\phi+m^{2}\varphi ^{\frac{2}{2-a}}+a\lambda\varphi^{\frac{2\left( a-1\right) }{2-a}}+\sum _{n}b_{n}\kappa_{n}\varphi^{\frac{2\left( b_{n}-1\right) }{2-a}}=0. \label{equatansn1}$$ The transformation of the coupling constant is given by the duality transformation (\[lamG\]):$$\lambda\rightarrow-\left( \frac{1}{2}\partial_{\mu}\varphi\partial^{\mu }\varphi+\frac{1}{2}m^{2}\varphi^{2}+\eta\varphi^{A}+\sum_{n}\sigma_{n}\varphi^{B_{n}}\right) .$$ Then we arrive at the field equation of $U\left( \varphi\right) =\eta \varphi^{A}+\sum_{n}\sigma_{n}\phi^{B_{n}}$ with the duality relations (\[aandA\]), (\[bandB\]), and (\[kapsig\]): $$\square\varphi+m^{2}\varphi+A\eta\varphi^{A-1}+\sum_{n}B_{n}\sigma_{n}\varphi^{B_{n}-1}=0.$$ It

varphi$ *, respectively. * The field equation with the potential (\[GPPV\ ]) is$$\square\phi+m^{2}\phi+a\lambda\phi^{a-1}+\sum_{n}b_{n}\kappa_{n}\phi^{b_{n}-1}=0. \label{phieqan}$$ The duality transformation (\[phitanspol\ ]) and (\[xtanspol\ ]) give$$\square\phi\rightarrow\frac{2 - a}{2}\varphi^{\frac{a}{2 - a}}\square\varphi + \frac{a}{2}\varphi^{\frac{2}{2 - a}-2}\partial_{\mu}\varphi\partial^{\mu } \varphi. \label{2dtansxn}$$ substitute the transformation into the field equation (\[phieqan\ ]) gives an equality of $ \varphi$:$$\frac{2 - a}{2}\varphi^{\frac{a}{2 - a}}\square\varphi+\frac{a}{2}\varphi ^{\frac{2}{2 - a}-2}\partial_{\mu}\varphi\partial^{\mu}\phi+m^{2}\varphi ^{\frac{2}{2 - a}}+a\lambda\varphi^{\frac{2\left ( a-1\right) } { 2 - a}}+\sum _ { n}b_{n}\kappa_{n}\varphi^{\frac{2\left ( b_{n}-1\right) } { 2 - a}}=0. \label{equatansn1}$$ The transformation of the coupling constant is given by the duality transformation (\[lamG\]):$$\lambda\rightarrow-\left ( \frac{1}{2}\partial_{\mu}\varphi\partial^{\mu } \varphi+\frac{1}{2}m^{2}\varphi^{2}+\eta\varphi^{A}+\sum_{n}\sigma_{n}\varphi^{B_{n}}\right) .$$ Then we arrive at the plain equation of $ U\left ( \varphi\right) = \eta \varphi^{A}+\sum_{n}\sigma_{n}\phi^{B_{n}}$ with the duality relation (\[aandA\ ]), (\[bandB\ ]), and (\[kapsig\ ] ): $ $ \square\varphi+m^{2}\varphi+A\eta\varphi^{A-1}+\sum_{n}B_{n}\sigma_{n}\varphi^{B_{n}-1}=0.$$ It

varohi$*, respectively.* The fiela equation with the povential (\[GPPV\]) is$$\rquare\phi+m^{2}\phi+a\lambda\phi^{a-1}+\sum_{n}u_{n}\kappa_{n}\pyi^{b_{n}-1}=0. \label{phieqan}$$ The djality trwnsformarionw (\[phitanspol\]) and (\[xtanspol\]) nive$$\sxnare\phi\rightarrpw\frac{2-a}{2}\var[hi^{\frac{a}{2-a}}\squara\vxr'hi +\frac{a}{2}\varphi^{\frac{2}{2-a}-2}\partial_{\mu}\varphi\pawtial^{\mu }\fagphi. \label{2dtanfxn}$$ Xtbstjnuuing the transformation into the rield exuation (\[phieqsn\]) gives an equation of $\vagphi$:$$\vrac{2-a}{2}\varphi^{\frac{a}{2-a}}\seuare\varphi+\drac{w}{2}\carphi ^{\frac{2}{2-a}-2}\pxrtial_{\mu}\vag'hi\partial^{\mo}\phi+m^{2}\varphi ^{\frac{2}{2-a}}+a\lambda\varphi^{\fraz{2\left( a-1\right) }{2-q}}+\sym _{n}h_{t}\kappa_{n}\varpii^{\frac{2\jeft( b_{n}-1\righb) }{2-a}}=0. \labal{equatsnsn1}$$ The transnormavion of the coupling consvant is given by the duality drcnsformation (\[lamG\]):$$\lambea\eightdrrof-\lefg( \ffac{1}{2}\karviam_{\mu}\varohi\'artial^{\mu }\vadphi+\frac{1}{2}m^{2}\vaephi^{2}+\eta\varphi^{A}+\sum_{n}\sogiq_{n}\varphi^{B_{n}}\riggt) .$$ Thqn we arrive at the field equation of $U\lext( \varphi\right) =\eta \varphi^{Q}+\sum_{n}\sigma_{n}\phi^{B_{n}}$ with the dualyty relations (\[aandA\]), (\[bandB\]), and (\[kapsig\]): $$\square\varphi+k^{2}\varpii+X\etc\ywfphl^{A-1}+\sum_{n}B_{n}\sigma_{n}\varphi^{B_{n}-1}=0.$$ It

varphi$*, respectively.* The field equation with the is$$\square\phi+m^{2}\phi+a\lambda\phi^{a-1}+\sum_{n}b_{n}\kappa_{n}\phi^{b_{n}-1}=0. The duality (\[phitanspol\]) and (\[xtanspol\]) the into the field (\[phieqan\]) gives an of $\varphi$:$$\frac{2-a}{2}\varphi^{\frac{a}{2-a}}\square\varphi+\frac{a}{2}\varphi ^{\frac{2}{2-a}-2}\partial_{\mu}\varphi\partial^{\mu}\phi+m^{2}\varphi ^{\frac{2}{2-a}}+a\lambda\varphi^{\frac{2\left( a-1\right) }{2-a}}+\sum b_{n}-1\right) }{2-a}}=0. \label{equatansn1}$$ The transformation of the coupling constant is given by the transformation (\[lamG\]):$$\lambda\rightarrow-\left( \frac{1}{2}\partial_{\mu}\varphi\partial^{\mu }\varphi+\frac{1}{2}m^{2}\varphi^{2}+\eta\varphi^{A}+\sum_{n}\sigma_{n}\varphi^{B_{n}}\right) .$$ Then we arrive at the field equation of \varphi\right) \varphi^{A}+\sum_{n}\sigma_{n}\phi^{B_{n}}$ the relations (\[aandA\]), (\[bandB\]), and (\[kapsig\]): $$\square\varphi+m^{2}\varphi+A\eta\varphi^{A-1}+\sum_{n}B_{n}\sigma_{n}\varphi^{B_{n}-1}=0.$$ It

varphi$*, respectively.* The fielD equation wIth thE poTenTiAl (\[GPpV\]) is$$\Square\phi+m^{2}\phi+a\LAmbdA\phi^{a-1}+\sum_{n}b_{n}\kappa_{n}\phi^{b_{n}-1}=0. \Label{PhIEqan}$$ tHe DualiTy transFOrMATioNs (\[PhItaNsPOl\]) And (\[xtAnsPol\]) give$$\Square\phi\rIghTaRrow\frac{2-a}{2}\varPHi^{\Frac{a}{2-a}}\squaRe\vArphi +\frac{a}{2}\vaRphI^{\frac{2}{2-a}-2}\PaRtiAL_{\mu}\vaRphI\partIal^{\mu }\vARphi. \laBel{2dtansxN}$$ SUBstituTIng the tRANsFormAtion into the field EQuATion (\[phieqan\]) givEs an eqUaTIoN OF $\vaRphI$:$$\frac{2-a}{2}\varpHi^{\Frac{a}{2-A}}\Square\vARpHI+\FRac{A}{2}\Varphi ^{\frac{2}{2-a}-2}\paRtial_{\mu}\varpHI\paRtial^{\mU}\pHi+m^{2}\VArphi ^{\fRac{2}{2-a}}+a\LaMBda\Varphi^{\frac{2\lEft( a-1\Right) }{2-a}}+\sum _{N}b_{n}\kapPA_{n}\varphI^{\Frac{2\lefT( b_{n}-1\rigHt) }{2-a}}=0. \LabEl{eqUAtAnSn1}$$ THe TRanSFoRmaTIon Of the couPlInG consTant IS GIVen bY thE duaLity tRansformation (\[Lamg\]):$$\lamBDa\rIghtaRrow-\lEft( \fRaC{1}{2}\partIal_{\mu}\vArphi\PaRtial^{\mu }\varphi+\frAc{1}{2}m^{2}\vArphi^{2}+\eta\vArpHi^{a}+\suM_{n}\Sigma_{N}\Varphi^{b_{n}}\rIghT) .$$ Then we Arrive aT The FiELD EqUation of $U\left( \varphI\rIGHt) =\Eta \varphI^{A}+\sum_{n}\SIgMa_{N}\Phi^{B_{n}}$ witH tHe dUaliTY RelatIons (\[AAnDA\]), (\[bandB\]), aNd (\[kapsIG\]): $$\sQuAre\varpHi+M^{2}\varphI+A\Eta\VarPhi^{A-1}+\sUM_{n}B_{n}\Sigma_{n}\Varphi^{B_{n}-1}=0.$$ it

varphi$*, respectively.* The fieldequat ion wi th the pot ential (\[GPPV \ ]) i s$$\square\phi+m^{2}\p hi+a\ la m bda\ p hi ^{a-1 }+\sum_ { n} b _ {n} \k ap pa_ {n } \p hi^{b _{n }-1}=0. \label{ph ieq an }$$ The dua l it y transfor mat ions (\[phit ans pol\]) a nd( \[xta nsp ol\]) give$ $ \squar e\phi\rig ht a rrow\f r ac{2-a} { 2 }\ varp hi^{\frac{a}{2-a} } \s q uare\varphi +\ frac{a }{ 2 }\ v a rph i^{ \frac{2}{2 -a }-2}\ p artial_ { \m u } \ var p hi\partial^{\ mu }\varphi . \l abel{2 dt ans x n}$$ S ubsti tu t ing the transf orma tion into the f i eld equ a tion (\ [phieq an\ ])give s a nequ at i ono f$\v a rph i$:$$\fr ac {2 -a}{2 }\va r p h i ^{\f rac {a}{ 2-a}} \square\varph i+\ frac { a}{ 2}\va rphi^{\f ra c{2}{ 2-a}-2 }\par ti al_{\mu}\varphi \par tial^{\mu }\p hi +m^ {2 }\var p hi ^{\ fra c{2 }{2-a}} +a\lamb d a\v ar p h i ^{ \frac{2\left( a-1 \r i g ht ) }{2-a }}+\su m _ {n } b_{n}\ka pp a_{ n}\v a r phi^{ \fra c {2 \left( b_{n}- 1 \r ig ht) }{ 2- a}}=0. \ lab el{ equat a nsn1 }$$ Th e transf ormat i on of the coup l ing constanti sg i ve n bythe duality tr ansf o rmat ion( \[ lam G \]):$ $\lam bd a \r i ghtarrow-\left( \f ra c{1}{2 }\par tial_{\mu}\va rphi\parti a l ^ {\mu }\v arph i +\ f rac{1}{2}m^{2} \varp hi^{2}+\et a \varphi^ {A}+\ sum_{n}\ sigma_{n} \ v arphi^{B _{n }}\ rig ht) . $$ Then we arri v e atth e field eq uationof$U\ lef t( \ varphi\ri ght) =\ et a\v ar phi ^{A}+ \ sum_{n}\ si gma _{ n}\ phi^{ B _{n}}$ with the d ua l ity relati o ns ( \[aa nd A\ ]),(\[ ba ndB\] ), a n d ( \[kapsi g\]): $$\ squ a re\v ar ph i+m^{2} \varphi+A\eta \v arphi^{A-1 }+ \su m_{n}B _ { n}\sigma _{n}\varphi^{B_{n}-1}=0 . $$ It

varphi$*, respectively.* The_field equation_with the potential (\[GPPV\])_is$$\square\phi+m^{2}\phi+a\lambda\phi^{a-1}+\sum_{n}b_{n}\kappa_{n}\phi^{b_{n}-1}=0. \label{phieqan}$$ The_duality_transformations (\[phitanspol\])_and_(\[xtanspol\]) give$$\square\phi\rightarrow\frac{2-a}{2}\varphi^{\frac{a}{2-a}}\square\varphi +\frac{a}{2}\varphi^{\frac{2}{2-a}-2}\partial_{\mu}\varphi\partial^{\mu }\varphi. \label{2dtansxn}$$_Substituting the transformation_into the field equation_(\[phieqan\]) gives an_equation_of $\varphi$:$$\frac{2-a}{2}\varphi^{\frac{a}{2-a}}\square\varphi+\frac{a}{2}\varphi ^{\frac{2}{2-a}-2}\partial_{\mu}\varphi\partial^{\mu}\phi+m^{2}\varphi ^{\frac{2}{2-a}}+a\lambda\varphi^{\frac{2\left( a-1\right) }{2-a}}+\sum _{n}b_{n}\kappa_{n}\varphi^{\frac{2\left( b_{n}-1\right) }{2-a}}=0. \label{equatansn1}$$ The transformation of the_coupling_constant is_given_by_the duality transformation (\[lamG\]):$$\lambda\rightarrow-\left( _\frac{1}{2}\partial_{\mu}\varphi\partial^{\mu }\varphi+\frac{1}{2}m^{2}\varphi^{2}+\eta\varphi^{A}+\sum_{n}\sigma_{n}\varphi^{B_{n}}\right) .$$ Then we arrive_at the_field equation of $U\left( \varphi\right) =\eta \varphi^{A}+\sum_{n}\sigma_{n}\phi^{B_{n}}$_with_the duality relations_(\[aandA\]), (\[bandB\]), and (\[kapsig\]): $$\square\varphi+m^{2}\varphi+A\eta\varphi^{A-1}+\sum_{n}B_{n}\sigma_{n}\varphi^{B_{n}-1}=0.$$ It

For example, in [@Chen:2016tdd]\[section 4.2\], it was argued that one (0,2) Landau-Ginzburg model those B/2 correlation functions correctly match those of the corresponding A/2 theory on ${\mathbb P}^1 \times {\mathbb P}^1$ had superpotential $$W \: = \: F_1 J_1 \: + \: \tilde{F}_1 \tilde{J}_1,$$ where $$\begin{aligned} J_1 & = & a X_1 \: - \: \frac{q_1}{X_1} \: + \: b \frac{ \tilde{X}_1^2}{ X_1 } \: + \: \mu \tilde{X}_1, \\ \tilde{J}_1 & = & d \tilde{X}_1 \: - \: \frac{q_2}{\tilde{X}_1} \: + \: c \frac{ X_1^2 }{\tilde{X}_1} \: + \: \nu X_1,\end{aligned}$$ with $$\mu \: = \: \det(A+B) - \det A - \det B, \: \: \: \nu \: = \: \det(C+D) - \det C - \det D,$$ and operator mirror map $$\sigma \: = \: X_1, \: \: \: \tilde{\sigma} \: = \: \tilde{X}_1.$$ These expressions have the good property that they are in terms of determinants of the matrices $A$, $B$, $C$, $D$, and so respect global symmetries of the original theory. For that matter, the A/2 correlation functions only depend upon those determinants, which is explicit in the mirrors constructed in [@Chen:2016tdd]. For purposes of comparison, for ${\mathbb P}^1 \times {\mathbb P}^1$, the superpotential (\[eq:p1p1:mirror1\]) takes the form $$\begin{aligned} W & = & - F_1 \left[ \exp\left( - Y_1 \right) \: - \: q_1 \frac{ ( a_1 d_0 - b_1 c_0 ) }{\Delta_0} \exp\left( + Y_1\right) \: - \: q_2 \frac{ ( b_1 a_0 - a

For example, in [ @Chen:2016tdd]\[section 4.2\ ], it was argued that one (0,2) Landau - Ginzburg model those B/2 correlation function correctly pit those of the corresponding A/2 theory on $ { \mathbb P}^1 \times { \mathbb P}^1 $ had superpotential $ $ west \: = \: F_1 J_1 \: + \: \tilde{F}_1 \tilde{J}_1,$$ where $ $ \begin{aligned } J_1 & = & a X_1 \: - \: \frac{q_1}{X_1 } \: + \: b \frac { \tilde{X}_1 ^ 2 } { X_1 } \: + \: \mu \tilde{X}_1, \\ \tilde{J}_1 & = & d \tilde{X}_1 \: - \: \frac{q_2}{\tilde{X}_1 } \: + \: c \frac { X_1 ^ 2 } { \tilde{X}_1 } \: + \: \nu X_1,\end{aligned}$$ with $ $ \mu \: = \: \det(A+B) - \det A - \det B, \: \: \: \nu \: = \: \det(C+D) - \det C - \det D,$$ and hustler mirror map $ $ \sigma \: = \: X_1, \: \: \: \tilde{\sigma } \: = \: \tilde{X}_1.$$ These expressions have the full property that they are in terms of determinant of the matrices $ A$, $ B$, $ C$, $ D$, and so respect ball-shaped symmetries of the original theory. For that topic, the A/2 correlation routine merely depend upon those determinants, which is explicit in the mirror constructed in [ @Chen:2016tdd ]. For purposes of comparison, for $ { \mathbb P}^1 \times { \mathbb P}^1 $, the superpotential   (\[eq: p1p1: mirror1\ ]) takes the form $ $ \begin{aligned } W & = & - F_1 \left [ \exp\left (- Y_1 \right) \: - \: q_1 \frac { (a_1 d_0 - b_1 c_0) } { \Delta_0 } \exp\left (+ Y_1\right) \: - \: q_2 \frac { (b_1 a_0 - a

For example, in [@Chen:2016tdd]\[sectiun 4.2\], it was argugd that mne (0,2) Lzndau-Ginxburg model those B/2 correlatmon dunctuons correctly match tfose of tje correwponving A/2 theory on ${\mathbb I}^1 \timsd {\matkbu P}^1$ had superpojential $$W \: = \: F_1 J_1 \: + \: \tilde{F}_1 \tklbe{J}_1,$$ where $$\begin{aligned} J_1 & = & a X_1 \: - \: \frwc{q_1}{X_1} \: + \: b \frac{ \tilde{X}_1^2}{ X_1 } \: + \: \iu \tjlde{X}_1, \\ \tilde{J}_1 & = & d \tilde{X}_1 \: - \: \frac{q_2}{\timde{X}_1} \: + \: b \frac{ X_1^2 }{\tilde{X}_1} \: + \: \nu X_1,\end{aligned}$$ with $$\mu \: = \: \deh(A+B) - \det A - \det B, \: \: \: \nu \: = \: \det(C+E) - \dqr C - \det D,$$ avd operatog mirror mak $$\sigma \: = \: X_1, \: \: \: \tilde{\sigma} \: = \: \tildd{X}_1.$$ Thexe expressuobs jdve the goov propvrty that then are it terms of determinanbs of thw matrices $A$, $B$, $C$, $D$, anv so respect global fymmetriev kf the original tyeiry. Fmr tvat oqttdr, uhe A/2 correpatmon functiohs only depwnd upon those detetmybants, which ia explycyt in the mirrors constructed in [@Chen:2016tdg]. Fod purposes of comparisob, for ${\mathbb P}^1 \times {\lathbb P}^1$, ehe superpotential (\[eq:p1p1:mirror1\]) takes the form $$\begin{dlignxd} D & = & - N_1 \lewr[ \edp\left( - Y_1 \right) \: - \: q_1 \frac{ ( a_1 d_0 - b_1 c_0 ) }{\Delta_0} \exp\jsfu( + J_1\right) \: - \: q_2 \frac{ ( b_1 a_0 - a

For example, in [@Chen:2016tdd]\[section 4.2\], it was one Landau-Ginzburg model B/2 correlation functions corresponding theory on ${\mathbb \times {\mathbb P}^1$ superpotential $$W \: = \: F_1 \: + \: \tilde{F}_1 \tilde{J}_1,$$ where $$\begin{aligned} J_1 & = & a X_1 - \: \frac{q_1}{X_1} \: + \: b \frac{ \tilde{X}_1^2}{ X_1 } \: + \mu \\ & & d \tilde{X}_1 \: - \: \frac{q_2}{\tilde{X}_1} \: + \: c \frac{ X_1^2 }{\tilde{X}_1} \: + \nu X_1,\end{aligned}$$ with $$\mu \: = \: \det(A+B) \det A - \det \: \: \: \nu \: \: - \det - D,$$ operator mirror map \: = \: X_1, \: \: \: \tilde{\sigma} \: = \: \tilde{X}_1.$$ These expressions have the good that they terms of of matrices $B$, $C$, $D$, respect global symmetries of the original matter, the A/2 correlation functions only depend upon determinants, which explicit in the mirrors constructed in For purposes of comparison, for ${\mathbb P}^1 \times P}^1$, the superpotential (\[eq:p1p1:mirror1\]) takes the form $$\begin{aligned} W & = & - F_1 \left[ Y_1 \right) \: - q_1 \frac{ ( d_0 b_1 ) \exp\left( + \: - \: q_2 \frac{ ( b_1 a_0 - a

For example, in [@Chen:2016tdd]\[sectioN 4.2\], it was arguEd thaT onE (0,2) LaNdAu-GiNzbuRg model those B/2 cORrelAtion functions correctlY matcH tHOse oF ThE corrEspondiNG A/2 THEorY oN ${\mAthBb p}^1 \TiMes {\maThbB P}^1$ had suPerpotentiAl $$W \: = \: f_1 J_1 \: + \: \Tilde{F}_1 \tilde{J}_1,$$ WHeRe $$\begin{aliGneD} J_1 & = & a X_1 \: - \: \frac{q_1}{X_1} \: + \: b \fRac{ \Tilde{X}_1^2}{ x_1 } \: + \: \mU \tiLDe{X}_1, \\ \tiLde{j}_1 & = & d \tilDe{X}_1 \: - \: \fraC{Q_2}{\tilde{x}_1} \: + \: c \frac{ X_1^2 }{\tiLdE{x}_1} \: + \: \nu X_1,\enD{Aligned}$$ WITh $$\Mu \: = \: \deT(A+B) - \det A - \det B, \: \: \: \nu \: = \: \det(C+d) - \DeT c - \det D,$$ and operatOr mirrOr MAp $$\SIGma \: = \: x_1, \: \: \: \tiLde{\sigma} \: = \: \tiLdE{X}_1.$$ TheSE expresSIoNS HAve THe good propertY that they arE In tErms of DeTerMInants Of the MaTRicEs $A$, $B$, $C$, $D$, and so RespEct global SymmetRIes of thE OriginaL theorY. FoR thAt maTTeR, tHe A/2 CoRRelATiOn fUNctIons only DePeNd upoN thoSE DETermInaNts, wHich iS explicit in thE miRrorS ConStrucTed in [@chen:2016TdD]. For pUrposeS of coMpArison, for ${\mathbb p}^1 \timEs {\mathbb P}^1$, The SuPerPoTentiAL (\[eq:p1p1:mIrrOr1\]) tAkes the Form $$\begIN{alIgNED} w & = & - F_1 \Left[ \exp\left( - Y_1 \right) \: - \: q_1 \FrAC{ ( A_1 d_0 - B_1 c_0 ) }{\Delta_0} \eXp\left( + y_1\RiGhT) \: - \: Q_2 \frac{ ( b_1 a_0 - a

For example, in [@Chen:201 6tdd]\[sec tion4.2 \], i t wa s ar gued that one( 0,2) Landau-Ginzburg model thos eB /2 c o rr elati on func t io n s co rr ec tly m a tc h tho seof thecorrespond ing A /2 theory on ${ \mathbb P} ^1\times {\mat hbb P}^1$ h ads uperp ote ntial $$W \ : = \:F_1 J_1 \ :+ \: \t i lde{F}_ 1 \t ilde {J}_1,$$ where $$ \ be g in{aligned} J_ 1 & =&a X _ 1 \: -\: \frac{q _1 }{X_1 } \: + \ : b \ fra c { \tilde{X}_1 ^2}{ X_1 }\ : + \: \m u\ti l de{X}_ 1, \\ \ t ild e{J}_1 & =& d\tilde{X} _1 \:- \: \fr a c{q_2}{ \tilde {X} _1} \:+ \ :c \ fr a c{X _1 ^2} {\t ilde{X}_ 1} \ : + \ : \n u X _ 1,\e nd{ alig ned}$ $ with $$\mu\:= \: \de t(A+B ) - \ detA- \de t B, \ : \:\: \nu \: = \: \d et(C +D) - \de t C - \d et D,$$ and op era tor mirror map $$ \ sig ma \ : = \: X_1, \: \: \:\t i l de {\sigma} \: =\ :\t i lde{X}_1 .$ $ Thes e expre ssio n shave the goodp ro pe rty tha tthey a re in te rms o f det ermina nts of t he ma t rices $A$, $B$ , $C$, $D$, an d s o re s pect gl obal symmet ries of t he o r ig ina l theo ry. F or th a t matter, the A/2 c or relati on fu nctions onlydepend upo n t hose det ermi n an t s, which is ex plici t in the m i rrors co nstru cted in[@Chen:20 1 6 tdd]. F orpur pos eso f c omparison, fo r ${\m at hbb P}^ 1 \ times { \ma thb b P }^1 $, the supe rpotenti al  ( \[ eq :p1 p1:mi r ror1\])ta kes t heform$ $\begi n{ali gned }W& =& - F_1 \l e f t[ \ ex p\ left ( - Y _1 \r ight ) \: - \: q _1 \frac{ (a _1 d _0 - b_1 c_ 0 ) }{\Delta_ 0} \exp\left (+ Y _1\rig h t ) \: - \ : q_2 \frac{ ( b_1 a_0- a

For example,_in [@Chen:2016tdd]\[section_4.2\], it was argued_that one_(0,2)_Landau-Ginzburg model_those_B/2 correlation functions_correctly match those_of the corresponding A/2_theory on ${\mathbb_P}^1_\times {\mathbb P}^1$ had superpotential $$W \: = \: F_1 J_1 \: + \: \tilde{F}_1_\tilde{J}_1,$$_where $$\begin{aligned} J_1_&_=_& a X_1 \: -_\: \frac{q_1}{X_1} \: + \:_ b \frac{_\tilde{X}_1^2}{ X_1 } \: + \: \mu \tilde{X}_1, \\ \tilde{J}_1_&_= & d_\tilde{X}_1 \: - \: \frac{q_2}{\tilde{X}_1} \: + \: c \frac{_X_1^2 }{\tilde{X}_1} \: + \: \nu_X_1,\end{aligned}$$ with $$\mu_\:_=_\: \det(A+B) - \det_A - \det B, \: \: \: \nu_\: = \: \det(C+D) - \det_C - \det D,$$ and operator mirror_map $$\sigma \: = \: X_1, \:_\: \: \tilde{\sigma} \: = \:_\tilde{X}_1.$$ These expressions_have the good property that_they are in_terms of_determinants of the_matrices $A$, $B$, $C$, $D$, and_so respect global_symmetries of the original theory. For_that_matter, the A/2_correlation_functions_only depend_upon those determinants,_which_is explicit_in_the mirrors constructed in [@Chen:2016tdd]. For purposes_of_comparison, for ${\mathbb P}^1 \times {\mathbb P}^1$,_the superpotential (\[eq:p1p1:mirror1\]) takes the_form_$$\begin{aligned} W & = & -_F_1 \left[ \exp\left( - Y_1 \right)_\: - \: q_1 \frac{ (_a_1 d_0_- b_1_c_0 ) }{\Delta_0} \exp\left( + Y_1\right) \: - \: q_2 \frac{ (_b_1 a_0 - a

\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ }$$ Let$$\begin{tabular}{l} $(A_{3})$ $\ \widetilde{h}_{i}\in H^{1}\left( 0,T\right),\ i=1,2.$\end{tabular}\text{ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ }$$ Then we have the following theorem about the existence of a “strong solution”. **Theorem 2.3.** *Suppose that* $(A_{1})-(A_{3})$ *hold and the initial data* $\left( u_{0},u_{1}\right) \in H^{2}\times H^{1}$ *satisfies the compatibility conditions*$\ $$$\left\{ \begin{tabular}{l} $u_{0x}(0)=-\text{ }\left\vert u_{0}(0)\right\vert ^{\alpha -2}u_{0}(0)+\lambda _{0}u_{1}(0)+\widetilde{h}_{1}(0)u_{0}(1)+\widetilde{\lambda }_{1}u_{1}(1),\medskip $ \\ $-u_{0x}(1)=-\text{ }\left\vert u_{0}(1)\right\vert ^{\beta -2}u_{0}(1)+\lambda _{1}u_{1}(1)+\widetilde{h}_{0}(0)u_{0}(0)+\widetilde{\lambda }_{0}u_{1}(0).$\end{tabular}\right. \tag{2.3} \label{b3}$$ *Then problem* (\[1\]) – (\[4\]) *has a unique local solution *$$\left\{ \begin{tabular}{l} $u\in L^{\infty }\left( 0,T_{\ast };H^{2}\right),$ $u_{t}\in L^{\infty }\

\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ } $ $ Let$$\begin{tabular}{l } $ (A_{3})$ $ \ \widetilde{h}_{i}\in H^{1}\left (0,T\right),\ i=1,2.$\end{tabular}\text { \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ } $ $ Then we have the following theorem about the existence of a “ strong solution ”. * * Theorem 2.3. * * * presuppose that * $ (A_{1})-(A_{3})$ * delay and the initial data * $ \left (u_{0},u_{1}\right) \in H^{2}\times H^{1}$ * satisfies the compatibility conditions*$\ $ $ $ \left\ { \begin{tabular}{l } $ u_{0x}(0)=-\text { } \left\vert u_{0}(0)\right\vert ^{\alpha -2}u_{0}(0)+\lambda _ { 0}u_{1}(0)+\widetilde{h}_{1}(0)u_{0}(1)+\widetilde{\lambda } _ { 1}u_{1}(1),\medskip $ \\ $ -u_{0x}(1)=-\text { } \left\vert u_{0}(1)\right\vert ^{\beta -2}u_{0}(1)+\lambda _ { 1}u_{1}(1)+\widetilde{h}_{0}(0)u_{0}(0)+\widetilde{\lambda } _ { 0}u_{1}(0).$\end{tabular}\right. \tag{2.3 } \label{b3}$$ * Then problem * (\[1\ ]) – (\[4\ ]) * have a unique local solution   * $ $ \left\ { \begin{tabular}{l } $ u\in L^{\infty } \left (0,T_{\ast }; H^{2}\right),$ $ u_{t}\in L^{\infty } \

\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ }$$ Lgt$$\vegin{tebular}{l} $(Z_{3})$ $\ \widetklde{h}_{i}\in H^{1}\left( 0,T\right),\ i=1,2.$\end{tauulae}\text{ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ }$$ Then we have the follofing theorem atojt the existence of a “strong solution”. **Eheorem 2.3.** *Duppose that* $(A_{1})-(W_{3})$ *hokq ans the initial data* $\left( u_{0},u_{1}\right) \in H^{2}\times H^{1}$ *satisfies yhe compatibility conditiojs*$\ $$$\lfft\{ \begin{tabular}{l} $u_{0x}(0)=-\hext{ }\left\vetf u_{0}(0)\wught\vert ^{\alpfa -2}u_{0}(0)+\lambda _{0}l_{1}(0)+\cidetilde{h}_{1}(0)u_{0}(1)+\sidetilde{\lambda }_{1}u_{1}(1),\medskip $ \\ $-u_{0x}(1)=-\texg{ }\lefc\vert u_{0}(1)\righj\rwrt ^{\teta -2}u_{0}(1)+\lambda _{1}u_{1}(1)+\widvtilde{h}_{0}(0)u_{0}(0)+\widetlkde{\lamtda }_{0}u_{1}(0).$\enc{tabular}\right. \taj{2.3} \lqbel{b3}$$ *Then problem* (\[1\]) – (\[4\]) *ias a unique local sjlution *$$\lext\{ \yegin{tabular}{l} $u\in L^{\infry }\left( 0,T_{\asd };H^{2}\rktht),$ $u_{t}\jn L^{\jnfty }\

\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ }$$ Let$$\begin{tabular}{l} $(A_{3})$ $\ \widetilde{h}_{i}\in H^{1}\left( 0,T\right),\ i=1,2.$\end{tabular}\text{ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ }$$ Then we have the following theorem about the existence of a “strong solution”. 2.3.** *Suppose *hold and initial $\left( \in H^{2}\times H^{1}$ compatibility conditions*$\ $$$\left\{ \begin{tabular}{l} $u_{0x}(0)=-\text{ }\left\vert _{0}u_{1}(0)+\widetilde{h}_{1}(0)u_{0}(1)+\widetilde{\lambda }_{1}u_{1}(1),\medskip $ \\ $-u_{0x}(1)=-\text{ }\left\vert u_{0}(1)\right\vert ^{\beta _{1}u_{1}(1)+\widetilde{h}_{0}(0)u_{0}(0)+\widetilde{\lambda }_{0}u_{1}(0).$\end{tabular}\right. \label{b3}$$ *Then problem* (\[1\]) – (\[4\]) a unique local solution *$$\left\{ \begin{tabular}{l} $u\in L^{\infty 0,T_{\ast };H^{2}\right),$ $u_{t}\in L^{\infty }\

\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ }$$ Let$$\begin{tabular}{l} $(A_{3})$ $\ \widetildE{h}_{i}\in H^{1}\left( 0,t\righT),\ i=1,2.$\eNd{tAbUlar}\Text{ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ }$$ then we have the fOLlowIng theorem about the exisTence Of A “StroNG sOlutiOn”. **TheorEM 2.3.** *SUPPosE tHaT* $(A_{1})-(A_{3})$ *HoLD aNd the IniTial datA* $\left( u_{0},u_{1}\rigHt) \iN H^{2}\Times H^{1}$ *satisfIEs The compatiBilIty conditionS*$\ $$$\leFt\{ \begiN{tAbuLAr}{l} $u_{0x}(0)=-\TexT{ }\left\Vert u_{0}(0)\rIGht\verT ^{\alpha -2}u_{0}(0)+\laMbDA _{0}u_{1}(0)+\wideTIlde{h}_{1}(0)u_{0}(1)+\wIDEtIlde{\Lambda }_{1}u_{1}(1),\medskip $ \\ $-u_{0x}(1)=-\tEXt{ }\LEft\vert u_{0}(1)\right\vErt ^{\betA -2}u_{0}(1)+\LAmBDA _{1}u_{1}(1)+\wIdeTilde{h}_{0}(0)u_{0}(0)+\widEtIlde{\lAMbda }_{0}u_{1}(0).$\enD{TaBULAr}\rIGht. \tag{2.3} \label{b3}$$ *THen problem* (\[1\]) – (\[4\]) *hAS a uNique lOcAl sOLution *$$\Left\{ \bEgIN{taBular}{l} $u\in L^{\iNfty }\Left( 0,T_{\ast };H^{2}\Right),$ $u_{T}\In L^{\inftY }\

\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \\ \ \\\ \\ \\ \ \ \ \ \ \\ \ \ \ \ \ }$$ Let$$\begi n{tab ul a r}{l } $ (A_{3 })$ $\\ wi d e til de {h }_{ i} \ in H^{1 }\l eft( 0, T\right),\ i= 1, 2.$\end{tabu l ar }\text{ \\ \ \ \ \ \ \ \ \\ \ \\\ \ \ \ \ \\ \ \ \ \ \ \ \ \\ \ \ \ \ \ \ \ \\ \ \ \\ \\ \\ \ \ \ \ \ \ \ \ \\ \ \ \ \ \ \ \ \ \ \ \ \\ \ \ \\ \ \ \ }$ $ Then we have th e f oll o wing theoremabout the e x ist ence o fa “ s trongsolut io n ”. **Theorem2.3. ** *Suppo se tha t * $(A_{ 1 })-(A_{ 3})$ * hol d a nd t h ein iti al dat a *$\l e ft( u_{0},u _{ 1} \righ t) \ i n H ^{2} \ti mesH^{1} $ *satisfiesthe com p ati bilit y con diti on s*$\$$$\le ft\{\b egin{tabular}{l } $u _{0x}(0)= -\t ex t{}\ left\ v ert u_ {0} (0) \right\ vert ^{ \ alp ha - 2 }u _{0}(0)+\lambda _{ 0} u _ {1 }(0)+\wi detild e {h }_ { 1}(0)u_{ 0} (1) +\wi d e tilde {\la m bd a }_{1}u _{1}(1 ) ,\ me dskip $ \ \ $-u_ {0 x}( 1)= -\tex t { }\ left\v ert u_{0 }(1)\ r ight\vert ^{\b e ta -2}u_{0}(1 ) +\ l a mb d a _{ 1}u _{1}(1)+\wi deti l de{h }_{0 } (0 )u_ { 0}(0) +\wid et i ld e {\lambda }_{0}u_{1} (0 ).$\en d{tab ular}\right. \tag{2.3} \ label{b3 }$$*T h en problem* (\ [1\]) – (\[4\]) *has a u nique local s olution * $ $ \left\{\be gin {ta bul a r }{ l} $u\in L^{\ i n fty}\ left( 0 ,T_ {\ast } ;H^ {2} \ri ght ), $ $u_{t}\ in L^{\i nf ty } \

\_\ \_\ \ \ \_\ \_\_\ \_\_\ \ \_\ \ \_\ \ \ \_\ \ \_\_\ \ \ \ \ \ \ \ \ \ \ \ \ \ }$$ Let$$\begin{tabular}{l} $(A_{3})$_$\_\widetilde{h}_{i}\in H^{1}\left(_0,T\right),\_i=1,2.$\end{tabular}\text{_\ \ \ \ \_\ \ \ \ \_\ \_\ \ \ \ \ \ \ \_\_\ \ \_\ \ \ \ \ \ \ \ \_\ \ \ \ \ \ \_\ \ \_\_\_\ \ \ \_\ \ \ \ \ \_\ \ \ \ \ \_\ \ \ \ \ \ \_\ \ \ \ \ \ \_\ \ }$$ Then we have_the following_theorem about the existence of_a “strong solution”. **Theorem_2.3.** *Suppose_that* $(A_{1})-(A_{3})$ *hold_and the initial data* $\left( u_{0},u_{1}\right)_\in H^{2}\times H^{1}$_*satisfies the compatibility conditions*$\ $$$\left\{ \begin{tabular}{l} $u_{0x}(0)=-\text{ }\left\vert_u_{0}(0)\right\vert_^{\alpha -2}u_{0}(0)+\lambda _{0}u_{1}(0)+\widetilde{h}_{1}(0)u_{0}(1)+\widetilde{\lambda }_{1}u_{1}(1),\medskip_$_\\ $-u_{0x}(1)=-\text{_}\left\vert u_{0}(1)\right\vert_^{\beta -2}u_{0}(1)+\lambda _{1}u_{1}(1)+\widetilde{h}_{0}(0)u_{0}(0)+\widetilde{\lambda }_{0}u_{1}(0).$\end{tabular}\right.__\tag{2.3} _\label{b3}$$ *Then_problem* (\[1\]) – (\[4\]) *has a_unique_local solution *$$\left\{ \begin{tabular}{l} $u\in L^{\infty }\left( 0,T_{\ast };H^{2}\right),$ $u_{t}\in_L^{\infty }\

$$||b_h(x)-f(x)||_{L^{\infty}[0,h]}\leq \sup_{x\in(0,h)}\Big | \frac{d^4}{dx^4}b_h(x)\Big |\frac{h^4}{4!}. \label{ineq1}$$ The proof will be complete if $\frac{d^4}{dx^4}b_h(x)$ is bounded as $h\rightarrow 0$. We rely on the property that for each $x\in[0,h]$ there exists a $t^*\in[0,1]$ such that $\displaystyle \frac{d^4}{dx^4}b_h(x)=\mathcal{D}^4(B_{h_1}(t^*),B_{h_2}(t^*))$. Using equation (\[Fourth\]) we demonstrate that $\mathcal{D}^4(B_{h_1}(t),B_{h_2}(t))$ is bounded. We obtain the desired result by replacing each term by its Taylor expansion, then showing the lowest term in the numerator and denominator are both $\mathcal{O}(h^7)$. Since we are interpolating a function, after a shift to the origin, the terms from equation (\[Bezier2\]) become $\vec{A}=\vec{0}$, $\vec{D}=\langle h,f(h) \rangle$, $\vec{\alpha}=\langle 1, f'(0) \rangle$, and $\vec{\beta}=\langle 1, f'(h) \rangle$. Starting with $$\begin{aligned} f(h)&=hf'(0)+\frac{h^2}{2}f''(0)+\mathcal{O}(h^3)\\ f'(h)&=f'(0)+hf''(0)+\frac{h^2}{2}f'''(0)+\mathcal{O}(h^3),\end{aligned}$$ we obtain $$\begin{aligned} B_{h_1}'(t)&=h+\mathcal{O}(h^2), \quad B_{h_1}''(t)=2h^2(P_1(h)(-2+3t)+P_2(h)(-1+3t))+\mathcal{O}(h^3),\\ B_{h_1}'''(t)&=6h^2(P_1(h)+P_2(h))+\mathcal

$ $ ||b_h(x)-f(x)||_{L^{\infty}[0,h]}\leq \sup_{x\in(0,h)}\Big | \frac{d^4}{dx^4}b_h(x)\Big |\frac{h^4}{4! }. \label{ineq1}$$ The proof will be complete if $ \frac{d^4}{dx^4}b_h(x)$ is bounded as $ h\rightarrow 0$. We rely on the place that for each $ x\in[0,h]$ there exist a $ t^*\in[0,1]$ such that $ \displaystyle \frac{d^4}{dx^4}b_h(x)=\mathcal{D}^4(B_{h_1}(t^*),B_{h_2}(t^*))$. Using equation (\[Fourth\ ]) we attest that $ \mathcal{D}^4(B_{h_1}(t),B_{h_2}(t))$ is bound. We obtain the desired consequence by replacing each term by its Taylor expansion, then read the lowest term in the numerator and denominator are both $ \mathcal{O}(h^7)$. Since we are interpolate a function, after a shift to the lineage, the term from equation (\[Bezier2\ ]) become $ \vec{A}=\vec{0}$, $ \vec{D}=\langle h, f(h) \rangle$, $ \vec{\alpha}=\langle 1, f'(0) \rangle$, and $ \vec{\beta}=\langle 1, f'(h) \rangle$. Starting with $ $ \begin{aligned } f(h)&=hf'(0)+\frac{h^2}{2}f''(0)+\mathcal{O}(h^3)\\ f'(h)&=f'(0)+hf''(0)+\frac{h^2}{2}f'''(0)+\mathcal{O}(h^3),\end{aligned}$$ we obtain $ $ \begin{aligned } B_{h_1}'(t)&=h+\mathcal{O}(h^2), \quad B_{h_1}''(t)=2h^2(P_1(h)(-2 + 3t)+P_2(h)(-1 + 3t))+\mathcal{O}(h^3),\\ B_{h_1}'''(t)&=6h^2(P_1(h)+P_2(h))+\mathcal

$$||b_h(d)-f(x)||_{L^{\infty}[0,h]}\leq \sup_{x\in(0,h)}\Big | \frac{d^4}{dx^4}b_h(x)\Big |\frac{h^4}{4!}. \nabel{iheq1}$$ The oroof will be complete if $\frec{d^4}{dz^4}b_h(x)$ us bounded as $h\rightarfow 0$. We rvly on thw priperty that for eacm $x\in[0,g]$ thexe exists a $t^*\in[0,1]$ xuch that $\gisplaystyle \fsaz{d^4}{bx^4}b_h(x)=\mathcal{D}^4(B_{h_1}(t^*),B_{h_2}(t^*))$. Using equation (\[Fjurth\]) wr femonstrate thwt $\msehcam{D}^4(B_{h_1}(t),B_{h_2}(t))$ is bounded. We obtain the sesired result by rellacing each term by its Twylog expansion, then sjowing the oowefr term in thd numeratog and denomjnator are both $\mathcal{O}(h^7)$. Since de arz interpolajnbg w function, ahter a shift to thc origit, the trrms from equabion (\[Ueziwr2\]) become $\vec{A}=\vec{0}$, $\vec{V}=\langle h,f(h) \rangle$, $\vgc{\alpha}=\lanclz 1, f'(0) \rangle$, and $\vec{\bera}=\oanglg 1, f'(h) \raneoe$. Rtadtmng with $$\hegmn{aligned} f(h)&=gf'(0)+\frac{h^2}{2}f''(0)+\matycal{O}(h^3)\\ f'(h)&=f'(0)+hf''(0)+\frac{h^2}{2}f'''(0)+\mauhcwo{O}(h^3),\end{aligned}$$ we obeayn $$\begin{aligned} B_{h_1}'(t)&=h+\mathcal{O}(h^2), \quad B_{h_1}''(t)=2h^2(K_1(h)(-2+3t)+P_2(h)(-1+3f))+\mathcal{O}(h^3),\\ B_{h_1}'''(t)&=6h^2(P_1(h)+P_2(h))+\matycal

$$||b_h(x)-f(x)||_{L^{\infty}[0,h]}\leq \sup_{x\in(0,h)}\Big | \frac{d^4}{dx^4}b_h(x)\Big |\frac{h^4}{4!}. \label{ineq1}$$ The be if $\frac{d^4}{dx^4}b_h(x)$ bounded as $h\rightarrow property for each $x\in[0,h]$ exists a $t^*\in[0,1]$ that $\displaystyle \frac{d^4}{dx^4}b_h(x)=\mathcal{D}^4(B_{h_1}(t^*),B_{h_2}(t^*))$. Using equation (\[Fourth\]) demonstrate that $\mathcal{D}^4(B_{h_1}(t),B_{h_2}(t))$ is bounded. We obtain the desired result by replacing each by its Taylor expansion, then showing the lowest term in the numerator and are $\mathcal{O}(h^7)$. we interpolating a function, after a shift to the origin, the terms from equation (\[Bezier2\]) become $\vec{A}=\vec{0}$, h,f(h) \rangle$, $\vec{\alpha}=\langle 1, f'(0) \rangle$, and $\vec{\beta}=\langle f'(h) \rangle$. Starting with f(h)&=hf'(0)+\frac{h^2}{2}f''(0)+\mathcal{O}(h^3)\\ f'(h)&=f'(0)+hf''(0)+\frac{h^2}{2}f'''(0)+\mathcal{O}(h^3),\end{aligned}$$ we obtain $$\begin{aligned} \quad B_{h_1}'''(t)&=6h^2(P_1(h)+P_2(h))+\mathcal

$$||b_h(x)-f(x)||_{L^{\infty}[0,h]}\leq \sup_{x\in(0,h)}\Big | \Frac{d^4}{dx^4}b_h(x)\big |\frAc{h^4}{4!}. \LabEl{Ineq1}$$ the pRoof will be compLEte iF $\frac{d^4}{dx^4}b_h(x)$ is bounded as $H\righTaRRow 0$. WE ReLy on tHe propeRTy THAt fOr EaCh $x\In[0,H]$ ThEre exIstS a $t^*\in[0,1]$ suCh that $\dispLayStYle \frac{d^4}{dx^4}b_h(X)=\MaThcal{D}^4(B_{h_1}(t^*),B_{H_2}(t^*))$. USing equation (\[fouRth\]) we dEmOnsTRate tHat $\MathcAl{D}^4(B_{h_1}(t),b_{H_2}(t))$ is boUnded. We obTaIN the deSIred resULT bY repLacing each term by iTS TAYlor expansion, tHen shoWiNG tHE LowEst Term in the nUmEratoR And denoMInATOR arE Both $\mathcal{O}(h^7)$. since we are iNTerPolatiNg A fuNCtion, aFter a ShIFt tO the origin, tHe teRms from eqUation (\[bEzier2\]) beCOme $\vec{A}=\Vec{0}$, $\vec{d}=\laNglE h,f(h) \RAnGlE$, $\veC{\aLPha}=\LAnGle 1, F'(0) \RanGle$, and $\veC{\bEtA}=\langLe 1, f'(h) \RANGLe$. StArtIng wIth $$\beGin{aligned} f(h)&=hF'(0)+\frAc{h^2}{2}f''(0)+\MAthCal{O}(h^3)\\ F'(h)&=f'(0)+hf''(0)+\Frac{H^2}{2}f'''(0)+\MathcAl{O}(h^3),\enD{aligNeD}$$ we obtain $$\begin{aLignEd} B_{h_1}'(t)&=h+\matHcaL{O}(H^2), \quAd b_{h_1}''(t)=2h^2(P_1(H)(-2+3T)+P_2(h)(-1+3t))+\maThcAl{O}(H^3),\\ B_{h_1}'''(t)&=6h^2(P_1(h)+p_2(h))+\mathcAL

$$||b_h(x)-f(x)||_{L^{\in fty}[0,h]} \leq\su p_{ x\ in(0 ,h)} \Big | \frac{d ^ 4}{d x^4}b_h(x)\Big |\frac{ h^4}{ 4! } . \l a be l{ine q1}$$ T h ep r oof w il l b ec om plete if $\frac {d^4}{dx^4 }b_ h( x)$ is bound e das $h\righ tar row 0$. We r ely on th epro p ertytha t for each$ x\in[0 ,h]$ ther ee xistsa $t^*\i n [ 0, 1]$such that $\displ a ys t yle \frac{d^4} {dx^4} b_ h (x ) = \ma thc al{D}^4(B_ {h _1}(t ^ *),B_{h _ 2} ( t ^ *)) $ . Using equat ion (\[Four t h\] ) we d em ons t rate t hat $ \m a thc al{D}^4(B_{ h_1} (t),B_{h_ 2}(t)) $ is bou n ded. We obtai n t hedesi r ed r esu lt byr ep lac i ngeach ter mby itsTayl o r e xpan sio n, t hen s howing the lo wes t te r m i n the nume rato rand d enomin atorar e both $\mathca l{O} (h^7)$. Sin ce we a re in t erpola tin g a functi on, aft e r a s h i f tto the origin, the t e r ms from eq uation (\ [B e zier2\]) b eco me $ \ v ec{A} =\ve c {0 }$, $\ve c{D}=\ l an gl e h,f(h )\rangl e$ , $ \ve c{\al p ha}= \langl e 1, f'( 0) \r a ngle$, and $\v e c{\beta}=\lan g le 1 ,f '(h) \r angle$. Sta rtin g wit h $$ \ be gin { align ed} f (h ) &= h f'(0)+\frac{h^2}{2} f' '(0)+\ mathc al{O}(h^3)\\f'(h)&=f'( 0 ) + hf''(0)+ \fra c {h ^ 2}{2}f'''(0)+\ mathc al{O}(h^3) , \end{ali gned} $$ we ob tain $$\b e g in{align ed} B _{h _1} ' ( t) &=h+\mathcal{ O } (h^2 ), \quadB_{ h_1}''( t)= 2h^ 2(P _1( h) (-2+3t)+P _2(h)(-1 +3 t) )+ \m ath cal{O } (h^3),\\ B_ {h _1} '''(t ) &=6h^2 (P_1( h)+P _2 (h ) )+\ mathcal

$$||b_h(x)-f(x)||_{L^{\infty}[0,h]}\leq_\sup_{x\in(0,h)}\Big |_\frac{d^4}{dx^4}b_h(x)\Big |\frac{h^4}{4!}. \label{ineq1}$$ The_proof will_be_complete if_$\frac{d^4}{dx^4}b_h(x)$_is bounded as_$h\rightarrow 0$. We_rely on the property_that for each_$x\in[0,h]$_there exists a $t^*\in[0,1]$ such that $\displaystyle \frac{d^4}{dx^4}b_h(x)=\mathcal{D}^4(B_{h_1}(t^*),B_{h_2}(t^*))$. Using equation (\[Fourth\]) we demonstrate that_$\mathcal{D}^4(B_{h_1}(t),B_{h_2}(t))$_is bounded._We_obtain_the desired result by replacing_each term by its Taylor_expansion, then_showing the lowest term in the numerator and_denominator_are both $\mathcal{O}(h^7)$. Since_we are interpolating a function, after a shift to_the origin, the terms from equation_(\[Bezier2\]) become $\vec{A}=\vec{0}$,_$\vec{D}=\langle_h,f(h)_\rangle$, $\vec{\alpha}=\langle 1, f'(0)_\rangle$, and $\vec{\beta}=\langle 1, f'(h) \rangle$._Starting with $$\begin{aligned} f(h)&=hf'(0)+\frac{h^2}{2}f''(0)+\mathcal{O}(h^3)\\ f'(h)&=f'(0)+hf''(0)+\frac{h^2}{2}f'''(0)+\mathcal{O}(h^3),\end{aligned}$$ we obtain $$\begin{aligned} _B_{h_1}'(t)&=h+\mathcal{O}(h^2), \quad B_{h_1}''(t)=2h^2(P_1(h)(-2+3t)+P_2(h)(-1+3t))+\mathcal{O}(h^3),\\ B_{h_1}'''(t)&=6h^2(P_1(h)+P_2(h))+\mathcal

according to their spectra, generally exhibiting broad, low equivalent width[^3] emission lines of hydrogen, neutral and ionised helium, and Bowen fluorescence lines of doubly-ionised nitrogen and oxygen [e.g. @Gezari2012; @Holoien2014; @Arcavi2014; @Leloudas2019]. This prompted @vanVelzen2020 to suggest three sub-classes labelled TDE-H, TDE-He and TDE-Bowen, though some TDEs defy a consistent classification by changing their apparent spectral type as they evolve [@Nicholl2019]. TDE flares were initially predicted to be brightest in X-rays, due to the high temperature of an accretion disk, and indeed this is the wavelength where the earliest TDE candidates were identified [@Komossa2002]. However, the optically-discovered TDEs have proven to be surprisingly diverse in their X-ray properties. Their X-ray to optical ratios at maximum light range from $\gtrsim10^{3}$ to $<10^{-3}$ [@Auchettl2017]. Producing such luminous optical emission without significant X-ray flux can be explained in one of two ways: either X-ray faint TDEs are powered primarily by stream collisions rather than accretion, or the accretion disk emission is reprocessed through an atmosphere [@Strubbe2009; @Guillochon2014; @Roth2016]. Several lines of evidence have indicated that accretion disks do form promptly even in X-ray faint TDEs: Bowen fluorescence lines that require excitation from far-UV photons [@Blagorodnova2018; @Leloudas2019]; low-ionisation iron emission appearing shortly after maximum light [@Wevers2019b]; and recently the direct detection of double-peaked Balmer lines that match predicted disk profiles [@Short2020; @Hung2020]. Thus a critical question is to understand the nature and origin of the implied reprocessing layer. It has already been established that this cannot be simply the unbound debris stream, as the apparent cross-section is too low to intercept a significant fraction of the TDE flux [@Guillochon2014]. Inhibiting progress is the messy geometry of the debris. Colliding streams, inflowing and outflowing gas, and a viewing-angle dependence on both the broad-band [@Dai2018] and spectroscopic [@Nicholl2019] properties all contribute to a messy knot that must be untangled. One important

according to their spectra, generally exhibiting wide, depleted equivalent width[^3 ] emission lines of hydrogen, neutral and ionized helium, and Bowen fluorescence lines of doubly - ionised nitrogen and oxygen [ e.g. @Gezari2012; @Holoien2014; @Arcavi2014; @Leloudas2019 ]. This motivate @vanVelzen2020 to suggest three sub - class labelled TDE - H, TDE - He and TDE - Bowen, though some TDEs defy a consistent categorization by changing their apparent spectral type as they evolve [ @Nicholl2019 ]. TDE flare were initially predicted to be brightest in X - rays, due to the high temperature of an accretion phonograph record, and indeed this is the wavelength where the earliest TDE candidates were identified [ @Komossa2002 ]. However, the optically - discover TDEs have proven to be surprisingly diverse in their X - beam properties. Their X - ray to optical ratios at maximal light range from $ \gtrsim10^{3}$ to $ < 10^{-3}$ [ @Auchettl2017 ]. Producing such luminous optical emission without meaning ten - ray flux can be explained in one of two ways: either X - ray faint TDEs are powered primarily by stream collision rather than accretion, or the accretion disk discharge is recycle through an air [ @Strubbe2009; @Guillochon2014; @Roth2016 ]. Several lines of evidence have indicate that accretion phonograph record do form promptly even in X - re faint TDEs: Bowen fluorescence lines that require excitation from far - UV photons [ @Blagorodnova2018; @Leloudas2019 ]; low - ionization iron emission appearing shortly after maximum inner light [ @Wevers2019b ]; and recently the direct signal detection of doubly - top out Balmer lines that match predicted disk profiles [ @Short2020; @Hung2020 ]. Thus a critical question is to understand the nature and beginning of the implied reprocessing layer. It has already been established that this cannot be just the unbound debris stream, as the apparent cross - part is too abject to intercept a meaning fraction of the TDE flux [ @Guillochon2014 ]. Inhibiting advancement is the messy geometry of the debris. Colliding streams, inflowing and outflowing gas, and a viewing - slant dependence on both the broad - band [ @Dai2018 ] and spectroscopic [ @Nicholl2019 ] property all contribute to a messy knot that must be untangled. One important

acfording to their spectra, generally exhiyuting uroad, lkw equivxlent width[^3] emission lines oh hyerogeb, neutral and ionised felium, anf Bowen dluocescence lines oh doubly-ljnissf nicrigen and oxygek [e.g. @Gezari2012; @Holoien2014; @Arcavh2014; @Uepoudas2019]. This prompted @vanVelzen2020 to stggest yhgee sub-classes labtllqd TSV-H, TDE-He and TDE-Bowen, though soje TDEs defy a consixtent classification by chwngijg their apparent dpectral tyks af they evolve [@Nicholl2019]. TDT ylares were initially predicted to be brigftest in X-rays, euw tl the high txmperanure of an acggetion gisk, anc indeed this ls thx wacelength where the eacliest TDE candidatef were idancified [@Komossa2002]. Howevee, rhe okticanly-dkwcoxertd VDEa have prkven to be surprisingoy diverse in their X-wqy properties. Their X-way to optical ratios at maximum light gangs from $\gtrsim10^{3}$ to $<10^{-3}$ [@Auchertl2017]. Producing such lulinous opeical emission without significant X-ray flux can te ex'lxintd in ovw lf two ways: either X-ray faint TDEs are powereq ptikarily by strecm collisions rayhfr jhan accretion, or thz adcretion disk emisdion is repricessed trroubh an atmosphere [@Strubbe2009; @Guullochon2014; @Rotk2016]. Seceral lines of evibence have iudicatgd thay accretion disks do foxm projptly even ln X-ray fzknt TDEs: Bowen fuuogescance lintr that require exsitation hrom yar-UV phutonx [@Blagjrodnova2018; @Lfloudas2019]; low-ionisation irln emnssiot appearinh shortly after maximum light [@Wxters2019b]; and recgntny nhe direcc detegtion of double-[eaked Balmer kines tkat magch predicned disk 'rofiles [@Showt2020; @Hung2020]. Thus d critical quxstion is to yndeestand gfe nature and prigin of the impliwd reprocessing laner. Ij gas already beeu tsrablished that thks sajnpt te simply tha uncouvc debfis strtcm, cs the apparent cross-sectimn ia too low to interveit a signuficant sraction of tne TDE flux [@Guillofhon2014]. Iihibitmng prpgrgss is the messy geometry of ths debris. Foljiding streais, ikfloring and obtflowing gas, and a viewing-angle dependeice on both the broad-babd [@Dai2018] and spectroseokic [@Nicholl2019] pcopertyes all cmntribute to a messy knot that must bt untangled. One importaht

according to their spectra, generally exhibiting broad, width[^3] lines of neutral and ionised of nitrogen and oxygen @Gezari2012; @Holoien2014; @Arcavi2014; This prompted @vanVelzen2020 to suggest three labelled TDE-H, TDE-He and TDE-Bowen, though some TDEs defy a consistent classification by their apparent spectral type as they evolve [@Nicholl2019]. TDE flares were initially predicted be in due the high temperature of an accretion disk, and indeed this is the wavelength where the earliest candidates were identified [@Komossa2002]. However, the optically-discovered TDEs proven to be surprisingly in their X-ray properties. Their to ratios at light from to $<10^{-3}$ [@Auchettl2017]. such luminous optical emission without significant X-ray flux can be explained in one of two ways: either faint TDEs primarily by collisions than or the accretion is reprocessed through an atmosphere [@Strubbe2009; lines of evidence have indicated that accretion disks form promptly in X-ray faint TDEs: Bowen fluorescence that require excitation from far-UV photons [@Blagorodnova2018; @Leloudas2019]; iron emission appearing shortly after maximum light [@Wevers2019b]; and recently the direct detection of double-peaked that match predicted disk [@Short2020; @Hung2020]. Thus critical is understand nature and of the implied reprocessing layer. It has already been established that cannot be simply the unbound debris stream, as the apparent too to intercept a fraction of the TDE [@Guillochon2014]. progress is the messy the Colliding outflowing and viewing-angle dependence on both broad-band [@Dai2018] and spectroscopic [@Nicholl2019] all contribute to a untangled. One important

according to their spectra, geNerally exhIbitiNg bRoaD, lOw eqUivaLent width[^3] emissIOn liNes of hydrogen, neutral anD ioniSeD HeliUM, aNd BowEn fluorEScENCe lInEs Of dOuBLy-IonisEd nItrogen And oxygen [e.G. @GeZaRi2012; @Holoien2014; @ArcAVi2014; @leloudas2019]. ThIs pRompted @vanVeLzeN2020 to sugGeSt tHRee suB-clAsses LabellED TDE-H, TdE-He and TDe-BOWen, thoUGh some TdeS dEfy a Consistent classifICaTIon by changing tHeir apPaREnT SPecTraL type as theY eVolve [@nIcholl2019]. Tde fLAREs wERe initially prEdicted to be BRigHtest iN X-RayS, Due to tHe higH tEMpeRature of an aCcreTion disk, aNd indeED this is THe wavelEngth wHerE thE earLIeSt tDE CaNDidATeS weRE idEntified [@koMoSsa2002]. HoWeveR, THE OptiCalLy-diScoveRed TDEs have prOveN to bE SurPrisiNgly dIverSe In theIr X-ray PropeRtIes. Their X-ray to oPticAl ratios aT maXiMum LiGht raNGe from $\GtrSim10^{3}$ To $<10^{-3}$ [@AucheTtl2017]. ProdUCinG sUCH LuMinous optical emissIoN WItHout signIficanT x-rAy FLux can be ExPlaIned IN One of Two wAYs: Either X-rAy fainT tDes Are poweReD primaRiLy bY stReam cOLlisIons raTher than AccreTIon, or the accretIOn disk emissioN Is REPrOCessEd tHrough an atmOsphERe [@StRubbE2009; @guIllOChon2014; @ROth2016]. SeVeRAl LInes of evidence have iNdIcated That aCcretion disks Do form promPTLY even in X-Ray fAInT tDEs: Bowen fluorEscenCe lines thaT Require eXcitaTion from Far-UV photONS [@BlagoroDnoVa2018; @LEloUdaS2019]; LOw-Ionisation iroN EMissIoN appearIng Shortly AftEr mAxiMum LiGht [@Wevers2019B]; and receNtLy ThE dIreCt detECtion of dOuBle-PeAkeD BalmER lines That mAtch PrEdICteD disk prOFiLES [@ShoRt2020; @huNg2020]. ThUs a CrIticaL queSTioN is to unDerstand tHe nATure AnD oRigin of The implied repRoCessing layEr. it hAs alreADY been estAblished that this cannot bE Simply tHe uNbounD debRis stream, As tHe appaRenT Cross-sEction Is too LoW to INTercePT A sIgnIfIcant fractION of The TDe fLux [@GUillochOn2014]. Inhibiting progreSS is The messy geomeTry Of thE DEbRis. cOlLIdiNg STreAMS, inflowing and ouTflowing gaS, aND a Viewing-angLE dePeNdence oN both thE broaD-Band [@Dai2018] And spectrOscopic [@NiChOll2019] pROPerTies all conTribute tO a messy knOT that MUsT be unTanGled. OnE iMpoRtant

according to their spectr a, general ly ex hib iti ng bro ad,low equivalent widt h[^3] emission lines o f hyd ro g en,n eu traland ion i se d hel iu m, an dB ow en fl uor escence lines ofdou bl y-ionised ni t ro gen and ox yge n [e.g. @Gez ari 2012;@H olo i en201 4;@Arca vi2014 ; @Lelo udas2019] .T his pr o mpted @ v a nV elze n2020 to suggestt hr e e sub-classeslabell ed TD E - H,TDE -He and TD E- Bowen , though so m e TDE s defy a consi stent class i fic ationby ch a ngingtheir a p par ent spectra l ty pe as the y evol v e [@Nic h oll2019 ]. TD E f lar es w e re i nit ia l lyp re dic t edto be br ig ht est i n X- r a y s , du e t o th e hig h temperature of ana ccr etion disk , an dindee d this is t he wavelength whe re t he earlie stTD E c an didat e s were id ent ified [ @Komoss a 200 2] . H ow ever, the opticall y- d i sc overed T DEs ha v epr o ven to b esur pris i n gly d iver s ein their X-ray pr op erties. T heir X -r aytooptic a l ra tios a t maximu m lig h t range from $ \ gtrsim10^{3}$ to $ <1 0 ^{-3 }$[@Auchettl2 017] . Pro duci n gsuc h lumi nousop t ic a l emission withoutsi gnific ant X -ray flux can be explai n e d in oneof t w ow ays: either X- ray f aint TDEsa re power ed pr imarilyby stream c ollision s r ath ertha n ac cretion, or t h e acc re tion di skemissio n i s r epr oce ss ed throug h an atm os ph er e[@S trubb e 2009; @G ui llo ch on2 014;@ Roth20 16]. Sev er al lin es of e v id e n ce h av eindi cat ed that acc r eti on disk s do form pr o mptl yev en in X -ray faint TD Es : Bowen fl uo res cencel i nes that require excitation fro m far-UV ph otons [@B lagorodno va2 018; @ Lel o udas20 19]; l ow-io ni sat i o n iro n em iss io n appearin g sho rtlyaf termaximum light [@Wevers201 9 b]; and recently th e di r e ct de t ec t ion o f do u b le-peaked Balme r lines th at ma tch predic t eddi sk prof iles [@ Short 2 020; @H ung2020]. Thus a c ri tica l que stion is t o unders tand then ature an d ori gin of th eimp liedreproc e ssi ng la yer. I thas al ready b een esta blished that this canno t be s imply th e unbound de b ris stream,as t he apparen t c ros s-sec tio n is t oo l o wtoi nterc epta signific a nt fr a c ti on of the T D E flu x [@G uil l ochon2 014] . Inhibiting pro g ress is the me ssyg e ome try of t he debris. Colli din gs t reams, i nf lowing andoutflowi ng gas,and aviewin g-angle d ep e ndence onbot h the bro ad- ba n d [@Dai 20 18 ] and s pect ro scopic [@Nic h oll2 0 1 9] properties al l con t r ibute toa mes sy knot t h at m ust be unt angled. One impor tant

according_to their_spectra, generally exhibiting broad,_low equivalent_width[^3]_emission lines_of_hydrogen, neutral and_ionised helium, and_Bowen fluorescence lines of_doubly-ionised nitrogen and_oxygen_[e.g. @Gezari2012; @Holoien2014; @Arcavi2014; @Leloudas2019]. This prompted @vanVelzen2020 to suggest three sub-classes labelled TDE-H,_TDE-He_and TDE-Bowen,_though_some_TDEs defy a consistent classification_by changing their apparent spectral_type as_they evolve [@Nicholl2019]. TDE flares were initially predicted to_be_brightest in X-rays,_due to the high temperature of an accretion disk,_and indeed this is the wavelength_where the earliest_TDE_candidates_were identified [@Komossa2002]. However,_the optically-discovered TDEs have proven to_be surprisingly diverse in their X-ray_properties. Their X-ray to optical ratios at_maximum light range from $\gtrsim10^{3}$ to_$<10^{-3}$ [@Auchettl2017]. Producing such luminous_optical emission_without significant X-ray flux can_be explained in_one of_two ways: either_X-ray faint TDEs are powered primarily_by stream collisions_rather than accretion, or the accretion_disk_emission is reprocessed_through_an_atmosphere [@Strubbe2009;_@Guillochon2014; @Roth2016]. Several lines_of_evidence have_indicated_that accretion disks do form promptly_even_in X-ray faint TDEs: Bowen fluorescence lines_that require excitation from_far-UV_photons [@Blagorodnova2018; @Leloudas2019]; low-ionisation_iron emission appearing shortly after_maximum light [@Wevers2019b]; and recently the_direct detection_of double-peaked_Balmer lines that match predicted disk profiles [@Short2020; @Hung2020]. Thus a_critical question is to understand the_nature and origin of_the implied_reprocessing_layer. It has_already_been established_that this cannot be simply the unbound_debris stream,_as the apparent cross-section is too_low to intercept a_significant_fraction of the TDE flux [@Guillochon2014]. Inhibiting_progress is the messy geometry of_the debris. Colliding streams, inflowing_and_outflowing_gas, and a viewing-angle dependence_on both the broad-band [@Dai2018] and_spectroscopic [@Nicholl2019] properties_all contribute to a messy knot that_must_be untangled. One important

_g C = + \Omega \,C, \qquad s_g \bar C = - \Omega \,\bar C, \qquad s_g \beta = + \Omega \,\beta, \qquad s_g \bar \beta = - \Omega \bar \beta, \nonumber\\ && s_g\big(A_\mu, \phi_\mu, \rho, B, \bar B, R, \bar R\big) =0. \label{6.2}\end{aligned}$$ It is straightforward to check that under the above continuous ghost scale symmetry transformations (\[6.2\]) both the Lagrangian densities remain invariant (i.e. $s_g{\cal L}_b = s_g {\cal L}_{\bar b} = 0$). As a consequence, the existence of ghost scale symmetry leads to the following Noether’s conserved current ($J^\mu_g$) and charge ($Q_g$): $$\begin{aligned} J^\mu_g &=& i \Big[\bar C \cdot (D^\mu C) - (\partial^\mu \bar C) \cdot C + \bar \beta \cdot (D^\mu \beta) - (D^\mu \bar \beta) \cdot \beta \Big], \nonumber\\ Q_g &=& i \int d^2x \,\Big[\bar C \cdot (D^0 C) - (\partial^0 \bar C) \cdot C + \bar \beta \cdot (D^0 \beta) - (D^0 \bar \beta) \cdot \beta \Big].\end{aligned}$$ The conservation law $(\partial_\mu J^\mu_g = 0)$ can be proven by exploiting the E-L equations of motion (\[5.2\]). The ghost charge $Q_g$ also turns out to be the generator of the ghost scale symmetry transformations (\[6.2\]). For instance, one can check that $s_g C = - i [C, \, \Omega\,Q_g] = +\Omega\, C.$ The above ghost charge $Q_g$ together with the nilpotent (anti-)BRST charges $Q_{(a)b}$ obey a standard BRST algebra. In operator form, this algebra can be given as follow $$\begin{aligned} && Q^2_b = 0, \qquad Q^2_{ab

_ g C = + \Omega \,C, \qquad s_g \bar C = - \Omega \,\bar C, \qquad s_g \beta = + \Omega \,\beta, \qquad s_g \bar \beta = - \Omega \bar \beta, \nonumber\\ & & s_g\big(A_\mu, \phi_\mu, \rho, B, \bar B, R, \bar R\big) = 0. \label{6.2}\end{aligned}$$ It is straightforward to check that under the above continuous ghostwriter plate symmetry transformations (\[6.2\ ]) both the Lagrangian density stay invariant (i.e. $ s_g{\cal L}_b = s_g { \cal L}_{\bar b } = 0 $). As a consequence, the existence of touch plate symmetry leads to the following Noether ’s conserved current ($ J^\mu_g$) and charge ($ Q_g$ ): $ $ \begin{aligned } J^\mu_g & = & i \Big[\bar C \cdot (D^\mu C) - (\partial^\mu \bar C) \cdot C + \bar \beta \cdot (D^\mu \beta) - (D^\mu \bar \beta) \cdot \beta \Big ], \nonumber\\ Q_g & = & i \int d^2x \,\Big[\bar C \cdot (D^0 C) - (\partial^0 \bar C) \cdot C + \bar \beta \cdot (D^0 \beta) - (D^0 \bar \beta) \cdot \beta \Big].\end{aligned}$$ The conservation police $ (\partial_\mu J^\mu_g = 0)$ can be proven by exploiting the einsteinium - fifty equations of motion (\[5.2\ ]). The ghostwriter charge $ Q_g$ also sour out to be the generator of the ghost scale symmetry transformation (\[6.2\ ]). For instance, one can check that $ s_g C = - i [ C, \, \Omega\,Q_g ] = + \Omega\, C.$ The above ghost tear $ Q_g$ together with the nilpotent (anti-)BRST charges $ Q_{(a)b}$ obey a standard BRST algebra. In operator form, this algebra can be given as comply $ $ \begin{aligned } & & Q^2_b = 0, \qquad Q^2_{ab

_g C = + \Omega \,C, \qquad s_g \bar G = - \Omega \,\bar C, \qquad v_g \betz = + \Omeea \,\beta, \qquad s_g \bar \beta = - \Onega \var \beta, \nonumber\\ && s_g\bkg(A_\mu, \phi_\lu, \rho, B, \bar V, R, \bar R\bmf) =0. \label{6.2}\end{aljnned}$$ Nt is straightfotward to chewk that under dhd cbove continuous ghost scale symmetrr transgogmations (\[6.2\]) both the Jagrzngian densities remain invariant (j.e. $s_g{\can L}_b = s_g {\cal K}_{\bar b} = 0$). As a consequence, hhe fxistence of ghost scale symmgfry oeads to the following Noether’s cohserved current ($J^\mu_g$) and charge ($Q_g$): $$\bzgin{aligned} H^\my_g &=& h \Big[\bar C \rdot (D^\iu C) - (\partial^\mu \bar W) \cdot V + \bar \beta \cdpt (V^\mu \veta) - (D^\mu \bar \beta) \cvot \beta \Big], \nonumbet\\ Q_g &=& i \int d^2r \,\Big[\bar C \cdot (D^0 C) - (\paetial^0 \bar C) \cait Z + \gac \bsta \cdlt (V^0 \beta) - (D^0 \gar \beta) \cdit \beta \Big].\end{aligntd}$$ Eye conservatikn law $(\pwrtial_\mu J^\mu_g = 0)$ can be proven by exploiuing fhe E-L equations of motuon (\[5.2\]). The ghost charge $Q_g$ also eurns out to be the generator of the ghost scale vymmevrh txqnsfofnahions (\[6.2\]). For instance, one can check that $s_g C = - j [V, \, \Omega\,Q_g] = +\Omegc\, C.$ The above ghoxt cnwrge $Q_g$ togetfer wicg fhe nilpotent (anti-)HRST chwrges $Q_{(a)b}$ obey a syandard BRST algebra. In opeeator form, tkis algebra can be giren as folloc $$\begim{aligmed} && Q^2_b = 0, \qquad Q^2_{ab

_g C = + \Omega \,C, \qquad C - \Omega C, \qquad s_g \qquad \bar \beta = \Omega \bar \beta, && s_g\big(A_\mu, \phi_\mu, \rho, B, \bar R, \bar R\big) =0. \label{6.2}\end{aligned}$$ It is straightforward to check that under the continuous ghost scale symmetry transformations (\[6.2\]) both the Lagrangian densities remain invariant (i.e. L}_b s_g L}_{\bar = 0$). As a consequence, the existence of ghost scale symmetry leads to the following Noether’s current ($J^\mu_g$) and charge ($Q_g$): $$\begin{aligned} J^\mu_g &=& \Big[\bar C \cdot (D^\mu - (\partial^\mu \bar C) \cdot + \beta \cdot \beta) (D^\mu \beta) \cdot \beta \nonumber\\ Q_g &=& i \int d^2x \,\Big[\bar C \cdot (D^0 C) - (\partial^0 \bar C) \cdot C \bar \beta \beta) - \bar \cdot \Big].\end{aligned}$$ The conservation J^\mu_g = 0)$ can be proven E-L equations of motion (\[5.2\]). The ghost charge also turns to be the generator of the scale symmetry transformations (\[6.2\]). For instance, one can that $s_g C = - i [C, \, \Omega\,Q_g] = +\Omega\, C.$ The above ghost together with the nilpotent charges $Q_{(a)b}$ obey standard algebra. operator this algebra be given as follow $$\begin{aligned} && Q^2_b = 0, \qquad Q^2_{ab

_g C = + \Omega \,C, \qquad s_g \bar C = - \Omega \,\bAr C, \qquad s_g \Beta = + \OMegA \,\beTa, \QquaD s_g \bAr \beta = - \Omega \bar \BEta, \nOnumber\\ && s_g\big(A_\mu, \phi_\mu, \rhO, B, \bar b, R, \BAr R\bIG) =0. \lAbel{6.2}\eNd{alignED}$$ IT IS stRaIgHtfOrWArD to chEck That undEr the above ConTiNuous ghost scALe Symmetry trAnsFormations (\[6.2\]) boTh tHe LagrAnGiaN DensiTieS remaIn invaRIant (i.e. $S_g{\cal L}_b = s_g {\CaL l}_{\bar b} = 0$). AS A conseqUENcE, the Existence of ghost sCAlE Symmetry leads tO the foLlOWiNG noeTheR’s conserveD cUrrenT ($j^\mu_g$) and CHaRGE ($q_g$): $$\bEGin{aligned} J^\mu_G &=& i \Big[\bar C \cdOT (D^\mU C) - (\partIaL^\mu \BAr C) \cdoT C + \bar \BeTA \cdOt (D^\mu \beta) - (D^\mU \bar \Beta) \cdot \bEta \Big], \NOnumber\\ q_G &=& i \int d^2x \,\big[\bar c \cdOt (D^0 c) - (\parTIaL^0 \bAr C) \CdOT C + \bAR \bEta \CDot (d^0 \beta) - (D^0 \baR \bEtA) \cdot \Beta \bIG].\ENd{alIgnEd}$$ ThE consErvation law $(\paRtiAl_\mu j^\Mu_g = 0)$ Can be ProveN by eXpLoitiNg the E-l equaTiOns of motion (\[5.2\]). The gHost Charge $Q_g$ aLso TuRns OuT to be THe geneRatOr oF the ghoSt scale SYmmEtRY TRaNsformations (\[6.2\]). For insTaNCE, oNe can cheCk that $S_G C = - I [C, \, \oMega\,Q_g] = +\OmEgA\, C.$ THe abOVE ghosT chaRGe $q_g$ togethEr with THe NiLpotent (AnTi-)BRST ChArgEs $Q_{(A)b}$ obeY A staNdard BrST algebRa. In oPErator form, this ALgebra can be giVEn AS FoLLow $$\bEgiN{aligned} && Q^2_b = 0, \qQuad q^2_{Ab

_g C = + \Omega \,C, \qqua d s_g \bar C =- \Om eg a \, \bar C, \qquad s_g \bet a = + \Omega \,\beta, \qq ua d s_g \b ar \b eta = - \ O m ega \ ba r \ be t a, \non umb er\\ && s_g\big( A_\ mu , \phi_\mu,\ rh o, B, \bar B, R, \bar R\b ig) =0. \ la bel { 6.2}\ end {alig ned}$$ It isstraightf or w ard to check t h a tunde r the above conti n uo u s ghost scalesymmet ry tr a n sfo rma tions (\[6 .2 \]) b o th theL ag r a n gia n densities re main invari a nt(i.e.$s _g{ \ cal L} _b =s_ g {\ cal L}_{\ba r b} = 0$). A s a co n sequenc e , the e xisten ceofghos t s ca lesy m met r ylea d s t o the fo ll ow ing N oeth e r ’ s con ser vedcurre nt ($J^\mu_g$ ) a nd c h arg e ($Q _g$): $$\ be gin{a ligned } J^\ mu _g &=& i \Big[\ barC \cdot ( D^\ mu C) - (\pa r tial^\ mu\ba r C) \c dot C + \b ar \ b et a \cdot (D^\mu \be ta ) - (D^\mu\bar \ b et a) \cdot \b et a \ Big] , \nonu mber \ \Q_g &=&i \int d^ 2x \,\Big [\ bar C\c dot (D ^0 C) - (\ partia l^0 \bar C) \ c dot C + \bar\ beta \cdot (D ^ 0\ b et a ) - (D ^0 \bar \be ta)\ cdot \be t a\Bi g ].\en d{ali gn e d} $ $ The conservationla w $(\p artia l_\mu J^\mu_g = 0)$ can b e provenby e x pl o iting the E-Lequat ions of mo t ion (\[5 .2\]) . The gh ost charg e $Q_g$ al sotur nsout t obe the genera t o r of t he ghos t s cale sy mme try tr ans fo rmations(\[6.2\] ). F or i nst ance, one canch eck t hat $s_g C = -i [C, \,\O me g a\, Q_g] =+ \O m e ga\, C .$ Th e a bo ve gh ostc har ge $Q_g $ togethe r w i th t he n ilpoten t (anti-)BRST c harges $Q_ {( a)b }$ obe y a standa rd BRST algebra. In ope r ator fo rm, this alg ebra canbegivenasf ollow$$\beg in{al ig ned } && Q^ 2 _ b= 0 ,\qquad Q^2 _ { ab

_g C_= +_\Omega \,C, \qquad s_g_\bar C_=_- _\Omega_\,\bar C, \qquad_s_g \beta =_+ \Omega \,\beta,_ \qquad s_g \bar_\beta_= - \Omega \bar \beta, \nonumber\\ && s_g\big(A_\mu, \phi_\mu, \rho, B, \bar B,_R,_\bar R\big)_=0._\label{6.2}\end{aligned}$$_It is straightforward to check_that under the above continuous_ghost scale_symmetry transformations (\[6.2\]) both the Lagrangian densities remain_invariant_(i.e. $s_g{\cal L}_b_= s_g {\cal L}_{\bar b} = 0$). As a_consequence, the existence of ghost scale_symmetry leads to_the_following_Noether’s conserved current ($J^\mu_g$)_and charge ($Q_g$): $$\begin{aligned} J^\mu_g &=& i_\Big[\bar C \cdot (D^\mu C) -_(\partial^\mu \bar C) \cdot C + \bar_\beta \cdot (D^\mu \beta) - _(D^\mu \bar \beta) \cdot \beta_\Big], \nonumber\\ Q_g_&=& i \int d^2x \,\Big[\bar_C \cdot (D^0_C) -_(\partial^0 \bar C)_\cdot C + \bar \beta \cdot_(D^0 \beta) -_ (D^0 \bar \beta) \cdot \beta_\Big].\end{aligned}$$_The conservation law_$(\partial_\mu_J^\mu_g_= 0)$_can be proven_by_exploiting the_E-L_equations of motion (\[5.2\]). The ghost_charge_$Q_g$ also turns out to be the_generator of the ghost_scale_symmetry transformations (\[6.2\]). For_instance, one can check that_$s_g C = - i [C,_\, \Omega\,Q_g]_= +\Omega\,_C.$ The above ghost charge $Q_g$ together with the nilpotent (anti-)BRST charges_$Q_{(a)b}$ obey a standard BRST algebra._In operator form, this_algebra can_be_given as follow_$$\begin{aligned} &&_Q^2_b =_0, \qquad Q^2_{ab

rm g~cm^{-3}}$ (right panel) as a function of time in the $q=0.7$ non-AMR simulation (green curve), AMR simulation (red curve), SPH simulation (blue curve), and SPH simulation mapped to a grid (purple curve).[]{data-label="q07amrcofrac"}](f4.pdf){width="50.00000%"} Hybrid accretor, $M_{\rm tot}=0.71 M_\odot$, $q=0.5$ {#q05subsection} ---------------------------------------------------- ![High resolution non-AMR grid-based hybrid simulation with $q=0.5$, showing equatorial slices in the left column and slices perpendicular to this in the right column. The top row shows logarithm of density, the middle row shows temperature, and the bottom row shows the accretor mass fraction. The blob is clearly visible to the left of the core in the equatorial plots. The perpendicular slices in the right column stretches from the center of the grid to $5\times10^9~{\rm cm}$ in the radial direction, and from $-2\times10^9$ to $2\times10^9~{\rm cm}$ in the vertical direction, and it is made through the blob.[]{data-label="jan05hybrid"}](f5.pdf){width="85.00000%"} In the high resolution non-AMR grid simulation, we find that $\sim0.04 M_\odot$ accretor material is at $\rho<10^{5.2}~{\rm g/cm^3}$ ($\sim 0.02 M_\odot$ at $\rho<10^5~{\rm %0.024 Msun actually g/cm^3}$). If we assume that the accretor consists of equal amounts of $^{16}\mathrm{O}$ and $^{12}\mathrm{C}$, then half of this is $^{16}\mathrm{O}$. We also find that this amount keeps increasing with time past the merger event, indicating that there is some artificial diffusion of the mass fractions in our non-AMR grid-based simulations. In the merged core, we find densities of $\sim1.7\times10^6~{\rm g/cm^3}$, and temperatures in the SOF reaches $1.6\times10^8~{\rm K}$ in the high resolution simulation. For comparison, the amount of accretor mass below

rm g ~ cm^{-3}}$ (right panel) as a function of time in the $ q=0.7 $ non - AMR model (fleeceable curve), AMR simulation (crimson bend), SPH simulation (blue bend), and SPH pretense mapped to a power system (empurpled curve).[]{data - label="q07amrcofrac"}](f4.pdf){width="50.00000% " } Hybrid accretor, $ M_{\rm tot}=0.71 M_\odot$, $ q=0.5 $ { # q05subsection } ---------------------------------------------------- ! [ High solution non - AMR grid - based hybrid pretense with $ q=0.5 $, indicate equatorial slices in the left column and slices perpendicular to this in the right column. The top quarrel shows logarithm of density, the middle row show temperature, and the bottom row shows the accretor mass fraction. The blob is intelligibly visible to the left of the core in the equatorial plots. The perpendicular slices in the correct column stretches from the center of the grid to $ 5\times10 ^ 9~{\rm cm}$ in the radial direction, and from $ -2\times10 ^ 9 $ to $ 2\times10 ^ 9~{\rm cm}$ in the vertical commission, and it is seduce through the blob.[]{data - label="jan05hybrid"}](f5.pdf){width="85.00000% " } In the high resolution non - AMR grid simulation, we find that $ \sim0.04 M_\odot$ accretor material is at $ \rho<10^{5.2}~{\rm g / cm^3}$ ($ \sim 0.02 M_\odot$ at $ \rho<10 ^ 5~{\rm % 0.024 Msun actually g / cm^3}$). If we assume that the accretor consists of adequate sum of $ ^{16}\mathrm{O}$ and $ ^{12}\mathrm{C}$, then half of this is $ ^{16}\mathrm{O}$. We besides witness that this amount keeps increasing with meter past the merger event, indicating that there is some artificial dissemination of the mass fractions in our non - AMR grid - based simulations. In the merged congress of racial equality, we find densities of $ \sim1.7\times10 ^ 6~{\rm g / cm^3}$, and temperatures in the SOF reaches $ 1.6\times10 ^ 8~{\rm K}$ in the eminent resolution model. For comparison, the amount of accretor mass below

rm h~cm^{-3}}$ (right panel) as a fungtion of time in the $q=0.7$ ion-AMR aimulatiun (green curve), AMR simulatioi (ree curce), SPH simulation (blue curve), anf SPH sinulauion mapped to a jdid (purije chvve).[]{daca-oabel="q07amrcofrag"}](f4.pdf){width="50.00000%"} Hfbrid accretor, $M_{\fm tot}=0.71 M_\odot$, $q=0.5$ {#q05subsection} ---------------------------------------------------- ![High resolueion nom-ALR grid-based hibrid fimumation with $q=0.5$, showing equatorial smices ii the left colukn and slices perpendiculag to this in the right column. The top eow shows loearithm of density, the middle row shows temperature, avd thz bottom roq whoav the accrevor mafs fraction. Bne blot is clrarly visible bo thx ledt of the core in the equatorial plots. Jhe perpengieular slices in the rugyt conumn strdrchds rrpm the cfntxr of the gdid to $5\timew10^9~{\rm cm}$ in the radiak qptection, and rrom $-2\tymqs10^9$ to $2\times10^9~{\rm cm}$ in the vertical directpon, znd it is made through rhe blob.[]{data-label="jan05hibrid"}](f5.pdf){wydth="85.00000%"} In the high resolution non-AMR grid simulation, we fmna tkqt $\sio0.04 M_\ldot$ accretor material is at $\rho<10^{5.2}~{\rm g/cm^3}$ ($\sim 0.02 M_\jsou$ an $\rho<10^5~{\rm %0.024 Msun actuclly g/cm^3}$). If we asxule jhat the accrejor consisfs of equal amountd of $^{16}\majhrm{O}$ qnd $^{12}\mathri{C}$, tnen half of this is $^{16}\mathrm{O}$. We also finb tyat this amount kezps increasiug witn timr past the merger event, indjcating thah there ia some artificial dinfuvion of ufe mass fractions in our nin-AMX grid-bared ximulaeions. In tje mevced core, we find dfnsitnes ox $\sim1.7\times10^6~{\gm g/cm^3}$, and temperatures in the SOF reaches $1.6\tikev10^8~{\rm K}$ in thz high resolution symulation. For cpmparisjn, thd amount or accrevor mass beljw

rm g~cm^{-3}}$ (right panel) as a function in $q=0.7$ non-AMR (green curve), AMR (blue and SPH simulation to a grid curve).[]{data-label="q07amrcofrac"}](f4.pdf){width="50.00000%"} Hybrid accretor, $M_{\rm tot}=0.71 M_\odot$, {#q05subsection} ---------------------------------------------------- ![High resolution non-AMR grid-based hybrid simulation with $q=0.5$, showing equatorial slices the left column and slices perpendicular to this in the right column. The row logarithm density, middle row shows temperature, and the bottom row shows the accretor mass fraction. The blob is visible to the left of the core in equatorial plots. The perpendicular in the right column stretches the of the to cm}$ the radial direction, from $-2\times10^9$ to $2\times10^9~{\rm cm}$ in the vertical direction, and it is made through the blob.[]{data-label="jan05hybrid"}](f5.pdf){width="85.00000%"} In high resolution simulation, we that M_\odot$ material is at ($\sim 0.02 M_\odot$ at $\rho<10^5~{\rm %0.024 If we assume that the accretor consists of amounts of and $^{12}\mathrm{C}$, then half of this $^{16}\mathrm{O}$. We also find that this amount keeps with time past the merger event, indicating that there is some artificial diffusion of the in our non-AMR grid-based In the merged we densities $\sim1.7\times10^6~{\rm and temperatures the SOF reaches $1.6\times10^8~{\rm K}$ in the high resolution simulation. For the amount of accretor mass below

rm g~cm^{-3}}$ (right panel) as a functioN of time in tHe $q=0.7$ noN-AMr siMuLatiOn (grEen curve), AMR simULatiOn (red curve), SPH simulatioN (blue CuRVe), anD sPh simuLation mAPpED To a GrId (PurPlE CuRve).[]{daTa-lAbel="q07amRcofrac"}](f4.pdF){wiDtH="50.00000%"} Hybrid accreTOr, $m_{\rm tot}=0.71 M_\odoT$, $q=0.5$ {#q05Subsection} ---------------------------------------------------- ![HiGh rEsolutIoN noN-aMR grId-bAsed hYbrid sIMulatiOn with $q=0.5$, shOwINg equaTOrial slICEs In thE left column and sliCEs PErpendicular to This in ThE RiGHT coLumN. The top row ShOws loGArithm oF DeNSITy, tHE middle row shoWs temperatuRE, anD the boTtOm rOW shows The acCrETor Mass fractioN. The Blob is cleArly viSIble to tHE left of The corE in The EquaTOrIaL plOtS. the PErPenDIcuLar sliceS iN tHe rigHt coLUMN StreTchEs frOm the Center of the grId tO $5\timES10^9~{\rm Cm}$ in tHe radIal dIrEctioN, and frOm $-2\timEs10^9$ To $2\times10^9~{\rm cm}$ in thE verTical direCtiOn, And It Is madE ThrougH thE blOb.[]{data-lAbel="jan05HYbrId"}](F5.PDF){wIdth="85.00000%"} In the high resolUtION nOn-AMR griD simulATiOn, WE find thaT $\sIm0.04 M_\Odot$ ACCretoR matERiAl is at $\rhO<10^{5.2}~{\rm g/cm^3}$ ($\SIm 0.02 m_\oDot$ at $\rhO<10^5~{\rM %0.024 Msun aCtUalLy g/Cm^3}$). If wE AssuMe that The accreTor coNSists of equal amOUnts of $^{16}\mathrm{O}$ ANd $^{12}\MAThRM{C}$, thEn hAlf of this is $^{16}\MathRM{O}$. We Also FInD thAT this AmounT kEEpS Increasing with time pAsT the meRger eVent, indicatinG that there IS SOme artifIciaL DiFFusion of the masS fracTions in our NOn-AMR griD-baseD simulatIons. In the MERged core, We fInd DenSitIES oF $\sim1.7\times10^6~{\rm g/cM^3}$, ANd teMpEratureS in The SOF rEacHes $1.6\TimEs10^8~{\rM K}$ In the high ResolutiOn SiMuLaTioN. For cOMparison, ThE amOuNt oF accrETor masS beloW

rm g~cm^{-3}}$ (right pane l) as a fu nctio n o f t im e in the $q=0.7$ non-A M R si mulation (green curve) , AMR s i mula t io n (re d curve ) ,S P H s im ul ati on (b lue c urv e), and SPH simul ati on mapped to a gr id (purple cu rve).[]{data -la bel="q 07 amr c ofrac "}] (f4.p df){wi d th="50 .00000%"} H ybrida ccretor , $M _{\r m tot}=0.71 M_\od o t$ , $q=0.5$ {#q05 subsec ti o n} - --- --- ---------- -- ----- - ------- - -- - - - --- - ---------- ! [High resol u tio n non- AM R g r id-bas ed hy br i d s imulation w ith$q=0.5$,showin g equato r ial sli ces in th e l eftc ol um n a nd sli c es pe r pen dicularto t his i n th e r i ghtcol umn. Thetop row shows lo gari t hmof de nsity , th emiddl e rowshows t emperature, and the bottom r owsh ows t he ac c retormas s f raction . The b l obis c l ea rly visible to the l e f tof the c ore in th ee quatoria lplo ts.T h e per pend i cu lar slic es int he r ight co lu mn str et che s f rom t h e ce nter o f the gr id to $5\times10^9~{ \ rm cm}$ in th e r a d ia l dir ect ion, and fr om $ - 2\ti mes1 0 ^9 $ t o $2\t imes1 0^ 9 ~{ \ rm cm}$ in the vert ic al dir ectio n, and it ismade throu g h the blob .[]{ d at a -label="jan05h ybrid "}](f5.pdf ) {width=" 85.00 000%"} In the hi g h resolut ion no n-A MRg r id simulation,w e fin dthat $\ sim 0.04 M_ \od ot$ ac cre to r materia l is at$\ rh o< 10 ^{5 .2}~{ \ rm g/cm^ 3} $ ( $\ sim 0.02 M_\odo t$ at $\r ho <1 0 ^5~ {\rm %0 . 02 4 Msun a ct uall y g /c m^3}$ ). I f we assume that the ac c reto rco nsistsof equal amou nt s of $^{16 }\ mat hrm{O} $ and $^{1 2}\mathrm{C}$, then hal f of thi s i s $^{ 16}\ mathrm{O} $.We als o f i nd tha t this amou nt ke e p s inc r e as ing w ith time p a s t t he me rg er e vent, i ndicating that the r e i s some artifi cia l di f f us ion of the m a ssf r actions in ournon-AMR gr id - ba sed simula t ion s. In the merged core , we fin d densiti es of $\s im 1.7\ t i mes 10^6~{\rmg/cm^3}$ , and tem p eratu r es in t heSOF re ac hes $1.6 \times 1 0^8 ~{\rm K}$ i nthe hi gh re so lution s imulation. For compari son, t he am oun t of accr eto r ma ss below

rm g~cm^{-3}}$_(right panel)_as a function of_time in_the_$q=0.7$ non-AMR_simulation_(green curve), AMR_simulation (red curve),_SPH simulation (blue curve),_and SPH simulation_mapped_to a grid (purple curve).[]{data-label="q07amrcofrac"}](f4.pdf){width="50.00000%"} Hybrid accretor, $M_{\rm tot}=0.71 M_\odot$, $q=0.5$ {#q05subsection} ---------------------------------------------------- ![High resolution non-AMR grid-based_hybrid_simulation with_$q=0.5$,_showing_equatorial slices in the left_column and slices perpendicular to_this in_the right column. The top row shows logarithm_of_density, the middle_row shows temperature, and the bottom row shows the_accretor mass fraction. The blob is_clearly visible to_the_left_of the core in_the equatorial plots. The perpendicular slices_in the right column stretches from_the center of the grid to $5\times10^9~{\rm_cm}$ in the radial direction, and_from $-2\times10^9$ to $2\times10^9~{\rm cm}$_in the_vertical direction, and it is_made through the_blob.[]{data-label="jan05hybrid"}](f5.pdf){width="85.00000%"} In the_high resolution non-AMR_grid simulation, we find that $\sim0.04_M_\odot$ accretor material_is at $\rho<10^{5.2}~{\rm g/cm^3}$ ($\sim 0.02_M_\odot$_at $\rho<10^5~{\rm %0.024 Msun_actually g/cm^3}$)._If_we assume_that the accretor_consists_of equal_amounts_of $^{16}\mathrm{O}$ and $^{12}\mathrm{C}$, then half_of_this is $^{16}\mathrm{O}$. We also find that_this amount keeps increasing_with_time past the merger_event, indicating that there is_some artificial diffusion of the mass_fractions in_our non-AMR_grid-based simulations. In the merged core, we find densities of $\sim1.7\times10^6~{\rm_g/cm^3}$, and temperatures in the SOF_reaches $1.6\times10^8~{\rm K}$ in_the high_resolution_simulation. For comparison, the_amount_of accretor_mass below

but they are one-point spaces since $\bp \equiv 0$ (and $\Sigma \v$ cf.  \[t3\].) In all other cases, $\bp>0$ and, hence, $(H \setminus \Sigma, d_{\bp})$ and $(H \setminus \Sigma, d_{\bp^*})$ are homeomorphic to $(H \setminus \Sigma, g_H)$. Finally, we consider the Gromov boundaries of these hyperbolic spaces. Let us denote the one-point compactification of a hypersurface $H \in {\cal{H}}^{\R}_n$ by $\widehat{H}$. For the singular set $\Sigma_H$ of some $H \in {\cal{H}}^{\R}_n$ we *always* add $\infty_H$ to $\Sigma$ and define $\widehat\Sigma:=\Sigma \cup \infty_H$ (note that $\Sigma$ could already be compact). For $H \in {\cal{G}}^{c}_n$ we set $\widehat H=H$ and $\widehat\Sigma=\Sigma$. In Section \[grch\] we prove the following theorem rendering $\widehat\Sigma$ as the Gromov boundary of the hyperbolic unfoldings of $(H \setminus \Sigma, g_H)$. \[thm4\] For any non-totally geodesic $H \in {\cal{G}}$ the identity map on $H \setminus \Sigma$ extends to homeomorphisms between the one-point compactification $\widehat{H}$ and the Gromov compactifications $\overline{X}_G$ of $X=(H \setminus \Sigma,d_{\bp})$, $(H \setminus \Sigma,d_{\bp^*})$ and $(H \setminus \Sigma, k_{H \setminus \Sigma})$: $$\widehat{H}\cong\overline{(H \setminus \Sigma,d_{\bp})}_G \cong \overline{(H \setminus \Sigma,d_{\bp^*})}_G \cong \overline{(H \setminus \Sigma, k_{H \setminus \Sigma})}_G,$$ where $ \cong$ means homeomorphic. In particular, we find for the Gromov boundaries $\p_G(X)$: $$\widehat{\Sigma} \cong\p_G(H \setminus \Sigma,d_{\bp}) \cong \p_G(H \setminus \Sigma,d_{\bp^*}) \cong \p_G(H \setminus \Sigma

but they are one - point spaces since $ \bp \equiv 0 $ (and $ \Sigma \v$ cf.   \[t3\ ] .) In all other case, $ \bp>0 $ and, therefore, $ (H \setminus \Sigma, d_{\bp})$ and $ (H \setminus \Sigma, d_{\bp^*})$ are homeomorphic to $ (H \setminus \Sigma, g_H)$. Finally, we consider the Gromov limit of these hyperbolic space. Let us denote the one - compass point compactification of a hypersurface $ H \in { \cal{H}}^{\R}_n$ by $ \widehat{H}$. For the singular set up $ \Sigma_H$ of some $ H \in { \cal{H}}^{\R}_n$ we * constantly * add $ \infty_H$ to $ \Sigma$ and define $ \widehat\Sigma:=\Sigma \cup \infty_H$ (notice that $ \Sigma$ could already be compact). For $ H \in { \cal{G}}^{c}_n$ we set $ \widehat H = H$ and $ \widehat\Sigma=\Sigma$. In Section \[grch\ ] we rise the following theorem rendering $ \widehat\Sigma$ as the Gromov boundary of the hyperbolic unfoldings of $ (H \setminus \Sigma, g_H)$. \[thm4\ ] For any non - totally geodetic $ H \in { \cal{G}}$ the identity map on $ H \setminus \Sigma$ run to homeomorphisms between the one - point compactification $ \widehat{H}$ and the Gromov compactifications $ \overline{X}_G$ of $ X=(H \setminus \Sigma, d_{\bp})$, $ (H \setminus \Sigma, d_{\bp^*})$ and $ (H \setminus \Sigma, k_{H \setminus \Sigma})$: $ $ \widehat{H}\cong\overline{(H \setminus \Sigma, d_{\bp})}_G \cong \overline{(H \setminus \Sigma, d_{\bp^*})}_G \cong \overline{(H \setminus \Sigma, k_{H \setminus \Sigma})}_G,$$ where $ \cong$ means homeomorphic. In especial, we find for the Gromov boundary $ \p_G(X)$: $ $ \widehat{\Sigma } \cong\p_G(H \setminus \Sigma, d_{\bp }) \cong \p_G(H \setminus \Sigma, d_{\bp^ * }) \cong \p_G(H \setminus \Sigma

buh they are one-point spacts since $\bp \equiv 0$ (and $\Smgma \v$ df.  \[t3\].) In xll other cases, $\bp>0$ and, hence, $(H \setmunus \Sigma, d_{\bp})$ and $(H \sdtminus \Spgma, d_{\bp^*})$ qre iomeomorphic to $(I \setminmf \Sifla, g_K)$. Fmnally, we consicer the Grmmov boundariev uf these hyperbolic spaces. Let us denjte the oje-point compacjificseion of a hypersurface $H \in {\cal{H}}^{\R}_n$ by $\sidehat{I}$. For the singukar set $\Sigma_H$ of some $H \ij {\cap{H}}^{\R}_n$ we *always* add $\infty_H$ to $\Wigmw$ and define $\didehat\Sigma:=\Sigma \cup \jnfty_H$ (note that $\Sigma$ could alfeady be compacj). Rog $H \in {\cal{G}}^{c}_i$ we svt $\widehat H=H$ and $\widahat\Sigka=\Sigma$. In Secbion \[jrch\] we prove the followiig theorem rendering $\widehat\Shgja$ as the Gromov voyndari of dhe ftpefbomir uhfoldijgs of $(H \setmjnus \Sigma, t_H)$. \[thm4\] For any non-toualjj geodesic $H \jn {\cal{D}}$ ehe identity map on $H \setminus \Sigma$ exuends to homeomorphisms betwwen the one-point compwctificatyon $\widehat{H}$ and the Gromov compactifications $\ovesline{E}_G$ of $X=(M \segnijus \Sigma,d_{\bp})$, $(H \setminus \Sigma,d_{\bp^*})$ and $(H \setminta \Xinma, k_{H \setminus \Figma})$: $$\widehst{J}\cpgg\overline{(H \sgtminus \Sifma,d_{\bp})}_G \cong \overllne{(H \sejminus \Sigma,d_{\bp^*})}_D \comg \overline{(H \setminus \Sigma, k_{H \setminus \Wigma})}_G,$$ where $ \cong$ means homeumorkhic. Im particular, we find fox the Fromov bounfaries $\p_G(S)$: $$\widehat{\Sigma} \covg\p_N(H \vetminus \Sigma,d_{\bp}) \cong \p_G(H \setminus \Sigka,d_{\bp^*}) \cung \k_G(H \setiinus \Sigmw

but they are one-point spaces since $\bp (and \v$ cf. In all other \setminus d_{\bp})$ and $(H \Sigma, d_{\bp^*})$ are to $(H \setminus \Sigma, g_H)$. Finally, consider the Gromov boundaries of these hyperbolic spaces. Let us denote the one-point of a hypersurface $H \in {\cal{H}}^{\R}_n$ by $\widehat{H}$. For the singular set $\Sigma_H$ some \in we add $\infty_H$ to $\Sigma$ and define $\widehat\Sigma:=\Sigma \cup \infty_H$ (note that $\Sigma$ could already be compact). $H \in {\cal{G}}^{c}_n$ we set $\widehat H=H$ and In Section \[grch\] we the following theorem rendering $\widehat\Sigma$ the boundary of hyperbolic of \setminus \Sigma, g_H)$. For any non-totally geodesic $H \in {\cal{G}}$ the identity map on $H \setminus \Sigma$ extends to homeomorphisms the one-point and the compactifications of \setminus \Sigma,d_{\bp})$, $(H and $(H \setminus \Sigma, k_{H \setminus \Sigma,d_{\bp})}_G \cong \overline{(H \setminus \Sigma,d_{\bp^*})}_G \cong \overline{(H \setminus k_{H \setminus where $ \cong$ means homeomorphic. In we find for the Gromov boundaries $\p_G(X)$: $$\widehat{\Sigma} \setminus \Sigma,d_{\bp}) \cong \p_G(H \setminus \Sigma,d_{\bp^*}) \cong \p_G(H \setminus \Sigma

but they are one-point spaces sInce $\bp \equiV 0$ (and $\SIgmA \v$ cF.  \[t3\].) in alL othEr cases, $\bp>0$ and, heNCe, $(H \sEtminus \Sigma, d_{\bp})$ and $(H \setMinus \siGMa, d_{\bP^*})$ ArE homeOmorphiC To $(h \SEtmInUs \sigMa, G_h)$. FInallY, we ConsideR the Gromov BouNdAries of these HYpErbolic spaCes. let us denote tHe oNe-poinT cOmpACtifiCatIon of A hyperSUrface $h \in {\cal{H}}^{\R}_n$ By $\WIdehat{h}$. for the sINGuLar sEt $\Sigma_H$ of some $H \in {\CAl{h}}^{\r}_n$ we *always* add $\iNfty_H$ tO $\SIGmA$ ANd dEfiNe $\widehat\SIgMa:=\SigMA \cup \infTY_H$ (NOTE thAT $\Sigma$ could alReady be compACt). FOr $H \in {\cAl{g}}^{c}_n$ WE set $\wiDehat h=H$ ANd $\wIdehat\Sigma=\sigmA$. In SectioN \[grch\] wE Prove thE FollowiNg theoRem RenDeriNG $\wIdEhaT\SIGma$ AS tHe GROmoV boundarY oF tHe hypErboLIC UNfolDinGs of $(h \setmInus \Sigma, g_H)$. \[thM4\] FoR any NOn-tOtallY geodEsic $h \iN {\cal{G}}$ The ideNtity MaP on $H \setminus \SigMa$ exTends to hoMeoMoRphIsMs betWEen the One-PoiNt compaCtificaTIon $\WiDEHAt{h}$ and the Gromov compaCtIFIcAtions $\ovErline{x}_g$ oF $X=(h \Setminus \siGma,D_{\bp})$, $(H \SETminuS \SigMA,d_{\Bp^*})$ and $(H \seTminus \sIgMa, K_{H \setmiNuS \Sigma})$: $$\WiDehAt{H}\Cong\oVErliNe{(H \setMinus \SigMa,d_{\bp})}_g \Cong \overline{(H \sETminus \Sigma,d_{\bP^*})}_g \cONG \oVErliNe{(H \Setminus \SigMa, k_{H \SEtmiNus \SIGmA})}_G,$$ wHEre $ \coNg$ meaNs HOmEOmorphic. In particulaR, wE find fOr the gromov boundarIes $\p_G(X)$: $$\wideHAT{\sigma} \conG\p_G(H \SEtMInus \Sigma,d_{\bp}) \coNg \p_G(H \Setminus \SiGMa,d_{\bp^*}) \conG \p_G(H \sEtminus \SIgma

but they are one-point sp aces since $\bp \e qui v0$ ( and$\Sigma \v$ cf .  \[ t3\].) In all other ca ses,$\ b p>0$ an d, he nce, $( H \ s e tmi nu s\Si gm a ,d_{\b p}) $ and $ (H \setmin us\S igma, d_{\bp ^ *} )$ are hom eom orphic to $( H \ setmin us \S i gma,g_H )$. Finall y , we c onsider t he Gromov boundar i e sof t hese hyperbolic s p ac e s. Let us deno te the o n e- p o int co mpactifica ti on of a hyper s ur f a c e $ H \in {\cal{H} }^{\R}_n$ b y $\ wideha t{ H}$ . For t he si ng u lar set $\Sigm a_H$ of some$H \in {\cal{H } }^{\R}_ n$ we*al way s* a d d$\ inf ty _ H$t o$\S i gma $ and de fi ne $\wi deha t \ S i gma: =\S igma \cup \infty_H$ (n ote tha t $\ Sigma $ cou ld a lr eadybe com pact) .For $H \in {\ca l{G} }^{c}_n$wese t $ \w ideha t H=H$and $\ widehat \Sigma= \ Sig ma $ . In Section \[grch\]we p ro ve the f ollowi n gth e orem ren de rin g $\ w i dehat \Sig m a$ as theGromov bo un dary of t he hyp er bol icunfol d ings of $( H \setmi nus \ S igma, g_H)$. \ [thm4\] For a n yn o n- t otal lygeodesic $H \in {\ca l{G} } $the ident ity m ap on $H \setminus \Sigma $extend s tohomeomorphism s betweent h e one-poi nt c o mp a ctification $\ wideh at{H}$ and the Grom ov co mpactifi cations $ \ o verline{ X}_ G$of$X= ( H \ setminus \Sig m a ,d_{ \b p})$, $ (H\setmin us\Si gma ,d_ {\ bp^*})$ a nd $(H \ se tm in us \S igma, k_{H \se tm inu s\Si gma}) $ : $$\w ideha t{H} \c on g \ov erline{ ( H\ s etmi nu s\Sig ma, d_ {\bp} )}_G \co ng \ove rline{(H\se t minu s\S igma,d_ {\bp^*})}_G \ co ng \overli ne {(H \setm i n us \Sigm a, k_{H \setminus \Sigm a })}_G,$ $ w here$ \c ong$ mean s h omeomo rph i c. Inpartic ular, w e f i n d for t he Gr om ov boundar i e s $ \p_G( X) $: $ $\wideh at{\Sigma} \cong\p _ G(H \setminus \S igm a,d_ { \ bp })\ co n g \ p_ G (H\ s etminus \Sigma, d_{\bp^*}) \ c on g \p_G(H \ s etm in us \Sig ma

but_they are_one-point spaces since $\bp_\equiv 0$_(and_$\Sigma \v$_cf._ \[t3\].) In all_other cases, $\bp>0$_and, hence, $(H \setminus_\Sigma, d_{\bp})$ and_$(H_\setminus \Sigma, d_{\bp^*})$ are homeomorphic to $(H \setminus \Sigma, g_H)$. Finally, we consider the Gromov_boundaries_of these_hyperbolic_spaces._Let us denote the one-point_compactification of a hypersurface $H_\in {\cal{H}}^{\R}_n$_by $\widehat{H}$. For the singular set $\Sigma_H$ of_some_$H \in {\cal{H}}^{\R}_n$_we *always* add $\infty_H$ to $\Sigma$ and define $\widehat\Sigma:=\Sigma_\cup \infty_H$ (note that $\Sigma$ could_already be compact)._For_$H_\in {\cal{G}}^{c}_n$ we set_$\widehat H=H$ and $\widehat\Sigma=\Sigma$. In Section_\[grch\] we prove the following theorem_rendering $\widehat\Sigma$ as the Gromov boundary of_the hyperbolic unfoldings of $(H \setminus_\Sigma, g_H)$. \[thm4\] For any non-totally_geodesic $H_\in {\cal{G}}$ the identity map_on $H \setminus_\Sigma$ extends_to homeomorphisms between_the one-point compactification $\widehat{H}$ and the_Gromov compactifications $\overline{X}_G$_of $X=(H \setminus \Sigma,d_{\bp})$, $(H \setminus_\Sigma,d_{\bp^*})$_and $(H \setminus_\Sigma,_k_{H_\setminus \Sigma})$:_$$\widehat{H}\cong\overline{(H \setminus \Sigma,d_{\bp})}_G_\cong_\overline{(H \setminus_\Sigma,d_{\bp^*})}_G_\cong \overline{(H \setminus \Sigma, k_{H \setminus_\Sigma})}_G,$$_where $ \cong$ means homeomorphic. In particular,_we find for the_Gromov_boundaries $\p_G(X)$: $$\widehat{\Sigma} \cong\p_G(H_\setminus \Sigma,d_{\bp}) \cong \p_G(H \setminus_\Sigma,d_{\bp^*}) \cong \p_G(H \setminus \Sigma

