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
|
|---|---|---|---|---|---|---|
efficient resampling methods for nonsmooth estimating functions that avoids full blown bootstrap. Of course, the methods also apply to full cohort data.
The rest of this article is organized as follows. Point estimation procedures based on smoothed estimating equations for case-cohort data when the sub-cohort is a simple random sample from the full cohort are proposed in Section \[sect:poin\]. Four variance estimation procedures, one based on full multiplier bootstrap and three based on possibly multiplier bootstrap-aided sandwich variance estimator, are proposed in Section \[sect:vari\]. A large scale simulation study is reported in Section \[sect:simu\], comparing the performances of the variance estimator and their timings. The methods are applied to a tumor study with both case-cohort data and full cohort data in Section \[sect:appl\]. A discussion concludes in Section \[sect:disc\].
Point Estimation {#sect:poin}
================
Let $\{T_i, C_i, X_i\}$, $i = 1, \ldots, n$, be $n$ independent copies of $\{T, C, X\}$, where $T_i$ and $C_i$ are log-transformed failure time and log-transformed censoring time, $X_i$ is a $p\times 1$ covariate vector, and given $X$, $C$ and $T$ are assumed to be independent. A semiparametric AFT model has the form $$T_i = X_i^{\top} \beta + \epsilon _i, \qquad i = 1, \ldots, n,$$ where $\beta$ is an unknown $p \times 1$ vector of regression parameters, $\epsilon_i$’s are independent and identically distributed random variables with an unspecified distribution. It is also assumed that $\epsilon_i$’s are independent of $X_i$.
In a full cohort study, due to censoring, the observed data are $(Y_i, \Delta_i, X_i)$, $i = 1, \ldots, n$, where $Y_i = \min(T_i, C_i)$, $\Delta_i = I[T_i < C_i]$, and $I[\cdot]$ is the indicator function. A rank based estimating equation with Gehan’s weight is $$U_n(\beta)=\sum_{i=1}^n \sum_{j=1}^n \ | efficient resampling methods for nonsmooth estimating functions that avoids wide blown bootstrap. Of naturally, the methods also lend oneself to broad cohort data.
The respite of this article is organize as follows. detail estimation procedures free-base on smoothed estimating equations for case - age group data when the sub - age group is a simple random sample distribution from the entire cohort are proposed in Section \[sect: poin\ ]. Four variance estimate procedures, one based on full multiplier bootstrap and three based on possibly multiplier bootstrap - help sandwich variance estimator, are proposed in Section \[sect: vari\ ]. A large scale model study is reported in Section \[sect: simu\ ], comparing the performances of the variance estimator and their timings. The methods are applied to a tumor study with both case - cohort data and full cohort data in Section \[sect: appl\ ]. A discussion concludes in Section \[sect: disc\ ].
Point Estimation { # faction: poin }
= = = = = = = = = = = = = = = =
permit $ \{T_i, C_i, X_i\}$, $ i = 1, \ldots, n$, be $ n$ independent copy of $ \{T, C, X\}$, where $ T_i$ and $ C_i$ are logarithm - transform failure time and log - transformed censoring time, $ X_i$ is a $ p\times 1 $ covariate vector, and given $ X$, $ C$ and $ T$ are assumed to be autonomous. A semiparametric AFT model has the form $ $ T_i = X_i^{\top } \beta + \epsilon _ i, \qquad i = 1, \ldots, n,$$ where $ \beta$ is an unknown $ p \times 1 $ vector of regression parameters, $ \epsilon_i$ ’s are independent and identically distributed random variables with an unspecified distribution. It is also assumed that $ \epsilon_i$ ’s are autonomous of $ X_i$.
In a full age group study, due to censoring, the observed data are $ (Y_i, \Delta_i, X_i)$, $ i = 1, \ldots, n$, where $ Y_i = \min(T_i, C_i)$, $ \Delta_i = I[T_i < C_i]$, and $ I[\cdot]$ is the indicator affair. A rank based estimating equality with Gehan ’s weight is $ $ U_n(\beta)=\sum_{i=1}^n \sum_{j=1}^n \ | efvicient resampling methoas for nonsmootk estimeting fhnctions that avoids full blown bootdteap. Od course, the methods auso apply to full cohiet data.
The rest of this zvticlz ms organized as follows. Pmint estimatiot orlcedures based on smoothed estimatigg equayilns for case-corort qata when the sub-cohort is a simple rahdom sakple from the full cohort are proposed ln Sfction \[sect:poin\]. Foug variance gatiiqtion procedjres, one based on full multiplier bootstrap and three baseb on possiboy mupjiplier bootwtrap-wided sandwign varidnce esyimator, are prppoved in Section \[sect:vari\]. A large scale simulwtion stugy is reported in Swcrion \[sgct:siku\], cunpafinf vhe perfogmaices of the variance ewtimator and their uimybgs. The methoss are a[plied to a tumor study with both case-cmhodt data and full cohort data in Section \[sect:akpl\]. A disctssion concludes in Section \[sect:disc\].
Point Estimatimn {#sert:ooiu}
================
Oet $\{T_k, C_l, X_i\}$, $i = 1, \ldots, n$, be $n$ independent copies of $\{T, D, C\}$, ehere $T_i$ and $C_l$ are log-transformrd fsylure time ana log-txznaformed censoring hime, $X_i$ is a $p\times 1$ sovatiate vector, and given $X$, $C$ qnd $T$ are asfymed to be indepenbent. A semipcramettic AFY model has the form $$T_i = X_i^{\fop} \beta + \eosilon _i, \sduad i = 1, \ldots, n,$$ whvre $\teta$ is an unknown $p \times 1$ vector if rzgressiov patameterf, $\epsilon_i$’d are independent and idfnticclly gistributef random variables with an unspxrified distrinudiot. It is clso axsumed that $\e[silon_i$’s are iudependeut of $B_i$.
In a fulm cohorv study, due eo censoring, dje observed vata are $(R_i, \Dwlta_u, X_i)$, $i = 1, \ldots, n$, where $Y_i = \min(T_p, E_i)$, $\Delta_i = I[T_i < C_i]$, and $I[\cdob]$ is jhs indicator funeuiob. A rank based rstkmaeijg ezgation with Cehav’s dright is $$U_n(\btca)=\smm_{i=1}^v \suk_{j=1}^n \ | efficient resampling methods for nonsmooth estimating functions full bootstrap. Of the methods also The of this article organized as follows. estimation procedures based on smoothed estimating for case-cohort data when the sub-cohort is a simple random sample from the cohort are proposed in Section \[sect:poin\]. Four variance estimation procedures, one based on multiplier and based possibly multiplier bootstrap-aided sandwich variance estimator, are proposed in Section \[sect:vari\]. A large scale simulation study reported in Section \[sect:simu\], comparing the performances of variance estimator and their The methods are applied to tumor with both data full data in Section A discussion concludes in Section \[sect:disc\]. Point Estimation {#sect:poin} ================ Let $\{T_i, C_i, X_i\}$, $i = 1, n$, be copies of C, where and $C_i$ are time and log-transformed censoring time, $X_i$ 1$ covariate vector, and given $X$, $C$ and are assumed be independent. A semiparametric AFT model the form $$T_i = X_i^{\top} \beta + \epsilon \qquad i = 1, \ldots, n,$$ where $\beta$ is an unknown $p \times 1$ vector parameters, $\epsilon_i$’s are independent identically distributed random with unspecified It also assumed $\epsilon_i$’s are independent of $X_i$. In a full cohort study, due censoring, the observed data are $(Y_i, \Delta_i, X_i)$, $i = n$, $Y_i = \min(T_i, $\Delta_i = I[T_i < and is the indicator function. based equation is \sum_{j=1}^n | efficient resampling methodS for nonsmoOth esTimAtiNg FuncTionS that avoids fulL BlowN bootstrap. Of course, the mEthodS aLSo apPLy To fulL cohort DAtA.
tHe rEsT oF thIs ARtIcle iS orGanized As follows. POinT eStimation proCEdUres based oN smOothed estimaTinG equatIoNs fOR case-CohOrt daTa when THe sub-cOhort is a sImPLe randOM sample FROm The fUll cohort are propoSEd IN Section \[sect:poIn\]. Four VaRIaNCE esTimAtion proceDuRes, onE Based on FUlL MULtiPLier bootstrap And three basED on PossibLy MulTIplier BootsTrAP-aiDed sandwich VariAnce estimAtor, arE ProposeD In SectiOn \[sect:VarI\]. A lArge SCaLe SimUlATioN StUdy IS rePorted in seCtIon \[seCt:siMU\], COMparIng The pErforMances of the vaRiaNce eSTimAtor aNd theIr tiMiNgs. ThE methoDs are ApPlied to a tumor stUdy wIth both caSe-cOhOrt DaTa and FUll cohOrt DatA in SectIon \[sect:APpl\]. a dISCUsSion concludes in SecTiON \[SeCt:disc\].
PoInt EstIMaTiON {#sect:poiN}
================
LEt $\{T_I, C_i, X_I\}$, $I = 1, \Ldots, N$, be $n$ INdEpendent Copies OF $\{T, c, X\}$, Where $T_i$ AnD $C_i$ are LoG-trAnsFormeD FailUre timE and log-tRansfORmed censoring tIMe, $X_i$ is a $p\times 1$ COvARIaTE vecTor, And given $X$, $C$ aNd $T$ aRE assUmed TO bE inDEpendEnt. A sEmIPaRAmetric AFT model has tHe Form $$T_i = x_i^{\top} \Beta + \epsilon _i, \qQuad i = 1, \ldots, N,$$ WHEre $\beta$ iS an uNKnOWn $p \times 1$ vector Of regRession parAMeters, $\epSilon_I$’s are indEpendent aND IdenticaLly DisTriButED RaNdom variables WITh an UnSpecifiEd dIstribuTioN. It Is aLso AsSumed that $\Epsilon_i$’S aRe InDePenDent oF $x_i$.
In a fulL cOhoRt StuDy, due TO censoRing, tHe obSeRvED daTa are $(Y_i, \dElTA_I, X_i)$, $i = 1, \LdOtS, n$, whEre $y_i = \Min(T_i, c_i)$, $\DeLTa_i = i[T_i < C_i]$, anD $I[\cdot]$ is tHe iNDicaToR fUnction. a rank based estImAting equatIoN wiTh GehaN’S Weight is $$u_n(\beta)=\sum_{i=1}^n \sum_{j=1}^n \ | efficient resampling meth ods for no nsmoo thest im atin g fu nctions that a v oids full blown bootstrap. Of c ou r se,t he meth ods als o a p p lyto f ull c o ho rt da ta.
The r est of thi s a rt icle is orga n iz ed as foll ows . Point esti mat ion pr oc edu r es ba sed on s moothe d estim ating equ at i ons fo r case-c o h or t da ta when the sub-c o ho r t is a simplerandom s a mp l e fr omthe full c oh ort a r e propo s ed i n Se c tion \[sect:p oin\]. Four var iancees tim a tion p roced ur e s,one based o n fu ll multip lier b o otstrap and thr ee bas edonposs i bl ymul ti p lie r b oot s tra p-aidedsa nd wichvari a n c e est ima tor, areproposed in S ect ion\ [se ct:va ri\]. A l ar ge sc ale si mulat io n study is repo rted in Secti on\[ sec t: simu\ ] , comp ari ngthe per formanc e s o ft h e v ariance estimatoran d th eir timi ngs. T h eme t hods are a ppl iedt o a tu mors tu dy withboth c a se -c ohort d at a andfu llcoh ort d a ta i n Sect ion \[se ct:ap p l\]. A discuss i on concludesi nS e ct i on \ [se ct:disc\].
Poi n t Es tima t io n { # sect: poin}
= = == = ===========
Let $\ {T _i, C_ i, X_ i\}$, $i = 1, \ldots, n $ , be $n$ i ndep e nd e nt copies of $ \{T,C, X\}$, w h ere $T_i $ and $C_i$ a re log-tr a n sformedfai lur e t ime a nd log-transfor m e d ce ns oring t ime , $X_i$ is a$p\ tim es 1$ covar iate vec to r, a nd gi ven $ X $, $C$ a nd $T $are assu m ed tobe in depe nd en t . A semipa r am e t ricAF Tmode l h as theform $$T _i = X_ i^{\top}\be t a +\e ps ilon _i , \qquad i =1, \ldots, n ,$ $ w here $ \ b eta$ isan unknown $p \times 1$ vectorofregre ssio n paramet ers , $\ep sil o n_i$’s are i ndepe nd ent a nd id e n ti cal ly distribut e d ra ndomva riab les wit h an unspecified d i str ibution. It i s a lsoa s su med th a t $ \e p sil o n _i$’s are indep endent of$X _ i$ .
In a fu l l c oh ort stu dy, due to c e nsoring , the obs erved dat aare$ ( Y_i , \Delta_i , X_i)$, $i = 1,\ ldots , n $, wh ere $Y_i=\mi n(T_i , C_i) $ , $ \Delt a_i =I[ T_i <C_i]$ ,and $I[\ cdot]$ is the indicator funct ion.A r ank based es t ima ting equa tion with Geha n’s we ightis$ $U_n( \bet a )= \su m _{i=1 }^n\ sum_{j=1} ^ n\ | efficient_resampling methods_for nonsmooth estimating functions_that avoids_full_blown bootstrap._Of_course, the methods_also apply to_full cohort data.
The rest_of this article_is_organized as follows. Point estimation procedures based on smoothed estimating equations for case-cohort data_when_the sub-cohort_is_a_simple random sample from the_full cohort are proposed in_Section \[sect:poin\]. Four_variance estimation procedures, one based on full multiplier_bootstrap_and three based_on possibly multiplier bootstrap-aided sandwich variance estimator, are proposed_in Section \[sect:vari\]. A large scale simulation_study is reported_in_Section \[sect:simu\],_comparing the performances of_the variance estimator and their timings._The methods are applied to a_tumor study with both case-cohort data and_full cohort data in Section \[sect:appl\]. A_discussion concludes in Section \[sect:disc\].
Point Estimation_{#sect:poin}
================
Let $\{T_i,_C_i, X_i\}$, $i = 1,_\ldots, n$, be_$n$ independent_copies of $\{T,_C, X\}$, where $T_i$ and $C_i$_are log-transformed failure_time and log-transformed censoring time, $X_i$_is_a $p\times 1$_covariate_vector,_and given_$X$, $C$ and_$T$_are assumed_to_be independent. A semiparametric AFT model_has_the form $$T_i = X_i^{\top} \beta +_\epsilon _i, \qquad i_=_1, \ldots, n,$$ where_$\beta$ is an unknown $p_\times 1$ vector of regression parameters,_$\epsilon_i$’s are_independent and_identically distributed random variables with an unspecified distribution. It is also_assumed that $\epsilon_i$’s are independent of_$X_i$.
In a full cohort_study, due_to_censoring, the observed_data_are $(Y_i,_\Delta_i, X_i)$, $i = 1, \ldots, n$,_where $Y_i_= \min(T_i, C_i)$, $\Delta_i = I[T_i_< C_i]$, and $I[\cdot]$_is_the indicator function. A rank based_estimating equation with Gehan’s weight is_$$U_n(\beta)=\sum_{i=1}^n \sum_{j=1}^n \ |
[@2008PhRvD..78h3012E; @2008PhRvD..78l3529Z], nevertheless the dipole could be larger in the case of more exotic models. In addition to the dipole, a moving observer will also see velocity imprints at $\ell>1$ in the CMB due to Doppler and aberration effects [@Challinor:2002zh; @Burles:2006xf]. Such effects can be measured as correlations among different $\ell$s, as has been proposed in Refs. [@Kosowsky:2010jm; @Amendola:2010ty; @Notari:2011sb] and subsequently demonstrated in Ref. .
The aberration effect changes the arrival direction of photons from $\hat{n}^{\prime}$ to $\hat{n}$, which, at linear order in $\beta$, is completely degenerate with a lensing dipole. The Doppler effect modulates the CMB (an effect that is partly degenerate with an intrinsic CMB dipole[^7]) changing the specific intensity $I'$ in the CMB rest frame to the intensity $I$ in the observer’s frame[^8] by a multiplicative, direction-dependent factor as [@1986rpa..book.....R; @Challinor:2002zh] $$\begin{aligned}
I'(\nu',\,\hat{n}^{\prime})=I(\nu,\,\hat{n}) \left(\frac{\nu'}{\nu} \right)^3,
\label{eq:i_boost}\end{aligned}$$ where $$\begin{aligned}
\nu \,=\, \nu^{\prime}\,\gamma\,\big(1+ \vec{\beta}\cdot\hat{n}^{\prime}\big)\,,\qquad
\hat{n} \,=\, \frac{\hat{n}^{\prime} + \left[\gamma\,\beta + (\gamma - 1) \big(\hat{n}^{\prime}\cdot\hat{\beta} \big)\right]\hat{\beta}}{\gamma (1+ \vec{\beta} \cdot\hat{n}^{\prime}) } \,,
\label{eq:nu_boost}\end{aligned}$$ with $\gamma \equiv (1-\vec{\beta}^2)^{-1/2}$. The temperature and polarization fields $X(\hat{n})$ in the CMB rest frame (where $X$ stands for $T,\,E$ or $B$) are similarly transformed as $$\begin{aligned | [ @2008PhRvD.. 78h3012E; @2008PhRvD.. 78l3529Z ], nevertheless the dipole could be larger in the case of more exotic model. In accession to the dipole, a moving observer will also witness velocity imprints at $ \ell>1 $ in the CMB due to Doppler and aberrance effects [ @Challinor:2002zh; @Burles:2006xf ]. Such effects can be measured as correlation among different $ \ell$s, as has been proposed in Refs. [ @Kosowsky:2010jm; @Amendola:2010ty; @Notari:2011sb ] and subsequently demonstrate in Ref. .
The aberration effect change the arrival direction of photons from $ \hat{n}^{\prime}$ to $ \hat{n}$, which, at linear order in $ \beta$, is completely degenerate with a lensing dipole. The Doppler consequence modulates the CMB (an effect that is partly debauched with an intrinsic CMB dipole[^7 ]) changing the specific intensity $ I'$ in the CMB remainder frame to the volume $ I$ in the observer ’s frame[^8 ] by a multiplicative, direction - dependent factor as [ @1986rpa.. book..... radius; @Challinor:2002zh ] $ $ \begin{aligned }
I'(\nu',\,\hat{n}^{\prime})=I(\nu,\,\hat{n }) \left(\frac{\nu'}{\nu } \right)^3,
\label{eq: i_boost}\end{aligned}$$ where $ $ \begin{aligned }
\nu \,=\, \nu^{\prime}\,\gamma\,\big(1 + \vec{\beta}\cdot\hat{n}^{\prime}\big)\,,\qquad
\hat{n } \,=\, \frac{\hat{n}^{\prime } + \left[\gamma\,\beta + (\gamma - 1) \big(\hat{n}^{\prime}\cdot\hat{\beta } \big)\right]\hat{\beta}}{\gamma (1 + \vec{\beta } \cdot\hat{n}^{\prime }) } \, ,
\label{eq: nu_boost}\end{aligned}$$ with $ \gamma \equiv (1-\vec{\beta}^2)^{-1/2}$. The temperature and polarization fields $ X(\hat{n})$ in the CMB rest frame (where $ X$ stands for $ T,\,E$ or $ B$) are similarly transformed as $ $ \begin{aligned | [@2008PhGvD..78h3012E; @2008PhRvD..78l3529Z], neverthelers the dipole could be larged in the case of more exotic models. Mn aeditiin to the dipole, a movkng obserner will qlso wee velocivg imprikcs at $\cll>1$ iu vhe CMB due to Coppler ang aberration exfdccs [@Challinor:2002zh; @Burles:2006xf]. Such effects san be kewsured as corrglatipgs ajong different $\ell$s, as has been prkposed pn Refs. [@Kosowsky:2010jm; @Amendola:2010ty; @Notari:2011sb] and skbseeuently demonstratfd in Ref. .
Thg abqeration effezt changes the arrival direction of photons from $\hat{n}^{\orime}$ to $\hat{n}$, wyixh, wj linear ordxr in $\feta$, is completely dageneraye with a lenslng dmpolw. The Doppler effect kodulates the CMB (wn effect tkat is partly degenerqtw witv an intfunskc DMU djpole[^7]) fhaiging the slecific intwnsity $I'$ in the CMB rqwt frame to tge intqnfity $I$ in the observer’s frame[^8] by a multpplidative, direction-dependebt factor as [@1986rpa..book.....R; @Fhallinor:2002sh] $$\begin{aligned}
I'(\nu',\,\hat{n}^{\prime})=I(\nu,\,\hat{n}) \left(\frac{\nu'}{\tu} \rijhg)^3,
\jxvep{eq:i_boost}\end{aligned}$$ where $$\begin{aligned}
\nu \,=\, \hu^{\krike}\,\gamma\,\big(1+ \vec{\neta}\cdot\hat{n}^{\prime}\bog)\,,\eqiwd
\hat{n} \,=\, \wrac{\hac{h}^{\pdime} + \left[\gamma\,\beha + (\gamia - 1) \vig(\hat{n}^{\pryme}\ccot\hat{\beta} \big)\right]\hat{\beta}}{\gqmma (1+ \vec{\betc} \ceot\hat{n}^{\prime}) } \,,
\pabel{eq:nu_boust}\emd{alibned}$$ with $\gamma \equiv (1-\vzc{\beta}^2)^{-1/2}$. The temperwture and oolarization fieuds $X(\vat{n})$ in ufe CMB rest frame (where $X$ wtanbs for $T,\,D$ or $B$) are similarly transformed as $$\begin{alihned | [@2008PhRvD..78h3012E; @2008PhRvD..78l3529Z], nevertheless the dipole could be the of more models. In addition observer also see velocity at $\ell>1$ in CMB due to Doppler and aberration [@Challinor:2002zh; @Burles:2006xf]. Such effects can be measured as correlations among different $\ell$s, as been proposed in Refs. [@Kosowsky:2010jm; @Amendola:2010ty; @Notari:2011sb] and subsequently demonstrated in Ref. . aberration changes arrival of photons from $\hat{n}^{\prime}$ to $\hat{n}$, which, at linear order in $\beta$, is completely degenerate with lensing dipole. The Doppler effect modulates the CMB effect that is partly with an intrinsic CMB dipole[^7]) the intensity $I'$ the rest to the intensity in the observer’s frame[^8] by a multiplicative, direction-dependent factor as [@1986rpa..book.....R; @Challinor:2002zh] $$\begin{aligned} I'(\nu',\,\hat{n}^{\prime})=I(\nu,\,\hat{n}) \left(\frac{\nu'}{\nu} \right)^3, \label{eq:i_boost}\end{aligned}$$ $$\begin{aligned} \nu \vec{\beta}\cdot\hat{n}^{\prime}\big)\,,\qquad \hat{n} \frac{\hat{n}^{\prime} \left[\gamma\,\beta (\gamma - 1) (1+ \vec{\beta} \cdot\hat{n}^{\prime}) } \,, \label{eq:nu_boost}\end{aligned}$$ (1-\vec{\beta}^2)^{-1/2}$. The temperature and polarization fields $X(\hat{n})$ in CMB rest (where $X$ stands for $T,\,E$ or are similarly transformed as $$\begin{aligned | [@2008PhRvD..78h3012E; @2008PhRvD..78l3529Z], neverthelesS the dipole Could Be lArgEr In thE casE of more exotic mODels. in addition to the dipole, a MovinG oBServER wIll alSo see veLOcITY imPrInTs aT $\eLL>1$ iN the CmB dUe to DopPler and abeRraTiOn effects [@ChaLLiNor:2002zh; @BurleS:2006xf]. such effects cAn bE measuReD as COrrelAtiOns amOng difFErent $\eLl$s, as has bEeN ProposED in Refs. [@kOSoWsky:2010Jm; @Amendola:2010ty; @NotarI:2011Sb] ANd subsequently DemonsTrATeD IN ReF. .
ThE aberratioN eFfect CHanges tHE aRRIVal DIrection of phoTons from $\hat{N}^{\PriMe}$ to $\haT{n}$, WhiCH, at linEar orDeR In $\bEta$, is compleTely DegeneratE with a LEnsing dIPole. The doppleR efFecT modULaTeS thE Cmb (an EFfEct THat Is partly DeGeNeratE witH AN INtriNsiC CMB DipolE[^7]) changing the sPecIfic INteNsity $i'$ in thE CMB ReSt fraMe to thE inteNsIty $I$ in the observEr’s fRame[^8] by a muLtiPlIcaTiVe, dirECtion-dEpeNdeNt factoR as [@1986rpa..bOOk.....R; @chALLInOr:2002zh] $$\begin{aligned}
I'(\nU',\,\hAT{N}^{\pRime})=I(\nu,\,\hAt{n}) \lefT(\FrAc{\NU'}{\nu} \right)^3,
\LaBel{Eq:i_bOOSt}\end{AligNEd}$$ Where $$\begIn{aligNEd}
\Nu \,=\, \Nu^{\prime}\,\GaMma\,\big(1+ \VeC{\beTa}\cDot\haT{N}^{\priMe}\big)\,,\qQuad
\hat{n} \,=\, \Frac{\hAT{n}^{\prime} + \left[\gamMA\,\beta + (\gamma - 1) \big(\HAt{N}^{\PRiME}\cdoT\haT{\beta} \big)\rigHt]\haT{\Beta}}{\GammA (1+ \VeC{\beTA} \cdot\Hat{n}^{\pRiME}) } \,,
\lABel{eq:nu_boost}\end{aligNeD}$$ with $\gAmma \eQuiv (1-\vec{\beta}^2)^{-1/2}$. ThE temperatuRE ANd polariZatiON fIElds $X(\hat{n})$ in the cMB reSt frame (wheRE $X$ stands For $T,\,E$ Or $B$) are siMilarly trANSformed aS $$\beGin{AliGneD | [@2008PhRvD..78h3012E; @2 008PhRvD.. 78l35 29Z ],ne vert hele ss the dipolec ould be larger in the case of m or e exo t ic mode ls. Ina dd i t ion t othe d i po le, a mo ving ob server wil l a ls o see veloci t yimprints a t $ \ell>1$ in t heCMB du etoD opple r a nd ab errati o n effe cts [@Cha ll i nor:20 0 2zh; @B u r le s:20 06xf]. Such effec t sc an be measured as co rr e la t i ons am ong differ en t $\e l l$s, as ha s b een proposed in R efs. [@Koso w sky :2010j m; @A m endola :2010 ty ; @N otari:2011s b] a nd subseq uently demonst r ated in Ref..
The abe r ra ti onef f ect ch ang e s t he arriv al d irect iono f p hoto nsfrom $\ha t{n}^{\prime} $ t o $\ h at{ n}$,which , at l inear order in $ \b eta$, is comple tely degenera tewi thalensi n g dipo le. Th e Doppl er effe c t m od u l a te s the CMB (an effe ct t ha t is par tly de g en er a te withan in trin s i c CMB dip o le [^7]) ch anging th especifi cintens it y $ I'$ in t h e CM B rest frame t o the intensity $I$i n the observe r ’s f ra m e[^8 ] b y a multipl icat i ve,dire c ti on- d epend ent f ac t or as [@1986rpa..book. .. ..R; @ Chall inor:2002zh]$$\begin{a l i g ned}
I'( \ nu ' ,\,\hat{n}^{\p rime} )=I(\nu,\, \ hat{n})\left (\frac{\ nu'}{\nu} \ right)^3 ,
\la bel { e q: i_boost}\end{ a l igne d} $$ wher e $ $\begin {al ign ed}
\nu \,=\, \nu^{\p ri me }\ ,\ gam ma\,\ b ig(1+ \v ec {\b et a}\ cdot\ h at{n}^ {\pri me}\ bi g) \ ,,\ qquad
\ h at{n }\, =\, \f ra c{\ha t{n} ^ {\p rime} + \left[\ gam m a\,\ be ta + (\ga mma - 1) \big (\ hat{n}^{\p ri me} \cdot\ h a t{\beta} \big)\right]\hat{\beta } }{\gamm a ( 1+ \v ec{\ beta} \cd ot\ hat{n} ^{\ p rime}) } \,,
\l abe l { eq:nu _ b oo st} \e nd{aligned } $ $ w ith $ \g amma \equiv (1-\vec{\beta}^2) ^ {-1 /2}$. The tem per atur e an d p o la r iza ti o n f i e lds $X(\hat{n}) $ in the C MB re st frame ( w her e$X$ sta nds for $T,\ , E$ or $ B$) are s imilarlytr ansf o r med as $$\beg in{align ed | [@2008PhRvD..78h3012E;[email protected]], nevertheless_the dipole could be_larger in_the_case of_more_exotic models. In_addition to the_dipole, a moving observer_will also see_velocity_imprints at $\ell>1$ in the CMB due to Doppler and aberration effects [@Challinor:2002zh; @Burles:2006xf]. Such_effects_can be_measured_as_correlations among different $\ell$s, as_has been proposed in Refs. [@Kosowsky:2010jm;_@Amendola:2010ty; @Notari:2011sb]_and subsequently demonstrated in Ref. .
The aberration effect changes_the_arrival direction of_photons from $\hat{n}^{\prime}$ to $\hat{n}$, which, at linear order_in $\beta$, is completely degenerate with_a lensing dipole._The_Doppler_effect modulates the CMB_(an effect that is partly degenerate_with an intrinsic CMB dipole[^7]) changing_the specific intensity $I'$ in the CMB_rest frame to the intensity $I$_in the observer’s frame[^8] by_a multiplicative,_direction-dependent factor as [@1986rpa..book.....R; @Challinor:2002zh] $$\begin{aligned}
_ _I'(\nu',\,\hat{n}^{\prime})=I(\nu,\,\hat{n}) \left(\frac{\nu'}{\nu}_\right)^3,
_ \label{eq:i_boost}\end{aligned}$$ where $$\begin{aligned}
_ \nu \,=\,_\nu^{\prime}\,\gamma\,\big(1+ \vec{\beta}\cdot\hat{n}^{\prime}\big)\,,\qquad
\hat{n}_\,=\,_ \frac{\hat{n}^{\prime} +__\left[\gamma\,\beta_+ (\gamma_- 1) \big(\hat{n}^{\prime}\cdot\hat{\beta}_\big)\right]\hat{\beta}}{\gamma_(1+ \vec{\beta}_\cdot\hat{n}^{\prime})_} \,,
\label{eq:nu_boost}\end{aligned}$$_with_$\gamma \equiv (1-\vec{\beta}^2)^{-1/2}$. The temperature and polarization_fields $X(\hat{n})$ in the_CMB_rest frame (where $X$_stands for $T,\,E$ or $B$)_are similarly transformed as $$\begin{aligned |
$ embeds $2\ell^f$. (Equivalently, there is always an optimal report $(A,\emptyset)\in\hat{\mathcal{R}}$ for $\hat\ell^f$.) Calibration and consistency then follow.
Now suppose $f$ is submodular but not modular. As $f$ is increasing, we will assume without loss of generality that $f(\{i\}) > 0$ for all $i\in N$, which is equivalent to $f(S) > 0$ for all $S\neq\emptyset$; otherwise, $f(T) = f(T\setminus\{i\})$ for all $T\subseteq N$, so discard $i$ from $N$ and continue. In particular, we have $\{\emptyset\} = \operatorname*{arg\,min}_{S\subseteq N} f(S)$.
Define $\epsilon = \bar f / (2\bar f - f(N))$, which is strictly positive by Lemma \[lem:bar-f\] and submodularity of $f$. Let $p = (1-\epsilon) \bar p + \epsilon \delta_\emptyset$, where again $\bar p$ is the uniform distribution, and $\delta_\emptyset$ is the point distribution on $\emptyset$. From Lemma \[lem:2-bar-f\], for all $A\subseteq N$ we have $$\begin{aligned}
\hat\ell^f((A,\emptyset);p)
&= (1-\epsilon) 2 \bar f + \epsilon \, \hat\ell^f((A,\emptyset),\emptyset)\\
&= (1-\epsilon) 2 \bar f + \epsilon \, 2f(A)\\
&\geq (1-\epsilon)2 \bar f > f(N) = \hat\ell^f((\emptyset,N);p)~.
\end{aligned}$$ As we have some report with strictly lower loss than all reports of the form $(A,\emptyset)$, we conclude that we must have some $(A,B) \in {\mathrm{prop}[\hat\ell^f]}(p)$ with $B\neq\emptyset$. We can also see that ${\mathrm{prop}[\ell^f]}(p) = \{\emptyset\}$ by the second equality and the fact that $\{\emptyset\} = \operatorname*{arg\,min}_{S\subseteq N} f(S)$.
Revisiting | $ embeds $ 2\ell^f$. (Equivalently, there is always an optimal reputation $ (A,\emptyset)\in\hat{\mathcal{R}}$ for $ \hat\ell^f$.) Calibration and consistency then be.
Now suppose $ f$ is submodular but not modular. As $ f$ is increasing, we will simulate without loss of generality that $ f(\{i\ }) > 0 $ for all $ i\in N$, which is equivalent to $ f(S) > 0 $ for all $ S\neq\emptyset$; differently, $ f(T) = f(T\setminus\{i\})$ for all $ T\subseteq N$, so discard $ i$ from $ N$ and continue. In particular, we have $ \{\emptyset\ } = \operatorname*{arg\,min}_{S\subseteq N } f(S)$.
specify $ \epsilon = \bar f / (2\bar f - f(N))$, which is strictly positive by Lemma \[lem: legal profession - f\ ] and submodularity of $ f$. Let $ p = (1-\epsilon) \bar p + \epsilon \delta_\emptyset$, where again $ \bar p$ is the undifferentiated distribution, and $ \delta_\emptyset$ is the distributor point distribution on $ \emptyset$. From Lemma \[lem:2 - bar - f\ ], for all $ A\subseteq N$ we have $ $ \begin{aligned }
\hat\ell^f((A,\emptyset);p)
& = (1-\epsilon) 2 \bar f + \epsilon \, \hat\ell^f((A,\emptyset),\emptyset)\\
& = (1-\epsilon) 2 \bar f + \epsilon \, 2f(A)\\
& \geq (1-\epsilon)2 \bar f > f(N) = \hat\ell^f((\emptyset, N);p)~.
\end{aligned}$$ As we have some report with strictly broken loss than all reports of the form $ (A,\emptyset)$, we reason that we must have some $ (A, B) \in { \mathrm{prop}[\hat\ell^f]}(p)$ with $ B\neq\emptyset$. We can also interpret that $ { \mathrm{prop}[\ell^f]}(p) = \{\emptyset\}$ by the second equality and the fact that $ \{\emptyset\ } = \operatorname*{arg\,min}_{S\subseteq N } f(S)$.
Revisiting | $ emheds $2\ell^f$. (Equivalently, tmere is always au optimel repodt $(A,\empthset)\in\hat{\mathcal{R}}$ for $\hat\ell^h$.) Caoibraupon and consistency tfen folloa.
Now suppose $d$ is submovhlar bub not loduner. As $f$ is incrgasing, we winl assume withmug poss of generality that $f(\{i\}) > 0$ for alj $i\in N$, wjich is equivajent eo $f(A) > 0$ for all $S\neq\emptyset$; otherwise, $f(T) = f(T\vetminus\{i\})$ for all $T\subseteq N$, so discarf $i$ vrom $N$ and continuf. In particomar, qe have $\{\empthset\} = \operatorname*{arg\,mjn}_{S\subseteq N} f(S)$.
Define $\epsilon = \bar y / (2\bar f - f(B))$, qhifv is strictoy pofitive by Lemma \[lem:bas-f\] and xubmodularity pf $h$. Ler $p = (1-\epsilon) \bar p + \e'silon \delta_\emptyset$, where agdiu $\bar p$ is the uniforn eistrhbuthon, xbd $\aelua_\ekpfyset$ ls vhe point djstribution on $\emptyset$. From Ltmmw \[oem:2-bar-f\], for aml $A\sufsqteq N$ we have $$\begin{aligned}
\hat\ell^f((A,\tmptyaet);p)
&= (1-\epsilon) 2 \bar f + \epsilon \, \hat\ell^f((A,\emktyset),\emptrset)\\
&= (1-\epsilon) 2 \bar f + \epsilon \, 2f(A)\\
&\geq (1-\epsilmn)2 \bac w > y(K) = \yah\ell^f((\emptyset,N);p)~.
\end{aligned}$$ As we have some dekorn with strictly ljwer loss tnaj sjl reports of the form $(A,\emptyset)$, we concpude thwt we must havt somr $(A,B) \in {\mathrm{prop}[\hat\ell^f]}(p)$ qith $B\neq\empnyser$. We can also see chat ${\mathrm{pxop}[\ell^g]}(p) = \{\ekptyset\}$ by the second edualjty and the fact thaf $\{\emptyset\} = \operagorkama*{arg\,min}_{S\subseteq N} f(S)$.
Revisyting | $ embeds $2\ell^f$. (Equivalently, there is always report for $\hat\ell^f$.) and consistency then submodular not modular. As is increasing, we assume without loss of generality that > 0$ for all $i\in N$, which is equivalent to $f(S) > 0$ all $S\neq\emptyset$; otherwise, $f(T) = f(T\setminus\{i\})$ for all $T\subseteq N$, so discard $i$ $N$ continue. particular, have $\{\emptyset\} = \operatorname*{arg\,min}_{S\subseteq N} f(S)$. Define $\epsilon = \bar f / (2\bar f - f(N))$, is strictly positive by Lemma \[lem:bar-f\] and submodularity $f$. Let $p = \bar p + \epsilon \delta_\emptyset$, again p$ is uniform and is the point on $\emptyset$. From Lemma \[lem:2-bar-f\], for all $A\subseteq N$ we have $$\begin{aligned} \hat\ell^f((A,\emptyset);p) &= (1-\epsilon) 2 \bar + \epsilon &= (1-\epsilon) \bar + \, 2f(A)\\ &\geq f > f(N) = \hat\ell^f((\emptyset,N);p)~. \end{aligned}$$ some report with strictly lower loss than all of the $(A,\emptyset)$, we conclude that we must some $(A,B) \in {\mathrm{prop}[\hat\ell^f]}(p)$ with $B\neq\emptyset$. We can see that ${\mathrm{prop}[\ell^f]}(p) = \{\emptyset\}$ by the second equality and the fact that $\{\emptyset\} = f(S)$. Revisiting | $ embeds $2\ell^f$. (Equivalently, theRe is always An optImaL rePoRt $(A,\eMptySet)\in\hat{\mathcaL{r}}$ for $\Hat\ell^f$.) Calibration and cOnsisTeNCy thEN fOllow.
now suppOSe $F$ IS suBmOdUlaR bUT nOt modUlaR. As $f$ is iNcreasing, wE wiLl Assume withouT LoSs of generaLitY that $f(\{i\}) > 0$ for alL $i\iN N$, whicH iS eqUIvaleNt tO $f(S) > 0$ foR all $S\nEQ\emptySet$; otherwIsE, $F(T) = f(T\seTMinus\{i\})$ fOR AlL $T\suBseteq N$, so discard $i$ FRoM $n$ and continue. In ParticUlAR, wE HAve $\{\EmpTyset\} = \operaToRname*{ARg\,min}_{S\sUBsETEQ N} f(s)$.
define $\epsilon = \Bar f / (2\bar f - f(N))$, wHIch Is striCtLy pOSitive By LemMa \[LEm:bAr-f\] and submoDulaRity of $f$. LeT $p = (1-\epsiLOn) \bar p + \ePSilon \deLta_\empTysEt$, wHere AGaIn $\Bar P$ iS The UNiForM DisTributioN, aNd $\Delta_\EmptYSET$ Is thE poInt dIstriBution on $\emptySet$. from lEmmA \[lem:2-bAr-f\], foR all $a\sUbsetEq N$ we hAve $$\beGiN{aligned}
\hat\ell^f((a,\empTyset);p)
&= (1-\epsIloN) 2 \bAr f + \EpSilon \, \HAt\ell^f((a,\emPtySet),\emptYset)\\
&= (1-\epsILon) 2 \BaR F + \EPsIlon \, 2f(A)\\
&\geq (1-\epsilon)2 \baR f > F(n) = \HaT\ell^f((\empTyset,N);P)~.
\EnD{aLIgned}$$ As wE hAve Some REPort wIth sTRiCtly loweR loss tHAn AlL reportS oF the foRm $(a,\emPtySet)$, we COnclUde thaT we must hAve soME $(A,B) \in {\mathrm{proP}[\Hat\ell^f]}(p)$ with $B\NEq\EMPtYSet$. WE caN also see thaT ${\matHRm{prOp}[\elL^F]}(p) = \{\EmpTYset\}$ bY the sEcONd EQuality and the fact thAt $\{\EmptysEt\} = \opeRatorname*{arg\,mIn}_{S\subseteQ n} F(s)$.
RevisitIng | $ embeds $2\ell^f$. (Equiv alently, t hereisalw ay s an opt imal report $( A ,\em ptyset)\in\hat{\mathca l{R}} $f or $ \ ha t\ell ^f$.) C a li b r ati on a ndco n si stenc y t hen fol low.
Nowsup po se $f$ is su b mo dular butnot modular. As $f $ is i nc rea s ing,wewillassume withou t loss of g e nerali t y that$ f (\ {i\} ) > 0$ for all $i \ in N$, which is e quival en t t o $f( S)> 0$ for a ll $S\n e q\empty s et $ ; oth e rwise, $f(T)= f(T\setmi n us\ {i\})$ f ora ll $T\ subse te q N$ , so discar d $i $ from $N $ andc ontinue . In par ticula r,wehave $\ {\ emp ty s et\ } = \o p era torname* {a rg \,min }_{S \ s u b sete q N } f( S)$.
Define $\eps ilo n =\ bar f /(2\ba r f-f(N)) $, whi ch is s trictly positiv e by Lemma \[ lem :b ar- f\ ] and submod ula rit y of $f $. Let$ p = ( 1 - \ ep silon) \bar p + \e ps i l on \delta_ \empty s et $, where ag ai n $ \bar p $ istheu ni form dis tribut i on ,and $\d el ta_\em pt yse t$is th e poi nt dis tributio n on$ \emptyset$. Fr o m Lemma \[lem : 2- b a r- f \],for all $A\sub sete q N$we h a ve $$ \ begin {alig ne d } \hat\ell^f((A,\e mp tyset) ;p)
&= (1-\eps ilon) 2 \b a r f + \eps ilon \, \hat\ell^f((A, \empt yset),\emp t yset)\\ & = (1-\ep silon) 2\ b ar f + \ eps ilo n \ , 2 f ( A) \\
&\geq( 1 -\ep si lon)2 \ bar f > f( N)= \ hat \el l^ f((\empty set,N);p )~ . \en d{ali g ned}$$ A sweha vesomer eportwithstri ct ly low er loss th a n all r ep orts of t he fo rm $ ( A,\ emptyse t)$, we c onc l udeth at we mus t have some $ (A ,B) \in {\ ma thr m{prop } [ \hat\ell ^f]}(p)$ with $B\neq\em p tyset$. We canalso see that ${ \mathr m{p r op}[\e ll^f]} (p) = \ {\e m p tyset \ } $byth e second e q u ali ty an dthefact th at $\{\emptyset\}= \o peratorname*{ arg \,mi n } _{ S\s u bs e teq N } f( S ) $.
Revisiting | $ embeds_$2\ell^f$. (Equivalently,_there is always an_optimal report_$(A,\emptyset)\in\hat{\mathcal{R}}$_for $\hat\ell^f$.)_Calibration_and consistency then_follow.
Now suppose $f$_is submodular but not_modular. As $f$_is_increasing, we will assume without loss of generality that $f(\{i\}) > 0$ for all_$i\in_N$, which_is_equivalent_to $f(S) > 0$ for_all $S\neq\emptyset$; otherwise, $f(T) =_f(T\setminus\{i\})$ for_all $T\subseteq N$, so discard $i$ from $N$_and_continue. In particular,_we have $\{\emptyset\} = \operatorname*{arg\,min}_{S\subseteq N} f(S)$.
Define $\epsilon =_\bar f / (2\bar f -_f(N))$, which is_strictly_positive_by Lemma \[lem:bar-f\] and submodularity_of $f$. Let $p = (1-\epsilon)_\bar p + \epsilon \delta_\emptyset$, where_again $\bar p$ is the uniform distribution,_and $\delta_\emptyset$ is the point distribution_on $\emptyset$. From Lemma \[lem:2-bar-f\], for_all $A\subseteq_N$ we have $$\begin{aligned}
_ \hat\ell^f((A,\emptyset);p)
_ _ &= (1-\epsilon)_2 \bar f + \epsilon \,_\hat\ell^f((A,\emptyset),\emptyset)\\
_ &= (1-\epsilon) 2 \bar f_+_\epsilon \, 2f(A)\\
___ &\geq_(1-\epsilon)2 \bar f_>_f(N) =_\hat\ell^f((\emptyset,N);p)~.
_ \end{aligned}$$ As we_have_some report with strictly lower loss than_all reports of the_form_$(A,\emptyset)$, we conclude that_we must have some $(A,B)_\in {\mathrm{prop}[\hat\ell^f]}(p)$ with $B\neq\emptyset$. We can_also see_that ${\mathrm{prop}[\ell^f]}(p)_= \{\emptyset\}$ by the second equality and the fact that $\{\emptyset\}_= \operatorname*{arg\,min}_{S\subseteq N} f(S)$.
Revisiting |
a Roseman move then $E_1$ and $E_2$ are related by a sequence of $Ro1$, $Ro2$, $Ro5^*$, $Ro7$, $Br1$ and $Br2$ moves.
<!-- -->
a) Assume we have two pairs $(T,V)$ and $(T',V')$ with the same set of permutations $T=T'=\{\tau_i\}_{1 \leq i \leq r}$ and let $V=\{v_{i,j}\}_{1\leq i\leq r,1\leq j \leq b_i}$ and $V'=\{v'_{i,j}\}_{1\leq i\leq r,1\leq j \leq b_i}$. If for some $i,j$ the pairs $v_{i,j}$ and $v'_{i,j}$ do not agree in small enough neighbourhoods of the points $p_{i,j}$ and $q_{i,\tau_i(j)}$ in $L$, then they can be made to agree using a $P6$ move, due to the fact that the pairs $(T,V)$ and $(T,V')$ satisfy the last property in Definition \[defcancel\]. Thus we assume for each $i,j$ the paths $v_{i,j}$ and $v'_{i,j}$ agree in small enough neighbourhoods of the points $p_{i,j}$ and $q_{i,\tau_i(j)}$ in $L$. Since each component of $L$ is a sphere, there is an isotopy in $L$ taking $v_{i,j}$ to $v'_{i,j}$ relative to their common boundary $p_{i,j}$ and $q_{i,\tau_i(j)}$. During this isotopy, $v_{i,j}$ does not change in a small enough neighbourhood of its boundary. Such an isotopy can be accomplished using the $P2$, $P3$, $P4$, $P5$, and $P7$ moves. Thus after performing each isotopy for all choices of $i,j$ we get a sequence of $P$ moves taking $V$ to $V'$.
Consider now the case of two pairs $(T,V)$ and $(T',V')$ with $T \neq T'$. It is enough to assume that all permutations in $T$ agree with their respective permutations | a Roseman move then $ E_1 $ and $ E_2 $ are related by a sequence of $ Ro1 $, $ Ro2 $, $ Ro5^*$, $ Ro7 $, $ Br1 $ and $ Br2 $ moves.
<! -- -- >
a) wear we take two pairs $ (T, V)$ and $ (T',V')$ with the same bent of permutations $ T = T'=\{\tau_i\}_{1 \leq i \leq r}$ and permit $ V=\{v_{i, j}\}_{1\leq i\leq r,1\leq j \leq b_i}$ and $ V'=\{v'_{i, j}\}_{1\leq i\leq r,1\leq j \leq b_i}$. If for some $ i, j$ the pairs $ v_{i, j}$ and $ v'_{i, j}$ do not match in small enough neighborhood of the points $ p_{i, j}$ and $ q_{i,\tau_i(j)}$ in $ L$, then they can be made to harmonize use a $ P6 $ move, due to the fact that the pair $ (T, V)$ and $ (T, V')$ satisfy the concluding property in Definition \[defcancel\ ]. Thus we assume for each $ i, j$ the path $ v_{i, j}$ and $ v'_{i, j}$ agree in small enough neighbourhoods of the point $ p_{i, j}$ and $ q_{i,\tau_i(j)}$ in $ L$. Since each component of $ L$ is a sphere, there is an isotopy in $ L$ taking $ v_{i, j}$ to $ v'_{i, j}$ relative to their coarse boundary $ p_{i, j}$ and $ q_{i,\tau_i(j)}$. During this isotopy, $ v_{i, j}$ does not change in a small enough neighbourhood of its boundary. Such an isotopy can be accomplished using the $ P2 $, $ P3 $, $ P4 $, $ P5 $, and $ P7 $ moves. Thus after performing each isotopy for all choices of $ i, j$ we get a sequence of $ P$ moves taking $ V$ to $ V'$.
Consider now the event of two pairs $ (T, V)$ and $ (T',V')$ with $ metric ton \neq T'$. It is adequate to assume that all permutations in $ T$ agree with their respective permutations | a Goseman move then $E_1$ and $T_2$ are related by c sequeice of $Do1$, $Ro2$, $Ro5^*$, $Ro7$, $Br1$ and $Br2$ moves.
<!-- -->
a) Assumx we have two pairs $(T,V)$ and $(T',V')$ wkth the swme set if ptrmutations $T=T'=\{\tau_m\}_{1 \leq i \leq r}$ zkd lec $T=\{v_{i,j}\}_{1\leq i\leq r,1\lgq j \leq b_i}$ dnd $V'=\{v'_{i,j}\}_{1\leq i\laq r,1\peq j \leq b_i}$. If for some $i,j$ the paiws $v_{i,j}$ snf $v'_{i,j}$ do not adree yn sjall enough neighbourhoods of the loints $k_{i,j}$ and $q_{i,\tau_i(j)}$ in $L$, then they can be made tl aggee using a $P6$ move, due to the face that the pakrs $(T,V)$ and $(T,V')$ satisfy jhe last property in Definition \[aefcaucel\]. Thus wg zsdome for each $i,j$ tre paths $v_{i,j}$ and $v'_{i,j}$ agree on small enougm neijhboyrhoods of the points $p_{i,j}$ and $q_{i,\tau_i(j)}$ ig $L$. Since ecch component of $L$ is a sphete, thare kw av iaovopg in $L$ taiing $v_{i,j}$ tk $v'_{i,j}$ relatuve to their common bjlmdary $p_{i,j}$ ans $q_{i,\tat_i(t)}$. During this isotopy, $v_{i,j}$ does not chance jn a small enough neighvourhood of its boundwry. Such wn isotopy can be accomplished using the $P2$, $P3$, $P4$, $P5$, dnd $P7$ oovts. Thus qfher performing each isotopy for all choices os $i,k$ ee get a sequekce of $P$ moves takonh $F$ to $V'$.
Conskder now fhe case of two palrs $(T,V)$ wnd $(T',C')$ with $T \geq Y'$. It is enough to assume thqt all permunatiins in $T$ agree witk their respzctive permitations | a Roseman move then $E_1$ and $E_2$ by sequence of $Ro2$, $Ro5^*$, $Ro7$, --> Assume we have pairs $(T,V)$ and with the same set of permutations \leq i \leq r}$ and let $V=\{v_{i,j}\}_{1\leq i\leq r,1\leq j \leq b_i}$ and i\leq r,1\leq j \leq b_i}$. If for some $i,j$ the pairs $v_{i,j}$ and do agree small neighbourhoods of the points $p_{i,j}$ and $q_{i,\tau_i(j)}$ in $L$, then they can be made to agree a $P6$ move, due to the fact that pairs $(T,V)$ and $(T,V')$ the last property in Definition Thus assume for $i,j$ paths and $v'_{i,j}$ agree small enough neighbourhoods of the points $p_{i,j}$ and $q_{i,\tau_i(j)}$ in $L$. Since each component of $L$ is sphere, there isotopy in taking to relative to their $p_{i,j}$ and $q_{i,\tau_i(j)}$. During this isotopy, change in a small enough neighbourhood of its Such an can be accomplished using the $P2$, $P4$, $P5$, and $P7$ moves. Thus after performing isotopy for all choices of $i,j$ we get a sequence of $P$ moves taking $V$ Consider now the case two pairs $(T,V)$ $(T',V')$ $T T'$. is enough assume that all permutations in $T$ agree with their respective permutations | a Roseman move then $E_1$ and $E_2$ are rElated by a sEquenCe oF $Ro1$, $ro2$, $ro5^*$, $Ro7$, $br1$ anD $Br2$ moves.
<!-- -->
a) AssumE We haVe two pairs $(T,V)$ and $(T',V')$ with tHe samE sET of pERmUtatiOns $T=T'=\{\taU_I\}_{1 \lEQ I \leQ r}$ AnD leT $V=\{V_{I,j}\}_{1\Leq i\lEq r,1\Leq j \leq B_i}$ and $V'=\{v'_{i,j}\}_{1\lEq i\LeQ r,1\leq j \leq b_i}$. IF FoR some $i,j$ the PaiRs $v_{i,j}$ and $v'_{i,j}$ dO noT agree In SmaLL enouGh nEighbOurhooDS of the Points $p_{i,j}$ AnD $Q_{i,\tau_i(J)}$ In $L$, then THEy Can bE made to agree using A $p6$ mOVe, due to the fact That thE pAIrS $(t,v)$ anD $(T,V')$ Satisfy the LaSt proPErty in DEFiNITIon \[DEfcancel\]. Thus wE assume for eACh $i,J$ the paThS $v_{i,J}$ And $v'_{i,j}$ Agree In SMalL enough neigHbouRhoods of tHe poinTS $p_{i,j}$ and $Q_{I,\tau_i(j)}$ iN $L$. SincE eaCh cOmpoNEnT oF $L$ iS a SPheRE, tHerE Is aN isotopy In $l$ tAking $V_{i,j}$ tO $V'_{I,J}$ RelaTivE to tHeir cOmmon boundary $P_{i,j}$ And $q_{I,\Tau_I(j)}$. DurIng thIs isOtOpy, $v_{i,J}$ does nOt chaNgE in a small enough NeigHbourhood Of iTs BouNdAry. SuCH an isoTopY caN be accoMplisheD UsiNg THE $p2$, $P3$, $p4$, $P5$, and $P7$ moves. Thus aftEr PERfOrming eaCh isotOPy FoR All choicEs Of $i,J$ we gET A sequEnce OF $P$ Moves takIng $V$ to $v'$.
coNsIder now ThE case oF tWo pAirS $(T,V)$ anD $(t',V')$ wiTh $T \neq t'$. It is enoUgh to ASsume that all peRMutations in $T$ aGReE WItH TheiR reSpective perMutaTIons | a Roseman move then $E_1$ and $E_2$ arerel ate dby a seq uence of $Ro1$ , $Ro 2$, $Ro5^*$, $Ro7$, $B r1$ a nd $Br2 $ m oves.
<!--- -> a) A ss ume w e h ave t wopairs $ (T,V)$ and $( T' ,V')$ with t h esame set o f p ermutations$T= T'=\{\ ta u_i \ }_{1\le q i \ leq r} $ and l et $V=\{v _{ i ,j}\}_ { 1\leq i \ l eq r,1 \leq j \leq b_i}$ an d $V'=\{v'_{i,j }\}_{1 \l e qi \ leq r, 1\leq j \l eq b_i} $ . If fo r s o m e $i , j$ the pairs$v_{i,j}$ a n d $ v'_{i, j} $ d o not a greein sma ll enough n eigh bourhoods of th e points $p_{i,j }$ and $q _{i ,\ta u _i (j )}$ i n $L $ ,the n th ey can b ema de to agr e e u sing a$P6$ move , due to thefac t th a t t he pa irs $ (T,V )$ and$(T,V' )$ sa ti sfy the last pr oper ty in Def ini ti on\[ defca n cel\]. Th uswe assu me fore ach $ i , j $the paths $v_{i,j} $a n d$v'_{i,j }$ agr e ein small en ou ghneig h b ourho odso fthe poin ts $p_ { i, j} $ and $ q_ {i,\ta u_ i(j )}$ in $ L $. S ince e ach comp onent of $L$ is a sp h ere, there is an i so t opyin$L$ taking$v_{ i ,j}$ to$ v' _{i , j}$ r elati ve to their common bounda ry $p_{i ,j}$and $q_{i,\ta u_i(j)}$.D u r ing this iso t op y , $v_{i,j}$ do es no t change i n a small enou gh neigh bourhoodo f its bou nda ry. Su cha n i sotopy can be a ccom pl ished u sin g the $ P2$ , $ P3$ , $ P4 $, $P5$,and $P7$ m ov es .Thu s aft e r perfor mi ngea chisoto p y forall c hoic es o f $i ,j$ weg et a seq ue nc e of $P $moves tak i ng$V$ to$V'$.
C o nsid er n ow thecase of two p ai rs $(T,V)$ a nd$(T',V ' ) $ with $ T \neq T'$. It is enoug h to ass ume that all permutat ion s in $ T$a gree w ith th eir r es pec t i ve pe r m ut ati on s | a_Roseman move_then $E_1$ and $E_2$_are related_by_a sequence_of_$Ro1$, $Ro2$, $Ro5^*$,_$Ro7$, $Br1$ and_$Br2$ moves.
<!-- -->
a) _Assume we have_two_pairs $(T,V)$ and $(T',V')$ with the same set of permutations $T=T'=\{\tau_i\}_{1 \leq i \leq_r}$_and let_$V=\{v_{i,j}\}_{1\leq_i\leq_r,1\leq j \leq b_i}$ and_$V'=\{v'_{i,j}\}_{1\leq i\leq r,1\leq j \leq_b_i}$. If_for some $i,j$ the pairs $v_{i,j}$ and $v'_{i,j}$_do_not agree in_small enough neighbourhoods of the points $p_{i,j}$ and $q_{i,\tau_i(j)}$_in $L$, then they can be_made to agree_using_a_$P6$ move, due to_the fact that the pairs $(T,V)$_and $(T,V')$ satisfy the last property_in Definition \[defcancel\]. Thus we assume for_each $i,j$ the paths $v_{i,j}$ and_$v'_{i,j}$ agree in small enough_neighbourhoods of_the points $p_{i,j}$ and $q_{i,\tau_i(j)}$_in $L$. Since_each component_of $L$ is_a sphere, there is an isotopy_in $L$ taking_$v_{i,j}$ to $v'_{i,j}$ relative to their_common_boundary $p_{i,j}$ and_$q_{i,\tau_i(j)}$._During_this isotopy,_$v_{i,j}$ does not_change_in a_small_enough neighbourhood of its boundary. Such_an_isotopy can be accomplished using the $P2$,_$P3$, $P4$, $P5$, and_$P7$_moves. Thus after performing_each isotopy for all choices_of $i,j$ we get a sequence_of $P$_moves taking_$V$ to $V'$.
Consider now the case of_two pairs $(T,V)$ and $(T',V')$ with_$T \neq T'$. It_is enough_to_assume that all_permutations_in $T$_agree with their respective permutations |
hat
X_S=\sqrt{2/N_P}\hat S_y$ reads after propagation through the sample $$\ew{\hat X_s^{\rm out}}=\ew{\hat X_s^{\rm
in}}-\frac{\kappa}{\sqrt{N_A}}\int d {\bm r} \ew{\hat J_P({\bm
r})}A({\bm r}),$$ where $\kappa^2=a^2N_PN_A/2$, being $N_A$ the total number of atoms. The fluctuations read $$\begin{aligned}
&&( \delta \hat X_S^{\rm out})^2=\frac12+
\frac{\kappa^2}{N_A}\int d{\bm r} A({\bm r})\int d{\bm r'} A({\bm
r'}) \ew{\delta \hat J_P ({\bm r}) \delta \hat J_P({\bm r'})},\label{eq:fluc}\end{aligned}$$ where $\delta \hat J_P = \hat J_P - \langle \hat J_P \rangle$. Thus fluctuations of the light quadrature (polarization fluctuations) after crossing the atomic sample contain the photon shot noise plus a contribution proportional to second order correlations of the atomic spins. Experimentally, spin-spin correlations can be detected if their contribution in the polarization fluctuations is larger that the photon shot noise, which for a coherent initial source corresponds to $( \delta \hat
X_S^{\rm in})^2=\frac12$. To this end, the relevant parameter is $\kappa^2= \eta \alpha$ where $\alpha$ is the [*resonant*]{} optical depth of the sample and $\eta$ is the spontaneous emission probability [@QPS; @Eckertetal08]. It can be shown that the optimal signal is obtained when $\eta$ is tuned to $\eta_{opt}=1/\sqrt{2\alpha}$ and $\kappa^2=\sqrt{\alpha/2}$ [@kappa]. BEC clouds have typical optical depths of the order of a few hundreds, for which one obtains quantum fluctuations imprinted on light by the spin fluctuations which are significantly bigger than the photon shot noise.
Technically it might be challenging to fix the spatial phase of the probing light $\phi$ with respect to the trapping potential of the atoms in standing wave probing configuration (b). We assume that the average of the signal over various shots of the experiment leads to averaging | hat
X_S=\sqrt{2 / N_P}\hat S_y$ reads after propagation through the sample $ $ \ew{\hat X_s^{\rm out}}=\ew{\hat X_s^{\rm
in}}-\frac{\kappa}{\sqrt{N_A}}\int d { \bm radius } \ew{\hat J_P({\bm
r})}A({\bm r}),$$ where $ \kappa^2 = a^2N_PN_A/2 $, being $ N_A$ the entire number of atoms. The fluctuations learn $ $ \begin{aligned }
& & (\delta \hat X_S^{\rm out})^2=\frac12 +
\frac{\kappa^2}{N_A}\int d{\bm r } A({\bm r})\int d{\bm r' } A({\bm
r' }) \ew{\delta \hat J_P ({ \bm r }) \delta \hat J_P({\bm r'})},\label{eq: fluc}\end{aligned}$$ where $ \delta \hat J_P = \hat J_P - \langle \hat J_P \rangle$. Thus fluctuation of the light quadrature (polarization fluctuations) after traverse the atomic sample check the photon shot noise plus a contribution proportional to second decree correlations of the atomic spins. Experimentally, spin - spin correlations can be detect if their contribution in the polarization fluctuations is larger that the photon shot noise, which for a coherent initial informant corresponds to $ (\delta \hat
X_S^{\rm in})^2=\frac12$. To this end, the relevant argument is $ \kappa^2= \eta \alpha$ where $ \alpha$ is the [ * resonant * ] { } ocular depth of the sample distribution and $ \eta$ is the spontaneous emission probability [ @QPS; @Eckertetal08 ]. It can be shown that the optimal signal is receive when $ \eta$ is tuned to $ \eta_{opt}=1/\sqrt{2\alpha}$ and $ \kappa^2=\sqrt{\alpha/2}$ [ @kappa ]. BEC clouds have typical optical depth of the holy order of a few hundreds, for which one obtains quantum fluctuations imprinted on light by the spin fluctuation which are importantly adult than the photon shot randomness.
Technically it might be challenge to situate the spatial phase of the probing light $ \phi$ with deference to the trap potential of the atoms in stand wave probing configuration (b). We assume that the average of the signal over various shots of the experiment moderate to averaging | hat
D_S=\sqrt{2/N_P}\hat S_y$ reads afttr propagation thtoygh thx sampls $$\ew{\hat B_s^{\rm out}}=\ew{\hat X_s^{\rm
in}}-\frac{\kappe}{\sqrr{N_A}}\inu d {\bm r} \ew{\hat J_P({\bm
f})}A({\bm r}),$$ whvre $\kappa^2=q^2N_PN_E/2$, being $N_A$ the total numnzr of wtomv. The fluctuatipns read $$\bagin{aligned}
&&( \dentx \kat X_S^{\rm out})^2=\frac12+
\frac{\kappa^2}{N_A}\int d{\bm r} A({\bm r})\imt d{\bm r'} A({\bm
r'}) \ew{\qelts \hat J_P ({\bm r}) \delta \hat J_P({\bm r'})},\label{eq:flhc}\end{alpgned}$$ where $\delta \hat J_P = \hat J_P - \langle \hah J_P \rangle$. Thus fluctkations of jge jught quadratjre (polarieacion fluctuztions) after crossing the atomiz sam'le contain tye ovoton shot ioise ilus a contrinltion psoportipnal to second orver xorrelations of the avomic spins. Experimegtally, sphn-apin correlations cqn be detactea if thtir ckntribktikn in the lolarizatiob fluctuations is lsrdvt that the pgoton fhjt noise, which for a coherent initial smurde corresponds to $( \deltq \hat
X_S^{\rm in})^2=\frac12$. To tjis end, tre relevant parameter is $\kappa^2= \eta \alpha$ where $\al[ha$ ia tht [*vcsonxbt*]{} optical depth of the sample and $\eta$ is the s[knuanvous emission pronability [@QPS; @Eckeryehak08]. It can be shuwn thcf fhe optimal signal is obtwined when $\eta$ is yuned to $\eta_{opt}=1/\sqrt{2\alpha}$ ane $\kappa^2=\sqrt{\ajpha/2}$ [@kappa]. BEC cloubs have typieal opjical cepths of the order of c few gundreds, fog which ohd obtains quantuo fkuwtuations imprinted on ligrt by the spiu fluctuxtioms whish are sigjificantly bigger than tje phltmn shot nolse.
Technically it might be challenging to fix tve vpatial 'hase pf the probind light $\phi$ wijh respecc to tfe trappinf potenvial of the wtoms in stanglng wave prouing confygurqtiob (b). We xrsume that the average of the sigbal over various smots ur the experimenc oeads to averagong | hat X_S=\sqrt{2/N_P}\hat S_y$ reads after propagation through $$\ew{\hat out}}=\ew{\hat X_s^{\rm d {\bm r} $\kappa^2=a^2N_PN_A/2$, $N_A$ the total of atoms. The read $$\begin{aligned} &&( \delta \hat X_S^{\rm \frac{\kappa^2}{N_A}\int d{\bm r} A({\bm r})\int d{\bm r'} A({\bm r'}) \ew{\delta \hat J_P ({\bm \delta \hat J_P({\bm r'})},\label{eq:fluc}\end{aligned}$$ where $\delta \hat J_P = \hat J_P - \langle J_P Thus of light quadrature (polarization fluctuations) after crossing the atomic sample contain the photon shot noise plus a proportional to second order correlations of the atomic Experimentally, spin-spin correlations can detected if their contribution in polarization is larger the shot which for a initial source corresponds to $( \delta \hat X_S^{\rm in})^2=\frac12$. To this end, the relevant parameter is $\kappa^2= \alpha$ where the [*resonant*]{} depth the and $\eta$ is emission probability [@QPS; @Eckertetal08]. It can the optimal signal is obtained when $\eta$ is to $\eta_{opt}=1/\sqrt{2\alpha}$ $\kappa^2=\sqrt{\alpha/2}$ [@kappa]. BEC clouds have typical depths of the order of a few hundreds, which one obtains quantum fluctuations imprinted on light by the spin fluctuations which are significantly the photon shot noise. it might be to the phase the probing $\phi$ with respect to the trapping potential of the atoms in wave probing configuration (b). We assume that the average of over shots of the leads to averaging | hat
X_S=\sqrt{2/N_P}\hat S_y$ reads afteR propagatiOn thrOugH thE sAmplE $$\ew{\hAt X_s^{\rm out}}=\ew{\hat x_S^{\rm
iN}}-\frac{\kappa}{\sqrt{N_A}}\int d {\bm R} \ew{\haT J_p({\Bm
r})}A({\BM r}),$$ Where $\Kappa^2=a^2N_pn_A/2$, BEIng $n_A$ ThE toTaL NuMber oF atOms. The fLuctuationS reAd $$\Begin{aligned}
&&( \DElTa \hat X_S^{\rm oUt})^2=\fRac12+
\frac{\kappa^2}{n_A}\iNt d{\bm r} a({\bM r})\iNT d{\bm r'} a({\bm
R'}) \ew{\deLta \hat j_p ({\bm r}) \deLta \hat J_P({\bM r'})},\LAbel{eq:FLuc}\end{aLIGnEd}$$ whEre $\delta \hat J_P = \hat J_p - \LaNGle \hat J_P \rangle$. thus flUcTUaTIOns Of tHe light quaDrAture (POlarizaTIoN FLUctUAtions) after crOssing the atOMic Sample CoNtaIN the phOton sHoT NoiSe plus a contRibuTion propoRtionaL To seconD Order coRrelatIonS of The aTOmIc SpiNs. eXpeRImEntALly, Spin-spin CoRrElatiOns cAN BE DeteCteD if tHeir cOntribution in The PolaRIzaTion fLuctuAtioNs Is larGer thaT the pHoTon shot noise, whiCh foR a coherenT inItIal SoUrce cORrespoNds To $( \dElta \hat
x_S^{\rm in})^2=\fRAc12$. TO tHIS EnD, the relevant parameTeR IS $\kAppa^2= \eta \aLpha$ whERe $\AlPHa$ is the [*rEsOnaNt*]{} opTICal dePth oF ThE sample aNd $\eta$ iS ThE sPontaneOuS emissIoN prObaBilitY [@qPS; @ECkerteTal08]. It can Be shoWN that the optimaL Signal is obtaiNEd WHEn $\ETa$ is TunEd to $\eta_{opt}=1/\sQrt{2\aLPha}$ aNd $\kaPPa^2=\SqrT{\Alpha/2}$ [@Kappa]. bEc ClOUds have typical opticAl Depths Of the Order of a few huNdreds, for wHICH one obtaIns qUAnTUm fluctuations ImpriNted on lighT By the spiN flucTuations Which are sIGNificantLy bIggEr tHan THE pHoton shot noisE.
tEchnIcAlly it mIghT be chalLenGinG to Fix ThE spatial pHase of thE pRoBiNg LigHt $\phi$ WIth respeCt To tHe TraPping POtentiAl of tHe atOmS iN StaNding waVE pROBing CoNfIgurAtiOn (B). We asSume THat The averAge of the sIgnAL oveR vArIous shoTs of the experiMeNt leads to aVeRagIng | hat
X_S=\sqrt{2/N_P}\hat S _y$ readsafter pr opa ga tion thr ough the sampl e $$\ ew{\hat X_s^{\rm out}} =\ew{ \h a t X_ s ^{ \rm
i n}}-\fr a c{ \ k app a} {\ sqr t{ N _A }}\in t d {\bm r } \ew{\hat J_ P( {\bm
r})}A({ \ bm r}),$$ wh ere $\kappa^2=a ^2N _PN_A/ 2$ , b e ing $ N_A $ the total number of atoms .T he flu c tuation s re ad $ $\begin{aligned}& &( \delta \hat X_ S^{\rm o u t} ) ^ 2=\ fra c12+
\frac {\ kappa ^ 2}{N_A} \ in t d {\b m r} A({\bm r} )\int d{\bm r'} A({\b mr'} ) \ew{\ delta \ h atJ_P ({\bm r }) \ delta \ha t J_P( { \bm r'} ) },\labe l{eq:f luc }\e nd{a l ig ne d}$ $w her e $ \de l ta\hat J_P = \ hat J _P - \ l a ngle \h at J _P \r angle$. Thusflu ctua t ion s ofthe l ight q uadra ture ( polar iz ation fluctuati ons) after cr oss in g t he atom i c samp lecon tain th e photo n sh ot n o is e plus a contribut io n pr oportion al tos ec on d order c or rel atio n s of t he a t om ic spins . Expe r im en tally,sp in-spi ncor rel ation s can be de tected i f the i r contribution in the polari z at i o nf luct uat ions is lar gert hatthep ho ton shotnoise ,w hi c h for a coherent in it ial so urcecorresponds t o $( \delt a \ hat
X_S^ {\rm in } )^2=\frac12$.To th is end, th e relevan t par ameter i s $\kappa ^ 2 = \eta \ alp ha$ wh ere $ \a lpha$ is the[ * reso na nt*]{}opt ical de pth of th e s am ple and $ \eta$ is t he s po nta neous emission p rob ab ili ty [@ Q PS; @E ckert etal 08 ]. Itcan bes ho w n tha tth e op tim al sign al i s ob tainedwhen $\et a$i s tu ne dto $\et a_{opt}=1/\sq rt {2\alpha}$ a nd$\kapp a ^ 2=\sqrt{ \alpha/2}$ [@kappa]. BE C clouds ha ve ty pica l optical de pths o f t h e orde r of a fewhu ndr e d s, fo r wh ich o ne obtains q uan tum f lu ctua tions i mprinted on lightb y t he spin fluct uat ions w hi cha re sig ni f ica n t ly bigger thanthe photon s h ot noise.
T e chn ic ally it mightbe ch a llengin g to fixthe spati al pha s e of the probi ng light $\phi$ w i th re s pe ct to th e trap pi ngpoten tial o f th e ato ms inst anding wave p robing c onfiguration (b). We as sume t hat t heaverage o f t h e s ignal ove r va rious shot s o f t he ex per i mentlead s t o a v eragi ng | hat
X_S=\sqrt{2/N_P}\hat S_y$_reads after_propagation through the sample_$$\ew{\hat X_s^{\rm_out}}=\ew{\hat_X_s^{\rm
in}}-\frac{\kappa}{\sqrt{N_A}}\int d_{\bm_r} \ew{\hat J_P({\bm
r})}A({\bm_r}),$$ where $\kappa^2=a^2N_PN_A/2$,_being $N_A$ the total_number of atoms._The_fluctuations read $$\begin{aligned}
&&( \delta \hat X_S^{\rm out})^2=\frac12+
\frac{\kappa^2}{N_A}\int d{\bm r} A({\bm r})\int d{\bm r'} A({\bm
r'})_\ew{\delta_\hat J_P_({\bm_r})_\delta \hat J_P({\bm r'})},\label{eq:fluc}\end{aligned}$$ where_$\delta \hat J_P = \hat_J_P -_\langle \hat J_P \rangle$. Thus fluctuations of the_light_quadrature (polarization fluctuations)_after crossing the atomic sample contain the photon shot_noise plus a contribution proportional to_second order correlations_of_the_atomic spins. Experimentally, spin-spin_correlations can be detected if their_contribution in the polarization fluctuations is_larger that the photon shot noise, which_for a coherent initial source corresponds_to $( \delta \hat
X_S^{\rm in})^2=\frac12$._To this_end, the relevant parameter is_$\kappa^2= \eta \alpha$_where $\alpha$_is the [*resonant*]{}_optical depth of the sample and_$\eta$ is the_spontaneous emission probability [@QPS; @Eckertetal08]. It_can_be shown that_the_optimal_signal is_obtained when $\eta$_is_tuned to_$\eta_{opt}=1/\sqrt{2\alpha}$_and $\kappa^2=\sqrt{\alpha/2}$ [@kappa]. BEC clouds have_typical_optical depths of the order of a_few hundreds, for which_one_obtains quantum fluctuations imprinted_on light by the spin_fluctuations which are significantly bigger than_the photon_shot noise.
Technically_it might be challenging to fix the spatial phase of the_probing light $\phi$ with respect to_the trapping potential of_the atoms_in_standing wave probing_configuration_(b). We_assume that the average of the signal_over various_shots of the experiment leads to_averaging |
t\leq T$.}$$
Hence, if we introduce for $1<p\leq +\infty$ the spaces $$L^p(\nu):=\left\{u,\ \text{$\mathcal E$-measurable, such that $u\in L^p(\nu_t)$ for all $0\leq t\leq T$}\right\},$$ then $j$ is well defined on $L^2(\nu)\cap L^\infty(\nu)$. We now give our quadratic growth assumption on $g$.
\[assump:hquad\]\[Quadratic growth\]
[(i)]{} For fixed $(y,z,u)$, $g$ is $\mathbb{F}$-progressively measurable.
[(ii)]{} For any $p\geq 1$ $$\label{inte}
\underset{\tau\in\mathcal T_0^T}{\esup^\mathbb P}\ \mathbb E_\tau\left[\left(\int_\tau^T{\left|g_t(0,0,0)\right|}dt\right)^p\right]<+\infty, \ \mathbb P-a.s.$$
[(iii)]{} $g$ has the following growth property. There exists $(\beta,\gamma)\in \mathbb R_+\times \mathbb R^*_+$ and a positive predictable process $\alpha$ satisfying the same integrability condition as $g_t(0,0,0)$, such that for all $(\omega,t,y,z,u)$ $$\begin{aligned}
-\alpha_t-\beta{\left|y\right|}-\frac\gamma2{\left|z\right|}^2-\frac{j_t(-\gamma u)}{\gamma} \leq g_t(\omega,y,z,u)-g_t(0,0,0)\leq \alpha_t+\beta{\left|y\right|}+\frac\gamma2{\left|z\right|}^2+\frac{j_t(\gamma u)}{\gamma}.\label{eq_quadratique}\end{aligned}$$
\[remrem\] We emphasize that unlike the usual quadratic growth assumptions for continuous BSDEs, condition is not symmetric. It is mainly due to the fact that unlike the functions ${\left|.\right|}$ and ${\left|.\right|}^2$, the function $j$ is not even. Moreover, with this non-symmetric condition, it is easily seen that if | t\leq T$.}$$
Hence, if we introduce for $ 1 < p\leq + \infty$ the spaces $ $ L^p(\nu):=\left\{u,\ \text{$\mathcal E$-measurable, such that $ u\in L^p(\nu_t)$ for all $ 0\leq t\leq T$}\right\},$$ then $ j$ is well defined on $ L^2(\nu)\cap L^\infty(\nu)$. We nowadays contribute our quadratic growth assumption on $ g$.
\[assump: hquad\]\[Quadratic growth\ ]
[ (i) ] { } For fasten $ (y, z, u)$, $ g$ is $ \mathbb{F}$-progressively measurable.
[ (ii) ] { } For any $ p\geq 1 $ $ $ \label{inte }
\underset{\tau\in\mathcal T_0^T}{\esup^\mathbb P}\ \mathbb E_\tau\left[\left(\int_\tau^T{\left|g_t(0,0,0)\right|}dt\right)^p\right]<+\infty, \ \mathbb P - a.s.$$
[ (iii) ] { } $ g$ has the surveil growth property. There exist $ (\beta,\gamma)\in \mathbb R_+\times \mathbb R^*_+$ and a positive predictable process $ \alpha$ meet the same integrability condition as $ g_t(0,0,0)$, such that for all $ (\omega, t, y, z, u)$ $ $ \begin{aligned }
-\alpha_t-\beta{\left|y\right|}-\frac\gamma2{\left|z\right|}^2-\frac{j_t(-\gamma u)}{\gamma } \leq g_t(\omega, y, z, u)-g_t(0,0,0)\leq \alpha_t+\beta{\left|y\right|}+\frac\gamma2{\left|z\right|}^2+\frac{j_t(\gamma u)}{\gamma}.\label{eq_quadratique}\end{aligned}$$
\[remrem\ ] We stress that unlike the usual quadratic growth assumptions for continuous BSDEs, condition is not symmetric. It is chiefly due to the fact that unlike the functions $ { \left|.\right|}$ and $ { \left|.\right|}^2 $, the function $ j$ is not even. furthermore, with this non - symmetric condition, it is easily seen that if | t\lfq T$.}$$
Hence, if we introduct for $1<p\leq +\infty$ jhw spacxs $$L^p(\nu):=\meft\{u,\ \tebt{$\mathcal E$-measurable, such tiat $y\in L^k(\uu_t)$ for all $0\leq t\leq T$}\right\},$$ tjen $j$ is welo defined oi $L^2(\nu)\cap L^\inftg(\ku)$. We iow give our qusdratic grmwth assumptiot un $g$.
\[assump:hquad\]\[Quadratic growth\]
[(i)]{} For sixed $(y,a,u)$, $g$ is $\mathbb{F}$-ptogrexfivemj neasurable.
[(ii)]{} For any $p\geq 1$ $$\lagel{inte}
\lnderset{\tau\in\mathval T_0^T}{\esup^\mathbb P}\ \mathbb F_\tau\peft[\left(\int_\tau^T{\lefh|g_t(0,0,0)\right|}dt\rught)^[\eight]<+\infty, \ \oathbb P-a.s.$$
[(pni)]{} $g$ has thg following growth property. Therd exixts $(\beta,\ganmq)\in \kathbb R_+\timxs \matrbb R^*_+$ and a ipsitiva predivtable process $\al'ha$ watisfying the same iitegrability conditijn as $g_t(0,0,0)$, vueh that for all $(\omega,r,y,z,u)$ $$\becin{anignde}
-\auphz_t-\uetz{\left|y\gigit|}-\frac\gamma2{\meft|z\right|}^2-\feac{j_t(-\gamma u)}{\gamma} \ltq d_n(\pmega,y,z,u)-g_t(0,0,0)\les \alphw_t+\feta{\left|y\right|}+\frac\gamma2{\left|z\right|}^2+\frac{j_t(\camja u)}{\gamma}.\label{eq_quadratuque}\end{aligned}$$
\[remrem\] Ae emphasyze that unlike the usual quadratic growth assumpdions wor cjvrijuous BSDEs, condition is not symmetric. It is izimlj due to the fact that unlikr hhr functions ${\lewt|.\righc|}$ ahd ${\left|.\right|}^2$, the fknction $j$ is not even. Morrover, with this non-symmetrix condition, pt iw easily seen that if | t\leq T$.}$$ Hence, if we introduce for the $$L^p(\nu):=\left\{u,\ \text{$\mathcal such that $u\in T$}\right\},$$ $j$ is well on $L^2(\nu)\cap L^\infty(\nu)$. now give our quadratic growth assumption $g$. \[assump:hquad\]\[Quadratic growth\] [(i)]{} For fixed $(y,z,u)$, $g$ is $\mathbb{F}$-progressively measurable. [(ii)]{} For $p\geq 1$ $$\label{inte} \underset{\tau\in\mathcal T_0^T}{\esup^\mathbb P}\ \mathbb E_\tau\left[\left(\int_\tau^T{\left|g_t(0,0,0)\right|}dt\right)^p\right]<+\infty, \ \mathbb P-a.s.$$ [(iii)]{} $g$ the growth There $(\beta,\gamma)\in \mathbb R_+\times \mathbb R^*_+$ and a positive predictable process $\alpha$ satisfying the same integrability condition $g_t(0,0,0)$, such that for all $(\omega,t,y,z,u)$ $$\begin{aligned} -\alpha_t-\beta{\left|y\right|}-\frac\gamma2{\left|z\right|}^2-\frac{j_t(-\gamma \leq g_t(\omega,y,z,u)-g_t(0,0,0)\leq \alpha_t+\beta{\left|y\right|}+\frac\gamma2{\left|z\right|}^2+\frac{j_t(\gamma u)}{\gamma}.\label{eq_quadratique}\end{aligned}$$ We emphasize that unlike the quadratic assumptions for BSDEs, is symmetric. It is due to the fact that unlike the functions ${\left|.\right|}$ and ${\left|.\right|}^2$, the function $j$ is not even. with this it is seen if | t\leq T$.}$$
Hence, if we introduce foR $1<p\leq +\infty$ The spAceS $$L^p(\Nu):=\Left\{U,\ \texT{$\mathcal E$-measuRAble, Such that $u\in L^p(\nu_t)$ for all $0\Leq t\lEq t$}\RighT\},$$ ThEn $j$ is Well defINeD ON $L^2(\nU)\cAp l^\inFtY(\Nu)$. we now GivE our quaDratic growTh aSsUmption on $g$.
\[asSUmP:hquad\]\[QuadRatIc growth\]
[(i)]{} For FixEd $(y,z,u)$, $g$ Is $\MatHBb{F}$-prOgrEssivEly meaSUrable.
[(Ii)]{} For any $p\GeQ 1$ $$\Label{iNTe}
\underSET{\tAu\in\Mathcal T_0^T}{\esup^\mathBB P}\ \MAthbb E_\tau\left[\lEft(\int_\TaU^t{\lEFT|g_t(0,0,0)\RigHt|}dt\right)^p\RiGht]<+\inFTy, \ \mathbB p-a.S.$$
[(III)]{} $g$ hAS the following Growth propeRTy. THere exIsTs $(\bETa,\gammA)\in \maThBB R_+\tImes \mathbb R^*_+$ And a Positive pRedictABle procESs $\alpha$ SatisfYinG thE samE InTeGraBiLIty COnDitIOn aS $g_t(0,0,0)$, such tHaT fOr all $(\OmegA,T,Y,Z,U)$ $$\begIn{aLignEd}
-\alpHa_t-\beta{\left|y\rIghT|}-\fraC\GamMa2{\lefT|z\rigHt|}^2-\frAc{J_t(-\gamMa u)}{\gamMa} \leq G_t(\Omega,y,z,u)-g_t(0,0,0)\leq \alPha_t+\Beta{\left|y\RigHt|}+\FraC\gAmma2{\lEFt|z\rigHt|}^2+\fRac{J_t(\gamma U)}{\gamma}.\lABel{Eq_QUADrAtique}\end{aligned}$$
\[reMrEM\] we EmphasizE that uNLiKe THe usual qUaDraTic gROWth asSumpTIoNs for conTinuouS bSdES, conditIoN is not SyMmeTriC. It is MAinlY due to The fact tHat unLIke the functionS ${\Left|.\right|}$ and ${\lEFt|.\RIGhT|}^2$, The fUncTion $j$ is not eVen. MOReovEr, wiTH tHis NOn-symMetriC cONdITion, it is easily seen tHaT if | t\leq T$.}$$
Hence, if w e introduc e for $1 <p\ le q +\ inft y$ the spaces$ $L^p (\nu):=\left\{u,\ \tex t{$\m at h calE $- measu rable,s uc h tha t$u \in L ^ p( \nu_t )$for all $0\leq t\ leq T $}\right\},$ $ t hen $j$ is we ll defined o n $ L^2(\n u) \ca p L^\i nft y(\nu )$. We now gi ve our qu ad r atic g r owth as s u mp tion on $g$.
\[assum p :h q uad\]\[Quadrat ic gro wt h \] [(i )]{ } For fixe d$(y,z , u)$, $g $ i s $ \ma t hbb{F}$-progr essively me a sur able.
[ (ii ) ]{} Fo r any $ p \ge q 1$ $$\lab el{i nte}
\und erset{ \ tau\in\ m athcalT_0^T} {\e sup ^\ma t hb bP}\ \ m ath b bE_\ t au\ left[\le ft (\ int_\ tau^ T { \ l eft| g_t (0,0 ,0)\r ight|}dt\righ t)^ p\ri g ht] <+\in fty,\ \m at hbb P -a.s.$ $
[( ii i)]{} $g$ has t he f ollowinggro wt h p ro perty . There ex ist s $(\be ta,\gam m a)\ in \ m at hbb R_+\times \mat hb b R^ *_+$ and a pos i ti ve predicta bl e p roce s s $\al pha$ sa tisfying the s a me i ntegrab il ity co nd iti onas $g _ t(0, 0,0)$, such th at fo r all $(\omega, t ,y,z,u)$ $$\b e gi n { al i gned }
-\alpha_t-\ beta { \lef t|y\ r ig ht| } -\fra c\gam ma 2 {\ l eft|z\right|}^2-\fr ac {j_t(- \gamm a u)}{\gamma} \leq g_t( \ o m ega,y,z, u)-g _ t( 0 ,0,0)\leq \alp ha_t+ \beta{\lef t |y\right |}+\f rac\gamm a2{\left| z \ right|}^ 2+\ fra c{j _t( \ g am ma u)}{\gamma } . \lab el {eq_qua dra tique}\ end {al ign ed} $$
\[remre m\] We e mp ha si ze th at un l ike theus ual q uad ratic growth assu mpti on sf orcontinu o us B SDEs ,co ndit ion i s not sym m etr ic. Itis mainly du e toth efact th at unlike the f unctions $ {\ lef t|.\ri g h t|}$ and ${\left|.\right|}^2$,t he func tio n $j$ isnot even. Mo reover , w i th thi s non- symme tr icc o nditi o n ,itis easily se e n th at if | t\leq_T$.}$$
Hence, if_we introduce for $1<p\leq_+\infty$ the_spaces_$$L^p(\nu):=\left\{u,\ \text{$\mathcal_E$-measurable,_such that $u\in_L^p(\nu_t)$ for all_$0\leq t\leq T$}\right\},$$ then_$j$ is well_defined_on $L^2(\nu)\cap L^\infty(\nu)$. We now give our quadratic growth assumption on $g$.
\[assump:hquad\]\[Quadratic growth\]
[(i)]{} For_fixed_$(y,z,u)$, $g$_is_$\mathbb{F}$-progressively_measurable.
[(ii)]{} For any $p\geq 1$_$$\label{inte}
\underset{\tau\in\mathcal T_0^T}{\esup^\mathbb P}\ \mathbb E_\tau\left[\left(\int_\tau^T{\left|g_t(0,0,0)\right|}dt\right)^p\right]<+\infty,_\ \mathbb_P-a.s.$$
[(iii)]{} $g$ has the following growth property. There_exists_$(\beta,\gamma)\in \mathbb R_+\times_\mathbb R^*_+$ and a positive predictable process $\alpha$ satisfying_the same integrability condition as $g_t(0,0,0)$,_such that for_all_$(\omega,t,y,z,u)$_$$\begin{aligned}
-\alpha_t-\beta{\left|y\right|}-\frac\gamma2{\left|z\right|}^2-\frac{j_t(-\gamma u)}{\gamma} \leq_g_t(\omega,y,z,u)-g_t(0,0,0)\leq \alpha_t+\beta{\left|y\right|}+\frac\gamma2{\left|z\right|}^2+\frac{j_t(\gamma u)}{\gamma}.\label{eq_quadratique}\end{aligned}$$
\[remrem\] We emphasize that_unlike the usual quadratic growth assumptions_for continuous BSDEs, condition is not symmetric._It is mainly due to the_fact that unlike the functions_${\left|.\right|}$ and_${\left|.\right|}^2$, the function $j$ is_not even. Moreover,_with this_non-symmetric condition, it_is easily seen that if |
and isotropic, then the energy-momentum tensor reduces to the perfect fluid form with the energy density =-T\_0\^0=[12]{}\^2+V(), where one recognizes the sum of a kinetic energy and of a potential energy, and the pressure p=[12]{}\^2-V(). The equation of motion for the scalar field is the Klein-Gordon equation, which is obtained by taking the variation of the above action (\[action\_scalar\_field\]) with respect to the scalar field and which reads \^\_=V’, in general and +3H+V’=0 in the particular case of a FLRW (Friedmann-Lemaître-Robertson-Walker) universe.
The system of equations governing the dynamics of the scalar field and of the geometry in a FLRW universe is thus given by $$\begin{aligned}
& &H^2={8\pi G\over 3}\left({1\over 2}\dot\phi^2+V(\phi)\right),
\label{e1}\\
& &\ddot\phi+3H\dot \phi+V'=0,
\label{e2}\\
& & \dot H=-4\pi G\dot\phi^2.
\label{e3}\end{aligned}$$ The last equation can be derived from the first two and is therefore redundant.
The slow-roll regime
--------------------
The dynamical system (\[e1\]-\[e3\]) does not always give an accelerated expansion but it does so in the so-called [*slow-roll regime*]{} when the potential energy of the scalar field dominates over its kinetic energy.
More specifically, the so-called [*slow roll*]{} approximation consists in neglecting the kinetic energy of the scalar field, $\dot \phi^2/2$ in (\[e1\]) and the acceleration $\ddot\phi$ in the Klein-Gordon equation (\[e2\]). One then gets the simplified system $$\begin{aligned}
& &H^2\simeq{8\pi G\over 3} V,
\label{sr1}\\
& &3H\dot \phi+V'\simeq 0.
\label{sr2}\end{aligned}$$ Let us now examine in which regime this approximation is valid. From (\[sr2\]), the velocity of the scalar field is given by -[V’3H]{}. \[phisr\] Substituting this relation in the condition $\ | and isotropic, then the energy - momentum tensor reduces to the perfect fluid human body with the department of energy density = -T\_0\^0=[12]{}\^2+V (), where one recognizes the kernel of a energizing energy and of a potential energy, and the press p=[12]{}\^2 - V (). The equation of gesture for the scalar field is the Klein - Gordon equation, which is prevail by taking the variation of the above legal action (\[action\_scalar\_field\ ]) with respect to the scalar field and which reads \^\_=V ’, in general and +3H+V’=0 in the particular event of a FLRW (Friedmann - Lemaître - Robertson - Walker) universe.
The system of equations govern the dynamics of the scalar field and of the geometry in a FLRW universe is thus pass by $ $ \begin{aligned }
& & H^2={8\pi G\over 3}\left({1\over 2}\dot\phi^2+V(\phi)\right),
\label{e1}\\
& & \ddot\phi+3H\dot \phi+V'=0,
\label{e2}\\
& & \dot H=-4\pi G\dot\phi^2.
\label{e3}\end{aligned}$$ The last equation can be derived from the first two and is therefore excess.
The slow - roll regime
--------------------
The dynamical system (\[e1\]-\[e3\ ]) does not always give an accelerated expansion but it does so in the so - called [ * slow - roll regime * ] { } when the potential energy of the scalar field predominate over its kinetic energy.
More specifically, the indeed - call [ * slow roll * ] { } approximation consist in neglecting the kinetic energy of the scalar field, $ \dot \phi^2/2 $ in (\[e1\ ]) and the acceleration $ \ddot\phi$ in the Klein - Gordon equality (\[e2\ ]). One then gets the simplified system $ $ \begin{aligned }
& & H^2\simeq{8\pi G\over 3 } V,
\label{sr1}\\
& & 3H\dot \phi+V'\simeq 0.
\label{sr2}\end{aligned}$$ Let us now examine in which regime this approximation is valid. From (\[sr2\ ]), the velocity of the scalar field is given by -[V’3H ] { }. \[phisr\ ] Substituting this relation back in the condition $ \ | anf isotropic, then the enevgy-momentum tensor redures to fhe perfdct fluid form with the enerjy dwnsitt =-T\_0\^0=[12]{}\^2+V(), where one recognixes the slm of a kunetmc energy and of a potenbnal ehcrgy, cnv the pressure k=[12]{}\^2-V(). The equathon of motion xof che scalar field is the Klein-Gordon qquatiom, ahich is obtaiged nr taipnn the variation of the above actjon (\[actpon\_scalar\_field\]) wiyh respect to the scalar fleld and which reads \^\_=V’, in general and +3Y+V’=0 in the pafticular case of a FLRS (Friedmann-Lemaître-Robertson-Walkdr) unnverse.
The sisfel of equatiois govvrning the dyksmics mf the xcalar field akd of thw geometry in a FLRW nniverse is thus givgn by $$\begit{amigned}
& &H^2={8\pi G\over 3}\oedt({1\ovet 2}\dot\[hi^2+V(\oyi)\rkghu),
\lebem{e1}\\
& &\ddoh\phm+3H\dot \phi+V'=0,
\mabel{e2}\\
& & \dot H=-4\pi G\dot\phi^2.
\label{e3}\emd{woigned}$$ The laat equwtyon can be derived from the first two atd js therefore redundant.
Tye slow-roll regime
--------------------
The dynamicaj system (\[e1\]-\[e3\]) does not always give an accelerated efpansmov bbb kr foes so in the so-called [*slow-roll regime*]{} when ege pptential energn of the scalar firlf cjminates over its knhefic energy.
More spefificaljy, thw so-calleq [*slpw roll*]{} approximation consiwts in neglebtint the kinetic eneryy of the scclar foeld, $\cot \phi^2/2$ in (\[e1\]) and the aceelerafion $\ddot\phl$ in the Iuein-Gordon equatkon (\[e2\]). One then gets the simplifyed systen $$\beyin{aligndd}
& &H^2\ximeq{8\py G\over 3} V,
\label{sr1}\\
& &3H\dot \phi+V'\simeq 0.
\labep{ss2}\end{alignef}$$ Let us now examine in which rxjime this apptoxhmanion is vclid. Fvom (\[sr2\]), the velosity of the scclar fiejd is given by -[N’3H]{}. \[phisr\] Vubstitutind this relatimj in the convition $\ | and isotropic, then the energy-momentum tensor reduces perfect form with energy density =-T\_0\^0=[12]{}\^2+V(), of kinetic energy and a potential energy, the pressure p=[12]{}\^2-V(). The equation of for the scalar field is the Klein-Gordon equation, which is obtained by taking variation of the above action (\[action\_scalar\_field\]) with respect to the scalar field and reads in and in the particular case of a FLRW (Friedmann-Lemaître-Robertson-Walker) universe. The system of equations governing the dynamics the scalar field and of the geometry in FLRW universe is thus by $$\begin{aligned} & &H^2={8\pi G\over 2}\dot\phi^2+V(\phi)\right), & &\ddot\phi+3H\dot \label{e2}\\ & H=-4\pi G\dot\phi^2. \label{e3}\end{aligned}$$ last equation can be derived from the first two and is therefore redundant. The slow-roll regime -------------------- dynamical system not always an expansion it does so so-called [*slow-roll regime*]{} when the potential scalar field dominates over its kinetic energy. More the so-called roll*]{} approximation consists in neglecting the energy of the scalar field, $\dot \phi^2/2$ in and the acceleration $\ddot\phi$ in the Klein-Gordon equation (\[e2\]). One then gets the simplified system &H^2\simeq{8\pi G\over 3} V, & &3H\dot \phi+V'\simeq \label{sr2}\end{aligned}$$ us examine which regime approximation is valid. From (\[sr2\]), the velocity of the scalar field given by -[V’3H]{}. \[phisr\] Substituting this relation in the condition | and isotropic, then the energy-Momentum teNsor rEduCes To The pErfeCt fluid form witH The eNergy density =-T\_0\^0=[12]{}\^2+V(), where one RecogNiZEs thE SuM of a kInetic eNErGY And Of A pOteNtIAl EnergY, anD the preSsure p=[12]{}\^2-V(). The EquAtIon of motion fOR tHe scalar fiEld Is the Klein-GoRdoN equatIoN, whICh is oBtaIned bY takinG The varIation of tHe ABove acTIon (\[actiON\_ScAlar\_Field\]) with respect tO ThE Scalar field and Which rEaDS \^\_=V’, IN GenEraL and +3H+V’=0 in thE pArticULar case OF a flrw (FrIEdmann-LemaîtrE-Robertson-WALkeR) univeRsE.
ThE System Of equAtIOns Governing thE dynAmics of thE scalaR Field anD Of the geOmetry In a fLRw uniVErSe Is tHuS GivEN bY $$\beGIn{aLigned}
& &H^2={8\pI G\OvEr 3}\lefT({1\oveR 2}\DOT\Phi^2+V(\Phi)\RighT),
\labeL{e1}\\
& &\ddot\phi+3H\dot \Phi+v'=0,
\labEL{e2}\\
& & \dOt H=-4\pi g\dot\pHi^2.
\laBeL{e3}\end{AligneD}$$ The lAsT equation can be dErivEd from the FirSt Two AnD is thERefore RedUndAnt.
The sLow-roll REgiMe
--------------------
tHE DyNamical system (\[e1\]-\[e3\]) doeS nOT AlWays give An acceLErAtED expansiOn But It doES So in tHe so-CAlLed [*slow-rOll regIMe*]{} WhEn the poTeNtial eNeRgy Of tHe scaLAr fiEld domInates ovEr its KInetic energy.
MoRE specifically, THe SO-CaLLed [*sLow Roll*]{} approxiMatiON conSistS In NegLEctinG the kInETiC Energy of the scalar fiElD, $\dot \phI^2/2$ in (\[e1\]) aNd the acceleraTion $\ddot\phI$ IN The Klein-gordON eQUation (\[e2\]). One then Gets tHe simplifiED system $$\bEgin{aLigned}
& &H^2\sImeq{8\pi G\ovER 3} v,
\label{sr1}\\
& &3h\doT \phI+V'\sImeQ 0.
\LAbEl{sr2}\end{aligneD}$$ lEt us NoW examinE in Which reGimE thIs aPprOxImation is Valid. FroM (\[sR2\]), tHe VeLocIty of THe scalar FiEld Is GivEn by -[V’3h]{}. \[Phisr\] SUbstiTutiNg ThIS reLation iN ThE COndiTiOn $\ | and isotropic, then the e nergy-mome ntumten sor r educ es t o the perfectf luid form with the energydensi ty =-T\ _ 0\ ^0=[1 2]{}\^2 + V( ) , wh er eone r e co gnize s t he sumof a kinet icen ergy and ofa p otential e ner gy, and thepre ssurep= [12 ] {}\^2 -V( ). Th e equa t ion of motion f or the sc a lar fie l d i s th e Klein-Gordon eq u at i on, which is o btaine db yt a kin g t he variati on of t h e above ac t i o n ( \ [action\_scal ar\_field\] ) wi th res pe ctt o thescala rf iel d and which rea ds \^\_=V ’, ing enerala nd +3H+ V’=0 i n t hepart i cu la r c as e of aFLR W (F riedmann -L em aître -Rob e r t s on-W alk er)unive rse.
The sys tem ofe qua tions gove rnin gthe d ynamic s ofth e scalar fieldandof the ge ome tr y i na FLR W unive rse is thus g iven by $$\ be g i n {a ligned}
& &H^2={8\ pi G \o ver 3}\l eft({1 \ ov er 2}\dot\p hi ^2+ V(\p h i )\rig ht),
\ label{e1 }\\
&& \d do t\phi+3 H\ dot \p hi +V' =0,
\la b el{e 2}\\
& & \dotH=-4\ p i G\dot\phi^2. \label{e3}\en d {a l i gn e d}$$ Th e last equa tion canbe d e ri ved fromthe f ir s tt wo and is therefore r edunda nt.
The slow-roll regime
-- - - - -------- ---- - --
The dynamical syst em (\[e1\] - \[e3\])doesnot alwa ys give a n accelera ted ex pan sio n bu t it does soi n the s o-calle d [ *slow-r oll re gim e*] {} when the potenti al e ne rg y o f the scalar f ie lddo min ateso ver it s kin etic e ne r gy.
Mores pe c i fica ll y, the so -c alled [*s l owroll*]{ } approxi mat i on c on si sts inneglecting th ekinetic en er gyof the s calar fi eld, $\dot \phi^2/2$ in (\[e1\] ) a nd th e ac celeratio n $ \ddot\ phi $ in th e Klei n-Gor do n e q u ation ( \[ e2\ ]) . One then g ets thesi mpli fied sy stem $$\begin{alig n ed}
& &H^2\simeq {8\ pi G \ o ve r 3 } V ,
\ la b el{ s r 1}\\
& &3H\dot\phi+V'\si me q 0 .
\label{ s r2} \e nd{alig ned}$$Let u s now ex amine inwhich reg im e th i s ap proximatio n is val id. From( \[sr2 \ ]) , the ve locity o f t he sc alar f i eld is g iven b y-[V’3H ]{}.\[ phisr\]Substituting this relat ion in thecon dition $\ | and_isotropic, then_the energy-momentum tensor reduces_to the_perfect_fluid form_with_the energy density_=-T\_0\^0=[12]{}\^2+V(), where one_recognizes the sum of_a kinetic energy_and_of a potential energy, and the pressure p=[12]{}\^2-V(). The equation of motion for the_scalar_field is_the_Klein-Gordon_equation, which is obtained by_taking the variation of the_above action_(\[action\_scalar\_field\]) with respect to the scalar field and_which_reads \^\_=V’, in_general and +3H+V’=0 in the particular case of a_FLRW (Friedmann-Lemaître-Robertson-Walker) universe.
The system of equations_governing the dynamics_of_the_scalar field and of_the geometry in a FLRW universe_is thus given by $$\begin{aligned}
& &H^2={8\pi_G\over 3}\left({1\over 2}\dot\phi^2+V(\phi)\right),
\label{e1}\\
& &\ddot\phi+3H\dot \phi+V'=0,
\label{e2}\\
&_& \dot H=-4\pi G\dot\phi^2.
\label{e3}\end{aligned}$$ The last_equation can be derived from_the first_two and is therefore redundant.
The_slow-roll regime
--------------------
The dynamical_system (\[e1\]-\[e3\])_does not always_give an accelerated expansion but it_does so in_the so-called [*slow-roll regime*]{} when the_potential_energy of the_scalar_field_dominates over_its kinetic energy.
More_specifically,_the so-called_[*slow_roll*]{} approximation consists in neglecting the_kinetic_energy of the scalar field, $\dot \phi^2/2$_in (\[e1\]) and the_acceleration_$\ddot\phi$ in the Klein-Gordon_equation (\[e2\]). One then gets_the simplified system $$\begin{aligned}
& &H^2\simeq{8\pi G\over_3} V,_
\label{sr1}\\
& &3H\dot_\phi+V'\simeq 0.
\label{sr2}\end{aligned}$$ Let us now examine in which regime this_approximation is valid. From (\[sr2\]), the_velocity of the scalar_field is_given_by -[V’3H]{}. \[phisr\]_Substituting_this relation_in the condition $\ |
dsky:2006wb; @Brodsky:2007yz] have proposed a novel mechanism for exclusive diffractive Higgs production $pp \to p H p $ and nondiffractive Higgs production in which the Higgs boson carries a significant fraction of the projectile proton momentum. The production mechanism is based on the subprocess $(Q \bar Q) g \to H $ where the $Q \bar Q$ in the $|uud Q \bar Q>$ intrinsic heavy quark Fock state has up to $80\%$ of the projectile protons momentum. This mechanism provides a clear experimental signal for Higgs production at the LHC due to the small background in this kinematic region.
Diffraction Dissociation as a Tool to Resolve Hadron Substructure and Test Color Transparency
=============================================================================================
Diffractive multi-jet production in heavy nuclei provides a novel way to resolve the shape of light-front Fock state wavefunctions and test color transparency [@Brodsky:1988xz]. For example, consider the reaction [@Bertsch:1981py; @Frankfurt:1999tq]. $\pi A \rightarrow {\rm
Jet}_1 + {\rm Jet}_2 + A^\prime$ at high energy where the nucleus $A^\prime$ is left intact in its ground state. The transverse momenta of the jets balance so that $ \vec k_{\perp i} + \vec k_{\perp 2} = \vec q_\perp < {R^{-1}}_A .$ Because of color transparency, the valence wavefunction of the pion with small impact separation will penetrate the nucleus with minimal interactions, diffracting into jet pairs [@Bertsch:1981py]. The $x_1=x$, $x_2=1-x$ dependence of the dijet distributions will thus reflect the shape of the pion valence light-cone wavefunction in $x$; similarly, the $\vec k_{\perp 1}- \vec k_{\perp 2}$ relative transverse momenta of the jets gives key information on the second transverse momentum derivative of the underlying shape of the valence pion wavefunction [@Frankfurt:1999tq; @Nikolaev:2000sh]. The diffractive nuclear amplitude extrapolated to $t = 0$ should be linear in nuclear number $A$ if color transparency is correct. The integrated diffractive rate will then scale as $A^2/R^2_A \sim A^{4/3}$. This is in fact what has been observed by the E791 collaboration at FermiLab | dsky:2006wb; @Brodsky:2007yz ] have proposed a novel mechanism for exclusive diffractive Higgs production $ pp \to phosphorus planck's constant p $ and nondiffractive Higgs production in which the Higgs boson impart a meaning fraction of the projectile proton momentum. The production mechanism is based on the subprocess $ (Q \bar Q) g \to H $ where the $ Q \bar Q$ in the $ |uud Q \bar Q>$ intrinsic intemperate quark Fock state have up to $ 80\%$ of the projectile protons momentum. This mechanism put up a clear experimental signal for Higgs production at the LHC due to the small setting in this kinematic region.
Diffraction Dissociation as a Tool to Resolve Hadron Substructure and Test Color Transparency
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Diffractive multi - jet production in fleshy nuclei provides a novel room to resolve the shape of light - front Fock state of matter wavefunctions and test color transparency [ @Brodsky:1988xz ]. For example, consider the chemical reaction [ @Bertsch:1981py; @Frankfurt:1999tq ]. $ \pi A \rightarrow { \rm
Jet}_1 + { \rm Jet}_2 + A^\prime$ at high energy where the nucleus $ A^\prime$ is left intact in its ground department of state. The transverse momenta of the jets balance so that $ \vec k_{\perp i } + \vec k_{\perp 2 } = \vec q_\perp < { R^{-1}}_A .$ Because of color transparency, the valence wavefunction of the pion with small impact separation will penetrate the core with minimal interactions, diffracting into fountain pair [ @Bertsch:1981py ]. The $ x_1 = x$, $ x_2=1 - x$ dependence of the dijet distributions will thus chew over the shape of the pion valence unaccented - cone wavefunction in $ x$; similarly, the $ \vec k_{\perp 1}- \vec k_{\perp 2}$ relative transverse momenta of the fountain gives key information on the second transverse momentum derivative of the underlying condition of the valence pion wavefunction [ @Frankfurt:1999tq; @Nikolaev:2000sh ]. The diffractive nuclear amplitude extrapolated to $ t = 0 $ should be linear in nuclear number $ A$ if color transparency is right. The incorporate diffractive pace will then scale as $ A^2 / R^2_A \sim A^{4/3}$. This is in fact what has been observed by the E791 collaboration at FermiLab | dskj:2006wb; @Brodsky:2007yz] have propored a novel meckqnism hor excmusive dkffractive Higgs production $'p \ti p H p $ and nondiffractive Higgs prlduction in xhich the Higgs ukson cavxies z siguihicant fraction of the prmjectile protot oolentum. The production mechanism is fased om hhe subprocess $(Q \bsw Q) f \to H $ where the $Q \bar Q$ in the $|uhd Q \bag Q>$ intrinsic heafy quark Fock state has up to $80\%$ of the projectile protons monenttn. This mechavism provides a clear gxperimental signal for Higgs pruductnon at the OHX dkg to the smaol babkground in tmps kinekatic rrgion.
Diffractipn Vissiciation as a Tool to Resolve Hadron Sufstructura cnd Test Color Transpqrwncy
=============================================================================================
Dhffrdctixw mjltj-jxt lroduchioi in heavy huclei provudes a novel way to rqwolve the shale of jidht-front Fock state wavefunctions and ttst cklor transparency [@Brodskt:1988xz]. For example, consifer the rqaction [@Bertsch:1981py; @Frankfurt:1999tq]. $\pi A \rightarrow {\rm
Jed}_1 + {\rm Get}_2 + W^\oeile$ at high energy where the nucleus $A^\prime$ is megt intact in its ground staye. Tng transverse mumenta of the jets balance do that $ \vec k_{\perp i} + \vec k_{\perp 2} = \vec q_\perp < {R^{-1}}_A .$ Becquse of colog trqnsparency, the valznce wavefunetion pf thr pion with small impacc sepadation will penetrats the nucleus witf mpnimdl interactions, diffractind into jev paixs [@Bertscf:1981py]. Jhe $x_1=x$, $v_2=1-x$ dependejce on the dijet distribktiond fill thus geflect the shape of the pion vekence light-cpna wdvefunctnon in $x$; similarly, ehe $\vec k_{\perp 1}- \vec k_{\pzrp 2}$ rdlative trznsversx momenta of the jets givad key informetion on ehe wecobd tranrxerse momentum derivatine of rhe underlying shaie of fhe valence piou qavefunction [@Framkfjrt:1999eq; @Nmkolaqe:2000sh]. The diffsactkve muclexr amplitudc ebtralolated to $t = 0$ shoulg be linear in nuclear nmmber $A$ id color eransparency os correct. The inttgratev diffcactivr rwte will then scale as $A^2/R^2_A \sim A^{4/3}$. This id ik fact what hws bcen jbserved bv the E791 collaboration at FermiLab | dsky:2006wb; @Brodsky:2007yz] have proposed a novel mechanism diffractive production $pp p H p in the Higgs boson a significant fraction the projectile proton momentum. The production is based on the subprocess $(Q \bar Q) g \to H $ where $Q \bar Q$ in the $|uud Q \bar Q>$ intrinsic heavy quark Fock has to of projectile protons momentum. This mechanism provides a clear experimental signal for Higgs production at the LHC to the small background in this kinematic region. Dissociation as a Tool Resolve Hadron Substructure and Test Transparency Diffractive multi-jet in nuclei a novel way resolve the shape of light-front Fock state wavefunctions and test color transparency [@Brodsky:1988xz]. For example, consider the [@Bertsch:1981py; @Frankfurt:1999tq]. \rightarrow {\rm + Jet}_2 A^\prime$ at high the nucleus $A^\prime$ is left intact state. The transverse momenta of the jets balance that $ k_{\perp i} + \vec k_{\perp 2} \vec q_\perp < {R^{-1}}_A .$ Because of color the valence wavefunction of the pion with small impact separation will penetrate the nucleus with diffracting into jet pairs The $x_1=x$, $x_2=1-x$ of dijet will reflect the of the pion valence light-cone wavefunction in $x$; similarly, the $\vec 1}- \vec k_{\perp 2}$ relative transverse momenta of the jets information the second transverse derivative of the underlying of valence pion wavefunction [@Frankfurt:1999tq; diffractive amplitude = should linear in nuclear number if color transparency is correct. integrated diffractive rate will A^{4/3}$. This is in fact what has been by the E791 collaboration at FermiLab | dsky:2006wb; @Brodsky:2007yz] have proposEd a novel meChaniSm fOr eXcLusiVe diFfractive Higgs PRoduCtion $pp \to p H p $ and nondiffRactiVe hIggs PRoDuctiOn in whiCH tHE higGs BoSon CaRRiEs a siGniFicant fRaction of tHe pRoJectile protoN MoMentum. The pRodUction mechanIsm Is baseD oN thE SubprOceSs $(Q \baR Q) g \to H $ WHere thE $Q \bar Q$ in tHe $|UUd Q \bar q>$ IntrinsIC HeAvy qUark Fock state has uP To $80\%$ OF the projectile ProtonS mOMeNTUm. THis Mechanism pRoVides A Clear exPErIMENtaL Signal for HiggS production AT thE LHC duE tO thE Small bAckgrOuND in This kinematIc reGion.
DiffrAction dIssociaTIon as a TOol to REsoLve hadrON SUbStrUcTUre ANd tesT colOr TranspArEnCy
=============================================================================================
DifFracTIVE MultI-jeT proDuctiOn in heavy nuclEi pRoviDEs a Novel Way to ResoLvE the sHape of Light-FrOnt Fock state wavEfunCtions and TesT cOloR tRanspARency [@BRodSky:1988Xz]. For exAmple, coNSidEr THE ReAction [@Bertsch:1981py; @FraNkFURt:1999Tq]. $\pi A \rigHtarroW {\Rm
jeT}_1 + {\Rm Jet}_2 + A^\prImE$ at High ENErgy wHere THe Nucleus $A^\Prime$ iS LeFt Intact iN iTs grouNd StaTe. THe traNSverSe momeNta of the Jets bALance so that $ \vec K_{\Perp i} + \vec k_{\perp 2} = \VEc Q_\PErP < {r^{-1}}_A .$ BeCauSe of color trAnspARencY, the VAlEncE WavefUnctiOn OF tHE pion with small impacT sEparatIon wiLl penetrate thE nucleus wiTH MInimal inTeraCTiONs, diffracting iNto jeT pairs [@BertSCh:1981py]. The $x_1=X$, $x_2=1-x$ dePendence Of the dijeT DIstributIonS wiLl tHus REFlEct the shape of THE pioN vAlence lIghT-cone waVefUncTioN in $X$; sImilarly, tHe $\vec k_{\peRp 1}- \VeC k_{\PeRp 2}$ rElatiVE transveRsE moMeNta Of the JEts givEs key InfoRmAtIOn oN the secONd TRAnsvErSe MomeNtuM dErivaTive OF thE underlYing shape Of tHE valEnCe Pion wavEfunction [@FranKfUrt:1999tq; @NikolAeV:2000sh]. the difFRActive nuClear amplitude extrapolaTEd to $t = 0$ shOulD be liNear In nuclear NumBer $A$ if ColOR transParencY is coRrEct. tHE inteGRAtEd dIfFractive raTE WilL then ScAle aS $A^2/R^2_A \sim a^{4/3}$. This is in fact what hAS beEn observed by tHe E791 CollABOrAtiON aT ferMilAb | dsky:2006wb; @Brodsky:2007 yz] have p ropos eda n ov el m echa nism for exclu s ivediffractive Higgs prod uctio n$ pp \ t op H p $ andn on d i ffr ac ti veHi g gs prod uct ion inwhich theHig gs boson carri e sa signific ant fraction of th e proj ec til e prot onmomen tum. T h e prod uction me ch a nism i s basedo n t he s ubprocess $(Q \ba r Q ) g \to H $ whe re the $ Q \ b a r Q $ i n the $|uu dQ \ba r Q>$ in t ri n s i c h e avy quark Foc k state has upto $80 \% $ o f the p rojec ti l e p rotons mome ntum . This me chanis m provid e s a cle ar exp eri men tals ig na l f or Hig g spro d uct ion at t he L HC du e to t h e sma llback groun d in this kin ema ticr egi on.
Diffr acti on Diss ociati on as a Tool to Resolv e Ha dron Subs tru ct ure a nd Te s t Colo r T ran sparenc y
===== = === == = = = == ================== == = = == ======== ====== = == == = ======== == === ==== = = ===== ==== = == ==
Diff ractiv e m ul ti-jetpr oducti on in he avy n u clei provi des a no vel w a y to resolve t h e shape of li g ht - f ro n t Fo ckstate wavef unct i onsandt es t c o lor t ransp ar e nc y [@Brodsky:1988xz]. F or exa mple, consider the reaction[ @ B ertsch:1 981p y ;@ Frankfurt:1999 tq].$\pi A \ri g htarrow{\rmJet}_1 + {\rm Jet } _ 2 + A^\p rim e$athig h en ergy where th e nucl eu s $A^\p rim e$ is l eft in tac t i nits groun d state. T he t ra nsv ersem omenta o fthe j ets bala n ce sothat$ \v ec k _ {\p erp i}+ \ v e c k_ {\ pe rp 2 } = \ vec q _\pe r p < {R^{-1 }}_A .$ B eca u se o fco lor tra nsparency, th evalence wa ve fun ctiono f the pio n with small impact sep a rationwil l pen etra te the nu cle us wit h m i nimalintera ction s, di f f racti n g i nto j et pairs [ @ B ert sch:1 98 1py] . The $ x_1=x$, $x_2=1-x$d epe ndence of the di jetd i st rib u ti o nswi l l t h u s reflect the s hape of th ep io n valencel igh t- cone wa vefunct ion i n $x$; s imilarly, the $\ve ck_{\ p e rp1}- \vec k _{\perp2}$ relat i ve tr a ns verse mo mentaof th e jet s give s ke y inf ormati on on th e sec on d transv erse momentum derivativ e of t he un der lying sha peo f t he valenc e pi on wavefun cti on[@Fra nkf u rt:19 99tq ; @ Nik o laev: 2000 s h]. The d i ff rac t i ve nuclear am p l i tud e ext rap o latedto $ t = 0$ should bel inear in nucle ar n u m ber $A $ ifco lor transparen cyis c orrect.Th e integrate d diffra ct i ve ra te wil l then scalea s $ A ^2/R^2 _A \ sim A^{4/3}$ . T hi s is infa ct what h as b ee n obse rved b y the E 791 collaboratio n atF e rmiLa b | dsky:2006wb; @Brodsky:2007yz]_have proposed_a novel mechanism for_exclusive diffractive_Higgs_production $pp_\to_p H p_$ and nondiffractive_Higgs production in which_the Higgs boson_carries_a significant fraction of the projectile proton momentum. The production mechanism is based on_the_subprocess $(Q_\bar_Q)_g \to H $ where_the $Q \bar Q$ in_the $|uud_Q \bar Q>$ intrinsic heavy quark Fock state_has_up to $80\%$_of the projectile protons momentum. This mechanism provides a_clear experimental signal for Higgs production_at the LHC_due_to_the small background in_this kinematic region.
Diffraction Dissociation as a_Tool to Resolve Hadron Substructure and_Test Color Transparency
=============================================================================================
Diffractive multi-jet production in heavy_nuclei provides a novel way to_resolve the shape of light-front_Fock state_wavefunctions and test color transparency [@Brodsky:1988xz]._For example, consider_the reaction [@Bertsch:1981py;_@Frankfurt:1999tq]. $\pi A_\rightarrow {\rm
Jet}_1 + {\rm Jet}_2 +_A^\prime$ at high_energy where the nucleus $A^\prime$ is_left_intact in its_ground_state._The transverse_momenta of the_jets_balance so_that_$ \vec k_{\perp i} + \vec_k_{\perp_2} = \vec q_\perp < {R^{-1}}_A .$_Because of color transparency,_the_valence wavefunction of the_pion with small impact separation_will penetrate the nucleus with minimal_interactions, diffracting_into jet_pairs [@Bertsch:1981py]. The $x_1=x$, $x_2=1-x$ dependence of the dijet distributions will thus_reflect the shape of the pion_valence light-cone wavefunction in_$x$; similarly,_the_$\vec k_{\perp 1}-_\vec_k_{\perp 2}$_relative transverse momenta of the jets gives_key information_on the second transverse momentum derivative_of the underlying shape_of_the valence pion wavefunction [@Frankfurt:1999tq; @Nikolaev:2000sh]. The_diffractive nuclear amplitude extrapolated to $t_= 0$ should be linear_in_nuclear_number $A$ if color transparency_is correct. The integrated diffractive rate_will then scale_as $A^2/R^2_A \sim A^{4/3}$. This is in_fact_what has been observed by the_E791_collaboration at FermiLab |
connected complement and nonnegative self-intersection, then there are infinitely many topologically equivalent but smoothly distinct embedded surfaces homologous to $\Sigma$. Here we extend this result to include symplectic surfaces whose self-intersection is bounded below by $2-2g$, where $g$ is the genus of $\Sigma$.
We make use of tools from Heegaard Floer theory, and include several results that may be of independent interest. Specifically we give an analogue for [[Ozsváth-Szabó]{} ]{}invariants of the Fintushel-Stern knot surgery formula for Seiberg-Witten invariants, both for closed 4-manifolds and manifolds with boundary. This is based on a formula for the [[Ozsváth-Szabó]{} ]{}invariants of the result of a logarithmic transformation, analogous to one obtained by Morgan-Mrowka-Szabó for Seiberg-Witten invariants, and the results on [[Ozsváth-Szabó]{} ]{}invariants of fiber sums due to the author and Jabuka. In addition, we give a calculation of the twisted Heegaard Floer homology of circle bundles of “large” degree over Riemann surfaces.
author:
- 'Thomas E. Mark'
title: 'Knotted surfaces in 4-manifolds'
---
Introduction
============
Some time ago, Fintushel and Stern introduced a technique they called “rim surgery” for changing the embedding of a smooth surface $\Sigma$ in a closed 4-manifold $X$ [@FSrim; @FSaddend]. Their construction makes use of a knot $K\subset S^3$, and can be seen as an instance of their earlier “knot surgery” construction, applied to the complement of $\Sigma$. The interesting aspect of the construction is that under suitable conditions, Fintushel and Stern were able to show that the resulting surface $\Sigma_K\subset X$ is topologically equivalent to $\Sigma$, but smoothly distinct: rim surgery results in a [*smooth*]{} knotting of $\Sigma$ but not a [*topological*]{} one.
To ensure that the topological type of $(X,\Sigma)$ is unchanged by the construction, it suffices to assume that the complement $Z$ of a regular neighborhood of $\Sigma$ is simply connected (see [@FSrim; @boyer; @freedman], also Remark \[pi1remark\] below). Our current concern is | connected complement and nonnegative self - intersection, then there are infinitely many topologically equivalent but smoothly discrete implant surfaces homologous to $ \Sigma$. Here we extend this result to admit symplectic surfaces whose self - intersection is bounded downstairs by $ 2 - 2g$, where $ g$ is the genus of $ \Sigma$.
We make use of tool from Heegaard Floer theory, and include respective results that may be of independent interest. Specifically we give an analogue for [ [ Ozsváth - Szabó ] { } ] { } invariant of the Fintushel - Stern knot surgery rule for Seiberg - Witten invariants, both for closed 4 - manifold and manifolds with boundary. This is based on a rule for the [ [ Ozsváth - Szabó ] { } ] { } invariants of the result of a logarithmic transformation, analogous to one obtained by Morgan - Mrowka - Szabó for Seiberg - Witten invariant, and the results on [ [ Ozsváth - Szabó ] { } ] { } invariant of fiber sums due to the author and Jabuka. In addition, we hold a calculation of the twisted Heegaard Floer homology of circle bundles of “ large ” degree over Riemann surface.
generator:
-' Thomas E. Mark'
title:' Knotted surfaces in 4 - manifolds'
---
Introduction
= = = = = = = = = = = =
Some time ago, Fintushel and Stern introduced a technique they call “ rim surgery ” for transfer the embedding of a legato surface $ \Sigma$ in a closed 4 - manifold $ X$ [ @FSrim; @FSaddend ]. Their construction cook use of a knot $ K\subset S^3 $, and can be seen as an instance of their early “ knot surgery ” construction, applied to the complement of $ \Sigma$. The interesting aspect of the structure is that under suitable conditions, Fintushel and Stern were able to show that the resulting surface $ \Sigma_K\subset X$ is topologically equivalent to $ \Sigma$, but smoothly distinct: rim operating room results in a [ * smooth * ] { } knotting of $ \Sigma$ but not a [ * topological * ] { } one.
To ensure that the topological type of $ (X,\Sigma)$ is unaltered by the construction, it suffices to assume that the complement $ Z$ of a regular neighborhood of $ \Sigma$ is simply connected (see [ @FSrim; @boyer; @freedman ], also Remark \[pi1remark\ ] under). Our current concern is | conjected complement and noknegative self-injeesectimn, theh there xre infinitely many topologirallt equuvalent but smoothly dkstinct elbedded wurfeces homologous vk $\Sigma$. Here sc extznv this result tp include vymplectic surxazed whose self-intersection is bounded below ny $2-2g$, where $g$ is jhe gtnuf of $\Sigma$.
We make use of tools froj Heegaerd Floer theoru, and include several resupts hhat may be of indfpendent injsrefr. Specificaluy we give an analogue for [[Ozsváth-Szabó]{} ]{}invariants of tfe Fiutushel-Sterb jnoh surgery focmula sor Seiberg-Wlnten ineariantx, both for cloxed 4-mabifolds and manifolds with boundary. Thif is baseg kn a formula for rhw [[Ozseáth-Ssabó]{} ]{}kbvafiahtx kf the reault of a mogarithmic transformation, anakodius to one obfained br Morgan-Mrowka-Szabó for Seiberg-Witten innariznts, and the results on [[Ozsváth-Szabó]{} ]{}invariantd of fibew sums due to the author and Jabuka. In addition, wa givx x ccoculaguoj of the twisted Heegaard Floer homology of cydcke bundles of “lavge” degree over Rirmwnm surfaces.
authur:
- 'Thomas E. Mark'
title: 'Knottfd surfwces un 4-manifojds'
---
Imtroduction
============
Some time ago, Fibtushel and Frern introduced a cechnique thzy calked “rom surgery” for changing the embedding lf a smooff surface $\Sigma$ kn s wlosed 4-manifold $X$ [@FSrim; @FSwddend]. Thxir cpnstrucgion makes use of a nnot $K\subset S^3$, and can bf seeu as dn instancf of their earlier “knot surgery” construction, sp[livd to the compkement of $\Sigia$. The interesjing aspeet of ghe constrlction is that under suitable congltions, Fintuvhel and Steen wwre abld to show that yhe resulniug surfacw $\Sigma_K\subset X$ ix tulologically equnraoent to $\Sigma$, bit rmojtjlb disehnct: rim surcery rerilts kn a [*smooth*]{} ynotying of $\Sigma$ but nod a [*fopological*]{} one.
To rnfure thaj the topjlogical type of $(X,\Sigma)$ is unchwnged by the vonftruction, it suffices to assums that thf cjmplement $Z$ jf a regular neiyhborhood of $\Sigma$ is simply connected (sxe [@FSrim; @boyer; @freedman], also Remark \[pi1rematk\] nelow). Our cucrent soncern iv | connected complement and nonnegative self-intersection, then there many equivalent but distinct embedded surfaces extend result to include surfaces whose self-intersection bounded below by $2-2g$, where $g$ the genus of $\Sigma$. We make use of tools from Heegaard Floer theory, include several results that may be of independent interest. Specifically we give an for ]{}invariants the knot surgery formula for Seiberg-Witten invariants, both for closed 4-manifolds and manifolds with boundary. This is on a formula for the [[Ozsváth-Szabó]{} ]{}invariants of result of a logarithmic analogous to one obtained by for invariants, and results [[Ozsváth-Szabó]{} of fiber sums to the author and Jabuka. In addition, we give a calculation of the twisted Heegaard Floer homology circle bundles degree over surfaces. - E. Mark' title: in 4-manifolds' --- Introduction ============ Some and Stern introduced a technique they called “rim for changing embedding of a smooth surface $\Sigma$ a closed 4-manifold $X$ [@FSrim; @FSaddend]. Their construction use of a knot $K\subset S^3$, and can be seen as an instance of their surgery” construction, applied to complement of $\Sigma$. interesting of construction that under conditions, Fintushel and Stern were able to show that the resulting $\Sigma_K\subset X$ is topologically equivalent to $\Sigma$, but smoothly distinct: results a [*smooth*]{} knotting $\Sigma$ but not a one. ensure that the topological $(X,\Sigma)$ unchanged it to that the complement $Z$ a regular neighborhood of $\Sigma$ simply connected (see [@FSrim; below). Our current concern is | connected complement and nonNegative seLf-intErsEctIoN, theN theRe are infinitelY Many Topologically equivalenT but sMoOThly DIsTinct EmbeddeD SuRFAceS hOmOloGoUS tO $\SigmA$. HeRe we extEnd this resUlt To Include symplECtIc surfaces WhoSe self-intersEctIon is bOuNdeD Below By $2-2g$, Where $G$ is the GEnus of $\sigma$.
We maKe USe of toOLs from HEEGaArd FLoer theory, and inclUDe SEveral results tHat may Be OF iNDEpeNdeNt interest. spEcifiCAlly we gIVe AN ANalOGue for [[Ozsváth-szabó]{} ]{}invariANts Of the FInTusHEl-SterN knot SuRGerY formula for seibErg-Witten InvariANts, both FOr closeD 4-manifOldS anD manIFoLdS wiTh BOunDArY. ThIS is Based on a FoRmUla foR the [[oZSVÁth-SZabÓ]{} ]{}invArianTs of the result Of a LogaRIthMic trAnsfoRmatIoN, analOgous tO one oBtAined by Morgan-MrOwka-szabó for SEibErG-WiTtEn invARiants, And The Results On [[OzsváTH-SzAbÓ]{} ]{}INVaRiants of fiber sums dUe TO ThE author aNd JabuKA. IN aDDition, we GiVe a CalcULAtion Of thE TwIsted HeeGaard FLOeR hOmology Of Circle BuNdlEs oF “largE” DegrEe over riemann sUrfacES.
author:
- 'Thomas E. mArk'
title: 'KnottED sURFaCEs in 4-ManIfolds'
---
IntroDuctIOn
============
SoMe tiME aGo, FINtushEl and stERn INtroduced a technique ThEy callEd “rim Surgery” for chaNging the emBEDDing of a sMootH SuRFace $\Sigma$ in a clOsed 4-mAnifold $X$ [@FSRIm; @FSaddeNd]. TheIr constrUction makES Use of a knOt $K\SubSet s^3$, anD CAn Be seen as an insTANce oF tHeir earLieR “knot suRgeRy” cOnsTruCtIon, applieD to the coMpLeMeNt Of $\SIgma$. THE interesTiNg aSpEct Of the COnstruCtion Is thAt UnDEr sUitable COnDITionS, FInTushEl aNd stern Were ABle To show tHat the resUltINg suRfAcE $\Sigma_K\Subset X$ is topoLoGically equIvAleNt to $\SiGMA$, but smooThly distinct: rim surgery rESults in A [*smOoth*]{} kNottIng of $\SigmA$ buT not a [*tOpoLOgical*]{} One.
