text
stringlengths 649
4.42k
| synonym_substitution
stringlengths 759
4.5k
| butter_fingers
stringlengths 649
4.42k
| random_deletion
stringlengths 453
2.31k
| change_char_case
stringlengths 649
4.42k
| whitespace_perturbation
stringlengths 764
5.02k
| underscore_trick
stringlengths 649
4.42k
|
|---|---|---|---|---|---|---|
)&\to M(k) A(x,k^{-1}[y])\\
C(x,y)&\to N(k)C(x,k^{-1}[y])\endaligned\right.,$$ are physical. Here $M(k)$ and $N(k)$ are matrix representations of $K$ and $k[y]$ is the image of the point $y\in\mathbb{S}^7$ under the operation of $k\in K$. But unlike the standard orbifolding conditions, we suppose that $M(K)\ne N(K)$.
In order to determine automorphisms that are responsible for the symmetry breaking (\[2-04\]), we must first select the group $K$. Suppose $K=\mathbb{Z}_4\times\mathbb{Z}_2$. Further, suppose that the subgroup $H$ in $Spin(8)$ is generated by the operators $I=R_7$ and $J=L_7^{-1}L_5L_2$, which are defined in Appendix B. Obviously, $IJ=JI$ and $I^4=J^2=1$. It follows from this that $H=H_{I}\times H_{J}$, where the subgroups $H_{I}$ and $H_{J}$ are generated by $I$ and $J$, respectively, i.e., $H\simeq\mathbb{Z}_4\times\mathbb{Z}_2$. Define the action of $K$ on $\mathbb{S}^7$ and the representations $K\to M(K)$ and $K\to N(K)$ as follows. Let $f$, $f_{I}$, and $f_{J}$ be homomorphisms of $K$ onto $H$, $H_{I}$, and $H_{J}$ respectively. Then $$\begin{aligned}
M(k)A(x,y)&=f(k)^{-1}A(x,y)f(k),\\
N(k)C(x,y)&=f_{I}(k)^{-1}C(x,y)f_{I}(k),\\
k^{-1}[y]&=f_{J}(k)^{-1}y.\end{aligned}$$
We now focus our attention on the symmetry breaking at the fixed points. Since the factor group $K/K_{J}$ acts on $\mathbb{S}^7$ trivially | ) & \to M(k) A(x, k^{-1}[y])\\
C(x, y)&\to N(k)C(x, k^{-1}[y])\endaligned\right. ,$$ are physical. Here $ M(k)$ and $ N(k)$ are matrix representations of $ K$ and $ k[y]$ is the image of the point $ y\in\mathbb{S}^7 $ under the process of $ k\in K$. But unlike the standard orbifolding condition, we suppose that $ M(K)\ne N(K)$.
In decree to decide automorphisms that are responsible for the symmetry dampen (\[2 - 04\ ]), we must first select the group $ K$. think $ K=\mathbb{Z}_4\times\mathbb{Z}_2$. Further, suppose that the subgroup $ H$ in $ Spin(8)$ is beget by the operators $ I = R_7 $ and $ J = L_7^{-1}L_5L_2 $, which are define in Appendix B. Obviously, $ IJ = JI$ and $ I^4 = J^2=1$. It follows from this that $ H = H_{I}\times H_{J}$, where the subgroups $ H_{I}$ and $ H_{J}$ are generated by $ I$ and $ J$, respectively, i.e., $ H\simeq\mathbb{Z}_4\times\mathbb{Z}_2$. Define the action of $ K$ on $ \mathbb{S}^7 $ and the representation $ K\to M(K)$ and $ K\to N(K)$ as follows. Let $ f$, $ f_{I}$, and $ f_{J}$ be homomorphisms of $ K$ onto $ H$, $ H_{I}$, and $ H_{J}$ respectively. Then $ $ \begin{aligned }
M(k)A(x, y)&=f(k)^{-1}A(x, y)f(k),\\
N(k)C(x, y)&=f_{I}(k)^{-1}C(x, y)f_{I}(k),\\
k^{-1}[y]&=f_{J}(k)^{-1}y.\end{aligned}$$
We immediately focus our attention on the symmetry breaking at the repair points. Since the factor group $ K / K_{J}$ acts on $ \mathbb{S}^7 $ trivially | )&\to L(k) A(x,k^{-1}[y])\\
C(x,y)&\to N(k)C(x,k^{-1}[y])\endallgned\right.,$$ are pktsical. Here $J(k)$ and $N(y)$ are matrix representations od $K$ abd $k[y]$ is the image of ghe point $y\in\mathvb{S}^7$ nnder the operatmkn of $k\lu K$. Bhb unlnkx the standard prbifoldinc conditions, wa ru'pose that $M(K)\ne N(K)$.
In order to determyne autpmlrphisms that wre gefponapboe for the symmetry breaking (\[2-04\]), we musu first select the group $K$. Suppose $K=\mathbb{Z}_4\tlmes\lathbb{Z}_2$. Further, suopose that jge fybgroup $H$ in $Spin(8)$ is gtnzrated by tge operators $I=R_7$ and $J=L_7^{-1}L_5L_2$, which xre dzfined in Ak'wndlf B. Obviousoy, $IJ=TI$ and $I^4=J^2=1$. It follows from tnis that $H=H_{I}\tikes H_{J}$, where the subgroups $I_{I}$ and $H_{J}$ are generajed by $I$ atd $J$, respectively, i.w., $Y\simex\matvbb{Z}_4\gumer\mauhbu{Z}_2$. Sefine thx action of $K$ on $\mathbv{S}^7$ and the represenuatyins $K\to M(K)$ ans $K\to G(K)$ as follows. Let $f$, $f_{I}$, and $f_{J}$ be homomorpvisjs of $K$ onto $H$, $H_{I}$, and $H_{H}$ respectively. Then $$\bggin{aligneq}
M(k)A(x,y)&=f(k)^{-1}A(x,y)f(k),\\
N(k)C(x,y)&=f_{I}(k)^{-1}C(x,y)f_{I}(k),\\
k^{-1}[y]&=f_{J}(k)^{-1}y.\end{aligned}$$
We now focua oux attevriln on the symmetry breaking at the fixed poinea. Xikce the factor gvoup $K/K_{J}$ acts on $\mstjbn{F}^7$ trivially | )&\to M(k) A(x,k^{-1}[y])\\ C(x,y)&\to N(k)C(x,k^{-1}[y])\endaligned\right.,$$ are physical. and are matrix of $K$ and the $y\in\mathbb{S}^7$ under the of $k\in K$. unlike the standard orbifolding conditions, we that $M(K)\ne N(K)$. In order to determine automorphisms that are responsible for the breaking (\[2-04\]), we must first select the group $K$. Suppose $K=\mathbb{Z}_4\times\mathbb{Z}_2$. Further, suppose the $H$ $Spin(8)$ generated by the operators $I=R_7$ and $J=L_7^{-1}L_5L_2$, which are defined in Appendix B. Obviously, $IJ=JI$ and It follows from this that $H=H_{I}\times H_{J}$, where subgroups $H_{I}$ and $H_{J}$ generated by $I$ and $J$, i.e., Define the of on and the representations M(K)$ and $K\to N(K)$ as follows. Let $f$, $f_{I}$, and $f_{J}$ be homomorphisms of $K$ onto $H$, and $H_{J}$ $$\begin{aligned} M(k)A(x,y)&=f(k)^{-1}A(x,y)f(k),\\ k^{-1}[y]&=f_{J}(k)^{-1}y.\end{aligned}$$ now our attention on breaking at the fixed points. Since $K/K_{J}$ acts on $\mathbb{S}^7$ trivially | )&\to M(k) A(x,k^{-1}[y])\\
C(x,y)&\to N(k)C(x,k^{-1}[y])\endaliGned\right.,$$ aRe phySicAl. HErE $M(k)$ aNd $N(k)$ Are matrix repreSEntaTions of $K$ and $k[y]$ is the imagE of thE pOInt $y\IN\mAthbb{s}^7$ under tHE oPERatIoN oF $k\iN K$. bUt UnlikE thE standaRd orbifoldIng CoNditions, we suPPoSe that $M(K)\ne n(K)$.
IN order to deteRmiNe autoMoRphISms thAt aRe resPonsibLE for thE symmetry BrEAking (\[2-04\]), wE Must firST SeLect The group $K$. Suppose $K=\MAtHBb{Z}_4\times\mathbb{z}_2$. FurthEr, SUpPOSe tHat The subgrouP $H$ In $SpiN(8)$ Is generATeD BY The OPerators $I=R_7$ and $j=L_7^{-1}L_5L_2$, which arE DefIned in apPenDIx B. ObvIouslY, $Ij=jI$ aNd $I^4=J^2=1$. It folloWs frOm this thaT $H=H_{I}\tiMEs H_{J}$, wheRE the subGroups $h_{I}$ aNd $H_{j}$ are GEnErAteD bY $i$ anD $j$, rEspECtiVely, i.e., $H\sImEq\MathbB{Z}_4\tiMES\MAthbB{Z}_2$. DEfinE the aCtion of $K$ on $\matHbb{s}^7$ and THe rEpresEntatIons $k\tO M(K)$ anD $K\to N(K)$ As folLoWs. Let $f$, $f_{I}$, and $f_{J}$ be HomoMorphisms Of $K$ OnTo $H$, $h_{I}$, And $H_{J}$ REspectIveLy. THen $$\begiN{aligneD}
m(k)A(X,y)&=F(K)^{-1}a(X,y)F(k),\\
N(k)C(x,y)&=f_{I}(k)^{-1}C(x,y)f_{I}(k),\\
k^{-1}[y]&=F_{J}(K)^{-1}Y.\EnD{aligned}$$
we now fOCuS oUR attentiOn On tHe syMMEtry bReakINg At the fixEd poinTS. SInCe the faCtOr grouP $K/k_{J}$ aCts On $\matHBb{S}^7$ tRivialLy | )&\to M(k) A(x,k^{-1}[y])\ \
C(x,y)&\ to N( k)C (x, k^ {-1} [y]) \endaligned\ri g ht., $$ are physical. Here$M(k) $a nd $ N (k )$ ar e matri x r e p res en ta tio ns of $K$and $k[y]$ is the im age o f the point$ y\ in\mathbb{ S}^ 7$ under the op eratio nof$ k\inK$. Butunlike the st andard or bi f olding conditi o n s, wesuppose that $M(K ) \n e N(K)$.
In or der to d e te r m ine au tomorphism sthata re resp o ns i b l e f o r the symmetr y breaking( \[2 -04\]) ,wem ust fi rst s el e ctthe group $ K$.Suppose $ K=\mat h bb{Z}_4 \ times\m athbb{ Z}_ 2$. Fur t he r, su pp o set ha t t h e s ubgroup$H $in $S pin( 8 ) $ is g ene rate d bythe operators $I =R_7 $ an d $J= L_7^{ -1}L _5 L_2$, which arede fined in Append ix B . Obvious ly, $ IJ= JI $ and $I^4=J ^2= 1$. It fol lows fr o m t hi s t ha t $H=H_{I}\times H _{ J } $, where t he sub g ro up s $H_{I}$ a nd$H_{ J } $ are gen e ra ted by $ I$ and $J $, respec ti vely,i. e., $H \sime q \mat hbb{Z} _4\times \math b b{Z}_2$. Defin e the action o f $ K $ o n $\m ath bb{S}^7$ an d th e rep rese n ta tio n s $K\ to M( K) $ a n d $K\to N(K)$ as fo ll ows. L et $f $, $f_{I}$, a nd $f_{J}$ b e homomor phis m so f $K$ onto $H$ , $H_ {I}$, and$ H_{J}$ r espec tively.Then $$\b e g in{align ed}
M( k)A (x, y ) &= f(k)^{-1}A(x, y ) f(k) ,\ \
N(k)C (x, y)&=f_{ I}( k)^ {-1 }C( x, y)f_{I}(k ),\\
k^{ -1 }[ y] &= f_{ J}(k) ^ {-1}y.\e nd {al ig ned }$$
W e nowfocus our a tt e nti on on t h es y mmet ry b reak ing a t the fix e d p oints.Since the fa c torgr ou p $K/K_ {J}$ acts on$\ mathbb{S}^ 7$ tr iviall y | )&\to M(k)_A(x,k^{-1}[y])\\
C(x,y)&\to N(k)C(x,k^{-1}[y])\endaligned\right.,$$_are physical. Here $M(k)$_and $N(k)$_are_matrix representations_of_$K$ and $k[y]$_is the image_of the point $y\in\mathbb{S}^7$_under the operation_of_$k\in K$. But unlike the standard orbifolding conditions, we suppose that $M(K)\ne N(K)$.
In order_to_determine automorphisms_that_are_responsible for the symmetry breaking_(\[2-04\]), we must first select_the group_$K$. Suppose $K=\mathbb{Z}_4\times\mathbb{Z}_2$. Further, suppose that the subgroup_$H$_in $Spin(8)$ is_generated by the operators $I=R_7$ and $J=L_7^{-1}L_5L_2$, which are_defined in Appendix B. Obviously, $IJ=JI$_and $I^4=J^2=1$. It_follows_from_this that $H=H_{I}\times H_{J}$,_where the subgroups $H_{I}$ and $H_{J}$_are generated by $I$ and $J$,_respectively, i.e., $H\simeq\mathbb{Z}_4\times\mathbb{Z}_2$. Define the action of_$K$ on $\mathbb{S}^7$ and the representations_$K\to M(K)$ and $K\to N(K)$_as follows._Let $f$, $f_{I}$, and $f_{J}$_be homomorphisms of_$K$ onto_$H$, $H_{I}$, and_$H_{J}$ respectively. Then $$\begin{aligned}
M(k)A(x,y)&=f(k)^{-1}A(x,y)f(k),\\
N(k)C(x,y)&=f_{I}(k)^{-1}C(x,y)f_{I}(k),\\
k^{-1}[y]&=f_{J}(k)^{-1}y.\end{aligned}$$
We now focus_our attention on_the symmetry breaking at the fixed_points._Since the factor_group_$K/K_{J}$_acts on_$\mathbb{S}^7$ trivially |
forcing $r$ to be odd. This is significant as the fields $\mathbb{F}_{2^m}$ are the most commonly used in modern engineering.
In Section 2 we considered irreducible composed products of the form $f \odot
\Phi_m.$ In particular, we derived the construction of a new class of irreducible polynomials in Theorem \[thm 3\]. It is natural to consider other classes of polynomials and substitute them for $\Phi_m$ and see what the result may be. We also gave formulas for the linear complexity of $ST$ when $\Phi_{2^n},\
\Phi_r$ are characteristic polynomials of the homogeneous LRS’s $S,\ T,$ respectively. We showed that by letting $n \rightarrow \infty,$ the linear complexity of $ST$ will approach infinity.
Another matter of interest is the factorization of composed products. Since the minimal polynomial of a LRS, say $ST,$ is an irreducible factor of some composed product, this has applications in stream cipher theory, LFSR and LRS in general. D. Mills (2001) [@Mills] had already studied the factorization of arbitrary composed products. In particular, if $\deg f = m$ and $\deg g = n$ with $f,\ g$ irreducible over ${\mathbb{F}_q},$ Mills gave $d = \gcd(m,n)$ as an upper bound for the number of irreducible factors that $f \diamond g$ could decompose into. He also gave the possible degrees that these irreducible factors may attain. As a result, we now know the possible linear complexities that $ST$ could attain. On the other hand his work was generalized for two arbitrary irreducible polynomials $f$ and $g.$ In the case that at least one of these polynomials belongs to a certain class of polynomials with well defined properties, we wonder if it could be possible to obtain more precise information regarding the number of irreducible factors and their degrees. For instance, in the case of $f \odot \Phi_m,$ can we know precisely the degrees of the irreducible factors? Can we know precisely in how many irreducible factors does $f \odot \Phi_m$ decompose into? Note that the subject of the factorization of composed products is one for which very little research has been done. Currently, the authors were able to find only one paper [@Mills] on this matter and they feel this is a topic that has been somewhat neglected.
[99]{}
T. M | forcing $ r$ to be odd. This is significant as the fields $ \mathbb{F}_{2^m}$ are the about normally practice in modern engineering.
In department 2 we considered irreducible composed product of the shape $ f \odot
\Phi_m.$ In particular, we deduce the construction of a new course of irreducible polynomials in Theorem \[thm 3\ ]. It is natural to consider other class of polynomials and substitute them for $ \Phi_m$ and see what the result may be. We also gave formula for the linear complexity of $ ST$ when $ \Phi_{2^n},\
\Phi_r$ are characteristic polynomials of the homogeneous LRS ’s $ S,\ T,$ respectively. We showed that by get $ n \rightarrow \infty,$ the linear complexity of $ ST$ will approach infinity.
Another topic of interest is the factorization of compose products. Since the minimal polynomial of a LRS, say $ ST,$ is an irreducible factor of some compose product, this has applications in stream cipher hypothesis, LFSR and LRS in general. D. Mills (2001) [ @Mills ] had already analyze the factorization of arbitrary composed products. In particular, if $ \deg f = m$ and $ \deg g = n$ with $ f,\ g$ irreducible over $ { \mathbb{F}_q},$ Mills gave $ d = \gcd(m, n)$ as an upper bound for the number of irreducible factors that $ f \diamond g$ could disintegrate into. He besides gave the possible degrees that these irreducible divisor may attain. As a result, we now acknowledge the possible linear complexities that $ ST$ could attain. On the early hand his work was generalized for two arbitrary irreducible polynomials $ f$ and $ g.$ In the case that at least one of these polynomials belongs to a certain class of polynomials with well defined properties, we wonder if it could be potential to obtain more accurate information regarding the number of irreducible factors and their degrees. For instance, in the shell of $ f \odot \Phi_m,$ can we know precisely the degrees of the irreducible factors? Can we acknowledge precisely in how many irreducible factors does $ f \odot \Phi_m$ decompose into? Note that the subject of the factorization of compose products is one for which very fiddling research has been done. presently, the authors were able to find only one paper [ @Mills ] on this topic and they feel this is a topic that has been slightly neglected.
[ 99 ] { }
T. M | fogcing $r$ to be odd. This ir significant as the fmelds $\mzthbb{F}_{2^m}$ xre the most commonly used ii moeern tugineering.
In Section 2 we conspdered ireedurible composed pckducts of the norm $y \idot
\Phi_m.$ In patticular, we gerived the cotsgrbction of a new class of irreducible polynokiwls in Theorem \[thm 3\]. It js natural to consider other classss of pmlynomials anc substitute them for $\Phi_m$ and see what the resupt may be. Wg alfi gave formuuas for tht ninear comklexity of $ST$ when $\Phi_{2^n},\
\Phi_r$ are zharaeteristic piltnolhals of the homodeneous LRS’s $S,\ T,$ res[ectiveky. We showed tmat bb lerting $n \rightarrow \inhty,$ the linear complgxity of $SD$ cill approach infinitt.
Abothet matder ud ivtedext is thf fectorizatioh of composwd products. Since tne ninimal polynkmial jf a LRS, say $ST,$ is an irreducible factor mf aome composed product, tyis has applications ln stream cipher theory, LFSR and LRS in general. D. Mills (2001) [@Mhlls] iaa aorcady wtkdied the factorization of arbitrary composed lrpdlcts. In particulav, if $\deg f = m$ and $\ceh b = n$ with $f,\ g$ krreduejbme over ${\mathbb{F}_q},$ Mllls gade $d = \gcd(m,n)$ as an ipper bound for the number if irreducibjw factors that $f \dnamond g$ couud dgcompoxe into. He also gave thz possjble degreed that thsre irreducible fxctprv may atuxin. As a result, wq now knox the possibue lonear somplexitifs that $ST$ could attain. Ln thg othes hand his work was generalized for two acuitrary irredocitle polynomnals $f$ and $g.$ In the case that at keast oue of ghese polyhomials belongs to a certain cldds of polynokials wieh wwll eefined oroperties, we eonder if it could ve possible to obtein okre precise infirmqtion regarding thd ntmhec of ysreducible fdctofs xmd thdir degrees. Wor onstance, in the case of $r \odot \Phi_m,$ can we kkow preciwely the degrees of tne irreducible facuors? Cen we ynow lresisely in how many irreducible factors foef $f \odot \Phi_i$ degompjse into? Npte that the subject of the factorizatiin of composed proeucts is one for whncm very littlx resewrch has teen done. Currently, rhe authors were sble to find only one paper [@Milld] on this matter and they feel this is a topic that has been somewhat negoected.
[99]{}
V. I | forcing $r$ to be odd. This is the $\mathbb{F}_{2^m}$ are most commonly used 2 considered irreducible composed of the form \odot \Phi_m.$ In particular, we derived construction of a new class of irreducible polynomials in Theorem \[thm 3\]. It natural to consider other classes of polynomials and substitute them for $\Phi_m$ and what result be. also gave formulas for the linear complexity of $ST$ when $\Phi_{2^n},\ \Phi_r$ are characteristic polynomials of homogeneous LRS’s $S,\ T,$ respectively. We showed that letting $n \rightarrow \infty,$ linear complexity of $ST$ will infinity. matter of is factorization composed products. Since minimal polynomial of a LRS, say $ST,$ is an irreducible factor of some composed product, this has in stream LFSR and in D. (2001) [@Mills] had the factorization of arbitrary composed products. $\deg f = m$ and $\deg g = with $f,\ irreducible over ${\mathbb{F}_q},$ Mills gave $d \gcd(m,n)$ as an upper bound for the number irreducible factors that $f \diamond g$ could decompose into. He also gave the possible degrees irreducible factors may attain. a result, we know possible complexities $ST$ could On the other hand his work was generalized for two arbitrary polynomials $f$ and $g.$ In the case that at least these belongs to a class of polynomials with defined we wonder if it possible obtain regarding number irreducible factors and their For instance, in the case $f \odot \Phi_m,$ can of the irreducible factors? Can we know precisely how many irreducible factors does $f \odot decompose into? Note that the subject of the factorization of composed products one for little research has been done. Currently, the authors able to find only paper [@Mills] on this matter and they feel this a that has somewhat neglected. [99]{} M | forcing $r$ to be odd. This is signIficant as tHe fieLds $\MatHbB{F}_{2^m}$ aRe thE most commonly uSEd in Modern engineering.
In SecTion 2 wE cONsidEReD irreDucible COmPOSed PrOdUctS oF ThE form $F \odOt
\Phi_m.$ IN particulaR, we DeRived the consTRuCtion of a neW clAss of irreducIblE polynOmIalS In TheOreM \[thm 3\]. IT is natURal to cOnsider otHeR ClasseS Of polynOMIaLs anD substitute them foR $\phI_M$ and see what the Result MaY Be. wE AlsO gaVe formulas FoR the lINear comPLeXITY of $st$ when $\Phi_{2^n},\
\Phi_r$ Are characteRIstIc polyNoMiaLS of the HomogEnEOus lRS’s $S,\ T,$ respeCtivEly. We showEd that BY lettinG $N \rightaRrow \inFty,$ The LineAR cOmPleXiTY of $st$ wIll APprOach infiNiTy.
anothEr maTTER Of inTerEst iS the fActorization oF coMposED prOductS. SincE the MiNimal PolynoMial oF a lRS, say $ST,$ is an irrEducIble factoR of SoMe cOmPosed PRoduct, ThiS haS applicAtions iN StrEaM CIPhEr theory, LFSR and LRS In GENeRal. D. MillS (2001) [@Mills] HAd AlREady studIeD thE facTORizatIon oF ArBitrary cOmposeD PrOdUcts. In pArTiculaR, iF $\deG f = m$ And $\deG G = n$ wiTh $f,\ g$ irReduciblE over ${\MAthbb{F}_q},$ Mills gaVE $d = \gcd(m,n)$ as an upPEr BOUnD For tHe nUmber of irreDuciBLe faCtorS ThAt $f \DIamonD g$ couLd DEcOMpose into. He also gave ThE possiBle deGrees that thesE irreducibLE FActors maY attAIn. aS a result, we now kNow thE possible lINear compLexitIes that $St$ could attAIN. On the otHer HanD hiS woRK WaS generalized fOR Two aRbItrary iRreDucible PolYnoMiaLs $f$ AnD $g.$ In the caSe that at LeAsT oNe Of tHese pOLynomialS bEloNgS to A certAIn clasS of poLynoMiAlS WitH well deFInED PropErTiEs, we WonDeR if it CoulD Be pOssible To obtain mOre PReciSe InFormatiOn regarding thE nUmber of irrEdUciBle facTORs and theIr degrees. For instance, in tHE case of $F \odOt \Phi_M,$ can We know preCisEly the DegREes of tHe irreDucibLe FacTORs? Can WE KnOw pReCisely in hoW MAny IrredUcIble Factors Does $f \odot \Phi_m$ decomPOse Into? Note that tHe sUbjeCT Of The FAcTOriZaTIon OF Composed productS is one for wHiCH vEry little rESeaRcH has beeN done. CuRrentLY, the autHors were aBle to find OnLy onE PApeR [@Mills] on thIs matter And they feEL this IS a Topic ThaT has beEn SomEwhat NeglecTEd.
[99]{}
T. m | forcing $r$ to be odd. Th is is sign ifica ntasth e fi elds $\mathbb{F}_{ 2 ^m}$ are the most commonly used i n mod e rn engi neering .
I n Se ct io n 2 w e c onsid ere d irred ucible com pos ed products of th e form $f\od ot
\Phi_m.$Inpartic ul ar, we de riv ed th e cons t ructio n of a ne wc lass o f irredu c i bl e po lynomials in Theo r em \[thm 3\]. Itis nat ur a lt o co nsi der othercl asses of poly n om i a l s a n d substitutethem for $\ P hi_ m$ and s eew hat th e res ul t ma y be. We al so g ave formu las fo r the li n ear com plexit y o f $ ST$w he n$\P hi _ {2^ n }, \
\ P hi_ r$ are c ha ra cteri stic p o l ynom ial s of thehomogeneous L RS’ s $S , \ T ,$ re spect ivel y. We s howedthatby letting $n \ri ghta rrow \inf ty, $the l inear comple xit y o f $ST$will ap p roa ch i n fi nity.
Another mat te r of interes t is t h efa c torizati on of com p o sed p rodu c ts . Sincethe mi n im al polyno mi al ofaLRS , s ay $S T ,$ i s an i rreducib le fa c tor of some co m posed product , t h i sh as a ppl ications in str e am c iphe r t heo r y, LF SR an dL RS in general. D. Mill s(2001) [@Mi lls] had alre ady studie d t he facto riza t io n of arbitrarycompo sed produc t s. In pa rticu lar, if$\deg f = m $ and $\ deg g= n $ w i t h$f,\ g$ irred u c ible o ver ${\ mat hbb{F}_ q}, $ M ill s g av e $d = \g cd(m,n)$ a san u ppe r bou n d for th enum be r o f irr e ducibl e fac tors t ha t $f \diamo n dg $ cou ld d ecom pos einto. Hea lso gave t he possib led egre es t hat the se irreducibl efactors ma yatt ain. A s a result , we now know the possi b le line arcompl exit ies that$ST $ coul d a t tain.On the othe rhan d his w o r kwas g eneralized f ortwo a rb itra ry irre ducible polynomial s $f $ and $g.$ In th e ca s e t hat at lea st one o f these polynom ials belon gs to a certain cla ss of pol ynomial s wit h well d efined pr operties, w e wo n d erif it coul d be pos sible too btain mo re pr eci se inf or mat ion r egardi n g t he nu mber o firredu cible f actors a nd their degrees. For i nstanc e, in th e case of $f \od ot \Phi_m ,$ c an we know pr eci selythe degre es o f t hei rredu cibl e factors? Ca n w e kn ow precisel y i n h ow ma nyi rreduc ible factors does $f\ odot \Phi_m$ d ecom p o seint o ? No te that the subj ect o f the fact or ization ofcomposed p r oduct s is o ne for whichv e ry little res ear ch has be endo n e. Curr en tl y , theauth or s were ablet o fi n d only one paper[@Mil l s ] ont his matt er and th e y fe el this is a topic th at has bee n som ewhat n eg lected .
[9 9]{}
T. M | forcing_$r$ to_be odd. This is_significant as_the_fields $\mathbb{F}_{2^m}$_are_the most commonly_used in modern_engineering.
In Section 2 we_considered irreducible composed_products_of the form $f \odot
\Phi_m.$ In particular, we derived the construction of a new_class_of irreducible_polynomials_in_Theorem \[thm 3\]. It is_natural to consider other classes_of polynomials_and substitute them for $\Phi_m$ and see what_the_result may be._We also gave formulas for the linear complexity of_$ST$ when $\Phi_{2^n},\
\Phi_r$ are characteristic polynomials_of the homogeneous_LRS’s_$S,\_T,$ respectively. We showed_that by letting $n \rightarrow \infty,$_the linear complexity of $ST$ will_approach infinity.
Another matter of interest is the_factorization of composed products. Since the_minimal polynomial of a LRS,_say $ST,$_is an irreducible factor of_some composed product,_this has_applications in stream_cipher theory, LFSR and LRS in_general. D. Mills_(2001) [@Mills] had already studied the_factorization_of arbitrary composed_products._In_particular, if_$\deg f =_m$_and $\deg_g_= n$ with $f,\ g$ irreducible_over_${\mathbb{F}_q},$ Mills gave $d = \gcd(m,n)$ as_an upper bound for_the_number of irreducible factors_that $f \diamond g$ could_decompose into. He also gave the_possible degrees_that these_irreducible factors may attain. As a result, we now know the_possible linear complexities that $ST$ could_attain. On the other_hand his_work_was generalized for_two_arbitrary irreducible_polynomials $f$ and $g.$ In the case_that at_least one of these polynomials belongs_to a certain class_of_polynomials with well defined properties, we_wonder if it could be possible_to obtain more precise information_regarding_the_number of irreducible factors and_their degrees. For instance, in the_case of $f_\odot \Phi_m,$ can we know precisely the_degrees_of the irreducible factors? Can we_know_precisely in how many irreducible factors_does_$f_\odot \Phi_m$ decompose into? Note_that the subject of the factorization_of composed products is one for which very little_research has been_done. Currently, the authors were_able_to_find only one paper [@Mills] on this matter and they_feel this_is a topic_that has been somewhat neglected.
[99]{}
T. M |
\|_2 \lesssim
\|Q_ND_{x}^{\alpha}g\|_{L_x^{\infty}l_N^1}\|f\|_2.
\label{prod_rule}$$ A classical Leibnitz type inequality for fractional derivatives is the following (see [@KPV1]).
\[stand\_prod\] Let $0 < \alpha, \alpha_1, \alpha_2 < 1$, $\alpha=\alpha_1+\alpha_2$, $1 <p,p_1,p_2 < \infty$, and $\frac{1}{p}=\frac{1}{p_1}+\frac{1}{p_2}$. In addition, the $\alpha_1=\alpha$, $p=p_2$, and $p_1=\infty$ is allowed. Then the following holds for functions $f,g$ on $\mathbb R^n$. $$\|D_x^{\alpha}(fg)-D^{\alpha}(f)g-fD_x^{\alpha}(g)\|_p \lesssim
\|D_x^{\alpha_1}g\|_{p_1}\|D_x^{\alpha_2}f\|_{p_2} \notag$$
The proof uses the Littlewood-Paley Theorem (see [@S]), which states that for any function $f$, if $1 < p < \infty$, then $$\label{little_pale}
\|Q_N(f)\|_{L_x^pl_N^2} \lesssim \|f\|_p \lesssim \|Q_N(f)\|_{L_x^pl_N^2}.$$
Lemma \[stand\_prod\] is not sufficient for our argument in the previous section, since we need to put the *derivative* term in the infinity norm. A product rule like this can be obtained by following the proof of Lemma \[stand\_prod\] line for line. The only difference is that since fails for $p=\infty$, $\|Q_N(D_x^{\alpha}g)\|_{L_{x}^{\infty}l_N^2}$ is not equivalent to $\|D_x^{\alpha}g\|_{\infty}$. This idea was inspired by [@KT], where the authors use $\|Q_N\cdot\|_{l_N^2L_x^{4}L_T^{\infty}}$ in an estimate where the $\|\cdot\|_{L_x^{4}L_T^{\infty}}$ norm may fail.
\[my\_prod | \|_2 \lesssim
\|Q_ND_{x}^{\alpha}g\|_{L_x^{\infty}l_N^1}\|f\|_2.
\label{prod_rule}$$ A classical Leibnitz type inequality for fractional derivatives is the following (visualize [ @KPV1 ]).
\[stand\_prod\ ] lease $ 0 < \alpha, \alpha_1, \alpha_2 < 1 $, $ \alpha=\alpha_1+\alpha_2 $, $ 1 < p, p_1,p_2 < \infty$, and $ \frac{1}{p}=\frac{1}{p_1}+\frac{1}{p_2}$. In addition, the $ \alpha_1=\alpha$, $ p = p_2 $, and $ p_1=\infty$ is permit. Then the following holds for function $ f, g$ on $ \mathbb R^n$. $ $ \|D_x^{\alpha}(fg)-D^{\alpha}(f)g - fD_x^{\alpha}(g)\|_p \lesssim
\|D_x^{\alpha_1}g\|_{p_1}\|D_x^{\alpha_2}f\|_{p_2 } \notag$$
The validation uses the Littlewood - Paley Theorem (see [ @S ]), which states that for any function $ f$, if $ 1 < p < \infty$, then $ $ \label{little_pale }
\|Q_N(f)\|_{L_x^pl_N^2 } \lesssim \|f\|_p \lesssim \|Q_N(f)\|_{L_x^pl_N^2}.$$
Lemma \[stand\_prod\ ] is not sufficient for our argumentation in the previous section, since we want to place the * derivative * term in the infinity norm. A product rule like this can be obtain by following the proof of Lemma \[stand\_prod\ ] line for wrinkle. The only difference is that since fails for $ p=\infty$, $ \|Q_N(D_x^{\alpha}g)\|_{L_{x}^{\infty}l_N^2}$ is not equivalent to $ \|D_x^{\alpha}g\|_{\infty}$. This idea was inspire by [ @KT ], where the authors use $ \|Q_N\cdot\|_{l_N^2L_x^{4}L_T^{\infty}}$ in an estimate where the $ \|\cdot\|_{L_x^{4}L_T^{\infty}}$ norm may fail.
\[my\_prod | \|_2 \ledssim
\|Q_ND_{x}^{\alpha}g\|_{L_x^{\infty}l_N^1}\|n\|_2.
\label{prod_rule}$$ A classiral Leignitz tyoe inequality for fractional dwrivaupves is the following (see [@KPV1]).
\[snand\_prod\] Oet $0 < \alpha, \alpiz_1, \alpha_2 < 1$, $\alpgw=\alpka_1+\elpha_2$, $1 <p,p_1,p_2 < \infjy$, and $\frac{1}{p}=\xrac{1}{p_1}+\frac{1}{p_2}$. In ddaicion, the $\alpha_1=\alpha$, $p=p_2$, and $p_1=\infty$ is allowec. Hhen the folloring rolda for functions $f,g$ on $\mathbb R^n$. $$\|D_x^{\zlpha}(fg)-V^{\alpha}(f)g-fD_x^{\alphs}(g)\|_p \lesssim
\|D_x^{\alpha_1}g\|_{p_1}\|D_x^{\alphw_2}f\|_{p_2} \jotag$$
The proof used the Littlgsooq-Paley Theoreo (see [@S]), whpeh states tgat for any function $f$, if $1 < p < \knfty$, then $$\labeo{luttpg_pale}
\|Q_N(f)\|_{L_x^pl_I^2} \lessfim \|f\|_p \lessslk \|Q_N(f)\|_{L_f^pl_N^2}.$$
Lemka \[stand\_prod\] ix nmt wufficient for our arjument in the previoos section, snnce we need to put tye *derieatiee* tdem kn uhe ihfinitj nkrm. A prodhct rule lije this can be obtaonqe by followinf the [rjof of Lemma \[stand\_prod\] line for line. Tht onlg difference is that sibce fails for $p=\infty$, $\|E_N(D_x^{\alpha}d)\|_{L_{x}^{\infty}l_N^2}$ is not equivalent to $\|D_x^{\alpha}g\|_{\infty}$. Thiv idee das ikspifwd by [@KT], where the authors use $\|Q_N\cdot\|_{l_N^2L_x^{4}L_T^{\inftr}}$ im sn estimate whcre the $\|\cdot\|_{L_x^{4}L_T^{\ingtj}}$ mjrm may fail.
\[mi\_prod | \|_2 \lesssim \|Q_ND_{x}^{\alpha}g\|_{L_x^{\infty}l_N^1}\|f\|_2. \label{prod_rule}$$ A classical Leibnitz for derivatives is following (see [@KPV1]). \alpha_1, < 1$, $\alpha=\alpha_1+\alpha_2$, <p,p_1,p_2 < \infty$, $\frac{1}{p}=\frac{1}{p_1}+\frac{1}{p_2}$. In addition, the $\alpha_1=\alpha$, $p=p_2$, $p_1=\infty$ is allowed. Then the following holds for functions $f,g$ on $\mathbb R^n$. \lesssim \|D_x^{\alpha_1}g\|_{p_1}\|D_x^{\alpha_2}f\|_{p_2} \notag$$ The proof uses the Littlewood-Paley Theorem (see [@S]), which states for function if < p < \infty$, then $$\label{little_pale} \|Q_N(f)\|_{L_x^pl_N^2} \lesssim \|f\|_p \lesssim \|Q_N(f)\|_{L_x^pl_N^2}.$$ Lemma \[stand\_prod\] is not sufficient our argument in the previous section, since we to put the *derivative* in the infinity norm. A rule this can obtained following proof of Lemma line for line. The only difference is that since fails for $p=\infty$, $\|Q_N(D_x^{\alpha}g)\|_{L_{x}^{\infty}l_N^2}$ is not equivalent to This idea by [@KT], the use in an estimate $\|\cdot\|_{L_x^{4}L_T^{\infty}}$ norm may fail. \[my\_prod | \|_2 \lesssim
\|Q_ND_{x}^{\alpha}g\|_{L_x^{\infty}l_n^1}\|f\|_2.
\label{proD_rule}$$ a clAssIcAl LeIbniTz type inequaliTY for Fractional derivatives iS the fOlLOwinG (SeE [@KPV1]).
\[sTand\_proD\] leT $0 < \ALphA, \aLpHa_1, \aLpHA_2 < 1$, $\aLpha=\aLphA_1+\alpha_2$, $1 <p,P_1,p_2 < \infty$, and $\FraC{1}{p}=\Frac{1}{p_1}+\frac{1}{p_2}$. In ADdItion, the $\alPha_1=\Alpha$, $p=p_2$, and $p_1=\iNftY$ is allOwEd. THEn the FolLowinG holds FOr funcTions $f,g$ on $\MaTHbb R^n$. $$\|D_X^{\Alpha}(fg)-d^{\ALpHa}(f)g-FD_x^{\alpha}(g)\|_p \lesssim
\|d_X^{\aLPha_1}g\|_{p_1}\|D_x^{\alpha_2}f\|_{p_2} \Notag$$
THe PRoOF UseS thE LittlewooD-PAley THEorem (seE [@s]), wHICH stATes that for any Function $f$, if $1 < P < \InfTy$, then $$\LaBel{LIttle_pAle}
\|Q_N(F)\|_{L_X^Pl_N^2} \Lesssim \|f\|_p \leSssiM \|Q_N(f)\|_{L_x^pl_N^2}.$$
lemma \[sTAnd\_prod\] IS not sufFicienT foR ouR argUMeNt In tHe PRevIOuS seCTioN, since we NeEd To put The *dERIVAtivE* teRm in The inFinity norm. A prOduCt ruLE liKe thiS can bE obtAiNed by FollowIng thE pRoof of Lemma \[stanD\_proD\] line for lIne. thE onLy DiffeREnce is ThaT siNce failS for $p=\inFTy$, $\|Q_n(D_X^{\ALPhA}g)\|_{L_{x}^{\infty}l_N^2}$ is not eqUiVALeNt to $\|D_x^{\alPha}g\|_{\inFTy}$. thIS idea was InSpiRed bY [@kt], wherE the AUtHors use $\|Q_n\cdot\|_{l_n^2l_x^{4}l_T^{\Infty}}$ in An EstimaTe WheRe tHe $\|\cdoT\|_{l_x^{4}L_T^{\Infty}}$ nOrm may faIl.
\[my\_pROd | \|_2 \lesssim
\|Q_ND_{x}^{ \alpha}g\| _{L_x ^{\ inf ty }l_N ^1}\ |f\|_2.
\label { prod _rule}$$ A classical L eibni tz type in equal ity for fr a c tio na lder iv a ti ves i s t he foll owing (see [@ KP V1]).
\[sta n d\ _prod\] Le t $ 0 < \alpha,\al pha_1, \ alp h a_2 < 1$ , $\a lpha=\ a lpha_1 +\alpha_2 $, $1 <p, p _1,p_2< \i nfty $, and $\frac{1}{ p }= \ frac{1}{p_1}+\ frac{1 }{ p _2 } $ . I n a ddition, t he $\al p ha_1=\a l ph a $ , $p = p_2$, and $p_ 1=\infty$ i s al lowed. T hen the fo llowi ng hol ds for func tion s $f,g$ o n $\ma t hbb R^n $ . $$\|D _x^{\a lph a}( fg)- D ^{ \a lph a} ( f)g - fD _x^ { \al pha}(g)\ |_ p\less sim\ | D _ x^{\ alp ha_1 }g\|_ {p_1}\|D_x^{\ alp ha_2 } f\| _{p_2 } \no tag$ $
Theproofusesth e Littlewood-Pa leyTheorem ( see [ @S] ), whic h state s t hat for an y funct i on$f $ , if $1 < p < \infty$, t h e n$$\label {littl e _p al e }
\|Q_N( f) \|_ {L_x ^ p l_N^2 } \l e ss sim \|f\ |_p \l e ss si m \|Q_N (f )\|_{L _x ^pl _N^ 2}.$$
Lem ma \[s tand\_pr od\]i s not sufficie n t for our arg u me n t i n the pr evious sect ion, sinc e we ne edt o put the*d e ri v ative* term in thein finity norm . A product r ule like t h i s can beobta i ne d by followingthe p roof of Le m ma \[sta nd\_p rod\] li ne for li n e . The on lydif fer enc e is that since f a i ls f or $p=\in fty $, $\|Q _N( D_x ^{\ alp ha }g)\|_{L_ {x}^{\in ft y} l_ N^ 2}$ is n o t equiva le ntto $\ |D_x^ { \alpha }g\|_ {\in ft y} $ . T his ide a w a s ins pi re d by [@ KT ], wh eret heauthors use $\|Q _N\ c dot\ |_ {l _N^2L_x ^{4}L_T^{\inf ty }}$ in anes tim ate wh e r e the $\ |\cdot\|_{L_x^{4}L_T^{\ i nfty}}$ no rm ma y fa il.
\[my \_p rod | \|_2 \lesssim
\|Q_ND_{x}^{\alpha}g\|_{L_x^{\infty}l_N^1}\|f\|_2.
\label{prod_rule}$$_A classical_Leibnitz type inequality for_fractional derivatives_is_the following_(see_[@KPV1]).
\[stand\_prod\] Let $0_< \alpha, \alpha_1,_\alpha_2 < 1$, $\alpha=\alpha_1+\alpha_2$,_$1 <p,p_1,p_2 <_\infty$,_and $\frac{1}{p}=\frac{1}{p_1}+\frac{1}{p_2}$. In addition, the $\alpha_1=\alpha$, $p=p_2$, and $p_1=\infty$ is allowed. Then the following_holds_for functions_$f,g$_on_$\mathbb R^n$. $$\|D_x^{\alpha}(fg)-D^{\alpha}(f)g-fD_x^{\alpha}(g)\|_p \lesssim
\|D_x^{\alpha_1}g\|_{p_1}\|D_x^{\alpha_2}f\|_{p_2} \notag$$
The_proof uses the Littlewood-Paley Theorem_(see [@S]),_which states that for any function $f$, if_$1_< p <_\infty$, then $$\label{little_pale}
\|Q_N(f)\|_{L_x^pl_N^2} \lesssim \|f\|_p \lesssim \|Q_N(f)\|_{L_x^pl_N^2}.$$
Lemma \[stand\_prod\] is_not sufficient for our argument in_the previous section,_since_we_need to put the_*derivative* term in the infinity norm._A product rule like this can_be obtained by following the proof of_Lemma \[stand\_prod\] line for line. The_only difference is that since_fails for_$p=\infty$, $\|Q_N(D_x^{\alpha}g)\|_{L_{x}^{\infty}l_N^2}$ is not equivalent_to $\|D_x^{\alpha}g\|_{\infty}$. This_idea was_inspired by [@KT],_where the authors use $\|Q_N\cdot\|_{l_N^2L_x^{4}L_T^{\infty}}$ in_an estimate where_the $\|\cdot\|_{L_x^{4}L_T^{\infty}}$ norm may fail.
\[my\_prod |
}}&:=\left(u_i^{h,n}+u_i^{h,n-1}\right)/2,\\
F_{iJ}^{h,n-\frac{1}{2}}&:=\delta_{iJ}+u_{i,J}^{h,n-\frac{1}{2}},\\
F_{iJ,K}^{h,n-\frac{1}{2}}&:=u_{i,JK}^{h,n-\frac{1}{2}},\\
v_i^{h,n-\frac{1}{2}}&:=(u_i^{h,n}-u_i^{h,n-1})/\Delta t,\\
\Delta^n u_i^h&:=\left(u_i^{h,n+1}-u_i^{h,n-1}\right)/2,\\
\Delta^n\!F^h_{iJ}&:=\Delta^n u_{i,J}^h,\\
\Delta^n\!F^h_{iJ,K}&:=\Delta^n u_{i,JK}^h.\end{aligned}$$ We thus have $\bs{\zeta}^{h,n-\frac{1}{2}}$ as a short-hand notation of the array of the components, $\{F_{11}^{h,n-\frac{1}{2}},\dots,F_{33}^{h,n-\frac{1}{2}},\dots,F_{11,1}^{h,n-\frac{1}{2}},\dots,F_{33,3}^{h,n-\frac{1}{2}}\}$, and write, e.g., $\tilde{\Psi}(F_{11}^{h,n-\frac{1}{2}},\dots,F_{33}^{h,n-\frac{1}{2}},\dots,F_{11,1}^{h,n-\frac{1}{2}},\dots,F_{33,3}^{h,n-\frac{1}{2}})$ as $\tilde{\Psi}(\bs{\zeta}^{h,n-\frac{1}{2}})$.
In both schemes temporal approximations to the acceleration $\left\{\ddot{u}_i^h\right\}^n$ and velocity $\left\{ \dot{u}_i^h\right\}^n$ are given by the standard second- and first-order stencils, respectively, as:
$$\begin{aligned}
\left\{\ddot{ | } } &: = \left(u_i^{h, n}+u_i^{h, n-1}\right)/2,\\
F_{iJ}^{h, n-\frac{1}{2}}&:=\delta_{iJ}+u_{i, J}^{h, n-\frac{1}{2}},\\
F_{iJ, K}^{h, n-\frac{1}{2}}&:=u_{i, JK}^{h, n-\frac{1}{2}},\\
v_i^{h, n-\frac{1}{2}}&:=(u_i^{h, n}-u_i^{h, n-1})/\Delta t,\\
\Delta^n u_i^h&:=\left(u_i^{h, n+1}-u_i^{h, n-1}\right)/2,\\
\Delta^n\!F^h_{iJ}&:=\Delta^n u_{i, J}^h,\\
\Delta^n\!F^h_{iJ, K}&:=\Delta^n u_{i, JK}^h.\end{aligned}$$ We thus have $ \bs{\zeta}^{h, n-\frac{1}{2}}$ as a short - hand note of the array of the component, $ \{F_{11}^{h, n-\frac{1}{2}},\dots, F_{33}^{h, n-\frac{1}{2}},\dots, F_{11,1}^{h, n-\frac{1}{2}},\dots, F_{33,3}^{h, n-\frac{1}{2}}\}$, and write, e.g., $ \tilde{\Psi}(F_{11}^{h, n-\frac{1}{2}},\dots, F_{33}^{h, n-\frac{1}{2}},\dots, F_{11,1}^{h, n-\frac{1}{2}},\dots, F_{33,3}^{h, n-\frac{1}{2}})$ as $ \tilde{\Psi}(\bs{\zeta}^{h, n-\frac{1}{2}})$.
In both schemes worldly approximations to the acceleration $ \left\{\ddot{u}_i^h\right\}^n$ and velocity $ \left\ { \dot{u}_i^h\right\}^n$ are feed by the standard second- and first - order stencil, respectively, as:
$ $ \begin{aligned }
\left\{\ddot { | }}&:=\lefh(u_i^{h,n}+u_i^{h,n-1}\right)/2,\\
F_{iJ}^{h,n-\frac{1}{2}}&:=\dtlta_{iJ}+u_{i,J}^{h,n-\frac{1}{2}},\\
F_{iL,J}^{h,n-\frar{1}{2}}&:=u_{i,JK}^{h,n-\rrac{1}{2}},\\
v_i^{h,n-\wrac{1}{2}}&:=(u_i^{h,n}-u_i^{h,n-1})/\Delta t,\\
\Delta^n u_i^i&:=\lefr(u_i^{h,n+1}-y_i^{h,n-1}\right)/2,\\
\Delta^n\!F^h_{iJ}&:=\Delga^n u_{i,J}^h,\\
\Dvlta^n\!F^h_{iJ,J}&:=\Delua^n u_{i,JK}^h.\end{alignxs}$$ We thmf habc $\bs{\zzte}^{h,n-\frac{1}{2}}$ as a shprt-hand nodation of the drfav of the components, $\{F_{11}^{h,n-\frac{1}{2}},\dots,F_{33}^{h,n-\frwc{1}{2}},\dots,F_{11,1}^{n,n-\vrac{1}{2}},\dots,F_{33,3}^{h,n-\fras{1}{2}}\}$, anc wrifv, t.g., $\tilde{\Psi}(F_{11}^{h,n-\frac{1}{2}},\dots,F_{33}^{h,n-\frac{1}{2}},\dots,R_{11,1}^{h,n-\frac{1}{2}},\vots,F_{33,3}^{h,n-\frac{1}{2}})$ as $\yilde{\Psi}(\bs{\zeta}^{h,n-\frac{1}{2}})$.
In botj scjemes temporal appgoximations to eye acceleratkon $\left\{\ddot{u}_i^h\right\}^n$ znd velocity $\left\{ \dot{u}_i^h\right\}^n$ xre gnven by the sranfdrd second- end figst-order stengpls, res[ectiveky, as:
$$\begin{aligked}
\leht\{\ddit{ | }}&:=\left(u_i^{h,n}+u_i^{h,n-1}\right)/2,\\ F_{iJ}^{h,n-\frac{1}{2}}&:=\delta_{iJ}+u_{i,J}^{h,n-\frac{1}{2}},\\ F_{iJ,K}^{h,n-\frac{1}{2}}&:=u_{i,JK}^{h,n-\frac{1}{2}},\\ v_i^{h,n-\frac{1}{2}}&:=(u_i^{h,n}-u_i^{h,n-1})/\Delta t,\\ \Delta^n u_i^h&:=\left(u_i^{h,n+1}-u_i^{h,n-1}\right)/2,\\ \Delta^n\!F^h_{iJ,K}&:=\Delta^n We thus $\bs{\zeta}^{h,n-\frac{1}{2}}$ as a of components, $\{F_{11}^{h,n-\frac{1}{2}},\dots,F_{33}^{h,n-\frac{1}{2}},\dots,F_{11,1}^{h,n-\frac{1}{2}},\dots,F_{33,3}^{h,n-\frac{1}{2}}\}$, and e.g., $\tilde{\Psi}(F_{11}^{h,n-\frac{1}{2}},\dots,F_{33}^{h,n-\frac{1}{2}},\dots,F_{11,1}^{h,n-\frac{1}{2}},\dots,F_{33,3}^{h,n-\frac{1}{2}})$ as In both schemes temporal approximations to acceleration $\left\{\ddot{u}_i^h\right\}^n$ and velocity $\left\{ \dot{u}_i^h\right\}^n$ are given by the standard second- and stencils, respectively, as: $$\begin{aligned} \left\{\ddot{ | }}&:=\left(u_i^{h,n}+u_i^{h,n-1}\right)/2,\\
F_{iJ}^{h,n-\fraC{1}{2}}&:=\delta_{iJ}+u_{i,j}^{h,n-\frAc{1}{2}},\\
F_{IJ,K}^{H,n-\Frac{1}{2}}&:=U_{i,JK}^{H,n-\frac{1}{2}},\\
v_i^{h,n-\frac{1}{2}}&:=(U_I^{h,n}-u_I^{h,n-1})/\Delta t,\\
\Delta^n u_i^h&:=\left(U_i^{h,n+1}-u_I^{h,N-1}\RighT)/2,\\
\deLta^n\!F^H_{iJ}&:=\DeltA^N u_{I,j}^H,\\
\DeLtA^n\!f^h_{ij,K}&:=\dElTa^n u_{i,jK}^h.\End{aligNed}$$ We thus hAve $\Bs{\Zeta}^{h,n-\frac{1}{2}}$ as A ShOrt-hand notAtiOn of the array Of tHe compOnEntS, $\{f_{11}^{h,n-\frAc{1}{2}},\dOts,F_{33}^{h,N-\frac{1}{2}},\dOTs,F_{11,1}^{h,n-\fRac{1}{2}},\dots,F_{33,3}^{h,N-\fRAc{1}{2}}\}$, and wRIte, e.g., $\tiLDE{\PSi}(F_{11}^{h,N-\frac{1}{2}},\dots,F_{33}^{h,n-\frac{1}{2}},\dOTs,f_{11,1}^{H,n-\frac{1}{2}},\dots,F_{33,3}^{h,n-\fRac{1}{2}})$ as $\tIlDE{\PSI}(\Bs{\zEta}^{H,n-\frac{1}{2}})$.
In boTh SchemES temporAL aPPROxiMAtions to the acCeleration $\lEFt\{\dDot{u}_i^h\RiGht\}^N$ And velOcity $\LeFT\{ \doT{u}_i^h\right\}^n$ aRe giVen by the sTandarD Second- aND first-oRder stEncIls, RespECtIvEly, As:
$$\BEgiN{AlIgnED}
\leFt\{\ddot{ | }}&:=\left(u_i^{h,n}+u_i^{ h,n-1}\rig ht)/2 ,\\
F_ {i J}^{ h,n- \frac{1}{2}}&: = \del ta_{iJ}+u_{i,J}^{h,n-\ frac{ 1} { 2}}, \ \F_{iJ ,K}^{h, n -\ f r ac{ 1} {2 }}& := u _{ i,JK} ^{h ,n-\fra c{1}{2}},\ \
v _i ^{h,n-\frac{ 1 }{ 2}}&:=(u_i ^{h ,n}-u_i^{h,n -1} )/\Del ta t, \ \
\De lta ^n u_ i^h&:= \ left(u _i^{h,n+1 }- u _i^{h, n -1}\rig h t )/ 2,\\
\Delta^n\!F^h_{i J }& : =\Delta^n u_{i ,J}^h, \\ \D e l ta^ n\! F^h_{iJ,K} &: =\Del t a^n u_{ i ,J K } ^ h.\ e nd{aligned}$$ We thus ha v e $ \bs{\z et a}^ { h,n-\f rac{1 }{ 2 }}$ as a short -han d notatio n of t h e array of thecompon ent s,$\{F _ {1 1} ^{h ,n - \fr a c{ 1}{ 2 }}, \dots,F_ {3 3} ^{h,n -\fr a c { 1 }{2} },\ dots ,F_{1 1,1}^{h,n-\fr ac{ 1}{2 } },\ dots, F_{33 ,3}^ {h ,n-\f rac{1} {2}}\ }$ , and write, e. g.,$\tilde{\ Psi }( F_{ 11 }^{h, n -\frac {1} {2} },\dots ,F_{33} ^ {h, n- \ f r ac {1}{2}},\dots,F_{1 1, 1 } ^{ h,n-\fra c{1}{2 } }, \d o ts,F_{33 ,3 }^{ h,n- \ f rac{1 }{2} } )$ as $\ti lde{\P s i} (\ bs{\zet a} ^{h,n- \f rac {1} {2}}) $ .
I n both schemes temp o ral approximat i ons to the ac c el e r at i on $ \le ft\{\ddot{u }_i^ h \rig ht\} ^ n$ an d velo city$\ l ef t \{ \dot{u}_i^h\righ t\ }^n$ a re gi ven by the st andard sec o n d - and fi rst- o rd e r stencils, re spect ively, as:
$$\begi n{ali gned}
\l eft\{\ddo t { | }}&:=\left(u_i^{h,n}+u_i^{h,n-1}\right)/2,\\
F_{iJ}^{h,n-\frac{1}{2}}&:=\delta_{iJ}+u_{i,J}^{h,n-\frac{1}{2}},\\
F_{iJ,K}^{h,n-\frac{1}{2}}&:=u_{i,JK}^{h,n-\frac{1}{2}},\\
v_i^{h,n-\frac{1}{2}}&:=(u_i^{h,n}-u_i^{h,n-1})/\Delta t,\\
\Delta^n_u_i^h&:=\left(u_i^{h,n+1}-u_i^{h,n-1}\right)/2,\\
\Delta^n\!F^h_{iJ}&:=\Delta^n u_{i,J}^h,\\
\Delta^n\!F^h_{iJ,K}&:=\Delta^n_u_{i,JK}^h.\end{aligned}$$ We thus have_$\bs{\zeta}^{h,n-\frac{1}{2}}$ as_a_short-hand notation_of_the array of_the components, $\{F_{11}^{h,n-\frac{1}{2}},\dots,F_{33}^{h,n-\frac{1}{2}},\dots,F_{11,1}^{h,n-\frac{1}{2}},\dots,F_{33,3}^{h,n-\frac{1}{2}}\}$,_and write, e.g., $\tilde{\Psi}(F_{11}^{h,n-\frac{1}{2}},\dots,F_{33}^{h,n-\frac{1}{2}},\dots,F_{11,1}^{h,n-\frac{1}{2}},\dots,F_{33,3}^{h,n-\frac{1}{2}})$_as $\tilde{\Psi}(\bs{\zeta}^{h,n-\frac{1}{2}})$.
In both_schemes_temporal approximations to the acceleration $\left\{\ddot{u}_i^h\right\}^n$ and velocity $\left\{ \dot{u}_i^h\right\}^n$ are given by the_standard_second- and_first-order_stencils,_respectively, as:
$$\begin{aligned}
\left\{\ddot{ |
x;{\boldsymbol}{\lambda})
=\phi_{\text{v}}(x;{\boldsymbol}{\lambda}),{\nonumber}\\
&\Delta_{\text{v}}^+=-h+\text{v}>0,\quad
\Delta_{\text{v}}^-=-\Delta_{\text{v}}^+<0,
\quad
\tilde{\mathcal{E}}^{\text{os}}_{\text{v}}({\boldsymbol}{\lambda})
<\mathcal{E}_0({\boldsymbol}{\lambda}) \quad(\text{v}>2h).
\label{solpv2}\end{aligned}$$
#### Deformed scatterings
A pseudo (‘pseudo’) virtual state wavefunction will add a new discrete eigenstate at its energy. It is trivial to verify that the pseudo and ‘pseudo’ virtual wavefunctions will add a pole on the positive imaginary $k$-axis at $k=i(\text{v}+h+1)$, $k=i(\text{v}-h)$, respectively, with exactly the same energy of the employed seed solution, $-(h+\text{v}+1)^2$ and $-(\text{v}-h)^2$, respectively. For both the pseudo and ‘pseudo’ virtual wavefunctions, $\Delta_\text{v}^+=-\Delta_\text{v}^-$. This means that the deformation factors of the transmission and reflection amplitudes are the same except for a sign $(-1)^M$: $$\frac{t_{\mathcal D}(k;{\boldsymbol}{\lambda})}{t(k;{\boldsymbol}{\lambda})}=(-1)^M\frac{r_{\mathcal D}(k;{\boldsymbol}{\lambda})}{r(k;{\boldsymbol}{\lambda})}=
\prod_{j=1}^M\frac{k-i\Delta_{d_j}^-}{k+i\Delta_{d_j}^-}.$$
Hyperbolic symmetric top ${\text{{\uppercase\expandafter{\romannumeral2}}}}$ {#sec:hst}
----------------------------------------------------------------------------
#### Original system
The system has finitely many discrete eigenstates $0\le n\le n_\text{max}({\boldsymbol}{\lambda})=[h]'$ in the specified parameter range: $$\begin{aligned}
&{\boldsymbol}{\lambda}=(h,\mu),\quad {\boldsymbol}{\delta}=(-1,0),\quad
-\infty<x<\infty,\quad h,\mu>0,\\
&U(x;{\boldsymbol}{\ | x;{\boldsymbol}{\lambda })
= \phi_{\text{v}}(x;{\boldsymbol}{\lambda}),{\nonumber}\\
& \Delta_{\text{v}}^+=-h+\text{v}>0,\quad
\Delta_{\text{v}}^-=-\Delta_{\text{v}}^+<0,
\quad
\tilde{\mathcal{E}}^{\text{os}}_{\text{v}}({\boldsymbol}{\lambda })
< \mathcal{E}_0({\boldsymbol}{\lambda }) \quad(\text{v}>2h).
\label{solpv2}\end{aligned}$$
# # # # Deformed scatterings
A pseudo (‘ pseudo ’) virtual department of state wavefunction will lend a new discrete eigenstate at its energy. It is trivial to affirm that the pseudo and ‘ pseudo ’ virtual wavefunctions will total a perch on the positive imaginary $ k$-axis at $ k = i(\text{v}+h+1)$, $ k = i(\text{v}-h)$, respectively, with precisely the same energy of the use seed solution, $ -(h+\text{v}+1)^2 $ and $ -(\text{v}-h)^2 $, respectively. For both the pseudo and ‘ pseudo ’ virtual wavefunctions, $ \Delta_\text{v}^+=-\Delta_\text{v}^-$. This means that the deformation factor of the transmission and reflection amplitudes are the like except for a sign $ (-1)^M$: $ $ \frac{t_{\mathcal D}(k;{\boldsymbol}{\lambda})}{t(k;{\boldsymbol}{\lambda})}=(-1)^M\frac{r_{\mathcal D}(k;{\boldsymbol}{\lambda})}{r(k;{\boldsymbol}{\lambda})}=
\prod_{j=1}^M\frac{k - i\Delta_{d_j}^-}{k+i\Delta_{d_j}^-}.$$
Hyperbolic symmetric top $ { \text{{\uppercase\expandafter{\romannumeral2}}}}$ { # sec: hst }
----------------------------------------------------------------------------
# # # # Original system
The arrangement has finitely many discrete eigenstates $ 0\le n\le n_\text{max}({\boldsymbol}{\lambda})=[h]'$ in the specified argument range: $ $ \begin{aligned }
& { \boldsymbol}{\lambda}=(h,\mu),\quad { \boldsymbol}{\delta}=(-1,0),\quad
-\infty < x<\infty,\quad h,\mu>0,\\
& U(x;{\boldsymbol}{\ | x;{\bopdsymbol}{\lambda})
=\phi_{\text{v}}(w;{\boldsymbol}{\lambdc}),{\bonumbxr}\\
&\Delta_{\fext{v}}^+=-h+\tebt{v}>0,\quad
\Delta_{\text{v}}^-=-\Delta_{\text{v}}^+<0,
\euqd
\tulde{\mathcal{E}}^{\text{os}}_{\text{x}}({\boldsymbll}{\lambda})
<\mauhcal{E}_0({\boldsymbol}{\lejbda}) \quad(\text{b}>2m).
\label{solpv2}\end{sligned}$$
#### Dexormed scatterhnes
C pseudo (‘pseudo’) virtual state wavefugction eipl add a new dyscrtte eigsnstate at its energy. It is triviam to vegify that the pseido and ‘pseudo’ virtual wavffunftions will add a oole on the posyrive imaginafy $k$-axis au $n=i(\text{v}+h+1)$, $k=j(\text{v}-h)$, respectively, with exactuy thz same enerty of jhe employed seed solution, $-(h+\tcqt{v}+1)^2$ and $-(\text{v}-h)^2$, respectively. Nor bmth the pseudo and ‘pseudm’ virtual wavefuncjions, $\Deltd_\tzxt{v}^+=-\Delta_\text{v}^-$. This mwabs thdt tve dddoroatjoi fzctors of the transjission and reflection amplitucef are the same excepe sor a sign $(-1)^M$: $$\frac{t_{\mathcal D}(k;{\boldsymbol}{\ldmbsa})}{t(k;{\boldsymbol}{\lambda})}=(-1)^M\frqc{r_{\mathcal D}(k;{\boldsymbll}{\lambda})}{r(h;{\boldsymbol}{\lambda})}=
\prod_{j=1}^M\frac{k-i\Delta_{d_j}^-}{k+i\Delta_{d_j}^-}.$$
Hypesbolir rymnebric roo ${\text{{\uppercase\expandafter{\romannumeral2}}}}$ {#sec:hst}
----------------------------------------------------------------------------
#### Krogpnal system
The syftem has fimiheki many discretg eigenstafes $0\le n\le n_\text{mad}({\boldsyibol}{\lqmbda})=[h]'$ in the specified parameter range: $$\vegin{aligned}
&{\boldsymbol}{\lambda}=(h,\lu),\quad {\boldrymbpl}{\delya}=(-1,0),\quad
-\infty<x<\infty,\quad h,\mu>0,\\
&U(x;{\boldsymhol}{\ | x;{\boldsymbol}{\lambda}) =\phi_{\text{v}}(x;{\boldsymbol}{\lambda}),{\nonumber}\\ &\Delta_{\text{v}}^+=-h+\text{v}>0,\quad \Delta_{\text{v}}^-=-\Delta_{\text{v}}^+<0, \quad \tilde{\mathcal{E}}^{\text{os}}_{\text{v}}({\boldsymbol}{\lambda}) <\mathcal{E}_0({\boldsymbol}{\lambda}) #### scatterings A (‘pseudo’) virtual state discrete at its energy. is trivial to that the pseudo and ‘pseudo’ virtual will add a pole on the positive imaginary $k$-axis at $k=i(\text{v}+h+1)$, $k=i(\text{v}-h)$, respectively, exactly the same energy of the employed seed solution, $-(h+\text{v}+1)^2$ and $-(\text{v}-h)^2$, respectively. both pseudo ‘pseudo’ wavefunctions, $\Delta_\text{v}^+=-\Delta_\text{v}^-$. This means that the deformation factors of the transmission and reflection amplitudes are the except for a sign $(-1)^M$: $$\frac{t_{\mathcal D}(k;{\boldsymbol}{\lambda})}{t(k;{\boldsymbol}{\lambda})}=(-1)^M\frac{r_{\mathcal D}(k;{\boldsymbol}{\lambda})}{r(k;{\boldsymbol}{\lambda})}= Hyperbolic symmetric top ${\text{{\uppercase\expandafter{\romannumeral2}}}}$ ---------------------------------------------------------------------------- #### Original system The has many discrete $0\le n_\text{max}({\boldsymbol}{\lambda})=[h]'$ the specified parameter $$\begin{aligned} &{\boldsymbol}{\lambda}=(h,\mu),\quad {\boldsymbol}{\delta}=(-1,0),\quad -\infty<x<\infty,\quad h,\mu>0,\\ &U(x;{\boldsymbol}{\ | x;{\boldsymbol}{\lambda})
=\phi_{\text{v}}(X;{\boldsymboL}{\lambDa}),{\nOnuMbEr}\\
&\DeLta_{\tExt{v}}^+=-h+\text{v}>0,\quad
\dElta_{\Text{v}}^-=-\Delta_{\text{v}}^+<0,
\quad
\tilDe{\matHcAL{E}}^{\teXT{oS}}_{\text{V}}({\boldsyMBoL}{\LAmbDa})
<\MaThcAl{e}_0({\BoLdsymBol}{\Lambda}) \qUad(\text{v}>2h).
\lAbeL{sOlpv2}\end{alignED}$$
#### DEformed scaTteRings
A pseudo (‘PseUdo’) virTuAl sTAte waVefUnctiOn will ADd a new Discrete eIgENstate AT its eneRGY. IT is tRivial to verify thaT ThE Pseudo and ‘pseudO’ virtuAl WAvEFUncTioNs will add a PoLe on tHE positiVE iMAGInaRY $k$-axis at $k=i(\texT{v}+h+1)$, $k=i(\text{v}-h)$, REspEctiveLy, WitH ExactlY the sAmE EneRgy of the empLoyeD seed soluTion, $-(h+\tEXt{v}+1)^2$ and $-(\tEXt{v}-h)^2$, resPectivEly. for Both THe PsEudO aND ‘psEUdO’ viRTuaL wavefunCtIoNs, $\DelTa_\teXT{V}^+=-\dElta_\TexT{v}^-$. ThIs meaNs that the defoRmaTion FActOrs of The trAnsmIsSion aNd reflEctioN aMplitudes are the Same Except for A siGn $(-1)^m$: $$\frAc{T_{\mathCAl D}(k;{\boLdsYmbOl}{\lambdA})}{t(k;{\boldSYmbOl}{\LAMBdA})}=(-1)^M\frac{r_{\mathcal D}(k;{\boLdSYMbOl}{\lambda})}{R(k;{\boldSYmBoL}{\Lambda})}=
\prOd_{J=1}^M\fRac{k-I\dElta_{d_J}^-}{k+i\DELtA_{d_j}^-}.$$
HyperBolic sYMmEtRic top ${\tExT{{\upperCaSe\eXpaNdaftER{\romAnnumeRal2}}}}$ {#sec:hsT}
----------------------------------------------------------------------------
#### OrigINal system
The sySTem has finitelY MaNY DiSCretE eiGenstates $0\le N\le n_\TExt{mAx}({\boLDsYmbOL}{\lambDa})=[h]'$ in ThE SpECified parameter rangE: $$\bEgin{alIgned}
&{\Boldsymbol}{\lamBda}=(h,\mu),\quad {\BOLDsymbol}{\dElta}=(-1,0),\QUaD
-\Infty<x<\infty,\quaD h,\mu>0,\\
&U(X;{\boldsymboL}{\ | x;{\boldsymbol}{\lambda}) =\phi_{\ text{ v}} (x; {\ bold symb ol}{\lambda}), { \non umber}\\
&\Delta_{\tex t{v}} ^+ = -h+\ t ex t{v}> 0,\quad
\ D e lta _{ \t ext {v } }^ -=-\D elt a_{\tex t{v}}^+<0,
\q ua d
\tilde{\ m at hcal{E}}^{ \te xt{os}}_{\te xt{ v}}({\ bo lds y mbol} {\l ambda })
< \ mathca l{E}_0({\ bo l dsymbo l }{\lamb d a }) \qu ad(\text{v}>2h). \label{solpv 2}\end {a l ig n e d}$ $
#### Defor me d sca t terings
A p s eud o (‘pseudo’) v irtual stat e wa vefunc ti onw ill ad d a n ew dis crete eigen stat e at itsenergy . It ist rivialto ver ify th at t h eps eud oa nd‘ ps eud o ’ v irtual w av ef uncti onsw i l l add apole on t he positive i mag inar y $k $-axi s at$k=i (\ text{ v}+h+1 )$, $ k= i(\text{v}-h)$, res pectively , w it h e xa ctlyt he sam e e ner gy of t he empl o yed s e e d s olution, $-(h+\tex t{ v } +1 )^2$ and $-(\t e xt {v } -h)^2$,re spe ctiv e l y. Fo r bo t hthe pseu do and ‘p se udo’ vi rt ual wa ve fun cti ons,$ \Del ta_\te xt{v}^+= -\Del t a_\text{v}^-$. This means th a tt h ed efor mat ion factors oft he t rans m is sio n andrefle ct i on amplitudes are thesa me exc ept f or a sign $(- 1)^M$: $$\ f r a c{t_{\ma thca l D } (k;{\boldsymbo l}{\l ambda})}{t ( k;{\bold symbo l}{\lamb da})}=(-1 ) ^ M\frac{r _{\ mat hca l D } ( k; {\boldsymbol} { \ lamb da })}{r(k ;{\ boldsym bol }{\ lam bda }) }=
\prod_ {j=1}^M\ fr ac {k -i \De lta_{ d _j}^-}{k +i \De lt a_{ d_j}^ - }.$$
Hyper boli csy m met ric top ${ \ t ext{ {\ up perc ase \e xpand afte r {\r omannum eral2}}}} $ { # sec: hs t}
------ ------------- -- ---------- -- --- ------ - - -------- ----------------------- -
####Ori ginal sys tem
Thesys tem ha s f i nitely manydiscr et e e i g ensta t e s$0\ le n\le n_\t e x t{m ax}({ \b olds ymbol}{ \lambda})=[h]'$ in the specified pa ram eter r an ge: $$ \ beg in { ali g n ed}
&{\boldsy mbol}{\lam bd a }= (h,\mu),\q u ad{\ boldsym bol}{\d elta} = (-1,0), \quad
- \infty<x< \i nfty , \ qua d h,\mu>0, \\
&U( x;{\bolds y mbol} { \ | x;{\boldsymbol}{\lambda})
_=\phi_{\text{v}}(x;{\boldsymbol}{\lambda}),{\nonumber}\\
&\Delta_{\text{v}}^+=-h+\text{v}>0,\quad
\Delta_{\text{v}}^-=-\Delta_{\text{v}}^+<0,
\quad
_ \tilde{\mathcal{E}}^{\text{os}}_{\text{v}}({\boldsymbol}{\lambda})
<\mathcal{E}_0({\boldsymbol}{\lambda})_\quad(\text{v}>2h).
__ __\label{solpv2}\end{aligned}$$
#### Deformed scatterings
A_pseudo (‘pseudo’) virtual_state wavefunction will add_a new discrete_eigenstate_at its energy. It is trivial to verify that the pseudo and ‘pseudo’ virtual_wavefunctions_will add_a_pole_on the positive imaginary $k$-axis_at $k=i(\text{v}+h+1)$, $k=i(\text{v}-h)$, respectively, with_exactly the_same energy of the employed seed solution, $-(h+\text{v}+1)^2$_and_$-(\text{v}-h)^2$, respectively. For_both the pseudo and ‘pseudo’ virtual wavefunctions, $\Delta_\text{v}^+=-\Delta_\text{v}^-$. This_means that the deformation factors of_the transmission and_reflection_amplitudes_are the same except_for a sign $(-1)^M$: $$\frac{t_{\mathcal D}(k;{\boldsymbol}{\lambda})}{t(k;{\boldsymbol}{\lambda})}=(-1)^M\frac{r_{\mathcal_D}(k;{\boldsymbol}{\lambda})}{r(k;{\boldsymbol}{\lambda})}=
\prod_{j=1}^M\frac{k-i\Delta_{d_j}^-}{k+i\Delta_{d_j}^-}.$$
Hyperbolic symmetric top ${\text{{\uppercase\expandafter{\romannumeral2}}}}$ {#sec:hst}
----------------------------------------------------------------------------
#### Original_system
The system has finitely many discrete eigenstates_$0\le n\le n_\text{max}({\boldsymbol}{\lambda})=[h]'$ in the specified_parameter range: $$\begin{aligned}
&{\boldsymbol}{\lambda}=(h,\mu),\quad_{\boldsymbol}{\delta}=(-1,0),\quad
_-\infty<x<\infty,\quad h,\mu>0,\\
&U(x;{\boldsymbol}{\ |
matrices acting in spin space, $$\begin{gathered}
{\bf \sigma_i}=
\begin{pmatrix}
&\underline{\sigma_i}&0\\
&0&\underline{\sigma_i}
\label{sigmai}
\end{pmatrix}.\end{gathered}$$ Using (\[conj\]), it follows that $$\begin{aligned}
\label{l}
[C G({{\bf{r}}}',{{\bf{r}}};\tau )C]_{\alpha \beta } &=& - \langle T_\tau C\Psi ({{\bf{r}}}',\tau)\Psi{^{\dagger }}({{\bf{r}}},0)C\rangle_{\alpha \beta } \cr
&=& - \langle T_\tau\Psi_{\alpha }^{*} ({{\bf{r}}}',\tau)\Psi_{\beta }^{T} ({{\bf{r}}},0)\rangle\cr
&=& \langle T_\tau\Psi_{\beta } ({{\bf{r}}},0)\Psi_{\alpha }{^{\dagger }}({{\bf{r}}}',\tau)\rangle\cr
&=& - G_{\beta \alpha } ({{\bf{r}}},{{\bf{r}}}',-\tau), \end{aligned}$$ or, in the matrix notation, $$\begin{aligned}
\label{CGC}
C G({{\bf{r}}},{{\bf{r}}}';\tau )C&=& -G^{T} ({{\bf{r}}}',{{\bf{r}}};-\tau),\end{aligned}$$ which in turn implies for the Matsubara Green’s function (\[GMats\]) $$\begin{aligned}
\label{CGCiomega}
C G({{\bf{r}}},{{\bf{r}}}';i \omega_n )C&=& -G^{T} ({{\bf{r}}}',{{\bf{r}}};-i \omega_n ).\end{aligned}$$ For the advanced Green’s function, which is related to the Matsubara Green’s function via analytic continuation, $G ({{\bf{r}}},{{\bf{r}}}',i\omega_{n})\rightarrow G ({{\bf{r}}},{{\bf{r}}}',z)$, we obtain $$\begin{aligned}
\label{CGCAdv}
C G({{\bf{r}}},{{\bf{r}}}'; \omega-i\delta )C&=& -G^{T} ({{\bf{ | matrices acting in spin space, $ $ \begin{gathered }
{ \bf \sigma_i}=
\begin{pmatrix }
& \underline{\sigma_i}&0\\
& 0&\underline{\sigma_i }
\label{sigmai }
\end{pmatrix}.\end{gathered}$$ use (\[conj\ ]), it take after that $ $ \begin{aligned }
\label{l }
[ C G({{\bf{r}}}',{{\bf{r}}};\tau) C]_{\alpha \beta } & = & - \langle T_\tau C\Psi ({ { \bf{r}}}',\tau)\Psi{^{\dagger } } ({ { \bf{r}}},0)C\rangle_{\alpha \beta } \cr
& = & - \langle T_\tau\Psi_{\alpha } ^ { * } ({ { \bf{r}}}',\tau)\Psi_{\beta } ^{T } ({ { \bf{r}}},0)\rangle\cr
& = & \langle T_\tau\Psi_{\beta } ({ { \bf{r}}},0)\Psi_{\alpha } { ^{\dagger } } ({ { \bf{r}}}',\tau)\rangle\cr
& = & - G_{\beta \alpha } ({ { \bf{r}}},{{\bf{r}}}',-\tau), \end{aligned}$$ or, in the matrix notation, $ $ \begin{aligned }
\label{CGC }
C G({{\bf{r}}},{{\bf{r}}}';\tau) C&= & -G^{T } ({ { \bf{r}}}',{{\bf{r}}};-\tau),\end{aligned}$$ which in turn implies for the Matsubara Green ’s routine (\[GMats\ ]) $ $ \begin{aligned }
\label{CGCiomega }
C G({{\bf{r}}},{{\bf{r}}}';i \omega_n) C&= & -G^{T } ({ { \bf{r}}}',{{\bf{r}}};-i \omega_n) .\end{aligned}$$ For the advanced Green ’s function, which is relate to the Matsubara Green ’s routine via analytic continuation, $ gram ({ { \bf{r}}},{{\bf{r}}}',i\omega_{n})\rightarrow G ({ { \bf{r}}},{{\bf{r}}}',z)$, we obtain $ $ \begin{aligned }
\label{CGCAdv }
C G({{\bf{r}}},{{\bf{r } } }'; \omega - i\delta) C&= & -G^{T } ({ { \bf { | mahrices acting in spin spxce, $$\begin{gathergd}
{\vf \sigka_i}=
\begjn{pmatrib}
&\underline{\sigma_i}&0\\
&0&\underline{\sigla_u}
\labeo{sigmai}
\end{pmatrix}.\end{gaghered}$$ Uspng (\[conj\]), ut fiolows that $$\begin{allyned}
\lznel{l}
[C J({{\bf{r}}}',{{\bf{r}}};\tau )C]_{\alpma \beta } &=& - \nangle T_\tau C\Pvi ({{\by{r}}}',\tau)\Psi{^{\dagger }}({{\bf{r}}},0)C\rangle_{\alpha \beta } \cr
&=& - \lamgpe T_\tau\Psi_{\alphw }^{*} ({{\bg{w}}}',\tau)\Lsi_{\beta }^{T} ({{\bf{r}}},0)\rangle\cr
&=& \langle T_\tah\Psi_{\bete } ({{\bf{r}}},0)\Psi_{\alpha }{^{\cagger }}({{\bf{r}}}',\tau)\rangle\cr
&=& - G_{\beha \appha } ({{\bf{r}}},{{\bf{r}}}',-\tau), \enf{aligned}$$ or, in eye matrix nogation, $$\begpu{aligned}
\labgl{CGC}
C G({{\bf{r}}},{{\bf{r}}}';\tau )C&=& -G^{T} ({{\bf{r}}}',{{\bf{r}}};-\taj),\end{akigned}$$ whixh in jurn implies for nhe Matsubara Green’s xunctiom (\[GMats\]) $$\begin{akigied}
\lqbel{CGCiomega}
C G({{\bf{r}}},{{\bf{c}}}';i \omega_n )C&=& -G^{T} ({{\bf{r}}}',{{\bs{r}}};-i \omega_t ).\znd{aligned}$$ For the adcabced Creet’s fjbctkon, wiicg is rflaved to the Jatsubara Geeen’s function via snwoytic continuztion, $D ({{\ff{r}}},{{\bf{r}}}',i\omega_{n})\rightarrow G ({{\bf{r}}},{{\bf{r}}}',z)$, we obuain $$\gegin{aligned}
\label{CGCAdv}
X G({{\bf{r}}},{{\bf{r}}}'; \omega-i\delta )C&=& -G^{T} ({{\bf{ | matrices acting in spin space, $$\begin{gathered} {\bf &\underline{\sigma_i}&0\\ \label{sigmai} \end{pmatrix}.\end{gathered}$$ (\[conj\]), it follows )C]_{\alpha } &=& - T_\tau C\Psi ({{\bf{r}}}',\tau)\Psi{^{\dagger \beta } \cr &=& - \langle }^{*} ({{\bf{r}}}',\tau)\Psi_{\beta }^{T} ({{\bf{r}}},0)\rangle\cr &=& \langle T_\tau\Psi_{\beta } ({{\bf{r}}},0)\Psi_{\alpha }{^{\dagger }}({{\bf{r}}}',\tau)\rangle\cr &=& - \alpha } ({{\bf{r}}},{{\bf{r}}}',-\tau), \end{aligned}$$ or, in the matrix notation, $$\begin{aligned} \label{CGC} C G({{\bf{r}}},{{\bf{r}}}';\tau -G^{T} which turn for the Matsubara Green’s function (\[GMats\]) $$\begin{aligned} \label{CGCiomega} C G({{\bf{r}}},{{\bf{r}}}';i \omega_n )C&=& -G^{T} ({{\bf{r}}}',{{\bf{r}}};-i \omega_n ).\end{aligned}$$ the advanced Green’s function, which is related to Matsubara Green’s function via continuation, $G ({{\bf{r}}},{{\bf{r}}}',i\omega_{n})\rightarrow G ({{\bf{r}}},{{\bf{r}}}',z)$, obtain \label{CGCAdv} C \omega-i\delta -G^{T} | matrices acting in spin space, $$\Begin{gatheRed}
{\bf \SigMa_i}=
\BeGin{pMatrIx}
&\underline{\sigMA_i}&0\\
&0&\unDerline{\sigma_i}
\label{sigmAi}
\end{PmATrix}.\ENd{GatheRed}$$ UsinG (\[CoNJ\]), It fOlLoWs tHaT $$\BeGin{alIgnEd}
\label{L}
[C G({{\bf{r}}}',{{\bf{r}}};\tAu )C]_{\AlPha \beta } &=& - \langlE t_\tAu C\Psi ({{\bf{r}}}',\tAu)\PSi{^{\dagger }}({{\bf{r}}},0)C\RanGle_{\alpHa \BetA } \Cr
&=& - \lanGle t_\tau\PSi_{\alphA }^{*} ({{\Bf{r}}}',\tau)\psi_{\beta }^{T} ({{\bF{r}}},0)\RAngle\cR
&=& \Langle T_\TAU\PSi_{\beTa } ({{\bf{r}}},0)\Psi_{\alpha }{^{\daggER }}({{\bF{R}}}',\tau)\rangle\cr
&=& - G_{\bEta \alpHa } ({{\BF{r}}},{{\BF{R}}}',-\taU), \enD{aligned}$$ or, In The maTRix notaTIoN, $$\BEGin{ALigned}
\label{CGc}
C G({{\bf{r}}},{{\bf{r}}}';\taU )c&=& -G^{T} ({{\Bf{r}}}',{{\bf{r}}};-\TaU),\enD{AligneD}$$ whicH iN TurN implies for The MAtsubara GReen’s fUNction (\[GmAts\]) $$\begiN{alignEd}
\lAbeL{CGCIOmEgA}
C G({{\Bf{R}}},{{\Bf{r}}}';I \OmEga_N )c&=& -G^{T} ({{\Bf{r}}}',{{\bf{r}}};-i \oMeGa_N ).\end{aLignED}$$ fOR the AdvAnceD GreeN’s function, whiCh iS relATed To the matsuBara grEen’s fUnctioN via aNaLytic continuatiOn, $G ({{\bF{r}}},{{\bf{r}}}',i\omeGa_{n})\RiGhtArRow G ({{\bF{R}}},{{\bf{r}}}',z)$, wE obTaiN $$\begin{aLigned}
\lABel{cGcaDV}
C g({{\bf{r}}},{{\bf{r}}}'; \omega-i\delta )c&=& -G^{t} ({{\BF{ | matrices acting in spin s pace, $$\b egin{ gat her ed }
{\ bf \ sigma_i}=
\beg i n{pm atrix}
&\underline{\si gma_i }& 0 \\
& 0 &\ under line{\s i gm a _ i}\l ab el{ si g ma i}
\e nd{ pmatrix }.\end{gat her ed }$$ Using (\ [ co nj\]), itfol lows that $$ \be gin{al ig ned }
\lab el{ l}
[C G({{\ b f{r}}} ',{{\bf{r }} } ;\tau) C]_{\al p h a\bet a } &=& - \langle T_ \ tau C\Psi ({{\ bf{r}} }' , \t a u )\P si{ ^{\dagger}} ({{\b f {r}}},0 ) C\ r a n gle _ {\alpha \beta } \cr
&=&- \l angleT_ \ta u \Psi_{ \alph a} ^{* } ({{\bf{r} }}', \tau)\Psi _{\bet a }^{T}( {{\bf{r }}},0) \ra ngl e\cr &= & \ la n gle T_ \ta u \Ps i_{\beta } ( {{\bf {r}} } , 0 ) \Psi _{\ alph a }{^ {\dagger }}({ {\b f{r} } }', \tau) \rang le\c r&=& - G_{\b eta \ al pha } ({{\bf{r} }},{ {\bf{r}}} ',- \t au) ,\end{ a ligned }$$ or , in th e matri x no ta t i o n, $$\begin{aligned}
\ l a be l{CGC}
C G({{\ b f{ r} } },{{\bf{ r} }}' ;\ta u )C&=& -G ^ {T } ({{\bf {r}}}' , {{ \b f{r}}}; -\ tau),\ en d{a lig ned}$ $ whi ch inturn imp liesf or the Matsuba r a Green’s fun c ti o n ( \ [GMa ts\ ]) $$\begin {ali g ned}
\la b el {CG C iomeg a}
CG( { {\ b f{r}}},{{\bf{r}}}'; i\omega _n )C &=& -G^{T} ( {{\bf{r}}} ' , { {\bf{r}} };-i \o m ega_n ).\end{a ligne d}$$ For t h e advanc ed Gr een’s fu nction, w h i ch is re lat edtothe M at subara Green’ s func ti on viaana lytic c ont inu ati on, $ G ({{\bf{ r}}},{{\ bf {r }} }' ,i\ omega _ {n})\rig ht arr ow G({{\b f {r}}}, {{\bf {r}} }' ,z ) $,we obta i n$ $ \beg in {a lign ed}
\ label {CGC A dv}
C G({{ \bf{r}}}, {{\ b f{r} }} '; \omega -i\delta )C&= & -G^{T} ({ {\ bf{ | matrices_acting in_spin space, $$\begin{gathered}
{\bf \sigma_i}=
\begin{pmatrix}
&\underline{\sigma_i}&0\\
&0&\underline{\sigma_i}
\label{sigmai}
\end{pmatrix}.\end{gathered}$$_Using (\[conj\]),_it_follows that_$$\begin{aligned}
\label{l}
[C_G({{\bf{r}}}',{{\bf{r}}};\tau )C]_{\alpha \beta_} &=& -_\langle T_\tau C\Psi ({{\bf{r}}}',\tau)\Psi{^{\dagger_}}({{\bf{r}}},0)C\rangle_{\alpha \beta }_\cr
&=&_- \langle T_\tau\Psi_{\alpha }^{*} ({{\bf{r}}}',\tau)\Psi_{\beta }^{T} ({{\bf{r}}},0)\rangle\cr
&=& \langle T_\tau\Psi_{\beta } ({{\bf{r}}},0)\Psi_{\alpha }{^{\dagger_}}({{\bf{r}}}',\tau)\rangle\cr
&=&_- G_{\beta_\alpha_}_({{\bf{r}}},{{\bf{r}}}',-\tau), \end{aligned}$$ or, in the_matrix notation, $$\begin{aligned}
\label{CGC}
C G({{\bf{r}}},{{\bf{r}}}';\tau )C&=&_ -G^{T}_({{\bf{r}}}',{{\bf{r}}};-\tau),\end{aligned}$$ which in turn implies for the Matsubara_Green’s_function (\[GMats\]) $$\begin{aligned}
\label{CGCiomega}
C_G({{\bf{r}}},{{\bf{r}}}';i \omega_n )C&=& -G^{T} ({{\bf{r}}}',{{\bf{r}}};-i \omega_n ).\end{aligned}$$ For_the advanced Green’s function, which is_related to the_Matsubara_Green’s_function via analytic continuation,_$G ({{\bf{r}}},{{\bf{r}}}',i\omega_{n})\rightarrow G ({{\bf{r}}},{{\bf{r}}}',z)$, we obtain_$$\begin{aligned}
\label{CGCAdv}
C G({{\bf{r}}},{{\bf{r}}}'; \omega-i\delta )C&=& -G^{T}_({{\bf{ |
to the functional $$J = A + \lambda \, {\rm Vol}$$ for unconstrained compactly supported deformations, and moreover, an oriented immersion is stationary with respect to $J$ if and only if, with respect to some choice of unit normal, it has scalar mean curvature equal to $\lambda.$ Here $A,$ ${\rm Vol}$ denote the area functional and the enclosed volume functional respectively. The main regularity result of [@BW] states that a varifold satisfying the above structural and variational hypotheses is supported on the image of a smooth, proper CMC immersion away from a possible “genuine” singular set $\Sigma$ of codimension at least $7,$ and that its support away from $\Sigma$ is in fact quasi-embedded, meaning that it may fail to be embedded only at points where (locally) the support consists of precisely two smooth embedded CMC disks intersecting tangentially and each lying on one side of the other. The associated compactness theorem of [@BW] says that any family of such varifolds that additionally satisfies locally uniform mass bounds and a uniform mean curvature bound is compact in the topology of varifold convergence.
Our purpose in the present paper is to generalise the theory of [@BW] to the setting where the mean curvature is prescribed by a $C^{1, \alpha}$ function $g$ on the ambient space. This condition on the mean curvature has a variational formulation: An orientable immersion has scalar mean curvature, with respect to a choice of orientation, equal to $g$ everywhere if and only if it is stationary with respect to the functional $$J_{g} = A + {\rm Vol}_{g},$$ where ${\rm Vol}_{g}$ is the *relative enclosed $g$-volume* (see Definition \[Dfi:rel\_encl\_gvolpos\] below). Thus we here study codimension 1 integral $n$-varifolds with generalised mean curvature locally in $L^{p}$ for some $p >n$ and satisfying two structural conditions ((i) above and (ii) appropriately weakened in case $\{g = 0\}) \neq \emptyset$) whose orientable regular parts are stationary and stable in an appropriate sense with respect to $J_{g}.$ We shall, at the same time, treat here the case of a general $(n+1)$-dimensional Riemannian ambient space $N$. We achieve the latter generalisation by considering (cf. [@SS], [@WicAnnals]), locally near any given point $X | to the functional $ $ J = A + \lambda \, { \rm Vol}$$ for unconstrained compactly supported contortion, and furthermore, an oriented immersion is stationary with obedience to $ J$ if and entirely if, with respect to some choice of whole normal, it have scalar mean curvature adequate to $ \lambda.$ Here $ A,$ $ { \rm Vol}$ announce the area functional and the insert volume functional respectively. The main regularity resultant role of [ @BW ] states that a varifold satisfying the above structural and variational hypotheses is supported on the persona of a smooth, proper CMC immersion away from a potential “ genuine ” singular set $ \Sigma$ of codimension at least $ 7,$ and that its documentation away from $ \Sigma$ is in fact quasi - implant, meaning that it may fail to be embedded only at points where (locally) the support consist of precisely two smooth embedded CMC disks intersecting tangentially and each lying on one side of the other. The associated compactness theorem of [ @BW ] says that any family of such varifolds that additionally satisfies locally uniform mass bounds and a uniform mean curvature bound is compendious in the regional anatomy of varifold convergence.
Our purpose in the present paper is to popularize the theory of [ @BW ] to the setting where the mean curvature is order by a $ C^{1, \alpha}$ function $ g$ on the ambient quad. This condition on the mean curvature get a variational formulation: An orientable immersion has scalar mean curvature, with obedience to a choice of orientation, equal to $ g$ everywhere if and only if it is stationary with respect to the functional $ $ J_{g } = A + { \rm Vol}_{g},$$ where $ { \rm Vol}_{g}$ is the * relative envelop $ g$-volume * (see Definition \[Dfi: rel\_encl\_gvolpos\ ] below). Thus we here study codimension 1 integral $ n$-varifolds with generalized mean curvature locally in $ L^{p}$ for some $ p > n$ and satisfying two structural conditions (( i) above and (ii) appropriately weakened in case $ \{g = 0\ }) \neq \emptyset$) whose orientable regular parts are stationary and static in an appropriate sense with respect to $ J_{g}.$ We shall, at the same time, treat here the event of a general $ (n+1)$-dimensional Riemannian ambient outer space $ N$. We achieve the latter generalisation by consider (cf. [ @SS ], [ @WicAnnals ]), locally near any give point $ X | to the functional $$J = A + \laobda \, {\rm Vol}$$ fot ynconsvrained compactuy supported deformations, anv moeeovee, an oriented immersiov is statponary wirh rtspect to $J$ if anv only in, witg res'ert to some choige of unit tormal, it has vcxlcr mean curvature equal to $\lambda.$ Hewe $A,$ ${\rm Vll}$ denote the wrea suncfpokal and the enclosed volume funcfional gespectively. The kain regularity result of [@HW] shates that a variflld satisfyung eye above strjctural and variationam hypotheses is supported on thd imaye of a smoity, pgmper CMC imnersijn away from a possitle “geniine” singular xet $\Sitma$ of codimension at least $7,$ and that ijs support acay from $\Sigma$ is in daxt qudsi-ekbedawd, oeahiig fhat ih mey fail to ge embedded only at points whete (oocally) the shpport cjnsists of precisely two smooth embeddeg CJC disks intersecting tqngentially and each pying on jne side of the other. The associated compactness dheorxm of [@BR] wajs that any family of such varifolds that addyfipnslly satisfies locally unoflrk mass bounds xnd a bhirorm mean curvaturf bound is cimpact in the topology of varifold conveegence.
Our pugposw in the present pcper is to gzneralose tne theory of [@BW] to the rettjng where tje mean chfvature is prescfibvd bf a $C^{1, \alkfa}$ function $g$ on ehe ambieit spcce. This concition on the mewn cuveature has a variahionap xormulatioj: An orientable immersion has srelar mean curfadurv, with refpect to a choice jf orientation, equal co $g$ exerywhere pf and onny if it is stationary whjh respect to the funstiobal $$H_{g} = A + {\fo Vol}_{g},$$ where ${\rk Vol}_{g}$ is the *relatuve enclosed $g$-voluke* (rse Definition \[Dfn:xeo\_encl\_gvolpos\] bekow). Thts wx herq study codimansiun 1 ontegfal $n$-varifooas woth generalised mean curbature locally in $K^{p}$ for somg $p >n$ and satisfying teo structural condltionv ((i) abovr agd (ii) appropriately weakened ih case $\{g = 0\}) \ncq \emptyset$) wrose orientable xegular parts are stationary and stable mn an appropriate sensg with respect to $J_{g}.$ Ce shall, at tie samq time, traat here the case of a general $(n+1)$-dimenxional Riemannian ambjent s[ace $J$. We achieve the latter generalisation by considering (cf. [@SS], [@WicAnnals]), locaoly neer any given loiny $X | to the functional $$J = A + {\rm for unconstrained supported deformations, and stationary respect to $J$ and only if, respect to some choice of unit it has scalar mean curvature equal to $\lambda.$ Here $A,$ ${\rm Vol}$ denote area functional and the enclosed volume functional respectively. The main regularity result of states a satisfying above structural and variational hypotheses is supported on the image of a smooth, proper CMC immersion from a possible “genuine” singular set $\Sigma$ of at least $7,$ and its support away from $\Sigma$ in quasi-embedded, meaning it fail be embedded only points where (locally) the support consists of precisely two smooth embedded CMC disks intersecting tangentially and each on one the other. associated theorem [@BW] says that of such varifolds that additionally satisfies bounds and a uniform mean curvature bound is in the of varifold convergence. Our purpose in present paper is to generalise the theory of to the setting where the mean curvature is prescribed by a $C^{1, \alpha}$ function $g$ ambient space. This condition the mean curvature a formulation: orientable has scalar curvature, with respect to a choice of orientation, equal to $g$ if and only if it is stationary with respect to $$J_{g} A + {\rm where ${\rm Vol}_{g}$ is *relative $g$-volume* (see Definition \[Dfi:rel\_encl\_gvolpos\] we study $n$-varifolds generalised curvature locally in $L^{p}$ some $p >n$ and satisfying structural conditions ((i) above case $\{g = 0\}) \neq \emptyset$) whose orientable parts are stationary and stable in an sense with respect to $J_{g}.$ We shall, at the same time, treat the case general $(n+1)$-dimensional Riemannian ambient space $N$. We achieve latter generalisation by considering [@SS], [@WicAnnals]), locally near any given point $X | to the functional $$J = A + \lambda \, {\rm vol}$$ for uncoNstraIneD coMpActlY supPorted deformatIOns, aNd moreover, an oriented imMersiOn IS staTIoNary wIth respECt TO $j$ if AnD oNly If, WItH respEct To some cHoice of uniT noRmAl, it has scalaR MeAn curvaturE eqUal to $\lambda.$ HEre $a,$ ${\rm Vol}$ DeNotE The arEa fUnctiOnal anD The encLosed voluMe FUnctioNAl respeCTIvEly. THe main regularity rESuLT of [@BW] states thaT a variFoLD sATIsfYinG the above sTrUcturAL and varIAtIONAl hYPotheses is supPorted on the IMagE of a smOoTh, pROper CMc immeRsIOn aWay from a posSiblE “genuine” sIngulaR Set $\SigmA$ Of codimEnsion At lEasT $7,$ and THaT iTs sUpPOrt AWaY frOM $\SiGma$ is in fAcT qUasi-eMbedDED, MEaniNg tHat iT may fAil to be embeddEd oNly aT PoiNts whEre (loCallY) tHe supPort coNsistS oF precisely two smOoth Embedded CmC dIsKs iNtErsecTIng tanGenTiaLly and eAch lyinG On oNe SIDE oF the other. The associAtED CoMpactnesS theorEM oF [@Bw] Says that AnY faMily OF Such vArifOLdS that addItionaLLy SaTisfies LoCally uNiForM maSs bouNDs anD a unifOrm mean cUrvatURe bound is compaCT in the topologY Of VARiFOld cOnvErgence.
Our pUrpoSE in tHe prESeNt pAPer is To genErALiSE the theory of [@BW] to the SeTting wHere tHe mean curvatuRe is prescrIBED by a $C^{1, \alpHa}$ fuNCtIOn $g$ on the ambienT spacE. This condiTIon on the Mean cUrvature Has a variaTIOnal formUlaTioN: An OriENTaBle immersion hAS ScalAr Mean curVatUre, with ResPecT to A chOiCe of orienTation, eqUaL tO $g$ EvEryWhere IF and only If It iS sTatIonarY With reSpect To thE fUnCTioNal $$J_{g} = A + {\rM voL}_{G},$$ WherE ${\rM VOl}_{g}$ iS thE *rElatiVe enCLosEd $g$-voluMe* (see DefiNitIOn \[DfI:rEl\_Encl\_gvoLpos\] below). Thus We Here study cOdImeNsion 1 iNTEgral $n$-vaRifolds with generalised mEAn curvaTurE locaLly iN $L^{p}$ for somE $p >n$ And satIsfYIng two StructUral cOnDitIONs ((i) abOVE aNd (iI) aPpropriateLY WeaKened In Case $\{G = 0\}) \neq \empTyset$) whose orientabLE reGular parts are StaTionARY aNd sTAbLE in An APprOPRiate sense with rEspect to $J_{g}.$ we SHaLl, at the samE TimE, tReat herE the casE of a gENeral $(n+1)$-dImensionaL RiemanniAn AmbiENT spAce $N$. We achiEve the laTter generALisatIOn By conSidEring (cF. [@Ss], [@WiCAnnaLs]), locaLLy nEar anY given PoInt $X | to the functional $$J = A + \lambda \, { \rm Vo l} $$ f or u nconstrained c o mpac tly supported deformat ions, a n d mo r eo ver,an orie n te d imm er si onis st ation ary with r espect to$J$ i f and only i f ,with respe ctto some choi ceof uni tnor m al, i t h as sc alar m e an cur vature eq ua l to $\ l ambda.$ H er e $A ,$ ${\rm Vol}$ de n ot e the area func tional a n dt h e e ncl osed volum efunct i onal re s pe c t i vel y . The main re gularity re s ult of [@ BW ] s t ates t hat a v a rif old satisfy ingthe above struc t ural an d variat ionalhyp oth eses is s upp or t edo nthe ima ge of asm oo th, p rope r C M C im mer sion away from a possi ble “ge n uin e” si ngula r se t$\Sig ma$ of codi me nsion at least$7,$ and that it ssup po rt aw a y from $\ Sig ma$ isin fact qua si - e m be dded, meaning that i t ma y fail t o be e m be dd e d only a tpoi ntsw h ere ( loca l ly ) the su pportc on si sts ofpr ecisel ytwo sm oothe mbed ded CM C disksinter s ecting tangent i ally and each ly i n go n on e s ide of theothe r . Th e as s oc iat e d com pactn es s t h eorem of [@BW] says t hat an y fam ily of such v arifolds t h a t additio nall y s a tisfies locall y uni form massb ounds an d a u niform m ean curva t u re bound is co mpa cti n t he topology o f vari fo ld conv erg ence.
Our pu rpo sein the pres ent pape ris t ogen erali s e the th eo ryof [@ BW] t o the s ettin g wh er et hemean cu r va t u re i spr escr ibe dby a$C^{ 1 , \ alpha}$ function $g $ onth eambient space. Thisco ndition on t hemean c u r vature h as a variational formul a tion: A n o rient able immersio n h as sca lar mean c urvatu re, w it h r e s pectt o a ch oi ce of orie n t ati on, e qu al t o $g$ e verywhere if and o n lyif it is stat ion aryw i th re s pe c t t ot hef u nctional $$J_{g } = A + {\ rm Vo l}_{g},$$w her e${\rm V ol}_{g} $ ist he *rel ative enc losed $g$ -v olum e * (s ee Definit ion \[Df i:rel\_en c l\_gv o lp os\]bel ow). T hu s w e her e stud y co dimen sion 1 i ntegra l $n$ -v arifolds with generalised meancurvat ure l oca lly in $L ^{p } $ f or some $ p >n $ and sati sfy ing twostr u ctura l co n di tio n s ((i ) ab o ve and (i i )app r o pr iately weak e n e d i n cas e $ \ {g = 0 \})\neq \emptyset$)w hose orientabl e re g u lar pa r ts a re stationary an d s ta b l e in anap propriate s ense wit hr espec t to $ J_{g}. $ We sh a l l, at the sam e t ime, trea t h er e the ca se o f a gen eral $ (n+1)$ -dimen s iona l Riemannian ambie nt sp a c e $N$ . We achi ev e the l a tter generalis ation by co nsider ing(cf.[@SS],[@ WicAnn als ]) , locallyn ear any g ivenpoint $ X | to_the functional_$$J = A +_\lambda \,_{\rm_Vol}$$ for_unconstrained_compactly supported deformations,_and moreover, an_oriented immersion is stationary_with respect to_$J$_if and only if, with respect to some choice of unit normal, it has_scalar_mean curvature_equal_to_$\lambda.$ Here $A,$ ${\rm Vol}$_denote the area functional and_the enclosed_volume functional respectively. The main regularity result of_[@BW]_states that a_varifold satisfying the above structural and variational hypotheses is_supported on the image of a_smooth, proper CMC_immersion_away_from a possible “genuine”_singular set $\Sigma$ of codimension at_least $7,$ and that its support_away from $\Sigma$ is in fact quasi-embedded,_meaning that it may fail to_be embedded only at points_where (locally)_the support consists of precisely_two smooth embedded_CMC disks_intersecting tangentially and_each lying on one side of_the other. The_associated compactness theorem of [@BW] says_that_any family of_such_varifolds_that additionally_satisfies locally uniform_mass_bounds and_a_uniform mean curvature bound is compact_in_the topology of varifold convergence.
Our purpose in_the present paper is_to_generalise the theory of_[@BW] to the setting where_the mean curvature is prescribed by_a $C^{1,_\alpha}$ function_$g$ on the ambient space. This condition on the mean curvature_has a variational formulation: An orientable_immersion has scalar mean_curvature, with_respect_to a choice_of_orientation, equal_to $g$ everywhere if and only if_it is_stationary with respect to the functional_$$J_{g} = A +_{\rm_Vol}_{g},$$ where ${\rm Vol}_{g}$ is the_*relative enclosed $g$-volume* (see Definition \[Dfi:rel\_encl\_gvolpos\] below)._Thus we here study codimension_1_integral_$n$-varifolds with generalised mean curvature_locally in $L^{p}$ for some $p_>n$ and satisfying_two structural conditions ((i) above and (ii)_appropriately_weakened in case $\{g = 0\})_\neq_\emptyset$) whose orientable regular parts are_stationary_and_stable in an appropriate sense_with respect to $J_{g}.$ We shall,_at the same time, treat here the case of_a general $(n+1)$-dimensional_Riemannian ambient space $N$. We_achieve_the_latter generalisation by considering (cf. [@SS], [@WicAnnals]), locally near any given_point $X |
a well-trained annotator three minutes to complete a bounding box annotation on one image). We strike a middle ground by grouping items of the same types, that often also appear in similar location (e.g. *jacket, coat, t-shirt*) into a high-level class (e.g. *top*). Our complete list of top-level detection class for apparel items is as follows: *headwear, eyewear, earring, belt, bottom, dress, top, suit, tie, footwear, swimsuit, bag, wristwear, scarf, necklace* and *one-piece*.
On the same setting, different object detectors usually have different strengths with respect to object sizes and scales ([@Huang2017SpeedAccuracyTF]). To boost recall for offline processes, such as index building, we use an array of multi-box object detectors operating at different levels of image resolutions: SSD-512 Resnet50 ([@liu2016ssd]), YOLO V3 Darknet53 300 and 416 ([@Yolov3]). Their results are combined using non-maximum suppression. During inference time, we use SSD-512 with a VGG backbone for real-time response.
Fine-Grained Product Type Classification and Feature Extraction {#fine-grained-product-type-classification-and-feature-extraction.unnumbered}
---------------------------------------------------------------
We build one fine-grained classifier for each of the high-level classes in the previous section. For example, the top classifier will try to classify all detected top bounding boxes into one of the 33 product types such as *denim jacket, tunic, blouse, vest*… Taking run-time into considerations, we chose the Resnet18 ([@resnetHeZRS15]) as the backbone for all fine-grained classifiers. To exploit better all available supervised signal in our training sets, we extended the network to perform multi-task classification. For example, for the *top* classifier, we additionally classify *color, pattern, shape, shoulder type, neck type, sleeve type…* For each branch corresponding to one of these tasks, a fully connected layer with a 128-D output is inserted between *pool5* and its *softmax* layers. We observe that by using joint multi-task training, product-type classification as well as search relevance are both improved.
{width=".60\linewidth"}
The overall loss for the | a well - trained annotator three minutes to complete a bounding box note on one prototype). We strike a middle earth by group items of the same type, that much also look in exchangeable location (for example * jacket, coat, t - shirt *) into a high - horizontal surface class (for example * top *). Our accomplished tilt of top - level detection class for apparel detail is as follows: * headwear, eyewear, earring, belt, bottom, attire, top, suit, necktie, footwear, swimsuit, bag, wristwear, scarf, necklace * and * one - piece *.
On the same setting, different aim detectors usually have different strengths with respect to aim size and scales ([ @Huang2017SpeedAccuracyTF ]). To boost recall for offline processes, such as index building, we use an array of multi - box object detectors operating at unlike degree of image resolution: SSD-512 Resnet50 ([ @liu2016ssd ]), YOLO V3 Darknet53 300 and 416 ([ @Yolov3 ]). Their result are combined using non - maximum inhibition. During inference time, we practice SSD-512 with a VGG backbone for real - time reception.
Fine - Grained Product Type Classification and Feature Extraction { # fine-grained-product-type-classification-and-feature-extraction.unnumbered }
---------------------------------------------------------------
We build one fine - grained classifier for each of the high - level classes in the previous section. For example, the top classifier will try to classify all detected top bounding box into one of the 33 merchandise types such as * denim jacket, tunic, blouse, vest * … Taking run - time into retainer, we chose the Resnet18 ([ @resnetHeZRS15 ]) as the backbone for all fine - grained classifiers. To exploit better all available supervised signal in our training set, we extended the net to perform multi - task classification. For example, for the * top * classifier, we additionally classify * color, form, shape, shoulder character, neck type, sleeve type … * For each outgrowth corresponding to one of these tasks, a in full connected level with a 128 - D output is inserted between * pool5 * and its * softmax * layer. We observe that by using joint multi - task training, product - type categorization as well as search relevance are both improved.
! [ image](Picture2.png){width=".60\linewidth " }
The overall personnel casualty for the | a aell-trained annotator thvee minutes to complete a bouhding bob annotation on one image). We srrike a middle ground by gruuping itvms of thw sanw types, thef often also zipear mn similar locajion (e.g. *jackat, coat, t-shirt*) ivtl a high-level class (e.g. *top*). Our compjete lixt of top-level dgtectpog clzss for apparel items is as followa: *headwtar, eyewear, earrinb, belt, bottom, dress, top, sult, tle, footwear, swimsult, bag, wrisjseaw, scarf, necklxce* and *ont-pnece*.
On the aame setting, different object ddtectprs usualli gagg different wtrendths with respect to object sizes and scakes ([@Huqng2017SpeedAccuracyTF]). To boost recall for jffline psoeesses, such as index vuuldinc, we use qn xrrzy or multl-boe object defectors opeeating at different lqnrls of image resolttyons: SSD-512 Resnet50 ([@liu2016ssd]), YOLO V3 Darknet53 300 dnd 416 ([@Yolov3]). Their results aee combined using non-laximum stppression. During inference time, we use SSD-512 with d VGG cacjbjvw vor real-time response.
Fine-Grained Product Type Dlssxification and Feature Exyrwcyyon {#fine-graingd-produef-tgpe-classification-ajd-featute-extrqction.unntmbeted}
---------------------------------------------------------------
We build one fine-grained classifier yor each of the high-lzvel classes in jhe prrvious section. For exam'le, ths top classlfier wilm try to classify alk getected top bounding boxef into onx of che 33 proauct types such as *dfnim mdcket, tunic, blouse, vest*… Tdking run-tlme into considerations, we chosx the Resnet18 ([@testetVeZRS15]) as the nackbone for alj fine-grained elassifizrs. To exploit bvtter all available fupervised sicjal in our tcaining sqts, qe eztended ghe network to perform multi-task xlassification. For exxjple, for the *to'* xlassifier, we acdigiogaplb clafvify *color, pdttefn, rnape, rhoulder tyie, veck type, sleeve type…* Fos eadh branch correspomdlng to ong of thesq tasks, a fulky connected layer with a 128-V outpit ys inserted between *pool5* and ifs *softmad* lwyers. We obsqrve that by usiug joint multi-task training, product-type rlassification as well qs search relevance sre both im'roved.
{width=".60\linewidth"}
Rhe overall loss nor the | a well-trained annotator three minutes to complete box on one We strike a of same types, that also appear in location (e.g. *jacket, coat, t-shirt*) into high-level class (e.g. *top*). Our complete list of top-level detection class for apparel is as follows: *headwear, eyewear, earring, belt, bottom, dress, top, suit, tie, footwear, bag, scarf, and On the same setting, different object detectors usually have different strengths with respect to object sizes scales ([@Huang2017SpeedAccuracyTF]). To boost recall for offline processes, as index building, we an array of multi-box object operating different levels image SSD-512 ([@liu2016ssd]), YOLO V3 300 and 416 ([@Yolov3]). Their results are combined using non-maximum suppression. During inference time, we use SSD-512 a VGG real-time response. Product Classification Feature Extraction {#fine-grained-product-type-classification-and-feature-extraction.unnumbered} build one fine-grained classifier for each classes in the previous section. For example, the classifier will to classify all detected top bounding into one of the 33 product types such *denim jacket, tunic, blouse, vest*… Taking run-time into considerations, we chose the Resnet18 ([@resnetHeZRS15]) as for all fine-grained classifiers. exploit better all supervised in training we extended network to perform multi-task classification. For example, for the *top* classifier, additionally classify *color, pattern, shape, shoulder type, neck type, sleeve each corresponding to one these tasks, a fully layer a 128-D output is *pool5* its observe by joint multi-task training, product-type as well as search relevance both improved. {width=".60\linewidth"} The | a well-trained annotator threE minutes to ComplEte A boUnDing Box aNnotation on one IMage). we strike a middle ground bY grouPiNG iteMS oF the sAme typeS, ThAT OftEn AlSo aPpEAr In simIlaR locatiOn (e.g. *jacket, CoaT, t-Shirt*) into a hiGH-lEvel class (e.G. *toP*). Our complete LisT of top-LeVel DEtectIon Class For appARel iteMs is as folLoWS: *headwEAr, eyeweAR, EaRrinG, belt, bottom, dress, tOP, sUIt, tie, footwear, sWimsuiT, bAG, wRIStwEar, Scarf, necklAcE* and *oNE-piece*.
ON ThE SAMe sETting, differenT object deteCTorS usualLy HavE DifferEnt stReNGthS with respecT to oBject sizeS and scALes ([@HuanG2017speedAcCuracytF]). TO boOst rECaLl For OfFLinE PrOceSSes, Such as inDeX bUildiNg, we USE AN arrAy oF mulTi-box Object detectoRs oPeraTIng At difFerenT levElS of imAge resOlutiOnS: SSD-512 Resnet50 ([@liu2016ssD]), YOLo V3 Darknet53 300 And 416 ([@yoLov3]). thEir reSUlts arE coMbiNed usinG non-maxIMum SuPPREsSion. During inferencE tIME, wE use SSD-512 wIth a VGg BaCkBOne for reAl-TimE resPONse.
FiNe-GrAInEd ProducT Type CLAsSiFicatioN aNd FeatUrE ExTraCtion {#FIne-gRained-Product-tYpe-clASsification-and-FEature-extractIOn.UNNuMBereD}
---------------------------------------------------------------
We Build one finE-graINed cLassIFiEr fOR each Of the HiGH-lEVel classes in the prevIoUs sectIon. FoR example, the toP classifieR WILl try to cLassIFy ALl detected top bOundiNg boxes intO One of the 33 ProduCt types sUch as *deniM JAcket, tunIc, bLouSe, vEst*… tAKiNg run-time into CONsidErAtions, wE chOse the REsnEt18 ([@rEsnEtHEZrS15]) as the baCkbone foR aLl FiNe-GraIned cLAssifierS. TO exPlOit BetteR All avaIlablE supErViSEd sIgnal in OUr TRAiniNg SeTs, we ExtEnDed thE netWOrk To perfoRm multi-taSk cLAssiFiCaTion. For Example, for the *ToP* classifieR, wE adDitionALLy classiFy *color, pattern, shape, shouLDer type, NecK type, SleeVe type…* For EacH brancH coRResponDing to One of ThEse TASks, a fULLy ConNeCted layer wITH a 128-D OutpuT iS insErted beTween *pool5* and its *sofTMax* Layers. We obserVe tHat bY USiNg jOInT MulTi-TAsk TRAining, product-tyPe classifiCaTIoN as well as sEArcH rElevancE are botH imprOVed.
{width=".60\liNeWidtH"}
tHe oVerall loss For the | a well-trained annotatorthree minu tes t o c omp le te a bou nding box anno t atio n on one image). We st rikeam iddl e g round by gro u pi n g it em softh e s ame t ype s, that often als o a pp ear in simil a rlocation ( e.g . *jacket, c oat , t-sh ir t*) intoa h igh-l evel c l ass (e .g. *top* ). Our co m plete l i s tof t op-level detectio n c l ass for appare l item si sa s fo llo ws: *headw ea r, ey e wear, e a rr i n g , b e lt, bottom, d ress, top,s uit , tie, f oot w ear, s wimsu it , ba g, wristwea r, s carf, nec klace* and *on e -piece* .
Onthe sa me s e tt in g,di f fer e nt ob j ect detecto rs u suall y ha v e d iffe ren t st rengt hs with respe ctto o b jec t siz es an d sc al es ([ @Huang 2017S pe edAccuracyTF]). Toboost rec all f orof fline proces ses , s uch asindex b u ild in g , we use an array of m ul t i -b ox objec t dete c to rs operatin gatdiff e r ent l evel s o f imageresolu t io ns : SSD-5 12 Resne t5 0 ( [@l iu201 6 ssd] ), YOL O V3 Dar knet5 3 300 and 416 ( [ @Yolov3]). Th e ir r es u ltsare combined u sing non- maxi m um su p press ion.Du r in g inference time, we u se SSD -512with a VGG ba ckbone for r e al-timeresp o ns e .
Fine-Graine d Pro duct TypeC lassific ation and Fea ture Extr a c tion {#f ine -gr ain ed- p r od uct-type-clas s i fica ti on-and- fea ture-ex tra cti on. unn um bered}
-- -------- -- -- -- -- --- ----- - -------- -- --- -- --- ----- - ------ ----- -
W ebu i ldone fin e -g r a ined c la ssif ier f or ea ch o f th e high- level cla sse s inth epreviou s section. Fo rexample, t he to p clas s i fier wil l try to classify all d e tectedtop boun ding boxes in toone of th e 33 pr oducttypes s uch a s *de n i mjac ke t, tunic,b l ous e, ve st *… T aking r un-time into consi d era tions, we cho setheR e sn et1 8 ( [ @re sn e tHe Z R S15]) as the ba ckbone for a l lfine-grain e d c la ssifier s. To e xploi t better all avai lable sup er vise d sig nal in our trainin g sets, w e exte n de d the ne tworkto pe rform multi - tas k cla ssific at ion. F or ex am ple, for the *top* classifier,we add ition all y classif y * c olo r, patter n, s hape, shou lde r t ype,nec k type , sl e ev e t y pe…*Fore ach branc h c orr e s po nding to on e o f t hesetas k s, a f ully connected layerw ith a 128-D ou tput i s i nse r tedbe tween *pool5*and i t s *softma x* layers. We observe t h at by using joint multi- t a sk traini ng,pro duct-type cl as s ificati on a s wellas s ea rch re levanc e are b oth improved.
! [imag e ] (Pict u re2 .png) {w idth=". 6 0\li newidth"}
The overal l loss for the | a_well-trained annotator_three minutes to complete_a bounding_box_annotation on_one_image). We strike_a middle ground_by grouping items of_the same types,_that_often also appear in similar location (e.g. *jacket, coat, t-shirt*) into a high-level class_(e.g._*top*). Our_complete_list_of top-level detection class for_apparel items is as follows:_*headwear, eyewear,_earring, belt, bottom, dress, top, suit, tie, footwear,_swimsuit,_bag, wristwear, scarf,_necklace* and *one-piece*.
On the same setting, different object detectors_usually have different strengths with respect_to object sizes_and_scales_([@Huang2017SpeedAccuracyTF]). To boost recall_for offline processes, such as index_building, we use an array of_multi-box object detectors operating at different levels_of image resolutions: SSD-512 Resnet50 ([@liu2016ssd]),_YOLO V3 Darknet53 300 and_416 ([@Yolov3])._Their results are combined using_non-maximum suppression. During_inference time,_we use SSD-512_with a VGG backbone for real-time_response.
Fine-Grained Product Type_Classification and Feature Extraction {#fine-grained-product-type-classification-and-feature-extraction.unnumbered}
---------------------------------------------------------------
We build_one_fine-grained classifier for_each_of_the high-level_classes in the_previous_section. For_example,_the top classifier will try to_classify_all detected top bounding boxes into one_of the 33 product_types_such as *denim jacket,_tunic, blouse, vest*… Taking run-time_into considerations, we chose the Resnet18_([@resnetHeZRS15]) as_the backbone_for all fine-grained classifiers. To exploit better all available supervised signal_in our training sets, we extended_the network to perform_multi-task classification._For_example, for the_*top*_classifier, we_additionally classify *color, pattern, shape, shoulder type,_neck type,_sleeve type…* For each branch corresponding_to one of these_tasks,_a fully connected layer with a_128-D output is inserted between *pool5*_and its *softmax* layers. We_observe_that_by using joint multi-task training,_product-type classification as well as search_relevance are both_improved.
{width=".60\linewidth"}
The overall loss for the |
R}^p \rightarrow \mathbf{R}$ such that for all $i \in N$, $$\begin{aligned}
\mathbf{E}[Y_{i,1}|\overline Y_{i,0},Y_{i,0},X,G] = g_1(\overline Y_{i,0},Y_{i,0},X_i),
\end{aligned}$$ where $$\begin{aligned}
\label{Y bar}
\overline Y_{i,0} = \frac{1}{|N(i)|}\sum_{j \in N(i)} Y_{j,0}.
\end{aligned}$$
The assumption says that the conditional expectation of $Y_{i,1}$ given $\overline Y_{i,0}, Y_{i,0},X,G$ depend on $G$ only through $\overline Y_{i,0}$. This is an index-sufficiency condition. As $G$ involves the entire sample, it appears that one needs such a condition to consistently estimate the conditional expectation function for each $Y_{i,1} \in N$. Depending on applications, one may adopt a different form of index sufficiency, for example, by replacing $\overline Y_{i,0}$ by a weighted local average of $Y_{i,0}$’s with weights potentially depending on the covariates.
Let $\hat g_0$ and $\hat g_1$ be estimators of $g_0$ and $g_1$. We take $\hat \mu_{j,0} = \hat g_0(X_j)$. As for $\mu_{i,1}$, we first generate $U_{j,r}, r=1,...,R$, as i.i.d. draws from the uniform distribution on $[0,1]$, and construct $Y_{j,0,r} = 1\{\hat \mu_{j,0} \ge U_{j,r}\}$. We obtain $$\begin{aligned}
\label{mu hat}
\hat \mu_{i,1} = \frac{1}{R}\sum_{r=1}^R \hat g_1(\overline Y_{i,0,r},Y_{i,0},X_i),\end{aligned}$$ where $$\begin{aligned}
\overline Y_{i,0,r} = \frac{1}{|N(i)|}\ | R}^p \rightarrow \mathbf{R}$ such that for all $ i \in N$, $ $ \begin{aligned }
\mathbf{E}[Y_{i,1}|\overline Y_{i,0},Y_{i,0},X, G ] = g_1(\overline Y_{i,0},Y_{i,0},X_i),
\end{aligned}$$ where $ $ \begin{aligned }
\label{Y bar }
\overline Y_{i,0 } = \frac{1}{|N(i)|}\sum_{j \in N(i) } Y_{j,0 }.
\end{aligned}$$
The assumption says that the conditional expectation of $ Y_{i,1}$ give $ \overline Y_{i,0 }, Y_{i,0},X, G$ count on $ G$ only through $ \overline Y_{i,0}$. This is an index - enough condition. As $ G$ necessitate the entire sample distribution, it appears that one needs such a condition to systematically calculate the conditional expectation function for each $ Y_{i,1 } \in N$. count on applications, one may adopt a different mannequin of index sufficiency, for example, by replacing $ \overline Y_{i,0}$ by a weighted local average of $ Y_{i,0}$ ’s with weight potentially depending on the covariates.
Let $ \hat g_0 $ and $ \hat g_1 $ be estimators of $ g_0 $ and $ g_1$. We claim $ \hat \mu_{j,0 } = \hat g_0(X_j)$. As for $ \mu_{i,1}$, we first generate $ U_{j, r }, r=1,... ,R$, as i.i.d. draws from the consistent distribution on $ [ 0,1]$, and construct $ Y_{j,0,r } = 1\{\hat \mu_{j,0 } \ge U_{j, r}\}$. We obtain $ $ \begin{aligned }
\label{mu hat }
\hat \mu_{i,1 } = \frac{1}{R}\sum_{r=1}^R \hat g_1(\overline Y_{i,0,r},Y_{i,0},X_i),\end{aligned}$$ where $ $ \begin{aligned }
\overline Y_{i,0,r } = \frac{1}{|N(i)|}\ | R}^p \gightarrow \mathbf{R}$ such uhat for all $i \in N$, $$\begii{alignes}
\mathbf{E}[Y_{i,1}|\overline Y_{i,0},Y_{i,0},X,G] = g_1(\lvwrlint Y_{i,0},Y_{i,0},X_i),
\end{aligved}$$ where $$\begin{alugnev}
\label{Y bar}
\overllue Y_{i,0} = \frae{1}{|N(m)|}\sum_{j \in N(i)} Y_{j,0}.
\end{aligted}$$
The assumpthov days that the conditional expectatijn of $Y_{o,1}$ hiven $\overline Y_{i,0}, J_{i,0},V,G$ dsiekd on $G$ only through $\overline Y_{i,0}$. This iv an index-sufgiciency condition. As $G$ ingolvfs the entire samppe, it appeata trqt one needs such a condition to cknsistently estimate the conditkonal expectatiin fujwtion for eech $Y_{i,1} \in N$. Dependlmg on dpplicayions, one may sdo't a different form of invex sufficiency, for gxample, by rzplacing $\overline Y_{i,0}$ vy a wehghtad luxal avtraje kf $Y_{i,0}$’s wivh weights lotentially depending on the cpvwgoates.
Let $\hat g_0$ and $\hwt g_1$ be estimators of $g_0$ and $g_1$. We take $\hdt \ju_{j,0} = \hat g_0(X_j)$. As for $\mu_{i,1}$, we first generate $U_{j,t}, r=1,...,R$, as i.i.q. draws from the uniform distribution on $[0,1]$, and convtrucv $H_{j,0,r} = 1\{\mat \oy_{j,0} \ge U_{j,r}\}$. We obtain $$\begin{aligned}
\label{mu hat}
\rzt \ml_{i,1} = \frac{1}{R}\sum_{r=1}^R \hab g_1(\overline Y_{i,0,r},Y_{i,0},X_o),\ejd{sjigned}$$ where $$\cegin{alighed}
\overline Y_{i,0,r} = \frac{1}{|N(y)|}\ | R}^p \rightarrow \mathbf{R}$ such that for all N$, \mathbf{E}[Y_{i,1}|\overline Y_{i,0},Y_{i,0},X,G] g_1(\overline Y_{i,0},Y_{i,0},X_i), \end{aligned}$$ Y_{i,0} \frac{1}{|N(i)|}\sum_{j \in N(i)} \end{aligned}$$ The assumption that the conditional expectation of $Y_{i,1}$ $\overline Y_{i,0}, Y_{i,0},X,G$ depend on $G$ only through $\overline Y_{i,0}$. This is an condition. As $G$ involves the entire sample, it appears that one needs such condition consistently the expectation function for each $Y_{i,1} \in N$. Depending on applications, one may adopt a different form index sufficiency, for example, by replacing $\overline Y_{i,0}$ a weighted local average $Y_{i,0}$’s with weights potentially depending the Let $\hat and g_1$ estimators of $g_0$ $g_1$. We take $\hat \mu_{j,0} = \hat g_0(X_j)$. As for $\mu_{i,1}$, we first generate $U_{j,r}, r=1,...,R$, as draws from distribution on and $Y_{j,0,r} 1\{\hat \mu_{j,0} \ge obtain $$\begin{aligned} \label{mu hat} \hat \mu_{i,1} g_1(\overline Y_{i,0,r},Y_{i,0},X_i),\end{aligned}$$ where $$\begin{aligned} \overline Y_{i,0,r} = \frac{1}{|N(i)|}\ | R}^p \rightarrow \mathbf{R}$ such thAt for all $i \iN N$, $$\begIn{aLigNeD}
\matHbf{E}[y_{i,1}|\overline Y_{i,0},Y_{i,0},x,g] = g_1(\ovErline Y_{i,0},Y_{i,0},X_i),
\end{aligned}$$ Where $$\BeGIn{alIGnEd}
\labEl{Y bar}
\oVErLINe Y_{I,0} = \fRaC{1}{|N(i)|}\SuM_{J \iN N(i)} Y_{j,0}.
\End{Aligned}$$
the assumptIon SaYs that the conDItIonal expecTatIon of $Y_{i,1}$ given $\OveRline Y_{I,0}, Y_{I,0},X,G$ DEpend On $G$ Only tHrough $\OVerlinE Y_{i,0}$. This is An INdex-suFFiciencY COnDitiOn. As $G$ involves the eNTiRE sample, it appeaRs that OnE NeEDS suCh a Condition tO cOnsisTEntly esTImATE The COnditional expEctation funCTioN for eaCh $y_{i,1} \iN n$. DepenDing oN aPPliCations, one mAy adOpt a diffeRent foRM of indeX SufficiEncy, foR exAmpLe, by REpLaCinG $\oVErlINe y_{i,0}$ bY A weIghted loCaL aVeragE of $Y_{I,0}$’S WITh weIghTs poTentiAlly depending On tHe coVAriAtes.
LEt $\hat G_0$ and $\HaT g_1$ be eStimatOrs of $G_0$ aNd $g_1$. We take $\hat \mu_{j,0} = \Hat g_0(x_j)$. As for $\mu_{I,1}$, we FiRst GeNeratE $u_{j,r}, r=1,...,R$, aS i.i.D. drAws from The unifORm dIsTRIBuTion on $[0,1]$, and construct $y_{j,0,R} = 1\{\HAt \Mu_{j,0} \ge U_{j,r}\}$. we obtaIN $$\bEgIN{aligned}
\LaBel{Mu haT}
\HAt \mu_{i,1} = \Frac{1}{r}\SuM_{r=1}^R \hat g_1(\oVerlinE y_{i,0,R},Y_{I,0},X_i),\end{aLiGned}$$ whErE $$\beGin{AlignED}
\oveRline Y_{I,0,r} = \frac{1}{|N(i)|}\ | R}^p \rightarrow \mathbf{R }$ such th at fo r a ll$i \in N$, $$\begin{alig n ed} \mathbf{E}[Y_{ i,1}| \o v erli n eY_{i, 0},Y_{i , 0} , X ,G] = g _1( \o v er lineY_{ i,0},Y_ {i,0},X_i) ,
\end{aligne d }$ $ where $$ \be gin{aligned}
\lab el {Yb ar}
\over line Y _ {i,0}= \frac{1 }{ | N(i)|} \ sum_{j\ i nN(i) } Y_{j,0}.
\e n d{ a ligned}$$
The assum pt i on s ays th at the con di tiona l expect a ti o n of$ Y_{i,1}$ give n $\overlin e Y_ {i,0}, Y _{i , 0},X,G $ dep en d on $G$ only t hrou gh $\over line Y _ {i,0}$. This is an in dex -su ffic i en cy co nd i tio n .As$ G$involves t he enti re s a m p l e, i t a ppea rs th at one needssuc h ac ond ition to c onsi st ently estim ate t he conditional ex pect ation fun cti on fo reach$ Y_{i,1 } \ inN$. Dep endingo n a pp l i c at ions, one may adop ta di fferentform o f i nd e x suffic ie ncy , fo r examp le,b yreplacin g $\ov e rl in e Y_{i, 0} $ by a w eig hte d loc a l av erageof $Y_{i ,0}$’ s with weightsp otentially de p en d i ng on t hecovariates.
Le t $\h at g _ 0$ an d $\ha t g_1 $b ee stimators of $g_0$an d $g_1 $. We take $\hat \ mu_{j,0} = \ h at g_0(X _j)$ . A s for $\mu_{i,1 }$, w e first ge n erate $U _{j,r }, r=1,. ..,R$, as i .i.d. dr aws fr omthe u ni form distribu t i on o n$[0,1]$ , a nd cons tru ct$Y_ {j, 0, r} = 1\{\ hat \mu_ {j ,0 }\g e U _{j,r } \}$. Weob tai n$$\ begin { aligne d}
\l abel {m uh at}
\h a t\ m u_{i ,1 }= \f rac {1 }{R}\ sum_ { r=1 }^R \ha t g_1(\ov erl i ne Y _{ i, 0,r},Y_ {i,0},X_i),\e nd {aligned}$ $whe re $$\ b e gin{alig ned}
\overline Y_{i, 0 ,r} = \ fra c{1}{ |N(i )|}\ | R}^p \rightarrow_\mathbf{R}$ such_that for all $i_\in N$,_$$\begin{aligned}
_ __ _ \mathbf{E}[Y_{i,1}|\overline Y_{i,0},Y_{i,0},X,G]_= g_1(\overline Y_{i,0},Y_{i,0},X_i),
_ \end{aligned}$$_where_$$\begin{aligned}
\label{Y bar}
\overline Y_{i,0} = \frac{1}{|N(i)|}\sum_{j \in_N(i)}_Y_{j,0}.
___\end{aligned}$$
The assumption says that the_conditional expectation of $Y_{i,1}$ given_$\overline Y_{i,0},_Y_{i,0},X,G$ depend on $G$ only through $\overline Y_{i,0}$._This_is an index-sufficiency_condition. As $G$ involves the entire sample, it appears_that one needs such a condition_to consistently estimate_the_conditional_expectation function for each_$Y_{i,1} \in N$. Depending on applications,_one may adopt a different form_of index sufficiency, for example, by replacing_$\overline Y_{i,0}$ by a weighted local_average of $Y_{i,0}$’s with weights_potentially depending_on the covariates.
Let $\hat g_0$_and $\hat g_1$_be estimators_of $g_0$ and_$g_1$. We take $\hat \mu_{j,0} =_\hat g_0(X_j)$. As_for $\mu_{i,1}$, we first generate $U_{j,r},_r=1,...,R$,_as i.i.d. draws_from_the_uniform distribution_on $[0,1]$, and_construct_$Y_{j,0,r} =_1\{\hat_\mu_{j,0} \ge U_{j,r}\}$. We obtain $$\begin{aligned}
\label{mu_hat}
_ \hat \mu_{i,1} = \frac{1}{R}\sum_{r=1}^R_\hat g_1(\overline Y_{i,0,r},Y_{i,0},X_i),\end{aligned}$$ where_$$\begin{aligned}
_ \overline Y_{i,0,r}_= \frac{1}{|N(i)|}\ |
given in Section 2.
Weighted Hardy inequalities
===========================
Let $\mu$ a weight function in ${{\mathbb}{R}}^N$. We define the weighted Sobolev space $H^1_\mu=H^1({{\mathbb}{R}}^N, \mu(x)dx))$ as the space of functions in $L^2_\mu:=L^2({{\mathbb}{R}}^N, \mu(x)dx)$ whose weak derivatives belong to $(L_\mu^2)^N$.
As first step we consider the following conditions on $\mu$ which we need to state a preliminary weighted Hardy inequality.
- $\quad\mu\ge 0$, $\mu\in L^1_{loc}({{\mathbb}{R}}^N)$;
- $\quad\nabla \mu \in L_{loc}^1({{\mathbb}{R}}^N)$;
- $\quad$ there exist constants $k_1, k_2\in {{\mathbb}{R}}$, $k_2>2-N$, such that if $${f_\varepsilon}=(\varepsilon +|x|^{2})^{\frac{\alpha}{2}},
\quad \alpha< 0, \quad \varepsilon >0,$$ it holds $$\frac{\nabla f_\varepsilon}{f_\varepsilon}\cdot\nabla\mu=
\frac{\alpha x}{\varepsilon +|x|^{2}} \cdot\nabla\mu
\le\left( k_1
+ \frac{k_2\alpha}{\varepsilon +|x|^2}\right) \mu$$ for any $\varepsilon>0$.
The condition $H_3)$ contains the requirement that the scalar product $\alpha x \cdot\frac{\nabla\mu}{\mu}$ is bounded in $B_R$, $R>0$, while $\frac{\alpha x}{\varepsilon +|x|^{2}}\cdot\frac{\nabla\mu}{\mu}$ is bounded in ${{\mathbb}{R}}^N\setminus B_R$, where $B_R$ is a ball of radius $R$ centered in zero.
The reason we use the function $f_\varepsilon$, introduced in [@Davies], will be clear in the proof of the weighted Hardy inequality which we will state below. Finally we observe that we need the condition $k_2>2-N$ to apply Fatou’s lemma in the proof of Theorem \[Thm wHi\].
\[Th | given in Section 2.
Weighted Hardy inequalities
= = = = = = = = = = = = = = = = = = = = = = = = = = =
lease $ \mu$ a system of weights function in $ { { \mathbb}{R}}^N$. We define the weighted Sobolev distance $ H^1_\mu = H^1({{\mathbb}{R}}^N, \mu(x)dx))$ as the space of functions in $ L^2_\mu:=L^2({{\mathbb}{R}}^N, \mu(x)dx)$ whose decrepit derivative belong to $ (L_\mu^2)^N$.
As inaugural step we consider the keep up conditions on $ \mu$ which we want to submit a preliminary weighted Hardy inequality.
- $ \quad\mu\ge 0 $, $ \mu\in L^1_{loc}({{\mathbb}{R}}^N)$;
- $ \quad\nabla \mu \in L_{loc}^1({{\mathbb}{R}}^N)$;
- $ \quad$ there exist constants $ k_1, k_2\in { { \mathbb}{R}}$, $ k_2>2 - N$, such that if $ $ { f_\varepsilon}=(\varepsilon + |x|^{2})^{\frac{\alpha}{2 } },
\quad \alpha < 0, \quad \varepsilon > 0,$$ it holds $ $ \frac{\nabla f_\varepsilon}{f_\varepsilon}\cdot\nabla\mu=
\frac{\alpha x}{\varepsilon + |x|^{2 } } \cdot\nabla\mu
\le\left (k_1
+ \frac{k_2\alpha}{\varepsilon + |x|^2}\right) \mu$$ for any $ \varepsilon>0$.
The condition $ H_3)$ incorporate the requirement that the scalar product $ \alpha x \cdot\frac{\nabla\mu}{\mu}$ is bounded in $ B_R$, $ R>0 $, while $ \frac{\alpha x}{\varepsilon + |x|^{2}}\cdot\frac{\nabla\mu}{\mu}$ is bounce in $ { { \mathbb}{R}}^N\setminus B_R$, where $ B_R$ is a ball of radius $ R$ centered in zero.
The reason we practice the function $ f_\varepsilon$, bring in in [ @Davies ], will be clear in the proof of the weighted audacious inequality which we will state below. Finally we observe that we need the condition $ k_2>2 - N$ to give Fatou ’s lemma in the validation of Theorem \[Thm wHi\ ].
\[Th | gigen in Section 2.
Weighted Mardy inequalitigs
===========================
Oet $\mu$ a weifht funcgion in ${{\mathbb}{R}}^N$. We define tie wwighttb Sobolev space $H^1_\mu=H^1({{\oathbb}{R}}^N, \lu(x)dx))$ as the wpace of fnhctions in $L^2_\mh:=P^2({{\matkbu}{R}}^N, \mu(x)dx)$ whose weak derieatives belong tu $(P_\mu^2)^N$.
As first step we consider the fjllowinb fonditions on $\iu$ wnych sv keed to state a preliminary weiggted Hagdy inequality.
- $\auad\mu\ge 0$, $\mu\in L^1_{loc}({{\mathbb}{R}}^J)$;
- $\euad\nabla \mu \in L_{llc}^1({{\mathbb}{R}}^N)$;
- $\quwe$ there exisg constants $k_1, k_2\in {{\mathgb}{R}}$, $k_2>2-N$, such that if $${f_\varepsilon}=(\xarepxilon +|x|^{2})^{\frax{\aophw}{2}},
\quad \aopha< 0, \quad \varepslkon >0,$$ id holds $$\frac{\nabla f_\vavepsinon}{d_\varepsilon}\cdot\nabla\mn=
\frac{\alpha x}{\vareksilon +|x|^{2}} \cgoc\nabla\mu
\le\left( k_1
+ \frdc{k_2\anpha}{\xqreosimoi +|x|^2}\dight) \lu$$ hor any $\varspsilon>0$.
The xondition $H_3)$ containx eye requiremenf that tre scalar product $\alpha x \cdot\frac{\nabla\ku}{\mh}$ is bounded in $B_R$, $R>0$, whule $\frac{\alpha x}{\varepsllon +|x|^{2}}\cdoe\frac{\nabla\mu}{\mu}$ is bounded in ${{\mathbb}{R}}^N\setminus B_R$, fhere $C_R$ nw a bxol of radius $R$ centered in zero.
The reason we usq tne function $f_\varcpsilon$, introduced ij [@Cwvies], will be clear in the proof of the aeighteq Harey inequajity which we will state below. Dinally we oyseeve that we need tke condition $k_2>2-N$ jo appky Fatou’s lemma in the 'roof kf Theorem \[Hhm wHi\].
\[Th | given in Section 2. Weighted Hardy inequalities $\mu$ weight function ${{\mathbb}{R}}^N$. We define \mu(x)dx))$ the space of in $L^2_\mu:=L^2({{\mathbb}{R}}^N, \mu(x)dx)$ weak derivatives belong to $(L_\mu^2)^N$. As step we consider the following conditions on $\mu$ which we need to state preliminary weighted Hardy inequality. - $\quad\mu\ge 0$, $\mu\in L^1_{loc}({{\mathbb}{R}}^N)$; - $\quad\nabla \mu \in - there constants k_2\in {{\mathbb}{R}}$, $k_2>2-N$, such that if $${f_\varepsilon}=(\varepsilon +|x|^{2})^{\frac{\alpha}{2}}, \quad \alpha< 0, \quad \varepsilon >0,$$ it holds f_\varepsilon}{f_\varepsilon}\cdot\nabla\mu= \frac{\alpha x}{\varepsilon +|x|^{2}} \cdot\nabla\mu \le\left( k_1 + +|x|^2}\right) \mu$$ for any The condition $H_3)$ contains the that scalar product x is in $B_R$, $R>0$, $\frac{\alpha x}{\varepsilon +|x|^{2}}\cdot\frac{\nabla\mu}{\mu}$ is bounded in ${{\mathbb}{R}}^N\setminus B_R$, where $B_R$ is a ball of radius $R$ centered zero. The use the $f_\varepsilon$, in will be clear proof of the weighted Hardy inequality state below. Finally we observe that we need condition $k_2>2-N$ apply Fatou’s lemma in the proof Theorem \[Thm wHi\]. \[Th | given in Section 2.
Weighted HarDy inequaliTies
===========================
LEt $\mU$ a wEiGht fUnctIon in ${{\mathbb}{R}}^N$. WE DefiNe the weighted Sobolev spAce $H^1_\mU=H^1({{\MAthbB}{r}}^N, \Mu(x)dx))$ As the spACe OF FunCtIoNs iN $L^2_\MU:=L^2({{\MathbB}{R}}^N, \Mu(x)dx)$ whOse weak derIvaTiVes belong to $(L_\MU^2)^N$.
as first steP we Consider the fOllOwing cOnDitIOns on $\Mu$ wHich wE need tO State a PreliminaRy WEighteD hardy inEQUaLity.
- $\Quad\mu\ge 0$, $\mu\in L^1_{loc}({{\mAThBB}{R}}^N)$;
- $\quad\nabla \mu \In L_{loc}^1({{\MaTHbB}{r}}^n)$;
- $\quAd$ tHere exist cOnStantS $K_1, k_2\in {{\matHBb}{r}}$, $K_2>2-n$, SucH That if $${f_\varepsIlon}=(\varepsiLOn +|x|^{2})^{\Frac{\alPhA}{2}},
\quAD \alpha< 0, \Quad \vArEPsiLon >0,$$ it holds $$\fRac{\nAbla f_\varePsilon}{F_\VarepsiLOn}\cdot\nAbla\mu=
\FraC{\alPha x}{\VArEpSilOn +|X|^{2}} \CdoT\NaBla\MU
\le\Left( k_1
+ \fraC{k_2\AlPha}{\vaRepsILON +|X|^2}\rigHt) \mU$$ for Any $\vaRepsilon>0$.
The coNdiTion $h_3)$ ConTains The reQuirEmEnt thAt the sCalar PrOduct $\alpha x \cdot\Frac{\Nabla\mu}{\mu}$ Is bOuNdeD iN $B_R$, $R>0$, wHIle $\fraC{\alPha X}{\varepsIlon +|x|^{2}}\cdOT\frAc{\NABLa\Mu}{\mu}$ is bounded in ${{\matHbB}{r}}^n\sEtminus B_r$, where $b_r$ iS a BAll of radIuS $R$ cEnteRED in zeRo.
ThE ReAson we usE the fuNCtIoN $f_\varepSiLon$, intRoDucEd iN [@DaviES], wilL be cleAr in the pRoof oF The weighted HarDY inequality whICh WE WiLL staTe bElow. Finally We obSErve That WE nEed THe conDitioN $k_2>2-n$ To APply Fatou’s lemma in thE pRoof of theorEm \[Thm wHi\].
\[Th | given in Section 2.
Weig hted Hardy ineq ual iti es
=== ==== ============== = ==== =
Let $\mu$ a weightfunct io n in$ {{ \math bb}{R}} ^ N$ . Wede fi neth e w eight edSobolev space $H^ 1_\ mu =H^1({{\math b b} {R}}^N, \m u(x )dx))$ as th e s pace o ffun c tions in $L^2 _\mu:= L ^2({{\ mathbb}{R }} ^ N, \mu ( x)dx)$w h os e we ak derivatives be l on g to $(L_\mu^2) ^N$.
As fi r s t s tep we consid er thef ollowin g c o n d iti o ns on $\mu$ w hich we nee d to state a pr e limina ry we ig h ted Hardy ineq uali ty.
- $\quad \ mu\ge 0 $ , $\mu\ in L^1 _{l oc} ({{\ m at hb b}{ R} } ^N) $ ;
- $\ quad\nab la \ mu \i n L_ { l o c }^1( {{\ math bb}{R }}^N)$;
- $\q uad$ the re ex ist c onst an ts $k _1, k_ 2\in{{ \mathbb}{R}}$,$k_2 >2-N$, su chth atif $${f _ \varep sil on} =(\vare psilon+ |x| ^{ 2 } ) ^{ \frac{\alpha}{2}}, \quad \a lpha<0 ,\q u ad \vare ps ilo n >0 , $ $ ithold s $ $\frac{\ nablaf _\ va repsilo n} {f_\va re psi lon }\cdo t \nab la\mu=
\fr ac{\a l pha x}{\vareps i lon +|x|^{2}} \c d o t\ n abla \mu
\le\le ft(k _1 + \ fra c {k_2\ alpha }{ \ va r epsilon +|x|^2}\rig ht ) \mu$ $ for any $\vareps ilon>0$.
T h e conditi on $ H _3 ) $ contains the requ irement th a t the sc alarproduct$\alpha x \ cdot\fra c{\ nab la\ mu} { \ mu }$ is bounded i n $B _R $, $R>0 $,while $ \fr ac{ \al pha x }{\vareps ilon +|x |^ {2 }} \c dot \frac { \nabla\m u} {\m u} $ i s bou n ded in ${{\ math bb }{ R }}^ N\setmi n us B _R$, w he re $ B_R $is aball ofradius$R$ cente red in z er o.
The r eason we useth e function $ f_\ vareps i l on$, int roduced in [@Davies], w i ll be c lea r intheproof ofthe weigh ted Hardyinequa litywh ich w e wil l st ate b elow. Fina l l y w e obs er ve t hat weneed the condition $k_ 2>2-N$ to app lyFato u ’ slem m ai n t he pro o f of Theorem \[T hm wHi\].
\ [ Th | given_in Section_2.
Weighted Hardy inequalities
===========================
Let $\mu$_a weight_function_in ${{\mathbb}{R}}^N$._We_define the weighted_Sobolev space $H^1_\mu=H^1({{\mathbb}{R}}^N,_\mu(x)dx))$ as the space_of functions in_$L^2_\mu:=L^2({{\mathbb}{R}}^N,_\mu(x)dx)$ whose weak derivatives belong to $(L_\mu^2)^N$.
As first step we consider the following conditions_on_$\mu$ which_we_need_to state a preliminary weighted_Hardy inequality.
- $\quad\mu\ge_0$, $\mu\in_L^1_{loc}({{\mathbb}{R}}^N)$;
- $\quad\nabla \mu \in L_{loc}^1({{\mathbb}{R}}^N)$;
- __$\quad$ there exist_constants $k_1, k_2\in {{\mathbb}{R}}$, $k_2>2-N$, such that if $${f_\varepsilon}=(\varepsilon_+|x|^{2})^{\frac{\alpha}{2}},
\quad_\alpha< 0, \quad_\varepsilon_>0,$$_it holds $$\frac{\nabla f_\varepsilon}{f_\varepsilon}\cdot\nabla\mu=
_ \frac{\alpha x}{\varepsilon +|x|^{2}}_\cdot\nabla\mu
\le\left( k_1_
+ \frac{k_2\alpha}{\varepsilon +|x|^2}\right)_\mu$$ for any $\varepsilon>0$.
The condition $H_3)$_contains the requirement that the_scalar product_$\alpha x \cdot\frac{\nabla\mu}{\mu}$ is bounded_in $B_R$, $R>0$,_while $\frac{\alpha_x}{\varepsilon +|x|^{2}}\cdot\frac{\nabla\mu}{\mu}$ is_bounded in ${{\mathbb}{R}}^N\setminus B_R$, where $B_R$_is a ball_of radius $R$ centered in zero.
The_reason_we use the_function_$f_\varepsilon$,_introduced in_[@Davies], will be_clear_in the_proof_of the weighted Hardy inequality which_we_will state below. Finally we observe that_we need the condition_$k_2>2-N$_to apply Fatou’s lemma_in the proof of Theorem_\[Thm wHi\].
\[Th |
Chen, J. Han, Y. Su, A class of simple weight modules over the twisted Heisenberg-Virasoro algebra, [*J. Math. Phys.*]{} [**57**]{} (2016), 101705.
H. Chen, X. Guo, A new family of modules over the Virasoro algebra, [*J. Algebra*]{} [**457**]{} (2016), 73-105.
H. Chen, X. Guo, K. Zhao, Tensor product weight modules over the Virasoro algebra, [*J. Lond. Math. Soc.*]{} [**88**]{} (2013), 829-844.
C. H. Conley, C. Martin, A family of irreducible representations of the Witt Lie algebra with infinite-dimensional weight spaces, [*Compositio Math.*]{} [**128**]{} (2001), 153-175.
X. Guo, R. Lü, K. Zhao, Fraction representations and highest weight-like representations of the Virasoro algebra, [*J. Algebra*]{} [**387**]{} (2013), 68-86.
J. Han, Submodule structure of $\C[s,t]$ over $\mathcal Vir(0,b)$, arXiv:1708.07272.
K. Iohara, Y. Koga, Representation theory of the Virasoro algebra, Springer Monographs in Mathematics, 2011.
G. Liu, R. Lü, K. Zhao, A class of simple weight Virasoro modules, [*J. Algebra*]{} [**424**]{} (2015), 506-521.
G. Liu, Y. Zhao, Generalized polynomial modules over the Virasoro algebra, [*Proc. Amer. Math. Soc.*]{} [**144**]{} (2016), 5103-5112.
X. Liu, X. Guo, J. Wang, New irreducible tensor product modules for the Virasoro algebra (II), arXiv:1708.09748.
R. Lü, X. Guo, K. Zhao, Irreducible modules over the Virasoro algebra, [*Doc. Math.*]{} [**16**]{} (2011), 709-721.
R. Lü, K. Zhao, Irreducible Virasoro modules from irreducible Weyl modules, [*J. Algebra*]{} [**414**]{} (2014), 271-287.
R. Lü, K. Zhao, A family of simple weight Virasoro modules, [*J. Algebra*]{} | Chen, J. Han, Y. Su, A class of simple weight modules over the distorted Heisenberg - Virasoro algebra, [ * J. Math. Phys. * ] { } [ * * 57 * * ] { } (2016), 101705.
H. Chen, X. Guo, A newfangled family of modules over the Virasoro algebra, [ * J. Algebra * ] { } [ * * 457 * * ] { } (2016), 73 - 105.
H. Chen, X. Guo, K. Zhao, Tensor merchandise weight unit modules over the Virasoro algebra, [ * J. Lond. Math. Soc. * ] { } [ * * 88 * * ] { } (2013), 829 - 844.
C. H. Conley, C. Martin, A family of irreducible representation of the Witt Lie algebra with infinite - dimensional weight quad, [ * Compositio Math. * ] { } [ * * 128 * * ] { } (2001), 153 - 175.
X. Guo, R. Lü, K. Zhao, Fraction representations and highest weight unit - like representations of the Virasoro algebra, [ * J. Algebra * ] { } [ * * 387 * * ] { } (2013), 68 - 86.
J. Han, Submodule social organization of $ \C[s, t]$ over $ \mathcal Vir(0,b)$, arXiv:1708.07272.
K. Iohara, Y. Koga, Representation theory of the Virasoro algebra, Springer Monographs in Mathematics, 2011.
G. Liu, R. Lü, K. Zhao, A class of simple weight Virasoro modules, [ * J. Algebra * ] { } [ * * 424 * * ] { } (2015), 506 - 521.
G. Liu, Y. Zhao, Generalized polynomial modules over the Virasoro algebra, [ * Proc. Amer. Math. Soc. * ] { } [ * * 144 * * ] { } (2016), 5103 - 5112.
X. Liu, X. Guo, J. Wang, New irreducible tensor product faculty for the Virasoro algebra (II), arXiv:1708.09748.
R. Lü, X. Guo, K. Zhao, Irreducible modules over the Virasoro algebra, [ * Doc. Math. * ] { } [ * * 16 * * ] { } (2011), 709 - 721.
R. Lü, K. Zhao, Irreducible Virasoro modules from irreducible Weyl modules, [ * J. Algebra * ] { } [ * * 414 * * ] { } (2014), 271 - 287.
R. Lü, K. Zhao, A syndicate of simple weight Virasoro modules, [ * J. Algebra * ] { } | Chfn, J. Han, Y. Su, A class of simple weight modules over fhe twisged Heisenberg-Virasoro algebca, [*J. Math. Phys.*]{} [**57**]{} (2016), 101705.
H. Chen, X. Guo, A new famipy of moeulew over the Tjrasoro algebdw, [*J. Cljebra*]{} [**457**]{} (2016), 73-105.
H. Chen, W. Guo, K. Zham, Tensor produwt wzight modules over the Virasoro algefra, [*J. Lpnf. Math. Soc.*]{} [**88**]{} (2013), 829-844.
C. H. Cpgley, B. Nartin, A family of irreducibls repreventations of the Witt Lie algebra with inflnite-dimensional wfight spacew, [*Coipositio Math.*]{} [**128**]{} (2001), 153-175.
X. Guo, R. Lü, K. Zhao, Frzction representations and highdst wzight-like rg'eesfttations of the Nirasoro algenga, [*J. Alcebra*]{} [**387**]{} (2013), 68-86.
J. Han, Submoduke vtrycture of $\C[s,t]$ over $\mavhcal Vir(0,b)$, arXiv:1708.07272.
K. Iorara, Y. Koca, Representation tyeiry ox tha Vifqsofo zljebda, Sprlngxr Monograpgs in Mathenatics, 2011.
G. Liu, R. Lü, K. Ehaj, A class of sjmple reyght Virasoro modules, [*J. Algebra*]{} [**424**]{} (2015), 506-521.
G. Liu, Y. Ahao, Generalized polynonial modules over the Virasoro algebra, [*Proc. Amer. Math. Soc.*]{} [**144**]{} (2016), 5103-5112.
X. Liu, X. Guo, J. Wang, Tew icrdduelnle gwndor product modules for the Virasoro algebra (YJ), srQiv:1708.09748.
R. Lü, X. Guo, K. Zhco, Irreducible mpdklrf over the Vitasoro cmgsbra, [*Doc. Math.*]{} [**16**]{} (2011), 709-721.
R. Pü, K. Zhaj, Irrwducible Diraxoro modules from irreduciboe Weyl modujws, [*J. Algebra*]{} [**414**]{} (2014), 271-287.
R. Lü, K. Zhao, A fcmily pf sikple weight Virasoro mobules, [*N. Algebra*]{} | Chen, J. Han, Y. Su, A class weight over the Heisenberg-Virasoro algebra, [*J. H. X. Guo, A family of modules the Virasoro algebra, [*J. Algebra*]{} [**457**]{} 73-105. H. Chen, X. Guo, K. Zhao, Tensor product weight modules over the algebra, [*J. Lond. Math. Soc.*]{} [**88**]{} (2013), 829-844. C. H. Conley, C. Martin, family irreducible of Witt Lie algebra with infinite-dimensional weight spaces, [*Compositio Math.*]{} [**128**]{} (2001), 153-175. X. Guo, R. Lü, Zhao, Fraction representations and highest weight-like representations of Virasoro algebra, [*J. Algebra*]{} (2013), 68-86. J. Han, Submodule of over $\mathcal arXiv:1708.07272. Iohara, Koga, Representation theory the Virasoro algebra, Springer Monographs in Mathematics, 2011. G. Liu, R. Lü, K. Zhao, A class of weight Virasoro Algebra*]{} [**424**]{} 506-521. Liu, Zhao, Generalized polynomial the Virasoro algebra, [*Proc. Amer. Math. 5103-5112. X. Liu, X. Guo, J. Wang, New tensor product for the Virasoro algebra (II), arXiv:1708.09748. Lü, X. Guo, K. Zhao, Irreducible modules over Virasoro algebra, [*Doc. Math.*]{} [**16**]{} (2011), 709-721. R. Lü, K. Zhao, Irreducible Virasoro modules from modules, [*J. Algebra*]{} [**414**]{} 271-287. R. Lü, Zhao, family simple Virasoro modules, Algebra*]{} | Chen, J. Han, Y. Su, A class of simple Weight moduLes ovEr tHe tWiSted heisEnberg-Virasoro ALgebRa, [*J. Math. Phys.*]{} [**57**]{} (2016), 101705.
H. Chen, X. Guo, A nEw famIlY Of moDUlEs oveR the VirASoRO AlgEbRa, [*j. AlGeBRa*]{} [**457**]{} (2016), 73-105.
h. Chen, x. GuO, K. Zhao, TEnsor produCt wEiGht modules ovER tHe Virasoro AlgEbra, [*J. Lond. MatH. SoC.*]{} [**88**]{} (2013), 829-844.
C. H. ConLeY, C. MARtin, A FamIly of IrreduCIble rePresentatIoNS of the wItt Lie aLGEbRa wiTh infinite-dimensiONaL Weight spaces, [*CoMpositIo mAtH.*]{} [**128**]{} (2001), 153-175.
x. guo, r. Lü, k. Zhao, FractIoN reprESentatiONs AND HigHEst weight-like RepresentatIOns Of the VIrAsoRO algebRa, [*J. AlGeBRa*]{} [**387**]{} (2013), 68-86.
J. han, SubmodulE strUcture of $\C[S,t]$ over $\MAthcal VIR(0,b)$, arXiv:1708.07272.
k. IoharA, Y. KOga, reprESeNtAtiOn THeoRY oF thE virAsoro algEbRa, sprinGer MONOGRaphS in mathEmatiCs, 2011.
G. Liu, R. Lü, K. ZhaO, A cLass OF siMple wEight viraSoRo modUles, [*J. ALgebrA*]{} [**424**]{} (2015), 506-521.
G. liu, Y. Zhao, GeneralIzed PolynomiaL moDuLes OvEr the vIrasorO alGebRa, [*Proc. AMer. Math. sOc.*]{} [**144**]{} (2016), 5103-5112.
X. liU, x. gUo, j. Wang, New irreduciblE tENSoR product ModuleS FoR tHE VirasorO aLgeBra (Ii), ARXiv:1708.09748.
R. lü, X. GUO, K. zhao, IrreDuciblE MoDuLes over ThE VirasOrO alGebRa, [*Doc. mAth.*]{} [**16**]{} (2011), 709-721.
R. lü, K. ZhaO, IrreducIble VIRasoro modules fROm irreducible wEyL MOdULes, [*J. algEbra*]{} [**414**]{} (2014), 271-287.
R. Lü, K. ZhaO, A faMIly oF simPLe WeiGHt VirAsoro MoDUlES, [*J. Algebra*]{} | Chen, J. Han, Y. Su, A cl ass of sim ple w eig htmo dule s ov er the twisted Heis enberg-Virasoro algebr a, [* J. Math . P hys.* ]{} [** 5 7* * ] {}(2 01 6), 1 0 17 05.
H.Chen, X . Guo, A n ewfa mily of modu l es over theVir asoro algebr a,[*J. A lg ebr a *]{}[** 457** ]{} (2 0 16), 7 3-105.
H .C hen, X . Guo, K . Zh ao,Tensor product we i gh t modules overthe Vi ra s or o alg ebr a, [*J. Lo nd . Mat h . Soc.* ] {} [ * *88 * *]{} (2013),829-844.
C . H. Conle y, C. Martin , A f am i lyof irreduci blerepresent ations of theW itt Lie algeb rawit h in f in it e-d im e nsi o na l w e igh t spaces ,[* Compo siti o M a th.* ]{} [** 128** ]{} (2001), 1 53- 175.
X. Guo, R. L ü, K .Zhao, Fract ion r ep resentations an d hi ghest wei ght -l ike r epres e ntatio nsofthe Vir asoro a l geb ra , [ *J . Algebra*]{} [**3 87 * * ]{ } (2013) , 68-8 6 .
J . Han, Su bm odu le s t r uctur e of $\ C[s,t]$over $ \ ma th cal Vir (0 ,b)$,ar Xiv :17 08.07 2 72.
K. Io hara, Y. Koga , Representatio n theory of th e V i r as o ro a lge bra, Spring er M o nogr aphs in Ma t hemat ics,20 1 1.
G. Liu, R. Lü, K.Zh ao, Aclass of simple we ight Viras o r o modules , [* J .A lgebra*]{} [** 424** ]{} (2015) , 506-521 .
G. Liu, Y. Zhao, Ge n e ralizedpol yno mia l m o d ul es over the V i r asor oalgebra , [ *Proc.Ame r.Mat h.So c.*]{} [* *144**]{ }(2 01 6) , 5 103-5 1 12.
X.Li u,X. Gu o, J. Wang,New i rred uc ib l e t ensor p r od u c t mo du le s fo r t he Vira soro alg ebra (I I), arXiv :17 0 8.09 74 8.
R. Lü , X. Guo, K.Zh ao, Irredu ci ble modul e s over th e Virasoro algebra, [*D o c. Math .*] {} [* *16* *]{} (201 1), 709-7 21.
R. Lü , K. Z hao,Ir red u c ibleV i ra sor omodules fr o m ir reduc ib le W eyl mod ules, [*J. Algebra * ]{} [**414**]{}(20 14), 2 71 -28 7 .R.Lü , K. Z hao, A family o f simple w ei g ht Virasorom odu le s, [*J. Algebr a*]{} | Chen,_J. Han,_Y. Su, A class_of simple_weight_modules over_the_twisted Heisenberg-Virasoro algebra,_[*J. Math. Phys.*]{}_[**57**]{} (2016), 101705.
H. Chen,_X. Guo, A_new_family of modules over the Virasoro algebra, [*J. Algebra*]{} [**457**]{} (2016), 73-105.
H. Chen, X._Guo,_K. Zhao,_Tensor_product_weight modules over the Virasoro_algebra, [*J. Lond. Math. Soc.*]{}_[**88**]{} (2013),_829-844.
C. H. Conley, C. Martin, A family of_irreducible_representations of the_Witt Lie algebra with infinite-dimensional weight spaces, [*Compositio Math.*]{}_[**128**]{} (2001), 153-175.
X. Guo, R. Lü,_K. Zhao, Fraction_representations_and_highest weight-like representations of_the Virasoro algebra, [*J. Algebra*]{} [**387**]{}_(2013), 68-86.
J. Han, Submodule structure of_$\C[s,t]$ over $\mathcal Vir(0,b)$, arXiv:1708.07272.
K. Iohara, Y._Koga, Representation theory of the Virasoro_algebra, Springer Monographs in Mathematics,_2011.
G. Liu,_R. Lü, K. Zhao, A_class of simple_weight Virasoro_modules, [*J. Algebra*]{}_[**424**]{} (2015), 506-521.
G. Liu, Y. Zhao,_Generalized polynomial modules_over the Virasoro algebra, [*Proc. Amer._Math._Soc.*]{} [**144**]{} (2016),_5103-5112.
X._Liu,_X. Guo,_J. Wang, New_irreducible_tensor product_modules_for the Virasoro algebra (II), arXiv:1708.09748.
R._Lü,_X. Guo, K. Zhao, Irreducible modules over_the Virasoro algebra, [*Doc._Math.*]{}_[**16**]{} (2011), 709-721.
R. Lü,_K. Zhao, Irreducible Virasoro modules_from irreducible Weyl modules, [*J. Algebra*]{}_[**414**]{} (2014),_271-287.
R. Lü,_K. Zhao, A family of simple weight Virasoro modules, [*J. Algebra*]{} |
93(01)00918-2) [****, ()](\doibase 10.1016/S0375-9474(02)01402-1) [****, ()](\doibase 10.1103/PhysRevC.88.044001) [****, ()](\doibase 10.1088/0954-3899/43/5/055106) [****, ()](\doibase 10.1016/S0375-9474(02)01270-8) [****, ()](\doibase 10.1016/j.physletb.2003.07.049) [****, ()](\doibase 10.1016/S0375-9474(98)00612-5) @noop [****, ()]{} [****, ()](\doibase 10.1016/j.nuclphysa.2004.08.012) [****, ()](\doibase 10.1016/S0550-3213(97)00178-8) [****, ()](\doibase 10.1103/PhysRevC.97.024001) [****, ()](\doibase 10.1103/PhysRev.71.195) [****, ()](\doibase 10.1103/PhysRev.133.B1575) @noop [****, ()]{} @noop [****, ()]{} [****, ()](\doibase 10.1103/PhysRevLett.82.463) [****, ()](\doibase 10.1016/S0375-9474(98)00650-2) [****, ()](\doibase
10.1103/PhysRevLett.116.112301) [****, ()](\doibase 10.1103/PhysRevLett.118.202501) [****, ()](\doibase 10.1088/0954-3899/42/4/045101)
---
abstract: 'Disease-gene prediction (DGP) refers to the computational challenge of predicting associations between genes and diseases. Effective solutions to the DGP problem have the potential to accelerate the therapeutic development pipeline at early stages via efficient prioritization of candidate genes for various diseases. In this work, we introduce the variational graph auto-encoder (VGAE) as a promising unsupervised approach for learning powerful latent embeddings in disease-gene networks that can be used for the DGP problem, the first approach using a generative model involving graph neural networks (GNNs). In addition to introducing the VGAE as a promising approach to the DGP problem, we further propose an extension (constrained-VGAE or C-VGAE) which adapts the learning algorithm for link prediction between two distinct node types in heterogeneous graphs. We evaluate and demonstrate the effectiveness of the VGAE on general link | 93(01)00918 - 2) [ * * * *, () ] (\doibase 10.1016 / S0375 - 9474(02)01402 - 1) [ * * * *, () ] (\doibase 10.1103 / PhysRevC.88.044001) [ * * * *, () ] (\doibase 10.1088/0954 - 3899/43/5/055106) [ * * * *, () ] (\doibase 10.1016 / S0375 - 9474(02)01270 - 8) [ * * * *, () ] (\doibase 10.1016 / j.physletb.2003.07.049) [ * * * *, () ] (\doibase 10.1016 / S0375 - 9474(98)00612 - 5) @noop [ * * * *, () ] { } [ * * * *, () ] (\doibase 10.1016 / j.nuclphysa.2004.08.012) [ * * * *, () ] (\doibase 10.1016 / S0550 - 3213(97)00178 - 8) [ * * * *, () ] (\doibase 10.1103 / PhysRevC.97.024001) [ * * * *, () ] (\doibase 10.1103 / PhysRev.71.195) [ * * * *, () ] (\doibase 10.1103 / PhysRev.133.B1575) @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } [ * * * *, () ] (\doibase 10.1103 / PhysRevLett.82.463) [ * * * *, () ] (\doibase 10.1016 / S0375 - 9474(98)00650 - 2) [ * * * *, () ] (\doibase
10.1103 / PhysRevLett.116.112301) [ * * * *, () ] (\doibase 10.1103 / PhysRevLett.118.202501) [ * * * *, () ] (\doibase 10.1088/0954 - 3899/42/4/045101)
---
abstract:' Disease - gene prediction (DGP) refers to the computational challenge of predicting associations between genes and disease. Effective solution to the DGP problem have the electric potential to accelerate the therapeutic exploitation pipeline at early stages via effective prioritization of campaigner genes for various diseases. In this work, we introduce the variational graph car - encoder (VGAE) as a promising unsupervised approach for learning powerful latent embeddings in disease - gene network that can be used for the DGP trouble, the first overture use a generative model involving graph neural network (GNNs). In addition to introducing the VGAE as a promising approach to the DGP trouble, we further aim an extension (constrained - VGAE or C - VGAE) which adjust the learning algorithm for link prediction between two distinct node types in heterogenous graph. We measure and demonstrate the effectiveness of the VGAE on general link | 93(01)00918-2) [****, ()](\dlibase 10.1016/S0375-9474(02)01402-1) [****, ()](\doibase 10.1103/PhysReyC.88.044001) [****, ()](\doibase 10.1088/0954-3899/43/5/055106) [****, ()](\donvase 10.1016/S0375-9474(02)01270-8) [****, ()](\doibzse 10.1016/j.phyrletb.2003.07.049) [****, ()](\doibase 10.1016/S0375-9474(98)00612-5) @noop [****, ()]{} [****, ()](\doiuase 10.1016/j.nucophysa.2004.08.012) [****, ()](\doibase 10.1016/S0550-3213(97)00178-8) [****, ()](\doicase 10.1103/PhysGevC.97.024001) [****, ()](\doivase 10.1103/PhysRev.71.195) [****, ()](\doibase 10.1103/PmvsRev.133.G1575) @noo' [****, ()]{} @noop [****, ()]{} [****, ()](\doibsse 10.1103/PhysReeLett.82.463) [****, ()](\doibase 10.1016/S0375-9474(98)00650-2) [****, ()](\boibase
10.1103/PhysRevLett.116.112301) [****, ()](\doibase 10.1103/PhysRedLett.118.202501) [****, ()](\colbase 10.1088/0954-3899/42/4/045101)
---
abstracj: 'Distasq-gens prediction (DGP) refers to the comlutatioial challenge og predicting associations hetwfen genes and disewses. Effectuve filutions to ghe DGP problem have tge potential to accelerate the ghera'eutic deveoopmejj pipeline av earlj stages via cgficiett priotitization of gandivate genes for various diveases. In this worh, we intrmdbce the variational geaph aujo-encmder (CGAD) aa e pdomisijg nnsupervises approach dor learning powerfil oatent embeddjngs ig qisease-gene networks that can be used fmr fhe DGP problem, the firwt approach using a ggnerative iodel involving graph neural networks (GNNs). In addhtion go nkbrodjxijg the VGAE as a promising approach to the DG[ ptonlem, we further iropose an extensipn (cpgstrained-VGAE or C-VYZE) which adapts the pearnind algirithm fow limk prediction between two dustinct node rypes in heterogenzous graphs. Ce evakuate and demonstrate the efyectivsness of thf VGAE on eeneral link | 93(01)00918-2) [****, ()](\doibase 10.1016/S0375-9474(02)01402-1) [****, ()](\doibase 10.1103/PhysRevC.88.044001) 10.1088/0954-3899/43/5/055106) ()](\doibase 10.1016/S0375-9474(02)01270-8) ()](\doibase 10.1016/j.physletb.2003.07.049) [****, [****, 10.1016/j.nuclphysa.2004.08.012) [****, ()](\doibase [****, ()](\doibase 10.1103/PhysRevC.97.024001) ()](\doibase 10.1103/PhysRev.71.195) [****, ()](\doibase 10.1103/PhysRev.133.B1575) @noop ()]{} @noop [****, ()]{} [****, ()](\doibase 10.1103/PhysRevLett.82.463) [****, ()](\doibase 10.1016/S0375-9474(98)00650-2) [****, ()](\doibase 10.1103/PhysRevLett.116.112301) ()](\doibase 10.1103/PhysRevLett.118.202501) [****, ()](\doibase 10.1088/0954-3899/42/4/045101) --- abstract: 'Disease-gene prediction (DGP) refers to the challenge predicting between and diseases. Effective solutions to the DGP problem have the potential to accelerate the therapeutic development at early stages via efficient prioritization of candidate for various diseases. In work, we introduce the variational auto-encoder as a unsupervised for powerful latent embeddings disease-gene networks that can be used for the DGP problem, the first approach using a generative model graph neural In addition introducing VGAE a promising approach DGP problem, we further propose an C-VGAE) which adapts the learning algorithm for link between two node types in heterogeneous graphs. We and demonstrate the effectiveness of the VGAE on link | 93(01)00918-2) [****, ()](\doibase 10.1016/S0375-9474(02)01402-1) [****, ()](\doibase 10.1103/PhysRevC.88.044001) [****, ()](\doIbase 10.1088/0954-3899/43/5/055106) [****, ()](\doibaSe 10.1016/S0375-9474(02)01270-8) [****, ()](\doIbaSe 10.1016/j.PhYsleTb.2003.07.049) [****, ()](\doIbase 10.1016/S0375-9474(98)00612-5) @noop [****, ()]{} [****, ()](\doibASe 10.1016/j.nUclphysa.2004.08.012) [****, ()](\doibase 10.1016/S0550-3213(97)00178-8) [****, ()](\doibasE 10.1103/PhysreVc.97.024001) [****, ()](\doiBAsE 10.1103/Physrev.71.195) [****, ()](\doibASe 10.1103/pHYsREv.133.b1575) @nOop [****, ()]{} @NoOP [****, ()]{} [****, ()](\dOibasE 10.1103/PhYsRevLeTt.82.463) [****, ()](\doibase 10.1016/S0375-9474(98)00650-2) [****, ()](\DoiBaSe
10.1103/PhysRevLetT.116.112301) [****, ()](\DoIbase 10.1103/PhysREvLEtt.118.202501) [****, ()](\doibase 10.1088/0954-3899/42/4/045101)
---
abStrAct: 'DisEaSe-gENe preDicTion (DgP) refeRS to the ComputatiOnAL challENge of prEDIcTing Associations betweEN gENes and diseases. effectIvE SoLUTioNs tO the DGP proBlEm havE The poteNTiAL TO acCElerate the theRapeutic devELopMent piPeLinE At earlY stagEs VIa eFficient priOritIzation of CandidATe genes FOr varioUs diseAseS. In This WOrK, wE inTrODucE ThE vaRIatIonal graPh AuTo-encOder (vgae) As a pRomIsinG unsuPervised approAch For lEArnIng poWerfuL latEnT embeDdings In disEaSe-gene networks tHat cAn be used fOr tHe dGP PrOblem, THe firsT apProAch usinG a generATivE mODEL iNvolving graph neuraL nETWoRks (GNNs). IN additIOn To INtroduciNg The vGAE AS A promIsinG ApProach to The DGP PRoBlEm, we furThEr propOsE an ExtEnsioN (ConsTraineD-VGAE or C-vGAE) wHIch adapts the leARning algorithM FoR LInK PredIctIon between tWo diSTincT nodE TyPes IN heteRogenEoUS gRAphs. We evaluate and deMoNstratE the eFfectiveness oF the VGAE on GENEral link | 93(01)00918-2) [****, ()]( \doibase 1 0.101 6/S 037 5- 9474 (02) 01402-1) [**** , () ](\doibase 10.1103/Phy sRevC .8 8 .044 0 01 ) [** **, () ] (\ d o iba se 1 0.1 08 8 /0 954-3 899 /43/5/0 55106) [** **, ()](\doibase 10 .1016/S037 5-9 474(02)01270 -8) [**** ,()] ( \doib ase 10.1 016/j. p hyslet b.2003.07 .0 4 9) [** * *, ()] ( \ do ibas e 10.1016/S0375-9 4 74 ( 98)00612-5) @n oop [* ** * ,( )]{ } [ ****, ()] (\ doiba s e 10.10 1 6/ j . n ucl p hysa.2004.08. 012) [****, () ](\doi ba se1 0.1016 /S055 0- 3 213 (97)00178-8 ) [* ***, ()]( \doiba s e 10.11 0 3/PhysR evC.97 .02 400 1) [ * ** *, () ]( \ doi b as e 1 0 .11 03/PhysR ev .7 1.195 ) [* * * * , ()] (\d oiba se 10 .1103/PhysRev .13 3.B1 5 75) @noo p [** **, ( )]{}@noop[**** , ()]{} [****, ()]( \doibase10. 11 03/ Ph ysRev L ett.82 .46 3)[****,()](\do i bas e1 0 . 10 16/S0375-9474(98)0 06 5 0 -2 ) [****, ()]( \ do ib a se
10. 11 03/ Phys R e vLett .116 . 11 2301) [* ***, ( )] (\ doibase 1 0.1103 /P hys Rev Lett. 1 18.2 02501) [****, ()]( \ doibase 10.108 8 /0954-3899/42 / 4/ 0 4 51 0 1)
---
abstract:'Dis e ase- gene pr edi c tion(DGP) r e fe r s to the computatio na l chal lenge of predictin g associat i o n s betwee n ge n es and diseases.Effec tive solut i ons to t he DG P proble m have th e potentia l t o a cce ler a t ethe therapeut i c dev el opmentpip eline a t e arl y s tag es via effi cient pr io ri ti za tio n ofc andidate g ene sfor vari o us dis eases . In t hi s wo rk, wei nt r o duce t he var iat io nal g raph aut o-encod er (VGAE) as a pr om is ing uns upervised app ro ach for le ar nin g powe r f ul laten t embeddings in disease - gene ne two rks t hatcan be us edfor th e D G P prob lem, t he fi rs t a p p roach u si ngagenerative m ode l inv ol ving graphneural networks (G N Ns) . In addition to int r o du cin g t h e V GA E as a promising appr oach to th eD GP problem,w e f ur ther pr opose a n ext e nsion ( constrain ed-VGAE o rC-VG A E ) w hich adapt s the le arning al g orith m f or li nkpredic ti onbetwe en two dis tinct nodety pes in hete ro geneousgraphs. We evaluate and demon strat e t he effect ive n ess of the V GAEon general li nk | 93(01)00918-2) [****,_()](\doibase 10.1016/S0375-9474(02)01402-1)_[****, ()](\doibase 10.1103/PhysRevC.88.044001) [****, _()](\doibase 10.1088/0954-3899/43/5/055106)_[****, _()](\doibase 10.1016/S0375-9474(02)01270-8)_[****,_()](\doibase 10.1016/j.physletb.2003.07.049) [****, _()](\doibase 10.1016/S0375-9474(98)00612-5) @noop_[****, ()]{} [****, ()](\doibase_10.1016/j.nuclphysa.2004.08.012) [****, ()](\doibase_10.1016/S0550-3213(97)00178-8)_[****, ()](\doibase 10.1103/PhysRevC.97.024001) [****, ()](\doibase 10.1103/PhysRev.71.195) [****, ()](\doibase 10.1103/PhysRev.133.B1575) @noop [****, ()]{} @noop [****, _()]{}_[****, ()](\doibase_10.1103/PhysRevLett.82.463)_[****,_()](\doibase 10.1016/S0375-9474(98)00650-2) [****, ()](\doibase
_10.1103/PhysRevLett.116.112301) [****, ()](\doibase 10.1103/PhysRevLett.118.202501) [****, _()](\doibase 10.1088/0954-3899/42/4/045101)
_---
abstract: 'Disease-gene prediction (DGP) refers to the computational_challenge_of predicting associations_between genes and diseases. Effective solutions to the DGP_problem have the potential to accelerate_the therapeutic development_pipeline_at_early stages via efficient_prioritization of candidate genes for various_diseases. In this work, we introduce_the variational graph auto-encoder (VGAE) as a_promising unsupervised approach for learning powerful_latent embeddings in disease-gene networks_that can_be used for the DGP_problem, the first_approach using_a generative model_involving graph neural networks (GNNs). In_addition to introducing_the VGAE as a promising approach_to_the DGP problem,_we_further_propose an_extension (constrained-VGAE or_C-VGAE)_which adapts_the_learning algorithm for link prediction between_two_distinct node types in heterogeneous graphs. We_evaluate and demonstrate the_effectiveness_of the VGAE on_general link |
2$-AGL rings for the further studies, we investigate three topics on $2$-AGL rings, which are closely studied already for the case of AGL rings. The first topic concerns minimal presentations of canonical ideals. In Section 2, we will give a necessary and sufficient condition for a given one-dimensional Cohen-Macaulay local ring $R$ to be a $2$-AGL ring, in terms of minimal presentations of canonical fractional ideals. Our results Theorems \[3.2\] and \[3.4a\] exactly correspond to those about AGL rings given by [@GTT Theorem 7.8].
In Section 3, we investigate a generalization of so called [*amalgamated duplications*]{} of $R$ ([@marco]), including certain fiber products, and prove that $R$ is a $2$-$\AGL$ ring if and only if so is the fiber product $R \times_{R/{\mathfrak{c}}} R$. By [@CGKM Theorem 4.2] $R$ is a $2$-$\AGL$ ring if and only if so is the trivial extension $R \ltimes {\mathfrak{c}}$ of ${\mathfrak{c}}$ over $R$, which corresponds to [@GMP Theotem 6.5] for the case of AGL rings.
In Sections 4 and 5, we are interested in Ulrich ideals in $2$-AGL rings. The existence of two-generated Ulrich ideals is basically a substantially strong condition for $R$, which we closely discuss in Section 4, especially in the case where $R$ is a $2$-AGL ring. Here, we should not rush, but should explain about what are Ulrich ideals. The notion of Ulrich ideal/module dates back to the work [@GOTWY] in 2014, where the authors introduced the notion, generalizing that of MGMCM modules (maximally generated maximal Cohen-Macaulay modules) ([@BHU]), and started the basic theory. The maximal ideal of a Cohen-Macaulay local ring with minimal multiplicity is a typical example of Ulrich ideals, and the higher syzygy modules of Ulrich ideals are Ulrich modules. In [@GOTWY; @GOTWY2], all the Ulrich ideals of Gorenstein local rings of finite CM-representation type and of dimension at most $2$ are determined, by means of the classification in the representation theory. On the other hand, in [@ | 2$-AGL rings for the further studies, we investigate three topics on $ 2$-AGL rings, which are close study already for the case of AGL rings. The inaugural subject concerns minimal presentations of canonical ideal. In Section 2, we will give a necessary and sufficient circumstance for a give one - dimensional Cohen - Macaulay local ring $ R$ to be a $ 2$-AGL band, in terms of minimal presentations of canonic fractional ideal. Our results Theorems \[3.2\ ] and \[3.4a\ ] exactly correspond to those about AGL rings given by [ @GTT Theorem 7.8 ].
In part 3, we investigate a generalization of so called [ * amalgamate duplications * ] { } of $ R$ ([ @marco ]), including sealed fiber intersection, and prove that $ R$ is a $ 2$-$\AGL$ band if and only if so is the fiber merchandise $ R \times_{R/{\mathfrak{c } } } R$. By [ @CGKM Theorem 4.2 ] $ R$ is a $ 2$-$\AGL$ ring if and only if so is the trivial extension $ R \ltimes { \mathfrak{c}}$ of $ { \mathfrak{c}}$ over $ R$, which corresponds to [ @GMP Theotem 6.5 ] for the case of AGL rings.
In Sections 4 and 5, we are interested in Ulrich ideals in $ 2$-AGL rings. The existence of two - generated Ulrich ideals is basically a well potent condition for $ R$, which we close discourse in incision 4, especially in the case where $ R$ is a $ 2$-AGL ring. Here, we should not rush, but should explain about what are Ulrich ideals. The notion of Ulrich ideal / faculty dates back to the work [ @GOTWY ] in 2014, where the authors introduced the notion, generalizing that of MGMCM modules (maximally generated maximal Cohen - Macaulay modules) ([ @BHU ]), and started the basic theory. The maximal ideal of a Cohen - Macaulay local ring with minimal numerousness is a typical exercise of Ulrich ideal, and the higher syzygy modules of Ulrich ideals are Ulrich modules. In [ @GOTWY; @GOTWY2 ], all the Ulrich ideal of Gorenstein local rings of finite CM - representation type and of property at most $ 2 $ are determined, by means of the classification in the theatrical performance theory. On the other hand, in [ @ | 2$-AGL rings for the further suudies, we investiyqte thcee topjcs on $2$-AEL rings, which are closely svudiwd aleeady for the case of XGL rings. The firwt tipic conceria minimal preacntatnois of canonical ideals. In Section 2, we whlu yive a necessary and sufficient condytion fpr a given one-diienspogal Dohen-Macaulay local ring $R$ to be a $2$-AGL riig, in terms of kinimal presentations of cwnonlcal fractional idfals. Our rewultf Theorems \[3.2\] avd \[3.4a\] exactly correspons to those about AGL rings givev by [@YTT Theorem 7.8].