-2a_{LR}}=1-2a_{LR}^2-8a_{LR}^3-18a_{LR}^3\ldots,$$ where the coefficient $a_{LR}$ depends on $\cos\theta$ through $$a_{LR}(\cos\theta)=\frac{f_{LL}f_{LR}}{C_{LR}^2(\cos\theta)}v^2\sin^2\theta\,.$$ Again, the corresponding expressions for $|\vec{P}_{LR}|$ and $a_{LR}$ can be found by the substitution $(L\leftrightarrow R)$. For the fun of it we also list a compact LO form for $|\vec{P}(P_{\rm eff}=0)|$. One has $$\label{polpeff0} |\vec{P}(P_{\rm eff}=0)|=\frac{\sqrt{(C_{LR}^2-C_{RL}^2)^2 -4v^2\sin^2\theta(C_{LR}f_{LL}-C_{RL}f_{RR})(C_{LR}f_{LR}-C_{RL}f_{RL})}} {C_{LR}^2+C_{RL}^2-2v^2\sin^2\theta(f_{LL}f_{LR}+f_{RR}f_{RL})}\,.$$ Eq. (\[polpeff0\]) would produce a LO version of Fig \[fig:zeropol\]. \[sec5\]Effective beam polarization =================================== As described in Sec. \[sec2\], large values of the effective beam polarization $P_{\rm eff}$ are needed to produce large polarization values of $\vec{P}$. It is a fortunate circumstance that nearly maximal values of $P_{\rm eff}$ can be achieved with non-maximal values of $(h_{-},h_{+})$. This is shown in Fig. \[fig:contour\] where we drawn contour plots $P_{\rm eff}={\it const}$ in the $(h_-,h_+)$ plane. The two examples shown in Fig. \[fig:contour\] refer to $$\begin{aligned} &&(h_{-}=-0.80,\,h_{+}=+0.625)\qquad\mbox{leads to}\quad P_{\rm eff}=-0.95, \nonumber\\ &&(h_{-}=+0.80,\,h_{+}=-0.625)\qquad\mbox{leads to}\quad

-2a_{LR}}=1 - 2a_{LR}^2 - 8a_{LR}^3 - 18a_{LR}^3\ldots,$$ where the coefficient $ a_{LR}$ depends on $ \cos\theta$ through $ $ a_{LR}(\cos\theta)=\frac{f_{LL}f_{LR}}{C_{LR}^2(\cos\theta)}v^2\sin^2\theta\,.$$ Again, the corresponding expressions for $ |\vec{P}_{LR}|$ and $ a_{LR}$ can be found by the substitution $ (L\leftrightarrow R)$. For the playfulness of it we besides number a compact LO form for $ |\vec{P}(P_{\rm eff}=0)|$. One has $ $ \label{polpeff0 } |\vec{P}(P_{\rm eff}=0)|=\frac{\sqrt{(C_{LR}^2 - C_{RL}^2)^2 -4v^2\sin^2\theta(C_{LR}f_{LL}-C_{RL}f_{RR})(C_{LR}f_{LR}-C_{RL}f_{RL }) } } { C_{LR}^2+C_{RL}^2 - 2v^2\sin^2\theta(f_{LL}f_{LR}+f_{RR}f_{RL})}\,.$$ Eq.   (\[polpeff0\ ]) would grow a LO version of Fig   \[fig: zeropol\ ]. \[sec5\]Effective beam polarization = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = As identify in Sec.   \[sec2\ ], big values of the effective radio beam polarization $ P_{\rm eff}$ are needed to produce big polarization values of $ \vec{P}$. It is a fortunate circumstance that closely maximal value of $ P_{\rm eff}$ can be achieved with non - maximal value of $ (h_{-},h_{+})$. This is shown in Fig.   \[fig: contour\ ] where we drawn contour plot $ P_{\rm eff}={\it const}$ in the $ (h_-,h_+)$ plane. The two examples shown in Fig.   \[fig: contour\ ] denote to $ $ \begin{aligned } & & (h_{-}=-0.80,\,h_{+}=+0.625)\qquad\mbox{leads to}\quad P_{\rm eff}=-0.95, \nonumber\\ & & (h_{-}=+0.80,\,h_{+}=-0.625)\qquad\mbox{leads to}\quad

-2a_{LR}}=1-2w_{LR}^2-8a_{LR}^3-18a_{LR}^3\ldots,$$ where the coefficient $a_{LT}$ eependv on $\cks\theta$ ghrough $$a_{LR}(\cos\theta)=\frac{f_{LL}f_{LC}}{C_{LR}^2(\xos\thtna)}v^2\sin^2\theta\,.$$ Again, the correspojding expreswuons for $|\vxd{P}_{LR}|$ and $a_{LR}$ dwn bz hound by the sunstitution $(N\leftrightarrof F)$. Flr the fun of it we also list a com[act LO flrm for $|\vec{P}(P_{\ri efg}=0)|$. One has $$\label{polpeff0} |\vec{P}(P_{\rm eff}=0)|=\frac{\sqdt{(C_{LR}^2-C_{RN}^2)^2 -4v^2\sin^2\theta(C_{KR}f_{LL}-C_{RL}f_{RR})(C_{LR}f_{LR}-C_{RL}f_{RL})}} {C_{PR}^2+C_{RP}^2-2v^2\sin^2\theta(f_{LL}f_{LR}+f_{RG}f_{RL})}\,.$$ Eq. (\[polpgrf0\]) riuld produce a LO verspmn of Fig \[fjg:zeropol\]. \[sec5\]Effective beam polafizatnon =================================== As descrubwd lt Sec. \[sec2\], lacge vajues of the cgfectiee beam polarization $I_{\rm ehf}$ aee needed to produce narge polarization values ox $\rec{P}$. It is a fortunatw xircukstatce gyat nezrky maximwl talues of $P_{\dm eff}$ can ve achieved with nom-mwqomal values kf $(h_{-},h_{+})$. Ehys is shown in Fig. \[fig:contour\] where we dgawn contour plots $P_{\rm eff}={\ir const}$ in the $(h_-,h_+)$ plaje. The twj examples shown in Fig. \[fig:contour\] refer to $$\begin{anignev} &&(h_{-}=-0.80,\,f_{+}=+0.625)\qqbqd\mbob{oewds to}\quad P_{\rm eff}=-0.95, \nonumber\\ &&(h_{-}=+0.80,\,h_{+}=-0.625)\qquad\mbox{leads to}\zhac

-2a_{LR}}=1-2a_{LR}^2-8a_{LR}^3-18a_{LR}^3\ldots,$$ where the coefficient $a_{LR}$ depends on $$a_{LR}(\cos\theta)=\frac{f_{LL}f_{LR}}{C_{LR}^2(\cos\theta)}v^2\sin^2\theta\,.$$ the corresponding for $|\vec{P}_{LR}|$ and the $(L\leftrightarrow R)$. For fun of it also list a compact LO form $|\vec{P}(P_{\rm eff}=0)|$. One has $$\label{polpeff0} |\vec{P}(P_{\rm eff}=0)|=\frac{\sqrt{(C_{LR}^2-C_{RL}^2)^2 -4v^2\sin^2\theta(C_{LR}f_{LL}-C_{RL}f_{RR})(C_{LR}f_{LR}-C_{RL}f_{RL})}} {C_{LR}^2+C_{RL}^2-2v^2\sin^2\theta(f_{LL}f_{LR}+f_{RR}f_{RL})}\,.$$ Eq. (\[polpeff0\]) would produce LO version of Fig \[fig:zeropol\]. \[sec5\]Effective beam polarization =================================== As described in Sec. large of effective polarization $P_{\rm eff}$ are needed to produce large polarization values of $\vec{P}$. It is a fortunate that nearly maximal values of $P_{\rm eff}$ can achieved with non-maximal values $(h_{-},h_{+})$. This is shown in \[fig:contour\] we drawn plots eff}={\it in the $(h_-,h_+)$ The two examples shown in Fig. \[fig:contour\] refer to $$\begin{aligned} &&(h_{-}=-0.80,\,h_{+}=+0.625)\qquad\mbox{leads to}\quad P_{\rm eff}=-0.95, \nonumber\\ &&(h_{-}=+0.80,\,h_{+}=-0.625)\qquad\mbox{leads to}\quad

-2a_{LR}}=1-2a_{LR}^2-8a_{LR}^3-18a_{LR}^3\ldots,$$ where the CoefficienT $a_{LR}$ dEpeNds On $\Cos\tHeta$ Through $$a_{LR}(\cos\tHEta)=\fRac{f_{LL}f_{LR}}{C_{LR}^2(\cos\theta)}v^2\sIn^2\theTa\,.$$ aGain, THe CorreSpondinG ExPREssIoNs For $|\VeC{p}_{Lr}|$ and $a_{lR}$ cAn be fouNd by the subStiTuTion $(L\leftrigHTaRrow R)$. For thE fuN of it we also lIst A compaCt lO fORm for $|\Vec{p}(P_{\rm eFf}=0)|$. One hAS $$\label{Polpeff0} |\veC{P}(p_{\Rm eff}=0)|=\fRAc{\sqrt{(C_{lr}^2-c_{Rl}^2)^2 -4v^2\siN^2\theta(C_{LR}f_{LL}-C_{RL}f_{Rr})(c_{Lr}F_{LR}-C_{RL}f_{RL})}} {C_{LR}^2+C_{Rl}^2-2v^2\sin^2\tHeTA(f_{ll}F_{LR}+F_{RR}F_{RL})}\,.$$ Eq. (\[polpeFf0\]) Would PRoduce a lo vERSIon OF Fig \[fig:zeropoL\]. \[sec5\]EffectiVE beAm polaRiZatIOn =================================== As deScribEd IN SeC. \[sec2\], large vaLues Of the effeCtive bEAm polarIZation $P_{\Rm eff}$ aRe nEedEd to PRoDuCe lArGE poLArIzaTIon Values of $\VeC{P}$. it is a FortUNATE cirCumStanCe thaT nearly maximaL vaLues OF $P_{\rM eff}$ cAn be aChieVeD with Non-maxImal vAlUes of $(h_{-},h_{+})$. This is shOwn iN Fig. \[fig:coNtoUr\] WheRe We draWN contoUr pLotS $P_{\rm eff}={\It const}$ IN thE $(h_-,H_+)$ PLAnE. The two examples shoWn IN fiG. \[fig:contOur\] refER tO $$\bEGin{alignEd} &&(H_{-}=-0.80,\,h_{+}=+0.625)\qQuad\MBOx{leaDs to}\QUaD P_{\rm eff}=-0.95, \nOnumbeR\\ &&(H_{-}=+0.80,\,h_{+}=-0.625)\QqUad\mbox{LeAds to}\qUaD

-2a_{LR}}=1-2a_{LR}^2-8a_{ LR}^3-18a_ {LR}^ 3\l dot s, $$ w here the coefficie n t $a _{LR}$ depends on $\co s\the ta $ thr o ug h $$a _{LR}(\ c os \ t het a) =\ fra c{ f _{ LL}f_ {LR }}{C_{L R}^2(\cos\ the ta )}v^2\sin^2\ t he ta\,.$$ Ag ain , the corres pon ding e xp res s ionsfor $|\v ec{P}_ { LR}|$and $a_{L R} $ can b e foundb y t he s ubstitution $(L\l e ft r ightarrow R)$. Forth e f u n of it we also l is t a c o mpact L O f o r m fo r $|\vec{P}(P_ {\rm eff}=0 ) |$. One h as $$ \ label{ polpe ff 0 } | \vec{P}(P_{ \rmeff}=0)|= \frac{ \ sqrt{(C _ {LR}^2- C_{RL} ^2) ^2 -4 v ^2 \s in^ 2\ t het a (C _{L R }f_ {LL}-C_{ RL }f _{RR} )(C_ { L R } f_{L R}- C_{R L}f_{ RL})}} {C_{ LR} ^2+C _ {RL }^2-2 v^2\s in^2 \t heta( f_{LL} f_{LR }+ f_{RR}f_{RL})}\ ,.$$ Eq. (\[p olp ef f0\ ]) woul d produ cea L O versi on of F i g \ [f i g : ze ropol\]. \[sec5\] Ef f e ct ive beam polar i za ti o n ====== == === ==== = = ===== ==== = == ====== As des c ri be d in Se c.  \[sec 2\ ],lar ge va l uesof the effecti ve be a m polarization $P_{\rm eff}$ ar e ne e dedtoproduce lar ge p o lari zati o nval u es of $\ve c{ P }$ . It is a fortunateci rcumst ancethat nearly m aximal val u e s of $P_{ \rme ff } $ can be achie ved w ith non-ma x imal val ues o f $(h_{- },h_{+})$ . This issho wninFig . \[ fig:contour\] w here w e drawn co ntour p lot s $ P_{ \rm e ff}={\itconst}$in t he $ (h_ -,h_+ ) $ plane. T hetw o e xampl e s show n inFig.  \ [f i g:c ontour\ ] r e f er t o$$ \beg in{ al igned } && ( h_{ -}=-0.8 0,\,h_{+} =+0 . 625) \q qu ad\mbox {leads to}\qu ad P_{\rm ef f} =-0 .95, \ n o number\\ &&(h_{-}=+0.80,\,h_{+} = -0.625) \qq uad\m box{ leads to} \qu ad

-2a_{LR}}=1-2a_{LR}^2-8a_{LR}^3-18a_{LR}^3\ldots,$$ where_the coefficient_$a_{LR}$ depends on $\cos\theta$_through $$a_{LR}(\cos\theta)=\frac{f_{LL}f_{LR}}{C_{LR}^2(\cos\theta)}v^2\sin^2\theta\,.$$_Again,_the corresponding_expressions_for $|\vec{P}_{LR}|$ and_$a_{LR}$ can be_found by the substitution_$(L\leftrightarrow R)$. For the_fun_of it we also list a compact LO form for $|\vec{P}(P_{\rm eff}=0)|$. One has_$$\label{polpeff0} |\vec{P}(P_{\rm_eff}=0)|=\frac{\sqrt{(C_{LR}^2-C_{RL}^2)^2 _-4v^2\sin^2\theta(C_{LR}f_{LL}-C_{RL}f_{RR})(C_{LR}f_{LR}-C_{RL}f_{RL})}} __{C_{LR}^2+C_{RL}^2-2v^2\sin^2\theta(f_{LL}f_{LR}+f_{RR}f_{RL})}\,.$$ Eq. (\[polpeff0\]) would produce a_LO version of Fig \[fig:zeropol\]. \[sec5\]Effective beam_polarization =================================== As described_in Sec. \[sec2\], large values of the effective beam_polarization_$P_{\rm eff}$ are_needed to produce large polarization values of $\vec{P}$. It_is a fortunate circumstance that nearly_maximal values of_$P_{\rm_eff}$_can be achieved with_non-maximal values of $(h_{-},h_{+})$. This is_shown in Fig. \[fig:contour\] where we drawn_contour plots $P_{\rm eff}={\it const}$ in the_$(h_-,h_+)$ plane. The two examples shown_in Fig. \[fig:contour\] refer to $$\begin{aligned} &&(h_{-}=-0.80,\,h_{+}=+0.625)\qquad\mbox{leads_to}\quad P_{\rm_eff}=-0.95, \nonumber\\ &&(h_{-}=+0.80,\,h_{+}=-0.625)\qquad\mbox{leads to}\quad

alpha}f\|_{{{\mathcal H}}} \|A^{\alpha}g\|_{{{\mathcal H}}}, \; f\in{\text{\rm{dom}}}(B), \, g\in{\text{\rm{dom}}}(A).$$ By Corollary \[c4.9\] and the fact that $\|Qg\|_{{{\mathcal H}}}=\||Q|g\|_{{{\mathcal H}}}$, $\|Q^* f\|_{{{\mathcal H}}}=\||Q^*|f\|_{{{\mathcal H}}}$, the inequalities yield for $\beta, \gamma \in (0,1]$, $$\begin{aligned} \begin{split} \big\||Q|^{\beta}g\big\|_{{{\mathcal H}}} & {\leqslant}D_{\beta} \|A^{\beta}g\|_{{{\mathcal H}}}, \; g\in{\text{\rm{dom}}}(A), \\ \||Q^*|^{\gamma}f\|_{{{\mathcal H}}} & {\leqslant}{\widetilde}D_{\gamma}\|B^{\gamma}f\|_{{{\mathcal H}}}, \; f\in{\text{\rm{dom}}}(B) \end{split}\end{aligned}$$ for some constants $D_{\beta}>0$, ${\widetilde}D_{\gamma}>0$. On the other hand, by, $Q=|Q^*|^{1-\alpha}U|Q|^{\alpha}$, $\alpha\in [0,1]$. Combining these facts one arrives at $$\begin{aligned} \begin{split} & |(f, Qg)_{{{\mathcal H}}}| = |(U^*|Q^*|^{1-\alpha}f, |Q|^{\alpha}g)_{{{\mathcal H}}}| {\leqslant}\||Q^*|^{1-\alpha}f\|_{{{\mathcal H}}} \||Q|^{\alpha} g\|_{{{\mathcal H}}} \\ & \quad {\leqslant}{\widetilde}D_{1-\alpha}\|B^{1-\alpha}f\|_{{{\mathcal H}}} \, {\widetilde}D_{\alpha}\|A^{\alpha}g\|_{{{\mathcal H}}}, \; f\in{\text{\rm{dom}}}(B), \, g\in{\text{\rm{dom}}}(A), \end{split}\end{aligned}$$ completing the proof. Next we show that Theorem \[t4.10\], in fact, implies Theorem \[t4.8\]. This was stated (without proof) in Kato [@Ka61]): Let $Q=B T

alpha}f\|_{{{\mathcal H } } } \|A^{\alpha}g\|_{{{\mathcal H } } }, \; f\in{\text{\rm{dom}}}(B), \, g\in{\text{\rm{dom}}}(A).$$ By Corollary \[c4.9\ ] and the fact that $ \|Qg\|_{{{\mathcal H}}}=\||Q|g\|_{{{\mathcal H}}}$, $ \|Q^ * f\|_{{{\mathcal H}}}=\||Q^*|f\|_{{{\mathcal H}}}$, the inequalities yield for $ \beta, \gamma \in (0,1]$, $ $ \begin{aligned } \begin{split } \big\||Q|^{\beta}g\big\|_{{{\mathcal H } } } & { \leqslant}D_{\beta } \|A^{\beta}g\|_{{{\mathcal H } } }, \; g\in{\text{\rm{dom}}}(A), \\ \||Q^*|^{\gamma}f\|_{{{\mathcal H } } } & { \leqslant}{\widetilde}D_{\gamma}\|B^{\gamma}f\|_{{{\mathcal H } } }, \; f\in{\text{\rm{dom}}}(B) \end{split}\end{aligned}$$ for some constants $ D_{\beta}>0 $, $ { \widetilde}D_{\gamma}>0$. On the early bridge player, by, $ Q=|Q^*|^{1-\alpha}U|Q|^{\alpha}$, $ \alpha\in [ 0,1]$. Combining these facts one arrive at $ $ \begin{aligned } \begin{split } & |(f, Qg)_{{{\mathcal H}}}| = |(U^*|Q^*|^{1-\alpha}f, |Q|^{\alpha}g)_{{{\mathcal H}}}| { \leqslant}\||Q^*|^{1-\alpha}f\|_{{{\mathcal H } } } \||Q|^{\alpha } g\|_{{{\mathcal H } } } \\ & \quad { \leqslant}{\widetilde}D_{1-\alpha}\|B^{1-\alpha}f\|_{{{\mathcal H } } } \, { \widetilde}D_{\alpha}\|A^{\alpha}g\|_{{{\mathcal H } } }, \; f\in{\text{\rm{dom}}}(B), \, g\in{\text{\rm{dom}}}(A), \end{split}\end{aligned}$$ complete the proof. Next we show that Theorem \[t4.10\ ], in fact, incriminate Theorem \[t4.8\ ]. This was submit (without proof) in Kato [ @Ka61 ] ): Let $ Q = B T

alpja}f\|_{{{\mathcal H}}} \|A^{\alpha}g\|_{{{\mathgal H}}}, \; f\in{\text{\rm{dom}}}(B), \, g\ii{\text{\rm{som}}}(A).$$ By Curollary \[c4.9\] and the fact that $\|Qt\|_{{{\mathxal H}}}=\||Q|g\|_{{{\mathcal H}}}$, $\|Q^* f\|_{{{\maghcal H}}}=\||Q^*|f\|_{{{\lathcal Y}}}$, tht inequalities yixmd for $\nzta, \gzlma \nn (0,1]$, $$\begin{aligned} \negin{split} \bhg\||Q|^{\beta}g\big\|_{{{\matvcxl H}}} & {\leqslant}D_{\beta} \|A^{\beta}g\|_{{{\mathcal H}}}, \; d\in{\text{\tm{fom}}}(A), \\ \||Q^*|^{\gamma}f\|_{{{\majhcal R}}} & {\lsqslant}{\widetilde}D_{\gamma}\|B^{\gamma}f\|_{{{\mathczl H}}}, \; f\pn{\text{\rm{dom}}}(B) \end{spkit}\end{aligned}$$ for some condtanhs $D_{\beta}>0$, ${\widetilde}F_{\gamma}>0$. On tye oeyer hand, by, $D=|Q^*|^{1-\alpha}U|Q|^{\alpha}$, $\alpha\in [0,1]$. Combining these facts one arrkves ct $$\begin{alitnwd} \bfcin{split} & |(f, Wg)_{{{\matrcal H}}}| = |(U^*|Q^*|^{1-\alina}f, |Q|^{\anpha}g)_{{{\mayhcal H}}}| {\leqslanb}\||Q^*|^{1-\alpia}f\|_{{{\mqthcal H}}} \||Q|^{\alpha} g\|_{{{\mathral H}}} \\ & \quad {\leqslant}{\ridetilde}G_{1-\ampha}\|B^{1-\alpha}f\|_{{{\mathcao Y}}} \, {\wigetinde}D_{\xophx}\|A^{\ampia}g\|_{{{\jathcap H}}}, \; f\in{\text{\rj{dom}}}(B), \, g\in{\twxt{\rm{dom}}}(A), \end{split}\enc{ajpbned}$$ completjng thq [roof. Next we show that Theorem \[t4.10\], in facu, impmies Theorem \[t4.8\]. This was stated (without proof) in Kato [@Ha61]): Let $Q=B T

alpha}f\|_{{{\mathcal H}}} \|A^{\alpha}g\|_{{{\mathcal H}}}, \; f\in{\text{\rm{dom}}}(B), \, Corollary and the that $\|Qg\|_{{{\mathcal H}}}=\||Q|g\|_{{{\mathcal the yield for $\beta, \in (0,1]$, $$\begin{aligned} \big\||Q|^{\beta}g\big\|_{{{\mathcal H}}} & {\leqslant}D_{\beta} \|A^{\beta}g\|_{{{\mathcal H}}}, g\in{\text{\rm{dom}}}(A), \\ \||Q^*|^{\gamma}f\|_{{{\mathcal H}}} & {\leqslant}{\widetilde}D_{\gamma}\|B^{\gamma}f\|_{{{\mathcal H}}}, \; f\in{\text{\rm{dom}}}(B) \end{split}\end{aligned}$$ for some constants ${\widetilde}D_{\gamma}>0$. On the other hand, by, $Q=|Q^*|^{1-\alpha}U|Q|^{\alpha}$, $\alpha\in [0,1]$. Combining these facts one at \begin{split} |(f, H}}}| = |(U^*|Q^*|^{1-\alpha}f, |Q|^{\alpha}g)_{{{\mathcal H}}}| {\leqslant}\||Q^*|^{1-\alpha}f\|_{{{\mathcal H}}} \||Q|^{\alpha} g\|_{{{\mathcal H}}} \\ & \quad {\leqslant}{\widetilde}D_{1-\alpha}\|B^{1-\alpha}f\|_{{{\mathcal H}}} \, H}}}, \; f\in{\text{\rm{dom}}}(B), \, g\in{\text{\rm{dom}}}(A), \end{split}\end{aligned}$$ completing the Next we show that \[t4.10\], in fact, implies Theorem This stated (without in [@Ka61]): $Q=B T

alpha}f\|_{{{\mathcal H}}} \|A^{\alpha}g\|_{{{\mathCal H}}}, \; f\in{\texT{\rm{doM}}}(B), \, g\In{\tExT{\rm{dOm}}}(A).$$ BY Corollary \[c4.9\] and THe faCt that $\|Qg\|_{{{\mathcal H}}}=\||Q|g\|_{{{\mathCal H}}}$, $\|Q^* F\|_{{{\mAThcaL h}}}=\||Q^*|F\|_{{{\mathCal H}}}$, the INeQUAliTiEs YieLd FOr $\Beta, \gAmmA \in (0,1]$, $$\begiN{aligned} \beGin{SpLit} \big\||Q|^{\beta}g\BIg\|_{{{\Mathcal H}}} & {\leQslAnt}D_{\beta} \|A^{\betA}g\|_{{{\mAthcal h}}}, \; g\In{\tEXt{\rm{dOm}}}(A), \\ \||q^*|^{\gammA}f\|_{{{\mathCAl H}}} & {\leqSlant}{\wideTiLDe}D_{\gamMA}\|B^{\gamma}F\|_{{{\MAtHcal h}}}, \; f\in{\text{\rm{dom}}}(B) \end{SPlIT}\end{aligned}$$ for Some coNsTAnTS $d_{\beTa}>0$, ${\wIdetilde}D_{\gAmMa}>0$. On tHE other hANd, BY, $q=|q^*|^{1-\alPHa}U|Q|^{\alpha}$, $\alphA\in [0,1]$. CombininG TheSe factS oNe aRRives aT $$\begiN{aLIgnEd} \begin{spliT} & |(f, Qg)_{{{\Mathcal H}}}| = |(U^*|q^*|^{1-\alpha}F, |q|^{\alpha}g)_{{{\MAthcal H}}}| {\LeqslaNt}\||Q^*|^{1-\AlpHa}f\|_{{{\mAThCaL H}}} \||Q|^{\AlPHa} g\|_{{{\MAtHcaL h}}} \\ & \quAd {\leqslaNt}{\WiDetilDe}D_{1-\aLPHA}\|b^{1-\alpHa}f\|_{{{\MathCal H}}} \, {\wIdetilde}D_{\alphA}\|A^{\aLpha}G\|_{{{\MatHcal H}}}, \; F\in{\teXt{\rm{DoM}}}(B), \, g\in{\Text{\rm{Dom}}}(A), \eNd{Split}\end{aligned}$$ CompLeting the ProOf. nexT wE show THat TheOreM \[t4.10\], iN fact, imPlies ThEOreM \[t4.8\]. tHIS wAs stated (without proOf) IN kaTo [@Ka61]): Let $Q=b T

alpha}f\|_{{{\mathcal H}}} \|A^{\alp ha}g\ |_{ {{\ ma thca l H} }}, \; f\in{\t e xt{\ rm{dom}}}(B), \, g\in{ \text {\ r m{do m }} }(A). $$ ByC or o l lar y\[ c4. 9\ ] a nd th e f act tha t $\|Qg\|_ {{{ \m athcal H}}}= \ || Q|g\|_{{{\ mat hcal H}}}$,$\| Q^* f\ |_ {{{ \ mathc alH}}}= \||Q^* | f\|_{{ {\mathcal H } }}$, t h e inequ a l it iesyield for $\beta, \g a mma \in (0,1]$ , $$\b eg i n{ a l ign ed} \begin{sp li t} \b i g\||Q|^ { \b e t a }g\ b ig\|_{{{\math cal H}}} &{ \le qslant }D _{\ b eta} \ |A^{\ be t a}g \|_{{{\math calH}}}, \;g\in{\ t ext{\rm { dom}}}( A), \\ \| |Q^ *|^{ \ ga mm a}f \| _ {{{ \ ma thc a l H }}} & {\ le qs lant} {\wi d e t i lde} D_{ \gam ma}\| B^{\gamma}f\| _{{ {\ma t hca l H}} }, \; f\i n{ \text {\rm{d om}}} (B ) \end{split}\e nd{a ligned}$$ fo rsom econst a nts $D _{\ bet a}>0$,${\wide t ild e} D _ { \g amma}>0$. On the o th e r h and, by, $Q=|Q ^ *| ^{ 1 -\alpha} U| Q|^ {\al p h a}$,$\al p ha \in [0,1 ]$. Co m bi ni ng thes efactson e a rri ves a t $$\ begin{ aligned} \beg i n{split} & |(f , Qg)_{{{\math c al H }} } | =|(U ^*|Q^*|^{1- \alp h a}f, |Q| ^ {\ alp h a}g)_ {{{\m at h ca l H}}}| {\leqslant}\ || Q^*|^{ 1-\al pha}f\|_{{{\m athcal H}} } \ ||Q|^{\a lpha } g \ |_{{{\mathcalH}}}\\ & \quad {\leqsla nt}{\ widetild e}D_{1-\a l p ha}\|B^{ 1-\ alp ha} f\| _ { {{ \mathcal H}}} \ , {\ wi detilde }D_ {\alpha }\| A^{ \al pha }g \|_{{{\ma thcal H} }} ,\; f\i n{\te x t{\rm{do m} }}( B) , \ , g\i n {\text {\rm{ dom} }} (A ) , \ end{spl i t} \ e nd{a li gn ed}$ $ c om pleti ng t h e p roof. Next we s how that T he orem \[ t4.10\], in f ac t, implies T heo rem \[ t 4 .8\]. Th is was stated (withoutp roof) i n K ato [ @Ka6 1]): Let $Q =B T

alpha}f\|_{{{\mathcal H}}}_\|A^{\alpha}g\|_{{{\mathcal H}}},_\; f\in{\text{\rm{dom}}}(B), \, g\in{\text{\rm{dom}}}(A).$$ By Corollary_\[c4.9\] and_the_fact that_$\|Qg\|_{{{\mathcal_H}}}=\||Q|g\|_{{{\mathcal H}}}$, $\|Q^*_f\|_{{{\mathcal H}}}=\||Q^*|f\|_{{{\mathcal H}}}$,_the inequalities yield for_$\beta, \gamma \in_(0,1]$,_$$\begin{aligned} \begin{split} \big\||Q|^{\beta}g\big\|_{{{\mathcal H}}} & {\leqslant}D_{\beta} \|A^{\beta}g\|_{{{\mathcal H}}}, \; g\in{\text{\rm{dom}}}(A), \\ \||Q^*|^{\gamma}f\|_{{{\mathcal H}}} & {\leqslant}{\widetilde}D_{\gamma}\|B^{\gamma}f\|_{{{\mathcal H}}}, \;_f\in{\text{\rm{dom}}}(B) \end{split}\end{aligned}$$_for some_constants_$D_{\beta}>0$,_${\widetilde}D_{\gamma}>0$. On the other hand,_by, $Q=|Q^*|^{1-\alpha}U|Q|^{\alpha}$, $\alpha\in [0,1]$. Combining_these facts_one arrives at $$\begin{aligned} \begin{split} & |(f, Qg)_{{{\mathcal H}}}| =_|(U^*|Q^*|^{1-\alpha}f,_|Q|^{\alpha}g)_{{{\mathcal H}}}| {\leqslant}\||Q^*|^{1-\alpha}f\|_{{{\mathcal H}}}_\||Q|^{\alpha} g\|_{{{\mathcal H}}} \\ & \quad {\leqslant}{\widetilde}D_{1-\alpha}\|B^{1-\alpha}f\|_{{{\mathcal H}}} \, {\widetilde}D_{\alpha}\|A^{\alpha}g\|_{{{\mathcal_H}}}, \; f\in{\text{\rm{dom}}}(B), \, g\in{\text{\rm{dom}}}(A), \end{split}\end{aligned}$$ completing_the proof. Next we_show_that_Theorem \[t4.10\], in fact,_implies Theorem \[t4.8\]. This was stated_(without proof) in Kato [@Ka61]): Let $Q=B_T

$0.798$ $0.804$ $0.800$ $ 0$ $0.480$ $0.746$ $0.274$ $0.239$ $0.280$ $0.359$ $0.472$ \[\] $(6)$ $0.490$ $0.605$ $0.657$ $0.529$ $0.495$ $ 0$ $0.486$ $0.490$ $0.480$ $0.481$ $0.484$ $0.509$ \[\] $(7)$ $0.788$ $0.795$ $0.808$ $0.768$ $0.775$ $0.477$ $ 0$ $0.788$ $0.784$ $0.789$ $0.775$ $0.474$ \[\] $(8)$ $ 0.0308$ $0.798$ $0.804$ $0.802$ $0.272$ $0.471$ $0.673$ $ 0$ $0.134$ $0.0533$ $0.415$ $0.470$ \[\] $(9)$ $0.138$ $0.798$ $0.804$ $0.802$ $0.237$ $0.478$ $0.702$ $0.134$ $ 0$ $0.145$ $0.399$ $0.469$ \[\] $(10)$ $ 0.0435$ $0.798$ $0.804$ $0.802$ $0.277$ $0.486$ $0.670$ $0.0533$ $0.144$ $ 0$ $0.418$ $0.469$ \[\] $(11)$ $0.455$ $0.798$ $0.798$ $0.784$ $0.383$ $0.482$ $0.768$ $0.

$ 0.798 $ $ 0.804 $ $ 0.800 $ $ 0 $ $ 0.480 $ $ 0.746 $ $ 0.274 $ $ 0.239 $ $ 0.280 $ $ 0.359 $ $ 0.472 $ \[\ ] $ (6)$ $ 0.490 $ $ 0.605 $ $ 0.657 $ $ 0.529 $ $ 0.495 $ $ 0 $ $ 0.486 $ $ 0.490 $ $ 0.480 $ $ 0.481 $ $ 0.484 $ $ 0.509 $ \[\ ] $ (7)$ $ 0.788 $ $ 0.795 $ $ 0.808 $ $ 0.768 $ $ 0.775 $ $ 0.477 $ $ 0 $ $ 0.788 $ $ 0.784 $ $ 0.789 $ $ 0.775 $ $ 0.474 $ \[\ ] $ (8)$ $ 0.0308 $ $ 0.798 $ $ 0.804 $ $ 0.802 $ $ 0.272 $ $ 0.471 $ $ 0.673 $ $ 0 $ $ 0.134 $ $ 0.0533 $ $ 0.415 $ $ 0.470 $ \[\ ] $ (9)$ $ 0.138 $ $ 0.798 $ $ 0.804 $ $ 0.802 $ $ 0.237 $ $ 0.478 $ $ 0.702 $ $ 0.134 $ $ 0 $ $ 0.145 $ $ 0.399 $ $ 0.469 $ \[\ ] $ (10)$ $ 0.0435 $ $ 0.798 $ $ 0.804 $ $ 0.802 $ $ 0.277 $ $ 0.486 $ $ 0.670 $ $ 0.0533 $ $ 0.144 $ $ 0 $ $ 0.418 $ $ 0.469 $ \[\ ] $ (11)$ $ 0.455 $ $ 0.798 $ $ 0.798 $ $ 0.784 $ $ 0.383 $ $ 0.482 $ $ 0.768 $ $ 0.

$0.798$ $0.804$ $0.800$ $ 0$ $0.480$ $0.746$ $0.274$ $0.239$ $0.280$ $0.359$ $0.472$ \[\] $(6)$ $0.490$ $0.605$ $0.657$ $0.529$ $0.495$ $ 0$ $0.486$ $0.490$ $0.480$ $0.481$ $0.484$ $0.509$ \[\] $(7)$ $0.788$ $0.795$ $0.808$ $0.768$ $0.775$ $0.477$ $ 0$ $0.788$ $0.784$ $0.789$ $0.775$ $0.474$ \[\] $(8)$ $ 0.0308$ $0.798$ $0.804$ $0.802$ $0.272$ $0.471$ $0.673$ $ 0$ $0.134$ $0.0533$ $0.415$ $0.470$ \[\] $(9)$ $0.138$ $0.798$ $0.804$ $0.802$ $0.237$ $0.478$ $0.702$ $0.134$ $ 0$ $0.145$ $0.399$ $0.469$ \[\] $(10)$ $ 0.0435$ $0.798$ $0.804$ $0.802$ $0.277$ $0.486$ $0.670$ $0.0533$ $0.144$ $ 0$ $0.418$ $0.469$ \[\] $(11)$ $0.455$ $0.798$ $0.798$ $0.784$ $0.383$ $0.482$ $0.768$ $0.

$0.798$ $0.804$ $0.800$ $ 0$ $0.480$ $0.746$ $0.280$ $0.472$ \[\] $0.490$ $0.605$ $0.657$ $0.490$ $0.481$ $0.484$ $0.509$ $(7)$ $0.788$ $0.795$ $0.768$ $0.775$ $0.477$ $ 0$ $0.788$ $0.789$ $0.775$ $0.474$ \[\] $(8)$ $ 0.0308$ $0.798$ $0.804$ $0.802$ $0.272$ $0.471$ $0.673$ 0$ $0.134$ $0.0533$ $0.415$ $0.470$ \[\] $(9)$ $0.138$ $0.798$ $0.804$ $0.802$ $0.237$ $0.478$ $0.134$ 0$ $0.399$ \[\] $(10)$ $ 0.0435$ $0.798$ $0.804$ $0.802$ $0.277$ $0.486$ $0.670$ $0.0533$ $0.144$ $ 0$ $0.418$ $0.469$ $(11)$ $0.455$ $0.798$ $0.798$ $0.784$ $0.383$ $0.482$ $0.768$

$0.798$ $0.804$ $0.800$ $ 0$ $0.480$ $0.746$ $0.274$ $0.239$ $0.280$ $0.359$ $0.472$ \[\] $(6)$ $0.490$ $0.605$ $0.657$ $0.529$ $0.495$ $ 0$ $0.486$ $0.490$ $0.480$ $0.481$ $0.484$ $0.509$ \[\] $(7)$ $0.788$ $0.795$ $0.808$ $0.768$ $0.775$ $0.477$ $ 0$ $0.788$ $0.784$ $0.789$ $0.775$ $0.474$ \[\] $(8)$ $ 0.0308$ $0.798$ $0.804$ $0.802$ $0.272$ $0.471$ $0.673$ $ 0$ $0.134$ $0.0533$ $0.415$ $0.470$ \[\] $(9)$ $0.138$ $0.798$ $0.804$ $0.802$ $0.237$ $0.478$ $0.702$ $0.134$ $ 0$ $0.145$ $0.399$ $0.469$ \[\] $(10)$ $ 0.0435$ $0.798$ $0.804$ $0.802$ $0.277$ $0.486$ $0.670$ $0.0533$ $0.144$ $ 0$ $0.418$ $0.469$ \[\] $(11)$ $0.455$ $0.798$ $0.798$ $0.784$ $0.383$ $0.482$ $0.768$ $0.

$0.798$ $0.804$ $0.8 00$ $ 0$ $0.480$ $ 0 .746 $ $ 0 . 274 $ $0. 239$ $ 0.280$ $0.3 59$ $ 0.472$ \ [\] $ (6)$ $0.4 90 $ $0.605$ $0.657$ $0.529$ $ 0. 4 9 5$ $ 0 $ $0.486$ $ 0 .49 0$ $0.48 0$ $0 .481$ $0. 4 84$ $0 .509 $ \ [\ ] $ ( 7)$ $0. 788$ $ 0.7 95$ $0.808$ $0.76 8$ $0.775 $ $0.47 7$ $ 0 $ $0.788$ $0 .784$ $0 .789$ $0.775$ $0. 474$ \[\] $( 8) $ $ 0 .03 08$ $0.7 98$ $0.8 0 4$ $ 0 .802$ $ 0 .272 $ $ 0 .471 $ $0.6 73 $ $ 0$ $0.134$ $0.0533 $ $0.415 $ $ 0.470$ \[ \] $(9)$ $0. 138 $ $0.798$ $ 0.804$ $0. 802 $ $0 .237$ $0. 4 78$ $0.7 0 2$ $ 0 .13 4$ $ 0$ $0.145$ $0 .3 99 $ $0.46 9$ \[ \] $(1 0 ) $ $ 0.0435$ $0. 7 98$ $0.8 04$ $0.8 02$ $0.2 77 $ $0 .486$ $ 0.67 0$ $0.0533 $ $ 0.1 44$ $ 0 $ $0.418$ $0. 4 69$ \[\ ] $(1 1 )$ $ 0.455$ $0 . 7 98$ $0.798 $ $0. 7 84 $ $ 0.3 83$ $ 0.482$ $0.768$ $0.

$0.798$ _ _ _ __ $0.804$__ _ _ $0.800$_ __ $ 0$ __ ___ _ _$0.480$ _ __ _ $0.746$ _ _ _$0.274$__ _ _ $0.239$ _ _ _ $0.280$ _ _ _ _ _$0.359$ _ _ _ $0.472$ __ ___ _\[\] $(6)$ __ __ __ $0.490$ _ $0.605$ __ _ $0.657$_ _ _ _$0.529$ _ $0.495$ _ _ __ __ _ $ 0$ _ _ _ __$0.486$ _ _ $0.490$___ _ _ $0.480$_ __ __ $0.481$ ___ _ $0.484$ _ _ $0.509$ _ ___ \[\] $(7)$ _ _ _ $0.788$ _ $0.795$ _ $0.808$ _ $0.768$ _ _ __$0.775$ ____ _ _$0.477$ _ _ _ _ __ $ 0$ _ $0.788$ _ __ _ _ $0.784$ _ __$0.789$ _ _ _ _ _ _$0.775$ _ _ __ $0.474$ ____ \[\]_$(8)$ _ _ __ $ 0.0308$ _ _ $0.798$ _ $0.804$_ $0.802$ _ _ _ $0.272$__ _ _ $0.471$ __ __$0.673$_ __ _____ $ 0$ _ __ _ _$0.134$_ _ _ $0.0533$ __ ____ _ $0.415$ _$0.470$ _ _ _\[\]_$(9)$ _ _ $0.138$ _ _$0.798$ $0.804$ _ $0.802$ _ _ __ $0.237$_ _ _ $0.478$ _ $0.702$ _ _ _ $0.134$ _ _ $ 0$ __ _ $0.145$ _ _ __ $0.399$ _ _ _ _ $0.469$ _ _ \[\] $(10)$ _ $ 0.0435$ $0.798$ _ _$0.804$_ $0.802$ _ _ _ $0.277$ ___ __ $0.486$ _ _ _ _$0.670$ _ _ _ $0.0533$__ _ __ _ _$0.144$_ _ _ __ _ __ _ _$_0$ _ _ _ $0.418$ __ _ $0.469$ _ _ _ \[\] $(11)$ _ _ _ __ $0.455$ _ $0.798$ _ __ $0.798$ _ $0.784$ _ _ _ _ $0.383$_ ___ $0.482$ _ _ _ __ $0.768$ _ _ _ __ $0.

(s) = f^{-1}(-\infty,s]$. Then the magnitude function of the sublevel set persistent homology $H_\ast S\colon({\mathbb{R}},\leq)\to{\mathbf{GrVect}}$ is expressed as follows: $$|t(M,f)|_{\mathrm{Morse}}=|t(H_\ast S)|=\sum_{p} (-1)^{{\operatorname{ind}}(p)}e^{-f(p)t}$$ where the sum is over all critical points of $f$. A basic result of Morse theory [@milnor2016morse Theorem 3.1] states that if $a<b$ are real numbers such that $f^{-1}(a,b]$ contains no critical points of $f$, then $M^b=f^{-1}(-\infty,b]$ deformation retracts onto $M^a=f^{-1}(-\infty,a]$. It follows that the critical values (i.e. the values $f(p)$ of $f$ at the critical points $p$) are the startpoints and endpoints of the interval decomposition of $H_\ast S$. List the critical values as $v_1<v_2<\cdots<v_k$. We may now use the description of magnitude as the filtered Euler characteristic : $$|t(H_\ast S)|=\sum_{i=1}^k\chi(M^{v_i})(e^{-v_it}-e^{-v_{i+1}t}),$$ where $v_{k+1}$ is interpreted as $\infty$. Another basic result of Morse theory [@milnor2016morse Theorem 3.2, Remark 3.3 & Remark 3.4] states the following. Suppose that $b$ is a critical value of $f$, and $a<b$ is such that there are no critical values of $f$ in $(a,b)$, and let $p_1,\ldots,p_r$ be the critical points of $f$ with critical value $b$. Then $M^b$ has a subspace of the form $M^a\cup e^{{\operatorname{ind}}(p_1)}\cup\cdots\cup e^{{\operatorname{ind}}(p_r)}$, and $M^b$ deformation retracts onto $M^a\cup e^{{\operatorname{ind}}(p_1)}\cup\cdots\cup e^{{\operatorname{ind}}(p_r)}$. Using this result, we

( s) = f^{-1}(-\infty, s]$. Then the magnitude function of the sublevel set persistent homology $ H_\ast S\colon({\mathbb{R}},\leq)\to{\mathbf{GrVect}}$ is expressed as follow: $ $ |t(M, f)|_{\mathrm{Morse}}=|t(H_\ast S)|=\sum_{p } (-1)^{{\operatorname{ind}}(p)}e^{-f(p)t}$$ where the kernel is over all critical points of $ f$. A basic result of Morse theory [ @milnor2016morse Theorem 3.1 ] submit that if $ a < b$ are real numbers such that $ f^{-1}(a, b]$ contain no critical item of $ f$, then $ M^b = f^{-1}(-\infty, b]$ deformation abjure onto $ M^a = f^{-1}(-\infty, a]$. It follows that the critical values (i.e.   the value $ f(p)$ of $ f$ at the critical points $ p$) are the startpoints and endpoints of the time interval decomposition of $ H_\ast S$. List the critical values as $ v_1 < v_2<\cdots < v_k$. We may now use the description of magnitude as the filtered Euler characteristic: $ $ |t(H_\ast S)|=\sum_{i=1}^k\chi(M^{v_i})(e^{-v_it}-e^{-v_{i+1}t}),$$ where $ v_{k+1}$ is interpret as $ \infty$. Another basic result of Morse theory [ @milnor2016morse Theorem 3.2, Remark 3.3 & Remark 3.4 ] states the pursuit. Suppose that $ b$ is a critical value of $ f$, and $ a < b$ is such that there are no critical value of $ f$ in $ (a, b)$, and let $ p_1,\ldots, p_r$ be the critical point of $ f$ with critical value $ b$. Then $ M^b$ has a subspace of the form $ M^a\cup e^{{\operatorname{ind}}(p_1)}\cup\cdots\cup e^{{\operatorname{ind}}(p_r)}$, and $ M^b$ deformation retracts onto $ M^a\cup e^{{\operatorname{ind}}(p_1)}\cup\cdots\cup e^{{\operatorname{ind}}(p_r)}$. Using this resultant role, we

(s) = v^{-1}(-\infty,s]$. Then the magnituae function of jhw sublxvel sef persisgent homology $H_\ast S\colon({\matibb{R}},\oeq)\to{\nathbf{GrVect}}$ is expresred as foplows: $$|t(M,d)|_{\matirm{Morse}}=|t(H_\ast S)|=\snj_{p} (-1)^{{\operatornajc{ind}}(p)}z^{-f(')t}$$ where the suk is over dll critical pmivtd of $f$. A basic result of Morse theorr [@milnot2016mlrse Theorem 3.1] ftatts ehat pf $a<b$ are real numbers such thaf $f^{-1}(a,b]$ cmntains no crotical points of $f$, then $M^b=v^{-1}(-\infhy,b]$ deformation rehracts onto $M^a=f^{-1}(-\ybfty,a]$. It foluows that uhz critical balues (i.e. the values $f(p)$ of $f$ at ghe cxitical poibtw $p$) dre the stactpoinns and endpoikns of tve intetval decomposibion mf $Y_\ast S$. List the critiral values as $v_1<v_2<\cdotf<v_k$. We maf uow use the descriptiin of mdgnidude qs ghe fmltsred Eklec characterjstic : $$|t(H_\asr S)|=\sum_{i=1}^k\chi(M^{v_i})(e^{-v_it}-e^{-f_{i+1}e}),$$ where $v_{k+1}$ is jnterpweeed as $\infty$. Another basic result of Mogse fheory [@milnor2016morse Theoeem 3.2, Remark 3.3 & Remark 3.4] states tre following. Suppose that $b$ is a critical value ox $f$, aid $a<b$ if wufh that there are no critical values of $f$ in $(w,g)$, snc let $p_1,\ldots,p_r$ be the criyifak points of $f$ dith cxjtjcal value $b$. Then $L^b$ has w subwpace of uhe fprm $M^a\cup e^{{\operatorname{ind}}(p_1)}\xup\cdots\cup v^{{\opeeatorname{ind}}(p_r)}$, and $M^b$ deformacion rgtractx onto $M^a\cup e^{{\operatorncme{ind}}(l_1)}\cup\cdots\cuo e^{{\operatkfname{ind}}(p_r)}$. Using thps rasult, we

(s) = f^{-1}(-\infty,s]$. Then the magnitude function sublevel persistent homology S\colon({\mathbb{R}},\leq)\to{\mathbf{GrVect}}$ is expressed where sum is over critical points of A basic result of Morse theory Theorem 3.1] states that if $a<b$ are real numbers such that $f^{-1}(a,b]$ contains critical points of $f$, then $M^b=f^{-1}(-\infty,b]$ deformation retracts onto $M^a=f^{-1}(-\infty,a]$. It follows that critical (i.e. values of $f$ at the critical points $p$) are the startpoints and endpoints of the interval decomposition $H_\ast S$. List the critical values as $v_1<v_2<\cdots<v_k$. may now use the of magnitude as the filtered characteristic $$|t(H_\ast S)|=\sum_{i=1}^k\chi(M^{v_i})(e^{-v_it}-e^{-v_{i+1}t}),$$ $v_{k+1}$ interpreted $\infty$. Another basic of Morse theory [@milnor2016morse Theorem 3.2, Remark 3.3 & Remark 3.4] states the following. Suppose that $b$ a critical $f$, and is that are no critical $f$ in $(a,b)$, and let $p_1,\ldots,p_r$ points of $f$ with critical value $b$. Then has a of the form $M^a\cup e^{{\operatorname{ind}}(p_1)}\cup\cdots\cup e^{{\operatorname{ind}}(p_r)}$, $M^b$ deformation retracts onto $M^a\cup e^{{\operatorname{ind}}(p_1)}\cup\cdots\cup e^{{\operatorname{ind}}(p_r)}$. Using result, we

(s) = f^{-1}(-\infty,s]$. Then the magnitude fUnction of tHe subLevEl sEt PersIsteNt homology $H_\ast s\ColoN({\mathbb{R}},\leq)\to{\mathbf{GrVEct}}$ is ExPRessED aS follOws: $$|t(M,f)|_{\mAThRM{morSe}}=|T(H_\Ast s)|=\sUM_{p} (-1)^{{\OperaTorName{ind}}(P)}e^{-f(p)t}$$ where The SuM is over all crITiCal points oF $f$. A Basic result oF MoRse theOrY [@miLNor2016moRse theorEm 3.1] statES that iF $a<b$ are reaL nUMbers sUCh that $f^{-1}(A,B]$ CoNtaiNs no critical pointS Of $F$, Then $M^b=f^{-1}(-\infty,b]$ dEformaTiON rETRacTs oNto $M^a=f^{-1}(-\inftY,a]$. it folLOws that THe CRITicAL values (i.e. the vAlues $f(p)$ of $f$ aT The CriticAl PoiNTs $p$) are The stArTPoiNts and endpoInts Of the inteRval deCOmpositIOn of $H_\asT S$. List The CriTicaL VaLuEs aS $v_1<V_2<\CdoTS<v_K$. We MAy nOw use the DeScRiptiOn of MAGNItudE as The fIlterEd Euler characTerIstiC : $$|T(H_\aSt S)|=\suM_{i=1}^k\chI(M^{v_i})(E^{-v_It}-e^{-v_{i+1}T}),$$ where $V_{k+1}$ is iNtErpreted as $\infty$. anotHer basic rEsuLt Of MOrSe theORy [@milnOr2016mOrsE TheoreM 3.2, Remark 3.3 & rEmaRk 3.4] STATeS the following. SuppoSe THAt $B$ is a critIcal vaLUe Of $F$, And $a<b$ is sUcH thAt thERE are nO criTIcAl values Of $f$ in $(a,B)$, AnD lEt $p_1,\ldotS,p_R$ be the CrItiCal PointS Of $f$ wIth criTical valUe $b$. ThEN $M^b$ has a subspacE Of the form $M^a\cuP E^{{\oPERaTOrnaMe{iNd}}(p_1)}\cup\cdots\Cup e^{{\OPeraTornAMe{Ind}}(P_R)}$, and $M^B$ defoRmATiON retracts onto $M^a\cup e^{{\OpEratorName{iNd}}(p_1)}\cup\cdots\cuP e^{{\operatorNAME{ind}}(p_r)}$. UsIng tHIs REsult, we

(s) = f^{-1}(-\infty,s]$.Then the m agnit ude fu nc tion ofthe sublevel s e t pe rsistent homology $H_\ ast S \c o lon( { \m athbb {R}},\l e q) \ t o{\ ma th bf{ Gr V ec t}}$isexpress ed as foll ows :$$|t(M,f)|_{ \ ma thrm{Morse }}= |t(H_\ast S) |=\ sum_{p }(-1 ) ^{{\o per atorn ame{in d }}(p)} e^{-f(p)t }$ $ where the sum i sover all critical poi n ts of $f$. A bas ic res ul t o f Mor setheory [@m il nor20 1 6morseT he o r e m 3 . 1] states tha t if $a<b$a rereal n um ber s suchthat$f ^ {-1 }(a,b]$ con tain s no crit ical p o ints of $f$, th en $M^ b=f ^{- 1}(- \ in ft y,b ]$ def o rm ati o n r etractson to $M^a =f^{ - 1 } ( -\in fty ,a]$ . Itfollows thatthe cri t ica l val ues ( i.e.  t he va lues $ f(p)$ o f $f$ at the cr itic al points $p $) ar ethe s t artpoi nts an d endpo ints of the i n t e rv al decomposition o f$ H _\ ast S$.List t h ecr i tical va lu esas $ v _ 1<v_2 <\cd o ts <v_k$. W e mayn ow u se thede script io n o f m agnit u de a s thefiltered Eule r characteristi c : $$|t(H_\as t S ) | =\ s um_{ i=1 }^k\chi(M^{ v_i} ) (e^{ -v_i t }- e^{ - v_{i+ 1}t}) ,$ $ w h ere $v_{k+1}$ is in te rprete d as$\infty$. Ano ther basic r e sult ofMors e t h eory [@milnor2 016mo rse Theore m 3.2, Re mark3.3 & Re mark 3.4] s tates th e f oll owi ng. S up pose that $b$ i s acr iticalval ue of $ f$, an d $ a<b $is such t hat ther ear eno cr itica l valuesof $f $in$(a,b ) $, and let$p_1 ,\ ld o ts, p_r$ be th e crit ic al poi nts o f $f$ wit h cr iticalvalue $b$ . T h en $ M^ b$ has asubspace of t he form $M^a \c upe^{{\o p e ratornam e{ind}}(p_1)}\cup\cdots \ cup e^{ {\o perat orna me{ind}}( p_r )}$, a nd$ M^b$ d eforma tionre tra c t s ont o $M ^a\ cu p e^{{\ope r a tor name{ in d}}( p_1)}\c up\cdots\cup e^{{\ o per atorname{ind} }(p _r)} $ . U sin g t h isre s ult , we

(s) =_f^{-1}(-\infty,s]$. Then_the magnitude function of_the sublevel_set_persistent homology_$H_\ast_S\colon({\mathbb{R}},\leq)\to{\mathbf{GrVect}}$ is expressed_as follows: $$|t(M,f)|_{\mathrm{Morse}}=|t(H_\ast_S)|=\sum_{p} (-1)^{{\operatorname{ind}}(p)}e^{-f(p)t}$$ where the_sum is over_all_critical points of $f$. A basic result of Morse theory [@milnor2016morse Theorem 3.1] states that_if_$a<b$ are_real_numbers_such that $f^{-1}(a,b]$ contains no_critical points of $f$, then_$M^b=f^{-1}(-\infty,b]$ deformation_retracts onto $M^a=f^{-1}(-\infty,a]$. It follows that the critical_values_(i.e. the values $f(p)$_of $f$ at the critical points $p$) are the_startpoints and endpoints of the interval_decomposition of $H_\ast_S$._List_the critical values as_$v_1<v_2<\cdots<v_k$. We may now use the_description of magnitude as the filtered_Euler characteristic : $$|t(H_\ast S)|=\sum_{i=1}^k\chi(M^{v_i})(e^{-v_it}-e^{-v_{i+1}t}),$$ where $v_{k+1}$_is interpreted as $\infty$. Another basic_result of Morse theory [@milnor2016morse_Theorem 3.2,_Remark 3.3 & Remark 3.4]_states the following._Suppose that_$b$ is a_critical value of $f$, and $a<b$_is such that_there are no critical values of_$f$_in $(a,b)$, and_let_$p_1,\ldots,p_r$_be the_critical points of_$f$_with critical_value_$b$. Then $M^b$ has a subspace_of_the form $M^a\cup e^{{\operatorname{ind}}(p_1)}\cup\cdots\cup e^{{\operatorname{ind}}(p_r)}$, and $M^b$_deformation retracts onto $M^a\cup_e^{{\operatorname{ind}}(p_1)}\cup\cdots\cup_e^{{\operatorname{ind}}(p_r)}$. Using this result,_we

in the Belousov-Zhabotinsky reaction [@Sirimungkala].\ Designing appropriate controllers of these reaction-diffusion systems can reveal of great relevance within a reverse engineering approach for example towards the optimization of discharge-charge of lithium batteries (by for example enhancing the formation of solid phases during discharge more reversible upon charge) and the optimization of the structure of the fabricated electrodes as function of the fabrication parameters (e.g. temperature dynamics, reactant flow, etc.).\ In this paper, we consider the one-dimensional Allen-Cahn equation $$\begin{aligned} {\frac{\textstyle \partial u} {\textstyle \partial t}} -{\frac{\textstyle \partial^2 u} {\textstyle \partial x^2}} +{\frac{\textstyle 1} {\textstyle \epsilon^2}}f(u)=0 & x\in]0,1[, t >0,\\ u_x(0,t)=\alpha(t), u_x(1,t)=0& \forall t >0,\\ u(x,0)=u_0(x) & x\in ]0,1[.\end{aligned}$$\ This reaction-diffusion equation describes the process of phase separation in many situations. It was originally introduced in [@AllenCahn] by Allen and Cahn to model the motion of anti-phase boundaries in crystalline solids. In equation (1), $u$ represents the concentration of one of the possible phases, $\epsilon$ represents the interfacial width, supposed to be small as compared to the characteristic length of the laboratory scale. The homogenous Neumann boundary condition (when $\alpha(t)=0$) traduces that there is no loss of mass across the boundary walls. However, the Allen-Cahn equation is invoqued in a large number of complicated moving interface problems in materials science through a phase-field approach, therefore a large litterature in mathematical analysis and in numerical analysis is devoted to the study of the mathematical properties of this equation and of its simulation (see [@MPierre; @JShen] and the references therein).\ In equation (1), $f(u)$ represents the potential energy and $\alpha(t)$ represents the control flux at one of the boundaries; $f(u)$ is assumed having stable roots $\rho_i$, $i=1,\cdots, r$ such that $f(\rho_i)=0$ and $f'(\rho_i)>0$. It

in the Belousov - Zhabotinsky reaction [ @Sirimungkala].\ Designing appropriate controllers of these reaction - dissemination system can reveal of big relevance within a inverse engineering approach for exercise towards the optimization of release - charge of lithium battery (by for example enhancing the geological formation of solid phases during release more reversible upon bang) and the optimization of the structure of the manufacture electrodes as function of the lying parameters (e.g. temperature dynamics, reactant stream, etc.).\ In this paper, we consider the one - dimensional Allen - Cahn equality $ $ \begin{aligned } { \frac{\textstyle \partial u } { \textstyle \partial t } } -{\frac{\textstyle \partial^2 uranium } { \textstyle \partial x^2 } } + { \frac{\textstyle 1 } { \textstyle \epsilon^2}}f(u)=0 & x\in]0,1 [, t > 0,\\ u_x(0,t)=\alpha(t), u_x(1,t)=0 & \forall t > 0,\\ u(x,0)=u_0(x) & x\in ] 0,1[.\end{aligned}$$\ This chemical reaction - diffusion equation describes the process of phase legal separation in many situations. It was originally introduced in [ @AllenCahn ] by Allen and Cahn to model the motion of anti - phase boundaries in crystalline solid. In equation (1), $ u$ represents the concentration of one of the possible phases, $ \epsilon$ represents the interfacial width, supposed to be small as compared to the characteristic length of the lab scale. The homogenous Neumann boundary circumstance (when $ \alpha(t)=0 $) traduces that there be no passing of mass across the boundary walls. However, the Allen - Cahn equality is invoqued in a large numeral of complicated moving interface problems in material science through a phase - field approach, therefore a big litterature in mathematical analysis and in numerical analysis is devote to the study of the numerical property of this equation and of its simulation (see [ @MPierre; @JShen ] and the references therein).\ In equation (1), $ f(u)$ represents the potential department of energy and $ \alpha(t)$ represents the dominance flux at one of the boundaries; $ f(u)$ is bear having stable roots $ \rho_i$, $ i=1,\cdots, r$ such that $ f(\rho_i)=0 $ and $ f'(\rho_i)>0$. It

in the Belousov-Zhabotinsky reaction [@Sirimontkala].\ Dxsigninf appropfiate controllers of these rxactuon-didfusion systems can rexeal of ggeat relecanct within a reversx enginecxing ziproaeh for example tpwards the optimization mf dnscharge-charge of lithium batteries (fy for rxwmple enhancind tht fjrmafpok of solid phases during discharfe more reversible ulon charge) and the optimizwtioj of the structure of the fabtjcaewd electroder as functpmn of the rabrication parameters (e.g. tempefaturz dynamics, tzqctwtt flow, etc.).\ Mn thif paper, we consider dhe one-cimensional Alken-Rahn equation $$\begin{alignev} {\frac{\textstyle \partiwl u} {\textvtvle \partial t}} -{\frac{\texrsryle \kartidl^2 u} {\rexgstglx \pzrtial x^2}} +{\hrac{\textstyme 1} {\textstyoe \epsilon^2}}f(u)=0 & x\in]0,1[, t >0,\\ u_v(0,n)=\slpha(t), u_x(1,t)=0& \fkrall e >0,\\ t(x,0)=u_0(x) & x\in ]0,1[.\end{aligned}$$\ This reaction-diffuspon squation describes the process of phase sepatation in iany situations. It was originally introduced in [@AnlenCehv] bv Allev ajd Cahn to model the motion of anti-phase bounqzroex in crystallike solids. In equatooj (1), $o$ represents tfe conesnfration of one of hhe posfible phases, $\eksilom$ represents the interfaciao width, suppjwed to be small as compared tu thg charscteristic length of thz labodatory scalf. The homkeenous Neumann buuncasy condiukon (when $\alpha(t)=0$) twaduces tiat tkere is vo lpss of mass acrods thc boundary walls. Hoaever, tve Allen-Cajn equation is invoqued in a lacje number of vokplpcated moring ikterface probleis in materialx scienee thruugh a phaae-field approach, trerefore a lashe litteratuce in matremaricao analyrks and in numetical analysis is dwvoted to the studn of jhs mathematical 'xokeeties of this rquxtijn aid of hts simulatimn (sde [@OLierrd; @JShen] and ghe teferences therein).\ In equztion (1), $f(u)$ represenys the potgntial enqrgy and $\alphs(t)$ represents the fontrml hlux ay oge of the boundaries; $f(u)$ is asshmed havijg ftable roots $\rho_l$, $i=1,\cqots, r$ suck that $f(\rho_i)=0$ and $f'(\rho_i)>0$. It

in the Belousov-Zhabotinsky reaction [@Sirimungkala].\ Designing appropriate these systems can of great relevance for towards the optimization discharge-charge of lithium (by for example enhancing the formation solid phases during discharge more reversible upon charge) and the optimization of the of the fabricated electrodes as function of the fabrication parameters (e.g. temperature dynamics, flow, In paper, consider the one-dimensional Allen-Cahn equation $$\begin{aligned} {\frac{\textstyle \partial u} {\textstyle \partial t}} -{\frac{\textstyle \partial^2 u} {\textstyle x^2}} +{\frac{\textstyle 1} {\textstyle \epsilon^2}}f(u)=0 & x\in]0,1[, t u_x(0,t)=\alpha(t), u_x(1,t)=0& \forall t u(x,0)=u_0(x) & x\in ]0,1[.\end{aligned}$$\ This equation the process phase in situations. It was introduced in [@AllenCahn] by Allen and Cahn to model the motion of anti-phase boundaries in crystalline solids. equation (1), the concentration one the phases, $\epsilon$ represents width, supposed to be small as characteristic length of the laboratory scale. The homogenous boundary condition $\alpha(t)=0$) traduces that there is no of mass across the boundary walls. However, the equation is invoqued in a large number of complicated moving interface problems in materials science phase-field approach, therefore a litterature in mathematical and numerical is to the of the mathematical properties of this equation and of its simulation [@MPierre; @JShen] and the references therein).\ In equation (1), $f(u)$ potential and $\alpha(t)$ represents control flux at one the $f(u)$ is assumed having $\rho_i$, r$ and It

in the Belousov-Zhabotinsky rEaction [@SirImungKalA].\ DeSiGninG appRopriate controLLers Of these reaction-diffusiOn sysTeMS can REvEal of Great reLEvANCe wItHiN a rEvERsE engiNeeRing appRoach for exAmpLe Towards the opTImIzation of dIscHarge-charge oF liThium bAtTerIEs (by fOr eXamplE enhanCIng the Formation Of SOlid phASes duriNG DiSchaRge more reversible UPoN Charge) and the opTimizaTiON oF THe sTruCture of the FaBricaTEd electROdES AS fuNCtion of the fabRication parAMetErs (e.g. tEmPerATure dyNamicS, rEActAnt flow, etc.).\ IN thiS paper, we cOnsideR The one-dIMensionAl AlleN-CaHn eQuatIOn $$\BeGin{AlIGneD} {\FrAc{\tEXtsTyle \partIaL u} {\TextsTyle \PARTIal t}} -{\FraC{\texTstylE \partial^2 u} {\textStyLe \paRTiaL x^2}} +{\fraC{\textStylE 1} {\tExtstYle \epsIlon^2}}f(U)=0 & x\In]0,1[, t >0,\\ u_x(0,t)=\alpha(t), u_x(1,t)=0& \ForaLl t >0,\\ u(x,0)=u_0(x) & x\iN ]0,1[.\enD{aLigNeD}$$\ This REactioN-diFfuSion equAtion deSCriBeS THE pRocess of phase separAtION iN many sitUationS. it WaS OriginalLy IntRoduCED in [@AlLenCAHn] By Allen aNd Cahn TO mOdEl the moTiOn of anTi-PhaSe bOundaRIes iN crystAlline soLids. IN Equation (1), $u$ repreSEnts the concenTRaTIOn OF one Of tHe possible pHaseS, $\EpsiLon$ rEPrEseNTs the InterFaCIaL Width, supposed to be smAlL as comPared To the characteRistic lengTH OF the laboRatoRY sCAle. The homogenoUs NeuMann boundaRY conditiOn (wheN $\alpha(t)=0$) tRaduces thAT There is nO loSs oF maSs aCROsS the boundary wALLs. HoWeVer, the ALleN-Cahn eqUatIon Is iNvoQuEd in a largE number oF cOmPlIcAteD moviNG interfaCe ProBlEms In matERials sCiencE thrOuGh A PhaSe-field APpROAch, tHeReFore A laRgE littEratURe iN mathemAtical anaLysIS and In NuMerical Analysis is devOtEd to the stuDy Of tHe mathEMAtical prOperties of this equation aND of its sImuLatioN (see [@mPierre; @JSHen] And the RefERences ThereiN).\ In eqUaTioN (1), $F(U)$ reprESEnTs tHe Potential eNERgy And $\alPhA(t)$ rePresentS the control flux at oNE of The boundaries; $F(u)$ iS assUMEd HavINg STabLe ROotS $\RHo_i$, $i=1,\cdots, r$ such tHat $f(\rho_i)=0$ anD $f'(\RHo_I)>0$. It

in the Belousov-Zhabotins ky reactio n [@S iri mun gk ala] .\ D esigning appro p riat e controllers of these reac ti o n-di f fu sionsystems ca n rev ea lofgr e at rele van ce with in a rever seen gineering ap p ro ach for ex amp le towards t heoptimi za tio n of d isc harge -charg e of li thium bat te r ies (b y for ex a m pl e en hancing the forma t io n of solid phas es dur in g d i s cha rge more reve rs ibleu pon cha r ge ) a ndt he optimizati on of the s t ruc ture o fthe fabric atedel e ctr odes as fun ctio n of thefabric a tion pa r ameters (e.g. te mpe ratu r edy nam ic s , r e ac tan t fl ow, etc. ). \In th is p a p e r , we co nsid er th e one-dimensi ona l Al l en- Cahnequat ion$$ \begi n{alig ned}{\ frac{\textstyle \pa rtial u}{\t ex tst yl e \pa r tial t }}-{\ frac{\t extstyl e \p ar t i a l^ 2 u} {\textstyle \ pa r t ia l x^2}}+{\fra c {\ te x tstyle 1 }{\t exts t y le \e psil o n^ 2}}f(u)= 0 & x\ i n] 0, 1[, t > 0, \\ u_x (0 ,t) =\a lpha( t ), u _x(1,t )=0& \fo rallt >0,\\ u(x,0)= u _0(x) & x\in] 0, 1 [ .\ e nd{a lig ned}$$\ Thi s re a ctio n-di f fu sio n equa tionde s cr i bes the process ofph ase se parat ion in many s ituations. I t was ori gina l ly introduced in[@All enCahn] by Allen an d Cah n to mod el the mo t i on of an ti- pha sebou n d ar ies in crysta l l ineso lids. I n e quation (1 ),$u$ re pr esents th e concen tr at io nofone o f the pos si ble p has es, $ \ epsilo n$ re pres en ts the interf a ci a l wid th ,supp ose dto be sma l l a s compa red to th e c h arac te ri stic le ngth of the l ab oratory sc al e.The ho m o genous N eumann boundary conditi o n (when $\ alpha (t)= 0$) tradu ces thatthe r e is n o loss of m as s a c r oss t h e b oun da ry walls.H o wev er, t he All en-Cahn equation is invoq u edin a large nu mbe r of c om pli c at e d m ov i ngi n terface problem s in mater ia l sscience th r oug ha phase -fieldappro a ch, the refore alarge lit te ratu r e in mathemati cal anal ysis andi n num e ri cal a nal ysis i sdev otedto the stu dy of the m at hemati cal p ro pertiesof this equation and of its s imula tio n (see [@ MPi e rre ; @JShen] and the refer enc esthere in) . \ Inequa t io n ( 1 ), $f (u)$ represent s t hep o te ntial energ y a nd$\alp ha( t )$ rep rese nts the control f l ux at one of t he b o u nda rie s ; $f (u )$ is assumedhav in g stable r oo ts $\rho_i$ , $i=1,\ cd o ts, r $ such that$f(\rho _ i )= 0 $ and$f'( \rh o_i)>0$.It

in_the Belousov-Zhabotinsky_reaction [@Sirimungkala].\ Designing appropriate controllers_of these_reaction-diffusion_systems can_reveal_of great relevance_within a reverse_engineering approach for example_towards the optimization_of_discharge-charge of lithium batteries (by for example enhancing the formation of solid phases during_discharge_more reversible_upon_charge)_and the optimization of the_structure of the fabricated electrodes_as function_of the fabrication parameters (e.g. temperature dynamics, reactant_flow,_etc.).\ In this paper,_we consider the one-dimensional Allen-Cahn equation $$\begin{aligned} {\frac{\textstyle \partial u}_{\textstyle \partial t}} -{\frac{\textstyle \partial^2 u}_{\textstyle \partial x^2}}_+{\frac{\textstyle_1}_{\textstyle \epsilon^2}}f(u)=0 & x\in]0,1[,_t >0,\\ u_x(0,t)=\alpha(t), u_x(1,t)=0& \forall t >0,\\ u(x,0)=u_0(x)_& x\in ]0,1[.\end{aligned}$$\ This reaction-diffusion equation describes_the process of phase separation in many_situations. It was originally introduced in_[@AllenCahn] by Allen and Cahn_to model_the motion of anti-phase boundaries_in crystalline solids._In equation_(1), $u$ represents_the concentration of one of the_possible phases, $\epsilon$_represents the interfacial width, supposed to_be_small as compared_to_the_characteristic length_of the laboratory_scale._The homogenous_Neumann_boundary condition (when $\alpha(t)=0$) traduces that_there_is no loss of mass across the_boundary walls. However, the_Allen-Cahn_equation is invoqued in_a large number of complicated_moving interface problems in materials science_through a_phase-field approach,_therefore a large litterature in mathematical analysis and in numerical analysis_is devoted to the study of_the mathematical properties of_this equation_and_of its simulation_(see_[@MPierre; @JShen]_and the references therein).\ In equation (1), $f(u)$_represents the_potential energy and $\alpha(t)$ represents the_control flux at one_of_the boundaries; $f(u)$ is assumed having_stable roots $\rho_i$, $i=1,\cdots, r$ such_that $f(\rho_i)=0$ and $f'(\rho_i)>0$. It

, we let $\Omega_{1}:=[0, N_{1}^{2}]\times [0, \lambda N_{1}]^{d}$. Note that one can use finite overlapped balls of radius $N_{1}^{2}$ to cover $\Omega_{1}$ since $\lambda \geq N_{1}$. We want to prove $$\label{de1} \|Ef_{1} Ef_{2}\|_{L^2_{avg}(\omega_{\Omega_{1}})} \lesssim_{\epsilon} \lambda^{d/2}N_{2}^{\epsilon}\frac{N_{2}^{d-1}}{N_{1}}\prod_{j=1}^2\left(\sum_{|\theta|=\frac{1 }{\lambda N_1}}\|Ef_{j,\theta}\|_{L^4_{avg}(w_{\Omega_{1}})}^2\right)^{1/2}.$$ We first apply Corollary \[Tcor\] with $\delta= N_1^{-2}$, $\upsilon = \frac{N_2}{N_1}$, $R= N_{1}^{2}$. Note that $\delta\leq v$. Then we have $$\label{1111} \begin{aligned} \int |Ef_1 Ef_2|^2w_{B_{ N^{2}_{1}}}&\lesssim (N_1N_2)^{d-1}\Big|\frac{\log N_1}{\log N_1 -\log N_2}\Big|^{C} \sum_{|\theta_j|= \frac{1}{N_1^2}} \int |Ef_{1, \theta_1} Ef_{2,\theta_2}|^2 w_{B_{N^{2}_{1}}}\\ &\lesssim (N_1 N_2)^{d-1}N_{2}^{\epsilon} \sum_{|\theta_j|= \frac{1}{N_1^2}} \prod_{j=1}^2\|Ef_{j,\theta_j}\|_{L^4(w_{B_{ N^{2}_{1}}})}^2. \end{aligned}$$ \[rrr\] We avoid the case when $N_1=N_2$, and thus $\ln N_{1}-\ln N_{2}=0$, by first decomposing caps of diameter $N_{2}/N_{1}$ into caps of diameter $N_{

, we let $ \Omega_{1}:=[0, N_{1}^{2}]\times [ 0, \lambda N_{1}]^{d}$. Note that one can use finite overlapped ball of spoke $ N_{1}^{2}$ to cover $ \Omega_{1}$ since $ \lambda \geq N_{1}$. We want to rise $ $ \label{de1 } \|Ef_{1 } Ef_{2}\|_{L^2_{avg}(\omega_{\Omega_{1 } }) } \lesssim_{\epsilon } \lambda^{d/2}N_{2}^{\epsilon}\frac{N_{2}^{d-1}}{N_{1}}\prod_{j=1}^2\left(\sum_{|\theta|=\frac{1 } { \lambda N_1}}\|Ef_{j,\theta}\|_{L^4_{avg}(w_{\Omega_{1}})}^2\right)^{1/2}.$$ We foremost apply Corollary \[Tcor\ ] with $ \delta= N_1^{-2}$, $ \upsilon = \frac{N_2}{N_1}$, $ R= N_{1}^{2}$. Note that $ \delta\leq v$. Then we have $ $ \label{1111 } \begin{aligned } \int |Ef_1 Ef_2|^2w_{B _ { N^{2}_{1}}}&\lesssim (N_1N_2)^{d-1}\Big|\frac{\log N_1}{\log N_1 -\log N_2}\Big|^{C } \sum_{|\theta_j|= \frac{1}{N_1 ^ 2 } } \int |Ef_{1, \theta_1 } Ef_{2,\theta_2}|^2 w_{B_{N^{2}_{1}}}\\ & \lesssim (N_1 N_2)^{d-1}N_{2}^{\epsilon } \sum_{|\theta_j|= \frac{1}{N_1 ^ 2 } } \prod_{j=1}^2\|Ef_{j,\theta_j}\|_{L^4(w_{B _ { N^{2}_{1}}})}^2. \end{aligned}$$ \[rrr\ ] We avoid the font when $ N_1 = N_2 $, and thus $ \ln N_{1}-\ln N_{2}=0 $, by first decompose caps of diameter $ N_{2}/N_{1}$ into caps of diameter $ N _ {

, we let $\Omega_{1}:=[0, N_{1}^{2}]\times [0, \lambdx N_{1}]^{d}$. Note that one can use fjnite ovdrlapped balls of radius $N_{1}^{2}$ tl xover $\Omega_{1}$ since $\lambda \ged N_{1}$. We wajt to prive $$\oqbel{de1} \|Ef_{1} Eh_{2}\|_{M^2_{avg}(\omenc_{\Omegz_{1}})} \lesvwim_{\epsilon} \lamnda^{d/2}N_{2}^{\epsilot}\frac{N_{2}^{d-1}}{N_{1}}\prod_{j=1}^2\lafg(\sbm_{|\theta|=\frac{1 }{\lambda N_1}}\|Ef_{j,\theta}\|_{L^4_{avg}(w_{\Omeda_{1}})}^2\right)^{1/2}.$$ Wf first apply Sorokjary \[Ncir\] with $\delta= N_1^{-2}$, $\upsilon = \frac{H_2}{N_1}$, $R= N_{1}^{2}$. Iote that $\delta\keq v$. Then we have $$\label{1111} \behin{apigned} \int |Ef_1 Ef_2|^2w_{B_{ J^{2}_{1}}}&\lesssim (N_1N_2)^{e-1}\Big|\seac{\log N_1}{\log V_1 -\log N_2}\Big|^{C} \sum_{|\theta_n|= \frac{1}{N_1^2}} \int |Ef_{1, \theta_1} Ef_{2,\theta_2}|^2 w_{B_{V^{2}_{1}}}\\ &\lessxim (N_1 N_2)^{d-1}N_{2}^{\eksjllt} \sum_{|\theta_j|= \frac{1}{G_1^2}} \prod_{j=1}^2\|Ef_{j,\thcna_j}\|_{L^4(w_{B_{ T^{2}_{1}}})}^2. \end{alogned}$$ \[rrr\] We avold thx cawe when $N_1=N_2$, and thus $\li N_{1}-\ln N_{2}=0$, by first decjmposing wa's of diameter $N_{2}/N_{1}$ inti xaps mf dhamegwr $V_{

, we let $\Omega_{1}:=[0, N_{1}^{2}]\times [0, \lambda that can use overlapped balls of since \geq N_{1}$. We to prove $$\label{de1} Ef_{2}\|_{L^2_{avg}(\omega_{\Omega_{1}})} \lesssim_{\epsilon} \lambda^{d/2}N_{2}^{\epsilon}\frac{N_{2}^{d-1}}{N_{1}}\prod_{j=1}^2\left(\sum_{|\theta|=\frac{1 }{\lambda N_1}}\|Ef_{j,\theta}\|_{L^4_{avg}(w_{\Omega_{1}})}^2\right)^{1/2}.$$ We apply Corollary \[Tcor\] with $\delta= N_1^{-2}$, $\upsilon = \frac{N_2}{N_1}$, $R= N_{1}^{2}$. Note that v$. Then we have $$\label{1111} \begin{aligned} \int |Ef_1 Ef_2|^2w_{B_{ N^{2}_{1}}}&\lesssim (N_1N_2)^{d-1}\Big|\frac{\log N_1}{\log N_1 N_2}\Big|^{C} \frac{1}{N_1^2}} |Ef_{1, Ef_{2,\theta_2}|^2 w_{B_{N^{2}_{1}}}\\ &\lesssim (N_1 N_2)^{d-1}N_{2}^{\epsilon} \sum_{|\theta_j|= \frac{1}{N_1^2}} \prod_{j=1}^2\|Ef_{j,\theta_j}\|_{L^4(w_{B_{ N^{2}_{1}}})}^2. \end{aligned}$$ \[rrr\] We avoid the case when and thus $\ln N_{1}-\ln N_{2}=0$, by first decomposing of diameter $N_{2}/N_{1}$ into of diameter $N_{

, we let $\Omega_{1}:=[0, N_{1}^{2}]\times [0, \lambda N_{1}]^{d}$. NOte that one Can usE fiNitE oVerlAppeD balls of radius $n_{1}^{2}$ To coVer $\Omega_{1}$ since $\lambda \geq n_{1}$. We waNt TO proVE $$\lAbel{dE1} \|Ef_{1} Ef_{2}\|_{L^2_{aVG}(\oMEGa_{\OMeGa_{1}})} \LesSsIM_{\ePsiloN} \laMbda^{d/2}N_{2}^{\ePsilon}\frac{n_{2}^{d-1}}{N_{1}}\PrOd_{j=1}^2\left(\sum_{|\thETa|=\Frac{1 }{\lambda n_1}}\|Ef_{J,\theta}\|_{L^4_{avg}(w_{\OMegA_{1}})}^2\right)^{1/2}.$$ we FirST applY CoRollaRy \[Tcor\] WIth $\delTa= N_1^{-2}$, $\upsiloN = \fRAc{N_2}{N_1}$, $R= N_{1}^{2}$. nOte that $\DELtA\leq V$. Then we have $$\label{1111} \bEGiN{Aligned} \int |Ef_1 Ef_2|^2W_{B_{ N^{2}_{1}}}&\lesSsIM (N_1n_2)^{D-1}\big|\FraC{\log N_1}{\log N_1 -\lOg n_2}\Big|^{C} \SUm_{|\theta_J|= \FrAC{1}{n_1^2}} \Int |eF_{1, \theta_1} Ef_{2,\theta_2}|^2 W_{B_{N^{2}_{1}}}\\ &\lesssim (N_1 n_2)^{D-1}N_{2}^{\ePsilon} \SuM_{|\thETa_j|= \fraC{1}{N_1^2}} \proD_{j=1}^2\|eF_{j,\tHeta_j}\|_{L^4(w_{B_{ N^{2}_{1}}})}^2. \enD{aliGned}$$ \[rrr\] We Avoid tHE case whEN $N_1=N_2$, and tHus $\ln N_{1}-\Ln N_{2}=0$, By fIrst DEcOmPosInG CapS Of DiaMEteR $N_{2}/N_{1}$ into cApS oF diamEter $n_{

, we let $\Omega_{1}:=[0,N_{1}^{2}] \time s [ 0,\l ambd a N_ {1}]^{d}$. Not e tha t one can use finite o verla pp e d ba l ls of r adius $ N _{ 1 } ^{2 }$ t o c ov e r$\Ome ga_ {1}$ si nce $\lamb da\g eq N_{1}$. W e w ant to pro ve$$\label{de1 } \ |Ef_{1 }Ef_ { 2}\|_ {L^ 2_{av g}(\om e ga_{\O mega_{1}} )} \lesss i m_{\eps i l on } \l ambda^{d/2}N_{2}^ { \e p silon}\frac{N_ {2}^{d -1 } }{ N _ {1} }\p rod_{j=1}^ 2\ left( \ sum_{|\ t he t a | =\f r ac{1 }{\lamb da N_1}}\|E f _{j ,\thet a} \|_ { L^4_{a vg}(w _{ \ Ome ga_{1}})}^2 \rig ht)^{1/2} .$$ We first a p ply Cor ollary \[ Tco r\]w it h$\d el t a=N _1 ^{- 2 }$, $\upsil on = \fra c{N_ 2 } { N _1}$ , $ R= N _{1}^ {2}$. Note th at$\de l ta\ leq v $. Th en w ehave$$\lab el{11 11 } \begin{aligne d} \ int |Ef_1 Ef _2 |^2 w_ {B_{N ^{2}_{ 1}} }&\ lesssim (N_1N_ 2 )^{ d- 1 } \ Bi g|\frac{\log N_1}{ \l o g N _1 -\log N_2} \ Bi g| ^ {C} \su m_ {|\ thet a _ j|= \ frac { 1} {N_1^2}} \int| Ef _{ 1, \the ta _1} Ef _{ 2,\ the ta_2} | ^2 w _{B_{N ^{2}_{1} }}\\& \lesssim (N_1N _2)^{d-1}N_{2 } ^{ \ e ps i lon} \s um_{|\theta _j|= \fra c{1} { N_ 1^2 } } \pr od_{j =1 } ^2 \ |Ef_{j,\theta_j}\|_ {L ^4(w_{ B_{ N ^{2}_{1}}})}^ 2. \end{a l i g ned}$$ \[rr r \] We avoid thecasewhen $N_1= N _2$, and thus $\ln N_ {1}-\ln N _ { 2}=0$, b y f irs t d eco m p os ing caps of d i a mete r$N_{2}/ N_{ 1}$ int o c aps of di am eter $N_{

, we_let $\Omega_{1}:=[0,_N_{1}^{2}]\times [0, \lambda N_{1}]^{d}$._Note that_one_can use_finite_overlapped balls of_radius $N_{1}^{2}$ to_cover $\Omega_{1}$ since $\lambda_\geq N_{1}$. We_want_to prove $$\label{de1} \|Ef_{1} Ef_{2}\|_{L^2_{avg}(\omega_{\Omega_{1}})} \lesssim_{\epsilon} \lambda^{d/2}N_{2}^{\epsilon}\frac{N_{2}^{d-1}}{N_{1}}\prod_{j=1}^2\left(\sum_{|\theta|=\frac{1 }{\lambda N_1}}\|Ef_{j,\theta}\|_{L^4_{avg}(w_{\Omega_{1}})}^2\right)^{1/2}.$$ We first apply Corollary \[Tcor\] with_$\delta=_N_1^{-2}$, $\upsilon_=_\frac{N_2}{N_1}$,_$R= N_{1}^{2}$. Note that $\delta\leq_v$. Then we have $$\label{1111} \begin{aligned} \int_|Ef_1 Ef_2|^2w_{B_{_N^{2}_{1}}}&\lesssim (N_1N_2)^{d-1}\Big|\frac{\log N_1}{\log N_1 -\log N_2}\Big|^{C} _\sum_{|\theta_j|=_\frac{1}{N_1^2}} \int |Ef_{1,_\theta_1} Ef_{2,\theta_2}|^2 w_{B_{N^{2}_{1}}}\\ &\lesssim (N_1 N_2)^{d-1}N_{2}^{\epsilon} \sum_{|\theta_j|= \frac{1}{N_1^2}} \prod_{j=1}^2\|Ef_{j,\theta_j}\|_{L^4(w_{B_{ N^{2}_{1}}})}^2._ \end{aligned}$$ \[rrr\] We avoid the case when $N_1=N_2$,_and thus $\ln_N_{1}-\ln_N_{2}=0$,_by first decomposing caps_of diameter $N_{2}/N_{1}$ into caps of_diameter $N_{

tau^T{\left|\delta g_s\right|}ds\right]\right)^2&\leq 2\mu^2\left(\mathbb E_\tau\left[\int_\tau^T{\left|\delta z_s\right|}\left({\left|z^1_s\right|}+{\left|z^2_s\right|}\right)ds\right]\right)^2\\ &\hspace{0.9em}+2\mu^2\left(\mathbb E_\tau\left[\int_\tau^T{\left\|\delta u_s\right\|}_{L^2(\nu_s)}\left({\left\|u_s^1\right\|}_{L^2(\nu_s)}+{\left\|u_s^2\right\|}_{L^2(\nu_s)}\right)ds\right]\right)^2\\ &\leq 2\mu^2\left(\mathbb E_\tau\left[\int_\tau^T{\left|\delta z_s\right|}^2ds\right]\mathbb E_\tau\left[\int_\tau^T({\left|z^1_s\right|}+{\left|z^2_s\right|})^2ds\right]\right.\\ &\hspace{0.9em}\left.+\mathbb E_\tau\left[\int_\tau^T{\left\|\delta u_s\right\|}^2_{L^2(\nu_s)}ds\right]\mathbb E_\tau\left[\int_\tau^T\left({\left\|u_s^1\right\|}_{L^2(\nu_s)}+{\left\|u_s^2\right\|}_{L^2(\nu_s)}\right)^2ds\right]\right)\\ &\leq 4R^2\mu^2\left(\mathbb E_\tau\left[\int_\tau^T{\left|\delta z_s\right|}^2ds\right]+\mathbb E_\tau\left[\int_\tau^T\int_E\delta u_s^2(x)\nu(dx)ds\right]\right)\\ &\leq 32R^2\mu^2\left({\left\|\delta z\right\|}_{\mathbb H^2_{\rm{BMO}}}^2+{\left\|\delta u\right\|}_{\mathbb J^2_{\rm{BMO}}}^2\right

tau^T{\left|\delta g_s\right|}ds\right]\right)^2&\leq 2\mu^2\left(\mathbb E_\tau\left[\int_\tau^T{\left|\delta z_s\right|}\left({\left|z^1_s\right|}+{\left|z^2_s\right|}\right)ds\right]\right)^2\\ & \hspace{0.9em}+2\mu^2\left(\mathbb E_\tau\left[\int_\tau^T{\left\|\delta u_s\right\|}_{L^2(\nu_s)}\left({\left\|u_s^1\right\|}_{L^2(\nu_s)}+{\left\|u_s^2\right\|}_{L^2(\nu_s)}\right)ds\right]\right)^2\\ & \leq 2\mu^2\left(\mathbb E_\tau\left[\int_\tau^T{\left|\delta z_s\right|}^2ds\right]\mathbb E_\tau\left[\int_\tau^T({\left|z^1_s\right|}+{\left|z^2_s\right|})^2ds\right]\right.\\ & \hspace{0.9em}\left.+\mathbb E_\tau\left[\int_\tau^T{\left\|\delta u_s\right\|}^2_{L^2(\nu_s)}ds\right]\mathbb E_\tau\left[\int_\tau^T\left({\left\|u_s^1\right\|}_{L^2(\nu_s)}+{\left\|u_s^2\right\|}_{L^2(\nu_s)}\right)^2ds\right]\right)\\ & \leq 4R^2\mu^2\left(\mathbb E_\tau\left[\int_\tau^T{\left|\delta z_s\right|}^2ds\right]+\mathbb E_\tau\left[\int_\tau^T\int_E\delta u_s^2(x)\nu(dx)ds\right]\right)\\ & \leq 32R^2\mu^2\left({\left\|\delta z\right\|}_{\mathbb H^2_{\rm{BMO}}}^2+{\left\|\delta u\right\|}_{\mathbb J^2_{\rm{BMO}}}^2\right

tau^H{\left|\delta g_s\right|}ds\righu]\right)^2&\leq 2\mu^2\left(\mcrhbb E_\vau\left[\jnt_\tau^T{\ldft|\delta z_s\right|}\left({\left|z^1_s\rijht|}+{\lwft|z^2_s\eight|}\right)ds\right]\right)^2\\ &\fspace{0.9em}+2\ml^2\left(\mathvb E_\uau\left[\int_\tau^T{\lefv\|\selta u_s\right\|}_{M^2(\ku_s)}\leyt({\oeft\|u_s^1\right\|}_{L^2(\nu_x)}+{\left\|u_s^2\rigvt\|}_{L^2(\nu_s)}\right)ds\rhgft]\xight)^2\\ &\leq 2\mu^2\left(\mathbb E_\tau\left[\int_\tau^E{\left|\dektw z_s\right|}^2ds\rigrt]\mauhbf E_\tzl\ltft[\int_\tau^T({\left|z^1_s\right|}+{\left|z^2_s\right|})^2da\right]\rpght.\\ &\hspace{0.9em}\left.+\msthbb E_\tau\left[\int_\tau^T{\left\|\dflta u_s\right\|}^2_{L^2(\nu_s)}ds\righh]\mathbb E_\tao\mefe[\unt_\tau^T\left({\ldft\|u_s^1\right\|}_{L^2(\nu_s)}+{\left\|u_s^2\rifht\|}_{L^2(\nu_s)}\right)^2ds\right]\right)\\ &\leq 4R^2\mu^2\ueft(\mcthbb E_\tau\lgyr[\inh_\jau^T{\left|\delte z_s\ridht|}^2ds\right]+\mabnbb E_\tdu\left[\imt_\tau^T\int_E\delts u_v^2(x)\ny(dx)ds\right]\right)\\ &\leq 32R^2\mn^2\left({\left\|\delta z\righj\|}_{\mathbb H^2_{\rk{BJO}}}^2+{\left\|\delta u\righr\|}_{\mqthbb J^2_{\rm{TMO}}}^2\rktht

tau^T{\left|\delta g_s\right|}ds\right]\right)^2&\leq 2\mu^2\left(\mathbb E_\tau\left[\int_\tau^T{\left|\delta z_s\right|}\left({\left|z^1_s\right|}+{\left|z^2_s\right|}\right)ds\right]\right)^2\\ &\hspace{0.9em}+2\mu^2\left(\mathbb E_\tau\left[\int_\tau^T{\left\|\delta 2\mu^2\left(\mathbb z_s\right|}^2ds\right]\mathbb E_\tau\left[\int_\tau^T({\left|z^1_s\right|}+{\left|z^2_s\right|})^2ds\right]\right.\\ E_\tau\left[\int_\tau^T{\left\|\delta u_s\right\|}^2_{L^2(\nu_s)}ds\right]\mathbb E_\tau\left[\int_\tau^T\left({\left\|u_s^1\right\|}_{L^2(\nu_s)}+{\left\|u_s^2\right\|}_{L^2(\nu_s)}\right)^2ds\right]\right)\\ u_s^2(x)\nu(dx)ds\right]\right)\\ 32R^2\mu^2\left({\left\|\delta z\right\|}_{\mathbb H^2_{\rm{BMO}}}^2+{\left\|\delta J^2_{\rm{BMO}}}^2\right

tau^T{\left|\delta g_s\right|}ds\rigHt]\right)^2&\leq 2\Mu^2\lefT(\maThbB E_\Tau\lEft[\iNt_\tau^T{\left|\deltA Z_s\riGht|}\left({\left|z^1_s\right|}+{\left|Z^2_s\rigHt|}\RIght)DS\rIght]\rIght)^2\\ &\hspACe{0.9EM}+2\Mu^2\lEfT(\mAthBb e_\TaU\left[\Int_\Tau^T{\lefT\|\delta u_s\riGht\|}_{l^2(\nU_s)}\left({\left\|u_s^1\RIgHt\|}_{L^2(\nu_s)}+{\left\|U_s^2\rIght\|}_{L^2(\nu_s)}\righT)ds\Right]\rIgHt)^2\\ &\lEQ 2\mu^2\leFt(\mAthbb e_\tau\leFT[\int_\taU^T{\left|\delTa Z_S\right|}^2DS\right]\mATHbB E_\taU\left[\int_\tau^T({\left|z^1_S\RiGHt|}+{\left|z^2_s\right|})^2dS\right]\RiGHt.\\ &\HSPacE{0.9em}\Left.+\mathbb e_\tAu\lefT[\Int_\tau^T{\LEfT\|\DELta U_S\right\|}^2_{L^2(\nu_s)}ds\rIght]\mathbb E_\TAu\lEft[\int_\TaU^T\lEFt({\left\|U_s^1\rigHt\|}_{l^2(\Nu_s)}+{\Left\|u_s^2\right\|}_{l^2(\nu_s)}\Right)^2ds\riGht]\rigHT)\\ &\leq 4R^2\mu^2\LEft(\mathBb E_\tau\LefT[\inT_\tau^t{\LeFt|\DelTa Z_S\riGHt|}^2Ds\rIGht]+\Mathbb E_\tAu\LeFt[\int_\Tau^T\INT_e\DeltA u_s^2(X)\nu(dX)ds\riGht]\right)\\ &\leq 32R^2\mU^2\leFt({\leFT\|\deLta z\rIght\|}_{\mAthbB H^2_{\Rm{BMO}}}^2+{\Left\|\deLta u\rIgHt\|}_{\mathbb J^2_{\rm{BMO}}}^2\rIght

tau^T{\left|\delta g_s\rig ht|}ds\rig ht]\r igh t)^ 2& \leq 2\m u^2\left(\math b b E_ \tau\left[\int_\tau^T{ \left |\ d elta z_ s\rig ht|}\le f t( { \ lef t| z^ 1_s \r i gh t|}+{ \le ft|z^2_ s\right|}\ rig ht )ds\right]\r i gh t)^2\\ &\h spa ce{0.9em}+2\ mu^ 2\left (\ mat h bb E_ \ta u\lef t[\int _ \tau^T {\left\|\ de l ta u_s \ right\| } _ {L ^2(\ nu_s)}\left({\lef t \| u _s^1\right\|}_ {L^2(\ nu _ s) } + {\l eft \|u_s^2\ri gh t\|}_ { L^2(\nu _ s) } \ r igh t )ds\right]\ri ght)^2\\ &\ l eq2\mu^2 \l eft ( \mathb b E_\ ta u \le ft[\int_\ta u^T{ \left|\de lta z_ s \right| } ^2ds\ri ght]\m ath bbE_\t a u\ le ft[ \i n t_\ t au ^T( { \le ft|z^1_s \r ig ht|}+ {\le f t | z ^2_s \ri ght| })^2d s\right]\righ t.\ \ &\ h spa ce{0. 9em}\ left .+ \math bb E_\ tau\l ef t[\int_\tau^T{\ left \|\deltau_s \r igh t\ |}^2_ { L^2(\n u_s )}d s\right ]\mathb b E_ \t a u \ le ft[\int_\tau^T\lef t( { \ le ft\|u_s^ 1\righ t \| }_ { L^2(\nu_ s) }+{ \lef t \ |u_s^ 2\ri g ht \|}_{L^2 (\nu_s ) }\ ri ght)^2d s\ right] \r igh t)\ \ &\l e q 4R ^2\mu^ 2\left(\ mathb b E_\tau\left[\ i nt_\tau^T{\le f t| \ d el t a z_ s\r ight|}^2ds\ righ t ]+\m athb b E _\t a u\lef t[\in t_ \ ta u ^T\int_E\delta u_s^ 2( x)\nu( dx)ds \right]\right )\\ &\leq3 2 R ^2\mu^2\ left ( {\ l eft\|\delta z\ right \|}_{\math b b H^2_{\ rm{BM O}}}^2+{ \left\|\d e l ta u\rig ht\ |}_ {\m ath b b J ^2_{\rm{BMO}} } ^ 2\ri gh t

tau^T{\left|\delta g_s\right|}ds\right]\right)^2&\leq_2\mu^2\left(\mathbb E_\tau\left[\int_\tau^T{\left|\delta_z_s\right|}\left({\left|z^1_s\right|}+{\left|z^2_s\right|}\right)ds\right]\right)^2\\ &\hspace{0.9em}+2\mu^2\left(\mathbb E_\tau\left[\int_\tau^T{\left\|\delta u_s\right\|}_{L^2(\nu_s)}\left({\left\|u_s^1\right\|}_{L^2(\nu_s)}+{\left\|u_s^2\right\|}_{L^2(\nu_s)}\right)ds\right]\right)^2\\ &\leq 2\mu^2\left(\mathbb_E_\tau\left[\int_\tau^T{\left|\delta z_s\right|}^2ds\right]\mathbb_E_\tau\left[\int_\tau^T({\left|z^1_s\right|}+{\left|z^2_s\right|})^2ds\right]\right.\\ &\hspace{0.9em}\left.+\mathbb_E_\tau\left[\int_\tau^T{\left\|\delta u_s\right\|}^2_{L^2(\nu_s)}ds\right]\mathbb_E_\tau\left[\int_\tau^T\left({\left\|u_s^1\right\|}_{L^2(\nu_s)}+{\left\|u_s^2\right\|}_{L^2(\nu_s)}\right)^2ds\right]\right)\\ &\leq_4R^2\mu^2\left(\mathbb E_\tau\left[\int_\tau^T{\left|\delta z_s\right|}^2ds\right]+\mathbb_E_\tau\left[\int_\tau^T\int_E\delta u_s^2(x)\nu(dx)ds\right]\right)\\ &\leq 32R^2\mu^2\left({\left\|\delta_z\right\|}_{\mathbb H^2_{\rm{BMO}}}^2+{\left\|\delta u\right\|}_{\mathbb J^2_{\rm{BMO}}}^2\right

{H}_{1}\oplus\mathcal{H}_{2}}$$ is a C\*-algebra isomorphism, and hence $C^{\ast}\left( \left\{ \tilde{x}_{1}^{\ast}\tilde{x}_{2},x_{1}^{\ast}x_{1}\right\} \right) $ is isomorphic to the pullback $\mathcal{T}\oplus_{C\left( \mathbb{T}\right) }\mathcal{T}$ of two copies of $\mathcal{T}\overset{\sigma}{\rightarrow}C\left( \mathbb{T}\right) $ with $\tilde{x}_{1}^{\ast}\tilde{x}_{2}$ corresponding to $\mathcal{S}\oplus\mathcal{S}$. Proof. Clearly we only need to consider the case with $\mathcal{H}_{0}\neq0$. Since $\tilde{x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}$ is unitary and $C\left( \mathbb{T}\right) $ is the universal C\*-algebra generated by a single unitary generator, there is a unique C\*-algebra homomorphism $$h:C\left( \mathbb{T}\right) \rightarrow C^{\ast}\left( \left\{ \tilde {x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}\right\} \right)$$ sending $\mathrm{id}_{\mathbb{T}}$ to $\tilde{x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}$ while fixing all scalars in $\mathbb{C}\subset C\left( \mathbb{T}\right) $. Clearly with $x_{1}^{\ast}x_{1}|_{\mathcal{H}_{0}}=1$ and $x_{2}^{\ast}x_{2}|_{\mathcal{H}_{0}}=c$, $$h\circ\sigma:C^{\ast}\left( \left\{ \tilde{x}_{1}^{\ast}\tilde{x}_{2},x_{1}^{\ast}x_{1}\right\} \right) |_{\mathcal{H}_{1}\oplus\mathcal{H}_{2}}\rightarrow C^{\ast}\left( \left\{ \tilde{x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}\right\} \

{ H}_{1}\oplus\mathcal{H}_{2}}$$ is a C\*-algebra isomorphism, and hence $ C^{\ast}\left ( \left\ { \tilde{x}_{1}^{\ast}\tilde{x}_{2},x_{1}^{\ast}x_{1}\right\ } \right) $ is isomorphic to the pullback $ \mathcal{T}\oplus_{C\left ( \mathbb{T}\right) } \mathcal{T}$ of two copies of $ \mathcal{T}\overset{\sigma}{\rightarrow}C\left ( \mathbb{T}\right) $ with $ \tilde{x}_{1}^{\ast}\tilde{x}_{2}$ corresponding to $ \mathcal{S}\oplus\mathcal{S}$. validation. intelligibly we only need to think the lawsuit with $ \mathcal{H}_{0}\neq0$. Since $ \tilde{x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}$ is unitary and $ C\left ( \mathbb{T}\right) $ is the universal C\*-algebra beget by a unmarried unitary generator, there is a unique C\*-algebra homomorphism $ $ h: C\left ( \mathbb{T}\right) \rightarrow C^{\ast}\left ( \left\ { \tilde { x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}\right\ } \right)$$ transport $ \mathrm{id}_{\mathbb{T}}$ to $ \tilde{x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}$ while fixing all scalars in $ \mathbb{C}\subset C\left ( \mathbb{T}\right) $. distinctly with $ x_{1}^{\ast}x_{1}|_{\mathcal{H}_{0}}=1 $ and $ x_{2}^{\ast}x_{2}|_{\mathcal{H}_{0}}=c$, $ $ h\circ\sigma: C^{\ast}\left ( \left\ { \tilde{x}_{1}^{\ast}\tilde{x}_{2},x_{1}^{\ast}x_{1}\right\ } \right) |_{\mathcal{H}_{1}\oplus\mathcal{H}_{2}}\rightarrow C^{\ast}\left ( \left\ { \tilde{x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}\right\ } \

{H}_{1}\oppus\mathcal{H}_{2}}$$ is a C\*-algebrx isomorphism, aue hencx $C^{\ast}\lsft( \lefg\{ \tilde{x}_{1}^{\ast}\tilde{x}_{2},x_{1}^{\ast}x_{1}\right\} \eight) $ is isomorphic to thd pullbacn $\mathcao{T}\opoys_{C\left( \mefhbb{T}\rinkt) }\mzbhcal{C}$ if two copies pf $\mathcal{D}\overset{\sigma}{\rhgftcrrow}C\left( \mathbb{T}\right) $ with $\tilde{x}_{1}^{\wst}\tildr{x}_{2}$ corresponding to $\kwthczl{S}\oplus\mathcal{S}$. Proof. Clearly we ohly neev to consider tne case with $\mathcal{H}_{0}\neq0$. Sijce $\hilde{x}_{1}^{\ast}\tilde{x}_{2}|_{\matjcal{H}_{0}}$ is unutarr and $C\left( \oathbb{T}\right) $ is the oniversal C\*-algebra generated by x sinyle unitary gwnegdtor, there ms a ugique C\*-algebvs homokorphisk $$h:C\left( \mathnb{T}\rijht) \rightarrow C^{\ast}\left( \left\{ \tilde {x}_{1}^{\ast}\tijde{x}_{2}|_{\mathcdl{K}_{0}}\right\} \right)$$ sending $\mqthrm{hd}_{\madhbb{G}}$ to $\timdx{x}_{1}^{\aat}\tildf{x}_{2}|_{\methcal{H}_{0}}$ whime fixing aol scalars in $\mathbn{C}\flnset C\left( \mzthbb{T}\widht) $. Clearly with $x_{1}^{\ast}x_{1}|_{\mathcal{H}_{0}}=1$ and $x_{2}^{\asu}x_{2}|_{\matgcal{H}_{0}}=c$, $$h\circ\sigma:C^{\ast}\ledt( \left\{ \tilde{x}_{1}^{\ast}\tipde{x}_{2},x_{1}^{\ast}x_{1}\wight\} \right) |_{\mathcal{H}_{1}\oplus\mathcal{H}_{2}}\rightarrow C^{\asd}\left( \leyb\{ \ripde{x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}\right\} \

{H}_{1}\oplus\mathcal{H}_{2}}$$ is a C\*-algebra isomorphism, and hence \tilde{x}_{1}^{\ast}\tilde{x}_{2},x_{1}^{\ast}x_{1}\right\} $ is to the pullback copies $\mathcal{T}\overset{\sigma}{\rightarrow}C\left( \mathbb{T}\right) $ $\tilde{x}_{1}^{\ast}\tilde{x}_{2}$ corresponding to Proof. Clearly we only need to the case with $\mathcal{H}_{0}\neq0$. Since $\tilde{x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}$ is unitary and $C\left( \mathbb{T}\right) $ is universal C\*-algebra generated by a single unitary generator, there is a unique C\*-algebra $$h:C\left( \rightarrow \left\{ {x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}\right\} \right)$$ sending $\mathrm{id}_{\mathbb{T}}$ to $\tilde{x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}$ while fixing all scalars in $\mathbb{C}\subset C\left( \mathbb{T}\right) $. Clearly $x_{1}^{\ast}x_{1}|_{\mathcal{H}_{0}}=1$ and $x_{2}^{\ast}x_{2}|_{\mathcal{H}_{0}}=c$, $$h\circ\sigma:C^{\ast}\left( \left\{ \tilde{x}_{1}^{\ast}\tilde{x}_{2},x_{1}^{\ast}x_{1}\right\} \right) |_{\mathcal{H}_{1}\oplus\mathcal{H}_{2}}\rightarrow \left\{ \tilde{x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}\right\} \

{H}_{1}\oplus\mathcal{H}_{2}}$$ is a C\*-algebra IsomorphisM, and hEncE $C^{\aSt}\Left( \Left\{ \Tilde{x}_{1}^{\ast}\tilde{X}_{2},X_{1}^{\ast}X_{1}\right\} \right) $ is isomorphiC to thE pULlbaCK $\mAthcaL{T}\oplus_{c\LeFT( \MatHbB{T}\RigHt) }\MAtHcal{T}$ Of tWo copieS of $\mathcal{t}\ovErSet{\sigma}{\righTArRow}C\left( \maThbB{T}\right) $ with $\tIldE{x}_{1}^{\ast}\tIlDe{x}_{2}$ COrresPonDing tO $\mathcAL{S}\opluS\mathcal{S}$. prOOf. CleaRLy we onlY NEeD to cOnsider the case witH $\MaTHcal{H}_{0}\neq0$. Since $\tIlde{x}_{1}^{\aSt}\TIlDE{X}_{2}|_{\maThcAl{H}_{0}}$ is unitaRy And $C\lEFt( \mathbB{t}\rIGHT) $ is THe universal C\*-aLgebra generATed By a sinGlE unITary geNeratOr, THerE is a unique C\*-AlgeBra homomoRphism $$H:c\left( \maTHbb{T}\rigHt) \righTarRow c^{\ast}\LEfT( \lEft\{ \TiLDe {x}_{1}^{\ASt}\TilDE{x}_{2}|_{\mAthcal{H}_{0}}\rIgHt\} \Right)$$ SendING $\MAthrM{id}_{\MathBb{T}}$ to $\Tilde{x}_{1}^{\ast}\tildE{x}_{2}|_{\mAthcAL{H}_{0}}$ wHile fIxing All sCaLars iN $\mathbB{C}\subSeT C\left( \mathbb{T}\riGht) $. CLearly witH $x_{1}^{\aSt}X_{1}|_{\maThCal{H}_{0}}=1$ aND $x_{2}^{\ast}x_{2}|_{\MatHcaL{H}_{0}}=c$, $$h\cirC\sigma:C^{\ASt}\lEfT( \LEFt\{ \Tilde{x}_{1}^{\ast}\tilde{x}_{2},x_{1}^{\asT}x_{1}\RIGhT\} \right) |_{\maThcal{H}_{1}\OPlUs\MAthcal{H}_{2}}\rIgHtaRrow c^{\ASt}\lefT( \lefT\{ \TiLde{x}_{1}^{\ast}\tIlde{x}_{2}|_{\mAThCaL{H}_{0}}\right\} \

{H}_{1}\oplus\mathcal{H}_{ 2}}$$ is a C\*- alg ebr aisom orph ism, and hence $C^{ \ast}\left( \left\{ \tild e{ x }_{1 } ^{ \ast} \tilde{ x }_ { 2 },x _{ 1} ^{\ as t }x _{1}\ rig ht\} \ right) $isis omorphic tot he pullback$\m athcal{T}\op lus _{C\le ft ( \ mathb b{T }\rig ht) } \ mathca l{T}$ oftw o copie s of $\m a t hc al{T }\overset{\sigma} { \r i ghtarrow}C\lef t( \ma th b b{ T } \ri ght ) $ with$\ tilde { x}_{1}^ { \a s t } \ti l de{x}_{2}$ co rresponding to$\math ca l{S } \oplus \math ca l {S} $. Proof.Clea rly we on ly nee d to con s ider th e case wi th$\ma t hc al {H} _{ 0 }\n e q0 $.Sin ce $\til de {x }_{1} ^{\a s t } \ tild e{x }_{2 }|_{\ mathcal{H}_{0 }}$ isu nit ary a nd $C \lef t( \ma thbb{T }\rig ht ) $ is the uni vers al C\*-al geb ra ge ne rated by a s ing leunitary genera t or, t h e r eis a unique C\*-al ge b r ahomomorp hism $ $ h: C\ l eft( \m at hbb {T}\ r i ght) \ri g ht arrow C^ {\ast} \ le ft ( \lef t\ { \ti ld e { x}_ {1}^{ \ ast} \tilde {x}_{2}| _{\ma t hcal{H}_{0}}\r i ght\} \right ) $$ s en d ing$\m athrm{id}_{ \mat h bb{T }}$t o$\t i lde{x }_{1} ^{ \ as t }\tilde{x}_{2}|_{\m at hcal{H }_{0} }$ while fixi ng all sca l a r s in $\m athb b {C } \subset C\left ( \m athbb{T}\r i ght) $. Cle arly wit h $x_{1}^ { \ ast}x_{1 }|_ {\m ath cal { H }_ {0}}=1$ and $ x _ {2}^ {\ ast}x_{ 2}| _{\math cal {H} _{0 }}= c$ , $$h\cir c\sigma: C^ {\ as t} \le ft( \ left\{ \t ild e{ x}_ {1}^{ \ ast}\t ilde{ x}_{ 2} ,x _ {1} ^{\ast} x _{ 1 } \rig ht \} \r igh t) |_{ \mat h cal {H}_{1} \oplus\ma thc a l{H} _{ 2} }\right arrow C^{\ast }\ left( \le ft \{ \tild e { x}_{1}^{ \ast}\tilde{x}_{2}|_{\m a thcal{H }_{ 0}}\r ight \} \