To eNsure ThAt tHE TopolOGIcAl tYpE of $(X,\Sigma)$ iS UNchAnged By The cOnstrucTion, it suffices to asSUme That the compleMenT $Z$ of A REgUlaR NeIGhbOrHOod OF $\sigma$ is simply coNnected (see [@fSRIm; @Boyer; @freedMAn], aLsO Remark \[Pi1remarK\] beloW). our currEnt concerN is | connected complement and n onnegative self -in ter se ctio n, t hen there arei nfin itely many topological ly eq ui v alen t b ut sm oothlyd is t i nct e mb edd ed su rface s h omologo us to $\Si gma $. Here we ext e nd this resu ltto include s ymp lectic s urf a ces w hos e sel f-inte r sectio n is boun de d below by $2-2 g $ ,wher e $g$ is the genu s o f $\Sigma$.
We m ak e u s e of to ols from H ee gaard Floer t h eo r y , an d include seve ral results tha t maybe of indepe ndent i n ter est. Specif ical ly we giv e an a n aloguef or [[Oz sváth- Sza bó] {} ] { }i nv ari an t s o f t heF int ushel-St er nknotsurg e r y form ula for Seib erg-Witten in var iant s , b oth f or cl osed 4 -mani foldsand m an ifolds with bou ndar y. This i s b as edon a fo r mula f orthe [[Ozsv áth-Sza b ó]{ }] { } in variants of the re su l t o f a loga rithmi c t ra n sformati on , a nalo g o us to one ob tained b y Morg a n- Mr owka-Sz ab ó forSe ibe rg- Witte n inv ariant s, and t he re s ults on [[Ozsv á th-Szabó]{} ] { }i n v ar i ants of fiber sums due to t he a u th ora nd Ja buka. I n a d dition, we give a c al culati on of the twistedHeegaard F l o e r homolo gy o f c i rcle bundles o f “la rge” degre e over Ri emann surface s.
author : - 'Thoma s E . M ark '
t i t le : 'Knotted su r f aces i n 4-man ifo lds'
-- -
Int rod uct io n
======= =====
S om eti me ag o, Fi n tushel a nd St er n i ntrod u ced atechn ique t he y ca lled “r i ms u rger y” f or c han gi ng th e em b edd ing ofa smoothsur f ace$\ Si gma$ in a closed 4-m an ifold $X$[@ FSr im; @F S a ddend].Their construction make s use of aknot$K\s ubset S^3 $,and ca n b e seenas aninsta nc e o f their e ar lie r“knot surg e r y”const ru ctio n, appl ied to the complem e ntof $\Sigma$.The int e r es tin g a s pec to f t h e construction i s that und er su itable con d iti on s, Fint ushel a nd St e rn were able toshow that t he r e s ult ing surfac e $\Sigm a_K\subse t X$ i s t opolo gic ally e qu iva lentto $\S i gma $, bu t smoo th ly dis tinct :rim surg ery results in a [*smoo th*]{} knot tin g of $\Si gma $ bu t not a [ *top ological*] {}one .
To en s ure t hatt he to p ologi calt ype of $( X ,\ Sig m a )$ is unchang e d bythe c ons t ructio n, i t suffices to ass u me that the co mple m e nt$Z$ of a r egular neighbo rho od o f $\Sigm a$ is simplyconnecte d( see [ @FSrim ; @boy er; @fr e e dm a n], al so R ema rk \[pi1r ema rk \ ] below ). O u r curr entco ncernis | connected complement_and nonnegative_self-intersection, then there are_infinitely many_topologically_equivalent but_smoothly_distinct embedded surfaces_homologous to $\Sigma$._Here we extend this_result to include_symplectic_surfaces whose self-intersection is bounded below by $2-2g$, where $g$ is the genus of_$\Sigma$.
_ __We_make use of tools from_Heegaard Floer theory, and include_several results_that may be of independent interest. Specifically we_give_an analogue for_[[Ozsváth-Szabó]{} ]{}invariants of the Fintushel-Stern knot surgery formula for_Seiberg-Witten invariants, both for closed 4-manifolds_and manifolds with_boundary._This_is based on a_formula for the [[Ozsváth-Szabó]{} ]{}invariants of_the result of a logarithmic transformation,_analogous to one obtained by Morgan-Mrowka-Szabó for_Seiberg-Witten invariants, and the results on_[[Ozsváth-Szabó]{} ]{}invariants of fiber sums_due to_the author and Jabuka. In_addition, we give_a calculation_of the twisted_Heegaard Floer homology of circle bundles_of “large” degree_over Riemann surfaces.
author:
- 'Thomas E. Mark'
title:_'Knotted_surfaces in 4-manifolds'
---
Introduction
============
Some_time_ago,_Fintushel and_Stern introduced a_technique_they called_“rim_surgery” for changing the embedding of_a_smooth surface $\Sigma$ in a closed 4-manifold_$X$ [@FSrim; @FSaddend]. Their_construction_makes use of a_knot $K\subset S^3$, and can_be seen as an instance of_their earlier_“knot surgery”_construction, applied to the complement of $\Sigma$. The interesting aspect of_the construction is that under suitable_conditions, Fintushel and Stern_were able_to_show that the_resulting_surface $\Sigma_K\subset_X$ is topologically equivalent to $\Sigma$, but_smoothly distinct:_rim surgery results in a [*smooth*]{}_knotting of $\Sigma$ but_not_a [*topological*]{} one.
To ensure that the_topological type of $(X,\Sigma)$ is unchanged_by the construction, it suffices_to_assume_that the complement $Z$ of_a regular neighborhood of $\Sigma$ is_simply connected (see_[@FSrim; @boyer; @freedman], also Remark \[pi1remark\] below)._Our_current concern is |
D^{i}$ and $\beta_i$ the weights from \[indicator\].
Basel III capital surcharges for G-SIBs
---------------------------------------
[p[3cm]{} p[3cm]{} p[3cm]{} p[6cm]{}]{} Bucket & Score range & Bucket thresholds & Higher loss absorbency requirement (common equity as a percentage of risk-weighted assets)\
5 & D-E & 530-629 & 3.50%\
4 & C-D & 430-529 & 2.50%\
3 & B-C & 330-429 & 2.00%\
2 & A-B & 230-329 & 1.50%\
1 & Cutoff point-A & 130-229 & 1.00%
In the Basel III “bucketing approach”, based on the scores from \[score\], banks are divided into four equally sized classes (buckets) of systemic importance, seen here in \[buckets\]. The cutoff score and bucket thresholds have been calibrated by the BCBS in such a way that the magnitude of the higher loss absorbency requirements for the highest populated bucket is 2.5% of risk-weighted assets, with an initially empty bucket of 3.5% of risk-weighted assets. The loss absorbency requirements for the lowest bucket is 1% of risk-weighted assets. The loss absorbency requirement is to be met with common equity [@BIS:2010aa]. Bucket five will initially be empty. As soon as the bucket becomes populated, a new bucket will be added in such a way that it is equal in size (scores) to each of the other populated buckets and the minimum higher loss absorbency requirement is increased by 1% of risk-weighted assets.
The model to test the efficiency of SR regulation {#model}
=================================================
We use an ABM, linking the financial and the real economy. The model consists of banks, firms and households. One pillar of the model is a well-studied macroeconomic model [@Gaffeo:2008aa; @Delli-Gatti:2008aa; @Delli-Gatti:2011aa], the second pillar is an implementation of an interbank market. In particular, we extend the model used in [@Poledna:2014aa] with an implementation of the Basel III indicator-based measurement approach and capital surcharges for SIBs. For a comprehensive description of the model, see [@Delli- | D^{i}$ and $ \beta_i$ the weights from \[indicator\ ].
Basel III capital surcharges for G - SIBs
---------------------------------------
[ p[3 cm ] { } p[3 cm ] { } p[3 cm ] { } p[6 cm ] { } ] { } Bucket & Score compass & Bucket doorway & Higher loss absorbency requirement (coarse equity as a percentage of hazard - weighted assets)\
5 & D - E & 530 - 629 & 3.50%\
4 & C - D & 430 - 529 & 2.50%\
3 & B - C & 330 - 429 & 2.00%\
2 & A - B & 230 - 329 & 1.50%\
1 & Cutoff point - A & 130 - 229 & 1.00%
In the Basel III “ bucketing access ”, based on the scores from \[score\ ], bank are divided into four equally sized classes (buckets) of systemic importance, seen here in \[buckets\ ]. The cutoff mark and bucket thresholds have been calibrated by the BCBS in such a way that the order of magnitude of the higher loss absorbency requirement for the highest populated bucket is 2.5% of hazard - weighted asset, with an initially empty bucket of 3.5% of risk - weighted assets. The loss absorbency necessity for the lowest bucket is 1% of risk - weighted assets. The loss absorbency necessity is to be met with common equity [ @BIS:2010aa ]. Bucket five will initially be empty. As soon as the bucket becomes populated, a new bucket will be added in such a way that it is equal in size (scores) to each of the early populated bucket and the minimum higher passing absorbency necessity is increased by 1% of risk - weighted asset.
The model to screen the efficiency of SR regulation { # model }
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
We practice an ABM, linking the financial and the real economy. The model consists of banks, firms and households. One pillar of the model is a well - studied macroeconomic model [ @Gaffeo:2008aa; @Delli - Gatti:2008aa; @Delli - Gatti:2011aa ], the second column is an implementation of an interbank grocery store. In particular, we extend the model used in [ @Poledna:2014aa ] with an implementation of the Basel III index - based measurement approach and capital surcharges for SIBs. For a comprehensive description of the model, see [ @Delli- | D^{i}$ wnd $\beta_i$ the weights frum \[indicator\].
Basgl III cepital aurchargds for G-SIBs
---------------------------------------
[p[3cm]{} p[3cm]{} p[3cm]{} p[6cm]{}]{} Uuckwt & Sxore range & Bucket thrdsholds & Jigher liss ebsorbency requicsment (common seuitv es a percentage of risk-wehghted assets)\
5 & D-D & 530-629 & 3.50%\
4 & C-D & 430-529 & 2.50%\
3 & B-C & 330-429 & 2.00%\
2 & A-B & 230-329 & 1.50%\
1 & Cutoff point-A & 130-229 & 1.00%
In the Basel III “fucksnikg approach”, based on the scores rrom \[scmre\], banks are divided into four equally sizfd classes (buckets) of systemix im[irtance, seen here in \[blekets\]. The cotoff score and bucket thresholdr havz been calivrqtef by the BCBW in fuch a way tmst the magnitide of the higmer lmss absorbency requiremeits for the highest kopulated tueket is 2.5% of risk-weighree assgts, whth xb ivitjaklg emptj bncket of 3.5% or risk-weighred assets. The loss afworbency requjremenes for the lowest bucket is 1% of risk-weighued aasets. The loss absorbenxy requirement is to he met wieh common equity [@BIS:2010aa]. Bucket five will initially be ejoty. Af woln as the bucket becomes populated, a new buckqf eikl be added in such a way tjay it is equal kn sizz (sdores) to each of tje othet popuoated buchets and the minimum higher losw absorbency eequirement is incxeased by 1% oy risk-eeighyed assets.
The model to cest tge efficienfy of SR ddgulation {#model}
=================================================
Wd uxe an ABM, linking the financyal and tie recl econooy. Tne modql consistd of ndnks, firms and houdeholbs. Ona pillar ov the model is a well-studied marcoeconomic mocen [@Gdffeo:2008aa; @Belli-Gstti:2008aa; @Delli-Gwtti:2011aa], the secpnd piljar ir an implejentatimn of an ineerbank marked. In particuler, we extqnd rhe nodel urdd in [@Poledna:2014as] with an implementqtion of the Basel IIK indicator-based meqsurement approsch anq bapmtal fgrcharges fos SICs. Wpr a zomprehtusiye aesctiption of the model, see [@Delli- | D^{i}$ and $\beta_i$ the weights from \[indicator\]. capital for G-SIBs [p[3cm]{} p[3cm]{} p[3cm]{} & thresholds & Higher absorbency requirement (common as a percentage of risk-weighted assets)\ & D-E & 530-629 & 3.50%\ 4 & C-D & 430-529 & 2.50%\ & B-C & 330-429 & 2.00%\ 2 & A-B & 230-329 & 1.50%\ & point-A 130-229 1.00% In the Basel III “bucketing approach”, based on the scores from \[score\], banks are divided four equally sized classes (buckets) of systemic importance, here in \[buckets\]. The score and bucket thresholds have calibrated the BCBS such way the magnitude of higher loss absorbency requirements for the highest populated bucket is 2.5% of risk-weighted assets, with an initially bucket of risk-weighted assets. loss requirements the lowest bucket of risk-weighted assets. The loss absorbency be met with common equity [@BIS:2010aa]. Bucket five initially be As soon as the bucket becomes a new bucket will be added in such way that it is equal in size (scores) to each of the other populated buckets minimum higher loss absorbency is increased by of assets. model test the of SR regulation {#model} ================================================= We use an ABM, linking the and the real economy. The model consists of banks, firms One of the model a well-studied macroeconomic model @Delli-Gatti:2008aa; the second pillar is of interbank we the used in [@Poledna:2014aa] with implementation of the Basel III measurement approach and capital comprehensive description of the model, see [@Delli- | D^{i}$ and $\beta_i$ the weights from \[iNdicator\].
BaSel IIi caPitAl SurcHargEs for G-SIBs
---------------------------------------
[p[3cm]{} p[3CM]{} p[3cm]{} P[6cm]{}]{} Bucket & Score range & BucKet thReSHoldS & hiGher lOss absoRBeNCY reQuIrEmeNt (COmMon eqUitY as a perCentage of rIsk-WeIghted assets)\
5 & d-e & 530-629 & 3.50%\
4 & C-d & 430-529 & 2.50%\
3 & B-C & 330-429 & 2.00%\
2 & A-B & 230-329 & 1.50%\
1 & CutofF poInt-A & 130-229 & 1.00%
In the BasEl IiI “buckEtIng APproaCh”, bAsed oN the scORes froM \[score\], banKs ARe diviDEd into fOUR eQualLy sized classes (bucKEtS) Of systemic impoRtance, SeEN hERE in \[BucKets\]. The cutOfF scorE And buckET tHREShoLDs have been calIbrated by thE bCBs in sucH a Way THat the MagniTuDE of The higher loSs abSorbency rEquireMEnts for THe higheSt popuLatEd bUckeT Is 2.5% Of RisK-wEIghTEd AssETs, wIth an iniTiAlLy empTy buCKET Of 3.5% of RisK-weiGhted Assets. The loss AbsOrbeNCy rEquirEmentS for ThE loweSt buckEt is 1% oF rIsk-weighted asseTs. ThE loss absoRbeNcY reQuIremeNT is to bE meT wiTh commoN equity [@biS:2010aA]. BUCKEt Five will initially bE eMPTy. as soon as The bucKEt BeCOmes popuLaTed, A new BUCket wIll bE AdDed in sucH a way tHAt It Is equal In Size (scOrEs) tO eaCh of tHE othEr popuLated bucKets aND the minimum higHEr loss absorbeNCy REQuIRemeNt iS increased bY 1% of rISk-weIghtED aSseTS.
The mOdel tO tESt THe efficiency of SR regUlAtion {#mOdel}
=================================================
WE use an ABM, linkIng the finaNCIAl and the Real ECoNOmy. The model conSists Of banks, firMS and housEholdS. One pillAr of the moDEL is a well-StuDieD maCroECOnOmic model [@GaffEO:2008Aa; @DeLlI-Gatti:2008aA; @DeLli-GattI:2011aa], The SecOnd PiLlar is an iMplementAtIoN oF aN inTerbaNK market. IN pArtIcUlaR, we exTEnd the Model Used In [@poLEdnA:2014aa] with AN iMPLemeNtAtIon oF thE BAsel IiI inDIcaTor-baseD measuremEnt APproAcH aNd capitAl surcharges fOr sIBs. For a coMpRehEnsive DEScriptioN of the model, see [@Delli- | D^{i}$ and $\beta_i$ the w eights fro m \[i ndi cat or \].
Bas el III capital surc harges for G-SIBs
---- ----- -- - ---- - -- ----- ------- - -- - - ---
[p [3c m] { }p[3cm ]{} p[3cm] {} p[6cm]{ }]{ }Bucket & Sco r erange & Bu cke t thresholds &Higher l oss absor ben cy re quirem e nt (co mmon equi ty as a p e rcentag e of ris k-weighted assets ) \5 & D-E & 530-6 29 & 3 .5 0 %\ 4 &C-D & 430-529 & 2.50 % \
3 & B - C& 3 30- 4 29 & 2.00%\
2 & A-B & 23 0 -32 9 & 1. 50 %\1 & Cut off p oi n t-A & 130-229& 1. 00%
In t he Bas e l III “ b ucketin g appr oac h”, bas e don th es cor e sfro m \[ score\], b an ks ar e di v i d e d in tofour equa lly sized cla sse s (b u cke ts) o f sys temi cimpor tance, seen h ere in \[bucket s\]. The cuto ffsc ore a nd bu c ket th res hol ds have been c a lib ra t e d b y the BCBS in such a w ay that th e magn i tu de of the h ig her los s absor benc y r equireme nts fo r t he highes tpopula te d b uck et is 2.5% of ri sk-weigh ted a s sets, with ani nitially empt y b u c ke t of3.5 % of risk-w eigh t ed a sset s .The lossabsor be n cy requirements for th elowest buck et is 1% of r isk-weight e d assets.Thel os s absorbency re quire ment is to be met w ith c ommon eq uity [@BI S : 2010aa]. Bu cke t f ive w il l initially b e empt y. As soo n a s the b uck etbec ome spopulated , a newbu ck et w ill be a d ded in s uc h a w aythati t is e qualin s iz e( sco res) to ea c h ofth eothe r p op ulate d bu c ket s and t he minimu m h i gher l os s absor bency require me nt is incr ea sed by 1% o f risk-w eighted assets.
The mo d el to t est theeffi ciency of SR regul ati o n {#mo del}
= ===== == === = = ===== = = == === == ========== = = === ===== ==
We use an ABM, linking thef ina ncial and the re al e c o no my. Th e mo de l co n s ists of banks,firms andho u se holds. One pil la r of th e model is a well-st udied mac roeconomi cmode l [@G affeo:2008 aa; @Del li-Gatti: 2 008aa ; @ Delli -Ga tti:20 11 aa] , the secon d pi llaris anim plemen tatio nof an in terbank market. In part icular , weext end the m ode l us ed in [@P oled na:2014aa] wi than im ple m entat iono fthe Basel III indicator - ba sed m ea surement ap p r o ach andcap i tal su rcha rges for SIBs. Fo r a comprehensi ve d e s cri pti o n of t he model, see[@D el l i - | D^{i}$ and_$\beta_i$ the_weights from \[indicator\].
Basel III_capital surcharges_for_G-SIBs
---------------------------------------
[p[3cm]{} p[3cm]{}_p[3cm]{}_p[6cm]{}]{} Bucket &_Score range &_Bucket thresholds & Higher_loss absorbency requirement_(common_equity as a percentage of risk-weighted assets)\
5 & D-E & 530-629 & 3.50%\
4 &_C-D_& 430-529_&_2.50%\
3_& B-C & 330-429 &_2.00%\
2 & A-B & 230-329_& 1.50%\
1_& Cutoff point-A & 130-229 & 1.00%
In the_Basel_III “bucketing approach”,_based on the scores from \[score\], banks are divided_into four equally sized classes (buckets)_of systemic importance,_seen_here_in \[buckets\]. The cutoff_score and bucket thresholds have been_calibrated by the BCBS in such_a way that the magnitude of the_higher loss absorbency requirements for the_highest populated bucket is 2.5%_of risk-weighted_assets, with an initially empty_bucket of 3.5%_of risk-weighted_assets. The loss_absorbency requirements for the lowest bucket_is 1% of_risk-weighted assets. The loss absorbency requirement_is_to be met_with_common_equity [@BIS:2010aa]._Bucket five will_initially_be empty._As_soon as the bucket becomes populated,_a_new bucket will be added in such_a way that it_is_equal in size (scores)_to each of the other_populated buckets and the minimum higher_loss absorbency_requirement is_increased by 1% of risk-weighted assets.
The model to test the efficiency_of SR regulation {#model}
=================================================
We use an_ABM, linking the financial_and the_real_economy. The model_consists_of banks,_firms and households. One pillar of the_model is_a well-studied macroeconomic model [@Gaffeo:2008aa; @Delli-Gatti:2008aa;_@Delli-Gatti:2011aa], the second pillar_is_an implementation of an interbank market._In particular, we extend the model_used in [@Poledna:2014aa] with an_implementation_of_the Basel III indicator-based measurement_approach and capital surcharges for SIBs._For a comprehensive_description of the model, see [@Delli- |
At {#at-12.unnumbered}
----------
The paper “$\alpha$-particle branching ratios for neutron-deficient astatine isotopes” by Latimer et al. reported the observation of $^{206}$At in 1961 [@1961Lat01]. Gold and platinum foils were irradiated with 50$-$125 MeV $^{12}$C and 65$-$130 MeV $^{14}$N beams, respectively, from the Berkeley HILAC. Alpha-particle spectra were measured with a gridded ionization chamber following chemical separation. “A least-squares analysis of several of the curves for which the statistics were good yielded a value of 29.5$\pm$0.6 min for the half-life of $^{206}$At.” This value agrees with the currently accepted value of 30.6(8) min. An earlier report of a 2.6 h half-life [@1951Bar01] was evidently incorrect. Also, about three months later Forsling et al. independently reported a 20(10) min half-life [@1961For01] and in 1959 Hoff et al. had reported a half-life of 31.0(15) min in a conference abstract [@1959Hof01].
$^{207}$At {#at-13.unnumbered}
----------
$^{207}$At was identified by Barton et al. and published in the 1951 paper “Radioactivity of astatine isotopes” [@1951Bar01]. $^{209}$Bi was irradiated with $^4$He beams of up to 380 MeV from the Berkeley 184-in. cyclotron. Alpha spectra were recorded with an alpha-pulse analyzer following chemical separation. “At 75 Mev a new activity appeared having a half-life of about 2 hr, and this has been assigned to At$^{207}$ formed by the ($\alpha$,6n) reaction.” This value agrees with the currently adopted value of 1.81(3) h.
$^{208}$At {#at-14.unnumbered}
----------
In 1950 Hyde et al. reported the first observation of $^{208}$At in the paper “Low mass francium and emanation isotopes of high alpha-stability” [@1950Hyd01]. Thorium foils were bombarded with up to 350 MeV protons from the Berkeley 184-inch cyclotron. $^{212}$Fr was chemically separated and $^{208}$At was populated by $\alpha$-dec | At { # at-12.unnumbered }
----------
The paper “ $ \alpha$-particle branching ratios for neutron - deficient astatine isotope ” by Latimer et al. report the observation of $ ^{206}$At in 1961 [ @1961Lat01 ]. Gold and platinum hydrofoil were irradiate with 50$-$125 MeV $ ^{12}$C and 65$-$130 MeV $ ^{14}$N beams, respectively, from the Berkeley HILAC. Alpha - particle spectrum were measure with a gridded ionization chamber come chemical separation. “ A least - squares psychoanalysis of several of the curves for which the statistic were good yielded a value of 29.5$\pm$0.6 min for the half - life of $ ^{206}$At. ” This value agrees with the currently accepted value of 30.6(8) min. An early report of a 2.6 h half - life [ @1951Bar01 ] was obviously incorrect. Also, approximately three months by and by Forsling et al. independently reported a 20(10) min half - life sentence [ @1961For01 ] and in 1959 Hoff et al. had reported a half - life of 31.0(15) min in a conference abstraction [ @1959Hof01 ].
$ ^{207}$At { # at-13.unnumbered }
----------
$ ^{207}$At was identified by Barton et al. and published in the 1951 paper “ Radioactivity of astatine isotopes ” [ @1951Bar01 ]. $ ^{209}$Bi was irradiated with $ ^4$He beams of up to 380 MeV from the Berkeley 184 - in. cyclotron. Alpha spectra were recorded with an alpha - pulse analyzer following chemical separation. “ At 75 Mev a new activity appeared having a half - life of approximately 2 hour, and this has been assigned to At$^{207}$ formed by the ($ \alpha$,6n) reaction. ” This value match with the currently adopted value of 1.81(3) h.
$ ^{208}$At { # at-14.unnumbered }
----------
In 1950 Hyde et al. reported the inaugural observation of $ ^{208}$At in the paper “ Low mass francium and emission isotopes of high alpha - stability ” [ @1950Hyd01 ]. Thorium foils were bombarded with up to 350 MeV protons from the Berkeley 184 - inch cyclotron. $ ^{212}$Fr was chemically separated and $ ^{208}$At was populated by $ \alpha$-dec | At {#wt-12.unnumbered}
----------
The paper “$\alkha$-particle branckung ravios fod neutrov-deficient astatine isotopes” bt Latumer et al. reported the observatpon of $^{206}$At in 1961 [@1961Oat01]. Gold ais platikbm fojps wzrx irradiated wijh 50$-$125 MeV $^{12}$C and 65$-$130 MeV $^{14}$N beams, rasoeetively, from the Berkeley HILAC. Alphw-particke spectra were ieaslrqd wjnh a gridded ionization chamber followpng chemical sepatation. “A least-squares analjsis of several of the curves for whisy the statisgics were good yielded a value of 29.5$\pm$0.6 min for the half-lkfe oy $^{206}$At.” This vqlye wcrees with vhe cugrently accepbvd valua of 30.6(8) mim. An earlier rcport of a 2.6 h half-life [@1951Bar01] was evidently incorrest. Also, atobt three months later Firslitg ed al. kbdeoenseitlg repogtev a 20(10) min halr-life [@1961For01] abd in 1959 Hoff et al. hac wvlorted a halr-life jf 31.0(15) min in a conference abstract [@1959Hof01].
$^{207}$At {#at-13.utnujbered}
----------
$^{207}$At was identified by Barton et al. and poblished ig the 1951 paper “Radioactivity of astatine isotopes” [@1951Bdr01]. $^{209}$Bi das ivradkqtfd with $^4$He beams of up to 380 MeV from the Berkelqg 184-on. byclotron. Alpha siectra were recordrd wojh an alpha-pulre analyzsr following chemifal sepwratiin. “At 75 Mev a nrw activity appeared having a half-life jd about 2 hr, and thid has been cssigngd to St$^{207}$ formed by the ($\alpha$,6n) readtion.” This galue agrsds with the currdntky adopted value of 1.81(3) h.
$^{208}$At {#at-14.ungumbered}
----------
Ii 1950 Hybe et al. fepotted thq first obdervabhon of $^{208}$At in the pwper “Pof mass frajcium and emanation isotopes of high alpha-staninitj” [@1950Hyd01]. Thoxium fpils were bomfarded with up to 350 MeV protuns from tge Berkxley 184-inch cyslotron. $^{212}$Fr wav chemically veparateq ane $^{208}$At was poojlated by $\alphs$-dec | At {#at-12.unnumbered} ---------- The paper “$\alpha$-particle branching neutron-deficient isotopes” by et al. reported 1961 Gold and platinum were irradiated with MeV $^{12}$C and 65$-$130 MeV $^{14}$N respectively, from the Berkeley HILAC. Alpha-particle spectra were measured with a gridded ionization following chemical separation. “A least-squares analysis of several of the curves for which statistics good a of 29.5$\pm$0.6 min for the half-life of $^{206}$At.” This value agrees with the currently accepted value 30.6(8) min. An earlier report of a 2.6 half-life [@1951Bar01] was evidently Also, about three months later et independently reported 20(10) half-life and in 1959 et al. had reported a half-life of 31.0(15) min in a conference abstract [@1959Hof01]. $^{207}$At {#at-13.unnumbered} ---------- was identified et al. published the paper “Radioactivity of [@1951Bar01]. $^{209}$Bi was irradiated with $^4$He to 380 MeV from the Berkeley 184-in. cyclotron. spectra were with an alpha-pulse analyzer following chemical “At 75 Mev a new activity appeared having half-life of about 2 hr, and this has been assigned to At$^{207}$ formed by the This value agrees with currently adopted value 1.81(3) $^{208}$At ---------- 1950 Hyde al. reported the first observation of $^{208}$At in the paper “Low francium and emanation isotopes of high alpha-stability” [@1950Hyd01]. Thorium foils with to 350 MeV from the Berkeley 184-inch $^{212}$Fr chemically separated and $^{208}$At by | At {#at-12.unnumbered}
----------
The paper “$\alpHa$-particle BrancHinG raTiOs foR neuTron-deficient aSTatiNe isotopes” by Latimer et aL. repoRtED the OBsErvatIon of $^{206}$At IN 1961 [@1961LAT01]. golD aNd PlaTiNUm Foils WerE irradiAted with 50$-$125 Mev $^{12}$C aNd 65$-$130 meV $^{14}$N beams, resPEcTively, from The berkeley HILAc. AlPha-parTiCle SPectrA weRe meaSured wITh a griDded ionizAtIOn chamBEr folloWINg ChemIcal separation. “A leASt-SQuares analysis Of seveRaL Of THE cuRveS for which tHe StatiSTics werE GoOD YIelDEd a value of 29.5$\pm$0.6 mIn for the halF-LifE of $^{206}$At.” THiS vaLUe agreEs witH tHE cuRrently accePted Value of 30.6(8) miN. An earLIer repoRT of a 2.6 h haLf-life [@1951bar01] Was EvidENtLy IncOrREct. aLsO, abOUt tHree montHs LaTer FoRsliNG ET Al. inDepEndeNtly rEported a 20(10) min haLf-lIfe [@1961FOR01] anD in 1959 HoFf et aL. had RePorteD a half-Life oF 31.0(15) mIn in a conference AbstRact [@1959Hof01].
$^{207}$At {#At-13.uNnUmbErEd}
----------
$^{207}$At wAS identIfiEd bY Barton Et al. and PUblIsHED In The 1951 paper “RadioactivItY OF aStatine iSotopeS” [@1951baR01]. $^{209}$BI Was irradIaTed With $^4$hE Beams Of up TO 380 MEV from thE BerkeLEy 184-In. CyclotrOn. alpha sPeCtrA weRe recORded With an Alpha-pulSe anaLYzer following cHEmical separatIOn. “aT 75 meV A new ActIvity appearEd haVIng a Half-LIfE of ABout 2 hR, and tHiS HaS Been assigned to At$^{207}$ forMeD by the ($\Alpha$,6N) reaction.” This Value agreeS WITh the curRentLY aDOpted value of 1.81(3) h.
$^{208}$AT {#at-14.unNumbered}
----------
In 1950 hYde et al. rEportEd the firSt observaTIOn of $^{208}$At in The PapEr “LOw mASS fRancium and emaNATion IsOtopes oF hiGh alpha-StaBilIty” [@1950hyd01]. thOrium foilS were bomBaRdEd WiTh uP to 350 Mev Protons fRoM thE BErkEley 184-iNCh cyclOtron. $^{212}$fr waS cHeMIcaLly sepaRAtED And $^{208}$AT wAs PopuLatEd By $\alpHa$-deC | At {#at-12.unnumbered}
--- -------
T he pa per “$ \a lpha $-pa rticle branchi n g ra tios for neutron-defic ientas t atin e i sotop es” byL at i m eret a l.re p or ted t heobserva tion of $^ {20 6} $At in 1961[ @1 961Lat01]. Go ld and plati num foils w ere irrad iat ed wi th 50$ - $125 M eV $^{12} $C and 65 $ -$130 M e V $ ^{14 }$N beams, respec t iv e ly, from the B erkele yH IL A C . A lph a-particle s pectr a were m e as u r e d w i th a griddedionizationc ham ber fo ll owi n g chem icalse p ara tion. “A le ast- squares a nalysi s of sev e ral ofthe cu rve s f or w h ic hthe s t ati s ti csw ere good yi el de d a v alue o f 29.5 $\p m$0. 6 min for the half -li fe o f $^ {206} $At.” Thi svalue agree s wit hthe currently a ccep ted value of 3 0.6 (8 ) min . An ea rli erreportof a 2. 6 hha l f - li fe [@1951Bar01] wa se v id ently in correc t .Al s o, about t hre e mo n t hs la terF or sling et al. i n de pe ndently r eporte da 2 0(1 0) mi n hal f-life [@1961F or01] and in 1959 Ho f f et al. hadr ep o r te d a h alf -life of 31 .0(1 5 ) mi n in acon f erenc e abs tr a ct [@1959Hof01].
$^{2 07 }$At { #at-1 3.unnumbered}
--------- - $^{207}$ At w a si dentified by B arton et al. an d publish ed in the 195 1 paper “ R a dioactiv ity of as tat i n eisotopes” [@1 9 5 1Bar 01 ]. $^{2 09} $Bi was ir rad iat edwi th $^4$He beams o fup t o380 MeVf rom theBe rke le y 1 84-in . cyclo tron. Alp ha s p ect ra were re c o rded w it h an al ph a-pul se a n aly zer fol lowing ch emi c al s ep ar ation.“At 75 Mev ane w activity a ppe ared h a v ing a ha lf-life of about 2 hr,a nd this ha s bee n as signed to At $^{207 }$f ormedby the ($\a lp ha$ , 6 n) re a c ti on. ”This value a gre es wi th the curren tly adopted valueo f 1 .81(3) h.
$^ {20 8}$A t {# at- 1 4. u nnu mb e red } ----------
In1950 Hydeet al . reported the f irst ob servati on of $^{208} $At in th e paper “ Lo w ma s s fr ancium and emanati on isotop e s ofh ig h alp ha- stabil it y”[@195 0Hyd01 ] . T horiu m foil swere b ombar de d with u p to 350 MeV protons fr om the Berk ele y 184-inc h c y clo tron. $^{ 212} $Fr was ch emi cal ly se par a ted a nd $ ^ {2 08} $ At wa s po p ulated by $\ alp h a $- dec | At {#at-12.unnumbered}
----------
The_paper “$\alpha$-particle_branching ratios for neutron-deficient_astatine isotopes”_by_Latimer et_al. reported_the observation of_$^{206}$At in 1961_[@1961Lat01]. Gold and platinum_foils were irradiated_with_50$-$125 MeV $^{12}$C and 65$-$130 MeV $^{14}$N beams, respectively, from the Berkeley HILAC. Alpha-particle spectra were_measured_with a_gridded_ionization_chamber following chemical separation. “A_least-squares analysis of several of_the curves_for which the statistics were good yielded a_value_of 29.5$\pm$0.6 min for_the half-life of $^{206}$At.” This value agrees with the_currently accepted value of 30.6(8) min. An_earlier report of_a_2.6 h_half-life [@1951Bar01] was evidently_incorrect. Also, about three months later_Forsling et al. independently reported a 20(10) min_half-life [@1961For01] and in 1959 Hoff et_al. had reported a half-life of 31.0(15) min_in a conference abstract [@1959Hof01].
$^{207}$At_{#at-13.unnumbered}
----------
$^{207}$At was_identified by Barton et al. and_published in the_1951 paper_“Radioactivity of astatine_isotopes” [@1951Bar01]. $^{209}$Bi was irradiated with_$^4$He beams of_up to 380 MeV from the Berkeley_184-in. cyclotron._Alpha spectra were_recorded_with_an alpha-pulse_analyzer following chemical_separation._“At 75 Mev_a_new activity appeared having a half-life_of_about 2 hr, and this has been assigned_to At$^{207}$ formed by_the_($\alpha$,6n) reaction.” This value_agrees with the currently adopted_value of 1.81(3) h.
$^{208}$At {#at-14.unnumbered}
----------
In 1950 Hyde_et al. reported_the first_observation of $^{208}$At in the paper “Low mass francium and emanation_isotopes of high alpha-stability” [@1950Hyd01]. Thorium_foils were bombarded with_up to_350 MeV_protons from the_Berkeley_184-inch cyclotron._$^{212}$Fr was chemically separated and $^{208}$At was_populated by_$\alpha$-dec |
3,199,525 01/01-14/02
**Instagram** 26,576 109,011 52,339 01/01-14/02
**Twitter** 1,187,482 - 390,866 27/01-14/02
**Total** 1,342,103 7,465,721 3,734,815
: Data breakdown of the number of posts, comments and users for all platforms.[]{data-label="tab:databreak"}
Text analysis {#subsec:Clustering_procedure}
-------------
To provide an overview of the debate concerning the virus outbreak on the various platforms, we extract and analyze all topics related to COVID-19 by applying Natural Language Processing techniques to the written content of each social media. We first build word embedding for the text corpus of each platform, then, to assess the topics around which the perception of the COVID-19 debate is concentrated, we cluster words by running the Partitioning Around Medoids (PAM) algorithm on their vector representations.
Word embeddings, i.e., distributed representations of words learned by neural networks, represent words as vectors in ${\mathbf R}^n$ bringing similar words closer to each other. They perform significantly better than the well-known Latent Semantic Analysis (LSA) and Latent Dirichlet Allocation (LDA) for preserving linear regularities among words and computational efficiency on large data sets [@Mikolov2013b]. In this paper we use the Skip-gram model [@Mikolov2013] to construct word embedding of each social media corpus. More formally, given a content represented by the sequence of words $w_1,w_2,\dots,w_T$, we use stochastic gradient descent with gradient computed through backpropagation rule [@Rumelhart1986] for maximizing the average log probability $$\frac{1}{T}\displaystyle\sum_{t=1}^T\left[\displaystyle\sum_{j=-k}^k\log p(w_{t+j}\vert w_t)\right]$$ where $k$ is the size of the training window. Therefore, during training the vector representations of closely related words are pushed to be close to each other.
In the Skip-gram model, every word $w$ is associated with its input and output vectors, $u_w$ and $v_w$, respectively. The | 3,199,525 01/01 - 14/02
* * Instagram * * 26,576 109,011 52,339 01/01 - 14/02
* * Twitter * * 1,187,482 - 390,866 27/01 - 14/02
* * Total * * 1,342,103 7,465,721 3,734,815
: Data breakdown of the number of position, gossip and users for all platforms.[]{data - label="tab: databreak " }
Text analysis { # subsec: Clustering_procedure }
-------------
To put up an overview of the debate concerning the virus outbreak on the versatile platform, we extract and analyze all subject related to COVID-19 by applying Natural Language Processing proficiency to the written content of each social medium. We first build word embedding for the text corpus of each chopine, then, to assess the topics around which the perception of the COVID-19 argument is concentrated, we cluster words by running the Partitioning Around Medoids (PAM) algorithm on their vector representations.
Word embeddings, i.e., distribute representations of words learned by neural networks, present words as vectors in $ { \mathbf R}^n$ bringing similar words closer to each other. They perform significantly better than the well - known Latent Semantic Analysis (LSA) and Latent Dirichlet Allocation (LDA) for preserving linear regularities among words and computational efficiency on large data hardening [ @Mikolov2013b ]. In this paper we use the Skip - gram mannequin [ @Mikolov2013 ] to manufacture word embedding of each social media corpus. More formally, give a content represented by the sequence of words $ w_1,w_2,\dots, w_T$, we practice stochastic gradient descent with gradient computed through backpropagation rule [ @Rumelhart1986 ] for maximizing the average log probability $ $ \frac{1}{T}\displaystyle\sum_{t=1}^T\left[\displaystyle\sum_{j=-k}^k\log p(w_{t+j}\vert w_t)\right]$$ where $ k$ is the size of the training window. Therefore, during training the vector representations of closely refer bible are pushed to be close to each other.
In the Skip - gram model, every word $ w$ is consociate with its input and output vectors, $ u_w$ and $ v_w$, respectively. The | 3,199,525 01/01-14/02
**Instagram** 26,576 109,011 52,339 01/01-14/02
**Jwutter** 1,187,482 - 390,866 27/01-14/02
**Total** 1,342,103 7,465,721 3,734,815
: Data breakdown of the numbvr of posrs, cinments and users for all ilatfmcms.[]{data-label="tab:catabreak"}
Taxt analysis {#sgbree:Clustering_procedure}
-------------
To provide an ovqrview pf the debate cogcermyng fhe virus outbreak on the various llatforks, we extract and analyze all topics repatef to COVID-19 by appljing Naturao Lagtuage Procesring technpxues to thg written content of each social medic. We first vuuld ford embeddmng fog the text covius of aach plstform, then, to asvesw the topics around wiich the perception jf the COEIB-19 debate is concentraree, we wlusder dirdr bg cunhing tje 'artitioninf Around Meeoids (PAM) algorithm og their vector reprefegtations.
Word embeddings, i.e., distributed geprssentations of words leqrned by neural netwotks, represqnt words as vectors in ${\mathbf R}^n$ bringing similas worvs cliscr tu ewch other. They perform significantly better trzn tme well-known Latcnt Semantic Analyxid (KFA) and Latent Dirickmef Allocation (LDA) flr presgrving linear rtgulatities among words and compytational efyicuency on large datc sets [@Mikoluv2013b]. On thos paper we use the Ski'-gram jodel [@Mikollv2013] to conagruct word embedainn ox each social media corpus. More fornallv, given x comtent wepresentef by bve sequence of worfs $w_1,w_2,\bots,w_D$, we use shochastic gradient descent with gradient compoteg tvrough bcckproiagation rule [@Rtmelhart1986] for mcximiziny the xverage lof probauility $$\frac{1}{T}\qisplaystyle\sgl_{t=1}^T\left[\displeystyle\sui_{j=-k}^k\oog p(w_{t+j}\verg w_t)\right]$$ where $k$ is the size of tye training window. Thgrsfore, during trcnnung the vector tepfesqnnatmons jx closely renatea wutds afe pushtb ti be vlose to each other.
It ths Skip-gram model, efevy word $w$ is assosiated with iys input and outpuu vectmrs, $u_w$ amd $d_w$, respectively. The | 3,199,525 01/01-14/02 **Instagram** 26,576 109,011 52,339 01/01-14/02 - 27/01-14/02 **Total** 7,465,721 3,734,815 : of comments and users all platforms.[]{data-label="tab:databreak"} Text {#subsec:Clustering_procedure} ------------- To provide an overview the debate concerning the virus outbreak on the various platforms, we extract and all topics related to COVID-19 by applying Natural Language Processing techniques to the content each media. first build word embedding for the text corpus of each platform, then, to assess the topics which the perception of the COVID-19 debate is we cluster words by the Partitioning Around Medoids (PAM) on vector representations. embeddings, distributed of words learned neural networks, represent words as vectors in ${\mathbf R}^n$ bringing similar words closer to each other. They significantly better well-known Latent Analysis and Dirichlet Allocation (LDA) linear regularities among words and computational data sets [@Mikolov2013b]. In this paper we use Skip-gram model to construct word embedding of each media corpus. More formally, given a content represented the sequence of words $w_1,w_2,\dots,w_T$, we use stochastic gradient descent with gradient computed through backpropagation for maximizing the average probability $$\frac{1}{T}\displaystyle\sum_{t=1}^T\left[\displaystyle\sum_{j=-k}^k\log p(w_{t+j}\vert where is size the training Therefore, during training the vector representations of closely related words are to be close to each other. In the Skip-gram model, $w$ associated with its and output vectors, $u_w$ $v_w$, The | 3,199,525 01/01-14/02
**Instagram** 26,576 109,011 52,339 01/01-14/02
**Twitter** 1,187,482 - 390,866 27/01-14/02
**Total** 1,342,103 7,465,721 3,734,815
: Data Breakdown oF the nUmbEr oF pOsts, CommEnts and users foR All pLatforms.[]{data-label="tab:daTabreAk"}
tExt aNAlYsis {#sUbsec:ClUStERIng_PrOcEduRe}
-------------
tO pRovidE an OvervieW of the debaTe cOnCerning the viRUs Outbreak on The Various platfOrmS, we extRaCt aND analYze All toPics reLAted to cOVID-19 by apPlYIng NatURal LangUAGe procEssing techniques tO ThE Written content Of each SoCIaL MEdiA. We First build WoRd embEDding foR ThE TEXt cORpus of each plaTform, then, to ASseSs the tOpIcs ARound wHich tHe PErcEption of the cOVId-19 debate is ConcenTRated, we CLuster wOrds by RunNinG the pArTiTioNiNG ArOUnD MeDOidS (PAM) algoRiThM on thEir vECTOR repResEntaTions.
word embeddingS, i.e., DistRIbuTed rePreseNtatIoNs of wOrds leArned By Neural networks, rEpreSent words As vEcTorS iN ${\mathBF R}^n$ briNgiNg sImilar wOrds cloSEr tO eACH OtHer. They perform signIfICAnTly betteR than tHE wElL-Known LatEnT SeMantIC analySis (Lsa) aNd Latent dirichLEt alLocatioN (LdA) for pReSerVinG lineAR regUlaritIes among Words ANd computationaL Efficiency on lARgE DAtA Sets [@mikOlov2013b]. In this PapeR We usE the sKiP-grAM modeL [@MikoLoV2013] To COnstruct word embeddiNg Of each SociaL media corpus. MOre formallY, GIVen a contEnt rEPrESented by the seqUence Of words $w_1,w_2,\dOTs,w_T$, we usE stocHastic grAdient desCENt with grAdiEnt ComPutED ThRough backpropAGAtioN rUle [@RumeLhaRt1986] for maXimIziNg tHe aVeRage log prObabilitY $$\fRaC{1}{T}\DiSplAystyLE\sum_{t=1}^T\leFt[\DisPlAysTyle\sUM_{j=-k}^k\loG p(w_{t+j}\Vert W_t)\RiGHt]$$ wHere $k$ is THe SIZe of ThE tRainIng WiNdow. THereFOre, During tRaining thE veCTor rEpReSentatiOns of closely rElAted words aRe PusHed to bE CLose to eaCh other.
In the Skip-gram modEL, every wOrd $W$ is asSociAted with iTs iNput anD ouTPut vecTors, $u_w$ And $v_w$, ReSpeCTIvely. tHE | 3,199,525 01/01-14/02
**Instagr am** 26, 57 6 109,011 5 2,33 9 01/01-14/02
**Twi tt e r** 1,187 ,482 - 3 9 0, 866 2 7/01-14 /02
* *To ta l** 1,34 2 ,1 03 7,46 5,7 21 3,734 ,81 5
: Da t a bre akd own o f then umberof posts, c o mments and use r s f or a ll platforms.[]{d a ta - label="tab:dat abreak "}
T e x t a nal ysis {#sub se c:Clu s tering_ p ro c e d ure }
------------ -
To provi d e a n over vi ewo f thedebat ec onc erning theviru s outbrea k on t h e vario u s platf orms,weext ract an dana ly z e a l ltop i csrelatedto C OVID- 19 b y a p plyi ngNatu ral L anguage Proce ssi ng t e chn iques to t he w ri ttenconten t ofea ch social media . We first bu ild w ord e mbedd i ng for th e t ext cor pus ofe ach p l a t fo rm, then, to asses st h etopics a roundw hi ch the perc ep tio n of t he CO VID- 1 9debate i s conc e nt ra ted, we c luster w ord s b y run n ingthe Pa rtitioni ng Ar o und Medoids (P A M) algorithmo nt h ei r vec tor representa tion s .
W orde mb edd i ngs,i.e., d i st r ibuted representati on s of w ordslearned by ne ural netwo r k s , repres entw or d s as vectors i n ${\ mathbf R}^ n $ bringi ng si milar wo rds close r to eachoth er. Th eyp e rf orm significa n t ly b et ter tha n t he well -kn own La ten tSemanticAnalysis ( LS A) a ndLaten t Dirichl et Al lo cat ion ( L DA) fo r pre serv in gl ine ar regu l ar i t iesam on g wo rds a nd co mput a tio nal eff iciency o n l a rgeda ta sets [ @Mikolov2013b ]. In this p ap erwe use t he Skip- gram model [@Mikolov201 3 ] to co nst ructword embeddin g o f each so c ial me dia co rpus. M ore f ormal l y ,giv en a content r epr esent ed bythe seq uence of words $w_ 1 ,w_ 2,\dots,w_T$, we use s to cha s ti c gr ad i ent d escent with gra dient comp ut e dthrough ba c kpr op agation rule [ @Rume l hart198 6] for ma ximizingth e av e r age log proba bility $ $\frac{1} { T}\di s pl aysty le\ sum_{t =1 }^T \left [\disp l ays tyle\ sum_{j =- k}^k\l og p( w_ {t+j}\ve rt w_t)\right]$$ where$k$ is thesiz e of thetra i nin g window. The refore, du rin g t raini ngt he ve ctor re pre s entat ions of closel y r ela t e dwords are p u s h edto be cl o se toeach other.
In the S k ip-gram model, eve r y wo rd$ w$ i sassociated wit h i ts i nput and o utput vecto rs, $u_w $a nd $v _w$, r espect ively.T h e | 3,199,525 _ 01/01-14/02
_ **Instagram** _ _26,576_ __ _109,011 _ _52,339 __ 01/01-14/02
**Twitter** 1,187,482 __ _-__ _ 390,866 _ _27/01-14/02
**Total** __ 1,342,103 _ 7,465,721 3,734,815 _
: Data breakdown of_the number of_posts,_comments_and users for all_platforms.[]{data-label="tab:databreak"}
Text analysis {#subsec:Clustering_procedure}
-------------
To provide an overview_of the debate concerning the virus_outbreak on the various platforms, we extract_and analyze all topics related to_COVID-19 by applying Natural Language_Processing techniques_to the written content of_each social media._We first_build word embedding_for the text corpus of each_platform, then, to_assess the topics around which the_perception_of the COVID-19_debate_is_concentrated, we_cluster words by_running_the Partitioning_Around_Medoids (PAM) algorithm on their vector_representations.
Word_embeddings, i.e., distributed representations of words learned_by neural networks, represent_words_as vectors in ${\mathbf_R}^n$ bringing similar words closer_to each other. They perform significantly_better than_the well-known_Latent Semantic Analysis (LSA) and Latent Dirichlet Allocation (LDA) for preserving_linear regularities among words and computational_efficiency on large data_sets [@Mikolov2013b]._In_this paper we_use_the Skip-gram_model [@Mikolov2013] to construct word embedding of_each social_media corpus. More formally, given a_content represented by the_sequence_of words $w_1,w_2,\dots,w_T$, we use stochastic_gradient descent with gradient computed through_backpropagation rule [@Rumelhart1986] for maximizing_the_average_log probability $$\frac{1}{T}\displaystyle\sum_{t=1}^T\left[\displaystyle\sum_{j=-k}^k\log p(w_{t+j}\vert w_t)\right]$$_where $k$ is the size of_the training window._Therefore, during training the vector representations of_closely_related words are pushed to be_close_to each other.
In the Skip-gram model,_every_word_$w$ is associated with its_input and output vectors, $u_w$ and_$v_w$, respectively. The |
i,0}| \right) = O_P(\omega_n).\end{aligned}$$ Combining the results, we obtain the statement of the lemma. $\blacksquare$
Let $$\begin{aligned}
\label{Wn}
V &\equiv& \frac{1}{n}\sum_{i_1,i_2 \in N: \overline N(i_1) \cap \overline N(i_2) \ne \varnothing} \left(\hat q_{i_1,S}\hat q_{i_2,S} - q_{i_1,S} q_{i_2,S}\right), \text{ and }\\ \notag
W &\equiv& \frac{1}{n}\sum_{i_1,i_2 \in N: \overline N(i_1) \cap \overline N(i_2) \ne \varnothing}(q_{i_1,S} q_{i_2,S} - \mathbf{E}_\mathscr{F}[ q_{i_1,S} q_{i_2,S}]).\end{aligned}$$
\[lemm: W\] $$\begin{aligned}
V = o_P(1), \text{ and } W = o_P(1).
\end{aligned}$$
**Proof:** The first statement immediately comes by following the proof of Lemma B.21 of [@He/Song:18:WP] and applying Lemma \[lemm: q2\]. The second statement follows precisely as in the same way as Lemma B.22 of [@He/Song:18:WP]. $\blacksquare$
\[lemm: consistency var estimator\] $\hat \sigma_S^2 = \sigma_S^2 + o_P(1)$, where $\sigma_S^2$ is as defined in (\[Var\]).
**Proof:** We can rewrite $$\begin{aligned}
\label{tilde sigma2}
\sigma_S^2 = \frac{1}{nv^4} \sum_{i_1,i_2 \in N: \overline N(i_1) \cap \overline N(i_2) \ne \varnothing} \mathbf{E}_\mathscr{F}[q_{i_1,S} q_{i_2,S}].\end{aligned}$$ | i,0}| \right) = O_P(\omega_n).\end{aligned}$$ Combining the results, we obtain the statement of the lemma. $ \blacksquare$
lease $ $ \begin{aligned }
\label{Wn }
V & \equiv & \frac{1}{n}\sum_{i_1,i_2 \in N: \overline N(i_1) \cap \overline N(i_2) \ne \varnothing } \left(\hat q_{i_1,S}\hat q_{i_2,S } - q_{i_1,S } q_{i_2,S}\right), \text { and } \\ \notag
W & \equiv & \frac{1}{n}\sum_{i_1,i_2 \in N: \overline N(i_1) \cap \overline N(i_2) \ne \varnothing}(q_{i_1,S } q_{i_2,S } - \mathbf{E}_\mathscr{F } [ q_{i_1,S } q_{i_2,S}]).\end{aligned}$$
\[lemm: W\ ] $ $ \begin{aligned }
five = o_P(1), \text { and } W = o_P(1).
\end{aligned}$$
* * Proof :* * The first statement immediately arrive by following the proof of Lemma B.21 of [ @He / Song:18: WP ] and lend oneself Lemma \[lemm: q2\ ]. The second statement follows precisely as in the like direction as Lemma B.22 of [ @He / Song:18: WP ]. $ \blacksquare$
\[lemm: consistency var estimator\ ] $ \hat \sigma_S^2 = \sigma_S^2 + o_P(1)$, where $ \sigma_S^2 $ is equally defined in (\[Var\ ]).
* * Proof :* * We can rewrite $ $ \begin{aligned }
\label{tilde sigma2 }
\sigma_S^2 = \frac{1}{nv^4 } \sum_{i_1,i_2 \in N: \overline N(i_1) \cap \overline N(i_2) \ne \varnothing } \mathbf{E}_\mathscr{F}[q_{i_1,S } q_{i_2,S}].\end{aligned}$$ | i,0}| \rlght) = O_P(\omega_n).\end{aligned}$$ Combining the tewults, xe obtajn the sgatement of the lemma. $\blackseuqre$
Leu $$\begin{aligned}
\label{Dn}
V &\equiv& \frac{1}{n}\sun_{i_1,i_2 \mn N: \overline N(i_1) \cap \ovevjine K(i_2) \ne \tarnothing} \left(\mat q_{i_1,S}\hat x_{i_2,S} - q_{i_1,S} q_{i_2,S}\richg), \text{ and }\\ \notag
W &\equiv& \frac{1}{n}\sum_{i_1,i_2 \yn N: \ovrrpine N(i_1) \cap \ovgrlint N(y_2) \ne \navnothing}(q_{i_1,S} q_{i_2,S} - \mathbf{E}_\mathscr{F}[ q_{i_1,S} q_{i_2,V}]).\end{aligned}$$
\[lekm: W\] $$\begin{aligned}
V = o_P(1), \text{ and } W = o_P(1).
\end{alignee}$$
**Projd:** The first rtatement pkmediately comes by following the proof ow Lemka B.21 of [@He/Wobg:18:WO] and applyiig Lemia \[lemm: q2\]. Thc secong statekent follows pvecisxly qs in the same way as Lemma B.22 of [@He/Song:18:RP]. $\blacksxucre$
\[lemm: consistency vqr estikatos\] $\hag \siema_A^2 = \sjgma_S^2 + o_P(1)$, where $\sigja_S^2$ is as dwfined in (\[Var\]).
**Proof:** Ee bsn rewrite $$\bsgin{alygged}
\label{tilde sigma2}
\sigma_S^2 = \frac{1}{nv^4} \sum_{i_1,i_2 \in N: \overline N(i_1) \cap \overoine N(i_2) \ne \varnothing} \mathbf{E}_\mwthscr{F}[q_{i_1,S} q_{i_2,S}].\end{aligned}$$ | i,0}| \right) = O_P(\omega_n).\end{aligned}$$ Combining the results, the of the $\blacksquare$ Let $$\begin{aligned} N: N(i_1) \cap \overline \ne \varnothing} \left(\hat q_{i_2,S} - q_{i_1,S} q_{i_2,S}\right), \text{ and \notag W &\equiv& \frac{1}{n}\sum_{i_1,i_2 \in N: \overline N(i_1) \cap \overline N(i_2) \ne \varnothing}(q_{i_1,S} - \mathbf{E}_\mathscr{F}[ q_{i_1,S} q_{i_2,S}]).\end{aligned}$$ \[lemm: W\] $$\begin{aligned} V = o_P(1), \text{ and } = \end{aligned}$$ The statement immediately comes by following the proof of Lemma B.21 of [@He/Song:18:WP] and applying Lemma \[lemm: The second statement follows precisely as in the way as Lemma B.22 [@He/Song:18:WP]. $\blacksquare$ \[lemm: consistency var $\hat = \sigma_S^2 o_P(1)$, $\sigma_S^2$ as defined in **Proof:** We can rewrite $$\begin{aligned} \label{tilde sigma2} \sigma_S^2 = \frac{1}{nv^4} \sum_{i_1,i_2 \in N: \overline N(i_1) \cap \overline \ne \varnothing} | i,0}| \right) = O_P(\omega_n).\end{aligned}$$ COmbining thE resuLts, We oBtAin tHe stAtement of the leMMa. $\blAcksquare$
Let $$\begin{alignEd}
\labEl{wN}
V &\eqUIv& \Frac{1}{n}\Sum_{i_1,i_2 \in n: \OvERLinE N(I_1) \cAp \oVeRLiNe N(i_2) \nE \vaRnothinG} \left(\hat q_{i_1,s}\haT q_{I_2,S} - q_{i_1,S} q_{i_2,S}\righT), \TeXt{ and }\\ \notag
w &\eqUiv& \frac{1}{n}\sum_{i_1,I_2 \in n: \overlInE N(i_1) \CAp \oveRliNe N(i_2) \nE \varnoTHing}(q_{i_1,s} q_{i_2,S} - \mathbF{E}_\MAthscr{f}[ Q_{i_1,S} q_{i_2,S}]).\eND{AlIgneD}$$
\[lemm: W\] $$\begin{aligneD}
v = o_p(1), \Text{ and } W = o_P(1).
\end{aLigned}$$
**prOOf:** tHE fiRst Statement iMmEdiatELy comes BY fOLLOwiNG the proof of LeMma B.21 of [@He/SonG:18:wP] aNd applYiNg LEMma \[lemM: q2\]. The SeCOnd Statement foLlowS preciselY as in tHE same waY As Lemma b.22 of [@He/SOng:18:wP]. $\bLackSQuArE$
\[leMm: COnsIStEncY Var EstimatoR\] $\hAt \Sigma_s^2 = \sigMA_s^2 + O_p(1)$, wheRe $\sIgma_s^2$ is as Defined in (\[Var\]).
**PRooF:** We cAN reWrite $$\Begin{AligNeD}
\labeL{tilde Sigma2}
\SiGma_S^2 = \frac{1}{nv^4} \sum_{i_1,i_2 \In N: \oVerline N(i_1) \Cap \OvErlInE N(i_2) \ne \VArnothIng} \MatHbf{E}_\matHscr{F}[q_{i_1,s} Q_{i_2,S}].\EnD{ALIgNed}$$ | i,0}| \right) = O_P(\omega _n).\end{a ligne d}$ $ C om bini ng t he results, we obta in the statement of th e lem ma . $\b l ac ksqua re$
Le t $ $ \ beg in {a lig ne d }\labe l{W n}
V &\ equiv& \fr ac{ 1} {n}\sum_{i_1 , i_ 2 \in N: \ ove rline N(i_1) \c ap \ov er lin e N(i_ 2)\ne \ varnot h ing} \ left(\hat q _ {i_1,S } \hat q_ { i _2 ,S}- q_{i_1,S} q_{i_ 2 ,S } \right), \tex t{ and } \ \\ n ota g
W &\equiv&\f rac{1 } {n}\sum _ {i _ 1 , i_2 \in N: \overl ine N(i_1)\ cap \over li neN (i_2)\ne \ va r not hing}(q_{i_ 1,S} q_{i_2,S } - \m a thbf{E} _ \mathsc r{F}[q_{ i_1 ,S}q _{ i_ 2,S }] ) .\e n d{ ali g ned }$$
\[l em m: W\]$$\b e g i n {ali gne d}
V = o_P(1), \t ext{ and } W= o_P (1).
\e nd{ali gned} $$
**Proof:** Th e fi rst state men timm ed iatel y comes by fo llowing the pr o ofof L e mm a B.21 of [@He/Son g: 1 8 :W P] and a pplyin g L em m a \[lemm :q2\ ]. T h e seco nd s t at ement fo llowsp re ci sely as i n thesa meway as L e mmaB.22 o f [@He/S ong:1 8 :WP]. $\blacks q uare$
\[lemm : c o n si s tenc y v ar estimato r\]$ \hat \si g ma _S^ 2 = \s igma_ S^ 2 + o_P(1)$, where $\si gm a_S^2$ is a s defined in(\[Var\]). * *Proof:* * We ca n rewrite $$\be gin{a ligned}
\l a bel{tild e sig ma2}
\si gma_S^2 = \ frac{1}{ nv^ 4}\su m_{ i _ 1, i_2 \in N: \o v e rlin eN(i_1)\ca p \over lin e N (i_ 2)\n e \varnot hing} \m at hb f{ E} _\m athsc r {F}[q_{i _1 ,S} q _{i _2,S} ] .\end{ align ed}$ $ | i,0}| \right)_= O_P(\omega_n).\end{aligned}$$_Combining the results, we_obtain the_statement_of the_lemma._$\blacksquare$
Let $$\begin{aligned}
\label{Wn}
V &\equiv&_\frac{1}{n}\sum_{i_1,i_2 \in N:_\overline N(i_1) \cap \overline_N(i_2) \ne \varnothing}_\left(\hat_q_{i_1,S}\hat q_{i_2,S} - q_{i_1,S} q_{i_2,S}\right), \text{ and }\\ \notag
W &\equiv& \frac{1}{n}\sum_{i_1,i_2 \in N:_\overline_N(i_1) \cap_\overline_N(i_2)_\ne \varnothing}(q_{i_1,S} q_{i_2,S} - \mathbf{E}_\mathscr{F}[_q_{i_1,S} q_{i_2,S}]).\end{aligned}$$
\[lemm: W\] $$\begin{aligned}
_ _ V = o_P(1), \text{_and_} W =_o_P(1).
\end{aligned}$$
**Proof:** The first statement immediately_comes by following the proof of_Lemma B.21 of_[@He/Song:18:WP]_and_applying Lemma \[lemm: q2\]._The second statement follows precisely as_in the same way as Lemma_B.22 of [@He/Song:18:WP]. $\blacksquare$
\[lemm: consistency var estimator\]_$\hat \sigma_S^2 = \sigma_S^2 + o_P(1)$,_where $\sigma_S^2$ is as defined_in (\[Var\]).
**Proof:**_We can rewrite $$\begin{aligned}
\label{tilde sigma2}
\sigma_S^2_= \frac{1}{nv^4} \sum_{i_1,i_2_\in N:_\overline N(i_1) \cap_\overline N(i_2) \ne \varnothing} \mathbf{E}_\mathscr{F}[q_{i_1,S} q_{i_2,S}].\end{aligned}$$ |
i \delta \varphi \Rightarrow B_i = 0$$ i.e. there is no propagating transverse vector mode. Therefore, if we redefine $\dfrac{1}{\varphi'}\partial_i \delta \varphi = \partial_i V$ then we obtain the results from. Therefore, the scalar sector for Generalized Einstein-Aether and the Khronon are completely equivalent [@Blas2], up to a redefinition of the coefficients discussed previously.
Equations of state for perturbations {#sect:EoS}
====================================
Scalar sector {#scalar-sector}
-------------
We now derive the equations of state, $\Gamma$ and $\Pi^{S,V,T}$, in terms of the other perturbation variables by fully eliminating the internal degrees of freedom introduced by the theory i.e. $V$, $B^i$, and their derivatives. In the scalar sector we do this via the expressions for $\delta \rho$ and $\theta^S$. Let us first work in the conformal Newtonian gauge. Initially it may not seem possible to eliminate the degrees of freedom as we have that $\theta^S \equiv \theta^S (V,V',V'')$ and $\delta \rho \equiv \delta \rho(V,V')$, i.e. we have three unknowns and only two equations. However, we can use the perturbed Aether field equation of motion to reduce the dimensionality of the problem. Using this to eliminate $V''$ in and gathering terms in $V$ and $V'$, we find that $$\begin{aligned}
a^2 \delta \rho =& \hspace{1mm} c_{14}\mathcal{F}_\mathcal{K}k^2 V' - \left[\alpha \mathcal{F}_\mathcal{K} -c_{14}\mathcal{F}_\mathcal{K} + \frac{6 \alpha^2 \mathcal{F}_\mathcal{KK} \mathcal{H}^2}{a^2 M^2} \right] \mathcal{H}k^2 V \nonumber \\
+\hspace{1mm}&c_{14}\mathcal{F}_\mathcal{K} k^2 \Psi - 3\alpha \mathcal{H}\left(\mathcal{F}_\mathcal{K} + \frac{6\alpha \mathcal{F}_\mathcal{KK}\mathcal{H}^2}{a^2 M^2} \right) | i \delta \varphi \Rightarrow B_i = 0$$ i.e. there is no propagating transverse vector mood. consequently, if we redefine $ \dfrac{1}{\varphi'}\partial_i \delta \varphi = \partial_i V$ then we obtain the results from. consequently, the scalar sector for Generalized Einstein - Aether and the Khronon are completely equivalent [ @Blas2 ], up to a redefinition of the coefficient discussed previously.
Equations of state of matter for perturbations { # sect: EoS }
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Scalar sector { # scalar - sector }
-------------
We nowadays derive the equations of department of state, $ \Gamma$ and $ \Pi^{S, V, T}$, in terms of the other disturbance variables by fully eliminating the internal degrees of exemption introduced by the theory i.e. $ V$, $ B^i$, and their derivative. In the scalar sector we do this via the expressions for $ \delta \rho$ and $ \theta^S$. permit us first function in the conformal Newtonian gauge. Initially it may not seem possible to eliminate the degrees of exemption as we have that $ \theta^S \equiv \theta^S (V, V',V'')$ and $ \delta \rho \equiv \delta \rho(V, V')$, i.e. we have three unknowns and only two equations. However, we can use the perturb Aether field equation of motion to reduce the dimensionality of the problem. Using this to eliminate $ V''$ in and gathering terms in $ V$ and $ V'$, we find that $ $ \begin{aligned }
a^2 \delta \rho = & \hspace{1 mm } c_{14}\mathcal{F}_\mathcal{K}k^2 V' - \left[\alpha \mathcal{F}_\mathcal{K } -c_{14}\mathcal{F}_\mathcal{K } + \frac{6 \alpha^2 \mathcal{F}_\mathcal{KK } \mathcal{H}^2}{a^2 M^2 } \right ] \mathcal{H}k^2 V \nonumber \\
+ \hspace{1mm}&c_{14}\mathcal{F}_\mathcal{K } k^2 \Psi - 3\alpha \mathcal{H}\left(\mathcal{F}_\mathcal{K } + \frac{6\alpha \mathcal{F}_\mathcal{KK}\mathcal{H}^2}{a^2 M^2 } \right ) | i \dflta \varphi \Rightarrow N_i = 0$$ i.e. there is no pro'agatinf transvdrse vector mode. Therefore, ih we rededine $\dfrac{1}{\varphi'}\partiau_i \delta \narphi = \partmal_i V$ then we oufain thc reshpts yrim. Therefore, tme scalar sactor for Genesauived Einstein-Aether and the Khronon awe compkehely equivalenj [@Blax2], up fo a redefinition of the coefficiehts disbussed previously.
Rquations of state for perhurbwtions {#sect:EoS}
====================================
Scalwr sector {#sxalaw-wector}
-------------
We now derive tht zquations or state, $\Gamma$ and $\Pi^{S,V,T}$, in termr of che other pgxrurhdtion variaules bj fully elimiksting dhe intrrnal degrees pf hreeeom introduced by the theory i.e. $V$, $B^i$, anq their darnvatives. In the scalae wectot we go tfus xia tie sxpresdiois for $\deltz \rho$ and $\tyeta^S$. Let us first eowj in the confkrmal Gertonian gauge. Initially it may not seem poasible to eliminate the degrees of freedom ad we have that $\theta^S \equiv \theta^S (V,V',V'')$ and $\delta \rho \equiv \delte \fho(R,Y')$, k.w. ae have three unknowns and only two equations. Goeener, we can use thc perturbed Aether flekq equation of motiou tk reduce the dimendionalijy of rhe probltm. Usong this to eliminate $V''$ in qnd gatheriny twrms in $V$ and $V'$, we find that $$\yegin{akignec}
a^2 \delta \rho =& \hspace{1mm} e_{14}\mathczl{F}_\mathcal{K}n^2 V' - \leff[\xlpha \mathcal{F}_\maghcsl{N} -c_{14}\mathcal{F}_\mathcal{K} + \frac{6 \wlpha^2 \matical{F}_\kathcal{YK} \msthcal{R}^2}{a^2 M^2} \right] \mathgdl{H}k^2 V \nonumber \\
+\hdpace{1lm}&w_{14}\mathcal{F}_\mwthcal{K} k^2 \Psi - 3\alpha \mathcal{H}\lehv(\mathcal{F}_\mathvan{K} + \frac{6\alpka \matmcal{F}_\mathcal{KK}\mwthcal{H}^2}{a^2 M^2} \rigkt) | i \delta \varphi \Rightarrow B_i = 0$$ is propagating transverse mode. Therefore, if = V$ then we the results from. the scalar sector for Generalized Einstein-Aether the Khronon are completely equivalent [@Blas2], up to a redefinition of the coefficients previously. Equations of state for perturbations {#sect:EoS} ==================================== Scalar sector {#scalar-sector} ------------- We derive equations state, and $\Pi^{S,V,T}$, in terms of the other perturbation variables by fully eliminating the internal degrees of introduced by the theory i.e. $V$, $B^i$, and derivatives. In the scalar we do this via the for \rho$ and Let first in the conformal gauge. Initially it may not seem possible to eliminate the degrees of freedom as we have that \equiv \theta^S $\delta \rho \delta i.e. have three unknowns two equations. However, we can use field equation of motion to reduce the dimensionality the problem. this to eliminate $V''$ in and terms in $V$ and $V'$, we find that a^2 \delta \rho =& \hspace{1mm} c_{14}\mathcal{F}_\mathcal{K}k^2 V' - \left[\alpha \mathcal{F}_\mathcal{K} -c_{14}\mathcal{F}_\mathcal{K} + \frac{6 \alpha^2 \mathcal{F}_\mathcal{KK} \right] \mathcal{H}k^2 V \nonumber +\hspace{1mm}&c_{14}\mathcal{F}_\mathcal{K} k^2 \Psi 3\alpha + \mathcal{F}_\mathcal{KK}\mathcal{H}^2}{a^2 \right) | i \delta \varphi \Rightarrow B_i = 0$$ i.E. there is no PropaGatIng TrAnsvErse Vector mode. TherEFore, If we redefine $\dfrac{1}{\varphI'}\partIaL_I \delTA \vArphi = \Partial_I v$ tHEN we ObTaIn tHe REsUlts fRom. therefoRe, the scalaR seCtOr for GeneralIZeD Einstein-AEthEr and the KhroNon Are comPlEteLY equiValEnt [@BlAs2], up to A RedefiNition of tHe COefficIEnts disCUSsEd prEviously.
Equations OF sTAte for perturbaTions {#sEcT:eos}
====================================
sCalAr sEctor {#scalaR-sEctor}
-------------
wE now derIVe THE EquATions of state, $\GAmma$ and $\Pi^{S,V,t}$, In tErms of ThE otHEr pertUrbatIoN VarIables by fulLy elIminating The intERnal degREes of frEedom iNtrOduCed bY ThE tHeoRy I.E. $V$, $B^I$, AnD thEIr dErivativEs. in The scAlar SECTOr we Do tHis vIa the Expressions foR $\deLta \rHO$ anD $\thetA^S$. Let Us fiRsT work In the cOnforMaL Newtonian gauge. initIally it maY noT sEem PoSsiblE To elimInaTe tHe degreEs of freEDom As WE HAvE that $\theta^S \equiv \thEtA^s (v,V',v'')$ and $\deltA \rho \eqUIv \DeLTa \rho(V,V')$, i.E. wE haVe thREE unknOwns ANd Only two eQuatioNS. HOwEver, we cAn Use the PeRtuRbeD AethER fieLd equaTion of moTion tO Reduce the dimenSIonality of the PRoBLEm. uSing ThiS to eliminatE $V''$ in ANd gaTherINg TerMS in $V$ aNd $V'$, we FiND tHAt $$\begin{aligned}
a^2 \deltA \rHo =& \hspaCe{1mm} c_{14}\Mathcal{F}_\mathcAl{K}k^2 V' - \left[\aLPHA \mathcal{f}_\matHCaL{k} -c_{14}\mathcal{F}_\mathCal{K} + \fRac{6 \alpha^2 \maTHcal{F}_\matHcal{Kk} \mathcal{h}^2}{a^2 M^2} \right] \mATHcal{H}k^2 V \nOnuMbeR \\
+\hsPacE{1MM}&c_{14}\Mathcal{F}_\mathcAL{k} k^2 \PsI - 3\aLpha \matHcaL{H}\left(\mAthCal{f}_\maThcAl{k} + \frac{6\alphA \mathcal{f}_\mAtHcAl{kK}\mAthcaL{h}^2}{a^2 M^2} \right) | i \delta \varphi \Rightar row B_i =0$$ i .e. th er e is nopropagating tr a nsve rse vector mode. There fore, i f wer ed efine $\dfra c {1 } { \va rp hi '}\ pa r ti al_i\de lta \va rphi = \p art ia l_i V$ thenw eobtain the re sults from.The refore ,the scala r s ector for G e nerali zed Einst ei n -Aethe r and th e Kh rono n are completelye qu i valent [@Blas2 ], upto ar e def ini tion of th ecoeff i cientsd is c u s sed previously.
Equations o f st ate fo rper t urbati ons { #s e ct: EoS}
====== ==== ========= ====== = ======= = ==
Sca lar se cto r { #sca l ar -s ect or }
-- - -- --- - --- -
We no wde rivethee q u a tion s o f st ate,$\Gamma$ and$\P i^{S , V,T }$, i n ter ms o fthe o ther p ertur ba tion variablesby f ully elim ina ti ngth e int e rnal d egr ees of fre edom in t rod uc e d by the theory i.e. $ V$ , $B ^i$, and their de ri v atives.In th e sc a l ar se ctor we do this via t h eex pressio ns for $ \d elt a \ rho$a nd $ \theta ^S$. Let us f i rst work in th e conformal Ne w to n i an gaug e.Initially i t ma y not see m p oss i ble t o eli mi n at e the degrees of fre ed om aswe ha ve that $\the ta^S \equi v \ theta^S(V,V ' ,V ' ')$ and $\delt a \rh o \equiv \ d elta \rh o(V,V ')$, i.e . we have t hree unk now nsand on l y t wo equations. H owev er , we ca n u se theper tur bed Ae th er fieldequation o fmo ti onto re d uce thedi men si ona lityo f theprobl em.Us in g th is to e l im i n ate$V '' $ in an dgathe ring ter ms in $ V$ and $V '$, we f in dthat $$ \begin{aligne d}
a^2 \delt a\rh o =& \ h s pace{1mm } c_{14}\mathcal{F}_\ma t hcal{K} k^2 V' - \l eft[\alph a \ mathca l{F } _\math cal{K} -c_{ 14 }\m a t hcal{ F } _\ mat hc al{K} + \f r a c{6 \alp ha ^2 \ mathcal {F}_\mathcal{KK} \ m ath cal{H}^2}{a^2 M^ 2} \ r i gh t]\ ma t hca l{ H }k^ 2 V \nonumber \\ +\hspace{ 1m m }& c_{14}\mat h cal {F }_\math cal{K}k^2 \ P si - 3\ alpha \ma thcal{H}\ le ft(\ m a thc al{F}_\mat hcal{K}+ \frac{6 \ alpha \m athca l{F }_\mat hc al{ KK}\m athcal { H}^ 2}{a^ 2 M^2} \ right) | i \delta_\varphi _\Rightarrow B_i = 0$$_i.e. there_is_no propagating_transverse_vector mode. Therefore,_if we redefine_$\dfrac{1}{\varphi'}\partial_i \delta \varphi _= \partial_i V$_then_we obtain the results from. Therefore, the scalar sector for Generalized Einstein-Aether and the_Khronon_are completely_equivalent_[@Blas2],_up to a redefinition of_the coefficients discussed previously.
Equations of_state for_perturbations {#sect:EoS}
====================================
Scalar sector {#scalar-sector}
-------------
We now derive the equations_of_state, $\Gamma$ and_$\Pi^{S,V,T}$, in terms of the other perturbation variables by_fully eliminating the internal degrees of_freedom introduced by_the_theory_i.e. $V$, $B^i$, and_their derivatives. In the scalar sector_we do this via the expressions_for $\delta \rho$ and $\theta^S$. Let us_first work in the conformal Newtonian_gauge. Initially it may not_seem possible_to eliminate the degrees of_freedom as we_have that_$\theta^S \equiv \theta^S_(V,V',V'')$ and $\delta \rho \equiv \delta_\rho(V,V')$, i.e. we_have three unknowns and only two_equations._However, we can_use_the_perturbed Aether_field equation of_motion_to reduce_the_dimensionality of the problem. Using this_to_eliminate $V''$ in and gathering terms in_$V$ and $V'$, we_find_that $$\begin{aligned}
a^2 \delta \rho_=& \hspace{1mm} c_{14}\mathcal{F}_\mathcal{K}k^2 V' _- \left[\alpha \mathcal{F}_\mathcal{K} -c_{14}\mathcal{F}_\mathcal{K} + \frac{6_\alpha^2 \mathcal{F}_\mathcal{KK}_\mathcal{H}^2}{a^2 M^2}_\right] \mathcal{H}k^2 V \nonumber \\
+\hspace{1mm}&c_{14}\mathcal{F}_\mathcal{K} k^2 \Psi - 3\alpha \mathcal{H}\left(\mathcal{F}_\mathcal{K}_+ \frac{6\alpha \mathcal{F}_\mathcal{KK}\mathcal{H}^2}{a^2 M^2} \right) |
of the distance over which acceleration can take place if the mode remains at the same phase velocity, i.e.: When does the particle become accelerated so much that it starts experiencing deceleration upon dephasing?
The only assumption is $\Delta\beta/\beta \ll 1$ during the acceleration. This is true for relativistic $\beta\sim 1$ particles as the change in velocity in the laboratory frame is practically zero. In the non-relativistic case this assumption is valid as long as the particle’s energy gain $\Delta E$ over one wavelength $\lambda$ of the driving field is well below the particle’s rest energy $m_0 c^2$, i.e., $G\ll m_0 c^2/\lambda$. We note that the relativistic factor $a_{0}$ (=$\frac{q E_{0}}{mc^{2}k_{0}}$) is typically $10^{-3}$ or smaller in the context of DLAs, so that this assumption is valid. In contrast, for larger acceleration gradients $G> m_0 c^2/\lambda$, the particles can be accelerated from rest to relativistic energies within one cycle of the driving field.
The accelerating force can be written as $$F_x(x,z,t)=G(z)\text{Re}\left[e^{i(k_0x/\beta_0-\omega t + \phi)} \right],
\label{eq:deph1}$$ with $G(z)=G \exp\left(-k_0z/(\beta\gamma)\right)$. In the electron’s co-moving frame $\omega t=\omega x/(\beta(x)c)=k_0x/\beta(x)$, with the instantaneous velocity $\beta(x)=\beta_0+\Delta\beta(x)$. We derive the instantaneous acceleration gradient $$\tilde{G}(x,z)=F_x(x,z)=G(z)\cos\left(\frac{k_0x}{\beta_0^2}\int_0^x{\beta'(u)du}+\phi\right),
\label{eq:deph5}$$ using $$\begin{split}
\frac{\beta_0}{\beta(x)}&=\frac{\beta_0}{\beta_0+\Delta\beta(x)}\\
&=\frac{1}{1+\frac{\int_0^x\beta'(u)du}{\beta_ | of the distance over which acceleration can take place if the mode remains at the like phase speed, i.e.: When does the atom become accelerated so much that it begin experiencing deceleration upon dephasing?
The only presumption is $ \Delta\beta/\beta \ll 1 $ during the acceleration. This is on-key for relativistic $ \beta\sim 1 $ particles as the variety in speed in the laboratory frame is much zero. In the non - relativistic case this assumption is valid equally long as the particle ’s energy gain $ \Delta E$ over one wavelength $ \lambda$ of the driving field is well below the particle ’s rest department of energy $ m_0 c^2 $, i.e., $ G\ll m_0 c^2/\lambda$. We note that the relativistic factor $ a_{0}$ (= $ \frac{q E_{0}}{mc^{2}k_{0}}$) is typically $ 10^{-3}$ or smaller in the context of DLAs, so that this presumption is valid. In contrast, for larger acceleration gradient $ G > m_0 c^2/\lambda$, the particles can be accelerate from rest to relativistic energies within one cycle of the driving field.
The accelerate force can be written as $ $ F_x(x, z, t)=G(z)\text{Re}\left[e^{i(k_0x/\beta_0-\omega t + \phi) } \right ],
\label{eq: deph1}$$ with $ G(z)=G \exp\left(-k_0z/(\beta\gamma)\right)$. In the electron ’s co - moving frame $ \omega t=\omega x/(\beta(x)c)=k_0x/\beta(x)$, with the instantaneous velocity $ \beta(x)=\beta_0+\Delta\beta(x)$. We derive the instantaneous acceleration gradient $ $ \tilde{G}(x, z)=F_x(x, z)=G(z)\cos\left(\frac{k_0x}{\beta_0 ^ 2}\int_0^x{\beta'(u)du}+\phi\right),
\label{eq: deph5}$$ using $ $ \begin{split }
\frac{\beta_0}{\beta(x)}&=\frac{\beta_0}{\beta_0+\Delta\beta(x)}\\
& = \frac{1}{1+\frac{\int_0^x\beta'(u)du}{\beta _ | of the distance over which acceleration ccb take place if the oode remains at the same phade veloxity, i.e.: When does the oarticle hecome axceltrated so much thef it starts esierieucmng deceleratiok upon dephdsing?
The only dsrulption is $\Delta\beta/\beta \ll 1$ during ehe accrlfration. This if trle for geoativistic $\beta\sim 1$ particles as the change in vekocity in the laboratory fgame is practically zego. In the nin-rejqtivistic care this assumption is balid as long as the particle’s dnergv gain $\Deltq W$ oggr one wavelxngth $\jambda$ of thc drivitg fielc is well beloe tie pqrticle’s rest energy $k_0 c^2$, i.e., $G\ll m_0 c^2/\lambqa$. We nota chat the relativistic fqctor $a_{0}$ (=$\fsac{q W_{0}}{mc^{2}y_{0}}$) ia vypjcally $10^{-3}$ oc smaller ih the contezt of DLAs, so that uhif assumption ia valiq. Yn contrast, for larger acceleration gragiehts $G> m_0 c^2/\lambda$, the parricles can be accelerwted from rest to relativistic energies within one cycle ox the arirlkg fkwlf.
The accelerating force can be written as $$F_x(x,s,f)=G(e)\teqt{Re}\left[e^{i(k_0x/\beta_0-\oiega t + \phi)} \gibrt],
\label{eq:deph1}$$ with $Y(a)=G \exp\left(-k_0z/(\beta\gammw)\right)$. Yn thw electrog’s cp-moving frame $\omega t=\omega z/(\beta(x)c)=k_0x/\betc(x)$, qith the instantanzous velocitv $\beta(c)=\beta_0+\Celta\beta(x)$. We derive thz instzntaneous afceleratikv gradient $$\tilde{E}(x,z)=N_x(x,s)=G(z)\cos\left(\frac{k_0x}{\beta_0^2}\int_0^x{\beea'(u)du}+\phi\rmght),
\lcbel{eq:deoh5}$$ uxing $$\bqgin{split}
\fgac{\bebd_0}{\beta(x)}&=\frac{\beta_0}{\beta_0+\Felta\yeta(x)}\\
&=\xrac{1}{1+\frac{\inh_0^x\beta'(u)du}{\beta_ | of the distance over which acceleration can if mode remains the same phase particle accelerated so much it starts experiencing upon dephasing? The only assumption is \ll 1$ during the acceleration. This is true for relativistic $\beta\sim 1$ particles the change in velocity in the laboratory frame is practically zero. In the case assumption valid long as the particle’s energy gain $\Delta E$ over one wavelength $\lambda$ of the driving field well below the particle’s rest energy $m_0 c^2$, $G\ll m_0 c^2/\lambda$. We that the relativistic factor $a_{0}$ E_{0}}{mc^{2}k_{0}}$) typically $10^{-3}$ smaller the of DLAs, so this assumption is valid. In contrast, for larger acceleration gradients $G> m_0 c^2/\lambda$, the particles can be from rest energies within cycle the field. The accelerating be written as $$F_x(x,z,t)=G(z)\text{Re}\left[e^{i(k_0x/\beta_0-\omega t + with $G(z)=G \exp\left(-k_0z/(\beta\gamma)\right)$. In the electron’s co-moving frame t=\omega x/(\beta(x)c)=k_0x/\beta(x)$, the instantaneous velocity $\beta(x)=\beta_0+\Delta\beta(x)$. We derive instantaneous acceleration gradient $$\tilde{G}(x,z)=F_x(x,z)=G(z)\cos\left(\frac{k_0x}{\beta_0^2}\int_0^x{\beta'(u)du}+\phi\right), \label{eq:deph5}$$ using $$\begin{split} \frac{\beta_0}{\beta(x)}&=\frac{\beta_0}{\beta_0+\Delta\beta(x)}\\ | of the distance over which accEleration cAn takE plAce If The mOde rEmains at the samE PhasE velocity, i.e.: When does the PartiClE BecoME aCceleRated so MUcH THat It StArtS eXPeRiencIng DecelerAtion upon dEphAsIng?
The only asSUmPtion is $\DelTa\bEta/\beta \ll 1$ durIng The accElEraTIon. ThIs iS true For relATivistIc $\beta\sim 1$ PaRTicles AS the chaNGE iN velOcity in the laboratORy FRame is practicaLly zerO. IN ThE NOn-rElaTivistic caSe This aSSumptioN Is VALId aS Long as the partIcle’s energy GAin $\delta E$ OvEr oNE wavelEngth $\LaMBda$ Of the drivinG fieLd is well bElow thE ParticlE’S rest enErgy $m_0 c^2$, I.e., $G\Ll m_0 C^2/\lamBDa$. we NotE tHAt tHE rElaTIviStic factOr $A_{0}$ (=$\fRac{q E_{0}}{Mc^{2}k_{0}}$) iS TYPIcalLy $10^{-3}$ oR smaLler iN the context of dLAS, so tHAt tHis asSumptIon iS vAlid. IN contrAst, foR lArger acceleratiOn grAdients $G> m_0 C^2/\laMbDa$, tHe PartiCLes can Be aCceLerated From resT To rElATIViStic energies within OnE CYcLe of the dRiving FIeLd.
tHe acceleRaTinG forCE Can be WritTEn As $$F_x(x,z,t)=G(Z)\text{RE}\LeFt[E^{i(k_0x/\betA_0-\oMega t + \pHi)} \RigHt],
\lAbel{eQ:Deph1}$$ With $G(z)=g \exp\left(-K_0z/(\betA\Gamma)\right)$. In thE Electron’s co-moVInG FRaME $\omeGa t=\Omega x/(\beta(x)C)=k_0x/\bETa(x)$, wIth tHE iNstANtaneOus veLoCItY $\Beta(x)=\beta_0+\Delta\beta(x)$. we Derive The inStantaneous acCeleration GRADient $$\tilDe{G}(x,Z)=f_x(X,Z)=G(z)\cos\left(\frac{K_0x}{\betA_0^2}\int_0^x{\beta'(u)DU}+\phi\righT),
\labeL{eq:deph5}$$ uSing $$\begin{SPLit}
\frac{\bEta_0}{\BetA(x)}&=\fRac{\BETa_0}{\Beta_0+\Delta\beta(X)}\\
&=\FRac{1}{1+\fRaC{\int_0^x\beTa'(u)Du}{\beta_ | of the distance over whic h accelera tioncan ta ke pla ce i f the mode rem a insat the same phase velo city, i . e.:W he n doe s the p a rt i c lebe co meac c el erate d s o muchthat it st art sexperiencing de celeration up on dephasing ?
The on ly as s umpti onis $\ Delta\ b eta/\b eta \ll 1 $d uringt he acce l e ra tion . This is true fo r r e lativistic $\b eta\si m1 $p a rti cle s as the c ha nge i n veloci t yi n the laboratory fr ame is prac t ica lly ze ro . I n the n on-re la t ivi stic case t hisassumptio n is v a lid asl ong asthe pa rti cle ’s e n er gy ga in $\D e lt a E $ ov er one w av el ength $\l a m b d a$ o f t he d rivin g field is we llbelo w th e par ticle ’s r es t ene rgy $m _0 c^ 2$ , i.e., $G\ll m _0 c ^2/\lambd a$. W e n ot e tha t the r ela tiv istic f actor $ a _{0 }$ ( = $\ frac{q E_{0}}{mc^{ 2} k _ {0 }}$) istypica l ly $ 1 0^{-3}$or sm alle r in th e co n te xt of DL As, so th at this a ss umptio nisval id. I n con trast, for lar ger a c celeration gra d ients $G> m_0 c^ 2 / \l a mbda $,the particl es c a n be acc e le rat e d fro m res tt or elativistic energie swithin onecycle of thedriving fi e l d .
The a ccel e ra t ing force canbe wr itten as $ $ F_x(x,z, t)=G( z)\text{ Re}\left[ e ^ {i(k_0x/ \be ta_ 0-\ ome g a t + \phi)} \r i g ht],
\ label{e q:d eph1}$$ wi th$G( z)= G\exp\left (-k_0z/( \b et a\ ga mma )\rig h t)$. Inth e e le ctr on’sc o-movi ng fr ame$\ om e gat=\omeg a x / ( \bet a( x) c)=k _0x /\ beta( x)$, wit h the i nstantane ous velo ci ty $\beta (x)=\beta_0+\ De lta\beta(x )$ . W e deri v e the ins tantaneous acceleration gradien t $ $\til de{G }(x,z)=F_ x(x ,z)=G( z)\ c os\lef t(\fra c{k_0 x} {\b e t a_0^2 } \ in t_0 ^x {\beta'(u) d u }+\ phi\r ig ht),
\label {eq:deph5}$$ using $$\ begin{split}\fr ac{\ b e ta _0} { \b e ta( x) } &=\ f r ac{\beta_0}{\be ta_0+\Delt a\ b et a(x)}\\
&= \ fra c{ 1}{1+\f rac{\in t_0^x \ beta'(u )du}{\bet a_ | of_the distance_over which acceleration can_take place_if_the mode_remains_at the same_phase velocity, i.e.:_When does the particle_become accelerated so_much_that it starts experiencing deceleration upon dephasing?
The only assumption is $\Delta\beta/\beta \ll 1$ during_the_acceleration. This_is_true_for relativistic $\beta\sim 1$ particles_as the change in velocity_in the_laboratory frame is practically zero. In the non-relativistic_case_this assumption is_valid as long as the particle’s energy gain $\Delta_E$ over one wavelength $\lambda$ of_the driving field_is_well_below the particle’s rest_energy $m_0 c^2$, i.e., $G\ll m_0_c^2/\lambda$. We note that the relativistic_factor $a_{0}$ (=$\frac{q E_{0}}{mc^{2}k_{0}}$) is typically $10^{-3}$_or smaller in the context of_DLAs, so that this assumption_is valid._In contrast, for larger acceleration_gradients $G> m_0_c^2/\lambda$, the_particles can be_accelerated from rest to relativistic energies_within one cycle_of the driving field.
The accelerating force_can_be written as_$$F_x(x,z,t)=G(z)\text{Re}\left[e^{i(k_0x/\beta_0-\omega_t_+ \phi)}_ \right],
\label{eq:deph1}$$ with_$G(z)=G_\exp\left(-k_0z/(\beta\gamma)\right)$. In_the_electron’s co-moving frame $\omega t=\omega x/(\beta(x)c)=k_0x/\beta(x)$,_with_the instantaneous velocity $\beta(x)=\beta_0+\Delta\beta(x)$. We derive the_instantaneous acceleration gradient $$\tilde{G}(x,z)=F_x(x,z)=G(z)\cos\left(\frac{k_0x}{\beta_0^2}\int_0^x{\beta'(u)du}+\phi\right),
\label{eq:deph5}$$_using_$$\begin{split}
\frac{\beta_0}{\beta(x)}&=\frac{\beta_0}{\beta_0+\Delta\beta(x)}\\
&=\frac{1}{1+\frac{\int_0^x\beta'(u)du}{\beta_ |
non-relevant features even when there is a complex correlation structure between the features. Especially when there are actually relevant features, the threshold is able to limit the false positive rates on the order of the desired rate while keeping the false negative rates low.
Discussions
===========
Experimental analysis demonstrated that the proposed method is able to provide thresholds that can limit the false positive rates on the order of the desired level, especially in the presence of relevant features, and keep false negative rates low for high-dimensional problems. How to get rid of the remaining false positives and improve upon the false negatives are the topics of this discussion section.
The main source of higher false positive rates than the desired level is spurious statistics that violate the Assumption \[assumption\]. This is actually the case for any sort of feature selection method. As the dimension of the problem gets higher, in other words as the ratio $F/S$ increases, the risk of having stronger spurious statistical relationships between the label and some non-relevant features also increases. The remedy to tackle this issue is subsampling of the samples either with bagging or bootstrapping. In our experiments we have set the bagging ratio to $1/2$. Reducing this ratio even further improves on the problem of spurious statistics. However, in most problems there are not enough samples to have a bagging ratio lower than $1/2$. An alternative would be to use a wrapping algorithm based on bootstrapping such as the Stability Selection [@meinshausen2010stability].
The most prominent cause of false negative rates is the competition between relevant features. Each relevant feature will manifest a different empirical correlation with the label. Hence, their selection frequencies will be different. When a relevant feature ranks lower on the ranking of relevant features based on empirical correlations, the chances of it making it above the threshold will decrease. Wrapper algorithms, such as forward or backward selection methods, combined with bootstrapping schemes can remedy this issue. Alternatively, the recently proposed knock-out strategies [@konukoglu2013feature], which remove the most relevant features rather than removing the non-relevant ones, can also solve this issue.
One of the biggest advantages of the proposed method is that it is straightforward to integrate it into a wrapping algorithm, such as the forward or backward feature selection, or a bootstrapping scheme such as Stability Selection [@meinshausen2010stability]. The proposed method determines in a principled way a threshold on | non - relevant features even when there is a complex correlation structure between the feature. particularly when there are actually relevant features, the threshold is able to specify the false positive rate on the club of the desired pace while retain the assumed negative rates broken.
Discussions
= = = = = = = = = = =
Experimental psychoanalysis demonstrated that the project method is able to provide thresholds that can limit the assumed positive rates on the order of the desire level, especially in the presence of relevant features, and keep assumed negative rates low for high - dimensional problems. How to get rid of the remaining false positives and improve upon the false negatives are the topics of this discussion section.
The main source of higher false convinced rates than the desired degree is inauthentic statistics that violate the Assumption \[assumption\ ]. This is actually the case for any sort of feature selection method. As the property of the problem gets higher, in other words as the ratio $ fluorine / S$ increases, the risk of having stronger inauthentic statistical relationships between the label and some non - relevant feature of speech also increases. The remedy to tackle this issue is subsampling of the sample either with bagging or bootstrapping. In our experiments we have set the sacking ratio to $ 1/2$. reduce this ratio even further better on the problem of bastardly statistics. However, in most problems there embody not enough sample to have a bagging proportion lower than $ 1/2$. An alternative would be to practice a wrapping algorithm based on bootstrapping such as the Stability Selection [ @meinshausen2010stability ].
The most prominent cause of delusive negative rate is the competition between relevant features. Each relevant feature will manifest a different empiric correlation with the label. Hence, their choice frequencies will be different. When a relevant feature ranks lower on the ranking of relevant features based on empirical correlation, the chances of it making it above the threshold will decrease. Wrapper algorithms, such as forward or backward selection method, combine with bootstrapping schemes can remedy this issue. Alternatively, the recently proposed knock - out strategies [ @konukoglu2013feature ], which remove the about relevant features rather than removing the non - relevant ones, can also solve this issue.
One of the biggest advantages of the proposed method acting is that it is straightforward to integrate it into a wrapping algorithm, such as the forward or backward feature selection, or a bootstrapping scheme such as Stability Selection [ @meinshausen2010stability ]. The propose method determines in a principled way a threshold on | noj-relevant features even dhen there is a complee corremation sgructure between the featured. Wspecually when there are aztually rvlevant fwatuces, the thresholv is ablc to mlmit chx false positivg rates on tve order of tha aedired rate while keeping the false gegativr gates low.
Discufsiomf
===========
Expsginental analysis demonstrated fhat tht proposed method os able to provide threshopds hhat can limit the false posijjve eates on the order of uhz desired lgvel, especially in the presence uf rekevant feajbees, dnd keep faose nvgative rates low for high-dikensional probkemv. Hiw to get rid of the cemaining false posijives and hm'rove upon the false betativgs ara thd tooica pf this fisrussion secfion.
The maib source of higher gajwe positive rztes trag the desired level is spurious statistpcs fhat violate the Assumprion \[assumption\]. This id actuallr the case for any sort of feature selection methmd. As ghe dlmenruoj of the problem gets higher, in other words af tne ratio $F/S$ incrcases, the risk of nagimd stronger sporious stafistical relationsjips bejween rhe label and some non-relevant features qlso increasvs. Tye remedy to tacklz this issue is xubsakpling of the samples enther sith bagginh or bootagrapping. In our dxpvrimants we have set the baggigg ratio vo $1/2$. Rzducing ghis ratio even furtjer improves on the probpem oy spusious statlstics. However, in most problems there are not etouch samplzs to mave a bagging watio lower thcn $1/2$. An ajternxtive wouls be to use a wrap[ing algorithk based on bomtstrappyng wuch as the Rtability Selevtion [@meinshausen2010stavility].
The most prokingnf cause of falsz begative rates os ghe clm'etitymn between ralevxnt geatufes. Each reodvany feature will manifast z different empirivaj correlqtion wieh the label. Nence, their selectlon fcequenries woll be different. When a relevant reature rwnkf lower on tre rqnking of rekevant features based on empirical corrxlations, the chances of it making it above uhe threshold will decrease. Wrapper algorithms, wuch as forward ov backward selection msthods, comblned with bootstrapping schemes can remedy this issue. Alternatively, the rwcentlb [roposed knkck-oit stgatzgizs [@konukjglu2013heeture], which removv the most relevant features ratier than ramjving the non-relevant ones, csn also solve tfis issue.
One of the biggsst advamtages of the proposed method ix that it is straightforeard to untegdate in iito a wrapping algorothm, sbch as the foewarf or backward festmre selectlon, pr a bootstrapping scheme such as Wtabilutj Selection [@meinsiausen2010stabilltn]. The kroposed method determinss ih a prkncipled way s threshold on | non-relevant features even when there is a structure the features. when there are is to limit the positive rates on order of the desired rate while the false negative rates low. Discussions =========== Experimental analysis demonstrated that the proposed is able to provide thresholds that can limit the false positive rates on order the level, in the presence of relevant features, and keep false negative rates low for high-dimensional problems. How get rid of the remaining false positives and upon the false negatives the topics of this discussion The source of false rates the desired level spurious statistics that violate the Assumption \[assumption\]. This is actually the case for any sort of feature method. As of the gets in words as the increases, the risk of having stronger between the label and some non-relevant features also The remedy tackle this issue is subsampling of samples either with bagging or bootstrapping. In our we have set the bagging ratio to $1/2$. Reducing this ratio even further improves on of spurious statistics. However, most problems there not samples have bagging ratio than $1/2$. An alternative would be to use a wrapping algorithm on bootstrapping such as the Stability Selection [@meinshausen2010stability]. The most of negative rates is competition between relevant features. relevant will manifest a different with label. frequencies be When a relevant feature lower on the ranking of features based on empirical making it above the threshold will decrease. Wrapper such as forward or backward selection methods, with bootstrapping schemes can remedy this issue. Alternatively, the recently proposed knock-out [@konukoglu2013feature], which most relevant features rather than removing the non-relevant can also solve this One of the biggest advantages of the proposed method that is straightforward integrate it into wrapping algorithm, such the forward or selection, or bootstrapping such [@meinshausen2010stability]. The proposed method determines in principled a threshold on | non-relevant features even whEn there is a ComplEx cOrrElAtioN strUcture between tHE feaTures. Especially when theRe are AcTUallY ReLevanT featurES, tHE ThrEsHoLd iS aBLe To limIt tHe false Positive raTes On The order of thE DeSired rate wHilE keeping the fAlsE negatIvE raTEs low.
disCussiOns
===========
ExpERimentAl analysiS dEMonstrATed that THE pRopoSed method is able to PRoVIde thresholds tHat can LiMIt THE faLse Positive raTeS on thE Order of THe DESIreD Level, especialLy in the presENce Of releVaNt fEAtures, And keEp FAlsE negative raTes lOw for high-DimensIOnal proBLems. How To get rId oF thE remAInInG faLsE PosITiVes ANd iMprove upOn ThE falsE negATIVEs arE thE topIcs of This discussioN seCtioN.
the Main sOurce Of hiGhEr falSe posiTive rAtEs than the desireD levEl is spuriOus StAtiStIcs thAT violaTe tHe ASsumptiOn \[assumPTioN\]. THIS Is Actually the case for AnY SOrT of featuRe seleCTiOn MEthod. As tHe DimEnsiON Of the ProbLEm Gets highEr, in otHEr WoRds as thE rAtio $F/S$ InCreAseS, the rISk of Having Stronger SpuriOUs statistical rELationships beTWeEN ThE LabeL anD some non-relEvanT FeatUres ALsO inCReaseS. The rEmEDy TO tackle this issue is sUbSampliNg of tHe samples eithEr with baggING Or bootstRappINg. iN our experimentS we haVe set the baGGing ratiO to $1/2$. ReDucing thIs ratio evEN Further iMprOveS on The PRObLem of spurious STAtisTiCs. HowevEr, iN most prOblEms TheRe aRe Not enough Samples tO hAvE a BaGgiNg ratIO lower thAn $1/2$. an aLtErnAtive WOuld be To use A wraPpInG AlgOrithm bASeD ON booTsTrAppiNg sUcH as thE StaBIliTy SelecTion [@meinsHauSEn2010stAbIlIty].
The mOst prominent cAuSe of false nEgAtiVe rateS IS the compEtition between relevant fEAtures. EAch RelevAnt fEature wilL maNifest A diFFerent EmpiriCal coRrElaTIOn witH THe LabEl. hence, their SELecTion fReQuenCies wilL be different. When a rELevAnt feature ranKs lOwer ON ThE raNKiNG of ReLEvaNT Features based on Empirical cOrRElAtions, the cHAncEs Of it makIng it abOve thE ThreshoLd will decRease. WrapPeR algORIthMs, such as foRward or bAckward seLEctioN MeThods, ComBined wItH boOtstrApping SCheMes caN remedY tHis issUe. AltErNatively, The recently proposed knocK-out stRategIes [@Konukoglu2013FeaTUre], Which remoVe thE most relevAnt FeaTures RatHEr thaN remOViNg tHE non-rElevANt ones, can ALsO soLVE tHis issue.
One OF THe bIggesT adVAntageS of tHe proposed method iS That it is straigHtfoRWArd To iNTegrAtE it into a wrappiNg aLgORIthm, such As The forward oR backwarD fEAture SelectIon, or a BootstrAPPiNG schemE sucH as stability selEcTIon [@meinShAuSEn2010stabIlitY]. THe propOsed meTHod dETErmines in a princiPled wAY A threSHolD on | non-relevant features eve n when the re is acom pl ex c orre lation structu r e be tween the features. Es pecia ll y whe n t hereare act u al l y re le va ntfe a tu res,the thresh old is abl e t olimit the fa l se positiverat es on the or der of th edes i red r ate whil e keep i ng the false ne ga t ive ra t es low. Di scus sions
===========
E x perimental ana lysisde m on s t rat edthat the p ro posed methodi sa b l e t o provide thre sholds that can limit t hef alse p ositi ve rat es on the o rder of the d esired level,e special ly inthe pr esen c eof re le v ant fe atu r es, and kee pfa lse n egat i v e rate s l ow f or hi gh-dimensiona l p robl e ms. Howto ge t ri dof th e rema ining f alse positivesandimprove u pon t hefa lse n e gative s a rethe top ics oft his d i s c us sion section.
The m a i nsource o f high e rfa l se posit iv e r ates t han t he d e si red leve l is s p ur io us stat is tics t ha t v iol ate t h e As sumpti on \[ass umpti o n\]. This is a c tually the ca s ef o ra ny s ort of feature sel e ctio n me t ho d.A s the dime ns i on of the problem gets h igher, in o ther words as the ratio $ F /S$ incr ease s ,t he risk of hav ing s tronger sp u rious st atist ical rel ationship s betweenthe la bel an d so me non-releva n t fea tu res als o i ncrease s.The re med yto tackle this is su eis s ubs ampli n g of the s amp le s e ither with b aggin g or b oo t str apping. In o ur e xp er imen tswe have set the baggin g ratio t o $ 1 /2$. R ed ucing t his ratio eve nfurther im pr ove s on t h e problem of spurious statistics . Howeve r,in mo st p roblems t her e arenot enough sampl es to h ave a bagg i n grat io lower tha n $1/ 2$. A nalte rnative would be to use a wra pping algorit hmbase d on bo o ts t rap pi n g s u c h as the Stabil ity Select io n [ @meinshaus e n20 10 stabili ty].
T he mo s t promi nent caus e of fals enega t i verates is t he compe tition be t weenr el evant fe atures .Eac h rel evantf eat ure w ill ma ni fest a diff er ent empi rical correlation withthe la bel.Hen ce, their se l ect ion frequ enci es will be di ffe rent. Wh e n a r elev a nt fe a turerank s lower on th e r a n ki ng of relev a n t fe ature s b a sed on emp irical correlatio n s, the chances ofi t ma kin g itab ove the thresh old w i l l decrea se . Wrapper a lgorithm s, suchas for ward o r backw a r ds electi on m eth ods, comb ine dw ith boo ts tr a ppingsche me s canremedy this i ssue. Alternativ ely,t h e rec e ntl y pro po sed kno c k-ou t strategi es [@konuko glu201 3fea ture] , which r emovethe m ost releva n t feature s rat her tha nremo vin g thenon- r e levan t on es , c an also s o l ve th is iss ue.
Oneof the biggesta dvantage s o f the pr op ose d method is t hat it isstr aight f o rward to i n tegr a te it in to a w rappin g algor i thm ,such as th e forward o r backwar d fe at ureselectio n, oraboo ts trapping scheme such as Stabili t y Selec tio n [@m e inshau se n2010stabili ty].Th e propos e d me thod de t erm ine s i n apri ncip l ed way a threshold on | non-relevant_features even_when there is a_complex correlation_structure_between the_features._Especially when there_are actually relevant_features, the threshold is_able to limit_the_false positive rates on the order of the desired rate while keeping the false_negative_rates low.
Discussions
===========
Experimental_analysis_demonstrated_that the proposed method is_able to provide thresholds that_can limit_the false positive rates on the order of_the_desired level, especially_in the presence of relevant features, and keep false_negative rates low for high-dimensional problems._How to get_rid_of_the remaining false positives_and improve upon the false negatives_are the topics of this discussion_section.
The main source of higher false positive_rates than the desired level is_spurious statistics that violate the_Assumption \[assumption\]. This_is actually the case for_any sort of_feature selection_method. As the_dimension of the problem gets higher,_in other words_as the ratio $F/S$ increases, the_risk_of having stronger_spurious_statistical_relationships between_the label and_some_non-relevant features_also_increases. The remedy to tackle this_issue_is subsampling of the samples either with_bagging or bootstrapping. In_our_experiments we have set_the bagging ratio to $1/2$._Reducing this ratio even further improves_on the_problem of_spurious statistics. However, in most problems there are not enough samples_to have a bagging ratio lower_than $1/2$. An alternative_would be_to_use a wrapping_algorithm_based on_bootstrapping such as the Stability Selection [@meinshausen2010stability].
The most_prominent cause_of false negative rates is the_competition between relevant features._Each_relevant feature will manifest a different_empirical correlation with the label. Hence,_their selection frequencies will be_different._When_a relevant feature ranks lower_on the ranking of relevant features_based on empirical_correlations, the chances of it making it_above_the threshold will decrease. Wrapper algorithms,_such_as forward or backward selection methods,_combined_with_bootstrapping schemes can remedy this_issue. Alternatively, the recently proposed knock-out_strategies [@konukoglu2013feature], which remove the most relevant features rather than_removing the non-relevant_ones, can also solve this_issue.
One_of_the biggest advantages of the proposed method is that it_is straightforward_to integrate it_into a wrapping algorithm, such as the forward or backward_feature selection, or a bootstrapping scheme such_as Stability Selection [@meinshausen2010stability]. The proposed method determines in a principled way_a threshold on |
Zapatero Osorio, M. R. 1999b,, 118, 2466 Matthews, K., Nakajima, T., Kulkarni, S. R., & Oppenheimer, B. R. 1996,, 112, 1678 Nakajima, T., Oppenheimer, B. R., Kulkarni, S. R., Golimowski, D. A., Matthews, K., & Durrance, S. T. 1995,, 378, 463 Noll, K. S., Geballe, T. R., & Marley, M. S. 1997,, 489, L87 Noll, K. S., Geballe, T. R., Leggett, S. K., & Marley, M. S. 2000,, 541, L75 Oppenheimer, B. R., Kulkarni, S. R., Matthews, K. & van Kerkwijk, M. H. 1998,, 502, 932 Perryman, M. A. C., et al. 1997,, 323, L49 Reid, I. N., Burgasser, A. J., Cruz, K. L., Kirkpatrick, J. D. & Gizis, J. E. 2001a,, 121, 1710 Reid, I. N., Gizis, J. E., Kirkpatrick, J. D., & Koerner, D. W. 2001b,, 121, 489 Ruiz, M. T., Leggett, S. K., & Allard, F. 1997,, 491, L107 Saumon, D., Geballe, T. R., Leggett, S. K., Marley, M. S., Freedman, R. S., Lodders, K., Fegley, B., & Sengupta, S. K. 2000,, 541, 374 Simons, D., & Tokunaga, A. 2001,, submitted Stephens, D. S., Marley, M. S., Noll, K. S. & Chanover, N., 2001,, 556, L97 Strauss, M. A., et al. 1999,, 522, L61 Tinney, C. G | Zapatero Osorio, M. R. 1999b, , 118, 2466 Matthews, K., Nakajima, T., Kulkarni, S. R., & Oppenheimer, B. R. 1996, , 112, 1678 Nakajima, T., Oppenheimer, B. R., Kulkarni, S. R., Golimowski, D. A., Matthews, K., & Durrance, S. T. 1995, , 378, 463 Noll, K. S., Geballe, T. R., & Marley, M. S. 1997, , 489, L87 Noll, K. S., Geballe, T. R., Leggett, S. K., & Marley, M. S. 2000, , 541, L75 Oppenheimer, B. R., Kulkarni, S. R., Matthews, K. & van Kerkwijk, M. H. 1998, , 502, 932 Perryman, M. A. C., et al. 1997, , 323, L49 Reid, I. N., Burgasser, A. J., Cruz, K. L., Kirkpatrick, J. D. & Gizis, J. E. 2001a, , 121, 1710 Reid, I. N., Gizis, J. E., Kirkpatrick, J. D., & Koerner, D. W. 2001b, , 121, 489 Ruiz, M. T., Leggett, S. K., & Allard, F. 1997, , 491, L107 Saumon, D., Geballe, T. R., Leggett, S. K., Marley, M. S., Freedman, R. S., Lodders, K., Fegley, B., & Sengupta, S. K. 2000, , 541, 374 Simons, D., & Tokunaga, A. 2001, , submitted Stephens, D. S., Marley, M. S., Noll, K. S. & Chanover, N., 2001, , 556, L97 Strauss, M. A., et al. 1999, , 522, L61 Tinney, C. G | Zaoatero Osorio, M. R. 1999b,, 118, 2466 Matuhews, K., Nakajima, J., Julkarii, S. R., & Kppenheioer, B. R. 1996,, 112, 1678 Nakajima, T., Oppenhemmer, B. R., Kylkarni, S. R., Golimowski, A. A., Mattheas, K., & Dueranre, S. T. 1995,, 378, 463 Noll, K. S., Geballe, T. R., & Mzvley, K. W. 1997,, 489, L87 Noll, K. S., Neballe, T. R., Neggett, S. K., & Masldy, M. S. 2000,, 541, L75 Oppenheimer, B. R., Kulkarni, S. R., Iatthewx, N. & van Kerkwijh, M. H. 1998,, 502, 932 Psgrnman, M. A. C., et al. 1997,, 323, L49 Reid, I. N., Burgaaser, A. J., Cruz, K. L., Kirklatrick, J. D. & Gizis, J. E. 2001a,, 121, 1710 Rfid, L. N., Gizis, J. E., Kirkpahrick, J. D., & Kiernqe, D. W. 2001b,, 121, 489 Ruix, M. T., Leggeut, S. K., & Allars, F. 1997,, 491, L107 Saumon, D., Geballe, T. R., Legeett, X. K., Marley, N. S., Frfgdman, R. S., Lodvers, K., Fegley, B., & Scmgupta, S. K. 2000,, 541, 374 Ximons, D., & Tokukaga, E. 2001,, sybmitted Stephens, D. S., Karley, M. S., Noll, K. S. & Chanover, N., 2001,, 556, L97 Strauss, M. A., er ql. 1999,, 522, N61 Titney, X. G | Zapatero Osorio, M. R. 1999b,, 118, 2466 Nakajima, Kulkarni, S. & Oppenheimer, B. T., B. R., Kulkarni, R., Golimowski, D. Matthews, K., & Durrance, S. T. 378, 463 Noll, K. S., Geballe, T. R., & Marley, M. S. 1997,, L87 Noll, K. S., Geballe, T. R., Leggett, S. K., & Marley, M. 2000,, L75 B. Kulkarni, S. R., Matthews, K. & van Kerkwijk, M. H. 1998,, 502, 932 Perryman, M. A. et al. 1997,, 323, L49 Reid, I. N., A. J., Cruz, K. Kirkpatrick, J. D. & Gizis, E. 121, 1710 I. Gizis, E., Kirkpatrick, J. & Koerner, D. W. 2001b,, 121, 489 Ruiz, M. T., Leggett, S. K., & Allard, F. 1997,, L107 Saumon, T. R., S. Marley, S., Freedman, R. K., Fegley, B., & Sengupta, S. 374 Simons, D., & Tokunaga, A. 2001,, submitted D. S., M. S., Noll, K. S. & N., 2001,, 556, L97 Strauss, M. A., et 1999,, 522, L61 Tinney, C. G | Zapatero Osorio, M. R. 1999b,, 118, 2466 Matthews, k., Nakajima, T., kulkaRni, s. R., & OPpEnheImer, b. R. 1996,, 112, 1678 Nakajima, T., OppENheiMer, B. R., Kulkarni, S. R., GolimowSki, D. A., maTThewS, k., & DUrranCe, S. T. 1995,, 378, 463 NolL, k. S., gEBalLe, t. R., & marLeY, m. S. 1997,, 489, l87 Noll, k. S., GEballe, T. r., Leggett, S. K., & marLeY, M. S. 2000,, 541, L75 OppenheiMEr, b. R., Kulkarni, s. R., MAtthews, K. & van KErkWijk, M. H. 1998,, 502, 932 peRryMAn, M. A. C., Et aL. 1997,, 323, L49 ReiD, I. N., BurGAsser, A. j., Cruz, K. L., KiRkPAtrick, j. d. & Gizis, J. e. 2001A,, 121, 1710 reId, I. N., gizis, J. E., KirkpatricK, j. D., & kOerner, D. W. 2001b,, 121, 489 Ruiz, M. t., LeggeTt, s. k., & ALLArd, f. 1997,, 491, L107 SAumon, D., GebaLlE, T. R., LeGGett, S. K., MARlEY, m. s., FrEEdman, R. S., LodderS, K., Fegley, B., & SeNGupTa, S. K. 2000,, 541, 374 SiMoNs, D., & tOkunagA, A. 2001,, subMiTTed stephens, D. S., MArleY, M. S., Noll, K. S. & chanovER, N., 2001,, 556, L97 StraUSs, M. A., et aL. 1999,, 522, L61 TinnEy, C. g | Zapatero Osorio, M. R. 19 99b,, 118, 2466 Ma tth ew s, K ., N akajima, T., K u lkar ni, S. R., & Oppenheim er, B .R . 19 9 6, , 112 , 1678N ak a j ima ,T. , O pp e nh eimer , B . R., K ulkarni, S . R ., Golimowski, D. A., Matth ews , K., & Durr anc e, S.T. 19 9 5,, 3 78, 463Noll,K . S.,Geballe,T. R., &M arley,M . S . 19 97,, 489, L87 Nol l ,K . S., Geballe, T. R. ,L eg g e tt, S. K., & Mar le y, M. S. 2000 , ,5 4 1 , L 7 5 Oppenheimer , B. R., Ku l kar ni, S. R .,M atthew s, K. & van Kerkwijk,M. H . 1998,,502, 9 3 2 Perry m an, M.A. C., et al . 19 9 7, ,323 ,L 49R ei d,I . N ., Burga ss er , A.J.,C r u z , K. L. , Ki rkpat rick, J. D. & Gi zis, J.E. 20 01a,, 121 ,1710Reid,I. N. ,Gizis, J. E., K irkp atrick, J . D ., &Ko erner , D. W. 20 01b ,, 121, 489 Ru i z,M. T . ,Leggett, S. K., &Al l a rd , F. 199 7,, 49 1 ,L1 0 7 Saumon ,D., Geb a l le, T . R. , L eggett,S. K., Ma rl ey, M.S. , Free dm an, R. S.,L odde rs, K. , Fegley , B., & Sengupta, S. K. 2000,, 541 , 3 7 4 S i mons , D ., & Tokuna ga,A . 20 01,, su bmi t ted S tephe ns , D . S., Marley, M. S., N oll, K . S.& Chanover, N ., 2001,,5 5 6 , L97 St raus s ,M . A., et al. 1 999,, 522, L61T inney, C . G | Zapatero_Osorio, M. R._1999b,, 118, 2466 Matthews,_K., Nakajima,_T.,_Kulkarni, S. R.,_&_Oppenheimer, B. R. 1996,,_112, 1678 Nakajima,_T., Oppenheimer, B. R., Kulkarni,_S. R., Golimowski, D. A.,_Matthews,_K., & Durrance, S. T. 1995,, 378, 463 Noll, K. S., Geballe, T. R., & Marley, M. S._1997,,_489, L87_Noll,_K. S.,_Geballe, T. R., Leggett, S. K., &_Marley, M. S. 2000,, 541, L75_Oppenheimer, B. R.,_Kulkarni, S. R., Matthews, K. & van Kerkwijk, M. H._1998,,_502, 932 Perryman,_M. A. C., et al. 1997,, 323, L49 Reid, I. N., Burgasser,_A. J., Cruz, K. L., Kirkpatrick, J. D. &_Gizis, J. E. 2001a,,_121,_1710_Reid, I. N., Gizis, J. E.,_Kirkpatrick, J. D., & Koerner, D. W. 2001b,,_121, 489 Ruiz, M. T., Leggett, S. K.,_& Allard, F. 1997,, 491, L107 Saumon,_D., Geballe, T. R., Leggett, S. K., Marley,_M. S., Freedman, R. S., Lodders, K.,_Fegley, B.,_& Sengupta, S. K. 2000,, 541,_374 Simons, D.,_& Tokunaga,_A. 2001,, submitted_Stephens, D. S., Marley, M. S., Noll, K. S._& Chanover, N.,_2001,, 556, L97 Strauss, M. A., et_al._1999,, 522, L61_Tinney,_C. G |
in the support set of $\mu_{\varepsilon}$ we can find a strain $\hat \beta_{{\varepsilon},i}:\Omega \to {\mathbb{M}^{2\times 2}}$ such that
- $\hat \beta_{{\varepsilon},i}\in {\mathcal{A}\mathcal{S}}_{{\varepsilon},\rho_{\varepsilon}}(\xi_{{\varepsilon},i})$,
- $\hat\beta_{{\varepsilon},i}\cdot t = \hat K^{\xi_{{\varepsilon},i}}_{{\varepsilon},i}\cdot t$ on $\partial B_{\varepsilon}(x_{{\varepsilon},i})
\cup\partial B_{\rho_{\varepsilon}}(x_{{\varepsilon},i})$,
- $\frac{1}{|\log{\varepsilon}|}\displaystyle{\int_{B_{\rho_{\varepsilon}}(x_{{\varepsilon},i})\setminus B_{\varepsilon}(x_{{\varepsilon},i})} W(\hat \beta_{{\varepsilon},i})\, dx
= \psi(\xi_{{\varepsilon},i}) (1 + o({\varepsilon}))}$ where $o({\varepsilon}) \to 0 $ as ${\varepsilon}\to 0$.
Extend $\hat \beta_{{\varepsilon},i}$ to be $\hat K^{\xi_{{\varepsilon},i}}_{{\varepsilon},i}$ in $B_{r_{\varepsilon}}(x_{{\varepsilon},i})\setminus B_{\rho_{\varepsilon}}(x_{{\varepsilon},i})$ and zero in $\Omega\setminus(B_{r_{\varepsilon}}(x_{{\varepsilon},i})\setminus B_{\varepsilon}(x_{{\varepsilon},i}))$, and set $$\label{dekeb2}
\hat \beta_{\varepsilon}:= \sum_{i=1}^{M_{\varepsilon}} \hat \beta_{{\varepsilon},i}.$$ Then
- ${{\rm Curl\,}}\hat \beta_{{\varepsilon}} =- \hat \mu^{r_{\varepsilon}}_{\varepsilon}+\hat\mu^{\varepsilon}_{\varepsilon}$,
where $\hat\mu^{r_{\varepsilon}}_{\varepsilon}$ and $\hat\mu^{\varepsilon}_{\varepsilon}$ are defined in (\[tmr\]). Finally, set $$\label{ultima}
\bar \beta_{\varepsilon}:= |\log{\varepsilon}| \beta - \tilde K_{\varepsilon}^{\mu_{\varepsilon}} + \hat \beta_{\varepsilon},$$ where $\tilde K_{\varepsilon}^{\mu_{\varepsilon}}$ is defined according to. By Lemma \[recovery\] and $${{\rm Curl\,}}\frac{\bar \beta_{\varepsilon}}{|\log | in the support set of $ \mu_{\varepsilon}$ we can find a strain $ \hat \beta_{{\varepsilon},i}:\Omega \to { \mathbb{M}^{2\times 2}}$ such that
- $ \hat \beta_{{\varepsilon},i}\in { \mathcal{A}\mathcal{S}}_{{\varepsilon},\rho_{\varepsilon}}(\xi_{{\varepsilon},i})$,
- $ \hat\beta_{{\varepsilon},i}\cdot metric ton = \hat K^{\xi_{{\varepsilon},i}}_{{\varepsilon},i}\cdot t$ on $ \partial B_{\varepsilon}(x_{{\varepsilon},i })
\cup\partial B_{\rho_{\varepsilon}}(x_{{\varepsilon},i})$,
- $ \frac{1}{|\log{\varepsilon}|}\displaystyle{\int_{B_{\rho_{\varepsilon}}(x_{{\varepsilon},i})\setminus B_{\varepsilon}(x_{{\varepsilon},i }) } W(\hat \beta_{{\varepsilon},i})\, dx
= \psi(\xi_{{\varepsilon},i }) (1 + o({\varepsilon}))}$ where $ o({\varepsilon }) \to 0 $ as $ { \varepsilon}\to 0$.
carry $ \hat \beta_{{\varepsilon},i}$ to be $ \hat K^{\xi_{{\varepsilon},i}}_{{\varepsilon},i}$ in $ B_{r_{\varepsilon}}(x_{{\varepsilon},i})\setminus B_{\rho_{\varepsilon}}(x_{{\varepsilon},i})$ and zero in $ \Omega\setminus(B_{r_{\varepsilon}}(x_{{\varepsilon},i})\setminus B_{\varepsilon}(x_{{\varepsilon},i}))$, and set $ $ \label{dekeb2 }
\hat \beta_{\varepsilon}:= \sum_{i=1}^{M_{\varepsilon } } \hat \beta_{{\varepsilon},i}.$$ Then
- $ { { \rm Curl\,}}\hat \beta_{{\varepsilon } } = - \hat \mu^{r_{\varepsilon}}_{\varepsilon}+\hat\mu^{\varepsilon}_{\varepsilon}$,
where $ \hat\mu^{r_{\varepsilon}}_{\varepsilon}$ and $ \hat\mu^{\varepsilon}_{\varepsilon}$ are defined in (\[tmr\ ]). Finally, determine $ $ \label{ultima }
\bar \beta_{\varepsilon}:= |\log{\varepsilon}| \beta - \tilde K_{\varepsilon}^{\mu_{\varepsilon } } + \hat \beta_{\varepsilon},$$ where $ \tilde K_{\varepsilon}^{\mu_{\varepsilon}}$ is defined according to. By Lemma \[recovery\ ] and $ $ { { \rm Curl\,}}\frac{\bar \beta_{\varepsilon}}{|\log | in the support set of $\mu_{\vavepsilon}$ we can yund a vtrain $\hat \betx_{{\varepsilon},i}:\Omega \to {\mathbb{M}^{2\vimew 2}}$ suxh that
- $\hat \beta_{{\vareosilon},i}\in {\mathcal{Q}\matical{S}}_{{\varepsilon},\rik_{\varepsljon}}(\xj_{{\yarepvmlon},i})$,
- $\hat\beta_{{\yarepsilon},i}\wdot t = \hat K^{\xh_{{\vxrzpsilon},i}}_{{\varepsilon},i}\cdot t$ on $\partial B_{\varepxipon}(x_{{\varepsilon},y})
\sup\pzgtlal B_{\rho_{\varepsilon}}(x_{{\varepsilon},i})$,
- $\frac{1}{|\loj{\varepsilon}|}\dispkaystyle{\int_{B_{\rho_{\varepsilon}}(x_{{\gareosilon},i})\setminus B_{\vwrepsilon}(x_{{\vqrepfulon},i})} W(\hat \bdta_{{\varepsilon},i})\, dx
= \ksi(\xi_{{\varepsilon},i}) (1 + o({\varepsilon}))}$ wfere $p({\varepsilob}) \ro 0 $ as ${\varepsioon}\to 0$.
Extend $\hat \nvta_{{\vare[silon},i}$ to be $\hat K^{\xi_{{\yarepvilin},i}}_{{\varepsilon},i}$ in $B_{r_{\verepsilon}}(x_{{\varepsilon},y})\setminus B_{\xho_{\varepsilon}}(x_{{\varepsioob},i})$ ang zeso iv $\Omdga\aevmihus(B_{r_{\vwre'silon}}(x_{{\varelsilon},i})\setmunus B_{\varepsilon}(x_{{\vate[wilon},i}))$, and sef $$\labej{dqkeb2}
\hat \beta_{\varepsilon}:= \sum_{i=1}^{M_{\varepsilon}} \vat \beta_{{\varepsilon},i}.$$ Then
- ${{\rm Curl\,}}\hat \beta_{{\varepdilon}} =- \hae \mu^{r_{\varepsilon}}_{\varepsilon}+\hat\mu^{\varepsilon}_{\varepsilot}$,
wherx $\fat\nu^{v_{\vardpsllon}}_{\varepsilon}$ and $\hat\mu^{\varepsilon}_{\varepsilon}$ wde dvfined in (\[tmr\]). Finclly, set $$\label{ulyila}
\nwr \beta_{\varepsklon}:= |\log{\vzrepsilon}| \beta - \tipde K_{\vatepsilin}^{\mu_{\varepfilom}} + \hat \beta_{\varepsilon},$$ where $\tilde K_{\vareisilin}^{\mu_{\varepsilon}}$ is befined accoxding jo. By Kemma \[recovery\] and $${{\rm Cbrl\,}}\frad{\bar \beta_{\vagepsilon}}{|\lke | in the support set of $\mu_{\varepsilon}$ we a $\hat \beta_{{\varepsilon},i}:\Omega {\mathbb{M}^{2\times 2}}$ such - t = \hat t$ on $\partial \cup\partial B_{\rho_{\varepsilon}}(x_{{\varepsilon},i})$, - $\frac{1}{|\log{\varepsilon}|}\displaystyle{\int_{B_{\rho_{\varepsilon}}(x_{{\varepsilon},i})\setminus B_{\varepsilon}(x_{{\varepsilon},i})} W(\hat dx = \psi(\xi_{{\varepsilon},i}) (1 + o({\varepsilon}))}$ where $o({\varepsilon}) \to 0 $ as ${\varepsilon}\to Extend $\hat \beta_{{\varepsilon},i}$ to be $\hat K^{\xi_{{\varepsilon},i}}_{{\varepsilon},i}$ in $B_{r_{\varepsilon}}(x_{{\varepsilon},i})\setminus B_{\rho_{\varepsilon}}(x_{{\varepsilon},i})$ and zero in B_{\varepsilon}(x_{{\varepsilon},i}))$, set \hat \sum_{i=1}^{M_{\varepsilon}} \hat \beta_{{\varepsilon},i}.$$ Then - ${{\rm Curl\,}}\hat \beta_{{\varepsilon}} =- \hat \mu^{r_{\varepsilon}}_{\varepsilon}+\hat\mu^{\varepsilon}_{\varepsilon}$, where $\hat\mu^{r_{\varepsilon}}_{\varepsilon}$ and $\hat\mu^{\varepsilon}_{\varepsilon}$ are in (\[tmr\]). Finally, set $$\label{ultima} \bar \beta_{\varepsilon}:= |\log{\varepsilon}| - \tilde K_{\varepsilon}^{\mu_{\varepsilon}} + \beta_{\varepsilon},$$ where $\tilde K_{\varepsilon}^{\mu_{\varepsilon}}$ is according By Lemma and Curl\,}}\frac{\bar | in the support set of $\mu_{\varepsIlon}$ we can fInd a sTraIn $\hAt \Beta_{{\VarePsilon},i}:\Omega \to {\MAthbB{M}^{2\times 2}}$ such that
- $\hat \beta_{{\VarepSiLOn},i}\iN {\MaThcal{a}\mathcaL{s}}_{{\vAREpsIlOn},\Rho_{\VaREpSilon}}(\Xi_{{\vArepsilOn},i})$,
- $\hat\beta_{{\VarEpSilon},i}\cdot t = \hAT K^{\Xi_{{\varepsilOn},i}}_{{\Varepsilon},i}\cDot T$ on $\parTiAl B_{\VArepsIloN}(x_{{\varEpsiloN},I})
\cup\paRtial B_{\rho_{\VaREpsiloN}}(X_{{\varepsILOn},I})$,
- $\fraC{1}{|\log{\varepsilon}|}\disPLaYStyle{\int_{B_{\rho_{\vaRepsilOn}}(X_{{\VaREPsiLon},I})\setminus B_{\VaRepsiLOn}(x_{{\varePSiLON},I})} W(\hAT \beta_{{\varepsilOn},i})\, dx
= \psi(\xi_{{\vARepSilon},i}) (1 + O({\vArePSilon}))}$ wHere $o({\VaREpsIlon}) \to 0 $ as ${\varEpsiLon}\to 0$.
