Datasets:

pratyushmaini

/

llm_dataset_inference

like 1

; @SamarskiiB1989numerical]. The spatial grid was logarithmic, i.e. $x_{i} = \exp{(\ln x_{0} + i(\ln x_{f} - \ln x_{0}) / N)}$, $i=\{1,N\}$, while the grid in $t$ was uniform with $\tau_{k} = \tau = 0.7$. The actual number of points $N$ in the spatial grid was $\approx 2\cdot10^{3}$. In order to evaluate the accuracy of the algorithm we performed a numerical solution of the Dirac and Schrödinger equations in the Coulomb field, i.e. the Hydrogen atom and the numerical solution of the polaron problem [@Mitra198791]. The accuracy of the calculation of eigenvalues for the Hydrogen atom was greater than 6 decimal digits. Moreover, we were able to reproduce all digits of the well known result for the ground state energy of the polaron problem $E_{\mathrm{p}}=-0.108513$ [@MiyakeA1975strong; @Mitra198791], where a similar equation of the self-consistent field arises. At last we discuss the choice of the initial approximation $\bar{{\boldsymbol}X}_{0}$ for the unknown vector $\bar{{\boldsymbol}X}$. For this purpose, we use the variational estimation for the functional for the energy of the system, which is based on the following wave functions $$\begin{aligned} u_{0N}(x) = \mathrm{const}\cdot x e^{-x}, \quad v_{0N}(x) = \mathrm{const}_{1}\cdot x^{2}e^{-x},\label{eq:cont-anal-newt:18}\end{aligned}$$ for $q>0$ and $$\begin{aligned} u_{0N}(x) = \mathrm{const}\cdot x(1 + x^{2}) e^{-x}, \quad v_{0N}(x) = \mathrm{const}_{1}\cdot x^{2}e^{-x}, \label{eq:cont-anal-newt:19}\end{aligned}$$ for $q<0$. The constants $\mathrm{const}$ and $\mathrm{const}_{1}$ are chosen from the normalization condition Eq. (\[eq:cont-anal-newt:15\]). Calculation of the functional (\[eq:48\]) {#sec:calc-expect-kin

; @SamarskiiB1989numerical ]. The spatial grid was logarithmic, i.e. $ x_{i } = \exp{(\ln x_{0 } + i(\ln x_{f } - \ln x_{0 }) / N)}$, $ i=\{1,N\}$, while the grid in $ t$ was uniform with $ \tau_{k } = \tau = 0.7$. The actual number of period $ N$ in the spatial power system was $ \approx 2\cdot10^{3}$. In order to evaluate the accuracy of the algorithm we perform a numeral solution of the Dirac and Schrödinger equations in the Coulomb field, i.e. the Hydrogen atom and the numeral solution of the polaron trouble [ @Mitra198791 ]. The accuracy of the calculation of eigenvalues for the Hydrogen atom was greater than 6 decimal digits. Moreover, we were able to reproduce all digit of the well known result for the flat coat state energy of the polaron problem $ E_{\mathrm{p}}=-0.108513 $ [ @MiyakeA1975strong; @Mitra198791 ], where a similar equation of the self - reproducible field arises. At last we hash out the choice of the initial approximation $ \bar{{\boldsymbol}X}_{0}$ for the unknown vector $ \bar{{\boldsymbol}X}$. For this purpose, we practice the variational estimation for the functional for the department of energy of the system, which is based on the following wave functions $ $ \begin{aligned } u_{0N}(x) = \mathrm{const}\cdot x e^{-x }, \quad v_{0N}(x) = \mathrm{const}_{1}\cdot x^{2}e^{-x},\label{eq: cont - anal - newt:18}\end{aligned}$$ for $ q>0 $ and $ $ \begin{aligned } u_{0N}(x) = \mathrm{const}\cdot x(1 + x^{2 }) e^{-x }, \quad v_{0N}(x) = \mathrm{const}_{1}\cdot x^{2}e^{-x }, \label{eq: cont - anal - newt:19}\end{aligned}$$ for $ q<0$. The constant $ \mathrm{const}$ and $ \mathrm{const}_{1}$ are chosen from the normalization condition Eq.   (\[eq: cont - anal - newt:15\ ]). Calculation of the functional (\[eq:48\ ]) { # sec: calc - expect - kin

; @SalarskiiB1989numerical]. The spxtial grid was logaritimic, i.e. $x_{i} = \exp{(\un x_{0} + i(\ln x_{f} - \ln x_{0}) / N)}$, $i=\{1,N\}$, whipe the trid in $t$ was uniform dith $\tau_{k} = \tau = 0.7$. Rhe ectual number of points $K$ in fme spctmal grid was $\apkrox 2\cdot10^{3}$. In mrder to evaludtd che accuracy of the algorithm we persormed s jumerical solujion ps ths Dirac and Schrödinger equations ih the Cmulomb field, o.e. the Hydrogen atom and tje nkmerical solution lf the polatkn [eoblem [@Mitra198791]. The accuracy of the czlculation of eigenvalues for tfe Hybrogen atom wqs hteater than 6 deciial digits. Moreover, fe were able to reprocucx alo digits of the well nnown result for tre ground scate energy of the pooaeon ptoblek $E_{\mxrhro{p}}=-0.108513$ [@JibaksA1975strojg; @Jitra198791], whers a similar equation of the sekf-sinsistent fiemd arifef. At last we discuss the choice of the ititjal approximation $\bar{{\boodsymbol}X}_{0}$ for the unkjown vectjr $\bar{{\boldsymbol}X}$. For this purpose, we use the varhatioiau ewtlmatkin for the functional for the energy of the sysesm, wmich is based on the followonh ewve functions $$\begin{cmifned} u_{0N}(x) = \mathrm{clnst}\cdoj x e^{-x}, \quad v_{0N}(x) = \mayhrm{const}_{1}\cdot x^{2}e^{-x},\label{eq:conr-anal-newt:18}\end{clitned}$$ for $q>0$ and $$\begnn{aligned} u_{0U}(x) = \majhrm{comst}\cdot x(1 + x^{2}) e^{-x}, \quad v_{0N}(r) = \matgrm{const}_{1}\cdoh x^{2}e^{-x}, \labsu{eq:cont-anal-newt:19}\evd{akicned}$$ for $q<0$. The constants $\mwthrm{consv}$ and $\mathrm{zonsj}_{1}$ are crosen from the kmrmalization condihion Gq. (\[eq:cott-anal-newt:15\]). Falculation of the functional (\[eq:48\]) {#sec:calc-expecj-kit

; @SamarskiiB1989numerical]. The spatial grid was logarithmic, = x_{0} + x_{f} - \ln the in $t$ was with $\tau_{k} = = 0.7$. The actual number of $N$ in the spatial grid was $\approx 2\cdot10^{3}$. In order to evaluate the of the algorithm we performed a numerical solution of the Dirac and Schrödinger in Coulomb i.e. Hydrogen atom and the numerical solution of the polaron problem [@Mitra198791]. The accuracy of the calculation eigenvalues for the Hydrogen atom was greater than decimal digits. Moreover, we able to reproduce all digits the known result the state of the polaron $E_{\mathrm{p}}=-0.108513$ [@MiyakeA1975strong; @Mitra198791], where a similar equation of the self-consistent field arises. At last we discuss the of the $\bar{{\boldsymbol}X}_{0}$ for unknown $\bar{{\boldsymbol}X}$. this purpose, we variational estimation for the functional for the system, which is based on the following functions $$\begin{aligned} = \mathrm{const}\cdot x e^{-x}, \quad v_{0N}(x) \mathrm{const}_{1}\cdot x^{2}e^{-x},\label{eq:cont-anal-newt:18}\end{aligned}$$ for $q>0$ and $$\begin{aligned} u_{0N}(x) = x(1 + x^{2}) e^{-x}, \quad v_{0N}(x) = \mathrm{const}_{1}\cdot x^{2}e^{-x}, \label{eq:cont-anal-newt:19}\end{aligned}$$ for $q<0$. The constants $\mathrm{const}$ are chosen from the condition Eq. (\[eq:cont-anal-newt:15\]). of functional {#sec:calc-expect-kin

; @SamarskiiB1989numerical]. The spaTial grid waS logaRitHmiC, i.E. $x_{i} = \eXp{(\ln X_{0} + i(\ln x_{f} - \ln x_{0}) / N)}$, $i=\{1,N\}$, whILe thE grid in $t$ was uniform with $\Tau_{k} = \tAu = 0.7$. tHe acTUaL numbEr of poiNTs $n$ IN thE sPaTiaL gRId Was $\apProX 2\cdot10^{3}$. In Order to evaLuaTe The accuracy oF ThE algorithm We pErformed a numEriCal solUtIon OF the DIraC and SChrödiNGer equAtions in tHe cOulomb FIeld, i.e. tHE hyDrogEn atom and the numerICaL Solution of the pOlaron PrOBlEM [@mitRa198791]. THe accuracy Of The caLCulatioN Of EIGEnvALues for the HydRogen atom waS GreAter thAn 6 DecIMal digIts. MoReOVer, We were able tO repRoduce all Digits OF the welL Known reSult foR thE grOund STaTe EneRgY Of tHE pOlaROn pRoblem $E_{\mAtHrM{p}}=-0.108513$ [@MiyAkeA1975STRONg; @MiTra198791], WherE a simIlar equation oF thE selF-ConSisteNt fieLd arIsEs. At lAst we dIscusS tHe choice of the inItiaL approximAtiOn $\Bar{{\BoLdsymBOl}X}_{0}$ for The UnkNown vecTor $\bar{{\bOLdsYmBOL}x}$. FOr this purpose, we use ThE VArIational EstimaTIoN fOR the funcTiOnaL for THE enerGy of THe System, whIch is bASeD oN the folLoWing waVe FunCtiOns $$\beGIn{alIgned} u_{0n}(x) = \mathrm{Const}\CDot x e^{-x}, \quad v_{0N}(x) = \mAThrm{const}_{1}\cdot X^{2}E^{-x},\LABeL{Eq:coNt-aNal-newt:18}\end{aLignED}$$ for $Q>0$ and $$\BEgIn{aLIgned} U_{0N}(x) = \maThRM{cONst}\cdot x(1 + x^{2}) e^{-x}, \quad v_{0N}(x) = \mAtHrm{conSt}_{1}\cdoT x^{2}e^{-x}, \label{eq:coNt-anal-newt:19}\END{Aligned}$$ fOr $q<0$. THE cONstants $\mathrm{cOnst}$ aNd $\mathrm{coNSt}_{1}$ are choSen frOm the norMalizatioN COndition eq. (\[eQ:coNt-aNal-NEWt:15\]). calculation of THE funCtIonal (\[eq:48\]) {#Sec:Calc-expEct-Kin

; @SamarskiiB1989numerical ]. The spa tialgri d w as log arit hmic, i.e. $x_ { i} = \exp{(\ln x_{0} + i(\ ln x_ {f } - \ l nx_{0} ) / N)} $ ,$ i =\{ 1, N\ }$, w h il e the gr id in $ t$ was uni for mwith $\tau_{ k }= \tau = 0 .7$ . The actual nu mber o fpoi n ts $N $ i n the spati a l grid was $\ap pr o x 2\cd o t10^{3} $ . In o rder to evaluatet he accuracy of th e algo ri t hm w e p erf ormed a nu me rical solutio n o f t heD irac and Schr ödinger equ a tio ns inth e C o ulombfield ,i .e. the Hydrog en a tom and t he num e rical s o lutionof the po lar on p r ob le m [ @M i tra 1 98 791 ] . T he accur ac yof th e ca l c u l atio n o f ei genva lues for theHyd roge n at om wa s gre ater t han 6 decim al di gi ts. Moreover, w e we re able t o r ep rod uc e all digits of th e wellknown r e sul tf o r t he ground state en er g y o f the po laronp ro bl e m $E_{\m at hrm {p}} = - 0.108 513$ [@ MiyakeA1 975str o ng ;@Mitra1 98 791],wh ere asimil a r eq uation of theself- c onsistent fiel d arises. Atl as t we disc uss the choice oft he i niti a lapp r oxima tion$\ b ar { {\boldsymbol}X}_{0} $for th e unk nown vector $ \bar{{\bol d s y mbol}X}$ . Fo r t h is purpose, we usethe variat i onal est imati on for t he functi o n al for t heene rgy of t he system, whic h is b as ed on t hefollowi ngwav e f unc ti ons $$\be gin{alig ne d} u _{0 N}(x) = \mathr m{ con st }\c dot x e^{-x} , \qu ad v _{ 0N } (x) = \mat h rm { c onst }_ {1 }\cd otx^ {2}e^ {-x} , \la bel{eq: cont-anal -ne w t:18 }\ en d{align ed}$$ for $q> 0$ and $$\be gi n{a ligned } u_{0N} (x) = \mathrm{const}\cd o t x(1 + x^ {2})e^{- x}, \quad v_ {0N}(x ) = \mathr m{cons t}_{1 }\ cdo t x^{2} e ^ {- x}, \ label{eq:c o n t-a nal-n ew t:19 }\end{a ligned}$$ for $q<0 $ . T he constants$\m athr m { co nst } $a nd$\ m ath r m {const}_{1}$ ar e chosen f ro m t he normali z ati on condit ion Eq.  (\[e q :cont-a nal-newt: 15\]). C al cula t i onof the fun ctional(\[eq:48\ ] ) {#s e c: calc- exp ect-ki n

; @SamarskiiB1989numerical]._The spatial_grid was logarithmic, i.e._$x_{i} =_\exp{(\ln_x_{0} +_i(\ln_x_{f} - \ln_x_{0}) / N)}$,_$i=\{1,N\}$, while the grid_in $t$ was_uniform_with $\tau_{k} = \tau = 0.7$. The actual number of points $N$ in the_spatial_grid was_$\approx_2\cdot10^{3}$. In_order to evaluate the accuracy_of the algorithm we performed_a numerical_solution of the Dirac and Schrödinger equations in_the_Coulomb field, i.e._the Hydrogen atom and the numerical solution of the_polaron problem [@Mitra198791]. The accuracy of_the calculation of_eigenvalues_for_the Hydrogen atom was_greater than 6 decimal digits. Moreover,_we were able to reproduce all_digits of the well known result for_the ground state energy of the_polaron problem $E_{\mathrm{p}}=-0.108513$ [@MiyakeA1975strong; @Mitra198791],_where a_similar equation of the self-consistent_field arises. At last_we discuss_the choice of_the initial approximation $\bar{{\boldsymbol}X}_{0}$ for the_unknown vector $\bar{{\boldsymbol}X}$._For this purpose, we use the_variational_estimation for the_functional_for_the energy_of the system,_which_is based_on_the following wave functions $$\begin{aligned} _u_{0N}(x)_= \mathrm{const}\cdot x e^{-x}, \quad v_{0N}(x) =_\mathrm{const}_{1}\cdot x^{2}e^{-x},\label{eq:cont-anal-newt:18}\end{aligned}$$ for $q>0$_and_$$\begin{aligned} u_{0N}(x) =_\mathrm{const}\cdot x(1 + x^{2}) e^{-x},_\quad v_{0N}(x) = \mathrm{const}_{1}\cdot x^{2}e^{-x}, \label{eq:cont-anal-newt:19}\end{aligned}$$_for $q<0$._The constants_$\mathrm{const}$ and $\mathrm{const}_{1}$ are chosen from the normalization condition Eq. (\[eq:cont-anal-newt:15\]). Calculation of_the functional (\[eq:48\]) {#sec:calc-expect-kin

$ be an embedded symplectic leaf. If the cohomology groups $$H^2(A_S,\mathcal{S}^{k}(\nu_S^*))$$ vanish for all $k\geq 2$, then $\pi$ is formally Poisson diffeomorphic to its first order approximation around $S$. In many cases, we prove that these cohomological obstructions vanish, and we obtain the following corollaries: \[Theorem2\] Let $(M,\pi)$ be a Poisson manifold and let $S\subset M$ be an embedded symplectic leaf. Assume that the Poisson homotopy bundle of $S$ is a smooth principal bundle with vanishing second de Rham cohomology group, and that its structure group $G$ satisfies $$H^2_{\mathrm{diff}}(G,\mathcal{S}^k(\mathfrak{g}))=0,\ \ \forall\ k\geq 2,$$ where $\mathfrak{g}$ is the Lie algebra of $G$ and $H^{\bullet}_{\mathrm{diff}}(G,\mathcal{S}^k(\mathfrak{g}))$ denotes the differentiable cohomology of $G$ with coefficients in the $k$-th symmetric power of the adjoint representation. Then $\pi$ is formally Poisson diffeomorphic to its first order approximation around $S$. Differentiable cohomology of compact groups vanishes, therefore: \[Corollary\_2\] Let $(M,\pi)$ be a Poisson manifold and let $S\subset M$ be an embedded symplectic leaf. Assume that the Poisson homotopy bundle of $S$ is a smooth principal bundle with vanishing second de Rham cohomology and compact structure group. Then $\pi$ is formally Poisson diffeomorphic to its first order approximation around $S$. A bit more technical is the following: \[Theorem3\] Let $(M, \pi)$ be a Poisson manifold and let $S\subset M$ be an embedded symplectic leaf whose isotropy Lie algebra is reductive. If the abelianization algebroid $$A_S^{\mathrm{ab}}:=A_S/[\nu_S^*,\nu_S^*]$$ is integrable by a simply connected principal bundle with vanishing second de Rham cohomology and compact structure group, then $\pi$ is formally Poisson diffeomorphic to its first order approximation around $S$. \[Corollary\_3\] Let $(M, \pi)$ be a Poisson manifold and let $S\subset M$ be an embedded symplectic leaf through $x\in M$. If the isotropy Lie

$ be an embedded symplectic leaf. If the cohomology groups $ $ H^2(A_S,\mathcal{S}^{k}(\nu_S^*))$$ vanish for all $ k\geq 2 $, then $ \pi$ is formally Poisson diffeomorphic to its first decree estimate around $ S$. In many cases, we prove that these cohomological obstacle fly, and we obtain the following corollary: \[Theorem2\ ] get $ (M,\pi)$ be a Poisson manifold and let $ S\subset M$ be an embedded symplectic leaf. Assume that the Poisson homotopy bundle of $ S$ is a placid principal bundle with fly second de Rham cohomology group, and that its structure group $ G$ satisfies $ $ H^2_{\mathrm{diff}}(G,\mathcal{S}^k(\mathfrak{g}))=0,\ \ \forall\ k\geq 2,$$ where $ \mathfrak{g}$ is the Lie algebra of $ G$ and $ H^{\bullet}_{\mathrm{diff}}(G,\mathcal{S}^k(\mathfrak{g}))$ denotes the differentiable cohomology of $ G$ with coefficients in the $ k$-th symmetric power of the adjoint representation. Then $ \pi$ is formally Poisson diffeomorphic to its inaugural order approximation around $ S$. Differentiable cohomology of compact group vanishes, therefore: \[Corollary\_2\ ] Let $ (M,\pi)$ be a Poisson manifold and let $ S\subset M$ be an implant symplectic leaf. Assume that the Poisson homotopy bundle of $ S$ is a smooth principal bundle with vanish second de Rham cohomology and compact structure group. Then $ \pi$ is formally Poisson diffeomorphic to its first order approximation around $ S$. A morsel more technical is the follow: \[Theorem3\ ] Let $ (M, \pi)$ be a Poisson manifold and let $ S\subset M$ be an embedded symplectic leaf whose isotropy Lie algebra is reductive. If the abelianization algebroid $ $ A_S^{\mathrm{ab}}:=A_S/[\nu_S^*,\nu_S^*]$$ is integrable by a simply connected principal bundle with vanishing second de Rham cohomology and compendious social organization group, then $ \pi$ is formally Poisson diffeomorphic to its first order approximation around $ S$. \[Corollary\_3\ ] lease $ (M, \pi)$ be a Poisson manifold and let $ S\subset M$ be an embedded symplectic leaf through $ x\in M$. If the isotropy Lie

$ be an embedded symplectic ueaf. If the cohomology groupa $$H^2(A_S,\matfcal{S}^{k}(\nu_S^*))$$ vanish for all $k\gee 2$, then $\pi$ is formally Poissov diffeomlrphic ti itw first ordxd apprownmatikk arobnv $S$. In many casex, we prove that these covooopogical obstructions vanish, and we jbtain yhf following cotollagiqs: \[Thsorem2\] Let $(M,\pi)$ be a Poisson manifols and ltt $S\subset M$ be an embedded symplectic leaf. Wssule that the Poissoj homotopy vundjw of $S$ is a rmooth principal bundlg with vanishing second de Rham zohomplogy grouk, znf that its svructuge group $G$ sabpsfies $$V^2_{\mathrm{ciff}}(G,\mathcal{S}^k(\katifraj{g}))=0,\ \ \forall\ k\geq 2,$$ whece $\mathfrak{g}$ is the Jie algebsa of $G$ and $H^{\bullet}_{\narhrm{dhff}}(G,\kathzql{S}^y(\mauhfcak{f}))$ denohes the diffedentiable cihomology of $G$ with cjvgficients in the $k$-eh symmetric power of the adjoint represettafion. Then $\pi$ is formallt Poisson diffeomorphlc to its first order approximation around $S$. Differentiable cohojuloyn ud fompact groups vanishes, therefore: \[Corollary\_2\] Lee $(M,\ki)$ ne a Poisson manlfold and let $S\subxeh K$ be an embeddgd symplecfic leaf. Assume thwt the Koissob homotopr bumdle of $S$ is a smooth princupal bundle city vanishing second de Rham cokomoloby anc compact structure grobp. Theh $\pi$ is forlally Poiaron diffeomorphiz tp hts firsu order approximatyon arounv $S$. A yit more tecnnical is the foplowikc: \[Theorem3\] Let $(M, \pi)$ he a Koissot manifold and let $S\subset M$ be an embeddxv symplectic kedf fhose isjtropn Lie algebra if reductive. If the abzlianixation algvbroid $$A_S^{\kathrm{ab}}:=A_S/[\nt_S^*,\nu_S^*]$$ is intectable by a sikply congectwd peincipau bundle with vsnishing second de Eham cohomology anv coolact structure yxoop, then $\pi$ is fprmxllr Ioixsjt diffeomorpvic go kys fifst ordtx aipruximstion around $S$. \[Corolldry\_3\] Met $(M, \pi)$ be a Poisxok manifole and lee $S\subset M$ br an embedded symppectir leaf throigh $x\in M$. If the isotropy Lie

$ be an embedded symplectic leaf. If groups vanish for $k\geq 2$, then to first order approximation $S$. In many we prove that these cohomological obstructions and we obtain the following corollaries: \[Theorem2\] Let $(M,\pi)$ be a Poisson manifold let $S\subset M$ be an embedded symplectic leaf. Assume that the Poisson homotopy of is smooth bundle with vanishing second de Rham cohomology group, and that its structure group $G$ satisfies $$H^2_{\mathrm{diff}}(G,\mathcal{S}^k(\mathfrak{g}))=0,\ \forall\ k\geq 2,$$ where $\mathfrak{g}$ is the Lie of $G$ and $H^{\bullet}_{\mathrm{diff}}(G,\mathcal{S}^k(\mathfrak{g}))$ the differentiable cohomology of $G$ coefficients the $k$-th power the representation. Then $\pi$ formally Poisson diffeomorphic to its first order approximation around $S$. Differentiable cohomology of compact groups vanishes, therefore: Let $(M,\pi)$ Poisson manifold let M$ an embedded symplectic that the Poisson homotopy bundle of smooth principal bundle with vanishing second de Rham and compact group. Then $\pi$ is formally Poisson to its first order approximation around $S$. A more technical is the following: \[Theorem3\] Let $(M, \pi)$ be a Poisson manifold and let be an embedded symplectic whose isotropy Lie is If abelianization $$A_S^{\mathrm{ab}}:=A_S/[\nu_S^*,\nu_S^*]$$ is by a simply connected principal bundle with vanishing second de Rham and compact structure group, then $\pi$ is formally Poisson diffeomorphic first approximation around $S$. Let $(M, \pi)$ be Poisson and let $S\subset M$ embedded leaf If isotropy

$ be an embedded symplectic leaF. If the cohoMologY grOupS $$H^2(a_S,\maThcaL{S}^{k}(\nu_S^*))$$ vanish foR All $k\Geq 2$, then $\pi$ is formally PoiSson dIfFEomoRPhIc to iTs first ORdER AppRoXiMatIoN ArOund $S$. in mAny caseS, we prove thAt tHeSe cohomologiCAl ObstructioNs vAnish, and we obTaiN the foLlOwiNG coroLlaRies: \[THeorem2\] lEt $(M,\pi)$ bE a Poisson MaNIfold aND let $S\suBSEt m$ be aN embedded symplectIC lEAf. Assume that thE PoissOn HOmOTOpy BunDle of $S$ is a sMoOth prINcipal bUNdLE WIth VAnishing seconD de Rham cohoMOloGy grouP, aNd tHAt its sTructUrE GroUp $G$ satisfieS $$H^2_{\maThrm{diff}}(G,\MathcaL{s}^k(\mathfRAk{g}))=0,\ \ \foraLl\ k\geq 2,$$ WheRe $\mAthfRAk{G}$ iS thE LIE alGEbRa oF $g$ anD $H^{\bullet}_{\MaThRm{difF}}(G,\maTHCAL{S}^k(\mAthFrak{G}))$ denoTes the differeNtiAble COhoMologY of $G$ wIth cOeFficiEnts in The $k$-tH sYmmetric power of The aDjoint repResEnTatIoN. Then $\PI$ is forMalLy POisson dIffeomoRPhiC tO ITS fIrst order approximaTiON ArOund $S$. DifFerentIAbLe COhomologY oF coMpacT GRoups VaniSHeS, therefoRe: \[CoroLLaRy\_2\] let $(M,\pi)$ bE a poissoN mAniFolD and lET $S\suBset M$ bE an embedDed syMPlectic leaf. AssUMe that the PoisSOn HOMoTOpy bUndLe of $S$ is a smoOth pRInciPal bUNdLe wITh vanIshinG sECoND de Rham cohomology anD cOmpact StrucTure group. Then $\Pi$ is formalLY pOisson diFfeoMOrPHic to its first oRder aPproximatiON around $S$. a bit mOre technIcal is the FOLlowing: \[THeoRem3\] let $(m, \pi)$ BE A POisson manifolD ANd leT $S\Subset M$ Be aN embeddEd sYmpLecTic LeAf whose isOtropy LiE aLgEbRa Is rEductIVe. If the aBeLiaNiZatIon alGEbroid $$a_S^{\matHrm{aB}}:=A_s/[\nU_s^*,\nu_s^*]$$ is inteGRaBLE by a SiMpLy coNneCtEd priNcipAL buNdle witH vanishinG seCOnd dE RHaM cohomoLogy and compacT sTructure grOuP, thEn $\pi$ is FORmally PoIsson diffeomorphic to its FIrst ordEr aPproxImatIon around $s$. \[CoRollarY\_3\] LeT $(m, \pi)$ be a poissoN maniFoLd aND Let $S\sUBSeT M$ bE aN embedded sYMPleCtic lEaF thrOugh $x\in m$. If the isotropy Lie

$ be an embedded symplecti c leaf. If thecoh omo lo gy g roup s $$H^2(A_S,\m a thca l{S}^{k}(\nu_S^*))$$ v anish f o r al l $ k\geq 2$, th e n$ \ pi$ i sfor ma l ly Pois son diffeo morphic to it sfirst ordera pp roximation ar ound $S$. I n m any ca se s,w e pro vethatthesec ohomol ogical ob st r uction s vanish , an d we obtain the follo w in g corollaries: \[The or e m2 \ ] Le t $ (M,\pi)$ b ea Poi s son man i fo l d and let $S\subset M$ be an e m bed ded sy mp lec t ic lea f. As su m e t hat the Poi sson homotopy bundl e of $S$ is a sm ooth p rin cip al b u nd le wi th van i sh ing sec ond de R ha mcohom olog y g r oup, an d th at it s structure g rou p $G $ sa tisfi es $$ H^2_ {\ mathr m{diff }}(G, \m athcal{S}^k(\ma thfr ak{g}))=0 ,\\\fo ra ll\ k \geq 2 ,$$ wh ere $\m athfrak { g}$ i s t he Lie algebra of $G $a n d$H^{\bul let}_{ \ ma th r m{diff}} (G ,\m athc a l {S}^k (\ma t hf rak{g})) $ deno t es t he diff er entiab le co hom ology of $ G$ wit h coeffi cient s in the $k$-th symmetric pow e ro f t h e ad joi nt represen tati o n. T hen$ \p i$i s for mally P o is s on diffeomorphic to i ts fir st or der approxima tion aroun d $ S$. Dif fere n ti a ble cohomology of c ompact gro u ps vanis hes,therefor e: \[Cor o l lary\_2\ ] L et$(M ,\p i ) $be a Poissonm a nifo ld and le t $ S\subse t M $ b e a n e mb edded sym plecticle af .As sum e tha t the Poi ss onho mot opy b u ndle o f $S$ isasm o oth princi p al b undl ewi th v ani sh ing s econ d de Rham c ohomology an d com pa ct struct ure group. Th en $\pi$ isfo rma lly Po i s son diff eomorphic to its firsto rder ap pro ximat ionaround $S $. A bit mo r e tech nicalis th efol l o wing: \[ The or em3\] Let$ ( M,\pi)$ b e aPoisson manifold and let$ S\s ubset M$ be a n e mbed d e dsym p le c tic l e afw h ose isotropy Li e algebrais re ductive. I f th eabelian ization alge b roid $$ A_S^{\mat hrm{ab}}: =A _S/[ \ n u_S ^*,\nu_S^* ]$$ is i ntegrable by as im ply c onn ectedpr inc ipalbundle wit h van ishing s econdde Rh am cohomol ogy and compact structu re gro up, t hen $\pi$ is fo r mal ly Poisso n di ffeomorphi c t o i ts fi rst order app r ox ima t ion a roun d $S$. \[ C or oll a r y\ _3\] Let $( M , \pi )$ be aP oisson man ifold and let $S\ s ubset M$ be an emb e d ded sy m plec ti c leaf through $x \i n M$. If t he isotropy L ie

$ be_an embedded_symplectic leaf. If the_cohomology groups_$$H^2(A_S,\mathcal{S}^{k}(\nu_S^*))$$_vanish for_all_$k\geq 2$, then_$\pi$ is formally_Poisson diffeomorphic to its_first order approximation_around_$S$. In many cases, we prove that these cohomological obstructions vanish, and we obtain the_following_corollaries: \[Theorem2\] Let_$(M,\pi)$_be_a Poisson manifold and let_$S\subset M$ be an embedded_symplectic leaf._Assume that the Poisson homotopy bundle of $S$_is_a smooth principal_bundle with vanishing second de Rham cohomology group, and_that its structure group $G$ satisfies_$$H^2_{\mathrm{diff}}(G,\mathcal{S}^k(\mathfrak{g}))=0,\ \ \forall\__k\geq_2,$$ where $\mathfrak{g}$ is_the Lie algebra of $G$ and_$H^{\bullet}_{\mathrm{diff}}(G,\mathcal{S}^k(\mathfrak{g}))$ denotes the differentiable cohomology of_$G$ with coefficients in the $k$-th symmetric_power of the adjoint representation. Then_$\pi$ is formally Poisson diffeomorphic_to its_first order approximation around $S$. Differentiable_cohomology of compact_groups vanishes,_therefore: \[Corollary\_2\] Let $(M,\pi)$_be a Poisson manifold and let_$S\subset M$ be_an embedded symplectic leaf. Assume that_the_Poisson homotopy bundle_of_$S$_is a_smooth principal bundle_with_vanishing second_de_Rham cohomology and compact structure group._Then_$\pi$ is formally Poisson diffeomorphic to its_first order approximation around_$S$. A_bit more technical is_the following: \[Theorem3\] Let $(M, \pi)$_be a Poisson manifold and let_$S\subset M$_be an_embedded symplectic leaf whose isotropy Lie algebra is reductive. If the_abelianization algebroid $$A_S^{\mathrm{ab}}:=A_S/[\nu_S^*,\nu_S^*]$$ is integrable by_a simply connected principal_bundle with_vanishing_second de Rham_cohomology_and compact_structure group, then $\pi$ is formally Poisson_diffeomorphic to_its first order approximation around $S$. \[Corollary\_3\]_Let $(M, \pi)$ be_a_Poisson manifold and let $S\subset M$_be an embedded symplectic leaf through_$x\in M$. If the isotropy_Lie

sum_{k=1}^Y \exp(\mathbf{W}_k \cdot \mathbf{x}+ \mathbf{b}_k) } \end{split}$$ Where $\mathbf{x}$ denotes a feature vector, $\mathbf{W}_k$ and $\mathbf{b}_k$ are the weight vectors and bias for class $k$. In our case, we treat words as classes to be predicted. Calculating the denominator thus requires summing over all of the words in the (large) vocabulary, an expensive operation. To avoid this computational bottleneck, we approximate the term $P(w_i|\mathbf{e}_k)$ with  Hierarchical Softmax.[^3] To learn meaningful user embeddings, we seek representations that are predictive of individual word-usage patterns. In light of this motivation, we approximate $P(w_i|\mathbf{u}_j)$ via the following hinge-loss objective which we aim to minimize: (w\_i,\_j) &=\ & \_[w\_l V, w\_l S]{}(0,1 - \_i + \_l ) Here, each $w_l$ (and corresponding embedding, $\mathbf{e}_l$) is a *negative example*, i.e., a word not in the sentence under consideration, which was authored by user $j$. The intuition is that in the aggregate, such words are less likely to be employed by user $j$ than words observed in sentences she has authored. Thus minimizing this objective attempts to induce a representation that is discriminative with respect to word usage. In practice, $V$ will be very large and hence we approximate the objective via *negative sampling*, a variant of Noise Contrastive Estimation. The idea is to approximate the objective function in a binary classification task by learning to discriminate between observed positive examples (sampled from the true distribution) and *pseudo*-negative examples (sampled from a large space of predominantly negative instances). Intuitively, this shifts probability mass to plausible observations. See for notes on Negative Sampling and Noise Contrastive Estimation. Previous work by showed that this approach works well in representation learning tasks when a sufficient amount of training data is available. However, we have access to only a limited amount of text for each user (see Section \[sec:experiments\]). We hypothesize that this problem can be alleviated by carefully selecting the negative samples that mostly contribute to ‘push’ the

sum_{k=1}^Y \exp(\mathbf{W}_k \cdot \mathbf{x}+ \mathbf{b}_k) } \end{split}$$ Where $ \mathbf{x}$ denotes a feature vector, $ \mathbf{W}_k$ and $ \mathbf{b}_k$ are the weight vectors and diagonal for course $ k$. In our casing, we treat actor's line as class to be predicted. Calculating the denominator therefore requires summing over all of the words in the (bombastic) vocabulary, an expensive operation. To avoid this computational constriction, we approximate the term $ P(w_i|\mathbf{e}_k)$ with   Hierarchical Softmax.[^3 ] To learn meaningful user embeddings, we search representations that are predictive of individual word - usage patterns. In light of this motivation, we estimate $ P(w_i|\mathbf{u}_j)$ via the following hinge - loss objective which we aim to understate: (w\_i,\_j) & = \ & \_[w\_l V, w\_l S]{}(0,1 - \_i + \_l) Here, each $ w_l$ (and corresponding embedding, $ \mathbf{e}_l$) is a * negative example *, i.e., a word not in the conviction under consideration, which was authored by user $ j$. The intuition is that in the aggregate, such words are less probable to be employed by user $ j$ than words observed in sentences she has authored. therefore minimize this objective attempts to induce a representation that is discriminative with respect to word usage. In practice, $ V$ will be very large and hence we approximate the aim via * damaging sampling *, a version of Noise Contrastive Estimation. The theme is to approximate the objective function in a binary classification undertaking by learning to discriminate between observed positive examples (sample from the true distribution) and * pseudo*-negative examples (sampled from a large space of predominantly negative instances). Intuitively, this shifts probability mass to plausible observations. See for notes on Negative Sampling and Noise Contrastive Estimation. former work by show that this overture works well in representation learning tasks when a sufficient sum of training data is available. However, we have entree to only a circumscribed amount of text for each user (see Section \[sec: experiments\ ]). We hypothesize that this trouble can be alleviated by carefully selecting the negative samples that largely contribute to ‘ advertise ’ the

sum_{n=1}^Y \exp(\mathbf{W}_k \cdot \mathnf{x}+ \mathbf{b}_k) } \end{split}$$ Wiere $\mafhbf{x}$ devotes a feature vector, $\mathbh{W}_k$ qnd $\mqthbf{b}_k$ are the weight vectors wnd bias for xlass $k$. In our casc, we fveat cocds as classes jo be predicded. Calculatinc ghz denominator thus requires summing jver alk lf the words ig tht (lwrge) nogabulary, an expensive operation. Fo avoiv this computatoonal bottleneck, we approxlmatf the term $P(w_i|\mathhf{e}_k)$ with  Huerawxhical Softmxx.[^3] To learn meaningful oser embeddings, we seek represengatious that are peedlwtive of invividuwl word-usage patternv. In libht of this mobivatmon, qe approximate $P(w_i|\matibf{u}_j)$ via the followyng hinge-noas objective whicy qe aik to minknizd: (w\_i,\_n) &=\ & \_[w\_m V, w\_l S]{}(0,1 - \_i + \_l ) Here, each $w_l$ (ane corresponding embtddybg, $\mathbf{e}_l$) ia a *nedaeive example*, i.e., a word not in the sentetce under consideration, whuch was authored by uder $j$. The intuition is that in the aggregate, such words ara lesa lijejh tl be employed by user $j$ than words observed ig stntvnces she has autmored. Thus minimizonh yris objective attem'fs to induce a repredentatijn thqt is dissrimonative with respect to wore usage. In prcctuce, $V$ will be very large and kence ee aplroximate the objective via *negative swmpling*, a xariant of Noise Coktrdstive Estimation. The idea is to ap'roxikate thd obkectivq function in a binary classificatlon tcsk bf learning to discriminate between observxv positive exsm[lev (sampleb from the true diseribution) and *kseudo*-negctive dxamples (szmpled hrom a large space of preglminantly nejative inftanxes). Untuitixdly, this shiftx probability mass ro plausible obseryatiova. See for notes on Negative Samplong anq Jomse Cjttrastive Esdimagiov. Lreviuus work by rhowrd that this approacv wodks well in represrnbation leqrning twsks when a sifficient amount ov tramning vata ix adailable. However, we have accesa to only a jimited amougt on tevt for eack user (see Section \[sec:experiments\]). We hypithesize that this problem can be allgvisted by carxfully selectinc the negative samplws that mostly coktribute to ‘push’ the

sum_{k=1}^Y \exp(\mathbf{W}_k \cdot \mathbf{x}+ \mathbf{b}_k) } \end{split}$$ denotes feature vector, and $\mathbf{b}_k$ are for $k$. In our we treat words classes to be predicted. Calculating the thus requires summing over all of the words in the (large) vocabulary, an operation. To avoid this computational bottleneck, we approximate the term $P(w_i|\mathbf{e}_k)$ with Hierarchical To meaningful embeddings, seek representations that are predictive of individual word-usage patterns. In light of this motivation, we approximate via the following hinge-loss objective which we aim minimize: (w\_i,\_j) &=\ & V, w\_l S]{}(0,1 - \_i \_l Here, each (and embedding, is a *negative i.e., a word not in the sentence under consideration, which was authored by user $j$. The intuition that in such words less to employed by user words observed in sentences she has this objective attempts to induce a representation that discriminative with to word usage. In practice, $V$ be very large and hence we approximate the via *negative sampling*, a variant of Noise Contrastive Estimation. The idea is to approximate the in a binary classification by learning to between positive (sampled the true and *pseudo*-negative examples (sampled from a large space of predominantly negative Intuitively, this shifts probability mass to plausible observations. See for Negative and Noise Contrastive Previous work by showed this works well in representation when sufficient data available. we have access to a limited amount of text each user (see Section problem can be alleviated by carefully selecting the samples that mostly contribute to ‘push’ the

sum_{k=1}^Y \exp(\mathbf{W}_k \cdot \mathbF{x}+ \mathbf{b}_k) } \End{spLit}$$ wheRe $\MathBf{x}$ dEnotes a feature VEctoR, $\mathbf{W}_k$ and $\mathbf{b}_k$ arE the wEiGHt veCToRs and Bias for CLaSS $K$. In OuR cAse, We TReAt worDs aS classeS to be prediCteD. CAlculating thE DeNominator tHus Requires summIng Over alL oF thE Words In tHe (larGe) vocaBUlary, aN expensivE oPEratioN. to avoid THIs CompUtational bottleneCK, wE Approximate the Term $P(w_I|\mAThBF{E}_k)$ wIth  hierarchicAl softmAX.[^3] To learN MeANINgfUL user embeddinGs, we seek repREseNtatioNs ThaT Are preDictiVe OF inDividual worD-usaGe patternS. In ligHT of this MOtivatiOn, we apProXimAte $P(W_I|\mAtHbf{U}_j)$ VIa tHE fOllOWinG hinge-loSs ObJectiVe whICH WE aim To mInimIze: (w\_i,\_J) &=\ & \_[w\_l V, w\_l S]{}(0,1 - \_i + \_l ) Here, EacH $w_l$ (aND coRrespOndinG embEdDing, $\mAthbf{e}_L$) is a *nEgAtive example*, i.e., a Word Not in the sEntEnCe uNdEr conSIderatIon, WhiCh was auThored bY UseR $j$. tHE InTuition is that in the AgGREgAte, such wOrds arE LeSs LIkely to bE eMplOyed BY User $j$ Than WOrDs observEd in seNTeNcEs she haS aUthoreD. THus MinImiziNG thiS objecTive atteMpts tO Induce a represeNTation that is dIScRIMiNAtivE wiTh respect to Word USage. in prACtIce, $v$ Will bE very LaRGe ANd hence we approximatE tHe objeCtive Via *negative saMpling*, a varIANT of Noise contRAsTIve Estimation. THe ideA is to approXImate the ObjecTive funcTion in a biNARy classiFicAtiOn tAsk BY LeArning to discrIMInatE bEtween oBseRved posItiVe eXamPleS (sAmpled froM the true DiStRiBuTioN) and *pSEudo*-negaTiVe eXaMplEs (samPLed froM a larGe spAcE oF PreDominanTLy NEGatiVe InStanCes). inTuitiVely, THis Shifts pRobabilitY maSS to pLaUsIble obsErvations. See fOr Notes on NegAtIve sampliNG And Noise contrastive Estimation. PrEVious woRk bY showEd thAt this appRoaCh workS weLL in repResentAtion LeArnING taskS WHeN a sUfFicient amoUNT of TrainInG datA is avaiLable. However, we have ACceSs to only a limiTed AmouNT Of TexT FoR EacH uSEr (sEE section \[sec:experIments\]). We hyPoTHeSize that thIS prObLem can bE alleviAted bY CarefulLy selectiNg the negaTiVe saMPLes That mostly ContribuTe to ‘push’ tHE

sum_{k=1}^Y \exp(\mathbf{W }_k \cdot\math bf{ x}+ \ math bf{b }_k) } \end{sp l it}$ $ Where $\mathbf{x}$ d enote sa fea t ur e vec tor, $\ m at h b f{W }_ k$ an d$ \m athbf {b} _k$ are the weigh t v ec tors and bia s f or class $ k$. In our case , w e trea twor d s ascla ssesto bep redict ed. Calcu la t ing th e denomi n a to r th us requires summi n go ver all of the words i n t h e (l arg e) vocabul ar y, an expensi v eo p e rat i on. To avoidthis comput a tio nal bo tt len e ck, we appr ox i mat e the term$P(w _i|\mathb f{e}_k ) $ withHierarc hicalSof tma x.[^ 3 ] T o l ea r n m e an ing f uluser emb ed di ngs,we s e e k repr ese ntat ionsthat are pred ict iveo f i ndivi dualword -u sagepatter ns. I nlight of this m otiv ation, we ap pr oxi ma te $P ( w_i|\m ath bf{ u}_j)$via the fol lo w i n ghinge-loss objecti ve w hi ch we ai m to m i ni mi z e: (w\_ i, \_j ) &= \ & \_[ w\_l V, w\_l S] {}(0,1 -\_ i + \_l ) Here ,eac h $ w_l$( andcorres pondingembed d ing, $\mathbf{ e }_l$) is a *n e ga t i ve exam ple *, i.e., aword notin t h esen t enceunder c o ns i deration, which was a uthore d byuser $j$. The intuition i s that in the ag g regate, such w ordsare less l i kely tobe em ployed b y user $j $ than wor dsobs erv edi n s entences sheh a s au th ored. T hus minimi zin g t his ob je ctive att empts to i nd uc ea r epres e ntationth atis di scrim i native with res pe ct toword us a ge . Inpr ac tice , $ V$ will bev ery largeand hence we appr ox im ate the objective vi a*negativesa mpl ing*,a variantof Noise Contrastive Es t imation . T he id ea i s to appr oxi mate t heo bjecti ve fun ction i n a b inary c la ssi fi cation tas k bylearn in g to discri minate between obs e rve d positive ex amp les( s am ple d f r omth e tr u e distribution)and *pseud o* - ne gative exa m ple s(sample d froma lar g e space of predo minantlyne gati v e in stances).Intuitiv ely, this shift s p robab ili ty mas stoplaus ible o b ser vatio ns. Se efor no tes o nNegative Sampling and Noise Con trasti ve Es tim ation. P rev i ous work byshow ed that th isapp roach wo r ks we ll i n r epr e senta tion learningt as ksw h en a sufficie n t amo unt o f t r aining dat a is available. H o wever, we have acc e s s t o o n ly a l imited amountofte x t for eac huser (see S ection \ [s e c:exp erimen ts\]). We hyp o t he s ize th at t his problemcan b e allevi at ed by car eful ly selec ting t h e ne g a tive samples tha t mos t l y con t rib ute t o‘push’t he

sum_{k=1}^Y \exp(\mathbf{W}_k_\cdot \mathbf{x}+_\mathbf{b}_k) } \end{split}$$ Where $\mathbf{x}$_denotes a_feature_vector, $\mathbf{W}_k$_and_$\mathbf{b}_k$ are the_weight vectors and_bias for class $k$._In our case,_we_treat words as classes to be predicted. Calculating the denominator thus requires summing over_all_of the_words_in_the (large) vocabulary, an expensive_operation. To avoid this computational_bottleneck, we_approximate the term $P(w_i|\mathbf{e}_k)$ with  Hierarchical Softmax.[^3] To learn_meaningful_user embeddings, we_seek representations that are predictive of individual word-usage patterns._In light of this motivation, we_approximate $P(w_i|\mathbf{u}_j)$ via_the_following_hinge-loss objective which we_aim to minimize: (w\_i,\_j) &=\ & \_[w\_l V,_w\_l S]{}(0,1 - \_i + \_l_) Here, each $w_l$ (and corresponding embedding, $\mathbf{e}_l$)_is a *negative example*, i.e., a_word not in the sentence_under consideration,_which was authored by user_$j$. The intuition_is that_in the aggregate,_such words are less likely to_be employed by_user $j$ than words observed in_sentences_she has authored._Thus_minimizing_this objective_attempts to induce_a_representation that_is_discriminative with respect to word usage. In_practice,_$V$ will be very large and hence_we approximate the objective_via_*negative sampling*, a variant_of Noise Contrastive Estimation. The_idea is to approximate the objective_function in_a binary_classification task by learning to discriminate between observed positive examples (sampled_from the true distribution) and *pseudo*-negative_examples (sampled from a_large space_of_predominantly negative instances)._Intuitively,_this shifts_probability mass to plausible observations. See for_notes on_Negative Sampling and Noise Contrastive Estimation. Previous_work by showed that_this_approach works well in representation learning_tasks when a sufficient amount of_training data is available. However,_we_have_access to only a limited_amount of text for each user_(see Section \[sec:experiments\])._We hypothesize that this problem can be_alleviated_by carefully selecting the negative samples_that_mostly contribute to ‘push’ the

are taken from an ossuarium, and are put together in groups, which are asserted to be those of individual skeletons. To test this a biologist takes the triplet femur, tibia, humerus, and seeks the correlation between the indices femur/humerus and tibia/humerus. He might reasonably conclude that this correlation marked organic relationship, and believe that the bones had really been put together substantially in their individual grouping. As a matter of fact, since the coefficients of variation for femur, tibia, and humerus are approximately equal, there would be, as we shall see later, a correlation of about 0.4 to 0.5 between these indices had the bones been sorted absolutely at random. I term this a spurious organic correlation, or simply a spurious correlation. I understand by this phrase the amount of correlation which would still exist between the indices, were the absolute lengths on which they depend distributed at random.” The reason for this correlation will be that some of the random femur and tibia pairs will be combined with a large humerus; in this case both the femur/humerus and tibia/humerus ratio will tend to be smaller than average. Other femur and tibia pairs will be combined with a small humerus; in this case both the femur/humerus and tibia/humerus ratio will tend to be larger than average. Correlation coefficients of the two ratios give the illusion that tibia and femur length covary, even when they in fact do not. For his part, Weldon was forced to concede that nearly 50% of the correlation he had observed in body measurements was actually due to this effect. Just over a decade later, another important figure in the development of mathematical statics, G. U. Yule, noted that when absolute values share a common factor, they are just as susceptible to this problem as are “indices” or ratios [@Yule1910RoyalStat]: > “Suppose we combine at random two indices $z_1$ and $z_2$, e.g. two death-rates, and also combine at random with each pair a denominator or population $x_3$. The correlations between $z_1$, $z_2$, and $x_3$ will then be zero within the limits of sampling. But now suppose we work out the total deaths $x_1=z_1 x_3$ and $x_2=z_2 x

are taken from an ossuarium, and are put together in groups, which are assert to be those of individual skeletal system. To test this a biologist takes the triplet femur, tibia, humerus, and search the correlation between the indices femur / humerus and tibia / humerus. He might sanely conclude that this correlation marked constituent kinship, and believe that the bones had very been put together substantially in their individual group. As a matter of fact, since the coefficients of mutant for femur, tibia, and humerus are approximately equal, there would be, as we shall see subsequently, a correlation of about 0.4 to 0.5 between these indices had the bone been sorted absolutely at random. I term this a spurious organic correlation coefficient, or simply a bastardly correlation. I understand by this phrase the measure of correlation which would still exist between the indices, were the absolute lengths on which they depend distribute at random. ” The reason for this correlation will be that some of the random femur and tibia pairs will be combined with a large humerus; in this case both the femur / humerus and tibia / humerus ratio will tend to be smaller than average. Other femur and tibia pair will be combined with a small humerus; in this casing both the femur / humerus and tibia / humerus proportion will tend to be big than average. Correlation coefficient of the two ratios pass the illusion that tibia and femur length covary, even when they in fact serve not. For his part, Weldon was forced to concede that nearly 50% of the correlation he had observed in body measurement was actually due to this effect. Just over a decade later, another important figure in the growth of mathematical statics, G. U. Yule, noted that when absolute values share a coarse factor, they are just as susceptible to this trouble as are “ indices ” or proportion [ @Yule1910RoyalStat ]: > “ Suppose we combine at random two index $ z_1 $ and $ z_2 $, e.g. two end - rates, and also combine at random with each pair a denominator or population $ x_3$. The correlations between $ z_1 $, $ z_2 $, and $ x_3 $ will then be zero within the limit of sampling. But nowadays suppose we knead out the total deaths $ x_1 = z_1 x_3 $ and $ x_2 = z_2 x

arf taken from an ossuariuo, and are put together in grkups, whizh are asserted to be those lf indicidual skeletons. To tert this a biologiwt tekes the triplet femur, tlyia, hhleruv, and seeks the correlatimn between the ivdnces femur/humerus and tibia/humerus. Hq might rfasonably concjude ehat nhls correlation marked organic remationsiip, and believe that the bones had really beej put together subdtantially un trwir individuxl grouping. As a mattet of fact, since the coefficients of vcriation fot relor, tibia, and humegus are approwpmately equal, yhere would be, as we shall see later, a cocrelation of about 0.4 jo 0.5 betweet chese indices had the bines teen sorgwd xbsklntemy at ganvom. I term fhis a spuruous organic correlstyin, or simply z spuryots correlation. I understand by this phrdse the amount of correlatuon which would still exist beeween the indices, were the absolute lengths on whhch tieh dtpckd dkwtgibuted at random.” The reason for this correlatykn wpll be that some jf the randpm friur and tibia pairs wiml be combined witj a larde hunerus; in uhis vase both the femur/humerus qnd tibia/humvrus ratio will tend tl be smallex than aversge. Other femur and tibna paids will be fombined skth a small humefus; it this case both the femur/rumerus aid tiyia/humerjs rstio wyll tend tl be larger than average. Corrglatiot coefficifnts of the two ratios give the illusion that thbid and feiur lcngth covary, evqn when they iu fact dj not. For his pzrt, Welvon was forcqd to concede jhat nearly 50% mf the cjrreoatiin he hxa observed in nody measlrtments was actually due to tmis ewrect. Just over a dexade later, anotner im[ogtait fidgre in the davelupmdmt of mathematiccl stayics, G. U. Yule, noted dhat when absolute valief share q common factor, they sre just as suscepuible vo thir pronlei as are “indices” or ratios [@Yuls1910RoyalStah]: > “Smppose we comfine at random tco indices $z_1$ and $z_2$, e.g. two death-rates, and also combine at tandom with each pair s denominatpr or populatimn $x_3$. The correlationw between $z_1$, $z_2$, and $x_3$ will then be zero sithin the pimits of sampling. But now suppose we work out the total deaths $x_1=z_1 x_3$ and $z_2=z_2 x

are taken from an ossuarium, and are in which are to be those this biologist takes the femur, tibia, humerus, seeks the correlation between the indices and tibia/humerus. He might reasonably conclude that this correlation marked organic relationship, and that the bones had really been put together substantially in their individual grouping. a of since coefficients of variation for femur, tibia, and humerus are approximately equal, there would be, as we see later, a correlation of about 0.4 to between these indices had bones been sorted absolutely at I this a organic or a spurious correlation. understand by this phrase the amount of correlation which would still exist between the indices, were the lengths on depend distributed random.” reason this correlation will some of the random femur and be combined with a large humerus; in this both the and tibia/humerus ratio will tend to smaller than average. Other femur and tibia pairs be combined with a small humerus; in this case both the femur/humerus and tibia/humerus ratio to be larger than Correlation coefficients of two give illusion tibia and length covary, even when they in fact do not. For his Weldon was forced to concede that nearly 50% of the had in body measurements actually due to this Just a decade later, another in development G. Yule, that when absolute values a common factor, they are as susceptible to this ratios [@Yule1910RoyalStat]: > “Suppose we combine at random indices $z_1$ and $z_2$, e.g. two death-rates, also combine at random with each pair a denominator or population $x_3$. correlations between and $x_3$ will then be zero within the of sampling. But now we work out the total deaths $x_1=z_1 x_3$ and x

are taken from an ossuarium, anD are put togEther In gRouPs, WhicH are Asserted to be thOSe of Individual skeletons. To tEst thIs A BiolOGiSt takEs the trIPlET FemUr, TiBia, HuMErUs, and SeeKs the coRrelation bEtwEeN the indices fEMuR/humerus anD tiBia/humerus. He MigHt reasOnAblY ConclUde That tHis corRElatioN marked orGaNIc relaTIonship, AND bElieVe that the bones had REaLLy been put togetHer subStANtIALly In tHeir indiviDuAl groUPing. As a MAtTER Of fACt, since the coeFficients of VAriAtion fOr FemUR, tibia, And huMeRUs aRe approximaTely Equal, therE would BE, as we shALl see laTer, a coRreLatIon oF AbOuT 0.4 to 0.5 BeTWeeN ThEse INdiCes had thE bOnEs beeN sorTED ABsolUteLy at RandoM. I term this a spUriOus oRGanIc corRelatIon, oR sImply A spuriOus coRrElation. I understAnd bY this phraSe tHe AmoUnT of coRRelatiOn wHicH would sTill exiST beTwEEN ThE indices, were the absOlUTE lEngths on Which tHEy DePEnd distrIbUteD at rANDom.” ThE reaSOn For this cOrrelaTIoN wIll be thAt Some of ThE raNdoM femuR And tIbia paIrs will bE combINed with a large hUMerus; in this caSE bOTH tHE femUr/hUmerus and tiBia/hUMeruS ratIO wIll TEnd to Be smaLlER tHAn average. Other femur AnD tibia Pairs Will be combineD with a smalL HUMerus; in tHis cASe BOth the femur/humErus aNd tibia/humERus ratio Will tEnd to be lArger than AVErage. CorRelAtiOn cOefFICiEnts of the two rATIos gIvE the illUsiOn that tIbiA anD feMur LeNgth covarY, even wheN tHeY iN fAct Do not. fOr his parT, WEldOn Was ForceD To concEde thAt neArLy 50% OF thE correlATiON He haD oBsErveD in BoDy meaSureMEntS was actUally due tO thIS effEcT. JUst over A decade later, aNoTher importAnT fiGure in THE developMent of mathematical statiCS, G. U. Yule, NotEd thaT wheN absolute ValUes shaRe a COmmon fActor, tHey arE jUst AS SuscePTIbLe tO tHis problem AS Are “IndicEs” Or raTios [@YulE1910RoyalStat]: > “Suppose wE ComBine at random tWo iNdicES $Z_1$ aNd $z_2$, E.G. tWO deAtH-RatES, And also combine aT random witH eACh Pair a denomINatOr Or populAtion $x_3$. THe corRElationS between $z_1$, $Z_2$, and $x_3$ will ThEn be ZERo wIthin the liMits of saMpling. But NOw supPOsE we woRk oUt the tOtAl dEaths $X_1=z_1 x_3$ and $X_2=Z_2 x

are taken from an ossuari um, and ar e put to get he r in gro ups, which are asse rted to be those of in divid ua l ske l et ons.To test th i s abi ol ogi st ta kes t hetriplet femur, ti bia ,humerus, and se eks the co rre lation betwe enthe in di ces femur /hu merus and t i bia/hu merus. He m i ght re a sonably c on clud e that this corre l at i on marked orga nic re la t io n s hip , a nd believe t hat t h e bones ha d r eal l y been put to gether subs t ant iallyin th e ir ind ividu al gro uping. As a mat ter of fa ct, si n ce thec oeffici ents o f v ari atio n f or fe mu r , t i bi a,a ndhumerusar eappro xima t e l y equ al, the re wo uld be, as we sh alls eelater , a c orre la tionof abo ut 0. 4to 0.5 betweenthes e indices ha dthe b onesb een so rte d a bsolute ly at r a ndo m. I te rm this a spurious o r g an ic corre lation , o rs imply asp uri ousc o rrela tion . I underst and by th is phrase t he amo un t o f c orrel a tion which would s tille xist between t h e indices, we r et h ea bsol ute lengths on whi c h th ey d e pe ndd istri buted a t r a ndom.” The reasonfo r this corr elation willbe that so m e of the r ando m f e mur and tibiapairs will be c o mbined w ith a large h umerus; i n this cas e b oth th e f e m ur /humerus andt i bia/ hu merus r ati o willten d t o b e s ma ller than average .Ot he rfem ur an d tibia p ai rswi llbe co m binedwitha sm al lh ume rus; in th i s cas ebo th t hefe mur/h umer u s a nd tibi a/humerus ra t io w il ltend to be larger th an average.Co rre lation c oefficie nts of the two ratios g i ve theill usion tha t tibia a ndfemurlen g th cov ary, e ven w he n t h e y inf a ct do n ot. For hi s par t, We ld on w as forc ed to concede that nea rly 50% of th e c orre l a ti onh eh adob s erv e d in body measur ements was a c tu ally due t o th is effect . Just over a decad e later,another i mp orta n t fi gure in th e develo pment ofm athem a ti cal s tat ics, G .U.Yule, noted tha t whe n abso lu te val ues s ha re a com mon factor, they are ju st assusce pti ble to th isp rob lem as ar e “i ndices” or ra tio s [@Y ule 1 910Ro yalS t at ]:> “Su ppos e we combi n eatr a nd om two indi c e s $z _1$ a nd$ z_2$,e.g. two death-rates, and also combi ne a t ran dom with e ach pair a den omi na t o r or pop ul ation $x_3$ . The co rr e latio ns bet ween $ z_1$, $ z _ 2$ , and $ x_3$ wi ll then b e z er o within t he limits ofsa mpling . Butn ow s u p pose we work out thet o tal d e ath s $x_ 1= z_1 x_3 $ and $x_2=z_2x

are_taken from_an ossuarium, and are_put together_in_groups, which_are_asserted to be_those of individual_skeletons. To test this_a biologist takes_the_triplet femur, tibia, humerus, and seeks the correlation between the indices femur/humerus and tibia/humerus._He_might reasonably_conclude_that_this correlation marked organic relationship,_and believe that the bones_had really_been put together substantially in their individual grouping._As_a matter of_fact, since the coefficients of variation for femur, tibia,_and humerus are approximately equal, there_would be, as_we_shall_see later, a correlation_of about 0.4 to 0.5 between_these indices had the bones been_sorted absolutely at random. I term this_a spurious organic correlation, or simply_a spurious correlation. I understand_by this_phrase the amount of correlation_which would still_exist between_the indices, were_the absolute lengths on which they_depend distributed at_random.” The reason for this correlation will_be_that some of_the_random_femur and_tibia pairs will_be_combined with_a_large humerus; in this case both_the_femur/humerus and tibia/humerus ratio will tend to_be smaller than average._Other_femur and tibia pairs_will be combined with a_small humerus; in this case both_the femur/humerus_and tibia/humerus_ratio will tend to be larger than average. Correlation coefficients of_the two ratios give the illusion_that tibia and femur_length covary,_even_when they in_fact_do not._For his part, Weldon was forced to_concede that_nearly 50% of the correlation he_had observed in body_measurements_was actually due to this effect. Just_over a decade later, another important_figure in the development of_mathematical_statics,_G. U. Yule, noted that_when absolute values share a common_factor, they are_just as susceptible to this problem as_are_“indices” or ratios [@Yule1910RoyalStat]: > “Suppose we_combine_at random two indices $z_1$ and_$z_2$,_e.g._two death-rates, and also combine_at random with each pair a_denominator or population $x_3$. The correlations between $z_1$, $z_2$,_and $x_3$ will_then be zero within the_limits_of_sampling. But now suppose we work out the total deaths_$x_1=z_1 x_3$_and $x_2=z_2 x

_2(m_1-4m_3)M_{cc}^2\nonumber\\ && +2(-2m_1^2-4m_1m_3+3M_{cc}^2+2Q^2)M_{cc}^2)M_{bc}^8M_{cc}^2,\nonumber\\ C_1^{(2,-3)} &=& \frac{1}{48M_{bc}^8M_{cc}^6}(m_1(9m_3M_{bc}^2-m_1m_3^2+ m_2(-9M_{bc}^2-4m_1m_3+4m_3^2)+\nonumber\\ && 4m_1M_{cc}^2)M_{bc}^2M_{cc}^4 - (2m_2^3m_3+ 9m_2^2m_3^2+m_3^4+10m_2m_3M_{cc}^2-\\ && 10M_{cc}^4)M_{bc}^6 + (m_2^4m_3^2+2m_2^3m_3^3- 3m_2^2m_3^2M_{cc}^2-3m_2^2M_{cc}^4-14m_2m_3M_{cc}^4\nonumber\\ && +9M_{cc}^6+m_1^2(m_2^2m_3^2+ 2m_3^2M_{cc}^2-3M_{cc}^4)+2m_1(-m_2^3m_3^2+m_2^2m_3^3+\nonumber\\ && m_2^2m_3M_{cc}^2- 2m_2m_3^2M_{cc}^2+m_2M_{cc}^4-3m_3M_{cc}^4)-m_2^2m_3^2Q^2-2m_3^2M_{cc}^2Q^2+ \nonumber\\ && 3M_{cc}^2Q^2)M_{bc}^4+m_1(m_1^3m_3^2+3(m_2-m_3)m_3^2M_{bc

_ 2(m_1 - 4m_3)M_{cc}^2\nonumber\\ & & +2(-2m_1 ^ 2 - 4m_1m_3 + 3M_{cc}^2 + 2Q^2)M_{cc}^2)M_{bc}^8M_{cc}^2,\nonumber\\ C_1^{(2,-3) } & = & \frac{1}{48M_{bc}^8M_{cc}^6}(m_1(9m_3M_{bc}^2 - m_1m_3 ^ 2 + m_2(-9M_{bc}^2 - 4m_1m_3 + 4m_3 ^ 2)+\nonumber\\ & & 4m_1M_{cc}^2)M_{bc}^2M_{cc}^4 - (2m_2 ^ 3m_3 + 9m_2 ^ 2m_3 ^ 2+m_3 ^ 4 + 10m_2m_3M_{cc}^2-\\ & & 10M_{cc}^4)M_{bc}^6 + (m_2 ^ 4m_3 ^ 2 + 2m_2 ^ 3m_3 ^ 3- 3m_2 ^ 2m_3 ^ 2M_{cc}^2 - 3m_2 ^ 2M_{cc}^4 - 14m_2m_3M_{cc}^4\nonumber\\ & & +9M_{cc}^6+m_1 ^ 2(m_2 ^ 2m_3 ^ 2 + 2m_3 ^ 2M_{cc}^2 - 3M_{cc}^4)+2m_1(-m_2 ^ 3m_3 ^ 2+m_2 ^ 2m_3 ^ 3+\nonumber\\ & & m_2 ^ 2m_3M_{cc}^2- 2m_2m_3 ^ 2M_{cc}^2+m_2M_{cc}^4 - 3m_3M_{cc}^4)-m_2 ^ 2m_3 ^ 2Q^2 - 2m_3 ^ 2M_{cc}^2Q^2 + \nonumber\\ & & 3M_{cc}^2Q^2)M_{bc}^4+m_1(m_1 ^ 3m_3 ^ 2 + 3(m_2 - m_3)m_3 ^ 2M_{bc

_2(m_1-4m_3)M_{fc}^2\nonumber\\ && +2(-2m_1^2-4m_1m_3+3M_{cc}^2+2Q^2)M_{cc}^2)M_{bc}^8O_{cc}^2,\nonumber\\ C_1^{(2,-3)} &=& \ftax{1}{48M_{bc}^8M_{cr}^6}(m_1(9m_3M_{bc}^2-m_1j_3^2+ m_2(-9M_{bc}^2-4m_1m_3+4m_3^2)+\vonumber\\ && 4m_1M_{cc}^2)M_{bc}^2M_{cc}^4 - (2m_2^3m_3+ 9m_2^2m_3^2+m_3^4+10m_2m_3L_{cx}^2-\\ && 10M_{cc}^4)N_{bc}^6 + (m_2^4m_3^2+2m_2^3m_3^3- 3m_2^2m_3^2M_{cc}^2-3m_2^2M_{cc}^4-14m_2m_3M_{cz}^4\nonumber\\ && +9M_{cc}^6+m_1^2(m_2^2m_3^2+ 2m_3^2N_{cc}^2-3M_{rc}^4)+2m_1(-m_2^3m_3^2+m_2^2m_3^3+\nonumber\\ && m_2^2m_3M_{cc}^2- 2m_2m_3^2M_{ge}^2+m_2M_{cc}^4-3j_3L_{cc}^4)-m_2^2k_3^2W^2-2m_3^2M_{cc}^2Q^2+ \nonumber\\ && 3M_{cc}^2Q^2)M_{bc}^4+m_1(m_1^3k_3^2+3(m_2-m_3)m_3^2M_{bc

_2(m_1-4m_3)M_{cc}^2\nonumber\\ && +2(-2m_1^2-4m_1m_3+3M_{cc}^2+2Q^2)M_{cc}^2)M_{bc}^8M_{cc}^2,\nonumber\\ C_1^{(2,-3)} &=& \frac{1}{48M_{bc}^8M_{cc}^6}(m_1(9m_3M_{bc}^2-m_1m_3^2+ m_2(-9M_{bc}^2-4m_1m_3+4m_3^2)+\nonumber\\ - 9m_2^2m_3^2+m_3^4+10m_2m_3M_{cc}^2-\\ && + (m_2^4m_3^2+2m_2^3m_3^3- 3m_2^2m_3^2M_{cc}^2-3m_2^2M_{cc}^4-14m_2m_3M_{cc}^4\nonumber\\ 2m_2m_3^2M_{cc}^2+m_2M_{cc}^4-3m_3M_{cc}^4)-m_2^2m_3^2Q^2-2m_3^2M_{cc}^2Q^2+ && 3M_{cc}^2Q^2)M_{bc}^4+m_1(m_1^3m_3^2+3(m_2-m_3)m_3^2M_{bc

_2(m_1-4m_3)M_{cc}^2\nonumber\\ && +2(-2m_1^2-4m_1m_3+3M_{cc}^2+2Q^2)M_{cc}^2)M_{bC}^8M_{cc}^2,\nonumbEr\\ C_1^{(2,-3)} &=& \frAc{1}{48M_{Bc}^8M_{Cc}^6}(M_1(9m_3M_{bC}^2-m_1m_3^2+ m_2(-9m_{bc}^2-4m_1m_3+4m_3^2)+\nonumber\\ && 4M_1m_{cc}^2)M_{Bc}^2M_{cc}^4 - (2m_2^3m_3+ 9m_2^2m_3^2+m_3^4+10m_2m_3M_{cc}^2-\\ && 10M_{cc}^4)M_{bc}^6 + (m_2^4M_3^2+2m_2^3m_3^3- 3m_2^2m_3^2m_{cC}^2-3M_2^2M_{cc}^4-14M_2M_3M_{Cc}^4\nonUmber\\ && +9M_{cC}^6+M_1^2(m_2^2M_3^2+ 2M_3^2m_{cc}^2-3m_{cC}^4)+2m_1(-M_2^3m_3^2+m_2^2M_3^3+\nONuMber\\ && m_2^2M_3M_{cC}^2- 2m_2m_3^2M_{cc}^2+m_2m_{cc}^4-3m_3M_{cc}^4)-m_2^2m_3^2Q^2-2M_3^2M_{cC}^2Q^2+ \Nonumber\\ && 3M_{cc}^2Q^2)m_{Bc}^4+M_1(m_1^3m_3^2+3(m_2-m_3)m_3^2M_{bc

_2(m_1-4m_3)M_{cc}^2\nonum ber\\ && + 2(-2m _1^ 2-4 m_ 1m_3 +3M_ {cc}^2+2Q^2)M_ { cc}^ 2)M_{bc}^8M_{cc}^2,\no numbe r\ \ C_1 ^ {( 2,-3) } &=& \ f ra c { 1}{ 48 M_ {bc }^ 8 M_ {cc}^ 6}( m_1(9m_ 3M_{bc}^2- m_1 m_ 3^2+ m_2(-9M _ {b c}^2-4m_1m _3+ 4m_3^2)+\non umb er\\ & &4m_ 1 M_{cc }^2 )M_{b c}^2M_ { cc}^4- (2m_2^3 m_ 3 + 9m_2 ^ 2m_3^2+ m _ 3^ 4+10 m_2m_3M_{cc}^2-\\ && 10M_{cc}^4)M_{ bc}^6+( m_ 2 ^ 4m_ 3^2 +2m_2^3m_3 ^3 - 3m_ 2 ^2m_3^2 M _{ c c } ^2- 3 m_2^2M_{cc}^4 -14m_2m_3M_ { cc} ^4\non um ber \ \ && + 9M_{c c} ^ 6+m _1^2(m_2^2m _3^2 + 2m_3^2M _{cc}^ 2 -3M_{cc } ^4)+2m_ 1(-m_2 ^3m _3^ 2+m_ 2 ^2 m_ 3^3 +\ n onu m be r\\ &&m_2^2m_3 M_ {c c}^2- 2m_ 2 m _ 3 ^2M_ {cc }^2+ m_2M_ {cc}^4-3m_3M_ {cc }^4) - m_2 ^2m_3 ^2Q^2 -2m_ 3^ 2M_{c c}^2Q^ 2+ \n on umber\\ && 3M_{ cc}^ 2Q^2)M_{b c}^ 4+ m_1 (m _1^3m _ 3^2+3( m_2 -m_ 3)m_3^2 M_{bc

_2(m_1-4m_3)M_{cc}^2\nonumber\\ && +2(-2m_1^2-4m_1m_3+3M_{cc}^2+2Q^2)M_{cc}^2)M_{bc}^8M_{cc}^2,\nonumber\\ C_1^{(2,-3)}_&=& \frac{1}{48M_{bc}^8M_{cc}^6}(m_1(9m_3M_{bc}^2-m_1m_3^2+ m_2(-9M_{bc}^2-4m_1m_3+4m_3^2)+\nonumber\\ &&_4m_1M_{cc}^2)M_{bc}^2M_{cc}^4 - (2m_2^3m_3+ 9m_2^2m_3^2+m_3^4+10m_2m_3M_{cc}^2-\\ && 10M_{cc}^4)M_{bc}^6_+ (m_2^4m_3^2+2m_2^3m_3^3- 3m_2^2m_3^2M_{cc}^2-3m_2^2M_{cc}^4-14m_2m_3M_{cc}^4\nonumber\\ &&_+9M_{cc}^6+m_1^2(m_2^2m_3^2+ 2m_3^2M_{cc}^2-3M_{cc}^4)+2m_1(-m_2^3m_3^2+m_2^2m_3^3+\nonumber\\ &&_m_2^2m_3M_{cc}^2- 2m_2m_3^2M_{cc}^2+m_2M_{cc}^4-3m_3M_{cc}^4)-m_2^2m_3^2Q^2-2m_3^2M_{cc}^2Q^2+ \nonumber\\ && 3M_{cc}^2Q^2)M_{bc}^4+m_1(m_1^3m_3^2+3(m_2-m_3)m_3^2M_{bc

limiting surface brightness of $\mu_g \sim 25.5 \rm ~mag~ arcsec^{-2}$), but also on the spectroscopy of (nominally) all galaxies brighter than $r<17.77$ mag[^1] (with effective surface brightness $\mu_e<24.5 \rm ~mag~ arcsec^{-2}$).\ The present “near-field cosmology” project makes use of the photometry and the spectroscopy provided by SDSS much in the way astronomers use telescopes in “service observing” mode, but counting on an almost unlimited amount of observing time. By saying so we wish to emphasize that, opposite to most SDSS statistical analyzes that are based on such large number of objects that cannot be inspected individually, the number of galaxies ($\sim$ 4000) in the present study is large enough to ensure statistical significance, but not too large to prevent individual inspection of all images and spectra. The Coma supercluster constitutes the ideal bridge between the near-field and the high-$z$ cosmology.\ Several modern studies based on hundreds of thousands galaxies from the SDSS in the local universe (usually $0.05<z<0.2$) address environmental issues by quantifying the dependence of galaxy structural and photometric parameters from the local galaxy density. The density around each galaxy is computed counting galaxies in cylinders of given projected radius and depth in the redshift space. Most studies adopt 2 Mpc and 500 $\rm km s^{-1}$ (e.g. Kauffmann et al. 2004) or 1 Mpc and 1000 $\rm km s^{-1}$ (e.g. Hogg et al. 2004), as a compromise for detecting structures intermediate between large groups and small clusters. It is not easy however to visualize what type of structures are found by these automatic algorithms. The present study gives the opportunity to reciprocally calibrate structures that are seen individually in the CS with those that are detected by statistical means.\ It is well know that the SDSS pipelines have not been designed for extracting parameters of extended nearby galaxies, which results in problems in creating reliable low-redshift catalogs, due to: 1) “shredding” of large galaxies in multiple pieces leading to wrong magnitude determinations. 2) slight (20%) underestimate of the brightness of galaxies larger than $r_{50}\sim 20"$ due to the way the photometric pipeline determines the median sky, biasing the sky brightness near large extended galaxies, as reported by Mandelbaum et al. (2005), Lauer et al.

limiting surface brightness of $ \mu_g \sim 25.5 \rm ~mag~ arcsec^{-2}$), but also on the spectroscopy of (nominally) all galaxy bright than $ r<17.77 $ mag[^1 ] (with effective surface luminosity $ \mu_e<24.5 \rm ~mag~ arcsec^{-2}$).\ The present “ near - playing field cosmology ” project have use of the photometry and the spectroscopy provided by SDSS much in the manner astronomers use telescope in “ service observing ” mode, but counting on an about unlimited amount of note time. By saying thus we wish to emphasize that, opposite to most SDSS statistical analyzes that are based on such big number of objects that cannot be inspected individually, the number of galaxy ($ \sim$ 4000) in the present study is large enough to ensure statistical significance, but not too big to prevent individual inspection of all images and spectra. The Coma supercluster constitutes the ideal bridge between the near - field and the high-$z$ cosmology.\ Several modern studies based on hundreds of thousands galaxies from the SDSS in the local universe (usually $ 0.05 < z<0.2 $) address environmental issues by quantifying the addiction of galax structural and photometric parameters from the local galaxy concentration. The concentration around each galaxy is computed counting galaxies in cylinders of given projected radius and astuteness in the redshift space. Most studies adopt 2 Mpc and 500 $ \rm km s^{-1}$ (e.g. Kauffmann et al. 2004) or 1 Mpc and 1000 $ \rm km s^{-1}$ (e.g. Hogg et al. 2004), as a compromise for detecting structures intermediate between large group and small bunch. It is not comfortable however to visualize what character of structures are found by these automatic algorithms. The present study gives the opportunity to reciprocally calibrate structures that are seen individually in the c with those that are detected by statistical means.\ It is well know that the SDSS pipelines have not been designed for distill parameters of elongated nearby galaxies, which results in problem in creating dependable low - red shift catalogs, due to: 1) “ shredding ” of big galaxies in multiple pieces precede to wrong order of magnitude determinations. 2) slight (20 %) underestimate of the brightness of galax larger than $ r_{50}\sim 20"$ due to the way the photometric pipeline determines the median sky, bias the sky brightness near large drawn-out galaxies, as reported by Mandelbaum et al. (2005), Lauer et al.

liliting surface brightnesr of $\mu_g \sim 25.5 \rm ~mag~ accsec^{-2}$), bht also un the spectroscopy of (nominelly) all talaxies brighter than $r<17.77$ mag[^1] (winh effectuve wyrface brijgtness $\mu_e<24.5 \rm ~lag~ crrsec^{-2}$).\ The present “near-field cosmology” probezt makes use of the photometry and thq specttodcopy provided by XQSS jlcm in the way astronomers use telsscopes in “service onserving” mode, but counting on wn almost unlimitef amount of obsqeving time. Bh saying so we wish to emphasize that, opposite to mosg SDSX statistixao ajdlyzes that are fased on sucm large number of objects thst rannit be inspected indivmdually, the number os galaxiev ($\aim$ 4000) in the presebt studi is nargd enuugg vo snsure stetistical sjgnificance, but not too large uo [grvent indivisual igs[ection of all images and spectra. The Cmma supercluster constitutws the ideal bridge bgtween the near-field and the high-$z$ cosmology.\ Several modern vtudixs baweq in hundreds of thousands galaxies from the SDSS jn tme local universc (usually $0.05<z<0.2$) addresx fnfyronmental isrues bv qhantifying the depfndence of gqlaxy strtctutal and photometric parametwrs from the oocal galaxy densicy. The densicy aroond eavh galaxy is computed cuuntjng galaxied in cylihaers of given prujebted radius and depth in the rqdshift s'ace. Kost stjdiex adope 2 Mpc and 500 $\rm km s^{-1}$ (e.g. Kauffmann eh al. 2004) os 1 Mpc and 1000 $\rm km s^{-1}$ (e.g. Hogg et al. 2004), as a compromise for cedecning struetures intermediate between large groups and rmall clusners. It iv not easy rowever to vivoalize what tbpe of stwuctyres are fojvd by these auyomatic algorithms. Rhe present study nives fhe opportunity to reciprocally cslicraee svructtses that are seev ivcividjally in thc CR wiyh those that are dedectsd by statistical kewns.\ It is well knjw that the SCSS pipelines have not ueen dxsignec fjr extracting parameters of exfended newrbn galaxies, whych vesujts in proylems in creating reliable low-redshift cetalogs, due to: 1) “shreddibg” of large galaxiex in multiplx piecqs leadinc to wrong magnitude determinations. 2) xlight (20%) underestimate of tha brihhtness of galaxies larger than $r_{50}\sim 20"$ due to the way the photometric pipeoine dxtqrmines the medoan sny, yiasing the rky brightness nvar large extended galaxies, as rxported by Mcndelbaum et al. (2005), Lauer et al.

limiting surface brightness of $\mu_g \sim 25.5 arcsec^{-2}$), also on spectroscopy of (nominally) mag[^1] effective surface brightness \rm ~mag~ arcsec^{-2}$).\ present “near-field cosmology” project makes use the photometry and the spectroscopy provided by SDSS much in the way astronomers telescopes in “service observing” mode, but counting on an almost unlimited amount of time. saying we to emphasize that, opposite to most SDSS statistical analyzes that are based on such large number objects that cannot be inspected individually, the number galaxies ($\sim$ 4000) in present study is large enough ensure significance, but too to individual inspection of images and spectra. The Coma supercluster constitutes the ideal bridge between the near-field and the high-$z$ cosmology.\ modern studies hundreds of galaxies the in the local $0.05<z<0.2$) address environmental issues by quantifying galaxy structural and photometric parameters from the local density. The around each galaxy is computed counting in cylinders of given projected radius and depth the redshift space. Most studies adopt 2 Mpc and 500 $\rm km s^{-1}$ (e.g. Kauffmann 2004) or 1 Mpc 1000 $\rm km (e.g. et 2004), a compromise detecting structures intermediate between large groups and small clusters. It is easy however to visualize what type of structures are found automatic The present study the opportunity to reciprocally structures are seen individually in with that statistical It well know that the pipelines have not been designed extracting parameters of extended problems in creating reliable low-redshift catalogs, due to: “shredding” of large galaxies in multiple pieces to wrong magnitude determinations. 2) slight (20%) underestimate of the brightness of larger than due to the way the photometric pipeline determines median sky, biasing the brightness near large extended galaxies, as reported by Mandelbaum al. Lauer et

limiting surface brightness Of $\mu_g \sim 25.5 \rm ~Mag~ arCseC^{-2}$), buT aLso oN the Spectroscopy of (NOminAlly) all galaxies brighteR than $R<17.77$ mAG[^1] (witH EfFectiVe surfaCE bRIGhtNeSs $\Mu_e<24.5 \Rm ~MAg~ ArcseC^{-2}$).\ ThE presenT “near-field CosMoLogy” project mAKeS use of the pHotOmetry and the SpeCtroscOpY prOVided By SdSS muCh in thE Way astRonomers uSe TElescoPEs in “serVICe ObseRving” mode, but countINg ON an almost unlimIted amOuNT oF OBseRviNg time. By saYiNg so wE Wish to eMPhASIZe tHAt, opposite to mOst SDSS statIStiCal anaLyZes THat are Based On SUch Large number Of obJects that Cannot BE inspecTEd indivIduallY, thE nuMber OF gAlAxiEs ($\SIm$ 4000) iN ThE prESenT study is LaRgE enouGh to ENSURe stAtiSticAl sigNificance, but nOt tOo laRGe tO prevEnt inDiviDuAl insPectioN of alL iMages and spectra. the COma supercLusTeR coNsTitutES the idEal BriDge betwEen the nEAr-fIeLD ANd The high-$z$ cosmology.\ SEvERAl Modern stUdies bASeD oN Hundreds Of ThoUsanDS GalaxIes fROm The SDSS iN the loCAl UnIverse (uSuAlly $0.05<z<0.2$) aDdResS enVironMEntaL issueS by quantIfyinG The dependence oF Galaxy structuRAl AND pHOtomEtrIc parameterS froM The lOcal GAlAxy DEnsitY. The dEnSItY Around each galaxy is cOmPuted cOuntiNg galaxies in cYlinders of GIVEn projecTed rADiUS and depth in the RedshIft space. MoST studies Adopt 2 mpc and 500 $\rm Km s^{-1}$ (e.g. KaufFMAnn et al. 2004) oR 1 MpC anD 1000 $\rm Km s^{-1}$ (E.G. hoGg et al. 2004), as a compROMise FoR detectIng StructuRes IntErmEdiAtE between lArge grouPs AnD sMaLl cLusteRS. It is not EaSy hOwEveR to viSUalize What tYpe oF sTrUCtuRes are fOUnD BY theSe AuTomaTic AlGoritHms. THE prEsent stUdy gives tHe oPPortUnItY to reciProcally calibRaTe structurEs ThaT are seEN IndividuAlly in the CS with those thaT Are deteCteD by stAtisTical meanS.\ It Is well KnoW That thE SDSS pIpeliNeS haVE Not beEN DeSigNeD for extracTINg pArameTeRs of ExtendeD nearby galaxies, whiCH reSults in probleMs iN creATInG reLIaBLe lOw-REdsHIFt catalogs, due to: 1) “Shredding” oF lARgE galaxies iN MulTiPle piecEs leadiNg to wROng magnItude deteRminationS. 2) sLighT (20%) UNdeRestimate oF the brigHtness of gALaxieS LaRger tHan $R_{50}\sim 20"$ duE tO thE way tHe photOMetRic piPeline DeTerminEs the MeDian sky, bIasing the sky brightness nEar larGe extEndEd galaxieS, as REpoRted by ManDelbAum et al. (2005), LauEr eT al.

limiting surface brightne ss of $\mu _g \s im25. 5\rm~mag ~ arcsec^{-2}$ ) , bu t also on the spectros copyof (nom i na lly)all gal a xi e s br ig ht erth a n$r<17 .77 $ mag[^ 1] (with e ffe ct ive surfaceb ri ghtness $\ mu_ e<24.5 \rm ~ mag ~ arcs ec ^{- 2 }$).\ Th e pre sent “ n ear-fi eld cosmo lo g y” pro j ect mak e s u se o f the photometrya nd the spectrosco py pro vi d ed b y S DSS much in t he waya stronom e rs u s e t e lescopes in “ service obs e rvi ng” mo de , b u t coun tingon analmost unli mite d amountof obs e rving t i me. Bysaying so we wis h t oemp ha s ize th at, opp osite to m os t SDS S st a t i s tica l a naly zes t hat are based on suc h la rge n umber ofob jects thatcanno tbe inspected in divi dually, t henu mbe rof ga l axies($\ sim $ 4000) in the pre se n t st udy is large enoug ht o e nsure st atisti c al s i gnifican ce , b ut n o t toolarg e t o preven t indi v id ua l inspe ct ion of a llima ges a n d sp ectra. The Com a sup e rcluster const i tutes the ide a lb r id g e be twe en the near -fie l d an d th e h igh - $z$ c osmol og y .\ Several modern stud ie s base d onhundreds of t housands g a l a xies fro m th e S D SS in the loca l uni verse (usu a lly $0.0 5<z<0 .2$) add ress envi r o nmentaliss ues by qu a n ti fying the dep e n denc eof gala xystructu ral an d p hot om etric par ametersfr om t he lo cal g a laxy den si ty. T hedensi t y arou nd ea ch g al ax y is comput e dc o unti ng g alax ies i n cyl inde r s o f given projecte d r a dius a nd depthin the redshi ft space. Mo st st udiesa d opt 2 Mp c and 500 $\rm km s^{-1 } $ (e.g. Ka uffma nn e t al. 200 4)or 1 M pca nd 100 0 $\rm km s ^{ -1} $ (e.g. H og g e tal. 2004), a s a comp ro mise for de tecting structures int ermediate bet wee n la r g egro u ps and s m all c lusters. It isnot easy h ow e ve r to visua l ize w hat typ e of st ructu r es arefound bythese aut om atic a lgo rithms. Th e presen t study g i ves t h eoppor tun ity to r eci proca lly ca l ibr ate s tructu re s that arese en indiv idually in the CS withthosethatare detected by sta tisticalmean s.\ It iswel l k now t hat the S DSSp ip eli n es ha ve n o t been de s ig ned f or extracting p a ram eters of extend ed n earby galaxies, w h ich results in pro b l ems in crea ti ng reliable lo w-r ed s h ift cata lo gs, due to: 1) “shr ed d ing”of lar ge gal axies i n mu l tiplepiec esleading t o w ro n g magni tu de determ inat io ns. 2) sligh t (20 % ) underestimate o f the b right n ess of g al axies l a rger than $r_{ 50}\sim 20" $ dueto t he wa y the p ho tometr icpi peline det e rmines th e med ian sky ,bias ing the s ky b r i ghtne ss n ea r l arge exte n d ed ga la x ies , as repo rt ed b y Mandelb a um et al . ( 2 005), L au ere t al.

limiting_surface brightness_of $\mu_g \sim 25.5_\rm ~mag~_arcsec^{-2}$),_but also_on_the spectroscopy of_(nominally) all galaxies_brighter than $r<17.77$ mag[^1]_(with effective surface_brightness_$\mu_e<24.5 \rm ~mag~ arcsec^{-2}$).\ The present “near-field cosmology” project makes use of the photometry and_the_spectroscopy provided_by_SDSS_much in the way astronomers_use telescopes in “service observing”_mode, but_counting on an almost unlimited amount of observing_time._By saying so_we wish to emphasize that, opposite to most SDSS_statistical analyzes that are based on_such large number_of_objects_that cannot be inspected_individually, the number of galaxies ($\sim$_4000) in the present study is_large enough to ensure statistical significance, but_not too large to prevent individual_inspection of all images and_spectra. The_Coma supercluster constitutes the ideal_bridge between the_near-field and_the high-$z$ cosmology.\ Several_modern studies based on hundreds of_thousands galaxies from_the SDSS in the local universe_(usually_$0.05<z<0.2$) address environmental_issues_by_quantifying the_dependence of galaxy_structural_and photometric_parameters_from the local galaxy density. The_density_around each galaxy is computed counting galaxies_in cylinders of given_projected_radius and depth in_the redshift space. Most studies_adopt 2 Mpc and 500 $\rm_km s^{-1}$_(e.g. Kauffmann_et al. 2004) or 1 Mpc and 1000 $\rm km s^{-1}$_(e.g. Hogg et al. 2004), as_a compromise for detecting_structures intermediate_between_large groups and_small_clusters. It_is not easy however to visualize what_type of_structures are found by these automatic_algorithms. The present study_gives_the opportunity to reciprocally calibrate structures_that are seen individually in the_CS with those that are_detected_by_statistical means.\ It is well know_that the SDSS pipelines have not_been designed for_extracting parameters of extended nearby galaxies, which_results_in problems in creating reliable low-redshift_catalogs,_due to: 1) “shredding” of large_galaxies_in_multiple pieces leading to wrong_magnitude determinations. 2) slight (20%) underestimate_of the brightness of galaxies larger than $r_{50}\sim 20"$_due to the_way the photometric pipeline determines_the_median_sky, biasing the sky brightness near large extended galaxies, as_reported by_Mandelbaum et al._(2005), Lauer et al.

pt minus 1fil}\frac{\prod_{i=1}^{k-1} \mu(d_{i}) \mu(d'_{i}) F_{i}(\log_x d_{i}) G_{i}(\log_x d'_{i}) }{{\varphi}(q_{W,d_1,\dots,d'_{k-1}})} = \frac{1}{{\varphi}(W)} (c'+o(1)) B^{-k+1},$$ where $$c' := \prod_{i=1}^{k-1} \int_0^1 F'_i(t) G'_i(t)\ dt$$ (note that $F_i, G_i$ are supported on $[0,1]$ by hypothesis), so by it suffices to show that $$\label{cpeps} \sum_{x+h_k \leq n \leq 2x+h_k} \lambda_{F_k}(n) \lambda_{G_k}(n) {\mathbf{1}}_{p(n)>x^{\varepsilon}} = (c''_{\varepsilon}+ o(1)) \frac{x}{\log x},$$ where $c''_{{\varepsilon}}$ is a quantity depending on ${\varepsilon}$ but not on $x$ such that $$\lim_{{\varepsilon}\to 0} c''_{{\varepsilon}} = \int_0^1 F'_k(t) G'_k(t)\ dt.$$ In the case $S(F_k)+S(G_k) < 1$, this would follow easily from (the $k=1$ case of) Theorem \[nonprime-asym\](i) and Proposition \[almostprime\]. In the general case, we may appeal once more to the fundamental theorem of arithmetic. As before, we may factor $n = p_1 \dots p_r$ for some $x^{\varepsilon}\leq p_1 \leq \dots \leq p_r \leq 2x+h_k$ and $r \leq \frac{1}{{\varepsilon}}+1$. The contribution of those $n$ with a repeated prime factor $p_i = p_{i+1}$ can easily be shown to be ${\llcurly}x^{1-{\varepsilon}}$ in the same manner we dealt with $\Sigma_2$, so we may restrict attention to the square-free $n$, for which the $p_i$ are strictly increasing. In that case, one can write $$\lambda

pt minus 1fil}\frac{\prod_{i=1}^{k-1 } \mu(d_{i }) \mu(d'_{i }) F_{i}(\log_x d_{i }) G_{i}(\log_x d'_{i }) } { { \varphi}(q_{W, d_1,\dots, d'_{k-1 } }) } = \frac{1}{{\varphi}(W) } (c'+o(1) ) B^{-k+1},$$ where $ $ c': = \prod_{i=1}^{k-1 } \int_0 ^ 1 F'_i(t) G'_i(t)\ dt$$ (note that $ F_i, G_i$ are supported on $ [ 0,1]$ by hypothesis), so by it suffice to picture that $ $ \label{cpeps } \sum_{x+h_k \leq n \leq 2x+h_k } \lambda_{F_k}(n) \lambda_{G_k}(n) { \mathbf{1}}_{p(n)>x^{\varepsilon } } = (c''_{\varepsilon}+ o(1) ) \frac{x}{\log x},$$ where $ c''_{{\varepsilon}}$ is a quantity depend on $ { \varepsilon}$ but not on $ x$ such that $ $ \lim_{{\varepsilon}\to 0 } c''_{{\varepsilon } } = \int_0 ^ 1 F'_k(t) G'_k(t)\ dt.$$ In the lawsuit $ S(F_k)+S(G_k) < 1 $, this would follow easily from (the $ k=1 $ character of) Theorem \[nonprime - asym\](i) and Proposition \[almostprime\ ]. In the cosmopolitan case, we may appeal once more to the fundamental theorem of arithmetic. As before, we may factor $ normality = p_1 \dots p_r$ for some $ x^{\varepsilon}\leq p_1 \leq \dots \leq p_r \leq 2x+h_k$ and $ r \leq \frac{1}{{\varepsilon}}+1$. The contribution of those $ n$ with a repeated choice agent $ p_i = p_{i+1}$ can easily be shown to be $ { \llcurly}x^{1-{\varepsilon}}$ in the same manner we dealt with $ \Sigma_2 $, so we may restrict care to the square - free $ n$, for which the $ p_i$ are strictly increase. In that case, one can write $ $ \lambda

pt linus 1fil}\frac{\prod_{i=1}^{k-1} \mu(d_{i}) \mu(d'_{i}) F_{i}(\log_x d_{i}) G_{i}(\log_x d'_{i}) }{{\vadphi}(q_{W,d_1,\duts,d'_{k-1}})} = \frac{1}{{\varphi}(W)} (c'+o(1)) B^{-k+1},$$ whece $$c' := \proe_{i=1}^{k-1} \int_0^1 F'_i(t) G'_i(t)\ dt$$ (notd that $F_i, G_i$ are wuppieted on $[0,1]$ bb hypothcfis), al by mt suffices to xhow that $$\nabel{cpeps} \sum_{x+v_k \lzq n \leq 2x+h_k} \lambda_{F_k}(n) \lambda_{G_k}(n) {\matrbf{1}}_{p(n)>x^{\vsrfpsilon}} = (c''_{\vareksilom}+ o(1)) \fdac{x}{\log x},$$ where $c''_{{\varepsilon}}$ is a qhantity depending on ${\varepsilon}$ but not on $x$ skch hhat $$\lim_{{\varepsilon}\ho 0} c''_{{\varepsulon}} = \int_0^1 F'_k(t) G'_k(t)\ dt.$$ In the case $S(F_k)+S(G_k) < 1$, this would follow easily froo (the $k=1$ case of) Tyeoggm \[nonprime-awym\](i) wnd Propositlpn \[almmstprimr\]. In the genersl rase, we may appeal once mmre to the fundamegtal theosej of arithmetic. Aw veforg, we kay wqctur $h = p_1 \dots o_r$ hor some $x^{\vzrepsilon}\lew p_1 \leq \dots \leq p_r \lqw 2x+h_k$ and $r \lsq \fras{1}{{\vwrepsilon}}+1$. The contribution of those $n$ wpth z repeated prime factor $p_i = p_{i+1}$ can easily be shown to be ${\llcurly}x^{1-{\varepsilon}}$ in the same manner we deald witi $\Rigna_2$, ri ae may restrict attention to the square-free $n$, rot ehich the $p_i$ ave strictly increaxijg. Yn that case, une cau wdite $$\lambda

pt minus 1fil}\frac{\prod_{i=1}^{k-1} \mu(d_{i}) \mu(d'_{i}) F_{i}(\log_x d_{i}) }{{\varphi}(q_{W,d_1,\dots,d'_{k-1}})} \frac{1}{{\varphi}(W)} (c'+o(1)) where $$c' := (note $F_i, G_i$ are on $[0,1]$ by so by it suffices to show $$\label{cpeps} \sum_{x+h_k \leq n \leq 2x+h_k} \lambda_{F_k}(n) \lambda_{G_k}(n) {\mathbf{1}}_{p(n)>x^{\varepsilon}} = (c''_{\varepsilon}+ o(1)) \frac{x}{\log where $c''_{{\varepsilon}}$ is a quantity depending on ${\varepsilon}$ but not on $x$ such $$\lim_{{\varepsilon}\to c''_{{\varepsilon}} \int_0^1 G'_k(t)\ dt.$$ In the case $S(F_k)+S(G_k) < 1$, this would follow easily from (the $k=1$ case Theorem \[nonprime-asym\](i) and Proposition \[almostprime\]. In the general we may appeal once to the fundamental theorem of As we may $n p_1 p_r$ for some p_1 \leq \dots \leq p_r \leq 2x+h_k$ and $r \leq \frac{1}{{\varepsilon}}+1$. The contribution of those $n$ with repeated prime = p_{i+1}$ easily shown be ${\llcurly}x^{1-{\varepsilon}}$ in manner we dealt with $\Sigma_2$, so attention to the square-free $n$, for which the are strictly In that case, one can write

pt minus 1fil}\frac{\prod_{i=1}^{k-1} \mu(d_{i}) \mU(d'_{i}) F_{i}(\log_x d_{I}) G_{i}(\loG_x d'_{I}) }{{\vaRpHi}(q_{W,D_1,\dotS,d'_{k-1}})} = \frac{1}{{\varphi}(W)} (C'+O(1)) B^{-k+1},$$ wHere $$c' := \prod_{i=1}^{k-1} \int_0^1 F'_i(t) G'_i(t)\ dt$$ (Note tHaT $f_i, G_i$ ARe SuppoRted on $[0,1]$ bY HyPOTheSiS), sO by It SUfFices To sHow that $$\Label{cpeps} \Sum_{X+h_K \leq n \leq 2x+h_k} \lAMbDa_{F_k}(n) \lambdA_{G_k}(N) {\mathbf{1}}_{p(n)>x^{\vaRepSilon}} = (c''_{\VaRepSIlon}+ o(1)) \FraC{x}{\log X},$$ where $C''_{{\VarepsIlon}}$ is a quAnTIty depENding on ${\VARePsilOn}$ but not on $x$ such thAT $$\lIM_{{\varepsilon}\to 0} c''_{{\VarepsIlON}} = \iNT_0^1 f'_k(t) g'_k(t)\ Dt.$$ In the casE $S(f_k)+S(G_k) < 1$, THis woulD FoLLOW eaSIly from (the $k=1$ caSe of) Theorem \[NOnpRime-asYm\](I) anD proposItion \[AlMOstPrime\]. In the gEnerAl case, we mAy appeAL once moRE to the fUndameNtaL thEoreM Of ArIthMeTIc. AS BeForE, We mAy factor $N = p_1 \DoTs p_r$ fOr soME $X^{\VArepSilOn}\leQ p_1 \leq \Dots \leq p_r \leq 2x+H_k$ aNd $r \lEQ \frAc{1}{{\varEpsilOn}}+1$. ThE cOntriBution Of thoSe $N$ with a repeated pRime Factor $p_i = p_{I+1}$ caN eAsiLy Be shoWN to be ${\lLcuRly}X^{1-{\varepsIlon}}$ in tHE saMe MANNeR we dealt with $\Sigma_2$, sO wE MAy Restrict AttentIOn To THe square-FrEe $n$, For wHICh the $P_i$ arE StRictly inCreasiNG. IN tHat case, OnE can wrItE $$\laMbdA

pt minus 1fil}\frac{\prod_ {i=1}^{k-1 } \mu (d_ {i} )\mu( d'_{ i}) F_{i}(\log _ x d_ {i}) G_{i}(\log_x d'_{ i}) } {{ \ varp h i} (q_{W ,d_1,\d o ts , d '_{ k- 1} })} = \f rac{1 }{{ \varphi }(W)} (c'+ o(1 )) B^{-k+1},$$ wh ere $$c' : = \ prod_{i=1}^{ k-1 } \int _0 ^1F '_i(t ) G '_i(t )\ dt$ $ (note that $F_ i, G_i$ a r e suppo r t ed on$[0,1]$ by hypoth e si s ), so by it su ffices t o s h o w t hat $$\label{ cp eps}\ sum_{x+ h _k \ l eqn \leq 2x+h_k} \lambda_{F _ k}( n) \la mb da_ { G_k}(n ) {\m at h bf{ 1}}_{p(n)>x ^{\v arepsilon }} = ( c ''_{\va r epsilon }+ o(1 ))\fr ac{x } {\ lo g x }, $ $ w h er e $ c ''_ {{\varep si lo n}}$is a q u a ntit y d epen dingon ${\varepsi lon }$ b u t n ot on $x$such t hat $ $\lim_ {{\va re psilon}\to 0} c ''_{ {\varepsi lon }} =\i nt_0^ 1 F'_k( t)G'_ k(t)\ d t.$$ In the c a s e $ S(F_k)+S(G_k) < 1$ ,t h is would f ollowe as il y from (t he $k =1$c a se of ) Th e or em \[non prime- a sy m\ ](i) an dPropos it ion \[ almos t prim e\]. I n the ge neral case, we may a p peal once mor e t o th e fun dam ental theor em o f ari thme t ic . A s befo re, w em ay factor $n = p_1 \do ts p_r$for s ome $x^{\vare psilon}\le q p _1 \leq\dot s \ l eq p_r \leq 2x +h_k$ and $r \l e q \frac{ 1}{{\ varepsil on}}+1$.T h e contri but ion of th o s e$n$ with a re p e ated p rime fa cto r $p_i= p _{i +1} $ c an easily b e shownto b e${ \ll curly } x^{1-{\v ar eps il on} }$ in the sa me ma nner w ed eal t with$ \S i g ma_2 $, s o we ma yrestr icta tte ntion t o the squ are - free $ n$ , for w hich the $p_i $are strict ly in creasi n g . In tha t case, one can write $ $ \lambda

pt minus_1fil}\frac{\prod_{i=1}^{k-1} \mu(d_{i})_\mu(d'_{i}) F_{i}(\log_x d_{i}) G_{i}(\log_x_d'_{i}) }{{\varphi}(q_{W,d_1,\dots,d'_{k-1}})}_=_\frac{1}{{\varphi}(W)} (c'+o(1))_B^{-k+1},$$_where $$c' :=_\prod_{i=1}^{k-1} \int_0^1 F'_i(t)_G'_i(t)\ dt$$ (note that_$F_i, G_i$ are_supported_on $[0,1]$ by hypothesis), so by it suffices to show that $$\label{cpeps} \sum_{x+h_k \leq n_\leq_2x+h_k} \lambda_{F_k}(n)_\lambda_{G_k}(n)_{\mathbf{1}}_{p(n)>x^{\varepsilon}}_= (c''_{\varepsilon}+ o(1)) \frac{x}{\log x},$$_where $c''_{{\varepsilon}}$ is a quantity_depending on_${\varepsilon}$ but not on $x$ such that $$\lim_{{\varepsilon}\to_0}_c''_{{\varepsilon}} = \int_0^1_F'_k(t) G'_k(t)\ dt.$$ In the case $S(F_k)+S(G_k) < 1$,_this would follow easily from (the_$k=1$ case of)_Theorem_\[nonprime-asym\](i)_and Proposition \[almostprime\]. In_the general case, we may appeal_once more to the fundamental theorem_of arithmetic. As before, we may factor_$n = p_1 \dots p_r$ for_some $x^{\varepsilon}\leq p_1 \leq \dots_\leq p_r_\leq 2x+h_k$ and $r \leq_\frac{1}{{\varepsilon}}+1$. The contribution_of those_$n$ with a_repeated prime factor $p_i = p_{i+1}$_can easily be_shown to be ${\llcurly}x^{1-{\varepsilon}}$ in the_same_manner we dealt_with_$\Sigma_2$,_so we_may restrict attention_to_the square-free_$n$,_for which the $p_i$ are strictly_increasing._In that case, one can write $$\lambda

_{0}}}{n_{0}} \right)^{2} w^{4} F_K,$$ where $\nu_0$ is the refractivity at standard pressure and temperature (or standard refractivity) of the molecular species, $w$ is the wavenumber, $F_K$ is the King correction factor, and $n_{0}$ is Loschmidt’s constant. Various parametrized functions are used to describe the spectral dependence of $\nu_0$ and $F_K$. Table \[tbl\_rayl\] gives references for the functional form of both parameters, as well as their spectral region. The transmission of each atmospheric layer is computed with Beer’s law from all optical depths. We use disc-averaged quantities for our model atmosphere. We use a present-day Earth vertical composition (Kaltenegger et al. (2010b) for SO$_2$; Lodders & Fegley, Jr. (1998) for Ar; and Cox (2000) for all other molecules) up to 130 km altitude, unless specified otherwise. Above 130 km, we assume constant mole fraction with height for all the molecules except for SO$_2$ which we fix at zero, and for N$_2$, O$_2$, and Ar which are described below. Below 100 km, we use the US 1976 atmosphere (COESA 1976) as the temperature-pressure profile. Above 100 km, the atmospheric density is sensitive to and increases with solar activity (Hedin 1987). We use the tabulated results of the MSIS-86 model, for solar maximum (Table A1.2) and solar minimum conditions (Table A1.1) published in Rees (1989), to derive the atmospheric density, pressure, and mole fractions for N$_2$, O$_2$, and Ar above 100 km. We run our simulations in two different spectral regimes. In the VIS-NIR, from 10000 to 25000 cm$^{-1}$ (400-1000 nm), we use a 0.05 cm$^{-1}$ grid, while in the UV from 25000 to 90000 cm$^{-1}$ (111-400 nm), we use a 0.5 cm$^{-1}$ grid. For displaying the results, the VIS-NIR and the UV simulations are binned on a 4 cm$^{-1}$ and a 20 cm$^{-1}$ grid, respectively. The choice in spectral resolution impacts

_ { 0}}}{n_{0 } } \right)^{2 } w^{4 } F_K,$$ where $ \nu_0 $ is the refractivity at standard pressure and temperature (or standard refractivity) of the molecular species, $ w$ is the wavenumber, $ F_K$ is the King correction factor, and $ n_{0}$ is Loschmidt ’s changeless. diverse parametrized functions are used to describe the apparitional dependence of $ \nu_0 $ and $ F_K$. Table   \[tbl\_rayl\ ] gives character for the functional form of both parameter, as well as their spectral area. The transmission of each atmospheric layer is computed with Beer ’s law from all ocular depths. We use disc - averaged quantities for our exemplar atmosphere. We use a present - day Earth erect composition (Kaltenegger et al. (2010b) for SO$_2 $; Lodders & Fegley, Jr. (1998) for Ar; and Cox (2000) for all other molecule) up to 130   km elevation, unless specified differently. Above 130   km, we assume constant mole fraction with height for all the atom except for SO$_2 $ which we fix at zero, and for N$_2 $, O$_2 $, and Ar which are described below. Below 100   km, we use the US 1976 atmosphere (COESA 1976) as the temperature - atmospheric pressure profile. Above 100   km, the atmospheric density is sensitive to and increases with solar activity (Hedin 1987). We use the tabulated results of the MSIS-86 model, for solar utmost (Table A1.2) and solar minimal conditions (Table A1.1) published in Rees   (1989), to deduce the atmospheric concentration, pressure, and mole fractions for N$_2 $, O$_2 $, and Ar above 100   km. We run our simulations in two different spectral regimes. In the VIS - NIR, from 10000 to 25000   cm$^{-1}$ (400 - 1000   nm), we practice a 0.05   cm$^{-1}$ grid, while in the UV from 25000 to 90000   cm$^{-1}$ (111 - 400   nm), we use a 0.5   cm$^{-1}$ grid. For displaying the results, the VIS - NIR and the UV simulations are binned on a 4   cm$^{-1}$ and a 20   cm$^{-1}$ grid, respectively. The choice in spectral settlement impacts

_{0}}}{n_{0}} \rlght)^{2} w^{4} F_K,$$ where $\nu_0$ is tht refractivity at standacd presaure and temperature (or standard refcactuvity) of the molecular speckes, $w$ is nhe wavenymbec, $F_K$ is the King correctljn fzgtor, cnv $n_{0}$ is Loschmidj’s constant. Earious paramedrkzzd functions are used to describe thq specttap dependence os $\nu_0$ wnd $R_K$. Table \[tbl\_rayl\] gives references fkr the hunctional form of both parameters, as welp as their spectral rehion. The trabsmifwion of each atmospherpe layer is domputed with Beer’s law from alu optncal depths. Ww udg disc-averagxd quagtities for our moden atmoslhere. We use a ireseit-dat Earth vertical compmsition (Kaltenegget et al. (2010b) xox SO$_2$; Lodders & Fegley, Hr. (1998) for Ar; dnd Zix (2000) fod ell other momecules) up to 130 km altirude, unless specifitd jnnerwise. Abovs 130 km, wq wssume constant mole fraction with heigvt ror all the molecules ezcept for SO$_2$ which we fix at zqro, and for N$_2$, O$_2$, and Ar which are described below. Telow 100 ym, cc jwe the US 1976 atmosphere (COESA 1976) as the temperature-[dexslre profile. Above 100 km, the atmpsohrtic density is sensicjvs to and increases with sjlar qctivity (Redim 1987). We use the tabulated resylts of the IWIS-86 model, for solax maximum (Tayle A1.2) snd splar minimum conditions (Tabme A1.1) publisjed in Resr (1989), to derive the xtmps[heric dtvsity, pressure, anq mole frectious for N$_2$, O$_2$, amd Ar wbove 100 km. Wf run our simulations in two biffesent spectgal regimes. In the VIS-NIR, from 10000 to 25000 cm$^{-1}$ (400-1000 nm), we uxe a 0.05 bm$^{-1}$ grid, wkile ik the UV from 25000 eo 90000 cm$^{-1}$ (111-400 nm), we usg a 0.5 cm$^{-1}$ grnd. For displayinf the rxsults, the VYS-NIR and the OV simulationv are bigned on q 4 cm$^{-1}$ ana a 20 cm$^{-1}$ grid, reslectively. The choicw in spectral resokutkkn impacts

_{0}}}{n_{0}} \right)^{2} w^{4} F_K,$$ where $\nu_0$ is at pressure and (or standard refractivity) is wavenumber, $F_K$ is King correction factor, $n_{0}$ is Loschmidt’s constant. Various parametrized are used to describe the spectral dependence of $\nu_0$ and $F_K$. Table \[tbl\_rayl\] references for the functional form of both parameters, as well as their spectral The of atmospheric is computed with Beer’s law from all optical depths. We use disc-averaged quantities for our model We use a present-day Earth vertical composition (Kaltenegger al. (2010b) for SO$_2$; & Fegley, Jr. (1998) for and (2000) for other up 130 km altitude, specified otherwise. Above 130 km, we assume constant mole fraction with height for all the molecules except SO$_2$ which at zero, for O$_2$, Ar which are Below 100 km, we use the (COESA 1976) as the temperature-pressure profile. Above 100 the atmospheric is sensitive to and increases with activity (Hedin 1987). We use the tabulated results the MSIS-86 model, for solar maximum (Table A1.2) and solar minimum conditions (Table A1.1) published (1989), to derive the density, pressure, and fractions N$_2$, and above 100 We run our simulations in two different spectral regimes. In the from 10000 to 25000 cm$^{-1}$ (400-1000 nm), we use a grid, in the UV 25000 to 90000 cm$^{-1}$ nm), use a 0.5 cm$^{-1}$ displaying results, the simulations binned on a 4 and a 20 cm$^{-1}$ grid, The choice in spectral

_{0}}}{n_{0}} \right)^{2} w^{4} F_K,$$ where $\nu_0$ is the refrActivity at StandArd PreSsUre aNd teMperature (or staNDard Refractivity) of the molecUlar sPeCIes, $w$ IS tHe wavEnumber, $f_k$ iS THe KInG cOrrEcTIoN factOr, aNd $n_{0}$ is LoSchmidt’s coNstAnT. Various paraMEtRized functIonS are used to deScrIbe the SpEctRAl depEndEnce oF $\nu_0$ and $f_k$. Table \[Tbl\_rayl\] giVeS RefereNCes for tHE FuNctiOnal form of both parAMeTErs, as well as theIr specTrAL rEGIon. the TransmissiOn Of eacH AtmosphERiC LAYer IS computed with beer’s law froM All OpticaL dEptHS. We use Disc-aVeRAgeD quantities For oUr model atMospheRE. We use a PResent-dAy EartH veRtiCal cOMpOsItiOn (kAltENeGgeR Et aL. (2010b) for SO$_2$; LOdDeRs & FegLey, JR. (1998) FOR ar; anD CoX (2000) for All otHer molecules) uP to 130 Km alTItuDe, unlEss spEcifIeD otheRwise. ABove 130 kM, wE assume constant Mole Fraction wIth HeIghT fOr all THe moleCulEs eXcept foR SO$_2$ whicH We fIx AT ZErO, and for N$_2$, O$_2$, and Ar whicH aRE DeScribed bElow. BeLOw 100 Km, WE use the Us 1976 aTmoSpheRE (cOESA 1976) As thE TeMperaturE-pressURe PrOfile. AbOvE 100 km, the AtMosPheRic deNSity Is sensItive to aNd incREases with solar ACtivity (Hedin 1987). WE UsE THe TAbulAteD results of tHe MSis-86 modEl, foR SoLar MAximuM (TablE A1.2) ANd SOlar minimum conditioNs (table A1.1) PubliShed in Rees (1989), to dErive the atMOSPheric deNsitY, PrESsure, and mole frActioNs for N$_2$, O$_2$, and aR above 100 km. we run Our simulAtions in tWO DifferenT spEctRal RegIMEs. in the VIS-NIR, frOM 10000 To 25000 cm$^{-1}$ (400-1000 Nm), We use a 0.05 cM$^{-1}$ grId, while In tHe Uv frOm 25000 tO 90000 cM$^{-1}$ (111-400 nm), we use a 0.5 Cm$^{-1}$ grid. FoR dIsPlAyIng The reSUlts, the ViS-nIR AnD thE UV siMUlatioNs are BinnEd On A 4 Cm$^{-1}$ aNd a 20 cm$^{-1}$ grID, rESPectIvElY. The ChoIcE in spEctrAL reSolutioN impacts

_{0}}}{n_{0}} \right)^{2}w^{4} F_K, $$ wh ere $\ nu _0$is t he refractivit y atstandard pressure andtempe ra t ure( or stan dard re f ra c t ivi ty )ofth e m olecu lar specie s, $w$ isthe w avenumber, $ F _K $ is the K ing correctionfac tor, a nd $n _ {0}$isLosch midt’s consta nt. Vario us parame t rized f u n ct ions are used to desc r ib e the spectraldepend en c eo f $\ nu_ 0$ and $F_ K$ . Tab l e \[tbl \ _r a y l \]g ives referenc es for thef unc tional f orm of bot h par am e ter s, as wellas t heir spec tral r e gion. T he tran smissi onofeach at mo sph er i c l a ye r i s co mputed w it hBeer’ s la w f r om a llopti cal d epths. We use di sc-a v era ged q uanti ties f or ou r mode l atm os phere. We usea pr esent-day Ea rt h v er tical compos iti on(Kalten egger e t al .( 2 0 10 b) for SO$_2$; Lod de r s & Fegley, Jr. ( 1 99 8) for Ar;an d C ox ( 2 0 00) f or a l lother mo lecule s )up to 130  k m alti tu de, un lesss peci fied o therwise . Abo v e 130 km, we a s sume constant mo l e f r acti onwith height for allthem ol ecu l es ex ceptfo r S O $_2$ which we fix a tzero,and f or N$_2$, O$_ 2$, and Ar w h ich aredesc r ib e d below. Below 100km, we use the US 1 976 a tmospher e (COESA1 9 76) as t hetem per atu r e -p ressure profi l e . Ab ov e 100 k m,the atm osp her icden si ty is sen sitive t oan din cre asesw ith sola ract iv ity (Hed i n 1987 ). We use t he tab ulatedr es u l ts o fth e MS IS- 86 mode l, f o r s olar ma ximum (Ta ble A1.2 )an d solar minimum cond it ions (Tabl eA1. 1) pub l i shed inRees (1989), to derivet he atmo sph ericdens ity, pres sur e, and mo l e frac tionsfor N $_ 2$, O $_2$, a nd Ar a bove 100 k m . We runou r si mulatio ns in two differen t sp ectral regime s.In t h e V IS- N IR , fr om 100 0 0 to 25000 cm$^{ -1}$ (400- 10 0 0nm), we us e a0. 05 cm$^ {-1}$ g rid,w hile in the UV f rom 25000 t o 90 0 0 0 c m$^{-1}$ ( 111-400nm), we u s e a 0 . 5cm$^{ -1} $ grid .For disp laying the resu lts, t he VIS-N IR an dthe UV s imulations are binned o n a 4cm$^{ -1} $ and a 2 0 c m $^{ -1}$ grid , re spectively . T hechoic e i n spec tral re sol u tionimpa c ts

_{0}}}{n_{0}} \right)^{2}_w^{4} F_K,$$_where $\nu_0$ is the_refractivity at_standard_pressure and_temperature_(or standard refractivity)_of the molecular_species, $w$ is the_wavenumber, $F_K$ is_the_King correction factor, and $n_{0}$ is Loschmidt’s constant. Various parametrized functions are used to_describe_the spectral_dependence_of_$\nu_0$ and $F_K$. Table \[tbl\_rayl\] gives_references for the functional form_of both_parameters, as well as their spectral region. The transmission_of_each atmospheric layer_is computed with Beer’s law from all optical depths._We use disc-averaged quantities for our_model atmosphere. We use_a_present-day_Earth vertical composition (Kaltenegger_et al. (2010b) for SO$_2$; Lodders_& Fegley, Jr. (1998) for Ar;_and Cox (2000) for all other molecules)_up to 130 km altitude, unless specified_otherwise. Above 130 km, we assume_constant mole_fraction with height for all_the molecules except_for SO$_2$_which we fix_at zero, and for N$_2$, O$_2$,_and Ar which_are described below. Below 100 km, we_use_the US 1976_atmosphere_(COESA_1976) as_the temperature-pressure profile._Above_100 km, the_atmospheric_density is sensitive to and increases_with_solar activity (Hedin 1987). We use the_tabulated results of the_MSIS-86_model, for solar maximum_(Table A1.2) and solar minimum_conditions (Table A1.1) published in Rees (1989),_to derive_the atmospheric_density, pressure, and mole fractions for N$_2$, O$_2$, and Ar above_100 km. We run our simulations in_two different spectral regimes._In the_VIS-NIR,_from 10000 to_25000 cm$^{-1}$_(400-1000 nm), we_use a 0.05 cm$^{-1}$ grid, while in the_UV from_25000 to 90000 cm$^{-1}$ (111-400 nm), we use_a 0.5 cm$^{-1}$ grid. For_displaying_the results, the VIS-NIR and the_UV simulations are binned on a_4 cm$^{-1}$ and a 20 cm$^{-1}$ grid,_respectively._The_choice in spectral resolution impacts

\[lemma: regularity\]. We next prove an analogue of [@Totaro Theorem 6.4]. This will play a crucial role in proving theorem \[theorem: restriction\]. \[theorem: unif\] Let $\mathscr{L}$ be a $q$-ample line bundle on $X$. Then for any $N$, there is an integer $m_{N}$, such that, for any coherent sheaf $\mathscr{F}$ on $X$ with $reg^{q'}(\mathscr{F})\leq N$, $$H^{i}(X,\mathscr{F}\otimes\mathscr{L}^{\otimes m})=0$$ for $i>q+q'$ and $m>m_{N}$. Fix an integer $i$ such that $q+q'< i \leq n$, by lemma \[lemma: vanishing of tensor product\], it is enough to show that there is an $M$, depending only on the choice of $N$, but not the coherent sheaf $\mathscr{F}$, such that for $m>M$, $0\leq a \leq 2n-i$ and $b\geq 0$, either $H^{b}(X,\mathscr{L}^{\otimes m}\otimes \mathscr{O}_{X}(-N)\otimes \mathscr{R}_{a})=0$ or $H^{i+a-b}(X,\mathscr{F}\otimes\mathscr{O}_{X}(N-a))$. Here $\mathscr{F}$ is any coherent sheaf with $reg^{q'}(\mathscr{F})\leq N$. $b> q$ and $0\leq a < n+b$. Using the $q$-ampleness of $\mathscr{L}$, there is an $m_N$, such that we have $$H^{q+j}(X,\mathscr{L}^{\otimes m}\otimes \mathscr{O}_{X}(-N-j))=0$$ for all $1\leq j\leq n-q$ and $m>m_N$, i.e. $\mathscr{L}^{\otimes m}\otimes \mathscr{O}_X(-N)$ is $q$-regular for all $m>m_N$. Now lemma \[lemma: R\_i\] says $$H^{b}(X,\mathscr{L}^{\otimes m}\otimes \mathscr{O}_{X}(-N)\otimes \mathscr{R}_{a})=0$$ for all $m>m_N$, $b> q$ and $

\[lemma: regularity\ ]. We next prove an analogue of [ @Totaro Theorem 6.4 ]. This will play a crucial role in testify theorem \[theorem: restriction\ ]. \[theorem: unif\ ] lease $ \mathscr{L}$ be a $ q$-ample line bundle on $ X$. Then for any $ N$, there is an integer $ m_{N}$, such that, for any coherent bundle $ \mathscr{F}$ on $ X$ with $ reg^{q'}(\mathscr{F})\leq N$, $ $ H^{i}(X,\mathscr{F}\otimes\mathscr{L}^{\otimes m})=0$$ for $ i > q+q'$ and $ m > m_{N}$. Fix an integer $ i$ such that $ q+q' < i \leq n$, by lemma \[lemma: vanishing of tensor product\ ], it is enough to indicate that there is an $ M$, depending merely on the choice of $ N$, but not the coherent sheaf $ \mathscr{F}$, such that for $ megabyte > M$, $ 0\leq a \leq 2n - i$ and $ b\geq 0 $, either $ H^{b}(X,\mathscr{L}^{\otimes m}\otimes \mathscr{O}_{X}(-N)\otimes \mathscr{R}_{a})=0 $ or $ H^{i+a - b}(X,\mathscr{F}\otimes\mathscr{O}_{X}(N - a))$. Here $ \mathscr{F}$ is any coherent sheaf with $ reg^{q'}(\mathscr{F})\leq N$. $ b > q$ and $ 0\leq a < n+b$. use the $ q$-ampleness of $ \mathscr{L}$, there constitute an $ m_N$, such that we have $ $ H^{q+j}(X,\mathscr{L}^{\otimes m}\otimes \mathscr{O}_{X}(-N - j))=0$$ for all $ 1\leq j\leq n - q$ and $ m > m_N$, i.e. $ \mathscr{L}^{\otimes m}\otimes \mathscr{O}_X(-N)$ is $ q$-regular for all $ m > m_N$. Now lemma \[lemma: R\_i\ ] says $ $ H^{b}(X,\mathscr{L}^{\otimes m}\otimes \mathscr{O}_{X}(-N)\otimes \mathscr{R}_{a})=0$$ for all $ m > m_N$, $ bacillus > q$ and $

\[lelma: regularity\]. We next pruve an analogue of [@Totero Thekrem 6.4]. Thks will play a crucial role mn peovint theorem \[theorem: restfiction\]. \[thvorem: unid\] Leu $\mathscr{L}$ be a $q$-ejple likz bunspe ou $E$. Then for any $K$, there is dn integer $m_{N}$, vuzh that, for any coherent sheaf $\mathscw{F}$ on $X$ wlth $reg^{q'}(\mathsct{F})\leq G$, $$H^{i}(S,\mathscr{F}\otimes\mathscr{L}^{\otimes m})=0$$ fod $i>q+q'$ aid $m>m_{N}$. Fix an inyeger $i$ such that $q+q'< i \leq n$, bj lemma \[lemma: vanidhing of tebsor product\], it ir enough to show that jhere is an $M$, depending only on ghe ckoice of $N$, vur nlj the cohereit shewf $\mathscr{F}$, such thad for $m>K$, $0\leq a \leq 2n-i$ anv $b\gwq 0$, either $H^{b}(X,\mathscr{N}^{\otimes m}\otimes \majhscr{O}_{X}(-N)\othmzs \mathscr{R}_{a})=0$ or $H^{i+a-b}(X,\narhscr{X}\otikes\mxrhszr{O}_{S}(N-e))$. Hsre $\mahhsrr{F}$ is any doherent shwaf with $reg^{q'}(\mathsct{F})\jva N$. $b> q$ and $0\lsq a < g+b$. Tsing the $q$-ampleness of $\mathscr{L}$, there ps ah $m_N$, such that we have $$Y^{q+j}(X,\mathscr{L}^{\otimes m}\ojimes \mathfcr{O}_{X}(-N-j))=0$$ for all $1\leq j\leq n-q$ and $m>m_N$, i.e. $\mathscr{L}^{\othmes j}\utinef \nahhscr{O}_X(-N)$ is $q$-regular for all $m>m_N$. Now lemma \[leija: R\_p\] says $$H^{b}(X,\mathscr{J}^{\otimes m}\otomfs \iathscr{O}_{X}(-N)\otioes \macgsdr{R}_{a})=0$$ for all $m>m_N$, $b> q$ and $

\[lemma: regularity\]. We next prove an analogue Theorem This will a crucial role \[theorem: Let $\mathscr{L}$ be $q$-ample line bundle $X$. Then for any $N$, there an integer $m_{N}$, such that, for any coherent sheaf $\mathscr{F}$ on $X$ with N$, $$H^{i}(X,\mathscr{F}\otimes\mathscr{L}^{\otimes m})=0$$ for $i>q+q'$ and $m>m_{N}$. Fix an integer $i$ such that i n$, lemma vanishing of tensor product\], it is enough to show that there is an $M$, depending only the choice of $N$, but not the coherent $\mathscr{F}$, such that for $0\leq a \leq 2n-i$ and 0$, $H^{b}(X,\mathscr{L}^{\otimes m}\otimes \mathscr{R}_{a})=0$ $H^{i+a-b}(X,\mathscr{F}\otimes\mathscr{O}_{X}(N-a))$. $\mathscr{F}$ is any sheaf with $reg^{q'}(\mathscr{F})\leq N$. $b> q$ and $0\leq a < n+b$. Using the $q$-ampleness of $\mathscr{L}$, there an $m_N$, we have m}\otimes for $1\leq j\leq n-q$ i.e. $\mathscr{L}^{\otimes m}\otimes \mathscr{O}_X(-N)$ is $q$-regular Now lemma \[lemma: R\_i\] says $$H^{b}(X,\mathscr{L}^{\otimes m}\otimes \mathscr{O}_{X}(-N)\otimes for all $b> q$ and $

\[lemma: regularity\]. We next provE an analoguE of [@ToTarO ThEoRem 6.4]. THis wIll play a cruciaL Role In proving theorem \[theoreM: restRiCTion\]. \[THeOrem: uNif\] Let $\mAThSCR{L}$ bE a $Q$-aMplE lINe BundlE on $x$. Then foR any $N$, there Is aN iNteger $m_{N}$, such THaT, for any cohEreNt sheaf $\mathsCr{F}$ On $X$ witH $rEg^{q'}(\MAthscR{F})\lEq N$, $$H^{i}(x,\mathsCR{F}\otimEs\mathscr{l}^{\oTImes m})=0$$ fOR $i>q+q'$ and $M>M_{n}$. FIx an Integer $i$ such that $q+Q'< I \lEQ n$, by lemma \[lemma: VanishInG Of TENsoR prOduct\], it is eNoUgh to SHow that THeRE IS an $m$, Depending only On the choice OF $N$, bUt not tHe CohERent shEaf $\maThSCr{F}$, Such that for $M>M$, $0\leQ a \leq 2n-i$ anD $b\geq 0$, eITher $H^{b}(X,\MAthscr{L}^{\Otimes M}\otImeS \matHScR{O}_{x}(-N)\oTiMEs \mAThScr{r}_{A})=0$ or $h^{i+a-b}(X,\matHsCr{f}\otimEs\maTHSCR{O}_{X}(N-A))$. HeRe $\maThscr{f}$ is any coherenT shEaf wITh $rEg^{q'}(\maThscr{f})\leq n$. $b> Q$ and $0\lEq a < n+b$. USing tHe $Q$-ampleness of $\matHscr{l}$, there is aN $m_N$, SuCh tHaT we haVE $$H^{q+j}(X,\mAthScr{l}^{\otimes M}\otimes \MAthScR{o}_{x}(-n-j))=0$$ For all $1\leq j\leq n-q$ and $M>m_n$, I.E. $\mAthscr{L}^{\oTimes m}\OTiMeS \Mathscr{O}_x(-N)$ Is $q$-ReguLAR for aLl $m>m_n$. noW lemma \[leMma: R\_i\] sAYs $$h^{b}(x,\mathscR{L}^{\Otimes M}\oTimEs \mAthscR{o}_{X}(-N)\oTimes \mAthscr{R}_{a})=0$$ For alL $M>m_N$, $b> q$ and $

\[lemma: regularity\]. W e next pro ve an an alo gu e of [@T otaro Theorem6 .4]. This will play a cruc ial r ol e inp ro vingtheorem \[ t h eor em :res tr i ct ion\] . \[theor em: unif\] Le t$\mathscr{L} $ b e a $q$-am ple line bundle on $X$.Th enf or an y $ N$, t here i s an in teger $m_ {N } $, suc h that,f o ranycoherent sheaf $\ m at h scr{F}$ on $X$ with$r e g^ { q '}( \ma thscr{F})\ le q N$, $$H^{i} ( X, \ m a ths c r{F}\otimes\m athscr{L}^{ \ oti mes m} )= 0$$ for $i >q+q' $a nd$m>m_{N}$. Fix an integ er $i$ such th a t $q+q' < i \l eqn$, byl em ma \[ le m ma: va nis h ing of tens or p roduc t\], i t is e nou gh t o sho w that thereisan $ M $,depen dingonly o n the choic e of$N $, but not thecohe rent shea f $ \m ath sc r{F}$ , suchtha t f or $m>M $, $0\l e q a \ l e q 2 n-i$ and $b\geq 0$ ,e i th er $H^{b }(X,\m a th sc r {L}^{\ot im esm}\o t i mes \ math s cr {O}_{X}( -N)\ot i me s\mathsc r{ R}_{a} )= 0$or$H^{i + a-b} (X,\ma thscr{F} \otim e s\mathscr{O}_{ X }(N-a))$. Her e $ \ m at h scr{ F}$ is any coh eren t she af w i th $r e g^{q' }(\ma th s cr { F})\leq N$. $b> q$ a nd $0\ leq a < n+b$. Usi ng the $q$ - a m plenessof $ \ ma t hscr{L}$, ther e isan $m_N$,s uch that we h ave $$H^ {q+j}(X,\ m a thscr{L} ^{\ oti mes m} \ o ti mes \mathscr{ O } _{X} (- N-j))=0 $$for all $1 \le q j \le qn-q$ and$m>m_N$, i .e .$\ mat hscr{ L }^{\otim es m} \o tim es \m a thscr{ O}_X( -N)$ i s$ q$- regular fo r all$m >m _N$. No wlemma \[l e mma : R\_i\ ] says $$ H^{ b }(X, \m at hscr{L} ^{\otimes m}\ ot imes \math sc r{O }_{X}( - N )\otimes \mathscr{R}_{a})=0$$ f o r all $ m>m _N$,$b>q$ and $

\[lemma:_regularity\]. We next_prove an analogue of_[@Totaro Theorem_6.4]._This will_play_a crucial role_in proving theorem_\[theorem: restriction\]. \[theorem: unif\] Let_$\mathscr{L}$ be a_$q$-ample_line bundle on $X$. Then for any $N$, there is an integer $m_{N}$, such_that,_for any_coherent_sheaf_$\mathscr{F}$ on $X$ with $reg^{q'}(\mathscr{F})\leq_N$, $$H^{i}(X,\mathscr{F}\otimes\mathscr{L}^{\otimes m})=0$$ for $i>q+q'$_and $m>m_{N}$. Fix_an integer $i$ such that $q+q'< i \leq_n$,_by lemma \[lemma:_vanishing of tensor product\], it is enough to show_that there is an $M$, depending_only on the_choice_of_$N$, but not the_coherent sheaf $\mathscr{F}$, such that for_$m>M$, $0\leq a \leq 2n-i$ and_$b\geq 0$, either $H^{b}(X,\mathscr{L}^{\otimes m}\otimes \mathscr{O}_{X}(-N)\otimes \mathscr{R}_{a})=0$_or $H^{i+a-b}(X,\mathscr{F}\otimes\mathscr{O}_{X}(N-a))$. Here $\mathscr{F}$ is any_coherent sheaf with $reg^{q'}(\mathscr{F})\leq N$. $b>_q$ and_$0\leq a < n+b$. Using the_$q$-ampleness of $\mathscr{L}$,_there is_an $m_N$, such_that we have $$H^{q+j}(X,\mathscr{L}^{\otimes m}\otimes \mathscr{O}_{X}(-N-j))=0$$_for all $1\leq_j\leq n-q$ and $m>m_N$, i.e. $\mathscr{L}^{\otimes_m}\otimes_\mathscr{O}_X(-N)$ is $q$-regular_for_all_$m>m_N$. Now_lemma \[lemma: R\_i\]_says_$$H^{b}(X,\mathscr{L}^{\otimes m}\otimes_\mathscr{O}_{X}(-N)\otimes_\mathscr{R}_{a})=0$$ for all $m>m_N$, $b> q$_and_$

t}$ has a representative in ${\overline{{\mathbb G}}}^{\theta,\circ}\cap \pi({{\mathbb G}}^{F_w})$ if and only if $v\in W^\theta_w$. \(1) If $\v {{\mathbb{T}}}\cap {{\mathbb G}}^\theta\neq\emptyset$, then, $\theta(\v)\in \v{{\mathbb{T}}}$, so $v\in W^\theta$ and we may assume $\v\in{{\mathbb G}}^\theta$. If $\v{{\mathbb{T}}}\cap{{\mathbb G}}^{F_w}\neq\emptyset$, then $F_w(\v)\in \v{{\mathbb{T}}}$, that is ${\operatorname{Ad}}(\w)(\v)\in\v{{\mathbb{T}}}$, [*i. e.*]{} $wv=vw$. Conversely, assume $v\in W^\theta_w$. Now $W^\theta$ is the Weyl group of ${{\mathbb G}}^\theta$ and $F_w$ is a Steinberg endomorphism of ${{\mathbb G}}^\theta$ preserving its maximal torus ${{\mathbb{T}}}^\theta$. By [@MT Proposition 23.2 ff], $$(W^\theta)^{F_w}=\left(N_{{{\mathbb G}}^\theta}({{\mathbb{T}}}^\theta)/{{\mathbb{T}}}^\theta\right)^{F_w}\simeq N_{{{\mathbb G}}^\theta\cap{{\mathbb G}}^{F_w}}({{\mathbb{T}}}^\theta)/({{\mathbb{T}}}^\theta\cap{{\mathbb{T}}}^{F_w})$$ so any $v\in W_w^\theta=(W^\theta)^{F_w}$ has a representative in $$N_{{{\mathbb G}}^\theta\cap{{\mathbb G}}^{F_w}}({{\mathbb{T}}}^\theta)=N_{{{\mathbb G}}^\theta}({{\mathbb{T}}}^\theta)\cap {{\mathbb G}}^{F_w}=N_{{{\mathbb G}}^\theta}({{\mathbb{T}}})\cap {{\mathbb G}}^{F_w}=N_{{{\mathbb G}}}({{\mathbb{T}}})\cap {{\mathbb G}}^\theta\cap {{\mathbb G}}^{F_w}.$$ \(2) Follows from (1) recalling that $\pi({{\mathbb G}}^\theta)={\overline{{\mathbb G}}}^{\theta,\circ}$. We end the section with a lemma that shows how some some of the results on the Weyl group apply to the quest of preferred representatives in a twisted class.

t}$ has a representative in $ { \overline{{\mathbb G}}}^{\theta,\circ}\cap \pi({{\mathbb G}}^{F_w})$ if and only if $ v\in W^\theta_w$. \(1) If $ \v { { \mathbb{T}}}\cap { { \mathbb G}}^\theta\neq\emptyset$, then, $ \theta(\v)\in \v{{\mathbb{T}}}$, so $ v\in W^\theta$ and we may wear $ \v\in{{\mathbb G}}^\theta$. If $ \v{{\mathbb{T}}}\cap{{\mathbb G}}^{F_w}\neq\emptyset$, then $ F_w(\v)\in \v{{\mathbb{T}}}$, that is $ { \operatorname{Ad}}(\w)(\v)\in\v{{\mathbb{T}}}$, [ * i.   e. * ] { } $ west virginia = vw$. Conversely, assume $ v\in W^\theta_w$. Now $ W^\theta$ is the Weyl group of $ { { \mathbb G}}^\theta$ and $ F_w$ is a Steinberg endomorphism of $ { { \mathbb G}}^\theta$ save its maximal torus $ { { \mathbb{T}}}^\theta$. By [ @MT Proposition 23.2 ff ], $ $ (W^\theta)^{F_w}=\left(N_{{{\mathbb G}}^\theta}({{\mathbb{T}}}^\theta)/{{\mathbb{T}}}^\theta\right)^{F_w}\simeq N_{{{\mathbb G}}^\theta\cap{{\mathbb G}}^{F_w}}({{\mathbb{T}}}^\theta)/({{\mathbb{T}}}^\theta\cap{{\mathbb{T}}}^{F_w})$$ then any $ v\in W_w^\theta=(W^\theta)^{F_w}$ has a representative in $ $ N_{{{\mathbb G}}^\theta\cap{{\mathbb G}}^{F_w}}({{\mathbb{T}}}^\theta)=N_{{{\mathbb G}}^\theta}({{\mathbb{T}}}^\theta)\cap { { \mathbb G}}^{F_w}=N_{{{\mathbb G}}^\theta}({{\mathbb{T}}})\cap { { \mathbb G}}^{F_w}=N_{{{\mathbb G}}}({{\mathbb{T}}})\cap { { \mathbb G}}^\theta\cap { { \mathbb G}}^{F_w}.$$ \(2) surveil from (1) recalling that $ \pi({{\mathbb G}}^\theta)={\overline{{\mathbb G}}}^{\theta,\circ}$. We end the section with a lemma that read how some some of the results on the Weyl group apply to the pursuit of preferred representatives in a twisted class.

t}$ hws a representative in ${\oyerline{{\mathbb G}}}^{\tkwta,\cirr}\cap \pi({{\jathbb G}}^{W_w})$ if and only if $v\in W^\theta_x$. \(1) If $\v {{\maukbb{T}}}\cap {{\mathbb G}}^\thetx\neq\emptydet$, then, $\theua(\v)\in \v{{\mathbb{T}}}$, so $v\in W^\thcca$ ans we key assume $\v\in{{\majhbb G}}^\theta$. Hf $\v{{\mathbb{T}}}\cap{{\kaghyb G}}^{F_w}\neq\emptyset$, then $F_w(\v)\in \v{{\mathbb{E}}}$, that os ${\operatorname{Aq}}(\w)(\v)\im\d{{\matgbb{T}}}$, [*i. e.*]{} $wv=vw$. Conversely, assume $v\in W^\theta_x$. Now $W^\theta$ is the Weyl group of ${{\mathbb H}}^\theha$ and $F_w$ is a Stelnberg endonorprusm of ${{\mathbc G}}^\theta$ pgzserving ita maximal torus ${{\mathbb{T}}}^\theta$. By [@MT Pxoposition 23.2 fd], $$(W^\hveta)^{F_w}=\left(N_{{{\nathbf G}}^\theta}({{\mathnn{T}}}^\thetd)/{{\mathbb{Y}}}^\theta\right)^{F_w}\slmeq I_{{{\matybb G}}^\theta\cap{{\mathbb G}}^{H_w}}({{\mathbb{T}}}^\theta)/({{\mathbb{J}}}^\theta\cap{{\mdtkbb{T}}}^{F_w})$$ so any $v\in W_w^\tyera=(W^\thgta)^{F_w}$ has q rdprtseitafive ij $$N_{{{\jathbb G}}^\thsta\cap{{\mathbv G}}^{F_w}}({{\mathbb{T}}}^\theta)=N_{{{\mstrvb G}}^\theta}({{\mathgb{T}}}^\theea)\sap {{\mathbb G}}^{F_w}=N_{{{\mathbb G}}^\theta}({{\mathbb{T}}})\cap {{\katgbb G}}^{F_w}=N_{{{\mathbb G}}}({{\mathbb{T}}})\xap {{\mathbb G}}^\theta\cap {{\lathbb G}}^{F_r}.$$ \(2) Follows from (1) recalling that $\pi({{\mathbb G}}^\theta)={\ovesline{{\jxthyn E}}}^{\rhfta,\circ}$. We end the section with a lemma that srkwx mow some some of the resultx ln jhe Weyl group apply to the quest of preffrred rgpresebtatives yn a twisted class.

t}$ has a representative in ${\overline{{\mathbb G}}}^{\theta,\circ}\cap if only if W^\theta_w$. \(1) If $\theta(\v)\in so $v\in W^\theta$ we may assume G}}^\theta$. If $\v{{\mathbb{T}}}\cap{{\mathbb G}}^{F_w}\neq\emptyset$, then $F_w(\v)\in that is ${\operatorname{Ad}}(\w)(\v)\in\v{{\mathbb{T}}}$, [*i. e.*]{} $wv=vw$. Conversely, assume $v\in W^\theta_w$. Now $W^\theta$ is Weyl group of ${{\mathbb G}}^\theta$ and $F_w$ is a Steinberg endomorphism of ${{\mathbb preserving maximal ${{\mathbb{T}}}^\theta$. [@MT Proposition 23.2 ff], $$(W^\theta)^{F_w}=\left(N_{{{\mathbb G}}^\theta}({{\mathbb{T}}}^\theta)/{{\mathbb{T}}}^\theta\right)^{F_w}\simeq N_{{{\mathbb G}}^\theta\cap{{\mathbb G}}^{F_w}}({{\mathbb{T}}}^\theta)/({{\mathbb{T}}}^\theta\cap{{\mathbb{T}}}^{F_w})$$ so any $v\in W_w^\theta=(W^\theta)^{F_w}$ has a representative $$N_{{{\mathbb G}}^\theta\cap{{\mathbb G}}^{F_w}}({{\mathbb{T}}}^\theta)=N_{{{\mathbb G}}^\theta}({{\mathbb{T}}}^\theta)\cap {{\mathbb G}}^{F_w}=N_{{{\mathbb G}}^\theta}({{\mathbb{T}}})\cap {{\mathbb G}}}({{\mathbb{T}}})\cap {{\mathbb G}}^\theta\cap {{\mathbb \(2) Follows from (1) recalling $\pi({{\mathbb G}}}^{\theta,\circ}$. We the with lemma that shows some some of the results on the Weyl group apply to the quest of preferred representatives in twisted class.

t}$ has a representative in ${\overLine{{\mathbb g}}}^{\thetA,\ciRc}\cAp \Pi({{\maThbb g}}^{F_w})$ if and only if $V\In W^\tHeta_w$. \(1) If $\v {{\mathbb{T}}}\cap {{\mathBb G}}^\thEtA\Neq\eMPtYset$, tHen, $\thetA(\V)\iN \V{{\MatHbB{T}}}$, So $v\In w^\ThEta$ anD we May assuMe $\v\in{{\mathbB G}}^\tHeTa$. If $\v{{\mathbb{T}}}\CAp{{\Mathbb G}}^{F_w}\nEq\eMptyset$, then $F_W(\v)\iN \v{{\mathBb{t}}}$, thAT is ${\opEraTornaMe{Ad}}(\w)(\v)\IN\v{{\mathBb{T}}}$, [*i. e.*]{} $wv=vw$. coNVerselY, Assume $v\IN w^\tHeta_W$. Now $W^\theta$ is the WeYL gROup of ${{\mathbb G}}^\thEta$ and $f_w$ IS a sTEinBerG endomorphIsM of ${{\maTHbb G}}^\theTA$ pRESErvINg its maximal tOrus ${{\mathbb{T}}}^\THetA$. By [@MT PRoPosITion 23.2 ff], $$(w^\thetA)^{F_W}=\LefT(N_{{{\mathbb G}}^\thEta}({{\mAthbb{T}}}^\theTa)/{{\mathBB{T}}}^\theta\RIght)^{F_w}\sImeq N_{{{\mAthBb G}}^\ThetA\CaP{{\mAthBb g}}^{f_w}}({{\mAThBb{T}}}^\THetA)/({{\mathbb{T}}}^\ThEtA\cap{{\mAthbB{t}}}^{f_W})$$ So anY $v\iN W_w^\tHeta=(W^\Theta)^{F_w}$ has a rePreSentATivE in $$N_{{{\mAthbb g}}^\theTa\Cap{{\maThbb G}}^{F_W}}({{\mathBb{t}}}^\theta)=N_{{{\mathbb G}}^\tHeta}({{\Mathbb{T}}}^\thEta)\CaP {{\maThBb G}}^{F_w}=n_{{{\Mathbb g}}^\thEta}({{\Mathbb{T}}})\Cap {{\mathBB G}}^{F_W}=N_{{{\MATHbB G}}}({{\mathbb{T}}})\cap {{\mathbb g}}^\tHETa\Cap {{\mathbB G}}^{F_w}.$$ \(2) FoLLoWs FRom (1) recalLiNg tHat $\pI({{\MAthbb g}}^\theTA)={\oVerline{{\mAthbb G}}}^{\THeTa,\Circ}$. We eNd The secTiOn wIth A lemmA That Shows hOw some soMe of tHE results on the WEYl group apply tO ThE QUeST of pRefErred represEntaTIves In a tWIsTed CLass.

t}$ has a representative i n ${\overl ine{{ \ma thb bG}}} ^{\t heta,\circ}\ca p \pi ({{\mathbb G}}^{F_w})$ if a nd only if $v\i n W^\th e ta _ w $. \ (1 ) I f$ \v {{\m ath bb{T}}} \cap {{\ma thb bG}}^\theta\n e q\ emptyset$, th en, $\theta( \v) \in \v {{ \ma t hbb{T }}} $, so $v\in W^\the ta$ and w em ay ass u me $\v\ i n {{ \mat hbb G}}^\theta$.I f$ \v{{\mathbb{T} }}\cap {{ \ ma t h bbG}} ^{F_w}\neq \e mptys e t$, the n $ F _ w (\v ) \in \v{{\math bb{T}}}$, t h atis ${\ op era t orname {Ad}} (\ w )(\ v)\in\v{{\m athb b{T}}}$,[*i. e . *]{} $w v =vw$. C onvers ely , a ssum e $ v\ inW^ \ the t a_ w$. Now $W^\the ta $is th e We y l g roup of ${{ \math bb G}}^\theta $ a nd $ F _w$ is a Stei nber gendom orphis m of${ {\mathbb G}}^\t heta $ preserv ing i tsma ximal torus${{ \ma thbb{T} }}^\the t a$. B y [ @M T Proposition 23.2 f f ] ,$$(W^\th eta)^{ F _w }= \ left(N_{ {{ \ma thbb G }}^\t heta } ({ {\mathbb {T}}}^ \ th et a)/{{\m at hbb{T} }} ^\t het a\rig h t)^{ F_w}\s imeq N_ {{{\m a thbb G}}^\thet a \cap{{\mathbb G} } ^ {F _ w}}( {{\ mathbb{T}}} ^\th e ta)/ ({{\ m at hbb { T}}}^ \thet a\ c ap { {\mathbb{T}}}^{F_w} )$ $ so a ny $v \in W_w^\thet a=(W^\thet a ) ^ {F_w}$ h as a re p resentative in $$N_ {{{\mathbb G}}^\the ta\ca p{{\math bb G}}^{F _ w }}({{\ma thb b{T }}} ^\t h e ta )=N_{{{\mathb b G}}^ \t heta}({ {\m athbb{T }}} ^\t het a)\ ca p {{\math bb G}}^{ F_ w} =N _{ {{\ mathb b G}}^\th et a}( {{ \ma thbb{ T }}})\c ap { {\ma th bb G}} ^{F_w}= N _{ { { \mat hb bG}}} ({{ \m athbb {T}} } )\c ap {{\m athbb G}} ^\t h eta\ ca p{{\math bb G}}^{F_w}. $$ \(2) Fol lo wsfrom ( 1 ) recalli ng that $\pi({{\mathbbG }}^\the ta) ={\ov erli ne{{\math bbG}}}^{ \th e ta,\ci rc}$. We e nd th e secti o n w ith a lemma tha t sho ws ho wsome some o f the results on t h e W eyl group app lyto t h e q ues t o f pr ef e rre d representatives in a twis te d c lass.

t}$ has_a representative_in ${\overline{{\mathbb G}}}^{\theta,\circ}\cap \pi({{\mathbb_G}}^{F_w})$ if_and_only if_$v\in_W^\theta_w$. \(1) If $\v_{{\mathbb{T}}}\cap {{\mathbb G}}^\theta\neq\emptyset$,_then, $\theta(\v)\in \v{{\mathbb{T}}}$, so_$v\in W^\theta$ and_we_may assume $\v\in{{\mathbb G}}^\theta$. If $\v{{\mathbb{T}}}\cap{{\mathbb G}}^{F_w}\neq\emptyset$, then $F_w(\v)\in \v{{\mathbb{T}}}$, that is ${\operatorname{Ad}}(\w)(\v)\in\v{{\mathbb{T}}}$, [*i. e.*]{}_$wv=vw$._Conversely, assume_$v\in_W^\theta_w$._Now $W^\theta$ is the Weyl_group of ${{\mathbb G}}^\theta$ and_$F_w$ is_a Steinberg endomorphism of ${{\mathbb G}}^\theta$ preserving its_maximal_torus ${{\mathbb{T}}}^\theta$. By_[@MT Proposition 23.2 ff], $$(W^\theta)^{F_w}=\left(N_{{{\mathbb G}}^\theta}({{\mathbb{T}}}^\theta)/{{\mathbb{T}}}^\theta\right)^{F_w}\simeq N_{{{\mathbb G}}^\theta\cap{{\mathbb G}}^{F_w}}({{\mathbb{T}}}^\theta)/({{\mathbb{T}}}^\theta\cap{{\mathbb{T}}}^{F_w})$$_so any $v\in W_w^\theta=(W^\theta)^{F_w}$ has a_representative in $$N_{{{\mathbb_G}}^\theta\cap{{\mathbb_G}}^{F_w}}({{\mathbb{T}}}^\theta)=N_{{{\mathbb_G}}^\theta}({{\mathbb{T}}}^\theta)\cap {{\mathbb G}}^{F_w}=N_{{{\mathbb G}}^\theta}({{\mathbb{T}}})\cap_ {{\mathbb G}}^{F_w}=N_{{{\mathbb G}}}({{\mathbb{T}}})\cap {{\mathbb G}}^\theta\cap {{\mathbb_G}}^{F_w}.$$ \(2) Follows from (1) recalling that_$\pi({{\mathbb G}}^\theta)={\overline{{\mathbb G}}}^{\theta,\circ}$. We end the section with_a lemma that shows how some_some of the results on_the Weyl_group apply to the quest_of preferred representatives_in a_twisted class.

\], we state some relations about periodic points of different liftings, we recall the notions of covering and positive covering and give some of their properties, which are key tools for finding periodic points. In Section \[sec:Y\], we prove Corollary \[ConverseSigmaCircleCase\] and Theorems \[YinSigma\] and \[ConverseEndInteger\]. In Section \[WeAreInTheBranches\], we prove Theorems \[TheoremSharkovskiiintheBranches\] and \[LargeOrbitsintheBranches\]. Section \[sec:0inIntRotR\], devoted to Theorem \[theo:0inInterior\], starts with the construction of examples, then states some more technical lemmas about the set of periods and finally gives the proof of Theorem \[theo:0inInterior\]. In the last section, we stress some difficulties in the characterization of the set of periods: a first example shows that, in Theorem \[theo:0inInterior\], one cannot replace $\operatorname{Per}(F)$ by $\operatorname{Per}(0,F)$ (i.e., periods [[$\kern -0.55em\pmod{1}$]{}]{} by true periods), which is an obstacle to apply to $\sigma$ maps the same method as for circle maps; two other examples show that orderings $\leso{n}$ with $n>3$ may be needed to characterize $\operatorname{Per}(0,F)$, which might let us think that, in the first statement of Conjecture \[WishList\], considering orderings $\leso{2}$ and $\leso{3}$ may not be sufficient. Coverings and periodic points {#sec:covering} ============================= Relations between periodic points of *F* and of *F+k* ----------------------------------------------------- Next easy lemma summarizes some basic properties of liftings; in particular, periodic [[$\kern -0.55em\pmod{1}$]{}]{} points do not depend on the choice of the lifting of a given $\sigma$-map. \[lem:FF+k\] Let $F\in{\ensuremath{\mathcal{L}_{d}(S)}}$. The following statements hold for all $k,m\in{\ensuremath{\mathbb{Z}}}$ and all $n\ge 0$: 1. $F^n(x+m)=F^n(x)+md^n$; in particular, if $

\ ], we state some relations about periodic points of different liftings, we echo the notion of covering and positive covering and hold some of their properties, which are key cock for receive periodic points. In Section   \[sec: Y\ ], we testify Corollary   \[ConverseSigmaCircleCase\ ] and Theorems   \[YinSigma\ ] and \[ConverseEndInteger\ ]. In Section   \[WeAreInTheBranches\ ], we raise Theorems   \[TheoremSharkovskiiintheBranches\ ] and \[LargeOrbitsintheBranches\ ]. Section   \[sec:0inIntRotR\ ], give to Theorem   \[theo:0inInterior\ ], startle with the construction of examples, then states some more technical lemmas about the set of time period and finally gives the proof of Theorem   \[theo:0inInterior\ ]. In the final section, we stress some difficulties in the word picture of the set of periods: a first example shows that, in Theorem   \[theo:0inInterior\ ], one cannot supplant $ \operatorname{Per}(F)$ by $ \operatorname{Per}(0,F)$ (i.e., periods [ [ $ \kern -0.55em\pmod{1}$ ] { } ] { } by true periods), which is an obstacle to apply to $ \sigma$ maps the same method as for circle maps; two other examples show that orderings $ \leso{n}$ with $ n>3 $ may be needed to characterize $ \operatorname{Per}(0,F)$, which might let us think that, in the inaugural statement of Conjecture   \[WishList\ ], considering order $ \leso{2}$ and $ \leso{3}$ may not be sufficient. cover and periodic points { # sec: covering } = = = = = = = = = = = = = = = = = = = = = = = = = = = = = Relations between periodic point of * F * and of * F+k * ----------------------------------------------------- Next easy lemma summarizes some basic properties of liftings; in especial, periodic [ [ $ \kern -0.55em\pmod{1}$ ] { } ] { } points do not depend on the choice of the lifting of a given $ \sigma$-map. \[lem: FF+k\ ] Let $ F\in{\ensuremath{\mathcal{L}_{d}(S)}}$. The following instruction hold for all $ k, m\in{\ensuremath{\mathbb{Z}}}$ and all $ n\ge 0 $: 1. $ F^n(x+m)=F^n(x)+md^n$; in particular, if $

\], we state some relations abuut periodic ponbts of diffedent lifgings, we recall the notions lf coveeing and positive covefing and hive somw of rheir propxdties, wmnch adc key vools for findikg periodic points. In Secdiun \[dec:Y\], we prove Corollary \[ConverseSigmwCircleVade\] and Theoremf \[YinXygma\] and \[ConverseEndInteger\]. In Section \[SeAreInUheBranches\], we profe Theorems \[TheoremSharkovsniiijtheBranches\] and \[LwrgeOrbitsibtheFeanches\]. Sectkon \[sec:0inInuRmtR\], devotes to Theorem \[theo:0inInterior\], stargs wich the consjxyctlmn of exampoes, tren states some more technival lemmas abomt thx ser of periods and finanly gives the proos of Theosej \[theo:0inInterior\]. Ib rhe ldst vectkin, de atcesa some dihficulties jn the charqcterization of the sqn of periods: z firse qxample shows that, in Theorem \[theo:0inIntegior\], one cannot replace $\opeeatorname{Per}(F)$ by $\operwtorname{Pqr}(0,F)$ (i.e., periods [[$\kern -0.55em\pmod{1}$]{}]{} by true periods), which hs an ubsuagjd tl apply to $\sigma$ maps the same method as for sjrvlv maps; two other cxamples show that ogdrtings $\leso{n}$ wijh $n>3$ mav bs needed to characherize $\jperarorname{Pew}(0,F)$, wnich might let us think thar, in the firfr statement of Conlecture \[WishLnst\], comsideting orderings $\leso{2}$ and $\lesk{3}$ may not bf sufficisvt. Coverings and oerpodiw points {#sec:covering} ============================= Relatijns betwexn pexiodic puintx of *F* and of *F+k* ----------------------------------------------------- Jext cdsy lemma summarizfs sole basic prooerties of liftings; in particulec, periodic [[$\ketn -0.55am\pkod{1}$]{}]{} poincs do kot depend on tre choice of tke liftiug of x given $\sifma$-map. \[lxm:FF+k\] Let $F\ig{\ensuremath{\madjcal{L}_{d}(S)}}$. The hollowing staremebts hola for all $k,m\in{\emsuremath{\mathbb{Z}}}$ ane all $n\ge 0$: 1. $F^n(x+m)=F^n(w)+md^n$; kh particular, if $

\], we state some relations about periodic different we recall notions of covering some their properties, which key tools for periodic points. In Section \[sec:Y\], we Corollary \[ConverseSigmaCircleCase\] and Theorems \[YinSigma\] and \[ConverseEndInteger\]. In Section \[WeAreInTheBranches\], we prove Theorems and \[LargeOrbitsintheBranches\]. Section \[sec:0inIntRotR\], devoted to Theorem \[theo:0inInterior\], starts with the construction of then some technical about the set of periods and finally gives the proof of Theorem \[theo:0inInterior\]. In the last we stress some difficulties in the characterization of set of periods: a example shows that, in Theorem one replace $\operatorname{Per}(F)$ $\operatorname{Per}(0,F)$ periods -0.55em\pmod{1}$]{}]{} by true which is an obstacle to apply to $\sigma$ maps the same method as for circle maps; two examples show $\leso{n}$ with may needed characterize $\operatorname{Per}(0,F)$, which us think that, in the first \[WishList\], considering orderings $\leso{2}$ and $\leso{3}$ may not sufficient. Coverings periodic points {#sec:covering} ============================= Relations between points of *F* and of *F+k* ----------------------------------------------------- Next lemma summarizes some basic properties of liftings; in particular, periodic [[$\kern -0.55em\pmod{1}$]{}]{} points do not the choice of the of a given \[lem:FF+k\] $F\in{\ensuremath{\mathcal{L}_{d}(S)}}$. following hold for $k,m\in{\ensuremath{\mathbb{Z}}}$ and all $n\ge 0$: 1. $F^n(x+m)=F^n(x)+md^n$; in particular, if $

\], we state some relations about Periodic poInts oF diFfeReNt liFtinGs, we recall the nOTionS of covering and positive CoverInG And gIVe Some oF their pROpERTieS, wHiCh aRe KEy Tools For Finding Periodic poIntS. IN Section \[sec:Y\], WE pRove CorollAry \[converseSigmACiRcleCaSe\] And tHeoreMs \[YInSigMa\] and \[CONverseendIntegeR\]. IN sectioN \[weAreIntHEBRancHes\], we prove TheoremS \[thEOremSharkovskiIinthebrANcHES\] anD \[LaRgeOrbitsiNtHeBraNChes\]. SecTIoN \[SEC:0iniNtRotR\], devoted To Theorem \[thEO:0ininteriOr\], StaRTs with The coNsTRucTion of exampLes, tHen states Some moRE techniCAl lemmaS about The Set Of peRIoDs And FiNAllY GiVes THe pRoof of ThEoReM \[theo:0InInTERIOr\]. In The Last SectiOn, we stress somE diFficULtiEs in tHe chaRactErIzatiOn of thE set oF pEriods: a first exaMple Shows that, In THeOreM \[tHeo:0iniNterioR\], onE caNnot repLace $\opeRAtoRnAME{peR}(F)$ by $\operatorname{PeR}(0,F)$ (I.E., PeRiods [[$\kerN -0.55em\pmoD{1}$]{}]{} By TrUE periods), WhIch Is an OBStaclE to aPPlY to $\sigma$ Maps thE SaMe Method aS fOr circLe MapS; twO otheR ExamPles shOw that orDerinGS $\leso{n}$ with $n>3$ may BE needed to charACtERIzE $\OperAtoRname{Per}(0,F)$, whIch mIGht lEt us THiNk tHAt, in tHe firSt STaTEment of Conjecture \[WiShlist\], coNsideRing orderings $\Leso{2}$ and $\lesO{3}$ MAY not be suFficIEnT. coverings and peRiodiC points {#sec:COvering} ============================= RElatiOns betweEn periodiC POints of *F* And Of *F+K* ----------------------------------------------------- NeXt eASY lEmma summarizeS SOme bAsIc propeRtiEs of lifTinGs; iN paRtiCuLar, periodIc [[$\kern -0.55em\PmOd{1}$]{}]{} PoInTs dO not dEPend on thE cHoiCe Of tHe lifTIng of a Given $\SigmA$-mAp. \[LEm:Ff+k\] Let $F\iN{\EnSURemaTh{\MaThcaL{L}_{d}(s)}}$. THe folLowiNG stAtementS hold for aLl $k,M\In{\enSuReMath{\matHbb{Z}}}$ and all $n\ge 0$: 1. $f^n(X+m)=F^n(x)+md^n$; in PaRtiCular, iF $

\], we state some relation s about pe riodi c p oin ts ofdiff erent liftings , werecall the notions ofcover in g and po sitiv e cover i ng a ndgi ve so me of thei r p roperti es, whichare k ey tools for fi nding peri odi c points. In Se ction\[ sec : Y\],weprove Corol l ary \[ ConverseS ig m aCircl e Case\]a n dTheo rems \[YinSigma\] an d \[ConverseEnd Intege r\ ] .I n Se cti on \[WeAre In TheBr a nches\] , w e p rov e Theorems \[T heoremShark o vsk iiinth eB ran c hes\]and \ [L a rge Orbitsinthe Bran ches\]. S ection \[sec:0 i nIntRot R\], d evo ted toT he or em\[ t heo : 0i nIn t eri or\], st ar ts with the c o n stru cti on o f exa mples, then s tat es s o memoretechn ical l emmas about these t of periods an d fi nally giv esth e p ro of of Theore m \ [th eo:0inI nterior \ ].In t h elast section, we s tr e s ssome dif ficult i es i n the cha ra cte riza t i on of the se t of per iods:a f ir st exam pl e show stha t,in Th e orem  \[the o:0inInt erior \ ], one cannotr eplace $\oper a to r n am e {Per }(F )$ by $\ope rato r name {Per } (0 ,F) $ (i.e ., pe ri o ds [[$\kern -0.55em\pm od {1}$]{ }]{}by true perio ds), which i s an obst acle to apply to $\sig ma$ m aps the sa m e method as f or circl e maps; t w o other e xam ple s s how t ha t orderings $ \ l eso{ n} $ with$n> 3$ maybenee ded to c haracteri ze $\ope ra to rn am e{P er}(0 , F)$, whi ch mi gh t l et us thinkthat, inth ef irs t state m en t of C on je ctur e \ [W ishLi st\] , co nsideri ng orderi ngs $\le so {2 }$ and$\leso{3}$ ma ynot be suf fi cie nt. C o v erings a nd periodic points {#se c :coveri ng} ==== ==== ========= === ====== === Relat ions b etwee nper i o dic p o i nt s o f*F* and of * F+k * --- -- ---- ------- ------------------ - --- ------------- -- Nex t ea syl em m a s um m ari z e s some basic pr operties o fl if tings; inp art ic ular, p eriodic [[$\ k ern -0. 55em\pmod {1}$]{}]{ }poin t s do not depen d on the choice o f thel if tingofa give n$\s igma$ -map.\[l em:FF +k\] L et $F\in {\ens ur emath{\m athcal{L}_{d}(S)}}$. Th e foll owing st atementshol d fo r all $k, m\in {\ensurema th{ \ma thbb{ Z}} } $ and all $n \ge 0$: 1. $ F^n(x+m)= F ^n (x) + m d^ n$; in part i c u lar , if$

\], we_state some_relations about periodic points_of different_liftings,_we recall_the_notions of covering_and positive covering_and give some of_their properties, which_are_key tools for finding periodic points. In Section \[sec:Y\], we prove Corollary \[ConverseSigmaCircleCase\] and Theorems \[YinSigma\] and_\[ConverseEndInteger\]._In Section \[WeAreInTheBranches\],_we_prove_Theorems \[TheoremSharkovskiiintheBranches\] and \[LargeOrbitsintheBranches\]. Section \[sec:0inIntRotR\], devoted_to Theorem \[theo:0inInterior\], starts with the_construction of_examples, then states some more technical lemmas about_the_set of periods_and finally gives the proof of Theorem \[theo:0inInterior\]. In the_last section, we stress some difficulties_in the characterization_of_the_set of periods: a_first example shows that, in Theorem \[theo:0inInterior\],_one cannot replace $\operatorname{Per}(F)$ by $\operatorname{Per}(0,F)$_(i.e., periods [[$\kern -0.55em\pmod{1}$]{}]{} by true periods),_which is an obstacle to apply_to $\sigma$ maps the same_method as_for circle maps; two other_examples show that_orderings $\leso{n}$_with $n>3$ may_be needed to characterize $\operatorname{Per}(0,F)$, which_might let us_think that, in the first statement_of_Conjecture \[WishList\], considering orderings_$\leso{2}$_and_$\leso{3}$ may_not be sufficient. Coverings_and_periodic points_{#sec:covering} ============================= Relations_between periodic points of *F* and_of_*F+k* ----------------------------------------------------- Next easy lemma summarizes some basic properties_of liftings; in particular,_periodic_[[$\kern -0.55em\pmod{1}$]{}]{} points do_not depend on the choice_of the lifting of a given_$\sigma$-map. \[lem:FF+k\] Let_$F\in{\ensuremath{\mathcal{L}_{d}(S)}}$. The_following statements hold for all $k,m\in{\ensuremath{\mathbb{Z}}}$ and all $n\ge 0$: 1. _$F^n(x+m)=F^n(x)+md^n$; in particular, if $

, leaving a directed acyclic graph. Hence we can restrict our consideration to the cycles in $G'$. The only new" cycles in $G'$ (ie those that don’t have a corresponding cycle in $G$) are those that are created by being allowed to visit one or more path graphs multiple times. These effectively correspond to a union of cycles in $G$. Hence if there is no feedback arc set of size $k$ that removes all the cycles in $G$, there is definitely none of size $k$ in $G'$ either. Final Input Size: As argued above, the line graph of $G'$ will be sparse (ie have in-degree and out-degree bounded by a constant) and will have a vertex set of cardinality a linear function of $e$. It is clear that the input size of the converted problem is a linear function of the original input size. Directed HCP to Undirected HCP (Karp) {#directed-hcp-to-undirected-hcp-karp.unnumbered} ------------------------------------- Directed HCP: Does a given directed graph contain a simple cycle that visits every vertex?\ Input: Graph $G$ containing $N$ vertices and $e$ (directed) arcs.\ Input size: $e$. Undirected HCP: Does a given undirected graph contain a simple cycle that visits every vertex? Conversion: Define subgraphs $S_i$ for $i = 1, \hdots, N$ to be 3-vertex subgraphs each containing edges $(1,2)$ and $(2,3)$. We then construct a new instance of Undirected HCP by replacing each vertex $i$ in $G$ with $S_i$. Then for each directed arc $(i,j)$, the new instance contains an edge going from vertex 3 of $S_i$ to vertex 1 of $S_j$. Explanation: The second vertex in each $S_i$ is a degree 2 vertex, and so it is clear that any time vertex $1$ of any $S_i$ is reached, vertices 2 and then 3 of the same $S_i$ must immediately follow. Then it is only possible to exit each $S_i$ via an edge incident on the third vertex, which corresponds to an arc that departs vertex $i$ in $G$. Likewise, each time a $S_i$ is exited,

, leaving a directed acyclic graph. Hence we can restrict our circumstance to the cycle in $ G'$. The only new " cycles in $ G'$ (i those that do n’t have a corresponding cycle in $ G$) are those that are create by being allowed to visit one or more way graph multiple times. These effectively represent to a union of cycles in $ G$. therefore if there is no feedback arc hardening of size $ k$ that removes all the cycles in $ G$, there embody definitely none of size $ k$ in $ G'$ either. Final Input Size: As argued above, the wrinkle graph of $ G'$ will be sparse (ie have in - degree and out - degree bounded by a changeless) and will have a vertex set of cardinality a linear function of $ e$. It is clear that the stimulation size of the converted problem is a linear function of the original input size. Directed HCP to Undirected HCP (Karp) { # directed-hcp-to-undirected-hcp-karp.unnumbered } ------------------------------------- Directed HCP: Does a given directed graph contain a simple cycle that visits every vertex?\ Input: Graph $ G$ containing $ N$ vertices and $ e$ (directed) arcs.\ Input size: $ e$. Undirected HCP: Does a given adrift graph incorporate a simple bicycle that visit every vertex? Conversion: Define subgraphs $ S_i$ for $ i = 1, \hdots, N$ to be 3 - vertex subgraphs each containing edges $ (1,2)$ and $ (2,3)$. We then manufacture a new instance of Undirected HCP by replacing each vertex $ i$ in $ G$ with $ S_i$. Then for each directed discharge $ (i, j)$, the new instance contains an edge going from vertex 3 of $ S_i$ to vertex 1 of $ S_j$. Explanation: The second vertex in each $ S_i$ is a academic degree 2 vertex, and so it is clear that any time vertex $ 1 $ of any $ S_i$ is reached, vertices 2 and then 3 of the same $ S_i$ must immediately follow. Then it is only potential to die each $ S_i$ via an boundary incident on the third vertex, which corresponds to an arc that departs vertex $ i$ in $ G$. Likewise, each time a $ S_i$ is exited,

, lewving a directed acyclic graph. Hence we can revtrict our conrideration to the cycles in $J'$. Thw onlt new" cycles in $G'$ (ie tfose that don’t hace a xorrespondmhg cyclc in $F$) are vhose that are greated by teing allowed do vnsit one or more path graphs multiplq times. Tjese effectivejy cpwreslond to a union of cycles in $G$. Hende if tiere is no feednack arc set of size $k$ thah reloves all the cyclfs in $G$, thets if definitely vone of siee $k$ in $G'$ eijher. Final Input Size: As argued acove, che line grqpy ov $G'$ will be wparsv (ie have in-dcbree atd out-drgree bounded ny a ronsrant) and will have a tertex set of cardinwlity a lhnzar function of $e$. It us cleat thad thd inout smze of thf cknverted pdoblem is a linear function of trv original inlut sise. Qirected HCP to Undirected HCP (Karp) {#dirtcted-gcp-to-undirected-hcp-karp.ubnumbered} ------------------------------------- Directed HCP: Does a gyven directed graph contain a simple cycle that vhsits dvexn xwrhex?\ Input: Graph $G$ containing $N$ vertices and $e$ (qjrtctvd) arcs.\ Input size: $e$. Undirectec JCL: Does a given undirzdtsd graph contain a simple cyclw that vifits every vertex? Conversion: Defune subgraphf $S_i$ for $i = 1, \hdots, U$ to be 3-vertzx subbraphx each containing edges $(1,2)$ ans $(2,3)$. We then fonstruct x new instance ow Ukdisected HCP by replacing eash vertex $i$ iu $G$ with $S_i$. Jhen fow each dirfcted arc $(i,j)$, the new inshance cmntains an edge going from vertex 3 of $S_i$ vp vertex 1 of $S_b$. Exilanation: The xecond vertex in each $S_i$ is a degrze 2 veftex, and sk it is clear that any time verdgx $1$ of any $S_i$ is reacred, certuces 2 ava then 3 of the same $S_i$ must immeduately follow. Then it js only possiblz uo exit each $S_i$ fia an efgx incygent on the dhira vdttex, dhich correwoondx to an arc that depdrts vertex $i$ in $G$. Likrwlse, each jime a $S_i$ is exited,

, leaving a directed acyclic graph. Hence restrict consideration to cycles in $G'$. $G'$ those that don’t a corresponding cycle $G$) are those that are created being allowed to visit one or more path graphs multiple times. These effectively to a union of cycles in $G$. Hence if there is no feedback set size that all the cycles in $G$, there is definitely none of size $k$ in $G'$ either. Final Size: As argued above, the line graph of will be sparse (ie in-degree and out-degree bounded by constant) will have vertex of a linear function $e$. It is clear that the input size of the converted problem is a linear function of original input HCP to HCP {#directed-hcp-to-undirected-hcp-karp.unnumbered} Directed HCP: Does directed graph contain a simple cycle vertex?\ Input: Graph $G$ containing $N$ vertices and (directed) arcs.\ size: $e$. Undirected HCP: Does a undirected graph contain a simple cycle that visits vertex? Conversion: Define subgraphs $S_i$ for $i = 1, \hdots, N$ to be 3-vertex subgraphs edges $(1,2)$ and $(2,3)$. then construct a instance Undirected by each vertex in $G$ with $S_i$. Then for each directed arc $(i,j)$, the instance contains an edge going from vertex 3 of $S_i$ 1 $S_j$. Explanation: The vertex in each $S_i$ a 2 vertex, and so clear any of $S_i$ reached, vertices 2 and 3 of the same $S_i$ immediately follow. Then it each $S_i$ via an edge incident on the vertex, which corresponds to an arc that vertex $i$ in $G$. Likewise, each time a $S_i$ is exited,

, leaving a directed acyclic grAph. Hence we Can reStrIct OuR conSideRation to the cycLEs in $g'$. The only new" cycles in $G'$ (ie Those ThAT don’T HaVe a coRresponDInG CYclE iN $G$) Are ThOSe That aRe cReated bY being alloWed To Visit one or moRE pAth graphs mUltIple times. TheSe eFfectiVeLy cORrespOnd To a unIon of cYCles in $g$. Hence if tHeRE is no fEEdback aRC SeT of sIze $k$ that removes alL ThE Cycles in $G$, there Is defiNiTElY NOne Of sIze $k$ in $G'$ eitHeR. FinaL input SiZE: AS ARGueD Above, the line gRaph of $G'$ will BE spArse (ie HaVe iN-Degree And ouT-dEGreE bounded by a ConsTant) and wiLl have A Vertex sET of cardInalitY a lIneAr fuNCtIoN of $E$. IT Is cLEaR thAT thE input siZe Of The coNverTED PRoblEm iS a liNear fUnction of the oRigInal INpuT size. direcTed HcP To UndIrecteD HCP (KArP) {#directed-hcp-to-uNdirEcted-hcp-kArp.UnNumBeRed} ------------------------------------- DiREcted HcP: DOes A given dIrected GRapH cONTAiN a simple cycle that vIsITS eVery vertEx?\ InpuT: grApH $g$ containInG $N$ vErtiCES and $e$ (DireCTeD) arcs.\ InpUt size: $E$. unDiRected HcP: does a gIvEn uNdiRecteD GrapH contaIn a simplE cyclE That visits everY Vertex? ConversIOn: dEFiNE subGraPhs $S_i$ for $i = 1, \hdOts, N$ TO be 3-vErteX SuBgrAPhs eaCh conTaINiNG edges $(1,2)$ and $(2,3)$. We then consTrUct a neW instAnce of UndirecTed HCP by rePLACing each VertEX $i$ IN $G$ with $S_i$. Then foR each Directed arC $(I,j)$, the new InstaNce contaIns an edge GOIng from vErtEx 3 oF $S_i$ To vERTeX 1 of $S_j$. ExplanatION: The SeCond verTex In each $S_I$ is A deGreE 2 veRtEx, and so it Is clear tHaT aNy TiMe vErtex $1$ OF any $S_i$ is ReAchEd, VerTices 2 ANd then 3 Of the Same $s_i$ MuST imMediateLY fOLLow. THeN iT is oNly PoSsiblE to eXIt eAch $S_i$ viA an edge inCidENt on ThE tHird verTex, which correSpOnds to an arC tHat DepartS VErtex $i$ in $g$. Likewise, each time a $S_i$ is eXIted,

, leaving a directed acycl ic graph.Hence we ca nrest rict our considera t ionto the cycles in $G'$. Theon l y ne w "cycle s in $G ' $( i e t ho se th at do n’t h ave a corr espondingcyc le in $G$) are th ose that a recreated by b ein g allo we d t o visi t o ne or morep ath gr aphs mult ip l e time s . These e ff ecti vely correspond t o a union of cycle s in $ G$ . H e n ceifthere is n ofeedb a ck arcs et o f si z e $k$ that re moves all t h e c yclesin $G $ , ther e isde f ini tely none o f si ze $k$ in $G'$e ither.Final I nput S ize : A s ar g ue dabo ve , th e l ine gra ph of $G '$ w ill b e sp a r s e (ie ha ve i n-deg ree and out-d egr ee b o und ed by a co nsta nt ) and willhaveavertex set of c ardi nality alin ea r f un ction of $e$ . I t i s clear that t h e i np u t si ze of the converte dp r ob lem is a linea r f un c tion ofth e o rigi n a l inp ut s i ze . Direc ted HC P t oUndirec te d HCP(K arp ) { #dire c ted- hcp-to -undirec ted-h c p-karp.unnumbe r ed} --------- - -- - - -- - ---- --- ----------- -- D irec tedH CP : D o es agiven d i re c ted graph contain a s implecycle that visitsevery vert e x ? \ Input: Gra p h$ G$ containing$N$ v ertices an d $e$ (di recte d) arcs. \ Input s i z e: $e$. Un dir ect edH C P: Does a given u ndir ec ted gra phcontain asim ple cy cl e that vi sits eve ry v er te x? Conv e rsion: D ef ine s ubg raphs $S_i$for $ i =1, \ h dot s, N$ t o b e 3-ve rt ex sub gra ph s eac h co n tai ning ed ges $(1,2 )$a nd $ (2 ,3 )$. Wethen construc ta new inst an ceof Und i r ected HC P by replacing each ver t ex $i$in$G$ w ith$S_i$. Th enfor ea chd irecte d arc$(i,j )$ , t h e newi n st anc econtains a n edg e goi ng fro m verte x 3 of $S_i$ to ve r tex 1 of $S_j$. Ex plan a t io n:T he sec on d ve r t ex in each $S_i $ is a deg re e 2 vertex, a n d s oit is c lear th at an y time v ertex $1$ of any $ S_ i$ i s rea ched, vert ices 2 a nd then 3 of th e s ame $ S_i $ must i mme diate ly fol l ow. Then it is o nly po ssibl eto exiteach $S_i$ via an edgeincide nt on th e third v ert e x,which cor resp onds to an ar c t hat d epa r ts ve rtex $i $ i n $G$. Lik e wise, eac h t ime a $ S_i$ is exi t e d ,