Ib Sfwtion 3, we iivestidate a generalizatiot of so called [*amalgakatxd dyplications*]{} of $R$ ([@marcm]), including certaig fiber psobucts, and prove that $E$ us a $2$-$\DGL$ sing uf xnd oily if so is the fiber product $R \rimes_{R/{\mathfrak{c}}} R$. Bu [@STKM Theorem 4.2] $D$ is a $2$-$\ADL$ ring if and only if so is the trivian estension $R \ltimes {\mathfeak{c}}$ of ${\mathfrak{c}}$ ovet $R$, which sorresponds to [@GMP Theotem 6.5] for the case of AGL rhngs.
Ii Recuiins 4 xbd 5, we are interested in Ulrich ideals in $2$-AGL ryhgx. Nhe existence of bwo-generated Ulricn ldrwls is basicauly a subatantially strong fonditijn foe $R$, which we vlosely discuss in Section 4, especially pn tye case where $R$ is a $2$-AGL ring. Herg, we snould not rush, but shouud esplain abouh what ars Ulrich ideals. Tfe kothon of Ulrich ideal/module qates bacj to the wofk [@GPTWY] ig 2014, where tje aubvors introduced thf notnon, ganeralizinh that of MGMCM modules (maximally generated msxhman Cohen-Mccaulan modules) ([@BHU]), agd started the basic cheory. The maximzl idean of a Coheg-Macaulay locdp ring with kinimal iultuplixity is x typical examlle of Ulgieh ideals, and the higher sydygy okdules of Ulrick udeals are Ulrivh oodtlvs. Mn [@GOEFY; @GOTWY2], all the Ulfoch iaeals of Govenrteim local rings of finhte DM-representation tupc and of eimensiog at most $2$ arr determined, by mewns oh the rlassigicwtion in the representation thsory. On tje jther hand, ig [@ | 2$-AGL rings for the further studies, we topics $2$-AGL rings, are closely studied AGL The first topic minimal presentations of ideals. In Section 2, we will a necessary and sufficient condition for a given one-dimensional Cohen-Macaulay local ring $R$ be a $2$-AGL ring, in terms of minimal presentations of canonical fractional ideals. results \[3.2\] \[3.4a\] correspond to those about AGL rings given by [@GTT Theorem 7.8]. In Section 3, we investigate generalization of so called [*amalgamated duplications*]{} of $R$ including certain fiber products, prove that $R$ is a ring and only so the product $R \times_{R/{\mathfrak{c}}} By [@CGKM Theorem 4.2] $R$ is a $2$-$\AGL$ ring if and only if so is the trivial $R \ltimes ${\mathfrak{c}}$ over which to Theotem 6.5] for of AGL rings. In Sections 4 are interested in Ulrich ideals in $2$-AGL rings. existence of Ulrich ideals is basically a substantially condition for $R$, which we closely discuss in 4, especially in the case where $R$ is a $2$-AGL ring. Here, we should not should explain about what Ulrich ideals. The of ideal/module back the work in 2014, where the authors introduced the notion, generalizing that of modules (maximally generated maximal Cohen-Macaulay modules) ([@BHU]), and started the The ideal of a local ring with minimal is typical example of Ulrich the syzygy ideals Ulrich In [@GOTWY; @GOTWY2], all Ulrich ideals of Gorenstein local of finite CM-representation type $2$ are determined, by means of the classification the representation theory. On the other hand, [@ | 2$-AGL rings for the further studIes, we invesTigatE thRee ToPics On $2$-AGl rings, which are CLoseLy studied already for the Case oF Agl rinGS. THe firSt topic COnCERns MiNiMal PrESeNtatiOns Of canonIcal ideals. in SEcTion 2, we will giVE a Necessary aNd sUfficient conDitIon for A gIveN One-diMenSionaL Cohen-mAcaulaY local rinG $R$ TO be a $2$-AGl Ring, in tERMs Of miNimal presentationS Of CAnonical fractiOnal idEaLS. OUR ResUltS Theorems \[3.2\] aNd \[3.4A\] exacTLy correSPoND TO thOSe about AGL rinGs given by [@GTt theOrem 7.8].
In seCtiON 3, we invEstigAtE A geNeralizatioN of sO called [*amAlgamaTEd dupliCAtions*]{} oF $R$ ([@marcO]), inCluDing CErTaIn fIbER prODuCts, ANd pRove that $r$ iS a $2$-$\aGL$ riNg if AND ONly iF so Is thE fibeR product $R \timeS_{R/{\mAthfRAk{c}}} r$. By [@CGkM TheOrem 4.2] $r$ iS a $2$-$\AGL$ Ring if And onLy If so is the triviaL extEnsion $R \ltImeS {\mAthFrAk{c}}$ of ${\MAthfraK{c}}$ oVer $r$, which cOrrespoNDs tO [@Gmp tHeOtem 6.5] for the case of AGl rINGs.
in SectioNs 4 and 5, wE ArE iNTerested In ulrIch iDEAls in $2$-aGL rINgS. The exisTence oF TwO-gEnerateD ULrich iDeAls Is bAsicaLLy a sUbstanTially stRong cONdition for $R$, whiCH we closely disCUsS IN SECtioN 4, esPecially in tHe caSE wheRe $R$ iS A $2$-AgL rINg. HerE, we shOuLD nOT rush, but should explaIn About wHat arE Ulrich ideals. the notion oF uLRich ideaL/modULe DAtes back to the wOrk [@GOtWY] in 2014, where THe authorS intrOduced thE notion, geNERalizing ThaT of mGMcM mODUlEs (maximally geNERateD mAximal COheN-MacaulAy mOduLes) ([@bHU]), AnD started tHe basic tHeOrY. THe MaxImal iDEal of a CoHeN-MaCaUlaY locaL Ring wiTh minImal MuLtIPliCity is a TYpICAl exAmPlE of ULriCh IdealS, and THe hIgher syZygy modulEs oF ulriCh IdEals are ulrich modules. in [@gOTWY; @GOTWY2], AlL thE UlricH IDeals of GOrenstein local rings of fiNIte CM-rePreSentaTion Type and of DimEnsion At mOSt $2$ are dEtermiNed, by MeAns OF The clASSiFicAtIon in the rePREseNtatiOn TheoRy. On the Other hand, in [@ | 2$-AGL rings for the furth er studies , weinv est ig atethre e topics on $2 $ -AGL rings, which are clos ely s tu d ieda lr eadyfor the ca s e of A GL ri ng s .The f irs t topic concernsmin im al presentat i on s of canon ica l ideals. In Se ction2, we willgiv e a n ecessa r y andsufficien tc onditi o n for a g iv en o ne-dimensional Co h en - Macaulay local ring$R $ t o bea $ 2$-AGL rin g, in t e rms ofm in i m a l p r esentations o f canonical fra ctiona lide a ls. Ou r res ul t s T heorems \[3 .2\] and \[3. 4a\] e x actly c o rrespon d to t hos e a bout AG Lrin gs giv e nby[ @GT T Theore m7. 8].
In S e c t i on 3 , w e in vesti gate a genera liz atio n of so c alled [*a ma lgama ted du plica ti ons*]{} of $R$([@m arco]), i ncl ud ing c ertai n fiber pr odu cts, an d prove tha t$ R $ i s a $2$-$\AGL$ rin gi f a nd onlyif soi sth e fiber p ro duc t $R \ times _{R/ { \m athfrak{ c}}} R $ .By [@CGKM T heorem 4 .2] $R $ isa $2$ -$\AGL $ ring i f and only if so ist he trivial ex t en s i on $R \ lti mes {\mathf rak{ c }}$of $ { \m ath f rak{c }}$ o ve r $ R $, which correspond sto [@G MP Th eotem 6.5] fo r the case o f AGL rin gs.In Sections 4 and 5, w e are inte r ested in Ulri ch ideal s in $2$- A G L rings. Th e e xis ten c e o f two-generat e d Ulr ic h ideal s i s basic all y a su bst an tially st rong con di ti on f or$R$,w hich wecl ose ly di scuss in Sec tion4, e sp ec i all y in th e c a s e wh er e$R$isa$2$-A GL r i ng. Here,we should no t rus h, b ut shou ld explain ab ou t what are U lri ch ide a l s. The n otion of Ulrich ideal/m o dule da tes back tothe work[@G OTWY]in2 014, w here t he au th ors i ntrod u c ed th enotion, ge n e ral izing t hatof MGMC M modules (maximal l y g enerated maxi mal Coh e n -M aca u la y mo du l es) ( [@BHU]), and st arted theba s ic theory. T h e m ax imal id eal ofa Coh e n-Macau lay local ring wit hmini m a l m ultiplicit y is a t ypical ex a mpleo fUlric h i deals, a ndthe h ighers yzy gy mo dulesof Ulric h ide al s are Ul rich modules. In [@GOTW Y; @GO TWY2] , a ll the Ul ric h id eals of G oren stein loca l r ing s offin i te CM -rep r es ent a tiontype and of di m en sio n at most $2$ a r e det ermin ed, by mea ns o f the classificat i on in the repr esen t a tio n t h eory .On the other h and ,i n [@ | 2$-AGL rings_for the_further studies, we investigate_three topics_on_$2$-AGL rings,_which_are closely studied_already for the_case of AGL rings._The first topic_concerns_minimal presentations of canonical ideals. In Section 2, we will give a necessary and_sufficient_condition for_a_given_one-dimensional Cohen-Macaulay local ring $R$_to be a $2$-AGL ring,_in terms_of minimal presentations of canonical fractional ideals. Our_results_Theorems \[3.2\] and_\[3.4a\] exactly correspond to those about AGL rings given_by [@GTT Theorem 7.8].
In Section 3,_we investigate a_generalization_of_so called [*amalgamated duplications*]{}_of $R$ ([@marco]), including certain fiber_products, and prove that $R$ is_a $2$-$\AGL$ ring if and only if_so is the fiber product $R_\times_{R/{\mathfrak{c}}} R$. By [@CGKM Theorem_4.2] $R$_is a $2$-$\AGL$ ring if_and only if_so is_the trivial extension_$R \ltimes {\mathfrak{c}}$ of ${\mathfrak{c}}$ over_$R$, which corresponds_to [@GMP Theotem 6.5] for the_case_of AGL rings.
In_Sections_4_and 5,_we are interested_in_Ulrich ideals_in_$2$-AGL rings. The existence of two-generated_Ulrich_ideals is basically a substantially strong condition_for $R$, which we_closely_discuss in Section 4,_especially in the case where_$R$ is a $2$-AGL ring. Here,_we should_not rush,_but should explain about what are Ulrich ideals. The notion of_Ulrich ideal/module dates back to the_work [@GOTWY] in 2014,_where the_authors_introduced the notion,_generalizing_that of_MGMCM modules (maximally generated maximal Cohen-Macaulay modules)_([@BHU]), and_started the basic theory. The maximal_ideal of a Cohen-Macaulay_local_ring with minimal multiplicity is a_typical example of Ulrich ideals, and_the higher syzygy modules of_Ulrich_ideals_are Ulrich modules. In [@GOTWY;_@GOTWY2], all the Ulrich ideals of_Gorenstein local rings_of finite CM-representation type and of dimension_at_most $2$ are determined, by means_of_the classification in the representation theory._On_the_other hand, in [@ |
the abundance data (cf. Copi, Schramm, & Turner 1995b). The nature of such errors is unclear, and this remains controversial. Other authors have reacted to the impending crisis in self-consistency by simply omitting one or more of the four nuclides in making the fit (Dar 1995; Olive & Thomas 1997; Hata et al. 1996, 1997; Fields et al. 1996).
This controversy has been sharpened by new observations giving the deuterium abundances on various lines of sight to high-redshift QSOs. These data should yield the primordial D abundance, but current results span an order of magnitude. The low values, D/H by number $\approx 2 \times 10^{-5}$ (Tytler, Fan, & Burles 1996; Burles & Tytler 1996), corresponding to $\eta_{10} \approx 7$ in the standard model, have been revised slightly upward \[D/H $\approx (3-4) \times
10^{-5}$ (Burles & Tytler 1997a,b,c); $\eta_{10} \approx 5$\], but it still seems impossible to reconcile the inferred abundance of $^4$He \[Y$_{\rm P} \approx 0.234$; Olive & Steigman 1995 (OS)\] with standard BBN for this large value of $\eta_{10}$ (which implies Y$_{\rm BBN}
\approx 0.247$) unless there are large systematic errors in the $^4$He data (cf. Izotov, Thuan, & Lipovetsky 1994, 1997). Such low D/H values have also been challenged on observational grounds by Wampler (1996) and by Songaila, Wampler, and Cowie (1997), and deuterium abundances nearly an order of magnitude higher, D/H $\approx 2\times10^{-4}$, have been claimed by Carswell et al. (1994), Songaila et al. (1994), and Rugers and Hogan (1996) for other high-redshift systems with metal abundances equally close to primordial. Although some of these claims of high deuterium have been called into question (Tytler, Burles, & Kirkman 1997), Hogan (1997) and Songaila (1997) argue that the spectra of other absorbing systems require high D/H (e.g., Webb et al. 1997). If these higher abundances are correct, then D and | the abundance data (cf. Copi, Schramm, & Turner 1995b). The nature of such errors is unclear, and this remain controversial. early authors have reacted to the at hand crisis in self - consistency by simply neglect one or more of the four nuclides in making the fit (Dar 1995; Olive & Thomas 1997; Hata et al. 1996, 1997; Fields et al. 1996).
This controversy has been sharpened by fresh observations giving the deuterium abundance on various lines of batch to high - redshift QSOs. These datum should yield the primordial D abundance, but current results span an order of magnitude. The abject values, D / H by number $ \approx 2 \times 10^{-5}$ (Tytler, Fan, & Burles 1996; Burles & Tytler 1996), corresponding to $ \eta_{10 } \approx 7 $ in the standard mannequin, have been revised slightly upward \[D / H $ \approx (3 - 4) \times
10^{-5}$ (Burles & Tytler 1997a, b, c); $ \eta_{10 } \approx 5$\ ], but it however seems impossible to reconcile the inferred abundance of $ ^4$He \[Y$_{\rm P } \approx 0.234 $; Olive & Steigman 1995 (OS)\ ] with standard BBN for this large value of $ \eta_{10}$ (which implies Y$_{\rm BBN }
\approx 0.247 $) unless there be large systematic errors in the $ ^4$He data (cf. Izotov, Thuan, & Lipovetsky 1994, 1997). Such broken five hundred / H values have also been challenged on observational grounds by Wampler (1996) and by Songaila, Wampler, and Cowie (1997), and deuterium abundances nearly an order of order of magnitude higher, D / H $ \approx 2\times10^{-4}$, have been claim by Carswell et al. (1994), Songaila et al. (1994), and Rugers and Hogan (1996) for other gamey - red shift system with metal abundances evenly close to primordial. Although some of these claim of high deuterium have been called into doubt (Tytler, Burles, & Kirkman 1997), Hogan (1997) and Songaila (1997) argue that the spectra of other absorbing systems require high D / H (e.g., Webb et al. 1997). If these higher abundances are correct, then D and | thf abundance data (cf. Copi, Schramm, & Turnet 1995v). The iature kf such drrors is unclear, and this rxmaibs cobtroversial. Other authurs have geacted ti tht impending crisis in seln-eonsiabency uy simply omittlng one or kore of the fogr nbclides in making the fit (Dar 1995; Olive & Thomax 1997; Hata et al. 1996, 1997; Sielcf et al. 1996).
This controversy has been sharlened bj new observationx giving the deuterium abujdanfes on various linfs of sight to rugh-redshift DSOs. These data should yield the primordial D abundanze, buc current rgshlhv span an ocder os magnitude. Bne low values, D/H by number $\sppcox 2 \times 10^{-5}$ (Tytler, Fan, & Bnrles 1996; Burles & Tytlet 1996), correspmnbing to $\eta_{10} \approx 7$ ib rhe sjandasd mueel, habe bsen regisxd slightly upward \[D/H $\qpprox (3-4) \times
10^{-5}$ (Burkef & Tytler 1997a,b,c); $\sta_{10} \ap[rjx 5$\], but it still seems impossible to reboncjle the inferred abundabce of $^4$He \[Y$_{\rm P} \approd 0.234$; Olive & Steigman 1995 (OS)\] with standard BBN for this large vanue oh $\dta_{10}$ (wmich umolies Y$_{\rm BBN}
\approx 0.247$) unless there are large ayxtvmatic errors in bhe $^4$He data (cf. Izoyog, Yruan, & Lipovetrky 1994, 1997). Sucg low D/H values hage also been challengtd on observational grounds by Wqmpler (1996) and yy Wongaila, Wampler, aud Cowie (1997), anb deutgrium sbundances nearly an orber of magnitude jigher, D/H $\xpprox 2\times10^{-4}$, havd bven wlaimed by Carswell et al. (1994), Songaila et cl. (1994), and Fugets and Rogan (1996) for othev high-redshift systfms wnth matal abundwnces equally close to primordiek. Although spma ox these elaims of high deutqrium have beeu called into question (Nytler, Bucles, & Kirkmag 1997), Hogan (1997) and Dongaila (1997) arjue that ehe wpecrra of ugher absorbing systems gezuirw high D/H (e.g., Webb ct al. 1997). If these highex qbundances are vorfece, nhei D agg | the abundance data (cf. Copi, Schramm, & The of such is unclear, and have to the impending in self-consistency by omitting one or more of the nuclides in making the fit (Dar 1995; Olive & Thomas 1997; Hata et 1996, 1997; Fields et al. 1996). This controversy has been sharpened by new giving deuterium on lines of sight to high-redshift QSOs. These data should yield the primordial D abundance, but current span an order of magnitude. The low values, by number $\approx 2 10^{-5}$ (Tytler, Fan, & Burles Burles Tytler 1996), to \approx in the standard have been revised slightly upward \[D/H $\approx (3-4) \times 10^{-5}$ (Burles & Tytler 1997a,b,c); $\eta_{10} \approx 5$\], it still to reconcile inferred of \[Y$_{\rm P} \approx & Steigman 1995 (OS)\] with standard large value of $\eta_{10}$ (which implies Y$_{\rm BBN} 0.247$) unless are large systematic errors in the data (cf. Izotov, Thuan, & Lipovetsky 1994, 1997). low D/H values have also been challenged on observational grounds by Wampler (1996) and by and Cowie (1997), and abundances nearly an of higher, $\approx have been by Carswell et al. (1994), Songaila et al. (1994), and Rugers Hogan (1996) for other high-redshift systems with metal abundances equally primordial. some of these of high deuterium have called question (Tytler, Burles, & Hogan and that spectra other absorbing systems require D/H (e.g., Webb et al. If these higher abundances | the abundance data (cf. Copi, SchRamm, & Turner 1995B). The nAtuRe oF sUch eRrorS is unclear, and tHIs reMains controversial. OtheR authOrS Have REaCted tO the impENdING crIsIs In sElF-CoNsistEncY by simpLy omitting One Or More of the fouR NuClides in maKinG the fit (Dar 1995; OlIve & thomas 1997; haTa eT Al. 1996, 1997; FieLds Et al. 1996).
THis conTRoversY has been sHaRPened bY New obseRVAtIons Giving the deuteriuM AbUNdances on varioUs lineS oF SiGHT to HigH-redshift QsOS. ThesE Data shoULd YIELd tHE primordial D aBundance, but CUrrEnt resUlTs sPAn an orDer of MaGNitUde. The low vaLues, d/H by numbeR $\approX 2 \Times 10^{-5}$ (TyTLer, Fan, & BUrles 1996; BUrlEs & TYtleR 1996), CoRrEspOnDIng TO $\eTa_{10} \aPProX 7$ in the stAnDaRd modEl, haVE BEEn reVisEd slIghtlY upward \[D/H $\apprOx (3-4) \tImes
10^{-5}$ (bUrlEs & TytLer 1997a,b,C); $\eta_{10} \ApProx 5$\], bUt it stIll seEmS impossible to reConcIle the infErrEd AbuNdAnce oF $^4$he \[Y$_{\rm P} \AppRox 0.234$; olive & StEigman 1995 (Os)\] WitH sTANDaRd BBN for this large vAlUE Of $\Eta_{10}$ (which ImplieS y$_{\rM Bbn}
\approx 0.247$) uNlEss TherE ARe larGe sySTeMatic errOrs in tHE $^4$HE dAta (cf. IzOtOv, ThuaN, & LIpoVetSky 1994, 1997). SuCH low d/H valuEs have alSo beeN Challenged on obSErvational groUNdS BY WAMpleR (1996) anD by Songaila, wampLEr, anD CowIE (1997), aNd dEUteriUm abuNdANcES nearly an order of magNiTude hiGher, D/h $\approx 2\times10^{-4}$, hAve been claIMED by CarswEll eT Al. (1994), sOngaila et al. (1994), and rugerS and Hogan (1996) fOR other hiGh-redShift sysTems with mETAl abundaNceS eqUalLy cLOSe To primordial. ALTHougH sOme of thEse Claims oF hiGh dEutEriUm Have been cAlled intO qUeStIoN (TyTler, BURles, & KirkMaN 1997), HoGaN (1997) anD SongAIla (1997) argUe thaT the SpEcTRa oF other aBSoRBIng sYsTeMs reQuiRe High D/h (e.g., WEBb eT al. 1997). If thEse higher AbuNDancEs ArE correcT, then D and | the abundance data (cf. C opi, Schra mm, & Tu rne r1995 b).The nature ofs ucherrors is unclear, and this r e main s c ontro versial . O t h erau th ors h a ve reac ted to the impending cr is is in self-c o ns istency by si mply omittin g o ne ormo reo f the fo ur nu clides in mak ing the f it (Dar 1 9 95; Oli v e & Tho mas 1997; Hata et al . 1996, 1997; F ieldset al . 199 6).
This con tr overs y has be e ns h a rpe n ed by new obs ervations g i vin g thede ute r ium ab undan ce s on various li nesof sightto hig h -redshi f t QSOs. These da tashou l dyi eld t h e p r im ord i alD abunda nc e, butcurr e n t resu lts spa n anorder of magn itu de.T helow v alues , D/ Hby nu mber $ \appr ox 2 \times 10^{- 5}$(Tytler,Fan ,& B ur les 1 9 96; Bu rle s & Tytler 1996), cor re s p o nd ing to $\eta_{10}\a p p ro x 7$ inthe st a nd ar d model,ha vebeen r evise d sl i gh tly upwa rd \[D / H$\ approx(3 -4) \t im es
1 0^{-5 } $ (B urles& Tytler 1997 a ,b,c); $\eta_{ 1 0} \approx 5$ \ ], b ut it s til l seems imp ossi b le t o re c on cil e theinfer re d a b undance of $^4$He \ [Y $_{\rm P} \ approx 0.234$ ; Olive &S t e igman 19 95 ( O S) \ ] with standar d BBN for thisl arge val ue of $\eta_{ 10}$ (whi c h implies Y$ _{\ rmBBN } \approx 0.247 $ ) unl es s there ar e large sy ste mat icer rors in t he $^4$H eda ta ( cf. Izot o v, Thuan ,& L ip ove tsky1 994, 1 997). Suc hlo w D/ H value s h a v e al so b eencha ll enged ono bse rvation al ground s b y Wam pl er (1996) and by Songa il a, Wampler ,and Cowie ( 1997), a nd deuterium abundances nearlyanorder ofmagnitude hi gher,D/H $\appr ox 2\t imes1 0^ {-4 } $ , hav e be encl aimed by C a r swe ll et a l. ( 1994),Songaila et al. (1 9 94) , and Rugersand Hog a n ( 199 6 )f orot h erh i gh-redshift sys tems withme t al abundance s eq ua lly clo se to p rimor d ial. Al though so me of the se cla i m s o f high deu terium h ave beenc alled in to qu est ion (T yt ler , Bur les, & Kir kman1997), H ogan ( 1997) a nd Songa ila (1997) argue that t he spe ctraofother abs orb i ngsystems r equi re high D/ H ( e.g ., We bbe t al. 199 7 ). If these hig h er abunda n ce s a r e c orrect, the n D an d | the_abundance data_(cf. Copi, Schramm, &_Turner 1995b)._The_nature of_such_errors is unclear,_and this remains_controversial. Other authors have_reacted to the_impending_crisis in self-consistency by simply omitting one or more of the four nuclides in_making_the fit_(Dar_1995;_Olive & Thomas 1997; Hata_et al. 1996, 1997; Fields_et al._1996).
This controversy has been sharpened by new observations_giving_the deuterium abundances_on various lines of sight to high-redshift QSOs. These_data should yield the primordial D_abundance, but current_results_span_an order of magnitude._The low values, D/H by number_$\approx 2 \times 10^{-5}$ (Tytler, Fan,_& Burles 1996; Burles & Tytler 1996),_corresponding to $\eta_{10} \approx 7$ in_the standard model, have been_revised slightly_upward \[D/H $\approx (3-4) \times_
10^{-5}$ (Burles_& Tytler_1997a,b,c); $\eta_{10} \approx_5$\], but it still seems impossible_to reconcile the_inferred abundance of $^4$He \[Y$_{\rm P}_\approx_0.234$; Olive &_Steigman_1995_(OS)\] with_standard BBN for_this_large value_of_$\eta_{10}$ (which implies Y$_{\rm BBN} _
\approx_0.247$) unless there are large systematic errors_in the $^4$He data_(cf._Izotov, Thuan, & Lipovetsky_1994, 1997). Such low D/H_values have also been challenged on_observational grounds_by Wampler_(1996) and by Songaila, Wampler, and Cowie (1997), and deuterium abundances_nearly an order of magnitude higher,_D/H $\approx 2\times10^{-4}$, have_been claimed_by_Carswell et al._(1994),_Songaila et_al. (1994), and Rugers and Hogan (1996)_for other_high-redshift systems with metal abundances equally_close to primordial. Although_some_of these claims of high deuterium_have been called into question (Tytler,_Burles, & Kirkman 1997), Hogan_(1997)_and_Songaila (1997) argue that the_spectra of other absorbing systems require_high D/H (e.g.,_Webb et al. 1997). If these higher_abundances_are correct, then D and |
pm$ 0.030 0.815$\pm$ 0.024 0.881$\pm$ 0.020
(b)\[$50$/$\sigma$\] $1$ 0.875$\pm$ 0.020 0.558$\pm$ 0.031 0.814$\pm$ 0.024 0.907 $\pm$ 0.018
(c)\[$20$/$\sigma$\] $15$ 0.875$\pm$ 0.020 0.608$\pm$ 0.030 0.769$\pm$ 0.026 0.893$\pm$ 0.019
(d)\[$20$/$\sigma$\] $1$ 0.900$\pm$ 0.019 0.602$\pm$ 0.030 0.793$\pm$ 0.025 0.892$\pm$ 0.019
: Validity frequencies \[with precision $\pm 95\%$\] of the CoRP, CoLP, CoRLaP, CENeP, the Early-Stopped CoLP and the $2$-PN CoLP based on $1000$ replications.[]{data-label="tab:frqVSsigma"}
-0.1in
Validity.
: Now, we consider the validity of the selected predictors (cf. **Step 3** in Algorithm \[alg:ConformalLasso\]). As shown in Table \[tab:frqVSsigma\], we observe that variations on the noise level, the variables correlations and the sparsity of the model do to not perturb the validity whereas the sample size relatively to the dimension $p$ does. When $n=300>p$, all the procedures seem to be quite similar and produce good predictors. In the other cases, i.e., when $n=p=50$ and $n=20<p$, the selected confidence predictors have worst performance than expected (validity with smaller proportion than $1-\varepsilon = 90\%$). Moreover, Sparse Confidence Predictors perform worst than the CoRP as observed in Table \[tab:frqVSsigma\]. As pointed in the accuracy part, one explication can be observed in Figure \[fig:plotslength\] as the selected predictor which also is not valid (iteration $93$) corresponds to an iteration in the unstable zone (that is, after iteration $85$). Then in order to | pm$ 0.030 0.815$\pm$ 0.024 0.881$\pm$ 0.020
(b)\[$50$/$\sigma$\ ] $ 1 $ 0.875$\pm$ 0.020 0.558$\pm$ 0.031 0.814$\pm$ 0.024 0.907 $ \pm$ 0.018
(c)\[$20$/$\sigma$\ ] $ 15 $ 0.875$\pm$ 0.020 0.608$\pm$ 0.030 0.769$\pm$ 0.026 0.893$\pm$ 0.019
(d)\[$20$/$\sigma$\ ] $ 1 $ 0.900$\pm$ 0.019 0.602$\pm$ 0.030 0.793$\pm$ 0.025 0.892$\pm$ 0.019
: Validity frequencies \[with precision $ \pm 95\%$\ ] of the CoRP, CoLP, CoRLaP, CENeP, the Early - Stopped CoLP and the $ 2$-PN CoLP based on $ 1000 $ replications.[]{data - label="tab: frqVSsigma " }
-0.1 in
Validity.
: nowadays, we think the validity of the selected predictors (cystic fibrosis. * * Step 3 * * in Algorithm \[alg: ConformalLasso\ ]). As shown in Table \[tab: frqVSsigma\ ], we respect that pas seul on the noise level, the variables correlations and the sparseness of the model do to not perturb the validity whereas the sample size relatively to the dimension $ p$ suffice. When $ n=300 > p$, all the procedures seem to be quite similar and produce good forecaster. In the other cases, i.e., when $ n = p=50 $ and $ n=20 < p$, the selected assurance predictors have bad performance than expect (validity with smaller proportion than $ 1-\varepsilon = 90\%$). Moreover, Sparse Confidence Predictors perform bad than the CoRP as observed in Table \[tab: frqVSsigma\ ]. As pointed in the accuracy part, one explication can be observed in Figure \[fig: plotslength\ ] as the selected forecaster which besides is not valid (iteration $ 93 $) represent to an iteration in the unstable zone (that is, after iteration $ 85 $). Then in order to | pm$ 0.030 0.815$\pm$ 0.024 0.881$\pm$ 0.020
(b)\[$50$/$\sinma$\] $1$ 0.875$\km$ 0.020 0.558$\pm$ 0.031 0.814$\pj$ 0.024 0.907 $\po$ 0.018
(c)\[$20$/$\sigma$\] $15$ 0.875$\pm$ 0.020 0.608$\pm$ 0.030 0.769$\pm$ 0.026 0.893$\pm$ 0.019
(d)\[$20$/$\rigma$\] $1$ 0.900$\pm$ 0.019 0.602$\pm$ 0.030 0.793$\pm$ 0.025 0.892$\pm$ 0.019
: Vanmdity frequencigs \[with prechsion $\pm 95\%$\] of tve ClRP, CoLP, CoRLaP, CENeP, the Early-Stop[ed CoLL wnd the $2$-PN CoLP bastd jn $1000$ dvpoications.[]{data-label="tab:frqVSsigja"}
-0.1ii
Validity.
: Now, we consider the validity lf tje selected predichors (cf. **Stek 3** ig Algorithm \[ale:ConformalLasso\]). As shosn in Table \[tab:frqVSsigma\], we obsdrve chat variatuobs lt the noise levej, the variables corralationx and the sparxitb of the model do to not 'erturb the validity whereas dhz sample size relativwlt to jhe dhmenruon $p$ soxs. Shen $n=300>o$, aml the prodedures seen to be quite similsr qnd produce gkod prqdyctors. In the other cases, i.e., when $n=p=50$ ang $n=20<l$, the selected confidenxe predictors have wotst perforiance than expected (validity with smaller proporthon tiav $1-\vcvcpsiuin = 90\%$). Moreover, Sparse Confidence Predictors perfjdm wprst than the GoRP as observed im Hanje \[tab:frqVSsigoa\]. As 'kihted in the accurafy part, one wxplicatijn csn be observed in Figure \[fig:plotslength\] cs rhe selected predietor which auso os noy valid (iteration $93$) corrzsponda to an itegation in ghe unstable zond (tmat is, after iteration $85$). Then yn order vo | pm$ 0.030 0.815$\pm$ 0.024 0.881$\pm$ 0.020 (b)\[$50$/$\sigma$\] 0.020 0.031 0.814$\pm$ 0.907 $\pm$ 0.018 0.030 0.026 0.893$\pm$ 0.019 $1$ 0.900$\pm$ 0.019 0.030 0.793$\pm$ 0.025 0.892$\pm$ 0.019 : frequencies \[with precision $\pm 95\%$\] of the CoRP, CoLP, CoRLaP, CENeP, the Early-Stopped and the $2$-PN CoLP based on $1000$ replications.[]{data-label="tab:frqVSsigma"} -0.1in Validity. : Now, we the of selected (cf. **Step 3** in Algorithm \[alg:ConformalLasso\]). As shown in Table \[tab:frqVSsigma\], we observe that variations on noise level, the variables correlations and the sparsity the model do to perturb the validity whereas the size to the $p$ When all the procedures to be quite similar and produce good predictors. In the other cases, i.e., when $n=p=50$ and $n=20<p$, selected confidence worst performance expected with proportion than $1-\varepsilon Moreover, Sparse Confidence Predictors perform worst as observed in Table \[tab:frqVSsigma\]. As pointed in accuracy part, explication can be observed in Figure as the selected predictor which also is not (iteration $93$) corresponds to an iteration in the unstable zone (that is, after iteration $85$). order to | pm$ 0.030 0.815$\pm$ 0.024 0.881$\pm$ 0.020
(b)\[$50$/$\sigma$\] $1$ 0.875$\pm$ 0.020 0.558$\pm$ 0.031 0.814$\pm$ 0.024 0.907 $\pm$ 0.018
(c)\[$20$/$\sigmA$\] $15$ 0.875$\pm$ 0.020 0.608$\pm$ 0.030 0.769$\pm$ 0.026 0.893$\pm$ 0.019
(d)\[$20$/$\Sigma$\] $1$ 0.900$\Pm$ 0.019 0.602$\pM$ 0.030 0.793$\pm$ 0.025 0.892$\Pm$ 0.019
: valiDity Frequencies \[witH PrecIsion $\pm 95\%$\] of the CoRP, CoLP, CorLaP, CeNEp, the eArLy-StoPped CoLp AnD THe $2$-Pn COLp baSeD On $1000$ RepliCatIons.[]{datA-label="tab:fRqVssIgma"}
-0.1in
ValidiTY.
: NOw, we considEr tHe validity of The SelectEd PreDIctorS (cf. **step 3** iN AlgorIThm \[alg:conformallaSSo\]). As shOWn in TabLE \[TaB:frqvSsigma\], we observe tHAt VAriations on the Noise lEvEL, tHE VarIabLes correlaTiOns anD The sparSItY OF The MOdel do to not peRturb the valIDitY whereAs The SAmple sIze reLaTIveLy to the dimeNsioN $p$ does. WheN $n=300>p$, all THe proceDUres seeM to be qUitE siMilaR AnD pRodUcE GooD PrEdiCTorS. In the otHeR cAses, i.E., wheN $N=P=50$ ANd $n=20<p$, The SeleCted cOnfidence predIctOrs hAVe wOrst pErforMancE tHan exPected (ValidItY with smaller proPortIon than $1-\vaRepSiLon = 90\%$). moReoveR, sparse conFidEnce PreDictors PErfOrM WORsT than the CoRP as obseRvED In table \[tab:FrqVSsIGmA\]. AS Pointed iN tHe aCcurACY part, One eXPlIcation cAn be obSErVeD in FiguRe \[Fig:ploTsLenGth\] As the SElecTed preDictor whIch alSO is not valid (iteRAtion $93$) correspoNDs TO An ITeraTioN in the unstaBle zONe (thAt is, AFtEr iTEratiOn $85$). TheN iN OrDEr to | pm$ 0.030 0.815$\pm$ 0. 024 0. 881$\ pm$ 0. 02 0
(b)\[$50$/$\si g ma$\ ] $1$ 0.875 $\pm$ 0 . 020 0 .558$ \pm$ 0. 0 31 0 .8 14 $\p m$ 0. 024 0 .907 $\ pm$ 0.018 (c)\[$20$/$ \ si gma$\] $15$ 0.8 75$ \pm$ 0 .0 20 0.60 8$\ pm$ 0 .030 0.769 $\pm$ 0.0 26 0. 8 93$\pm$ 0 .0 19
(d)\[$20$/$\ s ig m a$\] $1$ 0. 9 00 $ \ pm$ 0. 019 0.60 2$ \pm$0 .030 0. 7 9 3 $\p m $ 0.025 0 .892$\pm$ 0 . 019
: V a lidity freq ue n cie s \[with pr ecis ion $\pm95\%$\ ] of the CoRP, C oLP, C oRL aP, CEN e P, t heEa r ly- S to ppe d Co LP and t he $ 2$-PN CoL P b a sedon$100 0$ re plications.[] {da ta-l a bel ="tab :frqV Ssig ma "}
-0. 1in
Va lidity.
: No w, w e conside r t he va li dityo f thesel ect ed pred ictors( cf. * * S t ep 3** in Algorithm\[ a l g: Conforma lLasso \ ]) .A s shownin Ta ble\ [ tab:f rqVS s ig ma\], we obser v eth at vari at ions o nthe no ise l e vel, the v ariables corr e lations and th e sparsity oft he m od e l do to not pertur b th e val idit y w her e as th e sam pl e s i ze relatively to th edimens ion $ p$ does. When $n=300>p$ , a ll the p roce d ur e s seem to be q uitesimilar an d produce good predict ors. In t h e other c ase s,i.e .,w h en $n=p=50$ and $ n=20 <p $, thesel ected c onf ide nce pr ed ictors ha ve worst p er fo rm anc e tha n expecte d(va li dit y wit h small er pr opor ti on tha n $1-\v a re p s ilon = 9 0\%$ ).Mo reove r, S p ars e Confi dence Pre dic t orspe rf orm wor st than the C oR P as obser ve d i n Tabl e \[tab:fr qVSsigma\]. As pointedi n the a ccu racypart , one exp lic ationcan be obs ervedin Fi gu re\ [ fig:p l o ts len gt h\] as the s ele ctedpr edic tor whi ch also is not val i d ( iteration $93 $)corr e s po nds to anit e rat i o n in the unstab le zone (t ha t i s, after i t era ti on $85$ ). Then in o r der to | pm$ 0.030_ _ 0.815$\pm$ 0.024 _ __0.881$\pm$ 0.020
__ _ (b)\[$50$/$\sigma$\] _ _ $1$ __ 0.875$\pm$ 0.020 0.558$\pm$ 0.031 0.814$\pm$ 0.024__ _0.907_$\pm$_0.018
_ (c)\[$20$/$\sigma$\] _ _ $15$ 0.875$\pm$ 0.020__ 0.608$\pm$ 0.030_ 0.769$\pm$ 0.026 _0.893$\pm$ 0.019
_ (d)\[$20$/$\sigma$\] ___ $1$_ 0.900$\pm$_0.019 0.602$\pm$ 0.030 _ 0.793$\pm$ 0.025 _ 0.892$\pm$ 0.019
_ : Validity frequencies_\[with precision_$\pm 95\%$\] of the CoRP,_CoLP, CoRLaP, CENeP,_the Early-Stopped_CoLP and the_$2$-PN CoLP based on $1000$ replications.[]{data-label="tab:frqVSsigma"}
_ -0.1in
Validity.
:_ Now, we consider the_validity_of the selected_predictors_(cf._**Step 3**_in Algorithm \[alg:ConformalLasso\]). As_shown_in Table \[tab:frqVSsigma\],_we_observe that variations on the noise_level,_the variables correlations and the sparsity of_the model do to_not_perturb the validity whereas_the sample size relatively to_the dimension $p$ does. When $n=300>p$,_all the_procedures seem_to be quite similar and produce good predictors. In the other_cases, i.e., when $n=p=50$ and $n=20<p$,_the selected confidence predictors_have worst_performance_than expected (validity_with_smaller proportion_than $1-\varepsilon = 90\%$). Moreover, Sparse Confidence_Predictors perform_worst than the CoRP as observed_in Table \[tab:frqVSsigma\]. As pointed_in_the accuracy part, one explication can_be observed in Figure \[fig:plotslength\] as the_selected predictor which also is_not_valid_(iteration $93$) corresponds to an_iteration in the unstable zone (that_is, after iteration_$85$). Then in order to |
much above the electroweak scale. However, if the family symmetry is global, and broken only spontaneously, then a massless Goldstone boson, the familon, will appear [@w82].
As an application of the mechanism of Fig. 3, consider the non-Abelian discrete symmetry $A_4$, the group of the even permutation of 4 objects which is also the symmetry group of the tetrahedron. It has been discussed [@mr01; @bmv03; @m04; @af05; @m05] as a family symmetry for the understanding of the neutrino mass matrix. Suppose it is combined with $SO(10)$. Then all quarks and leptons are naturally assigned as $({\bf 16}; \underline{3})$ under $SO(10) \times A_4$. \[There are 3 inequivalent irreducible singlet representations of $A_4$, $\underline{1}$, $\underline{1}'$, $\underline{1}''$, and 1 irreducible triplet representation $\underline{3}$.\] This assignment differs from the original one [@mr01; @bmv03] where $q,l \sim \underline{3}$ but $q^c,l^c
\sim \underline{1}, \underline{1}', \underline{1}''$, which cannot be embedded into $SO(10)$. The heavy scalar $H$ should then be assigned as $(\overline{\bf 10}; \underline{1}, \underline{1}', \underline{1}'')$, $\sigma$ as $(\overline{\bf 16}; \underline{1}, \underline{1}',
\underline{1}'')$, and $\phi$ as $(\overline{\bf 16}; \underline{1})$. For $a_{1,2,3} \sim \underline{3}$ and $b_{1,2,3} \sim
\underline{3}$ under $A_4$, $$\begin{aligned}
&& a_1 b_1 + a_2 b_2 + a_3 b_3 \sim \underline{1}, \\
&& a_1 b_1 + \omega^2 a_2 b_2 + \omega a_3 b_3 \sim \underline{1}', \\
&& a_1 b_1 + \omega a_2 b_2 + \omega^2 a_3 b_3 \sim \underline{1}'',\end{aligned | much above the electroweak scale. However, if the family symmetry is ball-shaped, and break in only spontaneously, then a massless Goldstone boson, the familon, will look [ @w82 ].
As an application of the mechanism of Fig. 3, regard the non - Abelian discrete symmetry $ A_4 $, the group of the tied permutation of 4 objects which is also the isotropy group of the tetrahedron. It has been discourse [ @mr01; @bmv03; @m04; @af05; @m05 ] as a family isotropy for the understanding of the neutrino mass matrix. Suppose it is combined with $ SO(10)$. Then all quark and leptons are naturally assigned as $ ({ \bf 16 }; \underline{3})$ under $ SO(10) \times A_4$. \[There are 3 inequivalent irreducible singlet representations of $ A_4 $, $ \underline{1}$, $ \underline{1}'$, $ \underline{1}''$, and 1 irreducible triplet representation $ \underline{3}$.\ ] This grant differs from the original one [ @mr01; @bmv03 ] where $ q, l \sim \underline{3}$ but $ q^c, l^c
\sim \underline{1 }, \underline{1 }', \underline{1}''$, which cannot be embedded into $ SO(10)$. The big scalar $ H$ should then be assigned as $ (\overline{\bf 10 }; \underline{1 }, \underline{1 }', \underline{1}'')$, $ \sigma$ as $ (\overline{\bf 16 }; \underline{1 }, \underline{1 }',
\underline{1}'')$, and $ \phi$ as $ (\overline{\bf 16 }; \underline{1})$. For $ a_{1,2,3 } \sim \underline{3}$ and $ b_{1,2,3 } \sim
\underline{3}$ under $ A_4 $, $ $ \begin{aligned }
& & a_1 b_1 + a_2 b_2 + a_3 b_3 \sim \underline{1 }, \\
& & a_1 b_1 + \omega^2 a_2 b_2 + \omega a_3 b_3 \sim \underline{1 }', \\
& & a_1 b_1 + \omega a_2 b_2 + \omega^2 a_3 b_3 \sim \underline{1}'',\end{aligned | mufh above the electroweak scale. However, nd the hamily aymmetry is global, and broken only s'ontqneouwly, then a massless Goudstone blson, the fammlon, will appear [@w82].
As an appliczbion mh the mechanism of Fig. 3, cotsider the non-Dbdlnan discrete symmetry $A_4$, the group of the evrn permutation os 4 ontecta which is also the symmetry group of the tetrahedron. Ot has been discussed [@mr01; @blv03; @m04; @af05; @m05] as a family dymmetry fot thq understandivg of the neutrino masa matrix. Suppose it is combined with $SO(10)$. Then aol quwtks and leptins age naturally assigned as $({\bf 16}; \underline{3})$ undcr $SO(10) \tines A_4$. \[There are 3 ineqnivalent irreducible singlet se'resentations of $A_4$, $\uneeeline{1}$, $\undarlivw{1}'$, $\uvdedlmne{1}''$, and 1 lrrxducible trjplet reprewentation $\underline{3}$.\] Trpx assignment diffews from the original one [@mr01; @bmv03] where $q,l \vim \underline{3}$ but $q^c,l^c
\sim \underline{1}, \underline{1}', \onderline{1}''$, rhich cannot be embedded into $SO(10)$. The heavy scalar $H$ shkjld tmen cw wssigned as $(\overline{\bf 10}; \underline{1}, \underline{1}', \ugsetlpne{1}'')$, $\sigma$ as $(\overjine{\bf 16}; \undrrpimg{1}, \underline{1}',
\unaerlinz{1}'')$, ahd $\phi$ as $(\overline{\hf 16}; \undgrline{1})$. For $a_{1,2,3} \sii \uncerline{3}$ and $b_{1,2,3} \sim
\underline{3}$ under $A_4$, $$\begpn{alugned}
&& a_1 b_1 + a_2 b_2 + a_3 y_3 \sim \underlnne{1}, \\
&& a_1 b_1 + \okega^2 a_2 b_2 + \omega a_3 b_3 \sim \undsrline{1}', \\
&& a_1 b_1 + \omega a_2 c_2 + \omega^2 a_3 b_3 \sim \uncesline{1}'',\end{aligned | much above the electroweak scale. However, if symmetry global, and only spontaneously, then familon, appear [@w82]. As application of the of Fig. 3, consider the non-Abelian symmetry $A_4$, the group of the even permutation of 4 objects which is the symmetry group of the tetrahedron. It has been discussed [@mr01; @bmv03; @m04; @m05] a symmetry the understanding of the neutrino mass matrix. Suppose it is combined with $SO(10)$. Then all quarks leptons are naturally assigned as $({\bf 16}; \underline{3})$ $SO(10) \times A_4$. \[There 3 inequivalent irreducible singlet representations $A_4$, $\underline{1}'$, $\underline{1}''$, 1 triplet $\underline{3}$.\] This assignment from the original one [@mr01; @bmv03] where $q,l \sim \underline{3}$ but $q^c,l^c \sim \underline{1}, \underline{1}', \underline{1}''$, which be embedded The heavy $H$ then assigned as $(\overline{\bf \underline{1}', \underline{1}'')$, $\sigma$ as $(\overline{\bf 16}; and $\phi$ as $(\overline{\bf 16}; \underline{1})$. For $a_{1,2,3} \underline{3}$ and \sim \underline{3}$ under $A_4$, $$\begin{aligned} && b_1 + a_2 b_2 + a_3 b_3 \sim \\ && a_1 b_1 + \omega^2 a_2 b_2 + \omega a_3 b_3 \sim \underline{1}', \\ b_1 + \omega a_2 + \omega^2 a_3 \sim | much above the electroweak scAle. However, If the FamIly SyMmetRy is Global, and brokeN Only Spontaneously, then a massLess GOlDStonE BoSon, thE familoN, WiLL AppEaR [@w82].
as aN aPPlIcatiOn oF the mecHanism of FiG. 3, coNsIder the non-AbELiAn discrete SymMetry $A_4$, the groUp oF the evEn PerMUtatiOn oF 4 objeCts whiCH is alsO the symmeTrY Group oF The tetrAHEdRon. IT has been discussed [@MR01; @bMV03; @m04; @af05; @m05] as a family SymmetRy FOr THE unDerStanding of ThE neutRIno mass MAtRIX. supPOse it is combinEd with $SO(10)$. TheN All Quarks AnD lePTons arE natuRaLLy aSsigned as $({\bf 16}; \UndeRline{3})$ undeR $SO(10) \timES A_4$. \[There ARe 3 inequIvalenT irRedUcibLE sInGleT rEPreSEnTatIOns Of $A_4$, $\underLiNe{1}$, $\UnderLine{1}'$, $\UNDERlinE{1}''$, anD 1 irrEduciBle triplet repResEntaTIon $\UnderLine{3}$.\] THis aSsIgnmeNt diffErs frOm The original one [@mR01; @bmv03] Where $q,l \siM \unDeRliNe{3}$ But $q^c,L^C
\sim \unDerLinE{1}, \underlIne{1}', \undeRLinE{1}''$, wHICH cAnnot be embedded intO $So(10)$. tHe Heavy scaLar $H$ shOUlD tHEn be assiGnEd aS $(\oveRLIne{\bf 10}; \UndeRLiNe{1}, \underlIne{1}', \undERlInE{1}'')$, $\sigma$ aS $(\oVerlinE{\bF 16}; \unDerLine{1}, \uNDerlIne{1}',
\undErline{1}'')$, anD $\phi$ aS $(\Overline{\bf 16}; \undeRLine{1})$. For $a_{1,2,3} \sim \unDErLINe{3}$ ANd $b_{1,2,3} \sIm
\uNderline{3}$ undEr $A_4$, $$\bEGin{aLignED}
&& a_1 B_1 + a_2 b_2 + A_3 B_3 \sim \uNderlInE{1}, \\
&& A_1 b_1 + \OMega^2 a_2 b_2 + \omega a_3 b_3 \sim \undErLine{1}', \\
&& a_1 b_1 + \Omega A_2 b_2 + \omega^2 a_3 b_3 \sim \uNderline{1}'',\enD{ALIgned | much above the electrowea k scale. H oweve r,ifth e fa mily symmetry is g l obal , and broken only spon taneo us l y, t h en a ma sslessG ol d s ton ebo son ,t he fami lon , willappear [@w 82] .
As an appli c at ion of the me chanism of F ig. 3, co ns ide r thenon -Abel ian di s cretesymmetry$A _ 4$, th e groupo f t he e ven permutation o f 4 objects whichis als ot he s ymm etr y group of t he te t rahedro n .I t has been discusse d [@mr01; @ b mv0 3; @m0 4; @a f 05; @m 05] a sa fa mily symmet ry f or the un dersta n ding of the neu trinomas s m atri x .Su ppo se iti scom b ine d with $ SO (1 0)$.Then a l l qua rks and lept ons are natur all y as s ign ed as $({\ bf 1 6} ; \un derlin e{3}) $under $SO(10) \ time s A_4$. \ [Th er e a re 3 in e quival ent ir reducib le sing l etre p r e se ntations of $A_4$, $ \ u nd erline{1 }$, $\ u nd er l ine{1}'$ ,$\u nder l i ne{1} ''$, an d 1 irre ducibl e t ri plet re pr esenta ti on$\u nderl i ne{3 }$.\]This ass ignme n t differs from the originalo ne [ @m r 01;@bm v03] where$q,l \sim \un d er lin e {3}$but $ q^ c ,l ^ c
\sim \underline{ 1} , \und erlin e{1}', \under line{1}''$ , w hich can notb ee mbedded into $ SO(10 )$. The he a vy scala r $H$ shouldthen be a s s igned as $( \ov erl ine { \ bf 10}; \underl i n e{1} ,\underl ine {1}', \ und erl ine {1} '' )$, $\sig ma$ as $ (\ ov er li ne{ \bf 1 6 }; \unde rl ine {1 },\unde r line{1 }', \und er li n e{1 }'')$,a nd $ \phi $as $(\ ove rl ine{\ bf 1 6 };\underl ine{1})$. Fo r $a_ {1 ,2 ,3} \si m \underline{ 3} $ and $b_{ 1, 2,3 } \sim \underli ne{3}$ under $A_4$, $$\ b egin{al ign ed}
& & a_ 1 b_1 + a _2b_2 +a_3 b_3 \s im \un derli ne {1} , \\
&& a _1 b_ 1+ \omega^2 a _2b_2 + \ omeg a a_3 b _3 \sim \underline { 1}' , \\
&& a_1 b _1+ \o m e ga a_ 2 b _ 2 + \ o meg a ^ 2 a_3 b_3 \sim\underline {1 } '' ,\end{alig n ed | much_above the_electroweak scale. However, if_the family_symmetry_is global,_and_broken only spontaneously,_then a massless_Goldstone boson, the familon,_will appear [@w82].
As_an_application of the mechanism of Fig. 3, consider the non-Abelian discrete symmetry $A_4$, the group_of_the even_permutation_of_4 objects which is also_the symmetry group of the_tetrahedron. It_has been discussed [@mr01; @bmv03; @m04; @af05; @m05]_as_a family symmetry_for the understanding of the neutrino mass matrix. Suppose_it is combined with $SO(10)$. Then_all quarks and_leptons_are_naturally assigned as $({\bf_16}; \underline{3})$ under $SO(10) \times A_4$._\[There are 3 inequivalent irreducible singlet_representations of $A_4$, $\underline{1}$, $\underline{1}'$, $\underline{1}''$, and_1 irreducible triplet representation $\underline{3}$.\] This_assignment differs from the original_one [@mr01;_@bmv03] where $q,l \sim \underline{3}$_but $q^c,l^c
\sim_\underline{1}, \underline{1}',_\underline{1}''$, which cannot_be embedded into $SO(10)$. The heavy_scalar $H$ should_then be assigned as $(\overline{\bf 10};_\underline{1},_\underline{1}', \underline{1}'')$, $\sigma$_as_$(\overline{\bf_16}; \underline{1},_\underline{1}',
\underline{1}'')$, and_$\phi$_as $(\overline{\bf_16};_\underline{1})$. For $a_{1,2,3} \sim \underline{3}$ and_$b_{1,2,3}_\sim
\underline{3}$ under $A_4$, $$\begin{aligned}
&& a_1 b_1_+ a_2 b_2 +_a_3_b_3 \sim \underline{1}, \\
&&_a_1 b_1 + \omega^2 a_2_b_2 + \omega a_3 b_3 \sim_\underline{1}', \\
&&_a_1 b_1_+ \omega a_2 b_2 + \omega^2 a_3 b_3 \sim \underline{1}'',\end{aligned |
_B(R/B)+1\leq n$. Then, by the same argument of proof of Theorem \[thB\], it follows that $\m_B$ is a self-dual ideal of $R$ satisfying $\ell_R(R/\m_B)\leq n$. In this case, Question \[Q1\] is affirmative for $R$.
0
Let $R=k[[t^3,t^{10},t^{11}]]$ be a numerical semigroup ring, where $k$ is a field. We give some observation on Question \[Q1\] fo $R$. $R$ has minimal multiplicity and is not almost Gorenstein. By Corollary, $bg(R)\geq 2$. In the case $bg(R)=2$, Question \[Q1\] is affirmative by Theorem \[thB\]. However, we don’t know whether $bg(R)=2$ or not.
Let $S=k[[t^3,t^{10}]]$ and $\mathfrak{n}$ be its maximal ideal. Then $S$ is Gorenstein and $S\subseteq R$ is module-finite residually rational birational extension with $\ell_S(R/S)=3$. In particular, $bg(R)\leq 3$. Put $B=\mathfrak{n}:\mathfrak{n}=k[[t^3,t^{10},t^{17}]]$. Then $\m_B:\m_B$ is equal to $k[[t^3,t^7]]$ and does not contain $R$. Therefore, we cannot apply Remark for this choice of $S$.
Let $S'=k[[t^6,t^9,t^{10}, t^{13},t^{14}]]$ and $\mathfrak{n}'$ be its maximal ideal. Then $S'$ is Gorenstein and $S'\subseteq R$ is module-finite residually rational birational extension with $\ell_{S'}(R/S')=3$. Put $B'=\mathfrak{n}':\mathfrak{n}'=k[[t^6,t^9,t^{10},t^{13},t^{14},t^{17}]]$. Then $\m_{B'}:\m_{B'}$ is equal to $k[[t^3,t^4]]$, which contains $R$. By Remark, $R$ has a self-dual ideal $I=\ | _ B(R / B)+1\leq n$. Then, by the same argument of proof of Theorem \[thB\ ], it follows that $ \m_B$ is a self - double ideal of $ R$ meet $ \ell_R(R/\m_B)\leq n$. In this case, Question \[Q1\ ] is affirmative for $ R$.
0
Let $ gas constant = k[[t^3,t^{10},t^{11}]]$ be a numerical semigroup ring, where $ k$ is a field. We give some notice on Question \[Q1\ ] fo $ R$. $ R$ has minimal multiplicity and is not almost Gorenstein. By Corollary, $ bg(R)\geq 2$. In the event $ bg(R)=2 $, Question \[Q1\ ] is affirmative by Theorem \[thB\ ]. However, we do n’t know whether $ bg(R)=2 $ or not.
lease $ S = k[[t^3,t^{10}]]$ and $ \mathfrak{n}$ be its maximal ideal. Then $ S$ is Gorenstein and $ S\subseteq R$ is module - finite residually intellectual birational extension with $ \ell_S(R / S)=3$. In particular, $ bg(R)\leq 3$. place $ B=\mathfrak{n}:\mathfrak{n}=k[[t^3,t^{10},t^{17}]]$. Then $ \m_B:\m_B$ is equal to $ k[[t^3,t^7]]$ and does not contain $ R$. consequently, we cannot apply Remark for this choice of $ S$.
Let $ S'=k[[t^6,t^9,t^{10 }, t^{13},t^{14}]]$ and $ \mathfrak{n}'$ be its maximal ideal. Then $ S'$ is Gorenstein and $ S'\subseteq R$ is faculty - finite residually rational birational extension with $ \ell_{S'}(R / S')=3$. Put $ B'=\mathfrak{n}':\mathfrak{n}'=k[[t^6,t^9,t^{10},t^{13},t^{14},t^{17}]]$. Then $ \m_{B'}:\m_{B'}$ is equal to $ k[[t^3,t^4]]$, which contains $ R$. By Remark, $ R$ have a self - dual ideal $ I=\ | _B(R/B)+1\peq n$. Then, by the same avgument of proof of Themrem \[tgB\], it foulows that $\m_B$ is a self-dual mdeao of $E$ satisfying $\ell_R(R/\m_B)\led n$. In thps case, Qyestmon \[Q1\] is affirmavjve for $R$.
0
Let $D=n[[t^3,t^{10},t^{11}]]$ ue a numerical xemigroup sing, where $k$ iv x yield. We give some observation on Quqstion \[A1\] vo $R$. $R$ has minymal iultjillcity and is not almost Gorenstejn. By Cmrollary, $bg(R)\grq 2$. In the case $bg(R)=2$, Questiln \[Q1\] is affirmative by Theorem \[thV\]. Horwver, we don’t know whether $bg(R)=2$ or nkt.
Let $S=k[[t^3,t^{10}]]$ and $\mathfrak{n}$ be its maxikal ideal. Jkwn $D$ is Gorenstxin anq $S\subseteq V$ is mogule-finote residually ravionql birational extensimn with $\ell_S(R/S)=3$. In karticular, $by(R)\leq 3$. Put $B=\mathfrak{n}:\narhfran{n}=k[[t^3,d^{10},t^{17}]]$. Tfwn $\o_B:\m_G$ ms squal ho $i[[t^3,t^7]]$ and doss not contqin $R$. Therefore, we vagbot apply Remzrk fow ehis choice of $S$.
Let $S'=k[[t^6,t^9,t^{10}, t^{13},t^{14}]]$ and $\mathfrdk{n}'$ be its maximal ideal. Tyen $S'$ is Gorenstein ajd $S'\subseeeq R$ is module-finite residually rational biratiotal eetdnsnin wigy $\fll_{S'}(R/S')=3$. Put $B'=\mathfrak{n}':\mathfrak{n}'=k[[t^6,t^9,t^{10},t^{13},t^{14},t^{17}]]$. Then $\m_{B'}:\i_{G'}$ os equal to $k[[t^3,t^4]]$, chich contains $R$. Bj Tgmark, $R$ has a relf-ducm jdeal $I=\ | _B(R/B)+1\leq n$. Then, by the same argument of \[thB\], it that $\m_B$ is satisfying n$. In this Question \[Q1\] is for $R$. 0 Let $R=k[[t^3,t^{10},t^{11}]]$ be numerical semigroup ring, where $k$ is a field. We give some observation on \[Q1\] fo $R$. $R$ has minimal multiplicity and is not almost Gorenstein. By $bg(R)\geq In case Question \[Q1\] is affirmative by Theorem \[thB\]. However, we don’t know whether $bg(R)=2$ or not. Let and $\mathfrak{n}$ be its maximal ideal. Then $S$ Gorenstein and $S\subseteq R$ module-finite residually rational birational extension $\ell_S(R/S)=3$. particular, $bg(R)\leq Put Then is equal to and does not contain $R$. Therefore, we cannot apply Remark for this choice of $S$. Let $S'=k[[t^6,t^9,t^{10}, and $\mathfrak{n}'$ maximal ideal. $S'$ Gorenstein $S'\subseteq R$ is rational birational extension with $\ell_{S'}(R/S')=3$. Put is equal to $k[[t^3,t^4]]$, which contains $R$. By $R$ has self-dual ideal $I=\ | _B(R/B)+1\leq n$. Then, by the same argumEnt of proof Of TheOreM \[thb\], iT folLows That $\m_B$ is a self-dUAl idEal of $R$ satisfying $\ell_R(R/\m_b)\leq n$. in THis cASe, questIon \[Q1\] is aFFiRMAtiVe FoR $R$.
0
LEt $r=K[[t^3,T^{10},t^{11}]]$ be a NumErical sEmigroup riNg, wHeRe $k$ is a field. WE GiVe some obseRvaTion on QuestiOn \[Q1\] Fo $R$. $R$ haS mIniMAl mulTipLicitY and is NOt almoSt GorenstEiN. by CoroLLary, $bg(R)\GEQ 2$. IN the Case $bg(R)=2$, Question \[Q1\] iS AfFIrmative by TheoRem \[thB\]. hoWEvER, We dOn’t Know whetheR $bG(R)=2$ or nOT.
Let $S=k[[t^3,T^{10}]]$ AnD $\MAThfRAk{n}$ be its maximAl ideal. Then $s$ Is GOrenstEiN anD $s\subseTeq R$ iS mODulE-finite resiDualLy rationaL biratIOnal extENsion wiTh $\ell_S(r/S)=3$. IN paRticULaR, $bG(R)\lEq 3$. pUt $B=\MAtHfrAK{n}:\mAthfrak{n}=K[[t^3,T^{10},t^{17}]]$. then $\m_b:\m_B$ iS EQUAl to $K[[t^3,t^7]]$ And dOes noT contain $R$. TherEfoRe, we CAnnOt appLy RemArk fOr This cHoice oF $S$.
Let $s'=k[[T^6,t^9,t^{10}, t^{13},t^{14}]]$ and $\mathfraK{n}'$ be Its maximaL idEaL. ThEn $s'$ is GoREnsteiN anD $S'\sUbseteq r$ is moduLE-fiNiTE REsIdually rational birAtIONaL extensiOn with $\ELl_{s'}(R/s')=3$. put $B'=\mathFrAk{n}':\MathFRAk{n}'=k[[t^6,T^9,t^{10},t^{13},t^{14},T^{17}]]$. thEn $\m_{B'}:\m_{B'}$ is Equal tO $K[[t^3,T^4]]$, wHich conTaIns $R$. By reMarK, $R$ hAs a seLF-duaL ideal $i=\ | _B(R/B)+1\leq n$. Then, by the sameargum ent of p roof ofTheorem \[thB\ ] , it follows that $\m_B$ i s a s el f -dua l i dealof $R$s at i s fyi ng $ \el l_ R (R /\m_B )\l eq n$.In this ca se, Q uestion \[Q1 \ ]is affirma tiv e for $R$.
0
Let $R =k [[t ^ 3,t^{ 10} ,t^{1 1}]]$b e a nu merical s em i groupr ing, wh e r e$k$is a field. We gi v es ome observatio n on Q ue s ti o n \[ Q1\ ] fo $R$.$R $ has minimal mu l t i pli c ity and is no t almost Go r ens tein.By Co r ollary , $bg (R ) \ge q 2$. In th e ca se $bg(R) =2$, Q u estion\ [Q1\] i s affi rma tiv e by Th eo rem \ [ thB \ ]. Ho w eve r, we do n’ tknowwhet h e r $bg( R)= 2$ o r not .
Let $S=k[[ t^3 ,t^{ 1 0}] ]$ an d $\m athf ra k{n}$ be it s max im al ideal. Then$S$is Gorens tei nand $ S\sub s eteq R $ i s m odule-f inite r e sid ua l l y r ational birational e x t en sion wit h $\el l _S (R / S)=3$. I npar ticu l a r, $b g(R) \ le q 3$. Pu t $B=\ m at hf rak{n}: \m athfra k{ n}= k[[ t^3,t ^ {10} ,t^{17 }]]$. Th en $\ m _B:\m_B$ is eq u al to $k[[t^3 , t^ 7 ] ]$ anddoe s not conta in $ R $. T here f or e,w e can not a pp l yR emark for this choi ce of $S $.
L et $S'=k[[t^6 ,t^9,t^{10 } , t^{13},t ^{14 } ]] $ and $\mathfra k{n}' $ be its m a ximal id eal.Then $S' $ is Gore n s tein and $S '\s ubs ete q R$ is module-fi n i te r es idually ra tionalbir ati ona l e xt ension wi th $\ell _{ S' }( R/ S') =3$.P ut $B'=\ ma thf ra k{n }':\m a thfrak {n}'= k[[t ^6 ,t ^ 9,t ^{10},t ^ {1 3 } ,t^{ 14 }, t^{1 7}] ]$ . The n $\ m _{B '}:\m_{ B'}$ is e qua l to$k [[ t^3,t^4 ]]$, which co nt ains $R$.By Re mark,$ R $ has aself-dual ideal $I=\ | _B(R/B)+1\leq n$._Then, by_the same argument of_proof of_Theorem_\[thB\], it_follows_that $\m_B$ is_a self-dual ideal_of $R$ satisfying $\ell_R(R/\m_B)\leq_n$. In this_case,_Question \[Q1\] is affirmative for $R$.
0
Let $R=k[[t^3,t^{10},t^{11}]]$ be a numerical semigroup ring, where $k$_is_a field._We_give_some observation on Question \[Q1\]_fo $R$. $R$ has minimal_multiplicity and_is not almost Gorenstein. By Corollary, $bg(R)\geq 2$._In_the case $bg(R)=2$,_Question \[Q1\] is affirmative by Theorem \[thB\]. However, we_don’t know whether $bg(R)=2$ or not.
Let_$S=k[[t^3,t^{10}]]$ and $\mathfrak{n}$_be_its_maximal ideal. Then $S$_is Gorenstein and $S\subseteq R$ is_module-finite residually rational birational extension with_$\ell_S(R/S)=3$. In particular, $bg(R)\leq 3$. Put $B=\mathfrak{n}:\mathfrak{n}=k[[t^3,t^{10},t^{17}]]$._Then $\m_B:\m_B$ is equal to $k[[t^3,t^7]]$_and does not contain $R$._Therefore, we_cannot apply Remark for this_choice of $S$.
Let_$S'=k[[t^6,t^9,t^{10}, t^{13},t^{14}]]$_and $\mathfrak{n}'$ be_its maximal ideal. Then $S'$ is_Gorenstein and $S'\subseteq_R$ is module-finite residually rational birational_extension_with $\ell_{S'}(R/S')=3$. Put_$B'=\mathfrak{n}':\mathfrak{n}'=k[[t^6,t^9,t^{10},t^{13},t^{14},t^{17}]]$._Then_$\m_{B'}:\m_{B'}$ is_equal to $k[[t^3,t^4]]$,_which_contains $R$._By_Remark, $R$ has a self-dual ideal_$I=\ |
O_h$) space group symmetry. In this [*whole-crystal*]{} symmetry, we first focus on the Co lattice by ignoring the first-neighboring Mn and Si atoms. The lattice is assumed to be a simple cubic composed by the second-neighboring Co at different sublattices in the primitive cell, which leads to the Co sitting at $O_h$ [*site*]{} symmetry. Second, our focus is turned on the tetrahedral ($T_d$) [*site*]{} symmetry. Neglecting the chemical atomic species, every atom forms a bcc lattice structure and is surrounded by a tetrahedral environment. The hybridization diagram of atomic orbital energy is discussed by following these two steps. Note that, for avoiding the confusion regarding the notations, the symmetric characters of the atomic orbital are unified using only representations for the $O_h$ site symmetry, which corresponds to the space group of the $L2_1$ full Heusler compound, as done also in the previous research [@prb-66-174429-2002].
We again start with the results of the standard GGA calculations for discussing the underlying electronic structure of Co$_2$MnSi. Figure \[fig\_band\_gga\] (a) shows the band structures for the minority spin states projected into the Co $e_{g(u)}$ ($d_{z^2}, d_{x^2-y^2}$) and $t_{2g(1u)}$ ($d_{xz}, d_{yz}, d_{xy}$), Mn $e_g$ ($d_{z^2}, d_{x^2-y^2}$) and $t_{2g}$ ($d_{xz}, d_{yz}, d_{xy}$), and Si $t_{1u}$ ($p_x, p_y, p_z$) orbitals. The lattice constant is set to the theoretically obtained value of 5.639 [Å]{}. The $d$ states of Co and Mn are visible around Fermi energy, while the Si $t_{1u}$ state can be seen only at very far from Fermi energy. To discuss the orbital hybridization mechanism, the eigenstates at $\Gamma$ point in Brillouin zone are focused. At 0.4 eV above the Fermi energy, the Co $e_u$ state appears but the other orbital components are not included in these eigenstates, which means that Co $e_u$ does not hybridize with the other atomic | O_h$) space group symmetry. In this [ * whole - crystal * ] { } isotropy, we foremost concenter on the Co lattice by ignore the beginning - neighboring Mn and Si atoms. The wicket is assumed to be a simple cubic composed by the second - neighbor Co at different sublattices in the archaic cell, which leads to the cobalt sit at $ O_h$ [ * site * ] { } symmetry. Second, our focus is turned on the tetrahedral ($ T_d$) [ * site * ] { } isotropy. Neglecting the chemical atomic species, every atom form a bcc lattice structure and is surround by a tetrahedral environment. The hybridization diagram of atomic orbital department of energy is discussed by follow these two steps. Note that, for avoiding the confusion involve the notations, the symmetric characters of the atomic orbital are unified using only representations for the $ O_h$ site symmetry, which corresponds to the space group of the $ L2_1 $ full Heusler compound, as done also in the previous research [ @prb-66 - 174429 - 2002 ].
We again get down with the results of the standard GGA calculations for hash out the underlie electronic structure of Co$_2$MnSi. Figure \[fig\_band\_gga\ ] (a) indicate the band structures for the minority spin states plan into the Co $ e_{g(u)}$ ($ d_{z^2 }, d_{x^2 - y^2}$) and $ t_{2g(1u)}$ ($ d_{xz }, d_{yz }, d_{xy}$), Mn $ e_g$ ($ d_{z^2 }, d_{x^2 - y^2}$) and $ t_{2g}$ ($ d_{xz }, d_{yz }, d_{xy}$), and Si $ t_{1u}$ ($ p_x, p_y, p_z$) orbitals. The lattice constant is set to the theoretically obtained value of 5.639 [ Å ] { }. The $ d$ states of Co and Mn are visible around Fermi energy, while the Si $ t_{1u}$ state can be attend only at very far from Fermi energy. To discuss the orbital hybridization mechanism, the eigenstates at $ \Gamma$ point in Brillouin zone are focus. At 0.4 eV above the Fermi department of energy, the Co $ e_u$ department of state appears but the other orbital components are not included in these eigenstates, which mean that Co $ e_u$ does not hybridize with the other atomic | O_h$) dpace group symmetry. In uhis [*whole-crystal*]{} symmetcy, we fjrst focjs on the Co lattice by ignocing the dirst-neighboring Mn ana Si atomd. The lartict is assumed to bx a simple cubjg com'owed by the secpnd-neighbosing Co at difxefeut sublattices in the primitive cell, which kewds to the Co fittpnd at $O_h$ [*site*]{} symmetry. Second, our focus is turied on the tetrshedral ($T_d$) [*site*]{} symmetry. Nfglefting the chemical atomic spexies, wvery atom furms a bcc lattice strocture and is surrounded by a tegrahebral envirobmwnt. Jhe hybridizetion qiagram of abpmic osbital rnergy is discmssed by following these two vteps. Note that, fot avoiding tke confusion regardint rhe nmtathons, rhe syjmxtrjc chagacvers of the atomic orbutal are unified usond only represehtatiogs for the $O_h$ site symmetry, which correspmnda to the space group of the $L2_1$ full Heusler clmpound, af done also in the previous research [@prb-66-174429-2002].
We again sdart xigh uhc fwsklts of the standard GGA calculations for disshsxikg the underlyinn electronic strucyuge jf Co$_2$MnSi. Figore \[fig\_bchd\_fga\] (a) shows the banf strucjures dor the mynoroty spin states projected ibto the Co $e_{g(l)}$ ($d_{z^2}, d_{x^2-y^2}$) and $t_{2g(1u)}$ ($d_{xz}, d_{vz}, d_{xy}$), Mn $e_g$ ($b_{z^2}, d_{x^2-y^2}$) and $y_{2g}$ ($d_{xz}, d_{yz}, d_{xy}$), and Si $t_{1u}$ ($p_x, l_y, p_z$) orbitwls. The lzgtice constant ir svt tm the thturetically obtainqd value if 5.639 [Å]{}. Che $d$ stxtes of Co and Mn arf visltle around Fermi ejergy, wvile the Sl $t_{1u}$ state can be seen only at txry far from Gesmi energy. Co disguss the orbitaj hybridizatiou mechannsm, thd eigenstanes at $\Gakma$ point ig Brillouin zmje are focusxd. At 0.4 eV wbovw thw Fermi dnergy, the Co $r_u$ state ckpears but the other orbital coolonents are not inxluded in these eieenftwtxs, whywh means thad Co $d_u$ apes nut hybridizc wkth yhe other atomic | O_h$) space group symmetry. In this [*whole-crystal*]{} first on the lattice by ignoring atoms. lattice is assumed be a simple composed by the second-neighboring Co at sublattices in the primitive cell, which leads to the Co sitting at $O_h$ symmetry. Second, our focus is turned on the tetrahedral ($T_d$) [*site*]{} symmetry. Neglecting chemical species, atom a bcc lattice structure and is surrounded by a tetrahedral environment. The hybridization diagram of atomic energy is discussed by following these two steps. that, for avoiding the regarding the notations, the symmetric of atomic orbital unified only for the $O_h$ symmetry, which corresponds to the space group of the $L2_1$ full Heusler compound, as done also in previous research again start the of standard GGA calculations the underlying electronic structure of Co$_2$MnSi. shows the band structures for the minority spin projected into Co $e_{g(u)}$ ($d_{z^2}, d_{x^2-y^2}$) and $t_{2g(1u)}$ d_{yz}, d_{xy}$), Mn $e_g$ ($d_{z^2}, d_{x^2-y^2}$) and $t_{2g}$ d_{yz}, d_{xy}$), and Si $t_{1u}$ ($p_x, p_y, p_z$) orbitals. The lattice constant is set to obtained value of 5.639 The $d$ states Co Mn visible Fermi energy, the Si $t_{1u}$ state can be seen only at very far Fermi energy. To discuss the orbital hybridization mechanism, the eigenstates point Brillouin zone are At 0.4 eV above Fermi the Co $e_u$ state the orbital included these which means that Co does not hybridize with the atomic | O_h$) space group symmetry. In thiS [*whole-crysTal*]{} syMmeTry, We FirsT focUs on the Co lattiCE by iGnoring the first-neighboRing MN aND Si aTOmS. The lAttice iS AsSUMed To Be A siMpLE cUbic cOmpOsed by tHe second-neIghBoRing Co at diffEReNt sublattiCes In the primitiVe cEll, whiCh LeaDS to thE Co SittiNg at $O_h$ [*SIte*]{} symMetry. SecoNd, OUr focuS Is turneD ON tHe teTrahedral ($T_d$) [*site*]{} syMMeTRy. Neglecting thE chemiCaL AtOMIc sPecIes, every atOm Forms A Bcc lattICe STRUctURe and is surrouNded by a tetrAHedRal envIrOnmENt. The hYbridIzATioN diagram of aTomiC orbital eNergy iS DiscussED by follOwing tHesE twO stePS. NOtE thAt, FOr aVOiDinG The ConfusioN rEgArdinG the NOTATionS, thE symMetriC characters of The AtomIC orBital Are unIfieD uSing oNly repResenTaTions for the $O_h$ siTe syMmetry, whiCh cOrResPoNds to THe spacE grOup Of the $L2_1$ fUll HeusLEr cOmPOUNd, As done also in the preViOUS rEsearch [@pRb-66-174429-2002].
We agAIn StARt with thE rEsuLts oF THe staNdarD gGa calculaTions fOR dIsCussing ThE underLyIng EleCtronIC strUcture Of Co$_2$MnSi. figurE \[Fig\_band\_gga\] (a) shoWS the band strucTUrES FoR The mInoRity spin staTes pROjecTed iNTo The cO $e_{g(u)}$ ($d_{Z^2}, d_{x^2-y^2}$) aNd $T_{2G(1u)}$ ($D_{Xz}, d_{yz}, d_{xy}$), Mn $e_g$ ($d_{z^2}, d_{x^2-y^2}$) anD $t_{2G}$ ($d_{xz}, d_{yZ}, d_{xy}$), aNd Si $t_{1u}$ ($p_x, p_y, p_z$) oRbitals. The LATTice consTant IS sET to the theoretiCally Obtained vaLUe of 5.639 [Å]{}. The $D$ statEs of Co anD Mn are visIBLe around ferMi eNerGy, wHILe The Si $t_{1u}$ state cAN Be seEn Only at vEry Far from ferMi eNerGy. TO dIscuss the Orbital hYbRiDiZaTioN mechANism, the eIgEnsTaTes At $\GamMA$ point In BriLlouIn ZoNE arE focuseD. at 0.4 Ev AbovE tHe fermI enErGy, the co $e_u$ STatE appearS but the otHer ORbitAl CoMponentS are not includEd In these eigEnStaTes, whiCH Means thaT Co $e_u$ does not hybridize wiTH the othEr aTomic | O_h$) space group symmetry . In this[*who le- cry st al*] {} s ymmetry, we fi r st f ocus on the Co lattice by i gn o ring th e fir st-neig h bo r i ngMn a ndSi at oms.The lattic e is assum edto be a simple cu bic compos edby the secon d-n eighbo ri ngC o atdif feren t subl a ttices in the p ri m itivec ell, wh i c hlead s to the Co sitti n ga t $O_h$ [*site *]{} s ym m et r y . S eco nd, our fo cu s ist urned o n t h e tet r ahedral ($T_d $) [*site*] { } s ymmetr y. Ne g lectin g the c h emi cal atomicspec ies, ever y atom forms a bcc lat tice s tru ctu re a n dis su rr o und e dbya te trahedra len viron ment . T h e hy bri diza tiondiagram of at omi c or b ita l ene rgy i s di sc ussed by fo llowi ng these two step s. N ote that, fo ravo id ing t h e conf usi onregardi ng then ota ti o n s ,the symmetric char ac t e rs of theatomic or bi t al are u ni fie d us i n g onl y re p re sentatio ns for th e$O_h$ s it e symm et ry, wh ich c o rres pondsto the s paceg roup of the $L 2 _1$ full Heus l er c om p ound , a s done also int he p revi o us re s earch [@pr b- 6 6- 1 74429-2002].
We ag ai n star t wit h the results of the st a n d ard GGAcalc u la t ions for discu ssing the under l ying ele ctron ic struc ture of C o $ _2$MnSi. Fi gur e \ [fi g \ _b and\_gga\] (a ) show sthe ban d s tructur esfor th e m in ority spi n states p ro je ct edintot he Co $e _{ g(u )} $ ( $d_{z ^ 2}, d_ {x^2- y^2} $) a n d $ t_{2g(1 u )} $ ($d_ {x z} , d_ {yz }, d_{x y}$) , Mn $e_g$($d_{z^2} , d _ {x^2 -y ^2 }$) and $t_{2g}$ ($d _{ xz}, d_{yz }, d_ {xy}$) , and Si $ t_{1u}$ ($p_x, p_y, p_z $ ) orbit als . The lat tice cons tan t is s ett o thetheore tical ly ob t a inedv a lu e o f5.639 [Å]{ } . Th e $d$ s tate s of Co and Mn are visibl e ar ound Fermi en erg y, w h i le th e S i $t _{ 1 u}$ s tate can be see n only atve r yfar from F e rmi e nergy.To disc uss t h e orbit al hybrid ization m ec hani s m , t he eigenst ates at$\Gamma$p ointi nBrill oui n zone a refocus ed. At 0.4 eV a bove t he Fermi ener gy , the Co $e_u$ state appears bu t theother or bital com pon e nts are notincl uded in th ese ei genst ate s , whi ch m e an s t h at Co $e_ u $ does no t h ybr i d iz e with theo t h eratomi c | O_h$) space_group symmetry._In this [*whole-crystal*]{} symmetry,_we first_focus_on the_Co_lattice by ignoring_the first-neighboring Mn_and Si atoms. The_lattice is assumed_to_be a simple cubic composed by the second-neighboring Co at different sublattices in the_primitive_cell, which_leads_to_the Co sitting at $O_h$_[*site*]{} symmetry. Second, our focus_is turned_on the tetrahedral ($T_d$) [*site*]{} symmetry. Neglecting the_chemical_atomic species, every_atom forms a bcc lattice structure and is surrounded_by a tetrahedral environment. The hybridization_diagram of atomic_orbital_energy_is discussed by following_these two steps. Note that, for_avoiding the confusion regarding the notations,_the symmetric characters of the atomic orbital_are unified using only representations for_the $O_h$ site symmetry, which_corresponds to_the space group of the_$L2_1$ full Heusler_compound, as_done also in_the previous research [@prb-66-174429-2002].
We again start with_the results of_the standard GGA calculations for discussing_the_underlying electronic structure_of_Co$_2$MnSi._Figure \[fig\_band\_gga\] (a) shows_the band structures_for_the minority_spin_states projected into the Co $e_{g(u)}$ ($d_{z^2},_d_{x^2-y^2}$)_and $t_{2g(1u)}$ ($d_{xz}, d_{yz}, d_{xy}$), Mn $e_g$ ($d_{z^2},_d_{x^2-y^2}$) and $t_{2g}$ ($d_{xz},_d_{yz},_d_{xy}$), and Si $t_{1u}$ ($p_x,_p_y, p_z$) orbitals. The lattice_constant is set to the theoretically_obtained value_of 5.639 [Å]{}._The $d$ states of Co and Mn are visible around Fermi_energy, while the Si $t_{1u}$ state_can be seen only_at very_far_from Fermi energy._To_discuss the_orbital hybridization mechanism, the eigenstates at $\Gamma$_point in_Brillouin zone are focused. At 0.4 eV_above the Fermi energy,_the_Co $e_u$ state appears but the_other orbital components are not included_in these eigenstates, which means_that_Co $e_u$_does not hybridize with the_other atomic |
aligned}$$
The above properties (\[eab0\]), (\[eab4\]), (\[eab\]), and $\mu^\emptyset$ an identity, are all we need from $\otimes$ to get the results of this work.
Note that commutation and associativity imply that $\bigotimes_{L\in {{\cal D}}}\mu_L\in {\cal P}_I$ is well-defined for a partition ${{\cal D}}$ of $I$.
For $\otimes$ the product measure we have $$\forall x\in \prod_{i\in I}A_i: \quad
\bigotimes_{L\in {{\cal D}}}\mu_L(x)=\prod_{L\in
{{\cal D}}}\mu_L(x_L).$$ If ${{\cal D}}={{\cal D}}^{si}=\{\{i\}: i\in I\}$, $\bigotimes_{i\in I} \mu_{\{i\}}$ is called Bernoulli and $(\mu^{\{i\}}: i\in I)$ are the one-site marginals.
From now on, we fix $\rho=(\rho_J: J\in {\mathbb{S}})$ a probability vector, so $\rho_J\ge 0$ for $J\in {\mathbb{S}}$ and $\sum_{J\in {\mathbb{S}}}\rho_J=1$, and that also satisfies $\rho_\emptyset=0$.
\[def1\] Define the following transformation $\Xi: {\cal P}_I\to {\cal P}_I$, $$\label{e4}
\Xi[\mu]= \sum_{J\in {\mathbb{S}}}
\rho_J \, \mu_J\otimes \mu_{J^c}=\sum_{J\in {\mathbb{S}}^{(\emptyset)}}
\rho_J \, \mu_J\otimes \mu_{J^c}.$$ $\Box$
Since $\mu_\emptyset\equiv 1$, then $$\label{e4x}
\Xi[\mu]= \rho_I \, \mu + \sum_{J\in {\mathbb{S}}^{\emptyset,I}}
\rho_J \, \mu_J\otimes \mu_{J^c}.$$
\[rem0ax\] We have $\rho_\emptyset=0$, but we can have $\rho_I>0$. On the other hand, since $\mu_J\otimes \mu_{ | aligned}$$
The above properties (\[eab0\ ]), (\[eab4\ ]), (\[eab\ ]), and $ \mu^\emptyset$ an identity, are all we need from $ \otimes$ to get the results of this workplace.
notice that commutation and associativity imply that $ \bigotimes_{L\in { { \cal D}}}\mu_L\in { \cal P}_I$ is well - define for a division $ { { \cal D}}$ of $ I$.
For $ \otimes$ the product measure we have $ $ \forall x\in \prod_{i\in I}A_i: \quad
\bigotimes_{L\in { { \cal D}}}\mu_L(x)=\prod_{L\in
{ { \cal D}}}\mu_L(x_L).$$ If $ { { \cal D}}={{\cal D}}^{si}=\{\{i\ }: i\in I\}$, $ \bigotimes_{i\in I } \mu_{\{i\}}$ is call Bernoulli and $ (\mu^{\{i\ } }: i\in I)$ are the one - site marginals.
From now along, we specify $ \rho=(\rho_J: J\in { \mathbb{S}})$ a probability vector, so $ \rho_J\ge 0 $ for $ J\in { \mathbb{S}}$ and $ \sum_{J\in { \mathbb{S}}}\rho_J=1 $, and that besides satisfies $ \rho_\emptyset=0$.
\[def1\ ] Define the following transformation $ \Xi: { \cal P}_I\to { \cal P}_I$, $ $ \label{e4 }
\Xi[\mu]= \sum_{J\in { \mathbb{S } } }
\rho_J \, \mu_J\otimes \mu_{J^c}=\sum_{J\in { \mathbb{S}}^{(\emptyset) } }
\rho_J \, \mu_J\otimes \mu_{J^c}.$$ $ \Box$
Since $ \mu_\emptyset\equiv 1 $, then $ $ \label{e4x }
\Xi[\mu]= \rho_I \, \mu + \sum_{J\in { \mathbb{S}}^{\emptyset, I } }
\rho_J \, \mu_J\otimes \mu_{J^c}.$$
\[rem0ax\ ] We have $ \rho_\emptyset=0 $, but we can accept $ \rho_I>0$. On the other hand, since $ \mu_J\otimes \mu _ { | alihned}$$
The above properties (\[eab0\]), (\[eab4\]), (\[eab\]), anb $\mu^\empvyset$ ah identigy, are all we need from $\otimxs$ ti get the results of this wurk.
Note tjat commytatmon and associatmbity imijy tgwt $\bngitimes_{L\in {{\cal C}}}\mu_L\in {\cal P}_I$ is well-defhndd for a partition ${{\cal D}}$ of $I$.
For $\otimqs$ the lrlduct measure re hsde $$\fkgaol x\in \prod_{i\in I}A_i: \quad
\bigotjmes_{L\in {{\cal D}}}\mu_L(x)=\proc_{L\in
{{\cal D}}}\mu_L(x_L).$$ If ${{\cal D}}={{\cap D}}^{sl}=\{\{i\}: i\in I\}$, $\bigotimes_{l\in I} \mu_{\{i\}}$ iw cajoed Bernoullk and $(\mu^{\{i\}}: p\nn I)$ are thg one-site marginals.
From now on, wd fix $\rho=(\rho_J: J\un {\mahvbb{S}})$ a probebilitj vector, so $\rmp_J\ge 0$ xor $J\in {\mathbb{S}}$ and $\smm_{J\in {\marhbb{S}}}\rho_J=1$, and that alvo satisfies $\rho_\emktyset=0$.
\[def1\] Geyine the following trqnwformdtiot $\Xi: {\xal P}_I\uo {\ral P}_I$, $$\lahel{x4}
\Xi[\mu]= \sum_{J\ih {\mathbb{S}}}
\ryo_J \, \mu_J\otimes \mu_{J^c}=\xui_{H\in {\mathbb{S}}^{(\emltyset)}}
\whj_J \, \mu_J\otimes \mu_{J^c}.$$ $\Box$
Since $\mu_\emptyset\exuib 1$, then $$\label{e4x}
\Xi[\mu]= \rho_U \, \mu + \sum_{J\in {\mathbb{S}}^{\gmptyset,I}}
\rro_J \, \mu_J\otimes \mu_{J^c}.$$
\[rem0ax\] We have $\rho_\emptyset=0$, but fe cai favt $\vmo_I>0$. Ub hhe other hand, since $\mu_J\otimes \mu_{ | aligned}$$ The above properties (\[eab0\]), (\[eab4\]), (\[eab\]), an are all need from $\otimes$ this Note that commutation associativity imply that {{\cal D}}}\mu_L\in {\cal P}_I$ is well-defined a partition ${{\cal D}}$ of $I$. For $\otimes$ the product measure we have x\in \prod_{i\in I}A_i: \quad \bigotimes_{L\in {{\cal D}}}\mu_L(x)=\prod_{L\in {{\cal D}}}\mu_L(x_L).$$ If ${{\cal D}}={{\cal D}}^{si}=\{\{i\}: I\}$, I} is Bernoulli and $(\mu^{\{i\}}: i\in I)$ are the one-site marginals. From now on, we fix $\rho=(\rho_J: J\in a probability vector, so $\rho_J\ge 0$ for $J\in and $\sum_{J\in {\mathbb{S}}}\rho_J=1$, and also satisfies $\rho_\emptyset=0$. \[def1\] Define following $\Xi: {\cal {\cal $$\label{e4} \sum_{J\in {\mathbb{S}}} \rho_J \mu_J\otimes \mu_{J^c}=\sum_{J\in {\mathbb{S}}^{(\emptyset)}} \rho_J \, \mu_J\otimes \mu_{J^c}.$$ $\Box$ Since $\mu_\emptyset\equiv 1$, then $$\label{e4x} \Xi[\mu]= \rho_I \, \mu \sum_{J\in {\mathbb{S}}^{\emptyset,I}} \mu_J\otimes \mu_{J^c}.$$ We $\rho_\emptyset=0$, we can have the other hand, since $\mu_J\otimes \mu_{ | aligned}$$
The above properties (\[Eab0\]), (\[eab4\]), (\[eab\]), aNd $\mu^\eMptYseT$ aN ideNtitY, are all we need fROm $\otImes$ to get the results of tHis woRk.
nOte tHAt CommuTation aND aSSOciAtIvIty ImPLy That $\bIgoTimes_{L\iN {{\cal D}}}\mu_L\in {\Cal p}_I$ Is well-defineD FoR a partitioN ${{\caL D}}$ of $I$.
For $\otimEs$ tHe prodUcT meASure wE haVe $$\forAll x\in \PRod_{i\in i}A_i: \quad
\biGoTImes_{L\iN {{\Cal D}}}\mu_L(X)=\PRoD_{L\in
{{\Cal D}}}\mu_L(x_L).$$ If ${{\cal D}}={{\caL d}}^{sI}=\{\{I\}: i\in I\}$, $\bigotimes_{I\in I} \mu_{\{I\}}$ iS CaLLEd BErnOulli and $(\mu^{\{I\}}: i\In I)$ arE The one-sITe MARGinALs.
From now on, we Fix $\rho=(\rho_J: J\IN {\maThbb{S}})$ a PrObaBIlity vEctor, So $\RHo_J\Ge 0$ for $J\in {\matHbb{S}}$ And $\sum_{J\in {\Mathbb{s}}}\Rho_J=1$, and THat also SatisfIes $\Rho_\EmptYSeT=0$.
\[dEf1\] DEfINe tHE fOllOWinG transfoRmAtIon $\Xi: {\Cal P}_i\TO {\CAl P}_I$, $$\LabEl{e4}
\XI[\mu]= \suM_{J\in {\mathbb{S}}}
\rhO_J \, \mU_J\otIMes \Mu_{J^c}=\sUm_{J\in {\MathBb{s}}^{(\emptYset)}}
\rhO_J \, \mu_J\OtImes \mu_{J^c}.$$ $\Box$
SincE $\mu_\eMptyset\eqUiv 1$, ThEn $$\lAbEl{e4x}
\XI[\Mu]= \rho_I \, \Mu + \sUm_{J\In {\mathbB{S}}^{\emptySEt,I}}
\RhO_j \, \MU_J\Otimes \mu_{J^c}.$$
\[rem0ax\] We hAvE $\RHo_\Emptyset=0$, But we cAN hAvE $\Rho_I>0$. On thE oTheR hanD, SInce $\mU_J\otIMeS \mu_{ | aligned}$$
The above prop erties (\[ eab0\ ]), (\ [e ab4\ ]),(\[eab\]), and $\mu ^\emptyset$ an identit y, ar ea ll w e n eed f rom $\o t im e s $ t oge t t he re sults of this w ork.
Note th at commutation an d associat ivi ty imply tha t $ \bigot im es_ { L\in{{\ cal D }}}\mu _ L\in { \cal P}_I $i s well - defined f or a p artition ${{\calD }} $ of $I$.
For$\otim es $ t h e pr odu ct measure w e hav e $$\for a ll x \ in\ prod_{i\in I} A_i: \quad\bi gotime s_ {L\ i n {{\c al D} }} \ mu_ L(x)=\prod_ {L\i n
{{\cal D}}}\ m u_L(x_L ) .$$ If${{\ca l D }}= {{\c a lD} }^{ si } =\{ \ {i \}: i\i n I\}$,$\ bi gotim es_{ i \ i n I}\mu _{\{ i\}}$ is called Be rno ulli and $(\m u^{\{ i\}} :i\inI)$ ar e the o ne-site margina ls.
From now on ,wefi x $\r h o=(\rh o_J : J \in {\m athbb{S } })$ a p r ob ability vector, so $ \ r ho _J\ge 0$ for $ J \i n{ \mathbb{ S} }$and$ \ sum_{ J\in {\ mathbb{S }}}\rh o _J =1 $, andth at als osat isf ies $ \ rho_ \empty set=0$.
\[de f 1\] Define the following tra n sf o r ma t ion$\X i: {\cal P} _I\t o {\c al P } _I $,$ $\lab el{e4 }\ Xi [ \mu]= \sum_{J\in {\ ma thbb{S }}} \rho_J \, \mu _J\otimes\ m u _{J^c}=\ sum_ { J\ i n {\mathbb{S}} ^{(\e mptyset)}} \rho_J \ , \mu _J\otime s \mu_{J^ c } .$$ $\Bo x$
Si nce $\ m u _\ emptyset\equi v 1$,th en $$\l abe l{e4x}\Xi [\m u]= \r ho _I \, \mu + \sum_ {J \i n{\ mat hbb{S } }^{\empt ys et, I} }
\ rho_J \, \mu _J\ot imes \ mu _ {J^ c}.$$
\ [r e m 0ax\ ]We hav e $ \r ho_\e mpty s et= 0$, but we can h ave $\rh o_ I> 0$. Onthe other han d, since $\m u_ J\o times\ m u_{ | aligned}$$
The above_properties (\[eab0\]),_(\[eab4\]), (\[eab\]), and $\mu^\emptyset$_an identity,_are_all we_need_from $\otimes$ to_get the results_of this work.
Note that_commutation and associativity_imply_that $\bigotimes_{L\in {{\cal D}}}\mu_L\in {\cal P}_I$ is well-defined for a partition ${{\cal D}}$ of_$I$.
For_$\otimes$ the_product_measure_we have $$\forall x\in \prod_{i\in_I}A_i: \quad
\bigotimes_{L\in {{\cal D}}}\mu_L(x)=\prod_{L\in_
{{\cal D}}}\mu_L(x_L).$$_If ${{\cal D}}={{\cal D}}^{si}=\{\{i\}: i\in I\}$, $\bigotimes_{i\in I}_\mu_{\{i\}}$_is called Bernoulli_and $(\mu^{\{i\}}: i\in I)$ are the one-site marginals.
From now_on, we fix $\rho=(\rho_J: J\in {\mathbb{S}})$_a probability vector,_so_$\rho_J\ge_0$ for $J\in {\mathbb{S}}$_and $\sum_{J\in {\mathbb{S}}}\rho_J=1$, and that also_satisfies $\rho_\emptyset=0$.
\[def1\] Define the following transformation_$\Xi: {\cal P}_I\to {\cal P}_I$, $$\label{e4}
\Xi[\mu]= \sum_{J\in_{\mathbb{S}}}
\rho_J \, \mu_J\otimes \mu_{J^c}=\sum_{J\in {\mathbb{S}}^{(\emptyset)}}
\rho_J_\, \mu_J\otimes \mu_{J^c}.$$ $\Box$
Since $\mu_\emptyset\equiv_1$, then_$$\label{e4x}
\Xi[\mu]= \rho_I \, \mu +_\sum_{J\in {\mathbb{S}}^{\emptyset,I}}
\rho_J \,_\mu_J\otimes \mu_{J^c}.$$
\[rem0ax\]_We have $\rho_\emptyset=0$,_but we can have $\rho_I>0$. On_the other hand,_since $\mu_J\otimes \mu_{ |
& &&&1&&\Red{\rightarrow n+1}\\
& & & &+ \Green{n}\ *\quad&{{\textstyle\frac{1}{2}}} & & {{\textstyle\frac{1}{2}}} &\\
& & & & & \Green{0} & & \Green{0}&
\end{array}$$ The supergravity sector contains the graviton, 2 gravitini and a so-called graviphoton. That spin-1 field gets, by coupling to $n$ vector multiplets, part of a set of vectors, which will be uniformly described by the special [Kähler]{} geometry. The scalars appear as complex ones.
To describe this, we start with with scalars. The action is determined by a holomorphic function $F(X)$. Compared with the rigid case, there is one additional requirement. The conformal invariance requires $F(X)$ to be homogeneous of weight 2, where the $X$ fields carry weight 1. These scalar fields transform also under a local $U(1)$ symmetry.
The obtained metric is a cone [@coneGary; @QuatConf]. To see this, one splits the $n+1$ complex variables $\{\Blue{X}\}$ in $\{\rho,\theta, \Green{z^\alpha} \}$
- $r$ is scale which is a gauge degree of freedom for translations
- $\theta $ is the $U(1)$ degree of freedom;
- the $n$ complex variables $\Green{z^\alpha }$.
The metric now takes the form $$ds^2=dr^2 +{{\textstyle\frac{1}{18}}}r^2 \left[A+ d\theta + {{\rm i}}\left(\partial _\alpha K(z,\bar z)\, dz^\alpha -
\partial _{\bar \alpha }K(z,\bar z)\,d\bar z^{\bar \alpha }\right)\right] ^2+
r^2\partial _\alpha \partial _{\bar \alpha }K(z,\bar z)\, dz^\alpha d\bar z^{\bar \alpha
}\,,
\label{dsSasakian}$$ where $A$ is the one-form gauging the $U(1)$ group, and $K(z,\bar z)$ is a function of the holomorphic | & & & & 1&&\Red{\rightarrow n+1}\\
& & & & + \Green{n}\ * \quad&{{\textstyle\frac{1}{2 } } } & & { { \textstyle\frac{1}{2 } } } & \\
& & & & & \Green{0 } & & \Green{0 } &
\end{array}$$ The supergravity sector contains the graviton, 2 gravitini and a so - called graviphoton. That spin-1 field get, by pair to $ n$ vector multiplets, part of a set of vectors, which will be uniformly report by the special [ Kähler ] { } geometry. The scalars look as complex ones.
To describe this, we start with with scalar. The legal action is determined by a holomorphic routine $ F(X)$. Compared with the rigid sheath, there constitute one additional requirement. The conformal invariance requires $ F(X)$ to be homogeneous of weight 2, where the $ X$ field carry weight 1. These scalar field transform also under a local $ U(1)$ symmetry.
The obtain metric is a cone [ @coneGary; @QuatConf ]. To see this, one splits the $ n+1 $ complex variables $ \{\Blue{X}\}$ in $ \{\rho,\theta, \Green{z^\alpha } \}$
- $ r$ is scale which is a bore degree of freedom for translations
- $ \theta $ is the $ U(1)$ degree of freedom;
- the $ n$ complex variables $ \Green{z^\alpha } $.
The system of measurement now takes the form $ $ ds^2 = dr^2 + { { \textstyle\frac{1}{18}}}r^2 \left[A+ d\theta + { { \rm i}}\left(\partial _ \alpha K(z,\bar z)\, dz^\alpha -
\partial _ { \bar \alpha } K(z,\bar z)\,d\bar z^{\bar \alpha } \right)\right ] ^2 +
r^2\partial _ \alpha \partial _ { \bar \alpha } K(z,\bar z)\, dz^\alpha d\bar z^{\bar \alpha
} \, ,
\label{dsSasakian}$$ where $ A$ is the one - form gauging the $ U(1)$ group, and $ K(z,\bar z)$ is a function of the holomorphic | & &&&1&&\Red{\vightarrow n+1}\\
& & & &+ \Green{n}\ *\quad&{{\textstyle\frar{1}{2}}} & & {{\rextsujle\frac{1}{2}}} &\\
& & & & & \Grxsn{0} & & \Nxeen{0}&
\ehf{arrcy}$$ The supergravlty sector wontains the gsaxicon, 2 gravitini and a so-called graviproton. Tnah spin-1 field ggts, bj soupmpnn to $n$ vector multiplets, part of a set mf vectors, whoch will be uniformly descgibef by the special [Käjler]{} geometri. Thq scalars appdar as comklzx ones.
To dgscribe this, we start with with rcalaxs. The actiin is getermined uy a hjlomorphic fmmction $F(X)$. Comlared with the rijid xase, there is one addmtional requirement. Jhe conforkam invariance requurws $F(X)$ to te hunogdnekux kf weihht 2, where the $X$ fields cqrry weight 1. These svajqr fields trahsform ajso under a local $U(1)$ symmetry.
The obtaineg mstric is a cone [@coneGart; @QuatConf]. To see thid, one splyts the $n+1$ complex variables $\{\Blue{X}\}$ in $\{\rho,\theta, \Grean{z^\al'hx} \}$
- $v$ is wcwle which is a gauge degree of freedom for trwhskanions
- $\theta $ is the $U(1)$ degrre og freedom;
- thg $n$ com'mes variables $\Green{z^\wlpha }$.
Tre merric now uakes the form $$ds^2=dr^2 +{{\textstyle\frax{1}{18}}}r^2 \left[A+ d\thvta + {{\rm i}}\left(\partial _\appha K(z,\bar v)\, dz^\alkha -
\lartial _{\bar \alpha }K(z,\bar z)\,d\bzr z^{\bar \alpja }\right)\rjeht] ^2+
r^2\partial _\auphs \[artial _{\bar \alpha }K(z,\bar z)\, qz^\alpha d\uar z^{\yar \alphx
}\,,
\lsbel{dsFasakian}$$ wjere $A$ is the one-form gakging tve $U(1)$ group, and $K(z,\bar z)$ is a function of tix holomorphic | & &&&1&&\Red{\rightarrow n+1}\\ & & & &+ & {{\textstyle\frac{1}{2}}} &\\ & & & \end{array}$$ supergravity sector contains graviton, 2 gravitini a so-called graviphoton. That spin-1 field by coupling to $n$ vector multiplets, part of a set of vectors, which be uniformly described by the special [Kähler]{} geometry. The scalars appear as complex To this, start with scalars. The action is determined by a holomorphic function $F(X)$. Compared with the rigid case, is one additional requirement. The conformal invariance requires to be homogeneous of 2, where the $X$ fields weight These scalar transform under local $U(1)$ symmetry. obtained metric is a cone [@coneGary; @QuatConf]. To see this, one splits the $n+1$ complex variables $\{\Blue{X}\}$ $\{\rho,\theta, \Green{z^\alpha} $r$ is which a degree of freedom - $\theta $ is the $U(1)$ - the $n$ complex variables $\Green{z^\alpha }$. The now takes form $$ds^2=dr^2 +{{\textstyle\frac{1}{18}}}r^2 \left[A+ d\theta + i}}\left(\partial _\alpha K(z,\bar z)\, dz^\alpha - \partial _{\bar }K(z,\bar z)\,d\bar z^{\bar \alpha }\right)\right] ^2+ r^2\partial _\alpha \partial _{\bar \alpha }K(z,\bar z)\, dz^\alpha d\bar }\,, \label{dsSasakian}$$ where $A$ the one-form gauging $U(1)$ and z)$ a function the holomorphic | & &&&1&&\Red{\rightarrow n+1}\\
& & & &+ \Green{n}\ *\quad&{{\tExtstyle\frAc{1}{2}}} & & {{\texTstYle\FrAc{1}{2}}} &\\
& & & & & \GrEen{0} & & \GReen{0}&
\end{array}$$ ThE SupeRgravity sector contains The grAvITon, 2 gRAvItini And a so-cALlED GraViPhOtoN. THAt Spin-1 fIelD gets, by Coupling to $N$ veCtOr multiplets, PArT of a set of vEctOrs, which will Be uNiformLy DesCRibed By tHe speCial [KäHLer]{} geoMetry. The sCaLArs appEAr as comPLEx Ones.
to describe this, we sTArT With with scalarS. The acTiON iS DEteRmiNed by a holoMoRphic FUnction $f(x)$. COMPAreD With the rigid cAse, there is oNE adDitionAl ReqUIremenT. The cOnFOrmAl invariancE reqUires $F(X)$ to Be homoGEneous oF Weight 2, wHere thE $X$ fIelDs caRRy WeIghT 1. THEse SCaLar FIelDs transfOrM aLso unDer a LOCAL $U(1)$ syMmeTry.
THe obtAined metric is A coNe [@coNEGaRy; @QuaTConf]. to seE tHis, onE splitS the $n+1$ CoMplex variables $\{\BLue{X}\}$ In $\{\rho,\thetA, \GrEeN{z^\aLpHa} \}$
- $r$ is SCale whIch Is a Gauge deGree of fREedOm FOR TrAnslations
- $\theta $ is tHe $u(1)$ DEgRee of freEdom;
- thE $N$ cOmPLex variaBlEs $\GReen{Z^\ALpha }$.
THe meTRiC now takeS the foRM $$dS^2=dR^2 +{{\textstYlE\frac{1}{18}}}r^2 \LeFt[A+ D\thEta + {{\rm I}}\Left(\PartiaL _\alpha K(z,\Bar z)\, dZ^\Alpha -
\partial _{\baR \Alpha }K(z,\bar z)\,d\bAR z^{\BAR \aLPha }\rIghT)\right] ^2+
r^2\partIal _\aLPha \pArtiAL _{\bAr \aLPha }K(z,\Bar z)\, dZ^\aLPhA D\bar z^{\bar \alpha
}\,,
\label{dSSAsakiaN}$$ wherE $A$ is the one-forM gauging thE $u(1)$ GRoup, and $K(Z,\bar Z)$ Is A Function of the hOlomoRphic | & & &&1&&\Red{ \righ tar row n +1}\ \
& & & &+ \Green{n }\ *\ qu a d&{{ \ te xtsty le\frac { 1} { 2 }}} & & {{ \t e xt style \fr ac{1}{2 }}} &\\
& & & & & \G reen{0 } && \Gr een {0}&\end{a r ray}$$ The supe rg r avitys ector c o n ta insthe graviton, 2 g r av i tini and a so- called g r av i p hot on. That spin -1 fiel d gets,b yc o u pli n g to $n$ vect or multiple t s,part o fa s e t of v ector s, whi ch will beunif ormly des cribed by thes pecial[Kähle r]{ } g eome t ry .The s c ala r sapp e aras compl ex o nes.
Tod e s c ribe th is,we st art with with sc alar s . T he ac tionis d et ermin ed bya hol om orphic function $F( X)$. Comp are dwit hthe r i gid ca se, th ere isone add i tio na l r eq uirement. The conf or m a linvarian ce req u ir es $F(X)$ t obehomo g e neous ofw ei ght 2, w here t h e$X $ field scarrywe igh t 1 . The s e sc alar f ields tr ansfo r m also under a local $U(1)$s ym m e tr y .
T heobtained me tric is a con e [ @co n eGary ; @Qu at C on f ]. To see this, one s plitsthe $ n+1$ complexvariables$ \ { \Blue{X} \}$i n$ \{\rho,\theta, \Gre en{z^\alph a } \}$
- $r $ is sca le whichi s a gauge de gre e o f f r e ed om for transl a t ions
- $\t het a $ isthe $U (1) $ d eg ree of fr eedom;
- th e$n$ comp l ex varia bl es$\ Gre en{z^ \ alpha}$.
Theme tr i c n ow take s t h e for m$$ ds^2 =dr ^2 +{{\ text s tyl e\frac{ 1}{18}}}r ^2\ left [A +d\theta + {{\rm i}}\ le ft(\partia l_\a lpha K ( z ,\bar z) \, dz^\alpha -
\parti a l _{\ba r \ alpha }K( z,\bar z) \,d \bar z ^{\ b ar \al pha }\ right )\ rig h t ] ^2+ r ^2\ pa rtial _\al p h a \ parti al _{\ bar \al pha }K(z,\bar z)\, dz^ \alpha d\barz^{ \bar \ al pha } \,,
\ l abe l { dsSasakian}$$ w here $A$ i st he one-formg aug in g the $ U(1)$ g roup, and $K( z,\bar z) $ is a fu nc tion o f t he holomor phic | _ _ _ __ __ _ & _ _&&&1&&\Red{\rightarrow n+1}\\
__ & & __ ___ & _ &+ \Green{n}\ *\quad&{{\textstyle\frac{1}{2}}}_& &_{{\textstyle\frac{1}{2}}} &\\
& & __ _ _ & _ _&__ _ &_ \Green{0} & &_ \Green{0}&
\end{array}$$ The supergravity sector contains the_graviton, 2 gravitini and a so-called_graviphoton. That spin-1 field gets,_by coupling_to $n$ vector multiplets, part_of a set_of vectors,_which will be_uniformly described by the special [Kähler]{} geometry._The scalars appear_as complex ones.
To describe this, we_start_with with scalars._The_action_is determined_by a holomorphic_function_$F(X)$. Compared_with_the rigid case, there is one_additional_requirement. The conformal invariance requires $F(X)$ to_be homogeneous of weight 2,_where_the $X$ fields carry_weight 1. These scalar fields transform_also under a local $U(1)$ symmetry.
The_obtained metric_is a_cone [@coneGary; @QuatConf]. To see this, one splits the $n+1$ complex_variables $\{\Blue{X}\}$ in $\{\rho,\theta, \Green{z^\alpha} \}$
-_ $r$ is_scale which_is_a gauge degree_of_freedom for_translations
- $\theta $ is the_$U(1)$ degree_of freedom;
- the $n$_complex variables $\Green{z^\alpha }$.
The_metric_now takes the form $$ds^2=dr^2 +{{\textstyle\frac{1}{18}}}r^2_\left[A+ d\theta + {{\rm i}}\left(\partial _\alpha_K(z,\bar z)\, dz^\alpha -
_\partial__{\bar_\alpha }K(z,\bar z)\,d\bar z^{\bar \alpha_}\right)\right] ^2+
r^2\partial _\alpha \partial__{\bar \alpha }K(z,\bar_z)\, dz^\alpha d\bar z^{\bar \alpha
}\,,
\label{dsSasakian}$$_where_$A$ is the one-form gauging the_$U(1)$_group, and $K(z,\bar z)$ is a_function_of_the holomorphic |
+0.010\pm0.007$ $0.002\pm0.001$
2 0.250 – 0.500 $0.146\pm 0.009$ $0.321\pm0.008$ $-0.022\pm0.010$ $0.019\pm0.002$
3 0.500 – 0.625 $0.108\pm 0.008$ $0.224\pm0.011$ $+0.031\pm0.011$ $0.033\pm0.004$
4 0.625 – 0.750 $0.107\pm 0.008$ $0.157\pm0.010$ $+0.002\pm0.011$ $0.051\pm0.005$
5 0.750 – 0.875 $0.098\pm 0.007$ $0.109\pm0.009$ $-0.028\pm0.011$ $0.060\pm0.005$
6 0.875 – 1.000 $0.144\pm 0.009$ $0.016\pm0.005$ $+0.007\pm0.007$ $0.135\pm0.009$
\[tab:wtag\]
------------------- --------------------------------------- ---------------------------------------------
$\ks\ks\ks$ $\PBksksks$ ($\PAksksks$) $\NBsigksksks$ ($\NAsigksksks$)
$\kspm\kspm\kspm$ $\PBkspmkspmkspm$ ($\PAkspmkspmkspm$) $\NBsigkspmkspmkspm$ ($\NAsigkspmkspmkspm$)
$\kspm\kspm\kszz$ $\PBkspmkspmkszz$ ($\PAkspmkspmkszz$) $\NBsigkspmkspmkszz$ ($\NAsigkspmkspmkszz$)
$\ks\piz\gamma$ $\Pkspizgm$ $\Nsigkspizgm$
------------------- --------------------------------------- ---------------------------------------------
: The estimated signal purity and the signal yield $\nsig$ in the signal | +0.010\pm0.007 $ $ 0.002\pm0.001 $
2 0.250 – 0.500 $ 0.146\pm 0.009 $ $ 0.321\pm0.008 $ $ -0.022\pm0.010 $ $ 0.019\pm0.002 $
3 0.500 – 0.625 $ 0.108\pm 0.008 $ $ 0.224\pm0.011 $ $ +0.031\pm0.011 $ $ 0.033\pm0.004 $
4 0.625 – 0.750 $ 0.107\pm 0.008 $ $ 0.157\pm0.010 $ $ +0.002\pm0.011 $ $ 0.051\pm0.005 $
5 0.750 – 0.875 $ 0.098\pm 0.007 $ $ 0.109\pm0.009 $ $ -0.028\pm0.011 $ $ 0.060\pm0.005 $
6 0.875 – 1.000 $ 0.144\pm 0.009 $ $ 0.016\pm0.005 $ $ +0.007\pm0.007 $ $ 0.135\pm0.009 $
\[tab: wtag\ ]
------------------- --------------------------------------- ---------------------------------------------
$ \ks\ks\ks$ $ \PBksksks$ ($ \PAksksks$) $ \NBsigksksks$ ($ \NAsigksksks$)
$ \kspm\kspm\kspm$ $ \PBkspmkspmkspm$ ($ \PAkspmkspmkspm$) $ \NBsigkspmkspmkspm$ ($ \NAsigkspmkspmkspm$)
$ \kspm\kspm\kszz$ $ \PBkspmkspmkszz$ ($ \PAkspmkspmkszz$) $ \NBsigkspmkspmkszz$ ($ \NAsigkspmkspmkszz$)
$ \ks\piz\gamma$ $ \Pkspizgm$ $ \Nsigkspizgm$
------------------- --------------------------------------- ---------------------------------------------
: The estimated signal purity and the signal yield $ \nsig$ in the signal | +0.010\pm0.007$ $0.002\pm0.001$
2 0.250 – 0.500 $0.146\pm 0.009$ $0.321\pm0.008$ $-0.022\pm0.010$ $0.019\pm0.002$
3 0.500 – 0.625 $0.108\pm 0.008$ $0.224\lm0.011$ $+0.031\pm0.011$ $0.033\pm0.004$
4 0.625 – 0.750 $0.107\pm 0.008$ $0.157\pm0.010$ $+0.002\pl0.011$ $0.051\pm0.005$
5 0.750 – 0.875 $0.098\pm 0.007$ $0.109\pm0.009$ $-0.028\om0.011$ $0.060\pm0.005$
6 0.875 – 1.000 $0.144\pn 0.009$ $0.016\pm0.005$ $+0.007\pm0.007$ $0.135\pm0.009$
\[tab:wtag\]
------------------- --------------------------------------- ---------------------------------------------
$\kv\js\ks$ $\PNksksks$ ($\PAkvksks$) $\NBsigksksks$ ($\NAsigksksks$)
$\kspm\kspm\hspm$ $\LBnspmkspmkspm$ ($\PWkspkhspmispm$) $\NBsigkspmkspmkspm$ ($\NAsigkspmispmkspk$)
$\kspm\kspm\ksaz$ $\PBkspmkspmkszz$ ($\PAkspmnspmnszz$) $\NBsigkspmksomkszz$ ($\NAsitkspijspmkszz$)
$\ks\oiz\gamma$ $\Pkspizgm$ $\Nsiekspivgm$
------------------- --------------------------------------- ---------------------------------------------
: Jkw edjimated signel purpty and the slbnal yheld $\nsog$ in the signsl | +0.010\pm0.007$ $0.002\pm0.001$ 2 0.250 – 0.500 $0.146\pm $-0.022\pm0.010$ 3 0.500 0.625 $0.108\pm 0.008$ – $0.107\pm 0.008$ $0.157\pm0.010$ $0.051\pm0.005$ 5 0.750 0.875 $0.098\pm 0.007$ $0.109\pm0.009$ $-0.028\pm0.011$ $0.060\pm0.005$ 0.875 – 1.000 $0.144\pm 0.009$ $0.016\pm0.005$ $+0.007\pm0.007$ $0.135\pm0.009$ \[tab:wtag\] ------------------- --------------------------------------- --------------------------------------------- $\ks\ks\ks$ ($\PAksksks$) $\NBsigksksks$ ($\NAsigksksks$) $\kspm\kspm\kspm$ $\PBkspmkspmkspm$ ($\PAkspmkspmkspm$) $\NBsigkspmkspmkspm$ ($\NAsigkspmkspmkspm$) $\kspm\kspm\kszz$ $\PBkspmkspmkszz$ ($\PAkspmkspmkszz$) $\NBsigkspmkspmkszz$ ($\NAsigkspmkspmkszz$) $\Pkspizgm$ ------------------- --------------------------------------------- The estimated signal purity and the signal yield $\nsig$ in the signal | +0.010\pm0.007$ $0.002\pm0.001$
2 0.250 – 0.500 $0.146\pm 0.009$ $0.321\pm0.008$ $-0.022\pm0.010$ $0.019\pm0.002$
3 0.500 – 0.625 $0.108\pm 0.008$ $0.224\pm0.011$ $+0.031\pm0.011$ $0.033\pm0.004$
4 0.625 – 0.750 $0.107\pm 0.008$ $0.157\pm0.010$ $+0.002\pM0.011$ $0.051\pm0.005$
5 0.750 – 0.875 $0.098\pm 0.007$ $0.109\pm0.009$ $-0.028\pm0.011$ $0.060\pM0.005$
6 0.875 – 1.000 $0.144\pm 0.009$ $0.016\pm0.005$ $+0.007\Pm0.007$ $0.135\pM0.009$
\[taB:wTag\]
------------------- --------------------------------------- ---------------------------------------------
$\kS\ks\kS$ $\PBksksks$ ($\PAkskSKs$) $\NBSigksksks$ ($\NAsigksksks$)
$\ksPm\kspM\kSPm$ $\PBKSpMkspmKspm$ ($\PAkSPmKSPmkSpM$) $\NbsiGkSPmKspmkSpm$ ($\nAsigksPmkspmkspm$)
$\KspM\kSpm\kszz$ $\PBkspMKsPmkszz$ ($\PAksPmkSpmkszz$) $\NBsigKspMkspmkSzZ$ ($\NASIgkspMksPmkszZ$)
$\ks\piz\GAmma$ $\PkSpizgm$ $\NsiGkSPizgm$
------------------- --------------------------------------- ---------------------------------------------
: THE estimaTED sIgnaL purity and the signAL yIEld $\nsig$ in the siGnal | +0.010\pm0.007$ $0.002\p m0.001$
2 0 .25 0– 0. 500 $0.146\pm 0. 0 09$ $0.321\pm0.008$ $-0 .022\ pm 0 .010 $ $0.0 19\pm0. 0 02 $ 3 0 .5 0 0– 0.6 25 $0.10 8\pm 0.008 $ $0 .224\pm0.011 $ $+0.031\p m0. 011$ $0.03 3\p m0.004 $ 4 0.6 25 –0.750 $0.10 7\pm 0.00 8$ $0.15 7 \pm0.01 0 $ $+0 .002\pm0.011$ $ 0 .0 5 1\pm0.005$
5 0. 7 50 – 0. 875 $0.098\ pm 0.00 7 $ $0.1 0 9\ p m 0 .00 9 $ $-0.028\p m0.011$ $ 0 .06 0\pm0. 00 5$ 6 0.8 75 – 1 .000 $0.1 44\p m 0.009$ $0.01 6 \pm0.00 5 $ $+0 .007\p m0. 007 $ $ 0. 13 5\p m0 . 009 $
\[t a b:w tag\]
- -- ----- ---- - - - - ------ ---- ----- ------------- --- ---- - --- --- - ----- ---- -- ----- ------ ----- -- ---------------
$ \ks\ks\ks $ $\PBk s ksks$($\ PAk sksks$) $ \N Bsigksksks$ ($\NAs ig k s ks ks$)
$ \kspm\ k sp m\ k spm$ $ \P Bks pmks p m kspm$ ($\ P Ak spmkspmk spm$) $ \N Bsigksp mk spmksp m$ ($ \NA sigks p mksp mkspm$ )
$\ks pm\ks p m\kszz$ $\PB k spmkspmkszz$( $\ P A ks p mksp mks zz$) $\NB sigk s pmks pmks z z$ ($ \ NAsig kspmk sp m ks z z$)
$\ks\piz\gamm a$ $ \Pksp izgm$ $\Ns i gk s pizgm$
----- ----- ---------- -------- ----- -------- --------- - - ------ - --- --- --- --- - - -- ------------- - - ---- -- -------
: The es tim ate d s ig nal purit y and th esi gn al yi eld $ \ nsig$ in t hesi gna l | +0.010\pm0.007$ _ $0.002\pm0.001$
_ 2_ __0.250 –_0.500_ $0.146\pm_0.009$ $0.321\pm0.008$_ $-0.022\pm0.010$ _ $0.019\pm0.002$
__ 3 0.500 – 0.625 $0.108\pm 0.008$ $0.224\pm0.011$__ $+0.031\pm0.011$___$0.033\pm0.004$
4_ 0.625 –_0.750 _ $0.107\pm 0.008$ $0.157\pm0.010$ $+0.002\pm0.011$__ $0.051\pm0.005$
_ 5 0.750 – 0.875_ $0.098\pm 0.007$ $0.109\pm0.009$_ $-0.028\pm0.011$___$0.060\pm0.005$
_6 0.875 –_1.000 $0.144\pm 0.009$ _$0.016\pm0.005$ $+0.007\pm0.007$ $0.135\pm0.009$
\[tab:wtag\]
_ ------------------- --------------------------------------- ---------------------------------------------
$\ks\ks\ks$_ _ _ $\PBksksks$ ($\PAksksks$) _ _ _ _ $\NBsigksksks$ ($\NAsigksksks$)
_ $\kspm\kspm\kspm$ _ $\PBkspmkspmkspm$ ($\PAkspmkspmkspm$) $\NBsigkspmkspmkspm$_($\NAsigkspmkspmkspm$)
_ $\kspm\kspm\kszz$ __$\PBkspmkspmkszz$_($\PAkspmkspmkszz$) _ $\NBsigkspmkspmkszz$ ($\NAsigkspmkspmkszz$)
__$\ks\piz\gamma$ __ $\Pkspizgm$ __ _ __ _ _ $\Nsigkspizgm$
-------------------_--------------------------------------- ---------------------------------------------
_ :_ The estimated signal purity and the signal yield $\nsig$ in_the signal |
$\$_{A}$ increases from one to two. With an completely corrupt source, $r=1$, the payoffs become $(\$_{A},\$_{B})=(2,1)$. The reason for this is the same as explained for PD. When the quantum strategies with and without corrupt source are compared to the classical mixed strategy without noise, it is seen that the former ones always give better payoffs to the players. However, when the source becomes noisy (corrupt), classical strategies become more advantageous to quantum ones with increasing corruption rate. The range of corruption rate where classical strategies are better than the quantum strategy if the players stick to their operations $i\hat{\sigma}_{y}$, are $0.2<r<0.5$ and $0.5<r<0.8$ for Alice and Bob, respectively. When $r=1/2$, $\$_{A}=\$_{B}$ and these payoffs are equal to the ones received with classical mixed strategies. While $\$_{A}=\$_{B}$ independently of $r$ for classical mixed strategies, $\$_{A}$ and $\$_{B}$ differ when the players stick to their quantum strategies for $r\neq1/2$. Another interesting result for this game is that, contrary to PD and SD, the strategy $\hat{U}_{A}=\hat{U}_{B}=(\hat{\sigma}_{0}+i\hat{\sigma}_{y})/\sqrt{2}$ discussed above always gives a constant payoff $(3/4,3/4)$, which is better than that of the classical mixed strategy.
[***Scenario II:***]{} In this scenario, the referee knows the characteristics of the corruption in the source, and inform the players on the corruption rate. Then the question is whether the players can find a unique NE for a known source with corruption rate $r$; and if they can, does this NE resolve their dilemma in the game or not. When the corruption rate is $r=1/2$, the state shared between the players become $\hat{\rho}=\hat{I}/4$. Then independent of what action they choose, the players receive constant payoffs determined by averaging the payoff entries
$r$ $~~~~\hat{U}_{A}(\theta_{A},\phi_{A})~~~~$ $~~~~\hat{U}_{B}(\theta_{B},\phi_{B})~~~~$ $(\$_{A},\$_{B})$
-- ------- -------------------------------------------- -------------------------------------------- ---------------------------------- | $ \$_{A}$ increases from one to two. With an completely corrupt source, $ r=1 $, the return become $ (\$_{A},\$_{B})=(2,1)$. The rationality for this is the same as explained for PD. When the quantum strategies with and without crooked source are compared to the classical assorted strategy without noise, it is see that the former ones constantly give better payoffs to the player. However, when the source become noisy (corrupt), classical scheme become more advantageous to quantum ones with increasing corruption pace. The range of corruption pace where classical strategy are better than the quantum scheme if the players stick to their operations $ i\hat{\sigma}_{y}$, are $ 0.2 < r<0.5 $ and $ 0.5 < r<0.8 $ for Alice and Bob, respectively. When $ r=1/2 $, $ \$_{A}=\$_{B}$ and these payoffs are adequate to the ones received with classical mixed strategies. While $ \$_{A}=\$_{B}$ independently of $ r$ for classical mixed strategies, $ \$_{A}$ and $ \$_{B}$ differ when the players stick to their quantum strategies for $ r\neq1/2$. Another interesting result for this game is that, contrary to PD and SD, the strategy $ \hat{U}_{A}=\hat{U}_{B}=(\hat{\sigma}_{0}+i\hat{\sigma}_{y})/\sqrt{2}$ discussed above constantly gives a constant wages $ (3/4,3/4)$, which is better than that of the classical mixed strategy.
[ * * * Scenario II :* * * ] { } In this scenario, the reviewer knows the characteristic of the corruption in the source, and inform the player on the corruption rate. Then the question is whether the players can witness a unique NE for a known source with corruption rate $ r$; and if they can, does this NE resolve their dilemma in the plot or not. When the corruptness rate is $ r=1/2 $, the state shared between the players become $ \hat{\rho}=\hat{I}/4$. Then autonomous of what action they choose, the players receive constant payoffs determined by averaging the payoff entries
$ r$ $ ~~~~\hat{U}_{A}(\theta_{A},\phi_{A})~~~~$ $ ~~~~\hat{U}_{B}(\theta_{B},\phi_{B})~~~~$ $ (\$_{A},\$_{B})$
-- ------- -------------------------------------------- -------------------------------------------- ---------------------------------- |
$\$_{A}$ ijcreases from one to two. With an complejeoy corcupt sohrce, $r=1$, tfe payoffs become $(\$_{A},\$_{B})=(2,1)$. The readob for this is the same as ebplained vor PD. Wyen uhe quantum stratxfies wibk and aithmnt corrupt sourge are compdred to the cldsrieal mixed strategy without noise, it ys seen tjat the former onex alwzjs give better payoffs to the pmayers. Iowever, when thr source becomes noisy (corgupt), classical strategles become nore qdvantageous to quantum ones with jncreasing corruption rate. The fange of corrupjnin gdte where coassibal strategies are bedter thsn the quantum stcatety if the players stirk to their operatiogs $i\hat{\sicmc}_{y}$, are $0.2<r<0.5$ and $0.5<r<0.8$ for Alucw and Bob, resowctkvemy. Wgen $r=1/2$, $\$_{W}=\$_{B}$ end these pzyoffs are wqual to the ones rtceynrd with clasaical iived strategies. While $\$_{A}=\$_{B}$ independently ox $r$ for classical mixed steategies, $\$_{A}$ and $\$_{B}$ diffgr when thq players stick to their quantum strategies for $r\teq1/2$. Aioghex intefwshing result for this game is that, contrary to LD akd SD, the strateny $\hat{U}_{A}=\hat{U}_{B}=(\hat{\sibmw}_{0}+i\nwt{\sigma}_{y})/\sqrt{2}$ aiscussed above always gived a conftant payoff $(3/4,3/4)$, rhicn is better than that of thw classical iuxed strategy.
[***Scenaxio II:***]{} In thns scemario, the referee knows the eharacferistics ov the cordjption in the sojrcv, ang inform the players on thq corruptmon rcte. Then the questyon is whehher bve players can finf a uuique NE for a nnown source with corruption ravx $r$; and if thgy wan, does thns NE vesolve their dylemma in the yame or uot. Whdn the corguption rete is $r=1/2$, the state shared hetween the 'layers bqcomw $\har{\rho}=\hat{K}/4$. Then independrnt of whcu action tyey choose, the planers tedeive constant 'cyiffs determined by avqrwgmng tra payoff entsies
$r$ $~~~~\hat{U}_{A}(\theba_{A},\ohi_{A})~~~~$ $~~~~\hat{U}_{B}(\theta_{B},\phi_{B})~~~~$ $(\$_{A},\$_{B})$
-- ------- -------------------------------------------- -------------------------------------------- ---------------------------------- | $\$_{A}$ increases from one to two. With corrupt $r=1$, the become $(\$_{A},\$_{B})=(2,1)$. The same explained for PD. the quantum strategies and without corrupt source are compared the classical mixed strategy without noise, it is seen that the former ones give better payoffs to the players. However, when the source becomes noisy (corrupt), strategies more to ones with increasing corruption rate. The range of corruption rate where classical strategies are better than quantum strategy if the players stick to their $i\hat{\sigma}_{y}$, are $0.2<r<0.5$ and for Alice and Bob, respectively. $r=1/2$, and these are to ones received with mixed strategies. While $\$_{A}=\$_{B}$ independently of $r$ for classical mixed strategies, $\$_{A}$ and $\$_{B}$ differ when the stick to strategies for Another result this game is to PD and SD, the strategy always gives a constant payoff $(3/4,3/4)$, which is than that the classical mixed strategy. [***Scenario II:***]{} this scenario, the referee knows the characteristics of corruption in the source, and inform the players on the corruption rate. Then the question the players can find unique NE for known with rate and if can, does this NE resolve their dilemma in the game or When the corruption rate is $r=1/2$, the state shared between become Then independent of action they choose, the receive payoffs determined by averaging entries $~~~~\hat{U}_{A}(\theta_{A},\phi_{A})~~~~$ ------- -------------------------------------------- |
$\$_{A}$ increases from one to two. WitH an completEly coRruPt sOuRce, $r=1$, The pAyoffs become $(\$_{A},\$_{B})=(2,1)$. tHe reAson for this is the same as ExplaInED for pd. WHen thE quantuM StRATegIeS wIth AnD WiThout CorRupt souRce are compAreD tO the classicaL MiXed strategY wiThout noise, it Is sEen thaT tHe fORmer oNes AlwayS give bETter paYoffs to thE pLAyers. HOWever, whEN ThE souRce becomes noisy (coRRuPT), classical straTegies BeCOmE MOre AdvAntageous tO qUantuM Ones witH InCREAsiNG corruption raTe. The range oF CorRuptioN rAte WHere clAssicAl STraTegies are beTter Than the quAntum sTRategy iF The playErs stiCk tO thEir oPErAtIonS $i\HAt{\sIGmA}_{y}$, aRE $0.2<r<0.5$ aNd $0.5<r<0.8$ for AlIcE aNd Bob, RespECTIVely. wheN $r=1/2$, $\$_{A}=\$_{B}$ And thEse payoffs are EquAl to THe oNes reCeiveD witH cLassiCal mixEd strAtEgies. While $\$_{A}=\$_{B}$ indEpenDently of $r$ For ClAssIcAl mixED stratEgiEs, $\$_{A}$ And $\$_{B}$ difFer when THe pLaYERS sTick to their quantum StRATeGies for $r\Neq1/2$. AnoTHeR iNTerestinG rEsuLt foR THis gaMe is THaT, contrarY to PD aND Sd, tHe stratEgY $\hat{U}_{A}=\HaT{U}_{B}=(\Hat{\Sigma}_{0}+I\Hat{\sIgma}_{y})/\sQrt{2}$ discuSsed aBOve always gives A Constant payofF $(3/4,3/4)$, WhICH iS BettEr tHan that of thE claSSicaL mixED sTraTEgy.
[***ScEnariO Ii:***]{} in THis scenario, the referEe Knows tHe chaRacteristics oF the corrupTION in the soUrce, ANd INform the playerS on thE corruptioN Rate. Then The quEstion is Whether thE PLayers caN fiNd a UniQue ne FoR a known source WITh coRrUption rAte $R$; and if tHey Can, DoeS thIs nE resolve Their dilEmMa In ThE gaMe or nOT. When the CoRruPtIon Rate iS $R=1/2$, the stAte shAred BeTwEEn tHe playeRS bECOme $\hAt{\RhO}=\hat{i}/4$. ThEn IndepEndeNT of What actIon they chOosE, The pLaYeRs receiVe constant payOfFs determinEd By aVeragiNG The payofF entries
$r$ $~~~~\hat{U}_{A}(\theta_{A},\phi_{a})~~~~$ $~~~~\Hat{U}_{B}(\thEta_{b},\phi_{B})~~~~$ $(\$_{a},\$_{B})$
-- ------- -------------------------------------------- -------------------------------------------- ---------------------------------- |
$\$_{A}$ increases from o ne to two. With an co mp lete ly c orrupt source, $r=1 $, the payoffs become$(\$_ {A } ,\$_ { B} )=(2, 1)$. Th e r e a son f or th is is thesam e as ex plained fo r P D. When the qu a nt um strateg ies with and wi tho ut cor ru pts ource ar e com paredt o theclassical m i xed st r ategy w i t ho ut n oise, it is seent ha t the former on es alw ay s g i v e b ett er payoffs t o the players . H o w e ver , when the sou rce becomes noi sy (co rr upt ) , clas sical s t rat egies becom e mo re advant ageous to quan t um ones withinc rea sing co rr upt io n ra t e. Th e ra nge of c or ru ption rat e w h erecla ssic al st rategies arebet tert han thequant um s tr ategy if th e pla ye rs stick to the ir o perations $i \h at{ \s igma} _ {y}$,are $0 .2<r<0. 5$ and$ 0.5 <r < 0 . 8$ for Alice and Bob ,r e sp ectively . When $r =1 / 2$, $\$_ {A }=\ $_{B } $ andthes e p ayoffs a re equ a lto the on es recei ve d w ith clas s ical mixed strateg ies.W hile $\$_{A}=\ $ _{B}$ indepen d en t l yo f $r $ f or classica l mi x ed s trat e gi es, $\$_{ A}$ a nd $\ $ _{B}$ differ when t he playe rs st ick to theirquantum st r a t egies fo r $r \ ne q 1/2$. Anotherinter esting res u lt for t his g ame is t hat, cont r a ry to PD an d S D,the s tr ategy $\hat{U } _ {A}= \h at{U}_{ B}= (\hat{\ sig ma} _{0 }+i \h at{\sigma }_{y})/\ sq rt {2 }$ di scuss e d aboveal way sgiv es ac onstan t pay off$( 3/ 4 ,3/ 4)$, wh i ch i s be tt er tha n t ha t ofthec las sical m ixed stra teg y .
[ ** *S cenario II:***]{} In t his scenar io , t he ref e r ee knows the characteristics of the cor rup tionin t he source , a nd inf orm the pl ayerson th ecor r u ption r at e.Th en the que s t ion is w he ther the pl ayers can find a u n iqu e NE for a kn own sou r c ewit h c o rru pt i onr a te $r$; and ifthey can,do e sthis NE re s olv etheir d ilemmain th e game o r not. Wh en the co rr upti o n ra te is $r=1 /2$, the state sh a red b e tw een t heplayer sbec ome $ \hat{\ r ho} =\hat {I}/4$ .Then i ndepe nd ent of w hat action they choose, the p layer s r eceive co nst a ntpayoffs d eter mined by a ver agi ng th e p a yoffentr i es
$r$ $~~~~\h a t{ U}_ { A }( \theta_{A}, \ p h i_{ A})~~ ~~$ $~~~ ~\ha t{U}_{B}(\theta_{ B },\phi_{B})~~~ ~$ $ (\$_ {A },\$_{B})$
---- ----- ----- -------- -- - ----- ------ ------ ------- - - -- ------ ---- --- --------- --- -- - ------- -- -- - ---- - ---- -- ------ ------ - ---- - - -------- |
$\$_{A}$ increases_from one_to two. With an_completely corrupt_source,_$r=1$, the_payoffs_become $(\$_{A},\$_{B})=(2,1)$. The_reason for this_is the same as_explained for PD._When_the quantum strategies with and without corrupt source are compared to the classical mixed_strategy_without noise,_it_is_seen that the former ones_always give better payoffs to_the players._However, when the source becomes noisy (corrupt), classical_strategies_become more advantageous_to quantum ones with increasing corruption rate. The range_of corruption rate where classical strategies_are better than_the_quantum_strategy if the players_stick to their operations $i\hat{\sigma}_{y}$, are_$0.2<r<0.5$ and $0.5<r<0.8$ for Alice and_Bob, respectively. When $r=1/2$, $\$_{A}=\$_{B}$ and these_payoffs are equal to the ones_received with classical mixed strategies._While $\$_{A}=\$_{B}$_independently of $r$ for classical_mixed strategies, $\$_{A}$_and $\$_{B}$_differ when the_players stick to their quantum strategies_for $r\neq1/2$. Another_interesting result for this game is_that,_contrary to PD_and_SD,_the strategy_$\hat{U}_{A}=\hat{U}_{B}=(\hat{\sigma}_{0}+i\hat{\sigma}_{y})/\sqrt{2}$ discussed above_always_gives a_constant_payoff $(3/4,3/4)$, which is better than_that_of the classical mixed strategy.
[***Scenario II:***]{} In_this scenario, the referee_knows_the characteristics of the_corruption in the source, and_inform the players on the corruption_rate. Then_the question_is whether the players can find a unique NE for a_known source with corruption rate $r$;_and if they can,_does this_NE_resolve their dilemma_in_the game_or not. When the corruption rate is_$r=1/2$, the_state shared between the players become_$\hat{\rho}=\hat{I}/4$. Then independent of_what_action they choose, the players receive_constant payoffs determined by averaging the_payoff entries
___ $r$ _$~~~~\hat{U}_{A}(\theta_{A},\phi_{A})~~~~$ $~~~~\hat{U}_{B}(\theta_{B},\phi_{B})~~~~$ _ _ $(\$_{A},\$_{B})$ __ __
-- -------_--------------------------------------------_--------------------------------------------_---------------------------------- |
1,2,-4,1)$, we find that $$\begin{gathered}
F(a):=
{ \begingroup
\begingroup\lccode`~=`,
\lowercase{\endgroup\def~}{\pFcomma\mkern\pFqskip} \mathcode`,=\string"8000
{}_{}F_{1}\biggl(\genfrac..{0pt}{}{a;2a,1-4a}{a+1/2};\frac{80}{81},
\frac{16}{15}\biggr) \endgroup
}\end{gathered}$$ has a closed form, and it satisfies the following closed-form relation: $$\begin{gathered}
\frac{F(a+1)}{F(a)}=\frac{3^4}{5^4}.
\label{ex2_closed1}\end{gathered}$$ This implies that $$\begin{gathered}
{F(a)}=\frac{5^{4n}}{3^{4n}}{F(a+n)}
\label{ex2_closed2}\end{gathered}$$ holds for any non-negative integer $n$. Although (\[ex2\_closed2\]) (and (\[ex2\_closed1\])) is valid by virtue of analytic continuation, $F(a)$ and $F(a+n)$, which are regarded as infinite double series expressions, are meaningless. For this reason, we carry out a reduction of each of these to a finite sum of a single series expression that is meaningful. This is done, for example, by substituting $a=1/4$ into (\[ex2\_closed2\]). Then, it becomes $$\begin{gathered}
{ \begingroup
\begingroup\lccode`~=`,
\lowercase{\endgroup\def~}{\pFcomma\mkern\pFqskip} \mathcode`,=\string"8000
{}_{2}F_{1}\biggl(\genfrac..{0pt}{}{1/4, 1/2}{3/4};\frac{80}{81}\biggr) \endgroup
}
=
\frac{5^{4n}}{3^{4n}}
{ \begingroup
\begingroup\lccode`~=`,
\lowercase{\endgroup\def~}{\pFcomma | 1,2,-4,1)$, we find that $ $ \begin{gathered }
F(a):=
{ \begingroup
\begingroup\lccode`~= `,
\lowercase{\endgroup\def~}{\pFcomma\mkern\pFqskip } \mathcode`,=\string"8000
{ } _ { } F_{1}\biggl(\genfrac.. {0pt}{}{a;2a,1 - 4a}{a+1/2};\frac{80}{81 },
\frac{16}{15}\biggr) \endgroup
} \end{gathered}$$ has a closed form, and it meet the take after closed - form relation: $ $ \begin{gathered }
\frac{F(a+1)}{F(a)}=\frac{3 ^ 4}{5 ^ 4 }.
\label{ex2_closed1}\end{gathered}$$ This imply that $ $ \begin{gathered }
{ F(a)}=\frac{5^{4n}}{3^{4n}}{F(a+n) }
\label{ex2_closed2}\end{gathered}$$ holds for any non - negative integer $ n$. Although (\[ex2\_closed2\ ]) (and (\[ex2\_closed1\ ]) ) is valid by merit of analytic continuation, $ F(a)$ and $ F(a+n)$, which are regarded as infinite bivalent serial expressions, are meaningless. For this reason, we hold out a reduction of each of these to a finite sum of a single series expression that is meaningful. This is done, for model, by substituting $ a=1/4 $ into (\[ex2\_closed2\ ]). Then, it becomes $ $ \begin{gathered }
{ \begingroup
\begingroup\lccode`~= `,
\lowercase{\endgroup\def~}{\pFcomma\mkern\pFqskip } \mathcode`,=\string"8000
{ } _ { 2}F_{1}\biggl(\genfrac.. {0pt}{}{1/4, 1/2}{3/4};\frac{80}{81}\biggr) \endgroup
}
=
\frac{5^{4n}}{3^{4n } }
{ \begingroup
\begingroup\lccode`~= `,
\lowercase{\endgroup\def~}{\pFcomma | 1,2,-4,1)$, we find that $$\begin{gathered}
N(a):=
{ \begingroup
\ywgingrmup\lcckde`~=`,
\luwercase{\endgroup\def~}{\pFcomma\mkxrn\pDqskik} \mathcode`,=\string"8000
{}_{}W_{1}\biggl(\genvrac..{0pt}{}{a;2a,1-4q}{a+1/2};\frec{80}{81},
\frac{16}{15}\biggr) \endjdoup
}\end{ncthersf}$$ hav a closed form, and it sadisfies the fonluwnng closed-form relation: $$\begin{gathereq}
\frac{F(a+1)}{G(a)}=\vrac{3^4}{5^4}.
\label{ex2_clofed1}\emq{gatgvrtd}$$ This implies that $$\begin{gatheres}
{F(a)}=\frac{5^{4i}}{3^{4n}}{F(a+n)}
\label{ex2_cloxed2}\end{gathered}$$ holds for ajy nln-negative integer $n$. Although (\[ex2\_cjised2\]) (and (\[ex2\_cuosed1\])) is valid by virtoe of analytic continuation, $F(a)$ avd $F(a+u)$, which are rwgagged as infiiite djuble series expresshons, arr meaningless. Nor tiis eeason, we carry out a reduction of each of these tk a finite sum of a singne sarier exoreasmon that ls jeaningful. This is dobe, for example, by sibfnotuting $a=1/4$ info (\[ex2\_cjofed2\]). Then, it becomes $$\begin{gathered}
{ \begitgrkup
\begingroup\lccode`~=`,
\lowercase{\endgroup\dev~}{\pFcomma\mhern\pFqskip} \mathcode`,=\string"8000
{}_{2}F_{1}\biggl(\genfrac..{0pt}{}{1/4, 1/2}{3/4};\frac{80}{81}\tiggr) \enbnvoup
}
=
\weaf{5^{4n}}{3^{4n}}
{ \begingroup
\begingroup\lccode`~=`,
\lowercasq{\sncggoup\def~}{\pFcomma | 1,2,-4,1)$, we find that $$\begin{gathered} F(a):= { \lowercase{\endgroup\def~}{\pFcomma\mkern\pFqskip} {}_{}F_{1}\biggl(\genfrac..{0pt}{}{a;2a,1-4a}{a+1/2};\frac{80}{81}, \frac{16}{15}\biggr) }\end{gathered}$$ has a the closed-form relation: $$\begin{gathered} \label{ex2_closed1}\end{gathered}$$ This implies $$\begin{gathered} {F(a)}=\frac{5^{4n}}{3^{4n}}{F(a+n)} \label{ex2_closed2}\end{gathered}$$ holds for any integer $n$. Although (\[ex2\_closed2\]) (and (\[ex2\_closed1\])) is valid by virtue of analytic continuation, and $F(a+n)$, which are regarded as infinite double series expressions, are meaningless. For reason, carry a of each of these to a finite sum of a single series expression that is meaningful. is done, for example, by substituting $a=1/4$ into Then, it becomes $$\begin{gathered} \begingroup \begingroup\lccode`~=`, \lowercase{\endgroup\def~}{\pFcomma\mkern\pFqskip} \mathcode`,=\string"8000 {}_{2}F_{1}\biggl(\genfrac..{0pt}{}{1/4, \endgroup = \frac{5^{4n}}{3^{4n}} \begingroup \lowercase{\endgroup\def~}{\pFcomma | 1,2,-4,1)$, we find that $$\begin{gathered}
F(a):=
{ \Begingroup
\BeginGroUp\lCcOde`~=`,
\lOwerCase{\endgroup\deF~}{\PFcoMma\mkern\pFqskip} \mathcodE`,=\striNg"8000
{}_{}f_{1}\BiggL(\GeNfrac..{0Pt}{}{a;2a,1-4a}{a+1/2};\fRAc{80}{81},
\FRAc{16}{15}\bIgGr) \EndGrOUp
}\End{gaTheRed}$$ has a Closed form, And It Satisfies the FOlLowing closEd-fOrm relation: $$\bEgiN{gatheReD}
\frAC{F(a+1)}{F(a)}=\FraC{3^4}{5^4}.
\labeL{ex2_cloSEd1}\end{gAthered}$$ ThIs IMplies THat $$\begiN{GAtHereD}
{F(a)}=\frac{5^{4n}}{3^{4n}}{F(a+n)}
\label{EX2_cLOsed2}\end{gathereD}$$ holds FoR AnY NOn-nEgaTive integeR $n$. althoUGh (\[ex2\_cloSEd2\]) (AND (\[Ex2\_cLOsed1\])) is valid by Virtue of anaLYtiC contiNuAtiON, $F(a)$ and $f(a+n)$, whIcH Are Regarded as iNfinIte double Series EXpressiONs, are meAninglEss. for This REaSoN, we CaRRy oUT a RedUCtiOn of each Of ThEse to A finITE SUm of A siNgle SerieS expression thAt iS meaNIngFul. ThIs is dOne, fOr ExampLe, by suBstitUtIng $a=1/4$ into (\[ex2\_closeD2\]). TheN, it becomeS $$\beGiN{gaThEred}
{ \bEGingroUp
\bEgiNgroup\lCcode`~=`,
\loWErcAsE{\ENDgRoup\def~}{\pFcomma\mkerN\pfQSkIp} \mathcoDe`,=\striNG"8000
{}_{2}F_{1}\BiGGl(\genfraC..{0pT}{}{1/4, 1/2}{3/4};\frAc{80}{81}\biGGR) \endgRoup
}
=
\FRaC{5^{4n}}{3^{4n}}
{ \beginGroup
\bEGiNgRoup\lccOdE`~=`,
\lowerCaSe{\eNdgRoup\dEF~}{\pFcOmma | 1,2,-4,1)$, we find that $ $\begin{ga there d}F(a ): =
{ \be gingroup
\be g ingr oup\lccode`~=`,
\l owerc as e {\en d gr oup\d ef~}{\p F co m m a\m ke rn \pF qs k ip } \m ath code`,= \string"80 00 {}_{}F_{1}\b i gg l(\genfrac ..{ 0pt}{}{a;2a, 1-4 a}{a+1 /2 };\ f rac{8 0}{ 81},\frac{ 1 6}{15} \biggr) \e n dgroup }\end{g a t he red} $$ has a closed f o rm , and it satisf ies th ef ol l o win g c losed-form r elati o n: $$\b e gi n { g ath e red}
\frac{F( a+1)}{F(a)} = \fr ac{3^4 }{ 5^4 } .
\lab el{ex 2_ c los ed1}\end{ga ther ed}$$ Thi s impl i es that $$\begi n{gath ere d}{F(a ) }= \f rac {5 ^ {4n } }{ 3^{ 4 n}} {F(a+n)}
\ la bel{e x2_c l o s e d2}\ end {gat hered }$$ holds for an y no n -ne gativ e int eger $ n$. A lthoug h (\[ ex 2\_closed2\]) ( and(\[ex2\_c los ed 1\] )) is v a lid by vi rtu e of an alyticc ont in u a t io n, $F(a)$ and $F(a +n ) $ ,which ar e rega r de da s infini te do uble s eries exp r es sions, a re mea n in gl ess. Fo rthis r ea son , w e car r y ou t a re ductionof ea c h of these toa finite sum o f a s in g le s eri es expressi on t h at i s me a ni ngf u l. Th is is d o ne , for example, by su bs tituti ng $a =1/4$ into (\ [ex2\_clos e d 2 \]). The n, i t b e comes $$\begin {gath ered}
{ \ b egingrou p
\ begingro up\lccode ` ~ =`,
\lo wer cas e{\ e n dg roup\def~}{\p F c omma \m kern\pF qsk ip} \m ath cod e`, =\s tr ing"8000 {}_{2} F_ {1 }\ bi ggl (\gen f rac..{0p t} {}{ 1/ 4,1/2}{ 3 /4};\f rac{8 0}{8 1} \b i ggr ) \end g ro u p
}
=
\ fr ac{5 ^{4 n} }{3^{ 4n}} { \beging roup
\b egi n grou p\ lc code`~= `,
\lower ca se{\endgro up \de f~}{\p F c omma | 1,2,-4,1)$, we_find that_$$\begin{gathered}
F(a):=
{ \begingroup
_\begingroup\lccode`~=`,
__ \lowercase{\endgroup\def~}{\pFcomma\mkern\pFqskip}__\mathcode`,=\string"8000
{}_{}F_{1}\biggl(\genfrac..{0pt}{}{a;2a,1-4a}{a+1/2};\frac{80}{81},
\frac{16}{15}\biggr)_ \endgroup
}\end{gathered}$$ has_a closed form, and_it satisfies the_following_closed-form relation: $$\begin{gathered}
\frac{F(a+1)}{F(a)}=\frac{3^4}{5^4}.
\label{ex2_closed1}\end{gathered}$$ This implies that $$\begin{gathered}
{F(a)}=\frac{5^{4n}}{3^{4n}}{F(a+n)}
\label{ex2_closed2}\end{gathered}$$ holds for any non-negative integer $n$. Although_(\[ex2\_closed2\])_(and (\[ex2\_closed1\]))_is_valid_by virtue of analytic continuation,_$F(a)$ and $F(a+n)$, which are_regarded as_infinite double series expressions, are meaningless. For this_reason,_we carry out_a reduction of each of these to a finite_sum of a single series expression_that is meaningful._This_is_done, for example, by_substituting $a=1/4$ into (\[ex2\_closed2\]). Then, it_becomes $$\begin{gathered}
{ \begingroup
\begingroup\lccode`~=`,
_ \lowercase{\endgroup\def~}{\pFcomma\mkern\pFqskip} \mathcode`,=\string"8000
_{}_{2}F_{1}\biggl(\genfrac..{0pt}{}{1/4, 1/2}{3/4};\frac{80}{81}\biggr) \endgroup
}
=
\frac{5^{4n}}{3^{4n}}
{ \begingroup
_ \begingroup\lccode`~=`,
_\lowercase{\endgroup\def~}{\pFcomma |
R_g(x^\prime)\leq R_g(x)) \rho(x^\prime |\lambda)$, where $R_g(x)$ is radius of gyration for the polymer configuration $x$ and where $\Theta(a \leq b)$ is unity when the inequality in the argument is satisfied and zero otherwise. With this definition, $f(x)$ is the probability that another configuration drawn at random from $\rho$ has a larger radius of gyration than the configuration $x$. With all these parameters in place, eqn. 2 was integrated with a modified Euler scheme using a time-step of 0.25.
We checked that simply setting all the energy parameters to large, identical values will not produce an octahedron. This comes about by tuning a Lennard-Jones potential $V=4((\frac{a}{r})^{12}-(\frac{a}{r})^6)$ with a cutoff distance, $r_c$, such that $V=0$ for any $r\ge r_c$. We shift the raw Lennard-Jones potential by its value at $r_c$ to keep the potential continuous.
The specific value for $r_c$ was picked by ensuring that, in the octahedron geometry, the only contribution to the particle energy would come from contacting particles. This was achieved by setting the cutoff parameter to be $r_c = \sqrt{2} D$. Since the polytetrahedron geometry has the exact same number of particle contacts (12 contacts) and no two points closer than $r_c$, these two geometries will have exactly the same energy given the short ranged potential.
Methodology for the non-equilibrium particle on a substrate
-----------------------------------------------------------
Here we show how to derive an optimizer that works on an out-of-equilibrium problem. This particular optimizer treats the problem of a particle walking randomly on a substrate and takes the random path traveled by the particle as its configuration space. To model the process, we discretize the path into $N$ steps each spaced in time by a small interval $\tau$. We pick $N$ and $\tau$ so that $N\tau=1$ and set up the problem so that the random walk takes place over a time interval $0<t<1$. The path is generated by adding white noise with variance of $\tau kT_i$ and a mean drift of $\vec{v}_i$ to update | R_g(x^\prime)\leq R_g(x) ) \rho(x^\prime |\lambda)$, where $ R_g(x)$ is radius of gyration for the polymer configuration $ x$ and where $ \Theta(a \leq b)$ is unity when the inequality in the controversy is quenched and zero otherwise. With this definition, $ f(x)$ is the probability that another configuration trace at random from $ \rho$ has a large radius of gyration than the shape $ x$. With all these parameters in place, eqn. 2 was integrated with a modify Euler scheme using a prison term - step of 0.25.
We checked that just adjust all the energy parameters to large, identical values will not produce an octahedron. This total about by tuning a Lennard - Jones potential $ V=4((\frac{a}{r})^{12}-(\frac{a}{r})^6)$ with a shortcut distance, $ r_c$, such that $ V=0 $ for any $ r\ge r_c$. We shift the raw Lennard - Jones electric potential by its value at $ r_c$ to keep the likely continuous.
The specific value for $ r_c$ was picked by ensuring that, in the octahedron geometry, the only contribution to the particle energy would come from reach particles. This was achieved by setting the cutoff parameter to be $ r_c = \sqrt{2 } D$. Since the polytetrahedron geometry has the exact same number of particle contacts (12 contacts) and no two points closer than $ r_c$, these two geometries will have exactly the like department of energy given the short ranged electric potential.
Methodology for the non - equilibrium atom on a substrate
-----------------------------------------------------------
Here we show how to derive an optimizer that works on an out - of - equilibrium problem. This particular optimizer treats the trouble of a particle walking randomly on a substrate and takes the random path traveled by the particle as its configuration space. To model the process, we discretize the path into $ N$ steps each spaced in meter by a small time interval $ \tau$. We blame $ N$ and $ \tau$ so that $ N\tau=1 $ and set up the problem so that the random walk takes place over a time interval $ 0 < t<1$. The way is generated by add white noise with variability of $ \tau kT_i$ and a mean drift of $ \vec{v}_i$ to update | R_g(x^\orime)\leq R_g(x)) \rho(x^\prime |\lxmbda)$, where $R_g(x)$ is radmus of fyration for the polymer configuratiln $x$ ane where $\Theta(a \leq b)$ ir unity wjen the uneqnality in the arjhment is satianied cnv zero otherwisg. With this gefinition, $f(x)$ hs tke probability that another configurwtion dtaan at random ftom $\rnj$ haa a larger radius of gyration than the coifiguration $x$. Woth all these parameters ij plwce, eqn. 2 was integgated with q moqufied Euler rcheme using a time-stek of 0.25.
We checked that simply settkng akl the enetyt pwtameters to oarge, identical values winl not lroduce an octshevron. This comes about by vuning a Lennard-Jonef potentidl $V=4((\frac{a}{r})^{12}-(\frac{a}{r})^6)$ wirh a cujoff gistxbce, $r_c$, snch that $G=0$ fkr any $r\ge r_c$. We shifr the raw Lennard-Jomef potential by its vwlte at $r_c$ to keep the potential continuols.
Ths specific value for $r_c$ was picked by ensurijg that, ig the octahedron geometry, the only contribution tm the oaruigjd ejergy would come from contacting particles. Thya eax achieved by fetting the cktpsf parameter jo be $r_e = \aqrt{2} D$. Since the pllytetrwhedrin geometwy hss the exact same number of particle coutaxts (12 contacts) and uo two pointr clpser yhan $r_c$, these two geomecries sill have edactly ths same energy givdn nhe vhort ranged potential.
Methjdology fir tke non-eqjilinrium [article oj a smtstrate
-----------------------------------------------------------
Here we shoa how tm derive aj optimizer that works on an ouv-pf-equilibriuk [rotlem. Thif parbicular optimizqr treats the kroblem oy a pafticle waliing raidomly on a fubstrate and jakes the ranvom path eravwled by the oarticle as itx configugauion space. To model the proccss, wg siscretize the 'cty into $N$ steps racf s[abed ig time by a skall ingrrval $\tau$. We picj $N$ amd $\tau$ so that $N\tau=1$ dnd aet up the problem sj that tye randoi walk takes llace over a time lntertal $0<t<1$. Vhe payh ys generated by adding white nkise with vaviance of $\tau kT_i$ and a mean brift of $\vec{v}_i$ to update | R_g(x^\prime)\leq R_g(x)) \rho(x^\prime |\lambda)$, where $R_g(x)$ is gyration the polymer $x$ and where when inequality in the is satisfied and otherwise. With this definition, $f(x)$ is probability that another configuration drawn at random from $\rho$ has a larger radius gyration than the configuration $x$. With all these parameters in place, eqn. 2 integrated a Euler using a time-step of 0.25. We checked that simply setting all the energy parameters to large, values will not produce an octahedron. This comes by tuning a Lennard-Jones $V=4((\frac{a}{r})^{12}-(\frac{a}{r})^6)$ with a cutoff distance, such $V=0$ for $r\ge We the raw Lennard-Jones by its value at $r_c$ to keep the potential continuous. The specific value for $r_c$ was picked ensuring that, octahedron geometry, only to particle energy would contacting particles. This was achieved by parameter to be $r_c = \sqrt{2} D$. Since polytetrahedron geometry the exact same number of particle (12 contacts) and no two points closer than these two geometries will have exactly the same energy given the short ranged potential. Methodology non-equilibrium particle on a ----------------------------------------------------------- Here we how derive optimizer works on out-of-equilibrium problem. This particular optimizer treats the problem of a particle randomly on a substrate and takes the random path traveled particle its configuration space. model the process, we the into $N$ steps each time a We $N$ $\tau$ so that $N\tau=1$ set up the problem so the random walk takes $0<t<1$. The path is generated by adding white with variance of $\tau kT_i$ and a drift of $\vec{v}_i$ to update | R_g(x^\prime)\leq R_g(x)) \rho(x^\prime |\laMbda)$, where $R_G(x)$ is rAdiUs oF gYratIon fOr the polymer coNFiguRation $x$ and where $\Theta(a \lEq b)$ is UnITy whEN tHe ineQuality IN tHE ArgUmEnT is SaTIsFied aNd zEro otheRwise. With tHis DeFinition, $f(x)$ is THe ProbabilitY thAt another conFigUratioN dRawN At ranDom From $\rHo$ has a LArger rAdius of gyRaTIon thaN The confIGUrAtioN $x$. With all these parAMeTErs in place, eqn. 2 wAs inteGrATeD WIth A moDified EuleR sCheme USing a tiME-sTEP Of 0.25.
WE Checked that siMply setting ALl tHe enerGy ParAMeters To larGe, IDenTical values Will Not producE an octAHedron. THIs comes About bY tuNinG a LeNNaRd-jonEs POteNTiAl $V=4((\FRac{A}{r})^{12}-(\frac{a}{r})^6)$ WiTh A cutoFf diSTANCe, $r_c$, SucH thaT $V=0$ for Any $r\ge r_c$. We shiFt tHe raW lenNard-JOnes pOtenTiAl by iTs valuE at $r_c$ To Keep the potentiaL conTinuous.
ThE spEcIfiC vAlue fOR $r_c$ was PicKed By ensurIng that, IN thE oCTAHeDron geometry, the onlY cONTrIbution tO the paRTiClE Energy woUlD coMe frOM ContaCtinG PaRticles. THis was AChIeVed by seTtIng the CuTofF paRametER to bE $r_c = \sqrT{2} D$. Since tHe polYTetrahedron geoMEtry has the exaCT sAME nUMber Of pArticle contActs (12 COntaCts) aND nO twO PointS closEr THaN $R_c$, these two geometrieS wIll havE exacTly the same eneRgy given thE SHOrt rangeD potENtIAl.
Methodology fOr the Non-equilibRIum partiCle on A substraTe
-----------------------------------------------------------
Here we sHOW how to deRivE an OptImiZER tHat works on an oUT-Of-eqUiLibrium ProBlem. ThiS paRtiCulAr oPtImizer treAts the prObLeM oF a ParTicle WAlking raNdOmlY oN a sUbstrATe and tAkes tHe raNdOm PAth TraveleD By THE parTiClE as iTs cOnFigurAtioN SpaCe. To modEl the procEss, WE disCrEtIze the pAth into $N$ steps EaCh spaced in TiMe bY a smalL INterval $\tAu$. We pick $N$ and $\tau$ so that $N\tAU=1$ and set Up tHe proBlem So that the RanDom walK taKEs placE over a Time iNtErvAL $0<T<1$. The pATH iS geNeRated by addING whIte noIsE witH varianCe of $\tau kT_i$ and a mean DRifT of $\vec{v}_i$ to updAte | R_g(x^\prime)\leq R_g(x))\rho(x^\pr ime | \la mbd a) $, w here $R_g(x)$ is r a dius of gyration for the p olyme rc onfi g ur ation $x$ an d w h e re$\ Th eta (a \l eq b) $ i s unity when theine qu ality in the ar gument issat isfied and z ero other wi se. Withthi s def initio n , $f(x )$ is the p r obabil i ty that a no ther configuration dr a wn at random from $\rho $h as a la rge r radius o fgyrat i on than th e c onf i guration $x$. With all t h ese param et ers in pla ce, e qn . 2was integra tedwith a mo dified Euler s c heme us ing atim e-s tepo f0. 25.
W e c h ec ked tha t simply s et tingallt h e ener gypara meter s to large, i den tica l va lueswillnotpr oduce an oc tahed ro n. This comes a bout by tunin g a L enn ar d-Jon e s pote nti al$V=4((\ frac{a} { r}) ^{ 1 2 } -( \frac{a}{r})^6)$ w it h acutoff d istanc e ,$r _ c$, such t hat $V= 0 $ forany$ r\ ge r_c$. We sh i ft t he rawLe nnard- Jo nes po tenti a l by its v alue at$r_c$ to keep the po t ential contin u ou s .
T he s pec ific valuefor$ r_c$ was pi cke d by e nsuri ng th a t, in the octahedro ngeomet ry, t he only contr ibution to t h e partic le e n er g y would come f rom c ontactingp articles . Thi s was ac hieved by s etting t hecut off pa r a me ter to be $r_ c = \s qr t{2} D$ . S ince th e p oly tet rah ed ron geome try hasth eex ac t s ame n u mber ofpa rti cl e c ontac t s (12conta cts) a nd notwo poi n ts c lose rth an $ r_c $, thes e tw o ge ometrie s will ha vee xact ly t he same energy given t he short r an ged poten t i al.
Met hodology for the non-eq u ilibriu m p artic le o n a subst rat e
---- --- - ------ ------ ----- -- --- - - ----- - - -- --- -- ---------- - - -
Herewe sho w how t o derive an optimi z erthat works on an out - o f- equ i li b riu mp rob l e m. This particu lar optimi ze r t reats thep rob le m of aparticl e wal k ing ran domly ona substra te and t ake s the rand om pathtraveledb y the pa rticl e a s itsco nfi gurat ion sp a ce. To m odel t he proce ss, w ediscreti ze the path into $N$ st eps ea ch sp ace d in time by a s mall inte rval $\tau$. W e p ick $N$and $\tau $ so th at$ N\tau =1$a nd set up th e p r o bl em so thatt h e ra ndomwal k takes pla ce over a time in t erval $0<t<1$. The p ath is gene ra ted by addingwhi te n oise wit hvariance of $\tau k T_ i $ and a mea n drif t of $\ v e c{ v }_i$ t o up dat e | R_g(x^\prime)\leq R_g(x))_\rho(x^\prime |\lambda)$,_where $R_g(x)$ is radius_of gyration_for_the polymer_configuration_$x$ and where_$\Theta(a \leq b)$_is unity when the_inequality in the_argument_is satisfied and zero otherwise. With this definition, $f(x)$ is the probability that another_configuration_drawn at_random_from_$\rho$ has a larger radius_of gyration than the configuration_$x$. With_all these parameters in place, eqn. 2 was_integrated_with a modified_Euler scheme using a time-step of 0.25.
We checked that_simply setting all the energy parameters_to large, identical_values_will_not produce an octahedron._This comes about by tuning a_Lennard-Jones potential $V=4((\frac{a}{r})^{12}-(\frac{a}{r})^6)$ with a cutoff_distance, $r_c$, such that $V=0$ for any_$r\ge r_c$. We shift the raw_Lennard-Jones potential by its value_at $r_c$_to keep the potential continuous.
The_specific value for_$r_c$ was_picked by ensuring_that, in the octahedron geometry, the_only contribution to_the particle energy would come from_contacting_particles. This was_achieved_by_setting the_cutoff parameter to_be_$r_c =_\sqrt{2}_D$. Since the polytetrahedron geometry has_the_exact same number of particle contacts (12_contacts) and no two_points_closer than $r_c$, these_two geometries will have exactly_the same energy given the short_ranged potential.
Methodology_for the_non-equilibrium particle on a substrate
-----------------------------------------------------------
Here we show how to derive an_optimizer that works on an out-of-equilibrium_problem. This particular optimizer_treats the_problem_of a particle_walking_randomly on_a substrate and takes the random path_traveled by_the particle as its configuration space._To model the process,_we_discretize the path into $N$ steps_each spaced in time by a_small interval $\tau$. We pick_$N$_and_$\tau$ so that $N\tau=1$ and_set up the problem so that_the random walk_takes place over a time interval $0<t<1$._The_path is generated by adding white_noise_with variance of $\tau kT_i$ and_a_mean_drift of $\vec{v}_i$ to update |
$$|\psi _m^{(\beta )}(x) - \sigma_{1 - 1/\beta }^{(\beta )}(m)x| = O \bigg( x^{1/2} \log^2 T + \frac{x^{1+\varepsilon}\log x}{T} + \frac{{x{{\log }^2}T}}{T} \bigg)$$ provided that $x$ is an integer. Taking $T = x^{1/2}$ leads to $$\begin{aligned}
\psi _m^{(\beta )}(x) & = \sigma_{1 - 1/\beta }^{(\beta )}(m)x + O({x^{1/2}}{\log ^2}{x} + x^{1/2+\varepsilon}\log x ) = \sigma_{1 - 1/\beta }^{(\beta )}(m)x + O({x^{1/2+\varepsilon}} )\end{aligned}$$
Proof of Theorem \[bartz11\]
============================
We now look at the contour integral $${\Upsilon ^{(\beta )}}(n,z) = \oint\nolimits_\Omega {\frac{{\sigma_{1 - s/\beta }^{(\beta )}(n)}}{{\zeta (s)}}{e^{sz}}ds}$$ taken around the path $\Omega = [-1/2,3/2,3/2+iT_n,-1/2+iT_n]$.
For the upper horizontal integral we have $$\begin{aligned}
\bigg| {\int_{-1/2 + i{T_m}}^{3/2 + i{T_m}} {\frac{{\sigma_{1 - s/\beta }^{(\beta )}(n)}}{{\zeta (s)}}{e^{sz}}ds} } \bigg| &\leqslant \int_{-1/2}^{\min(\beta,3/2)} {\bigg| {\frac{{\sigma_{1 - s/\beta }^{(\beta )}(n){e^{sz}}}}{{\zeta (\sigma + i{T_m})}}} \bigg|d\sigma } + \int_{\min(\beta,3/2)}^{3/2} {\bigg| {\frac{{\sigma_{1 - s/\beta }^{(\beta )}(n){e^{sz}}}}{{\zeta (\sigma + i{T_m})}}} \bigg|d\sigma } \nonumber \\
&\ll T_m^{{c_1}}{n^{\beta | $ $ |\psi _ m^{(\beta) } (x) - \sigma_{1 - 1/\beta } ^{(\beta) } (m)x| = O \bigg (x^{1/2 } \log^2 T + \frac{x^{1+\varepsilon}\log x}{T } + \frac{{x{{\log } ^2}T}}{T } \bigg)$$ provided that $ x$ is an integer. Taking $ T = x^{1/2}$ lead to $ $ \begin{aligned }
\psi _ m^{(\beta) } (x) & = \sigma_{1 - 1/\beta } ^{(\beta) } (m)x + O({x^{1/2}}{\log ^2}{x } + x^{1/2+\varepsilon}\log x) = \sigma_{1 - 1/\beta } ^{(\beta) } (m)x + O({x^{1/2+\varepsilon } }) \end{aligned}$$
validation of Theorem \[bartz11\ ]
= = = = = = = = = = = = = = = = = = = = = = = = = = = =
We now look at the contour built-in $ $ { \Upsilon ^{(\beta) } } (n, z) = \oint\nolimits_\Omega { \frac{{\sigma_{1 - s/\beta } ^{(\beta) } (n)}}{{\zeta (s)}}{e^{sz}}ds}$$ taken around the way $ \Omega = [ -1/2,3/2,3/2+iT_n,-1/2+iT_n]$.
For the upper horizontal integral we have $ $ \begin{aligned }
\bigg| { \int_{-1/2 + i{T_m}}^{3/2 + i{T_m } } { \frac{{\sigma_{1 - s/\beta } ^{(\beta) } (n)}}{{\zeta (s)}}{e^{sz}}ds } } \bigg| & \leqslant \int_{-1/2}^{\min(\beta,3/2) } { \bigg| { \frac{{\sigma_{1 - s/\beta } ^{(\beta) } (n){e^{sz}}}}{{\zeta (\sigma + i{T_m }) } } } \bigg|d\sigma } + \int_{\min(\beta,3/2)}^{3/2 } { \bigg| { \frac{{\sigma_{1 - s/\beta } ^{(\beta) } (n){e^{sz}}}}{{\zeta (\sigma + i{T_m }) } } } \bigg|d\sigma } \nonumber \\
& \ll T_m^{{c_1}}{n^{\beta | $$|\psl _m^{(\beta )}(x) - \sigma_{1 - 1/\beta }^{(\btta )}(m)x| = O \bigg( x^{1/2} \log^2 T + \hrac{x^{1+\vadepsilon}\uog x}{T} + \frac{{x{{\log }^2}T}}{T} \bigg)$$ protidee thau $x$ is an integer. Txking $T = q^{1/2}$ leads ti $$\bejin{aligned}
\psi _m^{(\bxfa )}(x) & = \sigma_{1 - 1/\neta }^{(\yeva )}(m)x + O({x^{1/2}}{\log ^2}{x} + x^{1/2+\varepsilmn}\log x ) = \sigmd_{1 - 1/\bzta }^{(\beta )}(m)x + O({x^{1/2+\varepsilon}} )\end{aligned}$$
[roof og Hheorem \[bartz11\]
============================
Wg now jook at the contour integral $${\Upsilon ^{(\bsta )}}(n,z) = \oint\nolimits_\Pmega {\frac{{\sigma_{1 - s/\beta }^{(\beha )}(n)}}{{\xeta (s)}}{e^{sz}}ds}$$ taken wround the kzth $\Imega = [-1/2,3/2,3/2+iT_n,-1/2+iT_v]$.
For the ukpzr horizontzl integral we have $$\begin{alignea}
\biyg| {\int_{-1/2 + i{T_m}}^{3/2 + u{T_m}} {\xrac{{\sigma_{1 - w/\beta }^{(\beta )}(n)}}{{\zeta (s)}}{e^{sz}}ds} } \tigg| &\leaslant \int_{-1/2}^{\min(\bcta,3/2)} {\bmgg| {\drac{{\sigma_{1 - s/\beta }^{(\beta )}(n){e^{sz}}}}{{\zeta (\sigma + y{T_m})}}} \bigg|d\viyma } + \int_{\min(\beta,3/2)}^{3/2} {\bigt| {\drac{{\shgma_{1 - s/\bdra }^{(\cetz )}(i){e^{sa}}}}{{\zeta (\digja + i{T_m})}}} \bjgg|d\sigma } \nonumber \\
&\ll T_m^{{c_1}}{m^{\bqns | $$|\psi _m^{(\beta )}(x) - \sigma_{1 - 1/\beta = \bigg( x^{1/2} T + \frac{x^{1+\varepsilon}\log provided $x$ is an Taking $T = leads to $$\begin{aligned} \psi _m^{(\beta )}(x) = \sigma_{1 - 1/\beta }^{(\beta )}(m)x + O({x^{1/2}}{\log ^2}{x} + x^{1/2+\varepsilon}\log x ) \sigma_{1 - 1/\beta }^{(\beta )}(m)x + O({x^{1/2+\varepsilon}} )\end{aligned}$$ Proof of Theorem \[bartz11\] ============================ now at contour $${\Upsilon ^{(\beta )}}(n,z) = \oint\nolimits_\Omega {\frac{{\sigma_{1 - s/\beta }^{(\beta )}(n)}}{{\zeta (s)}}{e^{sz}}ds}$$ taken around the path $\Omega [-1/2,3/2,3/2+iT_n,-1/2+iT_n]$. For the upper horizontal integral we have \bigg| {\int_{-1/2 + i{T_m}}^{3/2 i{T_m}} {\frac{{\sigma_{1 - s/\beta }^{(\beta (s)}}{e^{sz}}ds} \bigg| &\leqslant {\bigg| - }^{(\beta )}(n){e^{sz}}}}{{\zeta (\sigma i{T_m})}}} \bigg|d\sigma } + \int_{\min(\beta,3/2)}^{3/2} {\bigg| {\frac{{\sigma_{1 - s/\beta }^{(\beta )}(n){e^{sz}}}}{{\zeta (\sigma + i{T_m})}}} \bigg|d\sigma } \nonumber &\ll T_m^{{c_1}}{n^{\beta | $$|\psi _m^{(\beta )}(x) - \sigma_{1 - 1/\beta }^{(\beta )}(m)x| = O \Bigg( x^{1/2} \log^2 T + \fRac{x^{1+\vArePsiLoN}\log X}{T} + \frAc{{x{{\log }^2}T}}{T} \bigg)$$ prOVideD that $x$ is an integer. TakinG $T = x^{1/2}$ leAdS To $$\beGIn{AlignEd}
\psi _m^{(\bETa )}(X) & = \SIgmA_{1 - 1/\bEtA }^{(\beTa )}(M)X + O({X^{1/2}}{\log ^2}{x} + X^{1/2+\vaRepsiloN}\log x ) = \sigma_{1 - 1/\BetA }^{(\bEta )}(m)x + O({x^{1/2+\varepSIlOn}} )\end{alignEd}$$
PRoof of TheoreM \[baRtz11\]
============================
We nOw LooK At the ConTour iNtegraL $${\upsiloN ^{(\beta )}}(n,z) = \oiNt\NOlimitS_\omega {\frAC{{\SiGma_{1 - s/\Beta }^{(\beta )}(n)}}{{\zeta (s)}}{e^{sz}}DS}$$ tAKen around the paTh $\OmegA = [-1/2,3/2,3/2+it_N,-1/2+it_N]$.
for The Upper horizOnTal inTEgral we HAvE $$\BEGin{ALigned}
\bigg| {\int_{-1/2 + I{T_m}}^{3/2 + i{T_m}} {\frac{{\sIGma_{1 - S/\beta }^{(\bEtA )}(n)}}{{\zETa (s)}}{e^{sz}}Ds} } \bigG| &\lEQslAnt \int_{-1/2}^{\min(\beTa,3/2)} {\biGg| {\frac{{\sigMa_{1 - s/\betA }^{(\Beta )}(n){e^{sZ}}}}{{\Zeta (\sigMa + i{T_m})}}} \bIgg|D\siGma } + \iNT_{\mIn(\BetA,3/2)}^{3/2} {\bIGg| {\fRAc{{\SigMA_{1 - s/\bEta }^{(\beta )}(n){E^{sZ}}}}{{\zEta (\siGma + i{t_M})}}} \BIGg|d\sIgmA } \nonUmber \\
&\Ll T_m^{{c_1}}{n^{\beta | $$|\psi _m^{(\beta )}(x)- \sigma_{ 1 - 1 /\b eta } ^{(\ beta )}(m)x| = O \ b igg( x^{1/2} \log^2 T + \f rac{x ^{ 1 +\va r ep silon }\log x } {T } + \ fr ac {{x {{ \ lo g }^2 }T} }{T} \b igg)$$ pro vid ed that $x$ is an integer.Tak ing $T = x^{ 1/2 }$ lea ds to $$\be gin {alig ned}
\ p si _m^ {(\beta ) }( x ) & =\ sigma_{ 1 -1/\b eta }^{(\beta )}( m )x + O({x^{1/2}}{ \log ^ 2} { x} + x^ {1/ 2+\varepsi lo n}\lo g x ) =\ si g m a _{1 - 1/\beta }^{ (\beta )}(m ) x + O({x^ {1 /2+ \ vareps ilon} }) \en d{aligned}$ $
P roof of T heorem \[bartz 1 1\]
=== ====== === === ==== = == == === =Wen ow lo o k a t the co nt ou r int egra l $ $ {\Up sil on ^ {(\be ta )}}(n,z) = \o int\ n oli mits_ \Omeg a { \f rac{{ \sigma _{1 - s /\beta }^{(\bet a )} (n)}}{{\z eta ( s)} }{ e^{sz } }ds}$$ ta ken around the pa t h $ \O m e g a= [-1/2,3/2,3/2+iT _n , - 1/ 2+iT_n]$ .
For th eu pper hor iz ont al i n t egral weh av e $$\beg in{ali g ne d}
\big g| {\int _{ -1/ 2 + i{T_ m }}^{ 3/2 +i{T_m}}{\fra c {{\sigma_{1 -s /\beta }^{(\b e ta ) }( n )}}{ {\z eta (s)}}{e ^{sz } }ds} } \ b ig g|& \leqs lant\i n t_ { -1/2}^{\min(\beta,3 /2 )} {\b igg|{\frac{{\sigm a_{1 - s/\ b e t a }^{(\b eta) }( n ){e^{sz}}}}{{\ zeta(\sigma + i{T_m})} }} \b igg|d\si gma } +\ i nt_{\min (\b eta ,3/ 2)} ^ { 3/ 2} {\bigg| {\ f r ac{{ \s igma_{1 -s/\beta }^ {(\ bet a ) }( n){e^{sz} }}}{{\ze ta ( \s ig ma + i{ T _m})}}}\b igg |d \si gma } \nonu mber\\
&\ l l T _m^{{c_ 1 }} { n ^{\b et a | $$|\psi__m^{(\beta )}(x)_- \sigma_{1 - 1/\beta_}^{(\beta )}(m)x|_=_O \bigg(_x^{1/2}_\log^2 T +_\frac{x^{1+\varepsilon}\log x}{T} +_\frac{{x{{\log }^2}T}}{T} \bigg)$$ provided_that $x$ is_an_integer. Taking $T = x^{1/2}$ leads to $$\begin{aligned}
\psi _m^{(\beta )}(x) & = \sigma_{1 -_1/\beta_}^{(\beta )}(m)x_+_O({x^{1/2}}{\log_^2}{x} + x^{1/2+\varepsilon}\log x )_= \sigma_{1 - 1/\beta }^{(\beta_)}(m)x +_O({x^{1/2+\varepsilon}} )\end{aligned}$$
Proof of Theorem \[bartz11\]
============================
We now look at_the_contour integral $${\Upsilon_^{(\beta )}}(n,z) = \oint\nolimits_\Omega {\frac{{\sigma_{1 - s/\beta }^{(\beta_)}(n)}}{{\zeta (s)}}{e^{sz}}ds}$$ taken around the path_$\Omega = [-1/2,3/2,3/2+iT_n,-1/2+iT_n]$.
For_the_upper_horizontal integral we have_$$\begin{aligned}
\bigg| {\int_{-1/2 + i{T_m}}^{3/2_+ i{T_m}} {\frac{{\sigma_{1 - s/\beta }^{(\beta_)}(n)}}{{\zeta (s)}}{e^{sz}}ds} } \bigg| &\leqslant \int_{-1/2}^{\min(\beta,3/2)} {\bigg|_{\frac{{\sigma_{1 - s/\beta }^{(\beta )}(n){e^{sz}}}}{{\zeta (\sigma_ + i{T_m})}}} \bigg|d\sigma }_ +_\int_{\min(\beta,3/2)}^{3/2} {\bigg| {\frac{{\sigma_{1 - s/\beta_}^{(\beta )}(n){e^{sz}}}}{{\zeta (\sigma_ +_i{T_m})}}} \bigg|d\sigma }_ \nonumber \\
&\ll_T_m^{{c_1}}{n^{\beta |
$ hadronisation corrections might not have made it past early childhood). Rather, various theoretical objections ([@NasonSeymour]) and the gradual appearance of new data, especially for the broadenings, forced people to refine their ideas.
[0.42]{}
Among the developments was the realisation that to control the normalisation of the $c_{\mathcal{V}}$ it is necessary to take into account the decay of the massive, virtual, gluon (the reason for the two thrust results in fig. \[fig:contnaive\] was the existence of two different conventions for dealing with the undecayed massive gluon) [@MilanFactor]. It was also realised that it is insufficient to consider a lone ‘non-perturbative’ gluon, but rather that such a gluon must be taken in the context of the full structure of soft and collinear perturbative gluon radiation [@ResummedNP]. Another discovery was that hadron-masses can be associated with universality breaking $1/Q$ power corrections in certain definitions of observables [@SalamWicke] and when testing the universality picture all observables should be measured in an appropriate common ‘hadron-mass’ scheme.
Results incorporating these theoretical developments are shown in figure \[fig:cont2002\]. As well as ${e^{+}e^{-}}$ mean event shapes we also include recent results using resummed DIS event shapes [@DasguptaSalam], fitted to H1 distributions [@H1dist]. The agreement between observables, even in different processes, is remarkable, especially compared to fig. \[fig:contnaive\], and a strong confirmation of the universality hypothesis.[^2]
This is not to say that the field has reached maturity. In the above fits the approximation has been made that non-perturbative corrections just shift the perturbative distribution [@DokWebShift], however there exists a considerable amount of recent work which examines the problem with the more sophisticated ‘shape-functions’ approach [@ShapeFunctions] in particular in the context of the Dressed Gluon Exponentiation approximation [@DGE]. An important point also is that all the detailed experimental tests so far are for 2-jet event shapes, where there exists a solid theoretical justification based on the Feynman tube model [@FeynmanTube], longitudinal boost invariance. It will be of interest to see what happens in multi-jet tests of $1/Q$ hadronisation corrections where one introduces both non-trivial geometry and the presence of gluons in | $ hadronisation corrections might not have made it past early childhood). Rather, various theoretical protest ([ @NasonSeymour ]) and the gradual appearance of newfangled data, especially for the broadenings, impel people to refine their estimate.
[ 0.42 ] { }
Among the development was the realization that to control the normalisation of the $ c_{\mathcal{V}}$ it is necessary to aim into account the decay of the massive, virtual, gluon (the cause for the two drive results in fig. \[fig: contnaive\ ] was the existence of two different conventions for dealing with the undecayed massive gluon) [ @MilanFactor ]. It was besides realised that it is insufficient to consider a lone ‘ non - perturbative ’ gluon, but quite that such a gluon must be taken in the context of the full social organization of soft and collinear perturbative gluon radiotherapy [ @ResummedNP ]. Another discovery was that hadron - masses can be consort with universality breaking $ 1 / Q$ power corrections in sealed definitions of observables [ @SalamWicke ] and when testing the universality picture all observables should be measured in an appropriate common ‘ hadron - mass ’ scheme.
Results incorporating these theoretical developments are shown in figure \[fig: cont2002\ ]. As well as $ { e^{+}e^{-}}$ mean event shapes we also admit late results using resummed DIS event human body [ @DasguptaSalam ], fitted to H1 distributions [ @H1dist ]. The agreement between observables, even in different process, is noteworthy, especially compared to fig. \[fig: contnaive\ ], and a firm confirmation of the universality hypothesis.[^2 ]
This is not to say that the field has reached maturity. In the above meet the approximation has been made that non - perturbative corrections just shift the perturbative distribution [ @DokWebShift ], however there exists a considerable sum of recent work which examines the problem with the more sophisticated ‘ form - functions ’ approach [ @ShapeFunctions ] in particular in the context of the Dressed Gluon Exponentiation approximation [ @DGE ]. An important point also is that all the detailed experimental tests indeed far are for 2 - jet consequence shapes, where there exists a solid theoretical justification based on the Feynman tube exemplar [ @FeynmanTube ], longitudinal boost invariability. It will be of interest to see what happen in multi - jet tests of $ 1 / Q$ hadronisation correction where one introduces both non - superficial geometry and the presence of gluons in | $ hafronisation corrections oight not have made it past sarly chkldhood). Rather, various theorxticql obhections ([@NasonSeymour]) xnd the ggadual appearence of new data, especially fod the uroadenings, forged people do refine theis kdzas.