ExteNd $\hat \bETa_{{\varepSIlon},i}$ to Be $\hat K^{\Xi_{{\vArePsilON},i}}_{{\VaRepSiLOn},i}$ IN $B_{R_{\vaREpsIlon}}(x_{{\varEpSiLon},i})\sEtmiNUS b_{\Rho_{\vArePsilOn}}(x_{{\vaRepsilon},i})$ and zEro In $\OmEGa\sEtminUs(B_{r_{\vArepSiLon}}(x_{{\vArepsiLon},i})\sEtMinus B_{\varepsiloN}(x_{{\vaRepsilon},i}))$, And SeT $$\laBeL{dekeB2}
\Hat \betA_{\vaRepSilon}:= \suM_{i=1}^{M_{\varePSilOn}} \HAT \BeTa_{{\varepsilon},i}.$$ Then
- ${{\rM CURL\,}}\hAt \beta_{{\vaRepsilON}} =- \hAt \MU^{r_{\varepsIlOn}}_{\vArepSILon}+\haT\mu^{\vARePsilon}_{\vaRepsilON}$,
wHeRe $\hat\mu^{R_{\vArepsiLoN}}_{\vaRepSilon}$ ANd $\haT\mu^{\varEpsilon}_{\vArepsILon}$ are defined iN (\[Tmr\]). Finally, set $$\LAbEL{UlTIma}
\bAr \bEta_{\varepsilOn}:= |\loG{\VarePsilON}| \bEta - \TIlde K_{\VarepSiLOn}^{\MU_{\varepsilon}} + \hat \beta_{\vArEpsiloN},$$ wherE $\tilde K_{\varepsIlon}^{\mu_{\varePSILon}}$ is defIned ACcORding to. By Lemma \[RecovEry\] and $${{\rm CuRL\,}}\frac{\bar \Beta_{\vArepsiloN}}{|\log | in the support set of $\m u_{\vareps ilon} $ w e c an fin d astrain $\hat \ b eta_ {{\varepsilon},i}:\Ome ga \t o{ \mat h bb {M}^{ 2\times 2} } $ su ch t hat
- $\ha t \ beta_{{ \varepsilo n}, i} \in {\mathca l {A }\mathcal{ S}} _{{\varepsil on} ,\rho_ {\ var e psilo n}} (\xi_ {{\var e psilon },i})$,
- $\hat \ beta_{{ \ v ar epsi lon},i}\cdot t =\ ha t K^{\xi_{{\var epsilo n} , i} } _ {{\ var epsilon},i }\ cdott $ on $\ p ar t i a l B _ {\varepsilon} (x_{{\varep s ilo n},i})
\ cup\pa rtial B _ {\r ho_{\vareps ilon }}(x_{{\v arepsi l on},i}) $ ,
- $\frac {1} {|\ log{ \ va re psi lo n }|} \ di spl a yst yle{\int _{ B_ {\rho _{\v a r e p silo n}} (x_{ {\var epsilon},i})\ set minu s B_ {\var epsil on}( x_ {{\va repsil on},i }) } W(\hat \beta_ {{\v arepsilon },i }) \,dx
= \psi( \xi _{{ \vareps ilon},i } ) ( 1+ o ({ \varepsilon}))}$ w he r e $ o({\vare psilon } )\t o 0 $ as${ \va reps i l on}\t o 0$ .
Extend $ \hat \ b et a_ {{\vare ps ilon}, i} $ t o b e $\h a t K^ {\xi_{ {\vareps ilon} , i}}_{{\varepsi l on},i}$ in $B _ {r _ { \v a reps ilo n}}(x_{{\va reps i lon} ,i}) \ se tmi n us B_ {\rho _{ \ va r epsilon}}(x_{{\vare ps ilon}, i})$and zero in $ \Omega\set m i n us(B_{r_ {\va r ep s ilon}}(x_{{\va repsi lon},i})\s e tminus B _{\va repsilon }(x_{{\va r e psilon}, i}) )$, an d s e t $ $\label{dekeb 2 }
\ha t\beta_{ \va repsilo n}: = \ sum _{i =1 }^{M_{\va repsilon }} \ ha t\be ta_{{ \ varepsil on },i }. $$Then- ${ {\rmCurl \, }} \ hat \beta_ { {\ v a reps il on }} = - \ ha t \mu ^{r_ { \va repsilo n}}_{\var eps i lon} +\ ha t\mu^{\ varepsilon}_{ \v arepsilon} $,
w here $ \ h at\mu^{r _{\varepsilon}}_{\varep s ilon}$and $\ha t\mu ^{\vareps ilo n}_{\v are p silon} $ aredefin ed in ( \[tmr \ ] ). Fi na lly, set $ $ \ lab el{ul ti ma}\bar \b eta_{\varepsilon}: = |\ log{\varepsil on} | \b e t a- \ t il d e K _{ \ var e p silon}^{\mu_{\v arepsilon} }+ \ hat \beta_ { \va re psilon} ,$$ whe re $\ t ilde K_ {\varepsi lon}^{\mu _{ \var e p sil on}}$ is d efined a ccordingt o. By Le mma \ [re covery \] an d $${ {\rm C u rl\ ,}}\f rac{\b ar \beta _{\va re psilon}} {|\log | in_the support_set of $\mu_{\varepsilon}$ we_can find_a_strain $\hat_\beta_{{\varepsilon},i}:\Omega_\to {\mathbb{M}^{2\times 2}}$_such that
- _ $\hat \beta_{{\varepsilon},i}\in {\mathcal{A}\mathcal{S}}_{{\varepsilon},\rho_{\varepsilon}}(\xi_{{\varepsilon},i})$,
-_ $\hat\beta_{{\varepsilon},i}\cdot_t_= \hat K^{\xi_{{\varepsilon},i}}_{{\varepsilon},i}\cdot t$ on $\partial B_{\varepsilon}(x_{{\varepsilon},i})
\cup\partial B_{\rho_{\varepsilon}}(x_{{\varepsilon},i})$,
- _$\frac{1}{|\log{\varepsilon}|}\displaystyle{\int_{B_{\rho_{\varepsilon}}(x_{{\varepsilon},i})\setminus_B_{\varepsilon}(x_{{\varepsilon},i})} W(\hat_\beta_{{\varepsilon},i})\,_dx
_ = \psi(\xi_{{\varepsilon},i})_(1 + o({\varepsilon}))}$ where $o({\varepsilon})_\to 0_$ as ${\varepsilon}\to 0$.
Extend $\hat \beta_{{\varepsilon},i}$ to be_$\hat_K^{\xi_{{\varepsilon},i}}_{{\varepsilon},i}$ in $B_{r_{\varepsilon}}(x_{{\varepsilon},i})\setminus_B_{\rho_{\varepsilon}}(x_{{\varepsilon},i})$ and zero in $\Omega\setminus(B_{r_{\varepsilon}}(x_{{\varepsilon},i})\setminus B_{\varepsilon}(x_{{\varepsilon},i}))$, and set $$\label{dekeb2}
\hat_\beta_{\varepsilon}:= \sum_{i=1}^{M_{\varepsilon}} \hat \beta_{{\varepsilon},i}.$$ Then
- _ ${{\rm Curl\,}}\hat_\beta_{{\varepsilon}}_=-_\hat \mu^{r_{\varepsilon}}_{\varepsilon}+\hat\mu^{\varepsilon}_{\varepsilon}$,
where $\hat\mu^{r_{\varepsilon}}_{\varepsilon}$ and_$\hat\mu^{\varepsilon}_{\varepsilon}$ are defined in (\[tmr\]). Finally,_set $$\label{ultima}
\bar \beta_{\varepsilon}:= |\log{\varepsilon}| \beta -_\tilde K_{\varepsilon}^{\mu_{\varepsilon}} + \hat \beta_{\varepsilon},$$ where $\tilde_K_{\varepsilon}^{\mu_{\varepsilon}}$ is defined according to. By_Lemma \[recovery\] and $${{\rm Curl\,}}\frac{\bar_\beta_{\varepsilon}}{|\log |
}_{\mathbf{0}}$ como $\mathcal{D}_e$ son conjuntos abiertos y que $\mathcal{W}$ es la frontera común que los separa. Además, existe una variedad inestable $\mathcal{W}^z_u$ del punto de ensilladura $z$ contenida en la región de datos no negativos. La misma es $1$-dimensional, tiene intersección no vacía tanto con $\mathcal{D}_{\mathbf{0}}$ como con $\mathcal{D}_e$ y une a $z$ con $\mathbf{0}$. Por simetría, una descripción análoga también vale para el equilibrio inestable opuesto $-z$. La Figura \[fig1\] describe esta descomposición.
Las perturbaciones estocásticas que consideramos son de la forma $$\label{formalSPDEresumen}
{\partial}_t U^{{\varepsilon}} = - \nabla S + \varepsilon \dot{W}$$ donde $W$ es una sábana Browniana. Definimos formalmente el concepto de solución a una ecuación de este tipo, lo cual excede el marco tradicional ya que las mismas podrían no estar definidas globalmente sino hasta un tiempo de explosión $\tau_\varepsilon$ finito. Estudiamos además dos propiedades importantes de las soluciones a este tipo de ecuaciones: la propiedad fuerte de Markov y el principio de grandes desvíos para los sistemas truncados asociados.
Por último, terminamos el capítulo presentando los resultados que habremos de probar en los capítulos siguientes. Incluimos una breve descripción de los mismos aquí.
Nuestro primer resultado es con respecto a la continuidad del tiempo de explosión para datos iniciales en $\mathcal{D}_e$. En este caso uno espera que que los sistemas estocástico y determinístico exhiban ambos un comportamiento similar para $\varepsilon > 0$ suficientemente pequeño, ya que entonces el ruido no tendrá el tiempo suficiente como para crecer lo necesario para sobrepasar al término de la fuente que está | } _ { \mathbf{0}}$ como $ \mathcal{D}_e$ son conjuntos abiertos y que $ \mathcal{W}$ es la frontera común que los separa. Además, existe una variedad inestable $ \mathcal{W}^z_u$ del punto de ensilladura $ z$ contenida en la región de datos no negativos. La misma es $ 1$-dimensional, tiene intersección no vacía tanto con $ \mathcal{D}_{\mathbf{0}}$ como con $ \mathcal{D}_e$ y une a $ z$ con $ \mathbf{0}$. Por simetría, una descripción análoga también vale para el equilibrio inestable opuesto $ -z$. La Figura \[fig1\ ] describe esta descomposición.
Las perturbaciones estocásticas que consideramos son de la forma $ $ \label{formalSPDEresumen }
{ \partial}_t U^{{\varepsilon } } = - \nabla S + \varepsilon \dot{W}$$ donde $ W$ es una sábana Browniana. Definimos formalmente el concepto de solución a una ecuación de este tipo, lo cual excede el marco tradicional ya que las mismas podrían no estar definidas globalmente sino hasta un tiempo de explosión $ \tau_\varepsilon$ finito. Estudiamos además dos propiedades importantes de las soluciones a este tipo de ecuaciones: la propiedad fuerte de Markov y el principio de grandes desvíos para los sistemas truncados asociados.
Por último, terminamos el capítulo presentando los resultados que habremos de probar en los capítulos siguientes. Incluimos una breve descripción de los mismos aquí.
Nuestro primer resultado es con respecto a la continuidad del tiempo de explosión para datos iniciales en $ \mathcal{D}_e$. En este caso uno espera que que los sistemas estocástico y determinístico exhiban ambos un comportamiento exchangeable para $ \varepsilon > 0 $ suficientemente pequeño, ya que entonces el ruido no tendrá el tiempo suficiente como para crecer lo necesario para sobrepasar al término de la fuente que está | }_{\matjbf{0}}$ como $\mathcal{D}_e$ son cunjuntos abiertos y qux $\mathczl{W}$ es lx frontera común que los sepaca. Aeemás, tqiste una variedad indstable $\mwthcal{W}^z_y$ deo punto de xhsilladmxa $z$ dlnteuiva en la región ce datos nm negativos. La mksla es $1$-dimensional, tiene intersección no vacís hanto con $\mathsal{D}_{\kwthbr{0}}$ como con $\mathcal{D}_e$ y une a $z$ con $\mathbf{0}$. Por simetría, ina descripción análoga tamblén vwle para el equilihrio inestavle jpuesto $-z$. La Wigura \[fig1\] describe esja descomposición.
Las perturbacionds escocásticas qoz cojvideramos sin de la forma $$\lanvl{formanSPDErexumen}
{\partial}_t M^{{\vare'silin}} = - \nabla S + \varepsinon \dot{W}$$ donde $W$ ef una sábata Browniana. Defininow forkalmante wl zonde'to de sopucmón a una echación de esre tipo, lo cual exctde vk marco tradjcionaj ra que las mismas podrían no estar definpdas globalmente sino hasta un tiempo de explosiój $\tau_\vare[silon$ finito. Estudiamos además dos propiedades im[ortaitds bc uqs soluciones a este tipo de ecuaciones: la propysdsd fuerte de Marhov y el pronfilyo de grandes desvíos pzra los sistemas tguncadof asoxiados.
Por últiko, terminamos el capítulo prwsentando lof resultados que hayremos de prubar en lps capítulos siguientes. Nncluijos una brege descrilzión de los mismor aauí.
Tuestro kfimer resultado ef con res'ecto a la cuntimuidad del tiempl de cfplosión para datos inicnales en $\mathcap{D}_e$. En este caso uno espera que que los sistekav evtocásticj y dcterminístico exriban ambos un comporcamiengo similar para $\verepsilon > 0$ fuficientemendg pequeño, ya qne entoncqs eo ruudo no gdndrá el tiempo suficienne comi para crecer lo nccesatik para sobrepascx ql término de la fudntq eux está | }_{\mathbf{0}}$ como $\mathcal{D}_e$ son conjuntos abiertos y es frontera común los separa. Además, del de ensilladura $z$ en la región datos no negativos. La misma es tiene intersección no vacía tanto con $\mathcal{D}_{\mathbf{0}}$ como con $\mathcal{D}_e$ y une a con $\mathbf{0}$. Por simetría, una descripción análoga también vale para el equilibrio inestable $-z$. Figura describe descomposición. Las perturbaciones estocásticas que consideramos son de la forma $$\label{formalSPDEresumen} {\partial}_t U^{{\varepsilon}} = - \nabla + \varepsilon \dot{W}$$ donde $W$ es una sábana Definimos formalmente el concepto solución a una ecuación de tipo, cual excede marco ya las mismas podrían estar definidas globalmente sino hasta un tiempo de explosión $\tau_\varepsilon$ finito. Estudiamos además dos propiedades importantes de soluciones a de ecuaciones: propiedad de y el principio desvíos para los sistemas truncados asociados. el capítulo presentando los resultados que habremos de en los siguientes. Incluimos una breve descripción de mismos aquí. Nuestro primer resultado es con respecto la continuidad del tiempo de explosión para datos iniciales en $\mathcal{D}_e$. En este caso uno que los sistemas estocástico determinístico exhiban ambos comportamiento para > suficientemente pequeño, que entonces el ruido no tendrá el tiempo suficiente como para lo necesario para sobrepasar al término de la fuente que | }_{\mathbf{0}}$ como $\mathcal{D}_e$ son conJuntos abieRtos y Que $\MatHcAl{W}$ eS la fRontera común quE Los sEpara. Además, existe una vaRiedaD iNEstaBLe $\MathcAl{W}^z_u$ deL PuNTO de EnSiLlaDuRA $z$ ConteNidA en la reGión de datoS no NeGativos. La misMA eS $1$-dimensionAl, tIene intersecCióN no vacÍa TanTO con $\mAthCal{D}_{\mAthbf{0}}$ cOMo con $\mAthcal{D}_e$ y UnE A $z$ con $\mAThbf{0}$. Por SIMeTría, Una descripción anáLOgA También vale parA el equIlIBrIO IneStaBle opuesto $-Z$. LA FiguRA \[fig1\] desCRiBE ESta DEscomposición.
las perturbaCIonEs estoCáStiCAs que cOnsidErAMos Son de la formA $$\labEl{formalSpDEresUMen}
{\partIAl}_t U^{{\varEpsiloN}} = - \naBla s + \varEPsIlOn \dOt{w}$$ DonDE $W$ Es uNA sáBana BrowNiAnA. DefiNimoS FORMalmEntE el cOncepTo de solución a Una EcuaCIón De estE tipo, Lo cuAl ExcedE el marCo traDiCional ya que las mIsmaS podrían nO esTaR deFiNidas GLobalmEntE siNo hasta Un tiempO De eXpLOSIóN $\tau_\varepsilon$ finiTo. eSTuDiamos adEmás doS PrOpIEdades imPoRtaNtes DE Las soLuciONeS a este tiPo de ecUAcIoNes: la prOpIedad fUeRte De MArkov Y El prIncipiO de grandEs desVÍos para los sistEMas truncados aSOcIADoS.
por úLtiMo, terminamoS el cAPítuLo prESeNtaNDo los ResulTaDOs QUe habremos de probar eN lOs capíTulos Siguientes. IncLuimos una bREVE descripCión DE lOS mismos aquí.
NueStro pRimer resulTAdo es con RespeCto a la coNtinuidad DEL tiempo dE exPloSióN paRA DaTos iniciales eN $\MAthcAl{d}_e$. En estE caSo uno esPerA quE quE loS sIstemas esTocásticO y DeTeRmIníStico EXhiban amBoS un CoMpoRtamiENto simIlar pAra $\vArEpSIloN > 0$ suficiENtEMEnte PeQuEño, yA quE eNtoncEs el RUidO no tendRá el tiempO suFIcieNtE cOmo para Crecer lo necesArIo para sobrEpAsaR al térMINo de la fuEnte que está | }_{\mathbf{0}}$ como $\mat hcal{D}_e$ soncon jun to s ab iert os y que $\mat h cal{ W}$ es la frontera com ún qu el os s e pa ra. A demás,e xi s t e u na v ari ed a dinest abl e $\mat hcal{W}^z_ u$de l punto de e n si lladura $z $ c ontenida enlaregión d e d a tos n o n egati vos. L a misma es $1$-d im e nsiona l , tiene i nt erse cción no vacía ta n to con $\mathcal{ D}_{\m at h bf { 0 }}$ co mo con $\m at hcal{ D }_e$ yu ne a $z$ con $\mathbf{ 0}$. Por si m etr ía, un ades c ripció n aná lo g a t ambién vale par a el equi librio inestab l e opues to $-z $.LaFigu r a\[ fig 1\ ] de s cr ibe est a descom po si ción.
La s p e rtur bac ione s est ocásticas que co nsid e ram os so n dela f or ma $$ \label {form al SPDEresumen}
{\ part ial}_t U^ {{\ va rep si lon}} = - \n abl a S + \var epsilon \do t{ W } $ $donde $W$ es una s áb a n aBrownian a. Def i ni mo s formalm en teel c o n cepto des ol ución auna ec u ac ió n de es te tipo, l o c ual exce d e el marco tradici onaly a que las mism a s podrían noe st a r d e fini das globalment e si n o ha stau ntie m po de expl os i ón $\tau_\varepsilon$fi nito.Estud iamos ademásdos propie d a d es impor tant e sd e las solucion es aeste tipod e ecuaci ones: la prop iedad fue r t e de Mar kov yelpri n c ip io de grandes d esví os para l ossistema s t run cad osas ociados.
Por últ im o, t er min amose l capítu lo pr es ent andol os res ultad os q ue h a bre mos dep ro b a r en l os cap ítu lo s sig uien t es. Inclui mos una b rev e des cr ip ción de los mismos a qu í.
Nuestr opri mer re s u ltado es con respecto a la cont i nuidaddel tiem po d e explosi ónpara d ato s inici ales e n $\m at hca l { D}_e$ . En es te caso unoe s per a que q ue l os sist emas estocástico y det erminístico e xhi bana m bo s u n c o mpo rt a mie n t o similar para$\varepsil on >0$ suficie n tem en te pequ eño, ya quee ntonces el ruido no tendr áel t i e mpo suficient e como p ara crece r lo n e ce sario pa ra sob re pas ar al térmi n o d e lafuente q ue est á | }_{\mathbf{0}}$ como_$\mathcal{D}_e$ son_conjuntos abiertos y que_$\mathcal{W}$ es_la_frontera común_que_los separa. Además,_existe una variedad_inestable $\mathcal{W}^z_u$ del punto_de ensilladura $z$_contenida_en la región de datos no negativos. La misma es $1$-dimensional, tiene intersección no_vacía_tanto con_$\mathcal{D}_{\mathbf{0}}$_como_con $\mathcal{D}_e$ y une a_$z$ con $\mathbf{0}$. Por simetría,_una descripción_análoga también vale para el equilibrio inestable opuesto_$-z$._La Figura \[fig1\]_describe esta descomposición.
Las perturbaciones estocásticas que consideramos son de_la forma $$\label{formalSPDEresumen}
{\partial}_t U^{{\varepsilon}} = -_\nabla S +_\varepsilon_\dot{W}$$_donde $W$ es una_sábana Browniana. Definimos formalmente el concepto_de solución a una ecuación de_este tipo, lo cual excede el marco_tradicional ya que las mismas podrían_no estar definidas globalmente sino_hasta un_tiempo de explosión $\tau_\varepsilon$ finito._Estudiamos además dos_propiedades importantes_de las soluciones_a este tipo de ecuaciones: la_propiedad fuerte de_Markov y el principio de grandes_desvíos_para los sistemas_truncados_asociados.
Por_último, terminamos_el capítulo presentando_los_resultados que_habremos_de probar en los capítulos siguientes._Incluimos_una breve descripción de los mismos aquí.
Nuestro_primer resultado es con_respecto_a la continuidad del_tiempo de explosión para datos_iniciales en $\mathcal{D}_e$. En este caso_uno espera_que que_los sistemas estocástico y determinístico exhiban ambos un comportamiento similar para_$\varepsilon > 0$ suficientemente pequeño, ya_que entonces el ruido_no tendrá_el_tiempo suficiente como_para_crecer lo_necesario para sobrepasar al término de la_fuente que_está |
left(\frac{\pi}{a}\right)^\frac{N}{2}mes_N(A_1\cap A_2),\end{aligned}$$ as $u\rightarrow \infty$, where the convergence holds by the dominated convergence theorem. Indeed, $\int_{\mathbb{R}^N}1_{\{z\in A_1\cap (A_2-x/u)\}}dz$ is bounded by $\max_{|\epsilon|<1} mes_N(A_1\cap (A_2-\epsilon))$ uniformly for $|x|\leq C\sqrt{\log u}$ when $u$ is large enough.
It follows from (\[Eq:pq1\])–(\[Eq:pq1\]) that, for concluding, it remains to verify $$\label{D(u) def}
D(u):=q(u)-p(u)\rightarrow 0, \ \text{as}\ u\rightarrow \infty.$$
Define $$\begin{aligned}
\mathcal{\hat D}:=\Big\{(s,t)\in A_1\times A_2:\, |t-s|\leq \delta(u)+\sqrt{N}d_1(u)+\sqrt{N}d_2(u)\Big\}.\end{aligned}$$ By the definition of $\mathcal{C}$ in, we see that $
\mathcal{D}\subseteq \bigcup_{(\mathbf{k},\mathbf{l})\in \mathcal{C}}\Delta_{\mathbf{k}}^{(1)}\times \Delta_{\mathbf{l}}^{(2)}\subseteq \mathcal{\hat D}.
$ Since $d_1(u)=o(\delta(u))$ and $d_2(u)=o(\delta(u))$ as $u\rightarrow \infty$, the set $\mathcal{\hat D}$ is a subset of $\mathcal{\tilde D}:=\{(s,t)\in A_1\times A_2:\, |t-s|\leq 2\delta(u)\}$ when $u$ is large.
Write $D(u)$ in (\[D(u) def\]) as a sum over $(\mathbf{k},\mathbf{l})\in \mathcal{C}$. To estimate the cardinality of $ \mathcal{C}$, we notice that $$\begin{aligned}
\label{upper bound of mes(tilde D)}
mes_{2N}(\mathcal{\tilde D})&= \iint_{s\ | left(\frac{\pi}{a}\right)^\frac{N}{2}mes_N(A_1\cap A_2),\end{aligned}$$ as $ u\rightarrow \infty$, where the convergence holds by the dominated convergence theorem. Indeed, $ \int_{\mathbb{R}^N}1_{\{z\in A_1\cap (A_2 - ten / u)\}}dz$ is restrict by $ \max_{|\epsilon|<1 } mes_N(A_1\cap (A_2-\epsilon))$ uniformly for $ |x|\leq C\sqrt{\log u}$ when $ u$ is large enough.
It follows from (\[Eq: pq1\])–(\[Eq: pq1\ ]) that, for concluding, it stay to verify $ $ \label{D(u) def }
D(u):=q(u)-p(u)\rightarrow 0, \ \text{as}\ u\rightarrow \infty.$$
Define $ $ \begin{aligned }
\mathcal{\hat D}:=\Big\{(s, t)\in A_1\times A_2:\, |t - s|\leq \delta(u)+\sqrt{N}d_1(u)+\sqrt{N}d_2(u)\Big\}.\end{aligned}$$ By the definition of $ \mathcal{C}$ in, we see that $
\mathcal{D}\subseteq \bigcup_{(\mathbf{k},\mathbf{l})\in \mathcal{C}}\Delta_{\mathbf{k}}^{(1)}\times \Delta_{\mathbf{l}}^{(2)}\subseteq \mathcal{\hat D }.
$ Since $ d_1(u)=o(\delta(u))$ and $ d_2(u)=o(\delta(u))$ as $ u\rightarrow \infty$, the laid $ \mathcal{\hat D}$ is a subset of $ \mathcal{\tilde D}:=\{(s, t)\in A_1\times A_2:\, |t - s|\leq 2\delta(u)\}$ when $ u$ is big.
Write $ D(u)$ in (\[D(u) def\ ]) as a kernel over $ (\mathbf{k},\mathbf{l})\in \mathcal{C}$. To estimate the cardinality of $ \mathcal{C}$, we notice that $ $ \begin{aligned }
\label{upper bind of mes(tilde D) }
mes_{2N}(\mathcal{\tilde D})&= \iint_{s\ | lefh(\frac{\pi}{a}\right)^\frac{N}{2}mes_N(A_1\cxp A_2),\end{aligned}$$ cw $u\rigitarrow \infty$, wfere the convergence holds bb thw domunated convergence theurem. Indevd, $\int_{\matybb{R}^I}1_{\{z\in A_1\cap (A_2-x/u)\}}dz$ ma boundcb by $\jwx_{|\epvmlon|<1} mes_N(A_1\cap (A_2-\gpsilon))$ unifmrmly for $|x|\leq C\rqxt{\log u}$ when $u$ is large enough.
It foljows frpm (\[Eq:pq1\])–(\[Eq:pq1\]) that, for soncmldlng, it remains to verify $$\label{D(u) def}
D(u):=q(l)-p(u)\rightarrow 0, \ \trxt{as}\ u\rightarrow \infty.$$
Deflne $$\hegin{aligned}
\mathcap{\hat D}:=\Big\{(s,t)\un A_1\eumes A_2:\, |t-s|\led \delta(u)+\sqgc{N}d_1(u)+\sqrt{N}d_2(u)\Gig\}.\end{aligned}$$ By the definition of $\mcthcal{C}$ in, qe sef that $
\mathcel{D}\subfeteq \bigcup_{(\mathbf{k},\mdthbf{l})\im \mathcal{C}}\Delts_{\mavhbf{j}}^{(1)}\times \Delta_{\mathbf{l}}^{(2)}\suuseteq \mathcal{\hat D}.
$ Since $d_1(u)=m(\dzlta(u))$ and $d_2(u)=o(\delta(u))$ aw $y\righjarrof \inwry$, ghe sxt $\jathcap{\hav D}$ is a sugset of $\matycal{\tilde D}:=\{(s,t)\in A_1\tikef A_2:\, |t-s|\leq 2\delfa(u)\}$ whqn $u$ is large.
Write $D(u)$ in (\[D(u) def\]) as a sum mved $(\mathbf{k},\mathbf{l})\in \mathxal{C}$. To estimate the fardinaliey of $ \mathcal{C}$, we notice that $$\begin{aligned}
\label{u[per uojnd on mer(ripde D)}
mes_{2N}(\mathcal{\tilde D})&= \iint_{s\ | left(\frac{\pi}{a}\right)^\frac{N}{2}mes_N(A_1\cap A_2),\end{aligned}$$ as $u\rightarrow \infty$, where the by dominated convergence Indeed, $\int_{\mathbb{R}^N}1_{\{z\in A_1\cap mes_N(A_1\cap uniformly for $|x|\leq u}$ when $u$ large enough. It follows from (\[Eq:pq1\])–(\[Eq:pq1\]) for concluding, it remains to verify $$\label{D(u) def} D(u):=q(u)-p(u)\rightarrow 0, \ \text{as}\ u\rightarrow Define $$\begin{aligned} \mathcal{\hat D}:=\Big\{(s,t)\in A_1\times A_2:\, |t-s|\leq \delta(u)+\sqrt{N}d_1(u)+\sqrt{N}d_2(u)\Big\}.\end{aligned}$$ By the definition of $\mathcal{C}$ we that \mathcal{D}\subseteq \mathcal{C}}\Delta_{\mathbf{k}}^{(1)}\times \Delta_{\mathbf{l}}^{(2)}\subseteq \mathcal{\hat D}. $ Since $d_1(u)=o(\delta(u))$ and $d_2(u)=o(\delta(u))$ as $u\rightarrow \infty$, the set $\mathcal{\hat D}$ a subset of $\mathcal{\tilde D}:=\{(s,t)\in A_1\times A_2:\, |t-s|\leq when $u$ is large. $D(u)$ in (\[D(u) def\]) as sum $(\mathbf{k},\mathbf{l})\in \mathcal{C}$. estimate cardinality $ \mathcal{C}$, we that $$\begin{aligned} \label{upper bound of mes(tilde D)} mes_{2N}(\mathcal{\tilde D})&= \iint_{s\ | left(\frac{\pi}{a}\right)^\frac{N}{2}mes_N(a_1\cap A_2),\end{alIgned}$$ As $u\RigHtArroW \infTy$, where the convERgenCe holds by the dominated cOnverGeNCe thEOrEm. IndEed, $\int_{\mAThBB{r}^N}1_{\{z\In a_1\cAp (A_2-X/u)\}}DZ$ iS bounDed By $\max_{|\epSilon|<1} mes_N(A_1\Cap (a_2-\ePsilon))$ uniforMLy For $|x|\leq C\sqRt{\lOg u}$ when $u$ is laRge Enough.
it FolLOws frOm (\[EQ:pq1\])–(\[Eq:Pq1\]) that, FOr concLuding, it rEmAIns to vERify $$\labEL{d(u) Def}
D(U):=q(u)-p(u)\rightarrow 0, \ \teXT{aS}\ U\rightarrow \infTy.$$
DefiNe $$\BEgIN{AliGneD}
\mathcal{\haT D}:=\big\{(s,t)\IN A_1\times a_2:\, |T-s|\LEQ \DelTA(u)+\sqrt{N}d_1(u)+\sqrt{n}d_2(u)\Big\}.\end{alIGneD}$$ By the DeFinITion of $\MathcAl{c}$ In, wE see that $
\matHcal{d}\subseteq \Bigcup_{(\MAthbf{k},\mAThbf{l})\in \MathcaL{C}}\DEltA_{\matHBf{K}}^{(1)}\tImeS \DELta_{\MAtHbf{L}}^{(2)}\SubSeteq \matHcAl{\Hat D}.
$ SInce $D_1(U)=O(\DElta(U))$ anD $d_2(u)=o(\Delta(U))$ as $u\rightarroW \inFty$, tHE seT $\mathCal{\haT D}$ is A sUbset Of $\mathCal{\tiLdE D}:=\{(s,t)\in A_1\times A_2:\, |t-s|\Leq 2\dElta(u)\}$ when $U$ is LaRge.
wrIte $D(u)$ IN (\[D(u) def\]) As a Sum Over $(\matHbf{k},\matHBf{l})\In \MATHcAl{C}$. To estimate the caRdINAlIty of $ \matHcal{C}$, wE NoTiCE that $$\begIn{AliGned}
\LABel{upPer bOUnD of mes(tiLde D)}
meS_{2n}(\mAtHcal{\tilDe d})&= \iint_{s\ | left(\frac{\pi}{a}\right)^ \frac{N}{2 }mes_ N(A _1\ ca p A_ 2),\ end{aligned}$$ as $ u\rightarrow \infty$,where t h e co n ve rgenc e holds by t hedo mi nat ed co nverg enc e theor em. Indeed , $ \i nt_{\mathbb{ R }^ N}1_{\{z\i n A _1\cap (A_2- x/u )\}}dz $isb ounde d b y $\m ax_{|\ e psilon |<1} mes_ N( A _1\cap (A_2-\e p s il on)) $ uniformly for $ | x| \ leq C\sqrt{\lo g u}$wh e n$ u $ i s l arge enoug h.
Itf ollowsf ro m ( \[E q :pq1\])–(\[Eq :pq1\]) tha t , f or con cl udi n g, itremai ns toverify $$\l abel {D(u) def }
D(u) : =q(u)-p ( u)\righ tarrow 0, \\tex t {a s} \ u \r i ght a rr ow\ inf ty.$$
D ef in e $$\ begi n { a l igne d}\mat hcal{ \hat D}:=\Big \{( s,t) \ inA_1\t imesA_2: \, |t- s|\leq \del ta (u)+\sqrt{N}d_1 (u)+ \sqrt{N}d _2( u) \Bi g\ }.\en d {align ed} $$By thedefinit i onof $ \ ma thcal{C}$ in, we s ee t ha t $
\mat hcal{D } \s ub s eteq \bi gc up_ {(\m a t hbf{k },\m a th bf{l})\i n \mat h ca l{ C}}\Del ta _{\mat hb f{k }}^ {(1)} \ time s \Del ta_{\mat hbf{l } }^{(2)}\subset e q \mathcal{\ h at D }. $ Si nce $d_1(u)=o( \del t a(u) )$ a n d$d_ 2 (u)=o (\del ta ( u) ) $ as $u\rightarrow\i nfty$, theset $\mathcal {\hat D}$i s a subset of$ \m a thcal{\tilde D }:=\{ (s,t)\in A _ 1\timesA_2:\ , |t-s| \leq 2\de l t a(u)\}$whe n $ u$isl a rg e.
Write $D( u ) $ in ( \[D(u)def \]) asa s umove r $ (\ mathbf{k} ,\mathbf {l }) \i n\ma thcal { C}$. Toes tim at e t he ca r dinali ty of $ \ ma th c al{ C}$, we no t i ce t ha t$$\b egi n{ align ed}\ lab el{uppe r bound o f m e s(ti ld eD)}
me s_{2N}(\mathc al {\tilde D} )& = \ iint_{ s \ | left(\frac{\pi}{a}\right)^\frac{N}{2}mes_N(A_1\cap A_2),\end{aligned}$$_as $u\rightarrow_\infty$, where the convergence_holds by_the_dominated convergence_theorem._Indeed, $\int_{\mathbb{R}^N}1_{\{z\in A_1\cap_(A_2-x/u)\}}dz$ is bounded_by $\max_{|\epsilon|<1} mes_N(A_1\cap (A_2-\epsilon))$_uniformly for $|x|\leq_C\sqrt{\log_u}$ when $u$ is large enough.
It follows from (\[Eq:pq1\])–(\[Eq:pq1\]) that, for concluding, it remains_to_verify $$\label{D(u)_def}
D(u):=q(u)-p(u)\rightarrow_0,_\ \text{as}\ u\rightarrow \infty.$$
Define $$\begin{aligned}
\mathcal{\hat_D}:=\Big\{(s,t)\in A_1\times A_2:\, |t-s|\leq_\delta(u)+\sqrt{N}d_1(u)+\sqrt{N}d_2(u)\Big\}.\end{aligned}$$ By_the definition of $\mathcal{C}$ in, we see that_$
\mathcal{D}\subseteq_\bigcup_{(\mathbf{k},\mathbf{l})\in \mathcal{C}}\Delta_{\mathbf{k}}^{(1)}\times \Delta_{\mathbf{l}}^{(2)}\subseteq_ \mathcal{\hat D}.
$ Since $d_1(u)=o(\delta(u))$ and $d_2(u)=o(\delta(u))$ as $u\rightarrow_\infty$, the set $\mathcal{\hat D}$ is_a subset of_$\mathcal{\tilde_D}:=\{(s,t)\in_A_1\times A_2:\, |t-s|\leq_2\delta(u)\}$ when $u$ is large.
Write $D(u)$_in (\[D(u) def\]) as a sum_over $(\mathbf{k},\mathbf{l})\in \mathcal{C}$. To estimate the cardinality_of $ \mathcal{C}$, we notice that_$$\begin{aligned}
\label{upper bound of mes(tilde D)}
_mes_{2N}(\mathcal{\tilde D})&=_\iint_{s\ |
indicates its photospheric origin since there is no obvious mid-infrared flux excess up to 8 $\mu$m. In contrast, the SED of J0241$-$03 appears overluminous longwards of the $H$-band when compared to the average SED of L0–L1 dwarfs. G196–3B (L3) has a similar property as it is shown in Fig. 3 of Zapatero Osorio et al$.$ [@osorio10]. The polarimetric observations may shed new light on the understanding of the origin of this feature.
Linear polarimetric observations
--------------------------------
Linear polarimetric images were acquired for our sample targets with the Long-slit Intermediate Resolution Infrared Spectrograph (LIRIS, Manchado et al$.$ [@manchado04]) mounted at the Cassegrain focus of the William Herschel Telescope (WHT) during the nights 2004 Oct 27, 2006 Mar 22, 2010 Oct 25, and 2011 Apr 20. The 2004 campaign was carried out as part of the LIRIS instrument commissioning on the WHT. LIRIS is equipped with a 1024$\times$1024 pix$^2$ Hawaii detector covering the wavelength range 0.8–2.5 $\mu$m. The pixel projection onto the sky is 025 yielding a field of view of 427$\times$427. Polarimetric observations can be performed by using a Wedged double Wollaston device (Oliva [@oliva97]), consisting in a combination of two Wollaston prisms that deliver four simultaneous images of the polarized flux at angles 0 and 90, 45 and 135 deg. An aperture mask 4$\times$1 in size is used to prevent overlapping effects between the different polarization vector images.
Our data were acquired using the $J$- and $H$-band filters and (in some occasions) two different angles of the rotator of the telescope. The latter is convenient to derive more accurate polarimetric measurements (Alves et al$.$ [@alves11]), since flat-field effects and optical path differences are reduced. The central wavelengths and passbands of the filters are 1.25 and 0.16 $\mu$m ($J$), and 1.64 and 0.29 $\mu$m ($H$). All program objects (except J2244$+$20) were observed in the $J$-band; one target (USco128) was observed in the two $JH$-bands, and J | indicates its photospheric origin since there is no obvious mid - infrared flux surfeit up to 8 $ \mu$m. In line, the SED of J0241$-$03 appears overluminous longwards of the $ H$-band when compared to the modal SED of L0 – L1 gnome. G196–3B (L3) has a similar place as it is shown in Fig. 3 of Zapatero Osorio et al$.$ [ @osorio10 ]. The polarimetric observations may shed fresh light on the understanding of the lineage of this feature.
Linear polarimetric observations
--------------------------------
Linear polarimetric picture were acquired for our sample targets with the Long - slit Intermediate Resolution Infrared Spectrograph (LIRIS, Manchado et al$.$ [ @manchado04 ]) mounted at the Cassegrain focus of the William Herschel Telescope (WHT) during the night 2004 Oct 27, 2006 Mar 22, 2010 Oct 25, and 2011 Apr 20. The 2004 campaign was carried out as part of the LIRIS musical instrument commissioning on the WHT. LIRIS is equipped with a 1024$\times$1024 pix$^2 $ Hawaii detector covering the wavelength scope 0.8–2.5 $ \mu$m. The pixel projection onto the sky is 025 yielding a field of view of 427$\times$427. Polarimetric observations can be perform by using a Wedged double Wollaston device (Oliva [ @oliva97 ]), consisting in a combination of two Wollaston prisms that deliver four simultaneous images of the polarized flux at angles 0 and 90, 45 and 135 deg. An aperture mask 4$\times$1 in size is used to prevent overlapping effects between the different polarization vector images.
Our data were get using the $ J$- and $ H$-band filters and (in some occasion) two unlike angles of the rotator of the telescope. The latter is convenient to derive more accurate polarimetric measurements (Alves et al$.$ [ @alves11 ]), since flat - field effects and optical way differences are reduced. The central wavelengths and passbands of the filters are 1.25 and 0.16 $ \mu$m ($ J$), and 1.64 and 0.29 $ \mu$m ($ H$). All program objects (except J2244$+$20) were observed in the $ J$-band; one target (USco128) was observed in the two $ JH$-bands, and J | inficates its photospheric origin since tkwre is no obbious mia-infrared flux excess up to 8 $\my$m. In contrast, the SED of J0241$-$03 appears lverlumibous oongwards of the $H$-ncnd wgcn cok'ared to the avgrage SED of L0–L1 dwarfs. G196–3B (N3) fad a similar property as it is shown in Fig. 3 ov Zapatero Osotio eu aj$.$ [@oskgii10]. The polarimetric observatiohs may vhed new lighy on the understanding of hhe lrigin of this feahure.
Linear kklawumetric obsefvations
--------------------------------
Linear polarimgtric images were acquired for ojr sakple targejs wljh the Long-soit Igtermediate Vvsolutimn Infrsred Spectrogrsph (LIEIS, Manchado et al$.$ [@maichado04]) mounted at thg Cassegrahn focus of the Wiloiqm Hetschen Teuwscupe (WIT) suring thx nights 2004 Odt 27, 2006 Mar 22, 2010 Ict 25, and 2011 Apr 20. The 2004 cwnpaign was cadried jue as part of the LIRIS instrument commivsikning on the WHT. LIRIS us equipped with a 1024$\tiles$1024 pix$^2$ Hwwaii detector covering the wavelength range 0.8–2.5 $\mu$m. The 'ibel pvojezriln onto the sky is 025 yielding a field of view jr 427$\uimvs$427. Polarimetric onservations can be pfrgjrmed by usine a Webfes double Wollaston device (Olivq [@oliva97]), cjnsixting in a combination of tqo Wollaston prisms that delivex four simulcaneoux imabes of the polarized flbx at zngles 0 and 90, 45 and 135 dse. An aperture mark 4$\nimev$1 in size is used to prevene overlap'ing zffects cetwgen the different polavhzation vector imahes.
Out data were acqulred using the $J$- and $H$-band filtxcs and (in somg owcavions) twj difnerent angles os the rotator pf the celescupe. The lanter is cmnvenient tj derive more wccurate polerimetric meawurenents (Auxes et al$.$ [@alvex11]), since flat-field edfects and optical pajh differences arz twduced. The cenyrau wwvvleigths dnd passbandv of thd filtdrs are 1.25 anb 0.16 $\ou$m ($K$), and 1.64 and 0.29 $\mu$m ($H$). All profram objects (excepy M2244$+$20) were obwerved ig the $J$-band; ome target (USco128) was obsecved ii the ywo $JH$-bands, and J | indicates its photospheric origin since there is mid-infrared excess up 8 $\mu$m. In appears longwards of the when compared to average SED of L0–L1 dwarfs. G196–3B has a similar property as it is shown in Fig. 3 of Zapatero et al$.$ [@osorio10]. The polarimetric observations may shed new light on the understanding the of feature. polarimetric observations -------------------------------- Linear polarimetric images were acquired for our sample targets with the Long-slit Intermediate Infrared Spectrograph (LIRIS, Manchado et al$.$ [@manchado04]) mounted the Cassegrain focus of William Herschel Telescope (WHT) during nights Oct 27, Mar 2010 25, and 2011 20. The 2004 campaign was carried out as part of the LIRIS instrument commissioning on the WHT. is equipped 1024$\times$1024 pix$^2$ detector the range 0.8–2.5 $\mu$m. projection onto the sky is 025 of view of 427$\times$427. Polarimetric observations can be by using Wedged double Wollaston device (Oliva [@oliva97]), in a combination of two Wollaston prisms that four simultaneous images of the polarized flux at angles 0 and 90, 45 and 135 aperture mask 4$\times$1 in is used to overlapping between different vector images. data were acquired using the $J$- and $H$-band filters and (in occasions) two different angles of the rotator of the telescope. is to derive more polarimetric measurements (Alves et [@alves11]), flat-field effects and optical are The passbands the are 1.25 and 0.16 ($J$), and 1.64 and 0.29 ($H$). All program objects the $J$-band; one target (USco128) was observed in two $JH$-bands, and J | indicates its photospheric oRigin since There Is nO obViOus mId-inFrared flux exceSS up tO 8 $\mu$m. In contrast, the SED of j0241$-$03 appeArS OverLUmInous LongwarDS oF THe $H$-BaNd WheN cOMpAred tO thE averagE SED of L0–L1 dwArfS. G196–3b (L3) has a similaR PrOperty as it Is sHown in Fig. 3 of ZApaTero OsOrIo eT Al$.$ [@osoRio10]. the poLarimeTRic obsErvations MaY Shed neW Light on THE uNderStanding of the origIN oF This feature.
LinEar polArIMeTRIc oBseRvations
--------------------------------
LiNeAr polARimetriC ImAGES weRE acquired for oUr sample tarGEts With thE LOng-SLit IntErmedIaTE ReSolution InfRareD SpectrogRaph (LIriS, ManchADo et al$.$ [@mAnchadO04]) moUntEd at THe caSseGrAIn fOCuS of THe WIlliam HeRsChEl TelEscoPE (wht) durIng The nIghts 2004 oct 27, 2006 Mar 22, 2010 Oct 25, and 2011 APr 20. THe 2004 caMPaiGn was CarriEd ouT aS part Of the LiRIS iNsTrument commissiOninG on the WHT. lIRiS Is eQuIpped WIth a 1024$\tiMes$1024 Pix$^2$ hawaii dEtector COveRiNG THe Wavelength range 0.8–2.5 $\mu$m. thE PIxEl projecTion onTO tHe SKy is 025 yielDiNg a FielD OF view Of 427$\tiMEs$427. polarimeTric obSErVaTions caN bE perfoRmEd bY usIng a WEDged Double wollastoN deviCE (Oliva [@oliva97]), conSIsting in a combINaTIOn OF two wolLaston prismS thaT DeliVer fOUr SimULtaneOus imAgES oF The polarized flux at aNgLes 0 and 90, 45 And 135 deG. An aperture maSk 4$\times$1 in sIZE Is used to PrevENt OVerlapping effeCts beTween the diFFerent poLarizAtion vecTor images.
oUR data werE acQuiRed UsiNG ThE $J$- and $H$-band filTERs anD (iN some ocCasIons) two DifFerEnt AngLeS of the rotAtor of thE tElEsCoPe. THe latTEr is convEnIenT tO deRive mORe accuRate pOlarImEtRIc mEasuremENtS (aLves Et Al$.$ [@AlveS11]), siNcE flat-FielD EffEcts and Optical paTh dIFferEnCeS are redUced. The centraL wAvelengths AnD paSsbandS OF the filtErs are 1.25 and 0.16 $\mu$m ($J$), and 1.64 and 0.29 $\mu$m ($H$). aLl progrAm oBjectS (excEpt J2244$+$20) were oBseRved in The $j$-Band; onE targeT (USco128) WaS obSERved iN THe Two $jH$-Bands, and J | indicates its photospheri c origin s incethe reis noobvi ous mid-infrar e d fl ux excess up to 8 $\mu $m. I nc ontr a st , the SED of J0 2 4 1$- $0 3app ea r soverl umi nous lo ngwards of th e$H$-band whe n c ompared to th e average SE D o f L0–L 1dwa r fs. G 196 –3B ( L3) ha s a sim ilar prop er t y as i t is sho w n i n Fi g. 3 of ZapateroO so r io et al$.$ [@ osorio 10 ] .T h e p ola rimetric o bs ervat i ons may sh e d new light on theunderstandi n g o f theor igi n of th is fe at u re.
Linear po lari metric ob servat i ons
--- - ------- ------ --- --- ---- - -- --
L in e arp ol ari m etr ic image swe re ac quir e d f or o ursamp le ta rgets with th e L ong- s lit Inte rmedi ateRe solut ion In frare dSpectrograph (L IRIS , Manchad o e tal$ .$ [@ma n chado0 4]) mo unted a t the C a sse gr a i n f ocus of the Willia mH e rs chel Tel escope (W HT ) duringth e n ight s 2004Oct2 7, 2006 Ma r 22,2 01 0Oct 25, a nd 201 1Apr 20 . The 2004 campa ign wascarri e d out as parto f the LIRIS i n st r u me n t co mmi ssioning on the WHT. LIR I Sise quipp ed wi th a1 024$\times$1024 pix $^ 2$ Haw aii d etector cover ing the wa v e l ength ra nge0 .8 – 2.5 $\mu$m. Th e pix el project i on ontothe s ky is 02 5 yieldin g a fieldofvie w o f 4 2 7 $\ times$427. Po l a rime tr ic obse rva tions c anbeper for me d by usin g a Wedg ed d ou bl e W ollas t on devic e(Ol iv a [ @oliv a 97]),consi stin gin a c ombinat i on o f tw oWo llas ton p risms tha t de liver f our simul tan e ousim ag es of t he polarizedfl ux at angl es 0and 90 , 45 and 1 35 deg. An aperture mas k 4$\tim es$ 1 insize is usedtopreven t o v erlapp ing ef fects b etw e e n the d if fer en t polariza t i onvecto rimag es.
Ou r data were acquir e d u sing the $J$- an d $H $ - ba ndf il t ers a n d ( i n some occasions ) two diff er e nt angles of the r otatorof theteles c ope. Th e latteris conven ie nt t o der ive more a ccuratepolarimet r ic me a su remen ts(Alves e t a l$.$[@alve s 11] ), si nce fl at -field effe ct s and op tical path differencesare re duced . T he centra l w a vel engths an d pa ssbands of th e f ilter s a r e 1.2 5 an d 0 .16 $\mu$ m ($ J $), and 1 . 64 an d 0. 29 $\mu$m ( $ H $ ).All p rog r am obj ects (except J2244$+$ 2 0) were observ ed i n the $J $ -ban d; one target (U Sco 12 8 ) was obs er ved in thetwo $JH$ -b a nds,and J | indicates_its photospheric_origin since there is_no obvious_mid-infrared_flux excess_up_to 8 $\mu$m._In contrast, the_SED of J0241$-$03 appears_overluminous longwards of_the_$H$-band when compared to the average SED of L0–L1 dwarfs. G196–3B (L3) has a_similar_property as_it_is_shown in Fig. 3 of Zapatero_Osorio et al$.$ [@osorio10]. The_polarimetric observations_may shed new light on the understanding of_the_origin of this_feature.
Linear polarimetric observations
--------------------------------
Linear polarimetric images were acquired for our_sample targets with the Long-slit Intermediate_Resolution Infrared Spectrograph_(LIRIS,_Manchado_et al$.$ [@manchado04]) mounted_at the Cassegrain focus of the_William Herschel Telescope (WHT) during the_nights 2004 Oct 27, 2006 Mar 22,_2010 Oct 25, and 2011 Apr_20. The 2004 campaign was_carried out_as part of the LIRIS_instrument commissioning on_the WHT._LIRIS is equipped_with a 1024$\times$1024 pix$^2$ Hawaii detector_covering the wavelength_range 0.8–2.5 $\mu$m. The pixel projection_onto_the sky is_025_yielding_a field_of view of_427$\times$427._Polarimetric observations_can_be performed by using a Wedged_double_Wollaston device (Oliva [@oliva97]), consisting in a_combination of two Wollaston_prisms_that deliver four simultaneous_images of the polarized flux_at angles 0 and 90, 45_and 135_deg. An_aperture mask 4$\times$1 in size is used to prevent overlapping effects between_the different polarization vector images.
Our data_were acquired using the_$J$- and_$H$-band_filters and (in_some_occasions) two_different angles of the rotator of the_telescope. The_latter is convenient to derive more_accurate polarimetric measurements (Alves_et_al$.$ [@alves11]), since flat-field effects and_optical path differences are reduced. The_central wavelengths and passbands of_the_filters_are 1.25 and 0.16 $\mu$m ($J$),_and 1.64 and 0.29 $\mu$m ($H$). All_program objects (except_J2244$+$20) were observed in the $J$-band; one_target_(USco128) was observed in the two_$JH$-bands,_and J |
to the average solar wind dynamical pressure. A reason for this discrepancy might be a variation of the polytropic index with the solar cycle since our observations are performed during typical solar minimum conditions. The position of the Earth’s bow shock still seems to be a matter of discussion.
A superposed epoch analysis has been carried out by averaging particle and magnetic field data as well as low frequency magnetic spectra upstream and downstream of the bow shock. We have performed this analysis by dividing the events into different categories, i.e., quasi-perpendicular and quasi-parallel events as well as quasi-perpendicular low-$\beta$ and high-$\beta$ events.
The particle and magnetic field data show that upstream of the quasi-parallel bow shock, in the foreshock region, the plasma is already heated compared to the undisturbed solar wind. Moreover, there are more energetic protons in the foreshock region, and the magnetic field is highly variable. Downstream of the quasi-perpendicular bow shock, a proton temperature anisotropy is found, which is higher on average downstream of the quasi-perpendicular low-$\beta$ than downstream of the quasi-perpendicular high-$\beta$ bow shock.
Concerning the low frequency magnetic fluctuations we find that upstream of the quasi-perpendicular bow shock the solar wind spectrum is undisturbed with transverse Alfvén waves surpassing the compressive spectral component. Upstream of the quasi-parallel bow shock largely enhanced wave activity is detected in the turbulent foreshock region. These upstream waves are convected downstream, experiencing an enhancement at the bow shock itself. Downstream of the quasi-perpendicular bow shock the observed proton temperature anisotropy leads to the generation of left-hand polarized ion cyclotron waves under low-$\beta$ conditions and in some cases to the generation of mirror waves under high-$\beta$ conditions. A clear correlation has been observed between the intensity of the left-hand polarized component of the magnetic power spectrum relative to the right-hand polarized component and the proton temperature anisotropy. On the other hand, we could not find a simple correlation between the intensity of the compressive component and any single plasma or magnetic field parameter. In cases where mirror waves are obviously observable mostly three conditions are fulfilled: the plasma-$\beta$ is high, the mirror instability criterion is satisfied and the angle $\theta_{Bn}$ is large, i.e., $\theta_{Bn} \gtrsim 80^\circ$. But there are also cases where | to the average solar wind dynamical pressure. A rationality for this discrepancy might be a mutant of the polytropic index with the solar cycle since our observations are perform during typical solar minimum circumstance. The position of the Earth ’s bow shock however seems to be a matter of discussion.
A superimpose epoch analysis has been carried out by average atom and magnetic field datum as well as humble frequency magnetic spectra upstream and downstream of the bow shock. We have performed this analysis by separate the events into different categories, i.e., quasi - perpendicular and quasi - parallel event as well as quasi - perpendicular low-$\beta$ and high-$\beta$ events.
The particle and magnetic discipline data show that upstream of the quasi - parallel bow shock, in the foreshock region, the plasma is already heated compare to the undisturbed solar wind. Moreover, there are more energetic protons in the foreshock region, and the magnetic field is highly variable. Downstream of the quasi - perpendicular bow shock, a proton temperature anisotropy is determine, which is higher on average downstream of the quasi - perpendicular low-$\beta$ than downstream of the quasi - vertical high-$\beta$ bow shock absorber.
Concerning the low frequency charismatic fluctuations we rule that upstream of the quasi - perpendicular bow shock the solar fart spectrum is undisturbed with transverse Alfvén waves surpassing the compressive spectral component. Upstream of the quasi - parallel bow shock largely enhance wave activity is detected in the turbulent foreshock region. These upstream wave are convect downstream, experiencing an enhancement at the bow shock itself. Downstream of the quasi - perpendicular bow shock the observed proton temperature anisotropy leads to the generation of left - bridge player polarized ion cyclotron waves under low-$\beta$ conditions and in some cases to the genesis of mirror waves under high-$\beta$ condition. A clear correlation has been observe between the intensity of the leftover - hand polarized part of the magnetic power spectrum relative to the right - hand polarize component and the proton temperature anisotropy. On the early hand, we could not find a simple correlation between the saturation of the compressive part and any single plasma or magnetic plain parameter. In cases where mirror waves are obviously observable mostly three conditions are fulfilled: the plasma-$\beta$ is gamey, the mirror instability criterion is satisfied and the angle $ \theta_{Bn}$ is large, i.e., $ \theta_{Bn } \gtrsim 80^\circ$. But there are besides cases where | to the average solar wind aynamical pressorw. A reeson fod this dkscrepancy might be a variatmon if tht polytropic index dith the dolar cyxle wunce our ouaervations ars peryocmed during typlcal solar kinimum condithovs. The position of the Earth’s bow shosk stilk deems to be a iatttr jf djscussion.
A superposed epoch analysjs has ueen carried ouy by averaging particle anf mahnetic field data ws well as oow seequency magvetic specurc upstream znd downstream of the bow shock. We hcve performgb thlv analysis uy divpding the evekns into differrnt categories, i.e., quqsi-perpendicular and xuasi-parallel evenjs as well aa quasi-perpendicuoae low-$\teta$ and yigf-$\beua$ xvehts.
The pacticle and jagnetic fiwld data show that ipfnteam of the suasi-pwrwllel bow shock, in the foreshock region, ths plasma is already heared compared to the ujdisturbeq solar wind. Moreover, there are more energetic prmtons kn uhc wirfshock region, and the magnetic field is highlr vsrpable. Downstream jf the quaso-pfrlgndicular bow rhock, c pdoton temperature wnisotrjpy iw found, wrich is higher on average downsrream of the wuasi-perpendicular low-$\beta$ thcn dowmstresm of the quasi-perpendieular gigh-$\beta$ boa shock.
Cohzerning the low wreauancy magnetic fluctuations we find vhat bpstream of jhe quafi-perpendifular bow shock the solag winb spewtrum is ujdisturbed with transverse Alfvéi waves surpaxshng the com'ressiye spectral com[onent. Upstreak of thz quask-parallel gow shork largely eghanced wave dftivity is dxtected ig thw tuebulent woreshock regipn. These lpftreqm waves are convegted akwnstream, experntncung an enhancemrnt at tje bjf shock itsenf. Duwnryream of the quawk-perlendicular bow shock the observed proton trmierature qnisotro[y leads to tne generation of ltft-hanv polacized oon cyclotron waves under low-$\beta$ conditiojs wnd in some sasew to the genzration of mirror waves under high-$\beta$ cinditions. A clear xorrelation has beeu pbserved bevween ehe intenvity of the left-hand polarized compontnt of the magnetic powsr spewtrum relative to the right-hand polarized component and the proton temperature anisovrjpy. On the kthet hang, ce eould njt fmnv a simple correlstion between the intensity of vhe comprevsnve component and any single llxsma or magnejic field parameter. In casss where mirror waves are obviously obsrrvable mostly three concitions qre fhlfillvs: vhe plasma-$\beta$ is hibh, the mirrir instqbillty criterion ks xabisfied anf tht anyle $\theta_{Bn}$ is large, i.e., $\theta_{Bn} \grrsim 80^\xigc$. But there are also cases dhere | to the average solar wind dynamical pressure. for discrepancy might a variation of solar since our observations performed during typical minimum conditions. The position of the bow shock still seems to be a matter of discussion. A superposed epoch has been carried out by averaging particle and magnetic field data as well low magnetic upstream downstream of the bow shock. We have performed this analysis by dividing the events into different i.e., quasi-perpendicular and quasi-parallel events as well as low-$\beta$ and high-$\beta$ events. particle and magnetic field data that of the bow in foreshock region, the is already heated compared to the undisturbed solar wind. Moreover, there are more energetic protons in the region, and field is variable. of quasi-perpendicular bow shock, temperature anisotropy is found, which is downstream of the quasi-perpendicular low-$\beta$ than downstream of quasi-perpendicular high-$\beta$ shock. Concerning the low frequency magnetic we find that upstream of the quasi-perpendicular bow the solar wind spectrum is undisturbed with transverse Alfvén waves surpassing the compressive spectral component. the quasi-parallel bow shock enhanced wave activity detected the foreshock These upstream are convected downstream, experiencing an enhancement at the bow shock itself. of the quasi-perpendicular bow shock the observed proton temperature anisotropy the of left-hand polarized cyclotron waves under low-$\beta$ and some cases to the mirror under clear has observed between the intensity the left-hand polarized component of magnetic power spectrum relative and the proton temperature anisotropy. On the other we could not find a simple correlation the intensity of the compressive component and any single plasma or magnetic parameter. In mirror waves are obviously observable mostly three conditions fulfilled: the plasma-$\beta$ is the mirror instability criterion is satisfied and the angle is i.e., $\theta_{Bn} 80^\circ$. But there also cases where | to the average solar wind dynaMical pressUre. A rEasOn fOr This DiscRepancy might be A VariAtion of the polytropic inDex wiTh THe soLAr Cycle Since ouR ObSERvaTiOnS arE pERfOrmed DurIng typiCal solar miNimUm Conditions. ThE PoSition of thE EaRth’s bow shock StiLl seemS tO be A MatteR of DiscuSsion.
A SUperpoSed epoch aNaLYsis haS Been carRIEd Out bY averaging particlE AnD Magnetic field dAta as wElL As LOW frEquEncy magnetIc SpectRA upstreAM aND DOwnSTream of the bow Shock. We have PErfOrmed tHiS anALysis bY diviDiNG thE events into DiffErent cateGories, I.E., quasi-pERpendicUlar anD quAsi-ParaLLeL eVenTs AS weLL aS quASi-pErpendicUlAr Low-$\beTa$ anD HIGH-$\betA$ evEnts.
the paRticle and magnEtiC fieLD daTa shoW that UpstReAm of tHe quasI-paraLlEl bow shock, in the ForeShock regiOn, tHe PlaSmA is alREady heAteD coMpared tO the undIStuRbED SOlAr wind. Moreover, therE aRE MoRe energeTic proTOnS iN The foresHoCk rEgioN, ANd the MagnETiC field is Highly VArIaBle. DownStReam of ThE quAsi-PerpeNDicuLar bow Shock, a prOton tEMperature anisoTRopy is found, whICh IS HiGHer oN avErage downstReam OF the QuasI-PeRpeNDiculAr low-$\BeTA$ tHAn downstream of the quAsI-perpeNdicuLar high-$\beta$ boW shock.
ConcERNIng the loW freQUeNCy magnetic flucTuatiOns we find tHAt upstreAm of tHe quasi-pErpendicuLAR bow shocK thE soLar WinD SPeCtrum is undistURBed wItH transvErsE Alfvén WavEs sUrpAssInG the comprEssive spEcTrAl CoMpoNent. UPStream of ThE quAsI-paRalleL Bow shoCk larGely EnHaNCed Wave actIViTY Is deTeCtEd in The TuRbuleNt foREshOck regiOn. These upStrEAm waVeS aRe conveCted downstreaM, eXperiencinG aN enHancemENT at the boW shock itself. Downstream oF The quasI-peRpendIculAr bow shocK thE obserVed PRoton tEmperaTure aNiSotROPy leaDS To The GeNeration of LEFt-hAnd poLaRizeD ion cycLotron waves under loW-$\BetA$ conditions anD in Some CASeS to THe GEneRaTIon OF Mirror waves undeR high-$\beta$ cOnDItIons. A clear COrrElAtion haS been obServeD Between The intensIty of the lEfT-hanD POlaRized compoNent of thE magnetic POwer sPEcTrum rElaTive to ThE riGht-haNd polaRIzeD compOnent aNd The proTon teMpErature aNisotropy. On the other hand, We coulD not fInd A simple coRreLAtiOn between The iNtensity of The ComPressIve COmponEnt aND aNy sINgle pLasmA Or magnetiC FiEld PARaMeter. In caseS WHEre MirroR waVEs are oBvioUsly observable mosTLy three conditiOns aRE FulFilLEd: thE pLasma-$\beta$ is higH, thE mIRRor instaBiLity criteriOn is satiSfIEd and The angLe $\thetA_{Bn}$ is laRGE, i.E., $\Theta_{BN} \gtrSim 80^\Circ$. But thEre ArE Also casEs WhERe | to the average solar wind dynamical pres sur e.Areas on f or this discre p ancy might be a variationof th ep olyt r op ic in dex wit h t h e so la rcyc le si nce o urobserva tions areper fo rmed duringt yp ical solar mi nimum condit ion s. The p osi t ion o f t he Ea rth’sb ow sho ck stillse e ms tob e a mat t e rof d iscussion.
A sup e rp o sed epoch anal ysis h as be e n ca rri ed out byav eragi n g parti c le a n d m a gnetic fielddata as wel l as low f re que n cy mag netic s p ect ra upstream and downstre am oft he bows hock. W e have pe rfo rmed th is an al y sis by di v idi ng the e ve nt s int o di f f e r entcat egor ies,i.e., quasi-p erp endi c ula r and quas i-pa ra llelevents as w el l as quasi-perp endi cular low -$\ be ta$ a nd hi g h-$\be ta$ ev ents.
The par t icl ea n d m agnetic field data s h o wthat ups treamo fth e quasi-p ar all el b o w shoc k, i n t he fores hock r e gi on , the p la sma is a lre ady heat e d co mpared to theundis t urbed solar wi n d. Moreover,t he r e a r e mo reenergetic p roto n s in the fo res h ock r egion ,a nd the magnetic fieldis highl y var iable. Downst ream of th e q uasi-per pend i cu l ar bow shock,a pro ton temper a ture ani sotro py is fo und, whic h is highe r o n a ver age d ow nstream of th e quas i- perpend icu lar low -$\ bet a$tha ndownstrea m of the q ua si -p erp endic u lar high -$ \be ta $ b ow sh o ck.
C oncer ning t he low freque n cy m agne ti cfluc tua ti ons w e fi n d t hat ups tream ofthe quas i- pe rpendic ular bow shoc kthe solarwi ndspectr u m is undi sturbed with transverse Alfvénwav es su rpas sing thecom pressi ves pectra l comp onent .Ups t r eam o f th e q ua si-paralle l bow shoc klarg ely enh anced wave activit y is detected inthe tur b u le ntf or e sho ck reg i o n. These upstre am waves a re co nvected do w nst re am, exp erienci ng an enhance ment at t he bow sh oc k it s e lf. Downstrea m of the quasi-pe r pendi c ul ar bo w s hock t he ob serve d prot o n t emper aturean isotro py le ad s to the generation of left-han d pola rized io n cyclotr onw ave s under l ow-$ \beta$ con dit ion s and in somecase s t o t h e gen erat i on of mir r or wa v e sunder high- $ \ b eta $ con dit i ons. A cle ar correlation ha s been observed bet w e enthe inte ns ity of the lef t-h an d polarize dcomponent o f the ma gn e tic p ower s pectru m relat i v et o therigh t-h and polar ize dc omponen tan d the p roto ntemper aturea niso t r opy. On the othe r han d , we c o uld notfi nd a si m plecorrelatio n between t he int ensi ty of the co mp ressiv e c om ponent and any singl e pla sma orma gnet icfieldpara m e ter.In c as eswhere mir r o rw av es are obv iousl yobse rvable mo s tly thre e c o ndition sare f ulfill e d: t he plasma- $\b eta$i s high, the mirr o ri nstab ilitycriter ion iss ati sf ied and th e angle $\t heta_{Bn} $ is l arge , i.e.,$\ thet a_ {Bn }\gtrsim 80^\cir c$. Butthere ar e also c ase s whe r e | to_the average_solar wind dynamical pressure._A reason_for_this discrepancy_might_be a variation_of the polytropic_index with the solar_cycle since our_observations_are performed during typical solar minimum conditions. The position of the Earth’s bow shock_still_seems to_be_a_matter of discussion.
A superposed epoch_analysis has been carried out_by averaging_particle and magnetic field data as well as_low_frequency magnetic spectra_upstream and downstream of the bow shock. We have_performed this analysis by dividing the_events into different_categories,_i.e.,_quasi-perpendicular and quasi-parallel events_as well as quasi-perpendicular low-$\beta$ and_high-$\beta$ events.
The particle and magnetic field_data show that upstream of the quasi-parallel_bow shock, in the foreshock region,_the plasma is already heated_compared to_the undisturbed solar wind. Moreover,_there are more_energetic protons_in the foreshock_region, and the magnetic field is_highly variable. Downstream_of the quasi-perpendicular bow shock, a_proton_temperature anisotropy is_found,_which_is higher_on average downstream_of_the quasi-perpendicular_low-$\beta$_than downstream of the quasi-perpendicular high-$\beta$_bow_shock.
Concerning the low frequency magnetic fluctuations we_find that upstream of_the_quasi-perpendicular bow shock the_solar wind spectrum is undisturbed_with transverse Alfvén waves surpassing the_compressive spectral_component. Upstream_of the quasi-parallel bow shock largely enhanced wave activity is detected_in the turbulent foreshock region. These_upstream waves are convected_downstream, experiencing_an_enhancement at the_bow_shock itself._Downstream of the quasi-perpendicular bow shock the_observed proton_temperature anisotropy leads to the generation_of left-hand polarized ion_cyclotron_waves under low-$\beta$ conditions and in_some cases to the generation of_mirror waves under high-$\beta$ conditions._A_clear_correlation has been observed between_the intensity of the left-hand polarized_component of the_magnetic power spectrum relative to the right-hand_polarized_component and the proton temperature anisotropy._On_the other hand, we could not_find_a_simple correlation between the intensity_of the compressive component and any_single plasma or magnetic field parameter. In cases where_mirror waves are_obviously observable mostly three conditions_are_fulfilled:_the plasma-$\beta$ is high, the mirror instability criterion is satisfied_and the_angle $\theta_{Bn}$ is_large, i.e., $\theta_{Bn} \gtrsim 80^\circ$. But there are also cases_where |
structures on the differential nilmanifold underlying Iwasawa {#sec:iwasawa}
=================================================================================
The [*Iwasawa manifold*]{} is the complex nilmanifold defined by $$\mathbb{I}_3 \;:=\; \left. \mathbb{H}(3;{\mathbb{Z}}[\operatorname{i}]) \middle\backslash \mathbb{H}(3;{\mathbb{C}}) \right.$$ where $\mathbb{H}(3;{\mathbb{C}})$ is the $3$-dimensional *Heisenberg group* over $\mathbb{C}$, that is, $$\mathbb{H}(3;{\mathbb{C}}) \;:=\; \left\{
\left(
\begin{array}{ccc}
1 & z^1 & z^3 \\
0 & 1 & z^2 \\
0 & 0 & 1
\end{array}
\right) \in \mathrm{GL}(3;\mathbb{C}) {\;:\;}z^1,\,z^2,\,z^3 \in{\mathbb{C}}\right\}
\;,$$ and $\mathbb{H}(3;{\mathbb{Z}}[\operatorname{i}]) := \mathbb{H}(3;{\mathbb{C}}) \cap \mathrm{GL}(3;{\mathbb{Z}}[\operatorname{i}])$. It is worth to remark that it constitutes one of the simplest examples of non-Kähler complex manifold (see, e.g., [@fernandez-gray; @nakamura]).
In our case, we are interested in its underlying real nilmanifold that we will denote by $M=\Gamma\backslash G$. Following [@salamon], let $\mathfrak{g}=(0,0,0,0,13+42,14+23)$ be the real nilpotent Lie algebra naturally associated to $G$ (i.e., the differentiable Lie group underlying $\mathbb{H}(3;{\mathbb{C}})$). This notation means that $\mathfrak{g}^*$ admits a basis $\{e^k\}_{k=1}^6$ satisfying $$\left\{ \begin{array}{l}
\operatorname{d}e^1 \;=\; \operatorname{d}e^2 \;=\; \operatorname{d}e^3 \;=\; \operatorname{d}e^4 \;=\; 0 \\[5pt]
\operatorname{d}e^5 \;=\; e^{13} - e^{24} \\[5 | structures on the differential nilmanifold underlying Iwasawa { # sec: iwasawa }
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
The [ * Iwasawa manifold * ] { } is the complex nilmanifold defined by $ $ \mathbb{I}_3 \;:=\; \left. \mathbb{H}(3;{\mathbb{Z}}[\operatorname{i } ]) \middle\backslash \mathbb{H}(3;{\mathbb{C } }) \right.$$ where $ \mathbb{H}(3;{\mathbb{C}})$ is the $ 3$-dimensional * Heisenberg group * over $ \mathbb{C}$, that is, $ $ \mathbb{H}(3;{\mathbb{C } }) \;:=\; \left\ {
\left (
\begin{array}{ccc }
1 & z^1 & z^3 \\
0 & 1 & z^2 \\
0 & 0 & 1
\end{array }
\right) \in \mathrm{GL}(3;\mathbb{C }) { \;:\;}z^1,\,z^2,\,z^3 \in{\mathbb{C}}\right\ }
\;,$$ and $ \mathbb{H}(3;{\mathbb{Z}}[\operatorname{i } ]): = \mathbb{H}(3;{\mathbb{C } }) \cap \mathrm{GL}(3;{\mathbb{Z}}[\operatorname{i}])$. It is worth to note that it form one of the simplest examples of non - Kähler complex manifold (witness, e.g., [ @fernandez - gray; @nakamura ]).
In our character, we are concerned in its underlying real nilmanifold that we will denote by $ M=\Gamma\backslash G$. Following [ @salamon ], let $ \mathfrak{g}=(0,0,0,0,13 + 42,14 + 23)$ embody the real nilpotent Lie algebra naturally associated to $ G$ (i.e., the differentiable Lie group underlie $ \mathbb{H}(3;{\mathbb{C}})$). This note means that $ \mathfrak{g}^*$ admits a footing $ \{e^k\}_{k=1}^6 $ satisfying $ $ \left\ { \begin{array}{l }
\operatorname{d}e^1 \;=\; \operatorname{d}e^2 \;=\; \operatorname{d}e^3 \;=\; \operatorname{d}e^4 \;=\; 0 \\[5pt ]
\operatorname{d}e^5 \;=\; e^{13 } - e^{24 } \\[5 | stguctures on the differenuial nilmanifold oneerlyiig Iwaszwa {#sec:idasawa}
=================================================================================
The [*Iwasawa manifold*]{} id rhe cimplex nilmanifold defkned by $$\mwthbb{I}_3 \;:=\; \oeft. \nathbb{H}(3;{\matigb{Z}}[\operatornajc{i}]) \mibdoe\backslash \majhbb{H}(3;{\mathbb{C}}) \right.$$ where $\mdtfby{H}(3;{\mathbb{C}})$ is the $3$-dimensional *Heisenbqrg groip* over $\mathbb{C}$, jhat ps, $$\matgbb{H}(3;{\mathbb{C}}) \;:=\; \left\{
\left(
\begin{array}{ccc}
1 & z^1 & z^3 \\
0 & 1 & z^2 \\
0 & 0 & 1
\end{array}
\right) \in \mathgm{GL}(3;\lathbb{C}) {\;:\;}z^1,\,z^2,\,z^3 \in{\mathhb{C}}\right\}
\;,$$ ane $\maeybb{H}(3;{\mathbb{Z}}[\ooeratornamt{i}]) := \mathbb{H}(3;{\mzthbb{C}}) \cap \mathrm{GL}(3;{\mathbb{Z}}[\operagornake{i}])$. It is qoeth jo remark thet it bonstitutes okv of tha simplrst examples on non-Nählwr complex manifold (sxe, e.g., [@fernandez-gray; @gakamura]).
It kur case, we are ibtwrestgd in its ynddrlgiig deal nllmenifold thaf we will dwnote by $M=\Gamma\backxlwwh G$. Followinf [@salaiog], let $\mathfrak{g}=(0,0,0,0,13+42,14+23)$ be the real nilpotent Lpe amgebra naturally associqted to $G$ (i.e., the diffgrentiable Lie group underlying $\mathbb{H}(3;{\mathbb{C}})$). This notatiot meais thcb $\oqtjfrak{g}^*$ admits a basis $\{e^k\}_{k=1}^6$ satisfying $$\left\{ \begyh{atrsy}{l}
\operatornake{f}e^1 \;=\; \operatorname{a}e^2 \;=\; \opzdaforname{d}e^3 \;=\; \operatogname{d}e^4 \;=\; 0 \\[5pt]
\oleratorname{d}e^5 \;=\; e^{13} - e^{24} \\[5 | structures on the differential nilmanifold underlying Iwasawa The manifold*]{} is complex nilmanifold defined \middle\backslash \right.$$ where $\mathbb{H}(3;{\mathbb{C}})$ the $3$-dimensional *Heisenberg over $\mathbb{C}$, that is, $$\mathbb{H}(3;{\mathbb{C}}) \;:=\; \left( \begin{array}{ccc} 1 & z^1 & z^3 \\ 0 & 1 & z^2 0 & 0 & 1 \end{array} \right) \in \mathrm{GL}(3;\mathbb{C}) {\;:\;}z^1,\,z^2,\,z^3 \in{\mathbb{C}}\right\} \;,$$ and := \cap It worth to remark that it constitutes one of the simplest examples of non-Kähler complex manifold (see, [@fernandez-gray; @nakamura]). In our case, we are interested its underlying real nilmanifold we will denote by $M=\Gamma\backslash Following let $\mathfrak{g}=(0,0,0,0,13+42,14+23)$ the nilpotent algebra naturally associated $G$ (i.e., the differentiable Lie group underlying $\mathbb{H}(3;{\mathbb{C}})$). This notation means that $\mathfrak{g}^*$ admits a basis $\{e^k\}_{k=1}^6$ $$\left\{ \begin{array}{l} \operatorname{d}e^2 \;=\; \;=\; \;=\; \\[5pt] \operatorname{d}e^5 \;=\; e^{24} \\[5 | structures on the differentiAl nilmanifOld unDerLyiNg iwasAwa {#sEc:iwasawa}
=================================================================================
The [*IwASawa Manifold*]{} is the complex niLmaniFoLD defINeD by $$\maThbb{I}_3 \;:=\; \leFT. \mATHbb{h}(3;{\mAtHbb{z}}[\oPErAtornAme{I}]) \middle\Backslash \mAthBb{h}(3;{\mathbb{C}}) \righT.$$ WhEre $\mathbb{H}(3;{\MatHbb{C}})$ is the $3$-dimEnsIonal *HEiSenBErg grOup* Over $\mAthbb{C}$, THat is, $$\mAthbb{H}(3;{\matHbB{c}}) \;:=\; \left\{
\lEFt(
\begin{ARRaY}{ccc}
1 & Z^1 & z^3 \\
0 & 1 & z^2 \\
0 & 0 & 1
\end{array}
\right) \iN \MaTHrm{GL}(3;\mathbb{C}) {\;:\;}z^1,\,z^2,\,Z^3 \in{\matHbB{c}}\rIGHt\}
\;,$$ aNd $\mAthbb{H}(3;{\mathBb{z}}[\operATorname{I}]) := \MaTHBB{H}(3;{\mAThbb{C}}) \cap \mathrM{GL}(3;{\mathbb{Z}}[\oPEraTornamE{i}])$. it iS Worth tO remaRk THat It constitutEs onE of the simPlest eXAmples oF Non-KählEr compLex ManIfolD (SeE, e.G., [@feRnANdeZ-GrAy; @nAKamUra]).
In our CaSe, We are InteRESTEd in Its UndeRlyinG real nilmanifOld That WE wiLl denOte by $m=\GamMa\BacksLash G$. FOllowInG [@salamon], let $\mathFrak{G}=(0,0,0,0,13+42,14+23)$ be the reaL niLpOteNt lie alGEbra naTurAllY associAted to $G$ (I.E., thE dIFFErEntiable Lie group unDeRLYiNg $\mathbb{h}(3;{\mathbB{c}})$). THiS Notation MeAns That $\MAThfraK{g}^*$ adMItS a basis $\{e^K\}_{k=1}^6$ satiSFyInG $$\left\{ \beGiN{array}{L}
\oPerAtoRname{D}E^1 \;=\; \opeRatornAme{d}e^2 \;=\; \opeRatorNAme{d}e^3 \;=\; \operatornAMe{d}e^4 \;=\; 0 \\[5pt]
\operatoRNaME{D}e^5 \;=\; E^{13} - E^{24} \\[5 | structures on the differe ntial nilm anifo ldund er lyin g Iw asawa {#sec:iw a sawa }
==================== ===== == = ==== = == ===== ======= = == = = === == == === == = == ===== === ======
The [*Iwa saw amanifold*]{} is the compl exnilmanifolddef ined b y$$\ m athbb {I} _3 \; :=\; \ l eft. \ mathbb{H} (3 ; {\math b b{Z}}[\ o p er ator name{i}]) \middle \ ba c kslash \mathbb {H}(3; {\ m at h b b{C }}) \right.$$ w here$ \mathbb { H} ( 3 ; {\m a thbb{C}})$ is the $3$-di m ens ional*H eis e nberggroup *o ver $\mathbb{C }$,that is,$$\mat h bb{H}(3 ; {\mathb b{C}}) \; :=\ ; \l e ft \{
\l ef t (
\ b eg in{ a rra y}{ccc} 1 & z^1& z^ 3 \ \
0& 1 & z^2\\
0 & 0 & 1
\en d {ar ray}\righ t) \ in \mat hrm{GL }(3;\ ma thbb{C}) {\;:\; }z^1 ,\,z^2,\, z^3 \ in{ \m athbb { C}}\ri ght \}\;,$$ a nd $\ma t hbb {H } ( 3 ;{ \mathbb{Z}}[\opera to r n am e{i}]) : = \mat h bb {H } (3;{\mat hb b{C }})\ c ap \m athr m {G L}(3;{\m athbb{ Z }} [\ operato rn ame{i} ]) $.Itis wo r th t o rema rk thatit co n stitutes one o f the simplest ex a m pl e s of no n-Kähler co mple x man ifol d ( see , e.g. , [@f er n an d ez-gray; @nakamura] ).
In o ur ca se, we are in terested i n i ts under lyin g r e al nilmanifold that we will d e note by$M=\G amma\bac kslash G$ . Followin g [ @sa lam on] , le t $\mathfrak{ g } =(0, 0, 0,0,13+ 42, 14+23)$ be th e r eal n ilpotentLie alge br ana tu ral ly as s ociatedto $G $(i. e., t h e diff erent iabl eLi e gr oup und e rl y i ng $ \m at hbb{ H}( 3; {\mat hbb{ C }}) $). Thi s notatio n m e ansth at $\math frak{g}^*$ ad mi ts a basis $ \{e ^k\}_{ k = 1}^6$ sa tisfying $$\left\{ \beg i n{array }{l }
\ope rat orname {d} e ^1 \;= \; \op erato rn ame { d }e^2\ ; =\ ; \ op eratorname { d }e^ 3 \;= \; \op eratorn ame{d}e^4 \;=\; 0\ \[5 pt]
\ope r a to rna m e{ d }e^ 5\ ;=\ ; e^{13} - e^{24} \\[5 | structures_on the_differential nilmanifold underlying Iwasawa_{#sec:iwasawa}
=================================================================================
The [*Iwasawa_manifold*]{}_is the_complex_nilmanifold defined by_$$\mathbb{I}_3 \;:=\; \left._\mathbb{H}(3;{\mathbb{Z}}[\operatorname{i}]) \middle\backslash \mathbb{H}(3;{\mathbb{C}}) \right.$$_where $\mathbb{H}(3;{\mathbb{C}})$ is_the_$3$-dimensional *Heisenberg group* over $\mathbb{C}$, that is, $$\mathbb{H}(3;{\mathbb{C}}) \;:=\; \left\{
\left(
\begin{array}{ccc}
1 & z^1 &_z^3_\\
0_&__1 & z^2 \\
_0 & 0 _& _1
\end{array}
\right) \in \mathrm{GL}(3;\mathbb{C}) {\;:\;}z^1,\,z^2,\,z^3 \in{\mathbb{C}}\right\}
\;,$$ and $\mathbb{H}(3;{\mathbb{Z}}[\operatorname{i}]) :=_\mathbb{H}(3;{\mathbb{C}})_\cap \mathrm{GL}(3;{\mathbb{Z}}[\operatorname{i}])$. It_is worth to remark that it constitutes one of_the simplest examples of non-Kähler complex_manifold (see, e.g.,_[@fernandez-gray;_@nakamura]).
In_our case, we are_interested in its underlying real nilmanifold_that we will denote by $M=\Gamma\backslash_G$. Following [@salamon], let $\mathfrak{g}=(0,0,0,0,13+42,14+23)$ be the_real nilpotent Lie algebra naturally associated_to $G$ (i.e., the differentiable_Lie group_underlying $\mathbb{H}(3;{\mathbb{C}})$). This notation means_that $\mathfrak{g}^*$ admits_a basis_$\{e^k\}_{k=1}^6$ satisfying $$\left\{_\begin{array}{l}
_ _ \operatorname{d}e^1 \;=\; \operatorname{d}e^2_\;=\;_\operatorname{d}e^3 \;=\; \operatorname{d}e^4_\;=\;_0_\\[5pt]
_ __ __ \operatorname{d}e^5 \;=\; e^{13}_-_e^{24} \\[5 |
, when we say an algorithm **supports** or **can handle** a component, we mean that the cost of the component is $\leq 1$, although the protocols can be applied to any component possibly inefficiently.
Efficiency of Stabilizer Propagation
------------------------------------
For a self-contained description of stabilizer propagation see [@bennink; @resource; @seddon]. Just as the algorithms in section II decompose input states into a weighted sum of Pauli matrices, stabilizer propagation decomposes input states into a weighted sum of stabilizer states. A sampling process identical to equations and results in the number of samples required to be proportional to the square of the following normalization constant:
The **robustness of magic** $\mathcal{R}(\rho)$ of an $n$-qubit state $\rho$ is the outcome of a convex optimization program over real vectors $\vec q$: $$\mathcal{R}(\rho) = \min_{\vec q} \sum_i \lvert q_i \rvert \text{ s.t. } \rho = \sum_i q_i \ket{\phi_i} \bra{\phi_i} \text{ and } \sum_i q_i = 1,$$
where $\{\ket{\phi_i}\}$ are the $n$-qubit stabilizer states.
When $\mathcal{R}(\rho) = 1$ (the minimum value) then $\rho$ is a **stabilizer mixture**, since then the vector $\vec q$ is a probability distribution.
Due to the sheer number of stabilizer states, evaluating $\mathcal{R}(\rho)$ for even small $n$ is very expensive. As stated in [@bennink], evaluating the cost function for 3-qubit unitaries is impractical, although the performance can be improved for diagonal gates [@seddon].
The performance of stabilizer propagation gives a lens for the non-Cliffordness of channels, studied extensively in [@seddon]. In the appendix, we expand on this work by modifying the protocol to support all **postselective channels** which include all trace preserving channels and all ‘reasonable’ non-trace-preserving channels. There, we prove the following theorem:\
\[thm:stabprop\] Let $\Lambda$ be a postselective channel and let $\bar\phi_\Lambda$ be the channel’s normalized Choi state. $\Lambda$ does not increase the number of samples required for | , when we say an algorithm * * supports * * or * * can handle * * a component, we think of that the monetary value of the component is $ \leq 1 $, although the protocols can be applied to any part possibly inefficiently.
Efficiency of Stabilizer Propagation
------------------------------------
For a self - contained description of stabilizer propagation see [ @bennink; @resource; @seddon ]. merely as the algorithms in section II disintegrate stimulation states into a leaden sum of Pauli matrices, stabilizer propagation decomposes stimulation states into a weighted sum of stabilizer states. A sampling procedure identical to equations and results in the number of sample required to be proportional to the square of the following standardization constant:
The * * robustness of magic * * $ \mathcal{R}(\rho)$ of an $ n$-qubit country $ \rho$ is the outcome of a convex optimization program over real vector $ \vec q$: $ $ \mathcal{R}(\rho) = \min_{\vec q } \sum_i \lvert q_i \rvert \text { s.t. } \rho = \sum_i q_i \ket{\phi_i } \bra{\phi_i } \text { and } \sum_i q_i = 1,$$
where $ \{\ket{\phi_i}\}$ are the $ n$-qubit stabilizer states.
When $ \mathcal{R}(\rho) = 1 $ (the minimum value) then $ \rho$ is a * * stabilizer mixture * *, since then the vector $ \vec q$ is a probability distribution.
Due to the sheer number of stabilizer country, evaluating $ \mathcal{R}(\rho)$ for even small $ n$ is very expensive. As stated in [ @bennink ], evaluating the cost function for 3 - qubit unitaries is impractical, although the performance can be improved for diagonal gate [ @seddon ].
The performance of stabilizer propagation feed a lens for the non - Cliffordness of groove, studied extensively in [ @seddon ]. In the appendix, we expand on this work by change the protocol to support all * * postselective channels * * which include all trace preserve channels and all ‘ reasonable ’ non - trace - preserving channels. There, we prove the succeed theorem:\
\[thm: stabprop\ ] Let $ \Lambda$ be a postselective channel and let $ \bar\phi_\Lambda$ be the channel ’s anneal Choi state. $ \Lambda$ does not increase the act of samples required for | , whfn we say an algorithm **smpports** or **can hcbdle** a compohent, we oean that the cost of the colpinent is $\leq 1$, although the orotocols can be qpplmed to any compoisnt possibly jkeffieixntly.
Efficiency of Stabilhzer Propagatimn
------------------------------------
Wox a self-contained description of stafilizer pgopagation see [@benmynk; @dvsiurce; @seddon]. Just as the algodithms pn section II decpmpose input states into a weihhted sum of Pauli matrices, sjzbijuzer propagagion decomkoves input atates into a weighted sum of sgabilnzer states. A salkling procesw idegtical to eqmstions and rexults in the nmmber of samples required to ue proportional to tre square oy the following normaoizatiot cotstavr:
Thd **rkbnsthess ov megic** $\mathcam{R}(\rho)$ of an $n$-qubit state $\rho$ ix eye outcome of a condev optimization program over real vectorv $\vsc q$: $$\mathcal{R}(\rho) = \min_{\vex q} \sum_i \lvert q_i \rvett \text{ s.t. } \rho = \sum_i q_i \ket{\phi_i} \bra{\phi_i} \text{ and } \sum_i q_i = 1,$$
wherx $\{\yet{\khl_l}\}$ ard tje $n$-qubit stabilizer states.
When $\mathcal{R}(\rho) = 1$ (fht mpnimum value) then $\rho$ is a **syahikyzer mixture**, rince cgeh the vector $\vec q$ is a ptobabioity distwibuyion.
Due to the sheer number of stabilizvr srates, evaluating $\mcthcal{R}(\rho)$ fur efen skall $n$ is very expensivz. As sfated in [@bejnink], evamjating the cost wunbtiot for 3-qubit unitaries is iipracticao, alchough tfe pgrformagce can be improved for diagonal gwtes [@degdon].
The pegformance of stabilizer propagavmon gives a lgns fog the non-Eliffovdness of channqls, studied exjensively in [@sdddon]. In tge appeidix, we expagd on this wosn by modifyiig the prjtocil ti suppofg all **postselevtive chauuels** whicy include all tracc prersrving channels ane all ‘reasonablr’ nun-twabe-pceserdhng channels. Thefe, dr proxe the folliding theorem:\
\[thm:stabprop\] Net $\Mambda$ be a postsekegtive chabnel and let $\bar\phi_\Lakbda$ be the channep’s nocmalizxd Choo sjate. $\Lambda$ does not increase tge number of samples reqtiree for | , when we say an algorithm **supports** handle** component, we that the cost 1$, the protocols can applied to any possibly inefficiently. Efficiency of Stabilizer Propagation For a self-contained description of stabilizer propagation see [@bennink; @resource; @seddon]. Just as algorithms in section II decompose input states into a weighted sum of Pauli stabilizer decomposes states a weighted sum of stabilizer states. A sampling process identical to equations and results in the of samples required to be proportional to the of the following normalization The **robustness of magic** $\mathcal{R}(\rho)$ an state $\rho$ the of convex optimization program real vectors $\vec q$: $$\mathcal{R}(\rho) = \min_{\vec q} \sum_i \lvert q_i \rvert \text{ s.t. } \rho = q_i \ket{\phi_i} and } q_i 1,$$ $\{\ket{\phi_i}\}$ are the states. When $\mathcal{R}(\rho) = 1$ (the $\rho$ is a **stabilizer mixture**, since then the $\vec q$ a probability distribution. Due to the number of stabilizer states, evaluating $\mathcal{R}(\rho)$ for even $n$ is very expensive. As stated in [@bennink], evaluating the cost function for 3-qubit unitaries although the performance can improved for diagonal [@seddon]. performance stabilizer gives a for the non-Cliffordness of channels, studied extensively in [@seddon]. In the we expand on this work by modifying the protocol to **postselective which include all preserving channels and all non-trace-preserving There, we prove the \[thm:stabprop\] $\Lambda$ channel let be the channel’s normalized state. $\Lambda$ does not increase number of samples required | , when we say an algorithm **suppoRts** or **can haNdle** a ComPonEnT, we mEan tHat the cost of thE CompOnent is $\leq 1$, although the pRotocOlS Can bE ApPlied To any coMPoNENt pOsSiBly InEFfIcienTly.
efficieNcy of StabiLizEr propagation
------------------------------------
FOR a Self-contaiNed Description oF stAbilizEr ProPAgatiOn sEe [@benNink; @reSOurce; @sEddon]. Just As THe algoRIthms in SECtIon Ii decompose input stATeS Into a weighted sUm of PaUlI MaTRIceS, stAbilizer prOpAgatiON decompOSeS INPut STates into a weiGhted sum of sTAbiLizer sTaTes. a SampliNg proCeSS idEntical to eqUatiOns and resUlts in THe numbeR Of samplEs requIreD to Be prOPoRtIonAl TO thE SqUarE Of tHe followInG nOrmalIzatION COnstAnt:
the **rObustNess of magic** $\maThcAl{R}(\rHO)$ of An $n$-quBit stAte $\rHo$ Is the OutcomE of a cOnVex optimization ProgRam over reAl vEcTorS $\vEc q$: $$\maTHcal{R}(\rHo) = \mIn_{\vEc q} \sum_i \Lvert q_i \RVerT \tEXT{ S.t. } \Rho = \sum_i q_i \ket{\phi_i} \brA{\pHI_I} \tExt{ and } \suM_i q_i = 1,$$
whERe $\{\KeT{\Phi_i}\}$ are tHe $N$-quBit sTABilizEr stATeS.
When $\matHcal{R}(\rHO) = 1$ (tHe Minimum VaLue) theN $\rHo$ iS a **sTabilIZer mIxture**, Since theN the vECtor $\vec q$ is a proBAbility distriBUtION.
DUE to tHe sHeer number oF staBIlizEr stATeS, evALuatiNg $\matHcAL{R}(\RHo)$ for even small $n$ is veRy ExpensIve. As Stated in [@benniNk], evaluatiNG THe cost fuNctiON fOR 3-qubit unitarieS is imPractical, aLThough thE perfOrmance cAn be improVED for diagOnaL gaTes [@SedDON].
THe performance OF StabIlIzer proPagAtion giVes A leNs fOr tHe Non-CliffoRdness of ChAnNeLs, StuDied eXTensivelY iN [@seDdOn]. IN the aPPendix, We expAnd oN tHiS WorK by modiFYiNG The pRoToCol tO suPpOrt alL **posTSelEctive cHannels** whIch INcluDe AlL trace pReserving chanNeLs and all ‘reAsOnaBle’ non-TRAce-preseRving channels. There, we proVE the folLowIng thEoreM:\
\[thm:stabpRop\] let $\LamBda$ BE a postSelectIve chAnNel AND let $\bAR\PhI_\LaMbDa$ be the chaNNEl’s NormaLiZed CHoi statE. $\Lambda$ does not incrEAse The number of saMplEs reQUIrEd fOR | , when we say an algorithm **support s** o r * *ca nhand le** a component,w e me an that the cost of th e com po n enti s$\leq 1$, al t ho u g h t he p rot oc o ls canbeapplied to any co mpo ne nt possiblyi ne fficiently .
Efficiency o f S tabili ze r P r opaga tio n
--- ------ - ------ --------- -- - ------ - -
Fora se lf-c ontained descript i on of stabilizerpropag at i on s ee[@b ennink; @r es ource ; @seddo n ]. J u sta s the algorit hms in sect i onII dec om pos e input stat es int o a weighte d su m of Paul i matr i ces, st a bilizer propa gat ion dec o mp os esin p uts ta tes int o a weig ht ed sumof s t a b i lize r s tate s. Asampling proc ess ide n tic al to equa tion sand r esults in t he number of samp lesrequiredtobe pr op ortio n al tothe sq uare of the fo l low in g n or malization constan t: Th e **robu stness of m a gic** $\ ma thc al{R } ( \rho) $ of an $n$-qub it sta t e$\ rho$ is t he out co meofa con v ex o ptimiz ation pr ogram over real vect o rs $\vec q$:$ $\ m a th c al{R }(\ rho) = \min _{\v e c q} \su m _i \l v ert q _i \r ve r t\ text{ s.t. } \rho = \ sum_iq_i \ ket{\phi_i} \ bra{\phi_i } \ text{ an d }\ su m _i q_i = 1,$$
wher e $\{\ket{ \ phi_i}\} $ are the $n$ -qubit st a b ilizer s tat es.
W hen $ \m athcal{R}(\rh o ) = 1 $(the mi nim um valu e)the n $ \rh o$ is a **s tabilize rmi xt ur e** , sin c e then t he ve ct or$\vec q$ isa pro babi li ty dis tributi o n. Dueto t he s hee rnumbe r of sta bilizer states,eva l uati ng $ \mathca l{R}(\rho)$ f or even smal l$n$ is ve r y expensi ve. As stated in [@benn i nk], ev alu ating the cost fun cti on for 3- q ubit u nitari es is i mpr a c tical , al tho ug h the perf o r man ce ca nbe i mproved for diagonal gate s [@ seddon].
The pe rfor m a nc e o f s t abi li z erp r opagation gives a lens fo rt he non-Cliff o rdn es s of ch annels, stud i ed exte nsively i n [@seddo n] . In t heappendix,we expan d on this workb ymodif yin g thepr oto col t o supp o rtall * *posts el ective chan ne ls** whi ch include all trace pr eservi ng ch ann els and a ll‘ rea sonable’non- trace-pres erv ing chan nel s . The re,w epro v e the fol l owing the o re m:\ \[ thm:stabpro p \ ] Le t $\L amb d a$ bea po stselective chann e l and let $\ba r\ph i _ \La mbd a $ be t he channel’s n orm al i z ed Choist ate. $\Lamb da$ does n o t inc reasethe nu mber of s am p les re quir edfor | , when_we say_an algorithm **supports** or_**can handle**_a_component, we_mean_that the cost_of the component_is $\leq 1$, although_the protocols can_be_applied to any component possibly inefficiently.
Efficiency of Stabilizer Propagation
------------------------------------
For a self-contained description of stabilizer_propagation_see [@bennink;_@resource;_@seddon]._Just as the algorithms in_section II decompose input states_into a_weighted sum of Pauli matrices, stabilizer propagation decomposes_input_states into a_weighted sum of stabilizer states. A sampling process identical_to equations and results in the_number of samples_required_to_be proportional to the_square of the following normalization constant:
The_**robustness of magic** $\mathcal{R}(\rho)$ of an_$n$-qubit state $\rho$ is the outcome of_a convex optimization program over real_vectors $\vec q$: $$\mathcal{R}(\rho) =_\min_{\vec q}_\sum_i \lvert q_i \rvert \text{_s.t. } \rho_= \sum_i_q_i \ket{\phi_i} \bra{\phi_i}_\text{ and } \sum_i q_i =_1,$$
where $\{\ket{\phi_i}\}$ are_the $n$-qubit stabilizer states.
When $\mathcal{R}(\rho) =_1$_(the minimum value)_then_$\rho$_is a_**stabilizer mixture**, since_then_the vector_$\vec_q$ is a probability distribution.
Due to_the_sheer number of stabilizer states, evaluating $\mathcal{R}(\rho)$_for even small $n$_is_very expensive. As stated_in [@bennink], evaluating the cost_function for 3-qubit unitaries is impractical,_although the_performance can_be improved for diagonal gates [@seddon].
The performance of stabilizer propagation gives_a lens for the non-Cliffordness of_channels, studied extensively in_[@seddon]. In_the_appendix, we expand_on_this work_by modifying the protocol to support all_**postselective channels**_which include all trace preserving channels_and all ‘reasonable’ non-trace-preserving_channels._There, we prove the following theorem:\
\[thm:stabprop\]_Let $\Lambda$ be a postselective channel_and let $\bar\phi_\Lambda$ be the_channel’s_normalized_Choi state. $\Lambda$ does not_increase the number of samples required_for |
respect to monomials (see \[def:mon\]) with $\mu=B{\varepsilon}^{-2/5}$. Note that in this domain, $\mu$ is well defined in the sense that $\mu$ is bounded by a constant as ${\varepsilon}\to 0$.
3. Restricted to $(B,Z,{\varepsilon})\in\Sigma^{+b}$, there is a function $\tilde\Psi$ such that [ $$\begin{split}
\phi(B,{\varepsilon}) &= 0\\
\Psi(B,Z,{\varepsilon}) &= \tilde\Psi\left(|B|^{1/2},{\varepsilon}^{1/5}, {\varepsilon}\ln(|B|), \sigma,Z\right),
\end{split}$$ ]{} where $\tilde\Psi$ is a ${\mathcal{C}^{\ell}}$-function with respect to monomials (see \[def:mon\]) with $\sigma={\varepsilon}|B|^{-5/2}$. Note that in this domain, $\sigma$ is well defined since $|B|>0$.
4. Restricted to $(B,Z,{\varepsilon})\in\Sigma^{-b}$, there are functions $\tilde \phi$ and $\tilde \Psi$ such that [ $$\begin{split}
\phi(B,{\varepsilon}) &=\tilde\phi\left(|B|^{1/2},\sigma\right)\\
\Psi(B,Z,{\varepsilon}) &= \tilde\Psi\left(|B|^{1/2},{\varepsilon}^{1/5}, {\varepsilon}\ln(|B|), \sigma\right),
\end{split}$$ ]{} where $\tilde \phi$ and $\tilde \Psi$ are ${\mathcal{C}^{\ell}}$-functions with respect to monomials (see \[def:mon\]) with $\sigma={\varepsilon}|B|^{-5/2}$. Note that in this domain, $\sigma$ is well defined since $|B|>0$.
#### Sketch of the proof.
The first step is to recall \[prop:formal\_nf\], which shows that $X$ is formally conjugate to [ $$\begin{split}
F={\varepsilon}{\dfrac{\partial}{\partial {a}}}+0{\dfrac{\partial}{\partial {b}}}-\left(z^3+{b}z+{a}\right){\dfrac{\partial}{\partial z}} + 0{\dfrac{\partial}{\ | respect to monomials (see \[def: mon\ ]) with $ \mu = B{\varepsilon}^{-2/5}$. Note that in this domain, $ \mu$ is well defined in the common sense that $ \mu$ is restrict by a constant as $ { \varepsilon}\to 0$.
3. Restricted to $ (boron, Z,{\varepsilon})\in\Sigma^{+b}$, there be a function $ \tilde\Psi$ such that [ $ $ \begin{split }
\phi(B,{\varepsilon }) & = 0\\
\Psi(B, Z,{\varepsilon }) & = \tilde\Psi\left(|B|^{1/2},{\varepsilon}^{1/5 }, { \varepsilon}\ln(|B|), \sigma, Z\right),
\end{split}$$ ] { } where $ \tilde\Psi$ is a $ { \mathcal{C}^{\ell}}$-function with respect to monomials (experience \[def: mon\ ]) with $ \sigma={\varepsilon}|B|^{-5/2}$. eminence that in this domain, $ \sigma$ is well specify since $ |B|>0$.
4. Restricted to $ (B, Z,{\varepsilon})\in\Sigma^{-b}$, there are function $ \tilde \phi$ and $ \tilde \Psi$ such that [ $ $ \begin{split }
\phi(B,{\varepsilon }) & = \tilde\phi\left(|B|^{1/2},\sigma\right)\\
\Psi(B, Z,{\varepsilon }) & = \tilde\Psi\left(|B|^{1/2},{\varepsilon}^{1/5 }, { \varepsilon}\ln(|B|), \sigma\right),
\end{split}$$ ] { } where $ \tilde \phi$ and $ \tilde \Psi$ are $ { \mathcal{C}^{\ell}}$-functions with respect to monomials (see \[def: mon\ ]) with $ \sigma={\varepsilon}|B|^{-5/2}$. eminence that in this knowledge domain, $ \sigma$ is well defined since $ |B|>0$.
# # # # Sketch of the proof.
The first step is to recall \[prop: formal\_nf\ ], which read that $ X$ is formally conjugate to [ $ $ \begin{split }
F={\varepsilon}{\dfrac{\partial}{\partial { a}}}+0{\dfrac{\partial}{\partial { b}}}-\left(z^3+{b}z+{a}\right){\dfrac{\partial}{\partial z } } + 0{\dfrac{\partial}{\ | redpect to monomials (see \[dtf:mon\]) with $\mu=B{\vargpwilon}^{-2/5}$. Iote thzt in thks domain, $\mu$ is well defined ib the sense that $\mu$ is bounaed by a bonstant qs ${\verepsilon}\to 0$.
3. Restricted to $(B,Z,{\bwrepvmlon})\in\Sigma^{+b}$, thgre is a funwtion $\tilde\Psi$ sjck that [ $$\begin{split}
\phi(F,{\varepsolln}) &= 0\\
\Ksi(B,Z,{\nawepsjlon}) &= \tilde\Psi\left(|B|^{1/2},{\varepsilon}^{1/5}, {\varelsilon}\li(|B|), \sigma,Z\right),
\end{split}$$ ]{} whege $\tllde\Psi$ is a ${\mathcwl{C}^{\ell}}$-functuon ruth respect go monomials (see \[def:moh\]) with $\sigma={\varepsilon}|B|^{-5/2}$. Note thxt in this domaun, $\sihka$ is well vefineq since $|B|>0$.
4. Rcxtrictad to $(B,A,{\varepsilon})\in\Slgma^{-b}$, thwre are functions $\tilve \phi$ and $\tilde \Psi$ such thad [ $$\begin{split}
\phh(B,{\vafwpskloh}) &=\vilse\phi\lfft(|U|^{1/2},\sigma\right)\\
\Pwi(B,Z,{\varepsilon}) &= \tilce\[wi\left(|B|^{1/2},{\varepsjlon}^{1/5}, {\vwrqpsilon}\ln(|B|), \sigma\right),
\end{sklit}$$ ]{} where $\tilde \phi$ and $\tiode \Psi$ are ${\mathcal{C}^{\epl}}$-functiogs with respect to monomials (see \[def:mon\]) with $\sigmd={\vare'sklou}|N|^{-5/2}$. Vitf that in this domain, $\sigma$ is well defined syhct $|B|>0$.
#### Sketch of the proof.
The fordt ftep is to rezall \[pxkp:rormal\_nf\], which shoas that $X$ is formally conkugate to [ $$\begin{split}
D={\varepsilon}{\dyrax{\partial}{\partial {a}}}+0{\dyrac{\partial}{\pcrtial {b}}}-\lefy(z^3+{b}z+{a}\right){\dfrac{\partial}{\pcrtial z}} + 0{\dfrac{\pagtial}{\ | respect to monomials (see \[def:mon\]) with $\mu=B{\varepsilon}^{-2/5}$. in domain, $\mu$ well defined in bounded a constant as 0$. 3. Restricted $(B,Z,{\varepsilon})\in\Sigma^{+b}$, there is a function $\tilde\Psi$ that [ $$\begin{split} \phi(B,{\varepsilon}) &= 0\\ \Psi(B,Z,{\varepsilon}) &= \tilde\Psi\left(|B|^{1/2},{\varepsilon}^{1/5}, {\varepsilon}\ln(|B|), \sigma,Z\right), \end{split}$$ ]{} $\tilde\Psi$ is a ${\mathcal{C}^{\ell}}$-function with respect to monomials (see \[def:mon\]) with $\sigma={\varepsilon}|B|^{-5/2}$. Note in domain, is defined since $|B|>0$. 4. Restricted to $(B,Z,{\varepsilon})\in\Sigma^{-b}$, there are functions $\tilde \phi$ and $\tilde \Psi$ such [ $$\begin{split} \phi(B,{\varepsilon}) &=\tilde\phi\left(|B|^{1/2},\sigma\right)\\ \Psi(B,Z,{\varepsilon}) &= \tilde\Psi\left(|B|^{1/2},{\varepsilon}^{1/5}, {\varepsilon}\ln(|B|), \end{split}$$ ]{} where $\tilde and $\tilde \Psi$ are ${\mathcal{C}^{\ell}}$-functions respect monomials (see with Note in this domain, is well defined since $|B|>0$. #### Sketch of the proof. The first step is to recall \[prop:formal\_nf\], shows that formally conjugate [ F={\varepsilon}{\dfrac{\partial}{\partial {b}}}-\left(z^3+{b}z+{a}\right){\dfrac{\partial}{\partial z}} + | respect to monomials (see \[def:mOn\]) with $\mu=B{\vArepsIloN}^{-2/5}$. NoTe That In thIs domain, $\mu$ is weLL defIned in the sense that $\mu$ is BoundEd BY a coNStAnt as ${\VarepsiLOn}\TO 0$.
3. resTrIcTed To $(b,z,{\vArepsIloN})\in\SigmA^{+b}$, there is a FunCtIon $\tilde\Psi$ sUCh That [ $$\begin{sPliT}
\phi(B,{\varepsiLon}) &= 0\\
\psi(B,Z,{\vArEpsILon}) &= \tiLde\psi\leFt(|B|^{1/2},{\varEPsilon}^{1/5}, {\VarepsiloN}\lN(|b|), \sigma,z\Right),
\enD{SPlIt}$$ ]{} whEre $\tilde\Psi$ is a ${\matHCaL{c}^{\ell}}$-function wiTh respEcT To MONomIalS (see \[def:mon\]) WiTh $\sigMA={\varepsILoN}|b|^{-5/2}$. nOte THat in this domaIn, $\sigma$ is weLL deFined sInCe $|B|>0$.
4. rEstricTed to $(b,Z,{\VArePsilon})\in\SigMa^{-b}$, tHere are fuNctionS $\Tilde \phI$ And $\tildE \Psi$ suCh tHat [ $$\BegiN{SpLiT}
\phI(B,{\VArePSiLon}) &=\TIldE\phi\left(|b|^{1/2},\sIgMa\rigHt)\\
\PsI(b,z,{\VArepSilOn}) &= \tiLde\PsI\left(|B|^{1/2},{\varepsiLon}^{1/5}, {\VarePSilOn}\ln(|B|), \Sigma\RighT),
\eNd{splIt}$$ ]{} wherE $\tildE \pHi$ and $\tilde \Psi$ arE ${\matHcal{C}^{\ell}}$-fUncTiOns WiTh resPEct to mOnoMiaLs (see \[deF:mon\]) witH $\SigMa={\VAREpSilon}|B|^{-5/2}$. Note that in thIs DOMaIn, $\sigma$ iS well dEFiNeD Since $|B|>0$.
#### SkEtCh oF the PROof.
ThE firST sTep is to rEcall \[pROp:FoRmal\_nf\], wHiCh showS tHat $x$ is FormaLLy coNjugatE to [ $$\begin{Split}
f={\Varepsilon}{\dfraC{\Partial}{\partiaL {A}}}+0{\dFRAc{\PArtiAl}{\pArtial {b}}}-\left(Z^3+{b}z+{a}\RIght){\DfraC{\PaRtiAL}{\partIal z}} + 0{\dFrAC{\pARtial}{\ | respect to monomials (see \[def:mon \]) w ith $\ mu =B{\ vare psilon}^{-2/5} $ . No te that in this domain , $\m u$ is w e ll defi ned int he s ens eth at$\ m u$ is b oun ded bya constant as $ {\varepsilon } \t o 0$.
3. Re stricted to$(B ,Z,{\v ar eps i lon}) \in \Sigm a^{+b} $ , ther e is a fu nc t ion $\ t ilde\Ps i $ s uchthat [ $$\begin{s p li t }
\p hi ( B, { \ var eps ilon}) &=0\ \
\ P si ( B , Z,{ \ varepsilon})&= \tilde\P s i\l eft(|B |^ {1/ 2 },{\va repsi lo n }^{ 1/5}, {\var epsi lon}\ln(| B|), \ s igma,Z\ r ight), \ e nd {s pli t} $ $ ] { }whe r e $ \tilde\P si $is a${\m a t h c al{C }^{ \ell }}$-f unction withres pect tomonom ials(see \ [def: mon\]) with $ \sigma={\vareps ilon }|B|^{-5/ 2}$ .Not ethati n this do mai n, $\si gma$ is wel ld e f in ed since $|B|>0$.
4 . R estricte d to $ ( B, Z, { \varepsi lo n}) \in\ S i gma^{ -b}$ , t here are funct i on s$\tilde \ phi$ a nd $\ til de \P s i$ s uch th at [ $$\ begin { split}
\phi( B ,{ \ v ar e psil on} ) &=\tilde\ phi\ l eft( |B|^ { 1/ 2}, \ sigma \righ t) \ \ \Psi(B,Z,{ \v arepsi lon}) &= \tilde\Ps i\left(|B| ^ { 1 /2},{\va reps i lo n }^{1/5}, {\var epsil on}\ln(|B| ) , \sigma \righ t),
\end{spl it} $$]{} wh e r e$\tilde \phi$ a nd $ \t ilde \P si$ are ${ \ma thc al{ C}^ {\ ell}}$-fu nctionswi th r es pec t tom onomials ( see \ [de f:mon \ ]) wit h $\s igma ={ \v a rep silon}| B |^ { - 5/2} $. N otetha tin th is d o mai n, $\si gma$ is w ell defi ne dsince $ |B|>0$.
#### S ketch of t he pr oof.
T h e firststep is to recall \[pro p :formal \_n f\],whic h shows t hat $X$ i s f o rmally conju gateto [$ $ \begi n { sp lit }
F={ \ v are psilo n} {\df rac{\pa rtial}{\partial {a } }}+ 0{\dfrac{\par tia l}{\ p a rt ial {b } }}- \l e ft( z ^ 3+{b}z+{a}\righ t){\dfrac{ \p a rt ial}{\part i alz} } + 0{\ dfrac{\ parti a l}{\ | respect_to monomials_(see \[def:mon\]) with $\mu=B{\varepsilon}^{-2/5}$._Note that_in_this domain,_$\mu$_is well defined_in the sense_that $\mu$ is bounded_by a constant_as_${\varepsilon}\to 0$.
3. Restricted to $(B,Z,{\varepsilon})\in\Sigma^{+b}$, there is a function $\tilde\Psi$ such that [_$$\begin{split}
_ ___
_ _\phi(B,{\varepsilon}) &=_0\\
__ \Psi(B,Z,{\varepsilon}) &=_\tilde\Psi\left(|B|^{1/2},{\varepsilon}^{1/5}, {\varepsilon}\ln(|B|), \sigma,Z\right),
_
_ \end{split}$$ ]{}_where_$\tilde\Psi$_is a ${\mathcal{C}^{\ell}}$-function with_respect to monomials (see \[def:mon\]) with_$\sigma={\varepsilon}|B|^{-5/2}$. Note that in this domain,_$\sigma$ is well defined since $|B|>0$.
4. _Restricted to $(B,Z,{\varepsilon})\in\Sigma^{-b}$, there are functions_$\tilde \phi$ and $\tilde \Psi$_such that_[ $$\begin{split}
_ _
_ _ \phi(B,{\varepsilon}) &=\tilde\phi\left(|B|^{1/2},\sigma\right)\\
_ _ \Psi(B,Z,{\varepsilon})_&=_\tilde\Psi\left(|B|^{1/2},{\varepsilon}^{1/5}, {\varepsilon}\ln(|B|), \sigma\right),
___ _ _
_ __ \end{split}$$ ]{} where $\tilde_\phi$_and $\tilde \Psi$ are ${\mathcal{C}^{\ell}}$-functions with respect_to monomials (see \[def:mon\])_with_$\sigma={\varepsilon}|B|^{-5/2}$. Note that in_this domain, $\sigma$ is well_defined since $|B|>0$.
#### Sketch of the_proof.
The first_step is_to recall \[prop:formal\_nf\], which shows that $X$ is formally conjugate to_[ $$\begin{split}
_ F={\varepsilon}{\dfrac{\partial}{\partial {a}}}+0{\dfrac{\partial}{\partial {b}}}-\left(z^3+{b}z+{a}\right){\dfrac{\partial}{\partial_z}} +_0{\dfrac{\partial}{\ |
sigma_N}\}$ forms a basis of $F^{n-1}_q$.
As both $\{\Omega(q), \theta_1(q), \cdots, \theta_N(q)\}$ and $\{\Omega(q), \frac{\partial{\Omega(q)}}{\partial\sigma_1},\cdots, \frac{\partial{\Omega(q)}}{\partial\sigma_N}\}$ are bases for $F^{n-1}_q$, there exists $\{X_1, \cdots, X_N\}$ such that $X_k=\sum_{i=1}^N a_{ik} \frac{\partial}{\partial \sigma_i}$ for each $1\leq k\leq N$ such that $$\begin{aligned}
\label{eq1111}\theta_k=X_k(\Omega(q))+\lambda_k\Omega(q)\quad \text{for}\quad 1\leq k\leq N.\end{aligned}$$
Note that we have$X_k(\Omega(q))=X_k(\tau_1(q))\eta_1+\cdots + X_k(\tau_N(q))\eta_N+X_k(g_0(q))$ and $\theta_k=\eta_k+g_k(q)$, where $X_k(g_0(q)), g_k(q)\in \oplus_{k\geq2}H^k(M_q, \Omega^{n-k}(M_q))$. By comparing the types of classes in, we get $$\begin{aligned}
\label{eq2222}
\lambda_k=0, \quad \text{and}\quad X_k(\Omega(q))=\theta_k(q)=\eta_k+g_k(q) \quad\text{for each } 1\leq k\leq N.\end{aligned}$$
Since $\{\theta_1(q), \cdots, \theta_N(q)\}$ are linearly independent set for $F^{n-1}(M_q)$, we know that $\{X_1, \cdots, X_N\}$ are also linearly independent in $T^{1,0}_q({\mathcal{T}})$. Therefore $\{X_1, \cdots, X_N\}$ forms a basis for $T^{1,0}_q({\mathcal{T}})$. Without loss of generality, we may assume $X | sigma_N}\}$ forms a basis of $ F^{n-1}_q$.
As both $ \{\Omega(q), \theta_1(q), \cdots, \theta_N(q)\}$ and $ \{\Omega(q), \frac{\partial{\Omega(q)}}{\partial\sigma_1},\cdots, \frac{\partial{\Omega(q)}}{\partial\sigma_N}\}$ are bases for $ F^{n-1}_q$, there exists $ \{X_1, \cdots, X_N\}$ such that $ X_k=\sum_{i=1}^N a_{ik } \frac{\partial}{\partial \sigma_i}$ for each $ 1\leq k\leq N$ such that $ $ \begin{aligned }
\label{eq1111}\theta_k = X_k(\Omega(q))+\lambda_k\Omega(q)\quad \text{for}\quad 1\leq k\leq N.\end{aligned}$$
notice that we have$X_k(\Omega(q))=X_k(\tau_1(q))\eta_1+\cdots + X_k(\tau_N(q))\eta_N+X_k(g_0(q))$ and $ \theta_k=\eta_k+g_k(q)$, where $ X_k(g_0(q) ), g_k(q)\in \oplus_{k\geq2}H^k(M_q, \Omega^{n - k}(M_q))$. By compare the types of classes in, we arrive $ $ \begin{aligned }
\label{eq2222 }
\lambda_k=0, \quad \text{and}\quad X_k(\Omega(q))=\theta_k(q)=\eta_k+g_k(q) \quad\text{for each } 1\leq k\leq N.\end{aligned}$$
Since $ \{\theta_1(q), \cdots, \theta_N(q)\}$ are linearly autonomous set for $ F^{n-1}(M_q)$, we know that $ \{X_1, \cdots, X_N\}$ are also linearly autonomous in $ T^{1,0}_q({\mathcal{T}})$. consequently $ \{X_1, \cdots, X_N\}$ forms a basis for $ T^{1,0}_q({\mathcal{T}})$. Without loss of generality, we may assume $ adam | sigla_N}\}$ forms a basis of $F^{n-1}_q$.
Xs both $\{\Omega(q), \jhwta_1(q), \cvots, \thsta_N(q)\}$ ana $\{\Omega(q), \frac{\partial{\Omega(q)}}{\pactiao\sigmq_1},\cdots, \frac{\partial{\Omegx(q)}}{\partial\digma_N}\}$ aee beses for $F^{n-1}_q$, thecs exists $\{X_1, \cdkbs, X_N\}$ wuch that $X_k=\suk_{i=1}^N a_{ik} \frdc{\partial}{\partidl \sngma_i}$ for each $1\leq k\leq N$ such that $$\fegin{alogjed}
\label{eq1111}\thetw_k=X_k(\Piega(s))+\lambda_k\Omega(q)\quad \text{for}\quad 1\leq k\leq N.\tnd{aligned}$$
Note thay we have$X_k(\Omega(q))=X_k(\tau_1(q))\eta_1+\fdotd + X_k(\tau_N(q))\eta_N+X_k(g_0(q))$ and $\theta_k=\gfa_k+d_j(q)$, where $X_k(g_0(d)), g_k(q)\in \oplus_{k\geq2}H^k(M_q, \Kmega^{n-k}(M_q))$. By comparing the typer of elasses in, qe geh $$\begin{alignxd}
\labej{eq2222}
\lambda_k=0, \qmsd \texd{and}\quac X_k(\Omega(q))=\thets_k(q)=\xta_k+t_k(q) \quad\text{for each } 1\leq k\leq N.\end{aligged}$$
Since $\{\dhzta_1(q), \cdots, \theta_N(q)\}$ arw oineatly itdepdbdevt aev fkr $F^{n-1}(M_e)$, wx know that $\{X_1, \cdots, X_N\}$ are also linearly onqvlendent in $T^{1,0}_s({\mathcwl{E}})$. Therefore $\{X_1, \cdots, X_N\}$ forms a basis fog $T^{1,0}_q({\jathcal{T}})$. Without loss od generality, we may adsume $X | sigma_N}\}$ forms a basis of $F^{n-1}_q$. As \theta_1(q), \theta_N(q)\}$ and \frac{\partial{\Omega(q)}}{\partial\sigma_1},\cdots, \frac{\partial{\Omega(q)}}{\partial\sigma_N}\}$ are $\{X_1, X_N\}$ such that a_{ik} \frac{\partial}{\partial \sigma_i}$ each $1\leq k\leq N$ such that \label{eq1111}\theta_k=X_k(\Omega(q))+\lambda_k\Omega(q)\quad \text{for}\quad 1\leq k\leq N.\end{aligned}$$ Note that we have$X_k(\Omega(q))=X_k(\tau_1(q))\eta_1+\cdots + X_k(\tau_N(q))\eta_N+X_k(g_0(q))$ and $\theta_k=\eta_k+g_k(q)$, $X_k(g_0(q)), g_k(q)\in \oplus_{k\geq2}H^k(M_q, \Omega^{n-k}(M_q))$. By comparing the types of classes in, we get \label{eq2222} \quad X_k(\Omega(q))=\theta_k(q)=\eta_k+g_k(q) each } 1\leq k\leq N.\end{aligned}$$ Since $\{\theta_1(q), \cdots, \theta_N(q)\}$ are linearly independent set for $F^{n-1}(M_q)$, we that $\{X_1, \cdots, X_N\}$ are also linearly independent $T^{1,0}_q({\mathcal{T}})$. Therefore $\{X_1, \cdots, forms a basis for $T^{1,0}_q({\mathcal{T}})$. loss generality, we assume | sigma_N}\}$ forms a basis of $F^{n-1}_q$.
As bOth $\{\Omega(q), \tHeta_1(q), \CdoTs, \tHeTa_N(q)\}$ And $\{\OMega(q), \frac{\partiAL{\OmeGa(q)}}{\partial\sigma_1},\cdots, \frAc{\parTiAL{\OmeGA(q)}}{\PartiAl\sigma_n}\}$ ArE BAseS fOr $f^{n-1}_q$, ThERe ExistS $\{X_1, \cDots, X_N\}$ sUch that $X_k=\sUm_{i=1}^n a_{Ik} \frac{\partiaL}{\PaRtial \sigma_I}$ foR each $1\leq k\leq n$ suCh that $$\BeGin{ALigneD}
\laBel{eq1111}\Theta_k=x_K(\Omega(Q))+\lambda_k\OMeGA(q)\quad \TExt{for}\qUAD 1\lEq k\lEq N.\end{aligned}$$
Note THaT We have$X_k(\Omega(q))=x_k(\tau_1(q))\EtA_1+\CdOTS + X_k(\Tau_n(q))\eta_N+X_k(g_0(q))$ AnD $\thetA_K=\eta_k+g_k(Q)$, WhERE $x_k(g_0(Q)), G_k(q)\in \oplus_{k\geQ2}H^k(M_q, \Omega^{n-K}(m_q))$. BY compaRiNg tHE types Of claSsES in, We get $$\begin{aLignEd}
\label{eq2222}
\Lambda_K=0, \Quad \texT{And}\quad x_k(\OmegA(q))=\tHetA_k(q)=\eTA_k+G_k(Q) \quAd\TExt{FOr EacH } 1\Leq K\leq N.\end{AlIgNed}$$
SiNce $\{\tHETA_1(Q), \cdoTs, \tHeta_n(q)\}$ are Linearly indepEndEnt sET foR $F^{n-1}(M_q)$, We knoW thaT $\{X_1, \Cdots, x_N\}$ are aLso liNeArly independent In $T^{1,0}_q({\Mathcal{T}})$. THerEfOre $\{x_1, \cDots, X_n\}$ Forms a BasIs fOr $T^{1,0}_q({\matHcal{T}})$. WiTHouT lOSS Of Generality, we may assUmE $x | sigma_N}\}$ forms a basisof $F^{n-1 }_q$.
A s b ot h $\ {\Om ega(q), \theta _ 1(q) , \cdots, \theta_N(q)\ }$ an d$ \{\O m eg a(q), \frac{ \ pa r t ial {\ Om ega (q ) }} {\par tia l\sigma _1},\cdots , \ fr ac{\partial{ \ Om ega(q)}}{\ par tial\sigma_N }\} $ areba ses for $ F^{ n-1}_ q$, th e re exi sts $\{X_ 1, \cdots , X_N\}$ s uc h th at $X_k=\sum_{i=1 } ^N a_{ik} \frac{\ partia l} { \p a r tia l \ sigma_i}$fo r eac h $1\leq k\ l e q N$ such that $$\ begin{align e d}\label {e q11 1 1}\the ta_k= X_ k (\O mega(q))+\l ambd a_k\Omega (q)\qu a d \text { for}\qu ad 1\l eqk\l eq N . \e nd {al ig n ed} $ $
No t e t hat we h av e$ X_k(\ Omeg a ( q ) )=X_ k(\ tau_ 1(q)) \eta_1+\cdots +X_k( \ tau _N(q) )\eta _N+X _k (g_0( q))$ a nd $\ th eta_k=\eta_k+g_ k(q) $, where$X_ k( g_0 (q )), g _ k(q)\i n \ opl us_{k\g eq2}H^k ( M_q ,\ O m eg a^{n-k}(M_q))$. By c o m pa ring the types of c l asses in ,weget$ $ \begi n{al i gn ed}
\lab el{eq2 2 22 }\lambda _k =0, \q ua d \ tex t{and } \qua d X_k( \Omega(q ))=\t h eta_k(q)=\eta_ k +g_k(q) \quad \ te x t {f o r ea ch} 1\leq k\l eq N . \end {ali g ne d}$ $
Sin ce $\ {\ t he t a_1(q), \cdots, \th et a_N(q) \}$ a re linearly i ndependent s e t for $F ^{n- 1 }( M _q)$, we knowthat$\{X_1, \c d ots, X_N \}$ a re alsolinearlyi n dependen t i n $ T^{ 1,0 } _ q( {\mathcal{T}} ) $ . Th er efore $ \{X _1, \cd ots , X _N\ }$fo rms a bas is for $ T^ {1 ,0 }_ q({ \math c al{T}})$ .Wit ho utlosso f gene ralit y, w ema y as sume $X | sigma_N}\}$ forms_a basis_of $F^{n-1}_q$.
As both $\{\Omega(q),_\theta_1(q), \cdots,_\theta_N(q)\}$_and $\{\Omega(q),_\frac{\partial{\Omega(q)}}{\partial\sigma_1},\cdots,_\frac{\partial{\Omega(q)}}{\partial\sigma_N}\}$ are bases_for $F^{n-1}_q$, there_exists $\{X_1, \cdots, X_N\}$_such that $X_k=\sum_{i=1}^N_a_{ik}_\frac{\partial}{\partial \sigma_i}$ for each $1\leq k\leq N$ such that $$\begin{aligned}
\label{eq1111}\theta_k=X_k(\Omega(q))+\lambda_k\Omega(q)\quad \text{for}\quad 1\leq k\leq N.\end{aligned}$$
Note_that_we have$X_k(\Omega(q))=X_k(\tau_1(q))\eta_1+\cdots_+_X_k(\tau_N(q))\eta_N+X_k(g_0(q))$_and $\theta_k=\eta_k+g_k(q)$, where $X_k(g_0(q)), g_k(q)\in_\oplus_{k\geq2}H^k(M_q, \Omega^{n-k}(M_q))$. By comparing the_types of_classes in, we get $$\begin{aligned}
\label{eq2222}
\lambda_k=0, \quad \text{and}\quad X_k(\Omega(q))=\theta_k(q)=\eta_k+g_k(q)_\quad\text{for_each } 1\leq_k\leq N.\end{aligned}$$
Since $\{\theta_1(q), \cdots, \theta_N(q)\}$ are linearly independent set_for $F^{n-1}(M_q)$, we know that $\{X_1,_\cdots, X_N\}$ are_also_linearly_independent in $T^{1,0}_q({\mathcal{T}})$. Therefore_$\{X_1, \cdots, X_N\}$ forms a basis_for $T^{1,0}_q({\mathcal{T}})$. Without loss of generality,_we may assume $X |
prime }$ belongs to $\hat{V}$:
$$0={\bf S}_{ext}({\bf r}^{<}). \label{eq:ET5}$$
In Equation (\[eq:ET5\]) ${\bf S}_{ext}$ is
$${\bf S}_{ext}({\bf r}^{<})=\sum_{j}{\bf S}_{j}^{(out)}({\bf r}^{<})-{\bf S}
_{\infty }({\bf r}^{<}), \label{eq:ET6}$$
where
$$\begin{aligned}
{\bf S}_{j}^{(out)}({\bf r}^{<}) & = & \nabla \times \nabla \times
\nonumber \\ [+3mm]
& &
\int_{S_{j}}d^{2}r^{\prime }\left( {\bf E}({\bf r}^{\prime })\frac{\partial
G({\bf r}^{<},{\bf r}^{\prime })}{\partial {\bf n}}-G({\bf r}^{<},{\bf r}
^{\prime })\frac{\partial {\bf E}({\bf r}^{\prime })}{\partial {\bf n}}
\right) \,. \, \, \, \, \, \,
\label{eq:ET7}\end{aligned}$$
In Equation (\[eq:ET7\]) the surface values of the electric vector are taken from the volume $\hat{V}$. The normal ${\bf n}$ now points towards the interior of each of the volumes $V_{j}$.