{H}_{1}\oplus\mathcal{H}_{2}}$$ is_a C\*-algebra_isomorphism, and hence $C^{\ast}\left(_ \left\{__\tilde{x}_{1}^{\ast}\tilde{x}_{2},x_{1}^{\ast}x_{1}\right\} _\right)_ $ is_isomorphic to the_pullback $\mathcal{T}\oplus_{C\left( \mathbb{T}\right)_ }\mathcal{T}$ of_two_copies of $\mathcal{T}\overset{\sigma}{\rightarrow}C\left( \mathbb{T}\right) $ with $\tilde{x}_{1}^{\ast}\tilde{x}_{2}$ corresponding to $\mathcal{S}\oplus\mathcal{S}$. Proof. Clearly we only need_to_consider the_case_with_$\mathcal{H}_{0}\neq0$. Since $\tilde{x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}$ is unitary and_$C\left( \mathbb{T}\right) $_is the_universal C\*-algebra generated by a single unitary generator,_there_is a unique_C\*-algebra homomorphism $$h:C\left( \mathbb{T}\right) \rightarrow C^{\ast}\left( _\left\{ \tilde {x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}\right\} \right)$$ sending_$\mathrm{id}_{\mathbb{T}}$ to $\tilde{x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}$_while_fixing_all scalars in $\mathbb{C}\subset C\left(_ \mathbb{T}\right) $. Clearly with $x_{1}^{\ast}x_{1}|_{\mathcal{H}_{0}}=1$_and $x_{2}^{\ast}x_{2}|_{\mathcal{H}_{0}}=c$, $$h\circ\sigma:C^{\ast}\left( \left\{ _\tilde{x}_{1}^{\ast}\tilde{x}_{2},x_{1}^{\ast}x_{1}\right\} \right) |_{\mathcal{H}_{1}\oplus\mathcal{H}_{2}}\rightarrow C^{\ast}\left( _\left\{ \tilde{x}_{1}^{\ast}\tilde{x}_{2}|_{\mathcal{H}_{0}}\right\} \

this paper, we present a mesh neural network, named MeshNet, that learns on mesh data directly for 3D shape representation. To deal with the challenges in mesh data processing, the faces are regarded as the unit and connections between faces sharing common edges are defined, which enables us to solve the complexity and irregularity problem with per-face processes and a symmetry function. Moreover, the feature of faces is split into spatial and structural features. Based on these ideas, we design the network architecture, with two blocks named spatial and structural descriptors for learning the initial features, and a mesh convolution block for aggregating neighboring features. In this way, the proposed method is able to solve the complexity and irregularity problem of mesh and represent 3D shapes well. We apply our MeshNet method in the tasks of 3D shape classification and retrieval on the ModelNet40 [@wu20153d] dataset. And the experimental results show that MeshNet achieve significant improvement on 3D shape classification and retrieval using mesh data and comparable performance with recent methods using other types of 3D data. The key contributions of our work are as follows: - We propose a neural network using mesh for 3D shape representation and design blocks for capturing and aggregating features of polygon faces in 3D shapes. - We conduct extensive experiments to evaluate the performance of the proposed method, and the experimental results show that the proposed method performs well on the 3D shape classification and retrieval task. Related Work ============ Mesh Feature Extraction ----------------------- There are plenty of handcraft descriptors that extract features from mesh. @lien1984symbolic calculate moments of each tetrahedron in mesh [@lien1984symbolic], and @zhang2001efficient develop more functions applied to each triangle and add all the resulting values as features [@zhang2001efficient]. @hubeli2001multiresolution extend the features of surfaces to a multiresolution setting to solve the unstructured problem of mesh data [@hubeli2001multiresolution]. In SPH [@kazhdan2003rotation], a rotation invariant representation is presented with existing orientation dependent descriptors. Mesh difference of Gaussians (DOG) introduces the Gaussian filtering to shape functions.[@zaharescu2009surface] Intrinsic shape context (ISC) descriptor [@kokkinos2012intrinsic] develops a generalization to surfaces and solves the problem of orientational ambiguity. Deep Learning Methods for 3D Shape Representation ------------------------------------------------- With the construction

this paper, we present a mesh neural network, name MeshNet, that memorize on mesh data directly for 3D shape theatrical performance. To deal with the challenges in mesh data processing, the face are regarded as the unit and joining between faces sharing coarse edges are defined, which enable us to solve the complexity and irregularity problem with per - face processes and a symmetry function. furthermore, the feature of faces is split into spatial and morphologic features. Based on these theme, we design the net architecture, with two block named spatial and structural descriptors for learn the initial features, and a mesh convolution block for aggregating neighboring feature. In this room, the proposed method is able to solve the complexity and irregularity problem of mesh and represent 3D shapes well. We use our MeshNet method in the tasks of 3D condition categorization and retrieval on the ModelNet40 [ @wu20153d ] dataset. And the experimental results show that MeshNet achieve significant improvement on 3D shape classification and retrieval using net data and comparable performance with recent methods using other type of 3D data. The key contributions of our work are as play along: - We propose a nervous net use mesh for 3D shape representation and design blocks for capturing and aggregating feature of polygon faces in 3D shapes. - We conduct extensive experiment to evaluate the performance of the propose method, and the experimental consequence show that the propose method do well on the 3D shape classification and retrieval task. Related Work = = = = = = = = = = = = Mesh Feature Extraction ----------------------- There are batch of handcraft descriptors that extract feature from mesh. @lien1984symbolic calculate moments of each tetrahedron in mesh [ @lien1984symbolic ], and @zhang2001efficient develop more functions applied to each triangulum and add all the resulting value as features [ @zhang2001efficient ]. @hubeli2001multiresolution extend the features of surfaces to a multiresolution setting to solve the unstructured trouble of mesh data [ @hubeli2001multiresolution ]. In SPH [ @kazhdan2003rotation ], a rotation invariant representation is presented with existing orientation dependent form. engage difference of Gaussians (DOG) introduces the Gaussian filtering to shape functions.[@zaharescu2009surface ] Intrinsic shape context (ISC) descriptor [ @kokkinos2012intrinsic ] develops a abstraction to surfaces and solves the problem of orientational ambiguity. Deep Learning Methods for 3D Shape Representation ------------------------------------------------- With the construction

thls paper, we present a merh neural netwotk, named MeshNst, that uearns on mesh data directly fir 3D whape representation. Tu deal winh the chqlleiges in mesh date processing, tgc faczs are regarded ss the unid and connectimnr yetween faces sharing common edges awe defimef, which enablef us eo sklve the complexity and irregularify probnem with per-fsce processes and a symmetgy fknction. Moreover, tje feature if fwxes is split into spatpcl and strudtural features. Based on these kdeas, we design tye jgtwork archivecturv, with two blocks namad spatoal and structmral vesceiptors for learning vhe initial features, and a mevh convolution blocj dor acgrecativt ndiggbprjng fewtuces. In this way, the priposed method is abke np solve the dompleviey and irregularity problem of mesh and relresent 3D shapes well. We apply our MeshNet mejhod in thq tasks of 3D shape classification and retrieval ot the OodtlKct40 [@wj20153e] fataset. And the experimental results show thae MtshKet achieve signlficant improvemeny ln 3Q shape classkficatnkn and retrieval usijg mesh data and compwrabke performance with recent nethods usiny orher types of 3D daca. The key coutribujions pf our work are as folluws: - We proposf a neuram network using mdsh fmr 3D shakd representation wnd desigi bloeks for zaptoring agd aggregahing naatures of polygon faced hn 3D shaped. - We conduct extensive experiments to evalusta tve perfoxmance of the propofed method, and the ex'erimevtal resulns show tiat the propjsed method patforms well oi the 3D srape clawsificagkon and retriefal task. Rvlcted Work ============ Nesh Feature Extragtion ----------------------- Jhsre are plenty if yandcraft descropturs tjav extwdct features froo mdxh. @lidn1984symbolic galzulaye moments of each tatragedron in mesh [@liem1984snmbolic], abd @zhang2001qfficient devrlop more functiond appniev to esch triangle and add all the resumting valkes as features [@zhakg2001efsicient]. @huyeli2001multiresolution extend the features if surfaces to a myltiresolution settnnn to solve tie unseructured problem of mesh datq [@hubeli2001multiresokution]. In SPH [@kazhdan2003dotatimn], a gotation invariant representation is presented with existing orientation eependxne descriptods. Mrsh dpffzreuce of Dausdmans (DOG) introducvs the Gaussian filtering to sha'e functiots.[@vaharescu2009surface] Intrinsic shspd context (ISC) descriptor [@kokkinos2012intrjnsic] defelops a generalization to surfsces and solves the probkem of oeientztional amuiguity. Deep Learning Methobs for 3D Shapw Reiresentation ------------------------------------------------- Wigh yhc construchion

this paper, we present a mesh neural MeshNet, learns on data directly for with challenges in mesh processing, the faces regarded as the unit and connections faces sharing common edges are defined, which enables us to solve the complexity irregularity problem with per-face processes and a symmetry function. Moreover, the feature of is into and features. Based on these ideas, we design the network architecture, with two blocks named spatial and descriptors for learning the initial features, and a convolution block for aggregating features. In this way, the method able to the and problem of mesh represent 3D shapes well. We apply our MeshNet method in the tasks of 3D shape classification and on the dataset. And experimental show MeshNet achieve significant 3D shape classification and retrieval using comparable performance with recent methods using other types 3D data. key contributions of our work are follows: - We propose a neural network using for 3D shape representation and design blocks for capturing and aggregating features of polygon faces shapes. - We conduct experiments to evaluate performance the method, the experimental show that the proposed method performs well on the 3D shape and retrieval task. Related Work ============ Mesh Feature Extraction ----------------------- plenty handcraft descriptors that features from mesh. @lien1984symbolic moments each tetrahedron in mesh @zhang2001efficient more each and all the resulting values features [@zhang2001efficient]. @hubeli2001multiresolution extend the of surfaces to a unstructured problem of mesh data [@hubeli2001multiresolution]. In SPH a rotation invariant representation is presented with orientation dependent descriptors. Mesh difference of Gaussians (DOG) introduces the Gaussian filtering shape functions.[@zaharescu2009surface] context (ISC) descriptor [@kokkinos2012intrinsic] develops a generalization to and solves the problem orientational ambiguity. Deep Learning Methods for 3D Shape Representation With construction

this paper, we present a mesh neUral networK, nameD MeShNEt, That LearNs on mesh data diREctlY for 3D shape representatiOn. To dEaL With THe ChallEnges in MEsH DAta PrOcEssInG, ThE faceS arE regardEd as the uniT anD cOnnections beTWeEn faces shaRinG common edges Are DefineD, wHicH EnablEs uS to soLve the COmplexIty and irrEgULarity PRoblem wITH pEr-faCe processes and a syMMeTRy function. MoreOver, thE fEAtURE of FacEs is split iNtO spatIAl and stRUcTURAl fEAtures. Based on These ideas, wE DesIgn the NeTwoRK archiTectuRe, WIth Two blocks naMed sPatial and StructURal descRIptors fOr learNinG thE iniTIaL fEatUrES, anD A mEsh COnvOlution bLoCk For agGregATING neiGhbOrinG featUres. In this way, The PropOSed MethoD is abLe to SoLve thE complExity AnD irregularity prObleM of mesh anD rePrEseNt 3d shapES well. WE apPly Our Meshnet methOD in ThE TASkS of 3D shape classificAtION aNd retrieVal on tHE MOdELNet40 [@wu20153d] dAtAseT. And THE expeRimeNTaL results Show thAT MEsHNet achIeVe signIfIcaNt iMprovEMent On 3D shaPe classiFicatIOn and retrieval USing mesh data aND cOMPaRAble PerFormance witH recENt meThodS UsIng OTher tYpes oF 3D DAtA. the key contributions Of Our worK are aS follows: - We proPose a neuraL NETwork usiNg meSH fOR 3D shape represeNtatiOn and desigN Blocks foR captUring and AggregatiNG Features Of pOlyGon FacES In 3d shapes. - We condUCT extEnSive expEriMents to EvaLuaTe tHe pErFormance oF the propOsEd MeThOd, aNd the EXperimenTaL reSuLts Show tHAt the pRoposEd meThOd PErfOrms welL On THE 3D shApE cLassIfiCaTion aNd reTRieVal task. related WoRk ============ MESh FeAtUrE ExtracTion ----------------------- There are pLeNty of handcRaFt dEscripTORs that exTract features from mesh. @liEN1984symbolIc cAlculAte mOments of eAch TetrahEdrON in mesH [@lien1984sYmbolIc], And @ZHAng2001efFICiEnt DeVelop more fUNCtiOns apPlIed tO each trIangle and add all the REsuLting values as FeaTureS [@ZHaNg2001eFFiCIenT]. @hUBelI2001MUltiresolution eXtend the feAtUReS of surfaceS To a MuLtiresoLution sEttinG To solve The unstruCtured proBlEm of MESh dAta [@hubeli2001mUltiresoLution]. In Sph [@kazhDAn2003RotatIon], A rotatIoN inVariaNt reprESenTatioN is preSeNted wiTh exiStIng orienTation dependent descriptOrs. MesH diffEreNce of GausSiaNS (DOg) introducEs thE Gaussian fIltEriNg to sHapE FunctIons.[@ZAhAreSCu2009surFace] iNtrinsic sHApE coNTExT (ISC) descripTOR [@KokKinos2012IntRInsic] dEvelOps a generalizatioN To surfaces and sOlveS THe pRobLEm of OrIentational ambIguItY. dEep LearnInG Methods for 3d Shape RePrESentaTion ------------------------------------------------- WiTh the cOnstrucTIOn

this paper, we present amesh neura l net wor k,na medMesh Net, that lear n s on mesh data directly fo r 3Dsh a pe r e pr esent ation.T od e alwi th th ec ha lleng esin mesh data proc ess in g, the faces ar e regarded as the unit an d c onnect io nsb etwee n f acessharin g commo n edges a re define d , which e na bles us to solve thec om p lexity and irr egular it y p r o ble m w ith per-fa ce proc e sses an d a s y mme t ry function.Moreover, t h e f eature o f f a ces is spli ti nto spatial an d st ructuralfeatur e s. Base d on the se ide as, we des i gn t hene t wor k a rch i tec ture, wi th t wo bl ocks n a m ed s pat ialand s tructural des cri ptor s fo r lea rning the i nitia l feat ures, a nd a mesh convo luti on blockfor a ggr eg ating neighb ori ngfeature s. In t h iswa y , th e proposed methodis a bl e to sol ve the co mp l exity an dirr egul a r ity p robl e mof meshand re p re se nt 3D s ha pes we ll . Weapply ourMeshNe t method in t h e tasks of 3Ds hape classifi c at i o na nd r etr ieval on th e Mo d elNe t40[ @w u20 1 53d]datas et . A n d the experimentalre sultsshowthat MeshNetachieve si g n i ficant i mpro v em e nt on 3D shape clas sification and retr ieval using m esh dataa n d compar abl e p erf orm a n ce with recentm e thod susing o the r types of 3D da ta. The key c ontribut io ns o four work are as f ol low s: - We propos e a n eura lne t wor k using me s h for 3 Dshap e r ep resen tati o n a nd desi gn blocks fo r cap tu ri ng andaggregating f ea tures of p ol ygo n face s in 3D sh apes. - We conduct e x tensive ex perim ents to evalu ate the p erf o rmance of th e pro po sed m ethod , an d t he experimen t a l r esult sshow that t he proposed method per forms well on th e 3D s ha pec la s sif ic a tio n and retrieval t ask. Rela te d W ork ====== = === == MeshFeature Extr a ction - --------- --------- -- -- T h ere are plent y of han dcraft de s cript o rs that ex tractfe atu res f rom me s h.@lien 1984sy mb olic c alcul at e moment s of each tetrahedron i n mesh [@li en1 984symbol ic] , an d @zhang2 001e fficient d eve lop more fu n ction s ap p li edt o eac h tr i angle and ad d a l l t he resultin g v alu es as fe a tures[@zh ang2001efficient] . @hubeli2001mu ltir e s olu tio n ext en d the features of s u r faces to a multiresol ution se tt i ng to solve the u nstruct u r ed proble m of me sh data [ @hu be l i2001mu lt ir e soluti on]. I n SPH[@kazh d an20 0 3 rotation], a rot ation i nvari a ntrepre se ntation is p resented w ith existin g orie ntat ion d ependen tdescri pto rs . Mesh dif f erence of Gaus sians ( DO G) i ntr oduces the G aussi an f il ter ing to sh a p ef un ct i ons .[@z ahare sc u200 9surface] Intrinsi c s h ape con te xt( I SC) de s cr i p tor [@kokk ino s2012 i n trinsic] d e velo p sa gene raliza tion t o surfa c esan d solve s t h e problemof orient a tio na l am biguity. Deep L ear ni ng Methods for3D Shape Represe n t ation--- - ---- - ------ -- ------------ ----- -- - ------- - - --- Witht hecon str ucti on

this_paper, we_present a mesh neural_network, named_MeshNet,_that learns_on_mesh data directly_for 3D shape_representation. To deal with_the challenges in_mesh_data processing, the faces are regarded as the unit and connections between faces sharing_common_edges are_defined,_which_enables us to solve the_complexity and irregularity problem with_per-face processes_and a symmetry function. Moreover, the feature of_faces_is split into_spatial and structural features. Based on these ideas, we_design the network architecture, with two_blocks named spatial_and_structural_descriptors for learning the_initial features, and a mesh convolution_block for aggregating neighboring features. In_this way, the proposed method is able_to solve the complexity and irregularity_problem of mesh and represent_3D shapes_well. We apply our MeshNet method_in the tasks_of 3D_shape classification and_retrieval on the ModelNet40 [@wu20153d] dataset._And the experimental_results show that MeshNet achieve significant_improvement_on 3D shape_classification_and_retrieval using_mesh data and_comparable_performance with_recent_methods using other types of 3D_data. The_key contributions of our work are as_follows: - We_propose_a neural network using_mesh for 3D shape representation_and design blocks for capturing and_aggregating features_of polygon_faces in 3D shapes. - We conduct extensive experiments to_evaluate the performance of the proposed_method, and the experimental_results show_that_the proposed method_performs_well on_the 3D shape classification and retrieval task. Related_Work ============ Mesh Feature_Extraction ----------------------- There are plenty of handcraft descriptors_that extract features from_mesh._@lien1984symbolic calculate moments of each tetrahedron_in mesh [@lien1984symbolic], and @zhang2001efficient develop_more functions applied to each_triangle_and_add all the resulting values_as features [@zhang2001efficient]. @hubeli2001multiresolution extend the_features of surfaces_to a multiresolution setting to solve the_unstructured_problem of mesh data [@hubeli2001multiresolution]. In_SPH_[@kazhdan2003rotation], a rotation invariant representation is_presented_with_existing orientation dependent descriptors. Mesh_difference of Gaussians (DOG) introduces the_Gaussian filtering to shape functions.[@zaharescu2009surface] Intrinsic shape context (ISC)_descriptor [@kokkinos2012intrinsic] develops_a generalization to surfaces and_solves_the_problem of orientational ambiguity. Deep Learning Methods for 3D Shape Representation ------------------------------------------------- With_the construction

in §\[sect:discussion:exceed\] that this likely does not affect the final orbital period distributions in the simulations. - In all other cases, we mark the outcome as ‘no migration’. For each outcome, we show the corresponding fraction, based on the $N_\mathrm{MC}=10^3$ simulations, for the 2+2 and 3+1 configurations, and various sets of simulations: with the first (‘Orb. Distr. (1)’) and second (‘Orb. Distr. (2)’) assumptions about the orbital distributions (see §\[sect:IC:orbits\]), and with (‘F’) and without (‘NF’) the inclusion of flybys. Also, we show fractions after 10 Gyr of evolution (if applicable; otherwise, the end time is the stopping condition time), or after a random time between 0 and 10 Gyr ($t_x$), appropriate for continuous star formation. The most likely outcome in all cases is ‘no migration’. Note that the eccentricities may still have been excited (see §\[sect:pop\_syn:orb:e\] below). In a few per cent for the 2+2 systems, and in up to $\sim 14\%$ for the 3+1 systems, tidal migration occurs in orbit 1 or 2. In the 2+2 configuration, double migration, of which we showed an example in §\[sect:examples:1\], occurs in a few tenths of per cents, i.e., is relatively rare. RLOF is triggered in up to $\sim 11\%$ for the 2+2 systems, and $\sim 15\%$ for the 3+1 systems. In the 2+2 systems, dynamical instability rarely occurs, which can be understood by noting that dynamical instability can only be triggered in our simulations by flybys, or, more rarely, if the outer orbit eccentricity, $e_3$, is enhanced, which can only occur due to high-order terms (higher than quadrupole-order), and therefore only in compact systems. For the 3+1 systems, dynamical instability is relatively common for the orbit 1-2 pair (up to $\sim 14\%$). This can be attributed to enhanced eccentricity of orbit 2 if it is inclined relative to orbit 3. As noted above, the unbinding of the system due to flybys rarely occurs. The run time is exceeded mostly for 3+1 systems, and this is further addressed in §\[sect:discussion

in § \[sect: discussion: exceed\ ] that this likely does not affect the final orbital period distribution in the model. - In all other cases, we mark the consequence as ‘ no migration ’. For each outcome, we show the represent fraction, base on the $ N_\mathrm{MC}=10 ^ 3 $ simulations, for the 2 + 2 and 3 + 1 configurations, and various sets of pretense: with the first (‘ Orb. Distr. (1) ’) and second (‘ Orb. Distr. (2) ’) assumptions about the orbital distribution (determine § \[sect: IC: orbits\ ]), and with (‘ F ’) and without (‘ NF ’) the inclusion of flybys. Also, we show fraction after 10 Gyr of evolution (if applicable; otherwise, the end prison term is the stopping condition time), or after a random time between 0 and 10 Gyr ($ t_x$), appropriate for continuous ace formation. The most likely outcome in all cases is ‘ no migration ’. notice that the eccentricities may still have been excited (see § \[sect: pop\_syn: orb: e\ ] below). In a few per penny for the 2 + 2 organization, and in up to $ \sim 14\%$ for the 3 + 1 systems, tidal migration occurs in orbit 1 or 2. In the 2 + 2 configuration, double migration, of which we showed an example in § \[sect: examples:1\ ], occurs in a few tenths of per cent, i.e., is relatively rare. RLOF is triggered in up to $ \sim 11\%$ for the 2 + 2 systems, and $ \sim 15\%$ for the 3 + 1 system. In the 2 + 2 system, dynamical instability rarely occurs, which can be understand by note that dynamical instability can merely be triggered in our simulations by flybys, or, more rarely, if the outer orbit eccentricity, $ e_3 $, is enhanced, which can only happen due to high - order terms (higher than quadrupole - order), and therefore only in compendious system. For the 3 + 1 systems, dynamical instability is relatively coarse for the orbit 1 - 2 pair (up to $ \sim 14\%$). This can be attributed to enhance eccentricity of orbit 2 if it is inclined relative to orbit 3. As note above, the unbinding of the system due to flybys rarely occur. The run time is exceeded largely for 3 + 1 systems, and this is farther addressed in § \[sect: discussion

in §\[sect:discussion:exceed\] thxt this likely bies nov affecf the fival orbital period distributmons in tye simulations. - In alu other cwses, we nark rhe outcomx as ‘no migratjln’. Fox xach outcome, we show the worresponding xrxccion, based on the $N_\mathrm{MC}=10^3$ simulatijns, for tje 2+2 and 3+1 confidurauiogs, ahd various sets of simulations: witg the fprst (‘Orb. Distr. (1)’) amd second (‘Orb. Distr. (2)’) assumotiojs about the orbitwl distribujjonf (see §\[sect:IC:ofbits\]), and with (‘F’) and wjthout (‘NF’) the inclusion of flybhs. Alxo, we show feachhons after 10 Gyr jf evolution (if applhcable; ptherwise, the cnd tmme us the stopping condivion time), or after a random thmz between 0 and 10 Gyr ($t_z$), qpprokriata fof covtihupua star focmation. The jost likely outcome in all casts yw ‘no migratioh’. Note trat the eccentricities may still have bten escited (see §\[sect:pop\_syn:orv:e\] below). In a few per cent for the 2+2 systems, and in up to $\sim 14\%$ for the 3+1 systems, didal oigxqtion icfurs in orbit 1 or 2. In the 2+2 configuration, doubjs kinration, of which we showed sn ecwmple in §\[sect:gxamples:1\], odcurs in a few tenhhs of ker cebts, i.e., is relstively rare. RLOF is triggeeed in up to $\wim 11\%$ for the 2+2 systzms, and $\sim 15\%$ for the 3+1 systems. In the 2+2 systemr, dyhamical inshability dxrely occurs, whizh ban te understood by noting thwt dynamiral iustabiligy csn onlr be triggfred lt our simulations hy flibys, os, more rarfly, if the outer orbit eccentrirmty, $e_3$, is enhamcad, fhich cau only occur due to high-order terks (highzr thav quadrupome-order), and therefjre only in cmlpact systemv. For thq 3+1 ststens, dynaokcal instabiliyy is relcuively comnon for the orbit 1-2 pakd (up to $\sim 14\%$). Thnr can be attribuyed to ejhenced accentricity of urbky 2 if it is incllnea rekative to orbit 3. As totes above, the unbindonn of the wystem dte to flybys tarely occurs. The gun tmme is excerdeq mostly for 3+1 systems, and this is furthfr wddressed in §\[secb:dissussion

in §\[sect:discussion:exceed\] that this likely does not final period distributions the simulations. - mark outcome as ‘no For each outcome, show the corresponding fraction, based on $N_\mathrm{MC}=10^3$ simulations, for the 2+2 and 3+1 configurations, and various sets of simulations: the first (‘Orb. Distr. (1)’) and second (‘Orb. Distr. (2)’) assumptions about the distributions §\[sect:IC:orbits\]), with and without (‘NF’) the inclusion of flybys. Also, we show fractions after 10 Gyr of evolution applicable; otherwise, the end time is the stopping time), or after a time between 0 and 10 ($t_x$), for continuous formation. most outcome in all is ‘no migration’. Note that the eccentricities may still have been excited (see §\[sect:pop\_syn:orb:e\] below). In a per cent 2+2 systems, in to 14\%$ for the tidal migration occurs in orbit 1 the 2+2 configuration, double migration, of which we an example §\[sect:examples:1\], occurs in a few tenths per cents, i.e., is relatively rare. RLOF is in up to $\sim 11\%$ for the 2+2 systems, and $\sim 15\%$ for the 3+1 the 2+2 systems, dynamical rarely occurs, which be by that instability can be triggered in our simulations by flybys, or, more rarely, if outer orbit eccentricity, $e_3$, is enhanced, which can only occur high-order (higher than quadrupole-order), therefore only in compact For 3+1 systems, dynamical instability common the (up $\sim This can be attributed enhanced eccentricity of orbit 2 it is inclined relative above, the unbinding of the system due to rarely occurs. The run time is exceeded for 3+1 systems, and this is further addressed in §\[sect:discussion

in §\[sect:discussion:exceed\] thaT this likelY does Not AffEcT the FinaL orbital period DIstrIbutions in the simulatioNs. - In aLl OTher CAsEs, we mArk the oUTcOME as ‘No MiGraTiON’. FOr eacH ouTcome, we Show the corResPoNding fractioN, BaSed on the $N_\mAthRm{MC}=10^3$ simulatiOns, For the 2+2 AnD 3+1 coNFigurAtiOns, anD varioUS sets oF simulatiOnS: With thE First (‘OrB. dIsTr. (1)’) anD second (‘Orb. Distr. (2)’) asSUmPTions about the oRbital DiSTrIBUtiOns (See §\[sect:IC:oRbIts\]), anD With (‘F’) anD WiTHOUt (‘Nf’) The inclusion oF flybys. Also, WE shOw fracTiOns AFter 10 GyR of evOlUTioN (if applicabLe; otHerwise, thE end tiME is the sTOpping cOnditiOn tIme), Or afTEr A rAndOm TIme BEtWeeN 0 And 10 gyr ($t_x$), appRoPrIate fOr coNTINUous StaR forMatioN. The most likelY ouTcomE In aLl casEs is ‘nO migRaTion’. NOte thaT the eCcEntricities may sTill Have been eXciTeD (seE §\[sEct:poP\_Syn:orb:E\] beLow). in a few pEr cent fOR thE 2+2 sYSTEmS, and in up to $\sim 14\%$ for thE 3+1 sYSTeMs, tidal mIgratiON oCcURs in orbiT 1 oR 2. In The 2+2 cONFigurAtioN, DoUble migrAtion, oF WhIcH we showEd An examPlE in §\[SecT:examPLes:1\], oCcurs iN a few tenThs of PEr cents, i.e., is relATively rare. RLOf Is TRIgGEred In uP to $\sim 11\%$ for thE 2+2 sysTEms, aNd $\siM 15\%$ FoR thE 3+1 SysteMs. In tHe 2+2 SYsTEms, dynamical instabiLiTy rareLy occUrs, which can be Understood BY NOting thaT dynAMiCAl instability cAn onlY be triggerED in our siMulatIons by flYbys, or, morE RArely, if tHe oUteR orBit ECCeNtricity, $e_3$, is enHANced, WhIch can oNly Occur duE to HigH-orDer TeRms (higher Than quadRuPoLe-OrDer), And thERefore onLy In cOmPacT systEMs. For tHe 3+1 sysTems, DyNaMIcaL instabILiTY Is reLaTiVely ComMoN for tHe orBIt 1-2 pAir (up to $\Sim 14\%$). This caN be ATtriBuTeD to enhaNced eccentricItY of orbit 2 if It Is iNclineD RElative tO orbit 3. As noted above, the unBInding oF thE systEm duE to flybys RarEly occUrs. tHe run tIme is eXceedEd MosTLY for 3+1 sYSTeMs, aNd This is furtHER adDressEd In §\[seCt:discuSsion

in §\[sect:discussion:exc eed\] that this li kel ydoes not affect the fi n al o rbital period distribu tions i n the si mulat ions. - I n a ll o the rc as es, w e m ark the outcome a s ‘ no migration’. F or each ou tco me, we showthe corre sp ond i ng fr act ion,basedo n the$N_\mathr m{ M C}=10^ 3 $ simul a t io ns,for the 2+2 and 3 + 1c onfigurations, and v ar i ou s set s o f simulati on s: wi t h the f i rs t ( ‘Or b . Distr. (1)’ ) and secon d (‘ Orb. D is tr. (2)’)assum pt i ons about theorbi tal distr ibutio n s (see§ \[sect: IC:orb its \]) , an d w it h ( ‘F ’ ) a n dwit h out (‘NF’)th einclu sion o f flyb ys. Als o, we show fractio nsafte r 10 Gyrof ev olut io n (if appli cable ;otherwise, theendtime is t hest opp in g con d itiontim e), or aft er a ra n dom t i m e b etween 0 and 10 Gy r( $ t_ x$), app ropria t efo r continu ou s s tarf o rmati on.Th e most l ikelyo ut co me in a ll cases i s ‘ nomigra t ion’ . Note that th e ecc e ntricities may still have be e ne x ci t ed ( see §\[sect:po p\_s y n:or b:e\ ] b elo w ). In a fe wp er cent for the 2+2 sy st ems, a nd in up to $\sim14\%$ fort h e 3+1 sys tems , t i dal migrationoccur s in orbit 1 or 2.In th e 2+2 co nfigurati o n , double mi gra tio n,o f w hich we showe d an e xa mple in §\ [sect:e xam ple s:1 \], o ccurs ina few te nt hs o fper cent s , i.e.,is re la tiv ely r a re. RL OF is tri gg er e d i n up to $\ s i m 11 \% $forthe 2 +2 sy stem s , a nd $\si m 15\%$ f ort he 3 +1 s ystems. In the 2+2 s ys tems, dyna mi cal insta b i lity rar ely occurs, which can b e unders too d bynoti ng that d yna micalins t abilit y canonlybe tr i g gered i nour s imulations b y f lybys ,or,more ra rely, if the outer orb it eccentrici ty, $e_ 3 $ ,ise nh a nce d, whi c h can only occur due to hi gh - or der terms( hig he r thanquadrup ole-o r der), a nd theref ore onlyin com p a ctsystems. F or the 3 +1 system s , dyn a mi cal i nst abilit yisrelat ivelyc omm on fo r theor bit 1- 2 pai r(up to $ \sim 14\%$). This can b e attr ibute d t o enhance d e c cen tricity o f or bit 2 if i t i s i nclin edr elati ve t o o rbi t 3. A s no t ed above, th e u n b in ding of the s y ste m due to flybys rar ely occurs. The r u n time is exce eded m ost lyf or 3 +1 systems, andthi si s further a ddressed in §\[sect :d i scuss ion

in_§\[sect:discussion:exceed\] that_this likely does not_affect the_final_orbital period_distributions_in the simulations. -_ In_all other cases, we_mark the outcome_as_‘no migration’. For each outcome, we show the corresponding fraction, based on the $N_\mathrm{MC}=10^3$ simulations,_for_the 2+2_and_3+1_configurations, and various sets of_simulations: with the first (‘Orb._Distr. (1)’)_and second (‘Orb. Distr. (2)’) assumptions about the_orbital_distributions (see §\[sect:IC:orbits\]),_and with (‘F’) and without (‘NF’) the inclusion of_flybys. Also, we show fractions after_10 Gyr of_evolution_(if_applicable; otherwise, the end_time is the stopping condition time),_or after a random time between_0 and 10 Gyr ($t_x$), appropriate for_continuous star formation. The most likely outcome_in all cases is ‘no_migration’. Note_that the eccentricities may still_have been excited_(see §\[sect:pop\_syn:orb:e\]_below). In a_few per cent for the 2+2_systems, and in_up to $\sim 14\%$ for the_3+1_systems, tidal migration_occurs_in_orbit 1_or 2. In_the_2+2 configuration,_double_migration, of which we showed an_example_in §\[sect:examples:1\], occurs in a few tenths_of per cents, i.e.,_is_relatively rare. RLOF is_triggered in up to $\sim_11\%$ for the 2+2 systems, and_$\sim 15\%$_for the_3+1 systems. In the 2+2 systems, dynamical instability rarely occurs, which_can be understood by noting that_dynamical instability can only_be triggered_in_our simulations by_flybys,_or, more_rarely, if the outer orbit eccentricity, $e_3$,_is enhanced,_which can only occur due to_high-order terms (higher than_quadrupole-order),_and therefore only in compact systems._For the 3+1 systems, dynamical instability_is relatively common for the_orbit_1-2_pair (up to $\sim 14\%$)._This can be attributed to enhanced_eccentricity of orbit_2 if it is inclined relative to_orbit_3. As noted above, the unbinding_of_the system due to flybys rarely_occurs._The_run time is exceeded mostly_for 3+1 systems, and this is_further addressed in §\[sect:discussion

0(C^*(E \times_1 [m,n])) \cong \bigoplus_{v \in V} {\mathbb{Z}}[p_{(v,n)}]_0 \oplus \bigoplus_{k=0}^{n-m} \bigoplus_{v \in W} {\mathbb{Z}}[p_{(v,n-k)}]_0$$ and $K_0(C^*(E \times_1 [m,n]))^+$ is identified with $$\bigoplus_{v \in V} {\mathbb{N}}[p_{(v,n)}]_0 \oplus \bigoplus_{k=0}^{n-m} \bigoplus_{v \in W} {\mathbb{N}}[p_{(v,n-k)}]_0.$$ By the continuity of $K$-theory, one can let $m$ tend to $-\infty$ and deduce that $$\begin{aligned} K_0(C^*(E \times_1 [-\infty,n])) &\cong \bigoplus_{v \in V} {\mathbb{Z}}[p_{(v,n)}]_0 \oplus \bigoplus_{k=0}^{\infty} \bigoplus_{v \in W} {\mathbb{Z}}[p_{(v,n-k)}]_0 \\ &\cong {\mathbb{Z}}^V \oplus Z^W \oplus Z^W \oplus \ldots.\end{aligned}$$ Furthermore, it follows from [@RLL Theorem 6.3.2(ii)] that this isomorphism identifies $K_0(C^*(E \times_1 [-\infty,n]))^+$ with ${\mathbb{N}}^V \oplus {\mathbb{N}}^W \oplus {\mathbb{N}}^W \oplus \ldots$. This computation is used later in the proof of [@RS Theorem 3.1], where the $K_0$ functor is applied to a commutative diagram to obtain Figure (3.5) of [@RS], which we reproduce here: $$\xymatrix{ {\mathbb{Z}}^V \oplus {\mathbb{Z}}^W \oplus {\mathbb{Z}}^W \oplus \ldots \ar[d]_{1-D} \ar[r]^D & {\mathbb{Z}}^V \oplus {\mathbb{Z}}^W \oplus {\mathbb{Z}}^W \oplus \ldots \ar[d]_{1-D} \ar[r]^-{\iota_*^{n+1}} & K_0(C^*(

0(C^*(E \times_1 [ m, n ]) ) \cong \bigoplus_{v \in V } { \mathbb{Z}}[p_{(v, n)}]_0 \oplus \bigoplus_{k=0}^{n - m } \bigoplus_{v \in W } { \mathbb{Z}}[p_{(v, n - k)}]_0$$ and $ K_0(C^*(E \times_1 [ m, n]))^+$ is identified with $ $ \bigoplus_{v \in V } { \mathbb{N}}[p_{(v, n)}]_0 \oplus \bigoplus_{k=0}^{n - m } \bigoplus_{v \in W } { \mathbb{N}}[p_{(v, n - k)}]_0.$$ By the continuity of $ K$-theory, one can let $ m$ tend to $ -\infty$ and deduce that $ $ \begin{aligned } K_0(C^*(E \times_1 [ -\infty, n ]) ) & \cong \bigoplus_{v \in V } { \mathbb{Z}}[p_{(v, n)}]_0 \oplus \bigoplus_{k=0}^{\infty } \bigoplus_{v \in W } { \mathbb{Z}}[p_{(v, n - k)}]_0 \\ & \cong { \mathbb{Z}}^V \oplus Z^W \oplus Z^W \oplus \ldots.\end{aligned}$$ Furthermore, it follows from [ @RLL Theorem   6.3.2(ii) ] that this isomorphism identify $ K_0(C^*(E \times_1 [ -\infty, n]))^+$ with $ { \mathbb{N}}^V \oplus { \mathbb{N}}^W \oplus { \mathbb{N}}^W \oplus \ldots$. This calculation is used later in the proof of [ @RS Theorem   3.1 ], where the $ K_0 $ functor is apply to a commutative diagram to obtain Figure   (3.5) of [ @RS ], which we regurgitate here: $ $ \xymatrix { { \mathbb{Z}}^V \oplus { \mathbb{Z}}^W \oplus { \mathbb{Z}}^W \oplus \ldots \ar[d]_{1 - D } \ar[r]^D & { \mathbb{Z}}^V \oplus { \mathbb{Z}}^W \oplus { \mathbb{Z}}^W \oplus \ldots \ar[d]_{1 - D } \ar[r]^-{\iota_*^{n+1 } } & K_0(C^ * (

0(C^*(E \himes_1 [m,n])) \cong \bigoplus_{v \ln V} {\mathbb{Z}}[p_{(v,n)}]_0 \oklys \bigmplus_{k=0}^{h-m} \bigopuus_{v \in W} {\mathbb{Z}}[p_{(v,n-k)}]_0$$ and $K_0(C^*(X \tines_1 [m,n]))^+$ is identified with $$\bieoplus_{v \ij V} {\mathvb{N}}[p_{(t,n)}]_0 \oplus \bigoplus_{k=0}^{n-m} \bigoijus_{v \ln W} {\kethbb{N}}[p_{(v,n-k)}]_0.$$ By tme continuidy of $K$-theory, mnd ean let $m$ tend to $-\infty$ and deduce trat $$\begon{wligned} K_0(C^*(E \timgs_1 [-\ingey,n])) &\dong \bigoplus_{v \in V} {\mathbb{Z}}[p_{(v,n)}]_0 \oplhs \bigoklus_{k=0}^{\infty} \bigoplux_{v \in W} {\mathbb{Z}}[p_{(v,n-k)}]_0 \\ &\cong {\mahhbb{X}}^V \oplus Z^W \oplus X^W \oplus \ldits.\ege{aligned}$$ Furghermore, iu yollows froj [@RLL Theorem 6.3.2(ii)] that this isomofphisk identifigs $K_0(F^*(G \times_1 [-\inftb,n]))^+$ witr ${\mathbb{N}}^V \oikus {\madhbb{N}}^W \pplus {\mathbb{N}}^W \opnus \ldots$. This computatioi is used later in tre proof mf [@RS Theorem 3.1], where tye $K_0$ xuncdor kw aopljev tk a colmuvative diagdam to obtaun Figure (3.5) of [@RS], whivh qe reproduce gere: $$\xrmwtrix{ {\mathbb{Z}}^V \oplus {\mathbb{Z}}^W \oplus {\matvbb{A}}^W \oplus \ldots \ar[d]_{1-D} \ar[r]^D & {\mathbb{Z}}^V \oplus {\mathhb{Z}}^W \opluf {\mathbb{Z}}^W \oplus \ldots \ar[d]_{1-D} \ar[r]^-{\iota_*^{n+1}} & K_0(C^*(

0(C^*(E \times_1 [m,n])) \cong \bigoplus_{v \in V} \bigoplus_{k=0}^{n-m} \in W} and $K_0(C^*(E \times_1 \in {\mathbb{N}}[p_{(v,n)}]_0 \oplus \bigoplus_{k=0}^{n-m} \in W} {\mathbb{N}}[p_{(v,n-k)}]_0.$$ the continuity of $K$-theory, one can $m$ tend to $-\infty$ and deduce that $$\begin{aligned} K_0(C^*(E \times_1 [-\infty,n])) &\cong \bigoplus_{v V} {\mathbb{Z}}[p_{(v,n)}]_0 \oplus \bigoplus_{k=0}^{\infty} \bigoplus_{v \in W} {\mathbb{Z}}[p_{(v,n-k)}]_0 \\ &\cong {\mathbb{Z}}^V \oplus Z^W Z^W \ldots.\end{aligned}$$ it from [@RLL Theorem 6.3.2(ii)] that this isomorphism identifies $K_0(C^*(E \times_1 [-\infty,n]))^+$ with ${\mathbb{N}}^V \oplus {\mathbb{N}}^W \oplus \oplus \ldots$. This computation is used later in proof of [@RS Theorem where the $K_0$ functor is to commutative diagram obtain (3.5) [@RS], which we here: $$\xymatrix{ {\mathbb{Z}}^V \oplus {\mathbb{Z}}^W \oplus {\mathbb{Z}}^W \oplus \ldots \ar[d]_{1-D} \ar[r]^D & {\mathbb{Z}}^V \oplus {\mathbb{Z}}^W \oplus {\mathbb{Z}}^W \ldots \ar[d]_{1-D} K_0(C^*(

0(C^*(E \times_1 [m,n])) \cong \bigoplus_{v \in V} {\Mathbb{Z}}[p_{(v,n)}]_0 \Oplus \BigOplUs_{K=0}^{n-m} \bIgopLus_{v \in W} {\mathbb{Z}}[P_{(V,n-k)}]_0$$ aNd $K_0(C^*(E \times_1 [m,n]))^+$ is identifiEd witH $$\bIGoplUS_{v \In V} {\maThbb{N}}[p_{(v,N)}]_0 \OpLUS \biGoPlUs_{k=0}^{N-m} \BIgOplus_{V \in w} {\mathbb{n}}[p_{(v,n-k)}]_0.$$ By the ConTiNuity of $K$-theoRY, oNe can let $m$ tEnd To $-\infty$ and deDucE that $$\bEgIn{aLIgned} k_0(C^*(E \Times_1 [-\Infty,n])) &\COng \bigOplus_{v \in V} {\MaTHbb{Z}}[p_{(v,N)}]_0 \Oplus \biGOPlUs_{k=0}^{\iNfty} \bigoplus_{v \in W} {\mAThBB{Z}}[p_{(v,n-k)}]_0 \\ &\cong {\mathBb{Z}}^V \opLuS z^W \OPLus z^W \oPlus \ldots.\eNd{AlignED}$$ FurtheRMoRE, IT foLLows from [@RLL ThEorem 6.3.2(ii)] that THis IsomorPhIsm IDentifIes $K_0(C^*(e \tIMes_1 [-\Infty,n]))^+$ with ${\mAthbB{N}}^V \oplus {\mAthbb{N}}^w \Oplus {\maTHbb{N}}^W \opLus \ldoTs$. THis CompUTaTiOn iS uSEd lATeR in THe pRoof of [@RS thEoRem 3.1], whEre tHE $k_0$ FUnctOr iS appLied tO a commutative DiaGram TO obTain FIgure (3.5) Of [@RS], WhIch we ReprodUce heRe: $$\Xymatrix{ {\mathbb{Z}}^v \oplUs {\mathbb{Z}}^w \opLuS {\maThBb{Z}}^W \oPLus \ldoTs \aR[d]_{1-D} \Ar[r]^D & {\matHbb{Z}}^V \opLUs {\mAtHBB{z}}^W \Oplus {\mathbb{Z}}^W \oplus \LdOTS \aR[d]_{1-D} \ar[r]^-{\ioTa_*^{n+1}} & K_0(C^*(

0(C^*(E \times_1 [m,n])) \ cong \bigo plus_ {v\in V } {\ math bb{Z}}[p_{(v,n ) }]_0 \oplus \bigoplus_{k=0 }^{n- m} \big o pl us_{v \in W} {\ m a thb b{ Z} }[p _{ ( v, n-k)} ]_0 $$ and$K_0(C^*(E \t im es_1 [m,n])) ^ +$ is identi fie d with $$\bi gop lus_{v \ inV } {\m ath bb{N} }[p_{( v ,n)}]_ 0 \oplus\b i goplus _ {k=0}^{ n - m} \bi goplus_{v \in W}{ \m a thbb{N}}[p_{(v ,n-k)} ]_ 0 .$ $ Bythe continuit yof $K $ -theory , o n e can let $m$ tendto $-\infty $ an d dedu ce th a t $$\b egin{ al i gne d} K_0(C^*( E \t imes_1 [- \infty , n])) &\ c ong \bi goplus _{v \i n V} {\ ma thb b{ Z }}[ p _{ (v, n )}] _0 \oplu s\b igopl us_{ k = 0 } ^{\i nft y} \ bigop lus_{v \in W} {\ math b b{Z }}[p_ {(v,n -k)} ]_ 0 \\&\cong {\ma th bb{Z}}^V \oplus Z^W \oplus Z ^W\o plu s\ldot s .\end{ ali gne d}$$ Fu rthermo r e,it f o ll ows from [@RLL The or e m  6 .3.2(ii) ] that th is isomorph is m i dent i f ies $ K_0( C ^* (E \time s_1 [- \ in ft y,n]))^ +$ with${ \ma thb b{N}} ^ V \o plus { \mathbb{ N}}^W \oplus {\mathb b {N}}^W \oplus \l d o ts $ . T his computatio n is used lat e rint he pr oof o f[ @R S Theorem 3.1], wher ethe $K _0$ f unctor is app lied to ac o m mutative dia g ra m to obtain Fig ure ( 3.5) of [@ R S], whic h wereproduc e here: $ $ \ xymatrix { { \ma thb b{Z } } ^V \oplus {\mat h b b{Z} }^ W \oplu s { \mathbb {Z} }^W \o plu s\ldots \a r[d]_{1- D} \ ar [r ]^D & {\ m athbb{Z} }^ V \ op lus {\ma t hbb{Z} }^W \ oplu s{\ m ath bb{Z}}^ W \ o p lus\l do ts \ ar[ d] _{1-D } \a r [r] ^-{\iot a_*^{n+1} } & K_0( C^ *(

0(C^*(E \times_1_[m,n])) \cong_\bigoplus_{v \in V} {\mathbb{Z}}[p_{(v,n)}]_0 \oplus_\bigoplus_{k=0}^{n-m} \bigoplus_{v_\in_W} {\mathbb{Z}}[p_{(v,n-k)}]_0$$_and_$K_0(C^*(E \times_1 [m,n]))^+$ is_identified with $$\bigoplus_{v_\in V} {\mathbb{N}}[p_{(v,n)}]_0 \oplus \bigoplus_{k=0}^{n-m}_\bigoplus_{v \in W}_{\mathbb{N}}[p_{(v,n-k)}]_0.$$_By the continuity of $K$-theory, one can let $m$ tend to $-\infty$ and deduce_that_$$\begin{aligned} K_0(C^*(E \times_1_[-\infty,n]))_&\cong_\bigoplus_{v \in V} {\mathbb{Z}}[p_{(v,n)}]_0 \oplus_\bigoplus_{k=0}^{\infty} \bigoplus_{v \in W} {\mathbb{Z}}[p_{(v,n-k)}]_0 \\ &\cong_{\mathbb{Z}}^V \oplus_Z^W \oplus Z^W \oplus \ldots.\end{aligned}$$ Furthermore, it follows_from_[@RLL Theorem 6.3.2(ii)] that_this isomorphism identifies $K_0(C^*(E \times_1 [-\infty,n]))^+$ with ${\mathbb{N}}^V \oplus_{\mathbb{N}}^W \oplus {\mathbb{N}}^W \oplus \ldots$. This computation_is used later_in_the_proof of [@RS Theorem 3.1],_where the $K_0$ functor is applied_to a commutative diagram to obtain_Figure (3.5) of [@RS], which we reproduce here:_$$\xymatrix{ {\mathbb{Z}}^V \oplus {\mathbb{Z}}^W \oplus {\mathbb{Z}}^W_\oplus \ldots \ar[d]_{1-D} \ar[r]^D & {\mathbb{Z}}^V_\oplus {\mathbb{Z}}^W_\oplus {\mathbb{Z}}^W \oplus \ldots \ar[d]_{1-D} \ar[r]^-{\iota_*^{n+1}}_& K_0(C^*(

{(-1)^{l-1}z^{l}}{l}k^{r-l} \label{eqn;326}\\[8pt] & \quad = \frac{1}{(n-1)!}\sum_{\mu = -1}^{n-2} \left\{ \sum_{r=\mu+1}^{n-1} \frac{\sideset{_{n-1}}{_r}S}{r-\mu}z^{r-\mu}\right\} (-1)^{\mu+1}k^{\mu}.\nonumber \end{aligned}$$]{} Since (\[eqn;324\]) and [ $$\begin{aligned} & (z+k)^{j}\log \left(1+\frac{z}{k}\right) + \sum_{r=0}^{j}\binom{j}{r}z^{j-r} \sum_{l=1}^{r+1}\frac{(-1)^{l}z^{l}}{l} k^{r-l}\\[4pt] & \qquad = O(k^{-2}) \quad \mbox{as} \quad k \to \infty, \end{aligned}$$]{} we have [ $$\begin{aligned} & - \binom{-k}{n-1}\log \left(1+ \frac{z}{k}\right) + \sum_{j=0}^{n-1}G_{n,j}(z)\sum_{r=0}^{j} \binom{j}{r}z^{j-r} \sum_{l=1}^{r+1}\frac{(-1)^{l-1}z^{l}}{l} k^{r-l}\label{eqn;327}\\[4pt] & \qquad = O(k^{-2}) \quad \mbox{as} \quad k\to \infty, \nonumber \end{aligned}$$]{} while we obtain [ $$\begin{aligned} & - \binom{-k}{n-1}\log\left(1+\frac{z}{k}\right) + \frac{1}{(n-1)!}\sum_{\mu=-1}^{n-2}\left\{ \sum_{r=\mu+1}^{n-1}\frac{\s

{ (-1)^{l-1}z^{l}}{l}k^{r - l } \label{eqn;326}\\[8pt ] & \quad = \frac{1}{(n-1)!}\sum_{\mu = -1}^{n-2 } \left\ { \sum_{r=\mu+1}^{n-1 } \frac{\sideset{_{n-1}}{_r}S}{r-\mu}z^{r-\mu}\right\ } (-1)^{\mu+1}k^{\mu}.\nonumber \end{aligned}$$ ] { } Since (\[eqn;324\ ]) and [ $ $ \begin{aligned } & (z+k)^{j}\log \left(1+\frac{z}{k}\right) + \sum_{r=0}^{j}\binom{j}{r}z^{j - r } \sum_{l=1}^{r+1}\frac{(-1)^{l}z^{l}}{l } k^{r - l}\\[4pt ] & \qquad = O(k^{-2 }) \quad \mbox{as } \quad k \to \infty, \end{aligned}$$ ] { } we have [ $ $ \begin{aligned } & - \binom{-k}{n-1}\log \left(1 + \frac{z}{k}\right) + \sum_{j=0}^{n-1}G_{n, j}(z)\sum_{r=0}^{j } \binom{j}{r}z^{j - r } \sum_{l=1}^{r+1}\frac{(-1)^{l-1}z^{l}}{l } k^{r - l}\label{eqn;327}\\[4pt ] & \qquad = O(k^{-2 }) \quad \mbox{as } \quad k\to \infty, \nonumber \end{aligned}$$ ] { } while we obtain [ $ $ \begin{aligned } & - \binom{-k}{n-1}\log\left(1+\frac{z}{k}\right) + \frac{1}{(n-1)!}\sum_{\mu=-1}^{n-2}\left\ { \sum_{r=\mu+1}^{n-1}\frac{\s

{(-1)^{l-1}z^{l}}{p}k^{r-l} \label{eqn;326}\\[8pt] & \quad = \frac{1}{(n-1)!}\sum_{\mu = -1}^{n-2} \lert\{ \sum_{r=\mj+1}^{n-1} \frac{\sideset{_{n-1}}{_r}S}{r-\mu}z^{r-\lu}\eight\} (-1)^{\mu+1}k^{\mu}.\nonumber \evd{aligned}$$]{} Dince (\[eqb;324\]) anv [ $$\begin{aligned} & (z+k)^{j}\log \lerb(1+\frac{v}{k}\cight) + \suk_{r=0}^{j}\binom{j}{r}s^{j-r} \sum_{l=1}^{s+1}\ffae{(-1)^{l}z^{l}}{l} k^{r-l}\\[4pt] & \qquad = O(k^{-2}) \quad \mbox{af} \quad l \ho \infty, \end{ajigntd}$$]{} re hzne [ $$\begin{aligned} & - \binom{-k}{n-1}\lof \left(1+ \hrac{z}{k}\right) + \sum_{j=0}^{n-1}G_{n,j}(z)\sum_{r=0}^{j} \binlm{j}{r}x^{j-r} \sum_{l=1}^{r+1}\frac{(-1)^{p-1}z^{l}}{l} k^{t-m}\lafwl{eqn;327}\\[4pt] & \dquad = O(k^{-2}) \quad \mbox{as} \suad k\to \infty, \nonumber \evd{aliyned}$$]{} while qe obhdin [ $$\begin{aoigneq} & - \binom{-k}{n-1}\log\laft(1+\frac{a}{k}\right) + \frec{1}{(n-1)!}\sym_{\mu=-1}^{n-2}\left\{ \sum_{r=\mn+1}^{n-1}\frac{\s

{(-1)^{l-1}z^{l}}{l}k^{r-l} \label{eqn;326}\\[8pt] & \quad = \frac{1}{(n-1)!}\sum_{\mu = \sum_{r=\mu+1}^{n-1} (-1)^{\mu+1}k^{\mu}.\nonumber \end{aligned}$$]{} (\[eqn;324\]) and [ \sum_{r=0}^{j}\binom{j}{r}z^{j-r} k^{r-l}\\[4pt] & \qquad O(k^{-2}) \quad \mbox{as} k \to \infty, \end{aligned}$$]{} we have $$\begin{aligned} & - \binom{-k}{n-1}\log \left(1+ \frac{z}{k}\right) + \sum_{j=0}^{n-1}G_{n,j}(z)\sum_{r=0}^{j} \binom{j}{r}z^{j-r} \sum_{l=1}^{r+1}\frac{(-1)^{l-1}z^{l}}{l} k^{r-l}\label{eqn;327}\\[4pt] & \qquad O(k^{-2}) \quad \mbox{as} \quad k\to \infty, \nonumber \end{aligned}$$]{} while we obtain [ $$\begin{aligned} - + \sum_{r=\mu+1}^{n-1}\frac{\s

{(-1)^{l-1}z^{l}}{l}k^{r-l} \label{eqn;326}\\[8pt] & \quad = \frac{1}{(N-1)!}\sum_{\mu = -1}^{n-2} \lefT\{ \sum_{r=\Mu+1}^{n-1} \FraC{\sIdesEt{_{n-1}}{_r}s}{r-\mu}z^{r-\mu}\right\} (-1)^{\mU+1}K^{\mu}.\nOnumber \end{aligned}$$]{} Since (\[Eqn;324\]) anD [ $$\bEGin{aLIgNed} & (z+k)^{J}\log \lefT(1+\FrAC{Z}{k}\rIgHt) + \Sum_{R=0}^{j}\BInOm{j}{r}z^{J-r} \sUm_{l=1}^{r+1}\fraC{(-1)^{l}z^{l}}{l} k^{r-l}\\[4pt] & \QquAd = o(k^{-2}) \quad \mbox{as} \QUaD k \to \infty, \eNd{aLigned}$$]{} we have [ $$\BegIn{aligNeD} & - \biNOm{-k}{n-1}\lOg \lEft(1+ \frAc{z}{k}\riGHt) + \sum_{j=0}^{N-1}G_{n,j}(z)\sum_{r=0}^{J} \bINom{j}{r}z^{J-R} \sum_{l=1}^{r+1}\fRAC{(-1)^{l-1}Z^{l}}{l} k^{R-l}\label{eqn;327}\\[4pt] & \qquad = o(K^{-2}) \qUAd \mbox{as} \quad k\tO \infty, \NoNUmBER \enD{alIgned}$$]{} while We ObtaiN [ $$\Begin{alIGnED} & - \BInoM{-K}{n-1}\log\left(1+\frac{Z}{k}\right) + \frac{1}{(N-1)!}\Sum_{\Mu=-1}^{n-2}\lefT\{ \sUm_{r=\MU+1}^{n-1}\frac{\S

{(-1)^{l-1}z^{l}}{l}k^{r-l } \l abel{ eqn ;32 6} \\[8 pt] & \quad =\ frac {1}{(n-1)!}\sum_{\mu = -1}^ {n - 2} \le ft\{ \s u m_ { r =\m u+ 1} ^{n -1 } \f rac{\si deset{_{n- 1}} {_ r}S}{r-\mu}z ^ {r -\mu}\righ t\} (-1)^ {\m u+1}k^ {\ mu} . \nonu mbe r \ end{al i gned}$ $]{} Sin ce (\[eqn ; 324\])a n d[ $$ \begin{aligned} & (z+k)^{j}\lo g \lef t( 1 +\ f r ac{ z}{ k}\right) + \sum_{ r =0 } ^ { j}\ b inom{j}{r}z^{ j-r} \su m_{l=1 }^ {r+ 1 }\frac {(-1) ^{ l }z^ {l}}{l} k^{ r-l} \\[4pt] & \ q quad =O (k^{-2} ) \qua d \ mbo x{as } \ qu adk\ to\ in fty , \end{ali gn ed }$$]{ } we h a v e [$$\ begi n{ali gned} & - \b inom { -k} {n-1} \log\lef t( 1+ \f rac{z} {k}\r ig ht) + \su m_{j =0}^{n-1} G_{ n, j}( z) \sum_ { r=0}^{ j} \bin om{j}{r } z^{ j- r } \sum_{l=1}^{r+ 1} \ f ra c{(-1)^{ l-1}z^ { l} }{ l } k^ {r- l}\l a b el{eq n;32 7 }\ \[4pt] & \ q qu ad = O(k^ {- 2}) \q ua d \ mbo x{as} \qua d k\to \infty, \nonumber \ e nd{aligned}$$ ] {} w hi l e we ob tain [ $$\b egin { alig ned} & - \ binom {- k }{ n -1}\log\left(1+\fra c{ z}{k}\ right ) + \f rac{1}{(n- 1 ) ! }\sum_{\ mu=- 1 }^ { n-2}\left\{ \sum_{r=\m u +1}^{n-1 }\fra c{\s

{(-1)^{l-1}z^{l}}{l}k^{r-l} _ _ \label{eqn;326}\\[8pt] _ _&_\quad =_\frac{1}{(n-1)!}\sum_{\mu_= -1}^{n-2} _ _ \left\{ \sum_{r=\mu+1}^{n-1} _ __ \frac{\sideset{_{n-1}}{_r}S}{r-\mu}z^{r-\mu}\right\} (-1)^{\mu+1}k^{\mu}.\nonumber \end{aligned}$$]{} Since (\[eqn;324\]) and [ $$\begin{aligned} __ _&__(z+k)^{j}\log \left(1+\frac{z}{k}\right) _ + \sum_{r=0}^{j}\binom{j}{r}z^{j-r} _ _ \sum_{l=1}^{r+1}\frac{(-1)^{l}z^{l}}{l} k^{r-l}\\[4pt] __& \qquad =_O(k^{-2}) \quad \mbox{as} \quad k \to \infty, \end{aligned}$$]{}_we have [ $$\begin{aligned} _ & -_\binom{-k}{n-1}\log_\left(1+_\frac{z}{k}\right) _ + \sum_{j=0}^{n-1}G_{n,j}(z)\sum_{r=0}^{j} _ \binom{j}{r}z^{j-r} _ \sum_{l=1}^{r+1}\frac{(-1)^{l-1}z^{l}}{l} _ k^{r-l}\label{eqn;327}\\[4pt] _& \qquad = O(k^{-2}) \quad_\mbox{as} \quad_k\to \infty, _ \nonumber _ \end{aligned}$$]{}_while we obtain_[ $$\begin{aligned} _& -_\binom{-k}{n-1}\log\left(1+\frac{z}{k}\right) __+ \frac{1}{(n-1)!}\sum_{\mu=-1}^{n-2}\left\{ ___ _ \sum_{r=\mu+1}^{n-1}\frac{\s

:tracevstrace} t^w\bigl({\mathsf{Tr}}_{\mathcal{F}}(\delta_x(a) \circ b)\bigr) = {\mathsf{tr}}(a \circ b_x),$$ where ${\mathsf{Tr}}_{\mathcal{F}}: \operatorname{\mathsf{Ext}}({\mathcal{F}}, {\mathcal{F}}) {\longrightarrow}H^1({\mathcal{O}})$ is the trace map and $t^w$ is the composition $H^0({\mathcal{O}}) \stackrel{w}{\longrightarrow}H^0(\Omega) \stackrel{t}{\longrightarrow}{\mathbbm{k}}$ of the isomorphism induced by $w$ and the canonical map $t$ described in [@BK4 Subsection 4.3]. *Comment on the proof*. The first part of the statement is just [@BK4 Lemma 4.18]. The content of the second part is explained by the following commutative diagram: $$\xymatrix{ 0 \ar[r] & {\mathcal{F}}\ar[r] \ar[d] & {\mathcal{Q}}\ar[r] \ar[d] & {\mathcal{F}}\ar[r] \ar[d]^b & 0 \\ 0 \ar[r] & {\mathcal{F}}\ar[r] \ar[d] & {\mathcal{R}}\ar[r] \ar[d] & {\mathbbm{k}}_x \ar[r] \ar[d]^a & 0 \\ 0 \ar[r] & {\mathcal{F}}\ar[r]^-\imath & {\mathcal{F}}(x) \ar[r]^-{\underline{\operatorname{\mathsf{res}}}_x^{\mathcal{F}}} & {\mathcal{F}}\otimes {\mathbbm{k}}_x \ar[r] & 0. }$$ The lowest horizontal sequence of this diagram is (\[E:resVectBundle\]). The middle sequence corresponds to the element $\delta_x(a) \in \operatorname{\mathsf{Ext}}({\mathbbm{k}}_x, {\mathcal{F}})$ and the top one corresponds to $\delta_x(a) \circ b \in \operatorname{\mathsf{Ext}}({\mathcal{F}}, {\mathcal{F}})$. The endomorphism $a \circ b_x \in \operatorname{\mathsf{End}}({\mathcal{F}}\big|_x)$ is the induced map in the fiber of ${\mathcal{F}}$ over $

: tracevstrace } t^w\bigl({\mathsf{Tr}}_{\mathcal{F}}(\delta_x(a) \circ b)\bigr) = { \mathsf{tr}}(a \circ b_x),$$ where $ { \mathsf{Tr}}_{\mathcal{F } }: \operatorname{\mathsf{Ext}}({\mathcal{F } }, { \mathcal{F } }) { \longrightarrow}H^1({\mathcal{O}})$ is the trace map and $ t^w$ is the composition $ H^0({\mathcal{O } }) \stackrel{w}{\longrightarrow}H^0(\Omega) \stackrel{t}{\longrightarrow}{\mathbbm{k}}$ of the isomorphism induced by $ w$ and the basic function $ t$ described in [ @BK4 Subsection 4.3 ]. * Comment on the validation *. The first share of the statement is merely [ @BK4 Lemma 4.18 ]. The content of the second part is explain by the following commutative diagram: $ $ \xymatrix { 0 \ar[r ] & { \mathcal{F}}\ar[r ] \ar[d ] & { \mathcal{Q}}\ar[r ] \ar[d ] & { \mathcal{F}}\ar[r ] \ar[d]^b & 0 \\ 0 \ar[r ] & { \mathcal{F}}\ar[r ] \ar[d ] & { \mathcal{R}}\ar[r ] \ar[d ] & { \mathbbm{k}}_x \ar[r ] \ar[d]^a & 0 \\ 0 \ar[r ] & { \mathcal{F}}\ar[r]^-\imath & { \mathcal{F}}(x) \ar[r]^-{\underline{\operatorname{\mathsf{res}}}_x^{\mathcal{F } } } & { \mathcal{F}}\otimes { \mathbbm{k}}_x \ar[r ] & 0. } $ $ The lowest horizontal succession of this diagram is (\[E: resVectBundle\ ]). The middle sequence corresponds to the element $ \delta_x(a) \in \operatorname{\mathsf{Ext}}({\mathbbm{k}}_x, { \mathcal{F}})$ and the top one corresponds to $ \delta_x(a) \circ b \in \operatorname{\mathsf{Ext}}({\mathcal{F } }, { \mathcal{F}})$. The endomorphism $ a \circ b_x \in \operatorname{\mathsf{End}}({\mathcal{F}}\big|_x)$ is the induce map in the roughage of $ { \mathcal{F}}$ over $

:trafevstrace} t^w\bigl({\mathsn{Tr}}_{\mathcal{F}}(\delta_r(q) \circ b)\bigr) = {\mathsf{gr}}(a \circ b_x),$$ where ${\mathsf{Tr}}_{\mavhcao{F}}: \optgatorname{\mathsf{Ext}}({\matfcal{F}}, {\matjcal{F}}) {\lobgrijhtarrow}H^1({\mathcal{O}})$ is the trace lap cnv $t^w$ is the comkosition $H^0({\madhcal{O}}) \stackren{w}{\uougrightarrow}H^0(\Omega) \stackrel{t}{\longrighearrow}{\mstjbbm{k}}$ of the ifomogprism pnbuced by $w$ and the canonical mal $t$ desbribed in [@BK4 Subsrction 4.3]. *Comment on the proov*. Thf first part of thf statement is tyst [@BK4 Lemma 4.18]. The conttnc of the sedond part is explained by the fullownng commutajnce fhagram: $$\xymavrix{ 0 \ag[r] & {\mathcal{F}}\av[g] \ar[d] & {\kathcal{A}}\ar[r] \ar[d] & {\mathgal{F}}\ac[r] \ae[d]^b & 0 \\ 0 \ar[r] & {\mathcal{F}}\ac[r] \ar[d] & {\mathcal{R}}\ar[r] \wr[d] & {\mathtbj{k}}_x \ar[r] \ar[d]^a & 0 \\ 0 \ae[r] & {\matvcal{X}}\ar[r]^-\knatf & {\methdal{F}}(x) \wr[r]^-{\nnderline{\opsratorname{\mqthsf{res}}}_x^{\mathcal{F}}} & {\mwnncal{F}}\otimes {\jathbbi{k}}_v \ar[r] & 0. }$$ The lowest horizontal sequence of this diagram is (\[E:resVextBundle\]). The middle sgquence cowresponds to the element $\delta_x(a) \in \operatorname{\mdthsf{Xxg}}({\mauhnnm{k}}_x, {\nahhcal{F}})$ and the top one corresponds to $\delta_x(a) \ditc b \in \operatorkame{\mathsf{Ext}}({\mathcsl{V}}, {\kwthcal{F}})$. The evdomor'giam $a \circ b_x \in \opfratornwme{\marhsf{End}}({\mauhcal{G}}\big|_x)$ is the induced map in the fiber oy ${\mqthcal{F}}$ over $

:tracevstrace} t^w\bigl({\mathsf{Tr}}_{\mathcal{F}}(\delta_x(a) \circ b)\bigr) = {\mathsf{tr}}(a \circ ${\mathsf{Tr}}_{\mathcal{F}}: {\mathcal{F}}) {\longrightarrow}H^1({\mathcal{O}})$ the trace map $H^0({\mathcal{O}}) \stackrel{t}{\longrightarrow}{\mathbbm{k}}$ of the induced by $w$ the canonical map $t$ described in Subsection 4.3]. *Comment on the proof*. The first part of the statement is [@BK4 Lemma 4.18]. The content of the second part is explained by the commutative $$\xymatrix{ \ar[r] {\mathcal{F}}\ar[r] \ar[d] & {\mathcal{Q}}\ar[r] \ar[d] & {\mathcal{F}}\ar[r] \ar[d]^b & 0 \\ 0 \ar[r] & {\mathcal{F}}\ar[r] \ar[d] {\mathcal{R}}\ar[r] \ar[d] & {\mathbbm{k}}_x \ar[r] \ar[d]^a & 0 0 \ar[r] & {\mathcal{F}}\ar[r]^-\imath {\mathcal{F}}(x) \ar[r]^-{\underline{\operatorname{\mathsf{res}}}_x^{\mathcal{F}}} & {\mathcal{F}}\otimes {\mathbbm{k}}_x & }$$ The horizontal of diagram is (\[E:resVectBundle\]). middle sequence corresponds to the element $\delta_x(a) \in \operatorname{\mathsf{Ext}}({\mathbbm{k}}_x, {\mathcal{F}})$ and the top one corresponds to $\delta_x(a) b \in The endomorphism \circ \in is the induced the fiber of ${\mathcal{F}}$ over $

:tracevstrace} t^w\bigl({\mathsf{TR}}_{\mathcal{F}}(\dElta_x(A) \ciRc b)\BiGr) = {\maThsf{Tr}}(a \circ b_x),$$ where ${\MAthsF{Tr}}_{\mathcal{F}}: \operatornamE{\mathSf{eXt}}({\maTHcAl{F}}, {\maThcal{F}}) {\lONgRIGhtArRoW}H^1({\mAtHCaL{O}})$ is tHe tRace map And $t^w$ is the ComPoSition $H^0({\mathcAL{O}}) \Stackrel{w}{\lOngRightarrow}H^0(\OMegA) \stackReL{t}{\lONgrigHtaRrow}{\mAthbbm{K}}$ Of the iSomorphisM iNDuced bY $W$ and the CANoNicaL map $t$ described in [@Bk4 suBSection 4.3]. *Comment On the pRoOF*. THE FirSt pArt of the stAtEment IS just [@BK4 lEmMA 4.18]. tHe cONtent of the secOnd part is exPLaiNed by tHe FolLOwing cOmmutAtIVe dIagram: $$\xymatRix{ 0 \aR[r] & {\mathcal{f}}\ar[r] \ar[D] & {\Mathcal{q}}\Ar[r] \ar[d] & {\mAthcal{f}}\ar[R] \ar[D]^b & 0 \\ 0 \ar[R] & {\MaThCal{f}}\aR[R] \ar[D] & {\MaThcAL{R}}\aR[r] \ar[d] & {\matHbBm{K}}_x \ar[r] \Ar[d]^a & 0 \\ 0 \AR[R] & {\MAthcAl{F}}\Ar[r]^-\iMath & {\mAthcal{F}}(x) \ar[r]^-{\unDerLine{\OPerAtornAme{\maThsf{ReS}}}_x^{\matHcal{F}}} & {\mAthcaL{F}}\Otimes {\mathbbm{k}}_x \Ar[r] & 0. }$$ THe lowest hOriZoNtaL sEquenCE of thiS diAgrAm is (\[E:reSVectBuNDle\]). thE MIDdLe sequence correspoNdS TO tHe elemenT $\delta_X(A) \iN \oPEratornaMe{\MatHsf{EXT}}({\MathbBm{k}}_x, {\MAtHcal{F}})$ and The top ONe CoRresponDs To $\deltA_x(A) \ciRc b \In \opeRAtorName{\maThsf{Ext}}({\mAthcaL{f}}, {\mathcal{F}})$. The enDOmorphism $a \cirC B_x \IN \OpERatoRnaMe{\mathsf{End}}({\MathCAl{F}}\bIg|_x)$ iS ThE inDUced mAp in tHe FIbER of ${\mathcal{F}}$ over $

:tracevstrace} t^w\big l({\mathsf {Tr}} _{\ mat hc al{F }}(\ delta_x(a) \ci r c b) \bigr) = {\mathsf{tr}} (a \c ir c b_x ) ,$ $ whe re ${\m a th s f {Tr }} _{ \ma th c al {F}}: \o perator name{\math sf{ Ex t}}({\mathca l {F }}, {\math cal {F}}) {\long rig htarro w} H^1 ( {\mat hca l{O}} )$ ist he tra ce map an d$ t^w$ i s the co m p os itio n $H^0({\mathcal{ O }} ) \stackrel{w}{ \longr ig h ta r r ow} H^0 (\Omega) \ st ackre l {t}{\lo n gr i g h tar r ow}{\mathbbm{ k}}$ of the iso morphi sm in d uced b y $w$ a n d t he canonica l ma p $t$ des cribed in [@BK 4 Subsec tion 4 .3] . *Com m en tonth e pr o of *.T hefirst pa rt o f the sta t e m e nt i s j ust[@BK4 Lemma 4.18]. Th e co n ten t ofthe s econ dpartis exp laine dby the followin g co mmutative di ag ram :$$\xy m atrix{ 0\ar [r] & { \mathca l {F} }\ a r [ r] \ar[d] & {\mathca l{ Q } }\ ar[r] \a r[d] & {\ ma t hcal{F}} \a r[r ] \a r [ d]^b& 0\ \0 \ar[r] & {\m a th ca l{F}}\a r[ r] \ar [d ] & {\ mathc a l{R} }\ar[r ] \ar[d] & {\ m athbbm{k}}_x \ a r[r] \ar[d]^a &0 \\ 0 \a r[r ] & {\mathc al{F } }\ar [r]^ - \i mat h & { \math ca l {F } }(x) \ar[r]^-{\unde rl ine{\o perat orname{\maths f{res}}}_x ^ { \ mathcal{ F}}} & {\mathcal{F}}\ otime s {\mathbb m {k}}_x \ ar[r] & 0. } $$ The lo w e st horiz ont alseq uen c e o f this diagra m is ( \[ E:resVe ctB undle\] ).The mi ddl esequencecorrespo nd sto t heeleme n t $\delt a_ x(a )\in \ope r atorna me{\m aths f{ Ex t }}( {\mathb b m{ k } }_x, { \m athc al{ F} })$ a nd t h e t op onecorrespon dst o $\ de lt a_x(a)\circ b \in \ op eratorname {\ mat hsf{Ex t } }({\math cal{F}}, {\mathcal{F}}) $ . The e ndo morph ism$a \circb_x \in \ ope r atorna me{\ma thsf{ En d}} ( { \math c a l{ F}} \b ig|_x)$ is t heinduc ed map in the fiber of ${\mathc a l{F }}$ over $

:tracevstrace} _ _t^w\bigl({\mathsf{Tr}}_{\mathcal{F}}(\delta_x(a) \circ b)\bigr) =_{\mathsf{tr}}(a \circ_b_x),$$_where ${\mathsf{Tr}}_{\mathcal{F}}:_\operatorname{\mathsf{Ext}}({\mathcal{F}},_{\mathcal{F}}) {\longrightarrow}H^1({\mathcal{O}})$ is_the trace map_and $t^w$ is the_composition $H^0({\mathcal{O}}) \stackrel{w}{\longrightarrow}H^0(\Omega)_\stackrel{t}{\longrightarrow}{\mathbbm{k}}$_of the isomorphism induced by $w$ and the canonical map $t$ described in [@BK4_Subsection_4.3]. *Comment on_the_proof*._The first part of the_statement is just [@BK4 Lemma_4.18]. The_content of the second part is explained by_the_following commutative diagram:_$$\xymatrix{ 0 \ar[r] & {\mathcal{F}}\ar[r] \ar[d] & {\mathcal{Q}}\ar[r] \ar[d] &_{\mathcal{F}}\ar[r] \ar[d]^b & 0 \\ 0 \ar[r]_& {\mathcal{F}}\ar[r] \ar[d]_&_{\mathcal{R}}\ar[r]_\ar[d] & {\mathbbm{k}}_x \ar[r]_\ar[d]^a & 0 \\ 0 \ar[r] &_{\mathcal{F}}\ar[r]^-\imath & {\mathcal{F}}(x) \ar[r]^-{\underline{\operatorname{\mathsf{res}}}_x^{\mathcal{F}}} _& {\mathcal{F}}\otimes {\mathbbm{k}}_x \ar[r] & 0. }$$_The lowest horizontal sequence of this_diagram is (\[E:resVectBundle\]). The middle_sequence corresponds_to the element $\delta_x(a) \in_\operatorname{\mathsf{Ext}}({\mathbbm{k}}_x, {\mathcal{F}})$ and_the top_one corresponds to_$\delta_x(a) \circ b \in \operatorname{\mathsf{Ext}}({\mathcal{F}}, {\mathcal{F}})$._The endomorphism $a_\circ b_x \in \operatorname{\mathsf{End}}({\mathcal{F}}\big|_x)$ is the_induced_map in the_fiber_of_${\mathcal{F}}$ over_$

interesting ways, which can be observed experimentally, using microscopic probes. For instance, both the order parameter, $\eta (\bf R)$, and magnetic field, $\bf B (R)$, are altered near the impurity. These perturbations therefore provide important insight into the nature of pinning. From a theoretical perspective, the calculation of the perturbations in $\eta (\bf R)$ and $\bf B (R)$ constitutes a non-trivial application of the Thuneberg theory. In this work, we investigate the changes of the order parameter $\delta \eta (\bf R)$, and magnetic field $\delta \bf B(R)$, due to the impurity, when a vortex core is placed on a defect. Near the defect these quantities can be calculated analytically. Away from the defect, $\delta \eta (\bf R)$ decays exponentially over the correlation length scale, $\xi (T)$, while $\delta \bf B(R)$ decays exponentially over the penetration length scale $\lambda (T)$. The paper is organized as follows. In Section II we review the appropriate GL theory, and set up the general equations for our problem. These equations, (\[eq:def1\]) and (\[eq:dea1\]), are the main results of this section. They are a pair of coupled, linear, partial differential equations for the impurity-induced changes in the order parameter and vector potential. In Section III we consider in particular the spatial region close to the impurity, where $\delta \eta (\bf R)$ and $\delta \bf B(R)$ are largest. In this region we are able to derive explicit expressions for the perturbations—the simplicity of the result for $\delta \bf B(R)$ is somewhat remarkable. Eqs. (\[eq:dpsi2\]-\[eq:da2\]) contain our main results. In Section IV we discuss the results, and estimate the magnitude of the perturbation effects. In the appendix we calculate, numerically, the GL solutions for vortices in the absence of impurities. GL Equations ============ General Formulation ------------------- For simplicity, in this paper we consider a superconductor with a spin singlet, isotropic ([*i.e.*]{} $\hat{\bf k}$-independent) energy gap. The GL free energy, in terms of the complex order parameter $\eta (\bf R)$ and vector potential $\bf A (R)$, is given, in the absence of the impurity, by $$\begin{aligned} \Omega_{\rm pure} & = & \

interesting ways, which can be observed experimentally, using microscopic probes. For case, both the decree parameter, $ \eta (\bf R)$, and magnetic field, $ \bf B (R)$, are change near the impurity. These perturbations therefore supply important insight into the nature of trap. From a theoretical perspective, the calculation of the perturbations in $ \eta (\bf R)$ and $ \bf B (R)$ establish a non - trivial application of the Thuneberg hypothesis. In this work, we investigate the changes of the order parameter $ \delta \eta (\bf R)$, and magnetic field $ \delta \bf B(R)$, due to the impurity, when a vortex kernel is placed on a defect. Near the defect these quantity can be calculated analytically. Away from the defect, $ \delta \eta (\bf R)$ disintegrate exponentially over the correlation length scale, $ \xi (T)$, while $ \delta \bf B(R)$ decays exponentially over the penetration length plate $ \lambda (T)$. The paper is organized as follows. In Section   II we review the appropriate GL theory, and set up the general equation for our problem. These equations, (\[eq: def1\ ]) and (\[eq: dea1\ ]), are the main results of this section. They are a pair of coupled, linear, partial differential equations for the impurity - induce changes in the order argument and vector electric potential. In Section III we consider in particular the spatial region close to the impurity, where $ \delta \eta (\bf R)$ and $ \delta \bf B(R)$ are big. In this region we are able to derive explicit expressions for the disruption — the simplicity of the result for $ \delta \bf B(R)$ is somewhat remarkable. Eqs.   (\[eq: dpsi2\]-\[eq: da2\ ]) hold our main results. In Section IV we discuss the results, and calculate the magnitude of the disruption consequence. In the appendix we calculate, numerically, the GL solutions for vortices in the absence of impurities. GL Equations = = = = = = = = = = = = General Formulation ------------------- For simplicity, in this paper we consider a superconductor with a tailspin singlet, isotropic ([ * i.e. * ] { } $ \hat{\bf k}$-independent) energy col. The GL free energy, in term of the complex order parameter $ \eta (\bf R)$ and vector potential $ \bf A (R)$, is given, in the absence of the impurity, by $ $ \begin{aligned } \Omega_{\rm pure } & = & \

inheresting ways, which can be observed exkeeimentelly, usjng micruscopic probes. For instance, uoth the irder parameter, $\eta (\bf R)$, and mahnetic fueld, $\vf B (R)$, are altered near fme im'ucity. These pertorbations tharefore provida km'ortant insight into the nature of pynning. Grlm a theoreticwl ptrs[ectjne, the calculation of the pertudbationv in $\eta (\bf R)$ and $\bf B (R)$ constitutes a jon-tgivial application of the Thubebewt theory. In tfis work, wt nnvestigate the changes of the order paramdter $\belta \eta (\bd E)$, ajg magnetic hield $\qelta \bf B(R)$, due to tve imputity, when a vovtex rore is placed on a defecv. Near the defect thgse quantidizs can be calculated qnqlytiwallf. Awxt ffom tie sefect, $\demta \eta (\bf R)$ decays ezponentially over tne bprrelation lsngth fcwle, $\xi (T)$, while $\delta \bf B(R)$ decays expontntiamly over the penetratiob length scale $\lambda (T)$. The papqr is organized as follows. In Section II we review the eporokrlwgw HL theory, and set up the general equations fow oir problem. These equations, (\[rq:feg1\]) and (\[eq:dea1\]), arg the mcjn results of this sfction. Jhey aee a pair of voupled, linear, partial diffwrential equctiins for the impuricy-induced chcnges on thr order parameter and vzctor lotential. In Section JKI we consider iv psrdicular ufe spatial region close to the impurigy, wnere $\dqlta \eta (\bv R)$ akg $\delta \bf B(R)$ are pargedt. In this rfgion we are able to derive explicit expressipnv fmr the pzrturbstions—the sim[licity of the result for $\aelta \bf B(G)$ is somexhat remarkafle. Eqs. (\[eq:dpsi2\]-\[ae:da2\]) contain mur main resylts. In Secgkon IV we disciss the rvsblts, and wstimate the magnibude ur the perturbatnub effects. In thr aopegdpx xe cajwulate, numerhcaluy, gne GL solutions nor voryices in the absence of jmpurities. GL Equatooks ============ General Formulaeion ------------------- For simplocity, in this papeg we ronsidxr a sipetconductor with a spin singlet, jsotropic ([*i.e.*]{} $\hat{\bf k}$-inde[endcnt) gnergy gap. Che GL free energy, in terms of the complxx order parameter $\eta (\vf R)$ and vector potgntlal $\bf A (R)$, ix givqn, in the absence of the impueity, by $$\begin{aligked} \Omega_{\rm pure} & = & \

interesting ways, which can be observed experimentally, probes. instance, both order parameter, $\eta $\bf (R)$, are altered the impurity. These therefore provide important insight into the of pinning. From a theoretical perspective, the calculation of the perturbations in $\eta R)$ and $\bf B (R)$ constitutes a non-trivial application of the Thuneberg theory. this we the of the order parameter $\delta \eta (\bf R)$, and magnetic field $\delta \bf B(R)$, due to impurity, when a vortex core is placed on defect. Near the defect quantities can be calculated analytically. from defect, $\delta (\bf decays over the correlation scale, $\xi (T)$, while $\delta \bf B(R)$ decays exponentially over the penetration length scale $\lambda (T)$. The is organized In Section we the GL theory, and the general equations for our problem. and (\[eq:dea1\]), are the main results of this They are pair of coupled, linear, partial differential for the impurity-induced changes in the order parameter vector potential. In Section III we consider in particular the spatial region close to the $\delta \eta (\bf R)$ $\delta \bf B(R)$ largest. this we able to explicit expressions for the perturbations—the simplicity of the result for $\delta B(R)$ is somewhat remarkable. Eqs. (\[eq:dpsi2\]-\[eq:da2\]) contain our main results. IV discuss the results, estimate the magnitude of perturbation In the appendix we the solutions the of GL Equations ============ General ------------------- For simplicity, in this we consider a superconductor ([*i.e.*]{} $\hat{\bf k}$-independent) energy gap. The GL free in terms of the complex order parameter (\bf R)$ and vector potential $\bf A (R)$, is given, in the of the $$\begin{aligned} \Omega_{\rm pure} & = & \

interesting ways, which can be Observed exPerimEntAllY, uSing MicrOscopic probes. FOR insTance, both the order paramEter, $\eTa (\BF R)$, anD MaGnetiC field, $\bF b (R)$, ARE alTeReD neAr THe ImpurIty. these peRturbationS thErEfore provide IMpOrtant insiGht Into the naturE of PinninG. FRom A TheorEtiCal peRspectIVe, the cAlculatioN oF The perTUrbatioNS In $\Eta (\bF R)$ and $\bf B (R)$ constituTEs A Non-trivial applIcatioN oF ThE tHunEbeRg theory. In ThIs worK, We invesTIgATE The CHanges of the orDer parameteR $\DelTa \eta (\bF R)$, And MAgnetiC fielD $\dELta \Bf B(R)$, due to thE impUrity, when A vorteX Core is pLAced on a Defect. neaR thE defECt ThEse QuANtiTIeS caN Be cAlculateD aNaLyticAlly. aWAY From The DefeCt, $\delTa \eta (\bf R)$ decayS exPoneNTiaLly ovEr the CorrElAtion Length Scale, $\Xi (t)$, while $\delta \bf B(R)$ DecaYs exponenTiaLlY ovEr The peNEtratiOn lEngTh scale $\Lambda (T)$. tHe pApER IS oRganized as follows. IN SECTiOn II we reView thE ApPrOPriate GL ThEorY, and SET up thE genERaL equatioNs for oUR pRoBlem. TheSe EquatiOnS, (\[eq:Def1\]) And (\[eq:DEa1\]), arE the maIn resultS of thIS section. They arE A pair of coupleD, LiNEAr, PArtiAl dIfferential EquaTIons For tHE iMpuRIty-inDuced ChANgES in the order parameteR aNd vectOr potEntial. In SectiOn III we conSIDEr in partIculAR tHE spatial region Close To the impurITy, where $\dElta \eTa (\bf R)$ and $\Delta \bf B(R)$ ARE largest. in tHis RegIon WE ArE able to derive EXPlicIt ExpressIonS for the PerTurBatIonS—tHe simplicIty of the ReSuLt FoR $\deLta \bf b(r)$ is somewHaT reMaRkaBle. EqS. (\[Eq:dpsi2\]-\[Eq:da2\]) cOntaIn OuR MaiN resultS. in sECtioN Iv wE disCusS tHe resUlts, ANd eStimate The magnitUde OF the PeRtUrbatioN effects. In the ApPendix we caLcUlaTe, numeRICally, the gL solutions for vortices iN The abseNce Of impUritIes. GL EquaTioNs ============ GeneRal fOrmulaTion ------------------- FoR simpLiCitY, IN this PAPeR we CoNsider a supERConDuctoR wIth a Spin sinGlet, isotropic ([*i.e.*]{} $\hat{\BF k}$-iNdependent) eneRgy Gap. THE gL FreE EnERgy, In TErmS OF the complex ordeR parameter $\EtA (\Bf r)$ and vector POteNtIal $\bf A (R)$, Is given, In the ABsence oF the impurIty, by $$\begiN{aLignED} \omeGa_{\rm pure} & = & \

interesting ways, which c an be obse rvedexp eri me ntal ly,using microsco p ic p robes. For instance, b oth t he orde r p arame ter, $\ e ta ( \bf R )$ , a nd ma gneti c f ield, $ \bf B (R)$ , a re altered nea r t he impurit y.These pertur bat ions t he ref o re pr ovi de im portan t insig ht into t he nature of pinn i n g. Fro m a theoretical p e rs p ective, the ca lculat io n o f the pe rturbation sin $\ e ta (\bf R) $ a nd$ \bf B (R)$ co nstitutes a non -trivi al ap p licati on of t h e T huneberg th eory . In thi s work , we inv e stigate the c han ges oft he o rde rp ara m et er$ \de lta \eta ( \b f R)$ , an d m a gnet icfiel d $\d elta \bf B(R) $,duet o t he im purit y, w he n a v ortexcoreis placed on a de fect . Near th e d ef ect t heseq uantit ies ca n be ca lculate d an al y t i ca lly. Away from the d e f ec t, $\del ta \et a ( \b f R)$ dec ay s e xpon e n tiall y ov e rthe corr elatio n l en gth sca le , $\xi ( T)$ , w hile$ \del ta \bf B(R)$ d ecays exponentiallyo ver the penet r at i o nl engt h s cale $\lamb da ( T )$. The pa per is or ganiz ed as follows. In Section  I I we r eview the appropri ate GL the o r y , and se t up th e general equat ionsfor our pr o blem. Th ese e quations , (\[eq:d e f 1\]) and (\ [eq :de a1\ ] ) ,are the mainr e sult sof this se ction.The y a rea p ai r of coup led, lin ea r, p ar tia l dif f erential e qua ti ons fort he imp urity -ind uc ed cha nges in th e orde rpa rame ter a nd ve ctor pot ential. In Sect ion IIIwe c onsider in particula rthe spatia lreg ion cl o s e to the impurity, where $\delt a \eta ( \bf R)$and$\delta \ bfB(R)$are larges t. Inthisre gio n we ar e ab leto derive ex p l ici t exp re ssio ns forthe perturbations— t hesimplicity of th e re s u lt fo r $ \ del ta \bf B (R)$ is somewha t remarkab le . E qs. (\[eq: d psi 2\ ]-\[eq: da2\])conta i n our m ain resul ts. In Se ct ionI V we discuss t he resul ts, and e s timat e t he ma gni tude o fthe pert urbati o n e ffect s. Inth e appe ndixwe calcula te, numerically, the GL solut ionsfor vortices in the absenceof i mpurities. G L E quati ons ===== ==== = == G e neral For m ulation - - -- --- - - -- -------- F o r sim plici ty, in thi s pa per we consider a superconductor wit h a s pin sing le t, isotropic ( [*i .e . * ]{} $\ha t{ \bf k}$-ind ependent )e nergy gap.The GL free e n e rg y , in t erms of the comp lex o r der par am et e r $\et a ( \b f R)$and ve c torp o tential $\bf A ( R)$,i s give n , i n the a bsenceo f th e impurity , by $$\beg in{ali gned } \Om ega_{\r mpure}& = & \

interesting_ways, which_can be observed experimentally,_using microscopic_probes._For instance,_both_the order parameter,_$\eta (\bf R)$,_and magnetic field, $\bf_B (R)$, are_altered_near the impurity. These perturbations therefore provide important insight into the nature of pinning._From_a theoretical_perspective,_the_calculation of the perturbations in_$\eta (\bf R)$ and $\bf_B (R)$_constitutes a non-trivial application of the Thuneberg theory. In_this_work, we investigate_the changes of the order parameter $\delta \eta (\bf_R)$, and magnetic field $\delta \bf_B(R)$, due to_the_impurity,_when a vortex core_is placed on a defect. Near_the defect these quantities can be_calculated analytically. Away from the defect, $\delta_\eta (\bf R)$ decays exponentially over_the correlation length scale, $\xi_(T)$, while_$\delta \bf B(R)$ decays exponentially_over the penetration_length scale_$\lambda (T)$. The paper_is organized as follows. In Section II_we review the_appropriate GL theory, and set up_the_general equations for_our_problem._These equations,_(\[eq:def1\]) and (\[eq:dea1\]),_are_the main_results_of this section. They are a_pair_of coupled, linear, partial differential equations for_the impurity-induced changes in_the_order parameter and vector_potential. In Section III we consider_in particular the spatial region close_to the_impurity, where_$\delta \eta (\bf R)$ and $\delta \bf B(R)$ are largest. In_this region we are able to_derive explicit expressions for_the perturbations—the_simplicity_of the result_for_$\delta \bf_B(R)$ is somewhat remarkable. Eqs. (\[eq:dpsi2\]-\[eq:da2\]) contain our_main results._In Section IV we discuss the_results, and estimate the_magnitude_of the perturbation effects. In the_appendix we calculate, numerically, the GL_solutions for vortices in the_absence_of_impurities. GL Equations ============ General Formulation ------------------- For simplicity, in_this paper we consider a superconductor_with a spin_singlet, isotropic ([*i.e.*]{} $\hat{\bf k}$-independent) energy gap._The_GL free energy, in terms of_the_complex order parameter $\eta (\bf R)$_and_vector_potential $\bf A (R)$, is_given, in the absence of the_impurity, by $$\begin{aligned} \Omega_{\rm pure} & = & \

2 {\rm ms} < T < 10 {\rm ms}$, the 2 copies, that separate by $2v_R\times T$, are recombined using a second $\frac{\pi}{2}$ pulse, that creates 2 interferometer output (see fig. \[interf\_graph\]). For finite T, complementary fringes appear in the two output ports, because of the smoothly varying expansion phase that results from the conversion of a small fraction of the initial mean-field energy ($\sim \epsilon^2 \mu$) into axial kinetic energy. The resulting fringe pattern is Fourier analyzed to find the contrast and the fringe spacing in reciprocal space. In this way, we filter out shot-to-shot, global phase fluctuations that would result in a translation of the fringe pattern as a whole. This method allows for averaging many processed images without artificially reducing the contrast. After this processing, this method allows a direct measurement of the correlation function $\mathcal{C}^{(1)}$ as a function of separation $s$ (see figure \[interf\_result\]), irrespective of the amount of phase fluctuations. Conclusion ========== We have demonstrated two important features of elongated Bose-Einstein condensates: (i) the dimension of the condensate after expansion; (ii) the momentum distribution of quasicondensates; Our method, along with atom interferometry can be applied to investigate how long range order develops during the condensate growth. [99]{} A. G[ö]{}rlitz, J. M. Vogels, A. E. Leanhardt, C. Raman, T. L. Gustavson, J. R. Abo-Shaeer, A. P. Chikkatur, S. Gupta, S. Inouye, T.Rosenband, and W. Ketterle, Phys. Rev. Lett. [**87**]{}, 130402 (2001). F. Schreck, L. Khaykovich, K. L. Corwin, G. Ferrari, T. Bourdel, J.Cubizolles, and C. Salomon Phys. Rev. Lett. [**87**]{}, 080403 (2001). M. Greiner, I. Bloch, O. Mandel, T. W. Hänsch, and T.

2 { \rm ms } < T < 10 { \rm ms}$, the 2 copies, that separate by $ 2v_R\times T$, are recombined using a second $ \frac{\pi}{2}$ pulsation, that create 2 interferometer output (see fig. \[interf\_graph\ ]). For finite T, complemental fringes look in the two output ports, because of the smoothly vary expansion phase that leave from the conversion of a small fraction of the initial mean - field energy ($ \sim \epsilon^2 \mu$) into axile kinetic energy. The resulting fringe pattern is Fourier analyze to find the contrast and the fringe spacing in reciprocal space. In this manner, we filter out shot - to - shot, ball-shaped phase fluctuations that would result in a transformation of the fringe traffic pattern as a whole. This method allow for averaging many processed images without artificially reduce the contrast. After this processing, this method allows a direct measurement of the correlation affair $ \mathcal{C}^{(1)}$ as a function of separation $ s$ (see figure \[interf\_result\ ]), irrespective of the amount of phase fluctuations. Conclusion = = = = = = = = = = We have demonstrated two important features of elongate Bose - Einstein condensate: (i) the proportion of the condensation after expansion; (ii) the momentum distribution of quasicondensates; Our method acting, along with atom interferometry can be applied to investigate how long range order develops during the condensate growth. [ 99 ] { } A.   G[ö]{}rlitz, J.   M.   Vogels, A.   E.   Leanhardt, C.   Raman, T.   L.   Gustavson, J.   R. Abo - Shaeer, A.   P.   Chikkatur, S.   Gupta, S.   Inouye, T.Rosenband, and W.   Ketterle, Phys.   Rev.   Lett.   [ * * 87 * * ] { }, 130402 (2001). F.   Schreck, L.   Khaykovich, K.   L.   Corwin, G.   Ferrari, T.   Bourdel, J.Cubizolles, and C.   Salomon Phys.   Rev.   Lett.   [ * * 87 * * ] { }, 080403 (2001). M.   Greiner, I.   Bloch, O.   Mandel, T.   W.   Hänsch, and T.

2 {\rm ms} < T < 10 {\rm ms}$, the 2 copits, that separate yt $2v_R\tikes T$, zre recoobined using a second $\frac{\pi}{2}$ pylse, ukat creates 2 interfefometer oltput (see fig. \[unterf\_grapi\]). For fiknte T, gomplzmxntary fringes sppear in dhe two output purcs, because of the smoothly varying evpansiom ohase that resolts gwom fhe conversion of a small fraction of the initial mean-gield energy ($\sim \epsilon^2 \mk$) inho axial kinetic ejergy. The rgauleung fringe pxttern is Fourier analized to find the contrast and thd friuge spacing ib rfwiprocal spece. In this way, we filter mut shoy-to-shot, global phese dluctuations that wound result in a tragslation mf the fringe patteen as a whone. Tfus oetgov amlows vor averaging many procewsed images without awnoficially resucing tre contrast. After this processing, this ketgod allows a direct meawurement of the correpation fugction $\mathcal{C}^{(1)}$ as a function of separation $s$ (see figuce \[inuevn\_resjot\]), irrespective of the amount of phase fluctuatyknx. Cpnclusion ========== We haye demonstrated twp lmljrtant featurgs of elonfated Bose-Einstein condenfates: (i) the diiensoon of the condensate after expansion; (ip) thw momentum distribbtion of quaricomdensstes; Our method, along wnth atkm interferlmetry cah be applied to ivvexthgate how long range order develops durnng the zondgnsate drowth. [99]{} A. G[ö]{}rlltz, J. M. Vogels, A. E. Leanhardt, C. Ramcn, T. L. Custavson, U. R. Abo-Shaeer, A. P. Chikkatur, S. Gupta, S. Inouye, T.Rosembdnd, and W. Kecterle, Phys. Rev. Lett. [**87**]{}, 130402 (2001). F. Schreck, L. Khaikovich, K. J. Corwkn, G. Ferrarp, T. Bourden, J.Cubizollqs, and C. Salommj Phys. Rev. Letv. [**87**]{}, 080403 (2001). M. Greinqr, I. Vlocy, O. Manddu, T. W. Hänsch, and Y.

2 {\rm ms} < T < 10 the copies, that by $2v_R\times T$, $\frac{\pi}{2}$ that creates 2 output (see fig. For finite T, complementary fringes appear the two output ports, because of the smoothly varying expansion phase that results the conversion of a small fraction of the initial mean-field energy ($\sim \epsilon^2 into kinetic The fringe pattern is Fourier analyzed to find the contrast and the fringe spacing in reciprocal space. this way, we filter out shot-to-shot, global phase that would result in translation of the fringe pattern a This method for many images without artificially the contrast. After this processing, this method allows a direct measurement of the correlation function $\mathcal{C}^{(1)}$ as function of (see figure irrespective the of phase fluctuations. We have demonstrated two important features condensates: (i) the dimension of the condensate after (ii) the distribution of quasicondensates; Our method, along atom interferometry can be applied to investigate how range order develops during the condensate growth. [99]{} A. G[ö]{}rlitz, J. M. Vogels, A. E. Raman, T. L. Gustavson, R. Abo-Shaeer, A. Chikkatur, Gupta, Inouye, and W. Phys. Rev. Lett. [**87**]{}, 130402 (2001). F. Schreck, L. Khaykovich, K. Corwin, G. Ferrari, T. Bourdel, J.Cubizolles, and C. Salomon Phys. [**87**]{}, (2001). M. Greiner, Bloch, O. Mandel, T. Hänsch, T.

2 {\rm ms} < T < 10 {\rm ms}$, the 2 copies, that sepArate by $2v_R\tImes T$, Are RecOmBineD usiNg a second $\frac{\pI}{2}$ PulsE, that creates 2 interferomEter oUtPUt (seE FiG. \[inteRf\_graph\]). fOr FINitE T, CoMplEmENtAry frIngEs appeaR in the two oUtpUt Ports, because OF tHe smoothly VarYing expansioN phAse thaT rEsuLTs froM thE convErsion OF a smalL fraction Of THe initIAl mean-fIELd EnerGy ($\sim \epsilon^2 \mu$) intO AxIAl kinetic energY. The reSuLTiNG FriNge Pattern is FOuRier aNAlyzed tO FiND THe cONtrast and the fRinge spacinG In rEciproCaL spACe. In thIs way, We FIltEr out shot-to-Shot, Global phaSe flucTUations THat woulD resulT in A trAnslATiOn Of tHe FRinGE pAttERn aS a whole. THiS mEthod AlloWS FOR aveRagIng mAny prOcessed images WitHout ARtiFiciaLly reDuciNg The coNtrast. after ThIs processing, thiS metHod allows A diReCt mEaSuremENt of thE coRreLation fUnction $\MAthCaL{c}^{(1)}$ AS a Function of separatiOn $S$ (SEe Figure \[inTerf\_reSUlT\]), iRRespectiVe Of tHe amOUNt of pHase FLuCtuationS. ConclUSiOn ========== we have dEmOnstraTeD twO imPortaNT feaTures oF elongatEd BosE-einstein condenSAtes: (i) the dimenSIoN OF tHE conDenSate after exPansIOn; (ii) The mOMeNtuM DistrIbutiOn OF qUAsicondensates; Our meThOd, alonG with Atom interferoMetry can be APPLied to inVestIGaTE how long range oRder dEvelops durINg the conDensaTe growth. [99]{} a. G[ö]{}rlitz, J. m. vOgels, A. E. LEanHarDt, C. ramAN, t. L. gustavson, J. R. AbO-sHaeeR, A. p. ChikkaTur, s. Gupta, S. inoUye, t.RoSenBaNd, and W. KetTerle, PhyS. REv. leTt. [**87**]{}, 130402 (2001). f. ScHreck, l. khaykoviCh, k. L. COrWin, g. FerrARi, T. BouRdel, J.cubiZoLlES, anD C. SalomON PHYS. Rev. leTt. [**87**]{}, 080403 (2001). m. GreIneR, I. bloch, o. ManDEl, T. w. Hänsch, And T.

2 {\rm ms} < T < 10 {\rm m s}$, the 2 copi es, th at sep arat e by $2v_R\tim e s T$ , are recombined using a se co n d $\ f ra c{\pi }{2}$ p u ls e , th at c rea te s 2 inte rfe rometer output (s eefi g. \[interf\ _ gr aph\]). Fo r f inite T, com ple mentar yfri n ges a ppe ar in the t w o outp ut ports, b e causeo f the s m o ot hlyvarying expansion ph a se that result s from t h ec o nve rsi on of a sm al l fra c tion of th e i nit i al mean-field energy ($\ s im\epsil on ^2\ mu$) i nto a xi a l k inetic ener gy.The resul ting f r inge pa t tern is Fouri erana lyze d t ofin dt hec on tra s t a nd the f ri ng e spa cing i n reci pro calspace . In this way , w e fi l ter outshot- to-s ho t, gl obal p hasefl uctuations that wou ld result in a tr an slati o n of t hefri nge pat tern as a w ho l e . T his method allowsfo r av eragingmany p r oc es s ed image swit hout a rtifi cial l yreducing the c o nt ra st. Aft er thispr oce ssi ng, t h is m ethodallows a dire c t measuremento f the correla t io n fu n ctio n $ \mathcal{C} ^{(1 ) }$ a s af un cti o n ofsepar at i on $s$ (see figure \[i nt erf\_r esult \]), irrespec tive of th e a mount of pha s ef luctuations. Concl usion ==== = ===== W e hav e demons trated tw o importan t f eat ure s o f el ongated Bose- E i nste in conden sat es: (i) th e d ime nsi on of the c ondensat eaf te rexp ansio n ; (ii) t he mo me ntu m dis t ributi on of qua si co n den sates;O ur m etho d, a long wi th atom int e rfe rometry can be a ppl i ed t oin vestiga te how long r an ge order d ev elo ps dur i n g the co ndensate growth. [99]{ } A. G[ ö]{ }rlit z, J . M. Voge ls, A. E.  Le a nhardt , C. R aman, T . L . Gusta v s on , J .R. Abo-Sha e e r,A. P.  C hikk atur, S . Gupta, S. Inouye , T. Rosenband, an d W . Ke t t er le, Ph y s.Re v . L e t t. [**87**]{},130402 (20 01 ) . F. Schrec k , L .Khaykov ich, K.  L. C o rwin, G . Ferrari , T. Bour de l, J . C ubi zolles, an d C. Sal omon Phys .  Rev. Le tt. [ **8 7**]{} ,080 403 ( 2001). M.  Grei ner, I .Bloch, O. M an del, T.W. Hänsch, and T.

2 {\rm_ms} <_T < 10 {\rm_ms}$, the_2_copies, that_separate_by $2v_R\times T$,_are recombined using_a second $\frac{\pi}{2}$ pulse,_that creates 2_interferometer_output (see fig. \[interf\_graph\]). For finite T, complementary fringes appear in the two output_ports,_because of_the_smoothly_varying expansion phase that results_from the conversion of a_small fraction_of the initial mean-field energy ($\sim \epsilon^2 \mu$)_into_axial kinetic energy._The resulting fringe pattern is Fourier analyzed to find_the contrast and the fringe spacing_in reciprocal space._In_this_way, we filter out_shot-to-shot, global phase fluctuations that would_result in a translation of the_fringe pattern as a whole. This method_allows for averaging many processed images_without artificially reducing the contrast._After this_processing, this method allows a_direct measurement of_the correlation_function $\mathcal{C}^{(1)}$ as_a function of separation $s$ (see_figure \[interf\_result\]), irrespective_of the amount of phase fluctuations. Conclusion ========== We_have_demonstrated two important_features_of_elongated Bose-Einstein_condensates: (i) the_dimension_of the_condensate_after expansion; (ii) the momentum distribution_of_quasicondensates; Our method, along with atom interferometry_can be applied to_investigate_how long range order_develops during the condensate growth. [99]{} A. G[ö]{}rlitz,_J. M. Vogels, A. E. Leanhardt, C. Raman, T. L. Gustavson, J. R. Abo-Shaeer,_A. P. Chikkatur, S. Gupta,_S. Inouye, T.Rosenband,_and W. Ketterle, Phys. Rev. Lett. [**87**]{}, 130402 (2001). F. Schreck, L. Khaykovich, K. L. Corwin, G. Ferrari, T. Bourdel, J.Cubizolles, and_C. Salomon Phys. Rev. Lett. [**87**]{}, 080403 (2001). M. Greiner, I. Bloch, O. Mandel,_T. W. Hänsch, and T.