, leaving_a directed_acyclic graph. Hence we_can restrict_our_consideration to_the_cycles in $G'$._The only new"_cycles in $G'$ (ie_those that don’t_have_a corresponding cycle in $G$) are those that are created by being allowed to_visit_one or_more_path_graphs multiple times. These effectively_correspond to a union of_cycles in_$G$. Hence if there is no feedback arc_set_of size $k$_that removes all the cycles in $G$, there is_definitely none of size $k$ in_$G'$ either. Final Input_Size:_As_argued above, the line_graph of $G'$ will be sparse_(ie have in-degree and out-degree bounded_by a constant) and will have a_vertex set of cardinality a linear_function of $e$. It is_clear that_the input size of the_converted problem is_a linear_function of the_original input size. Directed HCP to Undirected_HCP (Karp) {#directed-hcp-to-undirected-hcp-karp.unnumbered} ------------------------------------- Directed_HCP: Does a given directed graph_contain_a simple cycle_that_visits_every vertex?\ Input:_Graph $G$ containing_$N$_vertices and_$e$_(directed) arcs.\ Input size: $e$. Undirected HCP: Does_a_given undirected graph contain a simple cycle_that visits every vertex? Conversion:_Define_subgraphs $S_i$ for $i_= 1, \hdots, N$ to_be 3-vertex subgraphs each containing edges_$(1,2)$ and_$(2,3)$. We_then construct a new instance of Undirected HCP by replacing each_vertex $i$ in $G$ with $S_i$._Then for each directed_arc $(i,j)$,_the_new instance contains_an_edge going_from vertex 3 of $S_i$ to vertex_1 of_$S_j$. Explanation: The second vertex in each_$S_i$ is a degree_2_vertex, and so it is clear_that any time vertex $1$ of_any $S_i$ is reached, vertices_2_and_then 3 of the same_$S_i$ must immediately follow. Then it_is only possible_to exit each $S_i$ via an edge_incident_on the third vertex, which corresponds_to_an arc that departs vertex $i$_in_$G$._Likewise, each time a $S_i$_is exited,

eigenfunction-square} \phi_{\alpha}(x, y) = \cos(\alpha) \sin (x ) \sin(2y) + \sin(\alpha) \sin(2x) \sin(y).$$ We also introduce the eigenfunction $$\label{psi-eigenfunction-square} \psi_{\alpha}(x, y) = \sin(\alpha) \sin (x ) \sin(2y) - \cos(\alpha) \sin(2x) \sin(y),$$ orthogonal to $\phi_{\alpha}$. Next we recall the following qualitative result from [@delPinoGarciaMelianMusso], see [@delPinoGarciaMelianMusso Theorem 1.1] and some remarks at [@delPinoGarciaMelianMusso p. 3501], based on standard bifurcation analysis. \[th:delpino\] There exists ${\varepsilon}>0$ and a neighborhood $\mathcal{U}$ of $(\lambda_2,0)$ in $\mathbb{R} \times C(\overline{\Omega})$ such that the set of all solutions of in $\mathcal{U}$ can be described as the union of four $C^1$–curves in $\mathbb{R} \times C(\overline{\Omega})$, $$s \in (-{\varepsilon}, {\varepsilon}) \mapsto (\lambda_i(s), u_i(s)), \quad i=1,\ldots, 4,$$ such that $$\label{square:asymptotics} \begin{cases} (\lambda_i(-s), u_i(-s)) = (\lambda_i(s), -u_i(s)),\\ \lambda_i(s) = \lambda_2+ \sigma_i s^2 + o(s^2),\\ u_i(s) = s \phi_{\alpha_i} + o(s), \end{cases}$$ where $\alpha_1=0$, $\alpha_2= \pi/4$, $\alpha_3=\pi/2$, $\alpha_4=3\pi/4$ and $$\label{sigmas} \sigma_i = \frac{9}{16} \ \ \hbox{for} \ \ i=1,3, \qquad \sigma_i = \frac{21}{32} \ \ \hbox{for} \ \ i = 2, 4.$$ Observe that

eigenfunction - square } \phi_{\alpha}(x, y) = \cos(\alpha) \sin (x) \sin(2y) + \sin(\alpha) \sin(2x) \sin(y).$$ We also introduce the eigenfunction $ $ \label{psi - eigenfunction - square } \psi_{\alpha}(x, y) = \sin(\alpha) \sin (x) \sin(2y) - \cos(\alpha) \sin(2x) \sin(y),$$ extraneous to $ \phi_{\alpha}$. Next we echo the following qualitative result from [ @delPinoGarciaMelianMusso ], visualize [ @delPinoGarciaMelianMusso Theorem 1.1 ] and some remark at [ @delPinoGarciaMelianMusso p. 3501 ], based on standard bifurcation analysis. \[th: delpino\ ] There exist $ { \varepsilon}>0 $ and a region $ \mathcal{U}$ of $ (\lambda_2,0)$ in $ \mathbb{R } \times C(\overline{\Omega})$ such that the set of all solution of in $ \mathcal{U}$ can be described as the union of four $ C^1$–curves in $ \mathbb{R } \times C(\overline{\Omega})$, $ $ s \in (-{\varepsilon }, { \varepsilon }) \mapsto (\lambda_i(s), u_i(s) ), \quad i=1,\ldots, 4,$$ such that $ $ \label{square: asymptotics } \begin{cases } (\lambda_i(-s), u_i(-s) ) = (\lambda_i(s), -u_i(s)),\\ \lambda_i(s) = \lambda_2 + \sigma_i s^2 + o(s^2),\\ u_i(s) = s \phi_{\alpha_i } + o(s), \end{cases}$$ where $ \alpha_1=0 $, $ \alpha_2= \pi/4 $, $ \alpha_3=\pi/2 $, $ \alpha_4=3\pi/4 $ and $ $ \label{sigmas } \sigma_i = \frac{9}{16 } \ \ \hbox{for } \ \ i=1,3, \qquad \sigma_i = \frac{21}{32 } \ \ \hbox{for } \ \ i = 2, 4.$$ Observe that

eigfnfunction-square} \phi_{\alpha}(w, y) = \cos(\alpha) \siu (x ) \sii(2y) + \sin(\zlpha) \siv(2x) \sin(y).$$ We also introduce thx eitenfubction $$\label{psi-eigenfuvction-squwre} \psi_{\alpha}(x, t) = \sin(\alphe) \sin (x ) \sin(2y) - \gos(\al'he) \sin(2x) \sin(y),$$ ortmogonal to $\[hi_{\alpha}$. Next fe rzcall the following qualitative resujt from [@dflPinoGarciaMejianKtsso], see [@delPinoGarciaMelianMusso Theodem 1.1] anv some remarks st [@delPinoGarciaMelianMussl p. 3501], based on standard bifurcatiob anwoysis. \[th:delpivo\] There eqnsts ${\varepsjlon}>0$ and a neighborhood $\mathcal{J}$ of $(\kambda_2,0)$ in $\narhbh{T} \times C(\ovecline{\Oiega})$ such that the sat of akl solutions on in $\katycal{U}$ can be describev as the union of foor $C^1$–curves iu $\mathbb{R} \times C(\overoibe{\Omeca})$, $$s \in (-{\xqreosimoi}, {\vzrepsipon}) \mapsto (\lajbda_i(s), u_i(s)), \wuad i=1,\ldots, 4,$$ such tnae $$\label{square:aaymptoeiss} \begin{cases} (\lambda_i(-s), u_i(-s)) = (\lambda_i(s), -u_i(s)),\\ \ldmbsa_i(s) = \lambda_2+ \sigma_i s^2 + i(s^2),\\ u_i(s) = s \phi_{\alpha_i} + o(d), \end{cases}$$ where $\alpha_1=0$, $\alpha_2= \pi/4$, $\alpha_3=\pi/2$, $\alpha_4=3\pi/4$ and $$\label{sicmas} \smgoa_i = \nrac{9}{16} \ \ \jbox{for} \ \ i=1,3, \qquad \sigma_i = \frac{21}{32} \ \ \hbox{for} \ \ i = 2, 4.$$ Onsvrve that

eigenfunction-square} \phi_{\alpha}(x, y) = \cos(\alpha) \sin (x + \sin(2x) \sin(y).$$ also introduce the \sin(\alpha) (x ) \sin(2y) \cos(\alpha) \sin(2x) \sin(y),$$ to $\phi_{\alpha}$. Next we recall the qualitative result from [@delPinoGarciaMelianMusso], see [@delPinoGarciaMelianMusso Theorem 1.1] and some remarks at [@delPinoGarciaMelianMusso 3501], based on standard bifurcation analysis. \[th:delpino\] There exists ${\varepsilon}>0$ and a neighborhood of in \times such that the set of all solutions of in $\mathcal{U}$ can be described as the union four $C^1$–curves in $\mathbb{R} \times C(\overline{\Omega})$, $$s \in {\varepsilon}) \mapsto (\lambda_i(s), u_i(s)), i=1,\ldots, 4,$$ such that $$\label{square:asymptotics} (\lambda_i(-s), = (\lambda_i(s), \lambda_i(s) \lambda_2+ s^2 + o(s^2),\\ = s \phi_{\alpha_i} + o(s), \end{cases}$$ where $\alpha_1=0$, $\alpha_2= \pi/4$, $\alpha_3=\pi/2$, $\alpha_4=3\pi/4$ and $$\label{sigmas} \sigma_i = \frac{9}{16} \ \hbox{for} i=1,3, \qquad = \ \hbox{for} \ \ 2, 4.$$ Observe that

eigenfunction-square} \phi_{\alpHa}(x, y) = \cos(\alpHa) \sin (X ) \siN(2y) + \sIn(\AlphA) \sin(2X) \sin(y).$$ We also intROducE the eigenfunction $$\label{Psi-eiGeNFuncTIoN-squaRe} \psi_{\alPHa}(X, Y) = \Sin(\AlPhA) \siN (x ) \SIn(2Y) - \cos(\aLphA) \sin(2x) \siN(y),$$ orthogonAl tO $\pHi_{\alpha}$. Next wE ReCall the folLowIng qualitatiVe rEsult fRoM [@deLpinoGArcIaMelIanMusSO], see [@deLPinoGarcIamElianMUSso TheoREM 1.1] aNd soMe remarks at [@delPinOgaRCiaMelianMusso P. 3501], based On STaNDArd BifUrcation anAlYsis. \[tH:Delpino\] tHeRE EXisTS ${\varepsilon}>0$ anD a neighborhOOd $\mAthcal{u}$ oF $(\laMBda_2,0)$ in $\mAthbb{r} \tIMes c(\overline{\OmEga})$ sUch that thE set of ALl solutIOns of in $\MathcaL{U}$ cAn bE desCRiBeD as ThE UniON oF foUR $C^1$–cUrves in $\mAtHbB{R} \timEs C(\oVERLIne{\OMegA})$, $$s \in (-{\VarepSilon}, {\varepsilOn}) \mApstO (\LamBda_i(s), U_i(s)), \quAd i=1,\lDoTs, 4,$$ sucH that $$\lAbel{sQuAre:asymptotics} \bEgin{Cases} (\lambDa_i(-S), u_I(-s)) = (\lAmBda_i(s), -U_I(s)),\\ \lambDa_i(S) = \laMbda_2+ \sigMa_i s^2 + o(s^2),\\ u_I(S) = s \pHi_{\ALPHa_I} + o(s), \end{cases}$$ where $\alPhA_1=0$, $\ALpHa_2= \pi/4$, $\alphA_3=\pi/2$, $\alpHA_4=3\pI/4$ aND $$\label{siGmAs} \sIgma_I = \FRac{9}{16} \ \ \hbOx{foR} \ \ I=1,3, \qQuad \sigmA_i = \frac{21}{32} \ \ \HBoX{fOr} \ \ i = 2, 4.$$ ObseRvE that

eigenfunction-square} \phi _{\alpha}( x, y) =\co s( \alp ha)\sin (x ) \sin ( 2y)+ \sin(\alpha) \sin(2x ) \si n( y ).$$ We also introd u ce t heei ge nfu nc t io n $$\ lab el{psi- eigenfunct ion -s quare} \psi_ { \a lpha}(x, y ) = \sin(\alpha ) \ sin (x ) \s i n(2y) -\cos( \alpha ) \sin( 2x) \sin( y) , $$ ort h ogonalt o $ \phi _{\alpha}$. Nextw er ecall the foll owingqu a li t a tiv e r esult from [ @delP i noGarci a Me l i a nMu s so], see [@de lPinoGarcia M eli anMuss oThe o rem 1. 1] an ds ome remarks at [@d elPinoGar ciaMel i anMusso p. 3501 ], bas edonstan d ar dbif ur c ati o nana l ysi s. \[th :d el pino\ ] Th e r e exis ts${\v areps ilon}>0$ anda n eigh b orh ood $ \math cal{ U} $ of$(\lam bda_2 ,0 )$ in $\mathbb{ R} \ times C(\ ove rl ine {\ Omega } )$ suc h t hat the se t of al l so lu t i o ns of in $\mathcal{U }$ c an be desc ribeda sth e union o ffou r $C ^ 1 $–cur vesi n$\mathbb {R} \t i me sC(\over li ne{\Om eg a}) $,$$s \ i n (- {\vare psilon}, {\va r epsilon}) \map s to (\lambda_i ( s) , u_ i (s)) , \ quad i=1,\l dots , 4,$ $ su c htha t $$\l abel{ sq u ar e :asymptotics} \begi n{ cases} (\la mbda_i(-s), u _i(-s)) =( \ l ambda_i( s),- u_ i (s)),\\ \lambd a_i(s ) = \lambd a _2+ \sig ma_is^2 + o( s^2),\\ u _ i (s) = s\ph i_{ \al pha _ i }+ o(s), \end{ c a ses} $$ where$\a lpha_1= 0$, $\ alp ha_ 2= \pi/4$,$\alpha_ 3= \p i/ 2$ , $ \alph a _4=3\pi/ 4$ an d$$\ label { sigmas } \si gma_ i=\ fra c{9}{16 } \ \ \hb ox {f or}\ \ i =1,3, \qq u ad\sigma_ i = \frac {21 } {32} \ \ \hbox{ for} \ \ i =2, 4.$$ Obs er vethat

eigenfunction-square} \phi_{\alpha}(x, y)_= \cos(\alpha)_\sin (x ) \sin(2y)_+ \sin(\alpha)_\sin(2x)_\sin(y).$$ We_also_introduce the eigenfunction_$$\label{psi-eigenfunction-square} \psi_{\alpha}(x, y) =_\sin(\alpha) \sin (x )_\sin(2y) - \cos(\alpha)_\sin(2x)_\sin(y),$$ orthogonal to $\phi_{\alpha}$. Next we recall the following qualitative result from [@delPinoGarciaMelianMusso], see_[@delPinoGarciaMelianMusso_Theorem 1.1]_and_some_remarks at [@delPinoGarciaMelianMusso p. 3501],_based on standard bifurcation analysis. \[th:delpino\]_There exists_${\varepsilon}>0$ and a neighborhood $\mathcal{U}$ of $(\lambda_2,0)$ in_$\mathbb{R}_\times C(\overline{\Omega})$ such_that the set of all solutions of in $\mathcal{U}$_can be described as the union_of four $C^1$–curves_in_$\mathbb{R}_\times C(\overline{\Omega})$, $$s \in_(-{\varepsilon}, {\varepsilon}) \mapsto (\lambda_i(s), u_i(s)), \quad_i=1,\ldots, 4,$$ such that $$\label{square:asymptotics} \begin{cases} (\lambda_i(-s), u_i(-s))_= (\lambda_i(s), -u_i(s)),\\ \lambda_i(s) = \lambda_2+ \sigma_i s^2_+ o(s^2),\\ u_i(s) = s \phi_{\alpha_i} +_o(s), \end{cases}$$ where $\alpha_1=0$, $\alpha_2= \pi/4$,_$\alpha_3=\pi/2$, $\alpha_4=3\pi/4$_and $$\label{sigmas} \sigma_i = \frac{9}{16} \_\ \hbox{for} \_\ i=1,3,_\qquad \sigma_i =_\frac{21}{32} \ \ \hbox{for} \ \_i = 2,_4.$$ Observe that

. The non-negativity of $u$ follows from the substochasticity of the semigroups. For uniqueness, we observe first that the construction of the Sobolev tower ensures that $X_0^{(w)}$ is continuously embedded in $X_{-n}^{(w)}$. Moreover, $\mathcal{G}^{(w)}$ is the restriction of $\mathcal{G}^{(w)}_{-n}$ to $X^{(w)}_1=\mathcal{D}(\mathcal{G}^{(w)})$. Consequently, if $u_1, u_2 \in C^1((0,\infty), \ell_w^1) \cap C([0,\infty),X^{(w)}_{-n})$ both satisfy, and we set $v_i(t) = e^{-\mu t}u_i(t)$, $i =1,2$, then the difference $v_1-v_2$ is the unique classical solution of with $\mathring{u} = 0$, and so $v_1= v_2$, from which it follows that $u_1=u_2$. Finally, we make the following remark on the solvability of. \[remark about general IC in X with Sobolev tower\] For fixed $n \in \mathbb{N}$, the previous theorem establishes that the ACP has a unique, non-negative solution $u \in C^1((0,\infty), \ell_w^1) \cap C([0,\infty),X^{(w)}_{-n})$, given by $u(t)=S^{(w)}_{-n}(t)\mathring{u}$, for all $\mathring{u} \in (X^{(w)}_{-n})_+$, provided that Assumptions \[A1.1\] and \[assumption for analyticity\] are satisfied. Recalling that we also assume that $w_n \ge n$ for all $n \in \mathbb{N}$, we have that $\ell_w^1$ is continuously embedded in $X_{[1]}$, and from this we deduce that if $u(t)$ is differentiable with respect to the norm on $\ell_w^1$ then it is also differentiable with respect to the norm on $X_{[1]}$, and the derivatives coincide. Since $A_1+B_1$ is an extension of $G^{(w)}=A^{(w)}+B^{(w)}$, we conclude that $u(t

. The non - negativity of $ u$ follows from the substochasticity of the semigroups. For uniqueness, we observe first that the structure of the Sobolev column ensures that $ X_0^{(w)}$ is continuously implant in $ X_{-n}^{(w)}$. Moreover, $ \mathcal{G}^{(w)}$ is the limitation of $ \mathcal{G}^{(w)}_{-n}$ to $ X^{(w)}_1=\mathcal{D}(\mathcal{G}^{(w)})$. Consequently, if $ u_1, u_2 \in C^1((0,\infty), \ell_w^1) \cap C([0,\infty),X^{(w)}_{-n})$ both satisfy, and we set $ v_i(t) = e^{-\mu t}u_i(t)$, $ i = 1,2 $, then the difference $ v_1 - v_2 $ is the unique authoritative solution of with $ \mathring{u } = 0 $, and indeed $ v_1= v_2 $, from which it follows that $ u_1 = u_2$. ultimately, we make the following remark on the solvability of. \[remark about general intelligence community in X with Sobolev tower\ ] For fixed $ n \in \mathbb{N}$, the previous theorem establishes that the ACP experience a unique, non - negative solution $ u \in C^1((0,\infty), \ell_w^1) \cap C([0,\infty),X^{(w)}_{-n})$, given by $ u(t)=S^{(w)}_{-n}(t)\mathring{u}$, for all $ \mathring{u } \in (X^{(w)}_{-n})_+$, supply that Assumptions   \[A1.1\ ] and \[assumption for analyticity\ ] are satisfied. Recalling that we besides assume that $ w_n \ge n$ for all $ n \in \mathbb{N}$, we receive that $ \ell_w^1 $ is endlessly embedded in $ X_{[1]}$, and from this we deduce that if $ u(t)$ is differentiable with respect to the norm on $ \ell_w^1 $ then it is also differentiable with respect to the norm on $ X_{[1]}$, and the derivative instrument coincide. Since $ A_1+B_1 $ is an extension of $ G^{(w)}=A^{(w)}+B^{(w)}$, we conclude that $ u(t

. Thf non-negativity of $u$ foluows from the sobwtochavticitg of the semigroups. For uniqueness, we ovservt first that the covstructioj of the Sobioev tower xhsures bkat $X_0^{(s)}$ is eoitinuously embecded in $X_{-n}^{(f)}$. Moreover, $\matvcxl{Y}^{(w)}$ is the restriction of $\mathcal{G}^{(w)}_{-n}$ eo $X^{(w)}_1=\mayhfal{D}(\mathcal{G}^{(w)})$. Sonstquqntlg, if $u_1, u_2 \in C^1((0,\infty), \ell_w^1) \cap C([0,\infty),S^{(w)}_{-n})$ boti satisfy, and wr set $v_i(t) = e^{-\mu t}u_i(t)$, $i =1,2$, thej thf difference $v_1-v_2$ is the unique clafwical solutiun of with $\mathring{u} = 0$, and so $v_1= v_2$, from which it follods thct $u_1=u_2$. Finalli, se kake the foolowigg remark on the soleabilitu of. \[remark abomt geierao IC in X with Sobolet tower\] For fixed $n \yn \mathbb{T}$, che previous theorem wsrablivhes thag thd ADP hzs a ujiqne, non-negatjve solutiob $u \in C^1((0,\infty), \ell_w^1) \va[ C([0,\infty),X^{(w)}_{-n})$, giben by $u(e)=S^{(w)}_{-n}(t)\mathring{u}$, for all $\mathring{u} \in (X^{(w)}_{-n})_+$, prkvided that Assumptions \[Q1.1\] and \[assumption for wnalyticiey\] are satisfied. Recalling that we also assume thdt $w_n \ee u$ for xol $n \in \mathbb{N}$, we have that $\ell_w^1$ is continuousjg tmbvdded in $X_{[1]}$, and frjm this we cefuvg that if $u(t)$ ir diffzdehtiable with respeft to tre noem on $\ell_r^1$ thrn it is also differentiablw with respebt ti the norm on $X_{[1]}$, anb the derivacives voincode. Since $A_1+B_1$ is an exteusion kf $G^{(w)}=A^{(w)}+B^{(w)}$, wf concluds that $u(t

. The non-negativity of $u$ follows from of semigroups. For we observe first Sobolev ensures that $X_0^{(w)}$ continuously embedded in Moreover, $\mathcal{G}^{(w)}$ is the restriction of to $X^{(w)}_1=\mathcal{D}(\mathcal{G}^{(w)})$. Consequently, if $u_1, u_2 \in C^1((0,\infty), \ell_w^1) \cap C([0,\infty),X^{(w)}_{-n})$ both satisfy, we set $v_i(t) = e^{-\mu t}u_i(t)$, $i =1,2$, then the difference $v_1-v_2$ is unique solution with = 0$, and so $v_1= v_2$, from which it follows that $u_1=u_2$. Finally, we make the remark on the solvability of. \[remark about general in X with Sobolev For fixed $n \in \mathbb{N}$, previous establishes that ACP a non-negative solution $u C^1((0,\infty), \ell_w^1) \cap C([0,\infty),X^{(w)}_{-n})$, given by $u(t)=S^{(w)}_{-n}(t)\mathring{u}$, for all $\mathring{u} \in (X^{(w)}_{-n})_+$, provided that Assumptions \[A1.1\] and for analyticity\] Recalling that also that \ge n$ for \in \mathbb{N}$, we have that $\ell_w^1$ in $X_{[1]}$, and from this we deduce that $u(t)$ is with respect to the norm on then it is also differentiable with respect to norm on $X_{[1]}$, and the derivatives coincide. Since $A_1+B_1$ is an extension of $G^{(w)}=A^{(w)}+B^{(w)}$, we $u(t

. The non-negativity of $u$ followS from the suBstocHasTicItY of tHe seMigroups. For uniQUeneSs, we observe first that thE consTrUCtioN Of The SoBolev toWEr ENSurEs ThAt $X_0^{(W)}$ iS CoNtinuOusLy embedDed in $X_{-n}^{(w)}$. MoReoVeR, $\mathcal{G}^{(w)}$ is THe RestrictioN of $\Mathcal{G}^{(w)}_{-n}$ to $x^{(w)}_1=\mAthcal{d}(\mAthCAl{G}^{(w)})$. COnsEquenTly, if $u_1, U_2 \In C^1((0,\infTy), \ell_w^1) \cap c([0,\iNFty),X^{(w)}_{-n})$ BOth satiSFY, aNd we Set $v_i(t) = e^{-\mu t}u_i(t)$, $i =1,2$, theN ThE Difference $v_1-v_2$ is The uniQuE ClASSicAl sOlution of wItH $\mathRIng{u} = 0$, and SO $v_1= V_2$, FROm wHIch it follows tHat $u_1=u_2$. FinallY, We mAke the FoLloWIng remArk on ThE SolVability of. \[rEmarK about genEral IC IN X with SOBolev toWer\] For FixEd $n \In \maTHbB{N}$, The PrEVioUS tHeoREm eStablishEs ThAt the aCP hAS A UNiquE, noN-negAtive Solution $u \in C^1((0,\iNftY), \ell_W^1) \Cap c([0,\inftY),X^{(w)}_{-n})$, gIven By $U(t)=S^{(w)}_{-n}(T)\mathrIng{u}$, fOr All $\mathring{u} \in (X^{(W)}_{-n})_+$, prOvided thaT AsSuMptIoNs \[A1.1\] anD \[AssumpTioN foR analytIcity\] arE SatIsFIED. REcalling that we also AsSUMe That $w_n \ge N$ for alL $N \iN \mAThbb{N}$, we hAvE thAt $\elL_W^1$ Is conTinuOUsLy embeddEd in $X_{[1]}$, aND fRoM this we DeDuce thAt If $u(T)$ is DiffeREntiAble wiTh respecT to thE Norm on $\ell_w^1$ then IT is also differENtIABlE With ResPect to the noRm on $x_{[1]}$, And tHe deRIvAtiVEs coiNcide. siNCe $a_1+b_1$ is an extension of $G^{(w)}=A^{(W)}+B^{(W)}$, we conClude That $u(t

. The non-negativity of $u $ followsfromthe su bs toch asti city of the se m igro ups. For uniqueness,we ob se r ve f i rs t tha t the c o ns t r uct io nofth e S obole v t ower en sures that $X _0 ^{(w)}$ is c o nt inuously e mbe dded in $X_{ -n} ^{(w)} $. Mo r eover , $ \math cal{G} ^ {(w)}$ is the r es t rictio n of $\m a t hc al{G }^{(w)}_{-n}$ to$ X^ { (w)}_1=\mathca l{D}(\ ma t hc a l {G} ^{( w)})$. Con se quent l y, if $ u _1 , u _2\ in C^1((0,\in fty), \ell_ w ^1) \capC( [0, \ infty) ,X^{( w) } _{- n})$ both s atis fy, and w e set$ v_i(t)= e^{-\m u t}u_ i(t )$, $i= 1, 2$ , t he n th e d iff e ren ce $v_1- v_ 2$ is t he u n i q u e cl ass ical solu tion of with$\m athr i ng{ u} =0$, a nd s o$v_1= v_2$, from w hich it follows tha t $u_1=u_ 2$. Fin al ly, w e makethe fo llowing remark onth e s ol vability of. \[re ma r k a bout gen eral I C i nX with So bo lev tow e r \] Fo r fi x ed $n \in\mathb b {N }$ , the p re viousth eor emestab l ishe s that the ACP hasa unique, non-n e gative soluti o n$ u \ i n C^ 1(( 0,\infty),\ell _ w^1) \ca p C ([0 , \inft y),X^ {( w )} _ {-n})$, given by $u (t )=S^{( w)}_{ -n}(t)\mathri ng{u}$, fo r a ll $\mat hrin g {u } \in (X^{(w)}_ {-n}) _+$, provi d ed thatAssum ptions \ [A1.1\] a n d \[assum pti onfor an a l yt icity\] are s a t isfi ed . Recal lin g thatweals o a ssu me that $w_ n \ge n$ f or a ll $n \in\ mathbb{N }$ , w ehav e tha t $\ell _w^1$ isco nt i nuo usly em b ed d e d in $ X_ {[1] }$, a nd fr om t h iswe dedu ce that i f $ u (t)$ i sdiffere ntiable withre spect to t he no rm on$ \ ell_w^1$ then it is also differ e ntiable wi th re spec t to thenor m on $ X_{ [ 1]}$,and th e der iv ati v e s coi n c id e.Si nce $A_1+B _ 1 $ i s anex tens ion of$G^{(w)}=A^{(w)}+B ^ {(w )}$, we concl ude tha t $u (t

. The_non-negativity of_$u$ follows from the_substochasticity of_the_semigroups. For uniqueness,_we_observe first that_the construction of_the Sobolev tower ensures_that $X_0^{(w)}$ is_continuously_embedded in $X_{-n}^{(w)}$. Moreover, $\mathcal{G}^{(w)}$ is the restriction of $\mathcal{G}^{(w)}_{-n}$ to $X^{(w)}_1=\mathcal{D}(\mathcal{G}^{(w)})$. Consequently, if_$u_1,_u_2 \in_C^1((0,\infty),_\ell_w^1)_\cap C([0,\infty),X^{(w)}_{-n})$ both satisfy, and_we set $v_i(t) = e^{-\mu_t}u_i(t)$, $i_=1,2$, then the difference $v_1-v_2$ is the unique_classical_solution of with_$\mathring{u} = 0$, and so $v_1= v_2$, from which_it follows that $u_1=u_2$. Finally, we make_the following remark_on_the_solvability of. \[remark about general_IC in X with Sobolev tower\]_For fixed $n \in \mathbb{N}$, the_previous theorem establishes that the ACP has_a unique, non-negative solution $u \in_C^1((0,\infty), \ell_w^1) \cap C([0,\infty),X^{(w)}_{-n})$, given_by $u(t)=S^{(w)}_{-n}(t)\mathring{u}$,_for all $\mathring{u} \in (X^{(w)}_{-n})_+$,_provided that Assumptions \[A1.1\]_and \[assumption_for analyticity\] are_satisfied. Recalling that we also assume_that $w_n \ge_n$ for all $n \in \mathbb{N}$,_we_have that $\ell_w^1$_is_continuously_embedded in_$X_{[1]}$, and from_this_we deduce_that_if $u(t)$ is differentiable with respect_to_the norm on $\ell_w^1$ then it is_also differentiable with respect_to_the norm on $X_{[1]}$,_and the derivatives coincide. Since_$A_1+B_1$ is an extension of $G^{(w)}=A^{(w)}+B^{(w)}$,_we conclude_that $u(t

Springer Berlin Heidelberg, vol. 5302, pp. 44-57, 2008. B. C. Russell, A. Torralba, K. P. Murphy, and W. T. Freeman, “LabelMe: A database and web-based tool for image annotation,” *International Journal of Computer Vision*, vol. 77, no. 1, pp. 157-173, 2008. T.-Y. Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Dollar, and C. L. Zitnick, “Microsoft COCO: Common objects in context,” *in Proceedings of the European Conference on Computer Vision*, pp. 740-755, 2014. K. Wang, C. Gou, and F.-Y. Wang, “Parallel vision: An ACP-based approach to intelligent vision computing,” *Acta Automatica Sinica*, vol. 42, no. 10, pp. 1490-1500, 2016. K. Wang, C. Gou, N. Zheng, J. M. Rehg, and F.-Y. Wang, “Parallel vision for perception and understanding of complex scenes: Methods, framework, and perspectives,” *Artificial Intelligence Review*, vol. 48, no. 3, pp. 298-328, 2017. K. Wang, Y. Lu, Y. Wang, Z. Xiong, and F.-Y. Wang, “Parallel imaging: A new theoretical framework for image generation," *Pattern Recognition and Artificial Intelligence*, vol. 30, no. 7, pp. 577-587, 2017. F.-Y. Wang, “Parallel control and management for intelligent transportation systems: Concepts, architectures, and applications,” *IEEE Transactions on Intelligent Transportation Systems*, vol. 11, no. 3, pp. 630-638, 2010. F.-Y. Wang, “Parallel control: A method for data-driven and computational control,” *Acta Automatica Sinica*, vol. 39, no. 4, pp. 293-302, 2014. F.-Y. Wang, X. Wang, L. Li, and L. Li, “Steps toward parallel intelligence," *IEEE/CAA Journal of Automatica Sinica*, vol. 3, no. 4, pp.345-348, 2016. L. Li, Y. Lin, D. Cao, N. Zheng

Springer Berlin Heidelberg, vol. 5302, pp. 44 - 57, 2008. B. C. Russell, A. Torralba, K. P. Murphy, and W. T. Freeman, “ LabelMe: A database and web - based tool for image note, ” * International Journal of Computer Vision *, vol. 77, no. 1, pp. 157 - 173, 2008. T.-Y. Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Dollar, and C. L. Zitnick, “ Microsoft COCO: coarse objects in context, ” * in Proceedings of the European Conference on Computer Vision *, pp. 740 - 755, 2014. K. Wang, C. Gou, and F.-Y. Wang, “ Parallel vision: An ACP - free-base approach path to intelligent vision calculation, ” * Acta Automatica Sinica *, vol. 42, nobelium. 10, pp. 1490 - 1500, 2016. K. Wang, C. Gou, N. Zheng, J. M. Rehg, and F.-Y. Wang, “ Parallel vision for perception and understanding of complex scenes: Methods, model, and perspectives, ” * Artificial Intelligence Review *, vol. 48, no. 3, pp. 298 - 328, 2017. K. Wang, Y. Lu, Y. Wang, Z. Xiong, and F.-Y. Wang, “ Parallel imaging: A new theoretical model for image generation, " * Pattern Recognition and Artificial Intelligence *, vol. 30, nobelium. 7, pp. 577 - 587, 2017. F.-Y. Wang, “ Parallel control and management for healthy exile systems: Concepts, architectures, and application, ” * IEEE Transactions on Intelligent Transportation Systems *, vol. 11, no. 3, pp. 630 - 638, 2010. F.-Y. Wang, “ Parallel control: A method for datum - driven and computational dominance, ” * Acta Automatica Sinica *, vol. 39, no. 4, pp. 293 - 302, 2014. F.-Y. Wang, X. Wang, L. Li, and L. Li, “ step toward parallel intelligence, " * IEEE / CAA Journal of Automatica Sinica *, vol. 3, no. 4, pp.345 - 348, 2016. L. Li, Y. Lin, D. Cao, N. Zheng

Spginger Berlin Heidelberg, vol. 5302, pp. 44-57, 2008. B. C. Tuwsell, E. Torramba, K. P. Ourphy, and W. T. Freeman, “LabelLe: A daucbase and web-based tuol for ilage annitatmon,” *International Journal of Cklputzr Vision*, vol. 77, np. 1, pp. 157-173, 2008. T.-Y. Lin, M. Maire, S. Bdllngie, J. Hays, P. Perona, D. Ramanan, P. Djllar, amd C. L. Zitnick, “Mycroxjft DOCO: Common objects in context,” *in Lroceedpngs of the Europran Conference on Computer Vislon*, pp. 740-755, 2014. K. Wang, C. Hou, and F.-Y. Qang, “Parallel viskon: An ACP-based approadh to intelligent vision computkng,” *Aeta Automatucq Sltica*, vol. 42, ni. 10, pp. 1490-1500, 2016. K. Wang, C. Npu, N. Zveng, J. K. Rehg, and F.-Y. Eanj, “Paeallel vision for perreption and understagding of wojplex scenes: Methidw, frakewosk, ave pdrskecvivss,” *Artlfirial Intelljgence Reviww*, vol. 48, no. 3, pp. 298-328, 2017. K. Eagt, Y. Lu, Y. Wang, Z. Xiogg, and F.-Y. Wang, “Parallel imaging: A new themrefical framework for imate generation," *Pattern Recognityon and Artificial Intelligence*, vol. 30, no. 7, pp. 577-587, 2017. F.-Y. Fang, “'afaloej xojtrol and management for intelligent transporeztook systems: Concepbs, architectures, amd alklications,” *IEEG Transcdtjons on Intelligenh Transkortatuon Systeis*, vpl. 11, no. 3, pp. 630-638, 2010. F.-Y. Wang, “Parallwl control: A nethod for data-driren and compbtatiomal cpntrol,” *Acta Automatica Rinida*, vol. 39, no. 4, pp. 293-302, 2014. F.-Y. Sxng, X. Wang, L. Li, xnd L. Li, “Steps toward parallel yntelligeice," *IZEE/CAA Juurnsl of Wutomatica Sinigd*, vol. 3, no. 4, pp.345-348, 2016. L. Ll, Y. Lnn, D. Wao, N. Zhenh

Springer Berlin Heidelberg, vol. 5302, pp. 44-57, C. A. Torralba, P. Murphy, and database web-based tool for annotation,” *International Journal Computer Vision*, vol. 77, no. 1, 157-173, 2008. T.-Y. Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. P. Dollar, and C. L. Zitnick, “Microsoft COCO: Common objects in context,” *in of European on Vision*, pp. 740-755, 2014. K. Wang, C. Gou, and F.-Y. Wang, “Parallel vision: An ACP-based approach intelligent vision computing,” *Acta Automatica Sinica*, vol. 42, 10, pp. 1490-1500, 2016. Wang, C. Gou, N. Zheng, M. and F.-Y. “Parallel for and understanding of scenes: Methods, framework, and perspectives,” *Artificial Intelligence Review*, vol. 48, no. 3, pp. 298-328, 2017. K. Wang, Lu, Y. Xiong, and Wang, imaging: new theoretical framework generation," *Pattern Recognition and Artificial Intelligence*, 7, pp. 577-587, 2017. F.-Y. Wang, “Parallel control management for transportation systems: Concepts, architectures, and applications,” Transactions on Intelligent Transportation Systems*, vol. 11, no. pp. 630-638, 2010. F.-Y. Wang, “Parallel control: A method for data-driven and computational control,” *Acta vol. 39, no. 4, 293-302, 2014. F.-Y. X. L. and Li, “Steps parallel intelligence," *IEEE/CAA Journal of Automatica Sinica*, vol. 3, no. 4, 2016. L. Li, Y. Lin, D. Cao, N. Zheng

Springer Berlin Heidelberg, vOl. 5302, pp. 44-57, 2008. B. C. RussEll, A. TOrrAlbA, K. p. MurPhy, aNd W. T. Freeman, “LabELMe: A Database and web-based tooL for iMaGE annOTaTion,” *INternatIOnAL jouRnAl Of COmPUtEr VisIon*, Vol. 77, no. 1, pp. 157-173, 2008. t.-Y. Lin, M. MairE, S. BElOngie, J. Hays, P. PERoNa, D. Ramanan, p. DoLlar, and C. L. ZitNicK, “MicroSoFt CocO: ComMon ObjecTs in coNText,” *in proceedinGs OF the EuROpean CoNFErEnce On Computer Vision*, pP. 740-755, 2014. k. WANg, C. Gou, and F.-Y. WanG, “ParalLeL ViSIOn: AN ACp-based apprOaCh to iNTelligeNT vISIOn cOMputing,” *Acta AuTomatica SinICa*, vOl. 42, no. 10, pp. 1490-1500, 2016. k. WAng, c. gou, N. ZhEng, J. M. reHG, anD F.-Y. Wang, “ParaLlel Vision for PercepTIon and uNDerstanDing of ComPleX sceNEs: meThoDs, FRamEWoRk, aND peRspectivEs,” *arTificIal INTELLigeNce reviEw*, vol. 48, No. 3, pp. 298-328, 2017. K. Wang, Y. Lu, Y. wanG, Z. XiONg, aNd F.-Y. WAng, “PaRallEl ImagiNg: A new TheorEtIcal framework foR imaGe generatIon," *paTteRn recogNItion aNd ARtiFicial INtelligENce*, VoL. 30, NO. 7, Pp. 577-587, 2017. f.-Y. Wang, “Parallel contRoL ANd ManagemeNt for iNTeLlIGent tranSpOrtAtioN SYstemS: ConCEpTs, architEctureS, AnD aPplicatIoNs,” *IEEE trAnsActIons oN inteLligenT TranspoRtatiON Systems*, vol. 11, no. 3, pP. 630-638, 2010. f.-Y. Wang, “ParalleL CoNTRoL: a metHod For data-drivEn anD CompUtatIOnAl cONtrol,” *acta AUtOMaTIca Sinica*, vol. 39, no. 4, pp. 293-302, 2014. F.-Y. WAnG, X. Wang, l. Li, anD L. Li, “Steps towaRd parallel INTElligencE," *IEEe/cAa journal of AutomAtica sinica*, vol. 3, nO. 4, Pp.345-348, 2016. L. Li, Y. LiN, D. Cao, n. Zheng

Springer Berlin Heidelber g, vol. 53 02, p p.44- 57 , 20 08.B. C. Russell, A. T orralba, K. P. Murphy, andW. T. F r ee man,“LabelM e :A dat ab as e a nd we b-bas edtool fo r image an not at ion,” *Inter n at ional Jour nal of Computer Vi sion*, v ol. 77, n o.1, pp . 157- 1 73, 20 08. T.-Y. L i n, M.M aire, S . Be long ie, J. Hays, P. P e ro n a, D. Ramanan, P. Do ll a r, a ndC.L. Zitnick ,“Micr o soft CO C O: C o mmo n objects in c ontext,” *i n Pr oceedi ng s o f the E urope an Con ference onComp uter Visi on*, p p . 740-7 5 5, 2014 . K. W ang , C . Go u ,an d F .- Y . W a ng , “ P ara llel vis io n: An A CP-b a s e d app roa ch t o int elligent visi oncomp u tin g,” * ActaAuto ma ticaSinica *, vo l. 42, no. 10, pp . 14 90-1500,201 6. K. W ang,C . Gou, N. Zh eng, J. M. Reh g , a nd F . -Y . Wang, “Parallelvi s i on for per ceptio n a nd understa nd ing ofc o mplex sce n es : Method s, fra m ew or k, andpe rspect iv es, ” * Artif i cial Intel ligenceRevie w *, vol. 48, no . 3, pp. 298-3 2 8, 2 01 7 . K. Wa ng, Y. Lu,Y. W a ng,Z. X i on g,a nd F. -Y. W an g ,“ Parallel imaging: A n ew the oreti cal framework for image g e neration ," * P at t ern Recognitio n and Artificia l Intelli gence *, vol.30, no. 7 , pp. 577- 587 , 2 017 . F . - Y. Wang, “Paral l e l co nt rol and ma nagemen t f orint ell ig ent trans portatio nsy st em s:Conce p ts, arch it ect ur es, anda pplica tions ,” * IE EE Tra nsactio n so n Int el li gent Tr an sport atio n Sy stems*, vol. 11, no . 3,pp .630-638 , 2010. F.-Y. W ang, “Para ll elcontro l : A metho d for data-driven and c o mputati ona l con trol ,” *ActaAut omatic a S i nica*, vol.39, n o. 4, p p. 29 3 - 30 2,20 14. F.-Y.W a ng, X. W an g, L . Li, a nd L. Li, “Steps t o war d parallel in tel lige n c e, " * I EE E /CA AJ our n a l of Automatica Sinica*,vo l .3, no. 4,p p.3 45 -348, 2 016. L. Li,Y . Lin,D. Cao, N . Zheng

Springer_Berlin Heidelberg,_vol. 5302, pp. 44-57,_2008. B._C._Russell, A._Torralba,_K. P. Murphy,_and W. T._Freeman, “LabelMe: A database_and web-based tool_for_image annotation,” *International Journal of Computer Vision*, vol. 77, no. 1, pp. 157-173, 2008._T.-Y._Lin, M._Maire,_S._Belongie, J. Hays, P. Perona,_D. Ramanan, P. Dollar, and_C. L._Zitnick, “Microsoft COCO: Common objects in context,” *in_Proceedings_of the European_Conference on Computer Vision*, pp. 740-755, 2014. K. Wang,_C. Gou, and F.-Y. Wang, “Parallel_vision: An ACP-based_approach_to_intelligent vision computing,” *Acta_Automatica Sinica*, vol. 42, no. 10,_pp. 1490-1500, 2016. K. Wang, C._Gou, N. Zheng, J. M. Rehg, and_F.-Y. Wang, “Parallel vision for perception_and understanding of complex scenes:_Methods, framework,_and perspectives,” *Artificial Intelligence Review*,_vol. 48, no._3, pp._298-328, 2017. K._Wang, Y. Lu, Y. Wang, Z._Xiong, and F.-Y._Wang, “Parallel imaging: A new theoretical_framework_for image generation,"_*Pattern_Recognition_and Artificial_Intelligence*, vol. 30,_no._7, pp._577-587,_2017. F.-Y. Wang, “Parallel control and_management_for intelligent transportation systems: Concepts, architectures, and_applications,” *IEEE Transactions on_Intelligent_Transportation Systems*, vol. 11,_no. 3, pp. 630-638, 2010._F.-Y. Wang, “Parallel control: A method_for data-driven_and computational_control,” *Acta Automatica Sinica*, vol. 39, no. 4, pp. 293-302, 2014._F.-Y. Wang, X. Wang, L. Li,_and L. Li, “Steps_toward parallel_intelligence,"_*IEEE/CAA Journal of_Automatica_Sinica*, vol._3, no. 4, pp.345-348, 2016. L. Li,_Y. Lin,_D. Cao, N. Zheng

\{ \Pi^{\text{c},\Lambda}_{\mu\nu\sigma\rho} \Gamma^\Lambda_{\nu\rho\gamma\delta} \Gamma^\Lambda_{\alpha\beta\sigma\mu} \nonumber \\ & & + \Pi^{\text{ph},\Lambda}_{\mu\nu\rho\sigma} \left[ \Gamma^\Lambda_{\beta\nu\gamma\rho} \Gamma^\Lambda_{\alpha\sigma\delta\mu} - \Gamma^\Lambda_{\alpha\nu\gamma\rho} \Gamma^\Lambda_{\beta\sigma\delta\mu} \right], \label{eq:flow:Gamma}\end{aligned}$$ where we have used the symmetries of $\Gamma$ to simplify the equations and abberviated $$\begin{aligned} \Pi^{\text{c},\Lambda}_{\mu\nu\sigma\rho} & := & P^{\Lambda}_{\mu\nu}(\Lambda) P^{\Lambda}_{\sigma\rho}(-\Lambda) \\ \Pi^{\text{ph},\Lambda}_{\mu\nu\sigma\rho} & := & P^{\Lambda}_{\mu\nu}(\Lambda) P^{\Lambda}_{\rho\sigma}(\Lambda) + P^{\Lambda}_{\mu\nu}(-\Lambda) P^{\Lambda}_{\rho\sigma}(-\Lambda).\end{aligned}$$ The full derivation may be found in Appendix \[app:T0StaticGamma\]. ### Initial conditions The initial conditions at $\Lambda\to\infty$ are given by $$\Sigma^{\Lambda\to\infty}_{\alpha\beta} = 0 \hspace{1em}\text{and}\hspace{1em} \Gamma^{\Lambda\to\infty}_{\alpha\beta\gamma\delta} = U_{\alpha\beta\gamma\delta}. \label{eq:initcond:SigmaGamma:Infty}$$ In order to solve the equations numerically, we need to choose an initial value $\Lambda_0$ that is still finite but larger than all other energy scales in the system. For $\Lambda > \Lambda_0$ one may assume a form of $({\mathrm{i}}\omega)^{-1}\mathds{1}$ for the propagator, allowing us to analytically integrate the flow equations from $\infty$ to $\Lambda_0$. In case of the flow equation for the vertex, power counting in $U$ and $\Lambda_0$

\ { \Pi^{\text{c},\Lambda}_{\mu\nu\sigma\rho } \Gamma^\Lambda_{\nu\rho\gamma\delta } \Gamma^\Lambda_{\alpha\beta\sigma\mu } \nonumber \\ & & + \Pi^{\text{ph},\Lambda}_{\mu\nu\rho\sigma } \left [ \Gamma^\Lambda_{\beta\nu\gamma\rho } \Gamma^\Lambda_{\alpha\sigma\delta\mu } - \Gamma^\Lambda_{\alpha\nu\gamma\rho } \Gamma^\Lambda_{\beta\sigma\delta\mu } \right ], \label{eq: flow: Gamma}\end{aligned}$$ where we have used the symmetries of $ \Gamma$ to simplify the equations and abberviated $ $ \begin{aligned } \Pi^{\text{c},\Lambda}_{\mu\nu\sigma\rho } &: = & P^{\Lambda}_{\mu\nu}(\Lambda) P^{\Lambda}_{\sigma\rho}(-\Lambda) \\ \Pi^{\text{ph},\Lambda}_{\mu\nu\sigma\rho } &: = & P^{\Lambda}_{\mu\nu}(\Lambda) P^{\Lambda}_{\rho\sigma}(\Lambda) + P^{\Lambda}_{\mu\nu}(-\Lambda) P^{\Lambda}_{\rho\sigma}(-\Lambda).\end{aligned}$$ The wide deriving may be found in Appendix   \[app: T0StaticGamma\ ]. # # # Initial conditions The initial conditions at $ \Lambda\to\infty$ are yield by $ $ \Sigma^{\Lambda\to\infty}_{\alpha\beta } = 0 \hspace{1em}\text{and}\hspace{1em } \Gamma^{\Lambda\to\infty}_{\alpha\beta\gamma\delta } = U_{\alpha\beta\gamma\delta }. \label{eq: initcond: SigmaGamma: Infty}$$ In order to solve the equation numerically, we need to choose an initial value $ \Lambda_0 $ that is still finite but bigger than all other energy scale in the system. For $ \Lambda > \Lambda_0 $ one may assume a form of $ ({ \mathrm{i}}\omega)^{-1}\mathds{1}$ for the propagator, allowing us to analytically integrate the flow equation from $ \infty$ to $ \Lambda_0$. In case of the flow equation for the vertex, might counting in $ U$ and $ \Lambda_0 $

\{ \Pi^{\text{c},\Lambda}_{\mu\nm\sigma\rho} \Gamka^\Lambsa_{\nu\rho\gxmma\delta} \Gamma^\Lambda_{\alpha\beva\sitma\mu} \nonumber \\ & & + \Pp^{\text{ph},\Lanbda}_{\ny\nu\rho\sigme} \leff[ \Gamke^\Lambda_{\beta\nu\gakma\rho} \Gamka^\Lambda_{\alpha\shgoa\belta\mu} - \Gamma^\Lambda_{\al[ha\nu\gakmw\rho} \Gamma^\Lambqa_{\beua\sygma\svlua\mu} \right], \label{eq:flkw:Gamma}\tnd{aligned}$$ where wr have used the symmetries of $\Hamma$ to simplify hhe equatiobs age abberviatea $$\begin{aligned} \Pi^{\text{c},\Mambda}_{\mu\nu\sigma\rho} & := & P^{\Lambda}_{\mu\vu}(\Lamyda) P^{\Lambda}_{\witma\gvo}(-\Lambda) \\ \Pm^{\text{pr},\Lambda}_{\mu\nu\slbma\rho} & := & P^{\Lakbda}_{\mu\nu}(\Lambda) P^{\Lembdq}_{\rho\sigma}(\Lambda) + P^{\Lqmvda}_{\mu\tu}(-\Lakbda) P^{\Laobdz}_{\rio\sjgma}(-\Lalbde).\end{aligned}$$ The full dwrivation may be foinq in Appendix \[alp:T0StaeisGamma\]. ### Initial conditions The initial cotdifions at $\Lambda\to\infty$ qre given by $$\Sigma^{\Lamhda\to\inftr}_{\alpha\beta} = 0 \hspace{1em}\text{and}\hspace{1em} \Camma^{\Mxmbbq\to\inwry}_{\wlpha\beta\gamma\delta} = U_{\alpha\beta\gamma\delta}. \labqm{ea:ikitcond:SigmaGammc:Infty}$$ In order yo spjve the equatkons nbjedically, we need to choose an ibitial vajue $\Kambda_0$ that is still finite but larger nhan all other energy dcales in tke sysjem. Fot $\Lambda > \Lambda_0$ one mav assuje a form ov $({\mathrm{i}}\koega)^{-1}\mathds{1}$ for tfe iropdgator, allowing us to analrtically mntegxate the floe equaeions from $\inftn$ to $\Lambda_0$. In case of tke flmw equatioj for the vertex, power counting in $U$ and $\Lambca_0$

\{ \Pi^{\text{c},\Lambda}_{\mu\nu\sigma\rho} \Gamma^\Lambda_{\nu\rho\gamma\delta} \Gamma^\Lambda_{\alpha\beta\sigma\mu} \nonumber \\ & \Pi^{\text{ph},\Lambda}_{\mu\nu\rho\sigma} \Gamma^\Lambda_{\beta\nu\gamma\rho} \Gamma^\Lambda_{\alpha\sigma\delta\mu} \Gamma^\Lambda_{\alpha\nu\gamma\rho} \Gamma^\Lambda_{\beta\sigma\delta\mu} \right], the of $\Gamma$ to the equations and $$\begin{aligned} \Pi^{\text{c},\Lambda}_{\mu\nu\sigma\rho} & := & P^{\Lambda}_{\mu\nu}(\Lambda) \\ \Pi^{\text{ph},\Lambda}_{\mu\nu\sigma\rho} & := & P^{\Lambda}_{\mu\nu}(\Lambda) P^{\Lambda}_{\rho\sigma}(\Lambda) + P^{\Lambda}_{\mu\nu}(-\Lambda) P^{\Lambda}_{\rho\sigma}(-\Lambda).\end{aligned}$$ The full derivation be found in Appendix \[app:T0StaticGamma\]. ### Initial conditions The initial conditions at $\Lambda\to\infty$ given $$\Sigma^{\Lambda\to\infty}_{\alpha\beta} 0 \Gamma^{\Lambda\to\infty}_{\alpha\beta\gamma\delta} = U_{\alpha\beta\gamma\delta}. \label{eq:initcond:SigmaGamma:Infty}$$ In order to solve the equations numerically, we need to choose an value $\Lambda_0$ that is still finite but larger all other energy scales the system. For $\Lambda > one assume a of for propagator, allowing us analytically integrate the flow equations from $\infty$ to $\Lambda_0$. In case of the flow equation for the power counting and $\Lambda_0$

\{ \Pi^{\text{c},\Lambda}_{\mu\nu\sigma\rho} \gamma^\LambdA_{\nu\rhO\gaMma\DeLta} \GAmma^\lambda_{\alpha\betA\SigmA\mu} \nonumber \\ & & + \Pi^{\text{ph},\LamBda}_{\mu\Nu\RHo\siGMa} \Left[ \GAmma^\LamBDa_{\BETa\nU\gAmMa\rHo} \gAmMa^\LamBda_{\Alpha\siGma\delta\mu} - \gamMa^\lambda_{\alpha\nU\GaMma\rho} \GammA^\LaMbda_{\beta\sigmA\deLta\mu} \rIgHt], \lABel{eq:FloW:GammA}\end{alIGned}$$ whEre we have UsED the syMMetries OF $\gaMma$ tO simplify the equatIOnS And abberviated $$\Begin{aLiGNeD} \pI^{\teXt{c},\lambda}_{\mu\nu\SiGma\rhO} & := & p^{\Lambda}_{\MU\nU}(\lAMbdA) p^{\Lambda}_{\sigma\rHo}(-\Lambda) \\ \Pi^{\tEXt{pH},\LambdA}_{\mU\nu\SIgma\rhO} & := & P^{\LamBdA}_{\Mu\nU}(\Lambda) P^{\LamBda}_{\rHo\sigma}(\LaMbda) + P^{\LAMbda}_{\mu\nU}(-\lambda) P^{\lambda}_{\Rho\SigMa}(-\LaMBdA).\eNd{aLiGNed}$$ tHe FulL DerIvation mAy Be Found In ApPENDIx \[apP:T0STatiCGammA\]. ### Initial condiTioNs ThE IniTial cOnditIons At $\lambdA\to\infTy$ are GiVen by $$\Sigma^{\LambdA\to\iNfty}_{\alpha\BetA} = 0 \hSpaCe{1Em}\texT{And}\hspAce{1Em} \GAmma^{\LamBda\to\inFTy}_{\aLpHA\BEtA\gamma\delta} = U_{\alpha\bEtA\GAmMa\delta}. \lAbel{eq:INiTcONd:SigmaGAmMa:INfty}$$ iN Order To soLVe The equatIons nuMErIcAlly, we nEeD to choOsE an IniTial vALue $\LAmbda_0$ tHat is stiLl finITe but larger thaN All other energY ScALEs IN the SysTem. For $\LambdA > \LamBDa_0$ onE may ASsUme A Form oF $({\mathRm{I}}\OmEGa)^{-1}\mathds{1}$ for the propaGaTor, allOwing Us to analyticaLly integraTE THe flow eqUatiONs FRom $\infty$ to $\LambDa_0$. In cAse of the flOW equatioN for tHe vertex, Power counTINg in $U$ and $\lamBda_0$

\{ \Pi^{\text{c },\Lambda} _{\mu \nu \si gm a\rh o} \Gam m a^\L ambda_{\nu\rho\gamma\d elta} \ G amma ^ \L ambda _{\alph a \b e t a\s ig ma \mu } \no number\\ & & + \Pi^{\tex t {p h},\Lambda }_{ \mu\nu\rho\s igm a} \lef t[\Gamm a^\Lam b da_{\b eta\nu\ga mm a \rho}\ Gamma^\ L a mb da_{ \alpha\sigma\delt a \m u } - \ Ga m ma ^ \ Lam bda _{\alpha\n u\ gamma \ rho} \G a mm a ^ \ Lam b da_{\beta\sig ma\delta\mu } \ri g ht], \l ab e l{e q:flow:Gamm a}\e nd{aligne d}$$ w h ere weh ave use d thesym met ries of $ \Ga mm a $ t o s imp l ify the equ at io ns an d ab b e r v iate d $ $\be gin{a ligned} \Pi^ {\t ext{ c },\ Lambd a}_{\ mu\n u\ sigma \rho}& :=&P^{\Lambda}_{\m u\nu }(\Lambda ) P ^{ \La mb da}_{ \ sigma\ rho }(- \Lambda ) \\ \ P i^{ \t e x t {p h},\Lambda}_{\mu\n u\ s i gm a\rho} & := &P ^{ \L a mbda}_{\ mu \nu }(\L a m bda)P^{\ L am bda}_{\r ho\sig m a} (\ Lambda) + P^{\Lamb d a}_{\mu\nu}(- \ La m b da ) P^{ \La mbda}_{\rho \sig m a}(- \Lam b da ).\ e nd{al igned }$ $ T h e full derivation m ay be fo und i n Appendix \[ app:T0Stat i c G amma\]. ### In i tial condition s Th e initialc ondition s at$\Lambda \to\infty $ are give n b y $ $\S igm a ^ {\ Lambda\to\inf t y }_{\ al pha\bet a} = 0 \h sp ace{1em}\ text{and }\ hs pa ce {1e m} \ G amma^{\L am bda \t o\i nfty} _ {\alph a\bet a\ga mm a\ d elt a} = U_ { \a l p ha\b et a\ gamm a\d el ta}. \la b el{ eq:init cond:Sigm aGa m ma:I nf ty }$$ Inorder to solv ethe equati on s n umeric a l ly, we n eed to choose an initia l value$\L ambda _0$that is s til l fini teb ut lar ger th an al loth e r ener g y s cal es in the sy s t em. For$\ Lamb da > \L ambda_0$ one may a s sum e a form of $ ({\ math r m {i }}\ o me g a)^ {- 1 }\m a t hds{1}$ for the propagato r, al lowing ust o a na lytical ly inte grate the flo w equatio ns from $ \i nfty $ to$\Lambda_0 $. In ca se of the flowe qu ation fo r theve rte x, po wer co u nti ng in $U$ a nd $\Lam bda_0 $

\{ _ _ _ __\Pi^{\text{c},\Lambda}_{\mu\nu\sigma\rho} __ _ _ \Gamma^\Lambda_{\nu\rho\gamma\delta} \Gamma^\Lambda_{\alpha\beta\sigma\mu}_ __ \nonumber \\ & & __ +_\Pi^{\text{ph},\Lambda}_{\mu\nu\rho\sigma} __ _ \left[ \Gamma^\Lambda_{\beta\nu\gamma\rho}_\Gamma^\Lambda_{\alpha\sigma\delta\mu} _ __ _ - \Gamma^\Lambda_{\alpha\nu\gamma\rho} \Gamma^\Lambda_{\beta\sigma\delta\mu} _ \right], _ _\label{eq:flow:Gamma}\end{aligned}$$_where_we have used the_symmetries of $\Gamma$ to simplify the_equations and abberviated $$\begin{aligned} \Pi^{\text{c},\Lambda}_{\mu\nu\sigma\rho} &_:= & P^{\Lambda}_{\mu\nu}(\Lambda) P^{\Lambda}_{\sigma\rho}(-\Lambda) \\ \Pi^{\text{ph},\Lambda}_{\mu\nu\sigma\rho} &_:= & P^{\Lambda}_{\mu\nu}(\Lambda) P^{\Lambda}_{\rho\sigma}(\Lambda) _ _ _ _ _ _ _ _ _ __ ___ _ _+_P^{\Lambda}_{\mu\nu}(-\Lambda) P^{\Lambda}_{\rho\sigma}(-\Lambda).\end{aligned}$$_The_full derivation may be found in_Appendix \[app:T0StaticGamma\]. ###_Initial conditions The initial conditions at $\Lambda\to\infty$ are_given by $$\Sigma^{\Lambda\to\infty}_{\alpha\beta} __ _ _ = 0 \hspace{1em}\text{and}\hspace{1em} \Gamma^{\Lambda\to\infty}_{\alpha\beta\gamma\delta} =_U_{\alpha\beta\gamma\delta}. \label{eq:initcond:SigmaGamma:Infty}$$_In order_to solve the equations numerically, we need to choose an initial_value $\Lambda_0$ that is still finite_but larger than all_other energy_scales_in the system._For_$\Lambda >_\Lambda_0$ one may assume a form of_$({\mathrm{i}}\omega)^{-1}\mathds{1}$ for_the propagator, allowing us to analytically_integrate the flow equations_from_$\infty$ to $\Lambda_0$. In case of_the flow equation for the vertex,_power counting in $U$ and_$\Lambda_0$

be a valuation domain and $\mathcal{V}$ be a definable coaisle in $\operatorname*{\mathbf{D}}(R)$. Then the two following conditions are equivalent: 1. the t-structure $(\mathcal{U},\mathcal{V})$ is compactly generated, 2. $\mathcal{V}$ arises from a chain of flat ring epimorphisms as in Proposition \[P:construction\]. Non-degeneracy and unbounded cosilting objects ---------------------------------------------- The final goal is to restrict Theorem \[T02\] to those definable coaisles, which belong to non-degenerate t-structures. In other words, to identify those definable coaisles, which are induced by a pure-injective cosilting object of $\operatorname*{\mathbf{D}}(R)$ (see Corollary \[C:rosie3\]). We will show that the right part of the non-degeneracy condition can only be achieved if the coaisle is cohomologically bounded below. On the other hand, we will exhibit in Example \[EX1\] a definable coaisle for which the left part of the non-degeneracy condition is achieved non-trivially. In other words, any coaisle induced by a pure-injective cosilting complex over a valuation domain is cohomologically bounded from below, but there are such which are not co-intermediate. In particular, any pure-injective cosilting complex over a valuation domain is cohomologically bounded below, but there are pure-injective cosilting complexes which are [not]{} bounded cosilting complexes. \[L:BW\] Let $R$ be a valuation domain and $\mathcal{V}$ be a definable coaisle Then $\bigcap_{n \in \mathbb{Z}} \mathcal{V}[n] = 0$ if and only if there is $l \in \mathbb{Z}$ such that $\mathcal{V} \subseteq \operatorname*{\mathbf{D}}^{>l}$. The “if” statement is trivial, thus we need to show just the “only if” implication. Let $\Theta(\mathcal{V}) = (\mathcal{X}_n \mid n \in \mathbb{Z})$ be the admissible filtration corresponding to $\mathcal{V}$. Since $\bigcap_{n \in \mathbb{Z}} \mathcal{V}[n] = 0$,

be a valuation domain and $ \mathcal{V}$ be a definable coaisle in $ \operatorname*{\mathbf{D}}(R)$. Then the two following conditions are equivalent: 1. the metric ton - social organization $ (\mathcal{U},\mathcal{V})$ is compactly generated, 2. $ \mathcal{V}$ arise from a range of flat ring epimorphisms as in Proposition   \[P: construction\ ]. Non - degeneracy and unbounded cosilting object ---------------------------------------------- The final goal is to qualify Theorem   \[T02\ ] to those definable coaisles, which belong to to non - degenerate t - structure. In other words, to identify those definable coaisles, which are induce by a pure - injective cosilting aim of $ \operatorname*{\mathbf{D}}(R)$ (see Corollary   \[C: rosie3\ ]). We will usher that the correct part of the non - degeneracy condition can only be achieve if the coaisle is cohomologically bounded below. On the early hand, we will show in Example   \[EX1\ ] a definable coaisle for which the left share of the non - degeneracy condition is achieved non - trivially. In other quarrel, any coaisle induced by a pure - injective cosilting complex over a valuation domain is cohomologically restrict from below, but there are such which are not co - intermediate. In particular, any pure - injective cosilting complex over a valuation domain is cohomologically bounded below, but there embody pure - injective cosilting complexes which are [ not ] { } bounce cosilting complex. \[L: BW\ ] Let $ R$ be a valuation domain and $ \mathcal{V}$ be a definable coaisle Then $ \bigcap_{n \in \mathbb{Z } } \mathcal{V}[n ] = 0 $ if and merely if there is $ l \in \mathbb{Z}$ such that $ \mathcal{V } \subseteq \operatorname*{\mathbf{D}}^{>l}$. The “ if ” statement is trivial, therefore we need to show just the “ only if ” implication. Let $ \Theta(\mathcal{V }) = (\mathcal{X}_n \mid n \in \mathbb{Z})$ be the admissible filtration corresponding to $ \mathcal{V}$. Since $ \bigcap_{n \in \mathbb{Z } } \mathcal{V}[n ] = 0 $,

be a valuation domain and $\oathcal{V}$ be a dgfunable coaisme in $\opdratorname*{\mathbf{D}}(R)$. Then the vwo dolloqing conditions are eqjivalent: 1. the t-steuctnre $(\mathcal{U},\mathrzl{V})$ is gjmpadbly gznxrated, 2. $\mathcal{Y}$ arises frmm a chain of xlxt ring epimorphisms as in Propositiog \[P:consttuftion\]. Non-degenetacy sgd uhbounded cosilting objects ---------------------------------------------- The finam goal ps to restrict Throrem \[T02\] to those definable foaidles, which belong ho non-degengdatq t-structures. In other words, to idehtify those definable coaisles, dhich are inducgb by d pure-injecvive cjsilting objcbt of $\o[eratormame*{\mathbf{D}}(R)$ (sce Cocollqry \[C:rosie3\]). We will shox that the right parj of the nmn-begeneracy condition xab onli be dchidced if tie doaislf ia cohomolofically boubded below. On the ouhew hand, we will exhibyt in Example \[EX1\] a definable coaisle for wvicg the left part of the bon-degeneracy conditiln is achyeved non-trivially. In other words, any coaisle indgced uy a kuvc-injdxtlve cosilting complex over a valuation domain js cphomologically bounded frpm brjow, but there are sbdh which are not co-ijtermedyate. Un partictlar, any pure-injective cosiltint complex ovvr a valuation domain ns cohomologncally bounced below, but there are purs-injective fosilting zomplexes which xre [nmt]{} boundta cosilting complqxes. \[L:BW\] Lxt $R$ ye a valjatipn domwin and $\mahhcal{Y}$ be a definable cowisle Tven $\bigcap_{j \in \mathbb{Z}} \mathcal{V}[n] = 0$ if and only if there iv $l \in \mathyb{Z}$ sugh that $\mathcal{D} \subseteq \opetatorname*{\iathbw{D}}^{>l}$. The “if” atatemeit is triviaj, thus we neeg to show jusv the “onlr if” impoicatiov. Let $\Theta(\mathval{V}) = (\matkeal{X}_n \mid n \in \mathbb{Z})$ be tme adojssible filtratnub corresponding to $\maehbal{T}$. Sinsa $\bigcap_{n \in \matfbb{X}} \mathzal{V}[n] = 0$,

be a valuation domain and $\mathcal{V}$ be coaisle $\operatorname*{\mathbf{D}}(R)$. Then two following conditions $(\mathcal{U},\mathcal{V})$ compactly generated, 2. arises from a of flat ring epimorphisms as in \[P:construction\]. Non-degeneracy and unbounded cosilting objects ---------------------------------------------- The final goal is to restrict \[T02\] to those definable coaisles, which belong to non-degenerate t-structures. In other words, identify definable which induced by a pure-injective cosilting object of $\operatorname*{\mathbf{D}}(R)$ (see Corollary \[C:rosie3\]). We will show that the part of the non-degeneracy condition can only be if the coaisle is bounded below. On the other we exhibit in \[EX1\] definable for which the part of the non-degeneracy condition is achieved non-trivially. In other words, any coaisle induced by a pure-injective complex over domain is bounded below, there are such not co-intermediate. In particular, any pure-injective a valuation domain is cohomologically bounded below, but are pure-injective complexes which are [not]{} bounded cosilting \[L:BW\] Let $R$ be a valuation domain and be a definable coaisle Then $\bigcap_{n \in \mathbb{Z}} \mathcal{V}[n] = 0$ if and only if $l \in \mathbb{Z}$ such $\mathcal{V} \subseteq \operatorname*{\mathbf{D}}^{>l}$. “if” is thus need to just the “only if” implication. Let $\Theta(\mathcal{V}) = (\mathcal{X}_n \mid n \mathbb{Z})$ be the admissible filtration corresponding to $\mathcal{V}$. Since $\bigcap_{n \mathcal{V}[n] 0$,

be a valuation domain and $\mathCal{V}$ be a defInablE coAisLe In $\opEratOrname*{\mathbf{D}}(R)$. tHen tHe two following conditioNs are EqUIvalENt: 1. The t-sTructurE $(\MaTHCal{u},\mAtHcaL{V})$ IS cOmpacTly GeneratEd, 2. $\mathcal{V}$ AriSeS from a chain oF FlAt ring epimOrpHisms as in ProPosItion \[P:CoNstRUctioN\]. NoN-degeNeracy ANd unboUnded cosiLtINg objeCTs ---------------------------------------------- The fiNAL gOal iS to restrict TheoreM \[t02\] tO Those definable CoaislEs, WHiCH BelOng To non-degenErAte t-sTRuctureS. in OTHEr wORds, to identify Those definaBLe cOaisleS, wHicH Are indUced bY a PUre-Injective coSiltIng object Of $\operATorname*{\MAthbf{D}}(R)$ (See CorOllAry \[c:rosIE3\]). WE wIll ShOW thAT tHe rIGht Part of thE nOn-DegenEracY CONDitiOn cAn onLy be aChieved if the cOaiSle iS CohOmoloGicalLy boUnDed beLow. On tHe othEr Hand, we will exhibIt in example \[EX1\] A deFiNabLe CoaisLE for whIch The Left parT of the nON-deGeNERAcY condition is achievEd NON-tRivially. in otheR WoRdS, Any coaisLe IndUced BY A pure-InjeCTiVe cosiltIng comPLeX oVer a valUaTion doMaIn iS coHomolOGicaLly bouNded from Below, BUt there are such WHich are not co-iNTeRMEdIAte. IN paRticular, any Pure-INjecTive COsIltINg comPlex oVeR A vALuation domain is cohoMoLogicaLly boUnded below, but There are puRE-INjective CosiLTiNG complexes whicH are [nOt]{} bounded cOSilting cOmpleXes. \[L:BW\] LeT $R$ be a valuATIon domaiN anD $\maThcAl{V}$ BE A dEfinable coaisLE then $\BiGcap_{n \in \MatHbb{Z}} \matHcaL{V}[n] = 0$ If aNd oNlY if there iS $l \in \mathBb{z}$ sUcH tHat $\MathcAL{V} \subsetEq \OpeRaTorName*{\mAThbf{D}}^{>l}$. the “if” StatEmEnT Is tRivial, tHUs WE Need To ShOw juSt tHe “Only iF” impLIcaTion. Let $\theta(\mathCal{v}) = (\MathCaL{X}_N \mid n \in \Mathbb{Z})$ be the aDmIssible filTrAtiOn corrESPonding tO $\mathcal{V}$. Since $\bigcap_{n \in \MAthbb{Z}} \mAthCal{V}[n] = 0$,

be a valuation domain and $\mathcal {V}$bea d ef inab le c oaisle in $\op e rato rname*{\mathbf{D}}(R)$ . The nt he t w ofollo wing co n di t i ons a re eq ui v al ent: 1. the t -structure $( \m athcal{U},\m a th cal{V})$ i s c ompactly gen era ted, 2. $ \ mathc al{ V}$ a risesf rom achain offl a t ring epimorp h i sm s as in Proposition \ [ P: c onstruction\]. Non- de g en e r acy an d unbounde dcosil t ing obj e ct s - --- - ------------- ----------- - --- ------ -- --- - - The fina lg oal is to rest rict Theorem\[T02\ ] to tho s e defin able c oai sle s, w h ic hbel on g to no n-d e gen erate t- st ru cture s. I n o t herwor ds,to id entify thosedef inab l e c oaisl es, w hich a re in ducedby apu re-injective co silt ing objec t o f$\o pe rator n ame*{\ mat hbf {D}}(R) $ (seeC oro ll a r y  \ [C:rosie3\]). We w il l sh ow thatthe ri g ht p a rt of th enon -deg e n eracy con d it ion canonly b e a ch ieved i fthe co ai sle is coho m olog ically bounded belo w . On the other hand, we will ex h i bi t inExa mple \[EX1\ ] ad efin able co ais l e for whic ht he left part of the no n- degene racycondition isachieved n o n - triviall y. I n o t her words, any coai sle induce d by a pu re-in jectivecosilting c omplex o ver aval uat i o ndomain is coh o m olog ic ally bo und ed from be low , b utth ere are s uch whic har eno t c o-int e rmediate .Inpa rti cular , any p ure-i njec ti ve cos iltingc om p l ex o ve ra va lua ti on do main iscohomol ogicallybou n dedbe lo w, butthere are pur e- injectiveco sil ting c o m plexes w hich are [not]{} bounde d cosilt ing comp lexe s. \[L:B W\] Let $ R$b e a va luatio n dom ai n a n d $\ma t h ca l{V }$ be a defi n a ble coai sl e Th en $\bi gcap_{n \in \mathb b {Z} } \mathcal{V} [n] = 0 $ if an d o n lyif the r e is $l \in \mat hbb{Z}$ su ch th at $\mathc a l{V }\subset eq \ope rator n ame*{\m athbf{D}} ^{>l}$. Th e “i f ” st atement is trivial , thus we needt oshowjus t the“o nly if”implic a tio n. Le t $\Th et a(\mat hcal{ V} ) = (\ma thcal{X}_n \mid n \in \ mathbb {Z})$ be the admi ssi b lefiltratio n co rrespondin g t o $ \math cal { V}$.Sinc e $ \bi g cap_{ n \i n \mathbb{ Z }} \m a t hc al{V}[n] =0 $ ,

be_a valuation_domain and $\mathcal{V}$ be_a definable_coaisle_in $\operatorname*{\mathbf{D}}(R)$._Then_the two following_conditions are equivalent: 1._ the t-structure $(\mathcal{U},\mathcal{V})$_is compactly generated, 2.__$\mathcal{V}$ arises from a chain of flat ring epimorphisms as in Proposition \[P:construction\]. Non-degeneracy and unbounded_cosilting_objects ---------------------------------------------- The final_goal_is_to restrict Theorem \[T02\] to those_definable coaisles, which belong to_non-degenerate t-structures._In other words, to identify those definable coaisles,_which_are induced by_a pure-injective cosilting object of $\operatorname*{\mathbf{D}}(R)$ (see Corollary \[C:rosie3\]). We_will show that the right part_of the non-degeneracy_condition_can_only be achieved if_the coaisle is cohomologically bounded below._On the other hand, we will_exhibit in Example \[EX1\] a definable coaisle for_which the left part of the_non-degeneracy condition is achieved non-trivially._In other_words, any coaisle induced by_a pure-injective cosilting_complex over_a valuation domain_is cohomologically bounded from below, but_there are such_which are not co-intermediate. In particular,_any_pure-injective cosilting complex_over_a_valuation domain_is cohomologically bounded_below,_but there_are_pure-injective cosilting complexes which are [not]{}_bounded_cosilting complexes. \[L:BW\] Let $R$ be a valuation_domain and $\mathcal{V}$ be_a_definable coaisle Then $\bigcap_{n_\in \mathbb{Z}} \mathcal{V}[n] = 0$_if and only if there is_$l \in_\mathbb{Z}$ such_that $\mathcal{V} \subseteq \operatorname*{\mathbf{D}}^{>l}$. The “if” statement is trivial, thus we need_to show just the “only if”_implication. Let $\Theta(\mathcal{V}) =_(\mathcal{X}_n \mid_n_\in \mathbb{Z})$ be_the_admissible filtration_corresponding to $\mathcal{V}$. Since $\bigcap_{n \in \mathbb{Z}}_\mathcal{V}[n] =_0$,

}{\omega^2}\right)^{1/2}\frac{dr}{r}=\frac{\pi(n+\alpha)}{\omega}. \label{eq-osc26}$$ Since the lower integral limit, $r_1$ depends only on the ratio $L/\omega$, then the whole left-hand side is a function of only one parameter, $L/\omega$, that is: $$F\left(\frac{L}{\omega}\right)=\frac{\pi(n+\alpha)}{\omega}.$$ This relation represents so-called Duvall’s law [@Duvall1982]. It means that a 2D dispersion relation $\omega=\omega(n,l)$ is reduced to the 1D relation between two ratios $L/\omega$ and $(n+\alpha)/\omega$. With an appropriate choice of parameter $\alpha$ (e.g. 1.5) these ratios can easily calculated from a table of observed solar frequencies. An example of such calculations shown in Fig. \[fig20\]) illustrates that the Duvall’s law holds quite well for the observed solar modes. The short bottom branch that separates from the main curve correspond to f-modes. ![The observed Duvall’s law relation for modes of $l=0-250$.[]{data-label="fig20"}](kosovichev_fig20.eps){width="0.6\linewidth"} Asymptotic sound-speed inversion -------------------------------- The Duvall’s law demonstrate that the asymptotic theory provides a rather accurate description of the observed solar p-modes. Thus, it can be used for solving the inverse problem of helioseismology - determination of the internal properties from the observed frequencies. Theoretically, the internal structure of the Sun is described by the stellar evolution theory [@Christensen-Dalsgaard1996]. This theory calculates the thermodynamic structure of the Sun during the evolution on the Main Sequence. The evolutionary model of the current age, $\approx 4.6\times 10^9$ years, is called the standard solar model. Helioseismology provides estimates of the interior properties, such as the sound-speed profiles, that can be compared with the predictions of the standard model. Our goal is to find corrections to a solar model from the observed frequency differences between the Sun and the model using the asymptotic formula for the Duvall’s law [@Christensen-Dalsgaard1988]. We consider a small perturbation of the sound-speed, $c\

} { \omega^2}\right)^{1/2}\frac{dr}{r}=\frac{\pi(n+\alpha)}{\omega }. \label{eq - osc26}$$ Since the lower integral limit, $ r_1 $ depends merely on the proportion $ L/\omega$, then the whole left - bridge player english is a function of only one argument, $ L/\omega$, that is: $ $ F\left(\frac{L}{\omega}\right)=\frac{\pi(n+\alpha)}{\omega}.$$ This relation represents so - call Duvall ’s law [ @Duvall1982 ]. It means that a 2D distribution relation $ \omega=\omega(n, l)$ is reduced to the 1D relation between two ratios $ L/\omega$ and $ (n+\alpha)/\omega$. With an appropriate option of parameter $ \alpha$ (e.g. 1.5) these ratio can easily calculated from a mesa of observed solar frequencies. An example of such calculation shown in Fig.   \[fig20\ ]) illustrates that the Duvall ’s law hold quite well for the observed solar modes. The short bottom branch that separates from the chief curve correspond to f - modes. ! [ The observed Duvall ’s law relation for modes of $ l=0 - 250$.[]{data - label="fig20"}](kosovichev_fig20.eps){width="0.6\linewidth " } Asymptotic sound - speed inversion -------------------------------- The Duvall ’s law demonstrate that the asymptotic theory provides a rather accurate description of the observed solar p - modes. therefore, it can be use for solving the inverse trouble of helioseismology - decision of the inner properties from the observed frequencies. Theoretically, the internal structure of the Sun is described by the stellar development theory [ @Christensen - Dalsgaard1996 ]. This theory calculates the thermodynamic structure of the Sun during the evolution on the Main Sequence. The evolutionary model of the current age, $ \approx 4.6\times 10 ^ 9 $ years, is called the standard solar model. Helioseismology provides estimates of the home properties, such as the heavy - speed profiles, that can be compared with the predictions of the standard model. Our finish is to find corrections to a solar model from the observed frequency differences between the Sun and the model using the asymptotic formula for the Duvall ’s law [ @Christensen - Dalsgaard1988 ]. We regard a small disruption of the sound - speed, $ c\

}{\omeha^2}\right)^{1/2}\frac{dr}{r}=\frac{\pi(n+\alpma)}{\omega}. \label{eq-ose26}$$ Since the lkwer intdgral limit, $r_1$ depends only oi thw ratuo $L/\omega$, then the whoue left-hajd side us a dunction oh only okz parzletex, $O/\omega$, that is: $$F\left(\frac{N}{\omega}\right)=\fraw{\pk(n+\clpha)}{\omega}.$$ This relation represents fo-callec Fuvall’s law [@Dudall1982]. Yt msans that a 2D dispersion relation $\kmega=\omtga(n,l)$ is reduced tp the 1D relation between tao rwtios $L/\omega$ and $(n+\wlpha)/\omega$. Qith qn appropriage choice of parameter $\alpha$ (e.g. 1.5) these ratios can easkly cclculated ftoj w table of ouserveq solar freqmvncies. Dn examlle of such cakcunatuons shown in Fig. \[fig20\]) mllustrates that the Duvall’s nac holds quite well foe rhe otsereed rilaf mkdxs. Fhe shlrt bottom brznch that swparates from the msig curve correslond tj s-modes. ![The observed Duvall’s law relation fod modes of $l=0-250$.[]{data-label="fit20"}](kosovichev_fig20.eps){widtj="0.6\linewidtr"} Asymptotic sound-speed inversion -------------------------------- The Duvall’s law damonsvrxte tmat gye asymptotic theory provides a rather accurate sexcgiption of the obferved solat o-mpqes. Thus, it cxn be baes for solving the lnverse proboem of hejiosrismology - determination of the internaj properties from tke observed yrequemcies. Theoretically, the intexnal sfructure of the Sun jr described by tfe xtallar evolution theory [@Chrystensen-Delsgacrd1996]. This thepry cajculates tje thcsmodynamic structuge of tve Sun durlng the evolution on the Main Sxauence. The efonutponary mobel of the current wge, $\approx 4.6\timgs 10^9$ years, is cxlled the atandarv solar modej. Helioseismonlgy provides estimatqs od thw interkur properties, xuch as tkt sound-spewd profiles, that cen bg dompared with tkt ptwdictions of tne rtagdwrv modqn. Our goal is to wina corrdctions to c rolat model from the obsarves frequency differrnges betwegn the Sug and the modrl using the asympuotic hormule for yhe Duvall’s law [@Christensen-Dalsgazrd1988]. We condidcr a small pewturnatijn of the xound-speed, $c\

}{\omega^2}\right)^{1/2}\frac{dr}{r}=\frac{\pi(n+\alpha)}{\omega}. \label{eq-osc26}$$ Since the lower integral limit, only the ratio then the whole of one parameter, $L/\omega$, is: $$F\left(\frac{L}{\omega}\right)=\frac{\pi(n+\alpha)}{\omega}.$$ This represents so-called Duvall’s law [@Duvall1982]. It that a 2D dispersion relation $\omega=\omega(n,l)$ is reduced to the 1D relation between ratios $L/\omega$ and $(n+\alpha)/\omega$. With an appropriate choice of parameter $\alpha$ (e.g. 1.5) ratios easily from table of observed solar frequencies. An example of such calculations shown in Fig. \[fig20\]) illustrates that Duvall’s law holds quite well for the observed modes. The short bottom that separates from the main correspond f-modes. ![The Duvall’s relation modes of $l=0-250$.[]{data-label="fig20"}](kosovichev_fig20.eps){width="0.6\linewidth"} sound-speed inversion -------------------------------- The Duvall’s law demonstrate that the asymptotic theory provides a rather accurate description of observed solar it can used solving inverse problem of determination of the internal properties from Theoretically, the internal structure of the Sun is by the evolution theory [@Christensen-Dalsgaard1996]. This theory calculates thermodynamic structure of the Sun during the evolution the Main Sequence. The evolutionary model of the current age, $\approx 4.6\times 10^9$ years, is standard solar model. Helioseismology estimates of the properties, as sound-speed that can compared with the predictions of the standard model. Our goal is find corrections to a solar model from the observed frequency the and the model the asymptotic formula for Duvall’s [@Christensen-Dalsgaard1988]. We consider a of sound-speed,

}{\omega^2}\right)^{1/2}\frac{dr}{r}=\frac{\pi(n+\aLpha)}{\omega}. \lAbel{eQ-osC26}$$ SiNcE the LoweR integral limit, $R_1$ DepeNds only on the ratio $L/\omegA$, then ThE WholE LeFt-hanD side is A FuNCTioN oF oNly OnE PaRametEr, $L/\Omega$, thAt is: $$F\left(\fRac{l}{\oMega}\right)=\fraC{\Pi(N+\alpha)}{\omegA}.$$ ThIs relation rePreSents sO-cAllED DuvaLl’s Law [@DuVall1982]. It MEans thAt a 2D dispeRsIOn relaTIon $\omegA=\OMeGa(n,l)$ Is reduced to the 1D reLAtIOn between two raTios $L/\oMeGA$ aND $(N+\alPha)/\Omega$. With aN aPpropRIate choICe OF PAraMEter $\alpha$ (e.g. 1.5) thEse ratios caN EasIly calCuLatED from a Table Of OBseRved solar frEqueNcies. An exAmple oF Such calCUlationS shown In FIg. \[fIg20\]) ilLUsTrAteS tHAt tHE DUvaLL’s lAw holds qUiTe Well fOr thE OBSErveD soLar mOdes. THe short bottom BraNch tHAt sEparaTes frOm thE mAin cuRve corRespoNd To f-modes. ![The obseRved duvall’s laW reLaTioN fOr modES of $l=0-250$.[]{daTa-lAbeL="fig20"}](kosOvichev_FIg20.ePs){WIDTh="0.6\Linewidth"} AsymptotiC sOUNd-Speed invErsion -------------------------------- tHe duVAll’s law dEmOnsTratE THat thE asyMPtOtic theoRy provIDeS a Rather aCcUrate dEsCriPtiOn of tHE obsErved sOlar p-modEs. ThuS, It can be used for SOlving the inveRSe PRObLEm of HelIoseismologY - detERminAtioN Of The INternAl proPeRTiES from the observed freQuEncies. theorEtically, the inTernal struCTURe of the SUn is DEsCRibed by the stelLar evOlution theORy [@ChristEnsen-dalsgaarD1996]. This theoRY CalculatEs tHe tHerModYNAmIc structure of THE Sun DuRing the EvoLution oN thE MaIn SEquEnCe. The evolUtionary MoDeL oF tHe cUrrenT Age, $\approX 4.6\tImeS 10^9$ yEarS, is caLLed the StandArd sOlAr MOdeL. HeliosEIsMOLogy PrOvIdes EstImAtes oF the INteRior proPerties, suCh aS The sOuNd-Speed prOfiles, that can Be Compared wiTh The PredicTIOns of the Standard model. Our goal is tO Find corRecTions To a sOlar model FroM the obSerVEd freqUency dIfferEnCes BETween THE SUn aNd The model usING thE asymPtOtic Formula For the Duvall’s law [@ChRIstEnsen-DalsgaarD1988]. We ConsIDEr A smALl PErtUrBAtiON Of the sound-speed, $C\

}{\omega^2}\right)^{1/2}\f rac{dr}{r} =\fra c{\ pi( n+ \alp ha)} {\omega}. \lab e l{eq -osc26}$$ Since the lo wer i nt e gral li mit,$r_1$ d e pe n d s o nl yonth e r atio$L/ \omega$ , then the wh ol e left-hands id e is a fun cti on of only o neparame te r,$ L/\om ega $, th at is: $$F\le ft(\frac{ L} { \omega } \right) = \ fr ac{\ pi(n+\alpha)}{\om e ga } .$$ This relat ion re pr e se n t s s o-c alled Duva ll ’s la w [@Duva l l1 9 8 2 ].I t means thata 2D disper s ion relat io n $ \ omega= \omeg a( n ,l) $ is reduce d to the 1D r elatio n betwee n two ra tios $ L/\ ome ga$a nd $ (n+ \a l pha ) /\ ome g a$. With an a pp ropri atec h o i ce o f p aram eter$\alpha$ (e.g . 1 .5)t hes e rat ios c an e as ily c alcula ted f ro m a table of ob serv ed solarfre qu enc ie s. An exampl e o f s uch cal culatio n s s ho w n in Fig. \[fig20\]) i ll u s tr ates tha t theD uv al l ’s law h ol dsquit e wellfort he observe d sola r m od es. The s hort b ot tom br ancht hatsepara tes from them ain curve corr e spond to f-mo d es . ! [ Theobs erved Duval l’sl aw r elat i on fo r mode s of$l = 0- 2 50$.[]{data-label=" fi g20"}] (koso vichev_fig20. eps){width = " 0 .6\linew idth " }Asymptotic sou nd-sp eed invers i on ----- ----- -------- --------- - - --- The Du val l’s la w de monstrate tha t theas ymptoti c t heory p rov ide s a ra th er accura te descr ip ti on o f t he ob s erved so la r p -m ode s. Th u s, itcan b e us ed f o r s olvingt he i nver se p robl emof heli osei s mol ogy - d eterminat ion of t he i nternal properties f ro m the obse rv edfreque n c ies. The oretically, the interna l struct ure of t he S un is des cri bed by th e stell ar evo lutio nthe o r y [@C h r is ten se n-Dalsgaar d 1 996 ]. Th is the ory cal culates the thermo d yna mic structure of the S un du r in g th ee vol u t ion on the Main Sequence. T h eevolutiona r y m od el of t he curr ent a g e, $\ap prox 4.6\ times 10^ 9$ yea r s , i s called t he stand ard solar model . H elios eis mology p rov idesestima t esof th e inte ri or pro perti es , such a s the sound-speed profi les, t hat c anbe compar edw ith the pred icti ons of the st and ard m ode l . Ou r go a list o fin d co r rectionst oa s o l ar model from t h e o bserv edf requen cy d ifferences betwee n the Sun and t he m o d elusi n g th easymptotic for mul af o r the Du va ll’s law [@ Christen se n -Dals gaard1 988]. We con s i de r a sma ll p ert urbationofth e sound- sp ee d , $c\

}{\omega^2}\right)^{1/2}\frac{dr}{r}=\frac{\pi(n+\alpha)}{\omega}. \label{eq-osc26}$$ Since_the lower_integral limit, $r_1$ depends_only on_the_ratio $L/\omega$,_then_the whole left-hand_side is a_function of only one_parameter, $L/\omega$, that_is:_$$F\left(\frac{L}{\omega}\right)=\frac{\pi(n+\alpha)}{\omega}.$$ This relation represents so-called Duvall’s law [@Duvall1982]. It means that a 2D dispersion_relation_$\omega=\omega(n,l)$ is_reduced_to_the 1D relation between two_ratios $L/\omega$ and $(n+\alpha)/\omega$. With_an appropriate_choice of parameter $\alpha$ (e.g. 1.5) these ratios_can_easily calculated from_a table of observed solar frequencies. An example of_such calculations shown in Fig. \[fig20\]) illustrates_that the Duvall’s_law_holds_quite well for the_observed solar modes. The short bottom_branch that separates from the main_curve correspond to f-modes. ![The observed Duvall’s law_relation for modes of $l=0-250$.[]{data-label="fig20"}](kosovichev_fig20.eps){width="0.6\linewidth"} Asymptotic sound-speed_inversion -------------------------------- The Duvall’s law demonstrate that_the asymptotic_theory provides a rather accurate_description of the_observed solar_p-modes. Thus, it_can be used for solving the_inverse problem of_helioseismology - determination of the internal_properties_from the observed_frequencies._Theoretically,_the internal_structure of the_Sun_is described_by_the stellar evolution theory [@Christensen-Dalsgaard1996]. This_theory_calculates the thermodynamic structure of the Sun_during the evolution on_the_Main Sequence. The evolutionary_model of the current age,_$\approx 4.6\times 10^9$ years, is called_the standard_solar model._Helioseismology provides estimates of the interior properties, such as the sound-speed_profiles, that can be compared with_the predictions of the_standard model. Our_goal_is to find_corrections_to a_solar model from the observed frequency differences_between the_Sun and the model using the_asymptotic formula for the_Duvall’s_law [@Christensen-Dalsgaard1988]. We consider a small perturbation_of the sound-speed, $c\

assume here that the atmosphere is homogeneous, we just use the reduced height that can be obtained for any real atmosphere just from the pressure and density at observer. Assuming furthermore ideal gas law for dry air we have: $$\label{eq:ell} \ell(T)={P \over {\rho \ g}}={P_0 \over {\rho_0 \ g_0}}{T \over T_0},$$ where ${\rho_0=1.293\ \mbox{kg}\,\mbox{m}^{-3}}$ for ${T_0=273.15\,\mbox{K}}$, ${P_0=101325\,\mbox{Pa}}$ and normal gravity ${g_0=9.80665\,\mbox{m}\,\mbox{s}^{-2}}$. The radius of curvature for Calern observatory (${\varphi=43\degr45\arcmin7\arcsec}$, ${h=1323\,\mbox{m}}$) was approximated by the minimum reference ellipsoid curvature at latitude $45\degr$ and sea level (Chollet [@Chollet], see Appendix \[App:curvature\]): $$r_c(45\degr,0)=6367.512\ \mathrm{km}$$ Ambient air refractivity was deduced from the refractive index $n_0(\lambda)$ under standard conditions and the partial pressure of water vapor $p$ by applying the formula recommended by the first resolution of the 13th General Assembly of the International Union of Geodesy and Geophysics (IUGG [@IUGG63]; Baldini [@Baldini]). After conversion to Pa (Pascal) as the pressure unit, the equation becomes: $$\label{eq:alpha} \alpha(T,P,f_h,\lambda)={T_0 \over T} \left\{(n_0(\lambda)-1) {P \over P_0} - 4.13\, 10^{-10}\ p(f_h,T)\right\}$$ Refractivity under standard condition (sea level, ${T=T_0}$, ${P=P_0}$, $0\%$ humidity, $0.03\%$ of carbon dioxide) was taken from the work of Barrel and Sears ([@Barrel]): $$\label{eq:Barrel} n_0(\lambda)-1=\left\{2876.04+{16.288 \over (10^6\lambda)^2}+{0.

assume here that the atmosphere is homogeneous, we just use the dilute altitude that can be obtained for any real atmosphere precisely from the pressure and density at perceiver. Assuming furthermore ideal gas law for dry breeze we have: $ $ \label{eq: ell } \ell(T)={P \over { \rho \ g}}={P_0 \over { \rho_0 \ g_0}}{T \over T_0},$$ where $ { \rho_0=1.293\ \mbox{kg}\,\mbox{m}^{-3}}$ for $ { T_0=273.15\,\mbox{K}}$, $ { P_0=101325\,\mbox{Pa}}$ and normal gravity $ { g_0=9.80665\,\mbox{m}\,\mbox{s}^{-2}}$. The spoke of curvature for Calern observatory ($ { \varphi=43\degr45\arcmin7\arcsec}$, $ { h=1323\,\mbox{m}}$) was approximated by the minimum reference ellipsoid curvature at latitude $ 45\degr$ and sea horizontal surface (Chollet [ @Chollet ], see Appendix \[App: curvature\ ] ): $ $ r_c(45\degr,0)=6367.512\ \mathrm{km}$$ Ambient tune refractivity was deduced from the refractive exponent $ n_0(\lambda)$ under standard conditions and the partial pressure of water system vapor $ p$ by applying the recipe recommended by the inaugural resolution of the 13th General Assembly of the International Union of Geodesy and Geophysics (IUGG [ @IUGG63 ]; Baldini [ @Baldini ]). After conversion to Pa (Pascal) as the pressure unit, the equation become: $ $ \label{eq: alpha } \alpha(T, P, f_h,\lambda)={T_0 \over T } \left\{(n_0(\lambda)-1) { P \over P_0 } - 4.13\, 10^{-10}\ p(f_h, T)\right\}$$ Refractivity under standard condition (sea level, $ { T = T_0}$, $ { P = P_0}$, $ 0\%$ humidity, $ 0.03\%$ of carbon dioxide) was taken from the work of Barrel and Sears ([ @Barrel ] ): $ $ \label{eq: Barrel } n_0(\lambda)-1=\left\{2876.04+{16.288 \over (10 ^ 6\lambda)^2}+{0.

asdume here that the atmoskhere is homogeneous, we oust uss the reauced height that can be obteinee for any real atmosphere jjst from nhe pressyre end density at ouaerver. Assumihn furchxrmore ideal gax law for gry air we hava: $$\lxbzl{eq:ell} \ell(T)={P \over {\rho \ g}}={P_0 \over {\rho_0 \ d_0}}{T \over T_0},$$ ahere ${\rho_0=1.293\ \mbox{hg}\,\mbpv{m}^{-3}}$ fkg ${U_0=273.15\,\mbox{K}}$, ${P_0=101325\,\mbox{Pa}}$ and normal gravity ${g_0=9.80665\,\mbox{m}\,\kbox{s}^{-2}}$. The radous of curvature for Calerj obdervatory (${\varphi=43\dehr45\arcmin7\arcwec}$, ${r=1323\,\nbox{m}}$) was aporoximated by the minijum reference ellipsoid curvatufe at latitude $45\eetr$ wtd sea leveo (Chojlet [@Chollet], see Appandix \[Alp:curvature\]): $$r_c(45\cegc,0)=6367.512\ \marhrm{km}$$ Ambient air rehractivity was deducgd from tha xefractive index $n_0(\lamvdq)$ undgr stdndafe cundjtmona and hhe partial pdessure of qater vapor $p$ by apklyybg the formulz recoimqnded by the first resolution of the 13th Geheral Assembly of the Ibternational Union of Geodesy wnd Geophysics (IUGG [@IUGG63]; Baldini [@Baldini]). After cotversmov ti Ia (Pxwcwl) as the pressure unit, the equation becomes: $$\jzbtl{ea:alpha} \alpha(T,P,f_m,\lambda)={T_0 \over T} \legt\{(j_0(\lsibda)-1) {P \over P_0} - 4.13\, 10^{-10}\ p(f_k,F)\rjght\}$$ Refractivity knder sjandare conditijn (sra level, ${T=T_0}$, ${P=P_0}$, $0\%$ humidity, $0.03\%$ od carbon dioqide) was taken from thz work of Baxrel amd Sesrs ([@Barrel]): $$\label{eq:Barreu} n_0(\lajbda)-1=\left\{2876.04+{16.288 \ovfr (10^6\lambda)^2}+{0.

assume here that the atmosphere is homogeneous, use reduced height can be obtained from pressure and density observer. Assuming furthermore gas law for dry air we $$\label{eq:ell} \ell(T)={P \over {\rho \ g}}={P_0 \over {\rho_0 \ g_0}}{T \over T_0},$$ where \mbox{kg}\,\mbox{m}^{-3}}$ for ${T_0=273.15\,\mbox{K}}$, ${P_0=101325\,\mbox{Pa}}$ and normal gravity ${g_0=9.80665\,\mbox{m}\,\mbox{s}^{-2}}$. The radius of curvature for observatory ${h=1323\,\mbox{m}}$) approximated the minimum reference ellipsoid curvature at latitude $45\degr$ and sea level (Chollet [@Chollet], see Appendix \[App:curvature\]): \mathrm{km}$$ Ambient air refractivity was deduced from the index $n_0(\lambda)$ under standard and the partial pressure of vapor by applying formula by first resolution of 13th General Assembly of the International Union of Geodesy and Geophysics (IUGG [@IUGG63]; Baldini [@Baldini]). After conversion Pa (Pascal) pressure unit, equation $$\label{eq:alpha} \over T} \left\{(n_0(\lambda)-1) P_0} - 4.13\, 10^{-10}\ p(f_h,T)\right\}$$ Refractivity (sea level, ${T=T_0}$, ${P=P_0}$, $0\%$ humidity, $0.03\%$ of dioxide) was from the work of Barrel and ([@Barrel]): $$\label{eq:Barrel} n_0(\lambda)-1=\left\{2876.04+{16.288 \over (10^6\lambda)^2}+{0.

assume here that the atmospheRe is homogeNeous, We jUst UsE the ReduCed height that cAN be oBtained for any real atmosPhere JuST froM ThE presSure and DEnSITy aT oBsErvEr. aSsUming FurThermorE ideal gas lAw fOr Dry air we have: $$\LAbEl{eq:ell} \ell(t)={P \oVer {\rho \ g}}={P_0 \over {\Rho_0 \ G_0}}{T \over t_0},$$ wHerE ${\Rho_0=1.293\ \mbOx{kG}\,\mbox{M}^{-3}}$ for ${T_0=273.15\,\mBOx{K}}$, ${P_0=101325\,\mbOx{Pa}}$ and noRmAL graviTY ${g_0=9.80665\,\mbox{m}\,\MBOx{S}^{-2}}$. The Radius of curvature FOr cAlern observatoRy (${\varpHi=43\DEgR45\ARcmIn7\aRcsec}$, ${h=1323\,\mbox{M}}$) wAs appROximateD By THE MinIMum reference eLlipsoid curVAtuRe at laTiTudE $45\Degr$ anD sea lEvEL (ChOllet [@CholleT], see appendix \[APp:curvATure\]): $$r_c(45\dEGr,0)=6367.512\ \mathrM{km}$$ AmbIenT aiR refRAcTiVitY wAS deDUcEd fROm tHe refracTiVe Index $N_0(\lamBDA)$ UNder StaNdarD condItions and the pArtIal pREssUre of Water VapoR $p$ By appLying tHe forMuLa recommended by The fIrst resolUtiOn Of tHe 13Th GenERal AssEmbLy oF the IntErnatioNAl UNiON OF GEodesy and GeophysicS (Iugg [@IuGG63]; BaldiNi [@BaldINi]). afTEr converSiOn tO Pa (PASCal) as The pREsSure unit, The equATiOn Becomes: $$\LaBel{eq:aLpHa} \aLphA(T,P,f_h,\LAmbdA)={T_0 \over t} \left\{(n_0(\laMbda)-1) {P \OVer P_0} - 4.13\, 10^{-10}\ p(f_h,T)\right\}$$ REFractivity undER sTANdARd coNdiTion (sea leveL, ${T=T_0}$, ${P=p_0}$, $0\%$ HumiDity, $0.03\%$ OF cArbON dioxIde) waS tAKeN From the work of Barrel AnD Sears ([@barreL]): $$\label{eq:BarreL} n_0(\lambda)-1=\leFT\{2876.04+{16.288 \OVer (10^6\lambdA)^2}+{0.

assume here that the atmo sphere ishomog ene ous ,we j ustuse the reduce d hei ght that can be obtain ed fo ra ny r e al atmo spherej us t fro mth e p re s su re an d d ensityat observe r.As suming furth e rm ore idealgas law for dry ai r we h av e:$$\la bel {eq:e ll} \e l l(T)={ P \over { \r h o \ g} } ={P_0 \ o v er {\r ho_0 \ g_0}}{T \o v er T_0},$$ where ${\rh o_ 0 =1 . 2 93\ \m box{kg}\,\ mb ox{m} ^ {-3}}$f or $ { T_0 = 273.15\,\mbox {K}}$, ${P_ 0 =10 1325\, \m box { Pa}}$and n or m algravity ${g _0=9 .80665\,\ mbox{m } \,\mbox { s}^{-2} }$. Th e r adi us o f c ur vat ur e fo r C ale r n o bservato ry ( ${\va rphi = 4 3 \ degr 45\ arcm in7\a rcsec}$, ${h= 132 3\,\ m box {m}}$ ) was app ro ximat ed bythe m in imum referenceelli psoid cur vat ur e a tlatit u de $45 \de gr$ and se a level (Ch ol l e t [ @Chollet], see App en d i x\[App:cu rvatur e \] ): $$r_c(45 \d egr ,0)= 6 3 67.51 2\ \ m at hrm{km}$ $ Ambi e nt a ir refr ac tivity w asded ucedf romthe re fractive inde x $n_0(\lambda) $ under standa r dc o nd i tion s a nd the part ialp ress ureo fwat e r vap or $p $b ya pplying the formula r ecomme ndedby the firstresolution o f the 13t h Ge n er a l Assembly ofthe I nternation a l Unionof Ge odesy an d Geophys i c s (IUGG[@I UGG 63] ; B a l di ni [@Baldini] ) . Aft er conver sio n to Pa (P asc al) as t he pressu re unit, t he e qu ati on be c omes: $$ \l abe l{ eq: alpha } \alph a(T,P ,f_h ,\ la m bda )={T_0\ ov e r T}\l ef t\{( n_0 (\ lambd a)-1 ) {P \overP_0} - 4. 13\ , 10^ {- 10 }\ p(f_ h,T)\right\}$ $Refractivi ty un der st a n dard con dition (sea level, ${T= T _0}$, $ {P= P_0}$ , $0 \%$ humid ity , $0.0 3\% $ of ca rbon d ioxid e) wa s taken f ro m t he work of B a r rel andSe ars([@Barr el]): $$\label{eq: B arr el} n_0(\lamb da) -1=\ l e ft \{2 8 76 . 04+ {1 6 .28 8 \over (10^6\lam bda)^2}+{0 .

assume_here that_the atmosphere is homogeneous,_we just_use_the reduced_height_that can be_obtained for any_real atmosphere just from_the pressure and_density_at observer. Assuming furthermore ideal gas law for dry air we have: $$\label{eq:ell} \ell(T)={P \over {\rho_\_g}}={P_0 \over_{\rho_0_\_g_0}}{T \over T_0},$$ where ${\rho_0=1.293\ \mbox{kg}\,\mbox{m}^{-3}}$_for ${T_0=273.15\,\mbox{K}}$, ${P_0=101325\,\mbox{Pa}}$ and normal_gravity ${g_0=9.80665\,\mbox{m}\,\mbox{s}^{-2}}$._The radius of curvature for Calern observatory (${\varphi=43\degr45\arcmin7\arcsec}$,_${h=1323\,\mbox{m}}$)_was approximated by_the minimum reference ellipsoid curvature at latitude $45\degr$ and_sea level (Chollet [@Chollet], see Appendix_\[App:curvature\]): $$r_c(45\degr,0)=6367.512\ \mathrm{km}$$_Ambient_air_refractivity was deduced from_the refractive index $n_0(\lambda)$ under standard_conditions and the partial pressure of_water vapor $p$ by applying the formula_recommended by the first resolution of_the 13th General Assembly of_the International_Union of Geodesy and Geophysics_(IUGG [@IUGG63]; Baldini_[@Baldini]). After_conversion to Pa_(Pascal) as the pressure unit, the_equation becomes: $$\label{eq:alpha} \alpha(T,P,f_h,\lambda)={T_0_\over T} \left\{(n_0(\lambda)-1) {P \over P_0}_-_4.13\, 10^{-10}\ p(f_h,T)\right\}$$_Refractivity_under_standard condition_(sea level, ${T=T_0}$,_${P=P_0}$,_$0\%$ humidity,_$0.03\%$_of carbon dioxide) was taken from_the_work of Barrel and Sears ([@Barrel]): $$\label{eq:Barrel} n_0(\lambda)-1=\left\{2876.04+{16.288_\over (10^6\lambda)^2}+{0.

between open subsets of $D$. Then we have a diagram $$\xymatrix{ J^r_{{\mathrm{imm}}}(V,D) \ar[r]^{h_*} \ar[d]_{\cong}^{k_*} & J^r(V,P) \ar[d]_{\cong}^{k_*} & V[B] \ar@{d->}[l] \ar[d]_{\cong}^{k_*}\\ J^r_{{\mathrm{imm}}}(W,D) \ar[r]^{h_*} & J^r(W,P) & W[B] \ar@{d->}[l] }$$ Now $j^r_x(k)$ in the top left corner is mapped by $k_*$ down to $j^r_y({\mathrm{id}})$. Hence we see that it is enough (equivalent) to check the transversality only at $j^r_y({\mathrm{id}})$’s for all $y\in D$ for which $h_*(j^r_y({\mathrm{id}}))=j^r_y(h)\in D[B]$. For such $y$ the same diagram shows that every $j^r_x(k)$ with target $y$ is mapped by $h_*$ to $D[B]$. Thus the whole fibre over $y$ of the target map $$J^r_{\mathrm{imm}}(D,D){\xrightarrow{\ \tau\ }} D$$ is mapped to $D[B]$. The target map $\tau$ has a section $$j^r({\mathrm{id}}):D\rightarrow J^r_{\mathrm{imm}}(D,D)$$ and so (i) is finally equivalent to the composite $$D{\xrightarrow{\ j^r({\mathrm{id}})\ }}J^r_{\mathrm{imm}}(D,D){\xrightarrow{\ h_*\ }}J^r(D,P)$$ being transverse to $D[B]$. This is (ii). We say that a map $g:M\rightarrow P$ is transverse to a ${\mathrm{Diff}}({\mathbb R}^d,0)$-invariant submanifold $B\subseteq J^r_0({\mathbb R}^d,P)$, denoted $g\pitchfork B$, if $$g_*:J^r_{0,{\mathrm{imm}}}({\mathbb R}^d,M)\rightarrow J^r_0({\mathbb R}^d,P)$$

between open subsets of $ D$. Then we have a diagram $ $ \xymatrix { J^r_{{\mathrm{imm}}}(V, D) \ar[r]^{h _ * } \ar[d]_{\cong}^{k _ * } & J^r(V, P) \ar[d]_{\cong}^{k _ * } & V[B ] \ar@{d->}[l ] \ar[d]_{\cong}^{k_*}\\ J^r_{{\mathrm{imm}}}(W, D) \ar[r]^{h _ * } & J^r(W, P) & W[B ] \ar@{d->}[l ] } $ $ Now $ j^r_x(k)$ in the top leftover recess is map by $ k_*$ down to $ j^r_y({\mathrm{id}})$. Hence we see that it is adequate (equivalent) to check the transversality only at $ j^r_y({\mathrm{id}})$ ’s for all $ y\in D$ for which $ h_*(j^r_y({\mathrm{id}}))=j^r_y(h)\in D[B]$. For such $ y$ the same diagram picture that every $ j^r_x(k)$ with target $ y$ is mapped by $ h_*$ to $ D[B]$. therefore the solid fibre over $ y$ of the aim map $ $ J^r_{\mathrm{imm}}(D, D){\xrightarrow{\ \tau\ } } D$$ is mapped to $ D[B]$. The target function $ \tau$ has a section $ $ j^r({\mathrm{id}}):D\rightarrow J^r_{\mathrm{imm}}(D, D)$$ and so (i) is finally equivalent to the composite $ $ D{\xrightarrow{\ j^r({\mathrm{id}})\ } } J^r_{\mathrm{imm}}(D, D){\xrightarrow{\ h_*\ } } J^r(D, P)$$ being transverse to $ D[B]$. This is (ii). We say that a function $ g: M\rightarrow P$ is transverse to a $ { \mathrm{Diff}}({\mathbb R}^d,0)$-invariant submanifold $ B\subseteq J^r_0({\mathbb R}^d, P)$, denoted $ g\pitchfork B$, if $ $ g_*:J^r_{0,{\mathrm{imm}}}({\mathbb R}^d, M)\rightarrow J^r_0({\mathbb R}^d, P)$$

behween open subsets of $D$. Uhen we have a dictram $$\xbmatrix{ N^r_{{\mathrm{kmm}}}(V,D) \ar[r]^{h_*} \ar[d]_{\cong}^{k_*} & J^r(V,P) \ar[v]_{\cont}^{k_*} & V[V] \ar@{d->}[l] \ar[d]_{\cong}^{k_*}\\ J^r_{{\mathro{imm}}}(W,D) \ar[g]^{h_*} & J^r(W,P) & W[B] \er@{d->}[l] }$$ Now $j^r_x(k)$ in the top left dlrnex ms mapped by $k_*$ cown to $j^r_f({\mathrm{id}})$. Henca de see that it is enough (equivalent) tj check tje transversalyty pgly zn $m^r_y({\mathrm{id}})$’s for all $y\in D$ for wgich $h_*(j^g_y({\mathrm{id}}))=j^r_y(h)\in C[B]$. For such $y$ the same diahram shows that every $u^r_x(k)$ with tqrgee $y$ is mapped by $h_*$ to $D[B]$. Thus the wgole fibre over $y$ of the target map $$L^r_{\mathrm{imm}}(E,D){\zrihvtarrow{\ \tau\ }} D$$ if mapped to $D[B]$. The tdrget msp $\tau$ has a scctioi $$j^r({\nathrm{id}}):D\rightarrow J^c_{\mathrm{imm}}(D,D)$$ and so (y) is finanlv equivalent to the cimpositg $$D{\xrhghtxerod{\ j^d({\methdm{id}})\ }}J^g_{\mavhrm{imm}}(D,D){\xrjghtarrow{\ h_*\ }}J^r(D,P)$$ being transvetsq to $D[B]$. This ia (ii). We swy that a map $g:M\rightarrow P$ is transvegse fo a ${\mathrm{Diff}}({\mathbb R}^e,0)$-invariant submanifolf $B\subsetqq J^r_0({\mathbb R}^d,P)$, denoted $g\pitchfork B$, if $$g_*:J^r_{0,{\mathrm{hmm}}}({\mavhcb X}^e,M)\rigfragrow J^r_0({\mathbb R}^d,P)$$

between open subsets of $D$. Then we diagram J^r_{{\mathrm{imm}}}(V,D) \ar[r]^{h_*} & J^r(V,P) \ar[d]_{\cong}^{k_*} \ar[r]^{h_*} J^r(W,P) & W[B] }$$ Now $j^r_x(k)$ the top left corner is mapped $k_*$ down to $j^r_y({\mathrm{id}})$. Hence we see that it is enough (equivalent) to the transversality only at $j^r_y({\mathrm{id}})$’s for all $y\in D$ for which $h_*(j^r_y({\mathrm{id}}))=j^r_y(h)\in D[B]$. such the diagram that every $j^r_x(k)$ with target $y$ is mapped by $h_*$ to $D[B]$. Thus the whole fibre $y$ of the target map $$J^r_{\mathrm{imm}}(D,D){\xrightarrow{\ \tau\ }} is mapped to $D[B]$. target map $\tau$ has a $$j^r({\mathrm{id}}):D\rightarrow and so is equivalent the composite $$D{\xrightarrow{\ }}J^r_{\mathrm{imm}}(D,D){\xrightarrow{\ h_*\ }}J^r(D,P)$$ being transverse to $D[B]$. This is (ii). We say that a map $g:M\rightarrow P$ transverse to R}^d,0)$-invariant submanifold J^r_0({\mathbb denoted B$, if $$g_*:J^r_{0,{\mathrm{imm}}}({\mathbb R}^d,P)$$

between open subsets of $D$. Then We have a diaGram $$\xYmaTriX{ J^R_{{\matHrm{iMm}}}(V,D) \ar[r]^{h_*} \ar[d]_{\conG}^{K_*} & J^r(V,p) \ar[d]_{\cong}^{k_*} & V[B] \ar@{d->}[l] \ar[d]_{\cong}^{K_*}\\ J^r_{{\maThRM{imm}}}(w,d) \aR[r]^{h_*} & J^r(w,P) & W[B] \ar@{d->}[L] }$$ noW $J^R_x(k)$ In ThE toP lEFt CorneR is Mapped bY $k_*$ down to $j^r_Y({\maThRm{id}})$. Hence we sEE tHat it is enoUgh (Equivalent) to CheCk the tRaNsvERsaliTy oNly at $J^r_y({\matHRm{id}})$’s fOr all $y\in D$ FoR Which $h_*(J^R_y({\mathrM{ID}}))=j^R_y(h)\iN D[B]$. For such $y$ the samE DiAGram shows that eVery $j^r_X(k)$ WItH TArgEt $y$ Is mapped by $H_*$ tO $D[B]$. ThUS the whoLE fIBRE ovER $y$ of the target Map $$J^r_{\mathrm{IMm}}(D,d){\xrighTaRroW{\ \Tau\ }} D$$ is MappeD tO $d[B]$. THe target map $\Tau$ hAs a sectioN $$j^r({\matHRm{id}}):D\riGHtarrow j^r_{\mathRm{iMm}}(D,d)$$ and SO (i) Is FinAlLY eqUIvAleNT to The compoSiTe $$d{\xrigHtarROW{\ J^R({\matHrm{Id}})\ }}J^r_{\MathrM{imm}}(D,D){\xrightaRroW{\ h_*\ }}J^r(d,p)$$ beIng trAnsveRse tO $D[b]$. This Is (ii). We Say thAt A map $g:M\rightarroW P$ is TransversE to A ${\mAthRm{diff}}({\mAThbb R}^d,0)$-InvAriAnt submAnifold $b\SubSeTEQ j^r_0({\Mathbb R}^d,P)$, denoted $g\pItCHFoRk B$, if $$g_*:J^r_{0,{\Mathrm{IMm}}}({\MaTHbb R}^d,M)\riGhTarRow J^R_0({\MAthbb r}^d,P)$$

between open subsets of $ D$. Then w e hav e a di ag ram$$\x ymatrix{ J^r_{ { \mat hrm{imm}}}(V,D) \ar[r] ^{h_* }\ ar[d ] _{ \cong }^{k_*} &J ^ r(V ,P )\ar [d ] _{ \cong }^{ k_*} &V[B] \ar@{ d-> }[ l] \ar[d]_{\ c on g}^{k_*}\\ J^ r_{{\mathrm{ imm }}}(W, D) \a r [r]^{ h_* } & J ^r(W,P ) & W[B ] \ar@{d- >} [ l] }$$ Now $j^ r _ x( k)$in the top left c o rn e r is mapped by $k_*$ d o wn t o $ j^r _y({\mathr m{ id}}) $ . Hence we s e e t h at it is enou gh (equival e nt) to ch ec k t h e tran svers al i tyonly at $j^ r_y( {\mathrm{ id}})$ ’ s for a l l $y\in D$ fo r w hic h $h _ *( j^ r_y ({ \ mat h rm {id } })) =j^r_y(h )\ in D[B] $. F o r s uch$y$ the same diagram show s t hate ver y $j^ r_x(k )$ w it h tar get $y $ isma pped by $h_*$ t o $D [B]$. Thu s t he wh ol e fib r e over $y $ o f the t arget m a p $ $J ^ r _ {\ mathrm{imm}}(D,D){ \x r i gh tarrow{\ \tau\ }} D $ $ is map pe d t o $D [ B ]$. T he t a rg et map $ \tau$h as a sectio n$$j^r( {\ mat hrm {id}} ) :D\r ightar row J^r_ {\mat h rm{imm}}(D,D)$ $ and so (i) i s f i n al l y eq uiv alent to th e co m posi te $ $ D{ \xr i ghtar row{\ j ^ r( { \mathrm{id}})\ }}J^ r_ {\math rm{im m}}(D,D){\xri ghtarrow{\ h _ *\ }}J^r (D,P ) $$ being transver se to $D[B]$. T h is is (i i). We say t hat a map $ g:M\righ tar row P$ is t ra nsverse to a$ { \mat hr m{Diff} }({ \mathbb R} ^d, 0)$ -in va riant sub manifold $ B\ su bs ete q J^r _ 0({\math bb R} ^d ,P) $, de n oted $ g\pit chfo rk B $ , i f $$g_* : J^ r _ {0,{ \m at hrm{ imm }} }({\m athb b R} ^d,M)\r ightarrow J^ r _0({ \m at hbb R}^ d,P)$$

between_open subsets_of $D$. Then we_have a_diagram_$$\xymatrix{ J^r_{{\mathrm{imm}}}(V,D) \ar[r]^{h_*}_\ar[d]_{\cong}^{k_*}_& J^r(V,P) \ar[d]_{\cong}^{k_*}_& V[B] \ar@{d->}[l]_\ar[d]_{\cong}^{k_*}\\ J^r_{{\mathrm{imm}}}(W,D) \ar[r]^{h_*} & J^r(W,P)_& W[B] \ar@{d->}[l] }$$_Now_$j^r_x(k)$ in the top left corner is mapped by $k_*$ down to $j^r_y({\mathrm{id}})$. Hence_we_see that_it_is_enough (equivalent) to check the_transversality only at $j^r_y({\mathrm{id}})$’s for_all $y\in_D$ for which $h_*(j^r_y({\mathrm{id}}))=j^r_y(h)\in D[B]$. For such $y$_the_same diagram shows_that every $j^r_x(k)$ with target $y$ is mapped by_$h_*$ to $D[B]$. Thus the whole_fibre over $y$_of_the_target map $$J^r_{\mathrm{imm}}(D,D){\xrightarrow{\ \tau\_}} D$$ is mapped to $D[B]$._The target map $\tau$ has a_section $$j^r({\mathrm{id}}):D\rightarrow J^r_{\mathrm{imm}}(D,D)$$ and so (i) is_finally equivalent to the composite $$D{\xrightarrow{\_j^r({\mathrm{id}})\ }}J^r_{\mathrm{imm}}(D,D){\xrightarrow{\ h_*\ }}J^r(D,P)$$ being_transverse to_$D[B]$. This is (ii). We say_that a map_$g:M\rightarrow P$_is transverse to_a ${\mathrm{Diff}}({\mathbb R}^d,0)$-invariant submanifold $B\subseteq J^r_0({\mathbb_R}^d,P)$, denoted $g\pitchfork_B$, if $$g_*:J^r_{0,{\mathrm{imm}}}({\mathbb R}^d,M)\rightarrow J^r_0({\mathbb R}^d,P)$$

$${}^{\vee}{G}(\lambda')={}^{\vee}{G}(\lambda)\quad {L}(\lambda')={L}(\lambda)\quad {N}(\lambda')={N}(\lambda).$$ Therefore, we can respectively define ${}^{\vee}{G}(\Lambda)={}^{\vee}{G}(\lambda),~{L}(\Lambda)={L}(\lambda),~N(\Lambda)={N}(\lambda)$ and $P(\Lambda)=P(\lambda)$. Now, for every ${}^{\vee}{G}$-orbit $\mathcal{O}$ of semi-simple elements in $\mathfrak{g}$ write $$X\left(\mathcal{O},{}^{\vee}{G}^{\Gamma}\right)=\left\{(y,\Lambda)\in X\left({}^{\vee}{G}^{\Gamma}\right):\Lambda\subset \mathcal{O}\right\}$$ and set $$\begin{aligned} \mathcal{F}(\mathcal{O})&=\text{set of canonical flats in }\mathcal{O}.\end{aligned}$$ Fix $\Lambda\in\mathcal{F}(\mathcal{O})$ and consider the sets $$\begin{aligned} \mathcal{C}(\mathcal{O})=\{ge(\Lambda)g^{-1}:g\in {}^{\vee}{G}\}\quad\text{ and }\quad \mathcal{I}(\mathcal{O})=\{y\in {}^{\vee}{G}^{\Gamma}-{}^{\vee}{G}:y^{2}\in\mathcal{C}(\mathcal{O})\}.\end{aligned}$$ From Proposition 6.13 [@ABV] the set $\mathcal{I}(\mathcal{O})$ decomposes into a finite number of ${}^{\vee}{G}$-orbits. List the orbits as $\mathcal{I}_{1}(\mathcal{O}),\cdots, \mathcal{I}_{r}(\mathcal{O})$ and for each $1\leq i\leq r$ choose a point $$y_{i}\in \mathcal{I}_{i}(\mathcal{O})\quad\text{with}\quad y_{i}^{2}=e(\Lambda).$$ Then conjugation by $y_{i}$ defines an involutive automorphism of ${}^{\vee}{G}(\Lambda)$ with fixed point set denoted by $K_{y_i}$. It is immediate from the definition of $X\left(\mathcal{O},{}^{\vee}{G}^{\Gamma}\right)$ that $$\begin{aligned} X\left(\mathcal{O},{}^{\vee}{G}^{\Gamma}\right)=\mathcal

$ $ { } ^{\vee}{G}(\lambda')={}^{\vee}{G}(\lambda)\quad { L}(\lambda')={L}(\lambda)\quad { N}(\lambda')={N}(\lambda).$$ Therefore, we can respectively define $ { } ^{\vee}{G}(\Lambda)={}^{\vee}{G}(\lambda),~{L}(\Lambda)={L}(\lambda),~N(\Lambda)={N}(\lambda)$ and $ P(\Lambda)=P(\lambda)$. Now, for every $ { } ^{\vee}{G}$-orbit $ \mathcal{O}$ of semi - simple chemical element in $ \mathfrak{g}$ spell $ $ X\left(\mathcal{O},{}^{\vee}{G}^{\Gamma}\right)=\left\{(y,\Lambda)\in X\left({}^{\vee}{G}^{\Gamma}\right):\Lambda\subset \mathcal{O}\right\}$$ and set $ $ \begin{aligned } \mathcal{F}(\mathcal{O})&=\text{set of canonical flats in } \mathcal{O}.\end{aligned}$$ Fix $ \Lambda\in\mathcal{F}(\mathcal{O})$ and study the sets $ $ \begin{aligned } \mathcal{C}(\mathcal{O})=\{ge(\Lambda)g^{-1}:g\in { } ^{\vee}{G}\}\quad\text { and } \quad \mathcal{I}(\mathcal{O})=\{y\in { } ^{\vee}{G}^{\Gamma}-{}^{\vee}{G}:y^{2}\in\mathcal{C}(\mathcal{O})\}.\end{aligned}$$ From Proposition 6.13 [ @ABV ] the set $ \mathcal{I}(\mathcal{O})$ decomposes into a finite issue of $ { } ^{\vee}{G}$-orbits. number the orbits as $ \mathcal{I}_{1}(\mathcal{O}),\cdots, \mathcal{I}_{r}(\mathcal{O})$ and for each $ 1\leq i\leq r$ choose a point $ $ y_{i}\in \mathcal{I}_{i}(\mathcal{O})\quad\text{with}\quad y_{i}^{2}=e(\Lambda).$$ Then conjugation by $ y_{i}$ defines an involutive automorphism of $ { } ^{\vee}{G}(\Lambda)$ with fixed compass point set denoted by $ K_{y_i}$. It is contiguous from the definition of $ X\left(\mathcal{O},{}^{\vee}{G}^{\Gamma}\right)$ that $ $ \begin{aligned } X\left(\mathcal{O},{}^{\vee}{G}^{\Gamma}\right)=\mathcal

$${}^{\vef}{G}(\lambda')={}^{\vee}{G}(\lambda)\quad {L}(\uambda')={L}(\lambda)\quce {N}(\lamuda')={N}(\lamgda).$$ Therdfore, we can respectively dehine ${}^{\vee}{G}(\Oambda)={}^{\vee}{G}(\lambda),~{L}(\Lambdx)={L}(\lambda),~N(\Pambda)={N}(\lqmbde)$ and $P(\Lambda)=P(\lambda)$. Now, for ebcry ${}^{\vze}{J}$-orbit $\mathcal{O}$ of semi-sikple elements hn $\mcthfrak{g}$ write $$X\left(\mathcal{O},{}^{\vee}{G}^{\Gammw}\right)=\lrfh\{(y,\Lambda)\in X\lest({}^{\vet}{G}^{\Gwmma}\dpgmt):\Lambda\subset \mathcal{O}\right\}$$ and set $$\bejin{aligned} \mathcsl{F}(\mathcal{O})&=\text{set of canojicap flats in }\mathcal{L}.\end{aligned}$$ Fix $\Oambda\in\mathzal{F}(\mathcal{O})$ and consiser the sets $$\begin{aligned} \mathcau{C}(\matkcal{O})=\{ge(\Lambea)t^{-1}:g\ij {}^{\vee}{G}\}\quad\teet{ and }\quad \mathcal{L}(\kathcan{O})=\{y\in {}^{\vre}{G}^{\Gamma}-{}^{\vee}{G}:y^{2}\ik\mathral{C}(\nathcal{O})\}.\end{aligned}$$ Frmm Proposition 6.13 [@ABD] the set $\mcthcal{I}(\mathcal{O})$ decompowes itto d fivute nujbxr kf ${}^{\vee}{H}$-oruits. List tge orbits aw $\mathcal{I}_{1}(\mathcal{O}),\ccoew, \mathcal{I}_{r}(\matgcal{O})$ wnq for each $1\leq i\leq r$ choose a point $$y_{i}\pn \mzthcal{I}_{i}(\mathcal{O})\quad\texr{with}\quad y_{i}^{2}=e(\Lambda).$$ Tjen conjudation by $y_{i}$ defines an involutive automorphism ox ${}^{\vee}{J}(\Lxmbbq)$ witf flxed point set denoted by $K_{y_i}$. It is immediate rrpm the definitiok of $X\left(\mathcal{O},{}^{\fef}{G}^{\Bwmma}\right)$ thaj $$\begin{cmifned} X\left(\mathcal{O},{}^{\vfe}{G}^{\Gammw}\righr)=\mathcal

$${}^{\vee}{G}(\lambda')={}^{\vee}{G}(\lambda)\quad {L}(\lambda')={L}(\lambda)\quad {N}(\lambda')={N}(\lambda).$$ Therefore, we can respectively and Now, for ${}^{\vee}{G}$-orbit $\mathcal{O}$ of $$X\left(\mathcal{O},{}^{\vee}{G}^{\Gamma}\right)=\left\{(y,\Lambda)\in \mathcal{O}\right\}$$ and set \mathcal{F}(\mathcal{O})&=\text{set of canonical in }\mathcal{O}.\end{aligned}$$ Fix $\Lambda\in\mathcal{F}(\mathcal{O})$ and consider sets $$\begin{aligned} \mathcal{C}(\mathcal{O})=\{ge(\Lambda)g^{-1}:g\in {}^{\vee}{G}\}\quad\text{ and }\quad \mathcal{I}(\mathcal{O})=\{y\in {}^{\vee}{G}^{\Gamma}-{}^{\vee}{G}:y^{2}\in\mathcal{C}(\mathcal{O})\}.\end{aligned}$$ From Proposition 6.13 [@ABV] the $\mathcal{I}(\mathcal{O})$ decomposes into a finite number of ${}^{\vee}{G}$-orbits. List the orbits as $\mathcal{I}_{1}(\mathcal{O}),\cdots, and each i\leq choose a point $$y_{i}\in \mathcal{I}_{i}(\mathcal{O})\quad\text{with}\quad y_{i}^{2}=e(\Lambda).$$ Then conjugation by $y_{i}$ defines an involutive automorphism of ${}^{\vee}{G}(\Lambda)$ fixed point set denoted by $K_{y_i}$. It is from the definition of that $$\begin{aligned} X\left(\mathcal{O},{}^{\vee}{G}^{\Gamma}\right)=\mathcal

$${}^{\vee}{G}(\lambda')={}^{\vee}{G}(\lambda)\quad {L}(\Lambda')={L}(\lamBda)\quAd {N}(\LamBdA')={N}(\laMbda).$$ therefore, we can REspeCtively define ${}^{\vee}{G}(\LambdA)={}^{\vee}{G}(\LaMBda),~{L}(\lAmBda)={L}(\lAmbda),~N(\LAMbDA)={n}(\laMbDa)$ And $p(\LAMbDa)=P(\laMbdA)$. Now, for Every ${}^{\vee}{G}$-oRbiT $\mAthcal{O}$ of semI-SiMple elemenTs iN $\mathfrak{g}$ wrIte $$x\left(\mAtHcaL{o},{}^{\vee}{G}^{\gamMa}\rigHt)=\left\{(Y,\lambda)\In X\left({}^{\veE}{G}^{\gAmma}\riGHt):\LambdA\SUbSet \mAthcal{O}\right\}$$ and seT $$\BeGIn{aligned} \mathcAl{F}(\matHcAL{O})&=\TEXt{sEt oF canonical FlAts in }\MAthcal{O}.\ENd{ALIGneD}$$ fix $\Lambda\in\maThcal{F}(\mathcAL{O})$ aNd consIdEr tHE sets $$\bEgin{aLiGNed} \Mathcal{C}(\matHcal{o})=\{ge(\Lambda)G^{-1}:g\in {}^{\veE}{g}\}\quad\teXT{ and }\quaD \mathcAl{I}(\MatHcal{o})=\{Y\iN {}^{\vEe}{G}^{\gaMMa}-{}^{\vEE}{G}:Y^{2}\in\MAthCal{C}(\mathCaL{O})\}.\End{alIgneD}$$ fROM ProPosItioN 6.13 [@ABV] tHe set $\mathcal{I}(\MatHcal{o})$ DecOmposEs intO a fiNiTe numBer of ${}^{\vEe}{G}$-orBiTs. List the orbits As $\maThcal{I}_{1}(\matHcaL{O}),\CdoTs, \MathcAL{I}_{r}(\matHcaL{O})$ aNd for eaCh $1\leq i\lEQ r$ cHoOSE A pOint $$y_{i}\in \mathcal{I}_{i}(\mAtHCAl{o})\quad\texT{with}\qUAd Y_{i}^{2}=E(\lambda).$$ ThEn ConJugaTIOn by $y_{I}$ defINeS an involUtive aUToMoRphism oF ${}^{\vEe}{G}(\LamBdA)$ wiTh fIxed pOInt sEt denoTed by $K_{y_i}$. it is iMMediate from the DEfinition of $X\lEFt(\MAThCAl{O},{}^{\vEe}{G}^{\gamma}\right)$ tHat $$\bEGin{aLignED} X\LefT(\MathcAl{O},{}^{\veE}{G}^{\gAmMA}\right)=\mathcal

$${}^{\vee}{G}(\lambda')= {}^{\vee}{ G}(\l amb da) \q uad{L}( \lambda')={L}( \ lamb da)\quad {N}(\lambda' )={N} (\ l ambd a ). $$ Th erefore , w e can r es pec ti v el y def ine ${}^{\ vee}{G}(\L amb da )={}^{\vee}{ G }( \lambda),~ {L} (\Lambda)={L }(\ lambda ), ~N( \ Lambd a)= {N}(\ lambda ) $ and$P(\Lambd a) = P(\lam b da)$. N o w ,forevery ${}^{\vee}{ G }$ - orbit $\mathca l{O}$of se m i -si mpl e elements i n $\m a thfrak{ g }$ w r ite $$X\left(\mat hcal{O},{}^ { \ve e}{G}^ {\ Gam m a}\rig ht)=\ le f t\{ (y,\Lambda) \inX\left({} ^{\vee } {G}^{\G a mma}\ri ght):\ Lam bda \sub s et \ mat hc a l{O } \r igh t \}$ $ and se t$$ \begi n{al i g n e d} \ mat hcal {F}(\ mathcal{O})&= \te xt{s e t o f can onica l fl at s in}\math cal{O }. \end{aligned}$$ Fix $\Lambda \in \m ath ca l{F}( \ mathca l{O })$ and co nsidert hese t s $$ \begin{aligned} \m at h c al {C}(\mat hcal{O } )= \{ g e(\Lambd a) g^{ -1}: g \ in {} ^{\v e e} {G}\}\qu ad\tex t {an d }\qua d\mathc al {I} (\m athca l {O}) =\{y\i n {}^{\v ee}{G } ^{\Gamma}-{}^{ \ vee}{G}:y^{2} \ in \ m at h cal{ C}( \mathcal{O} )\}. \ end{ alig n ed }$$ FromPropo si t io n 6.13 [@ABV] the se t$\math cal{I }(\mathcal{O} )$ decompo s e s into afini t en umber of ${}^{ \vee} {G}$-orbit s . List t he or bits as$\mathcal { I }_{1}(\m ath cal {O} ),\ c d ot s, \mathcal{I } _ {r}( \m athcal{ O}) $ and f oreac h $ 1\l eq i\leq r$ chooseapo in t$$y _{i}\ i n \mathc al {I} _{ i}( \math c al{O}) \quad \tex t{ wi t h}\ quad y_ { i} ^ { 2}=e (\ La mbda ).$ $Thenconj u gat ion by$y_{i}$ d efi n es a nin volutiv e automorphis mof ${}^{\v ee }{G }(\Lam b d a)$ with fixed point set denote d by $K_ {y_ i}$.It i s immedia tefrom t hed efinit ion of $X\l ef t(\ m a thcal { O }, {}^ {\ vee}{G}^{\ G a mma }\rig ht )$ t hat $$\ begin{aligned} X\l e ft( \mathcal{O},{ }^{ \vee } { G} ^{\ G am m a}\ ri g ht) = \ mathcal

$${}^{\vee}{G}(\lambda')={}^{\vee}{G}(\lambda)\quad_{L}(\lambda')={L}(\lambda)\quad {N}(\lambda')={N}(\lambda).$$_Therefore, we can respectively_define ${}^{\vee}{G}(\Lambda)={}^{\vee}{G}(\lambda),~{L}(\Lambda)={L}(\lambda),~N(\Lambda)={N}(\lambda)$_and_$P(\Lambda)=P(\lambda)$. Now,_for_every ${}^{\vee}{G}$-orbit $\mathcal{O}$_of semi-simple elements_in $\mathfrak{g}$ write $$X\left(\mathcal{O},{}^{\vee}{G}^{\Gamma}\right)=\left\{(y,\Lambda)\in_X\left({}^{\vee}{G}^{\Gamma}\right):\Lambda\subset \mathcal{O}\right\}$$ and_set_$$\begin{aligned} \mathcal{F}(\mathcal{O})&=\text{set of canonical flats in }\mathcal{O}.\end{aligned}$$ Fix $\Lambda\in\mathcal{F}(\mathcal{O})$ and consider the sets $$\begin{aligned} \mathcal{C}(\mathcal{O})=\{ge(\Lambda)g^{-1}:g\in {}^{\vee}{G}\}\quad\text{_and_}\quad \mathcal{I}(\mathcal{O})=\{y\in {}^{\vee}{G}^{\Gamma}-{}^{\vee}{G}:y^{2}\in\mathcal{C}(\mathcal{O})\}.\end{aligned}$$_From_Proposition_6.13 [@ABV] the set $\mathcal{I}(\mathcal{O})$_decomposes into a finite number_of ${}^{\vee}{G}$-orbits._List the orbits as $\mathcal{I}_{1}(\mathcal{O}),\cdots, \mathcal{I}_{r}(\mathcal{O})$ and for each_$1\leq_i\leq r$ choose_a point $$y_{i}\in \mathcal{I}_{i}(\mathcal{O})\quad\text{with}\quad y_{i}^{2}=e(\Lambda).$$ Then conjugation by $y_{i}$_defines an involutive automorphism of ${}^{\vee}{G}(\Lambda)$_with fixed point_set_denoted_by $K_{y_i}$. It is_immediate from the definition of $X\left(\mathcal{O},{}^{\vee}{G}^{\Gamma}\right)$_that $$\begin{aligned} X\left(\mathcal{O},{}^{\vee}{G}^{\Gamma}\right)=\mathcal

okritos”, GR-15310 Aghia Paraskevi, Attiki, Greece' - 'Institute of Nuclear Research and Nuclear Energy, Bulgarian Academy of Sciences, 72 Tzarigrad Road, 1784 Sofia, Bulgaria' - 'Department of Physics, University of Istanbul, 34134 Istanbul, Turkey' - 'Wright Laboratory, Yale University, New Haven, Connecticut 06520, USA' - 'Facility for Rare Isotope Beams, 640 South Shaw Lane, Michigan State University, East Lansing, MI 48824 USA' - 'Max-Planck-Institut für Kernphysik, Saupfercheckweg 1, D-69117 Heidelberg, Germany' author: - 'S. Sarantopoulou' title: 'Proxy-SU(3) symmetry in heavy nuclei: Prolate dominance and prolate-oblate shape transition' --- [1]{},,,,,,, [1]{} [2]{} [3]{} [4]{} [5]{} [6]{} Using a new approximate analytic parameter-free proxy-SU(3) scheme, simple predictions for the global feature of prolate dominance and for the locus of the prolate-oblate shape transition have been made and compared with empirical data. Emphasis is placed on the mechanism leading to the breaking of the particle-hole symmetry, which is instrumental in shaping up these predictions. It turns out that this mechanism is based on the SU(3) symmetry and the Pauli principle alone, without reference to any specific Hamiltonian. Intoduction =========== The recent introduction of the proxy-SU(3) scheme [@proxy1] has led to parameter independent predictions of the deformation parameters $\beta$ and $\gamma$ [@proxy2], as well as to predictions of the locus of the prolate to oblate shape transition on the nuclear chart and the prolate over oblate dominance in deformed rare earth nuclei [@proxy2]. In Section 2 of the present work we extend the study of the prolate to oblate transition to the actinides and superheavy elements, while in Section 3 we study in detail the particle-hole symmetry breaking, which leads to the prolate over oblate dominance and to the determination of the border of the prolate to oblate transition. Prolate to oblate transition in actinides and super

okritos ”, GR-15310 Aghia Paraskevi, Attiki, Greece' -' Institute of Nuclear Research and Nuclear Energy, Bulgarian Academy of Sciences, 72 Tzarigrad Road, 1784 Sofia, Bulgaria' -' Department of Physics, University of Istanbul, 34134 Istanbul, Turkey' -' Wright Laboratory, Yale University, New Haven, Connecticut 06520, USA' -' Facility for Rare Isotope Beams, 640 South Shaw Lane, Michigan State University, East Lansing, MI 48824 USA' -' Max - Planck - Institut für Kernphysik, Saupfercheckweg 1, D-69117 Heidelberg, Germany' author: -' S. Sarantopoulou' title:' Proxy - SU(3) symmetry in dense core: Prolate dominance and prolate - oblate form transition' --- [ 1 ] { }, ,, ,, ,, [ 1 ] { } [ 2 ] { } [ 3 ] { } [ 4 ] { } [ 5 ] { } [ 6 ] { } Using a modern approximate analytic argument - free proxy - SU(3) scheme, simple predictions for the global feature of speech of prolate dominance and for the locus of the prolate - oblate supreme headquarters allied powers europe transition have been draw and compare with empiric data. Emphasis is placed on the mechanism leave to the breaking of the particle - fix symmetry, which is implemental in shaping up these prediction. It turns out that this mechanism is based on the SU(3) symmetry and the Pauli rationale alone, without reference to any specific Hamiltonian. Intoduction = = = = = = = = = = = The recent introduction of the proxy - SU(3) scheme [ @proxy1 ] has led to parameter autonomous prediction of the deformation parameters $ \beta$ and $ \gamma$ [ @proxy2 ], as well as to predictions of the locus of the prolate to oblate shape transition on the nuclear chart and the prolate over oblate dominance in deformed rare land nuclei [ @proxy2 ]. In Section 2 of the present oeuvre we cover the study of the prolate to oblate transition to the actinides and superheavy element, while in Section 3 we study in detail the particle - fix symmetry breaking, which leads to the prolate over oblate dominance and to the determination of the border of the prolate to oblate conversion. Prolate to oblate transition in actinides and super

okrltos”, GR-15310 Aghia Paraskevi, Xttiki, Greece' - 'Iuwtitutx of Nudlear Rerearch and Nuclear Energy, Bupgqrian Academy of Sciences, 72 Gzarigrad Road, 1784 Sifia, Vulgaria' - 'Dxlartmenb of Lmysicv, University of Istanbul, 34134 Istanbul, Turkay' - 'Wxight Laboratory, Yale University, New Haven, Vojnecticut 06520, USA' - 'Facplyty ror Rare Isotope Beams, 640 South Shaw Lane, Mpchigan State Unifersity, East Lansing, MI 48824 UDA' - 'Mwx-Planck-Institut füg Kernphysij, Satpfercheckweg 1, D-69117 Heidelberg, Germany' zuthor: - 'S. Sarantopoulou' title: 'Proby-SU(3) xymmetry ib yeagi nuclei: Prooate qominance and prolata-oblate shape transitlon' --- [1]{},,,,,,, [1]{} [2]{} [3]{} [4]{} [5]{} [6]{} Usmng q new approximate ananytic parameter-freg proxy-SU(3) vckeme, simple predictiobs for jhe gnobau fextude or prolwte dominance and for thw locus of the prolstq-iblate shape fransieijn have been made and compared with emppriczl data. Emphasis is plaxed on the mechanism peading tj the breaking of the particle-hole symmetry, which is iisgrunektal un shaping up these predictions. It turns out thwf uhix mechanism is based on tne SI(3) symmetry and the Pchlj principle alone, aithout refeeence to wny xpecific Hamiltonian. Intoducrion =========== The receut untroduction of thz proxy-SU(3) sckeme [@ptoxy1] hss led to parameter indzpendeht predictilns of ths deformation parxmeners $\beta$ and $\gamma$ [@proxy2], as wqll as to prebictions of jhe locts of the orolaba to oblate shape hransntion on the nuflear chart and the prolate ovec oblate domimatce in defoxmed rsre earth nucjei [@proxy2]. In Seetion 2 oy the oresent wogk we extxnd the studr of the proldje to oblate vransitiog to the actiniads and superhesvy elemeuus, while ib Section 3 we studn in astail the partieut-hile symmetry bteayind, ahmch lqdds to the psolage ufer oclate domincnze amd to the determinathon kf the border of tne prolate to oblaee transition. Lrolate to oblate uransivion ii actimidgs and super

okritos”, GR-15310 Aghia Paraskevi, Attiki, Greece' - Nuclear and Nuclear Bulgarian Academy of Sofia, - 'Department of University of Istanbul, Istanbul, Turkey' - 'Wright Laboratory, Yale New Haven, Connecticut 06520, USA' - 'Facility for Rare Isotope Beams, 640 South Lane, Michigan State University, East Lansing, MI 48824 USA' - 'Max-Planck-Institut für Kernphysik, 1, Heidelberg, author: 'S. Sarantopoulou' title: 'Proxy-SU(3) symmetry in heavy nuclei: Prolate dominance and prolate-oblate shape transition' --- [1]{},,,,,,, [2]{} [3]{} [4]{} [5]{} [6]{} Using a new analytic parameter-free proxy-SU(3) scheme, predictions for the global feature prolate and for locus the shape transition have made and compared with empirical data. Emphasis is placed on the mechanism leading to the breaking of particle-hole symmetry, instrumental in up predictions. turns out that is based on the SU(3) symmetry principle alone, without reference to any specific Hamiltonian. =========== The introduction of the proxy-SU(3) scheme [@proxy1] led to parameter independent predictions of the deformation $\beta$ and $\gamma$ [@proxy2], as well as to predictions of the locus of the prolate shape transition on the chart and the over dominance deformed earth nuclei In Section 2 of the present work we extend the study the prolate to oblate transition to the actinides and superheavy in 3 we study detail the particle-hole symmetry which to the prolate over and the border the to oblate transition. Prolate oblate transition in actinides and

okritos”, GR-15310 Aghia Paraskevi, AtTiki, Greece' - 'instiTutE of nuCleaR ResEarch and NucleaR enerGy, Bulgarian Academy of ScIenceS, 72 TZArigRAd road, 1784 SOfia, BulGArIA' - 'depArTmEnt Of pHySics, UNivErsity oF Istanbul, 34134 IStaNbUl, Turkey' - 'WrigHT LAboratory, YAle university, NeW HaVen, ConNeCtiCUt 06520, USA' - 'facIlity For RarE isotopE Beams, 640 SouTh sHaw LanE, michigaN sTaTe UnIversity, East LansiNG, Mi 48824 uSA' - 'Max-Planck-InStitut FüR keRNPhySik, saupfercheCkWeg 1, D-69117 HEIdelberG, geRMANy' aUThor: - 'S. SarantopOulou' title: 'PROxy-sU(3) symmEtRy iN Heavy nUclei: prOLatE dominance aNd prOlate-oblaTe shapE TransitIOn' --- [1]{},,,,,,, [1]{} [2]{} [3]{} [4]{} [5]{} [6]{} Using A new apProXimAte aNAlYtIc pArAMetER-fRee PRoxY-SU(3) schemE, sImPle prEdicTIONS for The GlobAl feaTure of prolate DomInanCE anD for tHe locUs of ThE prolAte-oblAte shApE transition have Been Made and coMpaReD wiTh EmpirICal datA. EmPhaSis is plAced on tHE meChANISm Leading to the breakiNg OF ThE particlE-hole sYMmEtRY, which is InStrUmenTAL in shApinG Up These preDictioNS. IT tUrns out ThAt this MeChaNisM is baSEd on The SU(3) sYmmetry aNd the pAuli principle aLOne, without refEReNCE tO Any sPecIfic HamiltoNian. iNtodUctiON =========== THe rECent iNtrodUcTIoN Of the proxy-SU(3) scheme [@pRoXy1] has lEd to pArameter indepEndent predICTIons of thE defORmATion parameters $\Beta$ aNd $\gamma$ [@proXY2], as well aS to prEdictionS of the locUS Of the proLatE to OblAte SHApE transition on THE nucLeAr chart And The prolAte OveR obLatE dOminance iN deformeD rArE eArTh nUclei [@PRoxy2]. In SeCtIon 2 Of The PreseNT work wE exteNd thE sTuDY of The prolATe TO OblaTe TrAnsiTioN tO the aCtinIDes And supeRheavy eleMenTS, whiLe In section 3 We study in detaIl The particlE-hOle SymmetRY Breaking, Which leads to the prolate oVEr oblatE doMinanCe anD to the detErmInatioN of THe bordEr of thE prolAtE to OBLate tRANsItiOn. prolate to oBLAte TransItIon iN actiniDes and super

okritos”, GR-15310 Aghia P araskevi,Attik i,Gre ec e' - 'In stitute of Nuc l earResearch and Nuclear E nergy ,B ulga r ia n Aca demy of Sc i e nce s, 7 2 T za r ig rad R oad , 1784Sofia, Bul gar ia ' - 'Departm e nt of Physic s,University o f I stanbu l, 34 1 34 Is tan bul,Turkey ' - 'Wr ight Labo ra t ory, Y a le Univ e r si ty,New Haven, Connec t ic u t 06520, USA'- 'Fac il i ty f orRar e IsotopeBe ams,6 40 Sout h S h a w La n e, Michigan S tate Univer s ity , East L ans i ng, MI 4882 4U SA' - 'Max-Pla nck- Institutfür Ke r nphysik , Saupfe rcheck weg 1, D-6 9 11 7Hei de l ber g ,Ger m any ' author :-'S. S aran t o p o ulou ' t itle : 'Pr oxy-SU(3) sym met ry i n he avy n uclei : Pr ol ate d ominan ce an dprolate-oblateshap e transit ion '--- [1]{} , ,,,,,, [ 1]{ } [2]{ } [3]{ } [ 4] { } [ 5]{} [6]{} Using a n ew approxi mate a n al yt i c parame te r-f reep r oxy-S U(3) sc heme, si mple p r ed ic tions f or the g lo bal fe ature of p rolate dominan ce an d for the locus of the prolat e -o b l at e sha petransitionhave been mad e a ndc ompar ed wi th em p irical data. Emphas is is pl acedon the mechan ism leadin g t o the br eaki n go f the particle -hole symmetry, which is inst rumental in shapi n g up thes e p red ict ion s . I t turns out t h a t th is mechan ism is bas edonthe SU (3 ) symmetr y and th ePa ul ipri ncipl e alone,wi tho ut re feren c e to a ny sp ecif ic H a mil tonian. I n t oduc ti on === === == === Ther ece nt intr oductionoft he p ro xy -SU(3)scheme [@prox y1 ] has ledto pa ramete r independ ent predictions of thed eformat ion para mete rs $\beta $ a nd $\g amm a $ [@pr oxy2], as w el l a s to pr e d ic tio ns of the lo c u s o f the p rola te to o blate shape transi t ion on the nucle archar t an d t h ep rol at e ov e r oblate dominan ce in defo rm e drare earth nuc le i [@pro xy2]. In Se c tion 2of the pr esent wor kwe e x t end the study of theprolate t o obla t etrans iti on toth e a ctini des an d su perhe avy el em ents,while i n Sectio n 3 we study in detailthe pa rticl e-h ole symme try bre aking, wh ichleads to t hepro lateove r obla te d o mi nan c e and tot he determ i na tio n of the border o f th e pro lat e to ob late transition. Pro l ate to oblatetran s i tio n i n act in ides and super

okritos”, GR-15310_Aghia Paraskevi,_Attiki, Greece' - 'Institute of_Nuclear Research_and_Nuclear Energy,_Bulgarian_Academy of Sciences,_72 Tzarigrad Road,_1784 Sofia, Bulgaria' - 'Department_of Physics, University_of_Istanbul, 34134 Istanbul, Turkey' - 'Wright Laboratory, Yale University, New Haven, Connecticut 06520, USA' - 'Facility_for_Rare Isotope_Beams,_640_South Shaw Lane, Michigan State_University, East Lansing, MI 48824_USA' - 'Max-Planck-Institut_für Kernphysik, Saupfercheckweg 1, D-69117 Heidelberg, Germany' author: - 'S._Sarantopoulou' title:_'Proxy-SU(3) symmetry in_heavy nuclei: Prolate dominance and prolate-oblate shape transition' --- [1]{},,,,,,, [1]{} [2]{} [3]{} [4]{} [5]{} [6]{} Using a_new approximate analytic parameter-free proxy-SU(3) scheme,_simple predictions for_the_global_feature of prolate dominance_and for the locus of the_prolate-oblate shape transition have been made_and compared with empirical data. Emphasis is_placed on the mechanism leading to_the breaking of the particle-hole_symmetry, which_is instrumental in shaping up_these predictions. It_turns out_that this mechanism_is based on the SU(3) symmetry_and the Pauli_principle alone, without reference to any_specific_Hamiltonian. Intoduction =========== The recent introduction_of_the_proxy-SU(3) scheme_[@proxy1] has led_to_parameter independent_predictions_of the deformation parameters $\beta$ and_$\gamma$_[@proxy2], as well as to predictions of_the locus of the_prolate_to oblate shape transition_on the nuclear chart and_the prolate over oblate dominance in_deformed rare_earth nuclei_[@proxy2]. In Section 2 of the present work we extend the study_of the prolate to oblate transition_to the actinides and_superheavy elements,_while_in Section 3_we_study in_detail the particle-hole symmetry breaking, which leads_to the_prolate over oblate dominance and to_the determination of the_border_of the prolate to oblate transition. Prolate_to oblate transition in actinides and_super

. 2. By looking at $DY(0)$, there exists an $(m+2)$-dimensional center manifold ${\mathcal{W}_2^{^C}}$ [@Guckenheimer]. Let $M_0$ be the set of critical points of $Y$, that is [ $$\begin{split} M_0=\left\{ (u,v,w,z)\, | \, \Lambda(u,v,0,z)=0 \right\}. \end{split}$$ ]{} By definition, the manifold $M_0$ is invariant and normally hyperbolic. Now, assume $|w|\ll 1$. This condition appears naturally in our applications. By Fenichel’s theory [@Fenichel] the manifold $M_0$ persists as an invariant normally hyperbolic manifold $M_w$, for sufficiently small $w\neq 0$. We identify $M_w$ with ${\mathcal{W}_2^{^C}}$. In other words, there exists a $C^\ell$ function $m=m(u,v,w)$ such that the center manifold ${\mathcal{W}_2^{^C}}$ is given as a graph [ $$\begin{split} {\mathcal{W}_2^{^C}}=\operatorname{Graph}(u,v,w,m). \end{split}$$ ]{} Define $\zeta=z-m$, then $\zeta'=z'-m'$. But we know, due to invariance of ${\mathcal{W}_2^{^C}}$ under the flow of $Y$, that $\zeta'|_{\zeta=0}=0$. This is, there exists a $C^\ell$ function $H=H(u,v,w,\zeta)$ such that $\zeta'=-H\zeta$. With $H(0)=0$ and ${\dfrac{\partial H}{\partial \zeta}}(0)>0$. In conclusion of this step, there exists a $C^\ell$ transformation $\psi:(u,v,w,z)\mapsto (u,v,w,\zeta)$ that transforms the vector field $Y$ into [ $$\begin{split} \tilde Y:\begin{cases} u' &= \alpha w u + w \bar g\\ v_j' &= \beta_j w v_j \\ w' &= \gamma w^

. 2. By looking at $ DY(0)$, there exists an $ (m+2)$-dimensional center manifold $ { \mathcal{W}_2^{^C}}$ [ @Guckenheimer ]. lease $ M_0 $ be the hardening of critical points of $ Y$, that is [ $ $ \begin{split } M_0=\left\ { (u, v, w, z)\, | \, \Lambda(u, v,0,z)=0 \right\ }. \end{split}$$ ] { } By definition, the manifold $ M_0 $ is invariant and normally hyperbolic. immediately, assume $ |w|\ll 1$. This condition appear naturally in our application. By Fenichel ’s theory [ @Fenichel ] the manifold $ M_0 $ persists as an changeless normally hyperbolic manifold $ M_w$, for sufficiently humble $ w\neq 0$. We identify $ M_w$ with $ { \mathcal{W}_2^{^C}}$. In early words, there exists a $ C^\ell$ function $ m = m(u, five, w)$ such that the center manifold $ { \mathcal{W}_2^{^C}}$ is given as a graph [ $ $ \begin{split } { \mathcal{W}_2^{^C}}=\operatorname{Graph}(u, v, w, m). \end{split}$$ ] { } Define $ \zeta = z - m$, then $ \zeta'=z'-m'$. But we know, due to invariability of $ { \mathcal{W}_2^{^C}}$ under the flow of $ Y$, that $ \zeta'|_{\zeta=0}=0$. This is, there exists a $ C^\ell$ function $ H = H(u, vanadium, w,\zeta)$ such that $ \zeta'=-H\zeta$. With $ H(0)=0 $ and $ { \dfrac{\partial H}{\partial \zeta}}(0)>0$. In conclusion of this step, there exist a $ C^\ell$ transformation $ \psi:(u, v, w, z)\mapsto (u, v, w,\zeta)$ that transform the vector field $ Y$ into [ $ $ \begin{split } \tilde Y:\begin{cases } u' & = \alpha w u + w \bar g\\ v_j' & = \beta_j w v_j \\ w' & = \gamma w^

. 2. Bj looking at $DY(0)$, there exlsts an $(m+2)$-dimensional ceiter mahifold ${\mxthcal{W}_2^{^C}}$ [@Guckenheimer]. Let $M_0$ ue tye seu of critical pointr of $Y$, thwt is [ $$\bejin{split} M_0=\lefb\{ (u,v,w,a)\, | \, \Lcmuda(u,v,0,z)=0 \right\}. \etd{split}$$ ]{} By ddfnnition, the manifold $M_0$ is invariant wnd norkaply hyperbolic. Now, wssujv $|c|\ll 1$. This condition appears nathrally pn our applicatioms. By Fenichel’s theory [@Fenlchep] the manifold $M_0$ pfrsists as qn igcariant normxlly hyperbolic manifomd $M_w$, for sufficiently small $w\ndq 0$. Wz identify $N_w$ wihv ${\mathcal{W}_2^{^C}}$. In onher words, thcge exisds a $C^\ekl$ function $m=m(m,v,w)$ snch rhat the center manifmld ${\mathcal{W}_2^{^C}}$ is gyven as a gxaph [ $$\begin{split} {\mdthcxo{W}_2^{^C}}=\upedavorhame{Grwph}(n,v,w,m). \end{splir}$$ ]{} Define $\zeta=z-m$, trvm $\zeta'=z'-m'$. But we knjw, due to invariance of ${\mathcal{W}_2^{^C}}$ under tve rlow of $Y$, that $\zeta'|_{\zeta=0}=0$. This is, there exists a $C^\ell$ ftnction $H=H(u,v,w,\zeta)$ such that $\zeta'=-H\zeta$. With $H(0)=0$ and ${\gfrac{\'aftico H}{\pafriwl \zeta}}(0)>0$. In conclusion of this step, there evjsus s $C^\ell$ transfovmation $\psi:(u,v,w,z)\mapxtl (i,d,w,\zeta)$ that ttansforms fhe vector field $Y$ into [ $$\fegin{wplit} \yilde Y:\begin{cases} u' &= \alpya w u + w \bag g\\ v_j' &= \beta_j w v_j \\ w' &= \gamoa w^

. 2. By looking at $DY(0)$, there $(m+2)$-dimensional manifold ${\mathcal{W}_2^{^C}}$ Let $M_0$ be of that is [ M_0=\left\{ (u,v,w,z)\, | \Lambda(u,v,0,z)=0 \right\}. \end{split}$$ ]{} By definition, manifold $M_0$ is invariant and normally hyperbolic. Now, assume $|w|\ll 1$. This condition naturally in our applications. By Fenichel’s theory [@Fenichel] the manifold $M_0$ persists as invariant hyperbolic $M_w$, sufficiently small $w\neq 0$. We identify $M_w$ with ${\mathcal{W}_2^{^C}}$. In other words, there exists a $C^\ell$ $m=m(u,v,w)$ such that the center manifold ${\mathcal{W}_2^{^C}}$ is as a graph [ {\mathcal{W}_2^{^C}}=\operatorname{Graph}(u,v,w,m). \end{split}$$ ]{} Define $\zeta=z-m$, $\zeta'=z'-m'$. we know, to of under the flow $Y$, that $\zeta'|_{\zeta=0}=0$. This is, there exists a $C^\ell$ function $H=H(u,v,w,\zeta)$ such that $\zeta'=-H\zeta$. With $H(0)=0$ and H}{\partial \zeta}}(0)>0$. of this there a transformation $\psi:(u,v,w,z)\mapsto (u,v,w,\zeta)$ the vector field $Y$ into [ u' &= \alpha w u + w \bar v_j' &= w v_j \\ w' &= \gamma

. 2. By looking at $DY(0)$, there exists aN $(m+2)$-dimensioNal ceNteR maNiFold ${\MathCal{W}_2^{^C}}$ [@GuckenheiMEr]. LeT $M_0$ be the set of critical poInts oF $Y$, THat iS [ $$\BeGin{spLit} M_0=\lefT\{ (U,v,W,Z)\, | \, \lamBdA(u,V,0,z)=0 \rIgHT\}. \eNd{splIt}$$ ]{} BY definiTion, the manIfoLd $m_0$ is invariant ANd Normally hyPerBolic. Now, assuMe $|w|\Ll 1$. This CoNdiTIon apPeaRs natUrally IN our apPlicationS. BY fenichEL’s theorY [@fEnIcheL] the manifold $M_0$ persIStS As an invariant nOrmallY hYPeRBOliC maNifold $M_w$, foR sUfficIEntly smALl $W\NEQ 0$. We IDentify $M_w$ with ${\Mathcal{W}_2^{^C}}$. In OTheR words, ThEre EXists a $c^\ell$ fUnCTioN $m=m(u,v,w)$ such tHat tHe center mAnifolD ${\Mathcal{w}_2^{^c}}$ is giveN as a grAph [ $$\BegIn{spLIt} {\MaThcAl{w}_2^{^c}}=\opERaTorNAme{graph}(u,v,w,M). \eNd{Split}$$ ]{} defiNE $\ZETa=z-m$, TheN $\zetA'=z'-m'$. BuT we know, due to iNvaRianCE of ${\MathcAl{W}_2^{^C}}$ uNder ThE flow Of $Y$, thaT $\zeta'|_{\ZeTa=0}=0$. This is, there exIsts A $C^\ell$ funcTioN $H=h(u,v,W,\zEta)$ suCH that $\zEta'=-h\zeTa$. With $H(0)=0$ And ${\dfraC{\ParTiAL h}{\PaRtial \zeta}}(0)>0$. In conclusIoN OF tHis step, tHere exIStS a $c^\Ell$ transFoRmaTion $\PSI:(u,v,w,z)\MapsTO (u,V,w,\zeta)$ thAt tranSFoRmS the vecToR field $y$ iNto [ $$\BegIn{splIT} \tilDe Y:\begIn{cases} u' &= \Alpha W U + w \bar g\\ v_j' &= \beta_j w V_J \\ w' &= \gamma w^

. 2. By looking at $DY(0 )$, thereexist s a n $ (m +2)$ -dim ensional cente r man ifold ${\mathcal{W}_2^ {^C}} $[ @Guc k en heime r]. Let $M _ 0 $ b eth e s et of crit ica l point s of $Y$,tha tis [ $$ \ be gin{split} M _0= \left\ {(u, v ,w,z) \,| \,\Lambd a (u,v,0 ,z)=0 \r ig h t\}. \ end{ split}$$ ]{} By definition, th e mani fo l d$ M _0$ is invariant a nd no r mally h y pe r b o lic . Now, assume$|w|\ll 1$. Thi s cond it ion appear s nat ur a lly in our app lica tions. By Fenic h el’s th e ory [@F eniche l]the man i fo ld $M _0 $ pe r si sts asan invar ia nt norm ally h y p erbo lic man ifold $M_w$, for s uff icie n tly smal l $w\ neq0$ . Weidenti fy $M _w $ with ${\mathc al{W }_2^{^C}} $.In ot he r wor d s, the reexi sts a $ C^\ell$ fun ct i o n $ m=m(u,v,w)$ such t ha t th e center manif o ld $ { \mathcal {W }_2 ^{^C } } $ isgive n a s a grap h [$$ \begin{ sp lit} { \ math cal{W} _2^{^C}} =\ope r atorname{Graph } (u,v,w,m). \e nd{ split}$$ ]{ } De fine $\ zet a =z-m$ , the n$ \z e ta'=z'-m'$. But wekn ow, du e toinvariance of ${\mathca l { W }_2^{^C} }$ u n de r the flow of $ Y$, t hat $\zeta ' |_{\zeta =0}=0 $. Thisis, there e xists a$C^ \el l$fun c t io n $H=H(u,v,w, \ z eta) $such th at$\zeta' =-H \ze ta$ . W it h $H(0)=0 $ and ${ \d fr ac {\ par tialH }{\parti al \z et a}} (0)>0 $ . In c oncl us io n of this s t ep , ther eex ists a$C ^\ell $ tr a nsf ormatio n $\psi:( u,v , w,z) \m ap sto (u, v,w,\zeta)$ t ha t transfor ms th e vect o r field $ Y$ into [ $$\begin{spli t } \tild e Y: \begin{ca ses } u'& = \alp ha w u + w\b arg \ \ v _j ' & =\beta_j wv _ j \\ w' & = \gamm a w^

. 2. _By looking_at $DY(0)$, there exists_an $(m+2)$-dimensional_center_manifold ${\mathcal{W}_2^{^C}}$_[@Guckenheimer]._Let $M_0$ be_the set of_critical points of $Y$,_that is __ [ $$\begin{split} _M_0=\left\{_(u,v,w,z)\, |_\,_\Lambda(u,v,0,z)=0_ \right\}. _ _ _\end{split}$$ ]{} By definition, the_manifold_$M_0$ is invariant_and normally hyperbolic. Now, assume $|w|\ll 1$. This condition_appears naturally in our applications. By_Fenichel’s theory [@Fenichel]_the_manifold_$M_0$ persists as an_invariant normally hyperbolic manifold $M_w$, for_sufficiently small $w\neq 0$. We identify_$M_w$ with ${\mathcal{W}_2^{^C}}$. In other words, there_exists a $C^\ell$ function $m=m(u,v,w)$ such_that the center manifold ${\mathcal{W}_2^{^C}}$_is given_as a graph _ [ $$\begin{split} _ _ _ _{\mathcal{W}_2^{^C}}=\operatorname{Graph}(u,v,w,m). _ __ \end{split}$$ ]{} ___ Define_$\zeta=z-m$, then $\zeta'=z'-m'$._But_we know,_due_to invariance of ${\mathcal{W}_2^{^C}}$ under the_flow_of $Y$, that $\zeta'|_{\zeta=0}=0$. This is, there_exists a $C^\ell$ function_$H=H(u,v,w,\zeta)$_such that $\zeta'=-H\zeta$. With_$H(0)=0$ and ${\dfrac{\partial H}{\partial \zeta}}(0)>0$. _ In conclusion of_this step,_there exists_a $C^\ell$ transformation $\psi:(u,v,w,z)\mapsto (u,v,w,\zeta)$ that transforms the vector field $Y$_into [ $$\begin{split} _ _\tilde Y:\begin{cases} __ u'_&=_\alpha w_u + w \bar g\\ _ v_j'_&= \beta_j w v_j \\ _ w'_&=_\gamma w^

called the circular orthogonal ensemble (COE). We have $$\begin{aligned} &\left\langle \prod_{i=1}^L \det(I+\eta_i^{-1} M^{-1}) \cdot \prod_{i=1}^K \det(I+\eta_{L+i} M) \right\rangle_{S(n)} \\ =& \left\langle \prod_{i=1}^L \Psi^{{\mathrm{A}}}(\bz^{-1};\eta_i^{-1}) \cdot \prod_{i=1}^K \Psi^{{\mathrm{A}}}(\bz;\eta_{L+i}) \right\rangle_{n,1} = \prod_{i=1}^L \eta_i^{-n} \cdot P_{(n^L)}^{{\mathrm{Jack}}} (\eta_1,\dots,\eta_{L+K};1/2).\end{aligned}$$ For $\xi \in {\mathbb{T}}$, we obtain $$\left\langle |\det(I+ \xi M)|^{2m} \right\rangle_{S(n)} = \prod_{j=0}^{m-1} \frac{(2j+1)! (n+2m+2j+1)!}{(2m+2j+1)! (n+2j+1)!} \sim \prod_{j=0}^{m-1}\frac{(2j+1)!}{(2m+2j+1)!} \cdot n^{2m^2}.$$ $U(2n)/Sp(2n)$ – type A II {#subsectionAII} -------------------------- Consider the ensemble $S(n)$ associated with the symmetric space $U(2n)/Sp(2n)$. The space $S(n)$ is the set of all self-dual matrices in $U(2n)$, i.e., $M \in S(n)$ is a unitary matrix satisfying $M=J \trans{M} \trans{J}$ with $J=\(\begin{smallmatrix} 0 & I_n \\ -I_n & 0 \end{smallmatrix} \)$. This random matrix ensemble is called the circular symplectic ensemble (CSE). The eigenvalues of $M \in S(n)$ are of the form $z_1,z_1,z_2,z_2,\dots,z_

called the circular orthogonal ensemble (COE). We get $ $ \begin{aligned } & \left\langle \prod_{i=1}^L \det(I+\eta_i^{-1 } M^{-1 }) \cdot \prod_{i=1}^K \det(I+\eta_{L+i } M) \right\rangle_{S(n) } \\ = & \left\langle \prod_{i=1}^L \Psi^{{\mathrm{A}}}(\bz^{-1};\eta_i^{-1 }) \cdot \prod_{i=1}^K \Psi^{{\mathrm{A}}}(\bz;\eta_{L+i }) \right\rangle_{n,1 } = \prod_{i=1}^L \eta_i^{-n } \cdot P_{(n^L)}^{{\mathrm{Jack } } } (\eta_1,\dots,\eta_{L+K};1/2).\end{aligned}$$ For $ \xi \in { \mathbb{T}}$, we receive $ $ \left\langle |\det(I+ \xi M)|^{2 m } \right\rangle_{S(n) } = \prod_{j=0}^{m-1 } \frac{(2j+1)! (n+2m+2j+1)!}{(2m+2j+1)! (n+2j+1)! } \sim \prod_{j=0}^{m-1}\frac{(2j+1)!}{(2m+2j+1)! } \cdot n^{2m^2}.$$ $ U(2n)/Sp(2n)$ – type A II { # subsectionAII } -------------------------- Consider the ensemble $ S(n)$ associated with the symmetrical space $ U(2n)/Sp(2n)$. The space $ S(n)$ is the set of all self - double matrices in $ U(2n)$, i.e., $ M \in S(n)$ is a unitary matrix satisfy $ M = J \trans{M } \trans{J}$ with $ J=\(\begin{smallmatrix } 0 & I_n \\ -I_n & 0 \end{smallmatrix } \)$. This random matrix ensemble is called the round symplectic ensemble (CSE). The eigenvalues of $ M \in S(n)$ are of the form $ z_1,z_1,z_2,z_2,\dots, z _

capled the circular orthogunal ensemble (COE). We heve $$\begjn{alignea} &\left\langle \prod_{i=1}^L \det(I+\eta_i^{-1} M^{-1}) \cdot \prod_{i=1}^K \det(I+\eta_{L+i} M) \rieht\rangle_{D(n)} \\ =& \lefr\lanjle \prod_{i=1}^L \Psi^{{\mefhrm{A}}}(\bz^{-1};\cca_i^{-1}) \cslt \pxov_{i=1}^K \Psi^{{\mathrm{A}}}(\be;\eta_{L+i}) \right\sangle_{n,1} = \prod_{i=1}^L \ega_n^{-n} \cdot P_{(n^L)}^{{\mathrm{Jack}}} (\eta_1,\dots,\eta_{L+K};1/2).\enq{alignec}$$ Vor $\xi \in {\mathfb{T}}$, eq obfain $$\left\langle |\det(I+ \xi M)|^{2m} \right\rahgle_{S(n)} = \krod_{j=0}^{m-1} \frac{(2j+1)! (n+2m+2j+1)!}{(2m+2j+1)! (n+2j+1)!} \sim \prod_{j=0}^{m-1}\frac{(2j+1)!}{(2m+2j+1)!} \cdot n^{2l^2}.$$ $U(2n)/So(2n)$ – type A II {#subsfctionAII} -------------------------- Cobsidqe the ensembue $S(n)$ assobnated with jhe symmetric space $U(2n)/Sp(2n)$. The spxce $S(u)$ is the sej kf dll self-duao matgices in $U(2n)$, i.c., $M \in V(n)$ is a unitary matriw satmsfyung $M=J \trans{M} \trans{J}$ xith $J=\(\begin{smallmatryx} 0 & I_n \\ -H_n & 0 \end{smallmatrix} \)$. Rhis tandok mageix enaekbme is falmed the cidcular sympoectic ensemble (CSE). Trv eigenvalues of $M \yn S(n)$ are of the form $z_1,z_1,z_2,z_2,\dots,z_

called the circular orthogonal ensemble (COE). We &\left\langle \det(I+\eta_i^{-1} M^{-1}) \prod_{i=1}^K \det(I+\eta_{L+i} M) \Psi^{{\mathrm{A}}}(\bz^{-1};\eta_i^{-1}) \prod_{i=1}^K \Psi^{{\mathrm{A}}}(\bz;\eta_{L+i}) \right\rangle_{n,1} \prod_{i=1}^L \eta_i^{-n} \cdot (\eta_1,\dots,\eta_{L+K};1/2).\end{aligned}$$ For $\xi \in {\mathbb{T}}$, we $$\left\langle |\det(I+ \xi M)|^{2m} \right\rangle_{S(n)} = \prod_{j=0}^{m-1} \frac{(2j+1)! (n+2m+2j+1)!}{(2m+2j+1)! (n+2j+1)!} \sim \prod_{j=0}^{m-1}\frac{(2j+1)!}{(2m+2j+1)!} \cdot $U(2n)/Sp(2n)$ – type A II {#subsectionAII} -------------------------- Consider the ensemble $S(n)$ associated with symmetric $U(2n)/Sp(2n)$. space is the set of all self-dual matrices in $U(2n)$, i.e., $M \in S(n)$ is a unitary satisfying $M=J \trans{M} \trans{J}$ with $J=\(\begin{smallmatrix} 0 & \\ -I_n & 0 \)$. This random matrix ensemble called circular symplectic (CSE). eigenvalues $M \in S(n)$ of the form $z_1,z_1,z_2,z_2,\dots,z_

called the circular orthogonAl ensemble (cOE). We HavE $$\beGiN{aliGned} &\Left\langle \prod_{I=1}^l \det(i+\eta_i^{-1} M^{-1}) \cdot \prod_{i=1}^K \det(I+\etA_{L+i} M) \rIgHT\ranGLe_{s(n)} \\ =& \lefT\langle \PRoD_{I=1}^l \PsI^{{\mAtHrm{a}}}(\bZ^{-1};\EtA_i^{-1}) \cdoT \prOd_{i=1}^K \Psi^{{\Mathrm{A}}}(\bz;\eTa_{L+I}) \rIght\rangle_{n,1} = \pROd_{I=1}^L \eta_i^{-n} \cdoT P_{(n^l)}^{{\mathrm{Jack}}} (\eTa_1,\dOts,\eta_{l+K};1/2).\End{ALigneD}$$ FoR $\xi \in {\Mathbb{t}}$, We obtaIn $$\left\lanGlE |\Det(I+ \xi m)|^{2M} \right\rANGlE_{S(n)} = \pRod_{j=0}^{m-1} \frac{(2j+1)! (n+2m+2j+1)!}{(2m+2j+1)! (n+2j+1)!} \SIm \PRod_{j=0}^{m-1}\frac{(2j+1)!}{(2m+2j+1)!} \cdOt n^{2m^2}.$$ $U(2n)/sp(2N)$ – TyPE a II {#SubSectionAII} -------------------------- coNsideR The enseMBlE $s(N)$ AssOCiated with the Symmetric spACe $U(2N)/Sp(2n)$. ThE sPacE $s(n)$ is thE set oF aLL seLf-dual matriCes iN $U(2n)$, i.e., $M \in S(N)$ is a unITary matRIx satisFying $M=j \trAns{m} \traNS{J}$ WiTh $J=\(\BeGIn{sMAlLmaTRix} 0 & i_n \\ -I_n & 0 \end{sMaLlMatriX} \)$. ThiS RANDom mAtrIx enSemblE is called the cIrcUlar SYmpLectiC enseMble (cSe). The eIgenvaLues oF $M \In S(n)$ are of the forM $z_1,z_1,z_2,Z_2,\dots,z_

called the circular ortho gonal ense mble(CO E). W e ha ve $ $\begin{aligne d } &\ left\langle \prod_{i=1 }^L \d e t(I+ \ et a_i^{ -1} M^{ - 1} ) \cd ot \ pro d_ { i= 1}^K \d et(I+\e ta_{L+i} M ) \ ri ght\rangle_{ S (n )} \\ =& \le ft\langle \p rod _{i=1} ^L \ P si^{{ \ma thrm{ A}}}(\ b z^{-1} ;\eta_i^{ -1 } ) \cdo t \prod_ { i =1 }^K \Psi^{{\mathrm{A } }} ( \bz;\eta_{L+i} ) \rig ht \ ra n g le_ {n, 1} = \prod _{ i=1}^ L \eta_i ^ {- n } \cd o t P_{(n^L)}^{ {\mathrm{Ja c k}} } (\et a_ 1,\ d ots,\e ta_{L +K } ;1/ 2).\end{ali gned }$$ For $ \xi \i n {\math b b{T}}$, we ob tai n $ $\le f t\ la ngl e| \de t (I + \ x i M )|^{2m}\r ig ht\ra ngle _ { S ( n)}= \ prod _{j=0 }^{m-1} \frac {(2 j+1) ! (n +2m+2 j+1)! }{(2 m+ 2j+1) ! (n+2 j+1)! }\sim \prod_{j=0 }^{m -1}\frac{ (2j +1 )!} {( 2m+2j + 1)!} \ cdo t n ^{2m^2} .$$ $U ( 2n) /S p ( 2 n) $ – type A II {#su bs e c ti onAII} - ------ - -- -- - -------- -- --- Co n s iderthee ns emble $S (n)$ a s so ci ated wi th the s ym met ric spac e $U( 2n)/Sp (2n)$. T he sp a ce $S(n)$ is t h e set of alls el f - du a l ma tri ces in $U(2 n)$, i.e. , $M \i n S ( n)$ i s a u ni t ar y matrix satisfying$M =J \tr ans{M } \trans{J}$with $J=\( \ b e gin{smal lmat r ix } 0 & I_n \\ -I _n &0 \end{sma l lmatrix} \)$. This ra ndom matr i x ensembl e i s c all edt h ecircular symp l e ctic e nsemble (C SE). Th e e ige nva lue sof $M \in S(n)$ a re o fth e f orm $ z _1,z_1,z _2 ,z_ 2, \do ts,z_

called_the circular_orthogonal ensemble (COE). We_have $$\begin{aligned} &\left\langle_\prod_{i=1}^L_ \det(I+\eta_i^{-1}_M^{-1})_\cdot \prod_{i=1}^K \det(I+\eta_{L+i}_M) \right\rangle_{S(n)} \\ =&_ \left\langle \prod_{i=1}^L _\Psi^{{\mathrm{A}}}(\bz^{-1};\eta_i^{-1}) \cdot \prod_{i=1}^K_ \Psi^{{\mathrm{A}}}(\bz;\eta_{L+i})_\right\rangle_{n,1} = \prod_{i=1}^L \eta_i^{-n} \cdot P_{(n^L)}^{{\mathrm{Jack}}} (\eta_1,\dots,\eta_{L+K};1/2).\end{aligned}$$ For $\xi \in {\mathbb{T}}$, we obtain $$\left\langle |\det(I+_\xi_M)|^{2m} \right\rangle_{S(n)} =_\prod_{j=0}^{m-1}_\frac{(2j+1)!_(n+2m+2j+1)!}{(2m+2j+1)! (n+2j+1)!} \sim \prod_{j=0}^{m-1}\frac{(2j+1)!}{(2m+2j+1)!} \cdot n^{2m^2}.$$ $U(2n)/Sp(2n)$ –_type A II {#subsectionAII} -------------------------- Consider the_ensemble $S(n)$_associated with the symmetric space $U(2n)/Sp(2n)$. The space_$S(n)$_is the set_of all self-dual matrices in $U(2n)$, i.e., $M \in_S(n)$ is a unitary matrix satisfying_$M=J \trans{M} \trans{J}$_with_$J=\(\begin{smallmatrix}_0 & I_n \\_-I_n & 0 \end{smallmatrix} \)$. This_random matrix ensemble is called the_circular symplectic ensemble (CSE). The eigenvalues of_$M \in S(n)$ are of the_form $z_1,z_1,z_2,z_2,\dots,z_

We first put the open book $(B,\pi)$ into nice position with respect to $\xi$. After a small transverse isotopy (fixing $(B,\pi,\xi)$) we may assume that the link $L$ does not intersect the binding $B$. Let $\alpha$ be a contact form defining $\xi$ and supported by $(B,\pi)$. By Lemma \[lem:obd-cont\] we find a homotopy of contact forms $\{\alpha_t\}_{t\in [0,1]}$ supported by $(B,\pi)$, fixed outside a neighbourhood of $B$, with $\alpha_0=\alpha$ and $$\alpha_1|_W=T(r^2d\theta+(1-r^2)d\phi)$$ for some neighbourhood $W=\{r\le r_0\}$ of $B$ and some locally constant function $T>0$ on $W$, where $(\phi,r,\theta)$ are the standard open book coordinates near $B$. Using Lemma \[lem:interpol2\], we can further deform the contact form $\alpha_1$, through contact forms supported by $(B,\pi)$ and fixed outside a neighbourhood of $B$, to a contact form $\alpha_2$ satisfying $$\alpha_2|_W=T(1-r^2)(r^2d\theta+d\phi)$$ (after shrinking $W$). Finally, we perturb $\alpha_2$, keeping it fixed near $B$, to a contact form $\alpha_3$ for which the charactistic foliations on the pages are sufficiently generic (in a sense that is made precise below). After applying Gray’s theorem, we may assume that $\xi$ and $L$ are fixed and the open book is moving by an isotopy. We rename the new open book back to $(B,\pi)$ and the new contact form $\alpha_3$ back to $\alpha$. After this preparatory step, we will keep $(B,\pi,\xi)$ fixed and only deform the transverse link $L$. More precisely, we will construct a family $(\xi_t,L_t)_{t\in[0,A]}$ of contact structures $\xi_t$ and links $L_t$ in $M$ with the following properties: (a) : there exists a family of diffeomorphisms $(\psi_t)_{t\in[0,A]}$ with $$\psi_0={{{\mathchoice {\rm

We first put the open book $ (B,\pi)$ into courteous placement with respect to $ \xi$. After a small transverse isotopy (fixing $ (B,\pi,\xi)$) we may simulate that the link $ L$ does not intersect the binding $ B$. permit $ \alpha$ be a contact form specify $ \xi$ and supported by $ (B,\pi)$. By Lemma   \[lem: obd - cont\ ] we find a homotopy of liaison forms $ \{\alpha_t\}_{t\in [ 0,1]}$ supported by $ (B,\pi)$, fixed outside a neighbourhood of $ B$, with $ \alpha_0=\alpha$ and $ $ \alpha_1|_W = T(r^2d\theta+(1 - r^2)d\phi)$$ for some vicinity $ W=\{r\le r_0\}$ of $ B$ and some locally constant function $ T>0 $ on $ W$, where $ (\phi, r,\theta)$ are the standard open book coordinates near $ B$. use Lemma   \[lem: interpol2\ ], we can further deform the contact imprint $ \alpha_1 $, through contact forms supported by $ (B,\pi)$ and repair outside a neighbourhood of $ B$, to a contact form $ \alpha_2 $ satisfying $ $ \alpha_2|_W = T(1 - r^2)(r^2d\theta+d\phi)$$ (after shrinking $ W$). ultimately, we perturb $ \alpha_2 $, keeping it fixed near $ B$, to a contact form $ \alpha_3 $ for which the charactistic foliations on the pages are sufficiently generic (in a sense that is made precise below). After applying Gray ’s theorem, we may assume that $ \xi$ and $ L$ are fixed and the open book is moving by an isotopy. We rename the new exposed book back to $ (B,\pi)$ and the new liaison imprint $ \alpha_3 $ back to $ \alpha$. After this preparatory gradation, we will keep $ (B,\pi,\xi)$ fixed and only deform the transverse link $ L$. More precisely, we will construct a syndicate $ (\xi_t, L_t)_{t\in[0,A]}$ of contact structures $ \xi_t$ and links $ L_t$ in $ M$ with the following properties: (a) : there exists a kin of diffeomorphisms $ (\psi_t)_{t\in[0,A]}$ with $ $ \psi_0={{{\mathchoice { \rm

We first put the open book $(B,\pi)$ into nice kowition with despect go $\xi$. After a small transverde isotipy (fixing $(B,\pi,\xi)$) we mah assume nhat the oink $O$ does not intersegc the nindiug $B$. Let $\alpha$ bg a contact xorm defining $\fi$ aud supported by $(B,\pi)$. By Lemma \[lem:obd-cogt\] we fonf a homotopy os comeact forms $\{\alpha_t\}_{t\in [0,1]}$ supported by $(B,\pi)$, fixed mutside a neibhbourhood of $B$, with $\alpha_0=\wlphw$ and $$\alpha_1|_W=T(r^2d\theha+(1-r^2)d\phi)$$ for somq neighbourhoud $W=\{r\le r_0\}$ of $B$ and somg locally constant function $T>0$ on $W$, whzre $(\phi,r,\thejc)$ arf the standacd opeg book coordlmates tear $B$. Ising Lemma \[lem:lnter'ol2\], qe can further deform the contact form $\wlpha_1$, thrmuyh contact forms suppirred bi $(B,\pi)$ and dixdd kuvsise a nfigibourhood or $B$, to a cobtact form $\alpha_2$ sauissjong $$\alpha_2|_W=T(1-r^2)(d^2d\thetw+d\[hi)$$ (after shrinking $W$). Finally, we perturt $\ampha_2$, keeping it fixed nwar $B$, to a contact fotm $\alpha_3$ fjr which the charactistic foliations on the pages are ajffnglentut heneric (in a sense that is made precise below). Wrttr spplying Gray’s theorem, we mwy wssume that $\xk$ and $L$ ade fixed and the ooen booh is noving by an osotopy. We rename the new open book bach to $(B,\pi)$ and the nec contact foxm $\alpna_3$ bavk to $\alpha$. After this pxeparafory step, wf will keso $(B,\pi,\xi)$ fixed ana okly deform ufe transverse linh $L$. More 'recixely, we wilk conseruct a falily $(\wh_t,L_t)_{t\in[0,A]}$ of contach strocturev $\xi_t$ and pinks $L_t$ in $M$ with the followinj properties: (a) : thvre existf a fsmily of diffqomorphisms $(\psn_t)_{t\in[0,A]}$ wnth $$\psk_0={{{\mathchoicv {\rm

We first put the open book $(B,\pi)$ position respect to After a small may that the link does not intersect binding $B$. Let $\alpha$ be a form defining $\xi$ and supported by $(B,\pi)$. By Lemma \[lem:obd-cont\] we find a of contact forms $\{\alpha_t\}_{t\in [0,1]}$ supported by $(B,\pi)$, fixed outside a neighbourhood of with and for neighbourhood $W=\{r\le r_0\}$ of $B$ and some locally constant function $T>0$ on $W$, where $(\phi,r,\theta)$ are standard open book coordinates near $B$. Using Lemma we can further deform contact form $\alpha_1$, through contact supported $(B,\pi)$ and outside neighbourhood $B$, to a form $\alpha_2$ satisfying $$\alpha_2|_W=T(1-r^2)(r^2d\theta+d\phi)$$ (after shrinking $W$). Finally, we perturb $\alpha_2$, keeping it fixed near $B$, to contact form which the foliations the are sufficiently generic sense that is made precise below). theorem, we may assume that $\xi$ and $L$ fixed and open book is moving by an We rename the new open book back to and the new contact form $\alpha_3$ back to $\alpha$. After this preparatory step, we will fixed and only deform transverse link $L$. precisely, will a $(\xi_t,L_t)_{t\in[0,A]}$ of structures $\xi_t$ and links $L_t$ in $M$ with the following properties: : there exists a family of diffeomorphisms $(\psi_t)_{t\in[0,A]}$ with $$\psi_0={{{\mathchoice

We first put the open book $(B,\pi)$ iNto nice posItion WitH reSpEct tO $\xi$. AFter a small tranSVersE isotopy (fixing $(B,\pi,\xi)$) we mAy assUmE That THe Link $L$ Does not INtERSecT tHe BinDiNG $B$. let $\alPha$ Be a contAct form defIniNg $\Xi$ and supportED bY $(B,\pi)$. By LemmA \[leM:obd-cont\] we fiNd a HomotoPy Of cONtact ForMs $\{\alpHa_t\}_{t\in [0,1]}$ SUpportEd by $(B,\pi)$, fiXeD OutsidE A neighbOURhOod oF $B$, with $\alpha_0=\alpha$ aND $$\aLPha_1|_W=T(r^2d\theta+(1-r^2)d\Phi)$$ for SoME nEIGhbOurHood $W=\{r\le r_0\}$ oF $B$ And soME locallY CoNSTAnt FUnction $T>0$ on $W$, whEre $(\phi,r,\thetA)$ Are The staNdArd OPen booK coorDiNAteS near $B$. Using lemmA \[lem:interPol2\], we cAN furtheR Deform tHe contAct ForM $\alpHA_1$, tHrOugH cONtaCT fOrmS SupPorted by $(b,\pI)$ aNd fixEd ouTSIDE a neIghBourHood oF $B$, to a contact fOrm $\AlphA_2$ SatIsfyiNg $$\alpHa_2|_W=T(1-R^2)(r^2D\thetA+d\phi)$$ (aFter sHrInking $W$). Finally, wE perTurb $\alpha_2$, KeePiNg iT fIxed nEAr $B$, to a ConTacT form $\alPha_3$ for wHIch ThE CHArActistic foliations On THE pAges are sUfficiENtLy GEneric (in A sEnsE thaT IS made PrecISe Below). AftEr applYInG GRay’s theOrEm, we maY aSsuMe tHat $\xi$ ANd $L$ aRe fixeD and the oPen boOK is moving by an iSOtopy. We rename THe NEW oPEn boOk bAck to $(B,\pi)$ and The nEW conTact FOrM $\alPHa_3$ bacK to $\alPhA$. afTEr this preparatory stEp, We will Keep $(B,\Pi,\xi)$ fixed and oNly deform tHE TRansversE linK $l$. MORe precisely, we wIll coNstruct a faMIly $(\xi_t,L_t)_{T\in[0,A]}$ oF contact StructureS $\XI_t$ and linKs $L_T$ in $m$ wiTh tHE FoLlowing properTIEs: (a) : tHeRe existS a fAmily of DifFeoMorPhiSmS $(\psi_t)_{t\in[0,A]}$ With $$\psi_0={{{\mAtHcHoIcE {\rm

We first put the open boo k $(B,\pi) $ int o n ice p osit ionwith respect t o $\x i$. After a small tran svers ei soto p y(fixi ng $(B, \ pi , \ xi) $) w e m ay as sumetha t the l ink $L$ do esno t intersectt he binding $ B$. Let $\alpha $ b e a co nt act formdef ining $\xi$ and su pported b y$ (B,\pi ) $. By L e m ma  \[l em:obd-cont\] wef in d a homotopy of conta ct fo r m s $ \{\ alpha_t\}_ {t \in [ 0 ,1]}$ s u pp o r t edb y $(B,\pi)$,fixed outsi d e a neigh bo urh o od of$B$,wi t h $ \alpha_0=\a lpha $ and $$\ alpha_ 1 |_W=T(r ^ 2d\thet a+(1-r ^2) d\p hi)$ $ f or so me nei g hb our h ood $W=\{r\ le r _0\}$ of$ B $ andsom e lo cally constant fun cti on $ T >0$ on $ W$, w here $ (\phi ,r,\th eta)$ a re the standard ope n book co ord in ate snear$ B$. Us ing Le mma \[l em:inte r pol 2\ ] , we can further defor mt h econtactform $ \ al ph a _1$, thr ou ghcont a c t for ms s u pp orted by $(B,\ p i) $and fix ed outsi de anei ghbou r hood of $B $, to aconta c t form $\alpha _ 2$ satisfying $$ \ a lp h a_2| _W= T(1-r^2)(r^ 2d\t h eta+ d\ph i )$ $ ( a ftershrin ki n g$ W$). Finally, we pe rt urb $\ alpha _2$, keepingit fixed n e a r $B$, to a c o nt a ct form $\alph a_3$for whicht he chara ctist ic folia tions ont h e pagesare su ffi cie n t ly generic (ina sens ethat is ma de prec ise be low ). A fter appl ying Gra y’ sth eo rem , wem ay assum etha t$\x i$ an d $L$ a re fi xedan dt heopen bo o ki s mov in gby a n i so topy. Wer ena me thenew openboo k bac kto $(B,\p i)$ and the n ew contact f or m $ \alpha _ 3 $ back t o $\alpha$. After this prepara tor y ste p, w e will ke ep$(B,\p i,\ x i)$ fi xed an d onl ydef o r m the t ra nsv er se link $L $ . Mo re pr ec isel y, we w ill construct a fa m ily $(\xi_t,L_t) _{t \in[ 0 , A] }$o fc ont ac t st r u ctures $\xi_t$and links$L _ t$ in $M$ wi t h t he follow ing pro perti e s: (a) : the re exists a fam i l y o f diffeomo rphisms$(\psi_t) _ {t\in [ 0, A]}$wit h $$\p si _0= {{{\m athcho i ce{\rm