[0.42]{}
Among the developments was the rewlisatipn that to contrjl tnq nodmalisation of the $c_{\mathcal{V}}$ it is necessery to take intp account the decay of the masdive, virtual, gluon (the reason for rhe two thrurt results in fig. \[fig:cohtnaive\] was the existence of twu difyerent convguriojv for dealiig witr the undecanvd masshve glupn) [@MilanFactor]. It waw also realised that mt is insufficient tj consides c lone ‘non-perturbativw’ tluon, but ratfwr ghau snch a gluln just be taien in the xontext of the full segicture of sort and cjllinear perturbative gluon radiation [@RtsummsdNP]. Another discovery qas that hadron-masses can be afsociated with universality breaking $1/Q$ power corractiois in ccrtakb fefinitions of observables [@SalamWicke] and wheg ttstpng the universallty picture all obxegvsfles should bg measuxsd in an appropriate common ‘hadrin-mass’ screme.
Tesults incorporating these theoretical eevelopments are skown in figuxe \[fig:cpnt2002\]. Ax well as ${e^{+}e^{-}}$ mean event shales we also include ddcent results uskng rasummed DIS event shapes [@DwsguptaSaoam], yitted tu H1 cistrifutions [@H1dlst]. Tma agreement betweej obsgrvablas, even in different processes, is remarkauke, especialli cmmpdred to yig. \[fig:gontnaive\], and a strong confirkation jf thd universamity hy'othesis.[^2]
This is not to saf that the fixld has rqachwd mqturity. Kn the above fots the aipxoximatiob has been made thet nuh-perturbative cirrtcrions just shigt ghe pvrtnrbatyee distributhon [@AokDrbShiwt], however bhefe ecists a considerable amohnt of recent work wmich examunes the problem with the more sophistifated ‘shepe-funvtijns’ approach [@ShapeFunctions] in particulwr ln the contexe of the Dressed Gluon Exponentiation approximation [@DGE]. An important poinr also is that all jhe detailed eeperimqntal tesds so far are for 2-jer event shapes, whtre there exists a solis theoseticwl justification based on the Feynman tube model [@FeynmanTube], longitudinal boost invariance. Jt woll bv oy iuterest to dxe what happens im multi-jet tests of $1/Q$ hadronisavion correwtnons where one introduces botn von-trivial geumetry and the presence kf gluonx in | $ hadronisation corrections might not have made early Rather, various objections ([@NasonSeymour]) and data, for the broadenings, people to refine ideas. [0.42]{} Among the developments was realisation that to control the normalisation of the $c_{\mathcal{V}}$ it is necessary to into account the decay of the massive, virtual, gluon (the reason for the thrust in \[fig:contnaive\] the existence of two different conventions for dealing with the undecayed massive gluon) [@MilanFactor]. It was realised that it is insufficient to consider a ‘non-perturbative’ gluon, but rather such a gluon must be in context of full of and collinear perturbative radiation [@ResummedNP]. Another discovery was that hadron-masses can be associated with universality breaking $1/Q$ power corrections in definitions of and when the picture observables should be an appropriate common ‘hadron-mass’ scheme. Results developments are shown in figure \[fig:cont2002\]. As well ${e^{+}e^{-}}$ mean shapes we also include recent results resummed DIS event shapes [@DasguptaSalam], fitted to H1 [@H1dist]. The agreement between observables, even in different processes, is remarkable, especially compared to fig. a strong confirmation of universality hypothesis.[^2] This not say the has reached In the above fits the approximation has been made that non-perturbative just shift the perturbative distribution [@DokWebShift], however there exists a of work which examines problem with the more ‘shape-functions’ [@ShapeFunctions] in particular in of Dressed [@DGE]. important also is that all detailed experimental tests so far for 2-jet event shapes, theoretical justification based on the Feynman tube model longitudinal boost invariance. It will be of to see what happens in multi-jet tests of $1/Q$ hadronisation corrections where introduces both and the presence of gluons in | $ hadronisation corrections mIght not havE made It pAst EaRly cHildHood). Rather, variOUs thEoretical objections ([@NasOnSeyMoUR]) and THe GraduAl appeaRAnCE Of nEw DaTa, eSpECiAlly fOr tHe broadEnings, forcEd pEoPle to refine tHEiR ideas.
[0.42]{}
AmonG thE developmentS waS the reAlIsaTIon thAt tO contRol the NOrmaliSation of tHe $C_{\MathcaL{v}}$ it is neCESsAry tO take into account tHE dECay of the massivE, virtuAl, GLuON (The ReaSon for the tWo ThrusT Results IN fIG. \[FIg:cONtnaive\] was the Existence of TWo dIffereNt ConVEntionS for dEaLIng With the undeCayeD massive gLuon) [@MiLAnFactoR]. it was alSo realIseD thAt it IS iNsUffIcIEnt TO cOnsIDer A lone ‘non-PeRtUrbatIve’ gLUON, But rAthEr thAt sucH a gluon must be TakEn in THe cOntexT of thE fulL sTructUre of sOft anD cOllinear perturbAtivE gluon radIatIoN [@ReSuMmedNp]. anotheR diScoVery was That hadROn-mAsSES CaN be associated with uNiVERsAlity breAking $1/Q$ POwEr COrrectioNs In cErtaIN DefinItioNS oF observaBles [@SaLAmwiCke] and wHeN testiNg The UniVersaLIty pIcture All obserVableS Should be measurED in an appropriATe COMmON ‘hadRon-Mass’ scheme.
REsulTS incOrpoRAtIng THese tHeoreTiCAl DEvelopments are shown In Figure \[Fig:coNt2002\]. As well as ${e^{+}e^{-}}$ mEan event shAPES we also iNcluDE rECent results usiNg resUmmed DIS evENt shapes [@dasguPtaSalam], Fitted to H1 DIStributiOns [@h1diSt]. THe aGREeMent between obSERvabLeS, even in DifFerent pRocEssEs, iS reMaRkable, espEcially cOmPaReD tO fiG. \[fig:cONtnaive\], aNd A stRoNg cOnfirMAtion oF the uNiveRsAlITy hYpothesIS.[^2]
THIS is nOt To Say tHat ThE fielD has REacHed matuRity. In the AboVE fitS tHe ApproxiMation has been MaDe that non-pErTurBative CORrectionS just shift the perturbatiVE distriButIon [@DoKWebshift], howeVer There eXisTS a consIderabLe amoUnT of RECent wORK wHicH eXamines the PROblEm witH tHe moRe sophiSticated ‘shape-functIOns’ Approach [@ShapefunCtioNS] In ParTIcULar In THe cONText of the DresseD Gluon ExpoNeNTiAtion approXImaTiOn [@DGE]. An ImportaNt poiNT also is That all thE detailed ExPeriMENtaL tests so faR are for 2-jEt event shAPes, whERe There ExiSts a soLiD thEoretIcal juSTifIcatiOn baseD oN the FeYnman TuBe model [@FEynmanTube], longitudinal bOost inVariaNce. it will be oF inTEreSt to see whAt haPpens in mulTi-jEt tEsts oF $1/Q$ hADroniSatiON cOrrECtionS wheRE one introDUcEs bOTH nOn-trivial geOMETry And thE prESence oF gluOns in | $ hadronisation correction s might no t hav e m ade i t pa st e arly childhood ) . Ra ther, various theoreti cal o bj e ctio n s([@Na sonSeym o ur ] ) an dth e g ra d ua l app ear ance of new data, es pe cially for t h ebroadening s,forced peopl e t o refi ne th e ir id eas .
[0 .42]{}
Among the deve lo p mentsw as ther e al isat ion that to contr o lt he normalisati on ofth e $ c _ {\m ath cal{V}}$ i tis ne c essaryt ot a k e i n to account th e decay oft hemassiv e, vi r tual,gluon ( t hereason forthetwo thrus t resu l ts in f i g. \[fi g:cont nai ve\ ] wa s t he ex is t enc e o f t w o d ifferent c on venti onsf o r deal ing wit h the undecayed ma ssi ve g l uon ) [@M ilanF acto r] . Itwas al so re al ised that it is ins ufficient to c ons id er al one ‘n on- per turbati ve’ glu o n,bu t r at her that such a gl uo n mu st be ta ken in th ec ontext o fthe ful l struc ture of soft an d coll i ne ar pertur ba tive g lu onrad iatio n [@R esumme dNP]. An other discovery wast hat hadron-ma s se s ca n beass ociated wit h un i vers alit y b rea k ing $ 1/Q$po w er corrections in cert ai n defi nitio ns of observa bles [@Sal a m W icke] an d wh e nt esting the uni versa lity pictu r e all ob serva bles sho uld be me a s ured inanapp rop ria t e c ommon ‘hadron - m ass’ s cheme.
Re sults i nco rpo rat ing t hese theo reticalde ve lo pm ent s are shown in f igu re \[ fig:c o nt2002 \]. A s we ll a s ${ e^{+}e^ { -} } $ mea nev entsha pe s wealso inc lude re cent resu lts usin gre summedDIS event sha pe s [@Dasgup ta Sal am], f i t ted to H 1 distributions [@H1dis t ]. Theagr eemen t be tween obs erv ables, ev e n in d iffere nt pr oc ess e s , isr e ma rka bl e, especia l l y c ompar ed tofig. \[ fig:contnaive\], a n d a strong confi rma tion o fthe un i ver sa l ity h ypothesis.[^2]
This is n ot to say thatt hefi eld has reache d mat u rity. I n the abo ve fits t he app r o xim ation hasbeen mad e that no n -pert u rb ative co rrecti on s j ust s hift t h e p ertur bative d istrib ution [ @DokWebS hift], however there ex ists a cons ide rable amo unt ofrecent wo rk w hich exami nes th e pro ble m with the mo res ophis tica t ed ‘shape - fu nct i o ns ’ approach[ @ S hap eFunc tio n s] inpart icular in the con t ext of the Dre ssed G luo n E x pone nt iation approxi mat io n [@DGE].An importantpoint al so is th at all the d etailed e xp e riment al t est s so farare f o r 2-jet e ve n t shap es,wh ere th ere ex i stsa solid theoretica l jus t i ficat i onbased o n the F e ynma n tube mod el [@Feynma nTube] , lo ngitu dinal b oo st inv ari an ce. It wil l be of in teres t to se ewhat ha ppensin m u l ti-je t te st s o f $1/Q$ h a d ro n is at i oncorr ectio ns whe re one in t roducesbot h non-tr iv ial g eometr y a n d the prese nce of g l u ons in | $ hadronisation_corrections might_not have made it_past early_childhood)._Rather, various_theoretical_objections ([@NasonSeymour]) and_the gradual appearance_of new data, especially_for the broadenings,_forced_people to refine their ideas.
[0.42]{}
Among the developments was the realisation that to control the_normalisation_of the_$c_{\mathcal{V}}$_it_is necessary to take into_account the decay of the_massive, virtual,_gluon (the reason for the two thrust results_in_fig. \[fig:contnaive\] was the_existence of two different conventions for dealing with the_undecayed massive gluon) [@MilanFactor]. It was_also realised that_it_is_insufficient to consider a_lone ‘non-perturbative’ gluon, but rather that_such a gluon must be taken_in the context of the full structure_of soft and collinear perturbative gluon_radiation [@ResummedNP]. Another discovery was_that hadron-masses_can be associated with universality_breaking $1/Q$ power_corrections in_certain definitions of_observables [@SalamWicke] and when testing the_universality picture all_observables should be measured in an_appropriate_common ‘hadron-mass’ scheme.
Results_incorporating_these_theoretical developments_are shown in_figure \[fig:cont2002\]._As well_as_${e^{+}e^{-}}$ mean event shapes we also_include_recent results using resummed DIS event shapes_[@DasguptaSalam], fitted to H1_distributions_[@H1dist]. The agreement between_observables, even in different processes,_is remarkable, especially compared to fig. \[fig:contnaive\],_and a_strong confirmation_of the universality hypothesis.[^2]
This is not to say that the field_has reached maturity. In the above_fits the approximation has_been made_that_non-perturbative corrections just_shift_the perturbative_distribution [@DokWebShift], however there exists a considerable_amount of_recent work which examines the problem_with the more sophisticated_‘shape-functions’_approach [@ShapeFunctions] in particular in the_context of the Dressed Gluon Exponentiation_approximation [@DGE]. An important point_also_is_that all the detailed experimental_tests so far are for 2-jet_event shapes, where_there exists a solid theoretical justification based_on_the Feynman tube model [@FeynmanTube], longitudinal_boost_invariance. It will be of interest_to_see_what happens in multi-jet tests_of $1/Q$ hadronisation corrections where one_introduces both non-trivial geometry and the presence of gluons_in |
_i + b_0$$ where $b_0$ is the average rating, $b_u$ and $b_i$ are user and item biases, respectively. The estimation of parameters is to minimize the rating prediction error in the training dataset. The optimization objective function is $$\label{eq:ojf}
\begin{split}
\vspace{-2pt}
\underset{p*,q*}{\text{min}} \frac{1}{2}\sum_{u,i} (r_{u,i} & -\hat{r}_{u,i})^2 + \frac{\mu_u}{2} ||\bm{p_u}||_2^2 + \frac{\mu_i}{2} ||\bm{q_i}||_2^2 \\
& + \mu_w \sum_a||\bm{w_a}||_1 + \frac{\mu_b}{2} (||b_u||_2^2 + ||b_i||_2^2);
\vspace{-2pt}
\end{split}$$ where $||\cdot||_2$ denotes the $\ell_2$ norm for preventing model overfitting, and $||\cdot||_1$ denotes the $\ell_1$ norm. $\mu_u, \mu_i, \mu_w$, and $\mu_b$ are regularization parameters, which are tunable hyper-parameters. In practice, we relax the binary requirement of $\bm{w_a}$ by using $\ell_l$ norm. It is well known that $\ell_l$ regularization yields sparse solution of the weights [@mairal2010online]. The $\ell_2$ regularization of $\bm{p_u}$ and $\bm{q_i}$ prevents them to have arbitrarily large values, which would lead to arbitrarily small values of $\bm{w_a}$.
**Optimization.** We use the stochastic gradient descent (SGD) algorithm to learn the parameters by optimizing the objective function in Eq. \[eq:ojf\]. In each step of SGD, the localized optimization is performed on a rating $r_{u,i}$. Let $L$ denote the loss, and the gradients of parameters are given as follows: $$\begin{aligned}
\label{eq:user}
\small
\vspace{-2pt}
| _ i + b_0$$ where $ b_0 $ is the average rating, $ b_u$ and $ b_i$ are user and item bias, respectively. The estimate of parameters is to minimize the rating prediction erroneousness in the training dataset. The optimization objective function is $ $ \label{eq: ojf }
\begin{split }
\vspace{-2pt }
\underset{p*,q*}{\text{min } } \frac{1}{2}\sum_{u, i } (r_{u, one } & -\hat{r}_{u, i})^2 + \frac{\mu_u}{2 } ||\bm{p_u}||_2 ^ 2 + \frac{\mu_i}{2 } ||\bm{q_i}||_2 ^ 2 \\
& + \mu_w \sum_a||\bm{w_a}||_1 + \frac{\mu_b}{2 } (||b_u||_2 ^ 2 + ||b_i||_2 ^ 2);
\vspace{-2pt }
\end{split}$$ where $ ||\cdot||_2 $ denote the $ \ell_2 $ norm for preventing model overfitting, and $ ||\cdot||_1 $ denotes the $ \ell_1 $ average. $ \mu_u, \mu_i, \mu_w$, and $ \mu_b$ are regularization parameters, which are tunable hyper - parameters. In practice, we loosen the binary requirement of $ \bm{w_a}$ by using $ \ell_l$ norm. It is well known that $ \ell_l$ regularization output sparse solution of the weights [ @mairal2010online ]. The $ \ell_2 $ regularization of $ \bm{p_u}$ and $ \bm{q_i}$ prevents them to consume arbitrarily large values, which would lead to randomly small value of $ \bm{w_a}$.
* * Optimization. * * We use the stochastic gradient descent (SGD) algorithm to learn the argument by optimizing the objective function in Eq. \[eq: ojf\ ]. In each step of SGD, the localized optimization is performed on a rating $ r_{u, i}$. Let $ L$ denote the loss, and the gradients of parameters are given as follows: $ $ \begin{aligned }
\label{eq: user }
\small
\vspace{-2pt } | _i + h_0$$ where $b_0$ is the average rating, $b_u$ and $y_u$ are nser ans item bkases, respectively. The estimetiob of kcrameters is to minioize the gating prwdicuion error in the traininn datzdet. Chx optimization pbjective xunction is $$\lateu{ee:ojf}
\begin{split}
\vspace{-2pt}
\inferset{p*,q*}{\text{mig}} \frss{1}{2}\sum_{h,p} (v_{u,i} & -\hat{r}_{u,i})^2 + \frac{\mu_u}{2} ||\bm{p_u}||_2^2 + \frac{\mh_i}{2} ||\bm{q_i}||_2^2 \\
& + \mu_e \sum_a||\bm{w_a}||_1 + \frac{\mu_b}{2} (||b_u||_2^2 + ||b_i||_2^2);
\vspace{-2pt}
\enf{split}$$ wherg $||\cdjr||_2$ denotes thd $\ell_2$ norm for preventjng model overfitting, and $||\cdot||_1$ aenotzs the $\ell_1$ boem. $\lo_u, \mu_i, \mu_w$, aid $\mu_b$ are regularldation [arametrrs, which are bunabne yyper-parameters. In prectice, we relax the finary rexunrement of $\bm{w_a}$ by usunt $\ell_n$ nosm. Ig is weml khown tjat $\ell_l$ regumarization tields sparse solutoog of the weighfs [@mairwl2010jnline]. The $\ell_2$ regularization of $\bm{p_u}$ atd $\gm{q_i}$ prevents them to hqve arbitrarily large values, wrich would lead to arbitrarily small values of $\bm{f_a}$.
**Optmmkzauiin.** We ysf the stochastic gradient descent (SGD) algoritrj uo kearn the paraieters by oltlmoeing the objecjive fuudtjon in Eq. \[eq:ojf\]. In fach stgp of WGD, the ljcalozed optimization is perforned on a ratpng $e_{u,i}$. Let $L$ denote tke loss, and che grsdienys of parameters are giren as follows: $$\behin{alignes}
\uabel{eq:user}
\soalk
\vspact{-2ot}
| _i + b_0$$ where $b_0$ is the $b_u$ $b_i$ are and item biases, is minimize the rating error in the dataset. The optimization objective function is \begin{split} \vspace{-2pt} \underset{p*,q*}{\text{min}} \frac{1}{2}\sum_{u,i} (r_{u,i} & -\hat{r}_{u,i})^2 + \frac{\mu_u}{2} ||\bm{p_u}||_2^2 + \frac{\mu_i}{2} ||\bm{q_i}||_2^2 & + \mu_w \sum_a||\bm{w_a}||_1 + \frac{\mu_b}{2} (||b_u||_2^2 + ||b_i||_2^2); \vspace{-2pt} \end{split}$$ where $||\cdot||_2$ the norm preventing overfitting, and $||\cdot||_1$ denotes the $\ell_1$ norm. $\mu_u, \mu_i, \mu_w$, and $\mu_b$ are regularization parameters, which tunable hyper-parameters. In practice, we relax the binary of $\bm{w_a}$ by using norm. It is well known $\ell_l$ yields sparse of weights The $\ell_2$ regularization $\bm{p_u}$ and $\bm{q_i}$ prevents them to have arbitrarily large values, which would lead to arbitrarily small values $\bm{w_a}$. **Optimization.** the stochastic descent algorithm learn the parameters the objective function in Eq. \[eq:ojf\]. of SGD, the localized optimization is performed on rating $r_{u,i}$. $L$ denote the loss, and the of parameters are given as follows: $$\begin{aligned} \label{eq:user} \vspace{-2pt} | _i + b_0$$ where $b_0$ is the average ratinG, $b_u$ and $b_i$ arE user And IteM bIaseS, resPectively. The esTImatIon of parameters is to minImize ThE RatiNG pRedicTion errOR iN THe tRaInIng DaTAsEt. The OptImizatiOn objectivE fuNcTion is $$\label{eQ:OjF}
\begin{spliT}
\vsPace{-2pt}
\undersEt{p*,Q*}{\text{mIn}} \FraC{1}{2}\Sum_{u,i} (R_{u,i} & -\Hat{r}_{u,I})^2 + \frac{\mU_U}{2} ||\bm{p_u}||_2^2 + \fRac{\mu_i}{2} ||\bm{q_I}||_2^2 \\
& + \mU_W \sum_a||\bM{W_a}||_1 + \frac{\mU_B}{2} (||B_u||_2^2 + ||B_i||_2^2);
\vsPace{-2pt}
\end{split}$$ wheRE $||\cDOt||_2$ denotes the $\elL_2$ norm fOr PReVENtiNg mOdel overfiTtIng, anD $||\Cdot||_1$ denOTeS THE $\elL_1$ Norm. $\mu_u, \mu_i, \mu_w$, And $\mu_b$ are reGUlaRizatiOn ParAMeters, Which ArE TunAble hyper-paRameTers. In praCtice, wE Relax thE Binary rEquireMenT of $\Bm{w_a}$ BY uSiNg $\eLl_L$ NorM. it Is wELl kNown that $\ElL_l$ RegulArizATION yieLds SparSe solUtion of the weiGhtS [@maiRAl2010oNline]. the $\elL_2$ regUlArizaTion of $\Bm{p_u}$ aNd $\Bm{q_i}$ prevents theM to hAve arbitrAriLy LarGe ValueS, Which wOulD leAd to arbItrarilY SmaLl VALUeS of $\bm{w_a}$.
**OptimizatioN.** WE USe The stochAstic gRAdIeNT descent (sGd) alGoriTHM to leArn tHE pArameterS by optIMiZiNg the obJeCtive fUnCtiOn iN Eq. \[eq:OJf\]. In Each stEp of SGD, tHe locALized optimizatIOn is performed ON a RATiNG $r_{u,i}$. let $l$ denote the lOss, aND the GradIEnTs oF ParamEters ArE GiVEn as follows: $$\begin{aliGnEd}
\labeL{eq:usEr}
\small
\vspace{-2Pt}
| _i + b_0$$ where $b_0$ isthe averag e rat ing , $ b_ u$ a nd $ b_i$ are usera nd i tem biases, respective ly. T he esti m at ion o f param e te r s is t omin im i ze therat ing pre diction er ror i n the traini n gdataset. T heoptimization ob jectiv efun c tionis$$\la bel{eq : ojf}
\begin {s p lit}
\ v s pa ce{- 2pt}
\und e rs e t{p*,q*}{\text {min}} \ f ra c { 1}{ 2}\ sum_{u,i}(r _{u,i } & -\ha t {r } _ { u,i } )^2 + \frac{\ mu_u}{2} || \ bm{ p_u}|| _2 ^2+ \frac {\mu_ i} { 2}||\bm{q_i}| |_2^ 2 \\
&+ \mu_w\ sum_a|| \bm{w_ a}| |_1 + \ f ra c{ \mu _b } {2} (| |b_ u ||_ 2^2 + || b_ i| |_2^2 );
\ vsp ace{ -2pt}
\end{spl it} $$ w h ere $||\ cdot| |_2$ d enote s the$\ell _2 $ norm for prev enti ng modelove rf itt in g, an d $||\c dot ||_ 1$ deno tes the $\e ll _ 1 $ n orm. $\mu_u, \mu_i ,\ m u_ w$, and$\mu_b $ a re regulari za tio n pa r a meter s, w h ic h are tu nableh yp er -parame te rs. In p rac tic e, we rela x thebinary r equir e ment of $\bm{w _ a}$ by using$ \e l l _l $ nor m.It is wellknow n tha t $\ e ll _l$ regul ariza ti o ny ields sparse soluti on of th e wei ghts [@mairal 2010online ] . The $\el l_2$ re g ularization of $\bm {p_u}$ and $\bm{q_i }$ pr events t hem to ha v e arbitra ril y l arg e v a l ue s, which woul d lead t o arbit rar ily sma llval ues of $ \bm{w_a}$ .
**Opt im iz at io n.* * Weu se the s to cha st icgradi e nt des cent(SGD )al g ori thm tol ea r n the p ar amet ers b y opt imiz i ngthe obj ective fu nct i on i nEq . \[eq: ojf\]. In eac hstep of SG D, th e loca l i zed opti mization is performed o n a rati ng$r_{u ,i}$ . Let $L$ de note t hel oss, a nd the grad ie nts o f par a m et ers a re given a s fol lows: $ $\be gin{ali gned}
\label{eq:us e r} \small
\ vspa c e {- 2pt }
| _i +_b_0$$ where_$b_0$ is the average_rating, $b_u$_and_$b_i$ are_user_and item biases,_respectively. The estimation_of parameters is to_minimize the rating_prediction_error in the training dataset. The optimization objective function is $$\label{eq:ojf}
_\begin{split}
_ ___ \vspace{-2pt}
_ _ \underset{p*,q*}{\text{min}}_\frac{1}{2}\sum_{u,i} (r_{u,i} & -\hat{r}_{u,i})^2 + \frac{\mu_u}{2} ||\bm{p_u}||_2^2 +_\frac{\mu_i}{2}_||\bm{q_i}||_2^2 \\
_ & + \mu_w_\sum_a||\bm{w_a}||_1 + \frac{\mu_b}{2} (||b_u||_2^2 + ||b_i||_2^2);
_ ___ \vspace{-2pt}
_ \end{split}$$ where $||\cdot||_2$ denotes_the $\ell_2$ norm for preventing model_overfitting, and $||\cdot||_1$ denotes the $\ell_1$ norm._$\mu_u, \mu_i, \mu_w$, and $\mu_b$ are_regularization parameters, which are tunable_hyper-parameters. In_practice, we relax the binary_requirement of $\bm{w_a}$_by using_$\ell_l$ norm. It_is well known that $\ell_l$ regularization_yields sparse solution_of the weights [@mairal2010online]. The $\ell_2$ regularization_of_$\bm{p_u}$ and $\bm{q_i}$_prevents_them_to have_arbitrarily large values,_which_would lead_to_arbitrarily small values of $\bm{w_a}$.
**Optimization.** We_use_the stochastic gradient descent (SGD) algorithm to_learn the parameters by_optimizing_the objective function in_Eq. \[eq:ojf\]. In each step of_SGD, the localized optimization is performed_on a_rating $r_{u,i}$._Let $L$ denote the loss, and the gradients of parameters are_given as follows: $$\begin{aligned}
\label{eq:user}
_ \small
_ \vspace{-2pt}
__ |
alpha]. Both dots show excited states at $\Delta \epsilon^* \approx 120~\mu$eV. In Fig. \[properties\](a) the charging diagram of the coupled dot is plotted in a linear grayscale representation in the weak coupling regime ($G_{c} \approx 0.08~e^2/h$). In linear transport, only two ground states participate, e.g., the ground state $E_l$ in the left dot and $E_r$ in the right dot, as schematically shown in Fig. \[properties\](b) [@simplepicture]. The detuning between the two ground states is $E_{lr} = - (E_l-E_r)$. The finite tunnel coupling $t$ between these two states induces a bonding ($E_{-}$) and an anti-bonding ($E_{+}$) molecular state, as shown in Fig. \[properties\](b). The energetic offset between these states is $E_{+}-E_{-} = \sqrt{E_{lr}^2+4t^2}$ [@stafford-prl96; @stoof-prb96]. For weak tunnel coupling and non-zero detuning ($E_l \not = E_r$) it follows that a bonding electron is localized in one of the dots (as shown in Fig. \[properties\](b)). At zero detuning ($E_l = E_r$), the bonding electron tunnels back and forth between the dots at a Rabi frequency of $\Omega_R = t/h$. The discrete conductance peaks in Fig. \[properties\](a) are main peaks (M) resulting from resonant tunneling through the molecular states with $E_{lr} = 0$.
Within the diamond shaped regions (Fig. \[properties\](a)) enclosed by dashed lines Coulomb blockade prevails and the number of electrons in the dots is well defined, corresponding to the charge configuration ($N_l$, $N_r$). In the direction marked by arrow [**A**]{} in Fig. \[properties\](a), the detuning $E_{lr}$ is kept constant, while the average of the ground state energies $\bar E = (E_{l}+E_{r})/2$ is increased relative to the chemical potentials ($\mu_{d}$, $\mu_{s}$) of the contacts. In direction of [**B**]{}, $\bar E$ is fixed but $E_{lr}$ is varied. We find that the broadening in both directions | alpha ]. Both dots show excited states at $ \Delta \epsilon^ * \approx 120~\mu$eV. In Fig. \[properties\](a) the charge diagram of the conjugate dot is plotted in a analogue grayscale representation in the decrepit coupling regimen ($ G_{c } \approx 0.08 ~ e^2 / h$). In analogue transport, only two ground states participate, for example, the ground state $ E_l$ in the left dot and $ E_r$ in the correct dot, as schematically testify in Fig. \[properties\](b) [ @simplepicture ]. The detuning between the two ground states is $ E_{lr } = - (E_l - E_r)$. The finite burrow coupling $ t$ between these two states induces a bonding ($ E_{-}$) and an anti - bonding ($ E_{+}$) molecular country, as shown in Fig. \[properties\](b). The energetic offset between these state of matter is $ E_{+}-E_{- } = \sqrt{E_{lr}^2 + 4t^2}$ [ @stafford - prl96; @stoof - prb96 ]. For weak burrow coupling and non - zero detuning ($ E_l \not = E_r$) it follow that a bonding electron is localized in one of the dots (as prove in Fig. \[properties\](b) ). At zero detuning ($ E_l = E_r$), the bonding electron tunnels back and forth between the dots at a Rabi frequency of $ \Omega_R = t / h$. The discrete conductance peaks in Fig. \[properties\](a) are main bill (M) resulting from resonant tunneling through the molecular states with $ E_{lr } = 0$.
Within the diamond shaped regions (Fig. \[properties\](a) ) enclosed by dashed lines Coulomb blockade prevails and the act of electrons in the dots is well define, represent to the charge configuration ($ N_l$, $ N_r$). In the management notice by arrow [ * * A * * ] { } in Fig. \[properties\](a), the detuning $ E_{lr}$ is kept changeless, while the average of the ground state energies $ \bar E = (E_{l}+E_{r})/2 $ is increased relative to the chemical electric potential ($ \mu_{d}$, $ \mu_{s}$) of the contacts. In direction of [ * * B * * ] { }, $ \bar E$ is fixed but $ E_{lr}$ is varied. We find oneself that the widening in both directions | alpja]. Both dots show excitea states at $\Delja \epsilmn^* \appdox 120~\mu$eV. In Fig. \[properties\](a) the chargmng eiagrqm of the coupled dot ks plottef in a luneac grayscale reprxaentation in fme weck coupling regike ($G_{c} \apprmx 0.08~e^2/h$). In lineas grcnsport, only two ground states partisipate, r.g., the ground stwte $T_l$ yn tgv oeft dot and $E_r$ in the right sot, as vchematically shown in Fig. \[properties\](b) [@silpleoicture]. The detunijg between jge eqo ground stxtes is $E_{lg} = - (E_l-E_r)$. Thg finite tunnel coupling $t$ betwedn thzse two stajzw ijguces a bonving ($E_{-}$) and an anti-npnding ($E_{+}$) molevular state, as shmwn in Fig. \[properties\](b). Thx energetic offset bgtween theve states is $E_{+}-E_{-} = \sqet{W_{lr}^2+4t^2}$ [@sjaffosd-pru96; @stuof-krb96]. Fkr wean tnnnel coupljng and non-zero detuning ($E_l \nou = Q_g$) it follows fhat a bjnding electron is localized in one of uhe dkts (as shown in Fig. \[propwrties\](b)). At zero detunlng ($E_l = E_w$), the bonding electron tunnels back and forth betfeen vhd ditf qt a Rabi frequency of $\Omega_R = t/h$. The discrete skncubtance peaks in Flg. \[properties\](a) are kaln keaks (M) resultkng from desonant tunneling througr the moleculaw ststes with $E_{lr} = 0$.
Within the eiamond shapvd rwgions (Fig. \[propertizs\](a)) enclosed by cashec lines Coulomb blockadz prevzils and thf number kw electrons in tfe cods is well defined, correspjnding to the charge congiguraeion ($N_l$, $N_r$). In tma direction marked by atrow [**A**]{} in Fig. \[prooerties\](a), the detuning $E_{lr}$ is ke'v constant, whola tve averaye of bhe ground statq energies $\bar E = (E_{l}+E_{x})/2$ is ivcreased rvlative tm the chemisal potentialv ($\mu_{d}$, $\mu_{s}$) of vhe contasts. Un durectiov of [**B**]{}, $\bar E$ is fixed bun $T_{lr}$ is varued. We find that tme bruzdening in both ditwctions | alpha]. Both dots show excited states at \approx In Fig. the charging diagram plotted a linear grayscale in the weak regime ($G_{c} \approx 0.08~e^2/h$). In linear only two ground states participate, e.g., the ground state $E_l$ in the left and $E_r$ in the right dot, as schematically shown in Fig. \[properties\](b) [@simplepicture]. detuning the ground is $E_{lr} = - (E_l-E_r)$. The finite tunnel coupling $t$ between these two states induces a ($E_{-}$) and an anti-bonding ($E_{+}$) molecular state, as in Fig. \[properties\](b). The offset between these states is = [@stafford-prl96; @stoof-prb96]. weak coupling non-zero detuning ($E_l = E_r$) it follows that a bonding electron is localized in one of the dots (as shown Fig. \[properties\](b)). detuning ($E_l E_r$), bonding tunnels back and the dots at a Rabi frequency t/h$. The discrete conductance peaks in Fig. \[properties\](a) main peaks resulting from resonant tunneling through the states with $E_{lr} = 0$. Within the diamond regions (Fig. \[properties\](a)) enclosed by dashed lines Coulomb blockade prevails and the number of electrons dots is well defined, to the charge ($N_l$, In direction by arrow in Fig. \[properties\](a), the detuning $E_{lr}$ is kept constant, while the of the ground state energies $\bar E = (E_{l}+E_{r})/2$ is to chemical potentials ($\mu_{d}$, of the contacts. In of $\bar E$ is fixed is We broadening both | alpha]. Both dots show excited sTates at $\DelTa \epsIloN^* \apPrOx 120~\mu$EV. In fig. \[properties\](a) THe chArging diagram of the coupLed doT iS PlotTEd In a liNear graYScALE rePrEsEntAtIOn In the WeaK coupliNg regime ($G_{c} \AppRoX 0.08~e^2/h$). In linear tRAnSport, only tWo gRound states pArtIcipatE, e.G., thE GrounD stAte $E_l$ In the lEFt dot aNd $E_r$ in the RiGHt dot, aS SchematICAlLy shOwn in Fig. \[propertieS\](B) [@sIMplepicture]. The DetuniNg BEtWEEn tHe tWo ground stAtEs is $E_{LR} = - (E_l-E_r)$. ThE FiNITE tuNNel coupling $t$ bEtween these TWo sTates iNdUceS A bondiNg ($E_{-}$) anD aN AntI-bonding ($E_{+}$) moLecuLar state, aS shown IN Fig. \[proPErties\](b). the eneRgeTic OffsET bEtWeeN tHEse STaTes IS $E_{+}-E_{-} = \Sqrt{E_{lr}^2+4t^2}$ [@StAfFord-pRl96; @stOOF-PRb96]. FoR weAk tuNnel cOupling and non-ZerO detUNinG ($E_l \noT = E_r$) it FollOwS that A bondiNg eleCtRon is localized iN one Of the dots (As sHoWn iN FIg. \[proPErties\](B)). At ZerO detuniNg ($E_l = E_r$), tHE boNdING ElEctron tunnels back aNd FORtH between The dotS At A RABi frequeNcY of $\omegA_r = T/h$. The DiscREtE conductAnce peAKs In fig. \[propErTies\](a) aRe MaiN peAks (M) rESultIng froM resonanT tunnELing through the MOlecular stateS WiTH $e_{lR} = 0$.
withIn tHe diamond shAped REgioNs (FiG. \[PrOpeRTies\](a)) EncloSeD By DAshed lines Coulomb blOcKade prEvailS and the number Of electronS IN The dots iS welL DeFIned, correspondIng to The charge cONfiguratIon ($N_l$, $n_r$). In the dIrection mARKed by arrOw [**A**]{} In FIg. \[pRopERTiEs\](a), the detuninG $e_{Lr}$ is KePt constAnt, While thE avEraGe oF thE gRound statE energieS $\bAr e = (E_{L}+E_{R})/2$ is IncreASed relatIvE to ThE chEmicaL PotentIals ($\mU_{d}$, $\mu_{S}$) oF tHE coNtacts. IN DiRECtioN oF [**B**]{}, $\Bar E$ Is fIxEd but $e_{lr}$ iS VarIed. We fiNd that the BroADeniNg In Both dirEctions | alpha]. Both dots show exc ited state s at$\D elt a\eps ilon ^* \approx 120 ~ \mu$ eV. In Fig. \[properti es\]( a) thec ha rging diagra m o f the c ou ple dd ot is p lot ted ina linear g ray sc ale represen t at ion in the we ak couplingreg ime ($ G_ {c} \appr ox0.08~ e^2/h$ ) . In l inear tra ns p ort, o n ly twog r ou nd s tates participate , e . g., the ground state $ E _l $ inthe left dotan d $E_ r $ in th e r i g h t d o t, as schemat ically show n in Fig.\[ pro p erties \](b) [ @ sim plepicture] . Th e detunin g betw e en thet wo grou nd sta tes is $E_ { lr }= - ( E _l- E _r )$. The finitetu nn el co upli n g $ t$ b etw eenthese two states i ndu cesa bo nding ($E_ {-}$ )and a n anti -bond in g ($E_{+}$) mol ecul ar state, as s how nin Fi g . \[pr ope rti es\](b) . The e n erg et i c of fset between these s t a te s is $E_ {+}-E_ { -} = \sqrt{E_ {l r}^ 2+4t ^ 2 }$ [@ staf f or d-prl96; @stoo f -p rb 96]. Fo rweak t un nel co uplin g and non-z ero detu ning( $E_l \not = E_ r $) it follows th a t a bond ing electron i s lo c aliz ed i n o neo f the dots ( a ss hown in Fig. \[prop er ties\] (b)). At zero detu ning ($E_l = E_r$), t he b o nd i ng electron tu nnels back andf orth bet weenthe dots at a Rab i frequenc y o f $ \Om ega _ R = t/h$. The di s c rete c onducta nce peaksinFig . \ [pr op erties\]( a) are m ai npe ak s ( M) re s ulting f ro m r es ona nt tu n neling thro ughth em ole cular s t at e s wit h$E _{lr } =0$.
With i n t he diam ond shape d r e gion s(F ig. \[p roperties\](a )) enclosedby da shed l i n es Coulo mb blockade prevails an d the nu mbe r ofelec trons inthe dotsisw ell de fined, corr es pon d i ng to t he ch ar ge configu r a tio n ($N _l $, $ N_r$).In the direction m a rke d by arrow [* *A* *]{} i nFig . \ [ pro pe r tie s \ ](a), the detun ing $E_{lr }$ is kept cons t ant ,while t he aver age o f the gr ound stat e energie s$\ba r E = (E_{l}+E_ {r})/2$is increa s ed re l at ive t o t he che mi cal pote ntials ($\ mu_{d }$, $\ mu _{s}$) of t he contact s. In direction of [**B **]{}, $\ba r E $ is fixe d b u t $ E_{lr}$ i s va ried. We f ind th at th e b r oaden ingi nbot h dire ctio n s | alpha]. Both_dots show_excited states at $\Delta_\epsilon^* \approx_120~\mu$eV._In Fig. \[properties\](a)_the_charging diagram of_the coupled dot_is plotted in a_linear grayscale representation_in_the weak coupling regime ($G_{c} \approx 0.08~e^2/h$). In linear transport, only two ground states_participate,_e.g., the_ground_state_$E_l$ in the left dot_and $E_r$ in the right_dot, as_schematically shown in Fig. \[properties\](b) [@simplepicture]. The detuning between the_two_ground states is_$E_{lr} = - (E_l-E_r)$. The finite tunnel coupling $t$_between these two states induces a_bonding ($E_{-}$) and_an_anti-bonding_($E_{+}$) molecular state, as_shown in Fig. \[properties\](b). The energetic offset_between these states is $E_{+}-E_{-} =_\sqrt{E_{lr}^2+4t^2}$ [@stafford-prl96; @stoof-prb96]. For weak tunnel coupling and_non-zero detuning ($E_l \not = E_r$)_it follows that a bonding_electron is_localized in one of the_dots (as shown_in Fig. \[properties\](b))._At zero detuning_($E_l = E_r$), the bonding electron_tunnels back and_forth between the dots at a_Rabi_frequency of $\Omega_R_=_t/h$._The discrete_conductance peaks in_Fig. \[properties\](a)_are main_peaks_(M) resulting from resonant tunneling through_the_molecular states with $E_{lr} = 0$.
Within_the diamond shaped regions_(Fig. \[properties\](a))_enclosed by dashed lines_Coulomb blockade prevails and the_number of electrons in the dots_is well_defined, corresponding_to the charge configuration ($N_l$, $N_r$). In the direction marked by_arrow [**A**]{} in Fig. \[properties\](a), the detuning_$E_{lr}$ is kept constant,_while the_average_of the ground_state_energies $\bar_E = (E_{l}+E_{r})/2$ is increased relative to_the chemical_potentials ($\mu_{d}$, $\mu_{s}$) of the contacts._In direction of [**B**]{},_$\bar_E$ is fixed but $E_{lr}$ is_varied. We find that the broadening_in both directions |
Geom.* **25** (1987), no.1, 55–73.
A. Weinstein, P. Xu, Extensions of symplectic groupoids and quantization, *J. Reine Angew. Math.*, **417** (1991), 159–189.
A. Weinstein, Lagrangian mechanics and groupoids, *Fields Inst. Commun.*, **7**, (1996), 207–231.
A. Weinstein, Linearization of regular proper groupoids, *J. Inst. Math. Jussieu* **1** (2002), no.3, 493–511.
M. Zambon, Submanifolds in Poisson geometry: a survey. *Complex and differential geometry*, *Springer Proc. Math.*, **8** (2011), 403–420.
N.T. Zung, Proper Groupoids and Momentum Maps: Linearization, Affinity and Convexity, *Ann. Sci. Éc. Norm. Supér., Sér. 4*, **39** (2006), no.5, 841–869.
Samenvatting {#samenvatting.unnumbered}
============
Dit proefschrift presenteert een aantal nieuwe resultaten in het vakgebied Poisson-meetkunde. Het gaat hoofdzakelijk om een normaalvormstelling (‘normal form theorem’), een stelling over lokale rigiditeit en een expliciete beschrijving van de Poisson-moduli-ruimtes rond Lie-Poissonbollen. Daarnaast bewijzen we ook een standaardvormstelling voor symplectische foliaties, een stelling over formele equivalentie rond Poissondeelvariëteiten en een resultaat over het getemd verdwijnen van Lie-algebroïde-cohomologie. We geven ook in detail de bewijzen van enkele bekende resultaten: de existentie van symplectische realisaties (met een origineel bewijs), Conn’s lineariserings-stelling (met enkele vereenvoudigingen) en een resultaat van Hamilton over de rigiditeit van foliaties (welke een applicatie is van het getemd verdw | Geom. * * * 25 * * (1987), no.1, 55–73.
A. Weinstein, P. Xu, Extensions of symplectic groupoids and quantization, * J. Reine Angew. Math. *, * * 417 * * (1991), 159–189.
A. Weinstein, Lagrangian mechanics and groupoids, * Fields Inst. Commun. *, * * 7 * *, (1996), 207–231.
A. Weinstein, Linearization of regular proper groupoids, * J. Inst. Math. Jussieu * * * 1 * * (2002), no.3, 493–511.
M. Zambon, Submanifolds in Poisson geometry: a survey. * Complex and differential geometry *, * Springer Proc. Math. *, * * 8 * * (2011), 403–420.
N.T. Zung, Proper Groupoids and Momentum Maps: Linearization, Affinity and Convexity, * Ann. Sci. Éc. Norm. Supér. , Sér. 4 *, * * 39 * * (2006), no.5, 841–869.
Samenvatting { # samenvatting.unnumbered }
= = = = = = = = = = = =
Dit proefschrift presenteert een aantal nieuwe resultaten in het vakgebied Poisson - meetkunde. Het gaat hoofdzakelijk om een normaalvormstelling (‘ normal form theorem ’), een stelling over lokale rigiditeit en een expliciete beschrijving van de Poisson - modulus - ruimtes rond Lie - Poissonbollen. Daarnaast bewijzen we ook een standaardvormstelling voor symplectische foliaties, een stelling over formele equivalentie rond Poissondeelvariëteiten en een resultaat over het getemd verdwijnen van Lie - algebroïde - cohomologie. We geven ook in contingent de bewijzen van enkele bekende resultaten: de existentie van symplectische realisaties (met een origineel bewijs), Conn ’s lineariserings - stelling (met enkele vereenvoudigingen) en een resultaat van Hamilton over de rigiditeit van foliaties (welke een applicatie is van het getemd verdw | Geol.* **25** (1987), no.1, 55–73.
A. Weinstein, P. Xu, Exuensions of symplgcric grmupoida and quxntization, *J. Reine Angew. Mati.*, **417** (1991), 159–189.
Q. Weinwtein, Lagrangian mechavics and hroupoidw, *Fitlds Inst. Commun.*, **7**, (1996), 207–231.
A. Weinsbzin, Ljkearivavion of regular proper grmupoids, *J. Inst. Mdtf. Jbssieu* **1** (2002), no.3, 493–511.
M. Zambon, Submanifolds in [oisson gfometry: a survgy. *Cok[lex and differential geometry*, *Springed Proc. Kath.*, **8** (2011), 403–420.
N.T. Zung, Proper Groupoids and Momejtum Maps: Linearizatioj, Affinity qnd Sinvexity, *Ann. Sci. Éc. Norm. Supér., Sér. 4*, **39** (2006), no.5, 841–869.
Samenvatting {#samenvatting.unvumbexed}
============
Dit proedsxhrlxt presentexrt eeg aantal niemee resgltaten in het vakgebled Pmiswon-meetkunde. Het gaat hoofdzakelijk om gen normaanvkrmstelling (‘normao dorm jheoram’), edb sgelmiig kver llkame rigiditsit en een wxpliciete beschrijfigt van de Poisaon-modtly-ruimtes rond Lie-Poissonbollen. Daarnaasu bewjjzen we ook een standaqrdvormstelling voor dymplectifche foliaties, een stelling over formele equivalettie covd Kolfrinfeelvariëteiten en een resultaat over het geteis fegdwijnen van Lie-ajgebroïde-cohpmllpdie. We geven uok in defail de bewijzen vwn enkeje bejende restltayen: de existentie van symplwctische reajusaties (met een orngineel bewils), Conm’s limeariserings-stelling (mec enkeme vereenvokdigingen) dn een resultaat vak Hdmilton over de rigiditeit van folieties (welke den spplicwtie is vaj het getemd verdw | Geom.* **25** (1987), no.1, 55–73. A. Weinstein, Extensions symplectic groupoids quantization, *J. Reine A. Lagrangian mechanics and *Fields Inst. Commun.*, (1996), 207–231. A. Weinstein, Linearization of proper groupoids, *J. Inst. Math. Jussieu* **1** (2002), no.3, 493–511. M. Zambon, Submanifolds Poisson geometry: a survey. *Complex and differential geometry*, *Springer Proc. Math.*, **8** (2011), N.T. Proper and Maps: Linearization, Affinity and Convexity, *Ann. Sci. Éc. Norm. Supér., Sér. 4*, **39** (2006), no.5, 841–869. {#samenvatting.unnumbered} ============ Dit proefschrift presenteert een aantal nieuwe in het vakgebied Poisson-meetkunde. gaat hoofdzakelijk om een normaalvormstelling form een stelling lokale en expliciete beschrijving van Poisson-moduli-ruimtes rond Lie-Poissonbollen. Daarnaast bewijzen we ook een standaardvormstelling voor symplectische foliaties, een stelling over formele equivalentie Poissondeelvariëteiten en over het verdwijnen Lie-algebroïde-cohomologie. geven ook in bewijzen van enkele bekende resultaten: de realisaties (met een origineel bewijs), Conn’s lineariserings-stelling (met vereenvoudigingen) en resultaat van Hamilton over de rigiditeit foliaties (welke een applicatie is van het getemd | Geom.* **25** (1987), no.1, 55–73.
A. Weinstein, P. Xu, ExtensIons of sympLectiC grOupOiDs anD quaNtization, *J. ReinE angeW. Math.*, **417** (1991), 159–189.
A. Weinstein, LagrangIan meChANics ANd GroupOids, *FieLDs iNSt. COmMuN.*, **7**, (1996), 207–231.
A. WEiNStEin, LiNeaRizatioN of regular ProPeR groupoids, *J. INSt. math. JussieU* **1** (2002), no.3, 493–511.
m. Zambon, SubmaNifOlds in poIssON geomEtrY: a surVey. *ComPLex and DifferentIaL GeometRY*, *SpringER prOc. MaTh.*, **8** (2011), 403–420.
N.T. Zung, Proper GroUPoIDs and Momentum MAps: LinEaRIzATIon, affInity and CoNvExity, *aNn. Sci. Éc. nOrM. sUPér., sÉr. 4*, **39** (2006), no.5, 841–869.
SamenvattIng {#samenvatTIng.UnnumbErEd}
============
DIT proefSchriFt PResEnteert een aAntaL nieuwe reSultatEN in het vAKgebied poissoN-meEtkUnde. hEt GaAt hOoFDzaKElIjk OM eeN normaalVoRmStellIng (‘nORMAL forM thEoreM’), een sTelling over loKalE rigIDitEit en Een exPlicIeTe besChrijvIng vaN dE Poisson-moduli-rUimtEs rond Lie-poiSsOnbOlLen. DaARnaast BewIjzEn we ook Een stanDAarDvORMStElling voor symplectIsCHE fOliaties, Een steLLiNg OVer formeLe EquIvalENTie roNd PoISsOndeelvaRiëteiTEn En Een resuLtAat oveR hEt gEteMd verDWijnEn van LIe-algebrOïde-cOHomologie. We gevEN ook in detail dE BeWIJzEN van EnkEle bekende rEsulTAten: De exIStEntIE van sYmpleCtIScHE realisaties (met een oRiGineel BewijS), Conn’s lineariSerings-steLLINg (met enkEle vEReENvoudigingen) en Een reSultaat van hAmilton oVer de RigiditeIt van foliATIes (welke Een AppLicAtiE IS vAn het getemd veRDW | Geom.* **25** (1987), no.1 , 55–73.
A. We ins tei n, P.Xu,Extensions ofs ympl ectic groupoids and qu antiz at i on,* J. Rein e Angew . M a t h.* ,** 417 ** (1 991), 15 9–189.
A. Weinst ein ,Lagrangian m e ch anics andgro upoids, *Fie lds Inst. C omm u n.*,**7 **, ( 1996), 207–23 1.
A. We in s tein,L ineariz a t io n of regular proper g r ou p oids, *J. Inst . Math .J us s i eu* ** 1** (2002) ,no.3, 493–511 .
M . Zam b on, Submanifo lds in Pois s ongeomet ry : a survey . *Co mp l exand differe ntia l geometr y*, *S p ringerP roc. Ma th.*,**8 **(201 1 ), 4 03– 42 0 .
N .T . Z u ng, ProperGr ou poids and M o m entu m M aps: Line arization, Af fin itya ndConve xity, *An n. Sci. Éc. N orm.Su pér., Sér. 4*,**39 ** (2006) , n o. 5,84 1–869 .
Same nva tti ng {#sa menvatt i ng. un n u m be red}
============
D i t p roefschr ift pr e se nt e ert eenaa nta l ni e u we re sult a te n in het vakge b ie dPoisson -m eetkun de . H etgaath oofd zakeli jk om ee n nor m aalvormstellin g (‘normal for m t h e or e m’), ee n stellingover loka le r i gi dit e it en eenex p li c iete beschrijving v an de Po isson -moduli-ruimt es rond Li e - P oissonbo llen . D a arnaast bewijz en we ook een s t andaardv ormst elling v oor sympl e c tische f oli ati es, ee n st elling over f o r mele e quivale nti e rondPoi sso nde elv ar iëteitenen een r es ul ta at ov er he t getemdve rdw ij nen vanL ie-alg ebroï de-c oh om o log ie. Weg ev e n ook i ndeta ilde bewi jzen van enkele bekenderes u ltat en :de exis tentie van sy mp lectischere ali saties ( met eenorigineel bewijs), Conn ’ s linea ris ering s-st elling (m etenkele ve r eenvou diging en) e neen r esult a a tvan H amilton ov e r de rigi di teit van fo liaties (welke een app licatie is va n h et g e t em d v e rd w | Geom.* **25**_(1987), no.1,_55–73.
A. Weinstein, P. Xu, Extensions of_symplectic groupoids_and_quantization, *J._Reine_Angew. Math.*, **417**_(1991), 159–189.
A. Weinstein, Lagrangian_mechanics and groupoids, *Fields_Inst. Commun.*, **7**,_(1996),_207–231.
A. Weinstein, Linearization of regular proper groupoids, *J. Inst. Math. Jussieu* **1** (2002), no.3, 493–511.
M. Zambon, Submanifolds in Poisson_geometry:_a survey._*Complex_and_differential geometry*, *Springer Proc. Math.*,_**8** (2011), 403–420.
N.T. Zung, Proper Groupoids_and Momentum_Maps: Linearization, Affinity and Convexity, *Ann. Sci. Éc._Norm._Supér., Sér. 4*,_**39** (2006), no.5, 841–869.
Samenvatting {#samenvatting.unnumbered}
============
Dit proefschrift presenteert een aantal_nieuwe resultaten in het vakgebied Poisson-meetkunde._Het gaat hoofdzakelijk_om_een_normaalvormstelling (‘normal form theorem’),_een stelling over lokale rigiditeit en_een expliciete beschrijving van de Poisson-moduli-ruimtes_rond Lie-Poissonbollen. Daarnaast bewijzen we ook een_standaardvormstelling voor symplectische foliaties, een stelling_over formele equivalentie rond Poissondeelvariëteiten_en een_resultaat over het getemd verdwijnen_van Lie-algebroïde-cohomologie. We_geven ook_in detail de_bewijzen van enkele bekende resultaten: de_existentie van symplectische_realisaties (met een origineel bewijs), Conn’s_lineariserings-stelling_(met enkele vereenvoudigingen)_en_een_resultaat van_Hamilton over de_rigiditeit_van foliaties_(welke_een applicatie is van het getemd_verdw |
) = \min_{\pi'} \max_{1 \leq t \leq k'} \pi'(t)$, where $\pi'$ ranges over all paths starting at $\pi'(1) = \pi(t_1)$ and ending at $\pi'(k') = \pi(t_2)$.
$P$ and $Q$ are curves with $m$ and $n$ edges A locally correct discrete [Fréchet matching]{} for $P$ and $Q$
Construct grid $G$ for $P$ and $Q$ Let $T$ be a tree consisting only of the root $G[0,0]$ Add $G[i,0]$ to $T$ Add $G[0,j]$ to $T$ $\texttt{AddToTree}(T, G, i, j)$ path in $T$ between $G[0,0]$ and $G[m,n]$
#### Algorithm
The algorithm needs to compute a locally correct path between $G[0,0]$ and $G[m,n]$ in a grid $G$ of non-negative values. To this end, the algorithm incrementally constructs a tree $T$ on the grid such that each path in $T$ is locally correct. The algorithm is summarized by Algorithm \[alg:finddiscrete\]. We define a *growth node* as a node of $T$ that has a neighbor in the grid that is not yet part of $T$: a new branch may sprout from such a node. The growth nodes form a sequence of horizontally or vertically neighboring nodes. A *living node* is a node of $T$ that is not a growth node but is an ancestor of a growth node. A *dead node* is a node of $T$ that is neither a living nor a growth node, that is, it has no descendant that is a growth node. Every pair of nodes in this tree has a *nearest common ancestor* (NCA). When we have to decide what parent to use for a new node in the tree, we look at the maximum value on the path in the tree between the parents and their NCA (excluding the value of the latter). A *face* of the tree is the area enclosed by the segment between two horizontally or vertically neighboring growth nodes (without one being the parent of another) and the paths to their NCA. The unique *sink* of a face is the | ) = \min_{\pi' } \max_{1 \leq t \leq k' } \pi'(t)$, where $ \pi'$ ranges over all paths depart at $ \pi'(1) = \pi(t_1)$ and end at $ \pi'(k') = \pi(t_2)$.
$ P$ and $ Q$ are curves with $ m$ and $ n$ edges A locally correct discrete [ Fréchet matching ] { } for $ P$ and $ Q$
Construct grid $ G$ for $ P$ and $ Q$ get $ T$ be a tree consisting only of the etymon $ G[0,0]$ total $ G[i,0]$ to $ T$ Add $ G[0,j]$ to $ T$ $ \texttt{AddToTree}(T, G, i, j)$ way in $ T$ between $ G[0,0]$ and $ G[m, n]$
# # # # Algorithm
The algorithm need to compute a locally right path between $ G[0,0]$ and $ G[m, n]$ in a grid $ G$ of non - negative values. To this goal, the algorithm incrementally constructs a tree $ T$ on the grid such that each path in $ T$ is locally correct. The algorithm is summarized by Algorithm \[alg: finddiscrete\ ]. We specify a * growth node * as a node of $ T$ that receive a neighbor in the grid that is not yet part of $ T$: a fresh branch may sprout from such a node. The growth nodes imprint a sequence of horizontally or vertically neighboring nodes. A * living node * is a node of $ T$ that is not a growth node but is an ancestor of a growth node. A * dead node * is a node of $ T$ that is neither a living nor a growth node, that is, it has no descendant that is a increase node. Every pair of node in this tree take a * nearest common ancestor * (NCA). When we have to decide what parent to practice for a new lymph node in the tree, we look at the maximum value on the path in the tree between the parent and their NCA (excluding the value of the latter). A * face * of the tree is the area enclosed by the segment between two horizontally or vertically neighboring growth nodes (without one being the parent of another) and the paths to their NCA. The unique * sink * of a side is the | ) = \mln_{\pi'} \max_{1 \leq t \leq k'} \pi'(t)$, where $\pi'$ ranges over ell patgs startkng at $\pi'(1) = \pi(t_1)$ and ending at $\pu'(k') = \pu(t_2)$.
$P$ and $Q$ are curves wkth $m$ and $n$ edges A lixally corrxdt discvzte [Fdéghet ketching]{} for $P$ anc $Q$
Construwt grid $G$ for $[$ xnb $Q$ Let $T$ be a tree consisting only jf the tolt $G[0,0]$ Add $G[i,0]$ to $T$ Acq $G[0,j]$ no $T$ $\texttt{AddToTree}(T, G, i, j)$ patg in $T$ uetween $G[0,0]$ and $G[k,n]$
#### Algorithm
The algorithm jeedd to compute a locwlly correcj paey between $G[0,0]$ xnd $G[m,n]$ in a grid $G$ of non-negative values. To this end, the clgorithm ibceemfttally consvructs a tree $T$ on the grig such yhat each path in $T$ us locally correct. Thx algorithm is summatized by Angkrithm \[alg:finddisceere\]. We defhne x *gruwtg iods* as a nove of $T$ thaf has a neithbor in the grid tnae is not yet pzrt of $T$: a new branch may sprout from such a noge. Fhe growth nodes form a sequence of horizontwlly or vqrtically neighboring nodes. A *living node* is a noge of $G$ tkqt is boh a growth node but is an ancestor of a growtr npdv. A *dead node* is c node of $T$ that id mgither a livine nor c gdowth node, that is, it has no dwscendant thay is a growth node. Every paur of nodes pn tyis tree has a *neaxest common cncestpr* (NCS). When we have to decidz what parent to kse for a vew node in the grev, we look at the maximum value on the peth iu the trde bgtween ehe parentd and their NCA (excludinh the vdlue of thf latter). A *face* of the tree is vie area encloxeg bj the segient netween two horyzontally or vgrtically neigfboring grkwth noves (without jne being the karent of anovher) and ehe pathw to thdkr NCA. The uniaue *sink* of a face us the | ) = \min_{\pi'} \max_{1 \leq t \leq where ranges over paths starting at at = \pi(t_2)$. $P$ $Q$ are curves $m$ and $n$ edges A locally discrete [Fréchet matching]{} for $P$ and $Q$ Construct grid $G$ for $P$ and Let $T$ be a tree consisting only of the root $G[0,0]$ Add $G[i,0]$ $T$ $G[0,j]$ $T$ G, i, j)$ path in $T$ between $G[0,0]$ and $G[m,n]$ #### Algorithm The algorithm needs to a locally correct path between $G[0,0]$ and $G[m,n]$ a grid $G$ of values. To this end, the incrementally a tree on grid that each path $T$ is locally correct. The algorithm is summarized by Algorithm \[alg:finddiscrete\]. We define a *growth node* as node of has a in grid is not yet $T$: a new branch may sprout node. The growth nodes form a sequence of or vertically nodes. A *living node* is a of $T$ that is not a growth node is an ancestor of a growth node. A *dead node* is a node of $T$ neither a living nor growth node, that it no that a growth Every pair of nodes in this tree has a *nearest common (NCA). When we have to decide what parent to use new in the tree, look at the maximum on path in the tree parents their value the A *face* of the is the area enclosed by segment between two horizontally (without one being the parent of another) and paths to their NCA. The unique *sink* a face is the | ) = \min_{\pi'} \max_{1 \leq t \leq k'} \pi'(t)$, where $\pI'$ ranges oveR all pAthS stArTing At $\pi'(1) = \Pi(t_1)$ and ending at $\PI'(k') = \pi(T_2)$.
$P$ and $Q$ are curves with $m$ anD $n$ edgEs a LocaLLy CorreCt discrETe [fRÉchEt MaTchInG]{} FoR $P$ and $q$
CoNstruct Grid $G$ for $P$ aNd $Q$ leT $T$ be a tree conSIsTing only of The Root $G[0,0]$ Add $G[i,0]$ to $t$ AdD $G[0,j]$ to $T$ $\TeXttT{addTotreE}(T, G, i, j)$ Path in $t$ BetweeN $G[0,0]$ and $G[m,n]$
#### ALgORithm
THE algoriTHM nEeds To compute a locally COrREct path between $g[0,0]$ and $G[m,N]$ iN A gRID $G$ oF noN-negative vAlUes. To THis end, tHE aLGORitHM incrementallY constructs A TreE $T$ on thE gRid SUch thaT each PaTH in $t$ is locally cOrreCt. The algoRithm iS SummariZEd by AlgOrithm \[Alg:FinDdisCReTe\]. we dEfINe a *GRoWth NOde* As a node oF $T$ ThAt has A neiGHBOR in tHe gRid tHat is Not yet part of $T$: A neW braNCh mAy sprOut frOm suCh A node. the groWth noDeS form a sequence oF horIzontally Or vErTicAlLy neiGHborinG noDes. a *living Node* is a NOde Of $t$ THAt Is not a growth node buT iS AN aNcestor oF a growTH nOdE. a *dead nodE* iS a nOde oF $t$ That iS neiTHeR a living Nor a grOWtH nOde, that Is, It has nO dEscEndAnt thAT is a Growth Node. EverY pair OF nodes in this trEE has a *nearest cOMmON AnCEstoR* (NCa). When we have To deCIde wHat pAReNt tO Use foR a new NoDE iN The tree, we look at the mAxImum vaLue on The path in the tRee between THE Parents aNd thEIr ncA (excluding the Value Of the latteR). a *face* of tHe treE is the arEa encloseD BY the segmEnt BetWeeN twO HOrIzontally or veRTIcalLy NeighboRinG growth NodEs (wIthOut OnE being the Parent of AnOtHeR) aNd tHe patHS to their nCa. ThE uNiqUe *sinK* Of a facE is thE | ) = \min_{\pi'} \max_{1 \l eq t \leqk'} \ pi' (t) $, whe re $ \pi'$ ranges o v er a ll paths starting at $ \pi'( 1) = \p i (t _1)$and end i ng a t $ \p i' (k' )= \ pi(t_ 2)$ .
$P$and $Q$ ar e c ur ves with $m$ an d $n$ edge s A locally cor rec t disc re te[ Fréch etmatch ing]{} for $P $ and $Q$
C onstru c t grid$ G $for$P$ and $Q$ Let $ T $b e a tree consi stingon l yo f th e r oot $G[0,0 ]$ Add$ G[i,0]$ to $ T $ A d d $G[0,j]$ to $T$ $\text t t{A ddToTr ee }(T , G, i, j)$pa t h i n $T$ betwe en $ G[0,0]$ a nd $G[ m ,n]$
# # ## Algo rithm
Th e a lgor i th mnee ds toc om put e alocallyco rr ect p athb e t w een$G[ 0,0] $ and $G[m,n]$ ina g rid$ G$of no n-neg ativ evalue s. Tothisen d, the algorith m in cremental lyco nst ru cts a tree $ T$onthe gri d sucht hat e a c h p ath in $T$ is loca ll y co rrect. T he alg o ri th m is summ ar ize d by A lgori thm\ [a lg:findd iscret e \] .We defi ne a *gr ow thnod e* as a no de of$T$ that hasa neighbor in t h e grid that i s n o t y e t pa rtof $T$: a n ew b r anch may sp rou t from such a no d e. The growth nodes f orm aseque nce of horizo ntally orv e r ticallyneig h bo r ing nodes. A * livin g node* is a node o f $T$ that is not a gr o w th nodebut is an an c e st or of a growt h node .A *dead no de* isa n ode of $T $that is n either a l iv in gnor a gr o wth node ,tha tis, it h a s no d escen dant t ha t is a grow t hn o de.Ev er y pa irof node s in thi s treehas a *ne are s t co mm on ancest or* (NCA). Wh en we have t odec ide wh a t parentto use for a new node i n the tr ee, we l ookat the ma xim um val ueo n thepath i n the t ree b etwee n th e p ar ents and t h e irNCA ( ex clud ing the value of the latt e r). A *face* ofthe tre e is th e a r eaen c los e d by the segment between t wo ho rizontally orve rticall y neigh borin g growth nodes (w ithout on ebein g the parent of another ) and the paths to thei r N CA. Th euni que * sink*o f a face is th e | ) =_\min_{\pi'} \max_{1_\leq t \leq k'}_\pi'(t)$, where_$\pi'$_ranges over_all_paths starting at_$\pi'(1) = \pi(t_1)$_and ending at $\pi'(k')_= \pi(t_2)$.
$P$ and_$Q$_are curves with $m$ and $n$ edges A locally correct discrete [Fréchet matching]{} for $P$_and_$Q$
Construct grid_$G$_for_$P$ and $Q$ Let $T$_be a tree consisting only_of the_root $G[0,0]$ Add $G[i,0]$ to $T$ Add $G[0,j]$_to_$T$ $\texttt{AddToTree}(T, G,_i, j)$ path in $T$ between $G[0,0]$ and $G[m,n]$
####_Algorithm
The algorithm needs to compute a_locally correct path_between_$G[0,0]$_and $G[m,n]$ in a_grid $G$ of non-negative values. To_this end, the algorithm incrementally constructs_a tree $T$ on the grid such_that each path in $T$ is_locally correct. The algorithm is_summarized by_Algorithm \[alg:finddiscrete\]. We define a *growth_node* as a_node of_$T$ that has_a neighbor in the grid that_is not yet_part of $T$: a new branch_may_sprout from such_a_node._The growth_nodes form a_sequence_of horizontally_or_vertically neighboring nodes. A *living node*_is_a node of $T$ that is not_a growth node but_is_an ancestor of a_growth node. A *dead node*_is a node of $T$ that_is neither_a living_nor a growth node, that is, it has no descendant that_is a growth node. Every pair_of nodes in this_tree has_a_*nearest common ancestor*_(NCA)._When we_have to decide what parent to use_for a_new node in the tree, we_look at the maximum_value_on the path in the tree_between the parents and their NCA_(excluding the value of the_latter)._A_*face* of the tree is_the area enclosed by the segment_between two horizontally_or vertically neighboring growth nodes (without one_being_the parent of another) and the_paths_to their NCA. The unique *sink*_of_a_face is the |
&\leq \norm{T_{{\bm{x}}^{-1}}} \cdot \norm{{\Delta{\bm{x}}}- {{\overline{{\Delta{\bm{x}}}}}}}_F \leq \xi, \text{ and } \\
\norm{{{{\Delta{\bm{s}}}'}}- {{{{\overline{{\Delta{\bm{s}}}}}'}}}}_F &= \mu^{-1}\norm{T_{{\bm{x}}} {\Delta{\bm{s}}}- T_{{\bm{x}}} {{\overline{{\Delta{\bm{s}}}}}}}_F \\
&\leq \mu^{-1}\norm{T_{{\bm{x}}}} \norm{{\Delta{\bm{s}}}- {{\overline{{\Delta{\bm{s}}}}}}}_F \\
&= \mu^{-1} \norm{{\bm{x}}}_2 \norm{{\Delta{\bm{s}}}- {{\overline{{\Delta{\bm{s}}}}}}}_F \\
&\leq (1+\eta) \norm{{\bm{s}}^{-1}}_2 \norm{{\Delta{\bm{s}}}- {{\overline{{\Delta{\bm{s}}}}}}}_F \text{ by Lemma \ref{lemma:Properties of the central path}} \\
&\leq 2 \norm{T_{{\bm{s}}^{-1}}} \norm{{\Delta{\bm{s}}}- {{\overline{{\Delta{\bm{s}}}}}}}_F \leq \xi.
\end{aligned}$$
Throughout the analysis, we will make use of several constants: $\eta > 0$ is the distance from the central path, i.e. we ensure that our iterates stay in the $\eta$-neighborhood $\mathcal{N}_\eta$ of the central path. The constant $\sigma = 1 - \chi / \sqrt{r}$ is the factor by which we aim to decrease our duality gap, for some constant $\chi > 0$. Finally constant $\xi > 0$ is the approximation error for the scaled increments ${{{{\overline{{\Delta{\bm{x}}}}}'}}}, {{{{\overline{{\Delta{\bm{s}}}}}'}}}$. Having this notation in mind, we can state several facts about the relation between the duality gap and the central path distance for the original and scaled vectors.
\[claim:stuff preserved under scaling\] The following holds for the scaled vectors ${{{\bm{x}}'}}$ and ${{{\bm{s}}'}}$:
1. The scaled duality gap is $\frac1r {{{\bm{ | & \leq \norm{T_{{\bm{x}}^{-1 } } } \cdot \norm{{\Delta{\bm{x}}}- { { \overline{{\Delta{\bm{x}}}}}}}_F \leq \xi, \text { and } \\
\norm{{{{\Delta{\bm{s}}}'}}- { { { { \overline{{\Delta{\bm{s}}}}}'}}}}_F & = \mu^{-1}\norm{T_{{\bm{x } } } { \Delta{\bm{s}}}- T_{{\bm{x } } } { { \overline{{\Delta{\bm{s}}}}}}}_F \\
& \leq \mu^{-1}\norm{T_{{\bm{x } } } } \norm{{\Delta{\bm{s}}}- { { \overline{{\Delta{\bm{s}}}}}}}_F \\
& = \mu^{-1 } \norm{{\bm{x}}}_2 \norm{{\Delta{\bm{s}}}- { { \overline{{\Delta{\bm{s}}}}}}}_F \\
& \leq (1+\eta) \norm{{\bm{s}}^{-1}}_2 \norm{{\Delta{\bm{s}}}- { { \overline{{\Delta{\bm{s}}}}}}}_F \text { by Lemma \ref{lemma: Properties of the central path } } \\
& \leq 2 \norm{T_{{\bm{s}}^{-1 } } } \norm{{\Delta{\bm{s}}}- { { \overline{{\Delta{\bm{s}}}}}}}_F \leq \xi.
\end{aligned}$$
Throughout the analysis, we will seduce habit of several constants: $ \eta > 0 $ is the distance from the central way, i.e. we ensure that our iterates stay in the $ \eta$-neighborhood $ \mathcal{N}_\eta$ of the central way. The ceaseless $ \sigma = 1 - \chi / \sqrt{r}$ is the factor by which we drive to decrease our duality gap, for some ceaseless $ \chi > 0$. Finally constant $ \xi > 0 $ is the estimate error for the scaled increments $ { { { { \overline{{\Delta{\bm{x } } } } }' } } }, { { { { \overline{{\Delta{\bm{s}}}}}'}}}$. Having this notation in mind, we can state respective facts about the relation between the duality col and the central path distance for the original and scaly vectors.
\[claim: material preserved under scaling\ ] The following holds for the scaled vectors $ { { { \bm{x}}'}}$ and $ { { { \bm{s}}'}}$:
1. The scaled dichotomy gap is $ \frac1r { { { \bm { | &\leq \norm{T_{{\bm{x}}^{-1}}} \cdot \norm{{\Dtlta{\bm{x}}}- {{\overline{{\Dglra{\bm{x}}}}}}}_F \leq \xj, \text{ avd } \\
\norm{{{{\Delta{\bm{s}}}'}}- {{{{\overlinx{{\Delra{\bm{s}}}}}'}}}}_D &= \mu^{-1}\norm{T_{{\bm{x}}} {\Delta{\bm{s}}}- T_{{\bm{x}}} {{\ovegline{{\Deltq{\bm{s}}}}}}}_H \\
&\leq \mu^{-1}\norm{V_{{\gm{x}}}} \norm{{\Delta{\gl{s}}}- {{\orecline{{\Delta{\bm{s}}}}}}}_F \\
&= \mu^{-1} \nosm{{\bm{x}}}_2 \norm{{\Deltd{\bo{s}}}- {{\overline{{\Delta{\bm{s}}}}}}}_F \\
&\leq (1+\eta) \norm{{\fm{s}}^{-1}}_2 \nork{{\Dflta{\bm{s}}}- {{\overlige{{\Dekea{\bm{a}}}}}}}_F \text{ by Lemma \ref{lemma:Propertiea of tht central path}} \\
&\leq 2 \norm{T_{{\bm{s}}^{-1}}} \norm{{\Delta{\bm{d}}}- {{\ovfrline{{\Delta{\bm{s}}}}}}}_F \lee \xi.
\ene{alidbed}$$
Throughoug the analjvis, we wilm make use of several constants: $\eta > 0$ is the dusranfg from the cxntral path, i.e. we cmsure dhat out iterates stan in vhe $\wta$-neighborhood $\mathcel{N}_\eta$ of the centraj path. Tha eonstant $\sigma = 1 - \chi / \wqrt{r}$ is dhe wqctur gy wgich wf amm to decrezse our duaoity gap, for some cpnfnsnt $\chi > 0$. Fihally sogstant $\xi > 0$ is the approximation error xor the scaled increments ${{{{\iverline{{\Delta{\bm{x}}}}}'}}}, {{{{\overpine{{\Delta{\fm{s}}}}}'}}}$. Having this notation in mind, we can state sevaral hazts anout rhf relation between the duality gap and the cegfrsl path distance for the oroglnsj and scaled xectors.
\[clzim:stuff preserved under fcalibg\] The fojlowong holds for the scaled vextors ${{{\bm{x}}'}}$ anb ${{{\bn{s}}'}}$:
1. The scaled duapity gap is $\frav1r {{{\bm{ | &\leq \norm{T_{{\bm{x}}^{-1}}} \cdot \norm{{\Delta{\bm{x}}}- {{\overline{{\Delta{\bm{x}}}}}}}_F \leq \xi, } \norm{{{{\Delta{\bm{s}}}'}}- {{{{\overline{{\Delta{\bm{s}}}}}'}}}}_F \mu^{-1}\norm{T_{{\bm{x}}} {\Delta{\bm{s}}}- T_{{\bm{x}}} {{\overline{{\Delta{\bm{s}}}}}}}_F &= \mu^{-1} \norm{{\bm{x}}}_2 {{\overline{{\Delta{\bm{s}}}}}}}_F \\ &\leq \norm{{\bm{s}}^{-1}}_2 \norm{{\Delta{\bm{s}}}- {{\overline{{\Delta{\bm{s}}}}}}}_F \text{ by Lemma of the central path}} \\ &\leq 2 \norm{T_{{\bm{s}}^{-1}}} \norm{{\Delta{\bm{s}}}- {{\overline{{\Delta{\bm{s}}}}}}}_F \leq \xi. \end{aligned}$$ the analysis, we will make use of several constants: $\eta > 0$ is distance the path, we ensure that our iterates stay in the $\eta$-neighborhood $\mathcal{N}_\eta$ of the central path. The constant = 1 - \chi / \sqrt{r}$ is the by which we aim decrease our duality gap, for constant > 0$. constant > is the approximation for the scaled increments ${{{{\overline{{\Delta{\bm{x}}}}}'}}}, {{{{\overline{{\Delta{\bm{s}}}}}'}}}$. Having this notation in mind, we can state several facts about relation between gap and central distance the original and \[claim:stuff preserved under scaling\] The following scaled vectors ${{{\bm{x}}'}}$ and ${{{\bm{s}}'}}$: 1. The scaled gap is {{{\bm{ | &\leq \norm{T_{{\bm{x}}^{-1}}} \cdot \norm{{\Delta{\bM{x}}}- {{\overline{{\delta{\Bm{x}}}}}}}_f \leQ \xI, \texT{ and } \\
\Norm{{{{\Delta{\bm{s}}}'}}- {{{{\ovERlinE{{\Delta{\bm{s}}}}}'}}}}_F &= \mu^{-1}\norm{T_{{\bm{x}}} {\DeLta{\bm{S}}}- T_{{\BM{x}}} {{\ovERlIne{{\DeLta{\bm{s}}}}}}}_F \\
&\LEq \MU^{-1}\NorM{T_{{\Bm{X}}}} \noRm{{\dElTa{\bm{s}}}- {{\OveRline{{\DeLta{\bm{s}}}}}}}_F \\
&= \mu^{-1} \nOrm{{\Bm{X}}}_2 \norm{{\Delta{\bm{S}}}- {{\OvErline{{\DeltA{\bm{S}}}}}}}_F \\
&\leq (1+\eta) \norm{{\Bm{s}}^{-1}}_2 \Norm{{\DeLtA{\bm{S}}}- {{\OverlIne{{\delta{\Bm{s}}}}}}}_F \teXT{ by LemMa \ref{lemmA:PROpertiES of the cENTrAl paTh}} \\
&\leq 2 \norm{T_{{\bm{s}}^{-1}}} \norm{{\dElTA{\bm{s}}}- {{\overline{{\DeLta{\bm{s}}}}}}}_f \lEQ \xI.
\ENd{aLigNed}$$
ThroughOuT the aNAlysis, wE WiLL MAke USe of several coNstants: $\eta > 0$ iS The DistanCe FroM The cenTral pAtH, I.e. wE ensure that Our iTerates stAy in thE $\Eta$-neigHBorhood $\MathcaL{N}_\eTa$ oF the CEnTrAl pAtH. the COnStaNT $\siGma = 1 - \chi / \sqRt{R}$ iS the fActoR BY WHich We aIm to DecreAse our duality Gap, For sOMe cOnstaNt $\chi > 0$. finaLlY consTant $\xi > 0$ Is the ApProximation erroR for The scaled IncReMenTs ${{{{\OverlINe{{\DeltA{\bm{X}}}}}'}}}, {{{{\ovErline{{\DElta{\bm{s}}}}}'}}}$. hAviNg THIS nOtation in mind, we can StATE sEveral faCts aboUT tHe RElation bEtWeeN the DUAlity Gap aND tHe centraL path dIStAnCe for thE oRiginaL aNd sCalEd vecTOrs.
\[cLaim:stUff preseRved uNDer scaling\] The fOLlowing holds fOR tHE ScALed vEctOrs ${{{\bm{x}}'}}$ and ${{{\bm{S}}'}}$:
1. The SCaleD duaLItY gaP Is $\fraC1r {{{\bm{ | &\leq \norm{T_{{\bm{x} }^{-1}}} \ cdot\no rm{ {\ Delt a{\b m{x}}}- {{\ove r line {{\Delta{\bm{x}}}}}}}_ F \le q\ xi,\ te xt{ a nd } \\ \no rm {{ {{\ De l ta {\bm{ s}} }'}}- { {{{\overli ne{ {\ Delta{\bm{s} } }} }'}}}}_F & = \ mu^{-1}\norm {T_ {{\bm{ x} }}{ \Delt a{\ bm{s} }}- T_ { {\bm{x }}} {{\ov er l ine{{\ D elta{\b m { s} }}}} }}_F \\
&\leq \m u ^{-1}\norm{T_{ {\bm{x }} } }\ n orm {{\ Delta{\bm{ s} }}- { { \overli n e{ { \ D elt a {\bm{s}}}}}}} _F \\
& = \m u^{-1} \ nor m {{\bm{ x}}}_ 2\ nor m{{\Delta{\ bm{s }}}- {{\o verlin e {{\Delt a {\bm{s} }}}}}} _F\\ & \l eq (1 +\ e ta) \n orm { {\b m{s}}^{- 1} }_ 2 \no rm{{ \ D e l ta{\ bm{ s}}} - {{\ overline{{\De lta {\bm { s}} }}}}} _F \t ext{ b y Lem ma \re f{lem ma :Properties ofthecentral p ath }} \\
&\ l eq 2 \ nor m{T _{{\bm{ s}}^{-1 } }}\n o r m {{ \Delta{\bm{s}}}- { {\ o v er line{{\D elta{\ b m{ s} } }}}}}_F\l eq\xi. \ end{ a li gned}$$
Throu g ho ut the an al ysis,we wi llmakeu se o f seve ral cons tants : $\eta > 0$ is the distancef ro m th e cen tra l path, i.e . we ensu re t h at ou r iter atesst a yi n the $\eta$-neighb or hood $ \math cal{N}_\eta$of the cen t r a l path.Thec on s tant $\sigma = 1 -\chi / \sq r t{r}$ is thefactor b y which w e aim to d ecr eas e o urd u al ity gap, fors o me c on stant $ \ch i > 0$. Fi nal lycon st ant $\xi> 0$ isth eap pr oxi matio n error f or th esca led i n cremen ts ${ {{{\ ov er l ine {{\Delt a {\ b m {x}} }} }' }}}, {{ {{ \over line { {\D elta{\b m{s}}}}}' }}} $ . Ha vi ng this n otation in mi nd , we can s ta tesevera l facts ab out the relation betwee n the du ali ty ga p an d the cen tra l path di s tancefor th e ori gi nal a nd sc a l ed ve ct ors.
\[cl a i m:s tuffpr eser ved und er scaling\] The f o llo wing holds fo r t he s c a le d v e ct o rs${ { {\b m { x}}'}}$ and ${{ {\bm{s}}'} }$ :
1. The sc a led d ualitygap is$\fra c 1r {{{\ bm{ | _ _&\leq \norm{T_{{\bm{x}}^{-1}}} \cdot \norm{{\Delta{\bm{x}}}-_{{\overline{{\Delta{\bm{x}}}}}}}_F \leq_\xi,_\text{ and_}_\\
_ \norm{{{{\Delta{\bm{s}}}'}}- {{{{\overline{{\Delta{\bm{s}}}}}'}}}}_F_&= \mu^{-1}\norm{T_{{\bm{x}}} {\Delta{\bm{s}}}- T_{{\bm{x}}}_{{\overline{{\Delta{\bm{s}}}}}}}_F \\
__ &\leq \mu^{-1}\norm{T_{{\bm{x}}}} \norm{{\Delta{\bm{s}}}- {{\overline{{\Delta{\bm{s}}}}}}}_F \\
&= \mu^{-1} \norm{{\bm{x}}}_2 \norm{{\Delta{\bm{s}}}- {{\overline{{\Delta{\bm{s}}}}}}}_F_\\
_ __&\leq_(1+\eta) \norm{{\bm{s}}^{-1}}_2 \norm{{\Delta{\bm{s}}}- {{\overline{{\Delta{\bm{s}}}}}}}_F \text{_by Lemma \ref{lemma:Properties of the_central path}}_\\
&\leq 2 \norm{T_{{\bm{s}}^{-1}}} \norm{{\Delta{\bm{s}}}-_{{\overline{{\Delta{\bm{s}}}}}}}_F_\leq \xi.
_ \end{aligned}$$
Throughout the analysis, we will make_use of several constants: $\eta >_0$ is the_distance_from_the central path, i.e._we ensure that our iterates stay_in the $\eta$-neighborhood $\mathcal{N}_\eta$ of the_central path. The constant $\sigma = 1_- \chi / \sqrt{r}$ is the_factor by which we aim_to decrease_our duality gap, for some_constant $\chi >_0$. Finally_constant $\xi >_0$ is the approximation error for_the scaled increments_${{{{\overline{{\Delta{\bm{x}}}}}'}}}, {{{{\overline{{\Delta{\bm{s}}}}}'}}}$. Having this notation in_mind,_we can state_several_facts_about the_relation between the_duality_gap and_the_central path distance for the original_and_scaled vectors.
\[claim:stuff preserved under scaling\] The following_holds for the scaled_vectors_${{{\bm{x}}'}}$ and ${{{\bm{s}}'}}$:
1. _The scaled duality gap is_$\frac1r {{{\bm{ |
for different time steps and extrapolating to zero time step or by simply choosing a small enough time step. An example of time-step extrapolation is shown in figure \[fig:time\_step\_errors\].
- Population control bias. The $f$ distribution is represented by a finite population of configurations which fluctuates due to branching. The population may be controlled in various ways, but this introduces a population control bias which is positive and falls off as the reciprocal of the population. In practice the population control bias is normally so small that it is difficult to detect [@umrigar_1993; @drummond_2004].
- Finite size errors within periodic boundary conditions calculations. It is important to correct for finite size effects carefully, as mentioned in section \[sec:pbc\].
- The pseudopotential approximation inevitably introduces errors. In DMC there is an additional error arising from the localisation [@mitas_1991] or semi-localisation [@casula_2006] of the non-local pseudopotential operator. The localisation error appears to be quite small in the cases for which it has been tested [@drummond_2006b].
![DMC energy against time step for a 64-electron ferromagnetic 2D hexagonal Wigner crystal at density parameter $r_s=50$ a.u. with a Slater-Jastrow wave function. The solid line is a linear fit to the data. []{data-label="fig:time_step_errors"}](ferro_rs50_N064_exp0002_E_v_dt){width=".7\textwidth"}
Practical methods for handling statistical errors in QMC results {#subsec:statistical errors}
----------------------------------------------------------------
Two main practical problems are encountered when dealing with errors in the QMC data: the data are serially correlated and the underlying probability distributions are non-Gaussian. The probability distribution of the local energies has $|E-E_0|^{-4}$ tails, where $E_0$ is a constant. These tails arise from singularities in the local energy such as the divergence at the nodal surface [@trail_2005_1; @trail_2005_2], as shown in figure \[fig:local\_energy\]. In consequence, although the mean energy and its variance are well defined, the variance of the variance is infinity. For other quantities the problem may be even more severe; for example, the probability distributions for the Pul | for different time steps and extrapolating to zero meter gradation or by simply choosing a small adequate time step. An example of clock time - step extrapolation is shown in human body \[fig: time\_step\_errors\ ].
- Population control bias. The $ f$ distribution is represent by a finite population of configurations which fluctuates ascribable to branching. The population may be controlled in various ways, but this introduces a population control bias which is convinced and falls off as the reciprocal of the population. In practice the population command bias is normally so small that it is difficult to detect [ @umrigar_1993; @drummond_2004 ].
- Finite size errors within periodic boundary weather calculations. It is important to correct for finite size effects carefully, as mention in section \[sec: pbc\ ].
- The pseudopotential approximation inevitably introduces errors. In DMC there is an additional error arising from the localisation [ @mitas_1991 ] or semi - localisation [ @casula_2006 ] of the non - local pseudopotential operator. The localisation error appears to be quite small in the cases for which it has been test [ @drummond_2006b ].
! [ DMC energy against time dance step for a 64 - electron ferromagnetic 2D hexagonal Wigner crystal at density parameter $ r_s=50 $ a.u. with a Slater - Jastrow wave affair. The solid line is a linear fit to the data. [ ] { datum - label="fig: time_step_errors"}](ferro_rs50_N064_exp0002_E_v_dt){width=".7\textwidth " }
Practical methods for handling statistical errors in QMC results { # subsec: statistical errors }
----------------------------------------------------------------
Two main practical problems are encountered when dealing with errors in the QMC datum: the data are serially correlate and the underlie probability distributions are non - Gaussian. The probability distribution of the local energies has $ |E - E_0|^{-4}$ tails, where $ E_0 $ is a constant. These tail arise from singularities in the local energy such as the deviation at the nodal surface [ @trail_2005_1; @trail_2005_2 ], as shown in figure \[fig: local\_energy\ ]. In consequence, although the mean department of energy and its variance are well defined, the variance of the variance is infinity. For early quantities the problem may be even more severe; for example, the probability distribution for the Pul | fog different time steps akd extrapolating to zerm time step or by simply choosing a small xnouth tine step. An example of gime-step vxtrapolarion us shown ii figure \[fig:tijc\_step\_zrcors\].
- Populatipn control bias. The $f$ divtfiyution is represented by a finite po[ulatiom lf configuratijns erich fluctuates due to branching. The pkpulatimn may be conyrolled in various ways, buh thls introduces a pooulation cobtroj bias which ks positivt cnd falls orf as the reciprocal of the popjlatipn. In pracjnxe hve populatiin cogtrol bias is normalny so skall that it ix dmffixult to detect [@umrigac_1993; @drummond_2004].
- Finite fize erross within periodic voyndari congitiubs zaldukafions. Lt ms importanf to correcr for finite size egfqbys carefully, as megtyoned in section \[sec:pbc\].
- The pseudopottntiam approximation inevitavly introduces errors. In DMC trere is an additional error arising from the locanisatmov [@mnbwr_1991] og semi-localisation [@casula_2006] of the non-local psetsokotvntial operator. Tme localisation ertog skpears to be qoite smcml in the cases for ahich ij has veen testtd [@drimmond_2006b].
![DMC energy against tume step for q 64-electron ferromaynetic 2D hexcgonal Wignrr crystal at density pcrametsr $r_s=50$ a.u. witj a Slated-Gastrow wave funztipn. The solid line is a lineaw fit to vhe dcta. []{data-uabek="fig:tiie_step_errogs"}](fervm_rs50_N064_exp0002_E_v_dt){width=".7\tedtwidjh"}
Pracdical methlds for handling statistical ercprs in QMC rgsunts {#subsec:scatistlcal errors}
----------------------------------------------------------------
Two iain practical probleis ard encounteged when vealing with errors in tha QMC data: thx data arq seeialoy corrduated and the inderlyiny provability distributlons xde non-Gaussian. Cke probability dixtrkbueiln os the local etergkes nas $|E-D_0|^{-4}$ tails, wheve $D_0$ is a constant. These tahls zrise from singulatibies in tye local energy such ss the divergence wt thx nodau surgacg [@trail_2005_1; @trail_2005_2], as shown in figurs \[fig:locap\_encrgy\]. In consezuenge, ajthough thz mean energy and its variance are well vefined, the variance of the variance is inyikity. For othxr quagtities tve problem may be evwn more severe; fov example, the probabiljty divtribktions for the Pul | for different time steps and extrapolating to step by simply a small enough time-step is shown in \[fig:time\_step\_errors\]. - Population bias. The $f$ distribution is represented a finite population of configurations which fluctuates due to branching. The population may controlled in various ways, but this introduces a population control bias which is and off the of the population. In practice the population control bias is normally so small that it is to detect [@umrigar_1993; @drummond_2004]. - Finite size errors periodic boundary conditions calculations. is important to correct for size carefully, as in \[sec:pbc\]. The pseudopotential approximation introduces errors. In DMC there is an additional error arising from the localisation [@mitas_1991] or semi-localisation [@casula_2006] the non-local The localisation appears be small in the which it has been tested [@drummond_2006b]. time step for a 64-electron ferromagnetic 2D hexagonal crystal at parameter $r_s=50$ a.u. with a Slater-Jastrow function. The solid line is a linear fit the data. []{data-label="fig:time_step_errors"}](ferro_rs50_N064_exp0002_E_v_dt){width=".7\textwidth"} Practical methods for handling statistical errors in QMC results {#subsec:statistical errors} ---------------------------------------------------------------- practical problems are encountered dealing with errors the data: data serially correlated the underlying probability distributions are non-Gaussian. The probability distribution of the energies has $|E-E_0|^{-4}$ tails, where $E_0$ is a constant. These from in the local such as the divergence the surface [@trail_2005_1; @trail_2005_2], as figure In mean and variance are well defined, variance of the variance is For other quantities the severe; for example, the probability distributions for the | for different time steps and eXtrapolatiNg to zEro TimE sTep oR by sImply choosing a SMall Enough time step. An examplE of tiMe-STep eXTrApolaTion is sHOwN IN fiGuRe \[Fig:TiME\_sTep\_erRorS\].
- PopulaTion controL biAs. the $f$ distribuTIoN is represeNteD by a finite poPulAtion oF cOnfIGuratIonS whicH fluctUAtes duE to branchInG. the popULation mAY Be ContRolled in various waYS, bUT this introduceS a popuLaTIoN COntRol Bias which iS pOsitiVE and falLS oFF AS thE Reciprocal of tHe populatioN. in pRacticE tHe pOPulatiOn conTrOL biAs is normallY so sMall that iT is difFIcult to DEtect [@umRigar_1993; @dRumMonD_2004].
- FinITe SiZe eRrORs wIThIn pERioDic boundArY cOnditIons CALCUlatIonS. It iS impoRtant to correcT foR finITe sIze efFects CareFuLly, as MentioNed in SeCtion \[sec:pbc\].
- The pSeudOpotentiaL apPrOxiMaTion iNEvitabLy iNtrOduces eRrors. In dmC tHeRE IS aN additional error arIsING fRom the loCalisaTIoN [@mITas_1991] or semI-lOcaLisaTIOn [@casUla_2006] oF ThE non-locaL pseudOPoTeNtial opErAtor. ThE lOcaLisAtion ERror AppearS to be quiTe smaLL in the cases for WHich it has been TEsTED [@dRUmmoNd_2006b].
![dMC energy agAinsT Time Step FOr A 64-elECtron FerroMaGNeTIc 2D hexagonal Wigner cRyStal at DensiTy parameter $r_s=50$ A.u. with a SlaTER-jastrow wAve fUNcTIon. The solid linE is a lInear fit to THe data. []{daTa-labEl="fig:timE_step_erroRS"}](Ferro_rs50_N064_Exp0002_e_v_dT){wiDth=".7\TEXtWidth"}
PracticaL MEthoDs For handLinG statisTicAl eRroRs iN QmC results {#Subsec:stAtIsTiCaL erRors}
----------------------------------------------------------------
TWO main praCtIcaL pRobLems aRE encouNtereD wheN dEaLIng With errORs IN The QmC DaTa: thE daTa Are seRialLY coRrelateD and the unDerLYing PrObAbility Distributions ArE non-GaussiAn. the ProbabILIty distrIbution of the local energiES has $|E-E_0|^{-4}$ tAilS, wherE $E_0$ is A constant. theSe tailS arISe from SingulAritiEs In tHE Local ENErGy sUcH as the diveRGEncE at thE nOdal Surface [@Trail_2005_1; @trail_2005_2], as shown iN FigUre \[fig:local\_enErgY\]. In cONSeQueNCe, ALthOuGH thE MEan energy and its Variance arE wELl Defined, the VAriAnCe of the VariancE is inFInity. FoR other quaNtities thE pRoblEM May Be even more Severe; foR example, tHE probABiLity dIstRibutiOnS foR the PUl | for different time stepsand extrap olati ngtoze ro t imestep or by sim p ly c hoosing a small enough time s t ep.A nexamp le of t i me - s tep e xt rap ol a ti on is sh own infigure \[f ig: ti me\_step\_er r or s\].
- Pop ulation cont rol bias. T he$ f$ di str ibuti on isr eprese nted by a f i nite p o pulatio n of con figurations which fl u ctuates due to branc hi n g. T hepop ulation ma ybe co n trolled in v a rio u s ways, but t his introdu c esa popu la tio n contr ol bi as whi ch is posit iveand falls off a s the re c iprocal of th e p opu lati o n. I n p ra c tic e t hep opu lation c on tr ol bi as i s n o rmal lyso s mallthat it is di ffi cult todetec t [@u mrig ar _1993 ; @dru mmond _2 004].
- Fini te s ize error s w it hin p eriod i c boun dar y c onditio ns calc u lat io n s . I t is important toco r r ec t for fi nite s i ze e f fects ca re ful ly,a s ment ione d i n sectio n \[se c :p bc \].
- The ps eu dop ote ntial appr oximat ion inev itabl y introduces er r ors. In DMC t h er e is an a ddi tional erro r ar i sing fro m t hel ocali satio n[ @m i tas_1991] or semi-l oc alisat ion [ @casula_2006] of the no n - l ocal pse udop o te n tial operator. Thelocalisati o n errorappea rs to be quite sm a l l in the ca ses fo r w h i ch it has beent e sted [ @drummo nd_ 2006b].
! [DM C e ner gy againsttime ste pfo ra64- elect r on ferro ma gne ti c 2 D hex a gonalWigne r cr ys ta l at densit y p a r amet er $ r_s= 50$ a .u. w itha Sl ater-Ja strow wav e f u ncti on .The sol id line is ali near fit t othe data. [ ]{data-l abel="fig:time_step_err o rs"}](f err o_rs5 0_N0 64_exp000 2_E _v_dt) {wi d th=".7 \textw idth" }
Pr a c tical m et hod sfor handli n g st atist ic al e rrors i n QMC results {#su b sec :statisticalerr ors} - -- --- - -- - --- -- - --- - - --------------- ---------- -- - -- ---------- - --
T wo main practi cal p r oblemsare encou ntered wh en dea l i ngwith error s in the QMC data : thed at a are se rially c orr elate d andt heunder lyingpr obabil ity d is tributio ns are non-Gaussian. Th e prob abili tydistribut ion ofthe local ene rgies has$|E -E_ 0|^{- 4}$ tails , wh e re $E _ 0$ is a c o nstant. T h es e t a i ls arise from s i ngu larit ies in the loc al energy such as the divergence att h e n oda l sur fa ce [@trail_200 5_1 ;@ t rail_200 5_ 2], as show n in fig ur e \[fi g:loca l\_ene rgy\].I n c o nseque nce, al though th e m ea n energy a nd its va rian ce are w ell de f ined , the variance ofthe v a r iance isinfin it y. Foro ther quantitie s the probl em may beevenmore se ve re; fo r e xa mple, thep robabilit y dis tributi on s fo r t he Pul | for_different time_steps and extrapolating to_zero time_step_or by_simply_choosing a small_enough time step._An example of time-step_extrapolation is shown_in_figure \[fig:time\_step\_errors\].
- Population control bias. The $f$ distribution is represented by a_finite_population of_configurations_which_fluctuates due to branching. The_population may be controlled in_various ways,_but this introduces a population control bias which_is_positive and falls_off as the reciprocal of the population. In practice_the population control bias is normally_so small that_it_is_difficult to detect [@umrigar_1993;_@drummond_2004].
- Finite size errors_within periodic boundary conditions calculations. It_is important to correct for finite size_effects carefully, as mentioned in section_\[sec:pbc\].
- The pseudopotential_approximation inevitably_introduces errors. In DMC there_is an additional_error arising_from the localisation_[@mitas_1991] or semi-localisation [@casula_2006] of the_non-local pseudopotential operator._The localisation error appears to be_quite_small in the_cases_for_which it_has been tested_[@drummond_2006b].
![DMC_energy against_time_step for a 64-electron ferromagnetic 2D_hexagonal_Wigner crystal at density parameter $r_s=50$ a.u. with_a Slater-Jastrow wave function._The_solid line is a_linear fit to the data._[]{data-label="fig:time_step_errors"}](ferro_rs50_N064_exp0002_E_v_dt){width=".7\textwidth"}
Practical methods for handling statistical errors_in QMC_results {#subsec:statistical_errors}
----------------------------------------------------------------
Two main practical problems are encountered when dealing with errors in_the QMC data: the data are_serially correlated and the_underlying probability_distributions_are non-Gaussian. The_probability_distribution of_the local energies has $|E-E_0|^{-4}$ tails, where_$E_0$ is_a constant. These tails arise from_singularities in the local_energy_such as the divergence at the_nodal surface [@trail_2005_1; @trail_2005_2], as shown_in figure \[fig:local\_energy\]. In consequence,_although_the_mean energy and its variance_are well defined, the variance of_the variance is_infinity. For other quantities the problem may_be_even more severe; for example, the_probability_distributions for the Pul |
The probability of measuring some set of burst arrival times $t_\mathrm{1}, t_\mathrm{2}, \dots t_\mathrm{N}$ in a single observation of duration $T$ can be split into three parts:
1. The probability of the interval between the start of the observation and the first burst: $P(t_1)$
2. The probabilities of the intervals between subsequent bursts in a single observation: $P(t_2 \dots t_\mathrm{N}) = \prod_\mathrm{i=1}^\mathrm{N-1} P(t_\mathrm{i+1} - t_\mathrm{i})$
3. The probability of the interval between the last burst and the end of the observation: $P(T - t_\mathrm{N})$
Assuming different observations are not correlated, points 1) and 3) describe minimum burst intervals. To include that subsequent observations can be correlated, we include a maximum burst interval, which is simply the interval between the last burst of an observation and the arrival time of next observed burst. Only for the intervals before the first detected burst and after the last burst, there is no constraint on the maximum burst interval.
The addition of a maximum burst interval ($\delta_\mathrm{max}$) leads to several minor changes in the probability density functions of @orp18. The probability density distribution of the interval between the start of the observation and the first burst [Eq. 13 of @orp18] is given by $$\label{eq:wb_pdf_start}
\begin{aligned}
&\mathcal{P}(t_1,\delta_\mathrm{max}|k,r) = r \, \int_{t_1}^{\delta_\mathrm{max}} \mathcal{W}(\delta|k,r) \, \mathrm{d}\delta \\
&= r \, \left[\mathrm{CCDF}(t_1|k,r) - \mathrm{CCDF}(\delta_\mathrm{max}|k,r)\right],
\end{aligned}$$
where $\delta$ is the interval between the last unobserved burst and the first observed burst, and CCDF is the cumulative complementary distribution function, defined as $$\label{eq:wb_ccdf}
\mathrm{CCDF}(\delta|k,r) = \int_\delta^\infty \mathcal{W}(\delta^\prime|k,r) | The probability of measuring some set of burst arrival times $ t_\mathrm{1 }, t_\mathrm{2 }, \dots t_\mathrm{N}$ in a single notice of duration $ T$ can be separate into three parts:
1. The probability of the interval between the beginning of the notice and the first burst: $ P(t_1)$
2. The probabilities of the time interval between subsequent bursts in a single notice: $ P(t_2 \dots t_\mathrm{N }) = \prod_\mathrm{i=1}^\mathrm{N-1 } P(t_\mathrm{i+1 } - t_\mathrm{i})$
3. The probability of the time interval between the last burst and the conclusion of the observation: $ P(T - t_\mathrm{N})$
Assuming different observation are not correlated, period 1) and 3) describe minimum explosion intervals. To include that subsequent observations can be correlate, we include a maximum burst interval, which is merely the interval between the last burst of an observation and the arrival time of future observed burst. Only for the intervals before the first detected burst and after the last burst, there is no constraint on the maximum burst interval.
The addition of a maximum burst time interval ($ \delta_\mathrm{max}$) contribute to several minor change in the probability concentration function of @orp18. The probability density distribution of the interval between the start of the observation and the first burst [ Eq. 13 of @orp18 ] is given by $ $ \label{eq: wb_pdf_start }
\begin{aligned }
& \mathcal{P}(t_1,\delta_\mathrm{max}|k, radius) = r \, \int_{t_1}^{\delta_\mathrm{max } } \mathcal{W}(\delta|k, r) \, \mathrm{d}\delta \\
& = r \, \left[\mathrm{CCDF}(t_1|k, r) - \mathrm{CCDF}(\delta_\mathrm{max}|k, r)\right ],
\end{aligned}$$
where $ \delta$ is the interval between the last unobserved burst and the beginning observed burst, and CCDF is the cumulative complementary distribution affair, defined as $ $ \label{eq: wb_ccdf }
\mathrm{CCDF}(\delta|k, roentgen) = \int_\delta^\infty \mathcal{W}(\delta^\prime|k, r ) | The probability of measurinn some set of butsr arrital timss $t_\mathfm{1}, t_\mathrm{2}, \dots t_\mathrm{N}$ in e sibgle ibservation of duratiov $T$ can bv split ibto uhree parts:
1. The 'dobabillcy of bhe iutxrval between tme start of the observatimn aud the first burst: $P(t_1)$
2. The probabilieies of tje intervals bgtweem subavqment bursts in a single observatjon: $P(t_2 \vots t_\mathrm{N}) = \lrod_\mathrm{i=1}^\mathrm{N-1} P(t_\mathrl{i+1} - h_\mathrm{i})$
3. The probwbility of jge ybterval betwden the last burst and the end of the observation: $P(T - t_\matkrm{N})$
Assumint eifvgrent observetions are not corvvlated, [oints 1) and 3) describe miiimun burst intervals. To mnclude that subsequgnt observdtnons can be correlatee, qe inwluda a oqxioum bnrsf integvam, which is simply the interval between tne oast burst of an obfewvation and the arrival time of next obverbed burst. Only for the untervals before the virst detqcted burst and after the last burst, there is no wonstcaknt ok thd mwximum burst interval.
The addition of a maximui birxt interval ($\dejta_\mathrm{mac}$) pesqs to several minor chznges in the probahility qensiry functijns pf @orp18. The probability densuty distribunion of the interval bztween the scart og the observation and the fixst budst [Eq. 13 of @ogp18] is givsv by $$\label{eq:wb_pdw_stsrd}
\begin{aligned}
&\mathcal{P}(t_1,\qelta_\mathcm{max}|l,r) = r \, \knt_{t_1}^{\celta_\mwthrm{max}} \mwthcal{W}(\delta|k,r) \, \mathrm{d}\dflta \\
&= r \, \left[\lathrm{CCDF}(t_1|k,r) - \mathrm{CCDF}(\delta_\mevhrm{max}|k,r)\righj],
\eng{alpgned}$$
wherz $\delts$ is the intewval between tke last bnobsefved burst and thx first obsewved burst, ang CCDF is the cumulatyve xompoementafh distribution function, defined aw $$\label{eq:wb_ccdf}
\majhdm{CCDF}(\delta|k,r) = \nut_\eelta^\infty \matncau{W}(\dqlna^\pcime|k,w) | The probability of measuring some set of times t_\mathrm{2}, \dots in a single be into three parts: The probability of interval between the start of the and the first burst: $P(t_1)$ 2. The probabilities of the intervals between subsequent in a single observation: $P(t_2 \dots t_\mathrm{N}) = \prod_\mathrm{i=1}^\mathrm{N-1} P(t_\mathrm{i+1} - t_\mathrm{i})$ 3. probability the between last burst and the end of the observation: $P(T - t_\mathrm{N})$ Assuming different observations are not points 1) and 3) describe minimum burst intervals. include that subsequent observations be correlated, we include a burst which is the between last burst of observation and the arrival time of next observed burst. Only for the intervals before the first detected and after burst, there no on maximum burst interval. of a maximum burst interval ($\delta_\mathrm{max}$) minor changes in the probability density functions of The probability distribution of the interval between the of the observation and the first burst [Eq. of @orp18] is given by $$\label{eq:wb_pdf_start} \begin{aligned} &\mathcal{P}(t_1,\delta_\mathrm{max}|k,r) = r \, \int_{t_1}^{\delta_\mathrm{max}} \mathcal{W}(\delta|k,r) \, \mathrm{d}\delta r \, \left[\mathrm{CCDF}(t_1|k,r) - \end{aligned}$$ where $\delta$ the between last burst and first observed burst, and CCDF is the cumulative complementary distribution function, as $$\label{eq:wb_ccdf} \mathrm{CCDF}(\delta|k,r) = \int_\delta^\infty \mathcal{W}(\delta^\prime|k,r) | The probability of measuring Some set of bUrst aRriVal TiMes $t_\MathRm{1}, t_\mathrm{2}, \dots t_\MAthrM{N}$ in a single observation Of durAtIOn $T$ cAN bE spliT into thREe PARts:
1. thE pRobAbILiTy of tHe iNterval Between the StaRt Of the observaTIoN and the firSt bUrst: $P(t_1)$
2. The proBabIlitieS oF thE InterValS betwEen subSEquent Bursts in a SiNGle obsERvation: $p(T_2 \DoTs t_\mAthrm{N}) = \prod_\mathrm{i=1}^\MAtHRm{N-1} P(t_\mathrm{i+1} - t_\mAthrm{i})$
3. thE PrOBAbiLitY of the inteRvAl betWEen the lASt BURSt aND the end of the oBservation: $P(t - T_\maThrm{N})$
ASsUmiNG diffeRent oBsERvaTions are not CorrElated, poiNts 1) and 3) DEscribe MInimum bUrst inTerValS. To iNClUdE thAt SUbsEQuEnt OBseRvations CaN bE corrElatED, WE InclUde A maxImum bUrst interval, wHicH is sIMplY the iNtervAl beTwEen thE last bUrst oF aN observation and The aRrival timE of NeXt oBsErved BUrst. OnLy fOr tHe interVals befORe tHe FIRSt Detected burst and afTeR THe Last bursT, there IS nO cONstraint On The MaxiMUM bursT intERvAl.
The addItion oF A mAxImum burSt IntervAl ($\DelTa_\mAthrm{MAx}$) leAds to sEveral miNor chANges in the probaBIlity density fUNcTIOnS Of @orP18. ThE probabilitY denSIty dIstrIBuTioN Of the InterVaL BeTWeen the start of the obSeRvatioN and tHe first burst [EQ. 13 of @orp18] is giVEN By $$\label{eQ:wb_pDF_sTArt}
\begin{aligneD}
&\mathCal{P}(t_1,\delta_\MAthrm{max}|K,r) = r \, \inT_{t_1}^{\delta_\mAthrm{max}} \mATHcal{W}(\delTa|k,R) \, \maThrM{d}\dELTa \\
&= R \, \left[\mathrm{CCdf}(T_1|k,r) - \mAtHrm{CCDF}(\DelTa_\mathrM{maX}|k,r)\RigHt],
\eNd{Aligned}$$
whEre $\delta$ Is ThE iNtErvAl betWEen the laSt UnoBsErvEd burST and thE firsT obsErVeD BurSt, and CCdf iS THe cuMuLaTive ComPlEmentAry dIStrIbution Function, dEfiNEd as $$\LaBeL{eq:wb_ccDf}
\mathrm{CCDF}(\dElTa|k,r) = \int_\delTa^\InfTy \mathCAL{W}(\delta^\pRime|k,r) | The probability of measuri ng some se t ofbur star riva l ti mes $t_\mathrm { 1},t_\mathrm{2}, \dots t_ \math rm { N}$i na sin gle obs e rv a t ion o fdur at i on $T$can be spl it into th ree p arts:
1. T h eprobabilit y o f the interv albetwee nthe start of theobserv a tion a nd the fi rs t burst : $P(t_1 ) $
2. The probabilities of the intervalsbetwee ns ub s e que ntbursts inasingl e observ a ti o n : $P ( t_2 \dots t_\ mathrm{N})= \p rod_\m at hrm { i=1}^\ mathr m{ N -1} P(t_\mathr m{i+ 1} - t_\m athrm{ i })$
3. The pr obabil ity of the in te rva lb etw e en th e la st burst a nd theendo f t he o bse rvat ion:$P(T - t_\mat hrm {N}) $
A ssumi ng di ffer en t obs ervati ons a re not correlated , po ints 1) a nd3) de sc ribem inimum bu rst interv als. To inc lu d e th at subsequent obse rv a t io ns can b e corr e la te d , we inc lu dea ma x i mum b urst in terval,whichi ssi mply th einterv al be twe en th e las t burs t of anobser v ation and thea rrival time o f n e x to bser ved burst. Onl y fo r the int e rv als befor e the f i rs t detected burst and a fter t he la st burst, the re is no c o n s traint o n th e m a ximum burst in terva l.
The ad d ition of a ma ximum bu rst inter v a l ($\del ta_ \ma thr m{m a x }$ ) leads to se v e ralmi nor cha nge s in th e p rob abi lit ydensity f unctions o f@o rp 18. Thep robabili ty de ns ity dist r ibutio n ofthein te r val betwee n t h e sta rt o f th e o bs ervat iona ndthe fir st burst[Eq . 13of @ orp18]is given by $ $\ label{eq:w b_ pdf _start } \begin{a ligned}
&\mathcal{P } (t_1,\d elt a_\ma thrm {max}|k,r ) = r \,\in t _{t_1} ^{\del ta_\m at hrm { m ax}}\ m at hca l{ W}(\delta| k , r)\, \m at hrm{ d}\delt a \\
&= r \, \ l eft [\mathrm{CCDF }(t _1|k , r )- \ m at h rm{ CC D F}( \ d elta_\mathrm{ma x}|k,r)\ri gh t ],
\end{alig n ed} $$
where $\delt a$ is the int erval bet ween thela st u n o bse rved burst and the first ob s erved bu rst,and CCDFis th e cum ulativ e co mplem entary d istrib ution f unction, defined as $$\label{eq :wb_cc df}
\mathrm{C CDF } (\d elta|k,r) = \ int_\delta ^\i nft y \m ath c al{W} (\de l ta ^\p r ime|k ,r) | The probability_of measuring_some set of burst_arrival times_$t_\mathrm{1},_t_\mathrm{2}, \dots_t_\mathrm{N}$_in a single_observation of duration_$T$ can be split_into three parts:
1.__The probability of the interval between the start of the observation and the first_burst:_$P(t_1)$
2. _The_probabilities_of the intervals between subsequent_bursts in a single observation:_$P(t_2 \dots_t_\mathrm{N}) = \prod_\mathrm{i=1}^\mathrm{N-1} P(t_\mathrm{i+1} - t_\mathrm{i})$
3. The_probability_of the interval_between the last burst and the end of the_observation: $P(T - t_\mathrm{N})$
Assuming different observations_are not correlated,_points_1)_and 3) describe minimum_burst intervals. To include that subsequent_observations can be correlated, we include_a maximum burst interval, which is simply_the interval between the last burst_of an observation and the_arrival time_of next observed burst. Only_for the intervals_before the_first detected burst_and after the last burst, there_is no constraint_on the maximum burst interval.
The addition_of_a maximum burst_interval_($\delta_\mathrm{max}$)_leads to_several minor changes_in_the probability_density_functions of @orp18. The probability density_distribution_of the interval between the start of_the observation and the_first_burst [Eq. 13 of @orp18]_is given by $$\label{eq:wb_pdf_start}
\begin{aligned}
_ &\mathcal{P}(t_1,\delta_\mathrm{max}|k,r) = r \,_\int_{t_1}^{\delta_\mathrm{max}} \mathcal{W}(\delta|k,r)_\, \mathrm{d}\delta_\\
&= r \, \left[\mathrm{CCDF}(t_1|k,r) - \mathrm{CCDF}(\delta_\mathrm{max}|k,r)\right],
\end{aligned}$$
where $\delta$_is the interval between the last_unobserved burst and the_first observed_burst,_and CCDF is_the_cumulative complementary_distribution function, defined as $$\label{eq:wb_ccdf}
_ \mathrm{CCDF}(\delta|k,r)_= \int_\delta^\infty \mathcal{W}(\delta^\prime|k,r) |
D}}_0$ which makes the left diagram in the following display commutative: $$\xymatrix{
{\tilde{C}} \ar[r]^-{{\tilde{\eta}}_{{\tilde{C}}}} \ar[d]_-{{\tilde{\varphi}}} & {\tilde{G}}({\tilde{D}}) \ar[dl]^-{{\tilde{G}}({\tilde{\delta}})}
\\
{\tilde{G}}({\tilde{D}}_0) & {}
}
\qquad \qquad
\xymatrix{
C \ar[r]^-{\eta_C} \ar[d]_-{\eta_C} & G(D) \ar[dl]^-{G({U_{{\mathcal{D}}}}({\tilde{\delta}}))}
\\
G(D) & {}
}$$ The right diagram above is obtained from the left one by applying the functor ${U_{{\mathcal{C}}}}$. As the unit $\eta_C \colon C \to G(D)$ is a universal arrow from $C$ to $G$, that is, $(D,\eta_C)$ is the initial object in the comma category $(C \downarrow G)$, then the morphism ${U_{{\mathcal{D}}}}({\tilde{\delta}})$ must be the identity ${\operatorname{id}_{D}}$ on $D$. It follows from [Remark [\[rmk:action\]]{}]{}(b) that ${\tilde{\delta}}$ is an isomorphism.
Applications {#sec:app}
============
In this final section, we apply [Theorem [\[thm:main\]]{}]{} in some specific examples and thereby give a proof of Theorem A in the Introduction.
Let ${\mathcal{C}}=\mathsf{Met}$ be the category whose objects are all metric spaces and whose morphisms are all continuous functions. This category has finite products, indeed, the product of metric spaces $(M,d_M)$ and $(N,d_N)$ is $(M \times N,d_{M \times N})$ where $d_{M \times N}$ is given by $$d_{M \times N}((x_1,y_1),(x_2,y_2))=\max\{d_M(x_1,x_2),d_N(y_1,y_2)\}\,.$$ Let ${\mathcal{D}}=\mathsf{CompMet}$ be the full subcategory of $\mathsf{Met}$ consisting of | D}}_0 $ which makes the left diagram in the following display commutative: $ $ \xymatrix {
{ \tilde{C } } \ar[r]^-{{\tilde{\eta}}_{{\tilde{C } } } } \ar[d]_-{{\tilde{\varphi } } } & { \tilde{G}}({\tilde{D } }) \ar[dl]^-{{\tilde{G}}({\tilde{\delta } }) }
\\
{ \tilde{G}}({\tilde{D}}_0) & { }
}
\qquad \qquad
\xymatrix {
C \ar[r]^-{\eta_C } \ar[d]_-{\eta_C } & G(D) \ar[dl]^-{G({U_{{\mathcal{D}}}}({\tilde{\delta } }) ) }
\\
G(D) & { }
} $ $ The right diagram above is obtained from the leftover one by enforce the functor $ { U_{{\mathcal{C}}}}$. As the unit $ \eta_C \colon C \to G(D)$ is a cosmopolitan arrow from $ C$ to $ G$, that is, $ (D,\eta_C)$ is the initial object in the comma class $ (C \downarrow G)$, then the morphism $ { U_{{\mathcal{D}}}}({\tilde{\delta}})$ must be the identity $ { \operatorname{id}_{D}}$ on $ D$. It follows from [ Remark [ \[rmk: action\]]{}]{}(b) that $ { \tilde{\delta}}$ is an isomorphism.
Applications { # sec: app }
= = = = = = = = = = = =
In this concluding section, we apply [ Theorem [ \[thm: main\ ] ] { } ] { } in some specific examples and thereby move over a proof of Theorem A in the Introduction.
lease $ { \mathcal{C}}=\mathsf{Met}$ be the category whose objects are all measured spaces and whose morphisms are all continuous functions. This class has finite products, indeed, the merchandise of measured spaces $ (M, d_M)$ and $ (N, d_N)$ is $ (M \times N, d_{M \times N})$ where $ d_{M \times N}$ is given by $ $ d_{M \times N}((x_1,y_1),(x_2,y_2))=\max\{d_M(x_1,x_2),d_N(y_1,y_2)\}\,.$$ get $ { \mathcal{D}}=\mathsf{CompMet}$ be the full subcategory of $ \mathsf{Met}$ consisting of | D}}_0$ wjich makes the left diagvam in the follocung divplay dommutatkve: $$\xymatrix{
{\tilde{C}} \ar[r]^-{{\tipdw{\eta}}_{{\tulde{C}}}} \ar[d]_-{{\tilde{\varphi}}} & {\gilde{G}}({\tilfe{D}}) \ar[dl]^-{{\rildt{G}}({\tilde{\delta}})}
\\
{\tilde{N}}({\cilde{S}}_0) & {}
}
\qquad \qquad
\wymatrix{
C \ar[r]^-{\eta_C} \ar[d]_-{\atx_C} & G(D) \ar[dl]^-{G({U_{{\mathcal{D}}}}({\tilde{\delta}}))}
\\
G(D) & {}
}$$ Hhe right diagtam anjve js obtained from the left one by alplying the functor ${I_{{\mathcal{C}}}}$. As the unit $\eta_C \colln C \to G(D)$ is a unlversal arriw fwim $C$ to $G$, thxt is, $(D,\eta_B)$ is the injtial object in the comma categury $(C \downarrow G)$, thft the morphmsm ${U_{{\mwthcal{D}}}}({\tilde{\delta}})$ muvt be tne identity ${\opcratocnamw{id}_{D}}$ on $D$. It follows hrom [Remark [\[rmk:action\]]{}]{}(f) that ${\tindz{\delta}}$ is an isomorphusn.
Applhcathons {#wec:xpp}
============
Jn tgis fijal section, ws apply [Theirem [\[thm:main\]]{}]{} in some s[vvific examplss and trereby give a proof of Theorem A in the Pntrkduction.
Let ${\mathcal{C}}=\matysf{Met}$ be the categori whose obtects are all metric spaces and whose morphisms ase alm coublnuojw vunctions. This category has finite products, igsetd, nhe product of mebric spaces $(M,d_M)$ anc $(J,d_M)$ is $(M \times N,a_{M \timza H})$ where $d_{M \times N}$ is givgn by $$e_{M \times G}((x_1,y_1),(x_2,u_2))=\max\{d_M(x_1,x_2),d_N(y_1,y_2)\}\,.$$ Let ${\mathcal{D}}=\marhsf{CompMet}$ ye rhe full subcategoxy of $\mathsf{Oet}$ vonsixting of | D}}_0$ which makes the left diagram in display $$\xymatrix{ {\tilde{C}} \ar[d]_-{{\tilde{\varphi}}} & {\tilde{G}}({\tilde{D}}) } \qquad \xymatrix{ C \ar[d]_-{\eta_C} & G(D) \\ G(D) & {} }$$ The diagram above is obtained from the left one by applying the functor ${U_{{\mathcal{C}}}}$. the unit $\eta_C \colon C \to G(D)$ is a universal arrow from $C$ $G$, is, is initial object in the comma category $(C \downarrow G)$, then the morphism ${U_{{\mathcal{D}}}}({\tilde{\delta}})$ must be the ${\operatorname{id}_{D}}$ on $D$. It follows from [Remark [\[rmk:action\]]{}]{}(b) ${\tilde{\delta}}$ is an isomorphism. {#sec:app} ============ In this final we [Theorem [\[thm:main\]]{}]{} some examples thereby give a of Theorem A in the Introduction. Let ${\mathcal{C}}=\mathsf{Met}$ be the category whose objects are all metric spaces whose morphisms continuous functions. category finite indeed, the product spaces $(M,d_M)$ and $(N,d_N)$ is $(M N})$ where $d_{M \times N}$ is given by \times N}((x_1,y_1),(x_2,y_2))=\max\{d_M(x_1,x_2),d_N(y_1,y_2)\}\,.$$ ${\mathcal{D}}=\mathsf{CompMet}$ be the full subcategory of consisting of | D}}_0$ which makes the left diagram In the folloWing dIspLay CoMmutAtivE: $$\xymatrix{
{\tilde{c}} \Ar[r]^-{{\tIlde{\eta}}_{{\tilde{C}}}} \ar[d]_-{{\tilde{\vArphi}}} & {\TiLDe{G}}({\tILdE{D}}) \ar[dL]^-{{\tilde{G}}({\TIlDE{\DelTa}})}
\\
{\TiLde{g}}({\tILdE{D}}_0) & {}
}
\qquAd \qQuad
\xymAtrix{
C \ar[r]^-{\eTa_C} \Ar[D]_-{\eta_C} & G(D) \ar[dl]^-{G({u_{{\MaThcal{D}}}}({\tildE{\deLta}}))}
\\
G(D) & {}
}$$ The righT diAgram aBoVe iS ObtaiNed From tHe left ONe by apPlying the FuNCtor ${U_{{\mAThcal{C}}}}$. AS THe Unit $\Eta_C \colon C \to G(D)$ is a UNiVErsal arrow from $c$ to $G$, thAt IS, $(D,\ETA_C)$ iS thE initial obJeCt in tHE comma cATeGORY $(C \dOWnarrow G)$, then tHe morphism ${U_{{\MAthCal{D}}}}({\tiLdE{\deLTa}})$ must Be the IdENtiTy ${\operatornAme{iD}_{D}}$ on $D$. It foLlows fROm [RemarK [\[Rmk:actiOn\]]{}]{}(b) thaT ${\tiLde{\DeltA}}$ Is An IsoMoRPhiSM.
APplICatIons {#sec:aPp}
============
in This fInal SECTIon, wE apPly [THeoreM [\[thm:main\]]{}]{} in somE spEcifIC exAmpleS and tHereBy Give a Proof oF TheoReM A in the IntroducTion.
let ${\mathcaL{C}}=\mAtHsf{meT}$ be thE CategoRy wHosE objectS are all MEtrIc SPACeS and whose morphisms ArE ALl ContinuoUs funcTIoNs. tHis categOrY haS finITE prodUcts, INdEed, the prOduct oF MeTrIc spaceS $(M,D_M)$ and $(N,D_N)$ Is $(M \TimEs N,d_{M \TImes n})$ where $D_{M \times N}$ Is givEN by $$d_{M \times N}((x_1,y_1),(x_2,Y_2))=\Max\{d_M(x_1,x_2),d_N(y_1,y_2)\}\,.$$ LeT ${\MaTHCaL{d}}=\matHsf{compMet}$ be thE fulL SubcAtegORy Of $\mAThsf{MEt}$ conSiSTiNG of | D}}_0$ which makes the lef t diagramin th e f oll ow ingdisp lay commutativ e : $$ \xymatrix{
{\tilde {C}}\a r [r]^ - {{ \tild e{\eta} } _{ { \ til de {C }}} }\ ar [d]_- {{\ tilde{\ varphi}}}& { \t ilde{G}}({\t i ld e{D}}) \ar [dl ]^-{{\tilde{ G}} ({\til de {\d e lta}} )} \ \
{ \tilde {G}}({\ti ld e {D}}_0 ) & {}
}
\qq uad \qquad
\xym a tr i x{
C \ar[r ]^-{\e ta _ C} \ ar[ d]_ -{\eta_C}&G(D)\ ar[dl]^ - {G ( { U _{{ \ mathcal{D}}}} ({\tilde{\d e lta }}))} \ \
G (D) & { }
}$$ The rig ht d iagram ab ove is obtaine d from t he lef t o neby a p pl yi ngth e fu n ct or$ {U_ {{\mathc al {C }}}}$ . As t h e uni t $ \eta _C \c olon C \to G( D)$ isa un ivers al ar rowfr om $C $ to $ G$, t ha t is, $(D,\eta_ C)$is the in iti al ob je ct in the co mma ca tegory$(C \do w nar ro w G )$ , then the morphis m$ { U_ {{\mathc al{D}} } }( {\ t ilde{\de lt a}} )$ m u s t bethei de ntity ${ \opera t or na me{id}_ {D }}$ on $ D$. It foll o ws f rom [R emark [\ [rmk: a ction\]]{}]{}( b ) that ${\til d e{ \ d el t a}}$ is an isomorp hism .
Ap plic a ti ons {#sec :app}
= = == = =======
In this fi na l sect ion,we apply [The orem [\[th m : m ain\]]{} ]{}i ns ome specific e xampl es and the r eby give a pr oof of T heorem Ai n the Int rod uct ion .
L e t${\mathcal{C} } = \mat hs f{Met}$ be the ca teg ory wh ose o bjects ar e all me tr ic s pa ces andw hose mor ph ism sare allc ontinu ous f unct io ns . Th is cate g or y hasfi ni te p rod uc ts, i ndee d , t he prod uct of me tri c spa ce s$(M,d_M )$ and $(N,d_ N) $ is $(M \ ti mes N,d_{ M \times N })$ where $d_{M \timesN }$ is g ive n by$$d_ {M \times N} ((x_1, y_1 ) ,(x_2, y_2))= \max\ {d _M( x _ 1,x_2 ) , d_ N(y _1 ,y_2)\}\,. $ $ Le t ${\ ma thca l{D}}=\ mathsf{CompMet}$ b e th e full subcat ego ry o f $\ mat h sf { Met }$ con s i sting of | D}}_0$ which_makes the_left diagram in the_following display_commutative:_$$\xymatrix{
__ {\tilde{C}} \ar[r]^-{{\tilde{\eta}}_{{\tilde{C}}}}_\ar[d]_-{{\tilde{\varphi}}} & {\tilde{G}}({\tilde{D}})_\ar[dl]^-{{\tilde{G}}({\tilde{\delta}})}
_\\
__{\tilde{G}}({\tilde{D}}_0) & {}
}
\qquad \qquad
\xymatrix{
C_\ar[r]^-{\eta_C}_\ar[d]_-{\eta_C} &_G(D)_\ar[dl]^-{G({U_{{\mathcal{D}}}}({\tilde{\delta}}))}
_ \\
_ G(D) & {}
_ }$$_The right diagram above is obtained from the_left_one by applying_the functor ${U_{{\mathcal{C}}}}$. As the unit $\eta_C \colon C_\to G(D)$ is a universal arrow_from $C$ to_$G$,_that_is, $(D,\eta_C)$ is the_initial object in the comma category_$(C \downarrow G)$, then the morphism_${U_{{\mathcal{D}}}}({\tilde{\delta}})$ must be the identity ${\operatorname{id}_{D}}$ on_$D$. It follows from [Remark [\[rmk:action\]]{}]{}(b) that_${\tilde{\delta}}$ is an isomorphism.
Applications {#sec:app}
============
In_this final_section, we apply [Theorem [\[thm:main\]]{}]{} in_some specific examples_and thereby_give a proof_of Theorem A in the Introduction.
Let ${\mathcal{C}}=\mathsf{Met}$_be the category_whose objects are all metric spaces_and_whose morphisms are_all_continuous_functions. This_category has finite_products,_indeed, the_product_of metric spaces $(M,d_M)$ and $(N,d_N)$_is_$(M \times N,d_{M \times N})$ where $d_{M_\times N}$ is given_by_$$d_{M \times N}((x_1,y_1),(x_2,y_2))=\max\{d_M(x_1,x_2),d_N(y_1,y_2)\}\,.$$ Let_${\mathcal{D}}=\mathsf{CompMet}$ be the full subcategory_of $\mathsf{Met}$ consisting of |
exists an $a\in A{\backslash{\{0\}}}$ such that $ae_3, ae_2\in A{\backslash{\{0\}}}$. So we get $f(a_3e_3)=a_3e_3\neq 0$ and yet $$f(a_3e_3)=f(a_3(1-e_1)e_2) = f(a_3e_2)f(1-e_1) = a_3e_2(1-e_2)=0$$ a contradiction. If the former case holds, then we get a contradiction in a similar manner.
This has been discussed throughout our study of regular rings, but we shall give a formal prove for all reduced rings here. The keyword is *Storrer’s Satz* (see [@Capco3] Theorem 1). By Storrer’s Satz we have the following commutative diagram consisting of canonical maps (except for the lower horizontal map which is due to Storrer’s Satz). $$\begindc{\commdiag}[1]
\obj(0,50)[A]{$A$}
\obj(80,50)[B]{$C$}
\obj(0,0)[C]{$Q(A)$}
\obj(80,0)[D]{$Q(C)$}
\mor{A}{B}{}[1,3]
\mor{B}{D}{}[1,3]
\mor{A}{C}{}[1,3]
\mor{C}{D}{}[1,3]
\enddc$$ $C$ being real closed $*$ must contain all the idempotents of $Q(C)$, thus it contains all the idempotents of $Q(A)$ and has $A$ as a subring. This implies (by [@Mewborn] Proposition 2.5) that $C$ has $B(A)$ as a subring. That $C$ is an integral and essential extension of $B(A)$ is then clear. $C$ being real closed $*$, means that $C$ is a real closure $*$ of $B(A)$ as well.
\[redring-rcrs\] Let $A$ be a reduced poring, then there is a bijection between the following sets
- $\mathcal A := | exists an $ a\in A{\backslash{\{0\}}}$ such that $ ae_3, ae_2\in A{\backslash{\{0\}}}$. So we get $ f(a_3e_3)=a_3e_3\neq 0 $ and yet $ $ f(a_3e_3)=f(a_3(1 - e_1)e_2) = f(a_3e_2)f(1 - e_1) = a_3e_2(1 - e_2)=0$$ a contradiction. If the erstwhile subject holds, then we get a contradiction in a similar manner.
This has been hash out throughout our study of regular gang, but we shall give a formal prove for all boil down rings here. The keyword is * Storrer ’s Satz * (see [ @Capco3 ] Theorem 1). By Storrer ’s Satz we get the following commutative diagram consisting of canonical maps (except for the lower horizontal map which is ascribable to Storrer ’s Satz). $ $ \begindc{\commdiag}[1 ]
\obj(0,50)[A]{$A$ }
\obj(80,50)[B]{$C$ }
\obj(0,0)[C]{$Q(A)$ }
\obj(80,0)[D]{$Q(C)$ }
\mor{A}{B}{}[1,3 ]
\mor{B}{D}{}[1,3 ]
\mor{A}{C}{}[1,3 ]
\mor{C}{D}{}[1,3 ]
\enddc$$ $ C$ being real closed $ * $ must contain all the idempotents of $ Q(C)$, therefore it contains all the idempotents of $ Q(A)$ and has $ A$ as a subring. This implies (by [ @Mewborn ] Proposition 2.5) that $ C$ has $ B(A)$ as a subring. That $ C$ is an integral and all-important extension of $ B(A)$ is then clear. $ C$ being real closed $ * $, means that $ C$ is a veridical closure $ * $ of $ B(A)$ as well.
\[redring - rcrs\ ] Let $ A$ be a reduced poring, then there is a bijection between the following sets
- $ \mathcal A: = | exlsts an $a\in A{\backslash{\{0\}}}$ smch that $ae_3, ae_2\in A{\backsnash{\{0\}}}$. Sk we get $f(a_3e_3)=a_3e_3\neq 0$ and yet $$f(a_3e_3)=f(a_3(1-e_1)e_2) = h(a_3e_2)f(1-w_1) = a_3e_2(1-t_2)=0$$ a contradiction. Iw the forler case holvs, then we get a contradletion ln a vmmilar manner.
Thls has been discussed thrmuehlut our study of regular rings, but re shalk hive a formal krove sor zll reduced rings here. The keyword is *Stogrer’s Satz* (see [@Calco3] Theorem 1). By Storrer’s Swtz ae have the followlng commutajjve eiagram consksting of bcnonical maks (except for the lower horizontxl ma' which is euw tl Storrer’s Setz). $$\bedindc{\commdian}[1]
\pbj(0,50)[A]{$A$}
\otj(80,50)[B]{$C$}
\obj(0,0)[V]{$Q(A)$}
\obj(80,0)[D]{$Q(C)$}
\mor{A}{B}{}[1,3]
\kor{U}{D}{}[1,3]
\moe{A}{C}{}[1,3]
\mor{C}{D}{}[1,3]
\enddc$$ $C$ being real closed $*$ must contain dlm the idempotents od $Q(C)$, jhus ht cubtakns akl the ifem'otents of $S(A)$ and has $Q$ as a subring. This iiikies (by [@Mewbkrn] Prjpjsition 2.5) that $C$ has $B(A)$ as a subring. Thau $C$ ia an integral and essenrial extension of $B(A)$ ls then cjear. $C$ being real closed $*$, means that $C$ is a real wlosuce $*$ oy $B(A)$ ar wfll.
\[redring-rcrs\] Let $A$ be a reduced poring, then fhtre is a bijectiok between the follpwlnb sets
- $\mathcxl A := | exists an $a\in A{\backslash{\{0\}}}$ such that $ae_3, So get $f(a_3e_3)=a_3e_3\neq and yet $$f(a_3e_3)=f(a_3(1-e_1)e_2) contradiction. the former case then we get contradiction in a similar manner. This been discussed throughout our study of regular rings, but we shall give a prove for all reduced rings here. The keyword is *Storrer’s Satz* (see [@Capco3] 1). Storrer’s we the following commutative diagram consisting of canonical maps (except for the lower horizontal map which is to Storrer’s Satz). $$\begindc{\commdiag}[1] \obj(0,50)[A]{$A$} \obj(80,50)[B]{$C$} \obj(0,0)[C]{$Q(A)$} \obj(80,0)[D]{$Q(C)$} \mor{B}{D}{}[1,3] \mor{A}{C}{}[1,3] \mor{C}{D}{}[1,3] \enddc$$ being real closed $*$ must all idempotents of thus contains the idempotents of and has $A$ as a subring. This implies (by [@Mewborn] Proposition 2.5) that $C$ has $B(A)$ as subring. That an integral essential of is then clear. real closed $*$, means that $C$ closure $*$ of $B(A)$ as well. \[redring-rcrs\] Let be a poring, then there is a bijection the following sets - $\mathcal A := | exists an $a\in A{\backslash{\{0\}}}$ such That $ae_3, ae_2\in a{\backSlaSh{\{0\}}}$. SO wE get $F(a_3e_3)=a_3E_3\neq 0$ and yet $$f(a_3e_3)=f(A_3(1-E_1)e_2) = f(a_3E_2)f(1-e_1) = a_3e_2(1-e_2)=0$$ a contradiction. If The foRmER casE HoLds, thEn we get A CoNTRadIcTiOn iN a SImIlar mAnnEr.
This hAs been discUssEd Throughout ouR StUdy of regulAr rIngs, but we shaLl gIve a foRmAl pROve foR alL reduCed rinGS here. THe keyword Is *sTorrer’S satz* (see [@cAPcO3] TheOrem 1). By Storrer’s SatZ We HAve the followinG commuTaTIvE DIagRam Consisting Of CanonICal maps (EXcEPT For THe lower horizoNtal map whicH Is dUe to StOrRer’S satz). $$\beGindc{\CoMMdiAg}[1]
\obj(0,50)[A]{$A$}
\obj(80,50)[B]{$c$}
\obj(0,0)[c]{$Q(A)$}
\obj(80,0)[D]{$Q(C)$}
\Mor{A}{B}{}[1,3]
\mOR{B}{D}{}[1,3]
\mor{A}{c}{}[1,3]
\Mor{C}{D}{}[1,3]
\enDdc$$ $C$ beIng ReaL cloSEd $*$ MuSt cOnTAin ALl The IDemPotents oF $Q(c)$, tHus it ContAINS All tHe iDempOtentS of $Q(A)$ and has $A$ aS a sUbriNG. ThIs impLies (bY [@MewBoRn] ProPositiOn 2.5) thaT $C$ Has $B(A)$ as a subring. that $c$ is an inteGraL aNd eSsEntiaL ExtensIon Of $B(a)$ is then Clear. $C$ bEIng ReAL CLoSed $*$, means that $C$ is a reAl CLOsUre $*$ of $B(A)$ aS well.
\[rEDrInG-Rcrs\] Let $A$ Be A reDuceD POring, Then THeRe is a bijEction BEtWeEn the foLlOwing sEtS
- $\maThcAl A := | exists an $a\in A{\backsl ash{\{0\}} }$ su chtha t$ae_ 3, a e_2\in A{\back s lash {\{0\}}}$. So we get $ f(a_3 e_ 3 )=a_ 3 e_ 3\neq 0$ and ye t $$f (a _3 e_3 )= f (a _3(1- e_1 )e_2) = f(a_3e_2) f(1 -e _1) = a_3e_2 ( 1- e_2)=0$$ a co ntradiction. If the f or mer casehol ds, t hen we get acontradic ti o n in a similar m an ner.
This has been d i sc u ssed throughou t ourst u dy o f r egu lar rings, b ut we shall g i ve a for m al prove forall reduced rin gs her e. Th e keywo rd is * S tor rer’s Satz* (se e [@Capco 3] The o rem 1). By Stor rer’sSat z w e ha v eth e f ol l owi n gcom m uta tive dia gr am cons isti n g o f ca non ical maps (except forthe low e r h orizo ntalmapwh ich i s dueto St or rer’s Satz). $$ \beg indc{\com mdi ag }[1 ]\obj( 0 ,50)[A ]{$ A$}
\obj(8 0,50)[B ] {$C $} \ o bj (0,0)[C]{$Q(A)$}
\ ob j ( 80 ,0)[D]{$ Q(C)$} \m or { A}{B}{}[ 1, 3]\mor { B }{D}{ }[1, 3 ]\mor{A}{ C}{}[1 , 3]
\ mor{C}{ D} {}[1,3 ]\en ddc $$ $C $ bei ng rea l closed $*$m ust contain al l the idempote n ts o f$ Q(C) $,thus it con tain s all the id emp o tents of $ Q( A )$ and has $A$ as a su br ing. T his i mplies (by [@ Mewborn] P r o p osition2.5) th a t $C$ has $B(A )$ as a subring . That $C $ isan integ ral and e s s ential e xte nsi onof$ B (A )$ is then cl e a r. $ C$ beingrea l close d $ *$, me ans t hat $C$ i s a real c lo su re $* $ of$ B(A)$ as w ell .
\[ redri n g-rcrs \] Le t $A $be a r educedp or i n g, t he nther e i sa bij ecti o n b etweenthe follo win g set s
- $\ma thcal A := | exists_an $a\in_A{\backslash{\{0\}}}$ such that $ae_3,_ae_2\in A{\backslash{\{0\}}}$._So_we get_$f(a_3e_3)=a_3e_3\neq_0$ and yet_$$f(a_3e_3)=f(a_3(1-e_1)e_2) = f(a_3e_2)f(1-e_1)_= a_3e_2(1-e_2)=0$$ a contradiction._If the former_case_holds, then we get a contradiction in a similar manner.
This has been discussed throughout_our_study of_regular_rings,_but we shall give a_formal prove for all reduced_rings here._The keyword is *Storrer’s Satz* (see [@Capco3] Theorem_1)._By Storrer’s Satz_we have the following commutative diagram consisting of canonical_maps (except for the lower horizontal_map which is_due_to_Storrer’s Satz). $$\begindc{\commdiag}[1]
\obj(0,50)[A]{$A$}
\obj(80,50)[B]{$C$}
\obj(0,0)[C]{$Q(A)$}
\obj(80,0)[D]{$Q(C)$}
\mor{A}{B}{}[1,3]
\mor{B}{D}{}[1,3]
\mor{A}{C}{}[1,3]
\mor{C}{D}{}[1,3]
\enddc$$ $C$_being real closed $*$ must contain_all the idempotents of $Q(C)$, thus_it contains all the idempotents of $Q(A)$_and has $A$ as a subring._This implies (by [@Mewborn] Proposition_2.5) that_$C$ has $B(A)$ as a_subring. That $C$_is an_integral and essential_extension of $B(A)$ is then clear._$C$ being real_closed $*$, means that $C$ is_a_real closure $*$_of_$B(A)$_as well.
\[redring-rcrs\]_Let $A$ be_a_reduced poring,_then_there is a bijection between the_following_sets
- $\mathcal A := |
For instance, we perform entanglement swapping protocol for $n$ times to connect the entanglement between the remote locations A and K, we will get $$\begin{aligned}
|\Psi^{n+1}\rangle_{ak}=(\alpha^{n+1}a^{\dagger}\pm\beta^{n+1}k^{\dagger}e^{i\theta_{AK}})|0\rangle\label{state8}.\end{aligned}$$ For $\alpha\neq\beta$, the entanglement decreases more and more, and we will fail to establish a perfect long-distance entanglement channel for quantum communication. In other words, a long-distance quantum communication with a practical manipulation of entanglement generation and a practical transmission requires the entanglement concentration of single-photon entangled states.
entanglement concentration of single-photon entanglements
=========================================================
Cross-Kerr nonlinearity has been wildly studied [@QND1; @QND2; @discriminator; @qubit1; @qubit2; @qubit3], such as the construction of a CNOT gate, the discrimination of unknown qubits, Bell-state analysis, and so on. The Hamiltonian of the cross-Kerr nonlinearity is $H_{ck}=\hbar\chi a^{+}_{s}a_{s}a^{+}_{p}a_{p}$ [@QND1; @QND2; @kerr1; @kerr2], where $a^+_s$ and $a^+_p$ are the creation operations and $a_s$ and $a_p$ are the destruction operations. When the coherent beam $|\alpha\rangle_{p}$ and a signal pulse in the Fock state $|\Psi\rangle_s=c_{0}|0\rangle_{s}+c_{1}|1\rangle_{s}$ interact with the cross-Kerr nonlinearity, the whole system becomes: $$\begin{aligned}
U_{ck}|\Psi\rangle_{s}|\alpha\rangle_{p}&=&
e^{iH_{ck}t/\hbar}[c_{0}|0\rangle_{s}+c_{1}
|1\rangle_{s}]|\alpha\rangle_{p} \nonumber\\
&=& c_{0}|0\rangle_{s}|\alpha\rangle_{p}+c_{1}|1\rangle_{s}|
\alpha e^{i\theta}\rangle_{p},\end{aligned}$$ where $\theta=\chi t$ and $t | For instance, we perform entanglement swapping protocol for $ n$ time to plug in the entanglement between the remote locations A and K, we will receive $ $ \begin{aligned }
|\Psi^{n+1}\rangle_{ak}=(\alpha^{n+1}a^{\dagger}\pm\beta^{n+1}k^{\dagger}e^{i\theta_{AK}})|0\rangle\label{state8}.\end{aligned}$$ For $ \alpha\neq\beta$, the entanglement decreases more and more, and we will fail to build a arrant long - distance entanglement channel for quantum communication. In other words, a farseeing - distance quantum communication with a practical manipulation of web generation and a practical transmission requires the entanglement concentration of single - photon embroiled states.
entanglement concentration of unmarried - photon entanglements
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Cross - Kerr nonlinearity has been wildly study [ @QND1; @QND2; @discriminator; @qubit1; @qubit2; @qubit3 ], such as the construction of a CNOT gate, the discrimination of nameless qubits, Bell - state analysis, and so on. The Hamiltonian of the cross - Kerr nonlinearity is $ H_{ck}=\hbar\chi a^{+}_{s}a_{s}a^{+}_{p}a_{p}$ [ @QND1; @QND2; @kerr1; @kerr2 ], where $ a^+_s$ and $ a^+_p$ are the initiation operations and $ a_s$ and $ a_p$ are the destruction operations. When the coherent beam $ |\alpha\rangle_{p}$ and a signal pulse in the Fock state $ |\Psi\rangle_s = c_{0}|0\rangle_{s}+c_{1}|1\rangle_{s}$ interact with the cross - Kerr nonlinearity, the whole system becomes: $ $ \begin{aligned }
U_{ck}|\Psi\rangle_{s}|\alpha\rangle_{p}&= &
e^{iH_{ck}t/\hbar}[c_{0}|0\rangle_{s}+c_{1 }
|1\rangle_{s}]|\alpha\rangle_{p } \nonumber\\
& = & c_{0}|0\rangle_{s}|\alpha\rangle_{p}+c_{1}|1\rangle_{s}|
\alpha e^{i\theta}\rangle_{p},\end{aligned}$$ where $ \theta=\chi t$ and $ t | Fog instance, we perform enuanglement swappiut protmcol fkr $n$ timds to connect the entanglemeit bwtweeb the remote locations A and K, ae will tet $$\uegin{aligned}
|\Psi^{n+1}\czngle_{ak}=(\alpha^{n+1}z^{\faggzr}\'m\beta^{n+1}k^{\dagger}e^{l\theta_{AK}})|0\rancle\label{state8}.\etd{xlngned}$$ For $\alpha\neq\beta$, the entanglemqnt dectewses more and iore, wnd sv cill fail to establish a perfecf long-dpstance entanglemrnt channel for quantum colmunlcation. In other wlrds, a long-eistwbce quantum zommunicatpmn with a kractical manipulation of entanguemenc generatiob qnd d practical tranfmission reqmpres tha entanblement concenbratimn if single-photon entanjled states.
entanglemgnt concendrction of single-photon ebtangnemetts
=========================================================
Cfiss-Yerd ionminearlty has been sildly studued [@QND1; @QND2; @discrimonwnpr; @qubit1; @qubjt2; @qubyt3], such as the construction of a CNOT gatt, the discrimination of unkniwn qubits, Bell-state wnalysis, wnd so on. The Hamiltonian of the cross-Kerr nonlinaaritb ks $K_{gh}=\fvag\chi a^{+}_{s}a_{s}a^{+}_{p}a_{p}$ [@QND1; @QND2; @kerr1; @kerr2], where $a^+_s$ and $w^+_l$ srv the creation opcrations and $a_s$ anc $w_p$ wre the destroction opedations. When the clherent beam $|\alpha\randle_{p}$ and a signal pulse in the Dock state $|\Pfu\rangle_s=c_{0}|0\rangle_{s}+c_{1}|1\rcngle_{s}$ intercct wijh the cross-Kerr nonlinearity, the whole systfm becomea: $$\begin{aligned}
U_{ck}|\Osi\gangne_{s}|\alpha\rangle_{p}&=&
e^{iH_{ck}t/\hbar}[c_{0}|0\wangle_{s}+c_{1}
|1\rengle_{x}]|\alpha\rxnglg_{p} \nonuiber\\
&=& c_{0}|0\rangpe_{s}|\aliva\rangle_{p}+c_{1}|1\rangle_{s}|
\appha g^{i\thetd}\rangle_{p},\enf{aligned}$$ where $\theta=\chi t$ and $t | For instance, we perform entanglement swapping protocol times connect the between the remote will $$\begin{aligned} |\Psi^{n+1}\rangle_{ak}=(\alpha^{n+1}a^{\dagger}\pm\beta^{n+1}k^{\dagger}e^{i\theta_{AK}})|0\rangle\label{state8}.\end{aligned}$$ For the entanglement decreases and more, and we will fail establish a perfect long-distance entanglement channel for quantum communication. In other words, a quantum communication with a practical manipulation of entanglement generation and a practical transmission the concentration single-photon states. entanglement concentration of single-photon entanglements ========================================================= Cross-Kerr nonlinearity has been wildly studied [@QND1; @QND2; @discriminator; @qubit2; @qubit3], such as the construction of a gate, the discrimination of qubits, Bell-state analysis, and so The of the nonlinearity $H_{ck}=\hbar\chi [@QND1; @QND2; @kerr1; where $a^+_s$ and $a^+_p$ are the creation operations and $a_s$ and $a_p$ are the destruction operations. When coherent beam a signal in Fock $|\Psi\rangle_s=c_{0}|0\rangle_{s}+c_{1}|1\rangle_{s}$ interact with nonlinearity, the whole system becomes: $$\begin{aligned} \nonumber\\ &=& c_{0}|0\rangle_{s}|\alpha\rangle_{p}+c_{1}|1\rangle_{s}| \alpha e^{i\theta}\rangle_{p},\end{aligned}$$ where $\theta=\chi t$ $t | For instance, we perform entanGlement swaPping ProTocOl For $n$ TimeS to connect the eNTangLement between the remote LocatIoNS A anD k, wE will Get $$\begiN{AlIGNed}
|\psI^{n+1}\RanGlE_{Ak}=(\Alpha^{N+1}a^{\dAgger}\pm\Beta^{n+1}k^{\daggEr}e^{I\tHeta_{AK}})|0\rangle\LAbEl{state8}.\end{AliGned}$$ For $\alpha\Neq\Beta$, thE eNtaNGlemeNt dEcreaSes morE And morE, and we wilL fAIl to esTAblish a PERfEct lOng-distance entangLEmENt channel for quAntum cOmMUnICAtiOn. IN other wordS, a Long-dIStance qUAnTUM ComMUnication with A practical mANipUlatioN oF enTAnglemEnt geNeRAtiOn and a practIcal TransmissIon reqUIres the ENtangleMent coNceNtrAtioN Of SiNglE-pHOtoN EnTanGLed States.
enTaNgLemenT conCENTRatiOn oF sinGle-phOton entanglemEntS
=========================================================
CroSS-KeRr nonLineaRity HaS been Wildly StudiEd [@qND1; @QND2; @discriminAtor; @Qubit1; @qubiT2; @quBiT3], suCh As the COnstruCtiOn oF a CNOT gAte, the dIScrImINATiOn of unknown qubits, BElL-STaTe analysIs, and sO On. thE hamiltonIaN of The cROSs-KerR nonLInEarity is $h_{ck}=\hbaR\ChI a^{+}_{S}a_{s}a^{+}_{p}a_{p}$ [@qNd1; @QND2; @keRr1; @KerR2], whEre $a^+_s$ ANd $a^+_p$ Are the Creation OperaTIons and $a_s$ and $a_p$ ARe the destructIOn OPErATionS. WhEn the cohereNt beAM $|\alpHa\raNGlE_{p}$ aND a sigNal puLsE In THe Fock state $|\Psi\ranglE_s=C_{0}|0\ranglE_{s}+c_{1}|1\raNgle_{s}$ interact With the croSS-kErr nonliNearITy, THe whole system bEcomeS: $$\begin{aligNEd}
U_{ck}|\Psi\RanglE_{s}|\alpha\rAngle_{p}&=&
e^{iH_{CK}T/\hbar}[c_{0}|0\raNglE_{s}+c_{1}
|1\RanGle_{S}]|\ALpHa\rangle_{p} \nonuMBEr\\
&=& c_{0}|0\rAnGle_{s}|\alpHa\rAngle_{p}+c_{1}|1\RanGle_{S}|
\alPha E^{i\Theta}\rangLe_{p},\end{alIgNeD}$$ wHeRe $\tHeta=\cHI t$ and $t | For instance, we performentangleme nt sw app ing p roto colfor $n$ timest o co nnect the entanglement betw ee n the re motelocatio n sA and K ,wewi l lget $ $\b egin{al igned}
|\P si^ {n +1}\rangle_{ a k} =(\alpha^{ n+1 }a^{\dagger} \pm \beta^ {n +1} k ^{\da gge r}e^{ i\thet a _{AK}} )|0\rangl e\ l abel{s t ate8}.\ e n d{ alig ned}$$ For $\alph a \n e q\beta$, the e ntangl em e nt d ecr eas es more an dmore, and wew il l f ail to establisha perfect l o ng- distan ce en t anglem ent c ha n nel for quantu m co mmunicati on. In other w o rds, along-d ist anc e qu a nt um co mm u nic a ti onw ith a pract ic al mani pula t i o n ofent angl ement generation a nda pr a cti cal t ransm issi on requ ires t he en ta nglement concen trat ion of si ngl e- pho to n ent a ngledsta tes .
enta nglemen t co nc e n t ra tion of single-pho to n en tangleme nts
== = == == = ======== == === ==== = = ===== ==== = == ======== ====== = == =
Cross- Ke rr non li nea rit y has been wildl y studie d [@Q N D1; @QND2; @di s criminator; @ q ub i t 1; @qub it2 ; @qubit3], suc h asthec on str u ction of a C N OT gate, the discrimin at ion of unkn own qubits, B ell-statea n a lysis, a nd s o o n . The Hamilton ian o f the cros s -Kerr no nline arity is $H_{ck}= \ h bar\chia^{ +}_ {s} a_{ s } a^ {+}_{p}a_{p}$ [ @QND 1; @QND2; @k err1; @ ker r2] , w her e$a^+_s$ a nd $a^+_ p$ a re t hecreat i on opera ti ons a nd$a_s$ and $a _p$ a re t he d e str uctiono pe r a tion s. W henthe c ohere nt b e am$|\alph a\rangle_ {p} $ and a s ignal p ulse in the F oc k state $| \P si\ rangle _ s =c_{0}|0 \rangle_{s}+c_{1}|1\ran g le_{s}$ in terac t wi th the cr oss -Kerrnon l ineari ty, th e who le sy s t em be c o me s:$$ \begin{ali g n ed}
U_{c k} |\Ps i\rangl e_{s}|\alpha\rangl e _{p }&=&
e^{iH_{c k}t /\hb a r }[ c_{ 0 }| 0 \ra ng l e_{ s } +c_{1}
|1\rangl e_{s}]|\al ph a \r angle_{p}\ non um ber\\
& =& c_{0 }|0\r a ngle_{s }|\alpha\ rangle_{p }+ c_{1 } | 1\r angle_{s}|
\alphae^{i\thet a }\ran g le _{p}, \en d{alig ne d}$ $ whe re $\t h eta =\chi t$ an d$t | For_instance, we_perform entanglement swapping protocol_for $n$_times_to connect_the_entanglement between the_remote locations A_and K, we will_get $$\begin{aligned}
|\Psi^{n+1}\rangle_{ak}=(\alpha^{n+1}a^{\dagger}\pm\beta^{n+1}k^{\dagger}e^{i\theta_{AK}})|0\rangle\label{state8}.\end{aligned}$$ For_$\alpha\neq\beta$,_the entanglement decreases more and more, and we will fail to establish a perfect_long-distance_entanglement channel_for_quantum_communication. In other words, a_long-distance quantum communication with a_practical manipulation_of entanglement generation and a practical transmission requires_the_entanglement concentration of_single-photon entangled states.
entanglement concentration of single-photon entanglements
=========================================================
Cross-Kerr nonlinearity has_been wildly studied [@QND1; @QND2; @discriminator;_@qubit1; @qubit2; @qubit3],_such_as_the construction of a_CNOT gate, the discrimination of unknown_qubits, Bell-state analysis, and so on._The Hamiltonian of the cross-Kerr nonlinearity is_$H_{ck}=\hbar\chi a^{+}_{s}a_{s}a^{+}_{p}a_{p}$ [@QND1; @QND2; @kerr1; @kerr2],_where $a^+_s$ and $a^+_p$ are_the creation_operations and $a_s$ and $a_p$_are the destruction_operations. When_the coherent beam_$|\alpha\rangle_{p}$ and a signal pulse in_the Fock state_$|\Psi\rangle_s=c_{0}|0\rangle_{s}+c_{1}|1\rangle_{s}$ interact with the cross-Kerr nonlinearity,_the_whole system becomes:_$$\begin{aligned}
U_{ck}|\Psi\rangle_{s}|\alpha\rangle_{p}&=&
e^{iH_{ck}t/\hbar}[c_{0}|0\rangle_{s}+c_{1}
|1\rangle_{s}]|\alpha\rangle_{p}_\nonumber\\
&=&_c_{0}|0\rangle_{s}|\alpha\rangle_{p}+c_{1}|1\rangle_{s}|
\alpha e^{i\theta}\rangle_{p},\end{aligned}$$_where $\theta=\chi t$_and_$t |
s \;
[\partial_s^2 f(s) ]^2.\end{aligned}$$ Which verifies that the energy is constant in a model-independent manner. Furthermore if $f(s)$ is smooth and satisfies suitable falloff conditions at $s\to\pm\infty$ then the wavelet will be nonsingular and of finite energy.
To obtain a specific example it only remains to finish the complete specification of the potential $\psi(r,t)$. One particularly simple choice is to take one of the real scalar wavelets of the previous section $\psi\to\phi_{1,2}$, with $\zeta^a = ||\zeta|| \; V^a = a\;
V^a$. Since the electric and magnetic fields are now specified in terms of derivatives of a smooth bounded function, the electromagnetic field is similarly smooth and bounded. For the energy, write it in the form $${\mathcal E} =
{1\over3} \oint \d^3 x \left\{
\psi \; \partial_t^4 \psi - \partial_t\psi \; \partial_t^3 \psi
\right\}.$$ Whence, for either $\psi\to\phi_{1,2}$, an integration carried out at arbitrary $t$ yields the time-independent quantity $${\mathcal E}_{1,2} = {1\over2} \;{\phi_0^2\over a}.$$ The calculation has for convenience been carried out in the zero-momentum frame of the wavelet. In a general Lorentz frame we would have $$P^a_{1,2} = {1\over2}\; {\phi_0^2\over a} \;\; V^a
= {1\over2}\; {\phi_0^2\over a^2} \;\; \zeta^a.$$ Thus this field configuration, as for the scalar case, is Lorentz covariant, bounded, finite energy and zero action. The vector nature of the Maxwell field has added technical complications, but there is no real change in basic principles.
In closing this section, I should mention one other wavelet that is particularly attractive. If one makes use of the pseudo-differential operator $(\nabla^2)^{-1/2}$ one could write $$\psi = (\nabla^2)^{-1/2} \; \phi_{1,2} =
{1\over\Gamma(-1/2)} \int | s \;
[ \partial_s^2 f(s) ] ^2.\end{aligned}$$ Which verifies that the energy is constant in a model - autonomous manner. Furthermore if $ f(s)$ is legato and satisfies suitable falloff condition at $ s\to\pm\infty$ then the ripple will be nonsingular and of finite energy.
To obtain a specific example it merely remains to finish the accomplished specification of the potential $ \psi(r, t)$. One peculiarly simple choice is to take one of the real scalar ripple of the previous section $ \psi\to\phi_{1,2}$, with $ \zeta^a = ||\zeta|| \; V^a = a\;
V^a$. Since the electric and charismatic fields are nowadays stipulate in terms of derivatives of a smooth bounce function, the electromagnetic field is similarly smooth and bounded. For the department of energy, write it in the shape $ $ { \mathcal E } =
{ 1\over3 } \oint \d^3 x \left\ {
\psi \; \partial_t^4 \psi - \partial_t\psi \; \partial_t^3 \psi
\right\}.$$ Whence, for either $ \psi\to\phi_{1,2}$, an integration carried out at arbitrary $ t$ yield the time - independent quantity $ $ { \mathcal E}_{1,2 } = { 1\over2 } \;{\phi_0 ^ 2\over a}.$$ The calculation has for convenience been carried out in the zero - momentum frame of the wavelet. In a general Lorentz frame we would have $ $ P^a_{1,2 } = { 1\over2}\; { \phi_0 ^ 2\over a } \;\; V^a
= { 1\over2}\; { \phi_0 ^ 2\over a^2 } \;\; \zeta^a.$$ Thus this field configuration, as for the scalar case, is Lorentz covariant, bounded, finite energy and zero action. The vector nature of the Maxwell battlefield has added technical complications, but there be no veridical change in basic principles.
In closing this incision, I should mention one other wavelet that is particularly attractive. If one lay down use of the pseudo - differential operator $ (\nabla^2)^{-1/2}$ one could write $ $ \psi = (\nabla^2)^{-1/2 } \; \phi_{1,2 } =
{ 1\over\Gamma(-1/2) } \int | s \;
[\partial_s^2 f(s) ]^2.\end{aligned}$$ Dhich verifies jhqt the energg is conrtant in a model-independent labner. Durthermore if $f(s)$ is soooth and satisfiws snitable falloff rknditiokf at $d\to\pk\mnfty$ then the eavelet winl be nonsingunaf cnd of finite energy.
To obtain a specyfic exsmole it only reiainx to rpnlsh the complete specification or the pmtential $\psi(r,y)$. One particularly simple fhoife is to take one lf the real scajqr wavelets uf the prennous sectioh $\psi\to\phi_{1,2}$, with $\zeta^a = ||\zeta|| \; V^a = a\;
V^a$. Since the eoechtic and magnxtic fpelds are now specifiad in trrms of derivabives of a smooth bounded funrtion, the electromaggetic fiend is similarly smoity and boutded. Dor tht eierfy, wrihe mt in the fkrm $${\mathcal E} =
{1\over3} \oint \d^3 x \legt\{
\ixi \; \partial_t^4 \psi - \[awtial_t\psi \; \partial_t^3 \psi
\right\}.$$ Whence, for eifher $\psi\to\phi_{1,2}$, an integrqtion carried out at wrbitrary $t$ yields the time-independent quantity $${\mathcal E}_{1,2} = {1\over2} \;{\ohi_0^2\ivcr a}.$$ Rhf calculation has for convenience been carrieq oit in the zero-moientum framr lf jhe wavelet. In a genzdam Lorentz frame we would rave $$P^a_{1,2} = {1\over2}\; {\phi_0^2\pver a} \;\; V^a
= {1\over2}\; {\phi_0^2\over a^2} \;\; \zeta^a.$$ Thus rhis field configuxation, as fox the xcalat case, is Lorentz covarnant, bkunded, finihe energy xnd zero action. Ghe vactor naujre of the Maxwelj field hes adbed techvicak compjications, hut tmare is no real chajge iu bashc princippes.
In closing this section, I should mention ome otver wavejet tmat is particulwrly attractivg. If one iakes use of thv pseudo-dmfferential jperator $(\nabld^2)^{-1/2}$ one could wcite $$\psi = (\naboa^2)^{-1/2} \; \phi_{1,2} =
{1\ovdf\Gamma(-1/2)} \int | s \; [\partial_s^2 f(s) ]^2.\end{aligned}$$ Which verifies energy constant in model-independent manner. Furthermore satisfies falloff conditions at then the wavelet be nonsingular and of finite energy. obtain a specific example it only remains to finish the complete specification of potential $\psi(r,t)$. One particularly simple choice is to take one of the real wavelets the section with $\zeta^a = ||\zeta|| \; V^a = a\; V^a$. Since the electric and magnetic fields are specified in terms of derivatives of a smooth function, the electromagnetic field similarly smooth and bounded. For energy, it in form E} {1\over3} \oint \d^3 \left\{ \psi \; \partial_t^4 \psi - \partial_t\psi \; \partial_t^3 \psi \right\}.$$ Whence, for either $\psi\to\phi_{1,2}$, an integration out at yields the quantity E}_{1,2} {1\over2} \;{\phi_0^2\over a}.$$ has for convenience been carried out frame of the wavelet. In a general Lorentz we would $$P^a_{1,2} = {1\over2}\; {\phi_0^2\over a} \;\; = {1\over2}\; {\phi_0^2\over a^2} \;\; \zeta^a.$$ Thus this configuration, as for the scalar case, is Lorentz covariant, bounded, finite energy and zero action. nature of the Maxwell has added technical but is real in basic In closing this section, I should mention one other wavelet that particularly attractive. If one makes use of the pseudo-differential operator could $$\psi = (\nabla^2)^{-1/2} \phi_{1,2} = {1\over\Gamma(-1/2)} \int | s \;
[\partial_s^2 f(s) ]^2.\end{aligned}$$ WhicH verifies tHat thE enErgY iS conStanT in a model-indepENdenT manner. Furthermore if $f(s)$ Is smoOtH And sATiSfies SuitablE FaLLOff CoNdItiOnS At $S\to\pm\InfTy$ then tHe wavelet wIll Be Nonsingular aND oF finite eneRgy.
to obtain a speCifIc examPlE it ONly reMaiNs to fInish tHE complEte specifIcATion of THe potenTIAl $\Psi(r,T)$. One particularly sIMpLE choice is to takE one of ThE ReAL ScaLar Wavelets of ThE prevIOus sectIOn $\PSI\To\pHI_{1,2}$, with $\zeta^a = ||\zetA|| \; V^a = a\;
V^a$. Since THe eLectriC aNd mAGnetic FieldS aRE noW specified iN terMs of derivAtives OF a smootH Bounded FunctiOn, tHe eLectROmAgNetIc FIelD Is SimILarLy smooth AnD bOundeD. For THE ENergY, wrIte iT in thE form $${\mathcal E} =
{1\OveR3} \oinT \D^3 x \lEft\{
\psI \; \partIal_t^4 \PsI - \partIal_t\psI \; \partIaL_t^3 \psi
\right\}.$$ WhencE, for Either $\psi\To\pHi_{1,2}$, An iNtEgratIOn carrIed Out At arbitRary $t$ yiELds ThE TIMe-Independent quantitY $${\mATHcAl E}_{1,2} = {1\over2} \;{\pHi_0^2\over A}.$$ thE cALculatioN hAs fOr coNVEnienCe beEN cArried ouT in the ZErO-mOmentum FrAme of tHe WavEleT. In a gENeraL LorenTz frame wE woulD Have $$P^a_{1,2} = {1\over2}\; {\phi_0^2\oVEr a} \;\; V^a
= {1\over2}\; {\phi_0^2\oVEr A^2} \;\; \ZEtA^A.$$ ThuS thIs field confIgurATion, As foR ThE scALar caSe, is LOrENtZ Covariant, bounded, finItE energY and zEro action. The vEctor naturE OF The MaxweLl fiELd HAs added technicAl comPlications, BUt there iS no reAl change In basic prINCiples.
In CloSinG thIs sECTiOn, I should mentION one OtHer waveLet That is pArtIcuLarLy aTtRactive. If One makes UsE oF tHe PseUdo-diFFerentiaL oPerAtOr $(\nAbla^2)^{-1/2}$ oNE could Write $$\Psi = (\nAbLa^2)^{-1/2} \; \PHi_{1,2} =
{1\oVer\GammA(-1/2)} \InT | s \;
[\partial_s^2 f(s)]^2.\end{a ligne d}$ $ W hi ch v erif ies that the e n ergy is constant in a mode l-ind ep e nden t m anner . Furth e rm o r e i f$f (s) $i ssmoot h a nd sati sfies suit abl efalloff cond i ti ons at $s\ to\ pm\infty$ th enthe wa ve let willbenonsi ngular and of finite e ne r gy.
T o obtain a s peci fic example it on l yr emains to fini sh the c o mp l e tespe cification o f the potenti a l$ \ p si( r ,t)$. One par ticularly s i mpl e choi ce is to tak e one o f th e real scal ar w avelets o f thep revious section $\psi \to \ph i_{1 , 2} $, wi th $\z e ta ^a= || \zeta||\; V ^a =a\;V ^ a $ . Si nce the elec tric and magn eti c fi e lds arenow s peci fi ed in terms of d er ivatives of a s moot h bounded fu nc tio n, thee lectro mag net ic fiel d is si m ila rl y s mo oth and bounded. F or t he energy, write it i n the for m$${ \mat h c al E} =
{ 1 \o ver3} \o int \d ^ 3x\left\{ \psi \ ;\pa rti al_t^ 4 \ps i - \p artial_t \psi\ ; \partial_t^3 \psi
\right\} . $$ W he n ce,for either $\p si\t o \phi _{1, 2 }$ , a n inte grati on ca r ried out at arbitra ry $t$ y ields the time-ind ependent q u a n tity $${ \mat h ca l E}_{1,2} = {1 \over 2} \;{\phi _ 0^2\over a}.$ $ The ca lculation h as for c onv eni enc e b e e ncarried out i n theze ro-mome ntu m frame of th e w ave le t. In a g eneral L or en tz f ram e wew ould hav e$$P ^a _{1 ,2} = {1\ov er2}\ ; {\ ph i_ 0 ^2\ over a} \; \ ; V^a = {1\ ove r2 }\; { \phi _ 0^2 \over a ^2} \;\;\ze t a^a. $$ T hus thi s field confi gu ration, as f orthe sc a l ar case, is Lorentz covariant,b ounded, fi niteener gy and ze roaction . T h e vect or nat ure o fthe M axwel l fi eld h as added t e c hni cal c om plic ations, but there is no r e alchange in bas icprin c i pl es.
I n cl os i ngt h is section, I s hould ment io n o ne other w a vel et that i s parti cular l y attra ctive. If one make suseo f th e pseudo-d ifferent ial opera t or $( \ na bla^2 )^{ -1/2}$ o necould write $$\ psi = (\nab la ^2)^{- 1/2}\; \phi_{1 ,2} =
{1\over\Gamma(-1 /2)} \ int | s_\;
[\partial_s^2_f(s) ]^2.\end{aligned}$$ Which verifies_that the_energy_is constant_in_a model-independent manner._Furthermore if $f(s)$_is smooth and satisfies_suitable falloff conditions_at_$s\to\pm\infty$ then the wavelet will be nonsingular and of finite energy.
To obtain a specific_example_it only_remains_to_finish the complete specification of_the potential $\psi(r,t)$. One particularly_simple choice_is to take one of the real scalar_wavelets_of the previous_section $\psi\to\phi_{1,2}$, with $\zeta^a = ||\zeta|| \; V^a =_a\;
V^a$. Since the electric and magnetic_fields are now_specified_in_terms of derivatives of_a smooth bounded function, the electromagnetic_field is similarly smooth and bounded._For the energy, write it in the_form $${\mathcal E} =
{1\over3} \oint \d^3_x \left\{
\psi \; \partial_t^4_\psi -_\partial_t\psi \; \partial_t^3 \psi
\right\}.$$ Whence,_for either $\psi\to\phi_{1,2}$,_an integration_carried out at_arbitrary $t$ yields the time-independent quantity_$${\mathcal E}_{1,2} =_{1\over2} \;{\phi_0^2\over a}.$$ The calculation has_for_convenience been carried_out_in_the zero-momentum_frame of the_wavelet._In a_general_Lorentz frame we would have $$P^a_{1,2}__= {1\over2}\; {\phi_0^2\over a} \;\; V^a
=_ {1\over2}\; {\phi_0^2\over a^2}_\;\;_\zeta^a.$$ Thus this field_configuration, as for the scalar_case, is Lorentz covariant, bounded, finite_energy and_zero action._The vector nature of the Maxwell field has added technical complications,_but there is no real change_in basic principles.
In closing_this section,_I_should mention one_other_wavelet that_is particularly attractive. If one makes use_of the_pseudo-differential operator $(\nabla^2)^{-1/2}$ one could write_$$\psi = (\nabla^2)^{-1/2} \;_\phi_{1,2}_=
{1\over\Gamma(-1/2)} \int |
in L^2([0,1])$ with $f''\in
L^2([0,1])$ and with periodic boundary conditions. Then $B$ has (complex-valued) eigenfunctions $u_k(x)=\exp(2\pi kix)$ with eigenvalues $\nu_k=(2\pi k)^2$ such that $$H^r=\{f\in L^2([0,1])\,:\,
\sum_{k\in\operatorname{{\mathbb Z}}}\nu_k^{2r}{\lvert {\langle f,u_k \rangle} \rvert}^2<\infty\}$$ is the classical $L^2$-Sobolev space $H^{2r}_{per}$ of regularity (smoothness) $2r$ with periodic boundary conditions. See, e.g., Edmunds and Evans (1987) for many examples of generating smooth function spaces from differential operators.
For the typical choice ${\varphi}(t)=t^a$ for some $a>0$, Assumption \[K-smooth\] translates to ${\lVert Kh \rVert}_W{\leqslant}M{\lVert B^{-a}h \rVert}_X$, which means intuitively that the operator $K$ regularizes at least as much as $B^{-a}$. In the case $Bf(x):=-f''(x)$ the operator $K$ acts like integrating at least $(2a)$-times, i.e. maps $L^2$ to the $L^2$-Sobolev space of regularity $2a$.
In the statistics literature, for the standard nonparametric mean regression model (i.e., the model in which $K$ is the identity operator), the minimax risk lower and upper bounds have been established in mean integrated squared error loss for various classes of functions ${\mathscr}H$ such as a Sobolev ball (ellipsoid), a Hölder ball (hyperrectangle) or a Besov ball (ellipsoid or hyperrectangle or Besov body); see, e.g., Donoho, Liu and MacGibbon (1990), Yang and Barron (1999) and the references therein. As shown in these papers, what matters for minimax risk lower and upper bounds for nonparametric mean regression estimation is the complexity of the class of functions ${\mathscr}H$ that can be measured in terms of best finite dimensional approximation numbers. This motivates us to impose Assumption \[h- | in L^2([0,1])$ with $ f''\in
L^2([0,1])$ and with periodic boundary conditions. Then $ B$ has (complex - valued) eigenfunctions $ u_k(x)=\exp(2\pi kix)$ with eigenvalues $ \nu_k=(2\pi k)^2 $ such that $ $ H^r=\{f\in L^2([0,1])\,:\,
\sum_{k\in\operatorname{{\mathbb Z}}}\nu_k^{2r}{\lvert { \langle f, u_k \rangle } \rvert}^2<\infty\}$$ is the classical $ L^2$-Sobolev quad $ H^{2r}_{per}$ of regularity (smoothness) $ 2r$ with periodic boundary condition. See, for example, Edmunds and Evans (1987) for many examples of beget smooth function space from differential operators.
For the typical option $ { \varphi}(t)=t^a$ for some $ a>0 $, premise \[K - smooth\ ] translates to $ { \lVert Kh \rVert}_W{\leqslant}M{\lVert B^{-a}h \rVert}_X$, which intend intuitively that the operator $ K$ regularizes at least equally much as $ B^{-a}$. In the case $ Bf(x):=-f''(x)$ the operator $ K$ acts like integrating at least $ (2a)$-times, i.e. maps $ L^2 $ to the $ L^2$-Sobolev space of regularity $ 2a$.
In the statistics literature, for the standard nonparametric mean arrested development model (i.e., the model in which $ K$ is the identity operator), the minimax hazard lower and upper bounds have been established in mean integrated squared error personnel casualty for various classes of functions $ { \mathscr}H$ such as a Sobolev ball (ellipsoid), a Hölder ball (hyperrectangle) or a Besov musket ball (ellipsoid or hyperrectangle or Besov body); see, e.g., Donoho, Liu and MacGibbon (1990), Yang and Barron (1999) and the references therein. As shown in these papers, what matters for minimax risk lower and upper bounds for nonparametric mean regression estimation is the complexity of the class of functions $ { \mathscr}H$ that can be measured in terms of good finite dimensional approximation numbers. This motivates us to levy Assumption \[h- | in P^2([0,1])$ with $f''\in
L^2([0,1])$ and with perlodic boundary conditiois. Then $B$ has (cumplex-valued) eigenfunctions $n_k(x)=\ezp(2\pi jix)$ with eigenvalues $\nj_k=(2\pi k)^2$ subh that $$H^e=\{f\in O^2([0,1])\,:\,
\sum_{k\in\opecztornamc{{\iathgn Z}}}\nu_n^{2c}{\lvert {\langle f,o_k \rangle} \rvart}^2<\infty\}$$ is tha zlcssical $L^2$-Sobolev space $H^{2r}_{per}$ of regujarity (xmlothness) $2r$ witr pegijdic boundary conditions. See, e.g., Edmunda and Enans (1987) for many exsmples of generating smootj fujction spaces from differentiql o[wrators.
For tfe typical choice ${\varpgi}(t)=t^a$ for some $a>0$, Assumption \[K-smuoth\] cranslates jo ${\lGgrt Kh \rVert}_X{\leqslwnt}M{\lVert B^{-a}m \rVert}_F$, which means intuitiyely vhat the operator $K$ regulerizes at least as moch as $B^{-a}$. Hn the case $Bf(x):=-f''(x)$ tye operdtor $K$ azrs uikt iitefratinh av least $(2a)$-tijes, i.e. maps $L^2$ to the $L^2$-Sobolev xpwbr of regularjty $2a$.
Ig ehe statistics literature, for the standdrd nonparametric mean regeession model (i.e., the lodel in rhich $K$ is the identity operator), the minimax risk lowec xnd uiper voknds have been established in mean integrated aqiaged error loss fov various classes pf figctions ${\mathszr}H$ sueg zs a Sobolev ball (fllipsoyd), a Yölder balj (hylerrectangle) or a Besov balo (ellipsoid je hyperrectangle ox Besov body); see, e.g., Dpnoho, Liu and MacGibbon (1990), Yahg and Barrln (1999) and tgd references thefeik. Av shown in these papers, whwt matterw fox minimab rixk lowqr and uppfr bomtds for nonparametgic mgan recression edtimation is the complexity of vie class of foncdiots ${\mathser}H$ thst can be meafured in terms of besc finige dimensiknal ap'roximation gumbers. This kltivates us vo impose Assymptuon \[h- | in L^2([0,1])$ with $f''\in L^2([0,1])$ and with conditions. $B$ has eigenfunctions $u_k(x)=\exp(2\pi kix)$ that L^2([0,1])\,:\, \sum_{k\in\operatorname{{\mathbb Z}}}\nu_k^{2r}{\lvert f,u_k \rangle} \rvert}^2<\infty\}$$ the classical $L^2$-Sobolev space $H^{2r}_{per}$ of (smoothness) $2r$ with periodic boundary conditions. See, e.g., Edmunds and Evans (1987) for examples of generating smooth function spaces from differential operators. For the typical choice for $a>0$, \[K-smooth\] to ${\lVert Kh \rVert}_W{\leqslant}M{\lVert B^{-a}h \rVert}_X$, which means intuitively that the operator $K$ regularizes at least much as $B^{-a}$. In the case $Bf(x):=-f''(x)$ the $K$ acts like integrating least $(2a)$-times, i.e. maps $L^2$ the space of $2a$. the literature, for the nonparametric mean regression model (i.e., the model in which $K$ is the identity operator), the minimax risk and upper been established mean squared loss for various functions ${\mathscr}H$ such as a Sobolev Hölder ball (hyperrectangle) or a Besov ball (ellipsoid hyperrectangle or body); see, e.g., Donoho, Liu and (1990), Yang and Barron (1999) and the references As shown in these papers, what matters for minimax risk lower and upper bounds for regression estimation is the of the class functions that be in terms best finite dimensional approximation numbers. This motivates us to impose Assumption | in L^2([0,1])$ with $f''\in
L^2([0,1])$ and with periodiC boundary cOnditIonS. ThEn $b$ has (CompLex-valued) eigenFUnctIons $u_k(x)=\exp(2\pi kix)$ with eigEnvalUeS $\Nu_k=(2\pI K)^2$ sUch thAt $$H^r=\{f\in l^2([0,1])\,:\,
\SuM_{K\In\oPeRaTorNaME{{\mAthbb z}}}\nu_K^{2r}{\lvert {\Langle f,u_k \rAngLe} \Rvert}^2<\infty\}$$ is THe Classical $L^2$-sobOlev space $H^{2r}_{pEr}$ oF regulArIty (SMoothNesS) $2r$ witH perioDIc bounDary condiTiONs. See, e.G., edmunds AND EVans (1987) For many examples of GEnERating smooth fuNction SpACeS FRom DifFerential oPeRatorS.
for the tYPiCAL ChoICe ${\varphi}(t)=t^a$ foR some $a>0$, AssumPTioN \[K-smooTh\] TraNSlates To ${\lVeRt kH \rVErt}_W{\leqslanT}M{\lVErt B^{-a}h \rVeRt}_X$, whiCH means iNTuitiveLy that The OpeRatoR $k$ rEgUlaRiZEs aT LeAst AS muCh as $B^{-a}$. In ThE cAse $Bf(X):=-f''(x)$ tHE OPEratOr $K$ Acts Like iNtegrating at lEasT $(2a)$-tiMEs, i.E. maps $l^2$ to thE $L^2$-SoBoLev spAce of rEgulaRiTy $2a$.
In the statistIcs lIterature, For ThE stAnDard nONparamEtrIc mEan regrEssion mODel (I.e., THE MoDel in which $K$ is the idEnTITy Operator), The minIMaX rISk lower aNd UppEr boUNDs havE beeN EsTablisheD in meaN InTeGrated sQuAred erRoR loSs fOr varIOus cLasses Of functiOns ${\maTHscr}H$ such as a SoBOlev ball (ellipSOiD), A höLDer bAll (HyperrectanGle) oR A BesOv baLL (eLliPSoid oR hypeRrECtANgle or Besov body); see, e.G., DOnoho, LIu and macGibbon (1990), Yang And Barron (1999) aND THe refereNces THeREin. As shown in thEse paPers, what maTTers for mInimaX risk lowEr and uppeR BOunds for NonParAmeTriC MEaN regression esTIMatiOn Is the coMplExity of The ClaSs oF fuNcTions ${\mathScr}H$ that CaN bE mEaSurEd in tERms of besT fIniTe DimEnsioNAl apprOximaTion NuMbERs. THis motiVAtES Us to ImPoSe AsSumPtIon \[h- | in L^2([0,1])$ with $f''\i n
L^2([0,1 ])$ a ndwit hperi odic boundary cond i tion s. Then $B$ has (compl ex-va lu e d) e i ge nfunc tions $ u _k ( x )=\ ex p( 2\p ik ix )$ wi theigenva lues $\nu_ k=( 2\ pi k)^2$ suc h t hat $$H^r= \{f \in L^2([0,1 ])\ ,:\,
\ su m_{ k \in\o per atorn ame{{\ m athbbZ}}}\nu_k ^{ 2 r}{\lv e rt {\la n g le f,u _k \rangle} \rver t }^ 2 <\infty\}$$ is the c la s si c a l $ L^2 $-Sobolevsp ace $ H ^{2r}_{ p er } $ ofr egularity (sm oothness) $ 2 r$with p er iod i c boun daryco n dit ions. See,e.g. , Edmunds and E v ans (19 8 7) formany e xam ple s of ge ne rat in g sm o ot h f u nct ion spac es f rom d iffe r e n t ialope rato rs.
For the typic alchoi c e $ {\var phi}( t)=t ^a $ for some$a>0$ ,Assumption \[K- smoo th\] tran sla te s t o${\lV e rt Kh\rV ert }_W{\le qslant} M {\l Ve r t B^ {-a}h \rVert}_X$,wh i c hmeans in tuitiv e ly t h at the o pe rat or $ K $ regu lari z es at leas t as m u ch a s $B^{- a} $. Inth e c ase $Bf( x ):=- f''(x) $ the op erato r $K$ acts like integrating a t l e a st $(2a )$- times, i.e. map s $L^ 2$ t o t he$ L^2$- Sobol ev sp a ce of regularity $2 a$ .
Inthe s tatistics lit erature, f o r the stan dard no n parametric mea n reg ression mo d el (i.e. , the model i n which $ K $ is theide nti tyope r a to r), the minim a x ris klower a ndupper b oun dshav e b ee n establi shed inme an i nt egr ateds quared e rr orlo ssfor v a riousclass es o ffu n cti ons ${\ m at h s cr}H $su ch a s a S obole v ba l l ( ellipso id), a Hö lde r bal l(h yperrec tangle) or aBe sov ball ( el lip soid o r hyperrec tangle or Besov body);s ee, e.g .,Donoh o, L iu and Ma cGi bbon ( 199 0 ), Yan g andBarro n(19 9 9 ) and t he re fe rences the r e in. As s ho wn i n these papers, what matt e rsfor minimax r isk low e r a ndu pp e r b ou n dsf o r nonparametric mean regr es s io n estimati o n i sthe com plexity of t h e class of funct ions ${\m at hscr } H $ t hat can be measure d in term s of b e st fini tedimens io nal appr oximat i onnumbe rs. Th is motiv atesus to impo se Assumption \[h- | in L^2([0,1])$_with $f''\in
L^2([0,1])$_and with periodic boundary_conditions. Then_$B$_has (complex-valued)_eigenfunctions_$u_k(x)=\exp(2\pi kix)$ with_eigenvalues $\nu_k=(2\pi k)^2$_such that $$H^r=\{f\in L^2([0,1])\,:\,
\sum_{k\in\operatorname{{\mathbb_Z}}}\nu_k^{2r}{\lvert {\langle f,u_k_\rangle}_\rvert}^2<\infty\}$$ is the classical $L^2$-Sobolev space $H^{2r}_{per}$ of regularity (smoothness) $2r$ with periodic boundary_conditions._See, e.g.,_Edmunds_and_Evans (1987) for many examples_of generating smooth function spaces_from differential_operators.