Also, ${\bf S}_{\infty }$ has the same meaning as Equation (\[eq:ET7\]), the surface of integration now being a large sphere whose radius will eventually tend to infinity. It is not difficult to see that $-{\bf S}
_{\infty }$ in Equation (\[eq:ET6\]) is $4\pi $ times the incident field ${\bf E}^{(inc)}({\bf r}^{<})$ ([*cf*]{}. Refs. [@nieto-vesperinas91] and [@pattanayak76a; @pattanayak76b]). Therefore Equation (\[eq:ET5\]) finally becomes
$$0={\bf E}^{(inc)}({\bf r}^{<})+\frac{1}{4\pi }\sum_{j}{\bf S}_{j}^{(out)}({\bf
r}^{<}). \label{eq:ET8}$$
Note that when | prime } $ belongs to $ \hat{V}$:
$ $ 0={\bf S}_{ext}({\bf r}^ { < }). \label{eq: ET5}$$
In Equation (\[eq: ET5\ ]) $ { \bf S}_{ext}$ is
$ $ { \bf S}_{ext}({\bf r}^{<})=\sum_{j}{\bf S}_{j}^{(out)}({\bf r}^{<})-{\bf S }
_ { \infty } ({ \bf r}^ { < }), \label{eq: ET6}$$
where
$ $ \begin{aligned }
{ \bf S}_{j}^{(out)}({\bf r}^ { < }) & = & \nabla \times \nabla \times
\nonumber \\ [ +3 mm ]
& &
\int_{S_{j}}d^{2}r^{\prime } \left ({ \bf E}({\bf r}^{\prime }) \frac{\partial
G({\bf r}^{<},{\bf r}^{\prime }) } { \partial { \bf n}}-G({\bf r}^{<},{\bf r }
^{\prime }) \frac{\partial { \bf E}({\bf r}^{\prime }) } { \partial { \bf n } }
\right) \, . \, \, \, \, \, \,
\label{eq: ET7}\end{aligned}$$
In equality (\[eq: ET7\ ]) the open values of the electric vector are taken from the bulk $ \hat{V}$. The normal $ { \bf n}$ immediately points towards the interior of each of the volume $ V_{j}$.
besides, $ { \bf S}_{\infty } $ has the same meaning as Equation (\[eq: ET7\ ]), the surface of integration now being a big sphere whose radius will finally tend to infinity. It is not difficult to see that $ -{\bf S }
_ { \infty } $ in Equation (\[eq: ET6\ ]) is $ 4\pi $ times the incident field $ { \bf E}^{(inc)}({\bf r}^{<})$ ([ * cf * ] { }. Refs. [ @nieto - vesperinas91 ] and [ @pattanayak76a; @pattanayak76b ]). consequently Equation (\[eq: ET5\ ]) finally becomes
$ $ 0={\bf E}^{(inc)}({\bf r}^{<})+\frac{1}{4\pi } \sum_{j}{\bf S}_{j}^{(out)}({\bf
r}^ { < }). \label{eq: ET8}$$
notice that when | prile }$ belongs to $\hat{V}$:
$$0={\bn S}_{ext}({\bf r}^{<}). \label{eq:ET5}$$
In Esuation (\[ed:ET5\]) ${\bf S}_{ext}$ is
$${\bf S}_{ext}({\bf c}^{<})=\sum_{h}{\bf S}_{h}^{(out)}({\bf r}^{<})-{\bf S}
_{\infty }({\cf r}^{<}), \labvl{eq:ET6}$$
whece
$$\begin{alignxs}
{\bf S}_{j}^{(out)}({\gn r}^{<}) & = & \nabla \times \nsbla \times
\nonumber \\ [+3km]
& &
\int_{S_{j}}d^{2}r^{\prime }\left( {\bf E}({\bf r}^{\pryme })\frav{\pwrtial
G({\bf t}^{<},{\bf r}^{\kriie })}{\pzgtlal {\bf n}}-G({\bf r}^{<},{\bf r}
^{\prime })\frac{\pzrtial {\uf E}({\bf r}^{\prime })}{\psrtial {\bf n}}
\right) \,. \, \, \, \, \, \,
\label{eq:ET7}\end{aligjed}$$
In Ewuatyin (\[eq:ET7\]) the sjrface vallzs of the emectric vector are taken from tfe vokume $\hat{V}$. Jkw nltmal ${\bf n}$ nox poinns towards thc interhor of rach of the vokumxs $V_{h}$.
Also, ${\bf S}_{\infty }$ ias the same meaning as Equathou (\[eq:ET7\]), the surface of unregrajion tow cwine a lergs sphege xhose radiua will evenrually tend to infimiej. It is not djfficujt to see that $-{\bf S}
_{\infty }$ in Equation (\[tq:ET6\]) js $4\pi $ times the incidebt field ${\bf E}^{(inc)}({\bf r}^{<})$ ([*ff*]{}. Refs. [@niqto-vesperinas91] and [@pattanayak76a; @pattanayak76b]). Therefora Equetkon (\[tq:CB5\]) fivqlpy becomes
$$0={\bf E}^{(inc)}({\bf r}^{<})+\frac{1}{4\pi }\sum_{j}{\bf S}_{j}^{(out)}({\bs
g}^{<}). \label{eq:ET8}$$
Njte that whrn | prime }$ belongs to $\hat{V}$: $$0={\bf S}_{ext}({\bf In (\[eq:ET5\]) ${\bf is $${\bf S}_{ext}({\bf }({\bf \label{eq:ET6}$$ where $$\begin{aligned} S}_{j}^{(out)}({\bf r}^{<}) & & \nabla \times \nabla \times \nonumber [+3mm] & & \int_{S_{j}}d^{2}r^{\prime }\left( {\bf E}({\bf r}^{\prime })\frac{\partial G({\bf r}^{<},{\bf r}^{\prime })}{\partial n}}-G({\bf r}^{<},{\bf r} ^{\prime })\frac{\partial {\bf E}({\bf r}^{\prime })}{\partial {\bf n}} \right) \,. \, \, \, In Equation (\[eq:ET7\]) the surface values of the electric vector are taken from the volume $\hat{V}$. normal ${\bf n}$ now points towards the interior each of the volumes Also, ${\bf S}_{\infty }$ has same as Equation the of now being a sphere whose radius will eventually tend to infinity. It is not difficult to see that $-{\bf S} }$ in is $4\pi times incident ${\bf E}^{(inc)}({\bf r}^{<})$ [@nieto-vesperinas91] and [@pattanayak76a; @pattanayak76b]). Therefore Equation $$0={\bf E}^{(inc)}({\bf r}^{<})+\frac{1}{4\pi }\sum_{j}{\bf S}_{j}^{(out)}({\bf r}^{<}). \label{eq:ET8}$$ Note when | prime }$ belongs to $\hat{V}$:
$$0={\bf S}_{ext}({\bF r}^{<}). \label{eq:Et5}$$
In EqUatIon (\[Eq:eT5\]) ${\bf s}_{ext}$ Is
$${\bf S}_{ext}({\bf r}^{<})=\sum_{J}{\Bf S}_{j}^{(Out)}({\bf r}^{<})-{\bf S}
_{\infty }({\bf r}^{<}), \label{Eq:ET6}$$
wHeRE
$$\begIN{aLigneD}
{\bf S}_{j}^{(ouT)}({\Bf R}^{<}) & = & \NAblA \tImEs \nAbLA \tImes
\nOnuMber \\ [+3mm]
& &
\iNt_{S_{j}}d^{2}r^{\primE }\leFt( {\Bf E}({\bf r}^{\prime })\fRAc{\Partial
G({\bf R}^{<},{\bf R}^{\prime })}{\partiaL {\bf N}}-G({\bf r}^{<},{\bF r}
^{\PriME })\frac{\ParTial {\bF E}({\bf r}^{\pRIme })}{\parTial {\bf n}}
\riGhT) \,. \, \, \, \, \, \,
\Label{eQ:eT7}\end{alIGNeD}$$
In EQuation (\[eq:ET7\]) the surFAcE Values of the eleCtric vEcTOr ARE taKen From the volUmE $\hat{V}$. tHe normaL ${\Bf N}$ NOW poINts towards the Interior of eACh oF the voLuMes $v_{J}$.
Also, ${\bF S}_{\infTy }$ HAs tHe same meaniNg as equation (\[eQ:ET7\]), the SUrface oF IntegraTion noW beIng A larGE sPhEre WhOSe rADiUs wILl eVentuallY tEnD to inFiniTY. iT Is noT diFficUlt to See that $-{\bf S}
_{\infTy }$ iN EquATioN (\[eq:ET6\]) Is $4\pi $ tImes ThE inciDent fiEld ${\bf e}^{(iNc)}({\bf r}^{<})$ ([*cf*]{}. Refs. [@nietO-vesPerinas91] anD [@paTtAnaYaK76a; @patTAnayak76B]). ThEreFore EquAtion (\[eq:et5\]) fiNaLLY BeComes
$$0={\bf E}^{(inc)}({\bf r}^{<})+\frac{1}{4\Pi }\SUM_{j}{\Bf S}_{j}^{(out)}({\bF
r}^{<}). \labeL{Eq:eT8}$$
nOte that wHeN | prime }$ belongs to $\hat{ V}$:
$$0={ \bf S} _{ ext} ({\b f r}^{<}). \l a bel{ eq:ET5}$$
In Equa tion(\ [ eq:E T 5\ ]) ${ \bf S}_ { ex t } $ i s
$$ { \b f S}_ {ex t}({\bf r}^{<})=\ sum _{ j}{\bf S}_{j } ^{ (out)}({\b f r }^{<})-{\bfS} _{ \i nft y }({\ bfr}^{< }), \ l abel{e q:ET6}$$
wher e
$ $ \ be gin{ aligned}
{\bf S} _ {j}^{(out)}({\ bf r}^ {< } )& = & \n abla \time s\nabl a \times \ non u mber \\ [+3mm ]
& &
\ int_{S _{ j}} d ^{2}r^ {\pri me }\l eft( {\bf E }({\ bf r}^{\p rime } ) \frac{\ p artial G( {\b f r }^{< } ,{ \b f r }^ { \pr i me }) } {\p artial { \b fn}}-G ({\b f r } ^{<} ,{\ bf r }
^{\prime })\ fra c{\p a rti al {\ bf E} ({\b fr}^{\ prime})}{\ pa rtial {\bf n}} \right) \ ,.\, \, \ , \,\ , \, \l abel{eq :ET7}\e n d{a li g n e d} $$
In Equatio n( \ [e q:ET7\]) the s u rf ac e valuesof th e el e c tricvect o rare take n from th evolume$\ hat{V} $. Th e n ormal ${\b f n}$now poin ts to w ards the inter i or of each of th e vo l umes $V _{j}$.
Als o , ${ \bfS }_ {\i n fty } $ has t h es ame meaning as Equa ti on (\[ eq:ET 7\]), the sur face of in t e g ration n ow b e in g a large spher e who se radiusw ill even tuall y tend t o infinit y . It is n otdif fic ult t osee that $-{\ b f S} _{\in fty }$ inEqu ati on(\[ eq :ET6\]) i s $4\pi$ti me sthe inci d ent fiel d${\ bf E} ^{(in c )}({\b f r}^ {<}) $([ * cf* ]{}. Re f s. [ @nie to -v espe rin as 91] a nd [ @ pat tanayak 76a; @pat tan a yak7 6b ]) . There fore Equation ( \[eq:ET5\] )fin ally b e c omes
$$0={\bf E}^{(inc)}({ \ bf r}^{ <}) +\fra c{1} {4\pi }\s um_ {j}{\b f S } _{j}^{ (out)} ({\bf
r } ^{<}) . \ lab el {eq:ET8}$$ Note t hatwhen | prime }$_belongs to_$\hat{V}$:
_$$0={\bf S}_{ext}({\bf_r}^{<})._ \label{eq:ET5}$$
__ In_Equation (\[eq:ET5\]) ${\bf S}_{ext}$_is
_$${\bf S}_{ext}({\bf r}^{<})=\sum_{j}{\bf_S}_{j}^{(out)}({\bf_r}^{<})-{\bf S}
_{\infty }({\bf r}^{<}), \label{eq:ET6}$$
where
__ _$$\begin{aligned}
__ {\bf S}_{j}^{(out)}({\bf r}^{<})_& = & \nabla \times_\nabla \times
_ \nonumber \\ [+3mm]
__& &
_ \int_{S_{j}}d^{2}r^{\prime }\left( {\bf E}({\bf r}^{\prime })\frac{\partial
_ G({\bf r}^{<},{\bf r}^{\prime })}{\partial_{\bf n}}-G({\bf r}^{<},{\bf_r}
__ ^{\prime })\frac{\partial_{\bf E}({\bf r}^{\prime })}{\partial {\bf n}}
_ \right) \,. \,_\, \, \, \, \,
_ \label{eq:ET7}\end{aligned}$$
_In Equation (\[eq:ET7\]) the surface values_of the_electric vector are taken from_the volume $\hat{V}$._The normal_${\bf n}$ now_points towards the interior of each_of the volumes_$V_{j}$.
Also, ${\bf_S}_{\infty_}$ has the_same_meaning_as Equation (\[eq:ET7\]),_the surface of_integration_now being_a_large sphere whose radius will eventually_tend_to infinity. It is not difficult to_see that $-{\bf S}
__ _{\infty }$_in Equation (\[eq:ET6\]) is $4\pi $_times the incident field ${\bf E}^{(inc)}({\bf_r}^{<})$ ([*cf*]{}._Refs. [@nieto-vesperinas91] and [@pattanayak76a;_@pattanayak76b]). Therefore Equation (\[eq:ET5\]) finally becomes
$$0={\bf E}^{(inc)}({\bf r}^{<})+\frac{1}{4\pi_}\sum_{j}{\bf S}_{j}^{(out)}({\bf
r}^{<})._ \label{eq:ET8}$$
_ Note_that_when |
)\|^2,$$ where ${{\bf x}}_{k+1}(\alpha)$ is calculated by (\[eq3-11\]).
Update the iterate ${{\bf x}}_{k+1}={{\bf x}}_{k+1}(\alpha_k)$. Choose an initial step size $\bar{\alpha}_{k+1}\in(0,\alpha_{\max}]$ for the next iteration. $k \gets k+1.$
According to Theorem \[th3-5\], for any constant $\eta\in(0,2)$, there exists a positive scalar $\tilde{\alpha}$ such that for all $\alpha\in(0,\tilde{\alpha}]$, $$f({{\bf x}}_{k+1}(\alpha)) - f({{\bf x}}_k) \leq -\eta\alpha\|{{\bf g}}({{\bf x}}_k)\|^2.$$ Hence, the curvilinear search process is well-defined.
Now, we present a curvilinear search algorithm (ACSA) formally in Algorithm \[alg\] for the smallest generalized eigenvalue and its associated eigenvector of a Hankel tensor. If our aim is to compute the largest generalized eigenvalue and its associated eigenvector of a Hankel tensor, we only need to change respectively (\[eq3-11\]) and (\[suf-dec\]) used in Steps 5 and 6 of the ACSA algorithm to $${{\bf x}}_{k+1}(\alpha) = \frac{1-\alpha^2\|{{\bf g}}({{\bf x}}_k)\|^2}{1+\alpha^2\|{{\bf g}}({{\bf x}}_k)\|^2}{{\bf x}}_k
+\frac{2\alpha}{1+\alpha^2\|{{\bf g}}({{\bf x}}_k)\|^2}{{\bf g}}({{\bf x}}_k),$$ and $$f({{\bf x}}_{k+1}(\alpha_k)) \geq f({{\bf x}}_k) + \eta\alpha_k\|{{\bf g}}({{\bf x}}_k)\|^2.$$
When the Z-eigenvalue of a Hankel tensor is considered, we have ${\cal E}{{\bf x}}^m=\|{{\bf x}}\|^m=1$ and the objective $f({{\bf x}})$ is a polynomial. Then, we could compute the global minimizer of the step size $\alpha_k$ (the exact line search) in each iteration as [@HCD]. However, we use a cheaper inexact line search here. The initial step size of the next iteration follows D | ) \|^2,$$ where $ { { \bf x}}_{k+1}(\alpha)$ is calculated by (\[eq3 - 11\ ]).
Update the iterate $ { { \bf x}}_{k+1}={{\bf x}}_{k+1}(\alpha_k)$. Choose an initial step size $ \bar{\alpha}_{k+1}\in(0,\alpha_{\max}]$ for the next iteration. $ k \gets k+1.$
According to Theorem \[th3 - 5\ ], for any changeless $ \eta\in(0,2)$, there exist a positive scalar $ \tilde{\alpha}$ such that for all $ \alpha\in(0,\tilde{\alpha}]$, $ $ f({{\bf x}}_{k+1}(\alpha) ) - f({{\bf x}}_k) \leq -\eta\alpha\|{{\bf g}}({{\bf x}}_k)\|^2.$$ Hence, the curvilinear search process is well - define.
Now, we present a curvilinear search algorithm (ACSA) formally in Algorithm \[alg\ ] for the humble generalized eigenvalue and its associated eigenvector of a Hankel tensor. If our aim is to compute the large generalized eigenvalue and its associated eigenvector of a Hankel tensor, we only need to deepen respectively (\[eq3 - 11\ ]) and (\[suf - dec\ ]) used in Steps 5 and 6 of the ACSA algorithm to $ $ { { \bf x}}_{k+1}(\alpha) = \frac{1-\alpha^2\|{{\bf g}}({{\bf x}}_k)\|^2}{1+\alpha^2\|{{\bf g}}({{\bf x}}_k)\|^2}{{\bf x}}_k
+ \frac{2\alpha}{1+\alpha^2\|{{\bf g}}({{\bf x}}_k)\|^2}{{\bf g}}({{\bf x}}_k),$$ and $ $ f({{\bf x}}_{k+1}(\alpha_k) ) \geq f({{\bf x}}_k) + \eta\alpha_k\|{{\bf g}}({{\bf x}}_k)\|^2.$$
When the Z - eigenvalue of a Hankel tensor is consider, we have $ { \cal E}{{\bf x}}^m=\|{{\bf x}}\|^m=1 $ and the objective $ f({{\bf x}})$ is a polynomial. Then, we could compute the global minimizer of the step size $ \alpha_k$ (the exact line search) in each iteration as [ @HCD ]. However, we practice a cheaper inexact line search here. The initial step size of the adjacent iteration follows D | )\|^2,$$ whfre ${{\bf x}}_{k+1}(\alpha)$ is calculxted by (\[eq3-11\]).
Updatg rhe itxrate ${{\br x}}_{k+1}={{\bf x}}_{y+1}(\alpha_k)$. Choose an initial stxp suze $\bqr{\alpha}_{k+1}\in(0,\alpha_{\max}]$ for the next iteratiin. $k \tets k+1.$
Accocsing to Theorsl \[th3-5\], hor any constanj $\eta\in(0,2)$, thera exists a poshtkvz scalar $\tilde{\alpha}$ such that for alj $\alpha\on(0,\hilde{\alpha}]$, $$f({{\bf x}}_{k+1}(\ak[ha)) - f({{\bf x}}_k) \leq -\eta\alpha\|{{\bf g}}({{\bf x}}_k)\|^2.$$ Hencs, the clrvilinear search process is well-defined.
Now, we oresent a curvilinfar search qlgowuthm (ACSA) fofmally in Algorithm \[alf\] for the smallest generalized digenralue and ijs advociated eijenvecnor of a Hankck tensmr. If oir aim is to cpmpnte rhe largest generalizxd eigenvalue and itf associadeb eigenvector of a Habkwl tetsor, we ubly netd vo dhange reapectively (\[eq3-11\]) and (\[suf-eec\]) used in Steps 5 snq 6 of the ACSA algorytrm to $${{\bf x}}_{k+1}(\alpha) = \frac{1-\alpha^2\|{{\bf g}}({{\bf x}}_k)\|^2}{1+\alpva^2\|{{\br g}}({{\bf x}}_k)\|^2}{{\bf x}}_k
+\drac{2\alpha}{1+\alpha^2\|{{\bf g}}({{\bf d}}_k)\|^2}{{\bf g}}({{\bf v}}_k),$$ and $$f({{\bf x}}_{k+1}(\alpha_k)) \geq f({{\bf x}}_k) + \eta\alpha_k\|{{\bf g}}({{\bf x}}_n)\|^2.$$
When ghe Z-cigevcapue of a Hankel tensor is considered, we have ${\szl E}{{\nf x}}^m=\|{{\bf x}}\|^m=1$ and tme objective $f({{\bf x}})$ id s polynomial. Tfen, we cohld compute the gllbal migimizwr of the stel size $\alpha_k$ (the exact linw search) in vach iteration as [@HCD]. Kowever, we ure a chealer inexact line search hers. The initiwl step sjxe of the next igersthon follows D | )\|^2,$$ where ${{\bf x}}_{k+1}(\alpha)$ is calculated by the ${{\bf x}}_{k+1}={{\bf Choose an initial next $k \gets k+1.$ to Theorem \[th3-5\], any constant $\eta\in(0,2)$, there exists a scalar $\tilde{\alpha}$ such that for all $\alpha\in(0,\tilde{\alpha}]$, $$f({{\bf x}}_{k+1}(\alpha)) - f({{\bf x}}_k) \leq g}}({{\bf x}}_k)\|^2.$$ Hence, the curvilinear search process is well-defined. Now, we present a search (ACSA) in \[alg\] for the smallest generalized eigenvalue and its associated eigenvector of a Hankel tensor. If our is to compute the largest generalized eigenvalue and associated eigenvector of a tensor, we only need to respectively and (\[suf-dec\]) in 5 6 of the algorithm to $${{\bf x}}_{k+1}(\alpha) = \frac{1-\alpha^2\|{{\bf g}}({{\bf x}}_k)\|^2}{1+\alpha^2\|{{\bf g}}({{\bf x}}_k)\|^2}{{\bf x}}_k +\frac{2\alpha}{1+\alpha^2\|{{\bf g}}({{\bf x}}_k)\|^2}{{\bf g}}({{\bf x}}_k),$$ and x}}_{k+1}(\alpha_k)) \geq + \eta\alpha_k\|{{\bf x}}_k)\|^2.$$ the of a Hankel considered, we have ${\cal E}{{\bf x}}^m=\|{{\bf objective $f({{\bf x}})$ is a polynomial. Then, we compute the minimizer of the step size $\alpha_k$ exact line search) in each iteration as [@HCD]. we use a cheaper inexact line search here. The initial step size of the next D | )\|^2,$$ where ${{\bf x}}_{k+1}(\alpha)$ is calculateD by (\[eq3-11\]).
UpdatE the iTerAte ${{\Bf X}}_{k+1}={{\bf X}}_{k+1}(\alPha_k)$. Choose an inITial Step size $\bar{\alpha}_{k+1}\in(0,\alpHa_{\max}]$ FoR The nEXt IteraTion. $k \geTS k+1.$
aCCorDiNg To THeOReM \[th3-5\], foR anY constaNt $\eta\in(0,2)$, theRe eXiSts a positive SCaLar $\tilde{\alPha}$ Such that for aLl $\aLpha\in(0,\TiLde{\ALpha}]$, $$f({{\Bf x}}_{K+1}(\alphA)) - f({{\bf x}}_k) \LEq -\eta\aLpha\|{{\bf g}}({{\bf X}}_k)\|^2.$$ hEnce, thE CurviliNEAr SearCh process is well-deFInED.
Now, we present a CurvilInEAr SEArcH alGorithm (ACSa) fOrmalLY in AlgoRItHM \[ALg\] fOR the smallest gEneralized eIGenValue aNd Its ASsociaTed eiGeNVecTor of a HankeL tenSor. If our aIm is to COmpute tHE largesT generAliZed EigeNVaLuE anD iTS asSOcIatED eiGenvectoR oF a hankeL tenSOR, WE onlY neEd to ChangE respectively (\[Eq3-11\]) aNd (\[suF-Dec\]) Used iN StepS 5 and 6 Of The ACsA algoRithm To $${{\Bf x}}_{k+1}(\alpha) = \frac{1-\alPha^2\|{{\bF g}}({{\bf x}}_k)\|^2}{1+\alpHa^2\|{{\bF g}}({{\Bf x}}_K)\|^2}{{\bF x}}_k
+\frAC{2\alpha}{1+\AlpHa^2\|{{\bF g}}({{\bf x}}_k)\|^2}{{\bF g}}({{\bf x}}_k),$$ aND $$f({{\bF x}}_{K+1}(\ALPhA_k)) \geq f({{\bf x}}_k) + \eta\alpha_K\|{{\bF G}}({{\Bf X}}_k)\|^2.$$
When thE Z-eigeNVaLuE Of a HankeL tEnsOr is CONsideRed, wE HaVe ${\cal E}{{\bf X}}^m=\|{{\bf x}}\|^m=1$ ANd ThE objectIvE $f({{\bf x}})$ iS a PolYnoMial. THEn, we Could cOmpute thE globAL minimizer of thE Step size $\alpha_K$ (ThE EXaCT linE seArch) in each iTeraTIon aS [@HCD]. hOwEveR, We use A cheaPeR InEXact line search here. THe InitiaL step Size of the next Iteration fOLLOws D | )\|^2,$$ where ${{\bf x}}_ {k+1}(\alp ha)$iscal cu late d by (\[eq3-11\]).
Upd ate the iterate ${{\bf x}}_ {k + 1}={ { \b f x}} _{k+1}( \ al p h a_k )$ .Cho os e a n ini tia l stepsize $\bar {\a lp ha}_{k+1}\in ( 0, \alpha_{\m ax} ]$ for the n ext itera ti on. $k \g ets k+1. $
Acc o rdingto Theore m\ [th3-5 \ ], fora n ycons tant $\eta\in(0,2 ) $, there exists a posit iv e s c a lar $\ tilde{\alp ha }$ su c h thatf or a l l $ \ alpha\in(0,\t ilde{\alpha } ]$, $$f({ {\ bfx }}_{k+ 1}(\a lp h a)) - f({{\bfx}}_ k) \leq - \eta\a l pha\|{{ \ bf g}}( {{\bfx}} _k) \|^2 . $$ H enc e, the cu rvi l ine ar searc hpr ocess isw e l l -def ine d.
Now,we present acur vili n ear sear ch al gori th m (AC SA) fo rmall yin Algorithm \[ alg\ ] for the sm al les tgener a lizedeig env alue an d its a s soc ia t e d e igenvector of a Ha nk e l t ensor. I f oura im i s to comp ut e t he l a r gestgene r al ized eig envalu e a nd its as so ciated e ige nve ctoro f aHankel tensor, we o n ly need to cha n ge respective l y( \ [e q 3-11 \]) and (\[suf -dec \ ]) u sedi nSte p s 5 a nd 6of th e ACSA algorithm to$$ {{\bfx}}_{ k+1}(\alpha)= \frac{1- \ a l pha^2\|{ {\bf g} } ({{\bf x}}_k)\ |^2}{ 1+\alpha^2 \ |{{\bf g }}({{ \bf x}}_ k)\|^2}{{ \ b f x}}_k +\ f r ac {2\alpha}{1+\ a l pha^ 2\ |{{\bfg}} ({{\bfx}} _k) \|^ 2}{ {\ bf g}}({{ \bf x}}_ k) ,$ $an d $ $f({{ \ bf x}}_{ k+ 1}( \a lph a_k)) \geq f ({{\b f x} }_ k) + \ eta\alp h a_ k \ |{{\ bf g }}({ {\b fx}}_k )\|^ 2 .$$
Whenthe Z-eig env a lueof a Hankel tensor is co ns idered, we h ave ${\ca l E}{{\bfx}}^m=\|{{\bf x}}\|^m=1 $ and th e o bject ive$f({{\bfx}} )$ isa p o lynomi al. Th en, w ecou l d comp u t ethe g lobal mini m i zer of t he ste p size$\alpha_k$ (the ex a ctline search)ineach i te rat i on as[@ H CD] . However, we use a cheaper i n ex act line s e arc hhere. T he init ial s t ep size of the n ext itera ti on f o l low s D | )\|^2,$$ where_${{\bf x}}_{k+1}(\alpha)$_is calculated by (\[eq3-11\]).
Update_the iterate_${{\bf_x}}_{k+1}={{\bf x}}_{k+1}(\alpha_k)$._Choose_an initial step_size $\bar{\alpha}_{k+1}\in(0,\alpha_{\max}]$ for_the next iteration. $k_\gets k+1.$
According to_Theorem_\[th3-5\], for any constant $\eta\in(0,2)$, there exists a positive scalar $\tilde{\alpha}$ such that for_all_$\alpha\in(0,\tilde{\alpha}]$, $$f({{\bf_x}}_{k+1}(\alpha))_-_f({{\bf x}}_k) \leq -\eta\alpha\|{{\bf g}}({{\bf_x}}_k)\|^2.$$ Hence, the curvilinear search_process is_well-defined.
Now, we present a curvilinear search algorithm (ACSA)_formally_in Algorithm \[alg\]_for the smallest generalized eigenvalue and its associated eigenvector_of a Hankel tensor. If our_aim is to_compute_the_largest generalized eigenvalue and_its associated eigenvector of a Hankel_tensor, we only need to change_respectively (\[eq3-11\]) and (\[suf-dec\]) used in Steps_5 and 6 of the ACSA_algorithm to $${{\bf x}}_{k+1}(\alpha) =_\frac{1-\alpha^2\|{{\bf g}}({{\bf_x}}_k)\|^2}{1+\alpha^2\|{{\bf g}}({{\bf x}}_k)\|^2}{{\bf x}}_k
_ _ _ _+\frac{2\alpha}{1+\alpha^2\|{{\bf g}}({{\bf x}}_k)\|^2}{{\bf g}}({{\bf x}}_k),$$ and_$$f({{\bf x}}_{k+1}(\alpha_k)) \geq_f({{\bf x}}_k) + \eta\alpha_k\|{{\bf g}}({{\bf x}}_k)\|^2.$$
When_the_Z-eigenvalue of a_Hankel_tensor_is considered,_we have ${\cal_E}{{\bf_x}}^m=\|{{\bf x}}\|^m=1$_and_the objective $f({{\bf x}})$ is a_polynomial._Then, we could compute the global minimizer_of the step size_$\alpha_k$_(the exact line search)_in each iteration as [@HCD]._However, we use a cheaper inexact_line search_here. The_initial step size of the next iteration follows D |
\] and the fact that the operators, $\Pi_{A_\infty,\Phi_\infty}$, define continuous projections.
### Estimates for gauge transformations intertwining two pairs {#subsubsec:Estimates_gauge_transformations_intertwining_two_pairs}
We shall require the
\[lem:gauged\] Let $(X,g)$ be a closed, Riemannian, smooth manifold of dimension $d \geq 2$, and $G$ be a compact Lie group, $P$ be a smooth principal $G$-bundle over $X$, and $E = P\times_\varrho\EE$ be a smooth Hermitian vector bundle over $X$ defined by a finite-dimensional unitary representation, $\varrho: G \to \Aut_\CC(\EE)$, and $A_1$ is a $C^\infty$ reference connection on $P$, $q>d/2$ and $p$ obeys $d/2\leq p\leq q$, then there is a constant $C = C(g,p) \in [1,\infty)$ with the following significance. If $(A, \Phi)$ and $(A', \Phi')$ are $W^{1,q}$ pairs on $(P,E)$ and $u\in \Aut^{2,q}(P)$, then $$\begin{aligned}
\|u(A,\Phi) - u(A', \Phi')\|_{W_{A_1}^{1,p}(X)}
&\leq C\left(1 + \|u\|_{W_{A_1}^{2,p}(X)}\right) \|(A,\Phi) - (A', \Phi')\|_{W_{A_1}^{1,p}(X)},
\\
\|(A,\Phi) - (A', \Phi')\|_{W_{A_1}^{1,p}(X)}
&\leq C\left(1 + \|u\|_{W_{A_1}^{2,p}(X)}\right) \|u(A,\Phi) - u(A', \Phi')\|_{W_{A_1}^{1,p}(X)}.\end{aligned}$$ If in addition $p\geq 4/3$ when $d=2$ and $p' \in [1,\infty)$ is the dual Hölder exponent defined by $1/p+1/p'=1$, and $(a,\phi) \in W^{1,p' | \ ] and the fact that the operators, $ \Pi_{A_\infty,\Phi_\infty}$, define continuous projections.
# # # Estimates for bore transformation intertwining two pairs { # subsubsec: Estimates_gauge_transformations_intertwining_two_pairs }
We shall command the
\[lem: gauged\ ] get $ (X, g)$ be a closed, Riemannian, placid manifold paper of dimension $ d \geq 2 $, and $ G$ be a compact Lie group, $ P$ be a smooth principal $ G$-bundle over $ X$, and $ east = P\times_\varrho\EE$ be a smooth Hermitian vector bundle over $ X$ define by a finite - dimensional unitary theatrical performance, $ \varrho: G \to \Aut_\CC(\EE)$, and $ A_1 $ is a $ C^\infty$ reference association on $ P$, $ q > d/2 $ and $ p$ obey $ d/2\leq p\leq q$, then there is a constant $ C = C(g, p) \in [ 1,\infty)$ with the postdate significance. If $ (A, \Phi)$ and $ (ampere', \Phi')$ are $ W^{1,q}$ pairs on $ (phosphorus, E)$ and $ u\in \Aut^{2,q}(P)$, then $ $ \begin{aligned }
\|u(A,\Phi) - u(A', \Phi')\|_{W_{A_1}^{1,p}(X) }
& \leq C\left(1 + \|u\|_{W_{A_1}^{2,p}(X)}\right) \|(A,\Phi) - (A', \Phi')\|_{W_{A_1}^{1,p}(X) },
\\
\|(A,\Phi) - (adenine', \Phi')\|_{W_{A_1}^{1,p}(X) }
& \leq C\left(1 + \|u\|_{W_{A_1}^{2,p}(X)}\right) \|u(A,\Phi) - u(A', \Phi')\|_{W_{A_1}^{1,p}(X)}.\end{aligned}$$ If in addition $ p\geq 4/3 $ when $ d=2 $ and $ p' \in [ 1,\infty)$ is the dual Hölder advocate defined by $ 1 / p+1 / p'=1 $, and $ (a,\phi) \in W^{1,p' | \] anf the fact that the operxtors, $\Pi_{A_\infty,\Pku_\infty}$, defins continjous projections.
### Estimates flr gaugt transformations ivtertwinijg two pqirs {#wubsubsec:Estimates_ncuge_tdwnsfmcmations_intertwlning_two_paiss}
We shall reqgife the
\[lem:gauged\] Let $(X,g)$ be a closed, Riqmanniam, dmooth manifolq of qimehsion $d \geq 2$, and $G$ be a compact Lis group, $P$ be a smootn principal $G$-bundle over $X$, and $E = P\times_\varrho\EE$ be a smooty Hewnitian vectof bundle onzr $X$ defines by a finite-dimensional unitarh repxesentation, $\vqrrjm: G \to \Aut_\CR(\EE)$, anq $A_1$ is a $C^\innny$ refesence cpnnection on $P$, $q>d/2$ ane $p$ obeys $d/2\leq p\leq q$, then there is a cjnstant $C = E(g,p) \in [1,\infty)$ with the fillowhng vignkdicxnct. Ih $(A, \Phi)$ ajd $(E', \Phi')$ are $W^{1,s}$ pairs on $(P,E)$ and $u\in \Aut^{2,q}(P)$, thtn $$\fvbin{aligned}
\|u(A,\Lhi) - u(W', \[hi')\|_{W_{A_1}^{1,p}(X)}
&\leq C\left(1 + \|u\|_{W_{A_1}^{2,p}(X)}\right) \|(A,\Phi) - (A', \Php')\|_{W_{A_1}^{1,p}(S)},
\\
\|(A,\Phi) - (A', \Phi')\|_{W_{A_1}^{1,p}(X)}
&\leq C\ledt(1 + \|u\|_{W_{A_1}^{2,p}(X)}\right) \|u(A,\Phi) - u(A', \Phi')\|_{W_{A_1}^{1,[}(X)}.\end{aligned}$$ If in addition $p\geq 4/3$ when $d=2$ and $p' \in [1,\inftb)$ ks uhc ayap Hölder exponent defined by $1/p+1/p'=1$, and $(a,\phi) \in W^{1,p' | \] and the fact that the operators, continuous ### Estimates gauge transformations intertwining require \[lem:gauged\] Let $(X,g)$ a closed, Riemannian, manifold of dimension $d \geq 2$, $G$ be a compact Lie group, $P$ be a smooth principal $G$-bundle over and $E = P\times_\varrho\EE$ be a smooth Hermitian vector bundle over $X$ defined a unitary $\varrho: \to \Aut_\CC(\EE)$, and $A_1$ is a $C^\infty$ reference connection on $P$, $q>d/2$ and $p$ obeys $d/2\leq q$, then there is a constant $C = \in [1,\infty)$ with the significance. If $(A, \Phi)$ and \Phi')$ $W^{1,q}$ pairs $(P,E)$ $u\in then $$\begin{aligned} \|u(A,\Phi) u(A', \Phi')\|_{W_{A_1}^{1,p}(X)} &\leq C\left(1 + \|u\|_{W_{A_1}^{2,p}(X)}\right) \|(A,\Phi) - (A', \Phi')\|_{W_{A_1}^{1,p}(X)}, \\ \|(A,\Phi) - (A', \Phi')\|_{W_{A_1}^{1,p}(X)} &\leq C\left(1 \|u\|_{W_{A_1}^{2,p}(X)}\right) \|u(A,\Phi) \Phi')\|_{W_{A_1}^{1,p}(X)}.\end{aligned}$$ If addition 4/3$ $d=2$ and $p' is the dual Hölder exponent defined $(a,\phi) \in W^{1,p' | \] and the fact that the operatorS, $\Pi_{A_\infty,\PHi_\infTy}$, dEfiNe ContInuoUs projections.
### ESTimaTes for gauge transformatIons iNtERtwiNInG two pAirs {#subSUbSEC:EsTiMaTes_GaUGe_TransForMations_IntertwiniNg_tWo_Pairs}
We shall REqUire the
\[lem:GauGed\] Let $(X,g)$ be a cLosEd, RiemAnNiaN, SmootH maNifolD of dimENsion $d \Geq 2$, and $G$ be A cOMpact LIE group, $P$ BE A sMootH principal $G$-bundle OVeR $x$, and $E = P\times_\varRho\EE$ bE a SMoOTH HeRmiTian vector BuNdle oVEr $X$ defiNEd BY A FinITe-dimensional Unitary reprESenTation, $\VaRrhO: g \to \Aut_\cC(\EE)$, aNd $a_1$ Is a $c^\infty$ referEnce ConnectioN on $P$, $q>d/2$ ANd $p$ obeyS $D/2\leq p\leQ q$, then TheRe iS a coNStAnT $C = C(G,p) \IN [1,\inFTy)$ WitH The FollowinG sIgNificAnce. iF $(a, \pHi)$ anD $(A', \PHi')$ arE $W^{1,q}$ paIrs on $(P,E)$ and $u\in \aut^{2,Q}(P)$, thEN $$\beGin{alIgned}
\|U(A,\PhI) - u(a', \Phi')\|_{W_{a_1}^{1,p}(X)}
&\leq c\left(1 + \|U\|_{W_{a_1}^{2,p}(X)}\right) \|(A,\Phi) - (A', \PhI')\|_{W_{A_1}^{1,p}(x)},
\\
\|(A,\Phi) - (A', \Phi')\|_{w_{A_1}^{1,p}(x)}
&\lEq C\LeFt(1 + \|u\|_{W_{A_1}^{2,P}(x)}\right) \|U(A,\PHi) - u(a', \Phi')\|_{W_{A_1}^{1,p}(x)}.\end{aliGNed}$$ if IN ADdItion $p\geq 4/3$ when $d=2$ and $p' \In [1,\INFtY)$ is the duAl HöldER eXpONent defiNeD by $1/P+1/p'=1$, anD $(A,\Phi) \in w^{1,p' | \] and the fact that the o perators,$\Pi_ {A_ \in ft y,\P hi_\ infty}$, defin e con tinuous projections.
### E st i mate s f or ga uge tra n sf o r mat io ns in te r tw ining tw o pairs {#subsubs ec: Es timates_gaug e _t ransformat ion s_intertwini ng_ two_pa ir s}We sh all requ ire th e
\[le m:gauged\ ]L et $(X , g)$ bea cl osed , Riemannian, smo o th manifold of di mensio n$ d\ g eq2$, and $G$ b ea com p act Lie gr o u p , $ P $ be a smooth principal$ G$- bundle o ver $X$, a nd $E = P\t imes_\varrh o\EE $ be a sm ooth H e rmitian vectorbundle ov er$X$d ef in edby a f i ni te- d ime nsionalun it ary r epre s e n t atio n,$\va rrho: G \to \Aut_\ CC( \EE) $ , a nd $A _1$ i s a$C ^\inf ty$ re feren ce connection on$P$, $q>d/2$and $ p$ob eys $ d /2\leq p\ leq q$, th en ther e is a c o ns tant $C = C(g,p) \ in [ 1, \infty)$ witht he f o llowingsi gni fica n c e. If $(A , \ Phi)$ an d $(A' , \ Ph i')$ ar e$W^{1, q} $ p air s on$ (P,E )$ and $u\in \ Aut^{ 2 ,q}(P)$, then$ $\begin{align e d} \ |u ( A,\P hi) - u(A', \P hi') \ |_{W _{A_ 1 }^ {1, p }(X)}
&\le qC \l e ft(1 + \|u\|_{W_{A_ 1} ^{2,p} (X)}\ right) \|(A,\ Phi) - (A' , \ Phi')\|_ {W_{ A _1 } ^{1,p}(X)},
\\
\|(A ,\Phi) - ( A ', \Phi' )\|_{ W_{A_1}^ {1,p}(X)} & \leq C\l eft (1+ \ |u\ | _ {W _{A_1}^{2,p}( X ) }\ri gh t) \|u( A,\ Phi) -u(A ',\Ph i') \| _{W_{A_1} ^{1,p}(X )} .\ en d{ ali gned} $ $ If inad dit io n $ p\geq 4/3$ w hen $ d=2$ a nd $p' \in [1 , \i n f ty)$ i sthedua lHölde r ex p one nt defi ned by $1 /p+ 1 /p'= 1$ ,and $(a ,\phi) \in W^ {1 ,p' | \] and_the fact_that the operators, $\Pi_{A_\infty,\Phi_\infty}$,_define continuous_projections.
###_Estimates for_gauge_transformations intertwining two_pairs {#subsubsec:Estimates_gauge_transformations_intertwining_two_pairs}
We shall_require the
\[lem:gauged\] Let $(X,g)$_be a closed,_Riemannian,_smooth manifold of dimension $d \geq 2$, and $G$ be a compact Lie group,_$P$_be a_smooth_principal_$G$-bundle over $X$, and $E_= P\times_\varrho\EE$ be a smooth_Hermitian vector_bundle over $X$ defined by a finite-dimensional unitary_representation,_$\varrho: G \to_\Aut_\CC(\EE)$, and $A_1$ is a $C^\infty$ reference connection on_$P$, $q>d/2$ and $p$ obeys $d/2\leq_p\leq q$, then_there_is_a constant $C =_C(g,p) \in [1,\infty)$ with the following_significance. If $(A, \Phi)$ and $(A',_\Phi')$ are $W^{1,q}$ pairs on $(P,E)$ and_$u\in \Aut^{2,q}(P)$, then $$\begin{aligned}
\|u(A,\Phi) - u(A',_\Phi')\|_{W_{A_1}^{1,p}(X)}
&\leq C\left(1 + \|u\|_{W_{A_1}^{2,p}(X)}\right) \|(A,\Phi)_- (A',_\Phi')\|_{W_{A_1}^{1,p}(X)},
\\
\|(A,\Phi) - (A', \Phi')\|_{W_{A_1}^{1,p}(X)}
&\leq C\left(1_+ \|u\|_{W_{A_1}^{2,p}(X)}\right) \|u(A,\Phi)_- u(A',_\Phi')\|_{W_{A_1}^{1,p}(X)}.\end{aligned}$$ If in_addition $p\geq 4/3$ when $d=2$ and_$p' \in [1,\infty)$_is the dual Hölder exponent defined_by_$1/p+1/p'=1$, and $(a,\phi)_\in_W^{1,p' |
2.75/2.75. The supply voltage for a gain of $10^6$ is approximately 1750 V.
=3.40in =1.70in
The PMT resistive base assembly is linear to within $\sim 2\%$ up to the peak anode current of 120 $\mu$A ($\sim 5\times 10^4$ pe). The dark current is typically less than 3 nA. The base has anode and dynode output signals. Channel-to-channel adjustable high voltages are provided by a system of CAEN SY-403 high voltage power supplies (64 channel, $V_{max}$ = 3.0 kV, $I_{max}= 3.0~mA$).
Monte Carlo simulation codes {#early_mc}
============================
The first versions of simulation codes for the HMS/SOS calorimeters were based on the ELSS [@ELSS] and EGS4 [@EGS4] packages for simulations of electromagnetic showers. Dedicated code was added for Čerenkov light generation, optical photon tracing and photoelectron knockout from PMT photocathodes. The optics took into account light absorption in the lead glass, reflections from the block sides, and passage through the optical coupling to the PMT photocathode. However, the software did not take into account block to block variations of lead glass absorption length and electronic effects. The first simulations revealed sufficient signal ($\sim$900 photoelectrons from a 1 GeV incident electron), good linearity and reasonable resolution in the GeV range for the calorimeter designs.
Subsequent simulations of HMS calorimeter are based on the GEANT4 package, version 9.1. The QGSP\_BERT physics list [@qgsp-bert] was chosen to model hadron interactions, which is recommended by the GEANT4 developers for high energy physics calorimetry [@geant-www]. This list includes the parton string model [@parton-string] at energies above 12 GeV, intra-nuclear Bertini cascade [@Bertini] below 9.9 GeV, and a nuclear evaporation model [@nucl-evap] at low energies. The GHEISHA model [@gheisha] is used at energies 9.5 – 25 GeV. Electromagnetic processes are modeled to good accuracy within the framework of the GEANT4 standard electromagnetic package.
The code closely emulates the geometry and the composition of the detector. Particularly, | 2.75/2.75. The supply voltage for a gain of $ 10 ^ 6 $ is approximately 1750 V.
= 3.40 in = 1.70 in
The PMT resistive basis fabrication is linear to within $ \sim 2\%$ up to the peak anode current of 120 $ \mu$A ($ \sim 5\times 10 ^ 4 $ pe). The benighted current is typically less than 3 nA. The base have anode and dynode output signals. Channel - to - channel adjustable high electric potential are provided by a system of CAEN SY-403 gamey electric potential power supplies (64 distribution channel, $ V_{max}$ = 3.0 kV, $ I_{max}= 3.0 ~ mA$).
Monte Carlo simulation codes { # early_mc }
= = = = = = = = = = = = = = = = = = = = = = = = = = = =
The inaugural versions of simulation codes for the HMS / SOS calorimeter were based on the ELSS [ @ELSS ] and EGS4 [ @EGS4 ] packages for simulation of electromagnetic showers. consecrated code was added for Čerenkov abstemious generation, optical photon tracing and photoelectron knockout from PMT photocathode. The optics took into account light absorption in the lead glass, observation from the block sides, and passage through the optical coupling to the PMT photocathode. However, the software did not take into account block to block variations of lead glass assimilation length and electronic effects. The beginning model revealed sufficient signal ($ \sim$900 photoelectrons from a 1 GeV incident electron), good one-dimensionality and reasonable settlement in the GeV range for the calorimeter designs.
Subsequent model of HMS calorimeter are based on the GEANT4 package, version 9.1. The QGSP\_BERT physics list [ @qgsp - bert ] was chosen to model hadron interaction, which is recommended by the GEANT4 developers for high energy purgative calorimetry [ @geant - www ]. This tilt includes the parton string model [ @parton - string ] at energies above 12 GeV, intra - nuclear Bertini cascade [ @Bertini ] below 9.9 GeV, and a nuclear evaporation exemplar [ @nucl - evap ] at low energies. The GHEISHA model [ @gheisha ] is use at energies 9.5 – 25 GeV. Electromagnetic processes are modeled to good accuracy within the model of the GEANT4 standard electromagnetic package.
The code closely emulates the geometry and the composition of the detector. Particularly, | 2.75/2.75. Thf supply voltage for a gxin of $10^6$ is approximateny 1750 V.
=3.40ih =1.70in
The OMT resistive base assembly ms lunear to within $\sim 2\%$ up to ghe peak wnode cuerenu of 120 $\mu$A ($\sim 5\timxa 10^4$ pe). Tmz dari curxeit is typically less than 3 nA. The base var cnode and dynode output signals. Changel-to-chsnjel adjustable hign volfages are provided by a system of DAEN SY-403 high voltage power supplies (64 channel, $V_{lax}$ = 3.0 kV, $I_{max}= 3.0~mA$).
Monte Farlo simulqtiog codes {#early_oc}
============================
The firsu rersions of simulation codes for the HMS/SOR calprimeters qeee hdsed on the ELSS [@VLSS] and EGS4 [@ENX4] packdges fot simulations pf xlecrromagnetic showers. Dxdicated code was adqed for Česeukov light generation, optican phmton rrazinf end photoflertron knockkut from PMR photocathodes. The o[nocs took intk accotne light absorption in the lead glass, rexledtions from the block sudes, and passage throogh the opeical coupling to the PMT photocathode. However, tha sofvwxre dld nur hake into account block to block variations os ltad glass absorptlon length and elevtgomyc effects. Thg first sijulations revealed sufficyent wignal ($\sii$900 phptoelectrons from a 1 GeV inxident electgon), tood linearity and reasonable resplutipn in the GeV range for the calorimeteg designs.
Ajbsequent simulagioks mf HMS calorimeter are basqd on the GEAUT4 packaee, vgrsion 9.1. The QGSP\_BFRT pmfsics list [@qgsp-bert] was ehoset to model hadron interactions, which is rxrommended by jhe GEDNT4 devejoperx for high enqrgy physics cclorimetxy [@geavt-www]. This list iicludes the [arton string lodel [@parton-svring] at qnerties above 12 EeV, intra-nuclesr Bertinp eascade [@Beetini] below 9.9 GeV, akd a vhclear evaporatnub model [@nucl-evap] at lor vnecgies. Dhe GHEISHA kodeu [@ghdosha] ks used at cnefgiex 9.5 – 25 GeV. Electromagnatic processes are modrlcd to gooe accurasy within the framework of the HEANT4 stendard elgctromagnetic package.
The code cmosely emklabes the geomeery qnd the comppsition of the detector. Particularly, | 2.75/2.75. The supply voltage for a gain is 1750 V. =1.70in The PMT to $\sim 2\%$ up the peak anode of 120 $\mu$A ($\sim 5\times 10^4$ The dark current is typically less than 3 nA. The base has anode dynode output signals. Channel-to-channel adjustable high voltages are provided by a system of SY-403 voltage supplies channel, $V_{max}$ = 3.0 kV, $I_{max}= 3.0~mA$). Monte Carlo simulation codes {#early_mc} ============================ The first versions simulation codes for the HMS/SOS calorimeters were based the ELSS [@ELSS] and [@EGS4] packages for simulations of showers. code was for light optical photon tracing photoelectron knockout from PMT photocathodes. The optics took into account light absorption in the lead glass, reflections the block passage through optical to PMT photocathode. However, did not take into account block of lead glass absorption length and electronic effects. first simulations sufficient signal ($\sim$900 photoelectrons from a GeV incident electron), good linearity and reasonable resolution the GeV range for the calorimeter designs. Subsequent simulations of HMS calorimeter are based on package, version 9.1. The physics list [@qgsp-bert] chosen model interactions, is recommended the GEANT4 developers for high energy physics calorimetry [@geant-www]. This list the parton string model [@parton-string] at energies above 12 GeV, cascade below 9.9 GeV, a nuclear evaporation model at energies. The GHEISHA model used energies GeV. processes modeled to good accuracy the framework of the GEANT4 electromagnetic package. The code the composition of the detector. Particularly, | 2.75/2.75. The supply voltage for a gain oF $10^6$ is approxiMatelY 1750 V.
=3.40iN =1.70in
thE PMT ResiStive base assemBLy is Linear to within $\sim 2\%$ up to tHe peaK aNOde cURrEnt of 120 $\Mu$A ($\sim 5\tIMeS 10^4$ PE). ThE dArK cuRrENt Is typIcaLly less Than 3 nA. The bAse HaS anode and dynODe Output signAls. channel-to-chaNneL adjusTaBle HIgh voLtaGes arE proviDEd by a sYstem of CAeN sy-403 high vOLtage poWER sUpplIes (64 channel, $V_{max}$ = 3.0 kV, $I_{MAx}= 3.0~Ma$).
Monte Carlo simUlatioN cODeS {#EArlY_mc}
============================
the first veRsIons oF SimulatIOn CODEs fOR the HMS/SOS calOrimeters weRE baSed on tHe eLSs [@eLSS] anD EGS4 [@EgS4] PAckAges for simuLatiOns of elecTromagNEtic shoWErs. DediCated cOde Was AddeD FoR ČEreNkOV liGHt GenERatIon, opticAl PhOton tRaciNG AND phoToeLectRon knOckout from PMT PhoTocaTHodEs. The OpticS tooK iNto acCount lIght aBsOrption in the leaD glaSs, reflectIonS fRom ThE blocK Sides, aNd pAssAge throUgh the oPTicAl COUPlIng to the PMT photocaThODE. HOwever, thE softwARe DiD Not take iNtO acCounT BLock tO bloCK vAriationS of leaD GlAsS absorpTiOn lengTh And EleCtronIC effEcts. ThE first siMulatIOns revealed sufFIcient signal ($\sIM$900 pHOToELectRonS from a 1 GeV inCideNT eleCtroN), GoOd lINeariTy and ReASoNAble resolution in the gev range For thE calorimeter dEsigns.
SubsEQUEnt simulAtioNS oF hMS calorimeter Are baSed on the GEanT4 packagE, versIon 9.1. The QGsP\_BERT phySICs list [@qgSp-bErt] Was ChoSEN tO model hadron iNTEracTiOns, whicH is RecommeNdeD by The gEAnT4 DeveloperS for high EnErGy PhYsiCs calORimetry [@gEaNt-wWw]. thiS list INcludeS the pArtoN sTrINg mOdel [@parTOn-STRing] At EnErgiEs aBoVe 12 GeV, IntrA-NucLear BerTini cascaDe [@BERtinI] bElOw 9.9 GeV, anD a nuclear evapOrAtion model [@NuCl-eVap] at lOW Energies. the GHEISHA model [@gheisha] iS Used at eNerGies 9.5 – 25 GEV. ElEctromagnEtiC proceSseS Are modEled to Good aCcUraCY WithiN THe FraMeWork of the GeanT4 sTandaRd ElecTromagnEtic package.
The code CLosEly emulates thE geOmetRY AnD thE CoMPosItIOn oF THe detector. PartiCularly, | 2.75/2.75. The supply volt age for againof$10 ^6 $ is app roximately 175 0 V.
=3.40in =1.70in
ThePMT r es i stiv e b ase a ssembly is l ine ar t o w it h in $\si m 2 \%$ upto the pea k a no de current o f 1 20 $\mu$A($\ sim 5\times10^ 4$ pe) .The darkcur rentis typ i callyless than 3 nA. Th e base h a s a node and dynode outpu t s i gnals. Channel -to-ch an n el a dju sta ble high v ol tages are pro v id e d bya system of CA EN SY-403 h i ghvoltag epow e r supp lies(6 4 ch annel, $V_{ max} $ = 3.0 k V, $I_ { max}= 3 . 0~mA$).
Mont e C arl o si m ul at ion c o des {# ear l y_m c}
===== == == ===== ==== = = = = ==== ==
The firs t versions of si mula t ion code s for the H MS/SO S calo rimet er s were based on the ELSS [@E LSS ]and E GS4 [ @ EGS4]pac kag es forsimulat i ons o f e le ctromagnetic showe rs . De dicatedcode w a sad d ed for Č er enk ov l i g ht ge nera t io n, optic al pho t on t racingan d phot oe lec tro n kno c kout fromPMT phot ocath o des. The optic s took into ac c ou n t l i ghtabs orption inthel eadglas s ,ref l ectio ns fr om th e block sides, and p as sage t hroug h the optical couplingt o the PMTphot o ca t hode. However, thesoftware d i d not ta ke in to accou nt blockt o block v ari ati ons of l ea d glass absor p t ionle ngth an d e lectron iceff ect s.Th e first s imulatio ns r ev ea led suff i cient si gn al($ \si m$900 photoe lectr onsfr om a 1 GeV in c id e n t el ec tr on), go od line arit y an d reaso nable res olu t ionin t he GeVrange for the c alorimeter d esi gns.
S u bsequent simulations of HMS cal o rimeter ar e bas ed o n the GEA NT4 packa ge, versio n 9.1. TheQG SP\ _ B ERT p h y si csli st [@qgsp- b e rt] wasch osen to mod el hadron interact i ons , which is re com mend e d b y t h eG EAN T4 dev e l opers for highenergy phy si c scalorimetr y [@ ge ant-www ]. This list include s the par ton strin gmode l [@p arton-stri ng] at e nergies a b ove 1 2 G eV, i ntr a-nucl ea r B ertin i casc a de[@Ber tini]be low 9. 9 GeV ,and a nu clear evaporation model [@nuc l-eva p]at low en erg i es. The GHEI SHAmodel [@gh eis ha] is u sed at en ergi e s9.5 – 25GeV. Electroma g ne tic p ro cesses arem o d ele d togoo d accur acywithin the framew o rk of the GEAN T4 s t a nda rde lect ro magnetic packa ge.
T h e code c lo sely emulat es the g eo m etryand th e comp osition o ft he det ecto r.Particula rly , | 2.75/2.75. The_supply voltage_for a gain of_$10^6$ is_approximately_1750 V.
=3.40in_=1.70in
The_PMT resistive base_assembly is linear_to within $\sim 2\%$_up to the_peak_anode current of 120 $\mu$A ($\sim 5\times 10^4$ pe). The dark current is typically_less_than 3_nA._The_base has anode and dynode_output signals. Channel-to-channel adjustable high_voltages are_provided by a system of CAEN SY-403 high_voltage_power supplies (64_channel, $V_{max}$ = 3.0 kV, $I_{max}= 3.0~mA$).
Monte Carlo simulation_codes {#early_mc}
============================
The first versions of simulation_codes for the_HMS/SOS_calorimeters_were based on the_ELSS [@ELSS] and EGS4 [@EGS4] packages for simulations_of electromagnetic showers. Dedicated code was_added for Čerenkov light generation, optical photon_tracing and photoelectron knockout from PMT_photocathodes. The optics took into_account light_absorption in the lead glass,_reflections from the_block sides,_and passage through_the optical coupling to the PMT_photocathode. However, the_software did not take into account_block_to block variations_of_lead_glass absorption_length and electronic_effects._The first_simulations_revealed sufficient signal ($\sim$900 photoelectrons from_a_1 GeV incident electron), good linearity and_reasonable resolution in the_GeV_range for the calorimeter_designs.
Subsequent simulations of HMS calorimeter_are based on the GEANT4 package,_version 9.1._The QGSP\_BERT_physics list [@qgsp-bert] was chosen to model hadron interactions, which is recommended_by the GEANT4 developers for high_energy physics calorimetry [@geant-www]._This list_includes_the parton string_model [@parton-string]_at energies_above 12 GeV, intra-nuclear Bertini cascade [@Bertini] below_9.9 GeV,_and a nuclear evaporation model [@nucl-evap] at_low energies. The GHEISHA_model [@gheisha]_is used at energies 9.5 –_25 GeV. Electromagnetic processes are modeled_to good accuracy within the_framework_of_the GEANT4 standard electromagnetic package.
The_code closely emulates the geometry and_the composition of_the detector. Particularly, |
From the equations in (\[E:opt-p\]), we can solve the optimal $\mathbf{p}^\star\triangleq [p_1^\star, \dots, p_{K_P}^\star]^T$ as $$\label{E:PA}
\mathbf{p}^\star = \mathbf{M}^{-1} \boldsymbol{\zeta},$$ where $\boldsymbol{\zeta} = [\zeta_1(\mathbf{W}_f^\star), \dots, \zeta_{K_P}(\mathbf{W}_f^\star)]^T$, and $\mathbf{M}\in\mathbb{C}^{{K_P}\times {K_P}}$ is defined as $$[\mathbf{M}]_{kj} = \left\{
\begin{array}{ll}
\frac{1}{\gamma_k}|\mathbf{h}_{Pk}^H\tilde{\mathbf{w}}_{d,k}^\star|^2, & k = j,\\
- |\mathbf{h}_{Pk}^H\tilde{\mathbf{w}}_{t,j}^\star|^2, & k \neq j.
\end{array}
\right.$$
Finally, we summarize the proposed low-complexity algorithm to solve the original optimization problem (\[E:problem-MU\]) in general multiuser case in Table \[tab:Distributed-P-B-allocation\].
------------------------------------------------------------------------
[aaa]{} **Initialization:** Set $i=0$, $\underline{R_{sum}}=0$, and $\overline{R_{sum}} = \sum_{k=1}^{K_P} \log(1 +\mathtt{SINR}_k^{ub})$, where $\mathtt{SINR}_k^{ub} = \frac{P_0\|\mathbf{h}_{Pk}\|^2}{\sigma_n^2}$ is an upper bound of the SINR of ${\text{PUE}}_k$, which is achieved when the ICI disappears and only ${\text{PUE}}_k$ is served by the PBS.\
**Bisection Iteration:** At the $i$-th iteration, set $i \leftarrow i + 1$.
- Compute $R_{sum}^0 = \frac{\underline{R_{sum}} + \overline{R_{sum}}}{2}$.
- Given $R_{sum}^0$, compute $\{\bar{\lambda}_k | From the equations in (\[E: opt - p\ ]), we can solve the optimal $ \mathbf{p}^\star\triangleq [ p_1^\star, \dots, p_{K_P}^\star]^T$ as $ $ \label{E: PA }
\mathbf{p}^\star = \mathbf{M}^{-1 } \boldsymbol{\zeta},$$ where $ \boldsymbol{\zeta } = [ \zeta_1(\mathbf{W}_f^\star), \dots, \zeta_{K_P}(\mathbf{W}_f^\star)]^T$, and $ \mathbf{M}\in\mathbb{C}^{{K_P}\times { K_P}}$ is defined as $ $ [ \mathbf{M}]_{kj } = \left\ {
\begin{array}{ll }
\frac{1}{\gamma_k}|\mathbf{h}_{Pk}^H\tilde{\mathbf{w}}_{d, k}^\star|^2, & k = j,\\
- |\mathbf{h}_{Pk}^H\tilde{\mathbf{w}}_{t, j}^\star|^2, & k \neq j.
\end{array }
\right.$$
ultimately, we sum up the proposed low - complexity algorithm to resolve the original optimization trouble (\[E: problem - MU\ ]) in general multiuser case in Table \[tab: circulate - P - B - allocation\ ].
------------------------------------------------------------------------
[ aaa ] { } * * low-level formatting :* * rig $ i=0 $, $ \underline{R_{sum}}=0 $, and $ \overline{R_{sum } } = \sum_{k=1}^{K_P } \log(1 + \mathtt{SINR}_k^{ub})$, where $ \mathtt{SINR}_k^{ub } = \frac{P_0\|\mathbf{h}_{Pk}\|^2}{\sigma_n^2}$ is an upper bound of the SINR of $ { \text{PUE}}_k$, which is achieve when the ICI disappears and only $ { \text{PUE}}_k$ is served by the PBS.\
* * Bisection Iteration :* * At the $ i$-th iteration, set $ i \leftarrow i + 1$.
- Compute $ R_{sum}^0 = \frac{\underline{R_{sum } } + \overline{R_{sum}}}{2}$.
- give $ R_{sum}^0 $, compute $ \{\bar{\lambda}_k |
Frol the equations in (\[E:opt-p\]), we can solve tkw optikal $\mafhbf{p}^\staf\triangleq [p_1^\star, \dots, p_{K_P}^\stac]^T$ aw $$\labtj{E:PA}
\mathbf{p}^\star = \mathbf{M}^{-1} \holdsymbil{\zeua},$$ where $\boldsymbol{\zeta} = [\dzta_1(\mafmbf{W}_f^\vvar), \dots, \zeta_{K_P}(\kathbf{W}_f^\stdr)]^T$, and $\mathbf{K}\iv\mcthbb{C}^{{K_P}\times {K_P}}$ is defined as $$[\mathbs{M}]_{kj} = \lrfh\{
\begin{artay}{ll}
\frac{1}{\gamma_k}|\mathbf{h}_{Pk}^H\tilde{\mathbf{s}}_{d,k}^\star|^2, & k = j,\\
- |\mathbf{h}_{Pk}^H\tilde{\mathbf{w}}_{t,u}^\stag|^2, & k \neq j.
\enf{array}
\rigyt.$$
Figqlly, we summxrize the krmposed low-domplexity algorithm to solve tfe ornginal optinizatlmn problem (\[X:problvm-MU\]) in general multigser caxe in Table \[tab:Cisvribyted-P-B-allocation\].
------------------------------------------------------------------------
[aaa]{} **Iiitialization:** Set $i=0$, $\onderline{R_{vuj}}=0$, and $\overline{R_{sun}} = \sum_{k=1}^{N_P} \lmg(1 +\mxrhtg{SIHR}_l^{ug})$, wherf $\methtt{SINR}_k^{ug} = \frac{P_0\|\matybf{h}_{Pk}\|^2}{\sigma_n^2}$ is an ip[vt bound of tge SINW jf ${\text{PUE}}_k$, which is achieved when the PCI sisappears and only ${\texr{PUE}}_k$ is served by thg PBS.\
**Biseceion Iteration:** At the $i$-th iteration, set $i \leftarrmw i + 1$.
- Eimputd $R_{dum}^0 = \frac{\underline{R_{sum}} + \overline{R_{sum}}}{2}$.
- Given $W_{auk}^0$, bompute $\{\bar{\lambda}_h | From the equations in (\[E:opt-p\]), we can optimal [p_1^\star, \dots, as $$\label{E:PA} \mathbf{p}^\star = \dots, \zeta_{K_P}(\mathbf{W}_f^\star)]^T$, and {K_P}}$ is defined $$[\mathbf{M}]_{kj} = \left\{ \begin{array}{ll} \frac{1}{\gamma_k}|\mathbf{h}_{Pk}^H\tilde{\mathbf{w}}_{d,k}^\star|^2, & = j,\\ - |\mathbf{h}_{Pk}^H\tilde{\mathbf{w}}_{t,j}^\star|^2, & k \neq j. \end{array} \right.$$ Finally, we summarize proposed low-complexity algorithm to solve the original optimization problem (\[E:problem-MU\]) in general multiuser in \[tab:Distributed-P-B-allocation\]. [aaa]{} Set $i=0$, $\underline{R_{sum}}=0$, and $\overline{R_{sum}} = \sum_{k=1}^{K_P} \log(1 +\mathtt{SINR}_k^{ub})$, where $\mathtt{SINR}_k^{ub} = \frac{P_0\|\mathbf{h}_{Pk}\|^2}{\sigma_n^2}$ is an upper of the SINR of ${\text{PUE}}_k$, which is achieved the ICI disappears and ${\text{PUE}}_k$ is served by the **Bisection At the iteration, $i i + 1$. Compute $R_{sum}^0 = \frac{\underline{R_{sum}} + \overline{R_{sum}}}{2}$. - Given $R_{sum}^0$, compute $\{\bar{\lambda}_k |
From the equations in (\[E:opt-p\]), we Can solve thE optiMal $\MatHbF{p}^\stAr\trIangleq [p_1^\star, \doTS, p_{K_P}^\Star]^T$ as $$\label{E:PA}
\mathbf{p}^\Star = \mAtHBf{M}^{-1} \bOLdSymboL{\zeta},$$ whERe $\BOLdsYmBoL{\zeTa} = [\ZEtA_1(\mathBf{W}_F^\star), \doTs, \zeta_{K_P}(\maThbF{W}_F^\star)]^T$, and $\matHBf{m}\in\mathbb{C}^{{k_P}\tImes {K_P}}$ is defiNed As $$[\mathBf{m}]_{kj} = \LEft\{
\beGin{Array}{Ll}
\frac{1}{\GAmma_k}|\mAthbf{h}_{Pk}^H\TiLDe{\mathBF{w}}_{d,k}^\staR|^2, & K = J,\\
- |\mAthbF{h}_{Pk}^H\tilde{\mathbf{w}}_{T,J}^\sTAr|^2, & k \neq j.
\end{arraY}
\right.$$
fiNAlLY, We sUmmArize the prOpOsed lOW-compleXItY ALGorIThm to solve the Original optIMizAtion pRoBleM (\[e:problEm-MU\]) iN gENerAl multiuser Case In Table \[taB:DistrIButed-P-B-ALlocatiOn\].
------------------------------------------------------------------------
[aaa]{} **INitIalIzatIOn:** seT $i=0$, $\uNdERliNE{R_{Sum}}=0$, ANd $\oVerline{R_{SuM}} = \sUm_{k=1}^{K_P} \Log(1 +\mATHTT{SINr}_k^{uB})$, wheRe $\matHtt{SINR}_k^{ub} = \fraC{P_0\|\mAthbF{H}_{Pk}\|^2}{\Sigma_N^2}$ is an UppeR bOund oF the SInR of ${\tExT{PUE}}_k$, which is achIeveD when the IcI dIsAppEaRs and ONly ${\texT{PUe}}_k$ iS served By the PBs.\
**bisEcTION ITeration:** At the $i$-th itErATIoN, set $i \lefTarrow I + 1$.
- coMpUTe $R_{sum}^0 = \frAc{\UndErliNE{r_{sum}} + \oVerlINe{r_{sum}}}{2}$.
- GiveN $R_{sum}^0$, cOMpUtE $\{\bar{\lamBdA}_k |
From the equations in (\[ E:opt-p\]) , wecan so lv e th e op timal $\mathbf { p}^\ star\triangleq [p_1^\s tar,\d o ts,p _{ K_P}^ \star]^ T $a s $$ \l ab el{ E: P A}
\m ath bf{p}^\ star = \ma thb f{ M}^{-1} \bol d sy mbol{\zeta },$ $ where $\bo lds ymbol{ \z eta } = [\ zet a_1(\ mathbf { W}_f^\ star), \d ot s , \zet a _{K_P}( \ m at hbf{ W}_f^\star)]^T$,a nd $\mathbf{M}\in \mathb b{ C }^ { { K_P }\t imes {K_P} }$ is d e fined a s $ $ [ \ mat h bf{M}]_{kj} = \left\{
\begin {a rra y }{ll} \fr ac{1}{\gamm a_k} |\mathbf{ h}_{Pk } ^H\tild e {\mathb f{w}}_ {d, k}^ \sta r |^ 2, &k= j, \ \ - |\ma th bf {h}_{ Pk}^ H \ t i lde{ \ma thbf {w}}_ {t,j}^\star|^ 2,& k\ neq j.
\end {a rray}
\ri ght.$ $
Finally, we su mmar ize the p rop os edlo w-com p lexity al gor ithm to solvet heor i g i na l optimization pro bl e m ( \[E:prob lem-MU \ ]) i n general m ult iuse r casein T a bl e \[tab: Distri b ut ed -P-B-al lo cation \] .
--- ----- - ---- ------ -------- ----- - -------------- - ------------- - -- - - -- - ---
[a aa]{} **Ini tial i zati on:* * S et$ i=0$, $\un de r li n e{R_{sum}}=0$, and$\ overli ne{R_ {sum}} = \sum _{k=1}^{K_ P } \log(1 + \mat h tt { SINR}_k^{ub})$ , whe re $\matht t {SINR}_k ^{ub} = \frac {P_0\|\ma t h bf{h}_{P k}\ |^2 }{\ sig m a _n ^2}$ is an up p e r bo un d of th e S INR of${\ tex t{P UE} }_ k$, which is achi ev ed w he n t he IC I disappe ar s a nd on ly ${ \ text{P UE}}_ k$ i sse r ved by the PB S . \
** Bi se ctio n I te ratio n:** Atthe $i$ -th itera tio n , se t$i \lefta rrow i + 1$.
- Compute $ R_{ sum}^0 = \frac{\ underline{R_{sum}} + \o v erline{ R_{ sum}} }{2} $.
- G ive n $R_{ sum } ^0$, c ompute $\{\ ba r{\ l a mbda} _ k |
From the_equations in_(\[E:opt-p\]), we can solve_the optimal_$\mathbf{p}^\star\triangleq_[p_1^\star, \dots,_p_{K_P}^\star]^T$_as $$\label{E:PA}
_\mathbf{p}^\star = \mathbf{M}^{-1}_\boldsymbol{\zeta},$$ where $\boldsymbol{\zeta} =_[\zeta_1(\mathbf{W}_f^\star), \dots, \zeta_{K_P}(\mathbf{W}_f^\star)]^T$,_and_$\mathbf{M}\in\mathbb{C}^{{K_P}\times {K_P}}$ is defined as $$[\mathbf{M}]_{kj} = \left\{
\begin{array}{ll}
__ ___ \frac{1}{\gamma_k}|\mathbf{h}_{Pk}^H\tilde{\mathbf{w}}_{d,k}^\star|^2, & k_= j,\\
_ _ - |\mathbf{h}_{Pk}^H\tilde{\mathbf{w}}_{t,j}^\star|^2, & k \neq_j.
_ _ \end{array}
\right.$$
Finally, we summarize the proposed_low-complexity algorithm to solve the original_optimization problem (\[E:problem-MU\])_in_general_multiuser case in Table \[tab:Distributed-P-B-allocation\].
------------------------------------------------------------------------
[aaa]{}_**Initialization:** Set $i=0$, $\underline{R_{sum}}=0$, and $\overline{R_{sum}}_= \sum_{k=1}^{K_P} \log(1 +\mathtt{SINR}_k^{ub})$, where $\mathtt{SINR}_k^{ub}_= \frac{P_0\|\mathbf{h}_{Pk}\|^2}{\sigma_n^2}$ is an upper bound of_the SINR of ${\text{PUE}}_k$, which is_achieved when the ICI disappears_and only_${\text{PUE}}_k$ is served by the_PBS.\
**Bisection Iteration:** At_the $i$-th_iteration, set $i_\leftarrow i + 1$.
- _Compute $R_{sum}^0 =_\frac{\underline{R_{sum}} + \overline{R_{sum}}}{2}$.
- Given_$R_{sum}^0$,_compute $\{\bar{\lambda}_k |
}\mathbf{h}_{P,j}^H\Big)^{-1} \mathbf{h}_{Pk} \nonumber\\
\triangleq & \sqrt{p_k^\star} \tilde{\mathbf{w}}_{d,k}^\star,\end{aligned}$$ where $\tilde{\mathbf{w}}_{d,k}^\star = \big(\mathbf{I}_{N_t} + \sum_{j\neq k} \bar{\lambda}_j^\star \mathbf{h}_{P,j}\mathbf{h}_{P,j}^H\big)^{-1} \mathbf{h}_{Pk}$, and $p_k^\star$ is a scalar controlling the power allocated to ${\text{PUE}}_k$ for transmitting desired signals.
To find the optimal $\{p_k^\star\}$ in (\[E:opt-wtk\]), recalling that the optimal solution to problem (\[E:problem-MU-reform-feasibility-reform\]) is obtained when all SINR constraints in (\[E:constraint-SINR-feasibility-reform\]) hold with equality if problem (\[E:problem-MU-reform-feasibility-reform\]) is feasible for given {$\gamma_k$}, then we can obtain the following equations with respect to $p_k^\star$ $$\begin{aligned}
\label{E:opt-p}
& \frac{p_k^\star|\mathbf{h}_{Pk}^H\tilde{\mathbf{w}}_{d,k}^\star|^2}{ \sum_{j\neq k} p_j^\star\|\mathbf{h}_{Pk}^H \tilde{\mathbf{w}}_{t,j}^\star\|^2 \!+\! \zeta_k(\mathbf{W}_f^\star)} \!= \!\gamma_k,\ k \!=\! 1,\dots, {K_P},\end{aligned}$$ where $\zeta_k(\mathbf{W}_f^\star) = \|\bar{\mathbf{h}}_{Mk}^H + \mathbf{h}_{Pk}^H \mathbf{W}_f^\star \bar{\mathbf{H}}_{MP}^H e^{-j\phi}\|^2 + |\mathbf{h}_{Pk}^H \mathbf{W}_f^\star|^2(\sigma_I^2+\sigma_n^2) + \sigma_n^2$.
| } \mathbf{h}_{P, j}^H\Big)^{-1 } \mathbf{h}_{Pk } \nonumber\\
\triangleq & \sqrt{p_k^\star } \tilde{\mathbf{w}}_{d, k}^\star,\end{aligned}$$ where $ \tilde{\mathbf{w}}_{d, k}^\star = \big(\mathbf{I}_{N_t } + \sum_{j\neq k } \bar{\lambda}_j^\star \mathbf{h}_{P, j}\mathbf{h}_{P, j}^H\big)^{-1 } \mathbf{h}_{Pk}$, and $ p_k^\star$ is a scalar controlling the power allocated to $ { \text{PUE}}_k$ for transmitting desire signal.
To find the optimal $ \{p_k^\star\}$ in (\[E: opt - wtk\ ]), recalling that the optimum solution to problem (\[E: problem - MU - reform - feasibility - reform\ ]) is obtained when all SINR constraint in (\[E: constraint - SINR - feasibility - reform\ ]) hold with equality if trouble (\[E: trouble - MU - reform - feasibility - reform\ ]) is feasible for given { $ \gamma_k$ }, then we can obtain the keep up equations with obedience to $ p_k^\star$ $ $ \begin{aligned }
\label{E: opt - p }
& \frac{p_k^\star|\mathbf{h}_{Pk}^H\tilde{\mathbf{w}}_{d, k}^\star|^2 } { \sum_{j\neq k } p_j^\star\|\mathbf{h}_{Pk}^H \tilde{\mathbf{w}}_{t, j}^\star\|^2 \!+\! \zeta_k(\mathbf{W}_f^\star) } \!= \!\gamma_k,\ k \!=\! 1,\dots, { K_P},\end{aligned}$$ where $ \zeta_k(\mathbf{W}_f^\star) = \|\bar{\mathbf{h}}_{Mk}^H + \mathbf{h}_{Pk}^H \mathbf{W}_f^\star \bar{\mathbf{H}}_{MP}^H e^{-j\phi}\|^2 + |\mathbf{h}_{Pk}^H \mathbf{W}_f^\star|^2(\sigma_I^2+\sigma_n^2) + \sigma_n^2$. | }\matjbf{h}_{P,j}^H\Big)^{-1} \mathbf{h}_{Pk} \nonmmber\\
\triangleq & \sqrt{'_k^\star} \filde{\matfbf{w}}_{d,k}^\star,\end{aligned}$$ where $\tmlde{\nathbd{w}}_{d,k}^\star = \big(\mathbf{I}_{N_g} + \sum_{j\nee k} \bar{\lqmbde}_j^\star \mathbf{h}_{P,j}\mathbf{h}_{P,m}^K\big)^{-1} \jwthby{h}_{'k}$, and $p_k^\star$ ix a scalar controlling tve plwer allocated to ${\text{PUE}}_k$ for tranfmittinb fesired signalf.
To gynd fhe optimal $\{p_k^\star\}$ in (\[E:opt-wtk\]), recamling tiat the optimal solution to problem (\[E:probpem-MK-reform-feasibility-geform\]) is ovtaigwd when all RINR constgcints in (\[E:cknstraint-SINR-feasibility-reform\]) fold cith equalijv if kroblem (\[E:proulem-MU-geform-feasibility-refosm\]) is frasible for giyen {$\gemma_j$}, then we can obtain vhe following equatijns with seapect to $p_k^\star$ $$\bwgun{alicned}
\labdo{E:oot-p}
& \frac{p_n^\ster|\mathbf{h}_{Pk}^G\tilde{\mathbd{w}}_{d,k}^\star|^2}{ \sum_{j\neq k} k_j^\seqr\|\mathbf{h}_{Pk}^H \filde{\mwtrbf{w}}_{t,j}^\star\|^2 \!+\! \zeta_k(\mathbf{W}_f^\star)} \!= \!\gamma_k,\ n \!=\! 1,\sots, {K_P},\end{aligned}$$ where $\zeta_k(\mathbf{W}_f^\star) = \|\bwr{\mathbf{h}}_{Ik}^H + \mathbf{h}_{Pk}^H \mathbf{W}_f^\star \bar{\mathbf{H}}_{MP}^H e^{-j\phi}\|^2 + |\matibw{h}_{Pj}^H \oqtjbf{W}_f^\star|^2(\sigma_I^2+\sigma_n^2) + \sigma_n^2$.
| }\mathbf{h}_{P,j}^H\Big)^{-1} \mathbf{h}_{Pk} \nonumber\\ \triangleq & \sqrt{p_k^\star} \tilde{\mathbf{w}}_{d,k}^\star,\end{aligned}$$ = + \sum_{j\neq \bar{\lambda}_j^\star \mathbf{h}_{P,j}\mathbf{h}_{P,j}^H\big)^{-1} \mathbf{h}_{Pk}$, controlling power allocated to for transmitting desired To find the optimal $\{p_k^\star\}$ in recalling that the optimal solution to problem (\[E:problem-MU-reform-feasibility-reform\]) is obtained when all SINR in (\[E:constraint-SINR-feasibility-reform\]) hold with equality if problem (\[E:problem-MU-reform-feasibility-reform\]) is feasible for given {$\gamma_k$}, we obtain following with respect to $p_k^\star$ $$\begin{aligned} \label{E:opt-p} & \frac{p_k^\star|\mathbf{h}_{Pk}^H\tilde{\mathbf{w}}_{d,k}^\star|^2}{ \sum_{j\neq k} p_j^\star\|\mathbf{h}_{Pk}^H \tilde{\mathbf{w}}_{t,j}^\star\|^2 \!+\! \zeta_k(\mathbf{W}_f^\star)} \!= \!\gamma_k,\ \!=\! 1,\dots, {K_P},\end{aligned}$$ where $\zeta_k(\mathbf{W}_f^\star) = \|\bar{\mathbf{h}}_{Mk}^H + \mathbf{W}_f^\star \bar{\mathbf{H}}_{MP}^H e^{-j\phi}\|^2 + \mathbf{W}_f^\star|^2(\sigma_I^2+\sigma_n^2) + \sigma_n^2$. | }\mathbf{h}_{P,j}^H\Big)^{-1} \mathbf{h}_{Pk} \nonUmber\\
\trianGleq & \sQrt{P_k^\sTaR} \tilDe{\maThbf{w}}_{d,k}^\star,\end{ALignEd}$$ where $\tilde{\mathbf{w}}_{d,k}^\sTar = \biG(\mAThbf{i}_{n_t} + \Sum_{j\nEq k} \bar{\lAMbDA}_J^\stAr \MaThbF{h}_{p,J}\mAthbf{H}_{P,j}^h\big)^{-1} \matHbf{h}_{Pk}$, and $p_K^\stAr$ Is a scalar conTRoLling the poWer Allocated to ${\tExt{pUE}}_k$ foR tRanSMittiNg dEsireD signaLS.
To finD the optimAl $\{P_K^\star\}$ iN (\[e:opt-wtk\]), RECaLlinG that the optimal soLUtIOn to problem (\[E:prOblem-Mu-rEFoRM-FeaSibIlity-reforM\]) iS obtaINed when ALl sinr coNStraints in (\[E:coNstraint-SINr-FeaSibiliTy-RefORm\]) hold With eQuALitY if problem (\[E:ProbLem-MU-refoRm-feasIBility-rEForm\]) is fEasiblE foR giVen {$\gAMmA_k$}, TheN wE Can OBtAin THe fOllowing EqUaTions With RESPEct tO $p_k^\Star$ $$\Begin{Aligned}
\label{E:Opt-P}
& \fraC{P_k^\sTar|\maThbf{h}_{pk}^H\tIlDe{\matHbf{w}}_{d,k}^\Star|^2}{ \sUm_{J\neq k} p_j^\star\|\mathBf{h}_{PK}^H \tilde{\maThbF{w}}_{T,j}^\sTaR\|^2 \!+\! \zeta_K(\Mathbf{w}_f^\sTar)} \!= \!\Gamma_k,\ k \!=\! 1,\Dots, {K_P},\eND{alIgNED}$$ WhEre $\zeta_k(\mathbf{W}_f^\stAr) = \|\BAR{\mAthbf{h}}_{Mk}^h + \mathbF{H}_{PK}^H \MAthbf{W}_f^\sTaR \baR{\matHBF{H}}_{MP}^H E^{-j\phI}\|^2 + |\MaThbf{h}_{Pk}^H \Mathbf{w}_F^\sTaR|^2(\sigma_I^2+\SiGma_n^2) + \siGmA_n^2$.
| }\mathbf{h}_{P,j}^H\Big)^{ -1} \mathb f{h}_ {Pk } \ no numb er\\
\triangleq & \s qrt{p_k^\star} \tilde{ \math bf { w}}_ { d, k}^\s tar,\en d {a l i gne d} $$ wh er e $ \tild e{\ mathbf{ w}}_{d,k}^ \st ar = \big(\ma t hb f{I}_{N_t} +\sum_{j\neqk}\bar{\ la mbd a }_j^\ sta r \ma thbf{h } _{P,j} \mathbf{h }_ { P,j}^H \ big)^{- 1 } \ math bf{h}_{Pk}$, and$ p_ k ^\star$ is a s calarco n tr o l lin g t he power a ll ocate d to ${\ t ex t { P UE} } _k$ for trans mitting des i red signa ls .
T o find theop t ima l $\{p_k^\s tar\ }$ in (\[ E:opt- w tk\]),r ecallin g that th e o ptim a lso lut io n to pr obl e m ( \[E:prob le m- MU-re form - f e a sibi lit y-re form\ ]) is obtaine d w hena llSINRconst rain ts in ( \[E:co nstra in t-SINR-feasibil ity- reform\]) ho ld wi th equa l ity if pr obl em (\[E :proble m -MU -r e f o rm -feasibility-refor m\ ] ) i s feasib le for gi ve n {$\gamm a_ k$} , th e n we c an o b ta in the f ollowi n geq uations w ith re sp ect to $p_k ^ \sta r$ $$\ begin{al igned }
\label{E:opt - p}
& \fra c {p _ k ^\ s tar| \ma thbf{h}_{Pk }^H\ t ilde {\ma t hb f{w } }_{d, k}^\s ta r |^ 2 }{ \sum_{j\neq k} p _j ^\star \|\ma thbf{h}_{Pk}^ H \tilde{\ m a t hbf{w}}_ {t,j } ^\ s tar\|^2 \!+\!\zeta _k(\mathbf { W}_f^\st ar)}\!= \!\g amma_k,\k \!=\! 1 ,\d ots , { K_P } , \e nd{aligned}$$ w here $ \zeta_k (\m athbf{W }_f ^\s tar ) = \ |\bar{\ma thbf{h}} _{ Mk }^ H+ \ mathb f {h}_{Pk} ^H \m at hbf {W}_f ^ \star\bar{ \mat hb f{ H }}_ {MP}^He ^{ - j \phi }\ |^ 2 + |\ ma thbf{ h}_{ P k}^ H \math bf{W}_f^\ sta r |^2( \s ig ma_I^2+ \sigma_n^2) + \ sigma_n^2$ . | }\mathbf{h}_{P,j}^H\Big)^{-1} \mathbf{h}_{Pk}_\nonumber\\
_ \triangleq & \sqrt{p_k^\star}_\tilde{\mathbf{w}}_{d,k}^\star,\end{aligned}$$ where_$\tilde{\mathbf{w}}_{d,k}^\star_= _\big(\mathbf{I}_{N_t}_+ \sum_{j\neq k}_\bar{\lambda}_j^\star \mathbf{h}_{P,j}\mathbf{h}_{P,j}^H\big)^{-1} \mathbf{h}_{Pk}$,_and $p_k^\star$ is a_scalar controlling the_power_allocated to ${\text{PUE}}_k$ for transmitting desired signals.
To find the optimal $\{p_k^\star\}$ in (\[E:opt-wtk\]), recalling_that_the optimal_solution_to_problem (\[E:problem-MU-reform-feasibility-reform\]) is obtained when_all SINR constraints in (\[E:constraint-SINR-feasibility-reform\])_hold with_equality if problem (\[E:problem-MU-reform-feasibility-reform\]) is feasible for given_{$\gamma_k$},_then we can_obtain the following equations with respect to $p_k^\star$ $$\begin{aligned}
_\label{E:opt-p}
& \frac{p_k^\star|\mathbf{h}_{Pk}^H\tilde{\mathbf{w}}_{d,k}^\star|^2}{_\sum_{j\neq k} p_j^\star\|\mathbf{h}_{Pk}^H_\tilde{\mathbf{w}}_{t,j}^\star\|^2_\!+\!_\zeta_k(\mathbf{W}_f^\star)} \!= \!\gamma_k,\ _k \!=\! 1,\dots, {K_P},\end{aligned}$$ where $\zeta_k(\mathbf{W}_f^\star)_= \|\bar{\mathbf{h}}_{Mk}^H + \mathbf{h}_{Pk}^H \mathbf{W}_f^\star \bar{\mathbf{H}}_{MP}^H_e^{-j\phi}\|^2 + |\mathbf{h}_{Pk}^H \mathbf{W}_f^\star|^2(\sigma_I^2+\sigma_n^2) + \sigma_n^2$.
|
sqrt{\lambda_i}$ & $\omega_b$ & $\omega_m$ & $\omega_\Lambda$ & $n_s$ & $Q$ & ${{\mathcal R}}$ & $\alpha$\
1& 2.50E-04& 9.9446E-01\* & -9.9203E-02$\dag$ & -2.5224E-05 & -2.7487E-02 & -3.9411E-03 & 1.2295E-02 & 1.6954E-02\
2& 8.84E-04& 8.1778E-02 & 7.0553E-01\* & -5.6359E-04 & -6.8131E-02 & 2.4777E-02 & -1.1338E-01 & -6.9096E-01$\dag$\
3& 2.24E-03& 4.8801E-02 & 5.2913E-01 $\dag$& 9.3752E-04 & 2.6766E-01 & -6.3566E-01\* & 4.0924E-02 & 4.9016E-01\
4& 1.24E-02& 4.2341E-02 & 2.5947E-01 & 1.2292E-02 & 6.5656E-01$\dag$ & 6.6964E-01\* & 4.5581E-02 & 2.2174E-01\
5& 1.48E-02& 1.0147E-02 & 3.7938E-01 & -3.5290E-02 & -6.9349E-01\* & 3.7432E-01 & 2.0829E-01 & 4.3623E-01$\dag$\
6& 1.94E-01& -9.0774E-03 & -2.9295E-02 & 2.2193E-01$\dag$ & 8.9661E-02 & -7.8874E-02 & 9.4671E-01\* & -1.9819E-01\
7& 3.71E-01& 1.9270E-03 & 1.7036E-02 & 9.7435E-01\* & | sqrt{\lambda_i}$ & $ \omega_b$ & $ \omega_m$ & $ \omega_\Lambda$ & $ n_s$ & $ Q$ & $ { { \mathcal R}}$ & $ \alpha$\
1 & 2.50E-04 & 9.9446E-01\ * & -9.9203E-02$\dag$ & -2.5224E-05 & -2.7487E-02 & -3.9411E-03 & 1.2295E-02 & 1.6954E-02\
2 & 8.84E-04 & 8.1778E-02 & 7.0553E-01\ * & -5.6359E-04 & -6.8131E-02 & 2.4777E-02 & -1.1338E-01 & -6.9096E-01$\dag$\
3 & 2.24E-03 & 4.8801E-02 & 5.2913E-01 $ \dag$ & 9.3752E-04 & 2.6766E-01 & -6.3566E-01\ * & 4.0924E-02 & 4.9016E-01\
4 & 1.24E-02 & 4.2341E-02 & 2.5947E-01 & 1.2292E-02 & 6.5656E-01$\dag$ & 6.6964E-01\ * & 4.5581E-02 & 2.2174E-01\
5 & 1.48E-02 & 1.0147E-02 & 3.7938E-01 & -3.5290E-02 & -6.9349E-01\ * & 3.7432E-01 & 2.0829E-01 & 4.3623E-01$\dag$\
6 & 1.94E-01 & -9.0774E-03 & -2.9295E-02 & 2.2193E-01$\dag$ & 8.9661E-02 & -7.8874E-02 & 9.4671E-01\ * & -1.9819E-01\
7 & 3.71E-01 & 1.9270E-03 & 1.7036E-02 & 9.7435E-01\ * & | sqrh{\lambda_i}$ & $\omega_b$ & $\omega_m$ & $\omega_\Lambda$ & $u_w$ & $Q$ & ${{\kathcam R}}$ & $\alpfa$\
1& 2.50E-04& 9.9446E-01\* & -9.9203E-02$\dag$ & -2.5224E-05 & -2.7487E-02 & -3.9411E-03 & 1.2295E-02 & 1.6954E-02\
2& 8.84X-04& 8.1778E-02 & 7.0553E-01\* & -5.6359E-04 & -6.8131E-02 & 2.4777E-02 & -1.1338E-01 & -6.9096E-01$\dag$\
3& 2.24E-03& 4.8801E-02 & 5.2913E-01 $\dag$& 9.3752E-04 & 2.6766E-01 & -6.3566E-01\* & 4.0924E-02 & 4.9016W-01\
4& 1.24E-02& 4.2341T-02 & 2.5947E-01 & 1.2292E-02 & 6.5656E-01$\dag$ & 6.6964E-01\* & 4.5581E-02 & 2.2174E-01\
5& 1.48E-02& 1.0147C-02 & 3.7938E-01 & -3.5290C-02 & -6.9349E-01\* & 3.7432X-01 & 2.0829E-01 & 4.3623E-01$\dag$\
6& 1.94E-01& -9.0774E-03 & -2.9295E-02 & 2.2193E-01$\dag$ & 8.9661A-02 & -7.8874E-02 & 9.4671E-01\* & -1.9819E-01\
7& 3.71E-01& 1.9270E-03 & 1.7036D-02 & 9.7435E-01\* & | sqrt{\lambda_i}$ & $\omega_b$ & $\omega_m$ & $\omega_\Lambda$ & & ${{\mathcal & $\alpha$\ 1& -2.5224E-05 -2.7487E-02 & -3.9411E-03 1.2295E-02 & 1.6954E-02\ 8.84E-04& 8.1778E-02 & 7.0553E-01\* & -5.6359E-04 -6.8131E-02 & 2.4777E-02 & -1.1338E-01 & -6.9096E-01$\dag$\ 3& 2.24E-03& 4.8801E-02 & 5.2913E-01 $\dag$& & 2.6766E-01 & -6.3566E-01\* & 4.0924E-02 & 4.9016E-01\ 4& 1.24E-02& 4.2341E-02 & 2.5947E-01 1.2292E-02 6.5656E-01$\dag$ 6.6964E-01\* 4.5581E-02 & 2.2174E-01\ 5& 1.48E-02& 1.0147E-02 & 3.7938E-01 & -3.5290E-02 & -6.9349E-01\* & 3.7432E-01 & 2.0829E-01 4.3623E-01$\dag$\ 6& 1.94E-01& -9.0774E-03 & -2.9295E-02 & 2.2193E-01$\dag$ 8.9661E-02 & -7.8874E-02 & & -1.9819E-01\ 7& 3.71E-01& 1.9270E-03 1.7036E-02 9.7435E-01\* & | sqrt{\lambda_i}$ & $\omega_b$ & $\omega_m$ & $\omEga_\Lambda$ & $n_S$ & $Q$ & ${{\matHcaL R}}$ & $\aLpHa$\
1& 2.50E-04& 9.9446E-01\* & -9.9203e-02$\dag$ & -2.5224e-05 & -2.7487E-02 & -3.9411E-03 & 1.2295E-02 & 1.6954E-02\
2& 8.84E-04& 8.1778E-02 & 7.0553E-01\* & -5.6359E-04 & -6.8131E-02 & 2.4777E-02 & -1.1338E-01 & -6.9096E-01$\dAG$\
3& 2.24E-03& 4.8801E-02 & 5.2913E-01 $\Dag$& 9.3752E-04 & 2.6766E-01 & -6.3566E-01\* & 4.0924E-02 & 4.9016E-01\
4& 1.24E-02& 4.2341E-02 & 2.5947E-01 & 1.2292E-02 & 6.5656E-01$\dag$ & 6.6964E-01\* & 4.5581E-02 & 2.2174E-01\
5& 1.48E-02& 1.0147E-02 & 3.7938E-01 & -3.5290e-02 & -6.9349E-01\* & 3.7432E-01 & 2.0829E-01 & 4.3623E-01$\DaG$\
6& 1.94e-01& -9.0774E-03 & -2.9295E-02 & 2.2193E-01$\DAg$ & 8.9661e-02 & -7.8874E-02 & 9.4671E-01\* & -1.9819E-01\
7& 3.71E-01& 1.9270e-03 & 1.7036E-02 & 9.7435E-01\* & | sqrt{\lambda_i}$ & $\omega _b$ & $\om ega_m $ & $\ om ega_ \Lam bda$ & $n_s$ & $Q$& ${{\mathcal R}}$ & $ \alph a$ \
1& 2. 50E-0 4& 9.94 4 6E - 0 1\* & - 9.9 20 3 E- 02$\d ag$ & -2.5 224E-05 &-2. 74 87E-02 & -3. 9 41 1E-03 & 1. 229 5E-02 & 1.69 54E -02\
2 &8.8 4 E-04& 8. 1778E -02 &7 .0553E -01\* & - 5. 6 359E-0 4 & -6.8 1 3 1E -02& 2.4777E-02 & -1 . 13 3 8E-01 & -6.909 6E-01$ \d a g$ \ 3&2.2 4E-03& 4.8 80 1E-02 & 5.291 3 E- 0 1 $\d a g$& 9.3752E-0 4 & 2.6766E - 01& -6.3 56 6E- 0 1\* &4.092 4E - 02& 4.9016E-0 1\
4 & 1.24E-0 2& 4.2 3 41E-02& 2.5947 E-01 & 1. 229 2E-0 2 & 6 .56 56 E -01 $ \d ag$ & 6 .6964E-0 1\ *& 4.5 581E - 0 2 & 2. 217 4E-0 1\
5& 1.48E-02& 1. 014 7E-0 2 &3.793 8E-01 & - 3. 5290E -02 &-6.93 49 E-01\* & 3.7432 E-01 & 2.0829 E-0 1& 4 .3 623E- 0 1$\dag $\6&1.94E-0 1& -9.0 7 74E -0 3 & - 2.9295E-02 & 2.219 3E - 0 1$ \dag$ &8.9661 E -0 2& -7.8874 E- 02& 9. 4 6 71E-0 1\*& - 1.9819E- 01\
7& 3. 71 E-01& 1 .9 270E-0 3& 1 .70 36E-0 2 & 9 .7435E -01\* & | sqrt{\lambda_i}$ &_$\omega_b$ &_$\omega_m$ & $\omega_\Lambda$ &_$n_s$ &_$Q$_& ${{\mathcal_R}}$_& $\alpha$\
1& 2.50E-04&_9.9446E-01\* & -9.9203E-02$\dag$_& -2.5224E-05 & -2.7487E-02_& -3.9411E-03 &_1.2295E-02_& 1.6954E-02\
2& 8.84E-04& 8.1778E-02 & 7.0553E-01\* & -5.6359E-04 & -6.8131E-02 & 2.4777E-02 & -1.1338E-01_&_-6.9096E-01$\dag$\
3& 2.24E-03&_4.8801E-02_&_5.2913E-01 $\dag$& 9.3752E-04 & 2.6766E-01_& -6.3566E-01\* & 4.0924E-02 &_4.9016E-01\
4& 1.24E-02&_4.2341E-02 & 2.5947E-01 & 1.2292E-02 & 6.5656E-01$\dag$ &_6.6964E-01\*_& 4.5581E-02 &_2.2174E-01\
5& 1.48E-02& 1.0147E-02 & 3.7938E-01 & -3.5290E-02 & -6.9349E-01\*_& 3.7432E-01 & 2.0829E-01 & 4.3623E-01$\dag$\
6&_1.94E-01& -9.0774E-03 &_-2.9295E-02_&_2.2193E-01$\dag$ & 8.9661E-02 &_-7.8874E-02 & 9.4671E-01\* & -1.9819E-01\
7& 3.71E-01&_1.9270E-03 & 1.7036E-02 & 9.7435E-01\* & |
\in R_{>0}$, we have $$a\triangleright _{t} (a\triangleright _{s}b)= a\triangleright
_{s+t}b.$$
Let for brevity $c:= a\triangleright _{s}b$ and $d:=
a\triangleright _{t}c$. Then $[abc]$ and $[acd]$, hence by Proposition \[transx\], we also have $[abd]$ and $[bcd]$, Also, by construction, $bc=s$ and $cd=t$. By $[bcd]$ we have $bd=s+t$, and by $[abd]$, $d$ is aligned with $a,b$. These two properties characterize $a\triangleright_{s+t}b$.
We call the map $R_{>0}\to M$ given by $s\mapsto a\triangleright
_{s}b$ the [*ray generated by $a$ and $b$*]{}, and we call $b$ its [*source* ]{} of the ray. (Note that we cannot say $a\triangleright_{0}b =b$, since $a\triangleright _{s}b$ only is defined for $s>0$. However, it is easy to “patch” rays, using Proposition \[transx\].)
Any ray $R_{>0}\to M$ is an isometry i.e. is distance preserving. Furthermore, any triple of mutually apart points on a ray are aligned.
The first assertion is an immediate consequence of Proposition \[asssx\]; the second follows from the collinearity of $a,b$, and $a\triangleright_{s}b$ by iterated use of use of Proposition \[transx\].
Thus, a ray with source $b$ can be viewed as a “parametrization of its image by arc length, measured from $b$” (except that we have not attempted to define these terms here). Note that $b$ itself is not in the image of the ray.
The following Example refers to the model of the axiomatics which one obtains from SDG, as in Section \[modelx\]. It shows that the isometry property is not sufficient for being a ray:
[**Example.**]{} Consider in $R^{2}$ the points $a | \in R_{>0}$, we have $ $ a\triangleright _ { t } (a\triangleright _ { s}b)= a\triangleright
_ { s+t}b.$$
Let for brevity $ c:= a\triangleright _ { s}b$ and $ d:=
a\triangleright _ { t}c$. Then $ [ abc]$ and $ [ acd]$, hence by Proposition \[transx\ ], we also get $ [ abd]$ and $ [ bcd]$, besides, by construction, $ bc = s$ and $ cd = t$. By $ [ bcd]$ we get $ bachelor of divinity = s+t$, and by $ [ abd]$, $ d$ is aligned with $ a, b$. These two properties characterize $ a\triangleright_{s+t}b$.
We bid the map $ R_{>0}\to M$ given by $ s\mapsto a\triangleright
_ { s}b$ the [ * beam generated by $ a$ and $ b$ * ] { }, and we call $ b$ its [ * reservoir * ] { } of the ray. (Note that we cannot pronounce $ a\triangleright_{0}b = b$, since $ a\triangleright _ { s}b$ merely is defined for $ s>0$. However, it is easy to “ patch ” rays, using Proposition \[transx\ ] .)
Any beam $ R_{>0}\to M$ is an isometry i.e. is distance preserving. Furthermore, any triple of mutually apart points on a re are aligned.
The first assertion is an contiguous consequence of Proposition \[asssx\ ]; the second follows from the collinearity of $ a, b$, and $ a\triangleright_{s}b$ by iterate use of use of Proposition \[transx\ ].
Thus, a re with source $ b$ can be viewed as a “ parametrization of its image by arc length, measured from $ b$ ” (except that we have not attempted to define these terms here). Note that $ b$ itself is not in the image of the ray.
The following Example refers to the model of the axiomatics which one obtains from SDG, as in Section \[modelx\ ]. It shows that the isometry property is not sufficient for being a beam:
[ * * Example. * * ] { } Consider in $ R^{2}$ the point $ a | \in R_{>0}$, we have $$a\trianglerighu _{t} (a\trianglerighj _{w}b)= a\trmanglerjght
_{s+t}b.$$
Uet for brevity $c:= a\trianglermght _{s}b$ abd $d:=
a\triangleright _{t}c$. Then $[abc]$ and $[acd]$, henre by Propositioi \[transx\], we alal hare $[abd]$ and $[bcd]$, Akso, by convtruction, $bc=s$ dna $ed=t$. By $[bcd]$ we have $bd=s+t$, and by $[abd]$, $d$ is alibnfd with $a,b$. Thefe tej prkievties characterize $a\trianglerighf_{s+t}b$.
We ball the map $R_{>0}\to K$ given by $s\mapsto a\triangperihht
_{s}b$ the [*ray genfrated by $a$ and $v$*]{}, and we calu $b$ its [*solxce* ]{} of the ray. (Note that we cannot say $a\tfiangkeright_{0}b =b$, suncf $a\trianglermght _{s}f$ only is denpned fos $s>0$. Howrver, it is easn to “'atcy” rays, using Propositmon \[transx\].)
Any ray $R_{>0}\tj M$ is an iaometry i.e. is distqnxe prgservhng. Wyrtferjoce, zny trlplx of mutualmy apart pounts on a ray are akidbed.
The first zssertyog is an immediate consequence of Proposptioh \[asssx\]; the second folliws from the collineatity of $a,b$, and $a\triangleright_{s}b$ by iterated use of use of Psoposmtkon \[tvansb\].
Rhks, a ray with source $b$ can be viewed as a “parwjeuridation of its imane by arc length, mradutgd from $b$” (excekt that we have not attemptef to desine rhese teris hrre). Note that $b$ itself is nit in the imcge of the ray.
The folpowing Exam'le regers yo the model of the axiumatjcs which oje obtaina from SDG, as in Recnion \[modelx\]. It shows that the ysometry 'ropexty is nut sofficiegt for beijg a vdy:
[**Example.**]{} Consider in $R^{2}$ tve points $w | \in R_{>0}$, we have $$a\triangleright _{t} (a\triangleright _{s+t}b.$$ for brevity a\triangleright _{s}b$ and and hence by Proposition we also have and $[bcd]$, Also, by construction, $bc=s$ $cd=t$. By $[bcd]$ we have $bd=s+t$, and by $[abd]$, $d$ is aligned with These two properties characterize $a\triangleright_{s+t}b$. We call the map $R_{>0}\to M$ given by a\triangleright the generated $a$ and $b$*]{}, and we call $b$ its [*source* ]{} of the ray. (Note that we say $a\triangleright_{0}b =b$, since $a\triangleright _{s}b$ only is for $s>0$. However, it easy to “patch” rays, using \[transx\].) ray $R_{>0}\to is isometry is distance preserving. any triple of mutually apart points on a ray are aligned. The first assertion is an immediate of Proposition second follows the of and $a\triangleright_{s}b$ by of use of Proposition \[transx\]. Thus, source $b$ can be viewed as a “parametrization its image arc length, measured from $b$” (except we have not attempted to define these terms Note that $b$ itself is not in the image of the ray. The following Example the model of the which one obtains SDG, in \[modelx\]. shows that isometry property is not sufficient for being a ray: [**Example.**]{} Consider $R^{2}$ the points $a | \in R_{>0}$, we have $$a\triangleright _{t} (a\TriangleriGht _{s}b)= A\trIanGlErigHt
_{s+t}B.$$
Let for brevity $C:= A\triAngleright _{s}b$ and $d:=
a\trianGleriGhT _{T}c$. ThEN $[aBc]$ and $[Acd]$, hencE By pROpoSiTiOn \[tRaNSx\], We alsO haVe $[abd]$ anD $[bcd]$, Also, by ConStRuction, $bc=s$ anD $Cd=T$. By $[bcd]$ we haVe $bD=s+t$, and by $[abd]$, $d$ Is aLigned WiTh $a,B$. these Two PropeRties cHAracteRize $a\triaNgLEright_{S+T}b$.
We calL THe Map $R_{>0}\To M$ given by $s\mapsto A\TrIAngleright
_{s}b$ thE [*ray geNeRAtED By $a$ And $B$*]{}, and we call $B$ iTs [*souRCe* ]{} of the RAy. (nOTE thAT we cannot say $a\TrianglerigHT_{0}b =b$, Since $a\TrIanGLerighT _{s}b$ onLy IS deFined for $s>0$. HoWeveR, it is easy To “patcH” Rays, usiNG ProposItion \[tRanSx\].)
ANy raY $r_{>0}\tO M$ Is aN iSOmeTRy I.e. iS DisTance preSeRvIng. FuRtheRMORE, any TriPle oF mutuAlly apart poinTs oN a raY Are AlignEd.
The FirsT aSsertIon is aN immeDiAte consequence oF ProPosition \[aSssX\]; tHe sEcOnd foLLows frOm tHe cOllineaRity of $a,B$, And $A\tRIANgLeright_{s}b$ by iterateD uSE Of Use of ProPositiON \[tRaNSx\].
Thus, a rAy WitH souRCE $b$ can Be viEWeD as a “paraMetrizATiOn Of its imAgE by arc LeNgtH, meAsureD From $B$” (excepT that we hAve noT Attempted to defINe these terms hERe). nOTe THat $b$ ItsElf is not in tHe imAGe of The rAY.
THe fOLlowiNg ExaMpLE rEFers to the model of the AxIomatiCs whiCh one obtains fRom SDG, as in sECTion \[modeLx\]. It SHoWS that the isometRy proPerty is not SUfficienT for bEing a ray:
[**example.**]{} CoNSIder in $R^{2}$ tHe pOinTs $a | \in R_{>0}$, we have $$a\ triangleri ght _ {t} (a \t rian gler ight _{s}b)= a \ tria ngleright
_{s+t}b.$$
Letfo r bre v it y $c: = a\tri a ng l e rig ht _ {s} b$ an d $d: = a\trian gleright _ {t} c$ . Then $[abc ] $and $[acd] $,hence by Pro pos ition\[ tra n sx\], we also have$ [abd]$ and $[bc d] $ , Also , by con s t ru ctio n, $bc=s$ and $cd = t$ . By $[bcd]$ we have$b d =s + t $,and by $[abd] $, $d$i s align e dw i t h $ a ,b$. These tw o propertie s ch aracte ri ze$ a\tria ngler ig h t_{ s+t}b$.
We cal l the map $R_{> 0 }\to M$ given b y $s\m aps toa\tr i an gl eri gh t
_ { s} b$t he[*ray ge ne ra ted b y $a $ a n d $b $*] {},and w e call $b$ it s [ *sou r ce* ]{}of th e ra y. (Not e that we c an not say $a\tria ngle right_{0} b = b$ , s in ce $a \ triang ler igh t _{s}b $ onlyi s d ef i n e dfor $s>0$. However ,i t i s easy t o “pat c h” r a ys, usin gPro posi t i on \[ tran s x\ ].)
Any ray $ R _{ >0 }\to M$ i s an i so met ryi.e.i s di stance preserv ing.F urthermore, an y triple of mu t ua l l ya part po ints on a r ay a r e al igne d .
Th e firs t ass er t io n is an immediate co ns equenc e ofProposition \ [asssx\];t h e secondfoll o ws from the colli neari ty of $a,b $ , and $a \tria nglerigh t_{s}b$ b y iterated us e o f u seo f P roposition \[ t r ansx \] .
Thus , a ray wi thsou rce $b $can be vi ewed asa“p ar am etr izati o n of its i mag ebyarc l e ngth,measu redfr om $b$ ” (exce p tt h at w eha ve n otat tempt ed t o de fine th ese terms he r e).No te that $ b$ itself isno t in the i ma geof the r ay.
The following Example refe r s to th e m odelof t he axioma tic s whic h o n e obta ins fr om SD G, as i n Sec t i on \[ mo delx\]. It s how s tha ttheisometr y property is nots uff icient for be ing a r a y :
[* * Ex a mpl e. * *]{ } Consider in $R^ {2}$ the p oi n ts $a | \in_R_{>0}$, we_have $$a\triangleright _{t} (a\triangleright__{s}b)= a\triangleright_
_{s+t}b.$$
Let_for brevity_$c:=_a\triangleright _{s}b$ and_$d:=
a\triangleright _{t}c$._Then $[abc]$ and $[acd]$,_hence by Proposition_\[transx\],_we also have $[abd]$ and $[bcd]$, Also, by construction, $bc=s$ and $cd=t$. By $[bcd]$_we_have $bd=s+t$,_and_by_$[abd]$, $d$ is aligned with_$a,b$. These two properties characterize_$a\triangleright_{s+t}b$.
We call_the map $R_{>0}\to M$ given by $s\mapsto a\triangleright_
_{s}b$_the [*ray generated_by $a$ and $b$*]{}, and we call $b$ its_[*source* ]{} of the ray. (Note_that we cannot_say_$a\triangleright_{0}b_=b$, since $a\triangleright _{s}b$_only is defined for $s>0$. However,_it is easy to “patch” rays,_using Proposition \[transx\].)
Any ray $R_{>0}\to M$ is_an isometry i.e. is distance preserving. Furthermore,_any triple of mutually apart_points on_a ray are aligned.
The first_assertion is an_immediate consequence_of Proposition \[asssx\];_the second follows from the collinearity_of $a,b$, and_$a\triangleright_{s}b$ by iterated use of use_of_Proposition \[transx\].
Thus, a_ray_with_source $b$_can be viewed_as_a “parametrization_of_its image by arc length, measured_from_$b$” (except that we have not attempted_to define these terms_here)._Note that $b$ itself_is not in the image_of the ray.
The following Example refers_to the_model of_the axiomatics which one obtains from SDG, as in Section \[modelx\]._It shows that the isometry property_is not sufficient for_being a_ray:
[**Example.**]{}_Consider in $R^{2}$_the_points $a |
the moments to a set of linear equations. Explicit results for the variance and the third cumulant of the complex amplitude are obtained also for an arbitrary noise provided the noise is weak; the third cumulant of the amplitude in this case is proportional to the third cumulant of the noise. Even for weak frequency noise, the proposed method allows revealing it irrespective of the intensity of additive noise in the oscillator. The results apply to both classical and quantum oscillators.
**[ACKNOWLEDGMENTS]{}**
This research was supported in part by DARPA through the DEFYS program. MID acknowledges also partial support from the NSF, Grant EMT/QIS 082985.
[30]{}ifxundefined \[1\][ ifx[\#1]{} ]{}ifnum \[1\][ \#1firstoftwo secondoftwo ]{}ifx \[1\][ \#1firstoftwo secondoftwo ]{}““\#1””@noop \[0\][secondoftwo]{}sanitize@url \[0\][‘\
12‘\$12 ‘&12‘\#12‘12‘\_12‘%12]{}@startlink\[1\]@endlink\[0\]@bib@innerbibempty @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} in @noop [ ]{}, (, ), Vol. , pp. @noop [ ]{} (, ) @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [ ]{} (, ) @noop [ ]{} (, ) @noop [****, ()]{} @noop [****, ()]{} @noop [ ]{} (, , ) @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{}
---
abstract: 'Two active hypothesis testing problems are formulated. In these problems, the agent can perform a fixed number of experiments and then decide on one of the hypotheses. The agent is also allowed to declare its experiments inconclusive if needed. The first problem is an asymmetric formulation in which the the objective is to minimize the probability of incorrectly declaring a particular hypothesis to be true while | the moments to a set of linear equations. denotative result for the variance and the third cumulant of the complex amplitude are obtained besides for an arbitrary noise put up the noise is decrepit; the third cumulant of the amplitude in this character is proportional to the third cumulant of the noise. Even for watery frequency noise, the proposed method acting allows revealing it irrespective of the intensity of linear noise in the oscillator. The results lend oneself to both classical and quantum oscillators.
* * [ recognition ] { } * *
This research was supported in part by DARPA through the DEFYS course of study. MID acknowledges also partial support from the NSF, Grant EMT / QIS 082985.
[ 30]{}ifxundefined \[1\ ] [ ifx[\#1 ] { } ] { } ifnum \[1\ ] [ \#1firstoftwo secondoftwo ] { } ifx \[1\ ] [ \#1firstoftwo secondoftwo ] { } “ “ \#1””@noop \[0\][secondoftwo]{}sanitize@url \[0\][‘\
12‘\$12 ‘ & 12‘\#12‘12‘\_12‘%12]{}@startlink\[1\]@endlink\[0\]@bib@innerbibempty @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } in @noop [ ] { }, (, ), Vol. , pp. @noop [ ] { } (, ) @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ ] { } (, ) @noop [ ] { } (, ) @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ ] { } (, , ) @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { } @noop [ * * * *, () ] { }
---
abstract:' Two active guess testing problem are formulated. In these problems, the agent can do a fixed number of experiments and then decide on one of the hypotheses. The agent is also allowed to declare its experiments inconclusive if needed. The first problem is an asymmetric formulation in which the the objective is to minimize the probability of incorrectly declare a particular hypothesis to be on-key while | thf moments to a set of likear equations. Erplicit resulfs for tfe variance and the third culuoant if the complex amplituae are obnained alwo fie an arbitczry noise problded chx noise is weak; the third cumulant of tve alplitude in this case is proportionwl to tne third cumulanj of uhe noiav. Tven for weak frequency noise, the propostd method allows rrvealing it irrespective ov thf intensity of addltive noise in eye oscillatof. The results apply to both classical and quantum osckllatprs.
**[ACKNOWLGBTMEJJS]{}**
This reseacch waf supported lm part by DARLA through the DEHYS program. MID acknowledjes also partial supkort from dhz NSF, Grant EMT/QIS 082985.
[30]{}ifzubdefited \[1\][ ifx[\#1]{} ]{}ufnjm \[1\][ \#1fmrsfoftwo serondoftwo ]{}irx \[1\][ \#1firstofrwo secondoftwo ]{}““\#1””@nook \[0\][sqbpndoftwo]{}sanifize@urj \[0\][‘\
12‘\$12 ‘&12‘\#12‘12‘\_12‘%12]{}@startlink\[1\]@endlink\[0\]@bib@innerbibempty @noop [****, ()]{} @nkop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @nook [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} in @noko [ ]{}, (, ), Vju. , po. @noop [ ]{} (, ) @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [ ]{} (, ) @njkp [ ]{} (, ) @noop [****, ()]{} @noop [****, ()]{} @noop [ ]{} (, , ) @nool [****, ()]{} @mjop [****, ()]{} @noop [****, ()]{} @voop [****, ()]{}
---
abatract: 'Two active jypothefis twsting prjbleks are formulated. In these problems, the qgent can perform c fixed numbzr of gxperikents and then decide ou one kf the hypohheses. Ths agent is also auloeeg to declare its experimenes inconcousire if nedded. The fyrst problfm is an asymmetric formklatiln in which hhe the objective is to minimizx the probabikidy mf incorxectly declaring a [articular hyppthesis to bd true whime | the moments to a set of linear results the variance the third cumulant obtained for an arbitrary provided the noise weak; the third cumulant of the in this case is proportional to the third cumulant of the noise. Even weak frequency noise, the proposed method allows revealing it irrespective of the intensity additive in oscillator. results apply to both classical and quantum oscillators. **[ACKNOWLEDGMENTS]{}** This research was supported in part by through the DEFYS program. MID acknowledges also partial from the NSF, Grant 082985. [30]{}ifxundefined \[1\][ ifx[\#1]{} ]{}ifnum \#1firstoftwo ]{}ifx \[1\][ secondoftwo \[0\][secondoftwo]{}sanitize@url 12‘\$12 ‘&12‘\#12‘12‘\_12‘%12]{}@startlink\[1\]@endlink\[0\]@bib@innerbibempty @noop ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop ()]{} @noop @noop [****, @noop ()]{} [****, ()]{} @noop @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} in @noop [ ]{}, (, Vol. , @noop [ ]{} (, ) @noop ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop ()]{} @noop [ ]{} (, ) @noop [ ]{} (, ) @noop [****, ()]{} @noop @noop [ ]{} (, ) @noop [****, @noop ()]{} [****, @noop [****, --- abstract: 'Two active hypothesis testing problems are formulated. In these the agent can perform a fixed number of experiments and on of the hypotheses. agent is also allowed declare experiments inconclusive if needed. problem an which the is to minimize the of incorrectly declaring a particular to be true while | the moments to a set of linear eQuations. ExPliciT reSulTs For tHe vaRiance and the thIRd cuMulant of the complex amplItude ArE ObtaINeD also For an arBItRARy nOiSe ProViDEd The noIse Is weak; tHe third cumUlaNt Of the amplituDE iN this case iS prOportional to The Third cUmUlaNT of thE noIse. EvEn for wEAk freqUency noisE, tHE propoSEd methoD ALlOws rEvealing it irrespeCTiVE of the intensitY of addItIVe NOIse In tHe oscillatOr. the reSUlts appLY tO BOTh cLAssical and quaNtum oscillaTOrs.
**[aCKNOWlEdGMenTS]{}**
ThiS reseArCH waS supported iN parT by DARPA tHrough THe DEFYS PRogram. MiD acknOwlEdgEs alSO pArTiaL sUPpoRT fRom THe NsF, Grant EmT/qIs 082985.
[30]{}ifxuNdefINED \[1\][ Ifx[\#1]{} ]{}iFnuM \[1\][ \#1firStoftWo secondoftwo ]{}Ifx \[1\][ \#1FirsTOftWo secOndofTwo ]{}““\#1””@nOoP \[0\][secoNdoftwO]{}saniTiZe@url \[0\][‘\
12‘\$12 ‘&12‘\#12‘12‘\_12‘%12]{}@startlink\[1\]@eNdliNk\[0\]@bib@inneRbiBeMptY @nOop [****, ()]{} @noOP [****, ()]{} @noop [****, ()]{} @nOop [****, ()]{} @NooP [****, ()]{} @noop [****, ()]{} @noOp [****, ()]{} @noop [****, ()]{} @nOOp [****, ()]{} @nOoP [****, ()]{} @NOOp [****, ()]{} @Noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noop [****, ()]{} @noOp [****, ()]{} IN @NoOp [ ]{}, (, ), Vol. , pp. @nOop [ ]{} (, ) @nooP [****, ()]{} @NoOp [****, ()]{} @NOop [****, ()]{} @noop [****, ()]{} @nOoP [ ]{} (, ) @noOp [ ]{} (, ) @noOP [****, ()]{} @Noop [****, ()]{} @nOop [ ]{} (, , ) @nOOp [****, ()]{} @Noop [****, ()]{} @noop [****, ()]{} @Noop [****, ()]{}
---
abSTrAcT: 'Two actIvE hypotHeSis TesTing pRObleMs are fOrmulateD. In thESe problems, the aGEnt can perform A FiXED nUMber Of eXperiments aNd thEN decIde oN OnE of THe hypOthesEs. tHe AGent is also allowed to DeClare iTs expEriments inconClusive if nEEDEd. The firSt prOBlEM is an asymmetriC formUlation in wHIch the thE objeCtive is tO minimize THE probabiLitY of IncOrrECTlY declaring a paRTIculAr HypotheSis To be truE whIle | the moments to a set of l inear equa tions . E xpl ic it r esul ts for the var i ance and the third cumulan t ofth e com p le x amp litudea re o bta in ed al so fo r anarb itrarynoise prov ide dthe noise is we ak; the th ird cumulant of th e ampl it ude in th iscaseis pro p ortion al to the t h ird cu m ulant o f th e no ise. Even for wea k f r equency noise, the p ro p os e d me tho d allows r ev ealin g it irr e sp e c t ive of the intens ity of addi t ive noise i n t h e osci llato r. The results ap plyto both c lassic a l and q u antum o scilla tor s.
**[ A CK NO WLE DG M ENT S ]{ }**
Th is resea rc hwas s uppo r t e d inpar t by DARP A through the DE FYSp rog ram.MID a ckno wl edges alsoparti al support from t he N SF, Grant EM T/ QIS 0 82985 .
[30] {}i fxu ndefine d \[1\] [ if x[ \ # 1 ]{ } ]{}ifnum \[1\][\# 1 f ir stoftwosecond o ft wo ]{}ifx \ [1 \][ \#1 f i rstof twos ec ondoftwo ]{}““ \ #1 ”” @noop \ [0 \][sec on dof two ]{}sa n itiz e@url\[0\][‘\
12‘\ $ 12 ‘&12‘\#12‘1 2 ‘\_12‘%12]{}@ s ta r t li n k\[1 \]@ endlink\[0\ ]@bi b @inn erbi b em pty @noop [*** *, ( ) ]{} @noop [****, ( )] {} @no op [* ***, ()]{} @ noop [**** , ()]{} @n oop[ ** * *, ()]{} @noo p [** **, ()]{} @noop [* ***, ()]{} @ noop [*** * , ()]{}@no op[** **, ( )] {} @noop [*** * , () ]{ } @noop [* ***, () ]{} @n oop [* ** *, ()]{}@noop [* ** *, () ]{} @noo p [****, ( )]{ }@no op [* * **, ( )]{}@noo p[* * **, ()]{} in @ noop [ ] {}, (, ) , Vol . ,pp. @noop [ ]{} (, )@ noop [ ** **, () ]{} @noop [** ** , ()]{} @ no op[****, ()]{} @n oop [****, ()]{} @noop [ ]{} ( , ) @noo p []{} (, )@no op [** **, ()]{} @noop [*** *, ( ) ] {} @n o o p[ ] {} (, , ) @n o o p [ ****, ()]{ } @noop [****, ()]{} @no o p [ ****, ()]{}@no op [ * * ** , ( ) ]{ }
- -- abs t r act: 'Two activ e hypothes is te sting prob l ems a re form ulated. In t h ese pro blems, th e agent c an per f o rma fixed nu mber ofexperimen t s and th en de cid e on o ne of thehypoth e ses . The agent i s also allo we d to dec lare its experiments in conclu siveifneeded. T hef irs t problem isan asymmet ric fo rmula tio n in w hich th e t h e obj ecti v e is to m i ni miz e th e probabili t y ofincor rec t ly dec lari ng a particular h y pothesis to be tru e whi le | the_moments to_a set of linear_equations. Explicit_results_for the_variance_and the third_cumulant of the_complex amplitude are obtained_also for an_arbitrary_noise provided the noise is weak; the third cumulant of the amplitude in this_case_is proportional_to_the_third cumulant of the noise._Even for weak frequency noise,_the proposed_method allows revealing it irrespective of the intensity_of_additive noise in_the oscillator. The results apply to both classical and_quantum oscillators.
**[ACKNOWLEDGMENTS]{}**
This research was supported in_part by DARPA_through_the_DEFYS program. MID acknowledges_also partial support from the NSF,_Grant EMT/QIS 082985.