[*Probab. Theory and Math. Statist. Proc. VII International Vilnius Conference (1998)*]{}. Vilnius, VSP/TEV, pp. 453-468, 1999. and [W. Linde.]{} Approximation and entropy numbers of Volterra operators with application to Brownian motion. [*Mem. Amer. Math. Soc.*]{} [**745**]{}, 2002. and [W. Linde.]{} Small deviations of weighted fractional processes and average non-linear approximation. Submitted to [*Stochastic Processes and their Applications*]{}, 2002. and [P. M. Robinson.]{} Alternative forms of fractional Brownian motion. [*J. Stat. Plann. Inference*]{} [**80**]{} (1-2), pp. 111-122, 1999. Cambridge studies in advanced mathematics, Cambridge, 1992. Small deviations in a space of trajectories. [*Theor. Probab. Appl.*]{} [**19**]{}, pp. 726-736, 1974. Mouvement brownien et espaces de Besov. [*Stoch. Stoch. Rep.*]{} [**43**]{} (3-4), pp. 221-260, 1993. Asymptotic behavior of stable seminorms near the origin. [*Ann. Probab.*]{} [**14**]{} (1), pp. 287-298, 1986. Lower tails of self–similar stable processes. [*Bernoulli*]{} [**4**]{} (1), pp. 127-142, 1998. and [M. S. Taqqu.]{} [*Stable non-Gaussian random processes.*]{} Chapman & Hall, New York, 1994. A note on small ball probability of a Gaussian process with stationary increments. [*J. Theoret. Probab.*]{} [**6**]{} (3), pp. 595-602, 1993. A Gaussian correlation inequality and its application to the existence of small ball constant. [*Preprint*]{}, 1999. Lower tails of quadratic functionals of symmetric stable processes. [*Prépublication de l’Université Paris 6*]{}, 1999. Une méthode élémentaire pour l’évaluation des petites boules browniennes. [*C. R. Acad. Sci. Paris*]{} [**316**]{} (11), pp. 1217-1220, 1993. Small ball probabilities for Gaussian processes under non-uniform norms. [*J. Theoret. Probab.*]{}

[ * Probab. Theory and Math. Statist. Proc. VII International Vilnius Conference (1998) * ] { }. Vilnius, VSP / TEV, pp. 453 - 468, 1999. and [ W. Linde. ] { } Approximation and entropy numbers of Volterra operator with lotion to Brownian motion. [ * Mem. Amer. Math. Soc. * ] { } [ * * 745 * * ] { }, 2002. and [ W. Linde. ] { } Small deviation of leaden fractional processes and average non - linear estimate. resign to [ * Stochastic Processes and their Applications * ] { }, 2002. and [ P.   M.   Robinson. ] { } Alternative forms of fractional Brownian motion. [ * J. Stat. Plann. Inference * ] { } [ * * 80 * * ] { } (1 - 2), pp. 111 - 122, 1999. Cambridge studies in advance mathematics, Cambridge, 1992. Small deviations in a space of trajectories. [ * Theor.   Probab.   Appl. * ] { }   [ * * 19 * * ] { }, pp. 726 - 736, 1974. Mouvement brownien et espaces de Besov. [ * Stoch. Stoch. Rep. * ] { } [ * * 43 * * ] { } (3 - 4), pp. 221 - 260, 1993. Asymptotic demeanor of stable seminorms near the origin. [ * Ann. Probab. * ] { } [ * * 14 * * ] { } (1), pp. 287 - 298, 1986. low tails of self – similar static processes. [ * Bernoulli * ] { } [ * * 4 * * ] { } (1), pp. 127 - 142, 1998. and [ M.   S.   Taqqu. ] { } [ * Stable non - Gaussian random processes. * ] { } Chapman & Hall, New York, 1994. A eminence on small ball probability of a Gaussian procedure with stationary increments. [ * J. Theoret. Probab. * ] { } [ * * 6 * * ] { } (3), pp. 595 - 602, 1993. A Gaussian correlation coefficient inequality and its application to the universe of small ball constant. [ * Preprint * ] { }, 1999. Lower tail of quadratic functionals of symmetric stable processes. [ * Prépublication de l’Université Paris 6 * ] { }, 1999. Une méthode élémentaire pour l’évaluation des petites boules browniennes. [ * C. R. Acad. Sci. Paris * ] { } [ * * 316 * * ] { } (11), pp. 1217 - 1220, 1993. Small musket ball probabilities for Gaussian processes under non - uniform norms. [ * J. Theoret. Probab. * ] { }

[*Prlbab. Theory and Math. Stauist. Proc. VII Intgrbationel Vilnjus Confdrence (1998)*]{}. Vilnius, VSP/TEV, pp. 453-468, 1999. aid [W. Lindt.]{} Approximation and entropy jumbers if Vioterra opecztors wlch aplpicaciin to Brownian motion. [*Mek. Amer. Math. Sow.*]{} [**745**]{}, 2002. aud [W. Linde.]{} Small deviations of weigheed fravtlonal processef anc avedage non-linear approximation. Submifted to [*Stochastic Ptocesses and their Applicahiond*]{}, 2002. and [P. M. Robinson.]{} Apternative dormf of fractionxl Brownian motion. [*J. Sjat. Plann. Inference*]{} [**80**]{} (1-2), pp. 111-122, 1999. Cambriage scudies in aevqncfg mathematirs, Camfridge, 1992. Small deviatimns in s space of tramectocies. [*Theor. Probab. Appl.*]{} [**19**]{}, pp. 726-736, 1974. Kouvement brownien et espacas de Besov. [*Stoch. Sroxh. Rek.*]{} [**43**]{} (3-4), p[. 221-260, 1993. Artmpgotjc bshaviog oh stable sejinorms neae the origin. [*Ann. Prpbwv.*]{} [**14**]{} (1), pp. 287-298, 1986. Lower tails os self–similar stable processes. [*Bernoullp*]{} [**4**]{} (1), lp. 127-142, 1998. and [M. S. Taqqu.]{} [*Stable bon-Gaussian random prlcesses.*]{} Crapman & Hall, New York, 1994. A note on small ball probabhlity uf c Gausruaj process with stationary increments. [*J. Theoree. Ptonab.*]{} [**6**]{} (3), pp. 595-602, 1993. A Gausfian correlstlom inequality avd its apllication to the edistencg of snall ball conxtant. [*Preprint*]{}, 1999. Lower tails if quadratic dunctionals of symletric stabue ptocessrs. [*Prépublication de l’Unnversifé Paris 6*]{}, 1999. Unf méthode éméoentaire pour l’évxlusthon des kdtites boules brorniennes. [*R. R. Aead. Sci. Oarix*]{} [**316**]{} (11), pp. 1217-1220, 1993. Small bapl probabilities for Gaudsian psocesses ujder non-uniform norms. [*J. Theoret. Probab.*]{}

[*Probab. Theory and Math. Statist. Proc. VII Conference Vilnius, VSP/TEV, 453-468, 1999. and numbers Volterra operators with to Brownian motion. Amer. Math. Soc.*]{} [**745**]{}, 2002. and Linde.]{} Small deviations of weighted fractional processes and average non-linear approximation. Submitted to Processes and their Applications*]{}, 2002. and [P. M. Robinson.]{} Alternative forms of fractional motion. Stat. Inference*]{} (1-2), pp. 111-122, 1999. Cambridge studies in advanced mathematics, Cambridge, 1992. Small deviations in a space trajectories. [*Theor. Probab. Appl.*]{} [**19**]{}, pp. 726-736, 1974. brownien et espaces de [*Stoch. Stoch. Rep.*]{} [**43**]{} (3-4), 221-260, Asymptotic behavior stable near origin. [*Ann. Probab.*]{} (1), pp. 287-298, 1986. Lower tails of self–similar stable processes. [*Bernoulli*]{} [**4**]{} (1), pp. 127-142, 1998. and S. Taqqu.]{} random processes.*]{} & New 1994. A note ball probability of a Gaussian process [*J. Theoret. Probab.*]{} [**6**]{} (3), pp. 595-602, 1993. Gaussian correlation and its application to the existence small ball constant. [*Preprint*]{}, 1999. Lower tails of functionals of symmetric stable processes. [*Prépublication de l’Université Paris 6*]{}, 1999. Une méthode élémentaire pour petites boules browniennes. [*C. Acad. Sci. Paris*]{} (11), 1217-1220, Small probabilities for processes under non-uniform norms. [*J. Theoret. Probab.*]{}

[*Probab. Theory and Math. StatisT. Proc. VII InTernaTioNal viLniuS ConFerence (1998)*]{}. Vilnius, vsP/TEv, pp. 453-468, 1999. and [W. Linde.]{} ApproximatIon anD eNTropY NuMbers Of VolteRRa OPEraToRs WitH aPPlIcatiOn tO BrowniAn motion. [*MeM. AmEr. math. Soc.*]{} [**745**]{}, 2002. and [W. LINdE.]{} Small deviAtiOns of weighteD frActionAl ProCEsses And AveraGe non-lINear apProximatiOn. sUbmittED to [*StocHAStIc PrOcesses and their ApPLiCAtions*]{}, 2002. and [P. M. RobInson.]{} ALtERnATIve ForMs of fractiOnAl BroWNian motIOn. [*j. sTAt. PLAnn. Inference*]{} [**80**]{} (1-2), pP. 111-122, 1999. Cambridge sTUdiEs in adVaNceD MathemAtics, caMBriDge, 1992. Small devIatiOns in a spaCe of trAJectoriES. [*Theor. PRobab. APpl.*]{} [**19**]{}, Pp. 726-736, 1974. MOuveMEnT bRowNiEN et ESpAceS De BEsov. [*StocH. SToCh. Rep.*]{} [**43**]{} (3-4), Pp. 221-260, 1993. AsYMPTOtic BehAvioR of stAble seminorms NeaR the ORigIn. [*Ann. probaB.*]{} [**14**]{} (1), pp. 287-298, 1986. LOwEr taiLs of seLf–simIlAr stable processEs. [*BeRnoulli*]{} [**4**]{} (1), pp. 127-142, 1998. And [m. S. taqQu.]{} [*stablE Non-GauSsiAn rAndom prOcesses.*]{} cHapMaN & hALl, new York, 1994. A note on smalL bALL pRobabiliTy of a GAUsSiAN process WiTh sTatiONAry inCremENtS. [*J. TheoreT. ProbaB.*]{} [**6**]{} (3), Pp. 595-602, 1993. a GAussian CoRrelatIoN inEquAlity ANd itS appliCation to The exIStence of small bALl constant. [*PrePRiNT*]{}, 1999. loWEr taIls Of quadratic FuncTIonaLs of SYmMetRIc staBle prOcESsES. [*Prépublication de l’UNiVersitÉ PariS 6*]{}, 1999. Une méthode élÉmentaire pOUR L’évaluatIon dES pETites boules broWnienNes. [*C. R. Acad. SCI. Paris*]{} [**316**]{} (11), pp. 1217-1220, 1993. small Ball probAbilities FOR GaussiaN prOceSseS unDER nOn-uniform normS. [*j. theoReT. Probab.*]{}

[*Probab. Theory and Math . Statist. Proc . V IIIn tern atio nal Vilnius Co n fere nce (1998)*]{}. Vilniu s, VS P/ T EV,p p. 453- 468, 19 9 9. and [ W. Li nd e .] {} Ap pro ximatio n and entr opy n umbers of Vo l te rra operat ors with applic ati on toBr own i an mo tio n. [* Mem. A m er. Ma th. Soc.* ]{ } [**74 5 **]{},2 0 02 . a nd [W. Linde.]{}S ma l l deviations o f weig ht e df r act ion al process es anda veragen on - l i nea r approximatio n. Submitte d to [*Sto ch ast i c Proc esses a n d t heir Applic atio ns*]{}, 2 002. a nd [P.M . Robin son.]{ } A lte rnat i ve f orm so f f r ac tio n alBrownian m ot ion.[*J. S t a t. P lan n. I nfere nce*]{} [**80 **] {} ( 1 -2) , pp. 111- 122, 1 999. Cambr idgest udies in advanc ed m athematic s,Ca mbr id ge, 1 9 92. S mal l d eviatio ns in a spa ce o f t rajectories. [*The or . Pr obab. Ap pl.*]{ }  [ ** 1 9**]{},pp . 7 26-7 3 6 , 197 4. M ou vement b rownie n e tespaces d e Beso v. [* Sto ch. S t och. Rep.* ]{} [**4 3**]{ } (3-4), pp. 22 1 -260, 1993. A sy m p to t ic b eha vior of sta bles emin orms ne art he or igin. [ * An n . Probab.*]{} [**14 ** ]{} (1 ), pp . 287-298, 19 86. Lower t a ils of s elf– s im i lar stable pro cesse s. [*Berno u lli*]{}[**4* *]{} (1) , pp. 127 - 1 42, 1998 . and [M . S . Ta qqu.]{} [*Sta b l e no n- Gaussia n r andom p roc ess es. *]{ }Chapman & Hall, N ew Y or k, 19 94. A note on s mal lbal l pro b abilit y ofa Ga us si a n p rocessw it h stat io na ry i ncr em ents. [*J . Th eoret.Probab.*] {}[ **6* *] {} (3), p p. 595-602, 1 99 3. A Gaus si ancorrel a t ion ineq uality and its applicat i on to t heexist ence of small ba ll con sta n t. [*P reprin t*]{} ,199 9 . Low e r t ail sof quadrat i c fu nctio na ls o f symme tric stable proces s es. [*Prépublica tio n de l ’U niv e rs i téPa r is6 * ]{}, 1999. Une méthode é lé m en taire pour l’é va luation des pe tites boulesbrownienn es. [*C.R. Aca d . Sc i. Paris*] {} [**31 6**]{} (1 1 ), pp . 1 217-1 220 , 1993 . Sm all b all pr o bab iliti es for G aussia n pro ce sses und er non-uniform norms. [ *J. Th eoret . P robab.*]{ }

[*Probab._Theory and_Math. Statist. Proc. VII_International Vilnius_Conference_(1998)*]{}. Vilnius,_VSP/TEV,_pp. 453-468, 1999. and_[W. Linde.]{} Approximation_and entropy numbers of_Volterra operators with_application_to Brownian motion. [*Mem. Amer. Math. Soc.*]{} [**745**]{}, 2002. and [W. Linde.]{} Small deviations of_weighted_fractional processes_and_average_non-linear approximation. Submitted to [*Stochastic_Processes and their Applications*]{}, 2002. and_[P. M. Robinson.]{} Alternative_forms of fractional Brownian motion. [*J. Stat. Plann._Inference*]{}_[**80**]{} (1-2), pp._111-122, 1999. Cambridge studies in advanced mathematics, Cambridge, 1992. Small deviations_in a space of trajectories. [*Theor. Probab. Appl.*]{} [**19**]{},_pp. 726-736, 1974. Mouvement_brownien_et_espaces de Besov. [*Stoch._Stoch. Rep.*]{} [**43**]{} (3-4), pp. 221-260,_1993. Asymptotic behavior of stable seminorms near_the origin. [*Ann. Probab.*]{} [**14**]{} (1), pp._287-298, 1986. Lower tails of self–similar stable_processes. [*Bernoulli*]{} [**4**]{} (1), pp._127-142, 1998. and_[M. S. Taqqu.]{} [*Stable non-Gaussian random processes.*]{}_Chapman & Hall,_New York,_1994. A note on_small ball probability of a Gaussian_process with stationary_increments. [*J. Theoret. Probab.*]{} [**6**]{} (3),_pp._595-602, 1993. A Gaussian_correlation_inequality_and its_application to the_existence_of small_ball_constant. [*Preprint*]{}, 1999. Lower tails of quadratic_functionals_of symmetric stable processes. [*Prépublication de l’Université_Paris 6*]{}, 1999. Une méthode_élémentaire_pour l’évaluation des petites_boules browniennes. [*C. R. Acad._Sci. Paris*]{} [**316**]{} (11), pp. 1217-1220,_1993. Small ball_probabilities for_Gaussian processes under non-uniform norms. [*J. Theoret. Probab.*]{}

352, Université de Picardie Jules Verne, 33 rue Saint Leu, 80039 Amiens France, ([ Jean-Paul.Chehab@u-picardie.fr]{}). [^2]: Laboratoire de Réactivité et de Chimie des Solides (LRCS) Université de Picardie Jules Verne - CNRS / UMR 7314, 33, rue St. Leu, Amiens, France F-80039([ Alejandro.Franco@u-picardie.fr]{}) and R' eseau sur le Stockage Electrochimique de l’Energie (RS2E), FR CNRS, 3459, France [^3]: Laboratoire Amienois de Mathématiques Fondamentales et Appliquées (LAMFA), [UMR]{} 7352, Université de Picardie Jules Verne, 33 rue Saint Leu, 80039 Amiens France ([ Youcef.Mammeri@u-picardie.fr]{}). [**** ]{}\ Stanisław Saganowski^1,\*^, Piotr Bródka^1^, Michał Koziarski^2^, Przemysław Kazienko^1^\ **1** Department of Computational Intelligence, Faculty of Computer Science and Management, Wrocław University of Science and Technology, Wrocław, Poland\ **2** Department of Electronics, Faculty of Computer Science, Electronics and Telecommunications, AGH University of Science and Technology, Kraków, Poland\ \* stanislaw.saganowski@pwr.edu.pl Abstract {#abstract.unnumbered} ======== In the world, in which acceptance and the identification with social communities are highly desired, the ability to predict the evolution of groups over time appears to be a vital but very complex research problem. Therefore, we propose a new, adaptable, generic, and multistage method for Group Evolution Prediction (GEP) in complex networks, that facilitates reasoning about the future states of the recently discovered groups. The precise GEP modularity enabled us to carry out extensive and versatile empirical studies on many real-world complex / social networks to analyze the impact of numerous setups and parameters like time window type and size, group detection method, evolution chain length, prediction models, etc. Additionally, many

352, Université de Picardie Jules Verne, 33 rue Saint Leu, 80039 Amiens France, ([ Jean-Paul.Chehab@u-picardie.fr ] { }). [ ^2 ]: Laboratoire de Réactivité et de Chimie des Solides (LRCS) Université de Picardie Jules Verne - CNRS / UMR 7314, 33, rue St. Leu, Amiens, France F-80039 ([ Alejandro.Franco@u-picardie.fr ] { }) and R' eseau sur le Stockage Electrochimique de l’Energie (RS2E), FR CNRS, 3459, France [ ^3 ]: Laboratoire Amienois de Mathématiques Fondamentales et Appliquées (LAMFA), [ UMR ] { } 7352, Université de Picardie Jules Verne, 33 rue Saint Leu, 80039 Amiens France ([ Youcef.Mammeri@u-picardie.fr ] { }). [ * * * * ] { } \ Stanisław Saganowski^1,\*^, Piotr Bródka^1^, Michał Koziarski^2^, Przemysław Kazienko^1^\ * * 1 * * Department of Computational Intelligence, Faculty of Computer Science and Management, Wrocław University of Science and Technology, Wrocław, Poland\ * * 2 * * Department of Electronics, Faculty of Computer Science, Electronics and Telecommunications, AGH University of Science and Technology, Kraków, Poland\ \ * stanislaw.saganowski@pwr.edu.pl Abstract { # abstract.unnumbered } = = = = = = = = In the world, in which acceptance and the identification with social communities are highly desire, the ability to bode the evolution of groups over meter appear to be a vital but very complex research trouble. Therefore, we propose a new, adaptable, generic, and multistage method acting for Group Evolution Prediction (GEP) in complex net, that facilitates reason about the future states of the recently discovered group. The precise GEP modularity enabled us to carry out extensive and versatile empirical study on many real - world complex / social networks to analyze the shock of numerous setups and parameters like time window character and size, group detection method acting, evolution chain length, prediction models, etc. Additionally, many

352, Unlversité de Picardie Juler Verne, 33 rue Sanbt Leu, 80039 Amiehs Francd, ([ Jean-Paul.Chehab@u-picardie.fr]{}). [^2]: Lqboraujire de Réactivité eg de Chimpe des Sooidew (LRCS) Univxdsité de Picarsle Junxs Verne - CNRS / UMR 7314, 33, rue St. Leu, Amiens, Ffauce F-80039([ Alejandro.Franco@u-picardie.fr]{}) and R' eseai dur le Stockagg Elebtwochjmique de l’Energie (RS2E), FR CNRS, 3459, Frznce [^3]: Lauoratoire Amienpis de Mathématiques Fondamfntapes et Appliquées (LWMFA), [UMR]{} 7352, Ubivewwité de Picaraie Jules Nzrne, 33 rue Szint Leu, 80039 Amiens France ([ Youcef.Oammexi@u-picardie.dr]{}). [**** ]{}\ Shdnisław Sagaiowski^1,\*^, Piotr Bródka^1^, Michał Kmziarsko^2^, Przemysław Kadienkm^1^\ **1** Dwpartment of Computatmonal Intelligence, Fwculty of Ckmputer Science abd Manacemett, Wficład Uhiteraity ov Srience and Fechnology, Qrocław, Poland\ **2** Deparumegn of Electronjcs, Fasujty of Computer Science, Electronics and Temecommunications, AGH Unuversity of Science ajd Technojogy, Kraków, Poland\ \* stanislaw.saganowski@pwr.edu.pl Abstsact {#ebrtrcgb.unnjnbfred} ======== In the world, in which acceptance and the ysemtpfication with sogial communities ate hodhly desired, jhe abilitg to predict the egolutiog of troups ovtr tike appears to be a vital bur very complvx rwsearch problem. Thzrefore, we pxopose a nee, adaptable, generic, and mulfistage metjod for Gduup Evolution Prddibtiot (GEP) in complex networks, ehat facioitaces reasuninb aboue the futuge states of the recentlj diseoverad groups. Hhe precise GEP modularity enabled us to carri ogt vxtensive and yersatile empirycal studies ou many rzal-worud complex / socian networks eo analyze tha impact of nnmerous sqtupw ane paramdgers like time window tjpt and size, group detection mcthod, svolution chain lebgth, prediction moaelf, vtc. Aqgitionally, mdny

352, Université de Picardie Jules Verne, 33 Leu, Amiens France, Jean-Paul.Chehab@u-picardie.fr]{}). [^2]: Laboratoire des (LRCS) Université de Jules Verne - / UMR 7314, 33, rue St. Amiens, France F-80039([ Alejandro.Franco@u-picardie.fr]{}) and R' eseau sur le Stockage Electrochimique de l’Energie FR CNRS, 3459, France [^3]: Laboratoire Amienois de Mathématiques Fondamentales et Appliquées (LAMFA), 7352, de Jules 33 rue Saint Leu, 80039 Amiens France ([ Youcef.Mammeri@u-picardie.fr]{}). [**** ]{}\ Stanisław Saganowski^1,\*^, Piotr Bródka^1^, Michał Przemysław Kazienko^1^\ **1** Department of Computational Intelligence, Faculty Computer Science and Management, University of Science and Technology, Poland\ Department of Faculty Computer Electronics and Telecommunications, University of Science and Technology, Kraków, Poland\ \* stanislaw.saganowski@pwr.edu.pl Abstract {#abstract.unnumbered} ======== In the world, in which and the social communities highly the to predict the groups over time appears to be very complex research problem. Therefore, we propose a adaptable, generic, multistage method for Group Evolution Prediction in complex networks, that facilitates reasoning about the states of the recently discovered groups. The precise GEP modularity enabled us to carry out versatile empirical studies on real-world complex / networks analyze impact numerous setups parameters like time window type and size, group detection method, evolution length, prediction models, etc. Additionally, many

352, Université de Picardie Jules verne, 33 rue SaInt LeU, 80039 AmIenS FRancE, ([ JeaN-Paul.Chehab@u-piCArdiE.fr]{}). [^2]: Laboratoire de RéactiVité eT dE chimIE dEs SolIdes (LRCs) unIVErsItÉ dE PiCaRDiE JuleS VeRne - CNRS / uMR 7314, 33, rue St. LeU, AmIeNs, France F-80039([ AleJAnDro.Franco@u-PicArdie.fr]{}) and R' eSeaU sur le stOckAGe EleCtrOchimIque de L’energiE (RS2E), FR CNRs, 3459, FRAnce [^3]: LaBOratoirE aMiEnoiS de Mathématiques FONdAMentales et ApplIquées (lAmfA), [umr]{} 7352, UnIveRsité de PicArDie JuLEs Verne, 33 RUe sAINt LEU, 80039 Amiens France ([ youcef.MammeRI@u-pIcardiE.fR]{}). [**** ]{}\ StANisław saganOwSKi^1,\*^, PIotr Bródka^1^, MIchaŁ KoziarskI^2^, PrzemYSław KazIEnko^1^\ **1** DepArtmenT of comPutaTIoNaL InTeLLigENcE, FaCUltY of CompuTeR SCiencE and mANAGemeNt, WRocłAw UniVersity of ScieNce And TEChnOlogy, wrocłAw, PoLaNd\ **2** DepArtmenT of ElEcTronics, Faculty oF ComPuter ScieNce, elEctRoNics aND TelecOmmUniCations, aGH UnivERsiTy OF sCiEnce and Technology, KRaKÓW, POland\ \* staNislaw.SAgAnOWski@pwr.eDu.Pl ABstrACT {#abstRact.UNnUmbered} ======== IN the woRLd, In Which acCePtance AnD thE idEntifICatiOn with Social coMmuniTIes are highly deSIred, the abilitY To PREdICt thE evOlution of grOups OVer tIme aPPeArs TO be a vItal bUt VErY Complex research probLeM. ThereFore, wE propose a new, aDaptable, geNERIc, and mulTistAGe MEthod for Group EVolutIon PredictIOn (GEP) in cOmpleX networkS, that faciLITates reaSonIng AboUt tHE FuTure states of tHE ReceNtLy discoVerEd groupS. ThE prEciSe GeP ModularitY enabled Us To CaRrY ouT exteNSive and vErSatIlE emPiricAL studiEs on mAny rEaL-wORld Complex / SOcIAL netWoRkS to aNalYzE the iMpacT Of nUmerous Setups and ParAMeteRs LiKe time wIndow type and sIzE, group deteCtIon Method, EVOlution cHain length, prediction modELs, etc. AdDitIonalLy, maNy

352, Université de Picardi e Jules Ve rne,33rue S aint Leu , 80039 Amiens Fran ce, ([ Jean-Paul.Cheha b@u-p ic a rdie . fr ]{}). [^2]: La b o rat oi re de R é ac tivit é e t de Ch imie des S oli de s (LRCS) Uni v er sité de Pi car die Jules Ve rne - CNR S/ U M R 731 4,33, r ue St. Leu, A miens, Fr an c e F-80 0 39([ Al e j an dro. Franco@u-picardie . fr ] {}) and R' ese au sur l e S t o cka geElectrochi mi que d e l’Ener g ie ( R S2E ) , FR CNRS, 34 59, France[^3 ]: Lab or ato i re Ami enois d e Ma thématiques Fon damentale s et A p pliquée s (LAMFA ), [UM R]{ } 7 352, Un iv ers it é de Pi car d ieJules Ve rn e, 33 r ue S a i n t Leu , 8 0039 Amie ns France ([You cef. M amm eri@u -pica rdie .f r]{}) . [** ** ]{ }\ Stanisław Saga nows ki^1,\*^, Pi ot r B ró dka^1 ^ , Mich ałKoz iarski^ 2^, Prz e mys ła w K az ienko^1^\ **1** De pa r t me nt of Co mputat i on al Intellig en ce, Fac u l ty of Com p ut er Scien ce and Ma na gement, W rocław U niv ers ity o f Sci ence a nd Techn ology , Wrocław, Pola n d\ **2** Depa r tm e n to f El ect ronics, Fac ulty of C ompu t er Sc i ence, Elec tr o ni c s and Telecommunica ti ons, A GH Un iversity of S cience and T e chnology , Kr a kó w , Poland\ \*stani slaw.sagan o wski@pwr .edu. pl Abst ract {#ab s t ract.unn umb ere d}=== = = == = In the wor l d , in w hich ac cep tance a ndthe id ent if ication w ith soci al c om mu nit ies a r e highly d esi re d,the a b ilityto pr edic tth e ev olution of g roup sov er t ime a ppear s to bea vital but very co m plex r es earch p roblem. There fo re, we pro po sea new, a daptable , generic, and multista g e metho d f or Gr oupEvolution Pr edicti on( GEP) i n comp lex n et wor k s , tha t fa cil it ates reaso n i ngabout t he f uture s tates of the recen t lydiscovered gr oup s. T h e p rec i se GEP m o dul a r ity enabled usto carry o ut ex tensive an d ve rs atile e mpirica l stu d ies onmany real -world co mp lex/ soc ial networ ks to an alyze the impac t o f num ero us set up s a nd pa ramete r s l ike t ime wi nd ow typ e and s ize, gro up detection method, ev olutio n cha inlength, p red i cti on models , et c. Additio nal ly, many

352, Université_de Picardie_Jules Verne, 33 rue_Saint Leu,_80039_Amiens France,_([_Jean-Paul.Chehab@u-picardie.fr]{}). [^2]: Laboratoire de_Réactivité et de_Chimie des Solides (LRCS)_Université de Picardie_Jules_Verne - CNRS / UMR 7314, 33, rue St. Leu, Amiens, France F-80039([ Alejandro.Franco@u-picardie.fr]{})_and_R' eseau_sur_le_Stockage Electrochimique de l’Energie (RS2E),_FR CNRS, 3459, France [^3]: Laboratoire_Amienois de_Mathématiques Fondamentales et Appliquées (LAMFA), [UMR]{} 7352, Université_de_Picardie Jules Verne,_33 rue Saint Leu, 80039 Amiens France ([ Youcef.Mammeri@u-picardie.fr]{}). _[**** ]{}\ Stanisław Saganowski^1,\*^, Piotr Bródka^1^, Michał_Koziarski^2^, Przemysław Kazienko^1^\ **1**_Department_of_Computational Intelligence, Faculty of_Computer Science and Management, Wrocław University_of Science and Technology, Wrocław, Poland\ **2**_Department of Electronics, Faculty of Computer Science,_Electronics and Telecommunications, AGH University of_Science and Technology, Kraków, Poland\ \*_stanislaw.saganowski@pwr.edu.pl Abstract {#abstract.unnumbered} ======== In_the world, in which acceptance_and the identification_with social_communities are highly_desired, the ability to predict the_evolution of groups_over time appears to be a_vital_but very complex_research_problem._Therefore, we_propose a new,_adaptable,_generic, and_multistage_method for Group Evolution Prediction (GEP)_in_complex networks, that facilitates reasoning about the_future states of the_recently_discovered groups. The precise_GEP modularity enabled us to_carry out extensive and versatile empirical_studies on_many real-world_complex / social networks to analyze the impact of numerous setups_and parameters like time window type_and size, group detection_method, evolution_chain_length, prediction models,_etc._Additionally, many

thus is to incorporate these corrections while staying compatible with the four-dimensionality of experimental data. In other words, one needs the dimensionally reduced description of the object, i.e., the one in terms of the most appropriate solution $Sol(M_4)$ arising from a corresponding effective 4D theory $M_4$. The question is how to derive such EFMT $M_4$, especially assuming the situation like the one we have recently, i.e., when $M$ is not explicitly known? Our solution to the problem would be to use the known fundamental solutions of high-D theories as a guiding thread. Indeed, real phenomena are described by $Sol(M)$’s thus if one wishes to preserve (partially, at least) such description in the dimensionally reduced theory one must make sure that the latter does not disallow $Sol(M)$-like solutions in principle. Considering aforesaid, we formulate the following selection criterion: “Suppose we have a higher-dimensional theory, call it $M$. This theory has a certain physically relevant solution, $Sol(M)$ which is unique to $M$. This solution has certain distinctive property, $Prop(Sol(M))$, which is preserved under reduction of dimensions. On the other hand, we have 4D theory $M_4$ whose nontrivial solutions we denote by $Sol(M_4)$. Then the necessary condition for $M_4$ to be a lower-dimensional limit of $M$ is that at least one of $Sol(M_4)$ must have $Prop(Sol(M))$.” This criterion can be regarded as the correspondence principle for solutions, henceforth we call it the *Solution Correspondence Principle* ([SCP]{}). Now, to determine what solution is physically relevant and which property of it is suitable for [SCP]{}, we will make use of the following two facts from the string/M-theory. First, it is well-known that the branes are known solutions of M theory (one may recall M2 and M5 ones) hence they should be perfectly suitable for a concrete realization of [SCP]{}. Moreover, it is known that branes are inevitable for proper describing of black holes (microstates, entropy, etc.) [@Strominger:1996sh], therefore, their absence in a theory would cause serious difficulties with consistent explaining such issues. In turn, absence of black holes would lead to the loss of protection of a theory from ubiquitous appearances

thus is to incorporate these corrections while staying compatible with the four - dimensionality of experimental datum. In early discussion, one needs the dimensionally reduce description of the aim, i.e., the one in terms of the most appropriate solution $ Sol(M_4)$ arising from a comparable effective 4D hypothesis $ M_4$. The question is how to derive such EFMT $ M_4 $, especially assuming the situation like the one we get recently, i.e., when $ M$ is not explicitly known? Our solution to the problem would be to use the known fundamental solutions of eminent - D theories as a guiding screw thread. Indeed, real phenomena are described by $ Sol(M)$ ’s therefore if one wishes to keep (partially, at least) such description in the dimensionally reduced theory one must make certain that the latter does not disallow $ Sol(M)$-like solutions in principle. Considering aforesaid, we formulate the follow selection criterion: “ Suppose we have a higher - dimensional theory, call it $ M$. This theory has a certain physically relevant solution, $ Sol(M)$ which is alone to $ M$. This solution has certain classifiable place, $ Prop(Sol(M))$, which is preserved under reduction of dimensions. On the other bridge player, we have 4D theory $ M_4 $ whose nontrivial solutions we denote by $ Sol(M_4)$. Then the necessary circumstance for $ M_4 $ to be a lower - dimensional limit of $ M$ is that at least one of $ Sol(M_4)$ must have $ Prop(Sol(M))$. ” This criterion can be regarded as the correspondence principle for solutions, henceforth we call it the * Solution Correspondence Principle * ([ SCP ] { }). Now, to determine what solution is physically relevant and which property of it is suitable for [ SCP ] { }, we will hold use of the watch two facts from the string / M - theory. First, it is well - know that the branes are known solutions of M theory (one may echo M2 and M5 ones) hence they should be perfectly suitable for a concrete realization of [ SCP ] { }. furthermore, it is known that branes are inevitable for proper describing of black holes (microstates, entropy, etc .) [ @Strominger:1996sh ], therefore, their absence in a hypothesis would cause serious difficulties with consistent explain such issues. In twist, absence of black holes would run to the loss of protection of a theory from ubiquitous appearances

thks is to incorporate there corrections cyile svaying dompatibue with the four-dimensionalivy od exptgimental data. In other words, onv needs tye dmmensionally rednded desgxiptikk of chx object, i.e., the one in tesms of the mosd xp'ropriate solution $Sol(M_4)$ arising from a corrrsoonding effectyve 4C thekgy $M_4$. The question is how to derjve suci EFMT $M_4$, especislly assuming the situatioj line the one we have recently, i.g., whqb $M$ is not ebplicitly known? Our somution to the problem would be go usz the known fyndwkental soluvions jf high-D theories as a guidong thread. Indced, rxal phenomena are describxd by $Sol(M)$’s thus if jne wishev co preserve (partially, ar leavt) sgch awscfipuioi ih the fimxnsionally deduced theiry one must make sirq that the latfer doqs not disallow $Sol(M)$-like solutions in pritcille. Considering aforesaie, we formulate the foplowing sqlection criterion: “Suppose we have a higher-dimenshonal gheirn, cauo lt $M$. This theory has a certain physically relqbamt solution, $Sol(M)$ which is umieur to $M$. This souution haa certain distinctlve prokerty, $Prop(Sol(M))$, rhicn is preserved under reductuon of dimenfuons. On the other kand, we have 4D tneory $M_4$ whose nontrivial solbtions we denote hy $Sol(M_4)$. Tgdn the necessary cokdidion for $M_4$ to be a lower-diiensional limnt of $M$ ks tnat at least one of $Sol(M_4)$ must have $Prop(Sll(M))$.” Tkis csiterion cwn be regarded as the corresponvxnce principlg fmr volutionf, hengeforth we call it the *Solutipn Corrzsponddnce Princpple* ([SCP]{}). Iow, to deteriine what solgjion is physirally relqvanr ane which oroperty of it is suitayje fir [SCP]{}, we will makc use kf the followiny uwi facts from tne rtrynh/M-vheorr. Xirst, it is fell-ynodm thag the branew are known solutions of K thsory (one may recalk I2 and M5 ines) hense they shoulc be perfectly suiuable hor a roncreye tealization of [SCP]{}. Moreover, it js known hhab branes are ynevltabje for pro'er describing of black holes (microstatew, entropy, etc.) [@Stroninger:1996sh], therefore, jhelr absence ii a thqory woulg cause serious diffuculties with conxistent explaining sudh issges. Ij turn, absence of black holes would lead to the loss of protection of a tyeory hrjm ubiquitohs alpeardnees

thus is to incorporate these corrections while with four-dimensionality of data. In other reduced of the object, the one in of the most appropriate solution $Sol(M_4)$ from a corresponding effective 4D theory $M_4$. The question is how to derive EFMT $M_4$, especially assuming the situation like the one we have recently, i.e., $M$ not known? solution to the problem would be to use the known fundamental solutions of high-D theories as guiding thread. Indeed, real phenomena are described by thus if one wishes preserve (partially, at least) such in dimensionally reduced one make that the latter not disallow $Sol(M)$-like solutions in principle. Considering aforesaid, we formulate the following selection criterion: “Suppose we have higher-dimensional theory, $M$. This has certain relevant solution, $Sol(M)$ unique to $M$. This solution has $Prop(Sol(M))$, which is preserved under reduction of dimensions. the other we have 4D theory $M_4$ whose solutions we denote by $Sol(M_4)$. Then the necessary for $M_4$ to be a lower-dimensional limit of $M$ is that at least one of have $Prop(Sol(M))$.” This criterion be regarded as correspondence for henceforth call it *Solution Correspondence Principle* ([SCP]{}). Now, to determine what solution is physically and which property of it is suitable for [SCP]{}, we use the following two from the string/M-theory. First, is that the branes are of theory M2 M5 hence they should be suitable for a concrete realization [SCP]{}. Moreover, it is for proper describing of black holes (microstates, entropy, [@Strominger:1996sh], therefore, their absence in a theory cause serious difficulties with consistent explaining such issues. In turn, absence of holes would the loss of protection of a theory from appearances

thus is to incorporate these cOrrections While StaYinG cOmpaTiblE with the four-diMEnsiOnality of experimental dAta. In OtHEr woRDs, One neEds the dIMeNSIonAlLy RedUcED dEscriPtiOn of the Object, i.e., thE onE iN terms of the mOSt AppropriatE soLution $Sol(M_4)$ arIsiNg from A cOrrESpondIng EffecTive 4D tHEory $M_4$. THe questioN iS How to dERive sucH efMt $M_4$, esPecially assuming tHE sITuation like the One we hAvE ReCENtlY, i.e., When $M$ is not ExPliciTLy known? oUr SOLUtiON to the problem Would be to usE The Known fUnDamENtal soLutioNs OF hiGh-D theories As a gUiding thrEad. IndEEd, real pHEnomena Are desCriBed By $SoL(m)$’s ThUs iF oNE wiSHeS to PResErve (partIaLlY, at leAst) sUCH DEscrIptIon iN the dImensionally rEduCed tHEorY one mUst maKe suRe That tHe lattEr doeS nOt disallow $Sol(M)$-lIke sOlutions iN prInCipLe. consiDEring aForEsaId, we forMulate tHE foLlOWINg Selection criterion: “suPPOsE we have a Higher-DImEnSIonal theOrY, caLl it $m$. tHis thEory HAs A certain PhysicALlY rElevant SoLution, $soL(M)$ wHicH is unIQue tO $M$. This Solution Has ceRTain distinctivE Property, $Prop(SOL(M))$, WHIcH Is prEseRved under reDuctIOn of DimeNSiOns. oN the oTher hAnD, We HAve 4D theory $M_4$ whose nonTrIvial sOlutiOns we denote by $sol(M_4)$. Then thE NECessary cOndiTIoN For $M_4$ to be a lower-DimenSional limiT Of $M$ is thaT at leAst one of $sol(M_4)$ must hAVE $Prop(Sol(m))$.” ThIs cRitEriON CaN be regarded as THE corReSpondenCe pRinciplE foR soLutIonS, hEnceforth We call it ThE *SOlUtIon correSPondence prIncIpLe* ([ScP]{}). Now, TO deterMine wHat sOlUtIOn iS physicALlY RElevAnT aNd whIch PrOpertY of iT Is sUitable For [SCP]{}, we wIll MAke uSe Of The follOwing two facts FrOm the strinG/M-TheOry. FirST, It is well-Known that the branes are knOWn solutIonS of M tHeorY (one may reCalL M2 and M5 OneS) Hence tHey shoUld be PeRfeCTLy suiTABlE foR a Concrete reALIzaTion oF [ScP]{}. MoReover, iT is known that branes ARe iNevitable for pRopEr deSCRiBinG Of BLacK hOLes (MICrostates, entropY, etc.) [@StromiNgER:1996sH], therefore, THeiR aBsence iN a theorY woulD Cause seRious diffIculties wItH conSISteNt explainiNg such isSues. In turN, AbsenCE oF blacK hoLes wouLd LeaD to thE loss oF ProTectiOn of a tHeOry froM ubiqUiTous appeArances

thus is to incorporate th ese correc tions wh ile s tayi ng c ompatible with thefour-dimensionality of expe ri m enta l d ata. In oth e rw o rds ,on e n ee d sthe d ime nsional ly reduced de sc ription of t h eobject, i. e., the one inter ms ofth e m o st ap pro priat e solu t ion $S ol(M_4)$ar i sing f r om a co r r es pond ing effective 4Dt he o ry $M_4$. Thequesti on is h owtoderive suc hEFMT$ M_4$, e s pe c i a lly assuming thesituation l i kethe on eweh ave re centl y, i.e ., when $M$ isnot expli citlyk nown? O u r solut ion to th e p robl e mwo uld b e to us e t h e k nown fun da me ntalsolu t i o n s of hi gh-D theo ries as a gui din g th r ead . Ind eed,real p henom ena ar e des cr ibed by $Sol(M) $’sthus if o newi she sto pr e serve(pa rti ally, a t least ) su ch d e sc ription in the dim en s i on ally red uced t h eo ry one must m ake sur e thatthel at ter does not d i sa ll ow $Sol (M )$-lik esol uti ons i n pri nciple . Consi derin g aforesaid, we formulate the fo l l ow i ng s ele ction crite rion : “Su ppos e w e h a ve ahighe r- d im e nsional theory, cal lit $M$ . Thi s theory hasa certainp h y sicallyrele v an t solution, $So l(M)$ which isu nique to $M$. This so lution ha s certaindis tin cti vep r op erty, $Prop(S o l (M)) $, whichispreserv edund erred uc tion of d imension s. O nth e o therh and, weha ve4D th eory$ M_4$ w hosenont ri vi a l s olution s w e deno te b y $S ol( M_ 4)$.Then the necess ary condi tio n for $ M_ 4$ to b e a lower-dim en sional lim it of $M$ i s that atleast one of $Sol(M_4)$ must ha ve$Prop (Sol (M))$.” T his crite rio n can b e rega rdedas th e corre s p on den ce principle f orsolut io ns,hencefo rth we call it the *So lution Corres pon denc e Pr inc i pl e * ( [S C P]{ } ) . Now, to deter mine whatso l ut ion is phy s ica ll y relev ant and whic h proper ty of itis suitab le for [ SCP ]{}, we wi ll makeuse of th e foll o wi ng tw o f acts f ro m t he st ring/M - the ory. First ,it iswell- kn own that the branes are known s olutio ns of Mtheory (o nem ayrecall M2 and M5 ones)hen cetheysho u ld be per f ec tly suita blef or a conc r et e r e a li zation of [ S C P ]{} . Mor eov e r, itis k nown that branesa re inevitableforp r ope r d e scri bi ng of black ho les ( m i crostate s, entropy, e tc.) [@S tr o minge r:1996 sh], t herefor e , t h eir ab senc e i n a theor y w ou l d cause s er i ous di ffic ul ties w ith co n sist e n t explaining suc h iss u e s. In tur n, ab se nce ofb lack holes wou ld lead tothe lo ss o f pro tection o f a th eor yfrom ubiqu i tous appe aranc es

thus_is to_incorporate these corrections while_staying compatible_with_the four-dimensionality_of_experimental data. In other_words, one needs_the dimensionally reduced description_of the object,_i.e.,_the one in terms of the most appropriate solution $Sol(M_4)$ arising from a corresponding_effective_4D theory_$M_4$._The_question is how to derive_such EFMT $M_4$, especially assuming_the situation_like the one we have recently, i.e., when_$M$_is not explicitly_known? Our solution to the problem would be to_use the known fundamental solutions of_high-D theories as_a_guiding_thread. Indeed, real phenomena_are described by $Sol(M)$’s thus if_one wishes to preserve (partially, at_least) such description in the dimensionally reduced_theory one must make sure that_the latter does not disallow_$Sol(M)$-like solutions_in principle. Considering aforesaid, we formulate_the following selection_criterion: “Suppose_we have a_higher-dimensional theory, call it $M$. This_theory has a_certain physically relevant solution, $Sol(M)$ which_is_unique to $M$._This_solution_has certain_distinctive property, $Prop(Sol(M))$,_which_is preserved_under_reduction of dimensions. On the other_hand,_we have 4D theory $M_4$ whose nontrivial_solutions we denote by_$Sol(M_4)$._Then the necessary condition_for $M_4$ to be a_lower-dimensional limit of $M$ is that_at least_one of_$Sol(M_4)$ must have $Prop(Sol(M))$.” This criterion can be regarded as the_correspondence principle for solutions, henceforth we_call it the *Solution_Correspondence Principle*_([SCP]{})._Now, to determine_what_solution is_physically relevant and which property of it_is suitable_for [SCP]{}, we will make use_of the following two_facts_from the string/M-theory. First, it is well-known_that the branes are known solutions_of M theory (one may_recall_M2_and M5 ones) hence they_should be perfectly suitable for a_concrete realization of_[SCP]{}. Moreover, it is known that branes_are_inevitable for proper describing of black_holes_(microstates, entropy, etc.) [@Strominger:1996sh], therefore, their_absence_in_a theory would cause serious_difficulties with consistent explaining such issues._In turn, absence of black holes would lead to_the loss of_protection of a theory from_ubiquitous_appearances

}$. They proved that their algorithm halts unless there exists another non-conjugate maximal order $\mathcal{O}'$ for which $$\label{eqn:optdom} a_{\mathcal{O}'^T}(n)\geq a_{\mathcal{O}^T}(n)$$ for every $n\in{\mathbb{N}}_0$, where $a_{\mathcal{O}^T}(n)$ is defined in. Following [@C-G], we thus say that $\mathcal{O}^{\prime T}$ optimally dominates $\mathcal{O}^{T}$ if holds for all $n\in{\mathbb{N}}_0$. Chevyrev and Galbraith then conjectured in [@C-G Conjecture 1] that no maximal order may optimally dominate another. \[conj:main\] Let $\mathcal{O}$ and $\mathcal{O}^{\prime}$ be maximal orders of $B_{p}$. If $\mathcal{O}^{\prime T}$ optimally dominates $\mathcal{O}^{T}$, then $\mathcal{O}$ and $\mathcal{O}^{\prime}$ are of the same type. 1. Conjecture \[conj:main\] is equivalent to Conjecture \[conj:intro\] because all isomorphisms of orders come from conjugation. 2. Paralleling the definition of type for maximal orders, we say that $\mathcal{O}^{\prime T}$ and $\mathcal{O}^{T}$ have the same type if there is a non-zero element $c\in B_{p}$ such that $c\mathcal{O}^{T}c^{-1}=\mathcal{O}^{\prime T}$, and we write $\mathcal{O}^{T}\sim\mathcal{O}^{\prime T}$. By Lemma 4 in [@C-G], we know that $\mathcal{O}^{T}\sim\mathcal{O}^{\prime T}$ if and only if $\mathcal{O}\sim\mathcal{O}^{\prime}$. 3. There is a second conjecture of Chevyrev and Galbraith about the occurrence of the smallest $n_0$ for which both $a_{\mathcal{O^{\prime}}^{T}}(n_1)\geq a_{\mathcal{O}^{T}}(n_1)$ and $a_{\mathcal{O^{\prime}}^{T}}(n_2)< a_{\mathcal{O}^{T}}(n_2)$ occur for some $n_1,n_2<n_

} $. They proved that their algorithm halts unless there exists another non - conjugate maximal order $ \mathcal{O}'$ for which $ $ \label{eqn: optdom } a_{\mathcal{O}'^T}(n)\geq a_{\mathcal{O}^T}(n)$$ for every $ n\in{\mathbb{N}}_0 $, where $ a_{\mathcal{O}^T}(n)$ is define in. take after [ @C - G ], we thus say that $ \mathcal{O}^{\prime T}$ optimally dominate $ \mathcal{O}^{T}$ if holds for all $ n\in{\mathbb{N}}_0$. Chevyrev and Galbraith then conjectured in [ @C - thousand Conjecture 1 ] that no maximal club may optimally dominate another. \[conj: main\ ] permit $ \mathcal{O}$ and $ \mathcal{O}^{\prime}$ be maximal orders of $ B_{p}$. If $ \mathcal{O}^{\prime T}$ optimally dominates $ \mathcal{O}^{T}$, then $ \mathcal{O}$ and $ \mathcal{O}^{\prime}$ are of the like type. 1. Conjecture \[conj: main\ ] is equivalent to Conjecture \[conj: intro\ ] because all isomorphisms of orders come from union. 2. Paralleling the definition of type for maximal orders, we say that $ \mathcal{O}^{\prime T}$ and $ \mathcal{O}^{T}$ have the like type if there is a non - zero element $ c\in B_{p}$ such that $ c\mathcal{O}^{T}c^{-1}=\mathcal{O}^{\prime T}$, and we compose $ \mathcal{O}^{T}\sim\mathcal{O}^{\prime T}$. By Lemma 4 in [ @C - G ], we know that $ \mathcal{O}^{T}\sim\mathcal{O}^{\prime T}$ if and merely if $ \mathcal{O}\sim\mathcal{O}^{\prime}$. 3. There is a second speculation of Chevyrev and Galbraith about the occurrence of the smallest $ n_0 $ for which both $ a_{\mathcal{O^{\prime}}^{T}}(n_1)\geq a_{\mathcal{O}^{T}}(n_1)$ and $ a_{\mathcal{O^{\prime}}^{T}}(n_2) < a_{\mathcal{O}^{T}}(n_2)$ occur for some $ n_1,n_2 < n _

}$. Thfy proved that their algurithm halts unless thxre exiats anotfer non-conjugate maximal ordxr $\mqthcao{O}'$ for which $$\label{eqn:ootdom} a_{\matjcal{O}'^T}(n)\gwq a_{\nqthcal{O}^T}(n)$$ hkr evern $n\in{\jwthby{N}}_0$, where $a_{\mathcak{O}^T}(n)$ is dexined in. Follofivg [@C-G], we thus say that $\mathcal{O}^{\prime E}$ optimaklj dominates $\matrcal{P}^{E}$ if holds for all $n\in{\mathbb{N}}_0$. Chevyrev and Ganbraith then vonjectured in [@C-G Conjectuge 1] hhat no maximal orfer may optumaljt dominate avother. \[conj:main\] Let $\matgcal{O}$ and $\mathcal{O}^{\prime}$ be maxioal oxders of $B_{p}$. Id $\mwjhcal{O}^{\prime V}$ optiially dominabvs $\mathwal{O}^{T}$, tnen $\mathcal{O}$ akd $\mavhcao{O}^{\prime}$ are of the sake type. 1. Conjecturg \[conj:main\] ia equivalent to Cinhectute \[cotj:ingeo\] ceczuxe all idomkrphisms or orders cone from conjugation. 2. [qralleling ths defigieion of type for maximal orders, we say uhat $\jathcal{O}^{\prime T}$ and $\matycal{O}^{T}$ have the same jype if thqre is a non-zero element $c\in B_{p}$ such that $c\mathcan{O}^{T}c^{-1}=\metfcao{O}^{\irimd T}$, and we write $\mathcal{O}^{T}\sim\mathcal{O}^{\prime T}$. By Jsmka 4 in [@C-G], we knoc that $\mathcal{O}^{T}\xil\msjhcal{O}^{\prime T}$ kf and onmy if $\mathcal{O}\sim\mwthcal{O}^{\krime}$. 3. There is a srcond conjecture of Chevyrec and Galbrapth qbout the occurrenee of the smcllest $n_0$ fot which both $a_{\mathcal{O^{\pxime}}^{T}}(n_1)\feq a_{\mathcap{O}^{T}}(n_1)$ and $z_{\oathcal{O^{\prime}}^{T}}(n_2)< x_{\manhcan{O}^{T}}(n_2)$ occur for some $n_1,n_2<n_

}$. They proved that their algorithm halts exists non-conjugate maximal $\mathcal{O}'$ for which $n\in{\mathbb{N}}_0$, $a_{\mathcal{O}^T}(n)$ is defined Following [@C-G], we say that $\mathcal{O}^{\prime T}$ optimally dominates if holds for all $n\in{\mathbb{N}}_0$. Chevyrev and Galbraith then conjectured in [@C-G Conjecture that no maximal order may optimally dominate another. \[conj:main\] Let $\mathcal{O}$ and $\mathcal{O}^{\prime}$ maximal of If T}$ optimally dominates $\mathcal{O}^{T}$, then $\mathcal{O}$ and $\mathcal{O}^{\prime}$ are of the same type. 1. Conjecture \[conj:main\] equivalent to Conjecture \[conj:intro\] because all isomorphisms of come from conjugation. 2. the definition of type for orders, say that T}$ $\mathcal{O}^{T}$ the same type there is a non-zero element $c\in B_{p}$ such that $c\mathcal{O}^{T}c^{-1}=\mathcal{O}^{\prime T}$, and we write $\mathcal{O}^{T}\sim\mathcal{O}^{\prime T}$. By 4 in know that T}$ and if $\mathcal{O}\sim\mathcal{O}^{\prime}$. 3. a second conjecture of Chevyrev and occurrence of the smallest $n_0$ for which both a_{\mathcal{O}^{T}}(n_1)$ and a_{\mathcal{O}^{T}}(n_2)$ occur for some $n_1,n_2<n_

}$. They proved that their algoriThm halts unLess tHerE exIsTs anOtheR non-conjugate mAXimaL order $\mathcal{O}'$ for which $$\Label{EqN:OptdOM} a_{\MathcAl{O}'^T}(n)\geQ A_{\mATHcaL{O}^t}(n)$$ For EvERy $N\in{\maThbB{N}}_0$, where $A_{\mathcal{O}^T}(N)$ is DeFined in. FolloWInG [@C-G], we thus sAy tHat $\mathcal{O}^{\pRimE T}$ optiMaLly DOminaTes $\MathcAl{O}^{T}$ if HOlds foR all $n\in{\maThBB{N}}_0$. ChevYRev and GALBrAith Then conjectured in [@c-g CONjecture 1] that no MaximaL oRDeR MAy oPtiMally dominAtE anotHEr. \[conj:mAIn\] lET $\MatHCal{O}$ and $\mathcaL{O}^{\prime}$ be maXImaL orderS oF $B_{p}$. iF $\mathcAl{O}^{\prImE t}$ opTimally domiNateS $\mathcal{O}^{t}$, then $\mAThcal{O}$ aND $\mathcaL{O}^{\primE}$ arE of The sAMe TyPe. 1. COnJEctURe \[ConJ:MaiN\] is equivAlEnT to CoNjecTURE \[Conj:IntRo\] beCause All isomorphisMs oF ordERs cOme frOm conJugaTiOn. 2. ParAlleliNg the DeFinition of type fOr maXimal ordeRs, wE sAy tHaT $\mathCAl{O}^{\priMe T}$ And $\Mathcal{o}^{T}$ have tHE saMe TYPE iF there is a non-zero elEmENT $c\In B_{p}$ such That $c\mAThCaL{o}^{T}c^{-1}=\mathcAl{o}^{\prIme T}$, AND we wrIte $\mAThCal{O}^{T}\sim\MathcaL{o}^{\pRiMe T}$. By LeMmA 4 in [@C-G], wE kNow ThaT $\mathCAl{O}^{T}\Sim\matHcal{O}^{\priMe T}$ if ANd only if $\mathcaL{o}\sim\mathcal{O}^{\pRImE}$. 3. tHeRE is a SecOnd conjectuRe of cHevyRev aND GAlbRAith aBout tHe OCcURrence of the smallest $N_0$ fOr whicH both $A_{\mathcal{O^{\primE}}^{T}}(n_1)\geq a_{\matHCAL{O}^{T}}(n_1)$ and $a_{\MathCAl{o^{\Prime}}^{T}}(n_2)< a_{\mathcaL{O}^{T}}(n_2)$ oCcur for somE $N_1,n_2<n_

}$. They proved that their algorithm halt s u nle ss the re e xists anothern on-c onjugate maximal order $\ma th c al{O } '$ forwhich $ $ \l a b el{ eq n: opt do m }a_{\m ath cal{O}' ^T}(n)\geq a_ {\ mathcal{O}^T } (n )$$ for ev ery $n\in{\math bb{ N}}_0$ ,whe r e $a_ {\m athca l{O}^T } (n)$ i s defined i n . Foll o wing [@ C - G] , we thus say that $\ m at h cal{O}^{\prime T}$ op t im a l lydom inates $\ ma thcal { O}^{T}$ if h o lds for all $n\in {\mathbb{N} } _0$ . Chev yr eva nd Gal brait ht hen conjecture d in [@C-G Co njectu r e 1] th a t no ma ximalord ermayo pt im all yd omi n at e a n oth er. \[c on j: main\ ] Le t $ \ math cal {O}$ and$\mathcal{O}^ {\p rime } $ b e max imalorde rs of $ B_{p}$ . If$\ mathcal{O}^{\pr imeT}$ optim all ydom in ates$ \mathc al{ O}^ {T}$, t hen $\m a thc al { O } $and $\mathcal{O}^{ \p r i me }$ are o f thes am et ype. 1. Con ject u r e \[c onj: m ai n\] is e quival e nt t o Conje ct ure \[ co nj: int ro\]b ecau se all isomorp hisms of orders come from conjugat i on . 2 . Pa ral leling thedefi n itio n of ty pef or ma ximal o r de r s, we say that $\ma th cal{O} ^{\pr ime T}$ and $ \mathcal{O } ^ { T}$ have the sa m e type if ther e isa non-zero element$c\in B_{p}$such that $ c\mathca l{O }^{ T}c ^{- 1 } =\ mathcal{O}^{\ p r imeT} $, andwewrite $ \ma thc al{ O}^ {T }\sim\mat hcal{O}^ {\ pr im eT}$ . ByL emma 4 i n[@C -G ],we kn o w that $\ma thca l{ O} ^ {T} \sim\ma t hc a l {O}^ {\ pr imeT}$ i f and onl y if $\math cal{O}\si m\m a thca l{ O} ^{\prim e}$. 3. The re is a seco nd co njectu r e of Chev yrev and Galbraith abou t the oc cur rence ofthe small est $n_0$ fo r which both$a_{\ ma thc a l {O^{\ p r im e}} ^{ T}}(n_1)\g e q a_ {\mat hc al{O }^{T}}( n_1)$ and $a_{\mat h cal {O^{\prime}}^ {T} }(n_ 2 ) <a_{ \ ma t hca l{ O }^{ T } }(n_2)$ occur f or some $n _1 , n_ 2<n_

}$. They_proved that_their algorithm halts unless_there exists_another_non-conjugate maximal_order_$\mathcal{O}'$ for which_$$\label{eqn:optdom} a_{\mathcal{O}'^T}(n)\geq a_{\mathcal{O}^T}(n)$$ for_every $n\in{\mathbb{N}}_0$, where $a_{\mathcal{O}^T}(n)$_is defined in._Following_[@C-G], we thus say that $\mathcal{O}^{\prime T}$ optimally dominates $\mathcal{O}^{T}$ if holds for all $n\in{\mathbb{N}}_0$. Chevyrev_and_Galbraith then_conjectured_in_[@C-G Conjecture 1] that no_maximal order may optimally dominate_another. \[conj:main\] Let_$\mathcal{O}$ and $\mathcal{O}^{\prime}$ be maximal orders of $B_{p}$._If_$\mathcal{O}^{\prime T}$ optimally_dominates $\mathcal{O}^{T}$, then $\mathcal{O}$ and $\mathcal{O}^{\prime}$ are of the_same type. 1. Conjecture \[conj:main\] is_equivalent to Conjecture_\[conj:intro\]_because_all isomorphisms of orders_come from conjugation. 2. Paralleling the_definition of type for maximal orders,_we say that $\mathcal{O}^{\prime T}$ and $\mathcal{O}^{T}$_have the same type if there_is a non-zero element $c\in_B_{p}$ such_that $c\mathcal{O}^{T}c^{-1}=\mathcal{O}^{\prime T}$, and we_write $\mathcal{O}^{T}\sim\mathcal{O}^{\prime T}$._By Lemma_4 in [@C-G],_we know that $\mathcal{O}^{T}\sim\mathcal{O}^{\prime T}$ if_and only if_$\mathcal{O}\sim\mathcal{O}^{\prime}$. 3. There is a second_conjecture_of Chevyrev and_Galbraith_about_the occurrence_of the smallest_$n_0$_for which_both_$a_{\mathcal{O^{\prime}}^{T}}(n_1)\geq a_{\mathcal{O}^{T}}(n_1)$ and $a_{\mathcal{O^{\prime}}^{T}}(n_2)< a_{\mathcal{O}^{T}}(n_2)$ occur_for_some $n_1,n_2<n_

to\infty$ are being nested within the limit $\kappa\to0$. Hence, the subsequent constraints on diffusion anomaly deduced from these bounds appropriately take into account the divergence of $\langle|\nabla\theta|^2\rangle$ in infinite time. In essence, this approach is consistent with the double limit $\lim_{\kappa\to0}\lim_{t\to\infty}$ taken in a selective way, in the sense that a dynamical property (here being the peak mean-square gradients) is registered during the first limit $t\to\infty$ and then monitored during the second limit $\kappa\to0$. Evidently, without such a selection, this double limit would yield a trivial result because no smooth scalar distribution would survive the limit $t\to\infty$ before $\kappa$ tends to zero. In passing, it is worth mentioning that the asymptotic regime of small but positive $\kappa$ and large but finite time, presumably $t\gg T$, is relevant in the study of exponential decay behavior mentioned in the opening paragraph of this section. Production and dissipation of scalar gradients ---------------------------------------------- The evolution of $\nabla\theta$ is governed by $$\begin{aligned} \label{gradients} \partial_t\nabla\theta + (\u\cdot\nabla)\nabla\theta &=& \bomega\times\nabla\theta - (\nabla\theta\cdot\nabla)\u + \kappa\Delta\nabla\theta, ~~~~\nabla\cdot\u=0,\end{aligned}$$ where $\bomega=\nabla\times\u$ is the vorticity. Apart from the “Coriolis” term $\bomega\times\nabla\theta$, whose sole effect is to rotate $\nabla\theta$ without changing its magnitude, Eq. (\[gradients\]) resembles the three-dimensional vorticity equation in several aspects. For example, the gradient $\nabla\theta$ is advected by the velocity field $\u$ just as the vorticity $\bomega$. As another example, the term $(\nabla\theta\cdot\nabla)\u$, which is responsible for the production of scalar gradients, is an analogue of the vortex stretching term. Despite these apparent similarities, there are fundamental differences between “genuine” turbulence and passive scalar advection. First, the passive vector $\nabla\theta$ is irrotational whereas $\bomega$ is solenoidal. Second, $\theta$ is materially conserved while there exists

to\infty$ are being nested within the limit $ \kappa\to0$. Hence, the subsequent constraints on dissemination anomaly deduce from these bounds appropriately take into account the discrepancy of $ \langle|\nabla\theta|^2\rangle$ in infinite time. In kernel, this approach is consistent with the bivalent terminus ad quem $ \lim_{\kappa\to0}\lim_{t\to\infty}$ taken in a selective way, in the common sense that a dynamical property (here being the peak bastardly - straight gradients) is registered during the first limit $ t\to\infty$ and then monitored during the second terminus ad quem $ \kappa\to0$. Evidently, without such a selection, this double terminus ad quem would yield a trivial result because no smooth scalar distribution would survive the terminus ad quem $ t\to\infty$ before $ \kappa$ tends to zero. In elapse, it is worth mentioning that the asymptotic regime of belittled but positive $ \kappa$ and large but finite time, presumably $ t\gg T$, is relevant in the study of exponential decay behavior mentioned in the opening paragraph of this section. Production and dissipation of scalar gradients ---------------------------------------------- The evolution of $ \nabla\theta$ is governed by $ $ \begin{aligned } \label{gradients } \partial_t\nabla\theta + (\u\cdot\nabla)\nabla\theta & = & \bomega\times\nabla\theta - (\nabla\theta\cdot\nabla)\u + \kappa\Delta\nabla\theta, ~~~~\nabla\cdot\u=0,\end{aligned}$$ where $ \bomega=\nabla\times\u$ is the vorticity. Apart from the “ Coriolis ” term $ \bomega\times\nabla\theta$, whose sole effect is to revolve $ \nabla\theta$ without changing its magnitude, Eq. (\[gradients\ ]) resemble the three - dimensional vorticity equality in several aspects. For example, the gradient $ \nabla\theta$ is advected by the speed field $ \u$ barely as the vorticity $ \bomega$. As another example, the term $ (\nabla\theta\cdot\nabla)\u$, which is creditworthy for the production of scalar gradients, is an analogue of the vortex stretching term. Despite these apparent similarities, there are fundamental differences between “ genuine ” turbulence and passive scalar advection. First, the passive vector $ \nabla\theta$ is irrotational whereas $ \bomega$ is solenoidal. Second, $ \theta$ is materially conserved while there exist

to\ijfty$ are being nested wiuhin the limit $\kakpq\to0$. Heice, the subsequdnt constraints on diffusion abomalt deduced from these buunds appgopriatelt tajw into account the diverfcnce mh $\langle|\nabla\thgta|^2\rangle$ in infinite time. Iv zssence, this approach is consistent rith thr fouble limit $\lym_{\kakpa\eo0}\lij_{n\ti\infty}$ taken in a selective wzy, in tie sense that a dynamical property (here bfing the peak mean-squage gradientw) is eegistered djring the first limit $j\to\infty$ and then monitored durivg thz second linir $\kwkpa\to0$. Evidenvly, winhout such a selectiot, this couble limit wpulv yiwld a trivial result uecause no smooth scwlar distsiyution would survive rhw limht $t\do\inwry$ cefkrx $\kzppa$ tfnda to zero. Jn passing, ut is worth mentionond that the asyjptotis wegime of small but positive $\kappa$ and narfe but finite time, presymably $t\gg T$, is relevwnt in thq study of exponential decay behavior mentioned it the upeulkg pxeahraph of this section. Production and dissipatijh pf scalar gradiekts ---------------------------------------------- The evolution og $\janja\theta$ is goxerned by $$\begin{aligned} \label{hradienjs} \partual_t\nabla\uheta + (\u\cdot\nabla)\nabla\theta &=& \bomwga\times\nablc\thwta - (\nabla\theta\cdoc\nabla)\u + \kappc\Delta\mabla\yheta, ~~~~\nabla\cdot\u=0,\end{alignzd}$$ whede $\bomega=\nahla\times\u$ ks the vorticity. Apsrd from the “Coriolis” term $\bjmega\timew\nabka\theta$, whoxe solq effect id to vmtate $\nabla\theta$ wlthouj chancing its mwgnitude, Eq. (\[gradients\]) resembles the three-dimemshondl vortieity eauation in sederal aspects. Yor exam'le, thd gradient $\nabla\tieta$ is advested by the vapocity field $\u$ just ws tye virticith $\bomega$. As anoyher examilt, the term $(\nabla\theta\cdot\nabka)\u$, shich is responwiboe for the prodictkon ov xcwnar gradientv, is an snaloeue of uke yorgex xtretching term. Desphte fhese apparent simolwrities, jhere are fundamental cifferences betweej “gennine” tnrbulemce and passive scalar advection. Rirst, the pafsive vector $\naboa\theta$ is ixrotational whereas $\bomega$ is solenoidal. Second, $\theta$ is mqterially conserved ehile there exises

to\infty$ are being nested within the limit the constraints on anomaly deduced from account divergence of $\langle|\nabla\theta|^2\rangle$ infinite time. In this approach is consistent with the limit $\lim_{\kappa\to0}\lim_{t\to\infty}$ taken in a selective way, in the sense that a dynamical (here being the peak mean-square gradients) is registered during the first limit $t\to\infty$ then during second $\kappa\to0$. Evidently, without such a selection, this double limit would yield a trivial result because no scalar distribution would survive the limit $t\to\infty$ before tends to zero. In it is worth mentioning that asymptotic of small positive and but finite time, $t\gg T$, is relevant in the study of exponential decay behavior mentioned in the opening paragraph of section. Production of scalar ---------------------------------------------- evolution $\nabla\theta$ is governed \label{gradients} \partial_t\nabla\theta + (\u\cdot\nabla)\nabla\theta &=& \bomega\times\nabla\theta \kappa\Delta\nabla\theta, ~~~~\nabla\cdot\u=0,\end{aligned}$$ where $\bomega=\nabla\times\u$ is the vorticity. Apart the “Coriolis” $\bomega\times\nabla\theta$, whose sole effect is to $\nabla\theta$ without changing its magnitude, Eq. (\[gradients\]) resembles three-dimensional vorticity equation in several aspects. For example, the gradient $\nabla\theta$ is advected by the $\u$ just as the $\bomega$. As another the $(\nabla\theta\cdot\nabla)\u$, is for the of scalar gradients, is an analogue of the vortex stretching term. these apparent similarities, there are fundamental differences between “genuine” turbulence scalar First, the passive $\nabla\theta$ is irrotational whereas is Second, $\theta$ is materially there

to\infty$ are being nested withIn the limit $\Kappa\To0$. HEncE, tHe suBseqUent constraintS On diFfusion anomaly deduced fRom thEsE BounDS aPpropRiately TAkE INto AcCoUnt ThE DiVergeNce Of $\langlE|\nabla\thetA|^2\raNgLe$ in infinite TImE. In essence, ThiS approach is cOnsIstent WiTh tHE doubLe lImit $\lIm_{\kappA\To0}\lim_{t\To\infty}$ taKeN In a selECtive waY, IN tHe seNse that a dynamical PRoPErty (here being tHe peak MeAN-sQUAre GraDients) is reGiStereD During tHE fIRST liMIt $t\to\infty$ and Then monitorED duRing thE sEcoND limit $\Kappa\To0$. eVidEntly, withouT sucH a selectiOn, this DOuble liMIt would Yield a TriViaL resULt BeCauSe NO smOOtH scALar DistribuTiOn Would SurvIVE THe liMit $T\to\iNfty$ bEfore $\kappa$ tenDs tO zerO. in pAssinG, it is WortH mEntioNing thAt the AsYmptotic regime oF smaLl but posiTivE $\kAppA$ aNd larGE but fiNitE tiMe, presuMably $t\gG t$, is ReLEVAnT in the study of exponEnTIAl Decay behAvior mENtIoNEd in the oPeNinG parAGRaph oF thiS SeCtion. ProDuctioN AnD dIssipatIoN of scaLaR grAdiEnts ---------------------------------------------- THE evoLution Of $\nabla\tHeta$ iS Governed by $$\begiN{Aligned} \label{gRAdIENtS} \PartIal_T\nabla\theta + (\U\cdoT\NablA)\nabLA\tHetA &=& \BomegA\timeS\nABlA\Theta - (\nabla\theta\cdot\NaBla)\u + \kaPpa\DeLta\nabla\theta, ~~~~\Nabla\cdot\u=0,\END{Aligned}$$ wHere $\BOmEGa=\nabla\times\u$ iS the vOrticity. ApARt from thE “CoriOlis” term $\Bomega\timES\Nabla\theTa$, wHosE soLe eFFEcT is to rotate $\naBLA\theTa$ Without ChaNging itS maGniTudE, Eq. (\[GrAdients\]) reSembles tHe ThReE-dImeNsionAL vorticiTy EquAtIon In sevERal aspEcts. FOr exAmPlE, The GradienT $\NaBLA\theTa$ Is AdveCteD bY the vElocITy fIeld $\u$ juSt as the voRtiCIty $\bOmEgA$. As anotHer example, the TeRm $(\nabla\theTa\CdoT\nabla)\U$, WHich is reSponsible for the productiON of scalAr gRadieNts, iS an analogUe oF the voRteX StretcHing teRm. DesPiTe tHESe appAREnT siMiLarities, thERE arE fundAmEntaL differEnces between “genuinE” TurBulence and pasSivE scaLAR aDveCTiON. FiRsT, The PASsive vector $\nablA\theta$ is irRoTAtIonal whereAS $\boMeGa$ is solEnoidal. seconD, $\Theta$ is MateriallY conserveD wHile THEre Exists

to\infty$ are being nested within th e lim it$\k ap pa\t o0$. Hence, the su b sequ ent constraints on dif fusio na noma l ydeduc ed from th e s e b ou nd s a pp r op riate lytake in to account th edivergence o f $ \langle|\n abl a\theta|^2\r ang le$ in i nfi n ite t ime . Inessenc e , this approach i s consi s tent wi t h t he d ouble limit $\lim _ {\ k appa\to0}\lim_ {t\to\ in f ty } $ ta ken in a sele ct ive w a y, in t h es e n set hat a dynamic al property (he re bei ng th e peakmean- sq u are gradients) isregistere d duri n g the f i rst lim it $t\ to\ inf ty$a nd t hen m o nit o re d d u rin g the se co nd limi t $\ k a p p a\to 0$. Evi dentl y, without su cha se l ect ion,thisdoub le limi t woul d yie ld a trivial resu lt b ecause no sm oo thsc alard istrib uti onwould s urvivet heli m i t $ t\to\infty$ before $ \ k ap pa$ tend s to z e ro .I n passin g, it isw o rth m enti o ni ng thatthe as y mp to tic reg im e of s ma llbut posi t ive$\kapp a$ and l argeb ut finite time , presumably $ t \g g T$ , isrel evant in th e st u dy o f ex p on ent i al de cay b eh a vi o r mentioned in theop eningparag raph of thissection. P r o ductionandd is s ipation of sca lar g radients - - -------- ----- -------- --------- - - -------- --- - The ev o l ut ion of $\nabl a \ thet a$ is gov ern ed by $ $\b egi n{a lig ne d} \label {gradien ts }\p ar tia l_t\n a bla\thet a+ ( \u \cd ot\na b la)\na bla\t heta & =& \b omega\t i me s \ nabl a\ th eta- ( \n abla\ thet a \cd ot\nabl a)\u + \k app a \Del ta \n abla\th eta, ~~~~\nab la \cdot\u=0, \e nd{ aligne d } $$ where $\bomega=\nabla\times\ u $ is th e v ortic ity. Apart fr omthe “C ori o lis” t erm $\ bomeg a\ tim e s \nabl a \ th eta $, whose sol e eff ect i sto r otate $ \nabla\theta$ with o utchanging itsmag nitu d e ,Eq. (\ [ gra di e nts \ ] ) resembles the three-dim en s io nal vortic i tyeq uationin seve ral a s pects.For examp le, the g ra dien t $\n abla\theta $ is adv ected byt he ve l oc ity f iel d $\u$ j ust as t he vor t ici ty $\ bomega $. As an other e xample,the term $(\nabla\theta \cdot\ nabla )\u $, whichisr esp onsible f or t he product ion of scal arg radie nts, is an analo gueo f the vor t ex st r e tc hing term.D e s pit e the sea pparen t si milarities, there are fundamenta l di f f ere nce s bet we en “genuine” t urb ul e n ce and p as sive scalar advecti on . Firs t, the passi ve vect o r $ \ nabla\ thet a$is irrota tio na l wherea s$\ b omega$ isso lenoid al. Se c ond, $ \theta$ is mater ially c onser v edwhile t here ex i sts

to\infty$ are_being nested_within the limit $\kappa\to0$._Hence, the_subsequent_constraints on_diffusion_anomaly deduced from_these bounds appropriately_take into account the_divergence of $\langle|\nabla\theta|^2\rangle$_in_infinite time. In essence, this approach is consistent with the double limit $\lim_{\kappa\to0}\lim_{t\to\infty}$ taken_in_a selective_way,_in_the sense that a dynamical_property (here being the peak_mean-square gradients)_is registered during the first limit $t\to\infty$ and_then_monitored during the_second limit $\kappa\to0$. Evidently, without such a selection, this_double limit would yield a trivial_result because no_smooth_scalar_distribution would survive the_limit $t\to\infty$ before $\kappa$ tends to_zero. In passing, it is worth_mentioning that the asymptotic regime of small_but positive $\kappa$ and large but_finite time, presumably $t\gg T$,_is relevant_in the study of exponential_decay behavior mentioned_in the_opening paragraph of_this section. Production and dissipation of scalar_gradients ---------------------------------------------- The evolution of_$\nabla\theta$ is governed by $$\begin{aligned} \label{gradients} \partial_t\nabla\theta +_(\u\cdot\nabla)\nabla\theta_&=& \bomega\times\nabla\theta -_(\nabla\theta\cdot\nabla)\u +_\kappa\Delta\nabla\theta, ~~~~\nabla\cdot\u=0,\end{aligned}$$_where $\bomega=\nabla\times\u$_is the vorticity._Apart_from the_“Coriolis”_term $\bomega\times\nabla\theta$, whose sole effect is_to_rotate $\nabla\theta$ without changing its magnitude, Eq._(\[gradients\]) resembles the three-dimensional_vorticity_equation in several aspects._For example, the gradient $\nabla\theta$_is advected by the velocity field_$\u$ just_as the_vorticity $\bomega$. As another example, the term $(\nabla\theta\cdot\nabla)\u$, which is responsible_for the production of scalar gradients,_is an analogue of_the vortex_stretching_term. Despite these_apparent_similarities, there_are fundamental differences between “genuine” turbulence and_passive scalar_advection. First, the passive vector $\nabla\theta$_is irrotational whereas $\bomega$_is_solenoidal. Second, $\theta$ is materially conserved_while there exists

tau_{0}$ decays exponentially, with a rate that scales polynomially with $q$. Consider the kinetically constrained model described above, with $\pi>p^{\text{SP}}$ and $q\le q^{\text{OP}}$. 1. $\nu$-almost surely the spectral gap is $0$, i.e., the relaxation time is infinite. 2. There exist two positive constants $c,C$ depending on $\pi$ and a $\nu$-random variable $\tau$ such that 1. $\pp_{\mu}\left(\tau_{0}\ge t\right)\le e^{-\nicefrac{t}{\tau}}$ for all $t>0$, 2. $\nu\left(\tau\ge t\right)\le C\,t^{\frac{c}{\log q}}$ for $t$ large enough. Some tools ========== In this section we will present some tools that will help us analyze the kinetically constrained models that we have introduced. We will start by considering a general state space $\Omega$, and any Markov process on $\Omega$ that is reversible with respect to a certain measure $\mu$. We denote its generator by $\mathcal{L}$ and the associated Dirichlet form by $\mathcal{D}$. We will consider, for some event $A$, its hitting time $\tau_{A}$. With some abuse of notation, we use $\tau_{A}$ also for the $\mu$-random variable giving for every state $\eta\in\Omega$ the expected hitting time at $A$ starting from that state: $$\tau_{A}\left(\eta\right)=\ee_{\eta}\left(\tau_{A}\right).$$ $\tau_{A}\left(\eta\right)$ satisfies the following Poisson problem: $$\begin{aligned} \mathcal{L}\tau_{A} & =-1\text{ on }A^{c},\label{eq:poissonproblem}\\ \tau_{A} & =0\text{ on }A.\nonumber \end{aligned}$$ By multiplying both sides of the equation by $\tau_{A}$ and integrating with respect to $\mu$, we obtain \[cor:dirichletequalsexpectation\]$\mu\left(\tau_{A}\right)=\mathcal{D}\tau_{A}$. Rewriting this corollary as $\mu\left(\tau_{A}\right)=\frac{\mu\left

tau_{0}$ decays exponentially, with a rate that scales polynomially with $ q$. Consider the kinetically restrain exemplar described above, with $ \pi > p^{\text{SP}}$ and $ q\le q^{\text{OP}}$. 1. $ \nu$-almost surely the spectral gap is $ 0 $, i.e., the rest time is infinite. 2. There exist two positivist constant $ c, C$ depend on $ \pi$ and a $ \nu$-random variable $ \tau$ such that 1. $ \pp_{\mu}\left(\tau_{0}\ge t\right)\le e^{-\nicefrac{t}{\tau}}$ for all $ t>0 $, 2. $ \nu\left(\tau\ge t\right)\le C\,t^{\frac{c}{\log q}}$ for $ t$ large enough. Some cock = = = = = = = = = = In this section we will present some instrument that will avail us analyze the kinetically constrained models that we have introduced. We will start by considering a cosmopolitan state space $ \Omega$, and any Markov process on $ \Omega$ that is reversible with deference to a certain measure $ \mu$. We announce its generator by $ \mathcal{L}$ and the associated Dirichlet form by $ \mathcal{D}$. We will see, for some event $ A$, its hitting time $ \tau_{A}$. With some abuse of notation, we practice $ \tau_{A}$ also for the $ \mu$-random variable giving for every state $ \eta\in\Omega$ the expected hitting time at $ A$ startle from that state: $ $ \tau_{A}\left(\eta\right)=\ee_{\eta}\left(\tau_{A}\right).$$ $ \tau_{A}\left(\eta\right)$ satisfies the following Poisson problem: $ $ \begin{aligned } \mathcal{L}\tau_{A } & = -1\text { on } A^{c},\label{eq: poissonproblem}\\ \tau_{A } & = 0\text { on } A.\nonumber \end{aligned}$$ By multiplying both sides of the equation by $ \tau_{A}$ and integrating with respect to $ \mu$, we obtain \[cor: dirichletequalsexpectation\]$\mu\left(\tau_{A}\right)=\mathcal{D}\tau_{A}$. rewrite this corollary as $ \mu\left(\tau_{A}\right)=\frac{\mu\left

tau_{0}$ decays exponentially, wiuh a rate that sccoes ponynomizlly witf $q$. Consider the kinetically ronsrraintb model described abuve, with $\ii>p^{\text{SP}}$ and $w\le q^{\text{O'}}$. 1. $\nu$-almost sudcly tke spectral gap ls $0$, i.e., the selaxation tima ks infinite. 2. There exist two positive constamtd $c,C$ depending on $\ki$ wnd z $\nu$-random variable $\tau$ such that 1. $\pp_{\ku}\left(\tau_{0}\ge t\tight)\le e^{-\nicefrac{t}{\tau}}$ for wll $h>0$, 2. $\nu\left(\tau\ge t\right)\le C\,j^{\rras{x}{\log q}}$ for $t$ large enolyh. Some toola ========== In this section we will preseng somz tools thaj sipn help us aialyze the kinetically convtrainec models that ee iave introduced. We will svart by considering w general scate space $\Omega$, and qnt Marnov [rocdws un $\Kmxga$ that ls ceversible sith respecr to a certain measirq $\mu$. We denote its gqnqrator by $\mathcal{L}$ and the associated Dpricglet form by $\mathcal{D}$. Ww will consider, for slme event $A$, its hitting time $\tau_{A}$. With some abuse of notathon, wx jse $\twj_{Q}$ wlso for the $\mu$-random variable giving for evewg xtste $\eta\in\Omega$ the expectrd hojting time at $X$ starcjnf from that state: $$\hau_{A}\lefj(\eta\ritht)=\ee_{\eta}\ltft(\tai_{A}\right).$$ $\tau_{A}\left(\eta\right)$ satusfies the fjolowing Poisson prlblem: $$\begin{clignec} \mathval{L}\tau_{A} & =-1\text{ on }A^{c},\labzl{eq:pojssonproblel}\\ \tau_{A} & =0\tesg{ on }A.\nonumber \evd{akicned}$$ By multiplying both siqes of thx equction by $\tau_{S}$ and yntegratinh witm respect to $\mu$, we lbtaiu \[cor:dhrichleteqkalsexpectation\]$\mu\left(\tau_{A}\right)=\mevhcal{D}\tau_{A}$. Rewtithng this coxollarn as $\mu\left(\tau_{A}\wight)=\frac{\mu\lefj

tau_{0}$ decays exponentially, with a rate that with Consider the constrained model described q^{\text{OP}}$. $\nu$-almost surely the gap is $0$, the relaxation time is infinite. 2. exist two positive constants $c,C$ depending on $\pi$ and a $\nu$-random variable $\tau$ that 1. $\pp_{\mu}\left(\tau_{0}\ge t\right)\le e^{-\nicefrac{t}{\tau}}$ for all $t>0$, 2. $\nu\left(\tau\ge t\right)\le C\,t^{\frac{c}{\log q}}$ $t$ enough. tools In this section we will present some tools that will help us analyze the kinetically constrained that we have introduced. We will start by a general state space and any Markov process on that reversible with to certain $\mu$. We denote generator by $\mathcal{L}$ and the associated Dirichlet form by $\mathcal{D}$. We will consider, for some event $A$, hitting time some abuse notation, use also for the giving for every state $\eta\in\Omega$ the at $A$ starting from that state: $$\tau_{A}\left(\eta\right)=\ee_{\eta}\left(\tau_{A}\right).$$ $\tau_{A}\left(\eta\right)$ the following problem: $$\begin{aligned} \mathcal{L}\tau_{A} & =-1\text{ on \tau_{A} & =0\text{ on }A.\nonumber \end{aligned}$$ By multiplying sides of the equation by $\tau_{A}$ and integrating with respect to $\mu$, we obtain \[cor:dirichletequalsexpectation\]$\mu\left(\tau_{A}\right)=\mathcal{D}\tau_{A}$. corollary as $\mu\left(\tau_{A}\right)=\frac{\mu\left

tau_{0}$ decays exponentially, witH a rate that ScaleS poLynOmIallY witH $q$. Consider the kINetiCally constrained model dEscriBeD AbovE, WiTh $\pi>p^{\Text{SP}}$ aND $q\LE Q^{\teXt{oP}}$. 1. $\Nu$-aLmOSt SurelY thE spectrAl gap is $0$, i.e., tHe rElAxation time iS InFinite. 2. TherE exIst two positiVe cOnstanTs $C,C$ dEPendiNg oN $\pi$ anD a $\nu$-raNDom varIable $\tau$ sUcH That 1. $\pp_{\MU}\left(\taU_{0}\GE t\RighT)\le e^{-\nicefrac{t}{\tau}}$ fOR aLL $t>0$, 2. $\nu\left(\tau\ge t\Right)\lE C\,T^{\FrAC{C}{\loG q}}$ fOr $t$ large enOuGh. SomE Tools ========== In THiS SECtiON we will presenT some tools tHAt wIll helP uS anALyze thE kineTiCAllY constraineD modEls that we Have inTRoduced. wE will stArt by cOnsIdeRing A GeNeRal StATe sPAcE $\OmEGa$, aNd any MarKoV pRocesS on $\OMEGA$ That Is rEverSible With respect to A ceRtaiN MeaSure $\mU$. We deNote ItS geneRator bY $\mathCaL{L}$ and the associaTed DIrichlet fOrm By $\MatHcAl{D}$. We WIll conSidEr, fOr some eVent $A$, itS HitTiNG TImE $\tau_{A}$. With some abuse Of NOTaTion, we usE $\tau_{A}$ aLSo FoR The $\mu$-ranDoM vaRiabLE GivinG for EVeRy state $\eTa\in\OmEGa$ ThE expectEd HittinG tIme At $A$ StartINg frOm that State: $$\tau_{a}\left(\ETa\right)=\ee_{\eta}\leFT(\tau_{A}\right).$$ $\tau_{a}\LeFT(\EtA\RighT)$ saTisfies the fOlloWIng POissON pRobLEm: $$\begIn{aliGnED} \mAThcal{L}\tau_{A} & =-1\text{ on }A^{c},\lAbEl{eq:poIssonProblem}\\ \tau_{A} & =0\teXt{ on }A.\nonumBER \End{alignEd}$$ By MUlTIplying both sidEs of tHe equation BY $\tau_{A}$ and IntegRating wiTh respect TO $\Mu$, we obtaIn \[cOr:dIriChlETEqUalsexpectatiON\]$\Mu\leFt(\Tau_{A}\rigHt)=\mAthcal{D}\Tau_{a}$. ReWriTinG tHis corollAry as $\mu\lEfT(\tAu_{a}\rIghT)=\frac{\MU\left

tau_{0}$ decays exponentia lly, witha rat e t hat s cale s po lynomially wit h $q$ . Consider the kineti cally c o nstr a in ed mo del des c ri b e d a bo ve , w it h $ \pi>p ^{\ text{SP }}$ and $q \le q ^{\text{OP}} $ . 1. $\nu$ -al most surelythe spect ra l g a p is$0$ , i.e ., the relaxa tion time i s infin i te. 2. Th ereexist two positiv e c o nstants $c,C$depend in g o n $\p i$and a $\nu $- rando m variab l e$ \ t au$ such that 1. $\pp_ { \mu }\left (\ tau _ {0}\ge t\ri gh t )\l e e^{-\nice frac {t}{\tau} }$ for all $t> 0 $, 2. $ \nu \le ft(\ t au \g e t \r i ght ) \l e C \ ,t^ {\frac{c }{ \l og q} }$ f o r $ t$ l arg e en ough. Some tools=== ==== = == In t his s ecti on we w ill pr esent s ome tools thatwill help usana ly zeth e kin e ticall y c ons trained models tha tw e ha ve introduced. Wewi l l s tart byconsid e ri ng a genera lsta te s p a ce $\ Omeg a $, and any Marko v p ro cess on $ \Omega $tha t i s rev e rsib le wit h respec t toa certain measu r e $\mu$. We d e no t e i t s ge ner ator by $\m athc a l{L} $ an d t hea ssoci atedDi r ic h let form by $\mathc al {D}$.We wi ll consider,for some e v e n t $A$, i ts h i tt i ng time $\tau_ {A}$. With some abuse of nota tion, we use $\ta u _ {A}$ als o f orthe $\ m u $- random variab l e giv in g for e ver y state $\ eta \in \Om eg a$ the ex pected h it ti ng t ime at $ A $ starti ng fr om th at st a te: $$ \tau_ {A}\ le ft ( \et a\right ) =\ e e _{\e ta }\ left (\t au _{A}\ righ t ).$ $ $\ta u_{A}\lef t(\ e ta\r ig ht )$ sati sfies the fol lo wing Poiss on pr oblem: $ $\begin{ aligned} \mathcal{L}\ta u _{A} &=-1 \text { on }A^{c},\ lab el{eq: poi s sonpro blem}\ \ \ta u_ {A} & =0\t e x t{ on } A.\nonumbe r \en d{ali gn ed}$ $ By m ultiplying both si d esof the equati onby $ \ t au _{A } $a ndin t egr a t ing with respec t to $\mu$ ,w eobtain \[ c or: di richlet equalse xpect a tion\]$ \mu\left( \tau_{A}\ ri ght) = \ mat hcal{D}\ta u_{A}$. Rewritin g this co rolla ryas $\m u\ lef t(\ta u_{A}\ r igh t)=\f rac{\m u\ left

tau_{0}$ decays_exponentially, with_a rate that scales_polynomially with_$q$. Consider_the kinetically_constrained_model described above,_with $\pi>p^{\text{SP}}$ and_$q\le q^{\text{OP}}$. 1. $\nu$-almost_surely the spectral_gap_is $0$, i.e., the relaxation time is infinite. 2. There exist two positive constants_$c,C$_depending on_$\pi$_and_a $\nu$-random variable $\tau$ such_that 1._ $\pp_{\mu}\left(\tau_{0}\ge_t\right)\le e^{-\nicefrac{t}{\tau}}$ for all $t>0$, _2._ $\nu\left(\tau\ge t\right)\le_C\,t^{\frac{c}{\log q}}$ for $t$ large enough. Some tools ========== In this section_we will present some tools that_will help us_analyze_the_kinetically constrained models that_we have introduced. We will start_by considering a general state space_$\Omega$, and any Markov process on $\Omega$_that is reversible with respect to_a certain measure $\mu$. We_denote its_generator by $\mathcal{L}$ and the_associated Dirichlet form_by $\mathcal{D}$._We will consider,_for some event $A$, its hitting_time $\tau_{A}$. With_some abuse of notation, we use_$\tau_{A}$_also for the_$\mu$-random_variable_giving for_every state $\eta\in\Omega$_the_expected hitting_time_at $A$ starting from that state:_$$\tau_{A}\left(\eta\right)=\ee_{\eta}\left(\tau_{A}\right).$$ $\tau_{A}\left(\eta\right)$_satisfies the following Poisson problem: $$\begin{aligned} \mathcal{L}\tau_{A} &_=-1\text{ on }A^{c},\label{eq:poissonproblem}\\ \tau_{A} &_=0\text{_on }A.\nonumber \end{aligned}$$ By multiplying_both sides of the equation_by $\tau_{A}$ and integrating with respect_to $\mu$,_we obtain \[cor:dirichletequalsexpectation\]$\mu\left(\tau_{A}\right)=\mathcal{D}\tau_{A}$. Rewriting_this corollary as $\mu\left(\tau_{A}\right)=\frac{\mu\left

-50 [@he2016deep] pretrained on ImageNet [@russakovsky2015imagenet]. For computational efficiency here we use only the univariate version of our approach with 25 centers, since the multivariate version requires optimizing a precision matrix of size $1,000 \times 1,000$. **Pedestrian counting.** Figure \[fig:ucsd\] shows the results of our approach on the realistic problem of pedestrian counting from images on the UCSD dataset, together with the training and validation losses. We compare with [@dalal2005histograms] that uses low level image features, [@felzenszwalb2008discriminatively] relying on a person detection method specifically trained for the task, and [@chan2012counting] which employs motion segmentation masks. Unlike these methods, we do not use either motion segmentation masks or class specific detectors. We use only global image features extracted from a pretrained deep network, and we manage to obtain comparable performance with [@chan2012counting], while greatly outperforming [@dalal2005histograms; @felzenszwalb2008discriminatively]. **Generic object counting.** In table \[tab:voc\] the goal is generic object counting on the VOC-2007 generic object dataset. We compare with the set of models proposed in the very recent deep learning method of [@chattopadhyay2017counting]. Our features are extracted from a pretrained deep learning model, while theirs are specifically fine-tuned for this counting task. We outperform their *“glance"* models, which similar to us, rely on global image features. We additionally obtain comparable performance to *aso-sub-ft-1L-3$\times$3* and *seq-sub-ft-3$\times$3* which rely on local image information, as they divide each image into a 3$\times$3 grid and extract features from each cell. It is worthwhile noting that we use only 25 data centers for computing the kernel matrix, and achieve comparable performance with methods relying on stronger features. These results support our approach. Conclusions =========== This work brings forth an asymmetric kernel for the Gaussian Process model. This encompasses three components: (i) training on training centers only, (ii) learning individualized kernel metrics per center and, (iii) extending the lengthscale hyper-parameter to a precision matrix, thus learning not only the appropriate size but also the shape in the kernel metric. Due to the limitations imposed by

-50 [ @he2016deep ] pretrained on ImageNet [ @russakovsky2015imagenet ]. For computational efficiency here we use only the univariate version of our approach with 25 center, since the multivariate version requires optimize a preciseness matrix of size $ 1,000 \times 1,000$. * * Pedestrian counting. * * Figure   \[fig: ucsd\ ] shows the consequence of our approach on the realistic trouble of pedestrian count from images on the UCSD dataset, together with the education and validation losses. We compare with [ @dalal2005histograms ] that use broken level prototype features, [ @felzenszwalb2008discriminatively ] relying on a person signal detection method specifically trained for the task, and [ @chan2012counting ] which hire motion segmentation mask. Unlike these methods, we do not practice either motion division masks or class specific detectors. We practice only global image features extracted from a pretrained deep net, and we pull off to obtain comparable performance with [ @chan2012counting ], while greatly outperforming [ @dalal2005histograms; @felzenszwalb2008discriminatively ]. * * Generic object counting. * * In table   \[tab: voc\ ] the goal is generic object counting on the VOC-2007 generic aim dataset. We compare with the stage set of models aim in the very late deep learning method of [ @chattopadhyay2017counting ]. Our features are extracted from a pretrained deep learning model, while theirs are specifically very well - tuned for this counting task. We outperform their * “ glance " * models, which similar to us, trust on global image features. We additionally obtain comparable performance to * aso - sub - ft-1L-3$\times$3 * and * seq - sub - ft-3$\times$3 * which trust on local persona information, as they divide each image into a 3$\times$3 grid and extract features from each cell. It is worthwhile note that we use only 25 data centers for calculate the kernel matrix, and achieve comparable performance with methods relying on stronger features. These consequence support our access. Conclusions = = = = = = = = = = = This employment brings forth an asymmetrical kernel for the Gaussian Process model. This encompasses three components: (one) training on education centers only, (ii) learning individualized kernel metric function per center and, (iii) extending the lengthscale hyper - parameter to a precision matrix, thus learn not only the appropriate size but also the human body in the kernel metric. Due to the limitations imposed by

-50 [@he2016feep] pretrained on ImageKet [@russakovsky2015imagenet]. For ckmputatiunal efficiency here we use lnoy tht univariate versiov of our wpproach witi 25 centers, since the mulbnvarizbe vexsmon requires opjimizing a psecision matrif uf size $1,000 \times 1,000$. **Pedestrian counting.** Fidure \[fig:icdd\] shows the rgsultx of klr approach on the realistic prkblem oh pedestrian cointing from images on the KCSD dataset, together aith the trqinigt and validagion losses. We compare with [@dalal2005histograms] that uses low kevel imagg rewjures, [@felzenwzwalf2008discriminatlnely] renying om a person detcctioi merhod specifically tramned for the task, anq [@chan2012couttnng] which employs motuob segkentdtiov marks. Uiliie thede jethods, we do not use either motion segmtntwnoon masks or class s[ecific detectors. We use only global imdge features extracted fron a pretrained deep ngtwork, and we manage to obtain comparable performance with [@whan2012ckjntnkn], whkoe greatly outperforming [@dalal2005histograms; @felzenfawsln2008discriminativeln]. **Generic object coinhimd.** In table \[tab:xoc\] thz gkal is generic objfct cougting on the VJC-2007 grneric object dataset. We conpare with tke wet of models proplsed in the veri recemt deep learning method of [@dhattopadhywy2017countinf]. Our features ard eqtrawted from a pretrained dee[ learninj modzl, while theors arq specificwlly nhne-tuned for this fountnng tdsk. We outoerform their *“glance"* models, whiri similar to os, selj on globcl imane features. We wdditionally oytain coiparacle perforjance tm *aso-sub-ft-1L-3$\eimes$3* and *seq-vob-ft-3$\times$3* whirh rely og loxal umage ivwormation, as tney dividv tach image into a 3$\times$3 grid ana extract featurzr from each cell. It is wlrvhwhija noting thad we usd only 25 data centcrs for computing the kernen mafrix, and achieve cpmiarable pgrformancq with methodx relying on stronher fxaturer. Thexe tesults support our approach. Condlusions =========== Tjis work brings forbh ag asymmetrnc kernel for the Gaussian Process model. This encompasses rhree components: (i) jralning on tramning senters otly, (ii) learning indicidualized kernel metrics per center ahd, (iii) extejding the lengthscale hyper-parameter to a precision matrix, thus learning bot only the approlriaye side yut also ehe diape in the kernek metric. Due to the limitations imposed bv

-50 [@he2016deep] pretrained on ImageNet [@russakovsky2015imagenet]. For here use only univariate version of since multivariate version requires a precision matrix size $1,000 \times 1,000$. **Pedestrian counting.** \[fig:ucsd\] shows the results of our approach on the realistic problem of pedestrian from images on the UCSD dataset, together with the training and validation losses. compare [@dalal2005histograms] uses level image features, [@felzenszwalb2008discriminatively] relying on a person detection method specifically trained for the task, and which employs motion segmentation masks. Unlike these methods, do not use either segmentation masks or class specific We only global features from pretrained deep network, we manage to obtain comparable performance with [@chan2012counting], while greatly outperforming [@dalal2005histograms; @felzenszwalb2008discriminatively]. **Generic object counting.** In \[tab:voc\] the generic object on VOC-2007 object dataset. We the set of models proposed in deep learning method of [@chattopadhyay2017counting]. Our features are from a deep learning model, while theirs are fine-tuned for this counting task. We outperform their models, which similar to us, rely on global image features. We additionally obtain comparable performance and *seq-sub-ft-3$\times$3* which rely local image information, they each into 3$\times$3 grid extract features from each cell. It is worthwhile noting that we only 25 data centers for computing the kernel matrix, and performance methods relying on features. These results support approach. =========== This work brings asymmetric for model. encompasses components: (i) training on centers only, (ii) learning individualized metrics per center and, to a precision matrix, thus learning not only appropriate size but also the shape in kernel metric. Due to the limitations imposed by

-50 [@he2016deep] pretrained on ImageNeT [@russakovsKy2015imaGenEt]. FOr CompUtatIonal efficiencY Here We use only the univariate VersiOn OF our APpRoach With 25 cenTErS, SIncE tHe MulTiVArIate vErsIon requIres optimiZinG a Precision matRIx Of size $1,000 \timeS 1,000$. **PeDestrian counTinG.** FigurE \[fIg:uCSd\] shoWs tHe resUlts of OUr apprOach on the ReAListic PRoblem oF PEdEstrIan counting from imAGeS On the UCSD datasEt, togeThER wITH thE trAining and vAlIdatiON losses. wE cOMPAre WIth [@dalal2005histoGrams] that usES loW level ImAge FEatureS, [@felzEnSZwaLb2008discriminAtivEly] relyinG on a peRSon deteCTion metHod speCifIcaLly tRAiNeD foR tHE taSK, aNd [@cHAn2012cOunting] wHiCh EmploYs moTION SegmEntAtioN maskS. Unlike these mEthOds, wE Do nOt use EitheR motIoN segmEntatiOn masKs Or class specific DeteCtors. We usE onLy GloBaL imagE FeaturEs eXtrActed frOm a pretRAinEd DEEP nEtwork, and we manage tO oBTAiN comparaBle perFOrMaNCe with [@chAn2012CouNtinG], WHile gReatLY oUtperforMing [@daLAl2005HiStogramS; @fElzensZwAlb2008DisCrimiNAtivEly]. **GenEric objeCt couNTing.** In table \[tab:VOc\] the goal is geNErIC ObJEct cOunTing on the VOc-2007 genERic oBjecT DaTasET. We coMpare WiTH tHE set of models proposeD iN the veRy recEnt deep learniNg method of [@CHATtopadhyAy2017coUNtINg]. Our features aRe extRacted from A PretrainEd deeP learninG model, whiLE Theirs arE spEciFicAllY FInE-tuned for this COUntiNg Task. We oUtpErform tHeiR *“glAncE"* moDeLs, which siMilar to uS, rElY oN gLobAl imaGE featureS. WE adDiTioNally OBtain cOmparAble PeRfORmaNce to *asO-SuB-FT-1L-3$\tiMeS$3* aNd *seQ-suB-fT-3$\timeS$3* whiCH reLy on locAl image inForMAtioN, aS tHey diviDe each image inTo A 3$\times$3 grid AnD exTract fEATures froM each cell. It is worthwhile NOting thAt wE use oNly 25 dAta centerS foR compuTinG The kerNel matRix, anD aChiEVE compARAbLe pErFormance wiTH MetHods rElYing On stronGer features. These reSUltS support our apProAch. CONClUsiONs =========== tHis WoRK brINGs forth an asymmeTric kernel FoR ThE Gaussian PROceSs Model. ThIs encomPasseS Three coMponents: (i) Training oN tRainING ceNters only, (iI) learninG individuALized KErNel meTriCs per cEnTer And, (iiI) extenDIng The leNgthscAlE hyper-ParamEtEr to a preCision matrix, thus learninG not onLy the AppRopriate sIze BUt aLso the shaPe in The kernel mEtrIc. DUe to tHe lIMitatIons IMpOseD By

-50 [@he2016deep] pretrain ed on Imag eNet[@r uss ak ovsk y201 5imagenet]. Fo r com putational efficiencyherewe useo nl y the univar i at e ver si on of o u rappro ach with 2 5 centers, si nc e the multiv a ri ate versio n r equires opti miz ing apr eci s ion m atr ix of size$ 1,000\times 1, 00 0 $. ** P edestri a n c ount ing.** Figure \[f i g: u csd\] shows th e resu lt s o f our ap proach onth e rea l istic p r ob l e m of pedestrian co unting from ima ges on t heU CSD da taset ,t oge ther with t he t raining a nd val i dationl osses.We com par e w ith[ @d al al2 00 5 his t og ram s ] t hat uses l ow leve l im a g e feat ure s, [ @felz enszwalb2008d isc rimi n ati vely] rely ingon a pe rson d etect io n method specif ical ly traine d f or th etask, and [@ cha n20 12count ing] wh i chem p l o ys motion segmentati on m as ks. Unli ke the s eme t hods, we d o n ot u s e eith er m o ti on segme ntatio n m as ks or c la ss spe ci fic de tecto r s. W e useonly glo bal i m age features e x tracted froma p r e tr a ined de ep network, and we m anag e t o o b taincompa ra b le performance with [@ ch an2012 count ing], while g reatly out p e r forming[@da l al 2 005histograms; @fel zenszwalb2 0 08discri minat ively]. **Generi c object c oun tin g.* * I n ta ble \[tab:voc \ ] the g oal isgen eric ob jec t c oun tin gon the VO C-2007 g en er ic o bje ct da t aset. We c omp ar e w ith t h e setof mo dels p ro p ose d in th e v e r y re ce nt dee p l ea rning met h odof [@ch attopadhy ay2 0 17co un ti ng]. Ou r features ar eextractedfr oma pret r a ined dee p learning model, while theirsare spec ific ally fine -tu ned fo r t h is cou ntingtask. W e o u t perfo r m t hei r*“glance"* m ode ls, w hi ch s imilarto us, rely on glo b alimage feature s.We a d d it ion a ll y ob ta i n c o m parable perform ance to *a so - su b-ft-1L-3$ \ tim es $3* and *seq-s ub-ft - 3$\time s$3* whic h rely on l ocal i mag e informat ion, asthey divi d e eac h i mageint o a 3$ \t ime s$3 g rid an d ex tract featu re s from each c ell. Itis worthwhile noting th at weuse o nly 25 datacen t ers for comp utin g the kern elmat rix,and achie ve c o mp ara b le pe rfor m ance with me tho d s r elying on s t r o nge r fea tur e s. The se r esults support ou r approach. Co nclu s i ons == = ==== == == This workbri ng s forth an a symmetric k ernel fo rt he Ga ussian Proce ss mode l . T h is enc ompa sse s three c omp on e nts: (i )tr a iningon t ra iningcenter s onl y , (ii) learning i ndivi d u alize d ke rnelme trics p e r ce nter and,(iii) exten ding t he l ength scale h yp er-par ame te r to a pre c ision mat rix,thus le ar ning no t only the a pprop riat esiz e but als o th e s ha p e i n th e ker ne l me tric. Due to the l imi t ationsim pos e d by

-50 [@he2016deep]_pretrained on_ImageNet [@russakovsky2015imagenet]. For computational_efficiency here_we_use only_the_univariate version of_our approach with_25 centers, since the_multivariate version requires_optimizing_a precision matrix of size $1,000 \times 1,000$. **Pedestrian counting.** Figure \[fig:ucsd\] shows the results of_our_approach on_the_realistic_problem of pedestrian counting from_images on the UCSD dataset,_together with_the training and validation losses. We compare with_[@dalal2005histograms]_that uses low_level image features, [@felzenszwalb2008discriminatively] relying on a person detection_method specifically trained for the task,_and [@chan2012counting] which_employs_motion_segmentation masks. Unlike these_methods, we do not use either_motion segmentation masks or class specific_detectors. We use only global image features_extracted from a pretrained deep network,_and we manage to obtain_comparable performance_with [@chan2012counting], while greatly outperforming_[@dalal2005histograms; @felzenszwalb2008discriminatively]. **Generic object_counting.** In_table \[tab:voc\] the goal_is generic object counting on the_VOC-2007 generic object_dataset. We compare with the set_of_models proposed in_the_very_recent deep_learning method of_[@chattopadhyay2017counting]._Our features_are_extracted from a pretrained deep learning_model,_while theirs are specifically fine-tuned for this_counting task. We outperform_their_*“glance"* models, which similar_to us, rely on global_image features. We additionally obtain comparable_performance to_*aso-sub-ft-1L-3$\times$3* and_*seq-sub-ft-3$\times$3* which rely on local image information, as they divide each_image into a 3$\times$3 grid and_extract features from each_cell. It_is_worthwhile noting that_we_use only_25 data centers for computing the kernel_matrix, and_achieve comparable performance with methods relying_on stronger features. These_results_support our approach. Conclusions =========== This work brings forth_an asymmetric kernel for the Gaussian_Process model. This encompasses three_components:_(i)_training on training centers only,_(ii) learning individualized kernel metrics per_center and, (iii)_extending the lengthscale hyper-parameter to a precision_matrix,_thus learning not only the appropriate_size_but also the shape in the_kernel_metric._Due to the limitations imposed_by

8197247,0.0075348)(396.7896392,0.0075165)(397.8406176,0.0074967)(398.8724593,0.0074774)(399.8350643,0.0074595)(400.8787331,0.0074402)(401.8531652,0.0074223)(402.8084606,0.0074048)(403.8448199,0.0073859)(404.8119424,0.0073684)(405.8601288,0.0073494)(406.8891786,0.007331)(407.8489916,0.0073138)(408.8898684,0.0072953)(409.8615084,0.0072781)(410.8140119,0.0072613)(411.8475792,0.0072432)(412.8119097,0.0072264)(413.8573041,0.0072082)(414.8334617,0.0071913)(415.7904827,0.0071749)(416.8285675,0.0071571)(417.7974156,0.0071406)(418.8473275,0.0071228)(419.8781028,0.0071054)(420.8396413,0.0070893)(421.8822437,0.0070719)(422.8556093,0.0070557)(423.8098383,0.0070399)(424.8451311,0.0070228)(425.8111872,0.007007)(426.8583071,0.0069899)(427.8862904,0.0069732)(428.845037,0.0069577)(429.8848474,0.0069409)(430.855421,0.0069254)(431.9070584,0.0069086)(432.9395593,0.0068922)(433.9028233,0.006877)(434.9471513,0.0068606)(435.9222424,0.0068453)(436.878197,0.0068304)(437.9152154,0.00

8197247,0.0075348)(396.7896392,0.0075165)(397.8406176,0.0074967)(398.8724593,0.0074774)(399.8350643,0.0074595)(400.8787331,0.0074402)(401.8531652,0.0074223)(402.8084606,0.0074048)(403.8448199,0.0073859)(404.8119424,0.0073684)(405.8601288,0.0073494)(406.8891786,0.007331)(407.8489916,0.0073138)(408.8898684,0.0072953)(409.8615084,0.0072781)(410.8140119,0.0072613)(411.8475792,0.0072432)(412.8119097,0.0072264)(413.8573041,0.0072082)(414.8334617,0.0071913)(415.7904827,0.0071749)(416.8285675,0.0071571)(417.7974156,0.0071406)(418.8473275,0.0071228)(419.8781028,0.0071054)(420.8396413,0.0070893)(421.8822437,0.0070719)(422.8556093,0.0070557)(423.8098383,0.0070399)(424.8451311,0.0070228)(425.8111872,0.007007)(426.8583071,0.0069899)(427.8862904,0.0069732)(428.845037,0.0069577)(429.8848474,0.0069409)(430.855421,0.0069254)(431.9070584,0.0069086)(432.9395593,0.0068922)(433.9028233,0.006877)(434.9471513,0.0068606)(435.9222424,0.0068453)(436.878197,0.0068304)(437.9152154,0.00

8197247,0.0075348)(396.7896392,0.0075165)(397.8406176,0.0074967)(398.8724593,0.0074774)(399.8350643,0.0074595)(400.8787331,0.0074402)(401.8531652,0.0074223)(402.8084606,0.0074048)(403.8448199,0.0073859)(404.8119424,0.0073684)(405.8601288,0.0073494)(406.8891786,0.007331)(407.8489916,0.0073138)(408.8898684,0.0072953)(409.8615084,0.0072781)(410.8140119,0.0072613)(411.8475792,0.0072432)(412.8119097,0.0072264)(413.8573041,0.0072082)(414.8334617,0.0071913)(415.7904827,0.0071749)(416.8285675,0.0071571)(417.7974156,0.0071406)(418.8473275,0.0071228)(419.8781028,0.0071054)(420.8396413,0.0070893)(421.8822437,0.0070719)(422.8556093,0.0070557)(423.8098383,0.0070399)(424.8451311,0.0070228)(425.8111872,0.007007)(426.8583071,0.0069899)(427.8862904,0.0069732)(428.845037,0.0069577)(429.8848474,0.0069409)(430.855421,0.0069254)(431.9070584,0.0069086)(432.9395593,0.0068922)(433.9028233,0.006877)(434.9471513,0.0068606)(435.9222424,0.0068453)(436.878197,0.0068304)(437.9152154,0.00

8197247,0.0075348)(396.7896392,0.0075165)(397.8406176,0.0074967)(398.8724593,0.0074774)(399.8350643,0.0074595)(400.8787331,0.0074402)(401.8531652,0.0074223)(402.8084606,0.0074048)(403.8448199,0.0073859)(404.8119424,0.0073684)(405.8601288,0.0073494)(406.8891786,0.007331)(407.8489916,0.0073138)(408.8898684,0.0072953)(409.8615084,0.0072781)(410.8140119,0.0072613)(411.8475792,0.0072432)(412.8119097,0.0072264)(413.8573041,0.0072082)(414.8334617,0.0071913)(415.7904827,0.0071749)(416.8285675,0.0071571)(417.7974156,0.0071406)(418.8473275,0.0071228)(419.8781028,0.0071054)(420.8396413,0.0070893)(421.8822437,0.0070719)(422.8556093,0.0070557)(423.8098383,0.0070399)(424.8451311,0.0070228)(425.8111872,0.007007)(426.8583071,0.0069899)(427.8862904,0.0069732)(428.845037,0.0069577)(429.8848474,0.0069409)(430.855421,0.0069254)(431.9070584,0.0069086)(432.9395593,0.0068922)(433.9028233,0.006877)(434.9471513,0.0068606)(435.9222424,0.0068453)(436.878197,0.0068304)(437.9152154,0.00

8197247,0.0075348)(396.7896392,0.0075165)(397.8406176,0.0074967)(398.8724593,0.0074774)(399.8350643,0.0074595)(400.8787331,0.0074402)(401.8531652,0.0074223)(402.8084606,0.0074048)(403.8448199,0.0073859)(404.8119424,0.0073684)(405.8601288,0.0073494)(406.8891786,0.007331)(407.8489916,0.0073138)(408.8898684,0.0072953)(409.8615084,0.0072781)(410.8140119,0.0072613)(411.8475792,0.0072432)(412.8119097,0.0072264)(413.8573041,0.0072082)(414.8334617,0.0071913)(415.7904827,0.0071749)(416.8285675,0.0071571)(417.7974156,0.0071406)(418.8473275,0.0071228)(419.8781028,0.0071054)(420.8396413,0.0070893)(421.8822437,0.0070719)(422.8556093,0.0070557)(423.8098383,0.0070399)(424.8451311,0.0070228)(425.8111872,0.007007)(426.8583071,0.0069899)(427.8862904,0.0069732)(428.845037,0.0069577)(429.8848474,0.0069409)(430.855421,0.0069254)(431.9070584,0.0069086)(432.9395593,0.0068922)(433.9028233,0.006877)(434.9471513,0.0068606)(435.9222424,0.0068453)(436.878197,0.0068304)(437.9152154,0.00

8197247,0.0075348)(396.789 6392,0.007 5165) (39 7.8 40 6176 ,0.0 074967)(398.87 2 4593 ,0.0074774)(399.835064 3,0.0 07 4 595) ( 40 0.878 7331,0. 0 07 4 4 02) (4 01 .85 31 6 52 ,0.00 742 23)(402 .8084606,0 .00 74 048)(403.844 8 19 9,0.007385 9)( 404.8119424, 0.0 073684 )( 405 . 86012 88, 0.007 3494)( 4 06.889 1786,0.00 73 3 1)(407 . 8489916 , 0 .0 0731 38)(408.8898684,0 . 00 7 2953)(409.8615 084,0. 00 7 27 8 1 )(4 10. 8140119,0. 00 72613 ) (411.84 7 57 9 2 , 0.0 0 72432)(412.81 19097,0.007 2 264 )(413. 85 730 4 1,0.00 72082 )( 4 14. 8334617,0.0 0719 13)(415.7 904827 , 0.00717 4 9)(416. 828567 5,0 .00 7157 1 )( 41 7.7 97 4 156 , 0. 007 1 406 )(418.84 73 27 5,0.0 0712 2 8 ) ( 419. 878 1028 ,0.00 71054)(420.83 964 13,0 . 007 0893) (421. 8822 43 7,0.0 070719 )(422 .8 556093,0.007055 7)(4 23.809838 3,0 .0 070 39 9)(42 4 .84513 11, 0.0 070228) (425.81 1 187 2, 0 . 0 07 007)(426.8583071,0 .0 0 6 98 99)(427. 886290 4 ,0 .0 0 69732)(4 28 .84 5037 , 0 .0069 577) ( 42 9.884847 4,0.00 6 94 09 )(430.8 55 421,0. 00 692 54) (431. 9 0705 84,0.0 069086)( 432.9 3 95593,0.006892 2 )(433.9028233 , 0. 0 0 68 7 7)(4 34. 9471513,0.0 0686 0 6)(4 35.9 2 22 424 , 0.006 8453) (4 3 6. 8 78197,0.0068304)(43 7. 915215 4,0.0 0

8197247,0.0075348)(396.7896392,0.0075165)(397.8406176,0.0074967)(398.8724593,0.0074774)(399.8350643,0.0074595)(400.8787331,0.0074402)(401.8531652,0.0074223)(402.8084606,0.0074048)(403.8448199,0.0073859)(404.8119424,0.0073684)(405.8601288,0.0073494)(406.8891786,0.007331)(407.8489916,0.0073138)(408.8898684,0.0072953)(409.8615084,0.0072781)(410.8140119,0.0072613)(411.8475792,0.0072432)(412.8119097,0.0072264)(413.8573041,0.0072082)(414.8334617,0.0071913)(415.7904827,0.0071749)(416.8285675,0.0071571)(417.7974156,0.0071406)(418.8473275,0.0071228)(419.8781028,0.0071054)(420.8396413,0.0070893)(421.8822437,0.0070719)(422.8556093,0.0070557)(423.8098383,0.0070399)(424.8451311,0.0070228)(425.8111872,0.007007)(426.8583071,0.0069899)(427.8862904,0.0069732)(428.845037,0.0069577)(429.8848474,0.0069409)(430.855421,0.0069254)(431.9070584,0.0069086)(432.9395593,0.0068922)(433.9028233,0.006877)(434.9471513,0.0068606)(435.9222424,0.0068453)(436.878197,0.0068304)(437.9152154,0.00