For the typical choice ${\varphi}(t)=t^a$ for some $a>0$,_Assumption_\[K-smooth\] translates to_${\lVert Kh \rVert}_W{\leqslant}M{\lVert B^{-a}h \rVert}_X$, which means intuitively that_the operator $K$ regularizes at least_as much as_$B^{-a}$._In_the case $Bf(x):=-f''(x)$ the_operator $K$ acts like integrating at_least $(2a)$-times, i.e. maps $L^2$ to_the $L^2$-Sobolev space of regularity $2a$.
In the_statistics literature, for the standard nonparametric_mean regression model (i.e., the_model in_which $K$ is the identity_operator), the minimax_risk lower_and upper bounds_have been established in mean integrated_squared error loss_for various classes of functions ${\mathscr}H$_such_as a Sobolev_ball_(ellipsoid),_a Hölder_ball (hyperrectangle) or_a_Besov ball_(ellipsoid_or hyperrectangle or Besov body); see,_e.g.,_Donoho, Liu and MacGibbon (1990), Yang and_Barron (1999) and the_references_therein. As shown in_these papers, what matters for_minimax risk lower and upper bounds_for nonparametric_mean regression_estimation is the complexity of the class of functions ${\mathscr}H$ that_can be measured in terms of_best finite dimensional approximation_numbers. This_motivates_us to impose_Assumption_\[h- |
\right) \,.
\end{split}$$ This expression can be written as $I=\sum_r^d f(a,b,c,d,r)$, where $f$ is a well-behaved function. Furthermore, we can write this result as $$I_{abcd}=\sum_{r=0}^d \exp{\log{[f(a,b,c,d,r)]}}\,,$$ where $f$ must be positive.\
\
The logarithm $\log{(f)}$ is given by $$\begin{split}
\log{(f)}=&-\frac{1}{2}\log{(2)}-2\log{(\pi)}\\ &+\frac{1}{2}\left(\log{(c!)}+\log{(d!)}
-\log{(a!)}-\log{(b!)}\right)\\
&-\log{\left((d-r)!\right)}-\log{\left((c-d+r)!\right)}-\log{(r!)}\\
&+\log{\left(\Gamma\left(\frac{a+b-c+d+1}{2}-r\right)\right)}\\
&+\log{\left(\Gamma\left(\frac{a-b+c-d+1}{2}+r\right)\right)}\\
&+\log{\left(\Gamma\left(\frac{-a+b+c-d+1}{2}+r\right)\right)}\,.
\end{split}$$ Notice that due to the presence of the gamma functions, the previous expression could require the evaluation of a logarithm of a negative quantity. For this reason, we calculate instead $\log{(\abs{\Gamma(n)})}$, and compute the associated phase separately. This phase is introduced in the final expression after the exponentiation. Thus, the final expression of the integral taking into account the possible negative values of gamma functions reads $$I_{abcd}=\sum_{r=0}^d p\exp{\log{[f(a,b,c,d,r)]}}\,,$$ where $p$ is the phase, given by $p=\prod_{n=1}^3 p_n$, with $p_n$ being the phase generated by each gamma function.
![Energy spectrum of the two-particle system as a function of the interaction strength. The calculations are performed using 100 single-particle modes (solid green line). The analytical spectrum ( | \right) \, .
\end{split}$$ This expression can be written as $ I=\sum_r^d f(a, b, c, d, r)$, where $ f$ is a well - behaved function. Furthermore, we can spell this resultant role as $ $ I_{abcd}=\sum_{r=0}^d \exp{\log{[f(a, b, c, d, r)]}}\,,$$ where $ f$ must be positive.\
\
The logarithm $ \log{(f)}$ is given by $ $ \begin{split }
\log{(f)}=&-\frac{1}{2}\log{(2)}-2\log{(\pi)}\\ & + \frac{1}{2}\left(\log{(c!)}+\log{(d !) }
-\log{(a!)}-\log{(b!)}\right)\\
& -\log{\left((d - r)!\right)}-\log{\left((c - d+r)!\right)}-\log{(r!)}\\
& + \log{\left(\Gamma\left(\frac{a+b - c+d+1}{2}-r\right)\right)}\\
& + \log{\left(\Gamma\left(\frac{a - b+c - d+1}{2}+r\right)\right)}\\
& + \log{\left(\Gamma\left(\frac{-a+b+c - d+1}{2}+r\right)\right)}\, .
\end{split}$$ Notice that due to the presence of the gamma function, the previous expression could want the evaluation of a logarithm of a minus quantity. For this rationality, we calculate instead $ \log{(\abs{\Gamma(n)})}$, and calculate the associated phase separately. This phase is introduced in the final formula after the exponentiation. Thus, the final expression of the integral taking into history the possible negative values of da gamma functions reads $ $ I_{abcd}=\sum_{r=0}^d p\exp{\log{[f(a, b, c, d, r)]}}\,,$$ where $ p$ is the phase, give by $ p=\prod_{n=1}^3 p_n$, with $ p_n$ being the phase render by each gamma routine.
! [ Energy spectrum of the two - particle system as a affair of the interaction strength. The calculations are performed using 100 single - particle modes (solid green line). The analytical spectrum ( | \rigjt) \,.
\end{split}$$ This expresslon can be writtgn as $I=\snm_r^d f(a,g,c,d,r)$, whefe $f$ is a well-behaved functiln. Furtyermore, we can write tfis resuln as $$I_{abce}=\sum_{c=0}^d \exp{\log{[f(a,b,c,d,r)]}}\,,$$ xgere $f$ must bs posntmve.\
\
The logarithk $\log{(f)}$ is civen by $$\begin{vpuic}
\log{(f)}=&-\frac{1}{2}\log{(2)}-2\log{(\pi)}\\ &+\frac{1}{2}\left(\log{(c!)}+\lod{(d!)}
-\lpg{(w!)}-\log{(b!)}\right)\\
&-\jog{\ltft((q-r)!\rifht)}-\log{\left((c-d+r)!\right)}-\log{(r!)}\\
&+\log{\left(\Gzmma\lefu(\frac{a+b-c+d+1}{2}-r\right)\rignt)}\\
&+\log{\left(\Gamma\left(\frac{w-b+c-d+1}{2}+g\right)\right)}\\
&+\log{\peft(\Gamma\ledt(\frwx{-a+b+c-d+1}{2}+r\right)\rkght)}\,.
\end{splpc}$$ Notice thzt due to the presence of the gxmma yunctions, tye prfeious exprewsion could requivv the eealuatipn of a logaribhm oh a begative quantity. For this reason, we cajculate itscead $\log{(\abs{\Gamma(n)})}$, and cimputg the assuxiaged piass sepagatxly. This phzse is intriduced in the final evitession afted the qx[onentiation. Thus, the final expression mf fhe integral taking inti account the possiblg negative values of gamma functions reads $$I_{abcd}=\sum_{r=0}^d p\exp{\lmg{[f(a,b,r,d,f)]}}\,,$$ wkcve $p$ us the phase, given by $p=\prod_{n=1}^3 p_n$, with $p_n$ being trs khaxe generated bn each gamma functooj.
![Emgrgy spectrum uf the twk-particle system ad a funstion of the igtersction strength. The calcularions are pegforned using 100 single-pcrticle moder (sokid gteen line). The analyticau spsctrum ( | \right) \,. \end{split}$$ This expression can be $I=\sum_r^d where $f$ a well-behaved function. result $$I_{abcd}=\sum_{r=0}^d \exp{\log{[f(a,b,c,d,r)]}}\,,$$ where must be positive.\ The logarithm $\log{(f)}$ is given by \log{(f)}=&-\frac{1}{2}\log{(2)}-2\log{(\pi)}\\ &+\frac{1}{2}\left(\log{(c!)}+\log{(d!)} -\log{(a!)}-\log{(b!)}\right)\\ &-\log{\left((d-r)!\right)}-\log{\left((c-d+r)!\right)}-\log{(r!)}\\ &+\log{\left(\Gamma\left(\frac{a+b-c+d+1}{2}-r\right)\right)}\\ &+\log{\left(\Gamma\left(\frac{a-b+c-d+1}{2}+r\right)\right)}\\ &+\log{\left(\Gamma\left(\frac{-a+b+c-d+1}{2}+r\right)\right)}\,. \end{split}$$ Notice that due to the of the gamma functions, the previous expression could require the evaluation of a of negative For reason, we calculate instead $\log{(\abs{\Gamma(n)})}$, and compute the associated phase separately. This phase is introduced in final expression after the exponentiation. Thus, the final of the integral taking account the possible negative values gamma reads $$I_{abcd}=\sum_{r=0}^d where is phase, given by p_n$, with $p_n$ being the phase generated by each gamma function. ![Energy spectrum of the two-particle system a function interaction strength. calculations performed 100 single-particle modes line). The analytical spectrum ( | \right) \,.
\end{split}$$ This expressiOn can be wriTten aS $I=\sUm_r^D f(A,b,c,d,R)$, wheRe $f$ is a well-behaVEd fuNction. Furthermore, we can Write ThIS resULt As $$I_{abCd}=\sum_{r=0}^d \EXp{\LOG{[f(a,B,c,D,r)]}}\,,$$ WheRe $F$ MuSt be pOsiTive.\
\
The Logarithm $\lOg{(f)}$ Is Given by $$\begin{SPlIt}
\log{(f)}=&-\frac{1}{2}\Log{(2)}-2\Log{(\pi)}\\ &+\frac{1}{2}\lefT(\loG{(c!)}+\log{(d!)}
-\LoG{(a!)}-\lOG{(b!)}\rigHt)\\
&-\lOg{\lefT((d-r)!\rigHT)}-\log{\leFt((c-d+r)!\righT)}-\lOG{(r!)}\\
&+\log{\lEFt(\Gamma\LEFt(\Frac{A+b-c+d+1}{2}-r\right)\right)}\\
&+\loG{\LeFT(\Gamma\left(\frac{A-b+c-d+1}{2}+r\rIgHT)\rIGHt)}\\
&+\lOg{\lEft(\Gamma\leFt(\Frac{-a+B+C-d+1}{2}+r\righT)\RiGHT)}\,.
\End{SPlit}$$ Notice thaT due to the prESenCe of thE gAmmA FunctiOns, thE pREviOus expressiOn coUld requirE the evALuation OF a logarIthm of A neGatIve qUAnTiTy. FOr THis REaSon, WE caLculate iNsTeAd $\log{(\Abs{\GAMMA(N)})}$, and ComPute The asSociated phase SepAratELy. THis phAse is IntrOdUced iN the fiNal exPrEssion after the eXponEntiation. thuS, tHe fInAl expREssion Of tHe iNtegral Taking iNTo aCcOUNT tHe possible negative VaLUEs Of gamma fUnctioNS rEaDS $$I_{abcd}=\suM_{r=0}^D p\eXp{\loG{[F(A,b,c,d,r)]}}\,,$$ WherE $P$ iS the phasE, given BY $p=\PrOd_{n=1}^3 p_n$, wiTh $P_n$ beinG tHe pHasE geneRAted By each Gamma funCtion.
![eNergy spectrum oF The two-particlE SySTEm AS a fuNctIon of the intEracTIon sTrenGTh. the CAlculAtionS aRE pERformed using 100 single-pArTicle mOdes (sOlid green line). the analytiCAL Spectrum ( | \right) \,.
\end{split}$$This expre ssion ca n b ewrit tenas $I=\sum_r^d f(a, b,c,d,r)$, where $f$ i s a w el l -beh a ve d fun ction.F ur t h erm or e, we c a nwrite th is resu lt as $$I_ {ab cd }=\sum_{r=0} ^ d\exp{\log{ [f( a,b,c,d,r)]} }\, ,$$ wh er e $ f $ mus t b e pos itive. \
\
The logarith m$ \log{( f )}$ isg i ve n by $$\begin{split} \log{(f)}=&-\f rac{1} {2 } \l o g {(2 )}- 2\log{(\pi )} \\ &+ \ frac{1} { 2} \ l e ft( \ log{(c!)}+\lo g{(d!)}
-\l og{(a! )} -\l o g{(b!) }\rig ht ) \\ &-\log{ \lef t((d-r)!\ right) } -\log{\ l eft((c- d+r)!\ rig ht) }-\l o g{ (r !)} \\ &+ \lo g {\l eft(\Gam ma \l eft(\ frac { a + b -c+d +1} {2}- r\rig ht)\right)}\\
&+ \ log {\lef t(\Ga mma\ le ft(\f rac{a- b+c-d +1 }{2}+r\right)\r ight )}\\
&+\ lo g{\ le ft(\G a mma\le ft( \fr ac{-a+b +c-d+1} { 2}+ r\ r i g ht )\right)}\,.
\end{ sp l i t} $$ Notic e that du et o the pr es enc e of t he ga mmaf un ctions,the pr e vi ou s expre ss ion co ul d r equ ire t h e ev aluati on of alogar i thm of a negat i ve quantity.F or t hi s rea son , we calcul atei nste ad $ \ lo g{( \ abs{\ Gamma (n ) }) } $, and compute theas sociat ed ph ase separatel y. This ph a s e is intr oduc e di n the final ex press ion aftert he expon entia tion. Th us, the f i n al expre ssi onofthe i nt egral takingi n to a cc ount th e p ossible ne gat ive va lu es of gam ma funct io ns r ea ds$$I_{ a bcd}=\su m_ {r= 0} ^dp\exp { \log{[ f(a,b ,c,d ,r )] } }\, ,$$ whe r e$ p $ is t he pha se, g ivenby $ p =\p rod_{n= 1}^3 p_n$ , w i th $ p_ n$ beingthe phase gen er ated by ea ch ga mma fu n c tion.
! [Energy spectrum of the two-par tic le sy stem as a fun cti on ofthe intera ctionstren gt h.T h e cal c u la tio ns are perfo r m edusing 1 00 s ingle-p article modes (sol i d g reen line). T heanal y t ic als pe c tru m( | \right) \,.
\end{split}$$_This expression_can be written as_$I=\sum_r^d f(a,b,c,d,r)$,_where_$f$ is_a_well-behaved function. Furthermore,_we can write_this result as $$I_{abcd}=\sum_{r=0}^d_\exp{\log{[f(a,b,c,d,r)]}}\,,$$ where $f$_must_be positive.\
\
The logarithm $\log{(f)}$ is given by $$\begin{split}
\log{(f)}=&-\frac{1}{2}\log{(2)}-2\log{(\pi)}\\ &+\frac{1}{2}\left(\log{(c!)}+\log{(d!)}
__ -\log{(a!)}-\log{(b!)}\right)\\
___ &-\log{\left((d-r)!\right)}-\log{\left((c-d+r)!\right)}-\log{(r!)}\\
_&+\log{\left(\Gamma\left(\frac{a+b-c+d+1}{2}-r\right)\right)}\\
&+\log{\left(\Gamma\left(\frac{a-b+c-d+1}{2}+r\right)\right)}\\
_ _ &+\log{\left(\Gamma\left(\frac{-a+b+c-d+1}{2}+r\right)\right)}\,.
\end{split}$$ Notice that due to the presence_of_the gamma functions,_the previous expression could require the evaluation of a_logarithm of a negative quantity. For_this reason, we_calculate_instead_$\log{(\abs{\Gamma(n)})}$, and compute the_associated phase separately. This phase is_introduced in the final expression after_the exponentiation. Thus, the final expression of_the integral taking into account the_possible negative values of gamma_functions reads_$$I_{abcd}=\sum_{r=0}^d p\exp{\log{[f(a,b,c,d,r)]}}\,,$$ where $p$ is_the phase, given_by $p=\prod_{n=1}^3_p_n$, with $p_n$_being the phase generated by each_gamma function.
![Energy spectrum_of the two-particle system as a_function_of the interaction_strength._The_calculations are_performed using 100_single-particle_modes (solid_green_line). The analytical spectrum ( |
. Here we are interested in neutral stoichiometric clusters of typical materials with ionic bonding, that is $(AX)_n$ clusters, where $A$ is an alkali and X a halide atom. As [*ab initio*]{} studies on these clusters are computationally expensive, the first theoretical calculations were based on pairwise interaction models [@Mar83; @Die85; @Phi91]. Meanwhile, experimentalists moved forward using several techniques to produce and investigate these clusters: particle sputtering [@Cam81; @Bar82], where rare-gas ions are used to bombard a crystal surface, vapor condensation in an inert-gas atmosphere [@Ech81; @Pfl85; @Pfl86], and laser vaporization of a crystal surface [@Con88; @Twu90]. In the expanding molecular jet conditions, clusters undergo a rapid and efficient evaporative cooling, and this leads to a cluster size distribution determined almost exclusively by the cluster stability. The evaporative cooling process leads to the so-called abundance magic numbers as a result of the longer time that the most stable clusters remain in the beam before decaying. Alkali halide magic numbers are often explained in terms of cuboid structures resembling fragments of the crystalline lattice, but other possibilities like ring stackings, or even mixed structures exist, which could be competitive. Nevertheless, only recently the possibility of detecting different isomers has emerged in drift tube experiments [@Jar95; @Mai96]. Those experiments have led to a renewed interest in isomerization studies. In order to disentangle these interesting problems, [*ab initio*]{} calculations [@Ayuel; @Agu97; @Wei92; @Och94] provide an ideal complement to the experimental studies, which are restricted to charged species. For instance, a study of $(KCl)_n$ and $(LiF)_n$ clusters up to sizes of $n=32$ has been carried out in ref., using quantum-chemical methods including correlation effects at the MP2 level. We have performed calculations for $(NaCl)_n$ and $(NaI)_n$ using the [*ab initio*]{} Perturbed Ion (PI) model [@Ayuel; @Agu97].
The objective of the present paper is to give a global characterization of the structure and other related properties of small neutral alkali halide clusters. To this end we have carried out extensive PI calculations for the $(AX)_n$ $(n=1-10)$ clusters, identifying the most stable isomers, | . Here we are interested in neutral stoichiometric clusters of typical fabric with ionic bonding, that is $ (AX)_n$ bunch, where $ A$ is an alkali and X a halide atom. As [ * ab initio * ] { } report on these bunch are computationally expensive, the first theoretical calculations were based on pairwise interaction exemplar [ @Mar83; @Die85; @Phi91 ]. Meanwhile, experimentalists moved forward use several techniques to grow and investigate these clusters: particle spatter [ @Cam81; @Bar82 ], where rare - gas ions are used to bombard a crystal surface, vapor compression in an inert - gas atmosphere [ @Ech81; @Pfl85; @Pfl86 ], and laser vaporization of a crystal airfoil [ @Con88; @Twu90 ]. In the expanding molecular jet conditions, clusters undergo a rapid and effective evaporative cooling, and this lead to a cluster size distribution determined almost entirely by the cluster stability. The evaporative cooling process leads to the so - call abundance magic numbers as a result of the longer time that the most stable clusters remain in the beam before disintegrate. Alkali halide magic numbers are much excuse in terms of cuboid structures resembling fragment of the crystalline lattice, but other possibilities like ring stackings, or even mixed social organization exist, which could be competitive. Nevertheless, only recently the possibility of detecting unlike isomers has emerged in drift tube experiments [ @Jar95; @Mai96 ]. Those experiments have led to a renewed sake in isomerization studies. In order to disentangle these interesting problems, [ * ab initio * ] { } calculations [ @Ayuel; @Agu97; @Wei92; @Och94 ] leave an ideal complement to the experimental studies, which are restricted to charged coinage. For instance, a report of $ (KCl)_n$ and $ (LiF)_n$ clusters up to sizes of $ n=32 $ has been post out in ref. , use quantum - chemical methods including correlation coefficient effects at the MP2 level. We have performed calculations for $ (NaCl)_n$ and $ (NaI)_n$ practice the [ * ab initio * ] { } Perturbed Ion (PI) model [ @Ayuel; @Agu97 ].
The aim of the present paper is to give a global portrayal of the structure and early related properties of small neutral alkali halide bunch. To this end we have carried out extensive PI calculations for the $ (AX)_n$ $ (n=1 - 10)$ bunch, identifying the most stable isomers, | . Hege we are interested in keutral stoichiometric rlustera of typkcal materials with ionic boidint, thau is $(AX)_n$ clusters, wfere $A$ is an alkaoi aid X a halide atom. As [*ab initik*]{} stubixs on these cluxters are womputationallf dx'ensive, the first theoretical calculwtions eege based on payrwixq infvrcction models [@Mar83; @Die85; @Phi91]. Meanwgile, exkerimentalists movrd forward using several tfchnlques to produce ajd investigqte eyese clusterr: particle sputtering [@Dam81; @Bar82], where rare-gas ions are jsed co bombard q xrydjal surface, tapor bondensation lm an itert-gas atmosphere [@Ecm81; @Pfl85; @Pfo86], and laser vaporizatmon of a crystal sursace [@Con88; @Dwb90]. In the expanding mooexular jet conautiuns, ckuaters kndxrgo a rapis and efficuent evaporative coplybg, and this lsads tj w cluster size distribution determined dlmkst exclusively by the xluster stability. The evaporatyve cooling process leads to the so-called abundanwe majiz nbnbers qs a result of the longer time that the most stwglt ckusters remain in the beak hegjre decaying. Xlkali hamide magic numbers are ofjen explained ig tetms of cuboid structures rewembling fraymebts of the crystalpine latticz, but pther possibilities like riny staciings, or evfn mixed agructures exist, dhibh cmuld be competitive. Nevertreless, onoy rzcently ghe kossibijity of dehectikc different isomerd has ekerged in frift tube experiments [@Jar95; @Mai96]. Viose experimemtv hdve led co a rcnewed interest in isomerizatnon studnes. In order to sisentaigle these igteresting prmhlems, [*ab inivio*]{} calcujatiins [@Qyuel; @Aej97; @Wei92; @Och94] provode an idvaj conplement to the exierimgnfal studies, whiek qre restricted yo zhawgvd xpqwies. For insdancd, a xtudy of $(KCl)_n$ anb $(UiF)_n$ clusters up to sizev of $n=32$ has been carriec jut in rgf., using zuantum-chemicsl methods includijg cocrelatmon efgecjs at the MP2 level. We have perfkrmed calfulwtions for $(NwCl)_n$ and $(NaI)_n$ usnng the [*ab initio*]{} Perturbed Ion (PI) model [@Ayuel; @Agu97].
The objgctive of the present kaper is to gmve a dlobal chdracterization of thw structure and ouher related properties of smdll nfutral alkali halide clusters. To this end we have carried out extensive PU calcnlwtions for fhe $(SX)_n$ $(n=1-10)$ cjbsters, ydenvihying the most stsble isomers, | . Here we are interested in neutral of materials with bonding, that is an and X a atom. As [*ab studies on these clusters are computationally the first theoretical calculations were based on pairwise interaction models [@Mar83; @Die85; @Phi91]. experimentalists moved forward using several techniques to produce and investigate these clusters: particle [@Cam81; where ions used to bombard a crystal surface, vapor condensation in an inert-gas atmosphere [@Ech81; @Pfl85; @Pfl86], and vaporization of a crystal surface [@Con88; @Twu90]. In expanding molecular jet conditions, undergo a rapid and efficient cooling, this leads a size determined almost exclusively the cluster stability. The evaporative cooling process leads to the so-called abundance magic numbers as a result the longer the most clusters in beam before decaying. magic numbers are often explained in structures resembling fragments of the crystalline lattice, but possibilities like stackings, or even mixed structures exist, could be competitive. Nevertheless, only recently the possibility detecting different isomers has emerged in drift tube experiments [@Jar95; @Mai96]. Those experiments have led renewed interest in isomerization In order to these problems, initio*]{} [@Ayuel; @Agu97; @Och94] provide an ideal complement to the experimental studies, which are to charged species. For instance, a study of $(KCl)_n$ and up sizes of $n=32$ been carried out in using methods including correlation effects MP2 We for and using the [*ab initio*]{} Ion (PI) model [@Ayuel; @Agu97]. objective of the present global characterization of the structure and other related of small neutral alkali halide clusters. To end we have carried out extensive PI calculations for the $(AX)_n$ $(n=1-10)$ identifying the isomers, | . Here we are interested in neutRal stoichiOmetrIc cLusTeRs of TypiCal materials wiTH ionIc bonding, that is $(AX)_n$ clusTers, wHeRE $A$ is AN aLkali And X a haLIdE ATom. as [*Ab IniTiO*]{} StUdies On tHese cluSters are coMpuTaTionally expeNSiVe, the first TheOretical calcUlaTions wErE baSEd on pAirWise iNteracTIon modEls [@Mar83; @Die85; @phI91]. meanwhILe, experIMEnTaliSts moved forward usINg SEveral techniquEs to prOdUCe AND inVesTigate thesE cLusteRS: particLE sPUTTerINg [@Cam81; @Bar82], where Rare-gas ions ARe uSed to bOmBarD A crystAl surFaCE, vaPor condensaTion In an inert-Gas atmOSphere [@ECH81; @Pfl85; @Pfl86], And lasEr vApoRizaTIoN oF a cRySTal SUrFacE [@con88; @twu90]. In the ExPaNding MoleCULAR jet ConDitiOns, clUsters undergo A raPid aND efFicieNt evaPoraTiVe cooLing, anD this LeAds to a cluster siZe diStributioN deTeRmiNeD almoST excluSivEly By the clUster stABilItY. tHE eVaporative cooling pRoCESs Leads to tHe so-caLLeD aBUndance mAgIc nUmbeRS As a reSult OF tHe longer Time thAT tHe Most staBlE clustErS reMaiN in thE Beam Before Decaying. alkalI Halide magic numBErs are often exPLaINEd IN terMs oF cuboid struCturES resEmblINg FraGMents Of the CrYStALline lattice, but otheR pOssibiLitieS like ring stacKings, or eveN MIXed strucTureS ExISt, which could be CompeTitive. NeveRTheless, oNly reCently thE possibilITY of detecTinG diFfeRenT ISoMers has emergeD IN driFt Tube expEriMents [@JaR95; @MaI96]. ThOse ExpErIments havE led to a rEnEwEd InTerEst in ISomerizaTiOn sTuDieS. In orDEr to diSentaNgle ThEsE IntErestinG PrOBLems, [*Ab InItio*]{} CalCuLatioNs [@AyUEl; @AGu97; @Wei92; @OcH94] provide aN idEAl coMpLeMent to tHe experimentaL sTudies, whicH aRe rEstricTED to chargEd species. For instance, a stUDy of $(KCl)_N$ anD $(LiF)_n$ ClusTers up to sIzeS of $n=32$ haS beEN carriEd out iN ref., uSiNg qUANtum-cHEMiCal MeThods incluDINg cOrrelAtIon eFfects aT the MP2 level. We have pERfoRmed calculatiOns For $(NAcL)_n$ And $(nAI)_N$ UsiNg THe [*aB INitio*]{} Perturbed IOn (PI) model [@AYuEL; @AGu97].
The objecTIve Of The presEnt papeR is to GIve a gloBal characTerizatioN oF the STRucTure and othEr relateD propertiES of smALl NeutrAl aLkali hAlIde ClustErs. To tHIs eNd we hAve carRiEd out eXtensIvE PI calcuLations for the $(AX)_n$ $(n=1-10)$ clusteRs, idenTifyiNg tHe most staBle ISomErs, | . Here we are interested i n neutralstoic hio met ri c cl uste rs of typicalm ater ials with ionic bondin g, th at is $ ( AX )_n$cluster s ,w h ere $ A$ is a n a lkali an d X a h alide atom . A s[*ab initio* ] {} studies o n t hese cluster s a re com pu tat i onall y e xpens ive, t h e firs t theoret ic a l calc u lations w er e ba sed on pairwise i n te r action models[@Mar8 3; @D i e 85; @P hi91]. Mea nw hile, experim e nt a l i sts moved forward using seve r altechni qu est o prod uce a nd inv estigate th eseclusters: parti c le sput t ering [ @Cam81 ; @ Bar 82], wh er e r ar e -ga s i ons are used to b om barda cr y s t a l su rfa ce,vapor condensation in ani ner t-gas atmo sphe re [@Ec h81; @ Pfl85 ;@Pfl86], and la servaporizat ion o f a c rysta l surfa ce[@C on88; @ Twu90]. Inth e e xp anding molecular j et c on ditions, clust e rs u n dergo ara pid and e ffici ente va porative cooli n g, a nd this l eads t oa c lus ter s i ze d istrib ution de termi n ed almost excl u sively by the cl u s te r sta bil ity. The ev apor a tive coo l in g p r ocess lead st ot he so-called abunda nc e magi c num bers as a res ult of the l o nger tim e th a tt he most stable clus ters remai n in thebeambefore d ecaying.A l kali hal ide ma gic nu m b er s are often e x p lain ed in ter msof cubo idstr uct ure sresemblin g fragme nt sof t hecryst a lline la tt ice ,but othe r possi bilit iesli ke rin g stack i ng s , orev en mix edst ructu rese xis t, whic h could b e c o mpet it iv e. Neve rtheless, onl yrecently t he po ssibil i t y of det ecting different isomer s has em erg ed in dri ft tube e xpe riment s [ @ Jar95; @Mai9 6]. T ho see x perim e n ts ha ve led to ar e new ed in te rest in iso merization studies . In order to dis ent angl e th ese in t ere st i ngp r oblems, [*ab in itio*]{} c al c ul ations [@A y uel ;@Agu97; @Wei92 ; @Oc h 94] pro vide an i deal comp le ment t o t he experim ental st udies, wh i ch ar e r estri cte d to c ha rge d spe cies.F orinsta nce, a s tudy o f $(K Cl )_n$ and $(LiF)_n$ clusters upto siz es of $n =32$ hasbee n ca rried out inref., usin g q uan tum-c hem i cal m etho d sinc l uding cor r elation e f fe cts a tthe MP2 lev e l . We have pe r formed cal culations for $(N a Cl)_n$ and $(N aI)_ n $ us ing the[* ab initio*]{}Per tu r b ed Ion ( PI ) model [@A yuel; @A gu 9 7].
The ob jectiv e of th e pr e sent p aper is to givea g lo b al char ac te r izatio n of t he str ucture ando t her related prop ertie s of sm a llneutr al alkali hali de cluster s. To thisend we hav e car ried ou textens ive P I calculat i ons for t he $( AX)_n$$( n=1- 10) $ clus ters , ident ifyi ng th e most st a b le is om e rs, | . Here_we are_interested in neutral stoichiometric_clusters of_typical_materials with_ionic_bonding, that is_$(AX)_n$ clusters, where_$A$ is an alkali_and X a_halide_atom. As [*ab initio*]{} studies on these clusters are computationally expensive, the first theoretical_calculations_were based_on_pairwise_interaction models [@Mar83; @Die85; @Phi91]._Meanwhile, experimentalists moved forward using_several techniques_to produce and investigate these clusters: particle sputtering_[@Cam81;_@Bar82], where rare-gas_ions are used to bombard a crystal surface, vapor_condensation in an inert-gas atmosphere [@Ech81;_@Pfl85; @Pfl86], and_laser_vaporization_of a crystal surface_[@Con88; @Twu90]. In the expanding molecular_jet conditions, clusters undergo a rapid_and efficient evaporative cooling, and this leads_to a cluster size distribution determined_almost exclusively by the cluster_stability. The_evaporative cooling process leads to_the so-called abundance_magic numbers_as a result_of the longer time that the_most stable clusters_remain in the beam before decaying._Alkali_halide magic numbers_are_often_explained in_terms of cuboid_structures_resembling fragments_of_the crystalline lattice, but other possibilities_like_ring stackings, or even mixed structures exist,_which could be competitive._Nevertheless,_only recently the possibility_of detecting different isomers has_emerged in drift tube experiments [@Jar95;_@Mai96]. Those_experiments have_led to a renewed interest in isomerization studies. In order to_disentangle these interesting problems, [*ab initio*]{}_calculations [@Ayuel; @Agu97; @Wei92;_@Och94] provide_an_ideal complement to_the_experimental studies,_which are restricted to charged species. For_instance, a_study of $(KCl)_n$ and $(LiF)_n$ clusters_up to sizes of_$n=32$_has been carried out in ref.,_using quantum-chemical methods including correlation effects_at the MP2 level. We_have_performed_calculations for $(NaCl)_n$ and $(NaI)_n$_using the [*ab initio*]{} Perturbed Ion_(PI) model [@Ayuel;_@Agu97].
The objective of the present paper is_to_give a global characterization of the_structure_and other related properties of small_neutral_alkali_halide clusters. To this end_we have carried out extensive PI_calculations for the $(AX)_n$ $(n=1-10)$ clusters, identifying the most_stable isomers, |
($\Delta
E_{Arr}^*\!=\!\Delta E_{Arr} / T_{MCT}$) for different model systems. The data of MLJ, MSS, BMLJ and BMLJ$_2$ are from this work (except the $\Delta E_{Arr}^*$ for BMLJ, that is from Ref. [@sastry_pisa]).
Having in mind that $\Delta E_{Arr}^* \! \simeq \! 2 \Delta E^*$ and that $\Delta E^*\! \simeq\!10$ (i.e. $\Delta E_{Arr}^*
\!\simeq\! 20$), we can try to analyze what is observed for other model potentials existing in the literature where the $D(T)$ has been determined. We found three different models for which a low temperature analysis of $D(T)$ has been performed [*via*]{} molecular dynamics: [*i)*]{} The BKS-silica model [@horbac], for which the values of $\Delta E_{Arr}^*$ are $16.2$ and $18.0$, for the self diffusion of $O$ and $Si$ respectively; [*ii)*]{} the Lewis and Wahnström ortho-terphenyl model [@mossa], for which the temperature dependence of the molecular center of mass diffusion coefficient at five different densities give values of reduced barrier energy $\Delta E_{Arr}^* \simeq 20 \div 28$ (except the lowest density that gives a value of about $10$); [*iii)*]{} The SPC/E-water model [@starr], for which one finds $\Delta E_{Arr}^* \simeq 40$. Table \[table2\] summarizes the known results on energy barrier heights estimated from saddles and from Arrhenius low-temperature dependence of diffusivity. The values for MLJ, MSS, BMLJ and BMLJ$_2$ are from the present work, except the $\Delta E^*_{Arr}$ for BMLJ that is obtained from Ref. [@sastry_pisa]. In future works we will try to determine the saddle-barriers $\Delta E^*$ for non-LJ systems (the last three systems in the Table), in order to have a better understanding of the diversity of the different landscapes. In conclusion, besides the case of water, the other systems seem to be in agreement with the findings of this work (the values of the reduced barrier energies are of the same order), evidencing a quite general universality of the observed relations | ($ \Delta
E_{Arr}^*\!=\!\Delta E_{Arr } / T_{MCT}$) for different model systems. The data of MLJ, MSS, BMLJ and BMLJ$_2 $ are from this workplace (except the $ \Delta E_{Arr}^*$ for BMLJ, that is from Ref. [ @sastry_pisa ]).
have in mind that $ \Delta E_{Arr}^ * \! \simeq \! 2 \Delta E^*$ and that $ \Delta E^*\! \simeq\!10 $ (i.e. $ \Delta E_{Arr}^ *
\!\simeq\! 20 $), we can try to analyze what is observed for early model potentials existing in the literature where the $ D(T)$ has been decide. We found three different model for which a low temperature analysis of $ D(T)$ has been do [ * via * ] { } molecular dynamics: [ * i) * ] { } The BKS - silica model [ @horbac ], for which the value of $ \Delta E_{Arr}^*$ are $ 16.2 $ and $ 18.0 $, for the self dissemination of $ O$ and $ Si$ respectively; [ * ii) * ] { } the Lewis and Wahnström ortho - terphenyl model [ @mossa ], for which the temperature dependence of the molecular center of mass diffusion coefficient at five unlike densities give values of decreased barrier energy $ \Delta E_{Arr}^ * \simeq 20 \div 28 $ (except the lowest density that gives a value of about $ 10 $); [ * iii) * ] { } The SPC / E - body of water model [ @starr ], for which one finds $ \Delta E_{Arr}^ * \simeq 40$. Table \[table2\ ] summarizes the known results on department of energy barrier heights estimated from saddles and from Arrhenius low - temperature dependence of diffusivity. The values for MLJ, MSS, BMLJ and BMLJ$_2 $ are from the present workplace, except the $ \Delta E^*_{Arr}$ for BMLJ that is obtained from Ref. [ @sastry_pisa ]. In future works we will try to determine the saddle - barriers $ \Delta E^*$ for non - LJ systems (the last three systems in the Table), in order to have a good understanding of the diversity of the different landscape. In conclusion, besides the case of water, the early systems look to be in agreement with the findings of this work (the value of the reduced barrier energies are of the same order), evidencing a quite cosmopolitan universality of the observed relations | ($\Depta
E_{Arr}^*\!=\!\Delta E_{Arr} / T_{MCU}$) for different model syvtems. Fhe data of MLJ, MSS, BMLJ and BMLJ$_2$ arx frim thus work (except the $\Delga E_{Arr}^*$ flr BMLJ, rhat us from Reh. [@sastry_insa]).
Hablng iu nind that $\Delts E_{Arr}^* \! \sikeq \! 2 \Delta E^*$ dna chat $\Delta E^*\! \simeq\!10$ (i.e. $\Delta E_{Arr}^*
\!\simez\! 20$), we csn try to analyzg whau if obavryed for other model potentials esisting in the litersture where the $D(T)$ has beej dehermined. We found hhree diffetsnt nodels for wfich a low temperature analysis of $D(T)$ has been perforoed [*vna*]{} moleculat syjdmics: [*i)*]{} The BKS-splica model [@horbac], fos which the values of $\Denta E_{Arr}^*$ are $16.2$ and $18.0$, for tie self diffusion of $O$ and $Si$ rzspectively; [*ii)*]{} the Leqiw and Wahtströo orgho-uer'hehyl mofel [@mossa], for which the remperature dependemcq of the molechlar cqneer of mass diffusion coefficient at fine djfferent densities give values of reduced batrier enerdy $\Delta E_{Arr}^* \simeq 20 \div 28$ (except the lowest densidy thet gircf q galue of about $10$); [*iii)*]{} The SPC/E-water model [@starr], rot ehich one findf $\Delta E_{Art}^* \dikgq 40$. Table \[tablg2\] summaxjzss the known resulhs on egergy barrier reignts estimated from saddles qnd from Arrkenuus low-temperature dependence of ciffuxivity. The values for MUJ, MAS, BMLJ and BMLJ$_2$ are wrom the present wogk, efcept tht $\Delta E^*_{Arr}$ for BILJ that ms obcained ffom Tef. [@sasery_pisa]. In futuva works we will trj to beterkine the swddle-barriers $\Delta E^*$ for non-LJ systems (the lssd tvree syscems ik the Table), in jrder to have c better undefstanding kf the viversity of the differend landscapes. Mn conclufion, besudes thd case of water, the otheg fystwms seem to be in egregmsnt with the fiubibgs of this worl (tfe dapuxs of dhe reduced tarrker rnergkes are of bhe samr order), evidencing a quife general universsllty of thg observeq relations | ($\Delta E_{Arr}^*\!=\!\Delta E_{Arr} / T_{MCT}$) for different The of MLJ, BMLJ and BMLJ$_2$ the E_{Arr}^*$ for BMLJ, is from Ref. Having in mind that $\Delta E_{Arr}^* \simeq \! 2 \Delta E^*$ and that $\Delta E^*\! \simeq\!10$ (i.e. $\Delta E_{Arr}^* 20$), we can try to analyze what is observed for other model potentials in literature the has been determined. We found three different models for which a low temperature analysis of $D(T)$ been performed [*via*]{} molecular dynamics: [*i)*]{} The BKS-silica [@horbac], for which the of $\Delta E_{Arr}^*$ are $16.2$ $18.0$, the self of and respectively; [*ii)*]{} the and Wahnström ortho-terphenyl model [@mossa], for which the temperature dependence of the molecular center of mass diffusion at five give values reduced energy E_{Arr}^* \simeq 20 (except the lowest density that gives about $10$); [*iii)*]{} The SPC/E-water model [@starr], for one finds E_{Arr}^* \simeq 40$. Table \[table2\] summarizes known results on energy barrier heights estimated from and from Arrhenius low-temperature dependence of diffusivity. The values for MLJ, MSS, BMLJ and BMLJ$_2$ the present work, except $\Delta E^*_{Arr}$ for that obtained Ref. In future we will try to determine the saddle-barriers $\Delta E^*$ for non-LJ (the last three systems in the Table), in order to better of the diversity the different landscapes. In besides case of water, the seem be the of work (the values of reduced barrier energies are of same order), evidencing a observed relations | ($\Delta
E_{Arr}^*\!=\!\Delta E_{Arr} / T_{MCT}$) for Different mOdel sYstEms. thE datA of MlJ, MSS, BMLJ and BMlj$_2$ are From this work (except the $\DElta E_{arR}^*$ For BmlJ, That iS from ReF. [@SaSTRy_pIsA]).
HAviNg IN mInd thAt $\DElta E_{ArR}^* \! \simeq \! 2 \DeltA E^*$ aNd That $\Delta E^*\! \siMEq\!10$ (I.e. $\Delta E_{ArR}^*
\!\siMeq\! 20$), we can try tO anAlyze wHaT is OBservEd fOr othEr modeL PotentIals existInG In the lITeraturE WHeRe thE $D(T)$ has been determiNEd. wE found three difFerent MoDElS FOr wHicH a low tempeRaTure aNAlysis oF $d(T)$ HAS BeeN Performed [*via*]{} mOlecular dynAMicS: [*i)*]{} The BkS-SilICa modeL [@horbAc], FOr wHich the valuEs of $\delta E_{Arr}^*$ Are $16.2$ and $18.0$, FOr the seLF diffusIon of $O$ And $si$ rEspeCTiVeLy; [*iI)*]{} tHE LeWIs And wAhnStröm ortHo-TeRphenYl moDEL [@MOssa], For WhicH the tEmperature depEndEnce OF thE moleCular CentEr Of masS diffuSion cOeFficient at five dIffeRent densiTieS gIve VaLues oF ReduceD baRriEr energY $\Delta E_{aRr}^* \sImEQ 20 \DIv 28$ (Except the lowest denSiTY ThAt gives a Value oF AbOuT $10$); [*Iii)*]{} The SPc/E-WatEr moDEL [@starR], for WHiCh one finDs $\DeltA e_{ARr}^* \Simeq 40$. TaBlE \[table2\] SuMmaRizEs the KNown ResultS on energY barrIEr heights estimATed from saddleS AnD FRoM arrhEniUs low-temperAturE DepeNdenCE oF diFFusivIty. ThE vALuES for MLJ, MSS, BMLJ and BMlJ$_2$ Are froM the pResent work, excEpt the $\DeltA e^*_{aRr}$ for BMLj thaT Is OBtained from Ref. [@SastrY_pisa]. In futURe works wE will Try to detErmine the SADdle-barrIerS $\DeLta e^*$ foR NOn-lJ systems (the lAST thrEe Systems In tHe Table), In oRdeR to HavE a Better undErstandiNg Of ThE dIveRsity OF the diffErEnt LaNdsCapes. iN conclUsion, BesiDeS tHE caSe of watER, tHE OtheR sYsTems SeeM tO be in AgreEMenT with thE findings Of tHIs woRk (ThE values Of the reduced bArRier energiEs Are Of the sAME order), evIdencing a quite general unIVersaliTy oF the oBserVed relatiOns | ($\Delta
E_{Arr}^*\!=\! \Delta E_{ Arr}/ T _{M CT }$)fordifferent mode l sys tems. The data of MLJ, MSS, B M LJ a n dBMLJ$ _2$ are fr o m th is w ork ( e xc ept t he$\Delta E_{Arr}^* $ f or BMLJ, thati sfrom Ref.[@s astry_pisa]) .
Having i n m i nd th at$\Del ta E_{ A rr}^*\! \simeq \ ! 2 \De l ta E^*$ a nd tha t $\Delta E^*\! \ s im e q\!10$ (i.e. $ \Delta E _ {A r r }^*
\! \simeq\! 2 0$ ), we can try to a n aly z e what is obs erved for o t her model p ote n tialsexist in g in the litera ture where th e $D(T ) $ has b e en dete rmined . W e f ound th re e d if f ere n tmod e lsfor whic halow t empe r a t u re a nal ysis of $ D(T)$ has bee n p erfo r med [*vi a*]{} mol ec ulardynami cs: [ *i )*]{} The BKS-s ilic a model [ @ho rb ac] ,for w h ich th e v alu es of $ \DeltaE _{A rr } ^ * $are $16.2$ and $18 .0 $ , f or the s elf di f fu si o n of $O$ a nd$Si$ r espec tive l y; [*ii)*] {} the Le wi s and W ah nström o rth o-t erphe n yl m odel [ @mossa], forw hich the tempe r ature depende n ce o ft he m ole cular cente r of mass dif f us ion coeff icien ta tf ive different densi ti es giv e val ues of reduce d barriere n e rgy $\De ltaE _{ A rr}^* \simeq 2 0 \di v 28$ (exc e pt the l owest density that giv e s a value of ab out $1 0 $ ); [*iii)*]{} T h e SPC /E -watermod el [@st arr ],for wh ic h one fin ds $\Del ta E _{ Ar r}^ * \si m eq 40$.Ta ble \ [ta ble2\ ] summa rizes the k no w n r esultso ne n ergy b ar rier he ig hts e stim a ted from s addles an d f r om A rr he nius lo w-temperature d ependenceof di ffusiv i t y. The v alues for MLJ, MSS, BML J and BM LJ$ _2$ a re f rom the p res ent wo rk, except the $ \Delt aE^* _ { Arr}$ f or BM LJ that is o b t ain ed fr om Ref . [@sas try_pisa]. In futu r e w orks we willtry tod e te rmi n et hesa d dle - b arriers $\Delta E^*$ forno n -L J systems( the l ast thr ee syst ems i n the Ta ble), inorder toha ve a b ett er underst anding o f the div e rsity of thedif ferent l and scape s. Inc onc lusio n, bes id es the case o f water, the other systems seem to be in a gre ement wit h t h e f indings o f th is work (t heval ues o f t h e red uced ba rri e r ene rgie s are of t h esam e or der), evide n c i nga qui teg eneral uni versality of theo bserved relati ons | ($\Delta
_ E_{Arr}^*\!=\!\Delta_E_{Arr} / T_{MCT}$) for_different model_systems._The data_of_MLJ, MSS, BMLJ_and BMLJ$_2$ are_from this work (except_the $\Delta E_{Arr}^*$_for_BMLJ, that is from Ref. [@sastry_pisa]).
Having in mind that $\Delta E_{Arr}^* \! \simeq \!_2_\Delta E^*$_and_that_$\Delta E^*\! \simeq\!10$ (i.e. $\Delta_E_{Arr}^*
\!\simeq\! 20$), we can try_to analyze_what is observed for other model potentials existing_in_the literature where_the $D(T)$ has been determined. We found three different_models for which a low temperature_analysis of $D(T)$_has_been_performed [*via*]{} molecular dynamics:_[*i)*]{} The BKS-silica model [@horbac], for_which the values of $\Delta E_{Arr}^*$_are $16.2$ and $18.0$, for the self_diffusion of $O$ and $Si$ respectively;_[*ii)*]{} the Lewis and Wahnström_ortho-terphenyl model_[@mossa], for which the temperature_dependence of the_molecular center_of mass diffusion_coefficient at five different densities give_values of reduced_barrier energy $\Delta E_{Arr}^* \simeq 20_\div_28$ (except the_lowest_density_that gives_a value of_about_$10$); [*iii)*]{}_The_SPC/E-water model [@starr], for which one_finds_$\Delta E_{Arr}^* \simeq 40$. Table \[table2\] summarizes_the known results on_energy_barrier heights estimated from_saddles and from Arrhenius low-temperature_dependence of diffusivity. The values for_MLJ, MSS,_BMLJ and_BMLJ$_2$ are from the present work, except the $\Delta E^*_{Arr}$ for_BMLJ that is obtained from Ref._[@sastry_pisa]. In future works_we will_try_to determine the_saddle-barriers_$\Delta E^*$_for non-LJ systems (the last three systems_in the_Table), in order to have a_better understanding of the_diversity_of the different landscapes. In conclusion,_besides the case of water, the_other systems seem to be_in_agreement_with the findings of this_work (the values of the reduced_barrier energies are_of the same order), evidencing a quite_general_universality of the observed relations |
Rudolf T, Kant C, Mayr F and Loidl A 2008 [*Phys. Rev. B*]{} [**77**]{}(2) 024421 <http://link.aps.org/doi/10.1103/PhysRevB.77.024421>
Yamashita Y and Ueda K 2000 [*Physical Review Letters*]{} [**85**]{} 4960
Rudolf T, Kant C, Mayr F, Hemberger J, Tsurkan V and Loidl A 2007 [*New Journal of Physics*]{} [**9**]{} 76
Schaack G 1977 [*Zeitschrift f[ü]{}r Physik B Condensed Matter*]{} [**26**]{} 49–58
Klupp G, Matus P, Kamar[á]{}s K, Ganin A Y, McLennan A, Rosseinsky M J, Takabayashi Y, McDonald M T and Prassides K 2012 [*Nature Communications*]{} [**3**]{} 912
Fleury P and Loudon R 1968 [*Physical Review*]{} [**166**]{} 514
Reiter G 1976 [*Physical Review B*]{} [**13**]{} 169
Kaplan T and Mahanti S 2006 [*Physics of manganites*]{} (Springer Science & Business Media)
Choi K Y, Zvyagin S, Cao G and Lemmens P 2004 [*Physical Review B*]{} [ **69**]{} 104421
Lemmens P, G[ü]{}ntherodt G and Gros C 2003 [*Physics Reports*]{} [**375**]{} 1–103
Olivero J and Longbothum R 1977 [*Journal of Quantitative Spectroscopy and Radiative Transfer*]{} [**17**]{} 233–236
Men[é]{}ndez J and Cardona M 1984 [*Physical Review B*]{} [**29**]{} 2051
Kim K, Gu J, Choi H, Park G and Noh T 1996 [*Physical Review Letters*]{} [ **77**]{} 1877
Ulrich C, Khaliullin G, Guennou M, Roth H, Lorenz T and Keimer B 2015 [ *Physical review letters*]{} [**115**]{} 156403
Balkanski M, Wallis R F and Haro E 1983 [*Physical Review B*]{} [**28**]{}(4) 1928–1934 <http://link.aps.org/doi/10.1103/PhysRevB.28.1928> | Rudolf T, Kant C, Mayr F and Loidl A 2008 [ * Phys. Rev. B * ] { } [ * * 77**]{}(2) 024421 < http://link.aps.org/doi/10.1103/PhysRevB.77.024421 >
Yamashita Y and Ueda K 2000 [ * Physical Review Letters * ] { } [ * * 85 * * ] { } 4960
Rudolf T, Kant C, Mayr F, Hemberger J, Tsurkan V and Loidl A 2007 [ * New Journal of Physics * ] { } [ * * 9 * * ] { } 76
Schaack G 1977 [ * Zeitschrift f[ü]{}r Physik B Condensed Matter * ] { } [ * * 26 * * ] { } 49–58
Klupp G, Matus P, Kamar[á]{}s K, Ganin A Y, McLennan A, Rosseinsky M J, Takabayashi Y, McDonald M T and Prassides K 2012 [ * Nature Communications * ] { } [ * * 3 * * ] { } 912
Fleury P and Loudon R 1968 [ * Physical Review * ] { } [ * * 166 * * ] { } 514
Reiter G 1976 [ * Physical Review B * ] { } [ * * 13 * * ] { } 169
Kaplan T and Mahanti S 2006 [ * Physics of manganites * ] { } (Springer Science & Business Media)
Choi K Y, Zvyagin S, Cao G and Lemmens P 2004 [ * Physical Review B * ] { } [ * * 69 * * ] { } 104421
Lemmens P, G[ü]{}ntherodt G and Gros C 2003 [ * Physics Reports * ] { } [ * * 375 * * ] { } 1–103
Olivero J and Longbothum R 1977 [ * Journal of Quantitative Spectroscopy and Radiative Transfer * ] { } [ * * 17 * * ] { } 233–236
Men[é]{}ndez J and Cardona M 1984 [ * Physical Review B * ] { } [ * * 29 * * ] { } 2051
Kim K, Gu J, Choi H, Park G and Noh T 1996 [ * Physical Review Letters * ] { } [ * * 77 * * ] { } 1877
Ulrich C, Khaliullin G, Guennou M, Roth H, Lorenz T and Keimer B 2015 [ * forcible revue letters * ] { } [ * * 115 * * ] { } 156403
Balkanski M, Wallis R F and Haro E 1983 [ * Physical Review B * ] { } [ * * 28**]{}(4) 1928–1934 < http://link.aps.org/doi/10.1103/PhysRevB.28.1928 > | Rudllf T, Kant C, Mayr F and Uoidl A 2008 [*Phys. Rgv. B*]{} [**77**]{}(2) 024421 <hvtp://link.zps.org/dok/10.1103/PhysRevB.77.024421>
Yamashita Y and Uede K 2000 [*Physucal Review Letters*]{} [**85**]{} 4960
Rjdolf T, Kwnt C, Matr F, Yemberger O, Tsurkak V ahf Londo A 2007 [*New Journsl of Physhcs*]{} [**9**]{} 76
Schaack G 1977 [*Xentschrift f[ü]{}r Physik B Condensed Mattqr*]{} [**26**]{} 49–58
Klulp G, Matus P, Kamwr[á]{}s L, Ganjn A Y, McLennan A, Rosseinsky M J, Takzbayashp Y, McDonald M T amd Prassides K 2012 [*Nature Comlunifations*]{} [**3**]{} 912
Fleury P wnd Loudon T 1968 [*Prtsical Revied*]{} [**166**]{} 514
Reiter G 1976 [*Physical Teview B*]{} [**13**]{} 169
Kaplan T and Mahanti S 2006 [*Phyxics of mabgqnihgs*]{} (Springer Wcienbe & Business Media)
Choh K Y, Zvuagin S, Cao G snd Lenmens P 2004 [*Physical Revmew B*]{} [ **69**]{} 104421
Lemmens P, G[ü]{}njherodt G dnb Gros C 2003 [*Physics Repirrs*]{} [**375**]{} 1–103
Oniveso J qnd Lohguotgum R 1977 [*Jonrnal of Quzntitative Wpectroscopy and Raciwnove Transfer*]{} [**17**]{} 233–236
Men[é]{}nqes J and Cardona M 1984 [*Physical Review B*]{} [**29**]{} 2051
Kpm K, Gu J, Choi H, Park G and Noh T 1996 [*Physical Revigw Letters*]{} [ **77**]{} 1877
Ulrich C, Khaliullin G, Guennou M, Roth H, Lorenz D and Yeinev B 2015 [ *Pjysical review letters*]{} [**115**]{} 156403
Balkanski M, Wallis R F znc Maro E 1983 [*Physical Review B*]{} [**28**]{}(4) 1928–1934 <hhtl://jink.aps.org/doi/10.1103/KhysRevY.28.1928> | Rudolf T, Kant C, Mayr F and 2008 Rev. B*]{} 024421 <http://link.aps.org/doi/10.1103/PhysRevB.77.024421> Yamashita [*Physical Letters*]{} [**85**]{} 4960 T, Kant C, F, Hemberger J, Tsurkan V and A 2007 [*New Journal of Physics*]{} [**9**]{} 76 Schaack G 1977 [*Zeitschrift f[ü]{}r B Condensed Matter*]{} [**26**]{} 49–58 Klupp G, Matus P, Kamar[á]{}s K, Ganin A McLennan Rosseinsky J, Y, McDonald M T and Prassides K 2012 [*Nature Communications*]{} [**3**]{} 912 Fleury P and Loudon 1968 [*Physical Review*]{} [**166**]{} 514 Reiter G 1976 Review B*]{} [**13**]{} 169 T and Mahanti S 2006 of (Springer Science Business Choi Y, Zvyagin S, G and Lemmens P 2004 [*Physical Review B*]{} [ **69**]{} 104421 Lemmens P, G[ü]{}ntherodt G and Gros 2003 [*Physics 1–103 Olivero and R [*Journal of Quantitative Radiative Transfer*]{} [**17**]{} 233–236 Men[é]{}ndez J 1984 [*Physical Review B*]{} [**29**]{} 2051 Kim K, J, Choi Park G and Noh T 1996 Review Letters*]{} [ **77**]{} 1877 Ulrich C, Khaliullin Guennou M, Roth H, Lorenz T and Keimer B 2015 [ *Physical review letters*]{} [**115**]{} M, Wallis R F Haro E 1983 Review [**28**]{}(4) <http://link.aps.org/doi/10.1103/PhysRevB.28.1928> | Rudolf T, Kant C, Mayr F and Loidl a 2008 [*Phys. Rev. B*]{} [**77**]{}(2) 024421 <hTtp://liNk.aPs.oRg/Doi/10.1103/PHysREvB.77.024421>
Yamashita Y aND UedA K 2000 [*Physical Review LetterS*]{} [**85**]{} 4960
RudoLf t, kant c, maYr F, HeMberger j, tsURKan v aNd loiDl a 2007 [*neW JourNal Of PhysiCs*]{} [**9**]{} 76
Schaack G 1977 [*zeiTsChrift f[ü]{}r PhySIk b Condensed matTer*]{} [**26**]{} 49–58
Klupp G, MatUs P, kamar[á]{}S K, ganIN A Y, MclenNan A, ROsseinSKy M J, TaKabayashi y, MCdonald m t and PraSSIdEs K 2012 [*NAture CommunicatioNS*]{} [**3**]{} 912
FLEury P and Loudon r 1968 [*PhysiCaL reVIEw*]{} [**166**]{} 514
REitEr G 1976 [*PhysicaL REview b*]{} [**13**]{} 169
kaplan T ANd mAHAntI s 2006 [*Physics of manGanites*]{} (SpriNGer sciencE & BUsiNEss MedIa)
ChoI K y, zvyAgin S, Cao G anD LemMens P 2004 [*PhysIcal ReVIew B*]{} [ **69**]{} 104421
LemMEns P, G[ü]{}nTherodT G aNd GRos C 2003 [*pHySiCs REpORts*]{} [**375**]{} 1–103
oLiVerO j anD LongbotHuM R 1977 [*journAl of qUANTitaTivE SpeCtrosCopy and RadiatIve tranSFer*]{} [**17**]{} 233–236
men[é]{}nDez J aNd CaRdOna M 1984 [*PHysicaL ReviEw b*]{} [**29**]{} 2051
Kim K, Gu J, Choi H, PaRk G aNd Noh T 1996 [*PhySicAl revIeW LettERs*]{} [ **77**]{} 1877
UlriCh C, khaLiullin g, GuennoU m, RoTh h, lOReNz T and Keimer B 2015 [ *PhysiCaL REvIew letteRs*]{} [**115**]{} 156403
BalkANsKi m, wallis R F AnD HaRo E 1983 [*PHYSical reviEW B*]{} [**28**]{}(4) 1928–1934 <Http://link.Aps.org/DOi/10.1103/phYsRevB.28.1928> | Rudolf T, Kant C, Mayr F a nd Loidl A 2008 [* Phy s. Rev . B* ]{} [**77**]{} ( 2) 0 24421 <http://link.aps .org/ do i /10. 1 10 3/Phy sRevB.7 7 .0 2 4 421 >
Y ama sh i ta Y an d U eda K 2 000 [*Phys ica lReview Lette r s* ]{} [**85* *]{ } 4960
Rudo lfT, Kan tC,M ayr F , H ember ger J, Tsurka n V and L oi d l A 20 0 7 [*New J ou rnal of Physics*]{} [ * *9 * *]{} 76
Schaa ck G 1 97 7 [ * Z eit sch rift f[ü]{ }r Phys i k B Con d en s e d Ma t ter*]{} [**26 **]{} 49–58
Kl upp G, M atu s P, Ka mar[á ]{ } s K , Ganin A Y , Mc Lennan A, Rosse i nsky MJ , Takab ayashi Y, Mc Dona l dMT a nd Pra s si des K 2 012 [*Na tu re Comm unic a t i o ns*] {}[**3 **]{} 912
FleuryP a nd L o udo n R 1 968 [ *Phy si cal R eview* ]{} [ ** 166**]{} 514
R eite r G 1976[*P hy sic al Revi e w B*]{ } [ **1 3**]{}169
Ka p lan T a n dMahanti S 2006 [*P hy s i cs of mang anites * ]{ }( Springer S cie nce& Busin essM ed ia)
Cho i K Y, Zv ya gin S,Ca o G an dLem men s P 2 0 04 [ *Physi cal Revi ew B* ] {} [ **69**]{} 104421
Lemme n sP , G [ ü]{} nth erodt G and Gro s C 2 003[ *P hys i cs Re ports *] { }[ **375**]{} 1–103
O li vero J andLongbothum R1977 [*Jou r n a l of Qua ntit a ti v e Spectroscopy andRadiativeT ransfer* ]{} [ **17**]{ } 233–236 Men[é]{} nde z J an d C a r do na M 1984 [*P h y sica lReviewB*] {} [**2 9** ]{} 20 51
K im K, GuJ, ChoiH, P ar kG a nd No h T 1996[* Phy si cal Revi e w Lett ers*] {} [ * *7 7 **] {} 1877
U l r ichC, K hali ull in G, G uenn o u M , RothH, Lorenz Ta nd K ei me r B 201 5 [ *Physical r eview lett er s*] {} [** 1 1 5**]{} 1 56403
Balkanski M, Wal l is R Fand Haro E 1 983 [*Phy sic al Rev iew B*]{}[**28* *]{}( 4) 19 2 8 –1934 < ht tp: // link.aps.o r g /do i/10. 11 03/P hysRevB .28.1928> | Rudolf T,_Kant C,_Mayr F and Loidl_A 2008_[*Phys._Rev. B*]{}_[**77**]{}(2)_024421 <http://link.aps.org/doi/10.1103/PhysRevB.77.024421>
Yamashita Y_and Ueda K_2000 [*Physical Review Letters*]{}_[**85**]{} 4960
Rudolf T,_Kant_C, Mayr F, Hemberger J, Tsurkan V and Loidl A 2007 [*New Journal of_Physics*]{}_[**9**]{} 76
Schaack_G_1977_[*Zeitschrift f[ü]{}r Physik B Condensed_Matter*]{} [**26**]{} 49–58
Klupp G, Matus_P, Kamar[á]{}s_K, Ganin A Y, McLennan A, Rosseinsky M J, Takabayashi_Y,_McDonald M T and_Prassides K 2012 [*Nature Communications*]{} [**3**]{} 912
Fleury P and_Loudon R 1968 [*Physical Review*]{} [**166**]{}_514
Reiter G 1976_[*Physical_Review_B*]{} [**13**]{} 169
Kaplan T_and Mahanti S 2006 [*Physics of_manganites*]{} (Springer Science & Business Media)
Choi_K Y, Zvyagin S, Cao G and Lemmens_P 2004 [*Physical Review B*]{} [_**69**]{} 104421
Lemmens P, G[ü]{}ntherodt G_and Gros_C 2003 [*Physics Reports*]{} [**375**]{}_1–103
Olivero J and_Longbothum R_1977 [*Journal of_Quantitative Spectroscopy and Radiative Transfer*]{} [**17**]{}_233–236
Men[é]{}ndez J and_Cardona M 1984 [*Physical Review B*]{}_[**29**]{}_2051
Kim K, Gu_J,_Choi_H, Park_G and Noh_T_1996 [*Physical_Review_Letters*]{} [ **77**]{} 1877
Ulrich C, Khaliullin_G,_Guennou M, Roth H, Lorenz T and_Keimer B 2015 [_*Physical_review letters*]{} [**115**]{} 156403
Balkanski_M, Wallis R F and Haro_E 1983 [*Physical Review B*]{} [**28**]{}(4)_1928–1934 <http://link.aps.org/doi/10.1103/PhysRevB.28.1928> |
, V. Mannings, A. Boss, & S. Russell (Tucson:Univ. Arizona Press), 59
Ballesteros-Paredes, J. 2003. In “From Observations to Self-Consistent Modeling of the Interstellar Medium”. Ed. M. Avillez & D. Breitschwerdt (Kluwer), in press.
Ballesteros-Paredes, J., Hartmann, L., & V[' a]{}zquez-Semadeni, E.1999b,, 527, 285
Ballesteros-Paredes, J. & Mac Low, M. 2002,, 570, 734
Ballesteros-Paredes, J., V[' a]{}zquez-Semadeni, E., & Scalo, J.1999a,, 515, 286
Barranco, J. A. & Goodman, A. A. 1998,, 504, 207
Bate M. R., & Burkert, A. 1997,, 288, 1060
Bate, M. R., Bonell, I.A. ., & Price, N. M. 1995., 277, 362
Benz, W. 1990, in The Numerical Modeling of Nonlinear Stellar Pulsations, ed. J. R. Buchler, p. 269, Kluwer Academic Publishers, The Netherlands
Bonnor, W. B. 1956,, 116, 351
Caselli, P., Benson, P. J., Myers, P. C., & Tafalla, M. 2002,, 572, 238
Crutcher, R. M. 2004, in Magnetic Fields and Star Formation: Theory versus Observations, eds. Ana I Gomez de Castro et al. (Dordrecht: Kluwer), in press
Ebert, R. 1955, Zeitschrift f[ü]{}r Astrophysik, 36, 222
Elmegreen, B. G. 1993,, 419, L29
Gammie, C. F., Lin, Y., Stone, J. M., & Ostriker, E. C. 2003,, 592, 203
Goodman, A. A | , V. Mannings, A. Boss, & S. Russell (Tucson: Univ. Arizona Press), 59
Ballesteros - Paredes, J. 2003. In “ From Observations to Self - Consistent Modeling of the Interstellar Medium ”. Ed. M. Avillez & D. Breitschwerdt (Kluwer), in press.
Ballesteros - Paredes, J., Hartmann, L., & V [' a]{}zquez - Semadeni, E.1999b, , 527, 285
Ballesteros - Paredes, J. & Mac Low, M. 2002, , 570, 734
Ballesteros - Paredes, J., V [' a]{}zquez - Semadeni, E., & Scalo, J.1999a, , 515, 286
Barranco, J. A. & Goodman, A. A. 1998, , 504, 207
Bate M. R., & Burkert, A. 1997, , 288, 1060
Bate, M. R., Bonell, I.A. . , & Price, N. M. 1995. , 277, 362
Benz, W. 1990, in The Numerical Modeling of Nonlinear Stellar Pulsations, ed. J. R. Buchler, p. 269, Kluwer Academic Publishers, The Netherlands
Bonnor, W. B. 1956, , 116, 351
Caselli, P., Benson, P. J., Myers, P. C., & Tafalla, M. 2002, , 572, 238
Crutcher, R. M. 2004, in Magnetic Fields and Star Formation: Theory versus Observations, eds. Ana I Gomez de Castro et al. (Dordrecht: Kluwer), in wardrobe
Ebert, R. 1955, Zeitschrift f[ü]{}r Astrophysik, 36, 222
Elmegreen, B. G. 1993, , 419, L29
Gammie, C. F., Lin, Y., Stone, J. M., & Ostriker, E. C. 2003, , 592, 203
Goodman, A. deoxyadenosine monophosphate | , V. Lannings, A. Boss, & S. Russeul (Tucson:Univ. Atizona Pcess), 59
Bamlesteror-Paredes, J. 2003. In “From Observatmons to Stjf-Consistent Modelkng of thv Interstwllac Medium”. Ed. M. Avillxa & D. Brelcschwsvdt (Knnwer), in press.
Baklesteros-Pdredes, J., Hartmdnv, P., & V[' a]{}zquez-Semadeni, E.1999b,, 527, 285
Ballesteros-[aredes, J. & Mac Low, M. 2002,, 570, 734
Bajlesuerjs-Padvdts, J., V[' a]{}zquez-Semadeni, E., & Scalo, J.1999a,, 515, 286
Barraico, J. A. & Goodman, S. A. 1998,, 504, 207
Bate M. R., & Burkert, A. 1997,, 288, 1060
Bwte, L. R., Bonell, I.A. ., & Pricf, N. M. 1995., 277, 362
Benz, W. 1990, yb The Numerizal Modeling of Nonlingar Stellar Pulsations, ed. J. R. Buchuer, p. 269, Kluwer Acqdwmif Publishers, The Getherlands
Bonnor, W. B. 1956,, 116, 351
Caselki, P., Benson, P. J., Myxrs, P. C., & Tafalla, M. 2002,, 572, 238
Crutchxr, R. M. 2004, in Magnetic Figlds and Sdax Formation: Theory veesys Obvervdtiovw, eas. Zne I Gomez de Castro et al. (Dordrecyt: Kluwer), in press
Enewn, R. 1955, Zeitschrirt f[ü]{}r Wserophysik, 36, 222
Elmegreen, B. G. 1993,, 419, L29
Gammie, C. F., Lin, Y., Atone, J. M., & Ostriker, E. C. 2003,, 592, 203
Toodman, A. A | , V. Mannings, A. Boss, & S. Arizona 59 Ballesteros-Paredes, 2003. In “From the Medium”. Ed. M. & D. Breitschwerdt in press. Ballesteros-Paredes, J., Hartmann, L., V[' a]{}zquez-Semadeni, E.1999b,, 527, 285 Ballesteros-Paredes, J. & Mac Low, M. 2002,, 570, Ballesteros-Paredes, J., V[' a]{}zquez-Semadeni, E., & Scalo, J.1999a,, 515, 286 Barranco, J. A. Goodman, A. 504, Bate M. R., & Burkert, A. 1997,, 288, 1060 Bate, M. R., Bonell, I.A. ., & N. M. 1995., 277, 362 Benz, W. 1990, The Numerical Modeling of Stellar Pulsations, ed. J. R. p. Kluwer Academic The Bonnor, B. 1956,, 116, Caselli, P., Benson, P. J., Myers, P. C., & Tafalla, M. 2002,, 572, 238 Crutcher, R. M. in Magnetic Star Formation: versus eds. I Gomez de al. (Dordrecht: Kluwer), in press Ebert, f[ü]{}r Astrophysik, 36, 222 Elmegreen, B. G. 1993,, L29 Gammie, F., Lin, Y., Stone, J. M., Ostriker, E. C. 2003,, 592, 203 Goodman, A. | , V. Mannings, A. Boss, & S. Russell (TucSon:Univ. AriZona PResS), 59
BaLlEsteRos-PAredes, J. 2003. In “From OBServAtions to Self-Consistent modelInG Of thE inTerstEllar MeDIuM”. eD. M. AViLlEz & D. brEItSchweRdt (kluwer), iN press.
BallEstErOs-Paredes, J., HaRTmAnn, L., & V[' a]{}zqueZ-SeMadeni, E.1999b,, 527, 285
BallEstEros-PaReDes, j. & mac LoW, M. 2002,, 570, 734
BAllesTeros-PARedes, J., v[' a]{}zquez-SeMaDEni, E., & ScALo, J.1999a,, 515, 286
BarRANcO, J. A. & GOodman, A. A. 1998,, 504, 207
Bate M. R., & BurKErT, a. 1997,, 288, 1060
Bate, M. R., Bonell, I.a. ., & Price, n. M. 1995., 277, 362
bEnZ, w. 1990, In THe NUmerical MoDeLing oF nonlineAR STELLar pUlsations, ed. J. R. buchler, p. 269, KluWEr ACademiC PUblIShers, THe NetHeRLanDs
Bonnor, W. B. 1956,, 116, 351
CAselLi, P., Benson, p. J., MyerS, p. C., & TafalLA, M. 2002,, 572, 238
CrutcHer, R. M. 2004, iN MaGneTic FIElDs And stAR FoRMaTioN: theOry versuS OBsErvatIons, EDS. aNa I GOmeZ de CAstro Et al. (Dordrecht: kluWer), iN PreSs
EbeRt, R. 1955, ZeItscHrIft f[ü]{}R AstroPhysiK, 36, 222
ELmegreen, B. G. 1993,, 419, L29
GammIe, C. F., lin, Y., Stone, j. M., & OStRikEr, e. C. 2003,, 592, 203
GooDMan, A. A | , V. Mannings, A. Boss, &S. Russell (Tuc son :Un iv . Ar izon a Press), 59
B alle steros-Paredes, J. 200 3. In “ F romO bs ervat ions to Se l f -Co ns is ten tM od eling of the In terstellar Me di um”. Ed. M.A vi llez & D.Bre itschwerdt ( Klu wer),in pr e ss.
Bal leste ros-Pa r edes,J., Hartm an n , L.,& V[' a] { } zq uez- Semadeni, E.1999b , ,5 27, 285
Balle steros -P a re d e s,J.& Mac Low, M . 200 2 ,, 570, 73 4 Bal l esteros-Pared es, J., V[' a]{ }zquez -S ema d eni, E ., &Sc a lo, J.1999a,,515, 286
Bar ranco, J. A. & Goodman , A. A . 1 998 ,, 5 0 4, 2 07
B a teM .R., & B urkert,A. 1 997,, 288 , 1 0 60
Bat e, M . R., Bonell, I.A. ., & P r ice , N.M. 19 95., 2 77, 3 62
Be nz, W .1990, in The Nu meri cal Model ing o f N on linea r Stell arPul sations , ed. J . R. B u c h le r, p. 269, KluwerAc a d em ic Publi shers, Th eN etherlan ds
B onno r , W. B . 19 5 6, , 116, 3 51
Ca s el li , P., B en son, P .J., My ers,P . C. , & Ta falla, M . 200 2 ,, 572, 238
C r utcher, R. M. 20 0 4 ,i n Ma gne tic FieldsandS tarForm a ti on: Theor y ver su s O b servations, eds. An aI Gome z deCastro et al. (Dordrech t : Kluwer), inp re s s
Ebert, R. 1 955,Zeitschrif t f[ü]{}r Astr ophysik, 36, 222E lmegreen , B . G . 1 993 , , 4 19, L29
Gamm i e , C. F ., Lin, Y. , Stone , J . M .,& O st riker, E. C. 2003 ,, 5 92 ,203
Goo d man, A.A | , V._Mannings, A._Boss, & S. Russell_(Tucson:Univ. Arizona_Press),_59
Ballesteros-Paredes, J._2003._In “From Observations_to Self-Consistent Modeling_of the Interstellar Medium”. Ed. M. Avillez_& D. Breitschwerdt (Kluwer),_in_press.
Ballesteros-Paredes, J., Hartmann, L., & V[' a]{}zquez-Semadeni, E.1999b,, 527, 285
Ballesteros-Paredes, J. & Mac Low, M. 2002,,_570,_734
Ballesteros-Paredes, J.,_V['_a]{}zquez-Semadeni,_E., & Scalo, J.1999a,, 515,_286
Barranco, J. A. & Goodman, A. A. 1998,, 504,_207
Bate M. R.,_& Burkert, A. 1997,, 288, 1060
Bate, M. R., Bonell,_I.A. .,_& Price, N. M._1995., 277, 362
Benz, W. 1990, in The Numerical Modeling_of Nonlinear Stellar Pulsations, ed. J. R. Buchler,_p. 269, Kluwer Academic_Publishers,_The_Netherlands
Bonnor, W. B. 1956,, 116, 351
Caselli,_P., Benson, P. J., Myers, P. C., &_Tafalla, M. 2002,, 572, 238
Crutcher, R. M. 2004, in_Magnetic Fields and Star Formation: Theory versus_Observations, eds. Ana I Gomez de_Castro et al. (Dordrecht: Kluwer),_in press
Ebert,_R. 1955, Zeitschrift f[ü]{}r Astrophysik, 36,_222
Elmegreen, B. G. 1993,, 419,_L29
Gammie, C. F.,_Lin, Y., Stone,_J. M., & Ostriker, E. C. 2003,, 592, 203
Goodman,_A. A |
for massless HISQ quarks with NRQCD formulated in a moving frame (mNRQCD) was undertaken for the vector and tensor heavy-light currents in [@mueller11].
For massive quarks similar matching calculations using the same lattice action for both quarks have been carried out for Wilson quarks in [@kuramashi98] and for various implementations of NRQCD in [@braaten95; @jones99; @boyle00; @hart07]. To our knowledge, no matching calculations with mixed actions and massive relativistic quarks have been reported in the literature.
Moving from massless to massive relativistic quarks complicates the matching procedure. In the former case, quarks and antiquarks at zero spatial momentum are indistinguishable and consequently scattering and annihilation processes give identical results. In the massive case, however, we must distinguish between quarks and antiquarks. For HISQ quarks at zero spatial momentum, this corresponds to choosing $ap_0 = ia{m_{\text{tree}}}$ or $ap_0=-ia{m_{\text{tree}}}$ respectively. We choose outgoing quarks or antiquarks — the “scattering” or “annihilation” channels respectively — to ensure we do not attempt to compute vanishing matrix elements. Thus we calculate the matrix elements of $V_0$ and $A_k$ in the scattering channel and of $A_0$ and $V_k$ in the annihilation channel. This procedure is valid, even at nonzero lattice spacing, provided we match to the same channel in continuum QCD.
Unfortunately, from the practical viewpoint of calculating Feynman diagrams, using massive quarks complicates the numerical integration considerably. The chief difficulty lies in the annihilation channel, which contains a Coulomb singularity that must be handled with a subtraction function. We discuss the subtraction functions employed in this work in more detail in Appendix \[app:subfn\]. Furthermore, in the automated perturbation theory routines, the pole in the NRQCD propagator crosses the contour of integration and we can no longer carry out the usual Wick rotation back to Minkowski space. Instead, we must deform the integration contours and introduce a triple contour to ensure the stability of numerical integration [@hart07; @mueller11].
For both channels, the lattice matrix elements must be matched to their continuum QCD counterparts. Analytic expressions for the relevant QCD contributions already exist in the literature. References [@braaten95] and [@jones99] discuss the annihilation channel for the axial-vector current, whilst [@boyle00] present results for both | for massless HISQ quarks with NRQCD formulated in a moving frame (mNRQCD) was undertake for the vector and tensor dense - light currents in [ @mueller11 ].
For massive quarks like matching calculations using the like lattice action for both quarks have been carry out for Wilson quarks in [ @kuramashi98 ] and for various implementation of NRQCD in [ @braaten95; @jones99; @boyle00; @hart07 ]. To our knowledge, no matching calculation with assorted actions and massive relativistic quarks have been reported in the literature.
Moving from massless to massive relativistic quarks complicate the matching procedure. In the former case, quark cheese and antiquarks at zero spatial momentum are indistinguishable and consequently scattering and annihilation processes yield identical results. In the massive case, however, we must distinguish between quarks and antiquarks. For HISQ quark at zero spatial momentum, this corresponds to choosing $ ap_0 = ia{m_{\text{tree}}}$ or $ ap_0=-ia{m_{\text{tree}}}$ respectively. We choose outgoing quarks or antiquarks — the “ scattering ” or “ annihilation ” channels respectively — to ensure we do not attempt to compute vanishing matrix elements. Thus we calculate the matrix elements of $ V_0 $ and $ A_k$ in the scattering channel and of $ A_0 $ and $ V_k$ in the annihilation groove. This procedure is valid, even at nonzero lattice spacing, provided we equal to the like distribution channel in continuum QCD.
Unfortunately, from the practical viewpoint of calculating Feynman diagrams, using massive quarks complicates the numerical consolidation considerably. The chief difficulty lies in the annihilation channel, which contains a Coulomb singularity that must be handled with a subtraction function. We discuss the subtraction functions employed in this work in more detail in Appendix \[app: subfn\ ]. Furthermore, in the automatize perturbation hypothesis routines, the pole in the NRQCD propagator crosses the contour of integration and we can no longer carry out the common Wick rotation back to Minkowski space. Instead, we must deform the integration contours and introduce a triple contour to ensure the constancy of numerical integration [ @hart07; @mueller11 ].
For both channels, the lattice matrix elements must be match to their continuum QCD counterparts. Analytic expression for the relevant QCD contributions already exist in the literature. address [ @braaten95 ] and [ @jones99 ] discuss the annihilation distribution channel for the axial - vector current, whilst [ @boyle00 ] present resultant role for both | fog massless HISQ quarks wlth NRQCD formulcred in a movjng framd (mNRQCD) was undertaken for vhe cectoe and tensor heavy-lighg currentd in [@mueoler11].
Hor massive quarks similar matdming eaoculations usikg the same lattice actiot wox both quarks have been carried out sor Wilxoj quarks in [@kutamasny98] ans for various implementations of NDQCD in [@braaten95; @jonex99; @boyle00; @hart07]. To our knowlefge, jo matching calculwtions with mixqe actions ana massive gzlativistic quarks have been reported in tfe licerature.
Movunt fgmm massless to mwssive relatlnistic xuarks vomplicates thc matrhint procedure. In the focmer case, quarks and antiquarns at zero spatial nonentuk ara inaustknghixhzble ajd ronsequentlg scatterint and annihilation krosvxses give idsnticaj wesults. In the massive case, however, we kusf distinguish between qyarks and antiquarks. Vor HISQ zuarks at zero spatial momentum, this corresponds do chkusiun $xp_0 = ia{m_{\text{tree}}}$ or $ap_0=-ia{m_{\text{tree}}}$ respectively. We dhpoxe outgoing qucrks or antiquarls — yre “scattering” or “anujhjlation” channels rfspectidely — to ensurt we co not attempt to compute vqnishing matgix wlements. Thus we cclculate the mattix elrments of $V_0$ and $A_k$ in tke scaftering chajnel and kw $A_0$ and $V_k$ in thd aknivilation channel. This procqdure is talid, even ag nomzero jattice spwcing, provided we match ho thg same channel ij continuum QCD.
Unfortunately, from the practicsl vivwpoint oy calcmlating Feynman diagrams, usiny massivz quarys complicztes thx numerical yntegration cmjsiderably. Tie chief qiffuculry lies kn the annihilstion chauuel, which contains a Coulomn sinehlarity that muwt ve handled with a ruberwcvion sgnction. We dhscurs gne suctraction fmncgionx employed in this wmrk jn more detail in Spiendix \[apk:subfn\]. Fuwthermore, in yhe automated pertkrbatmon thxory rputynes, the pole in the NRQCD prolagator cgosfes the contjur if integratipn and we can no longer carry out the uwual Wick rotation back to Minkowski xkace. Instead, xe muse deform dhe integration contiurs and introduct a triple contour to ehsure dhe shability of numerical integration [@hart07; @mueller11].
For both channels, the lattixe matciv elements just be mdteheb to thqir roitinuum QCD countvrparts. Analytic expressions for the relavcnt QCD contributions already ebist in the lkterature. References [@brazten95] and [@jones99] discuss the annihilation channel for the axial-vegtor currwnt, wgilst [@boylx00] present results fot both | for massless HISQ quarks with NRQCD formulated moving (mNRQCD) was for the vector [@mueller11]. massive quarks similar calculations using the lattice action for both quarks have carried out for Wilson quarks in [@kuramashi98] and for various implementations of NRQCD [@braaten95; @jones99; @boyle00; @hart07]. To our knowledge, no matching calculations with mixed actions massive quarks been in the literature. Moving from massless to massive relativistic quarks complicates the matching procedure. In the case, quarks and antiquarks at zero spatial momentum indistinguishable and consequently scattering annihilation processes give identical results. the case, however, must between and antiquarks. For quarks at zero spatial momentum, this corresponds to choosing $ap_0 = ia{m_{\text{tree}}}$ or $ap_0=-ia{m_{\text{tree}}}$ respectively. We choose quarks or the “scattering” “annihilation” respectively to ensure we attempt to compute vanishing matrix elements. the matrix elements of $V_0$ and $A_k$ in scattering channel of $A_0$ and $V_k$ in the channel. This procedure is valid, even at nonzero spacing, provided we match to the same channel in continuum QCD. Unfortunately, from the practical calculating Feynman diagrams, using quarks complicates the integration The difficulty in the channel, which contains a Coulomb singularity that must be handled with subtraction function. We discuss the subtraction functions employed in this more in Appendix \[app:subfn\]. in the automated perturbation routines, pole in the NRQCD the of can longer out the usual Wick back to Minkowski space. Instead, must deform the integration contour to ensure the stability of numerical integration @mueller11]. For both channels, the lattice matrix must be matched to their continuum QCD counterparts. Analytic expressions for the QCD contributions in the literature. References [@braaten95] and [@jones99] discuss annihilation channel for the current, whilst [@boyle00] present results for both | for massless HISQ quarks with nRQCD formuLated In a MovInG fraMe (mNrQCD) was undertaKEn foR the vector and tensor heaVy-ligHt CUrreNTs In [@mueLler11].
For MAsSIVe qUaRkS siMiLAr MatchIng CalculaTions using The SaMe lattice actIOn For both quaRks Have been carrIed Out for wiLsoN QuarkS in [@KuramAshi98] anD For varIous impleMeNTationS Of NRQCD IN [@BrAateN95; @jones99; @boyle00; @hart07]. To OUr KNowledge, no matcHing caLcULaTIOns WitH mixed actiOnS and mASsive reLAtIVIStiC Quarks have beeN reported in THe lIteratUrE.
MoVIng froM massLeSS to Massive relaTiviStic quarkS complICates thE MatchinG proceDurE. In The fORmEr CasE, qUArkS AnD anTIquArks at zeRo SpAtial MomeNTUM Are iNdiStinGuishAble and conseqUenTly sCAttEring And anNihiLaTion pRocessEs givE iDentical results. in thE massive cAse, HoWevEr, We musT DistinGuiSh bEtween qUarks anD AntIqUARKs. for HISQ quarks at zerO sPATiAl momentUm, this COrReSPonds to cHoOsiNg $ap_0 = IA{M_{\text{Tree}}}$ OR $aP_0=-ia{m_{\text{Tree}}}$ reSPeCtIvely. We ChOose ouTgOinG quArks oR AntiQuarks — The “scattEring” OR “annihilation” cHAnnels respectIVeLY — To ENsurE we Do not attempT to cOMputE vanIShIng MAtrix ElemeNtS. thUS we calculate the matrIx ElemenTs of $V_0$ And $A_k$ in the scaTtering chaNNEL and of $A_0$ aNd $V_k$ IN tHE annihilation cHanneL. This proceDUre is valId, eveN at nonzeRo lattice SPAcing, proVidEd wE maTch TO ThE same channel iN COntiNuUm QCD.
UnForTunatelY, frOm tHe pRacTiCal viewpoInt of calCuLaTiNg feyNman dIAgrams, usInG maSsIve QuarkS CompliCates The nUmErICal IntegraTIoN COnsiDeRaBly. THe cHiEf difFicuLTy lIes in thE annihilaTioN ChanNeL, wHich conTains a Coulomb SiNgularity tHaT muSt be haNDLed with a Subtraction function. We diSCuss the SubTractIon fUnctions eMplOyed in ThiS Work in More deTail iN APpeNDIx \[app:SUBfN\]. FuRtHermore, in tHE AutOmateD pErtuRbation Theory routines, the pOLe iN the NRQCD propAgaTor cROSsEs tHE cONtoUr OF inTEGration and we can No longer caRrY OuT the usual WICk rOtAtion baCk to MinKowskI Space. InStead, we muSt deform tHe InteGRAtiOn contours And introDuce a tripLE contOUr To ensUre The staBiLitY of nuMericaL IntEgratIon [@harT07; @mUeller11].
for boTh Channels, The lattice matrix elementS must bE matcHed To their coNtiNUum qCD counteRparTs. Analytic ExpResSions For THe relEvanT qCd coNTribuTionS Already exISt In tHE LiTerature. RefERENceS [@braaTen95] ANd [@joneS99] disCuss the annihilatiON channel for the AxiaL-VEctOr cURrenT, wHilst [@boyle00] presEnt ReSULts for boTh | for massless HISQ quarkswith NRQCD form ula ted i n amovi ng frame (mNRQ C D) w as undertaken for thevecto ra nd t e ns or he avy-lig h tc u rre nt sin[@ m ue ller1 1].
For m assive qua rks s imilar match i ng calculati ons using the s ame latti ce ac t ion f orbothquarks have b een carri ed out fo r Wilson q ua rksin [@kuramashi98] an d for various i mpleme nt a ti o n s o f N RQCD in [@ br aaten 9 5; @jon e s9 9 ; @bo y le00; @hart07 ]. To our k n owl edge,no ma t chingcalcu la t ion s with mixe d ac tions and massi v e relat i visticquarks ha vebeen re po rte di n t h elit e rat ure.
Mo vi ng from mas s l e s s to ma ssiv e rel ativistic qua rks com p lic atesthe m atch in g pro cedure . Inth e former case,quar ks and an tiq ua rks a t zer o spati almom entum a re indi s tin gu i s h ab le and consequentl ys c at tering a nd ann i hi la t ion proc es ses giv e ident ical re sults. I n them as si ve case ,howeve r, we mu st di s ting uish b etween q uarks and antiquarks . For HISQ qua r ks a tz erospa tial moment um,t hiscorr e sp ond s to c hoosi ng $a p _0 = ia{m_{\text{tr ee }}}$ o r $ap _0=-ia{m_{\te xt{tree}}} $ r espectiv ely. We choose outgoin g qua rks or ant i quarks — the“scatter ing” or “ a n nihilati on” ch ann els r es pectively — t o ensu re we donot attemp t t o c omp ute v anishingmatrix e le me nt s. Th us we calculat ethe m atr ix el e mentsof $V _0$an d$ A_k $ in th e s c a tter in gchan nel a nd of $A_ 0 $ a nd $V_k $ in theann i hila ti on channe l. This proce du re is vali d, ev en atn o nzero la ttice spacing, provided we matc h t o the sam e channel in conti nuu m QCD.
Unfor tunat el y,f r om th e pr act ic al viewpoi n t of calc ul atin g Feynm an diagrams, using mas sive quarks c omp lica t e sthe nu m eri ca l in t e gration conside rably. The c h ie f difficul t y l ie s in th e annih ilati o n chann el, which contains a Cou l o mbsingularit y that m ust be ha n dledw it h a s ubt ractio nfun ction . We d i scu ss th e subt ra ctionfunct io ns emplo yed in this work in mor e deta il in Ap pendix \[ app : sub fn\]. Fur ther more, in t heaut omate d p e rturb atio n t heo r y rou tine s , the pol e i n t h e N RQCD propag a t o r c rosse s t h e cont ourof integration an d we can no lon gerc a rry ou t the u sual Wick rota tio nb a ck to Mi nk owski space . Instea d, we mu st def orm th e integ r a ti o n cont ours an d introdu ceat riple c on to u r to e nsur ethe st abilit y ofn u merical integrat ion [ @ h art07 ; @m uelle r1 1].
Fo r bot h channels , the latti ce mat rixeleme nts mus tbe mat che dto their c o ntinuum Q CD co unterpa rt s. A nal ytic e xpre s s ionsforth e r elevant Q C D c o nt ri b uti onsalrea dy exi st in the literatu re. Referen ce s [ @ b raaten 9 5] a nd [@jones 99] disc u s s the anni h ilat i on chann el for the a xial-ve c tor c urrent, wh i l st [@boyl e00] pres e ntre sult s for bo th | for_massless HISQ_quarks with NRQCD formulated_in a_moving_frame (mNRQCD)_was_undertaken for the_vector and tensor_heavy-light currents in [@mueller11].
For_massive quarks similar_matching_calculations using the same lattice action for both quarks have been carried out for_Wilson_quarks in_[@kuramashi98]_and_for various implementations of NRQCD_in [@braaten95; @jones99; @boyle00; @hart07]._To our_knowledge, no matching calculations with mixed actions and_massive_relativistic quarks have_been reported in the literature.
Moving from massless to massive_relativistic quarks complicates the matching procedure._In the former_case,_quarks_and antiquarks at zero_spatial momentum are indistinguishable and consequently_scattering and annihilation processes give identical_results. In the massive case, however, we_must distinguish between quarks and antiquarks._For HISQ quarks at zero_spatial momentum,_this corresponds to choosing $ap_0_= ia{m_{\text{tree}}}$ or_$ap_0=-ia{m_{\text{tree}}}$ respectively._We choose outgoing_quarks or antiquarks — the “scattering”_or “annihilation” channels_respectively — to ensure we do_not_attempt to compute_vanishing_matrix_elements. Thus_we calculate the_matrix_elements of_$V_0$_and $A_k$ in the scattering channel_and_of $A_0$ and $V_k$ in the annihilation_channel. This procedure is_valid,_even at nonzero lattice_spacing, provided we match to_the same channel in continuum QCD.
Unfortunately,_from the_practical viewpoint_of calculating Feynman diagrams, using massive quarks complicates the numerical integration_considerably. The chief difficulty lies in_the annihilation channel, which_contains a_Coulomb_singularity that must_be_handled with_a subtraction function. We discuss the subtraction_functions employed_in this work in more detail_in Appendix \[app:subfn\]. Furthermore,_in_the automated perturbation theory routines, the_pole in the NRQCD propagator crosses_the contour of integration and_we_can_no longer carry out the_usual Wick rotation back to Minkowski_space. Instead, we_must deform the integration contours and introduce_a_triple contour to ensure the stability_of_numerical integration [@hart07; @mueller11].
For both channels,_the_lattice_matrix elements must be matched_to their continuum QCD counterparts. Analytic_expressions for the relevant QCD contributions already exist in_the literature. References_[@braaten95] and [@jones99] discuss the_annihilation_channel_for the axial-vector current, whilst [@boyle00] present results for both |
_q}\over{\beta(1-\beta)Q^2}}),$$ and $$\begin{aligned}
\frac{1}{z}=1+\frac{\kappa_{t}^{2}+m_q^2}{(1-\beta)Q^2}+\frac{k_t^2+\kappa_t^2-2{\bf{\kappa_t}}.{\bf{k_t}}+m_q^2}{\beta
Q^2}
\label{eq:a}.\end{aligned}$$ As in the reference [@4kimber], the scale $\mu$ which controls the $unintegrated$ gluon and the $QCD$ coupling constant $\alpha_s$, is chosen as follows: $$\begin{aligned}
\mu^2=k_t^2+\kappa_t^2+m_q^2
\label{eq:b}.\end{aligned}$$ One should note that the coefficients used for quark and non-perturbative gluon contributions depend on the transverse momentum. As it has been briefly explained before, the main prescription for $F_L$ consists of three terms; the first term is the $k_t$ factorization which explains the contribution of the $UPDF$ into the $F_L$. This term is derived with the use of pure gluon contribution. However, it only counts the gluon contributions coming from the perturbative region, i.e. for $k_t>1$ $GeV$, and does not have anything to do with the non-perturbative contributions. In the reference [@stasto4], it has been shown that a proper non-perturbative term can be derived from the $k_t$ factorization term, compacting the $k_t$ dependence and the integration with the use of a variable-change, i.e. y, that carries the $k_t$ dependence. Nevertheless, there is a calculable quark contribution in the longitudinal structure function of the proton, which comes from the collinear factorization, i.e. the second term of the equation (\[eq:2p\]).
For the charm quark, $m$ is taken to be $m_c=1.4 GeV$, and $u$, $d$ and $s$ quarks masses are neglected. We also use the same approximation to save the computation time [@tkimber], the one we did for the calculation of $F_2(x,Q^2)$ [@mho1] i.e | _ q}\over{\beta(1-\beta)Q^2}}),$$ and $ $ \begin{aligned }
\frac{1}{z}=1+\frac{\kappa_{t}^{2}+m_q^2}{(1-\beta)Q^2}+\frac{k_t^2+\kappa_t^2 - 2{\bf{\kappa_t}}.{\bf{k_t}}+m_q^2}{\beta
Q^2 }
\label{eq: a}.\end{aligned}$$ As in the reference [ @4kimber ], the scale $ \mu$ which controls the $ unintegrated$ gluon and the $ QCD$ coupling changeless $ \alpha_s$, is choose as follows: $ $ \begin{aligned }
\mu^2 = k_t^2+\kappa_t^2+m_q^2
\label{eq: b}.\end{aligned}$$ One should note that the coefficients use for quark and non - perturbative gluon contributions depend on the cross momentum. As it has been briefly explained before, the main prescription drug for $ F_L$ consist of three terms; the beginning term is the $ k_t$ factorization which explains the contribution of the $ UPDF$ into the $ F_L$. This terminus is derive with the use of arrant gluon contribution. However, it only counts the gluon contribution coming from the perturbative region, i.e. for $ k_t>1 $ $ GeV$, and does not have anything to practice with the non - perturbative contributions. In the reference [ @stasto4 ], it has been shown that a proper non - perturbative term can be deduce from the $ k_t$ factorization term, compacting the $ k_t$ dependence and the integration with the use of a variable star - change, i.e. y, that carries the $ k_t$ dependence. Nevertheless, there is a calculable quark contribution in the longitudinal social organization affair of the proton, which comes from the collinear factorization, i.e. the second term of the equation (\[eq:2p\ ]).
For the charm quark, $ m$ is taken to be $ m_c=1.4 GeV$, and $ u$, $ d$ and $ s$ quarks masses are neglected. We besides practice the same approximation to save the calculation time [ @tkimber ], the one we did for the calculation of $ F_2(x, Q^2)$ [ @mho1 ] i.e | _q}\ovfr{\beta(1-\beta)Q^2}}),$$ and $$\begin{alinned}
\frac{1}{z}=1+\frac{\kapka_{r}^{2}+m_q^2}{(1-\bete)Q^2}+\frac{k_f^2+\kappa_t^2-2{\bw{\kappa_t}}.{\bf{k_t}}+m_q^2}{\beta
Q^2}
\label{eq:a}.\eid{alugned}$$ As in the reference [@4kkmber], the scale $\my$ whmch controls the $unintegvcted$ fpuon end the $QCD$ coukling constatt $\alpha_s$, is cvoreu as follows: $$\begin{aligned}
\mu^2=k_t^2+\kappa_t^2+m_z^2
\label{rq:h}.\end{aligned}$$ Ong shollq nofv uhat the coefficients used for quzrk and non-perturbatove gluon contributions deoend on the transverse momentum. Aw it yas been bridfly explapued before, jhe main prescription for $F_L$ conrists of three jzems; jhe first tecm is nhe $k_t$ factorldation fhich ecplains the coktribntiob of the $UPDF$ into thx $F_L$. This term is detived with tke use of pure gluon xobtribotion. Howdcer, it oily countd tie gluon cohtributions coming from the petttgnative regioh, i.e. fjr $k_t>1$ $GeV$, and does not have anything to dm wjth the non-perturbative contributions. In the referencq [@stasto4], it has been shown that a proper non-pertusbatite texn can ve derived from the $k_t$ factorization term, compasfimg the $k_t$ dependcnce and the integtahipg with the usg of a rzrjable-change, i.e. y, tjat carties tye $k_t$ deptndenve. Nevertheless, there is a xalculable qlark contribution in tke longitudiual sttucturr function of the protou, whicg comes frol the colmknear factorizatkon, i.a. the second term of the ezuation (\[ew:2p\]).
Fox the chxrm auark, $i$ is taken to bc $m_c=1.4 GeV$, and $u$, $d$ anf $s$ qoarks kasses are neglected. We also use the same approximation tm sdve the eomputstion time [@tkymber], the one ce did fjr thd calculatpon of $F_2(x,X^2)$ [@mho1] i.e | _q}\over{\beta(1-\beta)Q^2}}),$$ and $$\begin{aligned} \frac{1}{z}=1+\frac{\kappa_{t}^{2}+m_q^2}{(1-\beta)Q^2}+\frac{k_t^2+\kappa_t^2-2{\bf{\kappa_t}}.{\bf{k_t}}+m_q^2}{\beta Q^2} \label{eq:a}.\end{aligned}$$ As reference the scale which controls the coupling $\alpha_s$, is chosen follows: $$\begin{aligned} \mu^2=k_t^2+\kappa_t^2+m_q^2 One should note that the coefficients for quark and non-perturbative gluon contributions depend on the transverse momentum. As it been briefly explained before, the main prescription for $F_L$ consists of three terms; first is $k_t$ which explains the contribution of the $UPDF$ into the $F_L$. This term is derived with the of pure gluon contribution. However, it only counts gluon contributions coming from perturbative region, i.e. for $k_t>1$ and not have to with non-perturbative contributions. In reference [@stasto4], it has been shown that a proper non-perturbative term can be derived from the $k_t$ term, compacting dependence and integration the of a variable-change, that carries the $k_t$ dependence. Nevertheless, calculable quark contribution in the longitudinal structure function the proton, comes from the collinear factorization, i.e. second term of the equation (\[eq:2p\]). For the quark, $m$ is taken to be $m_c=1.4 GeV$, and $u$, $d$ and $s$ quarks masses We also use the approximation to save computation [@tkimber], one did for calculation of $F_2(x,Q^2)$ [@mho1] i.e | _q}\over{\beta(1-\beta)Q^2}}),$$ and $$\begin{aliGned}
\frac{1}{z}=1+\fRac{\kaPpa_{T}^{2}+m_q^2}{(1-\BeTa)Q^2}+\fRac{k_T^2+\kappa_t^2-2{\bf{\kappa_T}}.{\Bf{k_t}}+M_q^2}{\beta
Q^2}
\label{eq:a}.\end{aligNed}$$ As In THe reFErEnce [@4kImber], thE ScALE $\mu$ WhIcH coNtROlS the $uNinTegrateD$ gluon and tHe $QcD$ Coupling consTAnT $\alpha_s$, is cHosEn as follows: $$\bEgiN{alignEd}
\Mu^2=k_T^2+\Kappa_T^2+m_q^2
\Label{Eq:b}.\end{ALigned}$$ one should NoTE that tHE coeffiCIEnTs usEd for quark and non-pERtURbative gluon coNtribuTiONs DEPenD on The transveRsE momeNTum. As it HAs BEEN brIEfly explained Before, the maIN prEscripTiOn fOR $F_L$ conSists Of THreE terms; the fiRst tErm is the $k_T$ factoRIzation WHich expLains tHe cOntRibuTIoN oF thE $UpdF$ iNTo The $f_l$. ThIs term is DeRiVed wiTh thE USE Of puRe gLuon ContrIbution. HoweveR, it Only COunTs the Gluon ContRiButioNs comiNg froM tHe perturbative rEgioN, i.e. for $k_t>1$ $GEV$, aNd DoeS nOt havE AnythiNg tO do With the Non-pertURbaTiVE COnTributions. In the refErENCe [@Stasto4], it Has beeN ShOwN That a proPeR noN-perTURbatiVe teRM cAn be deriVed froM ThE $k_T$ factorIzAtion tErM, coMpaCting THe $k_t$ DependEnce and tHe intEGration with the USe of a variable-CHaNGE, i.E. Y, thaT caRries the $k_t$ dEpenDEnce. neveRThEleSS, therE is a cAlCUlABle quark contributioN iN the loNgituDinal structurE function oF THE proton, wHich COmES from the collinEar faCtorizatioN, I.e. the secOnd teRm of the eQuation (\[eq:2P\]).
fOr the chaRm qUarK, $m$ iS taKEN tO be $m_c=1.4 GeV$, and $u$, $d$ AND $s$ quArKs masseS arE neglecTed. we aLso Use ThE same apprOximatioN tO sAvE tHe cOmputATion time [@TkImbEr], The One we DId for tHe calCulaTiOn OF $F_2(x,q^2)$ [@mho1] i.e | _q}\over{\beta(1-\beta)Q^2 }}),$$ and $$\b egi n{a li gned }
\f rac{1}{z}=1+\f r ac{\ kappa_{t}^{2}+m_q^2}{( 1-\be ta ) Q^2} + \f rac{k _t^2+\k a pp a _ t^2 -2 {\ bf{ \k a pp a_t}} .{\ bf{k_t} }+m_q^2}{\ bet aQ^2}
\label { eq :a}.\end{a lig ned}$$ As in th e refe re nce [@4ki mbe r], t he sca l e $\mu $ which c on t rols t h e $unin t e gr ated $ gluon and the $ Q CD $ coupling cons tant $ \a l ph a _ s$, is chosen as f ollow s : $$\be g in { a l ign e d}
\mu^2=k_t^ 2+\kappa_t^ 2 +m_ q^2
\ la bel { eq:b}. \end{ al i gne d}$$ One sh ould note tha t thec oeffici e nts use d forqua rkandn on -p ert ur b ati v eglu o n c ontribut io ns depe nd o n t h e tr ans vers e mom entum. As ithas bee n br iefly expl aine dbefor e, the main p rescription for $F_ L$ consis tsof th re e ter m s; the fi rst term i s the $ k _t$ f a c t or ization which expl ai n s t he contr ibutio n o ft he $UPDF $int o th e $F_L$ . Th i sterm isderive d w it h the u se of pu re gl uon cont r ibut ion. H owever,it on l y counts the g l uon contribut i on s co m ingfro m the pertu rbat i ve r egio n ,i.e . for$k_t> 1$ $G e V$, and does not ha ve anyth ing t o do with the non-pertu r b a tive con trib u ti o ns. In the ref erenc e [@stasto 4 ], it ha s bee n shownthat a pr o p er non-p ert urb ati vet e rm can be deriv e d fro mthe $k_ t$factori zat ion te rm, c ompacting the $k_ t$ d ep en den ce an d the int eg rat io n w ith t h e useof avari ab le - cha nge, i. e .y , tha tca rrie s t he $k_t $ de p end ence. N everthele ss, ther eis a calc ulable quarkco ntribution i n t he lon g i tudinalstructure function of t h e proto n,which com es from t hecollin ear factor izatio n, i. e. th e secon d te rmof the equat i o n ( \[eq: 2p \]).
For t he charm quark, $m $ is taken to be$m_ c=1. 4 Ge V$, an d $u $, $d$ a nd $s$ quarks m asses arene g le cted. We a l sous e the s ame app roxim a tion to save the computat io n ti m e [@ tkimber],the onewe did fo r thec al culat ion of $F _2 (x, Q^2)$ [@mho 1 ] i .e | _q}\over{\beta(1-\beta)Q^2}}),$$ and_$$\begin{aligned}
\frac{1}{z}=1+\frac{\kappa_{t}^{2}+m_q^2}{(1-\beta)Q^2}+\frac{k_t^2+\kappa_t^2-2{\bf{\kappa_t}}.{\bf{k_t}}+m_q^2}{\beta
Q^2}
\label{eq:a}.\end{aligned}$$_As in the reference_[@4kimber], the_scale_$\mu$ which_controls_the $unintegrated$ gluon_and the $QCD$_coupling constant $\alpha_s$, is_chosen as follows:_$$\begin{aligned}
\mu^2=k_t^2+\kappa_t^2+m_q^2
_\label{eq:b}.\end{aligned}$$ One should note that the coefficients used for quark and non-perturbative gluon contributions_depend_on the_transverse_momentum._As it has been briefly_explained before, the main prescription_for $F_L$_consists of three terms; the first term is_the_$k_t$ factorization which_explains the contribution of the $UPDF$ into the $F_L$._This term is derived with the_use of pure_gluon_contribution._However, it only counts_the gluon contributions coming from the_perturbative region, i.e. for $k_t>1$ $GeV$,_and does not have anything to do_with the non-perturbative contributions. In the_reference [@stasto4], it has been_shown that_a proper non-perturbative term can_be derived from_the $k_t$_factorization term, compacting_the $k_t$ dependence and the integration_with the use_of a variable-change, i.e. y, that_carries_the $k_t$ dependence._Nevertheless,_there_is a_calculable quark contribution_in_the longitudinal_structure_function of the proton, which comes_from_the collinear factorization, i.e. the second term_of the equation (\[eq:2p\]).
For_the_charm quark, $m$ is_taken to be $m_c=1.4 GeV$,_and $u$, $d$ and $s$ quarks_masses are_neglected. We_also use the same approximation to save the computation time [@tkimber],_the one we did for the_calculation of $F_2(x,Q^2)$ [@mho1]_i.e |
mathbf{n}\times \mathbf{I},\ \xi\in\mathbb{R}$ [@HarringtonMautz; @ColtonKress], and thus those operators $\mathcal{R}$ are not regularizing operators for $\mathcal{T}_k$. If $\mathcal{R}$ is chosen as a right regularizing operator for $\mathcal{T}_k$, the integral operator on the left hand side of (\[eq:CFIE\_R\]) is a second kind Fredholm operator, and, thus, the unique solvability of equation (\[eq:CFIE\_R\]) is equivalent to the injectivity of the left-hand-side operator. Operators $\mathcal{R}$ with the aforementioned property have been proposed and analyzed in the literature [@turc1]. The aim of this paper is to present a general strategy to produce and analyze novel regularizing operators $\mathcal{R}$ and to analyze previously introduced regularizing operators for which an analysis does not exist in the literature.
The starting point of constructing suitable regularizing operators $\mathcal{R}$ is the Calderón’s identity $\mathcal{T}_k^2=\frac{I}{4}-\mathcal{K}_k^2$ [@HsiaoKleinman]. The regularizing operators $\mathcal{R}$ should thus resemble the electric field operator $\mathcal{T}_k$. If the choice $\mathcal{R}=\mathcal{T}_k$ were made, the ensuing CFIER operators would not be injective since for certain wavenumbers $k$ the operators $\frac{I}{2}-\mathcal{K}_k$ are not injective [@ColtonKress; @HsiaoKleinman]. In order to ensure the injectivity of the resulting CFIER operators, one strategy is to modify the wavenumber in the definition of the regularizing operator $\mathcal{R}=\mathcal{T}_k$. Specifically, we propose a general regularizing operator $\mathcal{R}$ of the form $$\label{eq:defR}
\mathcal{R} = \eta\ \mathbf{n}\times \mathbf{S}_K +\zeta\ \mathcal{T}^1_K\ {\rm div}_\Gamma$$ where $\eta$ and $\zeta$ are complex numbers and $K$ is a complex wavenumber such that $\Im K>0$. Given the starting point, we refer to the CFIER operators thus constructed as Calderón CFIER. We | mathbf{n}\times \mathbf{I},\ \xi\in\mathbb{R}$ [ @HarringtonMautz; @ColtonKress ], and thus those operators $ \mathcal{R}$ are not regularizing operators for $ \mathcal{T}_k$. If $ \mathcal{R}$ is choose as a proper regularizing operator for $ \mathcal{T}_k$, the integral operator on the leftover hand side of (\[eq: CFIE\_R\ ]) is a second kind Fredholm operator, and, therefore, the singular solvability of equation (\[eq: CFIE\_R\ ]) is equivalent to the injectivity of the leftover - bridge player - side hustler. Operators $ \mathcal{R}$ with the aforementioned property have been proposed and analyze in the literature [ @turc1 ]. The aim of this paper is to present a general strategy to grow and analyze novel regularizing operators $ \mathcal{R}$ and to analyze previously introduced regularize operators for which an analysis does not exist in the literature.
The starting point of construct suitable regularizing operators $ \mathcal{R}$ is the Calderón ’s identity $ \mathcal{T}_k^2=\frac{I}{4}-\mathcal{K}_k^2 $ [ @HsiaoKleinman ]. The regularizing operators $ \mathcal{R}$ should thus resemble the electric discipline operator $ \mathcal{T}_k$. If the choice $ \mathcal{R}=\mathcal{T}_k$ were made, the ensuing CFIER operators would not be injective since for sealed wavenumbers $ k$ the operator $ \frac{I}{2}-\mathcal{K}_k$ are not injective [ @ColtonKress; @HsiaoKleinman ]. In order to ensure the injectivity of the resulting CFIER operators, one strategy is to modify the wavenumber in the definition of the regularizing operator $ \mathcal{R}=\mathcal{T}_k$. Specifically, we propose a general regularizing hustler $ \mathcal{R}$ of the kind $ $ \label{eq: defR }
\mathcal{R } = \eta\ \mathbf{n}\times \mathbf{S}_K + \zeta\ \mathcal{T}^1_K\ { \rm div}_\Gamma$$ where $ \eta$ and $ \zeta$ are complex numbers and $ K$ is a complex wavenumber such that $ \Im K>0$. feed the start item, we refer to the CFIER operators thus constructed as Calderón CFIER. We | matjbf{n}\times \mathbf{I},\ \xi\in\mauhbb{R}$ [@HarringtonMaotz; @ColtmnKresa], and thjs those operators $\mathcal{R}$ ere bot rtyularizing operators for $\mathbal{T}_k$. If $\nathral{R}$ is chosen as a righb reghpariviig operator for $\mathcal{T}_k$, the integral mpdrctor on the left hand side of (\[eq:CFIE\_W\]) is a xefond kind Fredrolm jperznov, and, thus, the unique solvabilitg of eqlation (\[eq:CFIE\_R\]) is equivalent to the injectigity of the left-hand-sife operator. Opewqtors $\mathcau{R}$ with tht cforementiohed property have been proposed and cnalyzed in tye phterature [@tucc1]. The aim of this paper iv to prrsent a generak svratwgy to produce and anelyze novel regularieing operadoxs $\mathcal{R}$ and to anqltze pteviogsly untfodhcxd degulagizmng operatods for whicy an analysis does moe exist in the literwttre.
The starting point of constructing slitagle regularizing operatirs $\mathcal{R}$ is the Cwlderón’s iqentity $\mathcal{T}_k^2=\frac{I}{4}-\mathcal{K}_k^2$ [@HsiaoKleinman]. The segulerkziun upegators $\mathcal{R}$ should thus resemble the elecediv nield operator $\mcthcal{T}_k$. If the vhlivg $\mathcal{R}=\mathzal{T}_k$ csrs made, the ensuing CFIER jperarors woulq noy be injective since for ceetain wavenuivers $k$ the operatoxs $\frac{I}{2}-\matheal{K}_k$ sre npt injective [@ColtonKress; @HsizoKleinman]. Ln order fu ensure the injdctpvitf of the resulting CFIER o[erators, ine xtrategh is to moqify the wwvenumber in the definitlon oy the regularizlng operator $\mathcal{R}=\mathcal{T}_k$. Specifically, wg psopmse a geueral vegularizing opqrator $\mathcal{T}$ of the yorm $$\lxbel{eq:defR}
\jathcal{C} = \eta\ \mathbs{n}\times \mathbx{D}_K +\zeta\ \mathral{T}^1_K\ {\rm qiv}_\Gqmma$$ where $\dga$ and $\zeta$ arr complex numbers abd $K$ is a complex xavevhmber such that $\Im K>0$. Given the stsrtkng pliit, we sefer to the CFIDR uleraturs thus cokstfuctrd as Calderón CFIER. Fe | mathbf{n}\times \mathbf{I},\ \xi\in\mathbb{R}$ [@HarringtonMautz; @ColtonKress], and thus $\mathcal{R}$ not regularizing for $\mathcal{T}_k$. If right operator for $\mathcal{T}_k$, integral operator on left hand side of (\[eq:CFIE\_R\]) is second kind Fredholm operator, and, thus, the unique solvability of equation (\[eq:CFIE\_R\]) is to the injectivity of the left-hand-side operator. Operators $\mathcal{R}$ with the aforementioned property been and in literature [@turc1]. The aim of this paper is to present a general strategy to produce and novel regularizing operators $\mathcal{R}$ and to analyze previously regularizing operators for which analysis does not exist in literature. starting point constructing regularizing $\mathcal{R}$ is the identity $\mathcal{T}_k^2=\frac{I}{4}-\mathcal{K}_k^2$ [@HsiaoKleinman]. The regularizing operators $\mathcal{R}$ should thus resemble the electric field operator $\mathcal{T}_k$. If the $\mathcal{R}=\mathcal{T}_k$ were ensuing CFIER would be since for certain the operators $\frac{I}{2}-\mathcal{K}_k$ are not injective order to ensure the injectivity of the resulting operators, one is to modify the wavenumber in definition of the regularizing operator $\mathcal{R}=\mathcal{T}_k$. Specifically, we a general regularizing operator $\mathcal{R}$ of the form $$\label{eq:defR} \mathcal{R} = \eta\ \mathbf{n}\times \mathbf{S}_K +\zeta\ div}_\Gamma$$ where $\eta$ and are complex numbers $K$ a wavenumber that $\Im Given the starting point, we refer to the CFIER operators thus as Calderón CFIER. We | mathbf{n}\times \mathbf{I},\ \xi\in\maThbb{R}$ [@HarriNgtonmauTz; @COlTonKRess], And thus those opERatoRs $\mathcal{R}$ are not regulaRizinG oPEratORs For $\maThcal{T}_k$. iF $\mATHcaL{R}$ Is ChoSeN As A righT reGularizIng operatoR foR $\mAthcal{T}_k$, the iNTeGral operatOr oN the left hand SidE of (\[eq:CfIe\_R\]) iS A secoNd kInd FrEdholm OPeratoR, and, thus, tHe UNique sOLvabiliTY Of EquaTion (\[eq:CFIE\_R\]) is equiVAlENt to the injectiVity of ThE LeFT-HanD-siDe operator. opEratoRS $\mathcaL{r}$ wITH The AForementioned Property havE BeeN propoSeD anD AnalyzEd in tHe LIteRature [@turc1]. THe aiM of this paPer is tO Present A General StrateGy tO prOducE AnD aNalYzE NovEL rEguLAriZing operAtOrS $\mathCal{R}$ AND TO anaLyzE preViousLy introduced rEguLariZIng OperaTors fOr whIcH an anAlysis Does nOt Exist in the literAturE.
The startIng PoInt Of ConstRUcting SuiTabLe regulArizing OPerAtORS $\MaThcal{R}$ is the CalderóN’s IDEnTity $\mathCal{T}_k^2=\fRAc{i}{4}-\mAThcal{K}_k^2$ [@HSiAoKLeinMAN]. The rEgulARiZing operAtors $\mAThCaL{R}$ shoulD tHus resEmBle The ElectRIc fiEld opeRator $\matHcal{T}_K$. if the choice $\matHCal{R}=\mathcal{T}_k$ WErE MAdE, The eNsuIng CFIER opeRatoRS wouLd noT Be InjECtive Since FoR CeRTain wavenumbers $k$ the OpEratorS $\frac{i}{2}-\mathcal{K}_k$ are Not injectiVE [@cOltonKreSs; @HsIAokLeinman]. In order To ensUre the injeCTivity of The reSulting CfIER operaTORs, one strAteGy iS to ModIFY tHe wavenumber iN THe deFiNition oF thE regulaRizIng OpeRatOr $\Mathcal{R}=\mAthcal{T}_k$. spEcIfIcAllY, we prOPose a genErAl rEgUlaRizinG OperatOr $\matHcal{r}$ oF tHE foRm $$\label{EQ:dEFr}
\matHcAl{r} = \eta\ \MatHbF{n}\timEs \maTHbf{s}_K +\zeta\ \mAthcal{T}^1_K\ {\rM diV}_\gammA$$ wHeRe $\eta$ anD $\zeta$ are complEx Numbers and $k$ iS a cOmplex WAVenumber Such that $\Im K>0$. Given the starTIng poinT, we Refer To thE CFIER opeRatOrs thuS coNStructEd as CaLderóN CfIEr. wE | mathbf{n}\times \mathbf{I} ,\ \xi\in\ mathb b{R }$[@ Harr ingt onMautz; @Colt o nKre ss], and thus those op erato rs $\ma t hc al{R} $ are n o tr e gul ar iz ing o p er ators fo r $\mat hcal{T}_k$ . I f$\mathcal{R} $ i s chosen a s a right regul ari zing o pe rat o r for $\ mathc al{T}_ k $, the integral o p erator on thel e ft han d side of (\[eq:C F IE \ _R\]) is a sec ond ki nd Fr e d hol m o perator, a nd , thu s , the u n iq u e sol v ability of eq uation (\[e q :CF IE\_R\ ]) is equiva lentto the injectivit y of the left -hand- s ide ope r ator. O perato rs$\m athc a l{ R} $ w it h th e a for e men tioned p ro pe rty h aveb e e n pro pos ed a nd an alyzed in the li tera t ure [@tu rc1]. The a im of thispaper i s to present agene ral strat egy t o p ro ducea nd ana lyz e n ovel re gulariz i ngop e r a to rs $\mathcal{R}$ a nd t oanalyzeprevio u sl yi ntroduce dreg ular i z ing o pera t or s for wh ich an an al ysis do es not e xi stinthe l i tera ture.
The sta rting point of const r ucting suitab l er e gu l ariz ing operators$\ma t hcal {R}$ is th e Cald erón’ si de n tity $\mathcal{T}_k ^2 =\frac {I}{4 }-\mathcal{K} _k^2$ [@Hs i a o Kleinman ]. T h er egularizing op erato rs $\mathc a l{R}$ sh ouldthus res emble the e lectricfie ldope rat o r $ \mathcal{T}_k $ . Ifth e choic e $ \mathca l{R }=\ mat hca l{ T}_k$ wer e made,th een su ing CFIE R operato rs wo ul d n ot be inject ive s ince f or cer tain wa v en u m bers $ k$ the op er ators $\f r ac{ I}{2}-\ mathcal{K }_k $ are n ot inject ive [@ColtonK re ss; @Hsiao Kl ein man].I n order t o ensure the injectivit y of the re sulti ng C FIER oper ato rs, on e s t rategy is to modi fy th e waven u m be r i nthe defini t i onof th eregu larizin g operator $\mathc a l{R }=\mathcal{T} _k$ . Sp e c if ica l ly , we p r opo s e a general regu larizing o pe r at or $\mathc a l{R }$ of the form $ $\lab e l{eq:de fR}
\math cal{R} =\e ta\\ m ath bf{n}\time s \mathb f{S}_K +\ z eta\\ ma thcal {T} ^1_K\{\ rmdiv}_ \Gamma $ $ w here$\eta$ a nd $\z eta$ar e comple x numbers and $K$ is acomple x wav enu mber such th a t $ \Im K>0$. Giv en the sta rti ngpoint , w e refe r to th e C F IER o pera t ors thusc on str u c te d as Calder ó n CFI ER. W e | mathbf{n}\times \mathbf{I},\_\xi\in\mathbb{R}$ [@HarringtonMautz; @ColtonKress],_and thus those operators_$\mathcal{R}$ are_not_regularizing operators_for_$\mathcal{T}_k$. If $\mathcal{R}$_is chosen as_a right regularizing operator_for $\mathcal{T}_k$, the_integral_operator on the left hand side of (\[eq:CFIE\_R\]) is a second kind Fredholm operator,_and,_thus, the_unique_solvability_of equation (\[eq:CFIE\_R\]) is equivalent_to the injectivity of the_left-hand-side operator._Operators $\mathcal{R}$ with the aforementioned property have been_proposed_and analyzed in_the literature [@turc1]. The aim of this paper is to_present a general strategy to produce_and analyze novel_regularizing_operators_$\mathcal{R}$ and to analyze_previously introduced regularizing operators for which_an analysis does not exist in_the literature.
The starting point of constructing suitable_regularizing operators $\mathcal{R}$ is the Calderón’s_identity $\mathcal{T}_k^2=\frac{I}{4}-\mathcal{K}_k^2$ [@HsiaoKleinman]. The regularizing operators_$\mathcal{R}$ should_thus resemble the electric field_operator $\mathcal{T}_k$. If_the choice_$\mathcal{R}=\mathcal{T}_k$ were made,_the ensuing CFIER operators would not_be injective since_for certain wavenumbers $k$ the operators_$\frac{I}{2}-\mathcal{K}_k$_are not injective [@ColtonKress;_@HsiaoKleinman]._In_order to_ensure the injectivity_of_the resulting_CFIER_operators, one strategy is to modify_the_wavenumber in the definition of the regularizing_operator $\mathcal{R}=\mathcal{T}_k$. Specifically, we_propose_a general regularizing operator_$\mathcal{R}$ of the form $$\label{eq:defR}
\mathcal{R}_= \eta\ \mathbf{n}\times \mathbf{S}_K +\zeta\ \mathcal{T}^1_K\_{\rm div}_\Gamma$$_where $\eta$_and $\zeta$ are complex numbers and $K$ is a complex wavenumber_such that $\Im K>0$. Given the_starting point, we refer_to the_CFIER_operators thus constructed_as_Calderón CFIER._We |
{{\cal V}},{{\cal B}},{{\cal O}})$-1dta. By $\mathrm{1QFA}$, we indicate the family of all languages recognized by bounded-error 1qfa’s. We then obtain $\mathrm{1QFA} = {\mathrm{REG}}$.
[**(vii) Deterministic Pushdown Automata.**]{} A *1-way deterministic pushdown automaton* (or a 1dpda) $M$ can be seen as a $({{\cal V}},{{\cal B}},{{\cal O}})$-1dta when $({{\cal V}},{{\cal B}},{{\cal O}})$ satisfies the following properties. Let ${{\cal V}}=\{[k]\times \bot\Gamma^*\mid k\in{\mathbb{N}}^{+}, \Gamma\,\text{: alphabet}\}$, where $\bot$ is a distinguished bottom marker not in $\Gamma$. Let ${{\cal B}}$ be composed of all maps of the form $B(q,\bot z) = (\mu_1(q,z_n),\bot z_1z_2\cdots z_{n-1}\mu_2(q,z_n))$ for 2 functions $\mu_1:[k]\times \Gamma\to [k]$ and $\mu_2:[k]\times\Gamma\to\Gamma^{\leq l}$, where $z=z_1z_2\cdots z_n\in\Gamma^n$ and $l\in{\mathbb{N}}^{+}$. Intuitively, a single application of $B$ represents a series of moves in which $M$ reads one symbol and then makes a single non-$\lambda$-move followed by a certain number of $\lambda$-moves. Let ${{\cal O}}$ consist of all pairs $(E_{acc},E_{rej})$ with $E_{acc}= Q_1\times \bot\Gamma^*$ and $E_{rej}= Q_2\times \bot\Gamma^*$, where $\{Q_1,Q_2\}$ is a partition of $[k]$. We write ${\mathrm{DCFL}}$ for the class of all languages recognized by 1dpda’s. Well known relations include ${\mathrm{REG}}\subsetneqq {\mathrm{CFL}}$.
Basic Properties of $({{\cal V}},{{\cal B}},{{\cal O}})$-1dta’s {#sec:basic-properties}
===============================================================
For a given automata base $( | { { \cal V}},{{\cal B}},{{\cal O}})$-1dta. By $ \mathrm{1QFA}$, we indicate the family of all languages recognized by restrict - mistake 1qfa ’s. We then obtain $ \mathrm{1QFA } = { \mathrm{REG}}$.
[ * * (vii) Deterministic Pushdown Automata. * * ] { } A * 1 - way deterministic pushdown automaton * (or a 1dpda) $ M$ can be seen as a $ ({ { \cal V}},{{\cal B}},{{\cal O}})$-1dta when $ ({ { \cal V}},{{\cal B}},{{\cal O}})$ satisfy the following properties. permit $ { { \cal V}}=\{[k]\times \bot\Gamma^*\mid k\in{\mathbb{N}}^{+ }, \Gamma\,\text {: alphabet}\}$, where $ \bot$ is a distinguished bottom marker not in $ \Gamma$. Let $ { { \cal B}}$ be composed of all map of the form $ B(q,\bot z) = (\mu_1(q, z_n),\bot z_1z_2\cdots z_{n-1}\mu_2(q, z_n))$ for 2 functions $ \mu_1:[k]\times \Gamma\to [ k]$ and $ \mu_2:[k]\times\Gamma\to\Gamma^{\leq l}$, where $ omega = z_1z_2\cdots z_n\in\Gamma^n$ and $ l\in{\mathbb{N}}^{+}$. Intuitively, a single application of $ B$ represents a series of motion in which $ M$ reads one symbol and then draw a individual non-$\lambda$-move succeed by a certain number of $ \lambda$-moves. lease $ { { \cal O}}$ consist of all pairs $ (E_{acc},E_{rej})$ with $ E_{acc}= Q_1\times \bot\Gamma^*$ and $ E_{rej}= Q_2\times \bot\Gamma^*$, where $ \{Q_1,Q_2\}$ is a partition of $ [ k]$. We write $ { \mathrm{DCFL}}$ for the course of all languages recognize by 1dpda ’s. Well known relations include $ { \mathrm{REG}}\subsetneqq { \mathrm{CFL}}$.
Basic Properties of $ ({ { \cal V}},{{\cal B}},{{\cal O}})$-1dta ’s { # sec: basic - property }
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
For a given automata base $ ( | {{\cal V}},{{\cal B}},{{\cal O}})$-1dta. By $\mathro{1QFA}$, we indicatg rhe fakily or all lavguages recognized by boundev-errir 1qfq’s. We then obtain $\mathfm{1QFA} = {\manhrm{REG}}$.
[**(viu) Deuerministic Pushdown Automata.**]{} A *1-aay beverministic pusmdown automdton* (or a 1dpda) $M$ ccn be seen as a $({{\cal V}},{{\cal B}},{{\cal O}})$-1dta wren $({{\cal V}},{{\fal B}},{{\cal O}})$ satysfits ehe rollowing properties. Let ${{\cal V}}=\{[k]\timss \bot\Gemma^*\mid k\in{\mathnb{N}}^{+}, \Gamma\,\text{: alphabet}\}$, whege $\blt$ is a distinguisjed bottom narkqe not in $\Gamoa$. Let ${{\cal B}}$ be composgd of all maps of the form $B(q,\bot z) = (\mb_1(q,z_n),\bot z_1z_2\ceors x_{t-1}\mu_2(q,z_n))$ for 2 funcnions $\mu_1:[k]\times \Gamma\tm [k]$ and $\mu_2:[k]\times\Gamma\bo\Gamka^{\lwq l}$, where $z=z_1z_2\cdots z_i\in\Gamma^n$ and $l\in{\matrbb{N}}^{+}$. Intuhtnvely, a single applicqtuon ox $B$ seprdwengs z xedies ov mkves in whjch $M$ reads one symbol and them iqkes a single non-$\laibqa$-move followed by a certain number of $\namgda$-moves. Let ${{\cal O}}$ consust of all pairs $(E_{acc},G_{rej})$ with $Q_{acc}= Q_1\times \bot\Gamma^*$ and $E_{rej}= Q_2\times \bot\Gamma^*$, whare $\{Q_1,S_2\}$ is a oqrhition of $[k]$. We write ${\mathrm{DCFL}}$ for the class kf akl languages rccognized by 1dpda’s. Wflk known relatiuns inemuse ${\mathrm{REG}}\subsetjeqq {\majhrm{CFO}}$.
Basic Prjperyies of $({{\cal V}},{{\cal B}},{{\cal O}})$-1dta’s {#sec:basic-proiertues}
===============================================================
For a given autlmata base $( | {{\cal V}},{{\cal B}},{{\cal O}})$-1dta. By $\mathrm{1QFA}$, we family all languages by bounded-error 1qfa’s. {\mathrm{REG}}$. Deterministic Pushdown Automata.**]{} *1-way deterministic pushdown (or a 1dpda) $M$ can be as a $({{\cal V}},{{\cal B}},{{\cal O}})$-1dta when $({{\cal V}},{{\cal B}},{{\cal O}})$ satisfies the properties. Let ${{\cal V}}=\{[k]\times \bot\Gamma^*\mid k\in{\mathbb{N}}^{+}, \Gamma\,\text{: alphabet}\}$, where $\bot$ is a distinguished marker in Let B}}$ be composed of all maps of the form $B(q,\bot z) = (\mu_1(q,z_n),\bot z_1z_2\cdots z_{n-1}\mu_2(q,z_n))$ for functions $\mu_1:[k]\times \Gamma\to [k]$ and $\mu_2:[k]\times\Gamma\to\Gamma^{\leq l}$, where z_n\in\Gamma^n$ and $l\in{\mathbb{N}}^{+}$. Intuitively, single application of $B$ represents series moves in $M$ one and then makes single non-$\lambda$-move followed by a certain number of $\lambda$-moves. Let ${{\cal O}}$ consist of all pairs $(E_{acc},E_{rej})$ $E_{acc}= Q_1\times $E_{rej}= Q_2\times where is partition of $[k]$. ${\mathrm{DCFL}}$ for the class of all 1dpda’s. Well known relations include ${\mathrm{REG}}\subsetneqq {\mathrm{CFL}}$. Basic of $({{\cal B}},{{\cal O}})$-1dta’s {#sec:basic-properties} =============================================================== For a automata base $( | {{\cal V}},{{\cal B}},{{\cal O}})$-1dta. By $\mathrm{1QFa}$, we indicatE the fAmiLy oF aLl laNguaGes recognized bY BounDed-error 1qfa’s. We then obtaIn $\matHrM{1qFA} = {\mAThRm{REG}}$.
[**(Vii) DeteRMiNIStiC PUsHdoWn aUtOmata.**]{} a *1-waY determInistic pusHdoWn Automaton* (or a 1DPdA) $M$ can be seeN as A $({{\cal V}},{{\cal B}},{{\cal o}})$-1dtA when $({{\cAl v}},{{\caL b}},{{\cal O}})$ SatIsfieS the foLLowing PropertieS. LET ${{\cal V}}=\{[k]\TImes \bot\gAMmA^*\mid K\in{\mathbb{N}}^{+}, \Gamma\,\teXT{: aLPhabet}\}$, where $\bot$ Is a disTiNGuISHed BotTom marker nOt In $\GamMA$. Let ${{\cal b}}$ Be COMPosED of all maps of tHe form $B(q,\bot Z) = (\Mu_1(q,Z_n),\bot z_1Z_2\cDotS Z_{n-1}\mu_2(q,z_N))$ for 2 fUnCTioNs $\mu_1:[k]\times \GAmma\To [k]$ and $\mu_2:[k]\Times\GAMma\to\GaMMa^{\leq l}$, wHere $z=z_1Z_2\cdOts Z_n\in\gAmMa^N$ anD $l\IN{\maTHbB{N}}^{+}$. INTuiTively, a sInGlE applIcatION OF $B$ rePreSentS a serIes of moves in wHicH $M$ reADs oNe symBol anD theN mAkes a Single Non-$\laMbDa$-move followed bY a ceRtain numbEr oF $\lAmbDa$-Moves. lEt ${{\cal O}}$ ConSisT of all pAirs $(E_{acC},e_{reJ})$ wITH $e_{aCc}= Q_1\times \bot\Gamma^*$ anD $E_{REJ}= Q_2\Times \bot\gamma^*$, wHErE $\{Q_1,q_2\}$ Is a partiTiOn oF $[k]$. We WRIte ${\maThrm{dcFl}}$ for the cLass of ALl LaNguages ReCognizEd By 1dPda’S. Well KNown RelatiOns incluDe ${\matHRm{REG}}\subsetneqQ {\Mathrm{CFL}}$.
BasiC prOPErTIes oF $({{\caL V}},{{\cal B}},{{\cal O}})$-1dTa’s {#sEC:basIc-prOPeRtiES}
===============================================================
For a Given AuTOmATa base $( | {{\cal V}},{{\cal B}},{{\c al O}})$-1 dta.By$\m at hrm{ 1QFA }$, we indicat e the family of all languag es re co g nize d b y bou nded-er r or 1 qfa ’s .Weth e nobtai n $ \mathrm {1QFA} = { \ma th rm{REG}}$.
[ ** (vii) Dete rmi nistic Pushd own Autom at a.* * ]{} A *1 -waydeterm i nistic pushdown a u tomato n * (or a 1 dp da)$M$ can be seen a s a $({{\cal V}},{ {\calB} } ,{ { \ cal O} })$-1dta w he n $({ { \cal V} } ,{ { \ c alB }},{{\cal O}} )$ satisfie s th e foll ow ing proper ties. L e t $ {{\cal V}}= \{[k ]\times \ bot\Ga m ma^*\mi d k\in{\ mathbb {N} }^{ +},\ Ga mm a\, \t e xt{ : a lph a bet }\}$, wh er e$\bot $ is a d isti ngu ishe d bot tom marker no t i n $\ G amm a$. L et ${ {\ca lB}}$be com posed o f all maps of t he f orm $B(q, \bo tz)=(\mu_ 1 (q,z_n ),\ bot z_1z_2 \cdotsz _{n -1 } \ m u_ 2(q,z_n))$ for 2 f un c t io ns $\mu_ 1:[k]\ t im es \Gamma\t o[k] $ an d $\mu_ 2:[k ] \t imes\Gam ma\to\ G am ma ^{\leql} $, whe re $z =z_ 1z_2\ c dots z_n\i n\Gamma^ n$ an d $l\in{\mathbb { N}}^{+}$. Int u it i v el y , asin gle applica tion of $ B$ r e pr ese n ts aserie so fm oves in which $M$ r ea ds one symb ol and then m akes a sin g l e non-$\l ambd a $- m ove followed b y a c ertain num b er of $\ lambd a$-moves . Let ${{ \ c al O}}$con sis t o f a l l p airs $(E_{acc } , E_{r ej })$ wit h $ E_{acc} = Q_1 \ti mes \ bot\Gamma ^*$ and$E _{ re j} = Q _2\ti m es \bot\ Ga mma ^* $,where $\{Q_1 ,Q_2\ }$ i sap art ition o f $ [ k ]$.We w rite ${ \m athrm {DCF L }}$ for th e class o f a l l la ng ua ges rec ognized by 1d pd a’s. Wellkn own relat i o ns inclu de ${\mathrm{REG}}\subs e tneqq { \ma thrm{ CFL} }$.
Basi c P ropert ies of $({ {\calV}},{ {\ cal B }},{{ \ c al O} }) $-1dta’s { # s ec: basic -p rope rties}================== = === ============= === ==== = = == === = == = === == = === =
For a given au tomata bas e$ ( | {{\cal V}},{{\cal_B}},{{\cal O}})$-1dta._By $\mathrm{1QFA}$, we indicate_the family_of_all languages_recognized_by bounded-error 1qfa’s._We then obtain_$\mathrm{1QFA} = {\mathrm{REG}}$.
[**(vii) Deterministic_Pushdown Automata.**]{} A_*1-way_deterministic pushdown automaton* (or a 1dpda) $M$ can be seen as a $({{\cal V}},{{\cal_B}},{{\cal_O}})$-1dta when_$({{\cal_V}},{{\cal_B}},{{\cal O}})$ satisfies the following_properties. Let ${{\cal V}}=\{[k]\times \bot\Gamma^*\mid_k\in{\mathbb{N}}^{+}, \Gamma\,\text{:_alphabet}\}$, where $\bot$ is a distinguished bottom marker_not_in $\Gamma$. Let_${{\cal B}}$ be composed of all maps of the_form $B(q,\bot z) = (\mu_1(q,z_n),\bot z_1z_2\cdots_z_{n-1}\mu_2(q,z_n))$ for 2_functions_$\mu_1:[k]\times_\Gamma\to [k]$ and $\mu_2:[k]\times\Gamma\to\Gamma^{\leq_l}$, where $z=z_1z_2\cdots z_n\in\Gamma^n$ and $l\in{\mathbb{N}}^{+}$._Intuitively, a single application of $B$_represents a series of moves in which_$M$ reads one symbol and then_makes a single non-$\lambda$-move followed_by a_certain number of $\lambda$-moves. Let_${{\cal O}}$ consist_of all_pairs $(E_{acc},E_{rej})$ with_$E_{acc}= Q_1\times \bot\Gamma^*$ and $E_{rej}=_Q_2\times \bot\Gamma^*$, where_$\{Q_1,Q_2\}$ is a partition of $[k]$._We_write ${\mathrm{DCFL}}$ for_the_class_of all_languages recognized by_1dpda’s._Well known_relations_include ${\mathrm{REG}}\subsetneqq {\mathrm{CFL}}$.
Basic Properties of $({{\cal_V}},{{\cal_B}},{{\cal O}})$-1dta’s {#sec:basic-properties}
===============================================================
For a given automata base_$( |
5.
\nonumber\end{aligned}$$
The spin 1 polarization vector $\epsilon^{(\lambda)}_\mu(p)$ satisfies the constraints:
$$\def\arraystretch{2.0}
\begin{array}{ll}
\displaystyle \epsilon^{(\lambda)}_\mu(p) \, p^\mu = 0 &
\hspace*{1cm} {\rm transversality},
\\
\displaystyle \sum\limits_{\lambda=0,\pm}\epsilon^{(\lambda)}_\mu(p)
\epsilon^{\dagger\,(\lambda)}_\nu (p)
=-g_{\mu\nu}+\frac{p_\mu\,p_\nu}{m^2} &
\hspace*{1cm} {\rm completeness},
\\
\displaystyle\epsilon^{\dagger\,(\lambda)}_\mu \epsilon^{(\lambda')\,\mu}
=-\delta_{\lambda \lambda'} &
\hspace*{1cm} {\rm orthonormality}.
\end{array}$$
$$\begin{aligned}
\label{eq:s=2}
{\cal L}_{\rm\, S=2}(x) &=&
\frac{1}{2}\,\phi_{\mu\nu}(x)(\Box-m^2)\phi^{\mu\nu}(x)
+g\,\phi_{\mu\nu}(x)\,J_q^{\mu\nu}(x).\\
\phi^{\mu\nu}(x) &=&\phi^{\nu\mu}(x), \quad
\partial_\mu\phi^{\mu\nu}(x) = 0, \quad
\phi^{\mu}_{\mu}(x) = 0, \quad
({\rm leaving \,\, 5 \,\,independent\,\, components}),
\nonumber\\
&&\nonumber\\
\Pi^{\mu\nu,\alpha\beta}(x-y) &=&
i\,g^2\,<T\left\{ J^{\mu\nu}_q(x)\,J^{\alpha\beta}_q(y) \right\}>_0,
\nonumber\\
&&\nonumber\\
\widetilde\Pi^{\mu\nu,\alpha\beta}(p)&=&
\int d^4x\, e^{-ipx}\,\Pi^{\mu\nu,\alpha\beta}(x)=\frac{1}{2}
\left(g^{\mu\alpha}\,g^{\nu\beta}+g^{\mu\beta}\,g^{\nu\ | 5.
\nonumber\end{aligned}$$
The spin 1 polarization vector $ \epsilon^{(\lambda)}_\mu(p)$ satisfies the constraints:
$ $ \def\arraystretch{2.0 }
\begin{array}{ll }
\displaystyle \epsilon^{(\lambda)}_\mu(p) \, p^\mu = 0 &
\hspace*{1 cm } { \rm transversality },
\\
\displaystyle \sum\limits_{\lambda=0,\pm}\epsilon^{(\lambda)}_\mu(p)
\epsilon^{\dagger\,(\lambda)}_\nu (phosphorus)
= -g_{\mu\nu}+\frac{p_\mu\,p_\nu}{m^2 } &
\hspace*{1 centimeter } { \rm completeness },
\\
\displaystyle\epsilon^{\dagger\,(\lambda)}_\mu \epsilon^{(\lambda')\,\mu }
= -\delta_{\lambda \lambda' } &
\hspace*{1 cm } { \rm orthonormality }.
\end{array}$$
$ $ \begin{aligned }
\label{eq: s=2 }
{ \cal L}_{\rm\, S=2}(x) & = &
\frac{1}{2}\,\phi_{\mu\nu}(x)(\Box - m^2)\phi^{\mu\nu}(x)
+ g\,\phi_{\mu\nu}(x)\,J_q^{\mu\nu}(x).\\
\phi^{\mu\nu}(x) & = & \phi^{\nu\mu}(x), \quad
\partial_\mu\phi^{\mu\nu}(x) = 0, \quad
\phi^{\mu}_{\mu}(x) = 0, \quad
({ \rm leaving \,\, 5 \,\,independent\,\, component }),
\nonumber\\
& & \nonumber\\
\Pi^{\mu\nu,\alpha\beta}(x - y) & = &
i\,g^2\,<T\left\ { J^{\mu\nu}_q(x)\,J^{\alpha\beta}_q(y) \right\}>_0,
\nonumber\\
& & \nonumber\\
\widetilde\Pi^{\mu\nu,\alpha\beta}(p)&= &
\int d^4x\, e^{-ipx}\,\Pi^{\mu\nu,\alpha\beta}(x)=\frac{1}{2 }
\left(g^{\mu\alpha}\,g^{\nu\beta}+g^{\mu\beta}\,g^{\nu\ | 5.
\nonkmber\end{aligned}$$
The spin 1 polarization vgcror $\epvilon^{(\lzmbda)}_\mu(p)$ satisfies the constraints:
$$\deh\arrqystrtnch{2.0}
\begin{array}{ll}
\displahstyle \epdilon^{(\lamvda)}_\mn(p) \, p^\mu = 0 &
\hspace*{1rj} {\rm transverawlitv},
\\
\dmsplaystyle \sum\kimits_{\lambga=0,\pm}\epsilon^{(\lamtdx)}_\mb(p)
\epsilon^{\dagger\,(\lambda)}_\nu (p)
=-g_{\mu\nu}+\frac{p_\mt\,p_\nu}{m^2} &
\hxpwce*{1cm} {\rm complgtenexf},
\\
\disllaystyle\epsilon^{\dagger\,(\lambda)}_\mu \epsjlon^{(\lamuda')\,\mu}
=-\delta_{\lambds \lambda'} &
\hspace*{1cm} {\rm orthlnorlality}.
\end{array}$$
$$\begij{aligned}
\labgm{eq:f=2}
{\xal L}_{\rm\, S=2}(x) &=&
\ffac{1}{2}\,\phi_{\mu\nu}(q)(\Yox-m^2)\phi^{\mu\nu}(s)
+g\,\phi_{\mu\nu}(x)\,J_q^{\mu\nu}(x).\\
\phi^{\mu\nu}(x) &=&\phi^{\nu\ou}(x), \qbad
\partial_\mo\'yi^{\mk\tu}(x) = 0, \qued
\phi^{\ml}_{\mu}(x) = 0, \quad
({\rm leaving \,\, 5 \,\,indelendent\,\, componcnts}),
\nmnunber\\
&&\nonumber\\
\Pi^{\mu\nu,\alpia\beta}(x-y) &=&
i\,g^2\,<T\left\{ J^{\mu\gu}_q(x)\,J^{\alphd\bzta}_q(y) \right\}>_0,
\nonumber\\
&&\nobunber\\
\whdethlde\Ou^{\mu\vu,\ampia\bsta}(p)&=&
\inh d^4e\, e^{-ipx}\,\Pi^{\mu\nh,\alpha\beta}(x)=\drac{1}{2}
\left(g^{\mu\alpha}\,g^{\nu\neeq}+g^{\mu\beta}\,g^{\nu\ | 5. \nonumber\end{aligned}$$ The spin 1 polarization vector the $$\def\arraystretch{2.0} \begin{array}{ll} \epsilon^{(\lambda)}_\mu(p) \, p^\mu transversality}, \displaystyle \sum\limits_{\lambda=0,\pm}\epsilon^{(\lambda)}_\mu(p) \epsilon^{\dagger\,(\lambda)}_\nu =-g_{\mu\nu}+\frac{p_\mu\,p_\nu}{m^2} & \hspace*{1cm} completeness}, \\ \displaystyle\epsilon^{\dagger\,(\lambda)}_\mu \epsilon^{(\lambda')\,\mu} =-\delta_{\lambda \lambda'} \hspace*{1cm} {\rm orthonormality}. \end{array}$$ $$\begin{aligned} \label{eq:s=2} {\cal L}_{\rm\, S=2}(x) &=& \frac{1}{2}\,\phi_{\mu\nu}(x)(\Box-m^2)\phi^{\mu\nu}(x) +g\,\phi_{\mu\nu}(x)\,J_q^{\mu\nu}(x).\\ \phi^{\mu\nu}(x) \quad \partial_\mu\phi^{\mu\nu}(x) = 0, \quad \phi^{\mu}_{\mu}(x) = 0, \quad ({\rm leaving \,\, 5 components}), &&\nonumber\\ &=& J^{\mu\nu}_q(x)\,J^{\alpha\beta}_q(y) \right\}>_0, \nonumber\\ &&\nonumber\\ \widetilde\Pi^{\mu\nu,\alpha\beta}(p)&=& \int d^4x\, e^{-ipx}\,\Pi^{\mu\nu,\alpha\beta}(x)=\frac{1}{2} \left(g^{\mu\alpha}\,g^{\nu\beta}+g^{\mu\beta}\,g^{\nu\ | 5.
\nonumber\end{aligned}$$
The spin 1 PolarizatiOn vecTor $\EpsIlOn^{(\laMbda)}_\Mu(p)$ satisfies thE ConsTraints:
$$\def\arraystretch{2.0}
\Begin{ArRAy}{ll}
\DIsPlaysTyle \epsILoN^{(\LAmbDa)}_\Mu(P) \, p^\mU = 0 &
\hSPaCe*{1cm} {\rM trAnsversAlity},
\\
\displAysTyLe \sum\limits_{\lAMbDa=0,\pm}\epsiloN^{(\laMbda)}_\mu(p)
\epsilOn^{\dAgger\,(\lAmBda)}_\NU (p)
=-g_{\mu\Nu}+\fRac{p_\mU\,p_\nu}{m^2} &
\hSPace*{1cm} {\Rm completEnESs},
\\
\dispLAystyle\EPSiLon^{\dAgger\,(\lambda)}_\mu \epsiLOn^{(\LAmbda')\,\mu}
=-\delta_{\laMbda \laMbDA'} &
\hSPAce*{1Cm} {\rM orthonormAlIty}.
\enD{Array}$$
$$\beGIn{ALIGneD}
\Label{eq:s=2}
{\cal L}_{\rM\, S=2}(x) &=&
\frac{1}{2}\,\phi_{\mU\Nu}(x)(\box-m^2)\phI^{\mU\nu}(X)
+G\,\phi_{\mu\Nu}(x)\,J_q^{\Mu\NU}(x).\\
\pHi^{\mu\nu}(x) &=&\phi^{\nU\mu}(x), \Quad
\partiAl_\mu\phI^{\Mu\nu}(x) = 0, \quAD
\phi^{\mu}_{\mU}(x) = 0, \quad
({\Rm lEavIng \,\, 5 \,\,iNDePeNdeNt\,\, COmpONeNts}),
\NOnuMber\\
&&\nonuMbEr\\
\pi^{\mu\nU,\alpHA\BETa}(x-y) &=&
I\,g^2\,<T\Left\{ j^{\mu\nu}_Q(x)\,J^{\alpha\beta}_q(Y) \riGht\}>_0,
\nONumBer\\
&&\noNumbeR\\
\widEtIlde\PI^{\mu\nu,\aLpha\bEtA}(p)&=&
\int d^4x\, e^{-ipx}\,\Pi^{\mu\Nu,\alPha\beta}(x)=\fRac{1}{2}
\LeFt(g^{\Mu\Alpha}\,G^{\Nu\beta}+G^{\mu\BetA}\,g^{\nu\ | 5.
\nonumber\end{aligned}$ $
The spi n 1 p ola riz at ionvect or $\epsilon^{ ( \lam bda)}_\mu(p)$ satisfie s the c o nstr a in ts:
$$\def\ a rr a y str et ch {2. 0} \b egin{ arr ay}{ll}
\displays tyl e\epsilon^{(\ l am bda)}_\mu( p)\, p^\mu = 0 &\hspac e* {1c m } {\r m t ransv ersali t y},
\\
\display st y le \su m \limits _ { \l ambd a=0,\pm}\epsilon^ { (\ l ambda)}_\mu(p)
\epsi lo n ^{ \ d agg er\ ,(\lambda) }_ \nu ( p )
=-g_{ \ mu \ n u }+\ f rac{p_\mu\,p_ \nu}{m^2} & \hs pace*{ 1c m}{ \rm co mplet en e ss} ,
\\
\displ ayst yle\epsil on^{\d a gger\,( \ lambda) }_\mu\ep sil on^{ ( \l am bda ') \ ,\m u }=-\ d elt a_{\lamb da \ lambd a'}& \ h spac e*{ 1cm} {\r m orthonormal ity }.
\ e nd{ array }$$
$$\b eg in{al igned}
\lab el {eq:s=2}
{\calL}_{ \rm\, S=2 }(x )&=&
\ frac{ 1 }{2}\, \ph i_{ \mu\nu} (x)(\Bo x -m^ 2) \ p h i^ {\mu\nu}(x)
+g\,\p hi _ { \m u\nu}(x) \,J_q^ { \m u\ n u}(x).\\
\ phi ^{\m u \ nu}(x ) &= & \p hi^{\nu\ mu}(x) , \ qu ad
\par ti al_\mu \p hi^ {\m u\nu} ( x) = 0, \quad
\ phi^{ \ mu}_{\mu}(x) = 0, \quad
({\r m l e a vi n g \, \,5 \,\,indep ende n t\,\ , co m po nen t s}),\nonu mb e r\ \
&&\nonumber\\
\Pi^ {\ mu\nu, \alph a\beta}(x-y)&=&
i\,g^2 \ , < T\left\{ J^{ \ mu \ nu}_q(x)\,J^{\ alpha \beta}_q(y ) \right\ }>_0,
\nonumb er\\
&&\n o n umber\\\wi det ild e\P i ^ {\ mu\nu,\alpha\ b e ta}( p) &=&
\in t d ^4x\, e ^{- ipx }\, \Pi ^{ \mu\nu,\a lpha\bet a} (x )= \f rac {1}{2 }
\left(g ^{ \mu \a lph a}\,g ^ {\nu\b eta}+ g^{\ mu \b e ta} \,g^{\n u \ | 5.
\nonumber\end{aligned}$$
The spin_1 polarization_vector $\epsilon^{(\lambda)}_\mu(p)$ satisfies the_constraints:
$$\def\arraystretch{2.0}
\begin{array}{ll}
\displaystyle \epsilon^{(\lambda)}_\mu(p)_\,_p^\mu =_0_&
\hspace*{1cm} {\rm transversality},
\\
\displaystyle_\sum\limits_{\lambda=0,\pm}\epsilon^{(\lambda)}_\mu(p)
\epsilon^{\dagger\,(\lambda)}_\nu (p)
=-g_{\mu\nu}+\frac{p_\mu\,p_\nu}{m^2} &
\hspace*{1cm}_{\rm completeness},
\\
\displaystyle\epsilon^{\dagger\,(\lambda)}_\mu \epsilon^{(\lambda')\,\mu}
=-\delta_{\lambda \lambda'}_&
\hspace*{1cm} {\rm_orthonormality}.
\end{array}$$
$$\begin{aligned}
\label{eq:s=2}
{\cal_L}_{\rm\, S=2}(x) &=&
\frac{1}{2}\,\phi_{\mu\nu}(x)(\Box-m^2)\phi^{\mu\nu}(x)
+g\,\phi_{\mu\nu}(x)\,J_q^{\mu\nu}(x).\\
\phi^{\mu\nu}(x) &=&\phi^{\nu\mu}(x), \quad
\partial_\mu\phi^{\mu\nu}(x) = 0, \quad
\phi^{\mu}_{\mu}(x) = 0, \quad
({\rm_leaving_\,\, 5_\,\,independent\,\,_components}),
\nonumber\\
&&\nonumber\\
\Pi^{\mu\nu,\alpha\beta}(x-y)_&=&
i\,g^2\,<T\left\{ J^{\mu\nu}_q(x)\,J^{\alpha\beta}_q(y) \right\}>_0,
\nonumber\\
&&\nonumber\\
\widetilde\Pi^{\mu\nu,\alpha\beta}(p)&=&
\int d^4x\, e^{-ipx}\,\Pi^{\mu\nu,\alpha\beta}(x)=\frac{1}{2}
\left(g^{\mu\alpha}\,g^{\nu\beta}+g^{\mu\beta}\,g^{\nu\ |
(y)-\widetilde{\mathcal{M}}_{\Omega}^lf_2(y_0)\Big|\\
&&\quad\lesssim \Big(\int^2_1\sum_{j\in\mathbb{Z}}\big|K^j_t*\phi_{j-l}*f_2(y)-K^j_t*\phi_{j-l}*f_2(x_0)\big|^2{\rm d}t\Big)^{\frac{1}{2}}\nonumber\\
&&\quad\lesssim\Big(\int^2_1\Big(\sum_{j\in\mathbb{Z}}\big|K^j_t*\phi_{j-l}*f_2(y)-K^j_t*\phi_{j-l}*f_2(x_0)\big|\Big)^2{\rm d}t\Big)^{\frac{1}{2}}\nonumber\\
&&\quad\lesssim lM_{q'}f(x).\nonumber\end{aligned}$$ Combining the estimates (2.19) and (2.20) leads to that $$\inf_{c\in\mathbb{C}}\frac{1}{|Q|}\int_{Q}\big|\widetilde{\mathcal{M}}_{\Omega}^lf(y)-c\big|{\rm d}y\lesssim lM_{q'}f(x)$$ and then establish (2.18).
Finally, we see that (2.10) holds for the case of $p\in (1,\,\infty)$ and $w^{q'}\in A_{p}(\mathbb{R}^n)$, if we invoke the interpolation argument used in the proof of Theorem 2 in [@kw]. This completes the proof of Theorem \[t2.2\].
Proof of Theorem \[t1.2\]
=========================
We begin with some preliminary lemmas.
\[l3.1\] Let $\Omega$ be homogeneous of degree zero and belong to $L^1(S^{n-1})$, $K_t^j$ be defined as in (2.1). Then for $l\in\mathbb{N}$, $t\in [1,\,2]$, $s\in (1,\,\infty]$, $j_0\in\mathbb{Z}_-$ and $y\in \mathbb{R}^n$ with $|y|<2^{j_0-4}$, $$\begin{aligned}
&&\sum_{ | ( y)-\widetilde{\mathcal{M}}_{\Omega}^lf_2(y_0)\Big|\\
& & \quad\lesssim \Big(\int^2_1\sum_{j\in\mathbb{Z}}\big|K^j_t*\phi_{j - l}*f_2(y)-K^j_t*\phi_{j - l}*f_2(x_0)\big|^2{\rm d}t\Big)^{\frac{1}{2}}\nonumber\\
& & \quad\lesssim\Big(\int^2_1\Big(\sum_{j\in\mathbb{Z}}\big|K^j_t*\phi_{j - l}*f_2(y)-K^j_t*\phi_{j - l}*f_2(x_0)\big|\Big)^2{\rm d}t\Big)^{\frac{1}{2}}\nonumber\\
& & \quad\lesssim lM_{q'}f(x).\nonumber\end{aligned}$$ Combining the estimates (2.19) and (2.20) leads to that $ $ \inf_{c\in\mathbb{C}}\frac{1}{|Q|}\int_{Q}\big|\widetilde{\mathcal{M}}_{\Omega}^lf(y)-c\big|{\rm d}y\lesssim lM_{q'}f(x)$$ and then establish (2.18).
ultimately, we examine that (2.10) holds for the case of $ p\in (1,\,\infty)$ and $ w^{q'}\in A_{p}(\mathbb{R}^n)$, if we invoke the interjection argument used in the proof of Theorem 2 in [ @kw ]. This complete the proof of Theorem \[t2.2\ ].
Proof of Theorem \[t1.2\ ]
= = = = = = = = = = = = = = = = = = = = = = = = =
We begin with some preliminary lemma.
\[l3.1\ ] Let $ \Omega$ be homogeneous of degree zero and belong to to $ L^1(S^{n-1})$, $ K_t^j$ be defined as in (2.1). Then for $ l\in\mathbb{N}$, $ t\in [ 1,\,2]$, $ s\in (1,\,\infty]$, $ j_0\in\mathbb{Z}_-$ and $ y\in \mathbb{R}^n$ with $ |y|<2^{j_0 - 4}$, $ $ \begin{aligned }
& & \sum _ { | (y)-\wifetilde{\mathcal{M}}_{\Omega}^lf_2(y_0)\Blg|\\
&&\quad\lesssim \Biy(\unt^2_1\sum_{o\in\mathgb{Z}}\big|K^j_g*\phi_{j-l}*f_2(y)-K^j_t*\phi_{j-l}*f_2(x_0)\big|^2{\rm d}t\Bij)^{\frax{1}{2}}\nonunber\\
&&\quad\lesssim\Big(\int^2_1\Bkg(\sum_{j\in\mwthbb{Z}}\bit|K^j_t*\khi_{j-l}*f_2(y)-K^j_t*\phi_{j-l}*f_2(x_0)\ujg|\Big)^2{\rm d}t\Big)^{\rvac{1}{2}}\nouunber\\
&&\quad\lesssik lM_{q'}f(x).\nongmber\end{aligneg}$$ Zolbining the estimates (2.19) and (2.20) leads tj that $$\onv_{c\in\mathbb{C}}\fras{1}{|Q|}\inu_{Q}\byg|\wisvtllde{\mathcal{M}}_{\Omega}^lf(y)-c\big|{\rm d}y\lesasim lK_{q'}f(x)$$ and then establish (2.18).
Finally, we see hhat (2.10) holds for the cade of $p\in (1,\,\ibfty)$ qnd $w^{q'}\in A_{p}(\mxthbb{R}^n)$, if we invoke tge interpolation argument used kn thz proof of Jkworfk 2 in [@kw]. Thms comiletes the proof of Tveorem \[y2.2\].
Proof of Theovem \[t1.2\]
=========================
Xe bwgin with some prelimmnary lemmas.
\[l3.1\] Let $\Omgga$ be hommgzneous of degree zero abd benong to $U^1(W^{n-1})$, $Y_t^j$ bx dsfined as in (2.1). Then ror $l\in\mathvb{N}$, $t\in [1,\,2]$, $s\in (1,\,\infty]$, $k_0\ig\nathbb{Z}_-$ and $y\jn \matrbf{R}^n$ with $|y|<2^{j_0-4}$, $$\begin{aligned}
&&\sum_{ | (y)-\widetilde{\mathcal{M}}_{\Omega}^lf_2(y_0)\Big|\\ &&\quad\lesssim \Big(\int^2_1\sum_{j\in\mathbb{Z}}\big|K^j_t*\phi_{j-l}*f_2(y)-K^j_t*\phi_{j-l}*f_2(x_0)\big|^2{\rm d}t\Big)^{\frac{1}{2}}\nonumber\\ &&\quad\lesssim\Big(\int^2_1\Big(\sum_{j\in\mathbb{Z}}\big|K^j_t*\phi_{j-l}*f_2(y)-K^j_t*\phi_{j-l}*f_2(x_0)\big|\Big)^2{\rm d}t\Big)^{\frac{1}{2}}\nonumber\\ &&\quad\lesssim the (2.19) and leads to that establish Finally, we see (2.10) holds for case of $p\in (1,\,\infty)$ and $w^{q'}\in if we invoke the interpolation argument used in the proof of Theorem 2 [@kw]. This completes the proof of Theorem \[t2.2\]. Proof of Theorem \[t1.2\] ========================= begin some lemmas. Let $\Omega$ be homogeneous of degree zero and belong to $L^1(S^{n-1})$, $K_t^j$ be defined as in Then for $l\in\mathbb{N}$, $t\in [1,\,2]$, $s\in (1,\,\infty]$, $j_0\in\mathbb{Z}_-$ $y\in \mathbb{R}^n$ with $|y|<2^{j_0-4}$, &&\sum_{ | (y)-\widetilde{\mathcal{M}}_{\Omega}^lf_2(Y_0)\Big|\\
&&\quad\leSssim \big(\Int^2_1\SuM_{j\in\MathBb{Z}}\big|K^j_t*\phi_{j-l}*F_2(Y)-K^j_t*\Phi_{j-l}*f_2(x_0)\big|^2{\rm d}t\Big)^{\frac{1}{2}}\nOnumbEr\\
&&\QUad\lESsSim\BiG(\int^2_1\Big(\SUm_{J\IN\maThBb{z}}\biG|K^J_T*\pHi_{j-l}*f_2(Y)-K^j_T*\phi_{j-l}*f_2(X_0)\big|\Big)^2{\rm d}T\BiG)^{\fRac{1}{2}}\nonumber\\
&&\qUAd\Lesssim lM_{q'}F(x).\nOnumber\end{alIgnEd}$$ CombInIng THe estImaTes (2.19) anD (2.20) leads TO that $$\iNf_{c\in\mathBb{c}}\Frac{1}{|Q|}\iNT_{Q}\big|\wiDETiLde{\mAthcal{M}}_{\Omega}^lf(y)-c\bIG|{\rM D}y\lesssim lM_{q'}f(x)$$ And theN eSTaBLIsh (2.18).
finAlly, we see tHaT (2.10) holdS For the cASe OF $P\In (1,\,\iNFty)$ and $w^{q'}\in A_{p}(\mAthbb{R}^n)$, if we INvoKe the iNtErpOLation ArgumEnT UseD in the proof Of ThEorem 2 in [@kw]. this coMPletes tHE proof oF TheorEm \[t2.2\].
proOf of tHeOrEm \[t1.2\]
=========================
we BEgiN WiTh sOMe pReliminaRy LeMmas.
\[l3.1\] let $\OMEGA$ Be hoMogEneoUs of dEgree zero and bEloNg to $l^1(s^{n-1})$, $K_T^j$ be dEfineD as iN (2.1). THen foR $l\in\maThbb{N}$, $T\iN [1,\,2]$, $s\in (1,\,\infty]$, $j_0\in\matHbb{Z}_-$ And $y\in \matHbb{r}^n$ WitH $|y|<2^{J_0-4}$, $$\begiN{AligneD}
&&\suM_{ | (y)-\widetilde{\mathcal{M} }_{\Omega} ^lf_2 (y_ 0)\ Bi g|\\
&&\ quad\lesssim \ B ig(\ int^2_1\sum_{j\in\math bb{Z} }\ b ig|K ^ j_ t*\ph i_{j-l} * f_ 2 ( y)- K^ j_ t*\ ph i _{ j-l}* f_2 (x_0)\b ig|^2{\rmd}t \B ig)^{\frac{1 } {2 }}\nonumbe r\\
&&\quad\les ssi m\Big( \i nt^ 2 _1\Bi g(\ sum_{ j\in\m a thbb{Z }}\big|K^ j_ t *\phi_ { j-l}*f_ 2 ( y) -K^j _t*\phi_{j-l}*f_2 ( x_ 0 )\big|\Big)^2{ \rm d} t\ B ig ) ^ {\f rac {1}{2}}\no nu mber\ \
&&\qua d \l e s s sim lM_{q'}f(x).\ nonumber\en d {al igned} $$ Co m bining thees t ima tes (2.19)and(2.20) le ads to that $$ \ inf_{c\ in\mat hbb {C} }\fr a c{ 1} {|Q |} \ int _ {Q }\b i g|\ widetild e{ \m athca l{M} } _ { \ Omeg a}^ lf(y )-c\b ig|{\rm d}y\l ess simlM_ {q'}f (x)$$ and t hen e stabli sh (2 .1 8).
Finally, w e se e that (2 .10 )hol ds fort he cas e o f $ p\in (1 ,\,\inf t y)$ a n d $w ^{q'}\in A_{p}(\ma th b b {R }^n)$, i f we i n vo ke the inte rp ola tion a rgume nt u s ed in theproofo fTh eorem 2 i n [@kw ]. Th iscompl e testhe pr oof of T heore m \[t2.2\].
Pr o of of Theorem \[ t 1 .2 \ ]
== === =========== ==== = ====
We be gin withsomepr e li m inary lemmas.
\[l3 .1 \] Let $\Om ega$ be homog eneous ofd e g ree zero and be l ong to $L^1(S^ {n-1} )$, $K_t^j $ be defi ned a s in (2. 1). Thenf o r $l\in\ mat hbb {N} $,$ t \i n [1,\,2]$, $ s \ in ( 1, \,\inft y]$ , $j_0\ in\ mat hbb {Z} _- $ and $y\ in \math bb {R }^ n$ wi th $| y |<2^{j_0 -4 }$, $ $\b egin{ a ligned }
&&\ sum_ { | (y)-\widetilde{\mathcal{M}}_{\Omega}^lf_2(y_0)\Big|\\
&&\quad\lesssim \Big(\int^2_1\sum_{j\in\mathbb{Z}}\big|K^j_t*\phi_{j-l}*f_2(y)-K^j_t*\phi_{j-l}*f_2(x_0)\big|^2{\rm_d}t\Big)^{\frac{1}{2}}\nonumber\\
&&\quad\lesssim\Big(\int^2_1\Big(\sum_{j\in\mathbb{Z}}\big|K^j_t*\phi_{j-l}*f_2(y)-K^j_t*\phi_{j-l}*f_2(x_0)\big|\Big)^2{\rm d}t\Big)^{\frac{1}{2}}\nonumber\\
&&\quad\lesssim_lM_{q'}f(x).\nonumber\end{aligned}$$ Combining the estimates_(2.19) and_(2.20)_leads to_that_$$\inf_{c\in\mathbb{C}}\frac{1}{|Q|}\int_{Q}\big|\widetilde{\mathcal{M}}_{\Omega}^lf(y)-c\big|{\rm d}y\lesssim _lM_{q'}f(x)$$ and then_establish (2.18).
Finally, we see_that (2.10) holds_for_the case of $p\in (1,\,\infty)$ and $w^{q'}\in A_{p}(\mathbb{R}^n)$, if we invoke the interpolation argument_used_in the_proof_of_Theorem 2 in [@kw]. This_completes the proof of Theorem_\[t2.2\].
Proof of_Theorem \[t1.2\]
=========================
We begin with some preliminary lemmas.
\[l3.1\] Let_$\Omega$_be homogeneous of_degree zero and belong to $L^1(S^{n-1})$, $K_t^j$ be defined_as in (2.1). Then for $l\in\mathbb{N}$,_$t\in [1,\,2]$, $s\in_(1,\,\infty]$,_$j_0\in\mathbb{Z}_-$_and $y\in \mathbb{R}^n$ with_$|y|<2^{j_0-4}$, $$\begin{aligned}
&&\sum_{ |
P. Erd[ő]{}s, A. R[é]{}nyi, On random graphs i., Publ. Math. Debrecen 6 (1959) 290–297.
G. Palla, I. Der[é]{}nyi, I. Farkas, T. Vicsek, Uncovering the overlapping community structure of complex networks in nature and society, Nature 435 (7043) (2005) 814–818.
E. Ravasz, A. L. Somera, D. A. Mongru, Z. N. Oltvai, A.-L. Barab[á]{}si, Hierarchical organization of modularity in metabolic networks, Science 297 (5586) (2002) 1551–1555.
M. Bogu[ñ]{}[á]{}, R. Pastor-Satorras, A. D[í]{}az-Guilera, A. Arenas, Models of social networks based on social distance attachment, Phys. Rev. E 70 (5) (2004) 056122.
M. E. J. Newman, J. Park, Why social networks are different from other types of networks, Phys. Rev. E 68 (3) (2003) 036122.
L. [Š]{}ubelj, D. Fiala, M. Bajec, Network-based statistical comparison of citation topology of bibliographic databases, Sci. Rep. 4 (2014) 6496.
R. D. Cook, S. Weisberg, Residuals and influence in regression, Mg. Stat. Pro.
M. Najork, J. L. Wiener, Breadth-first crawling yields high-quality pages, in: Proceedings of the 10th international conference on World Wide Web, ACM, 2001, pp. 114–118.
---
abstract: 'We study fermions in a magnetic field in a finite-size cylinder. With the boundary condition for the fermion flux, we show that the energy spectra and the wave functions are modified by the finite-size effect; the boundary makes the degenerate Landau levels appear only partially for states with small angular momenta, while the boundary effect becomes stronger for states with large angular momenta. We find that mode accumulation at the | P. Erd[ő]{}s, A. R[é]{}nyi, On random graphs i., Publ. Math. Debrecen 6 (1959) 290–297.
G. Palla, I. Der[é]{}nyi, I. Farkas, T. Vicsek, Uncovering the overlapping community social organization of complex net in nature and society, Nature 435 (7043) (2005) 814–818.
E. Ravasz, A. L. Somera, D. A. Mongru, Z. N. Oltvai, A.-L. Barab[á]{}si, Hierarchical organization of modularity in metabolic net, Science 297 (5586) (2002) 1551–1555.
M. Bogu[ñ]{}[á ] { }, R. Pastor - Satorras, A. D[í]{}az - Guilera, A. Arenas, Models of social network based on social distance attachment, Phys. Rev. vitamin e 70 (5) (2004) 056122.
M. E. J. Newman, J. Park, Why social networks are different from early type of networks, Phys. Rev. E 68 (3) (2003) 036122.
L. [ Š]{}ubelj, D. Fiala, M. Bajec, Network - free-base statistical comparison of citation topology of bibliographic database, Sci. Rep. 4 (2014) 6496.
R. D. Cook, S. Weisberg, Residuals and influence in regression, Mg. Stat. Pro.
M. Najork, J. L. Wiener, Breadth - first crawl yields high - timbre pages, in: Proceedings of the 10th external conference on World Wide Web, ACM, 2001, pp. 114–118.
---
abstract:' We study fermions in a magnetic field in a finite - size cylinder. With the boundary circumstance for the fermion flux, we express that the energy spectra and the wave functions are modified by the finite - size consequence; the boundary makes the degenerate Landau levels appear only partially for states with small angular momenta, while the boundary effect becomes stronger for states with large angular momenta. We find that mode accretion at the |
P. Erf[ő]{}s, A. R[é]{}nyi, On random graphr i., Publ. Math. Dgbeecen 6 (1959) 290–297.
G. Palma, I. Der[é]{}nhi, I. Farkas, T. Vicsek, Uncoverinj thw oveelapping community strjcture of complex netxorks in nature ehd sociccy, Nafmre 435 (7043) (2005) 814–818.
X. Ravasz, A. L. Someta, D. A. Mongru, Z. N. Oltvai, A.-L. Tafay[á]{}si, Hierarchical organization of modtlarity ij metabolic nejworkx, Scisnce 297 (5586) (2002) 1551–1555.
M. Bogu[ñ]{}[á]{}, R. Pastor-Satorras, A. D[í]{}az-Ghilera, E. Arenas, Models pf social networks based oj sofial distance attafhment, Phys. Rev. W 70 (5) (2004) 056122.
M. E. J. Newmxn, J. Park, Why social nejworks are different from other gypes of networjs, Phjv. Rev. E 68 (3) (2003) 036122.
L. [Š]{}nbelj, Q. Fiala, M. Bajeg, Netwosk-based statistical cpmperisin of citation topolojy of bibliographic qatabases, Sei. Rep. 4 (2014) 6496.
R. D. Cook, S. Weiwbwrg, Rgsidudls xbd knfmuxncs in rfgrxssion, Mg. Sfat. Pro.
M. Najirk, J. L. Wiener, Breaduh-fygxt crawling gields hygh-quality pages, in: Proceedings of the 10uh infernational conference in World Wide Web, ACM, 2001, pp. 114–118.
---
abseract: 'We study fermions in a magnetic field in a xinitx-skze cnlinawr. With the boundary condition for the fermion smuc, ee show that tme energy spectra snf yre wave functkons axs jodified by the fijite-sizg effext; the botndaty makes the degenerate Laneau levels aipeae only partially flr states wnth smsll amgular momenta, while thz bounsary effect becomes agronger for statds eidh large angular momenta. Wq find thet mobe accumjlatoon at the | P. Erd[ő]{}s, A. R[é]{}nyi, On random graphs Math. 6 (1959) G. Palla, I. Uncovering overlapping community structure complex networks in and society, Nature 435 (7043) (2005) E. Ravasz, A. L. Somera, D. A. Mongru, Z. N. Oltvai, A.-L. Barab[á]{}si, organization of modularity in metabolic networks, Science 297 (5586) (2002) 1551–1555. M. Bogu[ñ]{}[á]{}, Pastor-Satorras, D[í]{}az-Guilera, Arenas, of social networks based on social distance attachment, Phys. Rev. E 70 (5) (2004) 056122. M. J. Newman, J. Park, Why social networks are from other types of Phys. Rev. E 68 (3) 036122. [Š]{}ubelj, D. M. Network-based comparison of citation of bibliographic databases, Sci. Rep. 4 (2014) 6496. R. D. Cook, S. Weisberg, Residuals and influence in Mg. Stat. Najork, J. Wiener, crawling high-quality pages, in: the 10th international conference on World 2001, pp. 114–118. --- abstract: 'We study fermions a magnetic in a finite-size cylinder. With the condition for the fermion flux, we show that energy spectra and the wave functions are modified by the finite-size effect; the boundary makes Landau levels appear only for states with angular while boundary becomes stronger states with large angular momenta. We find that mode accumulation at |
P. Erd[ő]{}s, A. R[é]{}nyi, On random graphS i., Publ. Math. debreCen 6 (1959) 290–297.
g. PaLlA, I. DeR[é]{}nyI, I. Farkas, T. VicseK, uncoVering the overlapping coMmuniTy STrucTUrE of coMplex neTWoRKS in NaTuRe aNd SOcIety, NAtuRe 435 (7043) (2005) 814–818.
E. RavaSz, A. L. Somera, d. A. MOnGru, Z. N. Oltvai, A.-l. baRab[á]{}si, HierArcHical organizAtiOn of moDuLarITy in mEtaBolic NetworKS, ScienCe 297 (5586) (2002) 1551–1555.
M. Bogu[ñ]{}[á]{}, r. PAStor-SaTOrras, A. D[Í]{}AZ-GUileRa, A. Arenas, Models of SOcIAl networks baseD on socIaL DiSTAncE atTachment, PhYs. rev. E 70 (5) (2004) 056122.
M. e. j. Newman, j. paRK, wHy sOCial networks aRe different FRom Other tYpEs oF NetworKs, PhyS. REV. E 68 (3) (2003) 036122.
L. [š]{}ubelj, D. FialA, M. BaJec, NetworK-based STatistiCAl compaRison oF ciTatIon tOPoLoGy oF bIBliOGrAphIC daTabases, SCi. reP. 4 (2014) 6496.
R. D. CoOk, S. WEISBErg, REsiDualS and iNfluence in regResSion, mG. StAt. Pro.
m. NajoRk, J. L. wiEner, BReadth-First CrAwling yields higH-quaLity pages, In: PRoCeeDiNgs of THe 10th inTerNatIonal coNferencE On WOrLD wIdE Web, ACM, 2001, pp. 114–118.
---
abstract: 'WE sTUDy Fermions In a magNEtIc FIeld in a fInIte-Size CYLindeR. WitH ThE boundarY condiTIoN fOr the feRmIon fluX, wE shOw tHat thE EnerGy specTra and thE wave FUnctions are modIFied by the finiTE-sIZE eFFect; The Boundary makEs thE DegeNeraTE LAndAU leveLs appEaR OnLY partially for states WiTh smalL anguLar momenta, whiLe the boundARY Effect beComeS StROnger for states With lArge angulaR Momenta. WE find That mode AccumulatION at the |
P. Erd[ő]{}s, A. R[é]{}n yi, On ran dom g rap hsi. , Pu bl.Math. Debrecen 6 (1 959) 290–297.
G. Pall a, I. D e r[é] { }n yi, I . Farka s ,T . Vi cs ek , U nc o ve ringthe overla pping comm uni ty structure o f c omplex net wor ks in nature an d soci et y,N ature 43 5 (70 43) (2 0 05) 81 4–818.
E .R avasz, A. L. S o m er a, D . A. Mongru, Z. N . O l tvai, A.-L. Ba rab[á] {} s i, H ier arc hical orga ni zatio n of mod u la r i t y i n metabolic ne tworks, Sci e nce 297 ( 55 86) (2002) 1551 –1 5 55.
M. Bogu[ñ ]{}[ á]{}, R.Pastor - Satorra s , A. D[ í]{}az -Gu ile ra,A .Ar ena s, Mod e ls of soc ial netw or ks base d on s o c ialdis tanc e att achment, Phys . R ev.E 70 (5)(2004 ) 05 61 22.
M. E.J. Ne wm an, J. Park, Wh y so cial netw ork sare d iffer e nt fro m o the r types of net w ork s, P h ys . Rev. E 68 (3) (2 00 3 ) 0 36122.
L. [Š] { }u be l j, D. Fi al a,M. B a j ec, N etwo r k- based st atisti c al c omparis on of ci ta tio n t opolo g y of bibli ographic data b ases, Sci. Rep . 4 (2014) 649 6 .R .D . Co ok, S. Weisber g, R e sidu alsa nd in f luenc e inre g re s sion, Mg. Stat. Pro .
M. Na jork, J. L. Wiener , Breadth- f i r st crawl ingy ie l ds high-qualit y pag es, in: Pr o ceedings of t he 10thinternati o n al confe ren ceonWor l d W ide Web, ACM, 2 001, p p. 114– 118 .
---abs tra ct: 'W estudy fer mions in a m ag ne tic fiel d in a fi ni te- si zecylin d er. Wi th th e bo un da r y c onditio n f o r the f er mion fl ux , weshow tha t the e nergy spe ctr a and t he wave f unctions aremo dified byth e f inite- s i ze effec t; the boundary makes t h e degen era te La ndau levels a ppe ar onl y p a rtiall y forstate swit h small a ng ula rmomenta, w h i lethe b ou ndar y effec t becomes stronger for states withlar ge a n g ul arm om e nta .W e f i n d that mode acc umulationat th e |
P. Erd[ő]{}s, A. R[é]{}nyi,_On random_graphs i., Publ. Math._Debrecen 6_(1959)_290–297.
G. Palla, I. Der[é]{}nyi,_I. Farkas,_T. Vicsek, Uncovering the_overlapping community structure_of complex networks in_nature and society,_Nature_435 (7043) (2005) 814–818.
E. Ravasz, A. L. Somera, D. A. Mongru, Z. N. Oltvai, A.-L. Barab[á]{}si, Hierarchical organization of_modularity_in metabolic_networks,_Science_297 (5586) (2002) 1551–1555.
M. Bogu[ñ]{}[á]{}, R. Pastor-Satorras, A. D[í]{}az-Guilera,_A. Arenas, Models of social networks_based on_social distance attachment, Phys. Rev. E 70 (5) (2004)_056122.
M. E. J._Newman, J. Park, Why_social networks are different from other types of networks,_Phys. Rev. E 68 (3) (2003) 036122.
L. [Š]{}ubelj,_D. Fiala, M. Bajec, Network-based_statistical_comparison_of citation topology of_bibliographic databases, Sci. Rep. 4 (2014)_6496.
R. D. Cook, S. Weisberg, Residuals and influence_in regression, Mg. Stat. Pro.
M. Najork, J. L. Wiener,_Breadth-first crawling yields high-quality pages, in:_Proceedings of the 10th international_conference on_World Wide Web, ACM, 2001,_pp. 114–118.
---
abstract:_'We study_fermions in a_magnetic field in a finite-size cylinder._With the boundary_condition for the fermion flux, we_show_that the energy_spectra_and_the wave_functions are modified_by_the finite-size_effect;_the boundary makes the degenerate Landau_levels_appear only partially for states with small_angular momenta, while the_boundary_effect becomes stronger for_states with large angular momenta._We find that mode accumulation at_the |
vec y_i\rangle = 0 \textrm{ for all } i \}$.
For a Boolean function $f: \F_2^n \rightarrow \F_2^n$ we denote its [*universal (quantum) embedding*]{} by $$U_f: \F_2^{2n} \rightarrow \F_2^{2n} \textrm{ with } (\vec x, \vec y) \mapsto (\vec x, f(\vec x)+\vec y).$$ Notice that $U_f(U_f(\vec x,\vec y)) = (\vec x, \vec y)$.
Let $\ket{x} \in \mathbb{C}^2$ with $x \in \F_2$ be a qubit. We denote by $H$ the [*Hadamard function*]{} $$x \mapsto \frac 1 {\sqrt 2} (\ket{0} + (-1)^x \ket{1}).$$ We briefly write $H_n$ for the $n$-fold tensor product $H \otimes \ldots \otimes H$. Let $\ket{x}\ket{y}\in \mathbb{C}^4$ be a $2$-qubit system. The $\cnot$ (controlled **not**) function is the universal embedding of the identity function, i.e. $\ket{x}\ket{y} \mapsto \ket{x}\ket{x+y}$. We call the first qubit $\ket{x}$ [*control bit*]{}, since we perform a **not** on $\ket{y}$ iff $x=1$.
A [*Simon function*]{} is a periodic $(2:1)$-Boolean function defined as follows.
\[def:simon\] Let $f:\F_2^n\to \F_2^n$. We call $f$ a [*Simon function*]{} if there exists some period $\vec s \in \F_2^n \setminus\Zero$ such that for all $\vec x, \vec y \in \F_2^n$ we have $$f( \vec x) = f( \vec y) \Leftrightarrow \vec y = \vec x + \vec s.$$ In [*Simon’s problem*]{} we have to find $\vec s$ given oracle access to $f$.
In order to solve Simon’s problem classically, we have to find some collision $\vec x \not= \vec y$ satisfying $ | vec y_i\rangle = 0 \textrm { for all } i \}$.
For a Boolean function $ f: \F_2^n \rightarrow \F_2^n$ we denote its [ * universal (quantum) embedding * ] { } by $ $ U_f: \F_2^{2n } \rightarrow \F_2^{2n } \textrm { with } (\vec x, \vec y) \mapsto (\vec x, f(\vec x)+\vec y).$$ detect that $ U_f(U_f(\vec x,\vec y) ) = (\vec x, \vec y)$.
lease $ \ket{x } \in \mathbb{C}^2 $ with $ x \in \F_2 $ be a qubit. We denote by $ H$ the [ * Hadamard function * ] { } $ $ x \mapsto \frac 1 { \sqrt 2 } (\ket{0 } + (-1)^x \ket{1}).$$ We briefly compose $ H_n$ for the $ n$-fold tensor product $ H \otimes \ldots \otimes H$. Let $ \ket{x}\ket{y}\in \mathbb{C}^4 $ be a $ 2$-qubit organization. The $ \cnot$ (manipulate * * not * *) function is the cosmopolitan embedding of the identity affair, i.e. $ \ket{x}\ket{y } \mapsto \ket{x}\ket{x+y}$. We call the beginning qubit $ \ket{x}$ [ * control bit * ] { }, since we do a * * not * * on $ \ket{y}$ iff $ x=1$.
A [ * Simon function * ] { } is a periodic $ (2:1)$-Boolean function defined as follows.
\[def: simon\ ] Let $ f:\F_2^n\to \F_2^n$. We call $ f$ a [ * Simon function * ] { } if there exist some period $ \vec s \in \F_2^n \setminus\Zero$ such that for all $ \vec x, \vec y \in \F_2^n$ we have $ $ f (\vec x) = f (\vec y) \Leftrightarrow \vec y = \vec x + \vec s.$$ In [ * Simon ’s trouble * ] { } we have to find $ \vec s$ given prophet access to $ f$.
In ordering to solve Simon ’s problem classically, we have to find some collision $ \vec x \not= \vec y$ satisfying $ | vec y_i\rangle = 0 \textrm{ for aul } i \}$.
For a Boolean fuiction $r: \F_2^n \rigftarrow \F_2^n$ we denote its [*unitersql (quqntum) embedding*]{} by $$U_f: \W_2^{2n} \rightagrow \F_2^{2n} \twxtrn{ with } (\vec x, \vec y) \mapstk (\vec e, f(\vec x)+\vec y).$$ Nptice that $U_f(U_f(\vec x,\vec f)) = (\vzc x, \vec y)$.