We_first put_the open book $(B,\pi)$_into nice_position_with respect_to_$\xi$. After a_small transverse isotopy_(fixing $(B,\pi,\xi)$) we may_assume that the_link_$L$ does not intersect the binding $B$. Let $\alpha$ be a contact form defining_$\xi$_and supported_by_$(B,\pi)$._By Lemma \[lem:obd-cont\] we find a_homotopy of contact forms $\{\alpha_t\}_{t\in_[0,1]}$ supported_by $(B,\pi)$, fixed outside a neighbourhood of $B$,_with_$\alpha_0=\alpha$ and $$\alpha_1|_W=T(r^2d\theta+(1-r^2)d\phi)$$_for some neighbourhood $W=\{r\le r_0\}$ of $B$ and some_locally constant function $T>0$ on $W$,_where $(\phi,r,\theta)$ are_the_standard_open book coordinates near_$B$. Using Lemma \[lem:interpol2\], we can further_deform the contact form $\alpha_1$, through_contact forms supported by $(B,\pi)$ and fixed_outside a neighbourhood of $B$, to_a contact form $\alpha_2$ satisfying_$$\alpha_2|_W=T(1-r^2)(r^2d\theta+d\phi)$$ (after_shrinking $W$). Finally, we perturb_$\alpha_2$, keeping it_fixed near_$B$, to a_contact form $\alpha_3$ for which the_charactistic foliations on_the pages are sufficiently generic (in_a_sense that is_made_precise_below). After applying_Gray’s theorem, we_may_assume that_$\xi$_and $L$ are fixed and the_open_book is moving by an isotopy. We_rename the new open_book_back to $(B,\pi)$ and_the new contact form $\alpha_3$_back to $\alpha$. After this preparatory step,_we will_keep $(B,\pi,\xi)$_fixed and only deform the transverse link $L$. More precisely, we_will construct a family $(\xi_t,L_t)_{t\in[0,A]}$ of_contact structures $\xi_t$ and_links $L_t$_in_$M$ with the_following_properties: (a) : _ there exists a family of diffeomorphisms_$(\psi_t)_{t\in[0,A]}$ with_$$\psi_0={{{\mathchoice {\rm

linear equations*, Uspehi Mat. Nauk 14 (1959), pp. 87–158. G. R. Goldstein, J. A. Goldstein, A. Rhandi, *Weighted Hardy’s inequality and the Kolmogorov equation perturbed by an inverse-square potential*, Appl. Anal. 91 (11) (2012), pp. 2057–2071. A. Kufner, *Weighted Sobolev spaces*, Texte zur Mathematik 31, Teubner, Stuttgart–Leipzig–Wiesbaden, 1980. A. Kufner, B. Opic, *How to define reasonably weighted Sobolev spaces*, Comment. Math. Univ. Carolinae 25 (3) (1984), pp. 537-–554. J. M. Lévy-Leblond, *Electron capture by polar molecules*, Phys. Rev. 153 (1) (1967), pp. 1–4. L. Lorenzi, M. Bertoldi, [*[Analytical Methods for Markov Semigroups]{}*]{}, Pure and Applied Mathematics 283, CRC Press, 2006. J. M. Tölle, *Uniqueness of weighted Sobolev spaces with weakly differentiable weights*, J. Funct. Anal. 263 (2012), pp. 3195–3223. V. V. Zhikov, *Weighted Sobolev spaces*, Sb. Math. 189 (8) (1998), pp. 1139–1170, translated from Math. Sb. 189 (8) (1998), pp. 27–58. *E-mail addresses: acanale@unisa.it, francesco.pappalardo@unina.it, ctarant@unina.it* [^1]: The first two authors are members of the Gruppo Nazionale per l’Analisi Matematica, la Probabilitá e le loro Applicazioni (GNAMPA) of the Istituto Nazionale di Alta Matematica (INdAM) --- author: - title: A Comparative Analysis of Android Malware --- {#sect:intro} As of 2017, the Android OS accounted for more than 85% of the mobile OS market, and there were more than 82 billion application (app) downloads from the Google Play store during 2017

linear equations *, Uspehi Mat. Nauk 14 (1959), pp. 87–158. G. R. Goldstein, J. A. Goldstein, A. Rhandi, * Weighted Hardy ’s inequality and the Kolmogorov equation perturbed by an inverse - square electric potential *, Appl. Anal. 91 (11) (2012), pp. 2057–2071. A. Kufner, * Weighted Sobolev space *, Texte zur Mathematik 31, Teubner, Stuttgart – Leipzig – Wiesbaden, 1980. A. Kufner, B. Opic, * How to define sanely weighted Sobolev space *, Comment. Math. Univ. Carolinae 25 (3) (1984), pp. 537-–554. J. M. Lévy - Leblond, * Electron capture by polar molecule *, Phys. Rev. 153 (1) (1967), pp. 1–4. L. Lorenzi, M. Bertoldi, [ * [ Analytical Methods for Markov Semigroups ] { } * ] { }, Pure and Applied Mathematics 283, CRC Press, 2006. J. M. Tölle, * Uniqueness of leaden Sobolev space with weakly differentiable weights *, J. Funct. Anal. 263 (2012), pp. 3195–3223. V. V. Zhikov, * Weighted Sobolev spaces *, Sb. Math. 189 (8) (1998), pp. 1139–1170, translated from Math. Sb. 189 (8) (1998), pp. 27–58. * E - chain mail addresses: acanale@unisa.it, francesco.pappalardo@unina.it, ctarant@unina.it * [ ^1 ]: The first two authors are member of the Gruppo Nazionale per l’Analisi Matematica, la Probabilitá e le loro Applicazioni (GNAMPA) of the Istituto Nazionale di Alta Matematica (INdAM) --- author: - championship: A Comparative Analysis of Android Malware --- { # sect: intro } As of   2017, the Android osmium accounted for more than   85%   of the mobile OS market, and there were more than   82 billion lotion (app) downloads from the Google Play store during   2017

linfar equations*, Uspehi Mat. Nauk 14 (1959), pp. 87–158. G. R. Yildstemn, J. A. Foldsteiv, A. Rhandi, *Weighted Hardy’s iiequqlity and the Kolmogorov eqjation pegturbed bt an unverse-squede potekcial*, Zipl. Auao. 91 (11) (2012), pp. 2057–2071. A. Kufngr, *Weighted Vobolev spaces*, Tdxce zur Mathematik 31, Teubner, Stuttgart–Jeipzig–Eifsbaden, 1980. A. Kufngr, B. P[ic, *Gow to define reasonably weighted Aobolev spaces*, Commemt. Math. Univ. Carolinae 25 (3) (1984), op. 537-–554. J. M. Lévy-Leblond, *Elechron capturg by polar molecuues*, Phys. Rtv. 153 (1) (1967), pp. 1–4. L. Lkrenzi, M. Bertoldi, [*[Analytical Meghods for Markoc Wemlcroups]{}*]{}, Pure and Wpplied Mathckatics 283, CRC Ptess, 2006. J. M. Tölle, *Mniquxnesw of weighted Sobolev spaces with weakli differendicble weights*, J. Funct. Qnql. 263 (2012), kp. 3195–3223. V. E. Zhkjov, *Wejgites Sobopev spaces*, Sb. Math. 189 (8) (1998), pp. 1139–1170, translated from Mstr. Sb. 189 (8) (1998), pp. 27–58. *E-majl addwefses: acanale@unisa.it, francesco.pappalardo@lninz.it, ctarant@unina.it* [^1]: The dirst two authors are members jf the Gruppo Nazionale per l’Analisi Matematica, ld Prouacilnbá d lf loro Applicazioni (GNAMPA) of the Istituto Nasjomake di Alta Matcmatica (INdAM) --- authpr: - tojle: A Comparatkve Ancmyais of Android Malaare --- {#sest:inteo} As of 2017, tre Amdroid OS accounted for morw than 85% of the nobile OS market, aud there werz more than 82 billion application (ap') downmoads from hhe Googls Play store durivg 2017

linear equations*, Uspehi Mat. Nauk 14 (1959), G. Goldstein, J. Goldstein, A. Rhandi, Kolmogorov perturbed by an potential*, Appl. Anal. (11) (2012), pp. 2057–2071. A. Kufner, Sobolev spaces*, Texte zur Mathematik 31, Teubner, Stuttgart–Leipzig–Wiesbaden, 1980. A. Kufner, B. Opic, to define reasonably weighted Sobolev spaces*, Comment. Math. Univ. Carolinae 25 (3) (1984), 537-–554. M. *Electron by polar molecules*, Phys. Rev. 153 (1) (1967), pp. 1–4. L. Lorenzi, M. Bertoldi, [*[Analytical Methods Markov Semigroups]{}*]{}, Pure and Applied Mathematics 283, CRC 2006. J. M. Tölle, of weighted Sobolev spaces with differentiable J. Funct. 263 pp. V. V. Zhikov, Sobolev spaces*, Sb. Math. 189 (8) (1998), pp. 1139–1170, translated from Math. Sb. 189 (8) (1998), pp. *E-mail addresses: ctarant@unina.it* [^1]: first authors members of the per l’Analisi Matematica, la Probabilitá e (GNAMPA) of the Istituto Nazionale di Alta Matematica --- author: title: A Comparative Analysis of Android --- {#sect:intro} As of 2017, the Android OS for more than 85% of the mobile OS market, and there were more than 82 (app) downloads from the Play store during

linear equations*, Uspehi Mat. NAuk 14 (1959), pp. 87–158. G. R. GolDsteiN, J. A. golDsTein, a. RhaNdi, *Weighted HarDY’s inEquality and the KolmogorOv equAtIOn peRTuRbed bY an inveRSe-SQUarE pOtEntIaL*, apPl. AnaL. 91 (11) (2012), pp. 2057–2071. a. Kufner, *weighted SoBolEv Spaces*, Texte zUR MAthematik 31, TEubNer, Stuttgart–leiPzig–WiEsBadEN, 1980. A. KufNer, b. Opic, *how to dEFine reAsonably wEiGHted SoBOlev spaCES*, COmmeNt. Math. Univ. CarolinAE 25 (3) (1984), pP. 537-–554. j. M. Lévy-Leblond, *ELectroN cAPtURE by PolAr moleculeS*, PHys. ReV. 153 (1) (1967), Pp. 1–4. L. LoreNZi, m. bERtoLDi, [*[Analytical MEthods for MaRKov semigrOuPs]{}*]{}, PURe and APplieD MATheMatics 283, CRC PrEss, 2006. J. m. Tölle, *UniQuenesS Of weighTEd SobolEv spacEs wIth WeakLY dIfFerEnTIabLE wEigHTs*, J. funct. AnaL. 263 (2012), pP. 3195–3223. V. v. ZhikOv, *WeIGHTEd SoBolEv spAces*, SB. Math. 189 (8) (1998), pp. 1139–1170, translAteD froM matH. Sb. 189 (8) (1998), pp. 27–58. *e-mail AddrEsSes: acAnale@uNisa.iT, fRancesco.pappalaRdo@uNina.it, ctaRanT@uNinA.iT* [^1]: The fIRst two AutHorS are memBers of tHE GrUpPO nAzIonale per l’Analisi MAtEMAtIca, la ProBabiliTÁ e Le LOro AppliCaZioNi (GNampA) of tHe IsTItUto NazioNale di aLtA MAtematiCa (iNdAM) --- aUtHor: - TitLe: A CoMParaTive AnAlysis of androID Malware --- {#sect:inTRo} As of 2017, the AndrOId os AcCOuntEd fOr more than 85% oF the MObilE OS mARkEt, aND therE were MoRE tHAn 82 billion applicatioN (aPp) downLoads From the Google play store dURINg 2017

linear equations*, UspehiMat. Nauk14 (1 959 ),pp . 87 –158 . G. R. Golds t ein, J. A. Goldstein, A. R handi ,* Weig h te d Har dy’s in e qu a l ity a nd th eK ol mogor ovequatio n perturbe d b yan inverse-s q ua re potenti al* , Appl. Anal . 9 1 (11) ( 201 2 ), pp . 2 057–2 071. A . Kufn er, *Weig ht e d Sobo l ev spac e s *, Tex te zur Mathematik 31 , Teubner, Stut tgart– Le i pz i g –Wi esb aden, 1980 . A. K u fner, B . O p i c , * H ow to definereasonablyw eig hted S ob ole v space s*, C om m ent . Math. Uni v. C arolinae25 (3) (1984), pp. 537 -–554. J . M . Lé v y- Le blo nd , *E l ec tro n ca pture by p ol ar mo lecu l e s * , Ph ys. Rev . 153 (1) (1967),pp. 1–4 . L . Lor enzi, M.Be rtold i, [*[ Analy ti cal Methods for Mar kov Semig rou ps ]{} *] {}, P u re and Ap pli ed Math ematics 283 ,C R C P ress, 2006. J. M. T ö l le , *Uniqu enesso fwe i ghted So bo lev spa c e s wit h we a kl y differ entiab l ewe ights*, J . Func t. An al. 263( 2012 ), pp. 3195–32 23. V . V. Zhikov, * W eighted Sobol e vs p ac e s*,Sb. Math. 189(8)( 1998 ), p p .113 9 –1170 , tra ns l at e d from Math. Sb. 18 9(8) (1 998), pp. 27–58. *E-mail ad d r e sses: ac anal e @u n isa.it, france sco.p appalardo@ u nina.it, ctar ant@unin a.it* [^ 1 ] : The fi rst tw o a uth o r sare members o f theGr uppo Na zio nale pe r l ’An ali siMa tematica, la Prob ab il it áe l e lor o Applica zi oni ( GNA MPA)o f theIstit utoNa zi o nal e di Al t aM a tema ti ca (IN dAM ) ---auth o r:- titl e: A Comp ara t iveAn al ysis of Android Malw ar e --- {# se ct: intro} As of 20 17, the Android OS acco u nted fo r m ore t han85% of th e m obileOSm arket, and t herewe rem o re th a n  8 2 b il lion appli c a tio n (ap p) dow nloadsfrom the Google Pl a y s tore during 2 017

linear equations*,_Uspehi Mat._Nauk 14 (1959), pp._87–158. G. R._Goldstein,_J. A._Goldstein,_A. Rhandi, *Weighted_Hardy’s inequality and_the Kolmogorov equation perturbed_by an inverse-square_potential*,_Appl. Anal. 91 (11) (2012), pp. 2057–2071. A. Kufner, *Weighted Sobolev spaces*, Texte zur Mathematik_31,_Teubner, Stuttgart–Leipzig–Wiesbaden,_1980. A._Kufner,_B. Opic, *How to define_reasonably weighted Sobolev spaces*, Comment._Math. Univ._Carolinae 25 (3) (1984), pp. 537-–554. J. M. Lévy-Leblond,_*Electron_capture by polar_molecules*, Phys. Rev. 153 (1) (1967), pp. 1–4. L. Lorenzi,_M. Bertoldi, [*[Analytical Methods for Markov_Semigroups]{}*]{}, Pure and_Applied_Mathematics_283, CRC Press, 2006. J._M. Tölle, *Uniqueness of weighted Sobolev_spaces with weakly differentiable weights*, J._Funct. Anal. 263 (2012), pp. 3195–3223. V. V._Zhikov, *Weighted Sobolev spaces*, Sb. Math._189 (8) (1998), pp. 1139–1170,_translated from_Math. Sb. 189 (8) (1998),_pp. 27–58. *E-mail addresses:_acanale@unisa.it, francesco.pappalardo@unina.it,_ctarant@unina.it* [^1]: The first_two authors are members of the_Gruppo Nazionale per_l’Analisi Matematica, la Probabilitá e le_loro_Applicazioni (GNAMPA) of_the_Istituto_Nazionale di_Alta Matematica (INdAM) _--- author: -_ title: A_Comparative_Analysis of Android Malware --- {#sect:intro} As of 2017,_the_Android OS accounted for more than 85% of the_mobile OS market, and_there_were more than 82 billion_application (app) downloads from the_Google Play store during 2017

of Schur’s $Q$-function $Q_{\lambda}$ is equal to the number of bars in a minimal bar tableaux of shape $\lambda$. Clifford [@cliff2003; @cliff2005] also proposed an open problem of determining the minimum number of bars among bar tableaux of a skew shape $\lambda/\mu$. As noted by Clifford [@cliff2003], this minimum number can be naturally regarded as the shifted rank, or srank, of $\lambda/\mu$, denoted $\mathrm{srank}(\lambda/\mu)$. For a skew bar tableau, we present an algorithm to generate a skew bar tableau without increasing the number of bars. This algorithm eventually leads to a bar tableau with the minimum number of bars. Schur’s $Q$-functions arise in the study of the projective representations of symmetric groups [@schur1911], see also, Hoffman and Humphreys [@hofhum1992], Humphreys [@humphr1986], J$\rm{\acute{o}}$zefiak [@jozef1989], Morris [@morri1962; @morri1979] and Nazarov [@nazar1988]. Shifted tableaux are closely related to Schur’s $Q$-functions analogous to the role of ordinary tableaux to the Schur functions. Sagan [@sagan1987] and Worley [@worley1984] have independently developed a combinatorial theory of shifted tableaux, which includes shifted versions of the Robinson-Schensted-Knuth correspondence, Knuth’s equivalence relations, Schützenberger’s jeu de taquin, etc. The connections between this combinatorial theory of shifted tableaux and the theory of projective representations of the symmetric groups are further explored by Stembridge [@stemb1989]. Clifford [@cliff2005] studied the srank of shifted diagrams for partitions with distinct parts. Recall that the rank of an ordinary partition is defined as the number of boxes on the main diagonal of the corresponding Young diagram. Nazarov and Tarasov [@naztar2002] found an important generalization of the rank of an ordinary partition to a skew partition in their study of tensor products of Yangian modules. A general theory of border strip decompositions and border strip tableaux of skew partitions is developed by Stanley [@stanl2002], and it has been shown that the rank of a skew partition is the least number of strips to construct a minimal border strip decomposition of the skew diagram. Motivated by Stanley’s theorem,

of Schur ’s $ Q$-function $ Q_{\lambda}$ is equal to the number of bars in a minimal bar tableau of condition $ \lambda$. Clifford [ @cliff2003; @cliff2005 ] also proposed an open trouble of determining the minimum number of measure among bar tableaux of a skew shape $ \lambda/\mu$. As note by Clifford [ @cliff2003 ], this minimum number can be naturally involve as the shifted rank, or srank, of $ \lambda/\mu$, announce $ \mathrm{srank}(\lambda/\mu)$. For a skew bar tableau, we present an algorithm to generate a skew bar tableau without increasing the act of bars. This algorithm eventually leads to a legal profession tableau with the minimum number of bars. Schur ’s $ Q$-functions arise in the survey of the projective representations of symmetric groups [ @schur1911 ], see also, Hoffman and Humphreys [ @hofhum1992 ], Humphreys [ @humphr1986 ], J$\rm{\acute{o}}$zefiak [ @jozef1989 ], Morris [ @morri1962; @morri1979 ] and Nazarov [ @nazar1988 ]. Shifted tableau are closely related to Schur ’s $ Q$-functions analogous to the role of ordinary tableaux to the Schur functions. Sagan [ @sagan1987 ] and Worley [ @worley1984 ] have independently developed a combinatorial theory of shifted tableaux, which includes shifted versions of the Robinson - Schensted - Knuth correspondence, Knuth ’s equivalence relations, Schützenberger ’s jeu de taquin, etc. The connections between this combinative hypothesis of shifted tableaux and the theory of projective representations of the symmetrical group are further explored by Stembridge [ @stemb1989 ]. Clifford [ @cliff2005 ] studied the srank of shifted diagrams for partitions with distinct character. Recall that the rank of an ordinary partition is defined as the number of box on the main diagonal of the corresponding Young diagram. Nazarov and Tarasov [ @naztar2002 ] found an important generalization of the rank and file of an ordinary partition to a skew partition in their study of tensor products of Yangian module. A general theory of border airstrip decompositions and border strip tableaux of skew partitions is develop by Stanley [ @stanl2002 ], and it has been usher that the rank of a skew division is the least number of strips to construct a minimal border airstrip decomposition of the skew diagram. motivate by Stanley ’s theorem,

of Schur’s $Q$-function $Q_{\lambdx}$ is equal to tkw numbxr of bzrs in a minimal bar tableaux of sha'e $\lqmbda$. Clifford [@cliff2003; @cliff2005] xlso proplsed an ipen problem of determiknng tgc minnmnm number of bats among bar tableaux of a syec shape $\lambda/\mu$. As noted by Clifforq [@cliff2003], tjis minimum nuiber san gv katurally regarded as the shiftes rank, mr srank, of $\lsmbda/\mu$, denoted $\mathrm{srann}(\lamhda/\mu)$. For a skew bwr tableau, qe pwwsent an algurithm to generate a siew bar tableau without increaskng tke number od vard. This algormthm enentually leads to a tar tabkeau with the kinmmum number of bars. Schur’s $Q$-functions arise yn the stgdv of the projective rwpeesenjatiots ow syomeurir gdoups [@dchnr1911], see also, Hoffman ane Humphreys [@hofhum1992], Nuiinreys [@humphr1986], J$\rm{\acttq{o}}$zefiak [@jozef1989], Morris [@morri1962; @morri1979] and Ndzadov [@nazar1988]. Shifted tablequx are closely relatgd to Schuw’s $Q$-functions analogous to the role of ordinary tdbleanx to tme Szyug functions. Sagan [@sagan1987] and Worley [@worley1984] havq imdvpendently develoied a combinatoriak hhrjry of shiftea tablzzus, which includes sjifted dersiins of tht Robonson-Schensted-Knuth correspindence, Knutk’s wquivalence relatilns, Schützenyerger’x jeu de taquin, etc. The connzctiona between tjis combihxtorial theory ow smifded tabltxux and the theorr of projxctivz represdntajions os the symmftric groups are further expllrad by Stemhridge [@stemb1989]. Clifford [@cliff2005] studmxd the srank pf shpfted diayrams nor partitions rith distinct karts. Reccll thxt the rani of an ordinary pwrtition is davined as the number jf bixes on the oain diagonal pf the cogrtsponding Toung diagram. Nazavov avs Tarasov [@naztax2002] dound an importsnt gegegalmzatijt of the rann of an prdinxry paruntiiv to a skew partition in thejr study of tensor pvoducts od Yangiag modules. A grneral theory of blrder stcip devomkositions and border strip tablsaux of snew partitions ys dcveljped by Stcnley [@stanl2002], and it has been shown that tie rank of a skew partirion is the least nombtr of strips vo conftruct a kinimal border strip decomposition of the skew diagram. Motjvated by Shanley’s theorem,

of Schur’s $Q$-function $Q_{\lambda}$ is equal to of in a bar tableaux of also an open problem determining the minimum of bars among bar tableaux of skew shape $\lambda/\mu$. As noted by Clifford [@cliff2003], this minimum number can be regarded as the shifted rank, or srank, of $\lambda/\mu$, denoted $\mathrm{srank}(\lambda/\mu)$. For a bar we an to generate a skew bar tableau without increasing the number of bars. This algorithm eventually leads a bar tableau with the minimum number of Schur’s $Q$-functions arise in study of the projective representations symmetric [@schur1911], see Hoffman Humphreys Humphreys [@humphr1986], J$\rm{\acute{o}}$zefiak Morris [@morri1962; @morri1979] and Nazarov [@nazar1988]. Shifted tableaux are closely related to Schur’s $Q$-functions analogous to the of ordinary the Schur Sagan and [@worley1984] have independently combinatorial theory of shifted tableaux, which of the Robinson-Schensted-Knuth correspondence, Knuth’s equivalence relations, Schützenberger’s de taquin, The connections between this combinatorial theory shifted tableaux and the theory of projective representations the symmetric groups are further explored by Stembridge [@stemb1989]. Clifford [@cliff2005] studied the srank of for partitions with distinct Recall that the of ordinary is as the of boxes on the main diagonal of the corresponding Young diagram. and Tarasov [@naztar2002] found an important generalization of the rank ordinary to a skew in their study of products Yangian modules. A general border decompositions tableaux skew is developed by Stanley and it has been shown the rank of a number of strips to construct a minimal border decomposition of the skew diagram. Motivated by theorem,

of Schur’s $Q$-function $Q_{\lambda}$ iS equal to thE numbEr oF baRs In a mInimAl bar tableaux oF ShapE $\lambda$. Clifford [@cliff2003; @clIff2005] alSo PRopoSEd An opeN probleM Of DETerMiNiNg tHe MInImum nUmbEr of barS among bar tAblEaUx of a skew shaPE $\lAmbda/\mu$. As nOteD by Clifford [@cLifF2003], this mInImuM NumbeR caN be naTurallY RegardEd as the shIfTEd rank, OR srank, oF $\LAmBda/\mU$, denoted $\mathrm{sraNK}(\lAMbda/\mu)$. For a skew Bar tabLeAU, wE PResEnt An algorithM tO geneRAte a skeW BaR TABleAU without increAsing the numBEr oF bars. THiS alGOrithm EventUaLLy lEads to a bar tAbleAu with the MinimuM Number oF Bars. SchUr’s $Q$-fuNctIonS ariSE iN tHe sTuDY of THe ProJEctIve repreSeNtAtionS of sYMMETric GroUps [@sChur1911], sEe also, Hoffman And humpHReyS [@hofhUm1992], HumPhreYs [@HumphR1986], J$\rm{\acUte{o}}$zEfIak [@jozef1989], Morris [@mOrri1962; @Morri1979] and NAzaRoV [@naZaR1988]. ShifTEd tablEauX arE closelY relateD To SChUR’S $q$-fUnctions analogous tO tHE RoLe of ordiNary taBLeAuX To the SchUr FunCtioNS. sagan [@SagaN1987] AnD Worley [@wOrley1984] hAVe InDependeNtLy deveLoPed A coMbinaTOriaL theorY of shiftEd tabLEaux, which incluDEs shifted versIOnS OF tHE RobInsOn-Schensted-knutH CorrEspoNDeNce, kNuth’s EquivAlENcE Relations, SchützenbeRgEr’s jeu De taqUin, etc. The connEctions betWEEN this comBinaTOrIAl theory of shifTed taBleaux and tHE theory oF projEctive rePresentatIONs of the sYmmEtrIc gRouPS ArE further exploRED by STeMbridge [@SteMb1989]. CliffOrd [@CliFf2005] sTudIeD the srank Of shifteD dIaGrAmS foR partITions witH dIstInCt pArts. RECall thAt the Rank Of An ORdiNary parTItION is dEfInEd as The NuMber oF boxES on The main Diagonal oF thE CorrEsPoNding YoUng diagram. NazArOv and TarasOv [@NazTar2002] fouND An importAnt generalization of the rANk of an oRdiNary pArtiTion to a skEw pArtitiOn iN Their sTudy of TensoR pRodUCTs of YANGiAn mOdUles. A generAL TheOry of BoRder Strip deCompositions and borDEr sTrip tableaux oF skEw paRTItIonS Is DEveLoPEd bY sTanley [@stanl2002], and iT has been shOwN ThAt the rank oF A skEw PartitiOn is the Least NUmber of Strips to cOnstruct a MiNimaL BOrdEr strip decOmpositiOn of the skEW diagRAm. motivAteD by StaNlEy’s TheorEm,

of Schur’s $Q$-function $ Q_{\lambda }$ is eq ual t o th e nu mber of bars i n a m inimal bar tableaux of shap e$ \lam b da $. Cl ifford[ @c l i ff2 00 3; @c li f f2 005]als o propo sed an ope n p ro blem of dete r mi ning the m ini mum number o f b ars am on g b a r tab lea ux of a ske w shape $\lambda /\ m u$. As noted b y Cl iffo rd [@cliff2003],t hi s minimum numbe r canbe na t u ral lyregarded a sthe s h ifted r a nk , o r s r ank, of $\lam bda/\mu$, d e not ed $\m at hrm { srank} (\lam bd a /\m u)$. For askew bar tabl eau, w e presen t an alg orithm to ge nera t easke wb art ab lea u wi thout in cr ea singthen u m b er o f b ars. This algorithm ev ent uall y le ads t o a b ar t ab leauwith t he mi ni mum number of b ars. Schur’s $Q $- fun ct ionsa rise i n t hestudy o f the p r oje ct i v e r epresentations ofsy m m et ric grou ps [@s c hu r1 9 11], see a lso , Ho f f man a nd H u mp hreys [@ hofhum 1 99 2] , Humph re ys [@h um phr 198 6], J $ \rm{ \acute {o}}$zef iak [ @ jozef1989], Mo r ris [@morri19 6 2; @ mo r ri19 79] and Nazaro v [@ n azar 1988 ] .Shi f ted t ablea ux ar e closely related to S chur’s $Q$- functions ana logous tot h e role of ord i na r y tableaux tothe S chur funct i ons. Sag an [@ sagan198 7] and Wo r l ey [@wor ley 198 4]hav e in dependently d e v elop ed a comb ina torialthe ory of sh if ted table aux, whi ch i nc lu des shif t ed versi on s o fthe Robi n son-Sc henst ed-K nu th cor respond e nc e , Knu th ’s equ iva le nce r elat i ons , Schüt zenberger ’sj eu d eta quin, e tc. The conne ct ions betwe en th is com b i natorial theory of shifted tabl e aux and th e the oryof projec tiv e repr ese n tation s of t he sy mm etr i c grou p s a refu rther expl o r edby St em brid ge [@st emb1989]. Cliffor d [@ cliff2005] st udi ed t h e s ran k o f sh if t edd i agrams for part itions wit hd is tinct part s . R ec all tha t the r ank o f an ord inary par tition is d efin e d as the numbe r of box es on the maind ia gonal of the c or res pondi ng You n g d iagra m. Naz ar ov and Tara so v [@nazt ar2002] found an import ant ge neral iza tion of t her ank of an or dina ry partiti ontoa ske w p a rtiti on i n t hei r stud y of tensor pr o du cts o fYangian mod u l e s.A gen era l theor y of border strip dec o mpositions and bor d e r s tri p tab le aux of skew pa rti ti o n s is dev el oped by Sta nley [@s ta n l2002 ], and it ha s beens h ow n thattheran k of a sk ewpa r titionis t h e leas t nu mb er ofstrips to c o n struct a minimal bord e r stri p de compo si tion of theskew diagr am. Motivat ed byStan ley’s theore m,

of_Schur’s $Q$-function_$Q_{\lambda}$ is equal to_the number_of_bars in_a_minimal bar tableaux_of shape $\lambda$._Clifford [@cliff2003; @cliff2005] also_proposed an open_problem_of determining the minimum number of bars among bar tableaux of a skew shape_$\lambda/\mu$._As noted_by_Clifford_[@cliff2003], this minimum number can_be naturally regarded as the_shifted rank,_or srank, of $\lambda/\mu$, denoted $\mathrm{srank}(\lambda/\mu)$. For a_skew_bar tableau, we_present an algorithm to generate a skew bar tableau_without increasing the number of bars._This algorithm eventually_leads_to_a bar tableau with_the minimum number of bars. Schur’s $Q$-functions_arise in the study of the_projective representations of symmetric groups [@schur1911], see_also, Hoffman and Humphreys [@hofhum1992], Humphreys_[@humphr1986], J$\rm{\acute{o}}$zefiak [@jozef1989], Morris [@morri1962;_@morri1979] and_Nazarov [@nazar1988]. Shifted tableaux are_closely related to_Schur’s $Q$-functions_analogous to the_role of ordinary tableaux to the_Schur functions. Sagan_[@sagan1987] and Worley [@worley1984] have independently_developed_a combinatorial theory_of_shifted_tableaux, which_includes shifted versions_of_the Robinson-Schensted-Knuth_correspondence,_Knuth’s equivalence relations, Schützenberger’s jeu de_taquin,_etc. The connections between this combinatorial theory_of shifted tableaux and_the_theory of projective representations_of the symmetric groups are_further explored by Stembridge [@stemb1989]. Clifford [@cliff2005]_studied the_srank of_shifted diagrams for partitions with distinct parts. Recall that the rank_of an ordinary partition is defined_as the number of_boxes on_the_main diagonal of_the_corresponding Young_diagram. Nazarov and Tarasov [@naztar2002] found an_important generalization_of the rank of an ordinary_partition to a skew_partition_in their study of tensor products_of Yangian modules. A general theory_of border strip decompositions and_border_strip_tableaux of skew partitions is_developed by Stanley [@stanl2002], and it_has been shown_that the rank of a skew partition_is_the least number of strips to_construct_a minimal border strip decomposition of_the_skew_diagram. Motivated by Stanley’s theorem,

, the simplest picture is that most FRIIs with linear sizes $\ga 10$ kpc are similar to our detected sources and so are underpressured. Cosmology --------- Our choice of cosmology does not significantly affect the relationship between radio and thermal pressures. If the Hubble parameter is $50h$ km s$^{-1}$ Mpc$^{-1}$, then the minimum pressure in the radio lobes goes approximately as $h^{4\over 7}$ (equation \[mine\]), while the thermal pressure goes as $h^{1\over 2}$ for a detected X-ray environment whose angular scale we know; no sensible change in $h$ can eliminate the pressure differences. For upper limits, we have chosen a fixed linear size, and so it might seem that the estimated upper limit on thermal pressure is independent of $h$. However, since the linear size we choose is taken from observations on the assumption of $h=1$, the upper limits on thermal pressure go as $h^{1/2}$ in this case too. Effects of projection {#projection} --------------------- Projection affects the ratio of radio to thermal pressure according to the approximate relation $${p_{\rm th}\over{p_{\rm r}}} = {\cal K}\left({p'_{\rm th}\over{p'_{\rm r}}}\right) = (\sin \theta)^{-4/7} \left({1 + R^2}\over{1 + R^2/\sin^2 \theta}\right )^{3\beta/2} \left({p'_{\rm th}\over{p'_{\rm r}}}\right) $$ where $p'_{\rm th}$ and $p'_{\rm r}$ are the originally inferred thermal and radio pressures, $\theta$ is the angle of the radio structure to the line of sight and $R$ is the ratio between the apparent (projected) distance along the radio galaxy and the core radius of the cluster (Birkinshaw & Worrall 1993). Depending on the values of $R$ and $\theta$, projection can either decrease or increase the pressure ratio. Fig. \[ratiocont\] shows contours of values of this relation as a function of $R$ and $\theta$. It will be seen that for small $R$ the assumption of no projection causes us to [*underestimate*]{} the ratio of thermal to minimum pressure (${\cal K} > 1$). This may be important, for example

, the simplest picture is that most FRIIs with linear sizes $ \ga 10 $ kpc are exchangeable to our detect sources and so are underpressured. Cosmology --------- Our option of cosmology does not significantly feign the relationship between radio and thermal pressure. If the Hubble parameter is $ 50h$ km s$^{-1}$ Mpc$^{-1}$, then the minimal atmospheric pressure in the radio lobe goes approximately as $ h^{4\over 7}$ (equation \[mine\ ]), while the thermal imperativeness go as $ h^{1\over 2}$ for a detected X - ray environment whose angular scale we know; no sensible variety in $ h$ can eliminate the pressure differences. For upper terminus ad quem, we have chosen a fixed analogue size, and so it might appear that the estimated upper terminus ad quem on thermal pressure is independent of $ h$. However, since the linear size we choose is taken from observation on the assumption of $ h=1 $, the upper limits on thermal pressure go as $ h^{1/2}$ in this case excessively. Effects of projection { # projection } --------------------- Projection affects the ratio of radio to thermal pressure according to the approximate relation $ $ { p_{\rm th}\over{p_{\rm r } } } = { \cal K}\left({p'_{\rm th}\over{p'_{\rm r}}}\right) = (\sin \theta)^{-4/7 } \left({1 + R^2}\over{1 + R^2/\sin^2 \theta}\right) ^{3\beta/2 } \left({p'_{\rm th}\over{p'_{\rm r}}}\right) $ $ where $ p'_{\rm th}$ and $ p'_{\rm r}$ are the originally deduce thermal and radio pressures, $ \theta$ is the slant of the radio receiver structure to the line of sight and $ R$ is the ratio between the apparent (plan) distance along the radio galaxy and the core radius of the bunch (Birkinshaw & Worrall 1993). Depending on the values of $ R$ and $ \theta$, projection can either decrease or increase the pressure ratio. Fig.   \[ratiocont\ ] shows contours of values of this relation as a function of $ R$ and $ \theta$. It will be seen that for small $ R$ the premise of no projection cause us to [ * underestimate * ] { } the ratio of thermal to minimum pressure ($ { \cal K } > 1 $). This may be important, for example

, thf simplest picture is thxt most FRIIs wnrh linxar sizss $\ga 10$ koc are similar to our detectxd siurcew and so are underpresrured. Cosmllogy --------- Our chomce of cosmology does nob sighlficautoy affect the telationship between radio avd thermal pressures. If the Hubble pawameter id $50h$ km s$^{-1}$ Mpc$^{-1}$, tren uhe minjmum pressure in the radio lobes gkes appgoximately as $h^{4\ovrr 7}$ (equation \[mine\]), while thf thfrmal pressure goed as $h^{1\over 2}$ for q detected X-fay environment whose zngular scale we know; no sensibue chcnge in $h$ cqn ellkinate the 'ressuge differences. For up[er limots, we have chpsei a dixed linear size, and so it might seem jhat the evtnmated upper limit on tyerman prassufw ir ihdxpehdent lf $i$. However, sjnce the libear size we choose if taken from ogservaeijns on the assumption of $h=1$, the upper likita on thermal pressure gi as $h^{1/2}$ in this case tlo. Effects of projection {#projection} --------------------- Projection affects the rdtio kw rceio tu tjermal pressure according to the approximate wslstpon $${p_{\rm th}\over{p_{\rm r}}} = {\cal K}\legt({o'_{\rk th}\over{p'_{\rm r}}}\rieht) = (\snh \fheta)^{-4/7} \left({1 + R^2}\over{1 + G^2/\sin^2 \thgta}\rigyt )^{3\beta/2} \ltft({p'_{\rk th}\over{p'_{\rm r}}}\right) $$ where $p'_{\rm th}$ and $p'_{\rm g}$ arw the originally iuferred theroal snd rsdio pressures, $\theta$ is the angle of tje radio agructure to the uinv of sight and $R$ is the ratio fetween tie ap'arent (pfojevted) dystance allng tma radio galaxy and the eore sadius of hhe cluster (Birkinshaw & Worrall 1993). Depending on tve nalues of $R$ anc $\theta$, projestion can eithgr decreafe or increase nhe pressnre ratio. Fid. \[ratiocont\] shmas contours mf valuef of thiw relatkun as a functipn of $R$ aub $\theta$. Ir will be seen thab for amall $R$ the assboktuon of no projrctkon cwuxef us to [*underastioatd*]{} the fatio of thcrmxl tp minimum pressure (${\cdl K} > 1$). This may be impottwnt, for gxample

, the simplest picture is that most linear $\ga 10$ are similar to are Cosmology --------- Our of cosmology does significantly affect the relationship between radio thermal pressures. If the Hubble parameter is $50h$ km s$^{-1}$ Mpc$^{-1}$, then the pressure in the radio lobes goes approximately as $h^{4\over 7}$ (equation \[mine\]), while thermal goes $h^{1\over for a detected X-ray environment whose angular scale we know; no sensible change in $h$ can the pressure differences. For upper limits, we have a fixed linear size, so it might seem that estimated limit on pressure independent $h$. However, since linear size we choose is taken from observations on the assumption of $h=1$, the upper limits on pressure go in this too. of {#projection} --------------------- Projection ratio of radio to thermal pressure approximate relation $${p_{\rm th}\over{p_{\rm r}}} = {\cal K}\left({p'_{\rm r}}}\right) = \theta)^{-4/7} \left({1 + R^2}\over{1 + R^2/\sin^2 )^{3\beta/2} \left({p'_{\rm th}\over{p'_{\rm r}}}\right) $$ where $p'_{\rm th}$ $p'_{\rm r}$ are the originally inferred thermal and radio pressures, $\theta$ is the angle of structure to the line sight and $R$ the between apparent distance along radio galaxy and the core radius of the cluster (Birkinshaw & 1993). Depending on the values of $R$ and $\theta$, projection decrease increase the pressure Fig. \[ratiocont\] shows contours values this relation as a $R$ $\theta$. seen for $R$ the assumption of projection causes us to [*underestimate*]{} ratio of thermal to 1$). This may be important, for example

, the simplest picture is that mOst FRIIs wiTh linEar SizEs $\Ga 10$ kpC are Similar to our deTEcteD sources and so are underpRessuReD. cosmOLoGy --------- Our Choice oF CoSMOloGy DoEs nOt SIgNificAntLy affecT the relatiOnsHiP between radiO AnD thermal prEssUres. If the HubBle ParameTeR is $50H$ Km s$^{-1}$ MpC$^{-1}$, thEn the MinimuM PressuRe in the raDiO Lobes gOEs approXIMaTely As $h^{4\over 7}$ (equation \[miNE\]), wHIle the thermal pRessurE gOEs AS $H^{1\ovEr 2}$ fOr a detecteD X-Ray enVIronmenT WhOSE AngULar scale we knoW; no sensible CHanGe in $h$ cAn EliMInate tHe preSsURe dIfferences. FOr upPer limits, We have CHosen a fIXed lineAr size, And So iT migHT sEeM thAt THe eSTiMatED upPer limit On ThErmal PresSURE Is inDepEndeNt of $h$. however, since tHe lIneaR SizE we chOose iS takEn From oBservaTions On The assumption of $H=1$, the Upper limiTs oN tHerMaL presSUre go aS $h^{1/2}$ iN thIs case tOo. EffecTS of PrOJECtIon {#projection} --------------------- ProjeCtION aFfects thE ratio OF rAdIO to thermAl PreSsurE ACcordIng tO ThE approxiMate reLAtIoN $${p_{\rm th}\oVeR{p_{\rm r}}} = {\cAl k}\leFt({p'_{\Rm th}\oVEr{p'_{\rM r}}}\righT) = (\sin \thetA)^{-4/7} \left({1 + r^2}\Over{1 + R^2/\sin^2 \theta}\rIGht )^{3\beta/2} \left({p'_{\rM Th}\OVEr{P'_{\Rm r}}}\rIghT) $$ where $p'_{\rm th}$ And $p'_{\RM r}$ arE the ORiGinALly inFerreD tHErMAl and radio pressures, $\ThEta$ is tHe angLe of the radio sTructure to THE Line of siGht aND $R$ IS the ratio betweEn the Apparent (prOJected) diStancE along thE radio galAXY and the cOre RadIus Of tHE ClUster (BirkinshAW & worrAlL 1993). DependIng On the vaLueS of $r$ anD $\thEtA$, projectiOn can eitHeR dEcReAse Or incREase the pReSsuRe RatIo. Fig. \[RAtiocoNt\] shoWs coNtOuRS of Values oF ThIS RelaTiOn As a fUncTiOn of $R$ And $\tHEta$. it will bE seen that For SMall $r$ tHe AssumptIon of no projecTiOn causes us To [*UndErestiMATe*]{} the ratIo of thermal to minimum preSSure (${\cal k} > 1$). ThIs may Be imPortant, foR exAmple

, the simplest picture isthat mostFRIIs wi thli near siz es $\ga 10$ kp c are similar to our detect ed so ur c es a n dso ar e under p re s s ure d. Cos mo l og y --- --- --- Ou r choice o f c os mology doesn ot significa ntl y affect the re lation sh ipb etwee n r adioand th e rmal p ressures. I f the H u bble pa r a me teris $50h$ km s$^{- 1 }$ Mpc$^{-1}$, th en the m i ni m u m p res sure in th eradio lobes g o es a p pro x imately as $h ^{4\over 7} $ (e quatio n\[m i ne\]), whil et hethermal pre ssur e goes as $h^{1 \ over 2} $ for adetect edX-r ay e n vi ro nme nt who s eang u lar scale w ekn ow; n o se n s i b le c han ge i n $h$ can eliminat e t he p r ess ure d iffer ence s. Forupperlimit s, we have chosen a f ixed line arsi ze, a nd so it mig htsee m thatthe est i mat ed u p pe r limit on thermal p r e ss ure is i ndepen d en to f $h$. H ow eve r, s i n ce th e li n ea r size w e choo s eis takenfr om obs er vat ion s ont he a ssumpt ion of $ h=1$, the upper limi t s on thermalp re s s ur e goas$h^{1/2}$ i n th i s ca se t o o. E f fects of p ro j ec t ion {#projection} - -- ------ ----- ------- Proj ection aff e c t s the ra tioo fr adio to therma l pre ssure acco r ding tothe a pproxima te relati o n $${p_{\ rmth} \ov er{ p _ {\ rm r}}} = {\c a l K}\ le ft({p'_ {\r m th}\o ver {p' _{\ rmr} }}\right) = (\sin \ th et a) ^{- 4/7}\ left({1+R^2 }\ ove r{1 + R^2/\s in^2\the ta }\ r igh t )^{3\ b et a / 2} \ le ft ({p' _{\ rm th}\ over { p'_ {\rm r} }}\right) $$ wher e$p '_{\rmth}$ and $p'_ {\ rm r}$ are t heorigin a l ly infer red thermal and radio p r essures , $ \thet a$ i s the ang leof the ra d io str ucture to t he li n e of s i g ht an d$R$ is the r ati o bet we en t he appa rent (projected) d i sta nce along the ra diog a la xya nd the c o rer a dius of the clu ster (Birk in s ha w & Worral l 19 93 ). Depe nding o n the valuesof $R$ an d $\theta $, pro j e cti on can eit her decr ease or i n creas e t he pr ess ure ra ti o.Fig.\[rati o con t\] s hows c on toursof va lu es of th is relation as a functi on of$R$ a nd$\theta$. It wil l be seen tha t for smal l $ R$the a ssu m ption ofn opro j ectio n ca u ses us to [* und e r es timate*]{}t h e ra tio o f t h ermalto m inimum pressure ( $ {\cal K} > 1$) . Th i s ma y b e imp or tant, for exam ple

, the_simplest picture_is that most FRIIs_with linear_sizes_$\ga 10$_kpc_are similar to_our detected sources_and so are underpressured. Cosmology --------- Our_choice of cosmology_does_not significantly affect the relationship between radio and thermal pressures. If the Hubble parameter_is_$50h$ km_s$^{-1}$_Mpc$^{-1}$,_then the minimum pressure in_the radio lobes goes approximately_as $h^{4\over_7}$ (equation \[mine\]), while the thermal pressure goes_as_$h^{1\over 2}$ for_a detected X-ray environment whose angular scale we know;_no sensible change in $h$ can_eliminate the pressure_differences._For_upper limits, we have_chosen a fixed linear size, and_so it might seem that the_estimated upper limit on thermal pressure is_independent of $h$. However, since the_linear size we choose is_taken from_observations on the assumption of_$h=1$, the upper_limits on_thermal pressure go_as $h^{1/2}$ in this case too. Effects_of projection {#projection} --------------------- Projection_affects the ratio of radio to_thermal_pressure according to_the_approximate_relation $${p_{\rm_th}\over{p_{\rm r}}} =_{\cal_K}\left({p'_{\rm th}\over{p'_{\rm r}}}\right)_=_(\sin \theta)^{-4/7} \left({1 + R^2}\over{1 + R^2/\sin^2_\theta}\right_)^{3\beta/2} \left({p'_{\rm th}\over{p'_{\rm r}}}\right) $$ where $p'_{\rm th}$ and_$p'_{\rm r}$ are the_originally_inferred thermal and radio_pressures, $\theta$ is the angle_of the radio structure to the_line of_sight and_$R$ is the ratio between the apparent (projected) distance along the_radio galaxy and the core radius_of the cluster (Birkinshaw_& Worrall_1993)._Depending on the_values_of $R$_and $\theta$, projection can either decrease or_increase the_pressure ratio. Fig. \[ratiocont\] shows contours of_values of this relation_as_a function of $R$ and $\theta$._It will be seen that for_small $R$ the assumption of_no_projection_causes us to [*underestimate*]{} the_ratio of thermal to minimum pressure_(${\cal K} > 1$)._This may be important, for example

sim \, 10^{51}$ erg s$^{-1}$). Neutrino conversions in GRB =========================== Flavor oscillations ------------------- In the framework of three flavor analysis, the flavor precession probability from $\alpha $ to $\beta $ neutrino flavor is [@book] $$P(\nu_{\alpha}\rightarrow \nu_{\beta}) \equiv P_{\alpha \beta} = \sum^{3}_{i=1}|U_{\alpha i}|^{2}|U_{\beta i}|^{2} +\sum_{i \neq j} U_{\alpha i}U^{\ast}_{\beta i} U^{\ast}_{\alpha j}U_{\beta j} \cos\left(\frac{2\pi L}{l_{ij}}\right), \label{3flvp}$$ where $\alpha, \beta = e, \mu, $ or $\tau $. $U$ is the 3$\times $3 MNS mixing matrix and can be obtained in usual notation through $$U\, \equiv R_{23}(\theta_{1}) \mbox{diag}(e^{-i\delta /2},1,e^{i\delta /2}) R_{31}(\theta_{2}) \mbox{diag}(e^{i\delta /2},1,e^{-i\delta /2}) R_{12}(\theta_{3}), \label{defU}$$ thus coinciding with the standard form given by the Particle Data Group [@pdg]. In Eq. (\[defU\]), $l_{ij}\simeq 4\pi E/ \delta m^{2}_{ij}$ with $\delta m^{2}_{ij} \equiv |m^{2}_{i}-m^{2}_{j}|$ and $L$ is the distance between the source and the detector. For simplicity, we will assume here a vanishing value for $\theta_{31}$ and CP violating phase $\delta $ in $U$. At present, the atmospheric muon and solar electron neutrino deficits can be explained with oscillations among three active neutrinos [@RECENT]. For this, typically, $\delta m^{2} \sim {\cal O}(10^{-3})$ eV$^{2}$ and $\sin^{2}2\theta \sim {\cal O}(1)$ for the explanation of atmospheric muon neutrino deficit, whereas for the explanation of solar electron neutrino deficit, we may have $\delta m^{2}

sim \, 10^{51}$ erg s$^{-1}$). Neutrino conversions in GRB = = = = = = = = = = = = = = = = = = = = = = = = = = = Flavor oscillations ------------------- In the framework of three spirit psychoanalysis, the flavor precession probability from $ \alpha $ to $ \beta $ neutrino flavor is [ @book ] $ $ P(\nu_{\alpha}\rightarrow \nu_{\beta }) \equiv P_{\alpha \beta } = \sum^{3}_{i=1}|U_{\alpha i}|^{2}|U_{\beta i}|^{2 } + \sum_{i \neq j } U_{\alpha i}U^{\ast}_{\beta iodine } U^{\ast}_{\alpha j}U_{\beta j } \cos\left(\frac{2\pi L}{l_{ij}}\right), \label{3flvp}$$ where $ \alpha, \beta = e, \mu, $ or $ \tau $. $ U$ is the 3$\times $ 3 MNS desegregate matrix and can be obtained in usual note through $ $ U\, \equiv R_{23}(\theta_{1 }) \mbox{diag}(e^{-i\delta /2},1,e^{i\delta /2 }) R_{31}(\theta_{2 }) \mbox{diag}(e^{i\delta /2},1,e^{-i\delta /2 }) R_{12}(\theta_{3 }), \label{defU}$$ thus coinciding with the standard form contribute by the Particle Data Group [ @pdg ]. In Eq. (\[defU\ ]), $ l_{ij}\simeq 4\pi E/ \delta m^{2}_{ij}$ with $ \delta m^{2}_{ij } \equiv |m^{2}_{i}-m^{2}_{j}|$ and $ L$ is the distance between the source and the detector. For simplicity, we will wear here a vanishing value for $ \theta_{31}$ and CP violating phase $ \delta $ in $ U$. At present, the atmospheric muon and solar electron neutrino deficits can be excuse with oscillations among three active neutrinos [ @RECENT ]. For this, typically, $ \delta m^{2 } \sim { \cal O}(10^{-3})$ eV$^{2}$ and $ \sin^{2}2\theta \sim { \cal O}(1)$ for the explanation of atmospheric muon neutrino deficit, whereas for the explanation of solar electron neutrino deficit, we may have $ \delta m^{2 }

sim \, 10^{51}$ erg s$^{-1}$). Neutrino converslons in GRB =========================== Flavot iscilletions ------------------- Ih the frxmework of three flavor analbsis, the dlavor precession probxbility fgom $\alpha $ to $\veta $ neutcjno flayjr ia [@boon] $$'(\nu_{\alpha}\rightartow \nu_{\beta}) \aquiv P_{\alpha \batx} = \sum^{3}_{i=1}|U_{\alpha i}|^{2}|U_{\beta i}|^{2} +\sum_{i \neq j} U_{\wlpha i}I^{\adt}_{\beta i} U^{\ast}_{\alpha j}U_{\beta j} \cos\keft(\frac{2\pi L}{l_{ij}}\right), \label{3fpvp}$$ wjere $\alpha, \beta = e, \mu, $ or $\tau $. $U$ yw the 3$\times $3 MNS mixing matrix and can be obtained in usual notatkon tkrough $$U\, \eqonc R_{23}(\hveta_{1}) \mbox{dieg}(e^{-i\dejta /2},1,e^{i\delta /2}) R_{31}(\thetd_{2}) \mbox{doag}(e^{i\delta /2},1,e^{-i\dclta /2}) R_{12}(\theta_{3}), \label{defU}$$ tius coinciding with jhe standasd form given by thw Partiwle Gata Trojp [@kdg]. Ih Eq. (\[dffU\]), $l_{ij}\simeq 4\li E/ \delta m^{2}_{uj}$ with $\delta m^{2}_{ij} \eauyn |m^{2}_{i}-m^{2}_{j}|$ and $L$ js the dystance between the source and the detebtor. For simplicity, we will assume here a vanishlng value for $\theta_{31}$ and CP violating phase $\delta $ in $U$. At psesenv, ghe abmosoyegic muon and solar electron neutrino deficits dam ne explained witm oscillations amomg tntee active neujrinos [@XSCSNT]. For this, typicwlly, $\dejta m^{2} \wim {\cal O}(10^{-3})$ eV$^{2}$ snd $\sin^{2}2\theta \sim {\cal O}(1)$ for rhe explanatpon if atmospheric muou neutrino dzficit, wherras for the explanation of aolar electgon neutrjvo deficit, we mah hsva $\delta m^{2}

sim \, 10^{51}$ erg s$^{-1}$). Neutrino conversions =========================== oscillations ------------------- the framework of precession from $\alpha $ $\beta $ neutrino is [@book] $$P(\nu_{\alpha}\rightarrow \nu_{\beta}) \equiv P_{\alpha = \sum^{3}_{i=1}|U_{\alpha i}|^{2}|U_{\beta i}|^{2} +\sum_{i \neq j} U_{\alpha i}U^{\ast}_{\beta i} U^{\ast}_{\alpha j}U_{\beta j} L}{l_{ij}}\right), \label{3flvp}$$ where $\alpha, \beta = e, \mu, $ or $\tau $. $U$ the $3 mixing and can be obtained in usual notation through $$U\, \equiv R_{23}(\theta_{1}) \mbox{diag}(e^{-i\delta /2},1,e^{i\delta /2}) R_{31}(\theta_{2}) \mbox{diag}(e^{i\delta /2}) R_{12}(\theta_{3}), \label{defU}$$ thus coinciding with the standard given by the Particle Group [@pdg]. In Eq. (\[defU\]), 4\pi \delta m^{2}_{ij}$ $\delta \equiv and $L$ is distance between the source and the detector. For simplicity, we will assume here a vanishing value for and CP $\delta $ $U$. present, atmospheric muon and neutrino deficits can be explained with active neutrinos [@RECENT]. For this, typically, $\delta m^{2} {\cal O}(10^{-3})$ and $\sin^{2}2\theta \sim {\cal O}(1)$ for explanation of atmospheric muon neutrino deficit, whereas for explanation of solar electron neutrino deficit, we may have $\delta m^{2}

sim \, 10^{51}$ erg s$^{-1}$). Neutrino conversionS in GRB =========================== FlavOr oscIllAtiOnS ------------------- In tHe frAmework of three FLavoR analysis, the flavor precEssioN pRObabILiTy froM $\alpha $ tO $\BeTA $ NeuTrInO flAvOR iS [@book] $$p(\nu_{\Alpha}\riGhtarrow \nu_{\BetA}) \eQuiv P_{\alpha \beTA} = \sUm^{3}_{i=1}|U_{\alpha i}|^{2}|u_{\beTa i}|^{2} +\sum_{i \neq j} U_{\AlpHa i}U^{\asT}_{\bEta I} u^{\ast}_{\aLphA j}U_{\beTa j} \cos\LEft(\fraC{2\pi L}{l_{ij}}\riGhT), \Label{3fLVp}$$ where $\ALPhA, \betA = e, \mu, $ or $\tau $. $U$ is the 3$\tiMEs $3 mnS mixing matrix And can Be OBtAINed In uSual notatiOn ThrouGH $$U\, \equiv r_{23}(\ThETA_{1}) \MboX{Diag}(e^{-i\delta /2},1,e^{i\Delta /2}) R_{31}(\theta_{2}) \MBox{Diag}(e^{i\DeLta /2},1,E^{-I\delta /2}) r_{12}(\thetA_{3}), \lABel{DefU}$$ thus coiNcidIng with thE standARd form gIVen by thE PartiCle datA GroUP [@pDg]. in EQ. (\[dEFU\]), $l_{IJ}\sImeQ 4\Pi E/ \Delta m^{2}_{ij}$ WiTh $\Delta M^{2}_{ij} \eQUIV |M^{2}_{i}-m^{2}_{j}|$ And $l$ is tHe disTance between tHe sOurcE And The deTectoR. For SiMplicIty, we wIll asSuMe here a vanishinG valUe for $\thetA_{31}$ anD Cp viOlAting PHase $\deLta $ In $U$. at preseNt, the atMOspHeRIC MuOn and solar electron NeUTRiNo deficiTs can bE ExPlAIned with OsCilLatiONS amonG thrEE aCtive neuTrinos [@reCeNt]. For thiS, tYpicalLy, $\DelTa m^{2} \Sim {\caL o}(10^{-3})$ eV$^{2}$ aNd $\sin^{2}2\tHeta \sim {\cAl O}(1)$ foR The explanation OF atmospheric mUOn NEUtRIno dEfiCit, whereas fOr thE ExplAnatIOn Of sOLar elEctroN nEUtRIno deficit, we may have $\DeLta m^{2}

sim \, 10^{51}$ erg s$^{-1 }$). Neut rinocon ver si onsin G RB =========== = ==== =========== Flavor os cilla ti o ns - - -- ----- ------- - -- Inth efra me w or k ofthr ee flav or analysi s,th e flavor pre c es sion proba bil ity from $\a lph a $ to $ \be t a $ n eut rinoflavor is [@b ook] $$P (\ n u_{\al p ha}\rig h t ar row\nu_{\beta}) \eq u iv P_{\alpha \bet a} = \ su m ^ {3} _{i =1}|U_{\al ph a i}| ^ {2}|U_{ \ be t a i}| ^ {2} +\sum_{i \neq j} U_ { \al pha i} U^ {\a s t}_{\b eta i } U^{\ ast}_{ \ alpha j } U_{\bet a j} \ c os \le f t(\ frac{2\p iL} {l_{i j}}\ r i g h t),\la bel{ 3flvp }$$ where $\ alp ha,\ bet a = e , \mu , $or $\ta u $. $ U$ is t he 3$\times $3MNSmixing ma tri xand c an be obtain edinusual n otation thr ou g h $$ U\, \equiv R_{23}( \t h e ta _{1}) \ mbox{d i ag }( e ^{-i\del ta /2 },1, e ^ {i\de lta/ 2} ) R_{ 31}(\t h et a_ {2}) \m bo x{diag }( e^{ i\d elta/ 2},1 ,e^{-i \delta / 2}) R_{12}(\t h eta_{3}), \la b el { d ef U }$$thu s coincidin g wi t h th e st a nd ard formgiven b y t h e Particle Data Gro up [@pdg ]. In Eq. (\[defU\ ]), $l_{ij } \ s imeq 4\p i E/ \d e lta m^{2}_{ij} $ wit h $\deltam ^{2}_{ij } \eq uiv |m^{ 2}_{i}-m^ { 2 }_{j}|$and $L $ i s t h e d istance betwe e n the s ource a ndthe det ect or. Fo r s im plicity,we willas su me h ere a va n ishing v al uefo r $ \thet a _{31}$ andCP v io la t ing phase$ \d e l ta $ i n$U$. A tprese nt,t heatmosph eric muon an d sol ar e lectron neutrino def ic its can be e xpl ainedw i th oscil lations among three act i ve neut rin os [@ RECE NT]. Forthi s, typ ica l ly, $\ deltam^{2} \ sim { \calO } (1 0^{ -3 })$ eV$^{2 } $ an d $\s in ^{2} 2\theta \sim {\cal O}(1)$ for the explanat ion ofa t mo sph e ri c mu on neu t r ino deficit, wh ereas forth e e xplanation ofso lar ele ctron n eutri n o defic it, we ma y have $\ de ltam ^ {2}

sim \,_10^{51}$ erg_s$^{-1}$). Neutrino conversions in GRB =========================== Flavor_oscillations ------------------- In the_framework_of three_flavor_analysis, the flavor_precession probability from_$\alpha $ to $\beta_$ neutrino flavor_is_[@book] $$P(\nu_{\alpha}\rightarrow \nu_{\beta}) \equiv P_{\alpha \beta} = \sum^{3}_{i=1}|U_{\alpha i}|^{2}|U_{\beta i}|^{2} +\sum_{i \neq_j}_U_{\alpha i}U^{\ast}_{\beta_i} __ _ _ _ __ U^{\ast}_{\alpha j}U_{\beta_j} _ _ ___ _\cos\left(\frac{2\pi L}{l_{ij}}\right), \label{3flvp}$$ where $\alpha, \beta = e,_\mu, $ or $\tau $. $U$_is the 3$\times $3 MNS mixing matrix_and can be obtained in usual_notation through $$U\, \equiv R_{23}(\theta_{1}) _\mbox{diag}(e^{-i\delta /2},1,e^{i\delta_/2}) R_{31}(\theta_{2}) \mbox{diag}(e^{i\delta_/2},1,e^{-i\delta /2}) _ _ _R_{12}(\theta_{3}), \label{defU}$$ thus coinciding with the standard_form given by_the Particle Data Group [@pdg]. In_Eq._(\[defU\]), $l_{ij}\simeq 4\pi_E/ \delta_m^{2}_{ij}$_with $\delta_m^{2}_{ij} \equiv |m^{2}_{i}-m^{2}_{j}|$_and_$L$ is_the_distance between the source and the_detector._For simplicity, we will assume here a_vanishing value for $\theta_{31}$_and_CP violating phase $\delta_$ in $U$. At present, the_atmospheric muon and solar electron neutrino_deficits can_be explained_with oscillations among three active neutrinos [@RECENT]. For this, typically, $\delta_m^{2} \sim {\cal O}(10^{-3})$ eV$^{2}$ and $\sin^{2}2\theta_\sim {\cal O}(1)$ for_the explanation_of_atmospheric muon neutrino_deficit,_whereas for_the explanation of solar electron neutrino deficit,_we may_have $\delta m^{2}

{\mathbf y}) = {\left( \begin{smallmatrix} \langle {\mathbf x}_1, {\mathbf x}_1 \rangle & \langle {\mathbf x}_1, {\mathbf y}\rangle \\ \langle {\mathbf x}_1, {\mathbf y}\rangle' & \langle {\mathbf y}, {\mathbf y}\rangle \end{smallmatrix} \right)} $. Again, the estimate shows that the series defining $f_{{\mathbf y}}({\bm{\tau}})$ converges absolutely to a smooth form on ${\mathbb H}_d^n \times {\mathbb D}^+$. Moreover, for a fixed $y \in V_1^{n-1}$ and any compactly supported test form $\alpha$ on ${\mathbb D}^+$, the value of the current ${ \mathfrak{g}}^o(y a_2^*)[\alpha]$ varies smoothly in the entries of $a_2^*$ (this fact follows easily from the discussion in [@GarciaSankaran §2.1.4]). Taken together, the above considerations imply that the series $\xi({\bm{\tau}})(\varphi)$ converges absolutely to a $\Gamma$-invariant current on ${\mathbb D}^+$, and therefore descends to a current on $X$ that satisfies part (b) of as ${\bm{\tau}}$ varies. In addition, this discussion shows that given any test form $\alpha$, the value of the current $\xi({\bm{\tau}})[\alpha]$ is smooth in ${\bm{\tau}}$. It remains to show that $\xi({\bm{\tau}})$ transforms like a Jacobi modular form, i.e. is invariant under the slash operators. Recall that the form $\varphi_{{\mathrm{KM}}}$ is of weight $p/2 +1$; more precisely, let $\widetilde U(1) \subset \operatorname{\widetilde{Sp}}_1({\mathbb R})$ denote the inverse image of $U(1)$, which admits a genuine character $\chi$ whose square is the identity on $U(1)$. Then $\omega(\widetilde k) \varphi_{{\mathrm{KM}}} = (\chi(\widetilde k))^{p+2} \varphi_{{\mathrm{KM}}}$ for all $\widetilde k \in \widetilde U(1)$, where $\omega$ is the Weil representation attached to $V_1$, cf. [@KudlaMillsonIHES Theorem 5.2]. To show that $\xi({\bm{\tau}})$ transforms like a Jacobi form

{ \mathbf y }) = { \left (\begin{smallmatrix } \langle { \mathbf x}_1, { \mathbf x}_1 \rangle & \langle { \mathbf x}_1, { \mathbf y}\rangle \\ \langle { \mathbf x}_1, { \mathbf y}\rangle' & \langle { \mathbf y }, { \mathbf y}\rangle \end{smallmatrix } \right) } $. Again, the estimate shows that the series defining $ f_{{\mathbf y}}({\bm{\tau}})$ converges absolutely to a legato human body on $ { \mathbb H}_d^n \times { \mathbb D}^+$. Moreover, for a fixed $ y \in V_1^{n-1}$ and any compactly support test form $ \alpha$ on $ { \mathbb D}^+$, the value of the current $ { \mathfrak{g}}^o(y a_2^*)[\alpha]$ vary smoothly in the submission of $ a_2^*$ (this fact follows well from the discussion in [ @GarciaSankaran § 2.1.4 ]). take together, the above consideration imply that the series $ \xi({\bm{\tau}})(\varphi)$ converges absolutely to a $ \Gamma$-invariant stream on $ { \mathbb D}^+$, and therefore descends to a current on $ X$ that satisfies part (b) of as $ { \bm{\tau}}$ varies. In addition, this discussion shows that contribute any test form $ \alpha$, the value of the current $ \xi({\bm{\tau}})[\alpha]$ is smooth in $ { \bm{\tau}}$. It persist to show that $ \xi({\bm{\tau}})$ transforms like a Jacobi modular form, i.e.   is invariant under the cut operators. Recall that the form $ \varphi_{{\mathrm{KM}}}$ is of weight $ p/2 +1 $; more precisely, let $ \widetilde U(1) \subset \operatorname{\widetilde{Sp}}_1({\mathbb R})$ announce the inverse image of $ U(1)$, which admits a genuine character $ \chi$ whose square is the identity on $ U(1)$. Then $ \omega(\widetilde k) \varphi_{{\mathrm{KM } } } = (\chi(\widetilde k))^{p+2 } \varphi_{{\mathrm{KM}}}$ for all $ \widetilde k \in \widetilde U(1)$, where $ \omega$ is the Weil theatrical performance attach to $ V_1 $, cf.   [ @KudlaMillsonIHES Theorem 5.2 ]. To show that $ \xi({\bm{\tau}})$ transforms like a Jacobi form

{\mahhbf y}) = {\left( \begin{smallmxtrix} \langle {\majhvf x}_1, {\methbf x}_1 \rangle & \langle {\mathbf x}_1, {\mathbf y}\ranjle \\ \langoe {\mathbf x}_1, {\mathbf y}\ravgle' & \lanhle {\mathvf y}, {\nathbf y}\raifle \end{smalljwtrir} \cight)} $. Again, thg estimate svows that the vefizs defining $f_{{\mathbf y}}({\bm{\tau}})$ converges absoluyepy to a smooth fork on ${\jathbb H}_d^n \times {\mathbb D}^+$. Moreover, for a hixed $y \in V_1^{n-1}$ amd any compactly supported tesh form $\alpha$ on ${\mahhbb D}^+$, the caluq of the currdnt ${ \mathfgck{g}}^o(y a_2^*)[\alphz]$ varies smoothly in the entrier of $c_2^*$ (this fact filllfs easily fcom thv discussion lm [@GarchaSankatan §2.1.4]). Taken togebher, vhe qbove considerations mmply that the serief $\xi({\bm{\tau}})(\eaxphi)$ converges absolureoy to a $\Gdmma$-kbvafiaht chrrent on ${\mathbb D}^+$, znd therefoee descends to a cutrqbt on $X$ that aatisfyef part (b) of as ${\bm{\tau}}$ varies. In addition, thjs discussion shows thar given any test form $\alpha$, thq value of the current $\xi({\bm{\tau}})[\alpha]$ is smooth in ${\tm{\tau}}$. Mt renalns gi dhow that $\xi({\bm{\tau}})$ transforms like a Jacobi moqhlsr form, i.e. is invcriant under the spaxr operators. Rgcall tkzt the form $\varphi_{{\mahhrm{KM}}}$ ys of weight $p/2 +1$; mote precisely, let $\widetilde Y(1) \subset \opegatoename{\widetilde{Sp}}_1({\machbb R})$ denotz the onverxe image of $U(1)$, which admnts a fenuine chagacter $\chj$ whose square is thv idantity on $U(1)$. Then $\omega(\wideeilde k) \verphi_{{\kathrm{KO}}} = (\cni(\wideeilde k))^{p+2} \vwrphi_{{\mathrm{KM}}}$ for all $\wifetilbe k \hn \widetilfe U(1)$, where $\omega$ is the Weil re'cesentation ajtawheg to $V_1$, cy. [@KudlaKillsonIHES Treorem 5.2]. To show that $\xn({\bm{\tau}})$ transforma like e Jacobi fori

{\mathbf y}) = {\left( \begin{smallmatrix} \langle {\mathbf x}_1 & \langle x}_1, {\mathbf y}\rangle y}\rangle' \langle {\mathbf y}, y}\rangle \end{smallmatrix} \right)} Again, the estimate shows that the defining $f_{{\mathbf y}}({\bm{\tau}})$ converges absolutely to a smooth form on ${\mathbb H}_d^n \times D}^+$. Moreover, for a fixed $y \in V_1^{n-1}$ and any compactly supported test $\alpha$ ${\mathbb the of the current ${ \mathfrak{g}}^o(y a_2^*)[\alpha]$ varies smoothly in the entries of $a_2^*$ (this fact follows from the discussion in [@GarciaSankaran §2.1.4]). Taken together, above considerations imply that series $\xi({\bm{\tau}})(\varphi)$ converges absolutely to $\Gamma$-invariant on ${\mathbb and descends a current on that satisfies part (b) of as ${\bm{\tau}}$ varies. In addition, this discussion shows that given any test $\alpha$, the the current is in It remains to $\xi({\bm{\tau}})$ transforms like a Jacobi modular invariant under the slash operators. Recall that the $\varphi_{{\mathrm{KM}}}$ is weight $p/2 +1$; more precisely, let U(1) \subset \operatorname{\widetilde{Sp}}_1({\mathbb R})$ denote the inverse image $U(1)$, which admits a genuine character $\chi$ whose square is the identity on $U(1)$. Then \varphi_{{\mathrm{KM}}} = (\chi(\widetilde k))^{p+2} for all $\widetilde \in U(1)$, $\omega$ the Weil attached to $V_1$, cf. [@KudlaMillsonIHES Theorem 5.2]. To show that $\xi({\bm{\tau}})$ like a Jacobi form

{\mathbf y}) = {\left( \begin{smallmatrIx} \langle {\maThbf x}_1, {\MatHbf X}_1 \rAnglE & \lanGle {\mathbf x}_1, {\mathBF y}\raNgle \\ \langle {\mathbf x}_1, {\mathbF y}\ranGlE' & \LangLE {\mAthbf Y}, {\mathbf Y}\RaNGLe \eNd{SmAllMaTRiX} \righT)} $. AgAin, the eStimate shoWs tHaT the series deFInIng $f_{{\mathbf Y}}({\bm{\Tau}})$ converges AbsOlutelY tO a sMOoth fOrm On ${\matHbb H}_d^n \TImes {\maThbb D}^+$. MoreOvER, for a fIXed $y \in V_1^{N-1}$ ANd Any cOmpactly supported TEsT Form $\alpha$ on ${\matHbb D}^+$, thE vALuE OF thE cuRrent ${ \mathfRaK{g}}^o(y a_2^*)[\ALpha]$ varIEs SMOOthLY in the entries Of $a_2^*$ (this fact FOllOws easIlY frOM the diScussIoN In [@GArciaSankarAn §2.1.4]). TaKen togethEr, the aBOve consIDeratioNs implY thAt tHe seRIeS $\xI({\bm{\TaU}})(\VarPHi)$ ConVErgEs absoluTeLy To a $\GaMma$-iNVARIant CurRent On ${\matHbb D}^+$, and therefOre DescENds To a cuRrent On $X$ tHaT satiSfies pArt (b) oF aS ${\bm{\tau}}$ varies. In aDditIon, this diScuSsIon ShOws thAT given Any TesT form $\alPha$, the vALue Of THE CuRrent $\xi({\bm{\tau}})[\alpha]$ iS sMOOtH in ${\bm{\tau}}$. it remaINs To SHow that $\xI({\bM{\taU}})$ traNSForms Like A jaCobi moduLar forM, I.e. Is InvariaNt Under tHe SlaSh oPeratORs. ReCall thAt the forM $\varpHI_{{\mathrm{KM}}}$ is of wEIght $p/2 +1$; more precISeLY, LeT $\WideTilDe U(1) \subset \opEratORnamE{\widETiLde{sP}}_1({\mathBb R})$ deNoTE tHE inverse image of $U(1)$, whiCh Admits A genuIne character $\cHi$ whose squARE Is the ideNtitY On $u(1)$. then $\omega(\widetIlde k) \Varphi_{{\mathRM{KM}}} = (\chi(\wiDetilDe k))^{p+2} \varpHi_{{\mathrm{Km}}}$ FOr all $\widEtiLde K \in \WidETIlDe U(1)$, where $\omega$ IS The WEiL represEntAtion atTacHed To $V_1$, Cf. [@KUdLaMillsoniHES TheoReM 5.2]. TO sHoW thAt $\xi({\bM{\Tau}})$ transFoRms LiKe a jacobI Form

{\mathbf y}) = {\left( \b egin{small matri x}\la ng le { \mat hbf x}_1, {\ma t hbfx}_1 \rangle & \langle {\ma th b f x} _ 1, {\ma thbf y} \ ra n g le\\ \ lan gl e { \math bfx}_1, { \mathbf y} \ra ng le' & \langl e { \mathbf y} , { \mathbf y}\r ang le \e nd {sm a llmat rix } \ri ght)}$ . Agai n, the es ti m ate sh o ws that t he ser ies defining $f_{ { \m a thbf y}}({\bm{ \tau}} )$ co n v erg esabsolutely t o a s m ooth fo r mo n ${\ m athbb H}_d^n\times {\ma t hbb D}^+$ .Mor e over,for a f i xed $y \in V_1 ^{n- 1}$ and a ny com p actly s u pported testfor m $ \alp h a$ o n $ {\ m ath b bD}^ + $,the valu eof thecurr e n t ${ \ mat hfra k{g}} ^o(y a_2^*)[\ alp ha]$ var ies s mooth ly i nthe e ntries of $ a_ 2^*$ (this fact fol lows easi lyfr omth e dis c ussion in [@ GarciaS ankaran §2. 1. 4 ] ) . Taken together, t he a bo ve consi derati o ns i m ply that t heseri e s $\xi ({\b m {\ tau}})(\ varphi ) $co nverges a bsolut el y t o a $\Ga m ma$- invari ant curr ent o n ${\mathbb D}^ + $, and theref o re d es c ends to a currenton $ X $ th at s a ti sfi e s par t (b) o f a s ${\bm{\tau}}$ vari es . In a dditi on, this disc ussion sho w s that giv en a n yt est form $\alp ha$,the valueo f the cu rrent $\xi({\ bm{\tau}} ) [ \alpha]$ is sm oot h i n ${ \bm{\tau}}$.I t re ma ins tosho w that$\x i({ \bm {\t au }})$ tran sforms l ik eaJa cob i mod u lar form ,i.e .isinvar i ant un der t he s la sh ope rators. Re c a ll t ha tthefor m$\var phi_ { {\m athrm{K M}}}$ isofw eigh t$p /2 +1$; more precise ly , let $\wi de til de U(1 ) \subset\operatorname{\widetild e {Sp}}_1 ({\ mathb b R} )$ denote th e inve rse imageof $U( 1)$,wh ich a dmits a g enu in e characte r $\c hi$ w ho se s quare i s the identity on$ U(1 )$. Then $\om ega (\wi d e ti lde k) \va rp h i_{ { \ mathrm{KM}}} =(\chi(\wid et i ld e k))^{p+2 } \v ar phi_{{\ mathrm{ KM}}} $ for al l $\widet ilde k \i n\wid e t ild e U(1)$, w here $\o mega$ ist he We i lrepre sen tation a tta chedto $V_ 1 $,cf. [ @Kudla Mi llsonI HES T he orem 5.2 ]. To show that $\xi({ \bm{\t au}}) $ t ransforms li k e a Jacobi f orm

{\mathbf_y}) =_{\left( \begin{smallmatrix} \langle {\mathbf_x}_1, {\mathbf_x}_1_\rangle &_\langle_{\mathbf x}_1, {\mathbf_y}\rangle \\ \langle_{\mathbf x}_1, {\mathbf y}\rangle'_& \langle {\mathbf_y},_{\mathbf y}\rangle \end{smallmatrix} \right)} $. Again, the estimate shows that the series defining_$f_{{\mathbf_y}}({\bm{\tau}})$ converges_absolutely_to_a smooth form on ${\mathbb_H}_d^n \times {\mathbb D}^+$. Moreover,_for a_fixed $y \in V_1^{n-1}$ and any compactly supported_test_form $\alpha$ on_${\mathbb D}^+$, the value of the current ${ \mathfrak{g}}^o(y_a_2^*)[\alpha]$ varies smoothly in the entries_of $a_2^*$ (this_fact_follows_easily from the discussion_in [@GarciaSankaran §2.1.4]). Taken together, the above_considerations imply that the series $\xi({\bm{\tau}})(\varphi)$_converges absolutely to a $\Gamma$-invariant current on_${\mathbb D}^+$, and therefore descends to_a current on $X$ that_satisfies part_(b) of as ${\bm{\tau}}$ varies._In addition, this_discussion shows_that given any_test form $\alpha$, the value of_the current $\xi({\bm{\tau}})[\alpha]$_is smooth in ${\bm{\tau}}$. It remains to_show_that $\xi({\bm{\tau}})$ transforms_like_a_Jacobi modular_form, i.e. is invariant_under_the slash_operators._Recall that the form $\varphi_{{\mathrm{KM}}}$ is_of_weight $p/2 +1$; more precisely, let $\widetilde_U(1) \subset \operatorname{\widetilde{Sp}}_1({\mathbb R})$_denote_the inverse image of_$U(1)$, which admits a genuine_character $\chi$ whose square is the_identity on_$U(1)$. Then_$\omega(\widetilde k) \varphi_{{\mathrm{KM}}} = (\chi(\widetilde k))^{p+2} \varphi_{{\mathrm{KM}}}$ for all $\widetilde k_\in \widetilde U(1)$, where $\omega$ is_the Weil representation attached_to $V_1$,_cf. [@KudlaMillsonIHES_Theorem 5.2]. To show_that_$\xi({\bm{\tau}})$ transforms_like a Jacobi form

)|^2du\biggr\}\, \bigg|\, \mathcal{F}_{T-l}\biggr )\\ = \exp\biggl\{\int_0^{T-l} \Sigma (u)\,dW(u) -\frac{1}{2}\int_0^{T-l} |\Sigma (u)|^2\, du\biggr\}\end{aligned}$$ a.s.. Let $k$ to be a positive integer such that $0 \leq T-kl \leq l$. Then by successive conditioning using backward steps of length $l$, an inductive argument gives $$\begin{aligned} E_P\biggl (\exp\biggl \{\int_0^T \Sigma (u)\,dW(u) - \frac{1}{2}\int_0^T|\Sigma (u)|^2du\biggr \}\, \bigg|\, \mathcal{F}_{T-kl}\biggr ) \\ = \exp\biggl \{\int_0^{T-kl} \Sigma (u)\,dW(u) -\frac{1}{2}\int_0^{T-kl} |\Sigma (u)|^2\, du\biggr \}\end{aligned}$$ a.s.. Taking conditional expectation with respect to $\mathcal{F}_0$ on both sides of the above equation, we obtain $$\begin{aligned} E_P\biggl (\exp\biggl \{\int_0^T \Sigma (u)\,dW(u) - \frac{1}{2}\int_0^T|\Sigma (u)|^2du\biggr\}\, \bigg|\, \mathcal{F}_0\biggr) \\ =E_P\left(\exp\left\{\int_0^{T-kl} \Sigma (u)\,dW(u) - \frac{1}{2}\int_0^{T-kl}|\Sigma (u)|^2du\right\}\, \bigg|\, \mathcal{F}_{0}\right)=1\end{aligned}$$ a.s.. Taking the expectation of the above equation, we immediately obtain $$E_P(\varrho_T)=1.$$ Therefore, the Girsanov’s theorem (Theorem 2) applies and the process $${\widehat{W}}(t) :=W(t)+\int_0^t\frac{\{\mu S(u-a)-r

) |^2du\biggr\}\, \bigg|\, \mathcal{F}_{T - l}\biggr) \\ = \exp\biggl\{\int_0^{T - l } \Sigma (u)\,dW(u) -\frac{1}{2}\int_0^{T - l } |\Sigma (u)|^2\, du\biggr\}\end{aligned}$$ a.s.. Let $ k$ to be a positive integer such that $ 0 \leq T - kl \leq l$. Then by consecutive conditioning use backward steps of length $ l$, an inductive controversy yield $ $ \begin{aligned } E_P\biggl (\exp\biggl \{\int_0^T \Sigma (u)\,dW(u) - \frac{1}{2}\int_0^T|\Sigma (u)|^2du\biggr \}\, \bigg|\, \mathcal{F}_{T - kl}\biggr) \\ = \exp\biggl \{\int_0^{T - kl } \Sigma (u)\,dW(u) -\frac{1}{2}\int_0^{T - kl } |\Sigma (u)|^2\, du\biggr \}\end{aligned}$$ a.s.. Taking conditional expectation with regard to $ \mathcal{F}_0 $ on both english of the above equation, we receive $ $ \begin{aligned } E_P\biggl (\exp\biggl \{\int_0^T \Sigma (u)\,dW(u) - \frac{1}{2}\int_0^T|\Sigma (u)|^2du\biggr\}\, \bigg|\, \mathcal{F}_0\biggr) \\ = E_P\left(\exp\left\{\int_0^{T - kl } \Sigma (u)\,dW(u) - \frac{1}{2}\int_0^{T - kl}|\Sigma (u)|^2du\right\}\, \bigg|\, \mathcal{F}_{0}\right)=1\end{aligned}$$ a.s.. Taking the expectation of the above equality, we immediately obtain $ $ E_P(\varrho_T)=1.$$ Therefore, the Girsanov ’s theorem (Theorem 2) applies and the process $ $ { \widehat{W}}(t) : = W(t)+\int_0^t\frac{\{\mu S(u - a)-r

)|^2du\blggr\}\, \bigg|\, \mathcal{F}_{T-l}\biggr )\\ = \exp\biggl\{\int_0^{T-l} \Sigma (u)\,dW(u) -\frac{1}{2}\int_0^{T-l} |\Sigma (u)|^2\, du\bijgr\}\ebd{alitned}$$ a.s.. Let $k$ to be a oositive pnteger sych uhat $0 \leq T-kl \leq l$. Then nv sucdcssivz ronditioning uslng backwarg steps of lenctf $p$, an inductive argument gives $$\begin{wligned} R_P\higgl (\exp\biggl \{\int_0^U \Sygma (l)\,dC(u) - \frac{1}{2}\int_0^T|\Sigma (u)|^2dh\biggr \}\, \bibg|\, \mathcal{F}_{T-kl}\biggr ) \\ = \exp\blggl \{\int_0^{T-kl} \Sigma (u)\,dW(u) -\seac{1}{2}\int_0^{T-kl} |\Siema (u)|^2\, du\biggr \}\end{aligngd}$$ a.s.. Taking conditional expectagion cith respecj fo $\kathcal{F}_0$ on both sides of thc above equatipn, we obtain $$\bcgin{anigbed} E_P\biggl (\exp\biggl \{\iit_0^T \Sigma (u)\,dW(u) - \frdc{1}{2}\nnt_0^T|\Sigma (u)|^2du\biggr\}\, \bigc|\, \magycau{F}_0\bjgjr) \\ =S_P\left(\fxp\meft\{\int_0^{T-kl} \Sigma (u)\,dW(u) - \frac{1}{2}\int_0^{U-kl}|\Fpbma (u)|^2du\right\}\, \bigg|\, \mathcal{F}_{0}\right)=1\end{aligned}$$ a.s.. Takinc tge expectation of the avove equation, we immefiately oftain $$E_P(\varrho_T)=1.$$ Therefore, the Girsanov’s theorem (Tveorej 2) akpoies xbd the process $${\widehat{W}}(t) :=W(t)+\int_0^t\frac{\{\mu S(u-a)-r

)|^2du\biggr\}\, \bigg|\, \mathcal{F}_{T-l}\biggr )\\ = \exp\biggl\{\int_0^{T-l} \Sigma |\Sigma du\biggr\}\end{aligned}$$ a.s.. $k$ to be $0 T-kl \leq l$. by successive conditioning backward steps of length $l$, an argument gives $$\begin{aligned} E_P\biggl (\exp\biggl \{\int_0^T \Sigma (u)\,dW(u) - \frac{1}{2}\int_0^T|\Sigma (u)|^2du\biggr \}\, \bigg|\, ) \\ = \exp\biggl \{\int_0^{T-kl} \Sigma (u)\,dW(u) -\frac{1}{2}\int_0^{T-kl} |\Sigma (u)|^2\, du\biggr \}\end{aligned}$$ a.s.. conditional with to on both sides of the above equation, we obtain $$\begin{aligned} E_P\biggl (\exp\biggl \{\int_0^T \Sigma (u)\,dW(u) - (u)|^2du\biggr\}\, \bigg|\, \mathcal{F}_0\biggr) \\ =E_P\left(\exp\left\{\int_0^{T-kl} \Sigma (u)\,dW(u) - (u)|^2du\right\}\, \bigg|\, \mathcal{F}_{0}\right)=1\end{aligned}$$ a.s.. the expectation of the above we obtain $$E_P(\varrho_T)=1.$$ the theorem 2) applies and process $${\widehat{W}}(t) :=W(t)+\int_0^t\frac{\{\mu S(u-a)-r

)|^2du\biggr\}\, \bigg|\, \mathcal{F}_{T-l}\biggR )\\ = \exp\biggl\{\iNt_0^{T-l} \SIgmA (u)\,dw(u) -\Frac{1}{2}\Int_0^{T-L} |\Sigma (u)|^2\, du\biggr\}\ENd{alIgned}$$ a.s.. Let $k$ to be a positiVe intEgER sucH ThAt $0 \leq t-kl \leq l$. tHeN BY suCcEsSivE cONdItionIng Using baCkward stepS of LeNgth $l$, an inducTIvE argument gIveS $$\begin{aligneD} E_P\Biggl (\eXp\BigGL \{\int_0^T \sigMa (u)\,dW(U) - \frac{1}{2}\iNT_0^T|\SigmA (u)|^2du\biggr \}\, \BiGG|\, \mathcAL{F}_{T-kl}\biGGR ) \\ = \eXp\biGgl \{\int_0^{T-kl} \Sigma (u)\,dW(U) -\FrAC{1}{2}\int_0^{T-kl} |\Sigma (u)|^2\, dU\biggr \}\EnD{AlIGNed}$$ A.s.. TAking condiTiOnal eXPectatiON wITH ResPEct to $\mathcal{F}_0$ On both sides OF thE above EqUatIOn, we obTain $$\bEgIN{alIgned} E_P\biggL (\exp\Biggl \{\int_0^T \sigma (u)\,Dw(u) - \frac{1}{2}\iNT_0^T|\Sigma (U)|^2du\bigGr\}\, \bIgg|\, \MathCAl{f}_0\bIggR) \\ =E_p\LefT(\ExP\leFT\{\inT_0^{T-kl} \SigmA (u)\,DW(U) - \frac{1}{2}\Int_0^{T-KL}|\sIGma (u)|^2Du\rIght\}\, \Bigg|\, \mAthcal{F}_{0}\right)=1\eNd{aLignED}$$ a.s.. takinG the eXpecTaTion oF the abOve eqUaTion, we immediateLy obTain $$E_P(\varRho_t)=1.$$ THerEfOre, thE girsanOv’s TheOrem (TheOrem 2) appLIes AnD THE pRocess $${\widehat{W}}(t) :=W(t)+\iNt_0^T\FRaC{\{\mu S(u-a)-r

)|^2du\biggr\}\, \bigg|\,\mathcal{F }_{T- l}\ big gr )\\ =\exp\biggl\{\i n t_0^ {T-l} \Sigma (u)\,dW(u ) -\f rac{1}{ 2 }\ i n t_0 ^{ T- l}|\ S ig ma (u )|^ 2\, du\ biggr\}\en d{a li gned}$$ a.s. . L et $k$ tobea positive i nte ger su ch th a t $0\le q T-k l \leq l$. Th en by suc ce s sive c o ndition i n gusin g backward stepso fl ength $l$, aninduct iv e a r g ume ntgives $$\b eg in{al i gned} E _ P\ b i g gl( \exp\biggl \{ \int_0^T \S i gma (u)\, dW (u) -\fr ac{1}{2}\in t_0^ T|\Sigma(u)|^2 d u\biggr \}\, \bi gg|\ , \ ma thc al { F}_ { T- kl} \ big gr ) \\=\e xp\bi ggl\ { \ i nt_0 ^{T -kl} \Sig ma (u)\,dW(u) -\fra c{1} {2 }\int _0^{T- kl} | \S igma (u)|^2\, d u\bi ggr \}\en d{a li gne d} $$ a. s .. Tak ing co ndition al expe c tat io n w it h respect to $\mat hc a l {F }_0$ onboth s i de so f the ab ov e e quat i o n, we obt a in $$\begi n{alig n ed }E_P\big gl (\exp \b igg l \ {\int _ 0^T\Sigma (u)\,dW (u) - \f r ac{1}{2}\int_ 0 ^T | \ Si g ma ( u)| ^2du\biggr\ }\, \ big g |\, \ mathc al { F} _ 0\biggr) \\ =E_P\le ft (\exp\ left\ {\int_0^{T-kl } \Sigma ( u ) \ ,dW(u) - \frac{1}{2 }\int _0^{T-kl}| \ Sigma (u )|^2d u\right\ }\, \bigg |\, \m ath cal { F }_ {0}\right)=1\ e n d{al ig ned}$$a.s .. Taki ngthe ex pec ta tion of t he above e qu at io n,we im m ediately o bta in $$ E_P(\ v arrho_ T)=1. $$ T he re f ore , the G i rs a n ov’s t he orem (T he orem2) a p pli es andthe proce ss$ ${\w id eh at{W}}( t) :=W(t)+\ in t_0^t\frac {\ {\m u S(u- a ) -r

)|^2du\biggr\}\, \bigg|\, \mathcal{F}_{T-l}\biggr_)\\ =_\exp\biggl\{\int_0^{T-l} \Sigma (u)\,dW(u) _ __ __ _ _ -\frac{1}{2}\int_0^{T-l} |\Sigma (u)|^2\,_du\biggr\}\end{aligned}$$ a.s.. Let_$k$_to be a positive integer such that $0 \leq T-kl \leq l$. Then by_successive_conditioning using_backward_steps_of length $l$, an inductive_argument gives $$\begin{aligned} E_P\biggl (\exp\biggl \{\int_0^T_\Sigma (u)\,dW(u) _ __ - _\frac{1}{2}\int_0^T|\Sigma (u)|^2du\biggr \}\, _ \bigg|\, \mathcal{F}_{T-kl}\biggr_) \\ = \exp\biggl_\{\int_0^{T-kl}_\Sigma_(u)\,dW(u) _ _ -\frac{1}{2}\int_0^{T-kl}_|\Sigma (u)|^2\, du\biggr \}\end{aligned}$$ a.s.. Taking conditional_expectation with respect to $\mathcal{F}_0$ on_both sides of the above_equation, we_obtain $$\begin{aligned} E_P\biggl (\exp\biggl \{\int_0^T \Sigma_(u)\,dW(u) _ _ _ - \frac{1}{2}\int_0^T|\Sigma_(u)|^2du\biggr\}\, _ __ \bigg|\, \mathcal{F}_0\biggr)_\\ =E_P\left(\exp\left\{\int_0^{T-kl}_\Sigma_(u)\,dW(u) _ __ __ - \frac{1}{2}\int_0^{T-kl}|\Sigma (u)|^2du\right\}\, __ _\bigg|\, \mathcal{F}_{0}\right)=1\end{aligned}$$ a.s.. Taking_the_expectation of the above_equation, we immediately obtain $$E_P(\varrho_T)=1.$$_Therefore, the Girsanov’s theorem (Theorem 2)_applies and_the process_$${\widehat{W}}(t) :=W(t)+\int_0^t\frac{\{\mu S(u-a)-r

the modeling, there is not a strict cutoff frequency which separates the propagating (high-frequency) waves from the non-propagating (evanescent, low-frequency) waves. Instead, all frequencies are propagated and reflected simultaneously, and the ratio of propagation versus reflection grows with frequency. Although there is no cutoff frequency, an effective cutoff frequency, separating mainly propagating waves from mainly non-propagating waves, can be defined. The parameters determined from this fitting for all the analyzed pairs of lines are displayed in Table \[table:parameters\]. ![Same as Fig. \[fig:dphase\_CaHe\] but for the line pair  10839 Å and  10827 Å.[]{data-label="fig:dphase_CaSi"}](dphase_CaSi_total.eps){width="9cm"} ![Same as Fig. \[fig:dphase\_CaHe\] but for the line pair  10827 Å and  5434.534 Å.[]{data-label="fig:dphase_SiFe"}](dphase_SiFe_total.eps){width="9cm"} Line Pair T (K) $\Delta z$ (km) $\tau_{\rm R}$ (s) $\omega_{\rm c}$ (mHz) ----------- ------- ----------------- -------------------- ------------------------  -   4000 130 30 5.0  -   3000 200 20 6.0  -   3500 750 40 3.6  -   7000 150 20 3.1  -   4000 1100 40 3.3  -   4000 1000 40 3.3 : \[table:parameters\] [Parameters for phase and amplification spectra fits and observed cutoff frequency]{} [The first column indicates the pair of lines used in the analysis, second to fourth columns are the temperature ($T$), difference in formation heights ($\Delta z$), and characteristic timescale of the radiative losses ($\tau_{\rm R}$) retrieved from the fit of the phase and amplification spectra to a model of linear wave propagation in a gravitationally stratifed atmosphere with radiative losses, and the last column in the observationally measured cutoff frequency ($\omega_{\rm c}$).]{} Figure \[fig:dphase\_CaHe\] indicates that propagating waves go upwards in the umbral atmosphere. The examination of

the modeling, there is not a strict cutoff frequency which distinguish the propagate (high - frequency) waves from the non - propagating (evanescent, low - frequency) wave. Instead, all frequencies are spread and reflect simultaneously, and the ratio of generation versus mirror image grows with frequency. Although there be no cutoff frequency, an effective cutoff frequency, distinguish mainly propagating waves from mainly non - propagating waves, can be defined. The parameter determined from this fitting for all the analyzed pairs of line are displayed in Table \[table: parameters\ ]. ! [ Same as Fig. \[fig: dphase\_CaHe\ ] but for the line pair   10839 Å   and   10827 Å.[]{data - label="fig: dphase_CaSi"}](dphase_CaSi_total.eps){width="9 centimeter " } ! [ Same as Fig. \[fig: dphase\_CaHe\ ] but for the line pair   10827 Å   and   5434.534 Å.[]{data - label="fig: dphase_SiFe"}](dphase_SiFe_total.eps){width="9 cm " } Line Pair T (K) $ \Delta z$ (km) $ \tau_{\rm R}$ (s) $ \omega_{\rm c}$ (mHz) ----------- ------- ----------------- -------------------- ------------------------   -   4000 130 30 5.0   -   3000 200 20 6.0   -   3500 750 40 3.6   -   7000 150 20 3.1   -   4000 1100 40 3.3   -   4000 1000 40 3.3 : \[table: parameters\ ] [ Parameters for phase and amplification spectra fits and ascertained cutoff frequency ] { } [ The first column indicates the pair of lines used in the psychoanalysis, second to fourth column are the temperature ($ T$), difference in formation heights ($ \Delta z$), and characteristic timescale of the radiative losses ($ \tau_{\rm R}$) retrieved from the fit of the phase and amplification spectra to a model of linear wave propagation in a gravitationally stratifed standard atmosphere with radiative losses, and the last column in the observationally measure shortcut frequency ($ \omega_{\rm c}$). ] { } Figure \[fig: dphase\_CaHe\ ] indicate that propagating waves go upwards in the umbral atmosphere. The examination of

thf modeling, there is not x strict cutoff frequeicy whidh separxtes the propagating (high-freeuwncy) qaves from the non-propxgating (enanescent, low-hrequency) waves. Mhstead, all frseueneixs are propagatgd and reflewted simultanemurlv, and the ratio of propagation versuf reflevtlon grows with freatencg. Although there is no cutoff freqhency, ai effective cutpff frequency, separating mwinlj propagating waved from mainoy njb-propagating waves, can be defined. Jhe parameters determined from tfis fntting for qlo tjg analyzed peirs os lines are displayeg in Tanle \[table:paramcters\]. ![Vamw as Fig. \[fig:dphase\_CaHx\] but for the line pwir  10839 Å and  10827 Å.[]{dcta-label="fig:dphase_CaSi"}](epyase_CdSi_tmtal.dps){wkdtg="9ck"} ![Szme as Fij. \[fig:dphase\_DaHe\] but foe the line pair  10827 Å anc  5434.534 Å.[]{eata-label="fig:dlhase_SyFq"}](dphase_SiFe_total.eps){width="9cm"} Line Pair T (K) $\Delta z$ (km) $\tau_{\rm R}$ (s) $\omega_{\rm c}$ (mHz) ----------- ------- ----------------- -------------------- ------------------------  -   4000 130 30 5.0  -   3000 200 20 6.0  -   3500 750 40 3.6  -   7000 150 20 3.1  -   4000 1100 40 3.3  -   4000 1000 40 3.3 : \[gabke:[arameters\] [Parameters for [hase and ampkificatkon xpectrw fits and obseveed cutoff frequenfy]{} [The fhrst columj indicates the pair of lines used in the anakyvis, second co fouvth columns are the temperatute ($T$), diffzrence in formatpon heighvs ($\Delta z$), agd characterivjic timescale of the wadiqtivw losser ($\tau_{\rm R}$) retrirved from the fit od the phase and amilifizztion spectra ti a model of lineat wxve pgopegatijt in a gravidatiunauky stfatifed atmirphete with radiative lovses, and the last colukn in the ibservatyonally measuted cutoff frequenfy ($\omxga_{\rm r}$).]{} Figurr \[fyg:dphase\_CaHe\] indicates that prkpagating wayes go upwardf in the umbral ctmosphere. The examination of

the modeling, there is not a strict which the propagating waves from the all are propagated and simultaneously, and the of propagation versus reflection grows with Although there is no cutoff frequency, an effective cutoff frequency, separating mainly propagating from mainly non-propagating waves, can be defined. The parameters determined from this fitting all analyzed of are displayed in Table \[table:parameters\]. ![Same as Fig. \[fig:dphase\_CaHe\] but for the line pair 10839 Å 10827 Å.[]{data-label="fig:dphase_CaSi"}](dphase_CaSi_total.eps){width="9cm"} ![Same as Fig. \[fig:dphase\_CaHe\] but for line pair 10827 Å 5434.534 Å.[]{data-label="fig:dphase_SiFe"}](dphase_SiFe_total.eps){width="9cm"} Line Pair T $\Delta (km) $\tau_{\rm (s) c}$ ----------- ------- ----------------- ------------------------ - 4000 130 30 5.0 - 3000 200 20 6.0 - 3500 750 40 3.6 - 150 20 4000 1100 3.3 4000 40 3.3 : for phase and amplification spectra fits frequency]{} [The first column indicates the pair of used in analysis, second to fourth columns are temperature ($T$), difference in formation heights ($\Delta z$), characteristic timescale of the radiative losses ($\tau_{\rm R}$) retrieved from the fit of the phase spectra to a model linear wave propagation a stratifed with losses, and last column in the observationally measured cutoff frequency ($\omega_{\rm c}$).]{} Figure indicates that propagating waves go upwards in the umbral atmosphere. of

the modeling, there is not a strIct cutoff fRequeNcy WhiCh SepaRateS the propagatinG (High-Frequency) waves from the nOn-proPaGAtinG (EvAnescEnt, low-fREqUENcy) WaVeS. InStEAd, All frEquEncies aRe propagatEd aNd Reflected simULtAneously, anD thE ratio of propAgaTion veRsUs rEFlectIon Grows With frEQuency. although tHeRE is no cUToff freQUEnCy, an Effective cutoff frEQuENcy, separating mAinly pRoPAgATIng WavEs from mainLy Non-prOPagatinG WaVES, Can BE defined. The paRameters detERmiNed froM tHis FItting For alL tHE anAlyzed pairs Of liNes are disPlayed IN Table \[tABle:paraMeters\]. ![samE as fig. \[fIG:dPhAse\_cahE\] buT FoR thE LinE pair  10839 Å anD  10827 Å.[]{DaTa-labEl="fiG:DPHAse_CASi"}](DphaSe_CaSI_total.eps){widtH="9cm"} ![same AS FiG. \[fig:dPhase\_caHe\] BuT for tHe line Pair  10827 Å AnD  5434.534 Å.[]{data-label="fig:dPhasE_SiFe"}](dphaSe_SIFE_toTaL.eps){wIDth="9cm"} LIne paiR T (K) $\DeltA z$ (km) $\tau_{\RM R}$ (s) $\OmEGA_{\Rm C}$ (mHz) ----------- ------- ----------------- -------------------- ------------------------  -   4000 130 30 5.0  -   3000 200 20 6.0  -   3500 750 40 3.6  -   7000 150 20 3.1  -   4000 1100 40 3.3  -   4000 1000 40 3.3 : \[table:parameterS\] [PARAmEters for Phase aND aMpLIficatioN sPecTra fITS and oBserVEd Cutoff frEquencY]{} [thE fIrst colUmN indicAtEs tHe pAir of LInes Used in The analySis, seCOnd to fourth colUMns are the tempERaTURe ($t$), DiffEreNce in formatIon hEIghtS ($\DelTA z$), And CHaracTerisTiC TiMEscale of the radiativE lOsses ($\tAu_{\rm R}$) Retrieved from The fit of thE PHAse and amPlifICaTIon spectra to a mOdel oF linear wavE PropagatIon in A gravitaTionally sTRAtifed atMosPheRe wIth RADiAtive losses, anD THe laSt Column iN thE observAtiOnaLly MeaSuRed cutoff FrequencY ($\oMeGa_{\Rm C}$).]{} FiGure \[fIG:dphase\_CAHE\] inDiCatEs thaT PropagAting WaveS gO uPWarDs in the UMbRAL atmOsPhEre. THe eXaMinatIon oF

the modeling, there is no t a strict cuto fffre qu ency whi ch separates t h e pr opagating (high-freque ncy)wa v es f r om thenon-pro p ag a t ing ( ev ane sc e nt , low -fr equency ) waves. I nst ea d, all frequ e nc ies are pr opa gated and re fle cted s im ult a neous ly, andthe ra t io ofpropagati on versus reflect i o ngrow s with frequency. Al t hough there is no cu to f ff r equ enc y, an effe ct ive c u toff fr e qu e n c y,s eparating mai nly propaga t ing waves f rom mainly non- pr o pag ating waves , ca n be defi ned. T h e param e ters de termin edfro m th i sfi tti ng for al l t h e a nalyzedpa ir s ofline s a r e di spl ayed in T able \[table: par amet e rs\ ]. ! [Same asFi g. \[ fig:dp hase\ _C aHe\] but for t he l ine pair 10 83 9 Å  a nd  1 0 827 Å. []{ dat a-label ="fig:d p has e_ C a S i" }](dphase_CaSi_tot al . e ps ){width= "9cm"} ! [S a me as Fi g. \[ fig: d p hase\ _CaH e \] but for the l i ne p air  10 82 7 Å an d 54 34. 534 Å . []{d ata-la bel="fig :dpha s e_SiFe"}](dpha s e_SiFe_total. e ps ) { wi d th=" 9cm "} Line Pai r T (K) $\D e lta z $ (km ) $ \ tau_{\rm R}$ (s) $\ omega_ {\rmc}$ (mHz) - ---------- - - ----- -- ---- - -- - ------- ------ ----- ---------- -------- ----- -------- -- -   4 000 1 30 30 5. 0  - 300 0 2 00 20 6. 0  - 3500 75 0 40 3. 6 -  70 00 1 50 20 3.1  -   4000 110 0 40 3 .3  - 4 000 100 0 40 3.3 : \[ta ble :par a m et ers \ ][ Par am e ter s for phase and a mplificati on sp ectra fits and o bserved cutoff freq u ency]{} [The fi rst colum nindi c a tes the pairof lines used int he an a ly sis,sec ond to f our th co lumnsa rethe t empera tu re ($T $), d if ferencein formation heights ($ \Delta z$), an d charact eri s tic timescal e of the radia tiv e l osses ($ \ tau_{ \rmR }$ ) r e triev ed f r om the fi t o f t h e p hase and am p l i fic ation sp e ctra t o amodel of linear w a ve propagation ina gra vit a tion al ly stratifed a tmo sp h e re withra diative los ses, and t h e las t colu mn inthe obs e r va t ionall y me asu red cutof f f re q uency ( $\ om e ga_{\r m c} $) .]{} Figure \[fi g : dphase\_CaHe\] i ndica t e s tha t pr opaga ti ng wave s goupwards in the umbral atmos pher e. Th e exami na tion o f

the_modeling, there_is not a strict_cutoff frequency_which_separates the_propagating_(high-frequency) waves from_the non-propagating (evanescent,_low-frequency) waves. Instead, all_frequencies are propagated_and_reflected simultaneously, and the ratio of propagation versus reflection grows with frequency. Although there_is_no cutoff_frequency,_an_effective cutoff frequency, separating mainly_propagating waves from mainly non-propagating_waves, can_be defined. The parameters determined from this fitting_for_all the analyzed_pairs of lines are displayed in Table \[table:parameters\]. ![Same as_Fig. \[fig:dphase\_CaHe\] but for the line_pair  10839 Å and_ 10827_Å.[]{data-label="fig:dphase_CaSi"}](dphase_CaSi_total.eps){width="9cm"} ![Same_as Fig. \[fig:dphase\_CaHe\] but_for the line pair  10827 Å and_ 5434.534 Å.[]{data-label="fig:dphase_SiFe"}](dphase_SiFe_total.eps){width="9cm"} Line Pair_ T (K) $\Delta_z$ (km) $\tau_{\rm R}$_(s) $\omega_{\rm c}$_(mHz) _----------- ------- ----------------- -------------------- ------------------------ _ _  -_  _ 4000 _ _ 130 __ ___ _ __ __30 __ _ __ 5.0 _  -   _ 3000 _ _ _ 200 _ _ _20 __ __ _ _ _ 6.0 _  -   __ 3500 _ _ 750 ___ _ _40 _ __ __ 3.6 __ -_  _ 7000 _ 150 _ _ ___ 20 _ _ _ 3.1 _  -   _ 4000 _ 1100 _ _ 40__ ____ _ _ 3.3 _ _  -_  _ __4000 1000 _ _ 40 __ _ _ _ 3.3 : \[table:parameters\] [Parameters for phase and_amplification_spectra fits_and observed cutoff frequency]{} [The_first column_indicates the_pair of lines_used in_the analysis,_second to_fourth columns are the temperature ($T$), difference_in_formation heights ($\Delta z$), and characteristic timescale of the_radiative_losses_($\tau_{\rm_R}$) retrieved from_the fit of_the phase and_amplification spectra to_a_model of linear wave propagation in_a gravitationally stratifed_atmosphere with radiative losses,_and the last column in the observationally measured cutoff frequency_($\omega_{\rm c}$).]{} Figure \[fig:dphase\_CaHe\] indicates that propagating waves go upwards in the umbral atmosphere. The examination of

Let ${\left\{ {X_t; t\geq 0} \right\}}$ be a continuous process, self-similar with index $h > 0$ such that for every $T>0$, $$\{X_{T-t} - X_{T}; t\in [0,T]\} \overset{\text(d)}{=}\{X_t; t\in [0,T]\}.$$ Moreover assume that for every $\theta>0$, $$\EE\Big[\exp{\Big(\theta \max_{t\in[0,1]} |X_t|\Big)} \Big] <+\infty.$$ Then, as $T\rightarrow +\infty$, $$\EE\left[\left( \int_0^T e^{X_t}\ dt\right)^{-1} \right] = h T^{-(1-h)} \Big( \EE\Big[\max_{t\in[0,1]} X_t \Big] +o(1) \Big).$$ Before applying this result to our random process $\Delta$ we first establish some useful facts concerning it. First we show that the process satisfies $$\{\Delta_{T-t} - \Delta_{T}; t\in [0,T]\} \overset{\text(d)}{=}\{\Delta_t; t\in [0,T]\}$$ and has stationary increments (Lemma \[statincr\]). Next we provide an exponential upper bound for the tail of $\Delta_1$, from which we deduce the Kolmogorov-Centsov continuity criterion (Lemma \[PropDelta1\]). Finally we show that the process satisfies a maximal inequality (Lemma \[PropInegmax\]). \[statincr\] Assume $(H1)$ and $(H3)$. The process $\Delta$ satisfies for every $T>0$, $$\label{RenvTemps} \{\Delta_{T-t} - \Delta_{T}; t\in [0,T]\} \overset{\text(d)}{=}\{\Delta_t; t\in [0,T]\} \,.$$ Therefore, it has stationary increments, i.e. for every $s>0$, $\{\Delta_{t+s} - \Delta_{s}; t\geq 0\} \overset{\text(d)}{=}\{\Delta_t; t\geq 0\}$. Conditionally to $\{Y(t); t \geq 0\}$, the process $\{\Delta_{

Let $ { \left\ { { X_t; t\geq 0 } \right\}}$ be a continuous process, self - similar with index $ planck's constant > 0 $ such that for every $ T>0 $, $ $ \{X_{T - metric ton } - X_{T }; t\in [ 0,T]\ } \overset{\text(d)}{=}\{X_t; t\in [ 0,T]\}.$$ Moreover assume that for every $ \theta>0 $, $ $ \EE\Big[\exp{\Big(\theta \max_{t\in[0,1 ] } |X_t|\Big) } \Big ] < + \infty.$$ Then, as $ T\rightarrow + \infty$, $ $ \EE\left[\left (\int_0^T e^{X_t}\ dt\right)^{-1 } \right ] = h T^{-(1 - h) } \Big (\EE\Big[\max_{t\in[0,1 ] } X_t \Big ] + o(1) \Big).$$ Before applying this solution to our random process $ \Delta$ we first build some utilitarian facts concerning it. First we show that the process meet $ $ \{\Delta_{T - t } - \Delta_{T }; t\in [ 0,T]\ } \overset{\text(d)}{=}\{\Delta_t; t\in [ 0,T]\}$$ and has stationary increase (Lemma \[statincr\ ]). Next we provide an exponential upper bound for the tail of $ \Delta_1 $, from which we deduce the Kolmogorov - Centsov continuity criterion (Lemma \[PropDelta1\ ]). Finally we prove that the process satisfies a maximal inequality (Lemma \[PropInegmax\ ]). \[statincr\ ] Assume $ (H1)$ and $ (H3)$. The process $ \Delta$ satisfies for every $ T>0 $, $ $ \label{RenvTemps } \{\Delta_{T - triiodothyronine } - \Delta_{T }; t\in [ 0,T]\ } \overset{\text(d)}{=}\{\Delta_t; t\in [ 0,T]\ } \,.$$ Therefore, it has stationary increments, i.e. for every $ s>0 $, $ \{\Delta_{t+s } - \Delta_{s }; t\geq 0\ } \overset{\text(d)}{=}\{\Delta_t; t\geq 0\}$. Conditionally to $ \{Y(t); t \geq 0\}$, the process $ \{\Delta _ {

Leh ${\left\{ {X_t; t\geq 0} \right\}}$ be a continuous ptoxess, sxlf-simimar with index $h > 0$ such that for evecy $T>0$, $$\{X_{T-t} - X_{T}; t\in [0,T]\} \overset{\text(d)}{=}\{B_t; t\in [0,T]\}.$$ Loreover assnme that for evecg $\theta>0$, $$\EE\Big[\swp{\Big(\chxta \max_{t\in[0,1]} |X_t|\Bin)} \Big] <+\infty.$$ Then, as $T\righdafrlw +\infty$, $$\EE\left[\left( \int_0^T e^{X_t}\ dt\righe)^{-1} \right] = j T^{-(1-h)} \Big( \EE\Big[\iax_{t\pn[0,1]} X_t \Gpg] +o(1) \Big).$$ Before applying this reault to our random ptocess $\Delta$ we first estahlisj some useful factd concernint it. Dirst we shod that the process satjsfies $$\{\Delta_{T-t} - \Delta_{T}; t\in [0,T]\} \ovdrset{\cext(d)}{=}\{\Delta_t; t\un [0,H]\}$$ and has stetionagy increments (Lemma \[sdatincr\]). Next we provice en ezponential upper bounv for the tail of $\Dejta_1$, from fhnch we deduce the Kolnotorov-Wentvov zintknujtb cditeriln (Memma \[PropSelta1\]). Finaloy we show that the pwicess satisfiss a mwxymal inequality (Lemma \[PropInegmax\]). \[statinbr\] Aasume $(H1)$ and $(H3)$. The procews $\Delta$ satisfies fot every $T>0$, $$\jabel{RenvTemps} \{\Delta_{T-t} - \Delta_{T}; t\in [0,T]\} \overset{\text(d)}{=}\{\Gelta_v; g\in [0,T]\} \,.$$ Rhfrefore, it has stationary increments, i.e. for edsru $x>0$, $\{\Delta_{t+s} - \Delba_{s}; t\geq 0\} \overset{\trxh(d)}{=}\{\Cglta_t; t\geq 0\}$. Conaitioncmlg to $\{Y(t); t \geq 0\}$, the procesf $\{\Delra_{

Let ${\left\{ {X_t; t\geq 0} \right\}}$ be process, with index > 0$ such - t\in [0,T]\} \overset{\text(d)}{=}\{X_t; [0,T]\}.$$ Moreover assume for every $\theta>0$, $$\EE\Big[\exp{\Big(\theta \max_{t\in[0,1]} |X_t|\Big)} <+\infty.$$ Then, as $T\rightarrow +\infty$, $$\EE\left[\left( \int_0^T e^{X_t}\ dt\right)^{-1} \right] = h T^{-(1-h)} \EE\Big[\max_{t\in[0,1]} X_t \Big] +o(1) \Big).$$ Before applying this result to our random process we establish useful concerning it. First we show that the process satisfies $$\{\Delta_{T-t} - \Delta_{T}; t\in [0,T]\} \overset{\text(d)}{=}\{\Delta_t; t\in and has stationary increments (Lemma \[statincr\]). Next we an exponential upper bound the tail of $\Delta_1$, from we the Kolmogorov-Centsov criterion \[PropDelta1\]). we show that process satisfies a maximal inequality (Lemma \[PropInegmax\]). \[statincr\] Assume $(H1)$ and $(H3)$. The process $\Delta$ satisfies for $T>0$, $$\label{RenvTemps} \Delta_{T}; t\in \overset{\text(d)}{=}\{\Delta_t; [0,T]\} Therefore, it has i.e. for every $s>0$, $\{\Delta_{t+s} - \overset{\text(d)}{=}\{\Delta_t; t\geq 0\}$. Conditionally to $\{Y(t); t \geq the process

Let ${\left\{ {X_t; t\geq 0} \right\}}$ be a contInuous procEss, seLf-sImiLaR witH indEx $h > 0$ such that for EVery $t>0$, $$\{X_{T-t} - X_{T}; t\in [0,T]\} \overset{\text(d)}{=}\{x_t; t\in [0,t]\}.$$ MOReovER aSsume That for EVeRY $\TheTa>0$, $$\eE\big[\ExP{\biG(\thetA \maX_{t\in[0,1]} |X_t|\BIg)} \Big] <+\infty.$$ theN, aS $T\rightarrow +\INfTy$, $$\EE\left[\leFt( \iNt_0^T e^{X_t}\ dt\righT)^{-1} \riGht] = h T^{-(1-h)} \biG( \EE\bIg[\max_{T\in[0,1]} x_t \Big] +O(1) \Big).$$ BeFOre appLying this ReSUlt to oUR random PROcEss $\DElta$ we first establISh SOme useful facts ConcerNiNG iT. fIrsT we Show that thE pRocesS SatisfiES $$\{\DELTA_{T-t} - \dElta_{T}; t\in [0,T]\} \overSet{\text(d)}{=}\{\DelTA_t; t\In [0,T]\}$$ and HaS stATionarY incrEmENts (lemma \[statinCr\]). NeXt we proviDe an exPOnentiaL Upper boUnd for The TaiL of $\DELtA_1$, fRom WhICh wE DeDucE The kolmogorOv-ceNtsov ContINUITy crIteRion (lemma \[propDelta1\]). FinaLly We shOW thAt the ProceSs saTiSfies A maximAl ineQuAlity (Lemma \[PropINegmAx\]). \[statincR\] AsSuMe $(H1)$ AnD $(H3)$. The PRocess $\delTa$ sAtisfieS for eveRY $T>0$, $$\lAbEL{rEnVTemps} \{\Delta_{T-t} - \Delta_{t}; t\IN [0,t]\} \oVerset{\teXt(d)}{=}\{\DelTA_t; T\iN [0,t]\} \,.$$ TherefoRe, It hAs stATIonarY incREmEnts, i.e. foR every $S>0$, $\{\deLtA_{t+s} - \DeltA_{s}; T\geq 0\} \ovErSet{\TexT(d)}{=}\{\DelTA_t; t\gEq 0\}$. CondItionallY to $\{Y(t); T \Geq 0\}$, the process $\{\DELta_{

Let ${\left\{ {X_t; t\geq 0} \right \}}$bea c on tinu ousprocess, self- s imil ar with index $h > 0$suchth a t fo r e very$T>0$,$ $\ { X _{T -t }- X _{ T }; t\in [0 ,T]\} \ overset{\t ext (d )}{=}\{X_t;t \i n [0,T]\}. $$Moreover ass ume thatfo r e v ery $ \th eta>0 $, $$\ E E\Big[ \exp{\Big (\ t heta \ m ax_{t\i n [ 0, 1]}|X_t|\Big)} \Big] <+ \ infty.$$ Then, as $T \r i gh t a rro w + \infty$, $ $\ EE\le f t[\left ( \ i n t _0^ T e^{X_t}\ dt\ right)^{-1} \ri ght] = h T^ { -(1-h) } \Bi g( \EE \Big[\max_{ t\in [0,1]} X_ t \Big ] +o(1)\ Big).$$ Befo reapp lyin g t hi s r es u ltt oour ran dom proc es s$\Del ta$w e f irst es tabl ish s ome useful fa cts con c ern ing i t. Fi rstwe show thatthe p ro cess satisfies$$\{ \Delta_{T -t} - \D el ta_{T } ; t\in [0 ,T] \} \ove rset{\t e xt( d) } { = }\ {\Delta_t; t\in [0 ,T ] \ }$ $ and ha s stat i on ar y increme nt s ( Lemm a \[sta tinc r \] ). Nextwe pro v id ean expo ne ntialup per bo und f o r th e tail of $\De lta_1 $ , from which w e deduce the K o lm o g or o v-Ce nts ov continui ty c r iter ion( Le mma \[Pro pDelt a1 \ ]) . Finally we show th at the p roces s satisfies a maximal i n e q uality ( Lemm a \ [ PropInegmax\]) . \[ statincr\] Assume $ (H1)$ and $(H 3)$. Thep r ocess $\ Del ta$ sa tis f i es for every $T > 0 $, $ $\ label{R env Temps} \{ \De lta _{T -t } - \Delt a_{T}; t \i n[0 ,T ]\} \ove r set{\tex t( d)} {= }\{ \Delt a _t; t\ in [0 ,T]\ }\, . $$Therefo r e, i t ha sst atio nar yincre ment s , i .e. for every $s >0$ , $\{ \D el ta_{t+s } - \Delta_{s }; t\geq 0\} \ ove rset{\ t e xt(d)}{= }\{\Delta_t; t\geq 0\}$ . Condi tio nally to$\{Y(t);t \ geq 0\ }$, the pr ocess$\{\D el ta_ {

Let_${\left\{ {X_t;_t\geq 0} \right\}}$ be_a continuous_process,_self-similar with_index_$h > 0$_such that for_every $T>0$, $$\{X_{T-t} -_X_{T}; t\in [0,T]\}_\overset{\text(d)}{=}\{X_t;_t\in [0,T]\}.$$ Moreover assume that for every $\theta>0$, $$\EE\Big[\exp{\Big(\theta \max_{t\in[0,1]} |X_t|\Big)} \Big] <+\infty.$$ Then,_as_$T\rightarrow +\infty$,_$$\EE\left[\left(_\int_0^T_e^{X_t}\ dt\right)^{-1} \right] = h_T^{-(1-h)} \Big( \EE\Big[\max_{t\in[0,1]} X_t \Big]_+o(1) \Big).$$ Before_applying this result to our random process $\Delta$_we_first establish some_useful facts concerning it. First we show that the_process satisfies $$\{\Delta_{T-t} - \Delta_{T}; t\in_[0,T]\} \overset{\text(d)}{=}\{\Delta_t; t\in_[0,T]\}$$_and_has stationary increments (Lemma_\[statincr\]). Next we provide an exponential_upper bound for the tail of_$\Delta_1$, from which we deduce the Kolmogorov-Centsov_continuity criterion (Lemma \[PropDelta1\]). Finally we_show that the process satisfies_a maximal_inequality (Lemma \[PropInegmax\]). \[statincr\] Assume $(H1)$_and $(H3)$. The_process $\Delta$_satisfies for every_$T>0$, $$\label{RenvTemps} \{\Delta_{T-t} - \Delta_{T}; t\in_[0,T]\} \overset{\text(d)}{=}\{\Delta_t; t\in_[0,T]\} \,.$$ Therefore, it has stationary_increments,_i.e. for every_$s>0$,_$\{\Delta_{t+s}_- \Delta_{s};_t\geq 0\} \overset{\text(d)}{=}\{\Delta_t;_t\geq_0\}$. Conditionally to_$\{Y(t);_t \geq 0\}$, the process $\{\Delta_{

. V. 69(2). P. 145-204. Prokhorov Yu. G. On algebraic threefolds whose hyperplane sections are Enriques surfaces // Russ. Acad. Sci. Sb. Math. 1995. V. 186 (9). P. 1341-1352. Prokhorov Yu. G. On the degree of Fano threefolds with canonical Gorenstein singularities // Russ. Acad. Sci. Sb. Math. 2005. V. 196(1). P. 81-122. Prokhorov Yu. G. On Fano-Enriques threefolds // Russ. Acad. Sci. Sb. Math. 2007. V. 198(4). P. 559-574. Reid M. Canonical 3-folds // Algebraic Geometry, Angers. 1979. P. 273-310. Reid M. Chapters on algebraic surfaces // Complex algebraic geometry (Park City. Ut. 1993) P. 3-159. Reid M. Projective morphisms according to Kawamata // Preprint Univ. of Warwick. 1983. Sano T. Classification of non-Gorenstein $\mathbb{Q}$-Fano $3$-folds of index $1$ // J. Math. Soc. Japan. V. 47(2). 1995. P. 369–380. [^1]: The work was partially supported by RFFI grant No. 08-01-00395-a and grant N.Sh.-1987.2008.1. [^2]: We assume that all varieties are projective, normal and defined over $\mathbb{C}$. --- author: - 'Shaolin Ji[^1]' - 'Shuzhen Yang[^2]' title: 'Non-Markovian Fully Coupled Forward-Backward Stochastic Systems and Classical Solutions of Path-dependent PDEs[^3]' --- **Abstract**. This paper explores the relationship between non-Markovian fully coupled forward-backward stochastic systems and [path-dependent PDEs. The definition of classical solution for the path-dependent PDE is given within the framework of functional Itô calculus. Under mild hypotheses, we prove that the ]{}forward-backward stochastic system provides the unique classical solution to the path-dependent PDE. [**Keywords**: Functional Itô calculus, Non-Markovian

. V. 69(2). P. 145 - 204. Prokhorov Yu. G. On algebraic threefolds whose hyperplane sections are Enriques surfaces // Russ. Acad. Sci. Sb. Math. 1995. V. 186 (9). P. 1341 - 1352. Prokhorov Yu. G. On the degree of Fano threefolds with basic Gorenstein singularity // Russ. Acad. Sci. Sb. Math. 2005. V. 196(1). P. 81 - 122. Prokhorov Yu. G. On Fano - Enriques threefolds // Russ. Acad. Sci. Sb. Math. 2007. V. 198(4). P. 559 - 574. Reid M. Canonical 3 - folds // Algebraic Geometry, Angers. 1979. P. 273 - 310. Reid M. Chapters on algebraic surface // Complex algebraic geometry (Park City. Ut. 1993) P. 3 - 159. Reid M. Projective morphisms according to Kawamata // Preprint Univ. of Warwick. 1983. Sano T. Classification of non - Gorenstein $ \mathbb{Q}$-Fano $ 3$-folds of index $ 1 $ // J. Math. Soc. Japan. V. 47(2). 1995. P. 369–380. [ ^1 ]: The oeuvre was partially patronize by RFFI grant No. 08 - 01 - 00395 - a and grant N.Sh.-1987.2008.1. [ ^2 ]: We assume that all varieties are projective, normal and define over $ \mathbb{C}$. --- author: -' Shaolin Ji[^1 ]' -' Shuzhen Yang[^2 ]' title:' Non - Markovian in full Coupled Forward - Backward Stochastic Systems and Classical Solutions of way - dependent PDEs[^3 ]' --- * * Abstract * *. This paper explore the relationship between non - Markovian fully match forward - backward stochastic systems and [ path - subject PDEs. The definition of classical solution for the path - dependent PDE is given within the model of functional Itô tartar. Under mild hypotheses, we prove that the ] { } forward - backward stochastic organization provides the unique classical solution to the path - dependent PDE. [ * * Keywords * *: Functional Itô tartar, Non - Markovian

. V. 69(2). P. 145-204. Prokhorov Yu. G. On algtbraic threefolds whose iyperplzne sectkons are Enriques surfaces // Cuss. Acad. Sci. Sb. Math. 1995. V. 186 (9). P. 1341-1352. Prukhorov Yl. G. On thw dejree of Fano thrxsfolds with czkoniccl Gorenstein sikgularities // Russ. Acad. Sch. Rb. Math. 2005. V. 196(1). P. 81-122. Prokhorov Yu. G. On Fano-Egriques tjreefolds // Rusf. Acsq. Scj. Sb. Math. 2007. V. 198(4). P. 559-574. Reid M. Canonical 3-fklds // Angebraic Geomrtry, Angers. 1979. P. 273-310. Reid M. Chaphers on algebraic surfwces // Complgs ajtebraic geomdtry (Park Bnty. Ut. 1993) P. 3-159. Rgid M. Projective morphisms accoraing co Kawamata // Preotint Univ. of Warwpck. 1983. Sano T. Classificadion of non-Gorenstein $\mavhbb{W}$-Fano $3$-folds of index $1$ // J. Math. Soc. Japan. V. 47(2). 1995. P. 369–380. [^1]: Tve work was partialoy suppmrteg by EFFK gdait Ho. 08-01-00395-a ajd jrant N.Sh.-1987.2008.1. [^2]: Ws assume thqt all varieties art pwijective, normzl and dqfined over $\mathbb{C}$. --- author: - 'Shaolin Ji[^1]' - 'Svuzgen Yang[^2]' title: 'Non-Markovuan Fully Coupled Foraard-Backwwrd Stochastic Systems and Classical Solutions of Path-veoenbckt PAWs[^3]' --- **Wbstract**. This paper explores the relationship geuwevn non-Markovian fmlly coupled forwatd-havhward stochasjic syscsma and [path-dependenh PDEs. Jhe dedinition jf ckassical solution for the pqth-dependent PDE is given withiu the framewurk pf fumctional Itô calculus. Unber mimd hypothesfs, we probd that the ]{}forwafd-bscnward stochastic system prjvides thx uniaue clarsicsl solttion to tje pabv-dependent PDE. [**Keywlrds**: Yuncthonal Itô cwlculus, Non-Markovian

. V. 69(2). P. 145-204. Prokhorov Yu. algebraic whose hyperplane are Enriques surfaces Math. V. 186 (9). 1341-1352. Prokhorov Yu. On the degree of Fano threefolds canonical Gorenstein singularities // Russ. Acad. Sci. Sb. Math. 2005. V. 196(1). P. Prokhorov Yu. G. On Fano-Enriques threefolds // Russ. Acad. Sci. Sb. Math. 2007. 198(4). 559-574. M. 3-folds // Algebraic Geometry, Angers. 1979. P. 273-310. Reid M. Chapters on algebraic surfaces // Complex geometry (Park City. Ut. 1993) P. 3-159. Reid Projective morphisms according to // Preprint Univ. of Warwick. Sano Classification of $\mathbb{Q}$-Fano of $1$ // J. Soc. Japan. V. 47(2). 1995. P. 369–380. [^1]: The work was partially supported by RFFI grant No. and grant We assume all are normal and defined --- author: - 'Shaolin Ji[^1]' - 'Non-Markovian Fully Coupled Forward-Backward Stochastic Systems and Classical of Path-dependent --- **Abstract**. This paper explores the between non-Markovian fully coupled forward-backward stochastic systems and PDEs. The definition of classical solution for the path-dependent PDE is given within the framework Itô calculus. Under mild we prove that ]{}forward-backward system the classical solution the path-dependent PDE. [**Keywords**: Functional Itô calculus, Non-Markovian

. V. 69(2). P. 145-204. Prokhorov Yu. G. On algebraic Threefolds Whose HypErpLaNe seCtioNs are Enriques sURfacEs // Russ. Acad. Sci. Sb. Math. 1995. V. 186 (9). P. 1341-1352. PRokhoRoV yu. G. ON ThE degrEe of FanO ThREEfoLdS wIth CaNOnIcal GOreNstein sIngularitiEs // RUsS. Acad. Sci. Sb. MaTH. 2005. V. 196(1). p. 81-122. Prokhorov yu. G. on Fano-EnriquEs tHreefoLdS // RuSS. Acad. sci. sb. MatH. 2007. V. 198(4). P. 559-574. ReiD m. CanonIcal 3-folds // alGEbraic gEometry, aNGeRs. 1979. P. 273-310. REid M. Chapters on algEBrAIc surfaces // CompLex algEbRAiC GEomEtrY (Park City. UT. 1993) P. 3-159. reid M. pRojectiVE mORPHisMS according to KAwamata // PrepRInt univ. of waRwiCK. 1983. Sano T. classIfICatIon of non-GorEnstEin $\mathbb{q}$-Fano $3$-fOLds of inDEx $1$ // J. Math. soc. JapAn. V. 47(2). 1995. p. 369–380. [^1]: ThE worK WaS pArtIaLLy sUPpOrtED by rFFI granT NO. 08-01-00395-a And grAnt N.sH.-1987.2008.1. [^2]: wE AssuMe tHat aLl varIeties are projEctIve, nORmaL and dEfineD oveR $\mAthbb{c}$. --- authoR: - 'ShaoLiN Ji[^1]' - 'Shuzhen Yang[^2]' tItle: 'non-MarkovIan fuLly coUpled fOrward-bacKwaRd StochAstic SySTemS aND cLaSsical Solutions of PAtH-DEpEndent PDes[^3]' --- **AbstRAcT**. THIs paper eXpLorEs thE RElatiOnshIP bEtween noN-MarkoVIaN fUlly couPlEd forwArD-baCkwArd stOChasTic sysTems and [pAth-dePEndent PDEs. The dEFinition of claSSiCAL sOLutiOn fOr the path-dePendENt PDe is gIVeN wiTHin thE framEwORk OF functional Itô calcuLuS. Under Mild hYpotheses, we prOve that the ]{}FORWard-backWard SToCHastic system prOvideS the unique CLassical SolutIon to the Path-depenDENt PDE. [**KeyWorDs**: FUncTioNAL ITô calculus, Non-mARkovIaN

. V. 69(2). P. 145-204. P rokhorov Y u. G. On al ge brai c th reefolds whose hype rplane sections are En rique ss urfa c es // R uss. Ac a d. S ci. S b. Ma th . 1 995.V.186 (9) . P. 1341- 135 2. ProkhorovY u. G. On the de gree of Fano th reefol ds wi t h can oni cal G orenst e in sin gularitie s/ / Russ . Acad.S c i. Sb. Math. 2005. V. 1 9 6( 1 ). P. 81-122. Prokh or o vY u . G . O n Fano-Enr iq ues t h reefold s / / R uss . Acad. Sci. S b. Math. 20 0 7.V. 198 (4 ).P . 559- 574. R e idM. Canonica l 3- folds //Algebr a ic Geom e try, An gers.197 9.P. 2 7 3- 31 0. R e idM .Cha p ter s on alg eb ra ic su rfac e s / / Co mpl ex a lgebr aic geometry(Pa rk C i ty. Ut.1993) P.3- 159. ReidM. Pr oj ective morphism s ac cording t o K aw ama ta // P r eprint Un iv. of War wick. 1 9 83. S a n oT. Classificationof n on -Gorenst ein $\ m at hb b {Q}$-Fan o$3$ -fol d s of i ndex $1 $ // J.Math.S oc .Japan.V. 47(2) .199 5.P. 36 9 –380 . [^1 ]: The w ork w a s partially su p ported by RFF I g r a nt No.08- 01-00395-aandg rant N.S h .- 198 7 .2008 .1. [^ 2 ]: We assume that allva rietie s are projective,normal and d e fined ov er $ \ ma t hbb{C}$. ---autho r: - 'Shao l in Ji[^1 ]' -'Shuzhen Yang[^2] ' title: ' Non -Ma rko via n Fu lly Coupled F o r ward -B ackward St ochasti c S yst ems an dClassical Solutio ns o fPa th- depen d ent PDEs [^ 3]' - -- **Ab s tract* *. Th is p ap er exp lores t h er e lati on sh ip b etw ee n non -Mar k ovi an full y coupled fo r ward -b ac kward s tochastic sys te ms and [pa th -de penden t PDEs. Th e definition of classic a l solut ion forthepath-depe nde nt PDE is givenwithin thefr ame w o rk of f un cti on al Itô cal c u lus . Und er mil d hypot heses, we prove th a t t he ]{}forward -ba ckwa r d s toc h as t icsy s tem p rovides the uni que classi ca l s olution to the p ath-dep endentPDE.[**Keyw ords**: F unctional I tô c a l cul us, Non-Ma rkovian

. V._69(2). P._145-204. Prokhorov Yu. G. On_algebraic threefolds_whose_hyperplane sections_are_Enriques surfaces //_Russ. Acad. Sci._Sb. Math. 1995. V._186 (9). P._1341-1352. Prokhorov_Yu. G. On the degree of Fano threefolds with canonical Gorenstein singularities // Russ._Acad._Sci. Sb._Math._2005._V. 196(1). P. 81-122. Prokhorov Yu._G. On Fano-Enriques threefolds //_Russ. Acad._Sci. Sb. Math. 2007. V. 198(4). P. 559-574. Reid_M._Canonical 3-folds //_Algebraic Geometry, Angers. 1979. P. 273-310. Reid M. Chapters on_algebraic surfaces // Complex algebraic geometry_(Park City. Ut._1993)_P._3-159. Reid M. Projective morphisms_according to Kawamata // Preprint Univ._of Warwick. 1983. Sano T. Classification of_non-Gorenstein $\mathbb{Q}$-Fano $3$-folds of index $1$ //_J. Math. Soc. Japan. V. 47(2)._1995. P. 369–380. [^1]: The work_was partially_supported by RFFI grant No._08-01-00395-a and grant_N.Sh.-1987.2008.1. [^2]: We_assume that all_varieties are projective, normal and defined_over $\mathbb{C}$. --- author: -_'Shaolin Ji[^1]' - 'Shuzhen Yang[^2]' title: 'Non-Markovian Fully_Coupled_Forward-Backward Stochastic Systems_and_Classical_Solutions of_Path-dependent PDEs[^3]' --- **Abstract**. This_paper_explores the_relationship_between non-Markovian fully coupled forward-backward stochastic_systems_and [path-dependent PDEs. The definition of classical_solution for the path-dependent_PDE_is given within the_framework of functional Itô calculus._Under mild hypotheses, we prove that_the ]{}forward-backward_stochastic system_provides the unique classical solution to the path-dependent PDE. [**Keywords**: Functional Itô_calculus, Non-Markovian

is reported in Table \[maserfitparams\_67\]. The sub-synthesized-beam kinematic analysis facilitated by these fits will be discussed in §\[maserkin\] and §\[individualsources\]. For each EGO in our survey within the FWHP of the VLA primary beam at 6.7 GHz (67; 19 targeted EGOs plus 3 serendipitiously observed), Table \[67iweighted\_egos\] lists the properties of associated 6.7 GHz [CH$_3$OH]{} maser group(s) and the angular separation of the 6.7 GHz maser group(s) and 24 [$\mu$m]{} EGO counterpart. Only two EGOs are potentially associated with multiple 6.7 GHz maser groups: G11.92$-$0.61 and G49.42+0.33. The case of G49.42+0.33 is ambiguous: two spatially and kinematically distinct maser spots are detected, separated by [$\sim$]{}9[$^{\prime\prime}$]{} ([$\sim$]{}0.5 pc at 12.3 kpc, see also §\[g4942\]). The western maser spot (G49.416+0.326) is associated with the EGO, but lacks a discrete 24 [$\mu$m]{} counterpart; however, G49.42+0.33 is much more distant than the other EGOs in our sample, so the sensitivity of the MIPSGAL survey corresponds to an intrinsically brighter 24 [$\mu$m]{} upper limit. The eastern maser spot (G49.417+0.324) is associated with a 24 [$\mu$m]{} source, but may not be associated with the EGO. Table \[67iweighted\_egos\] quantifies the correlation of 6.7 GHz [CH$_3$OH]{} masers and 24 [$\mu$m]{} emission that is visually apparent in Figure \[3color\]. Excluding the ambiguous case of G49.42+0.33, of the 21 remaining EGOs in Table \[67iweighted\_egos\] (18 targeted plus 3 serendipitously observed), 2 do not have 6.7 GHz maser detections (G49.27$-$0.34 and G49.27$-$0.32) and 2 do not have discrete 24 [$\mu$m]{} counterparts (G10.29$-$0.13 and G28.28$-$0.36). Of the

is reported in Table   \[maserfitparams\_67\ ]. The sub - synthesized - beam kinematic analysis facilitated by these fit will be discourse in § \[maserkin\ ] and § \[individualsources\ ]. For each EGO in our survey within the FWHP of the VLA primary beam at 6.7 gigahertz (67; 19 targeted EGOs plus 3 serendipitiously observed), Table   \[67iweighted\_egos\ ] number the properties of associated 6.7 gigahertz [ CH$_3$OH ] { } maser group(s) and the angular interval of the 6.7 GHz maser group(s) and 24 [ $ \mu$m ] { } EGO counterpart. entirely two EGOs are potentially associated with multiple 6.7 gigahertz maser groups: G11.92$-$0.61 and G49.42 + 0.33. The case of G49.42 + 0.33 is ambiguous: two spatially and kinematically distinct maser spots are detected, separated by [ $ \sim$]{}9[$^{\prime\prime}$ ] { } ([ $ \sim$]{}0.5 personal computer at 12.3 kpc, see also § \[g4942\ ]). The western maser point (G49.416 + 0.326) is associated with the EGO, but lacks a discrete 24 [ $ \mu$m ] { } counterpart; however, G49.42 + 0.33 is a lot more distant than the early EGOs in our sample, so the sensitivity of the MIPSGAL survey corresponds to an intrinsically brighter 24 [ $ \mu$m ] { } upper limit. The easterly maser spot (G49.417 + 0.324) is associated with a 24 [ $ \mu$m ] { } source, but may not be associated with the EGO. Table   \[67iweighted\_egos\ ] quantifies the correlation of 6.7 GHz [ CH$_3$OH ] { } masers and 24 [ $ \mu$m ] { } emission that is visually apparent in Figure   \[3color\ ]. Excluding the ambiguous case of G49.42 + 0.33, of the 21 remaining EGOs in Table   \[67iweighted\_egos\ ] (18 targeted plus 3 serendipitously observed), 2 do not have 6.7 gigahertz maser detection (G49.27$-$0.34 and G49.27$-$0.32) and 2 do not have discrete 24 [ $ \mu$m ] { } counterpart (G10.29$-$0.13 and G28.28$-$0.36). Of the

is reported in Table \[maserfltparams\_67\]. The sub-synthesmzed-beaj kinemagic analysis facilitated by vhesw fitw will be discussed in §\[maserkin\] and §\[induvidnalsources\]. For eedh EGO lu our durvzy within the FWMP of the VNA primary beak xt 6.7 GHz (67; 19 targeted EGOs plus 3 serendi[itiousky observed), Tablg \[67iweibrted\_sgos\] lists the properties of assocjated 6.7 JHz [CH$_3$OH]{} maser broup(s) and the angular sepwratlon of the 6.7 GHz mader group(s) qnd 24 [$\nu$m]{} EGO coungerpart. Only two EGOs zre potentially associated with multnple 6.7 GHz mqswr htoups: G11.92$-$0.61 and J49.42+0.33. The base of G49.42+0.33 is ambiguouv: two slatially and klnemavicaoly distinct maser spmts are detected, sgparated bf [$\aim$]{}9[$^{\prime\prime}$]{} ([$\sim$]{}0.5 px at 12.3 kpc, see qlsu §\[g4942\]). Tie sesterj meser spot (G49.416+0.326) is associared with the EGO, buu lwbls a discrets 24 [$\mu$m]{} cjunterpart; however, G49.42+0.33 is much more distatt fhan the other EGOs in iur sample, so the senditivity jf the MIPSGAL survey corresponds to an intrinsicdlly urkghuev 24 [$\nu$l]{} upper limit. The eastern maser spot (G49.417+0.324) is assjdistvd with a 24 [$\mu$m]{} somrce, but may not br wsxjciated with jhe EGO. Czbme \[67iweighted\_egos\] quwntifief the correlatyon pf 6.7 GHz [CH$_3$OH]{} masers and 24 [$\mu$n]{} emission tkat is visually apparznt in Figurz \[3color\]. Excliding the ambiguous casz of G49.42+0.33, of the 21 relaining EFUs in Table \[67iweigftec\_ecos\] (18 targeted plus 3 serendypitously obszrved), 2 du noj have 6.7 GHz maser detegdions (G49.27$-$0.34 and G49.27$-$0.32) and 2 do nlt have discgete 24 [$\mu$m]{} counterparts (G10.29$-$0.13 and G28.28$-$0.36). Of the

is reported in Table \[maserfitparams\_67\]. The sub-synthesized-beam facilitated these fits be discussed in EGO our survey within FWHP of the primary beam at 6.7 GHz (67; targeted EGOs plus 3 serendipitiously observed), Table \[67iweighted\_egos\] lists the properties of associated GHz [CH$_3$OH]{} maser group(s) and the angular separation of the 6.7 GHz maser and [$\mu$m]{} counterpart. two EGOs are potentially associated with multiple 6.7 GHz maser groups: G11.92$-$0.61 and G49.42+0.33. The case G49.42+0.33 is ambiguous: two spatially and kinematically distinct spots are detected, separated [$\sim$]{}9[$^{\prime\prime}$]{} ([$\sim$]{}0.5 pc at 12.3 see §\[g4942\]). The maser (G49.416+0.326) associated with the but lacks a discrete 24 [$\mu$m]{} counterpart; however, G49.42+0.33 is much more distant than the other EGOs our sample, sensitivity of MIPSGAL corresponds an intrinsically brighter upper limit. The eastern maser spot with a 24 [$\mu$m]{} source, but may not associated with EGO. Table \[67iweighted\_egos\] quantifies the correlation 6.7 GHz [CH$_3$OH]{} masers and 24 [$\mu$m]{} emission is visually apparent in Figure \[3color\]. Excluding the ambiguous case of G49.42+0.33, of the 21 in Table \[67iweighted\_egos\] (18 plus 3 serendipitously 2 not 6.7 maser detections and G49.27$-$0.32) and 2 do not have discrete 24 [$\mu$m]{} counterparts and G28.28$-$0.36). Of the

is reported in Table \[maserfitParams\_67\]. The sUb-synTheSizEd-Beam KineMatic analysis fACiliTated by these fits will be DiscuSsED in §\[mASeRkin\] aNd §\[indivIDuALSouRcEs\]. for EaCH EgO in oUr sUrvey wiThin the FWHp of ThE VLA primary bEAm At 6.7 GHz (67; 19 targeTed eGOs plus 3 sereNdiPitiouSlY obSErved), tabLe \[67iweIghted\_EGos\] lisTs the propErTIes of aSSociateD 6.7 ghz [cH$_3$OH]{} Maser group(s) and the ANgULar separation oF the 6.7 GHZ mASeR GRouP(s) aNd 24 [$\mu$m]{} EGO coUnTerpaRT. Only twO eGoS ARe pOTentially assoCiated with mULtiPle 6.7 GHz MaSer GRoups: G11.92$-$0.61 And G49.42+0.33. THe CAse Of G49.42+0.33 is ambiguOus: tWo spatialLy and kINematicALly distInct maSer SpoTs arE DeTeCteD, sEParATeD by [$\SIm$]{}9[$^{\pRime\primE}$]{} ([$\sIm$]{}0.5 Pc at 12.3 kPc, seE ALSO §\[g4942\]). ThE weSterN maseR spot (G49.416+0.326) is assocIatEd wiTH thE EGO, bUt lacKs a dIsCrete 24 [$\Mu$m]{} couNterpArT; however, G49.42+0.33 is much More Distant thAn tHe OthEr eGOs iN Our samPle, So tHe sensiTivity oF The mIpsgaL Survey corresponds tO aN INtRinsicalLy brigHTeR 24 [$\mU$M]{} upper liMiT. ThE easTERn masEr spOT (G49.417+0.324) Is associAted wiTH a 24 [$\Mu$M]{} source, BuT may noT bE asSocIated WIth tHe EGO. TAble \[67iweiGhted\_EGos\] quantifies tHE correlation oF 6.7 gHZ [ch$_3$Oh]{} MaseRs aNd 24 [$\mu$m]{} emissiOn thAT is vIsuaLLy AppARent iN FiguRe \[3COlOR\]. Excluding the ambiguOuS case oF G49.42+0.33, of tHe 21 remaining EGos in Table \[67iWEIGhted\_egoS\] (18 tarGEtED plus 3 serendipiTouslY observed), 2 dO Not have 6.7 Ghz masEr detectIons (G49.27$-$0.34 and G49.27$-$0.32) AND 2 do not haVe dIscRetE 24 [$\mu$M]{} COuNterparts (G10.29$-$0.13 and g28.28$-$0.36). oF the

is reported in Table \[ma serfitpara ms\_6 7\] . T he sub -syn thesized-beamk inem atic analysis facilita ted b yt hese fi ts wi ll be d i sc u s sed i n§\[ ma s er kin\] an d §\[in dividualso urc es \]. For each EG O in our s urv ey within th e F WHP of t heV LA pr ima ry be am at6 .7 GHz (67; 19ta r getedE GOs plu s 3sere ndipitiously obse r ve d ), Table \[67i weight ed \ _e g o s\] li sts the pr op ertie s of ass o ci a t e d 6 . 7 GHz [CH$_3$ OH]{} maser gro up(s)an d t h e angu lar s ep a rat ion of the6.7GHz maser group ( s) and2 4 [$\mu $m]{}EGO co unte r pa rt . O nl y tw o E GOs are potenti al ly asso ciat e d w ithmul tipl e 6.7 GHz maser gr oup s: G 1 1.9 2$-$0 .61 a nd G 49 .42+0 .33. T he ca se of G49.42+0.33 isambiguous : t wo sp at ially and ki nem ati cally d istinct mas er s p ot s are detected, se pa r a te d by [$\ sim$]{ } 9[ $^ { \prime\p ri me} $]{} ( [$\si m$]{ } 0. 5 pc at12.3 k p c, s ee also § \[g494 2\ ]). Th e wes t ernmaserspot (G4 9.416 + 0.326) is asso c iated with th e E G O ,b ut l ack s a discret e 24 [$\m u$m] { }cou n terpa rt; h ow e ve r , G49.42+0.33 is mu ch moredista nt than the o ther EGOsi n our samp le,s ot he sensitivity of t he MIPSGAL survey c orres ponds to an intri n s ically b rig hte r 2 4 [ $ \ mu $m]{} upper l i m it.Th e easte rnmaser s pot (G 49. 417 +0 .324) isassociat ed w it ha 2 4 [$\ m u$m]{} s ou rce ,but mayn ot beassoc iate dwi t h t he EGO. T a b le \ [6 7i weig hte d\ _egos \] q u ant ifies t he correl ati o n of 6 .7 GHz [C H$_3$OH]{} ma se rs and 24[$ \mu $m]{}e m ission t hat is visually apparen t in Fig ure  \[3c olor \]. Exclu din g theamb i guouscase o f G49 .4 2+0 . 3 3, of t he 21 r emaining E G O s i n Tab le  \[6 7iweigh ted\_egos\] (18 ta r get ed plus 3 ser end ipit o u sl y o b se r ved ), 2 d o not have 6.7 GH z maser de te c ti ons (G49.2 7 $-$ 0. 34 andG49.27$ -$0.3 2 ) and 2 do not h ave discr et e 24 [ $\m u$m]{} cou nterpart s (G10.29 $ -$0.1 3 a nd G2 8.2 8$-$0. 36 ).Of th e

is_reported in_Table \[maserfitparams\_67\]. The sub-synthesized-beam kinematic_analysis facilitated_by_these fits_will_be discussed in_§\[maserkin\] and §\[individualsources\]._For each EGO in_our survey within_the_FWHP of the VLA primary beam at 6.7 GHz (67; 19 targeted EGOs plus_3_serendipitiously observed),_Table \[67iweighted\_egos\]_lists_the properties of associated 6.7_GHz [CH$_3$OH]{} maser group(s) and_the angular_separation of the 6.7 GHz maser group(s) and_24_[$\mu$m]{} EGO counterpart._Only two EGOs are potentially associated with multiple 6.7_GHz maser groups: G11.92$-$0.61 and G49.42+0.33._The case of_G49.42+0.33_is_ambiguous: two spatially and_kinematically distinct maser spots are detected,_separated by [$\sim$]{}9[$^{\prime\prime}$]{} ([$\sim$]{}0.5 pc at_12.3 kpc, see also §\[g4942\]). The western_maser spot (G49.416+0.326) is associated with_the EGO, but lacks a_discrete 24_[$\mu$m]{} counterpart; however, G49.42+0.33 is_much more distant_than the_other EGOs in_our sample, so the sensitivity of_the MIPSGAL survey_corresponds to an intrinsically brighter 24_[$\mu$m]{}_upper limit. The_eastern_maser_spot (G49.417+0.324)_is associated with_a_24 [$\mu$m]{}_source,_but may not be associated with_the_EGO. Table \[67iweighted\_egos\] quantifies the correlation of 6.7 GHz_[CH$_3$OH]{} masers and 24_[$\mu$m]{}_emission that is visually_apparent in Figure \[3color\]. Excluding the_ambiguous case of G49.42+0.33, of the_21 remaining_EGOs in_Table \[67iweighted\_egos\] (18 targeted plus 3 serendipitously observed), 2 do not have_6.7 GHz maser detections (G49.27$-$0.34 and_G49.27$-$0.32) and 2 do_not have_discrete_24 [$\mu$m]{} counterparts_(G10.29$-$0.13_and G28.28$-$0.36)._Of the

if there exists a function $\psi:[0,\infty[\to[0,\infty]$ with the property $\lim_{t\to 0}{\psi(t)}/{t}=0$ such that for every two points $x$ and $y$ in ${\ensuremath{\mathcal H}}$ and every $\lambda\in\ ]0,1[$ such that $(1-\lambda)x+\lambda y\in{\ensuremath{\operatorname{dom}}}f$, $$\label{uniform smoothness} f\big((1-\lambda)x+\lambda y\big)+\lambda(1-\lambda)\psi\big(\|x-y\|\big)\geq(1-\lambda)f(x)+\lambda f(y).$$ The smallest possible function $\psi$ satisfying  is called the *gauge of uniform smoothness* of $f$ and is defined by $$\psi_f(t)=\sup\bigg\{\frac{(1-\lambda)f(x)+\lambda f(y)-f((1-\lambda)x+\lambda y)}{\lambda(1-\lambda)}\ \bigg|\ \begin{array}{c} \lambda\in\ ]0,1[,\ x,y\in{\ensuremath{\mathcal H}},\ \|x-y\|=t, \\ \big((1-\lambda)x+\lambda y\big)\in{\ensuremath{\operatorname{dom}}}f \end{array}\ \bigg\}.$$ \[uniform convexity equivalent to uniform monotonicity\]*[@Zal Theorem 3.5.10, (i)$\Leftrightarrow$(v)]* Suppose that $f:{\ensuremath{\mathcal H}}\to{\ensuremath{\,\left]-\infty,+\infty\right]}}$ is proper, lower semicontinuous and convex. Then $f$ is uniformly convex if and only if $\partial f$ is uniformly monotone, in which case $\partial f$ is ${\ensuremath{\varphi}}$-uniformly monotone with ${\ensuremath{\varphi}}=2{\ensuremath{\varphi}}_f^{**}$. Consequently, we arrive at the following results: \[uniform convexity is dominant or recessive\] ***(uniform convexity is dominant w.r.t. $p_\mu$)*** Suppose that for each $i\in I$, $f_i:{\ensuremath{\mathcal H}}\to{\ensuremath{\,\left]-\infty,+\infty\right]}}$ is proper, lower semicontinuous, convex and there exists $i_0\

if there exists a function $ \psi:[0,\infty[\to[0,\infty]$ with the property $ \lim_{t\to 0}{\psi(t)}/{t}=0 $ such that for every two points $ x$ and $ y$ in $ { \ensuremath{\mathcal H}}$ and every $ \lambda\in\ ] 0,1[$ such that $ (1-\lambda)x+\lambda y\in{\ensuremath{\operatorname{dom}}}f$, $ $ \label{uniform smoothness } f\big((1-\lambda)x+\lambda y\big)+\lambda(1-\lambda)\psi\big(\|x - y\|\big)\geq(1-\lambda)f(x)+\lambda f(y).$$ The little potential function $ \psi$ satisfying   is called the * bore of uniform smoothness * of $ f$ and is define by $ $ \psi_f(t)=\sup\bigg\{\frac{(1-\lambda)f(x)+\lambda f(y)-f((1-\lambda)x+\lambda y)}{\lambda(1-\lambda)}\ \bigg|\ \begin{array}{c } \lambda\in\ ] 0,1[,\ x, y\in{\ensuremath{\mathcal H}},\ \|x - y\|=t, \\ \big((1-\lambda)x+\lambda y\big)\in{\ensuremath{\operatorname{dom}}}f \end{array}\ \bigg\}.$$ \[uniform convexity equivalent to uniform monotonicity\]*[@Zal Theorem 3.5.10, (i)$\Leftrightarrow$(v) ] * presuppose that $ f:{\ensuremath{\mathcal H}}\to{\ensuremath{\,\left]-\infty,+\infty\right]}}$ is proper, lower semicontinuous and convex. Then $ f$ is uniformly convex if and entirely if $ \partial f$ is uniformly monotone, in which event $ \partial f$ is $ { \ensuremath{\varphi}}$-uniformly monotone with $ { \ensuremath{\varphi}}=2{\ensuremath{\varphi}}_f^{**}$. Consequently, we arrive at the following resultant role: \[uniform convexity is dominant or recessive\ ] * * * (undifferentiated convex shape is dominant w.r.t. $ p_\mu$) * * * Suppose that for each $ i\in I$, $ f_i:{\ensuremath{\mathcal H}}\to{\ensuremath{\,\left]-\infty,+\infty\right]}}$ is proper, lower semicontinuous, convex and there exist $ i_0\

if there exists a function $\psi:[0,\infty[\to[0,\infti]$ qith tie propsrty $\lim_{g\to 0}{\psi(t)}/{t}=0$ such that for everb twi poibts $x$ and $y$ in ${\ensuremxth{\mathcap H}}$ and wverb $\lambda\in\ ]0,1[$ such that $(1-\lambda)x+\lzlbda b\in{\ensuremath{\opgratorname{dok}}}f$, $$\label{unifork rmlothness} f\big((1-\lambda)x+\lambda y\big)+\lambdw(1-\lambda)\lsl\big(\|x-y\|\big)\geq(1-\laibda)g(v)+\lamgda f(y).$$ The smallest possible functjon $\psi$ satisfying  ix called the *gauge of uniflrm dmoothness* of $f$ anf is definee by $$\psi_f(t)=\sup\bigg\{\wrac{(1-\lambda)f(x)+\lambda f(y)-f((1-\mambda)x+\lambda y)}{\lambda(1-\lambda)}\ \bige|\ \begnn{array}{c} \lamvdq\in\ ]0,1[,\ x,y\in{\ensurenath{\mwthcal H}},\ \|x-y\|=t, \\ \big((1-\lamtda)x+\lamnda y\big)\in{\ensuvemati{\opeeatorname{dom}}}f \end{array}\ \bigg\}.$$ \[uniform convgxity equieament to uniform minitoniwity\]*[@Sal Gyeofem 3.5.10, (m)$\Lertrighharcow$(v)]* Supposs that $f:{\ensyremath{\mathcal H}}\to{\emstgrmath{\,\left]-\inffy,+\inftr\ryght]}}$ is proper, lower semicontinuous and cohvex. Then $f$ is uniformlt convex if and only lf $\partiaj f$ is uniformly monotone, in which case $\partial f$ is ${\eisjrenabh{\vafphl}}$-uniformly monotone with ${\ensuremath{\varphi}}=2{\ensuwsmstm{\varphi}}_f^{**}$. Consequektly, we arrive at yhf gjllowing resuuts: \[uniykrj convexity is domlnant ot recewsive\] ***(unisorm convexity is dominant w.r.t. $p_\mu$)*** Suppose nhat for each $i\in I$, $f_i:{\znsuremath{\machcal N}}\to{\enxuremath{\,\left]-\infty,+\infty\rnght]}}$ ia proper, loaer semickvtinuous, convex xnd tvere exists $i_0\

if there exists a function $\psi:[0,\infty[\to[0,\infty]$ with $\lim_{t\to such that every two points H}}$ every $\lambda\in\ ]0,1[$ that $(1-\lambda)x+\lambda y\in{\ensuremath{\operatorname{dom}}}f$, smoothness} f\big((1-\lambda)x+\lambda y\big)+\lambda(1-\lambda)\psi\big(\|x-y\|\big)\geq(1-\lambda)f(x)+\lambda f(y).$$ The smallest function $\psi$ satisfying is called the *gauge of uniform smoothness* of $f$ and defined by $$\psi_f(t)=\sup\bigg\{\frac{(1-\lambda)f(x)+\lambda f(y)-f((1-\lambda)x+\lambda y)}{\lambda(1-\lambda)}\ \bigg|\ \begin{array}{c} \lambda\in\ ]0,1[,\ x,y\in{\ensuremath{\mathcal H}},\ \|x-y\|=t, \\ y\big)\in{\ensuremath{\operatorname{dom}}}f \bigg\}.$$ convexity to uniform monotonicity\]*[@Zal Theorem 3.5.10, (i)$\Leftrightarrow$(v)]* Suppose that $f:{\ensuremath{\mathcal H}}\to{\ensuremath{\,\left]-\infty,+\infty\right]}}$ is proper, lower semicontinuous and convex. $f$ is uniformly convex if and only if f$ is uniformly monotone, which case $\partial f$ is monotone ${\ensuremath{\varphi}}=2{\ensuremath{\varphi}}_f^{**}$. Consequently, arrive the results: \[uniform convexity dominant or recessive\] ***(uniform convexity is dominant w.r.t. $p_\mu$)*** Suppose that for each $i\in I$, $f_i:{\ensuremath{\mathcal H}}\to{\ensuremath{\,\left]-\infty,+\infty\right]}}$ proper, lower and there $i_0\

if there exists a function $\psi:[0,\Infty[\to[0,\infTy]$ witH thE prOpErty $\Lim_{t\To 0}{\psi(t)}/{t}=0$ such thaT For eVery two points $x$ and $y$ in ${\enSuremAtH{\MathCAl h}}$ and eVery $\lamBDa\IN\ ]0,1[$ SucH tHaT $(1-\laMbDA)x+\LambdA y\iN{\ensureMath{\operatOrnAmE{dom}}}f$, $$\label{unIFoRm smoothneSs} f\Big((1-\lambda)x+\laMbdA y\big)+\lAmBda(1-\LAmbda)\Psi\Big(\|x-y\|\Big)\geq(1-\LAmbda)f(X)+\lambda f(y).$$ thE SmalleST possibLE FuNctiOn $\psi$ satisfying  is CAlLEd the *gauge of unIform sMoOThNESs* oF $f$ aNd is defineD bY $$\psi_f(T)=\Sup\bigg\{\FRaC{(1-\LAMbdA)F(x)+\lambda f(y)-f((1-\laMbda)x+\lambda Y)}{\LamBda(1-\lamBdA)}\ \biGG|\ \begin{Array}{C} \lAMbdA\in\ ]0,1[,\ x,y\in{\ensuRemaTh{\mathcal h}},\ \|x-y\|=t, \\ \biG((1-\Lambda)x+\LAmbda y\bIg)\in{\enSurEmaTh{\opERaToRnaMe{DOm}}}f \ENd{ArrAY}\ \biGg\}.$$ \[uniforM cOnVexitY equIVALEnt tO unIforM monoTonicity\]*[@Zal ThEorEm 3.5.10, (i)$\LEFtrIghtaRrow$(v)]* suppOsE that $F:{\ensurEmath{\MaThcal H}}\to{\ensuremAth{\,\lEft]-\infty,+\iNftY\rIghT]}}$ iS propER, lower SemIcoNtinuouS and conVEx. THeN $F$ IS uNiformly convex if anD oNLY iF $\partial F$ is uniFOrMlY Monotone, In WhiCh caSE $\PartiAl f$ iS ${\EnSuremath{\Varphi}}$-UNiFoRmly monOtOne witH ${\eNsuRemAth{\vaRPhi}}=2{\eNsuremAth{\varphI}}_f^{**}$. ConSEquently, we arriVE at the followiNG rESUlTS: \[uniForM convexity iS domINant Or reCEsSivE\] ***(UnifoRm conVeXItY Is dominant w.r.t. $p_\mu$)*** SupPoSe that For eaCh $i\in I$, $f_i:{\ensurEmath{\mathcAL h}}\To{\ensureMath{\,\LEfT]-\Infty,+\infty\righT]}}$ is prOper, lower sEMicontinUous, cOnvex and There exisTS $I_0\

if there exists a functio n $\psi:[0 ,\inf ty[ \to [0 ,\in fty] $ with the pro p erty $\lim_{t\to 0}{\psi(t )}/{t }= 0 $ su c hthatfor eve r yt w o p oi nt s $ x$ an d $y$ in ${\ens uremath{\m ath ca l H}}$ and e v er y $\lambda \in \ ]0,1[$ suc h t hat $( 1- \la m bda)x +\l ambda y\in{ \ ensure math{\ope ra t orname { dom}}}f $ , $ $\la bel{uniform smoot h ne s s} f\big((1-\l ambda) x+ \ la m b day\b ig)+\lambd a( 1-\la m bda)\ps i \b i g ( \|x - y\|\big)\geq( 1-\lambda)f ( x)+ \lambd af(y ) .$$ Th e sma ll e stpossible fu ncti on $\psi$ satis f ying  i s called the * gau geof u n if or m s mo o thn e ss * o f $f $ and is d ef inedby $ $ \ p s i_f( t)= \sup \bigg \{\frac{(1-\l amb da)f ( x)+ \lamb da f( y)-f (( 1-\la mbda)x +\lam bd a y)}{\lambda(1 -\la mbda)}\ \ big g| \ \ be gin{a r ray}{c } \ lam bda\in\ ]0,1[, \ x, y\ i n { \e nsuremath{\mathcal H } } ,\ \|x-y\| =t, \\ \ bi g ((1-\lam bd a)x +\la m b da y\ big) \ in {\ensure math{\ o pe ra torname {d om}}}f \ end {ar ray}\ \bi gg\}.$ $ \[uni formc onvexity equiv a lent to unifo r mm o no t onic ity \]*[@Zal Th eore m 3.5 .10, (i )$\ L eftri ghtar ro w $( v )]* Suppose that $f :{ \ensur emath {\mathcal H}} \to{\ensur e m a th{\,\le ft]- \ in f ty,+\infty\rig ht]}} $ is prope r , lowersemic ontinuou s and con v e x. Then$f$ is un ifo r m ly convex if an d only i f $\par tia l f$ is un ifo rml y m on otone, in which c as e$\ pa rti al f$ is ${\en su rem at h{\ varph i }}$-un iform ly m on ot o newith ${ \ en s u rema th {\ varp hi} }= 2{\en sure m ath {\varph i}}_f^{** }$. Con se qu ently,we arrive atth e followin gres ults:\ [uniform convexity is dominanto r reces siv e\] * **(u niform co nve xity i s d o minant w.r.t . $p_ \m u$) * * * Sup p o se th at for each$ i \in I$,$f _i:{ \ensure math{\mathcal H}}\ t o{\ ensuremath{\, \le ft]- \ i nf ty, + \i n fty \r i ght ] } }$ is proper, l ower semic on t in uous, conv e x a nd thereexists$i_0\

if_there exists_a function $\psi:[0,\infty[\to[0,\infty]$ with_the property_$\lim_{t\to_0}{\psi(t)}/{t}=0$ such_that_for every two_points $x$ and_$y$ in ${\ensuremath{\mathcal H}}$_and every $\lambda\in\_]0,1[$_such that $(1-\lambda)x+\lambda y\in{\ensuremath{\operatorname{dom}}}f$, $$\label{uniform smoothness} f\big((1-\lambda)x+\lambda y\big)+\lambda(1-\lambda)\psi\big(\|x-y\|\big)\geq(1-\lambda)f(x)+\lambda f(y).$$ The smallest possible function $\psi$ satisfying _is_called the_*gauge_of_uniform smoothness* of $f$ and_is defined by $$\psi_f(t)=\sup\bigg\{\frac{(1-\lambda)f(x)+\lambda f(y)-f((1-\lambda)x+\lambda_y)}{\lambda(1-\lambda)}\ \bigg|\_\begin{array}{c} \lambda\in\ ]0,1[,\ x,y\in{\ensuremath{\mathcal H}},\ \|x-y\|=t, \\ \big((1-\lambda)x+\lambda y\big)\in{\ensuremath{\operatorname{dom}}}f \end{array}\__\bigg\}.$$ \[uniform convexity equivalent_to uniform monotonicity\]*[@Zal Theorem 3.5.10, (i)$\Leftrightarrow$(v)]* Suppose that $f:{\ensuremath{\mathcal_H}}\to{\ensuremath{\,\left]-\infty,+\infty\right]}}$ is proper, lower semicontinuous and_convex. Then $f$_is_uniformly_convex if and only_if $\partial f$ is uniformly monotone,_in which case $\partial f$ is_${\ensuremath{\varphi}}$-uniformly monotone with ${\ensuremath{\varphi}}=2{\ensuremath{\varphi}}_f^{**}$. Consequently, we arrive at_the following results: \[uniform convexity is dominant_or recessive\] ***(uniform convexity is_dominant w.r.t._$p_\mu$)*** Suppose that for each_$i\in I$, $f_i:{\ensuremath{\mathcal_H}}\to{\ensuremath{\,\left]-\infty,+\infty\right]}}$ is_proper, lower semicontinuous,_convex and there exists $i_0\

D/L$, where $L$ is the height of the sample and $D$ its width, $g$ the gravitational acceleration, $\beta$ is the thermal expansion coefficient and $\nu$ and $\kappa$ the kinematic viscosity and the thermal diffusivity, respectively. RB convection is used as a simple model for all types of natural convection, regardless of geometry. For 2D RB, at a moderate ${\text{Ra}}= 10^8$, the heat flux is expected to be independent of the aspect-ratio, and thus of the sidewall boundary conditions, beyond $\Gamma \gtrsim 26$ [@poe12]. This value of $\Gamma$ is expected to be lower for 3D RB [@bai10]. Low aspect-ratio applications with a no-slip BC on the lateral walls, such as cooling fins and domestic heating, have moderate Rayleigh numbers (${\text{Ra}}< 10^{10}$) that are accessible with contemporary direct numerical simulations (DNS). However, Nature is dominated by flows with large aspect-ratios and inaccessible high Ra [@chi12]. In both experiments and numerics, the aspect-ratio $\Gamma$ is commonly reduced to decrease the volume of the setup, as Ra depends only on the height of the system. In simulations, reducing the aspect-ratio for fixed Ra decreases the computational complexity by $\mathcal{O}(\Gamma^2)$ in three-dimensions, while in experiments it means a smaller setup volume. In both cases it makes higher Rayleigh numbers more accesible. Despite the additional effort required for large aspect-ratio, research in this direction has been done in simulations [@bai10] and experiments [@fun05; @pui13; @hog13]. However, these endeavours were all limited in Ra due the large aspect-ratios used. In simulations one can approximate infinitely large $\Gamma$ by using periodic boundary conditions at the sidewalls. This leads to the question how large the simulated domain has to be to sufficiently approximate infinite $\Gamma$, as a truly infinite domain can in principle support any wavenumber while a periodic domain allows only for multiples of the fundamental one imposed by its size. This is addressed in this study in 2D RB, which is expected to require a larger domain than 3D and therefore is more demanding in terms of computational resources. Different temperature boundary conditions at the plates were studied by [@joh09], where no significant difference between constant flux and isothermal plates was found in the higher Ra

D / L$, where $ L$ is the height of the sample and $ D$ its width, $ g$ the gravitational acceleration, $ \beta$ is the thermal expansion coefficient and $ \nu$ and $ \kappa$ the kinematic viscosity and the thermal diffusivity, respectively. RB convection is used as a simple exemplar for all type of natural convection, regardless of geometry. For 2D RB, at a moderate $ { \text{Ra}}= 10 ^ 8 $, the heat flux is ask to be independent of the aspect - proportion, and thus of the sidewall boundary conditions, beyond $ \Gamma \gtrsim 26 $ [ @poe12 ]. This value of $ \Gamma$ is expected to be broken for 3D RB [ @bai10 ]. Low aspect - proportion applications with a no - slip BC on the lateral walls, such as cooling flipper and domestic heating, get moderate Rayleigh numbers ($ { \text{Ra } } < 10^{10}$) that are accessible with contemporary lineal numerical simulations (DNS). However, Nature is dominated by flows with big aspect - ratios and inaccessible eminent Ra [ @chi12 ]. In both experiments and numerics, the view - ratio $ \Gamma$ is commonly reduced to decrease the volume of the setup, as Ra depend only on the height of the system. In simulations, reducing the aspect - ratio for fixed Ra decreases the computational complexity by $ \mathcal{O}(\Gamma^2)$ in three - dimensions, while in experiments it means a smaller setup volume. In both cases it makes high Rayleigh numbers more accesible. Despite the additional campaign required for large aspect - ratio, inquiry in this direction has been suffice in simulations [ @bai10 ] and experiments [ @fun05; @pui13; @hog13 ]. However, these enterprise were all limited in Ra due the large aspect - ratios use. In simulations one can approximate infinitely large $ \Gamma$ by use periodic boundary condition at the sidewall. This leads to the question how large the simulate domain has to be to sufficiently approximate infinite $ \Gamma$, as a in truth infinite domain can in principle support any wavenumber while a periodic domain allow only for multiples of the fundamental one imposed by its size. This is addressed in this survey in 2D RB, which is expected to necessitate a larger knowledge domain than 3D and therefore is more demanding in term of computational resources. Different temperature boundary conditions at the plates were studied by [ @joh09 ], where no significant difference between constant flux and isothermal plates was found in the high Ra

D/L$, where $L$ is the height on the sample and $D$ its xidth, $g$ the graxitational acceleration, $\beta$ iw the thermal expansion coewficient wnd $\nu$ abd $\keppa$ the kinematmd viscosity ahf thz vhermal diffusiyity, respecdively. RB convewtkou is used as a simple model for all eypes og jatural convecjion, gedardmvsw of geometry. For 2D RB, at a mkderate ${\text{Ra}}= 10^8$, the neat flux is expected to bf infependent of the adpect-ratio, qnd eyus of the skdewall boludary condijions, beyond $\Gamma \gtrsim 26$ [@poe12]. Tfis vclue of $\Gamna$ is gxpected to ue lowvr for 3D RB [@bai10]. Low avpect-rayio applicatioks wivh a no-slip BC on the latxral walls, such as cjoling fits and domestic hearibg, haee mmderxre Faymemgh numbegs (${\vext{Ra}}< 10^{10}$) thaf are acceswible with contempotawj direct numedical fiiulations (DNS). However, Nature is dominattd by flows with large aspecr-ratios and inaccessihle high Wa [@chi12]. In both experiments and numerics, the aspect-satio $\Eamna$ kw fommonly reduced to decrease the volume of thq sttui, as Ra depends okly on the height pf tng system. In sioulations, reducing the aspeft-ratio for dixed Ra qecrrases the computational complexity by $\mcthxal{O}(\Gamma^2)$ in three-bimensions, wkile im exprriments it means a smauler setup volule. In botg cases it makes figmer Rayleigh numbers more accqsible. Dewpitz the adaitipnal esfort requlred nmr large aspect-ratlo, rededrch in thls direction has been done in smkulations [@bao10] dnd experimznts [@fmn05; @pui13; @hog13]. Howeder, these endecvours wzre alu limited pn Ra due the large wspect-ratios gded. In simuletions onq cab approximagd infinitely lsrge $\Gammc$ by ysing periodic boukdary donditions at tkt sudewalls. This lradr tj nhe qtastion how ldrge thd simuuated domaik hxs tp be to sufficiently appdoximate infinite $\Baima$, as a truly igfinite domaim can in principle suppmrt any eavgnumber while a periodic domain allows ojly for multiplqs on thg fundamentcl one imposed by its size. This is addrewsed in this studi in 2D RB, which is exkecued to requirx a lawger domahn than 3D and therefire is more demancing in terms of comphtatiotal rfsources. Different temperature boundary conditions at the plates were studued by [@joh09], where nk sibnifibanc dnfferense bxtxeen constant fluq and isothermal plates was founv in the hhgker Ra

D/L$, where $L$ is the height of and its width, the gravitational acceleration, coefficient $\nu$ and $\kappa$ kinematic viscosity and thermal diffusivity, respectively. RB convection is as a simple model for all types of natural convection, regardless of geometry. 2D RB, at a moderate ${\text{Ra}}= 10^8$, the heat flux is expected to independent the and of the sidewall boundary conditions, beyond $\Gamma \gtrsim 26$ [@poe12]. This value of $\Gamma$ is expected be lower for 3D RB [@bai10]. Low aspect-ratio with a no-slip BC the lateral walls, such as fins domestic heating, moderate numbers 10^{10}$) that are with contemporary direct numerical simulations (DNS). However, Nature is dominated by flows with large aspect-ratios and inaccessible Ra [@chi12]. experiments and the $\Gamma$ commonly reduced to volume of the setup, as Ra the height of the system. In simulations, reducing aspect-ratio for Ra decreases the computational complexity by in three-dimensions, while in experiments it means a setup volume. In both cases it makes higher Rayleigh numbers more accesible. Despite the additional for large aspect-ratio, research this direction has done simulations and [@fun05; @pui13; However, these endeavours were all limited in Ra due the large used. In simulations one can approximate infinitely large $\Gamma$ by boundary at the sidewalls. leads to the question large simulated domain has to sufficiently infinite truly domain in principle support any while a periodic domain allows for multiples of the size. This is addressed in this study in RB, which is expected to require a domain than 3D and therefore is more demanding in terms of computational Different temperature at the plates were studied by [@joh09], where significant difference between constant and isothermal plates was found in the higher Ra

D/L$, where $L$ is the height of the sAmple and $D$ iTs widTh, $g$ The GrAvitAtioNal acceleratioN, $\Beta$ Is the thermal expansion cOeffiCiENt anD $\Nu$ And $\kaPpa$ the kINeMATic ViScOsiTy ANd The thErmAl diffuSivity, respEctIvEly. RB convectIOn Is used as a sImpLe model for alL tyPes of nAtUraL ConveCtiOn, regArdlesS Of geomEtry. For 2D Rb, aT A moderATe ${\text{RA}}= 10^8$, THe Heat Flux is expected to bE InDEpendent of the aSpect-rAtIO, aND ThuS of The sidewalL bOundaRY conditIOnS, BEYonD $\gamma \gtrsim 26$ [@poE12]. This value oF $\gamMa$ is exPeCteD To be loWer foR 3D rb [@baI10]. Low aspect-rAtio ApplicatiOns witH A no-slip bc on the lAteral WalLs, sUch aS CoOlIng FiNS anD DoMesTIc hEating, haVe MoDeratE RayLEIGH numBerS (${\texT{Ra}}< 10^{10}$) thAt are accessibLe wIth cONteMporaRy dirEct nUmEricaL simulAtionS (DnS). However, Nature Is doMinated by FloWs WitH lArge aSPect-raTioS anD inacceSsible hIGh RA [@cHI12]. iN bOth experiments and nUmERIcS, the aspeCt-ratiO $\gaMmA$ Is commonLy RedUced TO DecreAse tHE vOlume of tHe setuP, As ra Depends OnLy on thE hEigHt oF the sYStem. in simuLations, rEduciNG the aspect-ratiO For fixed Ra decREaSES tHE comPutAtional compLexiTY by $\mAthcAL{O}(\gamMA^2)$ in thRee-diMeNSiONs, while in experimentS iT means A smalLer setup volumE. In both casES IT makes hiGher rAyLEigh numbers morE acceSible. DespiTE the addiTionaL effort rEquired foR LArge aspeCt-rAtiO, reSeaRCH iN this directioN HAs beEn Done in sImuLations [@Bai10] And ExpEriMeNts [@fun05; @pui13; @Hog13]. HowevEr, ThEsE eNdeAvourS Were all lImIteD iN Ra Due thE Large aSpect-RatiOs UsED. In SimulatIOnS ONe caN aPpRoxiMatE iNfiniTely LArgE $\Gamma$ bY using perIodIC bouNdArY conditIons at the sideWaLls. This leaDs To tHe quesTIOn how larGe the simulated domain has TO be to suFfiCientLy apProximate InfInite $\GAmmA$, As a truLy infiNite dOmAin CAN in prINCiPle SuPport any waVENumBer whIlE a peRiodic dOmain allows only for MUltIples of the funDamEntaL ONe ImpOSeD By iTs SIze. tHIs is addressed in This study iN 2D rb, wHich is expeCTed To Require A larger DomaiN Than 3D anD thereforE is more deMaNdinG IN teRms of compuTational Resources. dIfferENt TempeRatUre bouNdAry CondiTions aT The PlateS were sTuDied by [@Joh09], whErE no signiFicant difference between ConstaNt fluX anD isothermAl pLAteS was found In thE higher Ra

D/L$, where $L$ is the he ight of th e sam ple an d$D$itswidth, $g$ the grav itational acceleration , $\b et a $ is th e the rmal ex p an s i onco ef fic ie n tand $ \nu $ and $ \kappa$ th e k in ematic visco s it y and thethe rmal diffusi vit y, res pe cti v ely. RB conv ection is use d as a si mp l e mode l for al l ty pesof natural convec t io n , regardless o f geom et r y. F or2DRB, at a m od erate ${\text { Ra } } = 10 ^ 8$, the heatflux is exp e cte d to b eind e penden t ofth e as pect-ratio, and thus ofthe si d ewall b o undarycondit ion s,beyo n d$\ Gam ma \gt r si m 2 6 $ [ @poe12]. T hi s val ue o f $ \ Gamm a$is e xpect ed to be lowe r f or 3 D RB [@ba i10]. Low a spect -ratio appl ic ations with a n o-sl ip BC onthe l ate ra l wal l s, suc h a s c oolingfins an d do me s t i cheating, have mode ra t e R ayleighnumber s ( ${ \ text{Ra} }< 10 ^{10 } $ ) tha t ar e a ccessibl e with co nt emporar ydirect n ume ric al si m ulat ions ( DNS). Ho wever , Nature is dom i nated by flow s w i t hl arge as pect-ratios and inac cess i bl e h i gh Ra [@ch i1 2 ]. In both experiment sand nu meric s, the aspect -ratio $\G a m m a$ is co mmon l yr educed to decr easethe volume of the s etup, as Ra d epends on l y on thehei ght of th e sy stem. In simu l a tion s, reduci ngthe asp ect -ra tio fo rfixed Radecrease sth eco mpu tatio n al compl ex ity b y $ \math c al{O}( \Gamm a^2) $in thr ee-dime n si o n s, w hi le inexp er iment s it mea ns a sm aller set upv olum e. I n bothcases it make shigher Ray le igh numbe r s more ac cesible. Despite the ad d itional ef fortrequ ired forlar ge asp ect - ratio, resea rch i nthi s direc t i on ha sbeen donei n si mulat io ns [ @bai10] and experiments [ @ fun 05; @pui13; @ hog 13]. H ow eve r ,t hes ee nde a v ours were all l imited inRa du e the larg e as pe ct-rati os used . Ins imulati ons one c an approx im atei n fin itely larg e $\Gamm a$ by usi n g per i od ic bo und ary co nd iti ons a t thes ide walls . This l eads t o the q uestionhow large the simulated domai n has to be to su ffi c ien tly appro xima te infinit e $ \Ga mma$, as a tru ly i n fi nit e doma in c a n in prin c ip les u pp ort any wav e n u mbe r whi lea perio dicdomain allows onl y for multiples oft h e f und a ment al one imposed b y i ts s ize. Thi sis addresse d in thi ss tudyin 2DRB, wh ich ise x pe c ted to req uir e a large r d om a in than 3 Da nd the refo re is mo re dem a ndin g in terms of comp utati o n al re s our ces. D ifferen t tem perature b oundary con dition s at theplateswe re stu die dby [@joh09 ] , where n o sig nifican tdiff ere nce be twee n const antfl uxand isoth e r ma l p la t eswasfound i n th e higherR a

D/L$,_where $L$_is the height of_the sample_and_$D$ its_width,_$g$ the gravitational_acceleration, $\beta$ is_the thermal expansion coefficient_and $\nu$ and_$\kappa$_the kinematic viscosity and the thermal diffusivity, respectively. RB convection is used as a simple_model_for all_types_of_natural convection, regardless of geometry._For 2D RB, at a_moderate ${\text{Ra}}=_10^8$, the heat flux is expected to be_independent_of the aspect-ratio,_and thus of the sidewall boundary conditions, beyond $\Gamma_\gtrsim 26$ [@poe12]. This value of_$\Gamma$ is expected_to_be_lower for 3D RB_[@bai10]. Low aspect-ratio applications with a_no-slip BC on the lateral walls,_such as cooling fins and domestic heating,_have moderate Rayleigh numbers (${\text{Ra}}< 10^{10}$)_that are accessible with contemporary_direct numerical_simulations (DNS). However, Nature is_dominated by flows_with large_aspect-ratios and inaccessible_high Ra [@chi12]. In both experiments and_numerics, the aspect-ratio_$\Gamma$ is commonly reduced to decrease_the_volume of the_setup,_as_Ra depends_only on the_height_of the_system._In simulations, reducing the aspect-ratio for_fixed_Ra decreases the computational complexity by $\mathcal{O}(\Gamma^2)$_in three-dimensions, while in_experiments_it means a smaller_setup volume. In both cases_it makes higher Rayleigh numbers more_accesible. Despite_the additional_effort required for large aspect-ratio, research in this direction has been_done in simulations [@bai10] and experiments_[@fun05; @pui13; @hog13]. However,_these endeavours_were_all limited in_Ra_due the_large aspect-ratios used. In simulations one can_approximate infinitely_large $\Gamma$ by using periodic boundary_conditions at the sidewalls._This_leads to the question how large_the simulated domain has to be_to sufficiently approximate infinite $\Gamma$,_as_a_truly infinite domain can in_principle support any wavenumber while a_periodic domain allows_only for multiples of the fundamental one_imposed_by its size. This is addressed_in_this study in 2D RB, which_is_expected_to require a larger domain_than 3D and therefore is more_demanding in terms of computational resources. Different temperature boundary conditions_at the plates_were studied by [@joh09], where_no_significant_difference between constant flux and isothermal plates was found in_the higher_Ra

and a vector meson [@Bodwin:2013gca; @Kagan:2014ila; @Alte:2016yuw]. Thus far, the ${\PGg\PJGy}$, [$\PGg\PGy$(2S)]{}, [$\PGg\PGU$(nS)]{}, ${\ensuremath{\PGg\PGr}\xspace}$, and ${\ensuremath{\PGg\PGf}\xspace}$ decays have been searched for [@Sirunyan:2018fmm; @Aaboud:2018txb; @Aaboud:2017xnb]. The 95% upper limits on the branching fractions of the Higgs boson into ${\PGg\PJGy}$, ${\ensuremath{\PGg\PGr}\xspace}$, and ${\ensuremath{\PGg\PGf}\xspace}$ are 2 orders of magnitude larger than their expected values in the SM. For the [$\PGg\PGy$(2S)]{}and [$\PGg\PGU$(nS)]{}decays, the corresponding upper limits are, respectively, 3 and 5 orders of magnitude larger than the SM expectation. A related class of rare decays is that of the Higgs boson into a heavy vector boson and a vector meson (V) [@deFlorian:2016spz; @Isidori:2013cla]. Up to now only the decays of the Higgs boson into $\PZ\PJGy$ and $\PZ\PGhc$ have been studied experimentally [@Aad:2020hzm]. As indicated in Fig. \[fig:diagrams\], several processes contribute to the decay of the Higgs boson into a vector boson and a meson. The formation of a vector boson and a meson via ${\PH\to{\PZ\PZ^{*}}}$ or $\PH\to\PZ\PGg^{*}$ decays (Fig. \[fig:diagrams\], left and middle) are indirect contributions to this process. We refer to the decay of the Higgs boson into light quarks that radiate a vector boson and form a bound meson state (Fig. \[fig:diagrams\], right) as the direct process. In the SM the indirect processes contribute the most to the decay of the Higgs boson into a heavy vector boson and a vector meson. ![Feynman diagrams that contribute to the decay of a Higgs boson into a heavy vector boson and a vector meson. The grey oval shape represents the meson. The two indirect processes (left and middle), where the meson originates from an off

and a vector meson   [ @Bodwin:2013gca; @Kagan:2014ila; @Alte:2016yuw ]. Thus far, the $ { \PGg\PJGy}$, [ $ \PGg\PGy$(2S) ] { }, [ $ \PGg\PGU$(nS) ] { }, $ { \ensuremath{\PGg\PGr}\xspace}$, and $ { \ensuremath{\PGg\PGf}\xspace}$ decays have been searched for   [ @Sirunyan:2018fmm; @Aaboud:2018txb; @Aaboud:2017xnb ]. The 95% upper limits on the ramify fraction of the Higgs boson into $ { \PGg\PJGy}$, $ { \ensuremath{\PGg\PGr}\xspace}$, and $ { \ensuremath{\PGg\PGf}\xspace}$ are 2 orders of magnitude larger than their ask values in the SM. For the [ $ \PGg\PGy$(2S)]{}and [ $ \PGg\PGU$(nS)]{}decays, the corresponding upper limit are, respectively, 3 and 5 order of magnitude bigger than the SM expectation. A related course of rare decays is that of the Higgs boson into a heavy vector boson and a vector meson (V)   [ @deFlorian:2016spz; @Isidori:2013cla ]. Up to now only the decay of the Higgs boson into $ \PZ\PJGy$ and $ \PZ\PGhc$ have been studied experimentally   [ @Aad:2020hzm ]. As indicate in Fig.   \[fig: diagrams\ ], several processes lend to the decay of the Higgs boson into a vector boson and a meson. The formation of a vector boson and a meson via $ { \PH\to{\PZ\PZ^{*}}}$ or $ \PH\to\PZ\PGg^{*}$ decays (Fig.   \[fig: diagrams\ ], left and center) are indirect contributions to this process. We refer to the decay of the Higgs boson into unaccented quarks that radiate a vector boson and form a bound meson state (Fig.   \[fig: diagrams\ ], right) as the lineal process. In the SM the indirect processes contribute the most to the decay of the Higgs boson into a heavy vector boson and a vector meson. ! [ Feynman diagram that lend to the decay of a Higgs boson into a heavy vector boson and a vector meson. The grey oval shape represents the meson. The two indirect processes (left and middle), where the meson originates from an off

anf a vector meson [@Bodwin:2013gcx; @Kagan:2014ila; @Alte:2016iuq]. Thus far, tge ${\PGg\PJEy}$, [$\PGg\PGy$(2S)]{}, [$\PGg\PGU$(nS)]{}, ${\ensuremavh{\PGt\PGr}\xwpace}$, and ${\ensuremath{\PGe\PGf}\xspacv}$ decays yave veen searcisd for [@Slxunyah:2018nmm; @Acbiud:2018txb; @Aaboud:2017xkb]. The 95% uppar limits on tve bxanching fractions of the Higgs bosog into ${\LGh\PJGy}$, ${\ensuremajh{\PGg\KGr}\vspadv}$, cnd ${\ensuremath{\PGg\PGf}\xspace}$ are 2 orders of magnitude larger than their expectef vapues in the SM. For the [$\PGg\PGy$(2W)]{}and [$\PGg\PGU$(nS)]{}decahs, the corgzsponding ukper limits are, respectively, 3 ana 5 orbers of magbirudf larger thai the FM expectation. A reladed claxs of rare decsys is that of the Higgs bovon into a heavy vgctor bosot cnd a vector meson (V) [@dwFooriat:2016spz; @Isiairi:2013zla]. U' tk now lnlb the decaya of the Hitgs boson into $\PZ\PJBy$ qnd $\PZ\PGhc$ habe beeg ftudied experimentally [@Aad:2020hzm]. As indicattd in Fig. \[fig:diagrams\], several processes contribute to the dqcay of the Higgs boson into a vector boson and a mesoi. Ghe fjfnahion of a vector boson and a meson via ${\PH\to{\PZ\[A^{*}}}$ pr $\PH\to\PZ\PGg^{*}$ deccys (Fig. \[fig:diagraks\], lrst and middle) are iusidect contributions to thif proxess. We rtfer yo the decay of the Higgs bison into liyht quarks that radiace a vector yoson snd fprm a bound meson state (Fig. \[rig:diagrams\], right) as ghe direct procers. Pn tve SM tht indirect processqs contriuute che most to jhe decwy of the Jiggs boson into a heavy vectlr boson and a vector meson. ![Feynman diagrams that contribuje do nhe decay of a Higgs boson ynto a heavy vgctor bosjn ana a vector meson. Vhe grey ovaj shape reprevgnts the mesoi. The two indurecr procerres (left and moddle), whege the meson originates nrom xh off

and a vector meson [@Bodwin:2013gca; @Kagan:2014ila; @Alte:2016yuw]. the [$\PGg\PGy$(2S)]{}, [$\PGg\PGU$(nS)]{}, and ${\ensuremath{\PGg\PGf}\xspace}$ decays @Aaboud:2018txb; The 95% upper on the branching of the Higgs boson into ${\PGg\PJGy}$, and ${\ensuremath{\PGg\PGf}\xspace}$ are 2 orders of magnitude larger than their expected values in SM. For the [$\PGg\PGy$(2S)]{}and [$\PGg\PGU$(nS)]{}decays, the corresponding upper limits are, respectively, 3 and orders magnitude than SM expectation. A related class of rare decays is that of the Higgs boson into a vector boson and a vector meson (V) [@deFlorian:2016spz; Up to now only decays of the Higgs boson $\PZ\PJGy$ $\PZ\PGhc$ have studied [@Aad:2020hzm]. indicated in Fig. several processes contribute to the decay of the Higgs boson into a vector boson and a meson. formation of boson and meson ${\PH\to{\PZ\PZ^{*}}}$ $\PH\to\PZ\PGg^{*}$ decays (Fig. and middle) are indirect contributions to refer to the decay of the Higgs boson light quarks radiate a vector boson and form bound meson state (Fig. \[fig:diagrams\], right) as the process. In the SM the indirect processes contribute the most to the decay of the into a heavy vector and a vector ![Feynman that to decay of Higgs boson into a heavy vector boson and a vector meson. grey oval shape represents the meson. The two indirect processes middle), the meson originates an off

and a vector meson [@Bodwin:2013gca; @KAgan:2014ila; @AltE:2016yuw]. THus Far, ThE ${\PGg\pJGy}$, [$\pGg\PGy$(2S)]{}, [$\PGg\PGU$(ns)]{}, ${\EnsuRemath{\PGg\PGr}\xspace}$, and ${\eNsureMaTH{\PGg\pgf}\XspacE}$ decays HAvE BEen SeArCheD fOR [@SIrunyAn:2018fMm; @AabouD:2018txb; @Aaboud:2017Xnb]. thE 95% upper limits ON tHe branchinG frActions of the higGs bosoN iNto ${\pgg\PJGY}$, ${\enSuremAth{\PGg\pgr}\xspaCe}$, and ${\ensuReMAth{\PGg\pgf}\xspacE}$ ARe 2 OrdeRs of magnitude largER tHAn their expecteD valueS iN ThE sm. FoR thE [$\PGg\PGy$(2S)]{}anD [$\Pgg\PGU$(Ns)]{}decays, THe CORResPOnding upper liMits are, respECtiVely, 3 anD 5 oRdeRS of magNitudE lARgeR than the SM eXpecTation. A reLated cLAss of raRE decays Is that Of tHe HIggs BOsOn IntO a HEavY VeCtoR BosOn and a veCtOr Meson (v) [@deFLORIAn:2016spZ; @IsIdorI:2013cla]. UP to now only the DecAys oF The higgs Boson Into $\pZ\pJGy$ aNd $\PZ\PGHc$ havE bEen studied experImenTally [@Aad:2020hZm]. AS iNdiCaTed in fIg. \[fig:dIagRamS\], severaL procesSEs cOnTRIBuTe to the decay of the HIgGS BoSon into a Vector BOsOn ANd a meson. thE foRmatION of a vEctoR BoSon and a mEson viA ${\pH\To{\pZ\PZ^{*}}}$ or $\Ph\tO\PZ\PGg^{*}$ DeCayS (FiG. \[fig:dIAgraMs\], left And middlE) are iNDirect contribuTIons to this proCEsS. wE rEFer tO thE decay of the higgS BosoN intO LiGht QUarks That rAdIAtE A vector boson and form A bOund meSon stAte (Fig. \[fig:diagRams\], right) aS THE direct pRoceSS. IN The SM the indireCt proCesses contRIbute the Most tO the decaY of the HigGS Boson intO a hEavY veCtoR BOsOn and a vector mESOn. ![FeYnMan diagRamS that coNtrIbuTe tO thE dEcay of a HiGgs boson InTo A hEaVy vEctor BOson and a VeCtoR mEsoN. The gREy oval Shape ReprEsEnTS thE meson. THE tWO IndiReCt ProcEssEs (Left aNd miDDle), Where thE meson oriGinATes fRoM aN off

and a vector meson [@Bodw in:2013gca ; @Ka gan :20 14 ila; @Al te:2016yuw]. T h us f ar, the ${\PGg\PJGy}$, [$\P Gg \ PGy$ ( 2S )]{}, [$\PGg \ PG U $ (nS )] {} , $ {\ e ns urema th{ \PGg\PG r}\xspace} $,an d ${\ensurem a th {\PGg\PGf} \xs pace}$ decay s h ave be en se a rched fo r [@S irunya n :2018f mm; @Aabo ud : 2018tx b ; @Aabo u d :2 017x nb]. The 95% uppe r l i mits on the br anchin gf ra c t ion s o f the Higg sboson into ${ \ PG g \ P JGy } $, ${\ensurem ath{\PGg\PG r }\x space} $, an d ${\en surem at h {\P Gg\PGf}\xsp ace} $ are 2 o rderso f magni t ude lar ger th anthe ir e x pe ct edva l ues in th e SM . For th e[$ \PGg\ PGy$ ( 2 S ) ]{}a nd[$\P Gg\PG U$(nS)]{}deca ys, the cor respo nding upp er limi ts are , res pe ctively, 3 and5 or ders of m agn it ude l arger than t heSMexpecta tion. A re la t e d c lass of rare decay si s t hat of t he Hig g sbo s on intoahea vy v e c tor b oson an d a vect or mes o n(V ) [@deF lo rian:2 01 6sp z;@Isid o ri:2 013cla ]. Up to nowo nly the decays of the Higgsb os o n i n to $ \PZ \PJGy$ and$\PZ \ PGhc $ ha v ebee n stud ied e xp e ri m entally [@Aad:2020h zm ]. Asindic ated in Fig.\[fig:diag r a m s\], sev eral pr o cesses contrib ute t o the deca y of theHiggs boson i nto a vec t o r bosonand ames on. T he formation of a vec to r boson an d a mes onvia ${ \PH \t o{\PZ\PZ^ {*}}}$ o r$\ PH \t o\P Z\PGg ^ {*}$ dec ay s ( Fi g.\[fig : diagra ms\], lef tan d mi ddle) a r ei n dire ct c ontr ibu ti ons t o th i s p rocess. We refer to thede ca y of th e Higgs boson i nto lightqu ark s that r adiate a vector boson and forma boundmes on st ate(Fig. \[f ig: diagra ms\ ] , righ t) asthe d ir ect p roces s . I n t he SM the in d i rec t pro ce sses contri bute the most to t h e d ecay of the H igg s bo s o nint o a hea vy vec t o r boson and a v ector meso n. ! [Feynman d i agr am s thatcontrib ute t o the de cay of aHiggs bos on int o a h eavy vecto r bosonand a vec t or me s on . The gr ey ova lsha pe re presen t s t he me son. T he two i ndire ct process es (left and middle), w here t he me son originat esf rom an off

and_a vector_meson [@Bodwin:2013gca; @Kagan:2014ila; @Alte:2016yuw]. Thus_far, the_${\PGg\PJGy}$,_[$\PGg\PGy$(2S)]{}, [$\PGg\PGU$(nS)]{},_${\ensuremath{\PGg\PGr}\xspace}$,_and ${\ensuremath{\PGg\PGf}\xspace}$ decays_have been searched_for [@Sirunyan:2018fmm; @Aaboud:2018txb; @Aaboud:2017xnb]. The_95% upper limits_on_the branching fractions of the Higgs boson into ${\PGg\PJGy}$, ${\ensuremath{\PGg\PGr}\xspace}$, and ${\ensuremath{\PGg\PGf}\xspace}$ are 2_orders_of magnitude_larger_than_their expected values in the_SM. For the [$\PGg\PGy$(2S)]{}and [$\PGg\PGU$(nS)]{}decays,_the corresponding_upper limits are, respectively, 3 and 5 orders_of_magnitude larger than_the SM expectation. A related class of rare decays is_that of the Higgs boson into_a heavy vector_boson_and_a vector meson (V) [@deFlorian:2016spz;_@Isidori:2013cla]. Up to now only the_decays of the Higgs boson into_$\PZ\PJGy$ and $\PZ\PGhc$ have been studied experimentally [@Aad:2020hzm]._As indicated in Fig. \[fig:diagrams\], several processes_contribute to the decay of_the Higgs_boson into a vector boson_and a meson._The formation_of a vector_boson and a meson via ${\PH\to{\PZ\PZ^{*}}}$_or $\PH\to\PZ\PGg^{*}$ decays_(Fig. \[fig:diagrams\], left and middle) are indirect_contributions_to this process._We_refer_to the_decay of the_Higgs_boson into_light_quarks that radiate a vector boson_and_form a bound meson state (Fig. \[fig:diagrams\], right)_as the direct process._In_the SM the indirect_processes contribute the most to_the decay of the Higgs boson_into a_heavy vector_boson and a vector meson. ![Feynman diagrams that contribute to the decay_of a Higgs boson into a_heavy vector boson and_a vector_meson._The grey oval_shape_represents the_meson. The two indirect processes (left and_middle), where_the meson originates from an off

big{]}$. Recall that $f_\omega (0)>0$ for all $\omega$, and $f_\omega (x)$ has a local minima at: $$x_0 = \frac{\omega [1-W_0(\alpha)]}{2LN}.$$ When $\omega=\omega_0$, denote this local minima as $x_0=a$ and $f_{\omega_0} (a)=0$. From (\[Eq23\]), we have: $$a=\frac{ \omega_0 [1-W_0 (\alpha)]}{2LN} =-\frac{W_0(\alpha)}{1-W_0(\alpha)} > 0.$$ For a fixed $x$, taking the derivative of (\[EqA2\]) with respect to $\omega$, we obtain: $$g'_x (\omega) = \frac{\partial F (x, \omega)}{\partial \omega} = -\frac{2LNx^2}{\omega ^2}e^{-\frac{2LN}{\omega}x} <0.$$ Thus, the function $g_x (\omega)$ is monotonically decreasing with respect to $\omega$, for any fixed $x$. In particular, when $\omega_0 < \omega_{-1} < \omega$, we have: $$g_{x=a} (\omega)< g_{x=a} (\omega_0) = F (a, \omega_0) = f_{\omega_0} (a) =0,$$ which, however, implies: $$\label{EqA9} f_{\omega} (a) = F (a, \omega) = g_{x=a} (\omega) < 0.$$ Since we know $f_\omega (0)>0$, it follows from (\[EqA9\]) that the unique real root of $f_\omega (x)=0$ must lie in the interval $[0,a]= [ 0,-\frac{W_0 (\alpha)}{(1-W_0(\alpha)} ]$, as illustrated in Fig. \[figA2\]. [^1]: This work was supported by the National Science Foundation of China under Grants 61271215, 61001074, 61172065, 61071080, and 60825103, in part by Qualcomm Corporation Foundation, in part by the 973 program (2010CB328205, 2010CB328204), and in part by Shanghai 09XD1402200.

big{]}$. Recall that $ f_\omega (0)>0 $ for all $ \omega$, and $ f_\omega (x)$ has a local minima at: $ $ x_0 = \frac{\omega [ 1 - W_0(\alpha)]}{2LN}.$$ When $ \omega=\omega_0 $, denote this local minima as $ x_0 = a$ and $ f_{\omega_0 } (a)=0$. From (\[Eq23\ ]), we have: $ $ a=\frac { \omega_0 [ 1 - W_0 (\alpha)]}{2LN } = -\frac{W_0(\alpha)}{1 - W_0(\alpha) } > 0.$$ For a fixed $ x$, taking the derived function of (\[EqA2\ ]) with obedience to $ \omega$, we obtain: $ $ g'_x (\omega) = \frac{\partial F (x, \omega)}{\partial \omega } = -\frac{2LNx^2}{\omega ^2}e^{-\frac{2LN}{\omega}x } < 0.$$ Thus, the function $ g_x (\omega)$ is monotonically decrease with respect to $ \omega$, for any fixed $ x$. In finical, when $ \omega_0 < \omega_{-1 } < \omega$, we give birth: $ $ g_{x = a } (\omega) < g_{x = a } (\omega_0) = F (a, \omega_0) = f_{\omega_0 } (a) = 0,$$ which, however, implies: $ $ \label{EqA9 } f_{\omega } (a) = F (a, \omega) = g_{x = a } (\omega) < 0.$$ Since we know $ f_\omega (0)>0 $, it follows from (\[EqA9\ ]) that the unique veridical root of $ f_\omega (x)=0 $ must lie in the time interval $ [ 0,a]= [ 0,-\frac{W_0 (\alpha)}{(1 - W_0(\alpha) } ] $, as illustrated in Fig. \[figA2\ ]. [ ^1 ]: This work was back by the National Science Foundation of China under Grants 61271215, 61001074, 61172065, 61071080, and 60825103, in part by Qualcomm Corporation Foundation, in part by the 973 plan (2010CB328205, 2010CB328204), and in part by Shanghai 09XD1402200.

big{]}$. Gecall that $f_\omega (0)>0$ for xll $\omega$, and $f_\omega (x)$ has a local mknima at: $$x_0 = \frac{\omega [1-W_0(\alpha)]}{2PN}.$$ When $\omega=\omega_0$, denote thir local mpnima as $z_0=a$ aid $f_{\omega_0} (a)=0$. From (\[Eq23\]), we have: $$a=\fdwc{ \okxga_0 [1-W_0 (\alpha)]}{2LN} =-\ftac{W_0(\alpha)}{1-W_0(\al[ha)} > 0.$$ For a fifea $r$, taking the derivative of (\[EqA2\]) with wespect tl $\omega$, we obtwin: $$b'_v (\omsga) = \frac{\partial F (x, \omega)}{\partial \kmega} = -\hrac{2LNx^2}{\omega ^2}e^{-\ftac{2LN}{\omega}x} <0.$$ Thus, the funchion $g_x (\omega)$ is monotlnically dexreafung with resoect to $\omtgc$, for any fjxed $x$. In particular, when $\omega_0 < \omeya_{-1} < \omega$, wg gagg: $$g_{x=a} (\omega)< j_{x=a} (\omvga_0) = F (a, \omega_0) = f_{\omegd_0} (a) =0,$$ whoch, however, imilies: $$\lavel{EqA9} f_{\omega} (a) = F (a, \omega) = g_{x=a} (\omega) < 0.$$ Since wa inow $f_\omega (0)>0$, it filoows xrom (\[EqA9\]) rhag tge uhique geam root of $r_\omega (x)=0$ muwt lie in the interfaj $[0,a]= [ 0,-\frac{W_0 (\alpga)}{(1-W_0(\alpra)} ]$, as illustrated in Fig. \[figA2\]. [^1]: This work fas supported by the Natiobal Science Foundatioj of Chinw under Grants 61271215, 61001074, 61172065, 61071080, and 60825103, in part by Qualcomm Corposatioi Wouueatiov, ij part by the 973 program (2010CB328205, 2010CB328204), and in part by Srznbhsi 09XD1402200.

big{]}$. Recall that $f_\omega (0)>0$ for all $f_\omega has a minima at: $$x_0 denote local minima as and $f_{\omega_0} (a)=0$. (\[Eq23\]), we have: $$a=\frac{ \omega_0 [1-W_0 =-\frac{W_0(\alpha)}{1-W_0(\alpha)} > 0.$$ For a fixed $x$, taking the derivative of (\[EqA2\]) with to $\omega$, we obtain: $$g'_x (\omega) = \frac{\partial F (x, \omega)}{\partial \omega} = ^2}e^{-\frac{2LN}{\omega}x} Thus, function (\omega)$ is monotonically decreasing with respect to $\omega$, for any fixed $x$. In particular, when $\omega_0 \omega_{-1} < \omega$, we have: $$g_{x=a} (\omega)< g_{x=a} = F (a, \omega_0) f_{\omega_0} (a) =0,$$ which, however, $$\label{EqA9} (a) = (a, = (\omega) < 0.$$ we know $f_\omega (0)>0$, it follows from (\[EqA9\]) that the unique real root of $f_\omega (x)=0$ must in the [ 0,-\frac{W_0 ]$, illustrated Fig. \[figA2\]. [^1]: was supported by the National Science under Grants 61271215, 61001074, 61172065, 61071080, and 60825103, part by Corporation Foundation, in part by the program (2010CB328205, 2010CB328204), and in part by Shanghai

big{]}$. Recall that $f_\omega (0)>0$ for all $\Omega$, and $f_\oMega (x)$ Has A loCaL minIma aT: $$x_0 = \frac{\omega [1-W_0(\alPHa)]}{2LN}.$$ when $\omega=\omega_0$, denote thIs locAl MInimA As $X_0=a$ and $F_{\omega_0} (a)=0$. fRoM (\[eQ23\]), we HaVe: $$A=\frAc{ \OMeGa_0 [1-W_0 (\alPha)]}{2lN} =-\frac{W_0(\Alpha)}{1-W_0(\alphA)} > 0.$$ FoR a Fixed $x$, taking THe Derivative Of (\[EQA2\]) with respecT to $\Omega$, wE oBtaIN: $$g'_x (\omEga) = \Frac{\pArtial f (X, \omega)}{\Partial \omEgA} = -\Frac{2LNX^2}{\Omega ^2}e^{-\fRAC{2Ln}{\omeGa}x} <0.$$ Thus, the functioN $G_x (\OMega)$ is monotoniCally dEcREaSINg wIth Respect to $\oMeGa$, for ANy fixed $X$. in PARTicULar, when $\omega_0 < \oMega_{-1} < \omega$, we HAve: $$G_{x=a} (\omeGa)< G_{x=a} (\OMega_0) = F (a, \Omega_0) = F_{\oMEga_0} (A) =0,$$ which, howevEr, imPlies: $$\labeL{EqA9} f_{\oMEga} (a) = F (a, \oMEga) = g_{x=a} (\oMega) < 0.$$ SiNce We kNow $f_\OMeGa (0)>0$, It fOlLOws FRoM (\[Eqa9\]) ThaT the uniqUe ReAl rooT of $f_\OMEGA (x)=0$ muSt lIe in The inTerval $[0,a]= [ 0,-\frac{W_0 (\aLphA)}{(1-W_0(\alPHa)} ]$, aS illuStratEd in fiG. \[figA2\]. [^1]: this woRk was SuPported by the NatIonaL Science FOunDaTioN oF ChinA Under GRanTs 61271215, 61001074, 61172065, 61071080, aNd 60825103, in parT by QualCOmm coRPORaTion Foundation, in paRt BY ThE 973 program (2010cB328205, 2010CB328204), anD In PaRT by ShangHaI 09XD1402200.

big{]}$. Recall that $f_\ omega (0)> 0$ fo r a ll$\ omeg a$,and $f_\omega( x)$has a local minima at: $$x_ 0= \fr a c{ \omeg a [1-W_ 0 (\ a l pha )] }{ 2LN }. $ $When$\o mega=\o mega_0$, d eno te this localm in ima as $x_ 0=a $ and $f_{\o meg a_0} ( a) =0$ . From (\ [Eq23 \]), w e have: $$a=\fra c{ \omega _ 0 [1-W_ 0 (\ alph a)]}{2LN} =-\frac { W_ 0 (\alpha)}{1-W_ 0(\alp ha ) }> 0.$ $ F or a fixed $ x$, t a king th e d e r i vat i ve of (\[EqA2 \]) with re s pec t to $ \o meg a $, weobtai n: $$g '_x (\omega ) =\frac{\pa rtialF (x, \o m ega)}{\ partia l \ ome ga}= - \f rac {2 L Nx^ 2 }{ \om e ga^2}e^{-\ fr ac {2LN} {\om e g a } x} < 0.$ $ Th us, t he function $ g_x (\o m ega )$ is mono toni ca lly d ecreas ing w it h respect to $\ omeg a$, for a nyfi xed $ x$. I n parti cul ar, when $ \omega_ 0 <\o m e g a_ {-1} < \omega$, we h a v e: $$g_{x= a} (\o m eg a) < g_{x=a} ( \om ega_ 0 ) = F(a,\ om ega_0) = f_{\o m eg a_ 0} (a)=0 ,$$ wh ic h,how ever, impl ies: $ $\label{ EqA9} f_{\omega} ( a ) = F (a, \om e ga ) =g _{x= a}(\omega) <0.$$ Sinc e we kn ow$ f_\om ega ( 0) > 0$ , it follows from (\ [E qA9\]) that the unique r eal root o f $ f_\omega (x) = 0$ must lie in th e int erval $[0, a ]= [ 0,- \frac {W_0 (\a lpha)}{(1 - W _0(\alph a)} ]$ , a s i l l us trated in Fig . \[fi gA 2\]. [ ^1] : Thiswor k w assup po rted by t he Natio na lSc ie nce Foun d ation of C hin aund er Gr a nts 61 27121 5, 6 10 01 0 74, 611720 6 5, 6 1071 08 0, and 60 82 5103, inp art by Qua lcomm Cor por a tion F ou ndation , in part byth e 973 prog ra m ( 2010CB 3 2 8205, 20 10CB328204), and in par t by Sha ngh ai 09 XD14 02200.

big{]}$. Recall that_$f_\omega (0)>0$_for all $\omega$, and_$f_\omega (x)$_has_a local_minima_at: $$x_0 =_\frac{\omega [1-W_0(\alpha)]}{2LN}.$$ When_$\omega=\omega_0$, denote this local_minima as $x_0=a$_and_$f_{\omega_0} (a)=0$. From (\[Eq23\]), we have: $$a=\frac{ \omega_0 [1-W_0 (\alpha)]}{2LN} =-\frac{W_0(\alpha)}{1-W_0(\alpha)} > 0.$$ For_a_fixed $x$,_taking_the_derivative of (\[EqA2\]) with respect_to $\omega$, we obtain: $$g'_x_(\omega) =_\frac{\partial F (x, \omega)}{\partial \omega} = -\frac{2LNx^2}{\omega ^2}e^{-\frac{2LN}{\omega}x}_<0.$$_Thus, the function_$g_x (\omega)$ is monotonically decreasing with respect to $\omega$,_for any fixed $x$. In particular,_when $\omega_0 <_\omega_{-1}_<_\omega$, we have: $$g_{x=a}_(\omega)< g_{x=a} (\omega_0) = F (a,_\omega_0) = f_{\omega_0} (a) =0,$$ which,_however, implies: $$\label{EqA9} f_{\omega} (a) =_F (a, \omega) = g_{x=a} (\omega)_< 0.$$ Since we know_$f_\omega (0)>0$,_it follows from (\[EqA9\]) that_the unique real_root of_$f_\omega (x)=0$ must_lie in the interval $[0,a]= [_0,-\frac{W_0 (\alpha)}{(1-W_0(\alpha)} ]$,_as illustrated in Fig. \[figA2\]. [^1]: This_work_was supported by_the_National_Science Foundation_of China under_Grants_61271215, 61001074,_61172065,_61071080, and 60825103, in part by_Qualcomm_Corporation Foundation, in part by the 973_program (2010CB328205, 2010CB328204), and_in_part by Shanghai 09XD1402200.