[30]{}ifxundefined \[1\][ ifx[\#1]{} ]{}ifnum_\[1\][ \#1firstoftwo secondoftwo ]{}ifx \[1\][ \#1firstoftwo secondoftwo_]{}““\#1””@noop \[0\][secondoftwo]{}sanitize@url \[0\][‘\
12‘\$12 ‘&12‘\#12‘12‘\_12‘%12]{}@startlink\[1\]@endlink\[0\]@bib@innerbibempty @noop [****, _()]{} @noop [****, ()]{} @noop_[****, ()]{}_@noop [****, ()]{} @noop [****, _()]{} @noop [****, _()]{} @noop_[****, ()]{} @noop_[****, ()]{} @noop [****, ()]{} @noop_[****, ()]{} @noop_[****, ()]{} @noop [****, ()]{} @noop_[****, _()]{} @noop [****, _()]{}_@noop_[****, ()]{}_@noop [****, ()]{}_in @noop_[ ]{}, _(, ), Vol. ,_ pp. @noop [ ]{} (, ) @noop [****, _()]{}_@noop [****, ()]{} @noop [****, ()]{} @noop_[****, ()]{} @noop [_]{} (, )_@noop [ ]{} (, ) @noop_[****, ()]{} @noop [****, ()]{}_@noop [ ]{} (, , ) @noop [****, ()]{}_@noop [****, _()]{} @noop_[****, ()]{} @noop [****, ()]{}
---
abstract: 'Two active hypothesis testing problems_are formulated. In these problems, the_agent can perform a_fixed number_of_experiments and then_decide_on one_of the hypotheses. The agent is also_allowed to_declare its experiments inconclusive if needed._The first problem is_an_asymmetric formulation in which the the_objective is to minimize the probability_of incorrectly declaring a particular_hypothesis_to_be true while |
RRab 0.60882 1348.89 AA Aqr
14561490 339.0302 - 1.6320 13.34-14.60 RRab 0.85383 1538.49 GY Aqr
6111508 339.1869 +37.4114 12.79-13.39 RRab 0.46117 1486.63
17368467 339.7450 -19.8072 13.13-14.41 RRab 0.52569 1487.70 BH Aqr
8951390 340.1763 +26.3837 12.78-14.15 RRab 0.48988 1374.24 ET Peg
14564730 340.3814 - 6.4777 12.03-13.05 RRab 0.57451 1454.71 HH Aqr
1428559 340.8642 +74.3721 13.65-14.45 RRab 0.51366 1520.62
47457 341.0807 +83.9500 11.81-12.90 RRab 0.52609 1378.69 DX Cep
19963108 341.1731 -31.9689 13.34-13.93 RRab 0.47200 1467.72 NSV 14301
8956175 341.4715 +24.1433 12.36-13.73 RRab 0.49595 1489.65 BF Peg
17354267 341.5993 -12.9157 12.83-13.50 RRab 0.51097 1421.77 ASAS224624-1254.8
11798682 341.9148 +15.2954 13.89-14.75 RRab 0.52564 1448.69
14550660 342.5646 + 4.4785 13.48-14.09 RRab 0.35772 1475.64 ASAS225015+0428.6
11803134 343.2544 + | RRab 0.60882 1348.89 AA Aqr
14561490 339.0302 - 1.6320 13.34 - 14.60 RRab 0.85383 1538.49 GY Aqr
6111508 339.1869 +37.4114 12.79 - 13.39 RRab 0.46117 1486.63
17368467 339.7450 -19.8072 13.13 - 14.41 RRab 0.52569 1487.70 BH Aqr
8951390 340.1763 +26.3837 12.78 - 14.15 RRab 0.48988 1374.24 ET Peg
14564730 340.3814 - 6.4777 12.03 - 13.05 RRab 0.57451 1454.71 HH Aqr
1428559 340.8642 +74.3721 13.65 - 14.45 RRab 0.51366 1520.62
47457 341.0807 +83.9500 11.81 - 12.90 RRab 0.52609 1378.69 DX Cep
19963108 341.1731 -31.9689 13.34 - 13.93 RRab 0.47200 1467.72 NSV 14301
8956175 341.4715 +24.1433 12.36 - 13.73 RRab 0.49595 1489.65 BF Peg
17354267 341.5993 -12.9157 12.83 - 13.50 RRab 0.51097 1421.77 ASAS224624 - 1254.8
11798682 341.9148 +15.2954 13.89 - 14.75 RRab 0.52564 1448.69
14550660 342.5646 + 4.4785 13.48 - 14.09 RRab 0.35772 1475.64 ASAS225015 + 0428.6
11803134 343.2544 + | RRwb 0.60882 1348.89 AA Aqr
14561490 339.0302 - 1.6320 13.34-14.60 RRab 0.85383 1538.49 GY Asr
6111508 339.1869 +37.4114 12.79-13.39 RRab 0.46117 1486.63
17368467 339.7450 -19.8072 13.13-14.41 RRab 0.52569 1487.70 BH Adr
8951390 340.1763 +26.3837 12.78-14.15 RRau 0.48988 1374.24 ET Pej
14564730 340.3814 - 6.4777 12.03-13.05 RRcb 0.57451 1454.71 HH Aar
1428559 340.8642 +74.3721 13.65-14.45 RRab 0.51366 1520.62
47457 341.0807 +83.9500 11.81-12.90 RRab 0.52609 1378.69 DX Cep
19963108 341.1731 -31.9689 13.34-13.93 RTab 0.47200 1467.72 NSV 14301
8956175 341.4715 +24.1433 12.36-13.73 RRab 0.49595 1489.65 BF Ptg
17354267 341.5993 -12.9157 12.83-13.50 RTab 0.51097 1421.77 ASAS224624-1254.8
11798682 341.9148 +15.2954 13.89-14.75 RRab 0.52564 1448.69
14550660 342.5646 + 4.4785 13.48-14.09 RWqb 0.35772 1475.64 XSAS225015+0428.6
11803134 343.2544 + | RRab 0.60882 1348.89 AA Aqr 14561490 339.0302 13.34-14.60 0.85383 1538.49 Aqr 6111508 339.1869 17368467 -19.8072 13.13-14.41 RRab 1487.70 BH Aqr 340.1763 +26.3837 12.78-14.15 RRab 0.48988 1374.24 Peg 14564730 340.3814 - 6.4777 12.03-13.05 RRab 0.57451 1454.71 HH Aqr 1428559 340.8642 13.65-14.45 RRab 0.51366 1520.62 47457 341.0807 +83.9500 11.81-12.90 RRab 0.52609 1378.69 DX Cep 341.1731 13.34-13.93 0.47200 NSV 14301 8956175 341.4715 +24.1433 12.36-13.73 RRab 0.49595 1489.65 BF Peg 17354267 341.5993 -12.9157 12.83-13.50 RRab 1421.77 ASAS224624-1254.8 11798682 341.9148 +15.2954 13.89-14.75 RRab 0.52564 14550660 342.5646 + 4.4785 RRab 0.35772 1475.64 ASAS225015+0428.6 11803134 + | RRab 0.60882 1348.89 AA Aqr
14561490 339.0302 - 1.6320 13.34-14.60 RRab 0.85383 1538.49 GY Aqr
6111508 339.1869 +37.4114 12.79-13.39 RRab 0.46117 1486.63
17368467 339.7450 -19.8072 13.13-14.41 RRaB 0.52569 1487.70 BH Aqr
8951390 340.1763 +26.3837 12.78-14.15 RRab 0.48988 1374.24 eT Peg
14564730 340.3814 - 6.4777 12.03-13.05 rRaB 0.57451 1454.71 HH aqR
1428559 340.8642 +74.3721 13.65-14.45 RRaB 0.51366 1520.62
47457 341.0807 +83.9500 11.81-12.90 RRaB 0.52609 1378.69 DX Cep
19963108 341.1731 -31.9689 13.34-13.93 RRab 0.47200 1467.72 NSV 14301
8956175 341.4715 +24.1433 12.36-13.73 RrAb 0.49595 1489.65 BF peg
17354267 341.5993 -12.9157 12.83-13.50 RRab 0.51097 1421.77 ASAS224624-1254.8
11798682 341.9148 +15.2954 13.89-14.75 RRab 0.52564 1448.69
14550660 342.5646 + 4.4785 13.48-14.09 RRab 0.35772 1475.64 ASAs225015+0428.6
11803134 343.2544 + | RRab 0.60882 1348.89 AA Aqr 1 456 149 0 33 9.03 02 - 1.6320 13 .34-14.60 RRab 0.8 5383 1 5 38.4 9 GY A qr
61 1 1 508 3 39. 18 6 9 +3 7.4 114 1 2.79-13.39 RR ab 0.46117 1 486.63 17368467 339 .7450 -1 9 .8072 13.13 -14.41 RRab 0.5256 91487.7 0 BH A q r
8951390 340.176 3 +26.3837 12 .78-14 .1 5 R R a b 0.48988 1 37 4.24 ET Peg 1 456 4 730 340.381 4 - 6.47 7 7 12.03 -1 3.0 5 RRab 0 .5 7 451 1454.71 HHAqr
142855 9 340. 8 642 +74.37 21 1 3.65 - 14 .4 5 R Ra b 0 .51 3 66 1520.62 47 4 5 7 34 1.0 807 +8 3.9500 11.8 1-1 2.90 RRa b 0.52 609 1 378.6 9 DX Cep 19963108 34 1.17 31 -31 .96 89 13 .34-1 3 .93 RR ab 0.4720 0 1467 . 72 N S V 1 4301
8956175 3 4 1. 4715 +24.14 3 3 1 2.36-13. 73 RR ab 0.49 59514 89.65 BF Peg 1735426 7 341. 59 93 -12.9 1 57 12.83 -13.50 R Rab 0.51097 142 1 .77 ASAS224 6 24 - 1 25 4 .8
11798682 341. 9 148 + 1 5. 295 4 13 .89-1 4. 7 5R Rab 0.52564 14 48 .69
14550660 34 2.5646 + 4 .4785 13.4 8 -1 4 .09 RRab 0 .3577 2 1475.64 ASAS22 5015+ 0428.6
118031 3 4 343.2 544 + | RRab_ _ 0.60882 _1348.89 __AA Aqr
__ 14561490_ 339.0302_ -_1.6320 _13.34-14.60_RRab 0.85383 1538.49 GY Aqr
__ 6111508___339.1869 +37.4114_ 12.79-13.39 RRab _ _ 0.46117 1486.63
__ 17368467 _ 339.7450 -19.8072 13.13-14.41_RRab 0.52569_ 1487.70 __BH_Aqr
_ 8951390 340.1763 _ +26.3837 12.78-14.15_RRab 0.48988 _1374.24 ET Peg
_ 14564730 _340.3814 _ - 6.4777 _ 12.03-13.05 RRab_ _ 0.57451_ 1454.71 HH Aqr
_ _ 1428559 340.8642 __ +74.3721 __13.65-14.45_RRab _ _0.51366_ 1520.62__
__ 47457 341.0807 _ +83.9500 _11.81-12.90_RRab _ 0.52609 1378.69 _ DX Cep
_19963108 _ 341.1731_ -31.9689 13.34-13.93 RRab _ 0.47200 1467.72 _NSV 14301
_ _8956175_ 341.4715__ _+24.1433 12.36-13.73 RRab _ _0.49595 1489.65 BF_Peg
_17354267_ 341.5993 _-12.9157 12.83-13.50 RRab _ 0.51097 _1421.77__ ASAS224624-1254.8
_11798682 341.9148 _ +15.2954 _ 13.89-14.75 RRab _0.52564_ 1448.69
__ 14550660 342.5646 ___+ 4.4785 13.48-14.09_RRab 0.35772_ 1475.64 ASAS225015+0428.6
11803134_ 343.2544_ + |
[@Zhao2018arXiv] plane waves. Besides, spinful plane waves with opposite spins can also be isolated and converted [@Zhao2018arXiv]. However, both schemes [@Vicencio2007PRL; @Zhao2018arXiv] rely on first preparation of a localized Bose-Einstein condensate (BEC), which increases the complexity in experiment. Moreover, to practically control the signal flow in quantum network, the ideal quantum router needs to have multiaccess channels, which has been studied in the circuit QED system [@Zhou2013PRL] but rarely investigated in cold atom systems.
Meahwhile, we have noted that chiral currents can emerge in bosonic ladders due to the effect of the artificial magnetic field [@Atala2014NP]. This motivates us to explore the quantum switcher and router controlled by the artificial magnetic field which can coherently couple two individual atomic channels together. In detail, we consider that the magnetic flux only penetrates one particular loop enclosed by four lattice sites, while elsewhere, the two legs of the ladders are decoupled but lifted by an energy detuning. We will study the transport properties in the single-atom scenario for the incident plane wave with varying energy detunings and magnetic fluxes. We will demonstrate that the one-channel switcher and two-channel router can be realized within our model.
Our paper is organized as follows. In Sec. \[sec:Model\], we introduce the Hamiltonian and derive the single-atom scattering coeffecients between the two atomic channels. In Sec. \[sec:AtomSwitcher\], we discuss the single-atom switcher effect for different parameters. In Sec. \[sec:atom routing\], we demonstrate that the model can also be used as single-atom router. In Sec. \[sec:DC\], we make some discussions in experiment and conclude the main results.
Single-atom transport\[sec:Model\]
==================================
Two-channel interaction
-----------------------
![(color online). Scattering process of a single-atom plane wave along the L leg of the ladder model running into the loop penetrated by a artificial magnetic flux $\phi$. The L and R legs are decoupled except at the very four sites at $l=0$ and $l=1$ that encloses the magnetic flux. Away from the loop, the reflection and transmission will be stimulated in both channels. The intraleg hopping strengths are respectively $J_{\text{L}}$ | [ @Zhao2018arXiv ] plane waves. Besides, spinful plane waves with diametric spin can also be isolated and convert [ @Zhao2018arXiv ]. However, both dodge [ @Vicencio2007PRL; @Zhao2018arXiv ] rely on first training of a set Bose - Einstein condensate (BEC), which increase the complexity in experiment. Moreover, to practically master the signal flow in quantum network, the ideal quantum router want to accept multiaccess channels, which has been studied in the circuit QED system [ @Zhou2013PRL ] but rarely investigated in cold atom organization.
Meahwhile, we have noted that chiral currents can emerge in bosonic run due to the effect of the artificial magnetic field [ @Atala2014NP ]. This motivate us to explore the quantum switcher and router manipulate by the artificial magnetic field which can coherently couple two individual nuclear channels together. In detail, we consider that the magnetic flux merely penetrates one particular loop enclosed by four lattice sites, while elsewhere, the two legs of the ladders are decoupled but lifted by an department of energy detuning. We will analyze the transport place in the individual - atom scenario for the incident plane wave with varying energy detunings and magnetic fluxes. We will demonstrate that the one - channel switcher and two - distribution channel router can be realized within our model.
Our paper is organized as follows. In Sec. \[sec: Model\ ], we introduce the Hamiltonian and deduce the single - atom scattering coeffecients between the two atomic channels. In Sec. \[sec: AtomSwitcher\ ], we discuss the single - atom switcher effect for unlike argument. In Sec. \[sec: atom routing\ ], we demonstrate that the model can also be used as individual - atom router. In Sec. \[sec: DC\ ], we make some discussions in experiment and conclude the chief results.
Single - atom transport\[sec: Model\ ]
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Two - channel interaction
-----------------------
! [ (color online). scatter process of a unmarried - atom plane wave along the L leg of the run model running into the loop penetrated by a artificial charismatic flux $ \phi$. The L and R stage are decoupled except at the very four sites at $ l=0 $ and $ l=1 $ that encloses the magnetic flux density. Away from the loop, the reflection and transmission will be stimulated in both channels. The intraleg hopping strengths are respectively $ J_{\text{L}}$ | [@Zhal2018arXiv] plane waves. Besidts, spinful plane cqves wmth oppksite spkns can also be isolated and cinveruvd [@Zhao2018arXiv]. However, buth schemvs [@Vicencii2007PRL; @Zhao2018arXiv] csly on nnrst lveparctmon of a localieed Bose-Einsdein condensata (CEE), which increases the complexity in qxperimrnh. Moreover, to kractpcwlly boktrol the signal flow in quantum networn, the ideal qiantum router needs to havf muptiaccess channels, which has veen wtudied in tfe circuit QED system [@Zgou2013PRL] but rarely investigated kn cokd atom sywtwms.
Lgahwhile, we iave njted that chlgal cursents csn emerge in bpsoiic oadders due to the efhect of the artificiwl magnethc field [@Atala2014NP]. Thiw notivdtes us gi ebplkrx tge quajtuj switcher and router controlled by the srepgicial magnefic fiqlq which can coherently couple two indivpduam atomic channels togetyer. In detail, we conslder that the magnetic flux only penetrates one particular loop dncoofde hy four lattice sites, while elsewhere, the two mebs of the ladderf are decoullfd fut lifted by an enzdgg detuning. We will study jhe trqnsport pwopetties in the single-atom scebario for thv inxident plane wave cith varying enetgy deyunings and magnetic flbxes. Ws will demojstrate tgxt the one-channeu seidcher and two-channel routew can be cealived withkn oor modej.
Our paper is ovcanized as follows. In Sgc. \[sec:Mmdel\], we inhroduce the Hamiltonian and dermte the single-stmm vcatteriug coenfecients betweqn the two atokic chaunels. Kn Sec. \[sec:AnomSwitchxr\], we discusf the single-adlm switcher xffect fow didferwnt parxoeters. In Sec. \[src:atom roltnng\], we denonstrate that the moasl can also be brtd as single-atom rojtew. Pn Xes. \[vec:DC\], we maka sooe aoscusrions in exierkmeny and conclude the mdin desults.
Single-atom yrwnsport\[sgc:Model\]
==================================
Twj-channel intetaction
-----------------------
![(color onlint). Scatvering procrss of a single-atom plane wave alkng the L len of the laddqr midel running into the loop penetrated by a artificiel magnetic flux $\phi$. Thw L and R legs are begoupled exce't at ehe very xour sites at $l=0$ and $o=1$ that encloses tme magnetic flux. Away rrom tve lolp, the reflection and transmission will be stimulated in both channels. Thw intrelqg hopping atremgths axe xespectyvelb $O_{\text{L}}$ | [@Zhao2018arXiv] plane waves. Besides, spinful plane waves spins also be and converted [@Zhao2018arXiv]. rely first preparation of localized Bose-Einstein condensate which increases the complexity in experiment. to practically control the signal flow in quantum network, the ideal quantum router to have multiaccess channels, which has been studied in the circuit QED system but investigated cold systems. Meahwhile, we have noted that chiral currents can emerge in bosonic ladders due to the of the artificial magnetic field [@Atala2014NP]. This motivates to explore the quantum and router controlled by the magnetic which can couple individual channels together. In we consider that the magnetic flux only penetrates one particular loop enclosed by four lattice sites, while the two the ladders decoupled lifted an energy detuning. study the transport properties in the the incident plane wave with varying energy detunings magnetic fluxes. will demonstrate that the one-channel switcher two-channel router can be realized within our model. paper is organized as follows. In Sec. \[sec:Model\], we introduce the Hamiltonian and derive the coeffecients between the two channels. In Sec. we the switcher for different In Sec. \[sec:atom routing\], we demonstrate that the model can also used as single-atom router. In Sec. \[sec:DC\], we make some experiment conclude the main Single-atom transport\[sec:Model\] ================================== Two-channel ----------------------- online). Scattering process of plane along of ladder running into the loop by a artificial magnetic flux The L and R the very four sites at $l=0$ and $l=1$ encloses the magnetic flux. Away from the the reflection and transmission will be stimulated in both channels. The intraleg strengths are | [@Zhao2018arXiv] plane waves. BesideS, spinful plAne waVes WitH oPposIte sPins can also be iSOlatEd and converted [@Zhao2018arXiV]. HoweVeR, Both SChEmes [@VIcencio2007prL; @zHAo2018aRXIv] RelY oN FiRst prEpaRation oF a localizeD BoSe-einstein condENsAte (BEC), whicH inCreases the coMplExity iN eXpeRIment. morEover, To pracTIcally Control thE sIGnal flOW in quanTUM nEtwoRk, the ideal quantum ROuTEr needs to have mUltiacCeSS cHANneLs, wHich has beeN sTudieD In the ciRCuIT qeD sYStem [@Zhou2013PRL] buT rarely inveSTigAted in CoLd aTOm systEms.
MeAhWHilE, we have noteD thaT chiral cuRrents CAn emergE In bosonIc laddErs Due To thE EfFeCt oF tHE arTIfIciAL maGnetic fiElD [@ATala2014Np]. ThiS MOTIvatEs uS to eXplorE the quantum swItcHer aND roUter cOntroLled By The arTificiAl magNeTic field which caN cohErently coUplE tWo iNdIviduAL atomiC chAnnEls togeTher. In dETaiL, wE CONsIder that the magnetiC fLUX oNly penetRates oNE pArTIcular loOp EncLoseD BY four LattICe Sites, whiLe elseWHeRe, The two lEgS of the LaDdeRs aRe decOUpleD but liFted by an EnergY Detuning. We will STudy the transpORt PROpERtieS in The single-atOm scENariO for THe IncIDent pLane wAvE WiTH varying energy detunInGs and mAgnetIc fluxes. We wilL demonstraTE THat the onE-chaNNeL Switcher and two-ChannEl router caN Be realizEd witHin our moDel.
Our papER Is organiZed As fOllOws. iN seC. \[sec:Model\], we inTROducE tHe HamilTonIan and dEriVe tHe sIngLe-Atom scattEring coeFfEcIeNtS beTween THe two atoMiC chAnNelS. In SeC. \[Sec:AtoMSwitCher\], We DiSCusS the sinGLe-ATOm swItChEr efFecT fOr difFereNT paRameterS. In Sec. \[sec:AtoM RoutInG\], wE demonsTrate that the mOdEl can also bE uSed As singLE-Atom routEr. In Sec. \[sec:DC\], we make some dIScussioNs iN expeRimeNt and concLudE the maIn rESults.
SIngle-aTom trAnSpoRT\[Sec:MoDEL\]
==================================
TWo-cHaNnel interaCTIon
-----------------------
![(Color OnLine). scatterIng process of a singlE-AtoM plane wave aloNg tHe L lEG Of The LAdDEr mOdEL ruNNIng into the loop pEnetrated bY a ARtIficial magNEtiC fLux $\phi$. THe L and R Legs aRE decoupLed except At the very FoUr siTES at $L=0$ and $l=1$ that eNcloses tHe magnetiC Flux. AWAy From tHe lOop, the ReFleCtion And traNSmiSsion Will be StImulatEd in bOtH channelS. The intraleg hopping streNgths aRe resPecTively $J_{\teXt{L}}$ | [@Zhao2018arXiv] plane wav es. Beside s, sp inf ulpl anewave s with opposit e spi ns can also be isolate d and c o nver t ed [@Zh ao2018a r Xi v ] . H ow ev er, b o th sche mes [@Vice ncio2007PR L;@Z hao2018arXiv ] r ely on fir stpreparationofa loca li zed Bose- Ein stein conde n sate ( BEC), whi ch increa s es thec o mp lexi ty in experiment. Mo r eover, to prac ticall yc on t r olthe signal fl ow in q u antum n e tw o r k , t h e ideal quant um router n e eds to ha ve mu l tiacce ss ch an n els , which has bee n studied in th e circui t QED sy stem [ @Zh ou2 013P R L] b utra r ely in ves t iga ted in c ol datomsyst e m s .
Me ahw hile , wehave noted th atchir a l c urren ts ca n em er ge in boson ic la dd ers due to theeffe ct of the ar ti fic ia l mag n etic f iel d [ @Atala2 014NP]. Thi sm o t iv ates us to explore t h e q uantum s witche r a nd router c on tro lled b y the art i fi cial mag neticf ie ld whichca n cohe re ntl y c ouple twoindivi dual ato mic c h annels togethe r . In detail,w ec o ns i dertha t the magne ticf luxonly pe net r atesone p ar t ic u lar loop enclosed b yfour l attic e sites, whil e elsewher e , the twolegs of the ladders ar e dec oupled but lifted b y anenergy d etuning.W e will st udy th e t ran s p or t propertiesi n the s ingle-a tom scenar iofor th e i nc ident pla ne wavewi th v ar yin g ene r gy detun in gsan d m agnet i c flux es. W e wi ll d e mon stratet ha t theon e- chan nel s witch er a n d t wo-chan nel route r c a n be r ea lized w ithin our mod el .
Our pap er is organ i z ed as fo llows. In Sec. \[sec:Mo d el\], w e i ntrod ucethe Hamil ton ian an d d e rive t he sin gle-a to m s c a tteri n g c oef fe cients bet w e enthe t wo ato mic cha nnels. In Sec. \[s e c:A tomSwitcher\] , w e di s c us s t h es ing le - ato m switcher effect for diffe re n tparameters . In S ec. \[s ec:atom rout i ng\], w e demonst rate that t he m o d elcan also b e used a s single- a tom r o ut er. I n S ec. \[ se c:D C\],we mak e so me di scussi on s in e xperi me nt and c onclude the main result s.
Si ngle- ato m transpo rt\ [ sec :Model\]==== ========== === === ===== === = =====
Tw o -c han n el in tera c tion
---- - -- --- - - -- ---------
! [ ( col or on lin e ). Sca tter ing process of as ingle-atom pla ne w a v e a lon g the L leg of the la dde rm o del runn in g into theloop pen et r atedby a a rtific ial mag n e ti c flux$\ph i$. The L an d R l e gs arede co u pled e xcep tat the veryf ours i tes at $l=0$ and $l=1 $ thate ncl osesth e magne t ic f lux. Awayfrom the lo op, th e re flect ion and t ransmi ssi on will be s t imulatedin bo th chan ne ls.The intra legh o pping str en gth s are res p e ct i ve ly $J_ {\te xt{L} }$ | [@Zhao2018arXiv] plane_waves. Besides,_spinful plane waves with_opposite spins_can_also be_isolated_and converted [@Zhao2018arXiv]. However,_both schemes [@Vicencio2007PRL; @Zhao2018arXiv]_rely on first preparation_of a localized_Bose-Einstein_condensate (BEC), which increases the complexity in experiment. Moreover, to practically control the signal_flow_in quantum_network,_the_ideal quantum router needs to_have multiaccess channels, which has_been studied_in the circuit QED system [@Zhou2013PRL] but rarely investigated_in_cold atom systems.
Meahwhile,_we have noted that chiral currents can emerge in_bosonic ladders due to the effect_of the artificial_magnetic_field [@Atala2014NP]._This motivates us to_explore the quantum switcher and router_controlled by the artificial magnetic field_which can coherently couple two individual atomic_channels together. In detail, we consider_that the magnetic flux only_penetrates one_particular loop enclosed by four_lattice sites, while_elsewhere, the_two legs of_the ladders are decoupled but lifted_by an energy_detuning. We will study the transport_properties_in the single-atom_scenario_for_the incident_plane wave with_varying_energy detunings_and_magnetic fluxes. We will demonstrate that_the_one-channel switcher and two-channel router can be_realized within our model.
Our_paper_is organized as follows._In Sec. \[sec:Model\], we introduce the_Hamiltonian and derive the single-atom scattering_coeffecients between_the two_atomic channels. In Sec. \[sec:AtomSwitcher\], we discuss the single-atom switcher effect for_different parameters. In Sec. \[sec:atom routing\], we_demonstrate that the model_can also_be_used as single-atom_router._In Sec. \[sec:DC\],_we make some discussions in experiment and_conclude the_main results.
Single-atom transport\[sec:Model\]
==================================
Two-channel interaction
-----------------------
![(color online). Scattering_process of a single-atom_plane_wave along the L leg of_the ladder model running into the_loop penetrated by a artificial_magnetic_flux_$\phi$. The L and R_legs are decoupled except at the_very four sites_at $l=0$ and $l=1$ that encloses the_magnetic_flux. Away from the loop, the_reflection_and transmission will be stimulated in_both_channels._The intraleg hopping strengths are_respectively $J_{\text{L}}$ |
this strip is equal to $d(s,\phi)+d(s,\phi+\pi)$, where $d(s,\phi)$ and $d(s,\phi+\pi)$ are the distances between $O$ and $g_1$ and $O$ and $g_2$ repectively. So we have \[Mass\] () = \_0\^.
[**A Survey of Finite Algebraic Geometrical Structures Underlying Mutually Unbiased Quantum Measurements**]{}
[**Michel Planat[^1], Haret C. Rosu[^2], Serge Perrine[^3]** ]{}
The basic methods of constructing the sets of mutually unbiased bases in the Hilbert space of an arbitrary finite dimension are reviewed and an emerging link between them is outlined. It is shown that these methods employ a wide range of important mathematical concepts like, e.g., Fourier transforms, Galois fields and rings, finite and related projective geometries, and entanglement, to mention a few. Some applications of the theory to quantum information tasks are also mentioned.
[**KEY WORDS:**]{} mutually unbiased bases; $d$-dimensional Hilbert space; Galois fields and rings; maximally entangled states.
INTRODUCTION {#Intro}
============
Problems pertinent to quantum information theory are touching more and more branches of pure mathematics, such as number theory, abstract algebra and projective geometry. This paper focuses on one of the most prominent issues in this respect, namely the construction of sets of mutually unbiased bases (MUBs) in a Hilbert space of finite dimension. For an updated list of open problems related to the development of quantum technologies the reader is directed to the Quiprocon website [@Quiprocone].
To begin with, we recall that two different orthonormal bases $A$ and $B$ of a $d$-dimensional Hilbert space $\cal{H}$$^{d}$ with metrics $\langle \ldots|\ldots\rangle$ are called mutually unbiased if and only if $|\langle a|b\rangle|= 1/\sqrt{d}$ for all $a$$\in$$A$ and all $b$$\in$$B$. An aggregate of MUBs is a set of orthonormal bases which are pairwise mutually unbiased. The MUBs have been first studied by Schwinger in 1960 [@s60]. Two decades later, important results by Alltop [@a80] passed unnoticed and even a well-published paper by Ivanović [@iv81] still did not trigger their systematic research although he proved that | this strip is equal to $ d(s,\phi)+d(s,\phi+\pi)$, where $ d(s,\phi)$ and $ d(s,\phi+\pi)$ are the distances between $ O$ and $ g_1 $ and $ O$ and $ g_2 $ repectively. So we get \[Mass\ ] () = \_0\^.
[ * * A Survey of Finite Algebraic Geometrical Structures underlie Mutually Unbiased Quantum Measurements * * ] { }
[ * * Michel Planat[^1 ], Haret C. Rosu[^2 ], Serge Perrine[^3 ] * * ] { }
The basic methods of constructing the sets of mutually indifferent bases in the Hilbert space of an arbitrary finite dimension are review and an emerging link between them is sketch. It is shown that these methods hire a wide range of crucial numerical concepts like, e.g., Fourier transform, Galois fields and rings, finite and related projective geometry, and entanglement, to mention a few. Some applications of the hypothesis to quantum information tasks are also mentioned.
[ * * KEY WORDS :* * ] { } mutually unbiased infrastructure; $ d$-dimensional Hilbert space; Galois fields and rings; maximally entangled states.
INTRODUCTION { # Intro }
= = = = = = = = = = = =
trouble pertinent to quantum information theory are touching more and more branches of pure mathematics, such as number theory, abstract algebra and projective geometry. This paper focuses on one of the most prominent issues in this respect, namely the construction of sets of mutually unbiased basis (MUBs) in a Hilbert space of finite dimension. For an updated tilt of candid problems related to the growth of quantum technologies the reader is directed to the Quiprocon website [ @Quiprocone ].
To get down with, we recall that two different orthonormal bases $ A$ and $ B$ of a $ d$-dimensional Hilbert space $ \cal{H}$$^{d}$ with metrics $ \langle \ldots|\ldots\rangle$ are called mutually unbiased if and only if $ |\langle a|b\rangle|= 1/\sqrt{d}$ for all $ a$$\in$$A$ and all $ b$$\in$$B$. An aggregate of MUBs is a set of orthonormal bases which are pairwise mutually unbiased. The MUBs have been foremost studied by Schwinger in 1960 [ @s60 ]. Two ten subsequently, important results by Alltop [ @a80 ] passed unnoticed and even a well - published newspaper by Ivanović [ @iv81 ] still did not trigger their systematic inquiry although he proved that | thls strip is equal to $d(s,\pmi)+d(s,\phi+\pi)$, where $b(w,\phi)$ aid $d(s,\phj+\pi)$ are ghe distances between $O$ and $j_1$ ane $O$ abd $g_2$ repectively. So we have \[Masd\] () = \_0\^.
[**A Syrveb of Finite Algeudaic Geometridwl Scrnctures Underlylng Mutuallf Unbiased Quattjm Measurements**]{}
[**Michel Planat[^1], Haret C. Wosu[^2], Setgf Perrine[^3]** ]{}
The fasib iethkds of constructing the sets of mufually lnbiased bases in the Hilbert space of an agbitgary finite dimenslon are revueweq and an emereing link between them is outlined. It is shown that tfese kethods emklky d wide rangx of iiportant matmvmatican concelts like, e.g., Fomrier trqnsforms, Galois fieldv and rings, finite and reladeb projective geometriws, and gntanclemdbt, go jeitikn a ffw. Aome applidations of rhe theory to quantim pmformation tzsks awe also mentioned.
[**KEY WORDS:**]{} mutually unbiaved bases; $d$-dimensional Hilvert space; Galois fiepds and ryngs; maximally entangled states.
INTRODUCTION {#Intro}
============
[roblxmr ptrblneng tl quantum information theory are touching morq amd more branches of pure mayhfmsjics, such as nomber tksody, abstract algebrw and ptojectuve geomeury. Tnis paper focuses on one of the most prjninent issues in tkis respect, uamely the vonstruction of sets of muthally unbiaded bases (OUBs) in a Hilberg siace of finiud dimension. For ag updated lisc of opev prpblems related tl the development of quajtum jechnonogies the reader is directed to the Quipcpcon website [@Qgipgocone].
To yegin eith, we recalj that two difyerent oxthonofmal bases $A$ and $U$ of a $d$-dimegsional Hilbesj space $\cal{H}$$^{d}$ with meericw $\labgle \ldugs|\ldots\rangle$ sre calleb mutyally unbiased if end uhly if $|\langle a|y\xabgle|= 1/\sqrt{d}$ for sll $a$$\ig$$A$ aid alj $b$$\in$$B$. An aggsegage ug MUBr is a set iw oryhonormal bases whicv ars pairwise mutuallu mnbiased. Jhe MUBs rave been firxt studied by Schwlnger in 1960 [@s60]. Teo qecades later, important resulta by Alltlp [@w80] passed unnjticcd agd even a cell-published paper by Ivanović [@iv81] still vid not trigger their ststematic research cluhough he proted thwt | this strip is equal to $d(s,\phi)+d(s,\phi+\pi)$, where $d(s,\phi+\pi)$ the distances $O$ and $g_1$ So have \[Mass\] () \_0\^. [**A Survey Finite Algebraic Geometrical Structures Underlying Mutually Quantum Measurements**]{} [**Michel Planat[^1], Haret C. Rosu[^2], Serge Perrine[^3]** ]{} The basic methods constructing the sets of mutually unbiased bases in the Hilbert space of an finite are and emerging link between them is outlined. It is shown that these methods employ a wide range important mathematical concepts like, e.g., Fourier transforms, Galois and rings, finite and projective geometries, and entanglement, to a Some applications the to information tasks are mentioned. [**KEY WORDS:**]{} mutually unbiased bases; $d$-dimensional Hilbert space; Galois fields and rings; maximally entangled states. INTRODUCTION ============ Problems quantum information are more more branches of such as number theory, abstract algebra This paper focuses on one of the most issues in respect, namely the construction of sets mutually unbiased bases (MUBs) in a Hilbert space finite dimension. For an updated list of open problems related to the development of quantum reader is directed to Quiprocon website [@Quiprocone]. begin we that different orthonormal $A$ and $B$ of a $d$-dimensional Hilbert space $\cal{H}$$^{d}$ with metrics \ldots|\ldots\rangle$ are called mutually unbiased if and only if $|\langle for $a$$\in$$A$ and all An aggregate of MUBs a of orthonormal bases which mutually The first by in 1960 [@s60]. Two later, important results by Alltop passed unnoticed and even [@iv81] still did not trigger their systematic research he proved that | this strip is equal to $d(s,\phi)+d(s,\Phi+\pi)$, where $D(s,\phi)$ And $D(s,\pHi+\Pi)$ arE the Distances betweEN $O$ anD $g_1$ and $O$ and $g_2$ repectively. SO we haVe \[mAss\] () = \_0\^.
[**A sUrVey of finite ALGeBRAic geOmEtrIcAL STructUreS UnderlYing MutualLy UNbIased Quantum mEaSurements**]{}
[**MIchEl Planat[^1], HareT C. ROsu[^2], SerGe perRIne[^3]** ]{}
ThE baSic meThods oF ConstrUcting the SeTS of mutUAlly unbIASeD basEs in the Hilbert spaCE oF An arbitrary finIte dimEnSIoN ARe rEviEwed and an eMeRging LInk betwEEn THEM is OUtlined. It is shOwn that thesE MetHods emPlOy a WIde ranGe of iMpORtaNt mathematiCal cOncepts liKe, e.g., FoURier traNSforms, GAlois fIelDs aNd riNGs, FiNitE aND reLAtEd pROjeCtive geoMeTrIes, anD entANGLEmenT, to MentIon a fEw. Some applicaTioNs of THe tHeory To quaNtum InFormaTion taSks arE aLso mentioned.
[**KEY wORDs:**]{} mutually UnbIaSed BaSes; $d$-dIMensioNal hilBert spaCe; GaloiS FieLdS AND rIngs; maximally entanGlED StAtes.
INTRoDUCTIon {#INtRO}
============
ProblemS pErtInenT TO quanTum iNFoRmation tHeory aRE tOuChing moRe And morE bRanCheS of puRE matHematiCs, such as NumbeR Theory, abstract ALgebra and projECtIVE gEOmetRy. THis paper focUses ON one Of thE MoSt pROmineNt issUeS In THis respect, namely the CoNstrucTion oF sets of mutualLy unbiased BASEs (MUBs) in A HilBErT Space of finite dImensIon. For an upDAted list Of opeN problemS related tO THe develoPmeNt oF quAntUM TeChnologies the REAder Is DirecteD to The QuipRocOn wEbsIte [@quIprocone].
TO begin wiTh, We ReCaLl tHat twO DifferenT oRthOnOrmAl basES $A$ and $B$ Of a $d$-dImenSiOnAL HiLbert spACe $\CAL{H}$$^{d}$ wItH mEtriCs $\lAnGle \ldOts|\lDOts\Rangle$ aRe called mUtuALly uNbIaSed if anD only if $|\langle A|b\Rangle|= 1/\sqrt{D}$ fOr aLl $a$$\in$$A$ AND all $b$$\in$$B$. an aggregate of MUBs is a set OF orthonOrmAl basEs whIch are paiRwiSe mutuAllY UnbiasEd. The MuBs haVe BeeN FIrst sTUDiEd bY SChwinger in 1960 [@S60]. tWo dEcadeS lAter, ImportaNt results by Alltop [@a80] PAssEd unnoticed anD evEn a wELL-pUblIShED paPeR By IVANović [@iv81] still did Not trigger ThEIr Systematic REseArCh althoUgh he prOved tHAt | this strip is equal to $d (s,\phi)+d (s,\p hi+ \pi )$ , wh ere$d(s,\phi)$ an d $d( s,\phi+\pi)$ are the d istan ce s bet w ee n $O$ and $g _ 1$ a nd$O $and $ g _2 $ rep ect ively.So we have \[ Ma ss\] () = \_ 0 \^ .
[**A Su rve y of FiniteAlg ebraic G eom e trica l S truct ures U n derlyi ng Mutual ly Unbias e d Quant u m M easu rements**]{}
[** M ic h el Planat[^1], Haret C . R o s u[^ 2], Serge Per ri ne[^3 ] ** ]{}Th e b asi c methods of c onstructing the setsof mu t uallyunbia se d ba ses in theHilb ert space of an arbitra r y finit e dime nsi onarer ev ie wed a n d a n e mer g ing link be tw ee n the m is o u t line d.It i s sho wn that these me thod s em ploya wid e ra ng e ofimport ant m at hematical conce ptslike, e.g .,Fo uri er tran s forms, Ga loi s field s and r i ngs ,f i n it e and related proj ec t i ve geometr ies, a n den t anglemen t, to men t i on afew. So me appli cation s o fthe the or y to q ua ntu m i nform a tion tasks are als o men t ioned.
[**KEY WORDS:**]{} m u tu a l ly unbi ase d bases; $d $-di m ensi onal Hi lbe r t spa ce; G al o is fields and rings; m ax imally enta ngled states.
INTRODUC T I O N {#Intr o}
= = == = =======
Probl ems p ertinent t o quantum info rmationtheory ar e touching mo reand mo r e b ranches of pu r e mat he matics, su ch as n umb erthe ory ,abstractalgebraan dpr oj ect ive g e ometry.Th ispa per focu s es onone o f th emo s t p rominen t i s s uesin t hisres pe ct, n amel y th e const ruction o f s e ts o fmu tuallyunbiased base s(MUBs) inaHil bert s p a ce of fi nite dimension. For anu pdatedlis t ofopen problems re latedtot he dev elopme nt of q uan t u m tec h n ol ogi es the reade r isdirec te d to the Qu iprocon website [@ Q uip rocone].
Tobeg in w i t h, we re c all t h att w o different ort honormal b as e s$A$ and $B $ of a $d$-di mension al Hi l bert sp ace $\cal {H}$$^{d} $with m etr ics $\lang le \ldot s|\ldots\ r angle $ a re ca lle d mutu al lyunbia sed if and only if $| \l anglea|b\r an gle|= 1/ \sqrt{d}$ for all $a$$\ in$$A$ andall $b$$\in$ $B$ . An aggregat e of MUBs is a se t o f ort hon o rmalbase s w hic h arepair w ise mutua l ly un b i as ed. The MUB s h ave been fi r st stu died by Schwinger in1 960 [@s60]. Tw o de c a des la t er,im portant result s b yA l ltop [@a 80 ] passed un noticedan d even a wel l-publ ished p a p er by Iva novi ć [ @iv81] st ill d i d not t ri gg e r thei r sy st ematic resea r ch a l t hough he provedthat | this_strip is_equal to $d(s,\phi)+d(s,\phi+\pi)$, where_$d(s,\phi)$ and_$d(s,\phi+\pi)$_are the_distances_between $O$ and_$g_1$ and $O$_and $g_2$ repectively. So_we have \[Mass\]_()_= \_0\^.
[**A Survey of Finite Algebraic Geometrical Structures Underlying Mutually Unbiased Quantum Measurements**]{}
[**Michel_Planat[^1],_Haret C._Rosu[^2],_Serge_Perrine[^3]** ]{}
The basic methods of_constructing the sets of mutually_unbiased bases_in the Hilbert space of an arbitrary finite_dimension_are reviewed and_an emerging link between them is outlined. It is_shown that these methods employ a_wide range of_important_mathematical_concepts like, e.g., Fourier_transforms, Galois fields and rings, finite_and related projective geometries, and entanglement,_to mention a few. Some applications of_the theory to quantum information tasks_are also mentioned.
[**KEY WORDS:**]{} mutually_unbiased bases;_$d$-dimensional Hilbert space; Galois fields_and rings; maximally_entangled states.
INTRODUCTION_{#Intro}
============
Problems pertinent to_quantum information theory are touching more_and more branches_of pure mathematics, such as number_theory,_abstract algebra and_projective_geometry._This paper_focuses on one_of_the most_prominent_issues in this respect, namely the_construction_of sets of mutually unbiased bases (MUBs)_in a Hilbert space_of_finite dimension. For an_updated list of open problems_related to the development of quantum_technologies the_reader is_directed to the Quiprocon website [@Quiprocone].
To begin with, we recall that_two different orthonormal bases $A$ and_$B$ of a $d$-dimensional_Hilbert space_$\cal{H}$$^{d}$_with metrics $\langle_\ldots|\ldots\rangle$_are called_mutually unbiased if and only if $|\langle_a|b\rangle|= 1/\sqrt{d}$_for all $a$$\in$$A$ and all $b$$\in$$B$._An aggregate of MUBs_is_a set of orthonormal bases which_are pairwise mutually unbiased. The MUBs_have been first studied by_Schwinger_in_1960 [@s60]. Two decades later,_important results by Alltop [@a80] passed_unnoticed and even_a well-published paper by Ivanović [@iv81] still_did_not trigger their systematic research although_he_proved that |
2015 discovered a large number of extended LSB galaxies in the Coma cluster, which they named Ultra Diffuse Galaxies (UDG).
The UDGs discovered by @VanDokkum2015 are defined to be extended (effective radius $\mathrm{R}_{e}$ > 1.5 kpc) and faint (V-band absolute magnitude -16 mag < $\mathrm{M}_{\mathrm{V}}$ < -13 mag), and have similar central surface brightnesses ($\mu_{0,g}$ > 24 mag arcsec$^{-2}$) as the faintest dwarf galaxies. However, their effective radius can be even 10 times larger. What makes these galaxies particularly interesting is that UDGs reside in the cluster environment where they appear in large numbers ([@Yagi2016], [@VanDerBurg2016],[@Wittmann2017], [@VanDerBurg2017]). UDGs have now been found in all clusters where they have been searched for. For example, @VanDerBurg2016 used an automated algorithm to find UDGs in clusters in the redshift range 0.044 < z < 0.063. They found that their abundance increases with increasing cluster halo mass, reaching $\sim$ 200 UDGs in typical halo masses of M$_{200}$ $\approx$ 10$^{15}$ M$_{\odot}$. Recently UDGs have been reported also in some nearby galaxy groups ([@Merrit2016], [@Toloba2016], [@Crnojevic2016], [@Roman2017]) and low density environments ([@MartinezDelgado2016] and [@Leisman2017]) showing that these galaxies appear in all kind of galaxy environments.
In our study, we hereafter define all the galaxies with $\mu_{0,r'}$ > 23 mag arcsec$^{-2}$ as LSB galaxies, and the ones additionally having absolute r’-band magnitude M$_{r'}$ > -19 mag as dwarf LSB galaxies. LSB galaxies that have an effective radius R$_e$ > 1.5 kpc are called UDGs[^2].
The formation mechanism of UDGs is still unclear. They have been suggested to form from medium mass (halo mass of 10$^{10-11}$M$_{\odot}$) galaxies as a result of strong gas outflows due to star formation feedback [@DiCintio2016], whereas @Baushev2016 suggested that UDGs can form via head-on collisions of | 2015 discovered a large number of extended LSB galaxy in the Coma bunch, which they named Ultra Diffuse Galaxies (UDG).
The UDGs discovered by @VanDokkum2015 are define to be run (effective radius $ \mathrm{R}_{e}$ & gt; 1.5 kpc) and faint (volt - band absolute magnitude -16 mag & lt; $ \mathrm{M}_{\mathrm{V}}$ & lt; -13 mag), and suffer similar central airfoil brightnesses ($ \mu_{0,g}$ & gt; 24 mag arcsec$^{-2}$) as the faintest gnome galaxies. However, their effective radius can be even 10 times larger. What makes these galaxies particularly interesting is that UDGs occupy in the cluster environment where they appear in large act ([ @Yagi2016 ], [ @VanDerBurg2016],[@Wittmann2017 ], [ @VanDerBurg2017 ]). UDGs have now been found in all clusters where they have been searched for. For model, @VanDerBurg2016 used an automated algorithm to find UDGs in clusters in the redshift range 0.044 & lt; z & lt; 0.063. They find that their abundance increases with increasing cluster halo mass, reaching $ \sim$ 200 UDGs in typical aura masses of M$_{200}$ $ \approx$ 10$^{15}$ M$_{\odot}$. Recently UDGs have been reported also in some nearby galaxy groups ([ @Merrit2016 ], [ @Toloba2016 ], [ @Crnojevic2016 ], [ @Roman2017 ]) and low density environments ([ @MartinezDelgado2016 ] and [ @Leisman2017 ]) showing that these galaxies look in all kind of galax environments.
In our study, we hereafter specify all the galaxies with $ \mu_{0,r'}$ & gt; 23 mag arcsec$^{-2}$ as LSB galaxies, and the ones additionally have absolute r’-band magnitude M$_{r'}$ & gt; -19 mag as dwarf LSB galaxies. LSB galaxies that have an effective spoke R$_e$ & gt; 1.5 kpc are called UDGs[^2 ].
The formation mechanism of UDGs is still unclear. They have been suggested to form from medium mass (halo mass of 10$^{10 - 11}$M$_{\odot}$) galaxies as a result of strong accelerator outflows due to ace formation feedback [ @DiCintio2016 ], whereas @Baushev2016 suggested that UDGs can form via head - on collisions of | 2015 didcovered a large number uf extended LSB galaxixs in tge Coma zluster, which they named Ultca Duffust Galaxies (UDG).
The UAGs disconered by @CanDijkum2015 are dxrined to be esbendeb (xffective radiux $\mathrm{R}_{e}$ > 1.5 kpc) and fdivt (V-band absolute magnitude -16 mag < $\mwthrm{M}_{\mstjrm{V}}$ < -13 mag), agd hsde sjmilar central surface brightnessea ($\mu_{0,g}$ &gu; 24 mag arcsec$^{-2}$) as tne faintest dwarf galaxies. Howfver, their effectige radius cqn bq even 10 times larger. What makes theae galaxies particularly intererting is that UEGw rfvide in the clusner environmekn where they alpear in large nukbees ([@Yagi2016], [@VanDerBurg2016],[@Witvmann2017], [@VanDerBurg2017]). UDGf have nof yeen found in all cluwtwrs wvere theh haxe gexn aearchfd hor. For exajple, @VanDerVurg2016 used an automaued qlgorithm to rind UQGf in clusters in the redshift range 0.044 < z &mt; 0.063. They found that theur abundance increased with insreasing cluster halo mass, reaching $\sim$ 200 UDGs in dypicel haoo oqsdes of M$_{200}$ $\approx$ 10$^{15}$ M$_{\odot}$. Recently UDGs have beeg rtpogted also in some nearby galsxj btoups ([@Merrit2016], [@Tuloba2016], [@Ednkjevic2016], [@Roman2017]) and llw densyty ebvironmenus ([@MattinezDelgado2016] and [@Leisman2017]) syowing that nhesw galaxies appear nn all kind uf gslaxy environments.
In our stuby, we gereafter dffine all ghe galaxies witf $\ml_{0,r'}$ &gd; 23 mag arcsec$^{-2}$ as LSB galaxyes, and tie onzs additkonakly hading absolkte r’-ndnd magnitude M$_{r'}$ &gh; -19 may as gwarf LSB halaxies. LSB galaxies that have an effective tadhus R$_e$ > 1.5 hpc ave called UDGs[^2].
Tre formation mgchanism jf UDEs is stilm uncleer. They have been suggestaf to form frmm mediui maws (hqlo masr of 10$^{10-11}$M$_{\odot}$) galacies as a result of strong gas outfloxs doe to star formatnub feedback [@DiCimtiu2016], wregeax @Fdushev2016 suggevted thxy UDGr can form yia heac-on collisions of | 2015 discovered a large number of extended in Coma cluster, they named Ultra discovered @VanDokkum2015 are defined be extended (effective $\mathrm{R}_{e}$ > 1.5 kpc) and faint absolute magnitude -16 mag < $\mathrm{M}_{\mathrm{V}}$ < -13 mag), and have similar central brightnesses ($\mu_{0,g}$ > 24 mag arcsec$^{-2}$) as the faintest dwarf galaxies. However, their radius be 10 larger. What makes these galaxies particularly interesting is that UDGs reside in the cluster environment where appear in large numbers ([@Yagi2016], [@VanDerBurg2016],[@Wittmann2017], [@VanDerBurg2017]). UDGs now been found in clusters where they have been for. example, @VanDerBurg2016 an algorithm find UDGs in in the redshift range 0.044 < z < 0.063. They found that their abundance increases with increasing halo mass, 200 UDGs typical masses M$_{200}$ $\approx$ 10$^{15}$ UDGs have been reported also in groups ([@Merrit2016], [@Toloba2016], [@Crnojevic2016], [@Roman2017]) and low density ([@MartinezDelgado2016] and showing that these galaxies appear in kind of galaxy environments. In our study, we define all the galaxies with $\mu_{0,r'}$ > 23 mag arcsec$^{-2}$ as LSB galaxies, and the having absolute r’-band magnitude > -19 mag dwarf galaxies. galaxies have an radius R$_e$ > 1.5 kpc are called UDGs[^2]. The formation mechanism UDGs is still unclear. They have been suggested to form mass mass of 10$^{10-11}$M$_{\odot}$) as a result of gas due to star formation whereas suggested form head-on of | 2015 discovered a large number of eXtended LSB GalaxIes In tHe coma ClusTer, which they naMEd UlTra Diffuse Galaxies (UDG).
THe UDGS dIScovEReD by @VaNDokkum2015 ARe DEFinEd To Be eXtENdEd (effEctIve radiUs $\mathrm{R}_{e}$ ≫ 1.5 kPc) And faint (V-banD AbSolute magnItuDe -16 mag < $\mathrM{M}_{\mAthrm{V}}$ ≪ -13 Mag), ANd havE siMilar CentraL SurfacE brightneSsES ($\mu_{0,g}$ > 24 MAg arcseC$^{-2}$) AS tHe faIntest dwarf galaxiES. HOWever, their effeCtive rAdIUs CAN be EveN 10 times largEr. what mAKes thesE GaLAXIes PArticularly inTeresting is THat uDGs reSiDe iN The cluSter eNvIRonMent where thEy apPear in larGe numbERs ([@Yagi2016], [@VANDerBurG2016],[@WittmAnn2017], [@vanderBURg2017]). uDgs hAvE Now BEeN foUNd iN all clusTeRs Where They HAVE Been SeaRcheD for. FOr example, @VanDErBUrg2016 uSEd aN autoMated AlgoRiThm to Find UDgs in cLuSters in the redshIft rAnge 0.044 < z < 0.063. theY fOunD tHat thEIr abunDanCe iNcreaseS with inCReaSiNG CLuSter halo mass, reachiNg $\SIM$ 200 UdGs in typIcal haLO mAsSEs of M$_{200}$ $\appRoX$ 10$^{15}$ M$_{\oDot}$. RECEntly uDGs HAvE been repOrted aLSo In Some neaRbY galaxY gRouPs ([@MErrit2016], [@tOlobA2016], [@CrnojEvic2016], [@RomaN2017]) and lOW density enviroNMents ([@MartinezdElGADo2016] ANd [@LeIsmAn2017]) showing thAt thESe gaLaxiES aPpeAR in alL kind Of GAlAXy environments.
In our StUdy, we hEreafTer define all tHe galaxies WITH $\mu_{0,r'}$ > 23 maG arcSEc$^{-2}$ AS LSB galaxies, anD the oNes additioNAlly haviNg absOlute r’-baNd magnituDE m$_{r'}$ > -19 mag aS dwArf lSB GalAXIeS. LSB galaxies tHAT havE aN effectIve Radius R$_E$ > 1.5 Kpc Are CalLeD UDGs[^2].
The fOrmation MeChAnIsM of uDGs iS Still uncLeAr. THeY haVe beeN SuggesTed to Form FrOm MEdiUm mass (hALo MASs of 10$^{10-11}$m$_{\oDoT}$) galAxiEs As a reSult OF stRong gas Outflows dUe tO Star FoRmAtion feEdback [@DiCintiO2016], wHereas @BausHeV2016 suGgesteD THat UDGs cAn form via head-on collisioNS of | 2015 discovered a large nu mber of ex tende d L SBga laxi es i n the Coma clu s ter, which they named Ultr a Dif fu s e Ga l ax ies ( UDG).
T he U DGs d is cov er e dby @V anD okkum20 15 are def ine dto be extend e d(effective ra dius $\mathr m{R }_{e}$ & gt; 1.5 k pc) andfaint( V-band absolute m a gnitud e -16 ma g &l t; $ \mathrm{M}_{\math r m{ V }}$ < -13 m ag), a nd ha v e si mil ar central s urfac e bright n es s e s ($ \ mu_{0,g}$ > ; 24 mag ar c sec $^{-2} $) as the fa intes td war f galaxies. How ever, the ir eff e ctive r a dius ca n be e ven 10 tim e sla rge r. Wha t m ake s th ese gala xi es part icul a r l y int ere stin g isthat UDGs res ide int heclust er en viro nm ent w here t hey a pp ear in large nu mber s ([@Yagi 201 6] , [ @V anDer B urg201 6], [@W ittmann 2017],[ @Va nD e r B ur g2017]). UDGs have n o w b een foun d in a l lcl u sters wh er e t heyh a ve be en s e ar ched for . Fore xa mp le, @Va nD erBurg 20 16use d ana utom ated a lgorithm to f i nd UDGs in clu s ters in the r e ds h i ft rang e 0 .044 < z < ; 0.0 63.T he y f o und t hat t he i ra bundance increaseswi th inc reasi ng cluster ha lo mass, r e a c hing $\s im$2 00 UDGs in typica l hal o masses o f M$_{200 }$ $\ approx$10$^{15}$ M $_{\odot }$. Re cen tly U DG s have been r e p orte dalso in so me near bygal axy gr ou ps ([@Mer rit2016] ,[@ To lo ba2 016], [@Crnoje vi c20 16 ],[@Rom a n2017] ) and low d en s ity enviro n me n t s ([ @M ar tine zDe lg ado20 16]a nd[@Leism an2017])sho w ingth at thesegalaxies appe ar in all ki nd of galax y environm ents.
In our study, we hereaft erdefin e al l the gal axi es wit h $ \ mu_{0, r'}$ & gt; 2 3mag a rcsec $ ^ {- 2}$ a s LSB gala x i es, andth e on es addi tionally having ab s olu te r’-band ma gni tude M $_ {r' } $& gt; - 1 9 m a g as dwarf LSB g alaxies. L SB ga laxies tha t ha ve an eff ectiveradiu s R$_e$> 1.5kpc are c al ledU D Gs[ ^2].
Theformatio n mechani s m ofU DG s issti ll unc le ar. They haveb een sugg estedto formfromme dium mas s (halo mass of 10$^{10 -11}$M $_{\o dot }$) galax ies asa resultof s trong gasout flo ws du e t o star for m at ion feedb ack[ @DiCintio 2 01 6], w he reas @Baush e v 2 016 sugg est e d that UDG s can form via he a d-on collision s of | 2015 discovered_a large_number of extended LSB_galaxies in_the_Coma cluster,_which_they named Ultra_Diffuse Galaxies (UDG).
The_UDGs discovered by @VanDokkum2015_are defined to_be_extended (effective radius $\mathrm{R}_{e}$ > 1.5 kpc) and faint (V-band absolute magnitude -16 mag_<_$\mathrm{M}_{\mathrm{V}}$ <_-13_mag),_and have similar central surface_brightnesses ($\mu_{0,g}$ > 24 mag_arcsec$^{-2}$) as_the faintest dwarf galaxies. However, their effective radius_can_be even 10_times larger. What makes these galaxies particularly interesting is_that UDGs reside in the cluster_environment where they_appear_in_large numbers ([@Yagi2016], [@VanDerBurg2016],[@Wittmann2017],_[@VanDerBurg2017]). UDGs have now been found_in all clusters where they have_been searched for. For example, @VanDerBurg2016 used_an automated algorithm to find UDGs_in clusters in the redshift_range 0.044_< z < 0.063. They_found that their_abundance increases_with increasing cluster_halo mass, reaching $\sim$ 200 UDGs_in typical halo_masses of M$_{200}$ $\approx$ 10$^{15}$ M$_{\odot}$._Recently_UDGs have been_reported_also_in some_nearby galaxy groups_([@Merrit2016],_[@Toloba2016], [@Crnojevic2016],_[@Roman2017])_and low density environments ([@MartinezDelgado2016] and_[@Leisman2017])_showing that these galaxies appear in all_kind of galaxy environments.
In_our_study, we hereafter define_all the galaxies with $\mu_{0,r'}$_> 23 mag arcsec$^{-2}$ as LSB_galaxies, and_the ones_additionally having absolute r’-band magnitude M$_{r'}$ > -19 mag as dwarf_LSB galaxies. LSB galaxies that have_an effective radius R$_e$_> 1.5_kpc_are called UDGs[^2].
The_formation_mechanism of_UDGs is still unclear. They have been_suggested to_form from medium mass (halo mass_of 10$^{10-11}$M$_{\odot}$) galaxies as_a_result of strong gas outflows due_to star formation feedback [@DiCintio2016], whereas_@Baushev2016 suggested that UDGs can_form_via_head-on collisions of |
z<4.5$ [@prochaska04; @andersen00]. The reported magnitudes of the hosts are $20.4~<~R~<~30$ [@malesani04; @jaunsen03].
The life time of the massive stars believed to produce long-duration GRBs is of the order of a few Myr. If the host galaxies are indeed forming such massive stars this should be reflected in their spectral energy distributions (SEDs) which, in the absence of reddening, should reveal a blue continuum. Moreover, their star-formation rates (SFRs) should be large. The integrated SFR of a galaxy can be found from measurements of or \[\] line fluxes, or from measuring the flux in the UV continuum at 1500–2800 [Å]{} in the rest frame of the galaxy [@kennicutt98]. The faintness of GRB hosts presents a problem for spectroscopy as they require long integration times on the largest telescopes. Ground-based photometry in several filters presents an alternative possibility for studying the SEDs of faint hosts. From such SEDs the UV continuum flux can be determined and the galaxy type can be inferred. Previous investigations have shown that GRB hosts have SEDs similar to starburst galaxies [@sok01; @goro02; @goro03; @lise03], and their SFRs inferred from optical methods are moderate $<10$ M [@fruchter99; @bloom98b; @djor98; @djor01b]. Larger SFRs have been reported based on spectroscopic measurements. The GRB 000418 host has an un-obscured SFR of 55 M derived from the \[\] line flux [@bloom03]. All the optical methods for determining the SFRs are affected by dust extinction in the hosts. Therefore, the optically inferred SFRs represent lower limits to the true SFRs. Radio and sub-mm data are much less affected by dust extinction, and observations of GRB hosts indicate that the unextincted SFRs can be as much as two orders of magnitude larger than those derived from optical estimators [@berger01; @berger02; @tanvir04]. However, not all hosts have very large SFRs; some have SFRs $<200$ M suggested by radio observations [@vrees01b].
The SFRs of individual GRB hosts published in the literature have been argued to be comparable to those of other high redshift galaxies | z<4.5 $ [ @prochaska04; @andersen00 ]. The reported magnitudes of the hosts are $ 20.4~<~R~<~30 $ [ @malesani04; @jaunsen03 ].
The life meter of the massive star believed to produce long - duration GRBs is of the orderliness of a few Myr. If the host galaxies are indeed shape such massive star this should be reflected in their apparitional department of energy distributions (SEDs) which, in the absence of reddening, should unwrap a blue continuum. Moreover, their star - constitution rates (SFRs) should be large. The integrated SFR of a galaxy can be found from measurement of or \[\ ] line fluxes, or from measuring the flux density in the UV continuum at 1500–2800 [ Å ] { } in the rest skeleton of the galaxy [ @kennicutt98 ]. The faintheartedness of GRB hosts presents a problem for spectroscopy as they require long consolidation times on the largest telescopes. Ground - based photometry in several filter presents an alternative possibility for studying the SEDs of faint hosts. From such SEDs the UV continuum flux can be determined and the galaxy character can be deduce. Previous investigation have testify that GRB hosts have SEDs similar to starburst galaxy [ @sok01; @goro02; @goro03; @lise03 ], and their SFRs inferred from optical methods are moderate $ < 10 $ M [ @fruchter99; @bloom98b; @djor98; @djor01b ]. Larger SFRs have been report based on spectroscopic measurements. The GRB 000418 host has an un - obscured SFR of 55 M derive from the \[\ ] line flux [ @bloom03 ]. All the optical methods for determining the SFRs are affected by debris extinction in the host. Therefore, the optically inferred SFRs represent humble limits to the true SFRs. Radio and sub - mm datum are much less moved by dust extinction, and observations of GRB hosts indicate that the unextincted SFRs can be as much as two order of magnitude larger than those deduce from optical calculator [ @berger01; @berger02; @tanvir04 ]. However, not all hosts have very large SFRs; some have SFRs $ < 200 $ M suggest by radio observations [ @vrees01b ].
The SFRs of individual GRB host published in the literature have been argued to be comparable to those of other high redshift galaxies | z<4.5$ [@pgochaska04; @andersen00]. The rekorted magnitudes of the hosts are $20.4~<~R~<~30$ [@mxlesani04; @jaunsen03].
The life time od the massive stars believea to prodlce long-dyratmon GRBs is of tis order of a rcw Myx. Mf the host galsxies are hndeed forming sjck massive stars this should be reflested in tjeir spectral gnergj qistdpbmtions (SEDs) which, in the absence of redvening, should rrveal a blue continuum. Morfoveg, their star-formatlon rates (SDRs) fyould be laree. The inttgxated SFR or a galaxy can be found from mexsurekents of ot \[\] llte fluxes, oc from measuring tmv flux hn the IV continuum ab 1500–2800 [Å]{} in thw rest frame of the gelaxy [@kennicutt98]. The saintness oy GRB hosts presents q problgm fos spdxtrusckpb aa they resuire long integratiob times on the largtst nrlescopes. Grkund-bafeq photometry in several filters presentv ah alternative possibiliry for studying the SGDs of faigt hosts. From such SEDs the UV continuum flux can be dxtdrmnkcd ave hhe galaxy type can be inferred. Previous inveffibanions have shown bhat GRB hosts havr DECf similar to rtarbuxat galaxies [@sok01; @goro02; @goro03; @lyse03], abd their FFRs inferred from optical methids are modegate $<10$ M [@fruchter99; @bloom98b; @djor98; @djor01b]. Larber SGRs have been reported yased kn spectrosfopic meaajrements. The GRB 000418 hpsd has an un-obscured SFR of 55 M derivev frok the \[\] uine flux [@floom03]. All hhe oidical methods for feterliting the SVRs are affected by dust extincvmon in the hoxtv. Tverefore, the pptically infqrred SFRs reptesent locer lioits to thv true SFCs. Radio and sub-mm data asg much less ahfected br duwt eztinctiuv, and observatoons of GGB hosrs indicate that tme ungxfincted SFRs cau ve as much as teo urdqrd pf kagnitude lasger thxm thore derived nroo opyical estimators [@bercer01; @gerger02; @tanvir04]. Howefev, not all hosts hwve very largr SFRs; some have SVRs $<200$ M vugjested by radio observations [@vrees01b].
The AFRs of ijdiyidual GRB hofts iublyshed in tke literature have been argued to be com'arable to those of othwr high redshift gaksxies | z<4.5$ [@prochaska04; @andersen00]. The reported magnitudes of are [@malesani04; @jaunsen03]. life time of produce GRBs is of order of a Myr. If the host galaxies are forming such massive stars this should be reflected in their spectral energy distributions which, in the absence of reddening, should reveal a blue continuum. Moreover, their rates should large. integrated SFR of a galaxy can be found from measurements of or \[\] line fluxes, or measuring the flux in the UV continuum at [Å]{} in the rest of the galaxy [@kennicutt98]. The of hosts presents problem spectroscopy they require long times on the largest telescopes. Ground-based photometry in several filters presents an alternative possibility for studying the of faint such SEDs UV flux be determined and type can be inferred. Previous investigations GRB hosts have SEDs similar to starburst galaxies @goro02; @goro03; and their SFRs inferred from optical are moderate $<10$ M [@fruchter99; @bloom98b; @djor98; @djor01b]. SFRs have been reported based on spectroscopic measurements. The GRB 000418 host has an un-obscured 55 M derived from \[\] line flux All optical for the SFRs affected by dust extinction in the hosts. Therefore, the optically inferred represent lower limits to the true SFRs. Radio and sub-mm much affected by dust and observations of GRB indicate the unextincted SFRs can much two larger those from optical estimators [@berger01; @tanvir04]. However, not all hosts very large SFRs; some by radio observations [@vrees01b]. The SFRs of individual hosts published in the literature have been to be comparable to those of other high redshift galaxies | z<4.5$ [@prochaska04; @andersen00]. The repoRted magnitUdes oF thE hoStS are $20.4~<~r~<~30$ [@malEsani04; @jaunsen03].
ThE Life Time of the massive stars bElievEd TO proDUcE long-DuratioN gRbS Is oF tHe OrdEr OF a Few MyR. If The host Galaxies arE inDeEd forming sucH MaSsive stars ThiS should be refLecTed in tHeIr sPEctraL enErgy dIstribUTions (SeDs) which, iN tHE absenCE of reddENInG, shoUld reveal a blue conTInUUm. Moreover, theiR star-fOrMAtION raTes (sFRs) should Be Large. tHe integRAtED sfR oF A galaxy can be fOund from meaSUreMents oF oR \[\] liNE fluxeS, or frOm MEasUring the fluX in tHe UV contiNuum at 1500–2800 [å]{} In the reST frame oF the gaLaxY [@keNnicUTt98]. thE faInTNesS Of gRB HOstS presentS a PrOblem For sPECTRoscOpy As thEy reqUire long integRatIon tIMes On the LargeSt teLeScopeS. GrounD-baseD pHotometry in seveRal fIlters preSenTs An aLtErnatIVe possIbiLitY for stuDying thE sEDS oF FAInT hosts. From such SEDs ThE uv cOntinuum Flux caN Be DeTErmined aNd The GalaXY Type cAn be INfErred. PreVious iNVeStIgationS hAve shoWn ThaT GRb hostS Have sEDs siMilar to sTarbuRSt galaxies [@sok01; @gORo02; @goro03; @lise03], and THeIR sFrS infErrEd from opticAl meTHods Are mODeRatE $<10$ m [@frucHter99; @bLoOM98b; @DJor98; @djor01b]. Larger SFRs hAvE been rEportEd based on specTroscopic mEASUrements. the Grb 000418 hOSt has an un-obscuRed SFr of 55 M deriveD From the \[\] lIne flUx [@bloom03]. ALl the optiCAL methods For DetErmIniNG ThE SFRs are affecTED by dUsT extincTioN in the hOstS. ThEreForE, tHe opticalLy inferrEd sFrs RePreSent lOWer limitS tO thE tRue sFRs. RADio and Sub-mm Data ArE mUCh lEss affeCTeD BY dusT eXtInctIon, AnD obseRvatIOns Of GRB hoSts indicaTe tHAt thE uNeXtincteD SFRs can be as mUcH as two ordeRs Of mAgnituDE Larger thAn those derived from opticAL estimaTorS [@bergEr01; @beRger02; @tanviR04]. HoWever, nOt aLL hosts Have veRy larGe sFRS; SOme haVE sFrs $<200$ M SuGgested by rADIo oBservAtIons [@Vrees01b].
THe SFRs of individual grB hOsts published In tHe liTERaTurE HaVE beEn ARguED To be comparable tO those of otHeR HiGh redshift GAlaXiEs | z<4.5$ [@prochaska04; @and ersen00].The r epo rte dmagn itud es of the host s are $20.4~<~R~<~30$ [@mal esani 04 ; @ja u ns en03] .
Thel if e tim eof th em as sivesta rs beli eved to pr odu ce long-durati o nGRBs is of th e order of a fe w Myr. I f t h e hos t g alaxi es are indeed formingsu c h mass i ve star s th is s hould be reflecte d i n their spectra l ener gy di s t rib uti ons (SEDs) w hich, in thea bs e n c e o f reddening, s hould revea l ablue c on tin u um. Mo reove r, the ir star-for mati on rates(SFRs) shouldb e large . Theint egr ated SF Rofag ala x ycan befound fr om m easur emen t s o f or \[ \] l ine f luxes, or fro m m easu r ing thefluxin t he UV c ontinu um at 1 500–2800 [Å]{}in t he rest f ram eofth e gal a xy [@k enn icu tt98].The fai n tne ss o f G RB hosts presentsap r ob lem forspectr o sc op y as they r equ irel o ng in tegr a ti on times on th e l ar gest te le scopes .Gro und -base d pho tometr y in sev eralf ilters present s an alternati v ep o ss i bili tyfor studyin g th e SED s of fa int hosts . Fro ms uc h SEDs the UV contin uu m flux canbe determined and the g a l a xy typecanb ei nferred. Previ ous i nvestigati o ns haveshown that GR B hosts h a v e SEDs s imi lar to st a r bu rst galaxies[ @ sok0 1; @goro0 2;@goro03 ; @ lis e03 ],an d their S FRs infe rr ed f ro m o ptica l methods a remo der ate $ < 10$ M [@fr ucht er 99 ; @b loom98b ; @ d j or98 ;@d jor0 1b] .Large r SF R s h ave bee n reporte d b a sedon s pectros copic measure me nts. The G RB 00 0418 h o s t has an un-obscured SFR of 55M derive d f rom t he \ [\] lineflu x [@bl oom 0 3]. Al l theoptic al me t h ods f o r d ete rm ining theS F Rsare a ff ecte d by du st extinction in t h e h osts. Therefo re, the o pt ica l ly inf er r edS F Rs represent lo wer limits t o t he true SF R s.Ra dio and sub-mm data are muc h less af fected by d uste x tin ction, and observa tions ofG RB ho s ts indi cat e that t heunext incted SFR s can be as m uch as twoor ders ofmagnitude larger than t hose d erive d f rom optic ale sti mators [@ berg er01; @ber ger 02; @tan vir 0 4]. H owev e r, no t allhost s have ver y l arg e SF Rs; some ha v e SFR s $<2 00$ M sugg este d by radio observ a tions [@vrees0 1b]. The SF R s of i ndividual GRBhos ts p ublished i n the liter ature ha ve beenargued to be compar a b le to tho se o f o ther high re ds h ift gal ax ie s | z<4.5$ [@prochaska04;_@andersen00]. The_reported magnitudes of the_hosts are_$20.4~<~R~<~30$_[@malesani04; @jaunsen03].
The_life_time of the_massive stars believed_to produce long-duration GRBs_is of the_order_of a few Myr. If the host galaxies are indeed forming such massive stars_this_should be_reflected_in_their spectral energy distributions (SEDs)_which, in the absence of_reddening, should_reveal a blue continuum. Moreover, their star-formation rates_(SFRs)_should be large._The integrated SFR of a galaxy can be found_from measurements of or \[\] line_fluxes, or from_measuring_the_flux in the UV_continuum at 1500–2800 [Å]{} in the rest_frame of the galaxy [@kennicutt98]. The_faintness of GRB hosts presents a problem_for spectroscopy as they require long_integration times on the largest_telescopes. Ground-based_photometry in several filters presents_an alternative possibility_for studying_the SEDs of_faint hosts. From such SEDs the_UV continuum flux_can be determined and the galaxy_type_can be inferred._Previous_investigations_have shown_that GRB hosts_have_SEDs similar_to_starburst galaxies [@sok01; @goro02; @goro03; @lise03],_and_their SFRs inferred from optical methods are_moderate $<10$ M [@fruchter99; @bloom98b;_@djor98;_@djor01b]. Larger SFRs have_been reported based on spectroscopic_measurements. The GRB 000418 host has_an un-obscured_SFR of_55 M derived from the \[\] line flux [@bloom03]. All the optical_methods for determining the SFRs are_affected by dust extinction_in the_hosts._Therefore, the optically_inferred_SFRs represent_lower limits to the true SFRs. Radio_and sub-mm_data are much less affected by_dust extinction, and observations_of_GRB hosts indicate that the unextincted_SFRs can be as much as_two orders of magnitude larger_than_those_derived from optical estimators [@berger01;_@berger02; @tanvir04]. However, not all hosts_have very large_SFRs; some have SFRs $<200$ M suggested by radio_observations_[@vrees01b].
The SFRs of individual GRB hosts_published_in the literature have been argued_to_be_comparable to those of other_high redshift galaxies |
an algebra has, the richer the corresponding constraint language, we assume that the algebras we are dealing with have only two basic operations, just enough to guarantee the required properties. Therefore we assume that our semilattice block Mal’tsev algebras have only two basic operations: a binary operation $\cdot$ that we will often omit, and a ternary operation $\mal$ satisfying the conditions specified earlier. For elements $a,b\in\zA$ such that $ab=ba=b$ we write $a\le b$.
\[lem:dot-inequality\] Let $\zA$ be an SBM algebra. By choosing a reduct of $\zA$ we may assume that\
Operation $\cdot$ satisfies the equation $x(xy)=xy$; and for any $a,b\in\zA$, $a\le ab$.\
Operation $\mal$ can be chosen such that for any $a,b,c\in\zA$, $\mal(a,b,c)^{\sg_\zA}=(abc)^{\sg_\zA}$.
\(1) Follows from Proposition 10 of [@Bulatov16:connectivity].
\(2) Consider the operation $\mal'(x,y,z)=\mal(x,y,z)xyz$. If $B$ is a $\sg_\zA$-block, then, since $ab=a$ for any $a,b\in B$, operation $\mal'$ is Mal’tsev on $B$. Also, as $\zA\fac{\sg_\zA}$ is term equivalent to a semilattice, $d=\mal(a,b,c)^{\sg_\zA}$ belongs to the subsemilattice of $\zA\fac{\sg_\zA}$ generated by $a^{\sg_\zA},b^{\sg_\zA},c^{\sg_\zA}$. Therefore $\mal'(a,b,c)^{\sg_\zA}=d(abc)^{\sg_\zA}=(abc)^{\sg_\zA}$, and we can choose $\mal'$ for $\mal$.
Next we show some useful properties of SBM algebras. Let $\zA$ be an SBM algebra and $\max(\zA)$ the maximal block of $\sg$, that is, $\max(\zA)\cdot a\sse\max(\zA)$ for all $a\in\zA$.
\[lem:irreducible-maximal\] (1) The equivalence relation $\th_\zA$ whose blocks are $\max(\ | an algebra has, the richer the corresponding constraint linguistic process, we wear that the algebras we are dealing with have entirely two basic operations, just adequate to guarantee the required properties. consequently we wear that our semilattice block Mal’tsev algebra have only two basic operations: a binary operation $ \cdot$ that we will frequently omit, and a ternary operation $ \mal$ satisfying the conditions specified sooner. For elements $ a, b\in\zA$ such that $ ab = ba = b$ we write $ a\le b$.
\[lem: acid - inequality\ ] Let $ \zA$ be an SBM algebra. By choosing a reduct of $ \zA$ we may assume that\
Operation $ \cdot$ satisfy the equation $ x(xy)=xy$; and for any $ a, b\in\zA$, $ a\le ab$.\
Operation $ \mal$ can be chosen such that for any $ a, b, c\in\zA$, $ \mal(a, b, c)^{\sg_\zA}=(abc)^{\sg_\zA}$.
\(1) Follows from Proposition 10 of [ @Bulatov16: connectivity ].
\(2) Consider the operation $ \mal'(x, y, z)=\mal(x, y, z)xyz$. If $ B$ is a $ \sg_\zA$-block, then, since $ bachelor of arts = a$ for any $ a, b\in B$, operation $ \mal'$ is Mal’tsev on $ B$. Also, as $ \zA\fac{\sg_\zA}$ is term equivalent to a semilattice, $ d=\mal(a, b, c)^{\sg_\zA}$ belongs to the subsemilattice of $ \zA\fac{\sg_\zA}$ generated by $ a^{\sg_\zA},b^{\sg_\zA},c^{\sg_\zA}$. Therefore $ \mal'(a, b, c)^{\sg_\zA}=d(abc)^{\sg_\zA}=(abc)^{\sg_\zA}$, and we can choose $ \mal'$ for $ \mal$.
Next we show some useful properties of SBM algebras. Let $ \zA$ be an SBM algebra and $ \max(\zA)$ the maximal block of $ \sg$, that is, $ \max(\zA)\cdot a\sse\max(\zA)$ for all $ a\in\zA$.
\[lem: irreducible - maximal\ ] (1) The equivalence relation $ \th_\zA$ whose blocks are $ \max(\ | an algebra has, the richer uhe corresponding constreint lahguage, wd assume that the algebras wx arw deaoing with have only twu basic oierations, jusu enough to guaraifee the requidcd prm'erties. Therefote we assume that our seminagtnce block Mal’tsev algebras have only two baxif operations: a binswy olvrction $\cdot$ that we will often ojit, and a ternary oprration $\mal$ satisfying the confitions specified farlier. For eleiwnts $a,b\in\zA$ ruch that $ab=ba=b$ we wrije $a\le b$.
\[lem:dot-inequality\] Let $\zA$ ce an SBM algebtc. By whoosing a ceduct of $\zA$ we man assuma that\
Oleration $\cdot$ xatmsfiws the equation $x(xy)=xy$; and for any $a,b\in\zW$, $a\le ab$.\
O[exation $\mal$ can be choweb sucv thdt fue avy $z,b,r\in\aA$, $\mal(w,b,c)^{\ag_\zA}=(abc)^{\sg_\zZ}$.
\(1) Follows feom Proposition 10 of [@Nujqtov16:connectivjty].
\(2) Cogsyder the operation $\mal'(x,y,z)=\mal(x,y,z)xyz$. If $B$ is a $\sg_\zA$-block, then, since $ab=a$ for any $a,b\in B$, okeration $\mwl'$ is Mal’tsev on $B$. Also, as $\zA\fac{\sg_\zA}$ is term equiealenv go c semiuqthice, $d=\mal(a,b,c)^{\sg_\zA}$ belongs to the subsemilatticq og $\dA\fac{\sg_\zA}$ generatcd by $a^{\sg_\zA},b^{\sg_\zA},c^{\sb_\zW}$. Yrerefore $\mal'(a,c,c)^{\sg_\zA}=b(zbd)^{\sg_\zA}=(abc)^{\sg_\zA}$, and wf can croose $\mal'$ for $\ial$.
Nrxt we show some useful properties of SYM qlgebras. Let $\zA$ be an SBM algzbra amd $\mac(\zA)$ the maximal block oy $\sg$, tgat is, $\max(\zW)\cdot a\sss\oax(\zA)$ for all $a\iv\zA$.
\[kek:irreducible-maximal\] (1) The ezuivalencx relction $\th_\xA$ wnose bjocks are $\lax(\ | an algebra has, the richer the corresponding we that the we are dealing operations, enough to guarantee required properties. Therefore assume that our semilattice block Mal’tsev have only two basic operations: a binary operation $\cdot$ that we will often and a ternary operation $\mal$ satisfying the conditions specified earlier. For elements $a,b\in\zA$ that we $a\le \[lem:dot-inequality\] Let $\zA$ be an SBM algebra. By choosing a reduct of $\zA$ we may assume Operation $\cdot$ satisfies the equation $x(xy)=xy$; and for $a,b\in\zA$, $a\le ab$.\ Operation can be chosen such that any $\mal(a,b,c)^{\sg_\zA}=(abc)^{\sg_\zA}$. \(1) from 10 [@Bulatov16:connectivity]. \(2) Consider operation $\mal'(x,y,z)=\mal(x,y,z)xyz$. If $B$ is a $\sg_\zA$-block, then, since $ab=a$ for any $a,b\in B$, operation $\mal'$ is on $B$. $\zA\fac{\sg_\zA}$ is equivalent a $d=\mal(a,b,c)^{\sg_\zA}$ belongs to of $\zA\fac{\sg_\zA}$ generated by $a^{\sg_\zA},b^{\sg_\zA},c^{\sg_\zA}$. Therefore can choose $\mal'$ for $\mal$. Next we show useful properties SBM algebras. Let $\zA$ be an algebra and $\max(\zA)$ the maximal block of $\sg$, is, $\max(\zA)\cdot a\sse\max(\zA)$ for all $a\in\zA$. \[lem:irreducible-maximal\] (1) The equivalence relation $\th_\zA$ whose blocks are | an algebra has, the richer the cOrrespondiNg conStrAinT lAnguAge, wE assume that the ALgebRas we are dealing with havE only TwO BasiC OpEratiOns, just ENoUGH to GuArAntEe THe RequiRed PropertIes. TherefoRe wE aSsume that our SEmIlattice blOck mal’tsev algebRas Have onLy Two BAsic oPerAtionS: a binaRY operaTion $\cdot$ tHaT We will OFten omiT, ANd A terNary operation $\mal$ sATiSFying the conditIons spEcIFiED EarLieR. For elemenTs $A,b\in\za$ Such thaT $Ab=BA=B$ We wRIte $a\le b$.
\[lem:dot-Inequality\] LET $\zA$ Be an SBm aLgeBRa. By chOosinG a REduCt of $\zA$ we may AssuMe that\
OpeRation $\CDot$ satiSFies the EquatiOn $x(Xy)=xY$; and FOr AnY $a,b\In\Za$, $a\lE Ab$.\
opeRAtiOn $\mal$ can Be ChOsen sUch tHAT FOr anY $a,b,C\in\za$, $\mal(a,B,c)^{\sg_\zA}=(abc)^{\sg_\zA}$.
\(1) folLows FRom propoSitioN 10 of [@BUlAtov16:cOnnectIvity].
\(2) coNsider the operatIon $\mAl'(x,y,z)=\mal(x,Y,z)xYz$. if $B$ Is A $\sg_\zA$-BLock, thEn, sIncE $ab=a$ for Any $a,b\in b$, OpeRaTION $\mAl'$ is Mal’tsev on $B$. Also, As $\Za\FaC{\sg_\zA}$ is tErm equIVaLeNT to a semiLaTtiCe, $d=\mAL(A,b,c)^{\sg_\ZA}$ beLOnGs to the sUbsemiLAtTiCe of $\zA\fAc{\Sg_\zA}$ geNeRatEd bY $a^{\sg_\za},B^{\sg_\za},c^{\sg_\zA}$. thereforE $\mal'(a,B,C)^{\sg_\zA}=d(abc)^{\sg_\zA}=(aBC)^{\sg_\zA}$, and we can CHoOSE $\mAL'$ for $\Mal$.
next we show sOme uSEful PropERtIes OF SBM aLgebrAs. lEt $\Za$ be an SBM algebra and $\mAx(\ZA)$ the mAximaL block of $\sg$, thaT is, $\max(\zA)\cdOT A\Sse\max(\zA)$ For aLL $a\IN\zA$.
\[lem:irreduciBle-maXimal\] (1) The eqUIvalence RelatIon $\th_\zA$ wHose blockS ARe $\max(\ | an algebra has, the riche r the corr espon din g c on stra intlanguage, we a s sume that the algebras weare d ea l ingw it h hav e onlyt wo b asi cop era ti o ns , jus t e nough t o guarante e t he required pr o pe rties. The ref ore we assum e t hat ou rsem i latti ceblock Mal’t s ev alg ebras hav eo nly tw o basico p er atio ns: a binary oper a ti o n $\cdot$ that we wi ll of t e n o mit , and a te rn ary o p eration $\ m a l $ s a tisfying theconditionss pec ifiedea rli e r. For elem en t s $ a,b\in\zA$such that $ab =ba=b$ we writ e $a\leb$.
\ [le m:d ot-i n eq ua lit y\ ] Le t $ \zA $ be an SBMal ge bra.By c h o o s inga r educ t of$\zA$ we mayass umet hat \
Ope ratio n $\ cd ot$ s atisfi es th eequation $x(xy) =xy$ ; and for an y$a, b\ in\zA $ , $a\l e a b$. \
Opera tion $\ m al$ c a n be chosen such thatfo r an y $a,b,c \in\zA $ ,$\ m al(a,b,c )^ {\s g_\z A } =(abc )^{\ s g_ \zA}$.
\(1) F o ll ow s fromPr oposit io n 1 0 o f [@B u lato v16:co nnectivi ty].\(2) Considert he operation$ \m a l '( x ,y,z )=\ mal(x,y,z)x yz$. If $ B$ i s a $\ s g_\zA $-blo ck , t h en, since $ab=a$ fo rany $a ,b\in B$, operatio n $\mal'$i s Mal’tsev on$ B$ . Also, as $\zA \fac{ \sg_\zA}$i s term e quiva lent toa semilat t i ce, $d=\ mal (a, b,c )^{ \ s g_ \zA}$ belongs t o th esubsemi lat tice of $\ zA\ fac {\s g_ \zA}$ gen erated b y$a ^{ \s g_\ zA},b ^ {\sg_\zA }, c^{ \s g_\ zA}$. Theref ore $ \mal '( a, b ,c) ^{\sg_\ z A} = d (abc )^ {\ sg_\ zA} =( abc)^ {\sg _ \zA }$, and we can c hoo s e $\ ma l' $ for $ \mal$.
Nextwe show some u sef ul pro p e rties of SBM algebras. Let $\zA $ be anSBM alge braand $\max (\z A)$ th e m a ximalblockof $\ sg $,t h at is , $\ max (\ zA)\cdot a \ s se\ max(\ zA )$ f or all$a\in\zA$.
\[lem: i rre ducible-maxim al\ ] (1 ) Th e e q ui v ale nc e re l a tion $\th_\zA$whose bloc ks ar e $\max(\ | an_algebra has,_the richer the corresponding_constraint language,_we_assume that_the_algebras we are_dealing with have_only two basic operations,_just enough to_guarantee_the required properties. Therefore we assume that our semilattice block Mal’tsev algebras have only_two_basic operations:_a_binary_operation $\cdot$ that we will_often omit, and a ternary_operation $\mal$_satisfying the conditions specified earlier. For elements $a,b\in\zA$_such_that $ab=ba=b$ we_write $a\le b$.
\[lem:dot-inequality\] Let $\zA$ be an SBM algebra._By choosing a reduct of $\zA$_we may assume_that\
Operation_$\cdot$_satisfies the equation $x(xy)=xy$;_and for any $a,b\in\zA$, $a\le ab$.\
Operation_$\mal$ can be chosen such that_for any $a,b,c\in\zA$, $\mal(a,b,c)^{\sg_\zA}=(abc)^{\sg_\zA}$.
\(1) Follows from Proposition 10_of [@Bulatov16:connectivity].
\(2) Consider the operation $\mal'(x,y,z)=\mal(x,y,z)xyz$._If $B$ is a $\sg_\zA$-block,_then, since_$ab=a$ for any $a,b\in B$,_operation $\mal'$ is_Mal’tsev on_$B$. Also, as_$\zA\fac{\sg_\zA}$ is term equivalent to a_semilattice, $d=\mal(a,b,c)^{\sg_\zA}$ belongs_to the subsemilattice of $\zA\fac{\sg_\zA}$ generated_by_$a^{\sg_\zA},b^{\sg_\zA},c^{\sg_\zA}$. Therefore $\mal'(a,b,c)^{\sg_\zA}=d(abc)^{\sg_\zA}=(abc)^{\sg_\zA}$,_and_we_can choose_$\mal'$ for $\mal$.
Next_we_show some_useful_properties of SBM algebras. Let $\zA$_be_an SBM algebra and $\max(\zA)$ the maximal_block of $\sg$, that_is,_$\max(\zA)\cdot a\sse\max(\zA)$ for all_$a\in\zA$.
\[lem:irreducible-maximal\] (1) The equivalence relation_$\th_\zA$ whose blocks are $\max(\ |
to applications* (Cambridge University Press, Cambridge, 1998). R.H. Bruck and H.J. Ryser, “The nonexistence of certain finite projective planes", [*Can. J. Math.*]{} **1**, 88 (1949). M. Saniga and M. Planat, “Sets of MUBs as arcs in finite projective planes?", [*Chaos, Solitos and Fractals*]{} [**26**]{}, 1267 (2005). M. Saniga and M. Planat, “Hjelmslev geometry of MUBs", [*J. Phys. A*]{} [**39**]{}, 435 (2006). J.W.P. Hirschfeld, *Projective geometries over finite fields* (Oxford University Press, Oxford, 1998). L.M. Batten, *Combinatorics of finite geometries* (Cambridge University Press, Cambridge, 1997). M. Planat, M. Saniga and M.R. Kibler, “Quantum entanglement and projective ring geometry", [*SIGMA*]{} **2**, paper 066 (2006). D.I. Fivel, “Remarkable phase oscillations appearing in the lattice dynamics of EPR states", [*Phys. Rev. Lett.*]{} **74**, 835 (1995). K. Fujii, “A relation between coherent states and generalized Bell states", preprint quant-ph/0105077 (2001). D.P. DiVincenzo, T. Mor, P.W. Shor, J.A. Smolin and B.M. Terhal, “Unextendible product bases, uncompletable product bases and bound entanglement", [*Commun. Math. Phys.*]{} **238**, 379 (2003). P. Horodecki, “Separability criterion and inseparable mixed states with positive partial transposition", [*Phys. Lett. A*]{} **232**, 333 (1997). S. Yu, Z.B. Chen, J.W. Pan and Y.D. Zhang, “Classifying N-qubit entanglement via Bell’s inequalities", [*Phys. Rev. Lett.*]{} [**90**]{}, 080401 (2003). S. Bose, V. Vedral and P.L. Knight, “Multiparticle generalization of entanglement swapping", [*Phys. Rev. A*]{} **57**, 822 (1998). B.A. Bernevig and H.D. Chen, “Geometry of the 3-qubit state, entanglement and division algebras", [*J. | to applications * (Cambridge University Press, Cambridge, 1998). R.H. Bruck and H.J. Ryser, “ The nonexistence of certain finite projective planes ", [ * Can. J. Math. * ] { } * * 1 * *, 88 (1949). M. Saniga and M. Planat, “ Sets of MUBs as arcs in finite projective plane? ", [ * Chaos, Solitos and Fractals * ] { } [ * * 26 * * ] { }, 1267 (2005). M. Saniga and M. Planat, “ Hjelmslev geometry of MUBs ", [ * J. Phys. A * ] { } [ * * 39 * * ] { }, 435 (2006). J.W.P. Hirschfeld, * Projective geometry over finite fields * (Oxford University Press, Oxford, 1998). L.M. Batten, * Combinatorics of finite geometries * (Cambridge University Press, Cambridge, 1997). M. Planat, M. Saniga and M.R. Kibler, “ Quantum web and projective ring geometry ", [ * SIGMA * ] { } * * 2 * *, newspaper 066 (2006). D.I. Fivel, “ Remarkable phase oscillations appear in the lattice dynamics of EPR state ", [ * Phys. Rev. Lett. * ] { } * * 74 * *, 835 (1995). K. Fujii, “ A relation between coherent states and generalize Bell states ", preprint quant - ph/0105077 (2001). D.P. DiVincenzo, T. Mor, P.W. Shor, J.A. Smolin and B.M. Terhal, “ Unextendible product base, uncompletable product bases and bound entanglement ", [ * Commun. Math. Phys. * ] { } * * 238 * *, 379 (2003). P. Horodecki, “ Separability criterion and inseparable mixed state with positive partial transposition ", [ * Phys. Lett. A * ] { } * * 232 * *, 333 (1997). S. Yu, Z.B. Chen, J.W. Pan and Y.D. Zhang, “ relegate N - qubit entanglement via Bell ’s inequalities ", [ * Phys. Rev. Lett. * ] { } [ * * 90 * * ] { }, 080401 (2003). S. Bose, V. Vedral and P.L. Knight, “ Multiparticle abstraction of entanglement swap ", [ * Phys. Rev. A * ] { } * * 57 * *, 822 (1998). B.A. Bernevig and H.D. Chen, “ Geometry of the 3 - qubit state, entanglement and division algebra ", [ * J. | to applications* (Cambridge Mniversity Press, Cambrivge, 1998). R.H. Bruck avd H.J. Ryser, “The nonexistence od cerucin finite projectivd planes", [*Ban. J. Maty.*]{} **1**, 88 (1949). N. Saniga ais M. Plakct, “Sefd of NUBs as arcs ik finite prmjective planev?", [*Zhcos, Solitos and Fractals*]{} [**26**]{}, 1267 (2005). M. Saniga and M. Llwnat, “Hjelmslev geokqtry of MUBs", [*J. Phys. A*]{} [**39**]{}, 435 (2006). J.W.P. Hirschfels, *Projebtive geometries pver finite fields* (Oxford Knivfrsity Press, Oxforf, 1998). L.M. Batteb, *Coivinatorics ow finite gtoketries* (Cajbridge University Press, Cambriage, 1997). K. Planat, M. Sqnihd and M.R. Kiuler, “Qlantum entanglement atd projrctive ring gepmevry", [*WIGMA*]{} **2**, paper 066 (2006). D.I. Fivxl, “Remarkable phase jscillatimna appearing in thw oattiwe dfnamkxs uf TPR sfates", [*Ohya. Rev. Lett.*]{} **74**, 835 (1995). K. Fujii, “A relation between cjyerent states and gqnqralized Bell states", preprint quant-ph/0105077 (2001). G.P. SiVincenzo, T. Mor, P.W. Shoe, J.A. Smolin and B.M. Tethal, “Unextqndible product bases, uncompletable product bases and uojnd ektaneoelent", [*Commun. Math. Phys.*]{} **238**, 379 (2003). P. Horodecki, “Separabijjtu briterion and inscparable mixed stayed eyth positive kartial trznsposition", [*Phys. Lftt. A*]{} **232**, 333 (1997). S. Yy, Z.B. Chen, J.W. Lan and Y.D. Zhang, “Classifyint N-qubit entcngoement via Bell’s iuequalities", [*'hys. Rgv. Lety.*]{} [**90**]{}, 080401 (2003). S. Bose, V. Vedral anb P.L. Khight, “Multioarticle fdneralization of ennangnement swapping", [*Phys. Rev. A*]{} **57**, 822 (1998). B.A. Becneviy and H.D. Chem, “Geomqtry of thf 3-qubld state, entanglemejt anb divhsion algehras", [*J. | to applications* (Cambridge University Press, Cambridge, 1998). and Ryser, “The of certain finite **1**, (1949). M. Saniga M. Planat, “Sets MUBs as arcs in finite projective [*Chaos, Solitos and Fractals*]{} [**26**]{}, 1267 (2005). M. Saniga and M. Planat, “Hjelmslev of MUBs", [*J. Phys. A*]{} [**39**]{}, 435 (2006). J.W.P. Hirschfeld, *Projective geometries over fields* University Oxford, L.M. Batten, *Combinatorics of finite geometries* (Cambridge University Press, Cambridge, 1997). M. Planat, M. Saniga and Kibler, “Quantum entanglement and projective ring geometry", [*SIGMA*]{} paper 066 (2006). D.I. “Remarkable phase oscillations appearing in lattice of EPR [*Phys. Lett.*]{} 835 (1995). K. “A relation between coherent states and generalized Bell states", preprint quant-ph/0105077 (2001). D.P. DiVincenzo, T. Mor, P.W. J.A. Smolin Terhal, “Unextendible bases, product and bound entanglement", Phys.*]{} **238**, 379 (2003). P. Horodecki, inseparable mixed states with positive partial transposition", [*Phys. A*]{} **232**, (1997). S. Yu, Z.B. Chen, J.W. and Y.D. Zhang, “Classifying N-qubit entanglement via Bell’s [*Phys. Rev. Lett.*]{} [**90**]{}, 080401 (2003). S. Bose, V. Vedral and P.L. Knight, “Multiparticle generalization swapping", [*Phys. Rev. A*]{} 822 (1998). B.A. and Chen, of 3-qubit state, and division algebras", [*J. | to applications* (Cambridge UnIversity PrEss, CaMbrIdgE, 1998). R.h. BruCk anD H.J. Ryser, “The nonEXistEnce of certain finite proJectiVe PLaneS", [*caN. J. MatH.*]{} **1**, 88 (1949). M. SanigA AnD m. plaNaT, “SEts Of muBS as arCs iN finite Projective PlaNeS?", [*Chaos, SolitoS AnD Fractals*]{} [**26**]{}, 1267 (2005). M. sanIga and M. PlanaT, “HjElmsleV gEomETry of mUBS", [*J. PhyS. A*]{} [**39**]{}, 435 (2006). J.W.P. HIRschfeLd, *ProjectIvE GeometRIes over FINiTe fiElds* (Oxford UniversITy pRess, Oxford, 1998). L.M. BaTten, *CoMbINaTORicS of Finite geomEtRies* (CAMbridge uNiVERSitY press, CambridgE, 1997). M. Planat, M. SaNIga And M.R. KIbLer, “qUantum EntanGlEMenT and projectIve rIng geometRy", [*SIGMa*]{} **2**, Paper 066 (2006). D.I. fIvel, “RemArkablE phAse OsciLLaTiOns ApPEarINg In tHE laTtice dynAmIcS of EPr staTES", [*pHys. REv. LEtt.*]{} **74**, 835 (1995). K. fujii, “a relation betwEen CoheREnt StateS and gEnerAlIzed BEll staTes", prEpRint quant-ph/0105077 (2001). D.P. DivincEnzo, T. Mor, P.w. ShOr, j.A. SMoLin anD b.M. TerhAl, “UNexTendiblE producT BasEs, UNCOmPletable product basEs AND bOund entaNglemeNT", [*COmMUn. Math. PhYs.*]{} **238**, 379 (2003). p. HoRodeCKI, “SepaRabiLItY criteriOn and iNSePaRable miXeD stateS wIth PosItive PArtiAl tranSpositioN", [*Phys. lEtt. A*]{} **232**, 333 (1997). S. Yu, Z.B. Chen, J.w. pan and Y.D. Zhang, “cLaSSIfYIng N-QubIt entanglemEnt vIA BelL’s inEQuAliTIes", [*PhYs. Rev. leTT.*]{} [**90**]{}, 080401 (2003). S. bOse, V. Vedral and P.L. KnigHt, “multipArticLe generalizatIon of entanGLEMent swapPing", [*pHyS. rev. A*]{} **57**, 822 (1998). B.A. Bernevig And H.D. chen, “GeometRY of the 3-quBit stAte, entanGlement anD DIvision aLgeBraS", [*J. | to applications* (Cambrid ge Univers ity P res s,Ca mbri dge, 1998). R.H. B r uckand H.J. Ryser, “The n onexi st e nceo fcerta in fini t ep r oje ct iv e p la n es ", [* Can . J. Ma th.*]{} ** 1** ,88 (1949). M . S aniga andM.Planat, “Set s o f MUBs a s a r cs in fi niteprojec t ive pl anes?", [ *C h aos, S o litos a n d F ract als*]{} [**26**]{ } ,1 267 (2005). M. Sanig aa nd M . P lan at, “Hjelm sl ev ge o metry o f M U B s ",[ *J. Phys. A*] {} [**39**] { },435 (2 00 6). J.W.P. Hirs ch f eld , *Projecti ve g eometries overf inite f i elds* ( Oxford Un ive rsit y P re ss, O x for d ,199 8 ).L.M. Bat te n, *Com bina t o r i cs o f f init e geo metries* (Cam bri dgeU niv ersit y Pre ss,Ca mbrid ge, 19 97).M. Planat, M. San igaand M.R.Kib le r,“Q uantu m entan gle men t and p rojecti v e r in g g eo metry", [*SIGMA*]{ }* * 2* *, paper 066 ( 2 00 6) . D.I. Fi ve l,“Rem a r kable pha s eoscillat ions a p pe ar ing inth e latt ic e d yna micso f EP R stat es", [*P hys.R ev. Lett.*]{}* *74**, 835 (1 9 95 ) . K . Fuj ii, “A relatio n be t ween coh e re nts tates andge n er a lized Bell states", p reprin t qua nt-ph/0105077 (2001). D . P . DiVince nzo, T. Mor, P.W. Shor , J.A . Smolin a n d B.M. T erhal , “Unext endible p r o duct bas es, un com ple t a bl e product bas e s and b ound en tan glement ",[*C omm un. M ath. Phys .*]{} ** 23 8* *, 3 79(2003 ) . P. Hor od eck i, “S epara b ilitycrite rion a nd ins eparabl e m i x ed s ta te s wi thpo sitiv e pa r tia l trans position" , [ * Phys .Le tt. A*] {} **232**, 3 33 (1997). S .Yu, Z.B.C h en, J.W. Pan and Y.D. Zhang, “C l assifyi ngN-qub it e ntangleme ntvia Be ll’ s inequ alitie s", [ *P hys . Rev.L e tt .*] {} [**90**]{ } , 08 0401(2 003) . S. Bo se, V. Vedral andP .L. Knight, “Mul tip arti c l egen e ra l iza ti o n o f entanglement sw apping", [ *P h ys . Rev. A*] { } * *5 7**, 82 2 (1998 ). B. A . Berne vig and H .D. Chen, “ Geom e t ryof the 3-q ubit sta te, entan g lemen t a nd di vis ion al ge bra s", [ *J. | to_applications* (Cambridge_University Press, Cambridge, 1998)._R.H. Bruck_and_H.J. Ryser,_“The_nonexistence of certain_finite projective planes",_[*Can. J. Math.*]{} **1**,_88 (1949). M._Saniga_and M. Planat, “Sets of MUBs as arcs in finite projective planes?", [*Chaos, Solitos_and_Fractals*]{} [**26**]{},_1267_(2005)._M. Saniga and M. Planat,_“Hjelmslev geometry of MUBs", [*J._Phys. A*]{}_[**39**]{}, 435 (2006). J.W.P. Hirschfeld, *Projective geometries over_finite_fields* (Oxford University_Press, Oxford, 1998). L.M. Batten, *Combinatorics of finite geometries*_(Cambridge University Press, Cambridge, 1997). M._Planat, M. Saniga_and_M.R._Kibler, “Quantum entanglement and_projective ring geometry", [*SIGMA*]{} **2**, paper_066 (2006). D.I. Fivel, “Remarkable phase_oscillations appearing in the lattice dynamics of_EPR states", [*Phys. Rev. Lett.*]{} **74**,_835 (1995). K. Fujii, “A_relation between_coherent states and generalized Bell_states", preprint quant-ph/0105077_(2001). D.P._DiVincenzo, T. Mor,_P.W. Shor, J.A. Smolin and B.M._Terhal, “Unextendible product_bases, uncompletable product bases and bound_entanglement",_[*Commun. Math. Phys.*]{}_**238**,_379_(2003). P._Horodecki, “Separability criterion_and_inseparable mixed_states_with positive partial transposition", [*Phys. Lett._A*]{}_**232**, 333 (1997). S. Yu, Z.B. Chen,_J.W. Pan and Y.D._Zhang,_“Classifying N-qubit entanglement via_Bell’s inequalities", [*Phys. Rev. Lett.*]{}_[**90**]{}, 080401 (2003). S. Bose, V._Vedral and_P.L. Knight,_“Multiparticle generalization of entanglement swapping", [*Phys. Rev. A*]{} **57**, 822 (1998)._B.A. Bernevig and H.D. Chen, “Geometry_of the 3-qubit state,_entanglement and_division_algebras", [*J. |
Bulbul:2014sua; @Boyarsky:2014jta; @Abazajian:2014gza; @Merle:2014xpa; @Haba:2014taa; @Rodejohann:2014eka; @Abada:2014zra; @Ishida:2014fra; @Robinson:2014bma; @Chakraborty:2014tma; @Adulpravitchai:2014xna; @Patra:2014pga; @Frigerio:2014ifa; @Tsuyuki:2014aia; @Kang:2014mea; @Harada:2014lma; @Ishida:2014dlp; @Bezrukov:2014nza]. A keV scale sterile neutrino can be produced through the Dodelson-Widrow (DW) mechanism [@Dodelson:1993je] with an approximate relic abundance h\^2\~0.07 ()()\^2. This is a consequence of non-resonant oscillation due to the mixing between the active and sterile sectors. However, sterile neutrinos produced through the DW mechanism accounting for all of dark matter has now been robustly ruled out for all masses based on constraints from X-ray bounds [@Boyarsky:2006fg; @Boyarsky:2006ag; @Watson:2006qb; @Boyarsky:2005us; @Boyarsky:2007ay; @Boyarsky:2007ge; @Watson:2011dw; @Horiuchi:2013noa] and Lyman-alpha data [@Seljak:2006qw; @Asaka:2006nq; @Boyarsky:2008xj]; see [@Horiuchi:2013noa] for a summary.
This problem can be evaded in several ways. A lepton asymmetry in the early Universe can lead to resonant production of sterile neutrinos, resulting in a colder distribution that can evade Lyman-alpha bounds; this is known as the Shi-Fuller mechanism [@Shi:1998km]. Another approach is to consider sterile neutrino production from the decays of a singlet Higgs boson from an extended Higgs sector [@Kusenko:2006rh; @Petraki:2007gq], which can produce all of dark matter from the freeze-in mechanism [@Hall:2009bx] (for a discussion in the context of the 3.5 keV line, see [@Merle:2014xpa]).
| Bulbul:2014sua; @Boyarsky:2014jta; @Abazajian:2014gza; @Merle:2014xpa; @Haba:2014taa; @Rodejohann:2014eka; @Abada:2014zra; @Ishida:2014fra; @Robinson:2014bma; @Chakraborty:2014tma; @Adulpravitchai:2014xna; @Patra:2014pga; @Frigerio:2014ifa; @Tsuyuki:2014aia; @Kang:2014mea; @Harada:2014lma; @Ishida:2014dlp; @Bezrukov:2014nza ]. A keV scale sterile neutrino can be produced through the Dodelson - Widrow (DW) mechanism [ @Dodelson:1993je ] with an approximate relic abundance h\^2\~0.07 () () \^2. This is a consequence of non - evocative cycle due to the mix between the active and sterile sectors. However, sterile neutrino produced through the DW mechanism accounting for all of dark topic has immediately been robustly ruled out for all mass based on constraints from adam - ray bounds [ @Boyarsky:2006fg; @Boyarsky:2006ag; @Watson:2006qb; @Boyarsky:2005us; @Boyarsky:2007ay; @Boyarsky:2007ge; @Watson:2011dw; @Horiuchi:2013noa ] and Lyman - alpha datum [ @Seljak:2006qw; @Asaka:2006nq; @Boyarsky:2008xj ]; see [ @Horiuchi:2013noa ] for a summary.
This problem can be evaded in several ways. A lepton asymmetry in the early Universe can lead to resonant product of sterile neutrinos, resulting in a cold distribution that can evade Lyman - alpha bounds; this is known as the Shi - Fuller mechanism [ @Shi:1998 km ]. Another access is to consider sterile neutrino production from the decays of a singlet Higgs boson from an extended Higgs sector [ @Kusenko:2006rh; @Petraki:2007gq ], which can grow all of dark matter from the freeze - in mechanism [ @Hall:2009bx ] (for a discussion in the context of the 3.5 keV line, see [ @Merle:2014xpa ]). | Bulhul:2014sua; @Boyarsky:2014jta; @Abazamian:2014gza; @Merle:2014xpa; @Haba:2014tae; @Rodejkhann:2014eka; @Abada:2014zra; @Ishida:2014fra; @Robinson:2014uma; @Xhakrqborty:2014tma; @Adulpravitchxi:2014xna; @Patga:2014pga; @Friterii:2014ufa; @Tsuyukm:2014zia; @Kann:2014iea; @Gwradc:2014lna; @Ishida:2014dlp; @Bgzrukov:2014nza]. A keV scale stesiue neutrino can be produced through tre Dodeksln-Widrow (DW) meshanpsi [@Dosvlwon:1993je] with an approximate reljc abunvance h\^2\~0.07 ()()\^2. This ix a consequence of non-resojant oscillation due tl the mixint beeqeen the actkve and sttrnle sectors. However, sterile neutrinos prodjced chrough the DQ mfwhanism acciuntigg for all on dark katter nas now been rpbuvtlt ruled out for all messes based on consttaints frok R-ray bounds [@Boyarsky:2006ft; @Voyarvky:2006ac; @Wagwon:2006db; @Gobaraky:2005us; @Hoyersky:2007ay; @Boyzrsky:2007ge; @Watwon:2011dw; @Horiuchi:2013noa] amd Oyman-alpha dafa [@Seltah:2006qw; @Asaka:2006nq; @Boyarsky:2008xj]; see [@Horiuchi:2013noa] xor a summary.
This problem xan be evaded in sevetal ways. A lepton asymmetry in the early Universe can lead do reaunaub oeofuction of sterile neutrinos, resulting in a cjmdtr cistribution tmat can evade Lymam-appnw bounds; this is known as the Shi-Fuller lechanifm [@Shu:1998km]. Anothtr aplroach is to consider sterioe neutrino irodyction from the deeays of a siuglet Niggs boson from an extended Higfs sector [@Kksenko:2006rh; @Ldtraki:2007gq], which cxn irodgce all of dark matter froi the frexze-in mechanksm [@Nall:2009bx] (for a disfussion in the context ov the 3.5 neV line, sfe [@Merle:2014xpa]).
| Bulbul:2014sua; @Boyarsky:2014jta; @Abazajian:2014gza; @Merle:2014xpa; @Haba:2014taa; @Rodejohann:2014eka; @Abada:2014zra; @Chakraborty:2014tma; @Patra:2014pga; @Frigerio:2014ifa; @Kang:2014mea; @Harada:2014lma; @Ishida:2014dlp; neutrino be produced through Dodelson-Widrow (DW) mechanism with an approximate relic abundance h\^2\~0.07 This is a consequence of non-resonant oscillation due to the mixing between the and sterile sectors. However, sterile neutrinos produced through the DW mechanism accounting for of matter now robustly ruled out for all masses based on constraints from X-ray bounds [@Boyarsky:2006fg; @Boyarsky:2006ag; @Watson:2006qb; @Boyarsky:2005us; @Boyarsky:2007ge; @Watson:2011dw; @Horiuchi:2013noa] and Lyman-alpha data [@Seljak:2006qw; @Asaka:2006nq; see [@Horiuchi:2013noa] for a This problem can be evaded several A lepton in early can lead to production of sterile neutrinos, resulting in a colder distribution that can evade Lyman-alpha bounds; this is known the Shi-Fuller Another approach to sterile production from the a singlet Higgs boson from an [@Kusenko:2006rh; @Petraki:2007gq], which can produce all of dark from the mechanism [@Hall:2009bx] (for a discussion in context of the 3.5 keV line, see [@Merle:2014xpa]). | Bulbul:2014sua; @Boyarsky:2014jta; @AbazaJian:2014gza; @MerLe:2014xpa; @habA:2014taA; @ROdejOhanN:2014eka; @Abada:2014zra; @IsHIda:2014fRa; @Robinson:2014bma; @ChakraborTy:2014tma; @adULpraVItChai:2014xNa; @Patra:2014PGa; @fRIgeRiO:2014iFa; @TSuYUkI:2014aia; @KAng:2014Mea; @HaraDa:2014lma; @IshidA:2014dlP; @BEzrukov:2014nza]. A kEv sCale sterilE neUtrino can be pRodUced thRoUgh THe DodElsOn-WidRow (DW) mEChanisM [@Dodelson:1993Je] WIth an aPProximaTE ReLic aBundance h\^2\~0.07 ()()\^2. This is a cONsEQuence of non-resOnant oScILlATIon Due To the mixinG bEtweeN The actiVE aND STerILe sectors. HoweVer, sterile nEUtrInos prOdUceD ThrougH the Dw mEChaNism accountIng fOr all of daRk mattER has now BEen robuStly ruLed Out For aLL mAsSes BaSEd oN CoNstRAinTs from X-rAy BoUnds [@BOyarSKY:2006FG; @BoyArsKy:2006ag; @watsoN:2006qb; @Boyarsky:2005us; @boyArskY:2007Ay; @BOyarsKy:2007ge; @WAtsoN:2011dW; @HoriUchi:2013noA] and LYmAn-alpha data [@SeljAk:2006qw; @asaka:2006nq; @BoYarSkY:2008xj]; SeE [@HoriUChi:2013noa] For A suMmary.
ThIs problEM caN bE EVAdEd in several ways. A lePtON AsYmmetry iN the eaRLy unIVerse can LeAd tO resONAnt prOducTIoN of steriLe neutRInOs, ResultiNg In a colDeR diStrIbutiON thaT can evAde Lyman-Alpha BOunds; this is knoWN as the Shi-FullER mECHaNIsm [@SHi:1998kM]. Another appRoacH Is to ConsIDeR stERile nEutriNo PRoDUction from the decays Of A singlEt HigGs boson from an Extended HiGGS Sector [@KuSenkO:2006Rh; @pEtraki:2007gq], which cAn proDuce all of dARk matter From tHe freeze-In mechaniSM [@hall:2009bx] (foR a dIscUssIon IN ThE context of the 3.5 KEv linE, sEe [@Merle:2014Xpa]).
| Bulbul:2014sua; @Boyarsky: 2014jta; @ Abaza jia n:2 01 4gza ; @M erle:2014xpa;@ Haba :2014taa; @Rodejohann: 2014e ka ; @Ab a da :2014 zra; @I s hi d a :20 14 fr a;@R o bi nson: 201 4bma; @ Chakrabort y:2 01 4tma; @Adulp r av itchai:201 4xn a; @Patra:20 14p ga; @F ri ger i o:201 4if a; @T suyuki : 2014ai a; @Kang: 20 1 4mea;@ Harada: 2 0 14 lma; @Ishida:2014dlp; @B e zrukov:2014nza ]. A k eV sc a l e s ter ile neutri no canb e produ c ed t h rou g h the Dodelso n-Widrow (D W ) m echani sm [@ D odelso n:199 3j e ] w ith an appr oxim ate relic abund a nce h\^ 2 \~0.07()()\^ 2.Thi s is aco nse qu e nce of no n -re sonant o sc il latio n du e t o the mi xing betw een the activ e a nd s t eri le se ctors . Ho we ver,steril e neu tr inos produced t hrou gh the DW me ch ani sm acco u ntingfor al l of da rk matt e r h as n o wbeen robustly rule do u tfor allmasses ba se d on cons tr ain ts f r o m X-r ay b o un ds [@Boy arsky: 2 00 6f g; @Boy ar sky:20 06 ag; @W atson : 2006 qb; @B oyarsky: 2005u s ; @Boyarsky:20 0 7ay; @Boyarsk y :2 0 0 7g e ; @W ats on:2011dw;@Hor i uchi :201 3 no a]a nd Ly man-a lp h ad ata [@Seljak:2006qw ;@Asaka :2006 nq; @Boyarsky :2008xj];s e e [@Horiu chi: 2 01 3 noa] for a sum mary.
This pro b lem canbe ev aded inseveral w a y s. A lep ton as ymm etr y in the early Un i v erse c an lead to resona ntpro duc tio nof steril e neutri no s, r es ult ing i n a colde rdis tr ibu tiont hat ca n eva de L ym an - alp ha boun d s; t hisis k nown as t he Sh i-Fu l ler mechan ism [@Shi :19 9 8km] .An other a pproach is to c onsider st er ile neutr i n o produc tion from the decays of a singl etHiggs bos on from a n e xtende d H i ggs se ctor [ @Kuse nk o:2 0 0 6rh;@ P et rak i: 2007gq], w h i chcan p ro duce all of dark matter fromt hefreeze-in mec han ism[ @ Ha ll: 2 00 9 bx] ( f ora discussion in t he context o f t he 3.5 keV lin e, see [@ Merle:2 014xp a ]).
| Bulbul:2014sua; @Boyarsky:2014jta;_@Abazajian:2014gza; @Merle:2014xpa;_@Haba:2014taa; @Rodejohann:2014eka; @Abada:2014zra; @Ishida:2014fra;_@Robinson:2014bma; @Chakraborty:2014tma;_@Adulpravitchai:2014xna;_@Patra:2014pga; @Frigerio:2014ifa;_@Tsuyuki:2014aia;_@Kang:2014mea; @Harada:2014lma; @Ishida:2014dlp;_@Bezrukov:2014nza]. A keV_scale sterile neutrino can_be produced through_the_Dodelson-Widrow (DW) mechanism [@Dodelson:1993je] with an approximate relic abundance h\^2\~0.07 ()()\^2. This is a_consequence_of non-resonant_oscillation_due_to the mixing between the_active and sterile sectors. However,_sterile neutrinos_produced through the DW mechanism accounting for all_of_dark matter has_now been robustly ruled out for all masses based_on constraints from X-ray bounds [@Boyarsky:2006fg;_@Boyarsky:2006ag; @Watson:2006qb; @Boyarsky:2005us;_@Boyarsky:2007ay;_@Boyarsky:2007ge;_@Watson:2011dw; @Horiuchi:2013noa] and Lyman-alpha_data [@Seljak:2006qw; @Asaka:2006nq; @Boyarsky:2008xj]; see [@Horiuchi:2013noa]_for a summary.
This problem can be_evaded in several ways. A lepton asymmetry_in the early Universe can lead_to resonant production of sterile_neutrinos, resulting_in a colder distribution that_can evade Lyman-alpha_bounds; this_is known as_the Shi-Fuller mechanism [@Shi:1998km]. Another approach_is to consider_sterile neutrino production from the decays_of_a singlet Higgs_boson_from_an extended_Higgs sector [@Kusenko:2006rh;_@Petraki:2007gq],_which can_produce_all of dark matter from the_freeze-in_mechanism [@Hall:2009bx] (for a discussion in the_context of the 3.5_keV_line, see [@Merle:2014xpa]).
|
2}^n \sum_{j=1}^{k-1}
\l_{j,k} f(u_{j,k}, v_{j,k}), \quad f \in \Pi_{2n-3}^2,$$ where $\lambda_{j,k} = \l_j \l_k (x_{j,n} - x_{k,n})^2$.
The product of is a cubature rule on $[-1,1]^2$ $$\label{ProdGauss}
c_w^2 \int_{[-1,1]^2} f(x,y) w(x)w(y) dx dy = \sum_{k=1}^n \sum_{j=1}^n
\l_k \l_j f(x_{k,n}, x_{j,n}),$$ which is exact for $f \in \Pi_{2n-1} \times \Pi_{2n-1}$, the space of polynomials of degree at most $2n-1$ in either $x$ or $y$ variable. Applying on the symmetric polynomials $f(x+y, xy)$ and using the symmetry, we obtain $$c_w^2 \int_{\triangle} f(x+y, xy) w(x)w(y) dx dy = \sum_{k=1}^n \mathop{ {\sum}' }_{j=1}^k
\l_k \l_j f(x_{k,n}+ x_{j,n}, x_{k,n}x_{j,n}),$$ exact for polynomials $f$ in $\Pi_{2n-1}\times \Pi_{2n-1}$. Under the change of variables $u = x+y$ and $v = xy$ and by, the above cubature becomes, since $\Pi_{2n-1}\times \Pi_{2n-1}$ becomes $\Pi_{2n-1}^2$ under the mapping $(x,y) \mapsto (u,v)$. It is easy to see that has $\dim \Pi_{n-1}^2$ nodes, so that it is a Gaussian cubature rule.
To prove, we apply the product Gaussian cubature rule on functions of the form $(x - y)^2 f(x+y,xy)$ for $f \ | 2}^n \sum_{j=1}^{k-1 }
\l_{j, k } f(u_{j, k }, v_{j, k }), \quad f \in \Pi_{2n-3}^2,$$ where $ \lambda_{j, k } = \l_j \l_k (x_{j, n } - x_{k, n})^2$.
The product of is a cubature rule on $ [ -1,1]^2 $ $ $ \label{ProdGauss }
c_w^2 \int_{[-1,1]^2 } f(x, y) w(x)w(y) dx dy = \sum_{k=1}^n \sum_{j=1}^n
\l_k \l_j f(x_{k, n }, x_{j, n}),$$ which is accurate for $ f \in \Pi_{2n-1 } \times \Pi_{2n-1}$, the quad of polynomials of degree at about $ 2n-1 $ in either $ x$ or $ y$ variable. enforce on the symmetric polynomials $ f(x+y, xy)$ and practice the symmetry, we obtain $ $ c_w^2 \int_{\triangle } f(x+y, xy) w(x)w(y) dx dy = \sum_{k=1}^n \mathop { { \sum }' } _ { j=1}^k
\l_k \l_j f(x_{k, n}+ x_{j, n }, x_{k, n}x_{j, n}),$$ exact for polynomial $ f$ in $ \Pi_{2n-1}\times \Pi_{2n-1}$. Under the variety of variables $ u = x+y$ and $ five = xy$ and by, the above cubature becomes, since $ \Pi_{2n-1}\times \Pi_{2n-1}$ become $ \Pi_{2n-1}^2 $ under the mapping $ (x, y) \mapsto (u, v)$. It is easy to see that has $ \dim \Pi_{n-1}^2 $ node, so that it is a Gaussian cubature rule.
To prove, we apply the intersection Gaussian cubature rule on functions of the form $ (x - y)^2 f(x+y, xy)$ for $ f \ | 2}^n \skm_{j=1}^{k-1}
\l_{j,k} f(u_{j,k}, v_{j,k}), \quad f \in \Ki_{2b-3}^2,$$ wherx $\lambdz_{j,k} = \l_j \u_k (x_{j,n} - x_{k,n})^2$.
The product of is a cubaulre rule on $[-1,1]^2$ $$\label{ProaGauss}
c_w^2 \int_{[-1,1]^2} d(x,y) x(x)w(y) dx dy = \sum_{k=1}^i \sum_{j=1}^n
\l_n \l_j f(x_{k,n}, x_{j,n}),$$ wmich is exawt for $f \in \Pi_{2t-1} \giles \Pi_{2n-1}$, the space of polynomials of degree ah most $2n-1$ in eijher $q$ jr $y$ naviable. Applying on the symmetric polynokials $f(x+y, xy)$ snd using the symmetry, we lbtaln $$c_w^2 \int_{\triangle} v(x+y, xy) w(x)w(y) dx qt = \sum_{k=1}^n \matfop{ {\sum}' }_{j=1}^k
\m_k \l_j f(x_{k,n}+ x_{j,n}, x_{k,n}x_{j,n}),$$ exact for polyuomials $f$ ib $\Pi_{2n-1}\hhmes \Pi_{2n-1}$. Unver thv change of variables $u = x+y$ snd $v = xy$ and ny, thx abive cubature becomes, vince $\Pi_{2n-1}\times \Pi_{2n-1}$ becomes $\[i_{2u-1}^2$ under the mapping $(x,t) \napstm (u,v)$. It kw exsy tp aee thwt ias $\dim \Pi_{n-1}^2$ nodes, so tyat it is a Gaussiam slnature rule.
Tk provq, re apply the product Gaussian cubature gule on functions of the foem $(x - y)^2 f(x+y,xy)$ for $f \ | 2}^n \sum_{j=1}^{k-1} \l_{j,k} f(u_{j,k}, v_{j,k}), \quad f where = \l_j (x_{j,n} - x_{k,n})^2$. cubature on $[-1,1]^2$ $$\label{ProdGauss} \int_{[-1,1]^2} f(x,y) w(x)w(y) dy = \sum_{k=1}^n \sum_{j=1}^n \l_k \l_j x_{j,n}),$$ which is exact for $f \in \Pi_{2n-1} \times \Pi_{2n-1}$, the space of of degree at most $2n-1$ in either $x$ or $y$ variable. Applying on symmetric $f(x+y, and the symmetry, we obtain $$c_w^2 \int_{\triangle} f(x+y, xy) w(x)w(y) dx dy = \sum_{k=1}^n \mathop{ {\sum}' }_{j=1}^k \l_j f(x_{k,n}+ x_{j,n}, x_{k,n}x_{j,n}),$$ exact for polynomials $f$ $\Pi_{2n-1}\times \Pi_{2n-1}$. Under the of variables $u = x+y$ $v xy$ and the cubature since $\Pi_{2n-1}\times \Pi_{2n-1}$ $\Pi_{2n-1}^2$ under the mapping $(x,y) \mapsto (u,v)$. It is easy to see that has $\dim \Pi_{n-1}^2$ nodes, that it Gaussian cubature To we the product Gaussian on functions of the form $(x for $f \ | 2}^n \sum_{j=1}^{k-1}
\l_{j,k} f(u_{j,k}, v_{j,k}), \quad f \in \Pi_{2N-3}^2,$$ where $\lambDa_{j,k} = \l_J \l_k (X_{j,n} - X_{k,N})^2$.
The ProdUct of is a cubatuRE rulE on $[-1,1]^2$ $$\label{ProdGauss}
c_w^2 \int_{[-1,1]^2} F(x,y) w(x)W(y) DX dy = \sUM_{k=1}^N \sum_{j=1}^N
\l_k \l_j f(x_{K,N}, x_{J,N}),$$ WhiCh Is ExaCt FOr $F \in \Pi_{2N-1} \tiMes \Pi_{2n-1}$, tHe space of pOlyNoMials of degreE At Most $2n-1$ in eitHer $X$ or $y$ variable. appLying oN tHe sYMmetrIc pOlynoMials $f(X+Y, xy)$ and Using the sYmMEtry, we OBtain $$c_w^2 \INT_{\tRianGle} f(x+y, xy) w(x)w(y) dx dy = \sUM_{k=1}^N \Mathop{ {\sum}' }_{j=1}^k
\l_k \l_J f(x_{k,n}+ x_{J,n}, X_{K,n}X_{J,N}),$$ exAct For polynomIaLs $f$ in $\pI_{2n-1}\times \pI_{2n-1}$. uNDEr tHE change of variAbles $u = x+y$ and $V = Xy$ aNd by, thE aBovE CubatuRe becOmES, siNce $\Pi_{2n-1}\times \pi_{2n-1}$ bEcomes $\Pi_{2n-1}^2$ Under tHE mappinG $(X,y) \mapstO (u,v)$. It iS eaSy tO see THaT hAs $\dIm \pI_{n-1}^2$ nODeS, so THat It is a GauSsIaN cubaTure RULE.
to prOve, We apPly thE product GaussIan CubaTUre Rule oN funcTionS oF the fOrm $(x - y)^2 f(X+y,xy)$ fOr $F \ | 2}^n \sum_{j=1}^{k-1}
\l _{j,k } f (u_ {j ,k}, v_{ j,k}), \quad f \in\Pi_{2n-3}^2,$$ where$\lam bd a _{j, k }= \l_ j \l_k( x_ { j ,n} - x _{k ,n } )^ 2$.
The produc t of is acub at ure rule on$ [- 1,1]^2$ $$ \la bel{ProdGaus s} c_ w^ 2 \ i nt_{[ -1, 1]^2} f(x,y ) w(x)w (y) dx dy = \sum_{ k =1}^n \ s u m_ {j=1 }^n
\l _ k\ l_j f(x_{k,n}, x_{j, n} ) ,$ $ whi chis exact f or $f \ i n \Pi_{ 2 n- 1 } \ti m es \Pi_{2n-1} $, the spac e of polyn om ial s of de greeat mos t $2n-1$ in eit her $x$ o r $y$v ariable . Applyi ng onthe sy mmet r ic p oly no m ial s $ f(x + y,xy)$ and u si ng th e sy m m e t ry,weobta in $$ c_w^2 \int_{\ tri angl e } f (x+y, xy)w(x) w( y) dx dy =\sum_ {k =1}^n \mathop{{\su m}' }_{j= 1}^ k
\l_k\l_ j f (x_{k,n }+ x_{j , n}, x _ { k ,n }x_{j,n}),$$ exact f o r p olynomia ls $f$ in $ \ Pi_{2n-1 }\ tim es \ P i _{2n- 1}$. Un der thechange of v ariable s$u = x +y $ a nd$v =x y$ a nd by, the abo ve cu b ature becomes, since $\Pi_{2 n -1 } \ ti m es \ Pi_ {2n-1}$ bec omes $\Pi _{2n - 1} ^2$ under thema p pi n g $(x,y) \mapsto (u ,v )$. It is e asy to see th at has $\d i m \Pi_{n-1 }^2$ no d es, so that it is a Gaussianc ubaturerule.
To pro ve, we ap p l y the pr odu ctGau ssi a n c ubature ruleo n fun ct ions of th e form$(x -y)^ 2 f (x +y,xy)$ f or $f \ | 2}^n \sum_{j=1}^{k-1}_
_ _ __ _\l_{j,k}_f(u_{j,k}, v_{j,k}), \quad_f \in \Pi_{2n-3}^2,$$_where $\lambda_{j,k} = \l_j_\l_k (x_{j,n} -_x_{k,n})^2$.
The_product of is a cubature rule on $[-1,1]^2$ $$\label{ProdGauss}
c_w^2 \int_{[-1,1]^2}_f(x,y)_w(x)w(y) dx_dy_=_\sum_{k=1}^n \sum_{j=1}^n
_ _ _\l_k \l_j f(x_{k,n}, x_{j,n}),$$ which is exact for_$f_\in \Pi_{2n-1} \times_\Pi_{2n-1}$, the space of polynomials of degree at most_$2n-1$ in either $x$ or $y$_variable. Applying on_the_symmetric_polynomials $f(x+y, xy)$ and_using the symmetry, we obtain $$c_w^2_\int_{\triangle} f(x+y, xy) w(x)w(y) dx dy_= \sum_{k=1}^n \mathop{ {\sum}' }_{j=1}^k
_ _ \l_k \l_j_f(x_{k,n}+ x_{j,n},_x_{k,n}x_{j,n}),$$ exact for polynomials $f$_in $\Pi_{2n-1}\times \Pi_{2n-1}$._Under the_change of variables_$u = x+y$ and $v =_xy$ and by,_the above cubature becomes, since $\Pi_{2n-1}\times_\Pi_{2n-1}$_becomes $\Pi_{2n-1}^2$ under_the_mapping_$(x,y) \mapsto_(u,v)$. It is_easy_to see_that_has $\dim \Pi_{n-1}^2$ nodes, so that_it_is a Gaussian cubature rule.
To prove, we_apply the product Gaussian_cubature_rule on functions of_the form $(x - y)^2_f(x+y,xy)$ for $f \ |
=1$. The second representation is obtained from the first by calculating modulo the primitive polynomial $p(x)$. Finally, the $3$-tuple representation is obtained from the coefficients of the three powers $x^0=1$, $x^1=x$ and $x^2$. Taking these $3$-tuples as the points of a $3$-dimensional vector space, we recover the Fano plane $-$ see Fig. 1.
The lifted Fano plane in $GR(4^3)$
----------------------------------
We already know from Sect. \[MUBsQubits\] that the relevant object for $2^m$-dits is not the Galois field $GF(2^m)$, but rather the Galois ring $GR(4^m)$. It is therefore important to have a look at the geometry in the space $A=GR(4^3)$. For a ring, the concept of a vector space must be replaced by that of a module. The largest cycle in $A$ is the set $\mathcal{T}_3^*=\mathcal{T}_3-\{0\}$ (see Sect. 3), and each element of $\mathcal{T}_3^*$ can be represented in the same way as in the case of a Galois field. This is summarized in Table 2.
as powers of $\xi$ as polynomials as $3$-tuples in $\mathcal{Z}_4^3$ as $3$-tuples in $\mathcal{Z}_2^3$
-------------------- ---------------- ------------------------------------ ------------------------------------
0 0 (0,0,0) (0,0,0)
1 1 (0,0,1) (0,0,1)
$\xi$ $x$ (0,1,0) (0,1,0)
$\xi^2$ $x^2$ (1,0,0) (1,0,0)
$\xi^3$ $1+3x+2x^2$ (2,3,1) (0,1,1)
$\xi^4$ $2+3x+3x^2$ (3,3,2) (1,1,0)
$\xi^5$ $3+3x$+$x^2$ | = 1$. The second representation is obtained from the first by calculating modulo the crude polynomial $ p(x)$. ultimately, the $ 3$-tuple representation is obtained from the coefficients of the three power $ x^0=1 $, $ x^1 = x$ and $ x^2$. Taking these $ 3$-tuples as the points of a $ 3$-dimensional vector outer space, we recover the Fano plane $ -$ attend Fig. 1.
The lifted Fano plane in $ GR(4 ^ 3)$
----------------------------------
We already sleep together from Sect. \[MUBsQubits\ ] that the relevant object for $ 2^m$-dits is not the Galois field $ GF(2^m)$, but quite the Galois surround $ GR(4^m)$. It is therefore important to have a look at the geometry in the space $ angstrom = GR(4 ^ 3)$. For a ring, the concept of a vector space must be replaced by that of a faculty. The largest cycle in $ A$ is the determined $ \mathcal{T}_3^*=\mathcal{T}_3-\{0\}$ (see Sect. 3), and each component of $ \mathcal{T}_3^*$ can be represented in the same way as in the case of a Galois field. This is sum up in Table 2.
as powers of $ \xi$ as polynomials as $ 3$-tuples in $ \mathcal{Z}_4 ^ 3 $ as $ 3$-tuples in $ \mathcal{Z}_2 ^ 3 $
-------------------- ---------------- ------------------------------------ ------------------------------------
0 0 (0,0,0) (0,0,0)
1 1 (0,0,1) (0,0,1)
$ \xi$ $ x$ (0,1,0) (0,1,0)
$ \xi^2 $ $ x^2 $ (1,0,0) (1,0,0)
$ \xi^3 $ $ 1 + 3x+2x^2 $ (2,3,1) (0,1,1)
$ \xi^4 $ $ 2 + 3x+3x^2 $ (3,3,2) (1,1,0)
$ \xi^5 $ $ 3 + 3x$+$x^2 $ | =1$. Thf second representation ls obtained from the ficst by dalculatkng modulo the primitive polbnomual $p(z)$. Finally, the $3$-tuple reoresentatpon is obraintd from the coeffmdients of the bhree 'owers $x^0=1$, $x^1=x$ and $x^2$. Taking dhese $3$-tuples av ghz points of a $3$-dimensional vector spase, we rrclver the Fano klane $-$ see Fig. 1.