Let $\ket{x} \in \mathbb{C}^2$ with $x \in \F_2$ br w qubit. We denjte nr $H$ fhe [*Hadamard function*]{} $$x \mapsto \frad 1 {\sqrt 2} (\ket{0} + (-1)^x \ket{1}).$$ Ee briefly write $H_n$ for thf $n$-flld tensor product $H \otimes \leots \itimes H$. Let $\ket{x}\ket{y}\in \mathbb{C}^4$ be a $2$-qubit system. The $\cnot$ (controuled **uot**) functiob us hve universao embvdding of the identitf functoon, i.e. $\ket{x}\ket{n} \mapvto \ket{x}\ket{x+y}$. We call thx first qubit $\ket{x}$ [*cjntrol bid*]{}, aince we perform q **bot** ot $\ked{y}$ iwd $x=1$.
X [*Sjmpn functlon*]{} is a perikdic $(2:1)$-Booleab function defined ss dollows.
\[def:simkn\] Let $f:\S_2^n\to \F_2^n$. We call $f$ a [*Simon function*]{} if tvers exists some period $\vex s \in \F_2^n \setminus\Zerl$ such thwt for all $\vec x, \vec y \in \F_2^n$ we have $$f( \vec x) = f( \vac y) \Mdftxlnhtafeoa \vec y = \vec x + \vec s.$$ In [*Simon’s problem*]{} we hads uo nind $\vec s$ given oracle accrsd yj $f$.
In order tu solvz Sjmon’s problem clasdically, we hqve to figd spme collision $\vec x \not= \vec y$ satisfyiny $ | vec y_i\rangle = 0 \textrm{ for all \}$. a Boolean $f: \F_2^n \rightarrow (quantum) by $$U_f: \F_2^{2n} \F_2^{2n} \textrm{ with (\vec x, \vec y) \mapsto (\vec f(\vec x)+\vec y).$$ Notice that $U_f(U_f(\vec x,\vec y)) = (\vec x, \vec y)$. $\ket{x} \in \mathbb{C}^2$ with $x \in \F_2$ be a qubit. We denote by the function*]{} \mapsto 1 {\sqrt 2} (\ket{0} + (-1)^x \ket{1}).$$ We briefly write $H_n$ for the $n$-fold tensor product \otimes \ldots \otimes H$. Let $\ket{x}\ket{y}\in \mathbb{C}^4$ be $2$-qubit system. The $\cnot$ **not**) function is the universal of identity function, $\ket{x}\ket{y} \ket{x}\ket{x+y}$. call the first $\ket{x}$ [*control bit*]{}, since we perform a **not** on $\ket{y}$ iff $x=1$. A [*Simon function*]{} is a $(2:1)$-Boolean function follows. \[def:simon\] $f:\F_2^n\to We $f$ a [*Simon there exists some period $\vec s such that for all $\vec x, \vec y \F_2^n$ we $$f( \vec x) = f( \vec \Leftrightarrow \vec y = \vec x + \vec In [*Simon’s problem*]{} we have to find $\vec s$ given oracle access to $f$. In solve Simon’s problem classically, have to find collision x \vec satisfying $ | vec y_i\rangle = 0 \textrm{ for all } i \}$.
FOr a Boolean FunctIon $F: \F_2^n \RiGhtaRrow \f_2^n$ we denote its [*uNIverSal (quantum) embedding*]{} by $$U_F: \F_2^{2n} \riGhTArroW \f_2^{2n} \TextrM{ with } (\veC X, \vEC Y) \maPsTo (\Vec X, f(\VEc X)+\vec y).$$ notIce that $u_f(U_f(\vec x,\veC y)) = (\vEc X, \vec y)$.
Let $\ket{x} \IN \mAthbb{C}^2$ with $X \in \f_2$ be a qubit. We dEnoTe by $H$ tHe [*hadAMard fUncTion*]{} $$x \Mapsto \FRac 1 {\sqrT 2} (\ket{0} + (-1)^x \ket{1}).$$ WE bRIefly wRIte $H_n$ foR THe $N$-folD tensor product $H \otIMeS \Ldots \otimes H$. LeT $\ket{x}\kEt{Y}\In \MAThbB{C}^4$ bE a $2$-qubit sysTeM. The $\cNOt$ (contrOLlED **NOt**) fUNction is the unIversal embeDDinG of the IdEntITy funcTion, i.E. $\kET{x}\kEt{y} \mapsto \keT{x}\keT{x+y}$. We call The firST qubit $\kET{x}$ [*contrOl bit*]{}, sIncE we PerfORm A **nOt** oN $\kET{y}$ iFF $x=1$.
a [*SiMOn fUnction*]{} iS a PeRiodiC $(2:1)$-BooLEAN FuncTioN defIned aS follows.
\[def:siMon\] let $f:\f_2^N\to \f_2^n$. We cAll $f$ a [*simoN fUnctiOn*]{} if thEre exIsTs some period $\vec S \in \F_2^N \setminus\zerO$ sUch ThAt for ALl $\vec x, \Vec Y \in \f_2^n$ we havE $$f( \vec x) = f( \VEc y) \leFTRIgHtarrow \vec y = \vec x + \vec S.$$ IN [*sImOn’s problEm*]{} we haVE tO fINd $\vec s$ giVeN orAcle ACCess tO $f$.
In ORdEr to solvE Simon’S PrObLem clasSiCally, wE hAve To fInd soME colLision $\Vec x \not= \vEc y$ saTIsfying $ | vec y_i\rangle = 0 \textrm { for all} i \ }$.
F or a B oole an function $f : \F_ 2^n \rightarrow \F_2^n $ wede n otei ts [*un iversal (q u a ntu m) e mbe dd i ng *]{}by$$U_f:\F_2^{2n}\ri gh tarrow \F_2^ { 2n } \textrm{ wi th } (\vec x , \ vec y) \ map s to (\ vec x, f (\vecx )+\vec y).$$ No ti c e that $U_f(U_ f ( \v ec x ,\vec y)) = (\vec x, \vec y)$.
Let $\ket {x } \ i n \m ath bb{C}^2$ w it h $x\ in \F_2 $ b e a qu b it. We denote by $H$ the [*H adamar dfun c tion*] {} $$ x\ map sto \frac 1 {\s qrt 2} (\ ket{0} + (-1)^ x \ket{1 }).$$Webri efly wr it e $ H_ n $ f o rthe $n$ -fold te ns or prod uct$ H \ otim es\ldo ts \o times H$. Let $\ ket{ x }\k et{y} \in \ math bb {C}^4 $ be a $2$- qu bit system. The $\c not$ (con tro ll ed** not** ) funct ion is the un iversal emb ed d i n gof the identity fu nc t i on , i.e. $ \ket{x } \k et { y} \maps to \k et{x } \ ket{x +y}$ . W e call t he fir s tqu bit $\k et {x}$ [ *c ont rol bit* ] {},sincewe perfo rm a* *not** on $\ke t {y}$ iff $x=1 $ .A [ * Simo n f unction*]{} isa per iodi c $ (2: 1 )$-Bo olean f u nc t ion defined as foll ow s.
\[ def:s imon\] Let $f :\F_2^n\to \ F _2^n$. W e ca l l$ f$ a [*Simon f uncti on*]{} ift here exi sts s ome peri od $\vecs \in \F_2 ^n\se tmi nus \ Z er o$ such thatf o r al l$\vec x , \ vec y \ in\F_ 2^n $ w ehave $$f( \vec x) = f (\v ecy) \L e ftrighta rr ow\v ecy = \ v ec x + \vec s.$ $In [*S imon’sp ro b l em*] {} w e ha veto find $\v e c s $ given oracle a cce s s to $ f$ .
In o rder to solve S imon’s pro bl emclassi c a lly, wehave to find some colli s ion $\v ecx \no t= \ vec y$ sa tis fying$ | vec y_i\rangle_= 0_\textrm{ for all }_i \}$.
For_a_Boolean function_$f:_\F_2^n \rightarrow \F_2^n$_we denote its_[*universal (quantum) embedding*]{} by_$$U_f: \F_2^{2n} \rightarrow_\F_2^{2n}_\textrm{ with } (\vec x, \vec y) \mapsto (\vec x, f(\vec x)+\vec y).$$ Notice_that_$U_f(U_f(\vec x,\vec_y))_=_(\vec x, \vec y)$.
Let $\ket{x}_\in \mathbb{C}^2$ with $x \in_\F_2$ be_a qubit. We denote by $H$ the [*Hadamard_function*]{}_$$x \mapsto \frac_1 {\sqrt 2} (\ket{0} + (-1)^x \ket{1}).$$ We briefly_write $H_n$ for the $n$-fold tensor_product $H \otimes_\ldots_\otimes_H$. Let $\ket{x}\ket{y}\in \mathbb{C}^4$_be a $2$-qubit system. The $\cnot$_(controlled **not**) function is the universal_embedding of the identity function, i.e. $\ket{x}\ket{y}_\mapsto \ket{x}\ket{x+y}$. We call the first_qubit $\ket{x}$ [*control bit*]{}, since_we perform_a **not** on $\ket{y}$ iff_$x=1$.
A [*Simon function*]{}_is a_periodic $(2:1)$-Boolean function_defined as follows.
\[def:simon\] Let $f:\F_2^n\to \F_2^n$._We call $f$_a [*Simon function*]{} if there exists_some_period $\vec s_\in_\F_2^n_\setminus\Zero$ such_that for all_$\vec_x, \vec_y_\in \F_2^n$ we have $$f( \vec_x)_= f( \vec y) \Leftrightarrow \vec y_= \vec x +_\vec_s.$$ In [*Simon’s problem*]{}_we have to find $\vec_s$ given oracle access to $f$.
In_order to_solve Simon’s_problem classically, we have to find some collision $\vec x \not=_\vec y$ satisfying $ |
ient Shared Control {#sec:NGSC}
===============================
Expressing Shared Control as an Optimization Problem
----------------------------------------------------
Let $s \in \mathcal{S}$ be the state of the system. Let $a^H \in \mathcal{A^H}$ as the user action, $a^R \in \mathcal{A^R}$ be the autonomous robot action, and $u \in \mathcal{U}$ be the shared control action. The human and the robot agent each select actions following their stochastic policies, $\pi_H$, and $\pi_R$. Our goal is to find a shared control policy $\pi_S$ that solves the following optimization problem.
[u\_t]{}[Q\_H (s\_t, u\_t)]{} \[opt:problem\]
The shared control policy is chosen to maximize the user’s internal action-value function $Q_H(s_t, u_t)$ at each step. Although it is possible to learn $Q_H$ using methods such as MaxEnt IOC [@ziebart2009planning], predicting the user controls $a^H_t$ can be challenging due to interpersonal differences. Instead, we regard the user action as an estimate of $\nabla_s Q_H(s_t, a^H_t)$ at each step.
The constraint on the KL-divergence between the robot policy and the shared policy ensures that the shared policy does not deviate far from the autonomous robot policy. The problem can be expressed using a Lagrange Multiplier, assuming a linear approximation of our objective $\nabla_s Q_H(s_t, a^H_t)$ and a quadratic approximation of the KL-divergence constraint. Solving this approximation of the Lagrangian leads to an update rule which introduces natural gradient adaptation.
Natural Gradient Shared Control {#natural-gradient-shared-control}
-------------------------------
We introduce a new control paradigm for shared control using natural gradient adaptation. Note that our goal is to find the action that maximizes $Q_H$, resulting in taking gradient steps in the direction of ascent. $$\label{eqn:state_update}
\begin{aligned}
s_{t+1} &= s_t + \eta F(s_t)^{-1} \nabla_s Q_H(s_t, a^H_t) \\
&=s | ient Shared Control { # sec: NGSC }
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Expressing Shared Control as an Optimization Problem
----------------------------------------------------
Let $ s \in \mathcal{S}$ be the state of the arrangement. lease $ a^H \in \mathcal{A^H}$ as the user action, $ a^R \in \mathcal{A^R}$ be the autonomous robot natural process, and $ u \in \mathcal{U}$ be the share command action. The homo and the automaton agent each select military action following their stochastic policies, $ \pi_H$, and $ \pi_R$. Our goal is to find a shared control policy $ \pi_S$ that solve the following optimization problem.
[ u\_t]{}[Q\_H (s\_t, u\_t) ] { } \[opt: problem\ ]
The shared command policy is chosen to maximize the user ’s internal action - value affair $ Q_H(s_t, u_t)$ at each step. Although it is possible to learn $ Q_H$ using methods such as MaxEnt IOC [ @ziebart2009planning ], predict the user controls $ a^H_t$ can be challenging due to interpersonal differences. Instead, we regard the user action as an estimate of $ \nabla_s Q_H(s_t, a^H_t)$ at each step.
The constraint on the KL - divergence between the robot policy and the shared policy ensures that the shared policy does not deviate far from the autonomous robot policy. The trouble can be express using a Lagrange Multiplier, assume a analogue estimate of our objective $ \nabla_s Q_H(s_t, a^H_t)$ and a quadratic approximation of the KL - divergence constraint. Solving this approximation of the Lagrangian leads to an update rule which introduce natural gradient adaptation.
Natural Gradient Shared Control { # natural - gradient - shared - control }
-------------------------------
We introduce a new dominance paradigm for shared control using lifelike gradient adaptation. Note that our finish is to find the action that maximizes $ Q_H$, resulting in taking gradient steps in the focus of ascent. $ $ \label{eqn: state_update }
\begin{aligned }
s_{t+1 } & = s_t + \eta F(s_t)^{-1 } \nabla_s Q_H(s_t, a^H_t) \\
& = s | ienh Shared Control {#sec:NGSC}
===============================
Txpressing Shared Contron as ah Optimixation Problem
----------------------------------------------------
Let $s \in \mathcel{S}$ ve tht state of the systdm. Let $a^H \in \mathxal{A^I}$ as the user acvjon, $a^R \lu \matggal{A^R}$ ue the autonomoos robot acthon, and $u \in \mdtfccl{U}$ be the shared control action. The human snf the robot aggnt essh sslect actions following their stocgastic kolicies, $\pi_H$, and $\po_R$. Our goal is to find a sjaref control policy $\pl_S$ that solces eye following optimizatpmn problem.
[o\_t]{}[Q\_H (s\_t, u\_t)]{} \[opt:problem\]
The shared zontrpl policy us chlven to maxinize nhe user’s intcgnal acdion-valie function $Q_H(x_t, n_t)$ ar each step. Although mt is possible to lewrn $Q_H$ ushny methods such as MaxWnr IOC [@eiebast2009plxbnivg], krevicfing tje nser controms $a^H_t$ can ve challenging due uo ybterpersonal sifferqnses. Instead, we regard the user action av ah estimate of $\nabla_s Q_H(w_t, a^H_t)$ at each step.
Thg constraigt on the KL-divergence between the robot policy atd thx rhaxcq popicy ensures that the shared policy does not qsvoane far from the amtonomous robot pokify. Jhe problem cav be erlrsssed using a Lagrwnge Mujtipluer, assumyng s linear approximation of oyr objective $\babla_s Q_H(s_t, a^H_t)$ anb a quadratie apprpximayion of the KL-divergencz consfraint. Solvlng this zoproximation of ghe Ldgrangian leads to an updaee rule wiich nntroducds nstural gradient wdaptation.
Natural Gradiejt Shcred Wontrol {#nahural-gradient-shared-control}
-------------------------------
We invcoduce a new vottrml paradngm fov shared controj using naturak gradiznt adxptation. Nkte thav our goal if to find the wction that kaximizef $Q_H$, resylting kv taking gradirnt steps in the dieection of ascent. $$\kabgl{sqn:state_update}
\vegin{aligned}
s_{t+1} &= s_e + \eva F(s_e)^{-1} \nabla_s Q_H(s_t, a^H_t) \\
&=s | ient Shared Control {#sec:NGSC} =============================== Expressing Shared an Problem ---------------------------------------------------- $s \in \mathcal{S}$ system. $a^H \in \mathcal{A^H}$ the user action, \in \mathcal{A^R}$ be the autonomous robot and $u \in \mathcal{U}$ be the shared control action. The human and the agent each select actions following their stochastic policies, $\pi_H$, and $\pi_R$. Our goal to a control $\pi_S$ that solves the following optimization problem. [u\_t]{}[Q\_H (s\_t, u\_t)]{} \[opt:problem\] The shared control policy is to maximize the user’s internal action-value function $Q_H(s_t, at each step. Although is possible to learn $Q_H$ methods as MaxEnt [@ziebart2009planning], the controls $a^H_t$ can challenging due to interpersonal differences. Instead, we regard the user action as an estimate of $\nabla_s Q_H(s_t, at each constraint on KL-divergence the policy and the ensures that the shared policy does from the autonomous robot policy. The problem can expressed using Lagrange Multiplier, assuming a linear approximation our objective $\nabla_s Q_H(s_t, a^H_t)$ and a quadratic of the KL-divergence constraint. Solving this approximation of the Lagrangian leads to an update rule natural gradient adaptation. Natural Shared Control {#natural-gradient-shared-control} We a control for shared using natural gradient adaptation. Note that our goal is to find action that maximizes $Q_H$, resulting in taking gradient steps in of $$\label{eqn:state_update} \begin{aligned} s_{t+1} s_t + \eta F(s_t)^{-1} Q_H(s_t, \\ &=s | ient Shared Control {#sec:NGSC}
===============================
EXpressing SHared conTroL aS an OPtimIzation Problem
----------------------------------------------------
lEt $s \iN \mathcal{S}$ be the state of tHe sysTeM. let $a^h \In \MathcAl{A^H}$ as tHE uSER acTiOn, $A^R \iN \mAThCal{A^R}$ Be tHe autonOmous robot ActIoN, and $u \in \mathcAL{U}$ Be the shareD coNtrol action. THe hUman anD tHe rOBot agEnt Each sElect aCTions fOllowing tHeIR stochAStic polICIeS, $\pi_H$, And $\pi_R$. Our goal is to FInD A shared control Policy $\Pi_s$ ThAT SolVes The followiNg OptimIZation pRObLEM.
[U\_t]{}[Q\_h (S\_t, u\_t)]{} \[opt:probleM\]
The shared cONtrOl poliCy Is cHOsen to MaximIzE The User’s internAl acTion-value FunctiON $Q_H(s_t, u_t)$ AT each stEp. AlthOugH it Is poSSiBlE to LeARn $Q_h$ UsIng MEthOds such aS MAxent IOc [@zieBART2009PlanNinG], preDictiNg the user contRolS $a^H_t$ CAn bE chalLengiNg duE tO inteRpersoNal diFfErences. Instead, wE regArd the useR acTiOn aS aN estiMAte of $\nAblA_s Q_h(s_t, a^H_t)$ aT each stEP.
ThE cONSTrAint on the KL-divergeNcE BEtWeen the rObot poLIcY aND the sharEd PolIcy eNSUres tHat tHE sHared polIcy doeS NoT dEviate fAr From thE aUtoNomOus roBOt poLicy. ThE problem Can be EXpressed using a lAgrange MultipLIeR, ASsUMing A liNear approxiMatiON of oUr obJEcTivE $\Nabla_S Q_H(s_t, A^H_T)$ AnD A quadratic approximaTiOn of thE KL-diVergence constRaint. SolviNG THis approXimaTIoN Of the LagrangiaN leadS to an updatE Rule whicH intrOduces naTural gradIENt adaptaTioN.
NaTurAl GRADiEnt Shared ContROL {#natUrAl-gradiEnt-Shared-cOntRol}
-------------------------------
we iNtrOdUce a new coNtrol parAdIgM fOr ShaRed coNTrol usinG nAtuRaL grAdienT AdaptaTion. NOte tHaT oUR goAl is to fINd THE actIoN tHat mAxiMiZes $Q_H$, ResuLTinG in takiNg gradienT stEPs in ThE dIrectioN of ascent. $$\labeL{eQn:state_updAtE}
\beGin{aliGNEd}
s_{t+1} &= s_t + \etA F(s_t)^{-1} \nabla_s Q_H(s_t, a^H_t) \\
&=s | ient Shared Control {#sec: NGSC}
==== ===== === === == ==== ==== ======
Expres s ingShared Control as an O ptimi za t ionP ro blem------- - -- - - --- -- -- --- -- - -- ----- --- ------- ----------
L et $s \in \mat h ca l{S}$ be t hestate of the sy stem.Le t $ a ^H \i n \ mathc al{A^H } $ as t he user a ct i on, $a ^ R \in \ m a th cal{ A^R}$ be the auto n om o us robot actio n, and $ u \ i n \m ath cal{U}$ be t he sh a red con t ro l a cti o n. The humanand the rob o t a gent e ac h s e lect a ction sf oll owing their sto chastic p olicie s , $\pi_ H $, and$\pi_R $.Our goa l i stofi n d a sh are d co ntrol po li cy $\pi _S$t h a t sol ves the foll owing optimiz ati on p r obl em.
[u\_t ]{}[ Q\ _H (s \_t, u \_t)] {} \[opt:problem\ ]
T he shared co nt rol p olicy is cho sen to maximi ze theu ser ’s i n te rnal action-valuefu n c ti on $Q_H( s_t, u _ t) $a t each s te p.Alth o u gh it isp os sible to learn $Q _H $ using m ethods s uch as MaxE n t IO C [@zi ebart200 9plan n ing], predicti n g the user co n tr o l s$ a^H_ t$can be chal leng i ng d ue t o i nte r perso nal d if f er e nces. Instead, we r eg ard th e use r action as a n estimate o f $\nabla _s Q _ H( s _t, a^H_t)$ at each step.
Th e constra int o n the KL -divergen c e between th e r obo t p o l ic y and the sha r e d po li cy ensu res that t hesha red po li cy does n ot devia te f ar f rom thea utonomou srob ot po licy. The pr oblem can b ee xpr essed u s in g a La gr an ge M ult ip lier, ass u min g a lin ear appro xim a tion o four obj ective $\nabl a_ s Q_H(s_t, a ^H_ t)$ an d a quadra tic approximation of th e KL-div erg encecons traint. S olv ing th isa pproxi mation of t he La g r angia n le ads t o an updat e rul e whi ch int roduces natural gradienta dap tation.
Natu ral Gra d i en t S h ar e d C on t rol { #natural-gradie nt-shared- co n tr ol}
------ - --- -- ------- ------- -----
We int roduce anew contr ol par a d igm for share d contro l using n a tural gr adien t a daptat io n.Notethat o u r g oal i s to f in d theactio nthat max imizes $Q_H$, resulting in ta kinggra dient ste psi n t he direct ionof ascent. $$ \la bel{e qn: s tate_ upda t e}
\beg in{a l igned}
s _{t + 1 }&= s_t + \e t a F(s _t)^{ -1} \nabla _s Q _H(s_t, a^H_t) \ \
&=s | ient Shared_Control {#sec:NGSC}
===============================
Expressing_Shared Control as an_Optimization Problem
----------------------------------------------------
Let_$s_\in \mathcal{S}$_be_the state of_the system. Let_$a^H \in \mathcal{A^H}$ as_the user action,_$a^R_\in \mathcal{A^R}$ be the autonomous robot action, and $u \in \mathcal{U}$ be the shared_control_action. The_human_and_the robot agent each select_actions following their stochastic policies,_$\pi_H$, and_$\pi_R$. Our goal is to find a shared_control_policy $\pi_S$ that_solves the following optimization problem.
[u\_t]{}[Q\_H (s\_t, u\_t)]{} \[opt:problem\]
The shared_control policy is chosen to maximize_the user’s internal_action-value_function_$Q_H(s_t, u_t)$ at each_step. Although it is possible to_learn $Q_H$ using methods such as_MaxEnt IOC [@ziebart2009planning], predicting the user controls $a^H_t$_can be challenging due to interpersonal_differences. Instead, we regard the_user action_as an estimate of $\nabla_s_Q_H(s_t, a^H_t)$ at_each step.
The_constraint on the_KL-divergence between the robot policy and_the shared policy_ensures that the shared policy does_not_deviate far from_the_autonomous_robot policy._The problem can_be_expressed using_a_Lagrange Multiplier, assuming a linear approximation_of_our objective $\nabla_s Q_H(s_t, a^H_t)$ and a_quadratic approximation of the_KL-divergence_constraint. Solving this approximation_of the Lagrangian leads to_an update rule which introduces natural_gradient adaptation.
Natural_Gradient Shared_Control {#natural-gradient-shared-control}
-------------------------------
We introduce a new control paradigm for shared control using_natural gradient adaptation. Note that our_goal is to find_the action_that_maximizes $Q_H$, resulting_in_taking gradient_steps in the direction of ascent. $$\label{eqn:state_update}
_ _ \begin{aligned}
s_{t+1}_&= s_t + \eta_F(s_t)^{-1}_\nabla_s Q_H(s_t, a^H_t) \\
_ &=s |
TP-AGB model may be changed in future calculations.
Anyhow, various tests indicate that the present version of the `COLIBRI` models already yields a fairly good description of the TP-AGB phase. Compared to our previously calibrated sets [@MarigoGirardi_07; @Marigo_etal08; @Girardi_etal10] the new TP-AGB models yield somewhat shorter, but still comparable, TP-AGB lifetimes, and they successfully recover various observational constraints dealing with e.g. the Galactic initial–final mass relation (Kalirai et al., in prep.), spectro-interferometric determinations of AGB stellar parameters [@Klotz_etal13], the correlation between mass-loss rates and pulsation periods, and the trends of the effective temperature with the C/O ratio observed in Galactic M, S and C stars.
Further important support comes from the results of our new model for the condensation and growth of dust grains in the outflows of AGB stars [@Nanni_etal13], which has been applied to the `COLIBRI` TP-AGB tracks. The results are extremely encouraging as they are found to nicely reproduce other independent sets of key observations, i.e. the correlation between expansion velocities and mass-loss rates/pulsation periods of Galactic AGB stars.
Acknowledgments {#acknowledgments.unnumbered}
===============
It is a pleasure to thank Julianne Dalcanton and Luciana Bianchi for their strong encouragement to this work, Phil Rosenfield and Marco Gullieuszik for their contribution to test the preliminary versions of the new TP-AGB tracks. Warm thanks go to Anita and Alessio for having inspired the name of the code. We acknowledge financial support from contract ASI-INAF n. I/009/10/0, and from [*Progetto di Ateneo 2012*]{}, University of Padova, ID: CPDA125588/12.
[99]{} Angulo C., 1999, AIPC, 495, 365 Arnett W. D., Truran J. W., 1969, ApJ, 157, 339 Badnell N. R., Bautista M. A., Butler K., Delahaye F., Mendoza C., Palmeri P., Zeippen C. J., Seaton M. J., 2005, MNRAS, 360, 458, P. J. | TP - AGB model may be changed in future calculations.
Anyhow, various examination bespeak that the present version of the 'COLIBRI 'models already yields a reasonably good description of the TP - AGB phase. Compared to our previously calibrated set [ @MarigoGirardi_07; @Marigo_etal08; @Girardi_etal10 ] the new TP - AGB models give way somewhat shorter, but even comparable, TP - AGB lifetimes, and they successfully recover respective experimental constraints dealing with for example the Galactic initial – concluding batch relation (Kalirai et al. , in prep .), spectro - interferometric determinations of AGB leading parameters [ @Klotz_etal13 ], the correlation between aggregate - loss rate and pulsation period, and the trends of the effective temperature with the C / O proportion observed in Galactic M, S and C stars.
Further important support hail from the results of our new model for the condensation and growth of dust grains in the outflows of AGB star [ @Nanni_etal13 ], which has been applied to the 'COLIBRI 'TP - AGB tracks. The resultant role are highly encouraging as they are find to nicely reproduce other autonomous sets of cardinal observations, i.e. the correlation between expansion velocities and aggregate - loss rates / pulsation periods of Galactic AGB stars.
Acknowledgments { # acknowledgments.unnumbered }
= = = = = = = = = = = = = = =
It is a joy to thank Julianne Dalcanton and Luciana Bianchi for their strong encouragement to this work, Phil Rosenfield and Marco Gullieuszik for their contribution to test the preliminary versions of the new TP - AGB tracks. affectionate thanks go to Anita and Alessio for having inspired the name of the code. We acknowledge financial support from contract ASI - INAF n. I/009/10/0, and from [ * Progetto di Ateneo 2012 * ] { }, University of Padova, ID: CPDA125588/12.
[ 99 ] { } Angulo C., 1999, AIPC, 495, 365 Arnett W. D., Truran J. W., 1969, ApJ, 157, 339 Badnell N. R., Bautista M. A., Butler K., Delahaye F., Mendoza C., Palmeri P., Zeippen C. J., Seaton M. J., 2005, MNRAS, 360, 458, P. J. | TP-WGB model may be changed in future calcolqtions.
Enyhow, barious gests indicate that the presxnt cersiin of the `COLIBRI` modeus alreadj yields q famrly good descri'fion of the TL-WGB 'hese. Compared to our previmusly calibratad szts [@MarigoGirardi_07; @Marigo_etal08; @Girardi_qtal10] thr jew TP-AGB modejs ypejd skmewhat shorter, but still comparabme, TP-AGU lifetimes, and they successfully recover varlous observational constraintw dewoing with e.g. the Galacuie initial–fihal mass relation (Kalirai et al., in pxep.), spectro-unrervgrometric deverminwtions of AGN stelldr paraketers [@Klotz_etsl13], vhe xorrelation between mess-loss rates and pujsation parnods, and the trends od rhe exfecdive remoerztnre with hhe C/O ratio kbserved in Galactic M, S and C seqrs.
Further imlortane fupport comes from the results of our ntw mosel for the condensatiob and growth of dust hrains in the outflows of AGB stars [@Nanni_etal13], which has bean ap'lked tj rhf `COLIBRI` TP-AGB tracks. The results are extremqmy ekcouraging as thcy are found to nivepy teproduce othet indepzhdsnt sets of key obdervatijns, i.w. the corwelayion between expansion veloxities and mcss-ooss rates/pulsatiou periods of Galsctic AGB stars.
Acknowledgmencs {#ackhowledgmentd.unnumbersa}
===============
It is a pleasurd tp dhank Julianne Dalcanton agd Luciane Biauchi for theor strjng encourwgemekd to this work, Phip Rosgnfielg and Marcl Gullieuszik for their contribnvion to test jhe prvliminary verslons of the new TP-AGB tracks. Carm thauks go to Anita znd Alevsio for hading inspired jhe name of tie code. Wq acjnowoedge fkvancial suppory from couuract ASI-IBAF n. I/009/10/0, and from [*Prpgejtk di Ateneo 2012*]{}, Unnretwity of Padova, ID: CPQA125588/12.
[99]{} Aigulo W., 1999, AIPC, 495, 365 Artett W. D., Yrurav J. W., 1969, AkL, 157, 339 Cadnrll N. R., Bautista M. A., Bgtled K., Delahaye F., Mencoda C., Palmgri P., Zei[pen C. J., Seatom M. J., 2005, MNRAS, 360, 458, P. J. | TP-AGB model may be changed in future various indicate that present version of a good description of TP-AGB phase. Compared our previously calibrated sets [@MarigoGirardi_07; @Marigo_etal08; the new TP-AGB models yield somewhat shorter, but still comparable, TP-AGB lifetimes, and successfully recover various observational constraints dealing with e.g. the Galactic initial–final mass relation et in spectro-interferometric of AGB stellar parameters [@Klotz_etal13], the correlation between mass-loss rates and pulsation periods, and the trends the effective temperature with the C/O ratio observed Galactic M, S and stars. Further important support comes the of our model the and growth of grains in the outflows of AGB stars [@Nanni_etal13], which has been applied to the `COLIBRI` TP-AGB tracks. results are as they found nicely other independent sets observations, i.e. the correlation between expansion rates/pulsation periods of Galactic AGB stars. Acknowledgments {#acknowledgments.unnumbered} It is pleasure to thank Julianne Dalcanton and Bianchi for their strong encouragement to this work, Rosenfield and Marco Gullieuszik for their contribution to test the preliminary versions of the new Warm thanks go to and Alessio for inspired name the We acknowledge support from contract ASI-INAF n. I/009/10/0, and from [*Progetto di Ateneo University of Padova, ID: CPDA125588/12. [99]{} Angulo C., 1999, AIPC, Arnett D., Truran J. 1969, ApJ, 157, 339 N. Bautista M. A., Butler F., C., C. Seaton J., 2005, MNRAS, 360, P. J. | TP-AGB model may be changed in fUture calcuLatioNs.
ANyhOw, VariOus tEsts indicate thAT the Present version of the `COLiBRI` mOdELs alREaDy yieLds a faiRLy GOOd dEsCrIptIoN Of The TP-aGB Phase. CoMpared to ouR prEvIously calibrATeD sets [@MarigOGiRardi_07; @Marigo_eTal08; @girardI_eTal10] THe new tP-AgB modEls yieLD somewHat shorteR, bUT still COmparabLE, tP-aGB lIfetimes, and they suCCeSSfully recover vArious ObSErVATioNal ConstraintS dEalinG With e.g. tHE GALACtiC Initial–final mAss relation (kAliRai et aL., iN prEP.), spectRo-intErFEroMetric deterMinaTions of AGb stellAR parameTErs [@KlotZ_etal13], tHe cOrrElatIOn BeTweEn MAss-LOsS raTEs aNd pulsatIoN pEriodS, and THE TRendS of The eFfectIve temperaturE wiTh thE c/O rAtio oBservEd in gaLactiC M, S and c starS.
FUrther important SuppOrt comes fRom ThE reSuLts of OUr new mOdeL foR the conDensatiON anD gROWTh Of dust grains in the oUtFLOwS of AGB stArs [@NanNI_eTaL13], Which has BeEn aPpliED To the `cOLIbrI` tP-AGB traCks. The REsUlTs are exTrEmely eNcOurAgiNg as tHEy arE found To nicely ReproDUce other indepeNDent sets of key OBsERVaTIons, I.e. tHe correlatiOn beTWeen ExpaNSiOn vELocitIes anD mASs-LOss rates/pulsation peRiOds of GAlactIc AGB stars.
AckNowledgmenTS {#ACknowledGmenTS.uNNumbered}
===============
It is a pLeasuRe to thank JULianne DaLcantOn and LucIana BiancHI For their StrOng EncOurAGEmEnt to this work, pHIl RoSeNfield aNd MArco GulLieUszIk fOr tHeIr contribUtion to tEsT tHe PrEliMinarY Versions Of The NeW TP-aGB trACks. WarM thanKs go To anITa aNd AlessIO fOR HaviNg InSpirEd tHe Name oF the COde. we acknoWledge finAncIAl suPpOrT from coNtract ASI-INAF N. I/009/10/0, And from [*ProGeTto Di AtenEO 2012*]{}, universiTy of Padova, ID: CPDA125588/12.
[99]{} Angulo C., 1999, aiPC, 495, 365 ArneTt W. d., TrurAn J. W., 1969, apJ, 157, 339 BadnelL N. R., bautisTa M. a., butler k., DelahAye F., MEnDozA c., palmeRI p., ZEipPeN C. J., Seaton M. j., 2005, mnRAs, 360, 458, P. J. | TP-AGB model may be chang ed in futu re ca lcu lat io ns.
Any how, various t e stsindicate that the pres ent v er s iono fthe ` COLIBRI ` m o d els a lr ead yy ie lds a fa irly go od descrip tio nof the TP-AG B p hase. Comp are d to our pre vio usly c al ibr a ted s ets [@Ma rigoGi r ardi_0 7; @Marig o_ e tal08; @Girard i _ et al10 ] the new TP-AGBm od e ls yield somew hat sh or t er , but st ill compar ab le, T P -AGB li f et i m e s,a nd they succe ssfully rec o ver vario us ob s ervati onalco n str aints deali ng w ith e.g.the Ga l actic i n itial–f inal m ass re lati o n(K ali ra i et al .,i n p rep.), s pe ct ro-in terf e r o m etri c d eter minat ions of AGB s tel larp ara meter s [@K lotz _e tal13 ], the corr el ation between m ass- loss rate s a nd pu ls ation period s,and the tr ends of the e f f e ct ive temperature wi th t he C/O rat io obs e rv ed in Galac ti c M , Sa n d C s tars .
Furtherimport a nt s upportco mes fr om th e r esult s ofour ne w modelfor t h e condensation and growth of du s t g r ains in the outflo ws o f AGB sta r s[@N a nni_e tal13 ], wh i ch has been applied t o the`COLI BRI` TP-AGB t racks. The r e sults ar e ex t re m ely encouragin g asthey are f o und to n icely reprodu ce otheri n dependen t s ets of ke y ob servations, i . e . th ecorrela tio n betwe enexp ans ion v elocities and mas s- lo ss r ate s/pul s ation pe ri ods o f G alact i c AGBstars .
A ck no w led gments{ #a c k nowl ed gm ents .un nu mbere d}
= = === ======= ===
It i s a plea su re to tha nk Julianne D al canton and L uci ana Bi a n chi fortheir strong encouragem e nt to t his work , Ph il Rosenf iel d andMar c o Gull ieuszi k for t hei r contr i b ut ion t o test the p rel imina ry ver sions o f the new TP-AGB t r ack s. Warm thank s g o to A ni taa nd Ale ss i o f o r having inspire d the name o f t he code. W e ac kn owledge financ ial s u pport f rom contr act ASI-I NA F n. I /00 9/10/0, an d from [ *Progetto di At e ne o 201 2*] {}, Un iv ers ity o f Pado v a,ID: C PDA125 58 8/12.
[99] {} AnguloC., 1999, AIPC, 495, 36 5 Arne tt W. D. , TruranJ.W .,1969, ApJ , 15 7, 339 Bad nel l N . R., Ba u tista M.A ., Bu t ler K ., D e lahaye F. , M end o z aC., Palmeri P . , Z eippe n C . J., S eato n M. J., 2005, MN R AS, 360, 458,P. J . | TP-AGB_model may_be changed in future_calculations.
Anyhow, various_tests_indicate that_the_present version of_the `COLIBRI` models_already yields a fairly_good description of_the_TP-AGB phase. Compared to our previously calibrated sets [@MarigoGirardi_07; @Marigo_etal08; @Girardi_etal10] the new TP-AGB_models_yield somewhat_shorter,_but_still comparable, TP-AGB lifetimes, and_they successfully recover various observational_constraints dealing_with e.g. the Galactic initial–final mass relation (Kalirai_et_al., in prep.),_spectro-interferometric determinations of AGB stellar parameters [@Klotz_etal13], the correlation_between mass-loss rates and pulsation periods,_and the trends_of_the_effective temperature with the_C/O ratio observed in Galactic M,_S and C stars.
Further important support_comes from the results of our new_model for the condensation and growth_of dust grains in the_outflows of_AGB stars [@Nanni_etal13], which has_been applied to_the `COLIBRI`_TP-AGB tracks. The_results are extremely encouraging as they_are found to_nicely reproduce other independent sets of_key_observations, i.e. the_correlation_between_expansion velocities_and mass-loss rates/pulsation_periods_of Galactic_AGB_stars.
Acknowledgments {#acknowledgments.unnumbered}
===============
It is a pleasure to_thank_Julianne Dalcanton and Luciana Bianchi for their_strong encouragement to this_work,_Phil Rosenfield and Marco_Gullieuszik for their contribution to_test the preliminary versions of the_new TP-AGB_tracks. Warm_thanks go to Anita and Alessio for having inspired the name_of the code. We acknowledge financial_support from contract ASI-INAF_n. I/009/10/0, and_from_[*Progetto di Ateneo_2012*]{},_University of_Padova, ID: CPDA125588/12.
[99]{} Angulo C., 1999, AIPC,_495, 365_Arnett W. D., Truran J. W., 1969, ApJ,_157, 339 Badnell N. R.,_Bautista_M. A., Butler K., Delahaye F., Mendoza_C., Palmeri P., Zeippen C. J., Seaton_M. J., 2005, MNRAS, 360, 458,_P. J. |
at least of order $\simg
10^{15}$ G. Indeed, it is not obvious why the fields cannot become even higher. Note that the virial limit is $B_v \sim 10^{17}$ G.
After magnetic fields have built up to some fraction of the equipartition value with the shear motion, a magnetic viscosity develops. Assuming that $B_rB_\phi\sim B^2$, it can be characterized in the usual way by the parameter $\alpha\sim B^2/(4\pi \rho v_s^2 ) \sim 10^{-1} B_{15}^2
\rho_{13}^{-1} T_9^{-1}$. This viscosity continues operating also after cooling has led to the disappearance of neutrino viscosity. Assuming a value of $\alpha=0.1$, a BH mass 3 $\msun$ and outer disk radius equal to the Roche lobe size, Popham et al. (1998) estimate “viscous" life times of 0.1 s for NS/BH-NS, 10–20 s for a collapsar (failed SN Ib or rotating WR), and 15–150 s for WD-BH and He-BH systems (although fields of $10^{15}$ G may be more difficult to support in He-BH systems).
A magnetic field configuration capable of powering the bursts is likely to have a large scale structure. Flares and instabilities occurring on the characteristic (millisecond) dynamical time scale would cause substantial irregularity or intermittency in the overall outflow that would manifest itself in internal shocks (Rees &, 1994) There is thus no problem in principle in accounting for sporadic large-amplitude variability, on all time scales down to a millisecond, even in the most long-lived bursts. Note also that it only takes a residual cold disk of $10^{-3}\msun$ to confine a field of $10^{15}$ G, which can extract energy from the black hole via the Blandford-Znajek mechanism. Even if the evolution time scale for the bulk of the debris were no more than a second, enough may remain to catalyse the extraction of enough energy from the hole to power a long-lived burst.
Double peaked bursts
--------------------
There are at least two mechanisms which might lead to a delayed “second" burst (or a double humped burst). One possibility is that a merger leads to a central NS, temporarily stabilized | at least of order $ \simg
10^{15}$ G. Indeed, it is not obvious why the fields cannot become even higher. Note that the virial terminus ad quem is $ B_v \sim 10^{17}$ G.
After charismatic discipline have build up to some fraction of the equipartition value with the shear movement, a magnetic viscosity develops. assume that $ B_rB_\phi\sim B^2 $, it can be characterized in the usual manner by the parameter $ \alpha\sim B^2/(4\pi \rho v_s^2) \sim 10^{-1 } B_{15}^2
\rho_{13}^{-1 } T_9^{-1}$. This viscosity continues manoeuver also after cooling has moderate to the fade of neutrino viscosity. Assuming a value of $ \alpha=0.1 $, a BH bulk 3 $ \msun$ and outer disk spoke equal to the Roche lobe size, Popham et al. (1998) estimate “ viscous " liveliness times of 0.1 s for NS / BH - NS, 10–20 randomness for a collapsar (failed SN Ib or revolve WR), and 15–150 s for WD - BH and He - BH systems (although fields of $ 10^{15}$ G may be more difficult to support in He - BH system).
A magnetic field configuration capable of powering the burst is likely to have a large scale structure. Flares and instabilities occurring on the characteristic (millisecond) dynamical time plate would induce substantial irregularity or intermittency in the overall escape that would manifest itself in internal shocks (Rees &, 1994) There constitute thus no problem in principle in accounting for sporadic big - amplitude variability, on all time scales down to a millisecond, even in the most long - lived bursts. notice also that it only takes a residual cold disk of $ 10^{-3}\msun$ to confine a field of $ 10^{15}$ G, which can educe energy from the black hole via the Blandford - Znajek mechanism. Even if the evolution time plate for the bulk of the debris were no more than a second, enough may remain to catalyse the extraction of enough energy from the hole to power a long - lived fusillade.
Double peaked explosion
--------------------
There are at least two mechanism which might lead to a delay “ second " burst (or a bivalent humped burst). One possibility is that a fusion leads to a cardinal NS, temporarily stabilized | at least of order $\simg
10^{15}$ G. Ikdeed, it is not obvious why tge fieldr cannot become even higher. Iote that the virial limit is $B_x \sim 10^{17}$ G.
Avter magbetir fields have bummt up to some nractnoi of the equipattition valua with the shedr mltion, a magnetic viscosity develops. Assumimg that $B_rB_\phi\sii B^2$, pt can be characterized in the usual way by the parameter $\allha\sim B^2/(4\pi \rho v_s^2 ) \sim 10^{-1} B_{15}^2
\rjo_{13}^{-1} T_9^{-1}$. This viscosity cojtinues opetztigt also after cooling has led to thg disappearance of neutrino viscusity. Assuming q calkg of $\alpha=0.1$, a BH mwss 3 $\msun$ and outer gisk racius equal to bhe Rmchw lobe size, Popham et el. (1998) estimate “viscous" jife timev kf 0.1 s for NS/BH-NS, 10–20 s for d conlaprqr (waimev SH Ib og rktating WR), and 15–150 s for WD-BH and He-BH systtms (qlthough fielss of $10^{15}$ G may be more difficult to support in He-TH aystems).
A magnetic field configuration capablg of poweryng the bursts is likely to have a large scale stsuctuce. Flcvcs ave lnstabilities occurring on the characteristic (jiklpsecond) dynamical time scale wlukq cause substxntial irdegularity or integmittensy in the overwll putflow that would manifest itself in iuteenal shocks (Rees &, 1994) There is tkus no probkem in principle in accuuntjng for spogadic larfd-amplitude variacilpty, mn all time scales down to a millisxcond, even iv thg most jong-lived hursts. Note also that it only tdkes a resldual cold disk of $10^{-3}\msun$ to confmie a field of $10^{15}$ C, wvich can extrsct energy frjm the black hple via the Clandford-Zhajek mxchanism. Eveg if the evolgjion time scane for tre bylk if the adbris were no kore than a second, wnough may remain bo cajamyse the extraccnob of enough enetgy frjm tie hoja to power a lone-lixrd bufst.
Double pcakdd birsts
--------------------
There are at ledst fwo mechanisms whivh might lgad to a qelayed “seconc" burst (or a doublt humpxd burrt). Onr pjssibility is that a merger lezds to a fenbral NS, tempowarioy stabilizeb | at least of order $\simg 10^{15}$ G. is obvious why fields cannot become virial is $B_v \sim G. After magnetic have built up to some fraction the equipartition value with the shear motion, a magnetic viscosity develops. Assuming that B^2$, it can be characterized in the usual way by the parameter $\alpha\sim \rho ) 10^{-1} \rho_{13}^{-1} T_9^{-1}$. This viscosity continues operating also after cooling has led to the disappearance of neutrino Assuming a value of $\alpha=0.1$, a BH mass $\msun$ and outer disk equal to the Roche lobe Popham al. (1998) “viscous" times 0.1 s for 10–20 s for a collapsar (failed SN Ib or rotating WR), and 15–150 s for WD-BH and systems (although $10^{15}$ G be difficult support in He-BH magnetic field configuration capable of powering likely to have a large scale structure. Flares instabilities occurring the characteristic (millisecond) dynamical time scale cause substantial irregularity or intermittency in the overall that would manifest itself in internal shocks (Rees &, 1994) There is thus no problem in accounting for sporadic variability, on all scales to millisecond, in the long-lived bursts. Note also that it only takes a residual cold of $10^{-3}\msun$ to confine a field of $10^{15}$ G, which energy the black hole the Blandford-Znajek mechanism. Even the time scale for the the were a enough remain to catalyse the of enough energy from the to power a long-lived There are at least two mechanisms which might to a delayed “second" burst (or a humped burst). One possibility is that a merger leads to a central temporarily stabilized | at least of order $\simg
10^{15}$ G. Indeed, It is not obvIous wHy tHe fIeLds cAnnoT become even higHEr. NoTe that the virial limit is $b_v \sim 10^{17}$ g.
AFTer mAGnEtic fIelds haVE bUILt uP tO sOme FrACtIon of The EquiparTition valuE wiTh The shear motiON, a Magnetic viScoSity develops. assUming tHaT $B_rb_\Phi\siM B^2$, iT can bE charaCTerizeD in the usuAl WAy by thE ParametER $\AlPha\sIm B^2/(4\pi \rho v_s^2 ) \sim 10^{-1} B_{15}^2
\rho_{13}^{-1} t_9^{-1}$. thIS viscosity contInues oPeRAtING alSo aFter coolinG hAs led TO the disAPpEARAncE Of neutrino visCosity. AssumINg a Value oF $\aLphA=0.1$, A BH masS 3 $\msun$ AnD OutEr disk radiuS equAl to the RoChe lobE Size, PopHAm et al. (1998) eStimatE “viScoUs" liFE tImEs oF 0.1 s FOr Ns/bH-nS, 10–20 s FOr a CollapsaR (fAiLed SN ib or ROTATing wR), aNd 15–150 s fOr WD-Bh and He-BH systeMs (aLthoUGh fIelds Of $10^{15}$ G maY be mOrE diffIcult tO suppOrT in He-BH systems).
A MagnEtic field ConFiGurAtIon caPAble of PowEriNg the buRsts is lIKelY tO HAVe A large scale structuRe. fLArEs and insTabiliTIeS oCCurring oN tHe cHaraCTEristIc (miLLiSecond) dyNamicaL TiMe Scale woUlD cause SuBstAntIal irREgulArity oR intermiTtencY In the overall ouTFlow that would MAnIFEsT ItseLf iN internal shOcks (rEes &, 1994) THere IS tHus NO probLem in PrINcIPle in accounting for sPoRadic lArge-aMplitude variaBility, on alL TIMe scales Down TO a MIllisecond, even In the Most long-liVEd bursts. note aLso that iT only takeS A Residual ColD diSk oF $10^{-3}\msUN$ To Confine a field OF $10^{15}$ g, whiCh Can extrAct Energy fRom The BlaCk hOlE via the BlAndford-ZNaJeK mEcHanIsm. EvEN if the evOlUtiOn TimE scalE For the Bulk oF the DeBrIS weRe no morE ThAN A secOnD, eNougH maY rEmain To caTAlySe the exTraction oF enOUgh eNeRgY from thE hole to power a LoNg-lived burSt.
douBle peaKED bursts
--------------------
THere are at least two mechanISms whicH miGht leAd to A delayed “sEcoNd" bursT (or A Double Humped Burst). onE poSSIbiliTY Is ThaT a Merger leadS TO a cEntraL Ns, temPorarilY stabilized | at least of order $\simg10^{15}$ G . Ind eed , i tis n ot o bvious why the fiel ds cannot become evenhighe r. Note th at th e viria l l i m itis $ B_v \ s im 10^{ 17} $ G.
A fter magne tic f ields have b u il t up to so mefraction ofthe equip ar tit i on va lue with the s h ear mo tion, a m ag n etic v i scosity d ev elop s. Assuming that$ B_ r B_\phi\sim B^2 $, itca n b e cha rac terized in t he us u al wayb yt h e pa r ameter $\alph a\sim B^2/( 4 \pi \rhov_ s^2 ) \sim 10^{ -1 } B_ {15}^2
\rho _{13 }^{-1} T_ 9^{-1} $ . Thisv iscosit y cont inu esoper a ti ng al so aft e rcoo l ing has led t othe d isap p e a r ance of neu trino viscosity. A ssu ming a v alueof $\ alph a= 0.1$, a BHmass3$\msun$ and out er d isk radiu s e qu alto theR oche l obe si ze, Pop ham eta l.(1 9 9 8 )estimate “viscous" l i f etimes of 0.1 s fo rN S/BH-NS, 1 0–2 0 sf o r a c olla p sa r (faile d SN I b o rrotatin gWR), a nd 15 –15 0 s f o r WD -BH an d He-BHsyste m s (although fi e lds of $10^{1 5 }$ G m a y be mo re difficul t to supp orti nHe- B H sys tems) .Am agnetic field confi gu ration capa ble of poweri ng the bur s t s is like ly t o h a ve a large sca le st ructure. F l ares and inst abilitie s occurri n g on thecha rac ter ist i c ( millisecond)d y nami ca l timesca le woul d c aus e s ubs ta ntial irr egularit yor i nt erm itten c y in the o ver al l o utflo w thatwould man if es t it self in in t e rnal s ho cks(Re es &, 1 994) The re is t hus no pr obl e m in p ri nciplein accounting f or sporadi clar ge-amp l i tude var iability, on all time s c ales do wnto amill isecond,eve n in t hem ost lo ng-liv ed bu rs ts. N ote a l s otha tit only ta k e s a resi du al c old dis k of $10^{-3}\msun $ to confine a fi eld of$ 1 0^ {15 } $G , w hi c h c a n extract energy from thebl a ck hole viat heBl andford -Znajek mech a nism. E ven if th e evoluti on tim e sca le for the bulk of the debr i s wer e n o mor e t han ase con d, en ough m a y r emain to ca ta lyse t he ex tr action o f enough energy from th e hole to p owe r a long- liv e d b urst.
Do uble peaked bu rst s
- ----- --- - ----- ---- -
The r e are atl east twom ec han i s ms which migh t l ead to a de l ayed “ seco nd" burst (or a d o uble humped bu rst) . One po s sibi li ty is that a m erg er l eads toacentral NS, tempora ri l y sta bilize d | at_least of_order $\simg
10^{15}$ G. Indeed,_it is_not_obvious why_the_fields cannot become_even higher. Note_that the virial limit_is $B_v \sim_10^{17}$_G.
After magnetic fields have built up to some fraction of the equipartition value with_the_shear motion,_a_magnetic_viscosity develops. Assuming that $B_rB_\phi\sim_B^2$, it can be characterized_in the_usual way by the parameter $\alpha\sim B^2/(4\pi \rho_v_s^2_) \sim 10^{-1}_B_{15}^2
\rho_{13}^{-1} T_9^{-1}$. This viscosity continues operating also after cooling_has led to the disappearance of_neutrino viscosity. Assuming_a_value_of $\alpha=0.1$, a BH_mass 3 $\msun$ and outer disk_radius equal to the Roche lobe_size, Popham et al. (1998) estimate “viscous" life times_of 0.1 s for NS/BH-NS, 10–20_s for a collapsar (failed_SN Ib_or rotating WR), and 15–150_s for WD-BH_and He-BH_systems (although fields_of $10^{15}$ G may be more_difficult to support_in He-BH systems).
A magnetic field configuration_capable_of powering the_bursts_is_likely to_have a large_scale_structure. Flares_and_instabilities occurring on the characteristic (millisecond)_dynamical_time scale would cause substantial irregularity or_intermittency in the overall_outflow_that would manifest itself_in internal shocks (Rees &,_1994) There is thus no problem_in principle_in accounting_for sporadic large-amplitude variability, on all time scales down to a_millisecond, even in the most long-lived_bursts. Note also that_it only_takes_a residual cold_disk_of $10^{-3}\msun$_to confine a field of $10^{15}$ G,_which can_extract energy from the black hole_via the Blandford-Znajek mechanism._Even_if the evolution time scale for_the bulk of the debris were_no more than a second,_enough_may_remain to catalyse the extraction_of enough energy from the hole_to power a_long-lived burst.
Double peaked bursts
--------------------
There are at least_two_mechanisms which might lead to a_delayed_“second" burst (or a double humped_burst)._One_possibility is that a merger_leads to a central NS, temporarily_stabilized |
pa}^{\alpha}\; ds.$$ We first transform $$\int_{-D}^D {\lpa u + n - s \rpa}^{H'}\psi (u)\, du$$ through $\ell$ successive integrations by parts. Recalling (\[border\]), we see that each time the border terms $I^m\psi(\pm D)$, $0\leq m\leq \ell$, vanish. In the end, we obtain $$\begin{aligned}
\label{ijn1}
I^1_{n}& = & C\,\int_0^{n - 2D} {\lpa \int_{-D}^D {\lpa u + n - s \rpa}^{H'
- \ell}I^{\ell}\psi(u) \; du\rpa}^{\alpha}\; ds\nonumber\\
& \leq & C\,\int_0^{n - 2D} {\lpa D + n - s \rpa}^{\alpha(H' - \ell)}\;
ds\nonumber\\
& \leq & C\,\int_D^{+\infty} {s}^{\alpha(H - \ell) -1}\; ds,\end{aligned}$$ where in the first inequality we used $\ell>H> H'$. The integral on the right-hand side is clearly finite and independent of $n$.
The integral over the second domain is given, after a change of variable, by $$\begin{aligned}
\label{ijn2}
I^2_{n}= \int_0^{3D} {\lpa \int_{-D}^D {\lpa u + 2D - s\rpa}_+^{H'}
\psi(u) \; du\rpa}^{\alpha} ds &
\leq & C\, \int_0^{3D} {\lpa \int_0^{3D}u^{H'}\, du\rpa}^{\alpha} ds.\end{aligned}$$ Since $H' > -1/2$, the integral on the right-hand side is again finite and independent of $n$. Putting (\[ijn1\]) and (\[ijn2\]) together yields (\[sup\]) as desired.
The following upper bound on $||R||$ is a direct consequence of (\[ineqr2\]) and (\[sup\]): $$\label{iner}
||R||\; \leq\; C\sum_{j\geq 0} 2^{-(H-\beta - 1/p)j | pa}^{\alpha}\; ds.$$ We first transform $ $ \int_{-D}^D { \lpa u + n - s \rpa}^{H'}\psi (u)\, du$$ through $ \ell$ successive integration by part. Recalling (\[border\ ]), we see that each time the margin terms $ I^m\psi(\pm D)$, $ 0\leq m\leq \ell$, vanish. In the conclusion, we prevail $ $ \begin{aligned }
\label{ijn1 }
I^1_{n } & = & C\,\int_0^{n - 2D } { \lpa \int_{-D}^D { \lpa u + n - s \rpa}^{H'
- \ell}I^{\ell}\psi(u) \; du\rpa}^{\alpha}\; ds\nonumber\\
& \leq & C\,\int_0^{n - 2D } { \lpa D + n - second \rpa}^{\alpha(H' - \ell)}\;
ds\nonumber\\
& \leq & C\,\int_D^{+\infty } { s}^{\alpha(H - \ell) -1}\; ds,\end{aligned}$$ where in the first inequality we used $ \ell > H > H'$. The built-in on the right - hand side is clearly finite and autonomous of $ n$.
The integral over the second domain is give, after a variety of variable star, by $ $ \begin{aligned }
\label{ijn2 }
I^2_{n}= \int_0^{3D } { \lpa \int_{-D}^D { \lpa u + 2D - s\rpa}_+^{H' }
\psi(u) \; du\rpa}^{\alpha } ds &
\leq & C\, \int_0^{3D } { \lpa \int_0^{3D}u^{H'}\, du\rpa}^{\alpha } ds.\end{aligned}$$ Since $ H' > -1/2 $, the integral on the right - hand english is again finite and independent of $ n$. Putting (\[ijn1\ ]) and (\[ijn2\ ]) together yields (\[sup\ ]) as desire.
The following upper bound on $ ||R||$ is a direct consequence of (\[ineqr2\ ]) and (\[sup\ ] ): $ $ \label{iner }
||R||\; \leq\; C\sum_{j\geq 0 } 2^{-(H-\beta - 1 / p)j | pa}^{\appha}\; ds.$$ We first transfovm $$\int_{-D}^D {\lpa u + u - s \rpe}^{H'}\psi (u)\, du$$ throjgh $\ell$ successive integratilnw by kcrts. Recalling (\[bordef\]), we see nhat each timt the border terms $I^m\psi(\pm D)$, $0\les m\lex \ell$, vanish. In the end, wa obtain $$\begin{dlkgued}
\label{ijn1}
I^1_{n}& = & C\,\int_0^{n - 2D} {\lpa \int_{-D}^D {\l[a u + n - d \rpa}^{H'
- \ell}I^{\elj}\psi(l) \; du\rla}^{\alpha}\; ds\nonumber\\
& \leq & C\,\int_0^{n - 2D} {\lla D + n - s \rpa}^{\alpha(H' - \ell)}\;
ds\nonumber\\
& \leq & C\,\int_D^{+\lnftj} {s}^{\alpha(H - \ell) -1}\; ds,\fnd{aligned}$$ qherq in the firsg inequaliuy we used $\eml>H> H'$. The integral on the right-fand xide is clgcely xinite and mndepegdent of $n$.
Thc integsal ovet the second dpmamn iw given, after a changx of variable, by $$\begyn{aligned}
\nayel{ijn2}
I^2_{n}= \int_0^{3D} {\lpa \int_{-E}^D {\lpa o + 2D - s\rpx}_+^{Y'}
\psk(u) \; vu\rla}^{\alphw} da &
\leq & C\, \jnt_0^{3D} {\lpa \inr_0^{3D}u^{H'}\, du\rpa}^{\alpha} ds.\emd{woigned}$$ Since $G' > -1/2$, thq yntegral on the right-hand side is again fihite and independent of $n$. Putting (\[ijn1\]) and (\[ijj2\]) togethew yields (\[sup\]) as desired.
The following upper bound mn $||R||$ ms a blvect xojsequence of (\[ineqr2\]) and (\[sup\]): $$\label{iner}
||R||\; \leq\; C\sui_{n\gtq 0} 2^{-(H-\beta - 1/p)j | pa}^{\alpha}\; ds.$$ We first transform $$\int_{-D}^D {\lpa n s \rpa}^{H'}\psi du$$ through $\ell$ (\[border\]), see that each the border terms D)$, $0\leq m\leq \ell$, vanish. In end, we obtain $$\begin{aligned} \label{ijn1} I^1_{n}& = & C\,\int_0^{n - 2D} {\lpa \int_{-D}^D u + n - s \rpa}^{H' - \ell}I^{\ell}\psi(u) \; du\rpa}^{\alpha}\; ds\nonumber\\ & \leq C\,\int_0^{n 2D} D n - s \rpa}^{\alpha(H' - \ell)}\; ds\nonumber\\ & \leq & C\,\int_D^{+\infty} {s}^{\alpha(H - \ell) -1}\; ds,\end{aligned}$$ in the first inequality we used $\ell>H> H'$. integral on the right-hand is clearly finite and independent $n$. integral over second is after a change variable, by $$\begin{aligned} \label{ijn2} I^2_{n}= \int_0^{3D} {\lpa \int_{-D}^D {\lpa u + 2D - s\rpa}_+^{H'} \psi(u) \; du\rpa}^{\alpha} & \leq \int_0^{3D} {\lpa du\rpa}^{\alpha} Since > -1/2$, the the right-hand side is again finite $n$. Putting (\[ijn1\]) and (\[ijn2\]) together yields (\[sup\]) desired. The upper bound on $||R||$ is a consequence of (\[ineqr2\]) and (\[sup\]): $$\label{iner} ||R||\; \leq\; 0} 2^{-(H-\beta - 1/p)j | pa}^{\alpha}\; ds.$$ We first transform $$\Int_{-D}^D {\lpa u + n - S \rpa}^{H'}\Psi (U)\, du$$ ThRougH $\ell$ Successive inteGRatiOns by parts. Recalling (\[borDer\]), we SeE That EAcH time The bordER tERMs $I^M\pSi(\Pm D)$, $0\LeQ M\lEq \ell$, VanIsh. In thE end, we obtaIn $$\bEgIn{aligned}
\labEL{iJn1}
I^1_{n}& = & C\,\int_0^{n - 2D} {\Lpa \Int_{-D}^D {\lpa u + n - s \rPa}^{H'
- \Ell}I^{\elL}\pSi(u) \; DU\rpa}^{\aLphA}\; ds\noNumber\\
& \LEq & C\,\int_0^{N - 2D} {\lpa D + n - s \rPa}^{\ALpha(H' - \eLL)}\;
ds\nonuMBEr\\
& \Leq & C\,\Int_D^{+\infty} {s}^{\alpha(H - \eLL) -1}\; dS,\End{aligned}$$ wherE in the FiRSt INEquAliTy we used $\elL>H> h'$. The iNTegral oN ThE RIGht-HAnd side is cleaRly finite anD IndEpendeNt Of $n$.
tHe inteGral oVeR The Second domaiN is gIven, after A changE Of variaBLe, by $$\begIn{aligNed}
\LabEl{ijN2}
i^2_{n}= \InT_0^{3D} {\lPa \INt_{-D}^d {\LpA u + 2D - S\Rpa}_+^{h'}
\psi(u) \; du\rPa}^{\AlPha} ds &
\Leq & C\, \INT_0^{3d} {\Lpa \iNt_0^{3D}U^{H'}\, du\Rpa}^{\alPha} ds.\end{alignEd}$$ SInce $h' > -1/2$, The IntegRal on The rIgHt-hanD side iS agaiN fInite and indepenDent Of $n$. PuttinG (\[ijN1\]) aNd (\[iJn2\]) TogetHEr yielDs (\[sUp\]) aS desireD.
The folLOwiNg UPPEr Bound on $||R||$ is a direct cOnSEQuEnce of (\[inEqr2\]) and (\[SUp\]): $$\LaBEl{iner}
||R||\; \lEq\; c\suM_{j\geQ 0} 2^{-(h-\Beta - 1/p)J | pa}^{\alpha}\; ds.$$ We fi rst transf orm $ $\i nt_ {- D}^D {\l pa u + n - s \ r pa}^ {H'}\psi (u)\, du$$ th rough $ \ ell$ su ccess ive int e gr a t ion sby pa rt s .Recal lin g (\[bo rder\]), w e s ee that each t i me the borde r t erms $I^m\ps i(\ pm D)$ ,$0\ l eq m\ leq \ell $, van i sh. In the end, w e obtai n $$\beg i n {a lign ed}
\label{ijn1}I ^1 _ {n}& = & C\,\i nt_0^{ n- 2 D } {\ lpa \int_{-D} ^D {\lp a u + n- s \ r pa} ^ {H'
- \ell}I ^{\ell}\psi ( u)\; du\ rp a}^ { \alpha }\; d s\ n onu mber\\
& \l eq & C\,\int_ 0^{n - 2D} {\l p a D + n - s \ rpa }^{ \alp h a( H' -\e l l)} \ ;ds\ n onu mber\\
& \ le q & C \,\i n t _ D ^{+\ inf ty}{s}^{ \alpha(H - \e ll) -1} \ ; d s,\en d{ali gned }$ $ whe re inthe f ir st inequality w e us ed $\ell> H>H' $.Th e int e gral o n t heright-h and sid e is c l e a rl y finite and indep en d e nt of $n$.
Thei nt eg r al overth e s econ d domai n is gi ven, aft er a c h an ge of var ia ble, b y$$\ beg in{al i gned }
\lab el{ijn2}
I^2_ { n}= \int_0^{3D } {\lpa \int_{ - D} ^ D { \ lpau + 2D - s\rpa }_+^ { H'}\psi ( u) \; du\r pa}^{ \a l ph a } ds &
\leq & C\, \i nt_0^{ 3D} { \lpa \int_0^{ 3D}u^{H'}\ , d u\rpa}^{ \alp h a} ds.\end{aligne d}$$Since $H'> -1/2$,the i ntegralon the ri g h t-hand s ide is ag ain f in ite and indep e n dent o f $n$.Put ting (\ [ij n1\ ])and ( \[ijn2\]) togethe ryi el ds (\ [sup\ ] ) as des ir ed.
The foll o wing u pperboun don $|| R||$ is ad i rect c on sequ enc eof (\ [ine q r2\ ]) and(\[sup\]) : $ $ \lab el {i ner}
|| R||\; \leq\;C\ sum_{j\geq 0 } 2 ^{-(H- \ b eta - 1/ p)j | pa}^{\alpha}\; ds.$$_We first_transform $$\int_{-D}^D {\lpa u_+ n_-_s \rpa}^{H'}\psi_(u)\,_du$$ through $\ell$_successive integrations by_parts. Recalling (\[border\]), we_see that each_time_the border terms $I^m\psi(\pm D)$, $0\leq m\leq \ell$, vanish. In the end, we obtain_$$\begin{aligned}
\label{ijn1}
I^1_{n}&_= &_C\,\int_0^{n_-_2D} {\lpa \int_{-D}^D {\lpa u_+ n - s \rpa}^{H'
_- \ell}I^{\ell}\psi(u)_\; du\rpa}^{\alpha}\; ds\nonumber\\
& \leq & C\,\int_0^{n - 2D}_{\lpa_D + n_- s \rpa}^{\alpha(H' - \ell)}\;
ds\nonumber\\
& \leq & C\,\int_D^{+\infty} {s}^{\alpha(H_- \ell) -1}\; ds,\end{aligned}$$ where in_the first inequality_we_used_$\ell>H> H'$. The integral_on the right-hand side is clearly_finite and independent of $n$.
The integral_over the second domain is given, after_a change of variable, by $$\begin{aligned}
\label{ijn2}
I^2_{n}=_\int_0^{3D} {\lpa \int_{-D}^D {\lpa u_+ 2D_- s\rpa}_+^{H'}
\psi(u) \; du\rpa}^{\alpha}_ds &
\leq &_C\, _\int_0^{3D} {\lpa \int_0^{3D}u^{H'}\,_du\rpa}^{\alpha} ds.\end{aligned}$$ Since $H' > -1/2$,_the integral on_the right-hand side is again finite_and_independent of $n$._Putting_(\[ijn1\])_and (\[ijn2\])_together yields (\[sup\])_as_desired.
The following_upper_bound on $||R||$ is a direct_consequence_of (\[ineqr2\]) and (\[sup\]): $$\label{iner}
||R||\; \leq\; C\sum_{j\geq_0} 2^{-(H-\beta - 1/p)j |
Scientific, 2004.
Y. Aharonov and D. Bohm, Phys. Rev. [**115**]{}, 485 (1959).
---
abstract: 'The CM class number one problem for elliptic curves asked to find all elliptic curves defined over the rationals with non-trivial endomorphism ring. For genus-2 curves it is the problem of determining all CM curves of genus $2$ defined over the *reflex field*. We solve the problem by showing that the list given in Bouyer–Streng [@bouyer Tables 1a, 1b, 2b, and 2c] is complete.'
author:
- 'P[i]{}nar K[i]{}l[i]{}çer[^1] and Marco Streng[^2]'
bibliography:
-'mybib.bib'
title: |
The CM class number one problem for\
curves of genus $2$
---
Introduction
============
By a *curve*, we always mean a projective smooth geometrically irreducible algebraic curve. A curve $C$ over a field $k$ of genus $g$ has *complex multiplication* (CM) if the endomorphism ring of its Jacobian over $\overline{k}$ is an order ${\mathcal{O}}$ in a *CM field* of that is, a totally imaginary quadratic extension of a totally real number field of We say that $C$ by ${\mathcal{O}}$. For example, an elliptic has CM if ${\text{\normalfont{End}}}(E_{\overline{k}})$ is an order in an imaginary quadratic field $K$. An elliptic curve $E$ with CM by an order ${\mathcal{O}}_{K}$ can be defined over ${\mathbb{Q}}$ if and only if the class group ${\text{\normalfont{Cl}}}_{K}:=I_{K}/P_{K}$ is trivial. The CM class number one problem for elliptic curves asks to determine all imaginary quadratic fields of class number one. This problem was solved by Baker [@baker] (1966) and Stark [@stark] (1967); the fields are $K\cong{\mathbb{Q}}(\sqrt{-d})$ where $d\in \{1,\ 2,\ 3,\ 7,\ 11,\ 19,\ 43,\ 67,\ 163\}$.
We consider the analogous problem for curves of genus $2$.
In, we recall the definition of the *CM class group* $I_{K^ | Scientific, 2004.
Y. Aharonov and D. Bohm, Phys. Rev. [ * * 115 * * ] { }, 485 (1959).
---
abstract:' The CM class number one problem for elliptic curve ask to find all elliptic curves define over the rationals with non - trivial endomorphism band. For genus-2 crook it is the problem of determine all CM curves of genus $ 2 $ defined over the * automatic field *. We solve the problem by showing that the tilt given in Bouyer – Streng [ @bouyer Tables 1a, 1b, 2b, and 2c ] is complete.'
writer:
-' P[i]{}nar K[i]{}l[i]{}çer[^1 ] and Marco Streng[^2 ]'
bibliography:
-'mybib.bib'
title: |
The CM class act one problem for\
curves of genus $ 2 $
---
Introduction
= = = = = = = = = = = =
By a * bend *, we always mean a projective smooth geometrically irreducible algebraic curve. A curvature $ C$ over a field $ k$ of genus $ g$ has * complex multiplication * (CM) if the endomorphism ring of its Jacobian over $ \overline{k}$ is an order $ { \mathcal{O}}$ in a * CM discipline * of that is, a totally imaginary quadratic extension of a totally real numeral field of We say that $ C$ by $ { \mathcal{O}}$. For example, an elliptic has CM if $ { \text{\normalfont{End}}}(E_{\overline{k}})$ is an order in an imaginary quadratic field $ K$. An elliptic curve $ E$ with CM by an ordering $ { \mathcal{O}}_{K}$ can be defined over $ { \mathbb{Q}}$ if and only if the course group $ { \text{\normalfont{Cl}}}_{K}:=I_{K}/P_{K}$ is fiddling. The CM course number one problem for elliptic curves asks to determine all imaginary quadratic fields of course number one. This problem was solved by Baker [ @baker ] (1966) and Stark [ @stark ] (1967); the fields are $ K\cong{\mathbb{Q}}(\sqrt{-d})$ where $ d\in \{1,\ 2,\ 3,\ 7,\ 11,\ 19,\ 43,\ 67,\ 163\}$.
We consider the analogous problem for curves of genus $ 2$.
In, we recall the definition of the * CM class group * $ I_{K^ | Sclentific, 2004.
Y. Aharonov and D. Nohm, Phys. Rev. [**115**]{}, 485 (1959).
---
cvstracv: 'The CJ class vumber one problem for ellipvic xurvew asked to find all eluiptic cugves defibed icer the ravjonals with nkk-trivnao endomorphism ring. For cenus-2 curves id ks the problem of determining all CM surves pf genus $2$ defineq ovtr ehe *dvfoex field*. We solve the problej by shmwing that thr list given in Bouyer–Strejg [@bluyer Tables 1a, 1b, 2b, and 2c] is cimplqre.'
author:
- 'P[i]{}nxr K[i]{}l[i]{}çer[^1] and Marco Stteng[^2]'
bibliography:
-'mybib.bib'
title: |
The CM class bunbeg one problen for\
curves on genus $2$
---
Introdiction
============
By a *curye*, we alqays mean a projectivx smooth geometricaljy irreduwiyle algebraic curve. A cyrve $W$ ovar a dieud $i$ pf genus $g$ ias *complex multiplicarion* (CM) if the endokowinism ring of its Jwcjbian over $\overline{k}$ is an order ${\mathcan{O}}$ jn a *CM field* of that iw, a totally imaginary quadratis extension of a totally real number field of We vay tiag $C$ bn ${\magycwl{O}}$. For example, an elliptic has CM if ${\text{\norizlgokt{End}}}(E_{\overline{k}})$ ls an order in an omwgogary quadratiz fielb $K$. An elliptic curve $E$ with CM bt an ordew ${\mayhcal{O}}_{K}$ can be defined over ${\mathbb{Q}}$ if cnd only if the class group ${\text{\uormalgont{Ck}}}_{K}:=I_{K}/P_{K}$ is trivial. The CO clzss number lne problso for elliptic cjrvvs avks to dtgermine all imagigary quadcatic fields of vlass gumber one. This problem was solved by Bcker [@taker] (1966) and Stark [@stark] (1967); the fields are $K\cong{\mathbb{Q}}(\sqrt{-c})$ fhege $d\in \{1,\ 2,\ 3,\ 7,\ 11,\ 19,\ 43,\ 67,\ 163\}$.
We consider ehe analogous kroblem fjr cufves of gehus $2$.
In, xe recall thq definition mv the *CM clavs group* $I_{K^ | Scientific, 2004. Y. Aharonov and D. Bohm, [**115**]{}, (1959). --- 'The CM class curves to find all curves defined over rationals with non-trivial endomorphism ring. For curves it is the problem of determining all CM curves of genus $2$ over the *reflex field*. We solve the problem by showing that the list in [@bouyer 1a, 2b, and 2c] is complete.' author: - 'P[i]{}nar K[i]{}l[i]{}çer[^1] and Marco Streng[^2]' bibliography: -'mybib.bib' title: | CM class number one problem for\ curves of $2$ --- Introduction ============ a *curve*, we always mean projective geometrically irreducible curve. curve over a field of genus $g$ has *complex multiplication* (CM) if the endomorphism ring of its Jacobian over $\overline{k}$ is order ${\mathcal{O}}$ *CM field* that a imaginary quadratic extension totally real number field of We by ${\mathcal{O}}$. For example, an elliptic has CM ${\text{\normalfont{End}}}(E_{\overline{k}})$ is order in an imaginary quadratic field An elliptic curve $E$ with CM by an ${\mathcal{O}}_{K}$ can be defined over ${\mathbb{Q}}$ if and only if the class group ${\text{\normalfont{Cl}}}_{K}:=I_{K}/P_{K}$ is CM class number one for elliptic curves to all quadratic of class one. This problem was solved by Baker [@baker] (1966) and Stark (1967); the fields are $K\cong{\mathbb{Q}}(\sqrt{-d})$ where $d\in \{1,\ 2,\ 3,\ 19,\ 67,\ 163\}$. We the analogous problem for of $2$. In, we recall of *CM | Scientific, 2004.
Y. Aharonov and D. BoHm, Phys. Rev. [**115**]{}, 485 (1959).
---
aBstraCt: 'THe Cm cLass NumbEr one problem foR ElliPtic curves asked to find aLl ellIpTIc cuRVeS defiNed over THe RATioNaLs WitH nON-tRiviaL enDomorphIsm ring. For GenUs-2 Curves it is thE PrOblem of detErmIning all CM cuRveS of genUs $2$ DefINed ovEr tHe *refLex fieLD*. We solVe the probLeM By showINg that tHE LiSt giVen in Bouyer–Streng [@BOuYEr Tables 1a, 1b, 2b, and 2C] is comPlETe.'
AUThoR:
- 'P[i]{}Nar K[i]{}l[i]{}çer[^1] AnD MarcO streng[^2]'
bIBlIOGRapHY:
-'mybib.bib'
titlE: |
The CM class NUmbEr one pRoBleM For\
curVes of GeNUs $2$
---
INtroduction
============
by a *cUrve*, we alwAys meaN A projecTIve smooTh geomEtrIcaLly iRReDuCibLe ALgeBRaIc cURve. a curve $C$ oVeR a Field $K$ of gENUS $G$ has *ComPlex MultiPlication* (CM) if The EndoMOrpHism rIng of Its JAcObian Over $\ovErlinE{k}$ Is an order ${\mathcaL{O}}$ in A *CM field* oF thAt Is, a ToTally IMaginaRy qUadRatic exTension OF a tOtALLY rEal number field of We SaY THaT $C$ by ${\mathCal{O}}$. FoR ExAmPLe, an elliPtIc hAs CM IF ${\Text{\nOrmaLFoNt{End}}}(E_{\ovErline{K}})$ Is An Order in An ImaginArY quAdrAtic fIEld $K$. an elliPtic curvE $E$ witH cM by an order ${\matHCal{O}}_{K}$ can be defINeD OVeR ${\MathBb{Q}}$ If and only if The cLAss gRoup ${\TExT{\noRMalfoNt{Cl}}}_{K}:=i_{K}/p_{k}$ iS Trivial. The CM class nuMbEr one pRobleM for elliptic cUrves asks tO DETermine aLl imAGiNAry quadratic fiElds oF class numbER one. This ProblEm was solVed by BakeR [@BAker] (1966) and STarK [@stArk] (1967); The FIElDs are $K\cong{\matHBB{Q}}(\sqRt{-D})$ where $d\In \{1,\ 2,\ 3,\ 7,\ 11,\ 19,\ 43,\ 67,\ 163\}$.
WE considEr tHe aNalOgoUs Problem foR curves oF gEnUs $2$.
in, We rEcall THe definiTiOn oF tHe *Cm clasS Group* $I_{k^ | Scientific, 2004.
Y. Aha ronov andD. Bo hm, Ph ys . Re v. [**115**]{}, 4 8 5 (1 959).
---
abstract: ' The C Mc lass nu mberone pro b le m for e ll ipt ic cu rvesask ed to f ind all el lip ti c curves def i ne d over the ra tionals with no n-triv ia l e n domor phi sm ri ng. Fo r genus -2 curves i t is th e proble m of det ermining all CM c u rv e s of genus $2$ defin ed ov e r th e * reflex fie ld *. We solve t h ep r o ble m by showing t hat the lis t gi ven in B ouy e r–Stre ng [@ bo u yer Tables 1a, 1b, 2b, and2c] is complet e .'
auth or:
-'P[ i]{ }nar K[ i] {}l [i ] {}ç e r[ ^1] an d MarcoSt re ng[^2 ]'
b i b l i ogra phy :
-' mybib .bib'
title:|
T h e C M cla ss nu mber o ne pr oblemfor\ curves of gen us $ 2$
---
I ntr od uct io n
=== = ====== ==
By a *cur ve*, we alw ay s m ea n a projective smo ot h ge ometrica lly ir r ed uc i ble alge br aic cur v e . A c urve $C $ over a field $k $of genu s$g$ ha s*co mpl ex mu l tipl icatio n* (CM)if th e endomorphismr ing of its Ja c ob i a no ver$\o verline{k}$ isa n or der$ {\ mat h cal{O }}$ i na * C M field* of that is ,a tota lly i maginary quad ratic exte n s i on of atota l ly real number fi eld o f We say t h at $C$ b y ${\ mathcal{ O}}$. For e xample,anell ipt ich a sCM if ${\text { \ norm al font{En d}} }(E_{\o ver lin e{k }}) $is an ord er in an i ma gi na ryquadr a tic fiel d$K$ .Anellip t ic cur ve $E $ wi th C M by an ord e r$ { \mat hc al {O}} _{K }$ canbe d e fin ed over ${\mathb b{Q } }$ i fan d onlyif the classgr oup ${\tex t{ \no rmalfo n t {Cl}}}_{ K}:=I_{K}/P_{K}$ is tri v ial. Th e C M cla ss n umber one pr oblemfor ellipt ic cur ves a sk s t o deter m i ne al limaginaryq u adr aticfi elds of cla ss number one. Thi s pr oblem was sol ved byB a ke r [ @ ba k er] ( 1 966 ) and Stark [@sta rk] (1967) ;t he fields ar e $K \c ong{\ma thbb{Q} }(\sq r t{-d})$ where $d \in \{1,\ 2 ,\ 3 , \ 7, \ 11,\ 19, \ 43,\ 6 7,\ 163\} $ .
We co nside r t he ana lo gou s pro blem f o r c urves of ge nu s $2$.
In, w e recall the definition of the*CM cl ass g rou p* $I_{K^ | Scientific,_2004.
Y. Aharonov and_D. Bohm, Phys. Rev. [**115**]{}, 485_(1959).
---
abstract:_'The_CM class_number_one problem for_elliptic curves asked_to find all elliptic_curves defined over_the_rationals with non-trivial endomorphism ring. For genus-2 curves it is the problem of determining_all_CM curves_of_genus_$2$ defined over the *reflex_field*. We solve the problem_by showing_that the list given in Bouyer–Streng [@bouyer Tables_1a,_1b, 2b, and_2c] is complete.'
author:
- 'P[i]{}nar K[i]{}l[i]{}çer[^1] and Marco Streng[^2]'
bibliography:
-'mybib.bib'
title: |
_ The CM class_number one problem_for\
__ curves of_genus $2$
---
Introduction
============
By a *curve*, we always_mean a projective smooth geometrically irreducible_algebraic curve. A curve $C$ over a_field $k$ of genus $g$ has_*complex multiplication* (CM) if the_endomorphism ring_of its Jacobian over $\overline{k}$_is an order_${\mathcal{O}}$ in_a *CM field*_of that is, a totally imaginary_quadratic extension of_a totally real number field of_We_say that $C$_by_${\mathcal{O}}$._For example,_an elliptic has_CM_if ${\text{\normalfont{End}}}(E_{\overline{k}})$_is_an order in an imaginary quadratic_field_$K$. An elliptic curve $E$ with CM_by an order ${\mathcal{O}}_{K}$_can_be defined over ${\mathbb{Q}}$_if and only if the_class group ${\text{\normalfont{Cl}}}_{K}:=I_{K}/P_{K}$ is trivial. The_CM class_number one_problem for elliptic curves asks to determine all imaginary quadratic fields_of class number one. This problem_was solved by Baker_[@baker] (1966)_and_Stark [@stark] (1967);_the_fields are_$K\cong{\mathbb{Q}}(\sqrt{-d})$ where $d\in \{1,\ 2,\ 3,\ 7,\_11,\ 19,\_43,\ 67,\ 163\}$.
We consider the analogous_problem for curves of_genus_$2$.
In, we recall the definition of_the *CM class group* $I_{K^ |
this channel will provide a strong indication of the nuclear sticking effect. Thus, an increase in the average charge state of the nuclei, together with sharper positron peaks, could be used to demonstrate the existence of long sticking times and the existence of charged vacuum decay.
The calculations presented in this paper reported single-electron Dirac calculations with spin degeneracy. Thus, for small pair production probabilities it is correct to multiply them by a factor of two. For the (1S $e^-$, free $e^+$) correlated spectra one would have to take into account Pauli blocking effects and make use of many-electron atomic structure once the probabilities become appreciable.
The current work was performed in the monopole approximation to the two-center Coulomb potential. In previous work where static resonance parameters were calculated it was shown that the next-order effect is caused by the quadrupole coupling of the S$-$D states [@eackadconf2007]. While this coupling will certainly modify the resonance, e.g., increasing its energy, $E_{\rm res}$, it is unlikely that it will change the spectra significantly. This conclusion is based on the observation that substantial changes in $\Gamma$ due to higher-order contributions occur at larger internuclear separations where $\Gamma$ is small. It is expected that the main peak will be somewhat higher and broader, as indicated by the Db-Db results, due to the increased width $\Gamma$.
Acknowledgments
===============
The authors would like to thank Igor Khavkine for useful discussions. This work was supported by NSERC Canada, and was carried out using the Shared Hierarchical Academic Research Computing Network (SHARCNET:www.sharcnet.ca). E. Ackad was supported by the Ontario Graduate Scholarship program.
---
abstract: 'In this paper, we propose a new weight initialization method called [*even initialization*]{} for wide and deep nonlinear neural networks with the ReLU activation function. We prove that no poor local minimum exists in the initial loss landscape in the wide and deep nonlinear neural network initialized by the even initialization method that we propose. Specifically, in the initial loss landscape of such a wide and deep ReLU neural network model, the following four statements hold true: 1) the loss function is non-convex and non-concave; 2) every local minimum is a global minimum; 3) every critical point that is not a global minimum is a saddle point; and 4) bad saddle points exist. We also show that the weight values | this channel will provide a strong indication of the nuclear sticking consequence. therefore, an increase in the average cathexis country of the nuclei, together with sharp positron peak, could be used to show the existence of long sticking time and the existence of charged vacuum decay.
The calculations presented in this paper report single - electron Dirac calculations with spin corruption. Thus, for small couple production probability it is correct to reproduce them by a factor of two. For the (1S $ e^-$, free $ e^+$) correlated spectra one would own to take into account Pauli blocking effects and make habit of many - electron atomic structure once the probabilities become appreciable.
The current work was performed in the monopole approximation to the two - center Coulomb potential. In previous work where inactive resonance parameters were account it was show that the next - order consequence is caused by the quadrupole coupling of the S$-$D states [ @eackadconf2007 ]. While this coupling will surely modify the resonance, e.g., increasing its energy, $ E_{\rm res}$, it is unlikely that it will change the spectra significantly. This conclusion is based on the observation that solid changes in $ \Gamma$ ascribable to high - order contributions occur at larger internuclear separation where $ \Gamma$ is small. It is expected that the main acme will be somewhat higher and broader, as indicated by the Db - Db resultant role, due to the increased width $ \Gamma$.
citation
= = = = = = = = = = = = = = =
The authors would like to thank Igor Khavkine for utilitarian discussions. This work was supported by NSERC Canada, and was carried out practice the Shared Hierarchical Academic Research Computing Network (SHARCNET: www.sharcnet.ca). E. Ackad was supported by the Ontario Graduate Scholarship course of study.
---
abstract:' In this paper, we propose a new weight initialization method acting called [ * even initialization * ] { } for wide and thick nonlinear neural networks with the ReLU activation function. We prove that no hapless local minimum exists in the initial loss landscape in the wide and abstruse nonlinear neural network format by the even initialization method that we nominate. Specifically, in the initial loss landscape of such a wide and deep ReLU neural network model, the following four statements retain true: 1) the loss function is non - convex and non - concave; 2) every local minimum is a global minimum; 3) every critical point that is not a global minimum is a saddle detail; and 4) bad saddle points exist. We besides show that the system of weights values | thls channel will provide x strong indicajiin of vhe nucmear stizking effect. Thus, an increasx in the qverage charge state ow the nucpei, togerher qith sharpxd positvjn pswks, eonld be used to cemonstrata the existenca uf long sticking times and the existegce of vhwrged vacuum dgcay.
Tnq cambuoations presented in this papsr repogted single-electrpn Dirac calculations with spij degeneracy. Thus, vor small pqir [eoduction prubabilities it is corrgct to multiply them by a factor of tco. For the (1W $w^-$, fgge $e^+$) correlaved spvctra one would have do take into account Iauli blicking effects and mane use of many-elecjron atomiw atructure once thw probatilidies vecume a'prsciablf.
Thx current wkrk was perdormed in the monopplq approximatioh to tre two-center Coulomb potential. In previouv wkrk where static resonabce parameters were cwlculated it was shown that the next-order effect is caused by tie qucerupouw foupling of the S$-$D states [@eackadconf2007]. While thya volpling will certalnly modify the rexojamse, e.g., increaskng its ehergy, $E_{\rm res}$, it id unlikgly thqt it wilj chsnge the spectra significanrly. This conblusuon is based on thz observatiou that subsyantial changes in $\Gammc$ due fo higher-orfer contrjcutions occur at lagger internuclear separations rhere $\Gamna$ ix small. It os expqcted that the main peak will be slmewhct hicher and bgoader, as indicated by the Db-Db results, due tp dhe increaszd widbh $\Gamma$.
Acknowlqdgments
===============
The aujhors woujd liye to thani Igor Nhavkine fow useful discgdsions. This xork was fuppirtee by NSDFC Canada, and eas carrivd out using the Shared Mieratcgical Academic Xtseqrch Computing Metdorh (DHERCNEE:fww.sharcnet.cd). E. Xckxc was supporuzd ny ghe Pntario Graduate Schmlarahip program.
---
abstrscb: 'In this paper, wq propose a nrw weight initialieation mevhod csllgd [*even initialization*]{} for wide and deep noklinear neuraj nebworhs with thz ReLU activation function. We prove that no poor local minumum exists in the nnltial loss lendsca[e in the wide and deep nonlibear neural netwovk initialized by the sven ititiapization method that we propose. Specifically, in the initial loss landscapw of sncr a wide ans derp ReNU ueural getwlck model, the follpwing four statements hold true: 1) the lovs function is non-convex and npn-zoncave; 2) everi local minimum is a globam minimuk; 3) every critical point that is not a global minimum is a saddlw poiht; and 4) bav saddle points exisy. We anso syow thar thc weight valuer | this channel will provide a strong indication nuclear effect. Thus, increase in the nuclei, with sharper positron could be used demonstrate the existence of long sticking and the existence of charged vacuum decay. The calculations presented in this paper single-electron Dirac calculations with spin degeneracy. Thus, for small pair production probabilities it correct multiply by factor of two. For the (1S $e^-$, free $e^+$) correlated spectra one would have to take account Pauli blocking effects and make use of atomic structure once the become appreciable. The current work performed the monopole to two-center potential. In previous where static resonance parameters were calculated it was shown that the next-order effect is caused by the coupling of states [@eackadconf2007]. this will modify the resonance, its energy, $E_{\rm res}$, it is will change the spectra significantly. This conclusion is on the that substantial changes in $\Gamma$ due higher-order contributions occur at larger internuclear separations where is small. It is expected that the main peak will be somewhat higher and broader, by the Db-Db results, to the increased $\Gamma$. =============== authors like to Igor Khavkine for useful discussions. This work was supported by NSERC and was carried out using the Shared Hierarchical Academic Research (SHARCNET:www.sharcnet.ca). Ackad was supported the Ontario Graduate Scholarship --- 'In this paper, we new initialization initialization*]{} wide deep nonlinear neural networks the ReLU activation function. We that no poor local loss landscape in the wide and deep nonlinear network initialized by the even initialization method we propose. Specifically, in the initial loss landscape of such a wide deep ReLU model, the following four statements hold true: 1) loss function is non-convex non-concave; 2) every local minimum is a global minimum; every point that not a global is a saddle and 4) bad exist. We show the | this channel will provide a stRong indicaTion oF thE nuClEar sTickIng effect. Thus, aN IncrEase in the average charge State Of THe nuCLeI, togeTher witH ShARPer PoSiTroN pEAkS, coulD be Used to dEmonstrate The ExIstence of lonG StIcking timeS anD the existencE of ChargeD vAcuUM decaY.
ThE calcUlatioNS preseNted in thiS pAPer repORted sinGLE-eLectRon Dirac calculatiONs WIth spin degenerAcy. ThuS, fOR sMALl pAir Production PrObabiLIties it IS cORREct TO multiply them By a factor of TWo. FOr the (1S $E^-$, fRee $E^+$) CorrelAted sPeCTra One would havE to tAke into acCount PAUli blocKIng effeCts and MakE usE of mANy-ElEctRoN AtoMIc StrUCtuRe once thE pRoBabilItieS BECOme aPprEciaBle.
ThE current work wAs pErfoRMed In the MonopOle aPpRoximAtion tO the tWo-Center Coulomb poTentIal. In prevIouS wOrk WhEre stATic resOnaNce ParametErs were CAlcUlATED iT was shown that the neXt-ORDeR effect iS causeD By ThE QuadrupoLe CouPlinG OF the S$-$d staTEs [@EackadcoNf2007]. WhilE ThIs CouplinG wIll cerTaInlY moDify tHE resOnance, E.g., increaSing iTS energy, $E_{\rm res}$, iT Is unlikely thaT It WILl CHangE thE spectra sigNifiCAntlY. ThiS CoNclUSion iS baseD oN ThE Observation that subsTaNtial cHangeS in $\Gamma$ due to Higher-ordeR CONtributiOns oCCuR At larger internUcleaR separatioNS where $\GaMma$ is Small. It iS expected THAt the maiN peAk wIll Be sOMEwHat higher and bROAder, As IndicatEd bY the Db-DB reSulTs, dUe tO tHe increasEd width $\GAmMa$.
acKnOwlEdgmeNTs
===============
The autHoRs wOuLd lIke to THank IgOr KhaVkinE fOr USefUl discuSSiONS. ThiS wOrK was SupPoRted bY NSErc CaNada, and Was carrieD ouT UsinG tHe shared HIerarchical AcAdEmic ResearCh comPuting nETwork (SHArCNET:www.sharcnet.ca). E. AckaD Was suppOrtEd by tHe OnTario GradUatE ScholArsHIp progRam.
---
absTract: 'in ThiS PAper, wE PRoPosE a New weight iNITiaLizatIoN metHod callEd [*even initializatiON*]{} foR wide and deep nOnlIneaR NEuRal NEtWOrkS wITh tHE reLU activation fUnction. We pRoVE tHat no poor lOCal MiNimum exIsts in tHe iniTIal loss Landscape In the wide AnD deeP NOnlInear neuraL network InitializED by thE EvEn iniTiaLizatiOn MetHod thAt we prOPosE. SpecIficalLy, In the iNitiaL lOss landsCape of such a wide and deep RELU neuRal neTwoRk model, thE foLLowIng four stAtemEnts hold trUe: 1) tHe lOss fuNctIOn is nOn-coNVeX anD Non-coNcavE; 2) Every locaL MiNimUM Is A global miniMUM; 3) EveRy criTicAL point That Is not a global minimUM is a saddle poinT; and 4) BAD saDdlE PoinTs Exist. We also shoW thAt THE weight vAlUes | this channel will provide a strongindic ati onof the nuc lear stickinge ffec t. Thus, an increase i n the a v erag e c harge stateo ft h e n uc le i,to g et her w ith sharpe r positron pe ak s, could beu se d to demon str ate the exis ten ce oflo ngs ticki ngtimes and t h e exis tence ofch a rged v a cuum de c a y.
Th e calculations pr e se n ted in this pa per re po r te d sin gle -electronDi rac c a lculati o ns w i ths pin degenerac y. Thus, fo r sm all pa ir pr o ductio n pro ba b ili ties it iscorr ect to mu ltiply them by a facto r of t wo. Fo r th e ( 1S $e ^- $ , f r ee $e ^ +$) correla te dspect ra o n e w ould ha ve t o tak e into accoun t P auli blo cking effe ctsan d mak e useof ma ny -electron atomi c st ructure o nce t hepr obabi l itiesbec ome apprec iable.The c u r r en t work was perform ed i nthe mono pole a p pr ox i mation t othe two - c enter Cou l om b potent ial. I n p re vious w or k wher esta tic reso n ance param eters we re ca l culated it was shown that th e n e x t- o rder ef fect is cau sedb y th e qu a dr upo l e cou pling o f t h e S$-$D states [@ea ck adconf 2007] . While thiscoupling w i l l certain ly m o di f y the resonanc e, e. g., increa s ing itsenerg y, $E_{\ rm res}$, i t is unl ike lytha t i t wi ll change the s pect ra signif ica ntly. T his co ncl usi on is based on theob se rv at ion that substant ia l c ha nge s in$ \Gamma $ due tohi gh e r-o rder co n tr i b utio ns o ccur at l arger int e rnu clear s eparation s w h ere$\ Ga mma$ is small. It is e xpected th at th e main p eak will be somewhat higher and broader , a s ind icat ed by the Db -Db re sul t s, due to th e inc re ase d width $ \G amm a$ .
Acknowl e d gme nts
= == ==== ======= =
The authors wou l d l ike to thankIgo r Kh a v ki nef or use fu l di s c ussions. This w ork was su pp o rt ed by NSER C Ca na da, and was ca rried out usi ng the Sh ared Hier ar chic a l Ac ademic Res earch Co mputing N e twork (S HARCN ET: www.sh ar cne t.ca) . E. A c kad wassuppor te d by t he On ta rio Grad uate Scholarship progra m.
-- -
abs tra ct: 'In t his pap er, we pr opos e a new we igh t i nitia liz a tionmeth o dcal l ed [* even initializ a ti on* ] { }for wide an d d eep nonl ine a r neur al n etworks with theR eLU activation fun c t ion . W e pro ve that no poorloc al m inimum e xi sts in theinitiallo s s lan dscape in th e widea n dd eep no nlin ear neural n etw or k initia li ze d by th e ev en initi alizat i on m e t hod that we prop ose.S p ecifi c all y, in t he init i al l oss landsc ape of such a wid e an d dee p ReLUne ural n etw or k model, t h e followi ng fo ur stat em ents ho ld tru e: 1 ) the l ossfu nct ion is no n - co n ve xa ndnon- conca ve ; 2) every lo c al minim umi s a glo ba l m i n imum;3 )e v ery critic alpoint t hat is not a gl o ba l mini mum is a sad dle poi n t;an d 4) ba d s a d dle point s exist.W e a ls o sh ow thatth e we ig htva lues | this_channel will_provide a strong indication_of the_nuclear_sticking effect._Thus,_an increase in_the average charge_state of the nuclei,_together with sharper_positron_peaks, could be used to demonstrate the existence of long sticking times and the_existence_of charged_vacuum_decay.
The_calculations presented in this paper_reported single-electron Dirac calculations with_spin degeneracy._Thus, for small pair production probabilities it is_correct_to multiply them_by a factor of two. For the (1S $e^-$,_free $e^+$) correlated spectra one would_have to take_into_account_Pauli blocking effects and_make use of many-electron atomic structure_once the probabilities become appreciable.
The current_work was performed in the monopole approximation_to the two-center Coulomb potential. In_previous work where static resonance_parameters were_calculated it was shown that_the next-order effect_is caused_by the quadrupole_coupling of the S$-$D states [@eackadconf2007]._While this coupling_will certainly modify the resonance, e.g.,_increasing_its energy, $E_{\rm_res}$,_it_is unlikely_that it will_change_the spectra_significantly._This conclusion is based on the_observation_that substantial changes in $\Gamma$ due to_higher-order contributions occur at_larger_internuclear separations where $\Gamma$_is small. It is expected_that the main peak will be_somewhat higher_and broader,_as indicated by the Db-Db results, due to the increased width_$\Gamma$.
Acknowledgments
===============
The authors would like to thank_Igor Khavkine for useful_discussions. This_work_was supported by_NSERC_Canada, and_was carried out using the Shared Hierarchical_Academic Research_Computing Network (SHARCNET:www.sharcnet.ca). E. Ackad was_supported by the Ontario_Graduate_Scholarship program.
---
abstract: 'In this paper,_we propose a new weight initialization_method called [*even initialization*]{} for_wide_and_deep nonlinear neural networks with_the ReLU activation function. We prove_that no poor_local minimum exists in the initial loss_landscape_in the wide and deep nonlinear_neural_network initialized by the even initialization_method_that_we propose. Specifically, in the_initial loss landscape of such a_wide and deep ReLU neural network model, the following_four statements hold_true: 1) the loss function_is_non-convex_and non-concave; 2) every local minimum is a global minimum;_3) every_critical point that_is not a global minimum is a saddle point; and_4) bad saddle points exist. We also_show that the weight values |
which for power-law inflation can be derived using a trick of integrating by parts $$\tau \equiv \int \frac{dt}{a(t)} = \int \frac{da}{a^2 H} =
-\frac{1}{aH} + \int \frac{\epsilon \, da}{a^2 H} \,,$$ which for constant $\epsilon$ implies $$\tau = -\frac{1}{aH} \, \frac{1}{1-\epsilon} \,.$$
With these results, Eq. (\[ufield\]) for the perturbations reduces to a Bessel equation $$\label{bessel}
\left[ \frac{d^2}{d\tau^2} +k^2 -\frac{(\nu^2 - 1/4)}{\tau^2}
\right] u_k =0 \,,$$ where $$\nu \equiv \frac{3}{2} + \frac{\epsilon}{1-\epsilon} \,,$$ is a constant. The solution with the correct short-scale behaviour, shown in Eq. (\[short\]), is $$u_k(\tau) = \frac{\sqrt{\pi}}{2} e^{i(\nu +1/2)\pi/2}
(-\tau)^{1/2} H_{\nu}^{(1)} (-k\tau) \,,$$ where $H_{\nu}^{(1)}$ is the Hankel function of the first kind of order $\nu$.
The result we desire is the asymptotic form of the solution; taking $k/aH
\rightarrow 0$ gives the asymptotic form $$u_k \rightarrow e^{i(\nu -1/2)\pi /2} 2^{\nu -3/2}
\frac{\Gamma (\nu)}{\Gamma (3/2)} \frac{1}{\sqrt{2k}}
(-k \tau )^{-\nu +1/2} \,,$$ where $\Gamma$ is the usual gamma function. On substitution into the expression for the power spectrum, Eq. (\[pspec\]), this gives $$\label{scalaramp}
{\cal P}_{{\cal R}}^{1/2}(k) =2^{\nu -3/2} \frac{\Gamma(\nu)}{\Gamma(3/2)}
(\nu-1/2)^{1/2 - \nu} \frac{2}{m_{{\rm Pl}}^2} \left.
\frac{H^2}{|H' | which for power - law inflation can be derived use a whoremaster of integrating by parts $ $ \tau \equiv \int \frac{dt}{a(t) } = \int \frac{da}{a^2 H } =
-\frac{1}{aH } + \int \frac{\epsilon \, da}{a^2 henry } \,,$$ which for constant $ \epsilon$ implies $ $ \tau = -\frac{1}{aH } \, \frac{1}{1-\epsilon } \,.$$
With these resultant role, Eq. (\[ufield\ ]) for the perturbation reduces to a Bessel equality $ $ \label{bessel }
\left [ \frac{d^2}{d\tau^2 } + k^2 -\frac{(\nu^2 - 1/4)}{\tau^2 }
\right ] u_k = 0 \,,$$ where $ $ \nu \equiv \frac{3}{2 } + \frac{\epsilon}{1-\epsilon } \,,$$ is a constant. The solution with the right short - scale behavior, shown in Eq. (\[short\ ]), is $ $ u_k(\tau) = \frac{\sqrt{\pi}}{2 } e^{i(\nu +1/2)\pi/2 }
(-\tau)^{1/2 } H_{\nu}^{(1) } (-k\tau) \,,$$ where $ H_{\nu}^{(1)}$ is the Hankel function of the first kind of club $ \nu$.
The result we hope is the asymptotic form of the solution; take $ kelvin / aH
\rightarrow 0 $ gives the asymptotic form $ $ u_k \rightarrow e^{i(\nu -1/2)\pi /2 } 2^{\nu -3/2 }
\frac{\Gamma (\nu)}{\Gamma (3/2) } \frac{1}{\sqrt{2k } }
(-k \tau) ^{-\nu +1/2 } \,,$$ where $ \Gamma$ is the usual da gamma function. On substitution into the expression for the world power spectrum, Eq. (\[pspec\ ]), this gives $ $ \label{scalaramp }
{ \cal P}_{{\cal R}}^{1/2}(k) = 2^{\nu -3/2 } \frac{\Gamma(\nu)}{\Gamma(3/2) }
(\nu-1/2)^{1/2 - \nu } \frac{2}{m_{{\rm Pl}}^2 } \left.
\frac{H^2}{|H' | whlch for power-law inflatiun can be derivgd using a tridk of ingegrating by parts $$\tau \equiv \ibt \frqc{dt}{a(t)} = \int \frac{da}{a^2 H} =
-\frac{1}{wH} + \int \drac{\tpsilon \, da}{a^2 H} \,,$$ wijch for constzkt $\epvmlon$ implies $$\tao = -\frac{1}{aH} \, \fsac{1}{1-\epsilon} \,.$$
Witv ghzse results, Eq. (\[ufield\]) for the perturbwtions tefuces to a Besfel tquwtioh $$\label{bessel}
\left[ \frac{d^2}{d\tau^2} +k^2 -\frac{(\hu^2 - 1/4)}{\tau^2}
\right] u_k =0 \,,$$ where $$\nu \equiv \frac{3}{2} + \frwc{\epdilon}{1-\epsilon} \,,$$ is a constant. Tye sjoution with ghe correcu vhort-scale behaviour, shown in Eq. (\[short\]), is $$j_k(\tau) = \frac{\sqrt{\kn}}{2} e^{i(\jo +1/2)\pi/2}
(-\tau)^{1/2} H_{\nu}^{(1)} (-h\tau) \,,$$ where $M_{\mu}^{(1)}$ is dhe Hanlel function on the fiest kind of order $\nu$.
Tie result we desire ys the asfm'totic form of the soourion; jakinc $k/aF
\riehtzrcow 0$ gived tie asymptotjc form $$u_k \eightarrow e^{i(\nu -1/2)\pi /2} 2^{\nt -3/2}
\frac{\Gamja (\nu)}{\Gwmia (3/2)} \frac{1}{\sqrt{2k}}
(-k \tau )^{-\nu +1/2} \,,$$ where $\Gammd$ ia the usual gamma functuon. On substitution ijto the evpression for the power spectrum, Eq. (\[pspec\]), this givas $$\laueu{sccoaramo}
{\xap P}_{{\cal R}}^{1/2}(k) =2^{\nu -3/2} \frac{\Gamma(\nu)}{\Gamma(3/2)}
(\nu-1/2)^{1/2 - \nu} \fras{2}{j_{{\rk Il}}^2} \left.
\frac{H^2}{|M' | which for power-law inflation can be derived trick integrating by $$\tau \equiv \int = + \int \frac{\epsilon da}{a^2 H} \,,$$ for constant $\epsilon$ implies $$\tau = \, \frac{1}{1-\epsilon} \,.$$ With these results, Eq. (\[ufield\]) for the perturbations reduces to Bessel equation $$\label{bessel} \left[ \frac{d^2}{d\tau^2} +k^2 -\frac{(\nu^2 - 1/4)}{\tau^2} \right] u_k =0 \,,$$ $$\nu \frac{3}{2} \frac{\epsilon}{1-\epsilon} is a constant. The solution with the correct short-scale behaviour, shown in Eq. (\[short\]), is $$u_k(\tau) \frac{\sqrt{\pi}}{2} e^{i(\nu +1/2)\pi/2} (-\tau)^{1/2} H_{\nu}^{(1)} (-k\tau) \,,$$ where is the Hankel function the first kind of order The we desire the form the solution; taking \rightarrow 0$ gives the asymptotic form $$u_k \rightarrow e^{i(\nu -1/2)\pi /2} 2^{\nu -3/2} \frac{\Gamma (\nu)}{\Gamma (3/2)} \frac{1}{\sqrt{2k}} \tau )^{-\nu where $\Gamma$ the gamma On substitution into for the power spectrum, Eq. (\[pspec\]), {\cal P}_{{\cal R}}^{1/2}(k) =2^{\nu -3/2} \frac{\Gamma(\nu)}{\Gamma(3/2)} (\nu-1/2)^{1/2 - \frac{2}{m_{{\rm Pl}}^2} \frac{H^2}{|H' | which for power-law inflation Can be derivEd usiNg a TriCk Of inTegrAting by parts $$\taU \EquiV \int \frac{dt}{a(t)} = \int \frac{da}{a^2 h} =
-\frac{1}{AH} + \INt \frAC{\ePsiloN \, da}{a^2 H} \,,$$ whICh FOR coNsTaNt $\ePsILoN$ implIes $$\Tau = -\frac{1}{AH} \, \frac{1}{1-\epsiLon} \,.$$
wiTh these resulTS, EQ. (\[ufield\]) for The PerturbationS reDuces tO a besSEl equAtiOn $$\labEl{bessEL}
\left[ \fRac{d^2}{d\tau^2} +k^2 -\FrAC{(\nu^2 - 1/4)}{\tau^2}
\RIght] u_k =0 \,,$$ wHERe $$\Nu \eqUiv \frac{3}{2} + \frac{\epsiloN}{1-\EpSIlon} \,,$$ is a constanT. The soLuTIoN WIth The Correct shoRt-Scale BEhaviouR, ShOWN In EQ. (\[Short\]), is $$u_k(\tau) = \fRac{\sqrt{\pi}}{2} e^{i(\NU +1/2)\pi/2}
(-\Tau)^{1/2} H_{\nu}^{(1)} (-K\tAu) \,,$$ wHEre $H_{\nu}^{(1)}$ Is the haNKel Function of tHe fiRst kind of Order $\nU$.
the resuLT we desiRe is thE asYmpTotiC FoRm Of tHe SOluTIoN; taKIng $K/aH
\rightArRoW 0$ giveS the ASYMPtotIc fOrm $$u_K \righTarrow e^{i(\nu -1/2)\pi /2} 2^{\nU -3/2}
\frAc{\GaMMa (\nU)}{\GammA (3/2)} \frac{1}{\Sqrt{2K}}
(-k \Tau )^{-\nu +1/2} \,,$$ Where $\GAmma$ iS tHe usual gamma funCtioN. On substiTutIoN inTo The exPRessioN foR thE power sPectrum, eQ. (\[psPeC\]), THIs Gives $$\label{scalaramP}
{\cAL p}_{{\cAl R}}^{1/2}(k) =2^{\nu -3/2} \frAc{\GammA(\Nu)}{\gaMMa(3/2)}
(\nu-1/2)^{1/2 - \nu} \frAc{2}{M_{{\rm pl}}^2} \leFT.
\Frac{H^2}{|h' | which for power-law infla tion can b e der ive d u si ng a tri ck of integrat i ng b y parts $$\tau \equiv\int\f r ac{d t }{ a(t)} = \int \f r a c{d a} {a ^2H} =
-\f rac{1}{ aH} + \int \f ra c{\epsilon \ , d a}{a^2 H}\,, $$ which for co nstant $ \ep s ilon$ im plies $$\ta u = -\f rac{1}{aH }\ , \fra c {1}{1-\ e p si lon} \,.$$
With thes e r e sults, Eq. (\[ ufield \] ) f o r th e p erturbatio ns redu c es to a Be s s e l e q uation $$\lab el{bessel}\ lef t[ \fr ac {d^ 2 }{d\ta u^2}+k ^ 2 - \frac{(\nu^ 2 -1/4)}{\ta u^2} \rig h t] u_k=0 \,, $$whe re $ $ \n u\eq ui v \f r ac {3} { 2}+ \frac{ \e ps ilon} {1-\ e p s i lon} \, ,$$is aconstant. The so luti o n w ith t he co rrec tshort -scale beha vi our, shown in E q. ( \[short\] ),is $$ u_ k(\ta u ) = \f rac {\s qrt{\pi }}{2} e ^ {i( \n u + 1/ 2)\pi/2}
(-\t au ) ^ {1 /2} H_{\ nu}^{( 1 )} ( - k\tau) \ ,, $$wher e $H_{\ nu}^ { (1 )}$ is t he Han k el f unction o f thefi rst ki nd of orde r $\nu $.
Theresul t we desire ist he asymptotic fo r m o f the so lution; tak ing$ k/aH
\r i gh tar r ow 0$ give st he asymptotic form $$u _k \righ tarro w e^{i(\nu -1 /2)\pi /2} 2 ^ {\nu -3/ 2} \frac{\Gamma ( \nu)} {\Gamma (3 / 2)} \fra c{1}{ \sqrt{2k }}
( - k \tau )^ {-\ nu+1/ 2}\ , ,$ $ where $\Gam m a $ is t he usua l g amma fu nct ion . O n s ub stitution into th eex pr es sio n for the powe rspe ct rum , Eq. (\[psp ec\]) , th is g i ves $$\lab e l{ s c alar am p}
{\c alP} _{{\c al R } }^{ 1/2}(k) =2^{\nu-3/ 2 } \f ra c{ \Gamma( \nu)}{\Gamma( 3/ 2)}
( \n u-1 /2)^{1 / 2 - \nu}\frac{2}{m_{{\rm Pl}}^2 } \left.
\f rac{ H^2}{|H' | which_for power-law_inflation can be derived_using a_trick_of integrating_by_parts $$\tau \equiv_\int \frac{dt}{a(t)} =_\int \frac{da}{a^2 H} =_
__-\frac{1}{aH} + \int \frac{\epsilon \, da}{a^2 H} \,,$$ which for constant $\epsilon$ implies $$\tau_=_-\frac{1}{aH} \,_\frac{1}{1-\epsilon}_\,.$$
With_these results, Eq. (\[ufield\]) for the_perturbations reduces to a Bessel_equation $$\label{bessel}
\left[_\frac{d^2}{d\tau^2} +k^2 -\frac{(\nu^2 - 1/4)}{\tau^2}
__\right] u_k =0_\,,$$ where $$\nu \equiv \frac{3}{2} + \frac{\epsilon}{1-\epsilon} \,,$$ is_a constant. The solution with the_correct short-scale behaviour,_shown_in_Eq. (\[short\]), is $$u_k(\tau) =_\frac{\sqrt{\pi}}{2} e^{i(\nu +1/2)\pi/2}
_ (-\tau)^{1/2} H_{\nu}^{(1)} (-k\tau) \,,$$ where_$H_{\nu}^{(1)}$ is the Hankel function of the_first kind of order $\nu$.
The result_we desire is the asymptotic_form of_the solution; taking $k/aH
\rightarrow_0$ gives the_asymptotic form_$$u_k \rightarrow e^{i(\nu_-1/2)\pi /2} 2^{\nu -3/2}
_ \frac{\Gamma_(\nu)}{\Gamma (3/2)} \frac{1}{\sqrt{2k}}
__(-k \tau )^{-\nu_+1/2}_\,,$$_where $\Gamma$_is the usual_gamma_function. On_substitution_into the expression for the power_spectrum,_Eq. (\[pspec\]), this gives $$\label{scalaramp}
{\cal P}_{{\cal R}}^{1/2}(k) =2^{\nu_-3/2} \frac{\Gamma(\nu)}{\Gamma(3/2)}
__ (\nu-1/2)^{1/2 - \nu}_\frac{2}{m_{{\rm Pl}}^2} \left.
_ \frac{H^2}{|H' |
& 1.474 & 0.0027 & 28.300 & 17.038 & 12.474\
\
.01 & 1.575 & 0.0033 & 28.096 & 16.281 & 12.491 &.02 & 1.480 & 0.0029 & 28.099 & 16.353 & 12.354\
.03 & 1.527 & 0.0033 & 28.111 & 16.315 & 12.734 &.04 & 1.519 & 0.0030 & 28.106 & 16.342 & 12.453\
.05 & 1.541 & 0.0030 & 28.109 & 16.400 & 12.495 &.06 & 1.526 & 0.0029 & 28.116 & 16.374 & 12.524\
.07 & 1.532 & 0.0032 & 28.104 & 16.348 & 12.641 &.08 & 1.564 & 0.0039 & 28.095 & 16.266 & 12.659\
.09 & 1.456 & 0.0032 & 28.101 & 16.425 & 12.557 &.10 & 1.489 & 0.0029 & 28.108 & 16.347 & 12.581\
.11 & 1.486 & 0.0030 & 28.124 & 16.372 & 12.374 &.12 & 1.547 & 0.0037 & 28.122 & 16.398 & 12.655\
.13 & 1.495 & 0.0030 & 28.122 & 16.381 & 12.323 &.14 & 1.500 & 0.0029 & 28.127 & 16.386 & 12.367\
.15 & 1.466 & 0.0026 & 28.127 & 16.395 & 12.258 &.16 & 1.628 & 0.0032 & 28.113 & 16.186 & 12.530\
.17 & 1.653 & 0.0030 & 28.098 & 16.194 & 12.488 &.18 & 1.563 & 0.0034 & 28.105 & 16.273 & 12.800\
.19 & 1.579 & 0.0026 & 28.104 & 16.223 & 12.573 &.20 & 1.572 & 0.0029 & 28.117 & 16.266 & 12.567\
.21 & 1.581 & 0.0033 & 28. | & 1.474 & 0.0027 & 28.300 & 17.038 & 12.474\
\
.01 & 1.575 & 0.0033 & 28.096 & 16.281 & 12.491 & .02 & 1.480 & 0.0029 & 28.099 & 16.353 & 12.354\
.03 & 1.527 & 0.0033 & 28.111 & 16.315 & 12.734 & .04 & 1.519 & 0.0030 & 28.106 & 16.342 & 12.453\
.05 & 1.541 & 0.0030 & 28.109 & 16.400 & 12.495 & .06 & 1.526 & 0.0029 & 28.116 & 16.374 & 12.524\
.07 & 1.532 & 0.0032 & 28.104 & 16.348 & 12.641 & .08 & 1.564 & 0.0039 & 28.095 & 16.266 & 12.659\
.09 & 1.456 & 0.0032 & 28.101 & 16.425 & 12.557 & .10 & 1.489 & 0.0029 & 28.108 & 16.347 & 12.581\
.11 & 1.486 & 0.0030 & 28.124 & 16.372 & 12.374 & .12 & 1.547 & 0.0037 & 28.122 & 16.398 & 12.655\
.13 & 1.495 & 0.0030 & 28.122 & 16.381 & 12.323 & .14 & 1.500 & 0.0029 & 28.127 & 16.386 & 12.367\
.15 & 1.466 & 0.0026 & 28.127 & 16.395 & 12.258 & .16 & 1.628 & 0.0032 & 28.113 & 16.186 & 12.530\
.17 & 1.653 & 0.0030 & 28.098 & 16.194 & 12.488 & .18 & 1.563 & 0.0034 & 28.105 & 16.273 & 12.800\
.19 & 1.579 & 0.0026 & 28.104 & 16.223 & 12.573 & .20 & 1.572 & 0.0029 & 28.117 & 16.266 & 12.567\
.21 & 1.581 & 0.0033 & 28. | & 1.474 & 0.0027 & 28.300 & 17.038 & 12.474\
\
.01 & 1.575 & 0.0033 & 28.096 & 16.281 & 12.491 &.02 & 1.480 & 0.0029 & 28.099 & 16.353 & 12.354\
.03 & 1.527 & 0.0033 & 28.111 & 16.315 & 12.734 &.04 & 1.519 & 0.0030 & 28.106 & 16.342 & 12.453\
.05 & 1.541 & 0.0030 & 28.109 & 16.400 & 12.495 &.06 & 1.526 & 0.0029 & 28.116 & 16.374 & 12.524\
.07 & 1.532 & 0.0032 & 28.104 & 16.348 & 12.641 &.08 & 1.564 & 0.0039 & 28.095 & 16.266 & 12.659\
.09 & 1.456 & 0.0032 & 28.101 & 16.425 & 12.557 &.10 & 1.489 & 0.0029 & 28.108 & 16.347 & 12.581\
.11 & 1.486 & 0.0030 & 28.124 & 16.372 & 12.374 &.12 & 1.547 & 0.0037 & 28.122 & 16.398 & 12.655\
.13 & 1.495 & 0.0030 & 28.122 & 16.381 & 12.323 &.14 & 1.500 & 0.0029 & 28.127 & 16.386 & 12.367\
.15 & 1.466 & 0.0026 & 28.127 & 16.395 & 12.258 &.16 & 1.628 & 0.0032 & 28.113 & 16.186 & 12.530\
.17 & 1.653 & 0.0030 & 28.098 & 16.194 & 12.488 &.18 & 1.563 & 0.0034 & 28.105 & 16.273 & 12.800\
.19 & 1.579 & 0.0026 & 28.104 & 16.223 & 12.573 &.20 & 1.572 & 0.0029 & 28.117 & 16.266 & 12.567\
.21 & 1.581 & 0.0033 & 28. | & 1.474 & 0.0027 & 28.300 & 12.474\ .01 & & 0.0033 & &.02 1.480 & 0.0029 28.099 & 16.353 12.354\ .03 & 1.527 & 0.0033 28.111 & 16.315 & 12.734 &.04 & 1.519 & 0.0030 & 28.106 & & 12.453\ .05 & 1.541 & 0.0030 & 28.109 & 16.400 & 12.495 & & & & 16.374 & 12.524\ .07 & 1.532 & 0.0032 & 28.104 & 16.348 & 12.641 &.08 1.564 & 0.0039 & 28.095 & 16.266 & .09 & 1.456 & & 28.101 & 16.425 & &.10 1.489 & & & & 12.581\ .11 1.486 & 0.0030 & 28.124 & 16.372 & 12.374 &.12 & 1.547 & 0.0037 & 28.122 & & 12.655\ 1.495 & & & & 12.323 &.14 & 0.0029 & 28.127 & 16.386 & 1.466 & 0.0026 & 28.127 & 16.395 12.258 &.16 1.628 & 0.0032 & 28.113 & & 12.530\ .17 & 1.653 & 0.0030 & & 16.194 & 12.488 &.18 & 1.563 & 0.0034 & 28.105 & 16.273 & 12.800\ 1.579 & 0.0026 & & 16.223 & &.20 1.572 0.0029 28.117 & & 12.567\ .21 & 1.581 & 0.0033 & 28. | & 1.474 & 0.0027 & 28.300 & 17.038 & 12.474\
\
.01 & 1.575 & 0.0033 & 28.096 & 16.281 & 12.491 &.02 & 1.480 & 0.0029 & 28.099 & 16.353 & 12.354\
.03 & 1.527 & 0.0033 & 28.111 & 16.315 & 12.734 &.04 & 1.519 & 0.0030 & 28.106 & 16.342 & 12.453\
.05 & 1.541 & 0.0030 & 28.109 & 16.400 & 12.495 &.06 & 1.526 & 0.0029 & 28.116 & 16.374 & 12.524\
.07 & 1.532 & 0.0032 & 28.104 & 16.348 & 12.641 &.08 & 1.564 & 0.0039 & 28.095 & 16.266 & 12.659\
.09 & 1.456 & 0.0032 & 28.101 & 16.425 & 12.557 &.10 & 1.489 & 0.0029 & 28.108 & 16.347 & 12.581\
.11 & 1.486 & 0.0030 & 28.124 & 16.372 & 12.374 &.12 & 1.547 & 0.0037 & 28.122 & 16.398 & 12.655\
.13 & 1.495 & 0.0030 & 28.122 & 16.381 & 12.323 &.14 & 1.500 & 0.0029 & 28.127 & 16.386 & 12.367\
.15 & 1.466 & 0.0026 & 28.127 & 16.395 & 12.258 &.16 & 1.628 & 0.0032 & 28.113 & 16.186 & 12.530\
.17 & 1.653 & 0.0030 & 28.098 & 16.194 & 12.488 &.18 & 1.563 & 0.0034 & 28.105 & 16.273 & 12.800\
.19 & 1.579 & 0.0026 & 28.104 & 16.223 & 12.573 &.20 & 1.572 & 0.0029 & 28.117 & 16.266 & 12.567\
.21 & 1.581 & 0.0033 & 28. | & 1.474 & 0.0027 & 28.300 & 17.038& 12. 474 \
\
. 01 & 1.5 75 & 0.0033 &2 8.09 6 & 16.281 & 12.491 &. 02 &1. 4 80 & 0. 0029& 28.09 9 & 1 6.3 53 & 12 .3 5 4\
.03& 1 .527 &0.0033 & 2 8.1 11 & 16.315 &1 2. 734 &.04 & 1. 519 & 0.0030 &28.106 & 16 . 342 & 12 .453\
.05 & 1.541& 0.0030&2 8.109& 16.400 & 1 2.49 5 &.06 & 1.526 &0 .0 0 29 & 28.116 &16.374 & 12 . 5 24\
.0 7 & 1.532&0.003 2 & 28.1 0 4& 1 6.3 4 8 & 12.641 &. 08 & 1.564& 0. 0039 & 2 8.0 9 5 & 16 .266&1 2.6 59\
.09 & 1 .456 & 0.0032 & 28. 1 01 & 16 . 425 & 1 2.557&.1 0 & 1.4 8 9&0.0 02 9 &2 8. 108 & 1 6.347 &12 .5 81\
. 11 & 1 . 4 86 & 0. 0030 & 28 .124 & 16.372 &12.3 7 4 & .12 & 1.54 7 &0. 0037& 28.1 22 &16 .398 & 12.655\.13& 1.495 & 0. 00 30&28.12 2 & 16. 381 &12.323&.14 &1 .50 0& 0 .0 029 & 28.127 & 16. 38 6 &12.367\.15 &1 .4 66 & 0.0026 & 28 .127 & 16.3 95 & 12 .258 &.1 6 & 1. 6 28 & 0.0032 & 28.11 3& 1 6.1 86 &1 2.53 0\
.17 & 1.653 & 0. 0 030 & 28.098 & 16.194 & 12.4 8 8& . 18 & 1. 563 & 0.0034 & 28. 1 05 & 16. 2 73 &1 2.800 \
.19 & 1. 5 79 & 0.0026 & 28.10 4& 16.2 23 &12.573 &.20 & 1.572 & 0 . 0 0 29 & 28. 117& 1 6 .266 & 12.567\
.21& 1.581 &0 .0033 &28. | &_1.474 &_0.0027 & 28.300 &_17.038 &_12.474\
\
.01_& 1.575_&_0.0033 & 28.096_& 16.281 &_12.491 &.02 & 1.480_& 0.0029 &_28.099_& 16.353 & 12.354\
.03 & 1.527 & 0.0033 & 28.111 & 16.315 & 12.734_&.04_& 1.519_&_0.0030_& 28.106 & 16.342 &_12.453\
.05 & 1.541 & 0.0030_& 28.109_& 16.400 & 12.495 &.06 & 1.526 &_0.0029_& 28.116 &_16.374 & 12.524\
.07 & 1.532 & 0.0032 & 28.104_& 16.348 & 12.641 &.08 &_1.564 & 0.0039_&_28.095_& 16.266 & 12.659\
.09_& 1.456 & 0.0032 & 28.101_& 16.425 & 12.557 &.10 &_1.489 & 0.0029 & 28.108 & 16.347_& 12.581\
.11 & 1.486 & 0.0030_& 28.124 & 16.372 &_12.374 &.12_& 1.547 & 0.0037 &_28.122 & 16.398_& 12.655\
.13_& 1.495 &_0.0030 & 28.122 & 16.381 &_12.323 &.14 &_1.500 & 0.0029 & 28.127 &_16.386_& 12.367\
.15 &_1.466_&_0.0026 &_28.127 & 16.395_&_12.258 &.16_&_1.628 & 0.0032 & 28.113 &_16.186_& 12.530\
.17 & 1.653 & 0.0030 &_28.098 & 16.194 &_12.488_&.18 & 1.563 &_0.0034 & 28.105 & 16.273_& 12.800\
.19 & 1.579 & 0.0026_& 28.104_& 16.223_& 12.573 &.20 & 1.572 & 0.0029 & 28.117 & 16.266_& 12.567\
.21 & 1.581 & 0.0033_& 28. |
\theta };q)_{k}}{(q,abq^{\alpha },acq^{\alpha
},adq^{\alpha },abcdq^{\alpha -1};q)_{k}}\ q^{k} \notag \\
& \qquad \times ~_{10}W_{9}(abcdq^{2\alpha +k-1};q^{\alpha },bcq^{\alpha
-1},bdq^{\alpha -1},cdq^{\alpha -1},q^{k+1},abcdq^{2\alpha
+n+k-1},q^{k-n};q,a^{2}). \label{aaw4}\end{aligned}$$There is another useful representation of the associated Askey–Wilson polynomials in terms of a double series due to Rahman, $$\begin{aligned}
p_{n}^{\alpha }(x)& =p_{n}^{\alpha }(x;a,b,c,d|q)\smallskip \notag \\
& =\frac{(abcdq^{2\alpha -1},q^{\alpha +1};q)_{n}}{(q,abcdq^{\alpha
-1};q)_{n}}q^{-\alpha n}\sum_{k=0}^{n}\frac{(q^{-n},abcdq^{2\alpha
+n-1};q)_{k}}{(q^{\alpha +1},abq^{\alpha };q)_{k}}\smallskip \notag \\
& \qquad \times \frac{(aq^{\alpha }e^{i\theta },aq^{\alpha }e^{-i\theta
};q)_{k}}{(acq^{\alpha },acq^{\alpha };q)_{k}}\sum_{j=0}^{k}\frac{(q^{\alpha
},abq^{\alpha -1},acq^{\alpha -1},adq^{\alpha -1};q)_{j}}{(q,abcdq^{2\alpha
-2},aq^{\alpha }e^{i\theta },aq^{\alpha }e^{-i\theta };q)_{j}}q^{j},
\label{aaw5}\end{aligned}$$where $x=\cos \theta $ (see [@Ga:Ra Exercises 8.26–8.27] and [Rahman96]{}, [@Rah2000]). This formula will be the starting point for our investigation.
An Overview of the Main Result
==============================
To construct an eigenvalue problem for the associated | \theta }; q)_{k}}{(q, abq^{\alpha }, acq^{\alpha
}, adq^{\alpha }, abcdq^{\alpha -1};q)_{k}}\ q^{k } \notag \\
& \qquad \times ~_{10}W_{9}(abcdq^{2\alpha + k-1};q^{\alpha }, bcq^{\alpha
-1},bdq^{\alpha -1},cdq^{\alpha -1},q^{k+1},abcdq^{2\alpha
+ n+k-1},q^{k - n};q, a^{2 }). \label{aaw4}\end{aligned}$$There is another useful representation of the associated Askey – Wilson polynomials in term of a bivalent series due to Rahman, $ $ \begin{aligned }
p_{n}^{\alpha } (x) & = p_{n}^{\alpha } (x;a, b, c, d|q)\smallskip \notag \\
& = \frac{(abcdq^{2\alpha -1},q^{\alpha +1};q)_{n}}{(q, abcdq^{\alpha
-1};q)_{n}}q^{-\alpha n}\sum_{k=0}^{n}\frac{(q^{-n},abcdq^{2\alpha
+ n-1};q)_{k}}{(q^{\alpha +1},abq^{\alpha }; q)_{k}}\smallskip \notag \\
& \qquad \times \frac{(aq^{\alpha } e^{i\theta }, aq^{\alpha } e^{-i\theta
}; q)_{k}}{(acq^{\alpha }, acq^{\alpha }; q)_{k}}\sum_{j=0}^{k}\frac{(q^{\alpha
}, abq^{\alpha -1},acq^{\alpha -1},adq^{\alpha -1};q)_{j}}{(q, abcdq^{2\alpha
-2},aq^{\alpha } e^{i\theta }, aq^{\alpha } e^{-i\theta }; q)_{j}}q^{j },
\label{aaw5}\end{aligned}$$where $ x=\cos \theta $ (visualize [ @Ga: Ra Exercises 8.26–8.27 ] and [ Rahman96 ] { }, [ @Rah2000 ]). This convention will be the starting point for our probe.
An Overview of the Main Result
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
To construct an eigenvalue problem for the associated | \theha };q)_{k}}{(q,abq^{\alpha },acq^{\alpha
},aaq^{\alpha },abcdq^{\alkhq -1};q)_{k}}\ q^{n} \notzg \\
& \qquaa \times ~_{10}W_{9}(abcdq^{2\alpha +k-1};q^{\alpha },ucq^{\aopha
-1},beq^{\alpha -1},cdq^{\alpha -1},q^{k+1},abcaq^{2\alpha
+n+k-1},e^{k-n};q,a^{2}). \lqbel{eaw4}\end{aligned}$$Thecs is another hdefun representatiok of the asvociated Askey–Fiusln polynomials in terms of a double series dke to Rahman, $$\bggin{akygnes}
i_{n}^{\clpha }(x)& =p_{n}^{\alpha }(x;a,b,c,d|q)\smallskip \notag \\
& =\frac{(abcdq^{2\allha -1},q^{\alpha +1};q)_{n}}{(q,abcdq^{\alpha
-1};q)_{n}}e^{-\alpja n}\sum_{k=0}^{n}\frac{(q^{-n},abcfq^{2\alpha
+n-1};q)_{k}}{(q^{\qlphw +1},abq^{\alpha };q)_{k}}\rmallskip \notag \\
& \qquas \times \frac{(aq^{\alpha }e^{i\theta },aq^{\aupha }z^{-i\theta
};q)_{k}}{(acw^{\aophw },acq^{\alpha };q)_{j}}\sum_{j=0}^{h}\frac{(q^{\alpha
},ana^{\alpha -1},acq^{\alpna -1},adq^{\alpha -1};q)_{j}}{(a,abrdq^{2\aopha
-2},aq^{\alpha }e^{i\theta },ax^{\alpha }e^{-i\theta };q)_{j}}q^{t},
\label{aaw5}\anb{aligned}$$where $x=\cos \thwtq $ (seg [@Ga:Rd Exdecires 8.26–8.27] aid [Dahman96]{}, [@Rai2000]). This formhla will be the starting point fjg our investifation.
Wn Overview of the Main Result
==============================
To construcu an sigenvalue problem for rhe associated | \theta };q)_{k}}{(q,abq^{\alpha },acq^{\alpha },adq^{\alpha },abcdq^{\alpha -1};q)_{k}}\ q^{k} & \times ~_{10}W_{9}(abcdq^{2\alpha },bcq^{\alpha -1},bdq^{\alpha -1},cdq^{\alpha useful of the associated polynomials in terms a double series due to Rahman, p_{n}^{\alpha }(x)& =p_{n}^{\alpha }(x;a,b,c,d|q)\smallskip \notag \\ & =\frac{(abcdq^{2\alpha -1},q^{\alpha +1};q)_{n}}{(q,abcdq^{\alpha -1};q)_{n}}q^{-\alpha n}\sum_{k=0}^{n}\frac{(q^{-n},abcdq^{2\alpha +n-1};q)_{k}}{(q^{\alpha };q)_{k}}\smallskip \notag \\ & \qquad \times \frac{(aq^{\alpha }e^{i\theta },aq^{\alpha }e^{-i\theta };q)_{k}}{(acq^{\alpha },acq^{\alpha };q)_{k}}\sum_{j=0}^{k}\frac{(q^{\alpha -1},acq^{\alpha -1};q)_{j}}{(q,abcdq^{2\alpha }e^{i\theta }e^{-i\theta };q)_{j}}q^{j}, \label{aaw5}\end{aligned}$$where $x=\cos \theta $ (see [@Ga:Ra Exercises 8.26–8.27] and [Rahman96]{}, [@Rah2000]). This formula will the starting point for our investigation. An Overview the Main Result ============================== construct an eigenvalue problem for associated | \theta };q)_{k}}{(q,abq^{\alpha },acq^{\alpha
},aDq^{\alpha },abcDq^{\alpHa -1};q)_{K}}\ q^{k} \NoTag \\
& \qQuad \Times ~_{10}W_{9}(abcdq^{2\alpHA +k-1};q^{\aLpha },bcq^{\alpha
-1},bdq^{\alpha -1},cdQ^{\alphA -1},q^{K+1},AbcdQ^{2\AlPha
+n+k-1},Q^{k-n};q,a^{2}). \laBEl{AAW4}\enD{aLiGneD}$$THErE is anOthEr usefuL representAtiOn Of the associaTEd askey–WilsoN poLynomials in tErmS of a doUbLe sERies dUe tO RahmAn, $$\begiN{AligneD}
p_{n}^{\alpha }(x)& =P_{n}^{\ALpha }(x;a,B,C,d|q)\smalLSKiP \notAg \\
& =\frac{(abcdq^{2\alpha -1},q^{\ALpHA +1};q)_{n}}{(q,abcdq^{\alpha
-1};Q)_{n}}q^{-\alpHa N}\SuM_{K=0}^{N}\frAc{(q^{-N},abcdq^{2\alphA
+n-1};Q)_{k}}{(q^{\alPHa +1},abq^{\alPHa };Q)_{K}}\SMalLSkip \notag \\
& \qquaD \times \frac{(aQ^{\AlpHa }e^{i\thEtA },aq^{\ALpha }e^{-i\Theta
};Q)_{k}}{(ACq^{\aLpha },acq^{\alphA };q)_{k}}\sUm_{j=0}^{k}\frac{(q^{\Alpha
},aBQ^{\alpha -1},aCQ^{\alpha -1},aDq^{\alphA -1};q)_{j}}{(Q,abCdq^{2\aLPhA
-2},aQ^{\alPhA }E^{i\tHEtA },aq^{\ALphA }e^{-i\theta };Q)_{j}}Q^{j},
\Label{Aaw5}\eND{ALIgneD}$$whEre $x=\Cos \thEta $ (see [@Ga:Ra ExeRciSes 8.26–8.27] aND [RaHman96]{}, [@RAh2000]). ThiS forMuLa wilL be the StartInG point for our invEstiGation.
An OVerViEw oF tHe MaiN result
==============================
to cOnsTruct an EigenvaLUe pRoBLEM fOr the associated | \theta };q)_{k}}{(q,abq^{\ alpha },ac q^{\a lph a
} ,a dq^{ \alp ha },abcdq^{\a l pha-1};q)_{k}}\ q^{k} \n otag\\ & \q q ua d \ti mes ~_{ 1 0} W _ {9} (a bc dq^ {2 \ al pha + k-1 };q^{\a lpha },bcq ^{\ al pha
-1},bdq^ { \a lpha -1},c dq^ {\alpha -1}, q^{ k+1},a bc dq^ { 2\alp ha+n+k- 1},q^{ k -n};q, a^{2}). \l a bel{aa w 4}\end{ a l ig ned} $$There is anothe r u s eful represent ationof th e ass oci ated Askey –W ilson polynom i al s i n t e rms of a doub le series d u e t o Rahm an , $ $ \begin {alig ne d }
p _{n}^{\alph a }( x)& =p_{n }^{\al p ha }(x; a ,b,c,d| q)\sma lls kip \n o ta g\\&= \fr a c{ (ab c dq^ {2\alpha - 1} ,q^{\ alph a + 1 };q) _{n }}{( q,abc dq^{\alpha
-1 };q )_{n } }q^ {-\al pha n }\su m_ {k=0} ^{n}\f rac{( q^ {-n},abcdq^{2\a lpha
+n-1};q) _{k }} {(q ^{ \alph a +1},a bq^ {\a lpha }; q)_{k}} \ sma ll s k i p \notag \\
& \qqua d\ t im es \frac {(aq^{ \ al ph a }e^{i\t he ta},aq ^ { \alph a }e ^ {- i\theta};q)_{ k }} {( acq^{\a lp ha },a cq ^{\ alp ha }; q )_{k }}\sum _{j=0}^{ k}\fr a c{(q^{\alpha
} , abq^{\alpha - 1 }, a c q^ { \alp ha-1},adq^{\a lpha -1}; q)_{ j }} {(q , abcdq ^{2\a lp h a- 2},aq^{\alpha }e^{i \t heta } ,aq^{ \alpha }e^{-i \theta };q ) _ { j}}q^{j} ,
\l a be l {aaw5}\end{ali gned} $$where $x = \cos \th eta $ (see [@ Ga:Ra Exe r c ises 8.2 6–8 .27 ] a nd[ R ah man96]{}, [@R a h 2000 ]) . Thisfor mula wi llbethe st ar ting poin t for ou rin ve st iga tion.
An Over vi ewof th e Mai n Resul t
=== ==== == == = === ======= = == = = ===
T ocons tru ct an e igen v alu e probl em for th e a s soci at ed | \theta };q)_{k}}{(q,abq^{\alpha_},acq^{\alpha
},adq^{\alpha },abcdq^{\alpha_-1};q)_{k}}\ q^{k} \notag_\\
& \qquad_\times_~_{10}W_{9}(abcdq^{2\alpha +k-1};q^{\alpha_},bcq^{\alpha
-1},bdq^{\alpha_-1},cdq^{\alpha -1},q^{k+1},abcdq^{2\alpha
+n+k-1},q^{k-n};q,a^{2}). _\label{aaw4}\end{aligned}$$There is another_useful representation of the_associated Askey–Wilson polynomials_in_terms of a double series due to Rahman, $$\begin{aligned}
p_{n}^{\alpha }(x)& =p_{n}^{\alpha }(x;a,b,c,d|q)\smallskip \notag_\\
&_=\frac{(abcdq^{2\alpha -1},q^{\alpha_+1};q)_{n}}{(q,abcdq^{\alpha
-1};q)_{n}}q^{-\alpha_n}\sum_{k=0}^{n}\frac{(q^{-n},abcdq^{2\alpha
+n-1};q)_{k}}{(q^{\alpha_+1},abq^{\alpha };q)_{k}}\smallskip \notag \\
&_\qquad \times \frac{(aq^{\alpha }e^{i\theta },aq^{\alpha_}e^{-i\theta
};q)_{k}}{(acq^{\alpha },acq^{\alpha_};q)_{k}}\sum_{j=0}^{k}\frac{(q^{\alpha
},abq^{\alpha -1},acq^{\alpha -1},adq^{\alpha -1};q)_{j}}{(q,abcdq^{2\alpha
-2},aq^{\alpha }e^{i\theta },aq^{\alpha }e^{-i\theta };q)_{j}}q^{j},
\label{aaw5}\end{aligned}$$where_$x=\cos_\theta $ (see_[@Ga:Ra Exercises 8.26–8.27] and [Rahman96]{}, [@Rah2000]). This formula will be_the starting point for our investigation.
An_Overview of the_Main_Result
==============================
To_construct an eigenvalue problem_for the associated |
label{eq:py}\end{aligned}$$ $Z = \sum_{{\mathbf{y}}} \sum_{{\mathbf{z}}} \exp \left( \mathcal{E}({\mathbf{y}}, {\mathbf{z}}| {\mathbf{x}}) \right)$ is the partition function.
In the case of our Embedded Latent CRF model, as in the LDCRF model, latent states are deterministically partitioned to correspond to output values. That is, the number of latent states is a multiple of the number of output values, and $\psi_{zy}=0$ for pairs in the partitioning and $-\infty$ otherwise. The other scoring functions are learned as global bilinear parameter matrices.
In order to manage large numbers of latent states without overfitting, the Embedded Latent CRF enforces an additional restriction that the scoring function $\psi_{zz}$ possess a low-rank structure: that is, $\psi_{zz}(z_i,z_j)=z_i^\top U V^\top z_j$, where $U$ and $V$ are skinny rectangular matrices and $z_i,z_j$ are represented by one-hot vectors.
While inference in this model is tractable using tree belief propagation even when learning $\psi_{zy}$, the deterministic factors make it especially simple to implement.
Computing the quantities involved in can be carried out efficiently with dynamic programming using the forward algorithm, as in HMMs. To see this, note that to compute the numerator $\sum_{{\mathbf{z}}} \exp \left( \mathcal{E}({\mathbf{y}}, {\mathbf{z}}| {\mathbf{x}}) \right)$, given an output label ${\mathbf{y}}$, we can fold the local scores $\psi_{zf}$ and $\psi_{zy}$ into one score $\psi_{zfy}(f_t, z_t; y_t)$, and summing the resulting energy corresponds exactly to the forward algorithm in a CRF with $M$ states. The partition function can also be computed by dynamic programming: $$\begin{aligned}
Z &= \sum_{{\mathbf{y}}} \sum_{{\mathbf{z}}} \exp \left( \mathcal{E}({\mathbf{y}}, {\mathbf{z}}| {\mathbf{x}}) \right) \\
&= \sum_{{\mathbf{z}}} \sum_{{\mathbf{y}}} \prod_t \exp \left( \psi | label{eq: py}\end{aligned}$$ $ Z = \sum_{{\mathbf{y } } } \sum_{{\mathbf{z } } } \exp \left (\mathcal{E}({\mathbf{y } }, { \mathbf{z}}| { \mathbf{x } }) \right)$ is the partition function.
In the case of our Embedded Latent CRF model, as in the LDCRF exemplar, latent state are deterministically partitioned to correspond to output value. That is, the number of latent states is a multiple of the issue of output signal values, and $ \psi_{zy}=0 $ for pair in the partitioning and $ -\infty$ otherwise. The early scoring functions are learned as global bilinear argument matrices.
In order to oversee large numbers of latent state without overfitting, the Embedded Latent CRF enforces an additional restriction that the marking function $ \psi_{zz}$ possess a gloomy - rank social organization: that is, $ \psi_{zz}(z_i, z_j)=z_i^\top U V^\top z_j$, where $ U$ and $ V$ are tight-fitting rectangular matrices and $ z_i, z_j$ are represented by one - blistering vectors.
While inference in this model is tractable using tree belief propagation even when learning $ \psi_{zy}$, the deterministic factors make it especially simple to implement.
Computing the quantities involved in can be carried out efficiently with dynamic scheduling use the forward algorithm, as in HMMs. To see this, notice that to calculate the numerator $ \sum_{{\mathbf{z } } } \exp \left (\mathcal{E}({\mathbf{y } }, { \mathbf{z}}| { \mathbf{x } }) \right)$, given an output label $ { \mathbf{y}}$, we can fold the local scores $ \psi_{zf}$ and $ \psi_{zy}$ into one sexual conquest $ \psi_{zfy}(f_t, z_t; y_t)$, and summing the resulting energy corresponds precisely to the forward algorithm in a CRF with $ M$ states. The partition function can also be computed by dynamic scheduling: $ $ \begin{aligned }
Z & = \sum_{{\mathbf{y } } } \sum_{{\mathbf{z } } } \exp \left (\mathcal{E}({\mathbf{y } }, { \mathbf{z}}| { \mathbf{x } }) \right) \\
& = \sum_{{\mathbf{z } } } \sum_{{\mathbf{y } } } \prod_t \exp \left ( \psi | labfl{eq:py}\end{aligned}$$ $Z = \sum_{{\mxthbf{y}}} \sum_{{\mathby{z}}} \exp \neft( \mzthcal{E}({\mxthbf{y}}, {\mathbf{z}}| {\mathbf{x}}) \right)$ iw the partition function.
In ghe case lf our Enbedved Latent CRF model, as lu the PDCRY nodel, latent sjates are dederministicallf oaxtitioned to correspond to output vajues. Thst is, the number of kwtenf states is a multiple of the numbsr of oltput values, and $\lsi_{zy}=0$ for pairs in the parhitilning and $-\infty$ otjerwise. The othqe scoring fuvctions art nearned as global bilinear parameter matrkces.
Iu order to nabagf large numbxrs of latent statcx withmut ovetfitting, the Ekbevded Latent CRF enforces en additional restristion thad che scoring function $\psu_{zz}$ pmssevs a oow-fani xtducturf: tiat is, $\psi_{za}(z_i,z_j)=z_i^\top Y V^\top z_j$, where $U$ amd $N$ are skinny dectandujar matrices and $z_i,z_j$ are represented bj ons-hot vectors.
While inferwnce in this model is tractablq using tree belief propagation even when learninc $\psi_{vy}$, tht eeterounlstic factors make it especially simple to im[mekekt.
Computing the zuantities ongokded in can be carrizs kut efficiently wihh dynaiic peogrammind usong the forward algorithm, aw in HMMs. To wee this, note that to compute the numetator $\sum_{{\mathbf{z}}} \exp \leyt( \matgcal{E}({\mathbf{j}}, {\mathbf{z}}| {\oathbf{x}}) \right)$, gixen at output label ${\mathbf{y}}$, we san fold vhe lpcal scures $\psi_{zf}$ and $\psi_{zy}$ into one score $\psi_{zfy}(f_t, z_t; y_j)$, and vumming thf resulting energy corresponds xeactly to the fmrwdrd algoxithm ln a CRF with $M$ states. The pattition fbnctiov can also be com'uted by dynwmic programmhjg: $$\begin{aligied}
Z &= \sum_{{\iathvf{y}}} \wum_{{\mathcw{z}}} \exp \left( \mayhcal{E}({\matkyf{y}}, {\mathbd{z}}| {\mathbf{x}}) \right) \\
&= \xum_{{\ozthbf{z}}} \sum_{{\mathby{v}}} \krid_t \exp \left( \psk | label{eq:py}\end{aligned}$$ $Z = \sum_{{\mathbf{y}}} \sum_{{\mathbf{z}}} \exp \left( {\mathbf{x}}) is the function. In the CRF as in the model, latent states deterministically partitioned to correspond to output That is, the number of latent states is a multiple of the number output values, and $\psi_{zy}=0$ for pairs in the partitioning and $-\infty$ otherwise. The scoring are as bilinear parameter matrices. In order to manage large numbers of latent states without overfitting, the Embedded CRF enforces an additional restriction that the scoring $\psi_{zz}$ possess a low-rank that is, $\psi_{zz}(z_i,z_j)=z_i^\top U V^\top where and $V$ skinny matrices $z_i,z_j$ are represented one-hot vectors. While inference in this model is tractable using tree belief propagation even when learning $\psi_{zy}$, deterministic factors especially simple implement. the involved in can out efficiently with dynamic programming using as in HMMs. To see this, note that compute the $\sum_{{\mathbf{z}}} \exp \left( \mathcal{E}({\mathbf{y}}, {\mathbf{z}}| {\mathbf{x}}) given an output label ${\mathbf{y}}$, we can fold local scores $\psi_{zf}$ and $\psi_{zy}$ into one score $\psi_{zfy}(f_t, z_t; y_t)$, and summing the resulting exactly to the forward in a CRF $M$ The function also be by dynamic programming: $$\begin{aligned} Z &= \sum_{{\mathbf{y}}} \sum_{{\mathbf{z}}} \exp \left( \mathcal{E}({\mathbf{y}}, {\mathbf{x}}) \right) \\ &= \sum_{{\mathbf{z}}} \sum_{{\mathbf{y}}} \prod_t \exp \left( \psi | label{eq:py}\end{aligned}$$ $Z = \sum_{{\maThbf{y}}} \sum_{{\maThbf{z}}} \Exp \LefT( \mAthcAl{E}({\mAthbf{y}}, {\mathbf{z}}| {\mAThbf{X}}) \right)$ is the partition fuNctioN.
IN The cASe Of our embeddeD laTENt CrF MoDel, As IN tHe LDCrF mOdel, latEnt states aRe dEtErministicalLY pArtitioned To cOrrespond to oUtpUt valuEs. thaT Is, the NumBer of Latent STates iS a multiplE oF The numBEr of outPUT vAlueS, and $\psi_{zy}=0$ for pairs IN tHE partitioning aNd $-\inftY$ oTHeRWIse. the Other scoriNg FunctIOns are lEArNED As gLObal bilinear pArameter matRIceS.
In ordEr To mANage laRge nuMbERs oF latent statEs wiThout overFittinG, The EmbeDDed LateNt CRF eNfoRceS an aDDiTiOnaL rEStrICtIon THat The scoriNg FuNctioN $\psi_{ZZ}$ POSsesS a lOw-raNk strUcture: that is, $\pSi_{zZ}(z_i,z_J)=Z_i^\tOp U V^\tOp z_j$, wHere $u$ aNd $V$ arE skinnY rectAnGular matrices anD $z_i,z_J$ are repreSenTeD by OnE-hot vECtors.
WHilE inFerence In this mODel Is TRACtAble using tree belieF pROPaGation evEn when LEaRnINg $\psi_{zy}$, tHe DetErmiNIStic fActoRS mAke it espEciallY SiMpLe to impLeMent.
CoMpUtiNg tHe quaNTitiEs invoLved in caN be caRRied out efficieNTly with dynamiC PrOGRaMMing UsiNg the forwarD algORithM, as iN hMms. TO See thIs, notE tHAt TO compute the numeratoR $\sUm_{{\mathBf{z}}} \exP \left( \mathcal{E}({\Mathbf{y}}, {\matHBF{Z}}| {\mathbf{x}}) \RighT)$, GiVEn an output labeL ${\mathBf{y}}$, we can foLD the locaL scorEs $\psi_{zf}$ aNd $\psi_{zy}$ inTO One score $\Psi_{Zfy}(F_t, z_T; y_t)$, AND sUmming the resuLTIng eNeRgy corrEspOnds exaCtlY to The ForWaRd algoritHm in a CRF WiTh $m$ sTaTes. the paRTition fuNcTioN cAn aLso be COmputeD by dyNamiC pRoGRamMing: $$\begIN{aLIGned}
z &= \sUm_{{\MathBf{y}}} \SuM_{{\mathBf{z}}} \eXP \leFt( \mathcAl{E}({\mathbf{Y}}, {\maTHbf{z}}| {\MaThBf{x}}) \righT) \\
&= \sum_{{\mathbf{z}}} \suM_{{\mAthbf{y}}} \prod_T \eXp \lEft( \psi | label{eq:py}\end{aligned}$ $ $Z = \su m_{{\ mat hbf {y }}}\sum _{{\mathbf{z}} } \ex p \left( \mathcal{E}({ \math bf { y}}, {\ mathb f{z}}|{ \m a t hbf {x }} ) \ ri g ht )$ is th e parti tion funct ion .
In the case of our Embed ded Latent CRFmod el, as i n t h e LDC RFmodel , late n t stat es are de te r minist i cally p a r ti tion ed to correspondt oo utput values.That i s, th e num ber of latent s tates is a mu l ti p l e of the number of output val u es, and $ \p si_ { zy}=0$ forpa i rsin the part itio ning and$-\inf t y$ othe r wise. T he oth ersco ring fu nc tio ns are le arn e d a s global b il inear par a m e t er m atr ices .
In order to man age lar g e n umber s oflate nt stat es wit houtov erfitting, theEmbe dded Late ntCR F e nf orces an add iti ona l restr ictiont hat t h e sc oring function $\p si _ { zz }$ posse ss a l o w- ra n k struct ur e:that i s, $\ psi_ { zz }(z_i,z_ j)=z_i ^ \t op U V^\t op z_j$, w her e $ U$ an d $V$ are s kinny re ctang u lar matrices a n d $z_i,z_j$ a r er e pr e sent edby one-hotvect o rs.
Whi l einf e rence in t hi s m o del is tractable us in g tree beli ef propagatio n even whe n l earning$\ps i _{ z y}$, the deter minis tic factor s make it espe cially s imple toi m plement.
C omp uti ngt h equantities in v o lved i n can b e c arriedout ef fic ien tl y with dy namic pr og ra mm in g u singt he forwa rd al go rit hm, a s in HM Ms. T o se eth i s,note th a tt o com pu te the nu me rator $\s u m_{ {\mathb f{z}}} \e xp\ left (\m athcal{ E}({\mathbf{y }} , {\mathbf {z }}| {\mat h b f{x}}) \ right)$, given an outpu t label${\ mathb f{y} }$, we ca n f old th e l o cal sc ores $ \psi_ {z f}$ a nd $\ p s i_ {zy }$ into ones c ore $\ps i_ {zfy }(f_t,z_t; y_t)$, and su m min g the resulti ngener g y c orr e sp o nds e x act l y to the forward algorithm i n a CRF with$ M$st ates. T he part ition functio n can als o be comp ut ed b y dyn amic progr amming:$$\begin{ a ligne d }Z &=\su m_{{\m at hbf {y}}} \sum_ { {\m athbf {z}}}\e xp \le ft( \ ma thcal{E} ({\mathbf{y}}, {\mathbf {z}}|{\mat hbf {x}}) \ri ght ) \\
&= \sum_ {{\m athbf{z}}} \s um_ {{\ma thb f {y}}} \pr o d_ t \ exp \ left ( \psi | label{eq:py}\end{aligned}$$ $Z_= \sum_{{\mathbf{y}}}_\sum_{{\mathbf{z}}} \exp \left( \mathcal{E}({\mathbf{y}},_{\mathbf{z}}| {\mathbf{x}})_\right)$_is the_partition_function.
In the case_of our Embedded_Latent CRF model, as_in the LDCRF_model,_latent states are deterministically partitioned to correspond to output values. That is, the number_of_latent states_is_a_multiple of the number of_output values, and $\psi_{zy}=0$ for_pairs in_the partitioning and $-\infty$ otherwise. The other scoring_functions_are learned as_global bilinear parameter matrices.
In order to manage large numbers_of latent states without overfitting, the_Embedded Latent CRF_enforces_an_additional restriction that the_scoring function $\psi_{zz}$ possess a low-rank_structure: that is, $\psi_{zz}(z_i,z_j)=z_i^\top U V^\top_z_j$, where $U$ and $V$ are skinny_rectangular matrices and $z_i,z_j$ are represented_by one-hot vectors.
While inference in_this model_is tractable using tree belief_propagation even when_learning $\psi_{zy}$,_the deterministic factors_make it especially simple to implement.
Computing_the quantities involved_in can be carried out efficiently_with_dynamic programming using_the_forward_algorithm, as_in HMMs. To_see_this, note_that_to compute the numerator $\sum_{{\mathbf{z}}} \exp_\left(_\mathcal{E}({\mathbf{y}}, {\mathbf{z}}| {\mathbf{x}}) \right)$, given an output_label ${\mathbf{y}}$, we can_fold_the local scores $\psi_{zf}$_and $\psi_{zy}$ into one score_$\psi_{zfy}(f_t, z_t; y_t)$, and summing the_resulting energy_corresponds exactly_to the forward algorithm in a CRF with $M$ states. The_partition function can also be computed_by dynamic programming: $$\begin{aligned}
Z_&= \sum_{{\mathbf{y}}}_\sum_{{\mathbf{z}}}_\exp \left( \mathcal{E}({\mathbf{y}},_{\mathbf{z}}|_{\mathbf{x}}) \right)_\\
&= \sum_{{\mathbf{z}}} \sum_{{\mathbf{y}}} \prod_t \exp \left(_ \psi |
V/\partial z\right)=
\left( \partial V^* /\partial z^* \right)$, we have from Eq. (\[A1\]) for the quadratic variables $(\delta z)^2$ and $|\delta z|^2$ the following equations of motion $$\begin{aligned}
\label{A2}
i \frac{d}{d\tau}\left(\delta z\right)^2 &=&
-i\Gamma \left(\delta z\right)^2 +
2\frac{\partial V}{\partial z} \left(\delta z\right)^2 +
2\frac{\partial V}{\partial z^*} |\delta z|^2,\nonumber \\
%
i \frac{d}{d\tau} |\delta z|^2 &=&
-i\Gamma |\delta z|^2
-\frac{\partial V^*}{\partial z} \left(\delta z\right)^2 +
\frac{\partial V}{\partial z^*} \left(\delta z^*\right)^2.\end{aligned}$$ It is easy to see that Eqs. (\[A2\]) are equivalent to Eqs. (\[matrix\]) with $\epsilon=0$, if one makes substitutions $|\delta z|^2\rightarrow B$ and $(\delta z)^2\rightarrow C$.
It should be noticed that yet exists one difference between the linearization of the classical motion equations and the equations for quantum cumulants (\[9b\]), (\[9c\]): It is impossible to get the initial conditions (\[initial\_cond\]) for $C$ and $B$ from only initial conditions for the linearized classical equations of motion (see also discussion of this problem in [@9; @9']).
{#appendix2}
In this Appendix, we present the derivation of the time scale for the validity of the $1/N$-method for lossless ($\Gamma=0$) oscillators. Start with the case of Kerr nonlinearity $l=1$ and then generalize obtained results for arbitrary $l$. From Eqs. (\[z\_cl\]) and (\[B\_and\_C\]) for $\Gamma=0$ and $l=1$, we get $$\label{B1}
z_{cl}(\tau)=z_0\exp(-i\Omega\tau), \quad
\Omega\equiv\bar{\Delta}+|z_0|^2,$$ $$\label{B2}
C(\tau)=-\tau z_0^2 \left[ |z_ | V/\partial z\right)=
\left (\partial V^ * /\partial z^ * \right)$, we have from Eq. (\[A1\ ]) for the quadratic variables $ (\delta z)^2 $ and $ |\delta z|^2 $ the following equations of apparent motion $ $ \begin{aligned }
\label{A2 }
i \frac{d}{d\tau}\left(\delta z\right)^2 & = &
-i\Gamma \left(\delta z\right)^2 +
2\frac{\partial V}{\partial z } \left(\delta z\right)^2 +
2\frac{\partial V}{\partial z^ * } |\delta z|^2,\nonumber \\
%
i \frac{d}{d\tau } |\delta z|^2 & = &
-i\Gamma |\delta z|^2
-\frac{\partial V^*}{\partial omega } \left(\delta z\right)^2 +
\frac{\partial V}{\partial z^ * } \left(\delta z^*\right)^2.\end{aligned}$$ It is easy to see that Eqs. (\[A2\ ]) are equivalent to Eqs. (\[matrix\ ]) with $ \epsilon=0 $, if one make substitution $ |\delta z|^2\rightarrow B$ and $ (\delta z)^2\rightarrow C$.
It should be noticed that yet exist one dispute between the linearization of the classical motion equations and the equations for quantum cumulants (\[9b\ ]), (\[9c\ ] ): It is impossible to catch the initial conditions (\[initial\_cond\ ]) for $ C$ and $ B$ from alone initial condition for the linearized classical equations of motion (see besides discussion of this problem in [ @9; @9' ]).
{ # appendix2 }
In this Appendix, we present the derivation of the fourth dimension scale for the validity of the $ 1 / N$-method for lossless ($ \Gamma=0 $) oscillator. Start with the sheath of Kerr nonlinearity $ l=1 $ and then generalize receive results for arbitrary $ l$. From Eqs. (\[z\_cl\ ]) and (\[B\_and\_C\ ]) for $ \Gamma=0 $ and $ l=1 $, we get $ $ \label{B1 }
z_{cl}(\tau)=z_0\exp(-i\Omega\tau), \quad
\Omega\equiv\bar{\Delta}+|z_0|^2,$$ $ $ \label{B2 }
C(\tau)=-\tau z_0 ^ 2 \left [ |z _ | V/\pwrtial z\right)=
\left( \partiau V^* /\partial z^* \rntht)$, we have rrom Eq. (\[X1\]) for the quadratic variabled $(\eelta z)^2$ and $|\delta z|^2$ the foluowing eqlations od mouion $$\begin{aligned}
\label{A2}
i \nxac{d}{d\fwu}\leyt(\velta z\right)^2 &=&
-i\Gsmma \left(\dalta z\right)^2 +
2\frdc{\oaxtial V}{\partial z} \left(\delta z\right)^2 +
2\frwc{\partisl V}{\partial z^*} |\dejta e|^2,\nogumbsg \\
%
l \frac{d}{d\tau} |\delta z|^2 &=&
-i\Gamma |\delta z|^2
-\frac{\pertial V^*}{\partial z} \left(\delta z\right)^2 +
\frac{\pagtiap V}{\partial z^*} \left(\dflta z^*\right)^2.\ghd{ajugned}$$ It is dasy to set chat Eqs. (\[A2\]) zre equivalent to Eqs. (\[matrix\]) wigh $\epxilon=0$, if obe mangs substitutmons $|\dvlta z|^2\rightarvpw B$ atd $(\delts z)^2\rightarrow G$.
It sioule be noticed that yet exists one differgnce betwean the linearizatiob if thg clavsicxo mutikn esuatiojs end the equztions for wuantum cumulants (\[9b\]), (\[9c\]): Py is impossigle to gqt the initial conditions (\[initial\_cond\]) fmr $D$ and $B$ from only initiql conditions for the linearizqd classical equations of motion (see also discusshon oh ghiw irobuwm in [@9; @9']).
{#appendix2}
In this Appendix, we present the setination of the timc scale for the vakifiyi of the $1/N$-methud for loasless ($\Gamma=0$) oscilpators. Ftart with the casr of Kerr nonlinearity $l=1$ ane then generclize obtained resultd for arbitxary $l$. From Eqs. (\[z\_cl\]) and (\[B\_and\_C\]) for $\Yamma=0$ znd $l=1$, we geh $$\label{B1}
z_{du}(\tau)=z_0\exp(-i\Omega\taj), \qlad
\Okega\equiv\bar{\Delta}+|z_0|^2,$$ $$\label{B2}
C(\eau)=-\tau z_0^2 \oeft[ |z_ | V/\partial z\right)= \left( \partial V^* /\partial z^* have Eq. (\[A1\]) the quadratic variables the equations of motion \label{A2} i \frac{d}{d\tau}\left(\delta &=& -i\Gamma \left(\delta z\right)^2 + 2\frac{\partial z} \left(\delta z\right)^2 + 2\frac{\partial V}{\partial z^*} |\delta z|^2,\nonumber \\ % i \frac{d}{d\tau} z|^2 &=& -i\Gamma |\delta z|^2 -\frac{\partial V^*}{\partial z} \left(\delta z\right)^2 + \frac{\partial V}{\partial \left(\delta It easy see that Eqs. (\[A2\]) are equivalent to Eqs. (\[matrix\]) with $\epsilon=0$, if one makes substitutions $|\delta B$ and $(\delta z)^2\rightarrow C$. It should be that yet exists one between the linearization of the motion and the for cumulants (\[9c\]): It is to get the initial conditions (\[initial\_cond\]) for $C$ and $B$ from only initial conditions for the linearized equations of also discussion this in @9']). {#appendix2} In we present the derivation of the the validity of the $1/N$-method for lossless ($\Gamma=0$) Start with case of Kerr nonlinearity $l=1$ and generalize obtained results for arbitrary $l$. From Eqs. and (\[B\_and\_C\]) for $\Gamma=0$ and $l=1$, we get $$\label{B1} z_{cl}(\tau)=z_0\exp(-i\Omega\tau), \quad \Omega\equiv\bar{\Delta}+|z_0|^2,$$ $$\label{B2} C(\tau)=-\tau z_0^2 | V/\partial z\right)=
\left( \partial v^* /\partial z^* \rIght)$, wE haVe fRoM Eq. (\[A1\]) For tHe quadratic varIAbleS $(\delta z)^2$ and $|\delta z|^2$ the folLowinG eQUatiONs Of motIon $$\begiN{AlIGNed}
\LaBeL{A2}
i \FrAC{d}{D\tau}\lEft(\Delta z\rIght)^2 &=&
-i\Gamma \LefT(\dElta z\right)^2 +
2\frAC{\pArtial V}{\parTiaL z} \left(\delta z\RigHt)^2 +
2\frac{\PaRtiAL V}{\parTiaL z^*} |\delTa z|^2,\nonUMber \\
%
i \fRac{d}{d\tau} |\dElTA z|^2 &=&
-i\GamMA |\delta z|^2
-\FRAc{\PartIal V^*}{\partial z} \left(\dELtA Z\right)^2 +
\frac{\partIal V}{\paRtIAl Z^*} \LEft(\DelTa z^*\right)^2.\enD{aLigneD}$$ it is easY To SEE ThaT eqs. (\[A2\]) are equivaLent to Eqs. (\[maTRix\]) With $\epSiLon=0$, IF one maKes suBsTItuTions $|\delta z|^2\RighTarrow B$ anD $(\delta Z)^2\RightarROw C$.
It shOuld be NotIceD thaT YeT eXisTs ONe dIFfEreNCe bEtween thE lInEarizAtioN OF THe clAssIcal MotioN equations and The EquaTIonS for qUantuM cumUlAnts (\[9b\]), (\[9C\]): It is iMpossIbLe to get the initiAl coNditions (\[iNitIaL\_coNd\]) For $C$ aND $B$ from OnlY inItial coNditionS For ThE LINeArized classical equAtIONs Of motion (See alsO DiScUSsion of tHiS prObleM IN [@9; @9']).
{#appeNdix2}
iN tHis AppenDix, we pREsEnT the derIvAtion oF tHe tIme Scale FOr thE validIty of the $1/n$-methOD for lossless ($\GaMMa=0$) oscillators. sTaRT WiTH the CasE of Kerr nonlIneaRIty $l=1$ And tHEn GenERalizE obtaInED rESults for arbitrary $l$. FRoM Eqs. (\[z\_cL\]) and (\[B\_And\_C\]) for $\Gamma=0$ aNd $l=1$, we get $$\laBEL{b1}
z_{cl}(\tau)=z_0\Exp(-i\oMeGA\tau), \quad
\Omega\eQuiv\bAr{\Delta}+|z_0|^2,$$ $$\laBEl{B2}
C(\tau)=-\tAu z_0^2 \leFt[ |z_ | V/\partial z\right)=
\lef t( \partia l V^* /\ par ti al z ^* \ right)$, we ha v e fr om Eq. (\[A1\]) for th e qua dr a ticv ar iable s $(\de l ta z )^2 $an d $ |\ d el ta z| ^2$ the fo llowing eq uat io ns of motion $$ \begin{ali gne d}
\label{A2 }
i \frac {d }{d \ tau}\ lef t(\de lta z\ r ight)^ 2 &=&
-i\ Ga m ma \le f t(\delt a z\ righ t)^2 +
2\frac{\pa r ti a l V}{\partialz} \le ft ( \d e l taz\r ight)^2 +2\ frac{ \ partial V} { \ p art i al z^*} |\del ta z|^2,\no n umb er \\%i \ f rac{d} {d\ta u} |\d elta z|^2 & =&
- i\Gamma | \delta z|^2
-\ f rac{\pa rtialV^* }{\ part i al z } \ le f t(\ d el taz \ri ght)^2 +
\ fr ac{\p arti a l V }{\p art ialz^*}\left(\deltaz^* \rig h t)^ 2.\en d{ali gned }$ $ Itis eas y tose e that Eqs. (\[ A2\] ) are equ iva le ntto Eqs. (\[mat rix \]) with $ \epsilo n =0$ ,i f on e makes substituti on s $| \delta z |^2\ri g ht ar r ow B$ an d$(\ delt a z)^2\ righ t ar row C$.
It sh o ul dbe noti ce d that y etexi sts o n e di fferen ce betwe en th e linearization of the classi c al m ot i on e qua tions and t he e q uati onsf or qu a ntumcumul an t s( \[9b\]), (\[9c\]):It is im possi ble to get th e initialc o n ditions(\[i n it i al\_cond\]) fo r $C$ and $B$ f r om onlyiniti al condi tions for t he linea riz edcla ssi c a lequations ofm o tion ( see als o d iscussi onofthi s p ro blem in [ @9; @9'] ).
{ #a ppe ndix2 }
In thi sApp en dix , wep resent thederi va ti o n o f the t i me s cale f or the va li dityof t h e $ 1/N$-me thod forlos s less ( $\ Gamma=0 $) oscillator s. Start wit hthe caseo f Kerr no nlinearity $l=1$ and th e n gener ali ze ob tain ed result s f or arb itr a ry $l$ . From Eqs. ( \[z \ _ cl\]) a nd (\ [B \_and\_C\] ) for $\Ga mm a=0$ and $l =1$, we get $$\lab e l{B 1}
z_{cl}(\ta u)= z_0\ e x p( -i\ O me g a\t au ) , \ q u ad
\Omega\equiv \bar{\Delt a} + |z _0|^2,$$ $ $ \la be l{B2}
C (\tau)= -\tau z_0^2 \ left[ |z_ | V/\partial_z\right)=
\left( \partial_V^* /\partial z^* \right)$,_we have_from_Eq. (\[A1\])_for_the quadratic variables_$(\delta z)^2$ and_$|\delta z|^2$ the following_equations of motion_$$\begin{aligned}
\label{A2}
i_\frac{d}{d\tau}\left(\delta z\right)^2 &=&
-i\Gamma \left(\delta z\right)^2 +
2\frac{\partial V}{\partial z} \left(\delta z\right)^2 +
2\frac{\partial V}{\partial z^*} |\delta_z|^2,\nonumber_\\
%
i \frac{d}{d\tau}_|\delta_z|^2_&=&
-i\Gamma |\delta z|^2
-\frac{\partial V^*}{\partial z}_\left(\delta z\right)^2 +
\frac{\partial V}{\partial z^*}_\left(\delta z^*\right)^2.\end{aligned}$$_It is easy to see that Eqs. (\[A2\])_are_equivalent to Eqs._(\[matrix\]) with $\epsilon=0$, if one makes substitutions $|\delta z|^2\rightarrow_B$ and $(\delta z)^2\rightarrow C$.
It should_be noticed that_yet_exists_one difference between the_linearization of the classical motion equations_and the equations for quantum cumulants_(\[9b\]), (\[9c\]): It is impossible to get_the initial conditions (\[initial\_cond\]) for $C$_and $B$ from only initial_conditions for_the linearized classical equations of_motion (see also_discussion of_this problem in_[@9; @9']).
{#appendix2}
In this Appendix, we_present the derivation_of the time scale for the_validity_of the $1/N$-method_for_lossless_($\Gamma=0$) oscillators._Start with the_case_of Kerr_nonlinearity_$l=1$ and then generalize obtained results_for_arbitrary $l$. From Eqs. (\[z\_cl\]) and (\[B\_and\_C\])_for $\Gamma=0$ and $l=1$,_we_get $$\label{B1}
z_{cl}(\tau)=z_0\exp(-i\Omega\tau), \quad
\Omega\equiv\bar{\Delta}+|z_0|^2,$$ $$\label{B2}
C(\tau)=-\tau_z_0^2 \left[ |z_ |
eless such that ${\bf u}_1(x)={\bf m}(A,e,f)$ and ${\bf u}_\kappa(x)={\bf m}(A,E,F)$.
We shall exhibit a matrix ${\bf m}(A,0,0)={\bf n}(A,0,0)$ whose projective order is a multiple of $4$ and it is bigger than $8$. This will prove the statement for both $t=1,\kappa$.
Let ${{\mathbb F}}^\times_{q^2}=\langle \xi \rangle$ and consider the matrix $z={\operatorname{diag}}(\xi^{\frac{q-1}{2}},-\xi^{\frac{1-q}{2}},-\xi^{\frac{1-q}{2}},\xi^{
\frac{q-1}{2}}
)$ in $\SL_4({{\mathbb F}}_{q^2})$. The order of $z$ is $2({q+1})$ and $z^{\frac{q+1}{2}}={\operatorname{diag}}(\omega,\omega^{-1},\omega^{-1},\omega)$ for $\omega$ a primitive fourth root of $1$, hence the projective order of $z$ is $q+1$.
We claim that $z$ is ${\operatorname{PGL}}_4(\overline{\Fq})$-conjugate to $x={\bf m}(\left(\begin{smallmatrix}\Tr(z)/2&1\\
1&0\\
\end{smallmatrix}\right),0,0)$ and that $\Tr(z)\in{{\mathbb F}}_q^\times$. If this is the case, ${\bf u}_1(x)=x^2$ and its projective order is $\frac{q+1}{2}$ which is even as $q\equiv 3(4)$ and bigger than 4 since $q\geq11$.
The claim is proved if the following conditions hold, namely $$\begin{aligned}
&\det z=1; &&
\Tr z=2(\xi^{\frac{q-1}{2}}-\xi^{\frac{1-q}{2}})\in{{\mathbb F}}_q;
&& \xi^{\frac{q-1}2}\neq -\xi^{\frac{1-q}2}.\end{aligned}$$ Indeed, in this case, the matrix ${\bf m}(\left(\begin{smallmatrix}\Tr(z)/2&1\\
1&0\\
\end{ | eless such that $ { \bf u}_1(x)={\bf m}(A, e, f)$ and $ { \bf u}_\kappa(x)={\bf m}(A, E, F)$.
We shall exhibit a matrix $ { \bf m}(A,0,0)={\bf n}(A,0,0)$ whose projective order is a multiple of $ 4 $ and it is bigger than $ 8$. This will rise the instruction for both $ t=1,\kappa$.
Let $ { { \mathbb F}}^\times_{q^2}=\langle \xi \rangle$ and consider the matrix $ z={\operatorname{diag}}(\xi^{\frac{q-1}{2}},-\xi^{\frac{1 - q}{2}},-\xi^{\frac{1 - q}{2}},\xi^ {
\frac{q-1}{2 } }
) $ in $ \SL_4({{\mathbb F}}_{q^2})$. The order of $ z$ is $ 2({q+1})$ and $ z^{\frac{q+1}{2}}={\operatorname{diag}}(\omega,\omega^{-1},\omega^{-1},\omega)$ for $ \omega$ a archaic fourth root of $ 1 $, hence the projective club of $ z$ is $ q+1$.
We claim that $ z$ is $ { \operatorname{PGL}}_4(\overline{\Fq})$-conjugate to $ x={\bf m}(\left(\begin{smallmatrix}\Tr(z)/2&1\\
1&0\\
\end{smallmatrix}\right),0,0)$ and that $ \Tr(z)\in{{\mathbb F}}_q^\times$. If this is the character, $ { \bf u}_1(x)=x^2 $ and its projective order is $ \frac{q+1}{2}$ which is even as $ q\equiv 3(4)$ and adult than 4 since $ q\geq11$.
The claim is proved if the keep up conditions declare, namely $ $ \begin{aligned }
& \det z=1; & &
\Tr z=2(\xi^{\frac{q-1}{2}}-\xi^{\frac{1 - q}{2}})\in{{\mathbb F}}_q;
& & \xi^{\frac{q-1}2}\neq -\xi^{\frac{1 - q}2}.\end{aligned}$$ Indeed, in this case, the matrix $ { \bf m}(\left(\begin{smallmatrix}\Tr(z)/2&1\\
1&0\\
\end { | eleds such that ${\bf u}_1(x)={\bf m}(A,e,n)$ and ${\bf u}_\kappa(x)={\yd m}(A,E,F)$.
Xe shalm exhibig a matrix ${\bf m}(A,0,0)={\bf n}(A,0,0)$ whose 'rojwctivt order is a multipue of $4$ anf it is viggtr than $8$. This will prove bke stzbemenc hor both $t=1,\kappa$.
Ket ${{\mathbb F}}^\times_{q^2}=\langle \xk \xangle$ and consider the matrix $z={\operwtornamr{dlag}}(\xi^{\frac{q-1}{2}},-\xi^{\frwc{1-q}{2}},-\xp^{\fwac{1-q}{2}},\sp^{
\fvac{q-1}{2}}
)$ in $\SL_4({{\mathbb F}}_{q^2})$. The order of $z$ is $2({q+1})$ and $z^{\frac{q+1}{2}}={\oprratorname{diag}}(\omega,\omega^{-1},\omfga^{-1},\olega)$ for $\omega$ a pgimitive foodth eoot of $1$, henze the projective ordet of $z$ is $q+1$.
We claim that $z$ is ${\opdratoxname{PGL}}_4(\ovetljnf{\Xq})$-conjugate to $x={\ff m}(\left(\begik{xmallmdtrix}\Tr(a)/2&1\\
1&0\\
\end{smallmatriw}\righv),0,0)$ ane that $\Tr(z)\in{{\mathbb F}}_q^\vimes$. If this is the case, ${\bf g}_1(x)=r^2$ and its projective ireer iv $\frdc{q+1}{2}$ dyicf ia xveh as $q\fqumv 3(4)$ and bigfer than 4 sunce $q\geq11$.
The claim os itoved if the follorigg conditions hold, namely $$\begin{aligned}
&\dtt z=1; &&
\Fr z=2(\xi^{\frac{q-1}{2}}-\xi^{\frac{1-q}{2}})\in{{\matybb F}}_q;
&& \xi^{\frac{q-1}2}\neq -\xi^{\ftac{1-q}2}.\end{alidned}$$ Indeed, in this case, the matrix ${\bf m}(\left(\begin{vmalljxtrnw}\Br(z)/2&1\\
1&0\\
\eve{ | eless such that ${\bf u}_1(x)={\bf m}(A,e,f)$ and m}(A,E,F)$. shall exhibit matrix ${\bf m}(A,0,0)={\bf a of $4$ and is bigger than This will prove the statement for $t=1,\kappa$. Let ${{\mathbb F}}^\times_{q^2}=\langle \xi \rangle$ and consider the matrix $z={\operatorname{diag}}(\xi^{\frac{q-1}{2}},-\xi^{\frac{1-q}{2}},-\xi^{\frac{1-q}{2}},\xi^{ \frac{q-1}{2}} )$ $\SL_4({{\mathbb F}}_{q^2})$. The order of $z$ is $2({q+1})$ and $z^{\frac{q+1}{2}}={\operatorname{diag}}(\omega,\omega^{-1},\omega^{-1},\omega)$ for $\omega$ a fourth of hence projective order of $z$ is $q+1$. We claim that $z$ is ${\operatorname{PGL}}_4(\overline{\Fq})$-conjugate to $x={\bf m}(\left(\begin{smallmatrix}\Tr(z)/2&1\\ 1&0\\ and that $\Tr(z)\in{{\mathbb F}}_q^\times$. If this is the ${\bf u}_1(x)=x^2$ and its order is $\frac{q+1}{2}$ which is as 3(4)$ and than since The claim is if the following conditions hold, namely $$\begin{aligned} &\det z=1; && \Tr z=2(\xi^{\frac{q-1}{2}}-\xi^{\frac{1-q}{2}})\in{{\mathbb F}}_q; && \xi^{\frac{q-1}2}\neq -\xi^{\frac{1-q}2}.\end{aligned}$$ Indeed, this case, ${\bf m}(\left(\begin{smallmatrix}\Tr(z)/2&1\\ \end{ | eless such that ${\bf u}_1(x)={\bf m}(A,e,f)$ anD ${\bf u}_\kappa(x)={\Bf m}(A,E,f)$.
We ShaLl ExhiBit a Matrix ${\bf m}(A,0,0)={\bf n}(A,0,0)$ WHose Projective order is a multIple oF $4$ aND it iS BiGger tHan $8$. This WIlL PRovE tHe StaTeMEnT for bOth $T=1,\kappa$.
LEt ${{\mathbb F}}^\tImeS_{q^2}=\Langle \xi \rangLE$ aNd consider The Matrix $z={\operaTorName{diAg}}(\Xi^{\fRAc{q-1}{2}},-\xi^{\FraC{1-q}{2}},-\xi^{\fRac{1-q}{2}},\xi^{
\FRac{q-1}{2}}
)$ in $\sL_4({{\mathbb F}}_{Q^2})$. THE order OF $z$ is $2({q+1})$ anD $Z^{\FrAc{q+1}{2}}={\oPeratorname{diag}}(\omEGa,\OMega^{-1},\omega^{-1},\omega)$ For $\omeGa$ A PrIMItiVe fOurth root oF $1$, hEnce tHE projecTIvE ORDer OF $z$ is $q+1$.
We claim tHat $z$ is ${\operaTOrnAme{PGL}}_4(\OvErlINe{\Fq})$-coNjugaTe TO $x={\bF m}(\left(\begin{SmalLmatrix}\Tr(Z)/2&1\\
1&0\\
\end{smALlmatriX}\Right),0,0)$ anD that $\TR(z)\iN{{\maThbb f}}_Q^\tImEs$. IF tHIs iS ThE caSE, ${\bf U}_1(x)=x^2$ and itS pRoJectiVe orDER IS $\fraC{q+1}{2}$ wHich Is eveN as $q\equiv 3(4)$ and bIggEr thAN 4 siNce $q\gEq11$.
The ClaiM iS provEd if thE follOwIng conditions hoLd, naMely $$\begin{AliGnEd}
&\dEt Z=1; &&
\Tr z=2(\xI^{\Frac{q-1}{2}}-\xI^{\frAc{1-q}{2}})\In{{\mathbB F}}_q;
&& \xi^{\frAC{q-1}2}\nEq -\XI^{\FRaC{1-q}2}.\end{aligned}$$ Indeed, In THIs Case, the mAtrix ${\bF M}(\lEfT(\Begin{smaLlMatRix}\TR(Z)/2&1\\
1&0\\
\End{ | eless such that ${\bf u}_1 (x)={\bf m }(A,e ,f) $ a nd ${\ bf u }_\kappa(x)={\ b f m} (A,E,F)$.
We shall ex hibit a matr i x${\bf m}(A,0 , 0) = { \bf n }( A,0 ,0 ) $whose pr ojectiv e order is amu ltiple of $4 $ a nd it is b igg er than $8$. Th is wil lpro v e the st ateme nt for both $ t=1,\kapp a$ .
Let$ {{\math b b F }}^\ times_{q^2}=\lang l e\ xi \rangle$ an d cons id e rt h e m atr ix $z={\op er atorn a me{diag } }( \ x i ^{\ f rac{q-1}{2}}, -\xi^{\frac { 1-q }{2}}, -\ xi^ { \frac{ 1-q}{ 2} } ,\x i^{
\frac{q -1}{ 2}}
)$ in $\SL_ 4 ({{\mat h bb F}}_ {q^2}) $.The ord e rof $z $i s $ 2 ({ q+1 } )$and $z^{ \f ra c{q+1 }{2} } = { \ oper ato rnam e{dia g}}(\omega,\o meg a^{- 1 },\ omega ^{-1} ,\om eg a)$ f or $\o mega$ a primitive four th r oot of $1 $,he nce t he pr o jectiv e o rde r of $z $ is $q + 1$.
W e cl aim that $z$ is ${ \o p e ra torname{ PGL}}_ 4 (\ ov e rline{\F q} )$- conj u g ate t o $x = {\ bf m}(\l eft(\b e gi n{ smallma tr ix}\Tr (z )/2 &1\ \
1&0 \ \
\e nd{sma llmatrix }\rig h t),0,0)$ and t h at $\Tr(z)\in { {\ m a th b b F} }_q ^\times$. I f th i s is the ca se, ${\bf u}_1 (x ) =x ^ 2$ and its projecti ve order is $ \frac{q+1}{2} $ which is e v en as $q \equ i v3 (4)$ and bigge r tha n 4 since$ q\geq11$ .
Th e claimis proved i f the fo llo win g c ond i t io ns hold, name l y $$\ be gin{ali gne d}
&\de t z =1; &&
\T rz=2(\xi^{ \frac{q- 1} {2 }} -\ xi^ {\fra c {1-q}{2} }) \in {{ \ma thbbF }}_q;&& \x i^{\ fr ac { q-1 }2}\neq -\ x i ^{\f ra c{ 1-q} 2}. \e nd{al igne d }$$ Indeed , in this ca s e, t he m atrix $ {\bf m}(\left (\ begin{smal lm atr ix}\Tr ( z )/2&1\\1&0\\
\end{ | eless such_that ${\bf_u}_1(x)={\bf m}(A,e,f)$ and ${\bf_u}_\kappa(x)={\bf m}(A,E,F)$.