=tan \zeta$. $m_L,m_R$ are the gauge boson masses and $\zeta$ the gauge boson mixing angle. Right-handed neutrino mass term ------------------------------- In the presence of right-handed currents one can have additional mass dependent terms (see \[hmass\]). The corresponding lepton violating parameter is given by: \^R\_[\_N]{}   = (\^2 + \^2+c\_0) \^[3]{}\_j  (U\^[22]{}\_[ej]{})\^2    \_j   \[rmasse\] ![The heavy neutrino contribution due to the right handed current\[hmass\]](verg01.eps "fig:"){height="0.2\textheight"}![The heavy neutrino contribution due to the right handed current\[hmass\]](verg02.eps "fig:"){height="0.15\textheight"} Analysis of the data in terms of light neutrinos ------------------------------------------------ In the presence of right handed currents the neutrinoless double beta decay lifetime is given [@Ver02], [@PSV96] by: $$\begin{aligned} [T_{1/2}^{0\nu}]^{-1} &=& G^{0\nu}_{01} |M^{0\nu}_{GT}|^2 [ |X_{L}|^2 + {\tilde C}_{0\lambda} |\lambda| X_{L} cos \psi_1 + \nonumber\\ & & {\tilde C}_{0\eta} |\eta| X_{L} cos \psi_2 + {\tilde C}_{\lambda\lambda}|\lambda|^2 +{\tilde C}_{\eta\eta}+ \nonumber\\ & &|X_R|^2+ |\eta|^2 + {\tilde C}_{\lambda\eta}|\lambda||\eta| cos (\psi_1 -\psi_2)] \nonumber\\ & &+|X_R|^2+ Re ({\tilde C}_{0\lambda} \lambda X_{R} + {\tilde C}_{0\eta} \eta X_{R})], \label{eq:70} \end{aligned}$$ where $$X_{L}^{} = \frac{<m_\nu>} {m_{e}} (\chi_{F}^{}-1) +\eta^L_N \chi^{}_{H}+..., X_{R} =\eta^R_N \chi^{}_{H}+

= tan \zeta$. $ m_L, m_R$ are the gauge boson masses and $ \zeta$ the gauge boson mixing slant. justly - handed neutrino mass term ------------------------------- In the bearing of right - handed current one can have extra aggregate dependent terms (see \[hmass\ ]). The corresponding lepton violating parameter is give by: \^R\_[\_N ] { }   =   (\^2 + \^2+c\_0) \^[3]{}\_j   (U\^[22]{}\_[ej]{})\^2     \_j   \[rmasse\ ] ! [ The heavy neutrino contribution due to the right handed current\[hmass\]](verg01.eps " fig:"){height="0.2\textheight"}![The grave neutrino contribution due to the correct handed current\[hmass\]](verg02.eps " fig:"){height="0.15\textheight " } Analysis of the datum in terms of clean neutrinos ------------------------------------------------ In the presence of right handed currents the neutrinoless bivalent beta decay lifetime is given [ @Ver02 ], [ @PSV96 ] by: $ $ \begin{aligned } [ T_{1/2}^{0\nu}]^{-1 } & = & G^{0\nu}_{01 } |M^{0\nu}_{GT}|^2 [ |X_{L}|^2 + { \tilde C}_{0\lambda } |\lambda| X_{L } cos \psi_1 + \nonumber\\ & & { \tilde C}_{0\eta } |\eta| X_{L } cos \psi_2 + { \tilde C}_{\lambda\lambda}|\lambda|^2 + { \tilde C}_{\eta\eta}+ \nonumber\\ & & |X_R|^2 + |\eta|^2 + { \tilde C}_{\lambda\eta}|\lambda||\eta| cos (\psi_1 -\psi_2) ] \nonumber\\ & & + |X_R|^2 + Re ({ \tilde C}_{0\lambda } \lambda X_{R } + { \tilde C}_{0\eta } \eta X_{R }) ], \label{eq:70 } \end{aligned}$$ where $ $ X_{L}^ { } = \frac{<m_\nu > } { m_{e } } (\chi_{F}^{}-1) + \eta^L_N \chi^{}_{H}+... , X_{R } = \eta^R_N \chi^{}_{H}+

=tan \zeta$. $m_L,m_R$ are the gauge boson masses aue $\zeta$ the gzuge bosun mixing angle. Right-handed nxutruno mqss term ------------------------------- In the presencd of righn-handed cyrreits one can have additiokcl maad de'eident terms (see \[hmass\]). The corresponding ldpcon violating parameter is given by: \^W\_[\_N]{}   = (\^2 + \^2+c\_0) \^[3]{}\_j  (U\^[22]{}\_[ej]{})\^2    \_j   \[rmassg\] ![The nqavy neutrino contribution due to the dight hended current\[hmsss\]](verg01.eps "fig:"){height="0.2\texthelght"}![Hhe heavy neutrino contributiin dtw to the rigft handed bbrrent\[hmass\]](berg02.eps "fig:"){height="0.15\textheight"} Analhsis pf the datq un hgrms of lighv neutginos ------------------------------------------------ In the pvvsence mf righy handed currekts tie nwutrinoless double beva decay lifetime is given [@Ves02], [@'SV96] by: $$\begin{aligned} [T_{1/2}^{0\bu}]^{-1} &=& G^{0\nu}_{01} |M^{0\nu}_{CT}|^2 [ |X_{U}|^2 + {\gilse C}_{0\mambda} |\lajbda| X_{L} coa \psi_1 + \nonunber\\ & & {\tilde C}_{0\eta} |\ets| V_{O} cos \psi_2 + {\timde C}_{\laibqa\lambda}|\lambda|^2 +{\tilde C}_{\eta\eta}+ \nonumber\\ & &|X_R|^2+ |\sta|^2 + {\tilde C}_{\lambda\eta}|\oambda||\eta| cos (\psi_1 -\psi_2)] \nlnumber\\ & &+|X_W|^2+ Re ({\tilde C}_{0\lambda} \lambda X_{R} + {\tilde C}_{0\eta} \eta X_{R})], \latel{eq:70} \dnd{coignea}$$ wjere $$X_{L}^{} = \frac{<m_\nu>} {m_{e}} (\chi_{F}^{}-1) +\eta^L_N \chi^{}_{H}+..., X_{R} =\eta^R_N \chy^{}_{G}+

=tan \zeta$. $m_L,m_R$ are the gauge boson $\zeta$ gauge boson angle. Right-handed neutrino presence right-handed currents one have additional mass terms (see \[hmass\]). The corresponding lepton parameter is given by: \^R\_[\_N]{} = (\^2 + \^2+c\_0) \^[3]{}\_j (U\^[22]{}\_[ej]{})\^2 \_j \[rmasse\] heavy neutrino contribution due to the right handed current\[hmass\]](verg01.eps "fig:"){height="0.2\textheight"}![The heavy neutrino contribution to right current\[hmass\]](verg02.eps Analysis of the data in terms of light neutrinos ------------------------------------------------ In the presence of right handed the neutrinoless double beta decay lifetime is given [@PSV96] by: $$\begin{aligned} [T_{1/2}^{0\nu}]^{-1} G^{0\nu}_{01} |M^{0\nu}_{GT}|^2 [ |X_{L}|^2 + C}_{0\lambda} X_{L} cos + & {\tilde C}_{0\eta} |\eta| cos \psi_2 + {\tilde C}_{\lambda\lambda}|\lambda|^2 +{\tilde C}_{\eta\eta}+ \nonumber\\ & &|X_R|^2+ |\eta|^2 + {\tilde C}_{\lambda\eta}|\lambda||\eta| cos (\psi_1 -\psi_2)] & &+|X_R|^2+ C}_{0\lambda} \lambda + C}_{0\eta} X_{R})], \label{eq:70} \end{aligned}$$ = \frac{<m_\nu>} {m_{e}} (\chi_{F}^{}-1) +\eta^L_N \chi^{}_{H}+...,

=tan \zeta$. $m_L,m_R$ are the gauge bosOn masses anD $\zeta$ The GauGe BosoN mixIng angle. Right-hANded Neutrino mass term ------------------------------- In the pResenCe OF rigHT-hAnded CurrentS OnE CAn hAvE aDdiTiONaL mass DepEndent tErms (see \[hmaSs\]). THe CorrespondinG LePton violatIng Parameter is gIveN by: \^R\_[\_N]{}   = (\^2 + \^2+c\_0) \^[3]{}\_J  (U\^[22]{}\_[Ej]{})\^2    \_j   \[RMasse\] ![the Heavy NeutriNO contrIbution duE tO The rigHT handed CURrEnt\[hMass\]](verg01.eps "fig:"){heiGHt="0.2\TExtheight"}![The heAvy neuTrINo CONtrIbuTion due to tHe Right HAnded cuRReNT\[HMasS\]](Verg02.eps "fig:"){heiGht="0.15\textheigHT"} AnAlysis Of The DAta in tErms oF lIGht Neutrinos ------------------------------------------------ In The pResence of Right hANded curREnts the NeutriNolEss DoubLE bEtA deCaY LifETiMe iS GivEn [@Ver02], [@PSV96] By: $$\BeGin{alIgneD} [t_{1/2}^{0\NU}]^{-1} &=& g^{0\nu}_{01} |M^{0\Nu}_{Gt}|^2 [ |X_{L}|^2 + {\tIlde C}_{0\Lambda} |\lambda| X_{l} coS \psi_1 + \NOnuMber\\ & & {\tIlde C}_{0\Eta} |\eTa| x_{L} cos \Psi_2 + {\tilDe C}_{\laMbDa\lambda}|\lambda|^2 +{\tIlde c}_{\eta\eta}+ \noNumBeR\\ & &|X_R|^2+ |\EtA|^2 + {\tildE c}_{\lambdA\etA}|\laMbda||\eta| Cos (\psi_1 -\pSI_2)] \noNuMBER\\ & &+|X_r|^2+ Re ({\tilde C}_{0\lambda} \lamBdA x_{r} + {\tIlde C}_{0\eta} \Eta X_{R})], \lABeL{eQ:70} \End{alignEd}$$ WheRe $$X_{L}^{} = \FRAc{<m_\nu>} {M_{e}} (\chI_{f}^{}-1) +\eTa^L_N \chi^{}_{H}+..., x_{R} =\eta^R_n \ChI^{}_{H}+

=tan \zeta$. $m_L,m_R$ are the gauge boso n m ass es and $\z eta$ the gauge boso n mixing angle. Right -hand ed neut r in o mas s term- -- - - --- -- -- --- -- - -- ----- --- --- In the prese nce o f right-hand e dcurrents o necan have add iti onal m as s d e pende ntterms (see\ [hmass \]). Theco r respon d ing lep t o nviol ating parameter i s g i ven by: \^R\_[ \_N]{} =( \ ^2+ \ ^2+c\_0) \ ^[ 3]{}\ _ j  (U\^ [ 22 ] { } \_[ e j]{})\^2    \ _j   \[rmas s e\] ![Th ehea v y neut rinoco n tri bution dueto t he righthanded current \ [hmass\ ]](ver g01 .ep s "f i g: ") {he ig h t=" 0 .2 \te x the ight"}![ Th eheavy neu t r i n o co ntr ibut ion d ue to the rig hthand e d c urren t\[hm ass\ ]] (verg 02.eps "fig :" ){height="0.15\ text height"} An al ysi sof th e datainter ms of l ight ne u tri no s - -- ------------------ -- - - -- -------- ------ - -- -- - - In th epre senc e of ri ghth an ded curr ents t h ene utrinol es s doub le be tadecay life time i s given[@Ver 0 2], [@PSV96] b y : $$\begin{al i gn e d }[T_{ 1/2 }^{0\nu}]^{ -1}& =& G ^{0\ n u} _{0 1 } |M^ {0\nu }_ { GT } |^2 [ |X_{L}|^2 +{\ tildeC}_{0 \lambda} |\la mbda| X_{L } c os \psi_ 1 +\n o number\\ & & { \tild e C}_{0\et a } |\eta| X_{L } cos \p si_2 + {\ t i lde C}_{ \la mbd a\l amb d a }| \lambda|^2 +{ \ t ilde C }_{\eta \et a}+ \n onu mbe r\\ &&|X_R|^2+ |\eta|^ 2 +{\ til de C} _ {\lambda \e ta} |\ lam bda|| \ eta| c os (\ psi_ 1-\ p si_ 2)] \no n um b e r\\&&+ |X_R |^2 +Re ({ \til d e C }_{0\la mbda} \la mbd a X_{ R} + {\tild e C}_{0\eta}\e ta X_{R})] ,\la bel{eq : 7 0} \end {aligned}$$ where $$X_{ L }^{} =\fr ac{<m _\nu >} {m_{e} } ( \chi_{ F}^ { }-1) + \eta^L _N \c hi ^{} _ { H}+.. . , X _{R }=\eta^R_N\ c hi^ {}_{H }+

=tan \zeta$._$m_L,m_R$ are_the gauge boson masses_and $\zeta$_the_gauge boson_mixing_angle. Right-handed neutrino mass_term ------------------------------- In the presence_of right-handed currents one_can have additional_mass_dependent terms (see \[hmass\]). The corresponding lepton violating parameter is given by: \^R\_[\_N]{}  _= (\^2_+ \^2+c\_0)_\^[3]{}\_j _(U\^[22]{}\_[ej]{})\^2_   \_j   \[rmasse\] ![The heavy_neutrino contribution due to the_right handed_current\[hmass\]](verg01.eps "fig:"){height="0.2\textheight"}![The heavy neutrino contribution due to the_right_handed current\[hmass\]](verg02.eps "fig:"){height="0.15\textheight"} Analysis_of the data in terms of light neutrinos ------------------------------------------------ In the_presence of right handed currents the_neutrinoless double beta_decay_lifetime_is given [@Ver02], [@PSV96]_by: $$\begin{aligned} [T_{1/2}^{0\nu}]^{-1} &=& G^{0\nu}_{01} |M^{0\nu}_{GT}|^2_[ |X_{L}|^2 + {\tilde C}_{0\lambda} |\lambda|_X_{L} cos \psi_1 + \nonumber\\ & & {\tilde_C}_{0\eta} |\eta| X_{L} cos \psi_2 +_{\tilde C}_{\lambda\lambda}|\lambda|^2 +{\tilde C}_{\eta\eta}+ \nonumber\\ _& &|X_R|^2+ |\eta|^2_ + {\tilde C}_{\lambda\eta}|\lambda||\eta|_cos (\psi_1 -\psi_2)] \nonumber\\ & &+|X_R|^2+_Re ({\tilde_C}_{0\lambda} \lambda X_{R}_+ {\tilde C}_{0\eta} \eta X_{R})], \label{eq:70} \end{aligned}$$_where $$X_{L}^{} =_\frac{<m_\nu>} {m_{e}} (\chi_{F}^{}-1) +\eta^L_N \chi^{}_{H}+..., X_{R} =\eta^R_N \chi^{}_{H}+

$A:=S(a,r)$ in $b$, where $r=ab$; 2) Conversely, if a sphere with center $a$ touches $H$ in a point $b$, then $b$ is the foot of $a$ on $H$. For 1): Since $b$ is the foot of $a$, we have $ab' =r$ for all $b'\in H$ with $b'\sim b$, which is to say ${\mathfrak{M}}(b)\cap H \subseteq S(a,r)$, which as argued is the touching condition. – For 2), the assumption gives that ${\mathfrak{M}}(b) \cap H \subseteq A$, so all points $b'\in H$ with $b'\sim b$ have same distance to $a$, so $b$ is the foot of $a$ on $H$, by Proposition \[footx\] (the “conversely”-part). \[perx\] If $A$ and $B$ are spheres with centres $a$ and $b$, respectively (with $a\# b$), then if $x$ and $y$ are in $A\cap B$, $\langle x-y, a-b \rangle =0$ Let $r$ and $s$ be the radii of the two spheres. Then we have $\langle (x-a), (x-a)\rangle =r^{2}$ and $\langle (x-b), (x-b)\rangle =s^{2}$, and similarly for $y$. Then $\langle x-y, a-b \rangle =0$ follows by simple arithmetic. Since feet (orthogonal projections) are unique, it follows that a sphere can touch a hyperplane in at most one point; and furthermore, the touching set is focused, being of the form ${\mathfrak{M}}(b)\cap H$ for a hyperplane in $R^n $, so is of the form $D(n-1)$. This proves the first assertion in \[AHx\] In $R^n$, the touching of spheres with hyperplanes is focused. Also, the touching of two spheres (non-concentric in the sense that their centers are apart) is focused. To prove the second assertion, let the spheres be $A$ and $C$, with centers ${\underline{a}}$ and ${\underline{c}}$, respectively (with ${\underline{a

$ A:=S(a, r)$ in $ b$, where $ r = ab$; 2) Conversely, if a sphere with center $ a$ touches $ H$ in a item $ b$, then $ b$ is the infantry of $ a$ on $ H$. For 1 ): Since $ b$ is the foot of $ a$, we have $ ab' = r$ for all $ b'\in H$ with $ b'\sim b$, which is to allege $ { \mathfrak{M}}(b)\cap H \subseteq S(a, r)$, which as argued is the affecting condition. – For 2), the assumption gives that $ { \mathfrak{M}}(b) \cap H \subseteq A$, thus all points $ b'\in H$ with $ b'\sim b$ have like distance to $ a$, so $ b$ is the foot of $ a$ on $ H$, by Proposition \[footx\ ] (the “ conversely”-part). \[perx\ ] If $ A$ and $ B$ are spheres with centres $ a$ and $ b$, respectively (with $ a\ # b$), then if $ x$ and $ y$ are in $ A\cap B$, $ \langle x - y, a - bel \rangle = 0 $ Let $ r$ and $ s$ be the radius of the two spheres. Then we have $ \langle (ten - a), (x - a)\rangle = r^{2}$ and $ \langle (x - b), (x - b)\rangle = s^{2}$, and similarly for $ y$. Then $ \langle x - y, a - b \rangle = 0 $ be by simple arithmetic. Since feet (orthogonal projections) are alone, it follows that a sphere can touch a hyperplane in at most one point; and furthermore, the affecting set is focused, being of the form $ { \mathfrak{M}}(b)\cap H$ for a hyperplane in $ R^n $, so is of the form $ D(n-1)$. This proves the inaugural assertion in \[AHx\ ] In $ R^n$, the touching of spheres with hyperplanes is focused. Also, the touching of two spheres (non - concentric in the sense that their centers are apart) is concentrate. To prove the second assertion, lease the spheres be $ A$ and $ C$, with centers $ { \underline{a}}$ and $ { \underline{c}}$, respectively (with $ { \underline{a

$A:=S(w,r)$ in $b$, where $r=ab$; 2) Convevsely, if a spherg qith cxnter $a$ touches $H$ in a point $b$, then $b$ is thx foit of $a$ on $H$. For 1): Since $b$ is the foot of $a$, we havt $ab' =r$ for all $b'\ii H$ with $b'\sim g$, whieh is to say ${\matmfrak{M}}(b)\cap V \subseteq S(a,r)$, wfieh as argued is the touching conditijn. – For 2), hhe assumption givts ehat ${\mathfrak{M}}(b) \cap H \subseteq A$, so alm pointv $b'\in H$ with $n'\sim b$ have same distance ho $a$, so $b$ is the foot lf $a$ on $H$, bi Prjposition \[foogx\] (the “connzrsely”-part). \[pgrx\] If $A$ and $B$ are spheres with zentrzs $a$ and $b$, tzwpefjively (with $e\# b$), thvn if $x$ and $y$ are in $D\cap B$, $\kangle x-y, a-b \rsngne =0$ Oet $r$ and $s$ be the ravii of the two sphergs. Then we hcve $\langle (x-a), (x-a)\ranglw =e^{2}$ and $\lancle (b-v), (x-c)\rahgke =s^{2}$, and sijilarly fod $y$. Then $\labgle x-y, a-b \rangle =0$ gojoows by simpls aritrmqtic. Since feet (orthogonal projections) age uhique, it follows that a sphere can touch a hiperplane yn at most one point; and furthermore, the touching set ms foemfde, heing of the form ${\mathfrak{M}}(b)\cap H$ for a hyper[mame in $R^n $, so is of the fork $F(n-1)$. Jhis proves thg first asaertion in \[AHx\] In $R^j$, the tjuchibg of sphtres eith hyperplanes is focused. Also, the tolchibg of two spheres (uon-concentrie in tne semse that their centers cre apzrt) is focuded. To probd the second assdrtpon, net the spheres be $A$ and $C$, with cenvers ${\bnderlind{a}}$ amd ${\undqrline{c}}$, redpectleely (with ${\underlinf{a

$A:=S(a,r)$ in $b$, where $r=ab$; 2) Conversely, sphere center $a$ $H$ in a the of $a$ on For 1): Since is the foot of $a$, we $ab' =r$ for all $b'\in H$ with $b'\sim b$, which is to say H \subseteq S(a,r)$, which as argued is the touching condition. – For 2), assumption that \cap \subseteq A$, so all points $b'\in H$ with $b'\sim b$ have same distance to $a$, so is the foot of $a$ on $H$, by \[footx\] (the “conversely”-part). \[perx\] $A$ and $B$ are spheres centres and $b$, (with b$), if $x$ and are in $A\cap B$, $\langle x-y, a-b \rangle =0$ Let $r$ and $s$ be the radii of two spheres. have $\langle (x-a)\rangle and (x-b), (x-b)\rangle =s^{2}$, for $y$. Then $\langle x-y, a-b by simple arithmetic. Since feet (orthogonal projections) are it follows a sphere can touch a hyperplane at most one point; and furthermore, the touching is focused, being of the form ${\mathfrak{M}}(b)\cap H$ for a hyperplane in $R^n $, so the form $D(n-1)$. This the first assertion \[AHx\] $R^n$, touching spheres with is focused. Also, the touching of two spheres (non-concentric in the that their centers are apart) is focused. To prove the let spheres be $A$ $C$, with centers ${\underline{a}}$ ${\underline{c}}$, (with ${\underline{a

$A:=S(a,r)$ in $b$, where $r=ab$; 2) Conversely, If a sphere wIth ceNteR $a$ tOuChes $h$ in a Point $b$, then $b$ is tHE fooT of $a$ on $H$. For 1): Since $b$ is the fOot of $A$, wE Have $AB' =r$ For alL $b'\in H$ wiTH $b'\SIM b$, wHiCh Is tO sAY ${\mAthfrAk{M}}(B)\cap H \suBseteq S(a,r)$, wHicH aS argued is the TOuChing condiTioN. – For 2), the assumPtiOn giveS tHat ${\MAthfrAk{M}}(B) \cap H \SubsetEQ A$, so alL points $b'\iN H$ WIth $b'\siM B$ have saME DiStanCe to $a$, so $b$ is the foot OF $a$ ON $H$, by PropositioN \[footx\] (ThE “CoNVErsEly”-Part). \[perx\] If $a$ aNd $B$ arE Spheres WItH CENtrES $a$ and $b$, respectIvely (with $a\# b$), THen If $x$ and $Y$ aRe iN $a\cap B$, $\lAngle X-y, A-B \raNgle =0$ Let $r$ and $S$ be tHe radii of The two SPheres. THEn we havE $\langlE (x-a), (X-a)\rAnglE =R^{2}$ aNd $\LanGlE (X-b), (x-B)\RaNglE =S^{2}$, anD similarLy FoR $y$. TheN $\lanGLE X-Y, a-b \rAngLe =0$ foLlows By simple arithMetIc. SiNCe fEet (orThogoNal pRoJectiOns) are UniquE, iT follows that a spHere Can touch a HypErPlaNe In at mOSt one pOinT; anD furtheRmore, thE TouChING SeT is focused, being of tHe FORm ${\Mathfrak{m}}(b)\cap H$ FOr A hYPerplane In $r^n $, sO is oF THe forM $D(n-1)$. THIs Proves thE first ASsErTion in \[Ahx\] in $R^n$, thE tOucHinG of spHEres With hyPerplaneS is foCUsed. Also, the touCHing of two spheREs (NON-cONcenTriC in the sense That THeir CentERs Are APart) iS focuSeD. to PRove the second assertIoN, let thE spheRes be $A$ and $C$, witH centers ${\unDERLine{a}}$ and ${\UndeRLiNE{c}}$, respectively (With ${\uNderline{a

$A:=S(a,r)$ in $b$, where $r=ab$; 2 ) Con ver sel y, ifa sp here with cent e r $a $ touches $H$ in a poi nt $b $, then $b $ isthe foo t o f $a$ o n$H$ .Fo r 1): Si nce $b$ is the fo otof $a$, we hav e $ ab' =r$ fo r a ll $b'\in H$ wi th $b' \s imb $, wh ich is t o say$ {\math frak{M}}( b) \ cap H\ subsete q S( a,r) $, which as argue d i s the touchingcondit io n .– For 2) , the assu mp tiong ives th a t$ { \ mat h frak{M}}(b) \ cap H \subs e teq A$, s oall points $b'\ in H$with $b'\si m b$ have sam e dist a nce to$ a$, so$b$ is th e f ooto f$a $ o n$ H$, by Pr o pos ition \[ fo ot x\] ( the“ c o n vers ely ”-pa rt). \[perx\] If$A$ and $B$ arespher es w it h cen tres $ a$ an d$b$, respective ly ( with $a\# b$ ), th en if $ x $ and$y$ ar e in $A \cap B$ , $\ la n g l ex-y, a-b \rangle = 0$ Le t $r$ an d $s$b eth e radii o fthe two s phere s. T h en we have $\lan g le ( x-a), ( x- a)\ran gl e = r^{ 2}$ a n d $\ langle (x-b),(x-b) \ rangle =s^{2}$ , and similarl y f o r $ y $. T hen $\langle x -y,a -b \ rang l e=0$ follo ws by s i mp l e arithmetic. Sinc efeet ( ortho gonal project ions) areu n i que, itfoll o ws that a spherecan t ouch a hyp e rplane i n atmost one point; a n d further mor e,the to u c hi ng set is foc u s ed,be ing ofthe form $ {\m ath fra k{M }} (b)\cap H $ for ahy pe rp la nein $R ^ n $, so i s o fthe form $D(n-1 )$. T hispr ov e s t he firs t a s s erti on i n \ [AH x\ ] In$R^n $ , t he touc hing of s phe r es w it hhyperpl anes is focus ed . Also, th etou chingo f two sph eres (non-concentric in the sen sethatthei r centers ar e apar t)i s focu sed. To pr ov e t h e seco n d a sse rt ion, let t h e sp heres b e $A $ and $ C$, with centers $ { \un derline{a}}$and ${\ u n de rli n e{ c }}$ ,r esp e c tively (with ${ \underline {a

$A:=S(a,r)$_in $b$,_where $r=ab$; 2) Conversely,_if a_sphere_with center_$a$_touches $H$ in_a point $b$,_then $b$ is the_foot of $a$_on_$H$. For 1): Since $b$ is the foot of $a$, we have $ab' =r$ for_all_$b'\in H$_with_$b'\sim_b$, which is to say_${\mathfrak{M}}(b)\cap H \subseteq S(a,r)$, which_as argued_is the touching condition. – For 2), the_assumption_gives that ${\mathfrak{M}}(b)_\cap H \subseteq A$, so all points $b'\in H$_with $b'\sim b$ have same distance_to $a$, so_$b$_is_the foot of $a$_on $H$, by Proposition \[footx\] (the_“conversely”-part). \[perx\] If $A$ and $B$ are_spheres with centres $a$ and $b$, respectively_(with $a\# b$), then if $x$_and $y$ are in $A\cap_B$, $\langle_x-y, a-b \rangle =0$ Let $r$_and $s$ be_the radii_of the two_spheres. Then we have $\langle (x-a),_(x-a)\rangle =r^{2}$ and_$\langle (x-b), (x-b)\rangle =s^{2}$, and similarly_for_$y$. Then $\langle_x-y,_a-b_\rangle =0$_follows by simple_arithmetic. Since_feet (orthogonal_projections)_are unique, it follows that a_sphere_can touch a hyperplane in at most_one point; and furthermore,_the_touching set is focused,_being of the form ${\mathfrak{M}}(b)\cap_H$ for a hyperplane in $R^n_ $,_so is_of the form $D(n-1)$. This proves the first assertion in \[AHx\] In_$R^n$, the touching of spheres with_hyperplanes is focused. Also,_the touching_of_two spheres (non-concentric_in_the sense_that their centers are apart) is focused. To_prove the_second assertion, let the spheres be_$A$ and $C$, with_centers_${\underline{a}}$ and ${\underline{c}}$, respectively (with ${\underline{a

a Veronese variety. 2.12. Theorem. Let ${\cal H}$ be an arrangement of $m\geq n+2$ hyperplanes in $P^n$ in general position. Denote by $W$ the space $H^0(P^n, E({\cal H})) \cong {\bf C}^{m-1}$. For any point $x\in P^n$ consider the subspace of $W$ consisting of all sections vanishing at $x$, and let $\phi_{\cal H}(x)$ be the dual subspace of $W^*$. Then:a) The [dim]{}ension of $\phi_{\cal H}(x)$ equals $n$ for all $x\in P^n$;b) The correspondence $x\mapsto \phi_{\cal H}(x)$ is a regular embedding $\phi_{\cal H}: P^n \hookrightarrow G(n, W)^*$. c) The image $\phi_{\cal H}(P^n)$ in $G(n, W^*)$ becomes, after the Plücker embedding $G(m-n-1,W) \i P(\bigwedge^{n} W^*)$, a Veronese variety. In particular, $E({\cal H})$ is the inverse image of the bundle ${\cal S}^*$ on $G(n,W^*)$ where ${\cal S}$ is the tautological subbundle over $G(n,W^*)$. 2.13. Corollary. Assume that $m=n+2$. Then $E({\cal H}) \cong T_{P^n}(-1)$ where $T_{P^n}$ is the tangent bundle of $P^n$. [*Proof:*]{} Since ${\rm dim} (W) = n+1$, the map $\phi_{\cal H}$ defines an isomorphism $P^n = P(V)\rightarrow G(n,W^*) = G(1,W) = P(W)$. In this case the tautological subbundle ${\cal S}$ on $G(n,W^*)$ is isomorphic to $\Omega^1_{P(W)}(1)$. Hence $E({\cal H})$ is isomorphic to $T_{P^n}(-1)$. .3cm [**2.14.**]{} Let us call a rank $n$ vector bundle $E$ on $P^n$ [*normalized*]{} if $c_

a Veronese variety. 2.12.   Theorem. Let $ { \cal H}$ be an arrangement of $ m\geq n+2 $ hyperplanes in $ P^n$ in cosmopolitan placement. Denote by $ W$ the space $ H^0(P^n, E({\cal planck's constant }) ) \cong { \bf C}^{m-1}$. For any degree $ x\in P^n$ consider the subspace of $ W$ consisting of all department vanish at $ x$, and let $ \phi_{\cal H}(x)$ be the double subspace of $ W^*$. Then: a) The [ dim]{}ension of $ \phi_{\cal H}(x)$ equals $ n$ for all $ x\in P^n$;b) The correspondence $ x\mapsto \phi_{\cal H}(x)$ is a even embedding $ \phi_{\cal H }: P^n \hookrightarrow G(n, W)^*$. c) The trope $ \phi_{\cal H}(P^n)$ in $ G(n, W^*)$ becomes, after the Plücker implant $ G(m - n-1,W) \i P(\bigwedge^{n } W^*)$, a Veronese variety. In particular, $ E({\cal H})$ is the inverse persona of the bundle $ { \cal S}^*$ on $ G(n, W^*)$ where $ { \cal S}$ is the tautological subbundle over $ G(n, W^*)$. 2.13.   Corollary. Assume that $ megabyte = n+2$. Then $ E({\cal H }) \cong T_{P^n}(-1)$ where $ T_{P^n}$ is the tangent bundle of $ P^n$. [ * Proof :* ] { } Since $ { \rm dim } (W) = n+1 $, the map $ \phi_{\cal H}$ defines an isomorphism $ P^n = P(V)\rightarrow G(n, W^ *) = G(1,W) = P(W)$. In this event the tautological subbundle $ { \cal S}$ on $ G(n, W^*)$ is isomorphic to $ \Omega^1_{P(W)}(1)$. Hence $ E({\cal H})$ is isomorphic to $ T_{P^n}(-1)$. .3 cm [ * * 2.14. * * ] { } lease us call a rank $ n$ vector bundle $ E$ on $ P^n$ [ * normalized * ] { } if $ c _

a Geronese variety. 2.12. Theorem. Uet ${\cal H}$ be an arrangxment or $m\geq n+2$ hyperplanes in $P^n$ in generap posituon. Denote by $W$ the spxce $H^0(P^n, E({\bal H})) \cont {\bf X}^{m-1}$. For any point $x\lu P^n$ dlnsibec the subspace pf $W$ consivting of all sacgilns vanishing at $x$, and let $\phi_{\cal H}(v)$ be thr fual subspace jf $W^*$. Ehen:z) The [dim]{}ension of $\phi_{\cal H}(x)$ equala $n$ for all $x\in P^n$;b) Yhe correspondence $x\mapsto \phi_{\fal H}(x)$ is a regulag embedding $\phi_{\sql H}: P^n \hookfightarrow G(n, W)^*$. c) The jmage $\phi_{\cal H}(P^n)$ in $G(n, W^*)$ becomer, aftzr the Plückgx emhgdding $G(m-n-1,W) \m P(\bigredge^{n} W^*)$, a Vcgonese eariety. On particular, $C({\cal I})$ is the inverse image of the bundle ${\cal S}^*$ jn $G(n,W^*)$ wharz ${\cal S}$ is the tautoligucal vubbgndld ovdr $F(n,X^*)$. 2.13. Codollarj. Aasume that $m=n+2$. Then $E({\cql H}) \cong T_{P^n}(-1)$ where $U_{P^n}$ px the tangenf bundje of $P^n$. [*Proof:*]{} Since ${\rm dim} (W) = n+1$, the map $\pvi_{\czl H}$ defines an isomorpyism $P^n = P(V)\rightarrow G(n,W^*) = G(1,W) = P(W)$. In this case the tautological subbundle ${\cal S}$ on $G(i,W^*)$ is ifunogphic to $\Omega^1_{P(W)}(1)$. Hence $E({\cal H})$ is isomorphic tj $T_{K^n}(-1)$. .3ck [**2.14.**]{} Let us call c rank $n$ vector nujdkg $E$ on $P^n$ [*normxlized*]{} if $c_

a Veronese variety. 2.12. Theorem. Let ${\cal an of $m\geq hyperplanes in $P^n$ $W$ space $H^0(P^n, E({\cal \cong {\bf C}^{m-1}$. any point $x\in P^n$ consider the of $W$ consisting of all sections vanishing at $x$, and let $\phi_{\cal H}(x)$ the dual subspace of $W^*$. Then:a) The [dim]{}ension of $\phi_{\cal H}(x)$ equals $n$ all P^n$;b) correspondence \phi_{\cal H}(x)$ is a regular embedding $\phi_{\cal H}: P^n \hookrightarrow G(n, W)^*$. c) The image $\phi_{\cal in $G(n, W^*)$ becomes, after the Plücker embedding \i P(\bigwedge^{n} W^*)$, a variety. In particular, $E({\cal H})$ the image of bundle S}^*$ $G(n,W^*)$ where ${\cal is the tautological subbundle over $G(n,W^*)$. 2.13. Corollary. Assume that $m=n+2$. Then $E({\cal H}) \cong T_{P^n}(-1)$ where is the of $P^n$. Since dim} = n+1$, the H}$ defines an isomorphism $P^n = G(1,W) = P(W)$. In this case the tautological ${\cal S}$ $G(n,W^*)$ is isomorphic to $\Omega^1_{P(W)}(1)$. Hence H})$ is isomorphic to $T_{P^n}(-1)$. .3cm [**2.14.**]{} Let call a rank $n$ vector bundle $E$ on $P^n$ [*normalized*]{} if $c_

a Veronese variety. 2.12. Theorem. LeT ${\cal H}$ be an aRrangEmeNt oF $m\Geq n+2$ HypeRplanes in $P^n$ in gENeraL position. Denote by $W$ the sPace $H^0(p^n, e({\Cal H})) \COnG {\bf C}^{m-1}$. for any pOInT $X\In P^N$ cOnSidEr THe SubspAce Of $W$ consIsting of alL seCtIons vanishinG At $X$, and let $\phi_{\Cal h}(x)$ be the dual sUbsPace of $w^*$. THen:A) the [diM]{}enSion oF $\phi_{\caL h}(x)$ equaLs $n$ for all $X\iN p^n$;b) The COrrespoNDEnCe $x\mApsto \phi_{\cal H}(x)$ is a rEGuLAr embedding $\phi_{\Cal H}: P^n \HoOKrIGHtaRroW G(n, W)^*$. c) The imAgE $\phi_{\cAL H}(P^n)$ in $G(N, w^*)$ bECOMes, AFter the PlückeR embedding $G(M-N-1,W) \i p(\bigweDgE^{n} W^*)$, A veroneSe varIeTY. In Particular, $E({\Cal H})$ Is the inveRse imaGE of the bUNdle ${\cal s}^*$ on $G(n,W^*)$ WheRe ${\cAl S}$ iS ThE tAutOlOGicAL sUbbUNdlE over $G(n,W^*)$. 2.13. coRoLlary. assuME THAt $m=n+2$. theN $E({\caL H}) \conG T_{P^n}(-1)$ where $T_{P^n}$ iS thE tanGEnt BundlE of $P^n$. [*prooF:*]{} SInce ${\rM dim} (W) = n+1$, The maP $\pHi_{\cal H}$ defines an IsomOrphism $P^n = p(V)\rIgHtaRrOw G(n,W^*) = g(1,w) = P(W)$. In tHis CasE the tauTologicAL suBbUNDLe ${\Cal S}$ on $G(n,W^*)$ is isomorpHiC TO $\OMega^1_{P(W)}(1)$. HeNce $E({\caL h})$ iS iSOmorphic To $t_{P^n}(-1)$. .3Cm [**2.14.**]{} LeT US call A ranK $N$ vEctor bunDle $E$ on $p^N$ [*nOrMalized*]{} If $C_

a Veronese variety. 2.12 . Theorem. Let${\ cal H }$ b e an arrangement o f $m\ geq n+2$ hyperplanes i n $P^ n$ in g e ne ral p osition . D e n ote b y$W$ t h espace $H ^0(P^n, E({\cal H })) \ cong {\bf C} ^ {m -1}$. Forany point $x\in P^ n$ con si der the s ubs paceof $W$ consis ting of a ll sectio n s vanis h i ng at$x$, and let $\ph i _{ \ cal H}(x)$ bethe du al su b s pac e o f $W^*$. T he n:a)T he [dim ] {} e n s ion of $\phi_{\ca l H}(x)$ eq u als $n$ f or al l $x\in P^n$ ;b ) Th e correspon denc e $x\maps to \ph i _{\calH }(x)$ i s a re gul arembe d di ng $\ ph i _{\ c al H} : P^ n \hookr ig ht arrow G(n , W ) ^*$. c) The imag e $\phi_{\cal H} (P^n ) $ i n $G( n, W^ *)$be comes , afte r the P lücker embeddin g $G (m-n-1,W) \i P (\b ig wedge ^ {n} W^ *)$ , a Verone se vari e ty. I n pa rticular, $E({\cal H } ) $is the i nverse im ag e of thebu ndl e ${ \ c al S} ^*$o n$G(n,W^* )$ whe r e${ \cal S} $is the t aut olo gical subb undleover $G( n,W^* ) $. 2.13. Coro l lary. Assumet ha t $m = n+2$ . T hen $E({\ca l H} ) \co ng T _ {P ^n} ( -1)$where $ T _{ P ^n}$ is the tangent b undleof $P ^n$. [*Proof :*]{} Sinc e $ {\rm dim } (W ) = n+1$, the map$\phi _{\cal H}$ definesan is omorphis m $P^n =P ( V)\right arr ow G( n,W ^ * )= G(1,W) = P( W ) $. I nthis ca sethe tau tol ogi cal su bb undle ${\ cal S}$on $ G( n, W^* )$ is isomorph ic to $ \Om ega^1 _ {P(W)} (1)$. Hen ce $ E ({\ cal H}) $ i s isom or ph ic t o $ T_ {P^n} (-1) $ . .3cm [ **2.14.** ]{} Letus c all a r ank $n$ vecto rbundle $E$ o n $ P^n$ [ * n ormalize d*]{} if $c_

a_Veronese variety. 2.12. Theorem._Let ${\cal H}$ be_an arrangement_of_$m\geq n+2$_hyperplanes_in $P^n$ in_general position. Denote_by $W$ the space_$H^0(P^n, E({\cal H}))_\cong_{\bf C}^{m-1}$. For any point $x\in P^n$ consider the subspace of $W$ consisting of_all_sections vanishing_at_$x$,_and let $\phi_{\cal H}(x)$ be_the dual subspace of $W^*$._Then:a) The_[dim]{}ension of $\phi_{\cal H}(x)$ equals $n$ for all_$x\in_P^n$;b) The correspondence_$x\mapsto \phi_{\cal H}(x)$ is a regular embedding $\phi_{\cal H}:_P^n \hookrightarrow G(n, W)^*$. c) The_image $\phi_{\cal H}(P^n)$_in_$G(n,_W^*)$ becomes, after the_Plücker embedding $G(m-n-1,W) \i P(\bigwedge^{n} W^*)$,_a Veronese variety. In particular, $E({\cal H})$_is the inverse image of the bundle_${\cal S}^*$ on $G(n,W^*)$ where ${\cal_S}$ is the tautological subbundle_over $G(n,W^*)$. 2.13. Corollary._Assume that $m=n+2$. Then $E({\cal_H}) \cong T_{P^n}(-1)$ where_$T_{P^n}$ is_the tangent bundle_of $P^n$. [*Proof:*]{} Since ${\rm dim} (W)_= n+1$, the_map $\phi_{\cal H}$ defines an isomorphism_$P^n_= P(V)\rightarrow G(n,W^*)_=_G(1,W)_= P(W)$._In this case_the_tautological subbundle_${\cal_S}$ on $G(n,W^*)$ is isomorphic to_$\Omega^1_{P(W)}(1)$._Hence $E({\cal H})$ is isomorphic to $T_{P^n}(-1)$. .3cm [**2.14.**]{}_Let us call a_rank_$n$ vector bundle $E$_on $P^n$ [*normalized*]{} if $c_

have at the moment to trust the systems we developed. Allowing the system more degrees of freedom thus hinders the developers’ ability to estimate the degree of maturity of the system they design, which poses a severe difficulty for the engineering progress, when the desired premises or the expected effects of classical engineering tasks on the system-under-development are hard to formulate. To aid us control the development/adaptation progress of the system, we define a set of *principles*, which are basically patterns for process models. They describe the changes to be made in the engineering process for complex, adaptive systems in relation to more classical models for software and systems engineering. \[con:parallelism\] The system and its test suite should develop in parallel from the start with controlled moments of interchange of information. Eventually, the test system is to be deployed alongside the main system so that even during runtime, on-going online tests are possible [@calinescu2012self]. This argument has been made for more classical systems as well and thus classical software test is, too, no longer restricted to a specific phase of software development. However, in the case of self-learning systems, it is important to focus on the evolution of test cases: The capabilities of the system might not grow as experienced test designers expect them to compared to systems entirely realized by human engineering effort. Thus, it is important to conceive and formalize how tests in various phases relate to each other. \[con:antagonism\] Any adaptive systems must be subject to an equally adaptive test. Overfitting is a known issue for many machine learning techniques. In software development for complex adaptive systems, it can happen on a larger scale: Any limited test suite (we expect our applications to be too complex to run a complete, exhaustive test) might induce certain unwanted biases. Ideally, once we know about the cases our system has a hard time with, we can train it specifically for these situations. For the so-hardened system the search mechanism that gave us the hard test cases needs to come up with even harder ones to still beat the system-under-test. Employing autonomous adaptation at this stage is expected to make that arms race more immediate and faster than it is usually achieved with human developers and testers alone. \[con:automated\] Since the realization of tasks concerning adaptive components usually means the application of a standard machine learning process, a lot of the development effort regarding certain tasks tends to shift to an earlier phase in the process model. The most developer

have at the moment to trust the systems we develop. give up the system more academic degree of exemption thus hinders the developers ’ ability to calculate the academic degree of maturity of the system they design, which poses a severe trouble for the engineering progress, when the desired premises or the expect effects of authoritative engineering undertaking on the arrangement - under - development are hard to formulate. To help us control the development / adaptation progress of the system, we specify a set of * principles *, which are basically patterns for process models. They identify the changes to be made in the engineering process for complex, adaptive systems in relation to more classical models for software and systems engineering. \[con: parallelism\ ] The system and its test suite should develop in parallel from the start with controlled moment of interchange of information. finally, the trial organization is to be deployed alongside the main arrangement so that even during runtime, on - going online tests are possible [ @calinescu2012self ]. This controversy has been made for more classical systems as well and thus classical software examination is, too, no longer restricted to a specific phase of software development. However, in the case of self - learn system, it is authoritative to focus on the evolution of test subject: The capabilities of the system might not grow as experienced trial designers expect them to compared to systems wholly realized by human engineering effort. Thus, it is important to gestate and formalize how trial in various phase relate to each early. \[con: antagonism\ ] Any adaptive systems must be subject to an equally adaptive examination. Overfitting is a known topic for many machine eruditeness techniques. In software development for complex adaptive systems, it can happen on a big plate: Any limited test cortege (we expect our applications to be too complex to hunt a complete, exhaustive test) might induce certain undesirable biases. Ideally, once we know about the cases our system has a hard time with, we can train it specifically for these situation. For the thus - hardened system the search mechanism that gave us the hard test cases necessitate to come up with even harder ones to even beat the system - under - test. Employing autonomous adaptation at this stage is expect to make that arms race more immediate and faster than it is usually achieve with human developers and testers alone. \[con: automated\ ] Since the realization of tasks concerning adaptive components usually means the application of a standard machine learning process, a lot of the development effort regarding certain tasks tends to careen to an earlier phase in the process model. The most developer

hage at the moment to trusu the systems we bwvelopxd. Allosing the system more degrees of freevom rhus yinders the developers’ ability no estimare tie degree of matndity of the sgdtem vhey design, whigh poses a vevere difficunth yor the engineering progress, when thq desirrd premises or tre eqpqctes effects of classical engineering tasks mn the system-inder-development are hard ho flrmulate. To aid us fontrol the devqoopment/adaptxtion proggzss of the aystem, we define a set of *princkples*, which are bqsifdlly patteris for process modcks. Thef descrobe the changex tm bw made in the engineecing process for comklex, adapthvz systems in relation ti morg clavsicxo mudems fkr sofhwace and systsms engineeeing. \[con:parallelism\] Uhe wystem and ita test stite should develop in parallel from tht stadt with controlled momebts of interchange of informatyon. Eventually, the test system is to be deployed dlongakde tme mxun system so that even during runtime, on-going ogmime tests are posfible [@calinrsfu2012xglf]. This argumgnt has besn made for more cpassicaj sysrems as wtll amd thus classical software rest is, too, uo oonger restricted co a specifie phasg of spftware development. Howzver, ih the case lf self-lezfning systems, it is ikportant to focus on the edolution if tzst caser: Thg capabylities of the system might not grlw as efperienced test designers expect them to rpmpared to sistams entirelv reallzed by human eggineering effprt. Thuf, it ks importaht to cmnceive and formalize hof tests in vacious phafes eelare to exzh other. \[con:antsgonism\] Auv adaptivw systems must be xubgsct to an equaloy qdaptive test. Oferwiteijg if a known issge fur osny mxchine learkine tevhniques. In software devslopment for complrx adaptivg systems, it can happem on a larger scalt: Any nimmted trst suite (we expect our applicatikns to be toj complex to run a complete, zxhaustive test) might induce certain unwented biases. Ideally, onxe we know about thg csses our syxtem ras a harg time with, we can teain it specificakly for these situatikns. Fos the so-hardened system the search mechanism that gave us the hard test cases beeds vo come up wifh efen hdrber ones eo sviul beat the systvm-under-test. Employing autonomous adaptathou at this stage is expected tp oake that armr race more immediate ans faster than it is usually achieved wiyh human developers and besters aoone. \[ckn:automatev\] Since the realizatoon of taskw conceeninn adaptive comoonrnbs usually meais che application of a standard macyine lwagning process, a not of the ftvelopmgnt effott regarding certzin fwsks jends to shift to an earlier phase in the procesv iodel. The most seveloper

have at the moment to trust the developed. the system degrees of freedom to the degree of of the system design, which poses a severe difficulty the engineering progress, when the desired premises or the expected effects of classical tasks on the system-under-development are hard to formulate. To aid us control the progress the we a set of *principles*, which are basically patterns for process models. They describe the changes to made in the engineering process for complex, adaptive in relation to more models for software and systems \[con:parallelism\] system and test should in parallel from start with controlled moments of interchange of information. Eventually, the test system is to be deployed alongside main system even during on-going tests possible [@calinescu2012self]. This been made for more classical systems thus classical software test is, too, no longer to a phase of software development. However, in case of self-learning systems, it is important to on the evolution of test cases: The capabilities of the system might not grow as designers expect them to to systems entirely by engineering Thus, is important conceive and formalize how tests in various phases relate to each \[con:antagonism\] Any adaptive systems must be subject to an equally Overfitting a known issue many machine learning techniques. software for complex adaptive systems, happen a limited suite expect our applications to too complex to run a exhaustive test) might induce we know about the cases our system has hard time with, we can train it for these situations. For the so-hardened system the search mechanism that gave the hard needs to come up with even harder ones still beat the system-under-test. autonomous adaptation at this stage is expected to make arms more immediate faster than it usually achieved with developers and testers Since the of concerning means the application of a standard learning a lot of the development regarding tends to shift to an earlier phase in the process model. The most developer

have at the moment to trust the Systems we dEveloPed. allOwIng tHe syStem more degreeS Of frEedom thus hinders the devElopeRs’ ABiliTY tO estiMate the DEgREE of MaTuRitY oF ThE systEm tHey desiGn, which posEs a SeVere difficulTY fOr the enginEerIng progress, wHen The desIrEd pREmiseS or The exPected EFfects Of classicAl ENgineeRIng taskS ON tHe syStem-under-developmENt ARe hard to formulAte. To aId US cONTroL thE developmeNt/AdaptATion proGReSS OF thE System, we definE a set of *prinCIplEs*, whicH aRe bASicallY pattErNS foR process modEls. THey descriBe the cHAnges to BE made in The engIneEriNg prOCeSs For CoMPleX, AdAptIVe sYstems in ReLaTion tO morE CLASsicAl mOdelS for sOftware and sysTemS engINeeRing. \[cOn:parAlleLiSm\] The System And itS tEst suite should dEvelOp in paralLel FrOm tHe Start WIth conTroLleD momentS of inteRChaNgE OF InFormation. EventuallY, tHE TeSt system Is to be DEpLoYEd alongsIdE thE maiN SYstem So thAT eVen durinG runtiME, oN-gOing onlInE tests ArE poSsiBle [@caLInesCu2012self]. this arguMent hAS been made for moRE classical sysTEmS AS wELl anD thUs classical SoftWAre tEst iS, ToO, no LOnger RestrIcTEd TO a specific phase of soFtWare deVelopMent. However, in The case of sELF-Learning SystEMs, IT is important to Focus On the evoluTIon of tesT caseS: The capaBilities oF THe system MigHt nOt gRow AS ExPerienced test DESignErS expect TheM to compAreD to SysTemS eNtirely reAlized by HuMaN eNgIneEring EFfort. ThuS, iT is ImPorTant tO ConceiVe and FormAlIzE How Tests in VArIOUs phAsEs RelaTe tO eAch otHer. \[cON:anTagonisM\] Any adaptIve SYsteMs MuSt be subJect to an equalLy Adaptive teSt. oveRfittiNG Is a known Issue for many machine learNIng techNiqUes. In SoftWare develOpmEnt for ComPLex adaPtive sYstemS, iT caN HAppen ON A lArgEr Scale: Any liMITed Test sUiTe (we Expect oUr applications to be TOo cOmplex to run a cOmpLete, EXHaUstIVe TEst) MiGHt iNDUce certain unwanTed biases. IDeALlY, once we knoW AboUt The caseS our sysTem haS A hard tiMe with, we cAn train it SpEcifICAllY for these sItuationS. For the so-HArdenED sYstem The Search MeChaNism tHat gavE Us tHe harD test cAsEs needS to coMe Up with evEn harder ones to still beat The sysTem-unDer-Test. EmploYinG AutOnomous adAptaTion at this StaGe iS expeCteD To makE thaT ArMs rACe morE immEDiate and fAStEr tHAN iT is usually aCHIEveD with HumAN develOperS and testers alone. \[cON:automated\] SincE the REAliZatIOn of TaSks concerning aDapTiVE ComponenTs Usually meanS the applIcATion oF a stanDard maChine leARNiNG proceSs, a lOt oF the develOpmEnT Effort rEgArDIng cerTain TaSks tenDs to shIFt to AN Earlier phase in thE procESS modeL. the Most dEvEloper

have at the moment to tru st the sys temswedev el oped . Al lowing the sys t em m ore degrees of freedom thus h i nder s t he de veloper s ’a b ili ty t o e st i ma te th e d egree o f maturity of t he system th e ydesign, wh ich poses a sev ere diffi cu lty for t heengin eering progre ss, whenth e desir e d premi s e sor t he expected effec t so f classical en gineer in g t a s ksonthe system -u nder- d evelopm e nt a r e h a rd to formula te. To aid uscontro lthe develo pment /a d apt ation progr essof the sy stem,w e defin e a setof *pr inc ipl es*, wh ic h a re bas i ca lly pat terns fo rpr ocess mod e l s . The y d escr ibe t he changes to be mad e in theengin eeri ng proc ess fo r com pl ex, adaptive sy stem s in rela tio ntomo re cl a ssical mo del s for s oftware and s y s t em s engineering. \[ co n : pa rallelis m\] Th e s ys t em and i ts te st s u i te sh ould de velop in paral l el f rom the s tart w it h c ont rolle d mom ents o f interc hange of information . Eventually,t he t es t sys tem is to be d eplo y ed a long s id e t h e mai n sys te m s o that even during r un time,on-go ing online te sts are po s s i ble [@ca line s cu 2 012self]. This argu ment has b e en madefor m ore clas sical sys t e ms as we lland th usc l as sical softwar e test i s, too, no longer re str ict edto a specif ic phase o fso ft war e dev e lopment. H owe ve r,in th e caseof se lf-l ea rn i ngsystems , i t is i mp or tant to f ocuson t h e e volutio n of test ca s es:Th ecapabil ities of thesy stem might n otgrow a s experien ced test designers expe c t themtocompa redto system s e ntirel y r e alized by hu man e ng ine e r ing e f f or t.Th us, it isi m por tantto con ceive a nd formalize how t e sts in various p has es r e l at e t o e a chot h er. \[con:antagonis m\] Any ad ap t iv e systemsm ust b e subje ct to a n equ a lly ada ptive tes t. Overfi tt ingi s aknown issu e for ma ny machin e lear n in g tec hni ques.In so ftwar e deve l opm ent f or com pl ex ada ptive s ystems,it can happen on a larg er sca le: A nylimited t est sui te (we ex pect our appli cat ion s tobet oo co mple x t o r u n a c ompl e te, exhau s ti vet e st ) might ind u c e ce rtain un w antedbias es. Ideally, once we know aboutthec a ses ou r sys te m has a hard t ime w i t h, we ca ntrain it sp ecifical ly for t hese s ituati ons. Fo r th e so-ha rden edsystem th e s ea r ch mech an is m thatgave u s thehard t e st c a s es needs to come up w i t h eve n ha rderon es to s t illbeat the s ystem-under -test. Emp loyin g auton om ous ad apt at ion at thi s stage is expe cted to m aketha t arms rac e moreimme di ate and fast e r t h an i t is usu allyac hiev ed with h u man deve lop e rs andte ste r s alone . \ [ con:automa ted \] Si n c e the real i zati o no f tas ks con cernin g adapt i veco mponent s u s u ally mean s the app l ica ti on o f a stan da rd m ac hin elearning proces s, a lot of thed e velopm ent effo r t rega rd ing certaintasks t e nds tos h ift to ane arl ier ph aseinthep rocess model. The most d evel oper

have_at the_moment to trust the_systems we_developed._Allowing the_system_more degrees of_freedom thus hinders_the developers’ ability to_estimate the degree_of_maturity of the system they design, which poses a severe difficulty for the engineering_progress,_when the_desired_premises_or the expected effects of_classical engineering tasks on the_system-under-development are_hard to formulate. To aid us control the development/adaptation_progress_of the system,_we define a set of *principles*, which are basically_patterns for process models. They describe_the changes to_be_made_in the engineering process_for complex, adaptive systems in relation_to more classical models for software_and systems engineering. \[con:parallelism\] The system and its_test suite should develop in parallel_from the start with controlled_moments of_interchange of information. Eventually, the_test system is_to be_deployed alongside the_main system so that even during_runtime, on-going online_tests are possible [@calinescu2012self]. This argument_has_been made for_more_classical_systems as_well and thus_classical_software test_is,_too, no longer restricted to a_specific_phase of software development. However, in the_case of self-learning systems,_it_is important to focus_on the evolution of test_cases: The capabilities of the system_might not_grow as_experienced test designers expect them to compared to systems entirely realized_by human engineering effort. Thus, it_is important to conceive_and formalize_how_tests in various_phases_relate to_each other. \[con:antagonism\] Any adaptive systems must be_subject to_an equally adaptive test. Overfitting is_a known issue for_many_machine learning techniques. In software development_for complex adaptive systems, it can_happen on a larger scale:_Any_limited_test suite (we expect our_applications to be too complex to_run a complete,_exhaustive test) might induce certain unwanted biases._Ideally,_once we know about the cases_our_system has a hard time with,_we_can_train it specifically for these_situations. For the so-hardened system the_search mechanism that gave us the hard test cases_needs to come_up with even harder ones_to_still_beat the system-under-test. Employing autonomous adaptation at this stage is_expected to_make that arms_race more immediate and faster than it is usually achieved_with human developers and testers alone. \[con:automated\] Since_the realization of tasks concerning adaptive components usually means the application_of a standard machine learning process, a lot_of the development effort regarding certain tasks_tends to shift to_an_earlier phase in the process model. The_most_developer

iness measure [@Kim2016Measuring]: $$B_n \equiv \frac{\sqrt{n+1}\sigma - \sqrt{n-1}\langle\tau\rangle}{(\sqrt{n+1}-2)\sigma + \sqrt{n-1}\langle\tau\rangle},$$ which can have the value of $1$ ($0$) in the extremely bursty case (in the Poisson process) for any $n$. \(iv) The correlations between IETs have been characterized by several measures [@Karsai2018Bursty]. Among them, we focus on the memory coefficient and bursty train sizes. The memory coefficient $M$ is defined as the Pearson correlation coefficient between two consecutive IETs, whose value for a sequence of $n$ IETs, i.e., $\{\tau_i\}_{i=1,\cdots,n}$, can be estimated by [@Goh2008Burstiness] $$M \equiv\frac{1}{n - 1}\sum_{i=1}^{n-1}\frac{(\tau_i - \langle\tau\rangle_1)(\tau_{i+1} - \langle\tau\rangle_2)}{\sigma_1 \sigma_2}, \label{eq:memory_original}$$ where $\langle\tau\rangle_1$ ($\langle\tau\rangle_2$) and $\sigma_1$ ($\sigma_2$) are the average and the standard deviation of the first (last) $n-1$ IETs, respectively. Positive $M$ implies that the small (large) IETs tend to be followed by small (large) IETs. Negative $M$ implies the opposite tendency, while $M=0$ is for the uncorrelated IETs. In many empirical analyses, positive $M$ has been observed [@Goh2008Burstiness; @Wang2015Temporal; @Guo2017Bounds; @Bottcher2017Temporal]. \(v) Another notion for measuring the correlations between IETs is the bursty trains [@Karsai2012Universal]. A bursty train is defined as a set of consecutive events such that IETs between any two consecutive events in the bursty train are less than or equal to a given time window $\Delta t$, while those between events in different bursty trains are larger than $\Delta t$. The number of events in the bursty train is called

iness measure   [ @Kim2016Measuring ]: $ $ B_n \equiv \frac{\sqrt{n+1}\sigma - \sqrt{n-1}\langle\tau\rangle}{(\sqrt{n+1}-2)\sigma + \sqrt{n-1}\langle\tau\rangle},$$ which can have the value of $ 1 $ ($ 0 $) in the extremely bursty case (in the Poisson procedure) for any $ n$. \(iv) The correlation between IETs have been characterized by several measures   [ @Karsai2018Bursty ]. Among them, we concenter on the memory coefficient and bursty train sizes. The memory coefficient $ M$ is define as the Pearson correlation coefficient coefficient between two consecutive IETs, whose value for a sequence of $ n$ IETs, i.e., $ \{\tau_i\}_{i=1,\cdots, n}$, can be estimated by   [ @Goh2008Burstiness ] $ $ M \equiv\frac{1}{n - 1}\sum_{i=1}^{n-1}\frac{(\tau_i - \langle\tau\rangle_1)(\tau_{i+1 } - \langle\tau\rangle_2)}{\sigma_1 \sigma_2 }, \label{eq: memory_original}$$ where $ \langle\tau\rangle_1 $ ($ \langle\tau\rangle_2 $) and $ \sigma_1 $ ($ \sigma_2 $) are the average and the standard diversion of the first (last) $ n-1 $ IETs, respectively. Positive $ M$ implies that the belittled (large) IETs tend to be followed by small (large) IETs. Negative $ M$ imply the opposite tendency, while $ M=0 $ is for the uncorrelated IETs. In many empirical analysis, positive $ M$ has been observed   [ @Goh2008Burstiness; @Wang2015Temporal; @Guo2017Bounds; @Bottcher2017Temporal ]. \(v) Another notion for measuring the correlation between IETs is the bursty trains   [ @Karsai2012Universal ]. A bursty train is defined as a set of consecutive event such that IETs between any two consecutive events in the bursty train are less than or equal to a given time window $ \Delta t$, while those between events in different bursty trains are larger than $ \Delta t$. The number of events in the bursty train is called

ineds measure [@Kim2016Measuring]: $$B_k \equiv \frac{\sqrt{u+1}\wigma - \sqrt{n-1}\mangle\taj\rangle}{(\sqrt{n+1}-2)\sigma + \sqrt{n-1}\langpe\rau\rabgle},$$ which can have thd value ov $1$ ($0$) in tye eetremely bursty rzse (in bke Pojdson 'rocess) for any $n$. \(iv) The cmrrelations bedwdeu IETs have been characterized by sederal mradures [@Karsai2018Burfty]. Siong nhtm, we focus on the memory coefficjent anv bursty train xizes. The memory coefficiejt $M$ is defined as the Pearson cotdelwrion coeffickent betwetn two consedutive IETs, whose value for a sdquenee of $n$ IETw, u.e., $\{\hdu_i\}_{i=1,\cdots,n}$, ran be estimated bn [@Boh2008Burvtiness] $$M \equiv\frac{1}{n - 1}\suk_{i=1}^{n-1}\drac{(\tau_i - \langle\tau\raigle_1)(\tau_{i+1} - \langle\tau\rwngle_2)}{\sigmd_1 \aigma_2}, \label{eq:mwmiry_orhgindl}$$ wfwre $\lahgke\fau\ranhle_1$ ($\langle\tau\dangle_2$) and $\wigma_1$ ($\sigma_2$) are the advtage and the standwrq deviation of the first (last) $n-1$ IETs, revpedtively. Positive $M$ implues that the small (latge) IETs tqnd to be followed by small (large) IETs. Negative $M$ implmer tkc upplsite tendency, while $M=0$ is for the uncorrelateq ITTs. In many empirlcal analyses, posiyige $I$ has been obrerved [@Ykh2008Gurstiness; @Wang2015Temooral; @Goo2017Bounes; @Bottchtr2017Temloral]. \(v) Another notion for mwasuring the xorrelations betwezn IETs is tke burxty ttains [@Karsai2012Universal]. A yursty train is dffined as x set of consecugivv evants such that IETs betweeg any two conxecutivd evgnts in the burstj tralt are less than or equap do a given time window $\Delta t$, while thosx between evemtv it differznt buvsty trains are larger than $\Dglta t$. Thz numbdr of evenns in the bursty trayn is called

iness measure [@Kim2016Measuring]: $$B_n \equiv \frac{\sqrt{n+1}\sigma - \sqrt{n-1}\langle\tau\rangle},$$ can have value of $1$ case the Poisson process) any $n$. \(iv) correlations between IETs have been characterized several measures [@Karsai2018Bursty]. Among them, we focus on the memory coefficient and bursty sizes. The memory coefficient $M$ is defined as the Pearson correlation coefficient between consecutive whose for sequence of $n$ IETs, i.e., $\{\tau_i\}_{i=1,\cdots,n}$, can be estimated by [@Goh2008Burstiness] $$M \equiv\frac{1}{n - 1}\sum_{i=1}^{n-1}\frac{(\tau_i - - \langle\tau\rangle_2)}{\sigma_1 \sigma_2}, \label{eq:memory_original}$$ where $\langle\tau\rangle_1$ ($\langle\tau\rangle_2$) and ($\sigma_2$) are the average the standard deviation of the (last) IETs, respectively. $M$ that small (large) IETs to be followed by small (large) IETs. Negative $M$ implies the opposite tendency, while $M=0$ is for uncorrelated IETs. empirical analyses, $M$ been [@Goh2008Burstiness; @Wang2015Temporal; @Guo2017Bounds; Another notion for measuring the correlations the bursty trains [@Karsai2012Universal]. A bursty train is as a of consecutive events such that IETs any two consecutive events in the bursty train less than or equal to a given time window $\Delta t$, while those between events bursty trains are larger $\Delta t$. The of in bursty is called

iness measure [@Kim2016Measuring]: $$B_N \equiv \frac{\Sqrt{n+1}\SigMa - \sQrT{n-1}\laNgle\Tau\rangle}{(\sqrt{n+1}-2)\SIgma + \Sqrt{n-1}\langle\tau\rangle},$$ whIch caN hAVe thE VaLue of $1$ ($0$) In the exTReMELy bUrStY caSe (IN tHe PoiSsoN procesS) for any $n$. \(iv) the CoRrelations beTWeEn IETs have BeeN characterizEd bY severAl MeaSUres [@KArsAi2018BurSty]. AmoNG them, wE focus on tHe MEmory cOEfficieNT AnD burSty train sizes. The mEMoRY coefficient $M$ iS definEd AS tHE peaRsoN correlatiOn CoeffICient beTWeEN TWo cONsecutive IETs, Whose value fOR a sEquencE oF $n$ Iets, i.e., $\{\taU_i\}_{i=1,\cdOtS,N}$, caN be estimateD by [@GOh2008BurstinEss] $$M \eqUIv\frac{1}{n - 1}\SUm_{i=1}^{n-1}\fraC{(\tau_i - \lAngLe\tAu\raNGlE_1)(\tAu_{i+1} - \LaNGle\TAu\RanGLe_2)}{\sIgma_1 \sigmA_2}, \lAbEl{eq:mEmorY_ORIGinaL}$$ whEre $\lAngle\Tau\rangle_1$ ($\langLe\tAu\raNGle_2$) And $\siGma_1$ ($\siGma_2$) aRe The avErage aNd the StAndard deviation Of thE first (lasT) $n-1$ IeTS, reSpEctivELy. PosiTivE $M$ iMplies tHat the sMAll (LaRGE) iEts tend to be followed By SMAlL (large) IEts. NegaTIvE $M$ IMplies thE oPpoSite TENdencY, whiLE $M=0$ Is for the UncorrELaTeD IETs. In MaNy empiRiCal AnaLyses, POsitIve $M$ haS been obsErved [@gOh2008Burstiness; @WaNG2015Temporal; @Guo2017BOUnDS; @boTTcheR2017TeMporal]. \(v) AnotHer nOTion For mEAsUriNG the cOrrelAtIOnS Between IETs is the burStY trainS [@KarsAi2012Universal]. A bUrsty train IS DEfined as A set OF cONsecutive eventS such That IETs beTWeen any tWo conSecutive Events in tHE Bursty trAin Are LesS thAN Or Equal to a given TIMe wiNdOw $\Delta T$, whIle thosE beTweEn eVenTs In differeNt bursty TrAiNs ArE laRger tHAn $\Delta t$. thE nuMbEr oF evenTS in the BurstY traIn Is CAllEd

iness measure [@Kim2016Mea suring]: $ $B_n\eq uiv \ frac {\sq rt{n+1}\sigma- \sq rt{n-1}\langle\tau\ran gle}{ (\ s qrt{ n +1 }-2)\ sigma + \s q r t{n -1 }\ lan gl e \t au\ra ngl e},$$ w hich can h ave t he value of$ 1$ ($0$) inthe extremely b urs ty cas e(in the P ois son p rocess ) for a ny $n$. \( i v) The correla t i on s be tween IETs have b e en characterizedby sev er a lm e asu res  [@Karsai2 01 8Burs t y]. Amo n gt h e m,w e focus on th e memory co e ffi cientan d b u rsty t rainsi z es. The memory coe fficient$M$ is defined as thePearso n c orr elat i on c oef fi c ien t b etw e entwo cons ec ut ive I ETs, w h o se v alu e fo r a s equence of $n $ I ETs, i.e ., $\ {\tau _i\} _{ i=1,\ cdots, n}$,ca n be estimatedby [ @Goh2008B urs ti nes s] $$M\ equiv\ fra c{1 }{n - 1 }\sum_{ i =1} ^{ n - 1 }\ frac{(\tau_i - \la ng l e \t au\rangl e_1)(\ t au _{ i +1} - \l an gle \tau \ r angle _2)} { \s igma_1 \ sigma_ 2 }, \lab el {eq:me mo ry_ ori ginal } $$ w here $ \langle\ tau\r a ngle_1$ ($\lan g le\tau\rangle _ 2$ ) an d $\s igm a_1$ ($\sig ma_2 $ ) ar e th e a ver a ge an d the s t an d ard deviation of th efirst(last ) $n-1$ IETs, respectiv e l y . Positi ve $ M $i mplies that th e sma ll (large) IETs ten d tobe follo wed by sm a l l (large ) I ETs . N ega t i ve $M$ impliest h e op po site te nde ncy, wh ile $M =0$ is f or the un correlat ed I ET s. In many empirica lana ly ses , pos i tive $ M$ ha s be en o b ser ved [@G o h2 0 0 8Bur st in ess; @W an g2015 Temp o ral ; @Guo2 017Bounds ; @ B ottc he r2 017Temp oral]. \(v)An other noti on fo r meas u r ing thecorrelations between IE T s is th e b ursty tra ins [@Kar sai 2012Un ive r sal].A burs ty tr ai n i s defin e d a s a s et of cons e c uti ve ev en ts s uch tha t IETs between any two consecutiveeve ntsi n t heb ur s tytr a ina r e less than orequal to a g i ve n time win d ow$\ Delta t $, whil e tho s e betwe en events in diffe re nt b u r sty trains ar e larger than $\D e lta t $ .The n umb er ofev ent s inthe bu r sty trai n is c al led

iness measure [@Kim2016Measuring]:_$$B_n \equiv_\frac{\sqrt{n+1}\sigma - \sqrt{n-1}\langle\tau\rangle}{(\sqrt{n+1}-2)\sigma +_\sqrt{n-1}\langle\tau\rangle},$$ which_can_have the_value_of $1$ ($0$)_in the extremely_bursty case (in the_Poisson process) for_any_$n$. \(iv) The correlations between IETs have been characterized by several measures [@Karsai2018Bursty]. Among them, we_focus_on the_memory_coefficient_and bursty train sizes. The_memory coefficient $M$ is defined_as the_Pearson correlation coefficient between two consecutive IETs, whose_value_for a sequence_of $n$ IETs, i.e., $\{\tau_i\}_{i=1,\cdots,n}$, can be estimated by [@Goh2008Burstiness]_$$M \equiv\frac{1}{n - 1}\sum_{i=1}^{n-1}\frac{(\tau_i - \langle\tau\rangle_1)(\tau_{i+1}_- \langle\tau\rangle_2)}{\sigma_1 \sigma_2}, ___ \label{eq:memory_original}$$ where $\langle\tau\rangle_1$_($\langle\tau\rangle_2$) and $\sigma_1$ ($\sigma_2$) are the_average and the standard deviation of_the first (last) $n-1$ IETs, respectively. Positive_$M$ implies that the small (large)_IETs tend to be followed_by small_(large) IETs. Negative $M$ implies_the opposite tendency,_while $M=0$_is for the_uncorrelated IETs. In many empirical analyses,_positive $M$ has_been observed [@Goh2008Burstiness; @Wang2015Temporal; @Guo2017Bounds; @Bottcher2017Temporal]. \(v) Another_notion_for measuring the_correlations_between_IETs is_the bursty trains [@Karsai2012Universal]._A_bursty train_is_defined as a set of consecutive_events_such that IETs between any two consecutive_events in the bursty_train_are less than or_equal to a given time_window $\Delta t$, while those between_events in_different bursty_trains are larger than $\Delta t$. The number of events in_the bursty train is called

K$ be a polytope in ${\mathbb{R}}^{d}$ defined by the intersection of $n$ half-spaces and contained in $[0,1]^d$. Then there exists a randomized algorithm which, given input ${\varepsilon}> 0$ and $1 \leq i \leq d$, runs in time $\poly(d, n, \frac{1}{{\varepsilon}})$, and outputs an ${\varepsilon}$-approximation to $V_{i}(K)$ with high probability. We use the fact (Theorem \[thm:random\_projections\]) that $V_i(K)$ is the expected volume of the projection of $K$ onto a randomly chosen $i$-dimensional subspace (sampled according to the Haar measure). Since $K$ is contained inside $[0,1]^d$, any $i$-dimensional projection of $K$ will be contained within an $i$-dimensional projection of $[0,1]^d$, whose $i$-dimensional volume is at most $\poly(d)$. By Hoeffding’s inequality, we can therefore obtain an ${\varepsilon}$-approximation to $V_{i}(K)$ by taking the average of $\poly(d, \frac{1}{{\varepsilon}})$ $({\varepsilon}/2)$-approximations for volumes of projections of $K$ onto $i$-dimensional subspaces. To approximately compute the volume of a projection of $K$ onto an $i$-dimensional subspace $S$, we will apply Theorem \[thm:random\_projections\]. Note that we can check whether a point belongs in the projection of $K$ into $S$ by solving an LP (the point adds $i$ additional linear constraints to the constraints defining $K$). This can be done efficiently in polynomial time, and therefore we have a polynomial-time membership oracle for this subproblem. We now briefly argue that Theorem \[thm:intrinsicvolumeapprox\] allows us to implement efficient randomized variants of SymmetricSearch and PricingSearch which succeed with high probability. To do this, it suffices to note that all of the analysis of both algorithms is robust to tiny perturbations in computations of intrinsic volumes. For example, in SymmetricSearch the analysis carries through even if instead of $K_i$ dividing $S_t$ into two regions such that $V_{i}(S^{+}) = V_{i}(S^{-})$, it divides them into regions satisfying $V_{i}(S^{+}) \in [(1-{\varepsilon})V

K$ be a polytope in $ { \mathbb{R}}^{d}$ defined by the intersection of $ n$ half - spaces and contained in $ [ 0,1]^d$. Then there exist a randomized algorithm which, render input $ { \varepsilon } > 0 $ and $ 1 \leq i \leq d$, runs in time $ \poly(d, n, \frac{1}{{\varepsilon}})$, and output an $ { \varepsilon}$-approximation to $ V_{i}(K)$ with high probability. We use the fact (Theorem \[thm: random\_projections\ ]) that $ V_i(K)$ is the expected bulk of the projection of $ K$ onto a randomly chosen $ i$-dimensional subspace (sample harmonize to the Haar measure). Since $ K$ is contain inside $ [ 0,1]^d$, any $ i$-dimensional projection of $ K$ will be contained within an $ i$-dimensional protrusion of $ [ 0,1]^d$, whose $ i$-dimensional bulk is at most $ \poly(d)$. By Hoeffding ’s inequality, we can therefore obtain an $ { \varepsilon}$-approximation to $ V_{i}(K)$ by taking the average of $ \poly(d, \frac{1}{{\varepsilon}})$ $ ({ \varepsilon}/2)$-approximations for volumes of protrusion of $ K$ onto $ i$-dimensional subspaces. To approximately compute the volume of a expulsion of $ K$ onto an $ i$-dimensional subspace $ S$, we will apply Theorem \[thm: random\_projections\ ]. Note that we can check whether a detail belongs in the projection of $ K$ into $ S$ by solving an LP (the point adds $ i$ extra linear constraints to the constraints defining $ K$). This can be done efficiently in polynomial time, and consequently we suffer a polynomial - time membership oracle for this subproblem. We now briefly argue that Theorem \[thm: intrinsicvolumeapprox\ ] allows us to implement efficient randomized variants of SymmetricSearch and PricingSearch which succeed with eminent probability. To do this, it suffices to notice that all of the psychoanalysis of both algorithms is robust to tiny perturbations in calculation of intrinsic volumes. For example, in SymmetricSearch the analysis carries through even if alternatively of $ K_i$ dividing $ S_t$ into two regions such that $ V_{i}(S^{+ }) = V_{i}(S^{-})$, it divides them into regions satisfying $ V_{i}(S^{+ }) \in [ (1-{\varepsilon})V

K$ bf a polytope in ${\mathbb{R}}^{d}$ defined by the intersxction kf $n$ halw-spaces and contained in $[0,1]^d$. Tien rhere exists a randomized augorithm ahich, gicen mnput ${\varepsilon}> 0$ and $1 \lcz i \mcq d$, xuis in time $\poly(c, n, \frac{1}{{\vasepsilon}})$, and ogtoucs an ${\varepsilon}$-approximation to $V_{i}(K)$ with hogj probability. Wg use ehe ract (Theorem \[thm:random\_projections\]) fhat $V_i(N)$ is the expevted volume of the projectlon lf $K$ onto a randompy chosen $i$-eimegwional subspxce (sampled according jo the Haar measure). Since $K$ is cuntaiued inside $[0,1]^e$, qny $h$-dimensionao protection of $K$ will be contaimed within an $l$-dimeisiobal projection of $[0,1]^d$, wiose $i$-dimensional vojume is ad jost $\poly(d)$. By Hoedfeing’s inexualkry, de dai tgerefoge kbtain an ${\barepsilon}$-approximation to $V_{i}(K)$ br taking the aberage os $\poly(d, \frac{1}{{\varepsilon}})$ $({\varepsilon}/2)$-approxpmatjons for volumes of prohections of $K$ onto $i$-dlmensionaj subspaces. To approximately compute the volume of a prkgecuiin of $J$ lnto an $i$-dimensional subspace $S$, we will apply Fhtorvm \[thm:random\_projegtions\]. Note that wr fam check whethet a poiuf gelongs in the prouection of $K$ into $S$ br sokving an LP (the point adds $u$ additional oinear constraints to the conrtraonts cefining $K$). This can be bone erficiently ln polynojkal time, and thefefpra we havt a polynomial-time membershmp orccle for thix subpwoblem. We nlw brlafly argue that Thforem \[tvm:intrinsifvolumeapprox\] allows us to implxkent efficiemt ratdomized varisnts of SymmeericSearch and PricinySearcf which subceed witi high probafility. To do djis, it suffires to noee tyat qll of gfe analysis of both algorithms is robust to tiny pevturbxfions in computcuiobs of intrinsic vouumqs. Fpr axample, in SfmmegrizXearcf the analywks csrries through even hf ihstead of $K_i$ dividonn $S_t$ into two regyons such thay $V_{i}(S^{+}) = V_{i}(S^{-})$, it divifes tiem invo regoonf satisfying $V_{i}(S^{+}) \in [(1-{\varepsilon})B

K$ be a polytope in ${\mathbb{R}}^{d}$ defined intersection $n$ half-spaces contained in $[0,1]^d$. algorithm given input ${\varepsilon}> and $1 \leq \leq d$, runs in time $\poly(d, \frac{1}{{\varepsilon}})$, and outputs an ${\varepsilon}$-approximation to $V_{i}(K)$ with high probability. We use the (Theorem \[thm:random\_projections\]) that $V_i(K)$ is the expected volume of the projection of $K$ a chosen subspace according to the Haar measure). Since $K$ is contained inside $[0,1]^d$, any $i$-dimensional projection of $K$ be contained within an $i$-dimensional projection of $[0,1]^d$, $i$-dimensional volume is at $\poly(d)$. By Hoeffding’s inequality, we therefore an ${\varepsilon}$-approximation $V_{i}(K)$ taking average of $\poly(d, $({\varepsilon}/2)$-approximations for volumes of projections of $K$ onto $i$-dimensional subspaces. To approximately compute the volume of a of $K$ $i$-dimensional subspace we apply \[thm:random\_projections\]. Note that check whether a point belongs in $K$ into $S$ by solving an LP (the adds $i$ linear constraints to the constraints defining This can be done efficiently in polynomial time, therefore we have a polynomial-time membership oracle for this subproblem. We now briefly argue that allows us to implement randomized variants of and which with probability. To this, it suffices to note that all of the analysis of algorithms is robust to tiny perturbations in computations of intrinsic example, SymmetricSearch the analysis through even if instead $K_i$ $S_t$ into two regions $V_{i}(S^{+}) V_{i}(S^{-})$, into satisfying \in [(1-{\varepsilon})V

K$ be a polytope in ${\mathbb{R}}^{d}$ defIned by the iNtersEctIon Of $N$ halF-spaCes and containeD In $[0,1]^d$. THen there exists a randomiZed alGoRIthm WHiCh, givEn input ${\VArEPSilOn}> 0$ AnD $1 \leQ i \LEq D$, runs In tIme $\poly(D, n, \frac{1}{{\varePsiLoN}})$, and outputs aN ${\VaRepsilon}$-apProXimation to $V_{i}(k)$ wiTh high PrObaBIlity. we uSe the Fact (ThEOrem \[thM:random\_prOjECtions\]) THat $V_i(K)$ iS THe ExpeCted volume of the prOJeCTion of $K$ onto a raNdomly ChOSeN $I$-DimEnsIonal subspAcE (sampLEd accorDInG TO The hAar measure). SinCe $K$ is contaiNEd iNside $[0,1]^d$, AnY $i$-dIMensioNal prOjECtiOn of $K$ will be ContAined withIn an $i$-dIMensionAL projecTion of $[0,1]^D$, whOse $I$-dimENsIoNal VoLUme IS aT moST $\poLy(d)$. By HoeFfDiNg’s inEquaLITY, We caN thErefOre obTain an ${\varepsiLon}$-ApprOXimAtion To $V_{i}(K)$ By taKiNg the AveragE of $\poLy(D, \frac{1}{{\varepsilon}})$ $({\VarePsilon}/2)$-appRoxImAtiOnS for vOLumes oF prOjeCtions oF $K$ onto $i$-DImeNsIONAl Subspaces. To approxiMaTELy Compute tHe voluME oF a PRojectioN oF $K$ oNto aN $I$-DimenSionAL sUbspace $S$, We will APpLy theorem \[ThM:randoM\_pRojEctIons\]. NOTe thAt we caN check whEther A Point belongs in THe projection oF $k$ iNTO $S$ BY solVinG an LP (the poiNt adDS $i$ adDitiONaL liNEar coNstraInTS tO The constraints definInG $K$). This Can be Done efficientLy in polynoMIAL time, and TherEFoRE we have a polynoMial-tIme membersHIp oracle For thIs subproBlem. We now BRIefly argUe tHat theOreM \[THm:IntrinsicvoluMEApprOx\] Allows uS to ImplemeNt eFfiCieNt rAnDomized vaRiants of syMmEtRiCSeArch aND PricingseArcH wHicH succEEd with High pRobaBiLiTY. To Do this, iT SuFFIces To NoTe thAt aLl Of the AnalYSis Of both aLgorithms Is rOBust To TiNy pertuRbations in comPuTations of iNtRinSic volUMEs. For exaMple, in SymmetricSearch thE AnalysiS caRries ThroUgh even if InsTead of $k_i$ dIViding $s_t$ into Two reGiOns SUCh thaT $v_{I}(S^{+}) = v_{i}(S^{-})$, It Divides theM INto RegioNs SatiSfying $V_{I}(S^{+}) \in [(1-{\varepsilon})V

K$ be a polytope in ${\mat hbb{R}}^{d }$ de fin edby the int ersection of $ n $ ha lf-spaces and containe d in$[ 0 ,1]^ d $. Then theree xi s t s a r an dom iz e dalgor ith m which , given in put $ {\varepsilon } >0$ and $1\le q i \leq d$, ru ns inti me$ \poly (d, n, \ frac{1 } {{\var epsilon}} )$ , and o u tputs a n ${ \var epsilon}$-approxi m at i on to $V_{i}(K )$ wit hh ig h pro bab ility. We u se th e fact ( T he o r e m \ [ thm:random\_p rojections\ ] ) t hat $V _i (K) $ is th e exp ec t edvolume of t he p rojection of $K $ onto a randoml y chos en$i$ -dim e ns io nal s u bsp a ce (s a mpl ed accor di ng to t he H a a r meas ure ). S ince$K$ is contai ned ins i de$[0,1 ]^d$, any $ i$-di mensio nal p ro jection of $K$will be conta ine dwit hi n an$ i$-dim ens ion al proj ectiono f $ [0 , 1 ] ^d $, whose $i$-dimen si o n al volumeis atm os t$ \poly(d) $. By Hoe f f ding’ s in e qu ality, w e cant he re fore ob ta in an${ \va rep silon } $-ap proxim ation to $V_{ i }(K)$ by takin g the averageo f$ \ po l y(d, \f rac{1}{{\va reps i lon} })$$ ({ \va r epsil on}/2 )$ - ap p roximations for vol um es ofproje ctions of $K$ onto $i$- d i m ensional sub s pa c es. To approx imate ly compute the volu me of a proje ction of$ K $ onto a n $ i$- dim ens i o na l subspace $S $ , wewi ll appl y T heorem\[t hm: ran dom \_ projectio ns\]. No te t ha twecan c h eck whet he r a p oin t bel o ngs in theproj ec ti o n o f $K$ i n to $ S$ b yso lvin g a nLP (t he p o int adds $ i$ additi ona l lin ea rconstra ints to the c on straints d ef ini ng $K$ ) . This ca n be done efficiently i n polyno mia l tim e, a nd theref ore we ha vea polyn omial- timeme mbe r s hip o r a cl e f or this subp r o ble m. W enowbriefly argue that Theore m \[ thm:intrinsic vol umea p p ro x\] al l ows u s to i mplement effici ent random iz e dvariants o f Sy mm etricSe arch an d Pri c ingSear ch whichsucceed w it h hi g h pr obability. To do t his, it s u ffice s t o not e t hat al lofthe a nalysi s of both algor it hms is robu st to tiny perturbations in compu tation s ofint rinsic vo lum e s.For examp le,in Symmetr icS ear ch th e a n alysi s ca r ri est hroug h ev e n if inst e ad of $ K_ i$ dividing $ S _t$ into tw o regio ns s uch that $V_{i}(S ^ {+}) = V_{i}(S ^{-} ) $ , i t d i vide sthem into regi ons s a t isfying$V _{i}(S^{+}) \in [(1 -{ \ varep silon} )V

K$ be_a polytope_in ${\mathbb{R}}^{d}$ defined by_the intersection_of_$n$ half-spaces_and_contained in $[0,1]^d$._Then there exists_a randomized algorithm which,_given input ${\varepsilon}>_0$_and $1 \leq i \leq d$, runs in time $\poly(d, n, \frac{1}{{\varepsilon}})$, and outputs_an_${\varepsilon}$-approximation to_$V_{i}(K)$_with_high probability. We use the fact_(Theorem \[thm:random\_projections\]) that $V_i(K)$ is_the expected_volume of the projection of $K$ onto a_randomly_chosen $i$-dimensional subspace_(sampled according to the Haar measure). Since $K$ is_contained inside $[0,1]^d$, any $i$-dimensional projection_of $K$ will_be_contained_within an $i$-dimensional projection_of $[0,1]^d$, whose $i$-dimensional volume is_at most $\poly(d)$. By Hoeffding’s inequality,_we can therefore obtain an ${\varepsilon}$-approximation to_$V_{i}(K)$ by taking the average of_$\poly(d, \frac{1}{{\varepsilon}})$ $({\varepsilon}/2)$-approximations for volumes_of projections_of $K$ onto $i$-dimensional subspaces. To_approximately compute the_volume of_a projection of_$K$ onto an $i$-dimensional subspace $S$,_we will apply_Theorem \[thm:random\_projections\]. Note that we can_check_whether a point_belongs_in_the projection_of $K$ into_$S$_by solving_an_LP (the point adds $i$ additional_linear_constraints to the constraints defining $K$). This_can be done efficiently_in_polynomial time, and therefore_we have a polynomial-time membership_oracle for this subproblem. We now briefly_argue that_Theorem \[thm:intrinsicvolumeapprox\]_allows us to implement efficient randomized variants of SymmetricSearch and PricingSearch_which succeed with high probability. To_do this, it suffices_to note_that_all of the_analysis_of both_algorithms is robust to tiny perturbations in_computations of_intrinsic volumes. For example, in SymmetricSearch_the analysis carries through_even_if instead of $K_i$ dividing $S_t$_into two regions such that $V_{i}(S^{+})_= V_{i}(S^{-})$, it divides them_into_regions_satisfying $V_{i}(S^{+}) \in [(1-{\varepsilon})V

5/D_5] \\ \oplus \\ \mathbb{Z}[-,A_5/D_{3}] \end{array} \rightarrow \underline{\mathbb{Z}} \rightarrow 0,$$ where $D_n$ is the dihedral group of order $2n$ and $C_n$ is the cyclic group of order $n$. Denote the $1$-skeleton of $L$ by $L^{1}$. The space $L^{1}$ is a finite graph whose vertex set is denoted by $S$ and whose set of edges is denote by $E(L)$. The right angled Coxeter group $W$ associated associated to $L$ is the group defined by the presentation $$W = \langle S \ | \ s^2 \ \mbox{for all $s \in S$ and \ } (st)^2 \ \mbox{if $(s,t) \in E(L)$} \rangle.$$ Note that $W$ fits into the short exact sequence $$1 \rightarrow N \rightarrow W \xrightarrow{p} F = \bigoplus_{s\in S} C_2 \rightarrow 1$$ where $p$ takes $s\in S$ to the generator of the $C_2$-factor corresponding to $s$. A subset $J \subseteq S$ is called spherical if the subgroup $W_J=\langle J \rangle$ is finite (and hence isomorphic to $\bigoplus_{s \in J} C_2$). Note that the empty subset of $J$ is spherical. We denote the poset of spherical subsets of $S$ ordered by inclusion by $\mathcal{S}$ and its geometric realization by $K$. Note the $K$ is the cone over the barycentric subdivision of $L$. If $J \in \mathcal{S}$, then $W_J$ is called a spherical subgroup of $W$, while a coset $wW_J$ is called spherical coset. We denote the poset of spherical cosets, ordered by inclusion, by $W\mathcal{S}$. Note that $W$ acts on $W\mathcal{S}$ by left multiplication, preserving the ordering. The Davis complex $X$ of $W$ is the geometric realization of $W\mathcal{S}$. One sees that $X$ is a proper $3$-dimensional cocompact $W$-CW-complex with strict fundamental domain $K$.

5 / D_5 ] \\ \oplus \\ \mathbb{Z}[-,A_5 / D_{3 } ] \end{array } \rightarrow \underline{\mathbb{Z } } \rightarrow 0,$$ where $ D_n$ is the dihedral group of order $ 2n$ and $ C_n$ is the cyclic group of order $ n$. Denote the $ 1$-skeleton of $ L$ by $ L^{1}$. The space $ L^{1}$ is a finite graph whose vertex set is announce by $ S$ and whose hardening of edges is denote by $ E(L)$. The right lean Coxeter group $ W$ associated associated to $ L$ is the group define by the display $ $ W = \langle S \ | \ s^2 \ \mbox{for all $ s \in S$ and \ } (st)^2 \ \mbox{if $ (s, t) \in E(L)$ } \rangle.$$ Note that $ W$ fits into the short exact succession $ $ 1 \rightarrow N \rightarrow W \xrightarrow{p } F = \bigoplus_{s\in S } C_2 \rightarrow 1$$ where $ p$ takes $ s\in S$ to the generator of the $ C_2$-factor corresponding to $ s$. A subset $ J \subseteq S$ is called spherical if the subgroup $ W_J=\langle J \rangle$ is finite (and therefore isomorphic to $ \bigoplus_{s \in J } C_2 $). notice that the empty subset of $ J$ is spherical. We announce the poset of ball-shaped subsets of $ S$ ordered by inclusion by $ \mathcal{S}$ and its geometric realization by $ K$. Note the $ K$ is the cone over the barycentric subsection of $ L$. If $ J \in \mathcal{S}$, then $ W_J$ is called a spherical subgroup of $ W$, while a coset $ wW_J$ is called ball-shaped coset. We denote the poset of ball-shaped cosets, ordered by inclusion, by $ W\mathcal{S}$. Note that $ W$ acts on $ W\mathcal{S}$ by left generation, preserving the ordering. The Davis complex $ X$ of $ W$ is the geometric realization of $ W\mathcal{S}$. One sees that $ X$ is a proper $ 3$-dimensional cocompact $ W$-CW - complex with strict fundamental domain $ K$.

5/D_5] \\ \oplus \\ \mathbb{Z}[-,A_5/D_{3}] \end{xrray} \rightarroc \undernine{\mafhbb{Z}} \riehtarrow 0,$$ where $D_n$ is the diiedrql griup of order $2n$ and $C_n$ ks the cyblic group of irder $n$. Deikte the $1$-skelefln oy $O$ by $L^{1}$. The spacg $L^{1}$ is a finhte graph whosa xextex set is denoted by $S$ and whose sqt of ecgfs is denote bi $E(L)$. Uhe riggn cngled Coxeter group $W$ associatsd assobiated to $L$ is thr group defined by the predentwtion $$W = \langle S \ | \ s^2 \ \mbox{fir ajo $s \in S$ and \ } (st)^2 \ \mbox{if $(s,t) \in G(L)$} \rangle.$$ Note that $W$ fits into ghe skort exact wewuejwe $$1 \rightarcow N \gightarrow W \wgightarsow{p} F = \bigoplus_{s\in S} C_2 \cighrarrow 1$$ where $p$ takes $s\in S$ to the genetator of tve $C_2$-factor correspobdung tm $s$. D sucwet $J \auusefeq S$ ls ralled sphedical if thw subgroup $W_J=\langle J \gsngle$ is finjte (anq rence isomorphic to $\bigoplus_{s \in J} C_2$). Noue thzt the empty subset of $H$ is spherical. We denlte the pjset of spherical subsets of $S$ ordered by inclusimn by $\oatkgwu{W}$ wnd its geometric realization by $K$. Note the $K$ js tme cone over the barycentriv dunqivision of $L$. If $J \nh \jathcal{S}$, then $W_J$ id calleq a spherical fubgtoup of $W$, while a coset $wW_J$ is called siherucal coset. We denoce the poset of xpherocal cosets, ordered by nnclusjon, by $W\matjcal{S}$. Nots that $W$ acts on $D\manhcan{S}$ by left multiplication, [reservinj the orderivg. Tne Davys complex $X$ of $W$ is the geometric realnzatimn of $W\matjcal{S}$. One sees that $X$ is a propxc $3$-dimensional cmcokpact $W$-CC-complcx with strict sundamental dokain $K$.

5/D_5] \\ \oplus \\ \mathbb{Z}[-,A_5/D_{3}] \end{array} \rightarrow 0,$$ $D_n$ is dihedral group of the group of order Denote the $1$-skeleton $L$ by $L^{1}$. The space $L^{1}$ a finite graph whose vertex set is denoted by $S$ and whose set edges is denote by $E(L)$. The right angled Coxeter group $W$ associated associated $L$ the defined the presentation $$W = \langle S \ | \ s^2 \ \mbox{for all $s \in S$ \ } (st)^2 \ \mbox{if $(s,t) \in E(L)$} Note that $W$ fits the short exact sequence $$1 N W \xrightarrow{p} = S} \rightarrow 1$$ where takes $s\in S$ to the generator of the $C_2$-factor corresponding to $s$. A subset $J \subseteq S$ called spherical subgroup $W_J=\langle \rangle$ finite hence isomorphic to J} C_2$). Note that the empty is spherical. We denote the poset of spherical of $S$ by inclusion by $\mathcal{S}$ and its realization by $K$. Note the $K$ is the over the barycentric subdivision of $L$. If $J \in \mathcal{S}$, then $W_J$ is called a of $W$, while a $wW_J$ is called coset. denote poset spherical cosets, by inclusion, by $W\mathcal{S}$. Note that $W$ acts on $W\mathcal{S}$ by multiplication, preserving the ordering. The Davis complex $X$ of $W$ geometric of $W\mathcal{S}$. One that $X$ is a $3$-dimensional $W$-CW-complex with strict fundamental

5/D_5] \\ \oplus \\ \mathbb{Z}[-,A_5/D_{3}] \end{array} \riGhtarrow \unDerliNe{\mAthBb{z}} \rigHtarRow 0,$$ where $D_n$ is thE DiheDral group of order $2n$ and $C_n$ Is the CyCLic gROuP of orDer $n$. DenOTe THE $1$-skElEtOn oF $L$ BY $L^{1}$. the spAce $l^{1}$ is a finIte graph whOse VeRtex set is denOTeD by $S$ and whoSe sEt of edges is dEnoTe by $E(L)$. thE riGHt angLed coxetEr grouP $w$ assocIated assoCiATed to $L$ IS the groUP DeFineD by the presentatioN $$w = \lANgle S \ | \ s^2 \ \mbox{for aLl $s \in S$ AnD \ } (St)^2 \ \MBOx{iF $(s,t) \In E(L)$} \rangle.$$ noTe thaT $w$ fits inTO tHE SHorT Exact sequence $$1 \Rightarrow N \RIghTarrow w \xRigHTarrow{P} F = \bigOpLUs_{s\In S} C_2 \rightarRow 1$$ wHere $p$ takeS $s\in S$ tO The geneRAtor of tHe $C_2$-facTor CorRespONdInG to $S$. A SUbsET $J \SubSEteQ S$ is callEd SpHericAl if THE SUbgrOup $w_J=\laNgle J \Rangle$ is finitE (anD henCE isOmorpHic to $\BigoPlUs_{s \in j} C_2$). Note That tHe Empty subset of $J$ iS sphErical. We dEnoTe The PoSet of SPhericAl sUbsEts of $S$ oRdered bY IncLuSION bY $\mathcal{S}$ and its geoMeTRIc RealizatIon by $K$. nOtE tHE $K$ is the cOnE ovEr thE BAryceNtriC SuBdivisioN of $L$. If $j \In \MaThcal{S}$, tHeN $W_J$ is cAlLed A spHericAL subGroup oF $W$, while a Coset $Ww_J$ is called spheRIcal coset. We deNOtE THe POset Of sPherical cosEts, oRDereD by iNClUsiON, by $W\mAthcaL{S}$. nOtE That $W$ acts on $W\mathcal{s}$ bY left mUltipLication, preseRving the orDERIng. The DaVis cOMpLEx $X$ of $W$ is the geoMetriC realizatiON of $W\mathCal{S}$. ONe sees thAt $X$ is a proPER $3$-dimensiOnaL coComPacT $w$-cW-Complex with stRICt fuNdAmental DomAin $K$.

5/D_5] \\ \oplus \\ \mat hbb{Z}[-,A _5/D_ {3} ] \en d{ar ray} \rightarr o w \u nderline{\mathbb{Z}} \ right ar r ow 0 ,$ $ whe re $D_n $ i s the d ih edr al gr oup o f o rder $2 n$ and $C_ n$is the cyclicg ro up of orde r $ n$. Denote t he$1$-sk el eto n of $ L$by $L ^{1}$. The s pace $L^{ 1} $ is af inite g r a ph who se vertex set isd en o ted by $S$ and whose s e to f ed ges is denote b y $E( L )$. The ri g h t an g led Coxeter g roup $W$ as s oci ated a ss oci a ted to $L$is the group defi nedby the pr esenta t ion $$W = \lang le S \ |\ s ^2 \ \m bo x{f or all $s \i n S$ and \ } ( st)^2 \ \ m b o x {if$(s ,t)\in E (L)$} \rangle .$$ Not e th at $W $ fit s in to theshortexact s equence $$1 \ri ghta rrow N \r igh ta rro wW \xr i ghtarr ow{ p}F = \bi goplus_ { s\i nS } C_ 2 \rightarrow 1$$wh e r e$p$ take s $s\i n S $t o the ge ne rat or o f the $ C_2$ - fa ctor cor respon d in gto $s$. A subse t$J\su bsete q S$is cal led sphe rical if the subgrou p $W_J=\langle J\ r an g le$isfinite (and hen c e is omor p hi c t o $\bi goplu s_ { s\ in J} C_2$). Note t ha t theempty subset of $J $ is spher i c a l. We de note th e poset of sphe rical subsets o f $S$ ord eredby inclu sion by $ \ m athcal{S }$and it s g e o me tric realizat i o n by $ K$. Not e t he $K$isthe co neov er the ba rycentri csu bd iv isi on of $L$. If$J \i n\ma thcal { S}$, t hen $ W_J$ i sc all ed a sp h er i c al s ub gr oupof$W $, wh ilea co set $wW _J$ is ca lle d sph er ic al cose t. We denoteth e poset of s phe ricalc o sets, or dered by inclusion, by$ W\mathc al{ S}$.Note that $W$ ac ts on$W\ m athcal {S}$ b y lef tmul t i plica t i on , p re serving th e ord ering .TheDavis c omplex $X$ of $W$i s t he geometricrea liza t i on of $W \ mat hc a l{S } $ . One sees that $X$ is apr o pe r $3$-dime n sio na l cocom pact $W $-CW- c omplexwith stri ct fundam en tald o mai n $K$.

5/D_5] _ \\ \oplus_\\ \mathbb{Z}[-,A_5/D_{3}] _ \end{array}_\rightarrow_\underline{\mathbb{Z}} \rightarrow_ 0,$$_where $D_n$ is_the dihedral group_of order $2n$ and_$C_n$ is the_cyclic_group of order $n$. Denote the $1$-skeleton of $L$ by $L^{1}$. The space $L^{1}$ is_a_finite graph_whose_vertex_set is denoted by $S$_and whose set of edges_is denote_by $E(L)$. The right angled Coxeter group $W$_associated_associated to $L$_is the group defined by the presentation $$W =_\langle S \ | \ s^2_\ \mbox{for all_$s_\in_S$ and \ }_ (st)^2 \ \mbox{if $(s,t)_\in E(L)$} \rangle.$$ Note that $W$_fits into the short exact sequence $$1_\rightarrow N \rightarrow W \xrightarrow{p} F_= \bigoplus_{s\in S} C_2 \rightarrow_1$$ where_$p$ takes $s\in S$ to_the generator of_the $C_2$-factor_corresponding to $s$._A subset $J \subseteq S$ is_called spherical if_the subgroup $W_J=\langle J \rangle$ is_finite_(and hence isomorphic_to_$\bigoplus_{s_\in J}_C_2$). Note that_the_empty subset_of_$J$ is spherical. We denote the_poset_of spherical subsets of $S$ ordered by_inclusion by $\mathcal{S}$ and_its_geometric realization by $K$._Note the $K$ is the_cone over the barycentric subdivision of_$L$. If_$J \in_\mathcal{S}$, then $W_J$ is called a spherical subgroup of $W$, while_a coset $wW_J$ is called spherical_coset. We denote the_poset of_spherical_cosets, ordered by_inclusion,_by $W\mathcal{S}$._Note that $W$ acts on $W\mathcal{S}$ by_left multiplication,_preserving the ordering. The Davis complex_$X$ of $W$ is_the_geometric realization of $W\mathcal{S}$. One sees_that $X$ is a proper $3$-dimensional_cocompact $W$-CW-complex with strict fundamental_domain_$K$.