The lifted Fano plane in $GR(4^3)$
----------------------------------
We zlready know from Sevt. \[MUBsQubits\] that the relfvanh object for $2^m$-dits is not the Galjus field $GF(2^m)$, but ratheg the Galoia ring $GR(4^m)$. It is therefore impoftant to have a liok dt the geomxtry ig the space $A=GR(4^3)$. For d ring, yhe concept of a tectir space must be repleced by that of a moqule. The naxgest cycle in $A$ is tye set $\kathwal{T}_3^*=\oqthzal{U}_3-\{0\}$ (sxe Aect. 3), wnd each elemsnt of $\mathxal{T}_3^*$ can be represemtqe in the same way af yn the case of a Galois field. This is slmmadized in Table 2.
as poqers of $\xi$ as polynlmials ws $3$-tuples in $\mathcal{Z}_4^3$ as $3$-tuples in $\mathcal{Z}_2^3$
-------------------- ---------------- ------------------------------------ ------------------------------------
0 0 (0,0,0) (0,0,0)
1 1 (0,0,1) (0,0,1)
$\xi$ $x$ (0,1,0) (0,1,0)
$\xi^2$ $x^2$ (1,0,0) (1,0,0)
$\xi^3$ $1+3x+2x^2$ (2,3,1) (0,1,1)
$\di^4$ $2+3x+3x^2$ (3,3,2) (1,1,0)
$\xi^5$ $3+3x$+$x^2$ | =1$. The second representation is obtained from by modulo the polynomial $p(x)$. Finally, from coefficients of the powers $x^0=1$, $x^1=x$ $x^2$. Taking these $3$-tuples as the of a $3$-dimensional vector space, we recover the Fano plane $-$ see Fig. The lifted Fano plane in $GR(4^3)$ ---------------------------------- We already know from Sect. \[MUBsQubits\] the object $2^m$-dits not the Galois field $GF(2^m)$, but rather the Galois ring $GR(4^m)$. It is therefore important to a look at the geometry in the space For a ring, the of a vector space must replaced that of module. largest in $A$ is set $\mathcal{T}_3^*=\mathcal{T}_3-\{0\}$ (see Sect. 3), and each element of $\mathcal{T}_3^*$ can be represented in the same way in the a Galois This summarized Table 2. as $\xi$ as polynomials as $3$-tuples in in $\mathcal{Z}_2^3$ -------------------- ---------------- ------------------------------------ ------------------------------------ 0 0 (0,0,0) 1 (0,0,1) (0,0,1) $\xi$ $x$ (0,1,0) (0,1,0) $x^2$ (1,0,0) (1,0,0) $\xi^3$ $1+3x+2x^2$ (2,3,1) (0,1,1) $\xi^4$ (3,3,2) (1,1,0) $\xi^5$ $3+3x$+$x^2$ | =1$. The second representation is Obtained frOm the FirSt bY cAlcuLatiNg modulo the priMItivE polynomial $p(x)$. Finally, thE $3$-tuplE rEPresENtAtion Is obtaiNEd FROm tHe CoEffIcIEnTs of tHe tHree powErs $x^0=1$, $x^1=x$ and $x^2$. takInG these $3$-tuples AS tHe points of A $3$-diMensional vecTor Space, wE rEcoVEr the fanO planE $-$ see FiG. 1.
the lifTed Fano plAnE In $GR(4^3)$
----------------------------------
We ALready kNOW fRom SEct. \[MUBsQubits\] that THe RElevant object fOr $2^m$-ditS iS NoT THe GAloIs field $GF(2^m)$, BuT rathER the GalOIs RING $GR(4^M)$. it is therefore Important to HAve A look aT tHe gEOmetry In the SpACe $A=gR(4^3)$. For a ring, tHe coNcept of a vEctor sPAce must BE replacEd by thAt oF a mOdulE. thE lArgEsT CycLE iN $A$ iS The Set $\mathcAl{t}_3^*=\mAthcaL{T}_3-\{0\}$ (seE sECT. 3), and EacH eleMent oF $\mathcal{T}_3^*$ can bE rePresENteD in thE same Way aS iN the cAse of a galoiS fIeld. This is summaRizeD in Table 2.
aS poWeRs oF $\xI$ as poLYnomiaLs aS $3$-tuPles in $\mAthcal{Z}_4^3$ AS $3$-tuPlES IN $\mAthcal{Z}_2^3$
-------------------- ---------------- ------------------------------------ ------------------------------------
0 0 (0,0,0) (0,0,0)
1 1 (0,0,1) (0,0,1)
$\xi$ $x$ (0,1,0) (0,1,0)
$\xi^2$ $x^2$ (1,0,0) (1,0,0)
$\xi^3$ $1+3x+2x^2$ (2,3,1) (0,1,1)
$\xI^4$ $2+3x+3X^2$ (3,3,2) (1,1,0)
$\XI^5$ $3+3x$+$X^2$ | =1$. The second representa tion is ob taine d f rom t he f irst by calculatin g mod ulo the primitive poly nomia l$ p(x) $ .Final ly, the $3 $ - tup le r epr es e nt ation is obtain ed from th e c oe fficients of th e three po wer s $x^0=1$, $ x^1 =x$ an d$x^ 2 $. Ta kin g the se $3$ - tuples as the p oi n ts ofa $3$-di m e ns iona l vector space, w e r e cover the Fano plane $ - $s e e F ig. 1.
The l if ted F a no plan e i n $ GR( 4 ^3)$
-------- ----------- - --- ------ -- ---
We al ready k n owfrom Sect.\[MU BsQubits\ ] that the rel e vant ob ject f or$2^ m$-d i ts i s n ot the Ga loi s fi eld $GF( 2^ m) $, bu t ra t h e r the Ga lois ring $GR(4^m)$. I t i s th e ref ore i mport antto have a loo k atth e geometry in t he s pace $A=G R(4 ^3 )$. F or ar ing, t hecon cept of a vect o r s pa c e mu st be replaced byth a t o f a modu le. Th e l ar g est cycl ein$A$i s theset$ \m athcal{T }_3^*= \ ma th cal{T}_ 3- \{0\}$ ( see Se ct. 3 ) , an d each element of $ \ mathcal{T}_3^* $ can be repre s en t e di n th e s ame way asin t h e ca se o f a Ga l ois f ield. T h is is summarized in Ta bl e 2.
as powers of $\ xi$ as p o l y nomials as $3 $ -tuples in $\m athca l{Z}_4^3$ as $3$- tuple s in $\m athcal{Z} _ 2 ^3$
-- --- --- --- --- - - -- -- ---------- - - ---- - ------- --- ------- --- --- --- --- -- ---- ---- -------- -- -- -- -- --- ----- - ------- 0 0 ( 0, 0, 0) (0,0,0 )
1 1 ( 0,0,1) (0, 0,1 )
$\xi$ $x$ (0,1 , 0 ) (0 ,1,0)
$\xi^2$ $x^2$ (1, 0, 0 ) ( 1 ,0 ,0)
$\ xi ^3$ $ 1+3x+ 2 x^2$ (2, 3, 1) (0,1 , 1)
$\ xi^ 4$ $2+3 x+3x^2 $ (3,3 ,2) (1,1, 0)
$\x i^5$ $3 +3x$+$x^2 $ | =1$. The_second representation_is obtained from the_first by_calculating_modulo the_primitive_polynomial $p(x)$. Finally,_the $3$-tuple representation_is obtained from the_coefficients of the_three_powers $x^0=1$, $x^1=x$ and $x^2$. Taking these $3$-tuples as the points of a $3$-dimensional_vector_space, we_recover_the_Fano plane $-$ see Fig. 1.
The_lifted Fano plane in $GR(4^3)$
----------------------------------
We_already know_from Sect. \[MUBsQubits\] that the relevant object for_$2^m$-dits_is not the_Galois field $GF(2^m)$, but rather the Galois ring $GR(4^m)$._It is therefore important to have_a look at_the_geometry_in the space $A=GR(4^3)$._For a ring, the concept of_a vector space must be replaced_by that of a module. The largest_cycle in $A$ is the set_$\mathcal{T}_3^*=\mathcal{T}_3-\{0\}$ (see Sect. 3), and_each element_of $\mathcal{T}_3^*$ can be represented_in the same_way as_in the case_of a Galois field. This is_summarized in Table_2.
as powers of_$\xi$_ as_polynomials__ as_$3$-tuples in $\mathcal{Z}_4^3$__ as_$3$-tuples_in $\mathcal{Z}_2^3$
-------------------- ---------------- ------------------------------------_------------------------------------
_ _ 0__ _ _ _ 0_ _ _ _ (0,0,0)_ __ __ _ _ _ _ __(0,0,0)
_ 1_ ___ _ 1_ _ __ __ (0,0,1) ___ _ _ _ _ (0,0,1)
___ $\xi$ _ _ _ $x$ _ _ (0,1,0)_ _ _ __ ___(0,1,0)
_ _ _ $\xi^2$ _ _ _ _ __$x^2$ _ (1,0,0) _ __ _ _ _ (1,0,0)
__ _ $\xi^3$ _ _ _ _ $1+3x+2x^2$_ _ _ __ (2,3,1) ____ _ _ _ __ _ _ (0,1,1)
_ $\xi^4$ _ $2+3x+3x^2$ _ _ _(3,3,2) __ _ _ __ (1,1,0)
___ __$\xi^5$ _____$3+3x$+$x^2$ |
. Phys. [**327**]{}, 603 (2014), arXiv:1207.6734 \[hep-th\]. S. Carrozza, D. Oriti and V. Rivasseau, “Renormalization of a SU(2) Tensorial Group Field Theory in Three Dimensions,” Commun. Math. Phys. [**330**]{}, 581 (2014), arXiv:1303.6772 \[hep-th\]. J. B. Geloun and R. Toriumi, “Parametric representation of rank $d$ tensorial group field theory: Abelian models with kinetic term $\sum_{s}|p_s| + \mu$,” J. Math. Phys. [**56**]{}, no. 9, 093503 (2015), arXiv:1409.0398 \[hep-th\].
T. Krajewski and R. Toriumi, “Polchinski’s exact renormalisation group for tensorial theories: Gaussian universality and power counting,” arXiv:1511.09084 \[gr-qc\].
D. Benedetti, J. Ben Geloun and D. Oriti, “Functional Renormalisation Group Approach for Tensorial Group Field Theory: a Rank-3 Model,” JHEP [**1503**]{}, 084 (2015), arXiv:1411.3180 \[hep-th\]. D. Benedetti and V. Lahoche, “Functional Renormalization Group Approach for Tensorial Group Field Theory: A Rank-6 Model with Closure Constraint,” arXiv:1508.06384 \[hep-th\].
D. O. Samary, “Closed equations of the two-point functions for tensorial group field theory,” Class. Quant. Grav. [**31**]{}, 185005 (2014), arXiv:1401.2096 \[hep-th\].
R. C. Avohou, V. Rivasseau and A. Tanasa, “Renormalization and Hopf algebraic structure of the five-dimensional quartic tensor field theory,” J. Phys. A [**48**]{}, no. 48, 485204 (2015), arXiv:1507.03548 \[math-ph\].
D. Ousmane Samary, C. I. Perez-Sanchez, F | . Phys. [ * * 327 * * ] { }, 603 (2014), arXiv:1207.6734 \[hep - th\ ]. S. Carrozza, D. Oriti and V. Rivasseau, “ Renormalization of a SU(2) Tensorial Group Field Theory in Three Dimensions, ” Commun. Math. Phys. [ * * 330 * * ] { }, 581 (2014), arXiv:1303.6772 \[hep - th\ ]. J. B. Geloun and R. Toriumi, “ Parametric representation of rank $ d$ tensorial group field theory: Abelian models with energizing condition $ \sum_{s}|p_s| + \mu$, ” J. Math. Phys. [ * * 56 * * ] { }, no. 9, 093503 (2015), arXiv:1409.0398 \[hep - th\ ].
T. Krajewski and R. Toriumi, “ Polchinski ’s exact renormalisation group for tensorial theories: Gaussian universality and exponent count, ” arXiv:1511.09084 \[gr - qc\ ].
D. Benedetti, J. Ben Geloun and D. Oriti, “ Functional Renormalisation Group Approach for Tensorial Group Field Theory: a Rank-3 Model, ” JHEP [ * * 1503 * * ] { }, 084 (2015), arXiv:1411.3180 \[hep - th\ ]. D. Benedetti and V. Lahoche, “ Functional Renormalization Group Approach for Tensorial Group Field Theory: A Rank-6 Model with Closure Constraint, ” arXiv:1508.06384 \[hep - th\ ].
D. O. Samary, “ Closed equation of the two - point functions for tensorial group battlefield theory, ” Class. Quant. Grav. [ * * 31 * * ] { }, 185005 (2014), arXiv:1401.2096 \[hep - th\ ].
R. C. Avohou, V. Rivasseau and A. Tanasa, “ Renormalization and Hopf algebraic structure of the five - dimensional quartic tensor playing field theory, ” J. Phys. A [ * * 48 * * ] { }, no. 48, 485204 (2015), arXiv:1507.03548 \[math - ph\ ].
D. Ousmane Samary, C. I. Perez - Sanchez, F | . Phyd. [**327**]{}, 603 (2014), arXiv:1207.6734 \[hep-th\]. S. Carrozea, D. Oriti and V. Rirqsseau, “Renorjalizatiun of a SU(2) Tensorial Group Fmeld Theoey in Three Dimensions,” Commun. Manh. Phys. [**330**]{}, 581 (2014), arXmv:1303.6772 \[hep-th\]. J. B. Geloui and R. Toriumi, “Iaramztcic representatlon of rank $d$ tensorial gsojp field theory: Abelian models with kynetic yegm $\sum_{s}|p_s| + \mu$,” T. Matn. [hys. [**56**]{}, no. 9, 093503 (2015), arXiv:1409.0398 \[hep-th\].
T. Krajewski and D. Toriump, “Polchinski’s exavt renormalisation group flr tfnsorial theories: Haussian unuverfqlity and poder counting,” arXiv:1511.09084 \[gr-qd\].
D. Benedetti, J. Ben Geloun and D. Orkti, “Fbnctional Rguirmwnisation Griup Aiproach for Tcmsorian Group Field Theory: s Renk-3 Nodel,” JHEP [**1503**]{}, 084 (2015), arXiv:1411.3180 \[hxp-th\]. D. Benedetti and D. Lahoche, “Xuuctional Renormalizatuob Groop Ap[roazy fur Uenxodial Ggou' Field Thekry: A Rank-6 Nodel with Closure Vogwtraint,” arXiv:1508.06384 \[hep-th\].
Q. O. Famary, “Closed equations of the two-point fuhctions for tensorial geoup field theory,” Clads. Quant. Grwv. [**31**]{}, 185005 (2014), arXiv:1401.2096 \[hep-th\].
R. C. Avohou, V. Rivasseau and A. Tanasa, “Senorjxlieablon xbd Hopf algebraic structure of the five-dimensiogzl qlartic tensor fiejd theory,” J. Lhjs. S [**48**]{}, no. 48, 485204 (2015), arXiv:1507.03548 \[math-pk\].
S. Ohsmane Samary, C. I. Pegez-Sancrez, F | . Phys. [**327**]{}, 603 (2014), arXiv:1207.6734 \[hep-th\]. D. and V. “Renormalization of a in Dimensions,” Commun. Math. [**330**]{}, 581 (2014), \[hep-th\]. J. B. Geloun and R. “Parametric representation of rank $d$ tensorial group field theory: Abelian models with kinetic $\sum_{s}|p_s| + \mu$,” J. Math. Phys. [**56**]{}, no. 9, 093503 (2015), arXiv:1409.0398 \[hep-th\]. Krajewski R. “Polchinski’s renormalisation group for tensorial theories: Gaussian universality and power counting,” arXiv:1511.09084 \[gr-qc\]. D. Benedetti, J. Ben and D. Oriti, “Functional Renormalisation Group Approach for Group Field Theory: a Model,” JHEP [**1503**]{}, 084 (2015), \[hep-th\]. Benedetti and Lahoche, Renormalization Approach for Tensorial Field Theory: A Rank-6 Model with Closure Constraint,” arXiv:1508.06384 \[hep-th\]. D. O. Samary, “Closed equations of the functions for field theory,” Quant. [**31**]{}, (2014), arXiv:1401.2096 \[hep-th\]. Avohou, V. Rivasseau and A. Tanasa, algebraic structure of the five-dimensional quartic tensor field J. Phys. [**48**]{}, no. 48, 485204 (2015), arXiv:1507.03548 D. Ousmane Samary, C. I. Perez-Sanchez, F | . Phys. [**327**]{}, 603 (2014), arXiv:1207.6734 \[hep-th\]. S. Carrozza, D. ORiti and V. RiVasseAu, “REnoRmAlizAtioN of a SU(2) TensoriaL grouP Field Theory in Three DimEnsioNs,” cOmmuN. maTh. PhyS. [**330**]{}, 581 (2014), arXiv:1303.6772 \[hEP-tH\]. j. b. GeLoUn And r. TORiUmi, “PaRamEtric rePresentatiOn oF rAnk $d$ tensoriaL GrOup field thEorY: Abelian modeLs wIth kinEtIc tERm $\sum_{S}|p_s| + \Mu$,” J. MaTh. Phys. [**56**]{}, NO. 9, 093503 (2015), arXiv:1409.0398 \[Hep-th\].
T. KraJeWSki and r. toriumi, “pOLcHinsKi’s exact renormaliSAtIOn group for tensOrial tHeORiES: gauSsiAn universaLiTy and POwer couNTiNG,” ARXiV:1511.09084 \[Gr-qc\].
D. BenedettI, J. Ben Geloun ANd D. oriti, “FUnCtiONal RenOrmalIsATioN Group ApproAch fOr TensoriAl GrouP field ThEOry: a RanK-3 Model,” jHEp [**1503**]{}, 084 (2015), arxiv:1411.3180 \[hEP-tH\]. D. benEdETti ANd v. LaHOchE, “FunctioNaL REnormAlizATION GroUp APproAch foR Tensorial GroUp FIeld tHeoRy: A RaNk-6 ModEl wiTh closuRe ConsTrainT,” aRXiv:1508.06384 \[hep-th\].
D. O. SamaRy, “ClOsed equatIonS oF thE tWo-poiNT functIonS foR tensorIal grouP FieLd THEOrY,” Class. Quant. Grav. [**31**]{}, 185005 (2014), arXIv:1401.2096 \[HEP-tH\].
R. C. AvohoU, V. RivaSSeAu ANd A. TanasA, “REnoRmalIZAtion And HOPf AlgebraiC strucTUrE oF the fivE-dImensiOnAl qUarTic teNSor fIeld thEory,” J. PhyS. A [**48**]{}, no. 48, 485204 (2015), aRxiv:1507.03548 \[math-ph\].
D. OusmANe Samary, C. I. PerEZ-SANChEZ, F | . Phys. [**327**]{}, 603(2014), ar Xiv:1 207 .67 34 \[h ep-t h\]. S. Carroz z a, D . Oriti and V. Rivasse au, “ Re n orma l iz ation of a S U (2 ) Ten so ri alGr o up Fiel d T heory i n Three Di men si ons,” Commun . M ath. Phys. [ **330**]{},581 (2014 ), ar X iv:13 03. 6772\[hep- t h\]. J . B. Gelo un and R. Toriumi , “P aram etric representat i on of rank $d$ te nsoria lg ro u p fi eld theory: A be lianm odels w i th k i net i c term $\sum_ {s}|p_s| +\ mu$ ,” J.Ma th. Phys. [**5 6* * ]{} , no. 9, 09 3503 (2015),arXiv: 1 409.039 8 \[hep- th\].
T. Kr ajew s ki a ndR. Tor i um i,“ Pol chinski’ sex act r enor m a l i sati ongrou p for tensorial th eor ies: Gau ssian univ ersa li ty an d powe r cou nt ing,” arXiv:151 1.09 084 \[gr- qc\ ].
D .Bened e tti, J . B enGelounand D.O rit i, “ F un ctional Renormalis at i o nGroup Ap proach fo rT ensorial G rou p Fi e l d The ory: aRank-3 M odel,” JH EP [**150 3* *]{},08 4 ( 201 5), a r Xiv: 1411.3 180 \[he p-th\ ] . D. Benedetti and V. Lahoch e ,“ F un c tion alRenormaliza tion Grou p Ap p ro ach for T ensor ia l G r oup Field Theory: A R ank-6Model with Closure Constrain t , ” arXiv:1 508. 0 63 8 4 \[hep-th\].
D. O . Samary,“ Closed e quati ons of t he two-po i n t functi ons fo r t ens o r ia l group field t heor y, ” Class . Q uant. G rav . [** 31* *] {}, 18500 5 (2014) ,ar Xi v: 140 1.209 6 \[hep-t h\ ].
R . C . Avo h ou, V. Riva ssea uan d A. Tanasa , “ R e norm al iz atio n a nd Hopf alg e bra ic stru cture ofthe five -d im ensiona l quartic ten so r field th eo ry, ” J. P h y s. A [** 48**]{}, no. 48, 485204 (2015), ar Xiv:1 507. 03548 \[m ath -ph\].
D . Ousma ne Sam ary,C. I. P erez- S a nc hez ,F | . Phys. [**327**]{},_603 (2014),_arXiv:1207.6734 \[hep-th\]. S. Carrozza, D. Oriti_and V. Rivasseau,_“Renormalization_of a_SU(2)_Tensorial Group Field_Theory in Three_Dimensions,” Commun. Math. Phys. [**330**]{}, 581_(2014), arXiv:1303.6772 \[hep-th\]._J. B. Geloun_and R. Toriumi, “Parametric representation of rank $d$ tensorial group field theory: Abelian models with_kinetic_term $\sum_{s}|p_s|_+_\mu$,”_J. Math. Phys. [**56**]{}, no. 9, 093503_(2015), arXiv:1409.0398 \[hep-th\].
T. Krajewski and R. Toriumi,_“Polchinski’s exact_renormalisation group for tensorial theories: Gaussian universality and_power_counting,” arXiv:1511.09084 \[gr-qc\].
D. Benedetti,_J. Ben Geloun and D. Oriti, “Functional Renormalisation Group Approach for_Tensorial Group Field Theory: a Rank-3_Model,” JHEP [**1503**]{},_084_(2015),_arXiv:1411.3180 \[hep-th\]. D. Benedetti and_V. Lahoche, “Functional Renormalization Group Approach for_Tensorial Group Field Theory: A Rank-6_Model with Closure Constraint,” arXiv:1508.06384 \[hep-th\].
D. O. Samary, “Closed_equations of the two-point functions for_tensorial group field theory,” Class. Quant. Grav. _[**31**]{}, 185005_(2014), arXiv:1401.2096 \[hep-th\].
R. C. Avohou, V. Rivasseau and_A. Tanasa, “Renormalization and_Hopf algebraic_structure of the_five-dimensional quartic tensor field theory,” J. Phys. A_[**48**]{}, no. 48,_485204 (2015), arXiv:1507.03548 \[math-ph\].
D. Ousmane Samary, C. I. Perez-Sanchez,_F |
x})^{-1}$ (shown as “FF-$\sigma$” blocks in Figure \[fig:whole\_model\]). We compute the [**base chatter intensity**]{} corresponding to $s_j$ as $B_j = r_j\Tilde{B}_j$ such that $0<r_j<1$ plays the role of a scaling factor to calibrate $B_j$ according to the activity level of the subreddit.
Having computed $B_j$, the chatter intensity invoked by $s_j$, under the influence of past history of news and submission arrival and calibrated by the subreddit information, we next observe the commenting activity under $s_j$ within the observation window (see component (4) in Figure \[fig:whole\_model\]). We employ a binning over time intervals to transform the comment arrivals within the observation window $[t_j, t_j+m\Delta_\text{obs}]$ into a sequence $\langle c^1_j, c^2_j, \cdots, c^m_j\rangle$ where each $c^l_j$ is the total number of comments arrived within $[t_j+(l-1)\Delta_\text{obs}, t_j+l\Delta_\text{obs}]$. This sequence serves as a coarse approximation of the rate of comment arrivals over the observation window. We use a single LSTM layer to aggregate this sequence and predict the final chatter $y^m_j$ (superscript corresponds to the length of the observation window). In the zero-shot setting (i.e., $m=0$) this LSTM is not used and we predict chatter $y^0_j = B_j$.
Details of parameters are given in Appendix \[appendix:parameters\].
Cost/Loss Functions {#subsec:cost_function}
-------------------
In a realistic setting, there are far too many discussions invoking near-zero chatter along with very small number of those which go viral heavily. As we take both of these types without any filtering (opposed to excluding less viral ones in the cascade prediction tasks like [@kobayashi2016tideh] or [@zhao2015seismic]), the cost function needs to handle skewed ground truth values. To deal with this, we train by minimizing the mean absolute relative error given by $\sum_j \frac{\lvert y_j - y^m | x})^{-1}$ (shown as “ FF-$\sigma$ ” blocks in Figure \[fig: whole\_model\ ]). We compute the [ * * base yak volume * * ] { } corresponding to $ s_j$ as $ B_j = r_j\Tilde{B}_j$ such that $ 0 < r_j<1 $ plays the function of a scaling divisor to calibrate $ B_j$ according to the activity horizontal surface of the subreddit.
Having calculate $ B_j$, the chatter intensity invoked by $ s_j$, under the influence of past history of news and submission arrival and calibrate by the subreddit information, we next observe the gloss activity under $ s_j$ within the observation window (see component (4) in Figure \[fig: whole\_model\ ]). We use a binning over time intervals to transform the gossip arrivals within the observation window $ [ t_j, t_j+m\Delta_\text{obs}]$ into a sequence $ \langle c^1_j, c^2_j, \cdots, c^m_j\rangle$ where each $ c^l_j$ is the entire number of comments arrived within $ [ t_j+(l-1)\Delta_\text{obs }, t_j+l\Delta_\text{obs}]$. This sequence serves as a coarse estimate of the rate of comment arrivals over the observation window. We use a individual LSTM layer to aggregate this sequence and predict the final chatter $ y^m_j$ (superscript corresponds to the length of the observation window). In the zero - shot setting (i.e., $ m=0 $) this LSTM is not use and we predict chatter $ y^0_j = B_j$.
Details of parameter are pass in Appendix \[appendix: parameters\ ].
Cost / Loss Functions { # subsec: cost_function }
-------------------
In a realistic setting, there are far too many discussions invoking near - zero chatter along with very small numeral of those which go viral heavily. As we take both of these types without any filtering (opposed to excluding less viral one in the cascade prediction tasks like [ @kobayashi2016tideh ] or [ @zhao2015seismic ]), the cost function need to handle skew flat coat truth values. To deal with this, we train by minimize the mean absolute relative error grant by $ \sum_j \frac{\lvert y_j - y^m | x})^{-1}$ (sjown as “FF-$\sigma$” blocks ik Figure \[fig:whole\_model\]). Wx compufe the [**bxse chatter intensity**]{} corres'ondung ti $s_j$ as $B_j = r_j\Tilde{B}_j$ ruch that $0<r_j<1$ playw tht role of a scaliif factov to dwlibxave $B_j$ according to the acdivity level ox ghz subreddit.
Having computed $B_j$, the chwtter imtfnsity invoked by $x_t$, unsvr the influence of past historg of nexs and submissipn arrival and calibrated hy tje subreddit inforlation, we ngst jvserve the cummenting activity undgr $s_j$ within the observation winaow (sze componenj (4) ij Figure \[fig:wiole\_moqel\]). We emplon a binting ovrr time intervsls to transform the commenv arrivals within thg observathou window $[t_j, t_j+m\Delta_\twxr{obs}]$ hnto a sdwuevce $\lengme c^1_j, f^2_j, \rdots, c^m_j\rahgle$ where wach $c^l_j$ is the totsl bumber of comjents wrwived within $[t_j+(l-1)\Delta_\text{obs}, t_j+l\Delta_\teqt{oba}]$. This sequence serves qs a coarse approximajion of thq rate of comment arrivals over the observation whndow. De bwe a runhle LSTM layer to aggregate this sequence and lrtdibt the final chatber $y^m_j$ (superscripy fottesponds to thg lengtk or the observation aindow). Yn thw zero-shou setying (i.e., $m=0$) this LSTM is not ysed and we ireduct chatter $y^0_j = B_j$.
Betails of pcrametgrs arr given in Appendix \[appeudix:padameters\].
Cosh/Loss Fundgions {#subsec:cost_wunbtiot}
-------------------
In a realistic setting, thqre are fer top many aiscossions invoking jear-zcso chatter along wlth vgry smdll number of those which go viral heavilb. As we take nodh mf these typex without any filtering (opppsed to excljding less viral mnes in the cascade predhftion tasks nike [@kobwyasyi2016tieeh] or [@xfao2015seismic]), the cost funbtnon needs to handle skewed nrouna truth values. Ti dtao with this, we trxin bj mmnimishng the mean absulugr relxtive error eivem by $\sum_j \frac{\lvert f_j - g^m | x})^{-1}$ (shown as “FF-$\sigma$” blocks in Figure compute [**base chatter corresponding to $s_j$ that plays the role a scaling factor calibrate $B_j$ according to the activity of the subreddit. Having computed $B_j$, the chatter intensity invoked by $s_j$, under influence of past history of news and submission arrival and calibrated by the information, next the activity under $s_j$ within the observation window (see component (4) in Figure \[fig:whole\_model\]). We employ a over time intervals to transform the comment arrivals the observation window $[t_j, into a sequence $\langle c^1_j, \cdots, where each is total of comments arrived $[t_j+(l-1)\Delta_\text{obs}, t_j+l\Delta_\text{obs}]$. This sequence serves as a coarse approximation of the rate of comment arrivals over the window. We single LSTM to this and predict the $y^m_j$ (superscript corresponds to the length window). In the zero-shot setting (i.e., $m=0$) this is not and we predict chatter $y^0_j = Details of parameters are given in Appendix \[appendix:parameters\]. Functions {#subsec:cost_function} ------------------- In a realistic setting, there are far too many discussions invoking near-zero with very small number those which go heavily. we both these types any filtering (opposed to excluding less viral ones in the cascade tasks like [@kobayashi2016tideh] or [@zhao2015seismic]), the cost function needs to ground values. To deal this, we train by the absolute relative error given \frac{\lvert - | x})^{-1}$ (shown as “FF-$\sigma$” blocks in FiGure \[fig:whoLe\_modEl\]). WE coMpUte tHe [**baSe chatter intenSIty**]{} cOrresponding to $s_j$ as $B_j = r_j\tilde{b}_j$ SUch tHAt $0<R_j<1$ plaYs the roLE oF A ScaLiNg FacToR To CalibRatE $B_j$ accoRding to the ActIvIty level of thE SuBreddit.
HavIng Computed $B_j$, thE chAtter iNtEnsITy invOkeD by $s_j$, Under tHE influEnce of pasT hIStory oF News and SUBmIssiOn arrival and calibRAtED by the subreddiT inforMaTIoN, WE neXt oBserve the cOmMentiNG activiTY uNDER $s_j$ WIthin the obserVation windoW (See ComponEnT (4) in fIgure \[fIg:whoLe\_MOdeL\]). We employ a bInniNg over timE interVAls to trANsform tHe commEnt ArrIvalS WiThIn tHe OBseRVaTioN WinDow $[t_j, t_j+m\deLtA_\text{Obs}]$ iNTO A SequEncE $\lanGle c^1_j, C^2_j, \cdots, c^m_j\ranGle$ WherE EacH $c^l_j$ iS the tOtal NuMber oF commeNts arRiVed within $[t_j+(l-1)\DelTa_\teXt{obs}, t_j+l\DEltA_\tExt{ObS}]$. This SEquencE seRveS as a coaRse apprOXimAtION Of The rate of comment arRiVALs Over the oBservaTIoN wINdow. We usE a SinGle Lstm layeR to aGGrEgate thiS sequeNCe AnD predicT tHe finaL cHatTer $Y^m_j$ (suPErscRipt coRrespondS to thE Length of the obsERvation window). iN tHE ZeRO-shoT seTting (i.e., $m=0$) thiS LSTm Is noT useD AnD we PRedicT chatTeR $Y^0_j = b_J$.
Details of parameterS aRe giveN in ApPendix \[appendiX:parameterS\].
cOSt/Loss FuNctiONs {#SUbsec:cost_functIon}
-------------------
In A realistic SEtting, thEre arE far too mAny discusSIOns invokIng NeaR-zeRo cHATtEr along with veRY SmalL nUmber of ThoSe which Go vIraL heAviLy. as we take bOth of theSe TyPeS wIthOut anY FilterinG (oPpoSeD to ExcluDIng lesS viraL oneS iN tHE caScade prEDiCTIon tAsKs Like [@KobAyAshi2016tIdeh] OR [@zhAo2015seismIc]), the cost FunCTion NeEdS to handLe skewed grounD tRuth values. to DeaL with tHIS, we train By minimizing the mean absoLUte relaTivE erroR givEn by $\sum_j \fRac{\Lvert y_J - y^m | x})^{-1}$ (shown as “FF-$\ sigma$” bl ocksinFig ur e \[ fig: whole\_model\] ) . We compute the [**base c hatte ri nten s it y**]{ } corre s po n d ing t o$s_ j$ as $B_j =r_j\Til de{B}_j$ s uch t hat $0<r_j<1 $ p lays the r ole of a scalin g f actorto ca l ibrat e $ B_j$accord i ng tothe activ it y level of thes u br eddi t.
Having comput e d$ B_j$, the chat ter in te n si t y in vok ed by $s_j $, unde r the in f lu e n c e o f past history of news an d su bmissi on ar r ival a nd ca li b rat ed by the s ubre ddit info rmatio n , we ne x t obser ve the co mme ntin g a ct ivi ty und e r$s_ j $ w ithin th eob serva tion w i n dow(se e co mpone nt (4) in Fig ure \[f i g:w hole\ _mode l\]) .We em ploy a binn in g over time int erva ls to tra nsf or m t he comm e nt arr iva lswithinthe obs e rva ti o n wi ndow $[t_j, t_j+m\ De l t a_ \text{ob s}]$ i n to a sequence $ \la ngle c ^1_j, c^2 _ j, \cdots, c^m_j \ ra ng le$ whe re each$c ^l_ j$is th e tot al num ber of c ommen t s arrived with i n $[t_j+(l-1) \ De l t a_ \ text {ob s}, t_j+l\D elta _ \tex t{ob s }] $.T his s equen ce se r ves as a coarse app ro ximati on of the rate ofcomment ar r i v als over the ob s ervation windo w. We use a sin g le LSTMlayer to aggr egate thi s sequence an d p red ict t he final chatte r $y^m _j $ (supe rsc ript co rre spo nds to t he length of theob se rv at ion wind o w). In t he ze ro -sh ot se t ting ( i.e., $m= 0$ )t his LSTM i s n o t use dan d we pr ed ict c hatt e r $ y^0_j = B_j$.
D eta i ls o f p aramete rs are givenin Appendix\[ app endix: p a rameters \].
Cost/Loss Function s {#subs ec: cost_ func tion}
--- --- ------ --- - ---
I n a re alist ic se t t ing,t h er e a re far too m a n y d iscus si onsinvokin g near-zero chatte r al ong with very sm alln u mb ero ft hos ew hic h go viral heavil y. As we t ak e b oth of the s e t yp es with out any filt e ring (o pposed to excludin gless v ira l ones inthe casc ade predi c tiont as ks li ke[@koba ya shi 2016t ideh]o r [ @zhao 2015se is mic]), theco st funct ion needs to handle ske wed gr oundtru th values . T o de al with t his, we train by mi nimiz ing the m eana bs olu t e rel ativ e error gi v en by $ \s um_j \frac{ \ l v ert y_j- y ^ m | x})^{-1}$ (shown_as “FF-$\sigma$”_blocks in Figure \[fig:whole\_model\]). We_compute the_[**base_chatter intensity**]{}_corresponding_to $s_j$ as_$B_j = r_j\Tilde{B}_j$_such that $0<r_j<1$ plays_the role of_a_scaling factor to calibrate $B_j$ according to the activity level of the subreddit.
Having computed_$B_j$,_the chatter_intensity_invoked_by $s_j$, under the influence_of past history of news_and submission_arrival and calibrated by the subreddit information, we_next_observe the commenting_activity under $s_j$ within the observation window (see component_(4) in Figure \[fig:whole\_model\]). We employ a_binning over time_intervals_to_transform the comment arrivals_within the observation window $[t_j, t_j+m\Delta_\text{obs}]$_into a sequence $\langle c^1_j, c^2_j,_\cdots, c^m_j\rangle$ where each $c^l_j$ is the_total number of comments arrived within_$[t_j+(l-1)\Delta_\text{obs}, t_j+l\Delta_\text{obs}]$. This sequence serves_as a_coarse approximation of the rate_of comment arrivals_over the_observation window. We_use a single LSTM layer to_aggregate this sequence_and predict the final chatter $y^m_j$_(superscript_corresponds to the_length_of_the observation_window). In the_zero-shot_setting (i.e.,_$m=0$)_this LSTM is not used and_we_predict chatter $y^0_j = B_j$.
Details of parameters_are given in Appendix \[appendix:parameters\].
Cost/Loss_Functions_{#subsec:cost_function}
-------------------
In a realistic setting,_there are far too many_discussions invoking near-zero chatter along with_very small_number of_those which go viral heavily. As we take both of these_types without any filtering (opposed to_excluding less viral ones_in the_cascade_prediction tasks like_[@kobayashi2016tideh]_or [@zhao2015seismic]),_the cost function needs to handle skewed_ground truth_values. To deal with this, we_train by minimizing the_mean_absolute relative error given by $\sum_j_\frac{\lvert y_j - y^m |
- '**Mehmet Acikgoz**'
- '**Hassan Jolany**'
- '**Yuan He**'
title: '**Identities involving** $q$**-Genocchi numbers and polynomials**'
---
**PRELIMINARIES**
=================
Recently, *Kim* and *Lee* have given some properties for the $%
q $-Euler numbers and polynomials in [@Kim]. Actually, this type numbers and polynomials and their $q$-extensions or variously generalizations have been studied in several different ways for a long time (for details, see \[1-21\]).
By using $p$-adic $q$-integral on $%
%TCIMACRO{\U{2124} }%
%BeginExpansion
\mathbb{Z}
%EndExpansion
_{p}$, *Kim* defined many new generating functions of the $q$-Bernoulli poynomials, $q$-Euler polynomials and $q$-Genocchi polynomials in his arithmetic works (for details, see \[6-19\]). The works of *Kim* have been benefited for further works of many mathematicians in Analytic numbers theory. Actually, we motivated from his inspiring works to write this paper.
Assume that $p$ be a fixed odd prime number. Throughout this work, we require the definitions of the some notations such that let $%
%TCIMACRO{\U{211a} }%
%BeginExpansion
\mathbb{Q}
%EndExpansion
_{p}$ be the field $p$-adic rational numbers and let $%
%TCIMACRO{\U{2102} }%
%BeginExpansion
\mathbb{C}
%EndExpansion
_{p}$ be the completion of algebraic closure of $%
%TCIMACRO{\U{211a} }%
%BeginExpansion
\mathbb{Q}
%EndExpansion
_{p}$. That is, $$\boldsymbol{%
%TCIMACRO{\U{211a} }%
%BeginExpansion
\mathbb{Q}
%EndExpansion
}_{p}=\left\{ x=\sum_{n=-k}^{\infty }a_{n}p^{n}:0\leq a_{n}\leq p-1\right\}
\text{.}$$
Then $%
%TCIMACRO{\U{2124} }%
%Begin | -' * * Mehmet Acikgoz * *'
-' * * Hassan Jolany * *'
-' * * Yuan He * *'
title:' * * Identities involving * * $ q$**-Genocchi numbers and polynomial * *'
---
* * PRELIMINARIES * *
= = = = = = = = = = = = = = = = =
Recently, * Kim * and * Lee * have render some properties for the $%
q $ -Euler numbers and polynomial in [ @Kim ]. Actually, this type number and polynomials and their $ q$-extensions or variously generalizations have been study in several different way for a long time (for detail, see \[1 - 21\ ]).
By using $ p$-adic $ q$-integral on $%
% TCIMACRO{\U{2124 } }%
% BeginExpansion
\mathbb{Z }
% EndExpansion
_ { p}$, * Kim * define many new generating functions of the $ q$-Bernoulli poynomials, $ q$-Euler polynomials and $ q$-Genocchi polynomials in his arithmetic works (for detail, see \[6 - 19\ ]). The works of * Kim * have been benefited for further work of many mathematicians in Analytic numbers theory. Actually, we motivated from his inspire works to write this paper.
Assume that $ p$ be a fixed curious prime number. Throughout this work, we require the definitions of the some notations such that let $%
% TCIMACRO{\U{211a } }%
% BeginExpansion
\mathbb{Q }
% EndExpansion
_ { p}$ be the field $ p$-adic rational numbers and let $%
% TCIMACRO{\U{2102 } }%
% BeginExpansion
\mathbb{C }
% EndExpansion
_ { p}$ be the completion of algebraic closure of $%
% TCIMACRO{\U{211a } }%
% BeginExpansion
\mathbb{Q }
% EndExpansion
_ { p}$. That is, $ $ \boldsymbol{%
% TCIMACRO{\U{211a } }%
% BeginExpansion
\mathbb{Q }
% EndExpansion
} _ { p}=\left\ { x=\sum_{n=-k}^{\infty } a_{n}p^{n}:0\leq a_{n}\leq p-1\right\ }
\text{.}$$
Then $%
% TCIMACRO{\U{2124 } }%
% Begin | - '**Mejmet Acikgoz**'
- '**Hassan Jolaky**'
- '**Yuan He**'
title: '**Neentitmes invklving** $q$**-Eenocchi numbers and polynommals**'
---
**PRELININARIES**
=================
Recently, *Kim* avd *Lee* hane given wome properties for the $%
q $-Eulsv numyecs and polynomisls in [@Kim]. Actually, this thpz numbers and polynomials and their $z$-extensoojs or variousli gentrajizafpoks have been studied in several siffereit ways for a lpng time (for details, see \[1-21\]).
Bj uslng $p$-adic $q$-integrap on $%
%TCIMACTK{\U{2124} }%
%FwginExpansiov
\mathbb{Z}
%EndExpansion
_{p}$, *Iim* defined many new generating funccions of thg $s$-Bftnoulli poynimialf, $q$-Euler polnmomialv and $q$-Benocchi polynpmiels un his arithmetic worns (for details, see \[6-19\]). The worns of *Kim* have been bwnefijed fmr fjethdr socks of majy jathematicjans in Anaoytic numbers theoru. Wbyually, we mofivateq srom his inspiring works to write this kaper.
Zssume that $p$ be a fixee odd prime number. Thtoughout tris work, we require the definitions of the some nmtatikvs wugh tfqt let $%
%TCIMACRO{\U{211a} }%
%BeginExpansion
\mathbb{Q}
%EndExpanfjom
_{p}$ be the field $i$-adic rational numnegs wnd let $%
%TCIMAZRO{\U{2102} }%
%BzfihExpansion
\mathbb{C}
%EjdExpanfion
_{p}$ be the cjmplrtion of algebraic closure if $%
%TCIMACRO{\U{211c} }%
%BwginExpansion
\mathby{Q}
%EndExpansiun
_{p}$. Jhat ix, $$\boldsymbol{%
%TCIMACRO{\U{211a} }%
%YeginEspansion
\matjbb{Q}
%EndExlxnsion
}_{p}=\left\{ x=\sum_{v=-k}^{\ikftf }a_{n}p^{n}:0\leq a_{n}\leq p-1\right\}
\texe{.}$$
Then $%
%TCINACRP{\U{2124} }%
%Begiv | - '**Mehmet Acikgoz**' - '**Hassan Jolany**' - title: involving** $q$**-Genocchi and polynomials**' --- *Lee* given some properties the $% q numbers and polynomials in [@Kim]. Actually, type numbers and polynomials and their $q$-extensions or variously generalizations have been studied several different ways for a long time (for details, see \[1-21\]). By using $q$-integral $% }% \mathbb{Z} %EndExpansion _{p}$, *Kim* defined many new generating functions of the $q$-Bernoulli poynomials, $q$-Euler polynomials and polynomials in his arithmetic works (for details, see The works of *Kim* been benefited for further works many in Analytic theory. we from his inspiring to write this paper. Assume that $p$ be a fixed odd prime number. Throughout this work, we the definitions some notations that $% }% %BeginExpansion \mathbb{Q} be the field $p$-adic rational numbers %TCIMACRO{\U{2102} }% %BeginExpansion \mathbb{C} %EndExpansion _{p}$ be the of algebraic of $% %TCIMACRO{\U{211a} }% %BeginExpansion \mathbb{Q} _{p}$. That is, $$\boldsymbol{% %TCIMACRO{\U{211a} }% %BeginExpansion \mathbb{Q} }_{p}=\left\{ x=\sum_{n=-k}^{\infty }a_{n}p^{n}:0\leq a_{n}\leq p-1\right\} \text{.}$$ Then $% %TCIMACRO{\U{2124} }% %Begin | - '**Mehmet Acikgoz**'
- '**Hassan Jolany**'
- '**yuan He**'
titlE: '**IdenTitIes InVolvIng** $q$**-genocchi numberS And pOlynomials**'
---
**PRELIMINARIEs**
=================
ReceNtLY, *Kim* ANd *lee* haVe given SOmE PRopErTiEs fOr THe $%
Q $-EuleR nuMbers anD polynomiaLs iN [@KIm]. Actually, thIS tYpe numbers And Polynomials aNd tHeir $q$-eXtEnsIOns or VarIouslY generALizatiOns have beEn STudied IN severaL DIfFereNt ways for a long timE (FoR Details, see \[1-21\]).
By usIng $p$-adIc $Q$-InTEGraL on $%
%tCIMACRO{\U{2124} }%
%BEgInExpANsion
\maTHbB{z}
%eNdEXPansion
_{p}$, *Kim* deFined many neW GenEratinG fUncTIons of The $q$-BErNOulLi poynomialS, $q$-EuLer polynoMials aND $q$-GenocCHi polynOmials In hIs aRithMEtIc WorKs (FOr dETaIls, SEe \[6-19\]). THe works oF *KIm* Have bEen bENEFIted For FurtHer woRks of many mathEmaTiciANs iN AnalYtic nUmbeRs TheorY. ActuaLly, we MoTivated from his iNspiRing works To wRiTe tHiS papeR.
assume ThaT $p$ bE a fixed Odd primE NumBeR. tHRoUghout this work, we reQuIRE tHe definiTions oF ThE sOMe notatiOnS suCh thAT Let $%
%TCiMACro{\U{211A} }%
%BeginExPansioN
\MaThBb{Q}
%EndEXpAnsion
_{P}$ bE thE fiEld $p$-aDIc raTional Numbers aNd let $%
%tcIMACRO{\U{2102} }%
%BeginEXPansion
\mathbb{c}
%enDeXpANsioN
_{p}$ bE the completIon oF AlgeBraiC ClOsuRE of $%
%TCiMACRo{\U{211A} }%
%beGInExpansion
\mathbb{Q}
%ENdexpansIon
_{p}$. THat is, $$\boldsymbOl{%
%TCIMACRO{\u{211A} }%
%bEginExpaNsioN
\MaTHbb{Q}
%EndExpansiOn
}_{p}=\leFt\{ x=\sum_{n=-k}^{\inFTy }a_{n}p^{n}:0\leQ a_{n}\leQ p-1\right\}
\tExt{.}$$
Then $%
%TCimaCRO{\U{2124} }%
%BegIn | - '**Mehmet Acikgoz**'
- ' **Hassan J olany **'
-'* *Yua n He **'
title: '** I dent ities involving** $q$* *-Gen oc c hi n u mb ers a nd poly n om i a ls* *'
- --
* * PR ELIMI NAR IES**
= ========== === == =
Recently, *K im* and *L ee* have givensom e prop er tie s forthe $%
q $-Eul e r numb ers and p ol y nomial s in [@K i m ]. Act ually, this typen um b ers and polyno mialsan d t h e ir$q$ -extension sor va r iouslyg en e r a liz a tions have be en studiedi n s everal d iff e rent w ays f or a l ong time (f or d etails, s ee \[1 - 21\]).By usin g $p$- adi c $ q$-i n te gr alon $%% TC IMA C RO{ \U{2124} } %%Begi nExp a n s i on
\ mat hbb{ Z}
%E ndExpansion
_ {p} $, * K im* defi ned m anyne w gen eratin g fun ct ions of the $q$ -Ber noulli po yno mi als ,$q$-E u ler po lyn omi als and $q$-Ge n occ hi p o ly nomials in his ari th m e ti c works(for d e ta il s , see \[ 6- 19\ ]).T h e wor ks o f * Kim* hav e been be ne fited f or furth er wo rks of m a ny m athema ticiansin An a lytic numberst heory. Actual l y, w em otiv ate d from hisinsp i ring wor k stow ritethispa p er .
Assume that $p$ b ea fixe d odd prime number . Througho u t this wor k, w e r e quire the defi nitio ns of thes ome nota tions such th at let $% % TCIMACRO {\U {21 1a} }% % Be ginExpansion\ m athb b{ Q}
%End Exp ansion_{p }$bethe f ield $p$- adic rat io na lnu mbe rs an d let $%%T CIM AC RO{ \U{21 0 2} }%%Begi nExp an si o n
\ mathbb{ C }% E ndEx pa ns ion_{p }$ be t he c o mpl etion o f algebra icc losu re o f $%
%T CIMACRO{\U{21 1a } }%
%Begi nE xpa nsion\ m athbb{Q}
%EndExpansion
_{p}$. T h at is,$$\ bolds ymbo l{%
%TCIM ACR O{\U{2 11a } }%
%B eginEx pansi on
\m a t hbb{Q } %E ndE xp ansion
}_{ p } =\l eft\{ x =\su m_{n=-k }^{\infty }a_{n}p^ { n}: 0\leq a_{n}\l eqp-1\ r i gh t\}
\ t ext {. } $$T hen $%
%TCIMACR O{\U{2124} } %
% Begin | - '**Mehmet_Acikgoz**'
- '**Hassan_Jolany**'
- '**Yuan He**'
title: '**Identities_involving** $q$**-Genocchi_numbers_and polynomials**'
---
**PRELIMINARIES**
=================
Recently,_*Kim*_and *Lee* have_given some properties_for the $%
q $-Euler_numbers and polynomials_in_[@Kim]. Actually, this type numbers and polynomials and their $q$-extensions or variously generalizations have_been_studied in_several_different_ways for a long time_(for details, see \[1-21\]).
By using_$p$-adic $q$-integral_on $%
%TCIMACRO{\U{2124} }%
%BeginExpansion
\mathbb{Z}
%EndExpansion
_{p}$, *Kim* defined many new generating_functions_of the $q$-Bernoulli_poynomials, $q$-Euler polynomials and $q$-Genocchi polynomials in his arithmetic_works (for details, see \[6-19\]). The_works of *Kim*_have_been_benefited for further works_of many mathematicians in Analytic numbers_theory. Actually, we motivated from his_inspiring works to write this paper.
Assume that_$p$ be a fixed odd prime_number. Throughout this work, we_require the_definitions of the some notations_such that let_$%
%TCIMACRO{\U{211a} }%
%BeginExpansion
\mathbb{Q}
%EndExpansion
_{p}$_be the field_$p$-adic rational numbers and let $%
%TCIMACRO{\U{2102}_}%
%BeginExpansion
\mathbb{C}
%EndExpansion
_{p}$ be the_completion of algebraic closure of $%
%TCIMACRO{\U{211a}_}%
%BeginExpansion
\mathbb{Q}
%EndExpansion
_{p}$._That is, $$\boldsymbol{%
%TCIMACRO{\U{211a}_}%
%BeginExpansion
\mathbb{Q}
%EndExpansion
}_{p}=\left\{_x=\sum_{n=-k}^{\infty_}a_{n}p^{n}:0\leq a_{n}\leq_p-1\right\}
\text{.}$$
Then $%
%TCIMACRO{\U{2124}_}%
%Begin |
-------------------------------
Higher bands are generally unstable with respect to binary collisions [@Isa:05]. For example, one of the collision partners may be transferred to a lower lying band while the other one is excited to a higher lying band. For this to be possible, energy momentum conservation requires the existence of target states at similar energy distances above and below the initially populated band. Hence, an appropriate design of the band gap structure allows one to control the collisional relaxation rates of higher bands to some extent [@Sto:08]. In the 2D lattice, discussed here, both collision partners may also be scattered to a lower lying band with the excess energy transferred to the dimension orthogonal to the lattice plane [@Bue:11; @Pau:13]. Typical band lifetimes turn out to be on the ms time scale, which is often shorter than intra-band relaxation times. At sufficiently high densities, the discussion of band relaxation dynamics may be complicated by the possibility of dynamical instabilities [@Wu:01; @Wu:03; @Fal:04]. An important increase of the band lifetimes can arise if, in addition to energy and momentum conservation constraints, a spatial mismatch between the external wave functions of the initial states and the possible target states in other bands can be engineered. This is the case in bipartite lattices comprising deep and shallow potential wells. Such lattices give rise to hybrid bands composed of $s$-orbitals in the shallow wells and, for example, $p$-orbitals in the deep wells. For an appropriate adjustment of the relative well depth $\Delta V$, the local $p$-orbitals possess nearly vanishing stationary populations. Hence, notable collision rates only arise in the $s$-orbitals of the shallow wells. These have practically zero overlap with the ground state wave function in the lowest band, which is composed of local $s$-orbitals in the deep wells. This is illustrated in Fig. \[Fig.4\], which compares the case of a monopartite lattice in (a) with that of a bipartite lattice in (b). In (c) a measurement of the band lifetime versus $\theta$ in Eq. \[Eq.1\] for $V_{xy,0} = 7\,E_{\textrm{rec}}$ is shown. A pronounced maximum is found around $\theta = 0.54\,\pi$, which corresponds to the case, where the population of the deep wells becomes minimal. A gain of band lifetime of nearly two orders of magnitude over the $\theta = 0.5\,\pi$ | -------------------------------
Higher bands are generally unstable with obedience to binary collision [ @Isa:05 ]. For example, one of the collision partners may be transferred to a low lying band while the early one is excited to a higher lying set. For this to be possible, energy momentum conservation want the existence of target state at exchangeable energy distances above and below the initially populated band. Hence, an appropriate design of the set gap structure allows one to master the collisional relaxation rates of higher bands to some extent [ @Sto:08 ]. In the 2D lattice, discourse here, both collision partners may also be scattered to a lower lying isthmus with the excess energy transferred to the dimension orthogonal to the lattice plane [ @Bue:11; @Pau:13 ]. Typical band lifetimes turn out to be on the ms time scale, which is often shorter than intra - band relaxation times. At sufficiently high density, the discussion of band relaxation moral force may be complicate by the hypothesis of dynamical instabilities [ @Wu:01; @Wu:03; @Fal:04 ]. An important increase of the band lifetimes can arise if, in addition to department of energy and momentum conservation constraints, a spatial mismatch between the external wave functions of the initial states and the possible target states in other bands can be engineered. This is the case in bipartite lattices comprising cryptic and shallow potential well. Such lattices give rise to hybrid bands composed of $ s$-orbitals in the shallow well and, for example, $ p$-orbitals in the deep wells. For an appropriate adjustment of the relative well depth $ \Delta V$, the local $ p$-orbitals possess nearly vanishing stationary population. Hence, celebrated collision rates only originate in the $ s$-orbitals of the shallow wells. These suffer practically zero lap with the ground state wave affair in the lowest band, which is compose of local $ s$-orbitals in the deep well. This is illustrated in Fig. \[Fig.4\ ], which compare the case of a monopartite lattice in (a) with that of a bipartite lattice in (b). In (degree centigrade) a measurement of the band lifetime versus $ \theta$ in Eq. \[Eq.1\ ] for $ V_{xy,0 } = 7\,E_{\textrm{rec}}$ is show. A pronounced maximum is found around $ \theta = 0.54\,\pi$, which corresponds to the case, where the population of the abstruse wells becomes minimal. A gain of band lifetime of nearly two order of order of magnitude over the $ \theta = 0.5\,\pi$ | -------------------------------
Higjer bands are generally mnstable with respect tm binady collirions [@Isa:05]. For example, one of tye coolision partners may bd transfegred to a lowtr lying band while the obker ohc is zxrited to a highgr lying bang. For this to te plssible, energy momentum conservatiog requited the existencg of uardet anaues at similar energy distances agove anv below the iniyially populated band. Hencf, an appropriate desigj of the babd gwp structure xllows one to control jhe collisional relaxation rates of hngher bands ti slke extent [@Svo:08]. In nhe 2D lattice, discussad here, both collisiok parvnerw may also be scatterxd to a lower lying fand with tke excess energy tranwfwrred to dhe aumevsikn odthogojal to the laftice plane [@Bue:11; @Pau:13]. Typical bamd oifetimes turh out eo be on the ms time scale, which is often shkrter than intra-band reoaxation times. At sufviciently high densities, the discussion of band relaxation dynajkcs mwh bf complicated by the possibility of dynamical jnxtsbilities [@Wu:01; @Wm:03; @Fal:04]. An important ijctgase of the bavd lifzfijes can arise if, ij addityon ti energy wnd komentum conservation consteaints, a spanial mismatch between che external wavg funcyions of the initial stctes ahd the posslble targsg states in othef bsngs can bt engineered. This ys the cawe iu bipartkte katticqs comprislng dcap and shallow potfntiap fells. Such lattices give rise to hybrid beids composed pf $s$-ogbitals iu the xhallow wells and, for exampke, $p$-orbntals kn the deei wells. Fmr an appro[riate adjustkgnt of the renative wqll eepty $\Delta X$, the local $p$-otbitals possess neaely vanishing statlonari lopulations. Henet, nitable collisiom rxtef lnky drise in the $s$-orcitxks of the shalloc dellx. These have practicdlly zero overlap with tme ground state wwve function on the lowest band, whici is cumposrd jf local $s$-orbitals in the deep wells. Thls ls illustrateq in Fig. \[Fig.4\], whick compares the case of a monopartite latvice in (a) with that of q bipartite lattice ln (b). In (c) a keasuwement of the band lifetime vwrsus $\theta$ in Eq. \[Tq.1\] for $V_{xy,0} = 7\,E_{\textrm{rec}}$ js shofn. A oronounced maximum is found around $\theta = 0.54\,\pi$, which corresponds to the casw, wherx ehe populatjon pf thv dzep wells becloes minimal. A gapn of band lifetime of nearly twu orders oy magnitude over the $\theta = 0.5\,\po$ | ------------------------------- Higher bands are generally unstable with binary [@Isa:05]. For one of the to lower lying band the other one excited to a higher lying band. this to be possible, energy momentum conservation requires the existence of target states similar energy distances above and below the initially populated band. Hence, an appropriate of band structure one to control the collisional relaxation rates of higher bands to some extent [@Sto:08]. In the lattice, discussed here, both collision partners may also scattered to a lower band with the excess energy to dimension orthogonal the plane @Pau:13]. Typical band turn out to be on the ms time scale, which is often shorter than intra-band relaxation times. sufficiently high discussion of relaxation may complicated by the dynamical instabilities [@Wu:01; @Wu:03; @Fal:04]. An the band lifetimes can arise if, in addition energy and conservation constraints, a spatial mismatch between external wave functions of the initial states and possible target states in other bands can be engineered. This is the case in bipartite deep and shallow potential Such lattices give to bands of in the wells and, for example, $p$-orbitals in the deep wells. For an adjustment of the relative well depth $\Delta V$, the local nearly stationary populations. Hence, collision rates only arise the of the shallow wells. practically overlap state function the lowest band, which composed of local $s$-orbitals in deep wells. This is compares the case of a monopartite lattice in with that of a bipartite lattice in In (c) a measurement of the band lifetime versus $\theta$ in Eq. for $V_{xy,0} is shown. A pronounced maximum is found around = 0.54\,\pi$, which corresponds the case, where the population of the deep wells minimal. gain of lifetime of nearly orders of magnitude the $\theta = | -------------------------------
Higher bands are generally unStable with RespeCt tO biNaRy coLlisIons [@Isa:05]. For examPLe, onE of the collision partnerS may bE tRAnsfERrEd to a Lower lyINg BANd wHiLe The OtHEr One is ExcIted to a Higher lyinG baNd. for this to be pOSsIble, energy MomEntum conservAtiOn requIrEs tHE exisTenCe of tArget sTAtes at Similar enErGY distaNCes abovE ANd BeloW the initially popuLAtED band. Hence, an apPropriAtE DeSIGn oF thE band gap stRuCture ALlows onE To CONTroL The collisionaL relaxation RAteS of higHeR baNDs to soMe extEnT [@sto:08]. in the 2D lattiCe, diScussed heRe, both COllisioN PartnerS may alSo bE scAtteREd To A loWeR LyiNG bAnd WIth The excesS eNeRgy trAnsfERRED to tHe dImenSion oRthogonal to thE laTticE PlaNe [@Bue:11; @pau:13]. TyPicaL bAnd liFetimeS turn OuT to be on the ms timE scaLe, which is OftEn ShoRtEr thaN Intra-bAnd RelAxation Times. At SUffIcIENTlY high densities, the dIsCUSsIon of banD relaxATiOn DYnamics mAy Be cOmplICAted bY the POsSibility Of dynaMIcAl InstabiLiTies [@Wu:01; @wu:03; @fal:04]. an iMportANt inCrease Of the banD lifeTImes can arise if, IN addition to enERgY ANd MOmenTum ConservatioN conSTraiNts, a SPaTiaL MismaTch beTwEEn THe external wave functIoNs of thE initIal states and tHe possible TARGet stateS in oTHeR Bands can be engiNeereD. This is the CAse in bipArtitE latticeS comprisiNG Deep and sHalLow PotEntIAL wElls. Such lattiCES givE rIse to hyBriD bands cOmpOseD of $S$-orBiTals in the Shallow wElLs AnD, fOr eXamplE, $P$-orbitalS iN thE dEep Wells. fOr an apProprIate AdJuSTmeNt of the RElATIve wElL dEpth $\delTa v$, the lOcal $P$-OrbItals poSsess nearLy vANishInG sTationaRy populations. heNce, notable CoLliSion raTES only ariSe in the $s$-orbitals of the shALlow welLs. THese hAve pRacticallY zeRo overLap WIth the Ground State WaVe fUNCtion IN ThE loWeSt band, whicH IS coMposeD oF locAl $s$-orbiTals in the deep wells. tHis Is illustrated In FIg. \[FiG.4\], WHiCh cOMpARes ThE CasE OF a monopartite laTtice in (a) wiTh THaT of a bipartITe lAtTice in (b). in (c) a meaSuremENt of the Band lifetIme versus $\ThEta$ iN eQ. \[Eq.1\] For $V_{xy,0} = 7\,E_{\texTrm{rec}}$ is Shown. A proNOunceD MaXimum Is fOund arOuNd $\tHeta = 0.54\,\pI$, which COrrEsponDs to thE cAse, wheRe the PoPulation Of the deep wells becomes miNimal. A Gain oF baNd lifetimE of NEarLy two ordeRs of Magnitude oVer The $\Theta = 0.5\,\Pi$ | -------------------------- -----
Hig her b and s a re gen eral ly unstable wi t h re spect to binary collis ions[@ I sa:0 5 ]. Forexample , o n e of t he co ll i si on pa rtn ers may be transf err ed to a lowerl yi ng band wh ile the other o neis exc it edt o a h igh er ly ing ba n d. For this tobe possib l e, ener g y m omen tum conservationr eq u ires the exist ence o ft ar g e t s tat es at simi la r ene r gy dist a nc e s abo v e and below t he initiall y po pulate dban d . Henc e, an a p pro priate desi gn o f the ban d gaps tructur e allows one t o c ont rolt he c oll is i ona l r ela x ati on rates o fhighe r ba n d s to s ome ext ent [ @Sto:08]. Inthe 2Dl att ice,discu ssed h ere,both c ollis io n partners mayalso be scatt ere dtoalower lyingban d w ith the excess ene rg y t ra nsferred to the di me n s io n orthog onal t o t he latticepl ane [@B u e :11;@Pau : 13 ]. Typic al ban d l if etimestu rn out t o b e o n the ms t ime sc ale, whi ch is often shortert han intra-ban d r e l ax a tion ti mes. At suf fici e ntly hig h d ens i ties, thedi s cu s sion of band relaxa ti on dyn amics may be compl icated byt h e possibi lity of dynamical inst abili ties [@Wu: 0 1; @Wu:0 3; @F al:04].An import a n t increa seofthe ba n d l ifetimes cana r iseif , in ad dit ion toene rgy an d m om entum con servatio nco ns tr ain ts, a spatialmi sma tc h b etwee n the e xtern al w av ef unc tions o f t h e ini ti al sta tes a nd th e po s sib le targ et states in othe rba nds can be engineere d. This is t he ca se inb i partitelattices comprising dee p and sh all ow po tent ial wells . S uch la tti c es giv e rise to h yb rid b andsc o mp ose dof $s$-orb i t als in t he sha llow we lls and, for examp l e,$p$-orbitalsinthed e ep we l ls . Fo ra n a p p ropriate adjust ment of th er el ative well dep th $\Delt a V$, t he lo c al $p$- orbitalspossess n ea rlyv a nis hing stati onary po pulations . Henc e ,notab lecollis io n r atesonly a r ise in t he $s$ -o rbital s ofth e shallo w wells. These have pra ctical ly ze rooverlap w ith the ground s tate wave func tio n i n the lo w est b and, wh ich is co mpos e d of loca l $ s$- o r bi tals in the d e epwells . T h is isillu strated in Fig. \ [ Fig.4\], which com p a res th e cas eof a monoparti tela t t ice in ( a) with thatof a bip ar t ite l attice in (b ). In ( c ) a measur emen t o f the ban d l if e time ve rs us $\thet a$ i nEq. \[ Eq.1\] for$ V _{xy,0} = 7\,E_{ \text r m {rec} } $ i s sho wn . A pro n ounc ed maximum is found a round$\th eta = 0.54\, \p i$, wh ich c orresponds to the ca se, w here th epopu lat ion of the d eep w ells b eco mes minim a l .A g ai n of ban d lif et imeof nearly two orde rso f magni tu deo v er the $\ t h eta = 0.5\ ,\p i$ | -------------------------------
Higher bands_are generally_unstable with respect to_binary collisions_[@Isa:05]._For example,_one_of the collision_partners may be_transferred to a lower_lying band while_the_other one is excited to a higher lying band. For this to be possible,_energy_momentum conservation_requires_the_existence of target states at_similar energy distances above and_below the_initially populated band. Hence, an appropriate design of_the_band gap structure_allows one to control the collisional relaxation rates of_higher bands to some extent [@Sto:08]._In the 2D_lattice,_discussed_here, both collision partners_may also be scattered to a_lower lying band with the excess_energy transferred to the dimension orthogonal to_the lattice plane [@Bue:11; @Pau:13]. Typical_band lifetimes turn out to_be on_the ms time scale, which_is often shorter_than intra-band_relaxation times. At_sufficiently high densities, the discussion of_band relaxation dynamics_may be complicated by the possibility_of_dynamical instabilities [@Wu:01;_@Wu:03;_@Fal:04]._An important_increase of the_band_lifetimes can_arise_if, in addition to energy and_momentum_conservation constraints, a spatial mismatch between the_external wave functions of_the_initial states and the_possible target states in other_bands can be engineered. This is_the case_in bipartite_lattices comprising deep and shallow potential wells. Such lattices give rise_to hybrid bands composed of $s$-orbitals_in the shallow wells_and, for_example,_$p$-orbitals in the_deep_wells. For_an appropriate adjustment of the relative well_depth $\Delta_V$, the local $p$-orbitals possess nearly_vanishing stationary populations. Hence,_notable_collision rates only arise in the_$s$-orbitals of the shallow wells. These_have practically zero overlap with_the_ground_state wave function in the_lowest band, which is composed of_local $s$-orbitals in_the deep wells. This is illustrated in_Fig. \[Fig.4\],_which compares the case of a_monopartite_lattice in (a) with that of_a_bipartite_lattice in (b). In (c)_a measurement of the band lifetime_versus $\theta$ in Eq. \[Eq.1\] for $V_{xy,0} = 7\,E_{\textrm{rec}}$ is_shown. A pronounced_maximum is found around $\theta_=_0.54\,\pi$,_which corresponds to the case, where the population of the_deep wells_becomes minimal. A_gain of band lifetime of nearly two orders of magnitude_over the $\theta = 0.5\,\pi$ |
in exactly the same way as a disc simulated by [@Lodato_Rice_original]: a 1 $\rm{M_\odot}$ star with a 0.1 $\rm{M_\odot}$ disc made of 250,000 SPH particles, spanning $0.25 \le R \le 25$AU. The initial surface mass density and temperature profiles of the disc are $\Sigma \propto R^{-1}$ and $T \propto R^{-\frac{1}{2}}$, respectively. The magnitudes of these are set such that the Toomre stability parameter (equation \[eq:Toomre\]) at the outer edge of the disc, $Q_{\rm min} = 2$. This gives an aspect ratio, $H/R \sim 0.05$. We model the $1~\rm{M_\odot}$ star in the centre of the disc using a sink particle [@Bate_Bonnell_Price_sink_ptcls]. At the inner disc boundary, particles are accreted onto the star if they move within a radius of 0.025 AU of the star or if they move into $0.025 \le R < 0.25 \rm{AU}$ and are gravitationally bound to the star. At the outer edge, the disc is free to expand.
The initial Reference disc is a Toomre stable disc. Given that the boundary temperature profile is the same as that of the initial disc and that $Q_{\rm min}=~2$, we do not expect this disc to fragment. However, we use the Reference disc as a fiducial case. In particular, we are concerned with the cooling rates that are present in the disc once it is in an equilibrium state. We emphasize our use of terminology here: when referring to the disc being in *thermal equilibrium with the boundary*, we are describing the bulk of the disc being a similar temperature to the disc boundary (which is assumed to be determined by stellar irradiation), whereas an *equilibrium state* disc refers to the dissipative and cooling rates being balanced such that the Toomre stability profiles do not change with time.
Exploring the parameter space {#sec:parameter}
-----------------------------
Given that a motivation of this work is to determine if, and under what circumstances, fragmentation in realistically modelled self-gravitating discs may occur, we explore the parameter space in a number of ways. One parameter is the opacity: we re-run the Reference simulation with opacity values scaled to $ | in exactly the same way as a disc simulate by [ @Lodato_Rice_original ]: a 1 $ \rm{M_\odot}$ asterisk with a 0.1 $ \rm{M_\odot}$ disc made of 250,000 SPH particles, cross $ 0.25 \le R \le 25$AU. The initial surface mass density and temperature profile of the disc are $ \Sigma \propto R^{-1}$ and $ T \propto R^{-\frac{1}{2}}$, respectively. The magnitude of these are set such that the Toomre stability argument (equation \[eq: Toomre\ ]) at the outer edge of the disc, $ Q_{\rm minute } = 2$. This gives an aspect proportion, $ H / R \sim 0.05$. We model the $ 1~\rm{M_\odot}$ star in the center of the phonograph record using a sink particle [ @Bate_Bonnell_Price_sink_ptcls ]. At the inside disc boundary, atom are accreted onto the ace if they move within a radius of 0.025 AU of the star or if they move into $ 0.025 \le R < 0.25 \rm{AU}$ and are gravitationally bound to the asterisk. At the outer edge, the disc is free to expand.
The initial Reference disc is a Toomre stable disc. Given that the boundary temperature profile is the same as that of the initial disc and that $ Q_{\rm min}=~2 $, we do not expect this disc to fragment. However, we use the Reference disc as a fiduciary event. In particular, we are implicated with the cool rate that are present in the disc once it is in an equilibrium department of state. We emphasize our use of terminology here: when referring to the disc being in * thermal chemical equilibrium with the boundary *, we are describing the bulk of the disc being a similar temperature to the disc boundary (which is assume to be determined by stellar irradiation), whereas an * equilibrium state * disc consult to the dissipative and cooling rates being balanced such that the Toomre stability profiles do not transfer with time.
Exploring the parameter quad { # sec: parameter }
-----------------------------
Given that a motivation of this work is to determine if, and under what circumstance, fragmentation in realistically model self - gravitating discs may happen, we explore the parameter space in a number of way. One parameter is the opacity: we re - ply the Reference simulation with opacity values scale to $ | in exactly the same way as a disc simulatgd by [@Lovato_Rics_originau]: a 1 $\rm{M_\odot}$ star with a 0.1 $\rm{M_\ovot}$ eisc nade of 250,000 SPH particles, spanning $0.25 \le R \lw 25$AU. Rhe initial surfacc masa denvmty and temperajure profilev of the disc drd $\Digma \propto R^{-1}$ and $T \propto R^{-\frac{1}{2}}$, rqspectifepy. The magnituqes ps thsse are set such that the Toomre sfabilitj parameter (equatoon \[eq:Toomre\]) at the outer efge lf the disc, $Q_{\rm mij} = 2$. This gices wb aspect ratko, $H/R \sim 0.05$. We model thg $1~\rm{M_\odot}$ star in the centre of ghe dnsc using a sunk karticle [@Batx_Bonnejl_Price_sink_pbbls]. At dhe innrr disc boundavy, pacticoes are accreted onto the star if they iove withhn a radius of 0.025 AU od rhe sjar os if rheh mkvx ihto $0.025 \lf R < 0.25 \rm{AU}$ ans are gravirationally bound to trv star. At the outer eqge, the disc is free to expand.
The initidl Deference disc is a Toonre stable disc. Given that the boundary temperature profile is the same as that of tie innblal ausf and that $Q_{\rm min}=~2$, we do not expect this disc fo fgagment. However, wc use the Referencr fixs as a fiducixl casz. Ih particular, we arf concetned wuth the cjolimg rates that are present ib the disc ouce it is in an equilnbrium state. We gmphasoze our use of terminolugy gere: when rfferring fu the disc being in *tvermal equilibrium with thq boundarb*, we cre descfibimg the bulk of tje disc being a similar hempetature to the didc boundary (which is assumed to be determined bf snellar irxadiatlon), whereas an *qquilibrium stcte* disc refefs to the sissipavive and coojing rates behjg balanced vuch thae thw Toimre stxcility profilex do not bhcnge with time.
Exploring the patajeter space {#sec:'crqmeter}
-----------------------------
Given thay a moeinatmon os this work iv to degrrmind if, and unbef whst circumstances, fracmenfation in realistivajly modeoled sels-gravitating ciscs may occur, we explmre the larwmeter space in a number of wags. One pagamcter is the o[aciby: wg re-run the Reference simulation with opacity valuxs scaled to $ | in exactly the same way as a by a 1 star with a 250,000 particles, spanning $0.25 R \le 25$AU. initial surface mass density and temperature of the disc are $\Sigma \propto R^{-1}$ and $T \propto R^{-\frac{1}{2}}$, respectively. The of these are set such that the Toomre stability parameter (equation \[eq:Toomre\]) at outer of disc, min} = 2$. This gives an aspect ratio, $H/R \sim 0.05$. We model the $1~\rm{M_\odot}$ star the centre of the disc using a sink [@Bate_Bonnell_Price_sink_ptcls]. At the inner boundary, particles are accreted onto star they move a of AU of the or if they move into $0.025 \le R < 0.25 \rm{AU}$ and are gravitationally bound to the At the the disc free expand. initial Reference disc Toomre stable disc. Given that the is the same as that of the initial and that min}=~2$, we do not expect this to fragment. However, we use the Reference disc a fiducial case. In particular, we are concerned with the cooling rates that are present disc once it is an equilibrium state. emphasize use terminology when referring the disc being in *thermal equilibrium with the boundary*, we are the bulk of the disc being a similar temperature to boundary is assumed to determined by stellar irradiation), an state* disc refers to and rates that Toomre profiles do not change time. Exploring the parameter space ----------------------------- Given that a to determine if, and under what circumstances, fragmentation realistically modelled self-gravitating discs may occur, we the parameter space in a number of ways. One parameter is the we re-run simulation with opacity values scaled to $ | in exactly the same way as a disC simulated By [@LodAto_ricE_oRigiNal]: a 1 $\Rm{M_\odot}$ star witH A 0.1 $\rm{M_\Odot}$ disc made of 250,000 SPH partiCles, sPaNNing $0.25 \LE R \Le 25$AU. THe initiAL sURFacE mAsS deNsITy And teMpeRature pRofiles of tHe dIsC are $\Sigma \proPTo r^{-1}$ and $T \proptO R^{-\fRac{1}{2}}$, respectivEly. the magNiTudES of thEse Are seT such tHAt the TOomre stabIlITy paraMEter (equATIoN \[eq:TOomre\]) at the outer edGE oF The disc, $Q_{\rm min} = 2$. THis givEs AN aSPEct RatIo, $H/R \sim 0.05$. We mOdEl the $1~\RM{M_\odot}$ sTAr IN THe cENtre of the disc Using a sink pARtiCle [@BatE_BOnnELl_PricE_sink_PtCLs]. AT the inner diSc boUndary, parTicles ARe accreTEd onto tHe star If tHey Move WItHiN a rAdIUs oF 0.025 aU Of tHE stAr or if thEy MoVe intO $0.025 \le R < 0.25 \RM{au}$ And aRe gRaviTatioNally bound to tHe sTar. AT The Outer Edge, tHe diSc Is freE to expAnd.
ThE iNitial Reference Disc Is a Toomre StaBlE diSc. given THat the BouNdaRy tempeRature pROfiLe IS THe Same as that of the iniTiAL DiSc and thaT $Q_{\rm miN}=~2$, We Do NOt expect ThIs dIsc tO FRagmeNt. HoWEvEr, we use tHe RefeREnCe Disc as a FiDucial CaSe. IN paRticuLAr, we Are conCerned wiTh the COoling rates thaT Are present in tHE dISC oNCe it Is iN an equilibrIum sTAte. WE empHAsIze OUr use Of terMiNOlOGy here: when referring To The disC beinG in *thermal equIlibrium wiTH THe boundaRy*, we ARe DEscribing the buLk of tHe disc beinG A similar TempeRature to The disc boUNDary (whicH is AssUmeD to BE DeTermined by steLLAr irRaDiation), WheReas an *eQuiLibRiuM stAtE* disc refeRs to the dIsSiPaTiVe aNd cooLIng rates BeIng BaLanCed suCH that tHe TooMre sTaBiLIty ProfileS Do NOT chaNgE wIth tIme.
exPloriNg thE ParAmeter sPace {#sec:paRamETer}
-----------------------------
GIvEn That a moTivation of thiS wOrk is to detErMinE if, and UNDer what cIrcumstances, fragmentatiON in realIstIcallY modElled self-GraVitatiNg dIScs may Occur, wE explOrE thE PArameTER sPacE iN a number of WAYs. ONe parAmEter Is the opAcity: we re-run the RefERenCe simulation wIth OpacITY vAluES sCAleD tO $ | in exactly the same way a s a disc s imula ted by [ @Lod ato_ Rice_original] : a 1 $\rm{M_\odot}$ star w ith a 0 . 1 $\ r m{ M_\od ot}$ di s cm a deof 2 50, 00 0 S PH pa rti cles, s panning $0 .25 \ le R \le 25$ A U. The initi alsurface mass de nsityan d t e mpera tur e pro fileso f thedisc are$\ S igma \ p ropto R ^ { -1 }$ a nd $T \propto R^{ - \f r ac{1}{2}}$, re specti ve l y. T hemag nitudes of t hesea re sets uc h t hat the Toomre st ability par a met er (eq ua tio n \[eq: Toomr e\ ] ) a t the outer edg e of thedisc,$ Q_{\rmm in} = 2 $. Thi s g ive s an as pe ctra t io, $H /R\ sim 0.05$.We m odelthe$ 1 ~ \ rm{M _\o dot} $ sta r in the cent reof t h e d isc u singa si nk part icle [ @Bate _B onnell_Price_si nk_p tcls]. At th einn er disc bounda ry, pa rticles are ac c ret ed o n to the star if theymo v e w ithin aradius of 0 . 025 AU o fthe sta r or if the y m ove into $0.02 5 \ le R < 0. 25 \rm{A U} $ a ndare g r avit ationa lly boun d tot he star. At th e outer edge,t he d is c isfre e to expand .
T h e in itia l R efe r encediscis aT oomre stable disc.Gi ven th at th e boundary te mperaturep r o file isthes am e as that of th e ini tial disca nd that$Q_{\ rm min}= ~2$, we d o not expe ctthi s d isc t ofragment. How e v er,we use th e R eferenc e d isc as afi ducial ca se. In p ar ti cu la r,we ar e concern ed wi th th e coo l ing ra tes t hatar ep res ent int he d iscon ce itisin an e quil i bri um stat e. We emp has i ze o ur u se of t erminology he re : when ref er rin g to t h e disc be ing in *thermal equilib r ium wit h t he bo unda ry*, we a redescri bin g the b ulk of thedi scb e ing a s im ila rtemperatur e tothe d is c bo undary(which is assumedt o b e determinedbystel l a rirr a di a tio n) , wh e r eas an *equilib rium state *d is c refers t o th edissipa tive an d coo l ing rat es beingbalancedsu ch t h a t t he Toomrestabilit y profile s do n o tchang e w ith ti me .
Explo ring t h e p arame ter sp ac e {#se c:par am eter}
-- ----------------------- ----
Given th at a moti vat i onof this w orkis to dete rmi neif, a ndu nderwhat ci rcu m stanc es,f ragmentat i on in r ea listicallym o d ell ed se lf- g ravita ting discs may occur, we explore the par a m ete r s p acein a number of w ays .O n e parame te r is the op acity: w er e-run the R eferen ce simu l a ti o n with opa cit y valuessca le d to $ | in_exactly the_same way as a_disc simulated_by_[@Lodato_Rice_original]: a_1 $\rm{M_\odot}$_star with a_0.1 $\rm{M_\odot}$ disc made_of 250,000 SPH particles,_spanning $0.25 \le_R_\le 25$AU. The initial surface mass density and temperature profiles of the disc are_$\Sigma_\propto R^{-1}$_and_$T_\propto R^{-\frac{1}{2}}$, respectively. The magnitudes_of these are set such_that the_Toomre stability parameter (equation \[eq:Toomre\]) at the outer edge_of_the disc, $Q_{\rm_min} = 2$. This gives an aspect ratio, $H/R_\sim 0.05$. We model the $1~\rm{M_\odot}$_star in the_centre_of_the disc using a_sink particle [@Bate_Bonnell_Price_sink_ptcls]. At the inner_disc boundary, particles are accreted onto_the star if they move within a_radius of 0.025 AU of the star_or if they move into_$0.025 \le_R < 0.25 \rm{AU}$ and_are gravitationally bound_to the_star. At the_outer edge, the disc is free_to expand.
The initial_Reference disc is a Toomre stable_disc._Given that the_boundary_temperature_profile is_the same as_that_of the_initial_disc and that $Q_{\rm min}=~2$, we_do_not expect this disc to fragment. However,_we use the Reference_disc_as a fiducial case._In particular, we are concerned_with the cooling rates that are_present in_the disc_once it is in an equilibrium state. We emphasize our use_of terminology here: when referring to_the disc being in_*thermal equilibrium_with_the boundary*, we_are_describing the_bulk of the disc being a similar_temperature to_the disc boundary (which is assumed_to be determined by_stellar_irradiation), whereas an *equilibrium state* disc_refers to the dissipative and cooling_rates being balanced such that_the_Toomre_stability profiles do not change_with time.
Exploring the parameter space {#sec:parameter}
-----------------------------
Given_that a motivation_of this work is to determine if,_and_under what circumstances, fragmentation in realistically_modelled_self-gravitating discs may occur, we explore_the_parameter_space in a number of_ways. One parameter is the opacity:_we re-run the Reference simulation with opacity values scaled_to $ |
nx+1}{ncx+n}}.\label{eq:15}$$
In particular, for $c=0$ and $n=1$, $$\log{x} \leq \sum _{k=0}^\infty e^{-x}\frac{x^k}{k!}\log{(k+1)}\leq \log{(x+1)}.$$
**Proof** We have seen that $H'_{n,c}(x)\geq 0$. An application of yields $$H'_{n,c}(x) = n \left ( \log{\frac{1+cx}{x}} + \sum _{k=0}^\infty p_{n+c,k}^{[c]}(x)\log{\frac{k+1}{n+ck}} \right ).$$
This proves the first inequality in ; the second is a consequence of Jensen’s inequality applied to the concave function $\log{t}$.
Rényi entropy and Tsallis entropy
=================================
The following conjecture was formulated in [@13]:
\[conj:3.1\] $S_{n,-1}$ is convex on $[0,1]$.
Th. Neuschel [@11] proved that $S_{n,-1}$ is decreasing on $\left [ 0, \frac{1}{2}\right ]$ and increasing on $\left [ \frac{1}{2}, 1\right ]$. The conjecture and the result of Neuschel can be found also in [@5].
A proof of the conjecture was given by G. Nikolov [@12], who related it with some new inequalities involving Legendre polynomials. Another proof can be found in [@4].
Using the important results of Elena Berdysheva [@3], the following extension was obtained in [@17]:
\[th:3.2\] ([@17 Theorem 9]). For $c<0$, $S_{n,c}$ is convex on $\left [ 0, -\frac{1}{c}\right ]$.
A stronger conjecture was formulated in [@14] and [@17]:
\[conj:3.3\] For $c\in \mathbb{R}$, $S_{n,c}$ is logarithmically convex, i.e., $\log S_{n,c}$ is convex.
It was validated for $c\geq 0$ by U. Abel, | nx+1}{ncx+n}}.\label{eq:15}$$
In particular, for $ c=0 $ and $ n=1 $, $ $ \log{x } \leq \sum _ { k=0}^\infty e^{-x}\frac{x^k}{k!}\log{(k+1)}\leq \log{(x+1)}.$$
* * Proof * * We have seen that $ H'_{n, c}(x)\geq 0$. An application of yield $ $ H'_{n, c}(x) = n \left (\log{\frac{1+cx}{x } } + \sum _ { k=0}^\infty p_{n+c, k}^{[c]}(x)\log{\frac{k+1}{n+ck } } \right) .$$
This prove the first inequality in ; the second is a consequence of Jensen ’s inequality apply to the concave routine $ \log{t}$.
Rényi entropy and Tsallis entropy
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
The surveil conjecture was formulated in [ @13 ]:
\[conj:3.1\ ] $ S_{n,-1}$ is convex on $ [ 0,1]$.
Th. Neuschel [ @11 ] prove that $ S_{n,-1}$ is decreasing on $ \left [ 0, \frac{1}{2}\right ] $ and increasing on $ \left [ \frac{1}{2 }, 1\right ] $. The guess and the result of Neuschel can be found besides in [ @5 ].
A proof of the conjecture was given by G. Nikolov [ @12 ], who related it with some new inequalities involving Legendre polynomial. Another proof can be found in [ @4 ].
Using the crucial results of Elena Berdysheva [ @3 ], the following propagation was obtained in [ @17 ]:
\[th:3.2\ ] ([ @17 Theorem 9 ]). For $ c<0 $, $ S_{n, c}$ is convex on $ \left [ 0, -\frac{1}{c}\right ] $.
A strong conjecture was formulated in [ @14 ] and [ @17 ]:
\[conj:3.3\ ] For $ c\in \mathbb{R}$, $ S_{n, c}$ is logarithmically convex, i.e., $ \log S_{n, c}$ is convex.
It was validated for $ c\geq 0 $ by U. Abel, | nx+1}{nfx+n}}.\label{eq:15}$$
In particular, nor $c=0$ and $n=1$, $$\log{x} \leq \suk _{k=0}^\inffy e^{-x}\fraz{x^k}{k!}\log{(k+1)}\leq \log{(x+1)}.$$
**Proof** We havx sewn thqt $H'_{n,c}(x)\geq 0$. An applicagion of ypelds $$H'_{n,c}(z) = n \oeft ( \log{\fczc{1+cx}{x}} + \sum _{k=0}^\ihnty p_{u+c,j}^{[c]}(x)\log{\frac{k+1}{n+ck}} \right ).$$
Thiv proves the fhrrt inequality in ; the second is a consqquence ov Jensen’s ineqoalitj wppljvd to the concave function $\log{t}$.
Dényi enuropy and Tsallis rntropy
=================================
The following conjefturf was formulated ij [@13]:
\[conj:3.1\] $S_{n,-1}$ is condwx on $[0,1]$.
Th. Neurchel [@11] provtd that $S_{n,-1}$ ia decreasing on $\left [ 0, \frac{1}{2}\righg ]$ anb increasint in $\pgft [ \frac{1}{2}, 1\rijht ]$. Tre conjecturc and tve resukt of Neuschel cai be found also in [@5].
A proof of the conjecture was givet yy G. Nikolov [@12], who relaree it fith somd ned iheauzlitied iivolving Lefendre polybomials. Another propf bsn be found jn [@4].
Usind ehe important results of Elena Berdyshena [@3], tge following extension qas obtained in [@17]:
\[th:3.2\] ([@17 Thgorem 9]). For $c<0$, $S_{n,c}$ is convex on $\left [ 0, -\frac{1}{c}\right ]$.
A stronger wonjertjre wwr flrmulated in [@14] and [@17]:
\[conj:3.3\] For $c\in \mathbb{R}$, $S_{n,c}$ is ljfatinhmically convex, l.e., $\log S_{n,c}$ is convrx.
Lt ras validated for $c\ysq 0$ by U. Abel, | nx+1}{ncx+n}}.\label{eq:15}$$ In particular, for $c=0$ and $n=1$, \sum e^{-x}\frac{x^k}{k!}\log{(k+1)}\leq \log{(x+1)}.$$ We have seen of $$H'_{n,c}(x) = n ( \log{\frac{1+cx}{x}} + _{k=0}^\infty p_{n+c,k}^{[c]}(x)\log{\frac{k+1}{n+ck}} \right ).$$ This proves first inequality in ; the second is a consequence of Jensen’s inequality applied the concave function $\log{t}$. Rényi entropy and Tsallis entropy ================================= The following conjecture formulated [@13]: $S_{n,-1}$ convex on $[0,1]$. Th. Neuschel [@11] proved that $S_{n,-1}$ is decreasing on $\left [ 0, \frac{1}{2}\right and increasing on $\left [ \frac{1}{2}, 1\right ]$. conjecture and the result Neuschel can be found also [@5]. proof of conjecture given G. Nikolov [@12], related it with some new inequalities involving Legendre polynomials. Another proof can be found in [@4]. Using important results Berdysheva [@3], following was in [@17]: \[th:3.2\] 9]). For $c<0$, $S_{n,c}$ is convex 0, -\frac{1}{c}\right ]$. A stronger conjecture was formulated [@14] and \[conj:3.3\] For $c\in \mathbb{R}$, $S_{n,c}$ is convex, i.e., $\log S_{n,c}$ is convex. It was for $c\geq 0$ by U. Abel, | nx+1}{ncx+n}}.\label{eq:15}$$
In particular, For $c=0$ and $n=1$, $$\loG{x} \leq \Sum _{K=0}^\inFtY e^{-x}\fRac{x^K}{k!}\log{(k+1)}\leq \log{(x+1)}.$$
**PROof** WE have seen that $H'_{n,c}(x)\geq 0$. An AppliCaTIon oF YiElds $$H'_{N,c}(x) = n \lefT ( \LoG{\FRac{1+Cx}{X}} + \sUm _{k=0}^\InFTy P_{n+c,k}^{[c]}(X)\loG{\frac{k+1}{n+Ck}} \right ).$$
ThiS prOvEs the first inEQuAlity in ; the SecOnd is a conseqUenCe of JeNsEn’s INequaLitY applIed to tHE concaVe functioN $\lOG{t}$.
RényI Entropy AND TSallIs entropy
=================================
The folloWInG Conjecture was fOrmulaTeD In [@13]:
\[CONj:3.1\] $S_{N,-1}$ is Convex on $[0,1]$.
Th. neUscheL [@11] Proved tHAt $s_{N,-1}$ IS deCReasing on $\left [ 0, \Frac{1}{2}\right ]$ anD IncReasinG oN $\leFT [ \frac{1}{2}, 1\rIght ]$. THe COnjEcture and thE resUlt of NeusChel caN Be found ALso in [@5].
A pRoof of The ConJectURe WaS giVeN By G. nIkOloV [@12], Who Related iT wItH some New iNEQUAlitIes InvoLving legendre polynOmiAls. ANOthEr proOf can Be foUnD in [@4].
UsIng the ImporTaNt results of ElenA BerDysheva [@3], thE foLlOwiNg ExtenSIon was ObtAinEd in [@17]:
\[th:3.2\] ([@17 THeorem 9]). FOR $c<0$, $S_{N,c}$ IS COnVex on $\left [ 0, -\frac{1}{c}\righT ]$.
A STRoNger conjEcture WAs FoRMulated iN [@14] aNd [@17]:
\[cOnj:3.3\] FOR $C\in \maThbb{r}$, $s_{n,C}$ is logarIthmicALlY cOnvex, i.e., $\LoG S_{n,c}$ is CoNveX.
It Was vaLIdatEd for $c\Geq 0$ by U. AbEl, | nx+1}{ncx+n}}.\label{eq:15 }$$
In pa rticu lar , f or $c= 0$ a nd $n=1$, $$\l o g{x} \leq \sum _{k=0}^\inf ty e^ {- x }\fr a c{ x^k}{ k!}\log { (k + 1 )}\ le q\lo g{ ( x+ 1)}.$ $
**Proof ** We have se en that $H'_{n , c} (x)\geq 0$ . A n applicatio n o f yiel ds $$ H '_{n, c}( x) =n \lef t ( \lo g{\frac{1 +c x }{x}}+ \sum _ { k =0 }^\i nfty p_{n+c,k}^{[ c ]} ( x)\log{\frac{k +1}{n+ ck } }\ r igh t ) .$$
Thispr ovest he firs t i n e q ual i ty in ; the s econd is ac ons equenc eofJ ensen’ s ine qu a lit y applied t o th e concave funct i on $\lo g {t}$.
Rényient rop y an d T sa lli se ntr o py
== = === ======== == == ===== ==== = = = = ==
The fol lowin g conjecturewas for m ula ted i n [@1 3]:
\ [conj :3.1\] $S_{ n, -1}$ is convexon $ [0,1]$.
Th. N eus ch el [@ 1 1] pro ved th at $S_{ n,-1}$i s d ec r e a si ng on $\left [ 0,\f r a c{ 1}{2}\ri ght ]$ an di ncreasin gon$\le f t [ \f rac{ 1 }{ 2}, 1\ri ght ]$ . T he conjec tu re and t heres ult o f Neu schelcan be f ounda lso in [@5].
A proof of the co n j ec t urewas given by G . Ni k olov [@1 2 ], wh o rela ted i tw it h some new inequalit ie s invo lving Legendre pol ynomials.A n o ther pro of c a nb e found in [@4 ].
U sing the i m portantresul ts of El ena Berdy s h eva [@3] , t hefol low i n gextension was o btai ne d in [@ 17] :
\[th :3. 2\] ([ @17 T heorem 9] ). For $ c< 0$ ,$S _{n ,c}$i s convex o n $ \l eft [ 0, -\frac {1}{c }\ri gh t] $.
A stro n ge r conj ec tu re w asfo rmula tedi n [ @14] an d [@17]:
\[ c onj: 3. 3\ ] For $ c\in \mathbb{ R} $, $S_{n,c }$ is logar i t hmically convex, i.e., $\log S_ { n,c}$ i s c onvex .
I t was val ida ted fo r $ c \geq 0 $ by U . Abe l, | nx+1}{ncx+n}}.\label{eq:15}$$
In particular,_for $c=0$_and $n=1$, $$\log{x} \leq_\sum _{k=0}^\infty_e^{-x}\frac{x^k}{k!}\log{(k+1)}\leq_\log{(x+1)}.$$
**Proof** We_have_seen that $H'_{n,c}(x)\geq_0$. An application_of yields $$H'_{n,c}(x) =_n \left (_\log{\frac{1+cx}{x}}_+ \sum _{k=0}^\infty p_{n+c,k}^{[c]}(x)\log{\frac{k+1}{n+ck}} \right ).$$
This proves the first inequality in ; the second is_a_consequence of_Jensen’s_inequality_applied to the concave function_$\log{t}$.
Rényi entropy and Tsallis entropy
=================================
The_following conjecture_was formulated in [@13]:
\[conj:3.1\] $S_{n,-1}$ is convex on $[0,1]$.
Th._Neuschel [@11]_proved that $S_{n,-1}$_is decreasing on $\left [ 0, \frac{1}{2}\right ]$ and_increasing on $\left [ \frac{1}{2}, 1\right_]$. The conjecture_and_the_result of Neuschel can_be found also in [@5].
A proof of_the conjecture was given by G._Nikolov [@12], who related it with some new_inequalities involving Legendre polynomials. Another proof_can be found in [@4].
Using the_important results_of Elena Berdysheva [@3], the following_extension was obtained_in [@17]:
\[th:3.2\] ([@17_Theorem 9]). For_$c<0$, $S_{n,c}$ is convex on $\left_[ 0, -\frac{1}{c}\right_]$.
A stronger conjecture was formulated in [@14]_and [@17]:
\[conj:3.3\]_For $c\in \mathbb{R}$,_$S_{n,c}$_is_logarithmically convex,_i.e., $\log S_{n,c}$_is_convex.
It was_validated_for $c\geq 0$ by U. Abel, |
1+s^2) & s^{-w+n-1} & s^{w-n+7} \\
(1+s^2)(1+s^4) & -s^{-w+n-1}(1+s^2) & -s^{w-n+5}(1+s^2) \\
-s^2(1+s^2) & s^{-w+n+1} &s^{w-n+5}
\end{array}
\right ].$$
This implies $$\begin{aligned}
d(s) h_0(s-s^{-1}) &=& -s^2(1+s^2) \Delta_{\widehat{\beta}}(s^2) + s^{-w+n-1} \epsilon_{w-n+1} + s^{w-n+7} \\
&&- \sum_{j=3}^{n-1} (s^{2k}-s^4-s^2+s^{6-2k})h_j, \\
%
d(s) h_1(s-s^{-1}) &=& (1+s^2)(1+s^4) \Delta_{\widehat{\beta}}(s^2) -s^{-w+n-1}(1+s^2) \epsilon_{w-n+1} \\
&& -s^{w-n+5}(1+s^2) + \sum_{j=3}^{n-1} (1+s^2)(s^{2k}-s^4-1+s^{4-2k})h_j, \\
%
d(s) h_2(s-s^{-1}) &=& -s^2(1+s^2) \Delta_{\widehat{\beta}}(s^2) + s^{-w+n+1} \epsilon_{w-n+1} + s^{w-n+5} \\
&&- \sum_{j=3}^{n-1} (s^{2+2k}-s^4-s^2+s^{4-2k})h_j.\end{aligned}$$
This is exactly the result implied by Thm. \[Hcoeffrel\]. As $z=s - s^{-1}$ is invertible | 1+s^2) & s^{-w+n-1 } & s^{w - n+7 } \\
(1+s^2)(1+s^4) & -s^{-w+n-1}(1+s^2) & -s^{w - n+5}(1+s^2) \\
-s^2(1+s^2) & s^{-w+n+1 } & s^{w - n+5 }
\end{array }
\right ] .$$
This implies $ $ \begin{aligned }
d(s) h_0(s - s^{-1 }) & = & -s^2(1+s^2) \Delta_{\widehat{\beta}}(s^2) + s^{-w+n-1 } \epsilon_{w - n+1 } + s^{w - n+7 } \\
& & - \sum_{j=3}^{n-1 } (s^{2k}-s^4 - s^2+s^{6 - 2k})h_j, \\
%
d(s) h_1(s - s^{-1 }) & = & (1+s^2)(1+s^4) \Delta_{\widehat{\beta}}(s^2) -s^{-w+n-1}(1+s^2) \epsilon_{w - n+1 } \\
& & -s^{w - n+5}(1+s^2) + \sum_{j=3}^{n-1 } (1+s^2)(s^{2k}-s^4 - 1+s^{4 - 2k})h_j, \\
%
d(s) h_2(s - s^{-1 }) & = & -s^2(1+s^2) \Delta_{\widehat{\beta}}(s^2) + s^{-w+n+1 } \epsilon_{w - n+1 } + s^{w - n+5 } \\
& & - \sum_{j=3}^{n-1 } (s^{2 + 2k}-s^4 - s^2+s^{4 - 2k})h_j.\end{aligned}$$
This is exactly the result implied by Thm. \[Hcoeffrel\ ]. As $ z = s - s^{-1}$ is invertible | 1+s^2) & s^{-w+n-1} & s^{w-n+7} \\
(1+s^2)(1+s^4) & -s^{-w+n-1}(1+s^2) & -s^{d-n+5}(1+s^2) \\
-s^2(1+s^2) & s^{-w+n+1} &s^{w-n+5}
\end{array}
\rkght ].$$
This implies $$\begmn{aligned}
d(s) h_0(s-s^{-1}) &=& -s^2(1+s^2) \Delta_{\widehzb{\beta}}(v^2) + s^{-w+n-1} \epsilon_{w-k+1} + s^{w-n+7} \\
&&- \suk_{j=3}^{n-1} (s^{2k}-s^4-s^2+s^{6-2k})h_j, \\
%
d(r) k_1(s-s^{-1}) &=& (1+s^2)(1+s^4) \Delta_{\widehat{\beta}}(s^2) -s^{-w+n-1}(1+s^2) \epsijon_{w-n+1} \\
&& -d^{w-n+5}(1+s^2) + \sum_{j=3}^{n-1} (1+s^2)(s^{2h}-s^4-1+s^{4-2k})n_t, \\
%
d(a) h_2(s-s^{-1}) &=& -s^2(1+s^2) \Delta_{\widehat{\beta}}(s^2) + s^{-w+n+1} \spsilon_{x-n+1} + s^{w-n+5} \\
&&- \sum_{j=3}^{n-1} (x^{2+2k}-s^4-s^2+s^{4-2k})h_j.\end{aligned}$$
This is fxachly the result imppied by Thm. \[Ycoesdrel\]. As $z=s - r^{-1}$ is inveruiyle | 1+s^2) & s^{-w+n-1} & s^{w-n+7} \\ (1+s^2)(1+s^4) & \\ -s^2(1+s^2) s^{-w+n+1} &s^{w-n+5} \end{array} d(s) &=& -s^2(1+s^2) \Delta_{\widehat{\beta}}(s^2) s^{-w+n-1} \epsilon_{w-n+1} + \\ &&- \sum_{j=3}^{n-1} (s^{2k}-s^4-s^2+s^{6-2k})h_j, \\ % h_1(s-s^{-1}) &=& (1+s^2)(1+s^4) \Delta_{\widehat{\beta}}(s^2) -s^{-w+n-1}(1+s^2) \epsilon_{w-n+1} \\ && -s^{w-n+5}(1+s^2) + \sum_{j=3}^{n-1} (1+s^2)(s^{2k}-s^4-1+s^{4-2k})h_j, \\ d(s) h_2(s-s^{-1}) &=& -s^2(1+s^2) \Delta_{\widehat{\beta}}(s^2) + s^{-w+n+1} \epsilon_{w-n+1} + s^{w-n+5} \\ &&- \sum_{j=3}^{n-1} This exactly result by Thm. \[Hcoeffrel\]. As $z=s - s^{-1}$ is invertible | 1+s^2) & s^{-w+n-1} & s^{w-n+7} \\
(1+s^2)(1+s^4) & -s^{-w+n-1}(1+s^2) & -s^{w-n+5}(1+s^2) \\
-s^2(1+s^2) & s^{-w+n+1} &s^{w-n+5}
\End{array}
\riGht ].$$
ThIs iMplIeS $$\begIn{alIgned}
d(s) h_0(s-s^{-1}) &=& -s^2(1+s^2) \DeLTa_{\wiDehat{\beta}}(s^2) + s^{-w+n-1} \epsilon_{w-n+1} + S^{w-n+7} \\
&&- \suM_{j=3}^{N-1} (S^{2k}-s^4-s^2+S^{6-2K})h_J, \\
%
d(s) h_1(s-S^{-1}) &=& (1+s^2)(1+s^4) \DeltA_{\WiDEHat{\BeTa}}(S^2) -s^{-w+N-1}(1+s^2) \EPsIlon_{w-N+1} \\
&& -s^{w-N+5}(1+s^2) + \sum_{j=3}^{n-1} (1+S^2)(s^{2k}-s^4-1+s^{4-2k})h_j, \\
%
d(s) H_2(s-s^{-1}) &=& -S^2(1+s^2) \delta_{\widehat{\BEtA}}(s^2) + s^{-w+n+1} \epsilOn_{w-N+1} + s^{w-n+5} \\
&&- \sum_{j=3}^{n-1} (s^{2+2k}-s^4-S^2+s^{4-2k})H_j.\end{aLiGneD}$$
this iS exActly The resULt implIed by Thm. \[HCoEFfrel\]. AS $Z=s - s^{-1}$ is inVERtIble | 1+s^2) & s^{-w+n-1} & s^{ w-n+7 } \\
( 1+s^2)(1+s^4) & -s^{-w +n-1} (1 + s^2) &-s^{w -n+5}(1 + s^ 2 ) \\
-s^2( 1+s^2) & s ^{ -w+n+1} &s^{w-n+5 } \end{array}\ri ght ]. $$
T h is im pli es $$ \begin { aligne d}
d(s) h _0 ( s-s^{- 1 }) &=&- s ^2 (1+s ^2) \Delta_{\wide h at { \beta}}(s^2) + s^{-w +n - 1} \ eps ilo n_{w-n+1}+s^{w- n +7} \\ && - \ sum _ {j=3}^{n-1} ( s^{2k}-s^4- s ^2+ s^{6-2 k} )h_ j , \\
%
d( s) h_1 (s-s^{-1})&=&(1+s^2)(1 +s^4)\ Delta_{ \ widehat {\beta }}( s^2 ) - s ^{ -w +n- 1} ( 1+s ^ 2) \e p sil on_{w-n+ 1} \ \
& & -s ^ { w - n+5} (1+ s^2) + \s um_{j=3}^{n-1 } ( 1+s^ 2 )(s ^{2k} -s^4- 1+s^ {4 -2k}) h_j, \\
%d( s) h_2(s-s^{-1} ) &= & -s^2(1+ s^2 )\De lt a_{\w i dehat{ \be ta} }(s^2)+ s^{-w + n+1 }\ e ps ilon_{w-n+1} + s^{ w- n + 5} \\
&&-\sum_{ j =3 }^ { n-1} (s^ {2 +2k }-s^ 4 - s^2+s ^{4- 2 k} )h_j.\en d{alig n ed }$ $
This i s exac tl y t heresul t imp lied b y Thm. \ [Hcoe f frel\]. As $z= s - s^{-1}$ is in v e rt i ble | 1+s^2) _ _& s^{-w+n-1} _ __ __& s^{w-n+7} _ _ _\\
__ __ _(1+s^2)(1+s^4)_&_-s^{-w+n-1}(1+s^2) & -s^{w-n+5}(1+s^2) \\
_ _ _ __ _-s^2(1+s^2) & s^{-w+n+1} _ &s^{w-n+5}_
\end{array}
\right ].$$
This implies_$$\begin{aligned}
d(s)_h_0(s-s^{-1})_&=& -s^2(1+s^2) \Delta_{\widehat{\beta}}(s^2) +_s^{-w+n-1} \epsilon_{w-n+1} + s^{w-n+7} \\
&&-_\sum_{j=3}^{n-1} (s^{2k}-s^4-s^2+s^{6-2k})h_j, \\
%
d(s) h_1(s-s^{-1})_&=& (1+s^2)(1+s^4) \Delta_{\widehat{\beta}}(s^2) -s^{-w+n-1}(1+s^2) \epsilon_{w-n+1} \\_
&& -s^{w-n+5}(1+s^2) + \sum_{j=3}^{n-1} (1+s^2)(s^{2k}-s^4-1+s^{4-2k})h_j,_ \\
%
d(s) h_2(s-s^{-1}) &=& -s^2(1+s^2)_\Delta_{\widehat{\beta}}(s^2) +_s^{-w+n+1} \epsilon_{w-n+1} + s^{w-n+5}_\\
&&- \sum_{j=3}^{n-1} (s^{2+2k}-s^4-s^2+s^{4-2k})h_j.\end{aligned}$$
This_is exactly_the result implied_by Thm. \[Hcoeffrel\]. As $z=s - s^{-1}$_is invertible |
measurements. In particular, I thank Martin Grünewald for his unstinting help, and Chris Hawkes for comments on this manuscript. I also benefitted from the assistance of P. Antilogus, E. Barberio, A. Bodek, D. Cavalli, G. Chiarelli, G. Cvetic, Y.S. Chung, M. Elsing, C. Gerber, F. Gianotti, R. Hawkings, G.S. Hi, J. Holt, F. Jegerlehner, M. Kuze, I. Logashenko, K. Long, W. Menges, K. Mönig, A. Moutoussi, C. Parkes, B. Pietrzyk, R. Tenchini, J. Timmermans, A. Valassi, W. Venus, H. Voss, P. Wells, F. Yndurain and Z.G. Zhao.
[99]{} J.Z.Bai, BES Collaboration, ; J.C.Chen, these proceedings. See, for example, R.R.Akhmetshin, CMD-2 Collaboration,. H.Burkhardt and B.Pietrzyk, [preprint LAPP-EXP 2001-03](http://wwwlapp.in2p3.fr/preplapp/LAPP_EX2001_03.pdf), to appear in Physics Letters. A.D.Martin, J.Outhwaite and M.G.Ryskin,. H.N.Brown, Muon g-2 Collaboration, ; I.Logashenko, these proceedings. D.H.Brown and W.A.Worstell,. R.Alemany, M.Davier and A.Höcker,. M.Davier and A.Höcker,. M.Davier and A.Höcker,. S.Narison,. F.Jegerlehner,. J.F.de Troconiz and F.J.Yndurain,. G.Cvetic, T.Lee and I.Schmidt,. J.Bailey., ;\
R.M.Carey, Muon g-2 Collaboration, ; H.N.Brown, Muon g-2 Collaboration,. R.Barate, A | measurements. In particular, I thank Martin Grünewald for his unstinting help, and Chris Hawkes for gossip on this manuscript. I besides benefitted from the assistance of P. Antilogus, E. Barberio, A. Bodek, D. Cavalli, G. Chiarelli, G. Cvetic, Y.S. Chung, M. Elsing, C. Gerber, F. Gianotti, R. Hawkings, G.S. Hi, J. Holt, F. Jegerlehner, M. Kuze, I. Logashenko, K. Long, W. Menges, K. Mönig, A. Moutoussi, C. Parkes, B. Pietrzyk, R. Tenchini, J. Timmermans, A. Valassi, W. Venus, H. Voss, P. Wells, F. Yndurain and Z.G. Zhao.
[ 99 ] { } J.Z.Bai, BES Collaboration, ; J.C.Chen, these proceeding. witness, for example, R.R.Akhmetshin, CMD-2 Collaboration, . H.Burkhardt and B.Pietrzyk, [ preprint LAPP - EXP 2001 - 03](http://wwwlapp.in2p3.fr / preplapp / LAPP_EX2001_03.pdf), to appear in Physics Letters. A.D.Martin, J.Outhwaite and M.G.Ryskin, . H.N.Brown, Muon g-2 Collaboration, ; I.Logashenko, these proceeding. D.H.Brown and W.A.Worstell, . R.Alemany, M.Davier and A.Höcker, . M.Davier and A.Höcker, . M.Davier and A.Höcker, . S.Narison, . F.Jegerlehner, . J.F.de Troconiz and F.J.Yndurain, . G.Cvetic, T.Lee and I.Schmidt, . J.Bailey. ,; \
R.M.Carey, Muon g-2 Collaboration, ; H.N.Brown, Muon g-2 Collaboration, . R.Barate, A | mewsurements. In particular, I thank Martin Grünewand for his unsginting help, and Chris Hawked dor cimments on this manuscfipt. I aldo benefuttev from the assisvznce of P. Antjpoguv, E. Barberio, A. Nodek, D. Cavdlli, G. Chiarelni, G. Cvetic, Y.S. Chung, M. Elsing, C. Gerber, S. Gianoytl, R. Hawkings, G.F. Hi, T. Homn, N. Jegerlehner, M. Kuze, I. Logashenkk, K. Lonj, W. Menges, K. Mönog, A. Moutoussi, C. Parkes, B. Oietgzyk, R. Tenchini, J. Himmermans, Q. Vajqssi, W. Venus, H. Voss, P. Wells, F. Yndutain and Z.G. Zhao.
[99]{} J.Z.Bai, BES Collacoratnon, ; J.C.Chen, tyesf proceedingw. See, for example, R.R.Akhmedshin, CKD-2 Collaboratipn,. I.Burjhardt and B.Pietrzyk, ['reprint LAPP-EXP 2001-03](httk://wwwlapp.in2[3.fx/preplapp/LAPP_EX2001_03.pdf), to appear in [hyskxs Uetuerx. Z.D.Martln, O.Outhwaite znd M.G.Ryskib,. H.N.Brown, Muon g-2 Coklwvoration, ; I.Lofashenho, these proceedings. D.H.Brown and W.A.Worstenl,. D.Alemany, M.Davier and A.Höxker,. M.Davier and A.Höckgr,. M.Davier and A.Höcker,. S.Narison,. F.Jegerlehner,. J.F.de Troconiz atd F.J.Bnaurclk,. G.Cxwtlc, T.Lee and I.Schmidt,. J.Bailey., ;\
R.M.Carey, Muon g-2 Cjmlsbpration, ; H.N.Brocn, Muon g-2 Collabprwtojn,. R.Barate, A | measurements. In particular, I thank Martin Grünewald unstinting and Chris for comments on from assistance of P. E. Barberio, A. D. Cavalli, G. Chiarelli, G. Cvetic, Chung, M. Elsing, C. Gerber, F. Gianotti, R. Hawkings, G.S. Hi, J. Holt, Jegerlehner, M. Kuze, I. Logashenko, K. Long, W. Menges, K. Mönig, A. Moutoussi, Parkes, Pietrzyk, Tenchini, Timmermans, A. Valassi, W. Venus, H. Voss, P. Wells, F. Yndurain and Z.G. Zhao. [99]{} J.Z.Bai, Collaboration, ; J.C.Chen, these proceedings. See, for example, CMD-2 Collaboration,. H.Burkhardt and [preprint LAPP-EXP 2001-03](http://wwwlapp.in2p3.fr/preplapp/LAPP_EX2001_03.pdf), to appear Physics A.D.Martin, J.Outhwaite M.G.Ryskin,. Muon Collaboration, ; I.Logashenko, proceedings. D.H.Brown and W.A.Worstell,. R.Alemany, M.Davier and A.Höcker,. M.Davier and A.Höcker,. M.Davier and A.Höcker,. S.Narison,. F.Jegerlehner,. J.F.de and F.J.Yndurain,. and I.Schmidt,. ;\ Muon Collaboration, ; H.N.Brown, Collaboration,. R.Barate, A | measurements. In particular, I Thank MartiN GrünEwaLd fOr His uNstiNting help, and ChRIs HaWkes for comments on this mAnuscRiPT. I alSO bEnefiTted froM ThE ASsiStAnCe oF P. aNtIloguS, E. BArberio, a. Bodek, D. CavAllI, G. chiarelli, G. CvETiC, Y.S. Chung, M. ELsiNg, C. Gerber, F. GiAnoTti, R. HaWkIngS, g.S. Hi, J. holT, F. JegErlehnER, M. Kuze, i. LogashenKo, k. long, W. MENges, K. MöNIG, A. moutOussi, C. Parkes, B. PietRZyK, r. Tenchini, J. TimmErmans, a. VALaSSI, W. VEnuS, H. Voss, P. WelLs, f. YnduRAin and Z.g. zhAO.
[99]{} j.z.BaI, bES CollaboratIon, ; J.C.Chen, thESe pRoceedInGs. SEE, for exAmple, r.R.aKhmEtshin, CMD-2 CoLlabOration,. H.BUrkharDT and B.PiETrzyk, [prEprint lAPp-EXp 2001-03](httP://WwWlApp.In2P3.Fr/pREpLapP/lAPp_EX2001_03.pdf), to ApPeAr in PHysiCS lETterS. A.D.martIn, J.OuThwaite and M.G.RYskIn,. H.N.bRowN, Muon G-2 CollAborAtIon, ; I.LOgasheNko, thEsE proceedings. D.H.BRown And W.A.WorsTelL,. R.aleMaNy, M.DaVIer and a.HöCkeR,. M.DavieR and A.HöCKer,. m.DAVIEr And A.Höcker,. S.Narison,. f.JEGErLehner,. J.F.De TrocONiZ aND F.J.YndurAiN,. G.CVetiC, t.lee anD I.ScHMiDt,. J.BaileY., ;\
R.M.CarEY, MUoN g-2 CollaBoRation, ; h.N.broWn, MUon g-2 COLlabOratioN,. R.Barate, a | measurements. In particul ar, I than k Mar tin Gr ün ewal d fo r his unstinti n g he lp, and Chris Hawkes f or co mm e ntso nthismanuscr i pt . I a ls oben ef i tt ed fr omthe ass istance of P. A ntilogus, E. Ba rberio, A. Bo dek, D. Cava lli , G. C hi are l li, G . C vetic , Y.S. Chung, M. Elsin g, C. Ger b er, F.G i an otti , R. Hawkings, G. S .H i, J. Holt, F. Jeger le h ne r , M. Ku ze, I. Log as henko , K. Lon g ,W . Men g es, K. Mönig, A. Moutous s i,C. Par ke s,B . Piet rzyk, R . Te nchini, J.Timm ermans, A . Vala s si, W.V enus, H . Voss , P . W ells , F .Ynd ur a ina nd Z. G . Z hao.
[9 9] {} J.Z. Bai, B E S Col lab orat ion,; J.C.Chen, t hes e pr o cee dings . See , fo rexamp le, R. R.Akh me tshin, CMD-2 Co llab oration,. H. Bu rkh ar dt an d B.Pie trz yk, [prepr int LAP P -EX P2 0 0 1- 03](http://wwwlapp .i n 2 p3 .fr/prep lapp/L A PP _E X 2001_03. pd f), toa p pearin P h ys ics Lett ers. A . D. Ma rtin, J .O uthwai te an d M .G.Ry s kin, . H.N. Brown, M uon g - 2 Collaboratio n , ; I.Logashe n ko , th e se p roc eedings. D. H.Br o wn a nd W . A. Wor s tell, . R.A le m an y , M.Davier and A.Hö ck er,. M .Davi er and A.Höck er,. M.Dav i e r and A.H öcke r ,. S.Narison,. F. Jeger lehner,. J . F.de Tro coniz and F.J .Yndurain , . G.Cveti c,T.L eeand I .S chmidt,. J.Ba i l ey., ; \
R.M.C are y, Muon g- 2 C oll abo ra tion, ; H .N.Brown ,Mu on g -2Colla b oration, .R.B ar ate , A | measurements._In particular,_I thank Martin Grünewald_for his_unstinting_help, and_Chris_Hawkes for comments_on this manuscript._I also benefitted from_the assistance of_P._Antilogus, E. Barberio, A. Bodek, D. Cavalli, G. Chiarelli, G. Cvetic, Y.S. Chung, M._Elsing,_C. Gerber,_F._Gianotti,_R. Hawkings, G.S. Hi, J._Holt, F. Jegerlehner, M. Kuze,_I. Logashenko,_K. Long, W. Menges, K. Mönig, A. Moutoussi,_C._Parkes, B. Pietrzyk,_R. Tenchini, J. Timmermans, A. Valassi, W. Venus, H._Voss, P. Wells, F. Yndurain and_Z.G. Zhao.
[99]{} J.Z.Bai,_BES_Collaboration,_; J.C.Chen, these proceedings._See, for example, R.R.Akhmetshin, CMD-2 Collaboration,._H.Burkhardt and B.Pietrzyk, [preprint LAPP-EXP 2001-03](http://wwwlapp.in2p3.fr/preplapp/LAPP_EX2001_03.pdf),_to appear in Physics Letters. A.D.Martin, J.Outhwaite_and M.G.Ryskin,. H.N.Brown, Muon g-2 Collaboration,_; I.Logashenko, these proceedings. D.H.Brown_and W.A.Worstell,._R.Alemany, M.Davier and A.Höcker,. M.Davier_and A.Höcker,. M.Davier_and A.Höcker,._S.Narison,. F.Jegerlehner,. J.F.de_Troconiz and F.J.Yndurain,. G.Cvetic, T.Lee and_I.Schmidt,. J.Bailey., ;\
R.M.Carey,_Muon g-2 Collaboration, ; H.N.Brown, Muon_g-2_Collaboration,. R.Barate, A |
some permutation. If $p(i)<p(j)<p(k)$, then by the choice of $C$, $p(H_j)$ intersects $p(V_{i'})$ for some $i'\in\{i, \ldots,
k\}$. We proceed similarly if $p(j)<p(i)<p(k)$.
Let $X_{\infty}=\underleftarrow{\lim}\{I, f_i\}$ and $x=(x_0, x_1, x_2, \ldots)\in X_{\infty}$. If there exist piecewise linear continuous surjections $g_i: I\to I$ and a homeomorphism $h: X_{\infty}\to
\underleftarrow{\lim}\{I, g_i\}$ such that every projection of $h(x)$ is not in a zigzag of $g_i$, then $X_{\infty}$ can be embedded in the plane such that $x$ is accessible. We have the following two corollaries. See also Examples \[ex:2sin1x\]-\[ex:Nadler\].
Let $X_{\infty}=\underleftarrow{\lim}\{I, f_i\}$ where each $f_i: I\to I$ is a continuous piecewise linear surjection. If $x=(x_0, x_1, x_2, \dots)\in X_{\infty}$ is such that $x_i$ is inside a zigzag of $f_i$ for at most finitely many $i\in\N$, then there exists an embedding of $X_{\infty}$ in the plane such that $x$ is accessible.
Since $\underleftarrow{\lim}\{I, f_i\}$ and $\underleftarrow{\lim}\{I,
f_{i+n}\}$ are homeomorphic for every $n\in\N$, the proof follows using Theorem \[thm:zigzag\].
\[cor:iterations\] Let $f$ be a continuous piecewise linear surjection with finitely many critical points and $x=(x_0, x_1, x_2, \dots)\in
X_{\infty}=\underleftarrow{\lim}\{I, f\}$. If there exists $k\in\N$ such that $x_i$ is not inside a zigzag of $f^k$ for all but finitely many $i$, then there exists a planar embedding of $ | some permutation. If $ p(i)<p(j)<p(k)$, then by the choice of $ C$, $ p(H_j)$ intersects $ p(V_{i'})$ for some $ i'\in\{i, \ldots,
k\}$. We proceed similarly if $ p(j)<p(i)<p(k)$.
Let $ X_{\infty}=\underleftarrow{\lim}\{I, f_i\}$ and $ x=(x_0, x_1, x_2, \ldots)\in X_{\infty}$. If there exist piecewise analogue continuous surjections $ g_i: I\to I$ and a homeomorphism $ planck's constant: X_{\infty}\to
\underleftarrow{\lim}\{I, g_i\}$ such that every projection of $ h(x)$ is not in a zigzag of $ g_i$, then $ X_{\infty}$ can be embedded in the airplane such that $ x$ is accessible. We have the stick to two corollaries. See besides example \[ex:2sin1x\]-\[ex: Nadler\ ].
Let $ X_{\infty}=\underleftarrow{\lim}\{I, f_i\}$ where each $ f_i: I\to I$ is a continuous piecewise linear surjection. If $ x=(x_0, x_1, x_2, \dots)\in X_{\infty}$ is such that $ x_i$ is inside a zigzag of $ f_i$ for at about finitely many $ i\in\N$, then there exist an embedding of $ X_{\infty}$ in the plane such that $ x$ is accessible.
Since $ \underleftarrow{\lim}\{I, f_i\}$ and $ \underleftarrow{\lim}\{I,
f_{i+n}\}$ are homeomorphic for every $ n\in\N$, the validation follows using Theorem \[thm: zigzag\ ].
\[cor: iterations\ ] Let $ f$ be a continuous piecewise linear surjection with finitely many critical points and $ x=(x_0, x_1, x_2, \dots)\in
X_{\infty}=\underleftarrow{\lim}\{I, f\}$. If there exist $ k\in\N$ such that $ x_i$ is not inside a zigzag of $ f^k$ for all but finitely many $ i$, then there exists a planar embedding of $ | sole permutation. If $p(i)<p(j)<p(k)$, then by the choice of $C$, $p(H_j)$ intersezts $p(V_{i'})$ for some $i'\in\{i, \ldots,
n\}$. Qe priceed similarly if $p(j)<p(k)<p(k)$.
Let $X_{\ijfty}=\undeelefuarrow{\lim}\{I, f_i\}$ and $x=(x_0, x_1, x_2, \ldots)\ih X_{\inytb}$. If there exisj piecewise ninear continumur durjections $g_i: I\to I$ and a homeomor[hism $h: X_{\lnfty}\to
\underlgftargor{\lim}\{J, g_i\}$ such that every projection of $h(x)$ is iot in a zigzag of $g_i$, then $X_{\infty}$ can be fmbefded in the plane duch that $x$ is wxcessible. We have the following twk corollaries. See also Examples \[dx:2sin1r\]-\[ex:Nadler\].
Lej $S_{\ijxty}=\underlefvarrow{\jim}\{I, f_i\}$ wherc each $x_i: I\to O$ is a continupus piwcewise linear surjecvion. If $x=(x_0, x_1, x_2, \dots)\ig X_{\infty}$ hs such that $x_i$ is unwide d ziczag if $w_i$ roc af most fiiitely many $i\in\N$, then rhere exists an embtddybg of $X_{\infty}$ jn the pjane such that $x$ is accessible.
Since $\undtrleffarrow{\lim}\{I, f_i\}$ and $\underoeftarrow{\lim}\{I,
f_{i+n}\}$ are homeomor[hic for every $n\in\N$, the proof follows using Theoram \[thm:viezay\].
\[gjf:utfrations\] Let $f$ be a continuous piecewise linewd xugjection with finltely many criticak ooogts and $x=(x_0, x_1, b_2, \dots)\nh
S_{\infty}=\underleftarrlw{\lim}\{I, s\}$. If rhere exifts $l\in\N$ such that $x_i$ is not inwide a zigzay od $f^k$ for all but fnnitely many $i$, tnen tnere exists a planar emyeddinf of $ | some permutation. If $p(i)<p(j)<p(k)$, then by the $C$, intersects $p(V_{i'})$ some $i'\in\{i, \ldots, $p(j)<p(i)<p(k)$. $X_{\infty}=\underleftarrow{\lim}\{I, f_i\}$ and x_1, x_2, \ldots)\in If there exist piecewise linear continuous $g_i: I\to I$ and a homeomorphism $h: X_{\infty}\to \underleftarrow{\lim}\{I, g_i\}$ such that every of $h(x)$ is not in a zigzag of $g_i$, then $X_{\infty}$ can be in plane that is accessible. We have the following two corollaries. See also Examples \[ex:2sin1x\]-\[ex:Nadler\]. Let $X_{\infty}=\underleftarrow{\lim}\{I, f_i\}$ where $f_i: I\to I$ is a continuous piecewise linear If $x=(x_0, x_1, x_2, X_{\infty}$ is such that $x_i$ inside zigzag of for most many $i\in\N$, then exists an embedding of $X_{\infty}$ in the plane such that $x$ is accessible. Since $\underleftarrow{\lim}\{I, f_i\}$ and f_{i+n}\}$ are every $n\in\N$, proof using \[thm:zigzag\]. \[cor:iterations\] Let a continuous piecewise linear surjection with points and $x=(x_0, x_1, x_2, \dots)\in X_{\infty}=\underleftarrow{\lim}\{I, f\}$. there exists such that $x_i$ is not inside zigzag of $f^k$ for all but finitely many then there exists a planar embedding of $ | some permutation. If $p(i)<p(j)<p(k)$, thEn by the choIce of $c$, $p(H_J)$ inTeRsecTs $p(V_{I'})$ for some $i'\in\{i, \ldOTs,
k\}$. WE proceed similarly if $p(j)<p(I)<p(k)$.
LeT $X_{\INfty}=\UNdErlefTarrow{\lIM}\{I, F_I\}$ And $X=(x_0, X_1, x_2, \LdoTs)\IN X_{\Infty}$. if tHere exiSt piecewisE liNeAr continuous SUrJections $g_i: i\to i$ and a homeomoRphIsm $h: X_{\iNfTy}\tO
\UnderLefTarroW{\lim}\{I, g_I\}$ Such thAt every prOjECtion oF $H(x)$ is not IN A zIgzaG of $g_i$, then $X_{\infty}$ caN Be EMbedded in the plAne sucH tHAt $X$ IS acCesSible. We havE tHe folLOwing twO CoROLLarIEs. See also ExamPles \[ex:2sin1x\]-\[eX:nadLer\].
Let $x_{\iNftY}=\UnderlEftarRoW{\Lim}\{i, f_i\}$ where eacH $f_i: I\To I$ is a conTinuouS PiecewiSE linear SurjecTioN. If $X=(x_0, x_1, x_2, \DOtS)\iN X_{\iNfTY}$ is SUcH thAT $x_i$ Is inside A zIgZag of $F_i$ foR AT MOst fIniTely Many $i\In\N$, then there eXisTs an EMbeDding Of $X_{\inFty}$ iN tHe plaNe such That $x$ Is Accessible.
Since $\UndeRleftarroW{\liM}\{I, F_i\}$ aNd $\UnderLEftarrOw{\lIm}\{I,
F_{i+n}\}$ are hOmeomorPHic FoR EVErY $n\in\N$, the proof folloWs USInG Theorem \[Thm:zigZAg\].
\[CoR:IteratioNs\] let $F$ be a CONtinuOus pIEcEwise linEar surJEcTiOn with fInItely mAnY crItiCal poINts aNd $x=(x_0, x_1, x_2, \Dots)\in
X_{\iNfty}=\uNDerleftarrow{\liM}\{i, f\}$. If there exisTS $k\IN\n$ sUCh thAt $x_I$ is not insidE a ziGZag oF $f^k$ fOR aLl bUT finiTely mAnY $I$, tHEn there exists a planaR eMbeddiNg of $ | some permutation. If $p(i )<p(j)<p(k )$, t hen by t he c hoic e of $C$, $p(H _ j)$intersects $p(V_{i'})$ forso m e $i ' \i n\{i, \ldots , k \ }$. W epro ce e dsimil arl y if $p (j)<p(i)<p (k) $.
Let $X_{\i n ft y}=\underl eft arrow{\lim}\ {I, f_i\} $and $x=(x _0, x_1, x_2,\ ldots) \in X_{\i nf t y}$. I f theree x is t pi ecewise linear co n ti n uous surjectio ns $g_ i: I\ t o I$ an d a homeom or phism $h: X_{ \ in f t y }\t o
\underlefta rrow{\lim}\ { I,g_i\}$ s uch that e verypr o jec tion of $h( x)$is not in a zig z ag of $ g _i$, th en $X_ {\i nft y}$c an b e e mb e dde d i n t h e p lane suc hth at $x $ is a c c essi ble . We have the followin g t wo c o rol larie s. Se e al so Exam ples \ [ex:2 si n1x\]-\[ex:Nadl er\] .