We_shall_exhibit a_matrix_${\bf m}(A,0,0)={\bf n}(A,0,0)$_whose projective order_is a multiple of_$4$ and it_is_bigger than $8$. This will prove the statement for both $t=1,\kappa$.
Let ${{\mathbb F}}^\times_{q^2}=\langle \xi_\rangle$_and consider_the_matrix_$z={\operatorname{diag}}(\xi^{\frac{q-1}{2}},-\xi^{\frac{1-q}{2}},-\xi^{\frac{1-q}{2}},\xi^{
\frac{q-1}{2}}
)$ in $\SL_4({{\mathbb F}}_{q^2})$. The_order of $z$ is $2({q+1})$_and $z^{\frac{q+1}{2}}={\operatorname{diag}}(\omega,\omega^{-1},\omega^{-1},\omega)$_for $\omega$ a primitive fourth root of $1$,_hence_the projective order_of $z$ is $q+1$.
We claim that $z$ is ${\operatorname{PGL}}_4(\overline{\Fq})$-conjugate_to $x={\bf m}(\left(\begin{smallmatrix}\Tr(z)/2&1\\
1&0\\
\end{smallmatrix}\right),0,0)$ and that $\Tr(z)\in{{\mathbb_F}}_q^\times$. If this_is_the_case, ${\bf u}_1(x)=x^2$ and_its projective order is $\frac{q+1}{2}$ which_is even as $q\equiv 3(4)$ and_bigger than 4 since $q\geq11$.
The claim is_proved if the following conditions hold,_namely $$\begin{aligned}
&\det z=1; &&
\Tr z=2(\xi^{\frac{q-1}{2}}-\xi^{\frac{1-q}{2}})\in{{\mathbb_F}}_q;
&& \xi^{\frac{q-1}2}\neq_-\xi^{\frac{1-q}2}.\end{aligned}$$ Indeed, in this case,_the matrix ${\bf_m}(\left(\begin{smallmatrix}\Tr(z)/2&1\\
1&0\\
\end{ |
contribution. If such an enhancement were observed in the experiment, it would provide an important and unambiguous determination of the weak $\pi NN$ coupling constant $h_{\pi}^{1}$. However, a recent schematic calculation of $A_{\gamma}$ by Khriplovich and Korkin [@Khriplovich:2000mb], partly suggested by one of the present author, showed critical contradiction to Oka’s result, with a huge suppression of $A_{\gamma}$ at the energies $\omega_{\gamma}\gtrsim3$ MeV.
On the experimental side, a measurement of the asymmetry $A_{\gamma}$ in $\vec{\gamma}+d\rightarrow n+p$ was considered in the 80’s by E. D. Earle *et al.* [@Earle:1981; @Earle:1988fc] but no sensitive result was reported. However, due to advances in experimental techniques and instrumentation, the measurement of $A_{\gamma}$ becomes more feasible nowadays and several groups at JLab [@jlab-lett00], IASA (Athens), LEGS (BNL), TUNL, and SPring-8 show interest in such a measurement. It is therefore important to understand and improve previous estimates.
In this work, we carefully re-examine the $\vec{\gamma}+d\rightarrow n+p$ process with two main purposes:
1. Determine how the enhancement of the $h_{\pi}^{1}$ contribution in Oka’s results will change when the calculation is completed with missing parity-admixed components in the final state, in particular in the $^{3}P_{1}$ channel. The role of this last one was revealed by the schematic estimate of Ref. [@Khriplovich:2000mb].
2. Determine the uncertainty of Khriplovich and Korkin’s calculation in which very simple wave functions are used.
It is straightforward to deal with the point 1. In Ref. [@Khriplovich:2000mb], a nice and simple argument about the cancellation of the $h_{\pi}^{1}$ contribution from the final $^{3}P_{0}$, $^{3}P_{1}$, and $^{3}P_{2}$ states along with their parity-admixed partners was given. However, the argument assumed the absence of tensor as well as spin-orbit forces, which are important components of the $NN$ interaction. In order to address these two points (missing components | contribution. If such an enhancement were observed in the experiment, it would provide an important and unambiguous decision of the fallible $ \pi NN$ coupling constant $ h_{\pi}^{1}$. However, a recent conventional calculation of $ A_{\gamma}$ by Khriplovich and Korkin [ @Khriplovich:2000 mb ], partly suggested by one of the present author, picture critical contradiction to Oka ’s result, with a huge suppression of $ A_{\gamma}$ at the energy $ \omega_{\gamma}\gtrsim3 $ MeV.
On the experimental side, a measurement of the asymmetry $ A_{\gamma}$ in $ \vec{\gamma}+d\rightarrow n+p$ was considered in the 80 ’s by E. D. Earle * et al. * [ @Earle:1981; @Earle:1988fc ] but no sensible result was reported. However, due to advances in experimental techniques and instrumentality, the measurement of $ A_{\gamma}$ becomes more feasible nowadays and several groups at JLab [ @jlab - lett00 ], IASA (Athens), LEGS (BNL), TUNL, and SPring-8 show sake in such a measurement. It is therefore important to understand and better previous estimates.
In this work, we carefully re - analyze the $ \vec{\gamma}+d\rightarrow n+p$ process with two main purposes:
1. Determine how the enhancement of the $ h_{\pi}^{1}$ contribution in Oka ’s results will transfer when the calculation is completed with missing parity - admixed components in the final state, in particular in the $ ^{3}P_{1}$ channel. The role of this last one was revealed by the schematic estimate of Ref. [ @Khriplovich:2000 mb ].
2. Determine the uncertainty of Khriplovich and Korkin ’s calculation in which very simple wave function are used.
It is straightforward to share with the point 1. In Ref. [ @Khriplovich:2000 mb ], a nice and simple argument about the cancellation of the $ h_{\pi}^{1}$ contribution from the concluding $ ^{3}P_{0}$, $ ^{3}P_{1}$, and $ ^{3}P_{2}$ states along with their parity bit - admixed partners was given. However, the controversy assumed the absence of tensor as well as spin - orbit forces, which are important components of the $ NN$ interaction. In order to address these two points (missing component | cojtribution. If such an enmancement were oywerved in ths experioent, it would provide an implrrant qnd unambiguous determknation ov the weqk $\pm NN$ coupling coiatant $h_{\in}^{1}$. Howsyer, a cecent schematig calculatimn of $A_{\gamma}$ bf Yhxiplovich and Korkin [@Khriplovich:2000mb], pwrtly sighested by one jf tnq prssent author, showed critical contrzdictioi to Oka’s resuly, with a huge suppression lf $A_{\hamma}$ at the energles $\omega_{\ganma}\geesim3$ MeV.
On tfe experimtncal side, a jeasurement of the asymmetry $A_{\gxmma}$ nn $\vec{\gamma}+e\rughhdrrow n+p$ waw confidered in tmv 80’s by A. D. Earke *et al.* [@Earle:1981; @Eacle:1988fx] but no sensitive revult was reported. Rowever, dge to advances in ezpwrimettal tecfbiqjes aid jnstrulenvation, the jeasurement of $A_{\gamma}$ becomes kowv feasible nosadays agd several groups at JLab [@jlab-lett00], IASA (Atgens), LEGS (BNL), TUNL, and WPring-8 show interest ln such a measurement. It is therefore important to understdnd aid imkrive pfwvlous estimates.
In this work, we carefully re-exaijnt tme $\vec{\gamma}+d\righbarrow n+p$ process eihh jwo main purpores:
1. Dzfedmine how the enhajcement of tye $h_{\pi}^{1}$ cogtrinution in Oka’s results will change when rhe calculation is completed cith mossinb parity-admixed componeuts in the final dtate, in lxrticular in the $^{3}P_{1}$ bhantel. The role of this last jne was rxvealzd by thd scnematis estimate of Rcx. [@Khriplovich:2000mb].
2. Dftermnne tve uncertalnty of Khriplovich and Korkin’s calculation im fhibh very snmple eave functionf are used.
It ix straiyhtfordard to dezl with the point 1. In Ref. [@Khripnlvich:2000mb], a nire and siiple argyment acuut the cancelkation of the $h_{\pi}^{1}$ cintribution from tme fivzl $^{3}P_{0}$, $^{3}P_{1}$, and $^{3}P_{2}$ stcues along with theor oarytj-adkivad partners fas eivdm. Howdver, tht avguoent assumed the absence of fensor as well as xpln-orbit firces, whych are imporyant components of the $IN$ intxractipn. Yn order to address these two loints (midsikg components | contribution. If such an enhancement were observed experiment, would provide important and unambiguous NN$ constant $h_{\pi}^{1}$. However, recent schematic calculation $A_{\gamma}$ by Khriplovich and Korkin [@Khriplovich:2000mb], suggested by one of the present author, showed critical contradiction to Oka’s result, a huge suppression of $A_{\gamma}$ at the energies $\omega_{\gamma}\gtrsim3$ MeV. On the experimental a of asymmetry in $\vec{\gamma}+d\rightarrow n+p$ was considered in the 80’s by E. D. Earle *et al.* [@Earle:1981; @Earle:1988fc] no sensitive result was reported. However, due to in experimental techniques and the measurement of $A_{\gamma}$ becomes feasible and several at [@jlab-lett00], (Athens), LEGS (BNL), and SPring-8 show interest in such a measurement. It is therefore important to understand and improve previous In this carefully re-examine $\vec{\gamma}+d\rightarrow process two main purposes: how the enhancement of the $h_{\pi}^{1}$ results will change when the calculation is completed missing parity-admixed in the final state, in particular the $^{3}P_{1}$ channel. The role of this last was revealed by the schematic estimate of Ref. [@Khriplovich:2000mb]. 2. Determine the uncertainty of Khriplovich calculation in which very wave functions are It straightforward deal the point In Ref. [@Khriplovich:2000mb], a nice and simple argument about the cancellation the $h_{\pi}^{1}$ contribution from the final $^{3}P_{0}$, $^{3}P_{1}$, and $^{3}P_{2}$ with parity-admixed partners was However, the argument assumed absence tensor as well as which important $NN$ In to address these two (missing components | contribution. If such an enhanCement were ObserVed In tHe ExpeRimeNt, it would proviDE an iMportant and unambiguous DeterMiNAtioN Of The weAk $\pi NN$ cOUpLINg cOnStAnt $H_{\pI}^{1}$. hoWever, A reCent schEmatic calcUlaTiOn of $A_{\gamma}$ by kHrIplovich anD KoRkin [@KhriplovIch:2000Mb], partLy SugGEsted By oNe of tHe presENt authOr, showed cRiTIcal coNTradictION tO Oka’S result, with a huge sUPpREssion of $A_{\gamma}$ At the eNeRGiES $\OmeGa_{\gAmma}\gtrsim3$ mev.
On thE ExperimENtAL SIde, A Measurement of The asymmetrY $a_{\gaMma}$ in $\vEc{\GamMA}+d\righTarroW n+P$ Was Considered iN the 80’S by E. D. EarlE *et al.* [@EARle:1981; @EarlE:1988Fc] but no SensitIve ResUlt wAS rEpOrtEd. hOweVEr, Due TO adVances in ExPeRimenTal tECHNIqueS anD insTrumeNtation, the meaSurEmenT Of $A_{\Gamma}$ BecomEs moRe FeasiBle nowAdays AnD several groups aT JLaB [@jlab-lett00], iASa (ATheNs), lEGS (Bnl), TUNL, aNd SpriNg-8 show iNterest IN suCh A MEAsUrement. It is therefoRe IMPoRtant to uNderstANd AnD Improve pReVioUs esTIMates.
in thIS wOrk, we carEfully RE-eXaMine the $\VeC{\gamma}+D\rIghTarRow n+p$ PRoceSs with Two main pUrposES:
1. Determine how tHE enhancement oF ThE $H_{\Pi}^{1}$ COntrIbuTion in Oka’s rEsulTS wilL chaNGe WheN The caLculaTiON iS Completed with missinG pArity-aDmixeD components in The final stATE, In particUlar IN tHE $^{3}P_{1}$ channel. The roLe of tHis last one WAs revealEd by tHe schemaTic estimaTE Of Ref. [@KhrIplOviCh:2000mB].
2. DeTERmIne the uncertaINTy of khRiploviCh aNd KorkiN’s cAlcUlaTioN iN which verY simple wAvE fUnCtIonS are uSEd.
It is stRaIghTfOrwArd to DEal witH the pOint 1. in reF. [@khrIplovicH:2000Mb], A NIce aNd SiMple ArgUmEnt abOut tHE caNcellatIon of the $h_{\Pi}^{1}$ cONtriBuTiOn from tHe final $^{3}P_{0}$, $^{3}P_{1}$, and $^{3}P_{2}$ StAtes along wItH thEir parITY-admixed Partners was given. However, THe argumEnt AssumEd thE absence oF teNsor as WelL As spin-Orbit fOrces, WhIch ARE impoRTAnT coMpOnents of thE $nn$ inTeracTiOn. In Order to Address these two poiNTs (mIssing componeNts | contribution. If such anenhancemen t wer e o bse rv ed i n th e experiment,i t wo uld provide an importa nt an du namb i gu ous d etermin a ti o n of t he we ak $\ pi NN $ c oupling constant$h_ {\ pi}^{1}$. Ho w ev er, a rece ntschematic ca lcu lation o f $ A _{\ga mma }$ by Khrip l ovichand Korki n[ @Khrip l ovich:2 0 0 0m b],partly suggestedb yo ne of the pres ent au th o r, s how edcritical c on tradi c tion to Ok a ’ s re s ult, with a h uge suppres s ion of $A _{ \ga m ma}$ a t the e n erg ies $\omega _{\g amma}\gtr sim3$M eV.
On the exp erimen tal si de,a m ea sur em e nto fthe asy mmetry $ A_ {\ gamma }$ i n $ \ vec{ \ga mma} +d\ri ghtarrow n+p$ wa s co n sid eredin th e 80 ’s by E . D. E arle*e t al.* [@Earle: 1981 ; @Earle: 198 8f c]bu t nos ensiti veres ult was report e d.Ho w e v er , due to advancesin e xp erimenta l tech n iq ue s and ins tr ume ntat i o n, th e me a su rement o f $A_{ \ ga mm a}$ bec om es mor efea sib le no w aday s andseveralgroup s at JLab [@jla b -lett00], IAS A ( A t he n s),LEG S (BNL), TU NL,a nd S Prin g -8 sh o w int erest i n s u ch a measurement. I tis the refor e important t o understa n d and impr ovep re v ious estimates .
In this work , we care fully re-exam ine the $ \ v ec{\gamm a}+ d\r igh tar r o wn+p$ processw i th t wo main p urp oses:
1. De ter min ehow the e nhanceme nt o fth e $ h_{\p i }^{1}$ c on tri bu tio n inO ka’s r esult s wi ll c h ang e whent he c alcu la ti on i s c om plete d wi t h m issingparity-ad mix e d co mp on ents in the final st at e, in part ic ula r in t h e $^{3}P_ {1}$ channel. The roleo f thislas t one was revealed by the s che m atic e stimat e ofRe f.[ @ Khrip l o vi ch: 20 00mb].
2. Det ermin etheuncerta inty of Khriplovic h an d Korkin’s ca lcu lati o n i n w h ic h ve ry sim p l e wave function s are used .It is straig h tfo rw ard todeal wi th th e point1. In Ref . [@Khrip lo vich : 2 000 mb], a nic e and si mple argu m ent a b ou t the ca ncella ti onof th e $h_{ \ pi} ^{1}$ contr ib utionfromth e final$^{3}P_{0}$, $^{3}P_{1} $, and $^{3 }P_ {2}$ stat esa lon g with th eirparity-adm ixe d p artne rsw as gi ven. Ho wev e r, th e ar g ument ass u me d t h e a bsence of t e n s oras we lla s spin -orb it forces, whicha re important c ompo n e nts of the$N N$ interaction . I no r der to a dd ress thesetwo poin ts (miss ing co mponen ts | contribution._If such_an enhancement were observed_in the_experiment,_it would_provide_an important and_unambiguous determination of_the weak $\pi NN$_coupling constant $h_{\pi}^{1}$._However,_a recent schematic calculation of $A_{\gamma}$ by Khriplovich and Korkin [@Khriplovich:2000mb], partly suggested by_one_of the_present_author,_showed critical contradiction to Oka’s_result, with a huge suppression_of $A_{\gamma}$_at the energies $\omega_{\gamma}\gtrsim3$ MeV.
On the experimental side,_a_measurement of the_asymmetry $A_{\gamma}$ in $\vec{\gamma}+d\rightarrow n+p$ was considered in the_80’s by E. D. Earle *et_al.* [@Earle:1981; @Earle:1988fc]_but_no_sensitive result was reported._However, due to advances in experimental_techniques and instrumentation, the measurement of_$A_{\gamma}$ becomes more feasible nowadays and several_groups at JLab [@jlab-lett00], IASA (Athens),_LEGS (BNL), TUNL, and SPring-8_show interest_in such a measurement. It_is therefore important_to understand_and improve previous_estimates.
In this work, we carefully re-examine_the $\vec{\gamma}+d\rightarrow n+p$_process with two main purposes:
1. _Determine_how the enhancement_of_the_$h_{\pi}^{1}$ contribution_in Oka’s results_will_change when_the_calculation is completed with missing parity-admixed_components_in the final state, in particular in_the $^{3}P_{1}$ channel. The_role_of this last one_was revealed by the schematic_estimate of Ref. [@Khriplovich:2000mb].
2. Determine_the uncertainty_of Khriplovich_and Korkin’s calculation in which very simple wave functions are used.
It_is straightforward to deal with the_point 1. In Ref._[@Khriplovich:2000mb], a_nice_and simple argument_about_the cancellation_of the $h_{\pi}^{1}$ contribution from the final_$^{3}P_{0}$, $^{3}P_{1}$,_and $^{3}P_{2}$ states along with their_parity-admixed partners was given._However,_the argument assumed the absence of_tensor as well as spin-orbit forces,_which are important components of_the_$NN$_interaction. In order to address_these two points (missing components |
has allowed us to confirm the peculiar red colors of five sources in the sample. We can impose modest upper limits of 0.9% and 1.8% on the linear polarization degree for seven targets with a confidence of 99%. Only one source, 2MASSJ02411151$-$0326587 (L0), appears to be strongly polarized ($P \sim 3\,\%$) in the $J$-band with a significance level of $P/\sigma_P \sim 10$. The likely origin of its linearly polarized light and rather red infrared colors may reside in a surrounding disk with an asymmetric distribution of grains. Given its proximity (66$\pm$8 pc), this object becomes an excellent target for the direct detection of the disk.'
author:
- 'M$.$ R$.$ Zapatero Osorio'
- 'V$.$ J$.$ S$.$ Béjar'
- 'B$.$ Goldman'
- 'J$.$ A$.$ Caballero'
- 'R$.$ Rebolo, J$.$ A$.$ Acosta-Pulido, A$.$ Manchado, and K. Peña Ramírez'
title: 'Near-infrared Linear Polarization of Ultracool Dwarfs'
---
Introduction
============
There is growing evidence that dwarfs of spectral types L and T show a large spread of near- and mid-infrared colors that cannot be explained by simple models (e.g., Marley et al$.$ [@marley10]). It is believed that gravity, metallicity, and distribution of dust clouds play a critical role in defining the atmospheric properties of L- and T-type sources (see Kirkpatrick [@kirk05] for a review). Various groups have monitored photometrically a number of L and T objects aimed at characterizing cloud patchiness in brown dwarfs (e.g., Gelino et al$.$ [@gelino02]; Koen [@koen04a]; Koen et al$.$ [@koen04b]; Artigau et al. [@artigau06]). Polarimetric observations at optical wavelengths have also been attempted to confirm the presence of dusty clouds in the objects’ atmospheres (Ménard et al$.$ [@menard02]: Zapatero Osorio et al$.$ [@osorio06]; Goldman et al$.$ [@goldman09]; Tata et al$.$ [@tata09]). The main results | has allowed us to confirm the peculiar red colors of five source in the sample distribution. We can impose modest upper limit of 0.9% and 1.8% on the analogue polarization degree for seven targets with a confidence of 99% . merely one source, 2MASSJ02411151$-$0326587 (L0), appears to be powerfully polarize ($ P \sim 3\,\%$) in the $ J$-band with a meaning level of $ P/\sigma_P \sim 10$. The likely origin of its linearly polarized inner light and rather red infrared colors may reside in a surrounding phonograph record with an asymmetric distribution of grains. Given its proximity (66$\pm$8 personal computer), this object becomes an excellent prey for the direct signal detection of the disk.'
generator:
-' M$.$ R$.$ Zapatero Osorio'
-' V$.$ J$.$ S$.$ Béjar'
-' B$.$ Goldman'
-' J$.$ A$ .$ Caballero'
-' R$.$ Rebolo, J$.$ A$ .$ Acosta - Pulido, A$ .$ Manchado, and K. Peña Ramírez'
title:' Near - infrared Linear Polarization of Ultracool Dwarfs'
---
Introduction
= = = = = = = = = = = =
There be growing evidence that dwarfs of spectral types L and T show a big spread of near- and mid - infrared colors that cannot be explained by simple models (e.g., Marley et al$.$ [ @marley10 ]). It is believed that gravity, metallicity, and distribution of dust clouds meet a critical role in defining the atmospheric place of L- and T - character sources (see Kirkpatrick [ @kirk05 ] for a review). Various groups have monitored photometrically a number of L and T object aimed at characterizing cloud patchiness in brown dwarfs (e.g., Gelino et al$.$ [ @gelino02 ]; Koen [ @koen04a ]; Koen et al$.$ [ @koen04b ]; Artigau et al. [ @artigau06 ]). Polarimetric observation at optical wavelengths have also been attempted to confirm the presence of dusty cloud in the objects ’ atmospheres (Ménard et al$.$ [ @menard02 ]: Zapatero Osorio et al$.$ [ @osorio06 ]; Goldman et al$.$ [ @goldman09 ]; Tata et al$.$ [ @tata09 ]). The main results | had allowed us to confirm uhe peculiar red eilors mf fivs sourcer in the sample. We can imposx moeest ypper limits of 0.9% and 1.8% on the linewr polaruzatmon degree for sxben tarnzts wjbh a eoifidence of 99%. Onky one sousce, 2MASSJ02411151$-$0326587 (L0), ap[exrd to be strongly polarized ($P \sim 3\,\%$) ig the $J$-najd with a signyficsgce mvvtl of $P/\sigma_P \sim 10$. The likely orifin of pts linearly polatized light and rather red infgared colors may rfside in a wurrjynding disk dith an asjkmetric diatribution of grains. Given its oroxikity (66$\pm$8 pc), tyis mbject becones ag excellent bsrget xor the direct detectlon oh thw disk.'
author:
- 'M$.$ R$.$ Zapavero Osorio'
- 'V$.$ J$.$ S$.$ Béjwr'
- 'B$.$ Goldkau'
- 'J$.$ A$.$ Caballero'
- 'R$.$ Rebili, J$.$ A$.$ Acovta-Pjoidu, A$.$ Mencgado, ajd I. Peña Ramírsz'
title: 'Neae-infrared Linear Pokawpaation of Ulfracooj Qwarfs'
---
Introduction
============
There is growing evidtnce fhat dwarfs of spectral types L and T show a large spwead of near- and mid-infrared colors that cannot ba expmxintd by sknppe models (e.g., Marley et al$.$ [@marley10]). It is believqs uhan gravity, metalligity, and distributooj ps dust clouds play c cditical role in devining jhe atnospheric prolerties of L- and T-type sourxes (see Kirkiatruck [@kirk05] for a revnew). Various yroups have monitored photometricauly z number of L and T kcjects aimed at zhagactarizing cloud patchiness ig brown dxarfs (e.g., Gelkno gt al$.$ [@gqlino02]; Koen [@koen04a]; Koen et al$.$ [@koen04b]; Wrtigcu et al. [@artigak06]). Polarimetric observations at optical wavelemgdhs have alfo becn attempted to confirm the ptesence oy dusth clouds ih the oujects’ atmos[heres (Ménard aj al$.$ [@menard02]: Zepatero Oforii et al$.$ [@osofko06]; Goldman et sl$.$ [@goldmau09]; Tatq et al$.$ [@tata09]). The mein teaults | has allowed us to confirm the peculiar of sources in sample. We can 0.9% 1.8% on the polarization degree for targets with a confidence of 99%. one source, 2MASSJ02411151$-$0326587 (L0), appears to be strongly polarized ($P \sim 3\,\%$) in $J$-band with a significance level of $P/\sigma_P \sim 10$. The likely origin of linearly light rather infrared colors may reside in a surrounding disk with an asymmetric distribution of grains. Given its (66$\pm$8 pc), this object becomes an excellent target the direct detection of disk.' author: - 'M$.$ R$.$ Osorio' 'V$.$ J$.$ Béjar' 'B$.$ - 'J$.$ A$.$ - 'R$.$ Rebolo, J$.$ A$.$ Acosta-Pulido, A$.$ Manchado, and K. Peña Ramírez' title: 'Near-infrared Linear Polarization of Dwarfs' --- There is evidence dwarfs spectral types L show a large spread of near- that cannot be explained by simple models (e.g., et al$.$ It is believed that gravity, metallicity, distribution of dust clouds play a critical role defining the atmospheric properties of L- and T-type sources (see Kirkpatrick [@kirk05] for a review). have monitored photometrically a of L and objects at cloud in brown (e.g., Gelino et al$.$ [@gelino02]; Koen [@koen04a]; Koen et al$.$ [@koen04b]; et al. [@artigau06]). Polarimetric observations at optical wavelengths have also to the presence of clouds in the objects’ (Ménard al$.$ [@menard02]: Zapatero Osorio [@osorio06]; et et [@tata09]). main results | has allowed us to confirm the pEculiar red ColorS of FivE sOurcEs in The sample. We can IMposE modest upper limits of 0.9% anD 1.8% on thE lINear POlArizaTion degREe FOR seVeN tArgEtS WiTh a coNfiDence of 99%. only one souRce, 2mAsSJ02411151$-$0326587 (L0), appears tO Be Strongly poLarIzed ($P \sim 3\,\%$) in thE $J$-bAnd witH a SigNIficaNce Level Of $P/\sigMA_P \sim 10$. THe likely oRiGIn of itS LinearlY POlArizEd light and rather rED iNFrared colors maY residE iN A sURRouNdiNg disk with An AsymmETric disTRiBUTIon OF grains. Given iTs proximity (66$\PM$8 pc), This obJeCt bEComes aN exceLlENt tArget for the DireCt detectiOn of thE Disk.'
autHOr:
- 'M$.$ R$.$ ZapAtero OSorIo'
- 'V$.$ j$.$ S$.$ BéJAr'
- 'b$.$ GOldMaN'
- 'j$.$ A$.$ CABaLleRO'
- 'R$.$ REbolo, J$.$ A$.$ ACoStA-PuliDo, A$.$ MANCHAdo, aNd K. peña ramírEz'
title: 'Near-inFraRed LINeaR PolaRizatIon oF ULtracOol DwaRfs'
---
InTrOduction
============
There is GrowIng evidenCe tHaT dwArFs of sPEctral TypEs L And T shoW a large SPreAd OF NEaR- and mid-infrared colOrS THaT cannot bE explaINeD bY Simple moDeLs (e.G., MarLEY et al$.$ [@MarlEY10]). IT is belieVed thaT GrAvIty, metaLlIcity, aNd DisTriButioN Of duSt clouDs play a cRiticAL role in defininG The atmospheriC PrOPErTIes oF L- aNd T-type sourCes (sEE KirKpatRIcK [@kiRK05] for a RevieW). VARiOUs groups have monitorEd PhotomEtricAlly a number of l and T objecTS AImed at chAracTErIZing cloud patchIness In brown dwaRFs (e.g., GeliNo et aL$.$ [@gelino02]; KOen [@koen04a]; KOEN et al$.$ [@koeN04b]; ARtiGau Et aL. [@ARtIgau06]). PolarimetRIC obsErVations At oPtical wAveLenGthS haVe Also been aTtempted To CoNfIrM thE presENce of dusTy CloUdS in The obJEcts’ atMosphEres (méNaRD et Al$.$ [@menarD02]: zaPATero osOrIo et Al$.$ [@oSoRio06]; GoLdmaN Et aL$.$ [@goldmaN09]; Tata et al$.$ [@TatA09]). the mAiN rEsults | has allowed us to confirm the pecul iar r edcol or s of fiv e sources in t h e sa mple. We can impose mo destup p er l i mi ts of 0.9% a n d1 . 8%on t heli n ea r pol ari zationdegree for se ve n targets wi t ha confiden ceof 99%. Only on e sour ce , 2 M ASSJ0 241 1151$ -$0326 5 87 (L0 ), appear st o be s t ronglyp o la rize d ($P \sim 3\,\%$ ) i n the $J$-bandwith a s i gn i f ica nce level of$P /\sig m a_P \si m 1 0 $ . Th e likely origi n of its li n ear ly pol ar ize d light andra t her red infrar ed c olors may resid e in a s u rroundi ng dis k w ith ana sy mm etr ic dis t ri but i onof grain s. G ivenitsp r o x imit y ( 66$\ pm$8pc), this obj ect bec o mes an e xcell entta rgetfor th e dir ec t detection ofthedisk.'
au tho r:
-'M $.$ R $ .$ Zap ate roOsorio'
- 'V$. $ J$ .$ S $ .$ Béjar'
- 'B$.$ Go ld m a n'
- 'J$.$ A$.$C ab al l ero'
- ' R$ .$Rebo l o , J$. $ A$ . $Acosta-P ulido, A$ .$ Mancha do , andK. Pe ñaRamír e z'
t itle:'Near-in frare d Linear Polari z ation of Ultr a co o l D w arfs '
- --
Introdu ctio n
=== ==== = == ==There is g ro w in g evidence that dwar fs of sp ectra l types L and T show al a r ge sprea d of ne a r- and mid-inf rared colors th a t cannot be e xplained by simpl e models ( e.g .,Mar ley e tal$.$ [@marle y 1 0]). I t is be lie ved tha t g rav ity , m et allicity, and dis tr ib ut io n o f dus t cloudspl ayacri tical role i n def inin gth e at mospher i cp r oper ti es ofL-an d T-t ypes our ces (se e Kirkpat ric k [@k ir k0 5] fora review). Va ri ous groups h ave monit o r ed photo metrically a number ofL and Tobj ectsaime d at char act erizin g c l oud pa tchine ss in b row n dwarf s (e .g. ,Gelino eta l $.$ [@ge li no02 ]; Koen [@koen04a]; Koene t a l$.$ [@koen04 b]; Art i g au et al . [@ ar t iga u 0 6]). Polarimetr ic observa ti o ns at optica l wa ve lengths have a lso b e en atte mpted toconfirm t he pre s e nce of dustyclouds i n the obj e cts’a tm osphe res (Ména rd et al$. $ [@me n ard 02]:Zapate ro Osori o etal $.$ [@os orio06]; Goldman et al$ .$ [@g oldma n09 ]; Tata e t a l $.$ [@tata09 ]).The main r esu lts | has_allowed us_to confirm the peculiar_red colors_of_five sources_in_the sample. We_can impose modest_upper limits of 0.9% and_1.8% on the linear_polarization_degree for seven targets with a confidence of 99%. Only one source, 2MASSJ02411151$-$0326587 (L0),_appears_to be_strongly_polarized_($P \sim 3\,\%$) in the_$J$-band with a significance level_of $P/\sigma_P_\sim 10$. The likely origin of its linearly_polarized_light and rather_red infrared colors may reside in a surrounding disk_with an asymmetric distribution of grains._Given its proximity_(66$\pm$8_pc),_this object becomes an_excellent target for the direct detection_of the disk.'
author:
- 'M$.$ R$.$ Zapatero_Osorio'
- 'V$.$ J$.$ S$.$ Béjar'
- 'B$.$ Goldman'
-_'J$.$ A$.$ Caballero'
- 'R$.$ Rebolo, J$.$_A$.$ Acosta-Pulido, A$.$ Manchado, and_K. Peña_Ramírez'
title: 'Near-infrared Linear Polarization of_Ultracool Dwarfs'
---
Introduction
============
There is_growing evidence_that dwarfs of_spectral types L and T show_a large spread_of near- and mid-infrared colors that_cannot_be explained by_simple_models_(e.g., Marley_et al$.$ [@marley10])._It_is believed_that_gravity, metallicity, and distribution of dust_clouds_play a critical role in defining the_atmospheric properties of L-_and_T-type sources (see Kirkpatrick_[@kirk05] for a review). Various_groups have monitored photometrically a number_of L_and T_objects aimed at characterizing cloud patchiness in brown dwarfs (e.g., Gelino_et al$.$ [@gelino02]; Koen [@koen04a]; Koen_et al$.$ [@koen04b]; Artigau_et al._[@artigau06])._Polarimetric observations at_optical_wavelengths have_also been attempted to confirm the presence_of dusty_clouds in the objects’ atmospheres (Ménard_et al$.$ [@menard02]: Zapatero_Osorio_et al$.$ [@osorio06]; Goldman et al$.$_[@goldman09]; Tata et al$.$ [@tata09]). The_main results |
,j})-\phi(c_{2,j})\Big)a_ia_jcov_{ij}^k+O(|cov_{ij}^k|^{3/2}).\end{aligned}$$ In the last line, $\big(\phi(c_{1,i})-\phi(c_{2,i})\big)\big(\phi(c_{1,j})-\phi(c_{2,j})\big)a_ia_j$ is bounded by some constant except on a countable collection of measure zero sets. Let $C_i$ be defined as the set $\{z_{t/2}+\eta_i+\mu_i=0\}\cup\{z_{t/2}-\eta_i-\mu_i=0\}$. On the set $C_i^c$, $\big(\phi(c_{1,i})-\phi(c_{2,i})\big)a_i$ converges to zero as $a_i\rightarrow\infty$. Therefore, $\big(\phi(c_{1,i})-\phi(c_{2,i})\big)\big(\phi(c_{1,j})-\phi(c_{2,j})\big)a_ia_j$ is bounded by some constant on $(\bigcup_{i=1}^pC_i)^c$.
By the Cauchy-Schwartz inequality and $(C0)$ in Theorem 1, $p^{-2}\sum_{i,j}|cov_{i,j}^k|=O(p^{-\delta})$. Also we have $|cov_{ij}^k|^{3/2}<|cov_{ij}^k|$. On the set $(\bigcup_{i=1}^pC_i)^c$, we conclude that $${\mbox{Var}}\Big(p^{-1}\sum_{i=1}^pI(P_i\leq t|W_1,\cdots,W_k)\Big)=O_p(p^{-\delta}).$$ Hence by Lemma 1, for fixed $(w_1,\cdots,w_k)^T$, $$\label{g1}
p^{-1}\sum_{i=1}^p\big\{I(P_i\leq t|W_1=w_1,\cdots,W_k=w_k)
-P(P_i\leq t|W_1=w_1,\cdots,W_k=w_k)\big\}\stackrel{p\to | , j})-\phi(c_{2,j})\Big)a_ia_jcov_{ij}^k+O(|cov_{ij}^k|^{3/2}).\end{aligned}$$ In the last line, $ \big(\phi(c_{1,i})-\phi(c_{2,i})\big)\big(\phi(c_{1,j})-\phi(c_{2,j})\big)a_ia_j$ is bounded by some constant except on a countable solicitation of standard zero sets. Let $ C_i$ be define as the fixed $ \{z_{t/2}+\eta_i+\mu_i=0\}\cup\{z_{t/2}-\eta_i-\mu_i=0\}$. On the set $ C_i^c$, $ \big(\phi(c_{1,i})-\phi(c_{2,i})\big)a_i$ converges to zero as $ a_i\rightarrow\infty$. Therefore, $ \big(\phi(c_{1,i})-\phi(c_{2,i})\big)\big(\phi(c_{1,j})-\phi(c_{2,j})\big)a_ia_j$ is bounce by some constant on $ (\bigcup_{i=1}^pC_i)^c$.
By the Cauchy - Schwartz inequality and $ (C0)$ in Theorem 1, $ p^{-2}\sum_{i, j}|cov_{i, j}^k|=O(p^{-\delta})$. besides we have $ |cov_{ij}^k|^{3/2}<|cov_{ij}^k|$. On the set $ (\bigcup_{i=1}^pC_i)^c$, we conclude that $ $ { \mbox{Var}}\Big(p^{-1}\sum_{i=1}^pI(P_i\leq t|W_1,\cdots, W_k)\Big)=O_p(p^{-\delta}).$$ Hence by Lemma 1, for fixed $ (w_1,\cdots, w_k)^T$, $ $ \label{g1 }
p^{-1}\sum_{i=1}^p\big\{I(P_i\leq t|W_1 = w_1,\cdots, W_k = w_k)
-P(P_i\leq t|W_1 = w_1,\cdots, W_k = w_k)\big\}\stackrel{p\to | ,j})-\phl(c_{2,j})\Big)a_ia_jcov_{ij}^k+O(|cov_{ij}^k|^{3/2}).\ekd{aligned}$$ In the last lmne, $\big(\lhi(c_{1,i})-\phi(z_{2,i})\big)\big(\phi(c_{1,j})-\phi(c_{2,j})\big)a_ia_j$ is biundee by some constant excdpt on a bountable colowction of measure dzro ssbs. Lec $R_i$ be defined ax the set $\{s_{t/2}+\eta_i+\mu_i=0\}\cup\{z_{t/2}-\atx_i-\lu_i=0\}$. On the set $C_i^c$, $\big(\phi(c_{1,i})-\phi(c_{2,i})\big)w_i$ convrrhes to zero as $a_i\rpgrtardow\infty$. Therefore, $\big(\phi(c_{1,i})-\phi(c_{2,i})\bif)\big(\phi(b_{1,j})-\phi(c_{2,j})\big)a_ia_j$ is bounded by some constant ln $(\blgcup_{i=1}^pC_i)^c$.
By the Cwuchy-Schwarja igwquality and $(C0)$ in Theogzm 1, $p^{-2}\sum_{i,j}|ckv_{i,j}^k|=O(p^{-\delta})$. Also we have $|cov_{ij}^y|^{3/2}<|cov_{il}^k|$. On the sgc $(\bihwup_{i=1}^pC_i)^c$, we concjude that $${\mbox{Var}}\Big([^{-1}\sum_{i=1}^pI(L_i\leq t|W_1,\cdots,W_l)\Bij)=O_p(p^{-\eelta}).$$ Hence by Lemma 1, for fixed $(w_1,\cdots,w_h)^T$, $$\label{g1}
[^{-1}\sbm_{i=1}^p\big\{I(P_i\leq t|W_1=w_1,\cdotw,W_j=w_k)
-P(P_h\leq t|W_1=w_1,\zeotr,W_k=s_k)\uig\}\atackrfl{p\vo | ,j})-\phi(c_{2,j})\Big)a_ia_jcov_{ij}^k+O(|cov_{ij}^k|^{3/2}).\end{aligned}$$ In the last line, $\big(\phi(c_{1,i})-\phi(c_{2,i})\big)\big(\phi(c_{1,j})-\phi(c_{2,j})\big)a_ia_j$ is some except on countable collection of be as the set On the set $\big(\phi(c_{1,i})-\phi(c_{2,i})\big)a_i$ converges to zero as $a_i\rightarrow\infty$. $\big(\phi(c_{1,i})-\phi(c_{2,i})\big)\big(\phi(c_{1,j})-\phi(c_{2,j})\big)a_ia_j$ is bounded by some constant on $(\bigcup_{i=1}^pC_i)^c$. By the Cauchy-Schwartz inequality and in Theorem 1, $p^{-2}\sum_{i,j}|cov_{i,j}^k|=O(p^{-\delta})$. Also we have $|cov_{ij}^k|^{3/2}<|cov_{ij}^k|$. On the set $(\bigcup_{i=1}^pC_i)^c$, we that t|W_1,\cdots,W_k)\Big)=O_p(p^{-\delta}).$$ by 1, for fixed $(w_1,\cdots,w_k)^T$, $$\label{g1} p^{-1}\sum_{i=1}^p\big\{I(P_i\leq t|W_1=w_1,\cdots,W_k=w_k) -P(P_i\leq t|W_1=w_1,\cdots,W_k=w_k)\big\}\stackrel{p\to | ,j})-\phi(c_{2,j})\Big)a_ia_jcov_{ij}^k+O(|cov_{ij}^K|^{3/2}).\end{aligneD}$$ In thE laSt lInE, $\big(\Phi(c_{1,I})-\phi(c_{2,i})\big)\big(\phI(C_{1,j})-\phI(c_{2,j})\big)a_ia_j$ is bounded by sOme coNsTAnt eXCePt on a CountabLE cOLLecTiOn Of mEaSUrE zero SetS. Let $C_i$ bE defined as The SeT $\{z_{t/2}+\eta_i+\mu_i=0\}\cuP\{Z_{t/2}-\Eta_i-\mu_i=0\}$. On tHe sEt $C_i^c$, $\big(\phi(c_{1,I})-\phI(c_{2,i})\big)A_i$ ConVErges To zEro as $A_i\righTArrow\iNfty$. ThereFoRE, $\big(\phI(C_{1,i})-\phi(c_{2,i})\BIG)\bIg(\phI(c_{1,j})-\phi(c_{2,j})\big)a_ia_j$ is BOuNDed by some constAnt on $(\bIgCUp_{I=1}^Pc_i)^c$.
by tHe Cauchy-ScHwArtz iNEqualitY AnD $(c0)$ IN ThEOrem 1, $p^{-2}\sum_{i,j}|cov_{I,j}^k|=O(p^{-\delta})$. ALSo wE have $|cOv_{Ij}^k|^{3/2}<|COv_{ij}^k|$. ON the sEt $(\BIgcUp_{i=1}^pC_i)^c$, we coNcluDe that $${\mboX{Var}}\BiG(P^{-1}\sum_{i=1}^pI(p_I\leq t|W_1,\cDots,W_k)\big)=o_p(p^{-\DeltA}).$$ heNcE by leMMa 1, fOR fIxeD $(W_1,\cdOts,w_k)^T$, $$\laBeL{g1}
P^{-1}\sum_{i=1}^P\big\{i(p_I\LEq t|W_1=W_1,\cdOts,W_K=w_k)
-P(P_I\leq t|W_1=w_1,\cdots,W_K=w_k)\Big\}\sTAckRel{p\tO | ,j})-\phi(c_{2,j})\Big)a_i a_jcov_{ij }^k+O (|c ov_ {i j}^k |^{3 /2}).\end{alig n ed}$ $ In the last line, $\ big(\ ph i (c_{ 1 ,i })-\p hi(c_{2 , i} ) \ big )\ bi g(\ ph i (c _{1,j })- \phi(c_ {2,j})\big )a_ ia _j$ is bound e dby some co nst ant except o n a count ab lec ollec tio n ofmeasur e zerosets. Let $ C _i$ be defined a stheset $\{z_{t/2}+\e t a_ i +\mu_i=0\}\cup \{z_{t /2 } -\ e t a_i -\m u_i=0\}$.On thes et $C_i ^ c$ , $ \bi g (\phi(c_{1,i} )-\phi(c_{2 , i}) \big)a _i $ c o nverge s toze r o a s $a_i\righ tarr ow\infty$ . Ther e fore, $ \ big(\ph i(c_{1 ,i} )-\ phi( c _{ 2, i}) \b i g)\ b ig (\p h i(c _{1,j})- \p hi (c_{2 ,j}) \ b i g )a_i a_j $ is boun ded by some c ons tant on$(\bi gcup_ {i=1 }^ pC_i) ^c$.
By th eCauchy-Schwartz ine quality a nd$( C0) $in Th e orem 1 , $ p^{ -2}\sum _{i,j}| c ov_ {i , j } ^k |=O(p^{-\delta})$. A l s owe have$|cov_ { ij }^ k |^{3/2}< |c ov_ {ij} ^ k |$. O n th e s et $(\bi gcup_{ i =1 }^ pC_i)^c $, we co nc lud e t hat $ $ {\mb ox{Var }}\Big(p ^{-1} \ sum_{i=1}^pI(P _ i\leq t|W_1,\ c do t s ,W _ k)\B ig) =O_p(p^{-\d elta } ).$$ Hen c ebyL emma1, fo rf ix e d $(w_1,\cdots,w_k) ^T $, $$\ label {g1}
p^{-1}\s um_{i=1}^p \ b i g\{I(P_i \leq t| W _1=w_1,\cdots, W_k=w _k)
-P(P_i \ leq t|W_ 1=w_1 ,\cdots, W_k=w_k)\ b i g\}\stac kre l{p \to | ,j})-\phi(c_{2,j})\Big)a_ia_jcov_{ij}^k+O(|cov_{ij}^k|^{3/2}).\end{aligned}$$ In_the last_line, $\big(\phi(c_{1,i})-\phi(c_{2,i})\big)\big(\phi(c_{1,j})-\phi(c_{2,j})\big)a_ia_j$ is bounded_by some_constant_except on_a_countable collection of_measure zero sets._Let $C_i$ be defined_as the set_$\{z_{t/2}+\eta_i+\mu_i=0\}\cup\{z_{t/2}-\eta_i-\mu_i=0\}$._On the set $C_i^c$, $\big(\phi(c_{1,i})-\phi(c_{2,i})\big)a_i$ converges to zero as $a_i\rightarrow\infty$. Therefore, $\big(\phi(c_{1,i})-\phi(c_{2,i})\big)\big(\phi(c_{1,j})-\phi(c_{2,j})\big)a_ia_j$ is bounded_by_some constant_on_$(\bigcup_{i=1}^pC_i)^c$.
By_the Cauchy-Schwartz inequality and $(C0)$_in Theorem 1, $p^{-2}\sum_{i,j}|cov_{i,j}^k|=O(p^{-\delta})$. Also_we have_$|cov_{ij}^k|^{3/2}<|cov_{ij}^k|$. On the set $(\bigcup_{i=1}^pC_i)^c$, we conclude that_$${\mbox{Var}}\Big(p^{-1}\sum_{i=1}^pI(P_i\leq_t|W_1,\cdots,W_k)\Big)=O_p(p^{-\delta}).$$ Hence by_Lemma 1, for fixed $(w_1,\cdots,w_k)^T$, $$\label{g1}
p^{-1}\sum_{i=1}^p\big\{I(P_i\leq t|W_1=w_1,\cdots,W_k=w_k)
-P(P_i\leq t|W_1=w_1,\cdots,W_k=w_k)\big\}\stackrel{p\to |
.[]{data-label="fig:VoterResults"}](VoterResults.png){width="\columnwidth"}
Agent Based Models on a spatial domain
--------------------------------------
In the previous subsection, we have detailed how to create an ABM on a graph, how to create an associated problem, how to generate its code and how to run it. Simflowny 2 also includes the new family of ABM on a spatial domain, which we validate in this section. As many features are common with the ABM on a graph family, we will directly present an example without detailing the process of building the model and the problem In this example, a two-dimensional flocking model is implemented into Simflowny. We use the vectorial noise model of Gregoire and Chaté [@Gregoire2004]. The model is a variation of the Original Vicsek Model [@Vicsek1995] (OVM), devised to reproduce the collective motion - or flocking - we find in many biological and non-biological systems (see, for instance, references [@Toner1995],[@Gregoire2008], and [@Deutsch2012]and [@Vicsek2012] for reviews). In these systems long-range orientation order emerges after spontaneous symmetry breaking.
In the OVM, point-like agents move synchronously in discrete time steps, with a fixed common speed $v_0$. In 2D the orientation of agent $\alpha$ is an angle $\theta_{\alpha}$. The evolution rule provides the new angles at each time step, based on the angles of the agent’s neighbours (agents within a certain influence radius) in the previous time step. Essentially, the agent tries to align itself with its neighbours. This alignment is perturbed by a white noise.
The Gregoire and Chaté model is based on the OVM, but modifies the manner in which noise is incorporated into the model. They define *vectorial noise* as generated by errors when estimating interactions, in comparison with the *angular noise* in the OVM, related to errors in trying to follow the newly computed direction. Altogether, the update rule for the Gregoire and Chaté model is:
$$\theta^{t+1}_{\alpha} = arg \left[\sum_{\beta \sim \alpha}e^{i \theta^t_{\beta}} + \
\eta n^t_{\alpha} e^{i\xi^t_{\alpha}} \right],$$
where $\xi | .[]{data - label="fig: VoterResults"}](VoterResults.png){width="\columnwidth " }
Agent Based Models on a spatial domain
--------------------------------------
In the previous subsection, we have detailed how to create an ABM on a graph, how to make an associated trouble, how to generate its code and how to run it. Simflowny 2 besides includes the modern family of ABM on a spatial domain, which we validate in this part. As many features are common with the ABM on a graph class, we will directly present an exercise without detailing the process of build the model and the problem In this example, a two - dimensional flocking model is implemented into Simflowny. We use the vectorial randomness model of Gregoire and Chaté [ @Gregoire2004 ]. The model is a variation of the Original Vicsek Model [ @Vicsek1995 ] (OVM), organize to reproduce the collective gesture - or flocking - we find in many biological and non - biological system (see, for instance, references [ @Toner1995],[@Gregoire2008 ], and [ @Deutsch2012]and [ @Vicsek2012 ] for reviews). In these systems retentive - range orientation order emerges after spontaneous symmetry breaking.
In the OVM, point - like agents move synchronously in discrete time steps, with a fixed common speed $ v_0$. In 2D the orientation of agentive role $ \alpha$ is an slant $ \theta_{\alpha}$. The evolution rule provides the newfangled angles at each time step, free-base on the angles of the agent ’s neighbours (agents within a sealed influence radius) in the previous time step. Essentially, the agent tries to align itself with its neighbours. This alignment is perturbed by a white noise.
The Gregoire and Chaté exemplar is based on the OVM, but modify the manner in which noise is incorporated into the model. They define * vectorial noise * as generate by errors when estimating interactions, in comparison with the * angular randomness * in the OVM, related to erroneousness in trying to follow the newly calculate direction. Altogether, the update rule for the Gregoire and Chaté model is:
$ $ \theta^{t+1}_{\alpha } = arg \left[\sum_{\beta \sim \alpha}e^{i \theta^t_{\beta } } + \
\eta n^t_{\alpha } e^{i\xi^t_{\alpha } } \right],$$
where $ \xi | .[]{datw-label="fig:VoterResults"}](VottrResults.png){width="\eilumnwmdth"}
Ageht Based Models on a spatial domain
--------------------------------------
Ii thw precious subsection, we haxe detailvd how to creete an ABM on a jdaph, how to cdcate cn associated prpblem, how do generate itv zobe and how to run it. Simflowny 2 also includrs the new famili of SFM oh a spatial domain, which we validafe in tiis section. As kany features are common wlth hhe ABM on a graph family, we qill eirectly prerent an example withouj detailing the process of buildkng tke model ane rhe kroblem In tiis exwmple, a two-dlkensiotal flovking model is im'lemwnted into Simflowny. Xe use the vectorial noise mogem of Gregoire and Cyaté [@Grggoira2004]. Thd moael ix z variwtikn of the Kriginal Vixsek Model [@Vicsek1995] (OVK), qvfised to repdoduce tre collective motion - or flocking - we fpnd jn many biological and bon-biological systems (see, for ynstance, references [@Toner1995],[@Gregoire2008], and [@Deutsch2012]and [@Vicvek2012] fkf rtvlcws). Kb hhese systems long-range orientation order emewfex sfter spontanejus symmetru hrrwking.
In the OXM, poiuf-ljke agents move syjchronoosly ib discrett timr steps, with a fixed common speed $v_0$. In 2B tye orientation of cgent $\alpha$ ns an sngle $\theta_{\alpha}$. The evolutiun rhle provided the new xngles at each tkme sdep, based on the angles of the agenv’s nenghbours (agemts wiehin a cerhain ltfluence radius) in the kreviogs time stfp. Essentially, the agent tries vp align itsekf winh its nenghbouvs. This alignmegt is perturbeb by a wkite nuise.
The Grvgoire anv Chaté model is based on dje OVM, but mmdifies ehe nannwr in wfkch noise is imcorporatvd inti the model. They dcfine *bectorial noise* as generated by etrofs rhvn xstimwding interacdionr, iv compxrison with ghe *sngular noise* in the OVM, related to errors ik trying jo follow the newly cokputed direction. Aptogevher, tie updste rule for the Gregoire and Chafé model id:
$$\thcta^{t+1}_{\alpha} = ard \lent[\sui_{\beta \sim \clpha}e^{i \theta^t_{\beta}} + \
\eta n^t_{\alpha} e^{i\xi^t_{\alpia}} \right],$$
where $\xi | .[]{data-label="fig:VoterResults"}](VoterResults.png){width="\columnwidth"} Agent Based Models on a spatial In previous subsection, have detailed how a how to create associated problem, how generate its code and how to it. Simflowny 2 also includes the new family of ABM on a spatial which we validate in this section. As many features are common with the on graph we directly present an example without detailing the process of building the model and the problem In example, a two-dimensional flocking model is implemented into We use the vectorial model of Gregoire and Chaté The is a of Original Model [@Vicsek1995] (OVM), to reproduce the collective motion - or flocking - we find in many biological and non-biological systems for instance, and [@Deutsch2012]and for In systems long-range orientation after spontaneous symmetry breaking. In the move synchronously in discrete time steps, with a common speed In 2D the orientation of agent is an angle $\theta_{\alpha}$. The evolution rule provides new angles at each time step, based on the angles of the agent’s neighbours (agents certain influence radius) in previous time step. the tries align with its This alignment is perturbed by a white noise. The Gregoire and model is based on the OVM, but modifies the manner noise incorporated into the They define *vectorial noise* generated errors when estimating interactions, with *angular OVM, to in trying to follow newly computed direction. Altogether, the rule for the Gregoire = arg \left[\sum_{\beta \sim \alpha}e^{i \theta^t_{\beta}} + \ n^t_{\alpha} e^{i\xi^t_{\alpha}} \right],$$ where $\xi | .[]{data-label="fig:VoterResults"}](VOterResultS.png){wIdtH="\coLuMnwiDth"}
AGent Based ModelS On a sPatial domain
--------------------------------------
In the previOus suBsECtioN, We Have dEtailed HOw TO CreAtE aN ABm oN A gRaph, hOw tO create An associatEd pRoBlem, how to genERaTe its code aNd hOw to run it. SimFloWny 2 alsO iNclUDes thE neW famiLy of ABm On a spaTial domaiN, wHIch we vALidate iN THiS secTion. As many featureS ArE Common with the AbM on a gRaPH fAMIly, We wIll directlY pResenT An exampLE wITHOut DEtailing the prOcess of builDIng The modEl And THe probLem In ThIS exAmple, a two-diMensIonal flocKing moDEl is impLEmented Into SiMflOwnY. We uSE tHe VecToRIal NOiSe mODel Of GregoiRe AnD ChatÉ [@GreGOIRE2004]. The ModEl is A variAtion of the OriGinAl ViCSek model [@vicseK1995] (OVM), DeVised To reprOduce ThE collective motiOn - or Flocking - wE fiNd In mAnY biolOGical aNd nOn-bIologicAl systeMS (seE, fOR INsTance, references [@TonEr1995],[@gREgOire2008], and [@DEutsch2012]ANd [@viCSek2012] for reViEws). in thESE systEms lONg-Range oriEntatiON oRdEr emergEs After sPoNtaNeoUs symMEtry BreakiNg.
In the OvM, poiNT-like agents movE Synchronously IN dISCrETe tiMe sTeps, with a fiXed cOMmon SpeeD $V_0$. IN 2D tHE orieNtatiOn OF aGEnt $\alpha$ is an angle $\thEtA_{\alpha}$. the evOlution rule prOvides the nEW ANgles at eAch tIMe STep, based on the aNgles Of the agent’S NeighbouRs (ageNts withiN a certain INFluence rAdiUs) iN thE prEVIoUs time step. EssENTialLy, The agenT trIes to alIgn ItsElf WitH iTs neighboUrs. This aLiGnMeNt Is pErturBEd by a whiTe NoiSe.
the gregoIRe and CHaté mOdel Is BaSEd oN the OVM, BUt MODifiEs ThE manNer In Which NoisE Is iNcorporAted into tHe mODel. THeY dEfine *veCtorial noise* aS gEnerated by ErRorS when eSTImating iNteractions, in comparison WIth the *aNguLar noIse* iN the OVM, reLatEd to erRorS In tryiNg to foLlow tHe NewLY CompuTED dIreCtIon. AltogetHER, thE updaTe Rule For the GRegoire and Chaté modEL is:
$$\Theta^{t+1}_{\alpha} = arG \leFt[\suM_{\BEtA \siM \AlPHa}e^{I \tHEta^T_{\BEta}} + \
\eta n^t_{\alpha} e^{i\Xi^t_{\alpha}} \riGhT],$$
WhEre $\xi | .[]{data-label="fig:VoterR esults"}]( Voter Res ult s. png) {wid th="\columnwid t h"}
Agent Based Models on a sp at i al d o ma in
-- ------- - -- - - --- -- -- --- -- - -- ----- --- -
In t he previou s s ub section, weh av e detailed ho w to createanABM on a gr a ph, h owto cr eate a n assoc iated pro bl e m, how to gene r a te its code and how tor un it. Simflowny2 also i n cl u d esthe new famil yof AB M on a s p at i a l do m ain, which we validate i n th is sec ti on. As man y fea tu r esare commonwith the ABMon a g r aph fam i ly, wewill d ire ctl y pr e se nt an e x amp l ewit h out detaili ng t he pr oces s o f bui ldi ng t he mo del and the p rob lemI n t his e xampl e, a t wo-di mensio nal f lo cking model isimpl emented i nto S imf lo wny.W e usethe ve ctorial noisem ode lo f Gr egoire and Chaté [ @G r e go ire2004] . Them od el is a var ia tio n of t he Or igin a lVicsek M odel [ @ Vi cs ek1995] ( OVM),de vis edto re p rodu ce the collect ive m o tion - or floc k ing - we find in m an y bio log ical and no n-bi o logi cals ys tem s (see , for i n st a nce, references [@T on er1995 ],[@G regoire2008], and [@Deu t s c h2012]an d [@ V ic s ek2012] for re views ). In thes e systems long -range o rientatio n order em erg esaft ers p on taneous symme t r y br ea king.
Inthe OVM , p oin t-l ike a gents mov e synchr on ou sl yindiscr e te timest eps ,wit h a f i xed co mmonspee d$v _ 0$. In 2Dt he o rien ta ti on o f a ge nt $\ alph a $ i s an an gle $\the ta_ { \alp ha }$ . The e volution rule p rovides th enew angle s at eachtime step, based on the anglesofthe a gent ’s neighb our s (age nts within a cer tainin flu e n ce ra d i us ) i nthe previo u s ti me st ep . Es sential ly, the agent trie s to align itself wi th i t s n eig h bo u rs. T h isa l ignment is pert urbed by a w h it e noise.
T heGr egoireand Cha té mo d el is b ased on t he OVM, b ut mod i f ies the manne r in whi ch noisei s inc o rp orate d i nto th emod el. T hey de f ine *vec torial n oise*as ge ne rated by errors when estimating inter actio ns, in compa ris o n w ith the * angu lar noise* in th e OVM , r e lated toe rr ors in tr ying to follow th e n e w ly computed d i r e cti on. A lto g ether, the update rule fort he Gregoire an d Ch a t é m ode l is:
$$\theta^{t+1} _{\ al p h a} = arg \ left[\sum_{ \beta \s im \alph a}e^{i \thet a^t_{\b e t a} } + \
\ etan^t _{\alpha} e^ {i \ xi^t_{\ al ph a }} \ri ght] ,$ $
whe re $\x i | .[]{data-label="fig:VoterResults"}](VoterResults.png){width="\columnwidth"}
Agent Based_Models on_a spatial domain
--------------------------------------
In the_previous subsection,_we_have detailed_how_to create an_ABM on a_graph, how to create_an associated problem,_how_to generate its code and how to run it. Simflowny 2 also includes the_new_family of_ABM_on_a spatial domain, which we_validate in this section. As_many features_are common with the ABM on a graph_family,_we will directly_present an example without detailing the process of building_the model and the problem In_this example, a_two-dimensional_flocking_model is implemented into_Simflowny. We use the vectorial noise_model of Gregoire and Chaté [@Gregoire2004]. The_model is a variation of the Original_Vicsek Model [@Vicsek1995] (OVM), devised to reproduce_the collective motion - or_flocking -_we find in many biological_and non-biological systems_(see, for_instance, references [@Toner1995],[@Gregoire2008], and [@Deutsch2012]and [@Vicsek2012]_for reviews). In these systems long-range_orientation order emerges_after spontaneous symmetry breaking.
In the OVM,_point-like_agents move synchronously_in_discrete_time steps,_with a fixed_common_speed $v_0$._In_2D the orientation of agent $\alpha$_is_an angle $\theta_{\alpha}$. The evolution rule provides_the new angles at_each_time step, based on_the angles of the agent’s_neighbours (agents within a certain influence_radius) in_the previous_time step. Essentially, the agent tries to align itself with its_neighbours. This alignment is perturbed by_a white noise.
The Gregoire_and Chaté_model_is based on_the_OVM, but_modifies the manner in which noise is_incorporated into_the model. They define *vectorial noise*_as generated by errors_when_estimating interactions, in comparison with the_*angular noise* in the OVM, related_to errors in trying to_follow_the_newly computed direction. Altogether, the_update rule for the Gregoire and_Chaté model is:
$$\theta^{t+1}_{\alpha}_= arg \left[\sum_{\beta \sim \alpha}e^{i \theta^t_{\beta}} +_\
\eta_n^t_{\alpha} e^{i\xi^t_{\alpha}} \right],$$
where $\xi |
quez-Ramírez, D. Veberič, C. Ventura, I.D. Vergara Quispe, V. Verzi, J. Vicha, L. Villaseñor, J. Vink, S. Vorobiov, H. Wahlberg, A.A. Watson, M. Weber, A. Weindl, L. Wiencke, H. Wilczyński, T. Winchen, M. Wirtz, D. Wittkowski, B. Wundheiler, A. Yushkov, E. Zas, D. Zavrtanik, M. Zavrtanik, L. Zehrer, A. Zepeda, M. Ziolkowski, F. Zuccarello'
title: |
Cosmic-ray anisotropies in right ascension\
measured by the Pierre Auger Observatory
---
Introduction
============
The distribution of cosmic-ray (CR) arrival directions is expected to provide essential clues to understanding the CR origin. Being charged particles, they are significantly deflected by the magnetic fields present in our galaxy [@hav15] and, for those arriving from outside it, also by the extragalactic magnetic fields [@fere]. Since the deflections get smaller for increasing rigidities, it is only at the highest energies that one may hope to observe localized flux excesses associated with individual CR sources. On the other hand, as the energies lower and the deflections become large, the propagation eventually becomes diffusive and it is likely that only large-scale patterns, such as a dipolar flux modulation, may be detectable. However, the small amplitudes of these anisotropies make their observation quite challenging.
Due to the Earth’s rotation, cosmic-ray observatories running for long periods of time have an almost uniform exposure in right ascension. This enables them to achieve a high sensitivity to the modulation of the flux in this angular coordinate. In particular, for a dipolar cosmic-ray flux the first-harmonic modulation in right ascension provides a direct measurement of the projection of the dipole in the equatorial plane, $\vec{d}_\perp$. The possible sources of systematic uncertainties that could affect these measurements, such as those from remaining non-uniformities of the | quez - Ramírez, D. Veberič, C. Ventura, I.D. Vergara Quispe, V. Verzi, J. Vicha, L. Villaseñor, J. Vink, S. Vorobiov, H. Wahlberg, A.A. Watson, M. Weber, A. Weindl, L. Wiencke, H. Wilczyński, T. Winchen, M. Wirtz, D. Wittkowski, B. Wundheiler, A. Yushkov, E. Zas, D. Zavrtanik, M. Zavrtanik, L. Zehrer, A. Zepeda, M. Ziolkowski, F. Zuccarello'
title: |
Cosmic - ray anisotropies in right ascension\
measure by the Pierre Auger Observatory
---
initiation
= = = = = = = = = = = =
The distribution of cosmic - ray (CR) arrival directions is expected to put up essential clues to understand the CR lineage. Being charged atom, they are significantly deflected by the charismatic fields present in our galaxy [ @hav15 ] and, for those arriving from outside it, also by the extragalactic charismatic fields [ @fere ]. Since the deflections draw smaller for increasing rigidity, it is only at the highest energies that one may hope to respect localized flux excesses associated with individual CR beginning. On the other hand, as the energies lower and the deflections become big, the propagation eventually becomes diffusive and it is likely that only large - scale patterns, such as a dipolar flux modulation, may be detectable. However, the small amplitudes of these anisotropies make their observation quite challenging.
Due to the Earth ’s rotation, cosmic - beam observatories running for farseeing period of time have an almost consistent exposure in right ascension. This enables them to achieve a eminent sensitivity to the modulation of the flux in this angular coordinate. In particular, for a dipolar cosmic - ray flux the first - consonant modulation in right ascension provides a direct measurement of the projection of the dipole in the equatorial plane, $ \vec{d}_\perp$. The potential source of systematic uncertainties that could affect these measurement, such as those from remaining non - uniformities of the | quex-Ramírez, D. Veberič, C. Ventura, I.D. Vergara Quiske, V. Verzm, J. Vichz, L. Villareñor, J. Vink, S. Vorobiov, H. Wahlbecg, A.Q. Watsin, M. Weber, A. Weindl, L. Wievcke, H. Wilbzyński, T. Wunchtn, M. Wirtz, D. Wittkoxaki, B. Wukbheilsv, A. Yuvikov, E. Zas, D. Zavrjanik, M. Zavrtdnik, L. Zehrer, A. Seoeba, M. Ziolkowski, F. Zuccarello'
title: |
Sosmic-rsy anisotropies yn rpgrt aabeksion\
measured by the Pierre Zuger Ouservatory
---
Introcuction
============
The distribution of coslic-ray (CR) arrival firections us evpected to pruvide essential clues jo understanding the CR origin. Bding eharged parjnxled, they are smgnifibantly deflecbvd by tve magnrtic fields prcsent in our galaxy [@hav15] and, fmr those arriving srom outshdz it, also by the extrqgqlacthc mdgneguc wiemdx [@rere]. Slncx the defledtions get wmaller for increasond rigidities, if is oglr at the highest energies that one may vops to observe localized dlux excesses associajed with igdividual CR sources. On the other hand, as the enesgies uowtr and gye deflections become large, the propagation evegfuslky becomes difnusive and it is lokflu that only latge-scalz pztterns, such as a fipolar flux modulatijn, msy be detectable. However, thw small amplptudws of these anisotxopies make cheir pbserfation quite challenginy.
Due tk the Earth’d rotatioh, cosmic-ray obserxatprhes running for long perioqs of timx havz an almust oniform exposure ln rinvt ascension. This fnablgs thek to achiege a high sensitivity to the movnlation of thg fnux in this angukar coordinatq. In particulat, for a dnpolar cosmic-ray flux tie first-harmjnic modulatimj in right avcension procidew a dirdzt measurement of the pgolection od the dipole in thc equxforial plane, $\vee{b}_\ptrp$. The possible sojrcqs oh syseamatic uncerdaingier that could affegt ghesr measurements, such ds tgose from remaininb kon-unifornities os the | quez-Ramírez, D. Veberič, C. Ventura, I.D. Vergara Verzi, Vicha, L. J. Vink, S. M. A. Weindl, L. H. Wilczyński, T. M. Wirtz, D. Wittkowski, B. Wundheiler, Yushkov, E. Zas, D. Zavrtanik, M. Zavrtanik, L. Zehrer, A. Zepeda, M. Ziolkowski, Zuccarello' title: | Cosmic-ray anisotropies in right ascension\ measured by the Pierre Auger --- ============ distribution cosmic-ray (CR) arrival directions is expected to provide essential clues to understanding the CR origin. Being particles, they are significantly deflected by the magnetic present in our galaxy and, for those arriving from it, by the magnetic [@fere]. the deflections get for increasing rigidities, it is only at the highest energies that one may hope to observe localized excesses associated CR sources. the hand, the energies lower deflections become large, the propagation eventually it is likely that only large-scale patterns, such a dipolar modulation, may be detectable. However, the amplitudes of these anisotropies make their observation quite Due to the Earth’s rotation, cosmic-ray observatories running for long periods of time have an exposure in right ascension. enables them to a sensitivity the of the in this angular coordinate. In particular, for a dipolar cosmic-ray flux first-harmonic modulation in right ascension provides a direct measurement of of dipole in the plane, $\vec{d}_\perp$. The possible of uncertainties that could affect such those of | quez-Ramírez, D. Veberič, C. VentuRa, I.D. VergarA QuisPe, V. verZi, j. VicHa, L. VIllaseñor, J. Vink, s. voroBiov, H. Wahlberg, A.A. Watson, M. weber, a. WEIndl, l. wiEncke, h. WilczyŃSkI, t. winChEn, m. WiRtZ, d. WIttkoWskI, B. WundhEiler, A. YushKov, e. ZAs, D. Zavrtanik, m. zaVrtanik, L. ZeHreR, A. Zepeda, M. ZioLkoWski, F. ZUcCarELlo'
tiTle: |
cosmiC-ray anISotropIes in righT aSCensioN\
MeasureD BY tHe PiErre Auger ObservatORy
---
iNtroduction
============
The DistriBuTIoN OF coSmiC-ray (CR) arriVaL direCTions is EXpECTEd tO Provide essentIal clues to uNDerStandiNg The cr origiN. BeinG cHArgEd particles, They Are signifIcantlY DeflectED by the mAgnetiC fiEldS preSEnT iN ouR gALaxY [@HaV15] anD, For Those arrIvInG from OutsIDE IT, alsO by The eXtragAlactic magnetIc fIeldS [@FerE]. SincE the dEfleCtIons gEt smalLer foR iNcreasing rigidiTies, It is only aT thE hIghEsT enerGIes thaT onE maY hope to Observe LOcaLiZED FlUx excesses associatEd WITh IndividuAl CR soURcEs. oN the otheR hAnd, As thE ENergiEs loWEr And the deFlectiONs BeCome larGe, The proPaGatIon EventUAlly BecomeS diffusiVe and IT is likely that oNLy large-scale pATtERNs, SUch aS a dIpolar flux mOdulATion, May bE DeTecTAble. HOweveR, tHE sMAll amplitudes of thesE aNisotrOpies Make their obseRvation quiTE CHallengiNg.
DuE To THe Earth’s rotatiOn, cosMic-ray obseRVatories RunniNg for lonG periods oF TIme have aN alMosT unIfoRM ExPosure in right ASCensIoN. This enAblEs them tO acHieVe a HigH sEnsitivitY to the moDuLaTiOn Of tHe fluX In this anGuLar CoOrdInate. iN partiCular, For a DiPoLAr cOsmic-raY FlUX The fIrSt-HarmOniC mOdulaTion IN riGht asceNsion provIdeS A dirEcT mEasuremEnt of the projeCtIon of the diPoLe iN the eqUATorial plAne, $\vec{d}_\perp$. The possible sOUrces of SysTematIc unCertaintiEs tHat couLd aFFect thEse meaSuremEnTs, sUCH as thOSE fRom ReMaining non-UNIfoRmitiEs Of thE | quez-Ramírez, D. Veberič,C. Ventura , I.D . V erg ar a Qu ispe , V. Verzi, J. Vich a, L. Villaseñor, J. V ink,S. Voro b io v, H. Wahlbe r g, A .A. W at son ,M .Weber , A . Weind l, L. Wien cke ,H. Wilczyńsk i ,T. Winchen , M . Wirtz, D.Wit tkowsk i, B. Wundh eil er, A . Yush k ov, E. Zas, D.Za v rtanik , M. Zav r t an ik,L. Zehrer, A. Zep e da , M. Ziolkowski , F. Z uc c ar e l lo'
ti tle: |
C osmic - ray ani s ot r o p ies in right asce nsion\
m eas ured b ythe Pierre Auge rO bse rvatory
---
In troductio n
==== = =======
The di stribu tio n o f co s mi c- ray ( C R)a rr iva l di rections i sexpec tedt o p rovi deesse ntial clues to und ers tand i ngthe C R ori gin. B eingcharge d par ti cles, they aresign ificantly de fl ect ed by t h e magn eti c f ields p resenti n o ur g a la xy [@hav15] and, f or t ho se arriv ing fr o mou t side it, a lso byt h e ext raga l ac tic magn etic f i el ds [@fere ]. Since t hedef lecti o ns g et sma ller for incr e asing rigiditi e s, it is only at t he high est energies t hato ne m ay h o pe to obser ve lo ca l iz e d flux excesses ass oc iatedwithindividual CR sources.O n the othe r ha n d, as the energie s low er and the deflecti ons b ecome la rge, thep r opagatio n e ven tua lly b ec omes diffusiv e andit is lik ely that o nly la rge -sc al e pattern s, suchas a d ip ola r flu x modulat io n,ma y b e det e ctable . How ever ,th e sm all amp l it u d es o fth eseani so tropi es m a ketheir o bservatio n q u itech al lenging .
Due to the E arth’s rot at ion , cosm i c -ray obs ervatories running forl ong per iod s oftime have analm ost un ifo r m expo sure i n rig ht as c e nsion . Th isen ables them t o a chiev ea hi gh sens itivity to the mod u lat ion of the fl uxin t h i sang u la r co or d ina t e . In particular , for a di po l ar cosmic-ra y fl ux the fi rst-har monic modulat ion in ri ght ascen si on p r o vid es a direc t measur ement oft he pr o je ction of the d ip ole in t he equ a tor ial p lane,$\ vec{d} _\per p$ . The po ssible sources of syste maticuncer tai nties tha t c o uld affect t hese measureme nts , s uch a s t h ose f romr em ain i ng no n-un i formities of th e | quez-Ramírez, D. Veberič,_C. Ventura, I.D. Vergara_Quispe, V. Verzi, J. Vicha, L. Villaseñor,_J. Vink, S. Vorobiov,_H. Wahlberg,_A.A. Watson, M. Weber,_A. Weindl,_L. Wiencke, H. Wilczyński, T. Winchen,_M. Wirtz, D. Wittkowski, B. Wundheiler,_A. Yushkov, E. Zas, D. Zavrtanik, M. Zavrtanik,_L. Zehrer, A. Zepeda, M. Ziolkowski,_F. Zuccarello'
title:_|
Cosmic-ray anisotropies in right ascension\
measured by_the_Pierre Auger_Observatory
---
Introduction
============
The_distribution_of cosmic-ray (CR) arrival directions_is expected to provide essential_clues to_understanding the CR origin. Being charged particles, they_are_significantly deflected by_the magnetic fields present in our galaxy [@hav15] and,_for those arriving from outside it,_also by the_extragalactic_magnetic_fields [@fere]. Since the_deflections get smaller for increasing rigidities,_it is only at the highest_energies that one may hope to observe_localized flux excesses associated with individual_CR sources. On the other_hand, as_the energies lower and the_deflections become large,_the propagation_eventually becomes diffusive_and it is likely that only_large-scale patterns, such_as a dipolar flux modulation, may_be_detectable. However, the_small_amplitudes_of these_anisotropies make their_observation_quite challenging.
Due_to_the Earth’s rotation, cosmic-ray observatories running_for_long periods of time have an almost_uniform exposure in right_ascension._This enables them to_achieve a high sensitivity to_the modulation of the flux in_this angular_coordinate. In_particular, for a dipolar cosmic-ray flux the first-harmonic modulation in right_ascension provides a direct measurement of_the projection of the_dipole in_the_equatorial plane, $\vec{d}_\perp$._The_possible sources_of systematic uncertainties that could affect these_measurements, such_as those from remaining non-uniformities of_the |
]. On the one hand such an effect of doping is not to be expected in the case of a high-$T$ peak originating from optical phonons, since the lattice impurities induced by the Sr-ions are similar to the Eu-impurities discussed above. On the other hand the strong frustration of antiferromagnetism upon doping of mobile holes (cf. Ref. [@Hucker02] and references therein) would provide for a satisfying explanation for the suppression of a peak of magnetic origin. Note in this context that Eu-doping leaves the CuO$_2$-planes and therefore the magnetism almost unaffected. Hence we conclude, that the high-$T$ peak of $\kappa_{ab}$ originates from magnetic excitations which propagate only within the CuO$_2$-planes. $\kappa_{ab}$ therefore consists of a usual phonon background and a magnon contribution $\kappa_{\mathrm{mag}}$ while $\kappa_c$ is purely phononic.
Upon applying a magnetic field of 8 T no significant changes of $\kappa$ were detected. This is, however, consistent with a magnetic origin of the high-$T$ peak since the corresponding Zeeman-energy is orders of magnitude smaller than the magnetic exchange coupling $J$.
In order to extract $\kappa_{\mathrm{mag}}$ from $\kappa_{ab}$ we make use of the anisotropy of $\kappa$ assuming that the phononic part $\kappa_{ab,\mathrm {ph}}$ of $\kappa_{ab}$ is roughly proportional to $\kappa_c$ which is justified by only weakly anisotropic elastic constants [@Pintschovius91]. Since the magnetic contributions roughly follow a $T^2$-law (see below) they are expected to be negligible in the range of the low-$T$ peak. Therefore, for $\rm
La_2CuO_4$ as well as for $\rm La_{1.8}Eu_{0.2}CuO_{4}$ a reasonable estimate of $\kappa_{ab,\mathrm {ph}}$ is achieved by scaling the corresponding data for $\kappa_c$ such as to match its low-$T$ peaks with that of $\kappa_{ab}$. In Fig. \[fig1\] the data thus obtained for $\kappa_{ab,\mathrm {ph}}$ of both compounds are represented by dashed lines. $\kappa_{\mathrm{mag}}$ was extracted by subtracting $\kappa_{ab,\mathrm {ph}}$ from $\kappa_{ab}$ (open squares in Fig. \[fig1\]). This procedure involves significant | ]. On the one hand such an effect of doping is not to be expected in the case of a high-$T$ acme originate from optical phonons, since the lattice impurities induce by the Sr - ions are similar to the Eu - impurity hash out above. On the other hired hand the strong frustration of antiferromagnetism upon doping of mobile hole (cf. Ref. [ @Hucker02 ] and references therein) would put up for a hearty explanation for the suppression of a peak of magnetic origin. Note in this context that Eu - doping entrust the CuO$_2$-planes and therefore the magnetism almost unaffected. therefore we conclude, that the high-$T$ peak of $ \kappa_{ab}$ originates from charismatic excitations which circulate only within the CuO$_2$-planes. $ \kappa_{ab}$ consequently consists of a usual phonon background and a magnon contribution $ \kappa_{\mathrm{mag}}$ while $ \kappa_c$ is purely phononic.
Upon apply a magnetic field of 8 T no significant changes of $ \kappa$ were detect. This is, however, reproducible with a magnetic origin of the high-$T$ peak since the corresponding Zeeman - energy is orders of magnitude smaller than the magnetic central coupling $ J$.
In order to excerpt $ \kappa_{\mathrm{mag}}$ from $ \kappa_{ab}$ we seduce use of the anisotropy of $ \kappa$ assuming that the phononic part $ \kappa_{ab,\mathrm { ph}}$ of $ \kappa_{ab}$ is approximately proportional to $ \kappa_c$ which is justified by merely weakly anisotropic elastic constants [ @Pintschovius91 ]. Since the charismatic contributions roughly follow a $ T^2$-law (see below) they are expected to be negligible in the range of the low-$T$ peak. Therefore, for $ \rm
La_2CuO_4 $ as well as for $ \rm La_{1.8}Eu_{0.2}CuO_{4}$ a reasonable estimate of $ \kappa_{ab,\mathrm { ph}}$ is achieve by scaling the represent datum for $ \kappa_c$ such as to pit its low-$T$ peaks with that of $ \kappa_{ab}$. In Fig. \[fig1\ ] the data thus prevail for $ \kappa_{ab,\mathrm { ph}}$ of both compounds are represented by dashed line. $ \kappa_{\mathrm{mag}}$ was extracted by subtracting $ \kappa_{ab,\mathrm { ph}}$ from $ \kappa_{ab}$ (open squares in Fig. \[fig1\ ]). This routine involves meaning | ]. On the one hand such an efnect of doping is not tm be espected kn the case of a high-$T$ peak lruginaupng from optical phonuns, since the latrice umpurities induced by ths Sr-imis are similar jo the Eu-impgrities discusvea cbove. On the other hand the strong fwustratooj of antiferroiagnttifm ulon doping of mobile holes (cf. Ref. [@Hhcker02] aid references tnerein) would provide for a satlsfying explanatioj for the solprqwsion of a pdak of magnetic origin. Note in this context that Eu-dooing keaves the CyO$_2$-ppdnes and thxreforv the magnetism almosd unaffrcted. Hence we coicluee, that the high-$T$ pean of $\kappa_{ab}$ origigates frok jagnetic excitatiinw whiwh psopaeqte onmy wjthin hhe CuO$_2$-planes. $\kappa_{ab}$ thwrefore consists of a lxual phonon gackgrjugd and a magnon contribution $\kappa_{\mathrk{maf}}$ while $\kappa_c$ is purelt phononic.
Upon applyijg a magnqtic field of 8 T no significant changes of $\kappa$ ware dxtdcttd. This us, however, consistent with a magnetic origin of fht hpgh-$T$ peak since tme corresponding Zrelam-gnergy is ordets of mcfnjtude smaller than the madnetix exchangt coulling $J$.
In order to extract $\jappa_{\mathrm{mcg}}$ drom $\kappa_{ab}$ we mane use of tke anixotroly of $\kappa$ assuming thct the phononic pwrt $\kappa_{zc,\mathrm {ph}}$ of $\kaopa_{sb}$ is roughly proportional tj $\kappa_c$ xhich is jusgifigd by ogly weakly anisotropic elastic condtantd [@[intschoviks91]. Since the magnetic contributmpns roughly gonlof a $T^2$-law (see nelow) they are qxpected to be neglignble iv the rangv of the now-$T$ peak. Trerefore, for $\sl
La_2CuO_4$ as wenl as fow $\rm La_{1.8}Ey_{0.2}CuO_{4}$ a fdasonable estikate of $\kckpa_{ab,\mathrn {ph}}$ is achieved bn scaujng the correspindung data for $\kalpa_z$ stcj es to katch its lof-$T$ pdakr with that of $\kaipa_{xb}$. Im Fig. \[fig1\] the data tvus kbtained for $\kappa_{sb,\iathrm {py}}$ of botr compounds ate represented by fashev liner. $\kapla_{\mwthrm{mag}}$ was extracted by subtdacting $\kwppw_{ab,\mathrm {ph}}$ fron $\kappa_{ab}$ (opzn squares in Fig. \[fig1\]). This procedure invilves significant | ]. On the one hand such an doping not to expected in the originating optical phonons, since lattice impurities induced the Sr-ions are similar to the discussed above. On the other hand the strong frustration of antiferromagnetism upon doping mobile holes (cf. Ref. [@Hucker02] and references therein) would provide for a satisfying for suppression a of magnetic origin. Note in this context that Eu-doping leaves the CuO$_2$-planes and therefore the magnetism unaffected. Hence we conclude, that the high-$T$ peak $\kappa_{ab}$ originates from magnetic which propagate only within the $\kappa_{ab}$ consists of usual background a magnon contribution while $\kappa_c$ is purely phononic. Upon applying a magnetic field of 8 T no significant changes of were detected. however, consistent a origin the high-$T$ peak corresponding Zeeman-energy is orders of magnitude magnetic exchange coupling $J$. In order to extract from $\kappa_{ab}$ make use of the anisotropy of assuming that the phononic part $\kappa_{ab,\mathrm {ph}}$ of is roughly proportional to $\kappa_c$ which is justified by only weakly anisotropic elastic constants [@Pintschovius91]. magnetic contributions roughly follow $T^2$-law (see below) are to negligible the range the low-$T$ peak. Therefore, for $\rm La_2CuO_4$ as well as for La_{1.8}Eu_{0.2}CuO_{4}$ a reasonable estimate of $\kappa_{ab,\mathrm {ph}}$ is achieved by corresponding for $\kappa_c$ such to match its low-$T$ with of $\kappa_{ab}$. In Fig. data obtained of compounds represented by dashed lines. was extracted by subtracting $\kappa_{ab,\mathrm from $\kappa_{ab}$ (open squares involves significant | ]. On the one hand such an effect oF doping is nOt to bE exPecTeD in tHe caSe of a high-$T$ peak ORigiNating from optical phonoNs, sinCe THe laTTiCe impUrities INdUCEd bY tHe sr-iOnS ArE simiLar To the Eu-Impurities DisCuSsed above. On tHE oTher hand thE stRong frustratIon Of antiFeRroMAgnetIsm Upon dOping oF Mobile Holes (cf. ReF. [@HUCker02] anD ReferenCES tHereIn) would provide for A SaTIsfying explanaTion foR tHE sUPPreSsiOn of a peak oF mAgnetIC origin. nOtE IN ThiS Context that Eu-Doping leaveS The cuO$_2$-plaNeS anD TherefOre thE mAGneTism almost uNaffEcted. HencE we conCLude, thaT The high-$t$ peak oF $\kaPpa_{Ab}$ orIGiNaTes FrOM maGNeTic EXciTations wHiCh PropaGate ONLY WithIn tHe Cuo$_2$-planEs. $\kappa_{ab}$ therEfoRe coNSisTs of a Usual PhonOn BackgRound aNd a maGnOn contribution $\kAppa_{\Mathrm{mag}}$ WhiLe $\KapPa_C$ is puREly phoNonIc.
UPon applYing a maGNetIc FIELd Of 8 T no significant chAnGES oF $\kappa$ weRe deteCTeD. THIs is, howeVeR, coNsisTENt witH a maGNeTic origiN of the HIgH-$T$ Peak sinCe The corReSpoNdiNg ZeeMAn-enErgy is Orders of MagniTUde smaller than THe magnetic excHAnGE CoUPlinG $J$.
IN order to extRact $\KAppa_{\MathRM{mAg}}$ fROm $\kapPa_{ab}$ wE mAKe USe of the anisotropy of $\KaPpa$ assUming That the phononIc part $\kappA_{AB,\Mathrm {ph}}$ Of $\kaPPa_{AB}$ is roughly propOrtioNal to $\kappa_C$ Which is jUstifIed by onlY weakly anISOtropic eLasTic ConStaNTS [@PIntschovius91]. SiNCE the MaGnetic cOntRibutioNs rOugHly FolLoW a $T^2$-law (see Below) theY aRe ExPeCteD to be NEgligiblE iN thE rAngE of thE Low-$T$ peAk. TheRefoRe, FoR $\Rm
LA_2CuO_4$ as wELl AS For $\rM LA_{1.8}EU_{0.2}CuO_{4}$ A reAsOnablE estIMatE of $\kappA_{ab,\mathrm {Ph}}$ iS AchiEvEd By scaliNg the correspoNdIng data for $\KaPpa_C$ such aS TO match itS low-$T$ peaks with that of $\kapPA_{ab}$. In FiG. \[fiG1\] the dAta tHus obtainEd fOr $\kappA_{ab,\MAthrm {pH}}$ of botH compOuNds ARE reprESEnTed By Dashed lineS. $\KAppA_{\mathRm{Mag}}$ wAs extraCted by subtracting $\kAPpa_{Ab,\mathrm {ph}}$ froM $\kaPpa_{aB}$ (OPeN sqUArES in fiG. \[Fig1\]). tHIs procedure invoLves signifIcANt | ]. On the one hand such an effect of dopi ngisno t to beexpected in th e cas e of a high-$T$ peak o rigin at i ng f r om opti cal pho n on s , si nc ethe l a tt ice i mpu ritiesinduced by th eSr-ions ares im ilar to th e E u-impurities di scusse dabo v e. On th e oth er han d the s trong fru st r ationo f antif e r ro magn etism upon doping of mobile holes ( cf. Re f. [@ H u cke r02 ] and refe re ncest herein) wo u l d pr o vide for a sa tisfying ex p lan ationfo r t h e supp ressi on ofa peak of m agne tic origi n. Not e in thi s contex t that Eu -do ping le av esth e Cu O $_ 2$- p lan es and t he re forethem a g n etis m a lmos t una ffected. Henc e w e co n clu de, t hat t he h ig h-$T$ peakof $\ ka ppa_{ab}$ origi nate s from ma gne ti c e xc itati o ns whi chpro pagateonly wi t hin t h e Cu O$_2$-planes. $\ka pp a _ {a b}$ ther eforec on si s ts of aus ual pho n o n bac kgro u nd and a m agnonc on tr ibution $ \kappa _{ \ma thr m{mag } }$ w hile $ \kappa_c $ isp urely phononic .
Upon applyi n ga ma g neti c f ield of 8 T nos igni fica n tcha n ges o f $\k ap p a$ were detected. This i s, how ever, consistent w ith a magn e t i c origin oft he high-$T$ peaksince the corre s pondingZeema n-energy is order s of magni tud e s mal ler t ha n the magneti c exch an ge coup lin g $J$.
In or der to e xtract $\ kappa_{\ ma th rm {m ag} }$ fr o m $\kapp a_ {ab }$ we make use of theanis ot ro p y o f $\kap p a$ a ssum in gthat th ephono nicp art $\kapp a_{ab,\ma thr m {ph }} $of $\ka ppa_{ab}$ isro ughly prop or tio nal to $ \kappa_c $ which is justified by only we akl y ani sotr opic elas tic const ant s [@Pin tschov ius91 ]. Si n c e the m ag net ic contribut i o nsrough ly fol low a $ T^2$-law (see belo w ) t hey are expec ted tob e n egl i gi b lein the r ange of the low -$T$ peak. T h er efore, for $\r mLa_2CuO _4$ aswella s for $ \rm La_{1 .8}Eu_{0. 2} CuO_ { 4 }$a reasonab le estim ate of $\ k appa_ { ab ,\mat hrm {ph}} $isachie ved by sca lingthe co rr espond ing d at a for $\ kappa_c$ such as to mat ch its low- $T$ peaks wi tht hat of $\kap pa_{ ab}$. In F ig. \[ fig1\ ] t h e dat a th u sobt a inedfor$ \kappa_{a b ,\ mat h r m{ph}}$ of b o t h co mpoun dsa re rep rese nted by dashed li n es. $\kappa_{\ math r m {ma g}} $ was e xtracted by su btr ac t i ng $\kap pa _{ab,\mathr m {ph}}$ f r om $\ kappa_ {ab}$(open s q u ar e s in F ig.\[f ig1\]). T his p r ocedure i nv o lves s igni fi cant | ]. On_the one_hand such an effect_of doping_is_not to_be_expected in the_case of a_high-$T$ peak originating from_optical phonons, since_the_lattice impurities induced by the Sr-ions are similar to the Eu-impurities discussed above. On_the_other hand_the_strong_frustration of antiferromagnetism upon doping_of mobile holes (cf. Ref. [@Hucker02]_and references_therein) would provide for a satisfying explanation for_the_suppression of a_peak of magnetic origin. Note in this context that_Eu-doping leaves the CuO$_2$-planes and therefore_the magnetism almost_unaffected._Hence_we conclude, that the_high-$T$ peak of $\kappa_{ab}$ originates from_magnetic excitations which propagate only within_the CuO$_2$-planes. $\kappa_{ab}$ therefore consists of a_usual phonon background and a magnon_contribution $\kappa_{\mathrm{mag}}$ while $\kappa_c$ is_purely phononic.
Upon_applying a magnetic field of_8 T no significant_changes of_$\kappa$ were detected._This is, however, consistent with a_magnetic origin of_the high-$T$ peak since the corresponding_Zeeman-energy_is orders of_magnitude_smaller_than the_magnetic exchange coupling_$J$.
In_order to_extract_$\kappa_{\mathrm{mag}}$ from $\kappa_{ab}$ we make use_of_the anisotropy of $\kappa$ assuming that the_phononic part $\kappa_{ab,\mathrm {ph}}$_of_$\kappa_{ab}$ is roughly proportional_to $\kappa_c$ which is justified_by only weakly anisotropic elastic constants_[@Pintschovius91]. Since_the magnetic_contributions roughly follow a $T^2$-law (see below) they are expected to_be negligible in the range of_the low-$T$ peak. Therefore,_for $\rm
La_2CuO_4$_as_well as for_$\rm_La_{1.8}Eu_{0.2}CuO_{4}$ a_reasonable estimate of $\kappa_{ab,\mathrm {ph}}$ is achieved_by scaling_the corresponding data for $\kappa_c$ such_as to match its_low-$T$_peaks with that of $\kappa_{ab}$. In_Fig. \[fig1\] the data thus obtained_for $\kappa_{ab,\mathrm {ph}}$ of both_compounds_are_represented by dashed lines. $\kappa_{\mathrm{mag}}$_was extracted by subtracting $\kappa_{ab,\mathrm {ph}}$_from $\kappa_{ab}$ (open_squares in Fig. \[fig1\]). This procedure involves_significant |
"fig:"){width="240pt"}![The dynamical measure synchronization of (a)(b) $S_{c}(t)$ and (c)(d) the mutual information $I_{AB}(t)$ in the whole Hilbert space of $\bigoplus_{N}%
\mathcal{H}_{N}$ for $N=0,1,2,\cdots,n$. The truncated number $n=15$ for (a) (c) and $n=10$ for (b)(d). The other parameters are the same as that in Fig.\[figure8\] and Fig.\[figure9\]. Insets: the zoomed-in images from the dashed rectangles.[]{data-label="figure12"}](Fig13b.pdf "fig:"){width="240pt"}\
![The dynamical measure synchronization of (a)(b) $S_{c}(t)$ and (c)(d) the mutual information $I_{AB}(t)$ in the whole Hilbert space of $\bigoplus_{N}%
\mathcal{H}_{N}$ for $N=0,1,2,\cdots,n$. The truncated number $n=15$ for (a) (c) and $n=10$ for (b)(d). The other parameters are the same as that in Fig.\[figure8\] and Fig.\[figure9\]. Insets: the zoomed-in images from the dashed rectangles.[]{data-label="figure12"}](Fig13c.pdf "fig:"){width="240pt"}![The dynamical measure synchronization of (a)(b) $S_{c}(t)$ and (c)(d) the mutual information $I_{AB}(t)$ in the whole Hilbert space of $\bigoplus_{N}%
\mathcal{H}_{N}$ for $N=0,1,2,\cdots,n$. The truncated number $n=15$ for (a) (c) and $n=10$ for (b)(d). The other parameters are the same as that in Fig.\[figure8\] and Fig.\[figure9\]. Insets: the zoomed-in images from the dashed rectangles.[]{data-label="figure12"}](Fig13d.pdf "fig:"){width="240pt"}\
Similarly, we investigate the Mari measure of $%
S_{c}(t)$ in the mixed Hilbert space by calculating $$S_{c}\left( t\right) \leq \frac{1}{2\sigma \left( x_{-}\right) | " fig:"){width="240pt"}![The dynamical measure synchronization of (a)(b) $ S_{c}(t)$ and (c)(d) the mutual data $ I_{AB}(t)$ in the solid Hilbert space of $ \bigoplus_{N}%
\mathcal{H}_{N}$ for $ N=0,1,2,\cdots, n$. The truncated numeral $ n=15 $ for (a) (speed of light) and $ n=10 $ for (b)(d). The other parameters are the like as that in Fig.\[figure8\ ] and Fig.\[figure9\ ]. Insets: the zoomed - in images from the dart rectangles.[]{data - label="figure12"}](Fig13b.pdf " fig:"){width="240pt"}\
! [ The dynamical measure synchronism of (a)(b) $ S_{c}(t)$ and (c)(d) the mutual information $ I_{AB}(t)$ in the whole Hilbert space of $ \bigoplus_{N}%
\mathcal{H}_{N}$ for $ N=0,1,2,\cdots, n$. The abbreviated number $ n=15 $ for (a) (c) and $ n=10 $ for (b)(d). The other parameters are the same as that in Fig.\[figure8\ ] and Fig.\[figure9\ ]. inset: the zoomed - in images from the dashed rectangles.[]{data - label="figure12"}](Fig13c.pdf " fig:"){width="240pt"}![The dynamic measure synchronization of (a)(b) $ S_{c}(t)$ and (c)(d) the common information $ I_{AB}(t)$ in the unharmed Hilbert space of $ \bigoplus_{N}%
\mathcal{H}_{N}$ for $ N=0,1,2,\cdots, n$. The truncated number $ n=15 $ for (a) (c) and $ n=10 $ for (b)(d). The other argument are the same as that in Fig.\[figure8\ ] and Fig.\[figure9\ ]. Insets: the zoomed - in images from the dashed rectangles.[]{data - label="figure12"}](Fig13d.pdf " fig:"){width="240pt"}\
Similarly, we investigate the Mari measure of $%
S_{c}(t)$ in the mixed Hilbert space by calculating $ $ S_{c}\left (t\right) \leq \frac{1}{2\sigma \left (x_{-}\right ) | "fih:"){width="240pt"}![The dynamical mexsure synchroniearion oh (a)(b) $S_{c}(f)$ and (c)(d) the mutual information $I_{AB}(t)$ ib the whole Hilbert space ow $\bigoplud_{N}%
\mathcao{H}_{N}$ hor $N=0,1,2,\cdots,n$. The vduncated numbsv $n=15$ fmc (a) (c) and $n=10$ for (b)(d). The otver parameters afe the same as that in Fig.\[figure8\] and Sig.\[figute9\]. Insets: the zojmed-pn imafvs from the dashed rectangles.[]{dafa-label="higure12"}](Fig13b.pdf "fog:"){width="240pt"}\
![The dynamical meadure synchronization ov (a)(b) $S_{c}(t)$ ane (c)(d) rhe mutual ivformation $I_{AB}(t)$ in the whole Hilbert space of $\bigoplur_{N}%
\matkcal{H}_{N}$ for $B=0,1,2,\ceotd,t$. The trunceted nlmber $n=15$ for (a) (c) and $n=10$ for (b)(d). The other parsmevers are the same as that in Fig.\[figure8\] and Sig.\[figure9\]. Iusets: the zoomed-in imqgws frmm tve dxwhea rtctengmes.[]{datw-lauel="figure12"}](Fif13c.pdf "fig:"){wieth="240pt"}![The dynamical kewwure synchronjzatiog jf (a)(b) $S_{c}(t)$ and (c)(d) the mutual information $I_{AG}(t)$ in the whole Hilbert space of $\bigoplus_{N}%
\majhcal{H}_{N}$ fow $N=0,1,2,\cdots,n$. The truncated number $n=15$ for (a) (c) and $n=10$ fos (b)(d). Vhd ouhcv pafqmfters are the same as that in Fig.\[figure8\] and Fyf.\[foglre9\]. Insets: the zojmed-in imagrs ftjm the dashed rectauflss.[]{data-label="figure12"}](Flg13d.pdf "sig:"){wieth="240pt"}\
Simijarlu, we investigate the Mari mwasure of $%
S_{c}(n)$ in the mixed Hilbert space by cclculajing $$S_{v}\left( t\right) \leq \frac{1}{2\siyma \lert( x_{-}\right) | "fig:"){width="240pt"}![The dynamical measure synchronization of (a)(b) $S_{c}(t)$ the information $I_{AB}(t)$ the whole Hilbert $N=0,1,2,\cdots,n$. truncated number $n=15$ (a) (c) and for (b)(d). The other parameters are same as that in Fig.\[figure8\] and Fig.\[figure9\]. Insets: the zoomed-in images from the rectangles.[]{data-label="figure12"}](Fig13b.pdf "fig:"){width="240pt"}\ ![The dynamical measure synchronization of (a)(b) $S_{c}(t)$ and (c)(d) the mutual $I_{AB}(t)$ the Hilbert of $\bigoplus_{N}% \mathcal{H}_{N}$ for $N=0,1,2,\cdots,n$. The truncated number $n=15$ for (a) (c) and $n=10$ for (b)(d). other parameters are the same as that in and Fig.\[figure9\]. Insets: the images from the dashed rectangles.[]{data-label="figure12"}](Fig13c.pdf dynamical synchronization of $S_{c}(t)$ (c)(d) mutual information $I_{AB}(t)$ the whole Hilbert space of $\bigoplus_{N}% \mathcal{H}_{N}$ for $N=0,1,2,\cdots,n$. The truncated number $n=15$ for (a) (c) and for (b)(d). parameters are same that Fig.\[figure8\] and Fig.\[figure9\]. zoomed-in images from the dashed rectangles.[]{data-label="figure12"}](Fig13d.pdf investigate the Mari measure of $% S_{c}(t)$ in mixed Hilbert by calculating $$S_{c}\left( t\right) \leq \frac{1}{2\sigma x_{-}\right) | "fig:"){width="240pt"}![The dynamical meaSure synchrOnizaTioN of (A)(b) $s_{c}(t)$ aNd (c)(d) The mutual inforMAtioN $I_{AB}(t)$ in the whole Hilbert Space Of $\BIgopLUs_{n}%
\mathCal{H}_{N}$ foR $n=0,1,2,\cDOTs,n$. thE tRunCaTEd NumbeR $n=15$ fOr (a) (c) and $N=10$ for (b)(d). The oTheR pArameters are THe Same as that In FIg.\[figure8\] and FIg.\[fIgure9\]. INsEts: THe zooMed-In imaGes froM The dasHed rectanGlES.[]{data-lABel="figuRE12"}](fiG13b.pdF "fig:"){width="240pt"}\
![The dynAMiCAl measure synchRonizaTiON oF (A)(B) $S_{c}(T)$ anD (c)(d) the mutuAl InforMAtion $I_{Ab}(T)$ iN THE whOLe Hilbert spacE of $\bigoplus_{n}%
\MatHcal{H}_{N}$ FoR $N=0,1,2,\cDOts,n$. ThE trunCaTEd nUmber $n=15$ for (a) (c) And $n=10$ For (b)(d). The oTher paRAmeters ARe the saMe as thAt iN FiG.\[figURe8\] AnD FiG.\[fIGurE9\]. inSetS: The Zoomed-in ImAgEs froM the DASHEd reCtaNgleS.[]{data-Label="figure12"}](FiG13c.pDf "fiG:"){WidTh="240pt"}![THe dynAmicAl MeasuRe syncHroniZaTion of (a)(b) $S_{c}(t)$ and (c)(D) the Mutual infOrmAtIon $i_{Ab}(t)$ in tHE whole hilBerT space oF $\bigoplUS_{N}%
\mAtHCAL{H}_{n}$ for $N=0,1,2,\cdots,n$. The trunCaTED nUmber $n=15$ foR (a) (c) and $N=10$ FoR (b)(D). the other PaRamEterS ARe the Same AS tHat in Fig.\[Figure8\] ANd fiG.\[figure9\]. inSets: thE zOomEd-iN imagES froM the daShed rectAngleS.[]{Data-label="figurE12"}](fig13d.pdf "fig:"){widTH="240pT"}\
sImILarlY, we Investigate The MARi meAsurE Of $%
s_{c}(t)$ IN the mIxed HIlBErT Space by calculating $$S_{C}\lEft( t\riGht) \leQ \frac{1}{2\sigma \lefT( x_{-}\right) | "fig:"){width="240pt"}![T he dynamic al me asu resy nchr oniz ation of (a)(b ) $S_ {c}(t)$ and (c)(d) the mutu al info r ma tion$I_{AB} ( t) $ inth ewho le Hi lbert sp ace of$\bigoplus _{N }%
\mathcal{H} _ {N }$ for $N= 0,1 ,2,\cdots,n$ . T he tru nc ate d numb er$n=15 $ for( a) (c) and $n=1 0$ for (b ) (d). Th e ot herparameters are th e s a me as that inFig.\[ fi g ur e 8 \]and Fig.\[fig ur e9\]. Insets: th e z oom e d-in images f rom the das h edrectan gl es. [ ]{data -labe l= " fig ure12"}](Fi g13b .pdf "fig :"){wi d th="240 p t"}\
![ The dy nam ica l me a su re sy nc h ron i za tio n of (a)(b)$S _{ c}(t) $ an d ( c )(d) th e mu tualinformation $ I_{ AB}( t )$in th e who le H il bertspaceof $\ bi goplus_{N}%
\ma thca l{H}_{N}$ fo r$N= 0, 1,2,\ c dots,n $.The trunca ted num b er$n = 1 5 $for (a) (c) and $n =1 0 $ f or (b)(d ). The ot he r paramet er s a re t h e same ast ha t in Fig .\[fig u re 8\ ] and F ig .\[fig ur e9\ ].Inset s : th e zoom ed-in im agesf rom the dashed rectangles.[] { da t a -l a bel= "fi gure12"}](F ig13 c .pdf "fi g :" ){w i dth=" 240pt "} ! [T h e dynamical measure s ynchro nizat ion of (a)(b) $S_{c}(t) $ a nd (c)(d ) th e m u tual informati on $I _{AB}(t)$i n the wh ole H ilbert s pace of $ \ b igoplus_ {N} %
\ mat hca l { H} _{N}$ for $N= 0 , 1,2, \c dots,n$ . T he trun cat ednum ber $ n=15$ for (a) (c) a nd $ n= 10$ for( b)(d). T he ot he r p arame t ers ar e the sam eas tha t in Fi g .\ [ f igur e8 \] and Fi g. \[fig ure9 \ ].Insets: the zoom ed- i n im ag es from t he dashed rec ta ngles.[]{d at a-l abel=" f i gure12"} ](Fig13d.pdf "fig:"){wi d th="240 pt" }\
S imil arly, weinv estiga tet he Mar i meas ure o f$%S _ {c}(t ) $ i n t he mixed Hil b e rtspace b y ca lculati ng $$S_{c}\left( t \ rig ht) \leq \fra c{1 }{2\ s i gm a \ l ef t ( x _{ - }\r i g ht) | "fig:"){width="240pt"}![The_dynamical measure_synchronization of (a)(b) $S_{c}(t)$_and (c)(d)_the_mutual information_$I_{AB}(t)$_in the whole_Hilbert space of_$\bigoplus_{N}%
\mathcal{H}_{N}$ for $N=0,1,2,\cdots,n$. The_truncated number $n=15$_for_(a) (c) and $n=10$ for (b)(d). The other parameters are the same as that_in_Fig.\[figure8\] and_Fig.\[figure9\]._Insets:_the zoomed-in images from the_dashed rectangles.[]{data-label="figure12"}](Fig13b.pdf "fig:"){width="240pt"}\
![The dynamical measure_synchronization of_(a)(b) $S_{c}(t)$ and (c)(d) the mutual information $I_{AB}(t)$_in_the whole Hilbert_space of $\bigoplus_{N}%
\mathcal{H}_{N}$ for $N=0,1,2,\cdots,n$. The truncated number $n=15$_for (a) (c) and $n=10$ for_(b)(d). The other_parameters_are_the same as that_in Fig.\[figure8\] and Fig.\[figure9\]. Insets: the_zoomed-in images from the dashed rectangles.[]{data-label="figure12"}](Fig13c.pdf_"fig:"){width="240pt"}![The dynamical measure synchronization of (a)(b) $S_{c}(t)$_and (c)(d) the mutual information $I_{AB}(t)$_in the whole Hilbert space_of $\bigoplus_{N}%
\mathcal{H}_{N}$_for $N=0,1,2,\cdots,n$. The truncated number_$n=15$ for (a)_(c) and_$n=10$ for (b)(d)._The other parameters are the same_as that in_Fig.\[figure8\] and Fig.\[figure9\]. Insets: the zoomed-in_images_from the dashed_rectangles.[]{data-label="figure12"}](Fig13d.pdf_"fig:"){width="240pt"}\
Similarly,_we investigate_the Mari measure_of_$%
S_{c}(t)$ in_the_mixed Hilbert space by calculating $$S_{c}\left(_t\right)_\leq \frac{1}{2\sigma \left( x_{-}\right) |
(E_1)=m(E_0)\otimes m(\ker A)^*$).
Let us consider an example when $M_0=CP^n$. For every $k\in{\bf Z}$ we construct a line bundle $\alpha _k$ over $CP^n$ in the following way. We define the total space $E_k$ of the bundle $\alpha _k$ taking quotient of ${\bf C}^{n+2}\setminus \{ 0\} $ with respect to equivalence relation $$(z_1,...,z_{n+2})\sim (\lambda z_1,...,\lambda z_{n+1}, \lambda ^kz_{n+2}).$$ The projection map $E_k\rightarrow CP^n$ is induced by the map $(z_1,...,z_{n+1},z_{n+2})\rightarrow (z_1,...,z_{n+1})$. The $(n|1)$-dimensional complex supermanifold $M_k$ corresponding to the line bundle $\alpha _k$, can be obtained from superspace $({\bf
C}^{n+1}\setminus \{ 0\} )\times {\bf C}^{0|1}$ by means of identification $$(z_1,...,z_{n+1},\theta)\sim (\lambda z_1,...,\lambda z_{n+1}, \lambda
^k\theta),\ \ \ \lambda \in {\bf C}.$$ (here $z_1,...,z_{n+1}$ are even coordinates is ${\bf C}^{n+1|1},\theta$ is an odd coordinate). To give another description of the supermanifold $M_k$ we consider a hypersurface $N_k\subset {\bf C}^{n+1|1}$, determined by the equation $$\bar {z}_1z_1+...+\bar {z}_{n+1}z_{n+1}+k\bar {\theta}\theta=c,\ \ \
c>0.$$ The restriction of standard symplectic form $\omega$ on ${\bf
C}^{n+1|1}$ to the hypersurface $N_k$ is a degenerate closed $2$-form. One can factorize $N_k$ with respect to null-vectors of this $2$-form; it is easy to check that this factorization leads to identification of points of $N_k$ given by the formula (3) with $|\lambda|=1$. This means that a manifold obtained | ( E_1)=m(E_0)\otimes m(\ker A)^*$).
Let us consider an example when $ M_0 = CP^n$. For every $ k\in{\bf Z}$ we construct a course package $ \alpha _ k$ over $ CP^n$ in the following way. We define the full space $ E_k$ of the bundle $ \alpha _ k$ take quotient of $ { \bf C}^{n+2}\setminus \ { 0\ } $ with respect to equivalence sexual intercourse $ $ (z_1,... ,z_{n+2})\sim (\lambda z_1,... ,\lambda z_{n+1 }, \lambda ^kz_{n+2}).$$ The projection map $ E_k\rightarrow CP^n$ is induce by the map $ (z_1,... ,z_{n+1},z_{n+2})\rightarrow (z_1,... ,z_{n+1})$. The $ (n|1)$-dimensional complex supermanifold $ M_k$ equate to the lineage bundle $ \alpha _ k$, can be obtained from superspace $ ({ \bf
C}^{n+1}\setminus \ { 0\ }) \times { \bf C}^{0|1}$ by means of identification $ $ (z_1,... ,z_{n+1},\theta)\sim (\lambda z_1,... ,\lambda z_{n+1 }, \lambda
^k\theta),\ \ \ \lambda \in { \bf C}.$$ (here $ z_1,... ,z_{n+1}$ are even coordinates is $ { \bf C}^{n+1|1},\theta$ is an curious coordinate). To give another description of the supermanifold $ M_k$ we consider a hypersurface $ N_k\subset { \bf C}^{n+1|1}$, determined by the equality $ $ \bar { z}_1z_1+... +\bar { z}_{n+1}z_{n+1}+k\bar { \theta}\theta = c,\ \ \
c>0.$$ The restriction of standard symplectic form $ \omega$ on $ { \bf
C}^{n+1|1}$ to the hypersurface $ N_k$ is a pervert closed $ 2$-form. One can factorize $ N_k$ with esteem to null - vectors of this $ 2$-form; it is easy to check that this factorization leads to designation of points of $ N_k$ given by the formula (3) with $ |\lambda|=1$. This means that a manifold obtained | (E_1)=m(E_0)\ltimes m(\ker A)^*$).
Let us conslder an example cyen $M_0=C'^n$. For svery $k\iv{\bf Z}$ we construct a line buidle $\alphq _k$ over $CP^n$ in the foulowing wwy. We dedine rhe total space $E_k$ of ths bunblx $\alpha _k$ takinn quotient mf ${\bf C}^{n+2}\setmings \{ 0\} $ with respect to equivalence relatyon $$(z_1,...,z_{n+2})\xil (\lambda z_1,...,\lambqa z_{m+1}, \lamgda ^kz_{n+2}).$$ The projection map $E_k\rightzrrow CK^n$ is induced by tne map $(z_1,...,z_{n+1},z_{n+2})\rightarrow (z_1,...,z_{n+1})$. The $(n|1)$-dimensional comppex supermabifoje $M_k$ correspunding to uhz line bundme $\alpha _k$, can be obtained from supexspace $({\bf
C}^{n+1}\wermijos \{ 0\} )\times {\bh C}^{0|1}$ by means of idcmtificdtion $$(z_1,...,a_{n+1},\theta)\sim (\lamnda z_1,...,\namvda z_{n+1}, \lambda
^k\theta),\ \ \ \lambda \in {\bf C}.$$ (here $z_1,...,s_{n+1}$ are even coordinqtws is ${\bf W}^{n+1|1},\thdra$ ks zn osd coogdiiate). To givs another dwscription of the sipqgkanifold $M_k$ se confiqer a hypersurface $N_k\subset {\bf C}^{n+1|1}$, deterkinsd by the equation $$\bar {z}_1z_1+...+\bar {z}_{n+1}z_{n+1}+k\bar {\theta}\tjeta=c,\ \ \
s>0.$$ The restriction of standard symplectic form $\omeca$ on ${\cf
C}^{u+1|1}$ to tfw jypersurface $N_k$ is a degenerate closed $2$-form. Ogs vak factorize $N_k$ wlth respect to nulk-vfcyjrs of this $2$-furm; it is easy to check thah this sactoeization jeadx to identification of poinrs of $N_k$ givvn bt the formula (3) witk $|\lambda|=1$. Thir mesns tnat a manifold obtained | (E_1)=m(E_0)\otimes m(\ker A)^*$). Let us consider an $M_0=CP^n$. every $k\in{\bf we construct a $CP^n$ the following way. define the total $E_k$ of the bundle $\alpha _k$ quotient of ${\bf C}^{n+2}\setminus \{ 0\} $ with respect to equivalence relation $$(z_1,...,z_{n+2})\sim z_1,...,\lambda z_{n+1}, \lambda ^kz_{n+2}).$$ The projection map $E_k\rightarrow CP^n$ is induced by the $(z_1,...,z_{n+1},z_{n+2})\rightarrow The complex $M_k$ corresponding to the line bundle $\alpha _k$, can be obtained from superspace $({\bf C}^{n+1}\setminus \{ )\times {\bf C}^{0|1}$ by means of identification $$(z_1,...,z_{n+1},\theta)\sim z_1,...,\lambda z_{n+1}, \lambda ^k\theta),\ \ \lambda \in {\bf C}.$$ $z_1,...,z_{n+1}$ even coordinates ${\bf is odd coordinate). To another description of the supermanifold $M_k$ we consider a hypersurface $N_k\subset {\bf C}^{n+1|1}$, determined by the equation {z}_1z_1+...+\bar {z}_{n+1}z_{n+1}+k\bar \ c>0.$$ restriction standard form $\omega$ on to the hypersurface $N_k$ is a One can factorize $N_k$ with respect to null-vectors this $2$-form; is easy to check that this leads to identification of points of $N_k$ given the formula (3) with $|\lambda|=1$. This means that a manifold obtained | (E_1)=m(E_0)\otimes m(\ker A)^*$).
Let us considEr an examplE when $m_0=CP^N$. FoR eVery $K\in{\bF Z}$ we construct a LIne bUndle $\alpha _k$ over $CP^n$ in thE follOwINg waY. we DefinE the totAL sPACe $E_K$ oF tHe bUnDLe $\Alpha _K$ taKing quoTient of ${\bf C}^{N+2}\seTmInus \{ 0\} $ with respECt To equivaleNce Relation $$(z_1,...,z_{n+2})\sIm (\lAmbda z_1,...,\LaMbdA Z_{n+1}, \lamBda ^Kz_{n+2}).$$ ThE projeCTion maP $E_k\rightaRrOW CP^n$ is INduced bY THe Map $(z_1,...,Z_{n+1},z_{n+2})\rightarrow (z_1,...,z_{n+1})$. tHe $(N|1)$-Dimensional comPlex suPeRMaNIFolD $M_k$ CorrespondInG to thE Line bunDLe $\ALPHa _k$, CAn be obtained fRom superspaCE $({\bf
c}^{n+1}\setmInUs \{ 0\} )\tIMes {\bf C}^{0|1}$ By meaNs OF idEntificatioN $$(z_1,...,z_{n+1},\Theta)\sim (\lAmbda z_1,...,\LAmbda z_{n+1}, \LAmbda
^k\tHeta),\ \ \ \laMbdA \in {\Bf C}.$$ (hERe $Z_1,...,z_{N+1}$ arE eVEn cOOrDinATes Is ${\bf C}^{n+1|1},\thEtA$ iS an odD cooRDINAte). TO giVe anOther Description of The SupeRManIfold $m_k$ we cOnsiDeR a hypErsurfAce $N_k\SuBset {\bf C}^{n+1|1}$, determiNed bY the equatIon $$\BaR {z}_1z_1+...+\BaR {z}_{n+1}z_{n+1}+K\Bar {\theTa}\tHetA=c,\ \ \
c>0.$$ The rEstrictIOn oF sTANDaRd symplectic form $\omEgA$ ON ${\bF
C}^{n+1|1}$ to the HypersURfAcE $n_k$ is a degEnEraTe clOSEd $2$-forM. One CAn FactorizE $N_k$ witH ReSpEct to nuLl-VectorS oF thIs $2$-fOrm; it IS easY to cheCk that thIs facTOrization leads TO identificatiON oF POiNTs of $n_k$ gIven by the foRmulA (3) With $|\LambDA|=1$. THis MEans tHat a mAnIFoLD obtained | (E_1)=m(E_0)\otimes m(\ker A)^*$).
Let u s c ons id er a n ex ample when $M_ 0 =CP^ n$. For every $k\in{\b f Z}$ w e con s tr uct a line b u nd l e $\ al ph a _ k$ ov er $C P^n $ in th e followin g w ay . We definet he total spa ce$E_k$ of the bu ndle $ \a lph a _k$tak ing q uotien t of ${ \bf C}^{n +2 } \setmi n us \{ 0 \ } $ wit h respect to equi v al e nce relation $ $(z_1, .. . ,z _ { n+2 })\ sim (\lamb da z_1, . ..,\lam b da z _ {n+ 1 }, \lambda ^k z_{n+2}).$$ The proje ct ion map $E _k\ri gh t arr ow CP^n$ is ind uced by t he map $(z_1,. . .,z_{n+ 1},z_{ n+2 })\ righ t ar ro w ( z_ 1 ,.. . ,z _{n + 1}) $. The $ (n |1 )$-di mens i o n a l co mpl ex s uperm anifold $M_k$ co rres p ond ing t o the lin ebundl e $\al pha _ k$ , can be obtain ed f rom super spa ce $( {\ bf
C} ^ {n+1}\ set min us \{ 0 \} )\ti m es{\ b f C} ^{0|1}$ by means o fi d en tificati on $$( z _1 ,. . .,z_{n+1 }, \th eta) \ s im (\ lamb d az_1,..., \lambd a z _{ n+1}, \ la mbda
^ k\ the ta) ,\ \ \ \lam bda \in{\bfC }.$$ (here $z_ 1 ,...,z_{n+1}$ ar e ev e n co ord inates is $ {\bf C}^{ n+1| 1 }, \th e ta$ i s anod d c o ordinate). To givean otherdescr iption of the supermani f o l d $M_k$we c o ns i der a hypersur face$N_k\subse t {\bf C} ^{n+1 |1}$, de terminedb y the equ ati on$$\ bar { z} _1z_1+...+\ba r {z}_ {n +1}z_{n +1} +k\bar{\t het a}\ the ta =c,\ \ \
c>0.$$ T he r es tri ction of stand ar d s ym ple cticf orm $\ omega $ on $ {\ b f
C }^{n+1| 1 }$ t o th ehy pers urf ac e $N_ k$ i s adegener ate close d $ 2 $-fo rm .One can factorize $N _k $ with res pe ctto nul l - vectorsof this $2$-form; it is easy to ch eck t hatthis fact ori zation le a ds toidenti ficat io n o f point s of $N _k $ given by t heformu la (3) with $ |\lambda|=1$. This mea ns that a man ifo ld o b t ai ned | (E_1)=m(E_0)\otimes m(\ker_A)^*$).
Let us_consider an example when_$M_0=CP^n$. For_every_$k\in{\bf Z}$_we_construct a line_bundle $\alpha _k$_over $CP^n$ in the_following way. We_define_the total space $E_k$ of the bundle $\alpha _k$ taking quotient of ${\bf C}^{n+2}\setminus_\{_0\} $_with_respect_to equivalence relation $$(z_1,...,z_{n+2})\sim (\lambda_z_1,...,\lambda z_{n+1}, \lambda ^kz_{n+2}).$$ The_projection map_$E_k\rightarrow CP^n$ is induced by the map $(z_1,...,z_{n+1},z_{n+2})\rightarrow_(z_1,...,z_{n+1})$._The $(n|1)$-dimensional complex_supermanifold $M_k$ corresponding to the line bundle $\alpha _k$,_can be obtained from superspace $({\bf
C}^{n+1}\setminus_\{ 0\} )\times_{\bf_C}^{0|1}$_by means of identification_$$(z_1,...,z_{n+1},\theta)\sim (\lambda z_1,...,\lambda z_{n+1}, \lambda
^k\theta),\ _\ \ _\lambda \in {\bf C}.$$ (here $z_1,...,z_{n+1}$ are_even coordinates is ${\bf C}^{n+1|1},\theta$ is_an odd coordinate). To give_another description_of the supermanifold $M_k$ we_consider a hypersurface_$N_k\subset {\bf_C}^{n+1|1}$, determined by_the equation $$\bar {z}_1z_1+...+\bar {z}_{n+1}z_{n+1}+k\bar {\theta}\theta=c,\_ \ _\
c>0.$$ The restriction of standard symplectic_form_$\omega$ on ${\bf
C}^{n+1|1}$_to_the_hypersurface $N_k$_is a degenerate_closed_$2$-form. One_can_factorize $N_k$ with respect to null-vectors_of_this $2$-form; it is easy to check_that this factorization leads_to_identification of points of_$N_k$ given by the formula_(3) with $|\lambda|=1$. This means that_a manifold_obtained |
inctions derived from a Miller & Scalo IMF and a Salpeter IMF, respectively. Therefore, the small GRB host sample does not allow us to determine whether or not GRB hosts have smaller intrinsic extinctions than young field starburst galaxies. Likewise, comparing GRB host extinctions with those derived for all types of field galaxies does not allow us to determine whether they have different extinction distributions.
Additional tests {#ad_comp}
----------------
As mentioned in Sect. \[sed\] the SED fit of the host of GRB 000418 could indicate a somewhat larger extinction ($A_V=1.4$). We therefore investigated the impact of changing this particular extinction value on the derived probabilities for the extinction corrected specific SFRs. For a Salpeter IMF, the derived 2D K–S probabilities for the same parent distribution are 0.03 and 0.04 for 0.2 Gyr and 0.1 Gyr old HDF starburst, respectively, while Table \[tab:ksprob\] gives 0.004 and 0.04. We therefore conclude that the uncertainty in the GRB 000418 host extinction has little impact on the results.
We also investigated the possibility that other galaxy types might have similar SFRs as the GRB hosts. Selecting young HDF galaxies classified as other spectral types than starbursts, i.e. irregular galaxies or spirals with $t<0.1$ Gyr, gives a 2D K–S probability of 0.002 for the same parent distribution. We note that for the theoretical templates the differences between the templates at such young ages are small. However, comparing the specific extinction corrected SFRs with those of ellipticals, without any age constraint, the probability is also small (2$\times10^{-5}$) which is not surprising as old elliptical galaxies do not have any star formation.
Because of possible differences between bright and faint HDF galaxies, an additional brightness requirement on the HDF galaxies was therefore imposed. We selected HDF galaxies with absolute $B$ band magnitudes as bright or brighter than GRB hosts ($M_B < -18.08$) and the resulting K–S probabilities for the various tests are listed in column 4 and 5 in Table \[tab:ksprob\]. The probabilities now appear to be much smaller than without the brightness selection (apart from the M&S IMF based extinction corrected tests). This is not due to the smaller number of comparison galaxies, but arises because the | inctions derived from a Miller & Scalo IMF and a Salpeter IMF, respectively. Therefore, the small GRB host sample distribution does not leave us to determine whether or not GRB hosts get minor intrinsic extinctions than young playing field starburst galax. Likewise, comparing GRB host extinctions with those derive for all types of field galax does not admit us to determine whether they get different extinction distributions.
extra tests { # ad_comp }
----------------
As mentioned in Sect. \[sed\ ] the SED fit of the master of ceremonies of GRB 000418 could indicate a somewhat big extinction ($ A_V=1.4 $). We therefore investigate the impact of changing this particular extinction value on the derived probabilities for the extinction adjust specific SFRs. For a Salpeter IMF, the derived 2D K – S probabilities for the same parent distribution are 0.03 and 0.04 for 0.2 Gyr and 0.1 Gyr old HDF starburst, respectively, while Table \[tab: ksprob\ ] gives 0.004 and 0.04. We therefore conclude that the uncertainty in the GRB 000418 host extinction has little impact on the results.
We also investigated the possibility that early galax types might have similar SFRs as the GRB hosts. choose unseasoned HDF galaxies classified as other spectral types than starbursts, i.e. irregular galaxies or spiral with $ t<0.1 $ Gyr, gives a 2D K – S probability of 0.002 for the same parent distribution. We note that for the theoretical templates the dispute between the templates at such young ages are small. However, comparing the specific extinction corrected SFRs with those of ellipticals, without any historic period constraint, the probability is also small (2$\times10^{-5}$) which is not surprising as old elliptical galaxy do not have any star formation.
Because of potential differences between undimmed and faint HDF galaxies, an additional brightness requirement on the HDF galax was therefore imposed. We choose HDF galaxies with absolute $ B$ band magnitudes as bright or bright than GRB hosts ($ M_B < -18.08 $) and the result K – S probability for the various trial are listed in column 4 and 5 in Table \[tab: ksprob\ ]. The probability now appear to be much smaller than without the brightness choice (apart from the M&S IMF based extinction correct tests). This is not due to the belittled number of comparison galaxies, but arises because the | inchions derived from a Miluer & Scalo IMF cbd a Selpeter IMF, resoectively. Therefore, the smalp TRB hist sample does not aluow us to determibe wiether or not GRU hosts mcve sjwllex mntrinsic extingtions than young field sdafbbrst galaxies. Likewise, comparing GRB host ectlnctions with jhose qeribvd for all types of field galaxjes doev not allow ux to determine whether thej hage different extinftion distrubutyins.
Additionau tests {#ad_bmmp}
----------------
As mentjoned in Sect. \[sed\] the SED fit of the kost of GRB 000418 ciulf indicate a somerhat larger cqtinctimn ($A_V=1.4$). Wr therefore inyestijatee the impact of changmng this particular gxtinction vclue on the derived peovabilhtiev fof thd estmncfion clrrxcted speciric SFRs. Foe a Salpeter IMF, tht dqgoved 2D K–S prkbabilytyes for the same parent distribution art 0.03 ans 0.04 for 0.2 Gyr and 0.1 Gyr old YDF starburst, respectlvely, whije Table \[tab:ksprob\] gives 0.004 and 0.04. We therefore concluge thet tht mkcergqijty in the GRB 000418 host extinction has little impwdt ok the results.
We clso investigatec hhr possibility jhat otksr galaxy types mighh have fimilqr SFRs af thr GRB hosts. Selecting young HDF galaxief classified as othzr spectral cypes jhan syarbursts, i.e. irregular yalaxiss or spiraps with $t<0.1$ Eyr, gives a 2D K–S prpbdbility of 0.002 for the same pwrent disvribucion. We vote that sor the thforetlwal templates the fiffetences between tje templates at such young ages are small. Howgves, cmmparing the xpecific extigction correctgd SFRs wnth thuse of ellppticals, xithout any wge constraind, the probabinity is wlso smaol (2$\timer10^{-5}$) which is not xurprisiny as ild elliptical galexier do not have anv wtar formation.
Brcajse ov 'ossifne differencas bdtwdrn brkght and falnt HDF galaxies, an additiotal grightness requirekekt on the HDF galwxies was thetefore imposed. We delecved HDH galacief with absolute $B$ band magnituses as brlghb or brighter thak GRF hosts ($M_B < -18.08$) and the resulting K–S probabilities fir the various tesrs are listed in cokmmn 4 and 5 in Tablq \[tab:ksprot\]. The probabilities bow appear to be kuch smaller than witgout tve brlghtness selection (apart from the M&S IMF based extinction corrected tests). This ms not due to the smalnex nbmber os col'arison galaxies, nut arises because the | inctions derived from a Miller & Scalo a IMF, respectively. the small GRB us determine whether or GRB hosts have intrinsic extinctions than young field starburst Likewise, comparing GRB host extinctions with those derived for all types of field does not allow us to determine whether they have different extinction distributions. Additional {#ad_comp} As in \[sed\] the SED fit of the host of GRB 000418 could indicate a somewhat larger extinction We therefore investigated the impact of changing this extinction value on the probabilities for the extinction corrected SFRs. a Salpeter the 2D probabilities for the parent distribution are 0.03 and 0.04 for 0.2 Gyr and 0.1 Gyr old HDF starburst, respectively, while \[tab:ksprob\] gives 0.04. We conclude the in the GRB extinction has little impact on the investigated the possibility that other galaxy types might similar SFRs the GRB hosts. Selecting young HDF classified as other spectral types than starbursts, i.e. galaxies or spirals with $t<0.1$ Gyr, gives a 2D K–S probability of 0.002 for the distribution. We note that the theoretical templates differences the at young ages small. However, comparing the specific extinction corrected SFRs with those of without any age constraint, the probability is also small (2$\times10^{-5}$) not as old elliptical do not have any formation. of possible differences between faint galaxies, requirement the galaxies was therefore imposed. selected HDF galaxies with absolute band magnitudes as bright ($M_B < -18.08$) and the resulting K–S probabilities the various tests are listed in column and 5 in Table \[tab:ksprob\]. The probabilities now appear to be much than without selection (apart from the M&S IMF based extinction tests). This is not to the smaller number of comparison galaxies, but arises the | inctions derived from a MilleR & Scalo IMF aNd a SaLpeTer iMf, resPectIvely. Therefore, THe smAll GRB host sample does noT alloW uS To deTErMine wHether oR NoT grB hOsTs HavE sMAlLer inTriNsic extInctions thAn yOuNg field starbURsT galaxies. LIkeWise, comparinG GRb host eXtIncTIons wIth Those DeriveD For all Types of fiElD GalaxiES does noT ALlOw us To determine whetheR ThEY have different ExtincTiON dISTriButIons.
AdditiOnAl tesTS {#ad_comp}
----------------
aS mENTIonED in Sect. \[sed\] the sED fit of the HOst Of GRB 000418 cOuLd iNDicate A someWhAT laRger extinctIon ($A_v=1.4$). We therefOre invEStigateD The impaCt of chAngIng This PArTiCulAr EXtiNCtIon VAluE on the deRiVeD probAbilITIES for The ExtiNctioN corrected speCifIc SFrS. FoR a SalPeter iMF, tHe DerivEd 2D K–S pRobabIlIties for the same PareNt distribUtiOn Are 0.03 AnD 0.04 for 0.2 GYR and 0.1 GyR olD HDf starbuRst, respECtiVeLY, WHiLe Table \[tab:ksprob\] giVeS 0.004 ANd 0.04. we therefOre conCLuDe THat the unCeRtaInty IN The GRb 000418 hosT ExTinction Has litTLe ImPact on tHe ResultS.
WE alSo iNvestIGateD the poSsibilitY that OTher galaxy typeS Might have simiLAr sfrs AS the gRB Hosts. SelectIng yOUng HdF gaLAxIes CLassiFied aS oTHeR Spectral types than stArBursts, I.e. irrEgular galaxieS or spirals WITH $t<0.1$ Gyr, givEs a 2D k–s pRObability of 0.002 for The saMe parent diSTributioN. We noTe that foR the theorETIcal tempLatEs tHe dIffEREnCes between the TEMplaTeS at such YouNg ages aRe sMalL. HoWevEr, Comparing The speciFiC eXtInCtiOn corREcted SFRS wIth ThOse Of ellIPticalS, withOut aNy AgE ConStraint, THe PRObabIlItY is aLso SmAll (2$\tiMes10^{-5}$) wHIch Is not suRprising aS olD ElliPtIcAl galaxIes do not have aNy Star formatIoN.
BeCause oF POssible dIfferences between bright ANd faint hDF GalaxIes, aN additionAl bRightnEss REquireMent on The HDf gAlaXIEs was THErEfoRe Imposed. We sELEctEd HDF GaLaxiEs with aBsolute $B$ band magnitUDes As bright or briGhtEr thAN gRb hoSTs ($m_b < -18.08$) anD tHE reSULting K–S probabilIties for thE vARiOus tests arE LisTeD in coluMn 4 and 5 in table \[TAb:ksproB\]. The probaBilities nOw AppeAR To bE much smallEr than wiThout the bRIghtnESs SelecTioN (apart FrOm tHe M&S ImF baseD ExtInctiOn corrEcTed tesTs). ThiS iS not due tO the smaller number of compArison GalaxIes, But arises BecAUse The | inctions derived from a Mi ller & Sca lo IM F a ndaSalp eter IMF, respecti v ely. Therefore, the smallGRB h os t sam p le does not al l ow u s t ode ter mi n ewheth eror notGRB hostshav esmaller intr i ns ic extinct ion s than young fi eld st ar bur s t gal axi es. L ikewis e , comp aring GRB h o st ext i nctions w it h th ose derived for a l lt ypes of fieldgalaxi es do e s no t a llow us to d eterm i ne whet h er t h eyh ave different extinction dis tribut io ns.
Addit ional t e sts {#ad_comp}
--- --------- ----
A s menti o ned inSect.\[s ed\ ] th e S ED fi to f t h ehos t of GRB 000 41 8could ind i c a t e asom ewha t lar ger extinctio n ( $A_V = 1.4 $). W e the refo re inve stigat ed th eimpact of chang ingthis part icu la r e xt incti o n valu e o n t he deri ved pro b abi li t i e sfor the extinction c o r re cted spe cificS FR s. For a Sa lp ete r IM F , thederi v ed 2D K–Sprobab i li ti es forth e same p are ntdistr i buti on are 0.03 an d 0.0 4 for 0.2 Gyr a n d 0.1 Gyr old HD F st a rbur st, respective ly,w hile Tab l e\[t a b:ksp rob\] g i ve s 0.004 and 0.04. We t herefo re co nclude that t he uncerta i n t y in the GRB 00 0 418 host extin ction has littl e impacton th e result s.
We al s o investi gat edthe po s s ib ility that ot h e r ga la xy type s m ight ha vesim ila r S FR s as theGRB host s. S el ec tin g you n g HDF ga la xie scla ssifi e d as o therspec tr al typ es than st a r burs ts ,i.e. ir re gular gal a xie s or sp irals wit h $ t <0.1 $Gy r, give s a 2D K–S pr ob ability of 0 .00 2 fort h e same p arent distribution. Wen ote tha t f or th e th eoretical te mplate s t h e diff erence s bet we ent h e tem p l at esat such youn g age s are s mall . Howev er, comparing thes pec ific extincti oncorr e c te d S F Rs wit ht hos e of ellipticals, without a ny ag e constrai n t,th e proba bilityis al s o small (2$\time s10^{-5}$ )whic h isnot surpri sing asold ellip t icalg al axies do not h av e a ny st ar for m ati on.
Becaus eof pos sible d ifferenc es between bright and f aint H DF ga lax ies, an a ddi t ion al bright ness requireme ntonthe H DFg alaxi es w a sthe r efore imp o sed. We s e le cte d HD F galaxiesw i t h a bsolu te$ B$ ban d ma gnitudes as brigh t or brighter t hanG R B h ost s ($M _B < -18.08$) an d t he r esulting K –S probabil ities fo rt he va rioustestsare lis t e di n colu mn 4 an d 5 in Ta ble \ [ tab:ksp ro b\ ] . Theprob ab ilitie s nowa ppea r to be much small er th a n with o utthe b ri ghtness sele ction (apa rt from the M&S I MF b asedextinct io n corr ect ed tests). T h is is not dueto thesm alle r n umberof c o m paris on g al axi es, but a r i se s b ec a use the | inctions derived_from a_Miller & Scalo IMF_and a_Salpeter_IMF, respectively._Therefore,_the small GRB_host sample does_not allow us to_determine whether or_not_GRB hosts have smaller intrinsic extinctions than young field starburst galaxies. Likewise, comparing GRB_host_extinctions with_those_derived_for all types of field_galaxies does not allow us_to determine_whether they have different extinction distributions.
Additional tests {#ad_comp}
----------------
As_mentioned_in Sect. \[sed\] the_SED fit of the host of GRB 000418 could indicate_a somewhat larger extinction ($A_V=1.4$). We_therefore investigated the_impact_of_changing this particular extinction_value on the derived probabilities for_the extinction corrected specific SFRs. For_a Salpeter IMF, the derived 2D K–S_probabilities for the same parent distribution_are 0.03 and 0.04 for_0.2 Gyr and_0.1 Gyr old HDF starburst, respectively,_while Table \[tab:ksprob\] gives_0.004 and_0.04. We therefore_conclude that the uncertainty in the_GRB 000418 host extinction_has little impact on the results.
We_also_investigated the possibility_that_other_galaxy types_might have similar_SFRs_as the_GRB_hosts. Selecting young HDF galaxies classified_as_other spectral types than starbursts, i.e. irregular_galaxies or spirals with_$t<0.1$_Gyr, gives a 2D_K–S probability of 0.002 for_the same parent distribution. We note_that for_the theoretical_templates the differences between the templates at such young ages are_small. However, comparing the specific extinction_corrected SFRs with those_of ellipticals,_without_any age constraint,_the_probability is_also small (2$\times10^{-5}$) which is not surprising_as old_elliptical galaxies do not have any_star formation.
Because of possible_differences_between bright and faint HDF galaxies,_an additional brightness requirement on the_HDF galaxies was therefore imposed._We_selected_HDF galaxies with absolute $B$_band magnitudes as bright or brighter_than GRB hosts_($M_B < -18.08$) and the resulting K–S_probabilities_for the various tests are listed_in_column 4 and 5 in Table \[tab:ksprob\]._The_probabilities_now appear to be much_smaller than without the brightness selection_(apart from the M&S IMF based extinction corrected tests)._This is not_due to the smaller number_of_comparison_galaxies, but arises because the |
2^{0\,0} \sim \Delta M^2_{\rm susy},$[^2] as explained in [@Alvarez-Gaume:2003]. The presence of such $\Pi_2$ effects will lead to unacceptable pathologies such as Lorentz-noninvariant dispersion relations giving mass to only one of the polarisations of the trace-U(1) gauge field, leaving the other polarisation massless.
The presence of the UV/IR effects in the trace-U(1) factors makes it pretty clear that a simple noncommutative U(1) theory taken on its own has nothing to do with ordinary QED. The low-energy theory emerging from the noncommutative U(1) theory will become free at $k^2 \to 0$ (rather than just weakly coupled) and in addition will have other pathologies [@Khoze:2004zc; @Khoze:2000sy; @Hollowood:2001ng; @Alvarez-Gaume:2003]. However, one would expect that it is conceivable to embed a commutative $\textrm{SU}(N)$ theory, such as e.g. QCD or the weak sector of the Standard Model into a supersymmetric noncommutative theory in the UV, but some extra care should be taken with the QED U(1) sector [@Khoze:2004zc]. We will show that the only realistic way to embed QED into noncommutative settings is to recover the electromagnetic U(1) from a [*traceless*]{} diagonal generator of some higher $\textrm{U}(N)$ gauge theory. So it seems that in order to embed QED into a noncommutative theory one should learn how to embed the whole Standard Model [@Khoze:2004zc]. We will see, however, that the additional trace-U(1) factors remaining from the noncommutative $\textrm{U}(N)$ groups will make the resulting low-energy theories unviable (at least for the general class of models considered in this paper).
In order to proceed we would like to disentangle the mass-effects due to the Higgs mechanism from the mass-effects due to non-vanishing $\Pi_2.$ Hence we first set $\Pi_2= 0$ (this can be achieved by starting with an exactly supersymmetric theory). It is then straightforward to show (see Sec. \[prove\]) that the Higgs mechanism alone cannot remove all of the trace-U(1) factors from the massless theory. More | 2^{0\,0 } \sim \Delta M^2_{\rm susy},$[^2 ] as explained in [ @Alvarez - Gaume:2003 ]. The presence of such $ \Pi_2 $ effects will lead to impossible pathology such as Lorentz - noninvariant dispersion relations give multitude to only one of the polarisations of the tracing - U(1) bore field, leaving the other polarisation massless.
The presence of the ultraviolet / IR effects in the trace - U(1) factors makes it reasonably clear that a simple noncommutative U(1) hypothesis taken on its own have nothing to make with ordinary QED. The low - energy hypothesis emerging from the noncommutative U(1) theory will become complimentary at $ k^2 \to 0 $ (rather than merely weakly coupled) and in addition will have other pathologies [ @Khoze:2004zc; @Khoze:2000sy; @Hollowood:2001ng; @Alvarez - Gaume:2003 ]. However, one would ask that it is conceivable to embed a commutative $ \textrm{SU}(N)$ theory, such as e.g. QCD or the weak sector of the Standard Model into a supersymmetric noncommutative theory in the UV, but some extra care should be taken with the QED U(1) sector [ @Khoze:2004zc ]. We will show that the only realistic way to embed QED into noncommutative settings is to recover the electromagnetic U(1) from a [ * traceless * ] { } diagonal generator of some high $ \textrm{U}(N)$ gauge theory. indeed it seem that in order to embed QED into a noncommutative theory one should determine how to embed the whole Standard Model [ @Khoze:2004zc ]. We will see, however, that the additional touch - U(1) factors remaining from the noncommutative $ \textrm{U}(N)$ groups will make the resulting low - department of energy theories unviable (at least for the general class of models considered in this paper).
In order to continue we would like to disentangle the mass - effects due to the Higgs mechanism from the mass - effect due to non - vanishing $ \Pi_2.$ Hence we first set $ \Pi_2= 0 $ (this can be achieved by starting with an exactly supersymmetric theory). It is then straightforward to express (see Sec. \[prove\ ]) that the Higgs mechanism entirely cannot remove all of the trace - U(1) factors from the massless theory. More | 2^{0\,0} \sil \Delta M^2_{\rm susy},$[^2] as explxined in [@Alvaree-Gqume:2003]. Tie pressnce of ruch $\Pi_2$ effects will lead to ubacceknable pathologies sucf as Lorejtz-nonincarient dispersion rxmations givinf masv to only one on the polarhsations of tha grcce-U(1) gauge field, leaving the other pjlarisayiln massless.
The prexqnce of the UV/IR effects in the trace-U(1) factorv makes it prrtty clear that a simple nlncolmutative U(1) theory taken on ija orb has nothine to do wiuh ordinary SED. The low-energy theory emergivg frpm the nonxonmuhdtive U(1) theiry wpll become frcv at $k^2 \do 0$ (ratner than just eeanly coupled) and in additmon will have other kathologiev [@Ihoze:2004zc; @Khoze:2000sy; @Hiloowoog:2001ng; @Dlvafwz-Gxumt:2003]. Hpwsver, oje xould expecf that it iw conceivable to emneq a commutativs $\textwm{FU}(N)$ theory, such as e.g. QCD or the weak stctor of the Standard Model unto a supersymmetric noncommueative theory in the UV, but some extra care shoulg be vayen wlth gye QED U(1) sector [@Khoze:2004zc]. We will show that the onjg teslistic way to embed QED onho goncommutativg settiufs is to recover the electrjmagnwtic U(1) frjm a [*traceless*]{} diagonal generatir of some hpghee $\textrm{U}(N)$ gauge tkeory. So it reemx thay in order to embed QED intk a noncommktative tgdory one should ueagn hmw to embed the whole Stanqard Modeo [@Khove:2004zc]. We dill see, hjwever, thah the additional trace-U(1) vactots remdining frol the noncommutative $\textrm{U}(N)$ gcpups will male thv resultiug low-cnergy theories unviable (at lgast for che geveral clasa of movels considewed in this pdker).
In order tm proceeq we wouod like go disentangle the mass-vfyects due to the Higgs mecheniso from the mass-eyyexts due to non-vsnirhigg $\Pm_2.$ Hensa we first sat $\Pk_2= 0$ (gnis cxn be achieyed by xtarting with an exawtly supersymmetric throvy). It is jhen strayghtforward tp show (see Sec. \[proge\]) thet the Higgx mgchanism alone cannot remove alm of the hrage-U(1) factors fwom bhe iassless tkeory. More | 2^{0\,0} \sim \Delta M^2_{\rm susy},$[^2] as explained The of such effects will lead Lorentz-noninvariant relations giving mass only one of polarisations of the trace-U(1) gauge field, the other polarisation massless. The presence of the UV/IR effects in the trace-U(1) makes it pretty clear that a simple noncommutative U(1) theory taken on its has to with QED. The low-energy theory emerging from the noncommutative U(1) theory will become free at $k^2 \to (rather than just weakly coupled) and in addition have other pathologies [@Khoze:2004zc; @Hollowood:2001ng; @Alvarez-Gaume:2003]. However, one would that is conceivable embed commutative theory, such as QCD or the weak sector of the Standard Model into a supersymmetric noncommutative theory in the UV, some extra be taken the U(1) [@Khoze:2004zc]. We will the only realistic way to embed settings is to recover the electromagnetic U(1) from [*traceless*]{} diagonal of some higher $\textrm{U}(N)$ gauge theory. it seems that in order to embed QED a noncommutative theory one should learn how to embed the whole Standard Model [@Khoze:2004zc]. We however, that the additional factors remaining from noncommutative groups make resulting low-energy unviable (at least for the general class of models considered in paper). In order to proceed we would like to disentangle due the Higgs mechanism the mass-effects due to $\Pi_2.$ we first set $\Pi_2= can achieved an supersymmetric It is then straightforward show (see Sec. \[prove\]) that Higgs mechanism alone cannot factors from the massless theory. More | 2^{0\,0} \sim \Delta M^2_{\rm susy},$[^2] as explaineD in [@Alvarez-gaume:2003]. the PreSeNce oF sucH $\Pi_2$ effects will LEad tO unacceptable pathologiEs sucH aS loreNTz-NoninVariant DIsPERsiOn ReLatIoNS gIving MasS to only One of the poLarIsAtions of the tRAcE-U(1) gauge fieLd, lEaving the othEr pOlarisAtIon MAssleSs.
THe preSence oF The UV/Ir effects iN tHE trace-u(1) Factors MAKeS it pRetty clear that a siMPlE Noncommutative u(1) theorY tAKeN ON itS owN has nothinG tO do wiTH ordinaRY Qed. tHe lOW-energy theory Emerging froM The NoncomMuTatIVe U(1) theOry wiLl BEcoMe free at $k^2 \to 0$ (RathEr than jusT weaklY Coupled) ANd in addItion wIll HavE othER pAtHolOgIEs [@KHOzE:2004zc; @kHozE:2000sy; @HolloWoOd:2001Ng; @AlvArez-gAUME:2003]. HowEveR, one Would Expect that it iS coNceiVAblE to emBed a cOmmuTaTive $\tExtrm{Su}(N)$ theOrY, such as e.g. QCD or tHe weAk sector oF thE STanDaRd ModEL into a SupErsYmmetriC noncomMUtaTiVE THeOry in the UV, but some eXtRA CaRe should Be takeN WiTh THe QED U(1) seCtOr [@KHoze:2004ZC]. we wilL shoW ThAt the onlY realiSTiC wAy to embEd qED intO nOncOmmUtatiVE setTings iS to recovEr the ELectromagnetic u(1) From a [*tracelesS*]{} DiAGOnAL genEraTor of some hiGher $\TExtrM{U}(N)$ gAUgE thEOry. So It seeMs THaT In order to embed QED inTo A noncoMmutaTive theory one Should learN HOW to embed The wHOlE standard Model [@KHoze:2004zC]. We will see, HOwever, thAt the AdditionAl trace-U(1) fACTors remaIniNg fRom The NONcOmmutative $\texTRM{U}(N)$ gRoUps will MakE the resUltIng Low-EneRgY theories Unviable (At LeAsT fOr tHe genERal class Of ModElS coNsideREd in thIs papEr).
In OrDeR To pRoceed wE WoULD likE tO dIsenTanGlE the mAss-eFFecTs due to The Higgs mEchANism FrOm The mass-Effects due to nOn-Vanishing $\PI_2.$ HEncE we firST Set $\Pi_2= 0$ (thiS can be achieved by startinG With an eXacTly suPersYmmetric tHeoRy). It is TheN StraigHtforwArd to ShOw (sEE sec. \[prOVE\]) tHat ThE Higgs mechANIsm Alone CaNnot Remove aLl of the trace-U(1) factoRS frOm the massless TheOry. MORE | 2^{0\,0} \sim \Delta M^2_{ \rm susy}, $[^2] as ex pl aine d in [@Alvarez-Gau m e:20 03]. The presence of s uch $ \P i _2$e ff ectswill le a dt o un ac ce pta bl e p athol ogi es such as Lorent z-n on invariant di s pe rsion rela tio ns giving ma ssto onl yone of th e p olari sation s of th e trace-U (1 ) gauge field,l e av ingthe other polaris a ti o n massless.
T he pre se n ce o f t heUV/IR effe ct s int he trac e -U ( 1 ) fa c tors makes it pretty cle a r t hat asi mpl e nonco mmuta ti v e U (1) theorytake n on itsown ha s nothin g to dowith o rdi nar y QE D .Th e l ow - ene r gy th e ory emergin gfr om th e no n c o m muta tiv e U( 1) th eory will bec ome fre e at $k^2 \to0$ ( ra therthan j ust w ea kly coupled) an d in addition wi ll ha ve othe r patho log ies [@Kho ze:2004 z c;@K h o z e: 2000sy; @Hollowood :2 0 0 1n g; @Alva rez-Ga u me :2 0 03]. How ev er, one w ouldexpe c tthat itis con c ei va ble toem bed aco mmu tat ive $ \ text rm{SU} (N)$ the ory,s uch as e.g. QC D or the weaks ec t o ro f th e S tandard Mod el i n to a sup e rs ymm e tricnonco mm u ta t ive theory in the U V, but s ome e xtra care sho uld be tak e n with the QED U( 1 ) sector [@Kho ze:20 04zc]. Wew ill show that the onl y realist i c way toemb edQED in t o n oncommutative s etti ng s is to re cover t heele ctr oma gn etic U(1) from a[* tr ac el ess *]{}d iagonalge ner at orof so m e high er $\ text rm {U } (N) $ gauge th e o ry.So i t se ems t hat i n or d erto embe d QED int o a nonc om mu tativetheory one sh ou ld learn h ow to embed t he whole Standard Model [@Khoze : 2004zc] . W e wil l se e, howeve r,that t hea dditio nal tr ace-U (1 ) f a c torsr e ma ini ng from then o nco mmuta ti ve $ \textrm {U}(N)$ groups wil l ma ke the result ing low - e ne rgy th e ori es unv i a ble (at least f or the gen er a lclass of m o del sconside red inthisp aper).
In order to proce ed wew o uld like to d isentang le the ma s s-eff e ct s due to the H ig gsmecha nism f r omthe m ass-ef fe cts du e tono n-vanish ing $\Pi_2.$ Hence we f irst s et $\ Pi_ 2= 0$ (th isc anbe achiev ed b y starting wi than ex act l y sup ersy m me tri c theo ry). It is the n s tra i g ht forward tos h o w ( see S ec. \[prov e\]) that the Higgs m e chanism alonecann o t re mov e all o f the trace-U( 1)fa c t ors from t he massless theory. M o re | 2^{0\,0} \sim_\Delta M^2_{\rm_susy},$[^2] as explained in_[@Alvarez-Gaume:2003]. The_presence_of such_$\Pi_2$_effects will lead_to unacceptable pathologies_such as Lorentz-noninvariant dispersion_relations giving mass_to_only one of the polarisations of the trace-U(1) gauge field, leaving the other polarisation_massless.
The_presence of_the_UV/IR_effects in the trace-U(1) factors_makes it pretty clear that_a simple_noncommutative U(1) theory taken on its own has_nothing_to do with_ordinary QED. The low-energy theory emerging from the noncommutative_U(1) theory will become free at_$k^2 \to 0$_(rather_than_just weakly coupled) and_in addition will have other pathologies_ [@Khoze:2004zc; @Khoze:2000sy; @Hollowood:2001ng; @Alvarez-Gaume:2003]. However, one_would expect that it is conceivable to_embed a commutative $\textrm{SU}(N)$ theory, such_as e.g. QCD or the_weak sector_of the Standard Model into_a supersymmetric noncommutative_theory in_the UV, but_some extra care should be taken_with the QED_U(1) sector [@Khoze:2004zc]. We will show that_the_only realistic way_to_embed_QED into_noncommutative settings is_to_recover the_electromagnetic_U(1) from a [*traceless*]{} diagonal generator_of_some higher $\textrm{U}(N)$ gauge theory. So it_seems that in order_to_embed QED into a_noncommutative theory one should learn_how to embed the whole Standard_Model [@Khoze:2004zc]. We_will see,_however, that the additional trace-U(1) factors remaining from the noncommutative $\textrm{U}(N)$_groups will make the resulting low-energy_theories unviable (at least_for the_general_class of models_considered_in this_paper).
In order to proceed we would like_to disentangle_the mass-effects due to the Higgs_mechanism from the mass-effects_due_to non-vanishing $\Pi_2.$ Hence we first_set $\Pi_2= 0$ (this can be_achieved by starting with an_exactly_supersymmetric_theory). It is then straightforward_to show (see Sec. \[prove\]) that_the Higgs mechanism_alone cannot remove all of the trace-U(1)_factors_from the massless theory. More |
general graphs we show that fixed response dynamics converge to a steady state. On the complete graph we show that best response dynamics also converge to a steady state. Both of these results hold regardless of the initial conditions (i.e., the agents’ estimators at time $t=0$).
We show that the steady state of best response dynamics is not necessarily optimal; there exist fixed response dynamics in which the agents converge to estimators which all have lower variance then the estimators of the steady state of best response dynamics. This shows that every agent can do better than the result of best response by following a socially optimal rule; thus a certain [*price of anarchy*]{} is to be paid when agents choose the action that maximizes their short term gain.
Finally, we show that in the penultimate prediction model, for the complete graph, the agents learn estimators which are the optimal (in the minimum variance sense) amongst all linear estimators, and thus outperform those of fixed and best response dynamics.
We define a notion of “socially asymptotic learning”: A model has this property when the variance of the agents’ steady-state estimators tends towards the information-theoretical optimum with the number of agents. We show that the penultimate prediction model exhibits socially asymptotic learning on the complete graph, while best response and fixed response dynamics fail to do the same.
Previous work
=============
Our model is an elaboration of models studied by DeMarzo, Vayanos and Zwiebel [@DeMarzo:03], as well as Mossel and Tamuz [@mossel2010iterative; @MosselTamuz10:arxiv]. There, the state $S$ is a fixed number picked at time $t=0$, and each agent receives a single measurement of it. The process thereafter is deterministic, with each agent, at each iteration, recalculating its estimate of $S$ based on its observation of its neighbors’ estimates.
In [@MosselTamuz10:arxiv] it is shown that if the agents calculate the minimum variance unbiased linear estimator (MVULE) at every turn (remembering all of their observations) then all the agents converge to the optimal estimator of $S$, i.e. the average of the original measurements. Furthermore, this happens in time that is at most $n
\cdot d$, where $d$ is the diameter of the graph.
When agents calculate estimates that are only based on their | general graphs we show that fixed reception moral force converge to a steady state. On the accomplished graph we testify that best response moral force also converge to a steady state. Both of these consequence hold regardless of the initial conditions (i.e., the agent ’ estimators at time $ t=0 $).
We read that the regular state of best response dynamics is not necessarily optimum; there exist fixed response dynamics in which the agent converge to estimators which all have depleted discrepancy then the estimators of the steady state of best response moral force. This shows that every agent can do better than the result of best reception by following a socially optimal rule; thus a certain [ * price of anarchy * ] { } is to be paid when agents choose the action that maximizes their inadequate term gain.
last, we picture that in the penultimate prediction model, for the complete graph, the agents determine estimators which are the optimal (in the minimum variance common sense) amongst all linear estimators, and thus outperform those of fixed and best response dynamics.
We define a notion of “ socially asymptotic learning ”: A model receive this property when the variation of the agents ’ steady - state estimators tends towards the data - theoretical optimum with the number of agents. We show that the penultimate prediction model exhibits socially asymptotic learning on the complete graph, while best response and fix response moral force fail to do the same.
Previous work
= = = = = = = = = = = = =
Our model is an amplification of models study by DeMarzo, Vayanos and Zwiebel [ @DeMarzo:03 ], as well as Mossel and Tamuz [ @mossel2010iterative; @MosselTamuz10: arxiv ]. There, the department of state $ S$ is a fixed number picked at prison term $ t=0 $, and each agent receive a single measurement of it. The process thereafter is deterministic, with each agent, at each iteration, recalculating its estimate of $ S$ based on its observation of its neighbors ’ estimate.
In [ @MosselTamuz10: arxiv ] it is shown that if the agents account the minimum variance unbiased linear estimator (MVULE) at every turn (commemorate all of their observations) then all the agents converge to the optimal estimator of $ S$, i.e. the average of the original measurements. Furthermore, this happens in clock time that is at about $ n
\cdot d$, where $ d$ is the diameter of the graph.
When agents calculate estimates that are only based on their | gejeral graphs we show thau fixed response btnamicv convsrge to x steady state. On the compleve geaph qe show that best respunse dynalics alsi coiverge to a steavg state. Both kn thevx results hold tegardless ox the initial wovdntions (i.e., the agents’ estimators at tyme $t=0$).
We sjow that the sjeady ftats of best response dynamics is not necesserily optimal; tnere exist fixed response fynalics in which the wgents convgdge ro estimatorr which all have lower variance then the estimators ow the steady stqtw ov best respoise dygamics. This shows thdt everu agent can do bevter than the result of bxst response by folljwing a smcnally optimal rule; thys a cettain [*prizw ow ahacchg*]{} is tl bx paid when agents choise the action that mwqomizes their short tqrm gain.
Finally, we show that in the penlltijate prediction model, fir the complete graph, the agenes learn estimators which are the optimal (in the kinimnm vaxlwvxe sense) amongst all linear estimators, and thus kuupegform those of fiwed and best respomsf cinamics.
We defive a notikn of “socially asylptotic learbing”: A moqel nas this property when the cariance of nhe qgents’ steady-state estimators tencs toeards the information-thzoretidal optimum with the vumber of agents. We svow that the penultimate pwediction modzl exhibkts xocialjy asymptohic lcdrning on the comppete yraph, while besh response and fixed response dbiamics fail tp go nhe same.
Pxevioux work
=============
Our modql is an elabotation of modeus studied by DeMerzo, Vayanos and Zwiebel [@DaLarzo:03], as weln as Mosfel qnd Ramuz [@morrel2010iterative; @MpsselTamud10:axxiv]. Therw, the state $S$ is a fibsd number pickeb qt time $t=0$, and esch agqnn rxceivqv a single maasufemdmt of it. The progesr thrreafter is determinhstid, with each agent, st each itgration, rqcalculating ots estimate of $S$ hased on its pbsgrvation of its neighbors’ estimztes.
In [@MosdelBamuz10:arxiv] it is whown that iy the agents calculate the minimum variaice unbiased linear esjimator (MVULE) at everi tmrn (remembermng alj of theis observations) then qll the agents cokverge to the optimal sstimador ov $S$, i.e. the average of the original measurements. Furthermore, this happens un timx ehat is at jost $n
\cdon d$, chere $d$ is vhx diameter of the graph.
When agents calculate estmmates thad cre only based on their | general graphs we show that fixed response to steady state. the complete graph dynamics converge to a state. Both of results hold regardless of the initial (i.e., the agents’ estimators at time $t=0$). We show that the steady state best response dynamics is not necessarily optimal; there exist fixed response dynamics in the converge estimators all have lower variance then the estimators of the steady state of best response dynamics. This that every agent can do better than the of best response by a socially optimal rule; thus certain of anarchy*]{} to paid agents choose the that maximizes their short term gain. Finally, we show that in the penultimate prediction model, for the graph, the estimators which the (in minimum variance sense) linear estimators, and thus outperform those best response dynamics. We define a notion of asymptotic learning”: model has this property when the of the agents’ steady-state estimators tends towards the optimum with the number of agents. We show that the penultimate prediction model exhibits socially on the complete graph, best response and response fail do same. Previous ============= Our model is an elaboration of models studied by DeMarzo, and Zwiebel [@DeMarzo:03], as well as Mossel and Tamuz [@mossel2010iterative; the $S$ is a number picked at time and agent receives a single it. process with agent, each iteration, recalculating its of $S$ based on its of its neighbors’ estimates. that if the agents calculate the minimum variance linear estimator (MVULE) at every turn (remembering of their observations) then all the agents converge to the optimal estimator $S$, i.e. of the original measurements. Furthermore, this happens in that is at most \cdot d$, where $d$ is the diameter of the When calculate estimates are only based their | general graphs we show that fiXed responsE dynaMicS coNvErge To a sTeady state. On thE CompLete graph we show that besT respOnSE dynAMiCs alsO converGE tO A SteAdY sTatE. BOTh Of theSe rEsults hOld regardlEss Of The initial coNDiTions (i.e., the AgeNts’ estimatorS at Time $t=0$).
WE sHow THat thE stEady sTate of BEst resPonse dynaMiCS is not NEcessarILY oPtimAl; there exist fixed REsPOnse dynamics in Which tHe AGeNTS coNveRge to estimAtOrs whICh all haVE lOWER vaRIance then the eStimators of THe sTeady sTaTe oF Best reSponsE dYNamIcs. This showS thaT every ageNt can dO Better tHAn the reSult of BesT reSponSE bY fOllOwINg a SOcIalLY opTimal rulE; tHuS a cerTain [*PRICE of aNarChy*]{} iS to be Paid when agentS chOose THe aCtion That mAximIzEs theIr shorT term GaIn.
Finally, we show That In the penuLtiMaTe pReDictiON model, For The CompletE graph, tHE agEnTS LEaRn estimators which aRe THE oPtimal (in The minIMuM vARiance seNsE) amOngsT ALl linEar eSTiMators, anD thus oUTpErForm thoSe Of fixeD aNd bEst RespoNSe dyNamics.
we define A notiON of “socially asyMPtotic learninG”: a mODEl HAs thIs pRoperty when The vARianCe of THe AgeNTs’ steAdy-stAtE EsTImators tends towards ThE inforMatioN-theoretical oPtimum with THE Number of AgenTS. WE Show that the penUltimAte predictIOn model eXhibiTs socialLy asymptoTIC learninG on The ComPleTE GrAph, while best rESPonsE aNd fixed ResPonse dyNamIcs FaiL to Do The same.
PrEvious woRk
=============
ouR mOdEl iS an elABoration Of ModElS stUdied BY DeMarZo, VayAnos AnD ZWIebEl [@DeMarZO:03], aS WEll aS MOsSel aNd TAmUz [@mosSel2010iTEraTive; @MosSelTamuz10:aRxiV]. therE, tHe State $S$ iS a fixed number PiCked at time $T=0$, aNd eAch ageNT Receives A single measurement of it. THE procesS thEreafTer iS determinIstIc, with EacH Agent, aT each iTeratIoN, reCALculaTINg Its EsTimate of $S$ bASEd oN its oBsErvaTion of iTs neighbors’ estimatES.
In [@mosselTamuz10:arXiv] It is SHOwN thAT iF The AgENts CALculate the minimUm variance UnBIaSed linear eSTimAtOr (MVULE) At every Turn (rEMemberiNg all of thEir observAtIons) THEn aLl the agentS convergE to the optIMal esTImAtor oF $S$, i.E. the avErAge Of the OriginAL meAsureMents. FUrThermoRe, thiS hAppens in Time that is at most $n
\cdot d$, wHere $d$ iS the dIamEter of the GraPH.
WhEn agents cAlcuLate estimaTes ThaT are oNly BAsed oN theIR | general graphs we show th at fixed r espon sedyn am icsconv erge to a stea d y st ate. On the complete g raphwe show th at be st resp o ns e dyn am ic s a ls o c onver geto a st eady state . B ot h of these r e su lts hold r ega rdless of th e i nitial c ond i tions (i .e.,the ag e nts’ e stimators a t time$ t=0$).W eshow that the steadys ta t e of best resp onse d yn a mi c s is no t necessar il y opt i mal; th e re e x ist fixed respons e dynamicsi n w hich t he ag e nts co nverg et o e stimators w hich all have lower varianc e then t he est ima tor s of th este ad y st a te of bes t respon se d ynami cs.T h i s sho wsthat ever y agent can d o b ette r th an th e res ultof best respo nse b yfollowing a soc iall y optimal ru le ; t hu s a c e rtain[*p ric e of an archy*] { } i st o be paid when agentsch o o se the act ion th a tma x imizes t he irshor t termgain .
Finally, we sh o wth at in t he penul ti mat e p redic t ionmodel, for the comp l ete graph, the agents learne st i m at o rs w hic h are the o ptim a l (i n th e m ini m um va rianc es en s e) amongst all line ar estim ators , and thus ou tperform t h o s e of fix ed a n db est response d ynami cs.
We de f ine a no tionof “soci ally asym p t otic lea rni ng” : A mo d e lhas this prop e r ty w he n the v ari ance of th e a gen ts’ s teady-sta te estim at or ste nds towa r ds the i nf orm at ion -theo r etical opti mumwi th the number of a gent s. W e sh owth at th e pe n ult imate p rediction mo d el e xh ib its soc ially asympto ti c learning o n t he com p l ete grap h, while best responsea nd fixe d r espon se d ynamics f ail to do th e same.
Prev iouswo rk= = ===== = = == ==
O ur model i s anelabo ra tion of mod els studied by DeM a rzo , Vayanos and Zw iebe l [@ DeM a rz o :03 ], asw e ll as Mossel an d Tamuz [@ mo s se l2010itera t ive ;@Mossel Tamuz10 :arxi v ]. Ther e, the st ate $S$ i sa fi x e d n umber pick ed at ti me $t=0$, and e a ch agen t r eceive sa s ingle measu r eme nt of it. T he proce ss th er eafter i s deterministic, with e ach ag ent,ateach iter ati o n,recalcula ting its estim ate of $S$bas e d onitso bs erv a tionof i t s neighbo r s’ es t i ma tes.
In [@ M o s sel Tamuz 10: a rxiv]it i s shown that if t h e agents calcu late t hemin i mumva riance unbiase d l in e a r estima to r (MVULE) a t everytu r n (re member ing al l of th e i ro bserva tion s)then allthe a g ents co nv er g e to t he o pt imal e stimat o r of $ S$, i.e. the ave rageo f theo rig inalme asureme n ts.Furthermor e, this hap pens i n ti me th at is a tmost $ n
\ cd ot d$, whe r e $d$ isthe d iameter o f th e g raph.
Whe n agent s ca lc ula te estima t e st ha ta reonly base don t heir | general_graphs we_show that fixed response_dynamics converge_to_a steady_state._On the complete_graph we show_that best response dynamics_also converge to_a_steady state. Both of these results hold regardless of the initial conditions (i.e., the_agents’_estimators at_time_$t=0$).
We_show that the steady state_of best response dynamics is_not necessarily_optimal; there exist fixed response dynamics in which_the_agents converge to_estimators which all have lower variance then the estimators_of the steady state of best_response dynamics. This_shows_that_every agent can do_better than the result of best_response by following a socially optimal_rule; thus a certain [*price of anarchy*]{}_is to be paid when agents_choose the action that maximizes_their short_term gain.
Finally, we show that_in the penultimate_prediction model,_for the complete_graph, the agents learn estimators which_are the optimal_(in the minimum variance sense) amongst_all_linear estimators, and_thus_outperform_those of_fixed and best_response_dynamics.
We define_a_notion of “socially asymptotic learning”: A_model_has this property when the variance of_the agents’ steady-state estimators_tends_towards the information-theoretical optimum_with the number of agents._We show that the penultimate prediction_model exhibits_socially asymptotic_learning on the complete graph, while best response and fixed response_dynamics fail to do the same.
Previous_work
=============
Our model is an_elaboration of_models_studied by DeMarzo,_Vayanos_and Zwiebel [@DeMarzo:03],_as well as Mossel and Tamuz [@mossel2010iterative; @MosselTamuz10:arxiv]._There, the_state $S$ is a fixed number_picked at time $t=0$,_and_each agent receives a single measurement_of it. The process thereafter is_deterministic, with each agent, at_each_iteration,_recalculating its estimate of $S$_based on its observation of its_neighbors’ estimates.
In [@MosselTamuz10:arxiv] it_is shown that if the agents calculate_the_minimum variance unbiased linear estimator (MVULE)_at_every turn (remembering all of their_observations)_then_all the agents converge to_the optimal estimator of $S$, i.e._the average of the original measurements. Furthermore, this happens_in time that_is at most $n
\cdot d$,_where_$d$_is the diameter of the graph.
When agents calculate estimates that_are only_based on their |
/\bigl(nzn^{-\gamma}
\bigr) \,\mathrm{d}z+\mathrm{o}_{{\mathbb{P}}}\bigl(n^{-\gamma/2}\bigr).\qquad\end{aligned}$$ Since $V_i$’s are i.i.d. standard Gaussian, a central limit theorem for $\lambda_n^*-\mathbf{B}_n^*$ can be easily derived. Now Theorem \[thmnewtest\] follows from [(\[eqnewbias\])]{} and [(\[eqnewvar\])]{}. Details are omitted.
[Proof of Proposition \[proppower\]]{} By the Cauchy–Schwarz inequality, $$\begin{aligned}
\int_{{\mathbb{R}}}Q(c_{max},y)^2 \,
\mathrm{d}y&=&\int_{{\mathbb{R}}}\biggl[\int_{c_{\min}}^{c_{\max
}}
\bigl(\bigl[2K(y/z)-K\ast K(y/z)\bigr]/\sqrt{z}\bigr)\times
1/\sqrt{z} \,
\mathrm{d}z \biggr]^2 \,\mathrm{d}y
\\
&\le& \int_{{\mathbb{R}}}\biggl[\int_{c_{\min}}^{c_{\max}}\bigl[2K(y/z)-K\ast K(y/z)
\bigr]^2/z \,\mathrm{d}z\int_{c_{\min}}^{c_{\max}}1/z
\,\mathrm{d}z\biggr] \,\mathrm{d}y
\\
&=&\bigl(\log(c_{\max})-
\log(c_{\min})\bigr)\int_{c_{\min}}^{c_{\max}}\int
_{{\mathbb{R}}}\bigl[2K(y/z)-K\ast K(y/z)\bigr]^2/z \,
\mathrm{d}y \,\mathrm{d}z
\\
&=&\bigl(\log(c_{\max})-
\log(c_{\min})\bigr) (c_{\max}-c_{\min})\int
_{{{\mathbb{R}}}}\tilde{K}^2(t) \,\mathrm{d}t.\end{aligned}$$ Consider any fixed $c\in(0,\infty)$. Plugging the above inequality into [(\[powernewtest\])]{} and letting $c_{\max}\downarrow c$ and $c_{\min
}\uparrow c$, it follows that $\sup_{0<c_{\min}<c_{\max}<\ | /\bigl(nzn^{-\gamma }
\bigr) \,\mathrm{d}z+\mathrm{o}_{{\mathbb{P}}}\bigl(n^{-\gamma/2}\bigr).\qquad\end{aligned}$$ Since $ V_i$ ’s are i.i.d. standard Gaussian, a central limit theorem for $ \lambda_n^*-\mathbf{B}_n^*$ can be easily derived. nowadays Theorem \[thmnewtest\ ] take after from [ (\[eqnewbias\ ]) ] { } and [ (\[eqnewvar\ ]) ] { }. Details are omitted.
[ Proof of Proposition \[proppower\ ] ] { } By the Cauchy – Schwarz inequality, $ $ \begin{aligned }
\int_{{\mathbb{R}}}Q(c_{max},y)^2 \,
\mathrm{d}y&=&\int_{{\mathbb{R}}}\biggl[\int_{c_{\min}}^{c_{\max
} }
\bigl(\bigl[2K(y / z)-K\ast K(y / z)\bigr]/\sqrt{z}\bigr)\times
1/\sqrt{z } \,
\mathrm{d}z \biggr]^2 \,\mathrm{d}y
\\
& \le & \int_{{\mathbb{R}}}\biggl[\int_{c_{\min}}^{c_{\max}}\bigl[2K(y / z)-K\ast K(y / z)
\bigr]^2 / omega \,\mathrm{d}z\int_{c_{\min}}^{c_{\max}}1 / z
\,\mathrm{d}z\biggr ] \,\mathrm{d}y
\\
& = & \bigl(\log(c_{\max})-
\log(c_{\min})\bigr)\int_{c_{\min}}^{c_{\max}}\int
_ { { \mathbb{R}}}\bigl[2K(y / z)-K\ast K(y / z)\bigr]^2 / z \,
\mathrm{d}y \,\mathrm{d}z
\\
& = & \bigl(\log(c_{\max})-
\log(c_{\min})\bigr) (c_{\max}-c_{\min})\int
_ { { { \mathbb{R}}}}\tilde{K}^2(t) \,\mathrm{d}t.\end{aligned}$$ Consider any fixed $ c\in(0,\infty)$. plug the above inequality into [ (\[powernewtest\ ]) ] { } and letting $ c_{\max}\downarrow c$ and $ c_{\min
} \uparrow c$, it follows that $ \sup_{0 < c_{\min}<c_{\max}<\ | /\bigp(nzn^{-\gamma}
\bigr) \,\mathrm{d}z+\mauhrm{o}_{{\mathbb{P}}}\bigl(n^{-\yqmma/2}\bijr).\qquad\snd{aligndd}$$ Since $V_i$’s are i.i.d. standarv Gayssiab, a central limit theofem for $\lwmbda_n^*-\marhbf{U}_n^*$ can be easily derived. Now Tgcorem \[vhmnewtest\] follpws from [(\[exnewbias\])]{} and [(\[exndwrar\])]{}. Details are omitted.
[Proof of Propjsition \[pgoppower\]]{} By thg Caubhr–Schsarz inequality, $$\begin{aligned}
\int_{{\matgbb{R}}}Q(c_{mex},y)^2 \,
\mathrm{d}y&=&\int_{{\kathbb{R}}}\biggl[\int_{c_{\min}}^{c_{\max
}}
\bigp(\bigp[2K(y/z)-K\ast K(y/z)\bigr]/\sqgt{z}\bigr)\timew
1/\sqre{z} \,
\mathrm{d}z \bkggr]^2 \,\mathrm{d}y
\\
&\le& \int_{{\matgbb{R}}}\biggl[\int_{c_{\min}}^{c_{\max}}\bigl[2K(y/z)-K\ast K(y/z)
\bngr]^2/z \,\mathrm{e}z\unt_{f_{\kin}}^{c_{\max}}1/z
\,\matirm{d}z\bpggr] \,\mathrm{d}y
\\
&=&\npgl(\log(c_{\kax})-
\log(c_{\kin})\bigr)\int_{c_{\min}}^{g_{\max}}\iit
_{{\marhbb{R}}}\bigl[2K(y/z)-K\ast K(y/z)\bmgr]^2/z \,
\mathrm{d}y \,\mathrm{q}z
\\
&=&\bigl(\log(w_{\mcx})-
\log(c_{\min})\bigr) (c_{\max}-c_{\mib})\ibt
_{{{\matvbb{R}}}}\dildd{J}^2(t) \,\oatgrk{d}f.\end{allgnxd}$$ Consider any fixed $x\in(0,\infty)$. Plugging tne qbove inequaljty ineo [(\[powernewtest\])]{} and letting $c_{\max}\downarrow c$ znd $c_{\min
}\uparrow c$, it foolows that $\sup_{0<c_{\min}<c_{\mad}<\ | /\bigl(nzn^{-\gamma} \bigr) \,\mathrm{d}z+\mathrm{o}_{{\mathbb{P}}}\bigl(n^{-\gamma/2}\bigr).\qquad\end{aligned}$$ Since $V_i$’s are i.i.d. a limit theorem $\lambda_n^*-\mathbf{B}_n^*$ can be follows [(\[eqnewbias\])]{} and [(\[eqnewvar\])]{}. are omitted. [Proof Proposition \[proppower\]]{} By the Cauchy–Schwarz inequality, \int_{{\mathbb{R}}}Q(c_{max},y)^2 \, \mathrm{d}y&=&\int_{{\mathbb{R}}}\biggl[\int_{c_{\min}}^{c_{\max }} \bigl(\bigl[2K(y/z)-K\ast K(y/z)\bigr]/\sqrt{z}\bigr)\times 1/\sqrt{z} \, \mathrm{d}z \biggr]^2 \,\mathrm{d}y \\ &\le& K(y/z) \bigr]^2/z \,\mathrm{d}z\int_{c_{\min}}^{c_{\max}}1/z \,\mathrm{d}z\biggr] \,\mathrm{d}y \\ &=&\bigl(\log(c_{\max})- \log(c_{\min})\bigr)\int_{c_{\min}}^{c_{\max}}\int _{{\mathbb{R}}}\bigl[2K(y/z)-K\ast K(y/z)\bigr]^2/z \, \mathrm{d}y \,\mathrm{d}z &=&\bigl(\log(c_{\max})- (c_{\max}-c_{\min})\int \,\mathrm{d}t.\end{aligned}$$ any fixed $c\in(0,\infty)$. Plugging the above inequality into [(\[powernewtest\])]{} and letting $c_{\max}\downarrow c$ and $c_{\min }\uparrow it follows that $\sup_{0<c_{\min}<c_{\max}<\ | /\bigl(nzn^{-\gamma}
\bigr) \,\mathrm{d}z+\mAthrm{o}_{{\mathBb{P}}}\biGl(n^{-\GamMa/2}\Bigr).\QquaD\end{aligned}$$ SinCE $V_i$’s Are i.i.d. standard Gaussian, A centRaL LimiT ThEorem For $\lambDA_n^*-\MAThbF{B}_N^*$ cAn bE eASiLy derIveD. Now TheOrem \[thmnewTesT\] fOllows from [(\[eqNEwBias\])]{} and [(\[eqnEwvAr\])]{}. Details are OmiTted.
[PrOoF of pRoposItiOn \[proPpower\]]{} bY the CaUchy–SchwaRz INequalITy, $$\begin{ALIgNed}
\iNt_{{\mathbb{R}}}Q(c_{max},y)^2 \,
\maTHrM{D}y&=&\int_{{\mathbb{R}}}\biGgl[\int_{C_{\mIN}}^{c_{\MAX
}}
\biGl(\bIgl[2K(y/z)-K\ast k(y/Z)\bigr]/\SQrt{z}\bigR)\TiMES
1/\SqrT{Z} \,
\mathrm{d}z \biggR]^2 \,\mathrm{d}y
\\
&\le& \INt_{{\mAthbb{R}}}\BiGgl[\INt_{c_{\min}}^{C_{\max}}\bIgL[2k(y/z)-k\ast K(y/z)
\bigr]^2/Z \,\matHrm{d}z\int_{c_{\Min}}^{c_{\maX}}1/Z
\,\mathrm{D}Z\biggr] \,\mAthrm{d}Y
\\
&=&\biGl(\lOg(c_{\mAX})-
\lOg(C_{\miN})\bIGr)\iNT_{c_{\Min}}^{C_{\Max}}\Int
_{{\mathbB{R}}}\BiGl[2K(y/z)-k\ast k(Y/Z)\BIgr]^2/z \,
\MatHrm{d}Y \,\mathRm{d}z
\\
&=&\bigl(\log(c_{\mAx})-
\lOg(c_{\mIN})\biGr) (c_{\maX}-c_{\min})\Int
_{{{\mAtHbb{R}}}}\tIlde{K}^2(t) \,\MathrM{d}T.\end{aligned}$$ ConsIder Any fixed $c\In(0,\iNfTy)$. PLuGging THe abovE inEquAlity inTo [(\[powerNEwtEsT\])]{} AND lEtting $c_{\max}\downarroW c$ AND $c_{\Min
}\uparrOw c$, it fOLlOwS That $\sup_{0<c_{\MiN}<c_{\mAx}<\ | /\bigl(nzn^{-\gamma}
\bigr ) \,\mathr m{d}z +\m ath rm {o}_ {{\m athbb{P}}}\big l (n^{ -\gamma/2}\bigr).\qqua d\end {a l igne d }$ $ Sin ce $V_i $ ’s a rei. i. d.st a nd ard G aus sian, a central l imi ttheorem for$ \l ambda_n^*- \ma thbf{B}_n^*$ ca n be e as ily deriv ed. NowTheore m \[thm newtest\] f o llowsf rom [(\ [ e qn ewbi as\])]{} and [(\[ e qn e wvar\])]{}. De tailsar e o m i tte d.
[Proof of P ropos i tion \[ p ro p p o wer \ ]]{} By the C auchy–Schwa r z i nequal it y,$ $\begi n{ali gn e d}\int_{{\mat hbb{ R}}}Q(c_{ max},y ) ^2 \,
\ m athrm{d }y&=&\ int _{{ \mat h bb {R }}} \b i ggl [ \i nt_ { c_{ \min}}^{ c_ {\ max
} }
\b i g l ( \big l[2 K(y/ z)-K\ ast K(y/z)\bi gr] /\sq r t{z }\big r)\ti mes1/ \sqrt {z} \,
\mat hr m{d}z \biggr]^2 \,\ mathrm{d} y
\ \&\l e& \int _ {{\mat hbb {R} }}\bigg l[\int_ { c_{ \m i n } }^ {c_{\max}}\bigl[2K (y / z )- K\ast K( y/z)
\ b ig r] ^ 2/z \,\m at hrm {d}z \ i nt_{c _{\m i n} }^{c_{\m ax}}1/ z
\ ,\ mathrm{ d} z\bigg r] \, \ma thrm{ d }y
\ \
&=&\ bigl(\lo g(c_{ \ max})-
\log(c_ { \min})\bigr)\ i nt _ { c_ { \min }}^ {c_{\max}}\ int_ {{\m athb b {R }}} \ bigl[ 2K(y/ z) - K\ a st K(y/z)\bigr]^2/z \ ,
\mat hrm{d }y \,\mathrm{ d}z
\\
&=& \ b i gl(\log( c_{\ m ax } )-
\log(c_{\mi n})\b igr) (c_{\ m ax}-c_{\ min}) \int
_{{ {\mathbb{ R } }}}\tild e{K }^2 (t) \, \ m at hrm{d}t.\end{ a l igne d} $$ Cons ide r any f ixe d $ c\i n(0 ,\ infty)$.Plugging t he a bo veinequ a lity int o[(\ [p owe rnewt e st\])] {} an d le tt in g $c _{\max} \ do w n arro wc$ and $c _{ \min}\up a rro w c$, i t follows th a t $\ su p_ {0<c_{\ min}<c_{\max} <\ | /\bigl(nzn^{-\gamma}
\bigr) \,\mathrm{d}z+\mathrm{o}_{{\mathbb{P}}}\bigl(n^{-\gamma/2}\bigr).\qquad\end{aligned}$$_Since $V_i$’s_are i.i.d. standard Gaussian,_a central_limit_theorem for_$\lambda_n^*-\mathbf{B}_n^*$_can be easily_derived. Now Theorem_\[thmnewtest\] follows from [(\[eqnewbias\])]{}_and [(\[eqnewvar\])]{}. Details_are_omitted.
[Proof of Proposition \[proppower\]]{} By the Cauchy–Schwarz inequality, $$\begin{aligned}
\int_{{\mathbb{R}}}Q(c_{max},y)^2 \,
\mathrm{d}y&=&\int_{{\mathbb{R}}}\biggl[\int_{c_{\min}}^{c_{\max
}}
\bigl(\bigl[2K(y/z)-K\ast K(y/z)\bigr]/\sqrt{z}\bigr)\times
1/\sqrt{z} \,
\mathrm{d}z \biggr]^2 \,\mathrm{d}y
\\
&\le&_\int_{{\mathbb{R}}}\biggl[\int_{c_{\min}}^{c_{\max}}\bigl[2K(y/z)-K\ast_K(y/z)
\bigr]^2/z \,\mathrm{d}z\int_{c_{\min}}^{c_{\max}}1/z
\,\mathrm{d}z\biggr]_\,\mathrm{d}y
\\
&=&\bigl(\log(c_{\max})-
\log(c_{\min})\bigr)\int_{c_{\min}}^{c_{\max}}\int
_{{\mathbb{R}}}\bigl[2K(y/z)-K\ast_K(y/z)\bigr]^2/z_\,
\mathrm{d}y \,\mathrm{d}z
\\
&=&\bigl(\log(c_{\max})-
\log(c_{\min})\bigr) (c_{\max}-c_{\min})\int
_{{{\mathbb{R}}}}\tilde{K}^2(t) \,\mathrm{d}t.\end{aligned}$$ Consider_any fixed $c\in(0,\infty)$. Plugging the_above inequality_into [(\[powernewtest\])]{} and letting $c_{\max}\downarrow c$ and $c_{\min
}\uparrow_c$,_it follows that_$\sup_{0<c_{\min}<c_{\max}<\ |
ell}\right)^\frac{4}{3}~,$$ being $g_H$ the number of degrees of freedom that become non relativistic between typical BBN temperatures and $T_D$. The authors of [@Blennow:2012de] have shown that the cosmological constraints on ${N_{\textrm{eff}}}$ can be translated into the required heavy degrees of freedom heating the light dark sector plasma $g_h$ as a function of the dark sector decoupling temperature $T_D$ for a fixed value of $g_\ell$. Recent Planck data [@Ade:2013zuv], combined with measurements of the Hubble constant $H_0$ from the Hubble Space Telescope (HST), low multipole polarization measurements from the Wilkinson Microwave Anisotropy Probe (WMAP) 9 year data release [@Hinshaw:2012fq] and high multipole CMB data from both the Atacama Cosmology Telescope (ACT) [@Sievers:2013ica] and the South Pole Telescope (SPT) [@Hou:2012xq; @Story:2012wx] provide the constraint ${N_{\textrm{eff}}}$ is $3.62^{+0.50}_{-0.48}$. Using this constraint, the authors of Ref. [@DiValentino:2013qma] have found that having extra heavy degrees of freedom in the dark sector for low decoupling temperatures is highly disfavored.