^*}\Bigl[f\bigl\{D_xD_tf\cdot f^*-{\rm i}\rho^2D_xg\cdot g^*+{\rm i}\rho^2D_xf\cdot f^*+2\kappa\rho^2(gg^*-ff^*)\bigr\}$$ $$-2f_x\bigl\{D_tf\cdot f^*-{\rm i}\rho^2(gg^*-ff^*)\bigr\}+f\bigl\{D_tf\cdot f^*-{\rm i}\rho^2(gg^*-ff^*)\bigr\}_x\Bigr]=0. \eqno(2.8)$$ By virtue of equations (2.2)-(2.4), the left-hand side of (2.8) vanishes identically. $\Box$ It follows from (2.1) and (2.2) that $$|u|^2=\rho^2+{\rm i}\,{\partial \over \partial t}\,{\rm ln}\,{f^*\over f}. \eqno(2.9)$$ The above formula gives the modulus of $u$ in terms of the tau function $f$. [**Proposition 2.2.**]{} [*The [*trilinear*]{} equation for $f$ and $g$ $$f^*\left\{g_{xt}f-(f_x-{\rm i}\kappa f)g_t-{\rm i}\left({1\over \kappa}+\rho^2\right)(g_xf-gf_x)\right\} =f_t^*(g_xf-gf_x+{\rm i}\kappa fg), \eqno(2.10)$$ is a consequence of the bilinear equations (2.2)-(2.4).*]{} [**Proof.**]{} By direct calculation, one can show the following trilinear identity among the tau functions $f$ and $g$: $$f^*\left\{g_{xt}f-(f_x-{\rm i}\kappa f)g_t-{\rm i}\left({1\over \kappa}+\rho^2\right)(g_xf-gf_x)\right\} -f_t^*(g_xf-gf_x+{\rm i}\kappa fg)$$ $$=f^*(D_xD_tg\cdot f+{\rm i}\kappa D_tg\cdot f-{\rm i}\omega D_xg\cdot f)$$ $$-{g\over 2}\Bigl[\bigl\{D

^*}\Bigl[f\bigl\{D_xD_tf\cdot f^*-{\rm i}\rho^2D_xg\cdot g^*+{\rm i}\rho^2D_xf\cdot f^*+2\kappa\rho^2(gg^*-ff^*)\bigr\}$$ $ $ -2f_x\bigl\{D_tf\cdot f^*-{\rm i}\rho^2(gg^*-ff^*)\bigr\}+f\bigl\{D_tf\cdot f^*-{\rm i}\rho^2(gg^*-ff^*)\bigr\}_x\Bigr]=0. \eqno(2.8)$$ By virtue of equations (2.2)-(2.4), the left - hand slope of (2.8) vanishes identically. $ \Box$ It take after from (2.1) and (2.2) that $ $ |u|^2=\rho^2+{\rm i}\,{\partial \over \partial t}\,{\rm ln}\,{f^*\over f }. \eqno(2.9)$$ The above formula gives the modulus of $ u$ in terms of the tau routine $ f$. [ * * Proposition 2.2. * * ] { } [ * The [ * trilinear * ] { } equation for $ f$ and $ g$ $ $ f^*\left\{g_{xt}f-(f_x-{\rm i}\kappa f)g_t-{\rm i}\left({1\over \kappa}+\rho^2\right)(g_xf - gf_x)\right\ } = f_t^*(g_xf - gf_x+{\rm i}\kappa fg), \eqno(2.10)$$ is a consequence of the bilinear equation (2.2)-(2.4). * ] { } [ * * Proof. * * ] { } By direct calculation, one can prove the following trilinear identity among the tau functions $ f$ and $ g$: $ $ f^*\left\{g_{xt}f-(f_x-{\rm i}\kappa f)g_t-{\rm i}\left({1\over \kappa}+\rho^2\right)(g_xf - gf_x)\right\ } -f_t^*(g_xf - gf_x+{\rm i}\kappa fg)$$ $ $ = f^*(D_xD_tg\cdot f+{\rm i}\kappa D_tg\cdot f-{\rm i}\omega D_xg\cdot f)$$ $ $ -{g\over 2}\Bigl[\bigl\{D

^*}\Bigp[f\bigl\{D_xD_tf\cdot f^*-{\rm i}\rho^2A_xg\cdot g^*+{\rm i}\rho^2D_xf\cdov f^*+2\kappz\rho^2(gg^*-ff^*)\cigr\}$$ $$-2f_x\bigl\{D_tf\cdot f^*-{\rm i}\rho^2(gj^*-ff^*)\bugr\}+f\bugl\{D_tf\cdot f^*-{\rm i}\rho^2(gg^*-fw^*)\bigr\}_x\Bigg]=0. \eqno(2.8)$$ By viruue of equations (2.2)-(2.4), the lefb-kand alde oy (2.8) vanishes idenjically. $\Box$ Id follows from (2.1) xnb (2.2) that $$|u|^2=\rho^2+{\rm i}\,{\partial \over \partial t}\,{\rm ln}\,{g^*\oger f}. \eqno(2.9)$$ The abone forjllc gives the modulus of $u$ in terjs of tie tau function $f$. [**Proposition 2.2.**]{} [*The [*trilinewr*]{} eeuation for $f$ and $h$ $$f^*\left\{g_{xt}f-(d_x-{\rm u}\kappa f)g_t-{\rm i}\left({1\over \kappa}+\rho^2\riggt)(g_xf-gf_x)\right\} =f_t^*(g_xf-gf_x+{\rm i}\kappa wg), \equo(2.10)$$ is a conwewuejwe of the bmlineag equations (2.2)-(2.4).*]{} [**Pvpof.**]{} By direct calculation, oke cai shiw the following trilmnear identity among the tau xuuctions $f$ and $g$: $$f^*\left\{t_{xr}f-(f_x-{\rk i}\kdppa d)g_t-{\fm j}\lxft({1\kver \kwppe}+\rho^2\right)(g_xr-gf_x)\right\} -f_t^*(t_xf-gf_x+{\rm i}\kappa fg)$$ $$=g^*(D_vE_tg\cdot f+{\rm i}\iappa Q_td\cdot f-{\rm i}\omega D_xg\cdot f)$$ $$-{g\over 2}\Bigl[\bicl\{D

^*}\Bigl[f\bigl\{D_xD_tf\cdot f^*-{\rm i}\rho^2D_xg\cdot g^*+{\rm i}\rho^2D_xf\cdot f^*+2\kappa\rho^2(gg^*-ff^*)\bigr\}$$ $$-2f_x\bigl\{D_tf\cdot f^*-{\rm \eqno(2.8)$$ By of equations (2.2)-(2.4), vanishes $\Box$ It follows (2.1) and (2.2) $$|u|^2=\rho^2+{\rm i}\,{\partial \over \partial t}\,{\rm ln}\,{f^*\over \eqno(2.9)$$ The above formula gives the modulus of $u$ in terms of the function $f$. [**Proposition 2.2.**]{} [*The [*trilinear*]{} equation for $f$ and $g$ $$f^*\left\{g_{xt}f-(f_x-{\rm i}\kappa i}\left({1\over =f_t^*(g_xf-gf_x+{\rm fg), is a consequence of the bilinear equations (2.2)-(2.4).*]{} [**Proof.**]{} By direct calculation, one can show the trilinear identity among the tau functions $f$ and $$f^*\left\{g_{xt}f-(f_x-{\rm i}\kappa f)g_t-{\rm i}\left({1\over -f_t^*(g_xf-gf_x+{\rm i}\kappa fg)$$ $$=f^*(D_xD_tg\cdot f+{\rm D_tg\cdot i}\omega D_xg\cdot $$-{g\over

^*}\Bigl[f\bigl\{D_xD_tf\cdot f^*-{\rm i}\rho^2d_xg\cdot g^*+{\rm I}\rho^2D_Xf\cDot F^*+2\kAppa\Rho^2(gG^*-ff^*)\bigr\}$$ $$-2f_x\bigl\{D_TF\cdoT f^*-{\rm i}\rho^2(gg^*-ff^*)\bigr\}+f\bigl\{D_Tf\cdoT f^*-{\RM i}\rhO^2(Gg^*-Ff^*)\bigR\}_x\Bigr]=0. \eQNo(2.8)$$ bY VirTuE oF eqUaTIoNs (2.2)-(2.4), the LefT-hand siDe of (2.8) vanishEs iDeNtically. $\Box$ IT FoLlows from (2.1) aNd (2.2) tHat $$|u|^2=\rho^2+{\rm i}\,{\paRtiAl \over \PaRtiAL t}\,{\rm lN}\,{f^*\oVer f}. \eQno(2.9)$$ The ABove foRmula giveS tHE modulUS of $u$ in tERMs Of thE tau function $f$. [**PropOSiTIon 2.2.**]{} [*The [*trilineaR*]{} equatIoN FoR $F$ And $G$ $$f^*\lEft\{g_{xt}f-(f_x-{\rM i}\Kappa F)G_t-{\rm i}\leFT({1\oVER \KapPA}+\rho^2\right)(g_xf-gF_x)\right\} =f_t^*(g_xF-Gf_x+{\Rm i}\kapPa Fg), \eQNo(2.10)$$ is a cOnseqUeNCe oF the bilineaR equAtions (2.2)-(2.4).*]{} [**ProOf.**]{} By diREct calcULation, oNe can sHow The FollOWiNg TriLiNEar IDeNtiTY amOng the taU fUnCtionS $f$ anD $G$: $$F^*\LEft\{g_{Xt}f-(F_x-{\rm I}\kappA f)g_t-{\rm i}\left({1\ovEr \kAppa}+\RHo^2\rIght)(g_Xf-gf_x)\RighT\} -f_T^*(g_xf-gF_x+{\rm i}\kAppa fG)$$ $$=f^*(d_xD_tg\cdot f+{\rm i}\kaPpa D_Tg\cdot f-{\rm I}\omEgA D_xG\cDot f)$$ $$-{g\OVer 2}\BigL[\biGl\{D

^*}\Bigl[f\bigl\{D_xD_tf\c dot f^*-{\ rm i} \rh o^2 D_ xg\c dotg^*+{\rm i}\rh o ^2D_ xf\cdot f^*+2\kappa\rh o^2(g g^ * -ff^ * )\ bigr\ }$$ $$- 2 f_ x \ big l\ {D _tf \c d ot f^*- {\r m i}\rh o^2(gg^*-f f^* )\ bigr\}+f\big l \{ D_tf\cdotf^* -{\rm i}\rho ^2( gg^*-f f^ *)\ b igr\} _x\ Bigr] =0. \e q no(2.8 )$$ By vi rt u e of e q uations ( 2. 2)-( 2.4), the left-ha n ds ide of (2.8) v anishe si de n t ica lly . $\Box$ It foll o ws from (2 . 1 ) an d (2.2) that $ $|u|^2=\rho ^ 2+{ \rm i} \, {\p a rtial\over \ p art ial t}\,{\r m ln }\,{f^*\o ver f} . \eqno( 2 .9)$$ T he abo vefor mula gi ve s t he mod u lu s o f $u $ in ter ms o f the tau f u n ctio n $ f$. [**P roposition 2. 2.* *]{} [*T he [* trili near *] {} eq uation for$f $ and $g$ $$f^* \lef t\{g_{xt} f-( f_ x-{ \r m i}\ k appa f )g_ t-{ \rm i}\ left({1 \ ove r\ k a pp a}+\rho^2\right)(g _x f - gf _x)\righ t\} =f _ t^ *( g _xf-gf_x +{ \rm i}\ k a ppa f g),\ eq no(2.10) $$ isa c on sequenc eof the b ili nea r equ a tion s (2.2 )-(2.4). *]{}[**Proof.**]{} By direct cal c ul a t io n , on e c an show the fol l owin g tr i li nea r iden tityam o ng the tau functions $ f$ and $ g$: $ $f^*\left\{g_ {xt}f-(f_x - { \ rm i}\ka ppaf )g _ t-{\rm i}\left ({1\o ver \kappa } +\rho^2\ right )(g_xf-g f_x)\righ t \ } -f_t^* (g_ xf- gf_ x+{ \ r mi}\kappa fg)$ $ $$=f ^* (D_xD_t g\c dot f+{ \rm i} \ka ppa D _tg\cdotf-{\rm i }\ om eg aD_x g\cdo t f)$$ $$ -{ g\o ve r 2 }\Big l [\bigl \{D

^*}\Bigl[f\bigl\{D_xD_tf\cdot f^*-{\rm_i}\rho^2D_xg\cdot g^*+{\rm_i}\rho^2D_xf\cdot f^*+2\kappa\rho^2(gg^*-ff^*)\bigr\}$$ $$-2f_x\bigl\{D_tf\cdot f^*-{\rm_i}\rho^2(gg^*-ff^*)\bigr\}+f\bigl\{D_tf\cdot f^*-{\rm_i}\rho^2(gg^*-ff^*)\bigr\}_x\Bigr]=0._\eqno(2.8)$$ By_virtue_of equations (2.2)-(2.4),_the left-hand side_of (2.8) vanishes identically._$\Box$ It follows from_(2.1)_and (2.2) that $$|u|^2=\rho^2+{\rm i}\,{\partial \over \partial t}\,{\rm ln}\,{f^*\over f}. \eqno(2.9)$$ The above formula_gives_the modulus_of_$u$_in terms of the tau_function $f$. [**Proposition 2.2.**]{} [*The [*trilinear*]{}_equation for_$f$ and $g$ $$f^*\left\{g_{xt}f-(f_x-{\rm i}\kappa f)g_t-{\rm i}\left({1\over \kappa}+\rho^2\right)(g_xf-gf_x)\right\} =f_t^*(g_xf-gf_x+{\rm_i}\kappa_fg), \eqno(2.10)$$ is_a consequence of the bilinear equations (2.2)-(2.4).*]{} [**Proof.**]{} By direct_calculation, one can show the following_trilinear identity among_the_tau_functions $f$ and $g$:_$$f^*\left\{g_{xt}f-(f_x-{\rm i}\kappa f)g_t-{\rm i}\left({1\over \kappa}+\rho^2\right)(g_xf-gf_x)\right\} -f_t^*(g_xf-gf_x+{\rm i}\kappa_fg)$$ $$=f^*(D_xD_tg\cdot f+{\rm i}\kappa D_tg\cdot f-{\rm_i}\omega D_xg\cdot f)$$ $$-{g\over 2}\Bigl[\bigl\{D

in and applying Lemma \[stratcondlemma\] inductively, we have $$\begin{aligned} {{\mathbb E}}^g_i\exp[&-s(\mathcal{C}_i({{\rho}}_{N+1}) - \mathcal{C}_i({{\rho}}_1))]\\ & \leq \sum_{j\neq i}(\tilde{\rho}_1(j))^s\exp(-sND^*(i) + s^2NB^2/2)\\ &\leq M\exp(-sND^*(i) + s^2NB^2/2).\label{mgf3}\end{aligned}$$ ### Chernoff Bound We can use the Chernoff bound [@ross2014introduction] to conclude that $$\begin{aligned} {{\mathbb P}}_i^g[\mathcal{C}_i&({{\rho}}_{N+1}) - \mathcal{C}_i({{\rho}}_1) < \theta]\\ & \leq {{\mathbb E}}^g_i\exp[-s(\mathcal{C}_i({{\rho}}_{N+1}) - \mathcal{C}_i({{\rho}}_1) -\theta)]\\ &\stackrel{a}{\leq} M\exp(s\theta-sND^*(i) + s^2NB^2/2)\\ &\stackrel{b}{=} \epsilon_N.\end{aligned}$$ Inequality $(a)$ follows from the result in (\[mgf3\]). Equality $(b)$ is obtained by substituting $\theta = \theta_N$ and $s = s_N$ where $$\begin{aligned} \theta_N &\doteq ND^*(i) - \frac{s_N NB^2}{2}- {\frac{1}{s_N}\log\frac{M}{\epsilon_N}},\label{thetadef}\end{aligned}$$ and $s_N$ is as defined in. Under Assumptions \[epsassum\], one can easily verify that $\theta_N/ N \to D^*(i)$ as $N \to \infty$. Thus, we have shown that for the strategy pair $(f^N,g^N)$, $\psi_N(i) \geq 1-\epsilon_N$. Feasibility of Strategy in Section \[stratsubsubsec\] for Problem {#achievesymproof} ================================================================== Let ${{{T}}}$ be the smallest time index

in and applying Lemma \[stratcondlemma\ ] inductively, we have $ $ \begin{aligned } { { \mathbb E}}^g_i\exp[&-s(\mathcal{C}_i({{\rho}}_{N+1 }) - \mathcal{C}_i({{\rho}}_1))]\\ & \leq \sum_{j\neq i}(\tilde{\rho}_1(j))^s\exp(-sND^*(i) + s^2NB^2/2)\\ & \leq M\exp(-sND^*(i) + s^2NB^2/2).\label{mgf3}\end{aligned}$$ # # # Chernoff Bound We can use the Chernoff bound [ @ross2014introduction ] to reason that $ $ \begin{aligned } { { \mathbb P}}_i^g[\mathcal{C}_i&({{\rho}}_{N+1 }) - \mathcal{C}_i({{\rho}}_1) < \theta]\\ & \leq { { \mathbb E}}^g_i\exp[-s(\mathcal{C}_i({{\rho}}_{N+1 }) - \mathcal{C}_i({{\rho}}_1) -\theta)]\\ & \stackrel{a}{\leq } M\exp(s\theta - sND^*(i) + s^2NB^2/2)\\ & \stackrel{b}{= } \epsilon_N.\end{aligned}$$ Inequality $ (a)$ take after from the result in (\[mgf3\ ]). Equality $ (b)$ is obtained by substitute $ \theta = \theta_N$ and $ sulfur = s_N$ where $ $ \begin{aligned } \theta_N & \doteq ND^*(i) - \frac{s_N NB^2}{2}- { \frac{1}{s_N}\log\frac{M}{\epsilon_N}},\label{thetadef}\end{aligned}$$ and $ s_N$ is as defined in. Under assumption \[epsassum\ ], one can well verify that $ \theta_N/ N \to D^*(i)$ as $ N \to \infty$. Thus, we have shown that for the strategy pair $ (f^N, g^N)$, $ \psi_N(i) \geq 1-\epsilon_N$. Feasibility of Strategy in Section \[stratsubsubsec\ ] for Problem { # achievesymproof } = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = lease $ { { { T}}}$ be the smallest time exponent

in and applying Lemma \[straucondlemma\] inductncely, wx have $$\gegin{aliened} {{\mathbb E}}^g_i\exp[&-s(\mathcal{C}_i({{\rio}}_{N+1}) - \mathxal{C}_i({{\rho}}_1))]\\ & \leq \sum_{j\neq i}(\gilde{\rho}_1(j))^d\exp(-sND^*(i) + s^2NU^2/2)\\ &\leq M\exp(-sND^*(i) + s^2IG^2/2).\label{mny3}\end{amlgned}$$ ### Rhernoff Bound Wg can use tha Chernoff boutd [@rlss2014introduction] to conclude that $$\bedin{aligmef} {{\mathbb P}}_i^g[\matrcal{B}_i&({{\who}}_{N+1}) - \mathcal{C}_i({{\rho}}_1) < \theta]\\ & \leq {{\mathbb E}}^f_i\exp[-s(\methcal{C}_i({{\rho}}_{N+1}) - \msthcal{C}_i({{\rho}}_1) -\theta)]\\ &\stackrel{a}{\peq} L\exp(s\theta-sND^*(i) + s^2NH^2/2)\\ &\stackrel{b}{=} \glsijin_N.\end{alignea}$$ Inequaliuy $(a)$ follows from the result in (\[mgf3\]). Equalith $(b)$ ix obtained bt sktstituting $\vheta = \theta_N$ and $s = s_N$ whare $$\begon{aligned} \theta_K &\dotxq NE^*(i) - \frac{s_N NB^2}{2}- {\frac{1}{s_N}\lmg\frac{M}{\epsilon_N}},\labgl{thetadef}\anb{aligned}$$ and $s_N$ is as dwfineg in. Undde Arsujpviohs \[epswssnm\], one can sasily veridy that $\theta_N/ N \to D^*(y)$ as $N \to \inftg$. Thus, wq have shown that for the strategy pair $(f^N,f^N)$, $\psi_N(i) \geq 1-\epsilon_N$. Feqsibility of Strategy in Sectijn \[stratsubsubsec\] for Problem {#achievesymproof} ================================================================== Let ${{{T}}}$ be ghe sixolfst time index

in and applying Lemma \[stratcondlemma\] inductively, we {{\mathbb - \mathcal{C}_i({{\rho}}_1))]\\ \leq \sum_{j\neq i}(\tilde{\rho}_1(j))^s\exp(-sND^*(i) s^2NB^2/2).\label{mgf3}\end{aligned}$$ Chernoff Bound We use the Chernoff [@ross2014introduction] to conclude that $$\begin{aligned} {{\mathbb - \mathcal{C}_i({{\rho}}_1) < \theta]\\ & \leq {{\mathbb E}}^g_i\exp[-s(\mathcal{C}_i({{\rho}}_{N+1}) - \mathcal{C}_i({{\rho}}_1) -\theta)]\\ &\stackrel{a}{\leq} M\exp(s\theta-sND^*(i) s^2NB^2/2)\\ &\stackrel{b}{=} \epsilon_N.\end{aligned}$$ Inequality $(a)$ follows from the result in (\[mgf3\]). Equality $(b)$ obtained substituting = and $s = s_N$ where $$\begin{aligned} \theta_N &\doteq ND^*(i) - \frac{s_N NB^2}{2}- {\frac{1}{s_N}\log\frac{M}{\epsilon_N}},\label{thetadef}\end{aligned}$$ and $s_N$ is defined in. Under Assumptions \[epsassum\], one can easily that $\theta_N/ N \to as $N \to \infty$. Thus, have that for strategy $(f^N,g^N)$, \geq 1-\epsilon_N$. Feasibility Strategy in Section \[stratsubsubsec\] for Problem {#achievesymproof} ================================================================== Let ${{{T}}}$ be the smallest time index

in and applying Lemma \[stratcoNdlemma\] indUctivEly, We hAvE $$\begIn{alIgned} {{\mathbb E}}^g_i\EXp[&-s(\mAthcal{C}_i({{\rho}}_{N+1}) - \mathcal{C}_i({{\rHo}}_1))]\\ & \leq \SuM_{J\neq I}(\TiLde{\rhO}_1(j))^s\exp(-snd^*(i) + S^2nb^2/2)\\ &\leQ M\ExP(-sNd^*(i) + S^2nB^2/2).\Label{Mgf3}\End{aligNed}$$ ### ChernofF BoUnD We can use the cHeRnoff bound [@RosS2014introductioN] to ConcluDe ThaT $$\Begin{AliGned} {{\mAthbb P}}_I^G[\mathcAl{C}_i&({{\rho}}_{N+1}) - \mAtHCal{C}_i({{\rHO}}_1) < \theta]\\ & \lEQ {{\MaThbb e}}^g_i\exp[-s(\mathcal{C}_i({{\rHO}}_{N+1}) - \MAthcal{C}_i({{\rho}}_1) -\theTa)]\\ &\stacKrEL{a}{\LEQ} M\eXp(s\Theta-sND^*(i) + s^2nB^2/2)\\ &\StackREl{b}{=} \epsiLOn_n.\END{alIGned}$$ InequalitY $(a)$ follows frOM thE resulT iN (\[mgF3\]). equaliTy $(b)$ is ObTAinEd by substitUtinG $\theta = \theTa_N$ and $S = S_N$ where $$\BEgin{aliGned} \thEta_n &\doTeq Nd^*(I) - \fRaC{s_N nB^2}{2}- {\FRac{1}{S_n}\lOg\fRAc{M}{\Epsilon_N}},\LaBeL{thetAdef}\END{ALignEd}$$ aNd $s_N$ Is as dEfined in. Under assUmptIOns \[EpsasSum\], onE can EaSily vErify tHat $\thEtA_N/ N \to D^*(i)$ as $N \to \infTy$. ThUs, we have sHowN tHat FoR the sTRategy PaiR $(f^N,G^N)$, $\psi_N(i) \Geq 1-\epsiLOn_N$. feASIBiLity of Strategy in SeCtION \[sTratsubsUbsec\] fOR PRoBLem {#achieVeSymProoF} ================================================================== lEt ${{{T}}}$ be The sMAlLest time Index

in and applying Lemma \[s tratcondle mma\] in duc ti vely , we have $$\begin { alig ned} {{\mathbb E}}^g_i \exp[ &- s (\ma t hc al{C} _i({{\r h o} } _ {N+ 1} )- \ ma t hc al{C} _i( {{\rho} }_1))]\\ & \l eq \sum_{j\neq i} (\tilde{\r ho} _1(j))^s\exp (-s ND^*(i )+ s ^ 2NB^2 /2) \\ &\ leq M\ e xp(-sN D^*(i) +s^ 2 NB^2/2 ) .\label { m gf 3}\e nd{aligned}$$ ## # C h ernoff Bound We can u s et h e C her noff bound [ @ross 2 014intr o du c t i on] to conclude t hat $$\begi n {al igned} { {\m a thbb P }}_i^ g[ \ mat hcal{C}_i&( {{\r ho}}_{N+1 }) - \ m athcal{ C }_i({{\ rho}}_ 1)< \ thet a ]\ \& \ le q {{ \ ma thb b E} }^g_i\ex p[ -s (\mat hcal { C } _ i({{ \rh o}}_ {N+1} ) - \mathcal{ C}_ i({{ \ rho }}_1) -\th eta) ]\ \ &\s tackre l{a}{ \l eq} M\exp(s\the ta-s ND^*(i) + s^ 2N B^2 /2 )\\ & \ stackr el{ b}{ =} \eps ilon_N. \ end {a l i g ne d}$$ Inequality $( a) $ fo llows fr om the re su l t in (\[ mg f3\ ]).E q ualit y $( b )$ is obta ined b y s ub stituti ng $\the ta =\th eta_N $ and $s =s_N$ whe re $$ \ begin{aligned} \theta_N &\do t eq N D^ * (i)- \ frac{s_N NB ^2}{ 2 }- { \fra c {1 }{s _ N}\lo g\fra c{ M }{ \ epsilon_N}},\label{ th etadef }\end {aligned}$$ a nd $s_N$ i s a s define d in . U n der Assumption s \[e psassum\], one caneasil y verify that $\t h e ta_N/ N\to D^ *(i )$a s $ N \to \infty$ . Thus ,we have sh own tha t f orthe st ra tegy pair $(f^N,g ^N )$ ,$\ psi _N(i) \geq 1-\ ep sil on _N$ . Fe a sibili ty of Str at eg y in Sectio n \ [ s trat su bs ubse c\] f or Pr oble m { #achiev esymproof } = = ==== == == ======= ============= == ========== == === ====== = = ======== === Let ${{{T}}}$ be t h e small est time ind ex

in_and applying_Lemma \[stratcondlemma\] inductively, we_have $$\begin{aligned} {{\mathbb_E}}^g_i\exp[&-s(\mathcal{C}_i({{\rho}}_{N+1})_- \mathcal{C}_i({{\rho}}_1))]\\ &_\leq_\sum_{j\neq i}(\tilde{\rho}_1(j))^s\exp(-sND^*(i) +_s^2NB^2/2)\\ &\leq M\exp(-sND^*(i) +_s^2NB^2/2).\label{mgf3}\end{aligned}$$ ### Chernoff Bound We can_use the Chernoff_bound_[@ross2014introduction] to conclude that $$\begin{aligned} {{\mathbb P}}_i^g[\mathcal{C}_i&({{\rho}}_{N+1}) - \mathcal{C}_i({{\rho}}_1) < \theta]\\ & \leq {{\mathbb E}}^g_i\exp[-s(\mathcal{C}_i({{\rho}}_{N+1}) -_\mathcal{C}_i({{\rho}}_1)_-\theta)]\\ &\stackrel{a}{\leq} M\exp(s\theta-sND^*(i)_+_s^2NB^2/2)\\ &\stackrel{b}{=}_\epsilon_N.\end{aligned}$$ Inequality $(a)$ follows from_the result in (\[mgf3\]). Equality_$(b)$ is_obtained by substituting $\theta = \theta_N$ and $s_=_s_N$ where $$\begin{aligned} \theta_N_&\doteq ND^*(i) - \frac{s_N NB^2}{2}- {\frac{1}{s_N}\log\frac{M}{\epsilon_N}},\label{thetadef}\end{aligned}$$ and $s_N$ is_as defined in. Under Assumptions \[epsassum\],_one can easily_verify_that_$\theta_N/ N \to D^*(i)$_as $N \to \infty$. Thus, we_have shown that for the strategy_pair $(f^N,g^N)$, $\psi_N(i) \geq 1-\epsilon_N$. Feasibility of Strategy_in Section \[stratsubsubsec\] for Problem _{#achievesymproof} ================================================================== Let ${{{T}}}$ be the smallest_time index

1. No conjugates $hgh^{-1}$ of $g$ lie in a proper parabolic subgroup $P_\alpha \subsetneq GL_n$. 2. There are no nonzero proper $g$-stable ${{\mathbb{F}}}_q$-subspaces inside ${{\mathbb{F}}}_q^n$. 3. The characteristic polynomial $\det(xI_n-g)$ is irreducible in ${{\mathbb{F}}}_q[x]$. 4. The element $g$ is the image of some $\beta$ in ${{\mathbb{F}}}_{q^n}^\times$ satisfying ${{\mathbb{F}}}_q(\beta)={{\mathbb{F}}}_{q^n}$ (that is, a [*primitive element*]{} for ${{\mathbb{F}}}_{q^n}$) under one of the embeddings $ {{\mathbb{F}}}_{q^n}^\times \hookrightarrow GL_{{{\mathbb{F}}}_q}({{\mathbb{F}}}_{q^n}) \cong GL_n({{\mathbb{F}}}_q). $ The elements in $GL_n({{\mathbb{F}}}_q)$ satisfying these properties are called the [**regular elliptic elements**]{}. [(i) is equivalent to (ii).]{} A proper ${{\mathbb{F}}}_q$-subspace $U$, say with $\dim_{{{\mathbb{F}}}_q} U = d < n$, is $g$-stable if and only any $h$ in $GL_n({{\mathbb{F}}}_q)$ sending $U$ to the span of the first $d$ standard basis vectors in ${{\mathbb{F}}}_q^n$ has the property that $h g h^{-1}$ lies in a proper parabolic subgroup $P_{\alpha}$ with $\alpha_1=d$. .1in [(ii) implies (iii).]{} Argue the contrapositive: if $\det(xI_n-g)$ had a nonzero proper irreducible factor $f(x)$, then $\ker(f(g): V \rightarrow V)$ would be a nonzero proper $g$-stable subspace..1in [(iii) implies (iv).]{} If $f(x):=\det(xI_n-g)$ is irreducible in ${{\mathbb{F}}}_q[x]$, then $f(x)$ is also the minimal polynomial of $g$. Thus $g$ has rational canonical form over ${{\mathbb{F}}}_q$ equal to the companion matrix for $f(

1. No conjugates $ hgh^{-1}$ of $ g$ lie in a proper parabolic subgroup $ P_\alpha \subsetneq GL_n$. 2. There are no nonzero proper $ g$-stable $ { { \mathbb{F}}}_q$-subspaces inside $ { { \mathbb{F}}}_q^n$. 3. The characteristic polynomial $ \det(xI_n - g)$ is irreducible in $ { { \mathbb{F}}}_q[x]$. 4. The element $ g$ is the image of some $ \beta$ in $ { { \mathbb{F}}}_{q^n}^\times$ meet $ { { \mathbb{F}}}_q(\beta)={{\mathbb{F}}}_{q^n}$ (that is, a [ * crude element * ] { } for $ { { \mathbb{F}}}_{q^n}$) under one of the embeddings $ { { \mathbb{F}}}_{q^n}^\times \hookrightarrow GL_{{{\mathbb{F}}}_q}({{\mathbb{F}}}_{q^n }) \cong GL_n({{\mathbb{F}}}_q). $ The elements in $ GL_n({{\mathbb{F}}}_q)$ satisfying these property are called the [ * * regular egg-shaped elements * * ] { }. [ (i) is equivalent to (ii). ] { } A proper $ { { \mathbb{F}}}_q$-subspace $ U$, say with $ \dim_{{{\mathbb{F}}}_q } uracil = d < n$, is $ g$-stable if and only any $ h$ in $ GL_n({{\mathbb{F}}}_q)$ transport $ U$ to the span of the first $ d$ standard footing vectors in $ { { \mathbb{F}}}_q^n$ has the property that $ h g h^{-1}$ lies in a proper parabolic subgroup $ P_{\alpha}$ with $ \alpha_1 = d$. .1 in [ (ii) implies (iii). ] { } Argue the contrapositive: if $ \det(xI_n - g)$ had a nonzero proper irreducible factor $ f(x)$, then $ \ker(f(g ): V \rightarrow V)$ would be a nonzero proper $ g$-stable subspace.. 1 in [ (iii) imply (iv). ] { } If $ f(x):=\det(xI_n - g)$ is irreducible in $ { { \mathbb{F}}}_q[x]$, then $ f(x)$ is also the minimal polynomial of $ g$. Thus $ g$ has intellectual canonical form over $ { { \mathbb{F}}}_q$ adequate to the companion matrix for $ f (

1. Nl conjugates $hgh^{-1}$ of $g$ lit in a proper parcvolic vubgrohp $P_\alphx \subsetneq GL_n$. 2. There arx no nonztgo proper $g$-stable ${{\matfbb{F}}}_q$-subsiaces insude ${{\nqthbb{F}}}_q^n$. 3. Vge characteriabic pmoynomial $\det(xI_k-g)$ is irredgcible in ${{\mathtb{W}}}_q[r]$. 4. The element $g$ is the image of somq $\beta$ on ${{\mathbb{F}}}_{q^n}^\timef$ sauissyinf ${{\mathbb{F}}}_q(\beta)={{\mathbb{F}}}_{q^n}$ (that is, a [*lrimitine element*]{} for ${{\mayhbb{F}}}_{q^n}$) under one of the elbedfings $ {{\mathbb{F}}}_{q^j}^\times \hooktjgheqrrow GL_{{{\mathcb{F}}}_q}({{\mathbb{F}}}_{q^n}) \cong GL_n({{\jathbb{F}}}_q). $ The elements in $GL_n({{\oathby{F}}}_q)$ satisfyunt tjgse propertixs are called the [**vvgular alliptiv elements**]{}. [(i) is eqnivaoent to (ii).]{} A proper ${{\methbb{F}}}_q$-subspace $U$, sai with $\dim_{{{\kachbb{F}}}_q} U = d < n$, is $g$-stqboe if and onlh anh $h$ ii $GM_n({{\mathhb{F}}}_s)$ sending $H$ to the spqn of the first $d$ suanqqrd basis vecfors ig ${{\iathbb{F}}}_q^n$ has the property that $h g h^{-1}$ lpes jn a proper parabolic sybgroup $P_{\alpha}$ with $\appha_1=d$. .1in [(iy) implies (iii).]{} Argue the contrapositive: if $\det(xI_n-g)$ had e voneevj prlper irreducible factor $f(x)$, then $\ker(f(g): V \rightwdrpw V)$ would be a konzero proper $g$-stsbpe fubspace..1in [(iik) implnss (iv).]{} If $f(x):=\det(xI_n-g)$ id irredocible in ${{\mathbf{F}}}_q[x]$, then $f(x)$ is also the minimao polynomial if $g$. Thus $g$ has racional canonncal fprm ofer ${{\mathbb{F}}}_q$ equal to tke comlanion matrlx for $f(

1. No conjugates $hgh^{-1}$ of $g$ lie proper subgroup $P_\alpha GL_n$. 2. There ${{\mathbb{F}}}_q$-subspaces ${{\mathbb{F}}}_q^n$. 3. The polynomial $\det(xI_n-g)$ is in ${{\mathbb{F}}}_q[x]$. 4. The element $g$ the image of some $\beta$ in ${{\mathbb{F}}}_{q^n}^\times$ satisfying ${{\mathbb{F}}}_q(\beta)={{\mathbb{F}}}_{q^n}$ (that is, a [*primitive for ${{\mathbb{F}}}_{q^n}$) under one of the embeddings $ {{\mathbb{F}}}_{q^n}^\times \hookrightarrow GL_{{{\mathbb{F}}}_q}({{\mathbb{F}}}_{q^n}) \cong GL_n({{\mathbb{F}}}_q). The in satisfying properties are called the [**regular elliptic elements**]{}. [(i) is equivalent to (ii).]{} A proper ${{\mathbb{F}}}_q$-subspace $U$, with $\dim_{{{\mathbb{F}}}_q} U = d < n$, is if and only any in $GL_n({{\mathbb{F}}}_q)$ sending $U$ to span the first standard vectors ${{\mathbb{F}}}_q^n$ has the that $h g h^{-1}$ lies in a proper parabolic subgroup $P_{\alpha}$ with $\alpha_1=d$. .1in [(ii) implies (iii).]{} the contrapositive: had a proper factor then $\ker(f(g): V would be a nonzero proper $g$-stable (iv).]{} If $f(x):=\det(xI_n-g)$ is irreducible in ${{\mathbb{F}}}_q[x]$, then is also minimal polynomial of $g$. Thus $g$ rational canonical form over ${{\mathbb{F}}}_q$ equal to the matrix for $f(

1. No conjugates $hgh^{-1}$ of $g$ lie in a pRoper parabOlic sUbgRouP $P_\AlphA \subSetneq GL_n$. 2. There ARe no Nonzero proper $g$-stable ${{\maThbb{F}}}_Q$-sUBspaCEs InsidE ${{\mathbb{f}}}_Q^n$. 3. tHE chArAcTerIsTIc PolynOmiAl $\det(xI_N-g)$ is irreduCibLe In ${{\mathbb{F}}}_q[x]$. 4. THE eLement $g$ is tHe iMage of some $\beTa$ iN ${{\mathbB{F}}}_{Q^n}^\tIMes$ saTisFying ${{\Mathbb{f}}}_Q(\beta)={{\mAthbb{F}}}_{q^n}$ (tHaT Is, a [*priMItive elEMEnT*]{} for ${{\Mathbb{F}}}_{q^n}$) under one OF tHE embeddings $ {{\matHbb{F}}}_{q^n}^\TiMEs \HOOkrIghTarrow GL_{{{\maThBb{F}}}_q}({{\mAThbb{F}}}_{q^n}) \COnG gl_N({{\maTHbb{F}}}_q). $ The elemeNts in $GL_n({{\matHBb{F}}}_Q)$ satisFyIng THese prOpertIeS Are Called the [**reGulaR elliptic ElemenTS**]{}. [(i) is equIValent tO (ii).]{} A prOpeR ${{\maThbb{f}}}_Q$-sUbSpaCe $u$, Say WItH $\diM_{{{\MatHbb{F}}}_q} U = d < n$, Is $G$-sTable If anD ONLY any $H$ in $gL_n({{\mAthbb{f}}}_q)$ sending $U$ to tHe sPan oF The First $D$ stanDard BaSis veCtors iN ${{\mathBb{f}}}_q^n$ has the properTy thAt $h g h^{-1}$ lies In a PrOpeR pAraboLIc subgRouP $P_{\aLpha}$ witH $\alpha_1=d$. .1IN [(ii) ImPLIEs (Iii).]{} Argue the contrapOsITIvE: if $\det(xI_N-g)$ had a NOnZeRO proper iRrEduCiblE FActor $F(x)$, thEN $\kEr(f(g): V \rigHtarroW v)$ wOuLd be a noNzEro proPeR $g$-sTabLe subSPace..1In [(iii) iMplies (iv).]{} if $f(x):=\dET(xI_n-g)$ is irreducIBle in ${{\mathbb{F}}}_q[X]$, ThEN $F(x)$ IS alsO thE minimal polYnomIAl of $G$. ThuS $G$ hAs rATionaL canoNiCAl FOrm over ${{\mathbb{F}}}_q$ equaL tO the coMpaniOn matrix for $f(

1. No conjugates $hgh^{-1 }$ of $g$lie i n a pr op er p arab olic subgroup$ P_\a lpha \subsetneq GL _n$. 2 . Th e re areno nonz e ro p rop er $ g$- st a bl e ${{ \ma thbb{F} }}_q$-subs pac es inside ${{\ m at hbb{F}}}_q ^n$ . 3. The c har acteri st icp olyno mia l $\d et(xI_ n -g)$ i s irreduc ib l e in $ { {\mathb b { F} }}_q [x]$. 4. The el e me n t $g$ is the i mage o fs om e $\b eta $ in ${{\m at hbb{F } }}_{q^n } ^\ t i m es$ satisfying ${ {\mathbb{F} } }_q (\beta )= {{\ m athbb{ F}}}_ {q ^ n}$ (that is,a [* primitive eleme n t*]{} f o r ${{\m athbb{ F}} }_{ q^n} $ )un der o n e o f t hee mbe ddings $ {{\m athb b { F } }}_{ q^n }^\t imes\hookrightarr owGL_{ { {\m athbb {F}}} _q}( {{ \math bb{F}} }_{q^ n} ) \cong GL_n({{ \mat hbb{F}}}_ q). $ Thee lement s i n $ GL_n({{ \mathbb { F}} }_ q ) $ s atisfying these pr op e r ti es are c alledt he [ * *regular e lli ptic e lemen ts** ] {} . [(i)is equ i va le nt to ( ii ).]{}Apro per ${{\ m athb b{F}}} _q$-subs pace$ U$, say with $ \ dim_{{{\mathb b {F } } }_ q } U= d < n$, is $ g$-s t able ifa nd on l y any $h$in $G L _n({{\mathbb{F}}}_q )$ sendi ng $U $ to the span of the fi r s t $d$ sta ndar d b a sis vectors in ${{\ mathbb{F}} } _q^n$ ha s the propert y that $h g h^{-1}$ li esina p r o pe r parabolic s u b grou p$P_{\al pha }$ with $\ alp ha_ 1=d $. .1in [( ii) impl ie s(i ii ).] {} Ar g ue the c on tra po sit ive:i f $\de t(xI_ n-g) $ha d anonzero pr o p er i rr ed ucib lefa ctor$f(x ) $,then $\ ker(f(g): V\ righ ta rr ow V)$would be a no nz ero proper $ g$- stable s ubspace. .1in [(iii) implies (iv ) .]{} If $f (x):= \det (xI_n-g)$ is irred uci b le in${{\ma thbb{ F} }}_ q [ x]$,t h en $f (x )$ is also t heminim al pol ynomial of $g$. Thus $g$h asrational cano nic al f o r move r $ { {\m at h bb{ F } }}_q$ equal tothe compan io n m atrix for$ f(

1. _No conjugates_$hgh^{-1}$ of $g$ lie_in a_proper_parabolic subgroup_$P_\alpha _ _\subsetneq GL_n$. 2. _There are no nonzero_proper $g$-stable ${{\mathbb{F}}}_q$-subspaces_inside_${{\mathbb{F}}}_q^n$. 3. The characteristic polynomial $\det(xI_n-g)$ is irreducible in ${{\mathbb{F}}}_q[x]$. 4. The element $g$_is_the image_of_some_$\beta$ in ${{\mathbb{F}}}_{q^n}^\times$ satisfying ${{\mathbb{F}}}_q(\beta)={{\mathbb{F}}}_{q^n}$_(that is, a [*primitive element*]{}_for ${{\mathbb{F}}}_{q^n}$)_under one of the embeddings $ __{{\mathbb{F}}}_{q^n}^\times \hookrightarrow GL_{{{\mathbb{F}}}_q}({{\mathbb{F}}}_{q^n})_\cong GL_n({{\mathbb{F}}}_q). $ The elements in $GL_n({{\mathbb{F}}}_q)$_satisfying these properties are called the_[**regular elliptic elements**]{}. [(i)_is_equivalent_to (ii).]{} A proper_${{\mathbb{F}}}_q$-subspace $U$, say with $\dim_{{{\mathbb{F}}}_q} U_= d < n$, is $g$-stable_if and only any $h$ in $GL_n({{\mathbb{F}}}_q)$_sending $U$ to the span of_the first $d$ standard basis_vectors in_${{\mathbb{F}}}_q^n$ has the property that_$h g h^{-1}$_lies in_a proper parabolic_subgroup $P_{\alpha}$ with $\alpha_1=d$. .1in [(ii) implies_(iii).]{} Argue the_contrapositive: if $\det(xI_n-g)$ had a nonzero_proper_irreducible factor $f(x)$,_then_$\ker(f(g):_V \rightarrow_V)$ would be_a_nonzero proper_$g$-stable_subspace..1in [(iii) implies (iv).]{} If $f(x):=\det(xI_n-g)$_is_irreducible in ${{\mathbb{F}}}_q[x]$, then $f(x)$ is also_the minimal polynomial of_$g$._Thus $g$ has rational_canonical form over ${{\mathbb{F}}}_q$ equal_to the companion matrix for $f(

langle t^{\text{Ga}}_{\sigma \sigma} \rangle_{T}+ c\langle t^{\text{Mn}}_{\sigma \sigma} \rangle_{T}=0\end{aligned}$$ $\langle... \rangle_{\text{dis},T}$ denotes configuration average and thermal average at temperature $T$ for the spin operator, $c$ is the concentration of Mn impurities. After lengthy calculations we get, $$\begin{aligned} \Sigma_{\sigma}= \frac{\Large \langle \sum_{\lambda} x_{\lambda}\hat{\bf c}^{\lambda}_{\sigma}(\hat{\bf d}^{\lambda}_{\sigma})^{-1} \Large \rangle_{T}} {\Large \langle \sum_{\lambda} x_{\lambda}(\hat{\bf d}^{\lambda}_{\sigma})^{-1} \Large \rangle_{T}} \label{self}\end{aligned}$$ The sum runs over all constituents, in our binary system $x_{\text{Mn}}=c$ (resp. $x_{\text{Ga}}=1-c$). $$\begin{aligned} {\hat{\bf c}}^{\text{Mn}}_{\sigma}&=& z_{\sigma}\frac{J}{2}\hat{\bf S}^{z}+\large( \frac{J^{2}}{4} [S(S+1)-(\hat{\bf S}^{z})^2 -z_{\sigma}\hat{\bf S}^{z}] \nonumber \\ &&+ [z_{\sigma}\frac{J}{2}\hat{\bf S}^{z}+\Sigma_{\sigma}][\frac{J}{2}(z_{-\sigma}\hat{\bf S}^{z}-1 )-\Sigma_{-\sigma}] \Large) G_{-\sigma}\end{aligned}$$ and, $$\begin{aligned} {\hat{\bf d}}^{\text{Mn}}_{\sigma}&=&1-G_{-\sigma}[\frac{J}{2}(z_{-\sigma}\hat{\bf S}^{z}-1)-\Sigma_{-\sigma}] \nonumber \\ &&- G_{\sigma}[{\hat{\bf c}}^{\text{Mn}}_{\sigma}-z_{\sigma}\frac{J}{2}\hat{\bf S}^{z}]\end{aligned}$$ with $z_{\sigma}=1$ (resp.$-1$) for spin $\uparrow$ (resp.$\downarrow$). $\bar{G}_{\sigma} = \frac{1}{N} \sum_{\bf q} \bar{G}_{\

langle t^{\text{Ga}}_{\sigma \sigma } \rangle_{T}+ c\langle t^{\text{Mn}}_{\sigma \sigma } \rangle_{T}=0\end{aligned}$$ $ \langle... \rangle_{\text{dis},T}$ denotes configuration average and thermal average at temperature $ T$ for the spin operator, $ c$ is the concentration of Mn impurity. After drawn-out calculations we get, $ $ \begin{aligned } \Sigma_{\sigma}= \frac{\Large \langle \sum_{\lambda } x_{\lambda}\hat{\bf c}^{\lambda}_{\sigma}(\hat{\bf d}^{\lambda}_{\sigma})^{-1 } \Large \rangle_{T } } { \Large \langle \sum_{\lambda } x_{\lambda}(\hat{\bf d}^{\lambda}_{\sigma})^{-1 } \Large \rangle_{T } } \label{self}\end{aligned}$$ The sum run over all constituent, in our binary system $ x_{\text{Mn}}=c$ (resp. $ x_{\text{Ga}}=1 - c$). $ $ \begin{aligned } { \hat{\bf c}}^{\text{Mn}}_{\sigma}&= & z_{\sigma}\frac{J}{2}\hat{\bf S}^{z}+\large (\frac{J^{2}}{4 } [ S(S+1)-(\hat{\bf S}^{z})^2 -z_{\sigma}\hat{\bf S}^{z } ] \nonumber \\ & & + [ z_{\sigma}\frac{J}{2}\hat{\bf S}^{z}+\Sigma_{\sigma}][\frac{J}{2}(z_{-\sigma}\hat{\bf S}^{z}-1 ) -\Sigma_{-\sigma } ] \Large) G_{-\sigma}\end{aligned}$$ and, $ $ \begin{aligned } { \hat{\bf d}}^{\text{Mn}}_{\sigma}&=&1 - G_{-\sigma}[\frac{J}{2}(z_{-\sigma}\hat{\bf S}^{z}-1)-\Sigma_{-\sigma } ] \nonumber \\ & & - G_{\sigma}[{\hat{\bf c}}^{\text{Mn}}_{\sigma}-z_{\sigma}\frac{J}{2}\hat{\bf S}^{z}]\end{aligned}$$ with $ z_{\sigma}=1 $ (resp.$-1 $) for spin $ \uparrow$ (resp.$\downarrow$). $ \bar{G}_{\sigma } = \frac{1}{N } \sum_{\bf q } \bar{G}_{\

lanhle t^{\text{Ga}}_{\sigma \sigma} \rakgle_{T}+ c\langle t^{\terr{Mn}}_{\sigka \sigja} \rangld_{T}=0\end{aligned}$$ $\langle... \rangle_{\texv{dis},R}$ denites configuration avefage and nhermal aceraje at temperaturx $T$ for bke spjk opexavor, $c$ is the cokcentration of Mn impurither. Ayter lengthy calculations we get, $$\begyn{alignrd} \Digma_{\sigma}= \fras{\Larbq \lahgle \sum_{\lambda} x_{\lambda}\hat{\bf c}^{\lambda}_{\aigma}(\hau{\bf d}^{\lambda}_{\sigma})^{-1} \Latge \rangle_{T}} {\Large \langle \sul_{\lamhda} x_{\lambda}(\hat{\bf d}^{\pambda}_{\sigma})^{-1} \Oargq \rangle_{T}} \labdl{self}\end{aligned}$$ The som runs over all constituents, in our yinary systgm $x_{\hgxt{Mn}}=c$ (resp. $e_{\text{Gw}}=1-c$). $$\begin{aligkvd} {\hat{\bf c}}^{\text{Mm}}_{\sigma}&=& z_{\sigma}\frsc{J}{2}\iat{\bd S}^{z}+\large( \frac{J^{2}}{4} [S(S+1)-(\hat{\bh S}^{z})^2 -z_{\sigma}\hat{\bf S}^{z}] \nonumber \\ &&+ [v_{\sigma}\frac{J}{2}\hat{\bf S}^{z}+\Signa_{\wigma}][\xrac{B}{2}(z_{-\siena}\hxt{\br X}^{z}-1 )-\Aigma_{-\slgme}] \Large) G_{-\aigma}\end{alitned}$$ and, $$\begin{aligntd} {\hwn{\nf d}}^{\text{Mn}}_{\sifma}&=&1-G_{-\sidmw}[\frac{J}{2}(z_{-\sigma}\hat{\bf S}^{z}-1)-\Sigma_{-\sigma}] \nonumber \\ &&- F_{\sigma}[{\hat{\bf c}}^{\text{Mn}}_{\sigma}-z_{\sigma}\frac{J}{2}\hat{\bf S}^{z}]\enf{aligned}$$ rith $z_{\sigma}=1$ (resp.$-1$) for spin $\uparrow$ (resp.$\downarrow$). $\bdr{G}_{\sijmx} = \yvwz{1}{B} \dum_{\bf q} \bar{G}_{\

langle t^{\text{Ga}}_{\sigma \sigma} \rangle_{T}+ c\langle t^{\text{Mn}}_{\sigma \sigma} \rangle_{\text{dis},T}$ configuration average thermal average at operator, is the concentration Mn impurities. After calculations we get, $$\begin{aligned} \Sigma_{\sigma}= \frac{\Large \sum_{\lambda} x_{\lambda}\hat{\bf c}^{\lambda}_{\sigma}(\hat{\bf d}^{\lambda}_{\sigma})^{-1} \Large \rangle_{T}} {\Large \langle \sum_{\lambda} x_{\lambda}(\hat{\bf d}^{\lambda}_{\sigma})^{-1} \Large \rangle_{T}} The sum runs over all constituents, in our binary system $x_{\text{Mn}}=c$ (resp. $x_{\text{Ga}}=1-c$). {\hat{\bf z_{\sigma}\frac{J}{2}\hat{\bf \frac{J^{2}}{4} S}^{z})^2 -z_{\sigma}\hat{\bf S}^{z}] \nonumber \\ &&+ [z_{\sigma}\frac{J}{2}\hat{\bf S}^{z}+\Sigma_{\sigma}][\frac{J}{2}(z_{-\sigma}\hat{\bf S}^{z}-1 )-\Sigma_{-\sigma}] \Large) G_{-\sigma}\end{aligned}$$ and, $$\begin{aligned} {\hat{\bf d}}^{\text{Mn}}_{\sigma}&=&1-G_{-\sigma}[\frac{J}{2}(z_{-\sigma}\hat{\bf \nonumber \\ &&- G_{\sigma}[{\hat{\bf c}}^{\text{Mn}}_{\sigma}-z_{\sigma}\frac{J}{2}\hat{\bf S}^{z}]\end{aligned}$$ with $z_{\sigma}=1$ for spin $\uparrow$ (resp.$\downarrow$). = \frac{1}{N} \sum_{\bf q} \bar{G}_{\

langle t^{\text{Ga}}_{\sigma \sigma} \raNgle_{T}+ c\langLe t^{\teXt{MN}}_{\siGmA \sigMa} \raNgle_{T}=0\end{aligneD}$$ $\LangLe... \rangle_{\text{dis},T}$ denoteS confIgURatiON aVeragE and theRMaL AVerAgE aT teMpERaTure $T$ For The spin Operator, $c$ iS thE cOncentration OF MN impuritieS. AfTer lengthy caLcuLationS wE geT, $$\Begin{AliGned} \SIgma_{\siGMa}= \frac{\large \langLe \SUm_{\lambDA} x_{\lambdA}\HAt{\Bf c}^{\lAmbda}_{\sigma}(\hat{\bf d}^{\lAMbDA}_{\sigma})^{-1} \Large \ranGle_{T}} {\LaRgE \LaNGLe \sUm_{\lAmbda} x_{\lambDa}(\Hat{\bf D}^{\Lambda}_{\sIGmA})^{-1} \lARge \RAngle_{T}} \label{seLf}\end{aligneD}$$ the Sum runS oVer ALl consTitueNtS, In oUr binary sysTem $x_{\Text{Mn}}=c$ (reSp. $x_{\texT{ga}}=1-c$). $$\begiN{Aligned} {\Hat{\bf c}}^{\TexT{Mn}}_{\SigmA}&=& Z_{\sIgMa}\fRaC{j}{2}\haT{\Bf s}^{z}+\lARge( \Frac{J^{2}}{4} [S(S+1)-(\hAt{\Bf s}^{z})^2 -z_{\siGma}\hAT{\BF s}^{z}] \noNumBer \\ &&+ [z_{\Sigma}\Frac{J}{2}\hat{\bf S}^{z}+\SIgmA_{\sigMA}][\frAc{J}{2}(z_{-\sIgma}\hAt{\bf s}^{z}-1 )-\sigma_{-\Sigma}] \LArge) G_{-\SiGma}\end{aligned}$$ anD, $$\begIn{aligned} {\Hat{\Bf D}}^{\teXt{mn}}_{\sigMA}&=&1-G_{-\sigmA}[\frAc{J}{2}(Z_{-\sigma}\hAt{\bf S}^{z}-1)-\SIGma_{-\SiGMA}] \NoNumber \\ &&- G_{\sigma}[{\hat{\bf c}}^{\TeXT{mn}}_{\Sigma}-z_{\siGma}\fraC{j}{2}\hAt{\BF S}^{z}]\end{alIgNed}$$ With $Z_{\SIgma}=1$ (rEsp.$-1$) fOR sPin $\uparrOw$ (resp.$\DOwNaRrow$). $\bar{g}_{\sIgma} = \frAc{1}{n} \suM_{\bf Q} \bar{G}_{\

langle t^{\text{Ga}}_{\sig ma \sigma} \ran gle _{T }+ c\l angl e t^{\text{Mn} } _{\s igma \sigma} \rangle_{ T}=0\ en d {ali g ne d}$$ $\lang l e. . . \r an gl e_{ \t e xt {dis} ,T} $ denot es configu rat io n average an d t hermal ave rag e at tempera tur e $T$fo r t h e spi n o perat or, $c $ is th e concent ra t ion of Mn impu r i ti es. After lengthy ca l cu l ations we get, $$\be gi n {a l i gne d}\Sigma_{\s ig ma}=\ frac{\L a rg e \ lan g le \sum_{\lam bda} x_{\la m bda }\hat{ \b f c } ^{\lam bda}_ {\ s igm a}(\hat{\bf d}^ {\lambda} _{\sig m a})^{-1 } \Large \rang le_ {T} } {\ L ar ge \l an g le\ su m_{ \ lam bda} x_{ \l am bda}( \hat { \ b f d}^ {\l ambd a}_{\ sigma})^{-1}\La rge\ ran gle_{ T}} \ labe l{ self} \end{a ligne d} $$ The sum runs ove r all con sti tu ent s, in o u r bina rysys tem $x_ {\text{ M n}} =c $ ( re sp. $x_{\text{Ga}} =1 - c $) . $$\beg in{ali g ne d} {\hat{\b fc}} ^{\t e x t{Mn} }_{\ s ig ma}&=& z _{\sig m a} \f rac{J}{ 2} \hat{\ bf S} ^{z }+\la r ge(\frac{ J^{2}}{4 } [S( S +1)-(\hat{\bfS }^{z})^2 -z_{ \ si g m a} \ hat{ \bf S}^{z}] \ nonu m ber\\ & & +[z_ { \sigm a}\fr ac { J} { 2}\hat{\bf S}^{z}+\ Si gma_{\ sigma }][\frac{J}{2 }(z_{-\sig m a } \hat{\bf S}^ { z} - 1 )-\Sigma_{-\ sigma }] \Larg e ) G_{-\s igma} \end{ali gned}$$ a n d , $$\beg in{ ali gne d}{ \ ha t{\bf d}}^{\t e x t{Mn }} _{\sigm a}& =&1-G_{ -\s igm a}[ \fr ac {J}{2}(z_ {-\sigma }\ ha t{ \b f S }^{z} - 1)-\Sigm a_ {-\ si gma }] \n o number \\&&-G_ {\ s igm a}[{\ha t {\ b f c}} ^{ \t ext{ Mn} }_ {\sig ma}- z _{\ sigma}\ frac{J}{2 }\h a t{\b fS} ^{z}]\e nd{aligned}$$ w ith $z_{\s ig ma} =1$ (r e s p.$-1$)for spin $\uparrow$ (re s p.$\dow nar row$) . $\ bar{G}_{\ sig ma} =\fr a c{1}{N } \sum _{\bf q } \ b a r{G}_ { \

langle t^{\text{Ga}}_{\sigma \sigma}_\rangle_{T}+ c\langle t^{\text{Mn}}_{\sigma_\sigma} \rangle_{T}=0\end{aligned}$$ $\langle... \rangle_{\text{dis},T}$ denotes_configuration average_and_thermal average_at_temperature $T$ for_the spin operator,_$c$ is the concentration_of Mn impurities. After_lengthy_calculations we get, $$\begin{aligned} \Sigma_{\sigma}= \frac{\Large \langle \sum_{\lambda} x_{\lambda}\hat{\bf c}^{\lambda}_{\sigma}(\hat{\bf d}^{\lambda}_{\sigma})^{-1} \Large \rangle_{T}} {\Large \langle \sum_{\lambda} x_{\lambda}(\hat{\bf d}^{\lambda}_{\sigma})^{-1} \Large_\rangle_{T}}_\label{self}\end{aligned}$$ The_sum_runs_over all constituents, in our_binary system $x_{\text{Mn}}=c$ (resp. $x_{\text{Ga}}=1-c$)._$$\begin{aligned} {\hat{\bf c}}^{\text{Mn}}_{\sigma}&=& z_{\sigma}\frac{J}{2}\hat{\bf_S}^{z}+\large( \frac{J^{2}}{4} [S(S+1)-(\hat{\bf S}^{z})^2 -z_{\sigma}\hat{\bf S}^{z}] \nonumber \\ &&+_[z_{\sigma}\frac{J}{2}\hat{\bf S}^{z}+\Sigma_{\sigma}][\frac{J}{2}(z_{-\sigma}\hat{\bf_S}^{z}-1 )-\Sigma_{-\sigma}] _\Large) G_{-\sigma}\end{aligned}$$ and, $$\begin{aligned} {\hat{\bf d}}^{\text{Mn}}_{\sigma}&=&1-G_{-\sigma}[\frac{J}{2}(z_{-\sigma}\hat{\bf S}^{z}-1)-\Sigma_{-\sigma}] \nonumber \\ &&-_G_{\sigma}[{\hat{\bf c}}^{\text{Mn}}_{\sigma}-z_{\sigma}\frac{J}{2}\hat{\bf S}^{z}]\end{aligned}$$ with $z_{\sigma}=1$ (resp.$-1$) for_spin $\uparrow$ (resp.$\downarrow$)._$\bar{G}_{\sigma}_=_\frac{1}{N} \sum_{\bf q} \bar{G}_{\

+\epsilon_3}\right) + s_1(s,\,\epsilon_1,\,\epsilon_2,\,\epsilon_3) \sin\left(\frac{2\pi \epsilon_1} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\\ &+ s_1(s,\,\epsilon_2,\,\epsilon_3,\,\epsilon_1) \sin\left(\frac{2\pi \epsilon_2} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right) + s_1(s,\,\epsilon_3,\,\epsilon_1,\,\epsilon_2) \sin\left(\frac{2\pi \epsilon_3} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\\ &+ s_2(s,\,\epsilon_1,\,\epsilon_2,\,\epsilon_3) \sin\left(\frac{2\pi (\epsilon_1+\epsilon_2)} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right) + s_2(s,\,\epsilon_2,\,\epsilon_3,\,\epsilon_1) \sin\left(\frac{2\pi (\epsilon_2+\epsilon_3)} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\\ &+ s_2(s,\,\epsilon_3,\,\epsilon_1,\,\epsilon_2) \sin\left(\frac{2\pi (\epsilon_3+\epsilon_1)} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right) + s_3(s,\,\epsilon_1,\,\epsilon_2,\,\epsilon_3) \sin\left(\frac{2\pi (\epsilon_1+\epsilon_2+\epsilon_3)} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\Bigg)\,, \end{split}$$ where $$\notag \begin{split} &c_1(s,\,x,\,y,\,z)\\ = &4 x^2 \big( 3 s^6+6 s^5 x+9 s^5 y+9 s^5 z+3 s^4 x^2 +17 s^4 x y+17 s^4 x z+6 s^4 y^2+

+ \epsilon_3}\right) + s_1(s,\,\epsilon_1,\,\epsilon_2,\,\epsilon_3) \sin\left(\frac{2\pi \epsilon_1 } { s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\\ & + s_1(s,\,\epsilon_2,\,\epsilon_3,\,\epsilon_1) \sin\left(\frac{2\pi \epsilon_2 } { s+\epsilon_1+\epsilon_2+\epsilon_3}\right) + s_1(s,\,\epsilon_3,\,\epsilon_1,\,\epsilon_2) \sin\left(\frac{2\pi \epsilon_3 } { s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\\ & + s_2(s,\,\epsilon_1,\,\epsilon_2,\,\epsilon_3) \sin\left(\frac{2\pi (\epsilon_1+\epsilon_2) } { s+\epsilon_1+\epsilon_2+\epsilon_3}\right) + s_2(s,\,\epsilon_2,\,\epsilon_3,\,\epsilon_1) \sin\left(\frac{2\pi (\epsilon_2+\epsilon_3) } { s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\\ & + s_2(s,\,\epsilon_3,\,\epsilon_1,\,\epsilon_2) \sin\left(\frac{2\pi (\epsilon_3+\epsilon_1) } { s+\epsilon_1+\epsilon_2+\epsilon_3}\right) + s_3(s,\,\epsilon_1,\,\epsilon_2,\,\epsilon_3) \sin\left(\frac{2\pi (\epsilon_1+\epsilon_2+\epsilon_3) } { s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\Bigg)\, , \end{split}$$ where $ $ \notag \begin{split } & c_1(s,\,x,\,y,\,z)\\ = & 4 x^2 \big ( 3 s^6 + 6 s^5 x+9 s^5 y+9 s^5 z+3 s^4 x^2 +17 s^4 x y+17 s^4 x z+6 s^4 y^2 +

+\epsllon_3}\right) + s_1(s,\,\epsilon_1,\,\epsilok_2,\,\epsilon_3) \sin\left(\ftax{2\pi \epvilon_1} {s+\spsilon_1+\eosilon_2+\epsilon_3}\right)\\ &+ s_1(s,\,\epsilon_2,\,\e'silin_3,\,\epsulon_1) \sin\left(\frac{2\pi \epsiuon_2} {s+\epsilln_1+\epsilob_2+\epsmlon_3}\right) + s_1(s,\,\epsilon_3,\,\epsilok_1,\,\zpsilkk_2) \sin\lzfv(\frac{2\pi \epsilon_3} {x+\epsilon_1+\epvilon_2+\epsilon_3}\richg)\\ &+ s_2(d,\,\epsilon_1,\,\epsilon_2,\,\epsilon_3) \sin\left(\frac{2\pi (\epsilom_1+\eosilon_2)} {s+\epsilon_1+\gpsilpg_2+\epsjlon_3}\right) + s_2(s,\,\epsilon_2,\,\epsilon_3,\,\epsilon_1) \sih\left(\frec{2\pi (\epsilon_2+\epsolon_3)} {s+\epsilon_1+\epsilon_2+\epsilon_3}\gighh)\\ &+ s_2(s,\,\epsilon_3,\,\epsilon_1,\,\eosilon_2) \sin\ledt(\frwx{2\pi (\epsilon_3+\eosilon_1)} {s+\epspnon_1+\epsilon_2+\gpsilon_3}\right) + s_3(s,\,\epsilon_1,\,\epsilon_2,\,\epsiuon_3) \siu\left(\frac{2\pi (\epsipmn_1+\epsilon_2+\epwilon_3)} {f+\epsilon_1+\epsilon_2+\epsilmn_3}\right)\Nigg)\,, \end{split}$$ wmere $$\iotat \begin{split} &c_1(s,\,x,\,y,\,z)\\ = &4 x^2 \bij( 3 s^6+6 s^5 x+9 s^5 y+9 s^5 z+3 s^4 x^2 +17 f^4 x y+17 s^4 x z+6 s^4 y^2+

+\epsilon_3}\right) + s_1(s,\,\epsilon_1,\,\epsilon_2,\,\epsilon_3) \sin\left(\frac{2\pi \epsilon_1} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\\ &+ \epsilon_2} + s_1(s,\,\epsilon_3,\,\epsilon_1,\,\epsilon_2) \epsilon_3} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\\ &+ s_2(s,\,\epsilon_2,\,\epsilon_3,\,\epsilon_1) (\epsilon_2+\epsilon_3)} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\\ &+ \sin\left(\frac{2\pi (\epsilon_3+\epsilon_1)} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right) s_3(s,\,\epsilon_1,\,\epsilon_2,\,\epsilon_3) \sin\left(\frac{2\pi (\epsilon_1+\epsilon_2+\epsilon_3)} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\Bigg)\,, \end{split}$$ where \begin{split} &c_1(s,\,x,\,y,\,z)\\ = &4 x^2 \big( 3 s^6+6 s^5 x+9 s^5 y+9 s^5 s^4 x^2 +17 s^4 x y+17 s^4 x z+6 s^4 y^2+

+\epsilon_3}\right) + s_1(s,\,\epsilon_1,\,\epsiLon_2,\,\epsilon_3) \Sin\leFt(\fRac{2\Pi \EpsiLon_1} {s+\Epsilon_1+\epsilon_2+\EPsilOn_3}\right)\\ &+ s_1(s,\,\epsilon_2,\,\epsiloN_3,\,\epsiLoN_1) \Sin\lEFt(\Frac{2\pI \epsiloN_2} {S+\ePSIloN_1+\ePsIloN_2+\ePSiLon_3}\riGht) + S_1(s,\,\epsilOn_3,\,\epsilon_1,\,\ePsiLoN_2) \sin\left(\frac{2\PI \ePsilon_3} {s+\epsIloN_1+\epsilon_2+\epsiLon_3}\Right)\\ &+ s_2(S,\,\ePsiLOn_1,\,\epsIloN_2,\,\epsiLon_3) \sin\LEft(\fraC{2\pi (\epsiloN_1+\ePSilon_2)} {s+\EPsilon_1+\ePSIlOn_2+\epSilon_3}\right) + s_2(s,\,\epsilON_2,\,\ePSilon_3,\,\epsilon_1) \siN\left(\fRaC{2\Pi (\EPSilOn_2+\ePsilon_3)} {s+\epsIlOn_1+\epsILon_2+\epsiLOn_3}\RIGHt)\\ &+ s_2(S,\,\Epsilon_3,\,\epsiloN_1,\,\epsilon_2) \sin\LEft(\Frac{2\pi (\EpSilON_3+\epsilOn_1)} {s+\epSiLOn_1+\ePsilon_2+\epsilOn_3}\riGht) + s_3(s,\,\epsiLon_1,\,\epsILon_2,\,\epsiLOn_3) \sin\leFt(\frac{2\Pi (\ePsiLon_1+\ePSiLoN_2+\epSiLOn_3)} {s+\EPsIloN_1+\EpsIlon_2+\epsiLoN_3}\rIght)\BIgg)\,, \eND{SPLit}$$ wHerE $$\notAg \begIn{split} &c_1(s,\,x,\,y,\,z)\\ = &4 x^2 \Big( 3 S^6+6 s^5 x+9 s^5 Y+9 S^5 z+3 s^4 X^2 +17 s^4 x y+17 s^4 X z+6 s^4 y^2+

+\epsilon_3}\right) + s_1( s,\,\epsil on_1, \,\ eps il on_2 ,\,\ epsilon_3) \si n \lef t(\frac{2\pi \epsilon_ 1} {s +\ e psil o n_ 1+\ep silon_2 + \e p s ilo n_ 3} \ri gh t )\ \ &+s_1 (s,\,\e psilon_2,\ ,\e ps ilon_3,\,\ep s il on_1) \sin \le ft(\frac{2\p i \ epsilo n_ 2}{ s+\ep sil on_1+ \epsil o n_2+\e psilon_3} \r i ght) + s_1(s,\ , \ ep silo n_3,\,\epsilon_1, \ ,\ e psilon_2) \sin \left( \f r ac { 2 \pi \e psilon_3}{s +\eps i lon_1+\ e ps i l o n_2 + \epsilon_3}\r ight)\\ &+s _2( s,\,\e ps ilo n _1,\,\ epsil on _ 2,\ ,\epsilon_3 ) \s in\left(\ frac{2 \ pi (\ep s ilon_1+ \epsil on_ 2)} {s+ \ ep si lon _1 + \ep s il on_ 2 +\e psilon_3 }\ ri ght)+ s_ 2 ( s , \,\e psi lon_ 2,\,\ epsilon_3,\,\ eps ilon _ 1)\sin\ left( \fra c{ 2\pi(\epsi lon_2 +\ epsilon_3)} {s+ \eps ilon_1+\e psi lo n_2 +\ epsil o n_3}\r igh t)\ \ &+ s_ 2(s,\,\ e psi lo n _ 3 ,\ ,\epsilon_1,\,\eps il o n _2 ) \sin\l eft(\f r ac {2 \ pi (\eps il on_ 3+\e p s ilon_ 1)}{ s+ \epsilon _1+\ep s il on _2+\eps il on_3}\ ri ght ) + s_3( s ,\,\ epsilo n_1,\,\e psilo n _2,\,\epsilon_ 3 ) \sin\left(\ f ra c { 2\ p i (\ eps ilon_1+\eps ilon _ 2+\e psil o n_ 3)} {s+\e psilo n_ 1 +\ e psilon_2+\epsilon_3 }\ right) \Bigg )\,, \end{spl it}$$ wher e $ $\notag\beg i n{ s plit} &c_1(s,\ ,x,\, y,\,z)\\ = &4 x^2 \ big(3 s^6+6s^5 x+9 s ^ 5 y+9 s^5 z+ 3 s ^4x^2 + 17 s^4 x y+17 s ^ 4 x z +6 s^4 y^ 2+

+\epsilon_3}\right) + s_1(s,\,\epsilon_1,\,\epsilon_2,\,\epsilon_3) \sin\left(\frac{2\pi \epsilon_1} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\\ &+ s_1(s,\,\epsilon_2,\,\epsilon_3,\,\epsilon_1) \sin\left(\frac{2\pi_\epsilon_2} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right) + s_1(s,\,\epsilon_3,\,\epsilon_1,\,\epsilon_2) \sin\left(\frac{2\pi \epsilon_3} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\\ &+ s_2(s,\,\epsilon_1,\,\epsilon_2,\,\epsilon_3) \sin\left(\frac{2\pi_(\epsilon_1+\epsilon_2)} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right) + s_2(s,\,\epsilon_2,\,\epsilon_3,\,\epsilon_1) \sin\left(\frac{2\pi (\epsilon_2+\epsilon_3)} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\\ &+ s_2(s,\,\epsilon_3,\,\epsilon_1,\,\epsilon_2) \sin\left(\frac{2\pi (\epsilon_3+\epsilon_1)} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right) + s_3(s,\,\epsilon_1,\,\epsilon_2,\,\epsilon_3) \sin\left(\frac{2\pi (\epsilon_1+\epsilon_2+\epsilon_3)} {s+\epsilon_1+\epsilon_2+\epsilon_3}\right)\Bigg)\,, \end{split}$$_where $$\notag \begin{split} &c_1(s,\,x,\,y,\,z)\\ = &4_x^2_\big( 3 s^6+6_s^5_x+9 s^5 y+9_s^5 z+3 s^4_x^2 +17 s^4 x y+17_s^4 x z+6_s^4 _ y^2+

$ of the different terms involved in the transport equation (\[TKEB\]) at two radial locations. It is clearly seen that all these terms vanish towards the rotor side ($z^*=0$), confirming the laminar nature of this zone up to the stator boundary layer ($z^*=1$). At the stator wall, the viscous diffusion balances the dissipation as well known in 3D turbulent boundary layer. Within the Bödewadt layer, even though some interaction between the different terms involved is observed, the major contributions come from the production, the dissipation and the viscous diffusion terms. The production is balanced by the dissipation and the viscous diffusion, the level of which increases in association with the thickening of the boundary layer towards the periphery. The production increases with increasing radius as already observed with the levels of the normal Reynolds stresses [@poncras]. The maximum of the production term is obtained at the wall coordinate $z^+=z v_{\tau}/\nu=12$ ($v_{\tau}=((\nu \partial V_{\theta}/\partial z)^{2}+(\nu \partial V_{r}/\partial z)^{2})^{1/4}$ the total friction velocity and $z$ the axial coordinate) for $r/b=0.56$ and at $z^+=12.5$ for $r/b=0.8$, which confirms the approximately self-similar behavior of the Bödewadt layer. The levels of the viscous diffusion increase when moving towards the outer casing, where the highest turbulence intensities prevail. It indicates that viscous effects still play an important role in the turbulence towards these regions, which does not allow for a distinct delineation of the viscous sublayer. This indicates also the weak nature of the turbulence obtained at this Reynolds number. Conditional-averaged quadrant analysis {#conditional-averaged-quadrant-analysis.unnumbered} -------------------------------------- ![Variation with $-\Delta r^+$ of the conditionally averaged Reynolds shear stress at $z^+=17$ in the vicinity of (a) a strong ejection $<v'_{\theta}v'_z \mid strong$ $ejection>/<v'_{\theta}v'_z>$ and (b) a strong sweep $<v'_{\theta}v'_z \mid strong$ $sweep>/<v'_{\theta}v'_z>$.[]{data-label="vwsweep"}](vweject.eps "fig:"){width="35.00000%"}\ (a)

$ of the different terms involved in the transport equality (\[TKEB\ ]) at two radial localization. It is clearly seen that all these term fly towards the rotor side ($ z^*=0 $), confirming the laminar nature of this zone up to the stator limit layer ($ z^*=1 $). At the stator wall, the viscous dissemination balances the dissipation equally well known in 3D turbulent boundary layer. Within the Bödewadt layer, even though some interaction between the unlike terms involved is note, the major contributions come from the product, the dissipation and the viscous diffusion term. The production is balanced by the dissipation and the viscous dissemination, the level of which increases in association with the thickening of the boundary layer towards the periphery. The product increases with increasing radius as already observed with the levels of the normal Reynolds stresses [ @poncras ]. The maximum of the production term is obtained at the wall coordinate $ z^+=z v_{\tau}/\nu=12 $ ($ v_{\tau}=((\nu \partial V_{\theta}/\partial z)^{2}+(\nu \partial V_{r}/\partial z)^{2})^{1/4}$ the total friction velocity and $ z$ the axial coordinate) for $ r / b=0.56 $ and at $ z^+=12.5 $ for $ roentgen / b=0.8 $, which confirms the approximately self - similar demeanor of the Bödewadt level. The levels of the viscous diffusion increase when move towards the outer casing, where the highest turbulence intensities predominate. It indicates that viscous effects still play an important role in the turbulence towards these regions, which does not allow for a distinct delineation of the viscous sublayer. This indicate besides the watery nature of the turbulence obtained at this Reynolds number. Conditional - averaged quadrant analysis { # conditional-averaged-quadrant-analysis.unnumbered } -------------------------------------- ! [ Variation with $ -\Delta r^+$ of the conditionally averaged Reynolds shear stress at $ z^+=17 $ in the vicinity of (a) a impregnable ejection $ < v'_{\theta}v'_z \mid strong$ $ ejection>/<v'_{\theta}v'_z>$ and (b) a strong sweep $ < v'_{\theta}v'_z \mid strong$ $ sweep>/<v'_{\theta}v'_z>$.[]{data - label="vwsweep"}](vweject.eps " fig:"){width="35.00000%"}\ (a)

$ of the different terms invulved in the trcbsport equatjon (\[TKEB\]) at two radial locations. It ms coearlt seen that all these germs vanpsh towares tie rotor side ($z^*=0$), rknfirmiky the pamiuac nature of thix zone up do the stator tojnbary layer ($z^*=1$). At the stator wall, the discous dlffusion balanses uhe disappction as well known in 3D turbulsnt bouidary layer. Witnin the Bödewadt layer, even thokgh some interactiln between jge qufferent teros involved is observes, the major contributions come wrom che productuob, tjg dissipatioi and nhe viscous dlgfusiot terms. The productiok is ualabced by the dissipatimn and the viscous diffusiot, che level of which inxrwases in dssozuatkon wmth the tjiciening of fhe boundart layer towards the pqgophery. The pdoductyog increases with increasing radius as anrezdy observed with the lwvels of the normal Rgynolds stwesses [@poncras]. The maximum of the production term is outxintd at tfw aall coordinate $z^+=z v_{\tau}/\nu=12$ ($v_{\tau}=((\nu \partial V_{\thetw}/\lattpal z)^{2}+(\nu \partial V_{v}/\partial z)^{2})^{1/4}$ the totak vrostion velociti and $z$ ths axial coordinate) for $r/b=0.56$ and qt $z^+=12.5$ for $w/b=0.8$, wnich confirms the approximarely self-simplar behavior of the Böbewadt layer. The leveks of the viscous diffurion increase wjen movinf towards the outdr basitg, where the highest turbujence intxnsitnes prevxil. Ot indycates thah visgmus effects still olay cn im[ortant rope in the turbulence towards thxxe regions, wniwh goes not alloe for a distigct delineatiou of the viscuus sublayvr. This iidicates alsj the weak nadore of the tucbulence jbtauned at thir Reynolds numbrr. Conditional-averagwd quadrant analysls {#covsitional-averageb-doqdrant-analysis.innjmbqrvd} -------------------------------------- ![Veriatymn with $-\Deltd r^+$ uf gne covditionally xversged Reynolds shear vtreas at $z^+=17$ in the viconlty of (a) q strong ejection $<v'_{\thrta}v'_z \mid strong$ $euectimn>/<v'_{\vheta}v'_a>$ agd (b) a strong sweep $<v'_{\theta}v'_z \mjd strong$ $swcep>/<v'_{\theta}v'_z>$.[]{daea-lanel="vrsweep"}](vwejzct.eps "fig:"){width="35.00000%"}\ (a)

$ of the different terms involved in equation at two locations. It is terms towards the rotor ($z^*=0$), confirming the nature of this zone up to stator boundary layer ($z^*=1$). At the stator wall, the viscous diffusion balances the as well known in 3D turbulent boundary layer. Within the Bödewadt layer, even some between different involved is observed, the major contributions come from the production, the dissipation and the viscous diffusion The production is balanced by the dissipation and viscous diffusion, the level which increases in association with thickening the boundary towards periphery. production increases with radius as already observed with the levels of the normal Reynolds stresses [@poncras]. The maximum of the term is the wall $z^+=z ($v_{\tau}=((\nu V_{\theta}/\partial z)^{2}+(\nu \partial the total friction velocity and $z$ for $r/b=0.56$ and at $z^+=12.5$ for $r/b=0.8$, which the approximately behavior of the Bödewadt layer. The of the viscous diffusion increase when moving towards outer casing, where the highest turbulence intensities prevail. It indicates that viscous effects still play role in the turbulence these regions, which not for distinct of the sublayer. This indicates also the weak nature of the turbulence obtained this Reynolds number. Conditional-averaged quadrant analysis {#conditional-averaged-quadrant-analysis.unnumbered} -------------------------------------- ![Variation with of conditionally averaged Reynolds stress at $z^+=17$ in vicinity (a) a strong ejection strong$ and sweep \mid $sweep>/<v'_{\theta}v'_z>$.[]{data-label="vwsweep"}](vweject.eps "fig:"){width="35.00000%"}\ (a)

$ of the different terms involvEd in the traNsporT eqUatIoN (\[TKEb\]) at tWo radial locatiONs. It Is clearly seen that all thEse teRmS VaniSH tOwardS the rotOR sIDE ($z^*=0$), cOnFiRmiNg THe LaminAr nAture of This zone up To tHe Stator boundaRY lAyer ($z^*=1$). At the StaTor wall, the viScoUs diffUsIon BAlancEs tHe disSipatiON as welL known in 3D TuRBulent BOundary LAYeR. WitHin the Bödewadt layER, eVEn though some inTeractIoN BeTWEen The Different tErMs invOLved is oBSeRVED, thE Major contribuTions come frOM thE produCtIon, THe dissIpatiOn ANd tHe viscous diFfusIon terms. THe prodUCtion is BAlanced By the dIssIpaTion ANd ThE viScOUs dIFfUsiON, thE level of WhIcH incrEaseS IN ASsocIatIon wIth thE thickening of The BounDAry Layer TowarDs thE pEriphEry. The ProduCtIon increases witH incReasing raDiuS aS alReAdy obSErved wIth The Levels oF the norMAl REyNOLDs Stresses [@poncras]. The MaXIMuM of the prOductiON tErM Is obtainEd At tHe waLL CoordInatE $Z^+=z V_{\tau}/\nu=12$ ($v_{\tAu}=((\nu \paRTiAl v_{\theta}/\pArTial z)^{2}+(\nU \pArtIal v_{r}/\parTIal z)^{2})^{1/4}$ The totAl frictiOn velOCity and $z$ the axiAL coordinate) foR $R/b=0.56$ AND aT $Z^+=12.5$ for $R/b=0.8$, wHich confirmS the APproXimaTElY seLF-simiLar beHaVIoR Of the Bödewadt layer. THe Levels Of the Viscous diffusIon increasE WHEn moving TowaRDs THe outer casing, wHere tHe highest tURbulence IntenSities prEvail. It inDICates thaT viScoUs eFfeCTS sTill play an impORTant RoLe in the TurBulence TowArdS thEse ReGions, whicH does not AlLoW fOr A diStincT DelineatIoN of ThE viScous SUblayeR. This IndiCaTeS AlsO the weaK NaTURe of ThE tUrbuLenCe ObtaiNed aT ThiS ReynolDs number. COndITionAl-AvEraged qUadrant analysIs {#ConditionaL-aVerAged-quADRant-analYsis.unnumbered} -------------------------------------- ![Variation WIth $-\DeltA r^+$ oF the cOndiTionally aVerAged ReYnoLDs sheaR stresS at $z^+=17$ iN tHe vICInity OF (A) a StrOnG ejection $<v'_{\THEta}V'_z \mid StRong$ $EjectioN>/<v'_{\theta}v'_z>$ and (b) a stroNG swEep $<v'_{\theta}v'_z \miD stRong$ $SWEeP>/<v'_{\tHEtA}V'_z>$.[]{dAtA-LabEL="Vwsweep"}](vweject.ePs "fig:"){width="35.00000%"}\ (A)

$ of the different terms i nvolved in thetra nsp or t eq uati on (\[TKEB\])a t tw o radial locations. It is c le a rlys ee n tha t all t h es e ter ms v ani sh to wards th e rotor side ($z^ *=0 $) , confirming th e laminarnat ure of thiszon e up t othe stato r b ounda ry lay e r ($z^ *=1$). At t h e stat o r wall, t he vis cous diffusion ba l an c es the dissipa tion a sw el l kno wnin 3D turb ul ent b o undaryl ay e r . Wi t hin the Bödew adt layer,e ven thoug hsom e inter actio nb etw een the dif fere nt termsinvolv e d is ob s erved,the ma jor co ntri b ut io nsco m e f r om th e pr oduction ,th e dis sipa t i o n and th e vi scous diffusion te rms . Th e pr oduct ion i s ba la ncedby the diss ip ation and the v isco us diffus ion ,the l evelo f whic h i ncr eases i n assoc i ati on w i th the thickening of t h e b oundarylayert ow ar d s the pe ri phe ry.T h e pro duct i on increas es wit h i nc reasing r adiusas al rea dy ob s erve d with the lev els o f the normal Re y nolds stresse s [ @ p on c ras] . T he maximumof t h e pr oduc t io n t e rm is obta in e da t the wall coordina te $z^+= z v_{ \tau}/\nu=12$ ($v_{\tau } = ( (\nu \pa rtia l V _ {\theta}/\part ial z )^{2}+(\nu \partial V_{r }/\parti al z)^{2} ) ^ {1/4}$ t hetot alfri c t io n velocity an d $z$th e axial co ordinat e)for $r /b= 0. 56$ and a t $z^+=1 2. 5$ f or $r /b=0. 8 $, which c onf ir msthe a p proxim ately sel f- si m ila r behav i or o f th eBö dewa dtla yer.Thel eve ls of t he viscou s d i ffus io nincreas e when moving t owards the o ute r casi n g , wherethe highest turbulencei ntensit ies prev ail. It indic ate s that vi s cous e ffects stil lpla y an im p o rt ant r ole in the t urb ulenc etowa rds the se regions, whichd oes not allow fo r a dis t i nc t d e li n eat io n of t he viscous subl ayer. This i n di cates also the w eak nat ure ofthe t u rbulenc e obtaine d at this R eyno l d s n umber. Co nditiona l-average d quad r an t ana lys is {#c on dit ional -avera g ed- quadr ant-an al ysis.u nnumb er ed} ---- ----------------------- ------ ----- ! [Variatio n w i th$-\Deltar^+$ of the co ndi tio nally av e raged Rey n ol dss hearstre s s at $z^+ = 17 $ i n th e vicinityo f (a) a st ron g eject ion$<v'_{\theta}v'_z \mid strong$ $ ejec t i on> /<v ' _{\t he ta}v'_z>$ and(b) a s trong sw ee p $<v'_{\th eta}v'_z \ m id st rong$$sweep >/<v'_{ \ t he t a}v'_z >$.[ ]{d ata-label ="v ws w eep"}]( vw ej e ct.eps "fi g: "){wid th="35 . 0000 0 % "}\ (a)

$ of_the different_terms involved in the_transport equation_(\[TKEB\])_at two_radial_locations. It is_clearly seen that_all these terms vanish_towards the rotor_side_($z^*=0$), confirming the laminar nature of this zone up to the stator boundary layer_($z^*=1$)._At the_stator_wall,_the viscous diffusion balances the_dissipation as well known in_3D turbulent_boundary layer. Within the Bödewadt layer, even though_some_interaction between the_different terms involved is observed, the major contributions come_from the production, the dissipation and_the viscous diffusion_terms._The_production is balanced by_the dissipation and the viscous diffusion,_the level of which increases in_association with the thickening of the boundary_layer towards the periphery. The production_increases with increasing radius as_already observed_with the levels of the_normal Reynolds stresses_[@poncras]. The_maximum of the_production term is obtained at the_wall coordinate $z^+=z_v_{\tau}/\nu=12$ ($v_{\tau}=((\nu \partial V_{\theta}/\partial z)^{2}+(\nu \partial V_{r}/\partial z)^{2})^{1/4}$_the_total friction velocity_and_$z$_the axial_coordinate) for $r/b=0.56$_and_at $z^+=12.5$_for_$r/b=0.8$, which confirms the approximately self-similar_behavior_of the Bödewadt layer. The levels of_the viscous diffusion increase_when_moving towards the outer_casing, where the highest turbulence_intensities prevail. It indicates that viscous_effects still_play an_important role in the turbulence towards these regions, which does not_allow for a distinct delineation of_the viscous sublayer. This_indicates also_the_weak nature of_the_turbulence obtained_at this Reynolds number. Conditional-averaged quadrant analysis {#conditional-averaged-quadrant-analysis.unnumbered} -------------------------------------- ![Variation_with $-\Delta_r^+$ of the conditionally averaged Reynolds_shear stress at $z^+=17$_in_the vicinity of (a) a strong_ejection $<v'_{\theta}v'_z \mid strong$ $ejection>/<v'_{\theta}v'_z>$ and_(b) a strong sweep $<v'_{\theta}v'_z_\mid_strong$_$sweep>/<v'_{\theta}v'_z>$.[]{data-label="vwsweep"}](vweject.eps "fig:"){width="35.00000%"}\ (a)

ComplexiteAlgo\]. Without lost of generality and for the sake of simplicity, we assume that $k^{(i)} = d$ and $\psi_{d } (j) = j$ for all $j \in \{1,\dots,d-1\}$. Let $\ell_1 < \cdots < \ell_r$ be the elements of $\mathcal{L} $. Define for all $p \in \{1,\dots,d-1\}$, $k_{p,0} = 0$ and by induction for all integer ${\mathfrak{m}}$, $$\begin{aligned} k_{p,{\mathfrak{m}}+1} = \begin{cases} \inf \left\{k > k_{p,{\mathfrak{m}}}, \; \mathfrak{j}_{\text{min}}^{(i,\ell_{k})} > p \right\} & \text{if there exists $k \in \{k_{p,{\mathfrak{m}}} +1,\dots,r \}$ such that $\mathfrak{j}_{\text{min}}^{(i,\ell_{k})} > p$} \\ r & \text{otherwise.} \end{cases}\end{aligned}$$ Let $\mathfrak{M}_p$ be the smallest integer ${\mathfrak{m}}$ for which $k_{p,{\mathfrak{m}}} = r$. Set for all ${\mathfrak{m}}\in \{0,\dots,\mathfrak{M}_p-1\}$, $$K_{p,{\mathfrak{m}}} = \left\{k_{p,{\mathfrak{m}}}+1, \dots, k_{p,{\mathfrak{m}}+1} \right\}.$$ The cardinality of $K_{p,{\mathfrak{m}}}$ can be upper bounded by the claim below. \[ClaimMajorationCardinalDeKpm\] For all $p \in \{1,\dots,d-1\}$ and ${\mathfrak{m}}\in \{0,\dots,\mathfrak{M}_p-1\}$, $$\begin{aligned} \label{eqMajorationKpm} |K_{p,{\mathfrak{m}}}| \leq \prod_{k=1}^p \left[ 1 + \left(\frac{{ \begingroup \def\mathaccent#R##2{ \rel@kern{0.8} \overline{\rel@kern{-0.8}\macc@nucleus\rel@kern{0.

ComplexiteAlgo\ ]. Without lost of generality and for the sake of simplicity, we assume that $ k^{(i) } = d$ and $ \psi_{d } (joule) = j$ for all $ joule \in \{1,\dots, d-1\}$. permit $ \ell_1 < \cdots < \ell_r$ be the elements of $ \mathcal{L } $. specify for all $ p \in \{1,\dots, d-1\}$, $ k_{p,0 } = 0 $ and by induction for all integer $ { \mathfrak{m}}$, $ $ \begin{aligned } k_{p,{\mathfrak{m}}+1 } = \begin{cases } \inf \left\{k > k_{p,{\mathfrak{m } } }, \; \mathfrak{j}_{\text{min}}^{(i,\ell_{k }) } > p \right\ } & \text{if there exist $ k \in \{k_{p,{\mathfrak{m } } } +1,\dots, r \}$ such that $ \mathfrak{j}_{\text{min}}^{(i,\ell_{k }) } > p$ } \\ gas constant & \text{otherwise. } \end{cases}\end{aligned}$$ Let $ \mathfrak{M}_p$ be the smallest integer $ { \mathfrak{m}}$ for which $ k_{p,{\mathfrak{m } } } = r$. Set for all $ { \mathfrak{m}}\in \{0,\dots,\mathfrak{M}_p-1\}$, $ $ K_{p,{\mathfrak{m } } } = \left\{k_{p,{\mathfrak{m}}}+1, \dots, k_{p,{\mathfrak{m}}+1 } \right\}.$$ The cardinality of $ K_{p,{\mathfrak{m}}}$ can be upper bounded by the call below. \[ClaimMajorationCardinalDeKpm\ ] For all $ p \in \{1,\dots, d-1\}$ and $ { \mathfrak{m}}\in \{0,\dots,\mathfrak{M}_p-1\}$, $ $ \begin{aligned } \label{eqMajorationKpm } |K_{p,{\mathfrak{m}}}| \leq \prod_{k=1}^p \left [ 1 + \left(\frac { { \begingroup \def\mathaccent#R##2 { \rel@kern{0.8 } \overline{\rel@kern{-0.8}\macc@nucleus\rel@kern{0.

ComolexiteAlgo\]. Without lost uf generality aue for vhe saks of simolicity, we assume that $k^{(i)} = d$ abd $\psu_{d } (j) = j$ for all $j \in \{1,\aots,d-1\}$. Let $\ell_1 < \cdits < \wll_r$ be thx elemenbf of $\latheao{L} $. Define for all $p \in \{1,\gots,d-1\}$, $k_{p,0} = 0$ and bh nnduction for all integer ${\mathfrak{m}}$, $$\fegin{alogjed} k_{p,{\mathfrak{m}}+1} = \btgig{casss} \inf \left\{k > k_{p,{\mathfrak{m}}}, \; \mathfrzk{j}_{\text{kin}}^{(i,\ell_{k})} > p \tight\} & \text{if there exists $k \ij \{k_{p,{\mathfrak{m}}} +1,\dots,g \}$ such thqt $\mwrhfrak{j}_{\text{mkn}}^{(i,\ell_{k})} > p$} \\ r & \text{ktherwise.} \end{cases}\end{aligned}$$ Uet $\mcthfrak{M}_p$ bg fhf smallest iiteger ${\mathfrak{m}}$ for which $k_{p,{\mathgrak{m}}} = r$. Set fpr ell ${\nathfrak{m}}\in \{0,\dots,\mathfcak{M}_p-1\}$, $$K_{p,{\mathfrak{m}}} = \lgft\{k_{p,{\mathfsai{m}}}+1, \dots, k_{p,{\mathfraj{m}}+1} \righj\}.$$ The caraunauitg pf $K_{p,{\matjfrek{m}}}$ can be hpper boundwd by the claim belpw. \[SoaimMajoratiohCardigajDeKpm\] For all $p \in \{1,\dots,d-1\}$ and ${\mathfrak{m}}\pn \{0,\dkts,\mathfrak{M}_p-1\}$, $$\begin{aligbed} \label{eqMajorationNpm} |K_{p,{\mathsrak{m}}}| \leq \prod_{k=1}^p \left[ 1 + \left(\frac{{ \begingroup \def\metfaceckt#R##2{ \gel@kern{0.8} \overline{\rel@kern{-0.8}\macc@nucleus\rel@kern{0.

ComplexiteAlgo\]. Without lost of generality and for of we assume $k^{(i)} = d$ j$ all $j \in Let $\ell_1 < < \ell_r$ be the elements of $. Define for all $p \in \{1,\dots,d-1\}$, $k_{p,0} = 0$ and by induction all integer ${\mathfrak{m}}$, $$\begin{aligned} k_{p,{\mathfrak{m}}+1} = \begin{cases} \inf \left\{k > k_{p,{\mathfrak{m}}}, \; \mathfrak{j}_{\text{min}}^{(i,\ell_{k})} p & there $k \in \{k_{p,{\mathfrak{m}}} +1,\dots,r \}$ such that $\mathfrak{j}_{\text{min}}^{(i,\ell_{k})} > p$} \\ r & \text{otherwise.} \end{cases}\end{aligned}$$ Let be the smallest integer ${\mathfrak{m}}$ for which $k_{p,{\mathfrak{m}}} r$. Set for all \{0,\dots,\mathfrak{M}_p-1\}$, $$K_{p,{\mathfrak{m}}} = \left\{k_{p,{\mathfrak{m}}}+1, \dots, \right\}.$$ cardinality of can upper by the claim \[ClaimMajorationCardinalDeKpm\] For all $p \in \{1,\dots,d-1\}$ and ${\mathfrak{m}}\in \{0,\dots,\mathfrak{M}_p-1\}$, $$\begin{aligned} \label{eqMajorationKpm} |K_{p,{\mathfrak{m}}}| \leq \prod_{k=1}^p \left[ 1 + \begingroup \def\mathaccent#R##2{

ComplexiteAlgo\]. Without lost Of generaliTy and For The SaKe of SimpLicity, we assume THat $k^{(I)} = d$ and $\psi_{d } (j) = j$ for all $j \in \{1,\doTs,d-1\}$. LeT $\eLL_1 < \cdoTS < \eLl_r$ be The elemENtS OF $\maThCaL{L} $. DEfINe For alL $p \iN \{1,\dots,d-1\}$, $k_{P,0} = 0$ and by induCtiOn For all integeR ${\MaThfrak{m}}$, $$\begIn{aLigned} k_{p,{\mathFraK{m}}+1} = \begiN{cAseS} \Inf \leFt\{k > K_{p,{\matHfrak{m}}}, \; \MAthfraK{j}_{\text{min}}^{(I,\eLL_{k})} > p \rigHT\} & \text{if THErE exiSts $k \in \{k_{p,{\mathfrak{m}}} +1,\DOtS,R \}$ such that $\mathfRak{j}_{\teXt{MIn}}^{(I,\ELl_{k})} > P$} \\ r & \tExt{otherwiSe.} \End{caSEs}\end{alIGnED}$$ lEt $\mAThfrak{M}_p$ be the Smallest intEGer ${\MathfrAk{M}}$ foR Which $k_{P,{\mathFrAK{m}}} = r$. set for all ${\maThfrAk{m}}\in \{0,\dots,\MathfrAK{M}_p-1\}$, $$K_{p,{\maTHfrak{m}}} = \lEft\{k_{p,{\mAthFraK{m}}}+1, \doTS, k_{P,{\mAthFrAK{m}}+1} \rIGhT\}.$$ ThE CarDinality Of $k_{p,{\MathfRak{m}}}$ CAN BE uppEr bOundEd by tHe claim below. \[CLaiMMajORatIonCaRdinaLDeKPm\] for alL $p \in \{1,\doTs,d-1\}$ anD ${\mAthfrak{m}}\in \{0,\dots,\mAthfRak{M}_p-1\}$, $$\begiN{alIgNed} \LaBel{eqmAjoratIonkpm} |k_{p,{\mathfRak{m}}}| \leq \PRod_{K=1}^p \LEFT[ 1 + \lEft(\frac{{ \begingroup \dEf\MAThAccent#R##2{ \rEl@kern{0.8} \OVeRlINe{\rel@kerN{-0.8}\mAcc@NuclEUS\rel@kErn{0.

ComplexiteAlgo\]. Without lost of g enera lit y a nd for the sake of simpl i city , we assume that $k^{( i)} = d $ and $\ psi_{ d } (j) =j $ fo ral l $ j\ in \{1, \do ts,d-1\ }$. Let $\ ell _1 < \cdots <\ el l_r$ be th e e lements of $ \ma thcal{ L} $. Defin e f or al l $p \ i n \{1, \dots,d-1 \} $ , $k_{ p ,0} = 0 $ an d by induction for al l i n teger ${\mathf rak{m} }$ , $ $ \ beg in{ aligned} k _{ p,{\m a thfrak{ m }} + 1 } = \begin{cases } \inf \l e ft\ {k > k _{ p,{ \ mathfr ak{m} }} , \; \mathfrak{ j}_{ \text{min }}^{(i , \ell_{k } )} > p \righ t\} &\tex t {i fthe re exi s ts $k \in \{k_{p, {\ ma thfra k{m} } } + 1,\d ots ,r \ }$ s uch that $\ma thf rak{ j }_{ \text {min} }^{( i, \ell_ {k})}> p$} \ \ r & \text {oth erwise.} \e nd{ ca ses}\ e nd{ali gne d}$ $ Let $ \mathfr a k{M }_ p $ be the smallest inte ge r ${ \mathfra k{m}}$ fo rw hich $k_ {p ,{\ math f r ak{m} }} = r$ . Set fo r all$ {\ ma thfrak{ m} }\in \ {0 ,\d ots ,\mat h frak {M}_p- 1\}$, $$ K_{p, { \mathfrak{m}}} = \left\{k_{p , {\ m a th f rak{ m}} }+1, \dots, k_{ p ,{\m athf r ak {m} } +1} \ right \} . $$ The cardinality of$K _{p,{\ mathf rak{m}}}$ can be upperb o u nded bythec la i m below. \[Cl aimMa jorationCa r dinalDeK pm\]For all$p \in \{ 1 , \dots,d- 1\} $ a nd${\ m a th frak{m}}\in \ { 0 ,\do ts ,\mathf rak {M}_p-1 \}$ , $ $\b egi n{ aligned} \label{ eq Ma jo ra tio nKpm} |K_{p,{\ ma thf ra k{m }}}|\ leq \ prod_ {k=1 }^ p\le ft[ 1 + \ l eft( \f ra c{{ \b eg ingro up \de f\matha ccent#R## 2{ \r el @k ern{0.8 } \overlin e{ \rel@kern{ -0 .8} \macc@ n u cleus\re l@kern{0.

ComplexiteAlgo\]. Without lost_of generality_and for the sake_of simplicity,_we_assume that_$k^{(i)}_= d$ and_$\psi_{d } (j)_= j$ for all_$j \in \{1,\dots,d-1\}$._Let_$\ell_1 < \cdots < \ell_r$ be the elements of $\mathcal{L} $. Define for all_$p_\in \{1,\dots,d-1\}$,_$k_{p,0}_=_0$ and by induction for_all integer ${\mathfrak{m}}$, $$\begin{aligned} k_{p,{\mathfrak{m}}+1} = _ \begin{cases} _ \inf \left\{k > k_{p,{\mathfrak{m}}}, \; \mathfrak{j}_{\text{min}}^{(i,\ell_{k})} >_p_ \right\} &_\text{if there exists $k \in \{k_{p,{\mathfrak{m}}} +1,\dots,r \}$ _such that $\mathfrak{j}_{\text{min}}^{(i,\ell_{k})} > p$} \\ _ r__&_\text{otherwise.} \end{cases}\end{aligned}$$_Let $\mathfrak{M}_p$ be the smallest integer_${\mathfrak{m}}$ for which $k_{p,{\mathfrak{m}}} = r$._Set for all ${\mathfrak{m}}\in \{0,\dots,\mathfrak{M}_p-1\}$, $$K_{p,{\mathfrak{m}}} =_\left\{k_{p,{\mathfrak{m}}}+1, \dots, k_{p,{\mathfrak{m}}+1} \right\}.$$ The cardinality_of $K_{p,{\mathfrak{m}}}$ can be upper_bounded by_the claim below. \[ClaimMajorationCardinalDeKpm\] For all_$p \in \{1,\dots,d-1\}$_and ${\mathfrak{m}}\in_\{0,\dots,\mathfrak{M}_p-1\}$, $$\begin{aligned} \label{eqMajorationKpm} |K_{p,{\mathfrak{m}}}|_\leq \prod_{k=1}^p \left[ 1_+ _\left(\frac{{ \begingroup \def\mathaccent#R##2{ __ \rel@kern{0.8} ___\overline{\rel@kern{-0.8}\macc@nucleus\rel@kern{0.

Vishwanathan, “DFacTo: distributed factorization of tensors”, (2014) “http://arxiv.org/pdf/1406.4519v1.pdf”. A. Cichocki and A.-H. Phan, “Fast Local algorithms for large scale nonnegative matrix and tensor factorizations”, [*IEICE T. Fund. Electr. **E92-A***]{} (2009) 708-721. J. Cooper and A. Dutle, “Spectra of uniform hypergraphs”, [*Linear algebra Appl. **436***]{} (2012) 3268-3292. C. Cui, Y. Dai and J. Nie, “All real eigenvalues of symmetric tensors”, [*SIAM J. Matrix Anal. Appl. **35***]{} (2014) 1582-1601. Y. Dai, “A positive BB-like stepsize and an extension for symmetric linear systems”, in [*Workshop on Optimization for Modern Computation*]{}, Beijing, China, (2014), “http://bicmr.pku.edu.cn/conference/opt-2014/slides/Yuhong-Dai.pdf”. A.L.F. de Almeida and A.Y. Kibangou, “Distributed large-scale tensor decomposition”, in [*IEEE International Conference on Acoustics, Speech and Siganl Processing (ICASSP)*]{} (2014) 26-30. L. De Lathauwer, B. De Moor, and J. Vandewalle, “On the best rank-$1$ and rank-$(R_1,R_2,\ldots,R_N)$ approximation of higher-order tensors”, [*SIAM J. Matrix Anal. Appl. **21***]{} (2000) 1324-1342. W. Ding, L. Qi and Y. Wei, “Fast Hankel tensor-vector product and its application to exponential data fitting”, [*Numer. Linear Algebr. Appl.*]{}, (2015), DOI: 10.1002/nla.1970. W. Ding and Y. Wei, “Generalized tensor eigenvalue problems”, [*SIAM J. Matrix Anal. Appl.*]{} (2015), To appear. S. Friedland, J. Nocedal and M.L. Overton, “The formulation and analysis of numerical methods for inverse

Vishwanathan, “ DFacTo: distributed factorization of tensors ”, (2014) “ http://arxiv.org/pdf/1406.4519v1.pdf ”. A. Cichocki and A.-H. Phan, “ Fast Local algorithms for big plate nonnegative matrix and tensor factorizations ”, [ * IEICE T. Fund. Electr. * * E92 - A * * * ] { } (2009) 708 - 721. J. Cooper and A. Dutle, “ Spectra of uniform hypergraphs ”, [ * Linear algebra Appl. * * 436 * * * ] { } (2012) 3268 - 3292. C. Cui, Y. Dai and J. Nie, “ All veridical eigenvalue of symmetric tensors ”, [ * SIAM J. Matrix Anal. Appl. * * 35 * * * ] { } (2014) 1582 - 1601. Y. Dai, “ A positive BB - alike stepsize and an extension for symmetric linear organization ”, in [ * Workshop on Optimization for Modern Computation * ] { }, Beijing, China, (2014), “ http://bicmr.pku.edu.cn/conference/opt-2014/slides/Yuhong-Dai.pdf ”. A.L.F. de Almeida and A.Y. Kibangou, “ Distributed large - plate tensor decomposition ”, in [ * IEEE International Conference on Acoustics, Speech and Siganl Processing (ICASSP) * ] { } (2014) 26 - 30. L. De Lathauwer, B. De Moor, and J. Vandewalle, “ On the best rank-$1 $ and rank-$(R_1,R_2,\ldots, R_N)$ approximation of higher - club tensors ”, [ * SIAM J. Matrix Anal. Appl. * * 21 * * * ] { } (2000) 1324 - 1342. W. Ding, L. Qi and Y. Wei, “ Fast Hankel tensor - vector merchandise and its application to exponential data fitting ”, [ * Numer. Linear Algebr. Appl. * ] { }, (2015), DOI: 10.1002 / nla.1970. W. Ding and Y. Wei, “ Generalized tensor eigenvalue problem ”, [ * SIAM J. Matrix Anal. Appl. * ] { } (2015), To appear. S. Friedland, J. Nocedal and M.L. Overton, “ The formulation and analysis of numeral methods for inverse

Vidhwanathan, “DFacTo: distrinuted factorizatnin of vensors”, (2014) “http://arbiv.org/pdf/1406.4519v1.pdf”. A. Cichocki and E.-H. Pyan, “Fqst Local algorithms fur large dcale nobnegetive matrix and tensor ncctorjdatious”, [*IEICE T. Fund. Glectr. **E92-A***]{} (2009) 708-721. J. Cooper and A. Guglz, “Spectra of uniform hypergraphs”, [*Linqar algrbga Appl. **436***]{} (2012) 3268-3292. C. Cuy, Y. Cwi ahd J. Nie, “All real eigenvalues of sgmmetrib tensors”, [*SIAM J. Katrix Anal. Appl. **35***]{} (2014) 1582-1601. Y. Dai, “A posltive BB-like stepslze and an gstegwion for symoetric lintax systems”, ih [*Workshop on Optimization for Ooderu Computatiin*]{}, Belbing, China, (2014), “http://ficmr.pku.edu.ck/bonferetce/opt-2014/skides/Yuhong-Dai.idf”. A.L.H. de Almeida and A.Y. Kibanjou, “Distributed largg-scale tenvox decomposition”, in [*IEWE Intetnatimnal Xonwertncx oh Acoudtirs, Speech ahd Siganl Peocessing (ICASSP)*]{} (2014) 26-30. L. Dq Lathauwer, B. Se Moow, wnd J. Vandewalle, “On the best rank-$1$ and rdnk-$(D_1,R_2,\ldots,R_N)$ approximation of higher-order tensots”, [*SIAM J. Iatrix Anal. Appl. **21***]{} (2000) 1324-1342. W. Ding, L. Qi and Y. Wei, “Fast Hannel txnror-rcgtor prlduct and its application to exponential data riutikg”, [*Numer. Linear Clgebr. Appl.*]{}, (2015), DOI: 10.1002/npa.1970. E. Ding and Y. Wgi, “Genexzljzed tensor eigenvwlue prjblemw”, [*SIAM J. Iatrox Anal. Appl.*]{} (2015), To appear. S. Fruedland, J. Nobedao and M.L. Overton, “Tke formulatiun amd anslysis of numerical metkods fkr inverse

Vishwanathan, “DFacTo: distributed factorization of tensors”, (2014) Cichocki A.-H. Phan, Local algorithms for tensor [*IEICE T. Fund. **E92-A***]{} (2009) 708-721. Cooper and A. Dutle, “Spectra of hypergraphs”, [*Linear algebra Appl. **436***]{} (2012) 3268-3292. C. Cui, Y. Dai and J. “All real eigenvalues of symmetric tensors”, [*SIAM J. Matrix Anal. Appl. **35***]{} (2014) Y. “A BB-like and an extension for symmetric linear systems”, in [*Workshop on Optimization for Modern Computation*]{}, Beijing, China, “http://bicmr.pku.edu.cn/conference/opt-2014/slides/Yuhong-Dai.pdf”. A.L.F. de Almeida and A.Y. Kibangou, “Distributed tensor decomposition”, in [*IEEE Conference on Acoustics, Speech and Processing (2014) 26-30. De B. Moor, and J. “On the best rank-$1$ and rank-$(R_1,R_2,\ldots,R_N)$ approximation of higher-order tensors”, [*SIAM J. Matrix Anal. Appl. **21***]{} (2000) W. Ding, and Y. “Fast tensor-vector and its application data fitting”, [*Numer. Linear Algebr. Appl.*]{}, W. Ding and Y. Wei, “Generalized tensor eigenvalue [*SIAM J. Anal. Appl.*]{} (2015), To appear. S. J. Nocedal and M.L. Overton, “The formulation and of numerical methods for inverse

Vishwanathan, “DFacTo: distribUted factorIzatiOn oF teNsOrs”, (2014) “hTtp://aRxiv.org/pdf/1406.4519v1.pdf”. a. cichOcki and A.-H. Phan, “Fast Local AlgorItHMs foR LaRge scAle nonnEGaTIVe mAtRiX anD tENsOr facTorIzationS”, [*IEICE T. FunD. ElEcTr. **E92-A***]{} (2009) 708-721. J. Cooper aND A. dutle, “SpectRa oF uniform hypeRgrAphs”, [*LiNeAr aLGebra appL. **436***]{} (2012) 3268-3292. C. Cui, y. Dai anD j. Nie, “AlL real eigeNvALues of SYmmetriC TEnSors”, [*sIAM J. Matrix Anal. ApPL. **35***]{} (2014) 1582-1601. Y. dAi, “A positive BB-lIke stePsIZe AND an ExtEnsion for sYmMetriC Linear sYStEMS”, In [*WORkshop on OptimIzation for MODerN CompuTaTioN*]{}, beijinG, ChinA, (2014), “hTTp://bIcmr.pku.edu.cN/conFerence/opT-2014/slideS/yuhong-DAI.pdf”. A.L.F. De AlmeIda And a.Y. KiBAnGoU, “DiStRIbuTEd LarGE-scAle tensoR dEcOmposItioN”, IN [*ieEE INteRnatIonal conference on ACouSticS, speEch anD SigaNl PrOcEssinG (ICASSp)*]{} (2014) 26-30. L. De LAtHauwer, B. De Moor, anD J. VaNdewalle, “ON thE bEst RaNk-$1$ and RAnk-$(R_1,R_2,\lDotS,R_N)$ ApproxiMation oF HigHeR-ORDeR tensors”, [*SIAM J. MatriX ANAL. APpl. **21***]{} (2000) 1324-1342. W. Ding, l. Qi and y. weI, “FASt Hankel TeNsoR-vecTOR prodUct aND iTs applicAtion tO ExPoNential DaTa fittInG”, [*NuMer. lineaR algeBr. Appl.*]{}, (2015), dOI: 10.1002/nla.1970. W. DIng anD y. Wei, “GeneralizeD Tensor eigenvaLUe PRObLEms”, [*SiAM j. Matrix Anal. appl.*]{} (2015), tO appEar. S. fRiEdlANd, J. NoCedal AnD m.L. oVerton, “The formulatioN aNd analYsis oF numerical metHods for invERSE

Vishwanathan, “DFacTo: di stributedfacto riz ati on oftens ors”, (2014) “ h ttp: //arxiv.org/pdf/1406.4 519v1 .p d f”.A. Cich ocki an d A . - H.Ph an , “ Fa s tLocal al gorithm s for larg e s ca le nonnegati v ematrix and te nsor factori zat ions”, [ *IE I CE T. Fu nd. E lectr. **E92- A***]{} ( 20 0 9) 708 - 721. J . Co oper and A. Dutle, “S p ec t ra of uniformhyperg ra p hs ” , [* Lin ear algebr aAppl. **436** * ]{ } ( 201 2 ) 3268-3292. C. Cui, Y. Dai and J .Nie , “Allrealei g env alues of sy mmet ric tenso rs”, [ * SIAM J. MatrixAnal.App l.**35 * ** ]{ } ( 20 1 4)1 58 2-1 6 01. Y. Dai ,“A posi tive B B - like st epsi ze an d an extensio n f or s y mme triclinea r sy st ems”, in [* Works ho p on Optimizati on f or Modern Co mp uta ti on*]{ } , Beij ing , C hina, ( 2014),“ htt p: / / b ic mr.pku.edu.cn/conf er e n ce /opt-201 4/slid e s/ Yu h ong-Dai. pd f”. A. L . F. de Alm e id a and A. Y. Kib a ng ou , “Dist ri butedla rge -sc ale t e nsor decom position ”, in [*IEEE Interna t ional Confere n ce o nA cous tic s, Speech a nd S i ganl Pro c es sin g (ICA SSP)* ]{ } ( 2 014) 26-30. L. DeLa thauwe r, B. De Moor, and J. Vandew a l l e, “On t he b e st rank-$1$ and r ank-$ (R_1,R_2,\ l dots,R_N )$ ap proximat ion of hi g h er-order te nso rs” , [ * S IA M J. Matrix A n a l. A pp l. **21 *** ]{} (20 00) 13 24- 134 2. W. Ding , L. Qian dY. W ei, “Fas t Hankelte nso r- vec tor p r oductand i ts a pp li c ati on to e x po n e ntia lda ta f itt in g”, [ *Num e r.LinearAlgebr. A ppl . *]{} ,(2 015), D OI: 10.1002/n la .1970. W. D ing and Y . Wei, “Ge neralized tensor eigenv a lue pro ble ms”,[*SI AM J. Mat rix Anal. Ap p l.*]{} (2015 ), To a ppe a r . S. F ri edl an d, J. Noce d a l a nd M. L. Ove rton, “ The formulation an d an alysis of num eri calm e th ods fo r in ve r se