}\varphi_{1}^2 &=& 0 \\[5pt] \operatorname{d}\varphi_{1}^3 &=& - \varphi_{1}^2\wedge\bar\varphi_{1}^1 \end{array} \right. \;.$$ Observe that these last equations yield to the following equivalent definition of the complex structure $J_1$: $$J_1 e^1 \;=\; -e^2 \;, \qquad J_1 e^3 \;=\; e^4 \;, \qquad J_1 e^5 \;=\; e^6 \;.$$ As previously said, $J_1$ is complex-$\mathcal{C}^\infty$-pure-and-full at every stage in the sense of Li and Zhang (see [@latorre-otal-ugarte-villacampa Proposition 4] as regards to the first stage, see also [@latorre-ugarte]). In fact, one has $$\begin{aligned} H^1_{dR}(M;{\mathbb{C}}) &=& \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^1\right],\, \left[\varphi_{1}^2\right] \right\rangle}_{H^{(1,0)}_{J_1}(M)} \,\ \oplus \,\ \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{\bar1}\right],\, \left[\varphi_{1}^{\bar2}\right] \right\rangle}_{H^{(0,1)}_{J_1}(M)} \;, \\[5pt] H^2_{dR}(M;{\mathbb{C}}) &=& \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{12}\right],\, \left[\varphi_{1}^{23}\right] \right\rangle}_{H^{(2,0)}_{J_1}(M)} \,\ \oplus \,\ \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{\bar1\bar2}\right],\, \left[\varphi_{1}^{\bar2\bar3}\right] \right\rangle}_{H^{(0,2)}_{J_1}(M)} \\[5pt] & \oplus & \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{1\bar1}\right],\, \left[\varphi_{1}^{2\bar2}\right],\, \left[\

} \varphi_{1}^2 & = & 0 \\[5pt ] \operatorname{d}\varphi_{1}^3 & = & - \varphi_{1}^2\wedge\bar\varphi_{1}^1 \end{array } \right. \;.$$ Observe that these last equations yield to the follow equivalent definition of the complex social organization $ J_1 $: $ $ J_1 e^1 \;=\; -e^2 \; , \qquad J_1 e^3 \;=\; e^4 \; , \qquad J_1 e^5 \;=\; e^6 \;.$$ As previously said, $ J_1 $ is complex-$\mathcal{C}^\infty$-pure - and - full at every stagecoach in the common sense of Li and Zhang (see [ @latorre - otal - ugarte - villacampa Proposition 4 ] as regards to the first phase, see also [ @latorre - ugarte ]). In fact, one has $ $ \begin{aligned } H^1_{dR}(M;{\mathbb{C } }) & = & \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^1\right],\, \left[\varphi_{1}^2\right ] \right\rangle}_{H^{(1,0)}_{J_1}(M) } \,\ \oplus \,\ \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{\bar1}\right],\, \left[\varphi_{1}^{\bar2}\right ] \right\rangle}_{H^{(0,1)}_{J_1}(M) } \; , \\[5pt ] H^2_{dR}(M;{\mathbb{C } }) & = & \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{12}\right],\, \left[\varphi_{1}^{23}\right ] \right\rangle}_{H^{(2,0)}_{J_1}(M) } \,\ \oplus \,\ \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{\bar1\bar2}\right],\, \left[\varphi_{1}^{\bar2\bar3}\right ] \right\rangle}_{H^{(0,2)}_{J_1}(M) } \\[5pt ] & \oplus & \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{1\bar1}\right],\, \left[\varphi_{1}^{2\bar2}\right],\, \left[\

}\varohi_{1}^2 &=& 0 \\[5pt] \operatorname{d}\vavphi_{1}^3 &=& - \varphi_{1}^2\wedyw\bar\vacphi_{1}^1 \end{zrray} \rigft. \;.$$ Observe that these last eeuqtionw yield to the followivg equivapent defunitmon of the complxs strucbbre $J_1$: $$M_1 e^1 \;=\; -z^2 \;, \qquad J_1 e^3 \;=\; e^4 \;, \qquad J_1 e^5 \;=\; e^6 \;.$$ As previouvlh daid, $J_1$ is complex-$\mathcal{C}^\infty$-pure-agd-full st every stage ig tht sqnse of Li and Zhang (see [@latorre-otal-ugzrte-vilnacampa Propoxition 4] as regards to the virsh stage, see also [@lwtorre-ugartg]). In dact, one has $$\begin{aligned} H^1_{dR}(M;{\mathgb{C}}) &=& \underbrace{{\mathbb{C}}\left\langld \lefc[\varphi_{1}^1\righj],\, \mevj[\varphi_{1}^2\right] \righn\rangle}_{H^{(1,0)}_{J_1}(M)} \,\ \oikus \,\ \utderbrave{{\mathbb{C}}\left\lsngne \oeft[\varphi_{1}^{\bar1}\right],\, \leht[\varphi_{1}^{\bar2}\right] \rigrt\rangle}_{H^{(0,1)}_{B_1}(M)} \;, \\[5pt] H^2_{dR}(M;{\mathbb{C}}) &=& \ubderbtace{{\mdthbc{X}}\lewt\lznjle \left[\vwrpii_{1}^{12}\right],\, \leff[\varphi_{1}^{23}\righr] \right\rangle}_{H^{(2,0)}_{J_1}(M)} \,\ \okluf \,\ \underbrace{{\mzthbb{C}}\jest\langle \left[\varphi_{1}^{\bar1\bar2}\right],\, \left[\varpvi_{1}^{\bzr2\bar3}\right] \right\rangle}_{H^{(0,2)}_{H_1}(M)} \\[5pt] & \oplus & \underbrwce{{\mathbb{S}}\left\langle \left[\varphi_{1}^{1\bar1}\right],\, \left[\varphi_{1}^{2\bar2}\right],\, \left[\

}\varphi_{1}^2 &=& 0 \\[5pt] \operatorname{d}\varphi_{1}^3 &=& - \right. Observe that last equations yield of complex structure $J_1$: e^1 \;=\; -e^2 \qquad J_1 e^3 \;=\; e^4 \;, J_1 e^5 \;=\; e^6 \;.$$ As previously said, $J_1$ is complex-$\mathcal{C}^\infty$-pure-and-full at every in the sense of Li and Zhang (see [@latorre-otal-ugarte-villacampa Proposition 4] as regards the stage, also In fact, one has $$\begin{aligned} H^1_{dR}(M;{\mathbb{C}}) &=& \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^1\right],\, \left[\varphi_{1}^2\right] \right\rangle}_{H^{(1,0)}_{J_1}(M)} \,\ \oplus \,\ \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{\bar1}\right],\, \right\rangle}_{H^{(0,1)}_{J_1}(M)} \;, \\[5pt] H^2_{dR}(M;{\mathbb{C}}) &=& \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{12}\right],\, \left[\varphi_{1}^{23}\right] \,\ \oplus \,\ \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{\bar2\bar3}\right] \right\rangle}_{H^{(0,2)}_{J_1}(M)} \\[5pt] & \oplus \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{2\bar2}\right],\, \left[\

}\varphi_{1}^2 &=& 0 \\[5pt] \operatorname{d}\varpHi_{1}^3 &=& - \varphi_{1}^2\weDge\baR\vaRphI_{1}^1 \eNd{arRay} \rIght. \;.$$ Observe thaT ThesE last equations yield to tHe folLoWIng eQUiValenT definiTIoN OF thE cOmPleX sTRuCture $j_1$: $$J_1 e^1 \;=\; -E^2 \;, \qquad J_1 E^3 \;=\; e^4 \;, \qquad J_1 e^5 \;=\; e^6 \;.$$ as pReViously said, $J_1$ IS cOmplex-$\mathCal{c}^\infty$-pure-anD-fuLl at evErY stAGe in tHe sEnse oF Li and zHang (seE [@latorre-oTaL-Ugarte-VIllacamPA prOposItion 4] as regards to tHE fIRst stage, see alsO [@latorRe-UGaRTE]). In FacT, one has $$\begIn{AlignED} H^1_{dR}(M;{\maTHbB{c}}) &=& \UNdeRBrace{{\mathbb{C}}\lEft\langle \leFT[\vaRphi_{1}^1\riGhT],\, \leFT[\varphI_{1}^2\righT] \rIGht\Rangle}_{H^{(1,0)}_{J_1}(M)} \,\ \opLus \,\ \uNderbrace{{\Mathbb{c}}\Left\lanGLe \left[\vArphi_{1}^{\bAr1}\rIghT],\, \lefT[\VaRpHi_{1}^{\bAr2}\RIghT] \RiGht\RAngLe}_{H^{(0,1)}_{J_1}(M)} \;, \\[5pt] H^2_{DR}(m;{\mAthbb{c}}) &=& \undERBRAce{{\mAthBb{C}}\lEft\laNgle \left[\varphI_{1}^{12}\riGht],\, \lEFt[\vArphi_{1}^{23}\Right] \RighT\rAngle}_{h^{(2,0)}_{J_1}(M)} \,\ \oplUs \,\ \undErBrace{{\mathbb{C}}\lefT\lanGle \left[\vaRphI_{1}^{\bAr1\bAr2}\Right],\, \LEft[\varPhi_{1}^{\Bar2\Bar3}\righT] \right\rANglE}_{H^{(0,2)}_{j_1}(m)} \\[5PT] & \oPlus & \underbrace{{\mathBb{c}}\LEfT\langle \lEft[\varPHi_{1}^{1\BaR1}\Right],\, \lefT[\vArpHi_{1}^{2\baR2}\RIght],\, \lEft[\

}\varphi_{1}^2 &=& 0 \\[5p t] \opera torna me{ d}\ va rphi _{1} ^3 &=& - \varp h i_{1 }^2\wedge\bar\varphi_{ 1}^1\e n d{ar r ay } \ri ght. \; . $$ Obs er ve th at th ese l ast equati ons yieldtoth e followinge qu ivalent de fin ition of the co mplexst ruc t ure $ J_1 $: $$ J_1 e^ 1 \;=\; -e^2 \;, \ q quad J _ 1 e^3 \ ; = \; e^4 \;, \qquad J_1 e ^ 5\ ;=\; e^6 \;.$$ As p re v io u s lysai d, $J_1$ i scompl e x-$\mat h ca l { C }^\ i nfty$-pure-an d-full at e v ery stage i n t h e sens e ofLi and Zhang (see [@l atorre-ot al-uga r te-vill a campa P roposi tio n 4 ] as re ga rds t o th e f irs t st age, see a ls o [@l ator r e - u gart e]) . In fact , one has $$\ beg in{a l ign ed} H^1_{ dR}( M; {\mat hbb{C} }) &= &\underbrace{{\m athb b{C}}\lef t\l an gle \ left[ \ varphi _{1 }^1 \right] ,\, \le f t[\ va r p h i_ {1}^2\right] \righ t\ r a ng le}_{H^{ (1,0)} _ {J _1 } (M)} \, \\op lus\ , \ \un derb r ac e{{\math bb{C}} \ le ft \langle \ left[\ va rph i_{ 1}^{\ b ar1} \right ],\, \le ft[\v a rphi_{1}^{\bar 2 }\right] \rig h t\ r a ng l e}_{ H^{ (0,1)}_{J_1 }(M) } \;, \\[ 5 pt ] H ^2_{d R}(M; {\ m at h bb{C}}) &=& \underb ra ce{{\m athbb {C}}\left\lan gle \left[ \ v a rphi_{1} ^{12 } \r i ght],\, \left[ \varp hi_{1}^{23 } \right]\righ t\rangle }_{H^{(2, 0 ) }_{J_1}( M)} \, \ \ opl u s \ ,\ \underbrac e { {\ma th bb{C}}\ lef t\langl e \ lef t[\ var ph i_{1}^{\b ar1\bar2 }\ ri gh t] ,\, \lef t [\varphi _{ 1}^ {\ bar 2\bar 3 }\righ t] \r ight \r an g le} _{H^{(0 , 2) } _ {J_1 }( M) } \\ [5p t] & \ oplu s &\underb race{{\ma thb b {C}} \l ef t\langl e \left[\varp hi _{1}^{1\ba r1 }\r ight], \ , \left[\ varphi_{1}^{2\bar2}\rig h t],\, \ lef t[\

}\varphi_{1}^2 &=&_0 \\[5pt] _\operatorname{d}\varphi_{1}^3 &=& - \varphi_{1}^2\wedge\bar\varphi_{1}^1 \end{array} \right._\;.$$ Observe that_these_last equations_yield_to the following_equivalent definition of_the complex structure $J_1$:_$$J_1 e^1 \;=\;_-e^2_\;, \qquad J_1 e^3 \;=\; e^4 \;, \qquad J_1 e^5 \;=\; e^6 \;.$$ As previously_said,_$J_1$ is_complex-$\mathcal{C}^\infty$-pure-and-full_at_every stage in the sense_of Li and Zhang (see_[@latorre-otal-ugarte-villacampa Proposition_4] as regards to the first stage, see_also_[@latorre-ugarte]). In fact,_one has $$\begin{aligned} H^1_{dR}(M;{\mathbb{C}}) &=& \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^1\right],\, \left[\varphi_{1}^2\right] \right\rangle}_{H^{(1,0)}_{J_1}(M)}_ \,\ \oplus \,\ \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{\bar1}\right],\, \left[\varphi_{1}^{\bar2}\right]_\right\rangle}_{H^{(0,1)}_{J_1}(M)} \;, \\[5pt] _H^2_{dR}(M;{\mathbb{C}})_&=&_\underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{12}\right],\, \left[\varphi_{1}^{23}\right] \right\rangle}_{H^{(2,0)}_{J_1}(M)}_\,\ \oplus \,\ \underbrace{{\mathbb{C}}\left\langle \left[\varphi_{1}^{\bar1\bar2}\right],\, \left[\varphi_{1}^{\bar2\bar3}\right]_\right\rangle}_{H^{(0,2)}_{J_1}(M)} \\[5pt] & \oplus & \underbrace{{\mathbb{C}}\left\langle_\left[\varphi_{1}^{1\bar1}\right],\, \left[\varphi_{1}^{2\bar2}\right],\, \left[\

CBC\_SHA with TLS 1.0). Bhargavan et al. [@weak2016blockciphers] found that ciphers using 64-bit blocks (e.g., 3DES) are vulnerable to secret key disclosure. The official website toytalk.com can be used by parents to review the child’s recordings. The server certificate uses an RSA 2048-bit key and is signed by Amazon using SHA256 and an RSA 2048-bit key. The server does not support weak cipher suites and uses cryptographic libraries patched against known cryptographic attacks. However, there is no restriction on the number of login attempts, and the login can be bruteforced. We exploited this *[Online-password-bruteforce]{}* vulnerability during the course of testing to bruteforce our own account. This flaw was originally reported in January 2016[@Somerset2016barbie] and our experiments show that it is still not fixed. To investigate communication between the toy and the app, we use the following specific setup for [Hello Barbie]{}. We use two WiFi adapters on our test machine, one in AP mode as an impostor Hello Barbie, and the second to connect to the real Hello Barbie, exploiting its unprotected hotspot. We route all traffic from the impostor to the real toy. In the app, both toys, real and impostor, are listed, and we connect to the impostor. The smartphone hosting the app, meanwhile, is configured to forward all traffic through a Burp Suite proxy. This configuration requires the app to be patched to accept certificates signed by Burp Suite CA, and the app client certificate bundle added to Burp Suite as the client certificate. This allows the app to authenticate itself to the toy, enabling us to intercept communication between them. The toy authenticates to the app using a certificate issued to 192.168.10.1, signed by “ToyTalk CA”, and valid to 2030. Once the TLS handshake is complete, the app sends the parent account ID and WiFi configuration, including WiFi SSID and password, to the toy. The toy can be configured using any companion app, and no parent authentication is required to pair with the toy. Because the toy uses the parent’s account ID as the sole way to relate the child’s profile, including voice recordings, to the parent, an adversary need only reconfigure the toy to use the adversary’s account

CBC\_SHA with TLS 1.0). Bhargavan et al.   [ @weak2016blockciphers ] found that ciphers using 64 - bit block (for example, 3DES) are vulnerable to secret key disclosure. The official web site toytalk.com can be used by parent to review the child ’s recordings. The server security use an RSA 2048 - bit samara and is signed by Amazon using SHA256 and an RSA 2048 - morsel key. The server does not defend decrepit cipher suites and use cryptographic libraries patch against known cryptographic attacks. However, there exist no restriction on the number of login attempts, and the login can be bruteforced. We exploited this * [ Online - password - bruteforce ] { } * vulnerability during the path of testing to bruteforce our own account. This flaw was originally reported in January 2016[@Somerset2016barbie ] and our experiment show that it is still not fixed. To investigate communication between the toy and the app, we use the following specific setup for [ Hello Barbie ] { }. We use two WiFi adapters on our test machine, one in AP mode as an impostor Hello Barbie, and the second to connect to the real Hello Barbie, exploiting its unprotected hotspot. We route all dealings from the imposter to the real toy. In the app, both toys, veridical and impostor, are listed, and we connect to the imposter. The smartphone hosting the app, meanwhile, is configure to forward all traffic through a Burp Suite proxy. This configuration requires the app to be patch to accept certificates signed by Burp Suite CA, and the app client certificate bundle added to Burp Suite as the client certificate. This allows the app to authenticate itself to the toy, enable us to intercept communication between them. The plaything authenticates to the app using a certificate issued to 192.168.10.1, signed by “ ToyTalk CA ”, and valid to 2030. Once the TLS handshake is accomplished, the app sends the parent account ID and WiFi configuration, include WiFi SSID and password, to the toy. The toy can be configured using any companion app, and no parent authentication is want to pair with the miniature. Because the toy uses the parent ’s history ID as the sole manner to relate the child ’s visibility, including voice recording, to the parent, an adversary need only reconfigure the toy to practice the adversary ’s account

CBC\_DHA with TLS 1.0). Bhargavan tt al. [@weak2016blockcipkwrs] fonnd thaf cipherr using 64-bit blocks (e.g., 3DES) arx vuoneravle to secret key discuosure. The officiao weusite toytalk.com can be mfed gn parznvs to review thg child’s recmrdings. The sesvdr certificate uses an RSA 2048-bit key anq is sibnfd by Amazon ufing FHA256 znd an RSA 2048-bit key. The server does not sukport weak cipher xuites and uses cryptograpjic pibraries patched wgainst knoqn cwtptographic xttacks. However, there js no restriction on the number of lpgin attemkcw, ajg the login can fe bruteforccc. We efploitec this *[Online-psssxord-vruteforce]{}* vulnerabilmty during the coursg of testitg to bruteforce oue iwn awcoutt. Tfus wlas xas origijalmy reportes in Januart 2016[@Somerset2016barbie] anc jlt experimenta show trat it is still not fixed. To investigate cojmunication between the toy and the app, we ude the fojlowing specific setup for [Hello Barbie]{}. We use twm WiFm xdaktcvs ov okr test machine, one in AP mode as an impostor Geklp Barbie, and tme second to connevt tp the real Heluo Baryje, exploiting its unorotectgd hotwpot. We rjute all traffic from the imposror to the rval roy. In the app, botk toys, real cnd imkostor, are listed, and we connzct to the impostlr. The smzftphone hosting ghe a[p, meanwhile, is configured to forwacd alk traffkc tnrough a Burp Sulte pvmxy. This configurahion tequiras the app to be patched to accept certifmrates signed ny Bugp Suite EA, and the app cliegt certificate bundle addea to Burp Auite av the cliene certificate. Jhis allows tie app to autyentucate igrelf to the tou, enabliny us ro intercept commukicatkkn between them. Tht roy authenticayes to tje a[[ using a cestifkcagr issjed to 192.168.10.1, sigked by “YoyTalk CA”, and valid to 2030. Once the TLS handxhwke is cimplete, ehe app sends the parent accounu ID aid WiFm confogutation, including WiFi SSID and lassword, ho bhe toy. The tor cak be configureb using any companion app, and no parent euthentication is requieed to pair with thg tpy. Because vhe tor uses tha parent’s account ID as the sole way uo relate the child’s prkfile, hnclufing voice recordings, to the parent, an adversary need only reconfigure thw toy vo use the adberssry’s dceouut

CBC\_SHA with TLS 1.0). Bhargavan et al. that using 64-bit (e.g., 3DES) are The website toytalk.com can used by parents review the child’s recordings. The server uses an RSA 2048-bit key and is signed by Amazon using SHA256 and RSA 2048-bit key. The server does not support weak cipher suites and uses libraries against cryptographic However, there is no restriction on the number of login attempts, and the login can be We exploited this *[Online-password-bruteforce]{}* vulnerability during the course testing to bruteforce our account. This flaw was originally in 2016[@Somerset2016barbie] and experiments that is still not To investigate communication between the toy and the app, we use the following specific setup for [Hello We use adapters on test one AP mode as Hello Barbie, and the second to real Hello Barbie, exploiting its unprotected hotspot. We all traffic the impostor to the real toy. the app, both toys, real and impostor, are and we connect to the impostor. The smartphone hosting the app, meanwhile, is configured to traffic through a Burp proxy. This configuration the to patched accept certificates by Burp Suite CA, and the app client certificate bundle added Burp Suite as the client certificate. This allows the app itself the toy, enabling to intercept communication between The authenticates to the app certificate to “ToyTalk and to 2030. Once the handshake is complete, the app the parent account ID SSID and password, to the toy. The toy be configured using any companion app, and parent authentication is required to pair with the toy. Because the toy the parent’s as the sole way to relate the child’s including voice recordings, to parent, an adversary need only reconfigure the toy to the account

CBC\_SHA with TLS 1.0). Bhargavan et aL. [@weak2016blockCipheRs] fOunD tHat cIpheRs using 64-bit blocKS (e.g., 3DeS) are vulnerable to secreT key dIsCLosuRE. THe offIcial weBSiTE ToyTaLk.Com CaN Be Used bY paRents to Review the cHilD’s Recordings. ThE SeRver certifIcaTe uses an RSA 2048-bIt kEy and iS sIgnED by AmAzoN usinG SHA256 anD An RSA 2048-bIt key. The sErVEr does NOt suppoRT WeAk ciPher suites and uses CRyPTographic libraRies paTcHEd AGAinSt kNown cryptoGrAphic ATtacks. HOWeVER, TheRE is no restrictIon on the numBEr oF login AtTemPTs, and tHe logIn CAn bE bruteforceD. We eXploited tHis *[OnlINe-passwORd-bruteForce]{}* vUlnEraBiliTY dUrIng ThE CouRSe Of tEStiNg to brutEfOrCe our Own aCCOUNt. ThIs fLaw wAs oriGinally reportEd iN JanUAry 2016[@somerSet2016baRbie] AnD our eXperimEnts sHoW that it is still nOt fiXed. To inveStiGaTe cOmMunicATion beTweEn tHe toy anD the app, WE usE tHE FOlLowing specific setuP fOR [heLlo BarbiE]{}. We use TWo wifI adapterS oN ouR tesT MAchinE, one IN Ap mode as aN imposTOr heLlo BarbIe, And the SeConD to ConneCT to tHe real hello BarBie, exPLoiting its unprOTected hotspot. wE rOUTe ALl trAffIc from the imPostOR to tHe reAL tOy. IN The apP, both ToYS, rEAl and impostor, are lisTeD, and we ConneCt to the impostOr. The smartPHONe hostinG the APp, MEanwhile, is confIgureD to forward ALl traffiC throUgh a Burp suite proxY. tHis confiGurAtiOn rEquIREs The app to be patCHEd to AcCept cerTifIcates sIgnEd bY BuRp SUiTe CA, and thE app clieNt CeRtIfIcaTe bunDLe added tO BUrp suIte As the CLient cErtifIcatE. THiS AllOws the aPP tO AUtheNtIcAte iTseLf To the Toy, eNAblIng us to Intercept ComMUnicAtIoN betweeN them. The toy auThEnticates tO tHe aPp usinG A CertificAte issued to 192.168.10.1, signed by “ToyTALk CA”, and ValId to 2030. ONce tHe TLS handShaKe is coMplETe, the aPp sendS the pArEnt ACCount id AnD Wifi ConfiguratION, inCludiNg wiFi sSID and Password, to the toy. ThE Toy Can be configurEd uSing ANY cOmpANiON apP, aND no PARent authenticatIon is requiReD To Pair with thE Toy. beCause thE toy useS the pARent’s acCount ID as The sole waY tO relATE thE child’s proFile, inclUding voicE RecorDInGs, to tHe pArent, aN aDveRsary Need onLY reConfiGure thE tOy to usE the aDvErsary’s aCcount

CBC\_SHA with TLS 1.0). Bh argavan et al.[@w eak 20 16bl ockc iphers] foundt hatciphers using 64-bit b locks ( e .g., 3D ES) a re vuln e ra b l e t ose cre tk ey disc los ure. T he officia l w eb site toytalk . co m can be u sed by parentstoreview t hec hild’ s r ecord ings.T he ser ver certi fi c ate us e s an RS A 20 48-b it key and is sig n ed by Amazon usin g SHA2 56 an d anRSA 2048-bitke y. Th e server do e s not support weakcipher suit e s a nd use scry p tograp hic l ib r ari es patchedagai nst known crypt o graphic attacks . Howe ver , t here is n o r es t ric t io n o n th e number o flogin att e m p t s, a ndthelogin can be brute for ced. Weexplo itedthis * [Onli ne-pas sword -b ruteforce]{}* v ulne rabilitydur in g t he cour s e of t est ing to bru teforce our o w n ac count. This flaw w as o ri ginallyreport e din January20 16[ @Som e r set20 16ba r bi e] and o ur exp e ri me nts sho wthat i tissti ll no t fix ed. T o invest igate communicationb etween the to y a n d t h e ap p,we use thefoll o wing spe c if ics etupfor [ He l lo Barbie]{}. We use t wo WiFiadapt ers on our te st machine , o ne in AP mod e a s an impostor H elloBarbie, an d the sec ond t o connec t to ther e al Hello Ba rbi e,exp l o it ing its unpro t e cted h otspot. We routeall tr aff icfr om the im postor t oth ere altoy.I n the ap p, bo th to ys, r e al and impo stor ,ar e li sted, a n dw e con ne ct tothe i mpost or.T hesmartph one hosti ngt he a pp ,meanwhi le, is config ur ed to forw ar d a ll tra f f ic throu gh a Burp Suite proxy.T his con fig urati on r equires t heapp to be patche d to a ccept c ert i f icate s si gne dby Burp Su i t e C A, an dtheapp cli ent certificate bu n dle added to Bur p S uite a sthe cl i ent c e rti f i cate. This allo ws the app t o a uthenticat e it se lf to t he toy, enab l ing usto interc ept commu ni cati o n be tween them . The to y authent i cates to theapp using a ce rtifi cate i s sue d to192.16 8. 10.1,signe dby “ToyT alk CA”, and valid to 2 030. O nce t heTLS hands hak e is complete , th e app send s t heparen t a c count IDa nd Wi F i con figu r ation, in c lu din g Wi Fi SSID and p a ssw ord,tot he toy . T he toy can be con f igured using a ny c o m pan ion app, a nd no parent a uth en t i cation i srequired to pair wi th the t oy. Be causethe toy u se s the p aren t’s accountIDas the sol ewa y to re late t he chi ld’s p r ofil e , including voice reco r d ings, tothe p ar ent, an adve rsary need only recon figure the toyto useth e adve rsa ry ’s account

CBC\_SHA with_TLS 1.0)._Bhargavan et al. [@weak2016blockciphers] found_that ciphers_using_64-bit blocks_(e.g.,_3DES) are vulnerable_to secret key_disclosure. The official website toytalk.com_can be used_by_parents to review the child’s recordings. The server certificate uses an RSA 2048-bit key_and_is signed_by_Amazon_using SHA256 and an RSA_2048-bit key. The server does_not support_weak cipher suites and uses cryptographic libraries patched_against_known cryptographic attacks._However, there is no restriction on the number of_login attempts, and the login can_be bruteforced. We_exploited_this_*[Online-password-bruteforce]{}* vulnerability during the_course of testing to bruteforce our_own account. This flaw was originally_reported in January 2016[@Somerset2016barbie] and our experiments_show that it is still not_fixed. To investigate communication between the_toy and_the app, we use the_following specific setup_for [Hello_Barbie]{}. We use_two WiFi adapters on our test_machine, one in_AP mode as an impostor Hello_Barbie,_and the second_to_connect_to the_real Hello Barbie,_exploiting_its unprotected_hotspot._We route all traffic from the_impostor_to the real toy. In the app,_both toys, real and_impostor,_are listed, and we_connect to the impostor. The_smartphone hosting the app, meanwhile, is_configured to_forward all_traffic through a Burp Suite proxy. This configuration requires the app_to be patched to accept certificates_signed by Burp Suite_CA, and_the_app client certificate_bundle_added to_Burp Suite as the client certificate. This_allows the_app to authenticate itself to the_toy, enabling us to_intercept_communication between them. The toy authenticates_to the app using a certificate_issued to 192.168.10.1, signed by_“ToyTalk_CA”,_and valid to 2030. Once_the TLS handshake is complete, the_app sends the_parent account ID and WiFi configuration, including_WiFi_SSID and password, to the toy. The_toy_can be configured using any companion_app,_and_no parent authentication is required_to pair with the toy. Because_the toy uses the parent’s account ID as the_sole way to_relate the child’s profile, including_voice_recordings,_to the parent, an adversary need only reconfigure the toy_to use_the adversary’s account

R., Lopez, S., & Tytler, D., 2002, A&A, 382, 26 Rest, A., van den Bosch, F.C., Jaffe, W., Tran, H., Tsvetanov, Z., Ford, H.C., Davies, J., & Schafer, J.L, 2001, AJ, 121, 2431 Ros, E., Guirado, J.C., Marcaide, J.M., Perez-Torres, M.A., Falco, E.E., Munoz, J.A., Alberdi, A., & Lara L., 2000, A&A, 362, 845 Schechter, P. L. & Wambsganss, J. 2002,, 580, 685 Schneider, P., Ehlers, J., & Falco, E.E., 1992, Gravitational Lenses, (Springer Verlag: Berlin) Sykes, C.M., Browne, I.W.A., Jackson, N.J., Marlow, D.R., Nair, S., Wilkinson, P.N., Blandford, R.D., Cohen, J., Fassnacht, C.D., Hogg, D., Pearson, T.J., Readhead, A.C.S., Womble, D.S., Myers, S.T., de Bruyn, A.G., Bremer, M., Miley, G.K., & Schilizzi, R.T., 1998, MNRAS, 301, 310 Trotter, C.S., Winn, J.N. & Hewitt, J.N., 2000, ApJ, 535, 671 Walker, M. A. 1998, MNRAS, 294, 307 Weymann, R.J., Latham, D., Roger, J., Angel, P., Green, R.F., Liebert, J., Turnshek, D.A., Turnshek, D.E., & Tyson, J.A., 1980, Nature, 285, 641 Wisotzki, L., Christlieb, N., Liu, M.C., Maza, J., Morgan, N.D., & Schechter, P.L., 1999, A&A, 348, 41 Wisotzki,

R., Lopez, S., & Tytler, D., 2002, A&A, 382, 26 Rest, A., van den Bosch, F.C., Jaffe, W., Tran, H., Tsvetanov, Z., Ford, H.C., Davies, J., & Schafer, J.L, 2001, AJ, 121, 2431 Ros, E., Guirado, J.C., Marcaide, J.M., Perez - Torres, M.A., Falco, E.E., Munoz, J.A., Alberdi, A., & Lara L., 2000, A&A, 362, 845 Schechter, P.   L. & Wambsganss, J.   2002, , 580, 685 Schneider, P., Ehlers, J., & Falco, E.E., 1992, Gravitational Lenses, (Springer Verlag: Berlin) Sykes, C.M., Browne, I.W.A., Jackson, N.J., Marlow, D.R., Nair, S., Wilkinson, P.N., Blandford, R.D., Cohen, J., Fassnacht, C.D., Hogg, D., Pearson, T.J., Readhead, A.C.S., Womble, D.S., Myers, S.T., de Bruyn, A.G., Bremer, M., Miley, G.K., & Schilizzi, R.T., 1998, MNRAS, 301, 310 Trotter, C.S., Winn, J.N. & Hewitt, J.N., 2000, ApJ, 535, 671 Walker, M.   A.   1998, MNRAS, 294, 307 Weymann, R.J., Latham, D., Roger, J., Angel, P., Green, R.F., Liebert, J., Turnshek, D.A., Turnshek, D.E., & Tyson, J.A., 1980, Nature, 285, 641 Wisotzki, L., Christlieb, N., Liu, M.C., Maza, J., Morgan, N.D., & Schechter, P.L., 1999, A&A, 348, 41 Wisotzki,

R., Popez, S., & Tytler, D., 2002, A&A, 382, 26 Rtst, A., van den Bosey, F.C., Jeffe, W., Fran, H., Trvetanov, Z., Ford, H.C., Davies, J., & Sxhafee, J.L, 2001, AJ, 121, 2431 Ros, E., Guiradu, J.C., Marcwide, J.M., Peree-Torres, M.A., Falco, X.S., Munoz, J.A., Algcrdi, C., & Lara L., 2000, A&A, 362, 845 Sghechter, P. L. & Wambsganss, J. 2002,, 580, 685 Rckneider, P., Ehlers, J., & Falco, E.E., 1992, Gravitwtional Lfnses, (Springer Verkwg: Bsglln) Sykes, C.M., Browne, I.W.A., Jackson, N.J., Marlow, D.R., Nair, S., Wikkinson, P.N., Blandford, R.D., Cojen, U., Fassnacht, C.D., Hogh, D., Pearson, T.J., Wwadhead, A.C.S., Domble, D.S., Myers, S.T., de Gruyn, A.G., Bremer, M., Miley, G.K., & Schklizzn, R.T., 1998, MNRAS, 301, 310 Rrohjer, C.S., Winn, O.N. & Heritt, J.N., 2000, ApJ, 535, 671 Walker, M. A. 1998, MNRSS, 294, 307 Weymann, R.J., Lavham, D., Roger, J., Angel, P., Grxen, R.F., Liebert, J., Turgshek, D.A., Duxnshek, D.E., & Tyson, J.A., 1980, Barure, 285, 641 Wismtzkk, L., Zhrjsvlisb, N., Llu, J.C., Maza, J., Jorgan, N.D., & Wchechter, P.L., 1999, A&A, 348, 41 Wosjnaki,

R., Lopez, S., & Tytler, D., 2002, 26 A., van Bosch, F.C., Jaffe, Ford, Davies, J., & J.L, 2001, AJ, 2431 Ros, E., Guirado, J.C., Marcaide, Perez-Torres, M.A., Falco, E.E., Munoz, J.A., Alberdi, A., & Lara L., 2000, A&A, 845 Schechter, P. L. & Wambsganss, J. 2002,, 580, 685 Schneider, P., Ehlers, & E.E., Gravitational (Springer Verlag: Berlin) Sykes, C.M., Browne, I.W.A., Jackson, N.J., Marlow, D.R., Nair, S., Wilkinson, P.N., Blandford, Cohen, J., Fassnacht, C.D., Hogg, D., Pearson, T.J., A.C.S., Womble, D.S., Myers, de Bruyn, A.G., Bremer, M., G.K., Schilizzi, R.T., MNRAS, 310 C.S., Winn, J.N. Hewitt, J.N., 2000, ApJ, 535, 671 Walker, M. A. 1998, MNRAS, 294, 307 Weymann, R.J., Latham, D., J., Angel, R.F., Liebert, Turnshek, Turnshek, & Tyson, J.A., 285, 641 Wisotzki, L., Christlieb, N., J., Morgan, N.D., & Schechter, P.L., 1999, A&A, 41 Wisotzki,

R., Lopez, S., & Tytler, D., 2002, A&A, 382, 26 Rest, A., van dEn Bosch, F.C., JAffe, W., traN, H., TSvEtanOv, Z., FOrd, H.C., Davies, J., & ScHAfer, j.L, 2001, AJ, 121, 2431 Ros, E., Guirado, J.C., MarcaIde, J.M., peREz-ToRReS, M.A., FaLco, E.E., MuNOz, j.a., albErDi, a., & LaRa l., 2000, a&A, 362, 845 schecHteR, P. L. & WambSganss, J. 2002,, 580, 685 SchNeiDeR, P., Ehlers, J., & FalCO, E.e., 1992, GravitatiOnaL Lenses, (SprinGer verlag: beRliN) sykes, c.M., BRowne, i.W.A., JacKSon, N.J., MArlow, D.R., NaIr, s., wilkinSOn, P.N., BlaNDFoRd, R.D., cohen, J., Fassnacht, C.D., hOgG, d., Pearson, T.J., ReadHead, A.C.s., WOMbLE, d.S., MYerS, S.T., de Bruyn, a.G., bremeR, m., Miley, G.k., & scHILIzzI, r.T., 1998, MNRAS, 301, 310 TrotteR, C.S., Winn, J.N. & HeWItt, j.N., 2000, ApJ, 535, 671 WAlKer, m. a. 1998, MNRAS, 294, 307 weymaNn, r.j., LaTham, D., Roger, J., angeL, P., Green, R.F., lieberT, j., TurnshEK, D.A., TurnShek, D.E., & tysOn, J.a., 1980, NatURe, 285, 641 wiSotZkI, l., ChRIsTliEB, N., LIu, M.C., Maza, j., MOrGan, N.D., & scheCHTER, P.L., 1999, A&a, 348, 41 WiSotzKi,

R., Lopez, S., & Tytler,D., 2002,A&A,382 , 2 6 Res t, A ., van den Bos c h, F .C., Jaffe, W., Tran,H., T sv e tano v ,Z., F ord, H. C ., D avi es ,J., & Sc hafer , J .L, 200 1, AJ, 121 , 2 43 1 Ros, E.,G ui rado, J.C. , M arcaide, J.M .,Perez- To rre s , M.A .,Falco , E.E. , Munoz , J.A., A lb e rdi, A . , & Lar a L. , 20 00, A&A, 362, 845 S c hechter, P. L. & Wam bs g an s s , J . 2 002,, 580, 6 85 S c hneider , P . , Ehl e rs, J., & Fal co, E.E., 1 9 92, Gravi ta tio n al Len ses,(S p rin ger Verlag: Ber lin) Syk es, C. M ., Brow n e, I.W. A., Ja cks on, N.J . ,Ma rlo w, D.R . ,Nai r , S ., Wilki ns on , P.N ., B l a n d ford , R .D., Cohe n, J., Fassna cht , C. D .,Hogg, D.,Pear so n, T. J., Re adhea d, A.C.S., Womble , D. S., Myers , S .T .,de Bruy n , A.G. , B rem er, M., Miley, G.K ., & Sc hilizzi, R.T., 199 8, M NR AS, 301, 310 T ro tt e r, C.S., W inn , J. N . & He witt , J .N., 200 0, ApJ , 5 35 , 671 Wa lker,M.  A.  19 98, M N RAS, 294,307 Wey mann, R.J., Latham,D ., Roger, J., An g e l, P.,Gre en, R.F., L iebe r t, J ., T u rn she k , D.A ., Tu rn s he k , D.E., & Tyson, J. A. , 1980 , Nat ure, 285, 641 Wisotzki , L ., Chris tlie b ,N ., Liu, M.C.,Maza, J., Morga n , N.D.,& Sch echter,P.L., 199 9 , A&A, 34 8,41 Wi sot z k i,

R.,_Lopez, S.,_& Tytler, D., 2002,_A&A, 382,_26 Rest,_A., van_den_Bosch, F.C., Jaffe,_W., Tran, H.,_Tsvetanov, Z., Ford, H.C.,_Davies, J., &_Schafer,_J.L, 2001, AJ, 121, 2431 Ros, E., Guirado, J.C., Marcaide, J.M., Perez-Torres, M.A., Falco, E.E.,_Munoz,_J.A., Alberdi,_A.,_&_Lara L., 2000, A&A, 362,_845 Schechter, P. L. & Wambsganss, J. 2002,,_580, 685 Schneider,_P., Ehlers, J., & Falco, E.E., 1992, Gravitational_Lenses,_(Springer Verlag: Berlin) Sykes,_C.M., Browne, I.W.A., Jackson, N.J., Marlow, D.R., Nair, S.,_Wilkinson, P.N., Blandford, R.D., Cohen, J.,_Fassnacht, C.D., Hogg,_D.,_Pearson,_T.J., Readhead, A.C.S., Womble,_D.S., Myers, S.T., de Bruyn, A.G.,_Bremer, M., Miley, G.K., & Schilizzi,_R.T., 1998, MNRAS, 301, 310 Trotter, C.S., Winn,_J.N. & Hewitt, J.N., 2000, ApJ,_535, 671 Walker, M. A. 1998, MNRAS, 294,_307 Weymann, R.J.,_Latham, D., Roger, J., Angel,_P., Green, R.F.,_Liebert, J.,_Turnshek, D.A., Turnshek,_D.E., & Tyson, J.A., 1980, Nature,_285, 641 Wisotzki, L.,_Christlieb, N., Liu, M.C., Maza, J.,_Morgan,_N.D., & Schechter,_P.L.,_1999,_A&A, 348,_41 Wisotzki,

Theorem \[KTI\] to $\tilde \beta_n$, obtaining $$\label{ttt1} \int_{{\Omega}} |\tilde \beta_n - \tilde A_n|^2 \, dx \le C \left(\int_{{\Omega}} |\tilde \beta_n{^{\rm sym}}|^2 \, dx + \big(|\mu_n|({\Omega})\big)^2 \right) \le C|\log {\varepsilon_n}|^2,$$ where $\tilde A_n$ is the average of the anti-symmetric part of $\tilde \beta_n$. Since the average of $\beta_n$ is a symmetric matrix, we have $$\label{ttt2} \int_{{\Omega}_{{\varepsilon_n}}(\mu_n)} |\beta_n|^2 \, dx = \int_{{\Omega}_{{\varepsilon_n}}(\mu_n)} |\beta_n - \tilde A_n|^2 - |\tilde A_n|^2 \, dx \le \int_{{\Omega}_{{\varepsilon_n}}(\mu_n)} |\beta_n - \tilde A_n|^2 \, dx.$$ Finally, by and we conclude $$\int_{{\Omega}_{{\varepsilon_n}}(\mu_n)} |\beta_n|^2 \, dx \le \int_{{\Omega}_{{\varepsilon_n}}(\mu_n)} |\beta_n - \tilde A_n|^2 \, dx \le \int_{{\Omega}} |\tilde \beta_n - \tilde A_n|^2 \, dx \le C|\log {\varepsilon_n}|^2,$$ which gives the desired compactness property for $\beta_n/|\log{\varepsilon_n}|$ in $L^2({\Omega};{\mathbb{M}^{2\times 2}})$. [*Step $3.$ $\mu$ belongs to $H^{-1}({\Omega};{\mathbb{R}}^2)$ and ${{\rm Curl\,}}\beta=\mu$.*]{} Let ${\varphi}\in C^1_0({\Omega})$. It is easy to construct a sequence $\{{\varphi}_n\}\subset H^1_0({\Omega})$ converging to ${\varphi}$ uniformly and strongly in $H^1_0({\Omega})$ and satisfying the property $${\varphi}_n \equiv {\varphi}(x_{i,n}) \qquad \text{ in } B_{{\varepsilon_n}}(x_{i,n})\quad \text{for every } x_{i,

Theorem \[KTI\ ] to $ \tilde \beta_n$, obtaining $ $ \label{ttt1 } \int_{{\Omega } } |\tilde \beta_n - \tilde A_n|^2 \, dx \le C \left(\int_{{\Omega } } |\tilde \beta_n{^{\rm sym}}|^2 \, dx + \big(|\mu_n|({\Omega})\big)^2 \right) \le C|\log { \varepsilon_n}|^2,$$ where $ \tilde A_n$ is the average of the anti - symmetric part of $ \tilde \beta_n$. Since the average of $ \beta_n$ is a symmetric matrix, we get $ $ \label{ttt2 } \int_{{\Omega}_{{\varepsilon_n}}(\mu_n) } |\beta_n|^2 \, dx = \int_{{\Omega}_{{\varepsilon_n}}(\mu_n) } |\beta_n - \tilde A_n|^2 - |\tilde A_n|^2 \, dx \le \int_{{\Omega}_{{\varepsilon_n}}(\mu_n) } |\beta_n - \tilde A_n|^2 \, dx.$$ ultimately, by and we conclude $ $ \int_{{\Omega}_{{\varepsilon_n}}(\mu_n) } |\beta_n|^2 \, dx \le \int_{{\Omega}_{{\varepsilon_n}}(\mu_n) } |\beta_n - \tilde A_n|^2 \, dx \le \int_{{\Omega } } |\tilde \beta_n - \tilde A_n|^2 \, dx \le C|\log { \varepsilon_n}|^2,$$ which gives the desired concentration property for $ \beta_n/|\log{\varepsilon_n}|$ in $ L^2({\Omega};{\mathbb{M}^{2\times 2}})$. [ * Step $ 3.$ $ \mu$ belongs to $ H^{-1}({\Omega};{\mathbb{R}}^2)$ and $ { { \rm Curl\,}}\beta=\mu$. * ] { } permit $ { \varphi}\in C^1_0({\Omega})$. It is easy to construct a succession $ \{{\varphi}_n\}\subset H^1_0({\Omega})$ converge to $ { \varphi}$ uniformly and strongly in $ H^1_0({\Omega})$ and meet the property $ $ { \varphi}_n \equiv { \varphi}(x_{i, n }) \qquad \text { in } B_{{\varepsilon_n}}(x_{i, n})\quad \text{for every } x_{i,

Thforem \[KTI\] to $\tilde \beta_n$, obtaining $$\label{ttt1} \int_{{\Mmega}} |\filde \bega_n - \tilde A_n|^2 \, dx \le C \left(\inv_{{\Ometa}} |\tiode \beta_n{^{\rm sym}}|^2 \, dx + \bkg(|\mu_n|({\Omegw})\big)^2 \rigyt) \le X|\log {\varepsilon_n}|^2,$$ wmzre $\tjpde C_n$ is the averagg of the anth-symmetric pard uf $\tilde \beta_n$. Since the average of $\bqta_n$ is a symmetric mattix, wt hwve $$\mabel{ttt2} \int_{{\Omega}_{{\varepsilon_n}}(\mu_n)} |\betz_n|^2 \, dx = \pnt_{{\Omega}_{{\varepsilom_n}}(\mu_n)} |\beta_n - \tilde A_n|^2 - |\tipde W_n|^2 \, dx \le \int_{{\Omega}_{{\vagepsilon_n}}(\mu_b)} |\beeq_n - \tilde A_n|^2 \, dx.$$ Finally, by and we conclude $$\int_{{\Omega}_{{\varepsilon_n}}(\mu_v)} |\betc_n|^2 \, dx \le \inj_{{\Ojehd}_{{\varepsilon_i}}(\mu_n)} |\bvta_n - \tilde A_k|^2 \, dx \ne \int_{{\Okega}} |\tilde \bets_n - \tiode A_n|^2 \, dx \le C|\log {\vacepsilon_n}|^2,$$ which givef the deshrzd compactness properry for $\teta_t/|\log{\xqreosimoi_n}|$ jn $L^2({\Omfga};{\jathbb{M}^{2\timss 2}})$. [*Step $3.$ $\mu$ belongs to $H^{-1}({\Omega};{\mstrvb{R}}^2)$ and ${{\rm Cudl\,}}\beta=\iu$.*]{} Let ${\varphi}\in C^1_0({\Omega})$. It is easy to consuruct a sequence $\{{\varphi}_n\}\subswt H^1_0({\Omega})$ converging jo ${\varphi}$ tniformly and strongly in $H^1_0({\Omega})$ and satisfying tve prkoeruy $${\varpfu}_n \equiv {\varphi}(x_{i,n}) \qquad \text{ in } B_{{\varepsilon_n}}(x_{y,h})\qiac \text{for evern } x_{i,

Theorem \[KTI\] to $\tilde \beta_n$, obtaining $$\label{ttt1} \beta_n \tilde A_n|^2 dx \le C dx \big(|\mu_n|({\Omega})\big)^2 \right) \le {\varepsilon_n}|^2,$$ where $\tilde is the average of the anti-symmetric of $\tilde \beta_n$. Since the average of $\beta_n$ is a symmetric matrix, we $$\label{ttt2} \int_{{\Omega}_{{\varepsilon_n}}(\mu_n)} |\beta_n|^2 \, dx = \int_{{\Omega}_{{\varepsilon_n}}(\mu_n)} |\beta_n - \tilde A_n|^2 - |\tilde \, \le |\beta_n \tilde A_n|^2 \, dx.$$ Finally, by and we conclude $$\int_{{\Omega}_{{\varepsilon_n}}(\mu_n)} |\beta_n|^2 \, dx \le \int_{{\Omega}_{{\varepsilon_n}}(\mu_n)} |\beta_n \tilde A_n|^2 \, dx \le \int_{{\Omega}} |\tilde \beta_n \tilde A_n|^2 \, dx C|\log {\varepsilon_n}|^2,$$ which gives the compactness for $\beta_n/|\log{\varepsilon_n}|$ $L^2({\Omega};{\mathbb{M}^{2\times [*Step $\mu$ belongs to and ${{\rm Curl\,}}\beta=\mu$.*]{} Let ${\varphi}\in C^1_0({\Omega})$. It is easy to construct a sequence $\{{\varphi}_n\}\subset H^1_0({\Omega})$ converging to uniformly and $H^1_0({\Omega})$ and the $${\varphi}_n {\varphi}(x_{i,n}) \qquad \text{ B_{{\varepsilon_n}}(x_{i,n})\quad \text{for every } x_{i,

Theorem \[KTI\] to $\tilde \beta_n$, obtAining $$\labeL{ttt1} \iNt_{{\OMegA}} |\tIlde \Beta_N - \tilde A_n|^2 \, dx \le C \lEFt(\inT_{{\Omega}} |\tilde \beta_n{^{\rm sym}}|^2 \, dX + \big(|\mU_n|({\oMega})\BIg)^2 \Right) \Le C|\log {\vARePSIloN_n}|^2,$$ WhEre $\TiLDe a_n$ is tHe aVerage oF the anti-syMmeTrIc part of $\tildE \BeTa_n$. Since thE avErage of $\beta_n$ Is a SymmetRiC maTRix, we HavE $$\labeL{ttt2} \inT_{{\omega}_{{\vArepsilon_N}}(\mU_N)} |\beta_n|^2 \, DX = \int_{{\OmeGA}_{{\VaRepsIlon_n}}(\mu_n)} |\beta_n - \tildE a_n|^2 - |\TIlde A_n|^2 \, dx \le \int_{{\OMega}_{{\vaRePSiLON_n}}(\mU_n)} |\bEta_n - \tilde A_N|^2 \, dX.$$ FinaLLy, by and WE cONCLudE $$\Int_{{\Omega}_{{\varepSilon_n}}(\mu_n)} |\beTA_n|^2 \, dX \le \int_{{\omEga}_{{\VArepsiLon_n}}(\mU_n)} |\BEta_N - \tilde A_n|^2 \, dx \lE \int_{{\omega}} |\tildE \beta_n - \TIlde A_n|^2 \, dX \Le C|\log {\vArepsiLon_N}|^2,$$ whIch gIVeS tHe dEsIRed COmPacTNesS propertY fOr $\Beta_n/|\Log{\vAREPSiloN_n}|$ iN $L^2({\OmEga};{\maThbb{M}^{2\times 2}})$. [*SteP $3.$ $\mu$ BeloNGs tO $H^{-1}({\OmeGa};{\matHbb{R}}^2)$ AnD ${{\rm CuRl\,}}\beta=\Mu$.*]{} Let ${\VaRphi}\in C^1_0({\Omega})$. It iS easY to constrUct A sEquEnCe $\{{\varPHi}_n\}\subSet h^1_0({\OmEga})$ convErging tO ${\VarPhI}$ UNIfOrmly and strongly in $h^1_0({\OMEGa})$ And satisFying tHE pRoPErty $${\varpHi}_N \eqUiv {\vARPhi}(x_{i,N}) \qquAD \tExt{ in } B_{{\vaRepsilON_n}}(X_{i,N})\quad \teXt{For eveRy } X_{i,

Theorem \[KTI\] to $\tild e \beta_n$ , obt ain ing $ $\la bel{ ttt1} \int_{{\ O mega }} |\tilde \beta_n - \ tilde A _ n|^2 \, dx \ le C \l e ft ( \ int _{ {\ Ome ga } }|\til de\beta_n {^{\rm sym }}| ^2 \, dx + \bi g (| \mu_n|({\O meg a})\big)^2 \ rig ht) \l eC|\ l og {\ var epsil on_n}| ^ 2,$$ w here $\ti ld e A_n$i s the a v e ra ge o f the anti-symmet r ic part of $\tild e \bet a_ n $. S inc e t he average o f $\b e ta_n$ i s a s y mme t ric matrix, w e have $$\l a bel {ttt2} \ int _ {{\Ome ga}_{ {\ v are psilon_n}}( \mu_ n)} |\bet a_n|^2 \, dx = \int_{{ \Omega }_{ {\v arep s il on _n} }( \ mu_ n )} |\ b eta _n - \ti ld eA_n|^ 2 - | \ t ilde A_ n|^2 \, d x \le \int_{{ \Om ega} _ {{\ varep silon _n}} (\ mu_n) } |\be ta_n-\tilde A_n|^2 \ , dx .$$ Final ly, b y a nd we c o nclude $$ \in t_{{\Om ega}_{{ \ var ep s i l on _n}}(\mu_n)} |\bet a_ n | ^2 \, dx \ le \in t _{ {\ O mega}_{{ \v are psil o n _n}}( \mu_ n )} |\beta_ n - \t i ld eA_n|^2\, dx \ le \i nt_ {{\Om e ga}} |\til de \beta _n -\ tilde A_n|^2 \ , dx \le C|\lo g { \ v ar e psil on_ n}|^2,$$ wh ichg ives the de sir e d com pactn es s p r operty for $\beta_n /| \log{\ varep silon_n}|$ in $L^2({\Om e g a };{\math bb{M } ^{ 2 \times 2}})$. [*St ep $3.$ $\ m u$ belon gs to $H^{-1} ({\Omega} ; { \mathbb{ R}} ^2) $ a nd$ { {\ rm Curl\,}}\b e t a=\m u$ .*]{} L et${\varp hi} \in C^ 1_0 ({ \Omega})$ . It isea sy t ocon struc t a seque nc e $ \{ {\v arphi } _n\}\s ubset H^1 _0 ({ \ Ome ga})$ c o nv e r ging t o${\v arp hi }$ un ifor m lyand str ongly in$H^ 1 _0({ \O me ga})$ a nd satisfying t he propert y$${ \varph i } _n \equi v {\varphi}(x_{i,n}) \q q uad \te xt{ in } B_{ {\varepsi lon _n}}(x _{i , n})\qu ad \te xt{fo reve r y } x_ { i ,

Theorem_\[KTI\] to_$\tilde \beta_n$, obtaining $$\label{ttt1} \int_{{\Omega}}_|\tilde \beta_n_-_\tilde A_n|^2_\,_dx \le C \left(\int_{{\Omega}}_|\tilde \beta_n{^{\rm sym}}|^2_\, dx + \big(|\mu_n|({\Omega})\big)^2_\right) \le C|\log {\varepsilon_n}|^2,$$_where_$\tilde A_n$ is the average of the anti-symmetric part of $\tilde \beta_n$. Since the_average_of $\beta_n$_is_a_symmetric matrix, we have $$\label{ttt2} \int_{{\Omega}_{{\varepsilon_n}}(\mu_n)}_|\beta_n|^2 \, dx = \int_{{\Omega}_{{\varepsilon_n}}(\mu_n)} |\beta_n_- \tilde_A_n|^2 - |\tilde A_n|^2 \, dx \le \int_{{\Omega}_{{\varepsilon_n}}(\mu_n)} |\beta_n_-_\tilde A_n|^2 \,_dx.$$ Finally, by and we conclude $$\int_{{\Omega}_{{\varepsilon_n}}(\mu_n)} |\beta_n|^2 \,_dx \le \int_{{\Omega}_{{\varepsilon_n}}(\mu_n)} |\beta_n - \tilde A_n|^2_\, dx _\le_\int_{{\Omega}}_|\tilde \beta_n - \tilde_A_n|^2 \, dx \le C|\log {\varepsilon_n}|^2,$$_which gives the desired compactness property_for $\beta_n/|\log{\varepsilon_n}|$ in $L^2({\Omega};{\mathbb{M}^{2\times 2}})$. [*Step $3.$ $\mu$_belongs to $H^{-1}({\Omega};{\mathbb{R}}^2)$ and ${{\rm Curl\,}}\beta=\mu$.*]{}_Let ${\varphi}\in C^1_0({\Omega})$. It is_easy to_construct a sequence $\{{\varphi}_n\}\subset H^1_0({\Omega})$_converging to ${\varphi}$_uniformly and_strongly in $H^1_0({\Omega})$_and satisfying the property $${\varphi}_n \equiv_{\varphi}(x_{i,n}) \qquad \text{_in } B_{{\varepsilon_n}}(x_{i,n})\quad \text{for every }_x_{i,

}^{+}$ A $0 \subseteq 0 \subseteq 0 \subseteq \mathbb{Z}$ $0 \subseteq 0 \subseteq 0 \subseteq 0$ $0 \subseteq 2\mathbb{Z} \subseteq \mathbb{Z} \subseteq \mathbb{Z}$ $0 \subseteq 0 \subseteq 0 \subseteq 0$ $\hat{\mathcal{U}}_{0,-}^{+}$, $\hat{\mathcal{U}}_{T/2,+}^+$ AIII $0 \subseteq 0 \subseteq 0 \subseteq 0$ $0 \subseteq 0 \subseteq \mathbb{Z} \subseteq \mathbb{Z}^{2}$ $0 \subseteq 0 \subseteq 0 \subseteq 0$ $2\mathbb{Z} \subseteq \mathbb{Z} \subseteq \mathbb{Z} \subseteq \mathbb{Z}^{2}$ Symmetry Class $d_{\parallel}=0$ $d_{\parallel}=1$ $d_{\parallel}=2$ $d_{\parallel}=3$ ------------------------------------------------------------------- ------- ---------------------------------------------------- --------------------------------------------------------------- -------------------------------------------------------------------------- -------------------------------------------------------------------------- $\hat{\mathcal{A}}_{s}^{+}$ A $0 \subseteq 0 \subseteq 0 \subseteq 0$ $0 \subseteq 0 \subseteq 0 \subseteq 0$ $0 \subseteq 0 \subseteq \mathbb{Z}_2 \subseteq \mathbb{Z}_2$ $0 \subseteq 0 \subseteq 0 \subseteq 0$ $\hat{\mathcal{A}}_{0,+}^{+},\hat{\mathcal{A}}_{T/2,-}^{+}$ AIII $0 \subseteq 0 \subseteq 0 \subseteq 0$ $0 \subseteq 0 \subseteq 0 \subseteq \mathbb{Z}$ $0 \subseteq \mathbb{Z}_2 \subseteq \mathbb{Z}_2 \subseteq \mathbb{Z}_2$ $0 \subseteq 0 \subseteq 0 \subseteq 2\mathbb{Z}$ $\overline{\mathcal{A}}_{0}^{+},\overline{\mathcal{A}}_{T/2}^{-}$ A $0 \subseteq 0 \subseteq 0 \subseteq 0$ $0 \subseteq 0 \subseteq \mathbb{Z}_2 \subseteq \mathbb{Z}_2$ $0 \subseteq 0 \subseteq 0 \subseteq 0$ $0 \subseteq 0 \subseteq 0 \subseteq 0$ $\hat{\mathcal{A}}_{

} ^{+}$ A $ 0 \subseteq 0 \subseteq 0 \subseteq \mathbb{Z}$ $ 0 \subseteq 0 \subseteq 0 \subseteq 0 $ $ 0 \subseteq 2\mathbb{Z } \subseteq \mathbb{Z } \subseteq \mathbb{Z}$ $ 0 \subseteq 0 \subseteq 0 \subseteq 0 $ $ \hat{\mathcal{U}}_{0,-}^{+}$, $ \hat{\mathcal{U}}_{T/2,+}^+$ AIII $ 0 \subseteq 0 \subseteq 0 \subseteq 0 $ $ 0 \subseteq 0 \subseteq \mathbb{Z } \subseteq \mathbb{Z}^{2}$ $ 0 \subseteq 0 \subseteq 0 \subseteq 0 $ $ 2\mathbb{Z } \subseteq \mathbb{Z } \subseteq \mathbb{Z } \subseteq \mathbb{Z}^{2}$ Symmetry Class $ d_{\parallel}=0 $ $ d_{\parallel}=1 $ $ d_{\parallel}=2 $ $ d_{\parallel}=3 $ ------------------------------------------------------------------- ------- ---------------------------------------------------- --------------------------------------------------------------- -------------------------------------------------------------------------- -------------------------------------------------------------------------- $ \hat{\mathcal{A}}_{s}^{+}$ A $ 0 \subseteq 0 \subseteq 0 \subseteq 0 $ $ 0 \subseteq 0 \subseteq 0 \subseteq 0 $ $ 0 \subseteq 0 \subseteq \mathbb{Z}_2 \subseteq \mathbb{Z}_2 $ $ 0 \subseteq 0 \subseteq 0 \subseteq 0 $ $ \hat{\mathcal{A}}_{0,+}^{+},\hat{\mathcal{A}}_{T/2,-}^{+}$ AIII $ 0 \subseteq 0 \subseteq 0 \subseteq 0 $ $ 0 \subseteq 0 \subseteq 0 \subseteq \mathbb{Z}$ $ 0 \subseteq \mathbb{Z}_2 \subseteq \mathbb{Z}_2 \subseteq \mathbb{Z}_2 $ $ 0 \subseteq 0 \subseteq 0 \subseteq 2\mathbb{Z}$ $ \overline{\mathcal{A}}_{0}^{+},\overline{\mathcal{A}}_{T/2}^{-}$ A $ 0 \subseteq 0 \subseteq 0 \subseteq 0 $ $ 0 \subseteq 0 \subseteq \mathbb{Z}_2 \subseteq \mathbb{Z}_2 $ $ 0 \subseteq 0 \subseteq 0 \subseteq 0 $ $ 0 \subseteq 0 \subseteq 0 \subseteq 0 $ $ \hat{\mathcal{A } } _ {

}^{+}$ A $0 \subseteq 0 \subseteq 0 \vubsetsq \mathbc{Z}$ $0 \subseteq 0 \sybsettz 0 \subseteq 0$ $0 \subsetew 2\mauhbb{Z} \subseteq \mavgbb{Z} \sunfetes \matkbu{Z}$ $0 \subseteq 0 \sgbreceq 0 \subseteq 0$ $\hat{\mathcal{U}}_{0,-}^{+}$, $\hat{\matrcal{U}}_{T/2,+}^+$ AIII $0 \sunfetes 0 \subseteq 0 \subseteq 0$ $0 \shbseteq 0 \subseteq \mayhbb{Z} \subseteq \mathbb{Z}^{2}$ $0 \subsetee 0 \subseteq 0 \sufweteq 0$ $2\mathbb{Z} \sobseteq \mathbb{Z} \subseteq \mathbb{Z} \subszteq \mathbb{E}^{2}$ Snkmetry Class $d_{\parallel}=0$ $d_{\pxealuel}=1$ $d_{\karwolel}=2$ $d_{\larallel}=3$ ------------------------------------------------------------------- ------- ---------------------------------------------------- --------------------------------------------------------------- -------------------------------------------------------------------------- -------------------------------------------------------------------------- $\hat{\mathfal{A}}_{s}^{+}$ A $0 \subseteq 0 \subsetex 0 \suusdtew 0$ $0 \subseteq 0 \subseteq 0 \subseteq 0$ $0 \subseteq 0 \skbxgteq \mathbb{Z}_2 \sobseteq \mafhbb{Z}_2$ $0 \subseteq 0 \sunseteq 0 \subseteq 0$ $\hat{\mqthcal{A}}_{0,+}^{+},\hat{\manhcao{A}}_{T/2,-}^{+}$ AIII $0 \subsetzq 0 \sunsetea 0 \subseteq 0$ $0 \subseteq 0 \dubseteq 0 \rubseteq \mathbb{Z}$ $0 \subseteq \mathbb{Z}_2 \subfeteq \matibb{Z}_2 \xubseted \majhbb{Z}_2$ $0 \sunveteq 0 \subseteq 0 \skbsetgq 2\matvbb{Z}$ $\ovegline{\mathcal{A}}_{0}^{+},\overline{\mathcal{A}}_{T/2}^{-}$ A $0 \vubveteq 0 \sbbsetea 0 \subseteq 0$ $0 \subsejeq 0 \subszteq \mxthbb{Z}_2 \subaeteq \methbb{Z}_2$ $0 \dubseteq 0 \suuseteq 0 \stbsereq 0$ $0 \subswteq 0 \subseteq 0 \sunseted 0$ $\hat{\mathccu{Q}}_{

}^{+}$ A $0 \subseteq 0 \subseteq 0 $0 0 \subseteq \subseteq 0$ $0 \mathbb{Z}$ \subseteq 0 \subseteq \subseteq 0$ $\hat{\mathcal{U}}_{0,-}^{+}$, AIII $0 \subseteq 0 \subseteq 0 0$ $0 \subseteq 0 \subseteq \mathbb{Z} \subseteq \mathbb{Z}^{2}$ $0 \subseteq 0 \subseteq 0 0$ $2\mathbb{Z} \subseteq \mathbb{Z} \subseteq \mathbb{Z} \subseteq \mathbb{Z}^{2}$ Symmetry Class $d_{\parallel}=0$ $d_{\parallel}=1$ $d_{\parallel}=2$ ------------------------------------------------------------------- ---------------------------------------------------- -------------------------------------------------------------------------- $\hat{\mathcal{A}}_{s}^{+}$ A $0 \subseteq 0 \subseteq 0 \subseteq 0$ $0 \subseteq 0 \subseteq 0 \subseteq 0$ \subseteq 0 \subseteq \mathbb{Z}_2 \subseteq \mathbb{Z}_2$ $0 \subseteq \subseteq 0 \subseteq 0$ AIII $0 \subseteq 0 \subseteq \subseteq $0 \subseteq \subseteq \subseteq $0 \subseteq \mathbb{Z}_2 \mathbb{Z}_2 \subseteq \mathbb{Z}_2$ $0 \subseteq 0 \subseteq 0 \subseteq 2\mathbb{Z}$ $\overline{\mathcal{A}}_{0}^{+},\overline{\mathcal{A}}_{T/2}^{-}$ A $0 \subseteq 0 \subseteq 0 0$ $0 \subseteq \mathbb{Z}_2 \mathbb{Z}_2$ \subseteq \subseteq 0 \subseteq \subseteq 0 \subseteq 0 \subseteq 0$

}^{+}$ A $0 \subseteq 0 \subseteq 0 \subseteq \Mathbb{Z}$ $0 \subSeteq 0 \SubSetEq 0 \SubsEteq 0$ $0 \Subseteq 2\mathbb{z} \SubsEteq \mathbb{Z} \subseteq \matHbb{Z}$ $0 \sUbSEteq 0 \SUbSeteq 0 \SubseteQ 0$ $\HaT{\MAthCaL{U}}_{0,-}^{+}$, $\Hat{\MaTHcAl{U}}_{T/2,+}^+$ AiII $0 \SubseteQ 0 \subseteq 0 \sUbsEtEq 0$ $0 \subseteq 0 \suBSeTeq \mathbb{Z} \SubSeteq \mathbb{Z}^{2}$ $0 \SubSeteq 0 \sUbSetEQ 0 \subsEteQ 0$ $2\mathBb{Z} \subSEteq \maThbb{Z} \subsEtEQ \mathbB{z} \subsetEQ \MaThbb{z}^{2}$ Symmetry Class $d_{\paRAlLEl}=0$ $d_{\parallel}=1$ $d_{\paRallel}=2$ $D_{\pARaLLEl}=3$ ------------------------------------------------------------------- ------- ---------------------------------------------------- --------------------------------------------------------------- -------------------------------------------------------------------------- -------------------------------------------------------------------------- $\hAt{\mAthcal{A}}_{s}^{+}$ A $0 \sUbSeteq 0 \SUbseteq 0 \SUbSETEq 0$ $0 \sUBseteq 0 \subseteQ 0 \subseteq 0$ $0 \suBSetEq 0 \subsEtEq \mAThbb{Z}_2 \sUbsetEq \MAthBb{Z}_2$ $0 \subseteq 0 \SubsEteq 0 \subseTeq 0$ $\hat{\MAthcal{A}}_{0,+}^{+},\HAt{\mathcAl{A}}_{T/2,-}^{+}$ AIiI $0 \sUbsEteq 0 \SUbSeTeq 0 \SuBSetEQ 0$ $0 \sUbsETeq 0 \Subseteq 0 \SuBsEteq \mAthbB{z}$ $0 \SUBsetEq \mAthbB{Z}_2 \subSeteq \mathbb{Z}_2 \sUbsEteq \MAthBb{Z}_2$ $0 \suBseteQ 0 \subSeTeq 0 \suBseteq 2\MathbB{Z}$ $\Overline{\mathcal{a}}_{0}^{+},\oveRline{\mathCal{a}}_{T/2}^{-}$ a $0 \suBsEteq 0 \sUBseteq 0 \SubSetEq 0$ $0 \subseTeq 0 \subsETeq \MaTHBB{Z}_2 \Subseteq \mathbb{Z}_2$ $0 \subSeTEQ 0 \sUbseteq 0 \sUbseteQ 0$ $0 \SuBsETeq 0 \subseTeQ 0 \suBsetEQ 0$ $\Hat{\maThcaL{a}}_{

}^{+}$ A $0 \sub set eq0\sub sete q 0 \subseteq\ math bb{Z}$ $0 \s ub s eteq 0\subs eteq 0\ su b s ete q0$ $ 0 \subs eteq 2\mat hbb {Z } \subseteq\ ma thbb{Z} \s ubs eteq \mathbb {Z} $ $0 \s u bseteq 0 \subse te q 0 \su b seteq 0 $ $\ hat{\mathcal{U}}_ { 0, - }^{+}$, $\hat{ \mathc al { U} } _ {T/ 2,+ }^+$ AI II $0\ su b s e teq 0 \subseteq 0 \subseteq0 $ $0\ subset eq 0\s u bse teq \mathbb {Z}\subseteq \math b b{Z}^{2 } $ $0 \ s ub se teq 0 \su b se teq 0 \ subseteq 0 $ $2 \mat hbb{Z } \subseteq \ mat hbb{ Z } \ subse teq \ math bb {Z} \ subset eq \m at hbb{Z}^{2}$ Symmet ry Class $d_{\ parall e l} =0 $ $d_ { \p ar allel}= 1$ $d_{\p a rallel}=2$ $ d_ { \p a rallel}=3$ ------ -- ------ ----- ------------- ---------- - - - -------- ---- - -- - ------ ------- ---- ---------- - -------- ----- -------- --------- - - ----- -- --- --- --- --- - - -- ------------- - - ---- -- ------- --- ------- --- --- - - --- -- --------- -------- -- -- -- -- --- ----- - -------- -- --- -- --- ----- - ------ -------- -- -- - --- ------- - -- - - ---- -- -- ---- --- -- ----- ---- - --- ------- --------- --- - $\hat {\ mathcal{A} }_ {s} ^{+}$ A $0 \sub set eq 0\sub seteq 0 \ sub seteq0$ $ 0 \ s u bsete q 0\su bs eteq 0 \su b s ete q 0$ $0 \subseteq0 \s ubseteq \math bb{ Z}_2 \ su bse t eq \ma th b b{Z } _ 2$ $ 0\subseteq0 \s ub seteq 0 \subse teq 0 $ $\hat{\ma thcal{A}} _{ 0,+} ^ { +}, \hat{\math cal{A}}_ {T/2,-}^{ + }$ AIII $ 0\su bsete q 0 \s u bse teq 0 \subs et eq 0$ $ 0 \subseteq 0 \subseteq 0 \su bsete q \ mathbb{Z} $ $0 \s ubse teq \mathb b{Z }_2 \sub set e q \ma thbb { Z} _2\ subse teq\ mathbb{Z} _ 2$ $0 \ s u b set eq 0\su b seteq0 \s ubseteq 2\mathbb{ Z }$ $\overli ne{\ m a thc al{ A }}_{ 0} ^{+},\overline {\m at h c al{A}}_{ T/ 2}^{-}$ A $0 \ subset eq 0 \ subsete q 0\ subset eq 0 $ $0 \s ub s eteq 0\s ub s eteq \ math bb {Z}_2\subse t eq \ m a thbb{Z}_2$ $0\subs et eq 0 \s u bset eq 0 \subs eteq 0$ $ 0\subseteq0 \subsete q 0 \ subsete q0$ $\ha t{\m a t hcal{ A}}_ {

}^{+}$ _ _ _ __ __ _ _A _ _$0_\subseteq 0 \subseteq 0 \subseteq \mathbb{Z}$ __ ___ $0 \subseteq 0_\subseteq 0 \subseteq 0$ _ _ __ $0_\subseteq 2\mathbb{Z} \subseteq \mathbb{Z} \subseteq \mathbb{Z}$ _ _ ___ _ _$0 \subseteq 0 \subseteq 0 \subseteq_0$ $\hat{\mathcal{U}}_{0,-}^{+}$, $\hat{\mathcal{U}}_{T/2,+}^+$ _ AIII _ _ $0_\subseteq 0 \subseteq 0 \subseteq_0$ _ _ _ $0 \subseteq 0 \subseteq_\mathbb{Z} \subseteq \mathbb{Z}^{2}$_ __ ___ _ _$0_\subseteq 0_\subseteq_0 \subseteq 0$ __ _ __ $2\mathbb{Z} \subseteq_\mathbb{Z} \subseteq \mathbb{Z} \subseteq \mathbb{Z}^{2}$ _ _ _ _ _ _ Symmetry_ __ __ _ _ _ _ __ Class _ _ ___ $d_{\parallel}=0$ _ _ _ __ __ ___ _ $d_{\parallel}=1$ _ _ _ ___ _ _ _ _ $d_{\parallel}=2$ _ _ _ _ __ ____ _ _ _ _ _$d_{\parallel}=3$ _------------------------------------------------------------------- ------- ----------------------------------------------------_---------------------------------------------------------------_-------------------------------------------------------------------------- -------------------------------------------------------------------------- _ $\hat{\mathcal{A}}_{s}^{+}$_ __ _ _ _ A __ _$0 \subseteq 0 \subseteq_0 \subseteq_0$ _ _ _ _ _ __ $0 \subseteq 0 \subseteq 0 \subseteq 0$ ____ _ _ _ __ $0 \subseteq 0 \subseteq \mathbb{Z}_2_\subseteq \mathbb{Z}_2$ _ _ _ $0 \subseteq 0 \subseteq 0 \subseteq 0$ _ _ $\hat{\mathcal{A}}_{0,+}^{+},\hat{\mathcal{A}}_{T/2,-}^{+}$ _ __ AIII _ _ $0 \subseteq 0 \subseteq 0 \subseteq_0$_ ___$0 \subseteq 0_\subseteq_0 \subseteq \mathbb{Z}$ _____ $0 \subseteq \mathbb{Z}_2 \subseteq_\mathbb{Z}_2 \subseteq \mathbb{Z}_2$ __ _ __$0 \subseteq 0 \subseteq_0 \subseteq 2\mathbb{Z}$ _ $\overline{\mathcal{A}}_{0}^{+},\overline{\mathcal{A}}_{T/2}^{-}$ A __ ___$0_\subseteq 0 \subseteq 0_\subseteq 0$ $0 \subseteq 0 \subseteq \mathbb{Z}_2_\subseteq_\mathbb{Z}_2$ _ __ _ $0 \subseteq 0 \subseteq 0_\subseteq 0$ _ _ _ _ $0 \subseteq 0_\subseteq 0 \subseteq_0$ _ _ $\hat{\mathcal{A}}_{

orsanyi_2010; @Bazavov_2012]. This induces modifications to the spectral distributions of chiral partners in the light hadron sector such as the $\rho$ and $a_1$ mesons – ultimately leading to their degeneracy. A substantial melting of the chiral condensate is conjectured to be relevant already for HICs at low collision energies [@Klimt_1990; @Schaefer_2007]. Electromagnetic (EM) radiation represents an excellent probe for such investigations as it is emitted during the whole time evolution of a HIC and decouples from the strongly interacting collision zone once created. In contrast to real photons which are massless, virtual photons decaying into a pair of leptons ($e^+e^-$ or $\mu^+\mu^-$) have an additional observable in terms of their invariant-mass. Thus, one can access the information about the properties of matter produced inside the hot and dense fireball which is irretrievable from the spectra of final-state hadrons due to rescattering. In particular, the excess yield of low-mass dileptons above the contribution from the hadronic freeze-out cocktail was identified to be sensitive to the fireball lifetime, while the slope in the intermediate-mass region of the dilepton invariant-mass spectrum can serve as a thermometer which is not distorted by the collective expansion of the medium [@Rapp_vanHees_2016; @Specht_2010]. Coarse-grained transport approach for thermal dilepton rates ============================================================ For a proper theoretical description of the contribution of in-medium signals to the dilepton invariant-mass spectrum realistic thermal dilepton emission rates need to be convoluted with an accurate modeling of the chemical potentials and temperatures reached during the space-time evolution of the fireball. For HICs at ultrarelativistic energies (URHICs) a hydrodynamic description can be used to evaluate the thermodynamic properties of the medium and the thermal dilepton rates can directly be employed. The applicability of such an approach for the system evolution down to the SIS18 energy range of a few GeV is unclear. One issue is the justification of thermalization due to the long time it takes until a full overlap of the two incoming nuclei is reached. For this reason hadronic transport models like UrQMD are commonly used to describe the system evolution of HICs at SIS energies. The dilepton radiation is then obtained perturbatively by integrating for each created resonance the probability of a decay into a dilepton

orsanyi_2010; @Bazavov_2012 ]. This induces modifications to the spectral distributions of chiral partner in the faint hadron sector such as the $ \rho$ and $ a_1 $ mesons – ultimately leading to their degeneracy. A hearty melting of the chiral condensate is speculate to be relevant already for HICs at low collision energies [ @Klimt_1990; @Schaefer_2007 ]. Electromagnetic (EM) radiation represent an excellent probe for such investigations as it is emitted during the solid time evolution of a HIC and decouples from the powerfully interacting collision zone once created. In contrast to veridical photons which are massless, virtual photons disintegrate into a pair of leptons ($ e^+e^-$ or $ \mu^+\mu^-$) have an extra observable in terms of their changeless - mass. therefore, one can access the information about the properties of matter produced inside the hot and dense fireball which is irretrievable from the spectra of concluding - state hadrons due to rescattering. In particular, the excess yield of low - mass dileptons above the contribution from the hadronic freeze - out cocktail was identify to be sensible to the fireball lifetime, while the slope in the intermediate - mass region of the dilepton invariant - bulk spectrum can service as a thermometer which is not distorted by the collective expansion of the metier   [ @Rapp_vanHees_2016; @Specht_2010 ]. Coarse - granulate transport approach for thermal dilepton rates = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = For a proper theoretical description of the contribution of in - culture medium signals to the dilepton invariant - mass spectrum realistic thermal dilepton discharge rates need to be convoluted with an accurate modeling of the chemical potentials and temperatures reach during the space - time evolution of the fireball. For HICs at ultrarelativistic energies (URHICs) a hydrodynamic description can be used to measure the thermodynamic property of the medium and the thermal dilepton rates can directly be employed. The applicability of such an approach for the system evolution down to the SIS18 department of energy range of a few GeV is indecipherable. One issue is the justification of thermalization due to the long prison term it takes until a broad overlap of the two incoming nuclei is strive. For this reason hadronic conveyance models like UrQMD are normally used to describe the arrangement evolution of HICs at SIS energies. The dilepton radiotherapy is then obtained perturbatively by integrate for each created resonance the probability of a decay into a dilepton

orswnyi_2010; @Bazavov_2012]. This inducer modifications to the spectdal distfibutions of chiral partners ib the light hadron sector sjch as thv $\rho$ and $a_1$ mtsons – ultimately leading to thslr deyeieracy. A substaktial meltitg of the chirdl clndensate is conjectured to be reledant altewdy for HICs aj low solljsion energies [@Klimt_1990; @Schaefer_2007]. Electdomagneuic (EM) radiation rrpresents an excellent prohe flr such investigatlons as it us eiutted during the whole time evolutjon of a HIC and decouples from the xtrongly ibtwrafjing collisiin zoge once creabvd. In cmntrast to real photoks whmch qre massless, virtual 'hotons decaying intj a pair mf leptons ($e^+e^-$ or $\mu^+\nu^-$) have an dddiguonxl kbxedvable in terms of fheir invaruant-mass. Thus, one csn qccess the inrormatyog about the properties of matter productd inaide the hot and dense direball which is irrgtrievable from the spectra of final-state hadrons due to revcattxrkng. Nk oqrhicular, the excess yield of low-mass dileptons zbpvv the contributiok from the hadroniv vrrgze-out cocktaiu was nsehtified to be sensltive tj the fireball lifrtime, while the slope in thw intermediane-maws region of the dnlepton invaxiant-msss slectrum can serve as a chermojeter which is not djrtorted by the cullvctiee expansion of the medium [@Wapp_vanHexs_2016; @Spzcht_2010]. Coarre-grsined eransport wpproach for thermal dilfpton rdtes ============================================================ For a oroper theoretical description of the contribotimn mf in-mednum sinnals to the dijepton invariaut-mass s'ectruo realistib thermal dilepton eiission rates jeed to be cmnvoluteq wirh ab accurxge modeling of the chempccl potentuals and temperatuves rgadhed during the spqce-time evolutipn uf ehv fmrebajn. For HICs at ultfardkativkstic energles (URHOCs) a hydrodynamic dascrjption can be used tj evaluaje the thqrmodynamic ptoperties of the mtdium end thx therkal dilepton rates can directly bs employef. Tme applicabilyty if such an a'proach for the system evolution down to the SIS18 energy rabge of a few GeV is mnclear. One mssue ys the juvtification of thermqlization due to uhe long time it takes hntil d fulp overlap of the two incoming nuclei is reached. For this reason hadronic rransport models ljke IrQMD axe eommonlr usxd to describe the system evolution of HICs at SIR energias. The dilepton radiation is tnev obtained petturbatively by integratinf for eavh created resonance the probabolity of a decay into a cilepton

orsanyi_2010; @Bazavov_2012]. This induces modifications to the of partners in light hadron sector $a_1$ – ultimately leading their degeneracy. A melting of the chiral condensate is to be relevant already for HICs at low collision energies [@Klimt_1990; @Schaefer_2007]. Electromagnetic radiation represents an excellent probe for such investigations as it is emitted during whole evolution a and decouples from the strongly interacting collision zone once created. In contrast to real photons which massless, virtual photons decaying into a pair of ($e^+e^-$ or $\mu^+\mu^-$) have additional observable in terms of invariant-mass. one can the about properties of matter inside the hot and dense fireball which is irretrievable from the spectra of final-state hadrons due to In particular, yield of dileptons the from the hadronic was identified to be sensitive to while the slope in the intermediate-mass region of dilepton invariant-mass can serve as a thermometer which not distorted by the collective expansion of the [@Rapp_vanHees_2016; @Specht_2010]. Coarse-grained transport approach for thermal dilepton rates ============================================================ For a proper theoretical description contribution of in-medium signals the dilepton invariant-mass realistic dilepton rates to be with an accurate modeling of the chemical potentials and temperatures reached the space-time evolution of the fireball. For HICs at ultrarelativistic a description can be to evaluate the thermodynamic of medium and the thermal can be of an for the system evolution to the SIS18 energy range a few GeV is justification of thermalization due to the long time takes until a full overlap of the incoming nuclei is reached. For this reason hadronic transport models like UrQMD commonly used the system evolution of HICs at SIS energies. dilepton radiation is then perturbatively by integrating for each created resonance the probability a into a

orsanyi_2010; @Bazavov_2012]. This induces ModificatiOns to The SpeCtRal dIstrIbutions of chirAL parTners in the light hadron sEctor SuCH as tHE $\rHo$ and $A_1$ mesons – ULtIMAteLy LeAdiNg TO tHeir dEgeNeracy. A SubstantiaL meLtIng of the chirAL cOndensate iS coNjectured to bE reLevant AlReaDY for HiCs At low CollisIOn enerGies [@Klimt_1990; @scHAefer_2007]. ELEctromaGNEtIc (EM) Radiation represenTS aN Excellent probe For sucH iNVeSTIgaTioNs as it is emItTed duRIng the wHOlE TIMe eVOlution of a HIC And decoupleS FroM the stRoNglY InteraCting CoLLisIon zone once CreaTed. In contRast to REal photONs which Are masSleSs, vIrtuAL pHoTonS dECayINg IntO A paIr of leptOnS ($e^+E^-$ or $\mu^+\Mu^-$) haVE AN AddiTioNal oBservAble in terms of TheIr inVAriAnt-maSs. ThuS, one CaN acceSs the iNformAtIon about the propErtiEs of matteR prOdUceD iNside THe hot aNd dEnsE firebaLl which IS irReTRIEvAble from the spectra Of FINaL-state haDrons dUE tO rEScatteriNg. in pArtiCULar, thE excESs Yield of lOw-mass DIlEpTons aboVe The conTrIbuTioN from THe haDronic Freeze-ouT cockTAil was identifiED to be sensitivE To THE fIRebaLl lIfetime, whilE the SLope In thE InTerMEdiatE-mass ReGIoN Of the dilepton invariAnT-mass sPectrUm can serve as a ThermometeR WHIch is not DistORtED by the collectiVe expAnsion of thE Medium [@RaPp_vanhees_2016; @SpecHt_2010]. Coarse-gRAIned tranSpoRt aPprOacH FOr Thermal dileptON RateS ============================================================ FOr a propEr tHeoretiCal DesCriPtiOn Of the contRibution Of In-MeDiUm sIgnalS To the dilEpTon InVarIant-mASs specTrum rEaliStIc THerMal dilePToN EMissIoN rAtes NeeD tO be coNvolUTed With an aCcurate moDelINg of ThE cHemical Potentials and TeMperatures ReAchEd duriNG The space-Time evolution of the firebALl. For HIcs aT ultrArelAtivistic EneRgies (UrHIcS) a hydrOdynamIc desCrIptION can bE USeD to EvAluate the tHERmoDynamIc PropErties oF the medium and the thERmaL dilepton rateS caN dirECTlY be EMpLOyeD. THE apPLIcability of such An approach FoR ThE system evoLUtiOn Down to tHe SIS18 enErgy rANge of a fEw GeV is unClear. One iSsUe is THE juStificatioN of thermAlization DUe to tHE lOng tiMe iT takes UnTil A full OverlaP Of tHe two IncomiNg Nuclei Is reaChEd. For thiS reason hadronic transporT modelS like urQmD are commOnlY UseD to descriBe thE system evoLutIon Of HICS at siS eneRgieS. thE diLEpton RadiATion is theN ObTaiNED pErturbativeLY BY inTegraTinG For eacH creAted resonance the pRObability of a deCay iNTO a dIlePTon

orsanyi_2010; @Bazavov_201 2]. This i nduce s m odi fi cati onsto the spectra l dis tributions of chiral p artne rs in t h elight hadron se c t orsu ch as t h e$\rho $ a nd $a_1 $ mesons – ul ti mately leadi n gto their d ege neracy. A su bst antial m elt i ng of th e chi ral co n densat e is conj ec t ured t o be rel e v an t al ready for HICs at lo w collision ene rgies[@ K li m t _19 90; @Schaefer _2 007]. Electr o ma g n e tic (EM) radiatio n represent s an excel le ntp robe f or su ch inv estigations asit is emi tted d u ring th e wholetime e vol uti on o f a H ICan d de c ou ple s fr om the s tr on gly i nter a c t i ng c oll isio n zon e once create d.In c o ntr ast t o rea l ph ot ons w hich a re ma ss less, virtual p hoto ns decayi ngin toapairo f lept ons ($ e^+e^-$ or $\m u ^+\ mu ^ - $ )have an additional o b s er vable in terms of t h eir inva ri ant -mas s . Thus , on e c an acces s thei nf or mationab out th epro per tieso f ma tter p roducedinsid e the hot and d e nse fireballw hi c h i s irr etr ievable fro m th e spe ctra of fi n al-st ate h ad r on s due to rescatterin g. In p artic ular, the exc ess yieldo f low-mass dil e pt o ns above the c ontri bution fro m the had ronic freeze- out cockt a i l was id ent ifi edtob e s ensitive to t h e fir eb all lif eti me, whi lethe sl ope i n the int ermediat e- ma ss r egi on of the dile pt onin var iant- m ass sp ectru m ca nse r veas a th e rm o m eter w hi ch i s n ot dist orte d by the co llectiveexp a nsio nof the me dium [@Rapp_v an Hees_2016; @ Spe cht_20 1 0 ]. Coar se-grained transport ap p roach f ortherm al d ilepton r ate s ==== === = ====== ====== ===== == === = = ===== = = == === == ========== = = == Foraprop er theo retical descriptio n of the contribu tio n of i n- med i um sig na l s t o the dilepton in variant-ma ss sp ectrum rea l ist ic therma l dilep ton e m issionrates nee d to be c on volu t e d w ith an acc urate mo deling of the c h em icalpot ential sand temp eratur e s r eache d duri ng the s pace- ti me evolu tion of the fireball. For HI Cs at ul trarelati vis t icenergies(URH ICs) a hyd rod yna mic d esc r iptio n ca n b e u s ed to eva l uate thet he rmo d y na mic propert i e s of themed i um and the thermal dilepton rates can dire ctly b e e mpl o yed. T he applicabili tyof s uch an a pp roach for t he syste me volut ion do wn tothe SIS 1 8 e n ergy r ange of a few Ge V i su nclear. O ne issueis t he justi ficati o n of t hermalization du e tot h e lon g ti me it t akes un t il a full over lap of thetwo in comi ng nu clei is r eached . F or this reas o n hadroni c tra nsportmo dels li ke UrQ MD a r e comm only u sed to descr i b et he s y ste m ev oluti on ofHICs at S I S energi es. The dil ep ton r adiati o ni s then obta ine d per t u rbativelyb y in t eg r ating for e ach cr eated r e son an ce thepro b a bility of a decayi nto a dil epton

orsanyi_2010; @Bazavov_2012]._This induces_modifications to the spectral_distributions of_chiral_partners in_the_light hadron sector_such as the_$\rho$ and $a_1$ mesons_– ultimately leading_to_their degeneracy. A substantial melting of the chiral condensate is conjectured to be relevant_already_for HICs_at_low_collision energies [@Klimt_1990; @Schaefer_2007]. Electromagnetic (EM)_radiation represents an excellent probe_for such_investigations as it is emitted during the whole_time_evolution of a_HIC and decouples from the strongly interacting collision zone_once created. In contrast to real_photons which are_massless,_virtual_photons decaying into a_pair of leptons ($e^+e^-$ or $\mu^+\mu^-$)_have an additional observable in terms_of their invariant-mass. Thus, one can access_the information about the properties of_matter produced inside the hot_and dense_fireball which is irretrievable from_the spectra of_final-state hadrons_due to rescattering. In_particular, the excess yield of low-mass_dileptons above the_contribution from the hadronic freeze-out cocktail_was_identified to be_sensitive_to_the fireball_lifetime, while the_slope_in the_intermediate-mass_region of the dilepton invariant-mass spectrum_can_serve as a thermometer which is not_distorted by the collective_expansion_of the medium [@Rapp_vanHees_2016; @Specht_2010]. Coarse-grained_transport approach for thermal dilepton_rates ============================================================ For a proper theoretical description of_the contribution_of in-medium_signals to the dilepton invariant-mass spectrum realistic thermal dilepton emission rates_need to be convoluted with an_accurate modeling of the_chemical potentials_and_temperatures reached during_the_space-time evolution_of the fireball. For HICs at ultrarelativistic energies_(URHICs) a_hydrodynamic description can be used to_evaluate the thermodynamic properties_of_the medium and the thermal dilepton_rates can directly be employed. The_applicability of such an approach_for_the_system evolution down to the_SIS18 energy range of a few_GeV is unclear._One issue is the justification of thermalization_due_to the long time it takes_until_a full overlap of the two_incoming_nuclei_is reached. For this reason_hadronic transport models like UrQMD are_commonly used to describe the system evolution of HICs_at SIS energies._The dilepton radiation is then_obtained_perturbatively_by integrating for each created resonance the probability of a_decay into_a dilepton

morphismes de Calabi et graphe de Reeb sur le tore.* C. R. Acad. Sci. Paris, Ser. I vol.343 323-328 (2006). G. Reeb, *Sur les points singuliers d’une forme de Pfaff complètement intégrable ou d’une fonction numérique.* C.R. Acad. Sci. Paris, 222 :847-849, 1946. --- abstract: 'This paper presents [*Fiducia*]{}, a food review system involving a pipeline which processes restaurant-related reviews obtained from Zomato (India’s largest restaurant search and discovery service). Fiducia is specific to popular café food items and manages to identify relevant information pertaining to each item separately in the reviews. It uses a sentiment check on these pieces of text and accordingly suggests an appropriate restaurant for the particular item depending on user-item and item-item similarity. Experimental results show that the sentiment analyzer module of Fiducia achieves an accuracy of over 85% and our final recommender system achieves an RMSE of about $1.01$ beating other baselines.' author: - 'Mansi Goel, Ayush Agarwal, Deepak Thukral, Tanmoy Chakraborty' - Ayush Agarwal - Deepak Thukral - Tanmoy Chakraborty title: 'Fiducia: A Personalized Food Recommender System for Zomato' --- 0 Introduction ============ Zomato[^1] is the largest Indian restaurant directory application used to suggest nearby restaurants according to the favored cuisines, prices and other amenities like home delivery. It hosts vast information about each restaurant which is not clustered properly according to the users’ convenience. If we manage to get a collection of restaurants recognized according to user preferences, the usability of the application will greatly increase. Zomato fails to provide details at the individual item level. A restaurant can be very popular but it might not be famous for a certain food item; whereas a less popular restaurant might excel in that. Usually customers need to go through the reviews to find out what’s the best the restaurant serves and what should be avoided.\ [**Problem Definition:**]{} In this paper, we aim to design a system that maps customers’ choice of food to the place where they can be availed the best. The input of our system will be in the form of

morphismes de Calabi et graphe de Reeb sur le tore. * C. R. Acad. Sci. Paris, Ser. I vol.343 323 - 328 (2006). G. Reeb, * Sur les points singuliers d’une forme de Pfaff complètement intégrable ou d’une fonction numérique. * C.R. Acad. Sci. Paris, 222: 847 - 849, 1946. --- abstract:' This paper presents [ * Fiducia * ] { }, a food revue arrangement involving a pipeline which processes restaurant - refer reviews obtained from Zomato (India ’s big restaurant search and discovery serve). Fiducia is specific to popular café food item and manages to identify relevant data pertaining to each item separately in the reviews. It use a sentiment confirmation on these pieces of text and consequently suggests an appropriate restaurant for the particular item count on user - item and detail - item similarity. Experimental consequence show that the sentiment analyzer module of Fiducia achieves an accuracy of over 85% and our final recommender arrangement achieves an RMSE of about $ 1.01 $ beating other baselines.' author: -' Mansi Goel, Ayush Agarwal, Deepak Thukral, Tanmoy Chakraborty' - Ayush Agarwal - Deepak Thukral - Tanmoy Chakraborty title:' Fiducia: A Personalized Food Recommender System for Zomato' --- 0 Introduction = = = = = = = = = = = = Zomato[^1 ] is the largest Indian restaurant directory application used to suggest nearby restaurants according to the favored cuisines, prices and other comforts like home delivery. It host huge information about each restaurant which is not clustered properly according to the drug user ’ convenience. If we manage to get a collection of restaurant recognized according to user preferences, the usability of the application will greatly increase. Zomato fails to provide details at the individual item level. A restaurant can be very popular but it might not be celebrated for a certain food detail; whereas a less popular restaurant might excel in that. Usually customers need to go through the recapitulation to find out what ’s the best the restaurant serves and what should be avoided.\ [ * * Problem Definition :* * ] { } In this newspaper, we aim to design a system that maps customer ’ choice of food to the place where they can be availed the best. The input of our arrangement will be in the form of

morohismes de Calabi et grakhe de Reeb sur lg rore.* C. R. Acas. Sci. Pafis, Ser. I vol.343 323-328 (2006). G. Reeb, *Sur lxs piints singuliers d’une forme de Pfaff complètenent untégrable ou d’une njnctjln nbmécique.* C.R. Acad. Sgi. Paris, 222 :847-849, 1946. --- abstract: 'This pxpzr presents [*Fiducia*]{}, a food review syftem infopving a pipelige wnych lgogesses restaurant-related reviews obtaintd from Zomato (Indoa’s largest restaurant seagch wnd discovery servlce). Fiducia is fpecific to pupular café food items znd manages to identify relevang infprmation pgxraijhng to each item separately lm the seviews. It uses a senbimenv chwck on these pieces oh text and accordingjy suggesds an appropriate rwsrauratt fmr tfw pxrtjcnlad item de'ending on hser-item ane item-item similariuy. Qqlerimental rssults srow that the sentiment analyzer module mf Riducia achieves an accyracy of over 85% and out final resommender system achieves an RMSE of about $1.01$ beatitg otief bcwelindw.' akthor: - 'Mansi Goel, Ayush Agarwal, Deepak Thukral, Fammpy Chakraborty' - Ayush Agareap - Cgepak Thukral - Janmoy Egairaborty title: 'Fidufia: A Pgrsonaoized Fooq Revommender System for Zomato' --- 0 Untroduction ============ Domaro[^1] is the largest Nndian restabrant cirecyory application used tu sufgest nearbj restaurzvts according to thv faeored cuisines, prices and jther ameiitiex like fome delivqry. It hoshs vast information abouh eack resdaurant whlch is not clustered properly arrording to thg uverv’ convennence. Lf we manage to get a collectnon of rzstaurxnts recoghized arcording to tser preferenwgs, the usabilmty of thq applicqtion wkul greatly inctease. Zomcuo fails ti provide details et tfs individual itzo level. A restautang cwn bx verr popular but it oigfy not be famous nor a crrtain food item; wheseas a less popular rextwurant mught excql in that. Usially customers netd to jo thruugh yhe reviews to find out what’s the best the reftaurant serdes qnd what shobld be avoided.\ [**Problem Definition:**]{} In this paper, we aim to dwsign a system that kaps customxrs’ chjice of fmod to the place wheee they can be avsiled the best. The inlut of our dystem will be in the form of

morphismes de Calabi et graphe de Reeb tore.* R. Acad. Paris, Ser. I *Sur points singuliers d’une de Pfaff complètement ou d’une fonction numérique.* C.R. Acad. Paris, 222 :847-849, 1946. --- abstract: 'This paper presents [*Fiducia*]{}, a food review involving a pipeline which processes restaurant-related reviews obtained from Zomato (India’s largest restaurant and service). is to popular café food items and manages to identify relevant information pertaining to each item separately the reviews. It uses a sentiment check on pieces of text and suggests an appropriate restaurant for particular depending on and similarity. results show that sentiment analyzer module of Fiducia achieves an accuracy of over 85% and our final recommender system achieves RMSE of beating other author: 'Mansi Ayush Agarwal, Deepak Chakraborty' - Ayush Agarwal - Deepak Chakraborty title: 'Fiducia: A Personalized Food Recommender System Zomato' --- Introduction ============ Zomato[^1] is the largest restaurant directory application used to suggest nearby restaurants to the favored cuisines, prices and other amenities like home delivery. It hosts vast information restaurant which is not properly according to users’ If manage get a of restaurants recognized according to user preferences, the usability of the will greatly increase. Zomato fails to provide details at the level. restaurant can be popular but it might be for a certain food a popular in Usually need to go through reviews to find out what’s best the restaurant serves [**Problem Definition:**]{} In this paper, we aim to a system that maps customers’ choice of to the place where they can be availed the best. The input our system in the form of

morphismes de Calabi et graphE de Reeb sur Le torE.* C. R. acaD. SCi. PaRis, SEr. I vol.343 323-328 (2006). G. Reeb, *Sur LEs poInts singuliers d’une formE de PfAfF CompLÈtEment IntégraBLe OU D’unE fOnCtiOn NUmÉriquE.* C.R. acad. Sci. paris, 222 :847-849, 1946. --- abstrAct: 'thIs paper preseNTs [*fiducia*]{}, a foOd rEview system iNvoLving a PiPelINe whiCh pRocesSes resTAurant-Related reViEWs obtaINed from zOMaTo (InDia’s largest restauRAnT Search and discoVery seRvICe). fIDucIa iS specific tO pOpulaR Café fooD ItEMS And MAnages to identIfy relevant INfoRmatioN pErtAIning tO each ItEM seParately in tHe reViews. It usEs a senTIment chECk on theSe piecEs oF teXt anD AcCoRdiNgLY suGGeSts AN apPropriatE rEsTauraNt foR THE PartIcuLar iTem dePending on user-IteM and ITem-Item sImilaRity. exPerimEntal rEsultS sHow that the sentiMent Analyzer mOduLe Of FIdUcia aCHieves An aCcuRacy of oVer 85% and oUR fiNaL RECoMmender system achieVeS AN RmSE of aboUt $1.01$ beatINg OtHEr baseliNeS.' auThor: - 'mANsi GoEl, AyUSh agarwal, DEepak THUkRaL, Tanmoy chAkraboRtY' - AyUsh agarwAL - DeePak ThuKral - TanmOy ChaKRaborty title: 'FiDUcia: A PersonalIZeD fOoD recoMmeNder System fOr ZoMAto' --- 0 INtroDUcTioN ============ zomatO[^1] is thE lARgESt Indian restaurant dIrEctory AppliCation used to sUggest nearBY REstauranTs acCOrDIng to the favoreD cuisInes, prices ANd other aMenitIes like hOme deliveRY. it hosts vAst InfOrmAtiON AbOut each restauRANt whIcH is not cLusTered prOpeRly AccOrdInG to the useRs’ convenIeNcE. IF wE maNage tO Get a collEcTioN oF reStaurANts recOgnizEd acCoRdINg tO user prEFeRENces, ThE uSabiLitY oF the aPpliCAtiOn will gReatly incReaSE. ZomAtO fAils to pRovide details At The individUaL itEm leveL. a RestauraNt can be very popular but it MIght not Be fAmous For a Certain foOd iTem; wheReaS A less pOpular RestaUrAnt MIGht exCEL iN thAt. usually cusTOMerS need To Go thRough thE reviews to find out wHAt’s The best the resTauRant SERvEs aND wHAt sHoULd bE AVoided.\ [**Problem DeFinition:**]{} In ThIS pAper, we aim tO DesIgN a systeM that maPs cusTOmers’ chOice of fooD to the plaCe WherE THey Can be availEd the besT. The input OF our sYStEm wilL be In the fOrM of

morphismes de Calabi et gr aphe de Re eb su r l e t or e.*C. R . Acad. Sci. P a ris, Ser. I vol.343 323-32 8 (20 06 ) . G. Re eb, * Sur les po i n tssi ng uli er s d ’unefor me de P faff compl ète me nt intégrabl e o u d’une fo nct ion numériqu e.* C.R.Ac ad. Sci.Par is, 2 22 :84 7 -849,1946. -- -a bstrac t : 'This p ap er p resents [*Fiducia * ]{ } , a food revie w syst em in v o lvi nga pipeline w hichp rocesse s r e s t aur a nt-related re views obtai n edfrom Z om ato (India ’s la rg e strestaurantsear ch and di scover y servic e ). Fidu cia is sp eci fict opo pul ar caf é f ood ite ms and m an ag es to ide n t i f y re lev antinfor mation pertai nin g to eac h ite m sep arat el y inthe re views .It uses a senti ment check on th es e p ie ces o f textand ac cording ly sugg e sts a n a pp ropriate restauran tf o rthe part icular it em dependin gonuser - i tem a nd i t em -item si milari t y. E xperime nt al res ul tssho w tha t the senti ment ana lyzer module of Fidu c ia achieves a n a c c ur a cy o f o ver 85% and our fina l re c om men d er sy stemac h ie v es an RMSE of about $ 1.01$beati ng other base lines.' au t h o r: - 'Ma nsiG oe l , Ayush Agarwa l, De epak Thukr a l, Tanmo y Cha kraborty ' - Ayush A garwal - De epa k T huk r a l- Tanmoy Chak r a bort ytitle:'Fi ducia:A P ers ona liz ed Food Rec ommender S ys te mfor Zoma t o' --- 0 In tr odu ction ====== ===== = Z om at o [^1 ] is th e l a r gest I nd ianres ta urant dir e cto ry appl ication u sed to s ug ge st near by restaurant saccordingto th e favo r e d cuisin es, prices and other am e nitieslik e hom e de livery. I t h osts v ast inform ationabout e ach r estau r a nt wh ic h is not c l u ste red p ro perl y accor ding to the users’ con venience. Ifwemana g e t o g e ta co ll e cti o n of restaurants recognize da cc ording tou ser p referen ces, th e usa b ility o f the app licationwi ll g r e atl y increase . Zomato fails to provi d edetai lsat the i ndi vidua l item lev el. A resta ur ant ca n beve ry popul ar but it might not befamous fora c ertain fo odi tem ; whereas a l ess popula r r est auran t m i ght e xcel in th a t. Us uall y customer s n eed t ogo throught h e re views to find o ut w hat’s the best th e restaurant se rves a ndwha t sho ul d be avoided.\ [* *P r o blem Def in ition:**]{} In this p a per,we aim to de sign as y st e m that map s c ustomers’ ch oi c e of fo od t o the p lace w here t hey ca n bea v ailed the best.The i n p ut of our syst em will b e inthe form o f

morphismes de_Calabi et_graphe de Reeb sur_le tore.*_C._R. Acad._Sci._Paris, Ser. I_vol.343 323-328 (2006)._G. Reeb, *Sur les_points singuliers d’une_forme_de Pfaff complètement intégrable ou d’une fonction numérique.* C.R. Acad. Sci. Paris, 222 :847-849,_1946. _--- abstract: 'This_paper_presents_[*Fiducia*]{}, a food review system_involving a pipeline which processes_restaurant-related reviews_obtained from Zomato (India’s largest restaurant search and_discovery_service). Fiducia is_specific to popular café food items and manages to_identify relevant information pertaining to each_item separately in_the_reviews._It uses a sentiment_check on these pieces of text_and accordingly suggests an appropriate restaurant_for the particular item depending on user-item_and item-item similarity. Experimental results show_that the sentiment analyzer module_of Fiducia_achieves an accuracy of over_85% and our_final recommender_system achieves an_RMSE of about $1.01$ beating other_baselines.' author: - 'Mansi Goel,_Ayush Agarwal, Deepak Thukral, Tanmoy Chakraborty' -_Ayush_Agarwal - Deepak Thukral -_Tanmoy_Chakraborty title:_'Fiducia: A_Personalized Food Recommender_System_for Zomato' --- 0 Introduction ============ Zomato[^1]_is_the largest Indian restaurant directory application_used_to suggest nearby restaurants according to the_favored cuisines, prices and_other_amenities like home delivery._It hosts vast information about_each restaurant which is not clustered_properly according_to the_users’ convenience. If we manage to get a collection of restaurants_recognized according to user preferences, the_usability of the application_will greatly_increase._Zomato fails to_provide_details at_the individual item level. A restaurant can_be very_popular but it might not be_famous for a certain_food_item; whereas a less popular restaurant_might excel in that. Usually customers_need to go through the_reviews_to_find out what’s the best_the restaurant serves and what should_be avoided.\ [**Problem Definition:**]{}_In this paper, we aim to design_a_system that maps customers’ choice of_food_to the place where they can_be_availed_the best. The input of_our system will be in the_form of

$ form the sides of a right triangle (by Pythagoras theorem). Then $p_{1}+p_{2}\geq p$ follows by the triangle inequality. This proof heavily uses results specific to Euclidean geometry, and cannot be easily extended to hyperbolic geometry. As we remarked in the introduction, although the result is false in hyperbolic geometry as stated, we would expect a similar result to be true for polygons *bounded close to the origin*. Indeed, we have – \[thm\_weak\] Let $P,P_{1},P_{2}$ be regular hyperbolic $n$-gons, for $n\geq3$, with areas $A,A_{1},A_{2}$, and interior angles $\theta,\theta_{1},\theta_{2}$ respectively. Suppose that 1. $\theta\geq\cos^{-1}\left(-1+2\sin\left(\frac{\pi}{n}\right)\right)$ 2. $A_{1}+A_{2}=A$ Then $$\operatorname*{Perim}\left(P_{1}\right)+\operatorname*{Perim}\left(P_{2}\right)\geq\operatorname*{Perim}\left(P\right).$$ Here, equality only occurs if one of $A_1$ and $A_2$ is $0$, which is a degenerate case. Throughout the proof, when we refer to “$n-$gons”, we mean hyperbolic regular $n$-gons. Consider an $n$-gon $\Gamma=\Gamma\left(\alpha,p,a\right)$ with angle $\alpha$, perimeter $p$, area $a=\left(n-2\right)\pi-n\alpha$. We consider a triangulated section of $\Gamma$ as in Figure \[fig:1\]. ![A sector of the $n$-gon[]{data-label="fig:1"}](hyptriangle.png) Consider the right angled triangle $\Delta OAM$ in Figure \[fig:1\]. We know[^1] that (see formula (v) in Theorem 2.2.2 [@Bus]) $$\cosh\left(\frac{s}{2}\right)=\frac{\cos\left(\frac{\pi}{n}\right)}{\sin\left(\frac{\alpha}{2}\right)}.$$ Thus, $s=2\cosh^{-1}\left(\frac{\cos\

$ form the sides of a right triangle (by Pythagoras theorem). Then $ p_{1}+p_{2}\geq p$ follow by the triangulum inequality. This proof heavily use solution specific to Euclidean geometry, and cannot be easily widen to hyperbolic geometry. As we remarked in the introduction, although the result is false in hyperbolic geometry as stated, we would have a bun in the oven a similar result to be dependable for polygon * bounded close to the beginning *. Indeed, we have – \[thm\_weak\ ] Let $ phosphorus, P_{1},P_{2}$ be regular hyperbolic $ n$-gons, for $ n\geq3 $, with areas $ A, A_{1},A_{2}$, and interior lean $ \theta,\theta_{1},\theta_{2}$ respectively. Suppose that 1. $ \theta\geq\cos^{-1}\left(-1 + 2\sin\left(\frac{\pi}{n}\right)\right)$ 2. $ A_{1}+A_{2}=A$ Then $ $ \operatorname*{Perim}\left(P_{1}\right)+\operatorname*{Perim}\left(P_{2}\right)\geq\operatorname*{Perim}\left(P\right).$$ Here, equality only happen if one of $ A_1 $ and $ A_2 $ is $ 0 $, which is a degenerate event. Throughout the proof, when we refer to “ $ n-$gons ”, we mean hyperbolic regular $ n$-gons. Consider an $ n$-gon $ \Gamma=\Gamma\left(\alpha, phosphorus, a\right)$ with angle $ \alpha$, perimeter $ p$, area $ a=\left(n-2\right)\pi - n\alpha$. We consider a triangulated section of $ \Gamma$ as in Figure   \[fig:1\ ]. ! [ A sector of the $ n$-gon[]{data - label="fig:1"}](hyptriangle.png) Consider the correct slant triangle $ \Delta OAM$ in Figure   \[fig:1\ ]. We know[^1 ] that (see formula (v) in Theorem 2.2.2   [ @Bus ]) $ $ \cosh\left(\frac{s}{2}\right)=\frac{\cos\left(\frac{\pi}{n}\right)}{\sin\left(\frac{\alpha}{2}\right)}.$$ Thus, $ s=2\cosh^{-1}\left(\frac{\cos\

$ fogm the sides of a right uriangle (by Pythayiras tieorem). Fhen $p_{1}+p_{2}\gdq p$ follows by the triangle ibequaoity. This proof heavily uses resllts specufic ro Euclideeh geomebxy, ans canuov be easily extgnded to hyparbolic geometsy. Xs we remarked in the introduction, alehough yhf result is fajse pn hypsgbilic geometry as stated, we wohld exptct a similar resukt to be true for polygons *boujded close to the lrigin*. Indegs, wq have – \[thm\_weay\] Let $P,P_{1},P_{2}$ be regular hiperbolic $n$-gons, for $n\geq3$, with ardas $A,C_{1},A_{2}$, and intetnir wtgles $\theta,\vheta_{1},\treta_{2}$ respectlnely. Su[pose tnat 1. $\theta\geq\cps^{-1}\lxft(-1+2\sun\left(\frac{\pi}{n}\right)\rigit)$ 2. $A_{1}+A_{2}=A$ Then $$\operatorgame*{Perim}\neyt(P_{1}\right)+\operatorname*{Pwrum}\lefj(P_{2}\rigvt)\ged\ipefatkriams*{Perim}\pefv(P\right).$$ Hers, equality inly occurs if one pf $Q_1$ and $A_2$ is $0$, wgich if w degenerate case. Throughout the proof, wven we refer to “$n-$gons”, we mwan hyperbolic regulat $n$-gons. Cogsider an $n$-gon $\Gamma=\Gamma\left(\alpha,p,a\right)$ with ancle $\amoha$, pcrimdreg $p$, area $a=\left(n-2\right)\pi-n\alpha$. We consider a tryznbukated section jf $\Gamma$ as ij Gygure \[fig:1\]. ![A secjor of cge $n$-gon[]{data-label="fig:1"}](hjptriandle.pnt) Consider the right angled triangle $\Deltq OAM$ in Figlre \[fug:1\]. We know[^1] that (sez formula (v) nn Theprem 2.2.2 [@Nus]) $$\cosh\left(\frac{s}{2}\right)=\fxac{\cos\meft(\frac{\pi}{n}\gight)}{\sin\lswt(\frac{\alpha}{2}\right)}.$$ Ghux, $v=2\cosh^{-1}\lefu(\wrac{\cos\

$ form the sides of a right Pythagoras Then $p_{1}+p_{2}\geq follows by the uses specific to Euclidean and cannot be extended to hyperbolic geometry. As we in the introduction, although the result is false in hyperbolic geometry as stated, would expect a similar result to be true for polygons *bounded close to origin*. we – Let $P,P_{1},P_{2}$ be regular hyperbolic $n$-gons, for $n\geq3$, with areas $A,A_{1},A_{2}$, and interior angles $\theta,\theta_{1},\theta_{2}$ respectively. that 1. $\theta\geq\cos^{-1}\left(-1+2\sin\left(\frac{\pi}{n}\right)\right)$ 2. $A_{1}+A_{2}=A$ Then $$\operatorname*{Perim}\left(P_{1}\right)+\operatorname*{Perim}\left(P_{2}\right)\geq\operatorname*{Perim}\left(P\right).$$ Here, only occurs if one $A_1$ and $A_2$ is $0$, is degenerate case. the when refer to “$n-$gons”, mean hyperbolic regular $n$-gons. Consider an $n$-gon $\Gamma=\Gamma\left(\alpha,p,a\right)$ with angle $\alpha$, perimeter $p$, area $a=\left(n-2\right)\pi-n\alpha$. We consider triangulated section as in \[fig:1\]. sector the $n$-gon[]{data-label="fig:1"}](hyptriangle.png) Consider angled triangle $\Delta OAM$ in Figure that (see formula (v) in Theorem 2.2.2 [@Bus]) Thus, $s=2\cosh^{-1}\left(\frac{\cos\

$ form the sides of a right trianGle (by PythaGoras TheOreM). THen $p_{1}+P_{2}\geq P$ follows by the tRIangLe inequality. This proof hEavilY uSEs reSUlTs speCific to eUcLIDeaN gEoMetRy, ANd CannoT be Easily eXtended to hYpeRbOlic geometry. aS wE remarked iN thE introductioN, alThough ThE reSUlt is FalSe in hYperboLIc geomEtry as staTeD, We woulD Expect a SIMiLar rEsult to be true for pOLyGOns *bounded closE to the OrIGiN*. iNdeEd, wE have – \[thm\_weAk\] let $P,P_{1},p_{2}$ Be regulAR hYPERboLIc $n$-gons, for $n\geQ3$, with areas $A,a_{1},a_{2}$, anD interIoR anGLes $\theTa,\theTa_{1},\THetA_{2}$ respectiveLy. SuPpose that 1. $\Theta\gEQ\cos^{-1}\lefT(-1+2\Sin\left(\Frac{\pi}{N}\riGht)\RighT)$ 2. $a_{1}+A_{2}=a$ THen $$\OpERatORnAme*{pEriM}\left(P_{1}\riGhT)+\oPeratOrnaME*{pERim}\lEft(p_{2}\rigHt)\geq\Operatorname*{PEriM}\lefT(p\riGht).$$ HeRe, equAlitY oNly ocCurs if One of $a_1$ aNd $A_2$ is $0$, which is a deGeneRate case. THroUgHouT tHe proOF, when wE reFer To “$n-$gons”, We mean hYPerBoLIC ReGular $n$-gons. Consider An $N$-GOn $\gamma=\GamMa\left(\ALpHa,P,A\right)$ wiTh AngLe $\alPHA$, periMeteR $P$, aRea $a=\left(N-2\right)\PI-n\AlPha$. We coNsIder a tRiAngUlaTed seCTion Of $\GammA$ as in FigUre \[fiG:1\]. ![a sector of the $n$-gON[]{data-label="fig:1"}](HYpTRIaNGle.pNg) COnsider the rIght ANgleD triANgLe $\DELta OAm$ in FiGuRE \[fIG:1\]. We know[^1] that (see formuLa (V) in TheOrem 2.2.2 [@BUs]) $$\cosh\left(\fraC{s}{2}\right)=\fraC{\COS\left(\fraC{\pi}{n}\RIgHT)}{\sin\left(\frac{\alPha}{2}\riGht)}.$$ Thus, $s=2\coSH^{-1}\left(\fraC{\cos\

$ form the sides of a righ t triangle (byPyt hag or as t heor em). Then $p_{ 1 }+p_ {2}\geq p$ follows bythe t ri a ngle in equal ity. T h is p roo fhe avi ly us es re sul ts spec ific to Eu cli de an geometry, an d cannot b e e asily extend edto hyp er bol i c geo met ry. As wer emarke d in thein t roduct i on, alt h o ug h th e result is false in hyperbolic geo metryas st a t ed, we would exp ec t a s i milar r e su l t tob e true for po lygons *bou n ded close t o t h e orig in*.In d eed , we have – \[ thm\_weak \] Let $P,P_{1 } ,P_{2}$ be re gul arhype r bo li c $ n$ - gon s ,for $n\ geq3$, w it hareas $A, A _ { 1 },A_ {2} $, a nd in terior angles $\ thet a ,\t heta_ {1},\ thet a_ {2}$respec tivel y. Suppose that 1. $\theta\g eq\ co s^{ -1 }\lef t (-1+2\ sin \le ft(\fra c{\pi}{ n }\r ig h t ) \r ight)$ 2. $A_{1} +A _ { 2} =A$ The n $$\o p er at o rname*{P er im} \lef t ( P_{1} \rig h t) +\operat orname * {P er im}\lef t( P_{2}\ ri ght )\g eq\op e rato rname* {Perim}\ left( P \right).$$ Her e , equality on l yo c cu r s if on e of $A_1$and$ A_2$ is$ 0$ , w h ich i s a d eg e ne r ate case. Througho ut the p roof, when we refe r to “$n-$ g o n s”, we m eanh yp e rbolic regular $n$- gons. Cons i der an $ n$-go n $\Gamm a=\Gamma\ l e ft(\alph a,p ,a\ rig ht) $ wi th angle $\al p h a$,pe rimeter $p $, area $a =\l eft (n- 2\ right)\pi -n\alpha $. W eco nsi der a triangul at edse cti on of $\Gamm a$ as inFi gu r e \ [fig:1\ ] .! [A s ec to r of th e$n$-g on[] { dat a-label ="fig:1"} ](h y ptri an gl e.png) Consider the r ight angle dtri angle$ \ Delta OA M$ in Figure \[fig:1\]. We know [^1 ] tha t (s ee formul a ( v) inThe o rem 2. 2.2 [@ Bus]) $ $\c o s h\lef t ( \f rac {s }{2}\right ) = \fr ac{\c os \lef t(\frac {\pi}{n}\right)}{\ s in\ left(\frac{\a lph a}{2 } \ ri ght ) }. $ $ Th u s,$ s =2\cosh^{-1}\le ft(\frac{\ co s \

$ form_the sides_of a right triangle_(by Pythagoras_theorem)._Then $p_{1}+p_{2}\geq_p$_follows by the_triangle inequality. This proof_heavily uses results specific_to Euclidean geometry,_and_cannot be easily extended to hyperbolic geometry. As we remarked in the introduction, although the_result_is false_in_hyperbolic_geometry as stated, we would_expect a similar result to_be true_for polygons *bounded close to the origin*. Indeed,_we_have – \[thm\_weak\] Let_$P,P_{1},P_{2}$ be regular hyperbolic $n$-gons, for $n\geq3$, with areas_$A,A_{1},A_{2}$, and interior angles $\theta,\theta_{1},\theta_{2}$ respectively._Suppose that 1. _$\theta\geq\cos^{-1}\left(-1+2\sin\left(\frac{\pi}{n}\right)\right)$ 2.__$A_{1}+A_{2}=A$ Then $$\operatorname*{Perim}\left(P_{1}\right)+\operatorname*{Perim}\left(P_{2}\right)\geq\operatorname*{Perim}\left(P\right).$$ Here, equality_only occurs if one of $A_1$_and $A_2$ is $0$, which is_a degenerate case. Throughout the proof, when we_refer to “$n-$gons”, we mean hyperbolic_regular $n$-gons. Consider an $n$-gon_$\Gamma=\Gamma\left(\alpha,p,a\right)$ with_angle $\alpha$, perimeter $p$, area_$a=\left(n-2\right)\pi-n\alpha$. We consider_a triangulated_section of $\Gamma$_as in Figure \[fig:1\]. ![A sector of the_$n$-gon[]{data-label="fig:1"}](hyptriangle.png) Consider the right_angled triangle $\Delta OAM$ in Figure \[fig:1\]._We_know[^1] that (see_formula_(v)_in Theorem_2.2.2 [@Bus]) $$\cosh\left(\frac{s}{2}\right)=\frac{\cos\left(\frac{\pi}{n}\right)}{\sin\left(\frac{\alpha}{2}\right)}.$$ Thus, $s=2\cosh^{-1}\left(\frac{\cos\

with Hardy space function kernel, Acta Math. Sinica (English Series) [**16**]{} (2000), 593–600. Y. Ding, D. Fan and Y. Pan, Weighted boundedness for a class of rough Marcinkiewicz integral, Indiana Univ. Math. J. **48** (1999), 1037-1055. A problem on rough Marcinkiewicz functions, J. Austral. Math. Soc. [**71**]{} (2001), 1–9., J. Math. Anal. Appl. [**275**]{} (2002), 60–68. Y. Ding, Q. Xue and K. Yabuta, A remark to the $L^2$ boundedness of parametric Marcinkiewicz integral, J. Math. Anal. Appl. **387** (2012), 691-697. J. Duoandikoetxea, Weighted norm inequalities for homogeneous singular integrals, Trans. Amer. Math. Soc. **336** (1993), 869-880. J. Duoandikoetxea and J. L. Rubio de Francia, Maximal and singular integrals via Fourier transform estimates. Invent. Math. **84** (1986), 541-561. L. Grafakos, Modern Fourier analysis, GTM250, 2nd Edition, Springer, New York, 2008. X. Guo and G. Hu, Compactness of the commutators of homogeneous singular integral operators Sci. China Math. 58 (2015), 2347-2362. G. Hu, $L^p(\mathbb{R}^n)$ boundedness for a class of $g$-functions and its applications, J. Hokkaido Math. **32** (2003), 497-521. G. Hu and D. Yan, On commutator of the Marcinkiewicz integral, J. Math. Anal. Appl. [**283**]{} (2003), 351–361. Y. Komori and S. Shirai, Weighted Morrey spaces and a singular integral operator, Math. Nachr. **282** (2009), 219-232. D. S. Kurtz and R. Wheeden, Results on weighted norm inequalities for multipliers. Trans. Amer. Math. Soc. **255**(1979), 343-362. S. Mao, Y. Sawano and H. Wu, On the compactness of commutators for rough Marcinkiewicz integral operators, Taiwanese J. Math. **19** (

with Hardy space function kernel, Acta Math. Sinica (English Series) [ * * 16 * * ] { } (2000), 593–600. Y. Ding, D. Fan and Y. Pan, Weighted finiteness for a course of harsh Marcinkiewicz integral, Indiana Univ. Math. J. * * 48 * * (1999), 1037 - 1055. A trouble on rough Marcinkiewicz function, J. Austral. Math. Soc. [ * * 71 * * ] { } (2001), 1–9. , J. Math. Anal. Appl. [ * * 275 * * ] { } (2002), 60–68. Y. Ding, Q. Xue and K. Yabuta, A remark to the $ L^2 $ boundedness of parametric Marcinkiewicz built-in, J. Math. Anal. Appl. * * 387 * * (2012), 691 - 697. J. Duoandikoetxea, Weighted norm inequalities for homogeneous remarkable integrals, Trans. Amer. Math. Soc. * * 336 * * (1993), 869 - 880. J. Duoandikoetxea and J. L. Rubio de Francia, Maximal and singular integral via Fourier transform estimates. Invent. Math. * * 84 * * (1986), 541 - 561. L. Grafakos, Modern Fourier analysis, GTM250, 2nd Edition, Springer, New York, 2008. X. Guo and G. Hu, Compactness of the commutators of homogeneous remarkable integral operators Sci. China Math. 58 (2015), 2347 - 2362. G. Hu, $ L^p(\mathbb{R}^n)$ boundedness for a class of $ g$-functions and its applications, J. Hokkaido Math. * * 32 * * (2003), 497 - 521. G. Hu and D. Yan, On commutator of the Marcinkiewicz built-in, J. Math. Anal. Appl. [ * * 283 * * ] { } (2003), 351–361. Y. Komori and S. Shirai, Weighted Morrey spaces and a singular integral hustler, Math. Nachr. * * 282 * * (2009), 219 - 232. D. S. Kurtz and R. Wheeden, Results on weighted norm inequalities for multipliers. Trans. Amer. Math. Soc. * * 255**(1979), 343 - 362. S. Mao, Y. Sawano and H. Wu, On the concentration of commutators for grating Marcinkiewicz integral operators, Taiwanese J. Math. * * 19 * * (

wihh Hardy space function yernel, Acta Matk. Sinice (Engliah Serier) [**16**]{} (2000), 593–600. Y. Ding, D. Fan and Y. Pan, Xeigyted voundedness for a clasr of rougj Marcinjiewmcz integral, Indmzna Uniy. Matg. J. **48** (1999), 1037-1055. A problem on tough Marcinniewicz functimnr, L. Austral. Math. Soc. [**71**]{} (2001), 1–9., J. Math. Anal. Ap[l. [**275**]{} (2002), 60–68. Y. Dlng, Q. Xue and H. Yantta, Z remark to the $L^2$ boundedness of pzrametrpc Marcinkiewicz ontegral, J. Math. Anal. Appl. **387** (2012), 691-697. J. Duoandikoetxea, Welghted norm inezyalities for homogeneolv singular integrals, Trans. Amer. Math. Soc. **336** (1993), 869-880. J. Buoandikoetzeq ajg J. L. Rubio de Fgancia, Maximal and sitgular ontegrals via Nourixr teansform estimates. Intent. Math. **84** (1986), 541-561. L. Grafahos, Modert Yourier analysis, GTM250, 2bd Edithon, Vprivter, Nes Bori, 2008. X. Gko end G. Hu, Cojpactness od the commutators og rimogeneous sihgular igtegral operators Sci. China Math. 58 (2015), 2347-2362. G. Hl, $L^p(\jathbb{R}^n)$ boundedness foe a class of $g$-functiojs and itf applications, J. Hokkaido Math. **32** (2003), 497-521. G. Hu and D. Yan, Mn cojoutcbjf ov the Marcinkiewicz integral, J. Math. Anal. Appl. [**283**]{} (2003), 351–361. Y. Komori and S. Fhirai, Weigntfd Iorrey spaces and a sihgular integral opfrator, Iath. Bachr. **282** (2009), 219-232. Q. S. Lurtz and R. Wheeden, Results on weighted borm inequalities yor multiplizrs. Trsns. Aker. Math. Soc. **255**(1979), 343-362. S. Mao, Y. Rawaho and H. Wu, On the ckopactness of comoutstmrs for rough Marcinkiewics integrao opzrators, Gaiwsnese T. Math. **19** (

with Hardy space function kernel, Acta Math. Series) (2000), 593–600. Ding, D. Fan for class of rough integral, Indiana Univ. J. **48** (1999), 1037-1055. A problem rough Marcinkiewicz functions, J. Austral. Math. Soc. [**71**]{} (2001), 1–9., J. Math. Anal. [**275**]{} (2002), 60–68. Y. Ding, Q. Xue and K. Yabuta, A remark to $L^2$ of Marcinkiewicz J. Math. Anal. Appl. **387** (2012), 691-697. J. Duoandikoetxea, Weighted norm inequalities for homogeneous singular integrals, Amer. Math. Soc. **336** (1993), 869-880. J. Duoandikoetxea J. L. Rubio de Maximal and singular integrals via transform Invent. Math. (1986), L. Modern Fourier analysis, 2nd Edition, Springer, New York, 2008. X. Guo and G. Hu, Compactness of the commutators of homogeneous integral operators Math. 58 2347-2362. Hu, boundedness for a $g$-functions and its applications, J. Hokkaido 497-521. G. Hu and D. Yan, On commutator the Marcinkiewicz J. Math. Anal. Appl. [**283**]{} (2003), Y. Komori and S. Shirai, Weighted Morrey spaces a singular integral operator, Math. Nachr. **282** (2009), 219-232. D. S. Kurtz and R. Wheeden, weighted norm inequalities for Trans. Amer. Math. **255**(1979), S. Y. and H. On the compactness of commutators for rough Marcinkiewicz integral operators, Taiwanese Math. **19** (

with Hardy space function kerNel, Acta MatH. SiniCa (ENglIsH SerIes) [**16**]{} (2000), 593–600. Y. ding, D. Fan and Y. PaN, weigHted boundedness for a claSs of rOuGH MarCInKiewiCz integRAl, iNDiaNa unIv. MAtH. j. **48** (1999), 1037-1055. A ProblEm oN rough MArcinkiewiCz fUnCtions, J. AustrAL. MAth. Soc. [**71**]{} (2001), 1–9., J. MatH. AnAl. Appl. [**275**]{} (2002), 60–68. Y. Ding, Q. xue And K. YaBuTa, A REmark To tHe $L^2$ boUndednESs of paRametric MArCInkiewICz integRAL, J. math. anal. Appl. **387** (2012), 691-697. J. DuoandikOEtXEa, Weighted norm InequaLiTIeS FOr hOmoGeneous sinGuLar inTEgrals, TRAnS. aMEr. MATh. Soc. **336** (1993), 869-880. J. DuoandiKoetxea and J. l. rubIo de FrAnCia, mAximal And siNgULar Integrals viA FouRier transForm esTImates. INVent. MatH. **84** (1986), 541-561. L. GrafAkoS, MoDern fOuRiEr aNaLYsiS, gTm250, 2nd eDitIon, SprinGeR, NEw YorK, 2008. X. GuO AND g. Hu, COmpActnEss of The commutatorS of HomoGEneOus siNgulaR intEgRal opEratorS Sci. CHiNa Math. 58 (2015), 2347-2362. G. Hu, $L^p(\mathBb{R}^n)$ BoundedneSs fOr A clAsS of $g$-fUNctionS anD itS applicAtions, J. hOkkAiDO mAtH. **32** (2003), 497-521. G. Hu and D. Yan, On commuTaTOR oF the MarcInkiewICz InTEgral, J. MaTh. anaL. AppL. [**283**]{} (2003), 351–361. y. komorI and s. shIrai, WeigHted MoRReY sPaces anD a SingulAr IntEgrAl opeRAtor, math. NaChr. **282** (2009), 219-232. D. S. KurTz and r. wheeden, Results ON weighted norm INeQUAlITies For Multipliers. tranS. amer. math. sOc. **255**(1979), 343-362. s. MaO, y. SawaNo and h. WU, on THe compactness of commUtAtors fOr rouGh MarcinkiewiCz integral OPERators, TaIwanESe j. math. **19** (

with Hardy space function kernel, A cta M ath . S in ica(Eng lish Series) [ * *16* *]{} (2000), 593–600.Y. Di ng , D.F an andY. Pan, We i g hte dbo und ed n es s for aclass o f rough Ma rci nk iewicz integ r al , IndianaUni v. Math. J.**4 8** (1 99 9), 1037- 105 5. Aproble m on ro ugh Marci nk i ewiczf unction s , J . Au stral. Math. Soc. [* * 71**]{} (2001) , 1–9. ,J .M a th. An al. Appl.[* *275* * ]{} (20 0 2) , 6 0–6 8 . Y. Ding, Q. Xue and K. Yab uta, A r ema r k to t he $L ^2 $ bo undedness o f pa rametricMarcin k iewiczi ntegral , J. M ath . A nal. Ap pl . * *3 8 7** (2 012 ) , 6 91-697.J. D uoand ikoe t x e a , We igh tednorminequalitiesfor hom o gen eoussingu larin tegra ls, Tr ans.Am er. Math. Soc.**33 6** (1993 ),86 9-8 80 . J.D uoandi koe txe a and J . L. Ru b iode F r an cia, Maximal and s in g u la r integr als vi a F ou r ier tran sf orm est i m ates. Inv e nt . Math.**84** (1 98 6), 541 -5 61. L. G raf ako s, Mo d ernFourie r analys is, G T M250, 2nd Edit i on, Springer, Ne w Yo r k, 2 008 . X. Guo an d G. Hu,Comp a ct nes s of t he co mm u ta t ors of homogeneoussi ngular inte gral operator s Sci. Chi n a Math. 58 (20 1 5) , 2347-2362. G. Hu,$L^p(\math b b{R}^n)$ boun dednessfor a cla s s of $g$- fun cti ons an d it s application s , J.Ho kkaidoMat h. **32 **(20 03) , 4 97 -521. G.Hu and D .Ya n, O n c ommut a tor of t he Ma rc ink iewic z integ ral,J. M at h. Ana l. Appl . [ * * 283* *] {} (20 03) ,351–3 61.Y . K omori a nd S. Shi rai , Wei gh te d Morre y spaces andasingular i nt egr al ope r a tor, Mat h. Nachr. **282** (2009 ) , 219-2 32. D. S . Ku rtz and R . W heeden , R e sultson wei ghted n orm i nequa l i ti esfo r multipli e r s.Trans .Amer . Math. Soc. **255**(1979 ) , 3 43-362. S. Ma o,Y. S a w an o a n dH . W u, Ont h e compactness o f commutat or s f or rough M a rci nk iewiczintegra l ope r ators,Taiwanese J. Math. * *19* * (

with_Hardy space_function kernel, Acta Math._Sinica (English_Series)_[**16**]{} (2000),_593–600._Y. Ding, D._Fan and Y._Pan, Weighted boundedness for_a class of_rough_Marcinkiewicz integral, Indiana Univ. Math. J. **48** (1999), 1037-1055. A problem on rough Marcinkiewicz_functions,_J. Austral._Math._Soc._[**71**]{} (2001), 1–9., J. Math._Anal. Appl. [**275**]{} (2002), 60–68._Y. Ding,_Q. Xue and K. Yabuta, A remark to_the_$L^2$ boundedness of_parametric Marcinkiewicz integral, J. Math. Anal. Appl. **387** (2012),_691-697. J. Duoandikoetxea, Weighted norm inequalities_for homogeneous singular_integrals,_Trans._Amer. Math. Soc. **336**_(1993), 869-880. J. Duoandikoetxea and J._L. Rubio de Francia, Maximal and_singular integrals via Fourier transform estimates. Invent._Math. **84** (1986), 541-561. L. Grafakos,_Modern Fourier analysis, GTM250, 2nd_Edition, Springer,_New York, 2008. X. Guo_and G. Hu,_Compactness of_the commutators of_homogeneous singular integral operators Sci. China_Math. 58 (2015),_2347-2362. G. Hu, $L^p(\mathbb{R}^n)$ boundedness for_a_class of $g$-functions_and_its_applications, J._Hokkaido Math. **32**_(2003),_497-521. G._Hu_and D. Yan, On commutator of_the_Marcinkiewicz integral, J. Math. Anal. Appl. [**283**]{}_(2003), 351–361. Y. Komori_and_S. Shirai, Weighted Morrey_spaces and a singular integral_operator, Math. Nachr. **282** (2009), 219-232._D. S._Kurtz and_R. Wheeden, Results on weighted norm inequalities for multipliers. Trans. Amer._Math. Soc. **255**(1979), 343-362. S. Mao,_Y. Sawano and H._Wu, On_the_compactness of commutators_for_rough Marcinkiewicz_integral operators, Taiwanese J. Math. **19** (

_n: \; \pi (x)\ge r^n \} ) \le \deg^n \varrho(r)^n$. By Chebyshev’s inequality and the Borel–Cantelli lemma, for any $r>\underline{r}$ and $\P$-almost surely for all large $n$, $$\#\left\{ x\in \T_n: \; \pi (x) \ge r^n \right\} \le n^2 \deg^n \varrho(r)^n. \label{Jn-ub1}$$ On the other hand, by (\[chernoff\]), $$\P \left\{ \exists x\in \T_n: \pi (x) \ge r^n\right\} \le \deg^n \varrho (r)^n.$$ For $r> \overline{r}$, the expression on the right-hand side is summable in $n$. By the Borel–Cantelli lemma, for any $r>\overline{r}$ and $\P$-almost surely for all large $n$, $$\max_{x\in \T_n} \pi (x) < r^n. \label{Jn-ub}$$ [*Proof of Theorem \[t:posrec\]: upper bound.*]{} Fix $\varepsilon>0$ such that $q+ 3\varepsilon < {1\over \deg}$. We follow the strategy given in Liggett ([@liggett], p. 103) by introducing a positive recurrent birth-and-death chain $(\widetilde{X_j}, \, j\ge 0)$, starting from $0$, with transition probability from $i$ to $i+1$ (for $i\ge 1$) equal to $${1\over \widetilde{\pi} (i)} \, \sum_{x\in \T_i} \pi(x) (1- \omega(x, {\buildrel \leftarrow \over x})),$$ where $\widetilde{\pi} (i) := \sum_{x\in \T_i} \pi(x)$. We note that $\widetilde{\pi}$ is a finite invariant measure for $(\widetilde{X_j})$. Let $$\tau_n := \inf \left\{ i\ge 1: \, X_i \in \T_n\right\}, \qquad n\ge 0.$$ By Liggett ([@liggett

_ n: \; \pi (x)\ge r^n \ }) \le \deg^n \varrho(r)^n$. By Chebyshev ’s inequality and the Borel – Cantelli lemma, for any $ r>\underline{r}$ and $ \P$-almost surely for all large $ n$, $ $ \#\left\ { x\in \T_n: \; \pi (x) \ge r^n \right\ } \le n^2 \deg^n \varrho(r)^n. \label{Jn - ub1}$$ On the other hand, by (\[chernoff\ ]), $ $ \P \left\ { \exists x\in \T_n: \pi (x) \ge r^n\right\ } \le \deg^n \varrho (r)^n.$$ For $ r > \overline{r}$, the formula on the proper - hand side is summable in $ n$. By the Borel – Cantelli lemma, for any $ r>\overline{r}$ and $ \P$-almost surely for all large $ n$, $ $ \max_{x\in \T_n } \pi (x) < r^n. \label{Jn - ub}$$ [ * validation of Theorem \[t: posrec\ ]: upper bound. * ] { } Fix $ \varepsilon>0 $ such that $ q+ 3\varepsilon < { 1\over \deg}$. We follow the scheme impart in Liggett ([ @liggett ], p.   103) by introducing a convinced recurrent birth - and - death range $ (\widetilde{X_j }, \, j\ge 0)$, starting from $ 0 $, with transition probability from $ i$ to $ i+1 $ (for $ i\ge 1 $) equal to $ $ { 1\over \widetilde{\pi } (i) } \, \sum_{x\in \T_i } \pi(x) (1- \omega(x, { \buildrel \leftarrow \over x})),$$ where $ \widetilde{\pi } (iodine): = \sum_{x\in \T_i } \pi(x)$. We note that $ \widetilde{\pi}$ is a finite invariant measure for $ (\widetilde{X_j})$. Let $ $ \tau_n: = \inf \left\ { i\ge 1: \, X_i \in \T_n\right\ }, \qquad n\ge 0.$$ By Liggett ([ @liggett

_n: \; \oi (x)\ge r^n \} ) \le \deg^n \varrmo(r)^n$. By Chebysher’w ineqnality znd the Corel–Cantelli lemma, for any $c>\undwrlint{g}$ and $\P$-almost surely wor all lwrge $n$, $$\#\lwft\{ e\in \T_n: \; \pi (x) \ge c^h \right\} \le k^2 \deg^u \tarrho(r)^n. \labgl{Jn-ub1}$$ On the other hand, by (\[cfexnoff\]), $$\P \left\{ \exists x\in \T_n: \pi (x) \ge r^g\right\} \ke \deg^n \varrho (r)^g.$$ For $w> \ovsgllne{r}$, the expression on the right-gand sive is summable on $n$. By the Borel–Cantelli pemmw, for any $r>\overlinf{r}$ and $\P$-almist fyrely for alu large $n$, $$\max_{x\in \T_n} \pi (x) < r^n. \label{Jn-ub}$$ [*Proof of Theurem \[c:posrec\]: uppgx boktd.*]{} Fix $\vare'silon>0$ such that $q+ 3\varepsinon < {1\ovrr \deg}$. We folloe tie srrategy given in Liggxtt ([@liggett], p. 103) by inttoducing a pksitive recurrent burth-atd-dedth zyaiv $(\wjdxtimde{X_j}, \, j\gx 0)$, starting from $0$, with transition probabikiej from $i$ to $i+1$ (for $i\de 1$) equal to $${1\over \widetilde{\pi} (i)} \, \sum_{x\in \T_p} \pi(s) (1- \omega(x, {\buildrel \leftqrrow \over x})),$$ where $\widgtilde{\pi} (i) := \sum_{x\in \T_i} \pi(x)$. We note that $\widetilde{\pi}$ is a finhte iivxrickb mexwuge for $(\widetilde{X_j})$. Let $$\tau_n := \inf \left\{ i\ge 1: \, X_y \im \N_n\right\}, \qquad n\ge 0.$$ By Liggett ([@llgbgtt

_n: \; \pi (x)\ge r^n \} ) \varrho(r)^n$. Chebyshev’s inequality the Borel–Cantelli lemma, surely all large $n$, x\in \T_n: \; (x) \ge r^n \right\} \le n^2 \varrho(r)^n. \label{Jn-ub1}$$ On the other hand, by (\[chernoff\]), $$\P \left\{ \exists x\in \T_n: (x) \ge r^n\right\} \le \deg^n \varrho (r)^n.$$ For $r> \overline{r}$, the expression on right-hand is in By the Borel–Cantelli lemma, for any $r>\overline{r}$ and $\P$-almost surely for all large $n$, $$\max_{x\in \T_n} (x) < r^n. \label{Jn-ub}$$ [*Proof of Theorem \[t:posrec\]: bound.*]{} Fix $\varepsilon>0$ such $q+ 3\varepsilon < {1\over \deg}$. follow strategy given Liggett p. by introducing a recurrent birth-and-death chain $(\widetilde{X_j}, \, j\ge 0)$, starting from $0$, with transition probability from $i$ to $i+1$ $i\ge 1$) $${1\over \widetilde{\pi} \, \T_i} (1- \omega(x, {\buildrel x})),$$ where $\widetilde{\pi} (i) := \sum_{x\in note that $\widetilde{\pi}$ is a finite invariant measure $(\widetilde{X_j})$. Let := \inf \left\{ i\ge 1: \, \in \T_n\right\}, \qquad n\ge 0.$$ By Liggett ([@liggett

_n: \; \pi (x)\ge r^n \} ) \le \deg^n \varrho(r)^n$. By CHebyshev’s iNequaLitY anD tHe BoRel–CAntelli lemma, foR Any $r>\Underline{r}$ and $\P$-almost suRely fOr ALl laRGe $N$, $$\#\left\{ X\in \T_n: \; \pi (X) \Ge R^N \RigHt\} \Le N^2 \deG^n \VArRho(r)^n. \LabEl{Jn-ub1}$$ ON the other hAnd, By (\[Chernoff\]), $$\P \lefT\{ \ExIsts x\in \T_n: \pI (x) \gE r^n\right\} \le \deG^n \vArrho (r)^N.$$ FOr $r> \OVerliNe{r}$, The exPressiON on the Right-hand SiDE is sumMAble in $n$. bY ThE BorEl–Cantelli lemma, foR AnY $R>\overline{r}$ and $\P$-Almost SuRElY FOr aLl lArge $n$, $$\max_{x\iN \T_N} \pi (x) < r^N. \Label{Jn-UB}$$ [*PROOF of tHeorem \[t:posrec\]: Upper bound.*]{} FIX $\vaRepsilOn>0$ SucH That $q+ 3\vArepsIlON < {1\ovEr \deg}$. We follOw thE strategy Given iN liggett ([@LIggett], p. 103) By intrOduCinG a poSItIvE reCuRRenT BiRth-ANd-dEath chaiN $(\wIdEtildE{X_j}, \, j\GE 0)$, STArtiNg fRom $0$, wIth trAnsition probaBilIty fROm $i$ To $i+1$ (foR $i\ge 1$) eQual To $${1\Over \wIdetilDe{\pi} (i)} \, \SuM_{x\in \T_i} \pi(x) (1- \omega(x, {\BuilDrel \leftaRroW \oVer X})),$$ wHere $\wIDetildE{\pi} (I) := \suM_{x\in \T_i} \pI(x)$. We notE ThaT $\wIDETiLde{\pi}$ is a finite invaRiANT mEasure foR $(\widetILdE{X_J})$. let $$\tau_n := \iNf \LefT\{ i\ge 1: \, x_I \In \T_n\rIght\}, \QQuAd n\ge 0.$$ By LIggett ([@LIgGeTt

_n: \; \pi (x)\ge r^n \} ) \le \deg^ n \va rrh o(r )^ n$.By C hebyshev’s ine q uali ty and the Borel–Cante lli l em m a, f o rany $ r>\unde r li n e {r} $an d $ \P $ -a lmost su rely fo r all larg e $ n$ , $$\#\left\ { x \in \T_n:\;\pi (x) \ger^n \righ t\ } \le n^ 2 \de g^n \v a rrho(r )^n. \l a bel{Jn - ub1}$$O ntheother hand, by (\ [ ch e rnoff\]), $$\P \left \{ \e x i sts x\ in \T_n: \ pi (x)\ ge r^n\ r ig h t \ } \ l e \deg^n \var rho (r)^n.$ $ F or $r> \ ove r line{r }$, t he exp ression ontheright-han d side is summ a ble in$n$. B y t heBore l –C an tel li lem m a, fo r an y $r>\ov er li ne{r} $ an d $ \ P$-a lmo st s urely for all larg e $ n$,$ $\m ax_{x \in \ T_n} \ pi (x ) < r^ n. \label{Jn-ub}$$ [* Proof ofThe or em\[ t:pos r ec\]:upp erbound.* ]{} Fix $\v ar e p s il on>0$ such that $q +3 \ va repsilon < {1\ o ve r\ deg}$. We fo llow t he st rate g ygiven in Ligge t t([ @ligget t] , p. 1 03 ) b y i ntrod u cing a pos itive re curre n t birth-and-de a th chain $(\w i de t i ld e {X_j },\, j\ge 0)$ , st a rtin g fr o m$0$ , with tran si t io n probability from $ i$ to $i +1$ ( for $i\ge 1$) equal to$ $ { 1\over \ wide t il d e{\pi} (i)} \, \sum _{x\in \T_ i } \pi(x) (1-\omega(x , {\build r e l \lefta rro w \ ove r x } ) ), $$ where $\w i d etil de {\pi} ( i):= \sum _{x \in \T _i} \ pi(x)$. W e note t ha t$\ wi det ilde{ \ pi}$ isafin it e i nvari a nt mea surefor$( \w i det ilde{X_ j }) $ . Let $ $\ tau_ n : =\inf\lef t \{i\ge 1: \, X_i \ in\ T_n\ ri gh t\}, \q quad n\ge 0.$ $ By Ligget t([@ ligget t

_n: \;_\pi (x)\ge_r^n \} ) \le_\deg^n \varrho(r)^n$._By_Chebyshev’s inequality_and_the Borel–Cantelli lemma,_for any $r>\underline{r}$_and $\P$-almost surely for_all large $n$,_$$\#\left\{_x\in \T_n: \; \pi (x) \ge r^n \right\} \le n^2 \deg^n_\varrho(r)^n. _ __\label{Jn-ub1}$$ On_the other hand, by (\[chernoff\]),_$$\P \left\{ \exists x\in \T_n:_\pi (x)_\ge r^n\right\} \le \deg^n \varrho (r)^n.$$ For $r> \overline{r}$,_the_expression on the_right-hand side is summable in $n$. By the Borel–Cantelli_lemma, for any $r>\overline{r}$ and $\P$-almost_surely for all_large_$n$,_$$\max_{x\in \T_n} \pi (x)_< r^n. \label{Jn-ub}$$ [*Proof_of Theorem \[t:posrec\]: upper bound.*]{} Fix_$\varepsilon>0$ such that $q+ 3\varepsilon < {1\over_\deg}$. We follow the strategy given in_Liggett ([@liggett], p. 103) by introducing_a positive_recurrent birth-and-death chain $(\widetilde{X_j}, \,_j\ge 0)$, starting_from $0$,_with transition probability_from $i$ to $i+1$ (for $i\ge_1$) equal to_$${1\over \widetilde{\pi} (i)} \, \sum_{x\in \T_i}_\pi(x)_(1- \omega(x, {\buildrel_\leftarrow_\over_x})),$$ where $\widetilde{\pi}_(i) := \sum_{x\in_\T_i}_\pi(x)$. We_note_that $\widetilde{\pi}$ is a finite invariant_measure_for $(\widetilde{X_j})$. Let $$\tau_n := \inf \left\{_i\ge 1: \, X_i_\in_\T_n\right\}, \qquad n\ge 0.$$ By_Liggett ([@liggett

^{-2}$,$$\ell \left( s_{\ast },\widehat{s}_{\widehat{m}_{\mathrm{VFCV}}}\right) \leq \left( 1+\frac{L_{\left( \text{\textbf{SA}}\right),r}}{\sqrt{\ln n}}\right) \inf_{m\in \mathcal{M}_{n}}\left\{ \ell \left( s_{\ast },\widehat{s}_{m}^{\left( -1\right) }\right) \right\} +L_{\left( \text{\textbf{SA}}\right),r}\frac{\left( \ln n\right) ^{3}}{n}. $$ The proof of Theorem \[Theorem\_VFCV\] can be found in Section \[ssection\_proof\_VF\].\ In Theorem \[Theorem\_VFCV\], we show an oracle inequality with leading constant converging to one when the amount of data tends to infinity, that compares the excess risk of the model selected *via* VFCV to the excess risk of the best estimator learned with a fraction of the amount of data equal to $1-V^{-1}$. Thus, VFCV allows to asymptotically recover the oracle learned with a fraction of the amount of data equal to $1-V^{-1}$. This is natural since the $V$-fold criterion given in (\[def\_crit\_VFCV\]) is an unbiased estimate of the risk of estimators learned with a fraction $1-V^{-1}$ of the data. Consequently, it seems from Theorem \[Theorem\_VFCV\] that there is some room to improve the performances of VFCV for fixed $V$, since the oracle learned with all the data has better performances (smaller excess risk) than the oracle learned with only part of the initial data. Furthermore, using the concentration inequalities derived in Theorem [theorem\_excess\_risk\_strong\_loc]{}, we roughly have, for any $m\in \mathcal{M}_{n}$,$$\begin{aligned} \mathbb{E}\left[ \ell \left( s_{\ast },\widehat{s}_{m}^{\left( -1\right) }\right) \right] &=&\ell \left( s_{\ast },s_{m}\right) +\mathbb{E}\left[ \ell \left( s_{m},\widehat{s}_{m}^{\left( -1\right) }\right) \right] \\

^{-2}$,$$\ell \left (s_{\ast }, \widehat{s}_{\widehat{m}_{\mathrm{VFCV}}}\right) \leq \left (1+\frac{L_{\left (\text{\textbf{SA}}\right),r}}{\sqrt{\ln n}}\right) \inf_{m\in \mathcal{M}_{n}}\left\ { \ell \left (s_{\ast }, \widehat{s}_{m}^{\left (-1\right) } \right) \right\ } + L_{\left (\text{\textbf{SA}}\right),r}\frac{\left (\ln n\right) ^{3}}{n }. $ $ The proof of Theorem \[Theorem\_VFCV\ ] can be found in Section \[ssection\_proof\_VF\].\ In Theorem \[Theorem\_VFCV\ ], we show an prophet inequality with go constant converging to one when the sum of datum tends to infinity, that compares the excess risk of the exemplar selected * via * VFCV to the excess risk of the good estimator learned with a fraction of the sum of data equal to $ 1 - V^{-1}$. therefore, VFCV allows to asymptotically recover the prophet learned with a fraction of the sum of datum equal to $ 1 - V^{-1}$. This is natural since the $ V$-fold criterion yield in (\[def\_crit\_VFCV\ ]) is an unbiased estimate of the hazard of estimators determine with a fraction $ 1 - V^{-1}$ of the data. Consequently, it seem from Theorem \[Theorem\_VFCV\ ] that there is some room to improve the performances of VFCV for fasten $ V$, since the oracle learned with all the data has better performance (small excess risk) than the oracle learned with only part of the initial data. Furthermore, using the concentration inequality deduce in Theorem [ theorem\_excess\_risk\_strong\_loc ] { }, we roughly receive, for any $ m\in \mathcal{M}_{n}$,$$\begin{aligned } \mathbb{E}\left [ \ell \left (s_{\ast }, \widehat{s}_{m}^{\left (-1\right) } \right) \right ] & = & \ell \left (s_{\ast }, s_{m}\right) + \mathbb{E}\left [ \ell \left (s_{m},\widehat{s}_{m}^{\left (-1\right) } \right) \right ] \\

^{-2}$,$$\ell \left( s_{\ast },\widehat{s}_{\widehxt{m}_{\mathrm{VFCV}}}\riyyt) \leq \neft( 1+\fdac{L_{\left( \text{\textbf{SA}}\right),r}}{\sqrt{\ln n}}\rmght) \unf_{m\ib \mathcal{M}_{n}}\left\{ \ell \lewt( s_{\ast },\wpdehat{s}_{m}^{\lwft( -1\cight) }\right) \righv\} +L_{\left( \bzxt{\tesbbf{SA}}\xijht),r}\frac{\left( \ln n\right) ^{3}}{n}. $$ The proof of Dhdoxem \[Theorem\_VFCV\] can be found in Sectyon \[ssevtlon\_proof\_VF\].\ In Treortm \[Eheodvm\_YFCV\], we show an oracle inequalitg with neading constsnt converging to one when the amount of data tejds to infibity, rhat comparer the excess risk of tge model selected *via* VFCV to tfe exeess risk od rhe test estimavor lewrned with a fractiot of thr amount of daba eqnal ro $1-V^{-1}$. Thus, VFCV allows to asymptotically recover dhz oracle learned with a fracjion mf tfw aoouht or data eqnal to $1-V^{-1}$. Thjs is naturql since the $V$-fold vrynrrion given jn (\[def\_sryt\_VFCV\]) is an unbiased estimate of the rpsk kf estimators learned wuth a fraction $1-V^{-1}$ of tje data. Cogsequently, it seems from Theorem \[Theorem\_VFCV\] that therx ks woid rlom to improve the performances of VFCV for fysec $N$, since the oraclc learned with all tje qata has bettgr performznces (smaller exceds risk) than the oracje lrarned with only part of thw initial dana. Fyrthermore, using tke concentracion imequakities derived in Theorzm [thekrem\_excess\_rlsk\_strong\_muc]{}, we roughly haxe, nor any $m\in \mathcal{M}_{n}$,$$\begin{alidned} \mathbu{E}\lefc[ \ell \lewt( s_{\sst },\wiqehat{s}_{m}^{\lefh( -1\rigmd) }\right) \right] &=&\ell \lfft( s_{\cst },s_{k}\right) +\matjbb{E}\left[ \ell \left( s_{m},\widehat{s}_{m}^{\lefv( -1\right) }\right) \rhghn] \\

^{-2}$,$$\ell \left( s_{\ast },\widehat{s}_{\widehat{m}_{\mathrm{VFCV}}}\right) \leq \left( 1+\frac{L_{\left( \inf_{m\in \ell \left( },\widehat{s}_{m}^{\left( -1\right) }\right) ^{3}}{n}. The proof of \[Theorem\_VFCV\] can be in Section \[ssection\_proof\_VF\].\ In Theorem \[Theorem\_VFCV\], show an oracle inequality with leading constant converging to one when the amount data tends to infinity, that compares the excess risk of the model selected VFCV the risk the best estimator learned with a fraction of the amount of data equal to $1-V^{-1}$. Thus, allows to asymptotically recover the oracle learned with fraction of the amount data equal to $1-V^{-1}$. This natural the $V$-fold given (\[def\_crit\_VFCV\]) an unbiased estimate the risk of estimators learned with a fraction $1-V^{-1}$ of the data. Consequently, it seems from Theorem that there room to the of for fixed $V$, oracle learned with all the data (smaller excess risk) than the oracle learned with part of initial data. Furthermore, using the concentration derived in Theorem [theorem\_excess\_risk\_strong\_loc]{}, we roughly have, for $m\in \mathcal{M}_{n}$,$$\begin{aligned} \mathbb{E}\left[ \ell \left( s_{\ast },\widehat{s}_{m}^{\left( -1\right) }\right) \right] &=&\ell \left( s_{\ast },s_{m}\right) +\mathbb{E}\left[ s_{m},\widehat{s}_{m}^{\left( -1\right) }\right) \right]

^{-2}$,$$\ell \left( s_{\ast },\widehat{s}_{\widehaT{m}_{\mathrm{VFcV}}}\rigHt) \lEq \lEfT( 1+\fraC{L_{\leFt( \text{\textbf{SA}}\RIght),R}}{\sqrt{\ln n}}\right) \inf_{m\in \matHcal{M}_{N}}\lEFt\{ \elL \LeFt( s_{\asT },\widehaT{S}_{m}^{\LEFt( -1\rIgHt) }\RigHt) \RIgHt\} +L_{\leFt( \tExt{\textBf{SA}}\right),r}\FraC{\lEft( \ln n\right) ^{3}}{n}. $$ tHe Proof of TheOreM \[Theorem\_VFCV\] Can Be founD iN SeCTion \[sSecTion\_pRoof\_VF\].\ iN TheorEm \[Theorem\_vFcv\], we shoW An oraclE INeQualIty with leading conSTaNT converging to oNe when ThE AmOUNt oF daTa tends to iNfInity, THat compAReS THE exCEss risk of the mOdel selecteD *Via* vFCV to ThE exCEss risK of thE bESt eStimator leaRned With a fracTion of THe amounT Of data eQual to $1-v^{-1}$. ThUs, VfCV aLLoWs To aSyMPtoTIcAllY RecOver the oRaClE learNed wITH A FracTioN of tHe amoUnt of data equaL to $1-v^{-1}$. ThiS Is nAturaL sincE the $v$-fOld crIterioN giveN iN (\[def\_crit\_VFCV\]) is aN unbIased estiMatE oF thE rIsk of EStimatOrs LeaRned witH a fractIOn $1-V^{-1}$ Of THE DaTa. Consequently, it seEmS FRoM Theorem \[theoreM\_vFcV\] THat there Is SomE rooM TO imprOve tHE pErformanCes of VfcV FoR fixed $V$, SiNce the OrAclE leArned WIth aLl the dAta has beTter pERformances (smalLEr excess risk) tHAn THE oRAcle LeaRned with onlY parT Of thE iniTIaL daTA. FurtHermoRe, USiNG the concentration inEqUalitiEs derIved in Theorem [Theorem\_excESS\_Risk\_stroNg\_loC]{}, We ROughly have, for aNy $m\in \Mathcal{M}_{n}$,$$\bEGin{alignEd} \matHbb{E}\left[ \Ell \left( s_{\aST },\Widehat{s}_{M}^{\leFt( -1\rIghT) }\riGHT) \rIght] &=&\ell \left( s_{\aST },S_{m}\riGhT) +\mathbb{e}\leFt[ \ell \leFt( s_{M},\wiDehAt{s}_{M}^{\lEft( -1\right) }\rIght) \righT] \\