Another aspect of dark radiation is that it could interact with the dark matter sector. In asymmetric dark matter models (see Ref. [@Blennow:2010qp]), the dark matter production mechanism resembles to the one in the baryonic sector, with a particle-antiparticle asymmetry at high temperatures. The thermally symmetric dark matter component eventually annihilates and decays into dark radiation species. Due to the presence of such an interaction among the dark matter and dark radiation sectors, they behave as a tightly coupled fluid with pressure which will imprint oscillations in the matter power spectrum (as the acoustic oscillations in the photon-baryon fluid before the recombination era). The clustering properties of the dark radiation component may be modified within interacting schemes, and therefore the clustering parameters ${c_{\textrm{eff}}^2}$ and ${c_{\textrm{vis}}^2}$ may differ from their standard values for the neutrino case ${c_{\textrm{eff}}^2}= {c_{\textrm{vis}}^2}= 1/3$ (see Section \[sec:ceffcvis\]). In | ell}\right)^\frac{4}{3}~,$$ being $ g_H$ the number of degrees of freedom that become non relativistic between distinctive BBN temperature and $ T_D$. The authors of [ @Blennow:2012de ] have shown that the cosmologic restraint on $ { N_{\textrm{eff}}}$ can be translated into the required big degree of freedom heat the light dark sector plasma $ g_h$ as a routine of the dark sector decoupling temperature $ T_D$ for a specify value of $ g_\ell$. Recent Planck data [ @Ade:2013zuv ], combined with measurements of the Hubble constant $ H_0 $ from the Hubble Space Telescope (HST), low multipole polarization measurements from the Wilkinson Microwave Anisotropy Probe (WMAP) 9 class data release [ @Hinshaw:2012fq ] and high multipole CMB datum from both the Atacama Cosmology Telescope (ACT) [ @Sievers:2013ica ] and the South Pole Telescope (SPT) [ @Hou:2012xq; @Story:2012wx ] provide the constraint $ { N_{\textrm{eff}}}$ is $ 3.62^{+0.50}_{-0.48}$. Using this constraint, the authors of Ref. [ @DiValentino:2013qma ] have recover that having extra heavy degrees of freedom in the blue sector for low decoupling temperatures is highly disfavored.
Another aspect of dark radiation is that it could interact with the dark matter sector. In asymmetric dark matter models (see Ref. [ @Blennow:2010qp ]), the dark matter production mechanism resembles to the one in the baryonic sector, with a particle - antiparticle asymmetry at eminent temperatures. The thermally symmetric colored topic part eventually annihilates and decays into dark radiation species. Due to the presence of such an interaction among the dark matter and blue radiation sectors, they behave as a tightly coupled fluid with pressure which will imprint oscillation in the matter power spectrum (as the acoustic oscillations in the photon - baryon fluid before the recombination era). The cluster properties of the black radiation component may be modified within interacting schemes, and consequently the clustering parameters $ { c_{\textrm{eff}}^2}$ and $ { c_{\textrm{vis}}^2}$ may differ from their standard values for the neutrino case $ { c_{\textrm{eff}}^2}= { c_{\textrm{vis}}^2}= 1/3 $ (see Section \[sec: ceffcvis\ ]). In | ell}\gight)^\frac{4}{3}~,$$ being $g_H$ the nmmber of degrees of frexdom thzt becomd non relativistic between tbpicql BBB temperatures and $T_D$. Ghe authogs of [@Blebnow:2012ve] have shown thef the cosmoloflcal eoistraints on ${N_{\tgxtrm{eff}}}$ can be translated ivtl the required heavy degrees of freqdom hestlng the light qark fectkg klasma $g_h$ as a function of the dadk sectmr decoupling temperature $T_D$ for a fixef vapue of $g_\ell$. Recent Planck datq [@Ade:2013syv], combined dith measugzments of tge Hubble constant $H_0$ from the Hjbble Space Telgsdoog (HST), low muotipoje polarizatlpn meavuremenys from the Wikkiison Microwave Anisotropy Probe (WMAP) 9 year qata reledsz [@Hinshaw:2012fq] and high mylripolg CMB datx frum govh fhe Atwcaja Cosmolofy Telescopw (ACT) [@Sievers:2013ica] and trv South Pole Felescjpq (SPT) [@Hou:2012xq; @Story:2012wx] provide the constraitt ${H_{\textrm{eff}}}$ is $3.62^{+0.50}_{-0.48}$. Using thus constraint, the autjors of Rqf. [@DiValentino:2013qma] have found that having extra heaey dejrdes on frdwdlm in the dark sector for low decoupling tempqdauurvs is highly disfcvored.
Another asleft jf dark radiajion is thzt it could interaft with the eark matttr sevtor. In asymmetric dark matrer models (sve Rwf. [@Blennow:2010qp]), the daxk matter pruducjion mrchanism resembles to tke one in the barjonic secfur, with a particue-akti[article asymmetry at high temperatnres. Che theroalli symmeeric dark lattev component eventuaply aunihinates and fecays into dark radiation specmxs. Due to the psesvnce of sbch an interaction wmong the dark matter and aark radianion sectmrs, they berave as a tigvjly coupled fnuid witr prwssuee whicf will imprint pscillations in the matter power specbrum (xa the acoustic iscullations in thr pfotjn-hacyon snuid before dhe fecukbinagion era). Thc cuustrring properties of dhe sark radiation comlokent may ve modifyed within inyeracting schemes, wnd tierefoce the clostering parameters ${c_{\textrm{eff}}^2}$ znd ${c_{\textgm{vls}}^2}$ may differ fron their stanbard values for the neutrino case ${c_{\textrn{eff}}^2}= {c_{\textrm{vis}}^2}= 1/3$ (swe Section \[sec:ceffcrix\]). In | ell}\right)^\frac{4}{3}~,$$ being $g_H$ the number of degrees that non relativistic typical BBN temperatures [@Blennow:2012de] shown that the constraints on ${N_{\textrm{eff}}}$ be translated into the required heavy of freedom heating the light dark sector plasma $g_h$ as a function of dark sector decoupling temperature $T_D$ for a fixed value of $g_\ell$. Recent Planck [@Ade:2013zuv], with of Hubble constant $H_0$ from the Hubble Space Telescope (HST), low multipole polarization measurements from the Wilkinson Anisotropy Probe (WMAP) 9 year data release [@Hinshaw:2012fq] high multipole CMB data both the Atacama Cosmology Telescope [@Sievers:2013ica] the South Telescope [@Hou:2012xq; provide the constraint is $3.62^{+0.50}_{-0.48}$. Using this constraint, the authors of Ref. [@DiValentino:2013qma] have found that having extra heavy degrees freedom in sector for decoupling is disfavored. Another aspect radiation is that it could interact matter sector. In asymmetric dark matter models (see [@Blennow:2010qp]), the matter production mechanism resembles to the in the baryonic sector, with a particle-antiparticle asymmetry high temperatures. The thermally symmetric dark matter component eventually annihilates and decays into dark radiation to the presence of an interaction among dark and radiation they behave a tightly coupled fluid with pressure which will imprint oscillations in matter power spectrum (as the acoustic oscillations in the photon-baryon the era). The clustering of the dark radiation may modified within interacting schemes, the parameters may from standard values for the case ${c_{\textrm{eff}}^2}= {c_{\textrm{vis}}^2}= 1/3$ (see \[sec:ceffcvis\]). In | ell}\right)^\frac{4}{3}~,$$ being $g_H$ the numBer of degreEs of fReeDom ThAt beCome Non relativistiC BetwEen typical BBN temperatuRes anD $T_d$. the aUThOrs of [@blennow:2012DE] hAVE shOwN tHat ThE CoSmoloGicAl constRaints on ${N_{\tExtRm{Eff}}}$ can be tranSLaTed into the ReqUired heavy deGreEs of frEeDom HEatinG thE lighT dark sECtor plAsma $g_h$ as a FuNCtion oF The dark SECtOr deCoupling temperatuRE $T_d$ For a fixed value Of $g_\ell$. reCEnT pLanCk dAta [@Ade:2013zuv], cOmBined WIth measUReMENTs oF The Hubble consTant $H_0$ from thE hubBle SpaCe telEScope (HsT), low MuLTipOle polarizaTion MeasuremeNts froM The WilkINson MicRowave aniSotRopy pRoBe (wMAp) 9 yEAr dATa RelEAse [@hinshaw:2012fQ] aNd High mUltiPOLE cMB dAta From Both tHe Atacama CosmOloGy TeLEscOpe (ACt) [@SievErs:2013iCa] And thE South pole TElEscope (SPT) [@Hou:2012xq; @STory:2012Wx] provide The CoNstRaInt ${N_{\tEXtrm{efF}}}$ is $3.62^{+0.50}_{-0.48}$. usiNg this cOnstraiNT, thE aUTHOrS of Ref. [@DiValentino:2013qMa] HAVe Found thaT havinG ExTrA Heavy degReEs oF freEDOm in tHe daRK sEctor for Low decOUpLiNg tempeRaTures iS hIghLy dIsfavORed.
ANother Aspect of Dark rADiation is that iT Could interact WItH THe DArk mAttEr sector. In aSymmETric Dark MAtTer MOdels (See ReF. [@BLEnNOw:2010qp]), the dark matter prOdUction MechaNism resembles To the one in THE Baryonic SectOR, wITh a particle-antIpartIcle asymmeTRy at high TempeRatures. THe thermalLY SymmetriC daRk mAttEr cOMPoNent eventuallY ANnihIlAtes and DecAys into DarK raDiaTioN sPecies. Due To the preSeNcE oF sUch An intERaction aMoNg tHe DarK mattER and daRk radIatiOn SeCTorS, they beHAvE AS a tiGhTlY couPleD fLuid wIth pREssUre whicH will imprInt OScilLaTiOns in thE matter power sPeCtrum (as the AcOusTic oscILLations iN the photon-baryon fluid beFOre the rEcoMbinaTion Era). The cluSteRing prOpeRTies of The darK radiAtIon COMponeNT MaY be MoDified withIN IntEractInG schEmes, and Therefore the clusteRIng Parameters ${c_{\teXtrM{eff}}^2}$ AND ${c_{\TexTRm{VIs}}^2}$ mAy DIffER From their standaRd values foR tHE nEutrino casE ${C_{\teXtRm{eff}}^2}= {c_{\tExtrm{viS}}^2}= 1/3$ (see SECtion \[seC:ceffcvis\]). in | ell}\right)^\frac{4}{3}~,$ $ being $g _H$ t henum be r of deg rees of freedo m tha t become non relativis tic b et w eent yp icalBBN tem p er a t ure san d $ T_ D $. Theaut hors of [@Blennow :20 12 de] have sho w nthat the c osm ological con str aintson ${ N _{\te xtr m{eff }}}$ c a n be t ranslated i n to the require d he avydegrees of freedo m h e ating the ligh t dark s e ct o r pl asm a $g_h$ as a func t ion oft he d a rks ector decoupl ing tempera t ure $T_D$ f ora fixed valu eo f $ g_\ell$. Re cent Planck d ata [@ A de:2013 z uv], co mbined wi thmeas u re me nts o f th e H ubb l e c onstant$H _0 $ fro m th e H u bble Sp aceTeles cope (HST), l owmult i pol e pol ariza tion m easur ements from t he Wilkinson Mi crow ave Aniso tro py Pr ob e (WM A P) 9 y ear da ta rele ase [@H i nsh aw : 2 0 12 fq] and high multi po l e C MB datafrom b o th t h e Atacam aCos molo g y Tele scop e ( ACT) [@S ievers : 20 13 ica] an dthe So ut h P ole Tele s cope (SPT) [@Hou:2 012xq ; @Story:2012wx ] provide thec on s t ra i nt $ {N_ {\textrm{ef f}}} $ is$3.6 2 ^{ +0. 5 0}_{- 0.48} $. Us i ng this constraint, t he aut horsof Ref. [@DiV alentino:2 0 1 3 qma] hav e fo u nd that having ex tra h eavy degre e s of fre edomin the d ark secto r for lowdec oup lin g t e m pe ratures is hi g h ly d is favored .
Another as pec t o f d ar k radiati on is th at i tco uld inte r act with t heda rkmatte r secto r. In asy mm et r icdark ma t te r mode ls ( seeRef .[@Ble nnow : 201 0qp]),the darkmat t er p ro du ction m echanism rese mb les to the o nein the b aryonicsector, with a particle - antipar tic le as ymme try at hi ghtemper atu r es. Th e ther mally s ymm e t ric d a r kmat te r componen t eve ntual ly ann ihilate s and decays intod ark radiation sp eci es.D u etot he pre se n ceo f such an intera ction amon gt he dark matt e r a nd dark r adiatio n sec t ors, th ey behave as a tig ht ly c o u ple d fluid wi th press ure which willi mp rintosc illati on s i n the matte r po wer s pectru m(as th e aco us tic osci llations in the photon- baryon flui d b efore the re c omb ination e ra). The clust eri ngprope rti e s ofthed ar k r a diati on c o mponent m a ybem o di fied within i n ter actin g s c hemes, and therefore the cl u stering parame ters $ {c_ {\t e xtrm {e ff}}^2}$ and $ {c_ {\ t e xtrm{vis }} ^2}$ may di ffer fro mt heirstanda rd val ues for t he neutri no c ase ${c_{\te xtr m{ e ff}}^2} ={c _ {\text rm{v is }}^2}= 1/3$( seeS e ction \[sec:ceff cvis\ ] ) . In | ell}\right)^\frac{4}{3}~,$$ being_$g_H$ the_number of degrees of_freedom that_become_non relativistic_between_typical BBN temperatures_and $T_D$. The_authors of [@Blennow:2012de] have_shown that the_cosmological_constraints on ${N_{\textrm{eff}}}$ can be translated into the required heavy degrees of freedom heating_the_light dark_sector_plasma_$g_h$ as a function of_the dark sector decoupling temperature_$T_D$ for_a fixed value of $g_\ell$. Recent Planck data [@Ade:2013zuv],_combined_with measurements of_the Hubble constant $H_0$ from the Hubble Space Telescope_(HST), low multipole polarization measurements from_the Wilkinson Microwave_Anisotropy_Probe_(WMAP) 9 year data_release [@Hinshaw:2012fq] and high multipole CMB data_from both the Atacama Cosmology Telescope_(ACT) [@Sievers:2013ica] and the South Pole Telescope (SPT) [@Hou:2012xq;_@Story:2012wx] provide the constraint ${N_{\textrm{eff}}}$ is_$3.62^{+0.50}_{-0.48}$. Using this constraint, the_authors of_Ref. [@DiValentino:2013qma] have found that having_extra heavy degrees_of freedom_in the dark_sector for low decoupling temperatures is_highly disfavored.
Another aspect_of dark radiation is that it_could_interact with the_dark_matter_sector. In_asymmetric dark matter_models_(see Ref. [@Blennow:2010qp]),_the_dark matter production mechanism resembles to_the_one in the baryonic sector, with a_particle-antiparticle asymmetry at high_temperatures._The thermally symmetric dark_matter component eventually annihilates and_decays into dark radiation species. Due_to the_presence of_such an interaction among the dark matter and dark radiation sectors,_they behave as a tightly coupled_fluid with pressure which_will imprint_oscillations_in the matter_power_spectrum (as_the acoustic oscillations in the photon-baryon fluid_before the_recombination era). The clustering properties of_the dark radiation component_may_be modified within interacting schemes, and_therefore the clustering parameters ${c_{\textrm{eff}}^2}$ and_${c_{\textrm{vis}}^2}$ may differ from their_standard_values_for the neutrino case ${c_{\textrm{eff}}^2}=_{c_{\textrm{vis}}^2}= 1/3$ (see Section \[sec:ceffcvis\]). In |
a{K(z,\bar z)}={{\rm i}}\left( X^A\frac{\partial}{\partial\bar X^A}\bar F(\bar X)- \bar X^A
\frac{\partial}{\partial X^A} F(X)\right)\,.$$ On overlap of charts these functions should be related by (inhomogeneous) symplectic transformations $ISp(2n,{{\mbox{\rm$\mbox{I}\!\mbox{R}$}}})$: $$\left( \begin{array}{c}
X \\ \partial F \end{array}\right)_{(i)} = e^{{{\rm i}}c_{ij}} M_{ij}
\left( \begin{array}{c}
X \\ \partial F\end{array}\right)_{(j)}+ b_{ij}\,, \label{ISpn}$$ with $$c_{ij}\in {{\mbox{\rm$\mbox{I}\!\mbox{R}$}}}\ ;\qquad M_{ij} \in {\mbox{$Sp\left( 2n,{{\mbox{\rm$\mbox{I}\!\mbox{R}$}}}\right) $}}\ ;\qquad b_{ij} \in
{\mathord{\setlength{\unitlength}{1em}
\begin{picture}(0.6,0.7)(-0.1,0)
\put(-0.1,0){\rm C}
\thicklines
\put(0.2,0.05){\line(0,1){0.55}}
\end {picture}}}^{2n} \,,$$ satisfying the cocycle condition on overlaps of 3 charts.
There is, however, a second definition of rigid special [Kähler]{}manifolds, which is based on the symplectic structure, rather than on the prepotential.
A [Kähler]{} manifold is the base manifold of a $U(1)\times I{\mbox{$Sp\left( 2n,{{\mbox{\rm$\mbox{I}\!\mbox{R}$}}}\right) $}}$ bundle. A holomorphic section $\RawSienna{V}(z)$ defines the [Kähler]{} potential by $$\Magenta{K(z,\bar z)}={{\rm i}}{\mbox{$\langle \RawSienna{V}, \RawSienna{{\bar V}} \rangle$}}\,,$$ and it should satisfy the constraint $${\mbox{$\langle \partial_\alpha \Raw | a{K(z,\bar z)}={{\rm i}}\left (X^A\frac{\partial}{\partial\bar X^A}\bar F(\bar X)- \bar X^A
\frac{\partial}{\partial X^A } F(X)\right)\,.$$ On overlap of charts these functions should be related by (inhomogeneous) symplectic transformation $ ISp(2n,{{\mbox{\rm$\mbox{I}\!\mbox{R}$}}})$: $ $ \left (\begin{array}{c }
X \\ \partial F \end{array}\right)_{(i) } = e^{{{\rm i}}c_{ij } } M_{ij }
\left (\begin{array}{c }
ten \\ \partial F\end{array}\right)_{(j)}+ b_{ij}\, , \label{ISpn}$$ with $ $ c_{ij}\in { { \mbox{\rm$\mbox{I}\!\mbox{R}$}}}\; \qquad M_{ij } \in { \mbox{$Sp\left (2n,{{\mbox{\rm$\mbox{I}\!\mbox{R}$}}}\right) $ } } \; \qquad b_{ij } \in
{ \mathord{\setlength{\unitlength}{1em }
\begin{picture}(0.6,0.7)(-0.1,0)
\put(-0.1,0){\rm C }
\thicklines
\put(0.2,0.05){\line(0,1){0.55 } }
\end { picture}}}^{2n } \,,$$ satisfying the cocycle condition on overlaps of 3 chart.
There is, however, a second definition of rigid particular [ Kähler]{}manifolds, which is base on the symplectic structure, rather than on the prepotential.
A [ Kähler ] { } manifold is the base manifold of a $ U(1)\times I{\mbox{$Sp\left (2n,{{\mbox{\rm$\mbox{I}\!\mbox{R}$}}}\right) $ } } $ pile. A holomorphic section $ \RawSienna{V}(z)$ defines the [ Kähler ] { } electric potential by $ $ \Magenta{K(z,\bar z)}={{\rm i}}{\mbox{$\langle \RawSienna{V }, \RawSienna{{\bar V } } \rangle$}}\,,$$ and it should satisfy the constraint $ $ { \mbox{$\langle \partial_\alpha \Raw | a{K(z,\har z)}={{\rm i}}\left( X^A\frac{\partlal}{\partial\bar X^A}\yqr F(\bac X)- \bad X^A
\frac{\oartial}{\partial X^A} F(X)\right)\,.$$ On ocerlak of charts these fjnctions dhould bw reoqted by (inikmogeneous) syjilectnc transformatioks $ISp(2n,{{\mbox{\sm$\mbox{I}\!\mbox{R}$}}})$: $$\lafg( \yegin{array}{c}
X \\ \partial F \end{array}\rigrt)_{(i)} = e^{{{\rk l}}c_{ij}} M_{ij}
\left( \bggin{agrwy}{c}
S \\ \partial F\end{array}\right)_{(j)}+ b_{ij}\,, \labsl{ISpn}$$ xith $$c_{ij}\in {{\mbox{\tm$\mbox{I}\!\mbox{R}$}}}\ ;\qquad M_{ij} \in {\lbox{$Dp\left( 2n,{{\mbox{\rm$\mbox{L}\!\mbox{R}$}}}\right) $}}\ ;\qqtqd b_{ij} \in
{\matford{\setlength{\unitlengtg}{1em}
\begin{picture}(0.6,0.7)(-0.1,0)
\put(-0.1,0){\fm C}
\thixkoinfv
\puv(0.2,0.05){\line(0,1){0.55}}
\end {pictmge}}}^{2n} \,,$$ sadisfyinb the cocycle gondivion on overlaps of 3 charvs.
There is, however, a second dafnnition of rigid specuao [Kählgr]{}manhfolaw, wficg ms gased ln vhe symplecfic structuee, rather than on tne itepotential.
A [Kähler]{} iagifold is the base manifold of a $U(1)\times I{\mgox{$Sp\left( 2n,{{\mbox{\rm$\mbox{I}\!\mvox{R}$}}}\right) $}}$ bundle. A hllomorphis section $\RawSienna{V}(z)$ defines the [Kähler]{} potential ty $$\Majevta{J(z,\nar x)}={{\em i}}{\mbox{$\langle \RawSienna{V}, \RawSienna{{\bar V}} \rangle$}}\,,$$ znc pt should satisfy the constrsijt $${\ibox{$\langle \pattial_\al'ga \Raw | a{K(z,\bar z)}={{\rm i}}\left( X^A\frac{\partial}{\partial\bar X^A}\bar F(\bar X)- \frac{\partial}{\partial F(X)\right)\,.$$ On of charts these (inhomogeneous) transformations $ISp(2n,{{\mbox{\rm$\mbox{I}\!\mbox{R}$}}})$: $$\left( X \\ \partial \end{array}\right)_{(i)} = e^{{{\rm i}}c_{ij}} M_{ij} \left( X \\ \partial F\end{array}\right)_{(j)}+ b_{ij}\,, \label{ISpn}$$ with $$c_{ij}\in {{\mbox{\rm$\mbox{I}\!\mbox{R}$}}}\ ;\qquad M_{ij} \in {\mbox{$Sp\left( $}}\ ;\qquad b_{ij} \in {\mathord{\setlength{\unitlength}{1em} \begin{picture}(0.6,0.7)(-0.1,0) \put(-0.1,0){\rm C} \thicklines \put(0.2,0.05){\line(0,1){0.55}} \end {picture}}}^{2n} \,,$$ the condition overlaps 3 charts. There is, however, a second definition of rigid special [Kähler]{}manifolds, which is based on symplectic structure, rather than on the prepotential. A manifold is the base of a $U(1)\times I{\mbox{$Sp\left( 2n,{{\mbox{\rm$\mbox{I}\!\mbox{R}$}}}\right) bundle. holomorphic section defines [Kähler]{} by $$\Magenta{K(z,\bar z)}={{\rm \RawSienna{V}, \RawSienna{{\bar V}} \rangle$}}\,,$$ and it should satisfy the constraint $${\mbox{$\langle \partial_\alpha \Raw | a{K(z,\bar z)}={{\rm i}}\left( X^A\frac{\partiAl}{\partial\bAr X^A}\bAr F(\Bar x)- \bAr X^A
\Frac{\Partial}{\partial x^a} F(X)\rIght)\,.$$ On overlap of charts tHese fUnCTionS ShOuld bE relateD By (INHomOgEnEouS) sYMpLectiC trAnsformAtions $ISp(2n,{{\MboX{\rM$\mbox{I}\!\mbox{R}$}}})$: $$\lEFt( \Begin{array}{C}
X \\ \pArtial F \end{arRay}\Right)_{(i)} = E^{{{\rM i}}c_{IJ}} M_{ij}
\lEft( \Begin{Array}{c}
x \\ \PartiaL F\end{arraY}\rIGht)_{(j)}+ b_{iJ}\,, \Label{ISPN}$$ WiTh $$c_{iJ}\in {{\mbox{\rm$\mbox{I}\!\mboX{r}$}}}\ ;\qQUad M_{ij} \in {\mbox{$Sp\Left( 2n,{{\mBoX{\Rm$\MBOx{I}\!\MboX{R}$}}}\right) $}}\ ;\qquAd B_{ij} \in
{\MAthord{\sETlENGTh{\uNItlength}{1em}
\begIn{picture}(0.6,0.7)(-0.1,0)
\puT(-0.1,0){\Rm C}
\ThicklInEs
\pUT(0.2,0.05){\line(0,1){0.55}}
\eNd {picTuRE}}}^{2n} \,,$$ sAtisfying thE cocYcle condiTion on OVerlaps OF 3 charts.
there iS, hoWevEr, a sECoNd DefInITioN Of RigID spEcial [KähLeR]{}mAnifoLds, wHICH Is baSed On thE sympLectic structuRe, rAtheR ThaN on thE prepOtenTiAl.
A [KäHler]{} maNifolD iS the base manifolD of a $u(1)\times I{\mbOx{$SP\lEft( 2N,{{\mBox{\rm$\MBox{I}\!\mbOx{R}$}}}\RigHt) $}}$ bundlE. A holomORphIc SECTiOn $\RawSienna{V}(z)$ definEs THE [KÄhler]{} potEntial BY $$\MAgENta{K(z,\bar Z)}={{\rM i}}{\mBox{$\lANGle \RaWSieNNa{v}, \RawSienNa{{\bar V}} \RAnGlE$}}\,,$$ and it sHoUld satIsFy tHe cOnstrAInt $${\mBox{$\lanGle \partiAl_\alpHA \Raw | a{K(z,\bar z)}={{\rm i}}\l eft( X^A\f rac{\ par tia l} {\pa rtia l\bar X^A}\bar F(\b ar X)- \bar X^A
\frac {\par ti a l}{\ p ar tialX^A} F( X )\ r i ght )\ ,. $$On ov erlap of charts these fun cti on s should ber el ated by (i nho mogeneous) s ymp lectic t ran s forma tio ns $I Sp(2n, { {\mbox {\rm$\mbo x{ I }\!\mb o x{R}$}} } ) $: $$\ left( \begin{arra y }{ c }
X \\ \parti al F \ en d {a r r ay} \ri ght)_{(i)} = e^{{ { \rm i}} c _{ i j } } M _ {ij}
\left( \ begin{array } {c}
X \\ \ par t ial F\ end{a rr a y}\ right)_{(j) }+ b _{ij}\,,\label { ISpn}$$ with $$ c_{ij} \in {{ \mbo x {\ rm $\m bo x {I} \ !\ mbo x {R} $}}}\ ;\ qq ua d M_{ ij}\ i n {\mb ox{ $Sp\ left( 2n,{{\mbox{\ rm$ \mbo x {I} \!\mb ox{R} $}}} \r ight) $}}\;\qqu ad b_{ij} \in
{\m atho rd{\setle ngt h{ \un it lengt h }{1em}
\begin{ picture } (0. 6, 0 . 7 )( -0.1,0)
\p ut ( - 0. 1,0){\rm C}
\thickli ne s
\put (0.2 , 0. 05){\lin e(0,1) { 0. 55 }}
\ end {p ic tur e}} }^{2n } \,, $$ sat isfyingthe c o cycle conditio n on overlapso f3 ch a rts.
T here is, ho weve r , aseco n ddef i nitio n ofri g id special [Kähler]{}m an ifolds , whi ch is based o n the symp l e c tic stru ctur e ,r ather than onthe p repotentia l .
A [Kä hler] {} manif old is th e base man ifo ldofa $ U ( 1) \times I{\mbo x { $Sp\ le ft( 2n, {{\ mbox{\r m$\ mbo x{I }\! \m box{R}$}} }\right) $ }} $bu ndl e. Ah olomorph ic se ct ion $\Ra w Sienna {V}(z )$ d ef in e s t he [Käh l er ] { } po te nt ialby$$ \Mage nta{ K (z, \bar z) }={{\rm i }}{ \ mbox {$ \l angle \ RawSienna{V}, \ RawSienna{ {\ bar V}} \ r a ngle$}}\ ,,$$ and it should sati s fy thecon strai nt $ ${\mbox{$ \la ngle \ par t ial_\a lpha \ Raw | a{K(z,\bar z)}={{\rm_i}}\left( X^A\frac{\partial}{\partial\bar_X^A}\bar F(\bar X)- _\bar X^A
\frac{\partial}{\partial_X^A}_F(X)\right)\,.$$ On_overlap_of charts these_functions should be_related by (inhomogeneous) symplectic_transformations $ISp(2n,{{\mbox{\rm$\mbox{I}\!\mbox{R}$}}})$: $$\left(_\begin{array}{c}
_X \\ \partial F \end{array}\right)_{(i)} = e^{{{\rm i}}c_{ij}} M_{ij}
\left( \begin{array}{c}
X \\ \partial F\end{array}\right)_{(j)}+_b_{ij}\,,_\label{ISpn}$$ with_$$c_{ij}\in_{{\mbox{\rm$\mbox{I}\!\mbox{R}$}}}\_;\qquad M_{ij} \in {\mbox{$Sp\left( 2n,{{\mbox{\rm$\mbox{I}\!\mbox{R}$}}}\right)_$}}\ ;\qquad b_{ij} \in
{\mathord{\setlength{\unitlength}{1em}
_ _ \begin{picture}(0.6,0.7)(-0.1,0)
__\put(-0.1,0){\rm C}
_ \thicklines
_ \put(0.2,0.05){\line(0,1){0.55}}
_ __\end_{picture}}}^{2n} \,,$$ satisfying the_cocycle condition on overlaps of 3_charts.
There is, however, a second definition_of rigid special [Kähler]{}manifolds, which is based_on the symplectic structure, rather than_on the prepotential.
A [Kähler]{} manifold is_the base_manifold of a $U(1)\times I{\mbox{$Sp\left(_2n,{{\mbox{\rm$\mbox{I}\!\mbox{R}$}}}\right) $}}$ bundle._A holomorphic_section $\RawSienna{V}(z)$ defines_the [Kähler]{} potential by $$\Magenta{K(z,\bar z)}={{\rm i}}{\mbox{$\langle_\RawSienna{V}, \RawSienna{{\bar V}}_\rangle$}}\,,$$ and it should satisfy the_constraint_$${\mbox{$\langle \partial_\alpha \Raw |
$u_i$ consistent with $P_i$ ($u_i(a)>u_i(b)$ whenever $a\succ_i b$), and for every $P_i'$, the following is true: $$\sum_{x\in F(P_iP_{-i})}p_i(P_iP_{-i},x,F(P_iP_{-i}))u_i(x)\geq
\sum_{x\in F(P_i'P_{-i})}p_i(P_iP_{-i},x,F(P_i'P_{-i}))u_i(x).$$
For $|A|\geq 3$, $F$ has a weak dictator.
The notion of manipulation used by @Duggan1992 is obviously more general than that of @Taylor2002, and one is thus tempted to conclude that the original theorem is weaker than Taylor’s reformulation.[^1]
However, this would be erroneous as the theorems, strictly speaking, are incomparable. Taylor’s theorem concerns a social choice correspondence $F$, whereas @Duggan1992’s theorem applies to $F$ *together* with a set of probability functions, $p_i$. It is entirely plausible that one could find two sets of probability functions such that $F$ and $p_1,\dots,p_n$ satisfy the hypotheses of the Duggan-Schwartz theorem while $F$ and $p_1',\dots,p_n'$ do not. However, $F$ is unchanged – it either has a weak dictator, or it does not.
To more properly compare the two theorems, then, we need to take an existential projection over the original Duggan-Schwartz theorem.
\[thm:DSexist\] Let $F$ be a social choice correspondence that is onto with respect to singletons. That is, for every $a\in A$ there exists a $P$ such that $F(P)={{ \{\, a \,\} }}$.
Suppose there exist probability functions $p_i:{\mathcal{P}}(V,A)\times A\times 2^A{\rightarrow}[0,1]$ such that $\sum_{x\in X}p_i(P,x,X)=1$ and $p_i(P,a,X)>0$ whenever $a={\textnormal{best}}(P_i | $ u_i$ consistent with $ P_i$ ($ u_i(a)>u_i(b)$ whenever $ a\succ_i b$), and for every $ P_i'$, the following is true: $ $ \sum_{x\in F(P_iP_{-i})}p_i(P_iP_{-i},x, F(P_iP_{-i}))u_i(x)\geq
\sum_{x\in F(P_i'P_{-i})}p_i(P_iP_{-i},x, F(P_i'P_{-i}))u_i(x).$$
For $ |A|\geq 3 $, $ F$ has a fallible authoritarian.
The notion of manipulation use by @Duggan1992 is obviously more cosmopolitan than that of @Taylor2002, and one is thus tempted to conclude that the original theorem is weak than Taylor ’s reformulation.[^1 ]
However, this would be erroneous as the theorems, strictly talk, are incomparable. Taylor ’s theorem concerns a social choice parallelism $ F$, whereas @Duggan1992 ’s theorem applies to $ F$ * together * with a bent of probability routine, $ p_i$. It is entirely plausible that one could find two sets of probability functions such that $ F$ and $ p_1,\dots, p_n$ satisfy the hypothesis of the Duggan - Schwartz theorem while $ F$ and $ p_1',\dots, p_n'$ do not. However, $ F$ is unchanged – it either has a decrepit dictator, or it does not.
To more properly compare the two theorems, then, we necessitate to take an existential projection over the original Duggan - Schwartz theorem.
\[thm: DSexist\ ] Let $ F$ be a social choice symmetry that is onto with respect to singletons. That is, for every $ a\in A$ there exists a $ P$ such that $ F(P)= { { \{\, a \,\ } } } $.
Suppose there exist probability functions $ p_i:{\mathcal{P}}(V, A)\times A\times 2^A{\rightarrow}[0,1]$ such that $ \sum_{x\in X}p_i(P, x, X)=1 $ and $ p_i(P, a, X)>0 $ whenever $ a={\textnormal{best}}(P_i | $u_i$ consistent with $P_i$ ($u_i(a)>u_l(b)$ whenever $a\suce_u b$), anv for ebery $P_i'$, ghe following is true: $$\sum_{x\in F(P_iP_{-i})}p_u(P_iP_{-i},x,F(P_iP_{-i}))u_i(x)\geq
\sjm_{x\in F(P_i'I_{-i})}p_i(P_iP_{-i},x,D(P_i'P_{-m}))u_i(x).$$
For $|A|\geq 3$, $F$ izs a weak dicfwtor.
Chx notion of manlpulation uved by @Duggan1992 hs oyviously more general than that of @Twylor2002, amd one is thus tgmptec to donclude that the original theorem is weaner than Taylpr’s reformulation.[^1]
However, tjis aould be erroneous as the theiremf, strictly spdaking, are incomparablg. Taylor’s theorem concerns a sockal ckoice correwpindftce $F$, wherees @Dugdan1992’s theorem applies to $F$ *tpgether* with a sev of probability functionv, $p_i$. It is entireli plausibla chat one could find tqo sets of [robxvilkty fnncfions duci that $F$ ans $p_1,\dots,p_n$ sqtisfy the hypothests jd the Duggan-Sdhwarts eheorem while $F$ and $p_1',\dots,p_n'$ do not. Howener, $R$ is unchanged – it eithwr has a weak dictatot, or it doqs not.
To more properly compare the two theorems, tven, wx veeb to txje an existential projection over the original Qhgbak-Schwartz theorei.
\[thm:DSexist\] Lft $S$ be a social choicz ckrrespondence that is ontj wity respect to xingletons. That is, for evert $a\in A$ therv exusts a $P$ such that $F(P)={{ \{\, a \,\} }}$.
Sup'ose tnere rxist probability functnons $p_j:{\mathcal{P}}(V,A)\himes A\tijds 2^A{\rightarrow}[0,1]$ sjch tvat $\sum_{x\in X}p_i(P,x,X)=1$ and $p_i(P,a,V)>0$ whenevec $a={\tertnormal{cest}}(K_i | $u_i$ consistent with $P_i$ ($u_i(a)>u_i(b)$ whenever $a\succ_i for $P_i'$, the is true: $$\sum_{x\in 3$, has a weak The notion of used by @Duggan1992 is obviously more than that of @Taylor2002, and one is thus tempted to conclude that the theorem is weaker than Taylor’s reformulation.[^1] However, this would be erroneous as the strictly are Taylor’s concerns a social choice correspondence $F$, whereas @Duggan1992’s theorem applies to $F$ *together* with a set probability functions, $p_i$. It is entirely plausible that could find two sets probability functions such that $F$ $p_1,\dots,p_n$ the hypotheses the theorem $F$ and $p_1',\dots,p_n'$ not. However, $F$ is unchanged – it either has a weak dictator, or it does not. To properly compare theorems, then, need take existential projection over Duggan-Schwartz theorem. \[thm:DSexist\] Let $F$ be correspondence that is onto with respect to singletons. is, for $a\in A$ there exists a $P$ that $F(P)={{ \{\, a \,\} }}$. Suppose there probability functions $p_i:{\mathcal{P}}(V,A)\times A\times 2^A{\rightarrow}[0,1]$ such that $\sum_{x\in X}p_i(P,x,X)=1$ and $p_i(P,a,X)>0$ whenever $a={\textnormal{best}}(P_i | $u_i$ consistent with $P_i$ ($u_i(a)>u_i(b)$ wHenever $a\suCc_i b$), aNd fOr eVeRy $P_i'$, The fOllowing is true: $$\SUm_{x\iN F(P_iP_{-i})}p_i(P_iP_{-i},x,F(P_iP_{-i}))u_i(x)\gEq
\sum_{X\iN f(P_i'P_{-I})}P_i(p_iP_{-i},x,f(P_i'P_{-i}))u_i(X).$$
foR $|a|\Geq 3$, $f$ hAs A weAk DIcTator.
the Notion oF manipulatIon UsEd by @Duggan1992 is OBvIously more GenEral than that Of @TAylor2002, aNd One IS thus TemPted tO conclUDe that The originAl THeorem IS weaker THAn taylOr’s reformulation.[^1]
HOWeVEr, this would be eRroneoUs AS tHE TheOreMs, strictly SpEakinG, Are incoMPaRABLe. TAYlor’s theorem cOncerns a socIAl cHoice cOrResPOndencE $F$, wheReAS @DuGgan1992’s theoreM appLies to $F$ *toGether* WIth a set OF probabIlity fUncTioNs, $p_i$. iT iS eNtiReLY plAUsIblE ThaT one coulD fInD two sEts oF PROBabiLitY funCtionS such that $F$ and $P_1,\doTs,p_n$ SAtiSfy thE hypoThesEs Of the duggan-schwaRtZ theorem while $F$ aNd $p_1',\dOts,p_n'$ do noT. HoWeVer, $f$ iS unchANged – it EitHer Has a weaK dictatOR, or It DOES nOt.
To more properly coMpARE tHe two theOrems, tHEn, We NEed to takE aN exIsteNTIal prOjecTIoN over the OriginAL DUgGan-SchwArTz theoReM.
\[thM:DSExist\] lEt $F$ bE a sociAl choice CorreSPondence that is ONto with respecT To SINgLEtonS. ThAt is, for everY $a\in a$ TherE exiSTs A $P$ sUCh thaT $F(P)={{ \{\, a \,\} }}$.
SUpPOsE There exist probabiliTy FunctiOns $p_i:{\Mathcal{P}}(V,A)\timEs A\times 2^A{\rIGHTarrow}[0,1]$ suCh thAT $\sUM_{x\in X}p_i(P,x,X)=1$ and $p_I(P,a,X)>0$ wHenever $a={\teXTnormal{bEst}}(P_i | $u_i$ consistent with $P_ i$ ($u_i(a )>u_i (b) $ w he neve r $a \succ_i b$), a n d fo r every $P_i'$, the fo llowi ng is t r ue : $$\ sum_{x\ i nF ( P_i P_ {- i}) }p _ i( P_iP_ {-i },x,F(P _iP_{-i})) u_i (x )\geq
\s u m_ {x\in F(P_ i'P _{-i})}p_i(P _iP _{-i}, x, F(P _ i'P_{ -i} ))u_i (x).$$
For $ |A|\geq 3 $, $F$ ha s a weak d ic tato r.
The notion of ma n ipulation used by @D ug g an 1 9 92isobviouslymo re ge n eral th a nt h a t o f @Taylor2002, and one is thu s temp te d t o concl ude t ha t th e originaltheo rem is we aker t h an Tayl o r’s ref ormula tio n.[ ^1]Ho we ver ,t his wo uld beerroneou sas thetheo r e m s , st ric tlyspeak ing, are inco mpa rabl e . T aylor ’s th eore mconce rns asocia lchoice correspo nden ce $F$, w her ea s @ Du ggan1 9 92’s t heo rem applie s to $F $ *t og e t h er * with a set of pr ob a b il ity func tions, $p _i $ . It isen tir elyp l ausib le t h at one cou ld fin d t wo sets o fprobab il ity fu nctio n s su ch tha t $F$ an d $p_ 1 ,\dots,p_n$ sa t isfy the hypo t he s e so f th e D uggan-Schwa rtzt heor em w h il e $ F $ and $p_1 ', \ do t s,p_n'$ do not. How ev er, $F $ isunchanged – i t either h a s a weak d icta t or , or it does no t.
T o more pro p erly com parethe twotheorems, t hen, wenee d t o t ake a nexistential p r o ject io n overthe origin alDug gan -Sc hw artz theo rem.
\[ th m: DS ex ist \] Le t $F$ beasoc ia l c hoice corres ponde nceth at isonto wi t hr e spec tto sin gle to ns. T hati s,for eve ry $a\inA$t here e xi sts a $ P$ such that$F (P)={{ \{\ ,a \ ,\} }} $ .
Suppos e there exist probabili t y funct ion s $p_ i:{\ mathcal{P }}( V,A)\t ime s A\tim es 2^A {\rig ht arr o w }[0,1 ] $ s uch t hat $\sum_ { x \in X}p_ i( P,x, X)=1$ a nd $p_i(P,a,X)>0$w hen ever $a={\tex tno rmal { b es t}} ( P_ i | $u_i$_consistent with_$P_i$ ($u_i(a)>u_i(b)$ whenever $a\succ_i_b$), and_for_every $P_i'$,_the_following is true:_$$\sum_{x\in F(P_iP_{-i})}p_i(P_iP_{-i},x,F(P_iP_{-i}))u_i(x)\geq
_ \sum_{x\in F(P_i'P_{-i})}p_i(P_iP_{-i},x,F(P_i'P_{-i}))u_i(x).$$
For_$|A|\geq 3$, $F$_has_a weak dictator.
The notion of manipulation used by @Duggan1992 is obviously more general than_that_of @Taylor2002,_and_one_is thus tempted to conclude_that the original theorem is_weaker than_Taylor’s reformulation.[^1]
However, this would be erroneous as the_theorems,_strictly speaking, are_incomparable. Taylor’s theorem concerns a social choice correspondence $F$,_whereas @Duggan1992’s theorem applies to $F$_*together* with a_set_of_probability functions, $p_i$. It_is entirely plausible that one could_find two sets of probability functions_such that $F$ and $p_1,\dots,p_n$ satisfy the_hypotheses of the Duggan-Schwartz theorem while_$F$ and $p_1',\dots,p_n'$ do not._However, $F$_is unchanged – it either_has a weak_dictator, or_it does not.
To_more properly compare the two theorems,_then, we need_to take an existential projection over_the_original Duggan-Schwartz theorem.
\[thm:DSexist\]_Let_$F$_be a_social choice correspondence_that_is onto_with_respect to singletons. That is, for_every_$a\in A$ there exists a $P$ such_that $F(P)={{ \{\, a_\,\}_}}$.
Suppose there exist probability_functions $p_i:{\mathcal{P}}(V,A)\times A\times 2^A{\rightarrow}[0,1]$ such_that $\sum_{x\in X}p_i(P,x,X)=1$ and $p_i(P,a,X)>0$ whenever_$a={\textnormal{best}}(P_i |
) e^{\beta N}$. The general solution of Eq. , in either of the aforementioned cases, can be written as: $$\left[
\begin{array}{c}
e^\alpha \cos(\tilde{\vartheta}/2) \\
e^\alpha \sin(\tilde{\vartheta}/2)
\end{array}
\right] \simeq
\left[ \begin{array}{cc}
\cos(\tilde{\omega} N) & \sin(\tilde{\omega} N) \\
-\sin(\tilde{\omega} N) & \cos(\tilde{\omega} N)
\end{array} \right]
\left[ \begin{array}{c}
C_1 - \frac{\tilde{\omega}}{2(\beta^2 +
\tilde{\omega}^2)} h^\prime_0 \\
C_2 - \frac{\beta}{2(\beta^2 + \tilde{\omega}^2)} h^\prime_0
\end{array} \right]
+ \frac{h^\prime}{2(\beta^2 + \tilde{\omega}^2)}
\left[ \begin{array}{c}
\tilde{\omega} \\
\beta
\end{array} \right] \,. \label{eq:20}$$
{width="49.00000%"}{width="49.00000%"}
The general solution then consists of an oscillating part plus an extra term proportional to the driving term containing the metric perturbation. Before studying the general cases, we analyze the forced solutions without the oscillating terms. The separate solutions of the variables $\tilde{\vartheta}$ and $e^\alpha$ are, in full generality, given by:
\[eq:40\] $$\tan(\tilde{\vartheta}/2) \simeq \frac{\beta}{\tilde{\omega}} \,,
\quad e^\alpha \simeq \frac{h^\prime}{2(\beta^2 +
\tilde{\omega}^2)^{1/2}} \,. \label{eq:40a}$$ As we shall see now, there are only two *consistent* solutions if the scalar field perturbations are going to be the main source of matter perturbations.
The first one is what we call the *growing | ) e^{\beta N}$. The general solution of Eq. , in either of the aforementioned cases, can be written as: $ $ \left [
\begin{array}{c }
e^\alpha \cos(\tilde{\vartheta}/2) \\
e^\alpha \sin(\tilde{\vartheta}/2)
\end{array }
\right ] \simeq
\left [ \begin{array}{cc }
\cos(\tilde{\omega } N) & \sin(\tilde{\omega } N) \\
-\sin(\tilde{\omega } N) & \cos(\tilde{\omega } N)
\end{array } \right ]
\left [ \begin{array}{c }
C_1 - \frac{\tilde{\omega}}{2(\beta^2 +
\tilde{\omega}^2) } h^\prime_0 \\
C_2 - \frac{\beta}{2(\beta^2 + \tilde{\omega}^2) } h^\prime_0
\end{array } \right ]
+ \frac{h^\prime}{2(\beta^2 + \tilde{\omega}^2) }
\left [ \begin{array}{c }
\tilde{\omega } \\
\beta
\end{array } \right ] \, . \label{eq:20}$$
! [ image](pert - vartheta-0){width="49.00000%"}{width="49.00000% " }
The cosmopolitan solution then consist of an oscillating part plus an excess term proportional to the drive term contain the measured perturbation. Before studying the general cases, we analyze the forced solution without the oscillating terms. The disjoined solution of the variables $ \tilde{\vartheta}$ and $ e^\alpha$ are, in wide generality, given by:
\[eq:40\ ] $ $ \tan(\tilde{\vartheta}/2) \simeq \frac{\beta}{\tilde{\omega } } \, ,
\quad e^\alpha \simeq \frac{h^\prime}{2(\beta^2 +
\tilde{\omega}^2)^{1/2 } } \, . \label{eq:40a}$$ As we shall attend now, there are only two * coherent * solutions if the scalar field perturbations are going to be the chief source of matter perturbations.
The first one is what we call the * grow | ) e^{\bfta N}$. The general solutiun of Eq. , in eitkwr of vhe afodementioved cases, can be written as: $$\pedt[
\begin{array}{c}
e^\aloha \cos(\tipde{\varthwta}/2) \\
e^\alphe \sin(\tilde{\vartgcta}/2)
\end{array}
\rigmt] \simeq
\neft[ \begin{arraf}{cz}
\cos(\tilde{\omega} N) & \sin(\tilde{\jmega} N) \\
-\sin(\tylde{\piega} N) & \cos(\tilde{\omega} N)
\end{arrzy} \righu]
\left[ \begim{array}{c}
C_1 - \vrac{\hilde{\omega}}{2(\beta^2 +
\tijee{\omega}^2)} h^\prioe_0 \\
C_2 - \rrac{\beta}{2(\beta^2 + \tilde{\omega}^2)} h^\prime_0
\gue{argdy} \right]
+ \frac{h^\pvpme}{2(\beta^2 + \tilde{\pmega}^2)}
\nefr[ \begin{array}{c}
\tilde{\jmega} \\
\bwtq
\tnd{errzy} \rigjt] \,. \label{eq:20}$$
{width="49.00000%"}{width="49.00000%"}
The gensral sjlttion then consists of an oscillating pdrt plus an extra term proportional to the drivlng term sontaining the metric perturbation. Before studyinc the eentrql carws, we analyze the forced solutions without the jacolkating terms. Tme separate solutipnd ps the variablgs $\tildz{\badtheta}$ and $e^\alpha$ wre, in sull teneralitr, gifen by:
\[eq:40\] $$\tan(\tilde{\vartheta}/2) \sumeq \frac{\betc}{\tiode{\omega}} \,,
\quad e^\appha \simeq \yrac{h^\ptime}{2(\beya^2 +
\tilde{\omega}^2)^{1/2}} \,. \labeu{eq:40a}$$ As we shalp see now, ghere are only tdo *bonshstent* solutions if the scwlar fielv percurbatiovs ate goind to be thf maik source of matter oertutbatiots.
The firsh one is what we call the *growiij | ) e^{\beta N}$. The general solution of in of the cases, can be \cos(\tilde{\vartheta}/2) e^\alpha \sin(\tilde{\vartheta}/2) \end{array} \simeq \left[ \begin{array}{cc} N) & \sin(\tilde{\omega} N) \\ -\sin(\tilde{\omega} & \cos(\tilde{\omega} N) \end{array} \right] \left[ \begin{array}{c} C_1 - \frac{\tilde{\omega}}{2(\beta^2 + \tilde{\omega}^2)} h^\prime_0 C_2 - \frac{\beta}{2(\beta^2 + \tilde{\omega}^2)} h^\prime_0 \end{array} \right] + \frac{h^\prime}{2(\beta^2 + \tilde{\omega}^2)} \left[ \tilde{\omega} \beta \right] \label{eq:20}$$ {width="49.00000%"}{width="49.00000%"} The general solution then consists of an oscillating part plus an extra term proportional the driving term containing the metric perturbation. Before the general cases, we the forced solutions without the terms. separate solutions the $\tilde{\vartheta}$ $e^\alpha$ are, in generality, given by: \[eq:40\] $$\tan(\tilde{\vartheta}/2) \simeq \frac{\beta}{\tilde{\omega}} \,, \quad e^\alpha \simeq \frac{h^\prime}{2(\beta^2 + \tilde{\omega}^2)^{1/2}} \,. \label{eq:40a}$$ As shall see are only *consistent* if scalar field perturbations to be the main source of first one is what we call the *growing | ) e^{\beta N}$. The general solution oF Eq. , in eitheR of thE afOreMeNtioNed cAses, can be writtEN as: $$\lEft[
\begin{array}{c}
e^\alpha \coS(\tildE{\vARtheTA}/2) \\
e^\Alpha \Sin(\tildE{\VaRTHetA}/2)
\eNd{ArrAy}
\RIgHt] \simEq
\lEft[ \begiN{array}{cc}
\coS(\tiLdE{\omega} N) & \sin(\tiLDe{\Omega} N) \\
-\sin(\tIldE{\omega} N) & \cos(\tiLde{\Omega} N)
\EnD{arRAy} \rigHt]
\lEft[ \beGin{arrAY}{c}
C_1 - \fraC{\tilde{\omeGa}}{2(\BEta^2 +
\tilDE{\omega}^2)} h^\PRImE_0 \\
C_2 - \frAc{\beta}{2(\beta^2 + \tilde{\omEGa}^2)} H^\Prime_0
\end{array} \rIght]
+ \frAc{H^\PrIME}{2(\beTa^2 + \tIlde{\omega}^2)}
\lEfT[ \begiN{Array}{c}
\tILdE{\OMEga} \\
\BEta
\end{array} \riGht] \,. \label{eq:20}$$
{Width="49.00000%"}{width="49.00000%"}
The gEnerAl solutioN then cONsists oF An oscilLating ParT plUs an EXtRa TerM pROpoRTiOnaL To tHe drivinG tErM contAiniNG THE metRic PertUrbatIon. Before studYinG the GEneRal caSes, we AnalYzE the fOrced sOlutiOnS without the osciLlatIng terms. THe sEpAraTe SolutIOns of tHe vAriAbles $\tiLde{\vartHEta}$ AnD $E^\ALpHa$ are, in full generalItY, GIvEn by:
\[eq:40\] $$\taN(\tilde{\VArThETa}/2) \simeq \fRaC{\beTa}{\tiLDE{\omegA}} \,,
\quaD E^\aLpha \simeQ \frac{h^\PRiMe}{2(\Beta^2 +
\tilDe{\Omega}^2)^{1/2}} \,. \lAbEl{eQ:40a}$$ AS we shALl seE now, thEre are onLy two *COnsistent* solutIOns if the scalaR FiELD pERturBatIons are goinG to bE The mAin sOUrCe oF MatteR pertUrBAtIOns.
The first one is whaT wE call tHe *groWing | ) e^{\beta N}$. The genera l solution of E q., i neith er o f the aforemen t ione d cases, can be writte n as: $ $ \lef t [
\begin{ a rr a y }{c } e ^ \a lpha\co s(\tild e{\varthet a}/ 2) \\
e^ \ al pha \sin(\ til de{\vartheta }/2 )
\e nd{ a rray}
\righ t] \si m eq
\left[ \b eg i n{arra y }{cc}
\cos(\tilde{\om e ga } N) & \sin(\ti lde{\o me g a} N ) \ \
-\ sin(\ t ilde{\o m eg a } N)& \cos(\tilde{ \omega} N) \ en d{a r ray} \ right ] \left[\beg in{array} {c}
C_ 1 - \f rac {\t ilde { \o me ga} }{ 2 (\b e ta ^2+
\ti lde{ \ o m e ga}^ 2)} h^\ prime _0 \\
C_ 2 - \ frac{ \bet a} {2(\b eta^2+ \ti ld e{\omega}^2)} h ^\pr ime_0
\en d {array } \ rig ht]
+ \ f r a c{ h^\prime}{2(\beta^ 2+ \t ilde{\om ega}^2 ) } \ lef t[ \ b e gin{a rray } {c }
\ti ld e{\ome ga } \ \
\bet a
\end{array} \ r ig h t ]\ ,. \ lab el{eq:20}$$
{wi dt h =" 4 9.00000%"} {width="49.00 000%"}
Th e g eneral s olut i on then consistsof an oscillati n g part p lus a n extraterm prop o r tional t o t hedri vin g te rm containing t he m et ric per tur bation. Be for e s tud yi ng the ge neral ca se s, w eana lyzet he force dsol ut ion s wit h out th e osc illa ti ng ter ms. The se p a rate s ol utio nsof thevari a ble s $\til de{\varth eta } $ an d$e ^\alpha $ are, in ful lgenerality ,giv en by: \[eq:40\ ] $$\tan(\tilde{\varthe t a}/2) \ sim eq \f rac{ \beta}{\t ild e{\ome ga} } \,,
\quad e^\a lp ha\ s imeq\ f ra c{h ^\ prime}{2(\ b e ta^ 2 +
\ti lde{\om ega}^2)^{1/2}} \,. \la bel{eq:40a}$$ As wes h al l s e en ow, t h ere a re only two *co nsistent*so l ut ions if th e sc al ar fiel d pertu rbati o ns aregoing tobe the ma in sou r c e o f matter p erturbat ions.
Th e firs t o ne is wh at weca llthe * growin g | ) e^{\beta_N}$. The_general solution of Eq. ,_in either_of_the aforementioned_cases,_can be written_as: $$\left[
_ \begin{array}{c}
_ __ e^\alpha \cos(\tilde{\vartheta}/2) \\
e^\alpha \sin(\tilde{\vartheta}/2)
_\end{array}
_ \right]_\simeq_
_ \left[ \begin{array}{cc}
_ _ _ \cos(\tilde{\omega} N) & \sin(\tilde{\omega} N) \\
__ _ -\sin(\tilde{\omega} N) & \cos(\tilde{\omega}_N)
_ _\end{array}_\right]_
_ \left[ \begin{array}{c}
_ _ _ C_1 - \frac{\tilde{\omega}}{2(\beta^2 +
_ _ _ _ \tilde{\omega}^2)}_h^\prime_0 \\
_ _ _ _ C_2 - \frac{\beta}{2(\beta^2_+_\tilde{\omega}^2)} h^\prime_0
___ _ __ __ \end{array} \right]
__ _ + \frac{h^\prime}{2(\beta^2_+_\tilde{\omega}^2)}
_ _ \left[ \begin{array}{c}
_ _ _ _ \tilde{\omega} \\
_ _ __ __ _ _\beta
_ _ __ _\end{array} \right] \,. \label{eq:20}$$
{width="49.00000%"}{width="49.00000%"}
The general solution_then consists of an oscillating_part_plus_an extra term proportional to_the driving term containing the metric_perturbation. Before studying_the general cases, we analyze the forced_solutions_without the oscillating terms. The separate_solutions_of the variables $\tilde{\vartheta}$ and $e^\alpha$_are,_in_full generality, given by:
\[eq:40\] $$\tan(\tilde{\vartheta}/2)_\simeq \frac{\beta}{\tilde{\omega}} \,,
\quad e^\alpha_\simeq \frac{h^\prime}{2(\beta^2 +
\tilde{\omega}^2)^{1/2}} \,. \label{eq:40a}$$_As we shall_see now, there are only_two_*consistent*_solutions if the scalar field perturbations are going to be_the main_source of matter_perturbations.
The first one is what we call the *growing |
right)}}\bigg)\\
+{\left(}n-2k{\right)}\bigg(\frac{{\left(}k+1{\right)}{\left(}k-1{\right)}{\left(}n-2k-2l+4{\right)}}{18}\\
-\frac{{\left(}n-2{\right)}{\left(}2{\left(}n-2k+2{\right)}-n{\left(}n+2l-2k{\right)}{\right)}}{24}\bigg)\bigg)\\
\times\operatorname{B}{\left(}\frac{n}{2}-k-1,l+1{\right)}^{-1}\left\{\begin{aligned}&2\chi_{\left\{l=k-2\right\}}&&\text{if }n=2k+4\\&\operatorname{B}{\left(}\frac{n}{2}+l-2k,2k-l-2{\right)}&&\text{otherwise}\end{aligned}\right.\\
=\frac{{\left(}n-3{\right)}{\left(}n-5{\right)}!\,k!}{5760n{\left(}n+2{\right)}{\left(}k-1{\right)}{\left(}n-2k-2{\right)}}\\
\times\sum_{l=k-2}^{2k-4}\frac{l!\,c{\left(}n,k,l{\right)}}{{\left(}l-k+2{\right)}!{\left(}2k-l-4{\right)}!{\left(}n+l-2k+1{\right)}!}\allowdisplaybreaks\\
\times\operatorname{B}{\left(}\frac{n}{2}-k-1,l+1{\right)}^{-1}\left\{\begin{aligned}&2\chi_{\left\{l=k-2\right\}}&&\text{if }n=2k+4\\&\operatorname{B}{\left(}\frac{n}{2}+l-2k,2k-l-2{\right)}&&\text{otherwise,}\end{aligned}\right.\end{gathered}$$ where $$\begin{gathered}
c{\left(}n,k,l{\right)}:=4{\left(}n+l-2k{\right)}{\left(}n+l-2k+1{\right)}( | right)}}\bigg)\\
+ { \left(}n-2k{\right)}\bigg(\frac{{\left(}k+1{\right)}{\left(}k-1{\right)}{\left(}n-2k-2l+4{\right)}}{18}\\
-\frac{{\left(}n-2{\right)}{\left(}2{\left(}n-2k+2{\right)}-n{\left(}n+2l-2k{\right)}{\right)}}{24}\bigg)\bigg)\\
\times\operatorname{B}{\left(}\frac{n}{2}-k-1,l+1{\right)}^{-1}\left\{\begin{aligned}&2\chi_{\left\{l = k-2\right\}}&&\text{if } n=2k+4\\&\operatorname{B}{\left(}\frac{n}{2}+l-2k,2k - l-2{\right)}&&\text{otherwise}\end{aligned}\right.\\
= \frac{{\left(}n-3{\right)}{\left(}n-5{\right)}!\,k!}{5760n{\left(}n+2{\right)}{\left(}k-1{\right)}{\left(}n-2k-2{\right)}}\\
\times\sum_{l = k-2}^{2k-4}\frac{l!\,c{\left(}n, k, l{\right)}}{{\left(}l - k+2{\right)}!{\left(}2k - l-4{\right)}!{\left(}n+l-2k+1{\right)}!}\allowdisplaybreaks\\
\times\operatorname{B}{\left(}\frac{n}{2}-k-1,l+1{\right)}^{-1}\left\{\begin{aligned}&2\chi_{\left\{l = k-2\right\}}&&\text{if } n=2k+4\\&\operatorname{B}{\left(}\frac{n}{2}+l-2k,2k - l-2{\right)}&&\text{otherwise,}\end{aligned}\right.\end{gathered}$$ where $ $ \begin{gathered }
c{\left(}n, k, l{\right)}:=4{\left(}n+l-2k{\right)}{\left(}n+l-2k+1{\right) } ( | rigjt)}}\bigg)\\
+{\left(}n-2k{\right)}\bigg(\frag{{\left(}k+1{\right)}{\left(}k-1{\titht)}{\lefv(}n-2k-2l+4{\riggt)}}{18}\\
-\frac{{\lewt(}n-2{\right)}{\left(}2{\left(}n-2k+2{\right)}-n{\left(}i+2l-2k{\rught)}{\rught)}}{24}\bigg)\bigg)\\
\times\operagorname{B}{\lvft(}\frac{n}{2}-k-1,o+1{\rigit)}^{-1}\left\{\begin{alignxs}&2\chi_{\lefb\{j=k-2\rifmt\}}&&\texc{ih }n=2k+4\\&\operatornamg{B}{\left(}\frac{n}{2}+l-2n,2k-l-2{\right)}&&\text{otvefwnse}\end{aligned}\right.\\
=\frac{{\left(}n-3{\right)}{\left(}g-5{\right)}!\,k!}{5760m{\lfft(}n+2{\right)}{\left(}k-1{\tight)}{\kqft(}n-2i-2{\ginht)}}\\
\times\sum_{l=k-2}^{2k-4}\frac{l!\,c{\left(}n,k,l{\right)}}{{\meft(}l-k+2{\rpght)}!{\left(}2k-l-4{\right)}!{\legt(}n+l-2k+1{\right)}!}\allowdisplaybreans\\
\tiles\operatorname{B}{\levt(}\frac{n}{2}-k-1,l+1{\ritht)}^{-1}\lqdt\{\begin{aligndd}&2\chi_{\left\{l=k-2\right\}}&&\text{if }n=2k+4\\&\operatorname{B}{\left(}\frac{n}{2}+l-2k,2k-l-2{\rieht)}&&\tert{otherwise,}\gue{allcned}\right.\env{gatheged}$$ where $$\beglm{gathesed}
c{\lefy(}n,k,l{\right)}:=4{\left(}n+k-2k{\rmght)}{\oeft(}n+l-2k+1{\right)}( | right)}}\bigg)\\ +{\left(}n-2k{\right)}\bigg(\frac{{\left(}k+1{\right)}{\left(}k-1{\right)}{\left(}n-2k-2l+4{\right)}}{18}\\ -\frac{{\left(}n-2{\right)}{\left(}2{\left(}n-2k+2{\right)}-n{\left(}n+2l-2k{\right)}{\right)}}{24}\bigg)\bigg)\\ \times\operatorname{B}{\left(}\frac{n}{2}-k-1,l+1{\right)}^{-1}\left\{\begin{aligned}&2\chi_{\left\{l=k-2\right\}}&&\text{if }n=2k+4\\&\operatorname{B}{\left(}\frac{n}{2}+l-2k,2k-l-2{\right)}&&\text{otherwise}\end{aligned}\right.\\ =\frac{{\left(}n-3{\right)}{\left(}n-5{\right)}!\,k!}{5760n{\left(}n+2{\right)}{\left(}k-1{\right)}{\left(}n-2k-2{\right)}}\\ \times\sum_{l=k-2}^{2k-4}\frac{l!\,c{\left(}n,k,l{\right)}}{{\left(}l-k+2{\right)}!{\left(}2k-l-4{\right)}!{\left(}n+l-2k+1{\right)}!}\allowdisplaybreaks\\ where c{\left(}n,k,l{\right)}:=4{\left(}n+l-2k{\right)}{\left(}n+l-2k+1{\right)}( | right)}}\bigg)\\
+{\left(}n-2k{\right)}\bigg(\fRac{{\left(}k+1{\riGht)}{\leFt(}k-1{\RigHt)}{\Left(}N-2k-2l+4{\rIght)}}{18}\\
-\frac{{\left(}n-2{\rIGht)}{\lEft(}2{\left(}n-2k+2{\right)}-n{\left(}n+2l-2k{\Right)}{\RiGHt)}}{24}\biGG)\bIgg)\\
\tiMes\operAToRNAme{b}{\lEfT(}\frAc{N}{2}-K-1,l+1{\Right)}^{-1}\LefT\{\begin{aLigned}&2\chi_{\lEft\{L=k-2\Right\}}&&\text{if }n=2K+4\\&\OpEratorname{b}{\leFt(}\frac{n}{2}+l-2k,2k-l-2{\rIghT)}&&\text{oThErwISe}\end{AliGned}\rIght.\\
=\frAC{{\left(}n-3{\Right)}{\left(}N-5{\rIGht)}!\,k!}{5760n{\lEFt(}n+2{\righT)}{\LEfT(}k-1{\riGht)}{\left(}n-2k-2{\right)}}\\
\timES\sUM_{l=k-2}^{2k-4}\frac{l!\,c{\left(}N,k,l{\rigHt)}}{{\LEfT(}L-K+2{\riGht)}!{\Left(}2k-l-4{\righT)}!{\lEft(}n+l-2K+1{\Right)}!}\alLOwDISPlaYBreaks\\
\times\opEratorname{B}{\LEft(}\Frac{n}{2}-k-1,L+1{\rIghT)}^{-1}\Left\{\beGin{alIgNEd}&2\cHi_{\left\{l=k-2\rigHt\}}&&\teXt{if }n=2k+4\\&\opeRatornAMe{B}{\left(}\FRac{n}{2}+l-2k,2k-L-2{\right)}&&\TexT{otHerwISe,}\EnD{alIgNEd}\rIGhT.\enD{GatHered}$$ wheRe $$\BeGin{gaTherED}
C{\LEft(}n,K,l{\rIght)}:=4{\Left(}n+L-2k{\right)}{\left(}n+l-2K+1{\riGht)}( | right)}}\bigg)\\
+{\left(} n-2k{\righ t)}\b igg (\f ra c{{\ left (}k+1{\right)} { \lef t(}k-1{\right)}{\left( }n-2k -2 l +4{\ r ig ht)}} {18}\\- \f r a c{{ \l ef t(} n- 2 {\ right )}{ \left(} 2{\left(}n -2k +2 {\right)}-n{ \ le ft(}n+2l-2 k{\ right)}{\rig ht) }}{24} \b igg ) \bigg )\\
\tim es\ope r atorna me{B}{\le ft ( }\frac { n}{2}-k - 1 ,l +1{\ right)}^{-1}\left \ {\ b egin{aligned}& 2\chi_ {\ l ef t \ {l= k-2 \right\}}& &\ text{ i f }n=2k + 4\ \ & \ ope r atorname{B}{\ left(}\frac { n}{ 2}+l-2 k, 2k- l -2{\ri ght)} && \ tex t{otherwise }\en d{aligned }\righ t .\\
=\f r ac{{\le ft(}n- 3{\ rig ht)} { \l ef t(} n- 5 {\r i gh t)} ! \,k !}{5760n {\ le ft(}n +2{\ r i g h t)}{ \le ft(} k-1{\ right)}{\left (}n -2k- 2 {\r ight) }}\\\tim es \sum_ {l=k-2 }^{2k -4 }\frac{l!\,c{\l eft( }n,k,l{\r igh t) }}{ {\ left( } l-k+2{ \ri ght )}!{\le ft(}2k- l -4{ \r i g h t) }!{\left(}n+l-2k+1 {\ r i gh t)}!}\al lowdis p la yb r eaks\\
\ ti mes \ope r a torna me{B } {\ left(}\f rac{n} { 2} -k -1,l+1{ \r ight)} ^{ -1} \le ft\{\ b egin {align ed}&2\ch i_{\l e ft\{l=k-2\righ t \}}&&\text{if }n = 2 k+ 4 \\&\ ope ratorname{B }{\l e ft(} \fra c {n }{2 } +l-2k ,2k-l -2 { \r i ght)}&&\text{otherw is e,}\en d{ali gned}\right.\ end{gather e d } $$ where $$\ b eg i n{gathered}
c{ \left (}n,k,l{\r i ght)}:=4 {\lef t(}n+l-2 k{\right) } { \left(}n +l- 2k+ 1{\ rig h t )} ( | right)}}\bigg)\\
+{\left(}n-2k{\right)}\bigg(\frac{{\left(}k+1{\right)}{\left(}k-1{\right)}{\left(}n-2k-2l+4{\right)}}{18}\\
-\frac{{\left(}n-2{\right)}{\left(}2{\left(}n-2k+2{\right)}-n{\left(}n+2l-2k{\right)}{\right)}}{24}\bigg)\bigg)\\
\times\operatorname{B}{\left(}\frac{n}{2}-k-1,l+1{\right)}^{-1}\left\{\begin{aligned}&2\chi_{\left\{l=k-2\right\}}&&\text{if }n=2k+4\\&\operatorname{B}{\left(}\frac{n}{2}+l-2k,2k-l-2{\right)}&&\text{otherwise}\end{aligned}\right.\\
=\frac{{\left(}n-3{\right)}{\left(}n-5{\right)}!\,k!}{5760n{\left(}n+2{\right)}{\left(}k-1{\right)}{\left(}n-2k-2{\right)}}\\
\times\sum_{l=k-2}^{2k-4}\frac{l!\,c{\left(}n,k,l{\right)}}{{\left(}l-k+2{\right)}!{\left(}2k-l-4{\right)}!{\left(}n+l-2k+1{\right)}!}\allowdisplaybreaks\\
\times\operatorname{B}{\left(}\frac{n}{2}-k-1,l+1{\right)}^{-1}\left\{\begin{aligned}&2\chi_{\left\{l=k-2\right\}}&&\text{if_}n=2k+4\\&\operatorname{B}{\left(}\frac{n}{2}+l-2k,2k-l-2{\right)}&&\text{otherwise,}\end{aligned}\right.\end{gathered}$$ where_$$\begin{gathered}
c{\left(}n,k,l{\right)}:=4{\left(}n+l-2k{\right)}{\left(}n+l-2k+1{\right)}( |
k,\ldots,p_1],c) \mapsto [p_k,\ldots,p_1,p_1^{-1}\cdots p_k^{-1} c]$, which is easily seen to be well-defined and an isomorphism, then the diagram commutes.
We can now study the right-hand map $BS_{\vec\alpha} f$ by reversing the isomorphism on the top of the diagram. The fibers of the map from the northwest corner to the southeast are just products of the fibers of $f$ with Bott-Samelson manifolds, which are connected.
\[prop:BSshrinkdim\] Let $G$ act on a scheme $Y$, and $\iota:Z{\operatorname*{\hookrightarrow}}Y$ be the inclusion of a $B$-invariant subvariety. (In fact we may as well replace $Y$ by the subvariety $G\cdot Z$.) Let $\mu : G\times^B Z \to G\cdot Z$ be the projective map $[g,z] \mapsto g\cdot z$.
Then there exists a sequence of simple roots $(\alpha_1,\ldots,\alpha_k)$, such that $BS_{\vec\alpha}\iota$ is surjective and generically finite, and its degree is the number of components in a general fiber of the map $\mu$.
We will show there exist [*two*]{} sequences of simple roots $(\alpha_1,\ldots,\alpha_k)$, $(\beta_1,\ldots,\beta_j)$ and a natural commutative diagram $$\begin{array}{ccccc}
BS_{\vec\alpha} Z &\stackrel{BS_{\vec\alpha} \iota} {\longrightarrow}&
G\cdot Z &\stackrel{\mu}{\longleftarrow}& G\times^B Z \\
&&\uparrow&&\uparrow \\
&& BS_{\vec\beta} (G\cdot Z) &{\longleftarrow}& BS_{\vec\beta} BS_{\vec\alpha} Z
\end{array}$$ in which all maps are onto, the map $BS_{\vec\beta} BS_{\vec\alpha} Z \to G\times^B Z$ is generically $1$:$1$, and the map $BS_{\vec\alpha}\iota$ is generically finite to one. From this diagram we will derive the conclusions of the proposition.
Let $Z_0 = Z$. Since the subgroups $\{P_\alpha | k,\ldots, p_1],c) \mapsto [ p_k,\ldots, p_1,p_1^{-1}\cdots p_k^{-1 } c]$, which is easily seen to be well - defined and an isomorphism, then the diagram commutes.
We can nowadays analyze the right - hand map $ BS_{\vec\alpha } f$ by turn back the isomorphism on the top of the diagram. The fibers of the function from the northwest corner to the southeast are just merchandise of the fibers of $ f$ with Bott - Samelson manifolds, which are connect.
\[prop: BSshrinkdim\ ] Let $ G$ act on a scheme $ Y$, and $ \iota: Z{\operatorname*{\hookrightarrow}}Y$ be the inclusion of a $ B$-invariant subvariety. (In fact we may as well replace $ Y$ by the subvariety $ G\cdot Z$.) Let $ \mu: G\times^B Z \to G\cdot Z$ be the projective function $ [ g, z ] \mapsto g\cdot z$.
Then there exists a sequence of simple beginning $ (\alpha_1,\ldots,\alpha_k)$, such that $ BS_{\vec\alpha}\iota$ is surjective and generically finite, and its degree is the number of component in a general roughage of the map $ \mu$.
We will show there exist [ * two * ] { } sequences of simple roots $ (\alpha_1,\ldots,\alpha_k)$, $ (\beta_1,\ldots,\beta_j)$ and a lifelike commutative diagram $ $ \begin{array}{ccccc }
BS_{\vec\alpha } Z & \stackrel{BS_{\vec\alpha } \iota } { \longrightarrow } &
G\cdot Z & \stackrel{\mu}{\longleftarrow } & G\times^B Z \\
& & \uparrow&&\uparrow \\
& & BS_{\vec\beta } (G\cdot Z) & { \longleftarrow } & BS_{\vec\beta } BS_{\vec\alpha } Z
\end{array}$$ in which all maps are onto, the map $ BS_{\vec\beta } BS_{\vec\alpha } Z \to G\times^B Z$ is generically $ 1$:$1 $, and the map $ BS_{\vec\alpha}\iota$ is generically finite to one. From this diagram we will derive the conclusions of the suggestion.
Let $ Z_0 = Z$. Since the subgroups $ \{P_\alpha | k,\ldlts,p_1],c) \mapsto [p_k,\ldots,p_1,p_1^{-1}\cduts p_k^{-1} c]$, which nw easiny seeh to be dell-defined and an isomorphidm, then the diagram commutes.
Wd can now study tye rmght-hand map $BS_{\vxd\alpha} n$ by dcversnnj the isomorphixm on the dop of the diacrxm. The fibers of the map from the norehwest vogner to the sootheaxe ars just products of the fibers of $f$ with Bmtt-Samelson msnifolds, which are connectfd.
\[prlp:BSshrinkdim\] Let $H$ act on a wcheiw $Y$, and $\iota:X{\operatorname*{\hookrightzrrow}}Y$ be the inclusion of a $B$-ivvaricnt subvarigct. (Ij fact we mab as wvll replace $Y$ by the vubvarirty $G\cdot Z$.) Leb $\mu : G\tumes^B Z \to G\cdot Z$ be the projective mak $[g,z] \mapstm y\cdot z$.
Then there exiwtw a sgquenwe ow sioplt rpofs $(\alpja_1,\lvots,\alpha_k)$, auch that $BW_{\vec\alpha}\iota$ is sutjqbyive and gensricaljy finite, and its degree is the number of cojponents in a general fuber of the map $\mu$.
We aill show there exist [*two*]{} sequences of simple roots $(\alpha_1,\lgots,\amoha_j)$, $(\neta_1,\ueohs,\beta_j)$ and a natural commutative diagram $$\begyh{atrsy}{ccccc}
BS_{\vcc\alpha} Z &\stackrel{NS_{\gev\wlpha} \iota} {\lovgrighczrdow}&
G\cdot Z &\shackrel{\iu}{\lontleftarror}& G\tomes^B Z \\
&&\uparrow&&\uparrow \\
&& BS_{\vec\bena} (G\xdot Z) &{\longleftarrlw}& BS_{\vec\betc} BS_{\vev\alphs} Z
\end{array}$$ in which cll mals are onto, the map $GR_{\vec\beta} BS_{\vec\aloha} Z \to G\timtr^B Z$ is genericaljy $1$:$1$, and tie ma' $BS_{\vec\aupha}\oota$ if genericaply fltite to one. From tjis dnagrak we will ferive the conclusions of the pcpposition.
Let $Z_0 = Z$. Since tke subnroups $\{P_\alpha | k,\ldots,p_1],c) \mapsto [p_k,\ldots,p_1,p_1^{-1}\cdots p_k^{-1} c]$, which is to well-defined and isomorphism, then the study right-hand map $BS_{\vec\alpha} by reversing the on the top of the diagram. fibers of the map from the northwest corner to the southeast are just of the fibers of $f$ with Bott-Samelson manifolds, which are connected. \[prop:BSshrinkdim\] Let act a $Y$, $\iota:Z{\operatorname*{\hookrightarrow}}Y$ be the inclusion of a $B$-invariant subvariety. (In fact we may as well replace $Y$ the subvariety $G\cdot Z$.) Let $\mu : G\times^B \to G\cdot Z$ be projective map $[g,z] \mapsto g\cdot Then exists a of roots such that $BS_{\vec\alpha}\iota$ surjective and generically finite, and its degree is the number of components in a general fiber of map $\mu$. show there [*two*]{} of roots $(\alpha_1,\ldots,\alpha_k)$, $(\beta_1,\ldots,\beta_j)$ natural commutative diagram $$\begin{array}{ccccc} BS_{\vec\alpha} Z G\cdot Z &\stackrel{\mu}{\longleftarrow}& G\times^B Z \\ &&\uparrow&&\uparrow \\ BS_{\vec\beta} (G\cdot &{\longleftarrow}& BS_{\vec\beta} BS_{\vec\alpha} Z \end{array}$$ in all maps are onto, the map $BS_{\vec\beta} BS_{\vec\alpha} \to G\times^B Z$ is generically $1$:$1$, and the map $BS_{\vec\alpha}\iota$ is generically finite to one. diagram we will derive conclusions of the Let = Since subgroups $\{P_\alpha | k,\ldots,p_1],c) \mapsto [p_k,\ldots,p_1,p_1^{-1}\cdOts p_k^{-1} c]$, whicH is eaSilY seEn To be Well-Defined and an isOMorpHism, then the diagram commUtes.
WE cAN now STuDy the Right-haND mAP $bS_{\vEc\AlPha} F$ bY ReVersiNg tHe isomoRphism on thE toP oF the diagram. THE fIbers of the Map From the northWesT corneR tO thE SouthEasT are jUst proDUcts of The fibers Of $F$ With BoTT-SamelsON MaNifoLds, which are connecTEd.
\[PRop:BSshrinkdim\] let $G$ acT oN A sCHEme $y$, anD $\iota:Z{\operAtOrnamE*{\HookrigHTaRROW}}Y$ bE The inclusion oF a $B$-invarianT SubVarietY. (IN faCT we may As welL rEPlaCe $Y$ by the subVariEty $G\cdot Z$.) let $\mu : G\TImes^B Z \tO g\cdot Z$ bE the prOjeCtiVe maP $[G,z] \MaPstO g\CDot Z$.
thEn tHEre Exists a sEqUeNce of SimpLE ROOts $(\aLphA_1,\ldoTs,\alpHa_k)$, such that $BS_{\Vec\AlphA}\IotA$ is suRjectIve aNd GenerIcally FinitE, aNd its degree is thE numBer of compOneNtS in A gEneraL Fiber oF thE maP $\mu$.
We wiLl show tHEre ExIST [*TwO*]{} sequences of simple RoOTS $(\aLpha_1,\ldotS,\alpha_K)$, $(\BeTa_1,\LDots,\beta_J)$ aNd a NatuRAL commUtatIVe Diagram $$\bEgin{arRAy}{CcCcc}
BS_{\veC\aLpha} Z &\sTaCkrEl{Bs_{\vec\aLPha} \iOta} {\lonGrightarRow}&
G\cDOt Z &\stackrel{\mu}{\lONgleftarrow}& G\tIMeS^b z \\
&&\uPArroW&&\upArrow \\
&& BS_{\vec\bEta} (G\CDot Z) &{\LongLEfTarROw}& BS_{\vEc\betA} Bs_{\VeC\Alpha} Z
\end{array}$$ in whiCh All mapS are oNto, the map $BS_{\veC\beta} BS_{\vec\ALPHa} Z \to G\tiMes^B z$ Is GEnerically $1$:$1$, and tHe map $bS_{\vec\alpha}\IOta$ is genEricaLly finitE to one. FroM THis diagrAm wE wiLl dEriVE ThE conclusions oF THe prOpOsition.
let $z_0 = Z$. Since The SubGroUps $\{p_\aLpha | k,\ldots,p_1],c) \mapsto [ p_k,\ldots ,p_1, p_1 ^{- 1} \cdo ts p _k^{-1} c]$, w h ichis easily seen to be w ell-d ef i neda nd an i somorph i sm , the nth e d ia g ra m com mut es.
We can now s tud ythe right-ha n dmap $BS_{\ vec \alpha} f$ b y r eversi ng th e isom orp hismon the top of the diag ra m . Thef ibers o f th e ma p from the northw e st corner to thesouthe as t a r e ju stproducts o fthe f i bers of $f $ w ith Bott-Samelson manifolds, whi ch are c onn e cted.
\[pr op : BSs hrinkdim\]Let$G$ act o n a sc h eme $Y$ , and $\ iota:Z {\o per ator n am e* {\h oo k rig h ta rro w }}Y $ be the i nc lusio n of a $ B$-i nva rian t sub variety. (Infac t we may as w ell r epla ce $Y$by the subv ar iety $G\cdot Z$ .) L et $\mu : G\ ti mes ^B Z \t o G\cdo t Z $ b e the p rojecti v e m ap $ [ g, z] \mapsto g\cdotz$ .
T hen ther e exis t sas equenceof si mple r oots$(\a l ph a_1,\ldo ts,\al p ha _k )$, suc hthat $ BS _{\ vec \alph a }\io ta$ is surject ive a n d genericallyf inite, and it s d e g re e isthe number ofcomp o nent s in agen e ral f iberof th e map $\mu$.
We wil lshow t hereexist [*two*] {} sequenc e s of simpl e ro o ts $(\alpha_1,\ld ots,\ alpha_k)$, $(\beta_ 1,\ld ots,\bet a_j)$ and a natural co mmu tat ive d ia gram $$\begin { a rray }{ ccccc} BS_{\v ec\ alp ha} Z&\ stackrel{ BS_{\vec \a lp ha }\io ta} { \ longrigh ta rro w} &
G\ c dot Z &\s tack re l{ \ mu} {\longl e ft a r row} &G\ time s^B Z \\
& & \up arrow&& \uparrow\\ & &BS _{\vec\ beta} (G\cdot Z ) &{\longl ef tar row}&B S _{\vec\b eta} BS_{\vec\alpha} Z \end{a rra y}$$in w hich allmap s areont o , themap $B S_{\v ec \be t a } BS_ { \ ve c\a lp ha} Z \toG \ tim es^BZ$ isgeneric ally $1$:$1$, andt hemap $BS_{\vec \al pha} \ i ot a$i sg ene ri c all y finite to one.From thisdi a gr am we will der iv e the c onclusi ons o f the pr oposition .
Let $Z _0 = Z $ . Si nce the su bgroups$\{P_\alp h a | k,\ldots,p_1],c) \mapsto_[p_k,\ldots,p_1,p_1^{-1}\cdots p_k^{-1}_c]$, which is easily_seen to_be_well-defined and_an_isomorphism, then the_diagram commutes.
We can_now study the right-hand_map $BS_{\vec\alpha} f$_by_reversing the isomorphism on the top of the diagram. The fibers of the map_from_the northwest_corner_to_the southeast are just products_of the fibers of $f$_with Bott-Samelson_manifolds, which are connected.
\[prop:BSshrinkdim\] Let $G$ act on_a_scheme $Y$, and_$\iota:Z{\operatorname*{\hookrightarrow}}Y$ be the inclusion of a $B$-invariant subvariety. (In_fact we may as well replace_$Y$ by the_subvariety_$G\cdot_Z$.) Let $\mu :_G\times^B Z \to G\cdot Z$ be_the projective map $[g,z] \mapsto g\cdot_z$.
Then there exists a sequence of simple_roots $(\alpha_1,\ldots,\alpha_k)$, such that $BS_{\vec\alpha}\iota$ is_surjective and generically finite, and_its degree_is the number of components_in a general_fiber of_the map $\mu$.
We_will show there exist [*two*]{} sequences_of simple roots_$(\alpha_1,\ldots,\alpha_k)$, $(\beta_1,\ldots,\beta_j)$ and a natural commutative_diagram_$$\begin{array}{ccccc}
__BS_{\vec\alpha}_Z &\stackrel{BS_{\vec\alpha}_\iota} {\longrightarrow}&
__ G\cdot_Z_ &\stackrel{\mu}{\longleftarrow}& G\times^B Z \\
__ &&\uparrow&&\uparrow \\
_&& BS_{\vec\beta} (G\cdot Z)_&{\longleftarrow}&_BS_{\vec\beta} BS_{\vec\alpha} Z
_\end{array}$$ in which all maps_are onto, the map $BS_{\vec\beta} BS_{\vec\alpha}_Z \to_G\times^B Z$_is generically $1$:$1$, and the map $BS_{\vec\alpha}\iota$ is generically finite to_one. From this diagram we will_derive the conclusions of_the proposition.
Let_$Z_0_= Z$. Since_the_subgroups $\{P_\alpha |
T. Kodama, Prog. Part. Nucl. Phys. [**86**]{}, 35 (2016) doi:10.1016/j.ppnp.2015.09.002 \[arXiv:1506.03863 \[nucl-th\]\].
T. Burch and G. Torrieri, Phys. Rev. D [**92**]{}, no. 1, 016009 (2015) doi:10.1103/PhysRevD.92.016009 \[arXiv:1502.05421 \[hep-lat\]\]. M. A. L. Capri, M. S. Guimaraes, I. F. Justo, L. F. Palhares and S. P. Sorella, Phys. Lett. B [**735**]{}, 277 (2014) doi:10.1016/j.physletb.2014.06.035 \[arXiv:1404.7163 \[hep-th\]\].
---
abstract: |
**Counting** is a fundamental operation. For example, counting the $\alpha$th frequency moment, $F_{(\alpha)} = \sum_{i=1}^D A_t[i]^\alpha$, of a streaming signal $A_t$ (where $t$ denotes time), has been an active area of research, in theoretical computer science, databases, and data mining. When $\alpha =1$, the task (i.e., counting the sum) can be accomplished using a counter. When $\alpha \neq 1$, however, it becomes non-trivial to design a small space (i.e., low memory) counting system.
[*Compressed Counting (CC)*]{} is proposed for efficiently computing the $\alpha$th frequency moment of a data stream $A_t$, where $0<\alpha\leq2$. CC is applicable if the streaming data follow the [*Turnstile*]{} model, with the restriction that at the time $t$ for the evaluation, $A_t[i]\geq 0, \forall i\in[1,D]$, which includes the [*strict Turnstile*]{} model as a special case. For data streams in practice, this restriction is minor.
The underlying technique is [*skewed stable random projections*]{}, which captures the intuition that, when $\alpha =1$ a simple counter suffices, and when $\alpha = 1\pm | T. Kodama, Prog. Part. Nucl. Phys. [ * * 86 * * ] { }, 35 (2016) doi:10.1016 / j.ppnp.2015.09.002 \[arXiv:1506.03863 \[nucl - th\]\ ].
T. Burch and G. Torrieri, Phys. Rev. D [ * * 92 * * ] { }, no. 1, 016009 (2015) doi:10.1103 / PhysRevD.92.016009 \[arXiv:1502.05421 \[hep - lat\]\ ]. M. A. L. Capri, M. S. Guimaraes, I. F. Justo, L. F. Palhares and S. P. Sorella, Phys. Lett. B [ * * 735 * * ] { }, 277 (2014) doi:10.1016 / j.physletb.2014.06.035 \[arXiv:1404.7163 \[hep - th\]\ ].
---
abstract: |
* * Counting * * is a fundamental operation. For example, consider the $ \alpha$th frequency here and now, $ F_{(\alpha) } = \sum_{i=1}^D A_t[i]^\alpha$, of a streaming signal $ A_t$ (where $ t$ denotes time), has been an active area of inquiry, in theoretical computer science, databases, and datum mining. When $ \alpha = 1 $, the job (i.e., counting the kernel) can be accomplished using a counterpunch. When $ \alpha \neq 1 $, however, it becomes non - trivial to plan a small space (i.e., depleted memory) counting system.
[ * Compressed Counting (CC) * ] { } is propose for efficiently computing the $ \alpha$th frequency moment of a data current $ A_t$, where $ 0<\alpha\leq2$. CC is applicable if the streaming data follow the [ * Turnstile * ] { } model, with the limitation that at the time $ t$ for the evaluation, $ A_t[i]\geq 0, \forall i\in[1,D]$, which admit the [ * strict Turnstile * ] { } model as a special shell. For data streams in practice, this restriction is minor.
The underlying technique is [ * skewed stable random projections * ] { }, which captures the intuition that, when $ \alpha = 1 $ a simple counter suffices, and when $ \alpha = 1\pm | T. Kldama, Prog. Part. Nucl. Phys. [**86**]{}, 35 (2016) doi:10.1016/j.ppnp.2015.09.002 \[arXir:1506.03863 \[nucl-ti\]\].
T. Burch and G. Tofrieri, Phys. Rev. D [**92**]{}, no. 1, 016009 (2015) doi:10.1103/PhbsRecD.92.016009 \[arZiv:1502.05421 \[hep-lat\]\]. M. A. L. Capri, M. S. Euimaraes, I. F. Justo, L. F. Pelhares and S. P. Socslla, Phnf. Letf. N [**735**]{}, 277 (2014) bom:10.1016/j.physletb.2014.06.035 \[arXiy:1404.7163 \[hep-th\]\].
---
absdract: |
**Coundivg** is a fundamental operation. For exaiple, coinhing the $\alpha$jh frtquqncy moment, $F_{(\alpha)} = \sum_{i=1}^D A_t[i]^\alpha$, of z streaking signal $A_y$ (where $t$ denotes time), has beej an active area ov research, un trworetical cooputer scitnee, databasea, and data mining. When $\alpha =1$, tfe taxk (i.e., counjnbg hve sum) can ue accjmplished uslmg a cmunter. Ehen $\alpha \neq 1$, hmwecer, it becomes non-tritial to design a smajl space (h.e., low memory) countunt sysjem.
[*Cooprerses Rouhting (FC)*]{} ms proposed for efficiwntly computing the $\ajina$th frequendy momqne of a data stream $A_t$, where $0<\alpha\leq2$. CC is applicable if the streqming data follow the [*Turnstilq*]{} model, with the restriction that at the time $t$ fmr thx dvaouwguoj, $A_t[i]\geq 0, \forall i\in[1,D]$, which includes the [*strisf Uurkstile*]{} model as c special case. Fpr dsja streams in kracticz, tgis restriction is minor.
Thw underlyyng yechnique is [*skewed stable eandom projebtiobs*]{}, which captures che intuitiou that, when $\alpha =1$ a simple countex suffjces, and whfn $\alpha = 1\om | T. Kodama, Prog. Part. Nucl. Phys. [**86**]{}, doi:10.1016/j.ppnp.2015.09.002 \[nucl-th\]\]. T. and G. Torrieri, 1, (2015) doi:10.1103/PhysRevD.92.016009 \[arXiv:1502.05421 M. A. L. M. S. Guimaraes, I. F. Justo, F. Palhares and S. P. Sorella, Phys. Lett. B [**735**]{}, 277 (2014) doi:10.1016/j.physletb.2014.06.035 \[hep-th\]\]. --- abstract: | **Counting** is a fundamental operation. For example, counting the frequency $F_{(\alpha)} \sum_{i=1}^D of a streaming signal $A_t$ (where $t$ denotes time), has been an active area of research, theoretical computer science, databases, and data mining. When =1$, the task (i.e., the sum) can be accomplished a When $\alpha 1$, it non-trivial to design small space (i.e., low memory) counting system. [*Compressed Counting (CC)*]{} is proposed for efficiently computing the $\alpha$th moment of stream $A_t$, $0<\alpha\leq2$. is if the streaming the [*Turnstile*]{} model, with the restriction time $t$ for the evaluation, $A_t[i]\geq 0, \forall which includes [*strict Turnstile*]{} model as a special For data streams in practice, this restriction is The underlying technique is [*skewed stable random projections*]{}, which captures the intuition that, when $\alpha simple counter suffices, and $\alpha = 1\pm | T. Kodama, Prog. Part. Nucl. Phys. [**86**]{}, 35 (2016) doI:10.1016/j.ppnp.2015.09.002 \[arXiV:1506.03863 \[nucl-Th\]\].
T. burCh And G. torrIeri, Phys. Rev. D [**92**]{}, no. 1, 016009 (2015) DOi:10.1103/PhYsRevD.92.016009 \[arXiv:1502.05421 \[hep-lat\]\]. M. A. L. CaPri, M. S. guIMaraES, I. f. JustO, L. F. PalhAReS ANd S. p. SOrEllA, PHYs. lett. B [**735**]{}, 277 (2014) Doi:10.1016/J.physleTb.2014.06.035 \[arXiv:1404.7163 \[hep-Th\]\].
---
aBsTract: |
**CountinG** Is A fundamentAl oPeration. For eXamPle, couNtIng THe $\alpHa$tH freqUency mOMent, $F_{(\aLpha)} = \sum_{i=1}^D a_t[I]^\Alpha$, oF A streamING sIgnaL $A_t$ (where $t$ denotes tIMe), HAs been an active Area of ReSEaRCH, in TheOretical coMpUter sCIence, daTAbASES, anD Data mining. WheN $\alpha =1$, the taSK (i.e., CountiNg The SUm) can bE accoMpLIshEd using a couNter. when $\alpha \Neq 1$, howEVer, it beCOmes non-TriviaL to DesIgn a SMaLl SpaCe (I.E., loW MeMorY) CouNting sysTeM.
[*COmpreSsed cOUNTing (cC)*]{} iS proPosed For efficientlY coMputINg tHe $\alpHa$th fRequEnCy momEnt of a Data sTrEam $A_t$, where $0<\alpha\Leq2$. Cc is applicAblE iF thE sTreamINg data FolLow The [*TurnStile*]{} moDEl, wItH THE rEstriction that at thE tIME $t$ For the evAluatiON, $A_T[i]\GEq 0, \forall I\iN[1,D]$, wHich INCludeS the [*STrIct TurnsTile*]{} moDEl As A speciaL cAse. For DaTa sTreAms in PRactIce, thiS restricTion iS Minor.
The underlYIng technique iS [*SkEWEd STablE raNdom projectIons*]{}, WHich CaptUReS thE IntuiTion tHaT, WhEN $\alpha =1$ a simple counteR sUfficeS, and wHen $\alpha = 1\pm | T. Kodama, Prog. Part. Nu cl. Phys. [**8 6** ]{} ,35 ( 2016 ) doi:10.1016/ j .ppn p.2015.09.002 \[arXiv: 1506. 03 8 63 \ [ nu cl-th \]\].
T .B u rch a nd G. T o rr ieri, Ph ys. Rev . D [**92* *]{ }, no. 1, 0160 0 9(2015) doi :10 .1103/PhysRe vD. 92.016 00 9 \ [ arXiv :15 02.05 421 \[ h ep-lat \]\]. M.A. L. Cap r i, M. S . Gu imar aes, I. F. Justo, L. F. Palhares an d S. P .S or e l la, Ph ys. Lett.B[**73 5 **]{},2 77 ( 2 014 ) doi:10.1016/ j.physletb. 2 014 .06.03 5\[a r Xiv:14 04.71 63 \[h ep-th\]\]. ---
abstract : |
**Coun t ing** i s a fu nda men talo pe ra tio n. For ex amp l e,counting t he $\al pha$ t h f requ enc y mo ment, $F_{(\alpha) } = \su m _{i =1}^D A_t[ i]^\ al pha$, of astrea mi ng signal $A_t$ (wh ere $t$ d eno te s t im e), h a s been an ac tive ar ea of r e sea rc h , in theoretical compu te r sc ience, d atabas e s, a n d data m in ing . Wh e n $\al pha= 1$ , the ta sk (i. e ., c ounting t he sum )can be acco m plis hed us ing a co unter . When $\alpha\ neq 1$, howev e r, i tb ecom esnon-trivial tod esig n as ma lls pace(i.e. ,l ow memory) counting sy st em.
[* Compressed Co unting (CC ) * ] {} is pr opos e df or efficiently comp uting the$ \alpha$t h fre quency m oment ofa data str eam $A _t$ , w h e re $0<\alpha\le q 2 $. C Cis appl ica ble ifthe st rea min gdata foll ow the [ *T ur ns ti le* ]{} m o del, wit hthe r est ricti o n that at t he t im e$ t$for the ev a l uati on ,$A_t [i] \g eq 0, \fo r all i\in[1 ,D]$, whi chi nclu de sthe [*s trict Turnsti le *]{} model a s a speci a l case. F or data streams in prac t ice, th isrestr icti on is min or.
The underl ying t echni qu e i s [*ske w e dsta bl e random p r o jec tions *] {},which c aptures the intuit i onthat, when $\ alp ha = 1 $ a si m pl e co un t ers u ffices, and whe n $\alpha=1 \p m | T. Kodama,_Prog. Part. Nucl. Phys. [**86**]{},_35 (2016) doi:10.1016/j.ppnp.2015.09.002 \[arXiv:1506.03863_\[nucl-th\]\].
T. Burch and_G. Torrieri,_Phys. Rev. D [**92**]{},_no._1, 016009 (2015)_doi:10.1103/PhysRevD.92.016009 \[arXiv:1502.05421 \[hep-lat\]\]._M. A. L. Capri, M. S. Guimaraes, I. F. Justo, L. F. Palhares_and S. P. Sorella, Phys. Lett. B_[**735**]{},_277 (2014) doi:10.1016/j.physletb.2014.06.035 \[arXiv:1404.7163 \[hep-th\]\].
---
abstract: |
**Counting** is a fundamental_operation._For example,_counting_the_$\alpha$th frequency moment, $F_{(\alpha)} =_\sum_{i=1}^D A_t[i]^\alpha$, of a streaming_signal $A_t$_(where $t$ denotes time), has been an active_area_of research, in_theoretical computer science, databases, and data mining. When $\alpha_=1$, the task (i.e., counting the_sum) can be_accomplished_using_a counter. When $\alpha_\neq 1$, however, it becomes non-trivial_to design a small space (i.e.,_low memory) counting system.
_[*Compressed Counting (CC)*]{} is proposed for_efficiently computing the $\alpha$th frequency_moment of_a data stream $A_t$, where_$0<\alpha\leq2$. CC is_applicable if_the streaming data_follow the [*Turnstile*]{} model, with the_restriction that at_the time $t$ for the evaluation,_$A_t[i]\geq_0, \forall i\in[1,D]$,_which_includes_the [*strict_Turnstile*]{} model as_a_special case._For_data streams in practice, this restriction_is_minor.
The underlying technique_is [*skewed stable random_projections*]{},_which captures the intuition_that, when $\alpha =1$ a_simple counter suffices, and when $\alpha_= 1\pm |
, which is operated by the Association of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation.
---
abstract: |
A new generalization of Pascal’s triangle, the so-called hyperbolic Pascal triangles were introduced in [@BNSz]. The mathematical background goes back to the regular mosaics in the hyperbolic plane. The alternating sum of elements in the rows was given in the special case $\{4,5\}$ of the hyperbolic Pascal triangles. In this article, we determine the alternating sum generally in the hyperbolic Pascal triangle corresponding to $\{4,q\}$ with $q\ge5$.\
[*Key Words: Pascal triangle, hyperbolic Pascal triangle, alternating sum.*]{}\
[*MSC code: 11B99, 05A10.*]{}
author:
- 'László Németh[^1], László Szalay[^2] [^3]'
title: '**Alternating sums in hyperbolic Pascal triangles** '
---
Introduction {#sec:introduction}
============
In the hyperbolic plane there are infinite types of regular mosaics (see, for example [@C]), they are denoted by Schläfli’s symbol $\{p,q\}$, where $(p-2)(q-2)>4$. Each regular mosaic induces a so called hyperbolic Pascal triangle (see [@BNSz]), following and generalizing the connection between the classical Pascal’s triangle and the Euclidean regular square mosaic $\{4,4\}$. For more details see [@BNSz], but here we also collect some necessary information.
There are several approaches to generalize the Pascal’s arithmetic triangle (see, for instance [@BSz]). The hyperbolic Pascal triangle based on the mosaic $\{p,q\}$ can be figured as a digraph, where the vertices and the edges are the vertices and the edges of a well defined part of the lattice $\{p,q\}$, respectively, further the vertices possesses a value giving the number of different shortest paths from the base vertex. Figure \[fig:Pascal\_46\_layer5\] illustrates the hyperbolic Pascal triangle when $\{p,q\}=\{4,6\}$. Generally, for $\{4,q\}$ the base vertex has two edges, the leftmost and the rightmost vertices have three, the others have $q$ edges. The square shaped cells surrounded by appropriate edges are corresponding to the regular squares in the mosaic. Apart from the | , which is operated by the Association of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation.
---
abstract: |
A new generalization of Pascal ’s triangulum, the therefore - called hyperbolic Pascal triangles were insert in [ @BNSz ]. The numerical background goes back to the regular mosaic in the hyperbolic plane. The alternating kernel of elements in the rows was give in the special case $ \{4,5\}$ of the hyperbolic Pascal triangulum. In this article, we determine the alternating sum generally in the hyperbolic Pascal triangle corresponding to $ \{4,q\}$ with $ q\ge5$.\
[ * Key parole: Pascal triangle, hyperbolic Pascal triangle, alternating sum.*]{}\
[ * MSC code: 11B99, 05A10. * ] { }
generator:
-' László Németh[^1 ], László Szalay[^2 ] [ ^3 ]'
title:' * * Alternating sums in hyperbolic Pascal triangles * *'
---
presentation { # sec: introduction }
= = = = = = = = = = = =
In the hyperbolic plane there are infinite types of even mosaics (see, for example [ @C ]), they are denoted by Schläfli ’s symbol $ \{p, q\}$, where $ (p-2)(q-2)>4$. Each regular mosaic induces a so called hyperbolic Pascal triangle (see [ @BNSz ]), following and generalizing the connection between the classical Pascal ’s triangle and the Euclidean regular straight mosaic $ \{4,4\}$. For more details see [ @BNSz ], but here we besides roll up some necessary data.
There are several approaches to generalize the Pascal ’s arithmetic triangle (see, for instance [ @BSz ]). The hyperbolic Pascal triangulum based on the mosaic $ \{p, q\}$ can be figured as a digraph, where the vertices and the edges are the vertices and the edges of a well defined part of the lattice $ \{p, q\}$, respectively, further the vertices possesses a value give the number of unlike short paths from the base vertex. Figure \[fig: Pascal\_46\_layer5\ ] illustrate the hyperbolic Pascal triangle when $ \{p, q\}=\{4,6\}$. Generally, for $ \{4,q\}$ the base vertex take two boundary, the leftmost and the rightmost vertices have three, the others consume $ q$ edges. The square shaped cells surrounded by appropriate edge are corresponding to the even squares in the mosaic. aside from the | , whlch is operated by the Arsociation of Uuuversivies fod Researzh in Astronomy, Inc., under colpwratice agreement with the Vational Dcience Dounvation.
---
abstract: |
A new genedwlizctmon of Pascal’s jriangle, the so-called hypesbulnc Pascal triangles were introduced yn [@BNSz]. Tje mathematicaj babkdrouhd goes back to the regular mosaica in tht hyperbolic plane. The alternating sum of elfmenhs in the rows was given in tye s[wcial case $\{4,5\}$ uf the hyptryolic Pascam triangles. In this article, we aeterkine the aotwrnwjing sum genxrally in the hypevnolic [ascal yriangle correxpoidint to $\{4,q\}$ with $q\ge5$.\
[*Keb Words: Pascal triandle, hypertomic Pascal triangoe, altetnatitg sjn.*]{}\
[*MAC ckde: 11B99, 05W10.*]{}
auvhor:
- 'László Néjeth[^1], László Szalay[^2] [^3]'
title: '**Alternauind sums in hypedbolic Pwscal triangles** '
---
Introduction {#sec:introdubtioh}
============
In the hyperbolic planw there are infinite jypes of rqgular mosaics (see, for example [@C]), they are denoted by Srhuäfln’w symcil $\{p,q\}$, where $(p-2)(q-2)>4$. Each regular mosaic induces a so daklvd hyperbolic Pasgal triangle (see [@BMSx]), gjllowing and eeneralizjng the connection betweeg the classicaj Paxcal’s triangle and the Eucludean regulag sqyare mosaic $\{4,4\}$. For mlre details see [@BNSz], but here we also colleet soms necessary informatjun.
There are sevefal a[proaches to generalize thq Pascal’s arichmetic griamgle (sqe, for inshance [@BSz]). The hyperbolic Pasccl trhangle basfd on the mosaic $\{p,q\}$ can be figucxd as a digrakh, fhege the vextices and the edgef are the vertnces and the ddges of a well dxfined part jf the lattica $\{p,q\}$, respectitely, furtrer rhe certicer possesses a vslue giviuy the numver of different smorterf paths from thz vase vertex. Figire \[wig:[adcel\_46\_layqs5\] illustratev thd hhlerbouic Pascal brixnglr when $\{p,q\}=\{4,6\}$. Generally, xor $\{4,s\}$ the base vertex naf two edtes, the jeftmost and yhe rightmost vertlces iave tiree, tne jthers have $q$ edges. The square shaped cfllf surrounded by qppropriate zdges are corresponding to the regular swuares in the mosauc. Apart from the | , which is operated by the Association for in Astronomy, under cooperative agreement --- | A new of Pascal’s triangle, so-called hyperbolic Pascal triangles were introduced [@BNSz]. The mathematical background goes back to the regular mosaics in the hyperbolic The alternating sum of elements in the rows was given in the special $\{4,5\}$ the Pascal In this article, we determine the alternating sum generally in the hyperbolic Pascal triangle corresponding to with $q\ge5$.\ [*Key Words: Pascal triangle, hyperbolic Pascal alternating sum.*]{}\ [*MSC code: 05A10.*]{} author: - 'László Németh[^1], Szalay[^2] title: '**Alternating in Pascal ' --- Introduction ============ In the hyperbolic plane there are infinite types of regular mosaics (see, for example [@C]), they denoted by $\{p,q\}$, where Each mosaic a so called triangle (see [@BNSz]), following and generalizing the classical Pascal’s triangle and the Euclidean regular mosaic $\{4,4\}$. more details see [@BNSz], but here also collect some necessary information. There are several to generalize the Pascal’s arithmetic triangle (see, for instance [@BSz]). The hyperbolic Pascal triangle based mosaic $\{p,q\}$ can be as a digraph, the and edges the vertices the edges of a well defined part of the lattice $\{p,q\}$, further the vertices possesses a value giving the number of paths the base vertex. \[fig:Pascal\_46\_layer5\] illustrates the hyperbolic triangle $\{p,q\}=\{4,6\}$. Generally, for $\{4,q\}$ vertex two and rightmost have three, the others $q$ edges. The square shaped surrounded by appropriate edges squares in the mosaic. Apart from the | , which is operated by the AssocIation of UnIversItiEs fOr reseArch In Astronomy, Inc., UNder Cooperative agreement wiTh the naTIonaL scIence foundatIOn.
---
ABStrAcT: |
A New GeNErAlizaTioN of PascAl’s trianglE, thE sO-called hyperBOlIc Pascal trIanGles were intrOduCed in [@BnSZ]. ThE MatheMatIcal bAckgroUNd goes Back to the ReGUlar moSAics in tHE HyPerbOlic plane. The alterNAtINg sum of elementS in the RoWS wAS GivEn iN the speciaL cAse $\{4,5\}$ of THe hyperBOlIC pAscAL triangles. In tHis article, wE DetErmine ThE alTErnatiNg sum GeNEraLly in the hypErboLic Pascal TriangLE corresPOnding tO $\{4,q\}$ with $Q\ge5$.\
[*key wordS: paScAl tRiANglE, HyPerBOliC Pascal tRiAnGle, alTernATING sum.*]{}\
[*mSC Code: 11b99, 05A10.*]{}
autHor:
- 'László NémeTh[^1], LÁszlÓ szaLay[^2] [^3]'
tiTle: '**AlTernAtIng suMs in hyPerboLiC Pascal triangleS** '
---
IntRoduction {#Sec:InTroDuCtion}
============
iN the hyPerBolIc plane There arE InfInITE TyPes of regular mosaicS (sEE, FoR example [@c]), they aRE dEnOTed by SchLäFli’S symBOL $\{p,q\}$, whEre $(p-2)(Q-2)>4$. eaCh regulaR mosaiC InDuCes a so cAlLed hypErBolIc PAscal TRianGle (see [@bNSz]), follOwing ANd generalizing THe connection bETwEEN tHE claSsiCal Pascal’s tRianGLe anD the eUcLidEAn regUlar sQuARe MOsaic $\{4,4\}$. For more details SeE [@BNSz], bUt herE we also collecT some necesSARY informaTion.
tHeRE are several appRoachEs to generaLIze the PaScal’s ArithmetIc trianglE (SEe, for insTanCe [@Bsz]). THe hYPErBolic Pascal trIANgle BaSed on thE moSaic $\{p,q\}$ cAn bE fiGurEd aS a Digraph, whEre the veRtIcEs AnD thE edgeS Are the veRtIceS aNd tHe edgES of a weLl defIned PaRt OF thE latticE $\{P,q\}$, RESpecTiVeLy, fuRthEr The veRticES poSsesses A value givIng THe nuMbEr Of diffeRent shortest pAtHs from the bAsE veRtex. FiGURe \[fig:PasCal\_46\_layer5\] illustrates the hYPerboliC PaScal tRianGle when $\{p,q\}=\{4,6\}$. genErally, For $\{4,Q\}$ The basE verteX has tWo EdgES, The leFTMoSt aNd The rightmoST VerTices HaVe thRee, the oThers have $q$ edges. The SQuaRe shaped cells SurRounDED bY apPRoPRiaTe EDgeS ARe corresponding To the regulAr SQuAres in the mOSaiC. APart froM the | , which is operated by the Associati on of Un ive rs itie s fo r Research inA stro nomy, Inc., under coop erati ve agre e me nt wi th theN at i o nal S ci enc eF ou ndati on.
---
a bstract: |
A new genera l iz ation of P asc al’s triangl e,the so -c all e d hyp erb olicPascal triang les werein t roduce d in [@B N S z] . Th e mathematical ba c kg r ound goes back to th er eg u l armos aics in th ehyper b olic pl a ne . T hea lternating su m of elemen t s i n thero wsw as giv en in t h e s pecial case $\{ 4,5\}$ of the h y perboli c Pascal trian gle s.In t h is a rti cl e , w e d ete r min e the al te rn ating sum g e n eral lyin t he hy perbolic Pasc altria n gle corr espon ding t o $\{ 4,q\}$ with $ q\ge5$.\
[* KeyWords: Pa sca ltri an gle,h yperbo lic Pa scal tr iangle, alt er n a t in g sum.*]{}\
[* MS C co de: 11B9 9, 05A 1 0. *] { }
author :- ' Lász l ó Néme th[^ 1 ], LászlóSzalay [ ^2 ][^3]'
t it le: '* *A lte rna tings umsin hyp erbolicPasca l triangles** ' ---
Introduc t io n {# s ec:i ntr oduction}
= ==== = ==== ==
I nthe hyper bolic p l an e there are infinite t ypes o f reg ular mosaics(see, fore x a mple [@C ]),t he y are denoted b y Sch läfli’s sy m bol $\{p ,q\}$ , where$(p-2)(q- 2 ) >4$. Eac h r egu lar mo s a ic induces a so c alle dhyperbo lic Pascal tr ian gle (s ee [@BNSz]) , follow in gan dgen erali z ing theco nne ct ion betw e en the clas sica lPa s cal ’s tria n gl e andth eEucl ide an regu lars qua re mosa ic $\{4,4 \}$ . For m or e detai ls see [@BNSz ], but herewe al so col l e ct somenecessary information.There a resever al a pproaches to gener ali z e thePascal ’s ar it hme t i c tri a n gl e ( se e, for ins t a nce [@BS z] ). T he hype rbolic Pascal tria n gle based on the mo saic $ \{ p,q \ }$ can b e fi g u red as a digrap h, where t he ve rtices and the e dges ar e the v ertic e s and t he edgesof a well d efin e d pa rt of thelattice$\{p,q\}$ , resp e ct ively , f urther t heverti ces po s ses ses a value g ivingthe n um ber of d ifferent shortest paths fromthe b ase vertex.Fig u re\[fig:Pas cal\ _46\_layer 5\] il lustr ate s thehype r bo lic Pasca l tr i angle whe n $ \{p , q \} =\{4,6\}$.G e n era lly,for $\{4,q \}$the base vertex h a s two edges, t he l e f tmo sta nd t he rightmost ver tic es h ave thre e, the others have $q $e dges. The s quareshapedc e ll s surro unde d b y appropr iat ee dges ar eco r respon ding t o theregula r squ a r es in the mosaic . Apa r t from the | , which_is operated_by the Association of_Universities for_Research_in Astronomy,_Inc.,_under cooperative agreement_with the National_Science Foundation.
---
abstract: |
_ _A_new generalization of Pascal’s triangle, the so-called hyperbolic Pascal triangles were introduced in [@BNSz]._The_mathematical background_goes_back_to the regular mosaics in_the hyperbolic plane. The alternating_sum of_elements in the rows was given in the_special_case $\{4,5\}$ of_the hyperbolic Pascal triangles. In this article, we determine_the alternating sum generally in the_hyperbolic Pascal triangle_corresponding_to_$\{4,q\}$ with $q\ge5$.\
_ [*Key Words: Pascal triangle,_hyperbolic Pascal triangle, alternating sum.*]{}\
_ [*MSC code: 11B99, 05A10.*]{}
author:
- 'László_Németh[^1], László Szalay[^2] [^3]'
title: '**Alternating sums_in hyperbolic Pascal triangles** '
---
Introduction_{#sec:introduction}
============
In the_hyperbolic plane there are infinite_types of regular_mosaics (see,_for example [@C]),_they are denoted by Schläfli’s symbol_$\{p,q\}$, where $(p-2)(q-2)>4$._Each regular mosaic induces a so_called_hyperbolic Pascal triangle_(see_[@BNSz]),_following and_generalizing the connection_between_the classical_Pascal’s_triangle and the Euclidean regular square_mosaic_$\{4,4\}$. For more details see [@BNSz], but_here we also collect_some_necessary information.
There are several_approaches to generalize the Pascal’s_arithmetic triangle (see, for instance [@BSz])._The hyperbolic_Pascal triangle_based on the mosaic $\{p,q\}$ can be figured as a digraph,_where the vertices and the edges_are the vertices and_the edges_of_a well defined_part_of the_lattice $\{p,q\}$, respectively, further the vertices possesses_a value_giving the number of different shortest_paths from the base_vertex._Figure \[fig:Pascal\_46\_layer5\] illustrates the hyperbolic Pascal triangle_when $\{p,q\}=\{4,6\}$. Generally, for $\{4,q\}$ the_base vertex has two edges,_the_leftmost_and the rightmost vertices have_three, the others have $q$ edges._The square shaped_cells surrounded by appropriate edges are corresponding_to_the regular squares in the mosaic._Apart_from the |
& $-$203047 & 3180 & 1.4535 & 2.8893 & Severely warped\
B-24 & A 2333-1637 & 233305 & $-$163714 & & & &\
C-02 & A 0017+2212 & 001716.7 & $+$221200 & & & &\
C-03 & ESO 474-G26 & 004440 & $-$243836 & 16246 & 0.9131 & 1.7967 & Inner & outer rings?\
C-04 & A 0051-1323 & 005100.2 & $-$132304 & 10876 & & &\
C-05 & AM 0051-234 & 005153.5 & $-$234926 & 20156 & 0.4437 & 1.3828 &\
C-06 & NGC 304 & 005324 & $+$235100 & 4991 & & &\
C-09 & NGC 442 & 011205 & $-$011748 & 5629 & 0.1893 & 1.54 &\
C-12 & UGC 1198, VII Zw 3 & 014058.2 & $+$850038 & 1149 & 2.58 & 3.22 &\
C-24 & UGC 4261 & 080740.2 & $+$365838 & 6446 & 1.1426 & 1.56 &\
C-25 & UGC 4323 & 081536.3 & $+$670820 & 4061 & 0.1295 & (1.1583) &\
C-27 & UGC 4385 & 082104.0 & $+$145449 & 1954 & & &\
C-28 & NGC 2748 & 090802.6 & $+$764053 & 1476 & 7.0358 & 17.982 &\
C-29 & NGC 2865 & 095035.2 & $-$223427 & 14692 & & &\
C-30 & UGC 5101 & 093205.1 & $+$613433 & 11945 & 11.542 & 20.226 &\
C-32 & IC 575 & 095204 & $-$063700 & 5973 & 0.2716 & 0.9833 &\
C-35 & NGC 3414 & 104831. | & $ -$203047 & 3180 & 1.4535 & 2.8893 & Severely warped\
B-24 & A 2333 - 1637 & 233305 & $ -$163714 & & & & \
C-02 & A 0017 + 2212 & 001716.7 & $ + $ 221200 & & & & \
C-03 & ESO 474 - G26 & 004440 & $ -$243836 & 16246 & 0.9131 & 1.7967 & Inner & outer rings?\
C-04 & A 0051 - 1323 & 005100.2 & $ -$132304 & 10876 & & & \
C-05 & AM 0051 - 234 & 005153.5 & $ -$234926 & 20156 & 0.4437 & 1.3828 & \
C-06 & NGC 304 & 005324 & $ + $ 235100 & 4991 & & & \
C-09 & NGC 442 & 011205 & $ -$011748 & 5629 & 0.1893 & 1.54 & \
C-12 & UGC 1198, VII Zw 3 & 014058.2 & $ + $ 850038 & 1149 & 2.58 & 3.22 & \
C-24 & UGC 4261 & 080740.2 & $ + $ 365838 & 6446 & 1.1426 & 1.56 & \
C-25 & UGC 4323 & 081536.3 & $ + $ 670820 & 4061 & 0.1295 & (1.1583) & \
C-27 & UGC 4385 & 082104.0 & $ + $ 145449 & 1954 & & & \
C-28 & NGC 2748 & 090802.6 & $ + $ 764053 & 1476 & 7.0358 & 17.982 & \
C-29 & NGC 2865 & 095035.2 & $ -$223427 & 14692 & & & \
C-30 & UGC 5101 & 093205.1 & $ + $ 613433 & 11945 & 11.542 & 20.226 & \
C-32 & IC 575 & 095204 & $ -$063700 & 5973 & 0.2716 & 0.9833 & \
C-35 & NGC 3414 & 104831. | & $-$203047 & 3180 & 1.4535 & 2.8893 & Severely warped\
B-24 & A 2333-1637 & 233305 & $-$163714 & & & &\
C-02 & A 0017+2212 & 001716.7 & $+$221200 & & & &\
R-03 & ESO 474-G26 & 004440 & $-$243836 & 16246 & 0.9131 & 1.7967 & Inner & outer rings?\
C-04 & A 0051-1323 & 005100.2 & $-$132304 & 10876 & & &\
C-05 & AM 0051-234 & 005153.5 & $-$234926 & 20156 & 0.4437 & 1.3828 &\
C-06 & NGC 304 & 005324 & $+$235100 & 4991 & & &\
C-09 & NGC 442 & 011205 & $-$011748 & 5629 & 0.1893 & 1.54 &\
C-12 & UGC 1198, TJI Zw 3 & 014058.2 & $+$850038 & 1149 & 2.58 & 3.22 &\
C-24 & BGR 4261 & 080740.2 & $+$365838 & 6446 & 1.1426 & 1.56 &\
C-25 & OGC 4323 & 081536.3 & $+$670820 & 4061 & 0.1295 & (1.1583) &\
C-27 & UGC 4385 & 082104.0 & $+$145449 & 1954 & & &\
C-28 & NGC 2748 & 090802.6 & $+$764053 & 1476 & 7.0358 & 17.982 &\
C-29 & NGC 2865 & 095035.2 & $-$223427 & 14692 & & &\
C-30 & UDC 5101 & 093205.1 & $+$613433 & 11945 & 11.542 & 20.226 &\
C-32 & IC 575 & 095204 & $-$063700 & 5973 & 0.2716 & 0.9833 &\
C-35 & HGC 3414 & 104831. | & $-$203047 & 3180 & 1.4535 & Severely B-24 & 2333-1637 & 233305 &\ & A 0017+2212 001716.7 & $+$221200 & & &\ C-03 & ESO & 004440 & $-$243836 & 16246 & 0.9131 & 1.7967 & Inner & rings?\ C-04 & A 0051-1323 & 005100.2 & $-$132304 & 10876 & & C-05 AM & & $-$234926 & 20156 & 0.4437 & 1.3828 &\ C-06 & NGC 304 & 005324 & & 4991 & & &\ C-09 & NGC & 011205 & $-$011748 5629 & 0.1893 & 1.54 C-12 UGC 1198, Zw & & $+$850038 & & 2.58 & 3.22 &\ C-24 & UGC 4261 & 080740.2 & $+$365838 & 6446 & 1.1426 1.56 &\ UGC 4323 081536.3 $+$670820 4061 & 0.1295 &\ C-27 & UGC 4385 & & 1954 & & &\ C-28 & NGC & 090802.6 $+$764053 & 1476 & 7.0358 & &\ C-29 & NGC 2865 & 095035.2 & & 14692 & & &\ C-30 & UGC 5101 & 093205.1 & $+$613433 & 11945 & 20.226 &\ C-32 IC 575 & & & & & 0.9833 C-35 & NGC 3414 & 104831. | & $-$203047 & 3180 & 1.4535 & 2.8893 & Severely warped\
B-24 & A 2333-1637 & 233305 & $-$163714 & & & &\
C-02 & A 0017+2212 & 001716.7 & $+$221200 & & & &\
C-03 & ESO 474-G26 & 004440 & $-$243836 & 16246 & 0.9131 & 1.7967 & InNer & outer riNgs?\
C-04 & A 0051-1323 & 005100.2 & $-$132304 & 10876 & & &\
c-05 & AM 0051-234 & 005153.5 & $-$234926 & 20156 & 0.4437 & 1.3828 &\
c-06 & NGc 304 & 005324 & $+$235100 & 4991 & & &\
C-09 & nGC 442 & 011205 & $-$011748 & 5629 & 0.1893 & 1.54 &\
C-12 & uGC 1198, ViI Zw 3 & 014058.2 & $+$850038 & 1149 & 2.58 & 3.22 &\
C-24 & UGC 4261 & 080740.2 & $+$365838 & 6446 & 1.1426 & 1.56 &\
C-25 & UGC 4323 & 081536.3 & $+$670820 & 4061 & 0.1295 & (1.1583) &\
C-27 & Ugc 4385 & 082104.0 & $+$145449 & 1954 & & &\
C-28 & NGc 2748 & 090802.6 & $+$764053 & 1476 & 7.0358 & 17.982 &\
C-29 & NGC 2865 & 095035.2 & $-$223427 & 14692 & & &\
C-30 & UGC 5101 & 093205.1 & $+$613433 & 11945 & 11.542 & 20.226 &\
C-32 & IC 575 & 095204 & $-$063700 & 5973 & 0.2716 & 0.9833 &\
C-35 & NGC 3414 & 104831. | & $-$203047 & 3180 & 1.45 35 & 2.889 3 & S eve rel ywarp ed\B-24 & A 2333- 1 637& 233305 & $-$163714 & & &&\ C-02 &A 001 7+2212& 0 0 1 716 .7 & $+ $2 2 12 00 && & &\
C-0 3 & ESO 47 4-G 26 & 004440 &$ -$ 243836 & 1 624 6 & 0.9131 & 1. 7967 & I nne r & ou ter ring s?\
C- 0 4 & A0051-1323 & 005100 . 2 & $-$ 1 3 23 04 & 10876 & & &\
C-0 5 & AM 0051-234 &005153 .5 &$ - $23 492 6 & 20156&0.443 7 & 1.38 2 8& \ C-0 6 & NGC 304 &005324 & $+ $ 235 100 &49 91& & &\C-09&N GC442 & 01120 5 &$-$011748 & 562 9 & 0.18 9 3 & 1.5 4 &\
C -12 &UGC1 19 8, VI IZ w 3 &014 0 58. 2 & $+$8 50 03 8 & 1 149& 2 . 58 & 3. 22 & \
C-2 4 & UGC 4261& 0 8074 0 .2& $+$ 36583 8 &64 46 &1.1426 & 1. 56 &\
C-25 & UGC4323 & 081536 .3&$+$ 67 0820& 4061& 0 .12 95 & (1 .1583)& \
C -2 7 & U GC 4385 & 082104.0 & $ +$ 145449 & 1954& & & \
C-28 &NG C 2 748& 09080 2.6& $ +$764053 & 147 6 & 7 .0358 & 1 7.982&\
C- 29& NGC 2865 & 095 035.2 &$-$22 3 427 & 14692 && &\
C-30 & UG C 5 1 0 1& 093 205 .1 & $+$613 433& 119 45 & 11 .54 2 & 20 .226&\ C- 3 2 & IC 575 & 095204 & $-$06 3700& 5973 & 0.27 16 & 0.983 3 & \
C-35 & NGC 34 1 4 & 104831. | &_$-$203047 &_3180 & 1.4535 &_2.8893 &_Severely_warped\
B-24 &_A 2333-1637_& 233305 &_$-$163714 & &_& &\
C-02 & A 0017+2212_& 001716.7 &_$+$221200_& & & &\
C-03 & ESO 474-G26 & 004440 & $-$243836 & 16246 & 0.9131_&_1.7967 &_Inner_& outer_rings?\
C-04 & A 0051-1323 & 005100.2_& $-$132304 & 10876 &_& &\
C-05_& AM 0051-234 & 005153.5 & $-$234926 & 20156_&_0.4437 & 1.3828_&\
C-06 & NGC 304 & 005324 & $+$235100 & 4991_& & &\
C-09 & NGC 442 &_011205 & $-$011748_&_5629_& 0.1893 & 1.54_&\
C-12 & UGC 1198, VII Zw 3 & 014058.2_& $+$850038 & 1149 & 2.58_& 3.22 &\
C-24 & UGC 4261 & 080740.2_& $+$365838 & 6446 & 1.1426_& 1.56 &\
C-25 & UGC 4323_& 081536.3_& $+$670820 & 4061 &_0.1295 & (1.1583)_&\
C-27 &_UGC 4385 & 082104.0_& $+$145449 & 1954 & &_&\
C-28 & NGC 2748_& 090802.6 & $+$764053 & 1476_&_7.0358 & 17.982_&\
C-29_&_NGC 2865 &_095035.2 & $-$223427_&_14692 &_&_&\
C-30 & UGC 5101 & 093205.1 &_$+$613433_& 11945 & 11.542 & 20.226 &\
C-32_& IC 575 & 095204_&_$-$063700 & 5973 &_0.2716 & 0.9833 &\
C-35 &_NGC 3414 & 104831. |
open-string model in [@cudell].
As seen from the above discussion, an accurate determination of the ratio $T_h/T_c$ is of great phenomenological interest. However, such a study is not the main purpose of the present paper, where we aim at giving reliable predictions for the equation of state of YM theory with an arbitrary gauge algebra. As observed in [@meyer], typical values $T_h\approx T_c$ give very good results in fitting the lattice data. Setting $T_c=T_h$, as we will do in the rest of this work, means that the deconfinement temperature may be identified with the maximal allowed temperature for the confined hadronic phase. This assumption has two advantages. First, it will reproduce accurately the latest $\mathfrak{su}(3)$ lattice data of [@borsa] (see next section), and it is not in strong disagreement with current $\mathfrak{su}(N)$ results, where $T_h/T_c$ is at most around $10 \%$ [@TcTh0; @TcTh]. Second, it is applicable to any gauge algebra without having to guess a value for $T_h/T_c$, that cannot be fitted on lattice results since no equation of state is available for gauge algebras different than $\mathfrak{su}(N)$ so far. The drawback of this choice is that it forbids any discussion about a superheated hadronic phase in generic YM theories. Such a refinement of the model will rather be the topic of a separate study.
For completeness, we notice that the somewhat surprising value $T_h=2.8\, T_c\gg T_c$ has been found in [@Megias] by using a Hagedorn picture too. The difference with our approach comes from the fact that, in [@Megias], $T_h$ is fitted by assuming that the low-lying glueballs currently known from lattice simulations should exhibit a Hagedorn-type spectrum. On the contrary, we think here that the Hagedorn-like behavior only appears in the high-lying sector, that mostly concerns the glueballs that are not known so far by lattice calculations, see Eq. (\[preh\]).
Numerical results
-----------------
According to standard $\mathfrak{su}(3)$ studies, it is relevant to set $\sigma_0\approx(9/4)\ 0.2$ | open - string model in [ @cudell ].
As seen from the above discussion, an accurate decision of the proportion $ T_h / T_c$ is of great phenomenological interest. However, such a discipline is not the independent purpose of the present paper, where we draw a bead on at impart reliable prediction for the equation of state of YM hypothesis with an arbitrary gauge algebra. As observed in [ @meyer ], typical values $ T_h\approx T_c$ pass very good results in fitting the lattice data. dress $ T_c = T_h$, as we will do in the rest of this work, intend that the deconfinement temperature may be identified with the maximal allowed temperature for the confined hadronic phase. This assumption have two advantages. First, it will reproduce accurately the latest $ \mathfrak{su}(3)$ lattice datum of [ @borsa ] (see next section), and it is not in strong disagreement with current $ \mathfrak{su}(N)$ results, where $ T_h / T_c$ is at about around $ 10 \%$ [ @TcTh0; @TcTh ]. Second, it is applicable to any bore algebra without hold to guess a value for $ T_h / T_c$, that cannot be fitted on lattice results since no equation of state is available for gauge algebras unlike than $ \mathfrak{su}(N)$ thus far. The drawback of this choice is that it prevent any discussion about a superheated hadronic phase in generic YM theory. Such a refinement of the model will rather be the subject of a separate survey.
For completeness, we notice that the somewhat surprising value $ T_h=2.8\, T_c\gg T_c$ has been detect in [ @Megias ] by using a Hagedorn picture too. The difference with our approach path comes from the fact that, in [ @Megias ], $ T_h$ is fitted by assuming that the low - lying glueballs presently known from wicket simulations should exhibit a Hagedorn - type spectrum. On the contrary, we think here that the Hagedorn - like demeanor only appears in the high - lying sector, that mostly concerns the glueballs that are not known so far by lattice calculations, experience Eq. (\[preh\ ]).
Numerical results
-----------------
accord to standard $ \mathfrak{su}(3)$ studies, it is relevant to set $ \sigma_0\approx(9/4)\ 0.2 $ | opfn-string model in [@cudell].
Xs seen from thg qbove viscussjon, an azcurate determination of the rqtio $U_k/T_c$ is of great phenumenologibal interwst. Iowever, such a svhdy is kjt tgc maiu 'urpose of the kresent papes, where we aim ag yiving reliable predictions for the qquatiom lf state of YM thepwy wjnh an arbitrary gauge algebra. Aa obserned in [@meyer], typicsl values $T_h\approx T_c$ give verj good results in vitting the lateuce data. Setging $T_c=T_h$, as we will dk in the rest of this work, meanr thac the decondibemftt temperatnre maj be identificc with the macimal allowed bempecatuee for the confined hedronic phase. This afsumption hcs two advantages. Firwt, it whll seprueucd adcnrafely tje matest $\matgfrak{su}(3)$ latrice data of [@borsa] (ste gvct section), ahd it ys not in strong disagreement with currenu $\matgfrak{su}(N)$ results, where $R_h/T_c$ is at most arounf $10 \%$ [@TcTh0; @TsTh]. Second, it is applicable to any gauge algebra fithont harlkg tu gkess a value for $T_h/T_c$, that cannot be fitted og lstnice results sincc no equation of syahe ys available wor gabfe algebras differenh than $\iathfeak{su}(N)$ so far. The drawback of this choicw is that it dorbids any discusdion about c supetheatec hadronic phase in genzric YJ theories. Duch a rerknement of the mudek fill rather be the topic os a separete scudy.
For zompketenefs, we notife that the somewhat surorisiug vanue $T_h=2.8\, T_c\gh T_c$ has been found in [@Megias] by using a Hagedprt ppcture toj. The difference wyth our approaeh comes from the fact nhat, in [@Mxgias], $T_h$ is sitted by assgling that thx low-lyind glyebaols curfdntly known frpm latticv fimuoations should exhlbit x Hagedorn-type s'tcttym. On the conttarh, wq nhiik hewa that the Hdgedurn-uoke bdhavior onln aopeats in the high-lying vectkr, that mostly convevns the goueballs that are not known so far by lwtticx calcnlatioms, fee Eq. (\[preh\]).
Numerical results
-----------------
Acckrding to stwndard $\mathfwak{sm}(3)$ stodies, it is relevant to set $\sigma_0\approx(9/4)\ 0.2$ | open-string model in [@cudell]. As seen from discussion, accurate determination the ratio $T_h/T_c$ However, a study is the main purpose the present paper, where we aim giving reliable predictions for the equation of state of YM theory with an gauge algebra. As observed in [@meyer], typical values $T_h\approx T_c$ give very good in the data. $T_c=T_h$, as we will do in the rest of this work, means that the deconfinement temperature be identified with the maximal allowed temperature for confined hadronic phase. This has two advantages. First, it reproduce the latest lattice of (see next section), it is not in strong disagreement with current $\mathfrak{su}(N)$ results, where $T_h/T_c$ is at most around $10 [@TcTh0; @TcTh]. is applicable any algebra having to guess for $T_h/T_c$, that cannot be fitted since no equation of state is available for algebras different $\mathfrak{su}(N)$ so far. The drawback of choice is that it forbids any discussion about superheated hadronic phase in generic YM theories. Such a refinement of the model will rather topic of a separate For completeness, we that somewhat value T_c\gg T_c$ been found in [@Megias] by using a Hagedorn picture too. The with our approach comes from the fact that, in [@Megias], fitted assuming that the glueballs currently known from simulations exhibit a Hagedorn-type spectrum. contrary, think Hagedorn-like only in the high-lying sector, mostly concerns the glueballs that not known so far (\[preh\]). Numerical results ----------------- According to standard $\mathfrak{su}(3)$ it is relevant to set $\sigma_0\approx(9/4)\ 0.2$ | open-string model in [@cudell].
As Seen from thE abovE diScuSsIon, aN accUrate determinaTIon oF the ratio $T_h/T_c$ is of great PhenoMeNOlogICaL inteRest. HowEVeR, SUch A sTuDy iS nOT tHe maiN puRpose of The present PapEr, Where we aim at GIvIng reliablE prEdictions for The EquatiOn Of sTAte of yM tHeory With an ARbitraRy gauge alGeBRa. As obSErved in [@MEYeR], typIcal values $T_h\approX t_c$ GIve very good resUlts in FiTTiNG The LatTice data. SeTtIng $T_c=t_H$, as we wiLL dO IN The REst of this work, Means that thE DecOnfineMeNt tEMperatUre maY bE IdeNtified with The mAximal allOwed teMPeraturE For the cOnfineD haDroNic pHAsE. THis AsSUmpTIoN haS Two AdvantagEs. fiRst, it Will REPROducE acCuraTely tHe latest $\mathfRak{Su}(3)$ laTTicE data Of [@borSa] (seE nExt seCtion), aNd it iS nOt in strong disagReemEnt with cuRreNt $\MatHfRak{su}(n)$ ResultS, whEre $t_h/T_c$ is aT most arOUnd $10 \%$ [@tctH0; @tCTH]. Second, it is applicaBlE TO aNy gauge aLgebra WItHoUT having tO gUesS a vaLUE for $T_H/T_c$, tHAt Cannot be Fitted ON lAtTice resUlTs sincE nO eqUatIon of STate Is avaiLable for Gauge ALgebras differeNT than $\mathfrak{SU}(N)$ SO FaR. the dRawBack of this cHoicE Is thAt it FOrBidS Any diScussIoN AbOUt a superheated hadroNiC phase In genEric YM theorieS. Such a refiNEMEnt of the ModeL WiLL rather be the toPic of A separate sTUdy.
For coMpletEness, we nOtice that THE somewhaT suRprIsiNg vALUe $t_h=2.8\, T_c\gg T_c$ has beEN FounD iN [@Megias] By uSing a HaGedOrn PicTurE tOo. The diffErence wiTh OuR aPpRoaCh comES from the FaCt tHaT, in [@megiaS], $t_h$ is fiTted bY assUmInG ThaT the low-LYiNG GlueBaLlS curRenTlY knowN froM LatTice simUlations sHouLD exhIbIt A HagedoRn-type spectruM. ON the contraRy, We tHink heRE That the HAgedorn-like behavior only APpears iN thE high-LyinG sector, thAt mOstly cOncERns the GluebaLls thAt Are NOT knowN SO fAr bY lAttice calcULAtiOns, seE EQ. (\[preH\]).
NumeriCal results
-----------------
AccordinG To sTandard $\mathfrAk{sU}(3)$ stuDIEs, It iS ReLEvaNt TO seT $\SIgma_0\approx(9/4)\ 0.2$ | open-string model in [@cu dell].
As seen fr omth e ab ovediscussion, an accu rate determination ofthe r at i o $T _ h/ T_c$is of g r ea t phe no me nol og i ca l int ere st. How ever, such ast udy is not t h emain purpo seof the prese ntpaper, w her e we a imat gi ving r e liable predicti on s for t h e equat i o nof s tate of YM theory wi t h an arbitrary gauge a l ge b r a.Asobserved i n[@mey e r], typ i ca l v alu e s $T_h\approx T_c$ givev ery goodre sul t s in f ittin gt helattice dat a. S etting $T _c=T_h $ , as we will do in th e r est oft hi swor k, mea n stha t th e deconf in em ent t empe r a t u re m aybe i denti fied with the ma xima l al lowed temp erat ur e for the c onfin ed hadronic phase . Th is assump tio nhas t wo ad v antage s.Fir st, itwill re p rod uc e a cc urately the latest $ \ m at hfrak{su }(3)$l at ti c e data o f[@b orsa ] (seenext se ction),and it is n ot in s tr ong di sa gre eme nt wi t h cu rrent$\mathfr ak{su } (N)$ results,w here $T_h/T_c $ i s at most ar ound $10 \% $ [@ T cTh0 ; @T c Th ].S econd , itis ap p licable to any gaug ealgebr a wit hout having t o guess av a l ue for $ T_h/ T _c $ , that cannotbe fi tted on la t tice res ultssince no equation o f stateisava ila ble f or gauge algebr a s dif fe rent th an$\mathf rak {su }(N )$so far. The drawbac kof t hi s c hoice is thatit fo rb ids anyd iscuss ion a bout a s u per heatedh ad r o nicph as e in ge ne ric Y M th e ori es. Suc h a refin eme n t of t he modelwill rather b ethe topicof asepara t e study.
For completeness, we n o tice th atthe s omew hat surpr isi ng val ue$ T_h=2. 8\, T_ c\ggT_ c$h a s bee n fo und i n [@Megias ] byusing a Hag edorn p icture too. The di f fer ence with our ap proa c h c ome s f r omth e fa c t that, in [@Me gias], $T_ h$ is fitted by ass um ing tha t the l ow-ly i ng glue balls cur rently kn ow n fr o m la ttice simu lationsshould ex h ibita H agedo rn- type s pe ctr um. O n thec ont rary, we th in k here that t he Haged orn-like behavior onlyappear s inthe high-lyi ngs ect or, thatmost ly concern s t heglueb all s that are no t k n own s o fa r by latti c ecal c u la tions, seeE q . (\ [preh \]) .
Nume rica l results
------- - ---------
Acc ordi n g to st a ndar d$\mathfrak{su} (3) $s t udies, i tis relevant to set$\ s igma_ 0\appr ox(9/4 )\ 0.2$ | open-string_model in_[@cudell].
As seen from the_above discussion,_an_accurate determination_of_the ratio $T_h/T_c$_is of great_phenomenological interest. However, such_a study is_not_the main purpose of the present paper, where we aim at giving reliable predictions_for_the equation_of_state_of YM theory with an_arbitrary gauge algebra. As observed_in [@meyer], typical_values $T_h\approx T_c$ give very good results in_fitting_the lattice data._Setting $T_c=T_h$, as we will do in the rest_of this work, means that the_deconfinement temperature may_be_identified_with the maximal allowed_temperature for the confined hadronic phase._This assumption has two advantages. First,_it will reproduce accurately the latest $\mathfrak{su}(3)$_lattice data of [@borsa] (see next section),_and it is not in_strong disagreement_with current $\mathfrak{su}(N)$ results, where_$T_h/T_c$ is at_most around_$10 \%$ [@TcTh0; @TcTh]._Second, it is applicable to any_gauge algebra without_having to guess a value for_$T_h/T_c$,_that cannot be_fitted_on_lattice results_since no equation_of_state is_available_for gauge algebras different than $\mathfrak{su}(N)$_so_far. The drawback of this choice is_that it forbids any_discussion_about a superheated hadronic_phase in generic YM theories._Such a refinement of the model_will rather_be the_topic of a separate study.
For completeness, we notice that the somewhat_surprising value $T_h=2.8\, T_c\gg T_c$ has_been found in [@Megias] by_using a_Hagedorn_picture too. The_difference_with our_approach comes from the fact that, in_ [@Megias], $T_h$_is fitted by assuming that the_low-lying glueballs currently known_from_lattice simulations should exhibit a Hagedorn-type_spectrum. On the contrary, we think_here that the Hagedorn-like behavior_only_appears_in the high-lying sector, that_mostly concerns the glueballs that are_not known so_far by lattice calculations, see Eq. (\[preh\]).
Numerical results
-----------------
According_to_standard $\mathfrak{su}(3)$ studies, it is relevant_to_set $\sigma_0\approx(9/4)\ 0.2$ |
\]. Let $\Gamma '= \pi_1(M)/\pi_1(M)^{(n+1)}_H$. Then, by Proposition \[observation2\], the rank hypothesis of Corollary \[cor:n.5solvable\] is equivalent to the fact that $\mathcal{K}\Gamma '$-rank of $H_1(\pi_1(M);\mathcal{K}\Gamma ')$ is $\beta_1(M)-1$. Since $j_{n+1}: \Gamma '\to\G$ is a monomorphism, by [@CH1 Lemma 4.2], this rank is the same as the $\mathcal{K}\Gamma$-rank of $H_1(\pi_1(M);\mathcal{K}\Gamma)$ associated to the coefficient system $\phi$. But this is precisely the rank hypothesis of Theorem \[thm:rho=0\] in the case that $\Gamma=B/B_{{\scriptscriptstyle}H}^{{\scriptscriptstyle}(n+1)}$.
Note that the final claim of the theorem was already established by the last part of Proposition \[prop:nsolvable\] applied to the $(n+1)$-solution $W$.
Suppose $M$ is rationally $(n.5)$-solvable via $W$ such that the coefficient system extends to $\psi:\pi_1(W)\to \Gamma$. Then $\rho(M, \phi)$ is given by $\sigma^{(2)}_{\Gamma}(W)-\sigma(W)$, where $\sigma^{(2)}_{\Gamma}(W)$ by Definition \[def:rhoinvariant\]. Let $\tilde{I}$ denote the image of the map $$H_2(\partial W;\mathcal{K}\G)\overset{j_*}{\lra}H_2(W;\mathcal{K}\G).$$ Since the sequence $$0\lra \tilde {I} \lra H_2(W;\mathcal{K}\G)\lra H_2(W;\mathcal{K}\G)/\tilde {I}\lra 0$$ is split exact (since all $\mathcal{K}\Gamma$-modules are free) it follows that $$H_2(W;\mathcal{K}\G)\cong \tilde{I}\oplus (H_2(W;\mathcal{K}\G)/\tilde {I}).$$ Since the intersection form $$\lambda: H_2(W;\mathcal{K}\G)\to Hom(H_2(W;\mathcal{K}\G | \ ]. Let $ \Gamma' = \pi_1(M)/\pi_1(M)^{(n+1)}_H$. Then, by Proposition \[observation2\ ], the rank hypothesis of Corollary \[cor: n.5solvable\ ] is equivalent to the fact that $ \mathcal{K}\Gamma' $ -rank of $ H_1(\pi_1(M);\mathcal{K}\Gamma') $ is $ \beta_1(M)-1$. Since $ j_{n+1 }: \Gamma' \to\G$ is a monomorphism, by [ @CH1 Lemma 4.2 ], this rank is the like as the $ \mathcal{K}\Gamma$-rank of $ H_1(\pi_1(M);\mathcal{K}\Gamma)$ consociate to the coefficient system $ \phi$. But this is precisely the rank guess of Theorem \[thm: rho=0\ ] in the case that $ \Gamma = B / B_{{\scriptscriptstyle}H}^{{\scriptscriptstyle}(n+1)}$.
notice that the final claim of the theorem was already lay down by the final part of Proposition \[prop: nsolvable\ ] applied to the $ (n+1)$-solution $ W$.
Suppose $ M$ is rationally $ (n.5)$-solvable via $ W$ such that the coefficient system carry to $ \psi:\pi_1(W)\to \Gamma$. Then $ \rho(M, \phi)$ is given by $ \sigma^{(2)}_{\Gamma}(W)-\sigma(W)$, where $ \sigma^{(2)}_{\Gamma}(W)$ by Definition \[def: rhoinvariant\ ]. get $ \tilde{I}$ announce the image of the map $ $ H_2(\partial W;\mathcal{K}\G)\overset{j_*}{\lra}H_2(W;\mathcal{K}\G).$$ Since the succession $ $ 0\lra \tilde { I } \lra H_2(W;\mathcal{K}\G)\lra H_2(W;\mathcal{K}\G)/\tilde { I}\lra 0$$ is cleave exact (since all $ \mathcal{K}\Gamma$-modules are free) it follows that $ $ H_2(W;\mathcal{K}\G)\cong \tilde{I}\oplus (H_2(W;\mathcal{K}\G)/\tilde { I}).$$ Since the intersection imprint $ $ \lambda: H_2(W;\mathcal{K}\G)\to Hom(H_2(W;\mathcal{K}\G | \]. Leh $\Gamma '= \pi_1(M)/\pi_1(M)^{(n+1)}_H$. Then, bn Proposition \[obsgrcation2\], the rznk hypoghesis of Corollary \[cor:n.5solvaule\] us eqyivalent to the fact tfat $\mathcwl{K}\Gamma '$-ranj of $H_1(\pi_1(M);\mavgcal{K}\Gamma ')$ ia $\betc_1(M)-1$. Since $j_{n+1}: \Gamms '\to\G$ is a monomorphism, ty [@CK1 Lemma 4.2], this rank is the same as thq $\mathcsl{N}\Gamma$-rank of $R_1(\pi_1(M);\kwthczl{K}\Gamma)$ associated to the coefficjent syvtem $\phi$. But yhis is precisely the rank hyplthesis of Theorem \[hhm:rho=0\] in tye cwwe that $\Gammx=B/B_{{\scriptsbxiptstyle}H}^{{\sdriptscriptstyle}(n+1)}$.
Note that the winal claim of jkw tjgorem was alceady vstablished bn the ldst pary of Propositipn \[pcop:nwolvable\] applied to tie $(n+1)$-solution $W$.
Supposg $M$ is rathoually $(n.5)$-solvable via $W$ sych tvat dhe ziefwicjeit aystem exvends to $\psj:\pi_1(W)\to \Gammq$. Then $\rho(M, \phi)$ is bidvm by $\sigma^{(2)}_{\Gajma}(W)-\sidmw(W)$, where $\sigma^{(2)}_{\Gamma}(W)$ by Definition \[def:rhminbariant\]. Let $\tilde{I}$ denore the image of the mwp $$H_2(\partiwl W;\mathcal{K}\G)\overset{j_*}{\lra}H_2(W;\mathcal{K}\G).$$ Since the seqgence $$0\ura \tllde {U} \pra H_2(W;\mathcal{K}\G)\lra H_2(W;\mathcal{K}\G)/\tilde {I}\lra 0$$ is fllot exact (since ajl $\mathcal{K}\Balms$-iodules are ftee) it yklmows that $$H_2(W;\mathcap{K}\G)\cong \tildw{I}\oplus (H_2(R;\matncal{K}\G)/\tilde {I}).$$ Since the intwrsection fogm $$\lqmbda: H_2(W;\mathcal{K}\G)\tl Hom(H_2(W;\matheal{K}\G | \]. Let $\Gamma '= \pi_1(M)/\pi_1(M)^{(n+1)}_H$. Then, by the hypothesis of \[cor:n.5solvable\] is equivalent '$-rank $H_1(\pi_1(M);\mathcal{K}\Gamma ')$ is Since $j_{n+1}: \Gamma is a monomorphism, by [@CH1 Lemma this rank is the same as the $\mathcal{K}\Gamma$-rank of $H_1(\pi_1(M);\mathcal{K}\Gamma)$ associated to the system $\phi$. But this is precisely the rank hypothesis of Theorem \[thm:rho=0\] in case $\Gamma=B/B_{{\scriptscriptstyle}H}^{{\scriptscriptstyle}(n+1)}$. that final claim of the theorem was already established by the last part of Proposition \[prop:nsolvable\] applied the $(n+1)$-solution $W$. Suppose $M$ is rationally $(n.5)$-solvable $W$ such that the system extends to $\psi:\pi_1(W)\to \Gamma$. $\rho(M, is given $\sigma^{(2)}_{\Gamma}(W)-\sigma(W)$, $\sigma^{(2)}_{\Gamma}(W)$ Definition \[def:rhoinvariant\]. Let denote the image of the map $$H_2(\partial W;\mathcal{K}\G)\overset{j_*}{\lra}H_2(W;\mathcal{K}\G).$$ Since the sequence $$0\lra \tilde {I} \lra H_2(W;\mathcal{K}\G)\lra H_2(W;\mathcal{K}\G)/\tilde 0$$ is (since all are it that $$H_2(W;\mathcal{K}\G)\cong \tilde{I}\oplus Since the intersection form $$\lambda: H_2(W;\mathcal{K}\G)\to | \]. Let $\Gamma '= \pi_1(M)/\pi_1(M)^{(n+1)}_H$. Then, by ProPosition \[obServaTioN2\], thE rAnk hYpotHesis of CorollaRY \[cor:N.5solvable\] is equivalent tO the fAcT That $\MAtHcal{K}\gamma '$-raNK oF $h_1(\Pi_1(M);\MaThCal{k}\GAMmA ')$ is $\beTa_1(M)-1$. since $j_{n+1}: \gamma '\to\G$ is A moNoMorphism, by [@CH1 lEmMa 4.2], this rank Is tHe same as the $\mAthCal{K}\GaMmA$-raNK of $H_1(\pI_1(M);\mAthcaL{K}\GammA)$ AssociAted to the CoEFficieNT system $\PHI$. BUt thIs is precisely the rANk HYpothesis of TheOrem \[thM:rHO=0\] iN THe cAse That $\Gamma=B/b_{{\sCriptSCriptstYLe}h}^{{\SCRipTScriptstyle}(n+1)}$.
NOte that the fINal Claim oF tHe tHEorem wAs alrEaDY esTablished by The lAst part of proposITion \[proP:NsolvabLe\] applIed To tHe $(n+1)$-sOLuTiOn $W$.
suPPosE $m$ iS raTIonAlly $(n.5)$-solVaBlE via $W$ Such THAT The cOefFiciEnt syStem extends to $\Psi:\Pi_1(W)\tO \gamMa$. TheN $\rho(M, \Phi)$ iS gIven bY $\sigma^{(2)}_{\gamma}(w)-\sIgma(W)$, where $\sigma^{(2)}_{\gammA}(W)$ by DefinItiOn \[Def:RhOinvaRIant\]. LeT $\tiLde{i}$ denote The imagE Of tHe MAP $$h_2(\pArtial W;\mathcal{K}\G)\ovErSET{j_*}{\Lra}H_2(W;\matHcal{K}\G).$$ sInCe THe sequenCe $$0\Lra \TildE {i} \Lra H_2(W;\MathCAl{k}\G)\lra H_2(W;\mAthcal{k}\g)/\tIlDe {I}\lra 0$$ iS sPlit exAcT (siNce All $\maTHcal{k}\Gamma$-Modules aRe freE) It follows that $$H_2(w;\Mathcal{K}\G)\cong \TIlDE{i}\oPLus (H_2(w;\maThcal{K}\G)/\tildE {I}).$$ SiNCe thE intERsEctIOn forM $$\lambDa: h_2(w;\mAThcal{K}\G)\to Hom(H_2(W;\mathcAl{k}\G | \]. Let $\Gamma '= \pi_1(M )/\pi_1(M) ^{(n+ 1)} _H$ .Then , by Proposition \ [ obse rvation2\], the rank h ypoth es i s of Co rolla ry \[co r :n . 5 sol va bl e\] i s e quiva len t to th e fact tha t $ \m athcal{K}\Ga m ma '$-rank o f $ H_1(\pi_1(M) ;\m athcal {K }\G a mma ' )$is $\ beta_1 ( M)-1$. Since $j _{ n +1}: \ G amma '\ t o \G $ is a monomorphism,b y[ @CH1 Lemma 4.2 ], thi sr an k isthe same as t he $\ma t hcal{K} \ Ga m m a $-r a nk of $H_1(\p i_1(M);\mat h cal {K}\Ga mm a)$ associ atedto the coefficien t sy stem $\ph i$. Bu t this i s precis ely th e r ank hyp o th es isof The o re m \ [ thm :rho=0\] i nthe c aset h a t $\G amm a=B/ B_{{\ scriptscripts tyl e}H} ^ {{\ scrip tscri ptst yl e}(n+ 1)}$.
Note t hat the final c laim of the t heo re m w as alre a dy est abl ish ed by t he last par to f Pr oposition \[prop:n so l v ab le\] app lied t o t he $(n+1)$- so lut ion$ W $.
S uppo s e$M$ is r ationa l ly $ (n.5)$- so lvable v ia$W$ such that the c oefficie nt sy s tem extends to $\psi:\pi_1(W ) \t o \G a mma$ . T hen $\rho(M , \p h i)$is g i ve n b y $\si gma^{ (2 ) }_ { \Gamma}(W)-\sigma(W )$ , wher e $\s igma^{(2)}_{\ Gamma}(W)$ b y Definit ion\ [d e f:rhoinvariant \]. L et $\tilde { I}$ deno te th e imageof the ma p $$H_2(\p art ial W; \ma t h ca l{K}\G)\overs e t {j_* }{ \lra}H_ 2(W ;\mathc al{ K}\ G). $$Si nce the s equence$$ 0\ lr a\ti lde { I } \lra H _2 (W; \m ath cal{K } \G)\lr a H_2 (W;\ ma th c al{ K}\G)/\ t il d e {I} \l ra 0$$ is s plitexac t (s ince al l $\mathc al{ K }\Ga mm a$ -module s are free) i tfollows th at $$ H_2(W; \ m athcal{K }\G)\cong \tilde{I}\opl u s (H_2( W;\ mathc al{K }\G)/\til de{I}).$ $ S i nce th e inte rsect io n f o r m $$\ l a mb da: H _2(W;\math c a l{K }\G)\ to Hom (H_2(W; \mathcal{K}\G | \]. Let_$\Gamma '=_\pi_1(M)/\pi_1(M)^{(n+1)}_H$. Then, by Proposition \[observation2\],_the rank_hypothesis_of Corollary \[cor:n.5solvable\]_is_equivalent to the_fact that $\mathcal{K}\Gamma_'$-rank of $H_1(\pi_1(M);\mathcal{K}\Gamma ')$_is $\beta_1(M)-1$. Since_$j_{n+1}:_\Gamma '\to\G$ is a monomorphism, by [@CH1 Lemma 4.2], this rank is the same_as_the $\mathcal{K}\Gamma$-rank_of_$H_1(\pi_1(M);\mathcal{K}\Gamma)$_associated to the coefficient system_$\phi$. But this is precisely_the rank_hypothesis of Theorem \[thm:rho=0\] in the case that $\Gamma=B/B_{{\scriptscriptstyle}H}^{{\scriptscriptstyle}(n+1)}$.
Note_that_the final claim_of the theorem was already established by the last_part of Proposition \[prop:nsolvable\] applied to the_$(n+1)$-solution $W$.
Suppose $M$_is_rationally_$(n.5)$-solvable via $W$ such_that the coefficient system extends to_$\psi:\pi_1(W)\to \Gamma$. Then $\rho(M, \phi)$ is_given by $\sigma^{(2)}_{\Gamma}(W)-\sigma(W)$, where $\sigma^{(2)}_{\Gamma}(W)$ by Definition \[def:rhoinvariant\]._Let $\tilde{I}$ denote the image of_the map $$H_2(\partial W;\mathcal{K}\G)\overset{j_*}{\lra}H_2(W;\mathcal{K}\G).$$ Since_the sequence_$$0\lra \tilde {I} \lra H_2(W;\mathcal{K}\G)\lra_H_2(W;\mathcal{K}\G)/\tilde {I}\lra 0$$_is split_exact (since all_$\mathcal{K}\Gamma$-modules are free) it follows that_$$H_2(W;\mathcal{K}\G)\cong \tilde{I}\oplus (H_2(W;\mathcal{K}\G)/\tilde_{I}).$$ Since the intersection form $$\lambda:_H_2(W;\mathcal{K}\G)\to_Hom(H_2(W;\mathcal{K}\G |
close enough to permit the transition from two $j=0$ opposing simple loops into two opposing $j=\frac{1}{2}$ twisted loops. A possible geometric scenario for the transformation of two simple loops of current (yons) with opposite rotations into two $j=\frac{1}{2}$ twisted loops of current (preons) is shown in Fig. 3. To implement this scenario one would expect to go beyond the earlier considerations of this paper. Without attempting to do this, one notes according to Fig. 3 that the fusion of two yons may result in a doublet of preons as twisted loops, which might also qualify as Higgs doublets.
[c c c]{}\
& &\
\[-1.7cm\] & &\
\[-0.7cm\] & &\
Two $j=0$ neutral loops & gravitational attraction & interaction causing the crossing or\
\[-0.55cm\] with opposite topological & & redirection of neutral current flux\
\[-0.55cm\] rotation & & shown below\
\[-0.55cm\]
---------------------------------- ------------
\[-1cm\]
\[-1.1cm\]
\[-0.95cm\]
\[-0.9cm\] a preon c preon
\[-0.4cm\] $r=0$ $r=0$
\[-0.4cm\] $w_a = +1$ $w_c=-1$
\[-0.4cm\] $ Q_a = -\frac{e}{3}$ $ Q_c = 0$
---------------------------------- ------------
In the scenario suggested by Figure 3 the opposing states are quantum mechanically entangled and may undergo gravitational exchange scattering. The $\binom{c}{a}$ doublet of Fig. 3 is similar to the Higgs doublet which is independently required to be a SLq(2) singlet $(j=0)$ and a SU(2) charge doublet $(t=\frac{1}{2})$ by the mass term of the Lagrangian described in references 8 and 9. Since the Higgs mass contributes to the inertial mass, one should expect a fundamental connection with the gravitational field at this point.
If at an early cosmological time, only a fraction of the initial gas of the quantum loops, the yons, had been converted to preons and these in turn had led to | close enough to permit the transition from two $ j=0 $ opposing childlike loop into two opposing $ j=\frac{1}{2}$ twist loop. A possible geometric scenario for the transformation of two simple loop of current (yons) with opposite rotations into two $ j=\frac{1}{2}$ twist loops of current (preons) is shown in Fig. 3. To follow through this scenario one would expect to go beyond the early consideration of this paper. Without try to do this, one notes harmonize to Fig. 3 that the fusion of two yons may result in a doublet of preons as twisted loop, which might also qualify as Higgs doublets.
[ c c c]{}\
& & \
\[-1.7cm\ ] & & \
\[-0.7cm\ ] & & \
Two $ j=0 $ neutral cringle & gravitational attraction & interaction causing the crossing or\
\[-0.55cm\ ] with opposite topological & & redirection of neutral current flux\
\[-0.55cm\ ] rotation & & prove below\
\[-0.55cm\ ]
---------------------------------- ------------
\[-1cm\ ]
\[-1.1cm\ ]
\[-0.95cm\ ]
\[-0.9cm\ ] a preon c preon
\[-0.4cm\ ] $ r=0 $ $ r=0 $
\[-0.4cm\ ] $ w_a = +1 $ $ w_c=-1 $
\[-0.4cm\ ] $ Q_a = -\frac{e}{3}$ $ Q_c = 0 $
---------------------------------- ------------
In the scenario suggested by Figure 3 the opposing states are quantum mechanically entangled and may undergo gravitational substitution scattering. The $ \binom{c}{a}$ doublet of Fig. 3 is similar to the Higgs doublet which is independently required to be a SLq(2) singlet $ (j=0)$ and a SU(2) charge doublet $ (t=\frac{1}{2})$ by the mass term of the Lagrangian described in references 8 and 9. Since the Higgs mass contributes to the inertial batch, one should expect a fundamental joining with the gravitational sphere at this point.
If at an early cosmological time, merely a fraction of the initial gas of the quantum loops, the yons, had been converted to preons and these in turn had head to | cllse enough to permit the transition from two $j=0$ opposjng simpue loops into two opposing $j=\hrac{1}{2}$ twisuvd loops. A possible gdometric dcenario for rhe transformation of twk sim'lx loops of currgnt (yons) witv opposite rotdtkous into two $j=\frac{1}{2}$ twisted loops of ctrrent (lrfons) is shown yn Fpg. 3. To pmklement this scenario one would espect tm go beyond tne earlier considerations lf tjis paper. Without wttempting jk dj this, one noges accordpug to Fig. 3 jhat the fusion of two yons may fesulc in a douboer ov preons as vwisteq loops, whicm might also qialify as Higgx dmuboets.
[c c c]{}\
& &\
\[-1.7cm\] & &\
\[-0.7cm\] & &\
Two $j=0$ neutral loops & dravitatimncl attraction & interaxtuon cdusitg tfw cfosaiig kr\
\[-0.55cm\] wlth opposite fopological & & redirection of ntutwql current flhx\
\[-0.55cm\] rjtwtion & & shown below\
\[-0.55cm\]
---------------------------------- ------------
\[-1cm\]
\[-1.1cm\]
\[-0.95cm\]
\[-0.9fm\] a preon c preon
\[-0.4cm\] $r=0$ $r=0$
\[-0.4cm\] $w_a = +1$ $w_c=-1$
\[-0.4cm\] $ Q_a = -\frac{z}{3}$ $ Q_c = 0$
---------------------------------- ------------
In the scfnario fuggewted by Fygurr 3 the opposing states are wuantum mechcnixally entangled anb may undergu grsvitayional exchange scatternng. Ths $\binom{c}{a}$ dlublet of Wig. 3 is similar go nhe Viggs doublet which is indqpendentlb reqbired to be s SLq(2) finglet $(j=0)$ wnd a SU(2) charge doublet $(h=\frac{1}{2})$ bf the mass term of the Lagrangian describxv in referencgs 8 ang 9. Since the Miggs mass contwibutes to the inerticl masr, one shoumd expert a fundamegtal connectimj with the gcavitatiogal diele at thkr point.
If at am early cosmologicao time, only a fracbion ur the initial gcr of the quantum loups, tje yjts, had been wonvdrtdc to oreons and bhere im turn had led to | close enough to permit the transition from opposing loops into opposing $j=\frac{1}{2}$ twisted for transformation of two loops of current with opposite rotations into two $j=\frac{1}{2}$ loops of current (preons) is shown in Fig. 3. To implement this scenario would expect to go beyond the earlier considerations of this paper. Without attempting do one according Fig. 3 that the fusion of two yons may result in a doublet of preons as loops, which might also qualify as Higgs doublets. c c]{}\ & &\ & &\ \[-0.7cm\] & &\ $j=0$ loops & attraction interaction the crossing or\ with opposite topological & & redirection of neutral current flux\ \[-0.55cm\] rotation & & shown below\ \[-0.55cm\] ------------ \[-1cm\] \[-0.9cm\] a c \[-0.4cm\] $r=0$ \[-0.4cm\] $w_a $w_c=-1$ \[-0.4cm\] $ Q_a = -\frac{e}{3}$ 0$ ---------------------------------- ------------ In the scenario suggested by 3 the states are quantum mechanically entangled and undergo gravitational exchange scattering. The $\binom{c}{a}$ doublet of 3 is similar to the Higgs doublet which is independently required to be a SLq(2) and a SU(2) charge $(t=\frac{1}{2})$ by the term the described references 8 9. Since the Higgs mass contributes to the inertial mass, one expect a fundamental connection with the gravitational field at this at early cosmological time, a fraction of the gas the quantum loops, the been to in had to | close enough to permit the traNsition froM two $j=0$ OppOsiNg SimpLe loOps into two oppoSIng $j=\Frac{1}{2}$ twisted loops. A possiBle geOmETric SCeNario For the tRAnSFOrmAtIoN of TwO SiMple lOopS of currEnt (yons) witH opPoSite rotationS InTo two $j=\frac{1}{2}$ TwiSted loops of cUrrEnt (preOnS) is SHown iN FiG. 3. To imPlemenT This scEnario one WoULd expeCT to go beYONd The eArlier consideratiONs OF this paper. WithOut attEmPTiNG To dO thIs, one notes AcCordiNG to Fig. 3 tHAt THE FusIOn of two yons maY result in a dOUblEt of prEoNs aS TwisteD loopS, wHIch Might also quAlifY as Higgs dOubletS.
[C c c]{}\
& &\
\[-1.7cm\] & &\
\[-0.7cm\] & &\
tWo $j=0$ neutRal looPs & gRavItatIOnAl AttRaCTioN & InTerACtiOn causinG tHe CrossIng oR\
\[-0.55CM\] WIth oPpoSite TopolOgical & & redirecTioN of nEUtrAl curRent fLux\
\[-0.55cM\] rOtatiOn & & showN beloW\
\[-0.55cM\]
---------------------------------- ------------
\[-1cm\]
\[-1.1cm\]
\[-0.95cm\]
\[-0.9cm\] a preon C preOn
\[-0.4cm\] $r=0$ $r=0$
\[-0.4cm\] $w_A = +1$ $w_c=-1$
\[-0.4Cm\] $ q_a = -\fRaC{e}{3}$ $ Q_c = 0$
---------------------------------- ------------
IN The sceNarIo sUggesteD by FiguRE 3 thE oPPOSiNg states are quantum MeCHAnIcally enTangleD AnD mAY undergo GrAviTatiONAl excHangE ScAttering. the $\binOM{c}{A}$ dOublet oF FIg. 3 is siMiLar To tHe HigGS douBlet whIch is indEpendENtly required to BE a SLq(2) singlet $(j=0)$ ANd A su(2) cHArge DouBlet $(t=\frac{1}{2})$ by The mASs teRm of THe lagRAngiaN descRiBEd IN references 8 and 9. Since ThE Higgs Mass cOntributes to tHe inertial MASS, one shouLd exPEcT A fundamental coNnectIon with the GRavitatiOnal fIeld at thIs point.
If AT An early cOsmOloGicAl tIME, oNly a fraction oF THe inItIal gas oF thE quantuM loOps, The YonS, hAd been conVerted to PrEoNs AnD thEse in TUrn had leD tO | close enough to permit th e transiti on fr omtwo $ j=0$ opp osing simple l o opsinto two opposing $j=\ frac{ 1} { 2}$t wi stedloops.A p o s sib le g eom et r ic scen ari o for t he transfo rma ti on of two si m pl e loops of cu rrent (yons) wi th opp os ite rotat ion s int o two$ j=\fra c{1}{2}$tw i sted l o ops ofc u rr ent(preons) is shown in Fig. 3. To imp lement t h is s cen ari o one woul dexpec t to gob ey o n d th e earlier cons iderationso f t his pa pe r.W ithout atte mp t ing to do this , on e notes a ccordi n g to Fi g . 3 tha t thefus ion oft wo y ons m a y r e su lti n a doublet o fpreon s as t w i sted lo ops, whic h might alsoqua lify asHiggs doub lets .
[c c c]{}\
& &\
\ [-1.7cm\] & &\\[-0 .7cm\] &&\Tw o $ j= 0$ ne u tral l oop s & gravit ational att ra c t i on & interaction cau si n g t he cross ing or \
\ [- 0 .55cm\]wi thoppo s i te to polo g ic al & & r edirec t io nof neut ra l curr en t f lux \
\[- 0 .55c m\] ro tation & & sh o wn below\
\[-0 . 55cm\]
--- - -- - - -- - ---- --- ----------- ---- - --- ---- - -- -- \[-1cm\]
\ [-1.1cm\]
\[ -0.95 cm\] \[ -0. 9cm \]a pr eon c pre on
\[-0. 4cm \]$r= 0$ $r=0$
\ [-0 .4cm\ ] $w_a =+1 $ $w _ c=-1$ \[ -0.4 cm \] $ Q _a = -\ f ra c { e}{3 }$ $ Q _c=0$
---- - --- ------- --------- --- - ---- -- - ------- ----
In thesc enario sug ge ste d by F i g ure 3 th e opposing states are q u antum m ech anica llyentangled an d mayund e rgo gr avitat ional e xch a n ge sc a t te rin g. The $\bin o m {c} {a}$do uble t of Fi g. 3 is similar to the Higgs double t w hich i sind e pe n den tl y re q u ired to be a SL q(2) singl et $( j=0)$ anda SU (2 ) charg e doubl et $( t =\frac{ 1}{2})$ b y the mas sterm o f t he Lagrang ian desc ribed inr efere n ce s 8 a nd9. Sin ce th e Hig gs mas s co ntrib utes t othe in ertia lmass, on e should expect a funda mental conn ect ion withthe gra vitationa l fi eld at thi s p oin t.
I f a t an e arly co smo l ogica l ti m e, only a fr act i o nof the init i a l ga s ofthe quantu m lo ops, the yons, ha d been converte d to p reo nsa nd t he se in turn had le dt o | close_enough to_permit the transition from_two $j=0$_opposing_simple loops_into_two opposing $j=\frac{1}{2}$_twisted loops. A_possible geometric scenario for_the transformation of_two_simple loops of current (yons) with opposite rotations into two $j=\frac{1}{2}$ twisted loops of_current_(preons) is_shown_in_Fig. 3. To implement this_scenario one would expect to_go beyond_the earlier considerations of this paper. Without attempting_to_do this, one_notes according to Fig. 3 that the fusion of_two yons may result in a_doublet of preons_as_twisted_loops, which might also_qualify as Higgs doublets.
[c c c]{}\
&_&\
\[-1.7cm\] & &\
\[-0.7cm\] & &\
Two $j=0$_neutral loops & gravitational attraction & interaction_causing the crossing or\
\[-0.55cm\] with opposite_topological & & redirection of_neutral current_flux\
\[-0.55cm\] rotation & & shown_below\
\[-0.55cm\]
----------------------------------_------------
_ _ _ _ __ ___ _ __ __
__ _ \[-1cm\]__ _ _
_ _ _ \[-1.1cm\] _ _
_ __ __ _ \[-0.95cm\] _ _ _
__ _\[-0.9cm\] a preon _ ___c preon
_ _ \[-0.4cm\] $r=0$_ __ $r=0$
__ \[-0.4cm\]_$w_a_=_+1$ _ $w_c=-1$
_ \[-0.4cm\] $ Q_a = -\frac{e}{3}$ _$ Q_c =_0$
---------------------------------- ------------
In the_scenario_suggested_by Figure 3 the opposing states are quantum mechanically entangled_and may_undergo gravitational exchange_scattering. The $\binom{c}{a}$ doublet of Fig. 3 is similar to_the Higgs doublet which is independently required_to be a SLq(2) singlet $(j=0)$ and a SU(2) charge doublet_$(t=\frac{1}{2})$ by the mass term of the Lagrangian_described in references 8 and 9. Since_the Higgs mass contributes_to_the inertial mass, one should expect a_fundamental_connection_with_the gravitational field_at this point.
If_at an early_cosmological time,_only a fraction_of the_initial gas of_the_quantum loops, the yons, had been converted to preons and these in turn had led to |
-F bonds in the eclipsed conformation. Figure 2 shows as function of rotation of the carbon-sulfur bond the atom centered partial charges obtained for the triflic acid molecule. The sulfur-oxygen(hydroxyl) rotational potential energy surface possesses two physically equivalent minima (Figure 3). The lowest barrier between these minima corresponds to the proton directed axially, out into the solvent and has an energy of 2.0 kcal/mol. Electrostatic solvation substantially flattens that barrier so that the maximum free energy between the two minima is less than one kcal/mol. This stabilization of the barrier configuration is due to the better access of the proton to the solvent. The position of a probe water molecule interacting with triflic acid was determined by fully optimizing the pair of molecules at the MP2 level using both the 6-31G\*\* and D95\*\* basis sets. Recently, Ricchiardi and Ugliengo [@Ricchiardi] reported a geometry optimization of triflic acid plus a water molecule at the SCF/DZP level. Their result is qualitatively similar but quantitatively substantially different from that presented below. Because our results differed from those previous ones, several different basis sets and starting conditions were considered. The differences we observed among those optimizations were slight and do not account for the differences with the previous work that are likely traceable to different level of effort in achieving the optimum structure. Our results are shown in Figure 4. That structure is of lower energy than the structure reported by Ricchiardi and Ugliengo. However, our experience with this optimization is that the triflic-acid-water configurations are floppy as might be expected from noncovalent hydrogen-bonding interactions. For example, we several times found a local minimum of the energy surface with the exterior proton of Figure 4 directed into the plane of that drawing. Potential energy minima of that sort are only slightly higher in energy, of the order of one kcal/mol, than the structure shown. Note that the dihedral angle positioning the triflic acid proton is nearly the same in the minimum energy complex and the minimum energy triflic acid molecule (Figure 3). The floppiness of these structures is consistent with the prediction of the dielectric solvation model that free energy varies only gently with disposition of the acid proton in the region between the two minima (Figure 3). Attempts were made to determine a minimum energy conformation for triflate-hydronium ion pair but those configurations always relaxed to the optimum geometry of | -F bonds in the eclipsed conformation. Figure 2 show as affair of rotation of the carbon - sulfur bond the atom centered fond charges obtained for the triflic acid molecule. The sulfur - oxygen(hydroxyl) rotational likely department of energy surface possesses two physically equivalent minima (Figure 3). The lowest barrier between these minima corresponds to the proton directed axially, out into the solution and has an energy of 2.0 kcal / mol. Electrostatic solvation substantially flattens that barrier so that the maximal free energy between the two minima is less than one kcal / mol. This stabilization of the barrier shape is due to the good entree of the proton to the solvent. The position of a probe water atom interacting with triflic acid was determined by fully optimizing the pair of molecule at the MP2 level using both the 6 - 31G\*\ * and D95\*\ * basis sets. Recently, Ricchiardi and Ugliengo [ @Ricchiardi ] reported a geometry optimization of triflic acid plus a urine molecule at the SCF / DZP level. Their result is qualitatively similar but quantitatively substantially different from that presented below. Because our results differed from those previous ones, several different basis sets and starting condition were considered. The differences we observe among those optimization were slender and do not account for the differences with the previous work that are likely traceable to different level of campaign in achieving the optimum structure. Our results are shown in Figure 4. That structure is of lower department of energy than the structure reported by Ricchiardi and Ugliengo. However, our experience with this optimization is that the triflic - acid - water configurations are floppy as might be expected from noncovalent hydrogen - bonding interactions. For model, we several times found a local minimum of the energy surface with the exterior proton of trope 4 directed into the plane of that drawing. Potential energy minima of that sort are only slightly high in energy, of the decree of one kcal / mol, than the structure read. Note that the dihedral slant positioning the triflic acid proton is nearly the same in the minimum department of energy complex and the minimum energy triflic acid atom (Figure 3). The floppiness of these structures is consistent with the prediction of the dielectric solvation model that free energy varies only gently with disposition of the acid proton in the region between the two minima (Figure 3). Attempts were make to determine a minimum department of energy conformation for triflate - hydronium ion pair but those configurations always relaxed to the optimum geometry of | -F blnds in the eclipsed connormation. Figure 2 shows as fuhction ow rotation of the carbon-sulfnr bind tye atom centered partixl charged obtainwd fie the triflic acid molechpe. Tke sulfur-oxygen(hidroxyl) rotadional potentidl euergy surface possesses two physicaljy equifapent minima (Fidure 3). The lowest barrier between these minija corrtsponds to the proyon directed axially, out ijto hhe solvent and had an energy of 2.0 jcal/mol. Elecgrostatic solvation sugstantially flattens that barridr so that the nazimkk free enerjy betreen the two minima hs less than one kcal/kol. Thus stabilization of tie barrier configurajion is dua co the better access if the krotot to rhe somvxnt. The plsivion of a pdobe water nolecule interactinb rpyh triflic adid waf qetermined by fully optimizing the pair of molecules at the MP2 lecel using both the 6-31G\*\* wnd D95\*\* basys sets. Recently, Ricchiardi and Ugliengo [@Ricchiargi] re'ofteb a geunehry optimization of triflic acid plus a water jokebule at the SCF/DZI level. Their resukt ix qualitativeli similcd gut quantitatively substagtialoy differtnt ftom that presented below. Bexause our refylts differed from those prevnous omes, srveral different basis rets and startijg conditjuns were considefed. Tve differences we observed among thise pptimizxtioms werq slight ajd do not account for thf difyerenwes with tje previous work that are likelb traceable tp gifxerent lzvel on effort in achyeving the optnmum strbcture. Our resulns are shmwn in Figuwe 4. That struwjure is of loxer energr thqn tye struzgure reported ny Ricchicxdi and Utliengo. However, ouv expgrjence with this opuinization is thst ghe tgifkis-dcid-water cotfigjragoons xre flok'y cs mignt be expected from tonckvalent hydrogen-bomdlng interqctions. Sor example, wr several times foknd a loral mimimom of the energy surface with tge exterilr iroton of Figtre 4 directed inco the plane of that drawing. Potential eiergy minima of that soet are only slightli hlgher in enecgy, of the ordes of one kcal/mol, thab the structure smown. Note that the dihsdral dngle positioning the triflic acid proton is nearly the same in the minimum enwrgy complex and tge monimuk znexgy trislic ecid molecule (Figlre 3). The floppiness of these strnctures is cjnsistent with the predictiom uf the dielecjric solvation model that rree enetgy varies only gently with dislosition of the acid probon in thw regjon bensexn the two minima (Fibure 3). Cttemprs were madc to determine a kikimum enerhy cpnyormation for triflate-hydronium iin paie hut those confignrations alwwns relzxed tj the optimum gekmetdj of | -F bonds in the eclipsed conformation. Figure as of rotation the carbon-sulfur bond obtained the triflic acid The sulfur-oxygen(hydroxyl) rotational energy surface possesses two physically equivalent (Figure 3). The lowest barrier between these minima corresponds to the proton directed out into the solvent and has an energy of 2.0 kcal/mol. Electrostatic solvation flattens barrier that maximum free energy between the two minima is less than one kcal/mol. This stabilization of the configuration is due to the better access of proton to the solvent. position of a probe water interacting triflic acid determined fully the pair of at the MP2 level using both the 6-31G\*\* and D95\*\* basis sets. Recently, Ricchiardi and Ugliengo [@Ricchiardi] a geometry triflic acid a molecule the SCF/DZP level. is qualitatively similar but quantitatively substantially presented below. Because our results differed from those ones, several basis sets and starting conditions were The differences we observed among those optimizations were and do not account for the differences with the previous work that are likely traceable level of effort in the optimum structure. results shown Figure That structure of lower energy than the structure reported by Ricchiardi and Ugliengo. our experience with this optimization is that the triflic-acid-water configurations as be expected from hydrogen-bonding interactions. For example, several found a local minimum energy with of 4 into the plane of drawing. Potential energy minima of sort are only slightly order of one kcal/mol, than the structure shown. that the dihedral angle positioning the triflic proton is nearly the same in the minimum energy complex and the energy triflic (Figure 3). The floppiness of these structures is with the prediction of dielectric solvation model that free energy varies only gently disposition the acid in the region the two minima 3). Attempts were determine a energy for but those configurations always relaxed to optimum of | -F bonds in the eclipsed conforMation. FiguRe 2 shoWs aS fuNcTion Of roTation of the carBOn-suLfur bond the atom centereD partIaL CharGEs ObtaiNed for tHE tRIFliC aCiD moLeCUlE. The sUlfUr-oxygeN(hydroxyl) rOtaTiOnal potentiaL EnErgy surfacE poSsesses two phYsiCally eQuIvaLEnt miNimA (FiguRe 3). The lOWest baRrier betwEeN These mINima corRESpOnds To the proton directED aXIally, out into thE solveNt ANd HAS an EneRgy of 2.0 kcal/mOl. electROstatic SOlVATIon SUbstantially fLattens that BArrIer so tHaT thE MaximuM free EnERgy Between the tWo miNima is lesS than oNE kcal/moL. this staBilizaTioN of The bARrIeR coNfIGurATiOn iS Due To the betTeR aCcess Of thE PROTon tO thE solVent. THe position of a ProBe waTEr mOlecuLe intEracTiNg witH triflIc aciD wAs determined by fUlly OptimizinG thE pAir Of MolecULes at tHe Mp2 leVel usinG both thE 6-31g\*\* anD D95\*\* BASIs Sets. Recently, RicchiArDI AnD UgliengO [@RicchIArDi] REported a GeOmeTry oPTImizaTion OF tRiflic acId plus A WaTeR molecuLe At the ScF/dZP LevEl. TheIR resUlt is qUalitatiVely sIMilar but quantiTAtively substaNTiALLy DIffeRenT from that prEsenTEd beLow. BECaUse OUr resUlts dIfFErED from those previous oNeS, severAl difFerent basis seTs and startING ConditioNs weRE cONsidered. The difFerenCes we obserVEd among tHose oPtimizatIons were sLIGht and do Not AccOunT foR THe Differences wiTH The pReVious woRk tHat are lIkeLy tRacEabLe To differeNt level oF eFfOrT iN acHieviNG the optiMuM stRuCtuRe. Our REsults Are shOwn iN FIgURe 4. THat struCTuRE Is of LoWeR eneRgy ThAn the StruCTurE reportEd by RicchIarDI and ugLiEngo. HowEver, our experiEnCe with this OpTimIzatioN IS that the Triflic-acid-water configuRAtions aRe fLoppy As miGht be expeCteD from nOncOValent HydrogEn-bonDiNg iNTEractIONs. for ExAmple, we sevERAl tImes fOuNd a lOcal minImum of the energy surFAce With the exteriOr pRotoN OF FIguRE 4 dIRecTeD IntO THe plane of that drAwing. PotenTiAL eNergy minimA Of tHaT sort arE only slIghtlY Higher iN energy, of The order oF oNe kcAL/Mol, Than the strUcture shOwn. Note thAT the dIHeDral aNglE positIoNinG the tRiflic ACid ProtoN is neaRlY the saMe in tHe Minimum eNergy complex and the minimUm enerGy triFliC acid moleCulE (figUre 3). The floPpinEss of these StrUctUres iS coNSisteNt wiTH tHe pREdictIon oF The dielecTRiC soLVAtIon model thaT FREe eNergy VarIEs only GentLy with disposition OF the acid proton In thE REgiOn bETweeN tHe two minima (FigUre 3). atTEMpts were MaDe to determiNe a minimUm ENergy ConforMation For trifLATe-HYdroniUm ioN paIr but thosE coNfIGuratioNs AlWAys relAxed To The optImum geOMetrY OF | -F bonds in the eclipsed c onformatio n. Fi gur e 2 s hows asfunction of ro t atio n of the carbon-sulfur bond t h e at o mcente red par t ia l cha rg es ob ta i ne d for th e trifl ic acid mo lec ul e. The sulfu r -o xygen(hydr oxy l) rotationa l p otenti al en e rgy s urf ace p ossess e s twophysicall ye quival e nt mini m a ( Figu re 3). The lowest ba r rier between t hese m in i ma c orr esp onds to th eproto n direct e da x i all y , out into th e solvent a n d h as anen erg y of 2. 0 kca l/ m ol. Electrosta ticsolvation subst a ntially flatten s that ba rri er s o t ha t t he max i mu m f r eeenergy b et we en th e tw o m i nima is les s tha n one kcal/mo l.This sta biliz ation ofth e bar rier c onfig ur ation is due to the better a cce ss of t he pr o ton to th e s olvent. The po s iti on o f a probe water molec ul e in teractin g with tr if l ic acidwa s d eter m i ned b y fu l ly optimiz ing th e p ai r of mo le culesat th e M P2 le v el u sing b oth the6-31G \ *\* and D95\*\ * basis sets.R ec e n tl y , Ri cch iardi and U glie n go [ @Ric c hi ard i ] rep orted a ge o metry optimizationof trifl ic ac id plus a wat er molecul e a t the SC F/DZ P l e vel. Their res ult i s qualitat i vely sim ilarbut quan titativel y substant ial lydif fer e n tfrom that pre s e nted b elow. B eca use our re sul tsdif fe red fromthose pr ev io us o nes , sev e ral diff er ent b asi s set s and s tarti ng c on di t ion s werec on s i dere d. T he d iff er ences weo bse rved am ong those op t imiz at io ns were slight and d onot accoun tfor the d i f ferences with the previous work that ar e l ikely tra ceable to di fferen t l e vel of effor t inac hie v i ng th e op tim um structure . Our resu lt s ar e shown in Figure 4. That str ucture is oflow er e n e rg y t h an the s t ruc t u re reported byRicchiardi a n dUgliengo.H owe ve r, ourexperie nce w i th this optimiza tion is t ha t th e tri flic-acid- water co nfigurati o ns ar e f loppy as might b e e xpect ed fro m no ncova lent h yd rogen- bondi ng interac tions. For example, wesevera l tim esfound a l oca l mi nimum oftheenergy sur fac e w ith t hee xteri or p r ot ono f Fig ure4 directed in tot h eplane of th a t dra wing. Po t ential ene rgy minima of tha t sort are only sli g h tly hi g herin energy, of th e o rd e r of onekc al/mol, tha n the st ru c tureshown. Notethat th e di h edralangl e p ositionin g t he triflic a ci d proto n is n earlythe sa m e in t he minimum energ y com p l ex an d th e min im um ener g y tr iflic acid molecule ( Figure 3). Thefloppin es s of t hes estructures is consis tentwith th epred ict ion of the d ielec tric s olv ation mod e l t h at f r eeener gy va ri es o nly gentl y with di spo s ition o fthe a cid pr o to n in the reg ion betw e e n the twom inim a ( F igure 3). A ttempt s werem ade t o deter min e a minimum energy c o nfo rm atio n for tr if late -h ydr on ium ion pair bu t thoseconfigur a t ions a lwa y s re l axed t othe optimumgeome tr y of | -F bonds_in the_eclipsed conformation. Figure 2_shows as_function_of rotation_of_the carbon-sulfur bond_the atom centered_partial charges obtained for_the triflic acid_molecule._The sulfur-oxygen(hydroxyl) rotational potential energy surface possesses two physically equivalent minima (Figure 3). The_lowest_barrier between_these_minima_corresponds to the proton directed_axially, out into the solvent_and has_an energy of 2.0 kcal/mol. Electrostatic solvation substantially_flattens_that barrier so_that the maximum free energy between the two minima_is less than one kcal/mol. This_stabilization of the_barrier_configuration_is due to the_better access of the proton to_the solvent. The position of a_probe water molecule interacting with triflic acid_was determined by fully optimizing the_pair of molecules at the_MP2 level_using both the 6-31G\*\* and_D95\*\* basis sets._Recently, Ricchiardi_and Ugliengo [@Ricchiardi]_reported a geometry optimization of triflic_acid plus a_water molecule at the SCF/DZP level._Their_result is qualitatively_similar_but_quantitatively substantially_different from that_presented_below. Because_our_results differed from those previous ones,_several_different basis sets and starting conditions were_considered. The differences we_observed_among those optimizations were_slight and do not account_for the differences with the previous_work that_are likely_traceable to different level of effort in achieving the optimum structure._Our results are shown in Figure_4. That structure is_of lower_energy_than the structure_reported_by Ricchiardi_and Ugliengo. However, our experience with this_optimization is_that the triflic-acid-water configurations are floppy_as might be expected_from_noncovalent hydrogen-bonding interactions. For example, we_several times found a local minimum_of the energy surface with_the_exterior_proton of Figure 4 directed_into the plane of that drawing._Potential energy minima_of that sort are only slightly higher_in_energy, of the order of one_kcal/mol,_than the structure shown. Note that_the_dihedral_angle positioning the triflic acid_proton is nearly the same in_the minimum energy complex and the minimum energy triflic_acid molecule (Figure_3). The floppiness of these_structures_is_consistent with the prediction of the dielectric solvation model that_free energy_varies only gently_with disposition of the acid proton in the region between_the two minima (Figure 3). Attempts were_made to determine a minimum energy conformation for triflate-hydronium ion pair_but those configurations always relaxed to the optimum_geometry of |
It implies $$f'_{i_0}\in \sum_{j\neq i_0}\mathcal{O}f'_j+\mathcal{O}f,$$ which is impossible in the case of an isolated singularity $f$.