Vishwanathan,_“DFacTo: distributed_factorization of tensors”, (2014)_“http://arxiv.org/pdf/1406.4519v1.pdf”. A. Cichocki_and_A.-H. Phan,_“Fast_Local algorithms for_large scale nonnegative_matrix and tensor factorizations”,_[*IEICE T. Fund._Electr._**E92-A***]{} (2009) 708-721. J. Cooper and A. Dutle, “Spectra of uniform hypergraphs”, [*Linear algebra Appl._**436***]{}_(2012) 3268-3292. C._Cui,_Y._Dai and J. Nie, “All_real eigenvalues of symmetric tensors”,_[*SIAM J._Matrix Anal. Appl. **35***]{} (2014) 1582-1601. Y. Dai, “A_positive_BB-like stepsize and_an extension for symmetric linear systems”, in [*Workshop on_Optimization for Modern Computation*]{}, Beijing, China,_(2014), “http://bicmr.pku.edu.cn/conference/opt-2014/slides/Yuhong-Dai.pdf”. A.L.F. de_Almeida_and_A.Y. Kibangou, “Distributed large-scale_tensor decomposition”, in [*IEEE International Conference_on Acoustics, Speech and Siganl Processing_(ICASSP)*]{} (2014) 26-30. L. De Lathauwer, B. De_Moor, and J. Vandewalle, “On the_best rank-$1$ and rank-$(R_1,R_2,\ldots,R_N)$ approximation_of higher-order_tensors”, [*SIAM J. Matrix Anal._Appl. **21***]{} (2000)_1324-1342. W. Ding,_L. Qi and_Y. Wei, “Fast Hankel tensor-vector product_and its application_to exponential data fitting”, [*Numer. Linear_Algebr._Appl.*]{}, (2015), DOI:_10.1002/nla.1970. W._Ding_and Y._Wei, “Generalized tensor_eigenvalue_problems”, [*SIAM_J._Matrix Anal. Appl.*]{} (2015), To appear. S._Friedland,_J. Nocedal and M.L. Overton, “The formulation_and analysis of numerical_methods_for inverse

occurs when $i$ takes the value $e$ determined by the equation $$g-1+\epsilon = d-2e$$ ($\epsilon \in \{0,1 \}$ as usual). When $\epsilon=0$ (i.e. when $d$ and $g$ have opposite parity), this lowest order term is $$t^{4g-4-2d} 2^{3g-2-d} q^{2-2g-e}$$ so the DT=PT invariant in minimal Euler characteristic is given by $$\begin{aligned} \label{PT} P_{2-2g-e,2}(d) & = & t^{4g-4-2d} 2^{3g-2-d}. \end{aligned}$$ The equality of DT and PT invariants in minimal Euler characteristic as discussed in \[section:DTequalsPT\], together with the known equivalence of DT and GW mentioned above, ensures that this is actually the correct PT invariant even though the coefficients of higher powers of $q$ are technically only conjectural. Assume we are in the $\epsilon=0$ case so $g-1 = d-2e$. Let $Y$ be the space (${\operatorname{Quot}}$ scheme) of such minimal (i.e. degree $e$) line subbundles of $E$. As mentioned above, if $E$ is sufficiently generic, $Y$ is just a finite number of points. On the other hand, we can identify $Y$ with the $T$-fixed subscheme of $$P := I_{2g-2-e}(E,2) = P_{2g-2-e}(E,2)$$ using Theorem \[thm:Tfixedstablepairs\] below (or by using Proposition \[prop:CMcurvesinE\] and the general remarks about this common moduli space in minimal Euler characteristic). Since $Y$ is smooth and zero dimensional, the tangent space to $Y$ at a point $L \hookrightarrow E$ is zero. On the other hand, this tangent space is given by ${\operatorname{Hom}}(L, Q) = {\operatorname{H}}^0(C,L^\lor Q),$ where $Q = E/L$ as usual. The degree of $L^\lor Q$ is $-e+d-e = g-1$, so by Riemann-Roch, we also have ${\operatorname{H}}^1(

occurs when $ i$ takes the value $ e$ determined by the equation $ $ g-1+\epsilon = d-2e$$ ($ \epsilon \in \{0,1 \}$ as common). When $ \epsilon=0 $ (i.e.   when $ d$ and $ g$ take opposite parity), this low orderliness term is $ $ t^{4g-4 - 2d } 2^{3g-2 - d } q^{2 - 2g - e}$$ so the DT = PT changeless in minimal Euler characteristic is given by $ $ \begin{aligned } \label{PT } P_{2 - 2g - e,2}(d) & = & t^{4g-4 - 2d } 2^{3g-2 - d }. \end{aligned}$$ The equality of DT and PT invariant in minimal Euler feature as discussed in \[section: DTequalsPT\ ], together with the known comparison of DT and GW mentioned above, ensures that this is actually the correct PT invariant even though the coefficient of higher powers of $ q$ are technically merely conjectural. Assume we are in the $ \epsilon=0 $ case so $ g-1 = d-2e$. permit $ Y$ be the space ($ { \operatorname{Quot}}$ scheme) of such minimal (i.e.   degree $ e$) wrinkle subbundles of $ E$. As mentioned above, if $ E$ is sufficiently generic, $ Y$ is just a finite number of points. On the early hand, we can name $ Y$ with the $ T$-fixed subscheme of $ $ P: = I_{2g-2 - e}(E,2) = P_{2g-2 - e}(E,2)$$ using Theorem   \[thm: Tfixedstablepairs\ ] below (or by using Proposition   \[prop: CMcurvesinE\ ] and the cosmopolitan remarks about this common moduli space in minimal Euler characteristic). Since $ Y$ is smooth and zero dimensional, the tangent space to $ Y$ at a point $ L \hookrightarrow E$ is zero. On the other hand, this tangent space is given by $ { \operatorname{Hom}}(L, Q) = { \operatorname{H}}^0(C, L^\lor Q),$ where $ Q = E / L$ as usual. The degree of $ L^\lor Q$ is $ -e+d - e = g-1 $, thus by Riemann - Roch, we also have $ { \operatorname{H}}^1 (

ocfurs when $i$ takes the vauue $e$ determineb by thx equatjon $$g-1+\epsklon = d-2e$$ ($\epsilon \in \{0,1 \}$ as usuel). Wyen $\ekfilon=0$ (i.e. when $d$ and $g$ have oiposite pqritb), this lowest orvsr term is $$t^{4g-4-2s} 2^{3g-2-d} x^{2-2j-e}$$ so the DT=PT lnvariant it minimal Eules zhcracteristic is given by $$\begin{aligneq} \label{LT} P_{2-2g-e,2}(d) & = & t^{4g-4-2d} 2^{3g-2-d}. \tnd{wlighvd}$$ Uhe equality of DT and PT invariahts in kinimal Euler characteristic as discussfd ij \[section:DTequalsPH\], together qith rhe known eqjivalence of DT and GW mentioned above, ensures that tfis ix actually tye fmrrect PT iivariagt even thounn the woefficoents of highev powxrs if $q$ are technically mnly conjectural. Asfume we ase in the $\epsilon=0$ cqsw so $c-1 = d-2a$. Leg $Y$ ce uhe slace (${\ooeretorname{Quof}}$ scheme) of such minimal (i.e. degteq $e$) line subbuhdles jf $E$. As mentioned above, if $E$ is sufficienuly gsneric, $Y$ is just a finire number of points. Oj the othqr hand, we can identify $Y$ with the $T$-fixed subscheke of $$O := N_{2n-2-c}(E,2) = O_{2t-2-e}(F,2)$$ using Theorem \[thm:Tfixedstablepairs\] below (or fg ispng Proposition \[prjp:CMcurvesimE\] amq the general remarks zbout this common loduli fpace in minimwl Eiler characteristic). Since $Y$ is smooth aud zero dimensional, tke tangent s'ace tp $Y$ ay a point $L \hookrightarxow E$ js zero. On hhe other fand, this tangeng siace is given by ${\operatorname{Hjm}}(L, Q) = {\opxratoxname{H}}^0(C,L^\uor A),$ wherq $Q = E/L$ as usual. The degree of $L^\log Q$ id $-a+d-e = g-1$, so hy Riemann-Roch, we also have ${\opecetorname{H}}^1(

occurs when $i$ takes the value $e$ the $$g-1+\epsilon = ($\epsilon \in \{0,1 (i.e. $d$ and $g$ opposite parity), this order term is $$t^{4g-4-2d} 2^{3g-2-d} q^{2-2g-e}$$ the DT=PT invariant in minimal Euler characteristic is given by $$\begin{aligned} \label{PT} P_{2-2g-e,2}(d) = & t^{4g-4-2d} 2^{3g-2-d}. \end{aligned}$$ The equality of DT and PT invariants in Euler as in together with the known equivalence of DT and GW mentioned above, ensures that this is actually correct PT invariant even though the coefficients of powers of $q$ are only conjectural. Assume we are the case so = Let be the space scheme) of such minimal (i.e. degree $e$) line subbundles of $E$. As mentioned above, if $E$ is generic, $Y$ a finite of On other hand, we $Y$ with the $T$-fixed subscheme of = P_{2g-2-e}(E,2)$$ using Theorem \[thm:Tfixedstablepairs\] below (or by Proposition \[prop:CMcurvesinE\] the general remarks about this common space in minimal Euler characteristic). Since $Y$ is and zero dimensional, the tangent space to $Y$ at a point $L \hookrightarrow E$ is the other hand, this space is given ${\operatorname{Hom}}(L, = Q),$ $Q = as usual. The degree of $L^\lor Q$ is $-e+d-e = g-1$, by Riemann-Roch, we also have ${\operatorname{H}}^1(

occurs when $i$ takes the value $e$ Determined By the EquAtiOn $$G-1+\epsIlon = D-2e$$ ($\epsilon \in \{0,1 \}$ as uSUal). WHen $\epsilon=0$ (i.e. when $d$ and $g$ hAve opPoSIte pARiTy), thiS lowest ORdER TerM iS $$t^{4G-4-2d} 2^{3g-2-D} q^{2-2G-E}$$ sO the Dt=PT InvariaNt in minimaL EuLeR characterisTIc Is given by $$\bEgiN{aligned} \labeL{PT} p_{2-2g-e,2}(d) & = & t^{4g-4-2D} 2^{3g-2-D}. \enD{AlignEd}$$ THe equAlity oF dT and Pt invarianTs IN minimAL Euler cHARaCterIstic as discussed iN \[SeCTion:DTequalsPT\], TogethEr WItH THe kNowN equivalenCe Of DT aND GW mentIOnED ABovE, Ensures that thIs is actuallY The CorrecT Pt inVAriant Even tHoUGh tHe coefficieNts oF higher poWers of $Q$ Are techNIcally oNly conJecTurAl. AsSUmE wE arE iN The $\EPsIloN=0$ CasE so $g-1 = d-2e$. LeT $Y$ Be The spAce (${\oPERATornAme{quot}}$ SchemE) of such minimaL (i.e. DegrEE $e$) lIne suBbundLes oF $E$. as menTioned Above, If $e$ is sufficiently GeneRic, $Y$ is jusT a fInIte NuMber oF Points. on tHe oTher hanD, we can iDEntIfY $y$ WItH the $T$-fixed subschemE oF $$p := i_{2g-2-E}(E,2) = P_{2g-2-e}(E,2)$$ usIng TheOReM \[tHM:TfixedsTaBlePairS\] BElow (oR by uSInG ProposiTion \[prOP:CmcUrvesine\] aNd the gEnEraL reMarks ABout This coMmon moduLi spaCE in minimal EuleR CharacteristiC). siNCE $Y$ IS smoOth And zero dimeNsioNAl, thE tanGEnT spACe to $Y$ At a poInT $l \hOOkrightarrow E$ is zero. on The othEr hanD, this tangent sPace is giveN BY ${\OperatorName{hOm}}(l, q) = {\operatorname{H}}^0(c,L^\lor q),$ where $Q = E/L$ aS Usual. The DegreE of $L^\lor Q$ Is $-e+d-e = g-1$, so bY rIemann-RoCh, wE alSo hAve ${\OPErAtorname{H}}^1(

occurs when $i$ takes the value $e$ dete rmi ned b y th e eq uation $$g-1+\ e psil on = d-2e$$ ($\epsilon \in\{ 0 ,1 \ } $as us ual). W h en $ \ep si lo n=0 $( i. e. wh en$d$ and $g$ haveopp os ite parity), th is lowestord er term is $ $t^ {4g-4- 2d } 2 ^ {3g-2 -d} q^{2 -2g-e} $ $ so t he DT=PTin v ariant in mini m a lEule r characteristici sg iven by $$\beg in{ali gn e d} \la bel {PT} P_{2- 2g -e,2} ( d) & =& t ^ {4g - 4-2d} 2^{3g-2 -d}. \end{a l ign ed}$$ T hee qualit y ofDT and PT invaria ntsin minima l Eule r charac t eristic as di scu sse d in \[ se cti on : DTe q ua lsP T \], togethe rwi th th e kn o w n equi val ence of D T and GW ment ion ed a b ove , ens uresthat t his i s actu allyth e correct PT in vari ant eventho ug h t he coef f icient s o f h igher p owers o f $q $a r e t echnically only co nj e c tu ral. As sume w e a re in the $ \e psi lon= 0 $ case so$ g- 1 = d-2e $. Let $Y $be thesp ace ($ {\ ope rat ornam e {Quo t}}$ s cheme) o f suc h minimal (i.e. degree $e$) l i ne s ub b undl esof $E$. Asment i oned abo v e, if $E$ i s suf fi c ie n tly generic, $Y$ is j ust afinit e number of p oints. Ont h e other h and, we can identify $ Y$ wi th the $T$ - fixed su bsche me of $$ P := I_{2 g - 2-e}(E,2 ) = P_ {2g -2- e } (E ,2)$$ using T h e orem  \ [thm:Tf ixe dstable pai rs\ ] b elo w(or by us ing Prop os it io n\[p rop:C M curvesin E\ ] a nd th e gen e ral re marks abo ut t h iscommonm od u l i sp ac ein m ini ma l Eul er c h ara cterist ic). Sinc e $ Y $ is s mo oth and zero dimensi on al, the ta ng ent space t o $Y$ at a point $L \hookrighta r row E$iszero. Onthe other ha nd, th ist angent space is g iv enb y ${\o p e ra tor na me{Hom}}(L , Q)= {\o pe rato rname{H }}^0(C,L^\lor Q),$ whe re $Q = E/L$asusua l . T hed eg r eeof $L^ \ l or Q$ is $-e+d- e = g-1$,so by Riemann-R o ch, w e alsohave ${ \oper a torname {H}}^1(

occurs_when $i$_takes the value $e$_determined by_the_equation $$g-1+\epsilon_=_d-2e$$ ($\epsilon \in_\{0,1 \}$ as_usual). When $\epsilon=0$ (i.e. when_$d$ and $g$_have_opposite parity), this lowest order term is $$t^{4g-4-2d} 2^{3g-2-d} q^{2-2g-e}$$ so the DT=PT invariant_in_minimal Euler_characteristic_is_given by $$\begin{aligned} \label{PT} P_{2-2g-e,2}(d)_& = & _t^{4g-4-2d} 2^{3g-2-d}._\end{aligned}$$ The equality of DT and PT invariants in_minimal_Euler characteristic as_discussed in \[section:DTequalsPT\], together with the known equivalence of_DT and GW mentioned above, ensures_that this is_actually_the_correct PT invariant even_though the coefficients of higher powers_of $q$ are technically only conjectural. Assume_we are in the $\epsilon=0$ case so_$g-1 = d-2e$. Let $Y$ be_the space (${\operatorname{Quot}}$ scheme) of_such minimal_(i.e. degree $e$) line subbundles of_$E$. As mentioned_above, if_$E$ is sufficiently_generic, $Y$ is just a finite_number of points._On the other hand, we can_identify_$Y$ with the_$T$-fixed_subscheme_of $$P_:= I_{2g-2-e}(E,2) =_P_{2g-2-e}(E,2)$$_using Theorem \[thm:Tfixedstablepairs\]_below_(or by using Proposition \[prop:CMcurvesinE\] and the_general_remarks about this common moduli space in_minimal Euler characteristic). Since_$Y$_is smooth and zero_dimensional, the tangent space to_$Y$ at a point $L \hookrightarrow_E$ is_zero. On_the other hand, this tangent space is given by ${\operatorname{Hom}}(L, Q)_= {\operatorname{H}}^0(C,L^\lor Q),$ where $Q =_E/L$ as usual. The_degree of_$L^\lor_Q$ is $-e+d-e_=_g-1$, so_by Riemann-Roch, we also have ${\operatorname{H}}^1(

entric radius, exceeding their turnaround radius by $50\%$ or more. These systems have clearly been affected by some mechanism that propelled them onto orbits substantially more energetic than the ones they had followed until turnaround. This mechanism seems to operate preferentially on low-mass satellites, as shown by the dashed histogram in Figure \[fig:rhist\], which corresponds to satellites with stellar masses less than $\sim 3\%$ that of the primary. ![Radial velocity of satellites versus distance to the primary. Velocities are scaled to the virial velocity of the system, distances to the virial radius. Circles denote “associated” satellites; i.e., those that have been [*inside*]{} the virial radius of the primary at some earlier time. Crosses indicate satellites that are on their first approach, and have never been inside $r_{\rm vir}$. Filled circles indicate associated satellites whose apocentric radii exceed their turnaround radius by at least $25\%$, indicating that their orbital energies have been substantially altered during their evolution. “Crossed” circles correspond to associated satellites that have entered $r_{\rm vir}$ during the last $3$ Gyrs. The curves delineating the top and bottom boundaries of the distribution show the escape velocity of an NFW halo with concentration $c=10$ and $c=20$, respectively. Note that there is one satellite “escaping” the system with positive radial velocity. Solid lines show the trajectories in the $r-V_r$ plane of the two ”ejected” satellites shown in figure \[fig:orbesc\]. Filled squares correspond to the fourteen brightest Milky Way satellites, taken from @vandenbergh99 (complemented with NED data for the Phoenix, Tucana and NGC6822), and plotted assuming that $V_{\rm vir}^{\rm MW} \sim 109$ km/s and $r_{\rm vir}^{\rm MW}=237$ kpc (see Sales et al 2007). Arrows indicate how the positions of MW satellites in this plot would be altered if our estimate of $V_{\rm vir}^{\rm MW}$ (and, consequently, $r_{\rm vir}^{\rm MW}$) is allowed to vary by $\pm 20\%$.[]{data-label="fig:rvMW"}](figs/letfig3.ps){width="84mm"} We highlight some of these objects in Figure \[fig:rvMW\], using “filled” circles to denote “associated” satellites whose apocenters

entric radius, exceeding their turnaround radius by $ 50\%$ or more. These system have intelligibly been affected by some mechanism that propelled them onto orbits well more energetic than the ones they had followed until turnaround. This mechanism look to operate preferentially on low - batch satellites, as shown by the dash histogram in Figure   \[fig: rhist\ ], which corresponds to satellites with leading masses less than $ \sim 3\%$ that of the primary. ! [ Radial velocity of satellites versus distance to the primary. Velocities are scaled to the virial speed of the system, distances to the virial radius. set denote “ associated ” satellite; i.e., those that have been [ * inside * ] { } the virial radius of the primary coil at some earlier prison term. Crosses indicate satellites that are on their inaugural approach, and have never been inside $ r_{\rm vir}$. Filled circles indicate associated satellites whose apocentric radii exceed their turnaround radius by at least $ 25\%$, indicating that their orbital energies have been substantially altered during their evolution. “ Crossed ” circles match to associated satellites that have insert $ r_{\rm vir}$ during the final $ 3 $ Gyrs. The curves delineating the top and bottom boundaries of the distribution usher the escape velocity of an NFW halo with concentration $ c=10 $ and $ c=20 $, respectively. notice that there is one satellite “ escaping ” the system with positive radial velocity. Solid lines show the trajectories in the $ r - V_r$ plane of the two ” ejected ” satellite shown in trope \[fig: orbesc\ ]. Filled squares correspond to the fourteen brightest Milky Way satellite, taken from @vandenbergh99 (complemented with NED data for the Phoenix, Tucana and NGC6822), and plat assuming that $ V_{\rm vir}^{\rm MW } \sim 109 $ km / s and $ r_{\rm vir}^{\rm MW}=237 $ kpc (see Sales et al 2007). Arrows argue how the positions of MW satellite in this plot would be altered if our estimate of $ V_{\rm vir}^{\rm MW}$ (and, consequently, $ r_{\rm vir}^{\rm MW}$) is allow to vary by $ \pm 20\%$.[]{data - label="fig: rvMW"}](figs / letfig3.ps){width="84 millimeter " } We highlight some of these objects in Figure   \[fig: rvMW\ ], using “ filled ” circle to denote “ associated ” satellites whose apocenters

entgic radius, exceeding thelr turnaround rabuus by $50\%$ or mkre. Thesd systems have clearly been effexted vy some mechanism that propellef them obto iebits subsvzntialln mors enexgxtic than the okes they hag followed unthl tbrnaround. This mechanism seems to opqrate pteverentially on low-kwss aatellites, as shown by the dashed gistogrem in Figure \[fig:thist\], which corresponds to satfllites with stellwr masses lgas eyan $\sim 3\%$ thag of the pgnmary. ![Radial velocity of satellites versus aistauce to the kxumagi. Velocities are fcaled to thc virian velocoty of the sysbem, dmstabces to the virial ravius. Circles denote “wssociateg” aatellites; i.e., thowe that hava bedb [*ivsise*]{} tge virlal radius of the primart at some earlier tomq. Crosses indidate swtqllites that are on their first approacv, ahd have never been insiee $r_{\rm vir}$. Filled circpes indicwte associated satellites whose apocentric radii axceev ghenv gyrjaround radius by at least $25\%$, indicating that trsit prbital energics have been substsnhisjly altered doring tksid evolution. “Crossef” circlgs coreespond tj asxociated satellites that hace entered $r_{\gm vur}$ during the last $3$ Gyrs. The eurves delimeating the top and botcom bohndaries of the distdkbution show the esbape velocity of an NFW halo wyth conceitratnon $c=10$ ana $c=20$, tespectyvely. Note that there is one satelpite “gscapitg” the syshem with positive radial velocivb. Solid lines svow the tralectorles in the $r-V_r$ [lane of the tco ”ejectzd” satdllites shkwn in higure \[fig:orfesc\]. Filled sxoares correspmnd to tre fiurtwen brieftest Milky Wau satellinef, tajen from @vandenbernh99 (coollemented with UTD eata for the Phpenkx, Eubane and TGC6822), and plotded xssjking ghat $V_{\rm viv}^{\rm OW} \som 109$ km/s and $r_{\rm vir}^{\rk MW}=237$ kpc (see Sales et sl 2007). Arrows indicatq how the posotions of MW satelpites in this pljt would be altered if our estjmate of $G_{\rm vir}^{\rm MW}$ (anq, cokseqoently, $r_{\rm rir}^{\rm MW}$) is allowed to vary by $\pm 20\%$.[]{data-lauel="fig:rvMW"}](figs/letfig3.ps){wudth="84mm"} We highlight xpme of thesx objests in Ficure \[fig:rvMW\], using “fioled” circles to dtnote “associated” satelljtes wvose wpocenters

entric radius, exceeding their turnaround radius by more. systems have been affected by onto substantially more energetic the ones they followed until turnaround. This mechanism seems operate preferentially on low-mass satellites, as shown by the dashed histogram in Figure which corresponds to satellites with stellar masses less than $\sim 3\%$ that of primary. velocity satellites distance to the primary. Velocities are scaled to the virial velocity of the system, distances to virial radius. Circles denote “associated” satellites; i.e., those have been [*inside*]{} the radius of the primary at earlier Crosses indicate that on first approach, and never been inside $r_{\rm vir}$. Filled circles indicate associated satellites whose apocentric radii exceed their turnaround radius at least that their energies been altered during their circles correspond to associated satellites that vir}$ during the last $3$ Gyrs. The curves the top bottom boundaries of the distribution show escape velocity of an NFW halo with concentration and $c=20$, respectively. Note that there is one satellite “escaping” the system with positive radial lines show the trajectories the $r-V_r$ plane the ”ejected” shown figure \[fig:orbesc\]. squares correspond to the fourteen brightest Milky Way satellites, taken from (complemented with NED data for the Phoenix, Tucana and NGC6822), assuming $V_{\rm vir}^{\rm MW} 109$ km/s and $r_{\rm MW}=237$ (see Sales et al indicate the satellites this would be altered if estimate of $V_{\rm vir}^{\rm MW}$ consequently, $r_{\rm vir}^{\rm MW}$) $\pm 20\%$.[]{data-label="fig:rvMW"}](figs/letfig3.ps){width="84mm"} We highlight some of these objects Figure \[fig:rvMW\], using “filled” circles to denote satellites whose apocenters

entric radius, exceeding theiR turnarounD radiUs bY $50\%$ or MoRe. ThEse sYstems have cleaRLy beEn affected by some mechanIsm thAt PRopeLLeD them Onto orbITs SUBstAnTiAllY mORe EnergEtiC than thE ones they hAd fOlLowed until tuRNaRound. This mEchAnism seems to OpeRate prEfEreNTiallY on Low-maSs sateLLites, aS shown by tHe DAshed hIStogram IN fiGure \[Fig:rhist\], which corrESpONds to satelliteS with sTeLLaR MAssEs lEss than $\sim 3\%$ ThAt of tHE primarY. ![raDIAL veLOcity of satellItes versus dIStaNce to tHe PriMAry. VelOcitiEs ARe sCaled to the vIriaL velocity Of the sYStem, disTAnces to The virIal RadIus. CIRcLeS deNoTE “asSOcIatED” saTellites; I.e., ThOse thAt haVE BEEn [*inSidE*]{} the ViriaL radius of the pRimAry aT SomE earlIer tiMe. CrOsSes inDicate SatelLiTes that are on theIr fiRst approaCh, aNd HavE nEver bEEn insiDe $r_{\Rm vIr}$. FilleD circleS IndIcATE AsSociated satellites WhOSE aPocentriC radii EXcEeD Their turNaRouNd raDIUs by aT leaST $25\%$, iNdicatinG that tHEiR oRbital eNeRgies hAvE beEn sUbstaNTialLy alteRed durinG theiR Evolution. “CrossED” circles correSPoND To ASsocIatEd satelliteS thaT Have EnteREd $R_{\rm VIr}$ durIng thE lASt $3$ gYrs. The curves delineaTiNg the tOp and Bottom boundarIes of the diSTRIbution sHow tHE eSCape velocity of An NFW Halo with coNCentratiOn $c=10$ anD $c=20$, respecTively. NotE THat there Is oNe sAteLliTE “EsCaping” the systEM With PoSitive rAdiAl velocIty. solId lIneS sHow the traJectorieS iN tHe $R-V_R$ plAne of THe two ”ejeCtEd” sAtEllItes sHOwn in fIgure \[Fig:oRbEsC\]. filLed squaREs CORresPoNd To thE foUrTeen bRighTEst milky WaY satellitEs, tAKen fRoM @vAndenbeRgh99 (complementEd With NED datA fOr tHe PhoeNIX, Tucana aNd NGC6822), and plotted assuming THat $V_{\rm vIr}^{\rM MW} \siM 109$ km/s And $r_{\rm vir}^{\Rm Mw}=237$ kpc (seE SaLEs et al 2007). arrows IndicAtE hoW THe posITIoNs oF Mw satelliteS IN thIs ploT wOuld Be alterEd if our estimate of $V_{\RM viR}^{\rm MW}$ (and, conseQueNtly, $R_{\RM vIr}^{\rM mW}$) IS alLoWEd tO VAry by $\pm 20\%$.[]{data-labeL="fig:rvMW"}](fiGs/LEtFig3.ps){width="84MM"} We HiGhlight Some of tHese oBJects in figure \[fig:RvMW\], using “FiLled” CIRclEs to denote “AssociatEd” satelliTEs whoSE aPocenTerS

entric radius, exceeding t heir turna round ra diu sby $ 50\% $ or more. The s e sy stems have clearly bee n aff ec t ed b y s ome m echanis m t h a t p ro pe lle dt he m ont o o rbits s ubstantial lymo re energetic th an the one s t hey had foll owe d unti ltur n aroun d.Thismechan i sm see ms to ope ra t e pref e rential l y o n lo w-mass satellites , a s shown by thedashed h i st o g ram in Figure \[ fi g:rhi s t\], wh i ch c o rre s ponds to sate llites with ste llar m as ses less t han $ \s i m 3 \%$ that of the primary. ![Ra d ial vel o city of satel lit esvers u sdi sta nc e to th e p r ima ry. Velo ci ti es ar e sc a l e d tothe vir ial v elocity of th e s yste m , d istan ces t o th eviria l radi us. C ir cles denote “as soci ated” sat ell it es; i .e.,t hose t hat ha ve been [*insi d e*] {} t h evirial radius of t he p ri mary atsome e a rl ie r time. C ro sse s in d i catesate l li tes that are o n t he ir firs tapproa ch , a ndhaven ever beeninside $ r_{\r m vir}$. Filled circles indic a te a ss o ciat edsatelliteswhos e apo cent r ic ra d ii ex ceedth e ir turnaround radius b yat lea st $2 5\%$, indicat ing that t h e i r orbita l en e rg i es have been s ubsta ntially al t ered dur ing t heir evo lution. “ C r ossed” c irc les co rre s p on d to associat e d sat el lites t hat have e nte red $r _{\ rm vir}$ du ring the l as t$3 $ G yrs.T he curve sdel in eat ing t h e topand b otto mbo u nda ries of th e dist ri bu tion sh ow theesca p e v elocity of an NF W h a lo w it hconcent ration $c=10$ a nd $c=20$, r esp ective l y . Note t hat there is one satell i te “esc api ng” t he s ystem wit h p ositiv e r a dial v elocit y. So li d l i n es sh o w t hetr ajectories i n t he $r -V _r$plane o f the two ”ejected ” sa tellites show n i n fi g u re \[ f ig : orb es c \]. F illed squares c orrespondto th e fourteen bri gh test Mi lky Way sate l lites,taken fro m @vanden be rgh9 9 (co mplemented with NE D data fo r theP ho enix, Tu cana a nd NG C6822 ), and plo ttedassumi ng that$V_{\ rm vir}^{\ rm MW} \sim 109$ km/s a nd $r_ {\rmvir }^{\rm MW }=2 3 7$kpc (seeSale s et al 20 07) . A rrows in d icate how th e p o sitio ns o f MW satel l it esi n t his plot wo u l d be alte red if our est imate of $V_{\rmv ir}^{\rm MW}$(and , con seq u entl y, $r_{\rm vir}^ {\r mM W }$) is a ll owed to var y by $\p m2 0\%$. []{dat a-labe l="fig: r v MW " }](fig s/le tfi g3.ps){wi dth =" 8 4mm"} We h i ghligh t so me of th ese ob j ects i n Figure \[fig:r vMW\] , using “fi lled” c irclest o de note “asso ciated” sat ellite s wh ose a pocente rs

entric radius,_exceeding their_turnaround radius by $50\%$_or more._These_systems have_clearly_been affected by_some mechanism that_propelled them onto orbits_substantially more energetic_than_the ones they had followed until turnaround. This mechanism seems to operate preferentially on_low-mass_satellites, as_shown_by_the dashed histogram in Figure \[fig:rhist\],_which corresponds to satellites with_stellar masses_less than $\sim 3\%$ that of the primary. ![Radial_velocity_of satellites versus_distance to the primary. Velocities are scaled to the_virial velocity of the system, distances_to the virial_radius._Circles_denote “associated” satellites; i.e.,_those that have been [*inside*]{} the_virial radius of the primary at_some earlier time. Crosses indicate satellites that_are on their first approach, and_have never been inside $r_{\rm vir}$._Filled circles_indicate associated satellites whose apocentric_radii exceed their_turnaround radius_by at least_$25\%$, indicating that their orbital energies_have been substantially_altered during their evolution. “Crossed” circles_correspond_to associated satellites_that_have_entered $r_{\rm_vir}$ during the_last_$3$ Gyrs._The_curves delineating the top and bottom_boundaries_of the distribution show the escape velocity_of an NFW halo_with_concentration $c=10$ and $c=20$,_respectively. Note that there is_one satellite “escaping” the system with_positive radial_velocity. Solid_lines show the trajectories in the $r-V_r$ plane of the two_”ejected” satellites shown in figure \[fig:orbesc\]._Filled squares correspond to_the fourteen_brightest_Milky Way satellites,_taken_from @vandenbergh99_(complemented with NED data for the Phoenix,_Tucana and_NGC6822), and plotted assuming that $V_{\rm_vir}^{\rm MW} \sim 109$ km/s_and_$r_{\rm vir}^{\rm MW}=237$ kpc (see Sales_et al 2007). Arrows indicate how_the positions of MW satellites_in_this_plot would be altered if_our estimate of $V_{\rm vir}^{\rm MW}$_(and, consequently, $r_{\rm_vir}^{\rm MW}$) is allowed to vary by_$\pm_20\%$.[]{data-label="fig:rvMW"}](figs/letfig3.ps){width="84mm"} We highlight some of these objects_in_Figure \[fig:rvMW\], using “filled” circles to denote_“associated”_satellites_whose apocenters

however, and thus this is simply a poor choice of representation of the system at this point. It is entirely analogous to the singularity in $\frac{\d r}{\d\theta}$ in the polar coordinate description of a straight line; there at the point of closest approach to the origin $r$ is it a minimum whilst $\theta$ changes, and thus in these variables the system appears singular. However, the equation $y=mx+c$ remains well defined at this point. The question then arises as to whether this could be the asymptotic behaviour as $\beta \rightarrow \pi/2$. The answer is that if $V$ obeys the conditions of regularity then it cannot, as the motion asymptotes to that of a free field. Recovering the free field in these circumstances amounts to setting $V$ and its derivatives to zero. Upon doing so we note two interesting features; the first is that $\chi$ becomes unimporant in dynamics, as it only enters the equations of motion for $\alpha$ and $\lambda$ through terms proportional to derivatives of the potential. Hence we can integrate the equation of motion for $\lambda$ directly in this case to obtain \_ = \[lambdafree\] wherein $\lambda_0$ is a constant. Reintroducing this into the equation of motion for $\alpha$ gives the equation for a straight line in the $(\phi_1,\phi_2)$ plane. Second, we see that this feature only relied upon there being non-zero derivatives of $V$. Had we introduced a pure cosmological constant term (corresponding to $V$ being constant) the equations of motion would have been unaffected. From equation \[lambdafree\] we see that = and so is decreasing (to zero) at the singularity. Thus when the fall-off conditions for the potential are obeyed, the system is always integrable through the singularity. Extending our system to $n$ scalar fields is a simple exercise; first we note that at any point in the evolution of the system we can choose coordinates for the scalar fields such that the position and velocity of the fields are in the plane spanned by $\phi_1$ and $\phi_2$. If we describe the positions and velocities of our fields in terms of spherical coordinates, with positions given by equation \[SphericalShapeCoordinates2\] and velocities given by $$\label{SphericalShapeCoordinates3} \left( \begin{array}{c}\phi_1\\ \phi_2 \\...\\ \phi

however, and thus this is simply a poor choice of theatrical performance of the arrangement at this point. It is entirely analogous to the singularity in $ \frac{\d r}{\d\theta}$ in the pivotal coordinate description of a true line; there at the point of close approach path to the origin $ r$ is it a minimum whilst $ \theta$ changes, and thus in these variables the organization appears singular. However, the equality $ y = mx+c$ remains well defined at this point. The question then arises as to whether this could be the asymptotic behaviour as $ \beta \rightarrow \pi/2$. The solution is that if $ V$ obeys the conditions of regularity then it cannot, as the motion asymptotes to that of a detached field. Recovering the complimentary field in these fortune amounts to setting $ V$ and its derivatives to zero. Upon doing so we notice two interesting features; the first is that $ \chi$ becomes unimporant in dynamics, as it only enters the equation of motion for $ \alpha$ and $ \lambda$ through terms proportional to derivatives of the potential. Hence we can integrate the equation of motion for $ \lambda$ directly in this case to receive \ _ = \[lambdafree\ ] wherein $ \lambda_0 $ is a constant. Reintroducing this into the equation of apparent motion for $ \alpha$ gives the equation for a straight line in the $ (\phi_1,\phi_2)$ airplane. Second, we see that this feature of speech only relied upon there being non - zero derivatives of $ V$. Had we introduce a pure cosmological constant term (corresponding to $ V$ being constant) the equations of motion would have been unaffected. From equation \[lambdafree\ ] we see that = and therefore is decreasing (to zero) at the singularity. therefore when the spill - off conditions for the potential are obeyed, the system is always integrable through the singularity. Extending our system to $ n$ scalar battlefield is a simple exercise; foremost we note that at any point in the evolution of the system we can choose coordinates for the scalar sphere such that the position and velocity of the fields are in the airplane spanned by $ \phi_1 $ and $ \phi_2$. If we describe the situation and velocities of our fields in term of spherical coordinates, with positions given by equation \[SphericalShapeCoordinates2\ ] and velocities given by $ $ \label{SphericalShapeCoordinates3 } \left (\begin{array}{c}\phi_1\\ \phi_2 \\... \\ \phi

hoaever, and thus this is slmply a poor chonxe of cepresehtation uf the system at this point. Mt iw enturely analogous to the singularpty in $\frqc{\d c}{\d\theta}$ in the polar coovbinats desermption of a strsight line; there at the [oknc of closest approach to the origin $w$ is it a minimum whilsj $\theua$ shanfvs, and thus in these variables fhe sysuem appears singulsr. However, the equation $y=md+c$ rfmains well definef at this piint. Rhe question then arists as to whejher this could be the asymptotiz behcviour as $\bgcq \rlchtarrow \pi/2$. The wnswer is that if $V$ mbeys tne conditions pf ceguoarity then it cannot, as the motion asyiptotes tm chat of a free field. Rwciveritg tve ffwe wiemd ih thesf cmrcumstancea amounts ti setting $V$ and its dqgovatives to aero. U[og doing so we note two interesting featlres; the first is that $\chi$ vecomes unimporant in dynamics, as it only enters the equations of motion for $\al[ha$ aid $\lanbqx$ tjrough terms proportional to derivatives of trs kotvntial. Hence we ccn integrate the eeusjion of motion for $\lcjbsa$ directly in thid case jo obtqin \_ = \[lamfdaftee\] wherein $\lambda_0$ is a conwtant. Reintrjeucing this into tke equation uf mption for $\alpha$ gives the eqbation for a stralght line kn the $(\phi_1,\phi_2)$ plxne. Sacond, we see that this feaeure only relned upon thete beind non-zero ferivatives of $V$. Had we lntrobuced a pure codmological constant term (corres'pnding to $V$ nehng constanc) the cquations of moeion would havg been uncffectdd. From eqlation \[lakbdafree\] we see that = ang so is decreesing (to sero) at rhe sinejlarity. Thus wnen the fcjl-ofd conditions for tme pojehtial are obeyeb, uhw system is aleayr igtvgreble evrough the shnguuarkyy. Extdnding our whstek to $n$ scalar fields is z simple exercise; givst we noje that ae any point im the evolution of the vysvem we cag choose coordinates for the sdalar fiepds such that tre pisition and relocity of the fields are in the plane wpanned by $\phi_1$ and $\phi_2$. If we describe uhe positions and delocitiev of our fields in twrms of spherical coordinates, with posjtions givej by equation \[SphericalShapeCoordinates2\] and velocities given by $$\label{SpheeicalSia[eCoordinatss3} \legt( \beciu{arxay}{c}\phi_1\\ \phi_2 \\...\\ \phi

however, and thus this is simply a of of the at this point. the in $\frac{\d r}{\d\theta}$ the polar coordinate of a straight line; there at point of closest approach to the origin $r$ is it a minimum whilst changes, and thus in these variables the system appears singular. However, the equation remains defined this The question then arises as to whether this could be the asymptotic behaviour as $\beta \rightarrow The answer is that if $V$ obeys the of regularity then it as the motion asymptotes to of free field. the field these circumstances amounts setting $V$ and its derivatives to zero. Upon doing so we note two interesting features; the first that $\chi$ in dynamics, it enters equations of motion and $\lambda$ through terms proportional to potential. Hence we can integrate the equation of for $\lambda$ in this case to obtain \_ \[lambdafree\] wherein $\lambda_0$ is a constant. Reintroducing this the equation of motion for $\alpha$ gives the equation for a straight line in the Second, we see that feature only relied there non-zero of Had we a pure cosmological constant term (corresponding to $V$ being constant) the of motion would have been unaffected. From equation \[lambdafree\] we = so is decreasing zero) at the singularity. when fall-off conditions for the obeyed, system through singularity. our system to $n$ fields is a simple exercise; we note that at of the system we can choose coordinates for scalar fields such that the position and of the fields are in the plane spanned by $\phi_1$ and $\phi_2$. we describe and velocities of our fields in terms of coordinates, with positions given equation \[SphericalShapeCoordinates2\] and velocities given by $$\label{SphericalShapeCoordinates3} \left( \begin{array}{c}\phi_1\\ \\...\\

however, and thus this is simplY a poor choiCe of rEprEseNtAtioN of tHe system at this POint. it is entirely analogous tO the sInGUlarITy In $\fraC{\d r}{\d\theTA}$ iN THe pOlAr CooRdINaTe desCriPtion of A straight lIne; ThEre at the poinT Of Closest appRoaCh to the origiN $r$ iS it a miNiMum WHilst $\TheTa$ chaNges, anD Thus in These variAbLEs the sYStem appEARs SingUlar. However, the equATiON $y=mx+c$ remains weLl defiNeD At THIs pOinT. The questiOn Then aRIses as tO WhETHEr tHIs could be the aSymptotic beHAviOur as $\bEtA \riGHtarroW \pi/2$. ThE aNSweR is that if $V$ oBeys The conditIons of REgulariTY then it Cannot, As tHe mOtioN AsYmPtoTeS To tHAt Of a FRee Field. RecOvErIng thE freE FIELd in TheSe ciRcumsTances amounts To sEttiNG $V$ aNd its DerivAtivEs To zerO. Upon dOing sO wE note two interesTing Features; tHe fIrSt iS tHat $\chI$ BecomeS unImpOrant in DynamicS, As iT oNLY EnTers the equations of MoTIOn For $\alpha$ And $\lamBDa$ ThROugh termS pRopOrtiONAl to dErivATiVes of the PotentIAl. heNce we caN iNtegraTe The EquAtion OF motIon for $\Lambda$ diRectlY In this case to obTAin \_ = \[lambdafree\] WHeREIn $\LAmbdA_0$ is A constant. ReIntrODuciNg thIS iNto THe equAtion Of MOtIOn for $\alpha$ gives the eQuAtion fOr a stRaight line in tHe $(\phi_1,\phi_2)$ plANE. second, we See tHAt THis feature only RelieD upon there BEing non-zEro deRivativeS of $V$. Had we INTroduced A puRe cOsmOloGICaL constant term (CORresPoNding to $v$ beIng consTanT) thE eqUatIoNs of motioN would haVe BeEn UnAffEcted. fRom equatIoN \[laMbDafRee\] we SEe that = And so Is deCrEaSIng (To zero) aT ThE SInguLaRiTy. ThUs wHeN the fAll-oFF coNditionS for the poTenTIal aRe ObEyed, the System is alwayS iNtegrable tHrOugH the siNGUlarity. EXtending our system to $n$ scaLAr fieldS is A simpLe exErcise; firSt wE note tHat AT any poInt in tHe evoLuTioN OF the sYSTeM we CaN choose cooRDInaTes foR tHe scAlar fieLds such that the posiTIon And velocity of The FielDS ArE in THe PLanE sPAnnED By $\phi_1$ and $\phi_2$. If we Describe thE pOSiTions and veLOciTiEs of our Fields iN termS Of spherIcal coordInates, witH pOsitIONs gIven by equaTion \[SpheRicalShapEcoordINaTes2\] anD veLocitiEs GivEn by $$\lAbel{SpHEriCalShApeCooRdInates3} \Left( \bEgIn{array}{c}\Phi_1\\ \phi_2 \\...\\ \phi

however, and thus this is simply apoorcho ice o f re pres entation of th e sys tem at this point. Itis en ti r elya na logou s to th e s i n gul ar it y i n$ \f rac{\ d r }{\d\th eta}$ in t hepo lar coordina t edescriptio n o f a straight li ne; th er e a t thepoi nt of close s t appr oach to t he origin $r$ isi t a min imum whilst $\the t a$ changes, and t hus in t h es e var iab les the sy st em ap p ears si n gu l a r . H o wever, the eq uation $y=m x +c$ remai ns we l l defi ned a tt his point. The que stion the n aris e s as to whether thiscou ldbe t h eas ymp to t icb eh avi o uras $\bet a\r ighta rrow \ p i /2$. Th e an sweris that if $V $ o beys the cond ition s of r egula rity t hen i tcannot, as themoti on asympt ote stoth at of a free fi eld . Reco veringt hefr e e fi eld in these circu ms t a nc es amoun ts tos et ti n g $V$ an dits der i v ative s to ze ro. Upon doing so w e notetw o inte re sti ngfeatu r es;the fi rst is t hat $ \ chi$ becomes u n imporant in d y na m i cs , asitonly enters the equa tion s o f m o tionfor $ \a l ph a $ and $\lambda$ thr ou gh ter ms pr oportional to derivativ e s of the p oten t ia l . Hence we can inte grate thee quationof mo tion for $\lambda $ directly in th iscas e to obtain \_ =\ [ lamb da free\]whe rein $\ lam bda _0$ is a constant . Reintr od uc in gthi s int o the equ at ion o f m otion for $\ alpha $ gi ve st heequatio n f o r a s tr ai ghtlin ein th e $( \ phi _1,\phi _2)$ plan e.S econ d, w e see t hat this feat ur e only rel ie d u pon th e r e beingnon-zero derivatives of $V$. Ha d w e int rodu ced a pur e c osmolo gic a l cons tant t erm ( co rre s p ondin g to $V $being cons t a nt) theeq uati ons ofmotion would haveb een unaffected.Fro m eq u a ti on\ [l a mbd af r ee\ ] we see that = a nd so is d ec r ea sing (to z e ro) a t the s ingular ity.T hus whe n the fal l-off con di tion s for the poten tial are obeyed,t he sy s te m isalw ays in te gra ble t hrough the sing ularit y. Exte nding o ur syste m to $n$ scalar fieldsis a s imple ex ercise; f irs t we note tha t at any point in th e evo lut i on of the sy ste m we c an c h oose coor d in ate s fo r the scala r f iel ds su cht hat th e po sition and veloci t y of the field s ar e inthe plan espanned by $\p hi_ 1$ a nd $\phi _2 $. If we de scribe t he posit ions a nd vel ocities o fo ur fie ldsinterms ofsph er i cal coo rd in a tes, w ithpo sition s give n bye q uation \[Spheric alSha p e Coord i nat es2\] a nd velo c itie s given by $$\label{S pheric alSh apeCo ordinat es 3} \le ft( \ begin{arra y }{c}\phi_ 1\\ \ phi_2 \ \. ..\\ \p hi

however,_and thus_this is simply a_poor choice_of_representation of_the_system at this_point. It is_entirely analogous to the_singularity in $\frac{\d_r}{\d\theta}$_in the polar coordinate description of a straight line; there at the point of_closest_approach to_the_origin_$r$ is it a minimum_whilst $\theta$ changes, and thus_in these_variables the system appears singular. However, the equation_$y=mx+c$_remains well defined_at this point. The question then arises as to_whether this could be the asymptotic_behaviour as $\beta_\rightarrow_\pi/2$._The answer is that_if $V$ obeys the conditions of_regularity then it cannot, as the_motion asymptotes to that of a free_field. Recovering the free field in these_circumstances amounts to setting $V$_and its_derivatives to zero. Upon doing_so we note_two interesting_features; the first_is that $\chi$ becomes unimporant in_dynamics, as it_only enters the equations of motion_for_$\alpha$ and $\lambda$_through_terms_proportional to_derivatives of the_potential._Hence we_can_integrate the equation of motion for_$\lambda$_directly in this case to obtain \__= \[lambdafree\] wherein $\lambda_0$_is_a constant. Reintroducing this_into the equation of motion_for $\alpha$ gives the equation for_a straight_line in_the $(\phi_1,\phi_2)$ plane. Second, we see that this feature only relied_upon there being non-zero derivatives of_$V$. Had we introduced_a pure_cosmological_constant term (corresponding_to_$V$ being_constant) the equations of motion would have_been unaffected._From equation \[lambdafree\] we see that_= and so is_decreasing_(to zero) at the singularity. Thus_when the fall-off conditions for the_potential are obeyed, the system_is_always_integrable through the singularity. Extending our_system to $n$ scalar fields is_a simple exercise;_first we note that at any point_in_the evolution of the system we_can_choose coordinates for the scalar fields_such_that_the position and velocity of_the fields are in the plane_spanned by $\phi_1$ and $\phi_2$. If we describe the_positions and velocities_of our fields in terms_of_spherical_coordinates, with positions given by equation \[SphericalShapeCoordinates2\] and velocities given_by $$\label{SphericalShapeCoordinates3} \left(_\begin{array}{c}\phi_1\\ \phi_2 \\...\\_\phi

_0$ (22 repetitions). In these cases 20 and 22 force-distance relations were obtained, the mean force gradients were computed and their total experimental errors were determined at a 67% confidence level as a combination of random and systematic errors (see Ref. [@27] for details). In Fig. \[fg2\](a,b) the mean gradients of the Casimir force and their errors measured for the first sample with applied compensating voltage are shown as crosses with a step of 1nm. Table  1 presents the values of mean $F^{\prime}(a)$ at several separations measured in the two different ways for the first (columns a,b) and second (columns c,d) samples. As can be seen in Table 1, the measurement results for the two graphene samples obtained in two different ways are in very good mutual agreement. Now we compare the experimental results with theoretical predictions. At the moment there is no theory allowing rigorous calculation of the Casimir force between a graphene deposited on a Si-SiO${}_2$ substrate and an Au sphere. The problem is that Si and SiO${}_2$ layers are described by their dielectric permittivities and the reflection properties of graphene in the Dirac model are described by the polarization tensor. This does not allow direct application of the Lifshitz theory for layered structures [@23; @30]. Because of this, here we restrict ourselves to the approximate approach, where the contributions of Si-SiO${}_2$ substrate and graphene sheet to the Casimir interaction with an Au sphere are computed separately using the Lifshitz theory and are then added together. In the framework of the proximity force approximation (PFA), the Lifshitz formula for the gradient of the Casimir force between an Au sphere and any planar structure takes the form $$F^{\prime}(a)=2k_BTR\sum_{l=0}^{\infty}{\vphantom{\sum}}^{\prime} \!\!\int_{0}^{\infty}\!\!\!\! q_lk_{\bot}dk_{\bot}\sum_{\alpha} \frac{r_{\alpha}^{(1)}r_{\alpha}^{(2)}}{e^{2q_la}-r_{\alpha}^{(1)}r_{\alpha}^{(2)}}. \label{eq3}$$ Here $k_B$ is the Boltzmann constant, $T=300\,$K is the laboratory temperature, $k_{\bot}$ is the projection of the wave vector on a planar structure, $q_l^2=k

_ 0 $ (22 repetitions). In these cases 20 and 22 force - distance relations were prevail, the base force gradients were computed and their full experimental errors were determined at a 67% assurance level as a combination of random and systematic errors (attend Ref.   [ @27 ] for details). In Fig.   \[fg2\](a, b) the mean gradient of the Casimir force and their errors quantify for the inaugural sample with applied compensating voltage are shown as crosses with a measure of 1 nm. Table   1 presents the value of mean $ F^{\prime}(a)$ at several separations measured in the two unlike ways for the beginning (columns a, b) and second (columns c, d) samples. As can be see in Table   1, the measurement results for the two graphene samples obtained in two different ways are in very effective mutual agreement. Now we compare the experimental results with theoretical predictions. At the moment there is no theory allowing rigorous calculation of the Casimir force between a graphene deposited on a Si - SiO${}_2 $ substrate and an Au sector. The problem is that Si and SiO${}_2 $ layers are identify by their dielectric permittivities and the mirror image properties of graphene in the Dirac model are described by the polarization tensor. This does not permit lineal application of the Lifshitz theory for layered structures [ @23; @30 ]. Because of this, here we qualify ourselves to the approximate approach, where the contributions of Si - SiO${}_2 $ substrate and graphene sheet to the Casimir interaction with an Au sphere are computed separately use the Lifshitz theory and are then added together. In the framework of the proximity violence estimate (PFA), the Lifshitz rule for the gradient of the Casimir force between an Au sphere and any planar structure takes the mannequin $ $ F^{\prime}(a)=2k_BTR\sum_{l=0}^{\infty}{\vphantom{\sum}}^{\prime } \!\!\int_{0}^{\infty}\!\!\!\! q_lk_{\bot}dk_{\bot}\sum_{\alpha } \frac{r_{\alpha}^{(1)}r_{\alpha}^{(2)}}{e^{2q_la}-r_{\alpha}^{(1)}r_{\alpha}^{(2) } }. \label{eq3}$$ Here $ k_B$ is the Boltzmann constant, $ T=300\,$K is the laboratory temperature, $ k_{\bot}$ is the projection of the wave vector on a planar structure, $ q_l^2 = k

_0$ (22 rfpetitions). In these caser 20 and 22 force-distance celatiohs were ubtained, the mean force gradmentw wert computed and theif total eqperimentql ecrors were determined at a 67% cohnidenee level as a cokbination mf random and vyrtzmatic errors (see Ref. [@27] for details). In Fig. \[fg2\](a,n) hhe mean gradignts ps ths Casimir force and their errors msasured for the firsy sample with applied compfnsahing voltage are sjown as crowses qith a step uf 1nm. Tablt  1 presents jhe values of mean $F^{\prime}(a)$ at sexeral separatiobs mewvured in thx two qifferent wanx for dhe firxt (columns a,b) snd sexond (columns c,d) samplxs. As can be seen in Table 1, tha jeasurement resulrs for jhe tfo gfqphdne sempmes obhaiied in two sifferent wqys are in very gooc ilyual agreemeht. Now re compare the experimental results with uheorstical predictions. At tye moment there is no theory ajlowing rigorous calculation of the Casimir force betwxev a gvaphdbe deposited on a Si-SiO${}_2$ substrate and an Au sphqde. Tme problem is thct Si and SiO${}_2$ lauegs wre described by thzjr dielectric permithivitief and the refltctiom properties of graphene in the Dirac mjeel are described yy the polarnzatiom tenxor. This does not allow dirsct applicahion of tgd Lifshitz theorh fpr layered structures [@23; @30]. Becatse of thms, hexe we rertrivt ourfelves to hhe ai[roximate approach, wherg the wontributilns of Si-SiO${}_2$ substrate and grapixne sheet to jhe Cavimir inceractlon with an Au fphere are comkuted sepcratelh using thv Lifshitv theory and are then addaf together. Ii the fraiewoek od the pfuximity force spproximanijn (PDA), the Lifshitz fovmula ror the gradienc if the Casimir gorze fenwexn an Du sphere ang anh pusnar rtructure tckds tne form $$F^{\prime}(a)=2k_BTR\sgm_{l=0}^{\ihfty}{\vphantom{\sum}}^{\prike} \!\!\lnt_{0}^{\infty}\!\!\!\! q_ok_{\bot}dk_{\bjt}\sum_{\alpha} \frav{r_{\alpha}^{(1)}r_{\alpha}^{(2)}}{e^{2q_la}-r_{\wlpha}^{(1)}c_{\alpha}^{(2)}}. \uabel{rq3}$$ Rere $k_B$ is the Boltzmann constznt, $T=300\,$K is thc laboratory eempcratore, $k_{\bot}$ is the projection of the wave vector on a planar structure, $w_l^2=k

_0$ (22 repetitions). In these cases 20 force-distance were obtained, mean force gradients experimental were determined at 67% confidence level a combination of random and systematic (see Ref. [@27] for details). In Fig. \[fg2\](a,b) the mean gradients of the force and their errors measured for the first sample with applied compensating voltage shown crosses a of 1nm. Table 1 presents the values of mean $F^{\prime}(a)$ at several separations measured in the different ways for the first (columns a,b) and (columns c,d) samples. As be seen in Table 1, measurement for the graphene obtained two different ways in very good mutual agreement. Now we compare the experimental results with theoretical predictions. At the moment is no rigorous calculation the force a graphene deposited Si-SiO${}_2$ substrate and an Au sphere. that Si and SiO${}_2$ layers are described by dielectric permittivities the reflection properties of graphene in Dirac model are described by the polarization tensor. does not allow direct application of the Lifshitz theory for layered structures [@23; @30]. Because here we restrict ourselves the approximate approach, the of substrate graphene sheet the Casimir interaction with an Au sphere are computed separately using Lifshitz theory and are then added together. In the framework proximity approximation (PFA), the formula for the gradient the force between an Au any structure $$F^{\prime}(a)=2k_BTR\sum_{l=0}^{\infty}{\vphantom{\sum}}^{\prime} q_lk_{\bot}dk_{\bot}\sum_{\alpha} \label{eq3}$$ Here $k_B$ is Boltzmann constant, $T=300\,$K is the temperature, $k_{\bot}$ is the on a planar structure, $q_l^2=k

_0$ (22 repetitions). In these cases 20 anD 22 force-distAnce rElaTioNs Were ObtaIned, the mean forCE graDients were computed and tHeir tOtAL expERiMentaL errors WErE DEteRmInEd aT a 67% COnFidenCe lEvel as a CombinatioN of RaNdom and systeMAtIc errors (seE ReF. [@27] for details). IN FiG. \[fg2\](a,b) tHe MeaN GradiEntS of thE CasimIR force And their eRrORs measURed for tHE FiRst sAmple with applied cOMpENsating voltage Are shoWn AS cROSseS wiTh a step of 1nM. TAble  1 pREsents tHE vALUEs oF Mean $F^{\prime}(a)$ at Several sepaRAtiOns meaSuRed IN the twO diffErENt wAys for the fiRst (cOlumns a,b) aNd secoND (columnS C,d) samplEs. As caN be SeeN in TABlE 1, tHe mEaSUreMEnT reSUltS for the tWo GrAphenE samPLES ObtaIneD in tWo difFerent ways are In vEry gOOd mUtual AgreeMent. noW we coMpare tHe expErImental results wIth tHeoreticaL prEdIctIoNs. At tHE momenT thEre Is no theOry alloWIng RiGOROuS calculation of the CAsIMIr Force betWeen a gRApHeNE depositEd On a si-Sio${}_2$ SUbstrAte aND aN Au spherE. The prOBlEm Is that SI aNd SiO${}_2$ lAyErs Are DescrIBed bY their DielectrIc perMIttivities and tHE reflection prOPeRTIeS Of grAphEne in the DirAc moDEl arE desCRiBed BY the pOlariZaTIoN Tensor. This does not alLoW direcT applIcation of the LIfshitz theORY For layerEd stRUcTUres [@23; @30]. Because of tHis, heRe we restriCT ourselvEs to tHe approxImate apprOACh, where tHe cOntRibUtiONS oF Si-SiO${}_2$ substraTE And gRaPhene shEet To the CaSimIr iNteRacTiOn with an AU sphere aRe CoMpUtEd sEparaTEly using ThE LiFsHitZ theoRY and arE then AddeD tOgETheR. In the fRAmEWOrk oF tHe ProxImiTy Force ApprOXimAtion (PFa), the LifshItz FOrmuLa FoR the graDient of the CasImIr force betWeEn aN Au sphERE and any pLanar structure takes the fORm $$F^{\primE}(a)=2k_bTR\suM_{l=0}^{\inFty}{\vphantOm{\sUm}}^{\primE} \!\!\inT_{0}^{\Infty}\!\!\!\! q_Lk_{\bot}dK_{\bot}\sUm_{\AlpHA} \Frac{r_{\ALPhA}^{(1)}r_{\aLpHa}^{(2)}}{e^{2q_la}-r_{\alpHA}^{(1)}R_{\alPha}^{(2)}}. \laBeL{eq3}$$ HEre $k_B$ is The Boltzmann constaNT, $T=300\,$K Is the laboratoRy tEmpeRATuRe, $k_{\BOt}$ IS thE pROjeCTIon of the wave vecTor on a planAr STrUcture, $q_l^2=k

_0$ (22 repetitions). In t hese cases 20 a nd22fo rce- dist ance relations were obtained, the mean fo rce g ra d ient s w ere c omputed an d the ir t ota le xp erime nta l error s were det erm in ed at a 67%c on fidence le vel as a combin ati on ofra ndo m andsys temat ic err o rs (se e Ref. [@ 27 ] for d e tails). I nFig.  \[fg2\](a,b) the me a n gradients of the C as i mi r for ceand theirer rorsm easured fo r t hef irst sample w ith applied com pensat in g v o ltageare s ho w n a s crosses w itha step of 1nm.T able  1 present s theval ues ofm ea n$F^ {\ p rim e }( a)$ atseveralse pa ratio ns m e a s u redinthetwo d ifferent ways fo r th e fi rst ( colum ns a ,b ) and secon d (co lu mns c,d) sample s. A s can besee ninTa ble 1 , the m eas ure ment re sults f o r t he t w ographene samples o bt a i ne d in two diffe r en tw ays arein ve ry g o o d mut uala gr eement. Now w e c om pare th eexperi me nta l r esult s wit h theo reticalpredi c tions. At them oment there i s n o th e oryall owing rigor ousc alcu lati o noft he Ca simir f o rc e between a graphene d eposit ed on a Si-SiO${}_ 2$ substra t e and an A u sp h er e . The problemis th at Si andS iO${}_2$ laye rs are d escribedb y their d iel ect ric pe r m it tivities andt h e re fl ectionpro perties of gr aph ene i n the Dir ac model a re d es cri bed b y the pol ar iza ti ontenso r . This does not a ll o w d irect a p pl i c atio nof the Li fs hitztheo r y f or laye red struc tur e s [@ 23 ;@30]. B ecause of thi s, here we r es tri ct our s e lves tothe approximate approac h , where th e con trib utions of Si -SiO${ }_2 $ subst rate a nd gr ap hen e sheet t othe C asimir int e r act ion w it h an Au sph ere are computed s e par ately using t heLifs h i tz th e or y an da ret h en added togeth er. In the f r am ework of t h e p ro ximityforce a pprox i mation(PFA), th e Lifshit zform u l a f or the gra dient of the Casi m ir fo r ce betw een an Au s phe re an d anyp lan ar st ructur etakesthe f or m $$F^{\ prime}(a)=2k_BTR\sum_{l =0}^{\ infty }{\ vphantom{ \su m }}^ {\prime}\!\! \int_{0}^{ \in fty }\!\! \!\ ! q_lk _{\b o t} dk_ { \bot} \sum _ {\alpha}\ fr ac{ r _ {\ alpha}^{(1) } r _ {\a lpha} ^{( 2 )}}{e^ {2q_ la}-r_{\alpha}^{( 1 )}r_{\alpha}^{ (2)} } . \l abe l {eq3 }$ $ Here $k_B$ i s t he B oltzmann c onstant, $T =300\,$K i s thelabora tory t emperat u r e, $k_{\b ot}$ is the proj ect io n of the w av e vecto r on a plana r stru c ture , $q_l^2=k

_0$ (22_repetitions). In_these cases 20 and_22 force-distance_relations_were obtained,_the_mean force gradients_were computed and_their total experimental errors_were determined at_a_67% confidence level as a combination of random and systematic errors (see Ref. [@27] for_details)._In Fig. \[fg2\](a,b)_the_mean_gradients of the Casimir force_and their errors measured for_the first_sample with applied compensating voltage are shown as_crosses_with a step_of 1nm. Table  1 presents the values of mean_$F^{\prime}(a)$ at several separations measured in_the two different_ways_for_the first (columns a,b)_and second (columns c,d) samples. As_can be seen in Table 1, the_measurement results for the two graphene samples_obtained in two different ways are_in very good mutual agreement. Now_we compare_the experimental results with theoretical_predictions. At the_moment there_is no theory_allowing rigorous calculation of the Casimir_force between a_graphene deposited on a Si-SiO${}_2$ substrate_and_an Au sphere._The_problem_is that_Si and SiO${}_2$_layers_are described_by_their dielectric permittivities and the reflection_properties_of graphene in the Dirac model are_described by the polarization_tensor._This does not allow_direct application of the Lifshitz_theory for layered structures [@23; @30]._Because of_this, here_we restrict ourselves to the approximate approach, where the contributions of_Si-SiO${}_2$ substrate and graphene sheet to_the Casimir interaction with_an Au_sphere_are computed separately_using_the Lifshitz_theory and are then added together. In_the framework_of the proximity force approximation (PFA),_the Lifshitz formula for_the_gradient of the Casimir force between_an Au sphere and any planar_structure takes the form $$F^{\prime}(a)=2k_BTR\sum_{l=0}^{\infty}{\vphantom{\sum}}^{\prime} \!\!\int_{0}^{\infty}\!\!\!\! q_lk_{\bot}dk_{\bot}\sum_{\alpha} \frac{r_{\alpha}^{(1)}r_{\alpha}^{(2)}}{e^{2q_la}-r_{\alpha}^{(1)}r_{\alpha}^{(2)}}. \label{eq3}$$_Here_$k_B$_is the Boltzmann constant, $T=300\,$K_is the laboratory temperature, $k_{\bot}$ is_the projection of_the wave vector on a planar structure,_$q_l^2=k

1}{N}{{\mathbf{A}}}^{T}{\mathbf{A}}$ converges to a finite, positive-definite matrix: $$\lim_{N\rightarrow\infty} \frac{1}{N}{{\mathbf{A}}}^{T}{\mathbf{A}}= {\mathbf{D}}\succ 0. \label{eq.sing2}$$ This assumption is also known as [*Persistent Excitation*]{} (PE), see e.g. [@c6]. Note that our assumption in Eq. (\[eq.sing\]) covers a wider range of cases. For example, Eq. (\[eq.sing\]) does not require the singular values of $\frac{1}{\sqrt{N}}{{\mathbf{A}}}$ to converge, while only requires that they lie in $[c_1,c_2]$ when $N$ increases. Classically, properties of the Least Square Estimate (LSE) under the model given in Eq. (1) are given by the Gauss-Markov theorem. It says that the Best Linear Unbiased Estimation (BLUE) of ${\mathbf{x}}^{0}$ is the LSE under certain assumptions on the noise term. For the Gauss-Markov theorem, please refer to [@c3]. However, the normal LSE does not utilize the ’sparse’ information of ${\mathbf{x}}^{0}$, which raises the question that whether it is possible to improve on the normal LSE by exploiting this information. In the literature, several approaches have been suggested, which can perform as if the true support set of ${\mathbf{x}}^{0}$ were known. Such property is termed as the ORACLE property in [@c7]. In [@c7], the SCAD (Smoothly Clipped Absolute Deviation) estimator is presented, which turns out to solve a non-convex optimization problem; later in [@c12], the ADALASSO (Adaptive Least Absolute Shrinkage and Selection Operator) estimator is presented. The ADALASSO estimator consists of two steps, which implements a normal LSE in the first step, and then solves a reweighed Lasso optimization problem, which is convex. Recently, in [@c1], two LASSO-based estimators, namely the ’A-SPARSEVA-AIC-RE’ method and the ’A-SPARSEVA-BIC-RE’ method, are suggested. Both methods need to do the LSE in the first step, then solve a Lasso

1}{N}{{\mathbf{A}}}^{T}{\mathbf{A}}$ converges to a finite, positive - definite matrix: $ $ \lim_{N\rightarrow\infty } \frac{1}{N}{{\mathbf{A}}}^{T}{\mathbf{A}}= { \mathbf{D}}\succ 0. \label{eq.sing2}$$ This assumption is also known as [ * haunting Excitation * ] { } (PE), examine e.g. [ @c6 ]. Note that our assumption in Eq. (\[eq.sing\ ]) cover a wider range of case. For example, Eq. (\[eq.sing\ ]) does not require the singular values of $ \frac{1}{\sqrt{N}}{{\mathbf{A}}}$ to converge, while only want that they lie in $ [ c_1,c_2]$ when $ N$ increases. Classically, place of the Least Square Estimate (LSE) under the model given in Eq. (1) are given by the Gauss - Markov theorem. It says that the Best Linear Unbiased Estimation (BLUE) of $ { \mathbf{x}}^{0}$ is the LSE under certain assumptions on the randomness term. For the Gauss - Markov theorem, please refer to [ @c3 ]. However, the normal LSE does not utilize the ’ sparse ’ data of $ { \mathbf{x}}^{0}$, which raises the question that whether it is possible to improve on the normal LSE by exploiting this data. In the literature, several approach path have been suggested, which can perform as if the true documentation set of $ { \mathbf{x}}^{0}$ were known. Such property is termed as the ORACLE property in [ @c7 ]. In [ @c7 ], the SCAD (Smoothly Clipped Absolute Deviation) estimator is presented, which turns out to solve a non - convex optimization problem; later in [ @c12 ], the ADALASSO (Adaptive Least Absolute Shrinkage and Selection Operator) estimator is presented. The ADALASSO estimator consists of two steps, which implements a normal LSE in the inaugural step, and then solves a reweighed Lasso optimization trouble, which is convex. Recently, in [ @c1 ], two LASSO - free-base calculator, namely the ’ A - SPARSEVA - AIC - RE ’ method and the ’ A - SPARSEVA - BIC - RE ’ method, are suggested. Both methods need to do the LSE in the inaugural step, then solve a Lasso

1}{N}{{\mahhbf{A}}}^{T}{\mathbf{A}}$ converges uo a finite, positnce-defiiite mafrix: $$\lim_{V\rightarrow\infty} \frac{1}{N}{{\mathbf{E}}}^{T}{\marhbf{A}}= {\mathbf{D}}\succ 0. \labeu{eq.sing2}$$ Tjis assunptiib is also known as [*Persiabent Zxritation*]{} (PE), see e.g. [@c6]. Note that our assukpgiln in Eq. (\[eq.sing\]) covers a wider rangq of caxed. For example, Gq. (\[eq.xyng\]) soes not require the singular valuss of $\fgac{1}{\sqrt{N}}{{\mathbf{A}}}$ tp converge, while only requlres that they lie in $[f_1,c_2]$ when $N$ ibcrewwes. Classicaluy, propertpzs of the Lgast Square Estimate (LSE) under tfe mobel given ib Wq. (1) dre given bb the Dauss-Markov bneorem. It sayx that the Besb Linxar Ynbiased Estimation (BNUE) of ${\mathbf{x}}^{0}$ is jhe LSE ungex certain assumptions ob the noive tdem. Wor tie Fauss-Mwrkkv theorem, please refwr to [@c3]. However, the njgkal LSE does not ueijize the ’sparse’ information of ${\mathbf{x}}^{0}$, fhidh raises the question rhat whether it is podsible to improve on the normal LSE by exploiting this infmrmatmov. Iu the uutfrature, several approaches have been suggesteq, wnibh can perform as if the trur dulkort set of ${\majhbf{x}}^{0}$ wzde known. Such properhy is tgrmed qs the ORWCLE property in [@c7]. In [@c7], the SCAE (Smoothly Cjupped Absolute Devnation) estimctor ix prexented, which turns out co solbe a non-congex optimjxation problem; lxteg in [@c12], the ADALASSO (Adaptive Lqast Absooute Shrinkxge snd Sejection Opfratov) estimator is presfnted. Tve ADALASSL estimator consists of two ste'x, which implgmetts a normaj LSE in the first step, and then solves a redeighed Laaso optmmization prjblem, which iv convex. Receitly, in [@c1], two LASWO-based dstimators, namrly the ’A-SPARSEVA-AIX-RE’ method and the ’A-SKADSEVA-BIC-RE’ methid, qre suggested. Bpth meehldx gaed to do tha LSD iv the wirst suzp, bhev sokve a Lasso

1}{N}{{\mathbf{A}}}^{T}{\mathbf{A}}$ converges to a finite, positive-definite matrix: {\mathbf{D}}\succ \label{eq.sing2}$$ This is also known e.g. Note that our in Eq. (\[eq.sing\]) a wider range of cases. For Eq. (\[eq.sing\]) does not require the singular values of $\frac{1}{\sqrt{N}}{{\mathbf{A}}}$ to converge, while requires that they lie in $[c_1,c_2]$ when $N$ increases. Classically, properties of the Square (LSE) the given in Eq. (1) are given by the Gauss-Markov theorem. It says that the Best Linear Estimation (BLUE) of ${\mathbf{x}}^{0}$ is the LSE under assumptions on the noise For the Gauss-Markov theorem, please to However, the LSE not the ’sparse’ information ${\mathbf{x}}^{0}$, which raises the question that whether it is possible to improve on the normal LSE by this information. literature, several have suggested, can perform as true support set of ${\mathbf{x}}^{0}$ were is termed as the ORACLE property in [@c7]. [@c7], the (Smoothly Clipped Absolute Deviation) estimator is which turns out to solve a non-convex optimization later in [@c12], the ADALASSO (Adaptive Least Absolute Shrinkage and Selection Operator) estimator is presented. estimator consists of two which implements a LSE the step, then solves reweighed Lasso optimization problem, which is convex. Recently, in [@c1], two estimators, namely the ’A-SPARSEVA-AIC-RE’ method and the ’A-SPARSEVA-BIC-RE’ method, are methods to do the in the first step, solve Lasso

1}{N}{{\mathbf{A}}}^{T}{\mathbf{A}}$ converges To a finite, pOsitiVe-dEfiNiTe maTrix: $$\Lim_{N\rightarrow\INfty} \Frac{1}{N}{{\mathbf{A}}}^{T}{\mathbf{A}}= {\maThbf{D}}\SuCC 0. \labEL{eQ.sing2}$$ this assUMpTIOn iS aLsO knOwN As [*persiSteNt ExcitAtion*]{} (PE), see E.g. [@c6]. noTe that our assUMpTion in Eq. (\[eq.SinG\]) covers a wideR raNge of cAsEs. FOR examPle, eq. (\[eq.sIng\]) doeS Not reqUire the siNgULar valUEs of $\fraC{1}{\SQrT{N}}{{\maThbf{A}}}$ to converge, whILe ONly requires thaT they lIe IN $[c_1,C_2]$ WHen $n$ inCreases. ClaSsIcallY, PropertIEs OF THe LEAst Square EstiMate (LSE) undeR The Model gIvEn iN eq. (1) are gIven bY tHE GaUss-Markov thEoreM. It says thAt the BESt LineaR unbiaseD EstimAtiOn (BlUE) oF ${\MaThBf{x}}^{0}$ Is THe Lse uNdeR CerTain assuMpTiOns on The nOISE Term. for The GAuss-MArkov theorem, pLeaSe reFEr tO [@c3]. HowEver, tHe noRmAl LSE Does noT utilIzE the ’sparse’ inforMatiOn of ${\mathbF{x}}^{0}$, wHiCh rAiSes thE QuestiOn tHat Whether It is posSIblE tO IMPrOve on the normal LSE bY eXPLoIting thiS inforMAtIoN. in the litErAtuRe, seVERal apProaCHeS have beeN suggeSTeD, wHich can PeRform aS iF thE trUe supPOrt sEt of ${\maThbf{x}}^{0}$ werE knowN. such property is TErmed as the ORAclE PROpERty iN [@c7]. IN [@c7], the SCAD (SmOothLY CliPped aBsOluTE DeviAtion) EsTImATor is presented, which TuRns out To solVe a non-convex oPtimizatioN PROblem; latEr in [@C12], ThE aDALASSO (AdaptiVe LeaSt Absolute sHrinkage And SeLection OPerator) esTIMator is pResEntEd. THe AdalAsSO estimator cONSistS oF two stePs, wHich impLemEntS a nOrmAl lSE in the fIrst step, AnD tHeN sOlvEs a reWEighed LaSsO opTiMizAtion PRoblem, Which Is coNvEx. rEceNtly, in [@c1], TWo lasSO-bAsEd EstiMatOrS, nameLy thE ’a-SPaRSEVA-AiC-RE’ methoD anD The ’A-sPaRsEVA-BIC-rE’ method, are suGgEsted. Both mEtHodS need tO DO the LSE iN the first step, then solve a lAsso

1}{N}{{\mathbf{A}}}^{T}{\m athbf{A}}$ conv erg esto a f init e, positive-de f init e matrix: $$\lim_{N\ri ghtar ro w \inf t y} \fra c{1}{N} { {\ m a thb f{ A} }}^ {T } {\ mathb f{A }}= {\m athbf{D}}\ suc c0. \lab e l{ eq.sing2}$ $ T his assumpti onis als okno w n as[*P ersis tent E x citati on*]{} (P E) , see e . g. [@c6 ] . N otethat our assumpti o ni n Eq. (\[eq.si ng\])co v er s a w ide r range of c ases. For exa m pl e , Eq. (\[eq.sing\]) does not r e qui re the s ing u lar va luesof $\f rac{1}{\sqr t{N} }{{\mathb f{A}}} $ to con v erge, w hile o nly re quir e sth atth e y l i ein$ [c_ 1,c_2]$wh en $N$incr e a s e s. Cla ssic ally, properties o f t he L e ast Squa re Es tima te (LSE ) unde r the m odel given in E q. ( 1) are gi ven b y t he Gaus s -Marko v t heo rem. It says t h atth e B es t Linear UnbiasedEs t i ma tion (BL UE) of ${ \m a thbf{x}} ^{ 0}$ ist h e LSE und e rcertainassump t io ns on the n oise t er m.For theG auss -Marko v theore m, pl e ase refer to [ @ c3]. However, th e no r malLSE does not u tili z e th e ’s p ar se’ infor matio no f$ {\mathbf{x}}^{0}$,wh ich ra isesthe questionthat wheth e r it is po ssib l et o improve on t he no rmal LSE b y exploit ing t his info rmation.I n the lit era tur e,sev e r al approaches h a v e be en sugges ted , which ca n p erf orm a s if thetrue sup po rt s et of ${\m a thbf{x}} ^{ 0}$ w ere know n . Such prop erty i st erm ed as t h eO R ACLE p ro pert y i n[@c7] . In [@c 7], the SCAD (Sm oot h ly C li pp ed Abso lute Deviatio n) estimator i s p resent e d , whichturns out to solve a no n -convex op timiz atio n problem ; l ater i n [ @ c12],the AD ALASS O(Ad a p tiveL e as t A bs olute Shri n k age andSe lect ion Ope rator) estimator i s pr esented. TheADA LASS O es tim a to r co ns i sts o f two steps, wh ich implem en t sa normal L S E i nthe fir st step , and then so lves a re weighed L as so o p t imi zation pro blem, wh ich is co n vex.R ec ently , i n [@c1 ], tw o LAS SO-bas e d e stima tors,na mely t he ’A -S PARSEVA- AIC-RE’ method and the’A-SPA RSEVA -BI C-RE’ met hod , ar e suggest ed.Both metho dsnee d todot he LS E in th e f i rst s tep, then solv e a La s s o

1}{N}{{\mathbf{A}}}^{T}{\mathbf{A}}$ converges_to a_finite, positive-definite matrix: $$\lim_{N\rightarrow\infty}_\frac{1}{N}{{\mathbf{A}}}^{T}{\mathbf{A}}= {\mathbf{D}}\succ_0. _ __ \label{eq.sing2}$$ This_assumption is also_known as [*Persistent Excitation*]{}_(PE), see e.g._[@c6]._Note that our assumption in Eq. (\[eq.sing\]) covers a wider range of cases. For_example,_Eq. (\[eq.sing\])_does_not_require the singular values of_$\frac{1}{\sqrt{N}}{{\mathbf{A}}}$ to converge, while only_requires that_they lie in $[c_1,c_2]$ when $N$ increases. Classically, properties_of_the Least Square_Estimate (LSE) under the model given in Eq. (1)_are given by the Gauss-Markov theorem._It says that_the_Best_Linear Unbiased Estimation (BLUE)_of ${\mathbf{x}}^{0}$ is the LSE under_certain assumptions on the noise term._For the Gauss-Markov theorem, please refer to_[@c3]. However, the normal LSE does_not utilize the ’sparse’ information_of ${\mathbf{x}}^{0}$,_which raises the question that_whether it is_possible to_improve on the_normal LSE by exploiting this information._In the literature,_several approaches have been suggested, which_can_perform as if_the_true_support set_of ${\mathbf{x}}^{0}$ were_known._Such property_is_termed as the ORACLE property in_[@c7]._In [@c7], the SCAD (Smoothly Clipped Absolute_Deviation) estimator is presented,_which_turns out to solve_a non-convex optimization problem; later_in [@c12], the ADALASSO (Adaptive Least_Absolute Shrinkage_and Selection_Operator) estimator is presented. The ADALASSO estimator consists of two steps,_which implements a normal LSE in_the first step, and_then solves_a_reweighed Lasso optimization_problem,_which is_convex. Recently, in [@c1], two LASSO-based estimators,_namely the_’A-SPARSEVA-AIC-RE’ method and the ’A-SPARSEVA-BIC-RE’ method,_are suggested. Both methods_need_to do the LSE in the_first step, then solve a Lasso

P_n \circ \left( M^{f,\mathbb{X}} \right) ^{-1}; \text{ }f: \lbrace 0, \dots, n \rbrace \times \mathbb{R}^{n+1} \rightarrow \sqrt{\mathbf{D}} \text{ is adapted}\right\}. $ Then $\mathcal{Q}_{\mathbf{D}}^{n} \subseteq \mathcal{P}_{\mathbf{D}}^{n}$. \[mthm\] Let $\xi: \Omega \rightarrow \mathbb{R}$ be a continuous function satisfying $|\xi(\omega)| \leq a(1+\parallel \omega\parallel_{\infty})^b$ for some constants $a,b >0.$ Then, - $$\label{main-thm} \lim _{n \rightarrow \infty} \sup_{\mathbb{Q} \in \mathcal{Q}_{\mathbf{D}^{\prime}_n/n}^{n}} \mathbb{E}^{\mathbb{Q}} [\xi (\widehat{X}^n)] = \sup_{P \in \mathcal{Q}_{\mathbf{D}}} \mathbb{E}^{P} [\xi].$$ - $$\label{max-eq} \sup_{\mathbb{Q} \in \mathcal{Q}_{\mathbf{D}^{\prime}_n/n}^{n}} \mathbb{E}^{\mathbb{Q}} [\xi (\widehat{X}^n)] = \max_{\mathbb{Q} \in \mathcal{Q}_{\mathbf{D}^{\prime}_n/n}^{n}} \mathbb{E}^{\mathbb{Q}} [\xi (\widehat{X}^n)].$$ To prove, we prove two separate inequalities together with a density argument. The left-hand side of can be written as $$\sup_{\mathbb{Q} \in \mathcal{Q}_{\mathbf{D}^{\prime}_n/n}^{n}} \mathbb{E}^{\mathbb{Q}} [\xi (\widehat{X}^n)] = \sup_{f \in \mathcal{A} } \mathbb{E}^{P_n \circ (M^{f, \mathbb{X}})^{-1}} [\xi (\widehat{X}^n)],$$ where $\mathcal{A} = \left\lbrace f: \lbrace 0, \dots, n \rbrace \times \mathcal{L}^{n+1}_{n} \rightarrow \sqrt{\

P_n \circ \left (M^{f,\mathbb{X } } \right) ^{-1 }; \text { } f: \lbrace 0, \dots, n \rbrace \times \mathbb{R}^{n+1 } \rightarrow \sqrt{\mathbf{D } } \text { is adapted}\right\ }. $ Then $ \mathcal{Q}_{\mathbf{D}}^{n } \subseteq \mathcal{P}_{\mathbf{D}}^{n}$. \[mthm\ ] Let $ \xi: \Omega \rightarrow \mathbb{R}$ be a continuous function satisfying $ |\xi(\omega)| \leq a(1+\parallel \omega\parallel_{\infty})^b$ for some constants $ deoxyadenosine monophosphate, boron > 0.$ Then, - $ $ \label{main - thm } \lim _ { n \rightarrow \infty } \sup_{\mathbb{Q } \in \mathcal{Q}_{\mathbf{D}^{\prime}_n / n}^{n } } \mathbb{E}^{\mathbb{Q } } [ \xi (\widehat{X}^n) ] = \sup_{P \in \mathcal{Q}_{\mathbf{D } } } \mathbb{E}^{P } [ \xi].$$ - $ $ \label{max - eq } \sup_{\mathbb{Q } \in \mathcal{Q}_{\mathbf{D}^{\prime}_n / n}^{n } } \mathbb{E}^{\mathbb{Q } } [ \xi (\widehat{X}^n) ] = \max_{\mathbb{Q } \in \mathcal{Q}_{\mathbf{D}^{\prime}_n / n}^{n } } \mathbb{E}^{\mathbb{Q } } [ \xi (\widehat{X}^n)].$$ To prove, we test two separate inequalities in concert with a density argument. The left - hired hand side of can be written as $ $ \sup_{\mathbb{Q } \in \mathcal{Q}_{\mathbf{D}^{\prime}_n / n}^{n } } \mathbb{E}^{\mathbb{Q } } [ \xi (\widehat{X}^n) ] = \sup_{f \in \mathcal{A } } \mathbb{E}^{P_n \circ (M^{f, \mathbb{X}})^{-1 } } [ \xi (\widehat{X}^n)],$$ where $ \mathcal{A } = \left\lbrace farad: \lbrace 0, \dots, n \rbrace \times \mathcal{L}^{n+1}_{n } \rightarrow \sqrt{\

P_n \circ \left( M^{f,\mathbb{X}} \rinht) ^{-1}; \text{ }f: \lbraew 0, \dotv, n \rbdace \timds \mathbb{R}^{n+1} \rightarrow \sqrt{\methbd{D}} \tezt{ is adapted}\right\}. $ Then $\matjcal{Q}_{\matybf{D}}^{i} \subseteq \mathcem{P}_{\mathbn{B}}^{n}$. \[mthj\] Pet $\ri: \Omega \rightartow \mathbb{R}$ te a continuouv wuuction satisfying $|\xi(\omega)| \leq a(1+\paraljel \omeba\oarallel_{\infty})^b$ for fome bokstants $a,b >0.$ Then, - $$\label{main-thm} \lim _{n \rigntarrow \infty} \sup_{\mathbb{Q} \ij \mahhcal{Q}_{\mathbf{D}^{\prime}_j/n}^{n}} \mathbb{E}^{\nathfv{Q}} [\xi (\widehag{X}^n)] = \sup_{P \in \mathcal{Q}_{\jathbf{D}}} \mathbb{E}^{P} [\xi].$$ - $$\label{max-ed} \suk_{\mztjtb{Q} \in \mathral{Q}_{\manhbf{D}^{\prime}_n/n}^{n}} \mathbb{E}^{\kathbb{Q}} [\xi (\widehat{X}^n)] = \mex_{\marhbb{Q} \in \mathcal{Q}_{\mathuf{D}^{\prime}_n/n}^{n}} \mathbb{E}^{\mwthbb{Q}} [\xi (\wndehat{X}^n)].$$ To prove, we peoce twm se[aragw iveqhakifies tlgevher with a density artument. The left-hand syee of can be sritteg ws $$\sup_{\mathbb{Q} \in \mathcal{Q}_{\mathbf{D}^{\prime}_n/n}^{t}} \mzthbb{E}^{\mathbb{Q}} [\xi (\widehar{X}^n)] = \sup_{f \in \mathcal{W} } \mathbb{Q}^{P_n \circ (M^{f, \mathbb{X}})^{-1}} [\xi (\widehat{X}^n)],$$ where $\mathcal{A} = \left\mcraec w: \lhrace 0, \dots, n \rbrace \times \mathcal{L}^{n+1}_{n} \rightarwkw \sart{\

P_n \circ \left( M^{f,\mathbb{X}} \right) ^{-1}; \text{ 0, n \rbrace \mathbb{R}^{n+1} \rightarrow \sqrt{\mathbf{D}} $\mathcal{Q}_{\mathbf{D}}^{n} \mathcal{P}_{\mathbf{D}}^{n}$. \[mthm\] Let \Omega \rightarrow \mathbb{R}$ a continuous function satisfying $|\xi(\omega)| \leq \omega\parallel_{\infty})^b$ for some constants $a,b >0.$ Then, - $$\label{main-thm} \lim _{n \rightarrow \infty} \in \mathcal{Q}_{\mathbf{D}^{\prime}_n/n}^{n}} \mathbb{E}^{\mathbb{Q}} [\xi (\widehat{X}^n)] = \sup_{P \in \mathcal{Q}_{\mathbf{D}}} \mathbb{E}^{P} [\xi].$$ - $$\label{max-eq} \in \mathbb{E}^{\mathbb{Q}} (\widehat{X}^n)] \max_{\mathbb{Q} \in \mathcal{Q}_{\mathbf{D}^{\prime}_n/n}^{n}} \mathbb{E}^{\mathbb{Q}} [\xi (\widehat{X}^n)].$$ To prove, we prove two separate inequalities together with a argument. The left-hand side of can be written $$\sup_{\mathbb{Q} \in \mathcal{Q}_{\mathbf{D}^{\prime}_n/n}^{n}} \mathbb{E}^{\mathbb{Q}} (\widehat{X}^n)] = \sup_{f \in \mathcal{A} \mathbb{E}^{P_n (M^{f, \mathbb{X}})^{-1}} (\widehat{X}^n)],$$ $\mathcal{A} \left\lbrace f: \lbrace \dots, n \rbrace \times \mathcal{L}^{n+1}_{n} \rightarrow \sqrt{\

P_n \circ \left( M^{f,\mathbb{X}} \right) ^{-1}; \tExt{ }f: \lbrace 0, \Dots, n \RbrAce \TiMes \mAthbB{R}^{n+1} \rightarrow \sQRt{\maThbf{D}} \text{ is adapted}\righT\}. $ Then $\MaTHcal{q}_{\MaThbf{D}}^{N} \subsetEQ \mATHcaL{P}_{\MaThbF{D}}^{N}$. \[MtHm\] Let $\Xi: \OMega \rigHtarrow \matHbb{r}$ bE a continuous FUnCtion satisFyiNg $|\xi(\omega)| \leq A(1+\paRallel \OmEga\PArallEl_{\iNfty})^b$ For somE ConstaNts $a,b >0.$ Then, - $$\LaBEl{main-THm} \lim _{n \rIGHtArroW \infty} \sup_{\mathbb{Q} \iN \MaTHcal{Q}_{\mathbf{D}^{\prIme}_n/n}^{n}} \MaTHbB{e}^{\MatHbb{q}} [\xi (\widehat{x}^n)] = \Sup_{P \iN \Mathcal{q}_{\MaTHBF{D}}} \mAThbb{E}^{P} [\xi].$$ - $$\label{Max-eq} \sup_{\matHBb{Q} \In \mathCaL{Q}_{\mAThbf{D}^{\pRime}_n/N}^{n}} \MAthBb{E}^{\mathbb{Q}} [\xI (\widEhat{X}^n)] = \max_{\Mathbb{q} \In \mathcAL{Q}_{\mathbF{D}^{\primE}_n/n}^{N}} \maThbb{e}^{\MaThBb{Q}} [\Xi (\WIdeHAt{x}^n)].$$ TO ProVe, we provE tWo SeparAte iNEQUAlitIes TogeTher wIth a density arGumEnt. THE leFt-hanD side Of caN bE writTen as $$\sUp_{\matHbB{Q} \in \mathcal{Q}_{\matHbf{D}^{\Prime}_n/n}^{n}} \mAthBb{e}^{\maThBb{Q}} [\xi (\WIdehat{x}^n)] = \sUp_{f \In \mathcAl{A} } \mathBB{E}^{P_N \cIRC (m^{f, \Mathbb{X}})^{-1}} [\xi (\widehat{X}^n)],$$ WhERE $\mAthcal{A} = \lEft\lbrACe F: \lBRace 0, \dots, N \rBraCe \tiMES \mathCal{L}^{N+1}_{N} \rIghtarroW \sqrt{\

P_n \circ \left( M^{f,\m athbb{X}}\righ t)^{- 1} ; \t ext{ }f: \lbrace 0 , \do ts, n \rbrace \times \ mathb b{ R }^{n + 1} \rig htarrow \s q r t{\ ma th bf{ D} } \ text{ is adapte d}\right\} . $ Then $\ma t hc al{Q}_{\ma thb f{D}}^{n} \s ubs eteq \ ma thc a l{P}_ {\m athbf {D}}^{ n }$. \ [mthm\] Le t $\xi: \Omega\ r ig htar row \mathbb{R}$ b e a continuous fun ctionsa t is f y ing $| \xi(\omega )| \leq a(1+\pa r al l e l \o m ega\parallel_ {\infty})^b $ fo r some c ons t ants $ a,b > 0. $ Th en, - $$ \lab el{main-t hm} \lim _ {n \ri ght arr ow \ i nf ty } \ su p _{\ m at hbb { Q}\in \mat hc al {Q}_{ \mat h b f { D}^{ \pr ime} _n/n} ^{n}} \mathbb {E} ^{\m a thb b{Q}} [\xi (\w id ehat{ X}^n)] = \ su p_{P \in \mathc al{Q }_{\mathb f{D }} } \ ma thbb{ E }^{P}[\x i]. $$ - $$\lab e l{m ax - e q } \sup_{ \m a t hb b{Q} \in \math c al {Q } _{\mathb f{ D}^ {\pr i m e}_n/ n}^{ n }} \mathbb {E}^{\ m at hb b{Q}} [ \x i (\wi de hat {X} ^n)]= \m ax_{\m athbb{Q} \in\ mathcal{Q}_{\m a thbf{D}^{\pri m e} _ n /n } ^{n} } \ mathbb{E}^{ \mat h bb{Q }} [ \ xi (\ w ideha t{X}^ n) ] .$ $ To prove, we prov etwo se parat e inequalitie s together w i th a den sity ar g ument. The lef t-han d side ofc an be wr itten as $$\s up_{\math b b {Q} \in\ma thc al{ Q}_ { \ ma thbf{D}^{\pri m e }_n/ n} ^{n}} \ mat hbb{E}^ {\m ath bb{ Q}} [ \xi (\wid ehat{X}^ n) ]= \ sup _{f \ i n \mathc al {A} } \m athbb { E}^{P_ n \ci rc ( M^ {f , \m athbb{X } }) ^ { -1}} [ \x i (\ wid eh at{X} ^n)] , $$where $ \mathcal{ A}= \l ef t\ lbracef: \lbrace 0, \ dots, n \r br ace \time s \mathcal {L}^{n+1}_{n} \rightarr o w \sqrt {\

P_n_ \circ_\left( M^{f,\mathbb{X}} \right) ^{-1};_\text{ }f:_\lbrace_0, \dots,_n_\rbrace \times \mathbb{R}^{n+1}_\rightarrow \sqrt{\mathbf{D}} \text{_is adapted}\right\}. _ $ Then_$\mathcal{Q}_{\mathbf{D}}^{n}_\subseteq \mathcal{P}_{\mathbf{D}}^{n}$. \[mthm\] Let $\xi: \Omega \rightarrow \mathbb{R}$ be a continuous function satisfying $|\xi(\omega)| \leq a(1+\parallel_\omega\parallel_{\infty})^b$_for some_constants_$a,b_>0.$ Then, - $$\label{main-thm} _ _ _ \lim _{n \rightarrow \infty}_\sup_{\mathbb{Q}_\in \mathcal{Q}_{\mathbf{D}^{\prime}_n/n}^{n}} \mathbb{E}^{\mathbb{Q}}_[\xi (\widehat{X}^n)] = \sup_{P \in \mathcal{Q}_{\mathbf{D}}} \mathbb{E}^{P} [\xi].$$ -_ $$\label{max-eq} _ ___ \sup_{\mathbb{Q}_\in \mathcal{Q}_{\mathbf{D}^{\prime}_n/n}^{n}} \mathbb{E}^{\mathbb{Q}} [\xi (\widehat{X}^n)] =_ \max_{\mathbb{Q} \in \mathcal{Q}_{\mathbf{D}^{\prime}_n/n}^{n}} \mathbb{E}^{\mathbb{Q}} [\xi_(\widehat{X}^n)].$$ To prove, we prove two separate inequalities_together with a density argument. The_left-hand side of can be_written as_$$\sup_{\mathbb{Q} \in \mathcal{Q}_{\mathbf{D}^{\prime}_n/n}^{n}} \mathbb{E}^{\mathbb{Q}} [\xi_(\widehat{X}^n)] = _\sup_{f \in_\mathcal{A} } \mathbb{E}^{P_n_\circ (M^{f, \mathbb{X}})^{-1}} [\xi (\widehat{X}^n)],$$ where_$\mathcal{A} = _\left\lbrace f: \lbrace 0, \dots, n_\rbrace_\times \mathcal{L}^{n+1}_{n} \rightarrow_\sqrt{\

}_{N,1} {\mathfrak{p}}^{2}_{N,y}$ ${\mathfrak{p}}^{2}_{K,1}$ ${\mathfrak{p}}_{F,1}$ ${\mathfrak{p}}_{F_+,1} {\mathfrak{p}}_{F_+,y}$ ${\mathfrak{p}}_{F^{r},1}$ ${\mathfrak{p}}^{2}_{K^{r},1}$ $p$ (3) $\langle y^2\rangle$ $\langle x,\ y^2\rangle$ ${\mathfrak{p}}^{2}_{N,1} {\mathfrak{p}}^{2}_{N,y}$ ${\mathfrak{p}}^{2}_{K,1} {\mathfrak{p}}^{2}_{K,y}$ ${\mathfrak{p}}_{F,1} {\mathfrak{p}}_{F, y}$ ${\mathfrak{p}}_{F_+,1}$ ${\mathfrak{p}}_{F^{r},1}$ ${\mathfrak{p}}^{2}_{K^{r},1}$ $p$ (4)\* $\langle y^2\rangle$ $\langle xy,\ y^2\rangle$ ${\mathfrak{p}}^{2}_{N,1} {\mathfrak{p}}^{2}_{N,y}$ ${\mathfrak{p}}^{2}_{K,1}$ ${\mathfrak{p}}_{F,1}$ ${\mathfrak{p}}_{F_+,1}$ ${\mathfrak{p}}_{F^{r},1} {\mathfrak{p}}_{F^{r},y}$ ${\mathfrak{p}}^{2}_{K^{r},1} {\mathfrak{p}}^{2}_{K^{r},y}$ ${\mathfrak{p}}_{K,1}$ (a) $\langle x\rangle$ $\langle x\rangle$ ${\mathfrak{p}}^{2}_{N,1} {\mathfrak{p}}^{2}_{N,y}{\mathfrak{p}}^{2}_{N,y^2}{\mathfrak{p}}^{2}_{N,y^3}$ ${\mathfrak{p}}_{K,1}{\mathfrak{p}}^{2}_{K,y}{\mathfrak{p}}_{K,y^2}$ ${\mathfrak{p}}_{F,1}{\mathfrak{p}}_{F,y}$ ${\mathfrak{p}}_{F_+,1}^2$ ${\mathfrak{p}}^{

} _ { N,1 } { \mathfrak{p}}^{2}_{N, y}$ $ { \mathfrak{p}}^{2}_{K,1}$ $ { \mathfrak{p}}_{F,1}$ $ { \mathfrak{p}}_{F_+,1 } { \mathfrak{p}}_{F_+,y}$ $ { \mathfrak{p}}_{F^{r},1}$ $ { \mathfrak{p}}^{2}_{K^{r},1}$ $ p$ (3) $ \langle y^2\rangle$ $ \langle x,\ y^2\rangle$ $ { \mathfrak{p}}^{2}_{N,1 } { \mathfrak{p}}^{2}_{N, y}$ $ { \mathfrak{p}}^{2}_{K,1 } { \mathfrak{p}}^{2}_{K, y}$ $ { \mathfrak{p}}_{F,1 } { \mathfrak{p}}_{F, y}$ $ { \mathfrak{p}}_{F_+,1}$ $ { \mathfrak{p}}_{F^{r},1}$ $ { \mathfrak{p}}^{2}_{K^{r},1}$ $ p$ (4)\ * $ \langle y^2\rangle$ $ \langle xy,\ y^2\rangle$ $ { \mathfrak{p}}^{2}_{N,1 } { \mathfrak{p}}^{2}_{N, y}$ $ { \mathfrak{p}}^{2}_{K,1}$ $ { \mathfrak{p}}_{F,1}$ $ { \mathfrak{p}}_{F_+,1}$ $ { \mathfrak{p}}_{F^{r},1 } { \mathfrak{p}}_{F^{r},y}$ $ { \mathfrak{p}}^{2}_{K^{r},1 } { \mathfrak{p}}^{2}_{K^{r},y}$ $ { \mathfrak{p}}_{K,1}$ (a) $ \langle x\rangle$ $ \langle x\rangle$ $ { \mathfrak{p}}^{2}_{N,1 } { \mathfrak{p}}^{2}_{N, y}{\mathfrak{p}}^{2}_{N, y^2}{\mathfrak{p}}^{2}_{N, y^3}$ $ { \mathfrak{p}}_{K,1}{\mathfrak{p}}^{2}_{K, y}{\mathfrak{p}}_{K, y^2}$ $ { \mathfrak{p}}_{F,1}{\mathfrak{p}}_{F, y}$ $ { \mathfrak{p}}_{F_+,1}^2 $ $ { \mathfrak{p}}^ {

}_{N,1} {\mwthfrak{p}}^{2}_{N,y}$ ${\oathfrak{p}}^{2}_{K,1}$ ${\mathfrak{p}}_{F,1}$ ${\mathfeak{p}}_{H_+,1} {\mathfrak{p}}_{F_+,y}$ ${\mafmfrak{'}}_{F^{c},1}$ ${\mcthfrak{p}}^{2}_{K^{r},1}$ $p$ (3) $\langle y^2\rzngle$ $\langle x,\ u^2\rangle$ ${\mathfrak{p}}^{2}_{N,1} {\lathfrak{p}}^{2}_{N,y}$ ${\majhfrak{p}}^{2}_{K,1} {\mathfrak{p}}^{2}_{K,y}$ ${\oathfxak{p}}_{F,1} {\mathftcj{p}}_{F, i}$ ${\manhfrak{p}}_{F_+,1}$ ${\mathfral{p}}_{F^{c},1}$ ${\mwthfrak{p}}^{2}_{K^{s},1}$ $p$ (4)\* $\langle y^2\rangke$ $\langle xg,\ y^2\randlq$ ${\mathfrak{p}}^{2}_{T,1} {\mzthfrak{p}}^{2}_{N,y}$ ${\mathfrak{p}}^{2}_{K,1}$ ${\mavhwraj{p}}_{N,1}$ ${\mathfrak{p}}_{F_+,1}$ ${\mwfhgrsk{p}}_{F^{r},1} {\mathfrak{i}}_{F^{r},y}$ ${\mstjftwk{p}}^{2}_{K^{r},1} {\mathfray{p}}^{2}_{K^{r},y}$ ${\mathsrak{p}}_{J,1}$ (a) $\langle x\ranglw$ $\lcngoe x\rangle$ ${\mcthfrak{p}}^{2}_{N,1} {\machfrak{k}}^{2}_{N,y}{\matnfrak{p}}^{2}_{N,y^2}{\mathfrak{p}}^{2}_{N,y^3}$ ${\mcthfrai{p}}_{K,1}{\mathfrak{o}}^{2}_{K,y}{\mathfrzy{p}}_{K,y^2}$ ${\mathfrak{o}}_{F,1}{\mstvfrak{p}}_{F,y}$ ${\mathsrak{p}}_{F_+,1}^2$ ${\iathfrak{p}}^{

}_{N,1} {\mathfrak{p}}^{2}_{N,y}$ ${\mathfrak{p}}^{2}_{K,1}$ ${\mathfrak{p}}_{F,1}$ ${\mathfrak{p}}_{F_+,1} {\mathfrak{p}}_{F_+,y}$ ${\mathfrak{p}}_{F^{r},1}$ (3) y^2\rangle$ $\langle y^2\rangle$ ${\mathfrak{p}}^{2}_{N,1} {\mathfrak{p}}^{2}_{N,y}$ ${\mathfrak{p}}_{F_+,1}$ ${\mathfrak{p}}^{2}_{K^{r},1}$ $p$ (4)\* y^2\rangle$ $\langle xy,\ ${\mathfrak{p}}^{2}_{N,1} {\mathfrak{p}}^{2}_{N,y}$ ${\mathfrak{p}}^{2}_{K,1}$ ${\mathfrak{p}}_{F,1}$ ${\mathfrak{p}}_{F_+,1}$ ${\mathfrak{p}}_{F^{r},1} ${\mathfrak{p}}^{2}_{K^{r},1} {\mathfrak{p}}^{2}_{K^{r},y}$ ${\mathfrak{p}}_{K,1}$ (a) $\langle x\rangle$ $\langle x\rangle$ ${\mathfrak{p}}^{2}_{N,1} {\mathfrak{p}}^{2}_{N,y}{\mathfrak{p}}^{2}_{N,y^2}{\mathfrak{p}}^{2}_{N,y^3}$ ${\mathfrak{p}}_{K,1}{\mathfrak{p}}^{2}_{K,y}{\mathfrak{p}}_{K,y^2}$ ${\mathfrak{p}}_{F,1}{\mathfrak{p}}_{F,y}$ ${\mathfrak{p}}_{F_+,1}^2$

}_{N,1} {\mathfrak{p}}^{2}_{N,y}$ ${\mathfrak{p}}^{2}_{K,1}$ ${\matHfrak{p}}_{F,1}$ ${\matHfrak{P}}_{F_+,1} {\mAthFrAk{p}}_{F_+,Y}$ ${\matHfrak{p}}_{F^{r},1}$ ${\mathfrAK{p}}^{2}_{K^{r},1}$ $P$ (3) $\langle y^2\rangle$ $\langle x,\ y^2\RanglE$ ${\mAThfrAK{p}}^{2}_{n,1} {\mathFrak{p}}^{2}_{N,y}$ ${\MAtHFRak{P}}^{2}_{K,1} {\MaThfRaK{P}}^{2}_{K,Y}$ ${\mathFraK{p}}_{F,1} {\mathFrak{p}}_{F, y}$ ${\matHfrAk{P}}_{F_+,1}$ ${\mathfrak{p}}_{F^{R},1}$ ${\MaThfrak{p}}^{2}_{K^{r},1}$ $p$ (4)\* $\LanGle y^2\rangle$ $\laNglE xy,\ y^2\raNgLe$ ${\mAThfraK{p}}^{2}_{N,1} {\MathfRak{p}}^{2}_{N,y}$ ${\MAthfraK{p}}^{2}_{K,1}$ ${\mathfrAk{P}}_{f,1}$ ${\mathfRAk{p}}_{F_+,1}$ ${\matHFRaK{p}}_{F^{r},1} {\Mathfrak{p}}_{F^{r},y}$ ${\mathfRAk{P}}^{2}_{k^{r},1} {\mathfrak{p}}^{2}_{K^{r},y}$ ${\MathfrAk{P}}_{k,1}$ (a) $\LANglE x\rAngle$ $\langlE x\RanglE$ ${\MathfraK{P}}^{2}_{N,1} {\MATHfrAK{p}}^{2}_{N,y}{\mathfrak{p}}^{2}_{n,y^2}{\mathfrak{p}}^{2}_{n,Y^3}$ ${\maThfrak{P}}_{K,1}{\MatHFrak{p}}^{2}_{K,Y}{\mathFrAK{p}}_{K,Y^2}$ ${\mathfrak{p}}_{F,1}{\MathFrak{p}}_{F,y}$ ${\maThfrak{P}}_{f_+,1}^2$ ${\mathfrAK{p}}^{

}_{N,1} {\mathfrak{p}}^{2} _{N,y}$ ${\mathfrak{p}} ^{2}_ {K , 1}$ $ {\mat hfr ak{p}}_ {F,1}$ ${\math f ra k{p}}_{F_+ ,1} {\mathfrak{ p}} _{F_+, y} $ $ {\math f rak{p} }_{F^{r}, 1} $ $ { \mathfrak{p}}^ {2}_{K ^{ r }, 1 } $ $p$ ( 3) $\ la n gle y^2\rangle $ $\lang le x,\ y^2\ran g le$ $ {\m a th fra k {p} }^{2}_{N ,1 }{\mat hfra k { p } }^{2 }_{ N,y} $ $ {\ma th frak{ p}}^{2 }_{K, 1} {\mathfrak{p}} ^{2} _{K,y}$ ${\ma t hfrak{ p}} _{F ,1} {\m athfrak { p}} _{ F , y} $ ${\ ma t h fr ak{p}}_{ F_+,1} $ ${ \mat h fr ak{p}}_{ F^{r}, 1 }$ ${\mathf rak{p } }^{2}_{K^{r},1 } $ $p$ (4)\* $\la ng le y^2 \rang le$ $\la ngle xy,\y ^ 2 \rangle$ ${\mathf r ak{p}}^{ 2}_{N ,1} {\ma thfrak{p} } ^ {2}_{N,y }$ ${ \ma thf rak {p} }^ {2}_{K,1} $ ${ \m ath frak{ p }}_{F, 1}$ $ {\ma th fr ak{p }}_ {F _+,1} $ ${\mathf rak { p}}_ {F ^{ r},1} { \mathfrak{p}} _{ F^{r},y}$ $ {\mathfr ak{p}}^{2}_{K^{r},1} {\ m athfrak {p} }^{2} _{K^ {r},y}$ ${ \math fr ak{ p } }_{K, 1 } $ (a) $\l angle x \rangle$ $ \la ngle x\rangle $ $ {\ mat h fr a k{p }} ^ {2} _ { N,1} {\mathfrak {p}}^{2}_{ N, y }{ \mathfrak{ p }}^ {2 }_{N,y^ 2}{\mat hfrak { p}}^{2} _{N,y^3}$ ${\mat hf rak{ p } }_{ K,1}{\math frak{p}} ^{2}_{K,y } {\mat h fr ak{p} }_{ K,y^2} $ ${\ma thfrak { p}} _{F,1 }{\mat hf rak{p} }_{F, y} $ ${\mathfrak{p}}_ {F_+,1 }^2$ ${\ma thfr ak{p}}^{

}_{N,1} {\mathfrak{p}}^{2}_{N,y}$_ _ _ __ __ _ _ _ __ __ ___ ${\mathfrak{p}}^{2}_{K,1}$_ _ _ __ _ _ _${\mathfrak{p}}_{F,1}$ ___ _ ${\mathfrak{p}}_{F_+,1}_{\mathfrak{p}}_{F_+,y}$ _ _ ${\mathfrak{p}}_{F^{r},1}$ _ _ _ _ _ _ _ _ _ __ ${\mathfrak{p}}^{2}_{K^{r},1}$ ___ _ __ __ __ _ __ _ _ _ $p$_ _ _ _ _ __(3) __ _ $\langle y^2\rangle$ _ $\langle_x,\ y^2\rangle$ _ __ _ _ ___ ${\mathfrak{p}}^{2}_{N,1} {\mathfrak{p}}^{2}_{N,y}$ _ _ _ __ __ ___ ${\mathfrak{p}}^{2}_{K,1} {\mathfrak{p}}^{2}_{K,y}$ _ _ ${\mathfrak{p}}_{F,1} {\mathfrak{p}}_{F, y}$ _ _ ___ ${\mathfrak{p}}_{F_+,1}$ _ _ _ _ ${\mathfrak{p}}_{F^{r},1}$ _ _ _ _ __ ${\mathfrak{p}}^{2}_{K^{r},1}$____ _ _ _ _ _ _ __ _ _$p$ __ _ _ _ (4)\* __ $\langle_y^2\rangle$ _ _$\langle xy,\_y^2\rangle$ _ _ _ _ __ ____${\mathfrak{p}}^{2}_{N,1} {\mathfrak{p}}^{2}_{N,y}$ _ _ _ __ _ _ _ _ ${\mathfrak{p}}^{2}_{K,1}$ _ _ _ __ _ _ __ ${\mathfrak{p}}_{F,1}$ ___ __ _____ ${\mathfrak{p}}_{F_+,1}$ _ __ _${\mathfrak{p}}_{F^{r},1} {\mathfrak{p}}_{F^{r},y}$__ _ _ ${\mathfrak{p}}^{2}_{K^{r},1} {\mathfrak{p}}^{2}_{K^{r},y}$ __ ____ _ ${\mathfrak{p}}_{K,1}$ __ _ (a) __ _ $\langle x\rangle$ _ $\langle_x\rangle$ _ ${\mathfrak{p}}^{2}_{N,1} {\mathfrak{p}}^{2}_{N,y}{\mathfrak{p}}^{2}_{N,y^2}{\mathfrak{p}}^{2}_{N,y^3}$ ${\mathfrak{p}}_{K,1}{\mathfrak{p}}^{2}_{K,y}{\mathfrak{p}}_{K,y^2}$ ${\mathfrak{p}}_{F,1}{\mathfrak{p}}_{F,y}$ _ ${\mathfrak{p}}_{F_+,1}^2$_ _ __ _ _ ${\mathfrak{p}}^{

, 11, 13, 19, 25, 31, 37, 41, 43, 61, 67, 71, 73, 79, 121, 151,$ and $211.$ From [@PPR2], we know that the above is a complete list of genuine exceptions in Case 2, and $C_{2,0}=211.$ Analogously, we propose the following conjecture. We have $C_{1,1}=331$ and the list of prime powers not in $Q_{1,1}$ is shown in Case 1, as above. We complete this section with some remarks on the set $S$ of exceptional quadratic polynomials, whose members comprise quadratics of the form $f(x)=a(x+b)^2$, where $ab\neq 0$. In the context of Lemma \[CharBound\] their irreducible part is of degree 1 and hence the condition of Theorem \[main\] applies with $n=1.$ Here, if $(\alpha, f(\alpha))$ is primitive, then necessarily $a$ is a non square, in which case it suffices that $\alpha$ is primitive and $a(\alpha+b)^2$ is $L$-free, where $L$ is the odd part of $q-1$. Denote by $R_{1^2,0}$ the subset of $S$ for which $a$ is a non-square. By methods of this section this will lead to a better (smaller) lower bound for $C_{1^2,0}$ than the one shown in Theorem 4.1 for $C_{1,1}$. Case n=3, 4 and 5 ================= In this section, we demonstrate how to get at least one value $C_n$ for each $n\in \mathbb{N}$ and $n\geq 2$. Further, we provide some calculated values to reduce the bound $C_n$ for $n=3, 4$ and $5$. As described above, Theorem \[main\] and Lemma \[Wbound\] together imply that if $n (37.469)^2 q^{\frac{1}{3}}<q^{\frac{1}{2}}$ then $q\in Q_n$ i.e. $q>n^6(37.469)^{12}$ implies $q\in Q_n$. Hence, for each $n\in \mathbb{N}, n\geq 2$, one value of $

, 11, 13, 19, 25, 31, 37, 41, 43, 61, 67, 71, 73, 79, 121, 151,$ and $ 211.$ From [ @PPR2 ], we know that the above is a complete list of genuine exception in Case 2, and $ C_{2,0}=211.$ Analogously, we nominate the following conjecture. We have $ C_{1,1}=331 $ and the tilt of prime powers not in $ Q_{1,1}$ is shown in Case 1, equally above. We dispatch this section with some remarks on the set $ S$ of exceptional quadratic polynomials, whose member comprise quadratics of the class $ f(x)=a(x+b)^2 $, where $ ab\neq 0$. In the context of Lemma \[CharBound\ ] their irreducible share is of degree 1 and therefore the condition of Theorem \[main\ ] practice with $ n=1.$ Here, if $ (\alpha, f(\alpha))$ is crude, then necessarily $ a$ is a non square, in which case it suffice that $ \alpha$ is primitive and $ a(\alpha+b)^2 $ is $ L$-free, where $ L$ is the odd contribution of $ q-1$. Denote by $ R_{1 ^ 2,0}$ the subset of $ S$ for which $ a$ is a non - square. By method of this section this will lead to a better (smaller) lower bound for $ C_{1 ^ 2,0}$ than the one prove in Theorem 4.1 for $ C_{1,1}$. Case n=3, 4 and 5 = = = = = = = = = = = = = = = = = In this section, we demonstrate how to get at least one value $ C_n$ for each $ n\in \mathbb{N}$ and $ n\geq 2$. Further, we provide some forecast value to reduce the bound $ C_n$ for $ n=3, 4 $ and $ 5$. As described above, Theorem \[main\ ] and Lemma \[Wbound\ ] together imply that if $ n (37.469)^2 q^{\frac{1}{3}}<q^{\frac{1}{2}}$ then $ q\in Q_n$ i.e. $ q > n^6(37.469)^{12}$ imply $ q\in Q_n$. therefore, for each $ n\in \mathbb{N }, n\geq 2 $, one value of $

, 11, 13, 19, 25, 31, 37, 41, 43, 61, 67, 71, 73, 79, 121, 151,$ and $211.$ From [@PPV2], we know that tkw abovx is a domplete list of genuine exceptions mn Cqse 2, qnd $C_{2,0}=211.$ Analogously, we pfopose thv followibg cibjecture. We have $C_{1,1}=331$ and ths lisc if prime powerx not in $Q_{1,1}$ is shown in Cdsd 1, as above. We complete this section wyth somr gemarks on the set $F$ of vxgeptional quadratic polynomials, shose mtmbers comprise qusdratics of the form $f(x)=a(x+b)^2$, whege $ab\neq 0$. In the clntext of Lgjma \[XharBound\] thdir irredubnble part ia of degree 1 and hence the condktion of Theoren \[naij\] applies wivh $n=1.$ Hvre, if $(\alpha, n(\slpha))$ hs primotive, then neccssarmly $q$ is a non square, in xhich case it sufficgs that $\al[hc$ is primitive and $a(\aopya+b)^2$ iv $L$-fsee, dyerd $L$ ix fhe odf pert of $q-1$. Dehote by $R_{1^2,0}$ tye subset of $S$ for ehybn $a$ is a non-aquare. Br methods of this section this will leag tk a better (smaller) lowee bound for $C_{1^2,0}$ than thg one showg in Theorem 4.1 for $C_{1,1}$. Case n=3, 4 and 5 ================= In this section, we demoisgraue how gi het at least one value $C_n$ for each $n\in \mathbb{G}$ amd $n\geq 2$. Further, we provide slmr calculated vxlues ck deduce the bound $C_j$ for $n=3, 4$ and $5$. As descrybed above, Theorem \[main\] and Lemna \[Wbound\] toyetyer imply that if $u (37.469)^2 q^{\frac{1}{3}}<q^{\frae{1}{2}}$ then $q\in A_n$ i.e. $q>n^6(37.469)^{12}$ implies $q\in Q_n$. Hende, for each $n\in \mathgc{N}, n\geq 2$, one valje pf $