^{-2}$,$$\ell \left( s_{\a st },\wide hat{s }_{ \wi de hat{ m}_{ \mathrm{VFCV}} } \rig ht) \leq \left( 1+\fra c{L_{ \l e ft(\ te xt{\t extbf{S A }} \ r igh t) ,r }}{ \s q rt {\lnn}} \right) \inf_{m\i n \ ma thcal{M}_{n} } \l eft\{ \ell \l eft( s_{\ast }, \wideh at {s} _ {m}^{ \le ft( - 1\righ t ) }\ri ght) \rig ht \ } +L_{ \ left( \ t e xt {\te xtbf{SA}}\right), r }\ f rac{\left( \ln n\rig ht ) ^ { 3 }}{ n}. $$ Thepr oof o f Theore m \ [ T h eor e m\_VFCV\] can be found i n Se ction\[ sse c tion\_ proof \_ V F\] .\ In Theor em \ [Theorem\ _VFCV\ ] , we sh o w an or acle i neq ual ityw it hlea di n g c o ns tan t co nverging t oone w hent h e amou ntof d ata t ends to infin ity , th a t c ompar es th e ex ce ss ri sk ofthe m od el selected *vi a* V FCV to th e e xc ess r isk o f the b est es timator learne d wi th a fr action of the amou nt o fdata equ al to$ 1- V^ { -1}$. Th us , V FCVa l lowsto a s ym ptotical ly rec o ve rthe ora cl e lear ne d w ith a fr a ctio n of t he amoun t ofd ata equal to $ 1 -V^{-1}$. Thi s i s na t ural si nce the $V$ -fol d cri teri o ngiv e n in(\[de f\ _ cr i t\_VFCV\]) is an un bi ased e stima te of the ris k of estim a t o rs learn ed w i th a fraction $1- V^{-1 }$ of thed ata. Co nsequ ently, i t seems f r o m Theore m \ [Th eor em\ _ V FC V\] that ther e is s om e roomtoimprove th e p erf orm an ces of VF CV for f ix ed $ V$ , s incet he oracl elea rn edwitha ll the data has b et t erperform a nc e s (sm al le r ex ces srisk) tha n th e oracl e learned wi t h on ly p art ofthe initial d at a. Further mo re, using t he conce ntration inequalities d e rived i n T heore m [t heorem\_e xce ss\_ri sk\ _ strong \_loc] {}, w erou g h ly ha v e ,for a ny $m\in \ m a thc al{M} _{ n}$, $$\begi n{aligned} \mathbb { E}\ left[ \ell \l eft ( s_ { \ as t } , \w i deh at { s}_ { m }^{\left( -1\ri ght) }\rig ht ) \ right] &=& \ ell \ left( s _{\ast},s_{ m }\right ) +\mathb b{E}\left [\ell \ lef t( s_{m},\ widehat{ s}_{m}^{\ l eft(- 1\ right ) } \right )\ri ght]\\

^{-2}$,$$\ell \left(_s_{\ast },\widehat{s}_{\widehat{m}_{\mathrm{VFCV}}}\right)_\leq \left( 1+\frac{L_{\left( \text{\textbf{SA}}\right),r}}{\sqrt{\ln n}}\right) \inf_{m\in_\mathcal{M}_{n}}\left\{ \ell_\left(_s_{\ast },\widehat{s}_{m}^{\left(_-1\right)_}\right) \right\} +L_{\left(_\text{\textbf{SA}}\right),r}\frac{\left( \ln n\right)_^{3}}{n}. $$ The proof_of Theorem \[Theorem\_VFCV\]_can_be found in Section \[ssection\_proof\_VF\].\ In Theorem \[Theorem\_VFCV\], we show an oracle inequality with leading_constant_converging to_one_when_the amount of data tends_to infinity, that compares the_excess risk_of the model selected *via* VFCV to the_excess_risk of the_best estimator learned with a fraction of the amount_of data equal to $1-V^{-1}$. Thus,_VFCV allows to_asymptotically_recover_the oracle learned with_a fraction of the amount of_data equal to $1-V^{-1}$. This is_natural since the $V$-fold criterion given in_(\[def\_crit\_VFCV\]) is an unbiased estimate of_the risk of estimators learned_with a_fraction $1-V^{-1}$ of the data. Consequently,_it seems from_Theorem \[Theorem\_VFCV\]_that there is_some room to improve the performances_of VFCV for_fixed $V$, since the oracle learned_with_all the data_has_better_performances (smaller_excess risk) than_the_oracle learned_with_only part of the initial data._Furthermore,_using the concentration inequalities derived in Theorem_[theorem\_excess\_risk\_strong\_loc]{}, we roughly have,_for_any $m\in \mathcal{M}_{n}$,$$\begin{aligned} \mathbb{E}\left[ \ell_\left( s_{\ast },\widehat{s}_{m}^{\left( -1\right) }\right) \right]_&=&\ell \left( s_{\ast },s_{m}\right) +\mathbb{E}\left[ \ell \left(_s_{m},\widehat{s}_{m}^{\left( -1\right)_}\right) \right]_\\

focus on the case when the two sources are conditionally independent given the side information available at the encoders and/or the decoder. First, we identify the necessary and sufficient conditions under which separation is optimal when the side information is shared between the encoders and the decoder. Additionally, we show that separation is optimal for sources with reconstruction constraints, when only the decoder has access to the side information conditioned on which the two sources are independent. Next, we consider the case when the decoder is required to recover the common information shared by both encoders losslessly, but can tolerate some distortion for the parts known only at a single encoder. We show that separation is also optimal for this case. We then consider the transmission of sources with a common part in the sense of G[á]{}cs-K[ö]{}rner [@gacs1973common], and investigate the conditions under which separation is optimal in the absence of side information. Next, we provide necessary conditions for the transmission of correlated sources over a multiple access channel. This is achieved by providing a particular side information to the receiver and the transmitters, and using our results for transmitting correlated sources in the presence of side information. In particular, when the two sources are independent conditioned on the side information, our initial results indicate that the necessary and sufficient conditions can be achieved by considering separate source and channel coding. For the special case of transmitting bivariate Gaussian sources over a Gaussian multiple access channel, we provide comparisons of the necessary conditions obtained from our approach with the conditions from [@lapidoth2010sending] and [@7541654]. Our results show that the proposed technique provides the tightest known bound in certain scenarios. [*Related Work:*]{} Shannon proved the optimality of separate source and channel coding for transmitting a source through a noisy channel [@Shannon], known as the separation theorem. Separation was shown to be optimal for the lossy transmission of a source with decoder side information, in [@shamai1998systematic]. The point-to-point scenario was extended in [@gunduz2007correlated] to transmission of correlated sources through a multiple access channel, and separation was shown to be optimal when one of the sources is shared between the two encoders. For the lossless case, the optimality of separation was established in [@gunduz2009source] for transmitting correlated sources through a multiple access channel, whenever the decoder has access to some side information conditioned on which the two sources are independent. A joint source-channel coding scheme

focus on the case when the two sources are conditionally independent give the slope information available at the encoders and/or the decoder. First, we identify the necessary and sufficient condition under which separation is optimal when the english data is shared between the encoders and the decoder. Additionally, we show that separation is optimal for sources with reconstruction constraint, when only the decoder has entree to the side information stipulate on which the two sources are independent. Next, we see the case when the decoder is required to recover the coarse information shared by both encoders losslessly, but can tolerate some distortion for the part known only at a single encoder. We show that separation is besides optimal for this case. We then consider the transmission of sources with a common share in the common sense of G[á]{}cs - K[ö]{}rner [ @gacs1973common ], and investigate the conditions under which separation is optimal in the absence of side information. Next, we provide necessary conditions for the transmittance of correlate sources over a multiple access distribution channel. This is achieved by providing a particular side data to the receiver and the transmitter, and using our results for air correlated sources in the presence of side information. In particular, when the two source are independent conditioned on the side information, our initial results indicate that the necessary and sufficient condition can be achieve by considering separate source and channel coding. For the special case of transmitting bivariate Gaussian source over a Gaussian multiple entree channel, we provide comparisons of the necessary conditions obtained from our approach path with the conditions from [ @lapidoth2010sending ] and [ @7541654 ]. Our results prove that the proposed proficiency provides the tightest known bound in certain scenario. [ * Related Work :* ] { } Shannon proved the optimality of separate reservoir and channel tease for transmitting a source through a noisy channel [ @Shannon ], known as the separation theorem. Separation was shown to be optimal for the lossy transmission of a beginning with decoder side information, in [ @shamai1998systematic ]. The detail - to - point scenario was extended in [ @gunduz2007correlated ] to transmission of correlated sources through a multiple access distribution channel, and separation was shown to be optimal when one of the sources is shared between the two encoders. For the lossless case, the optimality of separation was prove in [ @gunduz2009source ] for convey correlated sources through a multiple access channel, whenever the decoder has access to some slope information conditioned on which the two sources are autonomous. A joint source - distribution channel coding scheme

fofus on the case when the two sources arg xonditmonally indepenaent given the side informatmon qvailqble at the encoders avd/or the fecoder. Dirsu, we identify the necessavv and duffncment conditions under whiwh separation hs o'timal when the side information is fhared nehween the encoqers wnd fhe decoder. Additionally, we show tgat seperation is optikal for sources with recondtruftion constraints, ahen only tye dqxoder has aczess to tht vide inforjation conditioned on which the two xources arg jnfgpendent. Nexv, we cjnsider the gsse whan the cecoder is reqmired to recover the common iiformation shared by both encmdzrs losslessly, but cab rolerdte vome eisgoruioi fkr the pacts known ohly at a sibgle encoder. We shoe eyat separatioh is ajsj optimal for this case. We then consideg ths transmission of sourcws with a common part in the sqnse of G[á]{}cs-K[ö]{}rner [@gacs1973common], and investigate the cmnditmovs bkqde ahich separation is optimal in the absence of aice information. Newt, we provide necexswru conditions fur the trznsmission of corrflated fourcws over a mulyiple access channel. This iw achieved bj prividing a particulcr side infoxmatiom to yhe receiver and the trcnsmitfers, and uslng our rsrults for transmkttpng worrelatta sources in the [resence if snde infofmatoon. In particulag, whek the two sources age inbepengent condihioned on the side information, our initial rexunts indicatz that the necessarr and sufficieut condicions zan be achpeved by ronsidering feparate sourwg and channel coding. Sor rhe wpecial zase of transmotting binaxiate Gauwsian sources over a Ezussian multiplz qccess channel, ee orodife cjkparisons of the nezrssarh condiunonw obtsined from our approdch sith the conditionx nrom [@lapieoth2010sendyng] and [@7541654]. Our tesults show that uhe prmpored trchgique provides the tightest knkwn bound in certain scegariis. [*Related Woxk:*]{} Shannon proved the optimality of sepacate source and channel coding for transmijtikg a source vhrougr a noisy channel [@Shannon], knoqn as the separatlon theorem. Separation was svown ho be optimal for the lossy transmission of a source with decoder side indormatmog, in [@shamai1998aystrmatib]. Tke 'oint-to-[oinv rcenario was extvnded in [@gunduz2007correlated] to traismission mf correlated sources through s oultiple accers channel, and separatioh was shpwn to be optimal when one of tne sources is shared beteeen the two sncodega. Hor the lossless casr, the mptimqlity od seiaration was ertanllshed in [@gkndue2009souxce] for transmitting correlated siurces tjrough a multiplx access chakntl, whengver the decoder has accsss fl somg side informayion conditionrd on which the two votrces are indekendent. A joint source-bhannel codinj scheme

focus on the case when the two conditionally given the information available at First, identify the necessary sufficient conditions under separation is optimal when the side is shared between the encoders and the decoder. Additionally, we show that separation optimal for sources with reconstruction constraints, when only the decoder has access to side conditioned which two sources are independent. Next, we consider the case when the decoder is required to recover common information shared by both encoders losslessly, but tolerate some distortion for parts known only at a encoder. show that is optimal this case. We consider the transmission of sources with a common part in the sense of G[á]{}cs-K[ö]{}rner [@gacs1973common], and investigate conditions under is optimal the of information. Next, we conditions for the transmission of correlated multiple access channel. This is achieved by providing particular side to the receiver and the transmitters, using our results for transmitting correlated sources in presence of side information. In particular, when the two sources are independent conditioned on the our initial results indicate the necessary and conditions be by separate source channel coding. For the special case of transmitting bivariate Gaussian sources a Gaussian multiple access channel, we provide comparisons of the obtained our approach with conditions from [@lapidoth2010sending] and Our show that the proposed the known scenarios. Work:*]{} proved the optimality of source and channel coding for a source through a the separation theorem. Separation was shown to be for the lossy transmission of a source decoder side information, in [@shamai1998systematic]. The point-to-point scenario was extended in [@gunduz2007correlated] transmission of through a multiple access channel, and separation was to be optimal when of the sources is shared between the two encoders. the case, the of separation was in [@gunduz2009source] for correlated sources through access channel, the has side information conditioned on which the sources independent. A joint source-channel coding

focus on the case when the two sOurces are cOnditIonAllY iNdepEndeNt given the side INforMation available at the enCoderS aND/or tHE dEcodeR. First, wE IdENTifY tHe NecEsSArY and sUffIcient cOnditions uNdeR wHich separatiON iS optimal whEn tHe side informAtiOn is shArEd bETween The EncodErs and THe decoDer. AdditiOnALly, we sHOw that sEPArAtioN is optimal for sourCEs WIth reconstructIon conStRAiNTS, whEn oNly the decoDeR has aCCess to tHE sIDE InfORmation conditIoned on whicH The Two souRcEs aRE indepEndenT. NEXt, wE consider thE casE when the dEcoder IS requirED to recoVer the ComMon InfoRMaTiOn sHaREd bY BoTh eNCodErs losslEsSlY, but cAn toLERATe soMe dIstoRtion For the parts knOwn Only AT a sIngle EncodEr. We ShOw thaT separAtion Is Also optimal for tHis cAse. We then ConSiDer ThE tranSMissioN of SouRces witH a commoN ParT iN THE sEnse of G[á]{}cs-K[ö]{}rner [@gaCs1973COMmOn], and invEstigaTE tHe COnditionS uNdeR whiCH SeparAtioN Is Optimal iN the abSEnCe Of side iNfOrmatiOn. nexT, we ProviDE necEssary ConditioNs for THe transmission OF correlated soURcES OvER a muLtiPle access chAnneL. this Is acHIeVed BY provIding A pARtICular side informatioN tO the reCeiveR and the transmItters, and uSING our resuLts fOR tRAnsmitting corrElateD sources in THe presenCe of sIde inforMation. In pARTicular, wHen The Two SouRCEs Are independenT COndiTiOned on tHe sIde infoRmaTioN, ouR inItIal resultS indicatE tHaT tHe NecEssarY And suffiCiEnt CoNdiTions CAn be acHieveD by cOnSiDEriNg separATe SOUrce AnD cHannEl cOdIng. FoR the SPecIal case Of transmiTtiNG bivArIaTe GaussIan sources oveR a gaussian muLtIplE accesS CHannel, we Provide comparisons of the NEcessarY coNditiOns oBtained frOm oUr apprOacH With thE condiTions FrOm [@lAPIdoth2010SENdIng] AnD [@7541654]. Our resultS SHow That tHe PropOsed tecHnique provides the tIGhtEst known bound In cErtaIN ScEnaRIoS. [*relAtED WoRK:*]{} shannon proved thE optimalitY oF SeParate sourCE anD cHannel cOding foR tranSMitting A source thRough a noiSy ChanNEL [@ShAnnon], known As the sepAration thEOrem. SEPaRatioN waS shown To Be oPtimaL for thE LosSy traNsmissIoN of a soUrce wItH decoder Side information, in [@shamai1998SystemAtic]. THe pOint-to-poiNt sCEnaRio was extEndeD in [@gunduz2007cOrrElaTed] to TraNSmissIon oF CoRreLAted sOurcES through a MUlTipLE AcCess channel, AND SepAratiOn wAS shown To be Optimal when one of tHE sources is sharEd beTWEen The TWo enCoDers. For the lossLesS cASE, the optiMaLity of separAtion was EsTAblisHed in [@gUnduz2009sOurce] foR TRaNSmittiNg coRreLated sourCes ThROugh a muLtIpLE accesS chaNnEl, whenEver thE DecoDER has access to some Side iNFOrmatIOn cOnditIoNed on whICh thE two sourceS are indepenDent. A jOint SourcE-channeL cOding sCheMe

focus on the case when th e two sour ces a recon di tion ally independent g i venthe side information a vaila bl e att he enco ders an d /o r the d ec ode r. Fi rst,weidentif y the nece ssa ry and suffici e nt condition s u nder which s epa ration i s o p timal wh en th e side inform ation issh a red be t ween th e en code rs and the decode r .A dditionally, w e show t h at s epa rat ion is opt im al fo r source s w i t h re c onstruction c onstraints, whe n only t hed ecoder hasac c ess to the sid e in formation condi t ioned o n whichthe tw o s our cesa re i nde pe n den t .Nex t , w e consid er t he ca se w h e n thedec oder is r equired to re cov er t h e c ommon info rmat io n sha red by both e ncoders lossles sly, but cantol er ate s ome d i storti onfor the pa rts kno w n o nl y a ta single encoder.We s ho w that s eparat i on i s also op ti mal for t his c ase. We then co nsider th etransmi ss ion of s our ces with a co mmon p art in t he se n se of G[á]{}cs - K[ö]{}rner [@ g ac s 1 97 3 comm on] , and inves tiga t e th e co n di tio n s und er wh ic h s e paration is optimal i n theabsen ce of side in formation. N ext, weprov i de necessary cond ition s for thet ransmiss ion o f correl ated sour c e s over a mu lti ple ac c e ss channel. Thi s is a ch ieved b y p rovidin g a pa rti cul ar side inf ormation t oth erec eiver and thetr ans mi tte rs, a n d usin g our res ul ts for transm i tt i n g co rr el ated so ur ces i n th e pr esenceof side i nfo r mati on .In part icular, whenth e two sour ce s a re ind e p endent c onditioned on the sidei nformat ion , our ini tial resu lts indic ate that t he nec essar yand s uffic i e nt co nd itions can b e a chiev ed byconside ring separate sour c e a nd channel co din g. F o r t hes pe c ial c a seo f transmitting b ivariate G au s si an sources ove ra Gauss ian mul tiple accesschannel,we provid ecomp a r iso ns of thenecessar y conditi o ns ob t ai ned f rom our a pp roa ch wi th the con ditio ns fro m[@lapi doth2 01 0sending ] and [@7541654]. Our r esults show th at the pr opo s edtechnique pro vides thetig hte st kn own bound inc er tai n scen ario s . [*Rela t ed Wo r k :* ]{} Shannon p r ove d the op t imalit y of separate sourcea nd channel cod ingf o r t ran s mitt in g a source thr oug ha noisy ch an nel [@Shann on], kno wn as th e sepa ration theore m . S e parati on w asshown tobeop t imal fo rth e lossy tra ns missio n of a sour c e with decoder si de in f o rmati o n,in [@ sh amai199 8 syst ematic]. T he point-to -point sce nario was ex te nded i n [ @g unduz2007c o rrelated] to t ransmis si on o f c orrela teds o urces thr ou gha multipl e ac c es sc han nel, andse para tion wass hown tobeo ptimalwh eno n e of t h es o urces is s har ed be t w een the tw o enc o de r s. Fo r thelossle ss case , th eoptimal ity o f separat ion was e s tab li shed in [@gu nd uz20 09 sou rc e] for transmit ting cor relateds o urcesthr o ugha multi pl e access cha nnel, w h enevert h e d ecoderh asacc ess tosom e si d e information conditione d on whic h th e two sour ce s are in d epen dent. Aj oint source-c hannel coding s c hem e

focus_on the_case when the two_sources are_conditionally_independent given_the_side information available_at the encoders_and/or the decoder. First,_we identify the_necessary_and sufficient conditions under which separation is optimal when the side information is shared_between_the encoders_and_the_decoder. Additionally, we show that_separation is optimal for sources_with reconstruction_constraints, when only the decoder has access to_the_side information conditioned_on which the two sources are independent. Next, we_consider the case when the decoder_is required to_recover_the_common information shared by_both encoders losslessly, but can tolerate_some distortion for the parts known_only at a single encoder. We show_that separation is also optimal for_this case. We then consider_the transmission_of sources with a common_part in the_sense of_G[á]{}cs-K[ö]{}rner [@gacs1973common], and_investigate the conditions under which separation_is optimal in_the absence of side information. Next, we_provide_necessary conditions for_the_transmission_of correlated_sources over a_multiple_access channel._This_is achieved by providing a particular_side_information to the receiver and the transmitters,_and using our results_for_transmitting correlated sources in_the presence of side information._In particular, when the two sources_are independent_conditioned on_the side information, our initial results indicate that the necessary and_sufficient conditions can be achieved by_considering separate source and_channel coding._For_the special case_of_transmitting bivariate_Gaussian sources over a Gaussian multiple access_channel, we_provide comparisons of the necessary conditions_obtained from our approach_with_the conditions from [@lapidoth2010sending] and [@7541654]._Our results show that the proposed_technique provides the tightest known_bound_in_certain scenarios. [*Related Work:*]{} Shannon proved_the optimality of separate source and_channel coding for_transmitting a source through a noisy channel_[@Shannon],_known as the separation theorem. Separation_was_shown to be optimal for the_lossy_transmission_of a source with decoder_side information, in [@shamai1998systematic]. The point-to-point_scenario was extended in [@gunduz2007correlated] to transmission of correlated_sources through a_multiple access channel, and separation_was_shown_to be optimal when one of the sources is shared_between the_two encoders. For_the lossless case, the optimality of separation was established in_[@gunduz2009source] for transmitting correlated sources through a_multiple access channel, whenever the decoder has access to some side_information conditioned on which the two sources are_independent. A joint source-channel coding scheme

the number of derivatives included into the analysis. As discussed in [@Showk], it implies that the minimum in the plots of Fig. 7 in [@Showk] should be merged with the apparent “kink” at $N \to \infty$. This “kink” is not really sharp at a finite $N$. Nevertheless, its location can be identified with the value of $\Delta(\sigma)$, at which the second derivative of the plot has a local maximum. The minimum of the plot is slightly varied with $N$, whereas the “kink” is barely moving [@Showk]. Apparently, the convergence to a certain asymptotic curve is remarkably faster than $1/N$, as it can be expected from Fig. 7 and other similar figures in [@Showk]. In particular, we have found that the location of the minimum in Fig. 7 of [@Showk] is varied almost linearly with $N^{-3}$. We have shown it in Fig. \[fig2\] by solid circles, the position of the “kink” being indicated by empty circles. The error bars of $\pm 0.000001$ correspond to the symbol size. The results for $N=153, 190, 231$ are presented, skipping the estimate for the location of the “kink” at $N=153$, which cannot be well determined from the corresponding plot in Fig. 7 of [@Showk]. The linear extrapolations (dashed lines) suggest that the minimum, very likely, is moved only slightly closer to the “kink” when $N$ is varied from $N=231$ to $N = \infty$. The linear extrapolation might be too inaccurate. Only in this case a refined numerical analysis for larger $N$ values can possibly confirm the hypothesis about the formation of a sharp kink at $N \to \infty$. The results of both [@Showk] and [@xx] strongly support the commonly accepted 3D Ising values of the critical exponents $\eta$ and $\nu$. In particular, the estimates $\eta=0.03631(3)$ and $\nu = 0.62999(5)$ have been reported in [@Showk]. However, these estimates are obtained, based on certain hypotheses. If these hypotheses are not used, then the conformal bootstrap analysis appears to be consistent even with the discussed here GFD values $\eta=1/8$ and $\nu =2/3$. Indeed, the corresponding

the number of derivatives included into the analysis. As discourse in   [ @Showk ], it entail that the minimum in the plots of Fig.   7 in   [ @Showk ] should be merged with the apparent “ kink ” at $ N \to \infty$. This “ crick ” is not really sharp at a finite $ N$. Nevertheless, its location can be identify with the value of $ \Delta(\sigma)$, at which the second derivative of the plot has a local utmost. The minimum of the plot is slenderly varied with $ N$, whereas the “ kink ” is barely move   [ @Showk ]. Apparently, the convergence to a certain asymptotic curvature is remarkably faster than $ 1 / N$, as it can be expected from Fig.   7 and other similar number in   [ @Showk ]. In particular, we have found that the location of the minimum in Fig.   7 of   [ @Showk ] is varied about linearly with $ N^{-3}$. We have shown it in Fig.   \[fig2\ ] by solid traffic circle, the position of the “ crick ” being indicated by empty lap. The error bars of $ \pm 0.000001 $ correspond to the symbol size. The results for $ N=153, 190, 231 $ are present, skipping the estimate for the location of the “ kink ” at $ N=153 $, which cannot be well determined from the correspond plot in Fig.   7 of   [ @Showk ]. The linear extrapolations (dashed lines) suggest that the minimum, very likely, is moved only slightly closer to the “ crick ” when $ N$ is change from $ N=231 $ to $ N = \infty$. The linear extrapolation might be besides inaccurate. merely in this case a refined numerical psychoanalysis for larger $ N$ value can possibly confirm the hypothesis about the formation of a abrupt kink at $ N \to \infty$. The results of both   [ @Showk ] and   [ @xx ] strongly support the commonly accept 3D Ising values of the critical exponents $ \eta$ and $ \nu$. In particular, the estimates $ \eta=0.03631(3)$ and $ \nu = 0.62999(5)$ have been reported in   [ @Showk ]. However, these estimate are obtained, based on certain hypotheses. If these hypotheses are not practice, then the conformal bootstrap analysis appears to be consistent even with the discussed here GFD values $ \eta=1/8 $ and $ \nu = 2/3$. Indeed, the corresponding

thf number of derivatives lncluded into thg qnalysms. As djscussed in [@Showk], it implies that the munimun in the plots of Fig. 7 kn [@Showk] sjould be merjed with the appedent “kikh” at $K \to \nnhty$. This “kink” ix not realny sharp at a xivice $N$. Nevertheless, its location can bq identofled with the vwlue jf $\Dslta(\sigma)$, at which the second deribative mf the plot hss a local maximum. The minlmum of the plot is sllghtly varigs wyrh $N$, whereas the “kink” pv barely mkving [@Showk]. Apparently, the convefgencz to a certqib adimptotic curte is gemarkably faster that $1/N$, as ot can be expegted hrom Fig. 7 and other similac figures in [@Showk]. In particuldr, we have found thqt the nocadion if ghe mmnijum in Fij. 7 of [@Showk] ia varied alnost linearly with $M^{-3}$. Rv have shown jt in Sid. \[fig2\] by solid circles, the position of tve “iink” being indicated by empty circles. The ertor bars os $\pm 0.000001$ correspond to the symbol size. The results fmr $N=153, 190, 231$ art ivesevref, skipping the estimate for the location of trs “likk” at $N=153$, which caknot be well deterkijec from the cortespondnhg plot in Fig. 7 of [@Shoak]. The jineae extrapojatipns (dashed lines) suggest thqt the minimlm, vwry likely, is moveb only slighcly clpser yo the “kink” when $N$ is vcried rrom $N=231$ to $N = \infty$. Tgd linear extrapouatpon kight be too inaccurate. Onjy in thiw caxe a rewinec numewical analjsis nmr larger $N$ values can kossibny confirm the hypothesis about the formavmon of a shark khnk at $N \to \inftn$. The results of both [@Showk] and [@rx] stronyly suoport the bommonly eccepted 3D Ifing values ox the critican exponegts $\wta$ qnd $\nu$. Kv particular, tne estimanef $\etq=0.03631(3)$ and $\nu = 0.62999(5)$ have becn rekodted in [@Showk]. Hoctvet, these estimatrs xre ohteined, tased on cerdain hyopthesds. If these fypoyheses are not used, dhen the conformal booysbrap analisis appewrs to be conxistent even with uhe divcursed nerg GFD values $\eta=1/8$ and $\nu =2/3$. Indeed, the corrfspjnding

the number of derivatives included into the discussed [@Showk], it that the minimum 7 [@Showk] should be with the apparent at $N \to \infty$. This “kink” not really sharp at a finite $N$. Nevertheless, its location can be identified the value of $\Delta(\sigma)$, at which the second derivative of the plot has local The of plot is slightly varied with $N$, whereas the “kink” is barely moving [@Showk]. Apparently, the convergence a certain asymptotic curve is remarkably faster than as it can be from Fig. 7 and other figures [@Showk]. In we found the location of minimum in Fig. 7 of [@Showk] is varied almost linearly with $N^{-3}$. We have shown it in \[fig2\] by the position the being by empty circles. bars of $\pm 0.000001$ correspond to The results for $N=153, 190, 231$ are presented, the estimate the location of the “kink” at which cannot be well determined from the corresponding in Fig. 7 of [@Showk]. The linear extrapolations (dashed lines) suggest that the minimum, very moved only slightly closer the “kink” when is from to = \infty$. linear extrapolation might be too inaccurate. Only in this case a numerical analysis for larger $N$ values can possibly confirm the the of a sharp at $N \to \infty$. results both [@Showk] and [@xx] the accepted of critical $\eta$ and $\nu$. In the estimates $\eta=0.03631(3)$ and $\nu 0.62999(5)$ have been reported are obtained, based on certain hypotheses. If these are not used, then the conformal bootstrap appears to be consistent even with the discussed here GFD values $\eta=1/8$ $\nu =2/3$. corresponding

the number of derivatives incLuded into tHe anaLysIs. AS dIscuSsed In [@Showk], it impliES thaT the minimum in the plots oF Fig. 7 iN [@SHOwk] sHOuLd be mErged wiTH tHE AppArEnT “kiNk” AT $N \To \infTy$. THis “kink” Is not reallY shArP at a finite $N$. NEVeRtheless, itS loCation can be iDenTified WiTh tHE valuE of $\delta(\Sigma)$, aT Which tHe second dErIVative OF the ploT HAs A locAl maximum. The minimUM oF The plot is slighTly varIeD WiTH $n$, whEreAs the “kink” iS bArely MOving [@ShOWk]. aPPAreNTly, the convergEnce to a certAIn aSymptoTiC cuRVe is reMarkaBlY FasTer than $1/N$, as iT can Be expecteD from FIG. 7 and othER similaR figurEs iN [@ShOwk]. IN PaRtIcuLaR, We hAVe FouND thAt the locAtIoN of thE minIMUM In FiG. 7 of [@showK] is vaRied almost linEarLy wiTH $N^{-3}$. WE have Shown It in fiG. \[fig2\] bY solid CirclEs, The position of thE “kinK” being indIcaTeD by EmPty ciRCles. ThE erRor Bars of $\pM 0.000001$ corresPOnd To THE SyMbol size. The results FoR $n=153, 190, 231$ ArE presentEd, skipPInG tHE estimatE fOr tHe loCATion oF the “KInK” at $N=153$, whicH cannoT Be WeLl deterMiNed froM tHe cOrrEsponDIng pLot in FIg. 7 of [@ShowK]. The lINear extrapolatIOns (dashed lineS) SuGGEsT That The Minimum, very LikeLY, is mOved ONlY slIGhtly CloseR tO ThE “Kink” when $N$ is varied frOm $n=231$ to $N = \inFty$. ThE linear extrapOlation migHT BE too inacCuraTE. ONLy in this case a rEfineD numerical ANalysis fOr larGer $N$ valuEs can possIBLy confirM thE hyPotHesIS AbOut the formatiON Of a sHaRp kink aT $N \tO \infty$. THe rEsuLts Of bOtH [@Showk] and [@Xx] strongLy SuPpOrT thE commONly accepTeD 3D ISiNg vAlues OF the crIticaL expOnEnTS $\etA$ and $\nu$. IN PaRTIculAr, ThE estImaTeS $\eta=0.03631(3)$ aNd $\nu = 0.62999(5)$ HAve Been repOrted in [@ShOwk]. hOwevEr, ThEse estiMates are obtaiNeD, based on ceRtAin HypothESEs. If thesE hypotheses are not used, thEN the conForMal boOtstRap analysIs aPpears To bE ConsisTent evEn witH tHe dISCusseD HErE GFd vAlues $\eta=1/8$ anD $\NU =2/3$. InDeed, tHe CorrEspondiNg

the number of derivatives includedintothe an al ysis . As discussed in[ @Sho wk], it implies that t he mi ni m um i n t he pl ots ofF ig . 7 i n[@ Sho wk ] s hould be merged with theapp ar ent “kink” a t $ N \to \inf ty$ . This “kink ” i s notre all y shar p a t a f inite$ N$. Ne vertheles s, its lo c ation c a n b e id entified with the va l ue of $\Delta( \sigma )$ , a t whi chthe second d eriva t ive oft he p l oth as a local ma ximum. Them ini mum of t hep lot is slig ht l y v aried with$N$, whereasthe “k i nk” isb arely m oving[@S how k].A pp ar ent ly , th e c onv e rge nce to a c er tainasym p t o t ic c urv e is rema rkably faster th an $ 1 /N$ , asit ca n be e xpect ed fro m Fig .7 and other sim ilar figuresin[@ Sho wk ]. In partic ula r,we have foundt hat t h e lo cation of the mini mu m in Fig. 7of [@S h ow k] is varie dalm ostl i nearl y wi t h$N^{-3}$ . We h a ve s hown it i n Fig.  \ [fi g2\ ] bys olid circl es, theposit i on of the “kin k ” being indic a te d by empt y c ircles. The err o r ba rs o f $ \pm 0.000 001$co r re s pond to the symbolsi ze. Th e res ults for $N=1 53, 190, 2 3 1 $ are pre sent e d, skipping the e stima te for the location of t he “kink ” at $N=1 5 3 $, which ca nno t b e w e l ldetermined fr o m the c orrespo ndi ng plot in Fi g.7 o f[@Showk]. The lin ea rex tr apo latio n s (dashe dlin es ) s ugges t thatthe m inim um ,v ery likely , i s move don ly s lig ht ly cl oser tothe “ki nk” when$N$ is v ar ie d from$N=231$ to $N = \infty$.Th e l ineare x trapolat ion might be too inaccu r ate. On lyin th is c ase a ref ine d nume ric a l anal ysis f or la rg er$ N $ val u e scan p ossibly co n f irm thehy poth esis ab out the formationo f a sharp kink a t $ N \t o \i nft y $. Th er esu l t s of both [@Sho wk] and [@ xx ] s trongly su p por tthe com monly a ccept e d 3D Is ing value s of thecr itic a l ex ponents $\ eta$ and $\nu$. I n part i cu lar,the estim at es$\eta =0.036 3 1(3 )$ an d $\nu = 0.629 99(5) $have bee n reported in [@Showk]. Howev er, t hes e estimat esa reobtained, bas ed on cert ain hy pothe ses . If t hese hy pot h esesaren ot used,t he n t h e c onformal bo o t s tra p ana lys i s appe arsto be consistente ven with the d iscu s s edher e GFD v alues $\eta=1/ 8$an d $\nu =2/ 3$ . Indeed, t he corre sp o nding

the_number of_derivatives included into the_analysis. As_discussed_in [@Showk], it_implies_that the minimum_in the plots_of Fig. 7 in [@Showk] should_be merged with_the_apparent “kink” at $N \to \infty$. This “kink” is not really sharp at a_finite_$N$. Nevertheless,_its_location_can be identified with the_value of $\Delta(\sigma)$, at which_the second_derivative of the plot has a local maximum._The_minimum of the_plot is slightly varied with $N$, whereas the “kink”_is barely moving [@Showk]. Apparently, the convergence_to a certain_asymptotic_curve_is remarkably faster than_$1/N$, as it can be expected_from Fig. 7 and other similar figures_in [@Showk]. In particular, we have found that_the location of the minimum in_Fig. 7 of [@Showk] is varied almost_linearly with_$N^{-3}$. We have shown it_in Fig. \[fig2\] by_solid circles,_the position of_the “kink” being indicated by empty_circles. The error_bars of $\pm 0.000001$ correspond to_the_symbol size. The_results_for_$N=153, 190,_231$ are presented,_skipping_the estimate_for_the location of the “kink” at_$N=153$,_which cannot be well determined from the_corresponding plot in Fig. 7_of [@Showk]._The linear extrapolations (dashed_lines) suggest that the minimum,_very likely, is moved only slightly_closer to_the “kink”_when $N$ is varied from $N=231$ to $N = \infty$. The_linear extrapolation might be too inaccurate._Only in this case_a refined_numerical_analysis for larger_$N$_values can_possibly confirm the hypothesis about the formation_of a_sharp kink at $N \to \infty$. The_results of both [@Showk] and [@xx]_strongly_support the commonly accepted 3D Ising_values of the critical exponents $\eta$_and $\nu$. In particular, the_estimates_$\eta=0.03631(3)$_and $\nu = 0.62999(5)$ have_been reported in [@Showk]. However, these estimates_are obtained, based_on certain hypotheses. If these hypotheses are_not_used, then the conformal bootstrap analysis_appears_to be consistent even with the_discussed_here_GFD values $\eta=1/8$ and $\nu_=2/3$. Indeed, the corresponding

r)_n=\prod_{k=1}^n (1-q^{kr}).$$ Thus, Theorem \[thm.CG\] is a special case of Problem \[prob.habiro\]. Equations appear in the [*dilogarithm ladders*]{} of Lewin and others, whose aim is to produce interesting elements of algebraic $K$-theory. For a detailed discussion, see [@Le] and [@Za2]. For the simplest knot $3_1$ and the simplest hyperbolic knot $4_1$, we have: $$\Lnp_{3_1,\SU(2)}(z)=\Lnp_{f_{3_1}}(z), \qquad \Lnp_{4_1,\SU(2)}(z)=\Lnp_{f_{4_1}}(z)$$ (and likewise, equality for the $\Lp$-series), where $$\begin{aligned} f_{3_1}(q) &=& \sum_{n=0}^\infty (q)_n \\ f_{4_1}(q) &=& \sum_{n=0}^\infty (-1)^n q^{-\frac{n(n+1)}{2}} (q)^2_n\end{aligned}$$ are both covered by Theorem \[thm.CG\]. The resurgence conclusion of Theorem \[thm.CG\] is valid for very general entire functions $F$, with some mild hypothesis. For a detailed discussion, see [@ES] and [@CG2]. Evidence ======== Some results ------------ Let us summarize what is known about Conjecture \[conj.1\]. Conjecture \[conj.1\] is known - for all 3-manifolds $M$ of the form $\S \times S^1$ where $\S$ is a closed surface and all compact groups $G$. Indeed, this follows from the fact that $Z_{M,G,n}$ is a polynomial in $n$. Thus, $\Lnp_{M,G}(z)$ is a rational function of $z$ with denominator a power of $1-z$. On the other hand, $\eLambda_{M,G}=\{0,1\}$. - for $\Lp_{M,\SU(2)}$ where $M$ is the [*Poincare homology sphere*]{}, or small Seifert fibered 3-manifolds, see [@CG1]. In this lucky case, one uses explicit formulas for

r)_n=\prod_{k=1}^n (1 - q^{kr}).$$ Thus, Theorem \[thm. CG\ ] is a special case of Problem \[prob.habiro\ ]. Equations look in the [ * dilogarithm ladder * ] { } of Lewin and others, whose bearing is to produce interesting chemical element of algebraic $ K$-theory. For a detailed discussion, witness [ @Le ] and [ @Za2 ]. For the simplest knot $ 3_1 $ and the childlike hyperbolic slub $ 4_1 $, we have: $ $ \Lnp_{3_1,\SU(2)}(z)=\Lnp_{f_{3_1}}(z), \qquad \Lnp_{4_1,\SU(2)}(z)=\Lnp_{f_{4_1}}(z)$$ (and similarly, equality for the $ \Lp$-series), where $ $ \begin{aligned } f_{3_1}(q) & = & \sum_{n=0}^\infty (q)_n \\ f_{4_1}(q) & = & \sum_{n=0}^\infty (-1)^n q^{-\frac{n(n+1)}{2 } } (q)^2_n\end{aligned}$$ are both covered by Theorem \[thm. CG\ ]. The resurgence termination of Theorem \[thm. CG\ ] is valid for very general entire affair $ F$, with some meek hypothesis. For a detailed discussion, see [ @ES ] and [ @CG2 ]. Evidence = = = = = = = = Some results ------------ Let us summarize what is know about Conjecture \[conj.1\ ]. Conjecture \[conj.1\ ] is known - for all 3 - manifolds $ M$ of the shape $ \S \times S^1 $ where $ \S$ is a closed surface and all compact groups $ G$. Indeed, this come from the fact that $ Z_{M, G, n}$ is a polynomial in $ n$. Thus, $ \Lnp_{M, G}(z)$ is a rational function of $ z$ with denominator a power of $ 1 - z$. On the other bridge player, $ \eLambda_{M, G}=\{0,1\}$. - for $ \Lp_{M,\SU(2)}$ where $ M$ is the [ * Poincare homology sphere * ] { }, or small Seifert fibered 3 - manifolds, see [ @CG1 ]. In this lucky case, one uses explicit formulas for

r)_n=\pgod_{k=1}^n (1-q^{kr}).$$ Thus, Theorem \[tmm.CG\] is a special case mf Proglem \[proc.habiro\]. Equations appear in tie [*dulogaeithm ladders*]{} of Lewin and othegs, whose qim ms to produce invsrestinn elejcnts mh algebraic $K$-thgory. For a datailed discusviun, see [@Le] and [@Za2]. For the simplest knot $3_1$ and tne simplest hypetbolib hnot $4_1$, we have: $$\Lnp_{3_1,\SU(2)}(z)=\Lnp_{f_{3_1}}(z), \qquad \Lnp_{4_1,\SU(2)}(z)=\Mnp_{f_{4_1}}(z)$$ (aid likewise, equslity for the $\Lp$-series), whege $$\bfgin{aligned} f_{3_1}(q) &=& \sum_{j=0}^\infty (q)_n \\ f_{4_1}(w) &=& \stn_{n=0}^\infty (-1)^n q^{-\ffac{n(n+1)}{2}} (q)^2_n\end{aligned}$$ are both covered by Theorem \[thm.CG\]. Tfe rexurgence cinxludhon of Theocem \[thi.CG\] is valid for verf genersl entire funcbions $F$, qith some mild hypothxsis. For a detailed qiscussiot, aee [@ES] and [@CG2]. Evidwnxe ======== Somg resglts ------------ Uwt js aukmzrize ahav is known zbout Conjexture \[conj.1\]. Conjectute \[bpnj.1\] is known - for ajl 3-manifolds $M$ of the form $\S \times S^1$ whtre $\S$ is a closed surface ane all compact groups $H$. Indeed, ehis follows from the fact that $Z_{M,G,n}$ is a polynomhal ii $v$. Tkmf, $\Ono_{M,G}(z)$ is a rational function of $z$ with denominwfot s power of $1-z$. Ok the other hand, $\eKalbcw_{M,G}=\{0,1\}$. - for $\Lp_{M,\RU(2)}$ whexs $J$ is the [*Poincare jomologi spheee*]{}, or smajl Srifert fibered 3-manifolds, sew [@CG1]. In this oucky case, one used explicit yormulss fot

r)_n=\prod_{k=1}^n (1-q^{kr}).$$ Thus, Theorem \[thm.CG\] is a of \[prob.habiro\]. Equations in the [*dilogarithm whose is to produce elements of algebraic For a detailed discussion, see [@Le] [@Za2]. For the simplest knot $3_1$ and the simplest hyperbolic knot $4_1$, we $$\Lnp_{3_1,\SU(2)}(z)=\Lnp_{f_{3_1}}(z), \qquad \Lnp_{4_1,\SU(2)}(z)=\Lnp_{f_{4_1}}(z)$$ (and likewise, equality for the $\Lp$-series), where $$\begin{aligned} f_{3_1}(q) &=& (q)_n f_{4_1}(q) \sum_{n=0}^\infty q^{-\frac{n(n+1)}{2}} (q)^2_n\end{aligned}$$ are both covered by Theorem \[thm.CG\]. The resurgence conclusion of Theorem \[thm.CG\] is valid very general entire functions $F$, with some mild For a detailed discussion, [@ES] and [@CG2]. Evidence ======== results Let us what known Conjecture \[conj.1\]. Conjecture is known - for all 3-manifolds $M$ of the form $\S \times S^1$ where $\S$ is a surface and groups $G$. this from fact that $Z_{M,G,n}$ polynomial in $n$. Thus, $\Lnp_{M,G}(z)$ is of $z$ with denominator a power of $1-z$. the other $\eLambda_{M,G}=\{0,1\}$. - for $\Lp_{M,\SU(2)}$ where $M$ the [*Poincare homology sphere*]{}, or small Seifert fibered see [@CG1]. In this lucky case, one uses explicit formulas for

r)_n=\prod_{k=1}^n (1-q^{kr}).$$ Thus, Theorem \[thm.cG\] is a speciAl casE of proBlEm \[prOb.haBiro\]. Equations aPPear In the [*dilogarithm ladderS*]{} of LeWiN And oTHeRs, whoSe aim is TO pRODucE iNtEreStINg ElemeNts Of algebRaic $K$-theorY. FoR a Detailed discUSsIon, see [@Le] anD [@Za2]. for the simpleSt kNot $3_1$ and ThE siMPlest HypErbolIc knot $4_1$, WE have: $$\LNp_{3_1,\SU(2)}(z)=\Lnp_{f_{3_1}}(Z), \qQUad \Lnp_{4_1,\su(2)}(z)=\Lnp_{f_{4_1}}(z)$$ (AND lIkewIse, equality for the $\lP$-sERies), where $$\begin{AligneD} f_{3_1}(Q) &=& \SuM_{N=0}^\InfTy (q)_N \\ f_{4_1}(q) &=& \sum_{n=0}^\infTy (-1)^N q^{-\fraC{N(n+1)}{2}} (q)^2_n\end{ALiGNED}$$ arE Both covered by theorem \[thm.Cg\]. the ResurgEnCe cONclusiOn of THeORem \[Thm.CG\] is valiD for Very generAl entiRE functiONs $F$, with Some miLd hYpoThesIS. FOr A deTaILed DIsCusSIon, See [@ES] and [@cG2]. evIdencE ======== SomE RESUlts ------------ let Us suMmariZe what is known AboUt CoNJecTure \[cOnj.1\]. CoNjecTuRe \[conJ.1\] is knoWn - for AlL 3-manifolds $M$ of thE forM $\S \times S^1$ wHerE $\S$ Is a ClOsed sURface aNd aLl cOmpact gRoups $G$. INDeeD, tHIS FoLlows from the fact thAt $z_{m,g,n}$ Is a polynOmial iN $N$. THuS, $\lnp_{M,G}(z)$ is A rAtiOnal FUNctioN of $z$ WItH denominAtor a pOWeR oF $1-z$. On the OtHer hanD, $\elamBda_{m,G}=\{0,1\}$. - for $\lP_{M,\SU(2)}$ Where $M$ Is the [*PoiNcare HOmology sphere*]{}, oR Small Seifert fIBeRED 3-mANifoLds, See [@CG1]. In this LuckY Case, One uSEs ExpLIcit fOrmulAs FOr

r)_n=\prod_{k=1}^n (1-q^{k r}).$$ Thu s, Th eor em\[ thm. CG\] is a specialc aseof Problem \[prob.habi ro\]. E quat i on s app ear int he [ *di lo ga rit hm la dders *]{ } of Le win and ot her s, whose aim i s t o produceint eresting ele men ts ofal geb r aic $ K$- theor y. For a deta iled disc us s ion, s e e [@Le] a nd [@Z a2]. For the sim p le s t knot $3_1$ a nd the s i mp l e sthyp erbolic kn ot $4_1 $ , we ha v e: $ $ \Ln p _{3_1,\SU(2)} (z)=\Lnp_{f _ {3_ 1}}(z) ,\qq u ad \L np_{4 _1 , \SU (2)}(z)=\Ln p_{f _{4_1}}(z )$$ (a n d likew i se, equ alityfor th e $\ L p$ -s eri es ) , w h er e $ $ \be gin{alig ne d} f_{3 _1}( q ) & =& \ sum _{n= 0}^\i nfty (q)_n \\ f_ {4_1 } (q) &=&\sum_ {n=0 }^ \inft y (-1) ^n q^ {- \frac{n(n+1)}{2 }} ( q)^2_n\en d{a li gne d} $$ ar e bothcov ere d by Th eorem \ [ thm .C G \ ] . The resurgence co nc l u si on of Th eorem\ [t hm . CG\] isva lid for v ery g ener a lentire f unctio n s$F $, with s ome mi ld hy pot hesis . For a det ailed di scuss i on, see [@ES]a nd [@CG2]. E v id e n ce ==== === = Some res ults ---- ---- - -- - L et us summ ar i ze what is known about C onject ure \ [conj.1\]. Co njecture \ [ c o nj.1\] i s kn o wn - for all 3 -mani folds $M$o f the fo rm $\ S \times S^1$ whe r e $\S$ is aclo sed su r f ac e and all com p a ct g ro ups $G$ . I ndeed,thi s f oll ows f rom the f act that $ Z_ {M ,G ,n} $ isa polynom ia l i n$n$ . Thu s , $\Ln p_{M, G}(z )$ i s arationa l f u n ctio nof $z$ wi th deno mina t ora power of $1-z$ . O n the o th er hand , $\eLambda_{ M, G}=\{0,1\} $. - for $ \Lp_{M,\ SU(2)}$ where $M$ is th e [*Poin car e hom olog y sphere* ]{} , or s mal l Seife rt fib ered3- man i f olds, s ee [@ CG 1]. In thi s luc ky ca se , on e usesexplicit formulasf or

r)_n=\prod_{k=1}^n (1-q^{kr}).$$_Thus, Theorem_\[thm.CG\] is a special_case of_Problem_\[prob.habiro\]. Equations appear_in_the [*dilogarithm ladders*]{}_of Lewin and_others, whose aim is_to produce interesting_elements_of algebraic $K$-theory. For a detailed discussion, see [@Le] and [@Za2]. For the simplest knot_$3_1$_and the_simplest_hyperbolic_knot $4_1$, we have: $$\Lnp_{3_1,\SU(2)}(z)=\Lnp_{f_{3_1}}(z),_\qquad \Lnp_{4_1,\SU(2)}(z)=\Lnp_{f_{4_1}}(z)$$ (and likewise, equality_for the_$\Lp$-series), where $$\begin{aligned} f_{3_1}(q) &=& \sum_{n=0}^\infty (q)_n \\ f_{4_1}(q) &=&_\sum_{n=0}^\infty_(-1)^n q^{-\frac{n(n+1)}{2}} (q)^2_n\end{aligned}$$_are both covered by Theorem \[thm.CG\]. The resurgence conclusion of_Theorem \[thm.CG\] is valid for very_general entire functions_$F$,_with_some mild hypothesis. For_a detailed discussion, see [@ES] and_[@CG2]. Evidence ======== Some results ------------ Let us summarize what is_known about Conjecture \[conj.1\]. Conjecture \[conj.1\] is_known - for all 3-manifolds_$M$ of the form $\S_\times S^1$_where $\S$ is a closed_surface and all_compact groups_$G$. Indeed, this_follows from the fact that $Z_{M,G,n}$_is a polynomial_in $n$. Thus, $\Lnp_{M,G}(z)$ is a_rational_function of $z$_with_denominator_a power_of $1-z$. On_the_other hand,_$\eLambda_{M,G}=\{0,1\}$. -_ for $\Lp_{M,\SU(2)}$ where $M$_is_the [*Poincare homology sphere*]{}, or small Seifert_fibered 3-manifolds, see [@CG1]._In_this lucky case, one_uses explicit formulas for

of the origin this is a smooth equivalence relation. That is $Y=\tfrac{1}{1+f_1}X$ reads as [ $$\begin{split} Y= {\varepsilon}{\dfrac{\partial}{\partial u}}+\sum_{j=1}^m{\varepsilon}\tilde g_j{\dfrac{\partial}{\partial v_j}}+\tilde H{\dfrac{\partial}{\partial z}}, \end{split}$$ ]{} where $\tilde g_j$, for $2\geq j\geq k-1$, and $\tilde H$ are smooth with $\tilde H(0)=0$ and ${\tfrac{\partial \tilde H}{\partial z}}(0)<0$. Now we note that the origin of ${\mathbb{R}}^{m+3}$ is a semyhyperbolic equilibrium point with $(u,v,{\varepsilon})$ being center coordinates and $z$ being the hyperbolic coordinate. We can now use Takens-Bonckaert results on normal forms of partially hyperbolic vector fields [@Bonckaert1; @Bonckaert2; @Takens_partially]. Thus, there exists a ${\mathcal{C}^{\ell}}$ change of coordinates (maybe respecting some constraints if required) under which $Y$ is conjugated to [ $$\begin{split} \bar Y={\varepsilon}{\dfrac{\partial}{\partial U}}+\sum_{j=1}^m{\varepsilon}\bar G_j{\dfrac{\partial}{\partial V_j}}+\bar H Z{\dfrac{\partial}{\partial Z}}, \end{split}$$ ]{} where $\bar G_j=\bar G_j(U,V,{\varepsilon})$, for $2\geq j\geq k-1$, and $\bar H=\bar H(U,V,{\varepsilon})$ are ${\mathcal{C}^{\ell}}$ functions, and where $\left\{ Z=0\right\}$ corresponds to a choice center manifold which we denote by ${\mathcal{W}^{^C}}$. We remark that in the vector field $\bar Y$, the functions $\bar G_j$ and $\bar H$ are independent of $Z$. Furthermore we have [ $$\begin{split} \bar H(0,0,0)={\dfrac{\partial \tilde H}{\partial z}}(0,0,0,0)<0. \end{split}$$ ]{} This means that in a small neighborhood of the origin $\bar Y$ can be

of the origin this is a smooth equivalence relation. That is $ Y=\tfrac{1}{1+f_1}X$ read as [ $ $ \begin{split } Y= { \varepsilon}{\dfrac{\partial}{\partial u}}+\sum_{j=1}^m{\varepsilon}\tilde g_j{\dfrac{\partial}{\partial v_j}}+\tilde H{\dfrac{\partial}{\partial z } }, \end{split}$$ ] { } where $ \tilde g_j$, for $ 2\geq j\geq k-1 $, and $ \tilde H$ are legato with $ \tilde H(0)=0 $ and $ { \tfrac{\partial \tilde H}{\partial z}}(0)<0$. Now we note that the beginning of $ { \mathbb{R}}^{m+3}$ is a semyhyperbolic balance point with $ (u, v,{\varepsilon})$ being center field coordinates and $ z$ being the hyperbolic coordinate. We can now practice Takens - Bonckaert result on normal forms of partially hyperbolic vector field [ @Bonckaert1; @Bonckaert2; @Takens_partially ]. Thus, there exist a $ { \mathcal{C}^{\ell}}$ variety of coordinates (maybe respect some constraints if required) under which $ Y$ is conjugate to [ $ $ \begin{split } \bar Y={\varepsilon}{\dfrac{\partial}{\partial U}}+\sum_{j=1}^m{\varepsilon}\bar G_j{\dfrac{\partial}{\partial V_j}}+\bar H Z{\dfrac{\partial}{\partial Z } }, \end{split}$$ ] { } where $ \bar G_j=\bar G_j(U, V,{\varepsilon})$, for $ 2\geq j\geq k-1 $, and $ \bar H=\bar H(U, V,{\varepsilon})$ are $ { \mathcal{C}^{\ell}}$ functions, and where $ \left\ { Z=0\right\}$ corresponds to a option center manifold which we announce by $ { \mathcal{W}^{^C}}$. We remark that in the vector airfield $ \bar Y$, the functions $ \bar G_j$ and $ \bar H$ are autonomous of $ Z$. Furthermore we have [ $ $ \begin{split } \bar H(0,0,0)={\dfrac{\partial \tilde H}{\partial z}}(0,0,0,0)<0. \end{split}$$ ] { } This means that in a small neighborhood of the lineage $ \bar Y$ can be

of the origin this is a smuoth equivalencg eelatimn. Thaf is $Y=\tffac{1}{1+f_1}X$ reads as [ $$\begin{split} T= {\vartisilon}{\dfrac{\partial}{\pargial u}}+\sum_{u=1}^m{\varepsulon}\uilde g_j{\dfrac{\partmzl}{\partial v_j}}+\tjpde K{\dhrac{\partial}{\partlal z}}, \end{vplit}$$ ]{} where $\tindd y_j$, for $2\geq j\geq k-1$, and $\tilde H$ are smjoth wiyh $\tilde H(0)=0$ and ${\tsrac{\kareial \niode H}{\partial z}}(0)<0$. Now we note thzt the mrigin of ${\matnbb{R}}^{m+3}$ is a semyhyperbolic fquipibrium point with $(u,v,{\varepsilin})$ bqung center cuordinates and $z$ being the hyperbolic coordinate. We cxn noc use Takenw-Bincndert resultw on gormal forms of parthally huperbolic vectpr hieles [@Bonckaert1; @Bonckaerv2; @Takens_partially]. Thos, there efiats a ${\mathcal{C}^{\ell}}$ cyange of wooraunages (meybs respfctmng some cohstraints id required) under whocr $Y$ is conjugafed to [ $$\fegin{split} \bar Y={\varepsilon}{\dfrac{\paruial}{\pzrtial U}}+\sum_{j=1}^m{\varepsilon}\var G_j{\dfrac{\partial}{\parjial V_j}}+\bar H Z{\dfrac{\partial}{\partial Z}}, \end{split}$$ ]{} where $\bar G_j=\tar G_o(U,X,{\vaxcisilub})$, vor $2\geq j\geq k-1$, and $\bar H=\bar H(U,V,{\varepsilon})$ are ${\jauhcsl{C}^{\ell}}$ functioks, and where $\left\{ A=0\rlgnj\}$ corresponds jo a choics center manifold ahich wg denore by ${\matrcal{E}^{^C}}$. We remark that in the vextor field $\bcr T$, the functions $\bax G_j$ and $\bar H$ ate indrpendent of $Z$. Furthermoxe we gave [ $$\begin{dplit} \bar H(0,0,0)={\dfrac{\partixl \nilda H}{\partial z}}(0,0,0,0)<0. \end{split}$$ ]{} Thys means vhat nn a smaul ngighborrood of thf orinhn $\bar Y$ can be

of the origin this is a smooth That $Y=\tfrac{1}{1+f_1}X$ reads [ $$\begin{split} Y= z}}, ]{} where $\tilde for $2\geq j\geq and $\tilde H$ are smooth with H(0)=0$ and ${\tfrac{\partial \tilde H}{\partial z}}(0)<0$. Now we note that the origin of is a semyhyperbolic equilibrium point with $(u,v,{\varepsilon})$ being center coordinates and $z$ being hyperbolic We now Takens-Bonckaert results on normal forms of partially hyperbolic vector fields [@Bonckaert1; @Bonckaert2; @Takens_partially]. Thus, there exists ${\mathcal{C}^{\ell}}$ change of coordinates (maybe respecting some constraints required) under which $Y$ conjugated to [ $$\begin{split} \bar U}}+\sum_{j=1}^m{\varepsilon}\bar V_j}}+\bar H Z}}, ]{} $\bar G_j=\bar G_j(U,V,{\varepsilon})$, $2\geq j\geq k-1$, and $\bar H=\bar H(U,V,{\varepsilon})$ are ${\mathcal{C}^{\ell}}$ functions, and where $\left\{ Z=0\right\}$ corresponds to a center manifold denote by We that the vector field the functions $\bar G_j$ and $\bar of $Z$. Furthermore we have [ $$\begin{split} \bar \tilde H}{\partial \end{split}$$ ]{} This means that in small neighborhood of the origin $\bar Y$ can

of the origin this is a smooth eQuivalence RelatIon. thaT iS $Y=\tfRac{1}{1+f_1}x$ reads as [ $$\begin{sPLit} Y= {\Varepsilon}{\dfrac{\partial}{\PartiAl U}}+\Sum_{j=1}^M{\VaRepsiLon}\tildE G_j{\DFRac{\PaRtIal}{\PaRTiAl v_j}}+\tIldE H{\dfrac{\Partial}{\parTiaL z}}, \End{split}$$ ]{} wherE $\TiLde g_j$, for $2\geQ j\gEq k-1$, and $\tilde H$ Are Smooth WiTh $\tILde H(0)=0$ aNd ${\tFrac{\pArtial \TIlde H}{\pArtial z}}(0)<0$. NoW wE Note thAT the oriGIN oF ${\matHbb{R}}^{m+3}$ is a semyhyperBOlIC equilibrium poInt witH $(u,V,{\VaREPsiLon})$ Being centeR cOordiNAtes and $Z$ BeING The HYperbolic coorDinate. We can NOw uSe TakeNs-bonCKaert rEsultS oN NorMal forms of pArtiAlly hyperBolic vECtor fieLDs [@BonckAert1; @BoNckAerT2; @TakENs_PaRtiAlLY]. ThUS, tHerE ExiSts a ${\mathCaL{C}^{\Ell}}$ chAnge OF COOrdiNatEs (maYbe reSpecting some cOnsTraiNTs iF requIred) uNder WhIch $Y$ iS conjuGated To [ $$\Begin{split} \bar Y={\vArepSilon}{\dfraC{\paRtIal}{\PaRtial u}}+\Sum_{j=1}^m{\vArePsiLon}\bar G_J{\dfrac{\pARtiAl}{\PARTiAl V_j}}+\bar H Z{\dfrac{\partIaL}{\PArTial Z}}, \end{Split}$$ ]{} wHErE $\bAR G_j=\bar G_j(u,V,{\VarEpsiLON})$, for $2\gEq j\gEQ k-1$, And $\bar H=\bAr H(U,V,{\vARePsIlon})$ are ${\MaThcal{C}^{\ElL}}$ fuNctIons, aND wheRe $\left\{ z=0\right\}$ coRrespONds to a choice ceNTer manifold whICh WE DeNOte bY ${\maThcal{W}^{^C}}$. We reMark THat iN the VEcTor FIeld $\bAr Y$, thE fUNcTIons $\bar G_j$ and $\bar H$ are InDependEnt of $z$. Furthermore wE have [ $$\begin{SPLIt} \bar H(0,0,0)={\dfRac{\pARtIAl \tilde H}{\partiaL z}}(0,0,0,0)<0. \end{Split}$$ ]{} This mEAns that iN a smaLl neighbOrhood of tHE Origin $\baR Y$ cAn bE

of the origin this is a s mooth equi valen cerel at ion. Tha t is $Y=\tfrac { 1}{1 +f_1}X$ reads as [ $$\ begin {s p lit} Y= {\vare p si l o n}{ \d fr ac{ \p a rt ial}{ \pa rtial u }}+\sum_{j =1} ^m {\varepsilon } \t ilde g_j{\ dfr ac{\partial} {\p artial v _j} } +\til deH{\df rac{\p a rtial} {\partial z } }, \end{sp l i t} $$ ] {} where $\tilde g_ j $, for $2\geqj\geqk- 1 $, a nd$\t ilde H$ ar esmoot h with $ \ ti l d e H( 0 )=0$ and ${\t frac{\parti a l \ tildeH} {\p a rtialz}}(0 )< 0 $.Now we note tha t the ori gin of ${\math b b{R}}^{ m+3}$isa s emyh y pe rb oli ce qui l ib riu m po int with $ (u ,v,{\ vare p s i l on}) $ b eing cent er coordinate s a nd $ z $ b eingthe h yper bo lic c oordin ate.We can now use Ta kens -Bonckaer t r es ult son no r mal fo rms of partia lly hyp e rbo li c v ec tor fields [@Bonck ae r t 1; @Boncka ert2;@ Ta ke n s_partia ll y]. Thu s , ther e ex i st s a ${\m athcal { C} ^{ \ell}}$ c hangeof co ord inate s (ma ybe re specting some constraints if required) und e rw h ic h $Y$ is conjugated to[ $$\ begi n {s pli t } \ ba r Y = {\varepsilon}{\dfra c{ \parti al}{\ partial U}}+\ sum_{j=1}^ m { \ varepsil on}\ b ar G_j{\dfrac{\pa rtial }{\partial V_j}}+\b ar HZ{\dfrac {\partial } { \partial Z} }, \e n d {s plit}$$ ]{} w h ere$\ bar G_j =\b ar G_j( U,V ,{\ var eps il on})$, fo r $2\geq j \g eq k -1$ , and $\bar H= \b arH( U,V ,{\va r epsilo n})$are${ \m a thc al{C}^{ \ el l } }$ f un ct ions , a nd wher e $\ l eft \{ Z=0\ right\}$cor r espo nd sto a ch oice center m an ifold whic hwedenote b y ${\mat hcal{W}^{^C}}$. We rema r k thatinthe v ecto r field $ \ba r Y$,the functi ons $\ bar G _j $ a n d $\ba r H$ ar eindependen t of$Z$.Fu rthe rmore w e have [ $$\begin{ s pli t} \ba r H (0,0 , 0 )= {\d f ra c {\p ar t ial \ tilde H}{\parti al z}}(0,0 ,0 , 0) <0. \e n d{s pl it}$$ ] {} Thi s mea n s thatin a smal l neighbo rh oodo f th e origin $ \bar Y$can be

of_the origin_this is a smooth_equivalence relation._That_is $Y=\tfrac{1}{1+f_1}X$_reads_as [ $$\begin{split} _ _ Y= {\varepsilon}{\dfrac{\partial}{\partial u}}+\sum_{j=1}^m{\varepsilon}\tilde g_j{\dfrac{\partial}{\partial_v_j}}+\tilde H{\dfrac{\partial}{\partial z}}, _ _ \end{split}$$ ]{} where $\tilde g_j$, for $2\geq j\geq k-1$, and $\tilde H$ are smooth_with_$\tilde H(0)=0$_and_${\tfrac{\partial_\tilde H}{\partial z}}(0)<0$. Now we_note that the origin of_${\mathbb{R}}^{m+3}$ is_a semyhyperbolic equilibrium point with $(u,v,{\varepsilon})$ being center_coordinates_and $z$ being_the hyperbolic coordinate. We can now use Takens-Bonckaert results_on normal forms of partially hyperbolic_vector fields [@Bonckaert1;_@Bonckaert2;_@Takens_partially]._Thus, there exists a_${\mathcal{C}^{\ell}}$ change of coordinates (maybe respecting_some constraints if required) under which_$Y$ is conjugated to [ $$\begin{split} _ \bar Y={\varepsilon}{\dfrac{\partial}{\partial_U}}+\sum_{j=1}^m{\varepsilon}\bar G_j{\dfrac{\partial}{\partial V_j}}+\bar H Z{\dfrac{\partial}{\partial_Z}}, _ \end{split}$$ ]{} where $\bar G_j=\bar_G_j(U,V,{\varepsilon})$, for $2\geq_j\geq k-1$,_and $\bar H=\bar_H(U,V,{\varepsilon})$ are ${\mathcal{C}^{\ell}}$ functions, and where_$\left\{ Z=0\right\}$ corresponds_to a choice center manifold which_we_denote by ${\mathcal{W}^{^C}}$._We_remark_that in_the vector field_$\bar_Y$, the_functions_$\bar G_j$ and $\bar H$ are_independent_of $Z$. Furthermore we have [ $$\begin{split} _ __\bar H(0,0,0)={\dfrac{\partial \tilde H}{\partial_z}}(0,0,0,0)<0. \end{split}$$ ]{} This_means that in a small neighborhood_of the_origin $\bar_Y$ can be

real materials, but yields the same physics as the $4f^1$ transition considered here [^2]: The 1-2-20 materials have an fcc structure, but the nature of FH order is independent of the structural details. --- abstract: 'We present a first study of the channel $H, A$ [$\;\rightarrow$]{}$~\tau\tau$ [$\;\rightarrow$]{}$~h^{\pm}+ h^{\mp} + X$ in CMS at high $m_A$ values where no triggering difficulties are expected with QCD jets. At present the $\tau$ selection is based solely on the presence of a hard isolated track in the “$\tau$” jet, but further refinements based on calorimeter collimation or impact parameter selections are obviously possible. The main irreducible background in these conditions is due to QCD jets with hard fragmentations. A large reduction of this background and improvement in the expected signal to background ratio is provided by $E_t^{miss}$ cuts. The expected high-mass reach in the $m_A$, $tan\beta$ parameter space for $3 \times 10^4pb^{-1}$ is shown. This $H$ [$\;\rightarrow$]{}$~\tau\tau$ channel provides the highest mass reach and the best mass resolution when compared to $\tau\tau$ [$\;\rightarrow$]{}$~l^{\pm}+h^{\mp} + X$ and $\tau\tau$ [$\;\rightarrow$]{}$~e^{\pm}+\mu^{\mp} + X$ final states. To the extent that with further calorimetric and impact parameter based selection criteria the QCD background can be kept under control, i.e. below the irreducible $Z,\gamma*$ [$\;\rightarrow$]{}$~\tau\tau$ background, we should strive to have a first level trigger allowing to explore the mass range down to $\sim$150 - 200 GeV.' --- [**July 7, 1999**]{} [**The $H_{SUSY}$ [$\;\rightarrow$]{}$~\tau\tau$ [$\;\rightarrow$]{}$~h^{\pm}+h^{\mp} + X$ channel, its advantages and potential instrumental drawbacks**]{} R. Kinnunen $^{a)}$\ Helsinki Institute of Physics, Helsinki, Finland\ D. Denegri $^{b)}$\ DAPNIA/SPP, CEN Saclay, France\ \

real materials, but yields the same purgative as the $ 4f^1 $ passage considered here [ ^2 ]: The 1 - 2 - 20 materials get an fcc social organization, but the nature of FH order is independent of the structural contingent. --- abstract:' We present a first survey of the channel $ H, A$ [ $ \;\rightarrow$]{}$~\tau\tau$ [ $ \;\rightarrow$]{}$~h^{\pm}+ h^{\mp } + X$ in CMS at gamey $ m_A$ values where no triggering difficulties are have a bun in the oven with QCD jet. At present the $ \tau$ selection is free-base solely on the presence of a heavily isolated track in the “ $ \tau$ ” jet, but further refinement based on calorimeter collimation or impact argument selections are obviously potential. The main irreducible background in these conditions is due to QCD jets with hard fragmentation. A large reduction of this background and improvement in the expected signal to background ratio is provided by $ E_t^{miss}$ cuts. The expected high - mass reach in the $ m_A$, $ tan\beta$ parameter space for $ 3 \times 10 ^ 4pb^{-1}$ is shown. This $ H$ [ $ \;\rightarrow$]{}$~\tau\tau$ channel provides the high mass reach and the dear bulk resolution when compared to $ \tau\tau$ [ $ \;\rightarrow$]{}$~l^{\pm}+h^{\mp } + X$ and $ \tau\tau$ [ $ \;\rightarrow$]{}$~e^{\pm}+\mu^{\mp } + X$ final states. To the extent that with further calorimetric and impact parameter based survival criteria the QCD background can be kept under control, i.e. below the irreducible $ Z,\gamma*$ [ $ \;\rightarrow$]{}$~\tau\tau$ background, we should strive to get a first level trigger allowing to explore the mass scope down to $ \sim$150 - 200 GeV.' --- [ * * July 7, 1999 * * ] { } [ * * The $ H_{SUSY}$ [ $ \;\rightarrow$]{}$~\tau\tau$ [ $ \;\rightarrow$]{}$~h^{\pm}+h^{\mp } + X$ channel, its advantages and potential instrumental drawbacks * * ] { } R.   Kinnunen   $ ^{a)}$\ Helsinki Institute of Physics, Helsinki, Finland\ D.   Denegri   $ ^{b)}$\ DAPNIA / SPP, CEN Saclay, France\ \

rewl materials, but yields uhe same physics cw the $4h^1$ transjtion covsidered here [^2]: The 1-2-20 materials hqve ab fcc structure, but thd nature lf FH oreer ms independent oh the stvbcturzp decamls. --- abstract: 'We present a first study ox ghz channel $H, A$ [$\;\rightarrow$]{}$~\tau\tau$ [$\;\rightwrrow$]{}$~h^{\pk}+ j^{\mp} + X$ in CMS wt hpgr $m_A$ naoues where no triggering diffjcultiev are expectec with QCD jets. At present the $\tau$ selection is hased soleli on rhe presence of a hard isolated trzck in the “$\tau$” jet, but further fefinzments basee in fdlorimeter rollimwtion or impact paraketer srlections are pbvmousoy possible. The main mrreducible backgrougd in theve conditions is duw ro QCG jeds wkrh fars hrafmentahiois. A large deduction od this background amd pkprovement ih the qx[ected signal to background ratio is prmvised by $E_t^{miss}$ cuts. The wxpected high-mass reafh in the $m_A$, $tan\beta$ parameter space for $3 \times 10^4pb^{-1}$ is showt. Thia $H$ [$\;\xlnhtafeoa$]{}$~\tau\tau$ channel provides the highest mass reasg snc the best masf resolutiom ahrg compared to $\tau\tab$ [$\;\rjghtarrow$]{}$~l^{\pm}+h^{\mp} + X$ and $\tao\tau$ [$\;\rughtarrow$]{}$~t^{\pm}+\mu^{\kp} + X$ final states. To the eztent that wpth durther calorimetrnc and impacc paraketer based selection criterna the QCD backgrlund can gd kept under congrok, h.e. below the irreducible $Z,\damma*$ [$\;\rigitarrpw$]{}$~\tau\taj$ bavkgrougd, we shoupd stvhve to have a firsh levgl tricger allowlng to explore the mass range down to $\sim$150 - 200 GgV.' --- [**Jgly 7, 1999**]{} [**The $H_{SBSY}$ [$\;\rinhtarrow$]{}$~\tau\tau$ [$\;\wightarrow$]{}$~h^{\pm}+h^{\kp} + X$ ckannel, its advannages and potential ynstrumental gtawbacks**]{} R. Kinnnnen $^{a)}$\ Helsynki Insritute uw Physics, Helsonki, Finlcud\ D. Denegru $^{b)}$\ DAPNIA/SPP, CEN Saglay, Wdance\ \

real materials, but yields the same physics $4f^1$ considered here The 1-2-20 materials the of FH order independent of the details. --- abstract: 'We present a study of the channel $H, A$ [$\;\rightarrow$]{}$~\tau\tau$ [$\;\rightarrow$]{}$~h^{\pm}+ h^{\mp} + X$ in CMS high $m_A$ values where no triggering difficulties are expected with QCD jets. At the selection based on the presence of a hard isolated track in the “$\tau$” jet, but further refinements based calorimeter collimation or impact parameter selections are obviously The main irreducible background these conditions is due to jets hard fragmentations. large of background and improvement the expected signal to background ratio is provided by $E_t^{miss}$ cuts. The expected high-mass reach in the $tan\beta$ parameter $3 \times is This [$\;\rightarrow$]{}$~\tau\tau$ channel provides mass reach and the best mass to $\tau\tau$ [$\;\rightarrow$]{}$~l^{\pm}+h^{\mp} + X$ and $\tau\tau$ [$\;\rightarrow$]{}$~e^{\pm}+\mu^{\mp} X$ final To the extent that with further and impact parameter based selection criteria the QCD can be kept under control, i.e. below the irreducible $Z,\gamma*$ [$\;\rightarrow$]{}$~\tau\tau$ background, we should strive a first level trigger to explore the range to - GeV.' --- 7, 1999**]{} [**The $H_{SUSY}$ [$\;\rightarrow$]{}$~\tau\tau$ [$\;\rightarrow$]{}$~h^{\pm}+h^{\mp} + X$ channel, its advantages potential instrumental drawbacks**]{} R. Kinnunen $^{a)}$\ Helsinki Institute of Physics, D. $^{b)}$\ DAPNIA/SPP, CEN France\ \

real materials, but yields the Same physicS as thE $4f^1$ tRanSiTion ConsIdered here [^2]: The 1-2-20 mATeriAls have an fcc structure, bUt the NaTUre oF fH Order Is indepENdENT of ThE sTruCtURaL detaIls. --- AbstracT: 'We present A fiRsT study of the cHAnNel $H, A$ [$\;\rightArrOw$]{}$~\tau\tau$ [$\;\righTarRow$]{}$~h^{\pm}+ H^{\mP} + X$ iN cMS at HigH $m_A$ vaLues whERe no trIggering dIfFIcultiES are expECTeD witH QCD jets. At present THe $\TAu$ selection is bAsed soLeLY oN THe pResEnce of a harD iSolatED track iN ThE “$\TAU$” jeT, But further refInements basED on CaloriMeTer COllimaTion oR iMPacT parameter sElecTions are oBviousLY possibLE. The maiN irredUciBle BackGRoUnD in ThESe cONdItiONs iS due to QCd jEtS with Hard FRAGMentAtiOns. A Large Reduction of thIs bAckgROunD and iMprovEmenT iN the eXpecteD signAl To background ratIo is Provided bY $E_t^{MiSs}$ cUtS. The eXPected HigH-maSs reach In the $m_A$, $TAn\bEtA$ PARaMeter space for $3 \times 10^4Pb^{-1}$ IS ShOwn. This $H$ [$\;\RightaRRoW$]{}$~\tAU\tau$ chanNeL prOvidES The hiGhesT MaSs reach aNd the bESt MaSs resolUtIon wheN cOmpAreD to $\taU\Tau$ [$\;\rIghtarRow$]{}$~l^{\pm}+h^{\mP} + X$ and $\TAu\tau$ [$\;\rightarroW$]{}$~E^{\pm}+\mu^{\mp} + X$ final STaTES. TO The eXteNt that with fUrthER calOrimETrIc aND impaCt parAmETeR Based selection criteRiA the QCd backGround can be kePt under conTROL, i.e. below The iRReDUcible $Z,\gamma*$ [$\;\riGhtarRow$]{}$~\tau\tau$ bACkground, We shoUld strivE to have a fIRSt level tRigGer AllOwiNG To Explore the masS RAnge DoWn to $\sim$150 - 200 geV.' --- [**july 7, 1999**]{} [**The $h_{SUsY}$ [$\;\rIghTarRoW$]{}$~\tau\tau$ [$\;\riGhtarrow$]{}$~H^{\pM}+h^{\Mp} + x$ cHanNel, itS AdvantagEs And PoTenTial iNStrumeNtal dRawbAcKs**]{} r. kinNunen $^{a)}$\ HELsINKi InStItUte oF PhYsIcs, HeLsinKI, FiNland\ D. DEnegri $^{b)}$\ DApNIa/sPP, CeN saClay, FraNce\ \

real materials, but yield s the same phys ics as t he $ 4f^1 $ transition c o nsid ered here [^2]: The 1 -2-20 m a teri a ls have an fcc st r u ctu re ,but t h enatur e o f FH or der is ind epe nd ent of the s t ru ctural det ail s. --- abst rac t: 'We p res e nt afir st st udy of the ch annel $H, A $ [$\;\ r ightarr o w $] {}$~ \tau\tau$ [$\;\ri g ht a rrow$]{}$~h^{\ pm}+ h ^{ \ mp } + X $ i n CMS at h ig h $m_ A $ value s w h e r e n o triggering d ifficulties are expec te d w i th QCD jets .A t p resent the$\ta u$ select ion is based s o lely on the p res enc e of aha rdis o lat e dtra c k i n the “$ \t au $” je t, b u t f urth errefi nemen ts based on c alo rime t ercolli matio n or i mpact param eterse lections are ob viou sly possi ble .The m ain i r reduci ble ba ckgroun d in th e seco n d i ti ons is due to QCDje t s w ith hard fragm e nt at i ons. A l ar geredu c t ion o f th i sbackgrou nd and im pr ovement i n theex pec ted sign a l to backg round ra tio i s provided by $ E _t^{miss}$ cu t s. T he expe cte d high-mass rea c h in the $m _A$ , $tan \beta $p ar a meter space for $3\t imes 1 0^4pb ^{-1}$ is sho wn. This $ H $ [$\;\rig htar r ow $ ]{}$~\tau\tau$ chan nel provid e s the hi ghest mass re ach and t h e best ma ssres olu tio n wh en compared t o $\ta u\ tau$ [$ \;\ rightar row $]{ }$~ l^{ \p m}+h^{\mp } + X$ a nd $ \t au \ta u$ [$ \ ;\righta rr ow$ ]{ }$~ e^{\p m }+\mu^ {\mp} + X $fi n alstates. To t he e xt en t th atwi th fu rthe r ca lorimet ric and i mpa c t pa ra me ter bas ed selectioncr iteria the Q CDbackgr o u nd can b e kept under control, i . e. belo w t he ir redu cible $Z, \ga mma*$[$\ ; \right arrow$ ]{}$~ \t au\ t a u$ ba c k gr oun d, we should s tri ve to h avea first level trigger all o win g to explorethe mas s ra nge do w n t o$ \si m $ 150 - 200 GeV.' --- [**J ul y 7 , 1999**]{ } [ ** The $H_ {SUSY}$ [$\; \ rightar row$]{}$~ \tau\tau$ [ $\;\ r i ght arrow$]{}$ ~h^{\pm} +h^{\mp}+ X$ c h an nel,its advan ta ges andpotent i alinstr umenta ldrawba cks** ]{ } R. Ki nnunen $^{a)}$\ Helsink i Inst itute of Physics, He l sin ki, Finla nd\D. Denegri  $^ {b) }$\ D APN I A/SPP , CE N S acl a y, Fr ance \ \

real_materials, but_yields the same physics_as the_$4f^1$_transition considered_here [^2]:_The 1-2-20 materials_have an fcc_structure, but the nature_of FH order_is_independent of the structural details. --- abstract: 'We present a first study of the channel_$H,_A$ [$\;\rightarrow$]{}$~\tau\tau$_[$\;\rightarrow$]{}$~h^{\pm}+_h^{\mp}_+ X$ in CMS at_high $m_A$ values where no_triggering difficulties_are expected with QCD jets. At present the_$\tau$_selection is based_solely on the presence of a hard isolated track_in the “$\tau$” jet, but further_refinements based on_calorimeter_collimation_or impact parameter selections_are obviously possible. The main irreducible_background in these conditions is due_to QCD jets with hard fragmentations. A_large reduction of this background and_improvement in the expected signal_to background_ratio is provided by $E_t^{miss}$_cuts. The expected_high-mass reach_in the $m_A$,_$tan\beta$ parameter space for $3 \times_10^4pb^{-1}$ is shown._This $H$ [$\;\rightarrow$]{}$~\tau\tau$ channel provides the_highest_mass reach and_the_best_mass resolution_when compared to_$\tau\tau$_[$\;\rightarrow$]{}$~l^{\pm}+h^{\mp} +_X$_and $\tau\tau$ [$\;\rightarrow$]{}$~e^{\pm}+\mu^{\mp} + X$ final_states._To the extent that with further calorimetric_and impact parameter based_selection_criteria the QCD background_can be kept under control,_i.e. below the irreducible $Z,\gamma*$ [$\;\rightarrow$]{}$~\tau\tau$_background, we_should strive_to have a first level trigger allowing to explore the mass_range down to $\sim$150 - 200_GeV.' --- [**July 7, 1999**]{} [**The $H_{SUSY}$_[$\;\rightarrow$]{}$~\tau\tau$ [$\;\rightarrow$]{}$~h^{\pm}+h^{\mp}_+_X$ channel, its_advantages_and potential_instrumental drawbacks**]{} R. Kinnunen $^{a)}$\ Helsinki Institute of Physics, Helsinki, Finland\ D. Denegri $^{b)}$\ DAPNIA/SPP,_CEN Saclay,_France\ \

Homeo}}_+({\mathbb{R}})$, endowed with the compact-open topology; let $\mathsf{Rep}_\#(G,{\mathsf{Homeo}}_+({\mathbb{R}}))$ be the subspace of representations with no global fixed points. A representation $\rho \in \mathsf{Rep}_\#(G,{\mathsf{Homeo}}_+({\mathbb{R}}))$ is said to be *flexible* if every open neighborhood of $\rho$ in $ \mathsf{Rep}_\#(G,{\mathsf{Homeo}}_+({\mathbb{R}}))$ contains a representation that is not semi-conjugate to $\rho$. The following lemma is implicit in work of Navas [@NavasFourier] as well as in [@RivasLO]. An explicit proof can be found in [@ABR Prop. 2.8]. \[r:Flexible\] Let $G$ be a discrete, countable group and let $\rho_0$ be the dynamical realization of an order $\preceq$ with basepoint $x\in {\mathbb{R}}$. If $\rho_0$ is flexible, then $\preceq$ is non-isolated in $\operatorname{\mathsf{LO}}(G)$. Though not needed for our work here, we note that a precise characterization of isolated circular and linear orders in terms of a strong form of rigidity (i.e. strong non-flexibility) of their dynamical realizations is given in [@MR]. As mentioned in the introduction, in order to prove Theorem \[LO\_thm\] we use the relationship between circular orders on groups and linear orders on their central extensions by ${\mathbb{Z}}$. For this purpose, we need the notion of [*cofinal elements*]{}. An element $h$ in a linearly-ordered group $(G, \preceq)$ is called *cofinal* if $$\label{eq:cofinal} \text{for all $g\in G$, there exist $m,n\in{\mathbb{Z}}$ such that $h^m \preceq g \preceq h^n$.}$$ \[r:cofinal\] Cofinal elements also have a characterization in terms of the dynamical realization: If $\rho$ is a dynamical realization of $\preceq$ with basepoint $x$, then $h \in G$ is cofinal if and only if $\rho(h)$ has no fixed point. Indeed

Homeo}}_+({\mathbb{R}})$, endowed with the compact - open topology; lease $ \mathsf{Rep}_\#(G,{\mathsf{Homeo}}_+({\mathbb{R}}))$ be the subspace of representation with no global specify point. A representation $ \rho \in \mathsf{Rep}_\#(G,{\mathsf{Homeo}}_+({\mathbb{R}}))$ is said to be * compromising * if every receptive neighborhood of $ \rho$ in $ \mathsf{Rep}_\#(G,{\mathsf{Homeo}}_+({\mathbb{R}}))$ incorporate a representation that is not semi - conjugate to $ \rho$. The following lemma is implicit in oeuvre of Navas [ @NavasFourier ] as well as in [ @RivasLO ]. An explicit validation can be found in [ @ABR Prop.   2.8 ]. \[r: Flexible\ ] Let $ G$ be a discrete, countable group and let $ \rho_0 $ be the dynamical realization of an order   $ \preceq$ with basepoint $ x\in { \mathbb{R}}$. If $ \rho_0 $ is flexible, then $ \preceq$ is non - isolated in $ \operatorname{\mathsf{LO}}(G)$. Though not necessitate for our work here, we note that a precise portrayal of isolated circular and linear orders in footing of a strong form of rigidity (i.e.   strong non - flexibility) of their dynamic realizations is given in [ @MR ]. As mentioned in the introduction, in order to prove Theorem   \[LO\_thm\ ] we practice the kinship between round orders on groups and linear orders on their central extensions by $ { \mathbb{Z}}$. For this purpose, we necessitate the impression of [ * cofinal component * ] { }. An element $ h$ in a linearly - arrange group $ (G, \preceq)$ is call * cofinal * if $ $ \label{eq: cofinal } \text{for all $ g\in G$, there exist $ m, n\in{\mathbb{Z}}$ such that $ h^m \preceq g \preceq h^n$.}$$ \[r: cofinal\ ] Cofinal elements also accept a characterization in terms of the dynamical realization: If $ \rho$ is a dynamical realization of $ \preceq$ with basepoint $ x$, then $ planck's constant \in G$ is cofinal if and only if $ \rho(h)$ has no fixed point. Indeed

Homfo}}_+({\mathbb{R}})$, endowed with tme compact-open topology; let $\mzthsf{Rep}_\#(E,{\mathsf{Homeo}}_+({\mathbb{R}}))$ be the snbspqce od representations with no globap fixed poinus. A representatioi $\rho \in \mathsr{Vep}_\#(G,{\mctisf{Homeo}}_+({\mathbb{R}}))$ is said tm be *flexible* hf erery open neighborhood of $\rho$ in $ \maehsf{Rep}_\#(B,{\mwthsf{Homeo}}_+({\mathfb{R}}))$ bogtaihs a representation that is not seji-conjujate to $\rho$. The gollowing lemma is implicih in work of Navas [@NavwsFourier] aw wejo as in [@RivarLO]. An explicit proof dan be found in [@ABR Prop. 2.8]. \[r:Flexibue\] Lec $G$ be a diwceetf, countable jroup wnd let $\rho_0$ nv the dfnamicak realization pf en oeder $\preceq$ with basepmint $x\in {\mathbb{R}}$. Is $\rho_0$ is xlzxible, then $\preceq$ is nin-isonateg in $\ipefatkriams{\mathsv{LO}}(J)$. Though not needed for our work here, we nptq that a preciae chawasterization of isolated circular and litead orders in terms of a wtrong form of rigidijy (i.e. strond non-flexibility) of their dynamical realizations hs gitev iu [@MR]. As nejtioned in the introduction, in order to prove Fhtorvm \[LO\_thm\] we use thc relationship beteefn sircular ordets on gxkuls and linear ordegs on treir xentral evtenxions by ${\mathbb{Z}}$. For this pyrpose, we nevd tye notion of [*cofincl elements*]{}. Au elemgnt $h$ on a linearly-ordered gruup $(F, \preceq)$ is called *ckwinal* if $$\label{eq:zofpnal} \dext{for all $g\in G$, there exift $m,n\in{\mavhbb{Z}}$ such tfat $n^m \preseq g \precfq h^n$.}$$ \[v:wofinal\] Cofinal elfmentd dlso have w characterization in terms of vie dynamical teanizdtion: If $\rho$ ls a dynamical wealization of $\preceq$ with basepoint $x$, then $h \in G$ is sofinal if ang only if $\rho(i)$ has no sixee pount. Indddd

Homeo}}_+({\mathbb{R}})$, endowed with the compact-open topology; let the of representations no global fixed \mathsf{Rep}_\#(G,{\mathsf{Homeo}}_+({\mathbb{R}}))$ said to be if every open of $\rho$ in $ \mathsf{Rep}_\#(G,{\mathsf{Homeo}}_+({\mathbb{R}}))$ contains representation that is not semi-conjugate to $\rho$. The following lemma is implicit in of Navas [@NavasFourier] as well as in [@RivasLO]. An explicit proof can be in Prop. \[r:Flexible\] $G$ be a discrete, countable group and let $\rho_0$ be the dynamical realization of an order with basepoint $x\in {\mathbb{R}}$. If $\rho_0$ is flexible, $\preceq$ is non-isolated in Though not needed for our here, note that precise of circular and linear in terms of a strong form of rigidity (i.e. strong non-flexibility) of their dynamical realizations is given [@MR]. As the introduction, order prove \[LO\_thm\] we use between circular orders on groups and their central extensions by ${\mathbb{Z}}$. For this purpose, need the of [*cofinal elements*]{}. An element $h$ a linearly-ordered group $(G, \preceq)$ is called *cofinal* $$\label{eq:cofinal} \text{for all $g\in G$, there exist $m,n\in{\mathbb{Z}}$ such that $h^m \preceq g \preceq h^n$.}$$ elements also have a in terms of dynamical If is dynamical realization $\preceq$ with basepoint $x$, then $h \in G$ is cofinal if only if $\rho(h)$ has no fixed point. Indeed

Homeo}}_+({\mathbb{R}})$, endowed with thE compact-opEn topOloGy; lEt $\MathSf{ReP}_\#(G,{\mathsf{Homeo}}_+({\mAThbb{r}}))$ be the subspace of represEntatIoNS witH No GlobaL fixed pOInTS. a rePrEsEntAtIOn $\Rho \in \MatHsf{Rep}_\#(G,{\Mathsf{HomeO}}_+({\maThBb{R}}))$ is said to bE *FlExible* if evEry Open neighborHooD of $\rho$ In $ \MatHSf{Rep}_\#(g,{\maThsf{HOmeo}}_+({\maTHbb{R}}))$ coNtains a rePrESentatIOn that iS NOt Semi-Conjugate to $\rho$. The FOlLOwing lemma is imPlicit In WOrK OF NaVas [@navasFouriEr] As welL As in [@RivASLo]. aN ExpLIcit proof can bE found in [@ABR pRop. 2.8]. \[R:FlexiBlE\] LeT $g$ be a diScretE, cOUntAble group anD let $\Rho_0$ be the dYnamicAL realizATion of aN order $\PreCeq$ With BAsEpOinT $x\IN {\maTHbB{R}}$. IF $\Rho_0$ Is flexibLe, ThEn $\preCeq$ iS NON-IsolAteD in $\oPeratOrname{\mathsf{Lo}}(G)$. THougH Not NeedeD for oUr woRk Here, wE note tHat a pReCise characterizAtioN of isolatEd cIrCulAr And liNEar ordErs In tErms of a Strong fORm oF rIGIDiTy (i.e. strong non-flexiBiLITy) Of their dYnamicAL rEaLIzations Is GivEn in [@mr]. as menTionED iN the intrOductiON, iN oRder to pRoVe TheoReM \[LO\_Thm\] We use THe reLationShip betwEen ciRCular orders on gROups and linear ORdERS oN TheiR ceNtral extensIons BY ${\matHbb{Z}}$. fOr ThiS PurpoSe, we nEeD ThE Notion of [*cofinal elemEnTs*]{}. An elEment $H$ in a linearly-oRdered grouP $(g, \PReceq)$ is cAlleD *CoFInal* if $$\label{eq:cOfinaL} \text{for alL $G\in G$, therE exisT $m,n\in{\matHbb{Z}}$ such tHAT $h^m \preceQ g \pRecEq h^N$.}$$ \[r:cOFInAl\] Cofinal elemENTs alSo Have a chAraCterizaTioN in TerMs oF tHe dynamicAl realizAtIoN: IF $\rHo$ iS a dynAMical reaLiZatIoN of $\PreceQ$ With baSepoiNt $x$, tHeN $h \IN G$ iS cofinaL If AND onlY iF $\rHo(h)$ hAs nO fIxed pOint. iNdeEd

Homeo}}_+({\mathbb{R}})$,endowed wi th th e c omp ac t-op en t opology; let $ \ math sf{Rep}_\#(G,{\mathsf{ Homeo }} _ +({\ m at hbb{R }}))$ b e t h e su bs pa ceof re prese nta tions w ith no glo bal f ixed points. A represent ati on $\rho \in \m athsf{ Re p}_ \ #(G,{ \ma thsf{ Homeo} } _+({\m athbb{R}} )) $ is sa i d to be * fl exib le* if every open ne i ghborhood of $ \rho$in $\ m ath sf{ Rep}_\#(G, {\ maths f {Homeo} } _+ ( { \ mat h bb{R}}))$ con tains a rep r ese ntatio ntha t is no t sem i- c onj ugate to $\ rho$ . The fo llowin g lemmai s impli cit in wo rkof N a va s[@N av a sFo u ri er] aswell asin [ @Riva sLO] . A n exp lic it p roofcan be foundin[@AB R Pr op. 2 .8]. \[r :F lexib le\] L et $G $be a discrete,coun table gro upan d l et $\rh o _0$ be th e d ynamica l reali z ati on o f a n order $\preceq$wi t h b asepoint $x\in {\ ma t hbb{R}}$ .If$\rh o _ 0$ is fle x ib le, then $\pre c eq $is non- is olated i n $ \op erato r name {\math sf{LO}}( G)$.Though not nee d ed for our wo r kh e re , wenot e that a pr ecis e cha ract e ri zat i on of isol at e dc ircular and linearor ders i n ter ms of a stron g form ofr i g idity (i .e.s tr o ng non-flexibi lity) of theird ynamical real izations is given i n [@MR]. A s m ent ion e d i n the introdu c t ion, i n order to proveThe ore m \ [LO \_ thm\] weuse there la ti on shi p bet w een circ ul aror der s ong roupsand l inea ror d ers on the i rc e ntra lex tens ion sby ${ \mat h bb{ Z}}$. F or this p urp o se,we n eed the notion of [* co final elem en ts* ]{}. A n element $h$ in a linearly-orde r ed grou p $ (G, \ prec eq)$ is c all ed *co fin a l* if$$\lab el{eq :c ofi n a l} \t e x t{ for a ll $g\in G $ , th ere e xi st $ m,n\in{ \mathbb{Z}}$ sucht hat $h^m \preceq g\pre c e qh^n $ .} $ $ \[ r :co f i nal\] Cofinal e lements al so ha ve a chara c ter iz ation i n terms of t h e dynam ical real ization:If $\r h o $ i s a dynami cal real ization o f $\pr e ce q$ wi thbasepo in t $ x$, t hen $h \in G$ i s cofi na l if a nd on ly if $\rh o(h)$ has no fixed poin t. Ind eed

Homeo}}_+({\mathbb{R}})$, endowed_with the_compact-open topology; let $\mathsf{Rep}_\#(G,{\mathsf{Homeo}}_+({\mathbb{R}}))$_be the_subspace_of representations_with_no global fixed_points. A representation $\rho_\in \mathsf{Rep}_\#(G,{\mathsf{Homeo}}_+({\mathbb{R}}))$ is said_to be *flexible*_if_every open neighborhood of $\rho$ in $ \mathsf{Rep}_\#(G,{\mathsf{Homeo}}_+({\mathbb{R}}))$ contains a representation that is not_semi-conjugate_to $\rho$. The_following_lemma_is implicit in work of_Navas [@NavasFourier] as well as_in [@RivasLO]._An explicit proof can be found in [@ABR_Prop. 2.8]. \[r:Flexible\]_Let $G$ be_a discrete, countable group and let $\rho_0$ be the_dynamical realization of an order $\preceq$ with_basepoint $x\in {\mathbb{R}}$._If_$\rho_0$_is flexible, then $\preceq$_is non-isolated in $\operatorname{\mathsf{LO}}(G)$. Though not needed_for our work here, we note_that a precise characterization of isolated circular_and linear orders in terms of_a strong form of rigidity_(i.e. strong non-flexibility)_of their dynamical realizations is_given in [@MR]. As_mentioned in_the introduction, in_order to prove Theorem \[LO\_thm\] we use_the relationship between_circular orders on groups and linear_orders_on their central_extensions_by_${\mathbb{Z}}$. For_this purpose, we_need_the notion_of_[*cofinal elements*]{}. An element $h$ in a_linearly-ordered_group $(G, \preceq)$ is called *cofinal* if_$$\label{eq:cofinal} \text{for all $g\in G$, there_exist_$m,n\in{\mathbb{Z}}$ such that $h^m_\preceq g \preceq h^n$.}$$ \[r:cofinal\] Cofinal_elements also have a characterization in_terms of_the dynamical_realization: If $\rho$ is a dynamical realization of $\preceq$ with basepoint_$x$, then $h \in G$ is_cofinal if and only_if $\rho(h)$_has_no fixed point._Indeed

j-1}, - z_{j},z_{j+1},\ldots,z_{2s-1}), \; j=1,...,2s-1.$$ Inside the group $H_{0}$, the only non-trivial elements acting with fixed points are $b_{1},\ldots, b_{2s-1}$ and $b_{2s}=b_{1}b_{2}\cdots b_{2s-1}$. The degree $2^{2s-1}$ holomorphic map $$P:D \to \widehat{\mathbb C}: (z_{1},\ldots,z_{2s-1}) \mapsto-\left( \dfrac{z_{2}}{z_{1}} \right)^{2}$$ is a branched regular cover with deck group being $H_{0}$. The projection under $P$ of the set of fixed points are as follows: $$P({\rm Fix}(b_{1}))=\infty, \quad P({\rm Fix}(b_{2}))=0, \quad P({\rm Fix}(b_{3}))=1,\quad P({\rm Fix}(b_{4}))=\lambda_{1},$$ $$P({\rm Fix}(b_{2k+3}))=\mu_{k,1}, \quad P({\rm Fix}(b_{2k+4}))=\mu_{k,2}, \quad k=1,\ldots, s.$$ In particular, the branch locus of $P$ is the set $$\{\infty,0,1,\lambda, \mu_{1,1}, \mu_{1,2},\ldots,\mu_{s-2,1}, \mu_{s-2,2}\}.$$ By the Riemann-Hurwitz formula, $D$ has genus $g_{D}=1+2^{2s-2}(s-2)$. Let us consider the surjective homomorphism $$\theta:H_{0} \to H=\langle a_{1},\ldots,a_{s-1}\rangle \cong {\mathbb Z}_{2}^{s-1}$$ $$\begin{array}{c} b_{1},b_{2} \mapsto a_{1}\\ b_{3},b_{4} \mapsto a_{2}\\ b_{5},b_{6} \mapsto a_{3}\\ \vdots\\ b_{2k-1}, b_{2k} \mapsto a_{k}\\ \vdots\\ b_{2s

j-1 }, - z_{j},z_{j+1},\ldots, z_{2s-1 }), \; j=1,... ,2s-1.$$ Inside the group $ H_{0}$, the only non - trivial elements acting with specify point are $ b_{1},\ldots, b_{2s-1}$ and $ b_{2s}=b_{1}b_{2}\cdots b_{2s-1}$. The degree $ 2^{2s-1}$ holomorphic map $ $ P :D \to \widehat{\mathbb speed of light }: (z_{1},\ldots, z_{2s-1 }) \mapsto-\left (\dfrac{z_{2}}{z_{1 } } \right)^{2}$$ is a branched regular covering with pack of cards group being $ H_{0}$. The projection under $ P$ of the set of fixed points are as follows: $ $ P({\rm Fix}(b_{1}))=\infty, \quad P({\rm Fix}(b_{2}))=0, \quad P({\rm Fix}(b_{3}))=1,\quad P({\rm Fix}(b_{4}))=\lambda_{1},$$ $ $ P({\rm Fix}(b_{2k+3}))=\mu_{k,1 }, \quad P({\rm Fix}(b_{2k+4}))=\mu_{k,2 }, \quad k=1,\ldots, s.$$ In finical, the branch locus of $ P$ is the hardened $ $ \{\infty,0,1,\lambda, \mu_{1,1 }, \mu_{1,2},\ldots,\mu_{s-2,1 }, \mu_{s-2,2}\}.$$ By the Riemann - Hurwitz rule, $ D$ has genus $ g_{D}=1 + 2^{2s-2}(s-2)$. Let us consider the surjective homomorphism $ $ \theta: H_{0 } \to H=\langle a_{1},\ldots, a_{s-1}\rangle \cong { \mathbb Z}_{2}^{s-1}$$ $ $ \begin{array}{c } b_{1},b_{2 } \mapsto a_{1}\\ b_{3},b_{4 } \mapsto a_{2}\\ b_{5},b_{6 } \mapsto a_{3}\\ \vdots\\ b_{2k-1 }, b_{2k } \mapsto a_{k}\\ \vdots\\ b_{2s

j-1}, - x_{j},z_{j+1},\ldots,z_{2s-1}), \; j=1,...,2s-1.$$ Inside tht group $H_{0}$, the onli bon-tritial elsments azting with fixed points are $u_{1},\ldors, b_{2s-1}$ and $b_{2s}=b_{1}b_{2}\cdots b_{2s-1}$. The aegree $2^{2s-1}$ jolomorpyic nqp $$P:D \to \wmsehat{\mabkbb C}: (d_{1},\ldotv,v_{2s-1}) \mapsto-\left( \dnrac{z_{2}}{z_{1}} \righd)^{2}$$ is a brancheg feyular cover with deck group being $H_{0}$. Ehe prokeftion under $P$ jf tnq sef of fixed points are as follows: $$P({\dm Fix}(b_{1}))=\pnfty, \quad P({\rm Fic}(b_{2}))=0, \quad P({\rm Fix}(b_{3}))=1,\quad P({\rm Flx}(b_{4}))=\lwmbda_{1},$$ $$P({\rm Fix}(b_{2k+3}))=\mu_{k,1}, \quad P({\rm Fix}(f_{2j+4}))=\mu_{k,2}, \quad k=1,\udots, s.$$ In kaxticular, thg branch locus of $P$ is the set $$\{\ivfty,0,1,\lcmbda, \mu_{1,1}, \mu_{1,2},\odits,\lo_{s-2,1}, \mu_{s-2,2}\}.$$ By the Riemwnn-Hurwitz formula, $D$ has gemus $g_{D}=1+2^{2s-2}(s-2)$. Let us coisidwr the surjective hommmorphism $$\theta:H_{0} \tj H=\langle a_{1},\mdots,a_{s-1}\rangle \cont {\nathbt Z}_{2}^{s-1}$$ $$\begkb{arfay}{d} b_{1},u_{2} \mzpsto w_{1}\\ b_{3},b_{4} \mapsto a_{2}\\ b_{5},g_{6} \mapsto a_{3}\\ \veots\\ b_{2k-1}, b_{2k} \mapsto a_{k}\\ \fdjnx\\ b_{2s

j-1}, - z_{j},z_{j+1},\ldots,z_{2s-1}), \; j=1,...,2s-1.$$ Inside the the non-trivial elements with fixed points b_{2s-1}$. degree $2^{2s-1}$ holomorphic $$P:D \to \widehat{\mathbb (z_{1},\ldots,z_{2s-1}) \mapsto-\left( \dfrac{z_{2}}{z_{1}} \right)^{2}$$ is a regular cover with deck group being $H_{0}$. The projection under $P$ of the of fixed points are as follows: $$P({\rm Fix}(b_{1}))=\infty, \quad P({\rm Fix}(b_{2}))=0, \quad P({\rm P({\rm $$P({\rm \quad Fix}(b_{2k+4}))=\mu_{k,2}, \quad k=1,\ldots, s.$$ In particular, the branch locus of $P$ is the set $$\{\infty,0,1,\lambda, \mu_{1,1}, \mu_{s-2,2}\}.$$ By the Riemann-Hurwitz formula, $D$ has genus Let us consider the homomorphism $$\theta:H_{0} \to H=\langle a_{1},\ldots,a_{s-1}\rangle {\mathbb $$\begin{array}{c} b_{1},b_{2} a_{1}\\ \mapsto b_{5},b_{6} \mapsto a_{3}\\ b_{2k-1}, b_{2k} \mapsto a_{k}\\ \vdots\\ b_{2s

j-1}, - z_{j},z_{j+1},\ldots,z_{2s-1}), \; j=1,...,2s-1.$$ Inside the grOup $H_{0}$, the onlY non-tRivIal ElEmenTs acTing with fixed pOInts Are $b_{1},\ldots, b_{2s-1}$ and $b_{2s}=b_{1}b_{2}\cdotS b_{2s-1}$. ThE dEGree $2^{2S-1}$ HoLomorPhic map $$p:d \tO \WIdeHaT{\mAthBb c}: (Z_{1},\lDots,z_{2S-1}) \maPsto-\lefT( \dfrac{z_{2}}{z_{1}} \riGht)^{2}$$ Is A branched regULaR cover with DecK group being $H_{0}$. the ProjecTiOn uNDer $P$ oF thE set oF fixed POints aRe as folloWs: $$p({\Rm Fix}(b_{1}))=\INfty, \quaD p({\Rm fix}(b_{2}))=0, \Quad P({\rm Fix}(b_{3}))=1,\quad P({\rM fiX}(B_{4}))=\lambda_{1},$$ $$P({\rm Fix}(b_{2K+3}))=\mu_{k,1}, \quAd p({\Rm fIX}(b_{2k+4}))=\Mu_{k,2}, \Quad k=1,\ldots, S.$$ IN partICular, thE BrANCH loCUs of $P$ is the set $$\{\Infty,0,1,\lambda, \MU_{1,1}, \mu_{1,2},\Ldots,\mU_{s-2,1}, \Mu_{s-2,2}\}.$$ bY the RiEmann-huRWitZ formula, $D$ haS genUs $g_{D}=1+2^{2s-2}(s-2)$. Let Us consIDer the sURjectivE homomOrpHisM $$\theTA:H_{0} \To h=\laNgLE a_{1},\lDOtS,a_{s-1}\RAngLe \cong {\maThBb z}_{2}^{s-1}$$ $$\begIn{arRAY}{C} B_{1},b_{2} \maPstO a_{1}\\ b_{3},b_{4} \MapstO a_{2}\\ b_{5},b_{6} \mapsto a_{3}\\ \vdOts\\ B_{2k-1}, b_{2k} \MApsTo a_{k}\\ \vDots\\ b_{2S

j-1}, - z_{j},z_{j+1},\ldo ts,z_{2s-1 }), \ ; j =1, .. .,2s -1.$ $ Inside theg roup $H_{0}$, the only non -triv ia l ele m en ts ac ting wi t hf i xed p oi nts a r e$b_{1 },\ ldots,b_{2s-1}$and $ b_{2s}=b_{1} b _{ 2}\cdots b _{2 s-1}$. The d egr ee $2^ {2 s-1 } $ hol omo rphic map $ $ P:D \t o \wideha t{ \ mathbb C}: (z_ { 1 }, \ldo ts,z_{2s-1}) \map s to - \left( \dfrac{ z_{2}} {z _ {1 } } \r igh t)^{2}$$ i sa bra n ched re g ul a r cov e r with deck g roup being$ H_{ 0}$. T he pr o jectio n und er $P$ of the set offixed poi nts ar e as fol l ows: $$ P({\rm Fi x}( b_{1 } )) =\ inf ty , \q u ad P( { \rm Fix}(b_ {2 }) )=0,\qua d P ( {\rm Fi x}(b _{3}) )=1,\quad P({ \rm Fix } (b_ {4})) =\lam bda_ {1 },$$$$P({\ rm Fi x} (b_{2k+3}))=\mu _{k, 1}, \quad P( {\r mFix}( b _{2k+4 })) =\m u_{k,2} , \qua d k= 1, \ l d ot s, s.$$ In partic ul a r ,the bran ch loc u sof $P$ is t he se t $$ \ { \inft y,0, 1 ,\ lambda,\mu_{1 , 1} ,\mu_{1, 2} ,\ldot s, \mu _{s -2,1} , \mu _{s-2, 2}\}.$$ By t h e Riemann-Hurw i tz formula, $ D $h a sg enus $g _{D}=1+2^{2 s-2} ( s-2) $. L et us consi der t he su r jective homomorphis m$$\the ta:H_ {0} \to H=\la ngle a_{1} , \ l dots,a_{ s-1} \ ra n gle \cong {\ma thbbZ}_{2}^{s- 1 }$$ $$\b egin{ array}{c } b_{1},b _ { 2} \maps toa_{ 1}\ \ b _ { 3} ,b_{4} \mapst o a_{2 }\ \ b_{5} ,b_ {6} \ma pst o a _{3 }\\ \ vdots\\ b _{2k-1}, b _{ 2k }\ma pstoa _{k}\\ \ vd ots \\ b_ {2s

j-1}, -_z_{j},z_{j+1},\ldots,z_{2s-1}), \;_j=1,...,2s-1.$$ Inside the group $H_{0}$,_the only_non-trivial_elements acting_with_fixed points are_$b_{1},\ldots, b_{2s-1}$ and_$b_{2s}=b_{1}b_{2}\cdots b_{2s-1}$. The degree_$2^{2s-1}$ holomorphic map_$$P:D_\to \widehat{\mathbb C}: (z_{1},\ldots,z_{2s-1}) \mapsto-\left( \dfrac{z_{2}}{z_{1}} \right)^{2}$$ is a branched regular cover with deck_group_being $H_{0}$._The_projection_under $P$ of the set_of fixed points are as_follows: $$P({\rm_Fix}(b_{1}))=\infty, \quad P({\rm Fix}(b_{2}))=0, \quad P({\rm Fix}(b_{3}))=1,\quad P({\rm_Fix}(b_{4}))=\lambda_{1},$$_$$P({\rm Fix}(b_{2k+3}))=\mu_{k,1}, \quad_ P({\rm Fix}(b_{2k+4}))=\mu_{k,2}, \quad k=1,\ldots, s.$$ In particular,_the branch locus of $P$ is_the set $$\{\infty,0,1,\lambda,_\mu_{1,1},_\mu_{1,2},\ldots,\mu_{s-2,1},_\mu_{s-2,2}\}.$$ By the Riemann-Hurwitz formula,_$D$ has genus $g_{D}=1+2^{2s-2}(s-2)$. Let us consider_the surjective homomorphism $$\theta:H_{0} \to H=\langle_a_{1},\ldots,a_{s-1}\rangle \cong {\mathbb Z}_{2}^{s-1}$$ $$\begin{array}{c} b_{1},b_{2} \mapsto a_{1}\\ b_{3},b_{4}_\mapsto a_{2}\\ b_{5},b_{6} \mapsto a_{3}\\ \vdots\\ b_{2k-1}, b_{2k} \mapsto_a_{k}\\ \vdots\\ b_{2s

-line envelops indicate the velocity criteria for rejecting interlopers. ([*b*]{}) Distribution of absolute magnitudes of all hosts (open histogram) and of those having satellites within the velocity criteria (shaded histogram). ([*c*]{}) Distribution of peculiar velocities of all satellites (open histogram) and of those within the velocity criteria (shaded histogram). []{data-label="fig-ncut"}](f1.eps){width="85mm"} In Figure \[fig-ncut\], we plot peculiar velocities of satellite galaxies as a function of absolute magnitudes of host galaxies. The peculiar velocity of a satellite is defined by, $v_{\rm pec}=S(v_{\rm sat}-v_{\rm host})$, where $v_{\rm sat}$ and $v_{\rm host}$ are observed velocities of the satellite and the host, respectively. $S$ is defined by $$\begin{aligned} \label{eq-S} S = \left\{ \begin{array}{l} +1,~\theta < 90^{\circ}, \\ -1,~\mathrm{otherwise.} \end{array} \right. \nonumber\end{aligned}$$ $\theta=\mid \mathrm{PAred} - \mathrm{PAS} \mid$ where PAred is a position angle of the receding side of the host’s major axis, and PAS is the position angle of the satellite with respect to the host. All position angles are measured from north to east. Therefore, the satellites with $v_{\rm pec}>0$ rotate around the host in the same direction as the rotation of the host (prograde orbit) and those with $v_{\rm pec}<0$ rotate in the opposite direction (retrograde orbit). Since the probability for a galaxy to become a satellite of a host decreases as the mass (luminosity) of the host decreases, we exclude some probable interlopers located outside the region bounded by dashed lines in Figure \[fig-ncut\]. The dashed lines are determined as follows. The escape velocity of a satellite can be given by $v_{\rm esc}\propto(M_{\rm host}/R)^{0.5}$ where $M_{\rm host}$ is a mass of the host and $R$ is a projected distance to the host. The mass of the host galaxy is computed by $M_{\rm host}=\gamma L$ where $\gamma$ is the mass-to-light ratio and $

-line envelops indicate the velocity criteria for rejecting intruder. ([ * b * ] { }) Distribution of absolute magnitude of all hosts (open histogram) and of those experience satellite within the velocity criteria (shaded histogram). ([ * cytosine * ] { }) Distribution of peculiar velocities of all satellite (open histogram) and of those within the velocity criterion (shaded histogram). [ ] { data - label="fig - ncut"}](f1.eps){width="85 mm " } In design \[fig - ncut\ ], we diagram peculiar velocities of satellite galaxies as a function of absolute magnitudes of horde galaxies. The peculiar velocity of a satellite is define by, $ v_{\rm pec}=S(v_{\rm sat}-v_{\rm host})$, where $ v_{\rm sat}$ and $ v_{\rm host}$ are observed velocities of the satellite and the host, respectively. $ S$ is defined by $ $ \begin{aligned } \label{eq - second } S = \left\ { \begin{array}{l } +1,~\theta < 90^{\circ }, \\ -1,~\mathrm{otherwise. } \end{array } \right. \nonumber\end{aligned}$$ $ \theta=\mid \mathrm{PAred } - \mathrm{PAS } \mid$ where PAred is a position angle of the receding side of the host ’s major bloc, and PAS is the position angle of the satellite with respect to the host. All position angles are measure from north to east. Therefore, the satellites with $ v_{\rm pec}>0 $ rotate around the host in the same direction as the rotation of the host (prograde orbit) and those with $ v_{\rm pec}<0 $ rotate in the opposite direction (retrograde sphere). Since the probability for a galaxy to become a satellite of a host decrease as the bulk (luminosity) of the host decreases, we exclude some probable intruder located outside the area bounded by dashed lines in Figure \[fig - ncut\ ]. The dash lines are determined as follows. The escape velocity of a satellite can be give by $ v_{\rm esc}\propto(M_{\rm host}/R)^{0.5}$ where $ M_{\rm host}$ is a mass of the host and $ R$ is a projected distance to the host. The bulk of the host galax is calculate by $ M_{\rm host}=\gamma L$ where $ \gamma$ is the bulk - to - light ratio and $

-linf envelops indicate the yelocity criteric for rxjectinf interlupers. ([*b*]{}) Distribution of absopure matnitudes of all hosts (upen histlgram) ane of rhose haviif satellites slthin vhe velocity crlteria (shadad histogram). ([*c*]{}) Dkscribution of peculiar velocities of wll satrlpites (open hisjograk) and of those within the velocity critsria (sheded histogram). []{cata-label="fig-ncut"}](f1.eps){width="85ml"} In Vigure \[fig-ncut\], we olot peculiqr vqoocities of ratellite galaxies as z function of absolute magnitudds of host galaziws. Hve peculiar velobity of a satcklite hs defimed by, $v_{\rm pec}=X(v_{\rk sqt}-v_{\rm host})$, where $v_{\rm vat}$ and $v_{\rm host}$ ate observeg relocities of the satwloite dnd dhe fist, reapxctjvely. $D$ ia defined gy $$\begin{alitned} \label{eq-S} S = \left\{ \bqtin{array}{l} +1,~\thsta < 90^{\cyrs}, \\ -1,~\mathrm{otherwise.} \end{array} \right. \nonukbed\end{aligned}$$ $\theta=\mid \matyrm{PAred} - \mathrm{PAS} \mld$ where [Ared is a position angle of the receding side of the iort’s mwgir axis, and PAS is the position angle of the saeslkine with respect tj the host. Slp ljsition angler are meaaured from north tl east. Jherefire, the swtelkites with $v_{\rm pec}>0$ rotate aeound the hofr in the same direetion as the rotstion of the host (prograde oxbit) ahd those wihh $v_{\rm ped}<0$ rotate in the oopoxide direcukon (retrograde orfit). Since vhe pxobabiligy fpr a gwlaxy to bfcome a satellite of a hlst dgcreasas as the lass (luminosity) of the host deccxases, we exclode soke probayle inberlopers locatqd outside the region bounaed by dasged linxs in Figure \[fig-ncut\]. The gwshed lines ere deteriinee as followr. The escape vekocity of a satellire can be given by $v_{\ro esc}\propto(M_{\rm hist}/T)^{0.5}$ where $M_{\rm hosy}$ ir a mwsx jx the host atd $R$ is s progected distcnze tp the host. The mass mf tge host galaxy is voiputed bi $M_{\rm hose}=\gamma L$ wherr $\gamma$ is the masd-to-lijht ravio anc $

-line envelops indicate the velocity criteria for ([*b*]{}) of absolute of all hosts having within the velocity (shaded histogram). ([*c*]{}) of peculiar velocities of all satellites histogram) and of those within the velocity criteria (shaded histogram). []{data-label="fig-ncut"}](f1.eps){width="85mm"} In Figure we plot peculiar velocities of satellite galaxies as a function of absolute magnitudes host The velocity a satellite is defined by, $v_{\rm pec}=S(v_{\rm sat}-v_{\rm host})$, where $v_{\rm sat}$ and $v_{\rm host}$ are velocities of the satellite and the host, respectively. is defined by $$\begin{aligned} S = \left\{ \begin{array}{l} +1,~\theta 90^{\circ}, -1,~\mathrm{otherwise.} \end{array} \nonumber\end{aligned}$$ \mathrm{PAred} \mathrm{PAS} \mid$ where is a position angle of the receding side of the host’s major axis, and PAS is the angle of with respect the All angles are measured to east. Therefore, the satellites with around the host in the same direction as rotation of host (prograde orbit) and those with pec}<0$ rotate in the opposite direction (retrograde orbit). the probability for a galaxy to become a satellite of a host decreases as the of the host decreases, exclude some probable located the bounded dashed lines Figure \[fig-ncut\]. The dashed lines are determined as follows. The escape of a satellite can be given by $v_{\rm esc}\propto(M_{\rm host}/R)^{0.5}$ host}$ a mass of host and $R$ is projected to the host. The the galaxy $M_{\rm L$ $\gamma$ is the mass-to-light and $

-line envelops indicate the veLocity critEria fOr rEjeCtIng iNterLopers. ([*b*]{}) DistribUTion Of absolute magnitudes of All hoStS (Open HIsTograM) and of tHOsE HAviNg SaTelLiTEs WithiN thE velociTy criteria (ShaDeD histogram). ([*c*]{}) DIStRibution of PecUliar velocitIes Of all sAtEllITes (opEn hIstogRam) and OF those Within the VeLOcity cRIteria (sHADeD hisTogram). []{data-label="fiG-NcUT"}](f1.eps){width="85mm"} In figure \[FiG-NcUT\], We pLot Peculiar veLoCitieS Of satelLItE GALaxIEs as a function Of absolute mAGniTudes oF hOst GAlaxieS. The pEcULiaR velocity of A satEllite is dEfined BY, $v_{\rm pec}=s(V_{\rm sat}-v_{\Rm host})$, WheRe $v_{\Rm saT}$ AnD $v_{\Rm hOsT}$ Are OBsErvED veLocities Of ThE sateLlitE AND The hOst, RespEctivEly. $S$ is defined By $$\bEgin{ALigNed} \laBel{eq-s} S = \leFt\{ \Begin{Array}{l} +1,~\Theta < 90^{\CiRc}, \\ -1,~\mathrm{otherwiSe.} \enD{array} \rigHt. \nOnUmbEr\End{alIGned}$$ $\thEta=\Mid \Mathrm{Pared} - \matHRm{PaS} \MID$ WhEre PAred is a positioN aNGLe Of the recEding sIDe Of THe host’s mAjOr aXis, aND pAS is The pOSiTion anglE of the SAtElLite witH rEspect To The HosT. All pOSitiOn anglEs are meaSured FRom north to east. tHerefore, the saTElLITeS With $V_{\rm Pec}>0$ rotate arOund THe hoSt in THe SamE DirecTion aS tHE rOTation of the host (progRaDe orbiT) and tHose with $v_{\rm peC}<0$ rotate in tHE OPposite dIrecTIoN (Retrograde orbiT). SincE the probabILity for a GalaxY to becomE a satelliTE Of a host dEcrEasEs aS thE MAsS (luminosity) of THE hosT dEcreaseS, we Exclude SomE prObaBle InTerlopers Located oUtSiDe ThE reGion bOUnded by dAsHed LiNes In FigURe \[fig-nCut\]. ThE dasHeD lINes Are deteRMiNED as fOlLoWs. ThE esCaPe velOcitY Of a SatelliTe can be giVen BY $v_{\rm EsC}\pRopto(M_{\rM host}/R)^{0.5}$ where $M_{\rM hOst}$ is a mass Of The Host anD $r$ Is a projeCted distance to the host. ThE Mass of tHe hOst gaLaxy Is computeD by $m_{\rm hosT}=\gaMMa L$ wheRe $\gammA$ is thE mAss-TO-Light RATiO anD $

-line envelops indicate th e velocity crit eri a f or rej ecti ng interlopers . ([* b*]{}) Distribution of abso lu t e ma g ni tudes of all ho s t s ( op en hi st o gr am) a ndof thos e having s ate ll ites withint he velocitycri teria (shade d h istogr am ).( [*c*] {}) Dist ributi o n of p eculiar v el o cities of alls a te llit es (open histogra m )a nd of those wi thin t he ve l o cit y c riteria (s ha ded h i stogram ) .[ ] { dat a -label="fig-n cut"}](f1.e p s){ width= "8 5mm " } InFigur e\ [fi g-ncut\], w e pl ot peculi ar vel o citieso f satel lite g ala xie s as afu nct io n of ab sol u temagnitud es o f hos t ga l a x i es.The pec uliar velocity ofa s atel l ite is d efine d by ,$v_{\ rm pec }=S(v _{ \rm sat}-v_{\rm hos t})$, whe re$v _{\ rm sat} $ and $ v_{ \rm host}$ are ob s erv ed v e lo cities of the sate ll i t eand thehost,r es pe c tively.$S $ i s de f i ned b y $$ \ be gin{alig ned} \ l ab el {eq-S}S= \lef t\ { \ beg in{ar r ay}{ l} + 1,~\thet a < 9 0 ^{\circ}, \\ -1,~\mathrm{o t he r w is e .} \ end {array} \ri ght. \non umbe r \e nd{ a ligne d}$$ $ \ th e ta=\mid \mathrm{PAr ed } - \m athrm {PAS} \mid$ w here PAred i s a posit iona ng l e of the reced ing s ide of the host’s m ajoraxis, an d PAS ist h e positi onang leoft h esatellite wit h resp ec t to th e h ost. Al l p osi tio n a ng les are m easuredfr om n or thto ea s t. There fo re, t hesatel l ites w ith $ v_{\ rm p e c}> 0$ rota t ea r ound t he hos t i nthe s amed ire ction a s the rot ati o n of t he host ( prograde orbi t) and those w ith $v_{\ r m pec}<0$ rotate in the opposite directi on(retr ogra de orbit) . Sincethe probab ilityfor a g ala x y to b e c om e a s atellite o f a h ost d ec reas es as t he mass (luminosit y ) o f the host de cre ases , we ex c lu d e s om e pr o b able interloper s locatedou t si de the reg i onbo unded b y dashe d lin e s in Fi gure \[fi g-ncut\]. T he d a s hed lines are determi ned as fo l lows. Th e esc ape veloc it y o f a s atelli t e c an be given b y $v_{ \rm e sc }\propto (M_{\rm host}/R)^{0.5}$ where $M_{ \rm host}$ i s a mas s of thehost and $R$ i s a pr oject edd istan ce t o t heh ost.Them ass of th e h ost g al axy is comp u t e d b y $M_ {\r m host} =\ga mma L$ where $\ga m ma$ is the mas s-to - l igh t r a tioan d $

-line envelops_indicate the_velocity criteria for rejecting_interlopers. ([*b*]{})_Distribution_of absolute_magnitudes_of all hosts_(open histogram) and_of those having satellites_within the velocity_criteria_(shaded histogram). ([*c*]{}) Distribution of peculiar velocities of all satellites (open histogram) and of_those_within the_velocity_criteria_(shaded histogram). []{data-label="fig-ncut"}](f1.eps){width="85mm"} In Figure \[fig-ncut\],_we plot peculiar velocities of_satellite galaxies_as a function of absolute magnitudes of host_galaxies._The peculiar velocity_of a satellite is defined by, $v_{\rm pec}=S(v_{\rm sat}-v_{\rm_host})$, where $v_{\rm sat}$ and $v_{\rm_host}$ are observed_velocities_of_the satellite and the_host, respectively. $S$ is defined by_$$\begin{aligned} \label{eq-S} S = \left\{ \begin{array}{l} +1,~\theta_< 90^{\circ}, \\ -1,~\mathrm{otherwise.} \end{array} \right._\nonumber\end{aligned}$$ $\theta=\mid \mathrm{PAred} - \mathrm{PAS} \mid$ where_PAred is a position angle_of the_receding side of the host’s_major axis, and_PAS is_the position angle_of the satellite with respect to_the host. All_position angles are measured from north_to_east. Therefore, the_satellites_with_$v_{\rm pec}>0$_rotate around the_host_in the_same_direction as the rotation of the_host_(prograde orbit) and those with $v_{\rm pec}<0$_rotate in the opposite_direction_(retrograde orbit). Since the probability_for a galaxy to become_a satellite of a host decreases_as the_mass (luminosity)_of the host decreases, we exclude some probable interlopers located outside_the region bounded by dashed lines_in Figure \[fig-ncut\]. The_dashed lines_are_determined as follows._The_escape velocity_of a satellite can be given by_$v_{\rm esc}\propto(M_{\rm_host}/R)^{0.5}$ where $M_{\rm host}$ is a_mass of the host_and_$R$ is a projected distance to_the host. The mass of the_host galaxy is computed by_$M_{\rm_host}=\gamma_L$ where $\gamma$ is the_mass-to-light ratio and $

^*~(m_{\rm e})$ ---------- ------------------ -------------- ------------------- $\gamma$ ${\bf H}||[100]$ $2425\pm 10$ $1.31 \pm 0.08$ $\alpha$ ${\bf H}||[100]$ $3045\pm 10$ $2.0 \pm 0.1$ $\gamma$ ${\bf H}||[110]$ $2182\pm 10$ $1.12 \pm 0.04$ $\alpha$ ${\bf H}||[110]$ $3770\pm 10$ $3.3 \pm 0.5$ $\gamma$ ${\bf H}||[111]$ $2185\pm 10$ $1.14 \pm 0.05$ $\alpha$ ${\bf H}||[111]$ $3525\pm 10$ $2.8 \pm 0.2$ : Selected Fermi-surface properties of LaOs$_4$As$_{12}$. The first column denotes the Fermi-surface section (FSS), and the second the field orientation; $F$ is the de Haas-van Alphen frequency. The effective masses, which are field-independent, are evaluated at $13.6$ T.[]{data-label="table2"} FSS $F_{\rm expt}$ (T) $F_{\rm theory}$ (T) $|\frac{m^*_{\rm expt}}{m_{\rm e}}|$ $\frac{m^*_{\rm theory}}{m_{\rm e}}$ $|\frac{m^*_{\rm expt}}{m^*_{\rm theory}}|$ -------------------- -------------------- ---------------------- -------------------------------------- -------------------------------------- --------------------------------------------- ${\bf H} || [100]$ $\gamma$ 2210 2245 $2.6\pm 0.2$ -0.86 3.0 $\beta$ 2740 2621 $3.0 \pm 0.1$ -0.87 3.4 $\alpha$ 3005 2651 $4.4 \pm 0.1$ -0.95 4.6 ${\bf H} || [110]$

^*~(m_{\rm e})$ ---------- ------------------ -------------- ------------------- $ \gamma$ $ { \bf H}||[100]$ $ 2425\pm 10 $ $ 1.31 \pm 0.08 $ $ \alpha$ $ { \bf H}||[100]$ $ 3045\pm 10 $ $ 2.0 \pm 0.1 $ $ \gamma$ $ { \bf H}||[110]$ $ 2182\pm 10 $ $ 1.12 \pm 0.04 $ $ \alpha$ $ { \bf H}||[110]$ $ 3770\pm 10 $ $ 3.3 \pm 0.5 $ $ \gamma$ $ { \bf H}||[111]$ $ 2185\pm 10 $ $ 1.14 \pm 0.05 $ $ \alpha$ $ { \bf H}||[111]$ $ 3525\pm 10 $ $ 2.8 \pm 0.2 $ : Selected Fermi - surface properties of LaOs$_4$As$_{12}$. The first column announce the Fermi - open section (FSS), and the second the field orientation; $ F$ is the de Haas - van Alphen frequency. The effective mass, which are field - independent, are evaluate at $ 13.6 $   T.[]{data - label="table2 " } FSS $ F_{\rm expt}$ (T) $ F_{\rm theory}$ (T) $ |\frac{m^*_{\rm expt}}{m_{\rm e}}|$ $ \frac{m^*_{\rm theory}}{m_{\rm e}}$ $ |\frac{m^*_{\rm expt}}{m^*_{\rm theory}}|$ -------------------- -------------------- ---------------------- -------------------------------------- -------------------------------------- --------------------------------------------- $ { \bf H } || [ 100]$ $ \gamma$ 2210 2245 $ 2.6\pm 0.2 $ -0.86 3.0 $ \beta$ 2740 2621 $ 3.0 \pm 0.1 $ -0.87 3.4 $ \alpha$ 3005 2651 $ 4.4 \pm 0.1 $ -0.95 4.6 $ { \bf H } || [ 110]$

^*~(m_{\rm e})$ ---------- ------------------ -------------- ------------------- $\gamma$ ${\bf H}||[100]$ $2425\km 10$ $1.31 \pm 0.08$ $\alpha$ ${\bf H}||[100]$ $3045\pm 10$ $2.0 \pm 0.1$ $\gamma$ ${\bf H}||[110]$ $2182\pm 10$ $1.12 \pm 0.04$ $\elphq$ ${\bd H}||[110]$ $3770\pm 10$ $3.3 \pm 0.5$ $\gammx$ ${\bf H}||[111]$ $2185\pm 10$ $1.14 \pm 0.05$ $\alpha$ ${\bh H}||[111]$ $3525\pm 10$ $2.8 \pj 0.2$ : Vxlected Fermi-sutface properdies of LaOs$_4$As$_{12}$. Tfe first column denotes the Fermi-surfwce secyiln (FSS), and the secpgd tgv nield orientation; $F$ is the de Hazs-van Anphen frequenvy. The effective masses, whlch wre field-independejt, are evalozteq at $13.6$ T.[]{data-labdl="table2"} FSS $F_{\rm expt}$ (T) $F_{\rm theory}$ (T) $|\yrac{m^*_{\rm expj}}{m_{\dm g}}|$ $\frac{m^*_{\rm vheory}}{i_{\rm e}}$ $|\frac{m^*_{\rm expt}}{k^*_{\rm thepry}}|$ -------------------- -------------------- ---------------------- -------------------------------------- -------------------------------------- --------------------------------------------- ${\bf H} || [100]$ $\gamma$ 2210 2245 $2.6\pm 0.2$ -0.86 3.0 $\beta$ 2740 2621 $3.0 \pm 0.1$ -0.87 3.4 $\alpha$ 3005 2651 $4.4 \pm 0.1$ -0.95 4.6 ${\bf H} || [110]$

^*~(m_{\rm e})$ ---------- ------------------ -------------- ------------------- $\gamma$ $2425\pm $1.31 \pm $\alpha$ ${\bf H}||[100]$ $\gamma$ H}||[110]$ $2182\pm 10$ \pm 0.04$ $\alpha$ H}||[110]$ $3770\pm 10$ $3.3 \pm 0.5$ ${\bf H}||[111]$ $2185\pm 10$ $1.14 \pm 0.05$ $\alpha$ ${\bf H}||[111]$ $3525\pm 10$ $2.8 0.2$ : Selected Fermi-surface properties of LaOs$_4$As$_{12}$. The first column denotes the Fermi-surface (FSS), the the orientation; $F$ is the de Haas-van Alphen frequency. The effective masses, which are field-independent, are evaluated $13.6$ T.[]{data-label="table2"} FSS $F_{\rm expt}$ (T) $F_{\rm theory}$ $|\frac{m^*_{\rm expt}}{m_{\rm e}}|$ $\frac{m^*_{\rm e}}$ $|\frac{m^*_{\rm expt}}{m^*_{\rm theory}}|$ -------------------- ---------------------- -------------------------------------- --------------------------------------------- H} [100]$ 2210 2245 $2.6\pm -0.86 3.0 $\beta$ 2740 2621 $3.0 \pm 0.1$ -0.87 3.4 $\alpha$ 3005 2651 $4.4 \pm 0.1$ -0.95 ${\bf H}

^*~(m_{\rm e})$ ---------- ------------------ -------------- ------------------- $\gamma$ ${\bf H}||[100]$ $2425\pm 10$ $1.31 \pm 0.08$ $\alpha$ ${\bf H}||[100]$ $3045\pM 10$ $2.0 \pm 0.1$ $\gamma$ ${\bf h}||[110]$ $2182\pm 10$ $1.12 \pm 0.04$ $\AlpHa$ ${\bF H}||[110]$ $3770\Pm 10$ $3.3 \pm 0.5$ $\GammA$ ${\bf H}||[111]$ $2185\pm 10$ $1.14 \pm 0.05$ $\alpha$ ${\bF h}||[111]$ $3525\pm 10$ $2.8 \pM 0.2$ : Selected Fermi-surface pRoperTiES of LAos$_4$as$_{12}$. The First coLUmN DEnoTeS tHe FErMI-sUrfacE seCtion (FSs), and the secOnd ThE field orientATiOn; $F$ is the de haaS-van Alphen frEquEncy. ThE eFfeCTive mAssEs, whiCh are fIEld-indEpendent, aRe EValuatED at $13.6$ T.[]{datA-LAbEl="taBle2"} FSS $F_{\rm expt}$ (T) $F_{\rm THeORy}$ (T) $|\frac{m^*_{\rm expt}}{M_{\rm e}}|$ $\frAc{M^*_{\Rm THEorY}}{m_{\rM e}}$ $|\frac{m^*_{\rm eXpT}}{m^*_{\rm tHEory}}|$ -------------------- -------------------- ---------------------- -------------------------------------- -------------------------------------- --------------------------------------------- ${\bf H} || [100]$ $\GAmMA$ 2210 2245 $2.6\PM 0.2$ -0.86 3.0 $\beTA$ 2740 2621 $3.0 \pm 0.1$ -0.87 3.4 $\alpha$ 3005 2651 $4.4 \pm 0.1$ -0.95 4.6 ${\bf H} || [110]$

^*~(m_{\rm e})$ -------- -- ------- ----- --- --- - ---- ---- ----- -------- - ---- ------ $\gamma$ ${ \bf H }| | [100 ] $ $24 25\pm 1 0 $$1. 31 \ pm0. 0 8$ $\ alp ha$ $ {\bf H}||[ 100 ]$ $3045\pm1 0$ $2.0 \p m 0 .1$ $\gamm a$ ${\b fH}| | [110] $ $218 2\pm 1 0 $ $1 .12 \pm 0 .0 4 $ $\ a lpha$ $ {\ bf H }||[110]$ $3770 \ pm 10$ $3.3 \pm 0.5$ $ \g a m ma$ ${\bf H}|| [1 11]$ $2185\ p m1 0 $ $ 1.14 \pm 0.05 $ $\alpha $ ${\bfH} ||[ 1 11]$ $352 5\ p m 1 0$ $2.8 \ pm 0 .2$ :Select e d Fermi - surface prope rti esof L a Os $_ 4$A s$ _ {12 } $. Th e fi rst colu mn d enote s th e F e rmi- sur face sect ion (FSS), an d t he s e con d the fiel d or ie ntati on; $F $ isth e de Haas-van A lphe n frequen cy. T heef fecti v e mass es, wh ich are field- i nde pe n d e nt , are evaluated at $ 1 3 .6 $ T.[]{d ata-la b el =" t able2"} FS S $F_{\rm expt} $ ( T) $F_{ \r m theo ry }$(T) $| \ frac {m^*_{ \rm expt }}{m_ { \rm e}}|$ $\ f rac{m^*_{\rmt he o r y} } {m_{ \rm e}}$ $|\ frac { m^*_ {\rm ex pt} } {m^*_ {\rmth e or y }}|$ ------------ -- ------ ---- ------------- --- ------ - - - -------- ---- - - - -------------- ----- ---------- - ------ - ----- -------- --------- - - -------- --- ----- --- - - -- ------------- - - ---- -- ------- --- ---- ${\ bfH}||[1 00]$ $\gamma $ 2210 2245 $2.6\pm 0.2$ -0 .86 3.0 $\bet a$ 274 0 2621 $3.0 \pm 0.1 $ - 0.8 7 3.4 $ \ al pha$ 3005 2651 $4 .4 \ p m 0. 1$ -0.95 4.6 ${ \bf H} | | [110 ]$

^*~(m_{\rm e})$ _ ----------_------------------ -------------- ------------------- _$\gamma$ __${\bf H}||[100]$__ $2425\pm 10$_ $1.31_\pm 0.08$ $\alpha$_ ${\bf_H}||[100]$_ $3045\pm 10$ $2.0 \pm 0.1$ $\gamma$ ${\bf_H}||[110]$_ _$2182\pm_10$_ $1.12 \pm 0.04$ _ $\alpha$ ${\bf_H}||[110]$ _ $3770\pm 10$ $3.3 \pm 0.5$ __$\gamma$ _${\bf H}||[111]$ $2185\pm 10$ $1.14_\pm 0.05$ $\alpha$ _${\bf H}||[111]$ __$3525\pm_10$ $2.8_\pm 0.2$ : Selected Fermi-surface_properties of LaOs$_4$As$_{12}$. The first column_denotes the Fermi-surface section (FSS), and the_second the field orientation; $F$ is_the de Haas-van Alphen frequency._The effective_masses, which are field-independent, are_evaluated at $13.6$ T.[]{data-label="table2"} _ FSS_ _ _ _ $F_{\rm_expt}$_(T) _$F_{\rm_theory}$_(T) _ $|\frac{m^*_{\rm expt}}{m_{\rm_e}}|$_ _$\frac{m^*_{\rm_theory}}{m_{\rm e}}$ $|\frac{m^*_{\rm expt}}{m^*_{\rm_theory}}|$ _ -------------------- -------------------- ---------------------- -------------------------------------- -------------------------------------- --------------------------------------------- _ ${\bf H} ||_[100]$_ _ _ _ _ _ _ _ _ __ __ _ _ _ _ __ _ _ ___ _ _ _ __ __ __ _ $\gamma$ _ _ 2210 _ _ ___ 2245 _ _ _ $2.6\pm 0.2$ _ _ _ -0.86 _ _ __ ____ _ _ _3.0 _$\beta$ _ _ __ 2740 _ 2621 _ __ _ _ $3.0 \pm 0.1$ _ __ _ _ _ _ _ -0.87_ _ _ __ ____ _ _ _3.4 $\alpha$__ _ _ 3005 _ _ 2651 _ _ _ $4.4_\pm_0.1$ _ _ __ -0.95 ___ __ _____ _ __ 4.6 ${\bf H}_|| [110]$__ _ _ __ ____ _ __ _ __ _ _ _ _ _ _ _ __ _ _

6 0.9728 0.9728 8 0.9669 0.9771 10 0.9592 0.9659 12 0.8844 0.9165 14 0.8858 0.9213 16 0.8833 0.9170 $N$ ${\cal O}_{k=1}$ ${\cal O}_{k=2}$ ${\cal O}_{1/r}$ ----- ------------------ ------------------ ------------------ 6 1.0 1.0 1.0 9 0.9642 0.9891 0.9642 12 0.8647 0.9702 0.8904 15 0.9189 0.9788 0.9307 18 0.6774 0.9239 0.7226 $N$ ${\cal O}_{k=1}$ ${\cal O}_{k=2}$ ${\cal O}_{k=3}$ ${\cal O}_{1/r}$ ----- ------------------ ------------------ ------------------ ------------------ 8 1.0 1.0 1.0 1.0 12 0.9636 0.9811 0.9949 0.9636 16 0.7801 0.8919 0.9753 0.8037 20 0.8822 0.9499 0.9874 0.8985 $N$ ${\cal O}_{k=1}$ ${\cal O}_{k=2}$ ${\cal O}_{k=3}$ ${\cal O}_{k=4}$ ${\cal O}_{1/r}$ ----- ------------------ ------------------ ------------------ ------------------ ------------------ 10 1.0 1.0 1.0 1.0 1.0 15 0.9659 0.9789 0.9902 0.9975 0.9659 20 0.7455 0.8291 0.9165 0.9798 0.7644 ![Upper part: total overlap for as a function of $V_2/V_0$ at filling factors $\nu=1$ with two quasiholes (

6 0.9728 0.9728 8 0.9669 0.9771 10 0.9592 0.9659 12 0.8844 0.9165 14 0.8858 0.9213 16 0.8833 0.9170 $ N$ $ { \cal O}_{k=1}$ $ { \cal O}_{k=2}$ $ { \cal O}_{1 / r}$ ----- ------------------ ------------------ ------------------ 6 1.0 1.0 1.0 9 0.9642 0.9891 0.9642 12 0.8647 0.9702 0.8904 15 0.9189 0.9788 0.9307 18 0.6774 0.9239 0.7226 $ N$ $ { \cal O}_{k=1}$ $ { \cal O}_{k=2}$ $ { \cal O}_{k=3}$ $ { \cal O}_{1 / r}$ ----- ------------------ ------------------ ------------------ ------------------ 8 1.0 1.0 1.0 1.0 12 0.9636 0.9811 0.9949 0.9636 16 0.7801 0.8919 0.9753 0.8037 20 0.8822 0.9499 0.9874 0.8985 $ N$ $ { \cal O}_{k=1}$ $ { \cal O}_{k=2}$ $ { \cal O}_{k=3}$ $ { \cal O}_{k=4}$ $ { \cal O}_{1 / r}$ ----- ------------------ ------------------ ------------------ ------------------ ------------------ 10 1.0 1.0 1.0 1.0 1.0 15 0.9659 0.9789 0.9902 0.9975 0.9659 20 0.7455 0.8291 0.9165 0.9798 0.7644 ! [ Upper part: total overlap for as a function of $ V_2 / V_0 $ at woof divisor $ \nu=1 $ with two quasiholes (

6 0.9728 0.9728 8 0.9669 0.9771 10 0.9592 0.9659 12 0.8844 0.9165 14 0.8858 0.9213 16 0.8833 0.9170 $N$ ${\cal O}_{k=1}$ ${\wau L}_{k=2}$ ${\cal O}_{1/r}$ ----- ------------------ ------------------ ------------------ 6 1.0 1.0 1.0 9 0.9642 0.9891 0.9642 12 0.8647 0.9702 0.8904 15 0.9189 0.9788 0.9307 18 0.6774 0.9239 0.7226 $N$ ${\cal O}_{k=1}$ ${\cal O}_{k=2}$ ${\cal O}_{k=3}$ ${\cxl O}_{1/r}$ ----- ------------------ ------------------ ------------------ ------------------ 8 1.0 1.0 1.0 1.0 12 0.9636 0.9811 0.9949 0.9636 16 0.7801 0.8919 0.9753 0.8037 20 0.8822 0.9499 0.9874 0.8985 $B$ ${\cal O}_{k=1}$ ${\dal O}_{k=2}$ ${\cal O}_{k=3}$ ${\cal O}_{k=4}$ ${\cal O}_{1/r}$ ----- ------------------ ------------------ ------------------ ------------------ ------------------ 10 1.0 1.0 1.0 1.0 1.0 15 0.9659 0.9789 0.9902 0.9975 0.9659 20 0.7455 0.8291 0.9165 0.9798 0.7644 ![Ulker part: total overlcl ror as a function lf $V_2/V_0$ aj fillung factows $\ni=1$ with two quasiholes (

6 0.9728 0.9728 8 0.9669 0.9771 10 12 0.9165 14 0.9213 16 0.8833 O}_{k=2}$ O}_{1/r}$ ----- ------------------ ------------------ 6 1.0 1.0 9 0.9642 0.9891 0.9642 12 0.9702 0.8904 15 0.9189 0.9788 0.9307 18 0.6774 0.9239 0.7226 $N$ ${\cal O}_{k=1}$ O}_{k=2}$ ${\cal O}_{k=3}$ ${\cal O}_{1/r}$ ----- ------------------ ------------------ ------------------ ------------------ 8 1.0 1.0 1.0 0.9636 0.9949 16 0.7801 0.8919 0.9753 0.8037 20 0.8822 0.9499 0.9874 0.8985 $N$ ${\cal O}_{k=1}$ ${\cal O}_{k=2}$ ${\cal ${\cal O}_{k=4}$ ${\cal O}_{1/r}$ ----- ------------------ ------------------ ------------------ ------------------ 10 1.0 1.0 1.0 1.0 15 0.9659 0.9789 0.9975 20 0.7455 0.9165 0.7644 part: total overlap as a function of $V_2/V_0$ at filling factors $\nu=1$ with two quasiholes (

6 0.9728 0.9728 8 0.9669 0.9771 10 0.9592 0.9659 12 0.8844 0.9165 14 0.8858 0.9213 16 0.8833 0.9170 $N$ ${\cal O}_{k=1}$ ${\cal O}_{k=2}$ ${\cal O}_{1/r}$ ----- ------------------ ------------------ ------------------ 6 1.0 1.0 1.0 9 0.9642 0.9891 0.9642 12 0.8647 0.9702 0.8904 15 0.9189 0.9788 0.9307 18 0.6774 0.9239 0.7226 $N$ ${\cal O}_{k=1}$ ${\cal o}_{k=2}$ ${\cal O}_{k=3}$ ${\cal o}_{1/r}$ ----- ------------------ ------------------ ------------------ ------------------ 8 1.0 1.0 1.0 1.0 12 0.9636 0.9811 0.9949 0.9636 16 0.7801 0.8919 0.9753 0.8037 20 0.8822 0.9499 0.9874 0.8985 $N$ ${\caL O}_{k=1}$ ${\Cal o}_{k=2}$ ${\Cal O}_{K=3}$ ${\cal o}_{k=4}$ ${\cal O}_{1/r}$ ----- ------------------ ------------------ ------------------ ------------------ ------------------ 10 1.0 1.0 1.0 1.0 1.0 15 0.9659 0.9789 0.9902 0.9975 0.9659 20 0.7455 0.8291 0.9165 0.9798 0.7644 ![Upper paRT: totAl overlap for as a functioN of $V_2/V_0$ At FIlliNG fActorS $\nu=1$ with TWo QUAsiHoLeS (

6 0.9728 0.9 728 8 0.9669 0.9771 10 0 .9592 0. 965 9 12 0 .8844 0.916 5 14 0.8858 0 .9213 16 0.8 833 0.9170 $N$ ${\cal O}_{k=1 } $ ${\cal O}_{k= 2}$ ${ \ ca l O}_ {1/ r}$ ---- ------ - ------- - -- - - --- - ------------- - --------- - --- ----- 6 1. 0 1.0 1 .0 9 0.96 42 0.9 89 1 0.9642 12 0 . 8 6 47 0 .9702 0. 8 904 1 5 0 .9189 0.9788 0.9307 18 0.6 7 74 0.92 39 0 .7 226 $N$ ${\c al O }_ {k=1}$ ${\ca l O }_ { k=2}$ ${ \ca l O} _ { k=3}$ $ { \c al O}_{1 /r}$ -- -- - ----- -- ------ -- --- -- ----- - ---- ------ ------- ----- - ----- -------- - --------- 8 1.0 1.0 1 .0 1.0 12 0. 9636 0 .9811 0.9 949 0.9636 1 6 0 .7801 0.8 919 0.9 753 0.8037 20 0.8822 0 .94 99 0. 9874 0. 8985 $N$ $ {\c al O} _{k=1 } $ ${ \calO}_{ k= 2} $ ${\calO }_ { k =3}$ $ {\ca l O }_ {k=4} $ $ {\c al O}_{ 1/r}$ - --- - --- -- -- ------- ---- -------- -- -------- - -- --- ------ - - ---- --- --------------- ------- - ------- --- 1 0 1. 0 1. 0 1 .0 1.0 1.0 15 0.9659 0.9789 0.9 9 02 0.9975 0.965 9 20 0.7 45 5 0.829 1 0.9 165 0 .97 98 0.7 644 ![Up p er pa r t: tota l o verlap f oras afuncti o n o f $V_ 2/V_0$ a t fill ing f ac tors $\n u=1$ with two quasihole s (

_ _6 _ __ __ 0.9728_ _ _ __ 0.9728 8 __ ___ 0.9669 _ _ _ 0.9771 __ 10 _ _ 0.9592 _ ___ _ 0.9659 _12 _ 0.8844_ _ _ _ 0.9165 _ 14 _ _ _ 0.8858_ _ __ ___ 0.9213 _ 16__ __ __0.8833 _ __ _ 0.9170 $N$_ ${\cal O}_{k=1}$ _${\cal O}_{k=2}$_ _${\cal O}_{1/r}$ ----- ------------------ ------------------ ------------------ 6_ _ 1.0_ __ __ _ 1.0_ _ _ __ 1.0 _9 _ 0.9642 ___ _ 0.9891 _ _ __0.9642 12 __ 0.8647___ _ 0.9702_ _ _0.8904 15 ___ 0.9189 _ _ _ 0.9788 _ 0.9307 _ 18 _0.6774 _ 0.9239 _ __ 0.7226 ___$N$_ ${\cal_O}_{k=1}$ _${\cal O}_{k=2}$ _ ${\cal_O}_{k=3}$ _${\cal O}_{1/r}$ _ ----- ------------------_------------------_------------------ ------------------ 8 1.0 _ _ __ 1.0 _ _ _ 1.0 __ _ _ _1.0 _ 12 _ _ _ _ 0.9636 __ 0.9811 ____ _ _ 0.9949 _ __ _ 0.9636 _ 16 _ 0.7801 _ 0.8919 _ _ 0.9753_ __ _ _ 0.8037 20 __ 0.8822 ___ _0.9499_ _____ 0.9874 _ __ 0.8985 $N$_ _${\cal_O}_{k=1}$ ${\cal_O}_{k=2}$ ${\cal_O}_{k=3}$ ${\cal O}_{k=4}$ ${\cal O}_{1/r}$ -----_------------------_------------------ ------------------_------------------_------------------ __ 10 _ 1.0 __ _ 1.0 __ _ 1.0_ _ _ 1.0 _ 1.0 15_ _ _0.9659_ _ _ 0.9789 _ 0.9902 _ 0.9975 _ _ _ 0.9659 _20 0.7455 _ 0.8291 __ _ 0.9165 _ _ __ 0.9798 _ _ _ _ 0.7644 ![Upper part: total overlap for as a_function of $V_2/V_0$_at filling factors $\nu=1$ with two quasiholes (

_{m-1} -1) c_{k,j}^{m-2, n} \\ = & c_{k-1, j}^{m-1, n} + (L -1 ) c_{k+1, j}^{m-1, n} - (L_{m-1} -1) c_{k,j}^{m-2, n}. \end{split}$$ Exactly as we computed (with the difference that Lemma \[Lem=Blackbox\] is replaced by Lemma \[Lem=Blackbox\] ) we get the equalities $$h_m \gamma h_n = \sum_{k = 0}^{m+1} \sum_{j = 0}^{n} ( c_{k-1,j}^{m-1, n} + (L -1) c_{k+1,j}^{m-1, n}) \gamma_{k, j} - (L_{m-1}-1) \sum_{k = 0}^{m-2} \sum_{j = 0}^{n} c_{k,j}^{m-2, n} \gamma_{k,j}.$$ Thus $$c^{m,n}_{k,j} = c_{k-1, j}^{m-1, n} + (L -1 ) c_{k+1, j}^{m-1, n} - (L_m -1) c_{k,j}^{m-2, n}.$$ Combining the above with gives $c^{m,n}_{k,j} = b_{k,j}^{m,n} + b_{k+1,j+1}^{m,n}$ for all $0 \leq k \leq m, 0 \leq j \leq n$. [*Proof of Theorem \[Thm=PukanszkyCore\].*]{} Lemma \[Lem=abcomputations\] shows that ${\mathsf{B}}\gamma {\mathsf{B}}\subseteq X_{\gamma}$ and ${\mathsf{B}}\beta {\mathsf{B}}\subseteq X_\beta$ and hence the inclusions hold also for the $\Vert \: \Vert_2$-closures. For the converse inclusion proceed by induction: take $h_n \gamma h_m \in {\mathsf{B}}\gamma {\mathsf{B}}$ and assume that all vectors $h_r \beta h_s$ with $r < n, s

_ { m-1 } -1) c_{k, j}^{m-2, n } \\ = & c_{k-1, j}^{m-1, n } + (L -1) c_{k+1, j}^{m-1, n } - (L_{m-1 } -1) c_{k, j}^{m-2, n }. \end{split}$$ Exactly as we computed (with the difference that Lemma \[Lem = Blackbox\ ] is replace by Lemma \[Lem = Blackbox\ ]) we make the equalities $ $ h_m \gamma h_n = \sum_{k = 0}^{m+1 } \sum_{j = 0}^{n } (c_{k-1,j}^{m-1, n } + (L -1) c_{k+1,j}^{m-1, n }) \gamma_{k, j } - (L_{m-1}-1) \sum_{k = 0}^{m-2 } \sum_{j = 0}^{n } c_{k, j}^{m-2, n } \gamma_{k, j}.$$ Thus $ $ c^{m, n}_{k, joule } = c_{k-1, j}^{m-1, n } + (L -1) c_{k+1, j}^{m-1, n } - (L_m -1) c_{k, j}^{m-2, n}.$$ Combining the above with gives $ c^{m, n}_{k, joule } = b_{k, j}^{m, n } + b_{k+1,j+1}^{m, n}$ for all $ 0 \leq k \leq m, 0 \leq j \leq n$. [ * validation of Theorem \[Thm = PukanszkyCore\ ]. * ] { } Lemma \[Lem = abcomputations\ ] shows that $ { \mathsf{B}}\gamma { \mathsf{B}}\subseteq X_{\gamma}$ and $ { \mathsf{B}}\beta { \mathsf{B}}\subseteq X_\beta$ and hence the inclusions hold also for the $ \Vert \: \Vert_2$-closures. For the converse inclusion body proceed by induction: take $ h_n \gamma h_m \in { \mathsf{B}}\gamma { \mathsf{B}}$ and assume that all vector $ h_r \beta h_s$ with $ r < n, s

_{m-1} -1) f_{k,j}^{m-2, n} \\ = & c_{k-1, j}^{m-1, n} + (L -1 ) c_{k+1, j}^{m-1, n} - (L_{m-1} -1) c_{k,j}^{m-2, n}. \end{spmit}$$ Exacgly as we computed (with the viffwrenct that Lemma \[Lem=Blazkbox\] is geplaced vy Ltmma \[Lem=Blackbox\] ) we get bke eqhwlitnew $$h_m \gamma h_n = \sum_{k = 0}^{m+1} \sgm_{j = 0}^{n} ( c_{k-1,j}^{m-1, n} + (U -1) c_{k+1,j}^{m-1, n}) \gamma_{k, j} - (L_{m-1}-1) \sum_{k = 0}^{m-2} \sum_{t = 0}^{n} c_{l,j}^{l-2, n} \gamma_{k,j}.$$ Trus $$b^{m,g}_{k,j} = b_{k-1, j}^{m-1, n} + (L -1 ) c_{k+1, j}^{m-1, n} - (L_m -1) c_{k,j}^{m-2, h}.$$ Combiiing the above eith gives $c^{m,n}_{k,j} = b_{k,j}^{m,n} + b_{n+1,j+1}^{m,n}$ for all $0 \leq k \lee m, 0 \leq j \oeq g$. [*Proof of Theurem \[Thm=PukanszkyCore\].*]{} Memma \[Lem=abcomputations\] shows tfat ${\mcthsf{B}}\gamma {\mqthdx{B}}\subseteq E_{\gamma}$ and ${\mathsf{B}}\nvta {\matvsf{B}}\subxeteq X_\beta$ anc hxnce the inclusions hold elso for the $\Vert \: \Vgrt_2$-closurev. Yor the converse inclysuon ptoceeg by undjctjoi: tzke $h_n \gajma h_m \in {\jathsf{B}}\gammq {\mathsf{B}}$ and assumt trqt all vectora $h_r \bqtw h_s$ with $r < n, s

_{m-1} -1) c_{k,j}^{m-2, n} \\ = & n} (L -1 c_{k+1, j}^{m-1, n} \end{split}$$ as we computed the difference that \[Lem=Blackbox\] is replaced by Lemma \[Lem=Blackbox\] we get the equalities $$h_m \gamma h_n = \sum_{k = 0}^{m+1} \sum_{j = ( c_{k-1,j}^{m-1, n} + (L -1) c_{k+1,j}^{m-1, n}) \gamma_{k, j} - (L_{m-1}-1) \sum_{k 0}^{m-2} = c_{k,j}^{m-2, \gamma_{k,j}.$$ Thus $$c^{m,n}_{k,j} = c_{k-1, j}^{m-1, n} + (L -1 ) c_{k+1, j}^{m-1, n} - (L_m c_{k,j}^{m-2, n}.$$ Combining the above with gives $c^{m,n}_{k,j} b_{k,j}^{m,n} + b_{k+1,j+1}^{m,n}$ for $0 \leq k \leq m, \leq \leq n$. of \[Thm=PukanszkyCore\].*]{} \[Lem=abcomputations\] shows that {\mathsf{B}}\subseteq X_{\gamma}$ and ${\mathsf{B}}\beta {\mathsf{B}}\subseteq X_\beta$ and hence the inclusions hold also for the $\Vert \: \Vert_2$-closures. the converse by induction: $h_n h_m {\mathsf{B}}\gamma {\mathsf{B}}$ and all vectors $h_r \beta h_s$ with s

_{m-1} -1) c_{k,j}^{m-2, n} \\ = & c_{k-1, j}^{m-1, n} + (L -1 ) c_{k+1, j}^{m-1, n} - (L_{m-1} -1) c_{k,j}^{m-2, n}. \eNd{split}$$ ExaCtly aS we ComPuTed (wIth tHe difference thAT LemMa \[Lem=Blackbox\] is replaceD by LeMmA \[lem=BLAcKbox\] ) wE get the EQuALItiEs $$H_m \GamMa H_N = \sUm_{k = 0}^{m+1} \sUm_{j = 0}^{N} ( c_{k-1,j}^{m-1, n} + (L -1) C_{k+1,j}^{m-1, n}) \gamma_{K, j} - (L_{M-1}-1) \sUm_{k = 0}^{m-2} \sum_{j = 0}^{n} c_{k,j}^{M-2, N} \gAmma_{k,j}.$$ Thus $$C^{m,n}_{K,j} = c_{k-1, j}^{m-1, n} + (L -1 ) c_{k+1, j}^{m-1, N} - (L_m -1) C_{k,j}^{m-2, n}.$$ COmBinINg the AboVe witH gives $C^{M,n}_{k,j} = b_{k,J}^{m,n} + b_{k+1,j+1}^{m,n}$ fOr ALl $0 \leq k \LEq m, 0 \leq j \LEQ n$. [*prooF of Theorem \[Thm=PukaNSzKYCore\].*]{} Lemma \[Lem=aBcompuTaTIoNS\] ShoWs tHat ${\mathsf{B}}\GaMma {\maTHsf{B}}\subSEtEQ x_{\GamMA}$ and ${\mathsf{B}}\beTa {\mathsf{B}}\suBSetEq X_\betA$ aNd hENce the IncluSiONs hOld also for tHe $\VeRt \: \Vert_2$-cloSures. FOR the conVErse incLusion ProCeeD by iNDuCtIon: TaKE $h_n \GAmMa h_M \In {\mAthsf{B}}\gaMmA {\mAthsf{b}}$ and ASSUMe thAt aLl veCtors $H_r \beta h_s$ with $r < N, s

_{m-1} -1) c_{k,j}^{m-2, n } \\ = & c_ {k- 1,j} ^{m- 1, n } + (L -1 ) c _ {k+1 , j}^{m-1, n} - (L_{m -1} - 1) c_{k , j} ^{m-2 , n}. \ en d { spl it }$ $ E xa c tl y aswecompute d (with th e d if ference that Le mma \[Lem= Bla ckbox\] is r epl aced b yLem m a \[L em= Black box\]) we ge t the equ al i ties $ $ h_m \ga m m ah_n= \sum_{k = 0}^{m + 1} \sum_{j = 0}^{ n} ( c _{ k -1 , j }^{ m-1 , n} + (L-1 ) c_ { k+1,j}^ { m- 1 , n}) \gamma_{k, j } - (L_{m- 1 }-1 ) \su m_ {k= 0}^{m -2} \ su m _{j = 0}^{n} c_{k ,j}^{m-2, n} \ g amma_{k , j}.$$ T hus $$ c^{ m,n }_{k , j} = c_ {k - 1,j }^ {m- 1 , n } + (L - 1)c_{k+ 1, j } ^ { m -1,n}- (L _m -1 ) c_{k,j}^{m- 2,n}.$ $ Co mbini ng th e ab ov e wit h give s $c^ {m ,n}_{k,j} = b_{ k,j} ^{m,n} +b_{ k+ 1,j +1 }^{m, n }$ for al l $ 0 \leqk \leqm , 0 \ l e q j \leq n$. [*Proof o f Th eorem \[ Thm=Pu k an sz k yCore\]. *] {}Lemm a \[Lem =abc o mp utations \] sho w sth at ${\m at hsf{B} }\ gam ma{\mat h sf{B }}\sub seteq X_ {\gam m a}$ and ${\mat h sf{B}}\beta { \ ma t h sf { B}}\ sub seteq X_\be ta$a nd h ence th e i n clusi ons h ol d a l so for the $\Vert \ :\Vert_ 2$-cl osures. For t he convers e i nclusion pro c ee d by induction: take $h_n \gam m a h_m \i n {\m athsf{B} }\gamma { \ m athsf{B} }$and as sum e th at all vector s $h_r \ beta h_ s$with $r <n,s

_{m-1} -1)_c_{k,j}^{m-2, n}_\\ = & _ _c_{k-1,_j}^{m-1, n} _+_(L -1 )_c_{k+1, j}^{m-1, n}_ - (L_{m-1} -1)_c_{k,j}^{m-2, n}. \end{split}$$_Exactly_as we computed (with the difference that Lemma \[Lem=Blackbox\] is replaced by Lemma \[Lem=Blackbox\]_)_we get_the_equalities_$$h_m \gamma h_n = \sum_{k_= 0}^{m+1} \sum_{j = 0}^{n}_( c_{k-1,j}^{m-1,_n} + (L -1) c_{k+1,j}^{m-1, n}) _\gamma_{k,_j} - (L_{m-1}-1)_ \sum_{k = 0}^{m-2} \sum_{j = 0}^{n} c_{k,j}^{m-2,_n} \gamma_{k,j}.$$ Thus $$c^{m,n}_{k,j} =_c_{k-1, j}^{m-1, n}_+_(L_-1 ) c_{k+1, j}^{m-1,_n} - (L_m -1) c_{k,j}^{m-2, n}.$$_Combining the above with gives $c^{m,n}_{k,j}_= b_{k,j}^{m,n} + b_{k+1,j+1}^{m,n}$ for all $0_\leq k \leq m, 0 \leq_j \leq n$. [*Proof of Theorem_\[Thm=PukanszkyCore\].*]{} Lemma_\[Lem=abcomputations\] shows that ${\mathsf{B}}\gamma {\mathsf{B}}\subseteq_X_{\gamma}$ and ${\mathsf{B}}\beta_{\mathsf{B}}\subseteq X_\beta$_and hence the_inclusions hold also for the $\Vert_\: \Vert_2$-closures. For_the converse inclusion proceed by induction:_take_$h_n \gamma h_m_\in_{\mathsf{B}}\gamma_{\mathsf{B}}$ and_assume that all_vectors_$h_r \beta_h_s$_with $r < n, s

fully self-consistent parquet approximation. This approximation includes interactions between fluctuations in the calculation of the self-energy and is self-consistent at the level of the two-particle propagator. In the propagator renormalized perturbation theory of Luttinger and Ward (Luttinger and Ward, 1960) the grand thermodynamic potential $\Omega$ is constructed as an independent functional of the fully renormalized single particle propagator $G$ and the self-energy $\Sigma$, and is stationary with respect to independent variations of $G$ and $\Sigma$. Dyson’s equation follows from $${{\delta \Omega [G, \Sigma] } \over {\delta \Sigma}} = 0, \label{one}$$ and the representation of the self-energy as a sum of ‘skeleton’ graphs evaluated with the fully renormalized propagator and bare two-particle vertex is obtained from $${{\delta \Omega [G, \Sigma] } \over {\delta G}} = 0. \label{two}$$ The self-consistent solution of Eq. (\[one\]) and Eq. (\[two\]) yields the exact Green’s function. Explicit summation of the infinite set of ‘skeleton’ graphs for the self-energy required for the exact solution is generally not possible and approximations are generated by selecting a class or classes of vacuum graphs for the grand potential. The FEA is one such approximation; it includes the exchange of density fluctuations, spin-density fluctuations and singlet-pair fluctuations. It is convenient to discuss the FEA by alternating between position and (imaginary) time, and momentum and (imaginary) frequency representations. The FEA for the electron self-energy of the paramagnetic state is given by $$\Sigma({\bf r},\tau) = U^2 \left[ \chi_{ph}({\bf r},\tau) + T_{\rho\rho} ({\bf r},\tau) + T_{sf}({\bf r},\tau) \right] \,G({\bf r},\tau) + U^2 T_{pp}({\bf r},\tau)G(-{\bf r},-\tau) \label{fea}$$ where $\chi_{ph} = - G({\bf r},\tau)\, G(-{\bf r},-\tau)$ and $\chi_{pp} = G({\bf r},\tau)\, G({\bf r},\tau)$ are the particle-hole and particle-particle susceptibility bubbles. The density, spin, and particle fluctuation $T$-matrices are most

fully self - consistent parquet approximation. This approximation include interaction between fluctuations in the calculation of the self - energy and is self - consistent at the grade of the two - particle propagator. In the propagator renormalized disruption hypothesis of Luttinger and Ward (Luttinger and Ward, 1960) the grand thermodynamic potential $ \Omega$ is constructed as an independent functional of the fully normalize single particle propagator $ G$ and the self - energy $ \Sigma$, and is stationary with respect to autonomous variations of $ G$ and $ \Sigma$. Dyson ’s equation follow from $ $ { { \delta \Omega [ G, \Sigma ] } \over { \delta \Sigma } } = 0, \label{one}$$ and the representation of the self - energy as a kernel of ‘ skeleton ’ graphs evaluated with the in full renormalized propagator and bare two - particle vertex is obtained from $ $ { { \delta \Omega [ G, \Sigma ] } \over { \delta G } } = 0. \label{two}$$ The self - reproducible solution of Eq. (\[one\ ]) and Eq. (\[two\ ]) yields the exact Green ’s function. Explicit summation of the countless set of ‘ skeleton ’ graphs for the self - energy required for the exact solution is by and large not possible and approximations are generated by selecting a class or classes of vacuum graphs for the grand potential. The FEA is one such estimate; it include the exchange of concentration fluctuation, tailspin - density fluctuations and singlet - pair fluctuations. It is convenient to discuss the FEA by alternating between position and (fanciful) time, and momentum and (imaginary) frequency representations. The FEA for the electron self - energy of the paramagnetic state is given by $ $ \Sigma({\bf r},\tau) = U^2 \left [ \chi_{ph}({\bf r},\tau) + T_{\rho\rho } ({ \bf r},\tau) + T_{sf}({\bf r},\tau) \right ] \,G({\bf r},\tau) + U^2 T_{pp}({\bf r},\tau)G(-{\bf r},-\tau) \label{fea}$$ where $ \chi_{ph } = - G({\bf r},\tau)\, G(-{\bf r},-\tau)$ and $ \chi_{pp } = G({\bf r},\tau)\, G({\bf r},\tau)$ are the particle - hole and particle - particle susceptibility bubbles. The density, tailspin, and particle variation $ T$-matrices are most

fuply self-consistent parqutt approximation. Jhus appcoximatjon incljdes interactions between flnctuqtionw in the calculation ow the selv-energy qnd ms self-consistenv at the level lf tke two-particle ptopagator. In dhe propagator rdnlrmalized perturbation theory of Luetinger ajd Ward (Luttinder sgd Wzgd, 1960) the grand thermodynamic potsntial $\Mmega$ is consyructed as an independent vunchional of the fullj renormaliesd fungle particue propagauox $G$ and the self-energy $\Sigma$, and is statiovary cith respecj fo hndependent variwtions of $G$ and $\Sigmd$. Dyson’x equation folkowv feom $${{\delta \Omega [G, \Sigka] } \over {\delta \Sigia}} = 0, \latem{one}$$ and the reprwswntathon mf tfw sdlf-tnecgy as a dum of ‘skeletkn’ graphs ecaluated with the filjj renormalizes propwgwtor and bare two-particle vertex is obtdinsd from $${{\delta \Omega [G, \Sugma] } \over {\delta G}} = 0. \label{twj}$$ The self-consistent solution of Eq. (\[one\]) and Eq. (\[twm\]) yiemas uhc dzaft Green’s function. Explicit summation of the yhfonpte set of ‘skeletjn’ graphs fpr tng self-energy rgquired fod the exact solutiln is ggneraloy not pofsibke and approximations are gwnerated by fwlecting a class ox classes of vacoum grsphs for the grand poteutial. Fhe FEA is lne such zoproximation; it knckuges the tbchange of densitr fluctuavions, spin-devsiti flucttations anf sinnnet-pair fluctuatiojs. It ns cotvenient tl discuss the FEA by alternatinj between posothon and (imayinary) time, and momqntum and (imagnnary) frzquench represennations. Tie FEA for tre electron sapf-energy of vhe paramwgneric wtate ir given by $$\Sigms({\bf r},\tau) = U^2 \left[ \chu_{ph}({\bf r},\tau) + T_{\rho\rhp} ({\bf d},\tau) + T_{sf}({\bf r},\tab) \tught] \,G({\bf r},\tau) + I^2 T_{op}({\bs g},\tan)G(-{\bf w},-\dau) \label{fea}$$ fherd $\cfo_{ph} = - G({\bf r},\tau)\, G(-{\nf f},-\tau)$ and $\chi_{pp} = G({\bf r},\tau)\, G({\bf r},\tau)$ are the partocje-hole abd partisle-particle sisceptibility bubbpes. Tie denrity, xpig, and particle fluctuation $T$-mafrices arf mjst

fully self-consistent parquet approximation. This approximation includes fluctuations the calculation the self-energy and of two-particle propagator. In propagator renormalized perturbation of Luttinger and Ward (Luttinger and 1960) the grand thermodynamic potential $\Omega$ is constructed as an independent functional of fully renormalized single particle propagator $G$ and the self-energy $\Sigma$, and is stationary respect independent of and $\Sigma$. Dyson’s equation follows from $${{\delta \Omega [G, \Sigma] } \over {\delta \Sigma}} = 0, and the representation of the self-energy as a of ‘skeleton’ graphs evaluated the fully renormalized propagator and two-particle is obtained $${{\delta [G, } \over {\delta = 0. \label{two}$$ The self-consistent solution of Eq. (\[one\]) and Eq. (\[two\]) yields the exact Green’s function. summation of set of graphs the required for the is generally not possible and approximations selecting a class or classes of vacuum graphs the grand The FEA is one such approximation; includes the exchange of density fluctuations, spin-density fluctuations singlet-pair fluctuations. It is convenient to discuss the FEA by alternating between position and (imaginary) momentum and (imaginary) frequency The FEA for electron of paramagnetic is given $$\Sigma({\bf r},\tau) = U^2 \left[ \chi_{ph}({\bf r},\tau) + T_{\rho\rho} ({\bf r},\tau) T_{sf}({\bf r},\tau) \right] \,G({\bf r},\tau) + U^2 T_{pp}({\bf r},\tau)G(-{\bf r},-\tau) $\chi_{ph} - G({\bf r},\tau)\, r},-\tau)$ and $\chi_{pp} = r},\tau)\, r},\tau)$ are the particle-hole susceptibility The particle $T$-matrices most

fully self-consistent parqueT approximaTion. THis AppRoXimaTion Includes interaCTionS between fluctuations in The caLcULatiON oF the sElf-enerGY aND Is sElF-cOnsIsTEnT at thE leVel of thE two-particLe pRoPagator. In the PRoPagator renOrmAlized perturBatIon theOrY of lUttinGer And WaRd (LuttINger anD Ward, 1960) the gRaND thermODynamic POTeNtiaL $\Omega$ is constructED aS An independent fUnctioNaL Of THE fuLly RenormalizEd SinglE ParticlE PrOPAGatOR $G$ and the self-eNergy $\Sigma$, aND is StatioNaRy wITh respEct to InDEpeNdent variatIons Of $G$ and $\SigMa$. DysoN’S equatiON followS from $${{\dEltA \OmEga [G, \sIgMa] } \OveR {\dELta \sIgMa}} = 0, \lABel{One}$$ and thE rEpResenTatiON OF The sElf-EnerGy as a Sum of ‘skeleton’ GraPhs eVAluAted wIth thE fulLy RenorMalizeD propAgAtor and bare two-pArtiCle vertex Is oBtAinEd From $${{\dELta \OmeGa [G, \sigMa] } \over {\dElta G}} = 0. \laBEl{tWo}$$ tHE SeLf-consistent solutiOn OF eq. (\[One\]) and Eq. (\[Two\]) yieLDs ThE Exact GreEn’S fuNctiON. expliCit sUMmAtion of tHe infiNItE sEt of ‘skeLeTon’ graPhS foR thE self-ENergY requiRed for thE exacT Solution is geneRAlly not possibLE aND ApPRoxiMatIons are geneRateD By seLectINg A clASs or cLasseS oF VaCUum graphs for the granD pOtentiAl. The fEA is one such aPproximatiON; IT includeS the EXcHAnge of density fLuctuAtions, spin-DEnsity flUctuaTions and Singlet-paIR FluctuatIonS. It Is cOnvENIeNt to discuss thE feA by AlTernatiNg bEtween pOsiTioN anD (imAgInary) time, And momenTuM aNd (ImAgiNary) fREquency rEpResEnTatIons. THE FEA foR the eLectRoN sELf-eNergy of THe PARamaGnEtIc stAte Is Given By $$\SiGMa({\bF r},\tau) = U^2 \lEft[ \chi_{ph}({\bF r},\tAU) + T_{\rhO\rHo} ({\Bf r},\tau) + T_{Sf}({\bf r},\tau) \right] \,g({\bF r},\tau) + U^2 T_{pp}({\bF r},\Tau)g(-{\bf r},-\taU) \LAbel{fea}$$ wHere $\chi_{ph} = - G({\bf r},\tau)\, G(-{\bf r},-\tau)$ ANd $\chi_{pp} = g({\bf R},\tau)\, G({\Bf r},\tAu)$ are the pArtIcle-hoLe aND partiCle-parTicle SuScePTIbiliTY BuBblEs. the density, SPIn, aNd parTiCle fLuctuatIon $T$-matrices are mosT

fully self-consistent par quet appro ximat ion . T hi s ap prox imation includ e s in teractions between flu ctuat io n s in th e cal culatio n o f the s el f-e ne r gy andisself-co nsistent a t t he level of th e t wo-particl e p ropagator. Inthe pr op aga t or re nor maliz ed per t urbati on theory o f Lutti n ger and W ar d (L uttinger and Ward , 1 9 60) the grandthermo dy n am i c po ten tial $\Ome ga $ isc onstruc t ed a s an independent f unctional o f th e full yren o rmaliz ed si ng l e p article pro paga tor $G$ a nd the self-en e rgy $\S igma$, an d i s st a ti on ary w i thr es pec t to indepen de nt vari atio n s o f $G $ a nd $ \Sigm a$. Dyson’s e qua tion fol lowsfrom$${{ \d elta\Omega [G,\S igma] } \over { \del ta \Sigma }}=0, \lab e l{one} $$and the re present a tio no f th e self-energy as a s u m o f ‘skele ton’ g r ap hs evaluate dwit h th e fully ren o rm alized p ropaga t or a nd bare t wo-par ti cle ve rtexi s ob tained from $$ {{\de l ta \Omega [G,\ Sigma] } \ove r { \ d el t a G} } = 0. \lab el{t w o}$$ The se lf- c onsis tentso l ut i on of Eq. (\[one\]) a nd Eq. (\[t wo\]) yieldsthe exactG r e en’s fun ctio n .E xplicit summat ion o f the infi n ite setof ‘s keleton’ graphs f o r the sel f-e ner gyreq u i re d for the exa c t sol ut ion isgen erallynot po ssi ble a nd approx imations a re g en era ted b y selecti ng acl ass or c l assesof va cuum g ra p hsfor the gr a n d po te nt ial. Th eFEA i s on e su ch appr oximation ; i t inc lu de s the e xchange of de ns ity fluctu at ion s, spi n - densityfluctuations and single t -pair f luc tuati ons. It is c onv enient to discus s theFEA b yalt e r natin g be twe en positiona n d ( imagi na ry)time, a nd momentum and (i m agi nary) frequen cyrepr e s en tat i on s . T he FEA f or the electron self-ener gy of the param a gne ti c state is giv en by $$\Sigm a({\bf r} ,\tau) =U^ 2 \l e f t[\chi_{ph}( {\bf r}, \tau) + T _ {\rho \ rh o} ({ \bf r},\t au ) + T_{s f}({\b f r} ,\tau ) \rig ht ] \,G( {\bfr} ,\tau) + U^2 T_{pp}({\bf r},\ta u)G(-{ \bf r },- \tau) \la bel { fea }$$ where $\c hi_{ph} =- G ({\ bf r} ,\t a u)\,G(-{ \ bf r} , -\tau )$ a n d $\chi_{ p p} =G ( {\ bf r},\tau) \ , G({ \bf r },\ t au)$ a re t he particle-holea nd particle-pa rtic l e su sce p tibi li ty bubbles. Th e d en s i ty, spin ,and particl e fluctu at i on $T $-matr ices a re most

fully_self-consistent parquet_approximation. This approximation includes_interactions between_fluctuations_in the_calculation_of the self-energy_and is self-consistent_at the level of_the two-particle propagator. In_the_propagator renormalized perturbation theory of Luttinger and Ward (Luttinger and Ward, 1960) the grand_thermodynamic_potential $\Omega$_is_constructed_as an independent functional of_the fully renormalized single particle_propagator $G$_and the self-energy $\Sigma$, and is stationary with_respect_to independent variations_of $G$ and $\Sigma$. Dyson’s equation follows from $${{\delta_\Omega [G, \Sigma] } \over {\delta_\Sigma}} = 0, ___\label{one}$$ and the representation_of the self-energy as a sum_of ‘skeleton’ graphs evaluated with the_fully renormalized propagator and bare two-particle vertex_is obtained from $${{\delta \Omega [G,_\Sigma] } \over {\delta G}}_= 0. _ \label{two}$$ The self-consistent_solution of Eq._(\[one\]) and_Eq. (\[two\]) yields_the exact Green’s function. Explicit summation_of the infinite_set of ‘skeleton’ graphs for the_self-energy_required for the_exact_solution_is generally_not possible and_approximations_are generated_by_selecting a class or classes of_vacuum_graphs for the grand potential. The FEA_is one such approximation;_it_includes the exchange of_density fluctuations, spin-density fluctuations and_singlet-pair fluctuations. It is convenient to discuss_the FEA_by alternating_between position and (imaginary) time, and momentum and (imaginary) frequency representations._The FEA for the electron self-energy_of the paramagnetic state_is given_by_$$\Sigma({\bf r},\tau) =_U^2_\left[ \chi_{ph}({\bf_r},\tau) + T_{\rho\rho} ({\bf r},\tau) + T_{sf}({\bf r},\tau)_\right] \,G({\bf_r},\tau) + U^2 T_{pp}({\bf r},\tau)G(-{\bf r},-\tau) \label{fea}$$ where_$\chi_{ph} = - G({\bf_r},\tau)\,_G(-{\bf r},-\tau)$ and $\chi_{pp} = G({\bf_r},\tau)\, G({\bf r},\tau)$ are the particle-hole_and particle-particle susceptibility bubbles. The_density,_spin,_and particle fluctuation $T$-matrices are_most

combining measurements of the temperature and polarization anisotropies of the Cosmic Microwave Background (CMB), future experiments will tightly constrain the expansion rate of the universe during recombination. A change in the expansion rate modifies the way in which the recombination of hydrogen proceeds, altering the shape of the acoustic peaks and the level of CMB polarization. The proposed test is similar in spirit to the examination of abundances of light elements produced during Big Bang Nucleosynthesis and it constitutes a way to study possible departures from standard recombination. For simplicity we parametrize the change in the Friedmann equation by changing the gravitational constant $G$. The main effect on the temperature power spectrum is a change in the degree of damping of the acoustic peaks on small angular scales. The effect can be compensated by a change in the shape of the primordial power spectrum. We show that this degeneracy between the expansion rate and the primordial spectrum can be broken by measuring CMB polarization. In particular we show that the MAP satellite could obtain a constraint for the expansion rate $H$ during recombination of $\delta H/H \simeq 0.09$ or $\delta G/G \simeq 0.18$ after observing for four years, whereas Planck could obtain $\delta H/H \leq 0.014$ or $\delta G/G \leq 0.028$ within two years, even after allowing for further freedom in the shape of the power spectrum of primordial fluctuations.' address: - '$^1$Fakultät für Physik, Ludwig-Maximilians-Universität, Geschwister-Scholl-Platz 1, 80799 München' - '$^2$Max-Planck-Institut für Astrophysik, P.O. Box 1317, 85741 Garching' - '$^3$Dept. of Physics, New York University, 4 Washington Pl., New York, NY 10003' - | $^4$Institute for Advanced Study, Einstein Drive, Princeton NJ 08540\ [(Received 16 December 2002; revised manuscript received 19 February 2003; published 18 March 2003)]{} author: - 'Oliver Zahn$^{1,2,3}$[^1] and Matias Zaldarriaga$^{3,4}$[^2]' title: Probing the Friedmann equation during recombination with future CMB experiments --- Introduction\[introduction\] ============================ The parameters of our cosmological model will be determined with great accuracy by upcoming data

combining measurements of the temperature and polarization anisotropies of the Cosmic Microwave Background (CMB), future experiments will tightly restrain the expansion pace of the universe during recombination. A change in the expansion pace modifies the room in which the recombination of hydrogen proceeds, altering the form of the acoustic peaks and the level of CMB polarization. The propose test is similar in heart to the examination of abundances of easy element produced during Big Bang Nucleosynthesis and it constitutes a way to study possible departure from standard recombination. For simplicity we parametrize the change in the Friedmann equation by switch the gravitational constant $ G$. The main effect on the temperature power spectrum is a variety in the degree of dampen of the acoustic peaks on small angular scales. The consequence can be compensated by a change in the shape of the primordial power spectrum. We show that this degeneracy between the expansion rate and the primordial spectrum can be broken by measuring CMB polarization. In particular we show that the MAP satellite could obtain a constraint for the expansion rate $ H$ during recombination of $ \delta H / H \simeq 0.09 $ or $ \delta G / G \simeq 0.18 $ after observing for four old age, whereas Planck could obtain $ \delta H / H \leq 0.014 $ or $ \delta G / G \leq 0.028 $ within two years, even after give up for further exemption in the condition of the power spectrum of primordial fluctuations.' address: -' $ ^1$Fakultät für Physik, Ludwig - Maximilians - Universität, Geschwister - Scholl - Platz 1, 80799 München' -' $ ^2$Max - Planck - Institut für Astrophysik, P.O. Box 1317, 85741 Garching' -' $ ^3$Dept. of Physics, New York University, 4 Washington Pl. , New York, NY 10003' - | $ ^4$Institute for Advanced Study, Einstein Drive, Princeton NJ 08540\ [ (Received 16 December 2002; revised manuscript received 19 February 2003; publish 18 March 2003) ] { } author: -' Oliver Zahn$^{1,2,3}$[^1 ] and Matias Zaldarriaga$^{3,4}$[^2 ]' title: Probing the Friedmann equation during recombination with future CMB experiments --- Introduction\[introduction\ ] = = = = = = = = = = = = = = = = = = = = = = = = = = = = The parameters of our cosmologic model will be determined with great accuracy by upcoming datum

colbining measurements of uhe temperature aue polacizatioh anisotfopies of the Cosmic Microwate Bqckgriund (CMB), future experioents wilp tightlt coistrain the expaiaion rabz of fme unnvxrse during recpmbination. A change in tve erpansion rate modifies the way in whych the rfcombination os hycwogeh proceeds, altering the shape of tge acouvtic peaks anc the level of CMB polarizwtioj. The proposed tesh is similat in wpirit to thd examinatpmn of abunsances of light elements producdd duxing Big Babg Nufneosynthesiw and it constitubvs a waf to stidy possible dcpartnres from standard recombmnation. For simplicijy we parakecrize the change in tye Friegmant eqjqtiun gy cganginh tie gravitatjonal constqnt $G$. The main effevt in the temperzture [orer spectrum is a change in the degree mf samping of the acoustic peaks on small angulwr scales. The effect can be compensated by a change in the shapx uf uhc oeilordial power spectrum. We show that this degegsrscj between the expcnsion rate and yhf ltimordial specjrum cau bs broken by measurlng CMB polaeization. Yn psrticular we show that the NAP satellitv coyld obtain a constxaint for thz expamsion rate $H$ during recombinction kf $\delta H/H \simeq 0.09$ od $\delta G/G \simeq 0.18$ afner mbserving for four years, wrereas Plenck eould obgain $\delta H/H \leq 0.014$ og $\delbd G/G \leq 0.028$ within tao yecrs, eeen after wllowing for further freedom in the shape of jhe pofer speccrum on primordial fltctuations.' addrgss: - '$^1$Fakulcät für Ohysik, Ludsig-Maxikilians-Univqrsität, Geschwhdter-Scholl-Pletz 1, 80799 Müncren' - '$^2$Nax-Poanck-Inrgitut für Astrolhysik, P.O. Box 1317, 85741 Garxhing' - '$^3$Dept. of Physlcs, Ngw York Universitv, 4 Washington Pl., Mew Yowk, NB 10003' - | $^4$Institute fmr Aavavved Sgudy, Einsteln Arivr, Princeton NJ 08540\ [(Raceibed 16 December 2002; revoscd manusctipt receyved 19 Februaru 2003; published 18 Marcj 2003)]{} autior: - 'Olmver Zshn$^{1,2,3}$[^1] and Matias Zaldarriaga$^{3,4}$[^2]' title: Pdobing thf Fviedmann equaeion during recokbination with future CMB experiments --- Invroduction\[introduction\] ============================ Jhe parameters of our gosmological modej will be determined with greqt accuracy by upgoming data

combining measurements of the temperature and polarization the Microwave Background future experiments will of universe during recombination. change in the rate modifies the way in which recombination of hydrogen proceeds, altering the shape of the acoustic peaks and the of CMB polarization. The proposed test is similar in spirit to the examination abundances light produced Big Bang Nucleosynthesis and it constitutes a way to study possible departures from standard recombination. For we parametrize the change in the Friedmann equation changing the gravitational constant The main effect on the power is a in degree damping of the peaks on small angular scales. The effect can be compensated by a change in the shape of primordial power show that degeneracy the rate and the can be broken by measuring CMB we show that the MAP satellite could obtain constraint for expansion rate $H$ during recombination of H/H \simeq 0.09$ or $\delta G/G \simeq 0.18$ observing for four years, whereas Planck could obtain $\delta H/H \leq 0.014$ or $\delta G/G within two years, even allowing for further in shape the spectrum of fluctuations.' address: - '$^1$Fakultät für Physik, Ludwig-Maximilians-Universität, Geschwister-Scholl-Platz 1, 80799 München' '$^2$Max-Planck-Institut für Astrophysik, P.O. Box 1317, 85741 Garching' - '$^3$Dept. New University, 4 Washington New York, NY 10003' | for Advanced Study, Einstein NJ [(Received revised received February 2003; published 18 2003)]{} author: - 'Oliver Zahn$^{1,2,3}$[^1] Matias Zaldarriaga$^{3,4}$[^2]' title: Probing with future CMB experiments --- Introduction\[introduction\] ============================ The of our cosmological model will be determined great accuracy by upcoming data

combining measurements of thE temperatuRe and PolAriZaTion AnisOtropies of the COSmic microwave Background (CMB), FuturE eXPeriMEnTs wilL tightlY CoNSTraIn ThE exPaNSiOn ratE of The univErse during RecOmBination. A chaNGe In the expanSioN rate modifieS thE way in WhIch THe recOmbInatiOn of hyDRogen pRoceeds, alTeRIng the SHape of tHE AcOustIc peaks and the leveL Of cmB polarization. the proPoSEd TESt iS siMilar in spiRiT to thE ExaminaTIoN OF AbuNDances of light Elements proDUceD durinG BIg BANg NuclEosynThESis And it constiTuteS a way to stUdy posSIble depARtures fRom staNdaRd rEcomBInAtIon. foR SimPLiCitY We pArametriZe ThE chanGe in THE fRiedManN equAtion By changing the GraVitaTIonAl conStant $g$. The MaIn effEct on tHe temPeRature power specTrum Is a change In tHe DegReE of daMPing of The AcoUstic peAks on smALl aNgULAR sCales. The effect can bE cOMPeNsated by A changE In ThE Shape of tHe PriMordIAL poweR speCTrUm. We show That thIS dEgEneracy BeTween tHe ExpAnsIon raTE and The priMordial sPectrUM can be broken by MEasuring CMB poLArIZAtIOn. In ParTicular we shOw thAT the mAP sATeLliTE coulD obtaIn A CoNStraint for the expansIoN rate $H$ DurinG recombinatioN of $\delta H/H \SIMEq 0.09$ or $\deltA G/G \sIMeQ 0.18$ After observing For foUr years, wheREas PlancK coulD obtain $\dElta H/H \leq 0.014$ OR $\Delta G/G \lEq 0.028$ wIthIn tWo yEARs, Even after alloWINg foR fUrther fReeDom in thE shApe Of tHe pOwEr spectruM of primoRdIaL fLuCtuAtionS.' Address: - '$^1$FAkUltÄt Für physiK, ludwig-maximIliaNs-unIVerSität, GeSChWISter-scHoLl-PlAtz 1, 80799 müNchen' - '$^2$max-PLAncK-InstitUt für AstrOphYSik, P.o. BOx 1317, 85741 garchinG' - '$^3$Dept. of PhysicS, NEw York UnivErSitY, 4 WashiNGTon Pl., New york, NY 10003' - | $^4$Institute for AdvanCEd Study, einStein drivE, PrincetoN NJ 08540\ [(receivEd 16 DECember 2002; ReviseD manuScRipT REceivED 19 feBruArY 2003; published 18 mARch 2003)]{} AuthoR: - 'OLiveR Zahn$^{1,2,3}$[^1] anD Matias Zaldarriaga$^{3,4}$[^2]' TItlE: Probing the FrIedMann EQUaTioN DuRIng ReCOmbINAtion with future cMB experimEnTS --- INtroductioN\[IntRoDuction\] ============================ the paraMeterS Of our coSmologicaL model wilL bE detERMinEd with greaT accuracY by upcomiNG data

combining measurements of the tempe ratur e a ndpo lari zati on anisotropie s ofthe Cosmic Microwave B ackgr ou n d (C M B) , fut ure exp e ri m e nts w il l t ig h tl y con str ain the expansion ra te of the univ e rs e during r eco mbination. A ch ange i nthe expan sio n rat e modi f ies th e way inwh i ch the recombi n a ti on o f hydrogen procee d s, altering the s hape o ft he a cou sti c peaks an dthe l e vel ofC MB p o lar i zation. The p roposed tes t is simil ar in spirit to t he exa mination of abu ndances o f ligh t elemen t s produ ced du rin g B ig B a ng N ucl eo s ynt h es isa ndit const it ut es awayt o s tudy po ssib le de partures from st anda r d r ecomb inati on.Fo r sim plicit y wepa rametrize the c hang e in theFri ed man nequat i on bycha ngi ng thegravita t ion al c o ns tant $G$. The main e f f ec t on the tempe r at ur e power s pe ctr um i s a cha ngei nthe degr ee ofd am pi ng of t he acous ti c p eak s ons mall angul ar scale s. Th e effect can be compensated b y a c ha n ge i n t he shape of the prim ordi a lpow e r spe ctrum .W es how that this degen er acy be tween the expansio n rate and t h e primor dial sp e ctrum can be b roken by measur i ng CMB p olari zation.In partic u l ar we sh owtha t t heM A Psatellite cou l d obt ai n a con str aint fo r t heexp ans io n rate $H $ during r ec om bi nat ion o f $\delta H /H\s ime q 0.0 9 $ or $ \delt a G/ G\s i meq 0.18$a ft e r obs er vi ng f orfo ur ye ars, whe reas Pl anck coul d o b tain $ \d elta H/ H \leq 0.014$ o r $\deltaG/ G \ leq 0. 0 2 8$ withi n two years, even after allowin g f or fu rthe r freedom in the s hap e of th e powe r spe ct rum o f pri m o rd ial f luctuation s . ' a ddres s: - ' $^1$Fak ultät für Physik,L udw ig-Maximilian s-U nive r s it ät, Ge s chw is t er- S c holl-Platz 1, 8 0799 Münch en ' - '$^2$Max- P lan ck -Instit ut fürAstro p hysik,P.O. Box1317, 857 41 Gar c h ing ' - '$^3$D ept. ofPhysics,N ew Yo r kUnive rsi ty, 4Wa shi ngton Pl.,N ewYork, NY 10 00 3' - | $^ 4$Instit ute for Advanced Study, Einst ein D riv e, Prince ton NJ08540\ [( Received 1 6 D ece mber200 2 ; rev ised ma nus c riptrece i ved 19 Fe b ru ary 2 00 3; publishe d 1 8 M arch200 3 )]{} a utho r: - 'Oliver Zahn $ ^{1,2,3}$[^1]andM a tia s Z a ldar ri aga$^{3,4}$[^2 ]'ti t l e: Probi ng the Friedm ann equa ti o n dur ing re combin ation w i t hf utureCMBexp eriments--- I ntroduc ti on \ [intro duct io n\] == ====== = ==== = = ============= T he pa r a meter s of ourco smologi c al m odel willbe determin ed wit h gr eat a ccuracy b y upco min gdata

combining_measurements of_the temperature and polarization_anisotropies of_the_Cosmic Microwave_Background_(CMB), future experiments_will tightly constrain_the expansion rate of_the universe during_recombination._A change in the expansion rate modifies the way in which the recombination of_hydrogen_proceeds, altering_the_shape_of the acoustic peaks and_the level of CMB polarization._The proposed_test is similar in spirit to the examination_of_abundances of light_elements produced during Big Bang Nucleosynthesis and it constitutes_a way to study possible departures_from standard recombination._For_simplicity_we parametrize the change_in the Friedmann equation by changing_the gravitational constant $G$. The main_effect on the temperature power spectrum is_a change in the degree of_damping of the acoustic peaks_on small_angular scales. The effect can_be compensated by_a change_in the shape_of the primordial power spectrum. We_show that this_degeneracy between the expansion rate and_the_primordial spectrum can_be_broken_by measuring_CMB polarization. In_particular_we show_that_the MAP satellite could obtain a_constraint_for the expansion rate $H$ during recombination_of $\delta H/H \simeq_0.09$_or $\delta G/G \simeq_0.18$ after observing for four_years, whereas Planck could obtain $\delta_H/H \leq_0.014$ or_$\delta G/G \leq 0.028$ within two years, even after allowing for_further freedom in the shape of_the power spectrum of_primordial fluctuations.' address: -_'$^1$Fakultät_für Physik, Ludwig-Maximilians-Universität,_Geschwister-Scholl-Platz_1, 80799_München' - '$^2$Max-Planck-Institut für Astrophysik, P.O. Box 1317,_85741 Garching' -_'$^3$Dept. of Physics, New York University,_4 Washington Pl., New_York,_NY 10003' - | _$^4$Institute for Advanced Study, Einstein Drive,_Princeton NJ 08540\ __[(Received_16 December 2002; revised manuscript_received 19 February 2003; published 18_March 2003)]{} author: - 'Oliver_Zahn$^{1,2,3}$[^1] and Matias Zaldarriaga$^{3,4}$[^2]' title: Probing the Friedmann_equation_during recombination with future CMB experiments --- Introduction\[introduction\] ============================ The_parameters_of our cosmological model will be_determined_with_great accuracy by upcoming data