If $E=1$, considering the leading term in $\bar{s}$ in the relation (\[eq3.3\]), we obtain that $1$ belongs to the maximal ideal of $W$, a contradiction.$\square$
*Remark :* After the preceding lemma, we have $\beta_1>n+1.$ Since in the smooth case, $\beta_1=n+1$, we conclude that $\beta_1$ characterizes the smoothness (among germs for which $0$ is at most an isolated singularity).
We are now ready to characterize the quasi-homogeneity. If $f$ is quasi homogeneous, the good operator has degree $1$ in $s$, is of the form $s-\theta$ with $\theta(f)=f$. From the resolution of $\textrm{bigr}N$ we constructed in Section 3.1, we deduce that $\textrm{ann}_W \delta$ is minimally generated by $f,(f'_i\partial_t),(f'_i\partial_{x_j}-f'_j\partial_{x_i}),s-\theta.$ The corresponding shifts for the filtration $F$ are respectively $0,1,1,1$.
If now $f$ is not quasi homogeneous, then $S_L$ has degree in $s$ greater than $1$. Indeed, suppose on the contrary that there exists a good operator anihilating $f^s$ of the form $s+P$ with $P=u+\theta$, $u\in\mathcal{O}$ and $\theta$ a vector field. We have $$0=(s+P)f^s=(s(1+\theta(f)\frac{1}{f})+u)f^s.$$ Substituting $s=0$, we get $u=0$. So $\theta(f)\frac{1}{f}=-1$ and $f\in J(f)$. This implies that $f$ is quasi homogeneous by [@saito71].
\[prop20\] $f$ is quasi homogeneous if and only if $\textrm{reg}_F N=0$.
#### Proof
If $f$ is not quasi homogeneous, | It implies $ $ f'_{i_0}\in \sum_{j\neq i_0}\mathcal{O}f'_j+\mathcal{O}f,$$ which is impossible in the case of an isolated singularity $ f$.
If $ E=1 $, considering the lead condition in $ \bar{s}$ in the relation (\[eq3.3\ ]), we obtain that $ 1 $ belongs to the maximal ideal of $ W$, a contradiction.$\square$
* Remark :* After the precede lemma, we have $ \beta_1 > n+1.$ Since in the smooth character, $ \beta_1 = n+1 $, we reason that $ \beta_1 $ characterizes the smoothness (among germs for which $ 0 $ is at most an isolated singularity).
We are now ready to qualify the quasi - homogeneity. If $ f$ is quasi homogeneous, the good operator has degree $ 1 $ in $ s$, is of the form $ s-\theta$ with $ \theta(f)=f$. From the settlement of $ \textrm{bigr}N$ we constructed in Section 3.1, we deduce that $ \textrm{ann}_W \delta$ is minimally generated by $ f,(f'_i\partial_t),(f'_i\partial_{x_j}-f'_j\partial_{x_i}),s-\theta.$ The corresponding shifts for the filtration $ F$ are respectively $ 0,1,1,1$.
If nowadays $ f$ is not quasi homogeneous, then $ S_L$ has degree in $ s$ greater than $ 1$. Indeed, suppose on the reverse that there exists a good hustler anihilating $ f^s$ of the form $ s+P$ with $ P = u+\theta$, $ u\in\mathcal{O}$ and $ \theta$ a vector airfield. We have $ $ 0=(s+P)f^s=(s(1+\theta(f)\frac{1}{f})+u)f^s.$$ substitute $ s=0 $, we get $ u=0$. So $ \theta(f)\frac{1}{f}=-1 $ and $ f\in J(f)$. This implies that $ f$ is quasi homogeneous by [ @saito71 ].
\[prop20\ ] $ f$ is quasi homogeneous if and entirely if $ \textrm{reg}_F N=0$.
# # # # Proof
If $ f$ is not quasi homogeneous, | It implies $$f'_{i_0}\in \sum_{j\neq i_0}\mxthcal{O}f'_j+\mathcal{O}f,$$ whirh is ijpossibld in the case of an isolated sungulqrity $f$.
If $E=1$, considerine the leafing tern in $\var{s}$ in thx relation (\[eq3.3\]), sc obtcii that $1$ belongs to the mafimal ideal of $W$, a contradiction.$\square$
*Remark :* After tre precrdlng lemma, we hwve $\nqta_1>n+1.$ Since in the smooth case, $\beta_1=n+1$, we concluve that $\beta_1$ chsracterizes the smoothness (amojg germs for which $0$ is at mosj an usolated sineularity).
We are now reasy to characterize the quasi-homugenenty. If $f$ is qyasl homogeneouw, the good operator has dagree $1$ on $s$, is of the focm $s-\rheta$ with $\theta(f)=f$. Frmm the resolution jf $\textrm{tiyr}N$ we constructed in Swctiot 3.1, wa deayce thzt $\tsxtrm{ajn}_W \delta$ is jinimally gwnerated by $f,(f'_i\partoaj_n),(g'_i\partial_{x_j}-f'_n\partiwl_{v_i}),s-\theta.$ The corresponding shifts for tve riltration $F$ are respecrively $0,1,1,1$.
If now $f$ is noj quasi hoiogeneous, then $S_L$ has degree in $s$ greater than $1$. Itdeed, rupkowe on rhf contrary that there exists a good operator whinikating $f^s$ of tme form $s+P$ with $P=u+\yhfts$, $u\in\mathcal{O}$ xnd $\thzfa$ a vector field. We have $$0=(s+K)f^s=(s(1+\thwta(f)\frac{1}{f})+t)f^s.$$ Xubstituting $s=0$, we get $u=0$. So $\rheta(f)\frac{1}{f}=-1$ cnd $f\in J(f)$. This implizs that $f$ is quaxi hokogeneous by [@saito71].
\[prop20\] $y$ is qhasi homogejeous if zvd only if $\textro{ren}_F T=0$.
#### Proof
If $f$ is not quasi hjmogeneouw, | It implies $$f'_{i_0}\in \sum_{j\neq i_0}\mathcal{O}f'_j+\mathcal{O}f,$$ which is the of an singularity $f$. If in in the relation we obtain that belongs to the maximal ideal of a contradiction.$\square$ *Remark :* After the preceding lemma, we have $\beta_1>n+1.$ Since in smooth case, $\beta_1=n+1$, we conclude that $\beta_1$ characterizes the smoothness (among germs for $0$ at an singularity). We are now ready to characterize the quasi-homogeneity. If $f$ is quasi homogeneous, the good has degree $1$ in $s$, is of the $s-\theta$ with $\theta(f)=f$. From resolution of $\textrm{bigr}N$ we constructed Section we deduce $\textrm{ann}_W is generated by $f,(f'_i\partial_t),(f'_i\partial_{x_j}-f'_j\partial_{x_i}),s-\theta.$ corresponding shifts for the filtration $F$ are respectively $0,1,1,1$. If now $f$ is not quasi homogeneous, then has degree greater than Indeed, on contrary that there good operator anihilating $f^s$ of the $P=u+\theta$, $u\in\mathcal{O}$ and $\theta$ a vector field. We $$0=(s+P)f^s=(s(1+\theta(f)\frac{1}{f})+u)f^s.$$ Substituting we get $u=0$. So $\theta(f)\frac{1}{f}=-1$ and J(f)$. This implies that $f$ is quasi homogeneous [@saito71]. \[prop20\] $f$ is quasi homogeneous if and only if $\textrm{reg}_F N=0$. #### Proof If not quasi homogeneous, | It implies $$f'_{i_0}\in \sum_{j\neq i_0}\mathCal{O}f'_j+\mathCal{O}f,$$ WhiCh iS iMposSiblE in the case of an ISolaTed singularity $f$.
If $E=1$, consIderiNg THe leADiNg terM in $\bar{s}$ IN tHE RelAtIoN (\[eq3.3\]), We OBtAin thAt $1$ bElongs tO the maximaL idEaL of $W$, a contradICtIon.$\square$
*REmaRk :* After the prEceDing leMmA, we HAve $\beTa_1>n+1.$ since In the sMOoth caSe, $\beta_1=n+1$, we CoNClude tHAt $\beta_1$ cHARaCterIzes the smoothness (AMoNG germs for which $0$ Is at moSt AN iSOLatEd sIngularity).
we Are noW Ready to CHaRACTerIZe the quasi-homOgeneity. If $f$ IS quAsi homOgEneOUs, the gOod opErATor Has degree $1$ in $S$, is oF the form $s-\Theta$ wITh $\theta(F)=F$. From thE resolUtiOn oF $\texTRm{BiGr}N$ We COnsTRuCteD In SEction 3.1, we DeDuCe thaT $\texTRM{ANn}_W \dEltA$ is mInimaLly generated bY $f,(f'_I\parTIal_T),(f'_i\paRtial_{X_j}-f'_j\PaRtial_{X_i}),s-\theTa.$ The CoRresponding shifTs foR the filtrAtiOn $f$ arE rEspecTIvely $0,1,1,1$.
IF noW $f$ iS not quaSi homogENeoUs, THEN $S_l$ has degree in $s$ greatEr THAn $1$. indeed, suPpose oN ThE cONtrary thAt TheRe exISTs a goOd opERaTor anihiLating $F^S$ oF tHe form $s+p$ wIth $P=u+\tHeTa$, $u\In\mAthcaL{o}$ and $\Theta$ a Vector fiEld. We HAve $$0=(s+P)f^s=(s(1+\theta(f)\FRac{1}{f})+u)f^s.$$ SubstiTUtING $s=0$, WE get $U=0$. So $\Theta(f)\frac{1}{f}=-1$ And $f\IN J(f)$. THis iMPlIes THat $f$ iS quasI hOMoGEneous by [@saito71].
\[prop20\] $f$ iS qUasi hoMogenEous if and only If $\textrm{reG}_f n=0$.
#### proof
If $f$ Is noT QuASi homogeneous, | It implies $$f'_{i_0}\in\sum_{j\ne q i_0 }\m ath ca l{O} f'_j +\mathcal{O}f, $ $ wh ich is impossible in t he ca se of a n i solat ed sing u la r i ty$f $.
I f$ E= 1$, c ons idering the leadi ngte rm in $\bar{ s }$ in the re lat ion (\[eq3.3 \]) , we o bt ain that$1$ belo ngs to the ma ximal ide al of $W$ , a cont r a di ctio n.$\square$
*Rem a rk :* After the p recedi ng le m m a,wehave $\bet a_ 1>n+1 . $ Since in t h e s m ooth case, $\ beta_1=n+1$ , we concl ud e t h at $\b eta_1 $c har acterizes t he s moothness (amon g germsf or whic h $0$isatmost an i sol at e d s i ng ula r ity ).
We a re n ow re adyt o c hara cte rize thequasi-homogen eit y. I f $f $ isquasi hom og eneou s, the good o perator has deg ree$1$ in $s $,is of t he fo r m $s-\ the ta$ with $ \theta( f )=f $. F r om the resolution of $ \ t ex trm{bigr }N$ we co ns t ructed i nSec tion 3 .1, w e de d uc e that $ \textr m {a nn }_W \de lt a$ ismi nim all y gen e rate d by $ f,(f'_i\ parti a l_t),(f'_i\par t ial_{x_j}-f'_ j \p a r ti a l_{x _i} ),s-\theta. $ Th e cor resp o nd ing shift s for t h ef iltration $F$ are r es pectiv ely $ 0,1,1,1$.
If now $f$ i s n ot quasi hom o ge n eous, then $S_ L$ ha s degree i n $s$ gre aterthan $1$ . Indeed, s uppose o n t hecon tra r y t hat there exi s t s ago od oper ato r anihi lat ing $f ^s$ o f the for m $s+P$wi th $ P= u+\ theta $ , $u\in\ ma thc al {O} $ and $\thet a$ avect or f i eld . We ha v e$ $ 0=(s +P )f ^s=( s(1 +\ theta (f)\ f rac {1}{f}) +u)f^s.$$ Su b stit ut in g $s=0$ , we get $u=0 $. So $\thet a( f)\ frac{1 } { f}=-1$ a nd $f\in J(f)$. This im p lies th at$f$ i s qu asi homog ene ous by [@ s aito71 ].
\[ prop2 0\ ] $ f $ is q u a si ho mo geneous if a ndonlyif $\t extrm{r eg}_F N=0$.
####P roo f
If $f$ isnot qua s i h omo g en e ous , | It_implies $$f'_{i_0}\in_\sum_{j\neq i_0}\mathcal{O}f'_j+\mathcal{O}f,$$ which is_impossible in_the_case of_an_isolated singularity $f$.
If_$E=1$, considering the_leading term in $\bar{s}$_in the relation_(\[eq3.3\]),_we obtain that $1$ belongs to the maximal ideal of $W$, a contradiction.$\square$
*Remark :*_After_the preceding_lemma,_we_have $\beta_1>n+1.$ Since in the_smooth case, $\beta_1=n+1$, we conclude_that $\beta_1$_characterizes the smoothness (among germs for which $0$_is_at most an_isolated singularity).
We are now ready to characterize the quasi-homogeneity._If $f$ is quasi homogeneous, the_good operator has_degree_$1$_in $s$, is of_the form $s-\theta$ with $\theta(f)=f$. From_the resolution of $\textrm{bigr}N$ we constructed_in Section 3.1, we deduce that $\textrm{ann}_W_\delta$ is minimally generated by $f,(f'_i\partial_t),(f'_i\partial_{x_j}-f'_j\partial_{x_i}),s-\theta.$_The corresponding shifts for the_filtration $F$_are respectively $0,1,1,1$.
If now $f$_is not quasi_homogeneous, then_$S_L$ has degree_in $s$ greater than $1$. Indeed,_suppose on the_contrary that there exists a good_operator_anihilating $f^s$ of_the_form_$s+P$ with_$P=u+\theta$, $u\in\mathcal{O}$ and_$\theta$_a vector_field._We have $$0=(s+P)f^s=(s(1+\theta(f)\frac{1}{f})+u)f^s.$$ Substituting $s=0$, we_get_$u=0$. So $\theta(f)\frac{1}{f}=-1$ and $f\in J(f)$. This_implies that $f$ is_quasi_homogeneous by [@saito71].
\[prop20\] $f$_is quasi homogeneous if and_only if $\textrm{reg}_F N=0$.
#### Proof
If $f$_is not_quasi homogeneous, |
{p}}}$ belongs to $(\mathcal{V}_{\operatorname*{\mathfrak{p}}})_n \subseteq \mathcal{V}_n$, again using Lemma \[L:local\] for the last inclusion. The natural map $\iota: M \rightarrow \prod_{\operatorname*{\mathfrak{p}}\in \operatorname*{Spec}(R)}M_{\operatorname*{\mathfrak{p}}}$ is a monomorphism. Since $R \rightarrow R_{\operatorname*{\mathfrak{p}}}$ is a flat ring epimorphism, the injective envelope $E(M_{\operatorname*{\mathfrak{p}}})$ in $\operatorname*{Mod-R_{\operatorname*{\mathfrak{p}}}}$ is an injective $R$-module. Therefore, we can use the injectivity to extend $\iota$ to a map $\varphi: E \rightarrow \prod_{\operatorname*{\mathfrak{p}}\in \operatorname*{Spec}(R)} E(M_{\operatorname*{\mathfrak{p}}})$. As $\varphi$ extends $\iota$, and $M$ is essential in $E$, it follows that $\varphi$ is a monomorphism. Therefore, $E$ is a direct summand in $\prod_{\operatorname*{\mathfrak{p}}\in \operatorname*{Spec}(R)} E(M_{\operatorname*{\mathfrak{p}}})$. But $\prod_{\operatorname*{\mathfrak{p}}\in \operatorname*{Spec}(R)} E(M_{\operatorname*{\mathfrak{p}}}) \in \mathcal{V}_n$, and thus $E \in \mathcal{V}_n$.
The converse implication follows again from [@BS Theorem 7.2].
Homological ring epimorphisms versus density {#S:nondense}
============================================
Let $R$ be a valuation domain and $\mathbb{X}=(\mathcal{X}_n \mid n \in \mathbb{Z})$ an admissible filtration in $\operatorname*{Spec}(R)$. We call such a sequence [non-dense]{} if the admissible system $\mathcal{X}_n$ is non-dense for all $n \in \mathbb{Z}$. Note that $\mathbb{X}$ is a non-dense admissible filtration if and only if it is just a nested sequence of non-dense admissible systems.
The aim is to show that the coaisles corresponding to non-dense admissible filtrations via Theorem \[T02\] are precisely those arising from a chain of homological epimorphisms via Proposition \[P:construction\]. The | { p}}}$ belongs to $ (\mathcal{V}_{\operatorname*{\mathfrak{p}}})_n \subseteq \mathcal{V}_n$, again using Lemma \[L: local\ ] for the last inclusion. The lifelike function $ \iota: M \rightarrow \prod_{\operatorname*{\mathfrak{p}}\in \operatorname*{Spec}(R)}M_{\operatorname*{\mathfrak{p}}}$ is a monomorphism. Since $ roentgen \rightarrow R_{\operatorname*{\mathfrak{p}}}$ is a flat ring epimorphism, the injective envelope $ E(M_{\operatorname*{\mathfrak{p}}})$ in $ \operatorname*{Mod - R_{\operatorname*{\mathfrak{p}}}}$ is an injective $ R$-module. consequently, we can use the injectivity to extend $ \iota$ to a map $ \varphi: E \rightarrow \prod_{\operatorname*{\mathfrak{p}}\in \operatorname*{Spec}(R) } E(M_{\operatorname*{\mathfrak{p}}})$. As $ \varphi$ widen $ \iota$, and $ M$ is essential in $ E$, it follows that $ \varphi$ is a monomorphism. Therefore, $ E$ is a direct summand in $ \prod_{\operatorname*{\mathfrak{p}}\in \operatorname*{Spec}(R) } E(M_{\operatorname*{\mathfrak{p}}})$. But $ \prod_{\operatorname*{\mathfrak{p}}\in \operatorname*{Spec}(R) } E(M_{\operatorname*{\mathfrak{p } } }) \in \mathcal{V}_n$, and thus $ east \in \mathcal{V}_n$.
The converse implication follows again from [ @BS Theorem 7.2 ].
homologic ring epimorphisms versus density { # south: nondense }
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Let $ R$ be a valuation sphere and $ \mathbb{X}=(\mathcal{X}_n \mid n \in \mathbb{Z})$ an admissible filtration in $ \operatorname*{Spec}(R)$. We call such a sequence [ non - dense ] { } if the admissible system $ \mathcal{X}_n$ is non - dense for all $ normality \in \mathbb{Z}$. Note that $ \mathbb{X}$ is a non - dense admissible filtration if and only if it is just a nested succession of non - dense admissible systems.
The bearing is to show that the coaisles corresponding to non - dense admissible filtrations via Theorem \[T02\ ] are precisely those get up from a chain of homological epimorphisms via Proposition \[P: construction\ ]. The | {p}}}$ bflongs to $(\mathcal{V}_{\operaturname*{\mathfrak{p}}})_u \subseveq \matgcal{V}_n$, aeain using Lemma \[L:local\] for tie lqst ibclusion. The natural mxp $\iota: M \rightareow \krod_{\operatorname*{\mefhfrak{p}}\lu \opedwtoruane*{Spec}(R)}M_{\operatprname*{\mathxrak{p}}}$ is a monmmur'hism. Since $R \rightarrow R_{\operatornaie*{\mathftan{p}}}$ is a flat ryng tpiiorpgpsn, the injective envelope $E(M_{\opsratorneme*{\mathfrak{p}}})$ in $\operatorname*{Mod-R_{\operatornwme*{\mwthfrak{p}}}}$ is an injfctive $R$-modome. Eyerefore, we zan use tht nnjectivity to extend $\iota$ to a map $\varphi: E \riyhtarrow \prid_{\ipegdtorname*{\matifrak{p}}\pn \operatorname*{Spec}(R)} A(M_{\operayorname*{\mathfral{p}}})$. Es $\vqrphi$ extends $\iota$, anv $M$ is essential in $G$, it follofs that $\varphi$ is a minomotphisk. Thdeefure, $E$ ia a digecv summand ih $\prod_{\operarorname*{\mathfrak{p}}\in \ppqgstorname*{Spec}(D)} E(M_{\opqrwtorname*{\mathfrak{p}}})$. But $\prod_{\operatorname*{\mdthrrak{p}}\in \operatorname*{Spex}(R)} E(M_{\operatorname*{\mathvrak{p}}}) \in \iathcal{V}_n$, and thus $E \in \mathcal{V}_n$.
The converse impnicatmov filjuqs again from [@BS Theorem 7.2].
Homological ring epimowlhosks versus denslty {#S:nondense}
============================================
Let $R$ bf s valuation dooain aus $\jathbb{X}=(\mathcal{X}_n \mld n \in \mathvb{Z})$ an adiissoble filtration in $\operatorbame*{Spec}(R)$. We xall such a sequenee [non-dense]{} nf the admixsible system $\mathcal{X}_n$ is hon-dense fog all $n \ih \mathbb{Z}$. Note thxt $\kadhbb{X}$ is a non-dense admissyble filtcatiou if and onli if it is just a nestcg sequence of non-dfnse cdmisvible systfms.
The aim is to show that the rpaisles corrgspmndpng to nou-dense admissible fyltrations via Theorei \[T02\] ard preciselj those acising from w chain of hokllogical epikorphismf viq Pripositiuv \[P:construction\]. The | {p}}}$ belongs to $(\mathcal{V}_{\operatorname*{\mathfrak{p}}})_n \subseteq \mathcal{V}_n$, again \[L:local\] the last The natural map is monomorphism. Since $R R_{\operatorname*{\mathfrak{p}}}$ is a ring epimorphism, the injective envelope $E(M_{\operatorname*{\mathfrak{p}}})$ $\operatorname*{Mod-R_{\operatorname*{\mathfrak{p}}}}$ is an injective $R$-module. Therefore, we can use the injectivity to extend to a map $\varphi: E \rightarrow \prod_{\operatorname*{\mathfrak{p}}\in \operatorname*{Spec}(R)} E(M_{\operatorname*{\mathfrak{p}}})$. As $\varphi$ extends $\iota$, $M$ essential $E$, follows that $\varphi$ is a monomorphism. Therefore, $E$ is a direct summand in $\prod_{\operatorname*{\mathfrak{p}}\in \operatorname*{Spec}(R)} E(M_{\operatorname*{\mathfrak{p}}})$. $\prod_{\operatorname*{\mathfrak{p}}\in \operatorname*{Spec}(R)} E(M_{\operatorname*{\mathfrak{p}}}) \in \mathcal{V}_n$, and thus $E \mathcal{V}_n$. The converse implication again from [@BS Theorem 7.2]. ring versus density ============================================ $R$ a valuation domain $\mathbb{X}=(\mathcal{X}_n \mid n \in \mathbb{Z})$ an admissible filtration in $\operatorname*{Spec}(R)$. We call such a sequence [non-dense]{} if admissible system non-dense for $n \mathbb{Z}$. that $\mathbb{X}$ is admissible filtration if and only if a nested sequence of non-dense admissible systems. The is to that the coaisles corresponding to non-dense filtrations via Theorem \[T02\] are precisely those arising a chain of homological epimorphisms via Proposition \[P:construction\]. The | {p}}}$ belongs to $(\mathcal{V}_{\operatoRname*{\mathfRak{p}}})_n \SubSetEq \MathCal{V}_N$, again using LemMA \[L:loCal\] for the last inclusion. the naTuRAl maP $\IoTa: M \riGhtarroW \PrOD_{\OpeRaToRnaMe*{\MAtHfrak{P}}\in \OperatoRname*{Spec}(R)}m_{\opErAtorname*{\mathFRaK{p}}}$ is a monomOrpHism. Since $R \riGhtArrow R_{\OpEraTOrnamE*{\maThfraK{p}}}$ is a fLAt ring EpimorphiSm, THe injeCTive envELOpE $E(M_{\oPeratorname*{\mathfrAK{p}}})$ IN $\operatorname*{MOd-R_{\opeRaTOrNAMe*{\mAthFrak{p}}}}$ is an iNjEctivE $r$-module. tHeREFOre, WE can use the injEctivity to eXTenD $\iota$ tO a Map $\VArphi: E \RightArROw \pRod_{\operatorName*{\Mathfrak{p}}\In \operATorname*{sPec}(R)} E(M_{\oPeratoRnaMe*{\mAthfRAk{P}}})$. AS $\vaRpHI$ exTEnDs $\iOTa$, aNd $M$ is essEnTiAl in $E$, It foLLOWS thaT $\vaRphi$ Is a moNomorphism. TheRefOre, $E$ IS a dIrect SummaNd in $\PrOd_{\opeRatornAme*{\maThFrak{p}}\in \operatorName*{spec}(R)} E(M_{\opEraToRnaMe*{\MathfRAk{p}}})$. But $\ProD_{\opEratornAme*{\mathFRak{P}}\iN \OPErAtorname*{Spec}(R)} E(M_{\opeRaTORnAme*{\mathfRak{p}}}) \in \MAtHcAL{V}_n$, and thUs $e \in \MathCAL{V}_n$.
ThE conVErSe implicAtion fOLlOwS again fRoM [@BS TheOrEm 7.2].
HOmoLogicAL rinG epimoRphisms vErsus DEnsity {#S:nondensE}
============================================
let $R$ be a valuatIOn DOMaIN and $\MatHbb{X}=(\mathcal{x}_n \miD N \in \mAthbB{z})$ aN adMIssibLe filTrATiON in $\operatorname*{Spec}(r)$. WE call sUch a sEquence [non-denSe]{} if the admISSIble systEm $\maTHcAL{X}_n$ is non-dense fOr all $N \in \mathbb{Z}$. nOte that $\mAthbb{x}$ is a non-dEnse admisSIBle filtrAtiOn iF anD onLY If It is just a nestED SequEnCe of non-DenSe admisSibLe sYstEms.
thE aim is to sHow that tHe CoAiSlEs cOrresPOnding to NoN-deNsE adMissiBLe filtRatioNs viA THeORem \[t02\] are preCIsELY thoSe ArIsinG frOm A chaiN of hOMolOgical ePimorphisMs vIA ProPoSiTion \[P:coNstruction\]. The | {p}}}$ belongs to $(\mathc al{V}_{\op erato rna me* {\ math frak {p}}})_n \subs e teq\mathcal{V}_n$, againusing L e mma\ [L :loca l\] for th e las tin clu si o n. Thenat ural ma p $\iota:M \ ri ghtarrow \pr o d_ {\operator nam e*{\mathfrak {p} }\in \ op era t ornam e*{ Spec} (R)}M_ { \opera torname*{ \m a thfrak { p}}}$ i s amono morphism. Since $ R \ r ightarrow R_{\ operat or n am e * {\m ath frak{p}}}$ i s a f l at ring ep i m o rph i sm, the injec tive envelo p e $ E(M_{\ op era t orname *{\ma th f rak {p}}})$ in$\op eratornam e*{Mod - R_{\ope r atornam e*{\ma thf rak {p}} } }$ i s a ni nje c ti ve$ R$- module.Th er efore , we c a n use th e in jecti vity to exten d $ \iot a $ t o a m ap $\ varp hi : E \ righta rrow\p rod_{\operatorn ame* {\mathfra k{p }} \in \ opera t orname *{S pec }(R)} E (M_{\op e rat or n a m e* {\mathfrak{p}}})$. A s $\ varphi$extend s $ \i o ta$, and $ M$is e s s entia l in $E $, it fo llowst ha t$\varph i$ is amo nom orp hism. Ther efore, $E$ isa dir e ct summand in$ \prod_{\opera t or n a me * {\ma thf rak{p}}\in\ope r ator name * {S pec } (R)}E(M_{ \o p er a torname*{\mathfrak{ p} }})$.But $ \prod_{\opera torname*{\ m a t hfrak{p} }\in \o p eratorname*{Sp ec}(R )} E(M_{\o p eratorna me*{\ mathfrak {p}}}) \i n \mathcal {V} _n$ , a ndt h us $E \in \math c a l{V} _n $.
The co nverseimp lic ati onfo llows aga in from[@ BS T he ore m 7.2 ] .
Homol og ica lrin g epi m orphis ms ve rsus d en s ity {#S:no n de n s e}
= == == ==== === == ===== ==== = === ======= ========= =
L et $ R$ b e a val uation domain a nd $\mathb b{ X}= (\math c a l{X}_n \ mid n \in \mathbb{Z})$a n admis sib le fi ltra tion in $ \op erator nam e *{Spec }(R)$. We c al l s u c h a s e q ue nce [ non-dense] { } if thead miss ible sy stem $\mathcal{X}_ n $ i s non-dense f orall$ n \ in\ ma t hbb {Z } $.N o te that $\mathb b{X}$ is a n o n- dense admi s sib le filtra tion if ando nly ifit is jus t a neste dsequ e n ceof non-den se admis sible sys t ems.Th e aim is to sh ow th at th e coai s les corr espond in g to n on-de ns e admiss ible filtrations via Th eorem\[T02 \]are preci sel y th ose arisi ng f rom a chai n o f h omolo gic a l epi morp h is msv ia Pr opos i tion \[P: c on str u c ti on\]. The | {p}}}$ belongs_to $(\mathcal{V}_{\operatorname*{\mathfrak{p}}})_n_\subseteq \mathcal{V}_n$, again using_Lemma \[L:local\] for_the_last inclusion._The_natural map $\iota:_M \rightarrow \prod_{\operatorname*{\mathfrak{p}}\in_\operatorname*{Spec}(R)}M_{\operatorname*{\mathfrak{p}}}$ is a monomorphism._Since $R \rightarrow_R_{\operatorname*{\mathfrak{p}}}$_is a flat ring epimorphism, the injective envelope $E(M_{\operatorname*{\mathfrak{p}}})$ in $\operatorname*{Mod-R_{\operatorname*{\mathfrak{p}}}}$ is an injective_$R$-module._Therefore, we_can_use_the injectivity to extend $\iota$_to a map $\varphi: E_\rightarrow \prod_{\operatorname*{\mathfrak{p}}\in_\operatorname*{Spec}(R)} E(M_{\operatorname*{\mathfrak{p}}})$. As $\varphi$ extends $\iota$, and $M$_is_essential in $E$,_it follows that $\varphi$ is a monomorphism. Therefore, $E$_is a direct summand in $\prod_{\operatorname*{\mathfrak{p}}\in_\operatorname*{Spec}(R)} E(M_{\operatorname*{\mathfrak{p}}})$. But_$\prod_{\operatorname*{\mathfrak{p}}\in_\operatorname*{Spec}(R)}_E(M_{\operatorname*{\mathfrak{p}}}) \in \mathcal{V}_n$, and_thus $E \in \mathcal{V}_n$.
The converse implication_follows again from [@BS Theorem 7.2].
Homological_ring epimorphisms versus density {#S:nondense}
============================================
Let $R$ be_a valuation domain and $\mathbb{X}=(\mathcal{X}_n \mid_n \in \mathbb{Z})$ an admissible_filtration in_$\operatorname*{Spec}(R)$. We call such a_sequence [non-dense]{} if_the admissible_system $\mathcal{X}_n$ is_non-dense for all $n \in \mathbb{Z}$._Note that $\mathbb{X}$_is a non-dense admissible filtration if_and_only if it_is_just_a nested_sequence of non-dense_admissible_systems.
The aim_is_to show that the coaisles corresponding_to_non-dense admissible filtrations via Theorem \[T02\] are precisely_those arising from a_chain_of homological epimorphisms via_Proposition \[P:construction\]. The |
After reconstructing the pre-clusters in the four projections, the final (full) cluster properties are determined by adding the pre-cluster energies and averaging the coordinates of the pre-cluster centres.
In case of more than one cluster per projection the situation is more difficult, as illustrated in figure \[fig:twocl\]:
![Projected views of a two-cluster event.[]{data-label="fig:twocl"}](VLQfig13.eps){width="14cm"}
Two particles hitting the calorimeter lead to eight pre-clusters from which four clusters can be reconstructed so that there is no unique solution. If however the two incident particles deposit clearly different amounts of energy in the calorimeter, combining the low energy pre-clusters versus the high energy ones allows a non-ambiguous solution. This method will of course fail if the particles have similar energies. In this case no unique solution is possible, and the decision needs to be left over to the VLQ track detector [@Hurling].
The Calibration Concept
=======================
Test Beam Calibration
---------------------
Both VLQ calorimeter modules were tested in an electron test beam at the Deutsches Elektronen - Synchrotron DESY in Hamburg. The beam, supplied by the DESY synchrotron, had energies adjustable between 1 and 6 GeV.
In the test setup, the VLQ module to be investigated was fixed at the moving mechanism constructed for operation in H1. As this device supplies only vertical movement, horizontal displacements were performed on a support table driven by a step motor. The vertical and horizontal displacement steps were 5 mm each, corresponding to the scintillator strip width.
The distance between beam collimator and calorimeter module was 5.70 m. The position of the beam in a vertical plane in front of the calorimeter was defined with the help of a silicon tracker telescope. The telescope consisted of two densely packed bundles, each equipped with four Si strip detector planes. The strip orientations were y - x - y - x - x - y - x - y. Each plane contained 384 Si strips of 50 $\mu$m width. The distance between the layers inside one bundle was 10 mm, the two bundles were 17 cm apart from each other. The cross section covered by the telescope was 2 x 2 cm$^2$. The Si - telecope was read out sequentially by a 6 - bit CAMAC ADC. It was sandwiched between two scint | After reconstructing the pre - clusters in the four projections, the final (wide) bunch properties are determined by lend the pre - cluster energy and averaging the coordinates of the pre - cluster center.
In case of more than one cluster per project the situation is more difficult, as illustrate in figure \[fig: twocl\ ]:
! [ Projected view of a two - cluster event.[]{data - label="fig: twocl"}](VLQfig13.eps){width="14 cm " }
Two particles hitting the calorimeter lead to eight pre - bunch from which four clusters can be reconstructed so that there is no singular solution. If however the two incident particles deposit intelligibly different sum of energy in the calorimeter, combining the low energy pre - clusters versus the eminent energy ones allows a non - ambiguous solution. This method will of course fail if the particles have similar energies. In this case no unique solution is possible, and the decision needs to be left over to the VLQ track detector [ @Hurling ].
The Calibration Concept
= = = = = = = = = = = = = = = = = = = = = = =
Test Beam Calibration
---------------------
Both VLQ calorimeter faculty were tested in an electron test balance beam at the Deutsches Elektronen - Synchrotron DESY in Hamburg. The radio beam, supplied by the DESY synchrotron, had energies adjustable between 1 and 6 GeV.
In the trial setup, the VLQ module to be investigated was fixed at the move mechanism constructed for operation in H1. As this device supplies only vertical movement, horizontal displacement were performed on a support table driven by a dance step motor. The vertical and horizontal supplanting steps were 5 mm each, corresponding to the scintillator strip width.
The distance between beam collimator and calorimeter module was 5.70 m. The position of the radio beam in a vertical plane in front of the calorimeter was define with the help of a silicon tracker telescope. The telescope consisted of two densely packed bundle, each equipped with four Si strip detector airplane. The strip orientation were y - x - y - x - x - y - x - y. Each plane contain 384 Si strips of 50 $ \mu$m width. The distance between the level inside one bundle was 10 mm, the two bundles were 17 curium apart from each early. The cross section covered by the telescope was 2 x 2 cm$^2$. The Si - telecope was learn out sequentially by a 6 - bit CAMAC ADC. It was sandwiched between two scint | Aftfr reconstructing the prt-clusters in the yiur prmjectikns, the winal (full) cluster propertied qre dtnermined by adding thd pre-clusner energues end averaging thx coordikctes kn the 're-cluster centtes.
In case ox more than ona zlbster per projection the situation if more civficult, as illostraued in rpgmre \[fig:twocl\]:
![Projected views of a two-cluvter event.[]{dats-label="fig:twocl"}](VLQfig13.eps){widhh="14cm"}
Hwo particles hittlng the calirimqrer lead to dight pre-clusters from which four clusters can be recunstrbcted so thqt thfte is no uniwue sjlution. If however tve two oncident partigles vepowit clearly different amounts of energy in the cdlkrimeter, combinint rhe lmw etergh prd-clhsvera versks vhe high ensrgy ones aolows a non-ambiguoux filution. This jethod wyll of course fail if the particles havt simjlar energies. In this cqse no unique solutioj is possyble, and the decision needs to be left over to tha VLQ graej detdxtlr [@Hurling].
The Calibration Concept
=======================
Test Beam Cajjbtanion
---------------------
Both VLQ calovimeter modules wete trfted in an elgctron cssf beam at the Deutdches Ejektrinen - Synshroyron DESY in Hamburg. The beqm, supplied yy rhe DESY synchrotrln, had eneryies acjustsble between 1 and 6 GeV.
Iu the fest setup, hhe VLQ mkaule to be invesgigstad was fixed at the moving mechanisn coustructea fot operaeion in H1. Ws thlv device supplies lnly rertiwal movemejt, horizontal displacements werx performed om d slpport tayle drlven by a step iotor. The vertncal and horixontal disilacement steps were 5 mm each, corsgsponding to vhe scintyllaror wtrip wkath.
The distancr between beam collumator and calorimcter okdule was 5.70 m. The powition of the bram in a vxrticwn plane in fsont of yhe cxlorimeuzr car degined with the help mf a silicon tracker trlcscope. Thg telesco[e consisted pf two densely pacned bnndles, each eqoipped with four Si strip detecfor planed. Tme strip oriegtatlons were y - x - y - x - x - y - x - y. Each plane contained 384 Si strips of 50 $\mo$m width. The distance netween the kayerf inside mne bundle was 10 mm, tye two bundles weve 17 cm apart from each other. The fross section covered by the telescope was 2 x 2 cm$^2$. The Si - telecope was reqd out sequentiallg by a 6 - tic CCMAC ADS. It xas sandwiched benween two scint | After reconstructing the pre-clusters in the four final cluster properties determined by adding the of the pre-cluster In case of than one cluster per projection the is more difficult, as illustrated in figure \[fig:twocl\]: ![Projected views of a two-cluster Two particles hitting the calorimeter lead to eight pre-clusters from which four clusters be so there no unique solution. If however the two incident particles deposit clearly different amounts of energy in calorimeter, combining the low energy pre-clusters versus the energy ones allows a solution. This method will of fail the particles similar In case no unique is possible, and the decision needs to be left over to the VLQ track detector [@Hurling]. The Concept ======================= Calibration --------------------- VLQ modules tested in an beam at the Deutsches Elektronen - Hamburg. The beam, supplied by the DESY synchrotron, energies adjustable 1 and 6 GeV. In the setup, the VLQ module to be investigated was at the moving mechanism constructed for operation in H1. As this device supplies only vertical displacements were performed on support table driven a motor. vertical horizontal displacement were 5 mm each, corresponding to the scintillator strip width. The between beam collimator and calorimeter module was 5.70 m. The the in a vertical in front of the was with the help of tracker The two packed each equipped with four strip detector planes. The strip were y - x x - y - x - y. Each contained 384 Si strips of 50 $\mu$m The distance between the layers inside one bundle was 10 mm, the bundles were apart from each other. The cross section covered the telescope was 2 2 cm$^2$. The Si - telecope was read out by 6 - CAMAC ADC. It sandwiched between two | After reconstructing the pre-Clusters in The foUr pRojEcTionS, the Final (full) clustER proPerties are determined by AddinG tHE pre-CLuSter eNergies ANd AVEraGiNg The CoORdInateS of The pre-cLuster centRes.
in Case of more thAN oNe cluster pEr pRojection the SitUation Is MorE DiffiCulT, as ilLustraTEd in fiGure \[fig:twOcL\]:
![projecTEd views OF A tWo-clUster event.[]{data-labEL="fIG:twocl"}](VLQfig13.epS){width="14Cm"}
tWo PARtiCleS hitting thE cAloriMEter leaD To EIGHt pRE-clusters from Which four clUSteRs can bE rEcoNStructEd so tHaT TheRe is no uniquE solUtion. If hoWever tHE two incIDent parTicles DepOsiT cleARlY dIffErENt aMOuNts OF enErgy in thE cAlOrimeTer, cOMBINing The Low eNergy Pre-clusters veRsuS the HIgh EnergY ones AlloWs A non-aMbiguoUs solUtIon. This method wiLl of Course faiL if ThE paRtIcles HAve simIlaR enErgies. IN this caSE no UnIQUE sOlution is possible, aNd THE dEcision nEeds to BE lEfT Over to thE VlQ tRack DETectoR [@HurLInG].
The CaliBratioN coNcEpt
=======================
Test beAm CaliBrAtiOn
---------------------
BOth VLq CaloRimeteR modules Were tESted in an electrON test beam at thE deUTScHEs ElEktRonen - SynchrOtroN dESY In HaMBuRg. THE beam, SupplIeD By THe DESY synchrotron, haD eNergieS adjuStable between 1 And 6 GeV.
In thE TESt setup, tHe VLq MoDUle to be investiGated Was fixed at THe moving MechaNism consTructed foR OPeration In H1. as tHis DevICE sUpplies only veRTIcal MoVement, hOriZontal dIspLacEmeNts WeRe performEd on a supPoRt TaBlE drIven bY A step motOr. the VeRtiCal anD HorizoNtal dIsplAcEmENt sTeps werE 5 Mm EACh, coRrEsPondIng To The scIntiLLatOr strip Width.
The dIstANce bEtWeEn beam cOllimator and cAlOrimeter moDuLe wAs 5.70 m. The POSition of The beam in a vertical plane IN front oF thE caloRimeTer was defIneD with tHe hELp of a sIlicon TrackEr TelESCope. THE TeLesCoPe consisteD OF twO densElY pacKed bundLes, each equipped witH FouR Si strip detecTor PlanES. thE stRIp ORieNtATioNS Were y - x - y - x - x - y - x - y. EacH plane contAiNEd 384 si strips of 50 $\MU$m wIdTh. The diStance bEtweeN The layeRs inside oNe bundle wAs 10 Mm, thE TWo bUndles were 17 Cm apart fRom each otHEr. The CRoSs secTioN coverEd By tHe telEscope WAs 2 x 2 Cm$^2$. The si - teleCoPe was rEad ouT sEquentiaLly by a 6 - bit CAMAC ADC. It was sAndwicHed beTweEn two scinT | After reconstructing the p re-cluster s inthe fo ur pro ject ions, the fina l (fu ll) cluster properties arede t ermi n ed by a dding t h ep r e-c lu st eren e rg ies a ndaveragi ng the coo rdi na tes of the p r e- cluster ce ntr es.
In case of moreth ano ne cl ust er pe r proj e ctionthe situa ti o n is m o re diff i c ul t, a s illustrated inf ig u re \[fig:twocl \]:
! [P r oj e c ted vi ews of a t wo -clus t er even t .[ ] { d ata - label="fig:tw ocl"}](VLQf i g13 .eps){ wi dth = "14cm" }
Tw op art icles hitti ng t he calori meterl ead toe ight pr e-clus ter s f romw hi ch fo ur clu s te rsc anbe recon st ru ctedso t h a t ther e i s no uniq ue solution.Ifhowe v erthe t wo in cide nt part iclesdepos it clearly differ entamounts o f e ne rgy i n the calori met er, combin ing the low e n e r gy pre-clusters vers us t he high en ergy o n es a l lows a n on -am bigu o u s sol utio n .This met hod wi l lof course f ail if t hepar ticle s hav e simi lar ener gies. In this case n o unique solut i on i sp ossi ble , and the d ecis i on n eeds to be leftoverto th e VLQ track detector [ @Hurli ng].
The Calibrat ion Concep t = ======== ==== = == = ======
Test B eam C alibration -------- ----- --------
Both VL Q calorime ter mo dul esw e re tested in an e lect ro n testbea m at th e D eut sch esEl ektronen- Synchr ot ro nDE SYin Ha m burg. Th ebea m, su pplie d by th e DES Y sy nc hr o tro n, hade ne r g iesad ju stab lebe tween 1 a n d 6 GeV.
In the te sts etup ,th e VLQ m odule to be i nv estigatedwa s f ixed a t the movi ng mechanism constructe d for op era tionin H 1. As thi s d evicesup p lies o nly ve rtica lmov e m ent,h o ri zon ta l displace m e nts were p erfo rmed on a support table d r ive n by a step m oto r. T h e v ert i ca l an dh ori z o ntal displaceme nt steps w er e 5 mm each,c orr es ponding to the scin t illator strip wi dth.
The d ista n c e b etween bea m collim ator andc alori m et er mo dul e was5. 70m. Th e posi t ion of t he bea min a v ertic al plane i n front of the calorime ter wa s def ine d with th e h e lpof a sili contracker te les cop e. Th e t e lesco pe c o ns ist e d oftwod ensely pa c ke d b u n dl es, each eq u i p ped with fo u r Si s trip detector planes. The strip orie ntat i o nswer e y - x - y - x - x - y-x - y. Eac hplane conta ined 384 S i stri ps of50 $\m u$m wid t h .T he dis tanc e b etween th e l ay e rs insi de o n e bund le w as 10 mm , thet wo b u n dles were 17 cmapart f rom e a chother .The cro s s se ction cove red by thetelesc opewas 2 x 2 cm $^ 2$. Th e S i- telecope was readout s equenti al ly b y a 6 - b it C A M AC AD C. I twas sandwich e d b e tw ee n tw o sc int | After reconstructing_the pre-clusters_in the four projections,_the final_(full)_cluster properties_are_determined by adding_the pre-cluster energies_and averaging the coordinates_of the pre-cluster_centres.
In_case of more than one cluster per projection the situation is more difficult, as_illustrated_in figure_\[fig:twocl\]:
![Projected_views_of a two-cluster event.[]{data-label="fig:twocl"}](VLQfig13.eps){width="14cm"}
Two particles_hitting the calorimeter lead to_eight pre-clusters_from which four clusters can be reconstructed so_that_there is no_unique solution. If however the two incident particles deposit_clearly different amounts of energy in_the calorimeter, combining_the_low_energy pre-clusters versus the_high energy ones allows a non-ambiguous_solution. This method will of course_fail if the particles have similar energies._In this case no unique solution_is possible, and the decision_needs to_be left over to the_VLQ track detector_[@Hurling].
The Calibration_Concept
=======================
Test Beam Calibration
---------------------
Both_VLQ calorimeter modules were tested in_an electron test_beam at the Deutsches Elektronen -_Synchrotron_DESY in Hamburg._The_beam,_supplied by_the DESY synchrotron,_had_energies adjustable_between_1 and 6 GeV.
In the test_setup,_the VLQ module to be investigated was_fixed at the moving_mechanism_constructed for operation in_H1. As this device supplies_only vertical movement, horizontal displacements were_performed on_a support_table driven by a step motor. The vertical and horizontal displacement_steps were 5 mm each, corresponding_to the scintillator strip_width.
The distance_between_beam collimator and_calorimeter_module was_5.70 m. The position of the beam in_a vertical_plane in front of the calorimeter_was defined with the_help_of a silicon tracker telescope. The_telescope consisted of two densely packed_bundles, each equipped with four_Si_strip_detector planes. The strip orientations_were y - x - y_- x -_x - y - x - y._Each_plane contained 384 Si strips of_50_$\mu$m width. The distance between the_layers_inside_one bundle was 10 mm,_the two bundles were 17 cm_apart from each other. The cross section covered by_the telescope was_2 x 2 cm$^2$. The_Si_-_telecope was read out sequentially by a 6 - bit_CAMAC ADC._It was sandwiched_between two scint |
00:Sharing; @Whittaker94:Informal]. This idea was recently further confirmed by Pentland *et al.*, in a study of workplace communication patterns at a Prague bank [@Pentland12:New]. They discovered a key characteristic of successful teams: members periodically interact with others outside of their team, and bring back new information. They dubbed this critical dimension of communication ‘exploration’, the tendency for a team to seek inter-group connections, and found that teams with high exploration tended to be more successful, demonstrating the power of serendipitous interactions in the office.
One factor that could clearly affect the ease of such interactions is the physical spaces of the workplace itself; for example, high-traffic areas such as coffee machines and photocopiers may be particularly likely places for inter-group meetings [@fayard2007photocopiers; @Isaacs96:Piazza]. In general, if spaces encourage the mixing and meeting of a diverse range of people, serendipitous meetings between individuals from different teams or social groups will occur more readily, which could be crucial; face-to-face communication has been shown to be more important than electronic means such as email or SMS [@Pentland12:New; @Stryker12:Facilitating].
Indeed, designing the layout of workspaces in accordance with office social dynamics resulting from the background of employees can help them work more effectively, as was shown by a recent study of offices across 11 different countries [@Steelcase13:Culture]. While this approach takes into account the national culture of the country where an office is situated, it does not consider what happens in very culturally diverse environments that accommodate workers from many different backgrounds.
The effects of cultural variations between countries on organisations have been studied by Geert Hofstede, by means of administering opinion surveys to IBM employees in over 70 countries. He derived five main factors, or cultural dimensions, that can account for most of the variance he observed [@Hofstede10:Cultures], and found that workers’ cultural backgrounds can shape the way they think, feel, and act. Besides the workplace environment itself, cultural differences between its occupants could affect office social dynamics and, therefore, the propensity of workers to engage in beneficial serendipitous inter-group interactions.
Since interactions between people with different areas of expertise or social circles are so beneficial for the exchange of information and ideas, it is important to understand | 00: Sharing; @Whittaker94: Informal ]. This idea was recently further confirmed by Pentland * et al. *, in a discipline of workplace communication design at a Prague bank [ @Pentland12: New ]. They discovered a key feature of successful teams: members periodically interact with others outside of their team, and institute back new information. They dub this critical dimension of communication ‘ exploration ’, the tendency for a team to try inter - group connections, and found that team with eminent exploration tended to be more successful, demonstrating the power of serendipitous interactions in the office.
One component that could clearly affect the ease of such interaction is the physical spaces of the workplace itself; for example, high - dealings areas such as chocolate machines and photocopiers may be particularly probable places for inter - group meetings [ @fayard2007photocopiers; @Isaacs96: Piazza ]. In general, if spaces encourage the mix and merging of a diverse range of people, serendipitous meetings between individuals from different teams or social groups will occur more promptly, which could be all-important; face - to - face communication has been shown to be more authoritative than electronic means such as email or SMS [ @Pentland12: New; @Stryker12: Facilitating ].
Indeed, designing the layout of workspace in accordance with position social dynamics resulting from the setting of employees can help them work more effectively, as was shown by a recent discipline of offices across 11 different countries [ @Steelcase13: Culture ]. While this approach takes into report the national culture of the country where an office is situated, it does not view what happens in very culturally diverse environments that accommodate workers from many different backgrounds.
The effects of cultural version between countries on organisations have been studied by Geert Hofstede, by means of administering public opinion surveys to IBM employee in over 70 countries. He derived five chief factors, or cultural dimensions, that can report for most of the variance he observed [ @Hofstede10: Cultures ], and found that workers ’ cultural backgrounds can shape the way they think, feel, and act. Besides the workplace environment itself, cultural differences between its resident could affect office social moral force and, therefore, the propensity of workers to betroth in beneficial serendipitous inter - group interactions.
Since interactions between people with unlike areas of expertise or social circles are so beneficial for the exchange of information and ideas, it is important to sympathize | 00:Shaging; @Whittaker94:Informal]. Tmis idea was recgnrly fucther cknfirmed by Pentland *et al.*, in a studb of workkjace communication patterns at a Prqgue vank [@Pentlais12:New]. Thcv disdlverzd a key charactgristic of sgccessful teamv: oelbers periodically interact with otrers ouyslde of their tgam, amq brjng back new information. They dubbsd this critical dimrnsion of communication ‘exolorwtion’, the tendency for a team to fwek inter-grojp connectpmns, and foond that teams with high exploragion cended to bg jogg successful, demogstrating thc power of serrndipitous intcractmons in the office.
One facvor that could clearjy affect tke ease of such interqcrions is dhe oyyskcam xpzces ov tie workplacs itself; foe example, high-traffoc qreas such as coffeq iachines and photocopiers may be particllarmy likely places for inrer-group meetings [@fayatd2007photocopyers; @Isaacs96:Piazza]. In general, if spaces encourage dhe mmxkng akd mdwtlng of a diverse range of people, serendipitouf mtetpngs between indiyiduals from diffetejt jeams or sociau grou'a sill occur more rewdily, wrich xould be srucoal; face-to-face communicatiob has been skowb to be more imporcant than elzctronoc mesns such as email or SMR [@Penfland12:New; @Stgyker12:Facimktating].
Indeed, derigkinc the layout of workspaces in accorvance with owficg sociaj dynamics resulting from the backhrounb of amployees fan help them work more effectitxly, as was shpwt bj a recenc studn of offices acwoss 11 differenj countrizs [@Steeucase13:Cultuge]. While vhis approacr takes into dfcount the netional ctltuee od the cujntry where an office is situated, it does not consiver dgat happens in rtry culturally divrrsd egvproimentf that accommmdatd wutkers from many biwferrnt backgrounds.
The exfecfs of cultural varoabions betqeen cougtries on orgsnisations have betn stuvied bb Geery Hjfstede, by means of administerjng opiniln furveys to IFM enployees in pver 70 countries. He derived five main fartors, or cultural dimenwions, that can accoont for most oh the dariance ve observed [@Hofstede10:Cyltures], and found that workers’ culturam backcrounfs can shape the way they think, feel, and act. Besides the workplace envirobment mtfelf, culturzl doffervnczs yetween its uccupants could sffect office social dynamics aid, therefose, the propensity of workers tp dngage in bengficial serendipitous inted-group imteractions.
Since interactions brtween people with diffevent areaw of sxpertpae or social circles are sm benwficial for the exchange of ikformation and ibeas, it is important to understane | 00:Sharing; @Whittaker94:Informal]. This idea was recently further Pentland al.*, in study of workplace bank They discovered a characteristic of successful members periodically interact with others outside their team, and bring back new information. They dubbed this critical dimension of ‘exploration’, the tendency for a team to seek inter-group connections, and found that with exploration to more successful, demonstrating the power of serendipitous interactions in the office. One factor that could clearly the ease of such interactions is the physical of the workplace itself; example, high-traffic areas such as machines photocopiers may particularly places inter-group meetings [@fayard2007photocopiers; In general, if spaces encourage the mixing and meeting of a diverse range of people, serendipitous meetings individuals from or social will more which could be communication has been shown to be electronic means such as email or SMS [@Pentland12:New; Indeed, designing layout of workspaces in accordance with social dynamics resulting from the background of employees help them work more effectively, as was shown by a recent study of offices across countries [@Steelcase13:Culture]. While this takes into account national of country an office situated, it does not consider what happens in very culturally diverse that accommodate workers from many different backgrounds. The effects of between on organisations have studied by Geert Hofstede, means administering opinion surveys to in 70 five factors, cultural dimensions, that can for most of the variance observed [@Hofstede10:Cultures], and found shape the way they think, feel, and act. the workplace environment itself, cultural differences between occupants could affect office social dynamics and, therefore, the propensity of workers engage in inter-group interactions. Since interactions between people with different of expertise or social are so beneficial for the exchange of information and it important to | 00:Sharing; @Whittaker94:Informal]. THis idea was RecenTly FurThEr coNfirMed by Pentland *eT Al.*, in A study of workplace commuNicatIoN PattERnS at a PRague baNK [@PENTlaNd12:neW]. ThEy DIsCoverEd a Key charActeristic Of sUcCessful teams: MEmBers periodIcaLly interact wIth Others OuTsiDE of thEir Team, aNd brinG Back neW informatIoN. they duBBed this CRItIcal Dimension of communICaTIon ‘exploration’, The tenDeNCy FOR a tEam To seek inteR-gRoup cONnectioNS, aND FOunD That teams with High exploraTIon Tended To Be mORe succEssfuL, dEMonStrating the PoweR of serendIpitouS InteracTIons in tHe offiCe.
ONe fActoR ThAt CouLd CLeaRLy AffECt tHe ease of SuCh InterActiONS IS the PhySicaL spacEs of the workplAce ItseLF; foR examPle, hiGh-trAfFic arEas sucH as coFfEe machines and phOtocOpiers may Be pArTicUlArly lIKely plAceS foR inter-gRoup meeTIngS [@fAYARd2007Photocopiers; @Isaacs96:piAZZa]. in generaL, if spaCEs EnCOurage thE mIxiNg anD MEetinG of a DIvErse rangE of peoPLe, SeRendipiToUs meetInGs bEtwEen inDIvidUals frOm differEnt teAMs or social grouPS will occur morE ReADIlY, WhicH coUld be cruciaL; facE-To-faCe coMMuNicATion hAs beeN sHOwN To be more important thAn ElectrOnic mEans such as emaIl or SMS [@PenTLANd12:New; @StrYker12:fAcILitating].
Indeed, DesigNing the layOUt of workSpaceS in accorDance with OFFice sociAl dYnaMicS reSULtIng from the bacKGRounD oF employEes Can help TheM woRk mOre EfFectively, As was shoWn By A rEcEnt Study OF offices AcRosS 11 dIffErent COuntriEs [@SteElcaSe13:cuLTurE]. While tHIs APProaCh TaKes iNto AcCount The nATioNal cultUre of the cOunTRy whErE aN office Is situated, it dOeS not considEr WhaT happeNS In very cuLturally diverse environmENts that AccOmmodAte wOrkers froM maNy diffEreNT backgRounds.
the efFeCts OF CultuRAL vAriAtIons betweeN COunTries On OrgaNisatioNs have been studied bY geeRt Hofstede, by mEanS of aDMInIstERiNG opInIOn sURVeys to IBM employEes in over 70 cOuNTrIes. He derivED fiVe Main facTors, or cUlturAL dimensIons, that cAn account FoR mosT OF thE variance hE observeD [@Hofstede10:cUlturES], aNd fouNd tHat worKeRs’ cUlturAl backGRouNds caN shape ThE way thEy thiNk, Feel, and aCt. Besides the workplace enVironmEnt itSelF, cultural DifFEreNces betweEn itS occupants CouLd aFfect OffICe socIal dYNaMicS And, thErefORe, the propENsIty OF WoRkers to engaGE IN beNeficIal SErendiPitoUs inter-group interACtions.
Since intEracTIOns BetWEen pEoPle with differeNt aReAS Of expertIsE or social ciRcles are So BEnefiCial foR the exChange oF INfORmatioN and IdeAs, it is impOrtAnT To underStAnD | 00:Sharing; @Whittaker94:I nformal].Thiside a w as rec entl y further conf i rmed by Pentland *et al.*, in a s t udyo fworkp lace co m mu n i cat io npat te r ns at a Pr ague ba nk [@Pentl and 12 :New]. Theyd is covered akey characteris tic of su cc ess f ul te ams : mem bers p e riodic ally inte ra c t with otherso u ts ideof their team, an d b r ing back new i nforma ti o n. T hey du bbed thiscr itica l dimens i on o f co m munication ‘e xploration’ , th e tend en cyf or a t eam t os eek inter-grou p co nnections , andf ound th a t teams withhig h e xplo r at io n t en d edt obem ore success fu l, demo nstr a t i n g th e p ower of s erendipitousint erac t ion s inthe o ffic e.
One facto r tha tcould clearly a ffec t the eas e o fsuc hinter a ctions is th e physi cal spa c esof t h eworkplace itself;fo r ex ample, h igh-tr a ff ic areas su ch as cof f e e mac hine s a nd photo copier s m ay be par ti cularl ylik ely plac e s fo r inte r-groupmeeti n gs [@fayard200 7 photocopiers; @I s a ac s 96:P iaz za]. In gen eral , ifspac e senc o urage themi x in g and meeting of a d iv erse r angeof people, se rendipitou s m eetingsbetw e en individuals fr om di fferent te a ms or so cialgroups w ill occur m ore read ily , w hic h c o u ld be crucial;f a ce-t o- face co mmu nicatio n h asbee n s ho wn to bemore imp or ta nt t han elec t ronic me an s s uc h a s ema i l or S MS [@ Pent la nd 1 2:N ew; @St r yk e r 12:F ac il itat ing ].
Ind eed, des igningthe layou t o f wor ks pa ces inaccordance wi th office so ci aldynami c s resulti ng from the backgroundo f emplo yee s can hel p them wo rkmore e ffe c tively , as w as sh ow n b y a rec e n tstu dy of office s acr oss 1 1diff erent c ountries [@Steelca s e13 :Culture]. Wh ile thi s ap pro a ch tak es int o account the nat ional cult ur e o f the coun t rywh ere anofficeis si t uated,it does n ot consid er wha t hap pens in ve ry cultu rally div e rse e n vi ronme nts thatac com modat e work e rsfrommany d if ferent back gr ounds.
The effects of cultural varia tions be tween cou ntr i eson organi sati ons have b een st udied by Geert Hof s te de, by me anso f adminis t er ing o pi nion survey s t o I BM em plo y ees in ove r 70 countries. H e derived fivemain f act ors , orcu ltural dimensi ons ,t h at can a cc ount for mo st of th ev arian ce heobserv ed [@Ho f s te d e10:Cu ltur es] , and fou ndth a t worke rs ’c ultura l ba ck ground s cans hape t he way they thin k, fe e l , and act . Bes id es thew orkp lace envir onment itse lf, cu ltur al di fferenc es betwe enit s occupant s could af fectofficeso cial dy namics and , there fore ,the propensi t y o f w or k ers toengag ein b eneficial serendip ito u s inter -g rou p intera c ti o n s.
Sinceint eract i o ns between peop l ew ith d iffere nt are as of e x per ti se or s oci a l circlesare so be n efi ci al f or the e xc hang eofin formation and i deas, it is impo r t ant to un d erst a nd | 00:Sharing; @Whittaker94:Informal]._This idea_was recently further confirmed_by Pentland_*et_al.*, in_a_study of workplace_communication patterns at_a Prague bank [@Pentland12:New]. They_discovered a key_characteristic_of successful teams: members periodically interact with others outside of their team, and bring_back_new information._They_dubbed_this critical dimension of communication_‘exploration’, the tendency for a_team to_seek inter-group connections, and found that teams with_high_exploration tended to_be more successful, demonstrating the power of serendipitous interactions_in the office.
One factor that could_clearly affect the_ease_of_such interactions is the_physical spaces of the workplace itself;_for example, high-traffic areas such as_coffee machines and photocopiers may be particularly_likely places for inter-group meetings [@fayard2007photocopiers; @Isaacs96:Piazza]._In general, if spaces encourage_the mixing_and meeting of a diverse_range of people,_serendipitous meetings_between individuals from_different teams or social groups will_occur more readily,_which could be crucial; face-to-face communication_has_been shown to_be_more_important than_electronic means such_as_email or_SMS [@Pentland12:New;_@Stryker12:Facilitating].
Indeed, designing the layout of workspaces_in_accordance with office social dynamics resulting from_the background of employees_can_help them work more_effectively, as was shown by_a recent study of offices across_11 different_countries [@Steelcase13:Culture]. While_this approach takes into account the national culture of the country_where an office is situated, it_does not consider what_happens in_very_culturally diverse environments_that_accommodate workers_from many different backgrounds.
The effects of cultural_variations between_countries on organisations have been studied_by Geert Hofstede, by_means_of administering opinion surveys to IBM_employees in over 70 countries. He_derived five main factors, or_cultural_dimensions,_that can account for most_of the variance he observed [@Hofstede10:Cultures], and_found that workers’_cultural backgrounds can shape the way they_think,_feel, and act. Besides the workplace_environment_itself, cultural differences between its occupants_could_affect_office social dynamics and, therefore,_the propensity of workers to engage_in beneficial serendipitous inter-group interactions.
Since interactions between people with_different areas of_expertise or social circles are_so_beneficial_for the exchange of information and ideas, it is important_to understand |
\over 2}(\cos(\theta_j) + \cos(\theta_{j+1}))\right],
\label{thetacenter}\end{aligned}$$ and the zone widths are $(\Delta r)_{i'} = r_{i+1} - r_i$ and $(\Delta \theta)_{j'} = \theta_{j+1} - \theta_j$. The definitions of $(\cos\vartheta)_{\beta'}$, $(\sin\vartheta)_{\beta'}$, and $(\cos\varphi)_{\gamma'}$ in (\[spaceDivergence\]) will be given below. Finally, the values of ${\cal N}$ on zone surfaces in (\[spaceDivergence\]) are given by a particular linear interpolation of zone center values. This linear interpolation—which depends on the neutrino mean free paths—has the effect of shifting from “diamond” differencing in diffusive regimes to “upwind” (or “donor-cell”) differencing in free-streaming regions; see [@liebendoerfer02] for details.
A conservative differencing of the momentum space divergence in (\[example\]) is $$\begin{aligned}
-{1\over {\cal V}_{\alpha',\beta',\gamma'}}\left(1\over r\right)_{i'} \left[({\cal A}_\vartheta\,{\cal N})_{\beta+1,\gamma'} -
({\cal A}_\vartheta\,{\cal N})_{\beta,\gamma'} \right] &-&
\nonumber \\
{1\over {\cal V}_{\alpha',\beta',\gamma'}}{(\cot\theta)_{j'} \over r_{i'}}
\left[({\cal A}_\varphi\,{\cal N})_{\beta',\gamma+1} -
({\cal A}_\varphi\,{\cal N})_{\beta',\gamma} \right]&,& \label{momentumDivergence}\end{aligned}$$ where the momentum space “geometric” factors are $$\begin{aligned}
{\cal V}_{\alpha',\beta',\gamma'}&=& (\epsilon_{\alpha'})^2 \sin(\vartheta_{\beta'}) (\Delta \epsilon)_{\alpha'} (\Delta\vartheta)_{\beta'} (\Delta\varphi)_{\gamma'}, \\
({\cal A}_\vartheta)_{\beta,\gamma'} &=& (\epsilon_{\alpha'})^2 [\sin(\vartheta_{\beta})]^2 (\Delta \epsilon)_{\alpha'} (\Delta\varphi)_{\gamma'}, \\
({\cal A}_\varphi)_{\beta',\gamma} &=& | \over 2}(\cos(\theta_j) + \cos(\theta_{j+1}))\right ],
\label{thetacenter}\end{aligned}$$ and the zone widths are $ (\Delta r)_{i' } = r_{i+1 } - r_i$ and $ (\Delta \theta)_{j' } = \theta_{j+1 } - \theta_j$. The definitions of $ (\cos\vartheta)_{\beta'}$, $ (\sin\vartheta)_{\beta'}$, and $ (\cos\varphi)_{\gamma'}$ in (\[spaceDivergence\ ]) will be given below. ultimately, the value of $ { \cal N}$ on zone surfaces in (\[spaceDivergence\ ]) are given by a particular analogue interpolation of zone center values. This analogue interjection — which depends on the neutrino mean free paths — has the impression of shifting from “ diamond ” differencing in diffusing regimes to “ upwind ” (or “ donor - cell ”) differencing in free - streaming regions; see [ @liebendoerfer02 ] for details.
A cautious differencing of the momentum space divergence in (\[example\ ]) is $ $ \begin{aligned }
-{1\over { \cal V}_{\alpha',\beta',\gamma'}}\left(1\over r\right)_{i' } \left[({\cal A}_\vartheta\,{\cal N})_{\beta+1,\gamma' } -
({ \cal A}_\vartheta\,{\cal N})_{\beta,\gamma' } \right ] & - &
\nonumber \\
{ 1\over { \cal V}_{\alpha',\beta',\gamma'}}{(\cot\theta)_{j' } \over r_{i' } }
\left[({\cal A}_\varphi\,{\cal N})_{\beta',\gamma+1 } -
({ \cal A}_\varphi\,{\cal N})_{\beta',\gamma } \right ] &, & \label{momentumDivergence}\end{aligned}$$ where the momentum space “ geometric ” agent are $ $ \begin{aligned }
{ \cal V}_{\alpha',\beta',\gamma'}&= & (\epsilon_{\alpha'})^2 \sin(\vartheta_{\beta' }) (\Delta \epsilon)_{\alpha' } (\Delta\vartheta)_{\beta' } (\Delta\varphi)_{\gamma' }, \\
({ \cal A}_\vartheta)_{\beta,\gamma' } & = & (\epsilon_{\alpha'})^2 [ \sin(\vartheta_{\beta})]^2 (\Delta \epsilon)_{\alpha' } (\Delta\varphi)_{\gamma' }, \\
({ \cal A}_\varphi)_{\beta',\gamma } & = & | \oveg 2}(\cos(\theta_j) + \cos(\theta_{j+1}))\rinht],
\label{thetacenjee}\end{almgned}$$ ahd the zune widths are $(\Delta r)_{i'} = r_{i+1} - r_u$ and $(\Delta \theta)_{j'} = \theta_{j+1} - \theta_j$. Tje definutiois of $(\cos\varthete)_{\geta'}$, $(\sik\rarthsba)_{\betc'}$, end $(\cos\varphi)_{\gakma'}$ in (\[spaweDivergence\]) whlu ye given below. Finally, the values of ${\cal N}$ pn zone surfaces in (\[x[aceSpvtrgence\]) are given by a particular linear interpolatiom of zone center values. Thls llnear interpolatioj—which depebds jb the neutrivo mean frte paths—has jhe effect of shifting from “diamund” dnfferencing ib dlxfusive regmmes tj “upwind” (or “donor-celn”) diffetencing in frec-streemint regions; see [@liebendmerfer02] for details.
W conservdtnve differencing of tye momettum spazw dkvedgxncs in (\[edam'le\]) is $$\begih{aligned}
-{1\ovee {\cal V}_{\alpha',\beta',\gamka'}}\jvgt(1\over r\righf)_{i'} \lefe[({\cwl A}_\vartheta\,{\cal N})_{\beta+1,\gamma'} -
({\cal A}_\varthetd\,{\cam N})_{\beta,\gamma'} \right] &-&
\nonunber \\
{1\over {\cal V}_{\alpha',\heta',\gamma'}}{(\sot\theta)_{j'} \over r_{i'}}
\left[({\cal A}_\varphi\,{\cal N})_{\beta',\gamma+1} -
({\can A}_\vacpfi\,{\cco N})_{\begq',\gwmma} \right]&,& \label{momentumDivergence}\end{aligned}$$ rgete the momentum fpace “geomeyrlc” sactors are $$\bggin{aliyhes}
{\cal V}_{\alpha',\beta',\gamla'}&=& (\epsijon_{\alpha'})^2 \sin(\vawtheya_{\beta'}) (\Delta \epsilon)_{\alpha'} (\Eelta\varthetc)_{\bera'} (\Delta\varphi)_{\gammc'}, \\
({\cal A}_\varthzta)_{\bets,\gamms'} &=& (\epsilon_{\alpha'})^2 [\sin(\vartketa_{\befa})]^2 (\Delta \epdilon)_{\alphz'} (\Delta\varphi)_{\gammx'}, \\
({\csl A}_\varphi)_{\beta',\gamma} &=& | \over 2}(\cos(\theta_j) + \cos(\theta_{j+1}))\right], \label{thetacenter}\end{aligned}$$ and the are r)_{i'} = - r_i$ and \theta_j$. definitions of $(\cos\vartheta)_{\beta'}$, and $(\cos\varphi)_{\gamma'}$ in will be given below. Finally, the of ${\cal N}$ on zone surfaces in (\[spaceDivergence\]) are given by a particular interpolation of zone center values. This linear interpolation—which depends on the neutrino mean paths—has effect shifting “diamond” differencing in diffusive regimes to “upwind” (or “donor-cell”) differencing in free-streaming regions; see [@liebendoerfer02] for A conservative differencing of the momentum space divergence (\[example\]) is $$\begin{aligned} -{1\over V}_{\alpha',\beta',\gamma'}}\left(1\over r\right)_{i'} \left[({\cal A}_\vartheta\,{\cal N})_{\beta+1,\gamma'} ({\cal N})_{\beta,\gamma'} \right] \nonumber {1\over V}_{\alpha',\beta',\gamma'}}{(\cot\theta)_{j'} \over r_{i'}} A}_\varphi\,{\cal N})_{\beta',\gamma+1} - ({\cal A}_\varphi\,{\cal N})_{\beta',\gamma} \right]&,& \label{momentumDivergence}\end{aligned}$$ where the momentum space “geometric” factors are $$\begin{aligned} {\cal (\epsilon_{\alpha'})^2 \sin(\vartheta_{\beta'}) (\Delta\vartheta)_{\beta'} (\Delta\varphi)_{\gamma'}, ({\cal &=& [\sin(\vartheta_{\beta})]^2 (\Delta \epsilon)_{\alpha'} ({\cal A}_\varphi)_{\beta',\gamma} &=& | \over 2}(\cos(\theta_j) + \cos(\theta_{j+1}))\rigHt],
\label{theTacenTer}\End{AlIgneD}$$ and The zone widths aRE $(\DelTa r)_{i'} = r_{i+1} - r_i$ and $(\Delta \theta)_{j'} = \Theta_{J+1} - \tHEta_j$. tHe DefinItions oF $(\CoS\VArtHeTa)_{\BetA'}$, $(\sIN\vArtheTa)_{\bEta'}$, and $(\cOs\varphi)_{\gaMma'}$ In (\[SpaceDivergeNCe\]) Will be giveN beLow. Finally, thE vaLues of ${\CaL N}$ oN Zone sUrfAces iN (\[spacedIvergeNce\]) are givEn BY a partICular liNEAr InteRpolation of zone ceNTeR Values. This lineAr inteRpOLaTIOn—wHicH depends on ThE neutRIno mean FReE PAThs—HAs the effect of Shifting froM “DiaMond” diFfEreNCing in DiffuSiVE reGimes to “upwiNd” (or “Donor-cell”) DifferENcing in FRee-streAming rEgiOns; See [@lIEbEnDoeRfER02] foR DeTaiLS.
A cOnservatIvE dIfferEnciNG OF The mOmeNtum Space Divergence in (\[eXamPle\]) iS $$\BegIn{aliGned}
-{1\oVer {\cAl v}_{\alphA',\beta',\gAmma'}}\lEfT(1\over r\right)_{i'} \lefT[({\cal a}_\vartheta\,{\Cal n})_{\bEta+1,\GaMma'} -
({\caL a}_\varthEta\,{\Cal n})_{\beta,\gaMma'} \righT] &-&
\NonUmBER \\
{1\OvEr {\cal V}_{\alpha',\beta',\gamMa'}}{(\COT\tHeta)_{j'} \oveR r_{i'}}
\lefT[({\CaL A}_\VArphi\,{\cal n})_{\bEta',\GammA+1} -
({\CAl A}_\vaRphi\,{\CAl n})_{\beta',\gamMa} \righT]&,& \LaBeL{momentUmdivergEnCe}\eNd{aLigneD}$$ WherE the moMentum spAce “geOMetric” factors aRE $$\begin{aligned}
{\CAl v}_{\ALpHA',\betA',\gaMma'}&=& (\epsilon_{\aLpha'})^2 \SIn(\vaRtheTA_{\bEta'}) (\dElta \ePsiloN)_{\aLPhA'} (\delta\vartheta)_{\beta'} (\DeLtA\varphI)_{\gammA'}, \\
({\cal A}_\vartheta)_{\Beta,\gamma'} &=& (\ePSILon_{\alpha'})^2 [\Sin(\vARtHEta_{\beta})]^2 (\Delta \epSilon)_{\Alpha'} (\Delta\VArphi)_{\gamMa'}, \\
({\cal a}_\varphi)_{\bEta',\gamma} &=& | \over 2}(\cos(\theta_j) +\cos(\thet a_{j+ 1}) )\r ig ht],
\la bel{thetacente r }\en d{aligned}$$ and the z one w id t hs a r e$(\De lta r)_ { i' } = r _{ i+ 1}-r _i $ and $( \Delta\theta)_{j '}=\theta_{j+1} -\theta_j$. Th e definition s o f $(\c os \va r theta )_{ \beta '}$, $ ( \sin\v artheta)_ {\ b eta'}$ , and $( \ c os \var phi)_{\gamma'}$ i n ( \ [spaceDivergen ce\])wi l lb e gi ven below. Fi na lly,t he valu e so f ${\ c al N}$ on zon e surfacesi n ( \[spac eD ive r gence\ ]) ar eg ive n by a part icul ar linear inter p olation of zone cente r v alu es.T hi slin ea r in t er pol a tio n—whichde pe nds o n th e n e utri nomean free paths—has th e e ffec t of shif tingfrom “ diamo nd” di ffere nc ing in diffusiv e re gimes to“up wi nd” ( or “d o nor-ce ll” ) d ifferen cing in fre e- s t r ea ming regions; see[@ l i eb endoerfe r02] f o rde t ails.
A c ons erva t i ve di ffer e nc ing of t he mom e nt um spacedi vergen ce in (\ [exam p le\] ) is $ $\begin{ align e d}
-{1\over {\ c al V}_{\alpha ' ,\ b e ta ' ,\ga mma '}}\left(1\ over r\ri ght) _ {i '}\ left[ ({\ca lA }_ \ vartheta\,{\cal N}) _{ \beta+ 1,\ga mma'} -
({\ca l A}_\vart h e t a\,{\cal N}) _ {\ b eta,\gamma'} \ right ] &-&
\non u mber \\{1\ov er {\ca l V}_{\al p h a',\beta ',\ gam ma' }}{ ( \ co t\theta)_{j'} \ over r _{i'}}\le ft[({\c alA}_ \va rph i\ ,{\cal N} )_{\beta ', \g am ma +1} -
({ \ cal A}_\ va rph i\ ,{\ cal N } )_{\be ta',\ gamm a} \ r igh t]&,& \ l ab e l {mom en tu mDiv erg en ce}\e nd{a l ign ed}$$ w here themom e ntum s pa ce “geo metric” facto rs are $$\be gi n{a ligned } {\cal V} _{\alpha',\beta',\gamma ' }&=& (\ eps ilon_ {\al pha'})^2\si n(\var the t a_{\be ta'})(\Del ta \e p s ilon) _ { \a lph a' } (\Delta \ v art heta) _{ \bet a'} (\D elta\varphi)_{\gam m a'} , \\
({\cal A }_\ vart h e ta )_{ \ be t a,\ ga m ma' } &=& (\epsilon_{ \alpha'})^ 2[ \s in(\varthe t a_{ \b eta})]^ 2 (\Del ta \e p silon)_ {\alpha'} (\Delta\ va rphi ) _ {\g amma'}, \\
({\calA}_\varph i )_{\b e ta ',\ga mma } &=& | \over 2}(\cos(\theta_j)_+ \cos(\theta_{j+1}))\right],
\label{thetacenter}\end{aligned}$$_and the zone widths_are $(\Delta_r)_{i'}_= r_{i+1}_-_r_i$ and $(\Delta_\theta)_{j'} = \theta_{j+1}_- \theta_j$. The definitions_of $(\cos\vartheta)_{\beta'}$, $(\sin\vartheta)_{\beta'}$,_and_$(\cos\varphi)_{\gamma'}$ in (\[spaceDivergence\]) will be given below. Finally, the values of ${\cal N}$ on_zone_surfaces in_(\[spaceDivergence\])_are_given by a particular linear_interpolation of zone center values._This linear_interpolation—which depends on the neutrino mean free paths—has_the_effect of shifting_from “diamond” differencing in diffusive regimes to “upwind” (or_“donor-cell”) differencing in free-streaming regions; see_[@liebendoerfer02] for details.
A_conservative_differencing_of the momentum space_divergence in (\[example\]) is $$\begin{aligned}
-{1\over {\cal_V}_{\alpha',\beta',\gamma'}}\left(1\over r\right)_{i'} \left[({\cal A}_\vartheta\,{\cal N})_{\beta+1,\gamma'} -
({\cal_A}_\vartheta\,{\cal N})_{\beta,\gamma'} \right] &-&
\nonumber \\
{1\over {\cal_V}_{\alpha',\beta',\gamma'}}{(\cot\theta)_{j'} \over r_{i'}}
\left[({\cal A}_\varphi\,{\cal N})_{\beta',\gamma+1} -
({\cal_A}_\varphi\,{\cal N})_{\beta',\gamma} \right]&,& \label{momentumDivergence}\end{aligned}$$ where_the momentum_space “geometric” factors are $$\begin{aligned}
{\cal_V}_{\alpha',\beta',\gamma'}&=& (\epsilon_{\alpha'})^2 \sin(\vartheta_{\beta'})_(\Delta \epsilon)_{\alpha'}_ (\Delta\vartheta)_{\beta'} (\Delta\varphi)_{\gamma'},_\\
({\cal A}_\vartheta)_{\beta,\gamma'} &=& (\epsilon_{\alpha'})^2 [\sin(\vartheta_{\beta})]^2 (\Delta_\epsilon)_{\alpha'} (\Delta\varphi)_{\gamma'}, \\
({\cal_A}_\varphi)_{\beta',\gamma} &=& |
_P^k g_1^m \prod_{s=1}^{q-2} h_s^{n_s}$. Equations,, combined with Lemmas \[lem:g1\] and \[lem:hs\] imply that $f \in L((2g(\cX)-2)P_\infty)$ if $$i(q^3+1)+j(q^3-q^2+q)+k(q^3+1)+m(q^4+q)+\sum_{s=1}^{q-2} n_s ((s+1)q-s)(q^3+1) \leq q^5-2q^3+q^2-2,$$ which is equivalent to $$\label{eq:inquality1}
i(q+1)+jq+k(q+1)+mq(q+1)+\sum_{s=1}^{q-2} n_s ((s+1)q-s)(q+1) \leq (q+1)(q^2-2).$$ On the other hand we have $$v_P(f)=iq^3+j+kq+m(q^2+1)+\sum_{s=1}^{q-2} n_s ((s+1)q^2).$$ Hence the claim follows from Lemma \[holom\].
\[obs:largestgapinG\] Inequality implies in particular that $i\leq q^2-2,j \leq q^2+q-3,k \leq q^2-2$, $m \leq q-1$ and $n_s \leq \lfloor (q+1)/(s+1)\rfloor$. This implies directly that the largest gap of $H(P)$ that is contained in $G$ is obtained by putting $i=q^2-2$ and all other remaining variables to $0$. In other words: the largest element in $G$ is $q^5-2q^3+1=2g(\cX)-q^2+1.$
\[obs:jsmall\] If $j\ge q$ and the tuple $(i,j,k,m,n_1,\dots,n_{q-2})$ satisfies inequality, then the tuple $(i,j-q,k+1,m,n_1,\dots,n_s)$ will also satisfy inequality. | _ P^k g_1^m \prod_{s=1}^{q-2 } h_s^{n_s}$. Equations, , combined with Lemmas \[lem: g1\ ] and \[lem: hs\ ] imply that $ f \in L((2g(\cX)-2)P_\infty)$ if $ $ i(q^3 + 1)+j(q^3 - q^2+q)+k(q^3 + 1)+m(q^4+q)+\sum_{s=1}^{q-2 } n_s (( s+1)q - s)(q^3 + 1) \leq q^5 - 2q^3+q^2 - 2,$$ which is equivalent to $ $ \label{eq: inquality1 }
i(q+1)+jq+k(q+1)+mq(q+1)+\sum_{s=1}^{q-2 } n_s (( s+1)q - s)(q+1) \leq (q+1)(q^2 - 2).$$ On the other hand we get $ $ v_P(f)=iq^3+j+kq+m(q^2 + 1)+\sum_{s=1}^{q-2 } n_s (( s+1)q^2).$$ therefore the claim follows from Lemma \[holom\ ].
\[obs: largestgapinG\ ] Inequality entail in finical that $ i\leq q^2 - 2,j \leq q^2+q-3,k \leq q^2 - 2 $, $ m \leq q-1 $ and $ n_s \leq \lfloor (q+1)/(s+1)\rfloor$. This implies directly that the big col of $ H(P)$ that is contained in $ G$ is obtained by putting $ i = q^2 - 2 $ and all other persist variables to $ 0$. In other actor's line: the largest element in $ G$ is $ q^5 - 2q^3 + 1=2g(\cX)-q^2 + 1.$
\[obs: jsmall\ ] If $ j\ge q$ and the tuple $ (i, j, k, m, n_1,\dots, n_{q-2})$ satisfies inequality, then the tuple $ (i, j - q, k+1,m, n_1,\dots, n_s)$ will also meet inequality. | _P^k g_1^m \prod_{s=1}^{q-2} h_s^{n_s}$. Equationr,, combined with Lemmas \[lem:g1\] znd \[lem:hr\] imply that $f \in L((2g(\cX)-2)P_\infty)$ id $$i(q^3+1)+j(w^3-q^2+q)+k(q^3+1)+m(q^4+q)+\sum_{s=1}^{q-2} n_s ((s+1)q-s)(q^3+1) \ueq q^5-2q^3+q^2-2,$$ wjich is wquitalent to $$\label{eq:inqualibv1}
i(q+1)+jq+i(e+1)+mq(q+1)+\vnm_{s=1}^{q-2} n_s ((s+1)q-s)(q+1) \lea (q+1)(q^2-2).$$ On tha other hand wa fare $$v_P(f)=iq^3+j+kq+m(q^2+1)+\sum_{s=1}^{q-2} n_s ((s+1)q^2).$$ Hence the cjaim fokllws from Lemma \[holpi\].
\[obs:margestgapinG\] Inequality implies ih partibular that $i\leq q^2-2,k \leq q^2+q-3,k \leq q^2-2$, $m \leq q-1$ anf $n_s \leq \lfloor (q+1)/(s+1)\rflolr$. This impoies eirectly thag the largtsc gap of $H(P)$ that is contained in $G$ is obtakned yy putting $u=q^2-2$ anf all other cemainpng variables to $0$. In mther wprds: the largext xlemwnt in $G$ is $q^5-2q^3+1=2g(\cX)-q^2+1.$
\[obs:osmall\] If $j\ge q$ and jhe tuple $(h,j,i,m,n_1,\dots,n_{q-2})$ satisfiws ineqoalitf, thdb tfe uupke $(i,j-q,k+1,m,j_1,\dovs,n_s)$ will amso satisfy inequality. | _P^k g_1^m \prod_{s=1}^{q-2} h_s^{n_s}$. Equations,, combined with and imply that \in L((2g(\cX)-2)P_\infty)$ if which equivalent to $$\label{eq:inquality1} n_s ((s+1)q-s)(q+1) \leq On the other hand we have n_s ((s+1)q^2).$$ Hence the claim follows from Lemma \[holom\]. \[obs:largestgapinG\] Inequality implies in that $i\leq q^2-2,j \leq q^2+q-3,k \leq q^2-2$, $m \leq q-1$ and $n_s \leq (q+1)/(s+1)\rfloor$. implies that largest gap of $H(P)$ that is contained in $G$ is obtained by putting $i=q^2-2$ and all remaining variables to $0$. In other words: the element in $G$ is \[obs:jsmall\] If $j\ge q$ and tuple satisfies inequality, the $(i,j-q,k+1,m,n_1,\dots,n_s)$ also satisfy inequality. | _P^k g_1^m \prod_{s=1}^{q-2} h_s^{n_s}$. Equations,, coMbined with lemmaS \[leM:g1\] aNd \[Lem:hS\] impLy that $f \in L((2g(\cX)-2)P_\INfty)$ If $$i(q^3+1)+j(q^3-q^2+q)+k(q^3+1)+m(q^4+q)+\sum_{s=1}^{q-2} n_s ((s+1)q-S)(q^3+1) \leq Q^5-2q^3+Q^2-2,$$ WhicH Is EquivAlent to $$\LAbEL{Eq:iNqUaLitY1}
i(Q+1)+Jq+K(q+1)+mq(q+1)+\Sum_{S=1}^{q-2} n_s ((s+1)q-s)(Q+1) \leq (q+1)(q^2-2).$$ On thE otHeR hand we have $$v_p(F)=iQ^3+j+kq+m(q^2+1)+\sum_{s=1}^{Q-2} n_s ((S+1)q^2).$$ Hence the clAim FollowS fRom lEmma \[hOloM\].
\[obs:lArgestGApinG\] INequality ImPLies in PArticulAR ThAt $i\lEq q^2-2,j \leq q^2+q-3,k \leq q^2-2$, $m \leQ Q-1$ aND $n_s \leq \lfloor (q+1)/(s+1)\Rfloor$. thIS iMPLieS diRectly that ThE largESt gap of $h(p)$ tHAT Is cONtained in $G$ is oBtained by puTTinG $i=q^2-2$ and AlL otHEr remaIning VaRIabLes to $0$. In otheR worDs: the largEst eleMEnt in $G$ iS $Q^5-2q^3+1=2g(\cX)-q^2+1.$
\[oBs:jsmaLl\] IF $j\gE q$ anD ThE tUplE $(i,J,K,m,n_1,\DOtS,n_{q-2})$ SAtiSfies ineQuAlIty, thEn thE TUPLe $(i,j-Q,k+1,m,N_1,\dotS,n_s)$ wiLl also satisfy IneQualITy. | _P^k g_1^m \prod_{s=1}^{q -2} h_s^{n _s}$. Eq uat io ns,, com bined with Lem m as \ [lem:g1\] and \[lem:hs \] im pl y tha t $ f \in L((2g( \ cX ) - 2)P _\ in fty )$ if $$i( q^3 +1)+j(q ^3-q^2+q)+ k(q ^3 +1)+m(q^4+q) + \s um_{s=1}^{ q-2 } n_s ((s+1) q-s )(q^3+ 1) \l e q q^5 -2q ^3+q^ 2-2,$$ whichis equiva le n t to $ $ \label{ e q :i nqua lity1}
i(q+1)+jq+ k (q + 1)+mq(q+1)+\su m_{s=1 }^ { q- 2 } n_ s ( (s+1)q-s)( q+ 1) \l e q (q+1) ( q^ 2 - 2 ).$ $ On the other hand we ha v e $ $v_P(f )= iq^ 3 +j+kq+ m(q^2 +1 ) +\s um_{s=1}^{q -2}n_s ((s+1 )q^2). $ $ Hence the cla im fol low s f romL em ma \[ ho l om\ ] .
\[ o bs: largestg ap in G\] I nequ a l i t y im pli es i n par ticular that$i\ leqq ^2- 2,j \ leq q ^2+q -3 ,k \l eq q^2 -2$,$m \leq q-1$ and$n_s \leq \lf loo r(q+ 1) /(s+1 ) \rfloo r$. Th is impl ies dir e ctl yt h a tthe largest gap of $ H ( P) $ that i s cont a in ed in $G$ i sobt aine d by pu ttin g $ i=q^2-2$ and a l lot her rem ai ning v ar iab les to $ 0 $. I n othe r words: thel argest element in $G$ is $q^ 5 -2 q ^ 3+ 1 =2g( \cX )-q^2+1.$
\[ob s :jsm all\ ] I f $ j \ge q $ and t h et uple $(i,j,k,m,n_1, \d ots,n_ {q-2} )$ satisfiesinequality , t hen thetupl e $ ( i,j-q,k+1,m,n_ 1,\do ts,n_s)$ w i ll alsosatis fy inequ ality. | _P^k _g_1^m \prod_{s=1}^{q-2}_h_s^{n_s}$. Equations,, combined with_Lemmas \[lem:g1\]_and_\[lem:hs\] imply_that_$f \in L((2g(\cX)-2)P_\infty)$_if $$i(q^3+1)+j(q^3-q^2+q)+k(q^3+1)+m(q^4+q)+\sum_{s=1}^{q-2} n_s_((s+1)q-s)(q^3+1) \leq q^5-2q^3+q^2-2,$$ which_is equivalent to_$$\label{eq:inquality1}
i(q+1)+jq+k(q+1)+mq(q+1)+\sum_{s=1}^{q-2}_n_s ((s+1)q-s)(q+1) \leq (q+1)(q^2-2).$$ On the other hand we have $$v_P(f)=iq^3+j+kq+m(q^2+1)+\sum_{s=1}^{q-2} n_s ((s+1)q^2).$$ Hence_the_claim follows_from_Lemma_\[holom\].
\[obs:largestgapinG\] Inequality implies in particular_that $i\leq q^2-2,j \leq q^2+q-3,k_\leq q^2-2$,_$m \leq q-1$ and $n_s \leq \lfloor (q+1)/(s+1)\rfloor$._This_implies directly that_the largest gap of $H(P)$ that is contained in_$G$ is obtained by putting $i=q^2-2$_and all other_remaining_variables_to $0$. In other_words: the largest element in $G$_is $q^5-2q^3+1=2g(\cX)-q^2+1.$
\[obs:jsmall\] If $j\ge q$ and_the tuple $(i,j,k,m,n_1,\dots,n_{q-2})$ satisfies inequality, then the_tuple $(i,j-q,k+1,m,n_1,\dots,n_s)$ will also satisfy inequality. |
1+\frac{\epsilon}{2}.$$ With this choice, $\delta_0$ in depends only on $\epsilon$. Thus we have to verify that $$\left(\frac{1}{\log|\log{|t|}|}+|c|\right)^{\frac{1}{k}} \leq |\log|t||^{\frac{\epsilon}{2}},$$ but this obviously holds for small $|t|$ since the left-hand side is bounded. Hence $\tilde{\Omega} \subset \Omega$ and $(0,0)$ is an irregular boundary point for $\Omega$ as well.
This implies that the upper Perron solution $\overline{H}_f$ does not always attain the boundary values at $(0, 0)$.
Barrier Families {#sec:appa}
================
In this Section, a sub/supersolution is a viscosity sub/supersolution of $$\label{eq:infinity_heat}
u_t -\Delta_{\infty}u = u_t -\sum_{i, j=1}^{n}u_{x_i}u_{x_ix_j}u_{x_j} =0,$$ the usual, non-normalised $\infty$- heat equation. Further, the upper and lower classes $\mathcal{U}_f$ and $\mathcal{L}_f$, and the Upper and Lower Perron solutions $\overline{H}_f$ and $\underline{H}_f$, now also refer to. As before, the boundary function $f$ is assumed to be continuous.
Note that if $u$ is a radial function $u(x, t)=G(|x|, t)$, then reads $$G_t(r, t)-G_r(r, t)^2G_{rr}(r, t)=0.$$ Also, if $u$ is a sub/supersolution to, then $-u$ is a\
super/subsolution.
The following definition is found in [@Bjorn2015]:
\[def:stronk\_barrier\] Let $\zeta_0 =(x_0, t_0) \in \partial \Omega$. A family of functions $\{w_j\}_j $ in $\Omega$ is called a *barrier family* in $\Omega$ at the point $\zeta_0$ if, for every $j$,
1. $w_j>0$ and $w_j$ a supers | 1+\frac{\epsilon}{2}.$$ With this choice, $ \delta_0 $ in depends only on $ \epsilon$. Thus we get to control that $ $ \left(\frac{1}{\log|\log{|t|}|}+|c|\right)^{\frac{1}{k } } \leq |\log|t||^{\frac{\epsilon}{2}},$$ but this obviously holds for small $ |t|$ since the left - handwriting side is bounded. therefore $ \tilde{\Omega } \subset \Omega$ and $ (0,0)$ is an irregular boundary point for $ \Omega$ as well.
This imply that the upper Perron solution $ \overline{H}_f$ does not always attain the boundary value at $ (0, 0)$.
Barrier Families { # sec: appa }
= = = = = = = = = = = = = = = =
In this Section, a sub / supersolution is a viscosity submarine / supersolution of $ $ \label{eq: infinity_heat }
u_t -\Delta_{\infty}u = u_t -\sum_{i, j=1}^{n}u_{x_i}u_{x_ix_j}u_{x_j } = 0,$$ the common, non - normalised $ \infty$- heat equality. Further, the upper and humble class $ \mathcal{U}_f$ and $ \mathcal{L}_f$, and the Upper and Lower Perron solutions $ \overline{H}_f$ and $ \underline{H}_f$, now also refer to. As ahead, the boundary function $ f$ is assumed to be continuous.
Note that if $ u$ is a radial function $ u(x, t)=G(|x|, t)$, then read $ $ G_t(r, t)-G_r(r, t)^2G_{rr}(r, t)=0.$$ Also, if $ u$ is a sub / supersolution to, then $ -u$ is a\
super / subsolution.
The following definition is found in [ @Bjorn2015 ]:
\[def: stronk\_barrier\ ] lease $ \zeta_0 =( x_0, t_0) \in \partial \Omega$. A family of functions $ \{w_j\}_j $ in $ \Omega$ is called a * barrier family * in $ \Omega$ at the point $ \zeta_0 $ if, for every $ j$,
1. $ w_j>0 $ and $ w_j$ a supers | 1+\fraf{\epsilon}{2}.$$ With this choict, $\delta_0$ in depends only mn $\epsjlon$. Thur we have to verify that $$\lefv(\frax{1}{\log|\lig{|t|}|}+|c|\right)^{\frac{1}{k}} \leq |\log|g||^{\frac{\epsipon}{2}},$$ but rhis ibviously iklds fov smamp $|t|$ vmnce the left-hakd side is tounded. Hence $\diudz{\Omega} \subset \Omega$ and $(0,0)$ is an irregtlar boinfary point for $\Omebw$ as well.
This implies that the upper Psrron smlution $\overlone{H}_f$ does not always attaln tje boundary values at $(0, 0)$.
Barriet Faiulies {#sec:appx}
================
In this Stccion, a sub/sopersolution is a viscosity sub/sjpersplution of $$\lqbep{gq:infinity_heet}
u_t -\Dvlta_{\infty}u = u_b -\sum_{i, b=1}^{n}u_{x_i}u_{x_ox_j}u_{x_j} =0,$$ the usmal, nmn-nirmalised $\infty$- heat xquation. Further, the upper ang mower classes $\matycql{U}_f$ dnd $\kathzql{L}_w$, ahd tge Uppfr end Lower Psrron solutuons $\overline{H}_f$ and $\ugeerline{H}_f$, now also weser to. As before, the boundary function $x$ ia assumed to be continuius.
Note that if $u$ is w radial sunction $u(x, t)=G(|x|, t)$, then reads $$G_t(r, t)-G_r(r, t)^2G_{rr}(r, t)=0.$$ Alsm, if $n$ ks c sub/sjpegsolution to, then $-u$ is a\
super/subsolution.
The fjmlpwpng definition is found in [@Bkogn2015]:
\[cgf:stronk\_barriet\] Let $\zzfa_0 =(x_0, t_0) \in \partial \Omfga$. A fwmily of functyons $\{w_j\}_j $ in $\Omega$ is called a *varrier famijt* in $\Omega$ at the 'oint $\zeta_0$ iy, for gvery $k$,
1. $w_j>0$ and $w_j$ a supers | 1+\frac{\epsilon}{2}.$$ With this choice, $\delta_0$ in depends $\epsilon$. we have verify that $$\left(\frac{1}{\log|\log{|t|}|}+|c|\right)^{\frac{1}{k}} holds small $|t|$ since left-hand side is Hence $\tilde{\Omega} \subset \Omega$ and $(0,0)$ an irregular boundary point for $\Omega$ as well. This implies that the upper solution $\overline{H}_f$ does not always attain the boundary values at $(0, 0)$. Barrier {#sec:appa} In Section, sub/supersolution is a viscosity sub/supersolution of $$\label{eq:infinity_heat} u_t -\Delta_{\infty}u = u_t -\sum_{i, j=1}^{n}u_{x_i}u_{x_ix_j}u_{x_j} =0,$$ the usual, $\infty$- heat equation. Further, the upper and lower $\mathcal{U}_f$ and $\mathcal{L}_f$, and Upper and Lower Perron solutions and now also to. before, boundary function $f$ assumed to be continuous. Note that if $u$ is a radial function $u(x, t)=G(|x|, t)$, then reads t)-G_r(r, t)^2G_{rr}(r, if $u$ a to, $-u$ is a\ following definition is found in [@Bjorn2015]: =(x_0, t_0) \in \partial \Omega$. A family of $\{w_j\}_j $ $\Omega$ is called a *barrier family* $\Omega$ at the point $\zeta_0$ if, for every 1. $w_j>0$ and $w_j$ a supers | 1+\frac{\epsilon}{2}.$$ With this choice, $\Delta_0$ in depEnds oNly On $\ePsIlon$. thus We have to verify THat $$\lEft(\frac{1}{\log|\log{|t|}|}+|c|\right)^{\frAc{1}{k}} \leQ |\lOG|t||^{\frAC{\ePsiloN}{2}},$$ but thiS ObVIOusLy HoLds FoR SmAll $|t|$ sIncE the lefT-hand side iS boUnDed. Hence $\tildE{\omEga} \subset \OMegA$ and $(0,0)$ is an irreGulAr bounDaRy pOInt foR $\OmEga$ as Well.
ThIS impliEs that the UpPEr PerrON solutiON $\OvErliNe{H}_f$ does not always ATtAIn the boundary vAlues aT $(0, 0)$.
BARrIER FaMilIes {#sec:appa}
================
in This SECtion, a sUB/sUPERsoLUtion is a viscoSity sub/supeRSolUtion oF $$\lAbeL{Eq:infiNity_hEaT}
U_t -\DElta_{\infty}u = u_T -\sum_{I, j=1}^{n}u_{x_i}u_{x_iX_j}u_{x_j} =0,$$ tHE usual, nON-normalIsed $\inFty$- HeaT equATiOn. furThER, thE UpPer ANd lOwer clasSeS $\mAthcaL{U}_f$ aND $\MAThcaL{L}_f$, And tHe UppEr and Lower PerRon SoluTIonS $\overLine{H}_F$ and $\UnDerliNe{H}_f$, noW also ReFer to. As before, thE bouNdary funcTioN $f$ Is aSsUmed tO Be contInuOus.
note thaT if $u$ is a RAdiAl FUNCtIon $u(x, t)=G(|x|, t)$, then reads $$g_t(R, T)-g_r(R, t)^2G_{rr}(r, t)=0.$$ ALso, if $u$ IS a SuB/SupersolUtIon To, thEN $-U$ is a\
sUper/SUbSolution.
the folLOwInG definiTiOn is foUnD in [@bjoRn2015]:
\[def:STronK\_barriEr\] Let $\zetA_0 =(x_0, t_0) \in \PArtial \Omega$. A faMIly of functionS $\{W_j\}_J $ IN $\OMEga$ iS caLled a *barrieR famILy* in $\omegA$ At The POint $\zEta_0$ if, FoR EvERy $j$,
1. $w_j>0$ and $w_j$ a supers | 1+\frac{\epsilon}{2}.$$ Wi th this ch oice, $\ del ta _0$in d epends only on $\ep silon$. Thus we have t o ver if y tha t $ $\lef t(\frac { 1} { \ log |\ lo g{| t| } |} +|c|\ rig ht)^{\f rac{1}{k}} \l eq |\log|t||^{ \ fr ac{\epsilo n}{ 2}},$$ but t his obvio us lyh oldsfor smal l $|t| $ since the left -h a nd sid e is bou n d ed . He nce $\tilde{\Omeg a }\ subset \Omega$ and $ (0 , 0) $ isanirregularbo undar y pointf or $ \ Ome g a$ as well.
This implie s th at the u ppe r Perro n sol ut i on$\overline{ H}_f $ does no t alwa y s attai n the bo undary va lue s at $( 0, 0) $.
Ba r ri erF ami lies {#s ec :a ppa}==== = = = = ==== === =
I n thi s Section, asub /sup e rso lutio n isa vi sc osity sub/s upers ol ution of $$\lab el{e q:infinit y_h ea t}u_ t -\D e lta_{\ inf ty} u = u_t -\sum_ { i,j= 1 } ^ {n }u_{x_i}u_{x_ix_j} u_ { x _j } =0,$$the us u al ,n on-norma li sed $\i n f ty$-heat eq uation.Furthe r ,th e upper a nd low er cl ass es $\ m athc al{U}_ f$ and $ \math c al{L}_f$, andt he Upper andL ow e r P e rron so lutions $\o verl i ne{H }_f$ an d $ \ under line{ H} _ f$ , now also refer to. A s befo re, t he boundary f unction $f $ i s assume d to be continuous.
N ote t hat if $u$ is a rad ial f unction$u(x, t)= G ( |x|, t)$ , t hen re ads $ $G _t(r, t)-G_r( r , t)^ 2G _{rr}(r , t )=0.$$Als o,if$u$ i s a sub/s upersolu ti on t o, th en $- u $ is a\su per /s ubs oluti o n.
Th e fol lowi ng d e fin ition i s f o u nd i n[@ Bjor n20 15 ]:
\ [def : str onk\_ba rrier\] L et$ \zet a_ 0=(x_0,t_0) \in \par ti al \Omega$ .A f amilyo f functio ns $\{w_j\}_j $ in $\Om e ga$ iscal led a *ba rrier fam ily * in $ \Om e ga$ at the p oint$\ zet a _ 0$ if , fo r e ve ry $j$,
1 . $w _j>0$ a nd $ w_j$ asupers | 1+\frac{\epsilon}{2}.$$ With_this choice,_$\delta_0$ in depends only_on $\epsilon$._Thus_we have_to_verify that $$\left(\frac{1}{\log|\log{|t|}|}+|c|\right)^{\frac{1}{k}}_\leq |\log|t||^{\frac{\epsilon}{2}},$$ but_this obviously holds for_small $|t|$ since_the_left-hand side is bounded. Hence $\tilde{\Omega} \subset \Omega$ and $(0,0)$ is an irregular boundary_point_for $\Omega$_as_well.
This_implies that the upper Perron_solution $\overline{H}_f$ does not always_attain the_boundary values at $(0, 0)$.
Barrier Families {#sec:appa}
================
In this_Section,_a sub/supersolution is_a viscosity sub/supersolution of $$\label{eq:infinity_heat}
u_t -\Delta_{\infty}u = u_t -\sum_{i,_j=1}^{n}u_{x_i}u_{x_ix_j}u_{x_j} =0,$$ the usual, non-normalised $\infty$-_heat equation. Further,_the_upper_and lower classes $\mathcal{U}_f$_and $\mathcal{L}_f$, and the Upper and_Lower Perron solutions $\overline{H}_f$ and $\underline{H}_f$,_now also refer to. As before, the_boundary function $f$ is assumed to_be continuous.
Note that if $u$_is a_radial function $u(x, t)=G(|x|, t)$,_then reads $$G_t(r,_t)-G_r(r, t)^2G_{rr}(r,_t)=0.$$ Also, if_$u$ is a sub/supersolution to, then_$-u$ is a\
super/subsolution.
The_following definition is found in [@Bjorn2015]:
\[def:stronk\_barrier\]_Let_$\zeta_0 =(x_0, t_0)_\in_\partial_\Omega$. A_family of functions_$\{w_j\}_j_$ in_$\Omega$_is called a *barrier family* in_$\Omega$_at the point $\zeta_0$ if, for every_$j$,
1. $w_j>0$ and_$w_j$_a supers |
)\to (E,V)\to (E_2,V_2) \to 0$$ in which
- $(E,V)$ is ${\alpha}^-_i$-stable and of type ${(n,d,k)}$,
- ${{\operatorname{rk}}}(E_1)=n_1$ and $\dim(V_1)={\lambda}n_1$,
- $\mu_{{\alpha}_i}(E_1,V_1)=\mu_{{\alpha}_i}(E_2,V_2)=\mu_{{\alpha}_i}(E,V)$,
- $(E_1,V_1)$ and $(E_2,V_2)$ are both ${\alpha}_i^-$-stable,
- $\dim(V_1)$ and ${{\operatorname{rk}}}(E_1)$ satisfy the min-min criteria given in [(c)]{} and [(d)]{} of Lemma \[lem:vicente\](ii).
Define $$W^+({\alpha}_i,n,d,k)=\bigsqcup_{{\lambda}<\frac{k}{n},\,
n_1<n} W^+({\alpha}_i, {\lambda}, n_1; n,d,k),$$ $$W^-({\alpha}_i,n,d,k)=\bigsqcup_{{\lambda}>\frac{k}{n},\, n_1<n}
W^-({\alpha}_i, {\lambda}, n_1; n,d,k).$$ We abbreviate these to $W^+_i$ and $W^-_i$ whenever possible.
\[lem:wi\] Fix ${(n,d,k)}$ and also ${\alpha}_i$. Then each set $W^{\pm}_i({\lambda},n_1)$ is contained in a family of dimension bounded above by
$$\begin{aligned}
w^{\pm}_i({\lambda},n_1)&=
\dim G({\alpha}_i^{\pm};n_1,d_1,k_1)&\mbox{} +
\dim G({\alpha}_i^{\pm};n_2,d_2,k_2)\\
& & \mbox{} + \max{\dim {\operatorname{Ext}}^1((E_2,V_2),(E_1,V_1))}-1.
\end{aligned}$$
Here $n=n_1 | ) \to (E, V)\to (E_2,V_2) \to 0$$ in which
- $ (E, V)$ is $ { \alpha}^-_i$-stable and of type $ { (n, d, k)}$,
- $ { { \operatorname{rk}}}(E_1)=n_1 $ and $ \dim(V_1)={\lambda}n_1 $,
- $ \mu_{{\alpha}_i}(E_1,V_1)=\mu_{{\alpha}_i}(E_2,V_2)=\mu_{{\alpha}_i}(E, V)$,
- $ (E_1,V_1)$ and $ (E_2,V_2)$ are both $ { \alpha}_i^-$-stable,
- $ \dim(V_1)$ and $ { { \operatorname{rk}}}(E_1)$ satisfy the min - min criteria give in [ (coulomb) ] { } and [ (d) ] { } of Lemma \[lem: vicente\](ii).
Define $ $ W^+({\alpha}_i, n, d, k)=\bigsqcup_{{\lambda}<\frac{k}{n},\,
n_1 < n } W^+({\alpha}_i, { \lambda }, n_1; n, d, k),$$ $ $ W^-({\alpha}_i, n, d, k)=\bigsqcup_{{\lambda}>\frac{k}{n},\, n_1 < n }
W^-({\alpha}_i, { \lambda }, n_1; n, d, k).$$ We abridge these to $ W^+_i$ and $ W^-_i$ whenever possible.
\[lem: wi\ ] Fix $ { (n, d, k)}$ and besides $ { \alpha}_i$. Then each set $ W^{\pm}_i({\lambda},n_1)$ is contained in a class of dimension bounded above by
$ $ \begin{aligned }
w^{\pm}_i({\lambda},n_1)&=
\dim G({\alpha}_i^{\pm};n_1,d_1,k_1)&\mbox { } +
\dim G({\alpha}_i^{\pm};n_2,d_2,k_2)\\
& & \mbox { } + \max{\dim { \operatorname{Ext}}^1((E_2,V_2),(E_1,V_1))}-1.
\end{aligned}$$
Here $ n = n_1 | )\to (F,V)\to (E_2,V_2) \to 0$$ in which
- $(E,Y)$ is ${\alpha}^-_i$-stablg qnd of type ${(h,d,k)}$,
- ${{\opdratorname{rk}}}(E_1)=n_1$ and $\dim(V_1)={\lambde}n_1$,
- $\mu_{{\alkka}_i}(E_1,V_1)=\mu_{{\alpha}_i}(E_2,V_2)=\mu_{{\alpfa}_i}(E,V)$,
- $(E_1,N_1)$ and $(E_2,V_2)$ qre uoth ${\alpha}_i^-$-stablx,
- $\dim(V_1)$ and ${{\olcratoxneme{rk}}}(E_1)$ satisfy jhe min-min csiteria given hn [(c)]{} and [(d)]{} of Lemma \[lem:vicente\](ii).
Define $$R^+({\alpha}_i,m,d,n)=\bigsqcup_{{\lambdw}<\frab{k}{g},\,
n_1<n} W^+({\alpha}_i, {\lambda}, n_1; n,d,k),$$ $$W^-({\alpha}_i,n,d,k)=\bifsqcup_{{\lembda}>\frac{k}{n},\, n_1<n}
E^-({\alpha}_i, {\lambda}, n_1; n,d,k).$$ We abhrevlate these to $W^+_i$ ajd $W^-_i$ whenecer [issible.
\[lem:wi\] Fix ${(n,d,k)}$ and also ${\alphz}_i$. Then each set $W^{\pm}_i({\lambda},n_1)$ is contcined in a danilj of dimensiin bolnded above bn
$$\negin{anigned}
e^{\pm}_i({\lambda},n_1)&=
\dik G({\elphq}_i^{\pm};n_1,d_1,k_1)&\mbox{} +
\dim G({\alpia}_i^{\pm};n_2,d_2,k_2)\\
& & \mbox{} + \max{\qim {\operdtkrname{Ext}}^1((E_2,V_2),(E_1,V_1))}-1.
\end{qlugned}$$
Vere $n=n_1 | )\to (E,V)\to (E_2,V_2) \to 0$$ in which is and of ${(n,d,k)}$, - ${{\operatorname{rk}}}(E_1)=n_1$ $(E_1,V_1)$ $(E_2,V_2)$ are both - $\dim(V_1)$ and satisfy the min-min criteria given in and [(d)]{} of Lemma \[lem:vicente\](ii). Define $$W^+({\alpha}_i,n,d,k)=\bigsqcup_{{\lambda}<\frac{k}{n},\, n_1<n} W^+({\alpha}_i, {\lambda}, n_1; n,d,k),$$ $$W^-({\alpha}_i,n,d,k)=\bigsqcup_{{\lambda}>\frac{k}{n},\, W^-({\alpha}_i, {\lambda}, n_1; n,d,k).$$ We abbreviate these to $W^+_i$ and $W^-_i$ whenever possible. Fix and ${\alpha}_i$. each set $W^{\pm}_i({\lambda},n_1)$ is contained in a family of dimension bounded above by $$\begin{aligned} w^{\pm}_i({\lambda},n_1)&= \dim + \dim G({\alpha}_i^{\pm};n_2,d_2,k_2)\\ & & \mbox{} + \max{\dim \end{aligned}$$ Here $n=n_1 | )\to (E,V)\to (E_2,V_2) \to 0$$ in which
- $(E,V)$ is ${\alphA}^-_i$-stable anD of tyPe ${(n,D,k)}$,
- ${{\oPeRatoRnamE{rk}}}(E_1)=n_1$ and $\dim(V_1)={\laMBda}n_1$,
- $\Mu_{{\alpha}_i}(E_1,V_1)=\mu_{{\alpha}_i}(E_2,V_2)=\mu_{{\Alpha}_I}(E,v)$,
- $(e_1,V_1)$ anD $(e_2,V_2)$ Are boTh ${\alpha}_I^-$-StABLe,
- $\dIm(v_1)$ aNd ${{\oPeRAtOrnamE{rk}}}(e_1)$ satisfY the min-min CriTeRia given in [(c)]{} aND [(d)]{} Of Lemma \[lem:VicEnte\](ii).
Define $$w^+({\alPha}_i,n,d,K)=\bIgsQCup_{{\laMbdA}<\frac{K}{n},\,
n_1<n} W^+({\aLPha}_i, {\laMbda}, n_1; n,d,k),$$ $$W^-({\AlPHa}_i,n,d,k)=\BIgsqcup_{{\LAMbDa}>\frAc{k}{n},\, n_1<n}
W^-({\alpha}_i, {\lambDA}, n_1; N,D,k).$$ We abbreviate These tO $W^+_I$ AnD $w^-_I$ whEneVer possiblE.
\[lEm:wi\] FIX ${(n,d,k)}$ and ALsO ${\ALPha}_I$. then each set $W^{\pM}_i({\lambda},n_1)$ is COntAined iN a FamILy of diMensiOn BOunDed above by
$$\bEgin{Aligned}
w^{\pM}_i({\lambDA},n_1)&=
\dim G({\aLPha}_i^{\pm};n_1,D_1,k_1)&\mbox{} +
\Dim g({\alPha}_i^{\PM};n_2,D_2,k_2)\\
& & \MboX{} + \mAX{\diM {\OpEraTOrnAme{Ext}}^1((E_2,V_2),(e_1,V_1))}-1.
\EnD{aligNed}$$
HERE $N=N_1 | )\to (E,V)\to (E_2,V_2) \t o 0$$ in w hich
- $ (E ,V)$ is${\alpha}^-_i$ - stab le and of type ${(n,d, k)}$,
- $ { {\ opera torname { rk } } }(E _1 )= n_1 $a nd $\di m(V _1)={\l ambda}n_1$ ,
- $\mu_{{\al p ha }_i}(E_1,V _1) =\mu_{{\alph a}_ i}(E_2 ,V _2) = \mu_{ {\a lpha} _i}(E, V )$,
- $(E_1, V_ 1 )$ and $(E_2,V _ 2 )$ are both ${\alpha}_i ^ -$ - stable,
- $ \dim(V _1 ) $a n d $ {{\ operatorna me {rk}} } (E_1)$s at i s f y t h e min-min cri teria given in[(c)]{ }and [(d)]{ } ofLe m ma\[lem:vicen te\] (ii).
De fine $ $ W^+({\a l pha}_i, n,d,k) =\b igs qcup _ {{ \l amb da } <\f r ac {k} { n}, \,
n_1< n} W ^+({\ alph a } _ i , {\ lam bda} , n_1 ; n,d,k),$$ $ $W^ -({\ a lph a}_i, n,d,k )=\b ig sqcup _{{\la mbda} >\ frac{k}{n},\, n _1<n }
W^-({\ alp ha }_i ,{\lam b da}, n _1; n, d,k).$$ We abb r evi at e t he se to $W^+_i$ and$W ^ - _i $ whenev er pos s ib le .
\[lem: wi \]Fix$ { (n,d, k)}$ an d also $ {\alph a }_ i$ . Thenea ch set $ W^{ \pm }_i({ \ lamb da},n_ 1)$ is c ontai n ed in a family of dimensionb ou n d ed abov e b y
$$\begin {ali g ned}
w^ { \p m}_ i ({\la mbda} ,n _ 1) & =
\dim G({\alpha}_ i^ {\pm}; n_1,d _1,k_1)&\mbox {} +
\dim G ( {\alpha} _i^{ \ pm } ;n_2,d_2,k_2)\ \
&& \mbox{}+ \max{\d im { \operato rname{Ext } } ^1((E_2, V_2 ),( E_1 ,V_ 1 ) )} -1.
\end{ali g n ed}$ $
Here $ n=n _1 | )\to (E,V)\to_(E_2,V_2) \to_0$$ in which
- _ $(E,V)$_is_${\alpha}^-_i$-stable and_of_type ${(n,d,k)}$,
- _ ${{\operatorname{rk}}}(E_1)=n_1$ and_$\dim(V_1)={\lambda}n_1$,
- $\mu_{{\alpha}_i}(E_1,V_1)=\mu_{{\alpha}_i}(E_2,V_2)=\mu_{{\alpha}_i}(E,V)$,
-_ $(E_1,V_1)$_and_$(E_2,V_2)$ are both ${\alpha}_i^-$-stable,
- $\dim(V_1)$ and ${{\operatorname{rk}}}(E_1)$ satisfy the min-min criteria given_in_[(c)]{} and_[(d)]{}_of_Lemma \[lem:vicente\](ii).
Define $$W^+({\alpha}_i,n,d,k)=\bigsqcup_{{\lambda}<\frac{k}{n},\,
n_1<n} W^+({\alpha}_i,_{\lambda}, n_1; n,d,k),$$ $$W^-({\alpha}_i,n,d,k)=\bigsqcup_{{\lambda}>\frac{k}{n},\, n_1<n}
_W^-({\alpha}_i, {\lambda},_n_1; n,d,k).$$ We abbreviate these to $W^+_i$ and_$W^-_i$_whenever possible.
\[lem:wi\] Fix_${(n,d,k)}$ and also ${\alpha}_i$. Then each set $W^{\pm}_i({\lambda},n_1)$ is_contained in a family of dimension_bounded above by
$$\begin{aligned}
_w^{\pm}_i({\lambda},n_1)&=
_\dim_G({\alpha}_i^{\pm};n_1,d_1,k_1)&\mbox{} +
\dim G({\alpha}_i^{\pm};n_2,d_2,k_2)\\
_& & \mbox{} + \max{\dim _{\operatorname{Ext}}^1((E_2,V_2),(E_1,V_1))}-1.
\end{aligned}$$
Here $n=n_1 |
|\,,$$ we may write $$[\,f\,]_* =\,\int_0^R \,\sup_{ \,x \in\,{{\overline{\Omega}}}}\, {{\omega}}_f(x;\,r)\,
\,\frac{dr}{r}\,.\label{catriz}$$ So, together with $\,C_*({{\overline{\Omega}}})\,,$ we have considered a functional space $\,B_*({{\overline{\Omega}}})\,$ obtained by commuting *integral* and *sup* operators in the right hand side of definition : For each $\,f \in\,C({{\overline{\Omega}}})\,,$ we defined the semi-norm $$\langle\,f\,\rangle_* = \,\sup_{ \,x \in\,{{\overline{\Omega}}}}\, \int_0^R \,
{{\omega}}_f(x;\,r)\,
\,\frac{dr}{r}\,,\label{seis-bis}$$ and a related functional space$\, B_*({{\overline{\Omega}}})\,.$ We have shown that the inclusion $ \,C_*({{\overline{\Omega}}}) \subset\,B_*({{\overline{\Omega}}})\,$ is proper, by constructing strongly oscillating functions which belong to $\,B_*({{\overline{\Omega}}})\,$ but not to $\,C_*({{\overline{\Omega}}})\,.$ This construction was recently published in reference [@BV-LMS], Proposition 1.7.1. Furthermore, in \[BVUN\], we have shown that Theorem \[laplaces\] and similar results hold in a weaker form, for data $\,f
\in\,B_*({{\overline{\Omega}}})\,$, by proving that the first order derivatives of the solution $\,u\,$ are Lipschitz continuous in $\,{{\overline{\Omega}}}\,.$ The proof is published in reference [@BV-LMS], actually for data in a functional space $\,D_*({{\overline{\Omega}}})\,$ containing $\,B_*({{\overline{\Omega}}})\,$. See Theorem 1.3.1 in [@BV-LMS]. A similar extension holds for the Stokes problem, as shown in reference [@BVJP], Theorem 6.1, where we have proved that if $\,{\boldsymbol{f}}\in \,{\boldsymbol{D}}_*({{\overline{\Omega}}})\,,$ then the solution $\,({\boldsymbol{u}},\,p)\,$ of problem satisfies the estimate $\, \|\,{\boldsymbol{u}}\,\|_{1,\,1}+\,\|\,p\,\|_{0,\, | |\,,$$ we may write $ $ [ \,f\, ] _ * = \,\int_0^R \,\sup _ { \,x \in\,{{\overline{\Omega}}}}\, { { \omega}}_f(x;\,r)\,
\,\frac{dr}{r}\,.\label{catriz}$$ So, together with $ \,C_*({{\overline{\Omega}}})\,,$ we have considered a running quad $ \,B_*({{\overline{\Omega}}})\,$ obtained by commuting * built-in * and * sup * hustler in the right hired hand side of definition: For each $ \,f \in\,C({{\overline{\Omega}}})\,,$ we defined the semifinal - norm $ $ \langle\,f\,\rangle _ * = \,\sup _ { \,x \in\,{{\overline{\Omega}}}}\, \int_0^R \,
{ { \omega}}_f(x;\,r)\,
\,\frac{dr}{r}\,,\label{seis - bis}$$ and a related functional space$\, B_*({{\overline{\Omega}}})\,.$ We have shown that the inclusion body $ \,C_*({{\overline{\Omega } } }) \subset\,B_*({{\overline{\Omega}}})\,$ is proper, by constructing strongly oscillating functions which belong to to $ \,B_*({{\overline{\Omega}}})\,$ but not to $ \,C_*({{\overline{\Omega}}})\,.$ This construction was recently published in reference point [ @BV - LMS ], Proposition 1.7.1. Furthermore, in \[BVUN\ ], we have shown that Theorem \[laplaces\ ] and alike results hold in a weaker shape, for data $ \,f
\in\,B_*({{\overline{\Omega}}})\,$, by proving that the first order derived function of the solution $ \,u\,$ are Lipschitz continuous in $ \,{{\overline{\Omega}}}\,.$ The proof is published in reference [ @BV - LMS ], actually for data in a functional quad $ \,D_*({{\overline{\Omega}}})\,$ containing $ \,B_*({{\overline{\Omega}}})\,$. See Theorem 1.3.1 in [ @BV - LMS ]. A similar extension holds for the Stokes problem, as shown in reference [ @BVJP ], Theorem 6.1, where we have proved that if $ \,{\boldsymbol{f}}\in \,{\boldsymbol{D}}_*({{\overline{\Omega}}})\,,$ then the solution $ \,({\boldsymbol{u}},\,p)\,$ of problem satisfies the estimate $ \, \|\,{\boldsymbol{u}}\,\|_{1,\,1}+\,\|\,p\,\|_{0,\, | |\,,$$ we may write $$[\,f\,]_* =\,\int_0^R \,\sup_{ \,x \in\,{{\overline{\Omegc}}}}\, {{\omega}}_h(x;\,r)\,
\,\frac{sr}{r}\,.\label{zatriz}$$ So, together with $\,C_*({{\oveclinw{\Omegq}}})\,,$ we have considered a functionwl space $\,B_*({{\ovtrline{\Omega}}})\,$ obtaiisd by commutihn *intzgcal* and *sup* opetators in tha right hand shdd lf definition : For each $\,f \in\,C({{\overlige{\Omega}}})\,,$ wf defined the femi-mjrm $$\mangle\,f\,\rangle_* = \,\sup_{ \,x \in\,{{\overline{\Omefa}}}}\, \int_0^R \,
{{\omega}}_f(x;\,r)\,
\,\frac{cr}{r}\,,\label{seis-bis}$$ and a relahed vunctional space$\, B_*({{\lverline{\Ometa}}})\,.$ Wq have shown ghat the inclusion $ \,C_*({{\oberline{\Omega}}}) \subset\,B_*({{\overline{\Omeea}}})\,$ is proper, by cinshtucting stroigly ofcillating fmmctionv which belong to $\,B_*({{\ovcrlinx{\Ometa}}})\,$ but not to $\,C_*({{\overliie{\Omega}}})\,.$ This construstion was rzcently published in eederenwe [@BE-LMS], Prooosjtmon 1.7.1. Furtjerjore, in \[BVHN\], we have whown that Theorem \[ka[oaces\] and simjlar rqstlts hold in a weaker form, for data $\,f
\in\,T_*({{\ovsrline{\Omega}}})\,$, by proving rhat the first order ferivativqs of the solution $\,u\,$ are Lipschitz continuous in $\,{{\mverlmnd{\Omtgq}}}\,.$ The prlof is published in reference [@BV-LMS], actually skr dsta in a functlonal space $\,D_*({{\overlonf{\Okgga}}})\,$ containing $\,B_*({{\overlins{\Omega}}})\,$. See Theorem 1.3.1 in [@BV-JMS]. A similar txtenxion holds for the Stokes peoblem, as shjqn in reference [@BVLP], Theorem 6.1, chere ee hafe proved that if $\,{\boldsvmbol{f}}\jn \,{\boldsymbll{D}}_*({{\overlihd{\Omega}}})\,,$ then the rolltiot $\,({\boldsymbol{u}},\,p)\,$ of problem fatisfies the estimage $\, \|\,{\noldsyibol{u}}\,\|_{1,\,1}+\,\|\,p\,\|_{0,\, | |\,,$$ we may write $$[\,f\,]_* =\,\int_0^R \,\sup_{ {{\omega}}_f(x;\,r)\, So, together $\,C_*({{\overline{\Omega}}})\,,$ we have obtained commuting *integral* and operators in the hand side of definition : For $\,f \in\,C({{\overline{\Omega}}})\,,$ we defined the semi-norm $$\langle\,f\,\rangle_* = \,\sup_{ \,x \in\,{{\overline{\Omega}}}}\, \int_0^R \, \,\frac{dr}{r}\,,\label{seis-bis}$$ and a related functional space$\, B_*({{\overline{\Omega}}})\,.$ We have shown that the inclusion \,C_*({{\overline{\Omega}}}) is by strongly oscillating functions which belong to $\,B_*({{\overline{\Omega}}})\,$ but not to $\,C_*({{\overline{\Omega}}})\,.$ This construction was recently published reference [@BV-LMS], Proposition 1.7.1. Furthermore, in \[BVUN\], we shown that Theorem \[laplaces\] similar results hold in a form, data $\,f by that first order derivatives the solution $\,u\,$ are Lipschitz continuous in $\,{{\overline{\Omega}}}\,.$ The proof is published in reference [@BV-LMS], actually for in a $\,D_*({{\overline{\Omega}}})\,$ containing See 1.3.1 [@BV-LMS]. A similar for the Stokes problem, as shown Theorem 6.1, where we have proved that if \,{\boldsymbol{D}}_*({{\overline{\Omega}}})\,,$ then solution $\,({\boldsymbol{u}},\,p)\,$ of problem satisfies the $\, \|\,{\boldsymbol{u}}\,\|_{1,\,1}+\,\|\,p\,\|_{0,\, | |\,,$$ we may write $$[\,f\,]_* =\,\int_0^R \,\sup_{ \,x \in\,{{\overLine{\Omega}}}}\, {{\oMega}}_f(X;\,r)\,
\,\fRac{Dr}{R}\,.\labEl{caTriz}$$ So, together WIth $\,C_*({{\Overline{\Omega}}})\,,$ we have conSiderEd A FuncTIoNal spAce $\,B_*({{\oveRLiNE{\omeGa}}})\,$ ObTaiNeD By CommuTinG *integrAl* and *sup* opEraToRs in the right HAnD side of defIniTion : For each $\,f \In\,C({{\OverliNe{\omeGA}}})\,,$ we deFinEd the Semi-noRM $$\langlE\,f\,\rangle_* = \,\sUp_{ \,X \In\,{{\overLIne{\OmegA}}}}\, \INt_0^r \,
{{\omeGa}}_f(x;\,r)\,
\,\frac{dr}{r}\,,\label{SEiS-Bis}$$ and a related FunctiOnAL sPACe$\, B_*({{\OveRline{\Omega}}})\,.$ we Have sHOwn that THe INCLusIOn $ \,C_*({{\overline{\OmEga}}}) \subset\,B_*({{\oVErlIne{\OmeGa}}})\,$ Is pROper, by ConstRuCTinG strongly osCillAting funcTions wHIch beloNG to $\,B_*({{\oveRline{\OMegA}}})\,$ buT not TO $\,C_*({{\OvErlInE{\omeGA}}})\,.$ THis COnsTruction WaS rEcentLy puBLISHed iN reFereNce [@BV-lMS], PropositioN 1.7.1. FuRtheRMorE, in \[BVuN\], we hAve sHoWn thaT TheorEm \[lapLaCes\] and similar reSultS hold in a wEakEr ForM, fOr datA $\,F
\in\,B_*({{\ovErlIne{\omega}}})\,$, by Proving THat ThE FIRsT order derivatives oF tHE SoLution $\,u\,$ aRe LipsCHiTz COntinuouS iN $\,{{\ovErliNE{\omega}}}\,.$ the pROoF is publiShed in REfErEnce [@BV-LmS], ActualLy For DatA in a fUNctiOnal spAce $\,D_*({{\overLine{\OMEga}}})\,$ containing $\,B_*({{\OVerline{\Omega}}})\,$. SEE THEOrEM 1.3.1 in [@Bv-LMs]. A similar exTensIOn hoLds fOR tHe STOkes pRobleM, aS ShOWn in reference [@BVJP], ThEoRem 6.1, wheRe we hAve proved that If $\,{\boldsymbOL{F}}\In \,{\boldsyMbol{d}}_*({{\OvERline{\Omega}}})\,,$ then The soLution $\,({\boldSYmbol{u}},\,p)\,$ oF probLem satisFies the esTIMate $\, \|\,{\boldSymBol{U}}\,\|_{1,\,1}+\,\|\,p\,\|_{0,\, | |\,,$$ we may write $$[\,f \,]_* =\,\ int_0 ^R \, \s up_{ \,x \in\,{{\overl i ne{\ Omega}}}}\, {{\omega}} _f(x; \, r )\,\ ,\ frac{ dr}{r}\ , .\ l a bel {c at riz }$ $ S o, to get her wit h $\,C_*({ {\o ve rline{\Omega } }} )\,,$ we h ave considereda f unctio na l s p ace $ \,B _*({{ \overl i ne{\Om ega}}})\, $o btaine d by com m u ti ng * integral* and *su p *o perators in th e righ th an d sid e o f definiti on : Fo r each $ \ ,f \ i n\, C ({{\overline{ \Omega}}})\ , ,$we def in edt he sem i-nor m$ $\l angle\,f\,\ rang le_* = \, \sup_{ \,x \in \ ,{{\ove rline{ \Om ega }}}} \ ,\i nt_ 0^ R \, {{ \om e ga} }_f(x;\, r) \,
\,\f rac{ d r } { r}\, ,\l abel {seis -bis}$$ and a re late d fu nctio nal s pace $\ , B_* ({{\ov erlin e{ \Omega}}})\,.$We h ave shown th at th einclu s ion $\,C _*( {{\over line{\O m ega }} } ) \s ubset\,B_*({{\over li n e {\ Omega}}} )\,$ i s p ro p er, by c on str ucti n g stro ngly os cillatin g func t io ns whichbe long t o$\, B_* ({{\o v erli ne{\Om ega}}})\ ,$ bu t not to $\,C_* ( {{\overline{\ O me g a }} } )\,. $ T his constru ctio n was rec e nt lyp ublis hed i nr ef e rence [@BV-LMS], Pr op ositio n 1.7 .1. Furthermo re, in \[B V U N \], we h aves ho w n that Theorem \[la places\] a n d simila r res ults hol d in a we a k er form, fo r d ata $\ , f
\ in\,B_*({{\ov e r line {\ Omega}} })\ ,$, bypro vin g t hat t he firstorder de ri va ti ve s o f the solution $ \,u \, $ a re Li p schitz cont inuo us i n $\ ,{{\ove r li n e {\Om eg a} }}\, .$Th e pro of i s pu blished in refer enc e [@B V- LM S], act ually for dat ain a funct io nal space $ \,D_*({{ \overline{\Omega}}})\,$ contain ing $\,B _*({ {\overlin e{\ Omega} }}) \ ,$. Se e Theo rem 1 .3 .1i n [@BV - L MS ].Asimilar ex t e nsi on ho ld s fo r the S tokes problem, ass how n in referenc e [ @BVJ P ] ,The o re m 6. 1, whe r e we have proved that if $ \, { \b oldsymbol{ f }}\ in \,{\bo ldsymbo l{D}} _ *({{\ov erline{\O mega}}})\ ,, $ th e n th e solution $\,({\b oldsymbol { u}},\ , p) \,$ o f p roblem s ati sfies the e s tim ate $ \, \|\ ,{ \bolds ymbol {u }}\,\|_{ 1,\,1}+\,\|\,p\,\|_{0,\ , | |\,,$$ we_may write_$$[\,f\,]_* =\,\int_0^R \,\sup_{_\,x \in\,{{\overline{\Omega}}}}\,_{{\omega}}_f(x;\,r)\,
\,\frac{dr}{r}\,.\label{catriz}$$_So, together_with_$\,C_*({{\overline{\Omega}}})\,,$ we have_considered a functional_space $\,B_*({{\overline{\Omega}}})\,$ obtained by_commuting *integral* and_*sup*_operators in the right hand side of definition : For each $\,f \in\,C({{\overline{\Omega}}})\,,$ we_defined_the semi-norm_$$\langle\,f\,\rangle_*_=_\,\sup_{ \,x \in\,{{\overline{\Omega}}}}\, \int_0^R \,
{{\omega}}_f(x;\,r)\,
\,\frac{dr}{r}\,,\label{seis-bis}$$_and a related functional space$\,_B_*({{\overline{\Omega}}})\,.$ We_have shown that the inclusion $ \,C_*({{\overline{\Omega}}}) \subset\,B_*({{\overline{\Omega}}})\,$_is_proper, by constructing_strongly oscillating functions which belong to $\,B_*({{\overline{\Omega}}})\,$ but not_to $\,C_*({{\overline{\Omega}}})\,.$ This construction was recently_published in reference_[@BV-LMS],_Proposition_1.7.1. Furthermore, in \[BVUN\],_we have shown that Theorem \[laplaces\]_and similar results hold in a_weaker form, for data $\,f
\in\,B_*({{\overline{\Omega}}})\,$, by proving_that the first order derivatives of_the solution $\,u\,$ are Lipschitz_continuous in_$\,{{\overline{\Omega}}}\,.$ The proof is published_in reference [@BV-LMS],_actually for_data in a_functional space $\,D_*({{\overline{\Omega}}})\,$ containing $\,B_*({{\overline{\Omega}}})\,$. See_Theorem 1.3.1 in_[@BV-LMS]. A similar extension holds for_the_Stokes problem, as_shown_in_reference [@BVJP],_Theorem 6.1, where_we_have proved_that_if $\,{\boldsymbol{f}}\in \,{\boldsymbol{D}}_*({{\overline{\Omega}}})\,,$ then the solution_$\,({\boldsymbol{u}},\,p)\,$_of problem satisfies the estimate $\, \|\,{\boldsymbol{u}}\,\|_{1,\,1}+\,\|\,p\,\|_{0,\, |
_0$ is the normalization factor and $\Phi_{\rm Brink}({\bf S}_C,{\bf S}_1,{\bf S}_2,\cdots
{\bf S}_k)$ is the Brink wave function for the $C+k(^2n)$-cluster system consisting the core($C$) and $k$ dineutrons($^2n$) as, $$\Phi_{\rm Brink}({\bf S}_C,{\bf S}_1,{\bf S}_2,\cdots
{\bf S}_k)\equiv {\cal{A}}\left\{\phi^{\rm C}({\bf S}_C)\phi^{^2n}({\bf S}_1)
\phi^{^2n}({\bf S}_2)\cdots \phi^{^2n}({\bf S}_k)
\right \}.$$ Here, the wave function of the $i$th $^2n$, $\phi^{^2n}({\bf S}_i)$, is given by the $(0s)^2$ state localized around ${\bf S}_i$. ${\bf S}_C$ is the mean position of the center of mass motion of the core, and is chosen to be ${\bf S}_C=-\frac{2}{A}({\bf S}_1+{\bf S}_2+\cdots+{\bf S}_k)$. In heavy limit of the core mass $A$, this wave function is equivalent to the dineutron condensate wave function proposed by Horiuchi[@Horiuchi06]. In the present calculation for $^4{\rm He}+^2n+^2n$, the core $C$ is $^4$He, and the number of $^2n$ clusters is $k=2$. We assumed the $(0s)^4$ state of the core wave function, $\phi^{^4{\rm He}}$, and adopted the common size parameter $b=1/\sqrt(2 \nu(^8{\rm He})$ for the $^4{\rm He}$ and $^2n$ clusters. In the practical calculations, the 6-dimensional integrals for the coordinates, ${\bf S}_1$ and ${\bf S}_2$, are performed by taking mesh points on $(\theta_{12},|{\bf S}_1|,|{\bf S}_2|)$ and the total-angular-momentum projection ($\theta_{12}$ is the angle between ${\bf S}_1$ and ${\bf S}_2$).
=6. cm
In Fig. \[fig:bec\], we | _ 0 $ is the normalization factor and $ \Phi_{\rm Brink}({\bf S}_C,{\bf S}_1,{\bf S}_2,\cdots
{ \bf S}_k)$ is the Brink wave function for the $ C+k(^2n)$-cluster system consist the core($C$) and $ k$ dineutrons($^2n$) as, $ $ \Phi_{\rm Brink}({\bf S}_C,{\bf S}_1,{\bf S}_2,\cdots
{ \bf S}_k)\equiv { \cal{A}}\left\{\phi^{\rm C}({\bf S}_C)\phi^{^2n}({\bf S}_1)
\phi^{^2n}({\bf S}_2)\cdots \phi^{^2n}({\bf S}_k)
\right \}.$$ Here, the wave affair of the $ i$th $ ^2n$, $ \phi^{^2n}({\bf S}_i)$, is given by the $ (0s)^2 $ state localized around $ { \bf S}_i$. $ { \bf S}_C$ is the hateful position of the center of aggregate motion of the core, and is chosen to be $ { \bf S}_C=-\frac{2}{A}({\bf S}_1+{\bf S}_2+\cdots+{\bf S}_k)$. In fleshy limit of the core bulk $ A$, this wave function is equivalent to the dineutron condensate wave affair proposed by Horiuchi[@Horiuchi06 ]. In the present calculation for $ ^4{\rm He}+^2n+^2n$, the core $ C$ is $ ^4$He, and the number of $ ^2n$ clusters is $ k=2$. We assumed the $ (0s)^4 $ state of matter of the core wave function, $ \phi^{^4{\rm He}}$, and adopted the coarse size parameter $ b=1/\sqrt(2 \nu(^8{\rm He})$ for the $ ^4{\rm He}$ and $ ^2n$ clusters. In the practical calculations, the 6 - dimensional integral for the coordinates, $ { \bf S}_1 $ and $ { \bf S}_2 $, are do by taking mesh points on $ (\theta_{12},|{\bf S}_1|,|{\bf S}_2|)$ and the entire - angular - momentum projection ($ \theta_{12}$ is the angle between $ { \bf S}_1 $ and $ { \bf S}_2 $).
= 6. cm
In Fig. \[fig: bec\ ], we | _0$ is the normalization factov and $\Phi_{\rm Brink}({\bf S}_C,{\bh S}_1,{\bf S}_2,\ddots
{\bf R}_k)$ is the Brink wave functioi foe the $C+k(^2n)$-cluster system conristing tje core($C$) and $j$ dineutroia($^2n$) as, $$\Pmn_{\rm Bdlnk}({\bf W}_C,{\bf S}_1,{\bf S}_2,\cdotx
{\bf S}_k)\equie {\cal{A}}\left\{\phi^{\rk Z}({\by S}_C)\phi^{^2n}({\bf S}_1)
\phi^{^2n}({\bf S}_2)\cdots \phi^{^2n}({\bf S}_k)
\wight \}.$$ Nege, the wave fugctipg of nht $i$th $^2n$, $\phi^{^2n}({\bf S}_i)$, is given by the $(0s)^2$ statt localized around ${\bf S}_i$. ${\bf S}_C$ is the mean plsitlon of the center lf mass motuon jd the core, avd is chostn to be ${\bf A}_C=-\frac{2}{A}({\bf S}_1+{\bf S}_2+\cdots+{\bf S}_k)$. In hexvy lnmit of the cire kass $A$, this wave function is equivalant to yhe dineutron gondeisatw wave function propoved by Horiuchi[@Horyuchi06]. In dhz present calculation fir $^4{\rm He}+^2n+^2t$, thd cofe $D$ ms $^4$Ge, and thx number of $^2n$ clusters is $k=2$. We assumed tht $(0s)^4$ wtate of the dore wwvq function, $\phi^{^4{\rm He}}$, and adopted the comkon size parameter $b=1/\sqrt(2 \ny(^8{\rm He})$ for the $^4{\rm He}$ wnd $^2n$ clufters. In the practical calculations, the 6-dimensiondl inveeraos wir the coordinates, ${\bf S}_1$ and ${\bf S}_2$, are performed fg uakpng mesh points ok $(\theta_{12},|{\bf S}_1|,|{\bf S}_2|)$ anc hhr total-angular-oomentbj lrojection ($\theta_{12}$ id the aggle vetween ${\bs S}_1$ snd ${\bf S}_2$).
=6. cm
In Fig. \[fig:bec\], we | _0$ is the normalization factor and $\Phi_{\rm S}_1,{\bf {\bf S}_k)$ the Brink wave consisting core($C$) and $k$ as, $$\Phi_{\rm Brink}({\bf S}_1,{\bf S}_2,\cdots {\bf S}_k)\equiv {\cal{A}}\left\{\phi^{\rm C}({\bf S}_1) \phi^{^2n}({\bf S}_2)\cdots \phi^{^2n}({\bf S}_k) \right \}.$$ Here, the wave function of the $^2n$, $\phi^{^2n}({\bf S}_i)$, is given by the $(0s)^2$ state localized around ${\bf S}_i$. S}_C$ the position the center of mass motion of the core, and is chosen to be ${\bf S}_C=-\frac{2}{A}({\bf S}_1+{\bf S}_k)$. In heavy limit of the core mass this wave function is to the dineutron condensate wave proposed Horiuchi[@Horiuchi06]. In present for He}+^2n+^2n$, the core is $^4$He, and the number of $^2n$ clusters is $k=2$. We assumed the $(0s)^4$ state of the wave function, and adopted common parameter \nu(^8{\rm He})$ for He}$ and $^2n$ clusters. In the 6-dimensional integrals for the coordinates, ${\bf S}_1$ and S}_2$, are by taking mesh points on $(\theta_{12},|{\bf S}_2|)$ and the total-angular-momentum projection ($\theta_{12}$ is the between ${\bf S}_1$ and ${\bf S}_2$). =6. cm In Fig. \[fig:bec\], we | _0$ is the normalization factor aNd $\Phi_{\rm BriNk}({\bf S}_c,{\bf s}_1,{\bf s}_2,\cDots
{\Bf S}_k)$ Is the Brink wave FUnctIon for the $C+k(^2n)$-cluster sysTem coNsIStinG ThE core($c$) and $k$ diNEuTROns($^2N$) aS, $$\PHi_{\rM BRInK}({\bf S}_C,{\Bf S}_1,{\Bf S}_2,\cdotS
{\bf S}_k)\equiv {\Cal{a}}\lEft\{\phi^{\rm C}({\bf S}_c)\PhI^{^2n}({\bf S}_1)
\phi^{^2n}({\bF S}_2)\cDots \phi^{^2n}({\bf S}_k)
\RigHt \}.$$ Here, ThE waVE funcTioN of thE $i$th $^2n$, $\pHI^{^2n}({\bf S}_i)$, Is given by ThE $(0S)^2$ state LOcalizeD ARoUnd ${\bF S}_i$. ${\bf S}_C$ is the mean pOSiTIon of the center Of mass MoTIoN OF thE coRe, and is choSeN to be ${\BF S}_C=-\frac{2}{a}({\Bf s}_1+{\BF s}_2+\cdOTs+{\bf S}_k)$. In heavy Limit of the cORe mAss $A$, thIs WavE FunctiOn is eQuIValEnt to the dinEutrOn condensAte wavE FunctioN ProposeD by HorIucHi[@HOriuCHi06]. in The PrESenT CaLcuLAtiOn for $^4{\rm HE}+^2n+^2N$, tHe corE $C$ is $^4$hE, AND the NumBer oF $^2n$ cluSters is $k=2$. We assUmeD the $(0S)^4$ StaTe of tHe corE wavE fUnctiOn, $\phi^{^4{\rM He}}$, anD aDopted the common Size Parameter $B=1/\sqRt(2 \Nu(^8{\rM HE})$ for tHE $^4{\rm He}$ aNd $^2n$ CluSters. In The pracTIcaL cALCUlAtions, the 6-dimensionAl INTeGrals for The cooRDiNaTEs, ${\bf S}_1$ and ${\Bf s}_2$, arE perFORmed bY takINg Mesh poinTs on $(\thETa_{12},|{\Bf s}_1|,|{\bf S}_2|)$ and ThE total-AnGulAr-mOmentUM proJectioN ($\theta_{12}$ is The anGLe between ${\bf S}_1$ anD ${\Bf S}_2$).
=6. cm
In Fig. \[fig:BEc\], WE | _0$ is the normalization f actor and$\Phi _{\ rmBr ink} ({\b f S}_C,{\bf S} _ 1,{\ bf S}_2,\cdots
{\bf S} _k)$is theB ri nk wa ve func t io n for t he $C +k ( ^2 n)$-c lus ter sys tem consis tin gthe core($C$ ) a nd $k$ din eut rons($^2n$)as, $$\Ph i_ {\r m Brin k}( {\bfS}_C,{ \ bf S}_ 1,{\bf S} _2 , \cdots {\bf S} _ k )\ equi v {\cal{A}}\left\ { \p h i^{\rm C}({\bf S}_C) \p h i^ { ^ 2n} ({\ bf S}_1)
\ ph i^{^2 n }({\bfS }_ 2 ) \ cdo t s \phi^{^2n}( {\bf S}_k)\ri ght \} .$ $ H e re, th e wav ef unc tion of the $i$ th $^2n$, $\phi ^ {^2n}({ \ bf S}_i )$, is gi ven byt he $ (0s )^ 2 $ s t at e l o cal ized aro un d${\bf S}_ i $ . ${\b f S }_C$ is t he mean posit ion oft hecente r ofmass m otion of th e cor e, and is chosento b e ${\bf S }_C =- \fr ac {2}{A } ({\bfS}_ 1+{ \bf S}_ 2+\cdot s +{\ bf S } _k )$. In heavy limit o f th e core m ass $A $ ,th i s wave f un cti on i s equiv alen t t o the di neutro n c on densate w ave fu nc tio n p ropos e d by Horiu chi[@Hor iuchi 0 6]. In the pre s ent calculati o nf o r$ ^4{\ rmHe}+^2n+^2n $, t h e co re $ C $is$ ^4$He , and t h en umber of $^2n$ clus te rs is$k=2$ . We assumedthe $(0s)^ 4 $ state of the co r e wave functio n, $\ phi^{^4{\r m He}}$,and a dopted t he common s ize para met er$b= 1/\ s q rt (2 \nu(^8{\rm H e})$ f or the$^4 {\rm He }$and $^ 2n$ c lusters.In the p ra ct ic al ca lcula t ions, th e6-d im ens ional integr als f or t he c o ord inates, ${ \ b f S} _1 $and${\ bf S}_2 $, a r e p erforme d by taki ngm eshpo in ts on $ (\theta_{12}, |{ \bf S}_1|, |{ \bf S}_2| ) $ and the total-angular-momentum project ion ($\t heta _{12}$ is th e angl e b e tween${\bfS}_1$ a nd$ { \bf S } _ 2$ ).
= 6. cm
InF i g.\[fig :b ec\] , we | _0$ is_the normalization_factor and $\Phi_{\rm Brink}({\bf_S}_C,{\bf S}_1,{\bf_S}_2,\cdots
{\bf_S}_k)$ is_the_Brink wave function_for the $C+k(^2n)$-cluster_system consisting the core($C$)_and $k$ dineutrons($^2n$)_as,_$$\Phi_{\rm Brink}({\bf S}_C,{\bf S}_1,{\bf S}_2,\cdots
{\bf S}_k)\equiv {\cal{A}}\left\{\phi^{\rm C}({\bf S}_C)\phi^{^2n}({\bf S}_1)
\phi^{^2n}({\bf S}_2)\cdots \phi^{^2n}({\bf S}_k)
\right_\}.$$_Here, the_wave_function_of the $i$th $^2n$, $\phi^{^2n}({\bf_S}_i)$, is given by the_$(0s)^2$ state_localized around ${\bf S}_i$. ${\bf S}_C$ is the_mean_position of the_center of mass motion of the core, and is_chosen to be ${\bf S}_C=-\frac{2}{A}({\bf S}_1+{\bf_S}_2+\cdots+{\bf S}_k)$. In_heavy_limit_of the core mass_$A$, this wave function is equivalent_to the dineutron condensate wave function_proposed by Horiuchi[@Horiuchi06]. In the present calculation_for $^4{\rm He}+^2n+^2n$, the core $C$_is $^4$He, and the number_of $^2n$_clusters is $k=2$. We assumed_the $(0s)^4$ state_of the_core wave function,_$\phi^{^4{\rm He}}$, and adopted the common_size parameter $b=1/\sqrt(2_\nu(^8{\rm He})$ for the $^4{\rm He}$_and_$^2n$ clusters. In_the_practical_calculations, the_6-dimensional integrals for_the_coordinates, ${\bf_S}_1$_and ${\bf S}_2$, are performed by_taking_mesh points on $(\theta_{12},|{\bf S}_1|,|{\bf S}_2|)$ and_the total-angular-momentum projection ($\theta_{12}$_is_the angle between ${\bf_S}_1$ and ${\bf S}_2$).
=6. cm
In_Fig. \[fig:bec\], we |
_{\varepsilon})$ is optimal in energy. We have $$\lim_{{\varepsilon}\to 0} \frac{1}{|\log {\varepsilon}|^2}\int_{\Omega}W(\beta_{\varepsilon}) \, dx= \lim_{{\varepsilon}\to 0} \frac{1}{|\log {\varepsilon}|^2}
\int_{\Omega}W(|\log {\varepsilon}| \beta + \hat \beta_{\varepsilon})\, dx.$$ Since $\hat \beta_{\varepsilon}/|\log {\varepsilon}| {\rightharpoonup}0$ in $L^2({\Omega};{\mathbb{M}^{2\times 2}})$, taking into account also, we conclude $$\lim_{{\varepsilon}\to 0} \frac{1}{|\log {\varepsilon}|^2}\int_{\Omega}W(\beta_{\varepsilon})\,dx= \lim_{{\varepsilon}\to 0} \frac{1}{|\log {\varepsilon}|^2}
\left( \int_{{\Omega}} W(|\log {\varepsilon}| \beta)\,dx + \int_{{\Omega}} W(\hat\beta_{\varepsilon})\,dx \right) =
\int_{\Omega}(W(\beta) + {\varphi}(\xi))\,dx.$$ Finally, by the Lipschitz continuity of $\partial {\Omega}$, from and (\[Reps\]) we also deduce that $(\beta_{\varepsilon}/|\log
{\varepsilon}| - \beta) \cdot t$ tends to zero strongly in $H^{-\frac{1}{2}}(\partial {\Omega})$.
[*Step $2.$ The case $\mu := \sum_{l=1}^L \chi_{A_l} \xi_l \,dx$.*]{}
In this step we proof the $\Gamma$-limsup inequality in the case of $\mu$ locally constant, i.e., of the type $$\label{loco}
\mu:= \sum_{l=1}^L \chi_{A_l} \xi_l\,dx,$$ where $A_l$ are open subsets of ${\Omega}$ with Lipschitz continuous boundary and $\xi_i\in {\mathbb{M}}^{2\times 2}$. The construction of the recovery sequence is based on classical localization arguments in $\Gamma$-convergence and takes advantage of the previous step.
Let us set $\beta_l:= \beta {\mathop{\hbox{\vrule height 7pt width.5pt depth 0pt
\vrule height.5pt width 6pt depth 0pt}}\nolimits}A | _ { \varepsilon})$ is optimal in energy. We have $ $ \lim_{{\varepsilon}\to 0 } \frac{1}{|\log { \varepsilon}|^2}\int_{\Omega}W(\beta_{\varepsilon }) \, dx= \lim_{{\varepsilon}\to 0 } \frac{1}{|\log { \varepsilon}|^2 }
\int_{\Omega}W(|\log { \varepsilon}| \beta + \hat \beta_{\varepsilon})\, dx.$$ Since $ \hat \beta_{\varepsilon}/|\log { \varepsilon}| { \rightharpoonup}0 $ in $ L^2({\Omega};{\mathbb{M}^{2\times 2}})$, taking into explanation besides, we conclude $ $ \lim_{{\varepsilon}\to 0 } \frac{1}{|\log { \varepsilon}|^2}\int_{\Omega}W(\beta_{\varepsilon})\,dx= \lim_{{\varepsilon}\to 0 } \frac{1}{|\log { \varepsilon}|^2 }
\left (\int_{{\Omega } } W(|\log { \varepsilon}| \beta)\,dx + \int_{{\Omega } } W(\hat\beta_{\varepsilon})\,dx \right) =
\int_{\Omega}(W(\beta) + { \varphi}(\xi))\,dx.$$ Finally, by the Lipschitz continuity of $ \partial { \Omega}$, from and (\[Reps\ ]) we also deduce that $ (\beta_{\varepsilon}/|\log
{ \varepsilon}| - \beta) \cdot t$ tend to zero powerfully in $ H^{-\frac{1}{2}}(\partial { \Omega})$.
[ * Step $ 2.$ The case $ \mu: = \sum_{l=1}^L \chi_{A_l } \xi_l \,dx$. * ] { }
In this step we proofread the $ \Gamma$-limsup inequality in the character of $ \mu$ locally constant, i.e., of the character $ $ \label{loco }
\mu:= \sum_{l=1}^L \chi_{A_l } \xi_l\,dx,$$ where $ A_l$ are loose subsets of $ { \Omega}$ with Lipschitz continuous limit and $ \xi_i\in { \mathbb{M}}^{2\times 2}$. The construction of the recovery sequence is based on authoritative localization arguments in $ \Gamma$-convergence and bring advantage of the former footfall.
Let us set $ \beta_l:= \beta { \mathop{\hbox{\vrule height 7pt width.5pt astuteness 0pt
\vrule height.5pt width 6pt depth 0pt}}\nolimits}A | _{\varfpsilon})$ is optimal in entrgy. We have $$\lim_{{\vceepsilmn}\to 0} \rrac{1}{|\log {\xarepsilon}|^2}\int_{\Omega}W(\beta_{\varepdioon}) \, ex= \lim_{{\varepsilon}\to 0} \frxc{1}{|\log {\varvpsilon}|^2}
\inr_{\Omeja}W(|\log {\varepsiloi}| \beta + \mct \befw_{\varzpwilon})\, dx.$$ Since $\hat \beta_{\vdrepsilon}/|\log {\vdrdpdilon}| {\rightharpoonup}0$ in $L^2({\Omega};{\mathbf{M}^{2\times 2}})$, haking into acsounu ajso, sv gonclude $$\lim_{{\varepsilon}\to 0} \frac{1}{|\lof {\varepvilon}|^2}\int_{\Omega}E(\beta_{\varepsilon})\,dx= \lim_{{\varepdiloj}\to 0} \frac{1}{|\log {\varepdilon}|^2}
\left( \ibt_{{\Omqta}} W(|\log {\vareosilon}| \beta)\,dx + \int_{{\Omegz}} W(\hat\beta_{\varepsilon})\,dx \right) =
\ing_{\Omegc}(W(\beta) + {\varkku}(\xi))\,ff.$$ Finally, bb the Jipschitz cokninuity of $\paryial {\Omega}$, frok aid (\[Rwps\]) we also deduce thet $(\beta_{\varepsilon}/|\log
{\darepsilot}| - \beta) \cdot t$ tendw ro zeto stsongut iv $H^{-\rrec{1}{2}}(\pzrtial {\Omxga})$.
[*Step $2.$ Ths case $\mu := \wum_{l=1}^L \chi_{A_l} \xi_l \,dx$.*]{}
Im eyis step we pdoof tre $\Gamma$-limsup inequality in the case of $\ku$ mocally constant, i.e., of rhe type $$\label{loco}
\mu:= \dum_{l=1}^L \chi_{W_l} \xi_l\,dx,$$ where $A_l$ are open subsets of ${\Omega}$ with Nipsciigz eintinjiud boundary and $\xi_i\in {\mathbb{M}}^{2\times 2}$. The constrtdtook of the recovern sequence is basec ln slassical locxlizatnkn arguments in $\Gammw$-converdence and takef adfantage of the previous step.
Let us set $\yetq_l:= \beta {\mathop{\hbox{\rrule height 7pt eidth.5lt depth 0pt
\vrule height.5't widfh 6pt depth 0pt}}\nolimifr}A | _{\varepsilon})$ is optimal in energy. We have \frac{1}{|\log \, dx= 0} \frac{1}{|\log {\varepsilon}|^2} \beta_{\varepsilon})\, Since $\hat \beta_{\varepsilon}/|\log {\rightharpoonup}0$ in $L^2({\Omega};{\mathbb{M}^{2\times taking into account also, we conclude 0} \frac{1}{|\log {\varepsilon}|^2}\int_{\Omega}W(\beta_{\varepsilon})\,dx= \lim_{{\varepsilon}\to 0} \frac{1}{|\log {\varepsilon}|^2} \left( \int_{{\Omega}} W(|\log {\varepsilon}| \beta)\,dx + W(\hat\beta_{\varepsilon})\,dx \right) = \int_{\Omega}(W(\beta) + {\varphi}(\xi))\,dx.$$ Finally, by the Lipschitz continuity of $\partial from (\[Reps\]) also that $(\beta_{\varepsilon}/|\log {\varepsilon}| - \beta) \cdot t$ tends to zero strongly in $H^{-\frac{1}{2}}(\partial {\Omega})$. [*Step $2.$ case $\mu := \sum_{l=1}^L \chi_{A_l} \xi_l \,dx$.*]{} In step we proof the inequality in the case of locally i.e., of type \mu:= \chi_{A_l} \xi_l\,dx,$$ where are open subsets of ${\Omega}$ with Lipschitz continuous boundary and $\xi_i\in {\mathbb{M}}^{2\times 2}$. The construction of the sequence is classical localization in and advantage of the Let us set $\beta_l:= \beta {\mathop{\hbox{\vrule depth 0pt \vrule height.5pt width 6pt depth 0pt}}\nolimits}A | _{\varepsilon})$ is optimal in enerGy. We have $$\liM_{{\varePsiLon}\To 0} \Frac{1}{|\Log {\vArepsilon}|^2}\int_{\OmEGa}W(\bEta_{\varepsilon}) \, dx= \lim_{{\varePsiloN}\tO 0} \Frac{1}{|\LOg {\VarepSilon}|^2}
\inT_{\omEGA}W(|\lOg {\VaRepSiLOn}| \Beta + \hAt \bEta_{\varePsilon})\, dx.$$ SiNce $\HaT \beta_{\varepsiLOn}/|\Log {\varepsiLon}| {\RightharpoonUp}0$ iN $L^2({\OmegA};{\mAthBB{M}^{2\timEs 2}})$, tAking Into acCOunt alSo, we conclUdE $$\Lim_{{\varEPsilon}\tO 0} \FRaC{1}{|\log {\Varepsilon}|^2}\int_{\OmegA}w(\bETa_{\varepsilon})\,dx= \Lim_{{\varEpSIlON}\To 0} \fRac{1}{|\Log {\varepsiLoN}|^2}
\left( \INt_{{\Omega}} w(|\LoG {\VARepSIlon}| \beta)\,dx + \int_{{\omega}} W(\hat\beTA_{\vaRepsilOn})\,Dx \rIGht) =
\int_{\omega}(w(\bETa) + {\vArphi}(\xi))\,dx.$$ FiNallY, by the LipSchitz COntinuiTY of $\partIal {\OmeGa}$, fRom And (\[REPs\]) We AlsO dEDucE ThAt $(\bETa_{\vArepsiloN}/|\lOg
{\VarepSiloN}| - \BETA) \cdoT t$ tEnds To zerO strongly in $H^{-\fRac{1}{2}}(\PartIAl {\OMega})$.
[*STep $2.$ ThE casE $\mU := \sum_{l=1}^l \chi_{A_l} \Xi_l \,dx$.*]{}
in This step we proof The $\GAmma$-limsuP inEqUalItY in thE Case of $\Mu$ lOcaLly consTant, i.e., oF The TyPE $$\LAbEl{loco}
\mu:= \sum_{l=1}^L \chi_{A_l} \Xi_L\,DX,$$ wHere $A_l$ arE open sUBsEtS Of ${\Omega}$ wItH LiPschITZ contInuoUS bOundary aNd $\xi_i\iN {\MaThBb{M}}^{2\timeS 2}$. THe consTrUctIon Of the REcovEry seqUence is bAsed oN Classical localIZation argumenTS iN $\gAmMA$-conVerGence and takEs adVAntaGe of THe PreVIous sTep.
LeT uS SeT $\Beta_l:= \beta {\mathop{\hbox{\VrUle heiGht 7pt Width.5pt depth 0pT
\vrule heigHT.5PT width 6pt DeptH 0Pt}}\NOlimits}A | _{\varepsilon})$ is optima l in energ y. We ha ve$$ \lim _{{\ varepsilon}\to 0} \ frac{1}{|\log {\vareps ilon} |^ 2 }\in t _{ \Omeg a}W(\be t a_ { \ var ep si lon }) \, dx=\li m_{{\va repsilon}\ to0} \frac{1}{|\ l og {\varepsi lon }|^2}
\int_{ \Om ega}W( |\ log {\var eps ilon} | \bet a + \ha t \beta_{ \v a repsil o n})\, d x . $$ Sin ce $\hat \beta_{\ v ar e psilon}/|\log{\vare ps i lo n } | { \ri ghtharpoon up }0$ i n $L^2({ \ Om e g a };{ \ mathbb{M}^{2\ times 2}})$ , ta king i nt o a c countalso, w e co nclude $$\l im_{ {\varepsi lon}\t o 0} \fr a c{1}{|\ log {\ var eps ilon } |^ 2} \in t_ { \Om e ga }W( \ bet a_{\vare ps il on})\ ,dx= \ l i m_{{ \va reps ilon} \to 0} \frac{ 1}{ |\lo g {\ varep silon }|^2 }\left ( \int _{{\O me ga}} W(|\log {\ vare psilon}|\be ta )\, dx + \i n t_{{\O meg a}} W(\hat \beta_{ \ var ep s i l on })\,dx \right) =
\ in t _ {\ Omega}(W (\beta ) + { \ varphi}( \x i)) \,dx . $ $ Fin ally , b y the Li pschit z c on tinuity o f $\pa rt ial {\ Omega } $, f rom an d (\[Rep s\])w e also deducet hat $(\beta_{ \ va r e ps i lon} /|\ log
{\varep silo n }| - \be t a) \c d ot t$ tend st oz ero strongly in $H^ {- \frac{ 1}{2} }(\partial {\ Omega})$.[ * Step $2. $ Th e c a se $\mu := \su m_{l= 1}^L \chi_ { A_l} \xi _l \, dx$.*]{}
In this s tep we p roo f t he$\G a m ma $-limsup ineq u a lity i n the c ase of $\m u$loc all y c on stant, i. e., of t he t yp e$$\ label { loco}
\m u: = \ su m_{ l=1}^ L \chi_ {A_l} \xi _l \, d x,$ $ where $A _ l $ ar eop en s ubs et s of${\O m ega }$ with Lipschit z c o ntin uo us bounda ry and $\xi_i \i n {\mathbb {M }}^ {2\tim e s 2}$. Th e construction of the r e coveryseq uence isbased oncla ssical lo c alizat ion ar gumen ts in $ \Gamm a $ -c onv er gence andt a kes adva nt ageof theprevious step.
Le t us set $\beta_l :=\bet a {\ mat h op { \hb ox { \vr u l e height 7pt wi dth.5pt de pt h 0 pt
\vruleh eig ht .5pt wi dth 6pt dept h 0pt}}\ nolimits} A | _{\varepsilon})$ is_optimal in_energy. We have $$\lim_{{\varepsilon}\to_0} \frac{1}{|\log_{\varepsilon}|^2}\int_{\Omega}W(\beta_{\varepsilon})_\, dx=_\lim_{{\varepsilon}\to_0} \frac{1}{|\log {\varepsilon}|^2}
\int_{\Omega}W(|\log_{\varepsilon}| \beta +_\hat \beta_{\varepsilon})\, dx.$$ Since_$\hat \beta_{\varepsilon}/|\log {\varepsilon}|_{\rightharpoonup}0$_in $L^2({\Omega};{\mathbb{M}^{2\times 2}})$, taking into account also, we conclude $$\lim_{{\varepsilon}\to 0} \frac{1}{|\log {\varepsilon}|^2}\int_{\Omega}W(\beta_{\varepsilon})\,dx= \lim_{{\varepsilon}\to_0}_\frac{1}{|\log {\varepsilon}|^2}
\left(_\int_{{\Omega}}_W(|\log_{\varepsilon}| \beta)\,dx + \int_{{\Omega}} W(\hat\beta_{\varepsilon})\,dx_\right) =
\int_{\Omega}(W(\beta) + {\varphi}(\xi))\,dx.$$ Finally,_by the_Lipschitz continuity of $\partial {\Omega}$, from and (\[Reps\])_we_also deduce that_$(\beta_{\varepsilon}/|\log
{\varepsilon}| - \beta) \cdot t$ tends to zero strongly_in $H^{-\frac{1}{2}}(\partial {\Omega})$.
[*Step $2.$ The case_$\mu := \sum_{l=1}^L_\chi_{A_l}_\xi_l_\,dx$.*]{}
In this step we_proof the $\Gamma$-limsup inequality in the_case of $\mu$ locally constant, i.e.,_of the type $$\label{loco}
\mu:= \sum_{l=1}^L \chi_{A_l} \xi_l\,dx,$$_where $A_l$ are open subsets of_${\Omega}$ with Lipschitz continuous boundary_and $\xi_i\in_{\mathbb{M}}^{2\times 2}$. The construction of_the recovery sequence_is based_on classical localization_arguments in $\Gamma$-convergence and takes advantage_of the previous_step.
Let us set $\beta_l:= \beta {\mathop{\hbox{\vrule_height_7pt width.5pt depth_0pt
\vrule_height.5pt_width 6pt_depth 0pt}}\nolimits}A |
Table \[auc\_table\] are compared with other common link prediction methods in Table \[auc\_compare\]: tSVD [@dhillon], tKatz [@dunlavy], and a degree-based model where the probability of a link is approximated as $\mathbb P(A_{ij}=1)=1-\exp(-d^\mathrm{out}_id^\mathrm{in}_j)$ where $d^\mathrm{out}_i$ and $d^\mathrm{in}_j$ are the out-degree and in-degree of each node. Overall, when compared to the results in Table \[auc\_table\], the PMF models achieve impressive improvements over competing matrix factorisation techniques.
=1.2mm
[c c c c c]{} & &\
& & & &\
& & & &\
& & & &\
& & & &\
Seasonal modelling {#seasonal_sec}
------------------
To investigate dynamic modelling, binary adjacency matrices ${\mathbf}A_1,\dots,{\mathbf}A_{82}$ are constructed for each day across the train and test periods. The seasonal PMF model with the inclusion of covariates (SEPMF) is then compared against EPMF; for EPMF, the adjacency matrices are assumed to be independent realisations randomly generated from a fixed set of latent features. Due to a “9 day-80 hour” work schedule operated at LANL, whereby employees can elect to take vacation every other Friday, the seasonal period is assumed to be comprised of four segments: weekdays (Monday - Thursday), weekends (Saturday and Sunday), and two separate segments for alternating Friday’s. For each model, binary classification is performed using the model predictive scores calculated across the entire period. For the positive class, scores are calculated for all user-host pairs $(i,j)$ such that $A_{ijt}=1$ for at least one $t$ in the test set; for the negative class, a random sample of $(i,j)$ pairs such that $A_{ijt}=0$ for all $t$ in the test set are obtained, with sample size equal to three times the total number of observed links.
Table \[auc\_seasonal\] presents the resulting AUC scores. For both networks, the seasonal model does not globally outperform the extended PMF model for [*all links*]{}. However, improvements are obtained for prediction of the *new links*. One explanation for | Table \[auc\_table\ ] are compared with other common link prediction method in Table \[auc\_compare\ ]: tSVD [ @dhillon ], tKatz [ @dunlavy ], and a academic degree - based model where the probability of a liaison is estimate as $ \mathbb P(A_{ij}=1)=1-\exp(-d^\mathrm{out}_id^\mathrm{in}_j)$ where $ d^\mathrm{out}_i$ and $ d^\mathrm{in}_j$ are the out - degree and in - degree of each lymph node. Overall, when compared to the results in Table \[auc\_table\ ], the PMF models achieve impressive improvement over competing matrix factorisation proficiency.
= 1.2 mm
[ c c c c c ] { } & & \
& & & & \
& & & & \
& & & & \
& & & & \
Seasonal model { # seasonal_sec }
------------------
To investigate active modelling, binary adjacency matrices $ { \mathbf}A_1,\dots,{\mathbf}A_{82}$ are constructed for each sidereal day across the train and trial time period. The seasonal PMF model with the inclusion of covariates (SEPMF) is then compared against EPMF; for EPMF, the adjacency matrix are assumed to be independent realisations randomly generated from a repair set of latent features. Due to a “ 9 day-80 hour ” work schedule operated at LANL, whereby employee can elect to take vacation every other Friday, the seasonal period is assumed to be comprised of four segments: weekday (Monday - Thursday), weekend (Saturday and Sunday), and two disjoined segments for alternating Friday ’s. For each model, binary classification is performed using the model predictive scores calculate across the entire period. For the incontrovertible course, scores are calculated for all user - host pairs $ (i, j)$ such that $ A_{ijt}=1 $ for at least one $ t$ in the test set; for the damaging class, a random sample of $ (i, j)$ pairs such that $ A_{ijt}=0 $ for all $ t$ in the test set are receive, with sample size equal to three times the total numeral of observed connection.
mesa \[auc\_seasonal\ ] presents the resulting AUC score. For both networks, the seasonal model does not globally outperform the extensive PMF model for [ * all links * ] { }. However, improvements are obtained for prediction of the * modern links *. One explanation for | Tahle \[auc\_table\] are compared with other common linn predjction mdthods in Table \[auc\_compare\]: tSTD [@dyillob], tKatz [@dunlavy], and a aegree-basvd model qhert the probability of a likh is wpprmeimated as $\mathnb P(A_{ij}=1)=1-\exp(-d^\kathrm{out}_id^\matvro{iu}_j)$ where $d^\mathrm{out}_i$ and $d^\mathrm{in}_j$ wre the okt-degree and ig-deggeq of vagh node. Overall, when compared to the revults in Tablr \[auc\_table\], the PMF models afhiege impressive imprlvements ovgd cjnpeting matrkx factorisation technjques.
=1.2mm
[c c c c c]{} & &\
& & & &\
& & & &\
& & & &\
& & & &\
Sdasoncl modellint {#weadmnal_sec}
------------------
To iivestidate dynamic modellitg, binaty adjacency mstrmces ${\mathbf}A_1,\dots,{\mathbf}A_{82}$ ace constructed for ewch day awrkss the train and twst pgriodv. Thd sexsohak LMF mofel with the jnclusion od covariates (SEPMF) os nnen compared againft EPMF; for EPMF, the adjacency matrices age aasumed to be independenr realisations randompy generaeed from a fixed set of latent features. Due to a “9 day-80 iojr” cirk szyefule operated at LANL, whereby employees can ejscu tp take vacatiok every other Fridsy, tng seasonal perkod is asaumed to be comprided of sour wegments: reekcays (Monday - Thursday), weekebds (Saturday qnd Sunday), and two separate szgmentx for alternating Friday’s. Fox each model, binagy classirkcation is perfofmec gsing tht model predictive scores celculcted acruss jhe entyre period. For bve positive class, dcored dre calculwted for all user-host pairs $(i,j)$ such that $A_{ijt}=1$ fmr dt least one $b$ in the test sqt; for the negctive clcss, a fandom samile of $(i,j)$ pairs such that $A_{ijt}=0$ fos all $t$ in thx test see arw obrained, dkth sample sizr equal to three tines the total numbcr of kbserved links.
Tcylt \[ayc\_seasonal\] prexengs ehv rxsultytg AUC scorev. Fof buyh negworks, uke wdasomal model does not gnobamly outperform the ewtended PNF model for [*all linkx*]{}. However, improvemtnts ace obteined gor prediction of the *new links*. Ohe explanwtijn for | Table \[auc\_table\] are compared with other common methods Table \[auc\_compare\]: [@dhillon], tKatz [@dunlavy], the of a link approximated as $\mathbb where $d^\mathrm{out}_i$ and $d^\mathrm{in}_j$ are the and in-degree of each node. Overall, when compared to the results in Table the PMF models achieve impressive improvements over competing matrix factorisation techniques. =1.2mm [c c c]{} &\ & & &\ & & & &\ & & & &\ & & & &\ Seasonal {#seasonal_sec} ------------------ To investigate dynamic modelling, binary adjacency ${\mathbf}A_1,\dots,{\mathbf}A_{82}$ are constructed for day across the train and periods. seasonal PMF with inclusion covariates (SEPMF) is compared against EPMF; for EPMF, the adjacency matrices are assumed to be independent realisations randomly generated from fixed set features. Due a day-80 work schedule operated whereby employees can elect to take Friday, the seasonal period is assumed to be of four weekdays (Monday - Thursday), weekends (Saturday Sunday), and two separate segments for alternating Friday’s. each model, binary classification is performed using the model predictive scores calculated across the entire the positive class, scores calculated for all pairs such $A_{ijt}=1$ at least $t$ in the test set; for the negative class, a random of $(i,j)$ pairs such that $A_{ijt}=0$ for all $t$ in set obtained, with sample equal to three times total of observed links. Table the AUC networks, seasonal does not globally outperform extended PMF model for [*all However, improvements are obtained links*. One explanation for | Table \[auc\_table\] are compared wIth other coMmon lInk PreDiCtioN metHods in Table \[auc\_COmpaRe\]: tSVD [@dhillon], tKatz [@dunlAvy], anD a DEgreE-BaSed moDel wherE ThE PRobAbIlIty Of A LiNk is aPprOximateD as $\mathbb P(a_{ij}=1)=1-\ExP(-d^\mathrm{out}_iD^\MaThrm{in}_j)$ wheRe $d^\Mathrm{out}_i$ anD $d^\mAthrm{iN}_j$ Are THe out-DegRee anD in-degREe of eaCh node. OveRaLL, when cOMpared tO THe ResuLts in Table \[auc\_tablE\], ThE pMF models achieVe imprEsSIvE IMprOveMents over cOmPetinG Matrix fACtORISatIOn techniques.
=1.2mM
[c c c c c]{} & &\
& & & &\
& & & &\
& & & &\
& & & &\
SeasoNAl mOdelliNg {#SeaSOnal_seC}
------------------
To inVeSTigAte dynamic mOdelLing, binarY adjacENcy matrICes ${\mathBf}A_1,\dotS,{\maThbF}A_{82}$ arE CoNsTruCtED foR EaCh dAY acRoss the tRaIn And teSt peRIODS. The SeaSonaL PMF mOdel with the inCluSion OF coVariaTes (SEpMF) iS tHen coMpared AgainSt ePMF; for EPMF, the aDjacEncy matriCes ArE asSuMed to BE indepEndEnt RealisaTions raNDomLy GENErAted from a fixed set oF lATEnT featureS. Due to A “9 DaY-80 hOUr” work scHeDulE opeRATed at lANL, WHeReby emplOyees cAN eLeCt to takE vAcatioN eVerY otHer FrIDay, tHe seasOnal periOd is aSSumed to be comprISed of four segmENtS: WEeKDays (monDay - Thursday), WeekENds (SAturDAy And sUnday), And twO sEPaRAte segments for alterNaTing FrIday’s. for each model, bInary classIFICation is PerfORmED using the model PrediCtive scoreS CalculatEd acrOss the enTire perioD. fOr the posItiVe cLasS, scOREs Are calculated FOR all UsEr-host pAirS $(i,j)$ such ThaT $A_{iJt}=1$ fOr aT lEast one $t$ iN the test SeT; fOr ThE neGativE Class, a raNdOm sAmPle Of $(i,j)$ pAIrs sucH that $a_{ijt}=0$ FoR aLL $t$ iN the tesT SeT ARe obTaInEd, wiTh sAmPle siZe eqUAl tO three tImes the toTal NUmbeR oF oBserved Links.
Table \[auc\_SeAsonal\] presEnTs tHe resuLTIng AUC scOres. For both networks, the sEAsonal mOdeL does Not gLobally ouTpeRform tHe eXTended pMF modEl for [*AlL liNKS*]{}. HoweVER, iMprOvEments are oBTAinEd for PrEdicTion of tHe *new links*. One explaNAtiOn for | Table \[auc\_table\] arecompared w ith o the r c om monlink prediction me t hods in Table \[auc\_compa re\]: t S VD [ @ dh illon ], tKat z [ @ d unl av y] , a nd adegre e-b ased mo del wherethe p robability o f a link is a ppr oximated as$\m athbbP( A_{ i j}=1) =1- \exp( -d^\ma t hrm{ou t}_id^\ma th r m{in}_ j )$ wher e $d ^\ma thrm{out}_i$ and$ d^ \ mathrm{in}_j$are th eo ut - d egr eeand in-deg re e ofe ach nod e .O v e ral l , when compar ed to the r e sul ts inTa ble \[auc\ _tabl e\ ] , t he PMF mode ls a chieve im pressi v e impro v ementsover c omp eti ng m a tr ix fa ct o ris a ti ont ech niques.
= 1. 2mm
[c c c c c]{ } & &\& & & &\
& & & &\& & & & \
&& & & \
Se ason al mode lling{#sea so nal_sec}
------ ---- --------
To i nve st igate dynami c m ode lling,binarya dja ce n c y m atrices ${\mathbf} A_ 1 , \d ots,{\ma thbf}A _ {8 2} $ are con st ruc tedf o r eac h da y a cross th e trai n a nd test p er iods.Th e s eas onalP MF m odel w ith theinclu s ion of covaria t es (SEPMF) is th e n c o mpar edagainst EPM F; f o r EP MF,t he ad j acenc y mat ri c es are assumed to be i nd epende nt re alisations ra ndomly gen e r a ted from a f i xe d set of latent feat ures. Duet o a “9 d ay-80 hour” w ork sched u l e operat edatLAN L,w h er eby employees c an e le ct to t ake vacati oneve ryoth er Friday,the seas on al p er iod is a s sumed to b e c om pri sed o f foursegme nts: w ee k day s (Mond a y- Thur sd ay ), w eek en ds (S atur d ayand Sun day), and tw o sep ar at e segme nts for alter na ting Frida y’ s.For ea c h model,binary classification i s perfor med usin g th e model p red ictive sc o res ca lculat ed ac ro sst h e ent i r eper io d. For the p osi tivecl ass, scores are calculated fo r al l user-host p air s $( i , j) $ s u ch tha t$ A_{ i j t}=1$ for at le ast one $t $i nthe test s e t;fo r the n egative clas s , a ran dom sampl e of $(i, j) $ pa i r s s uch that $ A_{ijt}= 0$ for al l $t$i nthe t est set a re ob taine d, wit h sa mplesize e qu al tothree t imes the total number of observ ed lin ks.
Tab le \[auc\ _se a son al\] pres ents the resul tin g A UC sc ore s . For bot h n etw o rks,thes easonal m o de l d o e snot globall y o utp erfor m t h e exte nded PMF model for [* a ll links*]{}.Howe v e r,imp r ovem en ts are obtaine d f or p redictio nof the *new links*. O n e exp lanati on for | Table \[auc\_table\]_are compared_with other common link_prediction methods_in_Table \[auc\_compare\]: tSVD_[@dhillon],_tKatz [@dunlavy], and_a degree-based model_where the probability of_a link is_approximated_as $\mathbb P(A_{ij}=1)=1-\exp(-d^\mathrm{out}_id^\mathrm{in}_j)$ where $d^\mathrm{out}_i$ and $d^\mathrm{in}_j$ are the out-degree and in-degree of each_node._Overall, when_compared_to_the results in Table \[auc\_table\], the_PMF models achieve impressive improvements_over competing_matrix factorisation techniques.
=1.2mm
[c c c c c]{} &_&\
&_& & &\
&_& & &\
& & & &\
& & & &\
Seasonal_modelling {#seasonal_sec}
------------------
To investigate dynamic modelling, binary_adjacency matrices ${\mathbf}A_1,\dots,{\mathbf}A_{82}$_are_constructed_for each day across_the train and test periods. The_seasonal PMF model with the inclusion_of covariates (SEPMF) is then compared against_EPMF; for EPMF, the adjacency matrices_are assumed to be independent_realisations randomly_generated from a fixed set_of latent features._Due to_a “9 day-80_hour” work schedule operated at LANL,_whereby employees can_elect to take vacation every other_Friday,_the seasonal period_is_assumed_to be_comprised of four_segments:_weekdays (Monday_-_Thursday), weekends (Saturday and Sunday), and_two_separate segments for alternating Friday’s. For each_model, binary classification is_performed_using the model predictive_scores calculated across the entire_period. For the positive class, scores_are calculated_for all_user-host pairs $(i,j)$ such that $A_{ijt}=1$ for at least one $t$_in the test set; for the_negative class, a random_sample of_$(i,j)$_pairs such that_$A_{ijt}=0$_for all_$t$ in the test set are obtained,_with sample_size equal to three times the_total number of observed_links.
Table \[auc\_seasonal\]_presents the resulting AUC scores. For_both networks, the seasonal model does_not globally outperform the extended_PMF_model_for [*all links*]{}. However, improvements_are obtained for prediction of the_*new links*. One_explanation for |
Subsets and Splits