, 11, 13, 19, 25, 31, 37, 61, 71, 73, 121, 151,$ and that above is a list of genuine in Case 2, and $C_{2,0}=211.$ Analogously, propose the following conjecture. We have $C_{1,1}=331$ and the list of prime powers in $Q_{1,1}$ is shown in Case 1, as above. We complete this section some on set of exceptional quadratic polynomials, whose members comprise quadratics of the form $f(x)=a(x+b)^2$, where $ab\neq 0$. In context of Lemma \[CharBound\] their irreducible part is degree 1 and hence condition of Theorem \[main\] applies $n=1.$ if $(\alpha, is then $a$ is a square, in which case it suffices that $\alpha$ is primitive and $a(\alpha+b)^2$ is $L$-free, where $L$ is odd part Denote by the of for which $a$ non-square. By methods of this section to a better (smaller) lower bound for $C_{1^2,0}$ the one in Theorem 4.1 for $C_{1,1}$. Case 4 and 5 ================= In this section, we how to get at least one value $C_n$ for each $n\in \mathbb{N}$ and $n\geq 2$. provide some calculated values reduce the bound for 4$ $5$. described above, \[main\] and Lemma \[Wbound\] together imply that if $n (37.469)^2 q^{\frac{1}{3}}<q^{\frac{1}{2}}$ $q\in Q_n$ i.e. $q>n^6(37.469)^{12}$ implies $q\in Q_n$. Hence, for each n\geq one value of

, 11, 13, 19, 25, 31, 37, 41, 43, 61, 67, 71, 73, 79, 121, 151,$ and $211.$ From [@PPR2], we know that the abOve is a compLete lIst Of gEnUine ExcePtions in Case 2, anD $c_{2,0}=211.$ AnaLogously, we propose the foLlowiNg COnjeCTuRe. We hAve $C_{1,1}=331$ and THe LISt oF pRiMe pOwERs Not in $q_{1,1}$ is Shown in case 1, as abovE. We CoMplete this seCTiOn with some RemArks on the set $s$ of ExceptIoNal QUadraTic PolynOmials, WHose meMbers compRiSE quadrATics of tHE FoRm $f(x)=A(x+b)^2$, where $ab\neq 0$. In thE CoNText of Lemma \[ChaRBound\] ThEIr IRRedUciBle part is oF dEgree 1 ANd hence THe CONDitIOn of Theorem \[maIn\] applies wiTH $n=1.$ HEre, if $(\aLpHa, f(\ALpha))$ is PrimiTiVE, thEn necessariLy $a$ iS a non squaRe, in whICh case iT SufficeS that $\aLphA$ is PrimITiVe And $A(\aLPha+B)^2$ Is $l$-frEE, whEre $L$ is thE oDd Part oF $q-1$. DeNOTE By $R_{1^2,0}$ tHe sUbseT of $S$ fOr which $a$ is a noN-sqUare. bY meThods Of thiS secTiOn thiS will lEad to A bEtter (smaller) lowEr boUnd for $C_{1^2,0}$ thAn tHe One ShOwn in tHeorem 4.1 For $c_{1,1}$. CaSe n=3, 4 and 5 ================= IN this seCTioN, wE DEMoNstrate how to get at lEaST OnE value $C_n$ For eacH $N\iN \mAThbb{N}$ and $N\gEq 2$. FUrthER, We proVide SOmE calculaTed valUEs To Reduce tHe Bound $C_N$ fOr $n=3, 4$ And $5$. as desCRibeD above, theorem \[mAin\] anD lemma \[Wbound\] togETher imply that IF $n (37.469)^2 Q^{\FRaC{1}{3}}<Q^{\fraC{1}{2}}$ thEn $q\in Q_n$ i.e. $q>n^6(37.469)^{12}$ ImplIEs $q\iN Q_n$. HENcE, foR Each $n\In \matHbB{n}, n\GEq 2$, one value of $

, 11, 13, 19, 25, 31, 37,41, 43, 61 , 67, 71 , 7 3, 79, 121 , 151,$ and $2 1 1.$ From [@PPR2], we know that t h e ab o ve is a comple t el i stof g enu in e e xcept ion s in Ca se 2, and$C_ {2 ,0}=211.$ An a lo gously, we pr opose the fo llo wing c on jec t ure. We have $C_{1 , 1}=331 $ and the l i st ofp rime po w e rs not in $Q_{1,1}$ iss ho w n in Case 1, a s abov e. W e com ple te this se ct ion w i th some re m a r kso n the set $S$ of excepti o nal quadr at icp olynom ials, w h ose members co mpri se quadra tics o f the fo r m $f(x) =a(x+b )^2 $,wher e $ ab \ne q0 $.I nthe con text ofLe mm a \[C harB o u n d \] t hei r ir reduc ible part isofdegr e e 1 andhence the c ondit ion of Theo re m \[main\] appl ieswith $n=1 .$He re, i f $(\ a lpha,f(\ alp ha))$ i s primi t ive ,t h e nnecessarily $a$ is a n on square, in wh i ch c a se it su ff ice s th a t $\al pha$ is primiti ve and $a (\ alpha+b )^ 2$ is$L $-f ree , whe r e $L $ is t he odd p art o f $q-1$. Denote by $R_{1^2,0} $ t h e s u bset of $S$ for wh ich$ a$ i s an on -sq u are.By me th o ds of this section thi swill l ead t o a better (s maller) lo w e r bound f or $ C _{ 1 ^2,0}$ than th e one shown inT heorem 4 .1 fo r $C_{1, 1}$. Cas e n=3, 4 a nd5 = === === = = == ====== In th i s sec ti on, wedem onstrat e h owtoget a t least o ne value $ C_ n$ f oreach$ n\in \ma th bb{ N} $ a nd $n \ geq 2$ . Fur ther ,we pro vide so m ec a lcul at ed val ues t o red ucet hebound $ C_n$ for$n= 3 , 4$ a nd $5$. As describedab ove, Theor em \[ main\] a nd Lemma \[Wbound\] together im p ly that if $n ( 37.4 69)^2 q^{ \fr ac{1}{ 3}} < q^{\fr ac{1}{ 2}}$th en$ q \in Q _ n $i.e .$q>n^6(37. 4 6 9)^ {12}$ i mpli es $q\i n Q_n$. Hence, for eac h $n\in \math bb{ N},n \ ge q 2 $ ,o neva l ueo f $

, 11,_13, 19,_25, 31, 37, 41,_43, 61,_67,_71, 73,_79,_121, 151,$ and_$211.$ From [@PPR2], we_know that the above_is a complete_list_of genuine exceptions in Case 2, and $C_{2,0}=211.$ Analogously, we propose the following conjecture. We_have_$C_{1,1}=331$ and_the_list_of prime powers not in_$Q_{1,1}$ is shown in Case_1, as_above. We complete this section with some remarks on_the_set $S$ of_exceptional quadratic polynomials, whose members comprise quadratics of the_form $f(x)=a(x+b)^2$, where $ab\neq 0$. In_the context of_Lemma_\[CharBound\]_their irreducible part is_of degree 1 and hence the_condition of Theorem \[main\] applies with_$n=1.$ Here, if $(\alpha, f(\alpha))$ is primitive,_then necessarily $a$ is a non_square, in which case it_suffices that_$\alpha$ is primitive and $a(\alpha+b)^2$_is $L$-free, where_$L$ is_the odd part_of $q-1$. Denote by $R_{1^2,0}$ the_subset of $S$_for which $a$ is a non-square._By_methods of this_section_this_will lead_to a better_(smaller)_lower bound_for_$C_{1^2,0}$ than the one shown in_Theorem_4.1 for $C_{1,1}$. Case n=3, 4 and 5 ================= In_this section, we demonstrate_how_to get at least_one value $C_n$ for each_$n\in \mathbb{N}$ and $n\geq 2$. Further,_we provide_some calculated_values to reduce the bound $C_n$ for $n=3, 4$ and $5$. As_described above, Theorem \[main\] and Lemma_\[Wbound\] together imply that_if $n_(37.469)^2_q^{\frac{1}{3}}<q^{\frac{1}{2}}$ then $q\in_Q_n$_i.e. $q>n^6(37.469)^{12}$_implies $q\in Q_n$. Hence, for each $n\in_\mathbb{N}, n\geq_2$, one value of $

elements much larger than $\langle N | \bar{s} s | N\rangle $; for $\langle N | \bar{u} u + \bar{d} d | N\rangle $, this is certainly the case. This can potentially offset the suppression of $m_{res} $. One may speculate that mixing with the operator $\bar{u} u + \bar{d} d$ constitutes the strongest effect, since the light quark fields are special in that they form the valence component of the nucleon, which has no counterpart in the vacuum expectation value that is subtracted off throughout, cf. (\[matel1\]). On the other hand, the presence of the valence quarks also strongly distorts the gluon field in the nucleon. No estimate of the gluonic admixture to $\bar{s} s$ is available, but the light quark admixture under renormalization will be argued below to constitute an effect of the order of 1%. In view of the statistical uncertainties associated with the present determination of $f_{T_s } $, which amount to about 20%, a putative gluonic mixing effect would have to be an order of magnitude larger than the light quark mixing effect in order to appreciably influence the final result for $f_{T_s } $. This seems a rather implausible scenario. For this reason, the strong constraint on mixing with light quarks derived below will be taken as indication that violations of (\[scalarinv\]) are negligible at the present level of statistical accuracy of $f_{T_s } $. Concentrating thus on the effect stemming from the mixing with the operator $\bar{u} u + \bar{d} d$, an estimate of the possible residual violation of (\[scalarinv\]) can be obtained from the Feynman-Hellmann theorem as follows [@jlqcd2]. The residual breaking of chiral symmetry can be parametrized to leading order via the additive mass renormalization $m_{res,q} $, which in general depends on the flavor $q$ for which one is considering the domain wall Dirac operator, $$m_q^{\mbox{\scriptsize renorm} } = Z_m (m_q^{\mbox{\scriptsize bare} } + m_{res,q} ) \.$$ Using (\[fhtheorem\]), one has $$\begin{aligned} \left. m_s \langle N |\bar{s} s| N\rangle \right|_{\mbox{\scriptsize bare} } &=& m_s^{\mbox{\scriptsize bare

elements much larger than $ \langle N | \bar{s } s | N\rangle $; for $ \langle N | \bar{u } u + \bar{d } d | N\rangle $, this is certainly the subject. This can potentially cancel the suppression of $ m_{res } $. One may speculate that mix with the hustler $ \bar{u } u + \bar{d } d$ constitutes the strongest effect, since the idle quark fields are special in that they shape the valence component of the nucleon, which has no counterpart in the void expectation value that is subtracted off throughout, cf.   (\[matel1\ ]). On the other hand, the bearing of the valence quarks also powerfully distorts the gluon discipline in the nucleon. No estimate of the gluonic admixture to $ \bar{s } s$ is available, but the light quark admixture under renormalization will be argue below to constitute an effect of the ordering of 1% . In view of the statistical uncertainties associated with the present determination of $ f_{T_s } $, which amount to approximately 20% , a putative gluonic mixing effect would have to be an order of magnitude larger than the abstemious quark cheese mixing effect in order to appreciably influence the final result for $ f_{T_s } $. This seems a preferably farfetched scenario. For this rationality, the strong restraint on mix with faint quarks derived below will be taken as indication that violations of (\[scalarinv\ ]) are negligible at the present level of statistical accuracy of $ f_{T_s } $. boil down thus on the effect stemming from the mixing with the operator $ \bar{u } u + \bar{d } d$, an estimate of the potential residual violation of (\[scalarinv\ ]) can be obtained from the Feynman - Hellmann theorem as follows [ @jlqcd2 ]. The residual breaking of chiral symmetry can be parametrized to leading order via the linear aggregate renormalization $ m_{res, q } $, which in cosmopolitan depends on the flavor $ q$ for which one is considering the domain wall Dirac operator, $ $ m_q^{\mbox{\scriptsize renorm } } = Z_m (m_q^{\mbox{\scriptsize bare } } + m_{res, q }) \.$$ Using (\[fhtheorem\ ]), one has $ $ \begin{aligned } \left. m_s \langle N |\bar{s } s| N\rangle \right|_{\mbox{\scriptsize unsheathed } } & = & m_s^{\mbox{\scriptsize bare

elfments much larger than $\uangle N | \bar{s} s | N\ranjle $; fod $\langle N | \bar{u} u + \bar{d} d | N\rangle $, tyis iw certainly the case. Tfis can pltentialoy ohfset the suppression of $m_{res} $. Lne key speculate thst mixing fith the operadof $\yar{u} u + \bar{d} d$ constitutes the strondest efgeft, since the lyght zuari fields are special in that they rorm tht valence componeny of the nucleon, which has no founterpart in the vacuum expgdtaeuon value thxt is subtgccted off tgroughout, cf. (\[matel1\]). On the other fand, che presencg kf jhe valence wuarkf also stronnky disdorts tne gluon field in thw nucleon. No estimate of the gluonic adiixture tm $\yar{s} s$ is available, byt the nighd quxek xdmjxvurs undeg rxnormalizatjon will be argued below to comsepyute an effedt of ehq order of 1%. In view of the statistical lncedtainties associated wirh the present determlnation os $f_{T_s } $, which amount to about 20%, a putative gluonic mixiig efycgt wuylf have to be an order of magnitude larger thag tne light quark mlxing effect in orceg yj appreciably influzhcs the final result for $f_{T_f } $. Tyis seems a rsther implausible scenario. Dor this reafin, the strong conscraint on miring woth loght quarks derived beluw wjll be takej as indidxtion that violagioks mf (\[scalarinv\]) are negligiblq at the 'reseut level of xtatiseical accugacy of $f_{T_s } $. Concentratijg thos on dhe effect stemming from the mixing with vie operator $\bsr{g} u + \bar{d} d$, an extimate of thq possible resndual vijlatiun of (\[scalzrinv\]) cen be obtainqd from the Fainman-Hellmann theorem as dolliws [@jlqza2]. The residual breaking of chiral symmetry can be peramgtdized to leadiny irder via the acdigivq laxs senormalizathon $o_{rer,a} $, whkch in geneval deprnds on the flavor $q$ for which one is consodcring the domain rall Dirac oprrator, $$m_q^{\mbox{\scripusize cenorm} } = Z_m (k_q^{\mfox{\scriptsize bare} } + m_{res,q} ) \.$$ Uaing (\[fhthforcm\]), one has $$\bedin{aoigned} \left. m_x \langle N |\bar{s} s| N\rangle \right|_{\mbox{\scri'tsize bare} } &=& m_s^{\mbox{\scriptsize bare

elements much larger than $\langle N | | $; for N | \bar{u} N\rangle this is certainly case. This can offset the suppression of $m_{res} $. may speculate that mixing with the operator $\bar{u} u + \bar{d} d$ constitutes strongest effect, since the light quark fields are special in that they form valence of nucleon, has no counterpart in the vacuum expectation value that is subtracted off throughout, cf. (\[matel1\]). On other hand, the presence of the valence quarks strongly distorts the gluon in the nucleon. No estimate the admixture to s$ available, the light quark under renormalization will be argued below to constitute an effect of the order of 1%. In view the statistical with the determination $f_{T_s $, which amount 20%, a putative gluonic mixing effect be an order of magnitude larger than the quark mixing in order to appreciably influence the result for $f_{T_s } $. This seems a implausible scenario. For this reason, the strong constraint on mixing with light quarks derived below taken as indication that of (\[scalarinv\]) are at present of accuracy of } $. Concentrating thus on the effect stemming from the mixing the operator $\bar{u} u + \bar{d} d$, an estimate of residual of (\[scalarinv\]) can obtained from the Feynman-Hellmann as [@jlqcd2]. The residual breaking symmetry be order the mass renormalization $m_{res,q} $, in general depends on the $q$ for which one Dirac operator, $$m_q^{\mbox{\scriptsize renorm} } = Z_m (m_q^{\mbox{\scriptsize } + m_{res,q} ) \.$$ Using (\[fhtheorem\]), has $$\begin{aligned} \left. m_s \langle N |\bar{s} s| N\rangle \right|_{\mbox{\scriptsize bare} } m_s^{\mbox{\scriptsize bare

elements much larger than $\lanGle N | \bar{s} s | N\RanglE $; foR $\laNgLe N | \bAr{u} u + \Bar{d} d | N\rangle $, thIS is cErtainly the case. This can PotenTiALly oFFsEt the SuppresSIoN OF $m_{rEs} $. onE maY sPEcUlate ThaT mixing With the opeRatOr $\Bar{u} u + \bar{d} d$ coNStItutes the sTroNgest effect, sIncE the liGhT quARk fieLds Are spEcial iN That thEy form the VaLEnce coMPonent oF THe NuclEon, which has no counTErPArt in the vacuum ExpectAtIOn VALue ThaT is subtracTeD off tHRoughouT, Cf. (\[MATEl1\]). ON The other hand, tHe presence oF The ValencE qUarKS also sTrongLy DIstOrts the gluoN fieLd in the nuCleon. NO EstimatE Of the glUonic aDmiXtuRe to $\BAr{S} s$ Is aVaILabLE, bUt tHE liGht quark AdMiXture UndeR RENOrmaLizAtioN will Be argued below To cOnstITutE an efFect oF the OrDer of 1%. in view Of the StAtistical uncertAintIes associAteD wIth ThE presENt deteRmiNatIon of $f_{T_S } $, which aMOunT tO ABOuT 20%, a putative gluonic mIxING eFfect wouLd have TO bE aN Order of mAgNitUde lARGer thAn thE LiGht quark Mixing EFfEcT in ordeR tO appreCiAblY inFluenCE the Final rEsult for $F_{T_s } $. ThIS seems a rather iMPlausible scenARiO. fOr THis rEasOn, the strong ConsTRainT on mIXiNg wITh ligHt quaRkS DeRIved below will be takeN aS indicAtion That violationS of (\[scalariNV\]) ARe negligIble AT tHE present level oF statIstical accURacy of $f_{T_S } $. ConcEntratinG thus on thE EFfect steMmiNg fRom The MIXiNg with the operATOr $\baR{u} U + \bar{d} d$, aN esTimate oF thE poSsiBle ReSidual vioLation of (\[ScAlArInV\]) caN be obTAined froM tHe FEyNmaN-HellMAnn theOrem aS folLoWs [@JLqcD2]. The resIDuAL BreaKiNg Of chIraL sYmmetRy caN Be pArametrIzed to leaDinG OrdeR vIa The addiTive mass renorMaLization $m_{rEs,Q} $, whIch in gENEral depeNds on the flavor $q$ for which ONe is conSidEring The dOmain wall dirAc operAtoR, $$M_q^{\mbox{\ScriptSize rEnOrm} } = z_M (M_q^{\mboX{\SCrIptSiZe bare} } + m_{res,Q} ) \.$$ uSinG (\[fhthEoRem\]), oNe has $$\beGin{aligned} \left. m_s \laNGle n |\bar{s} s| N\rangle \RigHt|_{\mbOX{\ScRipTSiZE baRe} } &=& M_S^{\mbOX{\Scriptsize bare

elements much larger than $\langleN | \ bar {s} s | N \ran gle $; for $\l a ngle N | \bar{u} u + \bar{ d} d|N \ran g le $, t his isc er t a inl yth e c as e .Thiscan potent ially offs etth e suppressio n o f $m_{res} $. One may spe cul ate th at mi x ing w ith theoperat o r $\ba r{u} u +\b a r{d} d $ consti t u te s th e strongest effec t ,s ince the light quark f i el d s ar e s pecial inth at th e y formt he v a len c e component o f the nucle o n,whichha s n o count erpar ti n t he vacuum e xpec tation va lue th a t is su b tracted off t hro ugh out, cf .(\[ ma t el1 \ ]) . O n th e otherha nd , the pre s e n c e of th e va lence quarks alsostr ongl y di stort s the glu on fiel d in t he nu cl eon. No estimat e of the gluo nic a dmi xt ure t o $\bar {s} s$ is ava ilable, but t h e li ght quark admixtur eu n de r renorm alizat i on w i ll be ar gu edbelo w to co nsti t ut e an eff ect of th eorder o f1%. In v iew of thes tati stical uncerta intie s associated wi t h the present de t e rm i nati onof $f_{T_s} $, whic h am o un t t o abou t 20% ,a p u tative gluonic mixi ng effec t wou ld have to be an ordero f magnitud e la r ge r than the ligh t qua rk mixinge ffect in orde r to app reciablyi n fluencethe fi nal re s u lt for $f_{T_s} $. T hi s seems aratherimp lau sib lesc enario. F or thisre as on ,the stro n g constr ai nton mi xingw ith li ght q uark sde r ive d below wi l l beta ke n as in di catio n th a t v iolatio ns of (\[ sca l arin v\ ]) are ne gligible at t he present l ev elof sta t i stical a ccuracy of $f_{T_s } $. Concen tra tingthus on the e ffe ct ste mmi n g from the m ixing w ith t he op e r at or$\ bar{u} u + \ bar {d} d $, anestimat e of the possibler esi dual violatio n o f (\ [ s ca lar i nv \ ])ca n be o btained from th e Feynman- He l lm ann theore m as f ollows[@jlqcd 2]. T h e resid ual break ing of ch ir al s y m met ry can beparametr ized to l e ading or der v iathe ad di tiv e mas s reno r mal izati on $m_ {r es,q}$, wh ic h in gen eral depends on the fla vor $q $ for wh ich one i s c o nsi dering th e do main wallDir acopera tor , $$m_ q^{\ m bo x{\ s cript size renorm} } =Z_m ( m_ q^{\mbox{\s c r i pts ize b are } } + m _{re s,q} ) \.$$ Using (\[fhtheorem\] ), o n e ha s $ $ \beg in {aligned} \lef t.m_ s \langleN|\bar{s} s| N\rangl e\ right |_{\mb ox{\sc riptsiz e ba r e} } & =& m _s^ {\mbox{\s cri pt s ize bar e

elements_much larger_than $\langle N |_\bar{s} s_|_N\rangle $;_for_$\langle N |_\bar{u} u +_\bar{d} d | N\rangle_$, this is_certainly_the case. This can potentially offset the suppression of $m_{res} $. One may speculate_that_mixing with_the_operator_$\bar{u} u + \bar{d} d$_constitutes the strongest effect, since_the light_quark fields are special in that they form_the_valence component of_the nucleon, which has no counterpart in the vacuum_expectation value that is subtracted off_throughout, cf. (\[matel1\]). On_the_other_hand, the presence of_the valence quarks also strongly distorts_the gluon field in the nucleon._No estimate of the gluonic admixture to_$\bar{s} s$ is available, but the_light quark admixture under renormalization_will be_argued below to constitute an_effect of the_order of_1%. In view_of the statistical uncertainties associated with_the present determination_of $f_{T_s } $, which amount_to_about 20%, a_putative_gluonic_mixing effect_would have to_be_an order_of_magnitude larger than the light quark_mixing_effect in order to appreciably influence the_final result for $f_{T_s_}_$. This seems a_rather implausible scenario. For this_reason, the strong constraint on mixing_with light_quarks derived_below will be taken as indication that violations of (\[scalarinv\]) are_negligible at the present level of_statistical accuracy of $f_{T_s_} $. Concentrating_thus_on the effect_stemming_from the_mixing with the operator $\bar{u} u +_\bar{d} d$,_an estimate of the possible residual_violation of (\[scalarinv\]) can_be_obtained from the Feynman-Hellmann theorem as_follows [@jlqcd2]. The residual breaking of_chiral symmetry can be parametrized_to_leading_order via the additive mass_renormalization $m_{res,q} $, which in general_depends on the_flavor $q$ for which one is considering_the_domain wall Dirac operator, $$m_q^{\mbox{\scriptsize renorm}_}_= Z_m (m_q^{\mbox{\scriptsize bare} } + m_{res,q}_)_\.$$_Using (\[fhtheorem\]), one has $$\begin{aligned} \left._m_s \langle N |\bar{s} s| N\rangle_\right|_{\mbox{\scriptsize bare} } &=& m_s^{\mbox{\scriptsize bare

$-type correction can be found efficiently for the surface code in terms of the graph $G_{\text{2D}}$ shown in Fig. \[fig:Surfaced5Graph\]. The graph $G_{\text{2D}}$ has a bulk node (black circle) for each $Z$ stabilizer generator, and a bulk edge (black) for each data qubit. A bulk edge coming from a bulk node corresponds to the edge’s data qubit being in the support of the node’s stabilizer. The graph also contains boundary nodes (white boxes) and boundary edges (blue), which do not correspond to stabilizers or data qubits. Each bulk and boundary edge is assigned weight one and zero respectively. The minimum weight decoder is then implemented as follows. After the error $E$ is applied, the nodes corresponding to unsatisfied stabilizers are highlighted. If an odd number of stabilizers was unsatisfied, one of the boundary nodes is also highlighted. Highlighted nodes are then efficiently paired together by the minimum weight connections in the graph, by Edmonds’ algorithm [@Edmonds65; @Kolmogorov09]. The correction $C$ is applied to the edges in the connection. Note that any single $\mathcal{O}(p)$ fault in this noise model corresponds to a weight one edge on the graph. [0.3]{}![Circuits for measuring (a) $Z$-type, and (b) $X$-type generators. Identity gates (black rectangles) are inserted in the $Z$-type stabilizer measurement circuits to ensure that all measurements are synchronized. Note that unlike in [@FMMC12], to be consistent with the other schemes in this paper, we assume that we can prepare and measure in both the $X$ and $Z$ basis.[]{data-label="fig:Surface17Circuits"}](ZstabMeasSurfaceCode.png "fig:"){width="\textwidth"} [0.3]{}![Circuits for measuring (a) $Z$-type, and (b) $X$-type generators. Identity gates (black rectangles) are inserted in the $Z$-type stabilizer measurement circuits to ensure that all measurements are synchronized. Note that unlike in [@FMMC12], to be consistent with the other schemes in this paper, we assume that we can prepare and measure in both the $X$ and $Z$ basis.[]{data-label="fig:Surface17Circuits"}](XstabMeasSurfaceCode.png "fig

$ -type correction can be found efficiently for the surface code in terms of the graph $ G_{\text{2D}}$ testify in Fig.   \[fig: Surfaced5Graph\ ]. The graph $ G_{\text{2D}}$ have a bulk node (black circle) for each $ Z$ stabilizer generator, and a bulk boundary (black) for each data qubit. A bulk edge come from a bulk node corresponds to the edge ’s datum qubit being in the documentation of the node ’s stabilizer. The graph besides contains boundary nodes (white boxes) and boundary edge (blue), which do not represent to stabilizers or data qubits. Each majority and boundary edge is assigned weight one and zero respectively. The minimal weight decoder is then implemented as follows. After the error $ E$ is applied, the node corresponding to unsatisfied stabilizers are highlighted. If an odd numeral of stabilizers was unsatisfied, one of the boundary nodes is also highlighted. Highlighted nodes are then efficiently paired together by the minimum weight connections in the graph, by Edmonds ’ algorithm [ @Edmonds65; @Kolmogorov09 ]. The correction $ C$ is applied to the edges in the connection. Note that any single $ \mathcal{O}(p)$ fault in this randomness exemplar corresponds to a weight unit one boundary on the graph. [ 0.3]{}![Circuits for measure (a) $ Z$-type, and (b) $ X$-type generators. Identity gate (black rectangles) are inserted in the $ Z$-type stabilizer measurement circuits to guarantee that all measurements are synchronized. Note that unlike in [ @FMMC12 ], to be consistent with the other schemes in this newspaper, we assume that we can prepare and measure in both the $ X$ and $ Z$ basis.[]{data - label="fig: Surface17Circuits"}](ZstabMeasSurfaceCode.png " fig:"){width="\textwidth " } [ 0.3]{}![Circuits for measuring (vitamin a) $ Z$-type, and (b) $ X$-type generator. Identity gates (black rectangles) are inserted in the $ Z$-type stabilizer measurement circuits to guarantee that all measurements are synchronized. Note that unlike in [ @FMMC12 ], to be reproducible with the other schemes in this paper, we assume that we can train and measure in both the $ X$ and $ Z$ basis.[]{data - label="fig: Surface17Circuits"}](XstabMeasSurfaceCode.png " libyan islamic fighting group

$-typf correction can be founa efficiently for the vurfacs code iv terms of the graph $G_{\text{2D}}$ dhiwn ib Fig. \[fig:Surfaced5Graph\]. Ghe graph $G_{\text{2D}}$ yas e bulk node (black circle) for ezgh $Z$ vvabilizer generstor, and a bulk edge (blawk) flr each data qubit. A bulk edge comigg from a bulk node cortespomqs tk the edge’s data qubit being in ths suppogt of the node’s syabilizer. The graph also clntalns boundary nodes (white boxew) anq boundary edees (blue), which do not dorrespond to stabilizers or daga quyits. Each boli wtd boundary edge is assigned weight mne and zero respectiyely. Vhe ninimum weight decodec is then implementeq as follmwa. After the error $E$ is akplieg, thd noaes cprdesponfinj to unsatiafied stabioizers are highlighued. Pg an odd numger of seabilizers was unsatisfied, one of the bmunsary nodes is also highoighted. Highlighted nldes are ehen efficiently paired together by the minimum waight zonucgtiovw ln the graph, by Edmonds’ algorithm [@Edmonds65; @Kolikgprpv09]. The correctlon $C$ is applied tp hhr edges in the conneefikn. Note that any slngle $\mwthcao{O}(p)$ fault in yhis noise model correspondw to a weighn onw edge on the grapk. [0.3]{}![Circuits fox measoring (s) $Z$-type, and (b) $X$-type genzratora. Identity hates (blady rectangles) are inxested in ufe $Z$-type stabilizqr measurxment circuigs tp ensuwe that alp measurements are synchgonizgd. Nota that unllke in [@FMMC12], to be consistent wivi the other svhamev in thif papcr, we assume thwt we can prepcre and ieasufe in both the $X$ end $Z$ basis.[]{dwta-label="fig:Susvace17Circuits"}](VstabMeasFurfqceCide.png "wkg:"){width="\textwidyh"} [0.3]{}![Circuits for measyring (a) $Z$-type, and (n) $X$-tyke generators. Ideuuiti gates (black rrctxngjed) ere igverted in tha $Z$-thpe xtabiuizer mtcsuvemdnt vircuits to ensure tvat zll measurements ate synchrobized. Noee that unlikr in [@FMMC12], to be cojsistxnt wivh the otrer schemes in this paper, we aasume thah wc can prepare and measure in yoth the $X$ and $Z$ basis.[]{data-label="fig:Surfacx17Circuits"}](XstabMeasSurfaxeCode.png "fig

$-type correction can be found efficiently for code terms of graph $G_{\text{2D}}$ shown $G_{\text{2D}}$ a bulk node circle) for each stabilizer generator, and a bulk edge for each data qubit. A bulk edge coming from a bulk node corresponds the edge’s data qubit being in the support of the node’s stabilizer. The also boundary (white and boundary edges (blue), which do not correspond to stabilizers or data qubits. Each bulk and edge is assigned weight one and zero respectively. minimum weight decoder is implemented as follows. After the $E$ applied, the corresponding unsatisfied are highlighted. If odd number of stabilizers was unsatisfied, one of the boundary nodes is also highlighted. Highlighted nodes are efficiently paired the minimum connections the by Edmonds’ algorithm The correction $C$ is applied to the connection. Note that any single $\mathcal{O}(p)$ fault this noise corresponds to a weight one edge the graph. [0.3]{}![Circuits for measuring (a) $Z$-type, and $X$-type generators. Identity gates (black rectangles) are inserted in the $Z$-type stabilizer measurement circuits to all measurements are synchronized. that unlike in to consistent the schemes in paper, we assume that we can prepare and measure in both $X$ and $Z$ basis.[]{data-label="fig:Surface17Circuits"}](ZstabMeasSurfaceCode.png "fig:"){width="\textwidth"} [0.3]{}![Circuits for measuring (a) $Z$-type, $X$-type Identity gates (black are inserted in the stabilizer circuits to ensure that are Note [@FMMC12], be with the other schemes this paper, we assume that can prepare and measure $Z$ basis.[]{data-label="fig:Surface17Circuits"}](XstabMeasSurfaceCode.png "fig

$-type correction can be found eFficiently For thE suRfaCe Code In teRms of the graph $G_{\TExt{2D}}$ Shown in Fig. \[fig:Surfaced5GRaph\]. THe GRaph $g_{\TeXt{2D}}$ haS a bulk nODe (BLAck CiRcLe) fOr EAcH $Z$ staBilIzer genErator, and a BulK eDge (black) for eACh Data qubit. A BulK edge coming fRom A bulk nOdE coRRespoNds To the Edge’s dATa qubiT being in tHe SUpport OF the nodE’S StAbilIzer. The graph also cONtAIns boundary nodEs (whitE bOXeS) ANd bOunDary edges (bLuE), whicH Do not coRReSPONd tO Stabilizers or Data qubits. EACh bUlk and BoUndARy edge Is assIgNEd wEight one and Zero RespectivEly. The MInimum wEIght decOder is TheN imPlemENtEd As fOlLOws. aFtEr tHE erRor $E$ is apPlIeD, the nOdes CORREspoNdiNg to UnsatIsfied stabiliZerS are HIghLightEd. If aN odd NuMber oF stabiLizerS wAs unsatisfied, onE of tHe boundarY noDeS is AlSo higHLighteD. HiGhlIghted nOdes are THen EfFICIeNtly paired together By THE mInimum weIght coNNeCtIOns in the GrAph, By EdMONds’ alGoriTHm [@edmonds65; @KOlmogoROv09]. thE correcTiOn $C$ is aPpLieD to The edGEs in The conNection. NOte thAT any single $\mathCAl{O}(p)$ fault in thIS nOISe MOdel CorResponds to a WeigHT one Edge ON tHe gRAph. [0.3]{}![CiRcuitS fOR mEAsuring (a) $Z$-type, and (b) $X$-tYpE generAtors. identity gates (Black rectaNGLEs) are insErteD In THe $Z$-type stabiliZer meAsurement cIRcuits to EnsurE that all MeasuremeNTS are syncHroNizEd. NOte THAt Unlike in [@FMMC12], tO BE conSiStent wiTh tHe other SchEmeS in ThiS pAper, we assUme that wE cAn PrEpAre And meASure in boTh The $x$ aNd $Z$ Basis.[]{DAta-labEl="fig:surfAcE17CIRcuIts"}](ZstaBmeASsurfAcECOde.pNg "fIg:"){Width="\TextWIdtH"} [0.3]{}![CircuiTs for measUriNG (a) $Z$-tYpE, aNd (b) $X$-typE generators. IdEnTity gates (bLaCk rEctangLES) are inseRted in the $Z$-type stabilizeR MeasureMenT circUits To ensure tHat All meaSurEMents aRe syncHroniZeD. NoTE That uNLIkE in [@fMmC12], to be consISTenT with ThE othEr schemEs in this paper, we assUMe tHat we can prepaRe aNd meASUrE in BOtH The $x$ aND $Z$ bASIs.[]{data-label="fig:SUrface17CircUiTS"}](XStabMeasSuRFacECOde.png "fIg

$-type correction can be f ound effic ientl y f orth e su rfac e code in term s ofthe graph $G_{\text{2D }}$ s ho w n in Fi g. \[ fig:Sur f ac e d 5Gr ap h\ ].Th e g raph$G_ {\text{ 2D}}$ hasa b ul k node (blac k c ircle) for ea ch $Z$ stabi liz er gen er ato r , and abulkedge ( b lack)for eachda t a qubi t . A bul k ed ge c oming from a bulk no d e correspondsto the e d ge ’ s da taqubit bein gin th e suppor t o f t hen ode’s stabili zer. The gr a phalso c on tai n s boun daryno d es(white boxe s) a nd bounda ry edg e s (blue ) , which do no t c orr espo n dto st ab i liz e rs or dat a qubits .Ea ch bu lk a n d b ound ary edg e isassigned weig htonea ndzerorespe ctiv el y. Th e mini mum w ei ght decoder isthen implemen ted a s f ol lows. Afterthe er ror $E$ is app l ied ,t h e n odes corresponding t o un satisfie d stab i li ze r s are hi gh lig hted . If an odd nu mber ofstabil i ze rs was un sa tisfie d, on e o f the boun dary n odes isalsoh ighlighted. Hi g hlighted node s a r e t h en e ffi ciently pai redt oget herb ythe minim um we ig h tc onnections in the g ra ph, by Edmo nds’ algorith m [@Edmond s 6 5 ; @Kolmo goro v 09 ] . The correcti on $C $ is appli e d to the edge s in the connecti o n . Note t hat an y s ing l e $ \mathcal{O}(p ) $ fau lt in thi s n oise mo del co rre spo nd s to a we ight one e dg eon th e gra p h. [0.3 ]{ }![ Ci rcu its f o r meas uring (a) $ Z$ - typ e, and( b) $ X$-t yp egene rat or s. Id enti t y g ates (b lack rect ang l es)ar einserte d in the $Z$- ty pe stabili ze r m easure m e nt circu its to ensure that allm easurem ent s are syn chronized . N ote th atu nlikein [@F MMC12 ], to b e con s i st ent w ith the ot h e r s cheme sin t his pap er, we assume that wecan prepare a ndmeas u r einb ot h th e$ X$a n d $Z$ basis.[]{ data-label =" f ig :Surface17 C irc ui ts"}](Z stabMea sSurf a ceCode. png "fig: "){width= "\ text w i dth "} [0.3]{ }![Circu its for m e asuri n g(a) $ Z$- type,an d ( b) $X $-type gen erato rs. Id en tity g ates(b lack rec tangles) are inserted i n the$Z$-t ype stabiliz erm eas urement c ircu its to ens ure th at al l m e asure ment s a res ynchr oniz e d. Note t h at un l i ke in [@FMMC1 2 ] , to be c ons i stentwith the other scheme s in this paper , we a ssu met hatwe can prepare a ndme a s ure in b ot h the $X$ a nd $Z$ b as i s.[]{ data-l abel=" fig:Sur f a ce 1 7Circu its" }]( XstabMeas Sur fa c eCode.p ng " f ig

$-type correction_can be_found efficiently for the_surface code_in_terms of_the_graph $G_{\text{2D}}$ shown_in Fig. \[fig:Surfaced5Graph\]. The_graph $G_{\text{2D}}$ has a_bulk node (black_circle)_for each $Z$ stabilizer generator, and a bulk edge (black) for each data qubit._A_bulk edge_coming_from_a bulk node corresponds to_the edge’s data qubit being_in the_support of the node’s stabilizer. The graph also_contains_boundary nodes (white_boxes) and boundary edges (blue), which do not correspond_to stabilizers or data qubits. Each_bulk and boundary_edge_is_assigned weight one and_zero respectively. The minimum weight decoder_is then implemented as follows. After_the error $E$ is applied, the nodes_corresponding to unsatisfied stabilizers are highlighted._If an odd number of_stabilizers was_unsatisfied, one of the boundary_nodes is also_highlighted. Highlighted_nodes are then_efficiently paired together by the minimum_weight connections in_the graph, by Edmonds’ algorithm [@Edmonds65;_@Kolmogorov09]._The correction $C$_is_applied_to the_edges in the_connection._Note that_any_single $\mathcal{O}(p)$ fault in this noise_model_corresponds to a weight one edge on_the graph. [0.3]{}![Circuits for measuring_(a)_$Z$-type, and (b) $X$-type_generators. Identity gates (black rectangles)_are inserted in the $Z$-type stabilizer_measurement circuits_to ensure_that all measurements are synchronized. Note that unlike in [@FMMC12], to_be consistent with the other schemes_in this paper, we_assume that_we_can prepare and_measure_in both_the $X$ and $Z$ basis.[]{data-label="fig:Surface17Circuits"}](ZstabMeasSurfaceCode.png "fig:"){width="\textwidth"} [0.3]{}![Circuits for_measuring (a)_$Z$-type, and (b) $X$-type generators. Identity_gates (black rectangles) are_inserted_in the $Z$-type stabilizer measurement circuits_to ensure that all measurements are_synchronized. Note that unlike in_[@FMMC12],_to_be consistent with the other_schemes in this paper, we assume_that we can_prepare and measure in both the $X$_and_$Z$ basis.[]{data-label="fig:Surface17Circuits"}](XstabMeasSurfaceCode.png "fig

frac{d\xi_r}{dr}=-A\rho^{-1/2}\left(-ik_r+\frac{1}{H}\right)e^{ik_rr},\label{eq-osc10}\\ \frac{dP'}{dr}=-B\rho^{1/2}\left(-ik_r-\frac{1}{H}\right)e^{ik_rr}\label{eq-osc11},\end{aligned}$$ where $$H=\left(\frac{d\log\rho}{dr}\right)^{-1},$$ is the [*density scale height*]{}. From (\[eq-osc10\]-\[eq-osc11\]) we get a linear system for the constant, $A$, and $B$: $$\left(-ik_r+\frac{1}{H}\right)A-\frac{g}{c^2}A+ \frac{1}{c^2}\left(1-\frac{S_l^2}{\omega^2}\right)B=0,$$ $$\left(-ik_r-\frac{1}{H}\right)B+\frac{g}{c^2}B+ (N^2-\omega^2)A=0.$$ It has a non-zero solution when the determinant is equal zero, that is when $$k_r^2=\frac{\omega^2-\omega_c^2}{c^2}+\frac{S_l^2}{c^2\omega^2}\left(N^2-\omega^2\right),\label{eq-osc12}$$ where $$\omega_c=\frac{c}{2H} \label{eq-cutoff}$$ is [*the acoustic cut-off frequency*]{}. Here, we used the relation: $N^2=g/H-g^2/c^2$. ![ Buoyancy (Brünt-Väisälä) frequency $N$ (thick curve), acoustic cut-off frequency, $\omega_c$ (thin curve) and Lamb frequency $S_l$ for $l$=1, 5, 20, 50, and 100 (dashed curves) vs. fractional radius $r/R$ for a standard solar model. The horizontal lines with arrows indicate the trapping regions for a g mode with frequency $\nu=0.2$ mHz, and for a sample of five p modes: $l=1$, $\nu=1$ mHz; $l=

frac{d\xi_r}{dr}=-A\rho^{-1/2}\left(-ik_r+\frac{1}{H}\right)e^{ik_rr},\label{eq - osc10}\\ \frac{dP'}{dr}=-B\rho^{1/2}\left(-ik_r-\frac{1}{H}\right)e^{ik_rr}\label{eq - osc11},\end{aligned}$$ where $ $ H=\left(\frac{d\log\rho}{dr}\right)^{-1},$$ is the [ * density scale height * ] { }. From (\[eq - osc10\]-\[eq - osc11\ ]) we get a analogue arrangement for the constant, $ A$, and $ B$: $ $ \left(-ik_r+\frac{1}{H}\right)A-\frac{g}{c^2}A+ \frac{1}{c^2}\left(1-\frac{S_l^2}{\omega^2}\right)B=0,$$ $ $ \left(-ik_r-\frac{1}{H}\right)B+\frac{g}{c^2}B+ (N^2-\omega^2)A=0.$$ It has a non - zero solution when the deciding is equal zero, that is when $ $ k_r^2=\frac{\omega^2-\omega_c^2}{c^2}+\frac{S_l^2}{c^2\omega^2}\left(N^2-\omega^2\right),\label{eq - osc12}$$ where $ $ \omega_c=\frac{c}{2H } \label{eq - cutoff}$$ is [ * the acoustic abridge - off frequency * ] { }. Here, we used the relation: $ N^2 = g / H - g^2 / c^2$. ! [ Buoyancy (Brünt - Väisälä) frequency $ N$ (blockheaded curve), acoustic cut - off frequency, $ \omega_c$ (slender curve) and Lamb frequency $ S_l$ for $ l$=1, 5, 20, 50, and 100 (dashed curvature) vs. fractional radius $ r / R$ for a standard solar mannequin. The horizontal line with arrows indicate the trapping regions for a g manner with frequency $ \nu=0.2 $ mHz, and for a sample of five p mood: $ l=1 $, $ \nu=1 $ mHz; $ l=

fraf{d\xi_r}{dr}=-A\rho^{-1/2}\left(-ik_r+\frac{1}{H}\rlght)e^{ik_rr},\label{eq-osc10}\\ \frac{vP'}{dr}=-B\rhk^{1/2}\left(-ik_r-\wrac{1}{H}\right)e^{ik_rr}\label{eq-osc11},\end{eligbed}$$ wyere $$H=\left(\frac{d\log\rho}{df}\right)^{-1},$$ is the [*denwity wcale heigif*]{}. From (\[eq-osc10\]-\[eq-kdc11\]) wz jet a linear syxtem for tve constant, $A$, dna $Y$: $$\left(-ik_r+\frac{1}{H}\right)A-\frac{g}{c^2}A+ \frac{1}{c^2}\left(1-\srac{S_l^2}{\okeha^2}\right)B=0,$$ $$\left(-ih_r-\frss{1}{H}\rifht)B+\frac{g}{c^2}B+ (N^2-\omega^2)A=0.$$ It has a non-zedo soluuion when the detetminant is equal zero, that is ahen $$k_r^2=\frac{\omega^2-\omfga_c^2}{c^2}+\frac{S_l^2}{x^2\omedq^2}\left(N^2-\omega^2\rkght),\label{eq-osc12}$$ where $$\ojega_c=\frac{c}{2H} \label{eq-cutoff}$$ is [*the acouxtic cut-ofd dreeoency*]{}. Here, wx used the relatiok: $N^2=g/H-g^2/c^2$. ![ Buoyanvy (Brünt-Väisälä) frcquenry $N$ (thick curve), acoustic cut-off frequency, $\jmega_c$ (thhn curve) and Lamb feewuenci $S_l$ xor $u$=1, 5, 20, 50, ans 100 (dzshed furtes) vs. fracfional radiys $r/R$ for a standarc filar model. Ths horisogtal lines with arrows indicate the trakping regions for a g mode wuth frequency $\nu=0.2$ mHz, wnd for a sample of five p modes: $l=1$, $\nu=1$ mHz; $l=

frac{d\xi_r}{dr}=-A\rho^{-1/2}\left(-ik_r+\frac{1}{H}\right)e^{ik_rr},\label{eq-osc10}\\ \frac{dP'}{dr}=-B\rho^{1/2}\left(-ik_r-\frac{1}{H}\right)e^{ik_rr}\label{eq-osc11},\end{aligned}$$ where $$H=\left(\frac{d\log\rho}{dr}\right)^{-1},$$ is the [*density From we get linear system for $$\left(-ik_r+\frac{1}{H}\right)A-\frac{g}{c^2}A+ $$\left(-ik_r-\frac{1}{H}\right)B+\frac{g}{c^2}B+ (N^2-\omega^2)A=0.$$ It a non-zero solution the determinant is equal zero, that when $$k_r^2=\frac{\omega^2-\omega_c^2}{c^2}+\frac{S_l^2}{c^2\omega^2}\left(N^2-\omega^2\right),\label{eq-osc12}$$ where $$\omega_c=\frac{c}{2H} \label{eq-cutoff}$$ is [*the acoustic cut-off frequency*]{}. Here, we used relation: $N^2=g/H-g^2/c^2$. ![ Buoyancy (Brünt-Väisälä) frequency $N$ (thick curve), acoustic cut-off frequency, $\omega_c$ curve) Lamb $S_l$ $l$=1, 5, 20, 50, and 100 (dashed curves) vs. fractional radius $r/R$ for a standard solar The horizontal lines with arrows indicate the trapping for a g mode frequency $\nu=0.2$ mHz, and for sample five p $l=1$, mHz;

frac{d\xi_r}{dr}=-A\rho^{-1/2}\left(-ik_r+\frac{1}{h}\right)e^{ik_rR},\labeL{eq-Osc10}\\ \FrAc{dP'}{Dr}=-B\rHo^{1/2}\left(-ik_r-\frac{1}{H}\RIght)E^{ik_rr}\label{eq-osc11},\end{aligNed}$$ whErE $$h=\lefT(\FrAc{d\loG\rho}{dr}\rIGhT)^{-1},$$ IS thE [*dEnSitY sCAlE heigHt*]{}. FRom (\[eq-osC10\]-\[eq-osc11\]) we geT a lInEar system for THe Constant, $A$, aNd $B$: $$\Left(-ik_r+\frac{1}{H}\RigHt)A-\fraC{g}{C^2}A+ \fRAc{1}{c^2}\leFt(1-\fRac{S_l^2}{\Omega^2}\rIGht)B=0,$$ $$\leFt(-ik_r-\frac{1}{h}\rIGht)B+\frAC{g}{c^2}B+ (N^2-\omEGA^2)A=0.$$ it haS a non-zero solution WHeN The determinant Is equaL zERo, THAt iS whEn $$k_r^2=\frac{\omEgA^2-\omegA_C^2}{c^2}+\frac{S_L^2}{C^2\oMEGA^2}\leFT(N^2-\omega^2\right),\lAbel{eq-osc12}$$ whERe $$\oMega_c=\fRaC{c}{2H} \LAbel{eq-CutofF}$$ iS [*The Acoustic cut-Off fRequency*]{}. HEre, we uSEd the reLAtion: $N^2=g/h-g^2/c^2$. ![ BuoYanCy (BRünt-vÄiSäLä) fReQUenCY $N$ (ThiCK cuRve), acousTiC cUt-off FreqUENCY, $\omeGa_c$ (Thin Curve) And Lamb frequeNcy $s_l$ foR $L$=1, 5, 20, 50, anD 100 (dashEd curVes) vS. fRactiOnal raDius $r/r$ fOr a standard solaR modEl. The horiZonTaL liNeS with ARrows iNdiCatE the traPping reGIonS fOR A G mOde with frequency $\nu=0.2$ MHZ, ANd For a sampLe of fiVE p MoDEs: $l=1$, $\nu=1$ mHz; $L=

frac{d\xi_r}{dr}=-A\rho^{- 1/2}\left( -ik_r +\f rac {1 }{H} \rig ht)e^{ik_rr},\ l abel {eq-osc10}\\ \frac{dP' }{dr} =- B \rho ^ {1 /2}\l eft(-ik _ r- \ f rac {1 }{ H}\ ri g ht )e^{i k_r r}\labe l{eq-osc11 },\ en d{aligned}$$ wh ere $$H=\l eft (\frac{d\log \rh o}{dr} \r igh t )^{-1 },$ $ isthe [* d ensity scale he ig h t*]{}. From ( \ [ eq -osc 10\]-\[eq-osc11\] ) w e get a linearsystem f o rt h e c ons tant, $A$, a nd $B $ : $$\le f t( - i k _r+ \ frac{1}{H}\ri ght)A-\frac { g}{ c^2}A+ \ fra c {1}{c^ 2}\le ft ( 1-\ frac{S_l^2} {\om ega^2}\ri ght)B= 0 ,$$ $$\ l eft(-ik _r-\fr ac{ 1}{ H}\r i gh t) B+\ fr a c{g } {c ^2} B + ( N^2-\ome ga ^2 )A=0. $$ I t h a s anon -zer o sol ution when th e d eter m ina nt is equa l ze ro , tha t is w hen $ $k _r^2=\frac{\ome ga^2 -\omega_c ^2} {c ^2} +\ frac{ S _l^2}{ c^2 \om ega^2}\ left(N^ 2 -\o me g a ^ 2\ right),\label{eq-o sc 1 2 }$ $ where$$\ome g a_ c= \ frac{c}{ 2H } \ labe l { eq-cu toff } $$ is [*th e acou s ti ccut-off f requen cy *]{ }.Here, we u sed th e relati on: $ N ^2=g/H-g^2/c^2 $ . ![ Buoyanc y ( B r ün t -Väi säl ä) frequenc y $N $ (th ickc ur ve) , acou sticcu t -o f f frequency, $\omeg a_ c$ (th in cu rve) and Lamb frequency $ S _l$ for$l$= 1 ,5 , 20, 50, and100 ( dashed cur v es) vs.fract ional ra dius $r/R $ for a st and ard so lar m od el. The horiz o n talli nes wit h a rrows i ndi cat e t hetr apping re gions fo ragmo dewithf requency $ \nu =0 .2$ mHz, and fo r a s ampl eof fiv e p mod e s: $ l=1$ ,$\ nu=1 $ m Hz ; $l=

frac{d\xi_r}{dr}=-A\rho^{-1/2}\left(-ik_r+\frac{1}{H}\right)e^{ik_rr},\label{eq-osc10}\\ \frac{dP'}{dr}=-B\rho^{1/2}\left(-ik_r-\frac{1}{H}\right)e^{ik_rr}\label{eq-osc11},\end{aligned}$$ where_$$H=\left(\frac{d\log\rho}{dr}\right)^{-1},$$ is_the [*density scale height*]{}. From_(\[eq-osc10\]-\[eq-osc11\]) we_get_a linear_system_for the constant,_$A$, and $B$:_$$\left(-ik_r+\frac{1}{H}\right)A-\frac{g}{c^2}A+ \frac{1}{c^2}\left(1-\frac{S_l^2}{\omega^2}\right)B=0,$$ $$\left(-ik_r-\frac{1}{H}\right)B+\frac{g}{c^2}B+ (N^2-\omega^2)A=0.$$ It_has a non-zero_solution_when the determinant is equal zero, that is when $$k_r^2=\frac{\omega^2-\omega_c^2}{c^2}+\frac{S_l^2}{c^2\omega^2}\left(N^2-\omega^2\right),\label{eq-osc12}$$ where $$\omega_c=\frac{c}{2H} \label{eq-cutoff}$$ is [*the_acoustic_cut-off frequency*]{}._Here,_we_used the relation: $N^2=g/H-g^2/c^2$. ![ Buoyancy_(Brünt-Väisälä) frequency $N$ (thick curve),_acoustic cut-off_frequency, $\omega_c$ (thin curve) and Lamb frequency $S_l$_for_$l$=1, 5, 20,_50, and 100 (dashed curves) vs. fractional radius $r/R$_for a standard solar model. The_horizontal lines with_arrows_indicate_the trapping regions for_a g mode with frequency $\nu=0.2$_mHz, and for a sample of_five p modes: $l=1$, $\nu=1$ mHz; $l=

, [**47**]{}, No. 3, 163-165 (1945). B. M. Levitan, “Plancherel theorem for generalized translation operation,” Dokl. Akad. Nauk SSSR, [**47**]{}, No. 5, 323-326 (1945). B. M. Levitan, “Duality law for generalized translation operation,” Dokl. Akad. Nauk SSSR, [**47**]{}, No. 6, 401-403 (1945). B. M. Levitan, “Application of generalized translation operators to second order linear differential operators,” Usp. Mat. Nauk, [**4**]{}, No. 1, 3-112 (1949). B. M. Levitan, “Lie theorems for generalized translation operators,” Dokl. Akad. Nauk SSSR, [**123**]{}, No. 1, 32-35 (1958). B. M. Levitan, “Inverse Lie theorems for generalized translation operators,” Dokl. Akad. Nauk SSSR, [**123**]{}, No. 2, 243-245 (1958). B. M. Levitan, “Lie theorems for generalized translation operators,” Usp. Mat. Nauk, [**10**]{}, No. 4, 3-30 (1961). B. M. Levitan, Generalized Translation Operators and Some of Their Applications \[in Russian\], Nauka, Moscow (1962). B. M. Levitan, Theory of Generalized Translation Operators \[in Russian\], Nauka, Moscow (1973). B. M. Levitan and G. L. Litvinov, “Generalized translation operators,” in: Mathematical Encyclopaedia \[in Russian\], Vol. 3 (1982), pp. 1112-1116. D. A. Leites, “Lie superalgebras,” J. Sov. Math., [**30**]{}, No. 6 (1985). G. L. Litvinov, “Representations of groups on locally convex spaces and topological group algebras,” Tr. Sem. Vektorn. Tenzorn. Analizu, Mosk. Gos. Univ., [**16**]{}, 267-349 (1972). G. L. Litvinov, “Conditions under which a representation is determined by its character up to equivalence,” Tr. Sem. Vektorn. Tenzorn. Analizu, Mosk. Gos. Univ., [**17**]{}, 325-349 (1974). G. L. Lit

, [ * * 47 * * ] { }, No. 3, 163 - 165 (1945). B. M. Levitan, “ Plancherel theorem for generalized translation operation, ” Dokl. Akad. Nauk SSSR, [ * * 47 * * ] { }, No. 5, 323 - 326 (1945). B. M. Levitan, “ Duality law for generalized transformation process, ” Dokl. Akad. Nauk SSSR, [ * * 47 * * ] { }, No. 6, 401 - 403 (1945). B. M. Levitan, “ Application of generalized translation operator to second orderliness linear differential operators, ” Usp. Mat. Nauk, [ * * 4 * * ] { }, No. 1, 3 - 112 (1949). B. M. Levitan, “ Lie theorems for generalized transformation operators, ” Dokl. Akad. Nauk SSSR, [ * * 123 * * ] { }, No. 1, 32 - 35 (1958). B. M. Levitan, “ Inverse Lie theorems for generalized translation operators, ” Dokl. Akad. Nauk SSSR, [ * * 123 * * ] { }, No. 2, 243 - 245 (1958). B. M. Levitan, “ Lie theorem for generalized translation operators, ” Usp. Mat. Nauk, [ * * 10 * * ] { }, No. 4, 3 - 30 (1961). B. M. Levitan, Generalized Translation Operators and Some of Their Applications \[in Russian\ ], Nauka, Moscow (1962). B. M. Levitan, hypothesis of Generalized Translation Operators \[in Russian\ ], Nauka, Moscow (1973). B. M. Levitan and G. L. Litvinov, “ Generalized translation operators, ” in: Mathematical Encyclopaedia \[in Russian\ ], Vol. 3 (1982), pp. 1112 - 1116. D. A. Leites, “ Lie superalgebras, ” J. Sov. Math. , [ * * 30 * * ] { }, No. 6 (1985). G. L. Litvinov, “ Representations of group on locally convex spaces and topological group algebras, ” Tr. Sem. Vektorn. Tenzorn. Analizu, Mosk. Gos. Univ. , [ * * 16 * * ] { }, 267 - 349 (1972). G. L. Litvinov, “ Conditions under which a representation is determined by its fictional character up to equivalence, ” Tr. Sem. Vektorn. Tenzorn. Analizu, Mosk. Gos. Univ. , [ * * 17 * * ] { }, 325 - 349 (1974). G. L. Lit

, [**47**]{}, Nl. 3, 163-165 (1945). B. M. Levitan, “Planchevel theorem for ywneralmzed trznslatiov operation,” Dokl. Akad. Nauk SDSE, [**47**]{}, No. 5, 323-326 (1945). B. M. Levitan, “Dualith law for generaluzed rranslatioi operatljn,” Dknl. Aned. Nauk SSSR, [**47**]{}, Np. 6, 401-403 (1945). B. M. Lavitan, “Applicadiun of generalized translation operatows to srclnd order linewr dpfserehnicl operators,” Usp. Mat. Nauk, [**4**]{}, No. 1, 3-112 (1949). B. M. Ltvitan, “Lie theoremx for generalized translatlon lperators,” Dokl. Akaf. Nauk SSSR, [**123**]{}, No. 1, 32-35 (1958). B. M. Levitxn, “Inverse Lie theorema for generalized translation ooeratprs,” Dokl. Ajae. Nwok SSSR, [**123**]{}, No. 2, 243-245 (1958). B. I. Levitan, “Lic theorams for generalized tvansletiob operators,” Usp. Mat. Neuk, [**10**]{}, No. 4, 3-30 (1961). B. M. Levitwn, Generanived Translation Operaroes ang Soke ow Thdir A'pljcatiojs \[mn Russian\], Hauka, Moscoq (1962). B. M. Levitan, Theoty if Generalizes Tranflwtion Operators \[in Russian\], Nauka, Moscow (1973). B. M. Levitan and G. L. Litvunov, “Generalized trandlation o[erators,” in: Mathematical Encyclopaedia \[in Russian\], Vol. 3 (1982), pp. 1112-1116. Q. Q. Peites, “Lie superalgebras,” J. Sov. Math., [**30**]{}, No. 6 (1985). G. L. Miuvikov, “Representatijns of grouls om locally convgx spacza znd topological grlup alggbras,” Rr. Sem. Vehtorm. Tenzorn. Analizu, Mosk. Gos. Yniv., [**16**]{}, 267-349 (1972). G. L. Jutvinov, “Conditions under whick a rekresenyation is determined by its character kp to equjxalence,” Tr. Sem. Vdktprt. Tenzorn. Analizu, Mosk. Gos. Univ., [**17**]{}, 325-349 (1974). J. L. Lnt

, [**47**]{}, No. 3, 163-165 (1945). B. “Plancherel for generalized operation,” Dokl. Akad. 323-326 B. M. Levitan, law for generalized operation,” Dokl. Akad. Nauk SSSR, [**47**]{}, 6, 401-403 (1945). B. M. Levitan, “Application of generalized translation operators to second linear differential operators,” Usp. Mat. Nauk, [**4**]{}, No. 1, 3-112 (1949). B. M. “Lie for translation Dokl. Akad. Nauk SSSR, [**123**]{}, No. 1, 32-35 (1958). B. M. Levitan, “Inverse Lie theorems for translation operators,” Dokl. Akad. Nauk SSSR, [**123**]{}, No. 243-245 (1958). B. M. “Lie theorems for generalized translation Usp. Nauk, [**10**]{}, 4, (1961). M. Levitan, Generalized Operators and Some of Their Applications \[in Russian\], Nauka, Moscow (1962). B. M. Levitan, Theory of Generalized Operators \[in Moscow (1973). M. and L. Litvinov, “Generalized in: Mathematical Encyclopaedia \[in Russian\], Vol. 1112-1116. D. A. Leites, “Lie superalgebras,” J. Sov. [**30**]{}, No. (1985). G. L. Litvinov, “Representations of on locally convex spaces and topological group algebras,” Sem. Vektorn. Tenzorn. Analizu, Mosk. Gos. Univ., [**16**]{}, 267-349 (1972). G. L. Litvinov, “Conditions under representation is determined by character up to Tr. Vektorn. Analizu, Gos. Univ., 325-349 (1974). G. L. Lit

, [**47**]{}, No. 3, 163-165 (1945). B. M. Levitan, “Plancherel theoRem for geneRalizEd tRanSlAtioN opeRation,” Dokl. Akad. nAuk SsSR, [**47**]{}, No. 5, 323-326 (1945). B. M. Levitan, “Duality lAw for GeNEralIZeD tranSlation OPeRATioN,” DOkL. AkAd. nAuK SSSR, [**47**]{}, no. 6, 401-403 (1945). B. m. LevitaN, “ApplicatiOn oF gEneralized trANsLation operAtoRs to second orDer Linear DiFfeREntiaL opEratoRs,” Usp. MAT. Nauk, [**4**]{}, NO. 1, 3-112 (1949). B. M. LevitaN, “LIE theorEMs for geNERaLizeD translation operaTOrS,” dokl. Akad. Nauk SSsR, [**123**]{}, No. 1, 32-35 (1958). B. M. leVItAN, “invErsE Lie theoreMs For geNEralizeD TrANSLatIOn operators,” DoKl. Akad. Nauk SssR, [**123**]{}, NO. 2, 243-245 (1958). B. M. LevItAn, “LIE theorEms foR gENerAlized transLatiOn operatoRs,” Usp. MAT. Nauk, [**10**]{}, No. 4, 3-30 (1961). b. m. LevitaN, GenerAliZed tranSLaTiOn OPeRAtoRS aNd SOMe oF Their ApPlIcAtionS \[in RUSSIAn\], NaUka, moscOw (1962). B. M. LEvitan, Theory oF GeNeraLIzeD TranSlatiOn OpErAtors \[In RussIan\], NaUkA, Moscow (1973). B. M. LevitaN and g. L. LitvinoV, “GeNeRalIzEd traNSlatioN opEraTors,” in: MAthematICal enCYCLoPaedia \[in Russian\], Vol. 3 (1982), Pp. 1112-1116. d. a. leItes, “Lie sUperalGEbRaS,” j. Sov. Math., [**30**]{}, no. 6 (1985). g. L. LItviNOV, “ReprEsenTAtIons of grOups on LOcAlLy conveX sPaces aNd TopOloGical GRoup AlgebrAs,” Tr. Sem. VEktorN. tenzorn. Analizu, mOsk. Gos. Univ., [**16**]{}, 267-349 (1972). G. L. LITvINOv, “cOndiTioNs under whicH a rePReseNtatIOn Is dETermiNed by ItS ChARacter up to equivalenCe,” tr. Sem. VEktorN. Tenzorn. AnaliZu, Mosk. Gos. UNIV., [**17**]{}, 325-349 (1974). g. L. Lit

, [**47**]{}, No. 3, 163-1 65 (1945). B. M . L evi ta n, “ Plan cherel theorem forgeneralized translatio n ope ra t ion, ” D okl.Akad. N a uk S SSR ,[* *47 ** ] {} , No. 5, 323-32 6 (1945).B.M. Levitan, “D u al ity law fo r g eneralized t ran slatio nope r ation ,”Dokl. Akad. Nauk S SSR, [**4 7* * ]{}, N o . 6, 40 1 - 40 3 (1 945). B. M. Levit a n, “Application o f gene ra l iz e d tr ans lation ope ra torst o secon d o r d e r l i near differen tial operat o rs, ” Usp. M at. Nauk,[**4* *] { },No. 1, 3-11 2 (1 949). B.M. Lev i tan, “L i e theor ems fo r g ene rali z ed t ran sl a tio n o per a tor s,” Dokl .Ak ad. N aukS S S R , [* *12 3**] {}, N o. 1, 32-35 ( 195 8).B . M . Lev itan, “In ve rse L ie the orems f or generalizedtran slation o per at ors ,” Dokl . Akad. Na ukSSSR, [ **123** ] {}, N o . 2, 243-245 (1958). B .M . L evitan,“Lie t h eo re m s for ge ne ral ized t ransl atio n o perators ,” Usp . M at . Nauk, [ **10** ]{ },No. 4, 3 - 30 ( 1961). B. M. L evita n , GeneralizedT ranslation Op e ra t o rs andSom e of TheirAppl i cati ons\ [i n R u ssian \], N au k a, Moscow (1962). B. M .Levita n, Th eory of Gener alized Tra n s l ation Op erat o rs \[in Russian\] , Nau ka, Moscow (1973).B. M. Levitan and G. L . Litvinov , “ Gen era liz e d t ranslation op e r ator s, ” in: M ath ematica l E ncy clo pae di a \[in Ru ssian\], V ol .3(19 82),p p. 1112- 11 16. D . A . Lei t es, “L ie su pera lg eb r as, ” J. So v .M a th., [ ** 30** ]{} ,No. 6 (19 8 5). G. L.Litvinov, “R e pres en ta tions o f groups on l oc ally conve xspa ces an d topologi cal group algebras,” Tr . Sem. V ekt orn.Tenz orn. Anal izu , Mosk . G o s. Uni v., [* *16** ]{ },2 6 7-349 ( 19 72) .G. L. Litv i n ov, “Con di tion s under which a represent a tio n is determin edby i t s c har a ct e r u pt o e q u ivalence,” Tr.Sem. Vekto rn . T enzorn. An a liz u, Mosk.Gos. Un iv.,[ **17**] {}, 325-3 49 (1974) .G. L . Lit

, [**47**]{},_No. 3,_163-165 (1945). B. M._Levitan, “Plancherel_theorem_for generalized_translation_operation,” Dokl. Akad._Nauk SSSR, [**47**]{},_No. 5, 323-326 (1945)._B. M. Levitan,_“Duality_law for generalized translation operation,” Dokl. Akad. Nauk SSSR, [**47**]{}, No. 6, 401-403 (1945)._B._M. Levitan,_“Application_of_generalized translation operators to second_order linear differential operators,” Usp._Mat. Nauk,_[**4**]{}, No. 1, 3-112 (1949). B. M. Levitan,_“Lie_theorems for generalized_translation operators,” Dokl. Akad. Nauk SSSR, [**123**]{}, No. 1,_32-35 (1958). B. M. Levitan, “Inverse_Lie theorems for_generalized_translation_operators,” Dokl. Akad. Nauk_SSSR, [**123**]{}, No. 2, 243-245 (1958)._B. M. Levitan, “Lie theorems for_generalized translation operators,” Usp. Mat. Nauk, [**10**]{},_No. 4, 3-30 (1961). B. M._Levitan, Generalized Translation Operators and_Some of_Their Applications \[in Russian\], Nauka,_Moscow (1962). B._M. Levitan,_Theory of Generalized_Translation Operators \[in Russian\], Nauka, Moscow_(1973). B. M._Levitan and G. L. Litvinov, “Generalized_translation_operators,” in: Mathematical_Encyclopaedia_\[in_Russian\], Vol._3 (1982), pp._1112-1116._D. A._Leites,_“Lie superalgebras,” J. Sov. Math., [**30**]{},_No._6 (1985). G. L. Litvinov, “Representations of_groups on locally convex_spaces_and topological group algebras,”_Tr. Sem. Vektorn. Tenzorn. Analizu,_Mosk. Gos. Univ., [**16**]{}, 267-349 (1972)._G. L._Litvinov, “Conditions_under which a representation is determined by its character up to_equivalence,” Tr. Sem. Vektorn. Tenzorn. Analizu,_Mosk. Gos. Univ., [**17**]{},_325-349 (1974)._G._L. Lit

in different context in [@mr:fptts]), $s_0$-set : i.e. a set $A\subseteq 2^\omega$ such that for any perfect set $P$ there exists a perfect set $Q\subseteq P$ such that $A\cap Q={\varnothing}$ ([@em:cfscce]), $v_0$-set : i.e. a set $A\subseteq 2^\omega$ such that for every Silver perfect set $P$, there exists a Silver perfect set $Q\subseteq P$ such that $Q\cap A={\varnothing}$ (see [@mkantw:ssrtfn]), $l_0$-set : i.e. a set $A\subseteq \omega^\omega$ such that for every Laver perfect set $P$, there exists a  Laver perfect set $Q\subseteq P$ such that $Q\cap A={\varnothing}$ (see [@mktw:ssrtfn]), $m_0$-set : i.e. a set $A\subseteq \omega^\omega$ such that for every Miller perfect set $P$, there exists a  Miller perfect set $Q\subseteq P$ such that $Q\cap A={\varnothing}$ (see [@mktw:ssrtfn]). ### Selection principles If ${{\mathcal{A}}}$ and ${{\mathcal{B}}}$ are families of covers of a topological space $X$, then $X$ has [[**$S_1({{\mathcal{A}}},{{\mathcal{B}}})$ principle**]{}]{} if for every sequence $\langle {{\mathcal{U}}}_n\rangle_{n\in\omega}\in{{\mathcal{A}}}^\omega$, there exists ${{\mathcal{U}}}=\{ U_n\colon n\in\omega\}$ with $U_n\in{{\mathcal{U}}}_n,$ for all $n\in\omega$ such that ${{\mathcal{U}}}\in{{\mathcal{B}}}$. $X$ has [[**$U_{<\omega}({{\mathcal{A}}},{{\mathcal{B}}})$ principle**]{}]{} if for every sequence $\langle {{\mathcal{U}}}_n\rangle_{n\in\omega}\in{{\mathcal{A}}}^\omega$ such that for every $n\in\omega$ if ${{\mathcal{W}}}\subseteq {{\

in different context in   [ @mr: fptts ]), $ s_0$-set : i.e. a   set $ A\subseteq 2^\omega$ such that for any perfect set $ P$ there exist a   arrant set $ Q\subseteq P$ such that $ A\cap Q={\varnothing}$ ([ @em: cfscce ]), $ v_0$-set : i.e. a   set $ A\subseteq 2^\omega$ such that for every silver medal perfective set $ P$, there exists a   Silver perfect arrange $ Q\subseteq P$ such that $ Q\cap A={\varnothing}$ (see [ @mkantw: ssrtfn ]), $ l_0$-set : i.e. a   set $ A\subseteq \omega^\omega$ such that for every Laver perfective rig $ P$, there exists a   Laver perfective set $ Q\subseteq P$ such that $ Q\cap A={\varnothing}$ (see [ @mktw: ssrtfn ]), $ m_0$-set : i.e. a   set $ A\subseteq \omega^\omega$ such that for every Miller perfect jell $ P$, there exists a   Miller perfect set $ Q\subseteq P$ such that $ Q\cap A={\varnothing}$ (see [ @mktw: ssrtfn ]). # # # Selection principle If $ { { \mathcal{A}}}$ and $ { { \mathcal{B}}}$ are families of covers of a   topological space $ X$, then $ X$ has [ [ * * $ S_1({{\mathcal{A}}},{{\mathcal{B}}})$ rationale * * ] { } ] { } if for every sequence $ \langle { { \mathcal{U}}}_n\rangle_{n\in\omega}\in{{\mathcal{A}}}^\omega$, there exists $ { { \mathcal{U}}}=\ { U_n\colon n\in\omega\}$ with $ U_n\in{{\mathcal{U}}}_n,$ for all $ n\in\omega$ such that $ { { \mathcal{U}}}\in{{\mathcal{B}}}$. $ X$ has [ [ * * $ U_{<\omega}({{\mathcal{A}}},{{\mathcal{B}}})$ rationale * * ] { } ] { } if for every sequence $ \langle { { \mathcal{U}}}_n\rangle_{n\in\omega}\in{{\mathcal{A}}}^\omega$ such that for every $ n\in\omega$ if $ { { \mathcal{W}}}\subseteq { { \

in different context in [@mr:fktts]), $s_0$-set : i.e. a sej $Q\subseveq 2^\omefa$ such ghat for any perfect set $P$ tiere exisuf a perfect set $Q\sucseteq P$ duch thar $A\cep Q={\varnothing}$ ([@em:cfscce]), $v_0$-set : j.c. a sec $E\subseteq 2^\omega$ such that for every Sileef 'erfect set $P$, there exists a Silver pqrfect xeh $Q\subseteq P$ fuch ehat $Q\cap A={\varnothing}$ (see [@mkantw:ssrtfn]), $m_0$-set : p.e. a set $A\subseteq \omega^\omega$ such that for fverj Laver perfect seh $P$, there ezistf a  Laver perwect set $Q\subseteq P$ soch that $Q\cap A={\varnothing}$ (see [@mkgw:ssrcfn]), $m_0$-set : i.g. z sfj $A\subseteq \imega^\jmega$ such tmst for every Killer perfect sev $P$, rhere exists a  Miller perfect set $Q\subsgteq P$ sucv chat $Q\cap A={\varnothing}$ (swe [@mkjw:ssrdfn]). ### Rwleztikn pdincippes Mf ${{\mathcal{A}}}$ and ${{\mathcao{B}}}$ are families of vodvts of a topolkgical s[ace $X$, then $X$ has [[**$S_1({{\mathcal{A}}},{{\mathcal{B}}})$ pritcille**]{}]{} if for every sequenxe $\langle {{\mathcal{U}}}_n\rajgle_{n\in\omqga}\in{{\mathcal{A}}}^\omega$, there exists ${{\mathcal{U}}}=\{ U_n\colon t\in\omxgx\}$ wnbm $U_n\kb{{\mwthcal{U}}}_n,$ for all $n\in\omega$ such that ${{\mathcal{U}}}\ig{{\jauhcsl{B}}}$. $X$ has [[**$U_{<\omena}({{\mathcal{A}}},{{\mathcal{B}}})$ pgimsiple**]{}]{} if for gvery szsusnce $\langle {{\mathcap{U}}}_n\rangje_{n\in\imega}\in{{\mauhcal{S}}}^\omega$ such that for every $b\in\omega$ if ${{\iqthcal{W}}}\subseteq {{\

in different context in [@mr:fptts]), $s_0$-set : set 2^\omega$ such for any perfect perfect $Q\subseteq P$ such $A\cap Q={\varnothing}$ ([@em:cfscce]), : i.e. a set $A\subseteq 2^\omega$ that for every Silver perfect set $P$, there exists a Silver perfect set P$ such that $Q\cap A={\varnothing}$ (see [@mkantw:ssrtfn]), $l_0$-set : i.e. a set $A\subseteq such for Laver set $P$, there exists a Laver perfect set $Q\subseteq P$ such that $Q\cap A={\varnothing}$ (see [@mktw:ssrtfn]), : i.e. a set $A\subseteq \omega^\omega$ such that every Miller perfect set there exists a Miller perfect $Q\subseteq such that A={\varnothing}$ [@mktw:ssrtfn]). Selection principles If and ${{\mathcal{B}}}$ are families of covers of a topological space $X$, then $X$ has [[**$S_1({{\mathcal{A}}},{{\mathcal{B}}})$ principle**]{}]{} if every sequence there exists U_n\colon with for all $n\in\omega$ ${{\mathcal{U}}}\in{{\mathcal{B}}}$. $X$ has [[**$U_{<\omega}({{\mathcal{A}}},{{\mathcal{B}}})$ principle**]{}]{} if $\langle {{\mathcal{U}}}_n\rangle_{n\in\omega}\in{{\mathcal{A}}}^\omega$ such that for every $n\in\omega$ if {{\

in different context in [@mr:fptTs]), $s_0$-set : i.e. a sEt $A\suBseTeq 2^\OmEga$ sUch tHat for any perfeCT set $p$ there exists a perfect seT $Q\subSeTEq P$ sUCh That $A\Cap Q={\varNOtHINg}$ ([@eM:cFsCce]), $V_0$-sET : i.E. a set $a\suBseteq 2^\oMega$ such thAt fOr Every Silver pERfEct set $P$, theRe eXists a Silver PerFect seT $Q\SubSEteq P$ SucH that $q\cap A={\vARnothiNg}$ (see [@mkanTw:SSrtfn]), $l_0$-SEt : i.e. a seT $a\SuBsetEq \omega^\omega$ such tHAt FOr every Laver peRfect sEt $p$, ThERE exIstS a  Laver perFeCt set $q\SubseteQ p$ sUCH ThaT $q\cap A={\varnothiNg}$ (see [@mktw:ssRTfn]), $M_0$-set : i.e. A sEt $A\SUbseteQ \omegA^\oMEga$ Such that for EverY Miller peRfect sET $P$, there EXists a  MIller pErfEct Set $Q\SUbSeTeq p$ sUCh tHAt $q\caP a={\vaRnothing}$ (SeE [@mKtw:ssRtfn]). ### sELECtioN prInciPles IF ${{\mathcal{A}}}$ and ${{\mAthCal{B}}}$ ARe fAmiliEs of cOverS oF a topOlogicAl spaCe $x$, then $X$ has [[**$S_1({{\mathcAl{A}}},{{\mAthcal{B}}})$ prIncIpLe**]{}]{} iF fOr eveRY sequeNce $\LanGle {{\mathCal{U}}}_n\raNGle_{N\iN\OMEgA}\in{{\mathcal{A}}}^\omega$, thErE EXiSts ${{\mathcAl{U}}}=\{ U_n\cOLoN n\IN\omega\}$ wiTh $u_n\iN{{\matHCAl{U}}}_n,$ fOr alL $N\iN\omega$ suCh that ${{\MAtHcAl{U}}}\in{{\maThCal{B}}}$. $X$ hAs [[**$u_{<\omEga}({{\MathcAL{A}}},{{\maThcal{B}}})$ PrinciplE**]{}]{} if foR Every sequence $\lANgle {{\mathcal{U}}}_n\RAnGLE_{n\IN\omeGa}\iN{{\mathcal{A}}}^\omEga$ sUCh thAt foR EvEry $N\In\omeGa$ if ${{\mAtHCaL{w}}}\subseteq {{\

in different context in [ @mr:fptts] ), $ s_0 $-s et : i. e. a set $A\su b sete q 2^\omega$ such thatfor a ny perf e ct set$P$ the r ee x ist saper fe c tset $ Q\s ubseteq P$ such t hat $ A\cap Q={\va r no thing}$ ([ @em :cfscce]), $v_ 0$-set : i.e.a s et $A \subse t eq 2^\ omega$ su ch that f o r every S il verperfect set $P$,t he r e exists a Sil ver pe rf e ct s et$Q\ subseteq P $sucht hat $Q\ c ap A = {\v a rnothing}$ (s ee [@mkantw : ssr tfn]), $l_ 0 $-set : i. e . a  set $A\sub sete q \omega^ \omega $ such t h at foreveryLav erperf e ct s et$P $ , t h er e e x ist s a  Lav er p erfec t se t $ Q \sub set eq P $ suc h that $Q\cap A= {\va r not hing} $ (se e [@ mk tw:ss rtfn]) , $m _0 $-set : i.e. a s et $A\sub set eq \o me ga^\o m ega$ s uch th at forevery M i lle rp e r fe ct set $P$, thereex i s ts a  Mill er per f ec ts et $Q\su bs ete q P$ s uch t hat$ Q\ cap A={\ varnot h in g} $ (see[@ mktw:s sr tfn ]). ### Sele ctionprincipl es I f ${{\mathcal{A } }}$ and ${{\m a th c a l{ B }}}$ ar e familiesof c o vers ofa  t opo l ogica l spa ce $X $ , then $X$ has [[** $S _1({{\ mathc al{A}}},{{\ma thcal{B}}} ) $ principl e**] { }] { } if for every sequ ence $\lan g le {{\ma thcal {U}}}_n\ rangle_{n \ i n\omega} \in {{\ mat hca l { A} }}^\omega$, t h e re e xi sts ${{ \ma thcal{U }}} =\{ U_ n\c ol on n\in\o mega\}$wi th $ U_ n\i n{{\m a thcal{U} }} _n, $for all$ n\in\o mega$ suc hth a t $ {{\math c al { U }}}\ in {{ \mat hca l{ B}}}$ . $X $ ha s [[**$ U_{<\omeg a}( { {\ma th ca l{A}}}, {{\mathcal{B} }} )$ princip le **] {}]{}i f for eve ry sequence $\langle {{ \ mathcal {U} }}_n\ rang le_{n\in\ ome ga}\in {{\ m athcal {A}}}^ \omeg a$ su c h that f or ev er y $n\in\om e g a$if ${ {\ math cal{W}} }\subseteq {{\

in_different context_in [@mr:fptts]), $s_0$-set : i.e._a set $A\subseteq_2^\omega$_such that_for_any perfect set_$P$ there exists_a perfect set $Q\subseteq P$_such that $A\cap_Q={\varnothing}$_([@em:cfscce]), $v_0$-set : i.e. a set $A\subseteq 2^\omega$ such that for every Silver perfect set_$P$,_there exists_a Silver_perfect_set $Q\subseteq P$ such that_$Q\cap A={\varnothing}$ (see [@mkantw:ssrtfn]), $l_0$-set : _ i.e._a set $A\subseteq \omega^\omega$ such that for every Laver_perfect_set $P$, there_exists a  Laver perfect set $Q\subseteq P$ such that_$Q\cap A={\varnothing}$ (see [@mktw:ssrtfn]), $m_0$-set : _i.e. a set $A\subseteq_\omega^\omega$_such_that for every Miller_perfect set $P$, there exists a _Miller perfect set $Q\subseteq P$ such_that $Q\cap A={\varnothing}$ (see [@mktw:ssrtfn]). ### Selection principles If_${{\mathcal{A}}}$ and ${{\mathcal{B}}}$ are families of_covers of a topological space $X$,_then $X$_has [[**$S_1({{\mathcal{A}}},{{\mathcal{B}}})$ principle**]{}]{} if for_every sequence $\langle_{{\mathcal{U}}}_n\rangle_{n\in\omega}\in{{\mathcal{A}}}^\omega$, there_exists ${{\mathcal{U}}}=\{ U_n\colon_n\in\omega\}$ with $U_n\in{{\mathcal{U}}}_n,$ for all $n\in\omega$_such that ${{\mathcal{U}}}\in{{\mathcal{B}}}$._$X$ has [[**$U_{<\omega}({{\mathcal{A}}},{{\mathcal{B}}})$ principle**]{}]{} if for_every_sequence $\langle {{\mathcal{U}}}_n\rangle_{n\in\omega}\in{{\mathcal{A}}}^\omega$_such_that_for every_$n\in\omega$ if ${{\mathcal{W}}}\subseteq_{{\

x,y,\ldots$ and continuation variables $k,l,\ldots$. 3ex \[adx54\] The [*terms*]{} $M$ and [*jumps*]{} $J$ of the CCV $\lambda \mu$-calculus are defined by the following syntax: 2ex 2ex where $x$ ranges over ordinary variables and $k$ over continuation variables. We call the prefix $[k]$ a [*jumper*]{}. 1ex The terms and jumps are syntactically separated. For example, $\lambda x.\,[k]M$ is disallowed. The lambda-abstraction and the mu-abstraction bind the abstracted variables as usual. Moreover, the let-sentences $\uplet LxM$ and $\uplet JxM$ bind the variable $x$ the scopes of which are $L$ and $J$. Definition of the set ${\it FV}(M)$ or ${\it FV}(J)$ of free (ordinary and continuation) variables is standard. For example, ${\it FV}(\uplet MxN)$ is given by $({\it FV}(M)-\{x\})\cup {\it FV}(N)$. The terms and the jumps obey ordinary $\alpha$-conversion rules. For example, $\uplet LxM$ equals $\uplet {L\{y/x\}}yM$ if $y$ is not free in $L$. 3ex \[eua38\] (1) We use curly braces for substitution to avoid too much of the overloading of the square brackets. We use two ways of substitution notation, $\{M/x\}$ and $\{x\mapsto M\}$. We select either from the aspect of readability. (2) We write $x\in M$ in place of $x\in {\rm FV}(M)$ for simplicity. 1ex As usual, we regard $\alpha$-convertible terms to be equal. In addition to this, we introduce new identification rules between terms involving the let notation. 3ex \[yoq54\] We assume the following equality axioms: 2ex 2ex The first rule is the associativity of let-constructs, the second the commutativity between $\mu$ and let, and the third the commutativity between jumper and let. To be precise, the first equality remains

x, y,\ldots$ and continuation variables $ k, l,\ldots$. 3ex \[adx54\ ] The [ * terms * ] { } $ M$ and [ * jumps * ] { } $ J$ of the CCV $ \lambda \mu$-calculus are define by the following syntax: 2ex 2ex where $ x$ compass over ordinary variable star and $ k$ over sequel variables. We call the prefix $ [ k]$ a [ * jumper * ] { }. 1ex The condition and jumps are syntactically separated. For model, $ \lambda x.\,[k]M$ is disallowed. The lambda - abstraction and the mu - abstractedness bind the abstracted variables as usual. furthermore, the let - sentences $ \uplet LxM$ and $ \uplet JxM$ oblige the variable $ x$ the scopes of which are $ L$ and $ J$. Definition of the fit $ { \it FV}(M)$ or $ { \it FV}(J)$ of free (ordinary and continuation) variable is standard. For example, $ { \it FV}(\uplet MxN)$ is given by $ ({ \it FV}(M)-\{x\})\cup { \it FV}(N)$. The terms and the jumps obey average $ \alpha$-conversion rules. For exercise, $ \uplet LxM$ equals $ \uplet { L\{y / x\}}yM$ if $ y$ is not free in $ L$. 3ex \[eua38\ ] (1) We use curly brace for substitution to avoid too much of the overloading of the square brackets. We use two way of substitution notation, $ \{M / x\}$ and $ \{x\mapsto M\}$. We select either from the aspect of readability. (2) We write $ x\in M$ in place of $ x\in { \rm FV}(M)$ for simplicity. 1ex As usual, we regard $ \alpha$-convertible terms to be adequate. In accession to this, we introduce modern designation rules between terms involving the lease notation. 3ex \[yoq54\ ] We assume the following equality axioms: 2ex 2ex The inaugural rule is the associativity of let - constructs, the second the commutativity between $ \mu$ and let, and the third the commutativity between jumper and let. To be precise, the beginning equality remains

x,y,\lfots$ and continuation vaviables $k,l,\ldots$. 3er \[qdx54\] Thx [*terms*]{} $M$ and [*jjmps*]{} $J$ of the CCV $\lambda \mu$-celcuous aee defined by the folluwing synnax: 2ex 2ex wyere $z$ ranges otsr ordikcry vzviablzs and $k$ over coktinuation eariables. We cdlu che prefix $[k]$ a [*jumper*]{}. 1ex The terms and jumps srf syntacticalli sepswates. For example, $\lambda x.\,[k]M$ is disallkwed. Tht lambda-abstractiom and the mu-abstraction bijd tje abstracted variwbles as usozl. Iireover, the uet-sentencts $\uplet LxM$ and $\uplet JxM$ bind the variabld $x$ tke scopes od qhifv are $L$ and $J$. Desinition of bne set ${\it FV}(M)$ or ${\it FV}(J)$ of nree (mrdunary and continuatioi) variables is standwrd. For efajple, ${\it FV}(\uplet MzN)$ is ghven by $({\kr FX}(M)-\{x\})\du' {\if FV}(N)$. Hhe terms and the jumps ibey ordinary $\alpha$-vognrrsion rules. For evaiple, $\uplet LxM$ equals $\uplet {L\{y/x\}}yM$ if $y$ ps nkt free in $L$. 3ex \[eua38\] (1) We usw curly braces for suhstitutiog to avoid too much of the overloading of the squdre bcazkeus. We urw hwo ways of substitution notation, $\{M/x\}$ and $\{x\mapffo M\}$. We select eitmer from the aspecy lf teadability. (2) We write $x\ih M$ in place of $x\ij {\rm FV}(I)$ for simpliciuy. 1ex Ax usual, we regard $\alpha$-convwrtible termf to be equal. In adbition to thns, we ontrocuce new identification rulss between herms invkuving the let nogatpon. 3ef \[yoq54\] We assume the followigg equalivy axnoms: 2ex 2ex The first rule is tje associativity of let-clnstrocts, tve second hhe commutativity between $\mu$ anv let, and the tvirg the coimutabivity between tumper and let. To be 'recisd, the firsn equalitb remains

x,y,\ldots$ and continuation variables $k,l,\ldots$. 3ex \[adx54\] $M$ [*jumps*]{} $J$ the CCV $\lambda following 2ex 2ex where ranges over ordinary and $k$ over continuation variables. We the prefix $[k]$ a [*jumper*]{}. 1ex The terms and jumps are syntactically separated. example, $\lambda x.\,[k]M$ is disallowed. The lambda-abstraction and the mu-abstraction bind the abstracted as Moreover, let-sentences LxM$ and $\uplet JxM$ bind the variable $x$ the scopes of which are $L$ and $J$. of the set ${\it FV}(M)$ or ${\it FV}(J)$ free (ordinary and continuation) is standard. For example, ${\it MxN)$ given by FV}(M)-\{x\})\cup FV}(N)$. terms and the obey ordinary $\alpha$-conversion rules. For example, $\uplet LxM$ equals $\uplet {L\{y/x\}}yM$ if $y$ is not free in 3ex \[eua38\] use curly for to too much of of the square brackets. We use substitution notation, $\{M/x\}$ and $\{x\mapsto M\}$. We select from the of readability. (2) We write $x\in in place of $x\in {\rm FV}(M)$ for simplicity. As usual, we regard $\alpha$-convertible terms to be equal. In addition to this, we introduce rules between terms involving let notation. 3ex We the equality 2ex 2ex first rule is the associativity of let-constructs, the second the commutativity $\mu$ and let, and the third the commutativity between jumper To precise, the first remains

x,y,\ldots$ and continuation varIables $k,l,\ldOts$. 3ex \[Adx54\] the [*TeRms*]{} $M$ And [*jUmps*]{} $J$ of the CCV $\lAMbda \Mu$-calculus are defined by The foLlOWing SYnTax: 2ex 2Ex where $X$ RaNGEs oVeR oRdiNaRY vAriabLes And $k$ oveR continuatIon VaRiables. We calL ThE prefix $[k]$ a [*jUmpEr*]{}. 1ex The terms And Jumps aRe SynTActicAllY sepaRated. FOR exampLe, $\lambda x.\,[K]M$ IS disalLOwed. The LAMbDa-abStraction and the mu-ABsTRaction bind the AbstraCtED vARIabLes As usual. MorEoVer, thE Let-sentENcES $\UPleT lxM$ and $\uplet Jxm$ bind the varIAblE $x$ the sCoPes OF which Are $L$ aNd $j$. defInition of thE set ${\It FV}(M)$ or ${\it fV}(J)$ of fREe (ordinARy and coNtinuaTioN) vaRiabLEs Is StaNdARd. FOR eXamPLe, ${\iT FV}(\uplet mxn)$ iS giveN by $({\iT fv}(m)-\{X\})\cup {\It Fv}(N)$. ThE termS and the jumps oBey OrdiNAry $\Alpha$-ConveRsioN rUles. FOr examPle, $\upLeT LxM$ equals $\uplet {l\{y/x\}}ym$ if $y$ is not FreE iN $L$. 3eX \[eUa38\] (1) We uSE curly BraCes For subsTitutioN To aVoID TOo Much of the overloadiNg OF ThE square bRacketS. we UsE Two ways oF sUbsTituTIOn notAtioN, $\{m/x\}$ And $\{x\mapsTo M\}$. We sELeCt Either fRoM the asPeCt oF reAdabiLIty. (2) WE write $X\in M$ in plAce of $X\In {\rm FV}(M)$ for simpLIcity. 1ex As usuaL, We REGaRD $\alpHa$-cOnvertible tErms TO be eQual. iN aDdiTIon to This, wE iNTrODuce new identificatiOn Rules bEtweeN terms involviNg the let noTATIon. 3ex \[yoq54\] we asSUmE The following eqUalitY axioms: 2ex 2eX the first Rule iS the assoCiativity OF Let-constRucTs, tHe sEcoND ThE commutativitY BEtweEn $\Mu$ and leT, anD the thiRd tHe cOmmUtaTiVity betweEn jumper AnD lEt. to Be pRecisE, The first EqUalItY reMains

x,y,\ldots$ and continuati on variabl es $k ,l, \ld ot s$. 3ex \[adx54\] Th e [*t erms*]{} $M$ and [*jum ps*]{ }$ J$ o f t he CC V $\lam b da \mu $- ca lcu lu s a re de fin ed by t he followi ngsy ntax: 2ex 2 ex where $x$ ra nges over or din ary va ri abl e s and $k $ ove r cont i nuatio n variabl es . We ca l l the p r e fi x $[ k]$ a [*jumper*]{ } .1ex The terms and j um p sa r e s ynt acticallyse parat e d. Fore xa m p l e,$ \lambda x.\,[ k]M$ is dis a llo wed. T he la m bda-ab strac ti o n a nd the mu-a bstr action bi nd the abstrac t ed vari ablesasusu al.M or eo ver ,t hel et -se n ten ces $\up le tLxM$and$ \ u p letJxM $ bi nd th e variable $x $ t he s c ope s ofwhich are $ L$ an d $J$. Defi ni tion of the set ${\ it FV}(M) $ o r${\ it FV}( J )$ offre e ( ordinar y and c o nti nu a t i on ) variables is sta nd a r d. For exa mple,$ {\ it FV}(\upl et Mx N)$i s give n by $( {\it FV} (M)-\{ x \} )\ cup {\i tFV}(N) $. Th e t ermsa nd t he jum ps obeyordin a ry $\alpha$-co n version rules . F o r e x ampl e,$\uplet LxM $ eq u als$\up l et {L \ {y/x\ }}yM$ i f $ y $ is not free in $L $. 3ex \[eu a38\] (1) We use curly b r aces for sub s ti t ution to avoid toomuch of th e overloa dingof the s quare bra c k ets. Weuse tw o w ays o fsubstitutionn o tati on , $\{M/ x\} $ and $ \{x \ma pst o M \} $. We sel ect eith er f ro mthe aspe c t of rea da bil it y. (2)W e writ e $x\ in M $in pla ce of $ x \i n {\rm F V} (M)$ fo rsimpl icit y . 1ex As usual, w e r e gard $ \a lpha$-c onvertible te rm s to be eq ua l.In add i t ion to t his, we introduce new i d entific ati on ru lesbetween t erm s invo lvi n g thelet no tatio n. 3 e x \[y o q 54 \]We assume th e fol lowin gequa lity ax ioms: 2ex 2ex Th e fi rst rule is t heasso c i at ivi t yo f l et - con s t ructs, the seco nd the com mu t at ivity betw e en$\ mu$ and let, a nd th e thirdthe commu tativitybe twee n jum per and le t. To be precise, the f i rs t equ ali ty rem ai ns

x,y,\ldots$ and_continuation variables_$k,l,\ldots$. 3ex \[adx54\] The [*terms*]{} $M$_and [*jumps*]{}_$J$_of the_CCV_$\lambda \mu$-calculus are_defined by the_following syntax: 2ex 2ex where $x$_ranges over ordinary_variables_and $k$ over continuation variables. We call the prefix $[k]$ a [*jumper*]{}. 1ex The terms and_jumps_are syntactically_separated._For_example, $\lambda x.\,[k]M$ is disallowed._The lambda-abstraction and the mu-abstraction_bind the_abstracted variables as usual. Moreover, the let-sentences $\uplet_LxM$_and $\uplet JxM$_bind the variable $x$ the scopes of which are_$L$ and $J$. Definition of the_set ${\it FV}(M)$_or_${\it_FV}(J)$ of free (ordinary_and continuation) variables is standard. For_example, ${\it FV}(\uplet MxN)$ is given_by $({\it FV}(M)-\{x\})\cup {\it FV}(N)$. The terms_and the jumps obey ordinary $\alpha$-conversion_rules. For example, $\uplet LxM$_equals $\uplet_{L\{y/x\}}yM$ if $y$ is not_free in $L$. 3ex \[eua38\] (1)_We use_curly braces for_substitution to avoid too much of_the overloading of_the square brackets. We use two_ways_of substitution notation,_$\{M/x\}$_and_$\{x\mapsto M\}$._We select either_from_the aspect_of_readability. (2) We write $x\in M$ in_place_of $x\in {\rm FV}(M)$ for simplicity. 1ex As usual,_we regard $\alpha$-convertible terms_to_be equal. In addition_to this, we introduce new_identification rules between terms involving the_let notation. 3ex \[yoq54\]_We assume_the following equality axioms: 2ex 2ex The first rule is the associativity of_let-constructs, the second the commutativity between_$\mu$ and let, and_the third_the_commutativity between jumper_and_let. To_be precise, the first equality remains

frac {3-{\alpha}}{q}-2+3{\alpha}}{\| u \|}_{L^{q}({\widehat{\mathbf A}}_R)}^{2-{\delta}}+ R^{-1+2{\alpha}} {\| u \|}_{L^{q}({\widehat{\mathbf A}}_R)}^{3-{\delta}}.$$ By Young’s inequality, $$|I_2+I_3| \le \frac12 {\| {{\nabla}}(u\zeta) \|}_{2}^2 + C J ^{\frac1{1-{\delta}/2}}.$$ Thus [$$\begin{split} \label{eq3-7} \int |{{\nabla}}(u\zeta)|^2 {\lesssim}{\left( R^{\frac {1+{\alpha}}2+\frac{{\alpha}-3}{q}} {\| u \|}_{L^{q}({\widehat{\mathbf A}}_R)} \right)}^2 + J ^{\frac1{1-{\delta}/2}}. \end{split}$$]{} To make the right side go to zero, it suffices to find a sequence $R_j\to \infty$, $j\in {{\mathbb N}}$, such that [$$\begin{split} \label{eq3-8} R_j^{\beta}{\| u \|}_{L^{q}( {\widehat{\mathbf A}}_{R_j})} \to 0, \end{split}$$]{} where $ {\widehat{\mathbf A}}_{R_j} = \{ x \in {{\mathbb R }}^3:\ R_j<|x|<R_j(1+8{\sigma}_0R_j^{-{\alpha}})\}$ and $${\beta}= {\beta}({\delta},{\alpha})=\max {\left\{ \frac {1+{\alpha}}2+\frac{{\alpha}-3}{q},\ \frac{\frac {3-{\alpha}}{q}-2+3{\alpha}}{2-{\delta}},\ \frac {-1+2{\alpha}}{3-{\delta}} \right\}}.$$ Note that the first argument is never greater than the second for ${\alpha}\ge0$ and ${\delta}\in [0,1]$, and they equal only if ${\alpha}=0$. Thus $${\beta}({\delta},{\alpha})= \max {\left\{ \frac{\frac {3-{\alpha}}{q}-2+3{\alpha}}{2-{\delta}},\ \frac {-1+2{\alpha}}{3-{\delta}} \right\}}.$$ If is true, then by,

frac { 3-{\alpha}}{q}-2 + 3{\alpha}}{\| u \|}_{L^{q}({\widehat{\mathbf A}}_R)}^{2-{\delta}}+ R^{-1 + 2{\alpha } } { \| u \|}_{L^{q}({\widehat{\mathbf A}}_R)}^{3-{\delta}}.$$ By Young ’s inequality, $ $ |I_2+I_3| \le \frac12 { \| { { \nabla}}(u\zeta) \|}_{2}^2 + C J ^{\frac1{1-{\delta}/2}}.$$ Thus [ $ $ \begin{split } \label{eq3 - 7 } \int |{{\nabla}}(u\zeta)|^2 { \lesssim}{\left (R^{\frac { 1+{\alpha}}2+\frac{{\alpha}-3}{q } } { \| u \|}_{L^{q}({\widehat{\mathbf A}}_R) } \right)}^2 + J ^{\frac1{1-{\delta}/2 } }. \end{split}$$ ] { } To make the proper slope go to zero, it suffices to find a succession $ R_j\to \infty$, $ j\in { { \mathbb N}}$, such that [ $ $ \begin{split } \label{eq3 - 8 } R_j^{\beta}{\| u \|}_{L^{q } ({ \widehat{\mathbf A}}_{R_j }) } \to 0, \end{split}$$ ] { } where $ { \widehat{\mathbf A}}_{R_j } = \ { x \in { { \mathbb R } } ^3:\ R_j<|x|<R_j(1 + 8{\sigma}_0R_j^{-{\alpha}})\}$ and $ $ { \beta}= { \beta}({\delta},{\alpha})=\max { \left\ { \frac { 1+{\alpha}}2+\frac{{\alpha}-3}{q},\ \frac{\frac { 3-{\alpha}}{q}-2 + 3{\alpha}}{2-{\delta}},\ \frac { -1 + 2{\alpha}}{3-{\delta } } \right\}}.$$ Note that the first argument is never greater than the second for $ { \alpha}\ge0 $ and $ { \delta}\in [ 0,1]$, and they adequate entirely if $ { \alpha}=0$. Thus $ $ { \beta}({\delta},{\alpha})= \max { \left\ { \frac{\frac { 3-{\alpha}}{q}-2 + 3{\alpha}}{2-{\delta}},\ \frac { -1 + 2{\alpha}}{3-{\delta } } \right\}}.$$ If is truthful, then by,

fraf {3-{\alpha}}{q}-2+3{\alpha}}{\| u \|}_{L^{q}({\widehxt{\mathbf A}}_R)}^{2-{\deltc}}+ R^{-1+2{\alphe}} {\| u \|}_{L^{s}({\widehat{\oathbf A}}_R)}^{3-{\delta}}.$$ By Young’s ineqnaliry, $$|I_2+I_3| \le \frac12 {\| {{\nabla}}(u\zeta) \|}_{2}^2 + C J ^{\frac1{1-{\felta}/2}}.$$ Thys [$$\btgin{split} \label{eq3-7} \mht |{{\nabla}}(u\zeta)|^2 {\mcsssik}{\oeft( R^{\frac {1+{\alpma}}2+\frac{{\alpha}-3}{x}} {\| u \|}_{L^{q}({\widehad{\mxtkbf A}}_R)} \right)}^2 + J ^{\frac1{1-{\delta}/2}}. \end{split}$$]{} Tj make yhf right side gj to sero, pt suffices to find a sequence $D_j\to \inhty$, $j\in {{\mathbb M}}$, such that [$$\begin{split} \labfl{eq3-8} G_j^{\beta}{\| u \|}_{L^{q}( {\widehwt{\mathbf A}}_{R_h})} \to 0, \end{split}$$]{} whdre $ {\widehat{\mathbf A}}_{R_j} = \{ x \in {{\mathbb R }}^3:\ R_j<|x|<R_j(1+8{\sigma}_0R_j^{-{\xlpha}})\}$ and $${\beta}= {\bwta}({\fglta},{\alpha})=\max {\left\{ \frac {1+{\alpha}}2+\ngac{{\alphd}-3}{q},\ \frac{\grac {3-{\alpha}}{q}-2+3{\alpma}}{2-{\delva}},\ \feac {-1+2{\alpha}}{3-{\delta}} \right\}}.$$ Iote that the first wrgument hs never greater thqn the vecotd fue ${\auphz}\gx0$ ahd ${\delha}\ii [0,1]$, and they equal only if ${\alpha}=0$. Thus $${\beta}({\cejns},{\alpha})= \max {\lsft\{ \srwc{\frac {3-{\alpha}}{q}-2+3{\alpha}}{2-{\delta}},\ \frac {-1+2{\alpha}}{3-{\delta}} \rifht\}}.$$ If is true, then by,

frac {3-{\alpha}}{q}-2+3{\alpha}}{\| u \|}_{L^{q}({\widehat{\mathbf A}}_R)}^{2-{\delta}}+ R^{-1+2{\alpha}} {\| A}}_R)}^{3-{\delta}}.$$ Young’s inequality, \le \frac12 {\| ^{\frac1{1-{\delta}/2}}.$$ [$$\begin{split} \label{eq3-7} \int {\lesssim}{\left( R^{\frac {1+{\alpha}}2+\frac{{\alpha}-3}{q}} u \|}_{L^{q}({\widehat{\mathbf A}}_R)} \right)}^2 + J \end{split}$$]{} To make the right side go to zero, it suffices to find sequence $R_j\to \infty$, $j\in {{\mathbb N}}$, such that [$$\begin{split} \label{eq3-8} R_j^{\beta}{\| u \|}_{L^{q}( A}}_{R_j})} 0, where {\widehat{\mathbf A}}_{R_j} = \{ x \in {{\mathbb R }}^3:\ R_j<|x|<R_j(1+8{\sigma}_0R_j^{-{\alpha}})\}$ and $${\beta}= {\beta}({\delta},{\alpha})=\max {\left\{ \frac {1+{\alpha}}2+\frac{{\alpha}-3}{q},\ {3-{\alpha}}{q}-2+3{\alpha}}{2-{\delta}},\ \frac {-1+2{\alpha}}{3-{\delta}} \right\}}.$$ Note that the first is never greater than second for ${\alpha}\ge0$ and ${\delta}\in and equal only ${\alpha}=0$. $${\beta}({\delta},{\alpha})= {\left\{ \frac{\frac {3-{\alpha}}{q}-2+3{\alpha}}{2-{\delta}},\ {-1+2{\alpha}}{3-{\delta}} \right\}}.$$ If is true, then by,

frac {3-{\alpha}}{q}-2+3{\alpha}}{\| u \|}_{L^{q}({\widehat{\Mathbf A}}_R)}^{2-{\deLta}}+ R^{-1+2{\aLphA}} {\| u \|}_{L^{Q}({\wIdehAt{\maThbf A}}_R)}^{3-{\delta}}.$$ By YOUng’s Inequality, $$|I_2+I_3| \le \frac12 {\| {{\nablA}}(u\zetA) \|}_{2}^2 + C j ^{\Frac1{1-{\DElTa}/2}}.$$ ThuS [$$\begin{sPLiT} \LAbeL{eQ3-7} \iNt |{{\nAbLA}}(u\Zeta)|^2 {\lEssSim}{\left( r^{\frac {1+{\alpha}}2+\FraC{{\aLpha}-3}{q}} {\| u \|}_{L^{q}({\wideHAt{\Mathbf A}}_R)} \riGht)}^2 + j ^{\frac1{1-{\delta}/2}}. \enD{spLit}$$]{} To mAkE thE Right SidE go to Zero, it SUfficeS to find a sEqUEnce $R_j\TO \infty$, $j\IN {{\MaThbb n}}$, such that [$$\begin{splIT} \lABel{eq3-8} R_j^{\beta}{\| u \|}_{L^{q}( {\WidehaT{\mAThBF a}}_{R_j})} \To 0, \eNd{split}$$]{} wheRe $ {\WidehAT{\mathbf a}}_{r_j} = \{ X \IN {{\MatHBb R }}^3:\ R_j<|x|<R_j(1+8{\sigma}_0r_j^{-{\alpha}})\}$ and $${\bETa}= {\bEta}({\delTa},{\AlpHA})=\max {\leFt\{ \fraC {1+{\aLPha}}2+\Frac{{\alpha}-3}{q},\ \fRac{\fRac {3-{\alpha}}{q}-2+3{\Alpha}}{2-{\dELta}},\ \frac {-1+2{\ALpha}}{3-{\delTa}} \righT\}}.$$ NoTe tHat tHE fIrSt aRgUMenT Is NevER grEater thaN tHe SeconD for ${\ALPHA}\ge0$ aNd ${\dElta}\In [0,1]$, and They equal only If ${\aLpha}=0$. tHus $${\Beta}({\dElta},{\aLpha})= \MaX {\left\{ \Frac{\frAc {3-{\alpHa}}{Q}-2+3{\alpha}}{2-{\delta}},\ \frac {-1+2{\AlphA}}{3-{\delta}} \rigHt\}}.$$ IF iS trUe, Then bY,

frac {3-{\alpha}}{q}-2+3{\ alpha}}{\| u \ |}_ {L^ {q }({\ wide hat{\mathbf A} } _R)} ^{2-{\delta}}+ R^{-1+2 {\alp ha } } {\ | u \|} _{L^{q} ( {\ w i deh at {\ mat hb f A }}_R) }^{ 3-{\del ta}}.$$ B y Y ou ng’s inequal i ty , $$|I_2+I _3| \le \frac12 {\ | {{\n ab la} } (u\ze ta) \|}_ {2}^2+ C J^{\frac1{ 1- { \delta } /2}}.$$ T hu s [$ $\begin{split} \l a be l {eq3-7} \int | {{\nab la } }( u \ zet a)| ^2 {\lesss im }{\le f t( R^{\ f ra c { 1+{ \ alpha}}2+\fra c{{\alpha}- 3 }{q }} {\| u \ | }_{L^{ q}({\ wi d eha t{\mathbf A }}_R )} \righ t)}^2+ J ^{\f r ac1{1-{ \delta }/2 }}. \e n d{ sp lit }$ $ ]{} To ma k e t he right s id e goto z e r o , itsuf fice s tofind a sequen ce$R_j \ to\inft y$, $ j\in { {\mat hbb N} }$, s uc h that [$$\begi n{sp lit} \lab el{ eq 3-8 }R_j^{ \ beta}{ \| u\|}_{L^ {q}( {\ w ide ha t { \ ma thbf A}}_{R_j})} \ to 0 , \end{sp lit}$$ ] {} w h ere $ {\ wi deh at{\ m a thbfA}}_ { R_ j} = \{x \in{ {\ ma thbb R}} ^3:\ R_ j<| x|< R_j(1 + 8{\s igma}_ 0R_j^{-{ \alph a }})\}$ and $${ \ beta}= {\bet a }( { \ de l ta}, {\a lpha})=\max {\l e ft\{ \f r ac {1 + {\alp ha}}2 +\ f ra c {{\alpha}-3}{q},\ \ fr ac{\fr ac {3 -{\alpha}}{q} -2+3{\alph a } } {2-{\del ta}} , \\ frac {-1+2{\al pha}} {3-{\delta } } \right \}}.$ $ Note t hat the f i r st argum ent is ne ver g re ater than the s econ dfor ${\ alp ha}\ge0 $ a nd${\ del ta }\in [0,1 ]$, andth ey e qu alonlyi f ${\alp ha }=0 $. Th us $$ { \beta} ({\de lta} ,{ \a l pha })= \ma x { \ l eft\ { \fra c{\ fr ac {3 -{\a l pha }}{q}-2 +3{\alpha }}{ 2 -{\d el ta }},\ \f rac {-1+2{\al ph a}}{3-{\de lt a}} \righ t \ }}.$$ If is true, then by,

frac {3-{\alpha}}{q}-2+3{\alpha}}{\|_ u_\|}_{L^{q}({\widehat{\mathbf A}}_R)}^{2-{\delta}}+ R^{-1+2{\alpha}} {\|_ u_\|}_{L^{q}({\widehat{\mathbf_A}}_R)}^{3-{\delta}}.$$ By Young’s_inequality,_$$|I_2+I_3| \le \frac12_{\| {{\nabla}}(u\zeta) \|}_{2}^2_ + C J ^{\frac1{1-{\delta}/2}}.$$_Thus [$$\begin{split} \label{eq3-7} \int_|{{\nabla}}(u\zeta)|^2 {\lesssim}{\left(_R^{\frac {1+{\alpha}}2+\frac{{\alpha}-3}{q}} {\| u \|}_{L^{q}({\widehat{\mathbf A}}_R)} \right)}^2 + J ^{\frac1{1-{\delta}/2}}. \end{split}$$]{} To make_the_right side_go_to_zero, it suffices to find_a sequence $R_j\to \infty$, $j\in_{{\mathbb N}}$,_such that [$$\begin{split} \label{eq3-8} R_j^{\beta}{\| u \|}_{L^{q}( {\widehat{\mathbf_A}}_{R_j})}_\to 0, \end{split}$$]{}_where $ {\widehat{\mathbf A}}_{R_j} = \{ x \in {{\mathbb_R }}^3:\ R_j<|x|<R_j(1+8{\sigma}_0R_j^{-{\alpha}})\}$ and $${\beta}=_ {\beta}({\delta},{\alpha})=\max {\left\{__\frac_{1+{\alpha}}2+\frac{{\alpha}-3}{q},\ \frac{\frac {3-{\alpha}}{q}-2+3{\alpha}}{2-{\delta}},\ \frac_{-1+2{\alpha}}{3-{\delta}} \right\}}.$$ Note that the first_argument is never greater than the_second for ${\alpha}\ge0$ and ${\delta}\in [0,1]$, and_they equal only if ${\alpha}=0$. Thus_$${\beta}({\delta},{\alpha})= \max {\left\{ _\frac{\frac {3-{\alpha}}{q}-2+3{\alpha}}{2-{\delta}},\_\frac {-1+2{\alpha}}{3-{\delta}} \right\}}.$$ If is_true, then by,