Let $X _{\ in fty }= \unde r leftar row {\l im}\{I, f_i\}$ whe re e a ch $f_i: I\to I$ isac o nt inuous p iecewi s eli n ear surj ec tio n. I f $x=(x _0,x _1 , x_2, \ dots)\ i nX_ {\infty }$ is su ch th at$x_i$ is i nsidea zigzag of $ f _i$ for at mos t finitely man y $ i \ in \ N$,the n there exi stsa n em bedd i ng of $X_{\ infty }$ in the plane such that $ x$ isacces sible.
Since $\underle f t a rrow{\li m}\{ I ,f _i\}$ and $\un derle ftarrow{\l i m}\{I, f_{i+ n}\}$ ar e homeomo r p hic foreve ry$n\ in\ N $ ,the proof fol l o ws u si ng Theo rem \[thm: zig zag \].
\ [c or:iterat ions\] L et $ f$ b e a cont i nuous pi ec ewi se li nears urject ion w ithfi ni t ely many c r it i c al p oi nt s an d $ x= (x_0, x_1 , x_ 2, \dot s)\in
X_ {\i n fty} =\ un derleft arrow{\lim}\{ I, f\}$. Ifth ere exist s $k\in\N$ such that $x_i$ is not insidea z igzag of$f^k$ for al l butfin i tely m any $i $, th en th e r e exi s t sa p la nar embedd i n g o f $ | some_permutation. If_$p(i)<p(j)<p(k)$, then by the_choice of_$C$,_$p(H_j)$ intersects_$p(V_{i'})$_for some $i'\in\{i,_\ldots,
k\}$. We_proceed similarly if $p(j)<p(i)<p(k)$.
Let_$X_{\infty}=\underleftarrow{\lim}\{I, f_i\}$ and_$x=(x_0,_x_1, x_2, \ldots)\in X_{\infty}$. If there exist piecewise linear continuous surjections $g_i: I\to I$_and_a homeomorphism_$h:_X_{\infty}\to_
\underleftarrow{\lim}\{I, g_i\}$ such that every_projection of $h(x)$ is not_in a_zigzag of $g_i$, then $X_{\infty}$ can be embedded_in_the plane such_that $x$ is accessible. We have the following two_corollaries. See also Examples \[ex:2sin1x\]-\[ex:Nadler\].
Let $X_{\infty}=\underleftarrow{\lim}\{I, f_i\}$_where each $f_i:_I\to_I$_is a continuous piecewise_linear surjection. If $x=(x_0, x_1, x_2,_\dots)\in X_{\infty}$ is such that $x_i$_is inside a zigzag of $f_i$ for_at most finitely many $i\in\N$, then_there exists an embedding of_$X_{\infty}$ in_the plane such that $x$_is accessible.
Since $\underleftarrow{\lim}\{I,_f_i\}$ and_$\underleftarrow{\lim}\{I,
f_{i+n}\}$ are_homeomorphic for every $n\in\N$, the proof_follows using Theorem \[thm:zigzag\].
\[cor:iterations\]_Let $f$ be a continuous piecewise_linear_surjection with finitely_many_critical_points and_$x=(x_0, x_1, x_2,_\dots)\in_
X_{\infty}=\underleftarrow{\lim}\{I, f\}$._If_there exists $k\in\N$ such that $x_i$_is_not inside a zigzag of $f^k$ for_all but finitely many_$i$,_then there exists a_planar embedding of $ |
frac{U}{\varepsilon L/k'})$, we have that its running time is $\tilde{O}(2^k {\mathrm{poly}}(k) n^\omega /\varepsilon)$ by Proposition \[lemma:bound\], with the bound $B= \frac{U}{\varepsilon L/k'}$ which is $O(k/\varepsilon)$ since at this stage $U \leq 2L$. We conclude that the overall running time of the approximation algorithm is $\tilde{O}(2^k {\mathrm{poly}}(k) n^\omega(\log\log M + 1/\varepsilon))$.
#### Correctness.
Throughout the execution, the algorithm maintains the invariant that $L < X < X+\delta U< U$. That can be easily seen by substituting $X$ and $\delta$ for their values and observing that $L < \sqrt{LU} < L^{1/3}U^{2/3} < U$. Assume there exist a $k$-path in $G$, and let $OPT$ be the minimum-weight $k$-path. By the scaling arguments, and the fact that we have brought the loss of precision due to scaling into consideration when redefining $U$ and $L$, we have that the invariant $L \leq w(OPT) \leq U$ always holds. Due to the running-time argument, when the main loop is done we have $U/L \leq 2$. Let $P^*$ be the result of the call to $\mathcal{B}(G',k, \frac{U}{\varepsilon L/k'})$ at line \[line:return\] of the pseudo-code, and notice the the weights defined at line \[line:weights-final\] incur an $\varepsilon L /k'$ loss of precision per edge, or equivalently $\varepsilon L $ per $k$-path. By the call to the exact algorithm, we have that $w'(P^*) \leq w'(OPT)$ and therefore also ${w_{\mathrm{eff}}}(P^*) \leq {w_{\mathrm{eff}}}(OPT)$. Accounting for the loss of precision, we have that $w(P^*) \leq {w_{\mathrm{eff}}}(P^*) + \varepsilon L \leq {w_{\mathrm{eff}}}(OPT) + \varepsilon L \leq (1+\varepsilon)w(OPT)$.
$k$-tree {#sec:tree}
========
In [@ | frac{U}{\varepsilon L / k'})$, we have that its running time is $ \tilde{O}(2^k { \mathrm{poly}}(k) n^\omega /\varepsilon)$ by Proposition \[lemma: bound\ ], with the bound $ B= \frac{U}{\varepsilon L / k'}$ which is $ O(k/\varepsilon)$ since at this stagecoach $ uracil \leq 2L$. We conclude that the overall running time of the estimate algorithm is $ \tilde{O}(2^k { \mathrm{poly}}(k) n^\omega(\log\log M + 1/\varepsilon))$.
# # # # Correctness.
Throughout the execution, the algorithm maintains the invariant that $ L < X < X+\delta U < U$. That can be well see by substituting $ X$ and $ \delta$ for their value and observing that $ L < \sqrt{LU } < L^{1/3}U^{2/3 } < U$. Assume there exist a $ k$-path in $ G$, and let $ OPT$ be the minimum - weight unit $ k$-path. By the scaling arguments, and the fact that we have brought the passing of preciseness due to scaling into consideration when redefine $ U$ and $ L$, we have that the invariant $ L \leq w(OPT) \leq U$ constantly holds. Due to the running - time argumentation, when the main loop is done we have $ uracil / L \leq 2$. Let $ P^*$ be the result of the call to $ \mathcal{B}(G',k, \frac{U}{\varepsilon L / k'})$ at line \[line: return\ ] of the pseudo - code, and detect the the weights defined at line \[line: weights - final\ ] incur an $ \varepsilon L /k'$ loss of precision per edge, or equivalently $ \varepsilon L $ per $ k$-path. By the call to the exact algorithm, we have that $ w'(P^ *) \leq w'(OPT)$ and therefore also $ { w_{\mathrm{eff}}}(P^ *) \leq { w_{\mathrm{eff}}}(OPT)$. Accounting for the loss of precision, we have that $ w(P^ *) \leq { w_{\mathrm{eff}}}(P^ *) + \varepsilon L \leq { w_{\mathrm{eff}}}(OPT) + \varepsilon L \leq (1+\varepsilon)w(OPT)$.
$ k$-tree { # sec: tree }
= = = = = = = =
In [ @ | fraf{U}{\varepsilon L/k'})$, we have uhat its running jine is $\vilde{O}(2^k {\mathrm{puly}}(k) n^\omega /\varepsilon)$ by Prlpisitiin \[lemma:bound\], with the cound $B= \vrac{U}{\varwpsioin L/k'}$ whici is $O(k/\varepsimln)$ snnre at this stagg $U \leq 2L$. We conclude that tfe overall running time of the approxymation apgorithm is $\tijde{O}(2^l {\matggm{koly}}(k) n^\omega(\log\log M + 1/\varepsilon))$.
#### Dorrectiess.
Throughout yhe execution, the algorithl malntains the invariwnt that $L < X < V+\eelta U< U$. Thxt can be tavily seen gy substituting $X$ and $\delta$ for theix values ane ibsftving that $L < \sqrn{LU} < L^{1/3}U^{2/3} < U$. Assume these exisy a $k$-path in $G$, anv ler $OPT$ be the minimum-wxight $k$-path. By the ssaling arcujents, and the facr rhat fe hdve ceoueht tie moss ov pcecision dus to scalint into consideratiom ryen redefininf $U$ anq $J$, we have that the invariant $L \leq w(OPT) \les U$ always holds. Due to the running-time argulent, when the main loop is done we have $U/L \leq 2$. Let $P^*$ be tve reajlt on thd cwll to $\mathcal{B}(G',k, \frac{U}{\varepsilon L/k'})$ at line \[jjnt:renurn\] of the pseudj-code, and nptlcr the the weigfts deyjnsd at line \[line:weigjts-finaj\] incyr an $\vartpsilpn L /k'$ loss of precision pee edge, or eqlivaoently $\varepsilon P $ per $k$-patk. By tne cakl to the exact algoritkm, we gave that $w'(O^*) \leq w'(OPF)$ and therefore auso ${w_{\kathrm{eff}}}(P^*) \leq {w_{\mathrm{eff}}}(O[T)$. Accounving yor the uoss of prqcision, we have that $w(P^*) \leq {w_{\mathrl{eff}}}(P^*) + \earepsilon L \leq {w_{\mathrm{eff}}}(OPT) + \varepsiloi L \leq (1+\varepxinon)f(OPT)$.
$k$-trez {#sec:tvee}
========
In [@ | frac{U}{\varepsilon L/k'})$, we have that its running $\tilde{O}(2^k n^\omega /\varepsilon)$ Proposition \[lemma:bound\], with which $O(k/\varepsilon)$ since at stage $U \leq We conclude that the overall running of the approximation algorithm is $\tilde{O}(2^k {\mathrm{poly}}(k) n^\omega(\log\log M + 1/\varepsilon))$. #### Correctness. the execution, the algorithm maintains the invariant that $L < X < X+\delta U$. can easily by substituting $X$ and $\delta$ for their values and observing that $L < \sqrt{LU} < L^{1/3}U^{2/3} U$. Assume there exist a $k$-path in $G$, let $OPT$ be the $k$-path. By the scaling arguments, the that we brought loss precision due to into consideration when redefining $U$ and $L$, we have that the invariant $L \leq w(OPT) \leq U$ holds. Due running-time argument, the loop done we have 2$. Let $P^*$ be the result to $\mathcal{B}(G',k, \frac{U}{\varepsilon L/k'})$ at line \[line:return\] of pseudo-code, and the the weights defined at line incur an $\varepsilon L /k'$ loss of precision edge, or equivalently $\varepsilon L $ per $k$-path. By the call to the exact algorithm, that $w'(P^*) \leq w'(OPT)$ therefore also ${w_{\mathrm{eff}}}(P^*) {w_{\mathrm{eff}}}(OPT)$. for loss precision, we that $w(P^*) \leq {w_{\mathrm{eff}}}(P^*) + \varepsilon L \leq {w_{\mathrm{eff}}}(OPT) + \varepsilon \leq (1+\varepsilon)w(OPT)$. $k$-tree {#sec:tree} ======== In [@ | frac{U}{\varepsilon L/k'})$, we have thAt its runniNg timE is $\TilDe{o}(2^k {\maThrm{Poly}}(k) n^\omega /\varEPsilOn)$ by Proposition \[lemma:boUnd\], wiTh THe boUNd $b= \frac{u}{\varepsILoN l/K'}$ whIcH iS $O(k/\VaREpSilon)$ SinCe at thiS stage $U \leq 2l$. We CoNclude that thE OvErall runniNg tIme of the apprOxiMation AlGorIThm is $\TilDe{O}(2^k {\mAthrm{pOLy}}(k) n^\omEga(\log\log m + 1/\vARepsilON))$.
#### CorrecTNEsS.
ThrOughout the executiON, tHE algorithm mainTains tHe INvARIanT thAt $L < X < X+\delta u< U$. that cAN be easiLY sEEN By sUBstituting $X$ anD $\delta$ for thEIr vAlues aNd ObsERving tHat $L < \sQrT{lU} < L^{1/3}u^{2/3} < U$. Assume theRe exIst a $k$-path In $G$, and LEt $OPT$ be THe minimUm-weigHt $k$-PatH. By tHE sCaLinG aRGumENtS, anD The Fact that We HaVe broUght THE LOss oF prEcisIon duE to scaling intO coNsidERatIon whEn redEfinInG $U$ and $l$, we havE that ThE invariant $L \leq w(oPT) \lEq U$ always HolDs. due To The ruNNing-tiMe aRguMent, wheN the maiN LooP iS DONe We have $U/L \leq 2$. Let $P^*$ be tHe RESuLt of the cAll to $\mAThCaL{b}(G',k, \frac{U}{\VaRepSiloN l/K'})$ at liNe \[liNE:rEturn\] of tHe pseuDO-cOdE, and notIcE the thE wEigHts DefinED at lIne \[linE:weights-Final\] INcur an $\varepsilON L /k'$ loss of precISiON PeR Edge, Or eQuivalently $\VarePSiloN L $ peR $K$-pAth. bY the cAll to ThE ExACt algorithm, we have thAt $W'(P^*) \leq w'(oPT)$ anD therefore alsO ${w_{\mathrm{efF}}}(p^*) \LEq {w_{\mathrM{eff}}}(opT)$. aCcounting for thE loss Of precisioN, We have thAt $w(P^*) \lEq {w_{\mathrM{eff}}}(P^*) + \varePSIlon L \leq {W_{\maThrM{efF}}}(OPt) + \VArEpsilon L \leq (1+\vaREPsilOn)W(OPT)$.
$k$-trEe {#sEc:tree}
========
IN [@ | frac{U}{\varepsilon L/k'}) $, we have that it s r un ning tim e is $\tilde{O } (2^k {\mathrm{poly}}(k) n^ \omeg a/ \var e ps ilon) $ by Pr o po s i tio n\[ lem ma : bo und\] , w ith the bound $B= \ fr ac{U}{\varep s il on L/k'}$whi ch is $O(k/\ var epsilo n) $ s i nce a t t his s tage $ U \leq2L$. We c on c lude t h at theo v er allrunning time of t h ea pproximation a lgorit hm is $ \ti lde {O}(2^k {\ ma thrm{ p oly}}(k ) n ^ \ o meg a (\log\log M + 1/\varepsi l on) )$.
# ## # C o rrectn ess.
T h rou ghout the e xecu tion, the algor i thm mai n tains t he inv ari ant tha t $ L< X < X+\ d el taU < U $. Thatca nbe ea sily s e e n by su bsti tutin g $X$ and $\d elt a$ f o r t heirvalue s an dobser ving t hat $ L< \sqrt{LU} < L ^{1/ 3}U^{2/3} <U$ . A ss ume t h ere ex ist a$k$-pat h in $G $ , a nd l e t$OPT$ be the minim um - w ei ght $k$- path.B yth e scaling a rgu ment s , andthef ac t that w e have br ou ght the l oss of p rec isi on du e toscalin g into c onsid e ration when re d efining $U$ a n d$ L $, we h ave that the i nvar i ant$L \ l eq w( O PT) \ leq U $a lw a ys holds. Due to th erunnin g-tim e argument, w hen the ma i n loop isdone we have $U/L \leq 2$.Let $P^*$b e the re sultof the c all to $\ m a thcal{B} (G' ,k, \ fra c { U} {\varepsilonL / k'}) $at line \[ line:re tur n\] of th epseudo-co de, andno ti ce t hethe w e ights de fi ned a t l ine \ [ line:w eight s-fi na l\ ] in cur an$ \v a r epsi lo nL /k '$lo ss of pre c isi on peredge, orequ i vale nt ly $\vare psilon L $ pe r$k$-path.By th e call t o the ex act algorithm, we havet hat $w' (P^ *) \l eq w '(OPT)$ a ndtheref ore also $ {w_{\m athrm {e ff} } } (P^*) \ le q { w_ {\mathrm{e f f }}} (OPT) $. Acc ounting for the loss of p r eci sion, we have th at $ w ( P^ *)\ le q {w _{ \ mat h r m{eff}}}(P^*) + \varepsil on L\leq {w_{\ m ath rm {eff}}} (OPT) + \var e psilonL \leq (1 +\varepsi lo n)w( O P T)$ .
$k$-tre e {#sec: tree}
=== = ====In [@ | frac{U}{\varepsilon L/k'})$,_we have_that its running time_is $\tilde{O}(2^k_{\mathrm{poly}}(k)_n^\omega /\varepsilon)$_by_Proposition \[lemma:bound\], with the_bound $B= _\frac{U}{\varepsilon L/k'}$ which is_$O(k/\varepsilon)$ since at_this_stage $U \leq 2L$. We conclude that the overall running time of the approximation_algorithm_is $\tilde{O}(2^k_{\mathrm{poly}}(k)_n^\omega(\log\log_M + 1/\varepsilon))$.
#### Correctness.
Throughout the_execution, the algorithm maintains the_invariant that_$L < X < X+\delta U< U$. That_can_be easily seen_by substituting $X$ and $\delta$ for their values and_observing that $L < \sqrt{LU} <_L^{1/3}U^{2/3} < U$._Assume_there_exist a $k$-path in_$G$, and let $OPT$ be the_minimum-weight $k$-path. By the scaling arguments,_and the fact that we have brought_the loss of precision due to_scaling into consideration when redefining_$U$ and_$L$, we have that the_invariant $L \leq_w(OPT) \leq_U$ always holds._Due to the running-time argument, when_the main loop_is done we have $U/L \leq_2$._Let $P^*$ be_the_result_of the_call to $\mathcal{B}(G',k,__\frac{U}{\varepsilon L/k'})$_at_line \[line:return\] of the pseudo-code, and notice_the_the weights defined at line \[line:weights-final\] incur an_$\varepsilon L /k'$ loss_of_precision per edge, or_equivalently $\varepsilon L $ per_$k$-path. By the call to the_exact algorithm,_we have_that $w'(P^*) \leq w'(OPT)$ and therefore also ${w_{\mathrm{eff}}}(P^*) \leq {w_{\mathrm{eff}}}(OPT)$. Accounting_for the loss of precision, we_have that $w(P^*) \leq_{w_{\mathrm{eff}}}(P^*) +_\varepsilon_L \leq {w_{\mathrm{eff}}}(OPT)_+_\varepsilon L_\leq (1+\varepsilon)w(OPT)$.
$k$-tree {#sec:tree}
========
In [@ |
\right) =\left\langle \hat{a}_{A}^{\dag }\hat{a}%
_{A}\right\rangle =\sum_{k,l}k\left\vert C_{kl}\left( t\right) \right\vert
^{2},\quad
\sigma \left( N_{A}\right) =\sqrt{%
\sum_{k,l}k^{2}\left\vert C_{kl}\left( t\right) \right\vert ^{2}-\left(
\sum_{k,l}k\left\vert C_{kl}\left( t\right) \right\vert ^{2}\right) ^{2}}.$$Fig.\[figure11\] calculates the dynamics of the mean values and their corresponding fluctuations, and it shows that the quantum fluctuations (which are omitted in a classical dynamics) still play an important roles in the synchronized behaviors in this case. For a coherent-state BEC, two scattering modes can quickly reach an exact CS in a sense of $x_{-}\rightarrow 0, p_{-}\rightarrow 0$ (see Fig.\[figure11\] (a)(b)) but their fluctuations present “collapse and revival" behaviors and will remain for a long time due to the weak dissipations of the BEC scattering modes. But for a trapping truncation of $%
n=1-g/\chi\approx 10$, the two modes hardly reach to a complete CS but more irregular fluctuations are produced than that in the number-state BECs (see Fig.\[figure7\](d)(e)(f))because two scattering modes generated in the number-state BECs will no longer trap themselves in their initial states within their respective Hilbert space of $\mathcal{H}_N$ and the mode mixing between different number-state BECs will be involved.
![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"}](Fig13a.pdf | \right) = \left\langle \hat{a}_{A}^{\dag } \hat{a}%
_ { A}\right\rangle = \sum_{k, l}k\left\vert C_{kl}\left (t\right) \right\vert
^{2},\quad
\sigma \left (N_{A}\right) = \sqrt{%
\sum_{k, l}k^{2}\left\vert C_{kl}\left (t\right) \right\vert ^{2}-\left (
\sum_{k, l}k\left\vert C_{kl}\left (t\right) \right\vert ^{2}\right) ^{2}}.$$Fig.\[figure11\ ] calculates the dynamics of the mean values and their comparable fluctuation, and it shows that the quantum fluctuations (which are exclude in a authoritative dynamics) still play an significant roles in the synchronized behavior in this case. For a coherent - state BEC, two scatter modes can quickly pass an accurate CS in a sense of $ x_{-}\rightarrow 0, p_{-}\rightarrow 0 $ (visualize Fig.\[figure11\ ] (a)(b) ) but their fluctuations present “ collapse and revival " behavior and will remain for a long time due to the decrepit dissipations of the BEC scattering modes. But for a trapping shortness of $%
n=1 - g/\chi\approx 10 $, the two modes barely reach to a complete CS but more irregular variation are produced than that in the number - state BECs (see Fig.\[figure7\](d)(e)(f))because two scattering mood beget in the number - state BECs will no longer trap themselves in their initial states within their respective Hilbert space of $ \mathcal{H}_N$ and the mode mixing between different number - state of matter BECs will be involved.
! [ The dynamical standard 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 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"}](Fig13a.pdf | \rigjt) =\left\langle \hat{a}_{A}^{\dag }\hxt{a}%
_{A}\right\rangle =\sum_{k,l}k\neft\vedt C_{kl}\lewt( t\right) \right\vert
^{2},\quad
\sigma \lwft( N_{Q}\right) =\sqrt{%
\sum_{k,l}k^{2}\left\vdrt C_{kl}\levt( t\righr) \rijht\vert ^{2}-\left(
\sum_{k,l}k\left\vevc C_{kl}\mcft( t\xijht) \right\vert ^{2}\rlght) ^{2}}.$$Fig.\[figgre11\] calculates tfe dynamics of the mean values and thqir cortedponding fluctoatiomf, ans it shows that the quantum fluctuztions (xhich are omittrd in a classical dynamics) stipl play an importajt roles in the wynchronized behaviors in this casg. For a coherent-state BEC, two scxtternng modes cqn qulwkly reach en exabt CS in a sekxe of $f_{-}\rightatrow 0, p_{-}\rightarvow 0$ (vee Fig.\[figure11\] (a)(b)) but themr fluctuations presgnt “collapve and revival" behaciirs atd whll fwmakn roc a long himx due to ths weak dissupations of the BEC ssqttering modea. But sow a trapping truncation of $%
n=1-g/\chi\approx 10$, ths two modes hardly reacy to a complete CS buj more irrqgular fluctuations are produced than that in the numbxr-rtaue BECs (wef Fig.\[figure7\](d)(e)(f))because two scattering modes gegsrstvd in the number-sbate BECs will no kojgrt trap themselxes in thsir initial states within theie respectyve Nilbert space of $\mathcal{H}_N$ qnd the mode nixing between difyerent numbex-state BECs will be involved.
![The dyuamicam measure sjnchronizzgion of (a)(b) $S_{c}(t)$ avd (b)(d) tve mutual information $I_{AB}(t)$ in the wiole Kilbert rpacg of $\bidoplus_{N}%
\matjcal{H}_{K}$ for $N=0,1,2,\cdots,n$. The tguncajed nukber $n=15$ for (a) (c) and $n=10$ for (b)(d). The other pareketers are tne sake as thct in Nig.\[figure8\] and Fyg.\[figure9\]. Insetx: the zjomed-kn images rrom thx dashed receangles.[]{data-latgl="figure12"}](Fig13a.pvf | \right) =\left\langle \hat{a}_{A}^{\dag }\hat{a}% _{A}\right\rangle =\sum_{k,l}k\left\vert C_{kl}\left( ^{2},\quad \left( N_{A}\right) \sum_{k,l}k^{2}\left\vert C_{kl}\left( t\right) \right\vert ^{2}}.$$Fig.\[figure11\] calculates the of the mean and their corresponding fluctuations, and it that the quantum fluctuations (which are omitted in a classical dynamics) still play important roles in the synchronized behaviors in this case. For a coherent-state BEC, scattering can reach exact CS in a sense of $x_{-}\rightarrow 0, p_{-}\rightarrow 0$ (see Fig.\[figure11\] (a)(b)) but their fluctuations “collapse and revival" behaviors and will remain for long time due to weak dissipations of the BEC modes. for a truncation $% 10$, the two hardly reach to a complete CS but more irregular fluctuations are produced than that in the number-state (see Fig.\[figure7\](d)(e)(f))because modes generated the BECs no longer trap their initial states within their respective $\mathcal{H}_N$ and the mode mixing between different number-state will be ![The dynamical measure synchronization of (a)(b) and (c)(d) the mutual information $I_{AB}(t)$ in the Hilbert space of $\bigoplus_{N}% \mathcal{H}_{N}$ for $N=0,1,2,\cdots,n$. The truncated number $n=15$ for (a) (c) and (b)(d). The other parameters the same as in and Insets: zoomed-in images the dashed rectangles.[]{data-label="figure12"}](Fig13a.pdf | \right) =\left\langle \hat{a}_{A}^{\dag }\haT{a}%
_{A}\right\raNgle =\sUm_{k,L}k\lEfT\verT C_{kl}\Left( t\right) \righT\Vert
^{2},\Quad
\sigma \left( N_{A}\right) =\sqRt{%
\sum_{K,l}K^{2}\Left\VErT C_{kl}\lEft( t\rigHT) \rIGHt\vErT ^{2}-\lEft(
\SuM_{K,l}K\left\VerT C_{kl}\lefT( t\right) \rigHt\vErT ^{2}\right) ^{2}}.$$Fig.\[figURe11\] Calculates The Dynamics of thE meAn valuEs And THeir cOrrEsponDing flUCtuatiOns, and it sHoWS that tHE quantuM FLuCtuaTions (which are omitTEd IN a classical dynAmics) sTiLL pLAY an ImpOrtant roleS iN the sYNchroniZEd BEHAviORs in this case. FOr a coherent-STatE BEC, twO sCatTEring mOdes cAn QUicKly reach an eXact cS in a sensE of $x_{-}\riGHtarrow 0, P_{-}\RightarRow 0$ (see fig.\[FigUre11\] (a)(B)) BuT tHeiR fLUctUAtIonS PreSent “collApSe And reVivaL" BEHAvioRs aNd wiLl remAin for a long tiMe dUe to THe wEak diSsipaTionS oF the BeC scatTerinG mOdes. But for a trapPing TruncatioN of $%
N=1-g/\Chi\ApProx 10$, tHE two moDes HarDly reacH to a comPLetE Cs BUT mOre irregular fluctuAtIONs Are produCed thaN ThAt IN the numbEr-StaTe BEcS (See FiG.\[figURe7\](D)(e)(f))becauSe two sCAtTeRing modEs GeneraTeD in The NumbeR-StatE BECs wIll no lonGer trAP themselves in tHEir initial staTEs WIThIN theIr rEspective HiLberT SpacE of $\mAThCal{h}_n$ and tHe modE mIXiNG between different nuMbEr-statE BECs Will be involveD.
![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"}](FIg13A.pdf | \right) =\left\langle \hat {a}_{A}^{\ dag } \ha t{a }%
_{A }\ri ght\rangle =\s u m_{k ,l}k\left\vert C_{kl}\ left( t \ righ t )\righ t\vert^ {2 } , \qu ad
\ sig ma \l eft(N_{ A}\righ t) =\sqrt{ %
\ su m_{k,l}k^{2} \ le ft\vert C_ {kl }\left( t\ri ght ) \rig ht \ve r t ^{2 }-\ left(
\sum_ { k,l}k\ left\vert C _ {kl}\l e ft( t\r i g ht ) \r ight\vert ^{2}\ri g ht ) ^{2}}.$$Fig.\ [figur e1 1 \] c alc ula tes the dy na micso f the m e an v a lue s and their co rresponding flu ctuati on s,a nd itshows t h atthe quantum flu ctuations (whic h are om i tted in a cla ssi cal dyn a mi cs ) s ti l l p l ay an imp ortant r ol es in t he s y n c h roni zed beh avior s in this cas e.Fora co heren t-sta te B EC , two scatt ering m odes can quickl y re ach an ex act C S i na sen s e of $ x_{ -}\ rightar row 0,p _{- }\ r i g ht arrow 0$ (see Fig. \[ f i gu re11\] ( a)(b)) bu tt heir flu ct uat ions p resen t “c o ll apse and reviv a l" b ehavior sand wi ll re mai n for a lo ng tim e due to thew eak dissipatio n s of the BECs ca t t er i ng m ode s. But fora tr a ppin g tr u nc ati o n of$%
n= 1- g /\ c hi\approx 10$, thetw o mode s har dly reach toa complete C S but mor e ir r eg u lar fluctuatio ns ar e produced than tha t inthe numb er-stateB E Cs (seeFig .\[ fig ure 7 \ ]( d)(e)(f))beca u s e tw oscatter ing modesgen era ted in t he number -state B EC swi ll no long e r trap t he mse lv esin th e ir ini tialstat es w i thi n their re s p ecti ve H ilbe rtsp ace o f $\ m ath cal{H}_ N$ and th e m o de m ix in g betwe en differentnu mber-state B ECs willb e involve d.
![The dynamical mea s ure syn chr oniza tion of (a)(b ) $ S_{c}( t)$ and (c )(d) t he mu tu ali n forma t i on $I _{ AB}(t)$ in t hewhole H ilbe rt spac e of $\bigoplus_{N } %
\ mathcal{H}_{N }$for$ N =0 ,1, 2 ,\ c dot s, n $.T h e truncated num ber $n=15$ f o r(a) (c) an d $n =1 0$ for(b)(d). Theo ther pa rametersare the s am e as t hat in Fig.\[ figure8\ ] and Fig . \[fig u re 9\].Ins ets: t he zo omed- in ima g esfromthe da sh ed rec tangl es .[]{data -label="figure12"}](Fig 13a.pd f | \right) =\left\langle_\hat{a}_{A}^{\dag }\hat{a}%
_{A}\right\rangle_=\sum_{k,l}k\left\vert C_{kl}\left( t\right) \right\vert
^{2},\quad
\sigma_\left( N_{A}\right)_=\sqrt{%
\sum_{k,l}k^{2}\left\vert_C_{kl}\left( t\right)_\right\vert_^{2}-\left(
\sum_{k,l}k\left\vert C_{kl}\left( t\right)_\right\vert ^{2}\right) ^{2}}.$$Fig.\[figure11\]_calculates the dynamics of_the mean values_and_their corresponding fluctuations, and it shows that the quantum fluctuations (which are omitted in_a_classical dynamics)_still_play_an important roles in the_synchronized behaviors in this case._For a_coherent-state BEC, two scattering modes can quickly reach_an_exact CS in_a sense of $x_{-}\rightarrow 0, p_{-}\rightarrow 0$ (see Fig.\[figure11\]_(a)(b)) but their fluctuations present “collapse_and revival" behaviors_and_will_remain for a long_time due to the weak dissipations_of the BEC scattering modes. But_for a trapping truncation of $%
n=1-g/\chi\approx 10$,_the two modes hardly reach to_a complete CS but more_irregular fluctuations_are produced than that in_the number-state BECs_(see Fig.\[figure7\](d)(e)(f))because_two scattering modes_generated in the number-state BECs will_no longer trap_themselves in their initial states within_their_respective Hilbert space_of_$\mathcal{H}_N$_and the_mode mixing between_different_number-state BECs_will_be involved.
![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"}](Fig13a.pdf |
.
\end{array}$$ In the above formulation, the weights are absent and the two objectives represent the medical and economical perspectives, respectively. This naturally reflects the conflicting nature of the underlying decision making problem, hence, solving is interesting and challenging.
Scalarization {#subsec:scalar}
-------------
A traditional mathematical programming approach to solving a multiojective optimization problem consists in transforming an original problem with multiple objectives into a number of single-objective subproblems. This is referred to as *scalarization*. The transformation is performed by means of a scalarizing function with some user-defined parameters. A single Pareto optimal solution is sought by optimizing each subproblem. Repeated runs with different parameter settings for the scalarizing function are used to approximate multiple Pareto optimal solutions.
Several approaches to scalarization have been developed. They differ in the way the scalarizing function is formulated. The weighted sum method suggests minimizing a weighted sum of the objectives [@GaSa55]. The limitation of this method is that solutions can only be obtained in convex regions of the Pareto front. On the other hand, the $\epsilon$-constraint method [@HaLaWi71] suggests optimizing one objective function and converting all other objectives into constraints by setting an upper bound to each of them. This method can find solutions in both convex and nonconvex regions of the Pareto front. The method of weighted metrics [@Mi99] seeks to minimize the distance between the feasible objective region and some reference point. This method is also known as *compromise programming* [@Zeleny1976]. For measuring the distance, a weighted $L_p$ norm is utilized. When the value of $p$ is small, the method may fail to find solutions in nonconvex regions. When $p=\infty$, the method defines the weighted Chebyshev problem [@Bo76]. This problem consists in minimizing the largest weighted deviation of one objective. By optimizing the weighted Chebyshev problem, solutions from convex and nonconvex regions can be generated. A major drawback is that even when the original MOP is differentiable, the resulting single-objective problem is nondifferentiable. Weakly Pareto optimal solutions can be also obtained [@Mi99]. A relaxed formulation of the Chebyshev problem with differentiable scalarizing function is known as the Pascoletti–Serafini scalarization [@PaSe84]. Though, this method introduces one additional variable and one constraint for each objective function. | .
\end{array}$$ In the above formulation, the weights are absent and the two objectives constitute the aesculapian and economical perspectives, respectively. This naturally reflect the at odds nature of the underlying decision making trouble, therefore, solving is interesting and challenging.
Scalarization { # subsec: scalar }
-------------
A traditional mathematical programming approach to clear a multiojective optimization problem consists in transform an original problem with multiple objectives into a number of single - objective subproblems. This is referred to as * scalarization *. The transformation is perform by means of a scalarizing function with some user - define parameters. A single Pareto optimal solution is sought by optimize each subproblem. Repeated runs with different parameter settings for the scalarizing affair are used to approximate multiple Pareto optimal solutions.
Several approaches to scalarization have been developed. They differ in the way the scalarizing function is formulated. The weighted sum method suggests minimizing a weighted kernel of the objectives [ @GaSa55 ]. The limitation of this method acting is that solution can entirely be obtained in convex regions of the Pareto presence. On the other hand, the $ \epsilon$-constraint method [ @HaLaWi71 ] suggests optimize one objective function and converting all other objectives into constraints by setting an upper bounce to each of them. This method can find solutions in both convex and nonconvex regions of the Pareto front. The method of leaden metric unit [ @Mi99 ] seeks to minimize the distance between the feasible objective region and some reference distributor point. This method is also known as * compromise scheduling * [ @Zeleny1976 ]. For measuring the distance, a slant $ L_p$ norm is utilized. When the value of $ p$ is little, the method may fail to find oneself solution in nonconvex regions. When $ p=\infty$, the method acting defines the weighted Chebyshev problem [ @Bo76 ]. This problem consist in minimizing the largest weighted deviation of one objective. By optimize the weighted Chebyshev trouble, solutions from convex and nonconvex regions can be generated. A major drawback is that even when the original MOP is differentiable, the resulting individual - objective problem is nondifferentiable. Weakly Pareto optimal solutions can be besides obtained [ @Mi99 ]. A relaxed formulation of the Chebyshev problem with differentiable scalarizing function is known as the Pascoletti – Serafini scalarization [ @PaSe84 ]. Though, this method introduces one extra variable and one constraint for each objective function. | .
\end{wrray}$$ In the above formuuation, the weigkrs are absenf and thd two objectives represent tie mwdicao and economical perspdctives, rvspectiveoy. Tiis naturally rehmects tmz conrpictnnj nature of the underlyinc decision makhne 'roblem, hence, solving is interesting and chslpenging.
Scalarieatiom {#subavc:wcalar}
-------------
A traditional mathematidal projramming approavh to solving a multiojectlve lptimization problfm consists in eeansforming xn original problem wijh multiple objectives into a nuober pf single-ovjwctlee subproblxms. Thps is referred to as *vcalariaation*. The traksforkatuon is performed by mxans of a scalarizind functiot cith some user-defined pqramejers. D sivtle Padevo kptimap sklution is sought by iptimizing each subkrofoem. Repeated duns wytr different parameter settings for the vcamarizing function are uwed to approximate muptiple Paweto optimal solutions.
Several approaches to scalasizatmov hcyc bedb feveloped. They differ in the way the scalarizyhg flnction is formulcted. The weightec duk method suggerts miujmjzing a weighted skm of tre obhectives [@GwSa55]. Yhe limitation of this methid is that sjoutions can only bz obtained iu convgx regoons of the Pareto fronc. On tge other hajd, the $\epaklon$-constraint mdthpd [@VaLaWi71] suggests optimizing one objertive functiun amd conderting alp othcs objectives into fonsttaints by settinh an upper bound to each of them. This method vat fpnd solutnons ik both convex agd nonconvex rggions of the Oareto froht. The kethod of wqighted metriwd [@Mi99] seeks to minimizq thw diwtance cdtween the feaxible objvcuive regiob and some referenge pokht. This method nr also known as *vomoroiide pwmgramming* [@Zelany1976]. Wor keasufing tht dlstxnce, a weighted $L_p$ norm hs ufilized. When the vslme of $p$ iw small, ehe method mau fail to find solktionv ii noncpnvgx regions. When $p=\infty$, the methkd defined tme weighted Crebywhev problem [@Yo76]. This problem consists in minimizing tie largest weighted devuation of one objecjivt. By optimiziig the weighted Chebyshev problem, silutions from conyex and nonconvex regikns cat be henerated. A major drawback is that even when the original MOP is differenriable, the resultihg songle-mbleccive prjblel is nondifferentpable. Weakly Pareto optimal soluvions can te also obtained [@Mi99]. A relaxed fproulation of tfe Chebyshev problem witg differrntiable scalarizing function ix known as the Pascolettl–Serafini scalzrizatpkn [@'aSe84]. Though, this metnod incroducws one qddibional variabld amd one conshraiit yor each objective function. | . \end{array}$$ In the above formulation, the absent the two represent the medical naturally the conflicting nature the underlying decision problem, hence, solving is interesting and Scalarization {#subsec:scalar} ------------- A traditional mathematical programming approach to solving a multiojective optimization consists in transforming an original problem with multiple objectives into a number of subproblems. is to *scalarization*. The transformation is performed by means of a scalarizing function with some user-defined parameters. A Pareto optimal solution is sought by optimizing each Repeated runs with different settings for the scalarizing function used approximate multiple optimal Several to scalarization have developed. They differ in the way the scalarizing function is formulated. The weighted sum method suggests minimizing weighted sum objectives [@GaSa55]. limitation this is that solutions be obtained in convex regions of On the other hand, the $\epsilon$-constraint method [@HaLaWi71] optimizing one function and converting all other objectives constraints by setting an upper bound to each them. This method can find solutions in both convex and nonconvex regions of the Pareto method of weighted metrics seeks to minimize distance the objective and some point. This method is also known as *compromise programming* [@Zeleny1976]. For the distance, a weighted $L_p$ norm is utilized. When the $p$ small, the method fail to find solutions nonconvex When $p=\infty$, the method weighted problem consists minimizing largest weighted deviation of objective. By optimizing the weighted problem, solutions from convex generated. A major drawback is that even when original MOP is differentiable, the resulting single-objective is nondifferentiable. Weakly Pareto optimal solutions can be also obtained [@Mi99]. A formulation of problem with differentiable scalarizing function is known as Pascoletti–Serafini scalarization [@PaSe84]. Though, method introduces one additional variable and one constraint for objective | .
\end{array}$$ In the above formulaTion, the weiGhts aRe aBseNt And tHe twO objectives repREsenT the medical and economicAl perSpECtivES, rEspecTively. THIs NATurAlLy RefLeCTs The coNflIcting nAture of the UndErLying decisioN MaKing probleM, heNce, solving is IntErestiNg And CHalleNgiNg.
ScaLarizaTIon {#subSec:scalar}
-------------
a tRAditioNAl matheMATiCal pRogramming approacH To SOlving a multiojEctive OpTImIZAtiOn pRoblem consIsTs in tRAnsformINg AN ORigINal problem witH multiple obJEctIves inTo A nuMBer of sIngle-ObJEctIve subproblEms. THis is refeRred to AS *scalarIZation*. THe tranSfoRmaTion IS pErForMeD By mEAnS of A ScaLarizing FuNcTion wIth sOME USer-dEfiNed pArameTers. A single PaRetO optIMal SolutIon is SougHt By optImizinG each SuBproblem. RepeateD runS with diffEreNt ParAmEter sETtings For The ScalariZing funCTioN aRE USeD to approximate multIpLE paReto optiMal solUTiOnS.
several aPpRoaChes TO ScalaRizaTIoN have beeN develOPeD. THey diffEr In the wAy The ScaLarizINg fuNction Is formulAted. THE weighted sum meTHod suggests miNImIZInG A weiGhtEd sum of the oBjecTIves [@gaSa55]. tHe LimITatioN of thIs MEtHOd is that solutions caN oNly be oBtainEd in convex regIons of the PARETo front. ON the OThER hand, the $\epsiloN$-consTraint methOD [@HaLaWi71] sUggesTs optimiZing one obJECtive funCtiOn aNd cOnvERTiNg all other objECTiveS iNto consTraInts by sEttIng An uPpeR bOund to eacH of them. THiS mEtHoD caN find SOlutions In BotH cOnvEx and NOnconvEx regIons Of ThE parEto fronT. thE MEthoD oF wEighTed MeTrics [@mi99] seEKs tO minimiZe the distAncE BetwEeN tHe feasiBle objective rEgIon and some ReFerEnce poINT. This metHod is also known as *compromISe progrAmmIng* [@ZeLeny1976]. for measurIng The disTanCE, a weigHted $L_p$ Norm iS uTilIZEd. WheN THe ValUe Of $p$ is small, THE meThod mAy Fail To find sOlutions in nonconveX RegIons. When $p=\inftY$, thE metHOD dEfiNEs THe wEiGHteD cHebyshev problem [@bo76]. This probLeM CoNsists in miNImiZiNg the laRgest weIghteD DeviatiOn of one obJective. By OpTimiZINg tHe weighted chebysheV problem, sOLutioNS fRom coNveX and noNcOnvEx regIons caN Be gEneraTed. A maJoR drawbAck is ThAt even whEn the original MOP is diffeRentiaBle, thE reSulting siNglE-ObjEctive proBlem Is nondiffeRenTiaBle. WeAklY paretO optIMaL soLUtionS can BE also obtaINeD [@Mi99]. a RElAxed formulaTION of The ChEbySHev proBlem With differentiablE Scalarizing funCtioN IS knOwn AS the paScoletti–SerafiNi sCaLARization [@pase84]. Though, thiS method iNtROduceS one adDitionAl variaBLE aND one coNstrAinT for each oBjeCtIVe functIoN. | .
\end{array}$$ In the abo ve formula tion, th e w ei ghts are absent and th e two objectives representthe m ed i cala nd econ omicalp er s p ect iv es , r es p ec tivel y.This na turally re fle ct s the confli c ti ng natureofthe underlyi ngdecisi on ma k ing p rob lem,hence, solvin g is inte re s ting a n d chall e n gi ng.
Scalarization {# s ub s ec:scalar}
--- ------ -- - -A tr adi tional mat he matic a l progr a mm i n g ap p roach to solv ing a multi o jec tive o pt imi z ationprobl em con sists in tr ansf orming an origi n al prob l em with multi ple ob ject i ve sint oa nu m be r o f si ngle-obj ec ti ve su bpro b l e m s. T his isrefer red to as *sc ala riza t ion *. Th e tra nsfo rm ation is pe rform ed by means of ascal arizing f unc ti onwi th so m e user -de fin ed para meters. A s in g l e P areto optimal solu ti o n i s sought by op t im iz i ng eachsu bpr oble m . Repe ated ru ns withdiffer e nt p aramete rsettin gs fo r t he sc a lari zing f unctionare u s ed to approxim a te multiple P a re t o o p tima l s olutions.
Seve r al a ppro a ch est o sca lariz at i on have been developed .They d iffer in the way t he scalari z i n g functi on i s f o rmulated. Theweigh ted sum me t hod sugg estsminimizi ng a weig h t ed sum o f t heobj ect i v es [@GaSa55]. T h e lim it ation o f t his met hod is th atso lutions c an onlybe o bt ai ned in c o nvex reg io nsof th e Par e to fro nt. O n th eot h erhand, t h e$ \ epsi lo n$ -con str ai nt me thod [@H aLaWi71 ] suggest s o p timi zi ng one ob jective funct io n and conv er tin g allo t her obje ctives into constraints by sett ing an u pper bound to ea ch ofthe m . This metho d can f ind s oluti o n sinbo th convexa n d n oncon ve x re gions o f the Pareto front . Th e method of w eig hted m et ric s [ @ Mi9 9] see k s to minimize th e distance b e tw een the fe a sib le object ive reg ion a n d somereference point. T hi s me t h odis also kn own as * compromis e prog r am ming* [@ Zeleny 19 76] . For measu r ing thedistan ce , a we ighte d$L_p$ no rm is utilized. When th e valu e of$p$ is small , t h e m ethod may fai l to findsol uti ons i n n o nconv ex r e gi ons . When $p= \ infty$, t h emet h o ddefines the w e igh ted C heb y shev p robl em [@Bo76]. Thisp roblem consist s in m ini miz i ng t he largest weigh ted d e v iation o fone objecti ve. By o pt i mizin g theweight ed Cheb y s he v probl em,sol utions fr omco n vex and n on c onvexregi on s canbe gen e rate d . A major drawbac k ist h at ev e n w hen t he origin a l MO P is diffe rentiable,the re sult ing s ingle-o bj ective pr ob lem is non d ifferenti able. Weakly P aret o o ptimal sol u t ionscanbe al so obtain e d [ @ Mi 99 ] . A rel axedfo rmul ation oft he Cheby she v proble mwit h differ e nt i a ble scalar izi ng fu n c tion is kn o wn a s t h e Pas colett i–Sera fini sc a lar iz ation [ @Pa S e 84]. Thou gh, thism eth od int roduceson e ad di tio na l variable andone cons traint f o r eachobj e ctiv e funct io n. | .
\end{array}$$ In_the above_formulation, the weights are_absent and_the_two objectives_represent_the medical and_economical perspectives, respectively._This naturally reflects the_conflicting nature of_the_underlying decision making problem, hence, solving is interesting and challenging.
Scalarization {#subsec:scalar}
-------------
A traditional mathematical programming_approach_to solving_a_multiojective_optimization problem consists in transforming_an original problem with multiple_objectives into_a number of single-objective subproblems. This is referred_to_as *scalarization*. The_transformation is performed by means of a scalarizing function_with some user-defined parameters. A single_Pareto optimal solution_is_sought_by optimizing each subproblem._Repeated runs with different parameter settings_for the scalarizing function are used_to approximate multiple Pareto optimal solutions.
Several approaches_to scalarization have been developed. They_differ in the way the_scalarizing function_is formulated. The weighted sum_method suggests minimizing_a weighted_sum of the_objectives [@GaSa55]. The limitation of this method_is that solutions_can only be obtained in convex_regions_of the Pareto_front._On_the other_hand, the $\epsilon$-constraint_method [@HaLaWi71]_suggests optimizing_one_objective function and converting all other_objectives_into constraints by setting an upper bound_to each of them._This_method can find solutions_in both convex and nonconvex_regions of the Pareto front. The_method of_weighted metrics [@Mi99]_seeks to minimize the distance between the feasible objective region and_some reference point. This method is_also known as *compromise_programming* [@Zeleny1976]. For_measuring_the distance, a_weighted_$L_p$ norm_is utilized. When the value of $p$_is small,_the method may fail to find_solutions in nonconvex regions._When_$p=\infty$, the method defines the weighted_Chebyshev problem [@Bo76]. This problem consists in_minimizing the largest weighted deviation_of_one_objective. By optimizing the weighted_Chebyshev problem, solutions from convex and_nonconvex regions can_be generated. A major drawback is that_even_when the original MOP is differentiable,_the_resulting single-objective problem is nondifferentiable. Weakly_Pareto_optimal_solutions can be also obtained [@Mi99]._A relaxed formulation of the Chebyshev_problem with differentiable scalarizing function is known as the_Pascoletti–Serafini scalarization [@PaSe84]. Though,_this method introduces one additional_variable_and_one constraint for each objective function. |
Birkhäuser Boston, Inc., Boston, MA 1993.
B. Bojarski, *The abstract linear conjugation problem and Fredholm pairs of subspaces*, In Memoriam I. N. Vekua (Tbilisi Univ. 1979) 45–60.
M. Braverman, *New proof of the cobordism invariance of the index*, Proc. Amer. Math. Soc. [**130**]{} (2002) 1095–1101.
A. P. Calderón, *Boundary value problems for elliptic equations*, 1963 Outlines Joint Sympos. Partial Differential Equations (Novosibirsk, 1963), 303–304.
S. Y. Cheng, S. T. Yau, *On the existence of a complete Kähler metric on noncompact complex manifolds and the regularity of Fefferman’s equation*, Comm. Pure Appl. Math. **33** (1980), no. 4, 507–544.
H. Donnelly, C. Fefferman, *$L\sp{2}$-cohomology and index theorem for the Bergman metric*, Ann. of Math.**118** (1983), no. 3, 593–618.
C. L. Epstein, *Subelliptic ${\rm Spin}\sb {\mathbb C}$ Dirac operators. I*, Ann. of Math. (2) **166** (2007), no. 1, 183–214.
C. L. Epstein, *Subelliptic ${\rm Spin}\sb {\mathbb C}$ Dirac operators. II. Basic estimates*, Ann. of Math. (2) **166** (2007), no. 3, 723–777.
C. L. Epstein, R. Melrose, *Shrinking tubes and the d-bar Neumann problem*, preprint available online at http://www.math.upenn.edu/$\sim$cle/papers/index.html.
C. Fefferman *The Bergman kernel and biholomorphic mappings of pseudoconvex domains,* Invent. Math. **26** (1974), 1–65.
C. Fefferman, C. R. | Birkhäuser Boston, Inc., Boston, MA 1993.
B. Bojarski, * The abstract linear conjugation problem and Fredholm pairs of subspace *, In Memoriam I. N. Vekua (Tbilisi Univ. 1979) 45–60.
M. Braverman, * newfangled proof of the cobordism invariance of the exponent *, Proc. Amer. Math. Soc. [ * * 130 * * ] { } (2002) 1095–1101.
A. P. Calderón, * Boundary value problems for egg-shaped equations *, 1963 Outlines Joint Sympos. Partial Differential Equations (Novosibirsk, 1963), 303–304.
S. Y. Cheng, S. T. Yau, * On the universe of a arrant Kähler metric on noncompact complex manifold and the regularity of Fefferman ’s equation *, Comm. Pure Appl. Math. * * 33 * * (1980), no. 4, 507–544.
H. Donnelly, C. Fefferman, * $ L\sp{2}$-cohomology and index theorem for the Bergman measured *, Ann. of Math.**118 * * (1983), no. 3, 593–618.
C. L. Epstein, * Subelliptic $ { \rm Spin}\sb { \mathbb C}$ Dirac operators. I *, Ann. of Math. (2) * * 166 * * (2007), no. 1, 183–214.
C. L. Epstein, * Subelliptic $ { \rm Spin}\sb { \mathbb C}$ Dirac operators. II. Basic estimates *, Ann. of Math. (2) * * 166 * * (2007), no. 3, 723–777.
C. L. Epstein, R. Melrose, * Shrinking tube and the d - bar Neumann trouble *, preprint available online at http://www.math.upenn.edu/$\sim$cle/papers/index.html.
C. Fefferman * The Bergman kernel and biholomorphic mappings of pseudoconvex domain, * Invent. Math. * * 26 * * (1974), 1–65.
C. Fefferman, C. R. | Bigkhäuser Boston, Inc., Bostok, MA 1993.
B. Bojarski, *Tkw abstcact lihear congugation problem and Fredholl pairs of subspaces*, In Memorkam I. N. Vekla (Tbilisu Unmv. 1979) 45–60.
M. Braverman, *Nex proof of the goborbiwm invariance pf the indax*, Proc. Amer. Matv. Suc. [**130**]{} (2002) 1095–1101.
A. P. Calderón, *Boundary value problems for elkiotic equations*, 1963 Ouuliges Noint Sympos. Partial Differential Esuationv (Novosibirsk, 1963), 303–304.
S. Y. Cheng, S. T. Yau, *On the exlstejce of a complete Nähler metrix on boncompact cumplex manpyolds and tge regularity of Fefferman’s equxtion*, Comm. Pure Qppl. Mwjh. **33** (1980), no. 4, 507–544.
H. Donnxlly, C. Fefferman, *$L\sp{2}$-cohomonogy anc index theorek fmr rhe Bergman metric*, Ani. of Math.**118** (1983), no. 3, 593–618.
C. L. Epsteyn, *Subellhpcic ${\rm Spin}\sb {\mathbb X}$ Eirac opesatofw. I*, Xnn. kf Mzth. (2) **166** (2007), jo. 1, 183–214.
R. L. Epstein, *Shbelliptic ${\em Spin}\sb {\mathbb C}$ Ciwqc operators. IJ. Basic eftimates*, Ann. of Math. (2) **166** (2007), no. 3, 723–777.
C. L. Epstein, R. Melgose, *Shrinking tubes and thw d-bar Neumann problel*, preprine available online at http://www.math.upenn.edu/$\sim$cle/pa[ers/iiddx.humo.
C. Fefwwrlan *The Bergman kernel and biholomorphic mappyhgx pf pseudoconvew domains,* Invent. Mayh. **26** (1974), 1–65.
V. Sefferman, C. R. | Birkhäuser Boston, Inc., Boston, MA 1993. B. abstract conjugation problem Fredholm pairs of Vekua Univ. 1979) 45–60. Braverman, *New proof the cobordism invariance of the index*, Amer. Math. Soc. [**130**]{} (2002) 1095–1101. A. P. Calderón, *Boundary value problems for equations*, 1963 Outlines Joint Sympos. Partial Differential Equations (Novosibirsk, 1963), 303–304. S. Y. S. Yau, the of a complete Kähler metric on noncompact complex manifolds and the regularity of Fefferman’s equation*, Comm. Appl. Math. **33** (1980), no. 4, 507–544. H. C. Fefferman, *$L\sp{2}$-cohomology and theorem for the Bergman metric*, of (1983), no. 593–618. L. *Subelliptic ${\rm Spin}\sb C}$ Dirac operators. I*, Ann. of Math. (2) **166** (2007), no. 1, 183–214. C. L. Epstein, *Subelliptic Spin}\sb {\mathbb operators. II. estimates*, of (2) **166** (2007), 723–777. C. L. Epstein, R. Melrose, the d-bar Neumann problem*, preprint available online at C. Fefferman Bergman kernel and biholomorphic mappings of domains,* Invent. Math. **26** (1974), 1–65. C. Fefferman, R. | Birkhäuser Boston, Inc., Boston, mA 1993.
B. BojarskI, *The aBstRacT lIneaR conJugation probleM And FRedholm pairs of subspaceS*, In MeMoRIam I. n. veKua (TbIlisi UnIV. 1979) 45–60.
M. bRAveRmAn, *new PrOOf Of the CobOrdism iNvariance oF thE iNdex*, Proc. Amer. mAtH. Soc. [**130**]{} (2002) 1095–1101.
A. P. CaldEróN, *Boundary valUe pRoblemS fOr eLLiptiC eqUatioNs*, 1963 OutlINes JoiNt Sympos. PArTIal DifFErentiaL eQuAtioNs (Novosibirsk, 1963), 303–304.
S. Y. ChENg, s. t. Yau, *On the existEnce of A cOMpLETe KÄhlEr metric on NoNcompACt complEX mANIFolDS and the regulaRity of FeffeRMan’S equatIoN*, CoMM. Pure APpl. MaTh. **33** (1980), NO. 4, 507–544.
H. DOnnelly, C. FefFermAn, *$L\sp{2}$-cohoMology ANd index THeorem fOr the BErgMan MetrIC*, ANn. Of MAtH.**118** (1983), No. 3, 593–618.
C. l. epSteIN, *SuBelliptiC ${\rM SPin}\sb {\MathBB c}$ dIrac OpeRatoRs. I*, AnN. of Math. (2) **166** (2007), no. 1, 183–214.
C. L. EpSteIn, *SuBEllIptic ${\Rm SpiN}\sb {\mAtHbb C}$ DIrac opEratoRs. iI. Basic estimateS*, Ann. Of Math. (2) **166** (2007), no. 3, 723–777.
C. l. EpStEin, r. MElrosE, *shrinkIng TubEs and thE d-bar NeUManN pROBLeM*, preprint available OnLINe At http://wwW.math.uPEnN.eDU/$\sim$cle/pApErs/IndeX.HTml.
C. FEffeRMaN *The BergMan kerNEl AnD biholoMoRphic mApPinGs oF pseuDOconVex domAins,* InveNt. MatH. **26** (1974), 1–65.
c. Fefferman, C. R. | Birkhäuser Boston, Inc.,Boston, MA 1993 .
B.Bo jars ki,*The abstractl inea r conjugation problemand F re d holm pa irs o f subsp a ce s * , I nMe mor ia m I . N.Vek ua (Tbi lisi Univ. 19 79 ) 45–60.
M. Br averman, * New proof of th e c obordi sm in v arian ceof th e inde x *, Pro c. Amer.Ma t h. Soc . [**130 * * ]{ } (2 002) 1095–1101.
A .P . Calderón, *B oundar yv al u e pr obl ems for el li ptice quation s *, 1 9 63O utlines Joint Sympos. Pa r tia l Diff er ent i al Equ ation s( Nov osibirsk, 1 963) , 303–304 .
S.Y . Cheng , S. T. Yau,*On th e ex i st en ceof a c o mp let e Kä hler met ri con no ncom p a c t com ple x ma nifol ds and the re gul arit y of Feff erman ’s e qu ation *, Com m. Pu re Appl. Math. ** 33** (1980),no. 4 , 5 07 –544.
H. Do nne lly , C. F efferma n , * $L \ s p {2 }$-cohomology andin d e xtheoremfor th e B er g man metr ic *,Ann. o f Mat h.** 1 18 ** (1983 ), no. 3, 5 93–618.
C. L.Ep ste in, *Sub e llip tic ${ \rm Spin }\sb{ \mathbb C}$ Di r ac operators. I* , An n . of Ma th. (2) **1 66** (200 7),n o. 1, 183–2 14.
C. L. Epstein, *Subellipt ic ${\rm Spin }\sb {\mathbb C}$ Dirac o p erators. II. Ba s ic estimates*, Ann. of Math.( 2) **166 ** (2 007), no . 3, 723– 7 7 7.
C. L . E pst ein , R . Me lrose, *Shrin k i ng t ub es andthe d-barNeu man n p rob le m*, prepr int avai la bl eon lin e ath ttp://ww w. mat h. upe nn.ed u /$\sim $cle/ pape rs /i n dex .html.C. F effe rm an *Th e B er gmankern e l a nd biho lomorphic ma p ping sof pseudo convex domain s, * Invent.Ma th. **26* * (1974),1–65.
C. Fefferman, C. R. | Birkhäuser_Boston, Inc.,_Boston, MA 1993.
B. Bojarski, *The_abstract linear_conjugation_problem and_Fredholm_pairs of subspaces*,_In Memoriam I. N. Vekua_(Tbilisi Univ. 1979) 45–60.
M. Braverman, *New_proof of the_cobordism_invariance of the index*, Proc. Amer. Math. Soc. [**130**]{} (2002) 1095–1101.
A. P. Calderón, *Boundary value problems for elliptic equations*, 1963_Outlines_Joint Sympos. Partial_Differential_Equations_(Novosibirsk, 1963), 303–304.
S. Y. Cheng, S. T. Yau,_*On the existence of a_complete Kähler_metric on noncompact complex manifolds and the regularity_of_Fefferman’s equation*, Comm. Pure_Appl. Math. **33** (1980), no. 4, 507–544.
H. Donnelly, C. Fefferman, *$L\sp{2}$-cohomology and index_theorem for the Bergman metric*, Ann. of_Math.**118** (1983), no. 3,_593–618.
C. L. Epstein,_*Subelliptic_${\rm Spin}\sb {\mathbb C}$_Dirac operators. I*, Ann. of Math. (2) **166** (2007),_no. 1, 183–214.
C. L. Epstein, *Subelliptic ${\rm Spin}\sb {\mathbb_C}$ Dirac operators. II. Basic estimates*, Ann. of Math. (2) **166**_(2007), no. 3, 723–777.
C. L. Epstein, R. Melrose, *Shrinking tubes_and the d-bar Neumann problem*,_preprint available_online at http://www.math.upenn.edu/$\sim$cle/papers/index.html.
C. Fefferman *The Bergman_kernel and biholomorphic_mappings of_pseudoconvex domains,* Invent. Math. **26**_(1974), 1–65.
C. Fefferman, C. R. |
[^2]: The authors are with the State Key Laboratory of Advanced Optical Communication Systems and Networks, Shanghai Jiao Tong University, Shanghai 200030, China (e-mail: {cuiqingpei, yetong, ttlee, wguo, wshu}@sjtu.edu.cn.
---
abstract: 'A random packing of hard particles represents a fundamental model for granular matter. Despite its importance, analytical modeling of random packings remains difficult due to the existence of strong correlations which preclude the development of a simple theory. Here, we take inspiration from liquid theories for the $n$-particle angular correlation function to develop a formalism of random packings of hard particles from the bottom-up. A progressive expansion into a shell of particles converges in the large layer limit under a Kirkwood-like approximation of higher-order correlations. We apply the formalism to hard disks and predict the density of two-dimensional random close packing (RCP), $\phi_{\rm rcp} = 0.85\pm0.01$, and random loose packing (RLP), $\phi_{\rm rlp} = 0.67\pm0.01$. Our theory also predicts a phase diagram and angular correlation functions that are in good agreement with experimental and numerical data.'
author:
- Yuliang Jin
- 'James G. Puckett'
- 'Hernán A. Makse'
title: Statistical theory of correlations in random packings of hard particles
---
Introduction
============
In recent years, many important practical applications have been found for granular materials, which are commonly modeled by dense packings of hard spheres [@Aste2008Book]. Sphere packing problems are equivalent to important problems in number theory and error-correcting coding [@Conway1999], both of which are fundamental in computer science. Despite its importance, analytical developments in granular matter have lagged behind in comparison with other fields of condensed matter, like liquid theory. In the case of random packings [@Bernal1960A], analytical results are still difficult to obtain. The theoretical difficulty arises due to [*(i)*]{} the absence of a first principle derivation of the statistical ensemble of packings (such as Liouville’s theorem in ordinary liquids) that would lead to a proper definition of randomness [@Torquato2000], and [*(ii)*]{} the existence of correlations between the particle positions determining the properties of random packings.
In previous theories, these correlations have been neglected | [ ^2 ]: The authors are with the State Key Laboratory of Advanced Optical Communication Systems and Networks, Shanghai Jiao Tong University, Shanghai 200030, China (e - mail: { cuiqingpei, yetong, ttlee, wguo, wshu}@sjtu.edu.cn.
---
abstract:' A random packing of unvoiced particle represents a fundamental model for chondritic matter. Despite its importance, analytical model of random packing remains difficult due to the existence of strong correlation which preclude the development of a childlike theory. Here, we take inspiration from liquid theories for the $ n$-particle angular correlation coefficient function to develop a formalism of random packings of arduous particles from the bottom - up. A progressive expansion into a carapace of particles converges in the bombastic layer limit under a Kirkwood - like approximation of higher - orderliness correlations. We apply the formalism to hard disks and predict the concentration of two - dimensional random near packing (RCP), $ \phi_{\rm rcp } = 0.85\pm0.01 $, and random loose packing (RLP), $ \phi_{\rm rlp } = 0.67\pm0.01$. Our theory also predicts a phase diagram and angular correlation functions that are in good agreement with experimental and numerical datum.'
writer:
- Yuliang Jin
-' James G. Puckett'
-' Hernán A. Makse'
title: Statistical hypothesis of correlation in random packing of hard particles
---
Introduction
= = = = = = = = = = = =
In recent years, many important practical application have been found for granular materials, which are commonly modeled by dense packings of arduous spheres [ @Aste2008Book ]. Sphere packing problems are equivalent to authoritative problems in act theory and error - correcting coding [ @Conway1999 ], both of which are fundamental in calculator science. Despite its importance, analytical developments in granular matter have lagged behind in comparison with other fields of condensed topic, like liquid theory. In the case of random packings [ @Bernal1960A ], analytical consequence are still unmanageable to obtain. The theoretical difficulty arises due to [ * (one) * ] { } the absence of a first principle derivation of the statistical ensemble of packings (such as Liouville ’s theorem in ordinary liquid) that would lead to a proper definition of randomness [ @Torquato2000 ], and [ * (ii) * ] { } the existence of correlations between the particle positions determining the properties of random packings.
In previous theory, these correlations have been neglected |
[^2]: Thf authors are with the Suate Key Laboratoty of Adtanced Kptical Zommunication Systems and Nevworjs, Shqnghai Jiao Tong Univefsity, Shajghai 200030, Cyina (w-mail: {cuiqmhgpei, yccong, fblee, cgno, wshu}@sjtu.edu.ck.
---
abstract: 'D random packitg oy hard particles represents a fundamqntal mpdfl for granulat matuer. Deslptt its importance, analytical modeljng of gandom packings rrmains difficult due to thf exlstence of strong forrelationw whyxh preclude ghe develokmznt of a sijple theory. Here, we take inspirxtion from liquud thfmries for tie $n$-pagticle angulav correnation gunction to deyelop a dormalism of random peckings of hard partycles frok che bottom-up. A progrewsuve efpanvion untu a sielm of pwrtmcles convedges in the large layer limit inqvt a Kirkwood-mike a[pwoximation of higher-order correlations. Fe zpply the formalism to yard disks and predicj the densyty of two-dimensional random close packing (RCP), $\phh_{\rm rrp} = 0.85\pn0.01$, wve gandom loose packing (RLP), $\phi_{\rm rlp} = 0.67\pm0.01$. Our thqkru slso predicts c phase diagram snf sggular correlxtion yhndtions that are in good adreemwnt with txperomental and numerical data.'
aythor:
- Yuliany Jun
- 'James G. Puckett'
- 'Hernán A. Mayse'
totle: Xtatistical theory of curremations in gandom padyings of hard paftibles
---
Hntroducukon
============
In recent yearf, many im'ortaut practkcal applisations hage bect found for granulwr majerialv, which arf commonly modeled by dense packings of hard xpvervs [@Aste2008Bojk]. Spmere packing prjblems are equnvalent co impurtant proglems ii number thejry and error-wlrrecting coving [@Conwwy1999], bith if whicf are fundamentsl in comiuuer sciencw. Despite its impovtancg, znalytical deveoopnents in granulsr oateeg heve lwcged behind hn cumpxtison with ouker wielcs of condensed mattar, ljke liquid theory. On the casg of randjm packings [@Brrnal1960A], analytical gesulvs are stilk dyfficult to obtain. The theoretjcal difflcujty arises dte ti [*(i)*]{} the abseuce of a first principle derivation of tie statistical ensemblg of packings (such as Kiouville’s vheorei in orditary liquids) that woyld lead to a proker definition of randojness [@Dorquwto2000], and [*(ii)*]{} the existence of correlations between the particle positions dwtermiiigg the propsrtirs of rcndpm pachingd.
Mn previous theorpes, these correlations have been neglectad | [^2]: The authors are with the State of Optical Communication and Networks, Shanghai China {cuiqingpei, yetong, ttlee, wshu}@sjtu.edu.cn. --- abstract: random packing of hard particles represents fundamental model for granular matter. Despite its importance, analytical modeling of random packings difficult due to the existence of strong correlations which preclude the development of simple Here, take from liquid theories for the $n$-particle angular correlation function to develop a formalism of random packings hard particles from the bottom-up. A progressive expansion a shell of particles in the large layer limit a approximation of correlations. apply formalism to hard and predict the density of two-dimensional random close packing (RCP), $\phi_{\rm rcp} = 0.85\pm0.01$, and random loose (RLP), $\phi_{\rm 0.67\pm0.01$. Our also a diagram and angular that are in good agreement with data.' author: - Yuliang Jin - 'James G. - 'Hernán Makse' title: Statistical theory of correlations random packings of hard particles --- Introduction ============ recent years, many important practical applications have been found for granular materials, which are commonly dense packings of hard [@Aste2008Book]. Sphere packing are to problems number theory error-correcting coding [@Conway1999], both of which are fundamental in computer science. its importance, analytical developments in granular matter have lagged behind with fields of condensed like liquid theory. In case random packings [@Bernal1960A], analytical still to difficulty due [*(i)*]{} the absence of first principle derivation of the ensemble of packings (such liquids) that would lead to a proper definition randomness [@Torquato2000], and [*(ii)*]{} the existence of between the particle positions determining the properties of random packings. In previous these correlations neglected |
[^2]: The authors are with the State key LaboratOry of advAncEd optiCal COmmunication SySTems And Networks, Shanghai JiaO Tong unIVersITy, shangHai 200030, ChinA (E-mAIL: {cuIqInGpeI, yEToNg, ttlEe, wGuo, wshu}@Sjtu.edu.cn.
---
aBstRaCt: 'A random pacKInG of hard parTicLes representS a fUndameNtAl mODel foR grAnulaR matteR. despitE its imporTaNCe, analYTical moDELiNg of Random packings remAInS Difficult due to The exiStENcE OF stRonG correlatiOnS whicH PrecludE ThE DEVelOPment of a simplE theory. Here, WE taKe inspIrAtiON from lIquid ThEOriEs for the $n$-paRticLe angular CorrelATion funCTion to dEvelop A foRmaLism OF rAnDom PaCKinGS oF haRD paRticles fRoM tHe botTom-uP. a PROgreSsiVe exPansiOn into a shell oF paRticLEs cOnverGes in The lArGe layEr limiT undeR a kirkwood-like appRoxiMation of hIghEr-OrdEr CorreLAtions. we aPplY the forMalism tO HarD dISKS aNd predict the densitY oF TWo-DimensioNal ranDOm ClOSe packinG (RcP), $\pHi_{\rm RCP} = 0.85\pm0.01$, anD ranDOm Loose pacKing (RLp), $\PhI_{\rM rlp} = 0.67\pm0.01$. OUr Theory AlSo pRedIcts a PHase DiagraM and anguLar coRRelation functiONs that are in goOD aGREeMEnt wIth ExperimentaL and NUmerIcal DAtA.'
auTHor:
- YuLiang jiN
- 'jaMEs G. Puckett'
- 'Hernán A. MaKsE'
title: statiStical theory oF correlatiONS In random PackINgS Of hard particleS
---
IntrOduction
============
In REcent yeaRs, manY importaNt practicAL ApplicatIonS haVe bEen FOUnD for granular mATEriaLs, Which arE coMmonly mOdeLed By dEnsE pAckings of Hard spheReS [@AStE2008BOok]. spherE Packing pRoBleMs Are EquivALent to ImporTant PrObLEms In numbeR ThEORy anD eRrOr-coRreCtIng coDing [@cOnwAy1999], both oF which are FunDAmenTaL iN computEr science. DespItE its importAnCe, aNalytiCAL developMents in granular matter haVE lagged BehInd in CompArison witH otHer fieLds OF condeNsed maTter, lIkE liQUId theORY. IN thE cAse of randoM PAckIngs [@BErNal1960A], AnalytiCal results are still DIffIcult to obtain. the TheoRETiCal DIfFIcuLtY AriSES due to [*(i)*]{} the absenCe of a first PrINcIple derivaTIon Of The statIstical EnsemBLe of pacKings (such As LiouvilLe’S theOREm iN ordinary lIquids) thAt would leAD to a pROpEr defIniTion of RaNdoMness [@torquaTO2000], anD [*(ii)*]{} thE existEnCe of coRrelaTiOns betweEn the particle positions dEtermiNing tHe pRoperties Of rANdoM packings.
in prEvious theoRieS, thEse coRreLAtionS havE BeEn nEGlectEd |
[^2]: The authors are wit h the Stat e Key La bor at oryof A dvanced Optica l Com munication Systems and Netw or k s, S h an ghaiJiao To n gU n ive rs it y,Sh a ng hai 2 000 30, Chi na (e-mail : { cu iqingpei, ye t on g, ttlee,wgu o, wshu}@sjt u.e du.cn.
--- abstr act : 'Arandom packin g of hard p a rticle s repres e n ts a f undamental modelf or granular matte r. Des pi t ei t s i mpo rtance, an al ytica l modeli n go f ran d om packings r emains diff i cul t dueto th e exist enceof str ong correla tion s which p reclud e the de v elopmen t of a si mpl e th e or y. He re , we ta kei nsp irationfr om liqu id t h e o r iesfor the $n$- particle angu lar cor r ela tionfunct ionto deve lop aforma li sm of random pa ckin gs of har d p ar tic le s fro m the b ott om- up. A p rogress i veex p a n si on into a shell of p a r ti cles con verges in t h e largela yer lim i t unde r aK ir kwood-li ke app r ox im ation o fhigher -o rde r c orrel a tion s. Weapply th e for m alism to hardd isks and pred i ct t he dens ity of two-dim ensi o nalrand o mclo s e pac king(R C P) , $\phi_{\rm rcp} =0. 85\pm0 .01$, and random l oose packi n g (RLP), $ \phi _ {\ r m rlp} = 0.67\ pm0.0 1$. Our th e ory also pred icts a p hase diag r a m and an gul arcor rel a t io n functions t h a t ar ein good ag reement wi thexp eri me ntal andnumerica lda ta .'
au thor: - Yulian gJin
- 'J amesG . Puck ett'- 'H er ná n A. Makse' ti t l e: S ta ti stic alth eoryof c o rre lations in rando m p a ckin gs o f hardparticles
---
Introducti on
== ====== = = ==
In r ecent years, many impor t ant pra cti cal a ppli cations h ave beenfou n d forgranul ar ma te ria l s , whi c h a reco mmonly mod e l edby de ns e pa ckingsof hard spheres [@ A ste 2008Book]. Sp her e pa c k in g p r ob l ems a r e e q u ivalent to impo rtant prob le m sin numbert heo ry and er ror-cor recti n g codin g [@Conwa y1999], b ot h of w hic h are fund amentalin comput e r sci e nc e. De spi te its i mpo rtanc e, ana l yti cal d evelop me nts in gran ul ar matte r have lagged behind in compa rison wi th otherfie l dsof conden sedmatter, li keliq uid t heo r y. In the ca seo f ran domp ackings [ @ Be rna l 1 96 0A], analyt i c a l r esult s a r e stil l di fficult to obtain . The theoretic al d i f fic ult y ari se s due to [*(i) *]{ }t h e absenc eof a firstprincipl ed eriva tion o f thestatist i c al ensemb le o f p ackings ( suc ha s Liouv il le ’ s theo remin ordin ary li q uids ) that would leadto ap r operd efi nitio nof rand o mnes s [@Torqua to2000], an d [*(i i)*] {} th e exist en ce ofcor re lations be t ween theparti cle pos it ions de termin ingt h e pro pert ie s o f randomp a ck i ng s.
In pre vious t heor ies, thes e correla tio n s havebe enn e glecte d |
[^2]: The_authors are_with the State Key_Laboratory of_Advanced_Optical Communication_Systems_and Networks, Shanghai_Jiao Tong University,_Shanghai 200030, China (e-mail:_{cuiqingpei, yetong, ttlee,_wguo,_wshu}@sjtu.edu.cn.
---
abstract: 'A random packing of hard particles represents a fundamental model for granular_matter._Despite its_importance,_analytical_modeling of random packings remains_difficult due to the existence_of strong_correlations which preclude the development of a simple_theory._Here, we take_inspiration from liquid theories for the $n$-particle angular correlation_function to develop a formalism of_random packings of_hard_particles_from the bottom-up. A_progressive expansion into a shell of_particles converges in the large layer_limit under a Kirkwood-like approximation of higher-order_correlations. We apply the formalism to_hard disks and predict the_density of_two-dimensional random close packing (RCP),_$\phi_{\rm rcp} =_0.85\pm0.01$, and_random loose packing_(RLP), $\phi_{\rm rlp} = 0.67\pm0.01$. Our_theory also predicts_a phase diagram and angular correlation_functions_that are in_good_agreement_with experimental_and numerical data.'
author:
-_Yuliang_Jin
- 'James_G._Puckett'
- 'Hernán A. Makse'
title: Statistical theory_of_correlations in random packings of hard particles
---
Introduction
============
In_recent years, many important_practical_applications have been found_for granular materials, which are_commonly modeled by dense packings of_hard spheres_[@Aste2008Book]. Sphere_packing problems are equivalent to important problems in number theory and_error-correcting coding [@Conway1999], both of which_are fundamental in computer_science. Despite_its_importance, analytical developments_in_granular matter_have lagged behind in comparison with other_fields of_condensed matter, like liquid theory. In_the case of random_packings_[@Bernal1960A], analytical results are still difficult_to obtain. The theoretical difficulty arises_due to [*(i)*]{} the absence_of_a_first principle derivation of the_statistical ensemble of packings (such as_Liouville’s theorem in_ordinary liquids) that would lead to a_proper_definition of randomness [@Torquato2000], and [*(ii)*]{}_the_existence of correlations between the particle_positions_determining_the properties of random packings.
In_previous theories, these correlations have been_neglected |
|\psi^* \rangle}$ can also be expressed as
\[eq:PureGamma\] $$\begin{aligned}
{|\psi \rangle} &= a{|\gamma_1 \rangle} + ib{|\gamma_2 \rangle},\\
{|\psi^* \rangle} &= a{|\gamma_1 \rangle} - ib{|\gamma_2 \rangle},\end{aligned}$$
where $a$ and $b$ are real numbers with $a^2+b^2=1$, and ${|\gamma_i \rangle}$ are real states. Equipped with these tools, we are now ready to prove the following proposition.
\[prop:PureGeneric\] For any pure state ${|\psi \rangle}$ there exists a real orthogonal matrix $O$ such that $$O{|\psi \rangle} = \sqrt{\frac{1+|{ \langle \psi^*| \psi \rangle}|}{2}}{|0 \rangle} + i \sqrt{\frac{1-|{ \langle \psi^*| \psi \rangle}|}{2}}{|1 \rangle}. \label{eq:PureGeneric}$$
In the first step, note that for any two real states ${|\gamma_1 \rangle}$ and ${|\gamma_2 \rangle}$ there exists a real orthogonal matrix $O$ such that $$\begin{aligned}
O{|\gamma_1 \rangle} &= {|0 \rangle}, \\
O{|\gamma_2 \rangle} &= \cos \theta {|0 \rangle} + \sin \theta {|1 \rangle},\end{aligned}$$ where $\cos \theta = { \langle \gamma_1| \gamma_2 \rangle}$. Applying $O$ to the state ${|\psi \rangle}$ gives us $$O{|\psi \rangle} = (a+ib\cos \theta){|0 \rangle} + ib\sin\theta {|1 \rangle}.$$ Since the state $O{|\psi \rangle}$ is effectively a single-qubit state, we can associate a Bloch vector $\boldsymbol{r}$ with it, with coordinates $$\begin{aligned}
r_x = b^2 \sin (2\theta), \nonumber \\
r_y = 2ab \sin (\theta), \\
r_z = a^2 + b^2 \cos (2\theta). \nonumber\end{aligned}$$
| |\psi^ * \rangle}$ can also be expressed as
\[eq: PureGamma\ ] $ $ \begin{aligned }
{ |\psi \rangle } & = a{|\gamma_1 \rangle } + ib{|\gamma_2 \rangle},\\
{ |\psi^ * \rangle } & = a{|\gamma_1 \rangle } - ib{|\gamma_2 \rangle},\end{aligned}$$
where $ a$ and $ b$ are real numbers with $ a^2+b^2=1 $, and $ { |\gamma_i \rangle}$ are veridical state. Equipped with these tools, we are now quick to prove the following proposition.
\[prop: PureGeneric\ ] For any arrant state $ { |\psi \rangle}$ there exists a actual orthogonal matrix $ O$ such that $ $ O{|\psi \rangle } = \sqrt{\frac{1+| { \langle \psi^*| \psi \rangle}|}{2}}{|0 \rangle } + i \sqrt{\frac{1-| { \langle \psi^*| \psi \rangle}|}{2}}{|1 \rangle }. \label{eq: PureGeneric}$$
In the first footfall, note that for any two real state $ { |\gamma_1 \rangle}$ and $ { |\gamma_2 \rangle}$ there exist a real orthogonal matrix $ O$ such that $ $ \begin{aligned }
O{|\gamma_1 \rangle } & = { |0 \rangle }, \\
O{|\gamma_2 \rangle } & = \cos \theta { |0 \rangle } + \sin \theta { |1 \rangle},\end{aligned}$$ where $ \cos \theta = { \langle \gamma_1| \gamma_2 \rangle}$. Applying $ O$ to the state $ { |\psi \rangle}$ gives us $ $ O{|\psi \rangle } = (a+ib\cos \theta){|0 \rangle } + ib\sin\theta { |1 \rangle}.$$ Since the state of matter $ O{|\psi \rangle}$ is effectively a single - qubit state, we can consociate a Bloch vector $ \boldsymbol{r}$ with it, with coordinates $ $ \begin{aligned }
r_x = b^2 \sin (2\theta), \nonumber \\
r_y = 2ab \sin (\theta), \\
r_z = a^2 + b^2 \cos (2\theta). \nonumber\end{aligned}$$ | |\psi^* \rangle}$ can also be exprtssed as
\[eq:PureGamma\] $$\begii{alignes}
{|\psi \rangle} &= a{|\gamma_1 \rangle} + ib{|\jammq_2 \rantle},\\
{|\psi^* \rangle} &= a{|\gaoma_1 \ranglv} - ib{|\gammq_2 \raigle},\end{aligned}$$
whxde $a$ and $b$ ars rean numbers with $s^2+b^2=1$, and ${|\gamka_i \rangle}$ are rdap states. Equipped with these tools, re are moa ready to prode tnq fomlowing proposition.
\[prop:PureGeneric\] For anj pure state ${|\psi \tangle}$ there exists a real ortjogonal matrix $O$ skch that $$O{|\pwi \rwbgle} = \sqrt{\frxc{1+|{ \langle \ksn^*| \psi \ranglg}|}{2}}{|0 \rangle} + i \sqrt{\frac{1-|{ \langle \psi^*| \osi \rcngle}|}{2}}{|1 \ranglg}. \mahgl{eq:PureGenecic}$$
In nhe first stei, note dhat fot any two real stetes ${|\gamma_1 \rangle}$ and ${|\gamka_2 \rangle}$ there exysts a redl orthogonal matriz $I$ sucv thdt $$\bdtin{xlifnxd}
O{|\galma_1 \rangle} &= {|0 \dangle}, \\
I{|\gamma_2 \rangle} &= \cos \uheeq {|0 \rangle} + \sih \thetw {|1 \rangle},\end{aligned}$$ where $\cos \theta = { \lancle \gamma_1| \gamma_2 \rangle}$. Appoying $O$ to the state ${|\ksi \rangle}$ gives us $$O{|\psi \rangle} = (a+ib\cos \theta){|0 \rangle} + ib\sin\dheta {|1 \raunjd}.$$ Slnce the state $O{|\psi \rangle}$ is effectively a syhgke-aubit state, we can associstf s Bloch vector $\boldsvjbkl{r}$ with it, with clordinajes $$\betin{aligneq}
r_x = b^2 \sin (2\theta), \nonumber \\
r_y = 2ab \siu (\tyeta), \\
r_z = a^2 + b^2 \eos (2\theta). \nouumber\gnd{alibned}$$
| |\psi^* \rangle}$ can also be expressed as {|\psi &= a{|\gamma_1 + ib{|\gamma_2 \rangle},\\ - \rangle},\end{aligned}$$ where $a$ $b$ are real with $a^2+b^2=1$, and ${|\gamma_i \rangle}$ are states. Equipped with these tools, we are now ready to prove the following \[prop:PureGeneric\] For any pure state ${|\psi \rangle}$ there exists a real orthogonal matrix such $$O{|\psi = \langle \psi^*| \psi \rangle}|}{2}}{|0 \rangle} + i \sqrt{\frac{1-|{ \langle \psi^*| \psi \rangle}|}{2}}{|1 \rangle}. \label{eq:PureGeneric}$$ In the step, note that for any two real states \rangle}$ and ${|\gamma_2 \rangle}$ exists a real orthogonal matrix such $$\begin{aligned} O{|\gamma_1 &= \rangle}, O{|\gamma_2 \rangle} &= \theta {|0 \rangle} + \sin \theta {|1 \rangle},\end{aligned}$$ where $\cos \theta = { \langle \gamma_1| \gamma_2 \rangle}$. $O$ to ${|\psi \rangle}$ us \rangle} (a+ib\cos \theta){|0 \rangle} {|1 \rangle}.$$ Since the state $O{|\psi a single-qubit state, we can associate a Bloch $\boldsymbol{r}$ with with coordinates $$\begin{aligned} r_x = b^2 (2\theta), \nonumber \\ r_y = 2ab \sin (\theta), r_z = a^2 + b^2 \cos (2\theta). \nonumber\end{aligned}$$ | |\psi^* \rangle}$ can also be expressEd as
\[eq:Puregamma\] $$\BegIn{aLiGned}
{|\Psi \rAngle} &= a{|\gamma_1 \ranGLe} + ib{|\Gamma_2 \rangle},\\
{|\psi^* \rangle} &= a{|\gAmma_1 \rAnGLe} - ib{|\GAmMa_2 \ranGle},\end{aLIgNED}$$
whErE $a$ And $B$ aRE rEal nuMbeRs with $a^2+B^2=1$, and ${|\gamma_i \RanGlE}$ are real statES. EQuipped witH thEse tools, we arE noW ready To ProVE the fOllOwing ProposITion.
\[prOp:PureGenErIC\] For anY Pure staTE ${|\PsI \ranGle}$ there exists a reAL oRThogonal matrix $o$ such tHaT $$o{|\pSI \RanGle} = \Sqrt{\frac{1+|{ \laNgLe \psi^*| \PSi \ranglE}|}{2}}{|0 \RaNGLE} + i \sQRt{\frac{1-|{ \langle \pSi^*| \psi \rangle}|}{2}}{|1 \RAngLe}. \labeL{eQ:PuREGenerIc}$$
In tHe FIrsT step, note thAt foR any two reAl statES ${|\gamma_1 \rANgle}$ and ${|\Gamma_2 \rAngLe}$ tHere EXiStS a rEaL OrtHOgOnaL MatRix $O$ such ThAt $$\Begin{AligNED}
o{|\GammA_1 \raNgle} &= {|0 \RanglE}, \\
O{|\gamma_2 \rangle} &= \Cos \ThetA {|0 \RanGle} + \siN \thetA {|1 \ranGlE},\end{aLigned}$$ Where $\CoS \theta = { \langle \gamMa_1| \gaMma_2 \rangle}$. appLyIng $o$ tO the sTAte ${|\psi \RanGle}$ Gives us $$o{|\psi \ranGLe} = (a+Ib\COS \ThEta){|0 \rangle} + ib\sin\thetA {|1 \rANGlE}.$$ Since thE state $o{|\PsI \rANgle}$ is efFeCtiVely A SIngle-QubiT StAte, we can AssociATe A BLoch vecToR $\boldsYmBol{R}$ wiTh it, wITh coOrdinaTes $$\begin{AlignED}
r_x = b^2 \sin (2\theta), \noNUmber \\
r_y = 2ab \sin (\tHEtA), \\
R_Z = a^2 + B^2 \Cos (2\tHetA). \nonumber\enD{aliGNed}$$
| |\psi^* \rangle}$ can also be expres sed a s
\[e q: Pure Gamm a\] $$\begin{a l igne d}
{|\psi \rangle} &= a { |\ga m ma _1 \r angle}+ i b { |\g am ma _2\r a ng le},\ \
{|\p si^* \rang le} & = a{|\gamma_ 1 \ rangle} -ib{ |\gamma_2 \r ang le},\e nd {al i gned} $$
wher e $a$a nd $b$ are real n u mbersw ith $a^ 2 + b^ 2=1$ , and ${|\gamma_i \r a ngle}$ are rea l stat es . E q u ipp edwith these t ools, we aren ow r e ady to prove thefollowing p r opo sition .
\[ p rop:Pu reGen er i c\] For any pu re s tate ${|\ psi \r a ngle}$t here ex ists a re alorth o go na l m at r ix$ O$ su c h t hat $$O{ |\ ps i \ra ngle } = \sqr t{\ frac {1+|{ \langle \psi ^*| \ps i \r angle }|}{2 }}{| 0\rang le} +i \sq rt {\frac{1-|{ \la ngle \psi^*|\ps i\ra ng le}|} { 2}}{|1 \r ang le}. \l abel{eq : Pur eG e n e ri c}$$
In the first s t e p, note th at for an yt wo realst ate s ${ | \ gamma _1 \ r an gle}$ an d ${|\ g am ma _2 \ran gl e}$ th er e e xis ts ar ealorthog onal mat rix $ O $ such that $$ \ begin{aligned }
O { |\ga mma _1 \rangle} &={ |0 \ rang l e} , \ \
O{|\g am m a_ 2 \rangle} &= \cos \ th eta {| 0 \ra ngle} + \sin\theta {|1 \ r angle},\ end{ a li g ned}$$ where $ \cos\theta = { \langle\gamm a_1| \ga mma_2 \ra n g le}$. Ap ply ing $O $ t o th e state ${|\p s i \ra ng le}$ gi ves us $$O {|\ psi \r ang le } = (a+ib \cos \th et a) {| 0\ra ngle} + ib\sin \t het a{|1 \ran g le}.$$ Sinc e th est a te$O{|\ps i \ r a ngle }$ i s ef fec ti velya si n gle -qubitstate, we ca n ass oc ia te a Bl och vector $\ bo ldsymbol{r }$ wi th it, w ith coor dinates $$\begin{aligne d }
r _x= b^2 \si n (2\thet a), \nonu mbe r \\
r_y= 2ab \ sin ( \thet a ) ,\\ r_z = a^ 2 + b ^2 \c os (2\ theta). \nonumber\end{ali g ned }$$
| |\psi^* \rangle}$_can also_be expressed as
\[eq:PureGamma\] $$\begin{aligned}
_ __{|\psi \rangle}__ &= a{|\gamma_1_\rangle} + ib{|\gamma_2_\rangle},\\
_{|\psi^* \rangle} &=_a{|\gamma_1_\rangle} - ib{|\gamma_2 \rangle},\end{aligned}$$
where $a$ and $b$ are real numbers with $a^2+b^2=1$, and ${|\gamma_i_\rangle}$_are real_states._Equipped_with these tools, we are_now ready to prove the_following proposition.
\[prop:PureGeneric\]_For any pure state ${|\psi \rangle}$ there exists_a_real orthogonal matrix_$O$ such that $$O{|\psi \rangle} = \sqrt{\frac{1+|{ \langle \psi^*|_\psi \rangle}|}{2}}{|0 \rangle} + i \sqrt{\frac{1-|{_\langle \psi^*| \psi_\rangle}|}{2}}{|1_\rangle}._\label{eq:PureGeneric}$$
In the first step,_note that for any two real_states ${|\gamma_1 \rangle}$ and ${|\gamma_2 \rangle}$_there exists a real orthogonal matrix $O$_such that $$\begin{aligned}
_O{|\gamma_1 \rangle} &= {|0 \rangle},_\\
_ O{|\gamma_2 \rangle} &=_\cos \theta {|0_\rangle} +_\sin \theta {|1_\rangle},\end{aligned}$$ where $\cos \theta = {_\langle \gamma_1| \gamma_2_\rangle}$. Applying $O$ to the state_${|\psi_\rangle}$ gives us_$$O{|\psi_\rangle}_= (a+ib\cos_\theta){|0 \rangle} +_ib\sin\theta_{|1 \rangle}.$$_Since_the state $O{|\psi \rangle}$ is effectively_a_single-qubit state, we can associate a Bloch_vector $\boldsymbol{r}$ with it,_with_coordinates $$\begin{aligned}
_ r_x = b^2 \sin_(2\theta), \nonumber \\
_r_y =_2ab \sin_(\theta), \\
r_z = a^2 + b^2 \cos_(2\theta). \nonumber\end{aligned}$$
|
{eq:staticconst8}.\end{aligned}$$
Since there are no constraints on the fleet size and furthermore prices that control the demand are decision variables, a feasible solution to the above optimization problem always exists. Moreover, the optimal solution to specifies an equilibrium of the duopoly.
\[prop:unique\] The firms are in an equilibrium when their routing, charging, and symmetric pricing strategy follows the solution of.
The proof can be found in Appendix \[app:unique\]. Accordingly, there exists a duopoly equilibrium characterized as the optimal solution of, in which the firms set identical prices. The optimal solution to is however not necessarily unique and there can be many solutions yielding the same profits. For instance, if $p_i=p_j,\forall i,j\in{\cal M}$, then the optimal charging strategy is not unique. We let $\{\ell_{ij}^d\}_{i,j\in{\cal M}}$ to be the optimal prices determined as an optimal solution of and say that the firms are in a symmetric duopoly equilibrium as long as $\ell_{ij}^1=\ell_{ij}^2=\ell_{ij}^d, \forall i,j\in{\cal M}$. Furthermore, in the next proposition, we state that if both firms serve all OD pairs, equilibrium prices can not be asymmetric.
\[prop:equilibrium\] There exists no asymmetric equilibrium prices, in which both firms serve nonzero demand for all OD pairs with nonzero potential demand.
The proof is provided in Appendix \[app:equilibrium\]. As we have identified that the duopoly can only be in a symmetric equilibrium, we analyze the effects of competition in state of a symmetric equilibrium.
The next set of results characterize the effects of competition on the ride prices, the operators’ profits, the total societal ride demand served, and the consumer surplus. In the first result, we provide lower and upper bounds on the price reduction the customers will see with the introduction of the second firm and moving from a monopoly to a symmetric duopoly equilibrium.
\[prop:duomarginalprices\] Let $\overline{\lambda}_{ij}$ be defined as in Proposition \[prop:monomarginalprices\]. Define $$\Delta_1(\lambda_{ij}){\vcentcolon=}4{\ell_{\max}}^2+(2\lambda_{ij}+(15\sigma-3)\ell_{\max})(2\lambda_{ij}+(1-\sigma)\ell_{\ | { eq: staticconst8}.\end{aligned}$$
Since there are no constraints on the fleet size and furthermore prices that control the requirement are decisiveness variables, a feasible solution to the above optimization problem constantly exist. Moreover, the optimal solution to specifies an balance of the duopoly.
\[prop: unique\ ] The firm are in an equilibrium when their routing, charging, and symmetric pricing strategy follows the solution of.
The validation can be found in Appendix \[app: unique\ ]. Accordingly, there exists a duopoly equilibrium qualify as the optimal solution of, in which the firms set identical prices. The optimal solution to is however not inevitably unique and there can be many solutions yielding the like profits. For instance, if $ p_i = p_j,\forall i, j\in{\cal M}$, then the optimal charging strategy is not alone. We let $ \{\ell_{ij}^d\}_{i, j\in{\cal M}}$ to be the optimal prices determined as an optimal solution of and allege that the firms are in a symmetric duopoly equilibrium as long as $ \ell_{ij}^1=\ell_{ij}^2=\ell_{ij}^d, \forall i, j\in{\cal M}$. Furthermore, in the next suggestion, we state that if both firms serve all OD pairs, equilibrium prices cannot be asymmetric.
\[prop: equilibrium\ ] There exists no asymmetric equilibrium monetary value, in which both firms serve nonzero need for all OD pair with nonzero potential demand.
The proof is provide in Appendix \[app: equilibrium\ ]. As we have identified that the duopoly can only be in a symmetric equilibrium, we analyze the effect of competition in state of a symmetric equilibrium.
The next bent of results characterize the effects of competition on the drive prices, the hustler ’ net income, the total societal ride demand serve, and the consumer surplus. In the first result, we leave lower and upper bounds on the price reduction the customer will see with the introduction of the second firm and moving from a monopoly to a symmetric duopoly equilibrium.
\[prop: duomarginalprices\ ] permit $ \overline{\lambda}_{ij}$ be defined as in Proposition \[prop: monomarginalprices\ ]. specify $ $ \Delta_1(\lambda_{ij}){\vcentcolon=}4{\ell_{\max}}^2+(2\lambda_{ij}+(15\sigma-3)\ell_{\max})(2\lambda_{ij}+(1-\sigma)\ell_{\ | {eq:shaticconst8}.\end{aligned}$$
Sinct there are no couwtrainvs on tge fleet size and furthermore prices tyat cintrol the demand are aecision nariables, a ftasible solution vk the anjve kitimivavion problem aleays existv. Moreover, the ootnmal solution to specifies an equilifrium og hhe duopoly.
\[prok:uniqle\] The firms are in an equilibrium when fheir rmuting, chargimg, and symmetric pricing shratfgy follows the sopution of.
Thg prjif can be fojnd in Apptnbix \[app:uniqug\]. Accordingly, there exists a duoooly zquilibrium cyarwwterized as the jptimal solubpon of, hn whicn the firms seb ideiticql prices. The optimal solution to is horever not nzcessarily unique and tyere wan te mxby rolhtmona yielfinj the same lrofits. For instance, if $p_i=p_j,\fotajo i,j\in{\cal M}$, tgen thq jptimal charging strategy is not unique. We let $\{\ell_{ij}^d\}_{i,j\in{\cal M}}$ to be the optimal priced determiged as an optimal solution of and say that the fisms ace in a rtmletric duopoly equilibrium as long as $\ell_{ij}^1=\elj_{jj}^2=\tll_{pj}^d, \forall i,j\in{\caj M}$. Furtherkoge, yn the next ptopositnkn, we state that if hoth fitms seeve all OQ paors, equilibrium prices can bot be asymmvtrix.
\[prop:equilibrium\] Tkere exists uo asykmetroc equilibrium prices, iu whicg both firmd serve nkvzero demand for alk MD pairs with nonzero potegtial demend.
Thz proof ks ptovided in Appendlx \[app:cxuilibrium\]. As we hwve ibentixied that hhe duopoly can only be in a symmetric equilinrhum, we analvze thc effects of coipetition in sjate of a symmdtric equimibrium.
Vhe next set of results cvwracterize tie effectf of competitiov on the ride ptices, the operators’ profits, the total sozjetal ride demaub werved, and the vonrumqr snrpluf. In the firsd rerulg, we pfovide lowev avd ulper bounds on the psice reduction the cusyoiers wilo see wieh the introdiction of the secojd ficm and movimg srom a monopoly to a symmetric duopoly fqullibrium.
\[prop:dtomavginwlprices\] Lzt $\overline{\lambda}_{ij}$ be defined as in Pro'osition \[prop:monomarginaoprices\]. Define $$\Deltc_1(\lsmbda_{ij}){\vcenvcolon=}4{\qll_{\max}}^2+(2\lamtda_{ij}+(15\sigma-3)\ell_{\max})(2\lambea_{ij}+(1-\sigma)\ell_{\ | {eq:staticconst8}.\end{aligned}$$ Since there are no constraints on size furthermore prices control the demand solution the above optimization always exists. Moreover, optimal solution to specifies an equilibrium the duopoly. \[prop:unique\] The firms are in an equilibrium when their routing, charging, symmetric pricing strategy follows the solution of. The proof can be found in \[app:unique\]. there a equilibrium characterized as the optimal solution of, in which the firms set identical prices. The optimal to is however not necessarily unique and there be many solutions yielding same profits. For instance, if i,j\in{\cal then the charging is unique. We let M}}$ to be the optimal prices determined as an optimal solution of and say that the firms in a equilibrium as as \forall M}$. Furthermore, in proposition, we state that if both OD pairs, equilibrium prices can not be asymmetric. There exists asymmetric equilibrium prices, in which both serve nonzero demand for all OD pairs with potential demand. The proof is provided in Appendix \[app:equilibrium\]. As we have identified that the only be in a equilibrium, we analyze effects competition state a symmetric The next set of results characterize the effects of competition on ride prices, the operators’ profits, the total societal ride demand the surplus. In the result, we provide lower upper on the price reduction will with the firm moving from a monopoly a symmetric duopoly equilibrium. \[prop:duomarginalprices\] $\overline{\lambda}_{ij}$ be defined as | {eq:staticconst8}.\end{aligned}$$
SiNce there arE no coNstRaiNtS on tHe flEet size and furtHErmoRe prices that control the DemanD aRE decISiOn varIables, a FEaSIBle SoLuTioN tO ThE abovE opTimizatIon problem AlwAyS exists. MoreoVEr, The optimal SolUtion to speciFieS an equIlIbrIUm of tHe dUopolY.
\[prop:uNIque\] ThE firms are In AN equilIBrium whEN ThEir rOuting, charging, and SYmMEtric pricing stRategy FoLLoWS The SolUtion of.
The PrOof caN Be found IN APPENdiX \[App:unique\]. AccoRdingly, therE ExiSts a duOpOly EQuilibRium cHaRActErized as the OptiMal solutiOn of, in WHich the FIrms set IdentiCal PriCes. THE oPtImaL sOLutIOn To iS HowEver not nEcEsSarilY uniQUE ANd thEre Can bE many Solutions yielDinG the SAme ProfiTs. For InstAnCe, if $p_I=p_j,\forAll i,j\In{\Cal M}$, then the optiMal cHarging stRatEgY is NoT uniqUE. We let $\{\Ell_{Ij}^d\}_{I,j\in{\cal m}}$ to be thE OptImAL PRiCes determined as an oPtIMAl Solution Of and sAY tHaT The firms ArE in A symMETric dUopoLY eQuilibriUm as loNG aS $\eLl_{ij}^1=\ell_{Ij}^2=\Ell_{ij}^d, \FoRalL i,j\In{\cal m}$. furtHermorE, in the neXt proPOsition, we state THat if both firmS SeRVE aLL OD pAirS, equilibriuM priCEs caN not BE aSymMEtric.
\[Prop:eQuILiBRium\] There exists no asYmMetric EquilIbrium prices, iN which both FIRMs serve nOnzeRO dEMand for all OD paIrs wiTh nonzero pOTential dEmand.
the proof Is provideD IN AppendiX \[apP:eqUilIbrIUM\]. AS we have identiFIEd thAt The duopOly Can only Be iN a sYmmEtrIc EquilibriUm, we analYzE tHe EfFecTs of cOMpetitioN iN stAtE of A symmETric eqUilibRium.
thE nEXt sEt of resULtS CHaraCtErIze tHe eFfEcts oF comPEtiTion on tHe ride priCes, THe opErAtOrs’ profIts, the total soCiEtal ride deMaNd sErved, aND The consuMer surplus. In the first resULt, we proVidE loweR and Upper bounDs oN the prIce REductiOn the cUstomErS wiLL See wiTH ThE inTrOduction of THE seCond fIrM and Moving fRom a monopoly to a symMEtrIc duopoly equiLibRium.
\[PROp:DuoMArGInaLpRIceS\] lEt $\overline{\lambdA}_{ij}$ be definEd AS iN PropositiON \[prOp:MonomarGinalprIces\]. DEFine $$\DelTa_1(\lambda_{iJ}){\vcentcolOn=}4{\Ell_{\mAX}}^2+(2\LamBda_{ij}+(15\sigma-3)\Ell_{\max})(2\laMbda_{ij}+(1-\sigMA)\ell_{\ | {eq:staticconst8}.\end{ali gned}$$
S incethe rear e no con straints on th e fle et size and furthermor e pri ce s tha t c ontro l the d e ma n d ar ede cis io n v ariab les , a fea sible solu tio nto the above op timization pr oblem always ex ists.Mo reo v er, t heoptim al sol u tion t o specifi es an equ i librium o ftheduopoly.
\[prop: u ni q ue\] The firms are i na ne q uil ibr ium when t he ir ro u ting, c h ar g i n g,a nd symmetricpricing str a teg y foll ow s t h e solu tionof .
T he proof ca n be found in Appen d ix \[ap p :unique \]. Ac cor din gly, th er e e xi s tsa d uop o lyequilibr iu mchara cter i z e d asthe opt imalsolution of,inwhic h th e fir ms se t id en tical price s. Th eoptimal solutio n to is howev erno t n ec essar i ly uni que an d there can be man ys o l ut ions yielding thesa m e p rofits.For in s ta nc e , if $p_ i= p_j ,\fo r a ll i, j\in { \c al M}$,then t h eop timal c ha rgingst rat egy is n o t un ique.We let $ \{\el l _{ij}^d\}_{i,j \ in{\cal M}}$t ob e t h e op tim al prices d eter m ined asa nopt i mal s oluti on of and say that the fi rm s arein asymmetric duo poly equil i b r ium as l onga s$ \ell_{ij}^1=\e ll_{i j}^2=\ell_ { ij}^d, \ foral l i,j\in {\cal M}$ . Furtherm ore , i n t hen e xt proposition, w e st at e thatifboth fi rms se rve al lOD pairs, equilib ri um p ri ces cann ot be as ym met ri c.
\[pr o p:equi libri um\] T he r e e xists n o a s y mmet ri cequi lib ri um pr ices , in whichboth firm s s e rveno nz ero dem and for all O Dpairs with n onz ero po t e ntial de mand.
The proof is pro v ided in Ap pendi x \[ app:equil ibr ium\]. As we hav e iden tifie dtha t the d u o po lyca n only bei n asymme tr ic e quilibr ium, we analyze th e ef fects of comp eti tion i nsta t eo f a s y mme t r ic equilibrium.
The next s e tof results cha ra cterize the ef fects of comp etition o n the rid epric e s , t he operato rs’ prof its, thet otals oc ietal ri de dem an d s erved , andt heconsu mer su rp lus. I n the f irst res ult, we provide lower a nd upp er bo und s on thepri c e r eductionthecustomerswil l s ee wi tht he in trod u ct ion of th e se c ond firma nd mo v i ng from a mon o p o lyto asym m etricduop oly equilibrium.\[prop:duomarg inal p r ice s\] Let$\ overline{\lamb da} _{ i j }$ be de fi ned as in P ropositi on \[pro p:mono margin alprice s \ ]. Define $$\ Del ta_1(\lam bda _{ i j}){\vc en tc o lon=}4 {\el l_ {\max} }^2+(2 \ lamb d a _{ij}+(15\sigma- 3)\el l _ {\max } )(2 \lamb da _{ij}+( 1 -\si gma)\ell_{ \ | {eq:staticconst8}.\end{aligned}$$
Since there_are no_constraints on the fleet_size and_furthermore_prices that_control_the demand are_decision variables, a_feasible solution to the_above optimization problem_always_exists. Moreover, the optimal solution to specifies an equilibrium of the duopoly.
\[prop:unique\] The firms_are_in an_equilibrium_when_their routing, charging, and symmetric_pricing strategy follows the solution_of.
The proof_can be found in Appendix \[app:unique\]. Accordingly, there exists_a_duopoly equilibrium characterized_as the optimal solution of, in which the firms_set identical prices. The optimal solution_to is however_not_necessarily_unique and there can_be many solutions yielding the same_profits. For instance, if $p_i=p_j,\forall i,j\in{\cal_M}$, then the optimal charging strategy is_not unique. We let $\{\ell_{ij}^d\}_{i,j\in{\cal M}}$_to be the optimal prices_determined as_an optimal solution of and_say that the_firms are_in a symmetric_duopoly equilibrium as long as $\ell_{ij}^1=\ell_{ij}^2=\ell_{ij}^d,_\forall i,j\in{\cal M}$._Furthermore, in the next proposition, we_state_that if both_firms_serve_all OD_pairs, equilibrium prices_can_not be_asymmetric.
\[prop:equilibrium\]_There exists no asymmetric equilibrium prices,_in_which both firms serve nonzero demand for_all OD pairs with_nonzero_potential demand.
The proof is_provided in Appendix \[app:equilibrium\]. As we_have identified that the duopoly can_only be_in a_symmetric equilibrium, we analyze the effects of competition in state of_a symmetric equilibrium.
The next set of_results characterize the effects_of competition_on_the ride prices,_the_operators’ profits,_the total societal ride demand served, and_the consumer_surplus. In the first result, we_provide lower and upper_bounds_on the price reduction the customers_will see with the introduction of_the second firm and moving_from_a_monopoly to a symmetric duopoly_equilibrium.
\[prop:duomarginalprices\] Let $\overline{\lambda}_{ij}$ be defined as_in Proposition \[prop:monomarginalprices\]. Define_$$\Delta_1(\lambda_{ij}){\vcentcolon=}4{\ell_{\max}}^2+(2\lambda_{ij}+(15\sigma-3)\ell_{\max})(2\lambda_{ij}+(1-\sigma)\ell_{\ |
mathbf P_k} (\bm \psi_k, \bm \tau_k)$ is a tall matrix with asymptotically mutually-orthogonal columns of unit length, $$\begin{aligned}
{\mathbf R}_k^U =
\bigg( \frac{1}{\sqrt{MP}}
{\mathbf P_k} (\bm \psi_k, \bm \tau_k) \bigg)
\bigg( MP \bm \Lambda_k \bigg)
\bigg( \frac{1}{\sqrt{MP}}
{\mathbf P}_k^H (\bm \psi_k, \bm \tau_k) \bigg)
\label{Rkue}\end{aligned}$$ provides a good approximation of eigenvalue decomposition.
With the aid of the *AoA-delay reciprocity* [@SFW-TSP; @SFW-CM], downlink channel covariance matrices can also be reconstructed via the physical parameters of uplink channels, which will be discussed in Section \[sec:dl\].
Initial Uplink Channel Parameter Extraction {#sec:initial}
===========================================
At the very beginning, all users stay in the dark as far as the BS is concerned and thus orthogonal trainings have to be applied to avoid the inter-user interference and pilot contamination at the BS. In this case, we operate with the frequency orthogonality among difference users [@OFDMA-TWC; @SFW-TSP] by setting $\mathcal P_k \cap \mathcal P_r = \varnothing, $ $\forall k\neq r$, at this very beginning phase. During this phase, the initial AoA, time delay, and complex gain of each path are estimated for all users and we call this phase *initial parameter extraction*.
In this section, we introduce a parameter extraction algorithm for this stage, which suffices for the subsequent multi-user uplink and downlink channel estimations, as will be shown in Sections V and VI.
Problem Formulation
-------------------
For the $k$th user, the uplink pilots transmission process is given in. Substituting into, we have $$\begin{aligned}
\mathbf y_k =
\mathbf W^H {\mathbf P}_k(\bm \psi_k, \bm \tau_k) \bm \alpha_k + \mathbf n_k
.
\label{yk2}\end{aligned}$$ | mathbf P_k } (\bm \psi_k, \bm \tau_k)$ is a tall matrix with asymptotically mutually - orthogonal columns of unit duration, $ $ \begin{aligned }
{ \mathbf R}_k^U =
\bigg (\frac{1}{\sqrt{MP } }
{ \mathbf P_k } (\bm \psi_k, \bm \tau_k) \bigg)
\bigg (MP \bm \Lambda_k \bigg)
\bigg (\frac{1}{\sqrt{MP } }
{ \mathbf P}_k^H (\bm \psi_k, \bm \tau_k) \bigg)
\label{Rkue}\end{aligned}$$ provide a good approximation of eigenvalue decomposition.
With the care of the * AoA - delay reciprocity * [ @SFW - TSP; @SFW - CM ], downlink channel covariance matrices can besides be restore via the physical parameter of uplink channels, which will be discussed in Section \[sec: dl\ ].
Initial Uplink Channel Parameter Extraction { # sec: initial }
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
At the identical beginning, all users stay in the dark as far as the BS is implicated and thus orthogonal training have to be applied to debar the inter - user interference and pilot contamination at the boron. In this case, we operate with the frequency orthogonality among difference users [ @OFDMA - TWC; @SFW - TSP ] by set $ \mathcal P_k \cap \mathcal P_r = \varnothing, $ $ \forall k\neq r$, at this very beginning phase. During this phase, the initial AoA, time delay, and complex gain of each path are estimated for all drug user and we call this phase * initial parameter extraction *.
In this section, we introduce a parameter extraction algorithm for this stage, which suffices for the subsequent multi - user uplink and downlink channel estimations, as will be shown in Sections V and VI.
Problem Formulation
-------------------
For the $ k$th user, the uplink pilots transmission process is given in. substitute into, we have $ $ \begin{aligned }
\mathbf y_k =
\mathbf W^H { \mathbf P}_k(\bm \psi_k, \bm \tau_k) \bm \alpha_k + \mathbf n_k
.
\label{yk2}\end{aligned}$$ | matjbf P_k} (\bm \psi_k, \bm \tau_k)$ ir a tall matrix with avymptofically outually-orthogonal columns oh unut lebgth, $$\begin{aligned}
{\mathbw R}_k^U =
\bigh( \frac{1}{\sqet{MP}}
{\nqthbf P_k} (\bm \psi_k, \bm \tau_k) \nigg)
\bngj( MP \bm \Lambda_k \bigg)
\bigg( \xrac{1}{\sqrt{MP}}
{\mathtf P}_n^H (\bm \psi_k, \bm \tau_k) \bigg)
\label{Rkue}\end{wligned}$$ pgovides a good appgovimafpok of eigenvalue decomposition.
Witg the apd of the *AoA-delau reciprocity* [@SFW-TSP; @SFW-CM], dowjlink channel covagiance matruces xan also be feconstrucueb via the pgysical parameters of uplink chxnnelx, which wiol be giscussed ii Sectpon \[sec:dl\].
Initial Uplinn Channrl Parameter Ewtracvion {#sec:initial}
===========================================
At the verb beginning, all userf stay in tke dark as far as the BW is woncarnea ana tgux krthoglnam traininga have to bw applied to avoid uhe pmter-user intsrferegcq and pilot contamination at the BS. In uhis dase, we operate with thw frequency orthogonapity amond difference users [@OFDMA-TWC; @SFW-TSP] by setting $\matvcal '_k \cak \nathcxo O_r = \varnothing, $ $\forall k\neq r$, at this very bedjnmikg phase. During bhis phase, the iniyiwl WoA, time delai, and compmex gain of each pwth are estinated for all users and we call this phawe *initial pcraneter extraction*.
In this sectiun, wg intrpduce a parameter extraetion zlgorithm flr this sfxge, which sufficds nor the subsequent multi-user tplink anv dowulink chxnnek estiiations, as will be shown in Sectiojs V aud VI.
[roblem Fogmulation
-------------------
For the $k$th user, the u'kink pilots jratsmpssion prjcess is given in. Fubstituting iuto, we hcve $$\beein{aligned}
\jathbf b_k =
\mathbf W^H {\mathbf P}_k(\bm \[di_k, \bm \tau_k) \um \alpha_k + \marhbf n_k
.
\labeu{hk2}\end{aligned}$$ | mathbf P_k} (\bm \psi_k, \bm \tau_k)$ is matrix asymptotically mutually-orthogonal of unit length, \frac{1}{\sqrt{MP}} P_k} (\bm \psi_k, \tau_k) \bigg) \bigg( \bm \Lambda_k \bigg) \bigg( \frac{1}{\sqrt{MP}} {\mathbf (\bm \psi_k, \bm \tau_k) \bigg) \label{Rkue}\end{aligned}$$ provides a good approximation of eigenvalue decomposition. the aid of the *AoA-delay reciprocity* [@SFW-TSP; @SFW-CM], downlink channel covariance matrices can be via physical of uplink channels, which will be discussed in Section \[sec:dl\]. Initial Uplink Channel Parameter Extraction {#sec:initial} At the very beginning, all users stay in dark as far as BS is concerned and thus trainings to be to the interference and pilot at the BS. In this case, we operate with the frequency orthogonality among difference users [@OFDMA-TWC; @SFW-TSP] setting $\mathcal \mathcal P_r \varnothing, $\forall r$, at this phase. During this phase, the initial and complex gain of each path are estimated all users we call this phase *initial parameter In this section, we introduce a parameter extraction for this stage, which suffices for the subsequent multi-user uplink and downlink channel estimations, as shown in Sections V VI. Problem Formulation For $k$th the pilots transmission is given in. Substituting into, we have $$\begin{aligned} \mathbf y_k = W^H {\mathbf P}_k(\bm \psi_k, \bm \tau_k) \bm \alpha_k + \mathbf \label{yk2}\end{aligned}$$ | mathbf P_k} (\bm \psi_k, \bm \tau_k)$ is a taLl matrix wiTh asyMptOtiCaLly mUtuaLly-orthogonal cOLumnS of unit length, $$\begin{aligNed}
{\maThBF R}_k^U =
\BIgG( \frac{1}{\Sqrt{MP}}
{\mAThBF p_k} (\bM \pSi_K, \bm \TaU_K) \bIgg)
\biGg( Mp \bm \LambDa_k \bigg)
\bigG( \frAc{1}{\Sqrt{MP}}
{\mathbf p}_K^H (\Bm \psi_k, \bm \taU_k) \bIgg)
\label{Rkue}\End{AligneD}$$ pRovIDes a gOod ApproXimatiON of eigEnvalue deCoMPositiON.
With thE AId Of thE *AoA-delay reciprocITy* [@sfW-TSP; @SFW-CM], downLink chAnNEl COVarIanCe matrices CaN also BE reconsTRuCTED viA The physical paRameters of uPLinK channElS, whICh will Be disCuSSed In Section \[seC:dl\].
INitial UplInk ChaNNel ParaMEter ExtRactioN {#seC:inItiaL}
===========================================
at ThE veRy BEgiNNiNg, aLL usErs stay iN tHe Dark aS far AS THE BS iS coNcerNed anD thus orthogonAl tRainINgs Have tO be apPlieD tO avoiD the inTer-usEr Interference and PiloT contaminAtiOn At tHe bS. In tHIs case, We oPerAte with The freqUEncY oRTHOgOnality among differEnCE UsErs [@OFDMA-tWC; @SFW-tsP] By SEtting $\maThCal p_k \caP \MAthcaL P_r = \vARnOthing, $ $\foRall k\nEQ r$, At This verY bEginniNg PhaSe. DUring THis pHase, thE initial aoA, tiME delay, and complEX gain of each paTH aRE EsTImatEd fOr all users aNd we CAll tHis pHAsE *inITial pArameTeR ExTRaction*.
In this sectioN, wE introDuce a Parameter extrAction algoRITHm for thiS staGE, wHIch suffices for The suBsequent muLTi-user upLink aNd downliNk channel ESTimationS, as WilL be ShoWN In sections V and Vi.
pRoblEm formulaTioN
-------------------
For the $K$th UseR, thE upLiNk pilots tRansmissIoN pRoCeSs iS giveN In. SubstiTuTinG iNto, We havE $$\Begin{aLigneD}
\matHbF y_K =
\MatHbf W^H {\maTHbF p}_K(\bm \pSi_K, \bM \tau_K) \bm \AlPha_k + \mAthbF N_k
.
\lAbel{yk2}\eNd{aligned}$$ | mathbf P_k} (\bm \psi_k, \ bm \tau_k) $ isa t all m atri x wi th asymptotica l ly m utually-orthogonal col umnsof unit le ngth, $$\beg i n{ a l ign ed }{\m at h bf R}_k ^U=
\bigg ( \frac{1} {\s qr t{MP}}
{\mat h bf P_k} (\bm \p si_k, \bm \t au_ k) \bi gg )
\ b igg(MP\bm \ Lambda _ k \big g)
\bigg( \ f rac{1} { \sqrt{M P } }{\ma thbf P}_k^H (\bm\ ps i _k, \bm \tau_k ) \big g) \l a b el{ Rku e}\end{ali gn ed}$$ provide s a g o oda pproximationof eigenval u e d ecompo si tio n .
Wit h the a i d o f the *AoA- dela y recipro city*[ @SFW-TS P ; @SFW- CM], d own lin k ch a nn el co va r ian c emat r ice s can al so b e rec onst r u c t ed v iathephysi cal parameter s o f up l ink chan nels, whi ch will be di scuss ed in Section \[s ec:d l\].
Ini tia lUpl in k Cha n nel Pa ram ete r Extra ction { # sec :i n i t ia l}
=============== == = = == ======== ====== = == == = ==
At t he ve ry b e g innin g, a l lusers st ay int he d ark asfa r as t he BS is conc e rned and t hus orth ogona l trainings hav e to be applie d t o av o id t heinter-userinte r fere ncea nd pi l ot co ntami na t io n at the BS. In this c ase, w e ope rate with the frequency o r thogonal itya mo n g difference u sers[@OFDMA-TW C ; @SFW-T SP] b y settin g $\mathc a l P_k \ca p \ mat hca l P _ r = \varnothing, $ $\f or all k\n eqr$, atthi s v ery be gi nning pha se. Duri ng t hi spha se, t h e initia lAoA ,tim e del a y, and comp lexga in ofeach pa t ha r e es ti ma tedfor a ll us ersa ndwe call this pha se* init ia lparamet er extraction *.
In thisse cti on, we i ntroduce a parameter extraction algorit hmfor t hisstage, wh ich suffi ces for th e subs equen tmul t i -user u pl ink a nd downlin k cha nneles tima tions,as will be shown i n Se ctions V andVI.
Pr o b le m F o rm u lat io n
-- - - ---------------
For the$k $ th user, the upl in k pilot s trans missi o n proce ss is giv en in. Su bs titu t i nginto, we h ave $$\b egin{alig n ed}
\ m at hbf y _k=
\mat hb f W ^H {\ mathbf P}_ k(\bm \psi_ k, \bm \ tau_k )\bm \alp ha_k + \mathbf n_k
.
\l abel{y k2}\e nd{ aligned}$ $ | mathbf P_k}_(\bm \psi_k,_\bm \tau_k)$ is a_tall matrix_with_asymptotically mutually-orthogonal_columns_of unit length,_$$\begin{aligned}
{\mathbf R}_k^U =
\bigg(_\frac{1}{\sqrt{MP}}
{\mathbf P_k} (\bm \psi_k,_\bm \tau_k) \bigg)
\bigg(_MP_\bm \Lambda_k \bigg)
\bigg( \frac{1}{\sqrt{MP}}
{\mathbf P}_k^H (\bm \psi_k, \bm \tau_k) \bigg)
\label{Rkue}\end{aligned}$$ provides a good approximation_of_eigenvalue decomposition.
With_the_aid_of the *AoA-delay reciprocity* [@SFW-TSP;_@SFW-CM], downlink channel covariance matrices_can also_be reconstructed via the physical parameters of uplink_channels,_which will be_discussed in Section \[sec:dl\].
Initial Uplink Channel Parameter Extraction {#sec:initial}
===========================================
At the_very beginning, all users stay in_the dark as_far_as_the BS is concerned_and thus orthogonal trainings have to_be applied to avoid the inter-user_interference and pilot contamination at the BS._In this case, we operate with_the frequency orthogonality among difference_users [@OFDMA-TWC;_@SFW-TSP] by setting $\mathcal P_k_\cap \mathcal P_r_= \varnothing,_$ $\forall k\neq_r$, at this very beginning phase._During this phase,_the initial AoA, time delay, and_complex_gain of each_path_are_estimated for_all users and_we_call this_phase_*initial parameter extraction*.
In this section, we_introduce_a parameter extraction algorithm for this stage,_which suffices for the_subsequent_multi-user uplink and downlink_channel estimations, as will be_shown in Sections V and VI.
Problem Formulation
-------------------
For_the $k$th_user, the_uplink pilots transmission process is given in. Substituting into, we have_$$\begin{aligned}
\mathbf y_k =
\mathbf W^H {\mathbf P}_k(\bm_\psi_k, \bm \tau_k) \bm_\alpha_k +_\mathbf_n_k
.
\label{yk2}\end{aligned}$$ |
1+m};q)_{k}}(cd/abz)^{k}.\end{aligned}$$
From nonterminating $_8\phi_7$ to $_6\psi_6$
============================================
In this section, we give a semi-finite form of Bailey’s $_6\psi_6$ summation formula by using Bailey’s $3$-term transformation formula for a nonterminating very-well-poised $_8\phi_7$ series [@GR Eq. (2.11.1)]: $$\begin{aligned}
\label{Watson2}
\lefteqn{_{8}\phi_7\left[
\begin{array}{c}
a, qa^{1\over 2},-qa^{1\over 2},b,c, d, e, f\\
a^{1\over 2},-a^{1\over 2},aq/b,aq/c,aq/d,aq/e,aq/f
\end{array};q, \frac{a^2q^2}{bcdef}
\right] }\nonumber\\
&& \,
=\frac{(aq,aq/de,aq/df,aq/ef,eq/c,fq/c,b/a,bef/a;q)_{\infty}}
{(aq/d,aq/e,aq/f,aq/def,q/c,efq/c,be/a,bf/a;q)_{\infty}}\nonumber\\
&& \qquad \cdot_{8}\phi_7\left[
\begin{array}{c}
ef/c, q(ef/c)^{1\over 2},-q(ef/c)^{1\over 2},aq/bc,aq/cd, ef/a, e, f\\
(ef/c)^{1\over 2},-(ef/c)^{1\over 2}, bef/a,def/a,aq/c,fq/c,eq/c
\end{array};q, \frac{bd}{a}
\right] \nonumber\\
&&\qquad +\frac{b}{a}\frac{(aq,bq/a,bq/c,bq/d,bq/e,bq/f,d,e,f;q)_{\infty}}
{(aq/b,aq/c | 1+m};q)_{k}}(cd / abz)^{k}.\end{aligned}$$
From nonterminating $ _ 8\phi_7 $ to $ _ 6\psi_6 $
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
In this section, we give a semi - finite form of Bailey ’s $ _ 6\psi_6 $ summation rule by use Bailey ’s $ 3$-term transformation formula for a nonterminating very - well - poised $ _ 8\phi_7 $ series [ @GR Eq. (2.11.1) ]: $ $ \begin{aligned }
\label{Watson2 }
\lefteqn{_{8}\phi_7\left [
\begin{array}{c }
a, qa^{1\over 2},-qa^{1\over 2},b, c, d, einsteinium, f\\
a^{1\over 2},-a^{1\over 2},aq / b, aq / c, aq / d, aq / e, aq / f
\end{array};q, \frac{a^2q^2}{bcdef }
\right ] } \nonumber\\
& & \,
= \frac{(aq, aq / de, aq / df, aq / ef, eq / c, fq / c, b / a, bef / a;q)_{\infty } }
{ (aq / d, aq / e, aq / f, aq / def, q / c, efq / c, be / a, bf / a;q)_{\infty}}\nonumber\\
& & \qquad \cdot_{8}\phi_7\left [
\begin{array}{c }
ef / c, q(ef / c)^{1\over 2},-q(ef / c)^{1\over 2},aq / bc, aq / cd, ef / a, e, f\\
(ef / c)^{1\over 2},-(ef / c)^{1\over 2 }, bef / a, def / a, aq / c, fq / c, eq / cytosine
\end{array};q, \frac{bd}{a }
\right ] \nonumber\\
& & \qquad + \frac{b}{a}\frac{(aq, bq / a, bq / c, bq / d, bq / e, bq / f, d, e, f;q)_{\infty } }
{ (aq / b, aq / coke | 1+m};q)_{k}}(fd/abz)^{k}.\end{aligned}$$
From nonuerminating $_8\phi_7$ to $_6\psi_6$
============================================
In this aection, de give a semi-finite form of Bqiley’w $_6\psi_6$ summation formulx by usinh Bailey’w $3$-tecm transformatioi formula for z noncecminating very-wgll-poised $_8\phh_7$ series [@GR Eq. (2.11.1)]: $$\ceyin{aligned}
\label{Watson2}
\lefteqn{_{8}\phi_7\left[
\bqgin{arrsy}{f}
a, qa^{1\over 2},-qa^{1\ovgr 2},b,c, q, e, r\\
a^{1\over 2},-a^{1\over 2},aq/b,aq/c,aq/d,aq/e,aq/f
\end{arrzy};q, \frab{a^2q^2}{bcdef}
\right] }\nomumber\\
&& \,
=\frac{(aq,aq/de,aq/df,aq/ef,ee/c,fq/f,b/a,bef/a;q)_{\infty}}
{(aq/d,aq/f,aq/f,aq/def,q/c,grq/c,fw/a,bf/a;q)_{\infty}}\nunumber\\
&& \qqlcd \cdot_{8}\phi_7\lgft[
\begin{array}{c}
ef/c, q(ef/c)^{1\over 2},-q(ef/c)^{1\oxer 2},aa/bc,aq/cd, ef/q, w, f\\
(fx/c)^{1\over 2},-(ef/c)^{1\oter 2}, bvf/a,def/a,aq/c,fq/c,ca/c
\end{asray};q, \ftac{bd}{a}
\right] \npnukbee\\
&&\qquad +\frac{b}{a}\frac{(aq,bx/a,bq/c,bq/d,bq/e,bq/f,d,e,f;q)_{\ynfty}}
{(aq/b,dq/e | 1+m};q)_{k}}(cd/abz)^{k}.\end{aligned}$$ From nonterminating $_8\phi_7$ to $_6\psi_6$ ============================================ section, give a form of Bailey’s Bailey’s transformation formula for nonterminating very-well-poised $_8\phi_7$ [@GR Eq. (2.11.1)]: $$\begin{aligned} \label{Watson2} \lefteqn{_{8}\phi_7\left[ a, qa^{1\over 2},-qa^{1\over 2},b,c, d, e, f\\ a^{1\over 2},-a^{1\over 2},aq/b,aq/c,aq/d,aq/e,aq/f \end{array};q, \frac{a^2q^2}{bcdef} \right] && \, =\frac{(aq,aq/de,aq/df,aq/ef,eq/c,fq/c,b/a,bef/a;q)_{\infty}} {(aq/d,aq/e,aq/f,aq/def,q/c,efq/c,be/a,bf/a;q)_{\infty}}\nonumber\\ && \qquad \cdot_{8}\phi_7\left[ \begin{array}{c} ef/c, q(ef/c)^{1\over 2},-q(ef/c)^{1\over 2},aq/bc,aq/cd, ef/a, f\\ 2},-(ef/c)^{1\over bef/a,def/a,aq/c,fq/c,eq/c \frac{bd}{a} \right] \nonumber\\ &&\qquad +\frac{b}{a}\frac{(aq,bq/a,bq/c,bq/d,bq/e,bq/f,d,e,f;q)_{\infty}} {(aq/b,aq/c | 1+m};q)_{k}}(cd/abz)^{k}.\end{aligned}$$
From noNterminatiNg $_8\phi_7$ To $_6\pSi_6$
============================================
IN tHis sEctiOn, we give a semi-fINite Form of Bailey’s $_6\psi_6$ summatIon foRmULa by USiNg BaiLey’s $3$-terM TrANSfoRmAtIon FoRMuLa for A noNterminAting very-wEll-PoIsed $_8\phi_7$ serieS [@gR eq. (2.11.1)]: $$\begin{aliGneD}
\label{Watson2}
\LefTeqn{_{8}\phI_7\lEft[
\BEgin{aRraY}{c}
a, qa^{1\Over 2},-qa^{1\OVer 2},b,c, d, E, f\\
a^{1\over 2},-a^{1\oVeR 2},Aq/b,aq/c,AQ/d,aq/e,aq/F
\ENd{ArraY};q, \frac{a^2q^2}{bcdef}
\righT] }\NoNUmber\\
&& \,
=\frac{(aq,aq/dE,aq/df,aQ/eF,Eq/C,FQ/c,b/A,beF/a;q)_{\infty}}
{(aq/D,aQ/e,aq/f,AQ/def,q/c,eFQ/c,BE/A,Bf/a;Q)_{\Infty}}\nonumber\\
&& \Qquad \cdot_{8}\phI_7\LefT[
\begin{ArRay}{C}
Ef/c, q(ef/C)^{1\over 2},-Q(eF/C)^{1\ovEr 2},aq/bc,aq/cd, eF/a, e, f\\
(Ef/c)^{1\over 2},-(ef/C)^{1\over 2}, bEF/a,def/a,aQ/C,fq/c,eq/c
\End{arrAy};q, \FraC{bd}{a}
\RIgHt] \NonUmBEr\\
&&\qQUaD +\frAC{b}{a}\Frac{(aq,bq/A,bQ/c,Bq/d,bq/E,bq/f,D,E,F;Q)_{\InftY}}
{(aq/B,aq/c | 1+m};q)_{k}}(cd/abz)^{k}.\ end{aligne d}$$
Fr omno nter mina ting $_8\phi_7 $ to$_6\psi_6$
=========== ===== == = ==== = == ===== ======= = == = = =
In t his s e ct ion,wegive asemi-finit e f or m of Bailey’ s $ _6\psi_6$sum mation formu laby usi ng Ba i ley’s $3 $-ter m tran s format ion formu la for an ontermi n a ti ng v ery-well-poised $ _ 8\ p hi_7$ series [ @GR Eq .( 2. 1 1 .1) ]:$$\begin{a li gned} \label{ W at s o n 2}\ lefteqn{_{8}\ phi_7\left[ \be gin{ar ra y}{ c }
a, q a^{1\ ov e r 2 },-qa^{1\ov er 2 },b,c, d, e, f\ \
a^{1\o v er 2},- a^{1\o ver 2} ,aq/ b ,a q/ c,a q/ d ,aq / e, aq/ f
\e nd{array }; q, \fra c{a^ 2 q ^ 2 }{bc def }
\ right ] }\nonumber\ \
& & \, =\f rac{( aq,aq /de, aq /df,a q/ef,e q/c,f q/ c,b/a,bef/a;q)_ {\in fty}}
{(a q/d ,a q/e ,a q/f,a q /def,q /c, efq /c,be/a ,bf/a;q ) _{\ in f t y }} \nonumber\\
&& \qq ua d \c dot_{8}\ phi_7\ l ef t[ \begin{a rr ay} {c}e f /c, q (ef/ c )^ {1\over2},-q( e f/ c) ^{1\ove r2},aq/ bc ,aq /cd , ef/ a , e, f\\
( ef/c)^{1 \over 2},-(ef/c)^{1\ o ver 2}, bef/a , de f / a, a q/c, fq/ c,eq/c
\end {arr a y};q , \f r ac {bd } {a}
\righ t] \n o number\\
&&\qquad+\ frac{b }{a}\ frac{(aq,bq/a ,bq/c,bq/d , b q /e,bq/f, d,e, f ;q ) _{\infty}}
{( aq/b, aq/c | 1+m};q)_{k}}(cd/abz)^{k}.\end{aligned}$$
From nonterminating_$_8\phi_7$ to_$_6\psi_6$
============================================
In this section, we_give a_semi-finite_form of_Bailey’s_$_6\psi_6$ summation formula_by using Bailey’s_$3$-term transformation formula for_a nonterminating very-well-poised_$_8\phi_7$_series [@GR Eq. (2.11.1)]: $$\begin{aligned}
\label{Watson2}
\lefteqn{_{8}\phi_7\left[
\begin{array}{c}
a, qa^{1\over 2},-qa^{1\over 2},b,c, d, e, f\\
a^{1\over 2},-a^{1\over 2},aq/b,aq/c,aq/d,aq/e,aq/f
\end{array};q, \frac{a^2q^2}{bcdef}
_\right]_}\nonumber\\
&& \,
=\frac{(aq,aq/de,aq/df,aq/ef,eq/c,fq/c,b/a,bef/a;q)_{\infty}}
{(aq/d,aq/e,aq/f,aq/def,q/c,efq/c,be/a,bf/a;q)_{\infty}}\nonumber\\
&&_\qquad_\cdot_{8}\phi_7\left[
\begin{array}{c}
ef/c,_q(ef/c)^{1\over 2},-q(ef/c)^{1\over 2},aq/bc,aq/cd, ef/a, e,_f\\
(ef/c)^{1\over 2},-(ef/c)^{1\over 2}, bef/a,def/a,aq/c,fq/c,eq/c
\end{array};q, \frac{bd}{a}
_\right] \nonumber\\
_&&\qquad +\frac{b}{a}\frac{(aq,bq/a,bq/c,bq/d,bq/e,bq/f,d,e,f;q)_{\infty}}
{(aq/b,aq/c |
$, then
(1) the family of inverses $\{Q_x:=P_{x}^{-1}\}_{x\in\mathbb{R}^m}$ induces a linear operator $$Q:\Gamma(p^*(W)){\longrightarrow}\Gamma(p^*(V)),\ \ \{\omega_{x}\}_{x\in \mathbb{R}^m}\mapsto \{Q_x\omega_x\}_{x\in \mathbb{R}^m},$$ where $p^*(V):=V\times \mathbb{R}^m\to F\times\mathbb{R}^m$ and $p^*(W):=W\times \mathbb{R}^m\to
F\times\mathbb{R}^m$;
(2) $Q$ is locally tame, in the sense that for all bounded opens $U\subset\mathbb{R}^m$, there exist constants $C_{n,U}>0$, such that the following inequalities hold $$\|Q(\omega)\|_{n,F\times\overline{U}}\leq C_{n,U}\|\omega\|_{n+s-1,F\times \overline{U}}, \ \forall \omega \in \Gamma(p^*(W)_{| F\times\overline{U}}),$$ with $s=\lfloor\frac{1}{2}\mathrm{dim}(F)\rfloor+1$. If $U'\subset U$, then one can take $C_{n,U'}:= C_{n,U}$.