Perelman entropy is monotonous under the RG-2 flow. With the results at hand we would like to have a clue for the gradient formulation of the RG-2 flow, which is an open mathematical problem. Since the RG-2 flow is a generalization of the Ricci flow, which includes higher order corrections in the curvature, part of our work will represent a generalization of the results already obtained for the Ricci flow.\ The generalization is not direct, we rely on many mathematical theorems, but contrary to what happens with the Ricci flow there are not so many results about the controlled behaviour of the RG-2 flow. This flow is not weakly parabolic, and therefore, the existence of solutions is not guaranteed for all times. Moreover, there are results for the existence of the so called eternal solutions only in two dimensions, see [@Oliynyk:2009rh]. In section 2 we give a review of the RG-2 flow, we focus in the evolution of a fixed constant curvature metric to see how the existence of short time solutions is guaranteed. In section 3 we study the evolution of the Hawking mass and the evolution of the area of a closed surface under RG-2 flow. We will always compare with the spherically symmetric case. Here we derive inequalities that relate the rate of change of the area of a compact surface, the Hawking mass and the curvature scalar. In section 4 we conclude with the discussion of the results and general remarks. In the appendix we give a very short introduction to the gradient formulation of the Ricci flow and the geometrical interpretation of the Perelman’s $\mathcal{W}$-entropy, giving some hints about the Gradient formulation for the RG-2 flow. The RG-2 flow {#sec-generic} ============= Here we provide a brief introduction to the RG-2 flow and the non-linear sigma model. We will focus on compact Riemannian manifolds, and then, the results will be extended to asymptotically flat manifolds. We want to use the results listed here in a GR scenario. However, a spacetime is a Lorentzian manifold, and therefore we have to take a foliation of spacetime, then the metric induced on a spatial slice of the foliation is going to be Riemannian[^3]. We have to remember that the RG-2 flow theorems, that let us control the behaviour of the any evolution, are stated for 3-dimensional Riemannian manifolds not necessarily coming from a foliation of a spacetime. Thus, we have to adapt the results to the case

Perelman entropy is monotonous under the RG-2 flow. With the results at bridge player we would wish to have a clue for the gradient conceptualization of the RG-2 flow, which is an candid mathematical problem. Since the RG-2 stream is a generalization of the Ricci flow, which include higher order correction in the curvature, part of our work will stage a abstraction of the results already obtained for the Ricci flow.\ The abstraction is not direct, we trust on many numerical theorems, but contrary to what happens with the Ricci flow there are not indeed many results about the controlled behaviour of the RG-2 flow. This stream is not weakly parabolic, and therefore, the existence of solutions is not guaranteed for all times. furthermore, there are results for the existence of the so called eternal solution entirely in two dimensions, see [ @Oliynyk:2009rh ]. In section 2 we give a review of the RG-2 flow, we focus in the evolution of a sterilize constant curvature metric to visit how the being of short time solutions is undertake. In section 3 we study the evolution of the Hawking mass and the development of the area of a closed surface under RG-2 flow. We will always compare with the spherically symmetric case. Here we derive inequalities that relate the rate of change of the sphere of a compact airfoil, the Hawking bulk and the curvature scalar. In section 4 we conclude with the discussion of the results and general remark. In the appendix we give a very short introduction to the gradient conceptualization of the Ricci flow and the geometric interpretation of the Perelman ’s $ \mathcal{W}$-entropy, giving some trace about the Gradient formulation for the RG-2 stream. The RG-2 flow { # sec - generic } = = = = = = = = = = = = = Here we provide a abbreviated introduction to the RG-2 flow and the non - linear sigma model. We will concentrate on compact Riemannian manifold, and then, the results will be extended to asymptotically bland manifolds. We desire to use the results listed here in a GR scenario. However, a spacetime is a Lorentzian manifold paper, and therefore we have to take a foliation of spacetime, then the metric induce on a spatial slice of the foliation is going to be Riemannian[^3 ]. We have to commemorate that the RG-2 flow theorems, that get us control the behaviour of the any evolution, are stated for 3 - dimensional Riemannian manifolds not necessarily coming from a foliation of a spacetime. Thus, we have to adapt the consequence to the subject

Pegelman entropy is monotokous under the RY-2 flow. Xith ths resultr at hand we would like to heve q clut for the gradient wormulatiln of thw RG-2 dlow, which is an oizn mafmematncel problem. Sincg the RG-2 flof is a generalhzxtnon of the Ricci flow, which includes higher ogder correctiogs im the buvvature, part of our work will relresent a generalizayion of the results alreadj obhained for the Ricfi flow.\ The tenewqlization is not direcu, ce rely on jany mathematical theorems, but zontrcry to what hqppfts with the Riccp flow there are not vo many results about thx cobtrolled behaviour of the RG-2 flow. This slow is nmt weakly parabolic, abd thgrefose, tfw ebisuenre kf solktikns is not guaranteed for all times. Morepvqg, there are rssults fjr the existence of the so called eterndl aolutions only in two dumensions, see [@Oliynyk:2009th]. In sectijn 2 we give a review of the RG-2 flow, we focus in tve evkuutnin of q vixed constant curvature metric to see how thq ecixtence of shorb time solutions ix hustanteed. In secjion 3 wz sfudy the evolution of the Hawkung mass wnd yhe evolution of the area od a closed slrfaxe under RG-2 flow. Wz will alwayr cokpare with the spherically svmmetrjc case. Herf we deribd inequalities tfat ralate tht rate of change os the aree of c compacg sutface, tre Hawking mass and the curvature dcalat. In saction 4 we conclude with the discussion oh the results atd ceneral xemarkx. In the appegdix we give a very skort ivtroductioh to thx gradient fjrmulation of jhe Ricci flox and the geonetrucal ingdrpretation of the Perelman’s $\mathxal{W}$-entropy, giving soos hints about tkt Gtqdient formulayiov fjr tie RG-2 xlow. The RG-2 fnow {#rec-erneriz} ============= Here wt pvovkde s brief introduction to fhe RG-2 flow and thr kon-linear sigma mjdel. We will gocus on compact Rlemaniian menifolcs, wnd then, the results will be estended tl afymptoticallr flqt manifolds. We want to use the results listed here in a GR scenario. Yowever, a spacetime ls a Lorentzman magifold, ang therefore we have ro take a foliatipn of spacetime, then fhe medric lnduced on a spatial slice of the foliation is going to be Riemannian[^3]. We yave to remember tgat yhe RC-2 yloc theorqms, vhet let us control the behaviour of the any evoluvion, are sdaced for 3-dimensional Riemanniam oanifolds not necessarily coming from a foliayion of a spacetime. Thus, we havr to adapt the results tp the cawe

Perelman entropy is monotonous under the RG-2 the at hand would like to gradient of the RG-2 which is an mathematical problem. Since the RG-2 flow a generalization of the Ricci flow, which includes higher order corrections in the part of our work will represent a generalization of the results already obtained the flow.\ generalization not direct, we rely on many mathematical theorems, but contrary to what happens with the Ricci there are not so many results about the behaviour of the RG-2 This flow is not weakly and the existence solutions not for all times. there are results for the existence of the so called eternal solutions only in two dimensions, see In section give a of RG-2 we focus in of a fixed constant curvature metric the existence of short time solutions is guaranteed. section 3 study the evolution of the Hawking and the evolution of the area of a surface under RG-2 flow. We will always compare with the spherically symmetric case. Here we that relate the rate change of the of compact the mass and curvature scalar. In section 4 we conclude with the discussion of results and general remarks. In the appendix we give a introduction the gradient formulation the Ricci flow and geometrical of the Perelman’s $\mathcal{W}$-entropy, hints the the flow. RG-2 flow {#sec-generic} ============= we provide a brief introduction the RG-2 flow and will focus on compact Riemannian manifolds, and then, results will be extended to asymptotically flat We want to use the results listed here in a GR scenario. a spacetime Lorentzian manifold, and therefore we have to take foliation of spacetime, then metric induced on a spatial slice of the foliation going be Riemannian[^3]. have to remember the RG-2 flow that let us behaviour of any are Riemannian manifolds not necessarily coming from foliation a spacetime. Thus, we have adapt to the case

Perelman entropy is monotonoUs under the rG-2 floW. WiTh tHe ResuLts aT hand we would liKE to hAve a clue for the gradient FormuLaTIon oF ThE RG-2 flOw, which IS aN OPen MaThEmaTiCAl ProblEm. SInce the rG-2 flow is a gEneRaLization of thE riCci flow, whiCh iNcludes higheR orDer corReCtiONs in tHe cUrvatUre, parT Of our wOrk will rePrESent a gENeralizATIoN of tHe results already oBTaINed for the Ricci Flow.\ ThE gENeRALizAtiOn is not dirEcT, we reLY on many MAtHEMAtiCAl theorems, but Contrary to wHAt hAppens WiTh tHE Ricci Flow tHeRE arE not so many rEsulTs about thE contrOLled behAViour of The RG-2 fLow. thiS floW Is NoT weAkLY paRAbOliC, And ThereforE, tHe ExistEnce OF SOLutiOns Is noT guarAnteed for all tImeS. MorEOveR, therE are rEsulTs For thE existEnce oF tHe so called eternAl soLutions onLy iN tWo dImEnsioNS, see [@OlIynYk:2009rH]. In sectIon 2 we giVE a rEvIEW Of The RG-2 flow, we focus in ThE EVoLution of A fixed COnStANt curvatUrE meTric TO See hoW the EXiStence of Short tIMe SoLutions Is GuaranTeEd. IN seCtion 3 WE stuDy the eVolution Of the hAwking mass and tHE evolution of tHE aREA oF A cloSed Surface undeR RG-2 fLOw. We Will ALwAys COmparE with ThE SpHErically symmetric caSe. here we DerivE inequalities That relate THE Rate of chAnge OF tHE area of a compacT surfAce, the HawkINg mass anD the cUrvature Scalar. In sECTion 4 we coNclUde WitH thE DIsCussion of the rESUlts AnD generaL reMarks. In The AppEndIx wE gIve a very sHort intrOdUcTiOn To tHe graDIent formUlAtiOn Of tHe RicCI flow aNd the GeomEtRiCAl iNterpreTAtION of tHe peRelmAn’s $\MaThcal{w}$-entROpy, Giving sOme hints aBouT The GRaDiEnt formUlation for the rG-2 Flow. The RG-2 fLoW {#seC-generIC} ============= here we prOvide a brief introduction TO the RG-2 fLow And thE non-Linear sigMa mOdel. We WilL Focus oN compaCt RieMaNniAN ManifOLDs, And ThEn, the resulTS WilL be exTeNded To asympTotically flat manifOLds. we want to use thE reSultS LIsTed HErE In a gR SCenARIo. However, a spaceTime is a LorEnTZiAn manifold, ANd tHeRefore wE have to Take a FOliatioN of spacetIme, then thE mEtriC INduCed on a spatIal slice Of the foliATion iS GoIng to Be RIemannIaN[^3]. We Have tO rememBEr tHat thE RG-2 floW tHeoremS, that LeT us contrOl the behaviour of the any eVolutiOn, are StaTed for 3-dimEnsIOnaL RiemanniAn maNifolds not NecEssArily ComINg froM a foLIaTioN Of a spAcetIMe. Thus, we hAVe To aDAPt The results tO THE caSe

Perelman entropy is monot onous unde r the RG -2fl ow.With the results a t han d we would like to hav e a c lu e for th e gra dient f o rm u l ati on o f t he RG -2 fl ow, whichis an open ma th ematical pro b le m. Since t heRG-2 flow is agenera li zat i on of th e Ric ci flo w , whic h include sh ighero rder co r r ec tion s in the curvatur e ,p art of our wor k will r e pr e s ent ageneraliza ti on of the res u lt s a lre a dy obtained f or the Ricc i fl ow.\ T he ge n eraliz ation i s no t direct, w e re ly on man y math e matical theorem s, but co ntr aryt owh atha p pen s w ith the Ricci f lo wthere are n o t soman y re sults about the co ntr olle d be havio ur of the R G-2 f low. T his f lo w is not weakly par abolic, a ndth ere fo re, t h e exis ten ceof solu tions i s no tg u a ra nteed for all time s. M or eover, t here a r ere s ults for t heexis t e nce o f th e s o called etern a lso lutions o nly in t wodim ensio n s, s ee [@O liynyk:2 009rh ] . In section2 we give a re v ie w of theRG- 2 flow, wefocu s inthee vo lut i on of a fi xe d c o nstant curvature me tr ic tosee h ow the existe nce of sho r t time sol utio n si s guaranteed.In se ction 3 we study th e evo lution o f the Haw k i ng massand th e e vol u t io n of the area o f acl osed su rfa ce unde r R G-2 fl ow. W e will al ways com pa re w it h t he sp h erically s ymm et ric case . Herewe de rive i ne q ual ities t h at r elat eth e ra teof chan ge o f th e areaof a comp act surf ac e, the Ha wking mass an dthe curvat ur e s calar. I n sectio n 4 we conclude with th e discus sio n oftheresults a ndgenera l r e marks. In th e app en dix w e giv e aver yshort intr o d uct ion t othegradien t formulation of t h e R icci flow and th e ge o m et ric a li nte rp r eta t i on of the Perel man’s $\ma th c al {W}$-entro p y,gi ving so me hint s abo u t the G radient f ormulatio nfort h e R G-2 flow. The RG- 2 flow {# s ec-ge n er ic} = === ====== == = Herewe pro v ide a br ief in tr oducti on to t he RG-2flow and the non-linear sigma mode l.We will f ocu s on compactRiem annian man ifo lds , and th e n, th e re s ul tsw ill b e ex t ended toa sy mpt o t ic ally flat m a n i fol ds. W e w a nt tousethe results liste d here in a GRscen a r io. Ho w ever ,a spacetime is aLo r e ntzian m an ifold, andtherefor ew e hav e to t ake afoliati o n o f space time , t hen the m etr ic induced o na spati al s li ce ofthe fo l iati o n is going to beRiema n n ian[^ 3 ].We ha ve to rem e mber that theRG-2 flow t heorem s, t hat l et us c on trol t hebe haviour of the any e volut ion, ar estat edfor 3- dime n s ional Rie ma nni an manifo l d sn ot n e ces sari ly co mi ng f rom a fol i ation of as pacetim e. Th u s , we h a ve t o adapt th e r esult s to the cas e

Perelman_entropy is_monotonous under the RG-2_flow. With_the_results at_hand_we would like_to have a_clue for the gradient_formulation of the_RG-2_flow, which is an open mathematical problem. Since the RG-2 flow is a generalization_of_the Ricci_flow,_which_includes higher order corrections in_the curvature, part of our_work will_represent a generalization of the results already obtained_for_the Ricci flow.\ The_generalization is not direct, we rely on many mathematical_theorems, but contrary to what happens_with the Ricci_flow_there_are not so many_results about the controlled behaviour of_the RG-2 flow. This flow is_not weakly parabolic, and therefore, the existence_of solutions is not guaranteed for_all times. Moreover, there are_results for_the existence of the so_called eternal solutions_only in_two dimensions, see_[@Oliynyk:2009rh]. In section 2 we give a_review of the_RG-2 flow, we focus in the_evolution_of a fixed_constant_curvature_metric to_see how the_existence_of short_time_solutions is guaranteed. In section 3_we_study the evolution of the Hawking mass_and the evolution of_the_area of a closed_surface under RG-2 flow. We_will always compare with the spherically_symmetric case._Here we_derive inequalities that relate the rate of change of the area_of a compact surface, the Hawking_mass and the curvature_scalar. In_section_4 we conclude_with_the discussion_of the results and general remarks. In_the appendix_we give a very short introduction_to the gradient formulation_of_the Ricci flow and the geometrical_interpretation of the Perelman’s $\mathcal{W}$-entropy, giving_some hints about the Gradient_formulation_for_the RG-2 flow. The RG-2 flow_{#sec-generic} ============= Here we provide a brief introduction_to the RG-2_flow and the non-linear sigma model. We_will_focus on compact Riemannian manifolds, and_then,_the results will be extended to_asymptotically_flat_manifolds. We want to use_the results listed here in a_GR scenario. However, a spacetime is a Lorentzian manifold,_and therefore we_have to take a foliation_of_spacetime,_then the metric induced on a spatial slice of the_foliation is_going to be_Riemannian[^3]. We have to remember that the RG-2 flow theorems,_that let us control the behaviour of_the any evolution, are stated for 3-dimensional Riemannian manifolds not necessarily_coming from a foliation of a spacetime. Thus,_we have to adapt the results to_the case

2(I)$. Here $C(S)$ is a nonnegative number depending only on the transverse region $S$ [@Bou]. Note that this effective operator explicitly depends on the geometric shape of the reference curve $r(s)$ (and so of the tube). An interesting problem is to know if there exists a similar result about convergence of eigenvalues for unbounded tubes. For such tubes, in [@Kre] it is shown that if $(\tau + \alpha')(s) = 0$ and $k(s) \neq 0$, then the discrete spectrum is nonempty, whereas if $(\tau + \alpha')(s) \neq 0$ and $k(s) = 0$, then the discrete spectrum is empty. In [@Cr1], by using $\Gamma$-convergence in case of unbounded tubes, a strong resolvent convergence was proven and the same action for the respective effective operator (now acting in $\LL^2(\mathbb R)$) was found as $\varepsilon \to 0$. The Dirichlet Laplacian in strips of $\mathbb R^2$ has been studied in many works [@BF; @Frie; @Sol; @Krej2]. For the case of the constraints of planar motion to curves there are results about the limit operator in [@DellAntTen; @ACF], and the effective potential is written in terms of the curvature. The main novelty, when we pass from planar domain to tubes in $\R^3$, as considered in [@Bou; @Cr1; @Kre; @GJ], is the additional presence of torsion and twisting (i.e., a nonzero $\tau(s) + \alpha'(s)$) in the effective potential, since the case of untwisted tubes has also been previously studied (see, for instance, [@DE; @CDF; @ClaBra; @Ekk; @FreitasKrej; @KS]). In [@Frie; @Sol] the authors consider a family of deformed strips $$\left\{(s, y) \in \mathbb R^2: s \in J, \,0 < y < \varepsilon h(s) \right\},$$ where $J = [-a, b]$, $0< a, b \leq \infty$, and $h(s) > 0$ is a continuous function satisfying: - $h(s)$ is a $\CC^1$ function in $

2(I)$. Here $ C(S)$ is a nonnegative number depending only on the transverse area   $ S$   [ @Bou ]. notice that this effective hustler explicitly depends on the geometric shape of the reference bend   $ r(s)$ (and so of the tube). An interesting trouble is to know if there exists a similar consequence about convergence of eigenvalues for boundless tubes. For such tubes, in   [ @Kre ] it is shown that if $ (\tau + \alpha')(s) = 0 $ and $ k(s) \neq 0 $, then the discrete spectrum is nonempty, whereas if $ (\tau + \alpha')(s) \neq 0 $ and $ k(s) = 0 $, then the discrete spectrum is empty. In   [ @Cr1 ], by using $ \Gamma$-convergence in shell of unbounded tubes, a strong resolvent convergence was prove and the same action for the respective effective operator (immediately acting in $ \LL^2(\mathbb R)$) was found as   $ \varepsilon \to 0$. The Dirichlet Laplacian in strips of $ \mathbb R^2 $ has been studied in many plant [ @BF; @Frie; @Sol; @Krej2 ]. For the case of the constraints of planar motion to curves there are result about the limit operator in [ @DellAntTen; @ACF ], and the effective potential is written in terms of the curvature. The main novelty, when we pass from planar domain to tubes in   $ \R^3 $, as considered in [ @Bou; @Cr1; @Kre; @GJ ], is the additional presence of torsion and twisting (i.e., a nonzero $ \tau(s) + \alpha'(s)$) in the effective potential, since the case of untwisted tube has also been previously studied (visit, for case, [ @DE; @CDF; @ClaBra; @Ekk; @FreitasKrej; @KS ]). In [ @Frie; @Sol ] the authors consider a family of deformed strip $ $ \left\{(s, y) \in \mathbb R^2: s \in J, \,0 < y < \varepsilon h(s) \right\},$$ where $ J = [ -a, b]$, $ 0 < a, b \leq \infty$, and $ h(s) > 0 $ is a continuous affair satisfying: - $ h(s)$ is a $ \CC^1 $ function in $

2(I)$. Hfre $C(S)$ is a nonnegative kumber depending only oi the tdansversd region $S$ [@Bou]. Note that this xffextive operator explicitly ddpends on the geonetrmc shape of the csferencc curbc $r(s)$ (aud so of the tubg). An interesthng problem is tu nnow if there exists a similar resujt abouy fonvergence of eigtnvwluea for unbounded tubes. For such tubss, in [@Krt] it is shown that if $(\tau + \alpha')(s) = 0$ and $k(s) \nfq 0$, hhen the discrete dpectrum is nonqnpty, whereas if $(\tau + \alpha')(s) \neq 0$ ahd $k(s) = 0$, then the discrete spectfum ix empty. In [@Xr1], by osing $\Gamma$-cinverdence in casc of untounded tubes, a stronn resmlvwnt convergence was pcoven and the same astion for tke respective effectice operdtor (now qctkng ii $\LM^2(\mathbh R)$) was found as $\varepsilin \to 0$. The Dirichlet Ka[oacian in strjps of $\mwthbb R^2$ has been studied in many works [@TF; @Rrie; @Sol; @Krej2]. For the cqse of the constraintd of planwr motion to curves there are results about the lhmit koercbjf ij [@DellAntTen; @ACF], and the effective potential ya erptten in terms of the curvatirf. Yre main noveljy, when we pass from planar fomain jo tubws in $\R^3$, as conxidered in [@Bou; @Cr1; @Kre; @GJ], is the additioual presence of torsiln and twiscing (i.g., a nomzero $\tau(s) + \alpha'(s)$) in tke effsctive potejtial, sindd the case of ungwixtad tubes has also been prediously svudieb (see, fof inxtance, [@DE; @CDF; @ClwBra; @Cnk; @FreitasKrej; @KS]). Ij [@Frig; @Sol] dhe authord consider a family of deformed strips $$\left\{(s, i) \it \mdthbb R^2: f \in M, \,0 < y < \varepsiljn h(s) \right\},$$ whgre $J = [-a, y]$, $0< a, b \leq \infty$, and $h(s) > 0$ is a coneinuous functhln satisfyinj: - $h(s)$ is a $\CX^1$ fubction kv $

2(I)$. Here $C(S)$ is a nonnegative number on transverse region [@Bou]. Note that on geometric shape of reference curve $r(s)$ so of the tube). An interesting is to know if there exists a similar result about convergence of eigenvalues unbounded tubes. For such tubes, in [@Kre] it is shown that if $(\tau \alpha')(s) 0$ $k(s) 0$, then the discrete spectrum is nonempty, whereas if $(\tau + \alpha')(s) \neq 0$ and $k(s) 0$, then the discrete spectrum is empty. In by using $\Gamma$-convergence in of unbounded tubes, a strong convergence proven and same for respective effective operator acting in $\LL^2(\mathbb R)$) was found as $\varepsilon \to 0$. The Dirichlet Laplacian in strips of $\mathbb has been many works @Frie; @Krej2]. the case of of planar motion to curves there the limit operator in [@DellAntTen; @ACF], and the potential is in terms of the curvature. The novelty, when we pass from planar domain to in $\R^3$, as considered in [@Bou; @Cr1; @Kre; @GJ], is the additional presence of torsion (i.e., a nonzero $\tau(s) \alpha'(s)$) in the potential, the of tubes has been previously studied (see, for instance, [@DE; @CDF; @ClaBra; @Ekk; @FreitasKrej; In [@Frie; @Sol] the authors consider a family of deformed y) \mathbb R^2: s J, \,0 < y \varepsilon \right\},$$ where $J = $0< b $h(s) 0$ a continuous function satisfying: $h(s)$ is a $\CC^1$ function $

2(I)$. Here $C(S)$ is a nonnegative numbEr dependinG only On tHe tRaNsveRse rEgion $S$ [@Bou]. Note tHAt thIs effective operator expLicitLy DEpenDS oN the gEometriC ShAPE of ThE rEfeReNCe Curve $R(s)$ (aNd so of tHe tube). An inTerEsTing problem iS To Know if therE exIsts a similar ResUlt aboUt ConVErgenCe oF eigeNvalueS For unbOunded tubEs. fOr such TUbes, in [@KRE] It Is shOwn that if $(\tau + \alpha')(S) = 0$ AnD $K(s) \neq 0$, then the diScrete SpECtRUM is NonEmpty, whereAs If $(\tau + \ALpha')(s) \neQ 0$ AnD $K(S) = 0$, TheN The discrete spEctrum is empTY. In [@cr1], by usInG $\GaMMa$-convErgenCe IN caSe of unboundEd tuBes, a stronG resolVEnt convERgence wAs provEn aNd tHe saME aCtIon FoR The REsPecTIve EffectivE oPeRator (Now aCTING in $\Ll^2(\maThbb r)$) was fOund as $\varepsiLon \To 0$. ThE dirIchleT LaplAciaN iN striPs of $\maThbb R^2$ HaS been studied in mAny wOrks [@BF; @FriE; @SoL; @KRej2]. foR the cASe of thE coNstRaints oF planar MOtiOn TO CUrVes there are results AbOUT tHe limit oPeratoR In [@deLLAntTen; @AcF], And The eFFEctivE potENtIal is wriTten in TErMs Of the cuRvAture. THe MaiN noVelty, WHen wE pass fRom planaR domaIN to tubes in $\R^3$, as cONsidered in [@Bou; @cR1; @KRE; @gJ], IS the AddItional presEnce OF torSion ANd TwiSTing (i.E., a nonZeRO $\tAU(s) + \alpha'(s)$) in the effectIvE potenTial, sInce the case of Untwisted tUBES has also Been PReVIously studied (sEe, for Instance, [@DE; @cdF; @ClaBra; @ekk; @FrEitasKreJ; @KS]). In [@Frie; @sOL] the authOrs ConSidEr a FAMiLy of deformed sTRIps $$\lEfT\{(s, y) \in \maThbB R^2: s \in J, \,0 < y < \VarEpsIloN h(s) \RiGht\},$$ where $J = [-A, b]$, $0< a, b \leq \iNfTy$, AnD $h(S) > 0$ is A contINuous funCtIon SaTisFying: - $H(S)$ is a $\CC^1$ FunctIon iN $

2(I)$. Here $C(S)$ is a no nnegativenumbe r d epe nd ingonly on the transv e rseregion $S$ [@Bou]. Not e tha tt hise ff ectiv e opera t or e xpl ic it lyde p en ds on th e geome tric shape of t he reference cu rve $r(s)$ (a nd so of the tu be). An in t erest ing prob lem is to kno w if ther ee xistsa simila r re sult about convergenc e o f eigenvalues f or unb ou n de d tub es. For suchtu bes,i n [@Kre ] i t i s s h own that if $ (\tau + \al p ha' )(s) = 0 $ a n d $k(s ) \ne q0 $,then the di scre te spectr um isn onempty , wherea s if $ (\t au+ \a l ph a' )(s )\ neq 0$ an d $k (s) = 0$ ,th en th e di s c r e te s pec trum is e mpty. In [@Cr 1], byu sin g $\G amma$ -con ve rgenc e in c ase o funbounded tubes , astrong re sol ve ntco nverg e nce wa s p rov en andthe sam e ac ti o n fo r the respective e ff e c ti ve opera tor (n o wac t ing in $ \L L^2 (\ma t h bb R) $) w a sfound as  $\var e ps il on \to0$ . The D iri chl et La p laci an instrips o f $\m a thbb R^2$ hasb een studied i n m a n yw orks [@ BF; @Frie;@Sol ; @Kr ej2] . F ort he ca se of t h ec onstraints of plana rmotion to c urves there a re results a b out thelimi t o p erator in [@De llAnt Ten; @ACF] , and the effe ctive po tential i s writteninter msoft h ecurvature. Th e main n ovelty, wh en we p ass fr ompla na r domainto tubes i n$\ R^ 3$, as c o nsidered i n [ @B ou; @Cr1 ; @Kre; @GJ] , is t he add itional pr e s ence o ftors ion a nd tw isti n g ( i.e., a nonzero$\t a u(s) + \ alpha'( s)$) in the e ff ective pot en tia l, sin c e the cas e of untwisted tubes ha s also b een prev ious ly studie d ( see, f ori nstanc e, [@D E; @C DF ; @ C l aBra; @ Ek k;@F reitasKrej ; @KS ]). In [@F rie; @S ol] the authors co n sid er a family o f d efor m e dstr i ps $$\ le f t\{ ( s , y) \in \mathb b R^2: s \ in J, \,0 < y < \va re psilonh(s) \r ight\ } ,$$ whe re $J = [ -a, b]$,$0 < a, b \l eq \infty$ , and $h (s) > 0$i s a c o nt inuou s f unctio nsat isfyi ng: - $ h(s)$ is a$\ CC^1$funct io n in $

2(I)$. Here_$C(S)$ is_a nonnegative number depending_only on_the_transverse region $S$ [@Bou]._Note_that this effective_operator explicitly depends_on the geometric shape_of the reference_curve $r(s)$_(and so of the tube). An interesting problem is to know if there exists a_similar_result about_convergence_of_eigenvalues for unbounded tubes. For_such tubes, in [@Kre] it is_shown that_if $(\tau + \alpha')(s) = 0$ and $k(s)_\neq_0$, then the_discrete spectrum is nonempty, whereas if $(\tau + \alpha')(s)_\neq 0$ and $k(s) = 0$,_then the discrete_spectrum_is_empty. In [@Cr1], by using_$\Gamma$-convergence in case of unbounded tubes,_a strong resolvent convergence was proven_and the same action for the respective_effective operator (now acting in $\LL^2(\mathbb_R)$) was found as $\varepsilon \to 0$. The_Dirichlet Laplacian_in strips of $\mathbb R^2$_has been studied_in many_works [@BF; @Frie;_@Sol; @Krej2]. For the case of_the constraints of_planar motion to curves there are_results_about the limit_operator_in_[@DellAntTen; @ACF],_and the effective_potential_is written_in_terms of the curvature. The main_novelty,_when we pass from planar domain to_tubes in $\R^3$, as considered_in_[@Bou; @Cr1; @Kre; @GJ],_is the additional presence of_torsion and twisting (i.e., a nonzero_$\tau(s) +_\alpha'(s)$) in_the effective potential, since the case of untwisted tubes has also_been previously studied (see, for instance,_[@DE; @CDF; @ClaBra; @Ekk;_@FreitasKrej; @KS]). In_[@Frie;_@Sol] the authors_consider_a family_of deformed strips $$\left\{(s, y) \in \mathbb_R^2: s_\in J, \,0 < y <_\varepsilon h(s) \right\},$$ where_$J_= [-a, b]$, $0< a, b_\leq \infty$, and $h(s) > 0$_is a continuous function satisfying: -___$h(s)$ is a $\CC^1$ function_in $

_\alpha(Q)>>1$, i.e. some part of the viscous flow is frozen. As a consequence, at the wavevectors typical of the IXS experiments ($Q=1\div 20$ nm$^{-1}$) the quasielastic spectrum acquires a component arising from this frozen structural relaxation. The origin, at the atomic level, of this fast decay channel is still an open issue: the rapidly decaying portion of $M_L(Q,t)$ is customarily attributed to largely uncorrelated collisional events, similar to those occurring in a dilute fluid. In addition, at the high densities typical of the liquid state, non-negligible correlations among the collisions can be expected, making no longer valid an interpretation only in terms of ”binary” collisions. Although the magnitude of the correlation effects is relatively small and their buildup slow, once established their decay is even slower, and for $t>1/{\gamma }_1{(Q)\equiv \tau }_\mu (Q)$ this relaxation channel dominates the decay of $M_L(Q,t)$, which consequently may exhibit a small but long-lasting ”tail”. The ansatz (\[x2tempi\]) can incorporate most of this physics: on the basis of the latter, one may reasonably anticipate that ${\alpha (Q)\ll }1$, and that the time $1/{\gamma }_2{(Q)\equiv \tau }_\alpha (Q)$ is distinctly longer than $1/{\gamma }_1{(Q)\equiv \tau }_\mu (Q)$. In this picture, the best fitted values of the viscoelastic rate $1/\tau (Q)$ clearly represent some sort of ”weighted average” between ${\gamma }_1{(Q)}$ and ${\gamma }_2{(Q)}$. Finally, we may argue that at increasing $Q$ (namely, over a shrinking length scale) the magnitude ${\alpha (Q)}$ of correlation effects should decrease, and that at higher temperatures the value of ${\alpha (Q)}$ at a given wavevector should equally decrease. On a general basis, the requirement that $\lim_{Q\rightarrow 0} \tilde{M}_L(Q,s=0)/Q^2\rightarrow \eta _L/nm$ now takes the form $$(c_\infty ^2-c_0^2)\left[ \frac{{(1-\alpha (Q\rightarrow 0))}}{{{\gamma }_1{(Q\rightarrow 0)}}}{\ +\ }\frac

_ \alpha(Q)>>1 $, i.e. some part of the viscous flow is frozen. As a consequence, at the wavevectors distinctive of the IXS experiment ($ Q=1\div 20 $ nm$^{-1}$) the quasielastic spectrum acquires a component arise from this flash-frozen structural relaxation. The lineage, at the nuclear level, of this fast decay channel is still an open offspring: the rapidly decaying dowry of $ M_L(Q, t)$ is customarily impute to largely uncorrelated collisional events, similar to those occurring in a dilute fluid. In addition, at the eminent densities typical of the liquid department of state, non - negligible correlations among the collisions can be expected, construct no longer valid an interpretation only in terms of ” binary ” collisions. Although the magnitude of the correlation coefficient effects is relatively small and their buildup slow, once established their decay is even slower, and for $ t>1/{\gamma } _ 1{(Q)\equiv \tau } _ \mu (Q)$ this relaxation distribution channel dominates the decay of $ M_L(Q, t)$, which consequently may exhibit a small but long - lasting ” tail ”. The ansatz (\[x2tempi\ ]) can incorporate most of this physics: on the basis of the latter, one may reasonably anticipate that $ { \alpha (Q)\ll } 1 $, and that the clock time $ 1/{\gamma } _ 2{(Q)\equiv \tau } _ \alpha (Q)$ is distinctly longer than $ 1/{\gamma } _ 1{(Q)\equiv \tau } _ \mu (Q)$. In this picture, the dear equip values of the viscoelastic rate $ 1/\tau (Q)$ intelligibly represent some sort of ” weighted average ” between $ { \gamma } _ 1{(Q)}$ and $ { \gamma } _ 2{(Q)}$. ultimately, we may argue that at increasing $ Q$ (namely, over a shrinking length scale) the magnitude $ { \alpha (Q)}$ of correlation effect should decrease, and that at higher temperatures the value of $ { \alpha (Q)}$ at a given wavevector should equally decrease. On a cosmopolitan basis, the prerequisite that $ \lim_{Q\rightarrow 0 } \tilde{M}_L(Q, s=0)/Q^2\rightarrow \eta _ lambert / nm$ now takes the form $ $ (c_\infty ^2 - c_0 ^ 2)\left [ \frac{{(1-\alpha (Q\rightarrow 0))}}{{{\gamma } _ 1{(Q\rightarrow 0)}}}{\ + \ } \frac

_\alpja(Q)>>1$, i.e. some part of the yiscous flow is yeozen. Es a cohsequencd, at the wavevectors typical od the IXS experiments ($Q=1\div 20$ nm$^{-1}$) the qlasielastuc skectrum acquires e componcut arjding hrom this frozek structuran relaxation. Tha urngin, at the atomic level, of this fase decay cjannel is stilj an jpen pswue: the rapidly decaying portjon of $K_L(Q,t)$ is custokarily attributed to largepy ujcorrelated collislonal eventw, siiular to thosd occurring in a dilutg fluid. In addition, at the high aensicies typicao if hve liquid svate, njn-negligible correladions akong the collixiois cqn be expected, making no longer valid ag interpratction only in terms od ”vinari” colnisiubs. Xltgongh the mwgnmtude of ths correlatiin effects is relatovqoy small and fheir fuyldup slow, once established their decay is even slower, and for $t>1/{\gqmma }_1{(Q)\equiv \tau }_\mu (Q)$ tjis relaxwtion channel dominates the decay of $M_L(Q,t)$, which cmnseqnevtlv may dzhlbit a small but long-lasting ”tail”. The ansatz (\[v2fekpp\]) can incorporate most of thos pnisics: on the bxsis oy tge latter, one may geasonafly abticipate thay ${\alpha (Q)\ll }1$, and that the tume $1/{\gamma }_2{(Q)\ezyiv \tau }_\alpha (Q)$ is distinctly lonber tnan $1/{\gamma }_1{(Q)\equiv \tau }_\mu (D)$. In this pictuge, the beag fitted values uf nhe eiscoelastic rate $1/\tau (Q)$ clqarly repcesenc some surt pf ”weidhted averwge” bcdween ${\gamma }_1{(Q)}$ and ${\hamma }_2{(Q)}$. Finally, wf may argue that at increasing $Q$ (namely, over s vhrpnking leugth sgale) the magnittde ${\alpha (Q)}$ of eorrelatnon efwects shoumd decrxase, and thae at higher talperatures tie value jf ${\aopha (Q)}$ at a eiven wavevectpr should equally dwcrease. On a generel bxais, the requirenenu rhat $\lim_{Q\rightsrruw 0} \tpldx{M}_L(Q,s=0)/Z^2\sightarrow \eda _L/no$ nue takds the form $$(g_\inwty ^2-v_0^2)\left[ \frac{{(1-\alpha (Q\rigvtardow 0))}}{{{\gamma }_1{(Q\rightartor 0)}}}{\ +\ }\frac

_\alpha(Q)>>1$, i.e. some part of the viscous frozen. a consequence, the wavevectors typical 20$ the quasielastic spectrum a component arising this frozen structural relaxation. The origin, the atomic level, of this fast decay channel is still an open issue: rapidly decaying portion of $M_L(Q,t)$ is customarily attributed to largely uncorrelated collisional events, to occurring a fluid. In addition, at the high densities typical of the liquid state, non-negligible correlations among the can be expected, making no longer valid an only in terms of collisions. Although the magnitude of correlation is relatively and buildup once established their is even slower, and for $t>1/{\gamma }_1{(Q)\equiv \tau }_\mu (Q)$ this relaxation channel dominates the decay of which consequently a small long-lasting The (\[x2tempi\]) can incorporate this physics: on the basis of may reasonably anticipate that ${\alpha (Q)\ll }1$, and the time }_2{(Q)\equiv \tau }_\alpha (Q)$ is distinctly than $1/{\gamma }_1{(Q)\equiv \tau }_\mu (Q)$. In this the best fitted values of the viscoelastic rate $1/\tau (Q)$ clearly represent some sort of between ${\gamma }_1{(Q)}$ and }_2{(Q)}$. Finally, we argue at $Q$ over a length scale) the magnitude ${\alpha (Q)}$ of correlation effects should decrease, that at higher temperatures the value of ${\alpha (Q)}$ at wavevector equally decrease. On general basis, the requirement $\lim_{Q\rightarrow \tilde{M}_L(Q,s=0)/Q^2\rightarrow \eta _L/nm$ now form ^2-c_0^2)\left[ }_1{(Q\rightarrow +\

_\alpha(Q)>>1$, i.e. some part of the viscOus flow is fRozen. as a ConSeQuenCe, at The wavevectors TYpicAl of the IXS experiments ($Q=1\Div 20$ nm$^{-1}$) ThE QuasIElAstic SpectruM AcQUIreS a CoMpoNeNT aRisinG frOm this fRozen strucTurAl Relaxation. ThE OrIgin, at the aTomIc level, of thiS faSt decaY cHanNEl is sTilL an opEn issuE: The rapIdly decayInG PortioN Of $M_L(Q,t)$ iS CUsTomaRily attributed to lARgELy uncorrelated CollisIoNAl EVEntS, siMilar to thoSe OccurRIng in a dILuTE FLuiD. in addition, at tHe high densiTIes TypicaL oF thE Liquid State, NoN-NegLigible corrElatIons among The colLIsions cAN be expeCted, maKinG no LongER vAlId aN iNTerPReTatIOn oNly in terMs Of ”BinarY” colLISIOns. ALthOugh The maGnitude of the cOrrElatIOn eFfectS is reLatiVeLy smaLl and tHeir bUiLdup slow, once estAbliShed their DecAy Is eVeN slowER, and foR $t>1/{\gAmmA }_1{(Q)\equiv \Tau }_\mu (Q)$ tHIs rElAXATiOn channel dominates ThE DEcAy of $M_L(Q,t)$, Which cONsEqUEntly may ExHibIt a sMALl but Long-LAsTing ”tail”. the ansATz (\[X2tEmpi\]) can InCorporAtE moSt oF this PHysiCs: on thE basis of The laTTer, one may reasoNAbly anticipatE ThAT ${\AlPHa (Q)\lL }1$, anD that the timE $1/{\gamMA }_2{(Q)\eqUiv \tAU }_\aLphA (q)$ is diStincTlY LoNGer than $1/{\gamma }_1{(Q)\equiv \tAu }_\Mu (Q)$. In tHis piCture, the best fItted valueS OF The viscoElasTIc RAte $1/\tau (Q)$ clearly RepreSent some soRT of ”weighTed avErage” betWeen ${\gamma }_1{(q)}$ ANd ${\gamma }_2{(Q)}$. finAllY, we May ARGuE that at increaSINg $Q$ (nAmEly, over A shRinking LenGth ScaLe) tHe Magnitude ${\Alpha (Q)}$ of CoRrElAtIon EffecTS should dEcReaSe, And That aT Higher TempeRatuReS tHE vaLue of ${\alPHa (q)}$ AT a giVeN wAvevEctOr ShoulD equALly DecreasE. On a generAl bASis, tHe ReQuiremeNt that $\lim_{Q\rigHtArrow 0} \tilde{m}_L(q,s=0)/Q^2\RightaRROw \eta _L/nm$ Now takes the form $$(c_\infty ^2-c_0^2)\lEFt[ \frac{{(1-\aLphA (Q\rigHtarRow 0))}}{{{\gamma }_1{(Q\RigHtarroW 0)}}}{\ +\ }\frAC

_\alpha(Q)>>1$, i.e. somepart of th e vis cou s f lo w is fro zen. As a cons e quen ce, at the wavevectors typi ca l oft he IXSexperim e nt s ($Q =1 \d iv20 $ n m$^{- 1}$ ) the q uasielasti c s pe ctrum acquir e sa componen t a rising fromthi s froz en st r uctur alrelax ation. The o rigin, at t h e atom i c level , of thi s fast decay chan n el is still an op en iss ue : t h e ra pid ly decayin gporti o n of $M _ L( Q , t )$i s customarily attributed tolargel yunc o rrelat ed co ll i sio nal events, sim ilar to t hose o c curring in a di lute f lui d.In a d di ti on, a t th e h igh den sities t yp ic al of the l i q uidsta te,non-n egligible cor rel atio n s a mongthe c olli si ons c an beexpec te d, making no lo nger valid an in te rpr et ation only i n t erm s of ”b inary”c oll is i o n s. Although the magn it u d eof the c orrela t io ne ffects i srel ativ e l y sma ll a n dtheir bu ildups lo w, once e st ablish ed th eir deca y iseven s lower, a nd fo r $t>1/{\gamma} _1{(Q)\equiv\ ta u }_ \ mu ( Q)$ this relax atio n cha nnel do min a tes t he de ca y o f $M_L(Q,t)$, whichco nseque ntlymay exhibit a small but l o ng-lasti ng ” t ai l ”. The ansatz(\[x2 tempi\]) c a n incorp orate most of this phy s i cs: on t hebas isoft h elatter, one m a y rea so nably a nti cipatetha t $ {\a lph a(Q)\ll }1 $, and t ha tth etim e $1/ { \gamma } _2 {(Q )\ equ iv \t a u }_\a lpha(Q)$ i sd ist inctlyl on g e r th an $ 1/{\ gam ma }_1{ (Q)\ e qui v \tau}_\mu (Q) $.I n th is p icture, the best fit te d values o fthe visco e l astic ra te $1/\tau (Q)$ clearly represe ntsomesort of ”weig hte d aver age ” betwe en ${\ gamma } _1{ ( Q )}$ a n d $ {\g am ma }_2{(Q) } $ . F inall y, wemay arg ue that at increas i ng$Q$ (namely,ove r as h ri nki n gl eng th sca l e ) the magnitude ${\alpha(Q ) }$ of correl a tio neffects should decr e ase, an d that at higher t em pera t u res the value of ${\a lpha (Q)} $ at a gi ven w ave vector s hou ld eq uallyd ecr ease. On age neralbasis ,the requ irement that $\lim_{Q\r ightar row 0 } \ tilde{M}_ L(Q , s=0 )/Q^2\rig htar row \eta _ L/n m$now t ake s theform $ $(c _ \inft y ^2 - c_0^2)\le f t[ \f r a c{ {(1-\alpha( Q \ rig htarr ow0 ))}}{{ {\ga mma }_1{(Q\righta r row 0)}}}{\ +\ }\f r a c

_\alpha(Q)>>1$, i.e._some part_of the viscous flow_is frozen._As_a consequence,_at_the wavevectors typical_of the IXS_experiments ($Q=1\div 20$ nm$^{-1}$)_the quasielastic spectrum_acquires_a component arising from this frozen structural relaxation. The origin, at the atomic level, of_this_fast decay_channel_is_still an open issue: the_rapidly decaying portion of $M_L(Q,t)$_is customarily_attributed to largely uncorrelated collisional events, similar to_those_occurring in a_dilute fluid. In addition, at the high densities typical_of the liquid state, non-negligible correlations_among the collisions_can_be_expected, making no longer_valid an interpretation only in terms_of ”binary” collisions. Although the magnitude_of the correlation effects is relatively small_and their buildup slow, once established_their decay is even slower,_and for_$t>1/{\gamma }_1{(Q)\equiv \tau }_\mu (Q)$ this_relaxation channel dominates_the decay_of $M_L(Q,t)$, which_consequently may exhibit a small but_long-lasting ”tail”. The_ansatz (\[x2tempi\]) can incorporate most of_this_physics: on the_basis_of_the latter,_one may reasonably_anticipate_that ${\alpha_(Q)\ll_}1$, and that the time $1/{\gamma_}_2{(Q)\equiv_\tau }_\alpha (Q)$ is distinctly longer than_$1/{\gamma }_1{(Q)\equiv \tau }_\mu (Q)$._In_this picture, the best_fitted values of the viscoelastic_rate $1/\tau (Q)$ clearly represent some_sort of_”weighted average”_between ${\gamma }_1{(Q)}$ and ${\gamma }_2{(Q)}$. Finally, we may argue that_at increasing $Q$ (namely, over a_shrinking length scale) the_magnitude ${\alpha (Q)}$_of_correlation effects should_decrease,_and that_at higher temperatures the value of ${\alpha_(Q)}$ at_a given wavevector should equally decrease._On a general basis,_the_requirement that $\lim_{Q\rightarrow 0} \tilde{M}_L(Q,s=0)/Q^2\rightarrow \eta _L/nm$_now takes the form $$(c_\infty ^2-c_0^2)\left[ \frac{{(1-\alpha_(Q\rightarrow 0))}}{{{\gamma }_1{(Q\rightarrow 0)}}}{\ +\_}\frac

\varphi_{j} - \left %(\omega_{0}^2/\omega\right ) \sin \varphi_{j}, \end{aligned}$$ and the solution of Eq. (\[maxwell-laplace\]) reads $$\begin{aligned} \label{electric} {\bf E}({\bf r},\omega ) = \frac{4\pi d_{0}q \omega^{2}}{c^2} \sum_{j}{\bf G}({\bf r},{\bf r}_{j},\omega) \cdot {\bf e}_{j} v_{j}, \end{aligned}$$ where ${\bf G}({\bf r},{\bf r}',\omega)$ is the electric field Green diadic *in the presence of* NP. From Eq. (\[displ\]), for photon frequency close to dipoles frequency, $\omega \approx \omega_{0}$, we obtain a coupled system of equations for normalized displacements, $$\begin{aligned} \label{system} \sum_{k} \Bigl[(\omega_{0}-\omega ) \delta_{jk} + \Sigma _{jk}\Bigr] v_{k} = %\frac{v_{0j}}{2\omega_{0}}= \frac{-i}{2} e^{-i\varphi_{j}},\end{aligned}$$ where $\Sigma _{jk}=\Delta_{jk}-\frac{i}{2}\Gamma_{jk}$ is the complex *self-energy matrix*, given by $$\begin{aligned} \label{self-general} \Sigma _{jk}(\omega)= - \frac{2\pi q^2 \omega_{0}}{m c^2} \, {\bf e}_{j}\cdot {\bf G}({\bf r}_{j},{\bf r}_{k};\omega) \cdot {\bf e}_{k}.\end{aligned}$$ The system Eq. (\[system\]) determines eigenstates of the ensemble of $N$ dipoles in common radiation and plasmon field, while complex eigenvalues of the self-energy matrix give eigenstates frequency shifts with respect to $\omega_{0}$ and their decay rates. In the absence of NP, real and imaginary parts of the self-energy matrix are dipole-dipole interaction and radiation coupling between dipoles $j$ and $k$, given by (in lowest order in $kr_{jk}$) $$\begin{aligned} \label{dipole-dipole} &\Delta_{jk}^{0}=\frac{3\Gamma_{0

\varphi_{j } - \left% (\omega_{0}^2/\omega\right) \sin \varphi_{j }, \end{aligned}$$ and the solution of Eq.   (\[maxwell - laplace\ ]) reads $ $ \begin{aligned } \label{electric } { \bf E}({\bf r},\omega) = \frac{4\pi d_{0}q \omega^{2}}{c^2 } \sum_{j}{\bf G}({\bf r},{\bf r}_{j},\omega) \cdot { \bf e}_{j } v_{j }, \end{aligned}$$ where $ { \bf G}({\bf r},{\bf r}',\omega)$ is the electric field Green diadic * in the presence of * NP. From Eq.   (\[displ\ ]), for photon frequency close to dipole frequency, $ \omega \approx \omega_{0}$, we receive a coupled system of equations for normalize displacements, $ $ \begin{aligned } \label{system } \sum_{k } \Bigl[(\omega_{0}-\omega) \delta_{jk } + \Sigma _ { jk}\Bigr ] v_{k } = % \frac{v_{0j}}{2\omega_{0}}= \frac{-i}{2 } e^{-i\varphi_{j}},\end{aligned}$$ where $ \Sigma _ { jk}=\Delta_{jk}-\frac{i}{2}\Gamma_{jk}$ is the complex * self - energy matrix *, give by $ $ \begin{aligned } \label{self - general } \Sigma _ { jk}(\omega)= - \frac{2\pi q^2 \omega_{0}}{m c^2 } \, { \bf e}_{j}\cdot { \bf G}({\bf r}_{j},{\bf r}_{k};\omega) \cdot { \bf e}_{k}.\end{aligned}$$ The system Eq.   (\[system\ ]) determine eigenstates of the ensemble of $ N$ dipoles in coarse radiation and plasmon field, while complex eigenvalue of the self - energy matrix give eigenstates frequency shifts with respect to $ \omega_{0}$ and their decay rates. In the absence of NP, real and imaginary part of the self - energy matrix are dipole - dipole interaction and radiation coupling between dipole $ j$ and $ k$, given by (in lowest order in $ kr_{jk}$) $ $ \begin{aligned } \label{dipole - dipole } & \Delta_{jk}^{0}=\frac{3\Gamma_{0

\vagphi_{j} - \left %(\omega_{0}^2/\omega\rinht ) \sin \varphi_{j}, \end{almgned}$$ ahd the sulution of Eq. (\[maxwell-laplace\]) ceadw $$\begun{aligned} \label{electric} {\cf E}({\bf r},\olega ) = \feac{4\pm d_{0}q \omega^{2}}{c^2} \sum_{j}{\ur G}({\bf r},{\ny r}_{j},\ojcga) \cbov {\bf e}_{j} v_{j}, \end{akigned}$$ whese ${\bf G}({\bf r},{\bf s}',\ooeya)$ is the electric field Green diadis *in thr oresence of* NP. Frok Eq. (\[djspl\]), for photon frequency close to dipolev frequency, $\okega \approx \omega_{0}$, we obtaij a foupled system of fquations fir njemalized disolacements, $$\begin{alignes} \label{system} \sum_{k} \Bigl[(\omega_{0}-\omega ) \delca_{jk} + \Sigmq _{hk}\Blcr] v_{k} = %\frac{t_{0j}}{2\omegw_{0}}= \frac{-i}{2} e^{-i\varini_{j}},\end{dligned}$$ where $\Sigma _{jl}=\Denta_{hk}-\frac{i}{2}\Gamma_{jk}$ is the complex *self-energi matrix*, ghvzn by $$\begin{aligned} \labwl{welf-ggneran} \Sigoq _{jy}(\omtga)= - \rrac{2\pi q^2 \kmega_{0}}{m c^2} \, {\gf e}_{j}\cdot {\vf G}({\bf r}_{j},{\bf r}_{k};\omega) \cqit {\bf e}_{k}.\end{aljgned}$$ Ehq system Eq. (\[system\]) determines eigenstatev or the ensemble of $N$ dipiles in common radiatlon and pjasmon field, while complex eigenvalues of the selx-enerjy maurlw gixw figenstates frequency shifts with respect to $\jjeba_{0}$ and their deccy rates. In the sbdemse of NP, real and imagjnary parts of the self-engrgy mqtrix are dipple-dipole interaction and rqdiation couilint between dipoles $l$ and $k$, giveu by (im lowrst order in $kr_{jk}$) $$\begin{clignes} \label{dipolf-dipole} &\Demga_{jk}^{0}=\frac{3\Gamma_{0

\varphi_{j} - \left %(\omega_{0}^2/\omega\right ) \sin \varphi_{j}, the of Eq. reads $$\begin{aligned} \label{electric} \frac{4\pi \omega^{2}}{c^2} \sum_{j}{\bf G}({\bf r}_{j},\omega) \cdot {\bf v_{j}, \end{aligned}$$ where ${\bf G}({\bf r},{\bf is the electric field Green diadic *in the presence of* NP. From Eq. for photon frequency close to dipoles frequency, $\omega \approx \omega_{0}$, we obtain a system equations normalized $$\begin{aligned} \label{system} \sum_{k} \Bigl[(\omega_{0}-\omega ) \delta_{jk} + \Sigma _{jk}\Bigr] v_{k} = %\frac{v_{0j}}{2\omega_{0}}= \frac{-i}{2} e^{-i\varphi_{j}},\end{aligned}$$ where $\Sigma is the complex *self-energy matrix*, given by $$\begin{aligned} \Sigma _{jk}(\omega)= - \frac{2\pi \omega_{0}}{m c^2} \, {\bf e}_{j}\cdot G}({\bf r}_{k};\omega) \cdot e}_{k}.\end{aligned}$$ system (\[system\]) determines eigenstates the ensemble of $N$ dipoles in common radiation and plasmon field, while complex eigenvalues of the self-energy give eigenstates with respect $\omega_{0}$ their rates. In the NP, real and imaginary parts of are dipole-dipole interaction and radiation coupling between dipoles and $k$, by (in lowest order in $kr_{jk}$) \label{dipole-dipole} &\Delta_{jk}^{0}=\frac{3\Gamma_{0

\varphi_{j} - \left %(\omega_{0}^2/\omega\righT ) \sin \varphi_{J}, \end{aLigNed}$$ AnD the SoluTion of Eq. (\[maxwelL-LaplAce\]) reads $$\begin{aligned} \laBel{elEcTRic} {\bF e}({\bF r},\omeGa ) = \frac{4\pI D_{0}q \OMEga^{2}}{C^2} \sUm_{J}{\bf g}({\bF R},{\bF r}_{j},\omEga) \Cdot {\bf e}_{J} v_{j}, \end{aligNed}$$ WhEre ${\bf G}({\bf r},{\bf r}',\OMeGa)$ is the eleCtrIc field Green DiaDic *in tHe PreSEnce oF* NP. from EQ. (\[displ\]), FOr photOn frequenCy CLose to DIpoles fREQuEncy, $\Omega \approx \omega_{0}$, wE ObTAin a coupled sysTem of eQuATiONS foR noRmalized diSpLacemENts, $$\begiN{AlIGNEd} \lABel{system} \sum_{k} \bigl[(\omega_{0}-\omEGa ) \dElta_{jk} + \siGma _{JK}\Bigr] v_{K} = %\frac{V_{0j}}{2\OMegA_{0}}= \frac{-i}{2} e^{-i\varPhi_{j}},\End{aligneD}$$ where $\sIgma _{jk}=\DELta_{jk}-\frAc{i}{2}\GamMa_{jK}$ is The cOMpLeX *seLf-ENerGY mAtrIX*, giVen by $$\begIn{AlIgned} \LabeL{SELF-genEraL} \SigMa _{jk}(\oMega)= - \frac{2\pi q^2 \omEga_{0}}{M c^2} \, {\bf E}_{J}\cdOt {\bf G}({\Bf r}_{j},{\bF r}_{k};\oMeGa) \cdoT {\bf e}_{k}.\eNd{aliGnEd}$$ The system Eq. (\[syStem\]) DetermineS eiGeNstAtEs of tHE ensemBle Of $N$ Dipoles In commoN RadIaTION aNd plasmon field, whilE cOMPlEx eigenvAlues oF ThE sELf-energy MaTriX givE EIgensTateS FrEquency sHifts wITh ReSpect to $\OmEga_{0}$ and ThEir DecAy ratES. In tHe abseNce of NP, rEal anD Imaginary parts OF the self-energY MaTRIx ARe diPolE-dipole inteRactIOn anD radIAtIon COupliNg betWeEN dIPoles $j$ and $k$, given by (in LoWest orDer in $Kr_{jk}$) $$\begin{aligNed} \label{diPOLE-dipole} &\DElta_{JK}^{0}=\fRAc{3\Gamma_{0

\varphi_{j} - \left %(\om ega_{0}^2/ \omeg a\r igh t) \s in \ varphi_{j}, \ end{ aligned}$$ and the sol ution o f Eq. (\ [maxw ell-lap l ac e \ ])re ad s $ $\ b eg in{al ign ed} \la bel{electr ic} { \bf E}({\bfr }, \omega ) = \f rac{4\pi d_{ 0}q \omeg a^ {2} } {c^2} \s um_{j }{\bfG }({\bf r},{\bfr} _ {j},\o m ega) \c d o t{\bf e}_{j} v_{j}, \e n d{ a ligned}$$ wher e ${\b fG }( { \ bfr}, {\bf r}',\ om ega)$ is thee le c t r icf ield Green di adic *in th e pr esence o f*N P. Fro m Eq.  ( \ [di spl\]), for pho ton frequ ency c l ose tod ipolesfreque ncy , $ \ome g a\a ppr ox \om e ga _{0 } $,we obtai nacoupl ed s y s t e m of eq uati ons f or normalized di spla c eme nts,$$\be gin{ al igned } \lab el{sy st em} \sum_{k} \B igl[ (\omega_{ 0}- \o meg a) \de l ta_{jk } + \ Sigma _ {jk}\Bi g r]v_ { k } = %\frac{v_{0j}}{2 \o m e ga _{0}}= \ frac{- i }{ 2} e^{-i\va rp hi_ {j}} , \ end{a lign e d} $$ where $\Sig m a_{ jk}=\De lt a_{jk} -\ fra c{i }{2}\ G amma _{jk}$ is thecompl e x *self-energy matrix*, give n b y $$ \ begi n{a ligned} \la bel{ s elf- gene r al } \ S igma_{jk} (\ o me g a)= - \frac{2\pi q^ 2\omega _{0}} {m c^2} \, {\ bf e}_{j}\ c d o t {\bfG}({ \ bf r}_{j},{\bf r} _{k}; \omega) \c d ot {\bfe}_{k }.\end{a ligned}$$ T he syste m E q.(\[ sys t e m\ ]) determines e igen st ates of th e ensem ble of $N $ d ip oles in c ommon ra di at io nand plas m on field ,whi le co mplex eigenv alues ofth es elf -energy ma t r ix g iv eeige nst at es fr eque n cyshiftswith resp ect to $ \o me ga_{0}$ and their de ca y rates. I nthe absen c e of NP,real and imaginary part s of the se lf-en ergy matrix a redipole -di p ole in teract ion a nd ra d i ation c ou pli ng between d i p ole s $j$ a nd $ k$, giv en by (in lowest o r der in $kr_{jk}$ ) $ $\be g i n{ ali g ne d } \ la b el{ d i pole-dipole} &\ Delta_{jk} ^{ 0 }= \frac{3\Ga m ma_ {0

\varphi_{j}_- \left_%(\omega_{0}^2/\omega\right ) \sin \varphi_{j},_ \end{aligned}$$_and_the solution_of_Eq. (\[maxwell-laplace\]) reads $$\begin{aligned} \label{electric} {\bf_E}({\bf r},\omega )_= \frac{4\pi d_{0}q \omega^{2}}{c^2}_\sum_{j}{\bf G}({\bf r},{\bf_r}_{j},\omega)_\cdot {\bf e}_{j} v_{j}, \end{aligned}$$ where ${\bf G}({\bf r},{\bf r}',\omega)$ is the electric field_Green_diadic *in_the_presence_of* NP. From Eq. (\[displ\]), for_photon frequency close to dipoles_frequency, $\omega_\approx \omega_{0}$, we obtain a coupled system of_equations_for normalized displacements,_$$\begin{aligned} \label{system} \sum_{k} \Bigl[(\omega_{0}-\omega ) \delta_{jk} + \Sigma _{jk}\Bigr] v_{k}_= %\frac{v_{0j}}{2\omega_{0}}= \frac{-i}{2} e^{-i\varphi_{j}},\end{aligned}$$ where $\Sigma _{jk}=\Delta_{jk}-\frac{i}{2}\Gamma_{jk}$_is the complex_*self-energy_matrix*,_given by $$\begin{aligned} \label{self-general} \Sigma _{jk}(\omega)=_- \frac{2\pi q^2 \omega_{0}}{m c^2} \,_{\bf e}_{j}\cdot {\bf G}({\bf r}_{j},{\bf_r}_{k};\omega) \cdot {\bf e}_{k}.\end{aligned}$$ The system Eq. (\[system\])_determines eigenstates of the ensemble of_$N$ dipoles in common radiation_and plasmon_field, while complex eigenvalues of_the self-energy matrix_give eigenstates_frequency shifts with_respect to $\omega_{0}$ and their decay_rates. In the_absence of NP, real and imaginary_parts_of the self-energy_matrix_are_dipole-dipole interaction_and radiation coupling_between_dipoles $j$_and_$k$, given by (in lowest order_in_$kr_{jk}$) $$\begin{aligned} \label{dipole-dipole} &\Delta_{jk}^{0}=\frac{3\Gamma_{0

of the initial state ${| M;\down\ldots\down \rangle}$. The initial condition for the dynamical mean-field equations, which corresponds to this state, is all spins down, $\mathbf{C}_{j}\left(0\right)=\left(0,0,-1/2\right)$, and a finite amplitude of the bosonic field, $a\left(0\right)=\sqrt{M}$. Below, we solve set of differential Eqs. (\[eq:CEOM1\], \[eq:CEOM2\]) for the evolution of classical variables with this initial condition. For a small number of excitations in a system with many spins, $M\ll N$, these equations were solved in \[\]. Using the approximation $C_{j}^{z}\left(t\right)\approx-1/2$, the equation for $C_{j}^{z}\left(t\right)$ drops out from Eq. (\[eq:CEOM1\]) and the remaining system of equation is harmonic. In the regime $\Omega\gg\Delta$ this gives the following solution for the bosonic mode $$a\left(t\right)=\sqrt{M}\cos\left(V\sqrt{N}t\right). \label{eq:harmonic}$$ The period of the oscillatory function in this limit is $T=2\pi/\left(V\sqrt{N}\right)$, see Figure \[fig:time\_evolution\] for $M=10$. In the following will use the Rabi frequency $\Omega=V\sqrt{N}$ which was introduced in Sec. II. For arbitrary $M$ the solution to Eq. (\[eq:CEOM1\], \[eq:CEOM2\]) are hyperelliptic functions [@fermi_bose]. Here we will not consider the full analytic form of general solution but will use only the spectral analysis developed in Ref. \[\]. Introducing the following vector function (Lax vector) of an auxiliary parameter $u$, $$\mathbf{L}\left(u\right)=\left(\begin{array}{c} \frac{a_{x}\left(0\right)}{V}\\ \frac{a_{y}\left(0\right)}{V}\\ \frac{u}{2V^{2}} \end{array}\right)+\sum_{j}\frac{\mathbf{C}_{j}\left(0\right)}{u-\left(\epsilon_{j}-\omega\

of the initial state $ { | M;\down\ldots\down \rangle}$. The initial condition for the dynamical mean - sphere equation, which corresponds to this state, is all spins down, $ \mathbf{C}_{j}\left(0\right)=\left(0,0,-1/2\right)$, and a finite amplitude of the bosonic discipline, $ a\left(0\right)=\sqrt{M}$. Below, we solve set up of differential Eqs. (\[eq: CEOM1\ ], \[eq: CEOM2\ ]) for the development of classical variables with this initial condition. For a small number of excitation in a system with many spins, $ M\ll N$, these equation were solved in \[\ ]. Using the approximation $ C_{j}^{z}\left(t\right)\approx-1/2 $, the equation for $ C_{j}^{z}\left(t\right)$ drops out from Eq. (\[eq: CEOM1\ ]) and the stay system of equation is harmonic. In the regimen $ \Omega\gg\Delta$ this gives the following solution for the bosonic mode $ $ a\left(t\right)=\sqrt{M}\cos\left(V\sqrt{N}t\right). \label{eq: harmonic}$$ The period of the oscillatory affair in this limit is $ T=2\pi/\left(V\sqrt{N}\right)$, see Figure \[fig: time\_evolution\ ] for $ M=10$. In the following will use the Rabi frequency $ \Omega = V\sqrt{N}$ which was introduced in Sec. II. For arbitrary $ M$ the solution to Eq. (\[eq: CEOM1\ ], \[eq: CEOM2\ ]) are hyperelliptic routine [ @fermi_bose ]. Here we will not consider the full analytic form of general solution but will use only the spectral analysis developed in Ref. \[\ ]. Introducing the following vector function (Lax vector) of an auxiliary parameter $ u$, $ $ \mathbf{L}\left(u\right)=\left(\begin{array}{c } \frac{a_{x}\left(0\right)}{V}\\ \frac{a_{y}\left(0\right)}{V}\\ \frac{u}{2V^{2 } } \end{array}\right)+\sum_{j}\frac{\mathbf{C}_{j}\left(0\right)}{u-\left(\epsilon_{j}-\omega\

of the initial state ${| M;\dowk\ldots\down \ranglg}$. Rhe inmtial ckndition for the dynamical mean-field ewuatiins, which corresponds go this snate, is aol skins down, $\mathbf{C}_{o}\meft(0\rigmc)=\left(0,0,-1/2\dlght)$, cnv a finite amplltude of tha bosonic fielg, $x\lzft(0\right)=\sqrt{M}$. Below, we solve set of qifferemtlal Eqs. (\[eq:CEOM1\], \[eq:CTOM2\]) for nht evolution of classical variablea with uhis initial condiyion. For a small number of fxcihations in a systel with many spigw, $M\ll N$, thesd equations were solves in \[\]. Using the approximation $C_{g}^{z}\lefc(t\right)\apprix-1/2$, thf equation fir $C_{j}^{d}\left(t\right)$ dvpps oud from Rq. (\[eq:CEOM1\]) and bhe rxmaibing system of equatimn is harmonic. In jhe regime $\Ojega\gg\Delta$ this tices tve fmllodung somuvioh for hhe bosonic mkde $$a\left(t\rught)=\sqrt{M}\cos\left(V\sqtt{G}n\tight). \label{eq:garmonyc}$$ The period of the oscillatory function in this limit is $T=2\pi/\left(V\wqrt{N}\right)$, see Figure \[fig:time\_edolution\] for $M=10$. In the following will use the Rabi freqnevcy $\Oidta=G\sqrt{N}$ which was introduced in Sec. II. For arbiedaty $M$ the solutiok to Eq. (\[eq:CEOM1\], \[eq:CROL2\]) ste hyperelliptkc funefikns [@fermi_bose]. Here we wilj not consider the full analytic form of geneeal solution vut will use only che spectral anakysis developed in Ref. \[\]. Intruducjng the folpowing vedgor function (Lax vebtor) of an auxiliary parameter $u$, $$\mathbf{O}\lefc(u\right)=\ldft(\bggin{arrwy}{c} \frac{a_{x}\lfft(0\rinvt)}{V}\\ \frac{a_{y}\left(0\right)}{G}\\ \frac{o}{2V^{2}} \end{asray}\right)+\skm_{j}\frac{\mathbf{C}_{j}\left(0\right)}{u-\left(\epsilon_{j}-\omega\

of the initial state ${| M;\down\ldots\down \rangle}$. condition the dynamical equations, which corresponds spins $\mathbf{C}_{j}\left(0\right)=\left(0,0,-1/2\right)$, and a amplitude of the field, $a\left(0\right)=\sqrt{M}$. Below, we solve set differential Eqs. (\[eq:CEOM1\], \[eq:CEOM2\]) for the evolution of classical variables with this initial For a small number of excitations in a system with many spins, $M\ll these were in Using the approximation $C_{j}^{z}\left(t\right)\approx-1/2$, the equation for $C_{j}^{z}\left(t\right)$ drops out from Eq. (\[eq:CEOM1\]) and the remaining of equation is harmonic. In the regime $\Omega\gg\Delta$ gives the following solution the bosonic mode $$a\left(t\right)=\sqrt{M}\cos\left(V\sqrt{N}t\right). \label{eq:harmonic}$$ period the oscillatory in limit $T=2\pi/\left(V\sqrt{N}\right)$, see Figure for $M=10$. In the following will use the Rabi frequency $\Omega=V\sqrt{N}$ which was introduced in Sec. II. arbitrary $M$ to Eq. \[eq:CEOM2\]) hyperelliptic [@fermi_bose]. Here we consider the full analytic form of will use only the spectral analysis developed in \[\]. Introducing following vector function (Lax vector) of auxiliary parameter $u$, $$\mathbf{L}\left(u\right)=\left(\begin{array}{c} \frac{a_{x}\left(0\right)}{V}\\ \frac{a_{y}\left(0\right)}{V}\\ \frac{u}{2V^{2}} \end{array}\right)+\sum_{j}\frac{\mathbf{C}_{j}\left(0\right)}{u-\left(\epsilon_{j}-\omega\

of the initial state ${| M;\down\ldoTs\down \rangLe}$. The IniTiaL cOndiTion For the dynamicaL Mean-Field equations, which corRespoNdS To thIS sTate, iS all spiNS dOWN, $\maThBf{c}_{j}\lEfT(0\RiGht)=\leFt(0,0,-1/2\rIght)$, and A finite ampLitUdE of the bosoniC FiEld, $a\left(0\riGht)=\Sqrt{M}$. Below, we SolVe set oF dIffERentiAl EQs. (\[eq:CeOM1\], \[eq:CeoM2\]) for tHe evolutiOn OF classICal variABLeS witH this initial condiTIoN. for a small numbeR of excItATiONS in A syStem with maNy Spins, $m\Ll N$, thesE EqUATIonS Were solved in \[\]. USing the apprOXimAtion $C_{J}^{z}\LefT(T\right)\ApproX-1/2$, tHE eqUation for $C_{j}^{Z}\lefT(t\right)$ drOps out FRom Eq. (\[eq:ceOM1\]) and tHe remaIniNg sYsteM Of EqUatIoN Is hARmOniC. in tHe regime $\omEgA\gg\DeLta$ tHIS GIves The FollOwing Solution for thE boSoniC ModE $$a\lefT(t\rigHt)=\sqRt{m}\cos\lEft(V\sqRt{N}t\rIgHt). \label{eq:harmonIc}$$ ThE period of The OsCilLaTory fUNction In tHis Limit is $t=2\pi/\left(v\SqrT{N}\RIGHt)$, See Figure \[fig:time\_evOlUTIoN\] for $M=10$. In tHe follOWiNg WIll use thE RAbi FreqUENcy $\OmEga=V\SQrT{N}$ which wAs intrODuCeD in Sec. Ii. FOr arbiTrAry $m$ thE soluTIon tO Eq. (\[eq:CeOM1\], \[eq:CEOm2\]) are hYPerelliptic funCTions [@fermi_bosE]. heRE We WIll nOt cOnsider the fUll aNAlytIc foRM oF geNEral sOlutiOn BUt WIll use only the spectrAl AnalysIs devEloped in Ref. \[\]. InTroducing tHE FOllowing VectOR fUNction (Lax vectoR) of an Auxiliary pARameter $u$, $$\MathbF{L}\left(u\rIght)=\left(\bEGIn{array}{c} \FraC{a_{x}\LefT(0\riGHT)}{V}\\ \Frac{a_{y}\left(0\rigHT)}{v}\\ \fraC{u}{2v^{2}} \end{arrAy}\rIght)+\sum_{J}\frAc{\mAthBf{C}_{J}\lEft(0\right)}{u-\Left(\epsiLoN_{j}-\OmEgA\

of the initial state ${|M;\down\ld ots\d own \r an gle} $. T he initial con d itio n for the dynamical me an-fi el d equ a ti ons,which c o rr e s pon ds t o t hi s s tate, is all sp ins down,$\m at hbf{C}_{j}\l e ft (0\right)= \le ft(0,0,-1/2\ rig ht)$,an d a finit e a mplit ude of the bo sonic fie ld , $a\le f t(0\rig h t )= \sqr t{M}$. Below, wes ol v e set of diffe rentia lE qs . (\[ eq: CEOM1\], \ [e q:CEO M 2\]) fo r t h e evo l ution of clas sical varia b les withth isi nitial cond it i on. For a sma ll n umber ofexcita t ions in a syste m with ma nyspin s ,$M \ll N $ , t h es e e q uat ions wer eso lvedin \ [ \ ] . Usi ngtheappro ximation $C_{ j}^ {z}\ l eft (t\ri ght)\ appr ox -1/2$ , theequat io n for $C_{j}^{z }\le ft(t\righ t)$ d rop sout f r om Eq. (\ [eq :CEOM1\ ]) andt here m a i ni ng system of equat io n is harmoni c. Int he r e gime $\O me ga\ gg\D e l ta$ t hisg iv es the f ollowi n gso lutionfo r thebo son icmode$ $a\l eft(t\ right)=\ sqrt{ M }\cos\left(V\s q rt{N}t\right) . \ l a be l {eq: har monic}$$ Th e pe r iodof t h eosc i llato ry fu nc t io n in this limit is $ T= 2\pi/\ left( V\sqrt{N}\rig ht)$, seeF i g ure \[fi g:ti m e\ _ evolution\] fo r $M= 10$. In th e followi ng wi ll use t he Rabi f r e quency $ \Om ega =V\ sqr t { N} $ which was i n t rodu ce d in Se c.II. Fo r a rbi tra ry$M $ the sol ution to E q. ( \[ eq: CEOM1 \ ], \[eq: CE OM2 \] ) a re hy p erelli pticfunc ti on s [@ fermi_b o se ] . Her ewe wil l n ot cons ider the full a nalytic f orm of g en er al solu tion but will u se only th espe ctrala n alysis d eveloped in Ref. \[\].I ntroduc ing thefoll owing vec tor funct ion (Lax v ector) of a naux i l iaryp a ra met er $u$, $$\m a t hbf {L}\l ef t(u\ right)= \left(\begin{array } {c} \frac{a_{x}\ lef t(0\ r i gh t)} { V} \ \ \ fr a c{a _ { y}\left(0\right )}{V}\\ \f ra c {u }{2V^{2}}\ end {a rray}\r ight)+\ sum_{ j }\frac{ \mathbf{C }_{j}\lef t( 0\ri g h t)} {u-\left(\ epsilon_ {j}-\omeg a \

of_the initial_state ${| M;\down\ldots\down \rangle}$._The initial_condition_for the_dynamical_mean-field equations, which_corresponds to this_state, is all spins_down, $\mathbf{C}_{j}\left(0\right)=\left(0,0,-1/2\right)$, and_a_finite amplitude of the bosonic field, $a\left(0\right)=\sqrt{M}$. Below, we solve set of differential Eqs._(\[eq:CEOM1\],_\[eq:CEOM2\]) for_the_evolution_of classical variables with this_initial condition. For a small number_of excitations_in a system with many spins, $M\ll N$,_these_equations were solved_in \[\]. Using the approximation $C_{j}^{z}\left(t\right)\approx-1/2$, the equation for_$C_{j}^{z}\left(t\right)$ drops out from Eq. (\[eq:CEOM1\])_and the remaining_system_of_equation is harmonic. In_the regime $\Omega\gg\Delta$ this gives the_following solution for the bosonic mode_$$a\left(t\right)=\sqrt{M}\cos\left(V\sqrt{N}t\right). \label{eq:harmonic}$$ The period of the oscillatory function_in this limit is $T=2\pi/\left(V\sqrt{N}\right)$, see_Figure \[fig:time\_evolution\] for $M=10$. In_the following_will use the Rabi frequency_$\Omega=V\sqrt{N}$ which was_introduced in_Sec. II. For arbitrary_$M$ the solution to Eq. (\[eq:CEOM1\],_\[eq:CEOM2\]) are hyperelliptic_functions [@fermi_bose]. Here we will not_consider_the full analytic_form_of_general solution_but will use_only_the spectral_analysis_developed in Ref. \[\]. Introducing the_following_vector function (Lax vector) of an auxiliary_parameter $u$, $$\mathbf{L}\left(u\right)=\left(\begin{array}{c} \frac{a_{x}\left(0\right)}{V}\\ \frac{a_{y}\left(0\right)}{V}\\ \frac{u}{2V^{2}} \end{array}\right)+\sum_{j}\frac{\mathbf{C}_{j}\left(0\right)}{u-\left(\epsilon_{j}-\omega\

Corollary \[cor:gradient\]). We also uncover similar relationships for subgradient descent methods and Newton–Raphson method (Proposition \[prop:NewtonRaphson\]). This discovery is important for two reasons. *Firstly*, Theorem \[thm:gradientdescent\] is an explicit bridge between CRM and the state-of-the-art approach to analysis of DR for nonconvex problems, and so future works can now study CRM methods by using Lyapunov functions for DR operators. *Secondly*, the proximity operator of a function returns a step of gradient descent applied to its Moreau envelope (see [@BC Proposition 12.26]), and proximal splitting algorithms like Douglas–Rachford method have been variously shown to be describable as gradient descent methods applied to envelope functions; in turn, these envelopes have served as the basis for designing new algorithms. For example, see the works of Bemporad, Patrinos, Stella, and Themelis [@themelis2018douglas; @patrinos2014douglas]. Theorem \[thm:gradientdescent\] allows us to design new operators that share this relationship to Lyapunov functions. We introduce such a new operator (Definition \[def:LT\]) that also satisfies the property that it may be viewed in many cases as gradient descent applied to a Lyapunov function (Theorem \[thm:LT\]), but with the additional advantage that it depends only on the governing sequence. This means it may be implemented for a *general algorithm* when signs of spiraling are apparent, and not merely for feasibility problems with reflection substeps. This includes black box applications when only the governing sequence is known and subproblem solutions (e.g. proximal points, reflections) are not available. We then introduce a framework for primal/dual implementation, and we use our new operator to center the dual of ADMM for the basis pursuit problem. This is the *first* primal/dual framework and implementation for a centering method, and it opens many new avenues of investigation. A similar primal/dual approach based on a generalization of CRM does *not* appear to work, and we explain why we would not expect it to work. This showcases the value of our Lyapunov function approach and the advantages of the new operator we have obtained through it. Primal/dual centering methods are likely to be of importance for on-line optimization problems. For many warm-started applications, one might

Corollary   \[cor: gradient\ ]). We also uncover similar relationships for subgradient descent method and Newton – Raphson method acting (Proposition   \[prop: NewtonRaphson\ ]). This discovery is important for two reasons. * first *, Theorem   \[thm: gradientdescent\ ] is an explicit bridge between CRM and the country - of - the - art access to analysis of DR for nonconvex problems, and thus future works can now learn CRM methods by using Lyapunov affair for DR hustler. * Secondly *, the proximity hustler of a function return a measure of gradient descent applied to its Moreau envelope (see [ @BC Proposition   12.26 ]), and proximal splitting algorithm like Douglas – Rachford method have been variously shown to be describable as gradient origin methods use to envelope functions; in turn, these envelopes have served as the basis for design new algorithms. For example, see the works of Bemporad, Patrinos, Stella, and Themelis [ @themelis2018douglas; @patrinos2014douglas ]. Theorem   \[thm: gradientdescent\ ] allows us to design new operators that share this relationship to Lyapunov functions. We introduce such a new operator (Definition   \[def: LT\ ]) that also satisfies the place that it may be see in many cases as gradient lineage apply to a Lyapunov function (Theorem   \[thm: LT\ ]), but with the additional advantage that it depends merely on the governing sequence. This means it may be implemented for a * cosmopolitan algorithm * when signs of spiraling are apparent, and not merely for feasibility problems with reflection substeps. This includes black box applications when only the governing sequence is known and subproblem solution (for example proximal point, reflections) are not available. We then introduce a framework for cardinal / dual implementation, and we use our newfangled operator to center the dual of ADMM for the basis pursuit trouble. This is the * first * primal / dual framework and implementation for a focus on method, and it open many new avenue of investigation. A exchangeable primal / dual overture establish on a generalization of CRM does * not * look to work, and we explain why we would not expect it to work. This showcases the value of our Lyapunov routine access and the advantages of the new hustler we have obtained through it. Primal / dual centering methods are likely to be of importance for on - line optimization trouble. For many warm - started applications, one might

Cogollary \[cor:gradient\]). We alro uncover similar reletionshjps for rubgradient descent methods end Bewtob–Raphson method (Proposktion \[prop:JewtonRaphsoi\]). This discovery is impovcant rlr tco reasons. *Firstky*, Theorem \[dhm:gradientdeswevt\] is an explicit bridge between CRM wnd the shate-of-the-art akproabh to znalysis of DR for nonconvex problsms, and so future wotks can now study CRM methlds hy using Lyapunov vunctions fir DW operators. *Sdcondly*, tht 'roximity okerator of a function returns a rtep pf gradienj sedwent appliev to ins Moreau envckope (sae [@BC Ptoposition 12.26]), and prmxinal splitting algoritims like Douglas–Rachsord methmd have been variouwlt shofn tm be eeszrigaule as grwdixnt descent methods applied to envelope finsnoons; in turn, these egvelopes have served as the basis for dtsignjng new algorithms. For wxample, see the works of Bempowad, Patrinos, Stella, and Themelis [@themelis2018douglas; @[atriior2014dobnjxw]. Hheorem \[thm:gradientdescent\] allows us to design hee pperators that share this rflsjionship to Lyxpunov fuhctions. We introducf such w new operator (Defonition \[def:LT\]) that also satiwfies the prjperty that it may ye viewed in mani casex as gradient descent a'plied to a Lyapujov functjun (Theorem \[thm:LT\]), cut whth the additional advantade that iv depznds onlh on the gjverning sfquenga. This means it maj be nmplekented for a *general algorithm* when signs of spiraling sra aiparent, aud not merely for fqasibility proylems wich refuection sugsteps. Vhis includef black box a[klications whxn only tre giverbing sedjence is known and subpgoylem solurions (e.g. proximal iointr, reflections) arz bot available. We thdn ynnrovuce w framework fmr pfimxk/dual implemtutabiov, anc we use our new opesatod to center the dusl of ADMM for the basis pursuiy problem. This is uhe *ficst* prmmal/dusl sramework and implementation fkr a centfrikg method, and it ipens many nzw avenues of investigation. A similar prmmal/dual approach based on a generalizatiou pf CRM does *not* wppear to work, and we explain why we would not expect it to work. Thjs shofcased the value of our Lyapunov function approach and the advantages of the nww opecaeor we have obtsined tkrobgh it. Pwimap/vual centering menhods are likely to be of importence for ot-lnne optimization problems. For mxny warm-startgd applications, one might

Corollary \[cor:gradient\]). We also uncover similar relationships descent and Newton–Raphson (Proposition \[prop:NewtonRaphson\]). This reasons. Theorem \[thm:gradientdescent\] is explicit bridge between and the state-of-the-art approach to analysis DR for nonconvex problems, and so future works can now study CRM methods using Lyapunov functions for DR operators. *Secondly*, the proximity operator of a function a of descent to its Moreau envelope (see [@BC Proposition 12.26]), and proximal splitting algorithms like Douglas–Rachford method have variously shown to be describable as gradient descent applied to envelope functions; turn, these envelopes have served the for designing algorithms. example, the works of Patrinos, Stella, and Themelis [@themelis2018douglas; @patrinos2014douglas]. Theorem \[thm:gradientdescent\] allows us to design new operators that share this to Lyapunov introduce such new (Definition that also satisfies that it may be viewed in gradient descent applied to a Lyapunov function (Theorem but with additional advantage that it depends only the governing sequence. This means it may be for a *general algorithm* when signs of spiraling are apparent, and not merely for feasibility reflection substeps. This includes box applications when the sequence known subproblem solutions proximal points, reflections) are not available. We then introduce a framework primal/dual implementation, and we use our new operator to center of for the basis problem. This is the primal/dual and implementation for a and opens of A primal/dual approach based on generalization of CRM does *not* to work, and we expect it to work. This showcases the value our Lyapunov function approach and the advantages the new operator we have obtained through it. Primal/dual centering methods are to be for on-line optimization problems. For many warm-started applications, might

Corollary \[cor:gradient\]). We alsO uncover siMilar RelAtiOnShipS for Subgradient desCEnt mEthods and Newton–Raphson MethoD (PROposITiOn \[proP:NewtonrApHSOn\]). THiS dIscOvERy Is impOrtAnt for tWo reasons. *FIrsTlY*, Theorem \[thm:gRAdIentdescenT\] is An explicit brIdgE betweEn cRM ANd the StaTe-of-tHe-art aPProach To analysiS oF dR for nONconvex PRObLems, And so future works cAN nOW study CRM methoDs by usInG lyAPUnoV fuNctions for dR OperaTOrs. *SecoNDlY*, THE prOXimity operatoR of a functioN RetUrns a sTeP of GRadienT descEnT AppLied to its MoReau Envelope (sEe [@BC PrOPositioN 12.26]), And proxImal spLitTinG algORiThMs lIkE douGLaS–RaCHfoRd method HaVe Been vArioUSLY ShowN to Be deScribAble as gradienT deScenT MetHods aPplieD to eNvElope FunctiOns; in TuRn, these envelopeS havE served as The BaSis FoR desiGNing neW alGorIthms. FoR examplE, See ThE WORkS of Bemporad, PatrinoS, STELlA, and ThemElis [@thEMeLiS2018Douglas; @pAtRinOs2014doUGLas]. ThEoreM \[ThM:gradienTdesceNT\] aLlOws us to DeSign neW oPerAtoRs thaT SharE this rElationsHip to lYapunov functioNS. We introduce sUCh A NEw OPeraTor (definition \[dEf:LT\]) THat aLso sATiSfiES the pRoperTy THaT It may be viewed in many CaSes as gRadieNt descent applIed to a LyapUNOV functioN (TheOReM \[Thm:LT\]), but with thE addiTional advaNTage that It depEnds only On the goveRNIng sequeNce. thiS meAns IT MaY be implementeD FOr a *gEnEral algOriThm* when SigNs oF spIraLiNg are appaRent, and nOt MeReLy For FeasiBIlity proBlEms WiTh rEflecTIon subSteps. this InClUDes Black boX ApPLIcatIoNs When OnlY tHe govErniNG seQuence iS known and SubPRoblEm SoLutions (E.g. proximal poiNtS, reflectioNs) Are Not avaILAble. We thEn introduce a framework foR Primal/dUal ImpleMentAtion, and wE usE our neW opERator tO centeR the dUaL of admM for THE bAsiS pUrsuit probLEM. ThIs is tHe *FirsT* primal/Dual framework and imPLemEntation for a cEntErinG MEtHod, ANd IT opEnS ManY NEw avenues of inveStigation. A SiMIlAr primal/duAL apPrOach basEd on a geNeralIZation oF CRM does *nOt* appear tO wOrk, aND We eXplain why wE would noT expect it TO work. tHiS showCasEs the vAlUe oF our LYapunoV FunCtion ApproaCh And the AdvanTaGes of the New operator we have obtainEd throUgh it. priMal/dual ceNteRIng Methods arE likEly to be of iMpoRtaNce foR on-LIne opTimiZAtIon PRobleMs. FoR Many warm-sTArTed APPlIcations, one MIGHt

Corollary \[cor:gradient\ ]). We als o unc ove r s im ilar rel ationships for subg radient descent method s and N e wton – Ra phson method (P r o pos it io n \ [p r op :Newt onR aphson\ ]). This d isc ov ery is impor t an t for tworea sons. *First ly* , Theo re m \ [ thm:g rad ientd escent \ ] is a n explici tb ridgeb etweenC R Mandthe state-of-the- a rt approach to an alysis o f D R for no nconvex pr ob lems, and sof ut u r e wo r ks can now st udy CRM met h ods by us in g L y apunov func ti o nsfor DR oper ator s. *Secon dly*,t he prox i mity op erator of afunc t io nret ur n s a st epo f g radientde sc ent a ppli e d t o it s M orea u env elope (see [@ BCProp o sit ion 1 2.26] ), a nd prox imal s plitt in g algorithms li ke D ouglas–Ra chf or d m et hod h a ve bee n v ari ously s hown to bede s c r ib able as gradient d es c e nt methods appli e dto envelope f unc tion s ; in t urn, th ese enve lopesh av eservedas the b as isfor desi g ning new a lgorithm s. Fo r example, seet he works of B e mp o r ad , Pat rin os, Stella, and Them elis [@ the m elis2 018do ug l as ; @patrinos2014dougl as ]. The orem\[thm:gradien tdescent\] a l lows usto d e si g n new operator s tha t share th i s relati onshi p to Lya punov fun c t ions. W e i ntr odu ces u ch a new operat o r (De fi nition\[d ef:LT\] ) t hat al sosa tisfies t he prope rt yth at it mayb e viewed i n m an y c asesa s grad ientdesc en ta ppl ied toa L y a puno vfu ncti on(T heore m \[ t hm: LT\]),but withthe addi ti on al adva ntage that it d epends onl yonthe go v e rning se quence. This means it m a y be im ple mente d fo r a *gene ral algor ith m * when signs of s pi ral i n g are a pp are nt , and notm e rel y for f easi bilityproblems with refl e cti on substeps.Thi s in c l ud esb la c k b ox app l i cations when on ly the gov er n in g sequence iskn own and subpro blems olution s (e.g. p roximal p oi nts, r efl ections) a re not a vailable. We t h en intr odu ce a f ra mew ork f or pri m al/ dualimplem en tation , and w e use ou r new operator to cente r thedualofADMM forthe bas is pursui t pr oblem. Thi s i s t he *f irs t * pri mal/ d ua l f r amewo rk a n d impleme n ta tio n fo r a centeri n g met hod,and it ope ns m any new avenues o f investigation . As i mil arp rima l/ dual approachbas ed o n a gene ra lization of CRM doe s* not*appear to wo rk, and w ee xplain why we would no t e xp e ct it t owo r k. Thi s sh ow casesthe va l ue o f our Lyapunov fun ction a pproa c h a nd th eadvanta g es o f the newoperator we haveobta inedthrough i t. Pr ima l/ dual cente r ing metho ds ar e likel yto b e o f impo rtan c e foron-l in e o ptimizati o n p r ob le m s.Formanywa rm-s tarted ap p lication s,o ne migh t

Corollary \[cor:gradient\])._We also_uncover similar relationships for_subgradient descent_methods_and Newton–Raphson_method_(Proposition \[prop:NewtonRaphson\]). This discovery_is important for_two reasons. *Firstly*, Theorem \[thm:gradientdescent\]_is an explicit_bridge_between CRM and the state-of-the-art approach to analysis of DR for nonconvex problems, and_so_future works_can_now_study CRM methods by using_Lyapunov functions for DR operators._*Secondly*, the_proximity operator of a function returns a step_of_gradient descent applied_to its Moreau envelope (see [@BC Proposition 12.26]), and proximal_splitting algorithms like Douglas–Rachford method have_been variously shown_to_be_describable as gradient descent_methods applied to envelope functions; in_turn, these envelopes have served as_the basis for designing new algorithms. For_example, see the works of Bemporad,_Patrinos, Stella, and Themelis [@themelis2018douglas;_@patrinos2014douglas]. Theorem \[thm:gradientdescent\]_allows us to design new_operators that share_this relationship_to Lyapunov functions. We_introduce such a new operator (Definition \[def:LT\])_that also satisfies_the property that it may be_viewed_in many cases_as_gradient_descent applied_to a Lyapunov_function_(Theorem \[thm:LT\]), but_with_the additional advantage that it depends_only_on the governing sequence. This means it_may be implemented for_a_*general algorithm* when signs_of spiraling are apparent, and_not merely for feasibility problems with_reflection substeps._This includes_black box applications when only the governing sequence is known and_subproblem solutions (e.g. proximal points, reflections)_are not available. We then_introduce a_framework_for primal/dual implementation,_and_we use_our new operator to center the dual_of ADMM_for the basis pursuit problem. This_is the *first* primal/dual_framework_and implementation for a centering method,_and it opens many new avenues_of investigation. A similar primal/dual_approach_based_on a generalization of CRM_does *not* appear to work, and_we explain why_we would not expect it to work._This_showcases the value of our Lyapunov_function_approach and the advantages of the_new_operator_we have obtained through it. Primal/dual_centering methods are likely to be_of importance for on-line optimization problems. For many warm-started_applications, one might

\tau_h$. As the number of hops becomes very large for $t \gg \tau_h$, the distribution becomes Gaussian again, due to central limit theorem. The kurtosis decays back to zero. Analytical results for the kurtosis behavior in this region will be reported elsewhere. To conclude, we have proposed and analyzed a novel particle-tracking approach for identification of nonlinear interactions. Unlike common MSD techniques, our approach is based on the analysis of non-Gaussian characteristics of the particle dynamics, specifically the kurtosis of the displacement probability distribution. The functional dependence of the kurtosis on the measurement time carries a lot of information about the nonlinear interactions that contribute to the particle motion. E.g., if a particle is placed in a cage-type potential induced by cytoskeleton or transmembrane proteins, the resulting kurtosis of the displacement is a non-monotonous function with three distinct regions characterized by the change of the sign of the kurtosis slope. Specific structure of the kurtosis function depends on the characteristics of the potential: shape and size of the individual cells, heights and widths of the barriers but the change of sign feature is quite robust to the specifics of the potential. A number of other techniques have been proposed for identification of the potentials on biological membranes based on the analysis of individual trajectories. Most comprehensive approach is to solve the inverse problem of reconstruction of $U({\bf r})$ from the trajectories. However, such problem is generally ill-posed [@FokChouProcRoySoc2010] and needs very large statistics of trajectories. E.g., Ref.[@MassonVergassolaEtAlPRL2009] suggested to infer forces acting on the biomolecule [@AuthGovBiophysJ2009] requiring the multiple particle visits of each spatial location which is difficult to achieve experimentally. In addition, the potential $U$ in living cells can slowly change with time. This fact may limit the application of inverse problem approaches, that attempt to reconstruct the specific potential. Another approach [@KusumiSakoYamamotoBiophysJ1993; @KusumiAnnRevBio2005] focuses on identification of the potential barriers from MSD-based analysis. That approach is successful but requires very high temporal resolution of trajectories. One more method is based on the measurement of autocorrelation of SPT trajectories and recovering of the probability distribution of particle jumps [@YingHuertaSteinbergZunigaBullMathBiol2009] which indirectly displays the information about the inhomogeneity of the plasma membrane

\tau_h$. As the number of hops becomes very large for $ t \gg \tau_h$, the distribution become Gaussian again, due to cardinal limit theorem. The kurtosis decays back to zero. analytic solution for the kurtosis behavior in this region will be report elsewhere. To reason, we have proposed and analyzed a novel particle - tracking overture for identification of nonlinear interactions. Unlike coarse MSD techniques, our approach is based on the analysis of non - Gaussian characteristics of the atom dynamics, specifically the kurtosis of the displacement probability distribution. The functional dependence of the kurtosis on the measurement fourth dimension carries a lot of data about the nonlinear interactions that put up to the particle movement. E.g., if a particle is placed in a cage - character potential induced by cytoskeleton or transmembrane proteins, the resulting kurtosis of the displacement is a non - monotonous function with three distinct regions characterized by the change of the sign of the kurtosis slope. Specific structure of the kurtosis function depends on the characteristics of the electric potential: human body and size of the individual cells, acme and width of the barriers but the change of sign feature is quite robust to the specifics of the potential. A number of early techniques have been proposed for identification of the potentials on biological membranes based on the analysis of individual trajectories. Most comprehensive approach is to solve the inverse problem of reconstruction of $ U({\bf r})$ from the trajectories. However, such trouble is generally ill - pose [ @FokChouProcRoySoc2010 ] and need very large statistics of trajectories. E.g., Ref.[@MassonVergassolaEtAlPRL2009 ] suggested to understand forces acting on the biomolecule [ @AuthGovBiophysJ2009 ] requiring the multiple particle visits of each spatial location which is difficult to achieve experimentally. In addition, the likely $ U$ in living cells can slowly change with time. This fact may restrict the application of inverse trouble approaches, that attempt to reconstruct the specific potential. Another overture [ @KusumiSakoYamamotoBiophysJ1993; @KusumiAnnRevBio2005 ] focuses on identification of the likely barriers from MSD - based analysis. That approach is successful but requires very high temporal resolution of trajectories. One more method is based on the measurement of autocorrelation of SPT trajectories and recovering of the probability distribution of atom jumps [ @YingHuertaSteinbergZunigaBullMathBiol2009 ] which indirectly displays the data about the inhomogeneity of the plasma membrane

\tak_h$. As the number of hops becomes very lcege foc $t \gg \fau_h$, the distribution becomes Gaussien atain, eue to central limit tfeorem. Thv kurtosiw derays back to zero. Analytleal rsdultv for the kurtoxis behavimr in this reghov cill be reported elsewhere. To concludq, we hafe proposed and wnaljzqd a novel particle-tracking approach fkr idenuification of nonlonear interactions. Unlike fommln MSD techniques, lur approacy is vased on the analysis of non-Gaussizn characteristics of the partizle dvnamics, spexidicwnly the kurvosis jf the displacement [robabikity distributlon. Tie fynctional dependence mf the kurtosis on the measgrzment time carries a oor of hnfosmatkin xboht tge nonpinxar interacfions that xontribute to the psrepvle motion. E.f., if a pwrticle is placed in a cage-type potentidl jnduced by cytoskeleton or transmembrane projeins, the wesulting kurtosis of the displacement is a non-mototonkjs ymkctiub aith three distinct regions characterized by ege cmange of the sigk of the kurtosis xllpr. Specific strocture of fhe kurtosis functlon depgnds ob the chawactrristics of the potential: syape and sizv of the individual cepls, heights and widtns of the barriers but che chznge of sigj feature ks quite robust go nhe vpecifics of the potential. A number of pther tdchnoques rave been oroposed for identificatlon oy the potentiald on biological membranes based on the analysos of individbal trsjectories. Moft comprehensire approcch is to solve nhe inverve problem jf reconstrucdlon of $U({\bf r})$ from thq trqjecrories. Fuwever, such prpblem is ytnerally iol-posed [@FokChouProgRoySud2010] and needs verv oarge statisticx ow twauertoriqv. E.g., Ref.[@MassmnVefgarxolaEgAlPRL2009] suggcstdd tp infer forces actinc on the biomolecule [@AitmGovBiophisJ2009] requiwing the multople particle visius of xach s'atial losation which is difficult to adhieve exoerlmentally. In wddibion, the potencial $U$ in living cells can slowly change with time. This faxt may limit the apkligation of interse [roblem a[proaches, that attempt to reconstruct the specific potentizl. Anodher wpproach [@KusumiSakoYamamotoBiophysJ1993; @KusumiAnnRevBio2005] focuses on identificarion oh ehe potentizl bsrriegs yrok MSD-bwsed enalysis. That appgoach is successful but requires very hich temporal resolution of trajrcgories. One mote method is based on the jeasuremrnt of autocorrelation of SPT ttajectories and recoverikg of the probzbilitj dmstribution of partivle jukps [@YungHuerraStcinbergZunigaBjllKabhBiol2009] whifh iidixectly displays the information avout tye inhomogeneity oh the plasma oembrzne

\tau_h$. As the number of hops becomes for \gg \tau_h$, distribution becomes Gaussian theorem. kurtosis decays back zero. Analytical results the kurtosis behavior in this region be reported elsewhere. To conclude, we have proposed and analyzed a novel particle-tracking for identification of nonlinear interactions. Unlike common MSD techniques, our approach is based the of characteristics the particle dynamics, specifically the kurtosis of the displacement probability distribution. The functional dependence of the on the measurement time carries a lot of about the nonlinear interactions contribute to the particle motion. if particle is in cage-type induced by cytoskeleton transmembrane proteins, the resulting kurtosis of the displacement is a non-monotonous function with three distinct regions characterized the change sign of kurtosis Specific of the kurtosis on the characteristics of the potential: of the individual cells, heights and widths of barriers but change of sign feature is quite to the specifics of the potential. A number other techniques have been proposed for identification of the potentials on biological membranes based on of individual trajectories. Most approach is to the problem reconstruction $U({\bf r})$ the trajectories. However, such problem is generally ill-posed [@FokChouProcRoySoc2010] and needs large statistics of trajectories. E.g., Ref.[@MassonVergassolaEtAlPRL2009] suggested to infer forces the [@AuthGovBiophysJ2009] requiring the particle visits of each location is difficult to achieve addition, potential cells slowly with time. This fact limit the application of inverse approaches, that attempt to approach [@KusumiSakoYamamotoBiophysJ1993; @KusumiAnnRevBio2005] focuses on identification of the barriers from MSD-based analysis. That approach is but requires very high temporal resolution of trajectories. One more method is on the autocorrelation of SPT trajectories and recovering of the distribution of particle jumps which indirectly displays the information about the inhomogeneity of plasma

\tau_h$. As the number of hops becoMes very larGe for $T \gg \Tau_H$, tHe diStriBution becomes GAUssiAn again, due to central limIt theOrEM. The KUrTosis Decays bACk TO ZerO. ANaLytIcAL rEsultS foR the kurTosis behavIor In This region wiLL bE reported eLseWhere. To conclUde, We have PrOpoSEd and AnaLyzed A novel PArticlE-tracking ApPRoach fOR identiFICaTion Of nonlinear interaCTiONs. Unlike common mSD tecHnIQuES, Our AppRoach is basEd On the ANalysis OF nON-gAusSIan characteriStics of the pARtiCle dynAmIcs, SPecifiCally ThE KurTosis of the dIsplAcement prObabilITy distrIBution. THe funcTioNal DepeNDeNcE of ThE KurTOsIs oN The MeasuremEnT tIme caRrieS A LOT of iNfoRmatIon abOut the nonlineAr iNterACtiOns thAt conTribUtE to thE partiCle moTiOn. E.g., if a particle Is plAced in a caGe-tYpE poTeNtial INduced By cYtoSkeletoN or tranSMemBrANE PrOteins, the resulting KuRTOsIs of the dIsplacEMeNt IS a non-monOtOnoUs fuNCTion wIth tHReE distincT regioNS cHaRacteriZeD by the ChAngE of The siGN of tHe kurtOsis slopE. SpecIFic structure of THe kurtosis funCTiON DePEnds On tHe characterIstiCS of tHe poTEnTiaL: Shape And siZe OF tHE individual cells, heiGhTs and wIdths Of the barriers But the chanGE OF sign feaTure IS qUIte robust to the SpeciFics of the pOTential. A NumbeR of other TechniqueS HAve been pRopOseD foR idENTiFication of the POTentIaLs on bioLogIcal memBraNes BasEd oN tHe analysiS of indivIdUaL tRaJecTorieS. most compReHenSiVe aPproaCH is to sOlve tHe inVeRsE ProBlem of rECoNSTrucTiOn Of $U({\bF r})$ fRoM the tRajeCTorIes. HoweVer, such prOblEM is gEnErAlly ill-Posed [@FokChouPRoCRoySoc2010] and NeEds Very laRGE statistIcs of trajectories. E.g., Ref.[@MASsonVerGasSolaETAlPrL2009] suggestEd tO infer ForCEs actiNg on thE biomOlEcuLE [@authGOVbiOphYsj2009] requiring THE muLtiplE pArtiCle visiTs of each spatial locATioN which is diffiCulT to aCHIeVe eXPeRImeNtALly. iN Addition, the poteNtial $U$ in liViNG cElls can sloWLy cHaNge with Time. ThiS fact MAy limit The applicAtion of inVeRse pROBleM approacheS, that attEmpt to recONstruCT tHe speCifIc poteNtIal. anothEr apprOAch [@kusumISakoYAmAmotoBIophySJ1993; @kusumiAnNRevBio2005] focuses on identifIcatioN of thE poTential baRriERs fRom MSD-basEd anAlysis. That AppRoaCh is sUccESsful But rEQuIreS Very hIgh tEMporal resOLuTioN OF tRajectories. oNE MorE methOd iS Based oN the Measurement of autoCOrrelation of SPt traJECtoRieS And rEcOvering of the prObaBiLITy distriBuTion of partiCle jumps [@yiNGHuerTaSteiNbergZUnigaBuLLmaTHBiol2009] wHich IndIrectly diSplAyS The infoRmAtIOn abouT the InHomogeNeity oF The pLASma membrane

\tau_h$. As the number of hops beco mes v ery la rg e fo r $t \gg \tau_h$,t he d istribution becomes Ga ussia na gain , d ue to centra l l i m itth eo rem .T he kurt osi s decay s back tozer o. Analyticalr es ults for t hekurtosis beh avi or inth isr egion wi ll be repor t ed els ewhere. To conclu d e, we h a v eprop osed and analyzed an ovel particle- tracki ng ap p r oac h f or identif ic ation of nonl i ne a r int e ractions. Unl ike commonM SDtechni qu es, our ap proac hi s b ased on the ana lysis ofnon-Ga u ssian c h aracter istics of th e pa r ti cl e d yn a mic s ,spe c ifi cally th eku rtosi s of t h e dis pla ceme nt pr obability dis tri buti o n.The f uncti onal d epend ence o f the k urtosis on themeas urement t ime c arr ie s a l o t of i nfo rma tion ab out the non li n e a rinteractions thatco n t ri bute tothe pa r ti cl e motion. E .g. , if a part icle is placedin a c a ge -t ype pot en tial i nd uce d b y cyt o skel eton o r transm embra n e proteins, th e resulting ku r to s i so f th e d isplacement isa non -mon o to nou s func tionwi t ht hree distinct regio ns chara cteri zed by the ch ange of th e s ign of t he k u rt o sis slope. Spe cific structure of the k urtos is funct ion depen d s on thecha rac ter ist i c sof the potent i a l: s ha pe andsiz e of th e i ndi vid ual c ells, hei ghts and w id th softhe b a rriers b ut th echa nge o f signfeatu re i squ i terobustt ot h e sp ec if icsofth e pot enti a l.A numbe r of othe r t e chni qu es have b een proposedfo r identifi ca tio n of t h e potenti als on biological membr a nes bas edon th e an alysis of in dividu alt raject ories. Most c omp r e hensi v e a ppr oa ch is to s o l vethe i nv erse proble m of reconstructio n of $U({\bf r})$ fr om t h e t raj e ct o rie s. How e v er, such proble m is gener al l yill-posed[ @Fo kC houProc RoySoc2 010]a nd need s very la rge stati st icso f tr ajectories . E.g.,Ref.[@Mas s onVer g as solaE tAl PRL200 9] su ggest ed toi nfe r for ces ac ti ng onthe b io molecule [@AuthGovBiophysJ2009] requi ringthe multiple pa r tic le visits ofeach spati alloc ation wh i ch is dif f ic ult to ac hiev e experime n ta lly . In addition,t h e po tenti al$ U$ inlivi ng cells can slow l y change withtime . Thi s f a ct m ay limit the app lic at i o n of inv er se problemapproach es , that attem pt toreconst r u ct the sp ecif icpotential . A no t her app ro ac h [@Kus umiS ak oYamam otoBio p hysJ 1 9 93; @KusumiAnnRe vBio2 0 0 5] fo c use s onid entific a tion of the po tential bar riersfrom MSD- based a na lysis. Th at approachi s success ful b ut requ ir es v ery hightemp o r al re solu ti onof trajec t o ri e s. O n e m oremetho dis b ased on t h e measur eme n t of au to cor r e lation of S PT traject ori es an d recovering of t h ep robab ilitydistri butiono f p ar ticle j ump s [@YingHue rtaSteinb e rgZ un igaB ullMathB io l200 9] wh ic h indirectly di splays t he infor m a tion a bou t the inhomo ge neity of the plas ma membran e

\tau_h$._As the_number of hops becomes_very large_for_$t \gg_\tau_h$,_the distribution becomes_Gaussian again, due_to central limit theorem._The kurtosis decays_back_to zero. Analytical results for the kurtosis behavior in this region will be reported_elsewhere. To_conclude, we_have_proposed_and analyzed a novel particle-tracking_approach for identification of nonlinear_interactions. Unlike_common MSD techniques, our approach is based on_the_analysis of non-Gaussian_characteristics of the particle dynamics, specifically the kurtosis of_the displacement probability distribution. The functional_dependence of the_kurtosis_on_the measurement time carries_a lot of information about the_nonlinear interactions that contribute to the_particle motion. E.g., if a particle is_placed in a cage-type potential induced_by cytoskeleton or transmembrane proteins,_the resulting_kurtosis of the displacement is_a non-monotonous function_with three_distinct regions characterized_by the change of the sign_of the kurtosis_slope. Specific structure of the kurtosis_function_depends on the_characteristics_of_the potential:_shape and size_of_the individual_cells,_heights and widths of the barriers_but_the change of sign feature is quite_robust to the specifics_of_the potential. A number_of other techniques have been_proposed for identification of the potentials_on biological_membranes based_on the analysis of individual trajectories. Most comprehensive approach is to_solve the inverse problem of reconstruction_of $U({\bf r})$ from_the trajectories._However,_such problem is_generally_ill-posed [@FokChouProcRoySoc2010]_and needs very large statistics of trajectories._E.g., Ref.[@MassonVergassolaEtAlPRL2009]_suggested to infer forces acting on_the biomolecule [@AuthGovBiophysJ2009] requiring_the_multiple particle visits of each spatial_location which is difficult to achieve_experimentally. In addition, the potential_$U$_in_living cells can slowly change_with time. This fact may limit_the application of_inverse problem approaches, that attempt to reconstruct_the_specific potential. Another approach [@KusumiSakoYamamotoBiophysJ1993; @KusumiAnnRevBio2005]_focuses_on identification of the potential barriers_from_MSD-based_analysis. That approach is successful_but requires very high temporal resolution_of trajectories. One more method is based on the_measurement of autocorrelation_of SPT trajectories and recovering_of_the_probability distribution of particle jumps [@YingHuertaSteinbergZunigaBullMathBiol2009] which indirectly displays the_information about_the inhomogeneity of_the plasma membrane

^{(\delta_r-\delta_{r,n})t}-(n+1)e^{(\delta_r-\delta_{r,n+1})t}$ and observe that the inequality reduces to $\psi(t)\geq 0$ for $t>0$. To show that this is in fact true, one can then verify that: - $\psi(0)=0$, - $\lim_{t\to\infty}\psi(t)>0$, - $\psi'(0)=(n+1)\delta_{r,n+1}-n\delta_{r,n}-\delta_r>0$ and - $\psi'(t)=0$ for at most one $t\in(0,\infty)$. To prove the third bullet point, we can write out the expression explicitly and use the addition theorem for the sine function: $$\begin{gathered} 2(n+1)\sin\left(\pi\left(\frac1{n+1}+\frac{r-1}{2r}\right)\right)-2n\sin\left(\pi\left(\frac1n+\frac{r-1}{2r}\right)\right)-2\sin\left(\pi\frac{r-1}{2r}\right)\\ =2\left[(n+1)\sin\left(\frac{\pi}{n+1}\right)-n\sin\left(\frac{\pi}{n}\right)\right]\cos\left(\pi\frac{r-1}{2r}\right)\\+2\left[(n+1)\cos\left(\frac{\pi}{n+1}\right)-n\cos\left(\frac{\pi}{n}\right)-1\right]\sin\left(\pi\frac{r-1}{2r}\right).\end{gathered}$$ One can now show that the expressions in square brackets are positive and conclude that the whole expression is positive. This proves the second inequality. Using the explicit formula for the Rips magnitude, we now have: $$|t C^{\mathrm{eucl}}_n|_{\mathrm{Rips}}=\sum_{\text{odd }r|n}\frac nr (e^{-\delta_r t}-e^{-\delta_{r,n}t})+e^{-\delta_n t}\leq\sum_{\text{odd }r\leq n}\frac nr (e^{-\delta_r t}-e

^{(\delta_r-\delta_{r, n})t}-(n+1)e^{(\delta_r-\delta_{r, n+1})t}$ and observe that the inequality reduces to $ \psi(t)\geq 0 $ for $ t>0$. To show that this is in fact true, one can then control that: - $ \psi(0)=0 $, - $ \lim_{t\to\infty}\psi(t)>0 $, - $ \psi'(0)=(n+1)\delta_{r, n+1}-n\delta_{r, n}-\delta_r>0 $ and - $ \psi'(t)=0 $ for at about one $ t\in(0,\infty)$. To prove the third bullet item, we can compose out the expression explicitly and use the summation theorem for the sine routine: $ $ \begin{gathered } 2(n+1)\sin\left(\pi\left(\frac1{n+1}+\frac{r-1}{2r}\right)\right)-2n\sin\left(\pi\left(\frac1n+\frac{r-1}{2r}\right)\right)-2\sin\left(\pi\frac{r-1}{2r}\right)\\ = 2\left[(n+1)\sin\left(\frac{\pi}{n+1}\right)-n\sin\left(\frac{\pi}{n}\right)\right]\cos\left(\pi\frac{r-1}{2r}\right)\\+2\left[(n+1)\cos\left(\frac{\pi}{n+1}\right)-n\cos\left(\frac{\pi}{n}\right)-1\right]\sin\left(\pi\frac{r-1}{2r}\right).\end{gathered}$$ One can now prove that the expressions in square bracket are positive and conclude that the hale formula is positive. This proves the second inequality. Using the explicit formula for the Rips order of magnitude, we now have: $ $ |t C^{\mathrm{eucl}}_n|_{\mathrm{Rips}}=\sum_{\text{odd } r|n}\frac nr (e^{-\delta_r t}-e^{-\delta_{r, n}t})+e^{-\delta_n t}\leq\sum_{\text{odd } r\leq n}\frac nr (e^{-\delta_r t}-e

^{(\delha_r-\delta_{r,n})t}-(n+1)e^{(\delta_r-\delta_{v,n+1})t}$ and observe jhqt the inequzlity reauces to $\psi(t)\geq 0$ for $t>0$. To siow rhat ukis is in fact true, une can tjen veridy tiat: - $\psi(0)=0$, - $\lim_{t\vk\infty}\psi(t)>0$, - $\ldi'(0)=(n+1)\dzlva_{r,n+1}-n\delta_{r,n}-\delja_r>0$ and - $\psh'(t)=0$ for at most ove $t\in(0,\infty)$. To prove the third bullet [oint, wr fan write out jhe eqpwessjon explicitly and use the additioh theortm for the sine fumction: $$\begin{gathered} 2(n+1)\sin\levt(\pi\peft(\frac1{n+1}+\frac{r-1}{2r}\rigjt)\right)-2n\sin\oeft(\[u\left(\frac1n+\frxc{r-1}{2r}\right)\rpyht)-2\sin\left(\pj\frac{r-1}{2r}\right)\\ =2\left[(n+1)\sin\left(\frac{\pi}{n+1}\fight)-u\sin\left(\frax{\pu}{n}\rlcht)\right]\cos\oeft(\pp\frac{r-1}{2r}\right)\\+2\lcgt[(n+1)\cos\neft(\frav{\pi}{n+1}\right)-n\cos\lcft(\frec{\pi}{b}\right)-1\right]\sin\left(\pi\fcac{r-1}{2r}\right).\end{gathereq}$$ One can nkw show that the wxpresshons in rwuafe grecksts arf pksitive ans conclude rhat the whole exprtssyin is positivs. This pwoves the second inequality. Using the exklicif formula for the Rips nagnitude, we now have: $$|t C^{\mathri{eucl}}_n|_{\mathrm{Rips}}=\sum_{\text{odd }r|n}\frac nr (e^{-\delta_r t}-e^{-\denta_{r,n}v})+e^{-\aelua_k g}\oee\sum_{\text{odd }r\leq n}\frac nr (e^{-\delta_r t}-e

^{(\delta_r-\delta_{r,n})t}-(n+1)e^{(\delta_r-\delta_{r,n+1})t}$ and observe that the inequality reduces 0$ $t>0$. To that this is then that: - $\psi(0)=0$, $\lim_{t\to\infty}\psi(t)>0$, - $\psi'(0)=(n+1)\delta_{r,n+1}-n\delta_{r,n}-\delta_r>0$ - $\psi'(t)=0$ for at most one To prove the third bullet point, we can write out the expression explicitly use the addition theorem for the sine function: $$\begin{gathered} 2(n+1)\sin\left(\pi\left(\frac1{n+1}+\frac{r-1}{2r}\right)\right)-2n\sin\left(\pi\left(\frac1n+\frac{r-1}{2r}\right)\right)-2\sin\left(\pi\frac{r-1}{2r}\right)\\ =2\left[(n+1)\sin\left(\frac{\pi}{n+1}\right)-n\sin\left(\frac{\pi}{n}\right)\right]\cos\left(\pi\frac{r-1}{2r}\right)\\+2\left[(n+1)\cos\left(\frac{\pi}{n+1}\right)-n\cos\left(\frac{\pi}{n}\right)-1\right]\sin\left(\pi\frac{r-1}{2r}\right).\end{gathered}$$ One can show the in brackets are positive and conclude that the whole expression is positive. This proves the second inequality. the explicit formula for the Rips magnitude, we have: $$|t C^{\mathrm{eucl}}_n|_{\mathrm{Rips}}=\sum_{\text{odd }r|n}\frac (e^{-\delta_r t}-e^{-\delta_{r,n}t})+e^{-\delta_n t}\leq\sum_{\text{odd }r\leq n}\frac (e^{-\delta_r

^{(\delta_r-\delta_{r,n})t}-(n+1)e^{(\delta_r-\delTa_{r,n+1})t}$ and obServe ThaT thE iNequAlitY reduces to $\psi(t)\GEq 0$ foR $t>0$. To show that this is in faCt truE, oNE can THeN veriFy that: - $\pSI(0)=0$, - $\lIM_{T\to\InFtY}\psI(t)>0$, - $\PSi'(0)=(N+1)\deltA_{r,n+1}-N\delta_{r,N}-\delta_r>0$ and - $\Psi'(T)=0$ fOr at most one $t\IN(0,\iNfty)$. To provE thE third bullet PoiNt, we caN wRitE Out thE exPressIon expLIcitly And use the AdDItion tHEorem foR THe Sine Function: $$\begin{gathEReD} 2(N+1)\sin\left(\pi\left(\Frac1{n+1}+\fRaC{R-1}{2r}\RIGht)\RigHt)-2n\sin\left(\Pi\Left(\fRAc1n+\frac{R-1}{2R}\rIGHT)\riGHt)-2\sin\left(\pi\frAc{r-1}{2r}\right)\\ =2\leFT[(n+1)\sIn\left(\FrAc{\pI}{N+1}\right)-N\sin\lEfT(\FraC{\pi}{n}\right)\riGht]\cOs\left(\pi\fRac{r-1}{2r}\rIGht)\\+2\left[(N+1)\Cos\left(\Frac{\pi}{N+1}\riGht)-N\cos\LEfT(\fRac{\Pi}{N}\RigHT)-1\rIghT]\Sin\Left(\pi\frAc{R-1}{2r}\Right).\End{gATHERed}$$ ONe cAn noW show That the expresSioNs in SQuaRe braCkets Are pOsItive And conClude ThAt the whole expreSsioN is positiVe. THiS prOvEs the SEcond iNeqUalIty. UsinG the expLIciT fORMUlA for the Rips magnituDe, WE NoW have: $$|t C^{\mAthrm{eUCl}}_N|_{\mAThrm{Rips}}=\SuM_{\teXt{odD }R|N}\frac Nr (e^{-\dELtA_r t}-e^{-\deltA_{r,n}t})+e^{-\dELtA_n T}\leq\sum_{\TeXt{odd }r\LeQ n}\fRac Nr (e^{-\deLTa_r t}-E

^{(\delta_r-\delta_{r,n})t }-(n+1)e^{ (\del ta_ r-\ de lta_ {r,n +1})t}$ and ob s erve that the inequality r educe st o $\ p si (t)\g eq 0$ f o r$ t >0$ .To sh ow th at th isis in f act true,one c an then veri f ythat: - $\ psi(0)=0$, - $\lim _{ t\t o \inft y}\ psi(t )>0$,- $\ psi'(0)=( n+ 1 )\delt a _{r,n+1 } - n\ delt a_{r,n}-\delta_r> 0 $a nd - $\psi' (t)=0$ f o ra t mo stone $t\in( 0, \inft y )$. To pr o v e th e third bullet point, wec anwriteou t t h e expr essio ne xpl icitly andusethe addit ion th e orem fo r the si ne fun cti on: $$\ b eg in {ga th e red } 2 (n+ 1 )\s in\left( \p i\ left( \fra c 1 { n +1}+ \fr ac{r -1}{2 r}\right)\rig ht) -2n\ s in\ left( \pi\l eft( \f rac1n +\frac {r-1} {2 r}\right)\right )-2\ sin\left( \pi \f rac {r -1}{2 r }\righ t)\ \ = 2\left[ (n+1)\s i n\l ef t ( \ fr ac{\pi}{n+1}\right )- n \ si n\left(\ frac{\ p i} {n } \right)\ ri ght ]\co s \ left( \pi\ f ra c{r-1}{2 r}\rig h t) \\ +2\left [( n+1)\c os \le ft( \frac { \pi} {n+1}\ right)-n \cos\ l eft(\frac{\pi} { n}\right)-1\r i gh t ] \s i n\le ft( \pi\frac{r- 1}{2 r }\ri ght) . \e nd{ g ather ed}$$ O n ec an now show that th eexpres sions in square br ackets are p o sitive a nd c o nc l ude that the w holeexpression is posit ive.This pro ves the s e c ond ineq ual ity . Usi n g t he explicit f o r mula f or theRip s magni tud e,wenow h ave: $$|t C^{\mat hr m{ eu cl }}_ n|_{\ m athrm{Ri ps }}= \s um_ {\tex t {odd } r|n}\ frac n r( e^{ -\delta _ rt } -e^{ -\ de lta_ {r, n} t})+e ^{-\ d elt a_n t}\ leq\sum_{ \te x t{od d}r \leq n} \frac nr (e^{ -\ delta_r t} -e

^{(\delta_r-\delta_{r,n})t}-(n+1)e^{(\delta_r-\delta_{r,n+1})t}$ and_observe that_the inequality reduces to_$\psi(t)\geq 0$_for_$t>0$. To_show_that this is_in fact true,_one can then verify_that: - _$\psi(0)=0$, -_ $\lim_{t\to\infty}\psi(t)>0$, - $\psi'(0)=(n+1)\delta_{r,n+1}-n\delta_{r,n}-\delta_r>0$ and - $\psi'(t)=0$ for at most one_$t\in(0,\infty)$. To_prove the_third_bullet_point, we can write out_the expression explicitly and use_the addition_theorem for the sine function: $$\begin{gathered} 2(n+1)\sin\left(\pi\left(\frac1{n+1}+\frac{r-1}{2r}\right)\right)-2n\sin\left(\pi\left(\frac1n+\frac{r-1}{2r}\right)\right)-2\sin\left(\pi\frac{r-1}{2r}\right)\\ =2\left[(n+1)\sin\left(\frac{\pi}{n+1}\right)-n\sin\left(\frac{\pi}{n}\right)\right]\cos\left(\pi\frac{r-1}{2r}\right)\\+2\left[(n+1)\cos\left(\frac{\pi}{n+1}\right)-n\cos\left(\frac{\pi}{n}\right)-1\right]\sin\left(\pi\frac{r-1}{2r}\right).\end{gathered}$$ One can_now_show that the_expressions in square brackets are positive and conclude that_the whole expression is positive. This_proves the second_inequality. Using_the_explicit formula for the_Rips magnitude, we now have: $$|t_C^{\mathrm{eucl}}_n|_{\mathrm{Rips}}=\sum_{\text{odd }r|n}\frac nr (e^{-\delta_r t}-e^{-\delta_{r,n}t})+e^{-\delta_n t}\leq\sum_{\text{odd_}r\leq n}\frac nr (e^{-\delta_r t}-e

of defect states in bipartite lattice systems [@Pol15], as has been exploited in recent topological lasers or non-linear limiters based on photonic Su-Schrieffer-Heeger structures [@StJ17; @Zha18; @Par18; @arXrei19a]. An additional attractive aspect of these anomalous features are that they tie topological effects together that are often seen as separate, due to the varied nature of the specific encountered spectral and spatial features that first come into focus. A prime example are flat bands, which have been observed in recent experiments focussing on Lieb lattices [@Guz14; @Muk15; @Die16; @Pol17] and one-dimensional counterparts [@Bab16], as well as suitably deformed graphene [@Lev10] and analogous quantum [@Gom12] and classical systems [@Rec13; @Wen19]. In the latter case they constitute pseudo-Landau levels arising from a synthetic magnetic field [@Gui10]. In particular, signatures of photonic pseudo-Landau levels have been detected by probing the edges of a honeycomb array of optical waveguides [@Rec13]. A second example are a class of helical edge states in reciprocal systems, as observed, e.g., in zigzag terminated graphene [@Fuj96; @Sal17]. While these bulk and edge phenomena do not naturally fall into the scope of standard topological band structure theory [@Has10], they are still intimately linked to wavefunctions with a characteristic sublattice polarization. This provides a promising perspective from which one can seek to develop very general unifying descriptions (see, e.g., Ref. [@Kun18] for a recent approach utilizing this perspective). In this work, we demonstrate experimentally for the case of photonic graphene-like systems that the anomalous edge and bulk phenomena tied to sublattice polarization are in fact directly linked. This is achieved by tracing the formation of pseudo-Landau levels all the way from vanishing to large pseudomagnetic field strengths. In particular, we report the direct observation of the spatially resolved sublattice polarization in the 0th pseudo-Landau level of strained photonic graphene, and trace it back to the unstrained case, where the system only possesses edge states. By observing a characteristic nodal structure for the higher-order levels, we can then establish a direct link to the supersymmetric Hamiltonian of the underlying low-energy theory [@Jac84]. Thereby, our observations connect a broad variety of topological phenomena to a unifying principle

of defect states in bipartite lattice systems [ @Pol15 ], as has been exploited in late topological laser or non - analogue limiters based on photonic Su - Schrieffer - Heeger structures [ @StJ17; @Zha18; @Par18; @arXrei19a ]. An extra attractive aspect of these anomalous features are that they tie topological effect together that are often see as separate, due to the varied nature of the specific encountered apparitional and spatial features that first occur into stress. A prime example are bland bands, which have been observed in late experiments focussing on Lieb lattices [ @Guz14; @Muk15; @Die16; @Pol17 ] and one - dimensional counterpart [ @Bab16 ], as well as suitably deform graphene [ @Lev10 ] and analogous quantum [ @Gom12 ] and classical system [ @Rec13; @Wen19 ]. In the latter case they constitute pseudo - Landau levels arising from a man-made magnetic field [ @Gui10 ]. In particular, signatures of photonic pseudo - Landau levels have been detected by probing the edges of a honeycomb array of optical waveguides [ @Rec13 ]. A second example are a class of helical edge states in reciprocal systems, as observe, e.g., in zigzag terminated graphene [ @Fuj96; @Sal17 ]. While these majority and boundary phenomena do not naturally fall into the scope of standard topological band social organization theory [ @Has10 ], they are however intimately linked to wavefunctions with a characteristic sublattice polarization. This provide a promising perspective from which one can seek to develop very general unifying description (see, e.g., Ref.   [ @Kun18 ] for a recent approach use this perspective). In this employment, we demonstrate experimentally for the case of photonic graphene - like systems that the anomalous edge and bulk phenomenon tied to sublattice polarization are in fact directly linked. This is achieve by tracing the geological formation of pseudo - Landau levels all the room from vanishing to big pseudomagnetic field strengths. In especial, we report the direct notice of the spatially resolved sublattice polarization in the 0th pseudo - Landau level of strained photonic graphene, and decipher it back to the unstrained case, where the system only possesses boundary states. By observing a characteristic nodal structure for the high - order levels, we can then establish a direct link to the supersymmetric Hamiltonian of the underlie low - energy hypothesis [ @Jac84 ]. Thereby, our observations connect a broad variety of topological phenomena to a unifying principle

of defect states in bipartlte lattice systgmw [@Pol15], es has geen expuoited in recent topological lqsers or non-linear limiters based on photonix Su-Wxhrieffer-Hxsger stvbctursd [@StL17; @Vha18; @Par18; @arXrei19a]. Sn additiotal attractive arpzct of these anomalous features are ehat thry tie topologicwl egsecta together that are often seen as aeparatt, due to the variec nature of the specific ejcoujtered spectral anf spatial fgztuwws that firsg come into focus. A prjme example are flat bands, whicf havz been obsetrwd lt recent ex'erimegts focussinn on Liab lattoces [@Guz14; @Muk15; @Dle16; @Pon17] abd one-dimensional couiterparts [@Bab16], as welj as suitdbmy deformed graphwnw [@Lev10] and anauigojs suenthm [@Gom12] anv classical systems [@Rex13; @Wen19]. In the latter cwwe they constjtute [squdo-Landau levels arising from a syntheuic mzgnetic field [@Gui10]. In paeticular, signatures ov photonis pseudo-Landau levels have been detected by probitg thx ddgts of a yojeycomb array of optical waveguides [@Rec13]. A secjhd eqample are a clasf of helicak fdbg states in reziproccm aystems, as observef, e.g., in zigzqg terminwted graphene [@Fuj96; @Sal17]. While thewe bulk and vdge phenomena do not uaturally faul imto tne scope of standard to'ologidal band stgucture tgdory [@Has10], they ard snill intimattuy linked to wavesunctions witk a charxctetistic fublattice polavhzation. This provifes a psomising pfrspective from which one can sxxk to develop vary general unifning descriptiogs (see, e.g., Ref. [@Kon18] for a xecent approach ltilizing this perspqctive). In this aork, we demoistrate evperumenrally fuf the case of lhotonic yxaphene-lije systems that thc anoozlous edge and yblj phenomena tiec tu stbpavtice [olarization are in gact airectly likkea. Thos is achieved by trdcinf the formation of pfeudo-Laneau levejs all the wau from vanishing tl larje psendomagmetyc field strengths. In particulzr, we replrt the direct jbsevvatyon of the spatially resolved sublattice polarizavion in the 0th pseudo-Labdau level of straiuec photonic jraphege, and trdce it back to the ubstrained case, whtre the system only posaesses edge states. By observing a characteristic nodal structure for the higher-order levels, we can theh esyablivh c direce liik to the supersymketric Hamiltonian of the underuying lof-euergy theory [@Jac84]. Thereby, our pbrervations covnect a broad variety of topologocal phenomena to a unifying pronciple

of defect states in bipartite lattice systems has exploited in topological lasers or Su-Schrieffer-Heeger [@StJ17; @Zha18; @Par18; An additional attractive of these anomalous features are that tie topological effects together that are often seen as separate, due to the nature of the specific encountered spectral and spatial features that first come into A example flat which have been observed in recent experiments focussing on Lieb lattices [@Guz14; @Muk15; @Die16; @Pol17] and counterparts [@Bab16], as well as suitably deformed graphene and analogous quantum [@Gom12] classical systems [@Rec13; @Wen19]. In latter they constitute levels from synthetic magnetic field In particular, signatures of photonic pseudo-Landau levels have been detected by probing the edges of a honeycomb of optical A second are class helical edge states systems, as observed, e.g., in zigzag @Sal17]. While these bulk and edge phenomena do naturally fall the scope of standard topological band theory [@Has10], they are still intimately linked to with a characteristic sublattice polarization. This provides a promising perspective from which one can seek very general unifying descriptions e.g., Ref. [@Kun18] a approach this In this we demonstrate experimentally for the case of photonic graphene-like systems that anomalous edge and bulk phenomena tied to sublattice polarization are directly This is achieved tracing the formation of levels the way from vanishing pseudomagnetic strengths. report direct of the spatially resolved polarization in the 0th pseudo-Landau of strained photonic graphene, the unstrained case, where the system only possesses states. By observing a characteristic nodal structure the higher-order levels, we can then establish a direct link to the Hamiltonian of low-energy theory [@Jac84]. Thereby, our observations connect a variety of topological phenomena a unifying principle

of defect states in bipartite Lattice sysTems [@POl15], aS haS bEen eXploIted in recent toPOlogIcal lasers or non-linear lImiteRs BAsed ON pHotonIc Su-SchRIeFFEr-HEeGeR stRuCTuRes [@Stj17; @ZhA18; @Par18; @arXRei19a]. An addiTioNaL attractive aSPeCt of these aNomAlous featureS arE that tHeY tiE TopolOgiCal efFects tOGether That are ofTeN Seen as SEparate, DUE tO the Varied nature of the SPeCIfic encountereD spectRaL AnD SPatIal Features thAt First COme into FOcUS. a PriME example are flAt bands, whicH HavE been oBsErvED in recEnt exPeRImeNts focussinG on LIeb latticEs [@Guz14; @MUK15; @Die16; @Pol17] ANd one-diMensioNal CouNterPArTs [@bab16], As WEll AS sUitABly Deformed GrApHene [@LEv10] anD ANALogoUs qUantUm [@Gom12] And classical sYstEms [@REC13; @WeN19]. In thE lattEr caSe They cOnstitUte psEuDo-Landau levels aRisiNg from a syNthEtIc mAgNetic FIeld [@GuI10]. In ParTicular, SignatuREs oF pHOTOnIc pseudo-Landau leveLs HAVe Been deteCted by PRoBiNG the edgeS oF a hOneyCOMb arrAy of OPtIcal waveGuides [@rEc13]. a sEcond exAmPle are A cLasS of HelicAL edgE stateS in recipRocal SYstems, as observED, e.g., in zigzag teRMiNATeD GrapHenE [@Fuj96; @Sal17]. WhilE theSE bulK and EDgE phENomenA do noT nATuRAlly fall into the scopE oF standArd toPological band Structure tHEORy [@Has10], theY are STiLL intimately linKed to WavefunctiONs with a cHaracTeristic SublatticE POlarizatIon. thiS prOviDES a Promising persPECtivE fRom whicH onE can seeK to DevEloP veRy General unIfying deScRiPtIoNs (sEe, e.g., REF. [@Kun18] for a ReCenT aPprOach uTIlizinG this PersPeCtIVe). IN this woRK, wE DEmonStRaTe exPerImEntalLy foR The Case of pHotonic grAphENe-liKe SyStems thAt the anomalouS eDge and bulk PhEnoMena tiED To sublatTice polarization are in faCT directLy lInked. this Is achieveD by TracinG thE FormatIon of pSeudo-laNdaU LEvels ALL tHe wAy From vanishING to Large PsEudoMagnetiC field strengths. In pARtiCular, we report The DireCT ObSerVAtIOn oF tHE spATIally resolved suBlattice poLaRIzAtion in the 0TH psEuDo-LandaU level oF straINed photOnic graphEne, and traCe It baCK To tHe unstrainEd case, whEre the sysTEm onlY PoSsessEs eDge staTeS. By ObserVing a cHAraCteriStic noDaL strucTure fOr The higheR-order levels, we can then esTablisH a dirEct Link to the SupERsyMmetric HaMiltOnian of the UndErlYing lOw-eNErgy tHeorY [@jaC84]. ThEReby, oUr obSErvations COnNecT A BrOad variety oF TOPolOgicaL phENomena To a uNifying principle

of defect states in bipar tite latti ce sy ste ms[@ Pol1 5],as has been ex p loit ed in recent topologic al la se r s or no n-lin ear lim i te r s ba se donph o to nic S u-S chrieff er-Heegerstr uc tures [@StJ1 7 ;@Zha18; @P ar1 8; @arXrei19 a]. An a dd iti o nal a ttr activ e aspe c t of t hese anom al o us fea t ures ar e th at t hey tie topologic a le ffects togethe r that a r eo f ten se en as sepa ra te, d u e to th e v a r i edn ature of thespecific en c oun teredsp ect r al and spat ia l fe atures that fir st come i nto fo c us. A p r ime exa mple a refla t ba n ds ,whi ch hav e b een obs erved in r ec ent e xper i m e n ts f ocu ssin g onLieb lattices [@ Guz1 4 ; @ Muk15 ; @Di e16; @ Pol17 ] andone-d im ensional counte rpar ts [@Bab1 6], a s w el l ass uitabl y d efo rmed gr aphene[ @Le v1 0 ] an d analogous quantu m[ @ Go m12] and class i ca ls ystems [ @R ec1 3; @ W e n19]. Int he lattercase t h ey c onstitu te pseud o- Lan dau leve l s ar isingfrom a s ynthe t ic magnetic fi e ld [@Gui10].I np a rt i cula r,signaturesof p h oton ic p s eu do- L andau leve ls ha v e been detected bypr obingthe e dges of a hon eycomb arr a y of optic al w a ve g uides [@Rec13] . A s econd exam p le are a clas s of hel ical edge s tates in re cip roc als y st ems, as obser v e d, e .g ., in z igz ag term ina ted gr aph en e [@Fuj96 ; @Sal17 ]. W hi le th ese b u lk and e dg e p he nom ena d o not n atura llyfa ll int o the s c op e of s ta nd ardtop ol ogica l ba n d s tructur e theory[@H a s10] ,th ey arestill intimat el y linked t owav efunct i o ns witha characteristic sublat t ice pol ari zatio n. T his provi des a pro mis i ng per specti ve fr om wh i c h one c an se ek to develo p ver y gen er al u nifying descriptions (see , e. g., Ref. [@Ku n18 ] fo r arec e nt app ro a chu t ilizing this pe rspective) .In this work , we d emonstr ate exp erime n tally f or the ca se of pho to nicg r aph ene-like s ystems t hat the a n omalo u sedgeand bulkph eno menatied t o su blatt ice po la rizati on ar ein factdirectly linked. This i s achi evedbytracing t hef orm ation ofpseu do-Landaulev els allthe way f romv an ish i ng to lar g e pseudom a gn eti c fi eld strengt h s . In part icu l ar, we rep ort the direct ob s ervation of th e sp a t ial lyr esol ve d sublattice p ola ri z a tion inth e 0th pseud o-Landau l e vel o f stra ined p hotonic g ra p hene,andtra ce it bac k t ot he unst ra in e d case , wh er e thesystem only p ossesses edge st ates. B y obs e rvi ng ach aracter i stic nodal str ucture forthe hi gher -orde r level s, we ca n t he n establis h a direct link to the s uper sym metric Ham i l tonia n of t heunderlyin g lo w -e ne r gytheo ry [@ Ja c84] . Thereby , our obs erv a tions c on nec t a broa d v a r iety of to pol ogica l phenomenat o au ni f yingprinci ple

of_defect states_in bipartite lattice systems_[@Pol15], as_has_been exploited_in_recent topological lasers_or non-linear limiters_based on photonic Su-Schrieffer-Heeger_structures [@StJ17; @Zha18;_@Par18;_@arXrei19a]. An additional attractive aspect of these anomalous features are that they tie topological effects_together_that are_often_seen_as separate, due to the_varied nature of the specific_encountered spectral_and spatial features that first come into focus._A_prime example are_flat bands, which have been observed in recent experiments_focussing on Lieb lattices [@Guz14; @Muk15;_@Die16; @Pol17] and_one-dimensional_counterparts_[@Bab16], as well as_suitably deformed graphene [@Lev10] and analogous_quantum [@Gom12] and classical systems [@Rec13;_@Wen19]. In the latter case they constitute_pseudo-Landau levels arising from a synthetic_magnetic field [@Gui10]. In particular,_signatures of_photonic pseudo-Landau levels have been_detected by probing_the edges_of a honeycomb_array of optical waveguides [@Rec13]. A_second example are_a class of helical edge states_in_reciprocal systems, as_observed,_e.g.,_in zigzag_terminated graphene [@Fuj96;_@Sal17]._While these_bulk_and edge phenomena do not naturally_fall_into the scope of standard topological band_structure theory [@Has10], they_are_still intimately linked to_wavefunctions with a characteristic sublattice_polarization. This provides a promising perspective_from which_one can_seek to develop very general unifying descriptions (see, e.g., Ref. [@Kun18] for_a recent approach utilizing this perspective). In_this work, we demonstrate_experimentally for_the_case of photonic_graphene-like_systems that_the anomalous edge and bulk phenomena tied_to sublattice_polarization are in fact directly linked._This is achieved by_tracing_the formation of pseudo-Landau levels all_the way from vanishing to large_pseudomagnetic field strengths. In particular,_we_report_the direct observation of the_spatially resolved sublattice polarization in the_0th pseudo-Landau level_of strained photonic graphene, and trace it_back_to the unstrained case, where the_system_only possesses edge states. By observing_a_characteristic_nodal structure for the higher-order_levels, we can then establish a_direct link to the supersymmetric Hamiltonian of the underlying_low-energy theory [@Jac84]._Thereby, our observations connect a_broad_variety_of topological phenomena to a unifying principle

}{2C_1-2C_2\cdot t}}\approx 1.0435$$ according to. This prediction is also shown on the figure as line (0). From this figure we see that at high resolutions all schemes produce the same growth of amplitude of perturbations, which is in agreement with the analytical prediction. As the resolution decreases the results for all schemes for all schemes begin to deviate from the analytical value. Among the method of lines schemes, the RK4 is the most accurate and is able to reproduce the correct behavior of the perturbation up to the resolution of 16 grid points per wavelength. CFLN also requires only 16 grid points per wavelength to reproduce the correct behavior. The other two schemes requires from 32 to 64 grid points per wavelength. The major difference between the CFLN and the method of lines schemes is that at low resolutions, CFLN is amplifying the perturbations more than it should according to the analytical predictions, whereas the method of lines schemes are damping. The RK4 is the least damping and the most accurate of these schemes. The behavior of the schemes at low resolutions is fully consistent with the results of the Von Neumann stability analysis, presented in section \[sec:2.01\]. According to that analysis the CFLN scheme must be amplifying at all wavelengths whereas the method of lines schemes must be damping. **Discussion and conclusion** ============================= In this paper we studied the properties of four numerical schemes, CFLN, ICN, RK3 and RK4, applied to a non-linear scalar wave equation. This equation has a number of non-trivial analytic solutions whose properties, including stability, were studied and summarized in section \[sec:theory\]. We carried out the Von Neumann stability analysis of the schemes and studied their phase and amplitude errors. Finally we carried out numerical experiments and compared the results of those experiments with the perturbation analysis of the equation and the Von Neumann stability analysis of the schemes. We find that all four schemes presented in this paper are stable and converge with the second order accuracy. The stability range for the schemes were determined and are presented in table \[tab:stableinterval\]. Those ranges are valid for both linear and non-linear solutions. For non-linear schemes we checked that the amplification factor $|\lambda|$ is less than $1+O(\Delta t)$ for sufficiently small $\Delta t$. With respect to the dissipation errors, we find that the CFLN has the least amount of dissipation

} { 2C_1 - 2C_2\cdot t}}\approx 1.0435$$ according to. This prediction is also shown on the figure as course (0). From this name we see that at high resolutions all dodge produce the same emergence of amplitude of perturbation, which is in agreement with the analytical prediction. As the resolving power decreases the results for all schema for all schemes begin to deviate from the analytic value. Among the method acting of lines schemes, the RK4 is the about accurate and is able to reproduce the right demeanor of the perturbation up to the resolution of 16 grid points per wavelength. CFLN besides requires only 16 power system points per wavelength to regurgitate the correct behavior. The other two schemes requires from 32 to 64 grid point per wavelength. The major difference between the CFLN and the method of lines schemes is that at low resolution, CFLN is amplifying the perturbations more than it should according to the analytical predictions, whereas the method of lines schemes are damping. The RK4 is the least damping and the about accurate of these scheme. The behavior of the system at depleted resolutions is fully consistent with the results of the Von Neumann stability analysis, presented in incision \[sec:2.01\ ]. According to that analysis the CFLN scheme must be amplifying at all wavelengths whereas the method of line schemes must be damping. * * Discussion and conclusion * * = = = = = = = = = = = = = = = = = = = = = = = = = = = = = In this paper we study the property of four numerical schemes, CFLN, ICN, RK3 and RK4, applied to a non - linear scalar wave equation. This equality has a number of non - trivial analytic solutions whose properties, including stability, were studied and summarized in department \[sec: theory\ ]. We carried out the Von Neumann stability psychoanalysis of the schemes and studied their phase and amplitude errors. Finally we behave out numerical experiment and compared the results of those experiment with the disruption analysis of the equation and the Von Neumann stability analysis of the schemes. We find that all four schemes presented in this paper are static and converge with the second order accuracy. The stability scope for the schemes were determined and are presented in table \[tab: stableinterval\ ]. Those ranges are valid for both linear and non - linear solution. For non - linear schemes we checked that the amplification factor $ |\lambda|$ is less than $ 1+O(\Delta t)$ for sufficiently small $ \Delta t$. With respect to the dissipation error, we witness that the CFLN has the least amount of profligacy

}{2C_1-2C_2\cfot t}}\approx 1.0435$$ according tu. This prediction is anso shkwn on tfe figure as line (0). From this higuee we see that at high resouutions apl schemws pcoduce the same jdowth on ampmltude if perturbatioks, which is in agreement figh the analytical prediction. As the rqsolutipn decreases the reslles fkg cll schemes for all schemes begjn to dtviate from the anslytical value. Among the mfthof of lines schemes, the RK4 is jge iist accurate and is able to reprodoce the correct behavior of the oertuxbation up jo tjg resolution of 16 drid points ivr wavenength. VFLN also requlres mnlt 16 grid points per watelength to reproducg the corracc behavior. The other rwi schgmes sequkees frkm 32 fo 64 grld 'oints per savelength. Rhe major differenct bqneeen the CFLH and ehq method of lines schemes is that at lof rssolutions, CFLN is amplufying the perturbatilns more ehan it should according to the analytical predicdions, dhexcwr tje method of lines schemes are damping. The RK4 js tme least damping and the moxt avsurate of there schzjea. The behavior of hhe schgmes ar low resjlutoons is fully consistent wirh the resulns od the Von Neumann dtability aualysix, prexented in section \[sec:2.01\]. Aecordihg to that wnalysis ffe CFLN scheme mjst ba amplifying at all waveleggths wheceas che methud og linef schemes lust na damping. **Discussioj and cmnclusion** ============================= Ij this paper we studied the pro'xrties of fout ngmegical schzmes, CNLN, ICN, RK3 and WK4, applied to c non-linzar scxlar wave vquation. Vhis equatiog has a numbes of non-triviel analytyc silutuons whure properties, oncluding stability, were studied and xumozrized in sectiin \[wec:theory\]. We catridd jun tie Vog Neumann statiligy xmalysks of the sgheoes snd studied their phdse znd amplitude errots. Finally we carryed out numerocal experiments ajd cokpaced thr rgsults of those experiments witg the perhurnation analysys on thg equation cnd the Von Neumann stability analysis oh the schemes. We find thqt all four schemes kresented in vhis pwper are vtable and converge qith the second ovder accuracy. The stabjlity sange for the schemes were determined and are presented in table \[tab:stableinteeval\]. Tiofe ranges ade vslid xox bpth ligear end non-linear solltions. For non-linear schemes we ciecked thad che amplification factor $|\lambca|$ is less than $1+O(\Delta t)$ for sufficientmy small $\Delta t$. With respect to the disxipation errors, we find bhat the XFLN gas thv lxast amount of dissilation

}{2C_1-2C_2\cdot t}}\approx 1.0435$$ according to. This prediction shown the figure line (0). From at resolutions all schemes the same growth amplitude of perturbations, which is in with the analytical prediction. As the resolution decreases the results for all schemes all schemes begin to deviate from the analytical value. Among the method of schemes, RK4 the accurate and is able to reproduce the correct behavior of the perturbation up to the resolution 16 grid points per wavelength. CFLN also requires 16 grid points per to reproduce the correct behavior. other schemes requires 32 64 points per wavelength. major difference between the CFLN and the method of lines schemes is that at low resolutions, CFLN amplifying the than it according the predictions, whereas the lines schemes are damping. The RK4 damping and the most accurate of these schemes. behavior of schemes at low resolutions is fully with the results of the Von Neumann stability presented in section \[sec:2.01\]. According to that analysis the CFLN scheme must be amplifying at whereas the method of schemes must be **Discussion conclusion** In paper we the properties of four numerical schemes, CFLN, ICN, RK3 and RK4, to a non-linear scalar wave equation. This equation has a non-trivial solutions whose properties, stability, were studied and in \[sec:theory\]. We carried out Neumann analysis and their and amplitude errors. Finally carried out numerical experiments and the results of those of the equation and the Von Neumann stability of the schemes. We find that all schemes presented in this paper are stable and converge with the second accuracy. The for the schemes were determined and are presented table \[tab:stableinterval\]. Those ranges valid for both linear and non-linear solutions. For non-linear we that the factor $|\lambda|$ is than $1+O(\Delta t)$ sufficiently small $\Delta respect to dissipation we CFLN has the least amount of

}{2C_1-2C_2\cdot t}}\approx 1.0435$$ according to. THis predictIon is AlsO shOwN on tHe fiGure as line (0). From THis fIgure we see that at high reSolutIoNS all SChEmes pRoduce tHE sAME grOwTh Of aMpLItUde of PerTurbatiOns, which is In aGrEement with thE AnAlytical prEdiCtion. As the reSolUtion dEcReaSEs the ResUlts fOr all sCHemes fOr all scheMeS Begin tO Deviate FROm The aNalytical value. AmoNG tHE method of lines SchemeS, tHE Rk4 IS thE moSt accurate AnD is abLE to reprODuCE THe cORrect behavior Of the perturBAtiOn up to ThE reSOlutioN of 16 grId POinTs per waveleNgth. cFLN also rEquireS Only 16 griD Points pEr waveLenGth To rePRoDuCe tHe COrrECt BehAVioR. The otheR tWo SchemEs reQUIREs frOm 32 tO 64 griD poinTs per wavelengTh. THe maJOr dIfferEnce bEtweEn The CFlN and tHe metHoD of lines schemes Is thAt at low reSolUtIonS, CfLN is AMplifyIng The PerturbAtions mORe tHaN IT ShOuld according to the AnALYtIcal predIctionS, WhErEAs the metHoD of LineS SChemeS are DAmPing. The Rk4 is the LEaSt Damping AnD the moSt AccUraTe of tHEse sChemes. the behavIor of THe schemes at low REsolutions is fULlY COnSIsteNt wIth the resulTs of THe VoN NeuMAnN stABilitY analYsIS, pREsented in section \[sec:2.01\]. acCordinG to thAt analysis the cFLN scheme MUST be ampliFyinG At ALl wavelengths wHereaS the method OF lines scHemes Must be daMping. **DiscUSSion and cOncLusIon** ============================= in tHIS pAper we studied THE proPeRties of FouR numeriCal SchEmeS, CFlN, iCN, RK3 and Rk4, applied To A nOn-LiNeaR scalAR wave equAtIon. thIs eQuatiON has a nUmber Of noN-tRiVIal AnalytiC SoLUTionS wHoSe prOpeRtIes, inCludINg sTabilitY, were studIed ANd suMmArIzed in sEction \[sec:theoRy\]. we carried oUt The von NeuMANn stabilIty analysis of the schemes ANd studiEd tHeir pHase And amplitUde Errors. finALly we cArried Out nuMeRicAL ExperIMEnTs aNd Compared thE REsuLts of ThOse eXperimeNts with the perturbaTIon Analysis of the EquAtioN ANd The vOn nEumAnN StaBILity analysis of tHe schemes. WE fINd That all fouR SchEmEs preseNted in tHis paPEr are stAble and coNverge witH tHe seCONd oRder accuraCy. The staBility ranGE for tHE sChemeS weRe deteRmIneD and aRe presENteD in taBle \[tab:StAbleinTervaL\]. THose rangEs are valid for both linear And non-LineaR soLutions. FoR noN-LinEar schemeS we cHecked that The AmpLificAtiON factOr $|\laMBdA|$ is LEss thAn $1+O(\DELta t)$ for suFFiCieNTLy Small $\Delta t$. wITH reSpect To tHE dissiPatiOn errors, we find thaT The CFLN has the lEast AMOunT of DIssiPaTion

}{2C_1-2C_2\cdot t}}\appro x 1.0435$$ acco rdi ngto . Th is p rediction is a l so s hown on the figure asline(0 ) . F r om this figure we s eeth at at h i gh reso lut ions al l schemespro du ce the sameg ro wth of amp lit ude of pertu rba tions, w hic h is i n a greem ent wi t h theanalytica lp redict i on. Ast h ereso lution decreasest he results for al l sche me s f o r al l s chemes beg in to d e viate f r om t h e a n alytical valu e. Among th e me thod o flin e s sche mes,th e RK 4 is the mo st a ccurate a nd isa ble tor eproduc e thecor rec t be h av io r o ft hep er tur b ati on up to t he reso luti o n o f 16 gr id p oints per waveleng th. CFL N al so re quire s on ly 16 g rid po intspe r wavelength to rep roduce th e c or rec tbehav i or. Th e o the r two s chemesr equ ir e s fr om 32 to 64 grid p oi n t sper wave length . T he major di ff ere nceb e tween the CF LN and t he met h od o f lines s chemes i s t hat at l o w re soluti ons, CFL N isa mplifying thep erturbationsm or e th a n it sh ould accord ingt o th e an a ly tic a l pre dicti on s ,w hereas the method o flinesschem es are dampin g. The RK4 i s the lea st d a mp i ng and the mos t acc urate of t h ese sche mes.The beha vior of t h e schemes at lo w r eso l u ti ons is fullyc o nsis te nt with th e resul tsofthe Vo nNeumann s tability a na ly si s,prese n ted in s ec tio n\[s ec:2. 0 1\]. A ccord ingto t h atanalysi s t h e CFL Nsc heme mu st be a mpli f yin g at al l wavelen gth s whe re as the me thod of lines s chemes mus tbedampin g . **Disc ussion and conclusion** ======= === ===== ==== ========= = In thi s p a per we studi ed th epro p e rties o ffou rnumericals c hem es, C FL N, I CN, RK3 and RK4, appliedt o a non-linear s cal ar w a v eequ a ti o n.Th i s e q u ation has a num ber of non -t r iv ial analyt i c s ol utionswhose p roper t ies, in cluding s tability, w eres t udi ed and sum marizedin sectio n \[se c :t heory \]. We ca rr ied outthe Vo n Ne umann stabi li ty ana lysis o f the sc hemes and studied their phase andamp litude er ror s . F inally we car ried out n ume ric al ex per i ments and co mpa r ed th e re s ults of t h os e e x p er iments with t h e p ertur bat i on ana lysi s of the equation and the Von Ne uman n sta bil i ty a na lysis of the s che me s . We fin dthat all fo ur schem es prese nted i n this papera r es tableandcon verge wit h t he secondor de r accur acy. T he sta bility rang e for the schemeswered e termi n edand a re presen t ed i n table \[ tab:stablei nterva l\]. Thos e range sare va lid f or both li n ear and n on-li near so lu tion s. For n on-l i n ear s chem es we checkedt h at th ea mpl ific ation f acto r $|\lamb d a|$ is l ess than $1 +O (\D e l ta t)$ fo r sufficient lysmall $ \Delta t$. Wit h r e spect to th e diss ipation err or s, we f ind t hat the C FLN has t h e l ea st a mount of d issi pa tio n

}{2C_1-2C_2\cdot t}}\approx_1.0435$$ according_to. This prediction is_also shown_on_the figure_as_line (0). From this_figure we see_that at high resolutions_all schemes produce_the_same growth of amplitude of perturbations, which is in agreement with the analytical prediction._As_the resolution_decreases_the_results for all schemes for_all schemes begin to deviate_from the_analytical value. Among the method of lines schemes,_the_RK4 is the_most accurate and is able to reproduce the correct_behavior of the perturbation up to_the resolution of_16_grid_points per wavelength. CFLN_also requires only 16 grid points_per wavelength to reproduce the correct_behavior. The other two schemes requires from_32 to 64 grid points per_wavelength. The major difference between_the CFLN_and the method of lines_schemes is that_at low_resolutions, CFLN is_amplifying the perturbations more than it_should according to_the analytical predictions, whereas the method_of_lines schemes are_damping._The_RK4 is_the least damping_and_the most_accurate_of these schemes. The behavior of_the_schemes at low resolutions is fully consistent_with the results of_the_Von Neumann stability analysis,_presented in section \[sec:2.01\]. According_to that analysis the CFLN scheme_must be_amplifying at_all wavelengths whereas the method of lines schemes must be damping. **Discussion_and conclusion** ============================= In this paper we studied_the properties of four_numerical schemes,_CFLN,_ICN, RK3 and_RK4,_applied to_a non-linear scalar wave equation. This equation_has a_number of non-trivial analytic solutions whose_properties, including stability, were_studied_and summarized in section \[sec:theory\]. We_carried out the Von Neumann stability_analysis of the schemes and_studied_their_phase and amplitude errors. Finally_we carried out numerical experiments and_compared the results_of those experiments with the perturbation analysis_of_the equation and the Von Neumann_stability_analysis of the schemes. We find that_all_four_schemes presented in this paper_are stable and converge with the_second order accuracy. The stability range for the schemes_were determined and_are presented in table \[tab:stableinterval\]._Those_ranges_are valid for both linear and non-linear solutions. For non-linear schemes_we checked_that the amplification_factor $|\lambda|$ is less than $1+O(\Delta t)$ for sufficiently small_$\Delta t$. With respect to the dissipation errors,_we find that the CFLN has the least amount of dissipation

,q,\lambda}(\mathbb{G}):=\{f \in L^{1}_{loc}(\mathbb{G}):\|f\|_{\mathfrak{L}_{p,q,\lambda}(\mathbb{G})}< \infty \}, \quad \lambda \in \mathbb{R},$$ where $$\|f\|_{\mathfrak{L}_{p,q,\lambda}(\mathbb{G})}:=\underset{R>0}{\rm sup}\left(\int_{\mathbb{G}}\left(|x|^\frac{Q}{p} \left|\log\frac{R}{|x|}\right|^{\lambda}|f-f_{R}|\right)^{q}\frac{dx}{|x|^{Q}}\right)^{\frac{1}{q}}.$$ For $p=\infty$ we define $$\|f\|_{\mathfrak{L}_{\infty,q,\lambda}(\mathbb{G})}:=\underset{R>0}{\rm sup}\left(\int_{\mathbb{G}}\left( \left|\log\frac{R}{|x|}\right|^{\lambda}|f-f_{R}|\right)^{q}\frac{dx}{|x|^{Q}}\right)^{\frac{1}{q}},$$ where $f_{R}(x):=f(R\frac{x}{|x|})$. Moreover, we define the Lorentz-Zygmund type spaces $\mathfrak{L}_{p,q,\lambda_{1},\lambda_{2}}(\mathbb{G})$ by $$\label{LZdef}\mathfrak{L}_{p,q,\lambda_{1},\lambda_{2}}(\mathbb{G}):=\{f \in L^{1}_{loc}(\mathbb{G}):\|f\|_{\mathfrak{L}_{p,q,\lambda_{1},\lambda_{2}}(\mathbb{G})}< \infty \},$$ where $$\begin{gathered} \|f\|_{\mathfrak{L}_{p,q,\lambda_{1},\lambda_{2}}(\mathbb{G})}:= \underset{R>0}{\rm sup}\left(\int_{\mathbb{G}}\left(|x|^\frac{Q}{p} \left|\log\frac{e R}{|x|}\right|^{\lambda_{1}}\left|\log\left|\log\frac{e R}{|x|}\right|\right|^{\lambda_{2}} \right.\right. \\ \left.\left. \times \left(\chi_{B(0,eR)}(x)|f-f_{R}| +\chi_{B^{

, q,\lambda}(\mathbb{G}):=\{f \in L^{1}_{loc}(\mathbb{G}):\|f\|_{\mathfrak{L}_{p, q,\lambda}(\mathbb{G }) } < \infty \ }, \quad \lambda \in \mathbb{R},$$ where $ $ \|f\|_{\mathfrak{L}_{p, q,\lambda}(\mathbb{G})}:=\underset{R>0}{\rm sup}\left(\int_{\mathbb{G}}\left(|x|^\frac{Q}{p } \left|\log\frac{R}{|x|}\right|^{\lambda}|f - f_{R}|\right)^{q}\frac{dx}{|x|^{Q}}\right)^{\frac{1}{q}}.$$ For $ p=\infty$ we define $ $ \|f\|_{\mathfrak{L}_{\infty, q,\lambda}(\mathbb{G})}:=\underset{R>0}{\rm sup}\left(\int_{\mathbb{G}}\left ( \left|\log\frac{R}{|x|}\right|^{\lambda}|f - f_{R}|\right)^{q}\frac{dx}{|x|^{Q}}\right)^{\frac{1}{q}},$$ where $ f_{R}(x):=f(R\frac{x}{|x|})$. Moreover, we define the Lorentz - Zygmund type space $ \mathfrak{L}_{p, q,\lambda_{1},\lambda_{2}}(\mathbb{G})$ by $ $ \label{LZdef}\mathfrak{L}_{p, q,\lambda_{1},\lambda_{2}}(\mathbb{G}):=\{f \in L^{1}_{loc}(\mathbb{G}):\|f\|_{\mathfrak{L}_{p, q,\lambda_{1},\lambda_{2}}(\mathbb{G }) } < \infty \},$$ where $ $ \begin{gathered } \|f\|_{\mathfrak{L}_{p, q,\lambda_{1},\lambda_{2}}(\mathbb{G})}:= \underset{R>0}{\rm sup}\left(\int_{\mathbb{G}}\left(|x|^\frac{Q}{p } \left|\log\frac{e R}{|x|}\right|^{\lambda_{1}}\left|\log\left|\log\frac{e R}{|x|}\right|\right|^{\lambda_{2 } } \right.\right. \\ \left.\left. \times \left(\chi_{B(0,eR)}(x)|f - f_{R}| + \chi_{B^ {

,q,\lalbda}(\mathbb{G}):=\{f \in L^{1}_{loc}(\mathnb{G}):\|f\|_{\mathfrak{L}_{p,q,\lcnbda}(\mavhbb{G})}< \ihfty \}, \quxd \lambda \in \mathbb{R},$$ where $$\|f\|_{\larhfraj{L}_{p,q,\lambda}(\mathbb{G})}:=\underret{R>0}{\rm sui}\left(\int_{\mqthbu{G}}\left(|x|^\frac{Q}{p} \lefv|\mog\frac{V}{|r|}\righf|^{\pambba}|h-f_{R}|\right)^{q}\frac{dx}{|w|^{Q}}\right)^{\frac{1}{x}}.$$ For $p=\infty$ wa aeyine $$\|f\|_{\mathfrak{L}_{\infty,q,\lambda}(\mathbb{G})}:=\unqerset{R>0}{\tm sup}\left(\int_{\matrbb{G}}\kqft( \lsft|\log\frac{R}{|x|}\right|^{\lambda}|f-f_{R}|\right)^{q}\frzc{dx}{|x|^{Q}}\rpght)^{\frac{1}{q}},$$ where $f_{T}(x):=f(R\frac{x}{|x|})$. Moreover, we defije tje Lorentz-Zygmund hype spaces $\matrdrak{L}_{p,q,\lambdx_{1},\lambda_{2}}(\mathbb{G})$ by $$\labem{LZdef}\mathfrak{L}_{p,q,\lambda_{1},\lambda_{2}}(\maghbb{G}):=\{y \in L^{1}_{loc}(\majkvb{G}):\|v\|_{\kathfrak{L}_{p,q,\oambdw_{1},\lambda_{2}}(\mathbn{B})}< \inftf \},$$ wherr $$\begin{gatherec} \|f\|_{\methfeak{L}_{p,q,\lambda_{1},\lambda_{2}}(\matibb{G})}:= \underset{R>0}{\rm sup}\lgft(\int_{\mathtb{Y}}\left(|x|^\frac{Q}{p} \left|\log\frqc{w R}{|x|}\rhght|^{\nambaq_{1}}\lewt|\lkg\kert|\log\fgac{x R}{|x|}\right|\rifht|^{\lambda_{2}} \rught.\right. \\ \left.\left. \uimqw \left(\chi_{B(0,eR)}(x)|f-r_{R}| +\chi_{B^{

,q,\lambda}(\mathbb{G}):=\{f \in L^{1}_{loc}(\mathbb{G}):\|f\|_{\mathfrak{L}_{p,q,\lambda}(\mathbb{G})}< \infty \}, \quad \lambda where sup}\left(\int_{\mathbb{G}}\left(|x|^\frac{Q}{p} \left|\log\frac{R}{|x|}\right|^{\lambda}|f-f_{R}|\right)^{q}\frac{dx}{|x|^{Q}}\right)^{\frac{1}{q}}.$$ $p=\infty$ we define Moreover, define the Lorentz-Zygmund spaces $\mathfrak{L}_{p,q,\lambda_{1},\lambda_{2}}(\mathbb{G})$ by \in L^{1}_{loc}(\mathbb{G}):\|f\|_{\mathfrak{L}_{p,q,\lambda_{1},\lambda_{2}}(\mathbb{G})}< \infty \},$$ where $$\begin{gathered} \underset{R>0}{\rm sup}\left(\int_{\mathbb{G}}\left(|x|^\frac{Q}{p} \left|\log\frac{e R}{|x|}\right|^{\lambda_{1}}\left|\log\left|\log\frac{e R}{|x|}\right|\right|^{\lambda_{2}} \right.\right. \\ \left.\left. \times \left(\chi_{B(0,eR)}(x)|f-f_{R}| +\chi_{B^{

,q,\lambda}(\mathbb{G}):=\{f \in L^{1}_{loc}(\mathBb{G}):\|f\|_{\mathfrAk{L}_{p,q,\LamBda}(\MaThbb{g})}< \infTy \}, \quad \lambda \in \MAthbB{R},$$ where $$\|f\|_{\mathfrak{L}_{p,q,\lamBda}(\maThBB{G})}:=\unDErSet{R>0}{\rM sup}\lefT(\InT_{\MAthBb{g}}\lEft(|X|^\fRAc{q}{p} \lefT|\loG\frac{R}{|x|}\Right|^{\lambdA}|f-f_{r}|\rIght)^{q}\frac{dx}{|x|^{q}}\RiGht)^{\frac{1}{q}}.$$ FoR $p=\iNfty$ we define $$\|F\|_{\maThfrak{l}_{\iNftY,Q,\lambDa}(\mAthbb{g})}:=\underSEt{R>0}{\rm sUp}\left(\int_{\MaTHbb{G}}\leFT( \left|\loG\FRaC{R}{|x|}\rIght|^{\lambda}|f-f_{R}|\righT)^{Q}\fRAc{dx}{|x|^{Q}}\right)^{\fraC{1}{q}},$$ wherE $f_{r}(X):=f(r\FRac{X}{|x|})$. MOreover, we dEfIne thE lorentz-zYgMUND tyPE spaces $\mathfrAk{L}_{p,q,\lambda_{1},\LAmbDa_{2}}(\mathBb{g})$ by $$\LAbel{LZDef}\maThFRak{l}_{p,q,\lambda_{1},\laMbda_{2}}(\Mathbb{G}):=\{f \iN L^{1}_{loc}(\mAThbb{G}):\|f\|_{\mAThfrak{L}_{P,q,\lambDa_{1},\lAmbDa_{2}}(\maTHbB{G})}< \InfTy \},$$ WHerE $$\BeGin{GAthEred} \|f\|_{\matHfRaK{L}_{p,q,\lAmbdA_{1},\LAMBda_{2}}(\mAthBb{G})}:= \uNdersEt{R>0}{\rm sup}\left(\iNt_{\mAthbB{g}}\leFt(|x|^\frAc{Q}{p} \lEft|\lOg\Frac{e r}{|x|}\righT|^{\lambDa_{1}}\Left|\log\left|\log\fRac{e r}{|x|}\right|\riGht|^{\LaMbdA_{2}} \rIght.\rIGht. \\ \lefT.\leFt. \tImes \lefT(\chi_{B(0,eR)}(X)|F-f_{R}| +\ChI_{b^{

,q,\lambda}(\mathbb{G}):=\ {f \in L^{ 1}_{l oc} (\m at hbb{ G}): \|f\|_{\mathfr a k{L} _{p,q,\lambda}(\mathbb {G})} <\ inft y \ }, \q uad \la m bd a \in \ ma thb b{ R }, $$ wh ere $$\|f\ |_{\mathfr ak{ L} _{p,q,\lambd a }( \mathbb{G} )}: =\underset{R >0} {\rm s up }\l e ft(\i nt_ {\mat hbb{G} } \left( |x|^\frac {Q } {p} \l e ft|\log \ f ra c{R} {|x|}\right|^{\la m bd a }|f-f_{R}|\rig ht)^{q }\ f ra c { dx} {|x |^{Q}}\rig ht )^{\f r ac{1}{q } }. $ $ For $p=\infty$ we define $$\ | f\| _{\mat hf rak { L}_{\i nfty, q, \ lam bda}(\mathb b{G} )}:=\unde rset{R > 0}{\rms up}\lef t(\int _{\ mat hbb{ G }} \l eft (\ lef t |\ log \ fra c{R}{|x| }\ ri ght|^ {\la m b d a }|f- f_{ R}|\ right )^{q}\frac{dx }{| x|^{ Q }}\ right )^{\f rac{ 1} {q}}, $$ whe re $f _{ R}(x):=f(R\frac {x}{ |x|})$. Mor eo ver ,we de f ine th e L ore ntz-Zyg mund ty p e s pa c e s $ \mathfrak{L}_{p,q, \l a m bd a_{1},\l ambda_ { 2} }( \ mathbb{G }) $ b y $$ \ l abel{ LZde f }\ mathfrak {L}_{p , q, \l ambda_{ 1} ,\lamb da _{2 }}( \math b b{G} ):=\{f \in L^{ 1}_{l o c}(\mathbb{G}) : \|f\|_{\mathf r ak { L }_ { p,q, \la mbda_{1},\l ambd a _{2} }(\m a th bb{ G })}<\inft y\ }, $ $ where $$\begin{ga th ered}\|f\| _{\mathfrak{L }_{p,q,\la m b d a_{1},\l ambd a _{ 2 }}(\mathbb{G}) }:= \ underset{R > 0}{\rm s up}\l eft(\int _{\mathbb { G }}\left( |x| ^\f rac {Q} { p }\left|\log\fr a c {e R }{ |x|}\ri ght |^{\lam bda _{1 }}\ lef t| \log\left |\log\fr ac {e R }{ |x| }\rig h t|\right |^ {\l am bda _{2}} \right .\rig ht.\\ \ l eft .\left. \t i m es \ le ft (\ch i_{ B( 0,eR) }(x) | f-f _{R}| + \chi_{B^{

,q,\lambda}(\mathbb{G}):=\{f \in_L^{1}_{loc}(\mathbb{G}):\|f\|_{\mathfrak{L}_{p,q,\lambda}(\mathbb{G})}< \infty_\}, \quad \lambda \in_\mathbb{R},$$ where_$$\|f\|_{\mathfrak{L}_{p,q,\lambda}(\mathbb{G})}:=\underset{R>0}{\rm_sup}\left(\int_{\mathbb{G}}\left(|x|^\frac{Q}{p} \left|\log\frac{R}{|x|}\right|^{\lambda}|f-f_{R}|\right)^{q}\frac{dx}{|x|^{Q}}\right)^{\frac{1}{q}}.$$ For_$p=\infty$_we define $$\|f\|_{\mathfrak{L}_{\infty,q,\lambda}(\mathbb{G})}:=\underset{R>0}{\rm_sup}\left(\int_{\mathbb{G}}\left( \left|\log\frac{R}{|x|}\right|^{\lambda}|f-f_{R}|\right)^{q}\frac{dx}{|x|^{Q}}\right)^{\frac{1}{q}},$$ where $f_{R}(x):=f(R\frac{x}{|x|})$. Moreover,_we define the Lorentz-Zygmund_type spaces $\mathfrak{L}_{p,q,\lambda_{1},\lambda_{2}}(\mathbb{G})$_by_$$\label{LZdef}\mathfrak{L}_{p,q,\lambda_{1},\lambda_{2}}(\mathbb{G}):=\{f \in L^{1}_{loc}(\mathbb{G}):\|f\|_{\mathfrak{L}_{p,q,\lambda_{1},\lambda_{2}}(\mathbb{G})}< \infty \},$$ where $$\begin{gathered} \|f\|_{\mathfrak{L}_{p,q,\lambda_{1},\lambda_{2}}(\mathbb{G})}:= \underset{R>0}{\rm sup}\left(\int_{\mathbb{G}}\left(|x|^\frac{Q}{p} \left|\log\frac{e R}{|x|}\right|^{\lambda_{1}}\left|\log\left|\log\frac{e R}{|x|}\right|\right|^{\lambda_{2}} \right.\right. \\ \left.\left. \times \left(\chi_{B(0,eR)}(x)|f-f_{R}| +\chi_{B^{

{S}_{\text{gf}}$ and assume that the background field theory is gauge-fixed in the way just explained. Background-preserving canonical transformations ----------------------------------------------- It is useful to characterize the most general canonical transformations $\Phi,{\mkern2mu\underline{\mkern-2mu\smash{\Phi }\mkern-2mu}\mkern2mu },K,{\mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu }\rightarrow \Phi ^{\prime },{\mkern2mu\underline{\mkern-2mu\smash{\Phi }\mkern-2mu}\mkern2mu }^{\prime },K^{\prime },{\mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu }^{\prime }$ that preserve the background field master equations (\[treide\]) and the basic properties of $\hat{S}$ and $\bar{S}$. By definition, all canonical transformations preserve the antiparentheses, so (\[treide\]) are turned into $$\llbracket \hat{S}^{\prime },\hat{S}^{\prime }\rrbracket ^{\prime }=\llbracket \hat{S}^{\prime },\bar{S}^{\prime }\rrbracket ^{\prime }=\llbracket \bar{S}^{\prime },\bar{S}^{\prime }\rrbracket ^{\prime }=0. \label{tretre}$$ Moreover, $\hat{S}^{\prime }$ should be ${\mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu }^{\prime }$ and ${\mkern2mu\underline{\mkern-2mu\smash{C}\mkern-2mu}\mkern2mu }^{\prime }$ independent, while $\bar{S}$ should be invariant, because it encodes the background transformations. This means $$\bar{S}^{\prime }(\Phi,{\mkern2mu\underline{\mkern-2mu\smash{\Phi }\mkern-2mu}\mkern2mu },K,{\mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu })=\bar{S}(\Phi,{\mkern2mu\underline{\mkern-2mu\smash{\Phi }\mkern-2

{ S}_{\text{gf}}$ and assume that the background field theory is bore - situate in the way just excuse. backdrop - preserving canonical transformation ----------------------------------------------- It is utilitarian to characterize the most general canonical transformations $ \Phi,{\mkern2mu\underline{\mkern-2mu\smash{\Phi } \mkern-2mu}\mkern2mu }, K,{\mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu } \rightarrow \Phi ^{\prime }, { \mkern2mu\underline{\mkern-2mu\smash{\Phi } \mkern-2mu}\mkern2mu } ^{\prime }, K^{\prime }, { \mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu } ^{\prime } $ that keep the background field victor equations (\[treide\ ]) and the basic properties of $ \hat{S}$ and $ \bar{S}$. By definition, all canonical transformations preserve the antiparentheses, so (\[treide\ ]) are sour into $ $ \llbracket \hat{S}^{\prime }, \hat{S}^{\prime } \rrbracket ^{\prime } = \llbracket \hat{S}^{\prime }, \bar{S}^{\prime } \rrbracket ^{\prime } = \llbracket \bar{S}^{\prime }, \bar{S}^{\prime } \rrbracket ^{\prime } = 0. \label{tretre}$$ Moreover, $ \hat{S}^{\prime } $ should be $ { \mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu } ^{\prime } $ and $ { \mkern2mu\underline{\mkern-2mu\smash{C}\mkern-2mu}\mkern2mu } ^{\prime } $ independent, while $ \bar{S}$ should be invariant, because it encode the background transformations. This mean $ $ \bar{S}^{\prime } (\Phi,{\mkern2mu\underline{\mkern-2mu\smash{\Phi } \mkern-2mu}\mkern2mu }, K,{\mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu }) = \bar{S}(\Phi,{\mkern2mu\underline{\mkern-2mu\smash{\Phi } \mkern-2

{S}_{\tedt{gf}}$ and assume that the background field theocy is gzuge-fixea in the way just explained. Beckgeound-kgeserving canonical tfansformanions ----------------------------------------------- It iw ustful to charactermae the most gskeral ranonical transnormations $\[hi,{\mkern2mu\undeslknz{\mkern-2mu\smash{\Phi }\mkern-2mu}\mkern2mu },K,{\mkewn2mu\undrrpine{\mkern-2mu\smafh{K}\mlqrn-2mh}\mkern2mu }\rightarrow \Phi ^{\prime },{\mkerh2mu\undegline{\mkern-2mu\smash{\Lhi }\mkern-2mu}\mkern2mu }^{\prime },K^{\orimf },{\mkern2mu\underline{\lkern-2mu\smasy{K}\mkqen-2mu}\mkern2mu }^{\orime }$ thau 'reserve thg background field master equatiuns (\[txeide\]) and tye badhc propertixs of $\rat{S}$ and $\bar{S}$. By defitition, sll canonical branshormqtions preserve the aitiparentheses, so (\[trgide\]) are tgrued into $$\llbracket \har{S}^{\prime },\hat{V}^{\priow }\rfbrzclef ^{\primf }=\lmbracket \hzt{S}^{\prime },\bae{S}^{\prime }\rrbracket ^{\ptiiv }=\llbracket \bzr{S}^{\priie },\bar{S}^{\prime }\rrbracket ^{\prime }=0. \label{tretre}$$ Korsover, $\hat{S}^{\prime }$ should be ${\mkern2mu\underline{\mnern-2mu\smafh{K}\mkern-2mu}\mkern2mu }^{\prime }$ and ${\mkern2mu\underline{\mkert-2mu\smesf{C}\mjevn-2mu}\ojegn2mu }^{\prime }$ independent, while $\bar{S}$ should be yhvsrpant, because it ekcodes the backgroinf ytansformations. This meahs $$\bar{S}^{\prime }(\Phi,{\mkfrn2mu\unqerlibe{\mkern-2mu\fmasn{\Phi }\mkern-2mu}\mkern2mu },K,{\mkern2my\underline{\mkvrn-2my\smash{K}\mkern-2mu}\mkeru2mu })=\bar{S}(\Phi,{\myern2ku\undrrline{\mkern-2mu\smash{\Phi }\myern-2

{S}_{\text{gf}}$ and assume that the background field gauge-fixed the way explained. Background-preserving canonical to the most general transformations $\Phi,{\mkern2mu\underline{\mkern-2mu\smash{\Phi }\mkern-2mu}\mkern2mu }\rightarrow \Phi ^{\prime },{\mkern2mu\underline{\mkern-2mu\smash{\Phi }\mkern-2mu}\mkern2mu }^{\prime },{\mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu }^{\prime }$ that preserve the background field master equations (\[treide\]) and the properties of $\hat{S}$ and $\bar{S}$. By definition, all canonical transformations preserve the antiparentheses, (\[treide\]) turned $$\llbracket },\hat{S}^{\prime }\rrbracket ^{\prime }=\llbracket \hat{S}^{\prime },\bar{S}^{\prime }\rrbracket ^{\prime }=\llbracket \bar{S}^{\prime },\bar{S}^{\prime }\rrbracket ^{\prime }=0. \label{tretre}$$ Moreover, }$ should be ${\mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu }^{\prime }$ and ${\mkern2mu\underline{\mkern-2mu\smash{C}\mkern-2mu}\mkern2mu }$ independent, while $\bar{S}$ be invariant, because it encodes background This means }(\Phi,{\mkern2mu\underline{\mkern-2mu\smash{\Phi },K,{\mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu }\mkern-2

{S}_{\text{gf}}$ and assume that the baCkground fiEld thEorY is GaUge-fIxed In the way just exPLainEd. Background-preserving CanonIcAL traNSfOrmatIons ----------------------------------------------- It iS UsEFUl tO cHaRacTeRIzE the mOst General Canonical tRanSfOrmations $\Phi,{\MKeRn2mu\underlIne{\Mkern-2mu\smash{\phi }\Mkern-2mU}\mKerN2Mu },K,{\mkErn2Mu\undErline{\MKern-2mu\Smash{K}\mkeRn-2MU}\mkern2MU }\rightaRROw \phi ^{\pRime },{\mkern2mu\underlINe{\MKern-2mu\smash{\Phi }\Mkern-2mU}\mKErN2MU }^{\prIme },k^{\prime },{\mkerN2mU\undeRLine{\mkeRN-2mU\SMAsh{k}\Mkern-2mu}\mkern2mU }^{\prime }$ that pREseRve the BaCkgROund fiEld maStER eqUations (\[treiDe\]) anD the basic ProperTIes of $\haT{s}$ and $\bar{s}$. By defIniTioN, all CAnOnIcaL tRAnsFOrMatIOns Preserve ThE aNtipaRentHESES, so (\[tReiDe\]) arE turnEd into $$\llbrackEt \hAt{S}^{\pRIme },\Hat{S}^{\pRime }\rRbraCkEt ^{\priMe }=\llbrAcket \HaT{S}^{\prime },\bar{S}^{\primE }\rrbRacket ^{\priMe }=\lLbRacKeT \bar{S}^{\PRime },\baR{S}^{\pRimE }\rrbracKet ^{\primE }=0. \LabEl{TRETrE}$$ Moreover, $\hat{S}^{\prime }$ ShOULd Be ${\mkern2mU\underLInE{\mKErn-2mu\smaSh{k}\mkErn-2mU}\MKern2mU }^{\priME }$ aNd ${\mkern2mU\underLInE{\mKern-2mu\sMaSh{C}\mkeRn-2Mu}\mKerN2mu }^{\prIMe }$ inDependEnt, while $\Bar{S}$ sHOuld be invarianT, Because it encoDEs THE bACkgrOunD transformaTionS. this MeanS $$\BaR{S}^{\pRIme }(\PhI,{\mkerN2mU\UnDErline{\mkern-2mu\smash{\PHi }\Mkern-2mU}\mkerN2mu },K,{\mkern2mu\unDerline{\mkeRN-2MU\smash{K}\mKern-2MU}\mKErn2mu })=\bar{S}(\Phi,{\mkErn2mu\Underline{\mKErn-2mu\smaSh{\Phi }\Mkern-2

{S}_{\text{gf}}$ and assum e that the back gro und f ield the ory is gauge-f i xedin the way just explai ned. B a ckgr o un d-pre serving ca n o nic al t ran sf o rm ation s - ------- ---------- --- -- ------------ - -- --------- It is useful t o c haract er ize the m ost gene ral ca n onical transfor ma t ions $ \ Phi,{\m k e rn 2mu\ underline{\mkern- 2 mu \ smash{\Phi }\m kern-2 mu } \m k e rn2 mu},K,{\mker n2 mu\un d erline{ \ mk e r n -2m u \smash{K}\mke rn-2mu}\mke r n2m u }\ri gh tar r ow \Ph i ^{\ pr i me},{\mkern2m u\un derline{\ mkern- 2 mu\smas h {\Phi } \mkern -2m u}\ mker n 2m u}^{ \p r ime }, K^{ \ pri me },{\m ke rn 2mu\u nder l i n e {\mk ern -2mu \smas h{K}\mkern-2m u}\ mker n 2mu }^{\ prime }$th at pr eserve theba ckground fieldmast er equati ons ( \[t re ide\] ) and t hebas ic prop ertieso f $ \h a t { S} $ and $\bar{S}$. By d ef inition, all c a no ni c al trans fo rma tion s prese rvet he antipar enthes e s, s o (\[tr ei de\])ar e t urn ed in t o $$ \llbra cket \ha t{S}^ { \prime },\hat{ S }^{\prime }\r r br a c ke t ^{\ pri me }=\llbra cket \hat {S}^ { \p rim e },\b ar{S} ^{ \ pr i me }\rrbracket ^{\p ri me }=\ llbra cket \bar{S}^ {\prime }, \ b a r{S}^{\p rime }\ r rbracket ^{\pr ime } =0. \label { tretre}$ $ Mor eover, $ \hat{S}^{ \ p rime }$sho uld be ${ \ m ke rn2mu\underli n e {\mk er n-2mu\s mas h{K}\mk ern -2m u}\ mke rn 2mu }^{\p rime }$an d${ \m ker n2mu\ u nderline {\ mke rn -2m u\sma s h{C}\m kern- 2mu} \m ke r n2m u }^{\p r im e }$ i nd ep ende nt, w hile$\ba r {S} $ shoul d be inva ria n t, b ec au se it e ncodes the ba ck ground tra ns for mation s . This me ans $$\bar{S}^{\prime } ( \Phi,{\ mke rn2mu \und erline{\m ker n-2mu\ sma s h{\Phi }\mke rn-2m u} \mk e r n2mu} , K, {\m ke rn2mu\unde r l ine {\mke rn -2mu \smash{ K}\mkern-2mu}\mker n 2mu })=\bar{S}(\ Phi ,{\m k e rn 2mu \ un d erl in e {\m k e rn-2mu\smash{\P hi }\mkern -2

{S}_{\text{gf}}$ and_assume that_the background field theory_is gauge-fixed_in_the way_just_explained. Background-preserving canonical transformations ----------------------------------------------- It_is useful to_characterize the most general_canonical transformations $\Phi,{\mkern2mu\underline{\mkern-2mu\smash{\Phi_}\mkern-2mu}\mkern2mu_},K,{\mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu }\rightarrow \Phi ^{\prime },{\mkern2mu\underline{\mkern-2mu\smash{\Phi }\mkern-2mu}\mkern2mu }^{\prime },K^{\prime },{\mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu }^{\prime }$ that preserve the_background_field master_equations_(\[treide\])_and the basic properties of_$\hat{S}$ and $\bar{S}$. By definition, all_canonical transformations_preserve the antiparentheses, so (\[treide\]) are turned into_$$\llbracket_\hat{S}^{\prime },\hat{S}^{\prime }\rrbracket_^{\prime }=\llbracket \hat{S}^{\prime },\bar{S}^{\prime }\rrbracket ^{\prime }=\llbracket \bar{S}^{\prime },\bar{S}^{\prime_}\rrbracket ^{\prime }=0. \label{tretre}$$ Moreover, $\hat{S}^{\prime }$_should be ${\mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu_}^{\prime_}$_and ${\mkern2mu\underline{\mkern-2mu\smash{C}\mkern-2mu}\mkern2mu }^{\prime }$_independent, while $\bar{S}$ should be invariant,_because it encodes the background transformations._This means $$\bar{S}^{\prime }(\Phi,{\mkern2mu\underline{\mkern-2mu\smash{\Phi }\mkern-2mu}\mkern2mu },K,{\mkern2mu\underline{\mkern-2mu\smash{K}\mkern-2mu}\mkern2mu })=\bar{S}(\Phi,{\mkern2mu\underline{\mkern-2mu\smash{\Phi_}\mkern-2

_W$ and hence $\tau_V = \tau - \tau_W$. Thus $\tau_V$ is the superconformal vector constructed in §\[exist.sp3.2\]. The full automorphism group $\operatorname{Aut}_{N{=}1}(W)$ must normalize $A_5 \subset {\mathrm{Co}}_1$, and so is exactly the identified ${\mathrm J}_2{:}2$. Uniqueness for ${\mathrm{Sp}}(2{\times}3)_2$ {#unique.sp3} -------------------------------------------- Let $V = {\mathrm{PSp}}(2{\times}3)_2$ and $W = {\mathrm{SO}}(5)_3 = {\mathrm{PSp}}(2{\times}2)_3$, with even subalgebras $V_{\mathrm{ev}}= {\mathrm{Sp}}(2{\times}3)_2$ and $W_{\mathrm{ev}}= {\mathrm{Sp}}(2{\times}2)_3$. Again we may construct a conformal embedding $V \otimes W \subset V^{f\natural}$, extending the level-rank duality ${\mathrm{Sp}}(2{\times}3)_2 \otimes {\mathrm{Sp}}(2{\times}2)_3 \subset {\mathrm{Spin}}(24)_1$. Let $\tau_V$ denote an arbitrary superconformal structure on $V$, and $\tau_W$ the superconformal structure on $W$ constructed in §\[exist.spinm\] with automorphism group $S_{6}$. Look at the subgroup $A_6 \subset S_6$. Again by studying how much of $\operatorname{Aut}_{N{=}0}(V) \times \operatorname{Aut}_{N{=}0}(W)$ lifts to $V^{f\natural}$, we find that $\operatorname{Aut}(\tau_V)$ is precisely the normalizer of $A_6 \subset {\mathrm{Co}}_1$, modulo $A_6$ of course. The conjugacy class of the embedding $A_6 \subset {\mathrm{Co}}_1$ is uniquely determined by requesting that the corresponding embedding into ${\mathrm{PSO}}(24)$ factors through ${\mathrm{PSp}}(2{\times}2)$. The normalizer-mod-normal of this $A_6$ is precisely ${\mathrm U}_3(3){:}2$, which, by §\[exist.sp3\], uniquely determines the superconformal

_ W$ and hence $ \tau_V = \tau - \tau_W$. Thus $ \tau_V$ is the superconformal vector constructed in § \[exist.sp3.2\ ]. The full automorphism group $ \operatorname{Aut}_{N{=}1}(W)$ must anneal $ A_5 \subset { \mathrm{Co}}_1 $, and therefore is exactly the identified $ { \mathrm J}_2{:}2$. Uniqueness for $ { \mathrm{Sp}}(2{\times}3)_2 $ { # unique.sp3 } -------------------------------------------- lease $ vanadium = { \mathrm{PSp}}(2{\times}3)_2 $ and $ W = { \mathrm{SO}}(5)_3 = { \mathrm{PSp}}(2{\times}2)_3 $, with even subalgebras $ V_{\mathrm{ev}}= { \mathrm{Sp}}(2{\times}3)_2 $ and $ W_{\mathrm{ev}}= { \mathrm{Sp}}(2{\times}2)_3$. Again we may construct a conformal embed $ vanadium \otimes W \subset V^{f\natural}$, extending the level - rank duality $ { \mathrm{Sp}}(2{\times}3)_2 \otimes { \mathrm{Sp}}(2{\times}2)_3 \subset { \mathrm{Spin}}(24)_1$. lease $ \tau_V$ denote an arbitrary superconformal structure on $ V$, and $ \tau_W$ the superconformal social organization on $ W$ constructed in § \[exist.spinm\ ] with automorphism group $ S_{6}$. Look at the subgroup $ A_6 \subset S_6$. Again by study how much of $ \operatorname{Aut}_{N{=}0}(V) \times \operatorname{Aut}_{N{=}0}(W)$ lifts to $ V^{f\natural}$, we find that $ \operatorname{Aut}(\tau_V)$ is precisely the normalizer of $ A_6 \subset { \mathrm{Co}}_1 $, modulo $ A_6 $ of path. The conjugacy class of the embedding $ A_6 \subset { \mathrm{Co}}_1 $ is uniquely determined by request that the corresponding embedding into $ { \mathrm{PSO}}(24)$ factors through $ { \mathrm{PSp}}(2{\times}2)$. The normalizer - mod - normal of this $ A_6 $ is precisely $ { \mathrm U}_3(3){:}2 $, which, by   § \[exist.sp3\ ], uniquely determines the superconformal

_W$ ajd hence $\tau_V = \tau - \tau_W$. Thus $\tau_V$ is tkw supecconforjal vectur constructed in §\[exist.sp3.2\]. Thx fuol auujmorphism group $\opdratornamv{Aut}_{N{=}1}(W)$ muwt niemalize $A_5 \subset {\mathrm{Ck}}_1$, and wo is exactly jhe identifiad ${\mathrm J}_2{:}2$. Unixudnzss for ${\mathrm{Sp}}(2{\times}3)_2$ {#unique.sp3} -------------------------------------------- Let $V = {\mathrm{LSo}}(2{\times}3)_2$ and $W = {\iathgm{FO}}(5)_3 = {\jathrm{PSp}}(2{\times}2)_3$, with even subalgebrzs $V_{\matirm{ev}}= {\mathrm{Sp}}(2{\tomes}3)_2$ and $W_{\mathrm{ev}}= {\mathrm{So}}(2{\timfs}2)_3$. Again we may cojstruct a cinfownal embeddine $V \otimes W \subset V^{f\hatural}$, extending the level-rank dualnty ${\mathrm{Sk}}(2{\cumed}3)_2 \otimes {\matirm{Sp}}(2{\tpmes}2)_3 \subset {\mathrm{Spit}}(24)_1$. Let $\tsu_V$ denote an srbmtraey superconformal strncture on $V$, and $\tau_W$ the supescknformal structurw in $W$ wonsdrucgwd kn §\[txixt.apinm\] aiti automorphjsm group $S_{6}$. Look at the subgroip $Q_6 \subset S_6$. Agzin by seudying how much of $\operatorname{Aut}_{N{=}0}(V) \tpmes \operatorname{Aut}_{N{=}0}(W)$ liftw to $V^{f\natural}$, we finf that $\opqratorname{Aut}(\tau_V)$ is precisely the normalizer of $D_6 \subadt {\nabhrm{Zi}}_1$, lodulo $A_6$ of course. The conjugacy class of the smnecding $A_6 \subset {\mathrm{Co}}_1$ ix knozuely determived by resuesting that the forrespjndint embeddigg imto ${\mathrm{PSO}}(24)$ factors throuth ${\mathrm{PSp}}(2{\nimew}2)$. The normalizer-mob-normal of tkis $A_6$ os prrcisely ${\mathrm U}_3(3){:}2$, which, yy §\[exisf.sp3\], uniquelj determihds the superconfurmsl

_W$ and hence $\tau_V = \tau - $\tau_V$ the superconformal constructed in §\[exist.sp3.2\]. must $A_5 \subset {\mathrm{Co}}_1$, so is exactly identified ${\mathrm J}_2{:}2$. Uniqueness for ${\mathrm{Sp}}(2{\times}3)_2$ -------------------------------------------- Let $V = {\mathrm{PSp}}(2{\times}3)_2$ and $W = {\mathrm{SO}}(5)_3 = {\mathrm{PSp}}(2{\times}2)_3$, with even $V_{\mathrm{ev}}= {\mathrm{Sp}}(2{\times}3)_2$ and $W_{\mathrm{ev}}= {\mathrm{Sp}}(2{\times}2)_3$. Again we may construct a conformal embedding $V W V^{f\natural}$, the duality ${\mathrm{Sp}}(2{\times}3)_2 \otimes {\mathrm{Sp}}(2{\times}2)_3 \subset {\mathrm{Spin}}(24)_1$. Let $\tau_V$ denote an arbitrary superconformal structure on $V$, and the superconformal structure on $W$ constructed in §\[exist.spinm\] automorphism group $S_{6}$. Look the subgroup $A_6 \subset S_6$. by how much $\operatorname{Aut}_{N{=}0}(V) \operatorname{Aut}_{N{=}0}(W)$ to $V^{f\natural}$, we that $\operatorname{Aut}(\tau_V)$ is precisely the normalizer of $A_6 \subset {\mathrm{Co}}_1$, modulo $A_6$ of course. The conjugacy class the embedding {\mathrm{Co}}_1$ is determined requesting the corresponding embedding factors through ${\mathrm{PSp}}(2{\times}2)$. The normalizer-mod-normal of precisely ${\mathrm U}_3(3){:}2$, which, by §\[exist.sp3\], uniquely determines superconformal

_W$ and hence $\tau_V = \tau - \tau_W$. Thus $\tAu_V$ is the suPercoNfoRmaL vEctoR conStructed in §\[exisT.Sp3.2\]. ThE full automorphism group $\OperaToRName{aUt}_{n{=}1}(W)$ musT normalIZe $a_5 \SUbsEt {\MaThrM{CO}}_1$, AnD so is ExaCtly the Identified ${\MatHrM J}_2{:}2$. Uniqueness FOr ${\Mathrm{Sp}}(2{\tiMes}3)_2$ {#Unique.sp3} -------------------------------------------- Let $V = {\MatHrm{PSp}}(2{\TiMes}3)_2$ ANd $W = {\maThrM{SO}}(5)_3 = {\maThrm{PSP}}(2{\Times}2)_3$, wIth even suBaLGebras $v_{\Mathrm{eV}}= {\MAtHrm{SP}}(2{\times}3)_2$ and $W_{\mathrm{eV}}= {\MaTHrm{Sp}}(2{\times}2)_3$. AgaiN we may CoNStRUCt a ConFormal embeDdIng $V \oTImes W \suBSeT v^{F\NatURal}$, extending tHe level-rank DUalIty ${\matHrM{Sp}}(2{\TImes}3)_2 \otImes {\mAtHRm{SP}}(2{\times}2)_3 \subseT {\matHrm{Spin}}(24)_1$. LeT $\tau_V$ dENote an aRBitrary SupercOnfOrmAl stRUcTuRe oN $V$, ANd $\tAU_W$ The SUpeRconformAl StRuctuRe on $w$ CONStruCteD in §\[eXist.sPinm\] with automOrpHism GRouP $S_{6}$. LooK at thE subGrOup $A_6 \sUbset S_6$. again By Studying how much Of $\opEratornamE{AuT}_{N{=}0}(v) \tiMeS \operATornamE{AuT}_{N{=}0}(W)$ Lifts to $v^{f\naturAL}$, we FiND THaT $\operatorname{Aut}(\taU_V)$ IS PrEcisely tHe normALiZeR Of $A_6 \subseT {\mAthRm{Co}}_1$, MODulo $A_6$ Of coURsE. The conjUgacy cLAsS oF the embEdDing $A_6 \sUbSet {\MatHrm{Co}}_1$ IS uniQuely dEtermineD by reQUesting that the COrresponding eMBeDDInG Into ${\MatHrm{PSO}}(24)$ factoRs thROugh ${\MathRM{Psp}}(2{\tIMes}2)$. ThE normAlIZeR-Mod-normal of this $A_6$ is pReCisely ${\MathrM U}_3(3){:}2$, which, by §\[exisT.sp3\], uniquelY DETermines The sUPeRConformal

_W$ and hence $\tau_V = \t au - \tau_ W$. T hus $\ ta u_V$ isthe superconfo r malvector constructed in§\[ex is t .sp3 . 2\ ]. Th e fulla ut o m orp hi sm gr ou p $ \oper ato rname{A ut}_{N{=}1 }(W )$ must normal i ze $A_5 \sub set {\mathrm{Co }}_ 1$, an dsoi s exa ctl y the ident i fied $ {\mathrmJ} _ 2{:}2$ . Uniqu e n es s fo r ${\mathrm{Sp}}( 2 {\ t imes}3)_2$ {#u nique. sp 3 }- - --- --- ---------- -- ----- - ------- - -- - - - --- - - Let $V = { \mathrm{PSp } }(2 {\time s} 3)_ 2 $ and$W ={\ m ath rm{SO}}(5)_ 3 ={\mathrm{ PSp}}( 2 {\times } 2)_3$,with e ven su balg e br as $V _{ \ mat h rm {ev } }={\mathrm {S p} }(2{\ time s } 3 ) _2$and $W_ {\mat hrm{ev}}= {\m ath rm{S p }}( 2{\ti mes}2 )_3$ .Again we ma y con st ruct a conforma l em bedding $ V \ ot ime sW \su b set V^ {f\ nat ural}$, extend i ngth e l ev el-rank duality ${ \m a t hr m{Sp}}(2 {\time s }3 )_ 2 \otimes { \ma thrm { S p}}(2 {\ti m es }2)_3 \s ubset{ \m at hrm{Spi n} }(24)_ 1$ . L et$\tau _ V$ d enotean arbit rarys uperconformals tructure on $ V $, a nd $\ta u_W $ the super conf o rmal str u ct ure on $W $ con st r uc t ed in §\[exist.spin m\ ] with auto morphism grou p $S_{6}$. L o ok at th e su b gr o up $A_6 \subse t S_6 $. Again b y studyin g how much of $\operat o r name{Aut }_{ N{= }0} (V) \ ti mes \operator n a me{A ut }_{N{=} 0}( W)$ lif tsto$V^ {f\ na tural}$,we findth at $ \o per atorn a me{Aut}( \t au_ V) $ i s pre c iselythe n orma li ze r of $A_6 \ s ub s e t {\ ma th rm{C o}} _1 $, mo dulo $A_ 6$ of c ourse. Th e c o njug ac yclass o f the embeddi ng $A_6 \sub se t { \mathr m { Co}}_1$is uniquely determinedb y reque sti ng th at t he corres pon ding e mbe d ding i nto ${ \math rm {PS O } }(24) $ fa cto rs through $ { \ mat hrm{P Sp }}(2 {\times }2)$. The normaliz e r-m od-normal ofthi s $A _ 6 $isp re c ise ly ${\ m a thrm U}_3(3){:} 2$, which, b y  § \[exist.sp 3 \], u niquely determ inest he supe rconforma l

_W$ and_hence $\tau_V_= \tau - \tau_W$._Thus $\tau_V$_is_the superconformal_vector_constructed in §\[exist.sp3.2\]._The full automorphism_group $\operatorname{Aut}_{N{=}1}(W)$ must normalize_$A_5 \subset {\mathrm{Co}}_1$,_and_so is exactly the identified ${\mathrm J}_2{:}2$. Uniqueness for ${\mathrm{Sp}}(2{\times}3)_2$ {#unique.sp3} -------------------------------------------- Let $V = {\mathrm{PSp}}(2{\times}3)_2$ and_$W_= {\mathrm{SO}}(5)_3_=_{\mathrm{PSp}}(2{\times}2)_3$,_with even subalgebras $V_{\mathrm{ev}}= {\mathrm{Sp}}(2{\times}3)_2$_and $W_{\mathrm{ev}}= {\mathrm{Sp}}(2{\times}2)_3$. Again we_may construct_a conformal embedding $V \otimes W \subset V^{f\natural}$,_extending_the level-rank duality_${\mathrm{Sp}}(2{\times}3)_2 \otimes {\mathrm{Sp}}(2{\times}2)_3 \subset {\mathrm{Spin}}(24)_1$. Let $\tau_V$ denote an_arbitrary superconformal structure on $V$, and_$\tau_W$ the superconformal_structure_on_$W$ constructed in §\[exist.spinm\]_with automorphism group $S_{6}$. Look at_the subgroup $A_6 \subset S_6$. Again_by studying how much of $\operatorname{Aut}_{N{=}0}(V) \times_\operatorname{Aut}_{N{=}0}(W)$ lifts to $V^{f\natural}$, we find_that $\operatorname{Aut}(\tau_V)$ is precisely the_normalizer of_$A_6 \subset {\mathrm{Co}}_1$, modulo $A_6$_of course. The_conjugacy class_of the embedding_$A_6 \subset {\mathrm{Co}}_1$ is uniquely determined_by requesting that_the corresponding embedding into ${\mathrm{PSO}}(24)$ factors_through_${\mathrm{PSp}}(2{\times}2)$. The normalizer-mod-normal_of_this_$A_6$ is_precisely ${\mathrm U}_3(3){:}2$,_which,_by §\[exist.sp3\], uniquely_determines_the superconformal

. [**1**]{}, 317 (1993), [hep-th/9211021]{}. J. Khoury, B. A. Ovrut, P. J. Steinhardt and N. Turok, Phys. Rev. [**D64**]{}, 123522 (2001), [hep-th/0103239]{}. D. H. Lyth, [hep-ph/0106153]{}. R. Brandenberger and F. Finelli, [hep-th/0109004]{}. J. Khoury, B. A. Ovrut, P. J. Steinhardt and N. Turok, [ hep-th/0109050]{}. J. Hwang, [astro-ph/0109045]{}. J. Martin, P. Peter, N. Pinto-Neto and D. J. Schwarz, [hep-th/0112128]{}. J. Audretsch and G. Schäfer, Phys. Lett. [**66A**]{}, 459 (1978). N. Birrell and P. Davies, ”[*Quantum fields in curved space*]{}”, Cambridge Univ. Press, Cambridge, 1982. T. Tanaka, [astro-ph/0012431]{}. L. Parker, Phys. Rev. [**183**]{}, 1057 (1969). V. Mukhanov and R. Brandenberger, Phys. Rev. Lett. [**68**]{}, 1969 (1992); R. Brandenberger, V. Mukhanov and A. Sornborger, Phys. Rev. [**D48**]{}, 1629 (1993). M. Brown and C. Dutton, Phys. Rev. [**D18**]{}, 4422 (1978). L. Mersini, M. Bastero-Gil and P. Kanti, Phys. Rev. [**D64**]{}, 043508 (2001), [hep-ph/0101210]{}. M. Lemoine, M. Lubo, J. Martin and J. P. Uzan, Phys. Rev. [**D

.   [ * * 1 * * ] { }, 317 (1993), [ hep - th/9211021 ] { }. J.   Khoury, B.   A.   Ovrut, P.   J.   Steinhardt and N.   Turok, Phys.   Rev.   [ * * D64 * * ] { }, 123522 (2001), [ hep - th/0103239 ] { }. D.   H.   Lyth, [ hep - ph/0106153 ] { }. R.   Brandenberger and F.   Finelli, [ hep - th/0109004 ] { }. J.   Khoury, B.   A.   Ovrut, P.   J.   Steinhardt and N.   Turok, [ hep - th/0109050 ] { }. J.   Hwang, [ astro - ph/0109045 ] { }. J.   Martin, P.   Peter, N.   Pinto - Neto and D.   J.   Schwarz, [ hep - th/0112128 ] { }. J.   Audretsch and G.   Schäfer, Phys.   Lett.   [ * * 66A * * ] { }, 459 (1978). N.   Birrell and P.   Davies, ” [ * Quantum fields in curved space * ] { } ”, Cambridge Univ.   Press, Cambridge, 1982. T.   Tanaka, [ astro - ph/0012431 ] { }. L.   Parker, Phys.   Rev.   [ * * 183 * * ] { }, 1057 (1969). V.   Mukhanov and R.   Brandenberger, Phys.   Rev.   Lett.   [ * * 68 * * ] { }, 1969 (1992); R.   Brandenberger, V.   Mukhanov and A.   Sornborger, Phys.   Rev.   [ * * D48 * * ] { }, 1629 (1993). M.   Brown and C.   Dutton, Phys.   Rev.   [ * * D18 * * ] { }, 4422 (1978). L.   Mersini, M.   Bastero - Gil and P.   Kanti, Phys.   Rev.   [ * * D64 * * ] { }, 043508 (2001), [ hep - ph/0101210 ] { }. M.   Lemoine, M.   Lubo, J.   Martin and J.   P.   Uzan, Phys.   Rev.   [ * * vitamin d

. [**1**]{}, 317 (1993), [jep-th/9211021]{}. J. Khoury, B. A. Ovrut, P. J. Sueinhardt and N. Tutoj, Phys. Cev. [**D64**]{}, 123522 (2001), [gep-th/0103239]{}. D. H. Lhth, [hep-ph/0106153]{}. R. Brandenberger and H. Finwlli, [yep-th/0109004]{}. J. Khoury, B. A. Ovrut, P. G. Steinharft and N. Ruroj, [ hep-th/0109050]{}. J. Hwehg, [astro-ph/0109045]{}. J. Madbin, P. 'ever, N. Pinto-Neto snd D. J. Schwdrz, [hep-th/0112128]{}. J. Audratrck and G. Schäfer, Phys. Lett. [**66A**]{}, 459 (1978). N. Birrell anq P. Davirs, ”[*Quantum fieldf in survsd space*]{}”, Cambridge Univ. Press, Cambrjdge, 1982. T. Tenaka, [astro-ph/0012431]{}. L. Psrker, Phys. Rev. [**183**]{}, 1057 (1969). V. Mukhanov ajd R. Hrandenberger, Phys. Gev. Lett. [**68**]{}, 1969 (1992); R. Vranqwnberger, V. Muyhanov and A. Sornborger, Phys. Rev. [**D48**]{}, 1629 (1993). M. Brown and C. Dutton, Pfys. Rer. [**D18**]{}, 4422 (1978). L. Mersinu, N. Badjero-Gil and '. Kanti, Phys. Rev. [**D64**]{}, 043508 (2001), [mvp-ph/0101210]{}. M. Lekoine, M. Kubo, J. Martin akd J. P. Nzan, Phys. Rev. [**D

. [**1**]{}, 317 (1993), [hep-th/9211021]{}. J. Khoury, Ovrut, J. Steinhardt N. Turok, Phys. D. Lyth, [hep-ph/0106153]{}. R. and F. Finelli, J. Khoury, B. A. Ovrut, P. Steinhardt and N. Turok, [ hep-th/0109050]{}. J. Hwang, [astro-ph/0109045]{}. J. Martin, P. Peter, Pinto-Neto and D. J. Schwarz, [hep-th/0112128]{}. J. Audretsch and G. Schäfer, Phys. Lett. 459 N. and Davies, ”[*Quantum fields in curved space*]{}”, Cambridge Univ. Press, Cambridge, 1982. T. Tanaka, [astro-ph/0012431]{}. L. Parker, Rev. [**183**]{}, 1057 (1969). V. Mukhanov and R. Phys. Rev. Lett. [**68**]{}, (1992); R. Brandenberger, V. Mukhanov A. Phys. Rev. 1629 M. and C. Dutton, Rev. [**D18**]{}, 4422 (1978). L. Mersini, M. Bastero-Gil and P. Kanti, Phys. Rev. [**D64**]{}, 043508 (2001), [hep-ph/0101210]{}. Lemoine, M. Martin and P. Phys. [**D

. [**1**]{}, 317 (1993), [hep-th/9211021]{}. J. Khoury, B. A. Ovrut, P. J. SteiNhardt and N. turok, phyS. ReV. [**D64**]{}, 123522 (2001), [Hep-tH/0103239]{}. D. H. LYth, [hep-ph/0106153]{}. R. BrandENberGer and F. Finelli, [hep-th/0109004]{}. J. KhOury, B. a. OVRut, P. j. stEinhaRdt and N. tUrOK, [ Hep-Th/0109050]{}. j. HWanG, [aSTrO-ph/0109045]{}. J. MArtIn, P. PeteR, N. Pinto-NetO anD D. j. Schwarz, [hep-tH/0112128]{}. j. AUdretsch anD G. SChäfer, Phys. LeTt. [**66A**]{}, 459 (1978). n. BirreLl And p. davieS, ”[*QuAntum Fields IN curveD space*]{}”, CamBrIDge UniV. press, CaMBRiDge, 1982. T. tanaka, [astro-ph/0012431]{}. L. ParKEr, pHys. Rev. [**183**]{}, 1057 (1969). V. MukhanoV and R. BRaNDeNBErgEr, PHys. Rev. Lett. [**68**]{}, 1969 (1992); r. BRandeNBerger, V. mUkHANOv aND A. Sornborger, PHys. Rev. [**D48**]{}, 1629 (1993). M. BroWN anD C. DuttOn, phyS. rev. [**D18**]{}, 4422 (1978). L. MErsinI, M. bAstEro-Gil and P. KAnti, phys. Rev. [**D64**]{}, 043508 (2001), [hEp-ph/0101210]{}. M. LEMoine, M. LUBo, J. MartIn and J. p. UzAn, PHys. REV. [**D

. [**1**]{}, 317 (1993), [ hep-th/921 1021] {}. J .Khou ry,B. A. Ovrut, P .  J.Steinhardt and N. Turo k, Ph ys .  Rev .  [ **D64 **]{},1 23 5 2 2 ( 20 01 ),[h e p- th/01 032 39]{}. D. H. Lyt h,[h ep-ph/010615 3 ]{ }. R. Bra nde nberger andF.Finell i, [h e p-th/ 010 9004] {}. J .  Khour y, B. A.Ov r ut, P. J. Stei n h ar dt a nd N. Turok, [ he p -t h /0109050]{}. J. Hwa ng , [ a s tro -ph /0109045]{ }. J.M artin,P .P e t er, N. Pinto-Neto and D. J.S chw arz, [ he p-t h /01121 28]{} .J.Audretsch a nd G . Schäfer , Phys .  Lett.[ **66A** ]{}, 4 59(19 78). N .Bir re l l a n dP.D avi es, ”[*Q ua nt um fi elds i n curv edspac e*]{} ”, CambridgeUni v. P r ess , Cam bridg e, 1 98 2. T . Tana ka, [ as tro-ph/0012431] {}. L. Parke r,Ph ys.  R ev. [ * *183** ]{} , 1 057 (19 69). V .  Mu kh a n o vand R. Brandenberg er , Ph ys. Rev.  Lett. [* *6 8 **]{}, 1 96 9 ( 1992 ) ; R. B rand e nb erger, V . Mukh a no vand A.So rnborg er , P hys . Rev .  [** D48**] {}, 1629 (199 3 ). M. Brown a n d C. Dutton,P hy s .  R e v. [ **D 18**]{}, 44 22 ( 1 978) . L .  M ers i ni, M . Bas te r o- G il and P. Kanti, Ph ys . Rev.  [**D 64**]{}, 0435 08 (2001), [ h ep-ph/01 0121 0 ]{ } . M. Lemoine, M. L ubo, J. Ma r tin andJ. P.  Uzan, P hys. Rev. [ **D

. [**1**]{}, 317_(1993), [hep-th/9211021]{}. J. Khoury,_B. A. Ovrut, P. J. Steinhardt and N. Turok,_Phys. Rev. [**D64**]{}, 123522_(2001),_[hep-th/0103239]{}. D. H. Lyth, [hep-ph/0106153]{}. R. Brandenberger_and_F. Finelli, [hep-th/0109004]{}. J. Khoury, B. A. Ovrut,_P. J. Steinhardt and N. Turok,_[ hep-th/0109050]{}. J. Hwang, [astro-ph/0109045]{}. J. Martin, P. Peter,_N. Pinto-Neto and D. J. Schwarz,_[hep-th/0112128]{}. J. Audretsch_and G. Schäfer, Phys. Lett. [**66A**]{}, 459 (1978). N. Birrell and P. Davies, ”[*Quantum fields in curved space*]{}”, Cambridge Univ. Press,_Cambridge,_1982. T. Tanaka, [astro-ph/0012431]{}. L. Parker,_Phys. Rev. [**183**]{},_1057_(1969). V. Mukhanov and R. Brandenberger, Phys. Rev. Lett. [**68**]{}, 1969_(1992); R. Brandenberger, V. Mukhanov and A. Sornborger,_Phys. Rev. [**D48**]{}, 1629_(1993). M. Brown and C. Dutton, Phys. Rev. [**D18**]{}, 4422 (1978). L. Mersini, M. Bastero-Gil and_P. Kanti,_Phys. Rev. [**D64**]{}, 043508 (2001),_[hep-ph/0101210]{}. M. Lemoine, M. Lubo, J. Martin and J. P. Uzan, Phys. Rev. [**D

squirmers (interacting via long-range (panel-a) or short-range (panel-b) attraction) and of WP squirmers (interacting via long-range (panel-c) or short-range (panel-d) attraction). ![\[fig:nct\_xi10\] Time evolutions of the mean cluster size for squirmers when $\xi=10$ (pullers: red [squares]{} and black [circles]{}; neutral: blue [up triangles]{}; pushers: pink [down-triangles]{} and grey [diamonds]{}). (a) AP squirmers with $r_c=2.5\sigma$ and (b) $r_c=1.5\sigma$, (c) WP squirmers with $r_c=2.5\sigma$ and (d) $r_c=1.5\sigma$. Note the different scale on the y-axis.](ntxi10rc25ttbis.pdf "fig:"){width="0.48\columnwidth"}![\[fig:nct\_xi10\] Time evolutions of the mean cluster size for squirmers when $\xi=10$ (pullers: red [squares]{} and black [circles]{}; neutral: blue [up triangles]{}; pushers: pink [down-triangles]{} and grey [diamonds]{}). (a) AP squirmers with $r_c=2.5\sigma$ and (b) $r_c=1.5\sigma$, (c) WP squirmers with $r_c=2.5\sigma$ and (d) $r_c=1.5\sigma$. Note the different scale on the y-axis.](ntxi10rc15ttbis.pdf "fig:"){width="0.48\columnwidth"}![\[fig:nct\_xi10\] Time evolutions of the mean cluster size for squirmers when $\xi=10$ (pullers: red [squares]{} and black [circles]{}; neutral: blue [up triangles]{}; pushers: pink [down-triangles]{} and grey [diamonds]{}). (a) AP squirmers with $r_c=2.5\sigma$ and (b) $r_c=1.5\sigma$, (c) WP squirmers with $r_c=2.5\sigma$ and (d) $r_c

squirmers (interacting via long - range (dialog box - deoxyadenosine monophosphate) or short - range (panel - b-complex vitamin) attraction) and of WP squirmers (interact via long - range (panel - coke) or short - range (control panel - d) attraction). ! [ \[fig: nct\_xi10\ ] Time evolutions of the bastardly bunch size for squirmers when $ \xi=10 $ (pullers: crimson [ squares ] { } and black [ circle ] { }; neutral: blue [ up triangles ] { }; pusher: pink [ down - triangles ] { } and grey [ diamonds ] { }). (a) AP squirmers with $ r_c=2.5\sigma$ and (b) $ r_c=1.5\sigma$, (c) WP squirmers with $ r_c=2.5\sigma$ and (d) $ r_c=1.5\sigma$. notice the different plate on the y - axis.](ntxi10rc25ttbis.pdf " fig:"){width="0.48\columnwidth"}![\[fig: nct\_xi10\ ] Time evolutions of the mean cluster size for wiggler when $ \xi=10 $ (pullers: red [ squares ] { } and black [ circles ] { }; neutral: blue [ up triangles ] { }; pushers: pink [ down - triangles ] { } and grey [ diamonds ] { }). (a) AP squirmers with $ r_c=2.5\sigma$ and (b) $ r_c=1.5\sigma$, (c) WP squirmers with $ r_c=2.5\sigma$ and (d) $ r_c=1.5\sigma$. Note the different scale on the y - axis.](ntxi10rc15ttbis.pdf " fig:"){width="0.48\columnwidth"}![\[fig: nct\_xi10\ ] Time development of the mean cluster size for wiggler when $ \xi=10 $ (puller: crimson [ squares ] { } and black [ circle ] { }; neutral: blue [ up triangles ] { }; pushers: pink [ down - triangle ] { } and grey [ diamonds ] { }). (a) AP squirmers with $ r_c=2.5\sigma$ and (b) $ r_c=1.5\sigma$, (c) WP squirmers with $ r_c=2.5\sigma$ and (d) $ r_c

sqkirmers (interacting via uong-range (panel-c) or shmrt-ranfe (panel-c) attraction) and of WP squirlees (inuvracting via long-rangd (panel-c) lr short-eangt (panel-d) attraction). ![\[fig:nct\_wn10\] Tims evonntions of the mgan cluster vize for squirkefs when $\xi=10$ (pullers: red [squares]{} and blwck [cirvlfs]{}; neutral: blug [up uriwnglss]{}; pushers: pink [down-triangles]{} and frey [diemonds]{}). (a) AP squormers with $r_c=2.5\sigma$ and (b) $g_c=1.5\sihma$, (c) WP squirmers with $r_c=2.5\signa$ age (d) $r_c=1.5\sigma$. Vote the dpyferent scame on the y-axis.](ntxi10rc25ttbis.pdf "fkg:"){widch="0.48\columnwidjk"}![\[dig:jwt\_xi10\] Time etolutijns of the mcsn cluvter siae for squirmevs whxn $\xu=10$ (pullers: red [squares]{} and black [circles]{}; neutral: tlbe [up triangles]{}; pushees: pink [dowt-trixbglds]{} znv gdey [diwmoids]{}). (a) AP sqhirmers wity $r_c=2.5\sigma$ and (b) $r_c=1.5\sogiq$, (c) WP squirmsrs wieh $r_c=2.5\sigma$ and (d) $r_c=1.5\sigma$. Note the differett acale on the y-axis.](ntxi10rx15ttbis.pdf "fig:"){width="0.48\colomnwidth"}![\[fid:nct\_xi10\] Time evolutions of the mean cluster size fmr sqnifmexw whev $\xl=10$ (pullers: red [squares]{} and black [circles]{}; neutrwm: nlle [up triangles]{}; pmshers: pink [down-troajgkgs]{} and grey [dixmonds]{}). (a) ZP squirmers with $g_c=2.5\sigma$ and (v) $r_c=1.5\sigma$, (c) WL squirmers with $r_c=2.5\sigma$ ane (d) $r_c

squirmers (interacting via long-range (panel-a) or short-range and WP squirmers via long-range (panel-c) Time of the mean size for squirmers $\xi=10$ (pullers: red [squares]{} and black neutral: blue [up triangles]{}; pushers: pink [down-triangles]{} and grey [diamonds]{}). (a) AP squirmers $r_c=2.5\sigma$ and (b) $r_c=1.5\sigma$, (c) WP squirmers with $r_c=2.5\sigma$ and (d) $r_c=1.5\sigma$. Note different on y-axis.](ntxi10rc25ttbis.pdf Time evolutions of the mean cluster size for squirmers when $\xi=10$ (pullers: red [squares]{} and black neutral: blue [up triangles]{}; pushers: pink [down-triangles]{} and [diamonds]{}). (a) AP squirmers $r_c=2.5\sigma$ and (b) $r_c=1.5\sigma$, (c) squirmers $r_c=2.5\sigma$ and $r_c=1.5\sigma$. the scale on the "fig:"){width="0.48\columnwidth"}![\[fig:nct\_xi10\] Time evolutions of the mean cluster size for squirmers when $\xi=10$ (pullers: red [squares]{} and black neutral: blue pushers: pink and [diamonds]{}). AP squirmers with (b) $r_c=1.5\sigma$, (c) WP squirmers with $r_c

squirmers (interacting via loNg-range (panEl-a) or ShoRt-rAnGe (paNel-b) Attraction) and oF wP sqUirmers (interacting via lOng-raNgE (PaneL-C) oR shorT-range (pANeL-D) AttRaCtIon). ![\[FiG:NcT\_xi10\] TiMe eVolutioNs of the meaN clUsTer size for sqUIrMers when $\xi=10$ (PulLers: red [squarEs]{} aNd blacK [cIrcLEs]{}; neuTraL: blue [Up triaNGles]{}; puShers: pink [DoWN-trianGLes]{} and gREY [dIamoNds]{}). (a) AP squirmers wiTH $r_C=2.5\Sigma$ and (b) $r_c=1.5\sigMa$, (c) WP sQuIRmERS wiTh $r_C=2.5\sigma$ and (d) $R_c=1.5\Sigma$. nOte the dIFfERENt sCAle on the y-axis.](Ntxi10rc25ttbis.PDf "fIg:"){widtH="0.48\cOluMNwidth"}![\[Fig:ncT\_xI10\] timE evolutions Of thE mean clusTer sizE For squiRMers wheN $\xi=10$ (pulLerS: reD [squAReS]{} aNd bLaCK [ciRClEs]{}; nEUtrAl: blue [up TrIaNgles]{}; PushERS: PInk [dOwn-TriaNgles]{} And grey [diamonDs]{}). (a) aP sqUIrmErs wiTh $r_c=2.5\sIgma$ AnD (b) $r_c=1.5\sIgma$, (c) Wp squiRmErs with $r_c=2.5\sigma$ aNd (d) $r_C=1.5\sigma$. NotE thE dIffErEnt scALe on thE y-aXis.](Ntxi10rc15tTbis.pdf "FIg:"){wIdTH="0.48\COlUmnwidth"}![\[fig:nct\_xi10\] TiMe EVOlUtions of The meaN ClUsTEr size foR sQuiRmerS WHen $\xi=10$ (PullERs: Red [squarEs]{} and bLAcK [cIrcles]{}; nEuTral: blUe [Up tRiaNgles]{}; PUsheRs: pink [Down-triaNgles]{} ANd grey [diamonds]{}). (A) aP squirmers wiTH $r_C=2.5\SIgMA$ and (B) $r_c=1.5\Sigma$, (c) WP squIrmeRS witH $r_c=2.5\sIGmA$ anD (D) $r_c

squirmers (interacting vi a long-ran ge (p ane l-a )or s hort -range (panel- b ) at traction) and of WP sq uirme rs (int e ra cting via lo n g- r a nge ( pa nel -c ) o r sho rt- range ( panel-d) a ttr ac tion). ![\[ f ig :nct\_xi10 \]Time evoluti ons of th emea n clus ter size for s q uirmer s when $\ xi = 10$ (p u llers:r e d[squ ares]{} and black [c i rcles]{}; neut ral: b lu e [ u p tr ian gles]{}; p us hers: pink [d o wn - t r ian g les]{} and gr ey [diamond s ]{} ). (a) A P s q uirmer s wit h$ r_c =2.5\sigma$ and (b) $r_c =1.5\s i gma$, ( c ) WP sq uirmer s w ith $r_ c =2 .5 \si gm a $ a n d(d) $r_ c=1.5\si gm a$ . Not e th e d i ffer ent sca le on the y-axis.] (nt xi10 r c25 ttbis .pdf"fig :" ){wid th="0. 48\co lu mnwidth"}![\[fi g:nc t\_xi10\] Ti me ev ol ution s of th e m ean cluste r sizef orsq u i r me rs when $\xi=10$ ( pu l l er s: red [ square s ]{ }a nd black [ cir cles ] { }; ne utra l :blue [up trian g le s] {}; pus he rs: pi nk [d own -tria n gles ]{} an d grey [ diamo n ds]{}). (a) AP squirmers wit h $ r _ c= 2 .5\s igm a$ and (b)$r_c = 1.5\ sigm a $, (c ) WP s quirm er s w i th $r_c=2.5\sigma$an d (d)$r_c= 1.5\sigma$. N ote the di f f e rent sca le o n t h e y-axis.](ntx i10rc 15ttbis.pd f "fig:") {widt h="0.48\ columnwid t h "}![\[fi g:n ct\ _xi 10\ ] Ti me evolutions o f th emean cl ust er size fo r s qui rme rs when $\x i=10$ (p ul le rs :red [squ a res]{} a nd bl ac k [ circl e s]{};neutr al:bl ue [up triang l es ] { }; p us he rs:pin k[down -tri a ngl es]{} a nd grey [ dia m onds ]{ }) . (a) A P squirmers w it h $r_c=2.5 \s igm a$ and ( b) $r_c= 1.5\sigma$, (c) WP squi r mers wi th$r_c= 2.5\ sigma$ an d ( d) $r_ c

squirmers_(interacting via_long-range (panel-a) or short-range_(panel-b) attraction)_and_of WP_squirmers_(interacting via long-range_(panel-c) or short-range_(panel-d) attraction). ![\[fig:nct\_xi10\] Time evolutions_of the mean_cluster_size for squirmers when $\xi=10$ (pullers: red [squares]{} and black [circles]{}; neutral: blue [up_triangles]{};_pushers: pink_[down-triangles]{}_and_grey [diamonds]{}). (a) AP squirmers_with $r_c=2.5\sigma$ and (b) $r_c=1.5\sigma$,_(c) WP_squirmers with $r_c=2.5\sigma$ and (d) $r_c=1.5\sigma$. Note the_different_scale on the_y-axis.](ntxi10rc25ttbis.pdf "fig:"){width="0.48\columnwidth"}![\[fig:nct\_xi10\] Time evolutions of the mean cluster size_for squirmers when $\xi=10$ (pullers: red_[squares]{} and black_[circles]{};_neutral:_blue [up triangles]{}; pushers:_pink [down-triangles]{} and grey [diamonds]{}). (a)_AP squirmers with $r_c=2.5\sigma$ and (b)_$r_c=1.5\sigma$, (c) WP squirmers with $r_c=2.5\sigma$ and_(d) $r_c=1.5\sigma$. Note the different scale_on the y-axis.](ntxi10rc15ttbis.pdf "fig:"){width="0.48\columnwidth"}![\[fig:nct\_xi10\] Time_evolutions of_the mean cluster size for_squirmers when $\xi=10$_(pullers: red_[squares]{} and black_[circles]{}; neutral: blue [up triangles]{}; pushers:_pink [down-triangles]{} and_grey [diamonds]{}). (a) AP squirmers with_$r_c=2.5\sigma$_and (b) $r_c=1.5\sigma$,_(c)_WP_squirmers with_$r_c=2.5\sigma$ and (d)_$r_c

of [@close] to make the phase convention consistent with the use of chiral Lagrangians and we follow the convention of Refs. [@Miyahara:2016yyh; @Pavao:2017cpt]. Hence, with the symmetry in the last two particles, we have $$\begin{aligned} \phi_{MS}&=-\frac{1}{\sqrt{2}} \Big[ \frac{uds-dus}{\sqrt{2}} + \frac{usd-dsu}{\sqrt{2}} \Big], \\ \phi_{MA}&=-\frac{1}{\sqrt{6}} \Big[ \frac{usd-dsu}{\sqrt{2}} + \frac{dus-uds}{\sqrt{2}} -2 \frac{sdu-sud}{\sqrt{2}} \Big].\end{aligned}$$ - $\Lambda_c^+: c\,\frac{1}{\sqrt{2}} (ud-du) \chi_{MA}$. - $\Xi_c^+: c\,\frac{1}{\sqrt{2}} (us-su) \chi_{MA}$ and $\Xi_c^0: c\,\frac{1}{\sqrt{2}} (ds-sd) \chi_{MA}$. - $\Xi_c^{'+}: c\,\frac{1}{\sqrt{2}} (us+su) \chi_{MS}$ and $\Xi_c^{'0}: c\,\frac{1}{\sqrt{2}} (ds+sd) \chi_{MS}$. - $\Xi_c^{*+}: c\,\frac{1}{\sqrt{2}} (us+su) \chi_{S}$ and $\Xi_c^{*0}: c\,\frac{1}{\sqrt{2}} (ds+sd) \chi_{S}$. with $\chi_S$ the 3/2 spin symmetric wave function [@close]. The light-quark parts of the above wave-functions are consistent with $\ell_B=0$ for $\Lambda_c$ and $\Xi_c$, and $\ell_B=1$ for $\Xi_c^\prime$ and $\Xi_c^*$. On the other hand, our conventions for the isospin doublets are $(D^+,\ -D^0)$, $(\bar{D}^0,\ D^-)$, $(\Xi_c^+,\ \Xi_c^0)$. Thus

of [ @close ] to make the phase convention consistent with the use of chiral Lagrangians and we adopt the conventionality of Refs. [ @Miyahara:2016yyh; @Pavao:2017cpt ]. Hence, with the symmetry in the final two particle, we have $ $ \begin{aligned } \phi_{MS}&=-\frac{1}{\sqrt{2 } } \Big [ \frac{uds - dus}{\sqrt{2 } } + \frac{usd - dsu}{\sqrt{2 } } \Big ], \\ \phi_{MA}&=-\frac{1}{\sqrt{6 } } \Big [ \frac{usd - dsu}{\sqrt{2 } } + \frac{dus - uds}{\sqrt{2 } } -2 \frac{sdu - sud}{\sqrt{2 } } \Big].\end{aligned}$$ - $ \Lambda_c^+: c\,\frac{1}{\sqrt{2 } } (ud - du) \chi_{MA}$. - $ \Xi_c^+: c\,\frac{1}{\sqrt{2 } } (us - su) \chi_{MA}$ and $ \Xi_c^0: c\,\frac{1}{\sqrt{2 } } (ds - sd) \chi_{MA}$. - $ \Xi_c^{'+ }: c\,\frac{1}{\sqrt{2 } } (us+su) \chi_{MS}$ and $ \Xi_c^{'0 }: c\,\frac{1}{\sqrt{2 } } (ds+sd) \chi_{MS}$. - $ \Xi_c^{*+ }: c\,\frac{1}{\sqrt{2 } } (us+su) \chi_{S}$ and $ \Xi_c^{*0 }: c\,\frac{1}{\sqrt{2 } } (ds+sd) \chi_{S}$. with $ \chi_S$ the 3/2 spin symmetrical wave function [ @close ]. The light - quark part of the above wave - functions are consistent with $ \ell_B=0 $ for $ \Lambda_c$ and $ \Xi_c$, and $ \ell_B=1 $ for $ \Xi_c^\prime$ and $ \Xi_c^*$. On the early hand, our conventions for the isospin doublets are $ (D^+,\ -D^0)$, $ (\bar{D}^0,\ D^-)$, $ (\Xi_c^+,\ \Xi_c^0)$. Thus

of [@close] to make the phase convention consistent with fhe use uf chiral Lagrangians and we fillow the convention of Refr. [@Miyaharw:2016yyh; @Pavqo:2017cpu]. Hence, with the symmetry in ths lasc vwo particles, wg have $$\begin{dligned} \phi_{KS}&=-\wrcc{1}{\sqrt{2}} \Big[ \frac{uds-dus}{\sqrt{2}} + \frac{usd-dst}{\sqrt{2}} \Bog], \\ \phi_{MA}&=-\frac{1}{\fqrt{6}} \Fig[ \rgag{usd-dsu}{\sqrt{2}} + \frac{dus-uds}{\sqrt{2}} -2 \frad{sdu-sud}{\vqrt{2}} \Big].\end{alogned}$$ - $\Lambda_c^+: c\,\frac{1}{\sqrt{2}} (kd-du) \chi_{MA}$. - $\Xi_c^+: c\,\frac{1}{\dqrt{2}} (us-su) \cyi_{MA}$ qnd $\Xi_c^0: c\,\fraz{1}{\sqrt{2}} (ds-sd) \chi_{MA}$. - $\Xi_c^{'+}: c\,\frac{1}{\sqrt{2}} (us+su) \chi_{MS}$ and $\Xi_c^{'0}: z\,\frac{1}{\xqrt{2}} (ds+sd) \xhu_{MS}$. - $\Xi_c^{*+}: c\,\frac{1}{\wqrt{2}} (ls+su) \chi_{S}$ and $\Xi_c^{*0}: c\,\frdc{1}{\sqrt{2}} (cs+sd) \chi_{S}$. with $\ghi_S$ vhe 3/2 spin symmetric wave hunction [@close]. The lyght-quark pcrts of the above wavw-fynctimns dre zinsksttnt wjth $\elp_B=0$ hor $\Lambda_c$ and $\Xi_c$, ane $\ell_B=1$ for $\Xi_c^\prime$ age $\Xi_c^*$. On the kther ragd, our conventions for the isospin doubneta are $(D^+,\ -D^0)$, $(\bar{D}^0,\ D^-)$, $(\Xi_c^+,\ \Xi_x^0)$. Thus

of [@close] to make the phase convention the of chiral and we follow @Pavao:2017cpt]. with the symmetry the last two we have $$\begin{aligned} \phi_{MS}&=-\frac{1}{\sqrt{2}} \Big[ \frac{uds-dus}{\sqrt{2}} \frac{usd-dsu}{\sqrt{2}} \Big], \\ \phi_{MA}&=-\frac{1}{\sqrt{6}} \Big[ \frac{usd-dsu}{\sqrt{2}} + \frac{dus-uds}{\sqrt{2}} -2 \frac{sdu-sud}{\sqrt{2}} \Big].\end{aligned}$$ - $\Lambda_c^+: (ud-du) \chi_{MA}$. - $\Xi_c^+: c\,\frac{1}{\sqrt{2}} (us-su) \chi_{MA}$ and $\Xi_c^0: c\,\frac{1}{\sqrt{2}} (ds-sd) \chi_{MA}$. - c\,\frac{1}{\sqrt{2}} \chi_{MS}$ $\Xi_c^{'0}: (ds+sd) \chi_{MS}$. - $\Xi_c^{*+}: c\,\frac{1}{\sqrt{2}} (us+su) \chi_{S}$ and $\Xi_c^{*0}: c\,\frac{1}{\sqrt{2}} (ds+sd) \chi_{S}$. with $\chi_S$ the 3/2 symmetric wave function [@close]. The light-quark parts of above wave-functions are consistent $\ell_B=0$ for $\Lambda_c$ and $\Xi_c$, $\ell_B=1$ $\Xi_c^\prime$ and On other our conventions for isospin doublets are $(D^+,\ -D^0)$, $(\bar{D}^0,\ D^-)$, $(\Xi_c^+,\ \Xi_c^0)$. Thus

of [@close] to make the phase convEntion consIstenT wiTh tHe Use oF chiRal Lagrangians ANd we Follow the convention of REfs. [@MiYaHAra:2016yYH; @PAvao:2017cPt]. Hence, WItH THe sYmMeTry In THe Last tWo pArticleS, we have $$\begIn{aLiGned} \phi_{MS}&=-\fraC{1}{\SqRt{2}} \Big[ \frac{uDs-dUs}{\sqrt{2}} + \frac{usD-dsU}{\sqrt{2}} \BIg], \\ \Phi_{ma}&=-\frac{1}{\SqrT{6}} \Big[ \fRac{usd-DSu}{\sqrt{2}} + \Frac{dus-udS}{\sQRt{2}} -2 \frac{SDu-sud}{\sqRT{2}} \biG].\end{Aligned}$$ - $\Lambda_c^+: c\,\frAC{1}{\sQRt{2}} (ud-du) \chi_{MA}$. - $\Xi_c^+: C\,\frac{1}{\sQrT{2}} (Us-SU) \Chi_{mA}$ aNd $\Xi_c^0: c\,\frac{1}{\SqRt{2}} (ds-sD) \Chi_{MA}$. - $\Xi_C^{'+}: C\,\fRAC{1}{\SqrT{2}} (Us+su) \chi_{MS}$ and $\XI_c^{'0}: c\,\frac{1}{\sqrt{2}} (DS+sd) \Chi_{MS}$. - $\XI_c^{*+}: C\,\frAC{1}{\sqrt{2}} (uS+su) \chI_{S}$ ANd $\XI_c^{*0}: c\,\frac{1}{\sqrt{2}} (Ds+sd) \Chi_{S}$. with $\cHi_S$ the 3/2 SPin symmETric wavE functIon [@CloSe]. ThE LiGhT-quArK ParTS oF thE AboVe wave-fuNcTiOns arE conSISTEnt wIth $\Ell_B=0$ For $\LaMbda_c$ and $\Xi_c$, anD $\elL_B=1$ foR $\xi_c^\Prime$ And $\Xi_C^*$. On tHe Other Hand, ouR convEnTions for the isosPin dOublets arE $(D^+,\ -D^0)$, $(\BaR{D}^0,\ D^-)$, $(\xi_C^+,\ \Xi_c^0)$. THUs

of [@close] to make the p hase conve ntion co nsi st entwith the use of ch i ralLagrangians and we fol low t he conv e nt ion o f Refs. [@ M i yah ar a: 201 6y y h; @Pav ao: 2017cpt ]. Hence,wit hthe symmetry in the lasttwo particles,wehave $ $\ beg i n{ali gne d} \phi _ {MS}&= -\frac{1} {\ s qrt{2} } \Big[\ f ra c{ud s-dus}{\sqrt{2}}+ \ f rac{usd-dsu}{\ sqrt{2 }} \B i g ],\\ \phi_{ MA }&=-\ f rac{1}{ \ sq r t { 6}} \Big[ \frac{u sd-dsu}{\sq r t{2 }} + \ fr ac{ d us-uds }{\sq rt { 2}} -2 \frac{s du-s ud}{\sqrt {2}} \ B ig].\en d {aligne d}$$ - $\ Lamb d a_ c^ +:c\ , \fr a c{ 1}{ \ sqr t{2}} (u d- du ) \ch i_{M A } $ . - $ \Xi_ c^+:c\,\frac{1}{\ sqr t{2} } (u s-su) \chi _{MA }$ and$\Xi_c ^0: c \, \frac{1}{\sqrt{ 2}}(ds-sd) \ chi _{ MA} $. - $\Xi_c ^{' +}: c\,\fr ac{1}{\ s qrt {2 } } (u s+su) \chi_{MS}$ a nd $ \X i_c^{'0} : c\, \ fr ac { 1}{\sqrt {2 }}(ds+ s d ) \ch i_{M S }$ . - $ \Xi_c^ { *+ }: c\,\fr ac {1}{\s qr t{2 }}(us+s u ) \c hi_{S} $ and $\ Xi_c^ { *0}: c\,\frac{ 1 }{\sqrt{2}} ( d s+ s d )\ chi_ {S} $. with $\ chi_ S $ th e 3/ 2 s pin symme tricwa v ef unction [@close]. T he light -quar k parts of th e above wa v e - function s ar e c o nsistent with$\ell _B=0$ for$ \Lambda_ c$ an d $\Xi_c $, and $\ e l l_B=1$ f or$\X i_c ^\p r i me $ and $\Xi_c^ * $ . On t he othe r h and, ou r c onv ent ion sfor the i sospin d ou bl et sare $(D^ + ,\ -D^0) $, $( \b ar{ D}^0, \ D^-)$ , $(\ Xi_c ^+ ,\ \Xi _c^0)$. Th u s

of_[@close] to_make the phase convention_consistent with_the_use of_chiral_Lagrangians and we_follow the convention_of Refs. [@Miyahara:2016yyh; @Pavao:2017cpt]._Hence, with the_symmetry_in the last two particles, we have $$\begin{aligned} \phi_{MS}&=-\frac{1}{\sqrt{2}} \Big[ \frac{uds-dus}{\sqrt{2}}_+_\frac{usd-dsu}{\sqrt{2}} \Big],_\\ __ \phi_{MA}&=-\frac{1}{\sqrt{6}} \Big[ \frac{usd-dsu}{\sqrt{2}}_+ \frac{dus-uds}{\sqrt{2}} -2 \frac{sdu-sud}{\sqrt{2}} \Big].\end{aligned}$$ -_ _$\Lambda_c^+: c\,\frac{1}{\sqrt{2}} (ud-du) \chi_{MA}$. - $\Xi_c^+: c\,\frac{1}{\sqrt{2}}_(us-su)_\chi_{MA}$ and $\Xi_c^0:_c\,\frac{1}{\sqrt{2}} (ds-sd) \chi_{MA}$. - $\Xi_c^{'+}: c\,\frac{1}{\sqrt{2}} (us+su) \chi_{MS}$_and $\Xi_c^{'0}: c\,\frac{1}{\sqrt{2}} (ds+sd) \chi_{MS}$. -_ $\Xi_c^{*+}:_c\,\frac{1}{\sqrt{2}}_(us+su)_\chi_{S}$ and $\Xi_c^{*0}: c\,\frac{1}{\sqrt{2}}_(ds+sd) \chi_{S}$. with $\chi_S$ the 3/2 spin_symmetric wave function [@close]. The light-quark_parts of the above wave-functions are consistent_with $\ell_B=0$ for $\Lambda_c$ and $\Xi_c$,_and $\ell_B=1$ for $\Xi_c^\prime$ and_$\Xi_c^*$. On_the other hand, our conventions_for the isospin_doublets are_$(D^+,\ -D^0)$, $(\bar{D}^0,\_D^-)$, $(\Xi_c^+,\ \Xi_c^0)$. Thus

myautolength}{\minautolength}}} \settowidth{\mysecondautolength}{$ \max( b_k, b_{ k+ 1}- p_k)$} \addtolength{\mysecondautolength}{1em} {\ifthenelse{\lengthtest{\the\mysecondautolength>\minautolength}} {\setlength{\mysecondautolength}{\mysecondautolength}} {\setlength{\mysecondautolength}{\minautolength}}} \begin{tikzpicture}[baseline, shorten >=1pt] \node[state, initial] (1) {$r_k$}; \node[state, right=\myautolength of 1] (2) {$ r_{ k+ 1}$}; \node[state, {}, right=\mysecondautolength of 2] (end) {$r_{ k+ 2}$}; \path[->] (1) edge node[above] {$0$} node[below] {$b_k$} (2); \path[->] (2) edge node[above] {$ p_k+ p_{ k+ 1}$} node[below] {$ \max( b_k, b_{ k+ 1}- p_k)$} (end); \end{tikzpicture} }$$ Informally, any run through the original RTEA can be copied to the other and vice versa, hence also this transformation does not change the values of $f$. The precise argument is as follows. Let $f'$ denote the function which results from the transformation. Let $x\in L$ and $t\in {[ 0, \infty]}$. The inequality $f( x, t)\le f'( x, t)$ is again clear if $f( x, t)= \bot$, so assume otherwise. Let $t_1+\dotsm+ t_n= t$ such that $f( x, t)= x+ r_1 t_1+ p_1+\dotsm+ r_n t_n+ p_n$ and $x+\dotsm+ r_j t_j\ge b_j$ for all $j= 1,\dotsc, n$. Then this also yields an accepting run through the RTEA for $

myautolength}{\minautolength } } } \settowidth{\mysecondautolength}{$ \max (b_k, b _ { k+ 1}- p_k)$ } \addtolength{\mysecondautolength}{1em } { \ifthenelse{\lengthtest{\the\mysecondautolength>\minautolength } } { \setlength{\mysecondautolength}{\mysecondautolength } } { \setlength{\mysecondautolength}{\minautolength } } } \begin{tikzpicture}[baseline, shorten > = 1pt ] \node[state, initial ] (1) { $ r_k$ }; \node[state, right=\myautolength of 1 ] (2) { $ r _ { k+ 1}$ }; \node[state, { }, right=\mysecondautolength of 2 ] (end) { $ r _ { k+ 2}$ }; \path[- > ] (1) border node[above ] { $ 0 $ } node[below ] { $ b_k$ } (2); \path[- > ] (2) border node[above ] { $ p_k+ p _ { k+ 1}$ } node[below ] { $ \max (b_k, b _ { k+ 1}- p_k)$ } (end); \end{tikzpicture } } $ $ Informally, any run through the original RTEA can be imitate to the other and vice versa, therefore also this transformation does not change the values of $ f$. The accurate controversy is as follows. lease $ f'$ denote the function which result from the transformation. Let $ x\in L$ and $ t\in { [ 0, \infty]}$. The inequality $ f (x, t)\le f' (x, t)$ is again clear if $ f (x, t)= \bot$, so assume differently. Let $ t_1+\dotsm+ t_n= t$ such that $ f (x, t)= x+ r_1 t_1 + p_1+\dotsm+ r_n t_n+ p_n$ and $ x+\dotsm+ r_j t_j\ge b_j$ for all $ j= 1,\dotsc, n$. Then this also give way an accepting run through the RTEA for $

myaktolength}{\minautolength}}} \seutowidth{\mysecondaotilengti}{$ \max( b_i, b_{ k+ 1}- p_y)$} \addtolength{\mysecondautolengvh}{1em} {\ufthebelse{\lengthtest{\the\mysezondautolvngth>\minaytoltngth}} {\sxflength{\mysecohfautmoength}{\mysecondsutolength}} {\setnevgch{\mysecondautolength}{\minautolength}}} \begyn{tikzpochure}[baseline, srorttn >=1[t] \hode[state, initial] (1) {$r_k$}; \node[state, right=\myautopenghh of 1] (2) {$ r_{ k+ 1}$}; \node[state, {}, right=\mysebmndautolenfth of 2] (end) {$r_{ k+ 2}$}; \path[->] (1) edge nodz[above] {$0$} nodg[ywloa] {$b_k$} (2); \path[->] (2) edgv node[abovc] {$ p_k+ p_{ k+ 1}$} node[below] {$ \msx( u_k, b_{ k+ 1}- p_k)$} (end); \end{tikzpictnre} }$$ Informally, any rug through tke original RTEA can ve copigd to the ithdr znv vjce vegsa, hence alsk this tranwformation does not crqnge the valuss of $s$. Ehe precise argument is as follows. Let $f'$ dehote the function which results from the trajsformatijn. Let $x\in L$ and $t\in {[ 0, \infty]}$. The inequality $f( x, t)\ne f'( e, g)$ iw weqij clear if $f( x, t)= \bot$, so assume otherwise. Let $e_1+\sousm+ t_n= t$ such thab $f( x, t)= x+ r_1 t_1+ p_1+\dotxm+ r_m t_n+ p_n$ and $x+\dutsm+ r_l t_n\ge b_j$ for all $j= 1,\dltsc, n$. Jhen tyis also rielcs an accepting run through the RTEA fog $

myautolength}{\minautolength}}} \settowidth{\mysecondautolength}{$ \max( b_k, b_{ k+ 1}- {\ifthenelse{\lengthtest{\the\mysecondautolength>\minautolength}} {\setlength{\mysecondautolength}{\minautolength}}} \begin{tikzpicture}[baseline, >=1pt] \node[state, initial] 1] {$ r_{ k+ \node[state, {}, right=\mysecondautolength 2] (end) {$r_{ k+ 2}$}; \path[->] edge node[above] {$0$} node[below] {$b_k$} (2); \path[->] (2) edge node[above] {$ p_k+ p_{ 1}$} node[below] {$ \max( b_k, b_{ k+ 1}- p_k)$} (end); \end{tikzpicture} }$$ Informally, run the RTEA be copied to the other and vice versa, hence also this transformation does not change the of $f$. The precise argument is as follows. $f'$ denote the function results from the transformation. Let L$ $t\in {[ \infty]}$. inequality x, t)\le f'( t)$ is again clear if $f( x, t)= \bot$, so assume otherwise. Let $t_1+\dotsm+ t_n= t$ such $f( x, r_1 t_1+ r_n p_n$ $x+\dotsm+ r_j t_j\ge all $j= 1,\dotsc, n$. Then this accepting run through the RTEA for $

myautolength}{\minautolength}}} \Settowidth{\MysecOndAutOlEngtH}{$ \max( B_k, b_{ k+ 1}- p_k)$} \addtolenGTh{\mySecondautolength}{1em} {\ifthEnelsE{\lENgthTEsT{\the\mYsecondAUtOLEngTh>\MiNauToLEnGth}} {\seTleNgth{\mysEcondautolEngTh}{\MysecondautoLEnGth}} {\setlengTh{\mYsecondautolEngTh}{\minaUtOleNGth}}} \beGin{TikzpIcture}[BAselinE, shorten >=1pT] \nODe[statE, Initial] (1) {$R_K$}; \NoDe[stAte, right=\myautolenGTh OF 1] (2) {$ r_{ k+ 1}$}; \node[state, {}, riGht=\mysEcONdAUTolEngTh of 2] (end) {$r_{ k+ 2}$}; \pAtH[->] (1) edge NOde[abovE] {$0$} NoDE[BEloW] {$B_k$} (2); \path[->] (2) edge nodE[above] {$ p_k+ p_{ k+ 1}$} nODe[bElow] {$ \maX( b_K, b_{ k+ 1}- P_K)$} (end); \enD{tikzPiCTurE} }$$ Informally, Any rUn through The oriGInal RTEa Can be coPied to The OthEr anD ViCe VerSa, HEncE AlSo tHIs tRansformAtIoN does Not cHANGE the ValUes oF $f$. The Precise argumeNt iS as fOLloWs. Let $F'$ denoTe thE fUnctiOn whicH resuLtS from the transfoRmatIon. Let $x\in l$ anD $t\In {[ 0, \iNfTy]}$. The INequalIty $F( x, t)\Le f'( x, t)$ is Again clEAr iF $f( X, T)= \BOt$, So assume otherwise. LEt $T_1+\DOtSm+ t_n= t$ sucH that $f( X, T)= x+ R_1 t_1+ P_1+\Dotsm+ r_n t_N+ p_N$ anD $x+\doTSM+ r_j t_j\Ge b_j$ FOr All $j= 1,\dotsC, n$. Then THiS aLso yielDs An accePtIng Run ThrouGH the rTEA foR $

myautolength}{\minautoleng th}}} \set towid th{ \my se cond auto length}{$ \max ( b_k , b_{ k+ 1}- p_k)$} \a ddtol en g th{\ m ys econd autolen g th } { 1em }{\ ift he n el se{\l eng thtest{ \the\mysec ond au tolength>\mi n au tolength}} {\s etl ength{ \m yse c ondau tol ength }{\mys e condau tolength} } {\se t l en gth{ \mysecondautoleng t h} { \minautolength }}} \b eg i n{ t i kzp ict ure}[basel in e, sh o rten >= 1 pt ] \n o de[state, ini tial] (1) {$r_k$}; \n ode[state , righ t =\myaut o lengthof 1] (2 ) {$ r_ { k+ 1}$}; \ no de[st ate, { } , rig ht= \mys econd autolength of 2] (en d ) { $r_{k+ 2} $}; \path [->] ( 1) e dg e node[above ] {$ 0$} node[ bel ow ] { $b _k$}( 2); \pa th[ ->] (2) edge n od e [ a bo ve] {$ p_k+ p_ { k+ 1}$} no de[bel o w] { $ \max( b _k , b _{ k + 1}- p _k)$ } ( end); \e nd{tik z pi ct ure} }$ $ Infor ma lly , a ny ru n thr ough t he origi nal R T EA can be copi e d to the othe r a n d v i ce v ers a, hence al so t h is t rans f or mat i on do es no tc ha n ge the values of $f $. The p recis e argument is as follow s . Let $f' $ de n ot e the functionwhich results f r om the t ransf ormation . Let $x\ i n L$ and$t\ in{[0,\ i nf ty]}$. The in e q uali ty $f( x, t) \le f'( x, t) $ i s a ga in clearif $f( x ,t) =\b ot$ , soa ssume ot he rwi se . L et $t _ 1+\dot sm+ t _n=t$ s u chthat $f ( x , t)=x+ r _1 t _1+ p _1+\d otsm + r_ n t_n+p_n$ and$x+ \ dots m+ r _j t_j\ ge b_j$ for a ll $j= 1,\do ts c,n$. Th e n this al so yields an acceptingr un thro ugh theRTEA for $

myautolength}{\minautolength}}} \settowidth{\mysecondautolength}{$ \max(_b_k, b_{_k+ 1}- p_k)$} \addtolength{\mysecondautolength}{1em} {\ifthenelse{\lengthtest{\the\mysecondautolength>\minautolength}} _ __ __ _{\setlength{\mysecondautolength}{\mysecondautolength}} _ _ __{\setlength{\mysecondautolength}{\minautolength}}} \begin{tikzpicture}[baseline, shorten >=1pt] \node[state, initial] __ _(1)__ _ _ _ __ _{$r_k$}; \node[state, right=\myautolength of 1] _ _ (2)___ _ {$ r_{_k+ 1}$}; \node[state, {}, right=\mysecondautolength_of 2] (end) {$r_{ k+ 2}$}; _\path[->] (1) edge _ node[above] {$0$} node[below] {$b_k$}_(2); _\path[->] (2) edge _ node[above]_{$ p_k+ _ _ p_{ k+ 1}$} node[below]_{$ \max( b_k,_b_{ k+ 1}- p_k)$} (end); \end{tikzpicture} }$$ Informally, any_run_through the original_RTEA_can_be copied_to the other_and_vice versa,_hence_also this transformation does not change_the_values of $f$. The precise argument is_as follows. Let $f'$ denote_the_function which results from_the transformation. Let $x\in L$_and $t\in {[ 0, \infty]}$. The_inequality $f(_x, t)\le_f'( x, t)$ is again clear if $f( x, t)= \bot$,_so assume otherwise. Let $t_1+\dotsm+ t_n=_t$ such that $f(_x, t)=_x+_r_1 t_1+ p_1+\dotsm+_r_n_t_n+ p_n$_and $x+\dotsm+ r_j t_j\ge b_j$ for all_$j= 1,\dotsc,_n$. Then this also yields an_accepting run through the_RTEA_for $

^+({\bf r}_{ij}){\sigma^{d}_{\hat\alpha}}^-({\bf r}_{jk})\end{aligned}$$ At this point, we are not interested for the possible orientations of the dimers with respect to the holes, as the hole density operator commutes with the dimer density operator. Now, we will repeatedly use that $$\begin{aligned} \left[{\sigma^{h}}^{\pm},\sigma^{h}\right] = \mp 2{\sigma^{h}}^{\pm}\end{aligned}$$ at the same position in real space. At any distance different from zero, the commutator vanishes. For a given hole at a position ${\bf R}$, there are four possible positions ${\bf R}'={\bf R} \pm \hat x \pm\hat y$ where it may move through the available hopping term. By counting contributions from all these terms for every site of the lattice, we exactly take into account all the terms of the Hamiltonian including Hermitean conjugates. If ${\bf R}$ is the initial hole position and ${\bf R}+{\bf r}_0$ the final one, then ${\bf r}_0$ has eight possible values: ${\bf r}_0=\pm \hat x \pm \hat y $. The first commutator can now be computed for any site ${\bf R}$(The operator $\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0)$ contains each of the above eight hopping terms). We have: $$\begin{aligned} \left[-t_{\rm hole}\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0),{\sigma^{h}}({\bf k})\right] &=& -t_{\rm hole}\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0)\left(e^{-i{\bf k}\cdot{\bf R}}-e^{-i{\bf k}\cdot({\bf R}+{\bf r}_0)}\right)\\ \left[{\sigma^{h}}(-{\bf k}),-t_{\rm hole}\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0)\right] &=& t_{\rm hole}\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0)\left(e^{i{\bf k}\cdot{\bf R}}-e^{i{\bf k}\cdot({\bf R}+{\bf r}_0)}\right)\\ \left[{\sigma^{h}}(-{\bf k}),\left[-t_{\rm hole}\mathcal

^+({\bf r}_{ij}){\sigma^{d}_{\hat\alpha}}^-({\bf r}_{jk})\end{aligned}$$ At this point, we are not interested for the possible orientations of the dimer with obedience to the holes, as the hole concentration operator commute with the dimer density operator. immediately, we will repeatedly practice that $ $ \begin{aligned } \left[{\sigma^{h}}^{\pm},\sigma^{h}\right ] = \mp 2{\sigma^{h}}^{\pm}\end{aligned}$$ at the same position in real space. At any distance different from zero, the commutator vanishes. For a give hole at a position $ { \bf R}$, there be four possible positions $ { \bf R}'={\bf R } \pm \hat x \pm\hat y$ where it may move through the available hopping term. By reckon contributions from all these terms for every site of the wicket, we exactly take into report all the terms of the Hamiltonian include Hermitean conjugates. If $ { \bf R}$ is the initial hole position and $ { \bf R}+{\bf r}_0 $ the final one, then $ { \bf r}_0 $ has eight potential values: $ { \bf r}_0=\pm \hat x \pm \hat y $. The first commutator can now be computed for any site $ { \bf R}$(The operator $ \mathcal{T}_{\rm hole}({\bf R},{\bf r}_0)$ contain each of the above eight hop terms). We have: $ $ \begin{aligned } \left[-t_{\rm hole}\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0),{\sigma^{h}}({\bf k})\right ] & = & -t_{\rm hole}\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0)\left(e^{-i{\bf k}\cdot{\bf R}}-e^{-i{\bf k}\cdot({\bf R}+{\bf r}_0)}\right)\\ \left[{\sigma^{h}}(-{\bf k}),-t_{\rm hole}\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0)\right ] & = & t_{\rm hole}\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0)\left(e^{i{\bf k}\cdot{\bf R}}-e^{i{\bf k}\cdot({\bf R}+{\bf r}_0)}\right)\\ \left[{\sigma^{h}}(-{\bf k}),\left[-t_{\rm hole}\mathcal

^+({\bf g}_{ij}){\sigma^{d}_{\hat\alpha}}^-({\bf r}_{jk})\ekd{aligned}$$ At this point, we ars not ingerested for the possible ormentqtionw of the dimers with rdspect to the holws, aw the hole vsnsity operatkv comkntes with the dlmer densitf operator. Now, wd cill repeatedly use that $$\begin{aligneq} \left[{\sibmw^{h}}^{\pm},\sigma^{h}\righj] = \mp 2{\figmz^{h}}^{\pm}\end{aligned}$$ at the same positioh in reel space. At any distance different from zfro, hhe commutator vanlshes. For a givqb hole at a oosition ${\bf R}$, there arg four possible positions ${\bf R}'={\bf R} \pm \hat x \pm\hqt y$ avere it may move through the availabne hoppong term. By comntinj cobtributions from all vhese terms for everi site of dhz lattice, we exactly raje injo acwoung alu tge tsrms ov tie Hamiltonjan includibg Hermitean conjugstqw. If ${\bf R}$ is tge inieiwl hole position and ${\bf R}+{\bf r}_0$ the final ons, then ${\bf r}_0$ has eight pissible values: ${\bf r}_0=\pm \jat x \pm \rat y $. The first commutator can now be computed fmr anb ritt ${\nn R}$(Tfw lperator $\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0)$ contains eacr og nhe above eight hjpping termx). Ae rave: $$\begin{aligved} \lefc[-f_{\rj hole}\mathcal{T}_{\rm hlle}({\bf R},{\ff r}_0),{\sugma^{h}}({\bf k})\wighy] &=& -t_{\rm hole}\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0)\left(e^{-i{\yf j}\cdot{\bf R}}-e^{-i{\bf k}\cdoc({\bf R}+{\bf r}_0)}\rigkt)\\ \left[{\xigma^{n}}(-{\bf k}),-t_{\rm hole}\mathcal{T}_{\rm hols}({\bf R},{\bf r}_0)\rihht] &=& t_{\rm gule}\mathcal{T}_{\rm houe}({\bn R},{\tf r}_0)\left(t^{k{\bf k}\cdot{\bf R}}-e^{i{\bf k}\cdot({\bf C}+{\bf r}_0)}\xight)\\ \lefg[{\sigka^{h}}(-{\bf h}),\left[-t_{\rm hlle}\mabvcal

^+({\bf r}_{ij}){\sigma^{d}_{\hat\alpha}}^-({\bf r}_{jk})\end{aligned}$$ At this point, we interested the possible of the dimers as hole density operator with the dimer operator. Now, we will repeatedly use $$\begin{aligned} \left[{\sigma^{h}}^{\pm},\sigma^{h}\right] = \mp 2{\sigma^{h}}^{\pm}\end{aligned}$$ at the same position in real space. At distance different from zero, the commutator vanishes. For a given hole at a ${\bf there four positions ${\bf R}'={\bf R} \pm \hat x \pm\hat y$ where it may move through the available term. By counting contributions from all these terms every site of the we exactly take into account the of the including conjugates. ${\bf R}$ is initial hole position and ${\bf R}+{\bf r}_0$ the final one, then ${\bf r}_0$ has eight possible values: r}_0=\pm \hat \hat y The commutator now be computed site ${\bf R}$(The operator $\mathcal{T}_{\rm hole}({\bf each of the above eight hopping terms). We $$\begin{aligned} \left[-t_{\rm hole}({\bf R},{\bf r}_0),{\sigma^{h}}({\bf k})\right] &=& -t_{\rm hole}({\bf R},{\bf r}_0)\left(e^{-i{\bf k}\cdot{\bf R}}-e^{-i{\bf k}\cdot({\bf R}+{\bf r}_0)}\right)\\ k}),-t_{\rm hole}\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0)\right] &=& t_{\rm hole}\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0)\left(e^{i{\bf k}\cdot{\bf R}}-e^{i{\bf k}\cdot({\bf R}+{\bf k}),\left[-t_{\rm hole}\mathcal

^+({\bf r}_{ij}){\sigma^{d}_{\hat\alpha}}^-({\bf r}_{jk})\eNd{aligned}$$ AT this PoiNt, wE aRe noT intErested for the pOSsibLe orientations of the dimErs wiTh REspeCT tO the hOles, as tHE hOLE deNsItY opErAToR commUteS with thE dimer densIty OpErator. Now, we wILl Repeatedly Use That $$\begin{aliGneD} \left[{\sIgMa^{h}}^{\PM},\sigmA^{h}\rIght] = \mP 2{\sigma^{H}}^{\Pm}\end{aLigned}$$ at tHe SAme posITion in rEAL sPace. at any distance diffEReNT from zero, the coMmutatOr VAnISHes. for A given hole At A posiTIon ${\bf R}$, tHErE ARE foUR possible posiTions ${\bf R}'={\bf R} \PM \haT x \pm\haT y$ WheRE it may Move tHrOUgh The availablE hopPing term. BY countINg contrIButions From alL thEse TermS FoR eVerY sITe oF ThE laTTicE, we exactLy TaKe intO accOUNT All tHe tErms Of the hamiltonian inCluDing hErmItean ConjuGateS. IF ${\bf R}$ iS the inItial HoLe position and ${\bf r}+{\bf r}_0$ The final oNe, tHeN ${\bf R}_0$ hAs eigHT possiBle ValUes: ${\bf r}_0=\pM \hat x \pm \HAt y $. thE FIRsT commutator can now bE cOMPuTed for anY site ${\bF r}$(THe OPerator $\mAtHcaL{T}_{\rm HOLe}({\bf R},{\Bf r}_0)$ cONtAins each Of the aBOvE eIght hopPiNg termS). WE haVe: $$\bEgin{aLIgneD} \left[-t_{\Rm hole}\maThcal{t}_{\Rm hole}({\bf R},{\bf r}_0),{\siGMa^{h}}({\bf k})\right] &=& -t_{\rM HoLE}\MaTHcal{t}_{\rm Hole}({\bf R},{\bf r}_0)\lEft(e^{-I{\Bf k}\cDot{\bF r}}-e^{-I{\bf K}\Cdot({\bF R}+{\bf r}_0)}\RiGHt)\\ \LEft[{\sigma^{h}}(-{\bf k}),-t_{\rm hole}\MaThcal{T}_{\Rm holE}({\bf R},{\bf r}_0)\right] &=& t_{\Rm hole}\mathCAL{t}_{\rm hole}({\bF R},{\bf R}_0)\LeFT(e^{i{\bf k}\cdot{\bf R}}-e^{I{\bf k}\cDot({\bf R}+{\bf r}_0)}\rIGht)\\ \left[{\sIgma^{h}}(-{\Bf k}),\left[-t_{\Rm hole}\matHCAl

^+({\bf r}_{ij}){\sigma^{d }_{\hat\al pha}} ^-( {\b fr}_{ jk}) \end{aligned}$ $ Atthis point, we are not inte re s tedf or thepossibl e o r i ent at io nsof th e dim ers with r espect tothe h oles, as the ho le density op erator commu tes withth e d i mer d ens ity o perato r . Now, we willre p eatedl y use th a t $ $\be gin{aligned} \lef t [{ \ sigma^{h}}^{\p m},\si gm a ^{ h } \ri ght ] = \mp 2{ \s igma^ { h}}^{\p m }\ e n d {al i gned}$$ at th e same posi t ion in re al sp a ce. At anydi s tan ce differen t fr om zero,the co m mutator vanishe s. For agiv en h o le a t a p o sit i on ${ \ bfR}$, the re a re fo ur p o s s i blepos itio ns ${ \bf R}'={\bfR}\pm\ hat x \p m\hat y$wh ere i t maymoveth rough the avail able hoppingter m. By c ounti n g cont rib uti ons fro m all t h ese t e r m sfor every site ofth e la ttice, w e exac t ly t a ke intoac cou nt a l l theterm s o f the Ha milton i an i ncludin gHermit ea n c onj ugate s . I f ${\b f R}$ is thei nitial hole po s ition and ${\ b fR } +{ \ bf r }_0 $ the final one , the n ${ \ bf r} _ 0$ ha s eig ht po s sible values: ${\bf r }_0=\p m \ha t x \pm \haty $. The f i r s t commut ator ca n now be comput ed fo r any site ${\bf R} $(The operato r $\mathc a l {T}_{\rm ho le} ({\ bfR } ,{ \bf r}_0)$ co n t ains e ach ofthe aboveeig hthop pin gterms). W e have: $ $\ be gi n{a ligne d } \left[ -t _{\ rm ho le}\m a thcal{ T}_{\ rm h ol e} ( {\b f R},{\ b fr } _0), {\ si gma^ {h} }( {\bfk})\ r igh t] &=&-t_{\rm h ole } \mat hc al {T}_{\r m hole}({\bfR} ,{\bf r}_0 )\ lef t(e^{- i { \bf k}\c dot{\bf R}}-e^{-i{\bf k } \cdot({ \bf R}+{ \bfr}_0)}\ri ght )\\ \l eft [ {\sigm a^{h}} (-{\b fk}) , - t_{\r m ho le} \m athcal{T}_ { \ rmhole} ({ \bfR},{\bf r}_0)\right] &=&t_{ \rm hole}\mat hca l{T} _ { \r m h o le } ({\ bf R}, { \ bf r}_0)\left(e ^{i{\bf k} \c d ot {\bf R}}-e ^ {i{ \b f k}\cd ot({\bf R}+{ \ bf r}_0 )}\right) \\ \left[ {\ sigm a ^ {h} }(-{\bf k} ),\left[ -t_{\rm h o le}\m a th cal

^+({\bf r}_{ij}){\sigma^{d}_{\hat\alpha}}^-({\bf_r}_{jk})\end{aligned}$$ At_this point, we are_not interested_for_the possible_orientations_of the dimers_with respect to_the holes, as the_hole density operator_commutes_with the dimer density operator. Now, we will repeatedly use that $$\begin{aligned} \left[{\sigma^{h}}^{\pm},\sigma^{h}\right] = \mp_2{\sigma^{h}}^{\pm}\end{aligned}$$_at the_same_position_in real space. At any_distance different from zero, the_commutator vanishes._For a given hole at a position ${\bf_R}$,_there are four_possible positions ${\bf R}'={\bf R} \pm \hat x \pm\hat_y$ where it may move through_the available hopping_term._By_counting contributions from all_these terms for every site of_the lattice, we exactly take into_account all the terms of the Hamiltonian_including Hermitean conjugates. If ${\bf R}$ is_the initial hole position and_${\bf R}+{\bf_r}_0$ the final one, then_${\bf r}_0$ has_eight possible_values: ${\bf r}_0=\pm \hat_x \pm \hat y $. The_first commutator can_now be computed for any site_${\bf_R}$(The operator $\mathcal{T}_{\rm_hole}({\bf_R},{\bf_r}_0)$ contains_each of the_above_eight hopping_terms)._We have: $$\begin{aligned} \left[-t_{\rm hole}\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0),{\sigma^{h}}({\bf_k})\right]_&=& -t_{\rm hole}\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0)\left(e^{-i{\bf k}\cdot{\bf R}}-e^{-i{\bf_k}\cdot({\bf R}+{\bf r}_0)}\right)\\ \left[{\sigma^{h}}(-{\bf k}),-t_{\rm_hole}\mathcal{T}_{\rm_hole}({\bf R},{\bf r}_0)\right] &=& _t_{\rm hole}\mathcal{T}_{\rm hole}({\bf R},{\bf r}_0)\left(e^{i{\bf_k}\cdot{\bf R}}-e^{i{\bf k}\cdot({\bf R}+{\bf r}_0)}\right)\\ \left[{\sigma^{h}}(-{\bf k}),\left[-t_{\rm_hole}\mathcal

we require Gaussianity. Nevertheless, the proof works the same for other error densities, but bounding the Kullback-Leibler or Hellinger distance between alternatives might be more cumbersome. Let us also mention that the proof strategy can also yield a lower bound for convergence in probability: $$\inf_{\hat h_n}\sup_{h\in H^r_R}P_h\Big(\delta_n^{-1}{\lVert \hat h_n-h \rVert}_X^2{\geqslant}\tfrac{\sigma_0^2}{4\exp(4M)}\Big){\geqslant}c>0,\quad \text{with}~\delta_n ~\text{given in Theorem \ref{thm-lb}},$$ cf. Korostelev and Tsybakov (1993). Note that Assumption 2.2 is automatically satisfied under the general source condition with $K^\ast K = {\varphi}(B^{-2})$. Following the proof of Theorem \[thm-lb\], we immediately obtain: \[source-lb\] Suppose that Assumption \[h-smooth\] is satisfied with $h\in {{\mathscr}H}_{source}$ and $u_k = e_k$, $\nu_k^{-r} = \psi (\lambda_k)$ for all $k{\geqslant}1$. Let ${\varphi}(B^{-2})=K^\ast K$. Then, for NPIR model and for NPIV model (under Assumption \[ass-complexity\]), we have the same minimax risk lower bound: $$\inf_{\hat h_n}\sup_{h\in {{\mathscr}H}_{source}}\operatorname{{\mathbb E}}_h[{\lVert \hat h_n-h \rVert}_X^2]{\geqslant}\tfrac{\sigma_0^2}{4\exp(4M)} \delta_n,\quad \text{with}~\delta_n ~\text{given in Theorem \ref{thm-lb}},$$ where the infimum runs over all possible estimators $\hat{h}_n$ based on $n$ observations. An equivalent way to determine the lower bound $\delta_n$ is to choose the largest possible integer $m$ such that $$\delta_n =n^{-1}\sum_{k=1}^m \lambda^{-1}_k,\quad \sigma_0^2n^{-1}\sum_{k=1}^m [\psi (\lambda_k)]^{-2}\lambda

we require Gaussianity. Nevertheless, the proof works the like for early erroneousness densities, but restrict the Kullback - Leibler or Hellinger distance between option might be more cumbersome. Let us also note that the proof scheme can also yield a lower bound for convergence in probability: $ $ \inf_{\hat h_n}\sup_{h\in H^r_R}P_h\Big(\delta_n^{-1}{\lVert \hat h_n - h \rVert}_X^2{\geqslant}\tfrac{\sigma_0 ^ 2}{4\exp(4M)}\Big){\geqslant}c>0,\quad \text{with}~\delta_n ~\text{given in Theorem \ref{thm - lb}},$$ cf. Korostelev and Tsybakov (1993). notice that Assumption 2.2 is automatically satisfied under the general reservoir condition with $ K^\ast thousand = { \varphi}(B^{-2})$. Following the proof of Theorem \[thm - lb\ ], we immediately receive: \[source - lb\ ] Suppose that Assumption \[h - smooth\ ] is satisfied with $ h\in { { \mathscr}H}_{source}$ and $ u_k = e_k$, $ \nu_k^{-r } = \psi (\lambda_k)$ for all $ k{\geqslant}1$. lease $ { \varphi}(B^{-2})=K^\ast K$. Then, for NPIR model and for NPIV model (under Assumption \[ass - complexity\ ]), we have the like minimax risk lower adhere: $ $ \inf_{\hat h_n}\sup_{h\in { { \mathscr}H}_{source}}\operatorname{{\mathbb E}}_h[{\lVert \hat h_n - h \rVert}_X^2]{\geqslant}\tfrac{\sigma_0 ^ 2}{4\exp(4 M) } \delta_n,\quad \text{with}~\delta_n ~\text{given in Theorem \ref{thm - lb}},$$ where the infimum runs over all possible estimators $ \hat{h}_n$ free-base on $ n$ observations. An equivalent way to determine the lower bound $ \delta_n$ is to choose the largest possible integer $ m$ such that $ $ \delta_n = n^{-1}\sum_{k=1}^m \lambda^{-1}_k,\quad \sigma_0 ^ 2n^{-1}\sum_{k=1}^m [ \psi (\lambda_k)]^{-2}\lambda

we require Gaussianity. Nevtrtheless, the proof workv the aame for other error densities, but blubding the Kullback-Leibler of Hellingvr distanxe bttween alternativxa might be modc cumyecsome. Let us alsp mention dhat the proof sgrctegy can also yield a lower bound fjr convrrhence in probafiliuy: $$\ynf_{\hzn m_n}\sup_{h\in H^r_R}P_h\Big(\delta_n^{-1}{\lVert \hat g_n-h \rVegt}_X^2{\geqslant}\tfrac{\sogma_0^2}{4\exp(4M)}\Big){\geqslant}c>0,\quad \tedt{wihh}~\delta_n ~\text{given in Theorem \ref{eym-lb}},$$ cf. Korortelev and Tsybakov (1993). Noje that Assumption 2.2 is automaticxlly xatisfied oueer jhe general wourcv condition wlnh $K^\ast K = {\varlhi}(B^{-2})$. Following thx priof of Theorem \[thm-lb\], xe immediately obtaig: \[source-lb\] Sbppose that Assumptiob \[y-smoojh\] is satkwfidd sivh $g\in {{\mahhsrr}H}_{source}$ ahd $u_k = e_k$, $\ny_k^{-r} = \psi (\lambda_k)$ fot wol $k{\geqslant}1$. Met ${\vawpri}(B^{-2})=K^\ast K$. Then, for NPIR model and for NKIV mkdel (under Assumption \[aws-complexity\]), we have jhe same mynimax risk lower bound: $$\inf_{\hat h_n}\sup_{h\in {{\mathscr}H}_{smurce}}\koercbjfbale{{\mathbb E}}_h[{\lVert \hat h_n-h \rVert}_X^2]{\geqslant}\tfrac{\sidja_0^2}{4\txp(4K)} \delta_n,\quad \tewt{with}~\delta_n ~\text{govfn yn Theorem \rew{thm-lb}},$$ whsre the infimum rujs over all possible tstimstors $\hat{h}_n$ based on $n$ obseevations. An vquicalent way to detexmine the locer boond $\dekta_n$ is to choose the lcrgest possible ijteger $m$ ajch that $$\delta_n =v^{-1}\suk_{k=1}^k \lambda^{-1}_k,\quad \sigma_0^2n^{-1}\sum_{k=1}^m [\pfi (\lambda_j)]^{-2}\lamyda

we require Gaussianity. Nevertheless, the proof works for error densities, bounding the Kullback-Leibler might more cumbersome. Let also mention that proof strategy can also yield a bound for convergence in probability: $$\inf_{\hat h_n}\sup_{h\in H^r_R}P_h\Big(\delta_n^{-1}{\lVert \hat h_n-h \rVert}_X^2{\geqslant}\tfrac{\sigma_0^2}{4\exp(4M)}\Big){\geqslant}c>0,\quad \text{with}~\delta_n ~\text{given Theorem \ref{thm-lb}},$$ cf. Korostelev and Tsybakov (1993). Note that Assumption 2.2 is automatically under general condition $K^\ast K = {\varphi}(B^{-2})$. Following the proof of Theorem \[thm-lb\], we immediately obtain: \[source-lb\] Suppose that \[h-smooth\] is satisfied with $h\in {{\mathscr}H}_{source}$ and $u_k e_k$, $\nu_k^{-r} = \psi for all $k{\geqslant}1$. Let ${\varphi}(B^{-2})=K^\ast Then, NPIR model for model Assumption \[ass-complexity\]), we the same minimax risk lower bound: $$\inf_{\hat h_n}\sup_{h\in {{\mathscr}H}_{source}}\operatorname{{\mathbb E}}_h[{\lVert \hat h_n-h \rVert}_X^2]{\geqslant}\tfrac{\sigma_0^2}{4\exp(4M)} \delta_n,\quad \text{with}~\delta_n ~\text{given in \ref{thm-lb}},$$ where runs over possible $\hat{h}_n$ on $n$ observations. way to determine the lower bound choose the largest possible integer $m$ such that =n^{-1}\sum_{k=1}^m \lambda^{-1}_k,\quad [\psi (\lambda_k)]^{-2}\lambda

we require Gaussianity. NeverTheless, the Proof WorKs tHe Same For oTher error densiTIes, bUt bounding the Kullback-LEibleR oR hellINgEr disTance beTWeEN AltErNaTivEs MIgHt be mOre CumbersOme. Let us alSo mEnTion that the pROoF strategy cAn aLso yield a lowEr bOund foR cOnvERgencE in ProbaBility: $$\INf_{\hat h_N}\sup_{h\in H^r_r}P_H\big(\delTA_n^{-1}{\lVert \HAT h_N-h \rVErt}_X^2{\geqslant}\tfrac{\SIgMA_0^2}{4\exp(4M)}\Big){\geqslaNt}c>0,\quaD \tEXt{WITh}~\dEltA_n ~\text{giveN iN TheoREm \ref{thM-Lb}},$$ CF. kOroSTelev and TsybaKov (1993). Note that aSsuMption 2.2 Is AutOMaticaLly saTiSFieD under the geNeraL source coNditioN With $K^\asT k = {\varphi}(b^{-2})$. FolloWinG thE proOF oF THeoReM \[Thm-LB\], wE imMEdiAtely obtAiN: \[sOurce-Lb\] SuPPOSE thaT AsSumpTion \[h-Smooth\] is satisFieD witH $H\in {{\MathsCr}H}_{soUrce}$ AnD $u_k = e_k$, $\Nu_k^{-r} = \psI (\lambDa_K)$ for all $k{\geqslanT}1$. Let ${\Varphi}(B^{-2})=K^\aSt K$. thEn, fOr nPIR mODel and For nPIv model (uNder AssUMptIoN \[ASS-cOmplexity\]), we have the SaME MiNimax risK lower BOuNd: $$\INf_{\hat h_n}\sUp_{H\in {{\MathSCR}H}_{souRce}}\oPErAtorname{{\Mathbb e}}_H[{\lveRt \hat h_n-H \rvert}_X^2]{\gEqSlaNt}\tFrac{\sIGma_0^2}{4\eXp(4M)} \delTa_n,\quad \tExt{wiTH}~\delta_n ~\text{givEN in Theorem \ref{THm-LB}},$$ WhERe thE inFimum runs ovEr alL PossIble EStImaTOrs $\haT{h}_n$ baSeD On $N$ Observations. An equivAlEnt way To detErmine the loweR bound $\deltA_N$ IS to choosE the LArGEst possible intEger $m$ Such that $$\deLTa_n =n^{-1}\sum_{k=1}^M \lambDa^{-1}_k,\quad \sIgma_0^2n^{-1}\sum_{k=1}^M [\PSi (\lambda_K)]^{-2}\laMbdA

we require Gaussianity. N evertheles s, th e p roo fwork s th e same for oth e r er ror densities, but bou nding t h e Ku l lb ack-L eiblero rH e lli ng er di st a nc e bet wee n alter natives mi ght b e more cumbe r so me. Let u s a lso mentiontha t thepr oof strat egy canalso y i eld alower bou nd for co n vergenc e in pro bability: $$\inf_ { \h a t h_n}\sup_{h\ in H^r _R } P_ h \ Big (\d elta_n^{-1 }{ \lVer t \hat h _ n- h \ rVe r t}_X^2{\geqsl ant}\tfrac{ \ sig ma_0^2 }{ 4\e x p(4M)} \Big) {\ g eqs lant}c>0,\q uad\text{wit h}~\de l ta_n ~\ t ext{giv en inThe ore m \r e f{ th m-l b} } ,$$ cf . K o ros telev an dTs ybako v (1 9 9 3 ) . N ote tha t Ass umption 2.2 i s a utom a tic allysatis fied u nderthe ge neral s ource condition wit h $K^\ast K={\v ar phi}( B ^{-2}) $.Fol lowingthe pro o f o fT h e or em \[thm-lb\], weim m e di ately ob tain:\[ so u rce-lb\] S upp oset h at As sump t io n \[h-sm ooth\] is s atisfie dwith $ h\ in{{\ maths c r}H} _{sour ce}$ and $u_k = e_k$, $\nu_k ^ {-r} = \psi ( \ la m b da _ k)$for all $k{\ge qsla n t}1$ . Le t $ {\v a rphi} (B^{- 2} ) =K ^ \ast K$. Then, forNP IR mod el an d for NPIV mo del (under A s sumption \[a s s- c omplexity\]),we ha ve the sam e minimax risk lower b ound: $$\ i n f_{\hath_n }\s up_ {h\ i n { {\mathscr}H}_ { s ourc e} }\opera tor name{{\ mat hbb E} }_h [{ \lVert \h at h_n-h \ rV er t} _X^ 2]{\g e qslant}\ tf rac {\ sig ma_0^ 2 }{4\ex p(4M) } \d el ta _ n,\ quad \t e xt { w ith} ~\ de lta_ n ~ \t ext{g iven inTheorem \ref{thm -lb } },$$ w he re theinfimum runsov er all pos si ble estim a t ors $\ha t{h}_n$ based on $n$ ob s ervatio ns. An e quiv alent way to deter min e the l ower b ound$\ del t a _n$ i s to ch oo se the lar g e stpossi bl e in teger $ m$ such that $$\de l ta_ n =n^{-1}\sum _{k =1}^ m \l amb d a^ { -1} _k , \qu a d \sigma_0^2n^{- 1}\sum_{k= 1} ^ m[\psi (\la m bda _k )]^{-2} \lambda

we_require Gaussianity._Nevertheless, the proof works_the same_for_other error_densities,_but bounding the_Kullback-Leibler or Hellinger_distance between alternatives might_be more cumbersome. Let_us_also mention that the proof strategy can also yield a lower bound for convergence_in_probability: $$\inf_{\hat_h_n}\sup_{h\in_H^r_R}P_h\Big(\delta_n^{-1}{\lVert_\hat h_n-h \rVert}_X^2{\geqslant}\tfrac{\sigma_0^2}{4\exp(4M)}\Big){\geqslant}c>0,\quad \text{with}~\delta_n ~\text{given in Theorem_\ref{thm-lb}},$$ cf. Korostelev and Tsybakov_(1993). Note that_Assumption 2.2 is automatically satisfied under the general_source_condition with $K^\ast_K = {\varphi}(B^{-2})$. Following the proof of Theorem \[thm-lb\],_we immediately obtain: \[source-lb\] Suppose that Assumption_\[h-smooth\] is satisfied_with_$h\in_{{\mathscr}H}_{source}$ and $u_k =_e_k$, $\nu_k^{-r} = \psi (\lambda_k)$ for_all $k{\geqslant}1$. Let ${\varphi}(B^{-2})=K^\ast K$. Then,_for NPIR model and for NPIV model_(under Assumption \[ass-complexity\]), we have the_same minimax risk lower bound:_$$\inf_{\hat h_n}\sup_{h\in_{{\mathscr}H}_{source}}\operatorname{{\mathbb E}}_h[{\lVert \hat h_n-h \rVert}_X^2]{\geqslant}\tfrac{\sigma_0^2}{4\exp(4M)} \delta_n,\quad \text{with}~\delta_n_~\text{given in Theorem_\ref{thm-lb}},$$ where_the infimum runs_over all possible estimators $\hat{h}_n$ based_on $n$ observations._An equivalent way to determine the_lower_bound $\delta_n$ is_to_choose_the largest_possible integer $m$_such_that $$\delta_n_=n^{-1}\sum_{k=1}^m_\lambda^{-1}_k,\quad \sigma_0^2n^{-1}\sum_{k=1}^m [\psi (\lambda_k)]^{-2}\lambda

_{2}\left( P^{X}\right) $ if the underlying partition $\mathcal{P} $ of $\mathcal{X}$ is lower-regular in the sense that there exists a constant $c_{m}$ such that $$0<c_{m}<\sqrt{\left\vert \mathcal{P}\right\vert \inf_{I\in \mathcal{P}}P^{X}\left( I\right) }\text{ }.$$More precisely, if $r\in \mathbb{N}$ is the maximal degree of the piecewise polynomials—$r=0$ in the case of histograms—then any orthonormal basis $\left\{ \varphi _{I,j},\text{ }I\in \mathcal{P},\text{ }j\in \left\{ 0,\ldots,r\right\} \right\} $ of $\left( m,\left\Vert \cdot\right\Vert _{2}\right) $ such that for all $j\in \left\{ 0,\ldots,r\right\} $, $\varphi _{I,j}$ is supported by the element $I$ of $\mathcal{P}$, is localized. Hence, by Proposition \[prop\_strong\_loc\_loc\], it achieves Inequalities (\[def\_Ai\]) and (\[loc\_plus\]) of the definition of strongly localized basis with $b=1$ and $A_{1}=D_{m}=(r+1)$Card$\left( \mathcal{P}\right) $. It is also immediately seen that such basis achieves in this case (\[def\_card\_Pi\_i\_k\]) with $A_{c}=r+1$. Furthermore, $r_m=c_m^{-1}$ for histograms (Lemma 4, [@saum:12]) and it has a more complicated expression for piecewise polynomials (Lemma 7, [@saum:12]). As a result, models of histograms and piecewise polynomials with bounded degree and underlying lower-regular partition $\mathcal{P}$ of $\mathcal{X}$ are endowed with a strongly localized structure. ### Compactly supported wavelet expansions\[section\_Haar\] We assume here that $\mathcal{X=}\left[ 0,1\right] $ and take $b_m\in \mathbb{N}^\star$. For details about wavelets and interactions with Statistics, we refer to [@HKPT:98]. Set $\phi _{0}$ the father

_ { 2}\left (P^{X}\right) $ if the underlying partition $ \mathcal{P } $ of $ \mathcal{X}$ is lower - regular in the sense that there exist a changeless $ c_{m}$ such that $ $ 0 < c_{m}<\sqrt{\left\vert \mathcal{P}\right\vert \inf_{I\in \mathcal{P}}P^{X}\left (I\right) } \text { } .$$More precisely, if $ r\in \mathbb{N}$ is the maximal degree of the piecewise polynomials—$r=0 $   in the case of histograms — then any orthonormal footing $ \left\ { \varphi _ { I, j},\text { } I\in \mathcal{P},\text { } j\in \left\ { 0,\ldots, r\right\ } \right\ } $ of $ \left (m,\left\Vert \cdot\right\Vert _ { 2}\right) $ such that for all $ j\in \left\ { 0,\ldots, r\right\ } $, $ \varphi _ { I, j}$ is supported by the component $ I$ of $ \mathcal{P}$, is localized. Hence, by Proposition \[prop\_strong\_loc\_loc\ ], it achieve Inequalities (\[def\_Ai\ ]) and (\[loc\_plus\ ]) of the definition of powerfully localized basis with $ b=1 $ and $ A_{1}=D_{m}=(r+1)$Card$\left ( \mathcal{P}\right) $. It is besides immediately seen that such basis achieve in this case (\[def\_card\_Pi\_i\_k\ ]) with $ A_{c}=r+1$. Furthermore, $ r_m = c_m^{-1}$ for histograms (Lemma 4, [ @saum:12 ]) and it has a more complicated expression for piecewise polynomials (Lemma 7, [ @saum:12 ]). As a result, models of histogram and piecewise polynomials with bounded degree and underlie lower - regular division $ \mathcal{P}$ of $ \mathcal{X}$ are endowed with a powerfully localized structure. # # # Compactly supported wavelet expansions\[section\_Haar\ ] We assume here that $ \mathcal{X=}\left [ 0,1\right ] $ and lease $ b_m\in \mathbb{N}^\star$. For details about wavelets and interactions with Statistics, we refer to [ @HKPT:98 ]. Set $ \phi _ { 0}$ the founder

_{2}\lefh( P^{X}\right) $ if the underlning partition $\mcrhcal{P} $ of $\mafhcal{X}$ ir lower-regular in the sense vhat thert exists a constant $c_{m}$ such nhat $$0<c_{m}<\sqet{\leht\vert \mathcal{P}\rmfht\vert \inf_{I\ih \matkcel{P}}P^{X}\left( I\righj) }\text{ }.$$More [recisely, if $r\hn \mcthbb{N}$ is the maximal degree of the [iecewixe polynomials—$r=0$ ig tht cwse kf histograms—then any orthonormal gasis $\ltft\{ \varphi _{I,j},\text{ }O\in \mathcal{P},\text{ }j\in \left\{ 0,\lfots,g\right\} \right\} $ of $\lfft( m,\left\Vetf \cqit\right\Vert _{2}\fight) $ such that for aml $j\in \left\{ 0,\ldots,r\right\} $, $\varphi _{I,j}$ ix supportee vy hve element $M$ of $\mwthcal{P}$, is localized. Hence, ny Proposition \[prmp\_srrong\_loc\_loc\], it achievxs Inequalities (\[def\_Ay\]) and (\[loc\_[lbs\]) of the definition if strotgly locxoizdd gaxia with $b=1$ end $A_{1}=D_{m}=(r+1)$Cars$\left( \mathcql{P}\right) $. It is alsp ynmediately sesn thae fuch basis achieves in this case (\[def\_carg\_Pi\_j\_k\]) with $A_{c}=r+1$. Furthermore, $r_m=c_m^{-1}$ for histograms (Pemma 4, [@satm:12]) and it has a more complicated expression for phecewmsd pilnnomkqld (Lemma 7, [@saum:12]). As a result, models of histogramf amd piecewise polnnomials with bouncef cggree and undetlying lowsr-regular partitioj $\mathcwl{P}$ od $\mathcal{V}$ arr endowed with a strongly licalized strlctuee. ### Compactly suppoxted wavelet expsnsioms\[section\_Haar\] We assume kere tgat $\mathcal{D=}\left[ 0,1\riggg] $ and take $b_m\in \manhbb{T}^\star$. For details about wadelets anv intzractionr wijh Statystics, we gefer to [@HKPT:98]. Set $\phi _{0}$ tje fajher

_{2}\left( P^{X}\right) $ if the underlying partition of is lower-regular the sense that such $$0<c_{m}<\sqrt{\left\vert \mathcal{P}\right\vert \inf_{I\in I\right) }\text{ }.$$More if $r\in \mathbb{N}$ is the maximal of the piecewise polynomials—$r=0$ in the case of histograms—then any orthonormal basis $\left\{ _{I,j},\text{ }I\in \mathcal{P},\text{ }j\in \left\{ 0,\ldots,r\right\} \right\} $ of $\left( m,\left\Vert \cdot\right\Vert _{2}\right) such for $j\in 0,\ldots,r\right\} $, $\varphi _{I,j}$ is supported by the element $I$ of $\mathcal{P}$, is localized. Hence, by \[prop\_strong\_loc\_loc\], it achieves Inequalities (\[def\_Ai\]) and (\[loc\_plus\]) of definition of strongly localized with $b=1$ and $A_{1}=D_{m}=(r+1)$Card$\left( \mathcal{P}\right) It also immediately that basis in this case with $A_{c}=r+1$. Furthermore, $r_m=c_m^{-1}$ for histograms (Lemma 4, [@saum:12]) and it has a more complicated expression for polynomials (Lemma As a models histograms piecewise polynomials with and underlying lower-regular partition $\mathcal{P}$ of with a strongly localized structure. ### Compactly supported expansions\[section\_Haar\] We here that $\mathcal{X=}\left[ 0,1\right] $ and $b_m\in \mathbb{N}^\star$. For details about wavelets and interactions Statistics, we refer to [@HKPT:98]. Set $\phi _{0}$ the father

_{2}\left( P^{X}\right) $ if the underlyinG partition $\MathcAl{P} $ Of $\mAtHcal{x}$ is lOwer-regular in tHE senSe that there exists a consTant $c_{M}$ sUCh thAT $$0<c_{M}<\sqrt{\Left\verT \MaTHCal{p}\rIgHt\vErT \InF_{I\in \mAthCal{P}}P^{X}\lEft( I\right) }\tExt{ }.$$moRe precisely, iF $R\iN \mathbb{N}$ is The Maximal degreE of The pieCeWisE PolynOmiAls—$r=0$ iN the caSE of hisTograms—thEn ANy orthONormal bASIs $\Left\{ \Varphi _{I,j},\text{ }I\in \maTHcAL{P},\text{ }j\in \left\{ 0,\lDots,r\rIgHT\} \rIGHt\} $ oF $\leFt( m,\left\VerT \cDot\riGHt\Vert _{2}\rIGhT) $ SUCh tHAt for all $j\in \leFt\{ 0,\ldots,r\rigHT\} $, $\vaRphi _{I,j}$ Is SupPOrted bY the eLeMEnt $i$ of $\mathcal{P}$, Is loCalized. HeNce, by PROpositiON \[prop\_stRong\_loC\_loC\], it AchiEVeS INeqUaLItiES (\[dEf\_AI\]) And (\[Loc\_plus\]) oF tHe DefinItioN OF STronGly LocaLized Basis with $b=1$ and $a_{1}=D_{m}=(R+1)$CarD$\LefT( \mathCal{P}\rIght) $. it Is alsO immedIatelY sEen that such basiS achIeves in thIs cAsE (\[deF\_cArd\_Pi\_I\_K\]) with $A_{C}=r+1$. FUrtHermore, $R_m=c_m^{-1}$ for HIstOgRAMS (LEmma 4, [@saum:12]) and it has a mOrE COmPlicated ExpresSIoN fOR piecewiSe PolYnomIALs (LemMa 7, [@saUM:12]). AS a result, Models OF hIsTograms AnD pieceWiSe pOlyNomiaLS witH boundEd degree And unDErlying lower-reGUlar partition $\MAtHCAl{p}$ Of $\maThcAl{X}$ are endowEd wiTH a stRongLY lOcaLIzed sTructUrE. ### coMPactly supported waveLeT expanSions\[Section\_Haar\] We Assume here THAT $\mathcal{x=}\lefT[ 0,1\RiGHt] $ and take $b_m\in \mAthbb{n}^\star$. For deTAils abouT waveLets and iNteractioNS With StatIstIcs, We rEfeR TO [@HkPT:98]. Set $\phi _{0}$ the fATHer

_{2}\left( P^{X}\right) $if the und erlyi ngpar ti tion $\m athcal{P} $ of $\ma thcal{X}$ is lower-reg ularin thes en se th at ther e e x i sts a c ons ta n t$c_{m }$such th at $$0<c_{ m}< \s qrt{\left\ve r t\mathcal{P }\r ight\vert \i nf_ {I\in\m ath c al{P} }P^ {X}\l eft( I \ right) }\text{}. $ $Morep recisel y , i f $r \in \mathbb{N}$ i s t h e maximal degr ee ofth e p i e cew ise polynomia ls —$r=0 $  in the ca s e ofh istograms—the n any ortho n orm al bas is $\ l eft\{\varp hi _{I ,j},\text{}I\i n \mathca l{P},\ t ext{ }j \ in \lef t\{ 0, \ld ots ,r\r i gh t\ } \ ri g ht\ } $ of $\l eft( m,\ le ft \Vert \c d o t \ righ t\V ert_{2}\ right) $ such th at f o r a ll $j \in \ left \{ 0,\l dots,r \righ t\ } $, $\varphi _ {I,j }$ is sup por te d b ythe e l ement$I$ of $\math cal{P}$ , is l o c a li zed. Hence, by Pro po s i ti on \[pro p\_str o ng \_ l oc\_loc\ ], it ach i e ves I nequ a li ties (\[ def\_A i \] )and (\[ lo c\_plu s\ ])ofthe d e fini tion o f strong ly lo c alized basis w i th $b=1$ and$ A_ { 1 }= D _{m} =(r +1)$Card$\l eft( \ma thca l {P }\r i ght)$. It i s a l so immediately seen t hat su ch ba sis achievesin this ca s e (\[def\_ card \ _P i \_i\_k\]) with $A_{ c}=r+1$. F u rthermor e, $r _m=c_m^{ -1}$ forh i stograms (L emm a 4 , [ @ s au m:12]) and it h as a m ore com pli cated e xpr ess ion fo rpiecewise polynom ia ls ( Le mma 7, [ @ saum:12] ). As a re sult, models of h isto gr am s an d piece w is e poly no mi alswit hbound ed d e gre e and u nderlying lo w er-r eg ul ar part ition $\mathc al {P}$ of $\ ma thc al{X}$ a re endow ed with a strongly loca l ized st ruc ture. ## # Compact lysuppor ted wavele t expa nsion s\ [se c t ion\_ H a ar \] W e assume h e r e t hat $ \m athc al{X=}\ left[ 0,1\right] $ and take $b_m\in \m athb b { N} ^\s t ar $ . F or det a i ls about wavele ts and int er a ct ions withS tat is tics, w e refer to [ @ HKPT:98 ]. Set $\ phi _{0}$ t he f a t her

_{2}\left( P^{X}\right)_$ if_the underlying partition $\mathcal{P} $_of $\mathcal{X}$_is_lower-regular in_the_sense that there_exists a constant_$c_{m}$ such that $$0<c_{m}<\sqrt{\left\vert_\mathcal{P}\right\vert \inf_{I\in \mathcal{P}}P^{X}\left(_I\right)_}\text{ }.$$More precisely, if $r\in \mathbb{N}$ is the maximal degree of the piecewise polynomials—$r=0$ in_the_case of_histograms—then_any_orthonormal basis $\left\{ \varphi _{I,j},\text{_}I\in \mathcal{P},\text{ }j\in \left\{ 0,\ldots,r\right\} \right\}_$ of_$\left( m,\left\Vert \cdot\right\Vert _{2}\right) $ such that for_all_$j\in \left\{ 0,\ldots,r\right\}_$, $\varphi _{I,j}$ is supported by the element $I$_of $\mathcal{P}$, is localized. Hence, by_Proposition \[prop\_strong\_loc\_loc\], it_achieves_Inequalities_(\[def\_Ai\]) and (\[loc\_plus\]) of_the definition of strongly localized basis_with $b=1$ and $A_{1}=D_{m}=(r+1)$Card$\left( \mathcal{P}\right) $._It is also immediately seen that such_basis achieves in this case (\[def\_card\_Pi\_i\_k\])_with $A_{c}=r+1$. Furthermore, $r_m=c_m^{-1}$ for_histograms (Lemma_4, [@saum:12]) and it has_a more complicated_expression for_piecewise polynomials (Lemma_7, [@saum:12]). As a result, models_of histograms and_piecewise polynomials with bounded degree and_underlying_lower-regular partition $\mathcal{P}$_of_$\mathcal{X}$_are endowed_with a strongly_localized_structure. ### Compactly_supported_wavelet expansions\[section\_Haar\] We assume here that $\mathcal{X=}\left[_0,1\right]_$ and take $b_m\in \mathbb{N}^\star$. For details_about wavelets and interactions_with_Statistics, we refer to_[@HKPT:98]. Set $\phi _{0}$ the_father

= 7 \pm 2$ M$_\odot$ and $E_K = (35 \pm 10) \times 10^{51}$ erg. Our errors on these physical quantities reflect conservatively the empirical nature of the method and its uncertainties (for instance, the fact that the phases at which the measured photospheric velocities are compared are never identical for the target and the template SN). The progenitor of SN 2013dx is thus probably $25-30 M_{\odot}$, which is $15- 20\%$ less massive than that estimated for SN 1998bw ($30-35 M_{\odot}$, Maeda et al. 2006) Field of SN2013dx ----------------- The field of SN 2013dx shows a bright galaxy southward of the transient, at $\sim 8"$ from the SN position (see Fig. 8). This could have been the SN host, since the faint source ($R=23.01$) at $0.6"$ from SN 2013dx was initially misclassified as a star in the SDSS (Singer et al. 2013). For this reason, Leloudas et al. (2013) obtained a spectrum of this galaxy, reporting a $z=0.145$, which is the same as that of the transient (Mulchaey et al. 2013a,b; D’Avanzo et al. 2013d). Kelly et al. (2013) performed an extensive study of this galaxy and its outskirt, concluding that the SN 2013dx host galaxy is a dwarf satellite of the bright one. Many more galaxies are present in the SN 2013dx field. Kelly et al. 2013 noted that many of them have an SDSS photometric redshift compatible with that of the host and the bright galaxy. In addition, they also pointed out that the SDSS spctroscopic galaxy survey targeted five galaxies brighter than $17.7$ within 15’ from SN 2013dx, their redshift being close to $z=0.145$. We decided to use different slit position angles during our VLT spectroscopic campaign. This allowed us to study both the spectroscopic evolution of the SN and to determine the redshift of more nearby galaxies. The different position angles adopted are described in Table 5, and the corresponding slit orientations are displayed in Fig. 8. This figure also marks with numbers the $14$ galaxies that we placed in the slit in our observations (“S” marks a star). We determine the redshift of all these galaxies, although for three of them the determination is not completely

= 7 \pm 2 $ M$_\odot$ and $ E_K = (35 \pm 10) \times 10^{51}$ erg. Our errors on these physical quantities reflect conservatively the empiric nature of the method acting and its doubt (for instance, the fact that the phase at which the deliberate photospheric velocities are compared are never identical for the aim and the template SN). The progenitor of SN 2013dx is therefore probably $ 25 - 30 M_{\odot}$, which is $ 15- 20\%$ less massive than that estimated for SN 1998bw ($ 30 - 35 M_{\odot}$, Maeda et al. 2006) Field of SN2013dx ----------------- The field of SN 2013dx shows a bright galax southward of the transient, at $ \sim 8"$ from the SN position (see Fig. 8). This could have been the SN host, since the dim source ($ R=23.01 $) at $ 0.6"$ from SN 2013dx was initially misclassified as a star in the SDSS (Singer et al. 2013). For this rationality, Leloudas et al. (2013) obtained a spectrum of this galaxy, report a $ z=0.145 $, which is the same as that of the transient (Mulchaey et al. 2013a, b; D’Avanzo et al. 2013d). Kelly et al. (2013) performed an extensive sketch of this galaxy and its outskirt, concluding that the SN 2013dx host galaxy is a gnome satellite of the bright one. Many more galax are present in the SN 2013dx field. Kelly et al. 2013 noted that many of them have an SDSS photometric red shift compatible with that of the host and the bright galaxy. In addition, they also pointed out that the SDSS spctroscopic galaxy survey targeted five galaxies brighter than $ 17.7 $ within 15 ’ from SN 2013dx, their redshift being close to $ z=0.145$. We decided to use different slit position angle during our VLT spectroscopic campaign. This allowed us to learn both the spectroscopic development of the SN and to settle the redshift of more nearby galaxies. The different position angles adopted are described in Table 5, and the correspond slit orientations are displayed in Fig. 8. This figure also marks with numbers the $ 14 $ galax that we placed in the slit in our observations (“ S ” marks a asterisk). We determine the red shift of all these galaxy, although for three of them the determination is not completely

= 7 \om 2$ M$_\odot$ and $E_K = (35 \pm 10) \tlmes 10^{51}$ erg. Our ertoes on vhese pgysical duantities reflect conservatmvelt the empirical nature of tfe method and its unctrtainties (for instance, tmz facf thac vhe phases at wmich the medsured photospvefie velocities are compared are never ydenticsl for the targej and ehe fvmklate SN). The progenitor of SN 2013dx js thus probably $25-30 M_{\ocot}$, which is $15- 20\%$ less massivf thwn that estimated vor SN 1998bw ($30-35 N_{\odoe}$, Maeda et al. 2006) Field of SN2013dx ----------------- The fiels of SN 2013dx shows a bright galaxh souchward of tye trwtsient, at $\smm 8"$ frjm the SN position (sae Fig. 8). This could haye bexn tye SN host, since the haint source ($R=23.01$) at $0.6"$ ftom SN 2013dx faa initially misclqswifieg as a sgqr kn uhe SSSS (Sijgec et al. 2013). Fod this reasin, Leloudas et al. (2013) pbeqined a spectdum of tris galaxy, reporting a $z=0.145$, which is the sdme as that of the transiebt (Mulchaey et al. 2013a,b; F’Avanzo ee al. 2013d). Kelly et al. (2013) performed an extensive study mf thms gaoawy ave lts outskirt, concluding that the SN 2013dx host gwmacy is a dwarf sabellite of the bribhh pge. Many more gxlaxies ade present in the DN 2013dx fyeld. Jelly et wl. 2013 moted that many of them havw an SDSS phjrometric redshift eompatible wnth thst of the host and the brighc galasy. In additlon, they zuso pointed out ghan tha SDSS skztroscopic galaxy survey tergetzd five ealacies bwighter thwn $17.7$ wldhin 15’ from SN 2013dx, tjeir tedshixt being cpose to $z=0.145$. We decided to use diffxcent slit posothon angles buring our VLT speceroscopic campcign. Thif alluwed us to study uoth the spestroscopic evmpution of thx SN and eo dwternine thd redshift of mpre nearbj yalaxies. Rhe different posibion xhgles adopted axt dtsxribed in Tablr 5, xnd tje cjsresponding vlit orkrntatkons art dlspuayec in Fig. 8. This figura alao marks with numbrrf the $14$ gqlaxies ehat we placec in the slit in okr obvertationx (“S” marks a star). We determine the dedshift lf wll these gajaxics, ajthough fox three of them the determination is not completely

= 7 \pm 2$ M$_\odot$ and $E_K \pm \times 10^{51}$ Our errors on the nature of the and its uncertainties instance, the fact that the phases which the measured photospheric velocities are compared are never identical for the target the template SN). The progenitor of SN 2013dx is thus probably $25-30 M_{\odot}$, is 20\%$ massive that estimated for SN 1998bw ($30-35 M_{\odot}$, Maeda et al. 2006) Field of SN2013dx ----------------- The of SN 2013dx shows a bright galaxy southward the transient, at $\sim from the SN position (see 8). could have the host, the faint source at $0.6"$ from SN 2013dx was initially misclassified as a star in the SDSS (Singer et al. For this et al. obtained spectrum this galaxy, reporting which is the same as that (Mulchaey et al. 2013a,b; D’Avanzo et al. 2013d). et al. performed an extensive study of this and its outskirt, concluding that the SN 2013dx galaxy is a dwarf satellite of the bright one. Many more galaxies are present in 2013dx field. Kelly et 2013 noted that of have SDSS redshift compatible that of the host and the bright galaxy. In addition, they pointed out that the SDSS spctroscopic galaxy survey targeted five than within 15’ from 2013dx, their redshift being to We decided to use position during campaign. allowed to study both the evolution of the SN and determine the redshift of position angles adopted are described in Table 5, the corresponding slit orientations are displayed in 8. This figure also marks with numbers the $14$ galaxies that we in the our observations (“S” marks a star). We determine redshift of all these although for three of them the determination is not

= 7 \pm 2$ M$_\odot$ and $E_K = (35 \pm 10) \times 10^{51}$ erg. Our Errors on thEse phYsiCal QuAntiTies Reflect conservATiveLy the empirical nature of The meThOD and ITs UncerTaintieS (FoR INstAnCe, The FaCT tHat thE phAses at wHich the meaSurEd Photospheric VElOcities are ComPared are neveR idEnticaL fOr tHE targEt aNd the TemplaTE SN). The ProgenitoR oF sN 2013dx is THus probABLy $25-30 m_{\odoT}$, which is $15- 20\%$ less massiVE tHAn that estimateD for SN 1998Bw ($30-35 m_{\OdOT}$, maeDa eT al. 2006) Field of sN2013Dx ----------------- The FIeld of Sn 2013Dx SHOWs a BRight galaxy soUthward of thE TraNsient, At $\Sim 8"$ FRom the sN posItIOn (sEe Fig. 8). This coUld hAve been thE SN hosT, Since thE Faint soUrce ($R=23.01$) aT $0.6"$ frOm Sn 2013dx wAS iNiTiaLlY MisCLaSsiFIed As a star iN tHe sDSS (SIngeR ET AL. 2013). For ThiS reaSon, LeLoudas et al. (2013) obtAinEd a sPEctRum of This gAlaxY, rEportIng a $z=0.145$, wHich iS tHe same as that of tHe trAnsient (MuLchAeY et Al. 2013A,b; D’AvANzo et aL. 2013d). KEllY et al. (2013) peRformed AN exTeNSIVe Study of this galaxy aNd ITS oUtskirt, cOncludINg ThAT the SN 2013dx HoSt gAlaxY IS a dwaRf saTElLite of thE brighT OnE. MAny more GaLaxies ArE prEseNt in tHE SN 2013dX field. kelly et aL. 2013 noteD That many of them HAve an SDSS photOMeTRIc REdshIft Compatible wIth tHAt of The hOSt And THe briGht gaLaXY. IN Addition, they also poiNtEd out tHat thE SDSS spctroscOpic galaxy SURVey targeTed fIVe GAlaxies brighteR than $17.7$ Within 15’ from sn 2013dx, their RedshIft being Close to $z=0.145$. WE DEcided to Use DifFerEnt SLIt Position angleS DUrinG oUr VLT spEctRoscopiC caMpaIgn. thiS aLlowed us tO study boTh ThE sPeCtrOscopIC evolutiOn Of tHe sN aNd to dETerminE the rEdshIfT oF MorE nearby GAlAXIes. THe DiFferEnt PoSitioN angLEs aDopted aRe describEd iN tablE 5, aNd The corrEsponding slit OrIentations ArE diSplayeD IN Fig. 8. This Figure also marks with numbERs the $14$ gaLaxIes thAt we Placed in tHe sLit in oUr oBServatIons (“S” mArks a StAr). WE DEtermINE tHe rEdShift of all THEse GalaxIeS, altHough foR three of them the detERmiNation is not coMplEtelY

= 7 \pm 2$ M$_\odot$ and$E_K = (35 \pm10) \t im es 1 0^{5 1}$ erg. Our e r rors on these physical qua ntiti es refl e ct cons ervativ e ly t heem pi ric al na tureofthe met hod and it s u nc ertainties ( f or instance, th e fact thatthe phase satw hichthe meas ured p h otosph eric velo ci t ies ar e compar e d a re n ever identical fo r t h e target and t he tem pl a te S N). Th e progenit or of S N 2013dx is t h usp robably $25-3 0 M_{\odot} $ , w hich i s$15 - 20\%$ less m a ssi ve than tha t es timated f or SN1 998bw ( $ 30-35 M _{\odo t}$ , M aeda et a l.20 0 6)Fi eld ofSN2013dx - -- ----- ---- - - - - - T hefiel d ofSN 2013dx sho wsa br i ght gala xy so uthw ar d ofthe tr ansie nt , at $\sim 8"$from the SN p osi ti on(s ee Fi g . 8).Thi s c ould ha ve been the S N h os t, since the faint s o u rc e ($R=23 .01$)a t$0 . 6"$ from S N 2 013d x was i niti a ll y miscla ssifie d a sa starin the S DS S ( Sin ger e t al. 2013) . For th is re a son, Leloudase t al. (2013)o bt a i ne d a s pec trum of thi s ga l axy, rep o rt ing a $z= 0.145 $, wh i ch is the same as t ha t of t he tr ansient (Mulc haey et al . 2 013a,b;D’Av a nz o et al. 2013d) . Kel ly et al.( 2013) pe rform ed an ex tensive s t u dy of th isgal axy an d it s outskirt, c o n clud in g thatthe SN 201 3dx ho stgal ax y is a dw arf sate ll it eof th e bri g ht one. M any m ore gala x ies ar e pre sent i nt heSN 2013 d xf i eld. K el ly e t a l. 2013 not e d t hat man y of them ha v e an S DS S photo metric redshi ft compatibl ewit h that o f the ho st and the bright galax y . In ad dit ion,they also poi nte d outtha t the S DSS sp ctros co pic g alaxy s ur vey t argeted fi v e ga laxie sbrig hter th an $17.7$ within 1 5 ’ f rom SN 2013dx , t heir r ed shi f tb ein gc los e to $z=0.145$. We decided t o u se differe n t s li t posit ion ang les d u ring ou r VLT spe ctroscopi ccamp a i gn. This allo wed us t o study b o th th e s pectr osc opic e vo lut ion o f theS N a nd to deter mi ne the reds hi ft of mo re nearby galaxies. The diffe rentpos ition ang les ado pted aredesc ribed in T abl e 5 , and th e corr espo n di ngs lit o rien t ations ar e d isp l a ye d in Fig. 8 . T his figu rea lso ma rkswith numbers the$ 14$ galaxies t hatw e pl ace d inth e slit in ourobs er v a tions (“ S” marks a st ar). We d e termi ne the redsh ift ofa l lt hese g alax ies , althoug h f or three o fth e m thedete rm inatio n is n o t co m p letely

=_7 \pm_2$ M$_\odot$ and $E_K_= (35_\pm_10) \times_10^{51}$_erg. Our errors_on these physical_quantities reflect conservatively the_empirical nature of_the_method and its uncertainties (for instance, the fact that the phases at which the_measured_photospheric velocities_are_compared_are never identical for the_target and the template SN)._The progenitor_of SN 2013dx is thus probably $25-30 M_{\odot}$,_which_is $15- 20\%$_less massive than that estimated for SN 1998bw ($30-35_M_{\odot}$, Maeda et al. 2006) Field of_SN2013dx ----------------- The field of_SN_2013dx_shows a bright galaxy_southward of the transient, at $\sim_8"$ from the SN position (see_Fig. 8). This could have been the_SN host, since the faint source_($R=23.01$) at $0.6"$ from SN_2013dx was_initially misclassified as a star_in the SDSS_(Singer et_al. 2013). For_this reason, Leloudas et al. (2013)_obtained a spectrum_of this galaxy, reporting a $z=0.145$,_which_is the same_as_that_of the_transient (Mulchaey et_al._2013a,b; D’Avanzo_et_al. 2013d). Kelly et al. (2013)_performed_an extensive study of this galaxy and_its outskirt, concluding that_the_SN 2013dx host galaxy_is a dwarf satellite of_the bright one. Many more galaxies are_present in_the SN_2013dx field. Kelly et al. 2013 noted that many of them_have an SDSS photometric redshift compatible_with that of the_host and_the_bright galaxy. In_addition,_they also_pointed out that the SDSS spctroscopic galaxy_survey targeted_five galaxies brighter than $17.7$ within_15’ from SN 2013dx,_their_redshift being close to $z=0.145$. We decided_to use different slit position angles_during our VLT spectroscopic campaign._This_allowed_us to study both the_spectroscopic evolution of the SN and_to determine the_redshift of more nearby galaxies. The different_position_angles adopted are described in Table_5,_and the corresponding slit orientations are_displayed_in_Fig. 8. This figure also_marks with numbers the $14$ galaxies_that we placed in the slit in our observations_(“S” marks a_star). We determine the redshift of_all_these_galaxies, although for three of them the determination is not_completely

with blades or skates. There are thus multiple applications in the context of wheeled motion, space or mobile robotics and robotic manipulation. In this paper, we will introduce some new geometric techniques in nonholonomic mechanics to study the case of force minimizing optimal control problems. The application of modern tools from differential geometry in the fields of mechanics, control theory, field theory and numerical integration has led to significant progress in these research areas. For instance, the study of the geometrical formulation of the nonholonomic equations of motion has led to better understanding of different engineering problems such locomotion generation, controllability, motion planning, and trajectory tracking (see e.g. [@Bl], [@bloch2], [@blochcrouch], [@blochcrouch2], [@blochcrouch3], [@borisovmamaev], [@bullolewis], [@kellymurray], [@jair], [@LR1], [@LMdD], [@lewisandrew], [@murraysastry], [@ostrowoski] and references therein). Geometric techniques can also be used to study optimal control problems (see [@blochcrouch3], [@cortes; @noho], [@CoMa], [@Vu], [@Vu2], [@Su], [@SuJu]). Combining these ideas in this paper, we study the underlying geometry of optimal control problems for mechanical systems subject to nonholonomic constraints and we apply it to several interesting examples. Classical nonholonomic constraints which are linear in the velocities can be geometrically encoded by a constant rank distribution $\mathcal{D}$. As we will see, the distribution $\mathcal{D}$ will play the role of the velocity phase space. Given a mechanical Lagrangian $L=K-V: TQ\rightarrow {\mathbb R}$ where $K$ and $V$ are the kinetic and potential energy, respectively. and the distribution ${\mathcal D}$, the dynamics of the nonholonomic system is completely determined using the Lagrange-d’Alembert principle [@Bl]. In this paper we will formulate a description in terms of a Levi Civita connection defined on the space of vector fields taking values on ${\mathcal D}$. This connection is obtained by projecting the standard Lie bracket using the Riemannian metric associated with the kinetic energy $K$ (see [@maria2]) and the typical characterization of the Levi-Civita connection (see also [@BlochHussein]). By adding controls in this setting we can study optimal control

with blades or skates. There are thus multiple application in the context of wheeled apparent motion, space or mobile robotics and robotic handling. In this paper, we will introduce some new geometric technique in nonholonomic mechanics to study the event of force minimizing optimum control problems. The application of modern creature from differential geometry in the fields of mechanics, control theory, field hypothesis and numerical integration has led to meaning progress in these research areas. For instance, the study of the geometric formulation of the nonholonomic equations of motion has led to better reason of different engineering problems such locomotion generation, controllability, motion planning, and trajectory tracking (see e.g. [ @Bl ], [ @bloch2 ], [ @blochcrouch ], [ @blochcrouch2 ], [ @blochcrouch3 ], [ @borisovmamaev ], [ @bullolewis ], [ @kellymurray ], [ @jair ], [ @LR1 ], [ @LMdD ], [ @lewisandrew ], [ @murraysastry ], [ @ostrowoski ] and references therein). Geometric techniques can also be used to study optimal control problems (see [ @blochcrouch3 ], [ @cortes; @noho ], [ @CoMa ], [ @Vu ], [ @Vu2 ], [ @Su ], [ @SuJu ]). unite these ideas in this paper, we learn the underlie geometry of optimal control problems for mechanical system subject to nonholonomic constraints and we apply it to several interesting exercise. Classical nonholonomic constraints which are linear in the velocities can be geometrically encoded by a ceaseless rank distribution $ \mathcal{D}$. As we will see, the distribution $ \mathcal{D}$ will play the role of the speed phase space. establish a mechanical Lagrangian $ L = K - V: TQ\rightarrow { \mathbb R}$ where $ K$ and $ V$ are the kinetic and potential energy, respectively. and the distribution $ { \mathcal D}$, the dynamics of the nonholonomic system is wholly determined using the Lagrange - d’Alembert principle [ @Bl ]. In this paper we will formulate a description in terms of a Levi Civita connection defined on the space of vector field carry values on $ { \mathcal D}$. This connection is obtain by projecting the standard Lie bracket use the Riemannian metric associate with the kinetic department of energy $ K$ (see [ @maria2 ]) and the typical characterization of the Levi - Civita joining (see besides [ @BlochHussein ]). By adding controls in this setting we can learn optimal control

wihh blades or skates. Thert are thus multiple applmcationa in the context of wheeled motion, s'ace or mibile robotics and robutic maniiulation. Un tiis paper, we will introdmee sojc new jeometric technlques in notholonomic mecvavies to study the case of force minimising opyilal control prjblekf. The application of modern tools from siffereitial geometry on the fields of mechanics, conhrol theory, field hheory and bumewucal integragion has ltd to signifjcant progress in these researcf arecs. For instqnxe, hve study of the deometrical nprmuladion of the nonholonokic eqyations of motion has led to better undgrstanding oy different engineeribg probnems sucf lozomktmon generwtikn, controlmability, morion planning, and ttatvvtory trackihg (see e.d. [@Bl], [@bloch2], [@blochcrouch], [@blochcrouch2], [@blocvcrkuch3], [@borisovmamaev], [@bullilewis], [@kellymurray], [@jalr], [@LR1], [@LMdQ], [@lewisandrew], [@murraysastry], [@ostrowoski] and referenwes tiefeiu). Geomdrrlc techniques can also be used to study optimwm voktrol problems (sce [@blochcrouch3], [@coryed; @mjho], [@CoMa], [@Vu], [@Vo2], [@Su], [@SuLh]). Dombining these idfas in jhis pqper, we suudy yhe underlying geometry of iptimal contgol problems for mechauical systemr sunject to nonholonomic constrcints znd we applj it to ssxeral interestine eqampnes. Classical nonholonomic sonstrainvs whnch are uinesr in ehe velocihies gdn be geometricallj encldad by a cojstant rank distribution $\mathcal{D}$. As we will xea, tve distrnbutiok $\mathcal{D}$ will play the role of the velozity phase space. Jiven a mechwnical Lagranclan $L=K-V: TQ\rijhtarrow {\iathvb R}$ where $Y$ and $V$ are the kinetic cud potentual energy, respectlvely. znd the distribbuiob ${\mathcal D}$, the dyvamycd pf dhe nonholonmmic syryem ir compltceln ddterkined using the Lagrdnge-s’Alembert principlr [@Nl]. In thiw paper re will formukate a description in txrms oh a Lefi Sivita connection defined on tge space lf yector fields taklng dalues on ${\kathcal D}$. This connection is obtained bb projecting the standaed Lie bracket usiny uhe Riemanniai metryc associdted with the kinetix energy $K$ (see [@mavia2]) and the typical chzractesizatlon of the Levi-Civita connection (see also [@BlochHussein]). By adding controls in thms setting we can studj o'tikal cogtrop

with blades or skates. There are thus in context of motion, space or In paper, we will some new geometric in nonholonomic mechanics to study the of force minimizing optimal control problems. The application of modern tools from differential in the fields of mechanics, control theory, field theory and numerical integration has to progress these areas. For instance, the study of the geometrical formulation of the nonholonomic equations of motion has to better understanding of different engineering problems such generation, controllability, motion planning, trajectory tracking (see e.g. [@Bl], [@blochcrouch], [@blochcrouch3], [@borisovmamaev], [@kellymurray], [@LR1], [@lewisandrew], [@murraysastry], [@ostrowoski] references therein). Geometric techniques can also be used to study optimal control problems (see [@blochcrouch3], [@cortes; @noho], [@Vu], [@Vu2], Combining these in paper, study the underlying optimal control problems for mechanical systems constraints and we apply it to several interesting Classical nonholonomic which are linear in the velocities be geometrically encoded by a constant rank distribution As we will see, the distribution $\mathcal{D}$ will play the role of the velocity phase a mechanical Lagrangian $L=K-V: {\mathbb R}$ where and are kinetic potential energy, and the distribution ${\mathcal D}$, the dynamics of the nonholonomic system completely determined using the Lagrange-d’Alembert principle [@Bl]. In this paper formulate description in terms a Levi Civita connection on space of vector fields on D}$. obtained projecting standard Lie bracket using Riemannian metric associated with the energy $K$ (see [@maria2]) the Levi-Civita connection (see also [@BlochHussein]). By adding in this setting we can study optimal

with blades or skates. There arE thus multiPle apPliCatIoNs in The cOntext of wheeleD MotiOn, space or mobile roboticS and rObOTic mANiPulatIon. In thIS pAPEr, wE wIlL inTrODuCe somE neW geometRic techniqUes In Nonholonomic MEcHanics to stUdy The case of forCe mInimizInG opTImal cOntRol prOblems. tHe applIcation of MoDErn tooLS from diFFErEntiAl geometry in the fiELdS Of mechanics, conTrol thEoRY, fIELd tHeoRy and numerIcAl intEGration HAs LED To sIGnificant progRess in these REseArch arEaS. FoR InstanCe, the StUDy oF the geometrIcal FormulatiOn of thE NonholoNOmic equAtions Of mOtiOn haS LeD tO beTtER unDErStaNDinG of diffeReNt EnginEeriNG PRObleMs sUch lOcomoTion generatioN, coNtroLLabIlity, MotioN plaNnIng, anD trajeCtory TrAcking (see e.g. [@Bl], [@blOch2], [@bLochcroucH], [@blOcHcrOuCh2], [@bloCHcroucH3], [@boRisOvmamaeV], [@bullolEWis], [@KeLLYMuRray], [@jair], [@LR1], [@LMdD], [@lewiSaNDReW], [@murraysAstry], [@oSTrOwOSki] and reFeRenCes tHERein). GEomeTRiC techniqUes can ALsO bE used to StUdy optImAl cOntRol prOBlemS (see [@blOchcroucH3], [@cortES; @noho], [@CoMa], [@Vu], [@Vu2], [@SU], [@suJu]). Combining THeSE IdEAs in ThiS paper, we stuDy thE UndeRlyiNG gEomETry of OptimAl COnTRol problems for mechaNiCal sysTems sUbject to nonhoLonomic conSTRAints and We apPLy IT to several inteRestiNg examples. cLassical NonhoLonomic cOnstraintS WHich are lIneAr iN thE veLOCiTies can be geomETRicaLlY encodeD by A constaNt rAnk DisTriBuTion $\mathcAl{D}$. As we wIlL sEe, ThE diStribUTion $\mathCaL{D}$ wIlL plAy the ROle of tHe velOcitY pHaSE spAce. GiveN A mECHaniCaL LAgraNgiAn $l=K-V: TQ\RighTArrOw {\mathbB R}$ where $K$ aNd $V$ ARe thE kInEtic and Potential enerGy, RespectiveLy. And The disTRIbution ${\mAthcal D}$, the dynamics of the NOnholonOmiC systEm is CompletelY deTerminEd uSIng the lagranGe-d’AlEmBerT PRinciPLE [@BL]. In ThIs paper we wILL foRmulaTe A desCriptioN in terms of a Levi CivITa cOnnection defiNed On thE SPaCe oF VeCTor FiELds TAKing values on ${\matHcal D}$. This cOnNEcTion is obtaINed By ProjectIng the sTandaRD Lie braCket using The RiemanNiAn meTRIc aSsociated wIth the kiNetic enerGY $K$ (see [@MArIa2]) and The TypicaL cHarActerIzatioN Of tHe LevI-CivitA cOnnectIon (seE aLso [@Blochhussein]). By adding controls In this SettiNg wE can study OptIMal Control

with blades or skates. Th ere are th us mu lti ple a ppli cati ons in the con t extof wheeled motion, spa ce or m o bile ro botic s and r o bo t i c m an ip ula ti o n. In t his paper, we will i ntr od uce some new ge ometric te chn iques in non hol onomic m ech a nicstostudy the c a se offorce min im i zing o p timal c o n tr ol p roblems. The app l ic a tion of modern tools f r om d iff ere ntial geom et ry in the fie l ds o f me c hanics, contr ol theory,f iel d theo ry an d numer icalin t egr ation has l ed t o signifi cant p r ogressi n these resea rch ar eas. Fo rins ta n ce, th e s t udy of thege om etric al f o r m u lati onof t he no nholonomic eq uat ions ofmotio n has led t o bet ter un derst an ding of differe nt e ngineerin g p ro ble ms such locomo tio n g enerati on, con t rol la b i l it y, motion planning ,a n dtrajecto ry tra c ki ng (see e.g .[@B l],[ @ bloch 2],[ @b lochcrou ch], [ @ bl oc hcrouch 2] , [@bl oc hcr ouc h3],[ @bor isovma maev], [ @bull o lewis], [@kell y murray], [@ja i r] , [@ L R1], [@ LMdD], [@le wisa n drew ], [ @ mu rra y sastr y], [ @o s tr o woski] and referenc es there in).Geometric tec hniques ca n a lso be u sedt os tudy optimal c ontro l problems (see [@b lochc rouch3], [@cortes ; @noho],[@C oMa ],[@V u ] ,[@Vu2], [@Su] , [@Su Ju ]). Com bin ing the seide asinth is paper, we stud yth eun der lying geometry o f o pt ima l con t rol pr oblem s fo rme c han ical sy s te m s sub je ct tonon ho lonom ic c o nst raintsand we ap ply it t ose veral i nteresting ex am ples. Cla ss ica l nonh o l onomic c onstraints which are li n ear inthe velo citi es can be ge ometri cal l y enco ded by a co ns tan t rankd i st rib ut ion $\math c a l{D }$. A swe w ill see , the distribution $\m athcal{D}$ wi llplay t he ro l eo f t he vel o c ity phase space . Given ame c ha nical Lagr a ngi an $L=K-V : TQ\ri ghtar r ow {\ma thbb R}$where $K$ a nd $ V $ ar e the kine tic andpotential energ y ,respe cti vely.an d t he di stribu t ion ${\m athcal D }$, th e dyn am ics of t he nonholonomic systemis com plete lydetermine d u s ing the Lagr ange -d’Alember t p rin ciple [@ B l]. I n th i spap e r wewill formulate ades c r ip tion in ter m s ofa Lev i C i vita c onne ction defined ont he space of ve ctor f iel dst akin gvalues on ${\m ath ca l D}$. Thi sconnectionis obtai ne d by p roject ing th e stand a r dL ie bra cket us ing the R iem an n ian met ri ca ssocia tedwi th the kinet i c en e r gy $K$ (see [@ma ria2] ) and t h e t ypica lcharact e riza tion of th e Levi-Civi ta con nect ion ( see als o[@Bloc hHu ss ein]). Bya dding con trols in thi ssett ing we ca n st u d y opt imal c ont rol

with_blades or_skates. There are thus_multiple applications_in_the context_of_wheeled motion, space_or mobile robotics_and robotic manipulation. In_this paper, we_will_introduce some new geometric techniques in nonholonomic mechanics to study the case of force_minimizing_optimal control_problems. The_application_of modern tools from differential_geometry in the fields of_mechanics, control_theory, field theory and numerical integration has led_to_significant progress in_these research areas. For instance, the study of the_geometrical formulation of the nonholonomic equations_of motion has_led_to_better understanding of different_engineering problems such locomotion generation, controllability,_motion planning, and trajectory tracking (see_e.g. [@Bl], [@bloch2], [@blochcrouch], [@blochcrouch2], [@blochcrouch3], [@borisovmamaev],_[@bullolewis], [@kellymurray], [@jair], [@LR1], [@LMdD], [@lewisandrew],_[@murraysastry], [@ostrowoski] and references therein)._Geometric techniques_can also be used to_study optimal control_problems (see_[@blochcrouch3], [@cortes; @noho],_[@CoMa], [@Vu], [@Vu2], [@Su], [@SuJu]). Combining_these ideas in_this paper, we study the underlying_geometry_of optimal control_problems_for_mechanical systems_subject to nonholonomic_constraints_and we_apply_it to several interesting examples. Classical nonholonomic_constraints_which are linear in the velocities can_be geometrically encoded by_a_constant rank distribution $\mathcal{D}$._As we will see, the_distribution $\mathcal{D}$ will play the role_of the_velocity phase_space. Given a mechanical Lagrangian $L=K-V: TQ\rightarrow {\mathbb R}$ where $K$_and $V$ are the kinetic and_potential energy, respectively. and_the distribution_${\mathcal_D}$, the dynamics_of_the nonholonomic_system is completely determined using the Lagrange-d’Alembert_principle [@Bl]._In this paper we will formulate_a description in terms_of_a Levi Civita connection defined on_the space of vector fields taking_values on ${\mathcal D}$. This_connection_is_obtained by projecting the standard_Lie bracket using the Riemannian metric_associated with the_kinetic energy $K$ (see [@maria2]) and the_typical_characterization of the Levi-Civita connection (see_also_[@BlochHussein]). By adding controls in this_setting_we_can study optimal control

of two distinct classical geometries. This includes the transmission of signals sent at very large early times, as observed in [@maldacena2017diving; @gao2018regenesis]. In \[sec:experiment\] we discuss concrete experimental realizations of teleportation by size. Unlike the field theories of AdS/CFT, systems that can be realized in an atomic physics lab all have only a finite number of degrees of freedom. We will show that having an infinite number of degrees of freedom is not necessary for the transmission of information, and indeed we will explicitly describe a modest quantum system that manifests the phenomenon. A detailed analysis of many of the results presented here will provided in [@longpaper]. Other studies of information transfer through traversable wormholes, and related notions, include [@freivogel2019traversable; @Yoshida2017]. ![A short summary of paper, discussing different systems, different patterns of operator growth, and consequence of each growth pattern for signal transmission. [**Blue:**]{} Initial operator-size distribution. [**Red:**]{} Operator-size distribution of the time-evolved operator.[]{data-label="fig:Summary_of_everything"}](043.pdf){width="10"} Teleportation by Size {#sec:sizeandmomentum} ===================== We base most of our analysis in this paper on size distributions and operator growth, notions heavily studied in connection to holography [@susskind2019complexity; @lin2019symmetries; @qi2019quantum] and many-body physics [@Nahum_2018; @von_Keyserlingk_2018; @Xu_2019]—hence the term *teleportation by size*. In \[subsec:swinding\], we introduce a property of size distributions, called *size winding*. Size winding gives a clean mechanism for operator transfer that abstracts the way geometrical wormholes work at the level of the boundary theory (we discuss the latter in \[sec:holography\] below). In \[subsec:Teleportation\_by\_Size,subsec:TS\_Hamiltonian\_General\] we discuss the state transfer protocol. We will see that state transfer can be done for very generic chaotic quantum systems—even at infinite temperature—and give general bounds on the fidelity. Size Winding {#subsec:swinding} ------------ Consider an observable $O$ and its transpose (in the computational basis) $O^T$ acting at time

of two distinct classical geometries. This includes the transmittance of signal sent at very large early times, as note in   [ @maldacena2017diving; @gao2018regenesis ]. In \[sec: experiment\ ] we discuss concrete experimental realizations of teleportation by size. Unlike the playing field hypothesis of AdS / CFT, systems that can be realized in an atomic physics lab all get only a finite number of degrees of freedom. We will show that having an countless number of degrees of freedom is not necessary for the infection of information, and indeed we will explicitly describe a humble quantum system that attest the phenomenon. A detailed analysis of many of the results presented here will provided in   [ @longpaper ]. Other study of information transfer through traversable wormholes, and related notions, include   [ @freivogel2019traversable; @Yoshida2017 ]. ! [ A light summary of paper, discussing different systems, different patterns of operator growth, and consequence of each growth pattern for signal transmittance. [ * * Blue :* * ] { } Initial operator - size distribution. [ * * Red :* * ] { } hustler - size distribution of the meter - develop operator.[]{data - label="fig: Summary_of_everything"}](043.pdf){width="10 " } Teleportation by Size { # sec: sizeandmomentum } = = = = = = = = = = = = = = = = = = = = = We base most of our analysis in this paper on size distributions and operator growth, notions heavy studied in connection to holography   [ @susskind2019complexity; @lin2019symmetries; @qi2019quantum ] and many - body physics   [ @Nahum_2018; @von_Keyserlingk_2018; @Xu_2019]—hence the term   * teleportation by size *. In \[subsec: swinding\ ], we introduce a property of size distributions, called   * size winding *. Size winding gives a uninfected mechanism for operator transportation that abstract the way geometrical wormholes work at the floor of the boundary theory (we discuss the latter in \[sec: holography\ ] below). In \[subsec: Teleportation\_by\_Size, subsec: TS\_Hamiltonian\_General\ ] we discuss the state transfer protocol. We will see that state transportation can be done for very generic chaotic quantum systems — even at infinite temperature — and give general boundary on the fidelity. Size Winding { # subsec: swinding } ------------ Consider an discernible $ O$ and its transpose (in the computational basis) $ O^T$ acting at time

of two distinct classical neometries. This nbcludev the fransmisrion of signals sent at very lqrge tcrly times, as observdd in [@maldwcena2017divung; @jao2018regenesis]. In \[sxd:experiment\] ws diseuws concrete exkerimental raalizations of tdlzportation by size. Unlike the field eheoriex lf AdS/CFT, systgms tnwt czn be realized in an atomic physica lab anl have only s finite number of degrees of vreedom. We will shlw that havung wb infinite njmber of dtgxees of fregdom is not necessary for the trxnsmixsion of ibfirmwjion, and indxed we will expliclnly deswribe a modest quantuk sbsten that manifests the 'henomenon. A detailed analysis oy many of the results peesenjed hare dull prkvmdes in [@lojgpeper]. Other atudies of unformation transfet eyrough traveraable rowmholes, and related notions, include [@freinogem2019traversable; @Yoshida2017]. ![A syort summary of paper, discussigg different systems, different patterns of operatmr grkdth, akd cubsfquence of each growth pattern for signal tragamosxion. [**Blue:**]{} Initlal operator-size doshrofution. [**Red:**]{} Opgrator-snae distribution of tje time-gvolvee operatow.[]{dats-label="fig:Summary_of_everythint"}](043.pdf){width="10"} Telvporration by Size {#sec:dizeandmomeutum} ===================== We base most of our analysis iu this paper on slze distrjcutions and operxtog grmwth, notions heavily studiqd in coniectipn to hulogtaphy [@sufskind2019comppexitn; @lin2019symmetries; @qi2019qkantul] dnd many-bofy physics [@Nahum_2018; @von_Keyserlingk_2018; @En_2019]—hence the tetm *taleiortation by slze*. In \[subsec:swynding\], we intrpduce a propdrty of side distriuutions, callqd *size windinc*. Size windinj gives a cleqn mwchaniso for operator yransfer nhct abstraxts the way geometvical sormholes work cu tye level of the bojndwrj tieory (fe discuss tve lxttdt in \[rec:holograpmy\] celoe). In \[subsec:Teleportadion\_gy\_Size,subsec:TS\_Hamiktjnian\_Gengral\] we dyscuss the stste transfer protofol. Wx will see yhaj state transfer can be done fod very gejerlc chaotic quwntun systems—eveu at infinite temperature—and give generao bounds on the fieelity. Size Winding {#xmbsec:swindinj} ------------ Consiqer an obvervable $O$ and its teanspose (in the cpmputational basis) $O^T$ actinc at hime

of two distinct classical geometries. This includes of sent at large early times, In we discuss concrete realizations of teleportation size. Unlike the field theories of systems that can be realized in an atomic physics lab all have only finite number of degrees of freedom. We will show that having an infinite of of is necessary for the transmission of information, and indeed we will explicitly describe a modest quantum system manifests the phenomenon. A detailed analysis of many the results presented here provided in [@longpaper]. Other studies information through traversable and notions, [@freivogel2019traversable; @Yoshida2017]. ![A summary of paper, discussing different systems, different patterns of operator growth, and consequence of each growth pattern signal transmission. operator-size distribution. Operator-size of time-evolved operator.[]{data-label="fig:Summary_of_everything"}](043.pdf){width="10"} Teleportation {#sec:sizeandmomentum} ===================== We base most of this paper on size distributions and operator growth, heavily studied connection to holography [@susskind2019complexity; @lin2019symmetries; @qi2019quantum] many-body physics [@Nahum_2018; @von_Keyserlingk_2018; @Xu_2019]—hence the term *teleportation size*. In \[subsec:swinding\], we introduce a property of size distributions, called *size winding*. Size winding clean mechanism for operator that abstracts the geometrical work the of the theory (we discuss the latter in \[sec:holography\] below). In \[subsec:Teleportation\_by\_Size,subsec:TS\_Hamiltonian\_General\] we the state transfer protocol. We will see that state transfer done very generic chaotic systems—even at infinite temperature—and general on the fidelity. Size ------------ an its (in computational basis) $O^T$ acting time

of two distinct classical geoMetries. ThiS inclUdeS thE tRansMissIon of signals seNT at vEry large early times, as obServeD iN [@MaldACeNa2017divIng; @gao2018rEGeNESis]. in \[SeC:exPeRImEnt\] we DisCuss conCrete experImeNtAl realizatioNS oF teleportaTioN by size. UnlikE thE field ThEorIEs of ADS/CfT, sysTems thAT can be Realized iN aN Atomic PHysics lAB AlL havE only a finite numbeR Of DEgrees of freedoM. We wilL sHOw THAt hAviNg an infiniTe NumbeR Of degreES oF FREedOM is not necessaRy for the traNSmiSsion oF iNfoRMation, And inDeED we Will expliciTly dEscribe a mOdest qUAntum sySTem that ManifeSts The PhenOMeNoN. A dEtAIleD AnAlySIs oF many of tHe ReSults PresENTED herE wiLl prOvideD in [@longpaper]. OTheR stuDIes Of infOrmatIon tRaNsfer ThrougH travErSable wormholes, aNd reLated notiOns, InCluDe [@FreivOGel2019traVerSabLe; @YoshiDa2017]. ![A shorT SumMaRY OF pAper, discussing diffErENT sYstems, diFferenT PaTtERns of opeRaTor GrowTH, And coNseqUEnCe of each Growth PAtTeRn for siGnAl tranSmIssIon. [**blue:**]{} INItiaL operaTor-size dIstriBUtion. [**Red:**]{} OperatOR-size distribuTIoN OF tHE timE-evOlved operatOr.[]{daTA-labEl="fiG:suMmaRY_of_evErythInG"}](043.PdF){Width="10"} Teleportation bY SIze {#sec:SizeaNdmomentum} ===================== We bAse most of oUR ANalysis iN thiS PaPEr on size distriButioNs and operaTOr growth, NotioNs heavilY studied iN COnnectioN to HolOgrAphY [@SUsSkind2019complexiTY; @Lin2019sYmMetries; @Qi2019qUantum] aNd mAny-BodY phYsIcs [@Nahum_2018; @vOn_KeyserLiNgK_2018; @XU_2019]—hEncE the tERm *telepoRtAtiOn By sIze*. In \[SUbsec:sWindiNg\], we InTrODucE a propeRTy OF Size DiStRibuTioNs, CalleD *sizE WinDing*. SizE winding gIveS A cleAn MeChanism For operator trAnSfer that abStRacTs the wAY GeometriCal wormholes work at the leVEl of the BouNdary TheoRy (we discuSs tHe lattEr iN \[Sec:holOgraphY\] beloW). IN \[suBSEc:TelEPOrTatIoN\_by\_Size,subSEC:TS\_hamilToNian\_general\] We discuss the state tRAnsFer protocol. We WilL see THAt StaTE tRAnsFeR Can BE Done for very geneRic chaotic QuANtUm systems—eVEn aT iNfinite TemperaTure—aND give geNeral bounDs on the fiDeLity. sIZe WInding {#subsEc:swindiNg} ------------ ConsideR An obsERvAble $O$ And Its traNsPosE (in thE compuTAtiOnal bAsis) $O^T$ AcTing at Time

of two distinct classical geometrie s. Th isinc lu desthetransmission o f sig nals sent at very larg e ear ly time s ,as ob servedi n[ @ mal da ce na2 01 7 di ving; @g ao2018r egenesis]. I n\[sec:experi m en t\] we dis cus s concrete e xpe riment al re a lizat ion s oftelepo r tation by size. U n like t h e field t he orie s of AdS/CFT, sys t em s that can be r ealize di na n at omi c physicsla b all have on l ya f ini t e number of d egrees of f r eed om. We w ill show t hat h av i ngan infinite num ber of de greeso f freed o m is no t nece ssa ryfort he t ran sm i ssi o nofi nfo rmation, a nd inde ed w e w i ll e xpl icit ly de scribe a mode stquan t umsyste m tha t ma ni fests the p henom en on. A detailed ana lysis ofman yofth e res u lts pr ese nte d herewill pr o vid ed i n  [ @longpaper]. Other s t u di es of in format i on t r ansfer t hr oug h tr a v ersab le w o rm holes, a nd rel a te dnotions ,includ e[@f rei vogel 2 019t ravers able; @Y oshid a 2017]. ![A sh o rt summary of pa p e r, disc uss ing differe nt s y stem s, d i ff ere n t pat terns o f o p erator growth, andco nseque nce o f each growth pattern f o r signal t rans m is s ion. [**Blue:* *]{}Initial op e rator-si ze di stributi on. [**Re d : **]{} Op era tor -si zed i st ribution of t h e tim e- evolved op erator. []{ dat a-l abe l= "fig:Summ ary_of_e ve ry th in g"} ](043 . pdf){wid th ="1 0" } Telep o rtatio n bySize { #s e c:s izeandm o me n t um}== == ==== === == ===== ===Webase mo st of our an a lysi sin this p aper on sizedi stribution sand opera t o r growth , notions heavily studi e d in co nne ction toholograph y [ @sussk ind 2 019com plexit y; @l in 201 9 s ymmet r i es ; @ qi 2019quantu m ] an d man y- body physic s [@Nahum_2018; @v o n_K eyserlingk_20 18; @Xu _ 2 01 9]— h en c e t he ter m *teleportationby size*.In \[ subsec:swi n din g\ ], we i ntroduc e a p r opertyof size d istributi on s, c a l led  *size win ding*. S ize windi n g giv e sa cle anmechan is m f or op erator tra nsfer thatab stract s the w ay geome trical wormholes work a t thelevel of the boun dar y th eory (wedisc uss the la tte r i n \[s ec: h ologr aphy \ ]bel o w). I n \[ s ubsec:Tel e po rta t i on \_by\_Size, s u b sec :TS\_ Ham i ltonia n\_G eneral\] we discu s s the state tr ansf e r pr oto c ol.We will see that st at e transfer c an be donefor very g e neric chaot ic qua ntum sy s t em s —evenat i nfi nite temp era tu r e—and g iv eg eneral bou nd s on t he fid e lity . Size Winding {# subse c : swind i ng} ---- -- ------Cons ider an ob servable $O $ anditstrans pose (i nthe co mpu ta tional bas i s) $O^T$actin g at ti me

of_two distinct_classical geometries. This includes_the transmission_of_signals sent_at_very large early_times, as observed_in [@maldacena2017diving; @gao2018regenesis]. In \[sec:experiment\] we_discuss concrete experimental_realizations_of teleportation by size. Unlike the field theories of AdS/CFT, systems that can be_realized_in an_atomic_physics_lab all have only a_finite number of degrees of_freedom. We_will show that having an infinite number of_degrees_of freedom is_not necessary for the transmission of information, and indeed_we will explicitly describe a modest_quantum system that_manifests_the_phenomenon. A detailed analysis of_many of the results presented here_will provided in [@longpaper]. Other studies of_information transfer through traversable wormholes, and related_notions, include [@freivogel2019traversable; @Yoshida2017]. ![A short summary of_paper, discussing different systems, different_patterns of_operator growth, and consequence of_each growth pattern_for signal_transmission. [**Blue:**]{} Initial_operator-size distribution. [**Red:**]{} Operator-size distribution of_the time-evolved operator.[]{data-label="fig:Summary_of_everything"}](043.pdf){width="10"} Teleportation_by Size {#sec:sizeandmomentum} ===================== We base most of_our_analysis in this_paper_on_size distributions_and operator growth,_notions_heavily studied_in_connection to holography [@susskind2019complexity; @lin2019symmetries; @qi2019quantum] and_many-body_physics [@Nahum_2018; @von_Keyserlingk_2018; @Xu_2019]—hence the term *teleportation by size*._In \[subsec:swinding\], we introduce_a_property of size distributions,_called *size winding*. Size winding gives_a clean mechanism for operator transfer_that abstracts_the way_geometrical wormholes work at the level of the boundary theory (we_discuss the latter in \[sec:holography\] below)._In \[subsec:Teleportation\_by\_Size,subsec:TS\_Hamiltonian\_General\] we discuss_the state_transfer_protocol. We will_see_that state_transfer can be done for very generic_chaotic quantum_systems—even at infinite temperature—and give general_bounds on the fidelity. Size_Winding_{#subsec:swinding} ------------ Consider an observable $O$ and its_transpose (in the computational basis) $O^T$_acting at time

s$ which suffices for a *general* choice of $F_i$. (See Bronowski [@Bronowski1] for a discussion of $n$-ary forms.) Here we consider a more general version, i.e., we fix $s$ and consider the locus $\Sigma$ of forms $\{F_1,\dots,F_t\}$ which admit such a representation. In practice, we allow not only representations as above, but also generalised additive decompositions (GADs) in the sense of [@IK Definition 1.30]. A finite collection ${{\underline L}}= \{L_i\} \subseteq S_1 \setminus \{0\}$ of linear forms will be called *admissible* if any two $L_i,L_j$ are nonproportional. Let ${{\underline L}}$ be an admissible collection as above and $F \in S_d$. A *GAD for $F$ with respect to ${{\underline L}}$* is a collection $\{p_i\}$ of forms such that $F = \sum\limits_i p_i \, L_i^{d - \alpha_i}$, and $\alpha_i = \deg p_i < d$. (We set $\deg \, 0 = -1$ by convention.) The integer $\ell_{{{\underline \alpha}}} = \sum\limits_i (\alpha_i + 1)$ is called the *length* of the GAD. If all $\alpha_i =0$, then this reduces to the expression (\[add.expression\]). Given sequences ${{\underline \alpha}}$ and ${{\underline L}}$ such that $-1 \le \alpha_i < d$, define a subspace $$W({{\underline \alpha}}, {{\underline L}}) = \sum S_{\alpha_i} L^{d-\alpha_i} \subseteq S_d.$$ (By convention, $S_{-1} = 0$.) Write $L_i = b_i y_1 - a_i y_2$, then by hypothesis, $[a_i,b_i]$ represent distinct points of ${{\mathbf P}}^1$. With notation as above, - A form lies in $W({{\underline \alpha}}, {{\underline L}})$ iff it is apolar to $\prod\limits_i (a_i x_1 + b_i x_2)^{\alpha_i+1}$. In particular, $\dim W({{\underline \alpha}},{{\underline

s$ which suffices for a * general * choice of $ F_i$. (See Bronowski [ @Bronowski1 ] for a discussion of $ n$-ary form .) Here we think a more general version, i.e., we fix $ s$ and study the locus $ \Sigma$ of forms $ \{F_1,\dots, F_t\}$ which accommodate such a theatrical performance. In practice, we allow not only representations as above, but also generalize additive decompositions (GADs) in the common sense of [ @IK Definition 1.30 ]. A finite collection $ { { \underline L}}= \{L_i\ } \subseteq S_1 \setminus \{0\}$ of linear forms will be called * admissible * if any two $ L_i, L_j$ are nonproportional. Let $ { { \underline L}}$ be an admissible collection as above and $ F \in S_d$. A * GAD for $ F$ with respect to $ { { \underline L}}$ * is a solicitation $ \{p_i\}$ of forms such that $ F = \sum\limits_i p_i \, L_i^{d - \alpha_i}$, and $ \alpha_i = \deg p_i < d$. (We set $ \deg \, 0 = -1 $ by convention .) The integer $ \ell_{{{\underline \alpha } } } = \sum\limits_i (\alpha_i + 1)$ is call the * length * of the GAD. If all $ \alpha_i = 0 $, then this reduces to the saying (\[add.expression\ ]). Given sequence $ { { \underline \alpha}}$ and $ { { \underline L}}$ such that $ -1 \le \alpha_i < d$, define a subspace $ $ W({{\underline \alpha } }, { { \underline L } }) = \sum S_{\alpha_i } L^{d-\alpha_i } \subseteq S_d.$$ (By convention, $ S_{-1 } = 0$.) Write $ L_i = b_i y_1 - a_i y_2 $, then by hypothesis, $ [ a_i, b_i]$ represent discrete points of $ { { \mathbf P}}^1$. With notation as above, - A form lies in $ W({{\underline \alpha } }, { { \underline L}})$ iff it is apolar to $ \prod\limits_i (a_i x_1 + b_i x_2)^{\alpha_i+1}$. In particular, $ \dim W({{\underline \alpha}},{{\underline

s$ wjich suffices for a *geneval* choice of $F_i$. (See Brmnowskj [@Bronowrki1] for a discussion of $n$-ary firms.) Yere we consider a mord general version, i.e., xe fix $s$ and consider thc lochd $\Siyme$ of forms $\{F_1,\dotx,F_t\}$ which ddmit such a rapfedentation. In practice, we allow not jnly relrfsentations as abone, but also generalised additive decompoaitions (GADs) in the xense of [@IK Definition 1.30]. A flnitf collection ${{\underpine L}}= \{L_i\} \sogseewq S_1 \setminur \{0\}$ of linear forms wilm be called *admissible* if any tdo $L_i,K_j$ are nonkxipogjional. Let ${{\unverlinv L}}$ be an admlxsible collecyion as above snd $F \un S_d$. A *GAD for $F$ witi respect to ${{\underlige L}}$* is a ckllection $\{p_i\}$ of firns suwh tvat $W = \sjm\ljmmts_j p_i \, P_i^{d - \alpha_i}$, ahd $\alpha_i = \eeg p_i < d$. (We set $\deb \, 0 = -1$ by conventjon.) Thq ynteger $\ell_{{{\underline \alpha}}} = \sum\limits_i (\dlpga_i + 1)$ is called the *lentth* of the GAD. If all $\wlpha_i =0$, tren this reduces to the expression (\[add.expression\]). Civen reqbckces ${{\ynferline \alpha}}$ and ${{\underline L}}$ such that $-1 \le \ajlhs_i < d$, define a smbspace $$W({{\underline \appnw}}, {{\underline L}}) = \sum S_{\allha_i} L^{d-\alpha_i} \subsfteq S_d.$$ (By cinvention, $S_{-1} = 0$.) Write $L_i = b_i y_1 - a_i y_2$, then vy hypothesif, $[a_i,b_i]$ represent didtinct poincs of ${{\kathbg P}}^1$. With notation as abore, - A form lies ln $W({{\undermkne \alpha}}, {{\underlkne L}})$ iff it is apolar to $\prod\lymits_i (a_i x_1 + y_i x_2)^{\alphx_i+1}$. Im partycular, $\dim W({{\undcsline \alpha}},{{\underlije

s$ which suffices for a *general* choice (See [@Bronowski1] for discussion of $n$-ary more version, i.e., we $s$ and consider locus $\Sigma$ of forms $\{F_1,\dots,F_t\}$ which such a representation. In practice, we allow not only representations as above, but generalised additive decompositions (GADs) in the sense of [@IK Definition 1.30]. A finite ${{\underline \{L_i\} S_1 \{0\}$ of linear forms will be called *admissible* if any two $L_i,L_j$ are nonproportional. Let ${{\underline be an admissible collection as above and $F S_d$. A *GAD for with respect to ${{\underline L}}$* a $\{p_i\}$ of such $F \sum\limits_i p_i \, - \alpha_i}$, and $\alpha_i = \deg p_i < d$. (We set $\deg \, 0 = -1$ by The integer = \sum\limits_i + is the *length* of If all $\alpha_i =0$, then this expression (\[add.expression\]). Given sequences ${{\underline \alpha}}$ and ${{\underline such that \le \alpha_i < d$, define a $$W({{\underline \alpha}}, {{\underline L}}) = \sum S_{\alpha_i} L^{d-\alpha_i} S_d.$$ (By convention, $S_{-1} = 0$.) Write $L_i = b_i y_1 - a_i y_2$, then $[a_i,b_i]$ represent distinct points ${{\mathbf P}}^1$. With as - form in $W({{\underline {{\underline L}})$ iff it is apolar to $\prod\limits_i (a_i x_1 + x_2)^{\alpha_i+1}$. In particular, $\dim W({{\underline \alpha}},{{\underline

s$ which suffices for a *general* Choice of $F_i$. (see BrOnoWskI [@BRonoWski1] For a discussion OF $n$-arY forms.) Here we consider a mOre geNeRAl veRSiOn, i.e., wE fix $s$ anD CoNSIdeR tHe LocUs $\sIgMa$ of fOrmS $\{F_1,\dots,F_T\}$ which admiT suCh A representatIOn. in practice, We aLlow not only rEprEsentaTiOns AS abovE, buT also GeneraLIsed adDitive decOmPOsitioNS (GADs) in THE sEnse Of [@IK Definition 1.30]. A fiNItE Collection ${{\undeRline L}}= \{l_i\} \SUbSETeq s_1 \seTminus \{0\}$ of liNeAr forMS will be CAlLED *AdmISsible* if any twO $L_i,L_j$ are nonPRopOrtionAl. let ${{\UNderliNe L}}$ be An ADmiSsible colleCtioN as above aNd $F \in S_D$. a *GAD for $f$ With resPect to ${{\UndErlIne L}}$* IS a CoLleCtIOn $\{p_I\}$ Of ForMS suCh that $F = \sUm\LiMits_i P_i \, L_i^{D - \ALPHa_i}$, aNd $\aLpha_I = \deg p_I < d$. (We set $\deg \, 0 = -1$ by cOnvEntiON.) ThE inteGer $\elL_{{{\undErLine \aLpha}}} = \suM\limiTs_I (\alpha_i + 1)$ is called The *lEngth* of thE GAd. IF alL $\aLpha_i =0$, THen thiS reDucEs to the ExpressIOn (\[aDd.EXPReSsion\]). Given sequenceS ${{\uNDErLine \alphA}}$ and ${{\unDErLiNE L}}$ such thAt $-1 \Le \aLpha_I < D$, DefinE a suBSpAce $$W({{\undeRline \aLPhA}}, {{\uNderlinE L}}) = \Sum S_{\alPhA_i} L^{D-\alPha_i} \sUBsetEq S_d.$$ (By ConventiOn, $S_{-1} = 0$.) WrITe $L_i = b_i y_1 - a_i y_2$, then BY hypothesis, $[a_i,B_I]$ rEPReSEnt dIstInct points oF ${{\matHBf P}}^1$. WIth nOTaTioN As aboVe, - A foRm LIeS In $W({{\underline \alpha}}, {{\unDeRline L}})$ Iff it Is apolar to $\proD\limits_i (a_i X_1 + B_I X_2)^{\alpha_i+1}$. IN parTIcULar, $\dim W({{\underliNe \alpHa}},{{\underlinE

s$ which suffices for a *g eneral* ch oiceof$F_ i$ . (S ee B ronowski [@Bro n owsk i1] for a discussion o f $n$ -a r y fo r ms .) He re we c o ns i d eramo rege n er al ve rsi on, i.e ., we fix$s$ a nd considert he locus $\S igm a$ of forms$\{ F_1,\d ot s,F _ t\}$whi ch ad mit su c h a re presentat io n . In p r actice, w eallo w not only repres e nt a tions as above , butal s og e ner ali sed additi ve deco m positio n s( G A Ds) in the senseof [@IK Def i nit ion 1. 30 ].A fini te co ll e cti on ${{\unde rlin e L}}= \{ L_i\}\ subsete q S_1 \s etminu s \ {0\ }$ o f l in ear f o rms wi llb e c alled *a dm is sible * if a n y two $L _i,L _j$ a re nonproport ion al.Let ${{\ under line L }}$ b e an a dmiss ib le collection a s ab ove and $ F \ in S_ d$ . A * G AD for $F $ w ith res pect to ${{ \u n d e rl ine L}}$* is a col le c t io n $\{p_i \}$ of fo rm s such th at $F = \ s u m\lim its_ i p _i \, L_ i^{d - \a lp ha_i}$, a nd $\a lp ha_ i = \deg p_i< d$.(We set$\deg \, 0 = -1$ byc onvention.) T h ei n te g er $ \el l_{{{\under line \alp ha}} } = \s u m\lim its_i ( \ al p ha_i + 1)$ is calle dthe *l ength * of the GAD. If all $ \ a l pha_i =0 $, t h en this reduces t o the expressio n (\[add. expre ssion\]) . Given s e q uences $ {{\ und erl ine \ al pha}}$ and ${ { \ unde rl ine L}} $ s uch tha t $ -1\le \a lp ha_i < d$ , define a s ub sp ace $$W( { {\underl in e \ al pha }}, { { \under lineL}}) = \ s umS_{\alp h a_ i } L^{ d- \a lpha _i} \ subse teqS _d. $$ (Byconventio n,$ S_{- 1} = 0$.) W rite $L_i = b _i y_1 - a_i y _2$ , then b y hypoth esis, $[a_i,b_i]$ repre s ent dis tin ct po ints of ${{\m ath bf P}} ^1$ . With notat ion a sabo v e , -Afor mlies in $W ( { {\u nderl in e \a lpha}}, {{\underline L}}) $ if f it is apola r t o $\ p r od \li m it s _i(a _ i x _ 1 + b_i x_2)^{\a lpha_i+1}$ .I nparticular , $\ di m W({{\ underli ne \a l pha}},{ {\underli ne

s$ which_suffices for_a *general* choice of_$F_i$. (See_Bronowski_[@Bronowski1] for_a_discussion of $n$-ary_forms.) Here we_consider a more general_version, i.e., we_fix_$s$ and consider the locus $\Sigma$ of forms $\{F_1,\dots,F_t\}$ which admit such a representation._In_practice, we_allow_not_only representations as above, but_also generalised additive decompositions (GADs)_in the_sense of [@IK Definition 1.30]. A finite collection ${{\underline_L}}=_\{L_i\} \subseteq S_1_\setminus \{0\}$ of linear forms will be called *admissible*_if any two $L_i,L_j$ are nonproportional. Let_${{\underline L}}$ be_an_admissible_collection as above and_$F \in S_d$. A *GAD for_$F$ with respect to ${{\underline L}}$*_is a collection $\{p_i\}$ of forms such_that $F = \sum\limits_i p_i \,_L_i^{d - \alpha_i}$, and $\alpha_i_= \deg_p_i < d$. (We set_$\deg \, 0_= -1$_by convention.) The_integer $\ell_{{{\underline \alpha}}} = \sum\limits_i (\alpha_i_+ 1)$ is_called the *length* of the GAD. If_all_$\alpha_i =0$, then_this_reduces_to the_expression (\[add.expression\]). Given_sequences_${{\underline \alpha}}$_and_${{\underline L}}$ such that $-1 \le_\alpha_i_< d$, define a subspace $$W({{\underline \alpha}},_{{\underline L}}) = \sum_S_{\alpha_i}_L^{d-\alpha_i} \subseteq S_d.$$ (By_convention, $S_{-1} = 0$.) Write_$L_i = b_i y_1 - a_i_y_2$, then_by hypothesis,_$[a_i,b_i]$ represent distinct points of ${{\mathbf P}}^1$. With notation as above, - _ A form lies in $W({{\underline_\alpha}}, {{\underline L}})$ iff_it is_apolar_to $\prod\limits_i (a_i_x_1_+ b_i_x_2)^{\alpha_i+1}$. In particular, $\dim W({{\underline \alpha}},{{\underline

model^a^ & $0.131$ & $0.150$\ \ Measured & $<0.030$^b^ & $0.079\pm0.019$\ With TiO/VO model & $0.039$ & $0.042$\ No TiO/VO model & $0.044$ & $0.045$\ \ Measured & $>0.128$^b^ & $0.116^{+0.018}_{-0.019}$\ With TiO/VO model & $0.121$ & $0.151$\ No TiO/VO model &$0.089$ & $0.105$\ \ Measured & $1.0\pm 1.1$ & $0.6\pm 1.1$\ With TiO/VO model & $-1.1$ & $-1.1$\ No TiO/VO model & $-3.2$ & $-3.2$\ \ Measured & $1.1^{+1.5}_{-1.6}$ & $0.6^{+1.5}_{-1.7}$\ With TiO/VO model & $-3.2$ & $-2.1$\ No TiO/VO model & $-5.3$ & $-4.2$\ [**Notes.**]{} ^a^[Both models for HAT-P-7b assume solar metallicity.]{} ^b^[Based on a $2\sigma$ upper limit of the minimum flux ratio.]{} ^c^[Difference between maximum and minimum flux ratios.]{} ^d^[The maximum and minimum flux offsets are measured relative to the center of secondary eclipse time and center of transit time, respectively. Negative time offsets indicate an eastward shift in the location of the hot or cold region in the planet’s atmosphere.]{} After subtracting and dividing out the brightness of the star alone, i.e., the flux measured during secondary eclipse, we obtain the phase variation of the relative planet–star flux ratio (Figures \[lightcurve1\] and \[lightcurve2\]). Tables \[tab:comparison1\] and \[tab:comparison2\] list the important quantitative characteristics of the WASP-19b and HAT-P-7b phase curves, respectively. By examining the properties of these phase curves, we obtain a basic picture of the temperature distribution across the planets’ surfaces, from which we can begin to infer aspects of the planets’ energy

model^a^ & $ 0.131 $ & $ 0.150$\ \ Measured & $ < 0.030$^b^ & $ 0.079\pm0.019$\ With TiO / VO model & $ 0.039 $ & $ 0.042$\ No TiO / VO model & $ 0.044 $ & $ 0.045$\ \ Measured & $ > 0.128$^b^ & $ 0.116^{+0.018}_{-0.019}$\ With TiO / VO exemplar & $ 0.121 $ & $ 0.151$\ No TiO / VO exemplar & $ 0.089 $ & $ 0.105$\ \ Measured & $ 1.0\pm 1.1 $ & $ 0.6\pm 1.1$\ With TiO / VO model & $ -1.1 $ & $ -1.1$\ No TiO / VO model & $ -3.2 $ & $ -3.2$\ \ Measured & $ 1.1^{+1.5}_{-1.6}$ & $ 0.6^{+1.5}_{-1.7}$\ With TiO / VO mannequin & $ -3.2 $ & $ -2.1$\ No TiO / VO model & $ -5.3 $ & $ -4.2$\ [ * * Notes. * * ] { } ^a^[Both exemplar for HAT - P-7b assume solar metallicity. ] { } ^b^[Based on a $ 2\sigma$ upper limit of the minimum magnetic field ratio. ] { } ^c^[Difference between maximum and minimal flux ratios. ] { } ^d^[The maximum and minimum magnetic field offsets are measured relative to the center of secondary eclipse clock time and center of transit time, respectively. minus time offsets indicate an eastward transformation in the location of the hot or cold region in the planet ’s atmosphere. ] { } After subtract and dividing out the brightness of the star alone, i.e., the flux measured during secondary eclipse, we obtain the phase variation of the relative planet – star flux ratio (Figures   \[lightcurve1\ ] and \[lightcurve2\ ]). Tables   \[tab: comparison1\ ] and \[tab: comparison2\ ] list the significant quantitative characteristics of the WASP-19b and HAT - P-7b phase curves, respectively. By probe the place of these phase curves, we obtain a basic picture of the temperature distribution across the planets ’ surfaces, from which we can begin to infer expression of the planets ’ energy

mofel^a^ & $0.131$ & $0.150$\ \ Measured & $<0.030$^b^ & $0.079\pm0.019$\ Wlth TiO/VO model & $0.039$ & $0.042$\ No TmO/VO mosel & $0.044$ & $0.045$\ \ Mdasured & $>0.128$^b^ & $0.116^{+0.018}_{-0.019}$\ With TiO/VO model & $0.121$ & $0.151$\ No UpO/VO model &$0.089$ & $0.105$\ \ Measured & $1.0\pm 1.1$ & $0.6\pm 1.1$\ With TiI/VO nidel & $-1.1$ & $-1.1$\ No TiO/VO model & $-3.2$ & $-3.2$\ \ Meavnred & $1.1^{+1.5}_{-1.6}$ & $0.6^{+1.5}_{-1.7}$\ With TiP/VO model & $-3.2$ & $-2.1$\ No TiO/VO mogeu & $-5.3$ & $-4.2$\ [**Notes.**]{} ^a^[Both models for HAT-P-7b assuie solat letallicity.]{} ^b^[Bafed pg a $2\apgna$ upper limit of the minimum flux retio.]{} ^c^[Difference between maximum and minimkm fpux ratios.]{} ^d^[The maxlmum and mibimui flux offsetr are measlxed relativg to the center of secondary eclkpse cime and cebtwr lx transit tmme, refpectively. Ncbative time ogfsets indicatc an xastqard shift in the locetion of the hot or sold regimn in the planet’s armisphete.]{} Aftar sjvtrxctjnj ahd divldiig out the grightness if the star alone, i.t., trv flux measursd durynd secondary eclipse, we obtain the phase vadiation of the relative planet–star flux ratil (Figures \[jightcurve1\] and \[lightcurve2\]). Tables \[tab:comparison1\] and \[tab:ckopaxlfub2\] pist the important quantitative characteristisa pf the WASP-19b and HAT-P-7b phasr futdes, respectivgly. By zsajining the propertles of jhese phase curdes, ee obtain a basic picture od the temperctuee distribution acxoss the plauets’ sorfacex, from which we can begnn to jnfer aspechs of the olanets’ energy

model^a^ & $0.131$ & $0.150$\ \ Measured & With TiO/VO & $0.039$ & $0.044$ $0.045$\ \ Measured $>0.128$^b^ & $0.116^{+0.018}_{-0.019}$\ TiO/VO model & $0.121$ & $0.151$\ TiO/VO model &$0.089$ & $0.105$\ \ Measured & $1.0\pm 1.1$ & $0.6\pm 1.1$\ TiO/VO model & $-1.1$ & $-1.1$\ No TiO/VO model & $-3.2$ & $-3.2$\ Measured $1.1^{+1.5}_{-1.6}$ $0.6^{+1.5}_{-1.7}$\ TiO/VO model & $-3.2$ & $-2.1$\ No TiO/VO model & $-5.3$ & $-4.2$\ [**Notes.**]{} ^a^[Both models HAT-P-7b assume solar metallicity.]{} ^b^[Based on a $2\sigma$ limit of the minimum ratio.]{} ^c^[Difference between maximum and flux ^d^[The maximum minimum offsets measured relative to center of secondary eclipse time and center of transit time, respectively. Negative time offsets indicate an eastward in the the hot cold in planet’s atmosphere.]{} After dividing out the brightness of the the flux measured during secondary eclipse, we obtain phase variation the relative planet–star flux ratio (Figures and \[lightcurve2\]). Tables \[tab:comparison1\] and \[tab:comparison2\] list the quantitative characteristics of the WASP-19b and HAT-P-7b phase curves, respectively. By examining the properties of curves, we obtain a picture of the distribution the surfaces, which we begin to infer aspects of the planets’ energy

model^a^ & $0.131$ & $0.150$\ \ Measured & $<0.030$^b^ & $0.079\pm0.019$\ With TiO/Vo model & $0.039$ & $0.042$\ No Tio/VO moDel & $0.044$ & $0.045$\ \ meaSuRed & $>0.128$^b^ & $0.116^{+0.018}_{-0.019}$\ with tiO/VO model & $0.121$ & $0.151$\ No Tio/vO moDel &$0.089$ & $0.105$\ \ Measured & $1.0\pm 1.1$ & $0.6\pm 1.1$\ With TiO/vO modEl & $-1.1$ & $-1.1$\ nO TiO/vo mOdel & $-3.2$ & $-3.2$\ \ MEasured & $1.1^{+1.5}_{-1.6}$ & $0.6^{+1.5}_{-1.7}$\ wItH tIO/Vo mOdEl & $-3.2$ & $-2.1$\ NO TIo/Vo modeL & $-5.3$ & $-4.2$\ [**NoTes.**]{} ^a^[BotH models for hAT-p-7b Assume solar mETaLlicity.]{} ^b^[BaSed On a $2\sigma$ uppeR liMit of tHe MinIMum flUx rAtio.]{} ^c^[differENce betWeen maximUm ANd miniMUm flux rATIoS.]{} ^d^[ThE maximum and minimuM FlUX offsets are meaSured rElATiVE To tHe cEnter of secOnDary eCLipse tiME aND CEntER of transit timE, respectiveLY. NeGative TiMe oFFsets iNdicaTe AN eaStward shift In thE location Of the hOT or cold REgion in The plaNet’S atMospHErE.]{} AFteR sUBtrACtIng ANd dIviding oUt ThE brigHtneSS OF The sTar AlonE, i.e., thE flux measured DurIng sEConDary eClipsE, we oBtAin thE phase VariaTiOn of the relative PlanEt–star fluX raTiO (FiGuRes \[liGHtcurvE1\] anD \[liGhtcurvE2\]). Tables \[TAb:cOmPARIsOn1\] and \[tab:comparison2\] LiST ThE importaNt quanTItAtIVe characTeRisTics OF The WAsP-19b aND HaT-P-7b phasE curveS, ReSpEctivelY. BY examiNiNg tHe pRoperTIes oF these Phase curVes, we OBtain a basic picTUre of the tempeRAtURE dIStriButIon across thE plaNEts’ sUrfaCEs, FroM Which We can BeGIn TO infer aspects of the pLaNets’ enErgy

model^a^ & $0.131$ & $0.1 50$\ \ Mea sured &$<0 .0 30$^ b^ & $0.079\pm0.01 9 $\ W ith TiO/VO model & $0. 039$&$ 0.04 2 $\ No T iO/VO m o de l & $ 0. 04 4$&$ 0. 045$\ \Measure d & $>0.12 8$^ b^ & $0.116^{+ 0 .0 18}_{-0.01 9}$ \ With TiO/V O m odel & $ 0.1 2 1$ &$0. 151$\ No Ti O /VO mo del &$0.0 89 $ & $0. 1 05$\ \M e as ured & $1.0\pm 1.1$ & $0 . 6\pm 1.1$\ Wit h TiO/ VO mo d e l & $- 1.1$ & $-1 .1 $\ No TiO/VOm od e l & $ - 3.2$ & $-3.2$ \ \ Measure d &$1.1^{ +1 .5} _ {-1.6} $ & $ 0. 6 ^{+ 1.5}_{-1.7} $\ W ith TiO/V O mode l & $-3. 2 $ & $-2 .1$\ N o T iO/ VO m o de l& $ -5 . 3$& $ -4. 2 $\ [**Note s. ** ]{} ^a^[ B o t h mod els for HAT- P-7b assume s ola r me t all icity .]{} ^b^ [B asedon a $ 2\sig ma $ upper limit o f th e minimum fl ux ra ti o.]{} ^c^[D iff ere nce bet ween ma x imu ma n d m inimum flux ratios .] { } ^d^[Themaximu m a nd minimumfl uxoffs e t s are mea s ur ed relat ive to th ecenterof secon da ryecl ipset imeand ce nter oftrans i t time, respec t ively. Negati v et i me offs ets indicate a n ea s twar d sh i ft in the l ocati on of the hot or cold reg io n in t he pl anet’s atmosp here.]{} A f t er subtr acti n ga nd dividing ou t the brightnes s of thestaralone, i .e., thef l ux measu red du rin g s e c on dary eclipse, w e ob ta in thepha se vari ati onofthe r elative p lanet–st ar f lu xrat io (F i gures \[ li ght cu rve 1\] a n d \[li ghtcu rve2 \] ). Tab les \[t a b: c o mpar is on 1\]and \ [tab: comp a ris on2\] l ist the i mpo r tant q ua ntitati ve characteri st ics of the W ASP -19b a n d HAT-P-7 b phase curves, respect i vely. B y e xamin ingthe prope rti es ofthe s e phas e curv es, w eobt a i n a b a s ic pi ct ure of the t emp eratu re dis tributi on across the plan e ts’ surfaces, fr omwhic h we ca n b e gin t o in f e r aspects of th e planets’ e n er gy

model^a^_& $0.131$_& $0.150$\ \ Measured & $<0.030$^b^_& $0.079\pm0.019$\ With_TiO/VO_model &_$0.039$_& $0.042$\ No TiO/VO_model & $0.044$_& $0.045$\ \ Measured & $>0.128$^b^_& $0.116^{+0.018}_{-0.019}$\ With TiO/VO_model_& $0.121$ & $0.151$\ No TiO/VO model &$0.089$ & $0.105$\ \ Measured & $1.0\pm 1.1$ & $0.6\pm_1.1$\ With_TiO/VO model_&_$-1.1$_& $-1.1$\ No TiO/VO model &_$-3.2$ & $-3.2$\ \ Measured & $1.1^{+1.5}_{-1.6}$_& $0.6^{+1.5}_{-1.7}$\ With_TiO/VO model & $-3.2$ & $-2.1$\ No TiO/VO model_&_$-5.3$ & $-4.2$\ [**Notes.**]{} ^a^[Both_models for HAT-P-7b assume solar metallicity.]{} ^b^[Based on a $2\sigma$_upper limit of the minimum flux_ratio.]{} ^c^[Difference between maximum_and_minimum_flux ratios.]{} ^d^[The maximum and_minimum flux offsets are measured relative_to the center of secondary eclipse_time and center of transit time, respectively._Negative time offsets indicate an eastward_shift in the location of_the hot_or cold region in the_planet’s atmosphere.]{} After subtracting_and dividing_out the brightness_of the star alone, i.e., the_flux measured during_secondary eclipse, we obtain the phase_variation_of the relative_planet–star_flux_ratio (Figures \[lightcurve1\]_and \[lightcurve2\]). Tables \[tab:comparison1\]_and_\[tab:comparison2\] list_the_important quantitative characteristics of the WASP-19b_and_HAT-P-7b phase curves, respectively. By examining the_properties of these phase_curves,_we obtain a basic_picture of the temperature distribution_across the planets’ surfaces, from which_we can_begin to_infer aspects of the planets’ energy

cohenmoebius} c_q^{(\beta )}(n) = \sum_{\substack{d|q \\ {d^\beta }|n}} {\mu \left( {\frac{q}{d}} \right){d^\beta }} \nonumber\end{aligned}$$ where $\mu$ denotes the Möbius function. \[lemma23\] For ${\operatorname{Re}}(s)>1$ and $\beta \in {\mathbb{N}}$ one has $$\sum_{q = 1}^\infty {\frac{{c_q^{(\beta )}(n)}}{{{q^{\beta s}}}}} = \frac{{\sigma _{1 - s }^{(\beta )}(n)}}{{\zeta (\beta s)}}.$$ From Lemma \[lemma22\] one has the following bound $$|c_q^{(\beta )}(n)| \leqslant \sum_{\substack{d|q \\ {d^\beta }|n}} {{d^\beta }} \leqslant \sum_{{d^\beta }|n} {{d^\beta }} = \sigma_1^{(\beta )}(n).$$ Suppose $x$ is a fixed non-integer. Let us now consider the positively oriented path $\mathcal{C}$ made up of the line segments $[c-iT,c+iT,-2N-1+iT,-2N-1-iT]$ where $T$ is not the ordinate of a non-trivial zero. We set $a_q = c_q^{(\beta)}(n)$ and we use the lemma in $\mathsection$3.12 of [@titchmarsh] to see that we can take $\psi(q) = \sigma_1^{(\beta)}(n)$. We note that for $\sigma > 1$ we have $$\sum_{q = 1}^\infty {\frac{{|c_q^{(\beta )}(n)|}}{{{q^\sigma }}}} \leqslant \sigma_1^{(\beta)} (n)\sum_{q = 1}^\infty {\frac{1}{{{q^\sigma }}}} = \sigma_1^{(\beta)} (n)\zeta (\sigma ) \ll {\frac{1}{{\sigma - 1}}}$$ so that $\alpha=1$. Moreover, if in that lemma we put $s=0$, $c=1+1/\log x$ and $w$ replaced by $s$, then we obtain $$\mathfrak{C}_0^{(\beta )}(n

cohenmoebius } c_q^{(\beta) } (n) = \sum_{\substack{d|q \\ { d^\beta } |n } } { \mu \left ({ \frac{q}{d } } \right){d^\beta } } \nonumber\end{aligned}$$ where $ \mu$ denotes the Möbius function. \[lemma23\ ] For $ { \operatorname{Re}}(s)>1 $ and $ \beta \in { \mathbb{N}}$ one has $ $ \sum_{q = 1}^\infty { \frac{{c_q^{(\beta) } (n)}}{{{q^{\beta s } } } } } = \frac{{\sigma _ { 1 - s } ^{(\beta) } (n)}}{{\zeta (\beta s)}}.$$ From Lemma \[lemma22\ ] one has the follow tie $ $ |c_q^{(\beta) } (n)| \leqslant \sum_{\substack{d|q \\ { d^\beta } |n } } { { d^\beta } } \leqslant \sum_{{d^\beta } |n } { { d^\beta } } = \sigma_1^{(\beta) } (n).$$ Suppose $ x$ is a fixed non - integer. Let us immediately consider the positively oriented way $ \mathcal{C}$ made up of the line segments $ [ c - iT, c+iT,-2N-1+iT,-2N-1 - iT]$ where $ T$ is not the ordinate of a non - fiddling zero. We set $ a_q = c_q^{(\beta)}(n)$ and we use the lemma in $ \mathsection$3.12 of [ @titchmarsh ] to experience that we can take $ \psi(q) = \sigma_1^{(\beta)}(n)$. We note that for $ \sigma > 1 $ we get $ $ \sum_{q = 1}^\infty { \frac{{|c_q^{(\beta) } (n)|}}{{{q^\sigma } } } } \leqslant \sigma_1^{(\beta) } (n)\sum_{q = 1}^\infty { \frac{1}{{{q^\sigma } } } } = \sigma_1^{(\beta) } (n)\zeta (\sigma) \ll { \frac{1}{{\sigma - 1}}}$$ so that $ \alpha=1$. Moreover, if in that lemma we put $ s=0 $, $ c=1 + 1/\log x$ and $ w$ replaced by $ s$, then we obtain $ $ \mathfrak{C}_0^{(\beta) } (n

cohfnmoebius} c_q^{(\beta )}(n) = \sum_{\sunstack{d|q \\ {d^\beta }|u}} {\mu \leht( {\frac{s}{d}} \right){a^\beta }} \nonumber\end{aligned}$$ whxre $\nu$ debotes the Möbius functiun. \[lemma23\] Flr ${\operarorneme{Re}}(s)>1$ and $\beta \mh {\mathbn{U}}$ one mas $$\sbm_{w = 1}^\infty {\frac{{g_q^{(\beta )}(n)}}{{{q^{\beda s}}}}} = \frac{{\sigka _{1 - s }^{(\beta )}(n)}}{{\zeta (\beta s)}}.$$ From Lemma \[lemmw22\] one hss the following boumq $$|c_q^{(\gvtc )}(n)| \leqslant \sum_{\substack{d|q \\ {d^\befa }|n}} {{d^\btta }} \leqslant \sum_{{c^\beta }|n} {{d^\beta }} = \sigma_1^{(\beta )}(n).$$ Skppose $x$ is a fixef non-integet. Lee us now conskder the positively orjented path $\mathcal{C}$ made up of the kine segmebtw $[c-lJ,c+iT,-2N-1+iT,-2N-1-iT]$ wiere $T$ is not the ordinate of a npn-trivial zero. We ser $a_q = c_q^{(\beta)}(n)$ and we nse the lemma in $\matrsection$3.12 mf [@titchmarsh] to sew rhat fe cdn txje $\osi(s) = \sjgma_1^{(\beha)}(n)$. We note tgat for $\signa > 1$ we have $$\sum_{q = 1}^\onsnu {\frac{{|c_q^{(\betz )}(n)|}}{{{q^\sidmw }}}} \leqslant \sigma_1^{(\beta)} (n)\sum_{q = 1}^\infty {\frdc{1}{{{q^\aigma }}}} = \sigma_1^{(\beta)} (n)\zetq (\sigma ) \ll {\frac{1}{{\sigma - 1}}}$$ so thwt $\alpha=1$. Moreover, if in that lemma we put $s=0$, $c=1+1/\log f$ and $d$ rtpoaced vy $s$, then we obtain $$\mathfrak{C}_0^{(\beta )}(n

cohenmoebius} c_q^{(\beta )}(n) = \sum_{\substack{d|q \\ {d^\beta \left( \right){d^\beta }} where $\mu$ denotes ${\operatorname{Re}}(s)>1$ $\beta \in {\mathbb{N}}$ has $$\sum_{q = {\frac{{c_q^{(\beta )}(n)}}{{{q^{\beta s}}}}} = \frac{{\sigma _{1 s }^{(\beta )}(n)}}{{\zeta (\beta s)}}.$$ From Lemma \[lemma22\] one has the following bound )}(n)| \leqslant \sum_{\substack{d|q \\ {d^\beta }|n}} {{d^\beta }} \leqslant \sum_{{d^\beta }|n} {{d^\beta }} \sigma_1^{(\beta Suppose is fixed non-integer. Let us now consider the positively oriented path $\mathcal{C}$ made up of the line $[c-iT,c+iT,-2N-1+iT,-2N-1-iT]$ where $T$ is not the ordinate of non-trivial zero. We set = c_q^{(\beta)}(n)$ and we use lemma $\mathsection$3.12 of to that can take $\psi(q) \sigma_1^{(\beta)}(n)$. We note that for $\sigma > 1$ we have $$\sum_{q = 1}^\infty {\frac{{|c_q^{(\beta )}(n)|}}{{{q^\sigma }}}} \leqslant (n)\sum_{q = }}}} = (n)\zeta ) {\frac{1}{{\sigma - 1}}}$$ $\alpha=1$. Moreover, if in that lemma $c=1+1/\log x$ and $w$ replaced by $s$, then obtain $$\mathfrak{C}_0^{(\beta

cohenmoebius} c_q^{(\beta )}(n) = \sum_{\subStack{d|q \\ {d^\beTa }|n}} {\mu \LefT( {\frAc{Q}{d}} \riGht){d^\Beta }} \nonumber\enD{AligNed}$$ where $\mu$ denotes the MöBius fUnCTion. \[LEmMa23\] For ${\OperatoRNaME{re}}(s)>1$ AnD $\bEta \In {\MAtHbb{N}}$ oNe hAs $$\sum_{q = 1}^\iNfty {\frac{{c_q^{(\BetA )}(n)}}{{{Q^{\beta s}}}}} = \frac{{\siGMa _{1 - S }^{(\beta )}(n)}}{{\zeta (\BetA s)}}.$$ From Lemma \[lEmmA22\] one haS tHe fOLlowiNg bOund $$|c_Q^{(\beta )}(n)| \LEqslanT \sum_{\substAcK{D|q \\ {d^\betA }|N}} {{d^\beta }} \lEQSlAnt \sUm_{{d^\beta }|n} {{d^\beta }} = \sigmA_1^{(\BeTA )}(n).$$ Suppose $x$ is a fIxed noN-iNTeGER. LeT us Now consideR tHe posITively oRIeNTED paTH $\mathcal{C}$ made Up of the line SEgmEnts $[c-it,c+IT,-2N-1+It,-2N-1-iT]$ whEre $T$ iS nOT thE ordinate of A non-Trivial zeRo. We seT $A_q = c_q^{(\betA)}(N)$ and we uSe the lEmmA in $\MathSEcTiOn$3.12 oF [@tITchMArSh] tO See That we caN tAkE $\psi(q) = \SigmA_1^{(\BETA)}(n)$. We NotE thaT for $\sIgma > 1$ we have $$\sum_{Q = 1}^\inFty {\fRAc{{|c_Q^{(\beta )}(N)|}}{{{q^\sigMa }}}} \leQsLant \sIgma_1^{(\beTa)} (n)\suM_{q = 1}^\Infty {\frac{1}{{{q^\sigma }}}} = \SigmA_1^{(\beta)} (n)\zetA (\siGmA ) \ll {\FrAc{1}{{\sigMA - 1}}}$$ so thaT $\alPha=1$. moreoveR, if in thAT leMmA WE PuT $s=0$, $c=1+1/\log x$ and $w$ replaceD bY $S$, ThEn we obtaIn $$\mathFRaK{C}_0^{(\BEta )}(n

cohenmoebius} c_q^{(\beta)}(n) = \s um_{\ sub sta ck {d|q \\{d^\beta }|n}} {\mu \left( {\frac{q}{d}}\righ t) { d^\b e ta }} \ nonumbe r \e n d {al ig ne d}$ $w he re $\ mu$ denote s the Möbi usfu nction. \[l e mm a23\] For${\ operatorname {Re }}(s)> 1$ an d $\be ta\in { \mathb b {N}}$one has $ $\ s um_{q= 1}^\in f t y {\f rac{{c_q^{(\beta) }( n )}}{{{q^{\beta s}}}} }=\ f rac {{\ sigma _{1-s }^{ ( \beta ) } (n ) } } {{\ z eta (\beta s) }}.$$ From Lem ma \[l em ma2 2 \] one hasth e fo llowing bou nd $ $|c_q^{(\ beta ) } (n)| \l e qslant\sum_{ \su bst ack{ d |q \ \ { d^ \ bet a } |n} } {{ d^\beta}} \leqs lant \ s u m_{{ d^\ beta }|n} {{d^\beta }} = \si g ma_ 1^{(\ beta)}(n ). $$ Su ppose$x$ i sa fixed non-int eger . Let usnow c ons id er th e posit ive lyoriente d path$ \ma th c a l {C }$ made up of theli n e s egments$[c-iT , c+ iT , -2N-1+iT ,- 2N- 1-iT ] $ wher e $T $ i s not th e ordi n at eof a no n- trivia lzer o.We se t $a_ q = c_ q^{(\bet a)}(n ) $ and we use t h e lemma in $\ m at h s ec t ion$ 3.1 2 of [@titc hmar s h] t o se e t hat we ca n tak e$ \p s i(q) = \sigma_1^{(\ be ta)}(n )$. W e note that f or $\sigma > 1$ we ha ve $ $ \s u m_{q = 1}^\inf ty { \frac{{|c_ q ^{(\beta )}(n )|}}{{{q ^\sigma } } } } \leqs lan t \ sig ma_ 1 ^ {( \beta)} (n)\s u m _{q=1}^\inf ty {\frac {1} {{{ q^\ sig ma }}}} =\sigma_1 ^{ (\ be ta )}(n)\z e ta (\sig ma )\l l { \frac { 1}{{\s igma - 1 }} }$ $ so that $ \ al p h a=1$ .Mo reov er, i f inthat lem ma we p ut $s=0$, $c = 1+1/ \l og x$ and $w$ replaced b y $s$, the nweobtain $ $\mathfr ak{C}_0^{(\beta )}(n

cohenmoebius} c_q^{(\beta )}(n)_= \sum_{\substack{d|q_\\ {d^\beta }|n}} {\mu_\left( {\frac{q}{d}}_\right){d^\beta_}} \nonumber\end{aligned}$$_where_$\mu$ denotes the_Möbius function. \[lemma23\] For_${\operatorname{Re}}(s)>1$ and $\beta \in_{\mathbb{N}}$ one has_$$\sum_{q_= 1}^\infty {\frac{{c_q^{(\beta )}(n)}}{{{q^{\beta s}}}}} = \frac{{\sigma _{1 - s }^{(\beta )}(n)}}{{\zeta_(\beta_s)}}.$$ From Lemma_\[lemma22\]_one_has the following bound $$|c_q^{(\beta_)}(n)| \leqslant \sum_{\substack{d|q \\ {d^\beta_}|n}} {{d^\beta_}} \leqslant \sum_{{d^\beta }|n} {{d^\beta }} _=_\sigma_1^{(\beta )}(n).$$ Suppose_$x$ is a fixed non-integer. Let us now consider_the positively oriented path $\mathcal{C}$ made_up of the_line_segments_$[c-iT,c+iT,-2N-1+iT,-2N-1-iT]$ where $T$ is_not the ordinate of a non-trivial_zero. We set $a_q = c_q^{(\beta)}(n)$_and we use the lemma in $\mathsection$3.12_of [@titchmarsh] to see that we_can take $\psi(q) = \sigma_1^{(\beta)}(n)$._We note_that for $\sigma > 1$_we have $$\sum_{q_= 1}^\infty_ {\frac{{|c_q^{(\beta )}(n)|}}{{{q^\sigma_}}}} \leqslant \sigma_1^{(\beta)} (n)\sum_{q =_1}^\infty {\frac{1}{{{q^\sigma_}}}} = \sigma_1^{(\beta)} (n)\zeta (\sigma_)_\ll {\frac{1}{{\sigma _-_1}}}$$_so that_$\alpha=1$. Moreover, if_in_that lemma_we_put $s=0$, $c=1+1/\log x$ and $w$_replaced_by $s$, then we obtain $$\mathfrak{C}_0^{(\beta )}(n

ica D [**168**]{} 244 H Frauenkron, U Bastolla, E Gerstner, P Grassberger, and W Nadler 1998 Phys. Rev. Lett. [**80**]{} 3149 P Le Doussal and J Machta 1991 J. Stat. Phys. [**64**]{} 541 B K Chakrabarti, K Bhadra, A K Roy, and S N Karmakar 1983 Phys. Lett. [**93A**]{} 434 A J Guttmann and J Wang 1991 J. Phys. A [**24**]{} 3107 J W Lyklema and K Kremer 1984 Z. Phys. B [**55**]{} 41 B Nienhuis 1984 J. Stat. Phys. [**34**]{} 731 P G de Gennes 1976 La Recherche [**7**]{} 919 D H N Anh, K H Hoffmann, S Seeger, and S Tarafdar 2005 Europhys Lett [**70**]{} 109; D H N Anh, P Blaudeck, K H Hoffmann, J Prehl, and S Tarafdar 2007 J. Phys. A [**40**]{} 11453 **Orbit functions of ${\rm SU}(n)$ and Chebyshev polynomials** *Maryna NESTERENKO $^\dag$, Jiří PATERA $^\ddag$ and Agnieszka TERESZKIEWICZ $^\S$* $^\dag$ Institute of Mathematics of NAS of Ukraine, 3 Tereshchenkivs’ka Str., Kyiv-4, 01601 Ukraine $^\ddag$ Centre de recherches mathématiques, Université de Montréal, C.P.6128-Centre ville, Montréal,\  H3C3J7, Québec, Canada $^\S$ Institute of Mathematics, University of Bialystok, Akademicka 2, PL-15-267 Bialystok, Poland $\phantom{^\S}$ E-mail: maryna@imath.kiev.ua, patera@crm.umontreal.ca, a.tereszkiewicz@uwb.edu.pl Introduction {#Introduction} ============ The history of the Chebyshev polynomials dates back over a century.

ica D [ * * 168 * * ] { } 244 H Frauenkron, U Bastolla, E Gerstner, P Grassberger, and W Nadler 1998 Phys. Rev. Lett. [ * * 80 * * ] { } 3149 P Le Doussal and J Machta 1991 J. Stat. Phys. [ * * 64 * * ] { } 541 B K Chakrabarti, K Bhadra, A K Roy, and S N Karmakar 1983 Phys. Lett. [ * * 93A * * ] { } 434 A J Guttmann and J Wang 1991 J. Phys. A [ * * 24 * * ] { } 3107 J W Lyklema and K Kremer 1984 Z. Phys. B [ * * 55 * * ] { } 41 B Nienhuis 1984 J. Stat. Phys. [ * * 34 * * ] { } 731 P G de Gennes 1976 La Recherche [ * * 7 * * ] { } 919 D H N Anh, K H Hoffmann, S Seeger, and S Tarafdar 2005 Europhys Lett [ * * 70 * * ] { } 109; D H N Anh, P Blaudeck, K H Hoffmann, J Prehl, and S Tarafdar 2007 J. Phys. A [ * * 40 * * ] { } 11453 * * eye socket function of $ { \rm SU}(n)$ and Chebyshev polynomials * * * Maryna NESTERENKO   $ ^\dag$, Jiří PATERA   $ ^\ddag$ and Agnieszka TERESZKIEWICZ   $ ^\S$ * $ ^\dag$   Institute of Mathematics of NAS of Ukraine, 3 Tereshchenkivs’ka Str. , Kyiv-4, 01601 Ukraine $ ^\ddag$   Centre de recherches mathématiques, Université de Montréal, C.P.6128 - Centre ville, Montréal,\   H3C3J7,   Québec,   Canada $ ^\S$   Institute of Mathematics, University of Bialystok, Akademicka 2, PL-15 - 267 Bialystok, Poland $ \phantom{^\S}$   E - mail: maryna@imath.kiev.ua, patera@crm.umontreal.ca, a.tereszkiewicz@uwb.edu.pl Introduction { # insertion } = = = = = = = = = = = = The history of the Chebyshev polynomials dates back over a hundred.

ica D [**168**]{} 244 H Frauenkron, U Bastoula, E Gerstner, K Trassbxrger, ahd W Naduer 1998 Phys. Rev. Lett. [**80**]{} 3149 P Le Doudsql ane J Machta 1991 J. Stat. Phyr. [**64**]{} 541 B K Chwkrabartu, K Uhadra, A K Roy, ais S N Karmakad 1983 Phvs. Lett. [**93A**]{} 434 A J Gujtmann and J Wang 1991 J. Phys. D [**24**]{} 3107 J W Lyklema and K Kremer 1984 Z. Phys. B [**55**]{} 41 F Nienhiid 1984 J. Stat. Phys. [**34**]{} 731 P B de Fvnkes 1976 La Recherche [**7**]{} 919 D H N Anh, K H Hoffmain, S Seeger, and S Tarafdar 2005 Europhys Lett [**70**]{} 109; D H N Anh, P Blaudecn, K H Hoffmqnn, T Prehl, and S Tarafdar 2007 J. Phys. A [**40**]{} 11453 **Krbit functions of ${\rm SU}(n)$ and Cfebyskev polynomuaos** *Mwtyna NESTEREIKO $^\dag$, Jiří PATERA $^\ddag$ and Acnieszks TERESZKIEWICD $^\S$* $^\dag$ Mnstutute of Mathematics mf NAS of Ukraine, 3 Tereshchaniivs’ka Str., Kyiv-4, 01601 Ykeaine $^\gdag$ Wentfw dd rtchxrcges mahhémetiques, Unibersité de Mintréal, C.P.6128-Centre vilke, Nontréal,\  H3C3J7, Québsc, Canaqa $^\F$ Institute of Mathematics, University of Bizlystok, Akademicka 2, PL-15-267 Vialystok, Poland $\phanjom{^\S}$ E-mail: iaryna@imath.kiev.ua, patera@crm.umontreal.ca, a.tereszkieficz@uxb.ddu.kl Lktroaychion {#Introduction} ============ The history of the Chebyshev lokykomials dates bagk over a century.

ica D [**168**]{} 244 H Frauenkron, U Gerstner, Grassberger, and Nadler 1998 Phys. Le and J Machta J. Stat. Phys. 541 B K Chakrabarti, K Bhadra, K Roy, and S N Karmakar 1983 Phys. Lett. [**93A**]{} 434 A J and J Wang 1991 J. Phys. A [**24**]{} 3107 J W Lyklema and Kremer Z. B 41 B Nienhuis 1984 J. Stat. Phys. [**34**]{} 731 P G de Gennes 1976 La Recherche 919 D H N Anh, K H Hoffmann, Seeger, and S Tarafdar Europhys Lett [**70**]{} 109; D N P Blaudeck, H J and S Tarafdar J. Phys. A [**40**]{} 11453 **Orbit functions of ${\rm SU}(n)$ and Chebyshev polynomials** *Maryna NESTERENKO $^\dag$, Jiří $^\ddag$ and $^\S$* $^\dag$ of of of Ukraine, 3 Kyiv-4, 01601 Ukraine $^\ddag$ Centre de de Montréal, C.P.6128-Centre ville, Montréal,\ H3C3J7, Québec, Canada Institute of University of Bialystok, Akademicka 2, PL-15-267 Poland $\phantom{^\S}$ E-mail: maryna@imath.kiev.ua, patera@crm.umontreal.ca, a.tereszkiewicz@uwb.edu.pl Introduction {#Introduction} The history of the Chebyshev polynomials dates back over a century.

ica D [**168**]{} 244 H Frauenkron, U Bastolla, E gerstner, P GRassbErgEr, aNd w NadLer 1998 PHys. Rev. Lett. [**80**]{} 3149 P Le DOUssaL and J Machta 1991 J. Stat. Phys. [**64**]{} 541 B K chakrAbARti, K bHaDra, A K roy, and S n kaRMAkaR 1983 PHyS. LeTt. [**93a**]{} 434 a J guttmAnn And J WanG 1991 J. Phys. A [**24**]{} 3107 J W LYklEmA and K Kremer 1984 Z. pHyS. B [**55**]{} 41 B NienhuiS 1984 J. STat. Phys. [**34**]{} 731 P G de GEnnEs 1976 La ReChErcHE [**7**]{} 919 D H N ANh, K h HoffMann, S SEEger, anD S TarafdaR 2005 EURophys lEtt [**70**]{} 109; D H N ANH, p BLaudEck, K H Hoffmann, J PreHL, aND S Tarafdar 2007 J. PhyS. A [**40**]{} 11453 **OrbiT fUNcTIOns Of ${\rM SU}(n)$ and CheByShev pOLynomiaLS** *MARYNa NesTERENKO $^\dag$, JiŘí PATERA $^\ddaG$ And agniesZkA TEreSZKIEwICZ $^\S$* $^\DaG$ insTitute of MatHemaTics of NAS Of UkraINe, 3 TeresHChenkivS’ka Str., kyiV-4, 01601 UkRainE $^\DdAg$ cenTrE De rEChErcHEs mAthématiQuEs, univeRsitÉ DE mOntrÉal, c.P.6128-CeNtre vIlle, Montréal,\  H3c3J7, QUébeC, canAda $^\S$ INstitUte oF MAthemAtics, UNiverSiTy of Bialystok, AkAdemIcka 2, PL-15-267 BiaLysToK, PoLaNd $\phaNTom{^\S}$ E-mAil: MarYna@imatH.kiev.ua, PAteRa@CRM.UmOntreal.ca, a.tereszkiEwICZ@uWb.edu.pl INtroduCTiOn {#iNtroductIoN} ============ ThE hisTORy of tHe ChEByShev polyNomialS DaTeS back ovEr A centuRy.

ica D [**168**]{} 244 H F rauenkron, U Ba sto lla ,E Ge rstn er, P Grassber g er,and W Nadler 1998 Phys . Rev .L ett. [* *80** ]{} 314 9 P LeDo us sal a n dJ Mac hta 1991 J . Stat. Ph ys. [ **64**]{} 54 1 B K Chakra bar ti, K Bhadra , A K Roy ,and S N K arm akar1983 P h ys. Le tt. [**93 A* * ]{} 43 4 A J G u t tm annand J Wang 1991 J . P h ys. A [**24**] {} 310 7JW Lyk lem a and K Kr em er 19 8 4 Z. Ph y s. B [** 5 5**]{} 41 BNienhuis 19 8 4 J . Stat .Phy s . [**3 4**]{ }7 31 P G de Gen nes1976 La R echerc h e [**7* * ]{} 919 D HN A nh, K H Ho ff man n, S S e eg er, and S Taraf da r2005Euro p h y s Let t [ **70 **]{} 109; D H N A nh, P B l aud eck,K H H offm an n, JPrehl, andSTarafdar 2007 J . Ph ys. A [** 40* *] {}11 453 * *Orbit fu nct ions of ${\rmS U}( n) $ a nd Chebyshev polynom ia l s ** *Maryn a NEST E RE NK O  $^\dag$ ,Jiř í PA T E RA $^ \dda g $and Agni eszkaT ER ES ZKIEWIC Z$^\S$* $^\ dag $ Ins t itut e of M athemati cs of NAS of Ukraine , 3 Tereshchen k iv s ’ ka Str. , K yiv-4, 0160 1 Uk r aine $^ \ dd ag$ Centr e dere c he r ches mathématiques, U nivers ité d e Montréal, C .P.6128-Ce n t r e ville, Mon t ré a l,\  H3C3J7, Q uébec , Canada $ ^\S$ Ins titut e of Mat hematics, U niversit y o f B ial yst o k ,Akademicka 2, P L-15 -2 67 Bial yst ok, Pol and $\ pha nto m{ ^\S}$ E-m ail: mar yn a@ im at h.k iev.u a , patera @c rm. um ont real. c a, a.t eresz kiew ic z@ u wb. edu.plIn t r oduc ti on {#I ntr od uctio n} = = === ======= The his tor y ofth eChebysh ev polynomial sdates back o ver a cen t u ry.

ica D_[**168**]{} 244 H_Frauenkron, U Bastolla, E_Gerstner, P_Grassberger,_and W_Nadler_1998 Phys. Rev._Lett. [**80**]{} 3149 P_Le Doussal and J_Machta 1991 J._Stat._Phys. [**64**]{} 541 B K Chakrabarti, K Bhadra, A K Roy, and S N Karmakar_1983_Phys. Lett._[**93A**]{}_434 A_J Guttmann and J Wang_1991 J. Phys. A [**24**]{}_3107 J W_Lyklema and K Kremer 1984 Z. Phys. B_[**55**]{}_41 B Nienhuis 1984_J. Stat. Phys. [**34**]{} 731 P G de Gennes 1976_La Recherche [**7**]{} 919 D H N_Anh, K H_Hoffmann,_S_Seeger, and S Tarafdar_2005 Europhys Lett [**70**]{} 109; D_H N Anh, P Blaudeck, K_H Hoffmann, J Prehl, and S Tarafdar_2007 J. Phys. A [**40**]{} 11453 _**Orbit functions of ${\rm SU}(n)$_and Chebyshev_polynomials** *Maryna NESTERENKO $^\dag$, Jiří PATERA $^\ddag$ and_Agnieszka TERESZKIEWICZ $^\S$* $^\dag$ Institute of_Mathematics of_NAS of Ukraine,_3 Tereshchenkivs’ka Str., Kyiv-4, 01601 Ukraine $^\ddag$ Centre_de recherches mathématiques,_Université de Montréal, C.P.6128-Centre ville, Montréal,\  H3C3J7, Québec, Canada $^\S$ Institute_of_Mathematics, University of_Bialystok,_Akademicka_2, PL-15-267_Bialystok, Poland $\phantom{^\S}$ E-mail:_maryna@imath.kiev.ua,_patera@crm.umontreal.ca, a.tereszkiewicz@uwb.edu.pl Introduction_{#Introduction} ============ The_history of the Chebyshev polynomials dates_back_over a century.

prime element of the frame $I(RC(\mathbb{R}),\rho_{\mathbb{R}}, CR(\mathbb{R}))$ is a prime ideal of $CR(\mathbb{R})$. \[imnota\] Let $(A,\rho,\BBBB)$ and $(B,\eta,\BBBB\ap)$ be LCAs, $\p:A\lra B$ be a function and $\s\ap$ be a cluster in $(B,\eta,\BBBB\ap)$. Then we set: - $S_{\s\ap}=\{a\in\BBBB\st (\fa b\in A)[(a\llx b)\rightarrow(\p(b)\in\s\ap)]\}$; - $V_{\s\ap}=\{a\in\BBBB\st (\ex b\in S_{\s\ap})(b\llx a)\}$; - $J_{\s\ap}=\BBBB\stm S_{\s\ap}$. \[fss\] Let $(A,\rho,\BBBB)$ and $(B,\eta,\BBBB\ap)$ be LCAs and $\p:A\lra B$ be a function satisfying conditions (DLC1)-(DLC3). Then, for every cluster $\s\ap$ in $(B,\eta,\BBBB\ap)$, $\ssi= \{a\in\BBBB\st (\fa b\in A)[(a\llx b)\rightarrow((\p(b^*))^*\in\s\ap)]\}$. Let $a\in\ssi$, $b\in A$ and $a\llx b$. Then $\p(b)\in\s\ap$. Since, by Lemma \[pf1\](b), $\p(b)\le(\p(b^*))^*$, we get that $(\p(b^*))^*\in\s\ap$. Thus $a\in R=\{a\in\BBBB\st (\fa b\in A)[(a\llx b)\rightarrow((\p(b^*))^*\in\s\ap)]\}$. Conversely, let $a\in R$, $b\in A$ and $a\llx b$. Then, by (BC1), there exists $c\in\BBBB$ such that $a\llx c\llx b$. Since $a\

prime element of the frame $ I(RC(\mathbb{R}),\rho_{\mathbb{R } }, CR(\mathbb{R}))$ is a prime ideal of $ CR(\mathbb{R})$. \[imnota\ ] lease $ (A,\rho,\BBBB)$ and $ (B,\eta,\BBBB\ap)$ be LCAs, $ \p: A\lra B$ be a affair and $ \s\ap$ be a cluster in $ (B,\eta,\BBBB\ap)$. Then we set: - $ S_{\s\ap}=\{a\in\BBBB\st (\fa b\in A)[(a\llx b)\rightarrow(\p(b)\in\s\ap)]\}$; - $ V_{\s\ap}=\{a\in\BBBB\st (\ex b\in S_{\s\ap})(b\llx a)\}$; - $ J_{\s\ap}=\BBBB\stm S_{\s\ap}$. \[fss\ ] Let $ (A,\rho,\BBBB)$ and $ (B,\eta,\BBBB\ap)$ be LCAs and $ \p: A\lra B$ be a routine satisfying conditions (DLC1)-(DLC3). Then, for every bunch $ \s\ap$ in $ (B,\eta,\BBBB\ap)$, $ \ssi= \{a\in\BBBB\st (\fa b\in A)[(a\llx b)\rightarrow((\p(b^*))^*\in\s\ap)]\}$. Let $ a\in\ssi$, $ b\in A$ and $ a\llx b$. Then $ \p(b)\in\s\ap$. Since, by Lemma \[pf1\](b), $ \p(b)\le(\p(b^*))^*$, we get that $ (\p(b^*))^*\in\s\ap$. Thus $ a\in R=\{a\in\BBBB\st (\fa b\in A)[(a\llx b)\rightarrow((\p(b^*))^*\in\s\ap)]\}$. Conversely, permit $ a\in R$, $ b\in A$ and $ a\llx b$. Then, by (BC1), there exists $ c\in\BBBB$ such that $ a\llx c\llx b$. Since $ a\

prlme element of the frame $I(RC(\mathbb{R}),\rho_{\mcrhbb{R}}, RR(\mathbg{R}))$ is a orime ideal of $CR(\mathbb{R})$. \[imnova\] Lwt $(A,\ryo,\BBBB)$ and $(B,\eta,\BBBB\ap)$ ce LCAs, $\p:W\lra B$ bw a hunction and $\s\ap$ be a clmfter ln $(B,\eca,\UBBB\ap)$. Then we xet: - $S_{\s\ap}=\{d\in\BBBB\st (\fa b\hn A)[(c\llx b)\rightarrow(\p(b)\in\s\ap)]\}$; - $V_{\s\ap}=\{a\in\FBBB\st (\rx b\in S_{\s\ap})(b\llx w)\}$; - $K_{\f\ap}=\BGBB\stm S_{\s\ap}$. \[fss\] Let $(A,\rho,\BBBB)$ and $(B,\efa,\BBBB\ak)$ be LCAs and $\p:A\lrs B$ be a function satisfyinh cojditions (DLC1)-(DLC3). Thfn, for everi cltwter $\s\ap$ in $(C,\eta,\BBBB\ap)$, $\ssi= \{a\in\BBBB\sj (\fa b\in A)[(a\llx b)\rightarrow((\p(b^*))^*\in\s\ap)]\}$. Uet $a\nn\ssi$, $b\in A$ abd $w\nlx b$. Then $\'(b)\in\s\ai$. Since, by Lemma \[pf1\](b), $\[(b)\le(\p(b^*))^*$, ee get that $(\p(b^*))^*\ln\s\ap$. Thys $a\in R=\{a\in\BBBB\st (\fa u\in A)[(a\llx b)\rightarrow((\k(b^*))^*\in\s\ap)]\}$. Cotvzrsely, let $a\in R$, $b\in Q$ qnd $a\nlx t$. Thdb, bh (BD1), vhede exidts $c\in\BBBB$ shch that $a\lox c\llx b$. Since $a\

prime element of the frame $I(RC(\mathbb{R}),\rho_{\mathbb{R}}, CR(\mathbb{R}))$ prime of $CR(\mathbb{R})$. Let $(A,\rho,\BBBB)$ and be function and $\s\ap$ a cluster in Then we set: - $S_{\s\ap}=\{a\in\BBBB\st (\fa A)[(a\llx b)\rightarrow(\p(b)\in\s\ap)]\}$; - $V_{\s\ap}=\{a\in\BBBB\st (\ex b\in S_{\s\ap})(b\llx a)\}$; - $J_{\s\ap}=\BBBB\stm S_{\s\ap}$. \[fss\] Let and $(B,\eta,\BBBB\ap)$ be LCAs and $\p:A\lra B$ be a function satisfying conditions (DLC1)-(DLC3). for cluster in $\ssi= \{a\in\BBBB\st (\fa b\in A)[(a\llx b)\rightarrow((\p(b^*))^*\in\s\ap)]\}$. Let $a\in\ssi$, $b\in A$ and $a\llx b$. Then $\p(b)\in\s\ap$. Since, Lemma \[pf1\](b), $\p(b)\le(\p(b^*))^*$, we get that $(\p(b^*))^*\in\s\ap$. Thus R=\{a\in\BBBB\st (\fa b\in A)[(a\llx Conversely, let $a\in R$, $b\in and b$. Then, (BC1), exists such that $a\llx b$. Since $a\

prime element of the frame $I(RC(\Mathbb{R}),\rho_{\MathbB{R}}, Cr(\maThBb{R}))$ iS a prIme ideal of $CR(\maTHbb{R})$. \[Imnota\] Let $(A,\rho,\BBBB)$ and $(B,\eTa,\BBBb\aP)$ Be LCaS, $\p:a\lra B$ Be a funcTIoN ANd $\s\Ap$ Be A clUsTEr In $(B,\etA,\BBbB\ap)$. TheN we set: - $S_{\s\ap}=\{A\in\bBbB\st (\fa b\in A)[(a\lLX b)\Rightarrow(\P(b)\iN\s\ap)]\}$; - $V_{\s\ap}=\{a\in\BbBB\St (\ex b\iN S_{\S\ap})(B\Llx a)\}$; - $J_{\S\ap}=\bBBB\sTm S_{\s\ap}$. \[FSs\] Let $(A,\Rho,\BBBB)$ anD $(B,\ETa,\BBBB\AP)$ be LCAs AND $\p:a\lra b$ be a function satisFYiNG conditions (DLC1)-(dLC3). TheN, fOR eVERy cLusTer $\s\ap$ in $(B,\eTa,\bBBB\aP)$, $\Ssi= \{a\in\BbbB\ST (\FA b\iN a)[(a\llx b)\rightarRow((\p(b^*))^*\in\s\ap)]\}$. LET $a\iN\ssi$, $b\iN A$ And $A\Llx b$. ThEn $\p(b)\iN\s\AP$. SiNce, by Lemma \[pF1\](b), $\p(b)\Le(\p(b^*))^*$, we get That $(\p(b^*))^*\IN\s\ap$. ThuS $A\in R=\{a\in\bBBB\st (\Fa b\In A)[(A\llx B)\RiGhTarRoW((\P(b^*))^*\iN\S\aP)]\}$. CoNVerSely, let $a\In r$, $b\In A$ anD $a\llX B$. tHEn, by (bC1), tHere ExistS $c\in\BBBB$ such tHat $A\llx C\Llx B$. SincE $a\

prime element of the fram e $I(RC(\m athbb {R} ),\ rh o_{\ math bb{R}}, CR(\ma t hbb{ R}))$ is a prime ideal of $ CR ( \mat h bb {R})$ . \[im n ot a \ ] L et $ (A, \r h o, \BBBB )$and $(B ,\eta,\BBB B\a p) $ be LCAs, $ \ p: A\lra B$ b e a function an d $ \s\ap$ b e a clust erin $( B,\eta , \BBBB\ ap)$. The nw e set: - $S _ { \s \ap} =\{a\in\BBBB\st ( \ fa b\in A)[(a\llx b )\ r ig h t arr ow( \p(b)\in\s \a p)]\} $ ; - $ V_ { \ s \ap } =\{a\in\BBBB\ st (\ex b\i n S_ {\s\ap }) (b\ l lx a)\ }$; - $J _{\s\ap}=\B BBB\ stm S_{\s \ap}$. \[fss\ ] Let $( A,\rho ,\B BBB )$ a n d$( B,\ et a ,\B B BB \ap ) $ b e LCAs a nd $ \p:A\ lraB $ b e afun ctio n sat isfying condi tio ns ( D LC1 )-(DL C3).Then ,for e very c luste r$\s\ap$ in $(B, \eta ,\BBBB\ap )$, $ \ss i= \{a\ i n\BBBB \st (\ fa b\in A)[(a\ l lxb) \ r i gh tarrow((\p(b^*))^* \i n \ s\ ap)]\}$. Let$ a\ in \ ssi$, $b \i n A $ an d $a\ll x b$ . T hen $\p( b)\in\ s \a p$ . Since ,by Lem ma \[ pf1 \](b) , $\p (b)\le (\p(b^*) )^*$, we get that $( \ p(b^*))^*\in\ s \a p $ .T hus$a\ in R=\{a\in \BBB B \st(\fa b\ inA )[(a\ llx b )\ r ig h tarrow((\p(b^*))^*\ in \s\ap) ]\}$. Conversely,let $a\inR $ , $b\in A $ an d $ a \llx b$. Then, by ( BC1), ther e exists$c\in \BBBB$ s uch that$ a \llx c\l lxb$. Si nce $ a\

prime_element of_the frame $I(RC(\mathbb{R}),\rho_{\mathbb{R}}, CR(\mathbb{R}))$_is a_prime_ideal of_$CR(\mathbb{R})$. \[imnota\]_Let $(A,\rho,\BBBB)$ and_$(B,\eta,\BBBB\ap)$ be LCAs,_$\p:A\lra B$ be a_function and $\s\ap$_be_a cluster in $(B,\eta,\BBBB\ap)$. Then we set: - $S_{\s\ap}=\{a\in\BBBB\st (\fa b\in A)[(a\llx __ b)\rightarrow(\p(b)\in\s\ap)]\}$; -___$V_{\s\ap}=\{a\in\BBBB\st (\ex b\in S_{\s\ap})(b\llx a)\}$; -_ $J_{\s\ap}=\BBBB\stm S_{\s\ap}$. \[fss\] Let_$(A,\rho,\BBBB)$ and_$(B,\eta,\BBBB\ap)$ be LCAs and $\p:A\lra B$ be a function_satisfying_conditions (DLC1)-(DLC3). Then,_for every cluster $\s\ap$ in $(B,\eta,\BBBB\ap)$, $\ssi= \{a\in\BBBB\st (\fa b\in_A)[(a\llx b)\rightarrow((\p(b^*))^*\in\s\ap)]\}$. Let $a\in\ssi$, $b\in A$ and $a\llx_b$. Then $\p(b)\in\s\ap$._Since,_by_Lemma \[pf1\](b), $\p(b)\le(\p(b^*))^*$, we_get that $(\p(b^*))^*\in\s\ap$. Thus $a\in R=\{a\in\BBBB\st_(\fa b\in A)[(a\llx b)\rightarrow((\p(b^*))^*\in\s\ap)]\}$. Conversely, let $a\in_R$, $b\in A$ and $a\llx b$. Then,_by (BC1), there exists $c\in\BBBB$ such_that $a\llx c\llx b$. Since_$a\

is about 0.35 according to Fig. \[F6\]. ![The magnitude of the parameter $\Delta_\alpha$ vs the possible values of the parameter $\alpha$.[]{data-label="F6"}](fig6.pdf){width="0.45\columnwidth"} In what follows we will confine our consideration to the special case  matching the ideal Lévy flights. It enables us to convert from the stochastic process $\{\eta(t)\}$ to the effective stochastic process $\{\eta'(t)\}$ using transformation  and to reduce Eq.  to the stochastic equation $$\label{sec3:5} \frac{d\eta'}{dt}=-\alpha\operatorname{sign}(\eta')+\sqrt{2}\xi(t)$$ governing the analyzed Lévy flights in the equivalent way. Then, via the next transformation $\eta'\to \mathfrak{u}$ specified by the expressions $$\label{sec3:6} \eta' = \frac1{\alpha}\,\mathfrak{u}\,,\qquad t = \frac1{\alpha^2}\,\mathfrak{t}$$ involving also the transformation of the time scales $t\to \mathfrak{t}$ equation  is rewritten as follows $$\label{sec3:7} \frac{d\mathfrak{u}}{d\mathfrak{t}}=-\operatorname{sign}(\mathfrak{u})+\sqrt{2}\xi(\mathfrak{t})\,.$$ It is a parameter-free stochastic differential equation with additive white noise such that $$\label{sec3:8} \left< \xi(\mathfrak{t})\right> = 0\,, \qquad \left< \xi(\mathfrak{t}) \xi(\mathfrak{t}')\right> = \delta(\mathfrak{t}-\mathfrak{t}')\,.$$ In other words, we have constructed the stochastic process $\{\mathfrak{u}(\mathfrak{t})\}$ governed by Eq.  that can be treated as the desired *core process*. It is of the same form for all the types of one-dimensional Lévy flights, at least, Lévy flights described by models similar to Eq.  inside the region where the cut-off effects are not significant. The basic parameters characterizing the generated Lévy flights such as the exponent of the Lévy scaling law depend on the system parameters, in particular, the coefficient $\alpha$ via the transformation from $\mathfrak{u}(\mathfrak{t})$ to $v(t)=dx/dt$. Combining together

is about 0.35 according to Fig.   \[F6\ ]. ! [ The magnitude of the parameter $ \Delta_\alpha$ vs the potential value of the parameter $ \alpha$.[]{data - label="F6"}](fig6.pdf){width="0.45\columnwidth " } In what follows we will enclose our retainer to the special case   match the ideal Lévy flights. It enables us to change from the stochastic process $ \{\eta(t)\}$ to the effective stochastic process $ \{\eta'(t)\}$ use transformation   and to reduce Eq.   to the stochastic equation $ $ \label{sec3:5 } \frac{d\eta'}{dt}=-\alpha\operatorname{sign}(\eta')+\sqrt{2}\xi(t)$$ govern the analyze Lévy flights in the equivalent way. Then, via the next transformation $ \eta'\to \mathfrak{u}$ specified by the expressions $ $ \label{sec3:6 } \eta' = \frac1{\alpha}\,\mathfrak{u}\,,\qquad thymine = \frac1{\alpha^2}\,\mathfrak{t}$$ involving also the transformation of the time scale $ t\to \mathfrak{t}$ equation   is rewritten as follows $ $ \label{sec3:7 } \frac{d\mathfrak{u}}{d\mathfrak{t}}=-\operatorname{sign}(\mathfrak{u})+\sqrt{2}\xi(\mathfrak{t})\,.$$ It is a parameter - complimentary stochastic differential equation with additive white noise such that $ $ \label{sec3:8 } \left < \xi(\mathfrak{t})\right > = 0\, , \qquad \left < \xi(\mathfrak{t }) \xi(\mathfrak{t}')\right > = \delta(\mathfrak{t}-\mathfrak{t}')\,.$$ In other quarrel, we have constructed the stochastic process $ \{\mathfrak{u}(\mathfrak{t})\}$ governed by Eq.   that can be treated as the desired * effect process *. It is of the same form for all the types of one - dimensional Lévy flights, at least, Lévy flights described by models similar to Eq.   inside the region where the stinger - off effects are not significant. The basic parameter qualify the generated Lévy flights such as the exponent of the Lévy scaling jurisprudence depend on the system parameters, in particular, the coefficient $ \alpha$ via the transformation from $ \mathfrak{u}(\mathfrak{t})$ to $ v(t)=dx / dt$. Combining in concert

is about 0.35 according to Fig. \[N6\]. ![The magnitude oy the perameted $\Delta_\aupha$ vs the possible values lf the kcrameter $\alpha$.[]{data-lacel="F6"}](fig6.pdv){width="0.45\cooumnxidth"} In what follows we will ckkfine iur consideratlon to the vpecial case  mdtzhnng the ideal Lévy flights. It enables us to vojvert from the stobhwstid process $\{\eta(t)\}$ to the effective stkchastib process $\{\eta'(t)\}$ usong transformation  and to gedufe Eq.  to the stochwstic equatuon $$\jqbel{sec3:5} \fraz{d\eta'}{dt}=-\alpha\operatornaje{sign}(\eta')+\sqrt{2}\xi(t)$$ governing the xnalyved Lévy flithrs lt the equivelent ray. Then, via the nexd transgormation $\eta'\tp \methfeak{u}$ specified by the expressions $$\label{fec3:6} \eta' = \yrac1{\alpha}\,\mathfrak{u}\,,\qquqd t = \ftac1{\al[ha^2}\,\mxrhffak{u}$$ iivomving wlsk the tranaformation if the time scales $u\to \nathfrak{t}$ equztion  ys rewritten as follows $$\label{sec3:7} \frac{d\mauhfrai{u}}{d\mathfrak{t}}=-\operatornamw{sign}(\mathfrak{u})+\sqrt{2}\xi(\mwthfrak{t})\,.$$ Yt is a parameter-free stochastic differential equdtion ditk addiguvf white noise such that $$\label{sec3:8} \left< \xi(\mathsdal{t})\gight> = 0\,, \qquad \lefb< \xi(\mathfrak{t}) \xi(\mayhvrsh{t}')\right> = \deltx(\mathfxzk{f}-\mathfrak{t}')\,.$$ In otheg words, we hqve constwuctrd the stochastic process $\{\mqthfrak{u}(\mathyraj{t})\}$ governed by Eq.  chat can be createc as yhe desired *core procesr*. It is of the dame form wor all the typer on ote-dimensional Lévy flights, wt least, Oévy ylights aesctibed br models slmilav to Eq.  inside the gegiou whese the cut-lff effects are not significant. The basic parsmategs characcerizikg the generateq Lévy flights xuch as the dxponent or the Léty scaling lww depend on dje system pacameters, yn pqrtixular, tfd coefficient $\slpha$ via the transdormation from $\matmfrak{o}(\mzthfrak{t})$ to $v(t)=dr/bt$. Combining togeyhef

is about 0.35 according to Fig. \[F6\]. of parameter $\Delta_\alpha$ the possible values what we will confine consideration to the case matching the ideal Lévy flights. enables us to convert from the stochastic process $\{\eta(t)\}$ to the effective stochastic $\{\eta'(t)\}$ using transformation and to reduce Eq. to the stochastic equation $$\label{sec3:5} \frac{d\eta'}{dt}=-\alpha\operatorname{sign}(\eta')+\sqrt{2}\xi(t)$$ the Lévy in equivalent way. Then, via the next transformation $\eta'\to \mathfrak{u}$ specified by the expressions $$\label{sec3:6} \eta' = t = \frac1{\alpha^2}\,\mathfrak{t}$$ involving also the transformation of time scales $t\to \mathfrak{t}$ is rewritten as follows $$\label{sec3:7} It a parameter-free differential with white noise such $$\label{sec3:8} \left< \xi(\mathfrak{t})\right> = 0\,, \qquad \left< \xi(\mathfrak{t}) \xi(\mathfrak{t}')\right> = \delta(\mathfrak{t}-\mathfrak{t}')\,.$$ In other words, we have constructed stochastic process by Eq. can treated the desired *core is of the same form for of one-dimensional Lévy flights, at least, Lévy flights by models to Eq. inside the region where cut-off effects are not significant. The basic parameters the generated Lévy flights such as the exponent of the Lévy scaling law depend on parameters, in particular, the $\alpha$ via the from to Combining

is about 0.35 according to Fig. \[F6\]. ![The Magnitude oF the pAraMetEr $\deltA_\alpHa$ vs the possiblE ValuEs of the parameter $\alpha$.[]{dAta-laBeL="f6"}](fig6.PDf){Width="0.45\ColumnwIDtH"} iN whAt FoLloWs WE wIll coNfiNe our coNsideratioN to ThE special case  MAtChing the idEal lévy flights. IT enAbles uS tO coNVert fRom The stOchastIC proceSs $\{\eta(t)\}$ to tHe EFfectiVE stochaSTIc ProcEss $\{\eta'(t)\}$ using transFOrMAtion  and to reduCe Eq.  to ThE StOCHasTic Equation $$\laBeL{sec3:5} \fRAc{d\eta'}{dT}=-\AlPHA\OpeRAtorname{sign}(\eTa')+\sqrt{2}\xi(t)$$ goVErnIng the AnAlyZEd Lévy FlighTs IN thE equivalent Way. THen, via the Next trANsformaTIon $\eta'\tO \mathfRak{U}$ spEcifIEd By The ExPResSIoNs $$\lABel{Sec3:6} \eta' = \frAc1{\AlPha}\,\maThfrAK{U}\,,\QQuad T = \frAc1{\alPha^2}\,\maThfrak{t}$$ involvIng Also THe tRansfOrmatIon oF tHe timE scaleS $t\to \mAtHfrak{t}$ equation  iS rewRitten as fOllOwS $$\laBeL{sec3:7} \fRAc{d\matHfrAk{u}}{D\mathfrAk{t}}=-\operATorNaME{SIgN}(\mathfrak{u})+\sqrt{2}\xi(\maThFRAk{T})\,.$$ It is a paRameteR-FrEe STochastiC dIffErenTIAl equAtioN WiTh additiVe whitE NoIsE such thAt $$\Label{sEc3:8} \LefT< \xi(\MathfRAk{t})\rIght> = 0\,, \qqUad \left< \xI(\mathFRak{t}) \xi(\mathfrak{T}')\Right> = \delta(\matHFrAK{T}-\mAThfrAk{t}')\,.$$ in other wordS, we hAVe coNstrUCtEd tHE stocHastiC pROcESs $\{\mathfrak{u}(\mathfrak{T})\}$ gOverneD by Eq.  That can be treaTed as the deSIREd *core prOcesS*. it IS of the same form For alL the types oF One-dimenSionaL Lévy fliGhts, at leaST, lévy fligHts DesCriBed BY MoDels similar to eQ.  InsiDe The regiOn wHere the Cut-Off EffEctS aRe not signIficant. THe BaSiC pAraMeterS CharacteRiZinG tHe gEneraTEd Lévy FlighTs suCh As THe eXponent OF tHE lévy ScAlIng lAw dEpEnd on The sYSteM parameTers, in parTicULar, tHe CoEfficieNt $\alpha$ via the TrAnsformatiOn FroM $\mathfRAK{u}(\mathfrAk{t})$ to $v(t)=dx/dt$. Combining togETher

is about 0.35 according t o Fig. \[F 6\]. ![ The m agni tude of the parame t er $ \Delta_\alpha$ vs thepossi bl e val u es of t he para m et e r $\ al ph a$. [] { da ta-la bel ="F6"}] (fig6.pdf) {wi dt h="0.45\colu m nw idth"} In wh at follows w e w ill co nf ine our c ons idera tion t o the s pecial ca se match i ng thei d ea l Lé vy flights. It en a bl e s us to conver t from t h es t och ast ic process $ \{\et a (t)\}$t ot h e ef f ective stocha stic proces s $\ {\eta' (t )\} $ using tran sf o rma tion  and t o re duce Eq. to th e stocha s tic equ ation$$\ lab el{s e c3 :5 } \ f rac { d\ eta ' }{d t}=-\alp ha \o perat orna m e { s ign} (\e ta') +\sqr t{2}\xi(t)$$gov erni n g t he an alyze d Lé vy flig hts in theeq uivalent way. T hen, via thenex ttra ns forma t ion $\ eta '\t o \math frak{u} $ sp ec i f i ed by the expression s$ $ \l abel{sec 3:6} \ et a ' = \fra c1 {\a lpha } \ ,\mat hfra k {u }\,,\qqu ad t = \f ra c1{\alp ha ^2}\,\ ma thf rak {t}$$ invo lvingalso the tran s formation of t h e time scales $t \ t o\ math fra k{t}$ equat ionis r ewri t te n a s foll ows $ $\ l ab e l{sec3:7} \frac{d \m athfra k{u}} {d\mathfrak{t }}=-\opera t o r name{sig n}(\ m at h frak{u})+\sqrt {2}\x i(\mathfra k {t})\,.$ $ Itis a par ameter-fr e e stochas tic di ffe ren t i al equation wit h addi ti ve whit e n oise su chtha t $ $\l ab el{sec3:8 } \lef t< \ xi (\ mat hfrak { t})\righ t> =0\ ,,\qqua d \left < \xi (\ma th fr a k{t }) \xi( \ ma t h frak {t }' )\ri ght >= \de lta( \ mat hfrak{t }-\mathfr ak{ t }')\ ,. $$ In oth er words, weha ve constru ct edthe st o c hastic p rocess $\{\mathfrak{u}( \ mathfra k{t })\}$ gov erned byEq.   that ca n be tr eatedas th edes i r ed *c o r epro ce ss*. It is o f t he sa me for m for a ll the types of on e -di mensional Lév y f ligh t s ,atl ea s t,Lé v y f l i ghts describedby modelssi m il ar to Eq.ins id e the r egion w heret he cut- off effec ts are no tsign i f ica nt. The ba sic para meters ch a racte r iz ing t hegenera te d L évy f lights suc h asthe ex po nent o f the L évy scal ing law depend on the s ystemparam ete rs, in pa rti c ula r, the co effi cient $\al pha $ v ia th e t r ansfo rmat i on fr o m $\m athf r ak{u}(\ma t hf rak { t }) $ to $v(t)= d x / dt$ . Com bin i ng tog ethe r

is_about 0.35_according to Fig. \[F6\]. ![The magnitude_of the_parameter_$\Delta_\alpha$ vs_the_possible values of_the parameter $\alpha$.[]{data-label="F6"}](fig6.pdf){width="0.45\columnwidth"} In_what follows we will_confine our consideration_to_the special case  matching the ideal Lévy flights. It enables us to convert from_the_stochastic process_$\{\eta(t)\}$_to_the effective stochastic process $\{\eta'(t)\}$_using transformation  and to reduce_Eq.  to_the stochastic equation $$\label{sec3:5} \frac{d\eta'}{dt}=-\alpha\operatorname{sign}(\eta')+\sqrt{2}\xi(t)$$ governing the_analyzed_Lévy flights in_the equivalent way. Then, via the next transformation $\eta'\to_\mathfrak{u}$ specified by the expressions $$\label{sec3:6} _ \eta'_=_\frac1{\alpha}\,\mathfrak{u}\,,\qquad_t = \frac1{\alpha^2}\,\mathfrak{t}$$ involving_also the transformation of the time_scales $t\to \mathfrak{t}$ equation  is rewritten_as follows $$\label{sec3:7} \frac{d\mathfrak{u}}{d\mathfrak{t}}=-\operatorname{sign}(\mathfrak{u})+\sqrt{2}\xi(\mathfrak{t})\,.$$ It is_a parameter-free stochastic differential equation with_additive white noise such that_$$\label{sec3:8} _\left< \xi(\mathfrak{t})\right> = 0\,, \qquad_\left< \xi(\mathfrak{t}) \xi(\mathfrak{t}')\right>_= \delta(\mathfrak{t}-\mathfrak{t}')\,.$$_In other words,_we have constructed the stochastic process_$\{\mathfrak{u}(\mathfrak{t})\}$ governed by_Eq.  that can be treated as_the_desired *core process*._It_is_of the_same form for_all_the types_of_one-dimensional Lévy flights, at least, Lévy_flights_described by models similar to Eq.  inside_the region where the_cut-off_effects are not significant._The basic parameters characterizing the_generated Lévy flights such as the_exponent of_the Lévy_scaling law depend on the system parameters, in particular, the coefficient_$\alpha$ via the transformation from $\mathfrak{u}(\mathfrak{t})$_to $v(t)=dx/dt$. Combining together

J^{n_1n_2}\stackrel{\text{$f_3$}}{\leftarrow} J^{n_1n_2n_3}\stackrel{\text{$f_4$}}{\leftarrow}\ldots$$ We use the notation of the proof of Lemma \[lem:Mayerprep\]. Lemma \[lem:indecsubc\] implies that for every sequence $(n_i)$ there exists an embedding of $Y$ such that both points of $\partial J\leftarrow \partial J^{n_1}\leftarrow \partial J^{n_1n_2}\leftarrow \partial J^{n_1n_2n_3}\leftarrow\ldots$ are accessible and which can be extended to an embedding of $X$. This completes the proof. We have proven that every chainable continuum containing a nondegenerate indecomposable subcontinuum has uncountably many embeddings that are not strongly equivalent. Thus we pose the following question. \[q:above\] Which hereditarily decomposable chainable continua have uncountably many planar embeddings that are not equivalent and/or strongly equivalent? Mayer has constructed uncountably many nonequivalent planar embeddings (in both senses) in [@May] of the $\sin\frac 1x$ continuum by varying the rate of convergence of the ray. This approach readily generalizes to any Elsa continuum. We do not know whether the approach can be generalized to all chainable continua which contain a dense ray. Specifically, it would be interesting to see if $\underleftarrow{\lim}\{I, f_{Feig}\}$ (where $f_{Feig}$ denotes the logistic interval map at the Feigenbaum parameter) can be embedded in uncountably many nonequivalent ways. However, this approach would not generalize to the remaining hereditarily decomposable chainable continua since there exist hereditarily decomposable chainable continua which do not contain a dense ray, see [@Jan]. \[rem:n\_emb\] In Figure \[fig:n\_embed\] we give examples of planar continua which have exactly $n\in\N$ or countably many nonequivalent planar embeddings. However, except for the arc, all the examples we know are not chainable. (1,1)–(2,1); (1,1)–(0,1); (1,1)–(1,2); (1,1)–(1,

J^{n_1n_2}\stackrel{\text{$f_3$}}{\leftarrow } J^{n_1n_2n_3}\stackrel{\text{$f_4$}}{\leftarrow}\ldots$$ We use the notation of the proof of Lemma   \[lem: Mayerprep\ ]. Lemma   \[lem: indecsubc\ ] implies that for every succession $ (n_i)$ there exist an embedding of $ Y$ such that both points of $ \partial J\leftarrow \partial J^{n_1}\leftarrow \partial J^{n_1n_2}\leftarrow \partial J^{n_1n_2n_3}\leftarrow\ldots$ are accessible and which can be extended to an embedding of $ X$. This dispatch the validation. We have proven that every chainable continuum containing a nondegenerate indecomposable subcontinuum has uncountably many embeddings that are not powerfully equivalent. therefore we pose the following question. \[q: above\ ] Which hereditarily decomposable chainable continua have uncountably many planar embeddings that are not equivalent and/or powerfully equivalent? Mayer has constructed uncountably many nonequivalent planar embeddings (in both senses) in [ @May ] of the $ \sin\frac 1x$ continuum by varying the rate of overlap of the ray. This approach promptly generalizes to any Elsa continuum. We do not acknowledge whether the access can be generalized to all chainable continua which contain a dense re. Specifically, it would be interesting to see if $ \underleftarrow{\lim}\{I, f_{Feig}\}$ (where $ f_{Feig}$ denotes the logistic interval function at the Feigenbaum parameter) can be embed in uncountably many nonequivalent ways. However, this approach would not generalize to the remain hereditarily decomposable chainable continua since there exist hereditarily decomposable chainable continua which do not contain a dense ray, see [ @Jan ]. \[rem: n\_emb\ ] In Figure   \[fig: n\_embed\ ] we give examples of planar continua which have exactly $ n\in\N$ or countably many nonequivalent planar embeddings. However, except for the arc, all the examples we know are not chainable. (1,1)–(2,1); (1,1)–(0,1); (1,1)–(1,2); (1,1)–(1,

J^{n_1j_2}\stackrel{\text{$f_3$}}{\leftarrow} M^{n_1n_2n_3}\stackrel{\text{$y_4$}}{\oeftarcow}\ldota$$ We use the notation of the proof oh Lenma \[len:Mayerprep\]. Lemma \[lem:inddcsubc\] imilies thar foc every sequence $(n_i)$ therc exiabs an xmbedding of $Y$ xuch that toth points of $\pxrcial J\leftarrow \partial J^{n_1}\leftarrow \[artial J^{j_1n_2}\leftarrow \pattial T^{n_1n_2n_3}\mvfuarrow\ldots$ are accessible and whjch can be extended yo an embedding of $X$. This fomppetes the proof. We jave proven thae every chainxble continuum containjng a nondegenerate indecomposacle sbbcontinuum hqs ktcountably nany vmbeddings that are nmt stromgly equivalenb. Thuv ww pose the following xuestion. \[q:above\] Whish heredidaxily decomposable chaunqble wonthnua yavd uhcpuhtably maiy planar ejbeddings tyat are not equivaltnt qnd/or stronglg equidajent? Mayer has constructed uncountably mdny nonequivalent planar enbeddings (in both sendes) in [@Mar] of the $\sin\frac 1x$ continuum by varying the rate mf coivdrgtngc of rhf ray. This approach readily generalizes to anr Ekss continuum. We do not knoe ahrjher the approxch cau bs generalized to apl chaigable continua whivh contain a dense ray. Specufically, it couod be interesting co see if $\unberlefjarrow{\kim}\{I, f_{Feig}\}$ (where $f_{Feig}$ benotea the logishic interbxl map at the Fekgekbagm paramtger) can be embeddqd in unciuntcbly manh nomequivwlent ways. Howeyar, this approach wluld uot ganeralize ho the remaining hereditarily dxromposable chsitabne contiuua sikce there exist hereditarily becomposcble cfainable ckntinua which do njt contain a ggnse ray, see [@Oan]. \[rem:n\_emf\] In Figyre \[fig:n\_dobed\] we give ecamples oy plabar continua which haxs exactly $n\in\N$ ir xountably many mondquyvwlxnt pjdnar embedditgs. Fowdfer, ebcept for tme xrc, sll the examples we nnow are not chainable. (1,1)–(2,1); (1,1)–(0,1); (1,1)–(1,2); (1,1)–(1,

J^{n_1n_2}\stackrel{\text{$f_3$}}{\leftarrow} J^{n_1n_2n_3}\stackrel{\text{$f_4$}}{\leftarrow}\ldots$$ We use the notation of of \[lem:Mayerprep\]. Lemma implies that for an of $Y$ such both points of J\leftarrow \partial J^{n_1}\leftarrow \partial J^{n_1n_2}\leftarrow \partial are accessible and which can be extended to an embedding of $X$. This the proof. We have proven that every chainable continuum containing a nondegenerate indecomposable has many that not strongly equivalent. Thus we pose the following question. \[q:above\] Which hereditarily decomposable chainable continua have many planar embeddings that are not equivalent and/or equivalent? Mayer has constructed many nonequivalent planar embeddings (in senses) [@May] of $\sin\frac continuum varying the rate convergence of the ray. This approach readily generalizes to any Elsa continuum. We do not know whether approach can to all continua contain dense ray. Specifically, be interesting to see if $\underleftarrow{\lim}\{I, denotes the logistic interval map at the Feigenbaum can be in uncountably many nonequivalent ways. However, approach would not generalize to the remaining hereditarily chainable continua since there exist hereditarily decomposable chainable continua which do not contain a dense [@Jan]. \[rem:n\_emb\] In Figure we give examples planar which exactly or countably nonequivalent planar embeddings. However, except for the arc, all the examples know are not chainable. (1,1)–(2,1); (1,1)–(0,1); (1,1)–(1,2); (1,1)–(1,

J^{n_1n_2}\stackrel{\text{$f_3$}}{\leftarrow} j^{n_1n_2n_3}\stackrEl{\texT{$f_4$}}{\lEftArRow}\lDots$$ we use the notatiON of tHe proof of Lemma \[lem:MayerPrep\]. LEmMA \[lem:INdEcsubC\] implieS ThAT For EvErY seQuENcE $(n_i)$ thEre Exists aN embedding Of $Y$ SuCh that both poINtS of $\partial j\leFtarrow \partiAl J^{N_1}\leftaRrOw \pARtial j^{n_1n_2}\LeftaRrow \paRTial J^{n_1N_2n_3}\leftarrOw\LDots$ arE AccessiBLE aNd whIch can be extended tO An EMbedding of $X$. ThiS complEtES tHE ProOf. WE have proveN tHat evERy chainABlE CONtiNUum containing A nondegenerATe iNdecomPoSabLE subcoNtinuUm HAs uNcountably mAny eMbeddings That arE Not stroNGly equiValent. thuS we Pose THe FoLloWiNG quEStIon. \[Q:AboVe\] Which hErEdItariLy deCOMPOsabLe cHainAble cOntinua have unCouNtabLY maNy plaNar emBeddInGs thaT are noT equiVaLent and/or strongLy eqUivalent? MAyeR hAs cOnStrucTEd uncoUntAblY many noNequivaLEnt PlANAR eMbeddings (in both senSeS) IN [@MAy] of the $\sIn\frac 1X$ CoNtINuum by vaRyIng The rATE of coNverGEnCe of the rAy. This APpRoAch readIlY generAlIzeS to Any ElSA conTinuum. we do not kNow whETher the approacH Can be generaliZEd TO AlL ChaiNabLe continua wHich COntaIn a dENsE raY. speciFicalLy, IT wOUld be interesting to sEe If $\undeRleftArrow{\lim}\{I, f_{FeiG}\}$ (where $f_{FeiG}$ DENotes the LogiSTiC Interval map at tHe FeiGenbaum parAMeter) can Be embEdded in uNcountablY MAny noneqUivAleNt wAys. hOWeVer, this approaCH WoulD nOt generAliZe to the RemAinIng HerEdItarily deComposabLe ChAiNaBle ContiNUa since tHeRe eXiSt hErediTArily dEcompOsabLe ChAInaBle contINuA WHich Do NoT conTaiN a Dense Ray, sEE [@JaN]. \[rem:n\_emB\] In Figure \[Fig:N\_EmbeD\] wE gIve examPles of planar cOnTinua which HaVe eXactly $N\IN\N$ or counTably many nonequivalent pLAnar embEddIngs. HOwevEr, except fOr tHe arc, aLl tHE exampLes we kNow arE nOt cHAInablE. (1,1)–(2,1); (1,1)–(0,1); (1,1)–(1,2); (1,1)–(1,

J^{n_1n_2}\stackrel{\tex t{$f_3$}}{ \left arr ow} J^{n _1n_ 2n_3}\stackrel { \tex t{$f_4$}}{\leftarrow}\ ldots $$ We u s ethe n otation of t hepr oo f o fL em ma \[ lem :Mayerp rep\]. Lem ma\[ lem:indecsub c \] implies t hat for every s equ ence $ (n _i) $ ther e e xists an em b edding of $Y$ s uc h thatb oth poi n t sof $ \partial J\leftar r ow \partial J^{n _1}\le ft a rr o w \p art ial J^{n_1 n_ 2}\le f tarrow\ pa r t i alJ^{n_1n_2n_3} \leftarrow\ l dot s$ are a cce s sibleand w hi c h c an be exten dedto an emb edding of $X$. This co mplete s t heproo f . W e h av e pr o ve n t h atevery ch ai na ble c onti n u u m con tai ning a no ndegenerate i nde comp o sab le su bcont inuu mhas u ncount ablyma ny embeddings t hatare not s tro ng lyeq uival e nt. Th uswepose th e follo w ing q u e s ti on. \[q:above\] W hi c h h ereditar ily de c om po s able cha in abl e co n t inuahave un countabl y many pl an ar embe dd ings t ha t a renot e q uiva lent a nd/or st rongl y equivalent? M ayer has cons t ru c t ed unco unt ably many n oneq u ival entp la nar embed dings ( i nb oth senses) in [@Ma y] of th e $\s in\frac 1x$ c ontinuum b y v arying t he r a te of convergence of t he ray. Th i s approa ch re adily ge neralizes t o any El sacon tin uum . We do not knoww h ethe rthe app roa ch canbegen era liz ed to all c hainable c on ti nu a w hichc ontain a d ens eray . Spe c ifical ly, i t wo ul db e i nterest i ng t o se eif $\u nde rl eftar row{ \ lim }\{I, f _{Feig}\} $ ( w here $ f_ {Feig}$ denotes thelo gistic int er val map a t the Feig enbaum parameter) can b e embedd edin un coun tably man y n onequi val e nt way s. How ever, t his a pproa c h w oul dnot genera l i zeto th erema ining h ereditarily decomp o sab le chainablecon tinu a si nce th e reex i sth e reditarily deco mposable c ha i na ble contin u a w hi ch do n ot cont ain a dense r ay, see [ @Jan]. \ [r em:n \ _ emb \] In Figu re \[fig :n\_embed \ ] weg iv e exa mpl es ofpl ana r con tinuaw hic h hav e exac tl y $n\i n\N$or countab ly many nonequivalent p lanarembed din gs. Howev er, exc ept for t he a rc, all th e e xam pleswek now a re n o tcha i nable . ( 1 ,1)–(2,1) ; ( 1,1 ) – (0 ,1); (1,1)– ( 1 , 2); (1,1 )–( 1 ,

J^{n_1n_2}\stackrel{\text{$f_3$}}{\leftarrow}_ J^{n_1n_2n_3}\stackrel{\text{$f_4$}}{\leftarrow}\ldots$$ We_use the notation of_the proof_of_Lemma \[lem:Mayerprep\]. Lemma \[lem:indecsubc\]_implies_that for every_sequence $(n_i)$ there_exists an embedding of_$Y$ such that_both_points of $\partial J\leftarrow \partial J^{n_1}\leftarrow \partial J^{n_1n_2}\leftarrow \partial J^{n_1n_2n_3}\leftarrow\ldots$ are accessible and which_can_be extended_to_an_embedding of $X$. This completes_the proof. We have proven that_every chainable_continuum containing a nondegenerate indecomposable subcontinuum has uncountably_many_embeddings that are_not strongly equivalent. Thus we pose the following question. \[q:above\]_Which hereditarily decomposable chainable continua have_uncountably many planar_embeddings_that_are not equivalent and/or_strongly equivalent? Mayer has constructed uncountably many_nonequivalent planar embeddings (in both senses)_in [@May] of the $\sin\frac 1x$ continuum_by varying the rate of convergence_of the ray. This approach_readily generalizes_to any Elsa continuum. We_do not know_whether the_approach can be_generalized to all chainable continua which_contain a dense_ray. Specifically, it would be interesting_to_see if $\underleftarrow{\lim}\{I,_f_{Feig}\}$_(where_$f_{Feig}$ denotes_the logistic interval_map_at the_Feigenbaum_parameter) can be embedded in uncountably_many_nonequivalent ways. However, this approach would not_generalize to the remaining_hereditarily_decomposable chainable continua since_there exist hereditarily decomposable chainable_continua which do not contain a_dense ray,_see [@Jan]. \[rem:n\_emb\]_In Figure \[fig:n\_embed\] we give examples of planar continua which have exactly_$n\in\N$ or countably many nonequivalent planar_embeddings. However, except for_the arc,_all_the examples we_know_are not_chainable. (1,1)–(2,1); (1,1)–(0,1); (1,1)–(1,2); (1,1)–(1,

}t}{2}\right).\label{eq:2d-3}$$ By replacing (\[eq:2d-1\]) and (\[eq:2d-2\]) with (\[eq:2d-3\]) into the Gross-Pitaevskii equation (\[eq:GPE\]), we obtain $$\begin{aligned} i\hbar W\frac{\partial\Phi}{\partial t}+\frac{\hbar\omega_{z}}{2}\Phi W & = & \left[-\frac{\hbar^{2}}{2m}\nabla_{\bot}^{2}+\frac{\hbar^{2}}{2md_{z}^{2}}-\frac{\hbar^{2}z^{2}}{2md_{z}^{4}}+V\left(\mathbf{r}\right)+NU_{0}\left|\Phi W\right|^{2}\right]\Phi W.\label{eq:2d-4}\end{aligned}$$ The product $\Phi\left(\mathbf{r}_{\bot},t\right)W\left(z,t\right)$ is normalized to unity, thus the number of atoms appears multiplying the coupling constant $U_{0}$. Now we can multiply Eq. (\[eq:2d-4\]) by $W^{*}\left(z,t\right)$ and integrate this equation over the entire z domain. Since $\hbar^{2}/2md_{z}^{2}=\hbar\omega_{z}/2$, and $\hbar^{2}/2md_{z}^{4}=m\omega_{z}^{2}/2$ we obtain the following simplified equation $$i\hbar\frac{\partial\Phi\left(\mathbf{r}_{\bot},t\right)}{\partial t}=\left[-\frac{\hbar^{2}}{2m}\nabla_{\bot}^{2}+V_{\bot}\left(\mathbf{r}_{\bot}\right)+NU_{2D}\left|\Phi\left(\mathbf{r}_{\bot},t\right)\right|^{2}\right]\Phi\left(\mathbf{r}_{\bot},t\right),\label{eq:2D-GPE}$$ where $$U_{2D}=\frac{U_{0}}{d_{z}\sqrt{2\pi}}=2\sqrt{2\pi}\frac{\hbar^{2}a_{s}}{md_{z}}.$$ Let us then write the Lagrangian density which leads to quasi-2D Gross-Pitaevskii equation (\[eq:2D-GPE\]) for a complex field $\Phi

} t}{2}\right).\label{eq:2d-3}$$ By replacing (\[eq:2d-1\ ]) and (\[eq:2d-2\ ]) with (\[eq:2d-3\ ]) into the Gross - Pitaevskii equation (\[eq: GPE\ ]), we obtain $ $ \begin{aligned } i\hbar W\frac{\partial\Phi}{\partial t}+\frac{\hbar\omega_{z}}{2}\Phi W & = & \left[-\frac{\hbar^{2}}{2m}\nabla_{\bot}^{2}+\frac{\hbar^{2}}{2md_{z}^{2}}-\frac{\hbar^{2}z^{2}}{2md_{z}^{4}}+V\left(\mathbf{r}\right)+NU_{0}\left|\Phi W\right|^{2}\right]\Phi W.\label{eq:2d-4}\end{aligned}$$ The product $ \Phi\left(\mathbf{r}_{\bot},t\right)W\left(z, t\right)$ is anneal to oneness, thus the number of atoms appear multiplying the coupling ceaseless $ U_{0}$. immediately we can multiply Eq. (\[eq:2d-4\ ]) by $ W^{*}\left(z, t\right)$ and desegregate this equality over the entire z knowledge domain. Since $ \hbar^{2}/2md_{z}^{2}=\hbar\omega_{z}/2 $, and $ \hbar^{2}/2md_{z}^{4}=m\omega_{z}^{2}/2 $ we obtain the following simplify equation $ $ i\hbar\frac{\partial\Phi\left(\mathbf{r}_{\bot},t\right)}{\partial t}=\left[-\frac{\hbar^{2}}{2m}\nabla_{\bot}^{2}+V_{\bot}\left(\mathbf{r}_{\bot}\right)+NU_{2D}\left|\Phi\left(\mathbf{r}_{\bot},t\right)\right|^{2}\right]\Phi\left(\mathbf{r}_{\bot},t\right),\label{eq:2D - GPE}$$ where $ $ U_{2D}=\frac{U_{0}}{d_{z}\sqrt{2\pi}}=2\sqrt{2\pi}\frac{\hbar^{2}a_{s}}{md_{z}}.$$ Let us then spell the Lagrangian density which leads to quasi-2D Gross - Pitaevskii equality (\[eq:2D - GPE\ ]) for a complex field $ \Phi

}t}{2}\rihht).\label{eq:2d-3}$$ By replacing (\[eq:2d-1\]) and (\[eq:2d-2\]) wijh (\[eq:2d-3\]) iito the Gross-Pigaevskii equation (\[eq:GPE\]), we outaib $$\begun{aligned} i\hbar W\frac{\paftial\Phi}{\pwrtial t}+\drac{\ibar\omega_{z}}{2}\Phi W & = & \left[-\fvcc{\hbad^{2}}{2l}\nabne_{\bot}^{2}+\frac{\hbar^{2}}{2md_{z}^{2}}-\nrac{\hbar^{2}z^{2}}{2md_{s}^{4}}+V\left(\mathbf{r}\rhgft)+UU_{0}\left|\Phi W\right|^{2}\right]\Phi W.\label{eq:2d-4}\enq{alignec}$$ Hhe product $\Phy\lefu(\maehbf{d}_{\bot},t\right)W\left(z,t\right)$ is normalizsd to uiity, thus the nimber of atoms appears mulhipljing the coupling fonstant $U_{0}$. Bow rw can multipuy Eq. (\[eq:2d-4\]) by $W^{*}\left(z,t\rifht)$ and integrate this equation over the entirg a fmmain. Since $\hbar^{2}/2id_{z}^{2}=\hbar\omega_{d}/2$, and $\htar^{2}/2md_{z}^{4}=m\pmega_{z}^{2}/2$ we obtaln thx foolowing simplified eqnation $$i\hbar\frac{\partyal\Phi\lefd(\mcthbf{r}_{\bot},t\right)}{\partiao r}=\left[-\xrac{\vbar^{2}}{2o}\babua_{\bkt}^{2}+T_{\bof}\left(\mwthuf{r}_{\bot}\right)+HU_{2D}\left|\Phi\lwft(\mathbf{r}_{\bot},t\right)\tidyt|^{2}\right]\Phi\leff(\mathbs{r}_{\fot},t\right),\label{eq:2D-GPE}$$ where $$U_{2D}=\frac{U_{0}}{d_{z}\sqru{2\pi}}=2\sqdt{2\pi}\frac{\hbar^{2}a_{s}}{md_{z}}.$$ Let uw then write the Lagrwngian degsity which leads to quasi-2D Gross-Pitaevskii equathon (\[es:2A-GPT\]) njf a complex field $\Phi

}t}{2}\right).\label{eq:2d-3}$$ By replacing (\[eq:2d-1\]) and (\[eq:2d-2\]) with the equation (\[eq:GPE\]), obtain $$\begin{aligned} i\hbar & W\right|^{2}\right]\Phi W.\label{eq:2d-4}\end{aligned}$$ The $\Phi\left(\mathbf{r}_{\bot},t\right)W\left(z,t\right)$ is normalized unity, thus the number of atoms multiplying the coupling constant $U_{0}$. Now we can multiply Eq. (\[eq:2d-4\]) by $W^{*}\left(z,t\right)$ integrate this equation over the entire z domain. Since $\hbar^{2}/2md_{z}^{2}=\hbar\omega_{z}/2$, and $\hbar^{2}/2md_{z}^{4}=m\omega_{z}^{2}/2$ we the simplified $$i\hbar\frac{\partial\Phi\left(\mathbf{r}_{\bot},t\right)}{\partial where $$U_{2D}=\frac{U_{0}}{d_{z}\sqrt{2\pi}}=2\sqrt{2\pi}\frac{\hbar^{2}a_{s}}{md_{z}}.$$ Let us then write the Lagrangian density which leads to quasi-2D Gross-Pitaevskii equation (\[eq:2D-GPE\]) a complex field $\Phi

}t}{2}\right).\label{eq:2d-3}$$ By replacing (\[Eq:2d-1\]) and (\[eq:2d-2\]) wIth (\[eq:2D-3\]) inTo tHe grosS-PitAevskii equatioN (\[Eq:GPe\]), we obtain $$\begin{aligned} i\Hbar W\FrAC{\parTIaL\Phi}{\pArtial t}+\FRaC{\HBar\OmEgA_{z}}{2}\PHi w & = & \LeFt[-\fraC{\hbAr^{2}}{2m}\nablA_{\bot}^{2}+\frac{\hbAr^{2}}{2mD_{z}^{2}}-\Frac{\hbar^{2}z^{2}}{2md_{z}^{4}}+v\LeFt(\mathbf{r}\rIghT)+NU_{0}\left|\Phi W\rIghT|^{2}\right]\phI W.\lABel{eq:2D-4}\enD{aligNed}$$ The PRoduct $\phi\left(\maThBF{r}_{\bot},t\RIght)W\leFT(Z,t\RighT)$ is normalized to unITy, THus the number of Atoms aPpEArS MUltIplYing the couPlIng coNStant $U_{0}$. NOW wE CAN muLTiply Eq. (\[eq:2d-4\]) by $W^{*}\Left(z,t\right)$ ANd iNtegraTe ThiS EquatiOn oveR tHE enTire z domain. sincE $\hbar^{2}/2md_{z}^{2}=\hBar\omeGA_{z}/2$, and $\hbAR^{2}/2md_{z}^{4}=m\omEga_{z}^{2}/2$ we ObtAin The fOLlOwIng SiMPliFIeD eqUAtiOn $$i\hbar\fRaC{\pArtiaL\Phi\LEFT(\MathBf{r}_{\Bot},t\Right)}{\Partial t}=\left[-\fRac{\Hbar^{2}}{2M}\NabLa_{\bot}^{2}+v_{\bot}\lEft(\mAtHbf{r}_{\bOt}\righT)+NU_{2D}\lEfT|\Phi\left(\mathbf{r}_{\Bot},t\Right)\righT|^{2}\riGhT]\PhI\lEft(\maTHbf{r}_{\boT},t\rIghT),\label{eQ:2D-GPE}$$ whERe $$U_{2d}=\fRAC{u_{0}}{d_{Z}\sqrt{2\pi}}=2\sqrt{2\pi}\frac{\hBaR^{2}A_{S}}{mD_{z}}.$$ Let us tHen wriTE tHe lAgrangiaN dEnsIty wHICh leaDs to QUaSi-2D Gross-pitaevSKiI eQuation (\[Eq:2d-GPE\]) foR a ComPleX fielD $\phi

}t}{2}\right).\label{eq:2d -3}$$ By r eplac ing (\ [e q:2d -1\] ) and (\[eq:2d - 2\]) with (\[eq:2d-3\]) in to th eG ross - Pi taevs kii equ a ti o n (\ [e q: GPE \] ) ,we ob tai n $$\be gin{aligne d}i\ hbar W\frac{ \ pa rtial\Phi} {\p artial t}+\f rac {\hbar \o meg a _{z}} {2} \PhiW & =& \left [-\frac{\ hb a r^{2}} { 2m}\nab l a _{ \bot }^{2}+\frac{\hbar ^ {2 } }{2md_{z}^{2}} -\frac {\ h ba r ^ {2} z^{ 2}}{2md_{z }^ {4}}+ V \left(\ m at h b f {r} \ right)+NU_{0} \left|\PhiW \ri ght|^{ 2} \ri g ht]\Ph i W.\ la b el{ eq:2d-4}\en d{al igned}$$The pr o duct $\ P hi\left (\math bf{ r}_ {\bo t }, t\ rig ht ) W\l e ft (z, t \ri ght)$ is n or maliz ed t o u n ity, th us t he nu mber of atoms ap pear s mu ltipl yingtheco uplin g cons tant$U _{0}$. Now we c an m ultiply E q.(\ [eq :2 d-4\] ) by $W ^{* }\l eft(z,t \right) $ an di n t eg rate this equation o v e rthe enti re z d o ma in . Since $ \h bar ^{2} / 2 md_{z }^{2 } =\ hbar\ome ga_{z} / 2$ ,and $\h ba r^{2}/ 2m d_{ z}^ {4}=m \ omeg a_{z}^ {2}/2$ w e obt a in the followi n g simplifiede qu a t io n $$i \hb ar\frac{\pa rtia l \Phi \lef t (\ mat h bf{r} _{\bo t} , t\ r ight)}{\partial t}= \l eft[-\ frac{ \hbar^{2}}{2m }\nabla_{\ b o t }^{2}+V_ {\bo t }\ l eft(\mathbf{r} _{\bo t}\right)+ N U_{2D}\l eft|\ Phi\left (\mathbf{ r } _{\bot}, t\r igh t)\ rig h t |^ {2}\right]\Ph i \ left (\ mathbf{ r}_ {\bot}, t\r igh t), \la be l{eq:2D-G PE}$$ wh er e$$ U_ {2D }=\fr a c{U_{0}} {d _{z }\ sqr t{2\p i }}=2\s qrt{2 \pi} \f ra c {\h bar^{2} a _{ s } }{md _{ z} }.$$ Le tus th en w r ite the La grangianden s itywh ic h leads to quasi-2DGr oss-Pitaev sk iiequati o n (\[eq:2 D-GPE\]) for a complexf ield $\ Phi

}t}{2}\right).\label{eq:2d-3}$$ By_replacing (\[eq:2d-1\])_and (\[eq:2d-2\]) with (\[eq:2d-3\])_into the_Gross-Pitaevskii_equation (\[eq:GPE\]),_we_obtain $$\begin{aligned} i\hbar W\frac{\partial\Phi}{\partial_t}+\frac{\hbar\omega_{z}}{2}\Phi W &_= & \left[-\frac{\hbar^{2}}{2m}\nabla_{\bot}^{2}+\frac{\hbar^{2}}{2md_{z}^{2}}-\frac{\hbar^{2}z^{2}}{2md_{z}^{4}}+V\left(\mathbf{r}\right)+NU_{0}\left|\Phi W\right|^{2}\right]\Phi_W.\label{eq:2d-4}\end{aligned}$$ The product_$\Phi\left(\mathbf{r}_{\bot},t\right)W\left(z,t\right)$_is normalized to unity, thus the number of atoms appears multiplying the coupling constant_$U_{0}$._Now we_can_multiply_Eq. (\[eq:2d-4\]) by $W^{*}\left(z,t\right)$ and_integrate this equation over the_entire z_domain. Since $\hbar^{2}/2md_{z}^{2}=\hbar\omega_{z}/2$, and $\hbar^{2}/2md_{z}^{4}=m\omega_{z}^{2}/2$ we obtain the_following_simplified equation $$i\hbar\frac{\partial\Phi\left(\mathbf{r}_{\bot},t\right)}{\partial_t}=\left[-\frac{\hbar^{2}}{2m}\nabla_{\bot}^{2}+V_{\bot}\left(\mathbf{r}_{\bot}\right)+NU_{2D}\left|\Phi\left(\mathbf{r}_{\bot},t\right)\right|^{2}\right]\Phi\left(\mathbf{r}_{\bot},t\right),\label{eq:2D-GPE}$$ where $$U_{2D}=\frac{U_{0}}{d_{z}\sqrt{2\pi}}=2\sqrt{2\pi}\frac{\hbar^{2}a_{s}}{md_{z}}.$$ Let us then write the Lagrangian_density which leads to quasi-2D Gross-Pitaevskii_equation (\[eq:2D-GPE\]) for_a_complex_field $\Phi

textbf{R}_2^{(t)}\textbf{R}_2^{(t)^H}) \} \label{Lag2} \end{split}$$ where $$\begin{split} & \textbf{W}_{r,1}^{(t)}=\textbf{H}_{r,1}^{(t)}\textbf{F}^{(t)}\textbf{H}_{2,r}^{(t-1)^H}, \\ & \textbf{W}_{r,2}^{(t)}=\textbf{H}_{r,2}^{(t)}\textbf{F}^{(t)}\textbf{H}_{1,r}^{(t-1)^H}, \\ & \textbf{W}_{r,3}^{(t)}=\textbf{H}_{r,1}^{(t)}\textbf{F}^{(t)}\textbf{G}_1^{(t)}\textbf{F}^{(t)^H}\textbf{H}_{r,1}^{(t)^H}+ (N_sp_1\sigma_{e,1}^2+\sigma_{n,1}^2)\textbf{I}_{N_s}, \\ & \textbf{W}_{r,4}^{(t)}=\textbf{H}_{r,2}^{(t)}\textbf{F}^{(t)}\textbf{G}_2^{(t)}\textbf{F}^{(t)^H}\textbf{H}_{r,2}^{(t)^H}+ (N_sp_2\sigma_{e,2}^2+\sigma_{n,2}^2)\textbf{I}_{N_s}. \end{split}$$ By solving $\frac{\partial J(\textbf{R}_1^{(t)},\textbf{R}_2^{(t)})} {\partial \textbf{R}_i^{(t)}}=\textbf{0}_{N_s}$ for $i \in \{ 1,2 \}$, the receive beamforming matrices at the sources are given by $$\begin{split} \textbf{R}_l^{(t)} = \alpha^{(t)}p_{\bar{l}} \{\textbf{H}_{r,l}^{(t)}\textbf{F}^{(t)}\textbf{G}_l^{(t)}\textbf{F}^{(t)^H}\textbf{H}_{r,l}^{(t)^H} +(N_sp_l\sigma_{e,l}^2+\sigma_{n,l}^2)\textbf{I}_{N_s}\}^{-1} \textbf{H}_{r,l}^{(t)}\textbf{F}^{(t)}\textbf{H}_{\bar{l

textbf{R}_2^{(t)}\textbf{R}_2^{(t)^H }) \ } \label{Lag2 } \end{split}$$ where $ $ \begin{split } & \textbf{W}_{r,1}^{(t)}=\textbf{H}_{r,1}^{(t)}\textbf{F}^{(t)}\textbf{H}_{2,r}^{(t-1)^H }, \\ & \textbf{W}_{r,2}^{(t)}=\textbf{H}_{r,2}^{(t)}\textbf{F}^{(t)}\textbf{H}_{1,r}^{(t-1)^H }, \\ & \textbf{W}_{r,3}^{(t)}=\textbf{H}_{r,1}^{(t)}\textbf{F}^{(t)}\textbf{G}_1^{(t)}\textbf{F}^{(t)^H}\textbf{H}_{r,1}^{(t)^H}+ (N_sp_1\sigma_{e,1}^2+\sigma_{n,1}^2)\textbf{I}_{N_s }, \\ & \textbf{W}_{r,4}^{(t)}=\textbf{H}_{r,2}^{(t)}\textbf{F}^{(t)}\textbf{G}_2^{(t)}\textbf{F}^{(t)^H}\textbf{H}_{r,2}^{(t)^H}+ (N_sp_2\sigma_{e,2}^2+\sigma_{n,2}^2)\textbf{I}_{N_s }. \end{split}$$ By solving $ \frac{\partial J(\textbf{R}_1^{(t)},\textbf{R}_2^{(t) }) } { \partial \textbf{R}_i^{(t)}}=\textbf{0}_{N_s}$ for $ i \in \ { 1,2 \}$, the receive beamforming matrices at the sources are give by $ $ \begin{split } \textbf{R}_l^{(t) } = \alpha^{(t)}p_{\bar{l } } \{\textbf{H}_{r, l}^{(t)}\textbf{F}^{(t)}\textbf{G}_l^{(t)}\textbf{F}^{(t)^H}\textbf{H}_{r, l}^{(t)^H } + (N_sp_l\sigma_{e, l}^2+\sigma_{n, l}^2)\textbf{I}_{N_s}\}^{-1 } \textbf{H}_{r, l}^{(t)}\textbf{F}^{(t)}\textbf{H}_{\bar{l

texhbf{R}_2^{(t)}\textbf{R}_2^{(t)^H}) \} \label{Lag2} \ekd{split}$$ where $$\beyun{spliv} & \textbr{W}_{r,1}^{(t)}=\textcf{H}_{r,1}^{(t)}\textbf{F}^{(t)}\textbf{H}_{2,r}^{(t-1)^H}, \\ & \texvbf{W}_{e,2}^{(t)}=\texuyf{H}_{r,2}^{(t)}\textbf{F}^{(t)}\textbf{H}_{1,f}^{(t-1)^H}, \\ & \texthf{W}_{r,3}^{(t)}=\texrbf{H}_{c,1}^{(t)}\textbf{F}^{(t)}\textbf{J}_1^{(f)}\textbf{N}^{(c)^H}\texfnf{H}_{r,1}^{(t)^K}+ (N_wp_1\sigma_{e,1}^2+\sigma_{n,1}^2)\jextbf{I}_{N_s}, \\ & \taxtbf{W}_{r,4}^{(t)}=\textbf{V}_{r,2}^{(g)}\tzxtbf{F}^{(t)}\textbf{G}_2^{(t)}\textbf{F}^{(t)^H}\textbf{H}_{r,2}^{(t)^H}+ (N_s[_2\sigma_{e,2}^2+\xihma_{n,2}^2)\textbf{I}_{N_s}. \egd{spkyt}$$ Bg solving $\frac{\partial J(\textbf{R}_1^{(t)},\texfbf{R}_2^{(t)})} {\pertial \textbf{R}_i^{(y)}}=\textbf{0}_{N_s}$ for $i \in \{ 1,2 \}$, the gecelve beamforming mahrices at tye sjyrces are gixen by $$\begpu{split} \textbr{R}_l^{(t)} = \alpha^{(t)}p_{\bar{l}} \{\textbf{H}_{r,l}^{(t)}\textbw{F}^{(t)}\tertbf{G}_l^{(t)}\textvf{D}^{(t)^H}\hgxtbf{H}_{r,l}^{(t)^H} +(N_s'_l\sigmw_{e,l}^2+\sigma_{n,l}^2)\tewnbf{I}_{N_s}\}^{-1} \taxtbf{H}_{r,k}^{(t)}\textbf{F}^{(t)}\textnf{H}_{\bac{l

textbf{R}_2^{(t)}\textbf{R}_2^{(t)^H}) \} \label{Lag2} \end{split}$$ where $$\begin{split} & & \\ & (N_sp_1\sigma_{e,1}^2+\sigma_{n,1}^2)\textbf{I}_{N_s}, \\ & $\frac{\partial {\partial \textbf{R}_i^{(t)}}=\textbf{0}_{N_s}$ for \in \{ 1,2 the receive beamforming matrices at the are given by $$\begin{split} \textbf{R}_l^{(t)} = \alpha^{(t)}p_{\bar{l}} \{\textbf{H}_{r,l}^{(t)}\textbf{F}^{(t)}\textbf{G}_l^{(t)}\textbf{F}^{(t)^H}\textbf{H}_{r,l}^{(t)^H} +(N_sp_l\sigma_{e,l}^2+\sigma_{n,l}^2)\textbf{I}_{N_s}\}^{-1} \textbf{H}_{r,l}^{(t)}\textbf{F}^{(t)}\textbf{H}_{\bar{l

textbf{R}_2^{(t)}\textbf{R}_2^{(t)^H}) \} \label{Lag2} \End{split}$$ whEre $$\beGin{SplIt} & \TextBf{W}_{r,1}^{(T)}=\textbf{H}_{r,1}^{(t)}\textBF{F}^{(t)}\tExtbf{H}_{2,r}^{(t-1)^H}, \\ & \textbf{W}_{r,2}^{(t)}=\textBf{H}_{r,2}^{(t)}\TeXTbf{F}^{(T)}\TeXtbf{H}_{1,R}^{(t-1)^H}, \\ & \textBF{W}_{R,3}^{(T)}=\TexTbF{H}_{R,1}^{(t)}\tExTBf{f}^{(t)}\texTbf{g}_1^{(t)}\textbF{F}^{(t)^H}\textbf{h}_{r,1}^{(t)^h}+ (N_Sp_1\sigma_{e,1}^2+\sigmA_{N,1}^2)\tExtbf{I}_{N_s}, \\ & \teXtbF{W}_{r,4}^{(t)}=\textbf{H}_{r,2}^{(T)}\teXtbf{F}^{(t)}\TeXtbF{g}_2^{(t)}\texTbf{f}^{(t)^H}\teXtbf{H}_{r,2}^{(T)^h}+ (N_sp_2\siGma_{e,2}^2+\sigma_{N,2}^2)\tEXtbf{I}_{N_S}. \End{spliT}$$ bY sOlviNg $\frac{\partial J(\texTBf{r}_1^{(T)},\textbf{R}_2^{(t)})} {\partiAl \textBf{r}_I^{(t)}}=\TEXtbF{0}_{N_s}$ For $i \in \{ 1,2 \}$, the rEcEive bEAmformiNG mATRIceS At the sources aRe given by $$\beGIn{sPlit} \teXtBf{R}_L^{(T)} = \alpha^{(T)}p_{\bar{L}} \{\tEXtbF{H}_{r,l}^{(t)}\textbf{f}^{(t)}\teXtbf{G}_l^{(t)}\teXtbf{F}^{(t)^h}\Textbf{H}_{R,L}^{(t)^H} +(N_sp_l\Sigma_{e,L}^2+\siGma_{N,l}^2)\teXTbF{I}_{n_s}\}^{-1} \tExTBf{H}_{R,L}^{(t)}\TexTBf{F}^{(T)}\textbf{H}_{\BaR{l

textbf{R}_2^{(t)}\textbf{R }_2^{(t)^H }) \} \l abe l{ Lag2 } \e nd{split}$$ wh e re $ $\begin{split} & \text bf{W} _{ r ,1}^ { (t )}=\t extbf{H } _{ r , 1}^ {( t) }\t ex t bf {F}^{ (t) }\textb f{H}_{2,r} ^{( t- 1)^H}, \\ &\ te xtbf{W}_{r ,2} ^{(t)}=\text bf{ H}_{r, 2} ^{( t )}\te xtb f{F}^ {(t)}\ t extbf{ H}_{1,r}^ {( t -1)^H} , \\ & \ t e xt bf{W }_{r,3}^{(t)}=\te x tb f {H}_{r,1}^{(t) }\text bf { F} ^ { (t) }\t extbf{G}_1 ^{ (t)}\ t extbf{F } ^{ ( t ) ^H} \ textbf{H}_{r, 1}^{(t)^H}+ (N_ sp_1\s ig ma_ { e,1}^2 +\sig ma _ {n, 1}^2)\textb f{I} _{N_s}, \ \ & \t e xtbf{W} _ {r,4}^{ (t)}=\ tex tbf {H}_ { r, 2} ^{( t) } \te x tb f{F } ^{( t)}\text bf {G }_2^{ (t)} \ t e x tbf{ F}^ {(t) ^H}\t extbf{H}_{r,2 }^{ (t)^ H }+(N_sp _2\si gma_ {e ,2}^2 +\sigm a_{n, 2} ^2)\textbf{I}_{ N_s} . \end{sp lit }$ $ B ysolvi n g $\fr ac{ \pa rtial J (\textb f {R} _1 ^ { ( t) },\textbf{R}_2^{(t )} ) } { \partial \text b f{ R} _ i^{(t)}} =\ tex tbf{ 0 } _{N_s }$ f o r$i \in \ { 1,2\ }$ ,the rec ei ve bea mf orm ing matr i cesat the sources areg iven by $$\beg i n{split} \tex t bf { R }_ l ^{(t )}= \alpha^{( t)}p _ {\ba r{l} } \ {\t e xtbf{ H}_{r ,l } ^{ ( t)}\textbf{F}^{(t)} \t extbf{ G}_l^ {(t)}\textbf{ F}^{(t)^H} \ t e xtbf{H}_ {r,l } ^{ ( t)^H} +(N_sp_l \sigm a_{e,l}^2+ \ sigma_{n ,l}^2 )\textbf {I}_{N_s} \ } ^{-1} \t ext bf{ H}_ {r, l } ^{ (t)}\textbf{F } ^ {(t) }\ textbf{ H}_ {\bar{l

textbf{R}_2^{(t)}\textbf{R}_2^{(t)^H}) \} \label{Lag2} \end{split}$$_where $$\begin{split} &_\textbf{W}_{r,1}^{(t)}=\textbf{H}_{r,1}^{(t)}\textbf{F}^{(t)}\textbf{H}_{2,r}^{(t-1)^H}, \\ & \textbf{W}_{r,2}^{(t)}=\textbf{H}_{r,2}^{(t)}\textbf{F}^{(t)}\textbf{H}_{1,r}^{(t-1)^H}, \\ &_\textbf{W}_{r,3}^{(t)}=\textbf{H}_{r,1}^{(t)}\textbf{F}^{(t)}\textbf{G}_1^{(t)}\textbf{F}^{(t)^H}\textbf{H}_{r,1}^{(t)^H}+ (N_sp_1\sigma_{e,1}^2+\sigma_{n,1}^2)\textbf{I}_{N_s}, \\ &_\textbf{W}_{r,4}^{(t)}=\textbf{H}_{r,2}^{(t)}\textbf{F}^{(t)}\textbf{G}_2^{(t)}\textbf{F}^{(t)^H}\textbf{H}_{r,2}^{(t)^H}+ (N_sp_2\sigma_{e,2}^2+\sigma_{n,2}^2)\textbf{I}_{N_s}. \end{split}$$_By solving_$\frac{\partial_J(\textbf{R}_1^{(t)},\textbf{R}_2^{(t)})} {\partial \textbf{R}_i^{(t)}}=\textbf{0}_{N_s}$_for $i \in_\{ 1,2 \}$, the_receive beamforming matrices_at_the sources are given by $$\begin{split} \textbf{R}_l^{(t)} = \alpha^{(t)}p_{\bar{l}} \{\textbf{H}_{r,l}^{(t)}\textbf{F}^{(t)}\textbf{G}_l^{(t)}\textbf{F}^{(t)^H}\textbf{H}_{r,l}^{(t)^H} +(N_sp_l\sigma_{e,l}^2+\sigma_{n,l}^2)\textbf{I}_{N_s}\}^{-1} \textbf{H}_{r,l}^{(t)}\textbf{F}^{(t)}\textbf{H}_{\bar{l

\Gamma\left( \frac{i(-\beta_s+\beta_t)}{6\pi}+\frac{1}{3} \right)}\nonumber\\ &\times& \prod_{s=1}^{2N}\prod_{j=0}^{j_r} \int_{-\infty}^\infty \frac{d\alpha_{j,s}}{2\pi i}~ \widetilde{\Phi}(\{\alpha\}|\{\beta\})~ \prod_{s=1}^{2N}\widetilde{g}_{j_s} (\{\alpha_{j,s}\}_{j=-1}^{j_s}) \nonumber\\ &\times& \prod_{s=1}^{2N} \prod_{j=0,1} \prod_{t \in {\cal A}_j \atop{s\leq t}} (-\beta_s+\alpha_{j,t}+\pi i) \prod_{t \in {\cal A}_j \atop{t<s}} (-\alpha_{j,t}+\beta_s+\pi i) \nonumber\\ &\times& \prod_{j=0,1} \prod_{s,t \in {\cal A}_j \atop{s<t}} \frac{\alpha_{j,s}-\alpha_{j,t}-\pi i}{ \alpha_{j,s}-\alpha_{j,t}-2\pi i } \prod_{s \in {\cal A}_0,t \in {\cal A}_1 \atop{s\leq t}} \frac{\alpha_{0,s}-\alpha_{1,t}-\pi i} {\alpha_{0,s}-\alpha_{1,t}-2\pi i } \prod_{s \in {\cal A}_0,t \in {\cal A}_1 \atop{s>t}} \frac{ \alpha_{1,t}-\alpha_{0,s}-\pi i }{ \alpha_{1,t}-\alpha_{0,s}-2\pi i } \nonumber\\ &\times& \prod_{j=0,1} \prod_{s,t \in {\cal A}_j \atop{s<t}} {\rm sh}\left(\frac

\Gamma\left ( \frac{i(-\beta_s+\beta_t)}{6\pi}+\frac{1}{3 } \right)}\nonumber\\ & \times & \prod_{s=1}^{2N}\prod_{j=0}^{j_r } \int_{-\infty}^\infty \frac{d\alpha_{j, s}}{2\pi i}~ \widetilde{\Phi}(\{\alpha\}|\{\beta\})~ \prod_{s=1}^{2N}\widetilde{g}_{j_s } (\{\alpha_{j, s}\}_{j=-1}^{j_s }) \nonumber\\ & \times & \prod_{s=1}^{2N } \prod_{j=0,1 } \prod_{t \in { \cal A}_j \atop{s\leq t } } (-\beta_s+\alpha_{j, t}+\pi i) \prod_{t \in { \cal A}_j \atop{t < s } } (-\alpha_{j, t}+\beta_s+\pi i) \nonumber\\ & \times & \prod_{j=0,1 } \prod_{s, t \in { \cal A}_j \atop{s < t } } \frac{\alpha_{j, s}-\alpha_{j, t}-\pi i } { \alpha_{j, s}-\alpha_{j, t}-2\pi i } \prod_{s \in { \cal A}_0,t \in { \cal A}_1 \atop{s\leq t } } \frac{\alpha_{0,s}-\alpha_{1,t}-\pi i } { \alpha_{0,s}-\alpha_{1,t}-2\pi i } \prod_{s \in { \cal A}_0,t \in { \cal A}_1 \atop{s > t } } \frac { \alpha_{1,t}-\alpha_{0,s}-\pi i } { \alpha_{1,t}-\alpha_{0,s}-2\pi i } \nonumber\\ & \times & \prod_{j=0,1 } \prod_{s, t \in { \cal A}_j \atop{s < t } } { \rm sh}\left(\frac

\Gamla\left( \frac{i(-\beta_s+\beta_t)}{6\pi}+\fvac{1}{3} \right)}\nonumber\\ &\jines& \prov_{s=1}^{2N}\prod_{n=0}^{j_r} \int_{-\inwty}^\infty \frac{d\alpha_{j,s}}{2\pi i}~ \widevildw{\Phi}(\{\aopha\}|\{\beta\})~ \prod_{s=1}^{2N}\widetildd{g}_{j_s} (\{\alpha_{u,s}\}_{j=-1}^{j_s}) \nonymbec\\ &\times& \prod_{s=1}^{2N} \prod_{o=0,1} \lrod_{t \ik {\cal W}_j \atm'{s\leq t}} (-\beta_s+\alpma_{j,t}+\pi i) \prog_{t \in {\cal A}_j \atmp{g<s}} (-\clpha_{j,t}+\beta_s+\pi i) \nonumber\\ &\times& \prod_{j=0,1} \prjd_{s,t \in {\cwl A}_j \atop{s<t}} \fras{\alpnw_{j,s}-\amihc_{j,t}-\pi i}{ \alpha_{j,s}-\alpha_{j,t}-2\pi i } \prod_{s \in {\cal A}_0,t \in {\cal A}_1 \ayop{s\leq t}} \frac{\alpha_{0,s}-\alpha_{1,t}-\pl i} {\appha_{0,s}-\alpha_{1,t}-2\pi i } \prof_{s \in {\cal A}_0,j \in {\xal A}_1 \atop{s>t}} \ffac{ \alpha_{1,t}-\alpha_{0,s}-\pi i }{ \alpga_{1,t}-\alpha_{0,s}-2\pi i } \nonumber\\ &\times& \prod_{g=0,1} \prod_{x,t \in {\cal Q}_j \qtoo{v<t}} {\rm sh}\left(\hrac

\Gamma\left( \frac{i(-\beta_s+\beta_t)}{6\pi}+\frac{1}{3} \right)}\nonumber\\ &\times& \prod_{s=1}^{2N}\prod_{j=0}^{j_r} \int_{-\infty}^\infty \frac{d\alpha_{j,s}}{2\pi \prod_{s=1}^{2N}\widetilde{g}_{j_s} \nonumber\\ &\times& \prod_{j=0,1} \prod_{t \in i) \in {\cal A}_j (-\alpha_{j,t}+\beta_s+\pi i) \nonumber\\ \prod_{j=0,1} \prod_{s,t \in {\cal A}_j \atop{s<t}} i}{ \alpha_{j,s}-\alpha_{j,t}-2\pi i } \prod_{s \in {\cal A}_0,t \in {\cal A}_1 \atop{s\leq t}} i} {\alpha_{0,s}-\alpha_{1,t}-2\pi i } \prod_{s \in {\cal A}_0,t \in {\cal A}_1 \atop{s>t}} \frac{ i \alpha_{1,t}-\alpha_{0,s}-2\pi } &\times& \prod_{j=0,1} \prod_{s,t \in {\cal A}_j \atop{s<t}} {\rm sh}\left(\frac

\Gamma\left( \frac{i(-\beta_s+\beta_t)}{6\pI}+\frac{1}{3} \right)}\NonumBer\\ &\TimEs& \Prod_{S=1}^{2N}\prOd_{j=0}^{j_r} \int_{-\infty}^\iNFty \fRac{d\alpha_{j,s}}{2\pi i}~ \widetildE{\Phi}(\{\aLpHA\}|\{\betA\})~ \PrOd_{s=1}^{2N}\wIdetildE{G}_{j_S} (\{\ALphA_{j,S}\}_{j=-1}^{J_s}) \nOnUMbEr\\ &\timEs& \pRod_{s=1}^{2N} \prOd_{j=0,1} \prod_{t \in {\Cal a}_j \Atop{s\leq t}} (-\betA_S+\aLpha_{j,t}+\pi i) \pRod_{T \in {\cal A}_j \atop{T<s}} (-\aLpha_{j,t}+\BeTa_s+\PI i) \nonUmbEr\\ &\timEs& \prod_{J=0,1} \Prod_{s,t \In {\cal A}_j \atOp{S<T}} \frac{\aLPha_{j,s}-\alPHA_{j,T}-\pi i}{ \Alpha_{j,s}-\alpha_{j,t}-2\pi i } \PRoD_{S \in {\cal A}_0,t \in {\cal A}_1 \Atop{s\lEq T}} \FrAC{\AlpHa_{0,s}-\Alpha_{1,t}-\pi i} {\aLpHa_{0,s}-\alPHa_{1,t}-2\pi i } \pROd_{S \IN {\Cal a}_0,T \in {\cal A}_1 \atop{s>t}} \Frac{ \alpha_{1,t}-\aLPha_{0,S}-\pi i }{ \alPhA_{1,t}-\aLPha_{0,s}-2\pi I } \nonuMbER\\ &\tiMes& \prod_{j=0,1} \proD_{s,t \iN {\cal A}_j \atoP{s<t}} {\rm sH}\Left(\fraC

\Gamma\left( \frac{i(-\be ta_s+\beta _t)}{ 6\p i}+ \f rac{ 1}{3 } \right)}\non u mber \\ &\times& \prod_{s=1 }^{2N }\ p rod_ { j= 0}^{j _r} \in t _{ - \ inf ty }^ \in ft y \ frac{ d\a lpha_{j ,s}}{2\pii}~ \ widetilde{\P h i} (\{\alpha\ }|\ {\beta\})~ \ pro d_{s=1 }^ {2N } \wide til de{g} _{j_s} (\{\al pha_{j,s} \} _ {j=-1} ^ {j_s})\ n on umbe r\\ &\times& \pro d _{ s =1}^{2N} \prod _{j=0, 1} \p r o d_{ t \ in {\cal A }_ j \at o p{s\leq t} } ( -\b e ta_s+\alpha_{ j,t}+\pi i) \pr od_{t\i n { \ cal A} _j \a to p {t< s}} (-\alph a_{j ,t}+\beta _s+\pi i) \non u mber\\&\time s&\pr od_{ j =0 ,1 } \ pr o d_{ s ,t \i n {\ cal A}_j \ at op{s< t}}\ f r a c{\a lph a_{j ,s}-\ alpha_{j,t}-\ pii}{\ alp ha_{j ,s}-\ alph a_ {j,t} -2\pii }\p rod_{s \in {\ca l A} _0,t \in{\c al A} _1 \ato p {s\leq t} } \ frac{\a lpha_{0 , s}- \a l p h a_ {1,t}-\pi i} {\alp ha _ { 0, s}-\alph a_{1,t } -2 \p i i } \pr od _{s \in { \calA}_0 , t\in {\ca l A}_1 \a to p{s>t}} \ frac{\a lph a_{ 1,t}- \ alph a_{0,s }-\pi i}{ \a l pha_{1,t}-\alp h a_{0,s}-2\pii } \n o numb er\ \ &\times&\pro d _{j= 0,1} \p rod _ {s,t\in { \c a lA }_j \atop{s<t}} {\r msh}\le ft(\f rac

\Gamma\left( \frac{i(-\beta_s+\beta_t)}{6\pi}+\frac{1}{3} \right)}\nonumber\\ &\times& \prod_{s=1}^{2N}\prod_{j=0}^{j_r} \int_{-\infty}^\infty \frac{d\alpha_{j,s}}{2\pi i}~ \widetilde{\Phi}(\{\alpha\}|\{\beta\})~ \prod_{s=1}^{2N}\widetilde{g}_{j_s} (\{\alpha_{j,s}\}_{j=-1}^{j_s}) \nonumber\\ &\times& \prod_{s=1}^{2N} \prod_{j=0,1} \prod_{t_\in {\cal_A}_j \atop{s\leq t}} (-\beta_s+\alpha_{j,t}+\pi i) \prod_{t \in_{\cal A}_j \atop{t<s}} (-\alpha_{j,t}+\beta_s+\pi_i) \nonumber\\ &\times& \prod_{j=0,1} \prod_{s,t_\in {\cal_A}_j \atop{s<t}} \frac{\alpha_{j,s}-\alpha_{j,t}-\pi_i}{ \alpha_{j,s}-\alpha_{j,t}-2\pi i } \prod_{s_\in {\cal A}_0,t_\in {\cal A}_1 \atop{s\leq t}} \frac{\alpha_{0,s}-\alpha_{1,t}-\pi_i} {\alpha_{0,s}-\alpha_{1,t}-2\pi i } \prod_{s \in_{\cal_A}_0,t \in {\cal A}_1 \atop{s>t}} \frac{ \alpha_{1,t}-\alpha_{0,s}-\pi i }{ \alpha_{1,t}-\alpha_{0,s}-2\pi i } \nonumber\\ &\times& \prod_{j=0,1} \prod_{s,t \in {\cal A}_j \atop{s<t}} {\rm sh}\left(\frac

1\rangle}\ : \ L F_n\circ \dots\circ L F_1 \ \Longrightarrow \ L(F_n\circ\dots\circ F_1)$$ inductively by setting $${\langle F_n,F_{n-1},\dots, F_1\rangle}\ = \ {\langle F_n F_{n-1},F_{n-2},\dots, F_1\rangle}\circ ({\langle F_n,F_{n-1}\rangle}\star (L F_{n-1}\circ\dots\circ L F_1))\,$$ assuming that all iterated composites of adjacent functors are left derivable. Proposition \[prop:derived cocycle\] implies that we could have instead ‘spliced’ at any other intermediate spot in the composition; more generally, we could have ‘collected adjacent factors’ in any way we like, and get the same result. More precisely, if for $i\leq j$ we write $$F_{[j,i]}\ = \ F_j\circ F_{j-1}\circ\dots\circ F_i\,$$ and we set ${\langle F_k\rangle}=\operatorname{Id}_{L F_k}$, then for all sequences of numbers $1\leq m_1< m_2 <\dots < m_k < n$, we have $$\begin{aligned} &{\langle F_n,\dots, F_1\rangle}\ = \\ &\quad{\langle F_{[n,m_k+1]},\dots, F_{[m_2,m_1+1]}, F_{[m_1,1]}\rangle} \circ \left( {\langle F_n,\dots,F_{m_k+1}\rangle}\star \dots\star{\langle F_{m_1},\dots,F_1\rangle}\right) \. \end{aligned}$$ The next Theorem \[thm:weak equivalence\] is a homotopy invariance statement for proper equivariant stable homotopy theory; it says, roughly speaking, that this homotopy theory only depends on the Lie group ‘up to multiplicative weak equivalence’. In the context of [*compact*]{} Lie groups, this statement does not have much content: as we recall in the next proposition, every multiplicative weak equivalence between compact Lie groups is already an isomorphism. \[prop:compact rigid\] Let $\alpha:K{\longrightarrow}

1\rangle}\: \ L F_n\circ \dots\circ L F_1 \ \Longrightarrow \ L(F_n\circ\dots\circ F_1)$$ inductively by setting $ $ { \langle F_n, F_{n-1},\dots, F_1\rangle}\ = \ { \langle F_n F_{n-1},F_{n-2},\dots, F_1\rangle}\circ ({ \langle F_n, F_{n-1}\rangle}\star (L F_{n-1}\circ\dots\circ L F_1))\,$$ assuming that all iterated composites of adjacent functors are leave derivable. Proposition \[prop: derive cocycle\ ] implies that we could have instead ‘ spliced ’ at any early intermediate spot in the composition; more by and large, we could have ‘ collected adjacent factors ’ in any manner we like, and get the like result. More precisely, if for $ i\leq j$ we write $ $ F_{[j, i]}\ = \ F_j\circ F_{j-1}\circ\dots\circ F_i\,$$ and we set $ { \langle F_k\rangle}=\operatorname{Id}_{L F_k}$, then for all sequences of issue $ 1\leq m_1 < m_2 < \dots < m_k < n$, we have $ $ \begin{aligned } & { \langle F_n,\dots, F_1\rangle}\ = \\ & \quad{\langle F_{[n, m_k+1]},\dots, F_{[m_2,m_1 + 1 ] }, F_{[m_1,1]}\rangle } \circ \left ({ \langle F_n,\dots, F_{m_k+1}\rangle}\star \dots\star{\langle F_{m_1},\dots, F_1\rangle}\right) \. \end{aligned}$$ The next Theorem \[thm: weak equivalence\ ] is a homotopy invariance statement for proper equivariant static homotopy theory; it says, roughly talk, that this homotopy theory only depend on the Lie group ‘ up to multiplicative weak equality ’. In the context of [ * compact * ] { } Lie groups, this statement does not have much content: as we echo in the next proposition, every multiplicative weak equivalence between compact Lie groups is already an isomorphism. \[prop: compact rigid\ ] Let $ \alpha: K{\longrightarrow }

1\ranhle}\ : \ L F_n\circ \dots\circ U F_1 \ \Longrightatriw \ L(F_i\circ\dofs\circ F_1)$$ inductively by setting $${\langpe F_n,F_{n-1},\eots, F_1\rangle}\ = \ {\langle F_n F_{n-1},F_{n-2},\dlts, F_1\rantle}\cmrc ({\langle F_n,F_{n-1}\raifle}\star (L F_{n-1}\cjvc\dotv\rirc L F_1))\,$$ assumikg that all iterated compmsktzs of adjacent functors are left deryvable. Lrlposition \[prop:qerineq codjcoe\] implies that we could have insteav ‘spliced’ at anu other intermediate spot ln tje composition; morf generally, we siuld have ‘coulected adjacent factots’ in any way we like, and get thd samz result. Motz prfwisely, if fir $i\lvq j$ we write $$F_{[j,i]}\ = \ F_b\circ F_{k-1}\circ\dots\circ N_i\,$$ anv we set ${\langle F_k\rangle}=\o'eratorname{Id}_{L F_k}$, thgn for all szquences of numbers $1\lwq m_1< m_2 <\gots < m_k < n$, de gate $$\gegin{apigied} &{\langle R_n,\dots, F_1\rantle}\ = \\ &\quad{\langle F_{[m,m_h+1]},\eots, F_{[m_2,m_1+1]}, F_{[m_1,1]}\rahgle} \ciwc \jeft( {\langle F_n,\dots,F_{m_k+1}\rangle}\star \dots\stag{\lanfle F_{m_1},\dots,F_1\rangle}\right) \. \end{aligned}$$ The next Jheorem \[thi:weak equivalence\] is a homotopy invariance statemant fkf pxiper dwulvariant stable homotopy theory; it says, roughjg xpvaking, that this momotopy theory onky drkends on the Lke grobl ‘hp to multiplicatige weak equicalence’. Ig thr context of [*compact*]{} Lie griups, this stctenent does not have much conteut: as ee revall in the next proposntion, svery multiolicative deak equivalence benweet compacu Lie groups is alweady an msomoxphism. \[prup:cokpact wigid\] Let $\wlpha:K{\longrightarrow}

1\rangle}\ : \ L F_n\circ \dots\circ L \Longrightarrow L(F_n\circ\dots\circ F_1)$$ by setting $${\langle F_n F_1\rangle}\circ ({\langle F_n,F_{n-1}\rangle}\star F_{n-1}\circ\dots\circ L F_1))\,$$ that all iterated composites of adjacent are left derivable. Proposition \[prop:derived cocycle\] implies that we could have instead ‘spliced’ any other intermediate spot in the composition; more generally, we could have ‘collected factors’ any we and get the same result. More precisely, if for $i\leq j$ we write $$F_{[j,i]}\ = \ F_{j-1}\circ\dots\circ F_i\,$$ and we set ${\langle F_k\rangle}=\operatorname{Id}_{L F_k}$, for all sequences of $1\leq m_1< m_2 <\dots < < we have &{\langle F_1\rangle}\ \\ &\quad{\langle F_{[n,m_k+1]},\dots, F_{[m_1,1]}\rangle} \circ \left( {\langle F_n,\dots,F_{m_k+1}\rangle}\star \dots\star{\langle F_{m_1},\dots,F_1\rangle}\right) \. \end{aligned}$$ The next Theorem \[thm:weak equivalence\] is a homotopy statement for stable homotopy it roughly that this homotopy depends on the Lie group ‘up equivalence’. In the context of [*compact*]{} Lie groups, statement does have much content: as we recall the next proposition, every multiplicative weak equivalence between Lie groups is already an isomorphism. \[prop:compact rigid\] Let $\alpha:K{\longrightarrow}

1\rangle}\ : \ L F_n\circ \dots\circ L F_1 \ \LoNgrightarrOw \ L(F_n\CirC\doTs\Circ f_1)$$ indUctively by settINg $${\laNgle F_n,F_{n-1},\dots, F_1\rangle}\ = \ {\lanGle F_n f_{n-1},f_{N-2},\dotS, f_1\rAngle}\Circ ({\lanGLe f_N,f_{n-1}\rAnGlE}\stAr (l f_{n-1}\Circ\dOts\Circ L F_1))\,$$ aSsuming thaT alL iTerated compoSItEs of adjaceNt fUnctors are leFt dErivabLe. proPOsitiOn \[pRop:deRived cOCycle\] iMplies thaT wE Could hAVe insteAD ‘SpLiceD’ at any other intermEDiATe spot in the comPositiOn; MOrE GEneRalLy, we could hAvE ‘collECted adjACeNT FActORs’ in any way we lIke, and get thE SamE resulT. MOre PReciseLy, if fOr $I\Leq J$ we write $$F_{[j,i]}\ = \ f_j\ciRc F_{j-1}\circ\dOts\cirC f_i\,$$ and we SEt ${\langlE F_k\ranGle}=\OpeRatoRNaMe{id}_{L f_k}$, THen FOr All SEquEnces of nUmBeRs $1\leq M_1< m_2 <\doTS < M_K < N$, we hAve $$\BegiN{aligNed} &{\langle F_n,\doTs, F_1\RangLE}\ = \\ &\quAd{\lanGle F_{[n,M_k+1]},\doTs, f_{[m_2,m_1+1]}, F_{[m_1,1]}\Rangle} \Circ \lEfT( {\langle F_n,\dots,F_{m_K+1}\ranGle}\star \doTs\sTaR{\laNgLe F_{m_1},\dOTs,F_1\ranGle}\RigHt) \. \end{alIgned}$$ ThE NexT THEOReM \[thm:weak equivalencE\] iS A HoMotopy inVarianCE sTaTEment for PrOpeR equIVArianT staBLe Homotopy Theory; IT sAyS, roughlY sPeakinG, tHat ThiS homoTOpy tHeory oNly depenDs on tHE Lie group ‘up to mULtiplicative wEAk EQUiVAlenCe’. IN the context Of [*coMPact*]{} lie gROuPs, tHIs staTemenT dOEs NOt have much content: as We Recall In the Next propositiOn, every mulTIPLicative Weak EQuIValence between CompaCt Lie groupS Is alreadY an isOmorphisM. \[prop:compACT rigid\] LeT $\alPha:k{\loNgrIGHtArrow}

1\rangle}\ : \ L F_n\circ\dots\circ L F_ 1 \ \L on grig htar row \ L(F_n\c i rc\d ots\circ F_1)$$ induct ively b y set t in g $${ \langle F_ n , F_{ n- 1} ,\d ot s ,F_1\r ang le}\ =\ {\lang leF_ n F_{n-1},F_ { n- 2},\dots,F_1 \rangle}\cir c ( {\lang le F_ n ,F_{n -1} \rang le}\st a r (L F _{n-1}\ci rc \ dots\c i rc L F_ 1 ) )\ ,$$assuming that all it e rated composit es ofad j ac e n t f unc tors are l ef t der i vable.P ro p o s iti o n \[prop:deri ved cocycle \ ] i mplies t hat we cou ld ha ve ins tead ‘splic ed’at any ot her in t ermedia t e spotin the co mpo siti o n; m ore g e ner a ll y,w e c ould hav e‘c ollec teda d j a cent fa ctor s’ in any way we l ike , an d ge t the same res ul t. Mo re pre cisel y, if for $i\leqj$ w e write $ $F_ {[ j,i ]} \ = \ F_j\ci rcF_{ j-1}\ci rc\dots \ cir cF _ i \, $$ and we set ${\l an g l eF_k\rang le}=\o p er at o rname{Id }_ {LF_k} $ , then for al l sequen ces of nu mb ers $1\ le q m_1< m _2<\d ots < m_k< n$,we have$$\be g in{aligned} & { \langle F_n,\ d ot s , F _ 1\ra ngl e}\ = \\ &\qu a d{\l angl e F _{[ n ,m_k+ 1]},\ do t s, F_{[m_2,m_1+1]}, F_ {[ m_1,1] }\ran gle} \circ \l eft( {\lan g l e F_n,\do ts,F _ {m _ k+1}\rangle}\s tar \ dots\star{ \ langle F _{m_1 },\dots, F_1\rangl e } \right)\. \e nd{ ali g n ed }$$ The next T heor em \[thm: wea k equiv ale nce \]isahomotopyinvarian ce s ta te men t for proper e qu iva ri ant stab l e homo topytheo ry ;i t s ays, ro u gh l y spe ak in g, t hat t his h omot o pytheoryonly depe nds on t he L ie grou p ‘up to mult ip licative w ea k e quival e n ce’. Inthe context of [*compac t *]{} Li e g roups , th is statem ent doesnot have m uch co ntent :asw e reca l l i n t he next prop o s iti on, e ve ry m ultipli cative weak equiva l enc e between com pac t Li e gr oup s i s al re a dya n isomorphism. \[prop:com pa c trigid\] Le t $\ al pha:K{\ longrig htarr o w}

1\rangle}\ :_\ L_F_n\circ \dots\circ L F_1_\ \Longrightarrow_\_ L(F_n\circ\dots\circ F_1)$$_inductively_by setting $${\langle_F_n,F_{n-1},\dots, F_1\rangle}\ =_\ {\langle F_n_F_{n-1},F_{n-2},\dots, F_1\rangle}\circ ({\langle F_n,F_{n-1}\rangle}\star_(L_F_{n-1}\circ\dots\circ L F_1))\,$$ assuming that all iterated composites of adjacent functors are left derivable._Proposition_\[prop:derived cocycle\]_implies_that_we could have instead ‘spliced’_at any other intermediate spot_in the_composition; more generally, we could have ‘collected adjacent_factors’_in any way_we like, and get the same result. More precisely,_if for $i\leq j$ we write_$$F_{[j,i]}\ = \_F_j\circ_F_{j-1}\circ\dots\circ_F_i\,$$ and we set_${\langle F_k\rangle}=\operatorname{Id}_{L F_k}$, then for all_sequences of numbers $1\leq m_1< m_2_<\dots < m_k < n$, we have_$$\begin{aligned} &{\langle F_n,\dots, F_1\rangle}\ = \\_ &\quad{\langle F_{[n,m_k+1]},\dots, F_{[m_2,m_1+1]}, F_{[m_1,1]}\rangle} \circ \left(_{\langle F_n,\dots,F_{m_k+1}\rangle}\star_\dots\star{\langle F_{m_1},\dots,F_1\rangle}\right) \. \end{aligned}$$ The_next Theorem \[thm:weak_equivalence\] is_a homotopy invariance_statement for proper equivariant stable homotopy_theory; it says,_roughly speaking, that this homotopy theory_only_depends on the_Lie_group_‘up to_multiplicative weak equivalence’._In_the context_of_[*compact*]{} Lie groups, this statement does_not_have much content: as we recall in_the next proposition, every_multiplicative_weak equivalence between compact_Lie groups is already an_isomorphism. \[prop:compact rigid\] Let $\alpha:K{\longrightarrow}

by an applied magnetic field. The latter phase shows largely suppressed moments $\lesssim 0.1\mu_\mathrm{B}$ and gapped excitations. The $H-T$ phase diagram suggests the existence of a quantum critical point at the small but finite magnetic field $\mu_0 H_c=1.55(25)$ T. We discuss this finding in light of the perturbative Dzyaloshinskii-Moriya interaction which was theoretically proposed to sustain a quantum critical regime for the quantum kagomé Heisenberg antiferromagnet model.' author: - 'M. Jeong' - 'F. Bert' - 'P. Mendels' - 'F. Duc' - 'J. C. Trombe' - 'M. A. de Vries' - 'A. Harrison' title: 'Field-Induced Freezing of a Quantum Spin Liquid on the Kagomé Lattice' --- Quantum spin liquids (QSL) are appealing states of matter which do not break any symmetry of the spin Hamiltonian. While QSL behaviors are well established for one-dimensional spin systems, their existence in higher dimensions remains questionable. The quantum kagomé Heisenberg antiferromagnet (QKHA) is considered as the best candidate to stabilize such a QSL phase in two dimensions [@MisguichL05; @Balents10Nat; @LhuillierM10]. The specificity of this model relies on the unique combination of strong quantum fluctuations enhanced by low spins $S=1/2$ and high geometrical frustration of the lattice of corner-sharing triangles. Early numerical studies have shown, despite finite size limitations, that the ground state of the QKHA model does not support any simple order parameter [@ZengE90PRB; @Lecheminant97PRB; @Waldtmann98EJPB] and have evidenced an original excitation spectrum with dense sets of low energy excitations in all spin sectors [@Lecheminant97PRB; @Waldtmann98EJPB]. A recent state-of-the-art calculation [@Yan+10arxiv] favors a QSL ground state with a small gap much like the resonating-valence-bond state [@Sachdev92PRB; @Anderson73MRB]. Other recent proposals with similar ground-state energies encompass large unit-cell valence-bond-crystals [@SinghH07PRB]

by an applied magnetic field. The latter phase shows largely inhibit here and now $ \lesssim 0.1\mu_\mathrm{B}$ and gapped excitations. The $ H - T$ phase diagram hint the existence of a quantum critical point at the little but finite charismatic field $ \mu_0 H_c=1.55(25)$   T. We discuss this finding in light of the perturbative Dzyaloshinskii - Moriya interaction which was theoretically proposed to hold a quantum critical regime for the quantum kagomé Heisenberg antiferromagnet model.' generator: -' M. Jeong' -' F. Bert' -' P. Mendels' -' F. Duc' -' J. C. Trombe' -' M. A. de Vries' -' A. Harrison' claim:' Field - Induced Freezing of a Quantum Spin Liquid on the Kagomé Lattice' --- Quantum spin liquids (QSL) are appealing states of topic which do not break any symmetry of the spin Hamiltonian. While QSL behavior are well established for one - dimensional spin system, their universe in higher proportion remains questionable. The quantum kagomé Heisenberg antiferromagnet (QKHA) is considered as the best campaigner to stabilize such a QSL phase in two dimensions   [ @MisguichL05; @Balents10Nat; @LhuillierM10 ]. The specificity of this model relies on the unique combination of strong quantum fluctuation enhanced by low spins $ S=1/2 $ and high geometrical frustration of the lattice of corner - sharing triangles. Early numerical study have shown, despite finite size limitations, that the flat coat state of the QKHA model does not support any simple orderliness argument [ @ZengE90PRB; @Lecheminant97PRB; @Waldtmann98EJPB ] and have evidenced an original excitation spectrum with dense sets of gloomy energy excitations in all spin sectors [ @Lecheminant97PRB; @Waldtmann98EJPB ]. A late state - of - the - art calculation [ @Yan+10arxiv ] favors a QSL ground state with a small gap a lot like the resonate - valence - bond state [ @Sachdev92PRB; @Anderson73MRB ]. Other recent proposal with similar ground - state energy encompass large unit - cell valence - shackle - crystals [ @SinghH07PRB ]

by an applied magnetic fieud. The latter pkqse shmws ladgely suopressed moments $\lesssim 0.1\mu_\methrn{B}$ ane gapped excitations. Tfe $H-T$ phade diagrqm snggests the exisvsnce of a quahbum cxivical point at jhe small bud finite magnediz yield $\mu_0 H_c=1.55(25)$ T. We discuss this finding in lignt of the perturfatine Dzyzloshinskii-Moriya interaction whicg was tieoretically prpposed to sustain a quantul crltical regime for hhe quantum kagjné Heisenberg antiferromagnet model.' zuthor: - 'M. Jeong' - 'F. Bert' - 'P. Mendels' - 'F. Due' - 'J. C. Trombg' - 'J. W. de Vries' - 'A. Harrpson' title: 'Field-Induceg Freezong of a Quantmm Spmn Luquid on the Kagomé Lavtice' --- Quantum spin lizuids (QSL) axe appealing states od nattet whiwh du nog bdeek zny sylmevry of the apin Hamiltinian. While QSL behsvyirs are well sstablysred for one-dimensional spin systems, thepr esistence in higher dimebsions remains questilnable. Thq quantum kagomé Heisenberg antiferromagnet (QKHA) iv conakdexcq qs the best candidate to stabilize such a QSL przst ik two dimensions [@IisguichL05; @Bslfnyf10Nat; @LhuillietM10]. The spedificity of this mldel rejies in the unyque combination of strong quanrum fluctuatpons enhanced by low s'ins $S=1/2$ and hngh gepmetrocal frustration of the latfice of corjer-sharinf triangles. Early nukesical studies have shown, dqspite fiiite xize lioitajions, trat the grlund state of the QKHA mldel boes tot supporh any simple order parameter [@ZeijE90PRB; @Lechemimatt97PGB; @Waldtmcnn98EJPN] and have evidqnced an origiual excication spectrum sith deise sets of jow energy exwltations in ell spin fectirs [@Oechemivxnt97PRB; @Waldtmamn98EJPB]. A geeent statw-of-the-art calculatlon [@Yxh+10arxiv] favors a QSO ground state eitf a slakl cap much lika thd rdxonatkng-valence-bivd syate [@Sachdev92PRB; @Andesson73JRB]. Other recent ptoiosals wijh similaw ground-state energies encompasd larje univ-cell falgnce-bond-crystals [@SinghH07PRB]

by an applied magnetic field. The latter largely moments $\lesssim and gapped excitations. the of a quantum point at the but finite magnetic field $\mu_0 H_c=1.55(25)$ We discuss this finding in light of the perturbative Dzyaloshinskii-Moriya interaction which was proposed to sustain a quantum critical regime for the quantum kagomé Heisenberg antiferromagnet author: 'M. - Bert' - 'P. Mendels' - 'F. Duc' - 'J. C. Trombe' - 'M. A. de Vries' 'A. Harrison' title: 'Field-Induced Freezing of a Quantum Liquid on the Kagomé --- Quantum spin liquids (QSL) appealing of matter do break symmetry of the Hamiltonian. While QSL behaviors are well established for one-dimensional spin systems, their existence in higher dimensions remains The quantum antiferromagnet (QKHA) considered the candidate to stabilize QSL phase in two dimensions [@MisguichL05; specificity of this model relies on the unique of strong fluctuations enhanced by low spins $S=1/2$ high geometrical frustration of the lattice of corner-sharing Early numerical studies have shown, despite finite size limitations, that the ground state of the does not support any order parameter [@ZengE90PRB; @Waldtmann98EJPB] have an excitation spectrum dense sets of low energy excitations in all spin sectors [@Lecheminant97PRB; A recent state-of-the-art calculation [@Yan+10arxiv] favors a QSL ground state small much like the state [@Sachdev92PRB; @Anderson73MRB]. Other proposals similar ground-state energies encompass valence-bond-crystals

by an applied magnetic field. THe latter phAse shOws LarGeLy suPpreSsed moments $\lesSSim 0.1\mU_\mathrm{B}$ and gapped excitAtionS. THE $H-T$ pHAsE diagRam suggEStS THe eXiStEncE oF A qUantuM crItical pOint at the sMalL bUt finite magnETiC field $\mu_0 H_c=1.55(25)$ t. We Discuss this fIndIng in lIgHt oF The peRtuRbatiVe DzyaLOshinsKii-Moriya InTEractiON which wAS ThEoreTically proposed to SUsTAin a quantum criTical rEgIMe FOR thE quAntum kagomÉ HEisenBErg antiFErROMAgnET model.' author: - 'M. jeong' - 'F. Bert' - 'P. mEndEls' - 'F. DuC' - 'J. c. TrOMbe' - 'M. A. dE VrieS' - 'A. hArrIson' title: 'FiEld-INduced FreEzing oF A QuantuM spin LiqUid on tHe KAgoMé LaTTiCe' --- quaNtUM spIN lIquIDs (QsL) are appEaLiNg staTes oF MATTer wHicH do nOt breAk any symmetry Of tHe spIN HaMiltoNian. WHile qSl behaViors aRe welL eStablished for onE-dimEnsional sPin SySteMs, Their EXistenCe iN hiGher dimEnsions REmaInS QUEsTionable. The quantum KaGOMé heisenbeRg antiFErRoMAgnet (QKHa) iS coNsidEREd as tHe beST cAndidate To stabILiZe Such a QSl pHase in TwO diMenSions [@mIsguIchL05; @BaLents10Nat; @lhuilLIerM10]. The specifiCIty of this modeL ReLIEs ON the UniQue combinatIon oF StroNg quANtUm fLUctuaTions EnHAnCEd by low spins $S=1/2$ and higH gEometrIcal fRustration of tHe lattice oF CORner-sharIng tRIaNGles. Early numerIcal sTudies have SHown, despIte fiNite size LimitatioNS, That the gRouNd sTatE of THE QkHA model does nOT SuppOrT any simPle Order paRamEteR [@ZeNgE90pRb; @LecheminAnt97PRB; @WaLdTmAnN98EjPB] And haVE evidencEd An oRiGinAl excITation SpectRum wItH dENse Sets of lOW eNERgy eXcItAtioNs iN aLl spiN secTOrs [@lechemiNant97PRB; @WaLdtMAnn98EjPb]. A Recent sTate-of-the-art cAlCulation [@YaN+10aRxiV] favorS A qSL grounD state with a small gap much LIke the rEsoNatinG-valEnce-bond sTatE [@SachdEv92Prb; @AnderSon73MRB]. other ReCenT PRoposALS wIth SiMilar grounD-STatE enerGiEs enCompass Large unit-cell valenCE-boNd-crystals [@SinGhH07pRB]

by an applied magnetic fi eld. The l atter ph ase s hows lar gely suppresse d mom ents $\lesssim 0.1\mu_ \math rm { B}$a nd gapp ed exci t at i o ns. T he $H -T $ p hasedia gram su ggests the ex is tence of a q u an tum critic alpoint at the sm all bu tfin i te ma gne tic f ield $ \ mu_0 H _c=1.55(2 5) $  T. We discuss t hi s fi nding in light of th e perturbativeDzyalo sh i ns k i i-M ori ya interac ti on wh i ch wast he o r e tic a lly proposedto sustaina qu antumcr iti c al reg ime f or the quantum ka gomé Heisenbe rg ant i ferroma g net mod el.' a uth or: - ' M .Je ong '- 'F . B ert ' -'P. Mend el s' - 'F . Du c ' - 'J. C. Tro mbe'- 'M. A. de V rie s' - 'A. Harr ison' tit le : 'Fi eld-In duced F reezing of a Qu antu m Spin Li qui donth e Kag o mé Lat tic e'--- Qu antum s p inli q u i ds (QSL) are appeali ng s ta tes of m atterw hi ch do not b re akanys y mmetr y of th e spin H amilto n ia n. WhileQS L beha vi ors ar e wel l est ablish ed for o ne-di m ensional spins ystems, their ex i s te n ce i n h igher dimen sion s rem ains qu est i onabl e. Th eq ua n tum kagomé Heisenbe rg antif errom agnet (QKHA)is conside r e d as thebest ca n didate to stab ilize such a QS L phase i n two dimensi ons [@Mis g u ichL05;@Ba len ts1 0Na t ; @ LhuillierM10] . Thesp ecifici tyof this mo del re lie son the un ique com bi na ti on of stro n g quantu mflu ct uat ionse nhance d bylowsp in s $S =1/2$ a n dh i gh g eo me tric alfr ustra tion ofthe lat tice of c orn e r-sh ar in g trian gles. Early n um erical stu di eshave s h o wn, desp ite finite size limitat i ons, th atthe g roun d state o f t he QKH A m o del do es not supp or t a n y simp l e o rde rparameter[ @ Zen gE90P RB ; @L echemin ant97PRB; @Waldtma n n98 EJPB] and hav e e vide n c ed an or i gin al exc i t ation spectrumwith dense s e ts of low en e rgy e xcitati ons inall s p in sect ors [@Lec heminant9 7P RB;@ W ald tmann98EJP B]. A re cent stat e -of-t h e- art c alc ulatio n[@Y an+10 arxiv] fav ors a QSL g ro und st ate w it h a smal l gap much like the res onatin g-val enc e-bond st ate [@S achdev92P RB;@Anderson7 3MR B]. Othe r r e centprop o sa lsw ith s imil a r ground- s ta tee n er gies encomp a s s la rge u nit - cell v alen ce-bond-crystals[ @SinghH07PRB]

by_an applied_magnetic field. The latter_phase shows_largely_suppressed moments_$\lesssim_0.1\mu_\mathrm{B}$ and gapped_excitations. The $H-T$_phase diagram suggests the_existence of a_quantum_critical point at the small but finite magnetic field $\mu_0 H_c=1.55(25)$ T. We discuss this_finding_in light_of_the_perturbative Dzyaloshinskii-Moriya interaction which was_theoretically proposed to sustain a_quantum critical_regime for the quantum kagomé Heisenberg antiferromagnet model.' author: -_'M._Jeong' - 'F. Bert' -_'P. Mendels' - 'F. Duc' - 'J. C. Trombe' - 'M. A._de Vries' - 'A. Harrison' title: 'Field-Induced Freezing_of a Quantum_Spin_Liquid_on the Kagomé Lattice' --- Quantum_spin liquids (QSL) are appealing states_of matter which do not break_any symmetry of the spin Hamiltonian. While_QSL behaviors are well established for_one-dimensional spin systems, their existence_in higher_dimensions remains questionable. The quantum_kagomé Heisenberg antiferromagnet_(QKHA) is_considered as the_best candidate to stabilize such a_QSL phase in_two dimensions [@MisguichL05; @Balents10Nat; @LhuillierM10]. The specificity_of_this model relies_on_the_unique combination_of strong quantum_fluctuations_enhanced by_low_spins $S=1/2$ and high geometrical frustration_of_the lattice of corner-sharing triangles. Early numerical_studies have shown, despite_finite_size limitations, that the_ground state of the QKHA_model does not support any simple_order parameter_[@ZengE90PRB; @Lecheminant97PRB;_@Waldtmann98EJPB] and have evidenced an original excitation spectrum with dense sets_of low energy excitations in all_spin sectors [@Lecheminant97PRB; @Waldtmann98EJPB]._A recent_state-of-the-art_calculation [@Yan+10arxiv] favors_a_QSL ground_state with a small gap much like_the resonating-valence-bond_state [@Sachdev92PRB; @Anderson73MRB]. Other recent proposals_with similar ground-state energies_encompass_large unit-cell valence-bond-crystals [@SinghH07PRB]

j}{\left(}x{\right)}=\sum_{m=0}^{\lfloor j/2\rfloor}\frac{2^{j-2m}}{m!{\left(}j-2m{\right)}!}\,\partial_r^{j-m}U{\left(}r{\right)}\\ \times\sum_{\sigma\in\mathfrak{S}(j)}\delta_{i_{\sigma{\left(}1{\right)}}i_{\sigma{\left(}2{\right)}}}\dotsm\delta_{i_{\sigma{\left(}2m-1{\right)}}i_{\sigma{\left(}2m{\right)}}}x_{i_{\sigma{\left(}2m+1{\right)}}}\dotsm x_{i_{\sigma{\left(}j{\right)}}}\quad\forall x\in{\mathbb{R}}^n,\end{gathered}$$ where $r:=\left|x\right|^2$, $U{\left(}r{\right)}:=U{\left(}x{\right)}={\left(}1+r{\right)}^{-{\left(}n-2k{\right)}/2}$ and $\mathfrak{S}{\left(}j{\right)}$ is the set of all permutations of ${\left(}1,\dotsc,j{\right)}$. Furthermore, it is easy to see that $$\begin{aligned} \label{Step3Eq16} \partial_r^{j}U{\left(}r{\right)}&={\left(}-1{\right)}^j2^{-j}{\left(}n-2k{\right)}{\left(}n-2k+2{\right)}\dotsm{\left(}n-2k+2j-2{\right)}{\left(}1+r{\right)}^{-\frac{n-2k+2j}{2}}\nonumber\allowdisplaybreaks\\ &=\frac{2{\left(}-1{\right)}^jj!}{{\left(}n-2k-2{\right)}}\operatorname{B}{\left(}\frac{n}{2}-k-1,j+1{\right)}^{-1}{\left(}1+r{\right)}^{-\frac{n-2k+2j}{2}}.\end{aligned}$$ Another induction yields $$\label{Step3Eq17} \Delta_0^l\,U{\left(}x{\right)}=\left\{\begin{

j}{\left(}x{\right)}=\sum_{m=0}^{\lfloor j/2\rfloor}\frac{2^{j-2m}}{m!{\left(}j-2m{\right)}!}\,\partial_r^{j - m}U{\left(}r{\right)}\\ \times\sum_{\sigma\in\mathfrak{S}(j)}\delta_{i_{\sigma{\left(}1{\right)}}i_{\sigma{\left(}2{\right)}}}\dotsm\delta_{i_{\sigma{\left(}2m-1{\right)}}i_{\sigma{\left(}2m{\right)}}}x_{i_{\sigma{\left(}2m+1{\right)}}}\dotsm x_{i_{\sigma{\left(}j{\right)}}}\quad\forall x\in{\mathbb{R}}^n,\end{gathered}$$ where $ r:=\left|x\right|^2 $, $ U{\left(}r{\right)}:=U{\left(}x{\right)}={\left(}1+r{\right)}^{-{\left(}n-2k{\right)}/2}$ and $ \mathfrak{S}{\left(}j{\right)}$ is the set of all permutations of $ { \left(}1,\dotsc, j{\right)}$. Furthermore, it is easy to see that $ $ \begin{aligned } \label{Step3Eq16 } \partial_r^{j}U{\left(}r{\right)}&={\left(}-1{\right)}^j2^{-j}{\left(}n-2k{\right)}{\left(}n-2k+2{\right)}\dotsm{\left(}n-2k+2j-2{\right)}{\left(}1+r{\right)}^{-\frac{n-2k+2j}{2}}\nonumber\allowdisplaybreaks\\ & = \frac{2{\left(}-1{\right)}^jj!}{{\left(}n-2k-2{\right)}}\operatorname{B}{\left(}\frac{n}{2}-k-1,j+1{\right)}^{-1}{\left(}1+r{\right)}^{-\frac{n-2k+2j}{2}}.\end{aligned}$$ Another evocation give $ $ \label{Step3Eq17 } \Delta_0^l\,U{\left(}x{\right)}=\left\{\begin {

j}{\levt(}x{\right)}=\sum_{m=0}^{\lfloor j/2\rflour}\frac{2^{j-2m}}{m!{\left(}j-2m{\titht)}!}\,\parvial_r^{j-m}H{\left(}r{\rieht)}\\ \times\sum_{\sigma\in\mathfrak{S}(j)}\veltq_{i_{\signa{\left(}1{\right)}}i_{\sigma{\left(}2{\rkght)}}}\dotsm\felta_{i_{\sitma{\ltft(}2m-1{\right)}}i_{\sigma{\lehf(}2m{\right)}}}w_{n_{\sigmz{\peft(}2k+1{\cight)}}}\dotsm x_{i_{\sinma{\left(}j{\rigvt)}}}\quad\forall x\hn{\oachbb{R}}^n,\end{gathered}$$ where $r:=\left|x\right|^2$, $T{\left(}r{\rogjt)}:=U{\left(}x{\right)}={\lgft(}1+r{\rpgrt)}^{-{\lern(}n-2j{\right)}/2}$ and $\mathfrak{S}{\left(}j{\righf)}$ is tht set of all permuyations of ${\left(}1,\dotsc,j{\right)}$. Furhhermore, it is easj to see thqt $$\bqtin{aligned} \lacel{Step3Eq16} \partial_r^{j}U{\lefj(}r{\right)}&={\left(}-1{\right)}^j2^{-j}{\left(}n-2k{\right)}{\lefg(}n-2k+2{\riyht)}\dotsm{\lefj(}u-2j+2j-2{\rlcht)}{\left(}1+r{\rigit)}^{-\frac{g-2k+2j}{2}}\nonumber\allowdispnaybreals\\ &=\frac{2{\left(}-1{\righb)}^jj!}{{\leht(}n-2k-2{\eight)}}\operatorname{B}{\lefv(}\frac{n}{2}-k-1,j+1{\right)}^{-1}{\left(}1+r{\ridht)}^{-\frac{n-2k+2b}{2}}.\eud{aligned}$$ Another indycrion iieldv $$\lacwl{Sgep3Tq17} \Dxltz_0^l\,U{\lefh(}x{\rmght)}=\left\{\begjn{

j}{\left(}x{\right)}=\sum_{m=0}^{\lfloor j/2\rfloor}\frac{2^{j-2m}}{m!{\left(}j-2m{\right)}!}\,\partial_r^{j-m}U{\left(}r{\right)}\\ \times\sum_{\sigma\in\mathfrak{S}(j)}\delta_{i_{\sigma{\left(}1{\right)}}i_{\sigma{\left(}2{\right)}}}\dotsm\delta_{i_{\sigma{\left(}2m-1{\right)}}i_{\sigma{\left(}2m{\right)}}}x_{i_{\sigma{\left(}2m+1{\right)}}}\dotsm x_{i_{\sigma{\left(}j{\right)}}}\quad\forall x\in{\mathbb{R}}^n,\end{gathered}$$ where $r:=\left|x\right|^2$, $\mathfrak{S}{\left(}j{\right)}$ the set all permutations of to that $$\begin{aligned} \label{Step3Eq16} &=\frac{2{\left(}-1{\right)}^jj!}{{\left(}n-2k-2{\right)}}\operatorname{B}{\left(}\frac{n}{2}-k-1,j+1{\right)}^{-1}{\left(}1+r{\right)}^{-\frac{n-2k+2j}{2}}.\end{aligned}$$ Another induction $$\label{Step3Eq17} \Delta_0^l\,U{\left(}x{\right)}=\left\{\begin{

j}{\left(}x{\right)}=\sum_{m=0}^{\lfloor j/2\rflOor}\frac{2^{j-2m}}{m!{\Left(}j-2M{\riGht)}!}\,\PaRtiaL_r^{j-m}u{\left(}r{\right)}\\ \timES\sum_{\Sigma\in\mathfrak{S}(j)}\delta_{I_{\sigmA{\lEFt(}1{\riGHt)}}I_{\sigmA{\left(}2{\riGHt)}}}\DOTsm\DeLtA_{i_{\sIgMA{\lEft(}2m-1{\rIghT)}}i_{\sigma{\Left(}2m{\right)}}}X_{i_{\sIgMa{\left(}2m+1{\right)}}}\DOtSm x_{i_{\sigma{\lEft(}J{\right)}}}\quad\foRalL x\in{\maThBb{R}}^N,\End{gaTheRed}$$ whEre $r:=\leFT|x\righT|^2$, $U{\left(}r{\riGhT)}:=u{\left(}x{\RIght)}={\lefT(}1+R{\RiGht)}^{-{\lEft(}n-2k{\right)}/2}$ and $\mathFRaK{s}{\left(}j{\right)}$ is tHe set oF aLL pERMutAtiOns of ${\left(}1,\dOtSc,j{\riGHt)}$. FurthERmORE, It iS Easy to see that $$\Begin{aligneD} \LabEl{Step3eq16} \ParTIal_r^{j}U{\Left(}r{\RiGHt)}&={\lEft(}-1{\right)}^j2^{-j}{\lEft(}n-2K{\right)}{\lefT(}n-2k+2{\rigHT)}\dotsm{\lEFt(}n-2k+2j-2{\riGht)}{\lefT(}1+r{\rIghT)}^{-\fraC{N-2k+2J}{2}}\nOnuMbER\alLOwDisPLayBreaks\\ &=\frAc{2{\LeFt(}-1{\rigHt)}^jj!}{{\LEFT(}N-2k-2{\riGht)}}\OperAtornAme{B}{\left(}\frac{n}{2}-K-1,j+1{\rIght)}^{-1}{\LEft(}1+R{\righT)}^{-\frac{N-2k+2j}{2}}.\eNd{AlignEd}$$ AnotHer inDuCtion yields $$\labeL{SteP3Eq17} \Delta_0^l\,u{\leFt(}X{\riGhT)}=\left\{\BEgin{

j}{\left(}x{\right)}=\sum_ {m=0}^{\lf loorj/2 \rf lo or}\ frac {2^{j-2m}}{m!{ \ left (}j-2m{\right)}!}\,\pa rtial _r ^ {j-m } U{ \left (}r{\ri g ht ) } \\\t im es\ su m _{ \sigm a\i n\mathf rak{S}(j)} \de lt a_{i_{\sigma { \l eft(}1{\ri ght )}}i_{\sigma {\l eft(}2 {\ rig h t)}}} \do tsm\d elta_{ i _{\sig ma{\left( }2 m -1{\ri g ht)}}i_ { \ si gma{ \left(}2m{\right) } }} x _{i_{\sigma{\l eft(}2 m+ 1 {\ r i ght )}} }\dotsm x_ {i _{\si g ma{\lef t (} j { \ rig h t)}}}\quad\fo rall x\in{\ m ath bb{R}} ^n ,\e n d{gath ered} $$ whe re $r:=\lef t|x\ right|^2$ , $U{\ l eft(}r{ \ right)} :=U{\l eft (}x {\ri g ht )} ={\ le f t(} 1 +r {\r i ght )}^{-{\l ef t( }n-2k {\ri g h t ) }/2} $ a nd $ \math frak{S}{\left (}j {\ri g ht) }$ is thesetof allpermut ation sof ${\left(}1,\ dots c,j{\righ t)} $. Fu rt hermo r e, itiseas y to se e that$ $\b eg i n { al igned} \label{Step 3E q 1 6} \partia l_r^{j } U{ \l e ft(}r{\r ig ht) }&={ \ l eft(} -1{\ r ig ht)}^j2^ {-j}{\ l ef t( }n-2k{\ ri ght)}{ \l eft (}n -2k+2 { \rig ht)}\d otsm{\le ft(}n - 2k+2j-2{\right ) }{\left(}1+r{ \ ri g h t) } ^{-\ fra c{n-2k+2j}{ 2}}\ n onum ber\ a ll owd i splay break s\ \ & = \frac{2{\left(}-1{\ ri ght)}^ jj!}{ {\left(}n-2k- 2{\right)} } \ o peratorn ame{ B }{ \ left(}\frac{n} {2}-k -1,j+1{\ri g ht)}^{-1 }{\le ft(}1+r{ \right)}^ { - \frac{n- 2k+ 2j} {2} }.\ e n d{ aligned}$$ An o t herin duction yi elds $$ \la bel {St ep3 Eq 17} \Delt a_0^l\,U {\ le ft (} x{\ right ) }=\left\ {\ beg in {

j}{\left(}x{\right)}=\sum_{m=0}^{\lfloor j/2\rfloor}\frac{2^{j-2m}}{m!{\left(}j-2m{\right)}!}\,\partial_r^{j-m}U{\left(}r{\right)}\\ \times\sum_{\sigma\in\mathfrak{S}(j)}\delta_{i_{\sigma{\left(}1{\right)}}i_{\sigma{\left(}2{\right)}}}\dotsm\delta_{i_{\sigma{\left(}2m-1{\right)}}i_{\sigma{\left(}2m{\right)}}}x_{i_{\sigma{\left(}2m+1{\right)}}}\dotsm_x_{i_{\sigma{\left(}j{\right)}}}\quad\forall x\in{\mathbb{R}}^n,\end{gathered}$$_where $r:=\left|x\right|^2$, $U{\left(}r{\right)}:=U{\left(}x{\right)}={\left(}1+r{\right)}^{-{\left(}n-2k{\right)}/2}$ and_$\mathfrak{S}{\left(}j{\right)}$ is_the_set of_all_permutations of ${\left(}1,\dotsc,j{\right)}$._Furthermore, it is_easy to see that_$$\begin{aligned} \label{Step3Eq16} \partial_r^{j}U{\left(}r{\right)}&={\left(}-1{\right)}^j2^{-j}{\left(}n-2k{\right)}{\left(}n-2k+2{\right)}\dotsm{\left(}n-2k+2j-2{\right)}{\left(}1+r{\right)}^{-\frac{n-2k+2j}{2}}\nonumber\allowdisplaybreaks\\ &=\frac{2{\left(}-1{\right)}^jj!}{{\left(}n-2k-2{\right)}}\operatorname{B}{\left(}\frac{n}{2}-k-1,j+1{\right)}^{-1}{\left(}1+r{\right)}^{-\frac{n-2k+2j}{2}}.\end{aligned}$$ Another induction_yields_$$\label{Step3Eq17} \Delta_0^l\,U{\left(}x{\right)}=\left\{\begin{

violation is sufficiently significant if $nS\geq c$ for some constant $c$, simulations show that the best choice of $c$ depends on the distribution of settings and outcomes in the experiment. The third and simplest approach is to block the data from the trials. Instead of updating the log-$p$-value after every trial, we process data $h$ trials at a time. The first block is used only for estimating the setting and outcome distribution of future trials. That is, we set $R_k(x)=1$ for $k=0,\ldots,(h-1)$. Subsequently, we have $R_{mh+k}=R_{mh}$ for $k=1,\ldots,(h-1)$ and all $m$. Note that neither the validity nor the asymptotic optimality of the calculated $p$-values requires updating the PBRs after each trial. Choosing $h$ large enough ensures that the first block’s trials have sufficient information for obtaining reasonable estimates of the distribution. An additional advantage of blocking the trials is that we avoid unnecessarily invoking the computationally costly optimizations required for updating the PBRs. We standardized the choice of block size so that if the total number of trials to be analyzed is $N$, $h$ is the maximum of $\lceil N/1000\rceil$ and $\lceil\ln(2d)d\rceil$, where $d$ is the number of possible setting and outcome combinations in a trial. The first expression ensures that we do not lose too much log-$p$-value by using the first block only for learning the setting and outcome distribution. The second one is chosen so that if $q$ is uniform, the probability that every setting and outcome combination occurs is at least $1/2$. We conclude this section with a note on implementing the PBR protocol. For monitoring an experiment and to adapt to changes in experimental configuration, the estimated setting and outcome distributions used in the PBRs should be based on recent trials only. This can be accomplished by windowing the trials with a window large enough to have statistically significant violation of LR (if there is violation), but small enough to avoid seeing significant changes in configuration. Our implementation of the local realism analysis engine uses a computationally simpler approach based on weighting the trials with exponentially decreasing weights in time determined by a configurable half-life. This feature was not used in the comparisons in Sec. \[sect:results\]. Effects of Suboptimal Estimates and LR Models {#sect

violation is sufficiently significant if $ nS\geq c$ for some constant $ c$, simulations picture that the dear choice of $ c$ depends on the distribution of settings and consequence in the experiment. The third and simplest approach is to obstruct the data from the trials. alternatively of updating the log-$p$-value after every trial, we serve data $ h$ trials at a fourth dimension. The inaugural block is used only for estimating the setting and outcome distribution of future test. That is, we set $ R_k(x)=1 $ for $ k=0,\ldots,(h-1)$. Subsequently, we have $ R_{mh+k}=R_{mh}$ for $ k=1,\ldots,(h-1)$ and all $ m$. Note that neither the robustness nor the asymptotic optimality of the calculated $ p$-values requires updating the PBRs after each trial. choose $ h$ large enough ensures that the first block ’s trial have sufficient information for obtaining reasonable estimates of the distribution. An extra advantage of stuff the trials is that we avoid unnecessarily invoking the computationally costly optimizations required for updating the PBRs. We standardized the choice of pulley size so that if the total number of test to be analyze is $ N$, $ h$ is the maximum of $ \lceil N/1000\rceil$ and $ \lceil\ln(2d)d\rceil$, where $ d$ is the number of potential setting and result combinations in a trial. The inaugural expression ensures that we do not lose too much log-$p$-value by using the first blockage only for learning the setting and outcome distribution. The second one is chosen so that if $ q$ is uniform, the probability that every mise en scene and outcome combination occurs is at least $ 1/2$. We conclude this section with a note on follow through the PBR protocol. For monitoring an experiment and to adapt to changes in experimental shape, the estimated context and outcome distributions use in the PBRs should be based on late trials only. This can be accomplished by windowing the trials with a window bombastic enough to have statistically meaning violation of LR (if there is irreverence), but small enough to avoid see significant change in configuration. Our implementation of the local realism analysis engine uses a computationally simpler approach establish on weighting the trials with exponentially decrease weights in time specify by a configurable half - life. This feature was not used in the comparison in Sec.   \[sect: results\ ]. Effects of Suboptimal Estimates and LR Models { # sect

villation is sufficiently rignificant if $uW\geq c$ for skme consgant $c$, simulations show that tye bewt choice of $c$ depends on the dpstributiin oh settings and onfcomes lu the cxpernmxnt. The third anc simplest approach is tm cllck the data from the trials. Insteaq of upcahing the log-$p$-vwlue wfted every trial, we process data $h$ trjals at a time. The forst block is used only fog eshimating the settijg and outcime qustribution uf future urnals. That ia, we set $R_k(x)=1$ for $k=0,\ldots,(h-1)$. Subseqjentlv, we have $R_{nh+j}=R_{mj}$ for $k=1,\ldots,(i-1)$ and wll $m$. Note tmst neidher thr validity nor thx astmptotic optimality oh the calculated $p$-vajues requhrzs updating the PBRs qfrer edch driau. Chuosjnj $h$ large enkugh ensurss that the first block’s trialx rqve sufficienf infowmwtion for obtaining reasonable estimatev or the distribution. An aeditional advantage ov blockind the trials is that we avoid unnecessarily invokhng tie conpmtatkinwlly costly optimizations required for updatigf uhe PBRs. We standcrdized the choive og block size su that if the total number lf triajs to be analyeed ix $N$, $h$ is the maximum of $\lceul N/1000\rceil$ anb $\lxeil\ln(2d)d\rceil$, wherz $d$ is the nbmber pf poxsible setting and outcume dombinationd in a trjxl. The first expfesximn ensurtr that we do not jose too nuch log-$p$-vauue ny usigg the firdt block only for learnijg thg setthng and ouhcome distribution. The second oix is chosen sp dhan if $q$ is unifprm, the probafility that evgry settiug and outcome ckmbinatmon occurs if at least $1/2$. We fonclude thiv sectiog wirh a note ov implementing yhe PBR pgouocol. For nonitoring an expevimenj znd to adapt to chqnges in experikengal clnhigurwdion, the esthmatdd rrttine and outcond dixtributions used in dhe LBRs should be basrd on recebt trialf only. This csn be accomplished by wmndowiig the tryals with a window large enougg to have stwtistically fignlficwnt violatnon of LR (if there is violation), but smalo enough to avoid weeing significant ehsnges in coifigurwtion. Our implementation of tye local realism snalysis engine uses z compgtatilnally simpler approach based on weighting the trials with exponentially eecreasing weights in yime gecerkined fy a ronfigurable half-kife. This feature was not used mn the com[axisons in Sec. \[sect:results\]. Effecys of Suboptimau Estimates and LR Modela {#sect

violation is sufficiently significant if $nS\geq c$ constant simulations show the best choice distribution settings and outcomes the experiment. The and simplest approach is to block data from the trials. Instead of updating the log-$p$-value after every trial, we data $h$ trials at a time. The first block is used only for the and distribution future trials. That is, we set $R_k(x)=1$ for $k=0,\ldots,(h-1)$. Subsequently, we have $R_{mh+k}=R_{mh}$ for $k=1,\ldots,(h-1)$ and $m$. Note that neither the validity nor the optimality of the calculated requires updating the PBRs after trial. $h$ large ensures the block’s trials have information for obtaining reasonable estimates of the distribution. An additional advantage of blocking the trials is that avoid unnecessarily computationally costly required updating PBRs. We standardized of block size so that if of trials to be analyzed is $N$, $h$ the maximum $\lceil N/1000\rceil$ and $\lceil\ln(2d)d\rceil$, where $d$ the number of possible setting and outcome combinations a trial. The first expression ensures that we do not lose too much log-$p$-value by first block only for the setting and distribution. second is so that $q$ is uniform, the probability that every setting and outcome combination is at least $1/2$. We conclude this section with a implementing PBR protocol. For an experiment and to to in experimental configuration, the and distributions PBRs be on recent trials only. can be accomplished by windowing trials with a window significant violation of LR (if there is violation), small enough to avoid seeing significant changes configuration. Our implementation of the local realism analysis engine uses a computationally approach based the trials with exponentially decreasing weights in time by a configurable half-life. feature was not used in the comparisons in Sec. Effects Suboptimal Estimates LR Models {#sect

violation is sufficiently siGnificant iF $nS\geQ c$ fOr sOmE conStanT $c$, simulations sHOw thAt the best choice of $c$ depeNds on ThE DistRIbUtion Of settiNGs AND ouTcOmEs iN tHE eXperiMenT. The thiRd and simplEst ApProach is to blOCk The data froM thE trials. InsteAd oF updatInG thE Log-$p$-vAluE afteR every TRial, we Process daTa $H$ Trials AT a time. THE FiRst bLock is used only for EStIMating the settiNg and oUtCOmE DIstRibUtion of futUrE triaLS. That is, WE sET $r_K(x)=1$ fOR $k=0,\ldots,(h-1)$. SubseQuently, we haVE $R_{mH+k}=R_{mh}$ fOr $K=1,\ldOTs,(h-1)$ and All $m$. NOtE ThaT neither the ValiDity nor thE asympTOtic optIMality oF the caLcuLatEd $p$-vALuEs ReqUiREs uPDaTinG The pBRs afteR eAcH triaL. ChoOSING $h$ laRge EnouGh ensUres that the fiRst BlocK’S trIals hAve suFficIeNt infOrmatiOn for ObTaining reasonabLe esTimates of The DiStrIbUtion. aN additIonAl aDvantagE of blocKIng ThE TRIaLs is that we avoid unnEcESSaRily invoKing thE CoMpUTationalLy CosTly oPTImizaTionS ReQuired foR updatINg ThE PBRs. We StAndardIzEd tHe cHoice OF bloCk size So that if The toTAl number of triaLS to be analyzed IS $N$, $H$ IS tHE maxImuM of $\lceil N/1000\rcEil$ aND $\lceIl\ln(2D)D\rCeiL$, Where $D$ is thE nUMbER of possible setting aNd OutcomE combInations in a trIal. The firsT EXPression EnsuREs THat we do not lose Too muCh log-$p$-valuE By using tHe firSt block oNly for leaRNIng the seTtiNg aNd oUtcOME dIstribution. ThE SEconD oNe is choSen So that iF $q$ iS unIfoRm, tHe ProbabiliTy that evErY sEtTiNg aNd outCOme combiNaTioN oCcuRs is aT Least $1/2$. WE concLude ThIs SEctIon with A NoTE On imPlEmEntiNg tHe pBR prOtocOL. FoR monitoRing an expEriMEnt aNd To Adapt to Changes in expeRiMental confIgUraTion, thE EStimated Setting and outcome distriBUtions uSed In the pBRs Should be bAseD on recEnt TRials oNly. ThiS can bE aCcoMPLisheD BY wIndOwIng the triaLS WitH a winDoW larGe enougH to have statisticalLY siGnificant violAtiOn of lr (If TheRE iS VioLaTIon), BUT small enough to aVoid seeing SiGNiFicant chanGEs iN cOnfigurAtion. OuR implEMentatiOn of the loCal realisM aNalySIS enGine uses a cOmputatiOnally simPLer apPRoAch baSed On weigHtIng The trIals wiTH exPonenTially DeCreasiNg weiGhTs in time Determined by a configurabLe half-Life. THis Feature waS noT UseD in the comPariSons in Sec. \[sEct:ResUlts\]. EFfeCTs of SUbopTImAl ESTimatEs anD lR Models {#sECt

violation is sufficiently significa nt if $n S\g eq c$forsome constant$ c$,simulations show thatthe b es t cho i ce of $ c$ depe n ds o n t he d ist ri b ut ion o f s ettings and outco mes i n the experi m en t. The th ird and simples t a pproac hist o blo ckthe d ata fr o m thetrials. I ns t ead of updatin g th e lo g-$p$-value after ev e ry trial, we p rocess d a ta $ h$tri als at a t im e. Th e firstb lo c k isu sed only forestimatingt hesettin gand outcom e dis tr i but ion of futu re t rials. Th at is, we set$ R_k(x)= 1$ for $k =0, \ldo t s, (h -1) $. Sub s eq uen t ly, we have $ R_ {mh+k }=R_ { m h } $ fo r $ k=1, \ldot s,(h-1)$ andall $m$ . No te th at ne ithe rthe v alidit y nor t he asymptotic o ptim ality ofthe c alc ul ated$ p$-val ues re quiresupdatin g th eP B R safter each trial.Ch o o si ng $h$ l arge e n ou gh ensuresth atthef i rst b lock ’ strials h ave su f fi ci ent inf or mation f orobt ainin g rea sonabl e estima tes o f the distribut i on. An additi o na l ad v anta geof blocking the tria ls i s t hat we av oid u nn e ce s sarily invoking the c omputa tiona lly costly op timization s r equiredforu pd a ting the PBRs. We s tandardize d the cho ice o f blocksize so t h a t if the to tal nu mbe r of trials to be a naly ze d is $N $,$h$ isthe ma xim umof $\lceilN/1000\r ce il $an d $ \lcei l \ln(2d)d \r cei l$ , w here$ d$ isthe n umbe rof pos sible s e tt i n g an dou tcom e c om binat ions ina trial . The fir ste xpre ss io n ensur es that we do n ot lose to omuc h log- $ p $-valueby using the first bloc k only f orlearn ingthe setti ngand ou tco m e dist ributi on. T he se c o nd on e is ch os en so that i f $ q$ is u nifo rm, the probability thate ver y setting and ou tcom e co mbi n at i onoc c urs i s at least $1/2 $. We con cl u de this sect i onwi th a no te on i mplem e nting t he PBR pr otocol. F or mon i t ori ng an expe riment a nd to ada p t toc ha ngesinexperi me nta l con figura t ion , the estim at ed set tingan d outcom e distributions used in the P BRs s hou ld be bas edo n r ecent tri alsonly. This ca n b e acc omp l ished byw in dow i ng th e tr i als witha w ind o w l arge enough t o ha ve st ati s ticall y si gnificant violati o n of LR (if th erei s vi ola t ion) ,but small enou ghto a void see in g significa nt chang es in co nfigur ation. Our im p l em e ntatio n of th e local r eal is m analys is e n gine u sesacomput ationa l ly s i m pler approach ba sed o n weigh t ing thetr ials wi t h ex ponentiall y decreasin g weig htsin ti me dete rm ined b y a c onfigurabl e half-lif e. Th is feat ur e wa s n ot use d in t he co mpar is ons in Sec.\ [ se c t: re s ult s\]. Eff ec ts o f Subopti m al Estim ate s and LR M ode l s {#sec t

violation_is sufficiently_significant if $nS\geq c$_for some_constant_$c$, simulations_show_that the best_choice of $c$_depends on the distribution_of settings and_outcomes_in the experiment. The third and simplest approach is to block the data from the_trials._Instead of_updating_the_log-$p$-value after every trial, we_process data $h$ trials at_a time._The first block is used only for estimating_the_setting and outcome_distribution of future trials. That is, we set $R_k(x)=1$_for $k=0,\ldots,(h-1)$. Subsequently, we have $R_{mh+k}=R_{mh}$_for $k=1,\ldots,(h-1)$ and_all_$m$._Note that neither the_validity nor the asymptotic optimality of_the calculated $p$-values requires updating the_PBRs after each trial. Choosing $h$ large_enough ensures that the first block’s_trials have sufficient information for_obtaining reasonable_estimates of the distribution. An_additional advantage of_blocking the_trials is that_we avoid unnecessarily invoking the computationally_costly optimizations required_for updating the PBRs. We standardized_the_choice of block_size_so_that if_the total number_of_trials to_be_analyzed is $N$, $h$ is the_maximum_of $\lceil N/1000\rceil$ and $\lceil\ln(2d)d\rceil$, where $d$_is the number of_possible_setting and outcome combinations_in a trial. The first_expression ensures that we do not_lose too_much log-$p$-value_by using the first block only for learning the setting and_outcome distribution. The second one is_chosen so that if_$q$ is_uniform,_the probability that_every_setting and_outcome combination occurs is at least $1/2$. We_conclude this_section with a note on implementing_the PBR protocol. For_monitoring_an experiment and to adapt to_changes in experimental configuration, the estimated_setting and outcome distributions used_in_the_PBRs should be based on_recent trials only. This can be_accomplished by windowing_the trials with a window large enough_to_have statistically significant violation of LR_(if_there is violation), but small enough_to_avoid_seeing significant changes in configuration._Our implementation of the local realism_analysis engine uses a computationally simpler approach based on_weighting the trials_with exponentially decreasing weights in_time_determined_by a configurable half-life. This feature was not used in_the comparisons_in Sec. \[sect:results\]. Effects of_Suboptimal Estimates and LR Models {#sect