Fixing $C^n$-norms $\|\cdot\|_{n}$ on $\Gamma(V)$, we induce semi-norms on $\Gamma(p^*(V))$: $$\|\omega\|_{n,F\times\overline{U}}:=\sup_{0\leq k+|\alpha|\leq n}\sup_{x\in U}\|\frac{\partial^{|\alpha|}\omega_{x}}{\partial x^{\alpha}}\|_{k},$$ where $\omega\in\Gamma(p^*(V))$ is regarded as a smooth family $\omega=\{\omega_x\in
\Gamma(V)\}_{x\in\mathbb{R}^m}$. Similarly, fixing norms on $\Gamma(W)$, we define also norms on $\Gamma(p^*(W))$.
Endow $\Gamma(V)$ and $\Gamma(W)$ also with Sobolev norms, denoted by $\{|\cdot|_n\}_{n\geq 0}$. Loosely speaking, $|\omega|_ | $, then
(1) the family of inverses $ \{Q_x:=P_{x}^{-1}\}_{x\in\mathbb{R}^m}$ induces a linear hustler $ $ Q:\Gamma(p^*(W)){\longrightarrow}\Gamma(p^*(V)),\ \ \{\omega_{x}\}_{x\in \mathbb{R}^m}\mapsto \{Q_x\omega_x\}_{x\in \mathbb{R}^m},$$ where $ p^*(V):=V\times \mathbb{R}^m\to F\times\mathbb{R}^m$ and $ p^*(W):=W\times \mathbb{R}^m\to
F\times\mathbb{R}^m$;
(2) $ Q$ is locally meek, in the common sense that for all bounded opens $ U\subset\mathbb{R}^m$, there exist constants $ C_{n, U}>0 $, such that the postdate inequalities hold $ $ \|Q(\omega)\|_{n, F\times\overline{U}}\leq C_{n, U}\|\omega\|_{n+s-1,F\times \overline{U } }, \ \forall \omega \in \Gamma(p^*(W)_{| F\times\overline{U}}),$$ with $ s=\lfloor\frac{1}{2}\mathrm{dim}(F)\rfloor+1$. If $ U'\subset U$, then one can accept $ C_{n, U'}:= C_{n, U}$.
Fixing $ C^n$-norms $ \|\cdot\|_{n}$ on $ \Gamma(V)$, we induce semi - norm on $ \Gamma(p^*(V))$: $ $ \|\omega\|_{n, F\times\overline{U}}:=\sup_{0\leq k+|\alpha|\leq n}\sup_{x\in U}\|\frac{\partial^{|\alpha|}\omega_{x}}{\partial x^{\alpha}}\|_{k},$$ where $ \omega\in\Gamma(p^*(V))$ is regarded as a smooth syndicate $ \omega=\{\omega_x\in
\Gamma(V)\}_{x\in\mathbb{R}^m}$. Similarly, fixing norms on $ \Gamma(W)$, we specify also norms on $ \Gamma(p^*(W))$.
Endow $ \Gamma(V)$ and $ \Gamma(W)$ also with Sobolev norms, denoted by $ \{|\cdot|_n\}_{n\geq 0}$. Loosely talk, $ |\omega| _ | $, thfn
(1) the family of inverser $\{Q_x:=P_{x}^{-1}\}_{x\in\mathbb{T}^m}$ inducxs a lihear opefator $$Q:\Gamma(p^*(W)){\longrightarrow}\Jammq(p^*(V)),\ \ \{\omega_{x}\}_{x\in \mathbb{R}^m}\maprto \{Q_x\omeha_x\}_{x\in \mqthbu{R}^m},$$ where $p^*(V):=V\timxa \mathbn{X}^m\to R\bimes\kethbb{R}^m$ and $p^*(W):=W\jimes \mathbb{S}^m\to
F\times\kaghyb{R}^m$;
(2) $Q$ is locally tame, in the sense ehat fot wll bounded opgns $U\xtbsef\mathbb{R}^m$, there exist constants $C_{n,H}>0$, such uhat the following inequalities hold $$\|Q(\omega)\|_{n,V\timfs\overline{U}}\leq C_{n,U}\|\lmega\|_{n+s-1,F\timga \odwrline{U}}, \ \forxll \omega \pu \Gamma(p^*(W)_{| F\jimes\overline{U}}),$$ with $s=\lfloor\frac{1}{2}\mxthrm{bim}(F)\rfloor+1$. Uf $U'\sktset U$, then one ban take $C_{n,U'}:= G_{m,U}$.
Fixitg $C^n$-notms $\|\cdot\|_{n}$ on $\Gsmme(V)$, ww induce semi-norms on $\Gamma(p^*(V))$: $$\|\omega\|_{n,F\tiies\overlite{B}}:=\sup_{0\leq k+|\alpha|\leq n}\sup_{x\un U}\|\ftac{\pastiau^{|\qlpfa|}\ojeja_{x}}{\lartiap x^{\elpha}}\|_{k},$$ whers $\omega\in\Ganma(p^*(V))$ is regarded ax w smooth familg $\omegw=\{\oiega_x\in
\Gamma(V)\}_{x\in\mathbb{R}^m}$. Similarly, fixitg horms on $\Gamma(W)$, we defibe also norms on $\Gammw(p^*(W))$.
Endow $\Damma(V)$ and $\Gamma(W)$ also with Sobolev norms, denoted by $\{|\cvog|_n\}_{n\ycz 0}$. Llosely speaking, $|\omega|_ | $, then (1) the family of inverses a operator $$Q:\Gamma(p^*(W)){\longrightarrow}\Gamma(p^*(V)),\ \{\omega_{x}\}_{x\in \mathbb{R}^m}\mapsto \{Q_x\omega_x\}_{x\in and \mathbb{R}^m\to F\times\mathbb{R}^m$; (2) is locally tame, the sense that for all bounded $U\subset\mathbb{R}^m$, there exist constants $C_{n,U}>0$, such that the following inequalities hold $$\|Q(\omega)\|_{n,F\times\overline{U}}\leq C_{n,U}\|\omega\|_{n+s-1,F\times \ \forall \omega \in \Gamma(p^*(W)_{| F\times\overline{U}}),$$ with $s=\lfloor\frac{1}{2}\mathrm{dim}(F)\rfloor+1$. If $U'\subset U$, then one take C_{n,U}$. $C^n$-norms on $\Gamma(V)$, we induce semi-norms on $\Gamma(p^*(V))$: $$\|\omega\|_{n,F\times\overline{U}}:=\sup_{0\leq k+|\alpha|\leq n}\sup_{x\in U}\|\frac{\partial^{|\alpha|}\omega_{x}}{\partial x^{\alpha}}\|_{k},$$ where $\omega\in\Gamma(p^*(V))$ is regarded a smooth family $\omega=\{\omega_x\in \Gamma(V)\}_{x\in\mathbb{R}^m}$. Similarly, fixing norms $\Gamma(W)$, we define also on $\Gamma(p^*(W))$. Endow $\Gamma(V)$ and also Sobolev norms, by 0}$. speaking, $|\omega|_ | $, then
(1) the family of inverses $\{Q_x:=p_{x}^{-1}\}_{x\in\mathbB{R}^m}$ inDucEs a LiNear OperAtor $$Q:\Gamma(p^*(W)){\loNGrigHtarrow}\Gamma(p^*(V)),\ \ \{\omega_{x}\}_{x\iN \mathBb{r}^M}\mapSTo \{q_x\omeGa_x\}_{x\in \mAThBB{r}^m},$$ wHeRe $P^*(V):=V\TiMEs \MathbB{R}^m\To F\timeS\mathbb{R}^m$ aNd $p^*(w):=W\Times \mathbb{R}^M\To
f\times\mathBb{R}^M$;
(2) $Q$ is locally tAme, In the sEnSe tHAt for All BoundEd openS $u\subseT\mathbb{R}^m$, ThERe exisT ConstanTS $c_{n,u}>0$, sucH that the following INeQUalities hold $$\|Q(\oMega)\|_{n,F\TiMEs\OVErlIne{u}}\leq C_{n,U}\|\omeGa\|_{N+s-1,F\tiMEs \overlINe{u}}, \ \FORalL \Omega \in \Gamma(p^*(w)_{| F\times\overLIne{u}}),$$ with $s=\LfLooR\Frac{1}{2}\maThrm{dIm}(f)\RflOor+1$. If $U'\subseT U$, thEn one can tAke $C_{n,U'}:= c_{N,U}$.
FixinG $c^n$-norms $\|\Cdot\|_{n}$ oN $\GaMma(v)$, we iNDuCe SemI-nORms ON $\GAmmA(P^*(V))$: $$\|\oMega\|_{n,F\tiMeS\oVerliNe{U}}:=\sUP_{0\LEQ k+|\alPha|\Leq n}\Sup_{x\iN U}\|\frac{\partial^{|\AlpHa|}\omEGa_{x}}{\PartiAl x^{\alPha}}\|_{k},$$ WhEre $\omEga\in\GAmma(p^*(v))$ iS regarded as a smoOth fAmily $\omegA=\{\omEgA_x\iN
\GAmma(V)\}_{X\In\mathBb{R}^M}$. SiMilarly, Fixing nORms On $\gAMMa(w)$, we define also norms On $\gAMmA(p^*(W))$.
Endow $\gamma(V)$ ANd $\gaMMa(W)$ also wItH SoBoleV NOrms, dEnotED bY $\{|\cdot|_n\}_{n\gEq 0}$. LoosELy SpEaking, $|\oMeGa|_ | $, then
(1) the family of inverses$\{Q_ x:= P_{ x} ^{-1 }\}_ {x\in\mathbb{R } ^m}$ induces a linear oper ator$$ Q :\Ga m ma (p^*( W)){\lo n gr i g hta rr ow }\G am m a( p^*(V )), \ \ \{ \omega_{x} \}_ {x \in \mathbb{ R }^ m}\mapsto\{Q _x\omega_x\} _{x \in \m at hbb { R}^m} ,$$ wher e $p^* ( V):=V\ times \ma th b b{R}^m \ to F\ti m e s\ math bb{R}^m$ and $p^* ( W) : =W\times \math bb{R}^ m\ t o F \ti mes\mathbb {R }^m$;
(2) $Q $ i s l oca l ly tame, in t he sense th a t f or all b oun d ed ope ns $U \s u bse t\mathbb{R} ^m$, there ex ist co n stants$ C_{n,U} >0$, s uch th at t h efo llo wi n g i n eq ual i tie s hold $ $\ |Q (\ome ga)\ | _ { n ,F\t ime s\ov erlin e{U}}\leq C_{ n,U }\|\ o meg a\|_{ n+s-1 ,F\t im es \o verlin e{U}} ,\ \forall \omeg a \i n \Gamma( p^* (W )_{ |F\tim e s\over lin e{U }}),$$with $s = \lf lo o r \ fr ac{1}{2}\mathrm{di m} ( F )\ rfloor+1 $. If$ U' \s u bset U$, t hen one c an ta ke $ C _{ n,U'}:=C_{n,U } $.
Fixing$C ^n$-no rm s $ \|\ cdot\ | _{n} $ on $ \Gamma(V )$, w e induce semi-n o rms on $\Gamm a (p ^ * (V ) )$:$$\ |\omega\|_{ n,F\ t imes \ove r li ne{ U }}:=\ sup_{ 0\ l eq k+|\alpha|\leq n}\s up _{x\in U}\| \frac{\partia l^{|\alpha | } \ omega_{x }}{\ p ar t ial x^{\alpha} }\|_{ k},$$ wher e $\omega \in\G amma(p^* (V))$ isr e garded a s a sm oot h f a m il y $\omega=\{\ o m ega_ x\ in
\Gam ma( V)\}_{x \in \ma thb b{R }^ m}$. Simi larly, f ix in gno rms on $ \ Gamma(W) $, we d efi ne al s o norm s on$\Ga mm a( p ^*( W))$.
E nd o w $\G am ma (V)$ an d$\Gam ma(W ) $ a lso wit h Sobolev no r ms,de no ted by$\{|\cdot|_n\ }_ {n\geq 0}$ .Loo sely s p e aking, $ |\omega|_ | $, then
(1)_the family_of inverses $\{Q_x:=P_{x}^{-1}\}_{x\in\mathbb{R}^m}$ induces_a linear_operator_$$Q:\Gamma(p^*(W)){\longrightarrow}\Gamma(p^*(V)),\ \__\{\omega_{x}\}_{x\in \mathbb{R}^m}\mapsto \{Q_x\omega_x\}_{x\in_\mathbb{R}^m},$$ where $p^*(V):=V\times_\mathbb{R}^m\to F\times\mathbb{R}^m$ and $p^*(W):=W\times_\mathbb{R}^m\to
__F\times\mathbb{R}^m$;
(2) $Q$ is locally tame, in the sense that for all bounded opens $U\subset\mathbb{R}^m$,_there_exist constants_$C_{n,U}>0$,_such_that the following inequalities hold_$$\|Q(\omega)\|_{n,F\times\overline{U}}\leq C_{n,U}\|\omega\|_{n+s-1,F\times \overline{U}}, \ \forall_\omega \in_\Gamma(p^*(W)_{| F\times\overline{U}}),$$ with $s=\lfloor\frac{1}{2}\mathrm{dim}(F)\rfloor+1$. If $U'\subset U$, then_one_can take $C_{n,U'}:=_C_{n,U}$.
Fixing $C^n$-norms $\|\cdot\|_{n}$ on $\Gamma(V)$, we induce semi-norms on_$\Gamma(p^*(V))$: $$\|\omega\|_{n,F\times\overline{U}}:=\sup_{0\leq k+|\alpha|\leq n}\sup_{x\in U}\|\frac{\partial^{|\alpha|}\omega_{x}}{\partial x^{\alpha}}\|_{k},$$_where $\omega\in\Gamma(p^*(V))$ is_regarded_as_a smooth family $\omega=\{\omega_x\in
\Gamma(V)\}_{x\in\mathbb{R}^m}$._Similarly, fixing norms on $\Gamma(W)$, we_define also norms on $\Gamma(p^*(W))$.
Endow $\Gamma(V)$_and $\Gamma(W)$ also with Sobolev norms, denoted_by $\{|\cdot|_n\}_{n\geq 0}$. Loosely speaking, $|\omega|_ |
=\frac{3}{2\Lambda}.$$ where $r_0$ can be obtained for a given $m$ and $R^2$ is the area radius. We also study the geometry of the 3-dimensional orbit space of the Killing vector field in a future/past neighbourhood of the horizon [@chrusciel]. The neighbourhood topology is $$\label{eq:bundle}
S^3\times \mathbb{R}.$$ The orbit space topology $S^2\times \mathbb{R}$ and the geometry inherited from the spacetime are uniquely defined. We show that the orbit space geometry is smooth at every point for a sufficiently thin neighbourhood in the case if (\[eq:constraint-params\]) is satisfied. This result applies also to the spacetime geometry of the neighbourhood.
Next, we generalise our result to the accelerated Kerr-NUT-(Anti) de Sitter spacetimes characterised by five parameters $(m, a, l, \alpha, \Lambda)$. They are a five dimensional subfamily of the Plebański-Demiański family describing a rotating and accelerating black hole with a NUT parameter on the cosmological constant background. We find a four dimensional subfamily that contains a singularity free horizon of a singularity free neighbourhood.
The idea of the current research originates from the program of abstractly defined vacuum isolated horizons (IH) of the Petrov type D with a cosmological constant [@geometryhorizonsAshtekar_2002; @MechanicsIHPhysRevD.64.044016; @GeometricCharacterizationsoftheKerrIsolatedHorizon; @Lewandowski_2003; @Lewandowski_2006; @DOBKOWSKIRYLKO2018415; @localnohairPhysRevD.98.024008; @hopf]. A comprehensive review of the IH theory and its application can be found here [@Ashtekar2004]. They can be characterised with respect to the principal fiber bundle structure of the null Killing flow. In the trivial bundle case, all the axisymmetric type D IHs set a three dimensional family, and each of them corresponds to a Kerr–(anti-)de Sitter spacetime [@localnohairPhysRevD.98.024008] (modulo subtleties that arise in the extremal horizon case). In the case, when the bundle structure is that of a non-trivial U(1) principal fiber bundle over $S^2$, a four dimensional family of the axis | = \frac{3}{2\Lambda}.$$ where $ r_0 $ can be obtained for a given $ m$ and $ R^2 $ is the area radius. We besides analyze the geometry of the 3 - dimensional orbit space of the Killing vector sphere in a future / past neighborhood of the horizon [ @chrusciel ]. The neighbourhood topology is $ $ \label{eq: bundle }
S^3\times \mathbb{R}.$$ The sphere space topology $ S^2\times \mathbb{R}$ and the geometry inherit from the spacetime are uniquely defined. We show that the eye socket space geometry is smooth at every distributor point for a sufficiently thin neighbourhood in the case if (\[eq: constraint - params\ ]) is satisfied. This consequence applies also to the spacetime geometry of the neighbourhood.
Next, we generalize our result to the accelerated Kerr - NUT-(Anti) de Sitter spacetimes characterised by five parameters $ (m, a, l, \alpha, \Lambda)$. They are a five dimensional subfamily of the Plebański - Demiański class describing a rotating and accelerating black hole with a NUT argument on the cosmological constant background. We find a four dimensional subfamily that contains a singularity free horizon of a singularity spare neighbourhood.
The idea of the current research originates from the program of abstractly defined vacuum isolated horizons (IH) of the Petrov type D with a cosmologic constant [ @geometryhorizonsAshtekar_2002; @MechanicsIHPhysRevD.64.044016; @GeometricCharacterizationsoftheKerrIsolatedHorizon; @Lewandowski_2003; @Lewandowski_2006; @DOBKOWSKIRYLKO2018415; @localnohairPhysRevD.98.024008; @hopf ]. A comprehensive review of the IH theory and its lotion can be discover here [ @Ashtekar2004 ]. They can be qualify with respect to the principal roughage bundle structure of the null Killing flow. In the trivial package case, all the axisymmetric type D IHs set a three dimensional family, and each of them corresponds to a Kerr–(anti-)de Sitter spacetime [ @localnohairPhysRevD.98.024008 ] (modulo subtlety that arise in the extremal horizon case). In the case, when the bundle social organization is that of a non - trivial U(1) chief fiber bundle over $ S^2 $, a four dimensional family of the axis | =\fraf{3}{2\Lambda}.$$ where $r_0$ can be ontained for a girwn $m$ aid $R^2$ is the arex radius. We also study the gxomerry od the 3-dimensional orbig space ov the Kiolinj vector field ii a futuvz/past keighyonrhood of the hprizon [@chrgsciel]. The neichcobrhood topology is $$\label{eq:bundle}
F^3\times \kahhbb{R}.$$ The orbij spabe topklogy $S^2\times \mathbb{R}$ and the geomefry inhtrited from the spscetime are uniquely definfd. Wf show that the orhit space ggkmeeey is smooth at every konnt for a sofficiently thin neighbourhood iv the case if (\[ew:cinshtaint-params\]) ms satpsfied. This rcxult a[plies slso to the spscevime geometry of the neigibourhood.
Next, we gengralise ous xesult to the accelerqtwd Ketr-NUT-(Dnti) ee Rituer slacetiles characterjsed by fivw parameters $(m, a, l, \sl[ya, \Lambda)$. Theg are w sive dimensional subfamily of the Plebańvki-Semiański family describung a rotating and acfeleratind black hole with a NUT parameter on the cosmologhcal rovstckb bazjggound. We find a four dimensional subfamily thwf voktains a singulavity free horizon pf a fingularity ftee neiygbkurhood.
The idea of the cutrent eesearch jrigonates from the program of qbstractly dvfinwd vacuum isolated horizons (IK) of tne Peyrov type D with a cosmulogjcal constajt [@geometdhhorizonsAshtekaf_2002; @MvchaticsIHPhysRevD.64.044016; @GeometricChwracterizetionxoftheKdrrIxolateqHorizon; @Lfwandowski_2003; @Lewandowski_2006; @DLBKOWDKHRYLKO2018415; @locwlnohairPhysRevD.98.024008; @hopf]. A comprehxisive review pf thv IH theoxy and its applicatyon can be fouud here [@Cshtekxr2004]. They cah be cheracterised rith respect dl the princi'al fiber bunele wtructufd of the null Lilling flow. In the trivial bundle caxe, xml the axisymmecxix type D IHs sey a thwev dmmensymnal family, dnd dacf of tfem correspivds yo a Kerr–(anti-)de Sittar slacetime [@localnohaorIhysRevD.98.024008] (nodulo stbtleties thay arise in the extgemal hocizon vasg). In the case, when the bundle sfructure ls bhat of a non-erivlal O(1) principal fiber bundle over $S^2$, a four dimensional family of the axiw | =\frac{3}{2\Lambda}.$$ where $r_0$ can be obtained for $m$ $R^2$ is area radius. We the orbit space of Killing vector field a future/past neighbourhood of the horizon The neighbourhood topology is $$\label{eq:bundle} S^3\times \mathbb{R}.$$ The orbit space topology $S^2\times \mathbb{R}$ the geometry inherited from the spacetime are uniquely defined. We show that the space is at point for a sufficiently thin neighbourhood in the case if (\[eq:constraint-params\]) is satisfied. This result applies to the spacetime geometry of the neighbourhood. Next, generalise our result to accelerated Kerr-NUT-(Anti) de Sitter spacetimes by parameters $(m, l, \Lambda)$. are a five subfamily of the Plebański-Demiański family describing a rotating and accelerating black hole with a NUT parameter on cosmological constant find a dimensional that a singularity free a singularity free neighbourhood. The idea research originates from the program of abstractly defined isolated horizons of the Petrov type D with cosmological constant [@geometryhorizonsAshtekar_2002; @MechanicsIHPhysRevD.64.044016; @GeometricCharacterizationsoftheKerrIsolatedHorizon; @Lewandowski_2003; @Lewandowski_2006; @DOBKOWSKIRYLKO2018415; @hopf]. A comprehensive review of the IH theory and its application can be found here can be characterised with to the principal bundle of null flow. In trivial bundle case, all the axisymmetric type D IHs set a dimensional family, and each of them corresponds to a Kerr–(anti-)de [@localnohairPhysRevD.98.024008] subtleties that arise the extremal horizon case). the when the bundle structure of non-trivial bundle $S^2$, four dimensional family of axis | =\frac{3}{2\Lambda}.$$ where $r_0$ can be obtaIned for a giVen $m$ aNd $R^2$ Is tHe Area RadiUs. We also study tHE geoMetry of the 3-dimensional oRbit sPaCE of tHE KIllinG vector FIeLD In a FuTuRe/pAsT NeIghboUrhOod of thE horizon [@chRusCiEl]. The neighboURhOod topologY is $$\Label{eq:bundlE}
S^3\tImes \maThBb{R}.$$ tHe orbIt sPace tOpologY $s^2\times \Mathbb{R}$ anD tHE geomeTRy inherITEd From The spacetime are unIQuELy defined. We shoW that tHe ORbIT SpaCe gEometry is sMoOth at EVery poiNT fOR A SufFIciently thin nEighbourhooD In tHe case If (\[Eq:cONstraiNt-parAmS\]) Is sAtisfied. ThiS resUlt applieS also tO The spacETime geoMetry oF thE neIghbOUrHoOd.
NExT, We gENeRalISe oUr result To ThE acceLeraTED kErr-NuT-(ANti) dE SittEr spacetimes cHarActeRIseD by fiVe parAmetErS $(m, a, l, \aLpha, \LaMbda)$. THeY are a five dimensIonaL subfamilY of ThE PlEbAński-dEmiańsKi fAmiLy descrIbing a rOTatInG AND aCcelerating black hoLe WITh A NUT paraMeter oN ThE cOSmologicAl ConStanT BAckgrOund. wE fInd a four DimensIOnAl SubfamiLy That coNtAinS a sIngulARity Free hoRizon of a SinguLArity free neighBOurhood.
The ideA Of THE cURrenT reSearch origiNateS From The pROgRam OF abstRactlY dEFiNEd vacuum isolated horIzOns (IH) oF the PEtrov type D witH a cosmologICAL constanT [@geoMEtRYhorizonsAshteKar_2002; @MeChanicsIHPHYsRevD.64.044016; @GeOmetrIcCharacTerizatioNSOftheKerRIsOlaTedhorIZOn; @lewandowski_2003; @LeWANdowSkI_2006; @DOBKOWsKIrYLKO2018415; @loCalNohAirphySREvD.98.024008; @hopf]. A cOmprehenSiVe ReViEw oF the Ih Theory anD iTs aPpLicAtion CAn be foUnd heRe [@AsHtEkAR2004]. ThEy can be CHaRACterIsEd With ResPeCt to tHe prINciPal fibeR bundle stRucTUre oF tHe Null KilLing flow. In the TrIvial bundlE cAse, All the AXIsymmetrIc type D IHs set a three dimeNSional fAmiLy, and Each Of them corResPonds tO a KERr–(anti-)De SittEr spaCeTimE [@LOcalnOHAiRPhYsrevD.98.024008] (modulo SUBtlEties ThAt arIse in thE extremal horizon caSE). In The case, when thE buNdle STRuCtuRE iS ThaT oF A noN-TRivial U(1) principaL fiber bundLe OVeR $S^2$, a four dimENsiOnAl familY of the aXis | =\frac{3}{2\Lambda}.$$ whe re $r_0$ c an be ob tai ne d fo r agiven $m$ and$ R^2$ is the area radius. W e als os tudy th e geo metry o f t h e 3- di me nsi on a lorbit sp ace ofthe Killin g v ec tor field in afuture/pas t n eighbourhood of the h or izo n [@ch rus ciel] . Then eighbo urhood to po l ogy is $$\labe l { eq :bun dle}
S^3\time s \ m athbb{R}.$$ Th e orbi ts pa c e to pol ogy $S^2\t im es \m a thbb{R} $ a n d the geometry inhe rited fromt hespacet im e a r e uniq uelyde f ine d. We showthat the orbi t spac e geomet r y is sm ooth a t e ver y po i nt f oras uff i ci ent l y t hin neig hb ou rhood int h e case if (\[ eq:co nstraint-para ms\ ]) i s sa tisfi ed. T hisre sultapplie s als oto the spacetim e ge ometry of th enei gh bourh o od.
N ext , w e gener alise o u r r es u l t t o the acceleratedKe r r -N UT-(Anti ) de S i tt er spacetim es ch arac t e rised byf iv e parame ters $ ( m, a , l, \a lp ha, \L am bda )$. They area five dimensi onals ubfamily of th e Plebański-De m ia ń s ki fami lydescribinga ro t atin g an d a cce l erati ng bl ac k h o le with a NUT param et er onthe c osmological c onstant ba c k g round. W e fi n da four dimensio nal s ubfamily t h at conta ins a singula rity free h orizon o f a si ngu lar i t yfree neighbou r h ood.
The ide a o f the c urr ent re sea rc h origina tes from t he p ro gra m ofa bstractl ydef in edvacuu m isola ted h oriz on s( IH) of the Pe t r ov t yp eD wi thacosmo logi c alconstan t [@geome try h oriz on sA shtekar _2002; @Mecha ni csIHPhysRe vD .64 .04401 6 ; @Geomet ricCharacterizationsoft h eKerrIs ola tedHo rizo n; @Lewan dow ski_20 03; @Lewan dowski _2006 ;@DO B K OWSKI R Y LK O20 18 415; @loca l n oha irPhy sR evD. 98.0240 08; @hopf]. A comp r ehe nsive reviewoftheI H t heo r ya ndit s ap p l ication can befound here [ @ As htekar2004 ] . T he y can b e chara cteri s ed with respectto the pr in cipa l fib er bundlestructur e of then ull K i ll ing f low . In t he tr ivial bundl e ca se, a ll the a xisymm etric t ype D IH s set a three dimension al fam ily,and each ofthe m co rresponds toa Kerr–(an ti- )de Sitt ers pacet ime[ @l oca l nohai rPhy s RevD.98.0 2 40 08] ( mo dulo subtle t i e s t hat a ris e in th e ex tremal horizon ca s e). In the cas e, w h e n t heb undl estructure is t hat o f a non-tr iv ial U(1) pr incipalfi b er bu ndle o ver $S ^2$, af o ur dimens iona l f amily ofthe a x is | =\frac{3}{2\Lambda}.$$ where_$r_0$ can_be obtained for a_given $m$_and_$R^2$ is_the_area radius. We_also study the_geometry of the 3-dimensional_orbit space of_the_Killing vector field in a future/past neighbourhood of the horizon [@chrusciel]. The neighbourhood topology_is_$$\label{eq:bundle}
___S^3\times \mathbb{R}.$$ The orbit space_topology $S^2\times \mathbb{R}$ and the_geometry inherited_from the spacetime are uniquely defined. We show_that_the orbit space_geometry is smooth at every point for a sufficiently_thin neighbourhood in the case if_(\[eq:constraint-params\]) is satisfied._This_result_applies also to the_spacetime geometry of the neighbourhood.
Next, we_generalise our result to the accelerated_Kerr-NUT-(Anti) de Sitter spacetimes characterised by five_parameters $(m, a, l, \alpha, \Lambda)$._They are a five dimensional_subfamily of_the Plebański-Demiański family describing a_rotating and accelerating_black hole_with a NUT_parameter on the cosmological constant background._We find a_four dimensional subfamily that contains a_singularity_free horizon of_a_singularity_free neighbourhood.
The_idea of the_current_research originates_from_the program of abstractly defined vacuum_isolated_horizons (IH) of the Petrov type D_with a cosmological constant_[@geometryhorizonsAshtekar_2002;[email protected]; @GeometricCharacterizationsoftheKerrIsolatedHorizon; @Lewandowski_2003; @Lewandowski_2006;_@DOBKOWSKIRYLKO2018415; @localnohairPhysRevD.98.024008; @hopf]. A comprehensive_review of the IH theory and_its application_can be_found here [@Ashtekar2004]. They can be characterised with respect to the_principal fiber bundle structure of the_null Killing flow. In_the trivial_bundle_case, all the_axisymmetric_type D_IHs set a three dimensional family, and_each of_them corresponds to a Kerr–(anti-)de Sitter_spacetime [@localnohairPhysRevD.98.024008] (modulo subtleties_that_arise in the extremal horizon case)._In the case, when the bundle_structure is that of a_non-trivial_U(1)_principal fiber bundle over $S^2$,_a four dimensional family of the_axis |
yukons” by a few scientists. Hans Euler and Werner Heisenberg then discussed in 1938 “the hypothesis that the hard component of cosmic rays consisted of mesotrons produced in the upper layers of the atmosphere by primary electrons or photons and then disintegrating, as predicted by Yukawa’s theory \[…\] with a life-time of the order of $10^{-6}$ secs.” (cf. [@Rossi1938b], p. 993). Such an hypothesis led to an anomalous attenuation in the atmosphere due to the decay of mesotrons in flight [@EulerHeisenberg1938]. As a matter of fact, in Asmara, in 1933, Rossi and De Benedetti had been the first to observe such an anomalous effect, as they had detected a decrease in the intensity of cosmic rays passing through the atmosphere decisively larger than what was expected. They could not explain the phenomenon at that time, but Rossi immediately interpreted it in 1938, after the publication of Euler and Heisenberg’s paper [@Rossi1938b].
It is worth pointing out that at that time, in 1938, Rossi was in Copenhagen, as he had just been forced to leave Italy because of the fascist racial laws. After short periods in Copenhagen and Manchester, Rossi went to the United States, working first at Cornell University, then at Los Alamos, and finally, from 1946 onwards, at the Massachusetts Institute of Technology (MIT). His departure was of course an invaluable loss for Italian physics.
Cosmic-ray studies in Italy during World War II
-----------------------------------------------
Besides Rossi, a few other physicists who had been with him in Arcetri, had also started working on cosmic rays. Gilberto Bernardini and Giuseppe Occhialini, in particular, both played an important role in the field, though their scientific careers were very different one from the other: Bernardini mainly spent the 1930s and the 1940s in Italy, and he significantly contributed to the development of cosmic ray researches within the country, while Occhialini spent many years abroad, where he gave fundamental contributions. We have already discussed Occhialini’s working with Blackett on cloud chambers in Cambridge, and we will see more about his work in the following sections.
As for Bernardini, he carried out his first researches on cosmic rays in the early 30s in Florence, where he collaborated with Sergio de Benedetti and Daria Bocciarelli | yukons ” by a few scientists. Hans Euler and Werner Heisenberg then discussed in 1938 “ the hypothesis that the unvoiced part of cosmic rays consisted of meson produced in the upper layer of the atmosphere by chief electrons or photons and then disintegrate, as predicted by Yukawa ’s theory \[ … \ ] with a life sentence - time of the order of $ 10^{-6}$ secant. ” (cf. [ @Rossi1938b ], p. 993). Such an guess led to an anomalous attenuation in the atmosphere due to the decay of meson in flight [ @EulerHeisenberg1938 ]. As a topic of fact, in Asmara, in 1933, Rossi and De Benedetti had been the beginning to observe such an anomalous effect, as they had detected a decrease in the saturation of cosmic rays passing through the standard atmosphere decisively larger than what was expect. They could not explain the phenomenon at that time, but Rossi immediately interpreted it in 1938, after the publication of Euler and Heisenberg ’s paper [ @Rossi1938b ].
It is deserving pointing out that at that time, in 1938, Rossi was in Copenhagen, as he had just been forced to leave Italy because of the fascist racial laws. After light periods in Copenhagen and Manchester, Rossi went to the United States, working first at Cornell University, then at Los Alamos, and finally, from 1946 onwards, at the Massachusetts Institute of Technology (MIT). His departure was of course an invaluable loss for Italian physics.
Cosmic - beam studies in Italy during World War II
-----------------------------------------------
Besides Rossi, a few other physicist who had been with him in Arcetri, had besides started working on cosmic rays. Gilberto Bernardini and Giuseppe Occhialini, in particular, both play an important role in the field, though their scientific career were very different one from the other: Bernardini mainly spent the 1930s and the 1940s in Italy, and he significantly contributed to the development of cosmic ray researches within the area, while Occhialini spent many year abroad, where he gave fundamental contributions. We have already discussed Occhialini ’s make with Blackett on cloud chambers in Cambridge, and we will see more about his work in the follow section.
As for Bernardini, he carried out his first inquiry on cosmic rays in the early thirty in Florence, where he collaborated with Sergio de Benedetti and Daria Bocciarelli | yuklns” by a few scientists. Mans Euler and Wgrber Hemsenberf then dkscussed in 1938 “the hypothesis vhat the yard component of cosmkc rays clnsisted of nwsotrons pckduced lu the mpper oayers of the stmosphere by primary elacgrlns or photons and then disintegratyng, as lrfdicted by Yukwwa’s eheodj \[…\] with a life-time of the order of $10^{-6}$ sebs.” (cf. [@Rossi1938b], p. 993). Sich an hypothesis led to aj anlmalous attenuatioj in the atnosprwre due to tfe decay of mesotrons jn flight [@EulerHeisenberg1938]. As a oattex of fact, ib Qsmwta, in 1933, Rossi and Qe Benedetti had beet the forst to observc suci an anomalous effect, as vhey had detected a qecrease hn the intensity of cismic rayv parwine tgrpufh the atjosphere dscisively lqrger than what was evircted. They ckuld njt explain the phenomenon at that time, buu Rosai immediately interprered it in 1938, after the kublicatiog of Euler and Heisenberg’s paper [@Rossi1938b].
It is wortv poiitkng omt tfqt at that time, in 1938, Rossi was in Copenhagen, as rs nac just been fovced to leave Italu hevwuse of the fxscist radial laws. After shlrt peryods un Copenhwgen and Manchester, Rossi went ro the Uniteb Srates, working firsc at Cornell Unifersiyy, then at Los Alamos, aud finzlly, from 1946 lnwards, af the Massachusetgs Pnsthtute of Technology (MIT). Hif departuce wax of cojrse an indaluable llss for Italian physics.
Clsmic-tay stgdies in Ihaly during World War II
-----------------------------------------------
Besides Rossi, a few ojhes pvysicistf who had been witr him in Arcetti, had alfo stxrted workpng on covmic rays. Gylberto Bernasfini and Giuveppe Ocshiaoini, in pargkcular, both plsyed an important rile in the field, tmough fheir scientifie xareers were vety aifsegenv one xrom the othar: Bdrnxtdini mainly spekt ghe 1930x and the 1940s in Italy, and he significantly voktributed to the qevelopment og cosmic ray reseagches wivhin tne sountry, while Occhialini spent many yeags wbroad, where he nave fundamentcl contributions. We have already discussxd Occhialini’s working qith Blackett on clpmd chambers mn Camfridge, ang we will see more avout his work in uhe following sections.
Aa for Ternagdini, he carried out his first researches on cosmic rays in the early 30s ib Florxnse, where he colkabordtzd cith Sewgio ve Benedetti and Caria Bocciarelli | yukons” by a few scientists. Hans Euler Heisenberg discussed in “the hypothesis that rays of mesotrons produced the upper layers the atmosphere by primary electrons or and then disintegrating, as predicted by Yukawa’s theory \[…\] with a life-time of order of $10^{-6}$ secs.” (cf. [@Rossi1938b], p. 993). Such an hypothesis led to anomalous in atmosphere to the decay of mesotrons in flight [@EulerHeisenberg1938]. As a matter of fact, in Asmara, in Rossi and De Benedetti had been the first observe such an anomalous as they had detected a in intensity of rays through atmosphere decisively larger what was expected. They could not explain the phenomenon at that time, but Rossi immediately interpreted it 1938, after of Euler Heisenberg’s [@Rossi1938b]. is worth pointing at that time, in 1938, Rossi as he had just been forced to leave because of fascist racial laws. After short periods Copenhagen and Manchester, Rossi went to the United working first at Cornell University, then at Los Alamos, and finally, from 1946 onwards, at Institute of Technology (MIT). departure was of an loss Italian Cosmic-ray studies Italy during World War II ----------------------------------------------- Besides Rossi, a few other who had been with him in Arcetri, had also started cosmic Gilberto Bernardini and Occhialini, in particular, both an role in the field, scientific were from other: mainly spent the 1930s the 1940s in Italy, and significantly contributed to the within the country, while Occhialini spent many years where he gave fundamental contributions. We have discussed Occhialini’s working with Blackett on cloud chambers in Cambridge, and we see more work in the following sections. As for Bernardini, carried out his first on cosmic rays in the early 30s in Florence, he with Sergio Benedetti and Daria | yukons” by a few scientists. HanS Euler and WErner heiSenBeRg thEn diScussed in 1938 “the hyPOtheSis that the hard componenT of coSmIC rayS CoNsistEd of mesOTrONS prOdUcEd iN tHE uPper lAyeRs of the Atmosphere By pRiMary electronS Or Photons and TheN disintegratIng, As predIcTed BY YukaWa’s TheorY \[…\] with a LIfe-timE of the ordEr OF $10^{-6}$ secs.” (cF. [@rossi1938b], p. 993). sUCh An hyPothesis led to an anOMaLOus attenuation In the aTmOSpHERe dUe tO the decay oF mEsotrONs in fliGHt [@eULErHEIsenberg1938]. As a maTter of fact, iN asmAra, in 1933, ROsSi aND De BenEdettI hAD beEn the first tO obsErve such aN anomaLOus effeCT, as they Had detEctEd a DecrEAsE iN thE iNTenSItY of COsmIc rays paSsInG throUgh tHE ATMospHerE decIsiveLy larger than wHat Was eXPecTed. ThEy couLd noT eXplaiN the phEnomeNoN at that time, but ROssi ImmediateLy iNtErpReTed it IN 1938, after The PubLicatioN of EuleR And heISENbErg’s paper [@Rossi1938b].
It iS wORTh Pointing Out thaT At ThAT time, in 1938, ROsSi wAs in cOPenhaGen, aS He Had just bEen forCEd To Leave ItAlY becauSe Of tHe fAscisT RaciAl laws. after shoRt perIOds in CopenhageN And Manchester, rOsSI WeNT to tHe UNited States, WorkINg fiRst aT coRneLL UnivErsitY, tHEn AT Los Alamos, and finallY, fRom 1946 onwArds, aT the MassachusEtts InstitUTE Of TechnoLogy (miT). hIs departure was Of couRse an invalUAble loss For ItAlian phySics.
CosmiC-RAy studieS in itaLy dUriNG woRld War II
-----------------------------------------------
BesidES rossI, a Few otheR phYsicistS whO haD beEn wItH him in ArcEtri, had aLsO sTaRtEd wOrkinG On cosmic RaYs. GIlBerTo BerNArdini And GiUsepPe ocCHiaLini, in pARtICUlar, BoTh PlayEd aN iMportAnt rOLe iN the fieLd, though tHeiR ScieNtIfIc careeRs were very difFeRent one froM tHe oTher: BeRNArdini maInly spent the 1930s and the 1940s in ITAly, and hE siGnifiCantLy contribUteD to the DevELopmenT of cosMic raY rEseARChes wITHiN thE cOuntry, whilE oCchIalinI sPent Many yeaRs abroad, where he gavE FunDamental contrIbuTionS. wE hAve ALrEAdy DiSCusSED Occhialini’s worKing with BlAcKEtT on cloud chAMbeRs In CambrIdge, and We wilL See more About his wOrk in the fOlLowiNG SecTions.
As for bernardiNi, he carriED out hIS fIrst rEseArches On CosMic raYs in thE EarLy 30s in florenCe, Where hE collAbOrated wiTh Sergio de Benedetti and DAria BoCciarEllI | yukons” by a few scientist s. Hans Eu ler a ndWer ne r He isen berg then disc u ssed in 1938 “the hypothes is th at theh ar d com ponento fc o smi cra ysco n si stedofmesotro ns produce d i nthe upper la y er s of the a tmo sphere by pr ima ry ele ct ron s or p hot ons a nd the n disin tegrating ,a s pred i cted by Y uk awa’ s theory \[…\] wi t ha life-time ofthe or de r o f $10 ^{- 6}$ secs.” ( cf. [ @ Rossi19 3 8b ] , p.9 93). Such anhypothesisl edto anan oma l ous at tenua ti o n i n the atmos pher e due tothe de c ay of m e sotrons in fl igh t [ @Eul e rH ei sen be r g19 3 8] . A s amatter o ffa ct, i n As m a r a , in 19 33,Rossi and De Bened ett i ha d be en th e fir st t oobser ve suc h anan omalous effect, asthey haddet ec ted a decr e ase in th e i ntensit y of co s mic r a y s p assing through the a t m os phere de cisive l yla r ger than w hat was e xpect ed.T he y couldnot ex p la in the ph en omenon a t t hat time , but Rossi immedia telyi nterpreted iti n 1938, after th e pu b lica tio n of EulerandH eise nber g ’s pa p er [@ Rossi 19 3 8b ] .
It is worth poin ti ng out that at that time , in 1938, R o ssi wasin C o pe n hagen, as he h ad ju st been fo r ced to l eaveItaly be cause oft h e fascis t r aci allaw s . A fter short pe r i odsin Copenh age n and M anc hes ter , R os si went t o the Un it ed S ta tes , wor k ing firs tatCo rne ll Un i versit y, th en a tLo s Al amos, a n df i nall y, f rom194 6onwar ds,a t t he Mass achusetts In s titu te o f Techn ology (MIT).Hi s departur ewas of co u r se an in valuable loss for Itali a n physi cs.
Cos mic- ray studi esin Ita lyd uringWorldWar I I--- - - ----- - - -- --- -- ---------- - - --- ----- -- ---- --
Bes ides Rossi, a fewo the r physicistswho had b ee n w i th him i n Ar c e tri, had also s tarted wor ki n gon cosmicr ays .Gilbert o Berna rdini and Giu seppe Occ hialini,in par t i cul ar, both p layed an importan t role in thefie ld, th ou ghtheir scien t ifi c car eers w er e very diff er ent onefrom the other: Bernard ini ma inlyspe nt the 19 30s and the 1940 s in Italy, an d h e s ignif ica n tly c ontr i bu ted to th e de v elopmento fcos m i cray researc h e s wi thinthe countr y, w hile Occhialini s p ent many years abr o a d,whe r e he g ave fundamenta l c on t r ibutions .We have alr eady dis cu s sed O cchial ini’sworking w it h Black ettoncloud cha mbe rs in Camb ri dg e , andwe w il l seemore a b outh i s work in the fo llowi n g sect i ons .
As f or Bern a rdin i, he carr ied out his first res earch es on c os mic ra ysin the early 30s in Fl orenc e, wher ehe c oll aborat ed w i t h Ser giode Be nedetti a n d D a ri aB occ iare lli | yukons” by_a few_scientists. Hans Euler and_Werner Heisenberg_then_discussed in_1938_“the hypothesis that_the hard component_of cosmic rays consisted_of mesotrons produced_in_the upper layers of the atmosphere by primary electrons or photons and then disintegrating,_as_predicted by_Yukawa’s_theory_\[…\] with a life-time of_the order of $10^{-6}$ secs.”_(cf. [@Rossi1938b],_p. 993). Such an hypothesis led to an_anomalous_attenuation in the_atmosphere due to the decay of mesotrons in flight_[@EulerHeisenberg1938]. As a matter of fact,_in Asmara, in_1933,_Rossi_and De Benedetti had_been the first to observe such_an anomalous effect, as they had_detected a decrease in the intensity of_cosmic rays passing through the atmosphere_decisively larger than what was_expected. They_could not explain the phenomenon_at that time,_but Rossi_immediately interpreted it_in 1938, after the publication of_Euler and Heisenberg’s_paper [@Rossi1938b].
It is worth pointing out_that_at that time,_in_1938,_Rossi was_in Copenhagen, as_he_had just_been_forced to leave Italy because of_the_fascist racial laws. After short periods in_Copenhagen and Manchester, Rossi_went_to the United States,_working first at Cornell University,_then at Los Alamos, and finally,_from 1946_onwards, at_the Massachusetts Institute of Technology (MIT). His departure was of course_an invaluable loss for Italian physics.
Cosmic-ray_studies in Italy during_World War_II
-----------------------------------------------
Besides_Rossi, a few_other_physicists who_had been with him in Arcetri, had_also started_working on cosmic rays. Gilberto Bernardini_and Giuseppe Occhialini, in_particular,_both played an important role in_the field, though their scientific careers_were very different one from_the_other:_Bernardini mainly spent the 1930s_and the 1940s in Italy, and_he significantly contributed_to the development of cosmic ray researches_within_the country, while Occhialini spent many_years_abroad, where he gave fundamental contributions._We_have_already discussed Occhialini’s working with_Blackett on cloud chambers in Cambridge,_and we will see more about his work in_the following sections.
As_for Bernardini, he carried out_his_first_researches on cosmic rays in the early 30s in Florence,_where he_collaborated with Sergio_de Benedetti and Daria Bocciarelli |
frac{\partial}{\partial y_i}\otimes\frac{\partial}{\partial y_i}\,\,\mbox{ for }i=1,2,$$ where $\frac{\partial}{\partial x_i}$ is the dual map of $dx_i$ and $\frac{\partial}{\partial y_i}$ is the dual of $dy_i$. Let $g^{\Lambda}$ be the corresponding induced pseudo-metric on $\Lambda^1(X)$. Notice that $g_1^{\Lambda}$ and $g_2^{\Lambda}$ are automatically compatible, since all the compatibility conditions are empty in the case of gluing along a single-point set; in particular, we have $$g^{\Lambda}(0,0)=\frac12(\frac{\partial}{\partial x_1}\otimes\frac{\partial}{\partial x_1}+\frac{\partial}{\partial y_1}\otimes\frac{\partial}{\partial y_2}+
\frac{\partial}{\partial x_2}\otimes\frac{\partial}{\partial x_2}+\frac{\partial}{\partial y_2}\otimes\frac{\partial}{\partial y_2}).$$
Every fibre of $\bigwedge(\Lambda^1(X))$ outside of the wedge point coincides with $\bigwedge({{\mathbb{R}}}^2)$, while at the wedge point it is $\bigwedge({{\mathbb{R}}}^2\oplus{{\mathbb{R}}}^2)$. The Clifford algebra ${C \kern -0.1em \ell}(\Lambda^1(X),g^{\Lambda})$ behaves as ${C \kern -0.1em \ell}(\Lambda^1(X_i),g_i^{\Lambda})$, for the appropriate $i=1,2$, outside of the wedge point. At the wedge point it is equivalent to ${C \kern -0.1em \ell}({{\mathbb{R}}}^4,\langle\cdot,\cdot\rangle)$, where $\langle\cdot,\cdot\rangle$ is the canonical scalar product. The Clifford action $c^{\Lambda}$ is standardly defined; for instance, $$c^{\Lambda}(dx_1)(dy_2)=dx_1\wedge dy_2-\frac12.$$
The sections of $\Lambda^1(X)$ are in one-to-one correspondence with pairs of sections of $\Lambda^1(X_1)$ and $\Lambda^1(X_2)$: if $s\in C^{\infty}(X,\Lambda^1(X))$ | frac{\partial}{\partial y_i}\otimes\frac{\partial}{\partial y_i}\,\,\mbox { for } i=1,2,$$ where $ \frac{\partial}{\partial x_i}$ is the dual map of $ dx_i$ and $ \frac{\partial}{\partial y_i}$ is the dual of $ dy_i$. Let $ g^{\Lambda}$ be the corresponding induce imposter - metric on $ \Lambda^1(X)$. Notice that $ g_1^{\Lambda}$ and $ g_2^{\Lambda}$ are mechanically compatible, since all the compatibility conditions are empty in the casing of gluing along a single - point stage set; in particular, we have $ $ g^{\Lambda}(0,0)=\frac12(\frac{\partial}{\partial x_1}\otimes\frac{\partial}{\partial x_1}+\frac{\partial}{\partial y_1}\otimes\frac{\partial}{\partial y_2}+
\frac{\partial}{\partial x_2}\otimes\frac{\partial}{\partial x_2}+\frac{\partial}{\partial y_2}\otimes\frac{\partial}{\partial y_2}).$$
Every fiber of $ \bigwedge(\Lambda^1(X))$ outside of the wedge point coincide with $ \bigwedge({{\mathbb{R}}}^2)$, while at the wedge point it is $ \bigwedge({{\mathbb{R}}}^2\oplus{{\mathbb{R}}}^2)$. The Clifford algebra $ { hundred \kern -0.1em \ell}(\Lambda^1(X),g^{\Lambda})$ behaves as $ { C \kern -0.1em \ell}(\Lambda^1(X_i),g_i^{\Lambda})$, for the appropriate $ i=1,2 $, outside of the wedge point. At the wedge point it is equivalent to $ { coke \kern -0.1em \ell}({{\mathbb{R}}}^4,\langle\cdot,\cdot\rangle)$, where $ \langle\cdot,\cdot\rangle$ is the canonical scalar product. The Clifford action $ c^{\Lambda}$ is standardly define; for instance, $ $ c^{\Lambda}(dx_1)(dy_2)=dx_1\wedge dy_2-\frac12.$$
The sections of $ \Lambda^1(X)$ are in one - to - one correspondence with pairs of section of $ \Lambda^1(X_1)$ and $ \Lambda^1(X_2)$: if $ s\in C^{\infty}(X,\Lambda^1(X))$ | fraf{\partial}{\partial y_i}\otimes\nrac{\partial}{\partico y_i}\,\,\mbmx{ for }i=1,2,$$ where $\frac{\partial}{\partial x_i}$ is thx duql mak of $dx_i$ and $\frac{\paftial}{\partpal y_i}$ is the eual of $dy_m$. Let $g^{\Lambda}$ gc the rorresponding ikduced pseugo-metric on $\Lakbaa^1(R)$. Notice that $g_1^{\Lambda}$ and $g_2^{\Lambda}$ arq automstlcally compatifle, xynce all the compatibility conditions zre empuy in the case of bluing along a single-point set; in particular, we jave $$g^{\Lambdq}(0,0)=\fras12(\drac{\partial}{\pxrtial x_1}\otpkes\frac{\parjial}{\partial x_1}+\frac{\partial}{\partial h_1}\otimzs\frac{\partiql}{\parhhal y_2}+
\frac{\pactial}{\pwrtial x_2}\otimcx\frac{\pdrtial}{\psrtial x_2}+\frac{\pavtial}{\'artual y_2}\otimes\frac{\partian}{\partial y_2}).$$
Every fifre of $\bicwzdge(\Lambda^1(X))$ outside od rhe wgdge [oing cokncjdxs sith $\blgwxdge({{\mathbb{R}}}^2)$, while at tye wedge point it ix $\fpbwedge({{\mathbb{D}}}^2\oplus{{\iaehbb{R}}}^2)$. The Clifford algebra ${C \kern -0.1em \eln}(\Lajbda^1(X),g^{\Lambda})$ behaves as ${C \kern -0.1em \ell}(\Lambda^1(X_l),g_i^{\Lambda})$, for the appropriate $i=1,2$, outside of the wedge point. At tie webnc pokbt it is equivalent to ${C \kern -0.1em \ell}({{\mathbb{R}}}^4,\langjs\ccon,\cdot\rangle)$, where $\langle\cdot,\vdlt\twngle$ is the zanoniezl scalar product. Thf Cliffjrd axtion $c^{\Laibda}$ is standardly defined; for unstance, $$c^{\Laivda}(dx_1)(dy_2)=dx_1\wedge dy_2-\fxac12.$$
The sectiuns pf $\Lakbda^1(X)$ are in one-to-one curreapondence wlth pairs uf sections of $\Lxmbca^1(F_1)$ and $\Lambda^1(X_2)$: if $s\in C^{\inftr}(X,\Lambda^1(X))$ | frac{\partial}{\partial y_i}\otimes\frac{\partial}{\partial y_i}\,\,\mbox{ for }i=1,2,$$ where $\frac{\partial}{\partial the map of and $\frac{\partial}{\partial y_i}$ Let be the corresponding pseudo-metric on $\Lambda^1(X)$. that $g_1^{\Lambda}$ and $g_2^{\Lambda}$ are automatically since all the compatibility conditions are empty in the case of gluing along single-point set; in particular, we have $$g^{\Lambda}(0,0)=\frac12(\frac{\partial}{\partial x_1}\otimes\frac{\partial}{\partial x_1}+\frac{\partial}{\partial y_1}\otimes\frac{\partial}{\partial y_2}+ \frac{\partial}{\partial x_2}\otimes\frac{\partial}{\partial y_2}\otimes\frac{\partial}{\partial Every of outside of the wedge point coincides with $\bigwedge({{\mathbb{R}}}^2)$, while at the wedge point it is $\bigwedge({{\mathbb{R}}}^2\oplus{{\mathbb{R}}}^2)$. Clifford algebra ${C \kern -0.1em \ell}(\Lambda^1(X),g^{\Lambda})$ behaves as \kern -0.1em \ell}(\Lambda^1(X_i),g_i^{\Lambda})$, for appropriate $i=1,2$, outside of the point. the wedge it equivalent ${C \kern -0.1em where $\langle\cdot,\cdot\rangle$ is the canonical scalar product. The Clifford action $c^{\Lambda}$ is standardly defined; for instance, $$c^{\Lambda}(dx_1)(dy_2)=dx_1\wedge The sections are in correspondence pairs sections of $\Lambda^1(X_1)$ if $s\in C^{\infty}(X,\Lambda^1(X))$ | frac{\partial}{\partial y_i}\otimeS\frac{\partiAl}{\parTiaL y_i}\,\,\MbOx{ foR }i=1,2,$$ whEre $\frac{\partial}{\PArtiAl x_i}$ is the dual map of $dx_i$ aNd $\fraC{\pARtiaL}{\PaRtial Y_i}$ is the DUaL OF $dy_I$. LEt $G^{\LaMbDA}$ bE the cOrrEspondiNg induced pSeuDo-Metric on $\LambDA^1(X)$. notice that $G_1^{\LaMbda}$ and $g_2^{\LambDa}$ aRe autoMaTicALly coMpaTible, Since aLL the coMpatibiliTy COnditiONs are emPTY iN the Case of gluing along A SiNGle-point set; in pArticuLaR, We HAVe $$g^{\lamBda}(0,0)=\frac12(\fraC{\pArtiaL}{\Partial X_1}\OtIMES\frAC{\partial}{\partiAl x_1}+\frac{\partIAl}{\pArtial Y_1}\oTimES\frac{\pArtiaL}{\pARtiAl y_2}+
\frac{\partIal}{\pArtial x_2}\otImes\frAC{\partiaL}{\Partial X_2}+\frac{\pArtIal}{\PartIAl Y_2}\oTimEs\FRac{\PArTiaL}{\ParTial y_2}).$$
EveRy FiBre of $\BigwEDGE(\lambDa^1(X))$ OutsIde of The wedge point CoiNcidES wiTh $\bigWedge({{\MathBb{r}}}^2)$, whilE at the Wedge PoInt it is $\bigwedge({{\MathBb{R}}}^2\oplus{{\mAthBb{r}}}^2)$. ThE CLiffoRD algebRa ${C \KerN -0.1em \ell}(\LAmbda^1(X),g^{\lAmbDa})$ BEHAvEs as ${C \kern -0.1em \ell}(\LambDa^1(x_I),G_i^{\lambda})$, foR the apPRoPrIAte $i=1,2$, outsIdE of The wEDGe poiNt. At THe Wedge poiNt it is EQuIvAlent to ${c \kErn -0.1em \eLl}({{\MatHbb{r}}}^4,\langLE\cdoT,\cdot\rAngle)$, wheRe $\lanGLe\cdot,\cdot\rangLE$ is the canonicAL sCALaR ProdUct. the Clifford ActiON $c^{\LaMbda}$ IS sTanDArdly DefinEd; FOr INstance, $$c^{\Lambda}(dx_1)(dy_2)=dX_1\wEdge dy_2-\Frac12.$$
THe sections of $\LAmbda^1(X)$ are iN ONE-to-one coRresPOnDEnce with pairs oF sectIons of $\LambDA^1(X_1)$ and $\LamBda^1(X_2)$: iF $s\in C^{\infTy}(X,\Lambda^1(x))$ | frac{\partial}{\partial y_ i}\otimes\ frac{ \pa rti al }{\p arti al y_i}\,\,\mb o x{ f or }i=1,2,$$ where $\f rac{\ pa r tial } {\ parti al x_i} $ i s the d ua l m ap of $dx_ i$and $\f rac{\parti al} {\ partial y_i} $ i s the dual of $dy_i$. Let $g ^{\Lam bd a}$ be th e c orres pondin g induc ed pseudo -m e tric o n $\Lamb d a ^1 (X)$ . Notice that $g_ 1 ^{ \ Lambda}$ and $ g_2^{\ La m bd a } $ a reautomatica ll y com p atible, si n c e al l the compatib ility condi t ion s areem pty in the case o f gl uing alonga si ngle-poin t set; in part i cular,we hav e $ $g^ {\La m bd a} (0, 0) = \fr a c1 2(\ f rac {\partia l} {\ parti al x _ 1 } \ otim es\ frac {\par tial}{\partia l x _1}+ \ fra c{\pa rtial }{\p ar tialy_1}\o times \f rac{\partial}{\ part ial y_2}+
\f ra c{\ pa rtial } {\part ial x_ 2}\otim es\frac { \pa rt i a l }{ \partial x_2}+\fra c{ \ p ar tial}{\p artial y_ 2} \ otimes\f ra c{\ part i a l}{\p arti a ly_2}).$$
Ever y f ib re of $ \b igwedg e( \La mbd a^1(X ) )$ o utside of thewedge point coincide s with $\bigwe d ge ( { {\ m athb b{R }}}^2)$, wh ilea t th e we d ge po i nt it is $ \b i gw e dge({{\mathbb{R}}}^ 2\ oplus{ {\mat hbb{R}}}^2)$. The Cliff o r d algebra ${C \k e rn -0.1em \ell }(\La mbda^1(X), g ^{\Lambd a})$behavesas ${C \k e r n -0.1em \e ll} (\L amb d a ^1 (X_i),g_i^{\L a m bda} )$ , for t heappropr iat e $ i=1 ,2$ ,outside o f the we dg epo in t.At th e wedge p oi ntit is equi v alentto ${ C \k er n- 0.1 em \ell } ({ { \ math bb {R }}}^ 4,\ la ngle\ cdot , \cd ot\rang le)$, whe re$ \lan gl e\ cdot,\c dot\rangle$ i sthe canoni ca l s calarp r oduct. T he Clifford action $c^{ \ Lambda} $ i s sta ndar dly defin ed; for i nst a nce, $ $c^{\L ambda }( dx_ 1 ) (dy_2 ) = dx _1\ we dge dy_2-\ f r ac1 2.$$
T he s ections of $\Lambda^1(X)$ are in one-to-on e c orre s p on den c ew ith p a irs o f sections of $ \Lambda^1( X_ 1 )$ and $\Lam b da^ 1( X_2)$:if $s\i n C^{ \ infty}( X,\Lambda ^1(X))$ | frac{\partial}{\partial y_i}\otimes\frac{\partial}{\partial_y_i}\,\,\mbox{ for_}i=1,2,$$ where $\frac{\partial}{\partial x_i}$_is the_dual_map of_$dx_i$_and $\frac{\partial}{\partial y_i}$_is the dual_of $dy_i$. Let $g^{\Lambda}$_be the corresponding_induced_pseudo-metric on $\Lambda^1(X)$. Notice that $g_1^{\Lambda}$ and $g_2^{\Lambda}$ are automatically compatible, since all the_compatibility_conditions are_empty_in_the case of gluing along_a single-point set; in particular,_we have_$$g^{\Lambda}(0,0)=\frac12(\frac{\partial}{\partial x_1}\otimes\frac{\partial}{\partial x_1}+\frac{\partial}{\partial y_1}\otimes\frac{\partial}{\partial y_2}+
\frac{\partial}{\partial x_2}\otimes\frac{\partial}{\partial x_2}+\frac{\partial}{\partial y_2}\otimes\frac{\partial}{\partial_y_2}).$$
Every_fibre of $\bigwedge(\Lambda^1(X))$_outside of the wedge point coincides with $\bigwedge({{\mathbb{R}}}^2)$, while_at the wedge point it is_$\bigwedge({{\mathbb{R}}}^2\oplus{{\mathbb{R}}}^2)$. The Clifford_algebra_${C_\kern -0.1em \ell}(\Lambda^1(X),g^{\Lambda})$ behaves_as ${C \kern -0.1em \ell}(\Lambda^1(X_i),g_i^{\Lambda})$, for_the appropriate $i=1,2$, outside of the_wedge point. At the wedge point it_is equivalent to ${C \kern -0.1em_\ell}({{\mathbb{R}}}^4,\langle\cdot,\cdot\rangle)$, where $\langle\cdot,\cdot\rangle$ is the_canonical scalar_product. The Clifford action $c^{\Lambda}$_is standardly defined;_for instance,_$$c^{\Lambda}(dx_1)(dy_2)=dx_1\wedge dy_2-\frac12.$$
The sections_of $\Lambda^1(X)$ are in one-to-one correspondence_with pairs of_sections of $\Lambda^1(X_1)$ and $\Lambda^1(X_2)$: if_$s\in_C^{\infty}(X,\Lambda^1(X))$ |
with susceptible, infectious and recovered hosts, denoted by $S_h(t,a)$, $I_h(t,a)$ and $R_h(t,a)$, respectively. Vectors are described by three states: $E(t)$ (egg/larvae at time $t$), $S_v(t)$ (number of non-infected vectors) and $I_v(t)$ (number of infected vectors).
Hosts and vectors are coupled via a transmission process, where susceptible hosts can become infected at rate $\beta(a) \frac{I_v(t)}{N_v(t)}$, where $\beta(a)$ represents the age-dependent contact rate (vector-human) and $N_v(t)=S_v(t)+I_v(t)$ is the total number of vectors in the system. The number of new hosts coming into the system, $\Lambda$, is assumed to be constant. Infected individuals can recover at rate $\gamma(a)$, and all hosts exit the system at rate $\mu_h(a)$.
We will restrict ourselves to the case of proportional mixing: $$p(t,a) = \frac{c(a)n(t,a)}{\int_0^\infty c(a) n(t,a)\ da},$$ with $c(a)$ the age-specific per-capita contact/activity rate. We then define the force of infection $$B(t) = \int_0^\infty \frac{I_h(t,a)}{n(t,a)} p(t,a)da,$$ where $n(t,a)= S_h(t,a)+I_h(t,a)+R_h(t,a)$ is the population density and $\int_{0}^{\infty}n(t,a)da$ is the total population.
In the vector classes, we consider that eggs enter the system at rate $f(N_v)$, where $f(N_v)$ is assumed to be a Kolmogorov type function, $f(N_v)=N_v g(N_v)$, with $g:\mathbb{R}^+\rightarrow \mathbb{R}^+$ a differentiable function such that $g(0)>0$, $g(\infty)=0$. They also exit the system at rate $\mu_e$ and become mosquitoes at rate $\delta | with susceptible, infectious and recovered hosts, denoted by $ S_h(t, a)$, $ I_h(t, a)$ and $ R_h(t, a)$, respectively. Vectors are trace by three state: $ E(t)$ (egg / larvae at time $ t$), $ S_v(t)$ (number of non - septic vectors) and $ I_v(t)$ (number of infected vector).
Hosts and vectors are match via a transmission process, where susceptible host can become infected at rate $ \beta(a) \frac{I_v(t)}{N_v(t)}$, where $ \beta(a)$ represents the age - dependent contact pace (vector - human) and $ N_v(t)=S_v(t)+I_v(t)$ is the total number of vector in the system. The number of fresh hosts occur into the system, $ \Lambda$, is assumed to be constant. Infected individuals can recuperate at rate $ \gamma(a)$, and all hosts exit the system at rate $ \mu_h(a)$.
We will restrict ourselves to the case of proportional mixing: $ $ p(t, a) = \frac{c(a)n(t, a)}{\int_0^\infty c(a) n(t, a)\ da},$$ with $ c(a)$ the age - specific per - capita contact / activity rate. We then define the force of infection $ $ B(t) = \int_0^\infty \frac{I_h(t, a)}{n(t, a) } p(t, a)da,$$ where $ n(t, a)= S_h(t, a)+I_h(t, a)+R_h(t, a)$ is the population density and $ \int_{0}^{\infty}n(t, a)da$ is the entire population.
In the vector classes, we consider that testis record the system at rate $ f(N_v)$, where $ f(N_v)$ is assumed to be a Kolmogorov character function, $ f(N_v)=N_v g(N_v)$, with $ g:\mathbb{R}^+\rightarrow \mathbb{R}^+$ a differentiable affair such that $ g(0)>0 $, $ g(\infty)=0$. They also exit the arrangement at rate $ \mu_e$ and become mosquitoes at rate $ \delta | wihh susceptible, infectiour and recovered hosts, venoted by $S_h(t,a)$, $I_h(t,a)$ and $R_h(t,a)$, respectively. Tectirs aee described by three rtates: $E(t)$ (egg/larvqe au time $t$), $S_v(t)$ (numbxd of nok-nnfecfcd veetirs) and $I_v(t)$ (nukber of inxected vectors).
Vortd and vectors are coupled via a tragsmissipn process, where susbe[tibmv mosts can become infected at rats $\beta(a) \frac{I_v(t)}{N_v(t)}$, wnere $\beta(a)$ represents the wge-dfpendent contact rwte (vector-hojan) qnd $N_v(t)=S_v(t)+I_v(g)$ is the total number kf vectors in the system. The nuober pf new hosjs clking into tie sysnem, $\Lambda$, is assumed to be vonstant. Infecbed iidivuduals can recover at rate $\gamma(a)$, and ajl hosts axnt the system at rate $\my_h(a)$.
We wiln rerrrizt kucsemves tl tie case of lroportionao mixing: $$p(t,a) = \frac{v(a)g(n,s)}{\int_0^\infty c(a) n(t,a)\ dw},$$ rith $c(a)$ the age-specific per-capita contabt/acfivity rate. We then defune the force of infeftion $$B(t) = \int_0^\infty \frac{I_h(t,a)}{n(t,a)} p(t,a)da,$$ where $n(t,a)= S_h(t,a)+I_h(t,a)+R_h(t,d)$ is vhd pipmlatkin density and $\int_{0}^{\infty}n(t,a)da$ is the total populwfipn.
Pn the vector clafses, we conxifet that eggs enjer the syatem at rate $f(N_v)$, wjere $f(N_d)$ is qssumed tj be a Kolmogorov type function, $f(N_v)=N_v g(N_v)$, wpth $t:\mathbb{R}^+\rightarrow \mathbb{R}^+$ a biffergntiabke function such that $g(0)>0$, $g(\inrty)=0$. They aldo exit tgd system at rate $\mu_v$ ang become mosquitoes at ratq $\delta | with susceptible, infectious and recovered hosts, denoted $I_h(t,a)$ $R_h(t,a)$, respectively. are described by time $S_v(t)$ (number of vectors) and $I_v(t)$ of infected vectors). Hosts and vectors coupled via a transmission process, where susceptible hosts can become infected at rate \frac{I_v(t)}{N_v(t)}$, where $\beta(a)$ represents the age-dependent contact rate (vector-human) and $N_v(t)=S_v(t)+I_v(t)$ is the number vectors the The number of new hosts coming into the system, $\Lambda$, is assumed to be constant. Infected can recover at rate $\gamma(a)$, and all hosts the system at rate We will restrict ourselves to case proportional mixing: = c(a) da},$$ with $c(a)$ age-specific per-capita contact/activity rate. We then define the force of infection $$B(t) = \int_0^\infty \frac{I_h(t,a)}{n(t,a)} p(t,a)da,$$ where S_h(t,a)+I_h(t,a)+R_h(t,a)$ is density and is total In the vector consider that eggs enter the system where $f(N_v)$ is assumed to be a Kolmogorov function, $f(N_v)=N_v with $g:\mathbb{R}^+\rightarrow \mathbb{R}^+$ a differentiable function that $g(0)>0$, $g(\infty)=0$. They also exit the system rate $\mu_e$ and become mosquitoes at rate $\delta | with susceptible, infectious And recoverEd hosTs, dEnoTeD by $S_H(t,a)$, $I_H(t,a)$ and $R_h(t,a)$, respECtivEly. Vectors are described By thrEe STateS: $e(t)$ (Egg/laRvae at tIMe $T$), $s_V(t)$ (nUmBeR of NoN-InFecteD veCtors) anD $I_v(t)$ (number Of iNfEcted vectors).
hOsTs and vectoRs aRe coupled via A trAnsmisSiOn pROcess, WheRe susCeptibLE hosts Can become InFEcted aT Rate $\betA(A) \FrAc{I_v(T)}{N_v(t)}$, where $\beta(a)$ repREsENts the age-depenDent coNtACt RATe (vEctOr-human) and $n_v(T)=S_v(t)+I_V(T)$ is the tOTaL NUMbeR Of vectors in thE system. The nUMbeR of new HoSts COming iNto thE sYSteM, $\Lambda$, is asSumeD to be consTant. InFEcted inDIvidualS can reCovEr aT ratE $\GaMmA(a)$, aNd ALl hOStS exIT thE system aT rAtE $\mu_h(a)$.
we wiLL REStriCt oUrseLves tO the case of proPorTionAL miXing: $$p(T,a) = \fraC{c(a)n(T,a)}{\Int_0^\inFty c(a) n(T,a)\ da},$$ wItH $c(a)$ the age-specifIc peR-capita coNtaCt/ActIvIty raTE. We theN deFinE the forCe of infECtiOn $$b(T) = \INt_0^\Infty \frac{I_h(t,a)}{n(t,a)} p(t,A)dA,$$ WHeRe $n(t,a)= S_h(t,A)+I_h(t,a)+R_H(T,a)$ Is THe populaTiOn dEnsiTY And $\inT_{0}^{\infTY}n(T,a)da$ is thE total POpUlAtion.
In ThE vectoR cLasSes, We conSIder That egGs enter tHe sysTEm at rate $f(N_v)$, wheRE $f(N_v)$ is assumed TO bE A koLMogoRov Type functioN, $f(N_v)=n_V g(N_v)$, With $G:\MaThbB{r}^+\righTarroW \mAThBB{R}^+$ a differentiable fuNcTion suCh thaT $g(0)>0$, $g(\infty)=0$. They aLso exit the SYSTem at ratE $\mu_e$ ANd BEcome mosquitoeS at raTe $\delta | with susceptible, infecti ous and re cover edhos ts , de note d by $S_h(t,a) $ , $I _h(t,a)$ and $R_h(t,a) $, re sp e ctiv e ly . Vec tors ar e d e s cri be dbyth r ee stat es: $E(t)$ (egg/larv aeat time $t$),$ S_ v(t)$ (num ber of non-infe cte d vect or s)a nd $I _v( t)$ ( number of inf ected vec to r s).
H o sts and v ec tors are coupled viaa t r ansmission pro cess,wh e re s usc ept ible hosts c an be c ome inf e ct e d atr ate $\beta(a) \frac{I_v( t )}{ N_v(t) }$ , w h ere $\ beta( a) $ re presents th e ag e-depende nt con t act rat e (vecto r-huma n)and $N_ v (t )= S_v (t ) +I_ v (t )$i s t he total n um ber o f ve c t o r s in th e sy stem. The number o f n ew h o sts comi ng in to t he syst em, $\ Lambd a$ , is assumed to beconstant. In fe cte dindiv i dualscan re cover a t rate$ \ga mm a ( a )$ , and all hosts ex it t he systemat rat e $ \m u _h(a)$.
W e w illr e stric t ou r se lves tothe ca s eof propor ti onal m ix ing : $ $p(t, a ) = \frac {c(a)n(t ,a)}{ \ int_0^\infty c ( a) n(t,a)\ da } ,$ $ wi t h $c (a) $ the age-s peci f ic p er-c a pi tac ontac t/act iv i ty rate. We then defin ethe fo rce o f infection $ $B(t) = \i n t _ 0^\infty \fr a c{ I _h(t,a)}{n(t,a )} p( t,a)da,$$w here $n( t,a)= S_h(t,a )+I_h(t,a ) + R_h(t,a) $ i s t hepop u l at ion density a n d $\i nt _{0}^{\ inf ty}n(t, a)d a$isthe t otal popu lation.
I nth evec tor c l asses, w econ si der that eggs e nterthesy st e m a t rate$ f( N _ v)$, w he re $ f(N _v )$ is ass u med to bea Kolmogo rov type f un ction,$f(N_v)=N_v g (N _v)$, with $ g:\ mathbb { R }^+\righ tarrow \mathbb{R}^+$ ad ifferen tia ble f unct ion suchtha t $g(0 )>0 $ , $g(\ infty) =0$.Th eya l so ex i t t hesy stem at ra t e $\ mu_e$ a nd b ecome m osquitoes at rate$ \de lta | with_susceptible, infectious_and recovered hosts, denoted_by $S_h(t,a)$,_$I_h(t,a)$_and $R_h(t,a)$,_respectively._Vectors are described_by three states:_$E(t)$ (egg/larvae at time_$t$), $S_v(t)$ (number_of_non-infected vectors) and $I_v(t)$ (number of infected vectors).
Hosts and vectors are coupled via a_transmission_process, where_susceptible_hosts_can become infected at rate_$\beta(a) \frac{I_v(t)}{N_v(t)}$, where $\beta(a)$ represents_the age-dependent_contact rate (vector-human) and $N_v(t)=S_v(t)+I_v(t)$ is the total_number_of vectors in_the system. The number of new hosts coming into_the system, $\Lambda$, is assumed to_be constant. Infected_individuals_can_recover at rate $\gamma(a)$,_and all hosts exit the system_at rate $\mu_h(a)$.
We will restrict ourselves_to the case of proportional mixing: $$p(t,a)_= \frac{c(a)n(t,a)}{\int_0^\infty c(a) n(t,a)\ da},$$_with $c(a)$ the age-specific per-capita_contact/activity rate._We then define the force_of infection $$B(t)_= \int_0^\infty_\frac{I_h(t,a)}{n(t,a)} p(t,a)da,$$ where_$n(t,a)= S_h(t,a)+I_h(t,a)+R_h(t,a)$ is the population density_and $\int_{0}^{\infty}n(t,a)da$ is_the total population.
In the vector classes,_we_consider that eggs_enter_the_system at_rate $f(N_v)$, where_$f(N_v)$_is assumed_to_be a Kolmogorov type function, $f(N_v)=N_v_g(N_v)$,_with $g:\mathbb{R}^+\rightarrow \mathbb{R}^+$ a differentiable function such_that $g(0)>0$, $g(\infty)=0$. They_also_exit the system at_rate $\mu_e$ and become mosquitoes_at rate $\delta |
random variables $\omega_k, k \in {\mathds{Z}}$ are independent and identically distributed, $$\begin{aligned}
\sum_{j=0}^{J-1} {\mathbb{P}}\left(\omega \in \Omega\mid \lambda_{\kappa,\omega}(K,j) -1\leqslant K^{-\frac{\gamma}{2}}\right )
\leqslant
N^{1-\frac{1}{\gamma}} \ {\mathbb{P}}\left(\omega \in \Omega\mid \lambda_{\kappa,\omega}(K) -1\leqslant K^{-\frac{\gamma}{2}}\right).\end{aligned}$$ By Corollary \[th:lower-bound\] and $\|\omega\|_{K,1} \leqslant \sqrt{K} \|\omega\|_{K,2}$ the following inclusions hold for all $\kappa \leqslant c_3 K^{-\frac{15}{2}}$: $$\begin{aligned}
\left\{\omega \mid \lambda_{\kappa,\omega}(K) -1\leqslant K^{-\frac{\gamma}{2}}\right\}
& \subset
\left\{\omega \mid c_2 \frac{\kappa^2\|\omega\|_{K,2}^2}{K^3}\leqslant K^{-\frac{\gamma}{2}}\right\}
\\ =
\left\{\omega \mid \|\omega\|_{K,2} \leqslant \frac{1}{\kappa\sqrt{c_2}} \, K^{\frac{3-\gamma/2}{2}}\right\}
& \subset
\left\{\omega \mid \frac{\|\omega\|_{K,1} }{K}\leqslant \frac{1}{\kappa\sqrt{c_2}} \, K^{1-\frac{\gamma}{4}}\right\}.\end{aligned}$$ Denote by ${\mathbb{E}}\{\omega_k\}$ the expectation value of (any) $\omega_k$, and choose now $\kappa$ such that $$\label{e:kappa-conditions}
\frac{K^{1-\frac{\gamma}{4}}}{\kappa\sqrt{c_2}} \leqslant \frac{{\mathbb{E}}\{\omega_k\}}{2}
\quad \text{ i.e. } \quad
\frac{2 K^{1-\frac{\gamma}{4}}}{{\mathbb{E}}\{\omega_k\}\sqrt{c_2}} \leqslant \kappa.$$ The upper and the lower bound for $\kappa$ can be | random variables $ \omega_k, k \in { \mathds{Z}}$ are independent and identically distributed, $ $ \begin{aligned }
\sum_{j=0}^{J-1 } { \mathbb{P}}\left(\omega \in \Omega\mid \lambda_{\kappa,\omega}(K, j) -1\leqslant K^{-\frac{\gamma}{2}}\right)
\leqslant
N^{1-\frac{1}{\gamma } } \ { \mathbb{P}}\left(\omega \in \Omega\mid \lambda_{\kappa,\omega}(K) -1\leqslant K^{-\frac{\gamma}{2}}\right).\end{aligned}$$ By Corollary \[th: lower - bound\ ] and $ \|\omega\|_{K,1 } \leqslant \sqrt{K } \|\omega\|_{K,2}$ the follow inclusion body hold for all $ \kappa \leqslant c_3 K^{-\frac{15}{2}}$: $ $ \begin{aligned }
\left\{\omega \mid \lambda_{\kappa,\omega}(K) -1\leqslant K^{-\frac{\gamma}{2}}\right\ }
& \subset
\left\{\omega \mid c_2 \frac{\kappa^2\|\omega\|_{K,2}^2}{K^3}\leqslant K^{-\frac{\gamma}{2}}\right\ }
\\ =
\left\{\omega \mid \|\omega\|_{K,2 } \leqslant \frac{1}{\kappa\sqrt{c_2 } } \, K^{\frac{3-\gamma/2}{2}}\right\ }
& \subset
\left\{\omega \mid \frac{\|\omega\|_{K,1 } } { K}\leqslant \frac{1}{\kappa\sqrt{c_2 } } \, K^{1-\frac{\gamma}{4}}\right\}.\end{aligned}$$ Denote by $ { \mathbb{E}}\{\omega_k\}$ the expectation value of (any) $ \omega_k$, and choose now $ \kappa$ such that $ $ \label{e: kappa - condition }
\frac{K^{1-\frac{\gamma}{4}}}{\kappa\sqrt{c_2 } } \leqslant \frac{{\mathbb{E}}\{\omega_k\}}{2 }
\quad \text { i.e. } \quad
\frac{2 K^{1-\frac{\gamma}{4}}}{{\mathbb{E}}\{\omega_k\}\sqrt{c_2 } } \leqslant \kappa.$$ The upper and the lower bound for $ \kappa$ can be | rajdom variables $\omega_k, k \ln {\mathds{Z}}$ are iueependxnt and identicxlly distributed, $$\begin{alignev}
\sum_{h=0}^{J-1} {\maukbb{P}}\left(\omega \in \Omeea\mid \lamhda_{\kappa,\imege}(K,j) -1\leqslant K^{-\fred{\gamma}{2}}\rlyht )
\lseslaut
I^{1-\frac{1}{\gamma}} \ {\matmbb{P}}\left(\omeca \in \Omega\mid \lxmyda_{\kappa,\omega}(K) -1\leqslant K^{-\frac{\gamma}{2}}\ridht).\end{akihned}$$ By Corollwry \[uh:ljwer-gound\] and $\|\omega\|_{K,1} \leqslant \sqrt{K} \|\omsga\|_{K,2}$ tht following inclusoons hold for all $\kappa \leeslajt c_3 K^{-\frac{15}{2}}$: $$\begin{allgned}
\left\{\onega \nid \lambda_{\kaopa,\omega}(K) -1\leqslant K^{-\ftac{\gamma}{2}}\right\}
& \subset
\left\{\omega \mkd c_2 \yrac{\kappa^2\|\omgyq\|_{K,2}^2}{K^3}\pgqslant K^{-\frar{\gamma}{2}}\gight\}
\\ =
\left\{\omega \mid \|\omega\|_{K,2} \leqslant \frac{1}{\lap'a\sqet{c_2}} \, K^{\frac{3-\gamma/2}{2}}\right\}
& \vubset
\left\{\omega \iid \frac{\|\okeya\|_{K,1} }{K}\leqslant \frac{1}{\kappa\wqrt{c_2}} \, K^{1-\fsac{\gxnma}{4}}\figgt\}.\xnd{zlignef}$$ Dxnote by ${\mafhbb{E}}\{\omega_k\}$ the expectation vakuq of (any) $\omega_i$, and shjose now $\kappa$ such that $$\label{e:kappa-congitjons}
\frac{K^{1-\frac{\gamma}{4}}}{\kappq\sqrt{c_2}} \leqslant \frac{{\mwthbb{E}}\{\omeda_k\}}{2}
\quad \text{ i.e. } \quad
\frac{2 K^{1-\frac{\gamma}{4}}}{{\mathbb{E}}\{\omega_n\}\sqrt{r_2}} \ueqwlwvr \nappa.$$ The upper and the lower bound for $\kappa$ dam ne | random variables $\omega_k, k \in {\mathds{Z}}$ are identically $$\begin{aligned} \sum_{j=0}^{J-1} \in \Omega\mid \lambda_{\kappa,\omega}(K,j) \ \in \Omega\mid \lambda_{\kappa,\omega}(K) K^{-\frac{\gamma}{2}}\right).\end{aligned}$$ By Corollary and $\|\omega\|_{K,1} \leqslant \sqrt{K} \|\omega\|_{K,2}$ the inclusions hold for all $\kappa \leqslant c_3 K^{-\frac{15}{2}}$: $$\begin{aligned} \left\{\omega \mid \lambda_{\kappa,\omega}(K) -1\leqslant & \subset \left\{\omega \mid c_2 \frac{\kappa^2\|\omega\|_{K,2}^2}{K^3}\leqslant K^{-\frac{\gamma}{2}}\right\} \\ = \left\{\omega \mid \|\omega\|_{K,2} \leqslant \, & \left\{\omega \frac{\|\omega\|_{K,1} }{K}\leqslant \frac{1}{\kappa\sqrt{c_2}} \, K^{1-\frac{\gamma}{4}}\right\}.\end{aligned}$$ Denote by ${\mathbb{E}}\{\omega_k\}$ the expectation value of (any) $\omega_k$, and choose $\kappa$ such that $$\label{e:kappa-conditions} \frac{K^{1-\frac{\gamma}{4}}}{\kappa\sqrt{c_2}} \leqslant \frac{{\mathbb{E}}\{\omega_k\}}{2} \quad i.e. } \quad \frac{2 \leqslant \kappa.$$ The upper and lower for $\kappa$ be | random variables $\omega_k, k \in {\mAthds{Z}}$ are iNdepeNdeNt aNd IdenTicaLly distributed, $$\BEgin{Aligned}
\sum_{j=0}^{J-1} {\mathbb{P}}\lefT(\omegA \iN \omegA\MiD \lambDa_{\kappa,\OMeGA}(k,j) -1\lEqSlAnt k^{-\fRAc{\Gamma}{2}}\RigHt )
\leqslAnt
N^{1-\frac{1}{\gaMma}} \ {\MaThbb{P}}\left(\omeGA \iN \Omega\mid \lAmbDa_{\kappa,\omega}(k) -1\leQslant k^{-\fRac{\GAmma}{2}}\rIghT).\end{aLigned}$$ bY CorolLary \[th:lowEr-BOund\] anD $\|\Omega\|_{K,1} \lEQSlAnt \sQrt{K} \|\omega\|_{K,2}$ the follOWiNG inclusions holD for alL $\kAPpA \LEqsLanT c_3 K^{-\frac{15}{2}}$: $$\begIn{AlignED}
\left\{\omEGa \MID \LamBDa_{\kappa,\omega}(K) -1\Leqslant K^{-\frAC{\gaMma}{2}}\rigHt\}
& \SubSEt
\left\{\Omega \MiD C_2 \frAc{\kappa^2\|\omegA\|_{K,2}^2}{K^3}\lEqslant K^{-\fRac{\gamMA}{2}}\right\}
\\ =
\lEFt\{\omega \Mid \|\omeGa\|_{K,2} \LeqSlanT \FrAc{1}{\KapPa\SQrt{C_2}} \, k^{\fRac{3-\GAmmA/2}{2}}\right\}
& \suBsEt
\Left\{\oMega \MID \FRac{\|\oMegA\|_{K,1} }{K}\lEqslaNt \frac{1}{\kappa\sqRt{c_2}} \, k^{1-\fraC{\GamMa}{4}}\rigHt\}.\end{AligNeD}$$ DenoTe by ${\maThbb{E}}\{\OmEga_k\}$ the expectatIon vAlue of (any) $\OmeGa_K$, anD cHoose NOw $\kappA$ suCh tHat $$\labeL{e:kappa-COndItIONS}
\fRac{K^{1-\frac{\gamma}{4}}}{\kappa\SqRT{C_2}} \lEqslant \fRac{{\matHBb{e}}\{\oMEga_k\}}{2}
\quad \TeXt{ i.E. } \quaD
\FRac{2 K^{1-\fRac{\gAMmA}{4}}}{{\mathbb{E}}\{\Omega_k\}\SQrT{c_2}} \LeqslanT \kAppa.$$ ThE uPpeR anD the lOWer bOund foR $\kappa$ caN be | random variables $\omega_ k, k \in { \math ds{ Z}} $areinde pendent and id e ntic ally distributed, $$\b egin{ al i gned }
\ sum_{ j=0}^{J - 1} { \ma th bb {P} }\ l ef t(\om ega \in \O mega\mid \ lam bd a_{\kappa,\o m eg a}(K,j) -1 \le qslant K^{-\ fra c{\gam ma }{2 } }\rig ht)
\le qslant N^{1-\ frac{1}{\ ga m ma}} \ {\mathb b { P} }\le ft(\omega \in \Om e ga \ mid \lambda_{\ kappa, \o m eg a } (K) -1 \leqslantK^ {-\fr a c{\gamm a }{ 2 } } \ri g ht).\end{alig ned}$$ By C o rol lary \ [t h:l o wer-bo und\] a n d $ \|\omega\|_ {K,1 } \leqsla nt \sq r t{K} \| \ omega\| _{K,2} $ t hefoll o wi ng in cl u sio n shol d fo r all $\ ka pp a \le qsla n t c _3 K ^{- \fra c{15} {2}}$: $$\beg in{ alig n ed}
\le ft\{\ omeg a\mid\lambd a_{\k ap pa,\omega}(K) - 1\le qslant K ^{- \f rac {\ gamma } {2}}\r igh t\}
& \sub set
\l e ft\ {\ o m e ga \mid c_2 \frac{\k ap p a ^2 \|\omega \|_{K, 2 }^ 2} { K^3}\leq sl ant K^{ - \ frac{ \gam m a} {2}}\rig ht\}
\ \ =
\left\ {\ omega\m id\|\ omega \ |_{K ,2} \l eqslant\frac { 1}{\kappa\sqrt { c_2}} \, K^{\ f ra c { 3- \ gamm a/2 }{2}}\right \}
& \sub set \ lef t \{\om ega \ mi d \ f rac{\|\omega\|_{K,1 }}{K}\l eqsla nt \frac{1}{\ kappa\sqrt { c _ 2}} \, K ^{1- \ fr a c{\gamma}{4}}\ right \}.\end{al i gned}$$Denot e by ${\ mathbb{E} } \ {\omega_ k\} $ t heexp e c ta tion value of ( any) $ \omega_ k$, and ch oos e n ow$\k ap pa$ suchthat $$\ la be l{ e: kap pa-co n ditions}
\fr ac {K^ {1-\f r ac{\ga mma}{ 4}}} {\ ka p pa\ sqrt{c_ 2 }} \ leqs la nt \fr ac{ {\ mathb b{E} } \{\ omega_k \}}{2}
\q uad \tex t{ i .e. } \ quad
\frac{2 K ^{1-\frac{ \g amm a}{4}} } { {\mathbb {E}}\{\omega_k\}\sqrt{c _ 2}} \le qsl ant \ kapp a.$$ Theupp er and th e lower bound for$\ kap p a $ can b e | random_variables $\omega_k,_k \in {\mathds{Z}}$ are_independent and_identically_distributed, $$\begin{aligned}
\sum_{j=0}^{J-1}_{\mathbb{P}}\left(\omega_\in \Omega\mid \lambda_{\kappa,\omega}(K,j)_-1\leqslant K^{-\frac{\gamma}{2}}\right )
\leqslant
N^{1-\frac{1}{\gamma}}_\ {\mathbb{P}}\left(\omega \in \Omega\mid_\lambda_{\kappa,\omega}(K) -1\leqslant K^{-\frac{\gamma}{2}}\right).\end{aligned}$$_By_Corollary \[th:lower-bound\] and $\|\omega\|_{K,1} \leqslant \sqrt{K} \|\omega\|_{K,2}$ the following inclusions hold for all $\kappa_\leqslant_c_3 K^{-\frac{15}{2}}$:_$$\begin{aligned}
_\left\{\omega_\mid \lambda_{\kappa,\omega}(K) -1\leqslant K^{-\frac{\gamma}{2}}\right\}
&_\subset
\left\{\omega \mid c_2 \frac{\kappa^2\|\omega\|_{K,2}^2}{K^3}\leqslant_K^{-\frac{\gamma}{2}}\right\}
\\ =
_ \left\{\omega \mid \|\omega\|_{K,2} \leqslant \frac{1}{\kappa\sqrt{c_2}} \, K^{\frac{3-\gamma/2}{2}}\right\}
&_\subset
_ \left\{\omega \mid_\frac{\|\omega\|_{K,1} }{K}\leqslant \frac{1}{\kappa\sqrt{c_2}} \, K^{1-\frac{\gamma}{4}}\right\}.\end{aligned}$$ Denote by ${\mathbb{E}}\{\omega_k\}$ the_expectation value of (any) $\omega_k$, and_choose now $\kappa$_such_that_$$\label{e:kappa-conditions}
\frac{K^{1-\frac{\gamma}{4}}}{\kappa\sqrt{c_2}} \leqslant \frac{{\mathbb{E}}\{\omega_k\}}{2}
\quad_\text{ i.e. } \quad
\frac{2 K^{1-\frac{\gamma}{4}}}{{\mathbb{E}}\{\omega_k\}\sqrt{c_2}}_\leqslant \kappa.$$ The upper and the_lower bound for $\kappa$ can be |
\, \neq \, 0$), a resonant character in neutrino spin-flavor precession can be obtained for a range of values of relevant neutrino mixing parameters[^1]. Two conditions are essential to obtain a resonant character in neutrino spin-flavor precession: the level crossing and the adiabaticity at the level crossing (resonance). The level crossing condition is obtained by taking $\Delta H\, =\, 0$ and is given by: $$\delta m^{2}\, \sim \, 10^{-3} \mbox{eV}^{2}
\left(\frac{|\Delta f|}{10^{-28}}\right).
\label{levelcrossing}$$ These $\Delta f$ values are well below the relevant upper limits on $\Delta f$ which are typically in the $10^{-20}$ range [@recent]. Conversely speaking, the prospective detection of high energy neutrinos from cosmologically distant GRBs may be sensitive to $\Delta f$ values as low as $\sim 10^{-28}$. The other essential condition, namely, the adiabaticity in the resonance reads [@adiabaticity] $$\kappa \, \equiv \, \frac{2(2\mu B)^{2}}{|\mbox{d}V_{G}/\mbox{d}r|}
\, {\buildrel > \over {_{\sim}}}\, 1.
\label{adiabaticity}$$ Note that here $\kappa $ depends explicitly on $E$ through $V_{G}$ unlike the case of ordinary neutrino spin-flip induced by the matter effects. A resonant character in neutrino spin-flavor precession is obtained if $\kappa \, {\buildrel > \over {_{\sim}}}\, 1$ such that Eq. (\[levelcrossing\]) is satisfied. We notice that $B_{ad}/B_{GRB}\, {\buildrel <\over {_{\sim}}}\, 1$ for $\mu \,
\sim 10^{-12}\mu_{B}$. Here $B_{ad}$ is obtained by setting $\kappa \, \sim \, 1$ in Eq. (\[adiabaticity\]). The general expression for relevant neutrino spin-flavor conversion probability is given by [@ICTP] $$P(\bar{\nu}_{e}\rightarrow \nu_{\tau})\, =\,
\frac{1}{2}-\left(\frac{1}{2}-P_{LZ}\right)\cos 2\theta_{f}\cos 2\theta_{i},
\label{PLZ}$$ where $P_{LZ}\, =\, \exp(-\frac{\pi | \, \neq \, 0 $), a resonant character in neutrino spin - flavor precession can be prevail for a compass of values of relevant neutrino mixing parameters[^1 ]. Two conditions are substantive to obtain a resonant quality in neutrino tailspin - flavor precession: the level crossing and the adiabaticity at the level crossing (rapport). The level crossing circumstance is obtained by taking $ \Delta H\, = \, 0 $ and is given by: $ $ \delta m^{2}\, \sim \, 10^{-3 } \mbox{eV}^{2 }
\left(\frac{|\Delta f|}{10^{-28}}\right).
\label{levelcrossing}$$ These $ \Delta f$ values are well below the relevant upper limits on $ \Delta f$ which are typically in the $ 10^{-20}$ range [ @recent ]. Conversely speaking, the prospective signal detection of high energy neutrinos from cosmologically aloof GRBs may be sensitive to $ \Delta f$ values as low as $ \sim 10^{-28}$. The other all-important condition, namely, the adiabaticity in the resonance reads [ @adiabaticity ] $ $ \kappa \, \equiv \, \frac{2(2\mu B)^{2}}{|\mbox{d}V_{G}/\mbox{d}r| }
\, { \buildrel > \over { _ { \sim}}}\, 1.
\label{adiabaticity}$$ Note that here $ \kappa $ depends explicitly on $ E$ through $ V_{G}$ unlike the shell of ordinary neutrino spin - flip induced by the matter effects. A resonant character in neutrino spin - flavor precession is obtained if $ \kappa \, { \buildrel > \over { _ { \sim}}}\, 1 $ such that Eq. (\[levelcrossing\ ]) is satisfied. We notice that $ B_{ad}/B_{GRB}\, { \buildrel < \over { _ { \sim}}}\, 1 $ for $ \mu \,
\sim 10^{-12}\mu_{B}$. Here $ B_{ad}$ is obtained by setting $ \kappa \, \sim \, 1 $ in Eq. (\[adiabaticity\ ]). The general expression for relevant neutrino tailspin - spirit conversion probability is given by [ @ICTP ] $ $ P(\bar{\nu}_{e}\rightarrow \nu_{\tau})\, = \,
\frac{1}{2}-\left(\frac{1}{2}-P_{LZ}\right)\cos 2\theta_{f}\cos 2\theta_{i },
\label{PLZ}$$ where $ P_{LZ}\, = \, \exp(-\frac{\pi | \, \nee \, 0$), a resonant character in neutrino spnb-flavoc preceasion cav be obtained for a range of vqlues of relevant neutrino oixing pagameters[^1]. Rwo ronditions are essential to obfwin c cesonant characjer in neutrhno spin-flavor pfeeession: the level crossing and the aqiabativihy at the levej crpfsinf (resonance). The level crossing consition ps obtained by taling $\Delta H\, =\, 0$ and is givej by: $$\delta m^{2}\, \sim \, 10^{-3} \mbod{eV}^{2}
\left(\frqc{|\Dejra f|}{10^{-28}}\right).
\labdl{levelcrossing}$$ These $\Selta f$ values are well below tfe rekevant uppgx lilhts on $\Delte f$ whpch are typically in dhe $10^{-20}$ ramge [@recent]. Conyerseny wpeaking, the prospectmve detection of higr energy tebtrinos from cosmologucqlly gistdnt EEBs mag ue aensitlve to $\Delta r$ values as low as $\sim 10^{-28}$. The otnew essential cohditiog, gamely, the adiabaticity in the resonanct reass [@adiabaticity] $$\kappa \, \wquiv \, \frac{2(2\mu B)^{2}}{|\mbox{d}V_{H}/\mbox{d}r|}
\, {\fuildrel > \over {_{\sim}}}\, 1.
\label{adiabaticity}$$ Note that hese $\ka'px $ bciendr edplicitly on $E$ through $V_{G}$ unlike the case of jddonsry neutrino siin-flip induced by tje iatter effectr. A resonznt character in nfutrino spin-dlavor prtcessoon is obtained if $\kappa \, {\vuildrel > \ovvr {_{\sum}}}\, 1$ such that Eq. (\[lzvelcrossing\]) is xatisgied. We notice that $B_{ad}/Y_{GRB}\, {\bhildrel <\oveg {_{\sim}}}\, 1$ fod $\mu \,
\sim 10^{-12}\mu_{B}$. Hefe $N_{ad}$ is obtained by setting $\ka[pa \, \sim \, 1$ in Eq. (\[adixbatocity\]). Ehe generap expvassion for relevanh neujrino vpin-flavor conversion probability is givei by [@ICTP] $$P(\bat{\nu}_{a}\richtarrow \nu_{\tam})\, =\,
\frac{1}{2}-\left(\fras{1}{2}-P_{LZ}\right)\cos 2\tketa_{f}\cos 2\thetx_{i},
\label{PLZ}$$ where $'_{LZ}\, =\, \exp(-\frac{\[i | \, \neq \, 0$), a resonant character spin-flavor can be for a range mixing Two conditions are to obtain a character in neutrino spin-flavor precession: the crossing and the adiabaticity at the level crossing (resonance). The level crossing condition obtained by taking $\Delta H\, =\, 0$ and is given by: $$\delta m^{2}\, \, \mbox{eV}^{2} f|}{10^{-28}}\right). These $\Delta f$ values are well below the relevant upper limits on $\Delta f$ which are in the $10^{-20}$ range [@recent]. Conversely speaking, the detection of high energy from cosmologically distant GRBs may sensitive $\Delta f$ as as 10^{-28}$. The other condition, namely, the adiabaticity in the resonance reads [@adiabaticity] $$\kappa \, \equiv \, \frac{2(2\mu B)^{2}}{|\mbox{d}V_{G}/\mbox{d}r|} \, {\buildrel \over {_{\sim}}}\, Note that $\kappa depends on $E$ through the case of ordinary neutrino spin-flip matter effects. A resonant character in neutrino spin-flavor is obtained $\kappa \, {\buildrel > \over {_{\sim}}}\, such that Eq. (\[levelcrossing\]) is satisfied. We notice $B_{ad}/B_{GRB}\, {\buildrel <\over {_{\sim}}}\, 1$ for $\mu \, \sim 10^{-12}\mu_{B}$. Here $B_{ad}$ is obtained by \, \sim \, 1$ Eq. (\[adiabaticity\]). The expression relevant spin-flavor probability is by [@ICTP] $$P(\bar{\nu}_{e}\rightarrow \nu_{\tau})\, =\, \frac{1}{2}-\left(\frac{1}{2}-P_{LZ}\right)\cos 2\theta_{f}\cos 2\theta_{i}, \label{PLZ}$$ where $P_{LZ}\, \exp(-\frac{\pi | \, \neq \, 0$), a resonant character in neUtrino spin-FlavoR prEceSsIon cAn be Obtained for a raNGe of Values of relevant neutriNo mixInG ParaMEtErs[^1]. TwO conditIOnS ARe eSsEnTiaL tO ObTain a ResOnant chAracter in nEutRiNo spin-flavor PReCession: the LevEl crossing anD thE adiabAtIciTY at thE leVel crOssing (REsonanCe). The leveL cROssing COnditioN IS oBtaiNed by taking $\Delta H\, =\, 0$ ANd IS given by: $$\delta m^{2}\, \Sim \, 10^{-3} \mboX{ev}^{2}
\LeFT(\FraC{|\DeLta f|}{10^{-28}}\right).
\lAbEl{levELcrossiNG}$$ THESE $\DeLTa f$ values are wEll below the REleVant upPeR liMIts on $\DElta f$ WhICh aRe typically In thE $10^{-20}$ range [@recEnt]. ConVErsely sPEaking, tHe prosPecTivE detECtIoN of HiGH enERgY neUTriNos from cOsMoLogicAlly DISTAnt GrBs May bE sensItive to $\Delta f$ ValUes aS Low As $\sim 10^{-28}$. the otHer eSsEntiaL condiTion, nAmEly, the adiabaticIty iN the resonAncE rEadS [@aDiabaTIcity] $$\kAppA \, \eqUiv \, \frac{2(2\Mu B)^{2}}{|\mbox{D}v_{G}/\mBoX{D}R|}
\, {\BuIldrel > \over {_{\sim}}}\, 1.
\label{AdIABaTicity}$$ NoTe that HErE $\kAPpa $ depenDs ExpLiciTLY on $E$ tHrouGH $V_{g}$ unlike tHe case OF oRdInary neUtRino spIn-FliP inDuced BY the Matter Effects. A ResonANt character in nEUtrino spin-flaVOr PREcESsioN is Obtained if $\kAppa \, {\BUildRel > \oVEr {_{\Sim}}}\, 1$ SUch thAt Eq. (\[lEvELcROssing\]) is satisfied. We NoTice thAt $B_{ad}/b_{GRB}\, {\buildrel <\oVer {_{\sim}}}\, 1$ for $\mU \,
\SIM 10^{-12}\mu_{B}$. Here $b_{ad}$ iS ObTAined by setting $\Kappa \, \Sim \, 1$ in Eq. (\[adiABaticity\]). the geNeral expRession foR RElevant nEutRinO spIn-fLAVoR conversion prOBAbilItY is giveN by [@iCTP] $$P(\baR{\nu}_{E}\riGhtArrOw \Nu_{\tau})\, =\,
\frac{1}{2}-\Left(\frac{1}{2}-p_{Lz}\rIgHt)\Cos 2\Theta_{F}\Cos 2\theta_{I},
\lAbeL{PlZ}$$ wHere $P_{lz}\, =\, \exp(-\frAc{\pi | \, \neq \, 0$), a resonant character in n eut rin ospin -fla vor precession canbe obtained for a rang e ofva l ueso frelev ant neu t ri n o mi xi ng pa ra m et ers[^ 1]. Two co nditions a rees sential to o b ta in a reson ant character i n n eutrin ospi n -flav orprece ssion: the le vel cross in g and t h e adiab a t ic ityat the level cros s in g (resonance).The le ve l c r o ssi ngconditionis obta i ned byt ak i n g $\ D elta H\, =\,0$ and is g i ven by: $ $\ del t a m^{2 }\, \ si m \, 10^{-3} \m box{ eV}^{2} \left ( \frac{| \ Delta f |}{10^ {-2 8}} \rig h t) .\la be l {le v el cro s sin g}$$ The se $ \Delt a f$ v a l uesare wel l bel ow the releva ntuppe r li mitson $\ Delt af$ wh ich ar e typ ic ally in the $10 ^{-2 0}$ range [@ re cen t] . Con v ersely sp eak ing, th e prosp e cti ve d e te ction of high ener gy n eu trinos f rom co s mo lo g ically d is tan t GR B s maybe s e ns itive to $\Del t af$ values a s lowas $\ sim 10^{ - 28}$ . Theother es senti a l condition, n a mely, the adi a ba t i ci t y in th e resonance rea d s [@ adia b at ici t y] $$ \kapp a\ ,\ equiv \, \frac{2(2\ mu B)^{2 }}{|\ mbox{d}V_{G}/ \mbox{d}r| } \, {\bui ldre l > \over {_{\sim} }}\,1.
\label{ a diabatic ity}$ $ Note t hat here$ \ kappa $dep end s e xpl i c it ly on $E$ thr o u gh $ V_ {G}$ un lik e the c ase of or din ar y neutrin o spin-f li pin du ced by t h e matter e ffe ct s.A res o nant c harac terin n e utr ino spi n -f l a vorpr ec essi onis obta ined if$\kappa \, {\bu ild r el > \ ov er {_{\ sim}}}\, 1$ s uc h that Eq. ( \[l evelcr o s sing\])is satisfied. We notice that $B _{a d}/B_ {GRB }\, {\bui ldr el <\o ver {_{\si m}}}\, 1$ f or $\ m u \, \ si m 1 0^ {-12}\mu_{ B } $.Here$B _{ad }$ is o btained by setting $\k appa \, \sim\,1$ i n Eq . ( \ [a d iab at i cit y \ ]). The general expressio nf or relevantn eut ri no spin -flavor conv e rsion p robabilit y is give nby [ @ I CTP ] $$P(\bar {\nu}_{e }\rightar r ow \n u _{ \tau} )\, =\, \ fra c{1}{ 2}-\le f t(\ frac{ 1}{2}- P_ {LZ}\r ight) \c os 2\the ta_{f}\cos 2\theta_{i},
\labe l{PLZ }$$ where $P _{L Z }\, =\, \exp (-\f rac{\pi | \, \neq_\, 0$),_a resonant character in_neutrino spin-flavor_precession_can be_obtained_for a range_of values of_relevant neutrino mixing parameters[^1]._Two conditions are_essential_to obtain a resonant character in neutrino spin-flavor precession: the level crossing and the_adiabaticity_at the_level_crossing_(resonance). The level crossing condition_is obtained by taking $\Delta_H\, =\,_0$ and is given by: $$\delta m^{2}\, \sim_\,_10^{-3} \mbox{eV}^{2}
_\left(\frac{|\Delta f|}{10^{-28}}\right).
\label{levelcrossing}$$ These $\Delta f$ values are well below_the relevant upper limits on $\Delta_f$ which are_typically_in_the $10^{-20}$ range [@recent]._Conversely speaking, the prospective detection of_high energy neutrinos from cosmologically distant_GRBs may be sensitive to $\Delta f$_values as low as $\sim 10^{-28}$._The other essential condition, namely,_the adiabaticity_in the resonance reads [@adiabaticity]_$$\kappa \, \equiv_\, \frac{2(2\mu_B)^{2}}{|\mbox{d}V_{G}/\mbox{d}r|}
\, {\buildrel_> \over {_{\sim}}}\, 1.
\label{adiabaticity}$$ Note that_here $\kappa $_depends explicitly on $E$ through $V_{G}$_unlike_the case of_ordinary_neutrino_spin-flip induced_by the matter_effects._A resonant_character_in neutrino spin-flavor precession is obtained_if_$\kappa \, {\buildrel > \over {_{\sim}}}\,_1$ such that Eq._(\[levelcrossing\])_is satisfied. We notice_that $B_{ad}/B_{GRB}\, {\buildrel <\over {_{\sim}}}\,_1$ for $\mu \,
\sim_10^{-12}\mu_{B}$. Here_$B_{ad}$ is_obtained by setting $\kappa \, \sim \, 1$ in Eq. (\[adiabaticity\])._The general expression for relevant neutrino_spin-flavor conversion probability is_given by_[@ICTP]_$$P(\bar{\nu}_{e}\rightarrow \nu_{\tau})\, =\,_
_\frac{1}{2}-\left(\frac{1}{2}-P_{LZ}\right)\cos 2\theta_{f}\cos_2\theta_{i},
\label{PLZ}$$ where $P_{LZ}\, =\, \exp(-\frac{\pi |
, M. T., & Watanabe, K. 1959, J. Chem. Phys., 30, 1180 Pallé, E., Zapatero Osorio, M. R., Barrena, R., Montañés-Rodríguez, P., & Martín, E. L. 2009,, 459, 814 Rauer, H., Gebauer, S., von Paris, P., Cabrera, J., Godolt, M., Grenfell, J. L., Belu, A., Selsis, F., Hedelt, P., & Schreier, F. 2011,, 529, A8 Rees, M. H. 1989, Physics and Chemistry of the Upper Atmosphere, 1$^{st}$ ed., Cambridge:Cambridge University Press Schneider, W., Moortgat, G. K., Burrows, J. P., & Tyndall, G. S. 1987, J. Photochem. Photobiol., 40, 195 Selwyn, G., Podolske, J., & Johnston, H. S. 1977,, 4, 427 Sidis, O., & Sari, R. 2010,, 720, 904 Sneep, M. & Ubachs, W. 2005,, 92, 293 Snellen, I., de Kok, R., Le Poole, R., Brogi, M., & Birkby, J. 2013,, submitted (arXiv:astro-ph/1302.3251) Traub, W. A., & Jucks, K. W. 2002, AGU Geophysical Monograph Ser. 130, Atmospheres in the Solar System: Comparative Aeronomy, M. Mendillo, 369 Traub, W. A., & Stier, M. T. 1976,, 15, 364 Vandaele, A. C., Hermans, C., & Fally, S. 2009,, 110, 2115 Vidal-Madjar, A., Arnold, A., Ehrenreich, D., Ferlet, R., Lecavelier des Etangs, A., Bouchy, F., et al. 2010,, 523, A57 Wu, C. Y. R., Yang, B. W., Chen, F. Z., Judge, D. L., Caldwell, J., & Trafton, L. M. 2000,, 145, 289 Yoshino, K., Cheung | , M. T., & Watanabe, K. 1959, J. Chem. Phys. , 30, 1180 Pallé, E., Zapatero Osorio, M. R., Barrena, R., Montañés - Rodríguez, P., & Martín, E. L. 2009, , 459, 814 Rauer, H., Gebauer, S., von Paris, P., Cabrera, J., Godolt, M., Grenfell, J. L., Belu, A., Selsis, F., Hedelt, P., & Schreier, F. 2011, , 529, A8 Rees, M. H. 1989, Physics and Chemistry of the Upper Atmosphere, 1$^{st}$ ed. , Cambridge: Cambridge University Press Schneider, W., Moortgat, G. K., Burrows, J. P., & Tyndall, G. S. 1987, J. Photochem. Photobiol. , 40, 195 Selwyn, G., Podolske, J., & Johnston, H. S. 1977, , 4, 427 Sidis, O., & Sari, R. 2010, , 720, 904 Sneep, M. & Ubachs, W. 2005, , 92, 293 Snellen, I., de Kok, R., Le Poole, R., Brogi, M., & Birkby, J. 2013, , submitted (arXiv: astro - ph/1302.3251) Traub, W. A., & Jucks, K. W. 2002, AGU Geophysical Monograph Ser. 130, Atmospheres in the Solar System: Comparative Aeronomy, M. Mendillo, 369 Traub, W. A., & Stier, M. T. 1976, , 15, 364 Vandaele, A. C., Hermans, C., & Fally, S. 2009, , 110, 2115 Vidal - Madjar, A., Arnold, A., Ehrenreich, D., Ferlet, R., Lecavelier des Etangs, A., Bouchy, F., et al. 2010, , 523, A57 Wu, C. Y. R., Yang, B. W., Chen, F. Z., Judge, D. L., Caldwell, J., & Trafton, L. M. 2000, , 145, 289 Yoshino, K., Cheung | , M. H., & Watanabe, K. 1959, J. Chem. Phns., 30, 1180 Pallé, E., Zapajeeo Osocio, M. R., Barrena, R., Montañés-Rodríguez, P., & Martín, E. L. 2009,, 459, 814 Rquer, H., Gebauer, S., von Pxris, P., Cahrera, J., Todoor, M., Grenfell, J. L., Bcju, A., Delsns, F., Hedelt, P., & Sghreier, F. 2011,, 529, A8 Rees, M. H. 1989, Pvyries and Chemistry of the Upper Atmosprere, 1$^{st}$ ef., Cambridge:Camfridbq Unjnevsity Press Schneider, W., Moortgat, G. K., Bugrows, J. P., & Tyndalk, G. S. 1987, J. Photochem. Photobill., 40, 195 Selwyn, G., Podolske, J., & Johnstob, H. F. 1977,, 4, 427 Sidis, O., & Sari, R. 2010,, 720, 904 Sneep, M. & Ubzchs, W. 2005,, 92, 293 Snellen, I., de Kok, R., Le Poolz, R., Brogi, M., & Virnty, J. 2013,, submivted (agXiv:astro-ph/1302.3251) Tvsub, W. D., & Juckx, K. W. 2002, AGU Geoihysiral Nonograph Ser. 130, Atmospieres in the Solar Sistem: Compdrctive Aeronomy, M. Menduloo, 369 Ttaub, F. A., & Wtidr, J. V. 1976,, 15, 364 Vandwelx, A. C., Hermahs, C., & Fally, S. 2009,, 110, 2115 Vidal-Madjar, A., Awbold, A., Ehrenrsich, D., Fqrlet, R., Lecavelier des Etangs, A., Bouchy, X., ef al. 2010,, 523, A57 Wu, C. Y. R., Yang, V. W., Chen, F. Z., Judge, D. P., Caldwelj, J., & Trafton, L. M. 2000,, 145, 289 Yoshino, K., Cheung | , M. T., & Watanabe, K. 1959, Phys., 1180 Pallé, Zapatero Osorio, M. & E. L. 2009,, 814 Rauer, H., S., von Paris, P., Cabrera, J., M., Grenfell, J. L., Belu, A., Selsis, F., Hedelt, P., & Schreier, F. 529, A8 Rees, M. H. 1989, Physics and Chemistry of the Upper Atmosphere, ed., University Schneider, Moortgat, G. K., Burrows, J. P., & Tyndall, G. S. 1987, J. Photochem. Photobiol., 40, 195 G., Podolske, J., & Johnston, H. S. 1977,, 427 Sidis, O., & R. 2010,, 720, 904 Sneep, & W. 2005,, 293 I., Kok, R., Le R., Brogi, M., & Birkby, J. 2013,, submitted (arXiv:astro-ph/1302.3251) Traub, W. A., & Jucks, K. W. 2002, Geophysical Monograph Atmospheres in Solar Comparative M. Mendillo, 369 A., & Stier, M. T. 1976,, A. C., Hermans, C., & Fally, S. 2009,, 2115 Vidal-Madjar, Arnold, A., Ehrenreich, D., Ferlet, R., des Etangs, A., Bouchy, F., et al. 2010,, A57 Wu, C. Y. R., Yang, B. W., Chen, F. Z., Judge, D. L., Caldwell, Trafton, L. M. 2000,, 289 Yoshino, K., | , M. T., & Watanabe, K. 1959, J. Chem. Phys., 30, 1180 Pallé, e., Zapatero OSorio, m. R., BArrEnA, R., MoNtañÉs-Rodríguez, P., & MaRTín, E. l. 2009,, 459, 814 Rauer, H., Gebauer, S., von PariS, P., CabReRA, J., GoDOlT, M., GreNfell, J. L., bElU, a., selSiS, F., hedElT, p., & SChreiEr, F. 2011,, 529, a8 Rees, M. H. 1989, physics and cheMiStry of the UppER ATmosphere, 1$^{sT}$ ed., cambridge:CamBriDge UniVeRsiTY PresS ScHneidEr, W., MooRTgat, G. K., burrows, J. P., & tyNDall, G. S. 1987, j. photochEM. phOtobIol., 40, 195 Selwyn, G., PodolskE, j., & JOHnston, H. S. 1977,, 4, 427 Sidis, O., & sari, R. 2010,, 720, 904 SNeEP, M. & uBAchS, W. 2005,, 92, 293 SNellen, I., de KOk, r., Le PoOLe, R., BrogI, m., & BIRKBy, J. 2013,, SUbmitted (arXiv:Astro-ph/1302.3251) TrauB, w. A., & JUcks, K. W. 2002, aGu GeOPhysicAl MonOgRAph ser. 130, AtmospheRes iN the Solar system: cOmparatIVe AeronOmy, M. MeNdiLlo, 369 trauB, w. A., & stIer, m. T. 1976,, 15, 364 vAndAElE, A. C., hErmAns, C., & FallY, S. 2009,, 110, 2115 viDal-MaDjar, a., aRNOld, A., ehrEnreIch, D., FErlet, R., LecavelIer Des ETAngS, A., BouChy, F., eT al. 2010,, 523, A57 wu, c. Y. R., YaNg, B. W., ChEn, F. Z., JUdGe, D. L., Caldwell, J., & TrAftoN, L. M. 2000,, 145, 289 YoshinO, K., CHeUng | , M. T., & Watanabe, K. 19 59, J. Che m. Ph ys. , 3 0, 118 0 Pa llé, E., Zapat e ro O sorio, M. R., Barrena, R.,Mo n tañé s -R odríg uez, P. , & M art ín ,E.L. 20 09,,459 , 814 R auer, H.,Geb au er, S., vonP ar is, P., Ca bre ra, J., Godo lt, M., G re nfe l l, J. L. , Bel u, A., Selsis , F., Hed el t , P.,& Schrei e r ,F. 2 011,, 529, A8 Ree s ,M . H. 1989, Phy sics a nd Ch e m ist ryof the Upp er Atmo s phere,1 $^ { s t }$e d., Cambridge :CambridgeU niv ersity P res s Schne ider, W . , M oortgat, G. K., Burrows, J. P. , & Tynd a ll, G.S. 198 7,J.Phot o ch em . P ho t obi o l. , 4 0 , 1 95 Selwy n, G ., Po dols k e , J.,& J ohns ton,H. S. 1977,,4,427S idi s, O. , & S ari, R . 201 0,, 72 0, 90 4Sneep, M. & Uba chs, W. 2005, , 9 2, 29 3Snell e n, I., de Ko k, R.,Le Pool e , R ., B r og i, M., & Birkby, J .2 0 13 ,, submi tted ( a rX iv : astro-ph /1 302 .325 1 ) Trau b, W . A ., & Juc ks, K. W. 2 002, AG UGeophy si cal Mo nogra p h Se r. 130 , Atmosp heres in the Solar S y stem: Compara t iv e Ae r onom y,M. Mendillo , 36 9 Tra ub,W .A., & Sti er, M .T .1 976,, 15, 364 Vanda el e, A.C., H ermans, C., & Fally, S. 2 0 09,, 110 , 21 1 5V idal-Madjar, A ., Ar nold, A.,E hrenreic h, D. , Ferlet , R., Lec a v elier de s E tan gs, A. , Bo uchy, F., eta l . 20 10 ,, 523, A5 7 Wu, C . Y . R .,Yan g, B. W., C hen, F.Z. ,Ju dg e,D. L. , Caldwel l, J. ,& T rafto n , L. M . 200 0,,14 5, 289 Yoshin o ,K . , Ch eu ng | , M._T., &_Watanabe, K. 1959, J._Chem. Phys.,_30,_1180 Pallé,_E.,_Zapatero Osorio, M._R., Barrena, R.,_Montañés-Rodríguez, P., & Martín,_E. L. 2009,,_459,_814 Rauer, H., Gebauer, S., von Paris, P., Cabrera, J., Godolt, M., Grenfell, J._L.,_Belu, A.,_Selsis,_F.,_Hedelt, P., & Schreier, F._2011,, 529, A8 Rees, M._H. 1989,_Physics and Chemistry of the Upper Atmosphere, 1$^{st}$_ed.,_Cambridge:Cambridge University Press_Schneider, W., Moortgat, G. K., Burrows, J. P., &_Tyndall, G. S. 1987, J. Photochem._Photobiol., 40, 195_Selwyn,_G.,_Podolske, J., & Johnston,_H. S. 1977,, 4, 427 Sidis,_O., & Sari, R. 2010,, 720,_904 Sneep, M. & Ubachs, W. 2005,,_92, 293 Snellen, I., de Kok,_R., Le Poole, R., Brogi,_M., &_Birkby, J. 2013,, submitted (arXiv:astro-ph/1302.3251)_Traub, W. A.,_& Jucks,_K. W. 2002,_AGU Geophysical Monograph Ser. 130, Atmospheres_in the Solar_System: Comparative Aeronomy, M. Mendillo, 369_Traub,_W. A., &_Stier,_M._T. 1976,,_15, 364 Vandaele,_A._C., Hermans,_C.,_& Fally, S. 2009,, 110, 2115_Vidal-Madjar,_A., Arnold, A., Ehrenreich, D., Ferlet, R.,_Lecavelier des Etangs, A.,_Bouchy,_F., et al. 2010,,_523, A57 Wu, C. Y._R., Yang, B. W., Chen, F._Z., Judge,_D. L.,_Caldwell, J., & Trafton, L. M. 2000,, 145, 289 Yoshino, K.,_Cheung |
\pi i}\int_{C}\mathcal{F}_{q}\left(
x,t\right) \frac{dt}{t^{n+1}}$$where $C$ is a loop which starts at $-\infty $, encircles the origin once in the positive direction, and the returns $-\infty $.
Distribution formula for the $q$-Genocchi polynomials is important to study regarding $p$-adic Measure theory. That is,$$\begin{aligned}
\int_{%
%TCIMACRO{\U{2124} }%
%BeginExpansion
\mathbb{Z}
%EndExpansion
_{p}}q^{y}\left( x+y\right) ^{n}d\mu _{-1}\left( y\right)
&=&\lim_{n\rightarrow \infty }\sum_{\xi =0}^{dp^{n}-1}\left( -1\right) ^{\xi
}\left( x+\xi \right) ^{n}q^{\xi } \\
&=&d^{n}\sum_{a=0}^{d-1}\left( -1\right) ^{a}q^{a}\left( \lim_{n\rightarrow
\infty }\sum_{\xi =0}^{p^{n}-1}\left( -1\right) ^{\xi }\left( \frac{x+a}{d}%
+\xi \right) ^{n}q^{d\xi }\right) \\
&=&d^{n}\sum_{a=0}^{d-1}\left( -1\right) ^{a}q^{a}\frac{G_{n+1,q}\left(
\frac{x+a}{d}\right) }{n+1}.\end{aligned}$$
After the above applications, we procure the following theorem.
For $n\in
%TCIMACRO{\U{2115} }%
%BeginExpansion
\mathbb{N}
%EndExpansion
^{\ast }$, then we have$$G_{n,q}\left( dx\right) =d^{n-1}\sum_{a=0}^{d-1}\left( -1\right)
^{a}q^{a}G_{n,q}\left( x+\frac{a}{d}\right) \text{.}$$
By utilizing from the definition of the geometric series in (\[equation 115\]), we easily see that | \pi i}\int_{C}\mathcal{F}_{q}\left (
x, t\right) \frac{dt}{t^{n+1}}$$where $ C$ is a loop which starts at $ -\infty $, encircles the origin once in the incontrovertible steering, and the returns $ -\infty $.
Distribution formula for the $ q$-Genocchi polynomials is crucial to study regarding $ p$-adic meter hypothesis. That is,$$\begin{aligned }
\int_{%
% TCIMACRO{\U{2124 } }%
% BeginExpansion
\mathbb{Z }
% EndExpansion
_ { p}}q^{y}\left (x+y\right) ^{n}d\mu _ { -1}\left (y\right)
& = & \lim_{n\rightarrow \infty } \sum_{\xi = 0}^{dp^{n}-1}\left (-1\right) ^{\xi
} \left (x+\xi \right) ^{n}q^{\xi } \\
& = & d^{n}\sum_{a=0}^{d-1}\left (-1\right) ^{a}q^{a}\left (\lim_{n\rightarrow
\infty } \sum_{\xi = 0}^{p^{n}-1}\left (-1\right) ^{\xi } \left (\frac{x+a}{d}%
+ \xi \right) ^{n}q^{d\xi } \right) \\
& = & d^{n}\sum_{a=0}^{d-1}\left (-1\right) ^{a}q^{a}\frac{G_{n+1,q}\left (
\frac{x+a}{d}\right) } { n+1}.\end{aligned}$$
After the above applications, we procure the following theorem.
For $ n\in
% TCIMACRO{\U{2115 } }%
% BeginExpansion
\mathbb{N }
% EndExpansion
^{\ast } $, then we have$$G_{n, q}\left (dx\right) = d^{n-1}\sum_{a=0}^{d-1}\left (-1\right)
^{a}q^{a}G_{n, q}\left (x+\frac{a}{d}\right) \text{.}$$
By utilizing from the definition of the geometric series in (\[equation 115\ ]), we well see that | \pi l}\int_{C}\mathcal{F}_{q}\left(
x,t\righu) \frac{dt}{t^{n+1}}$$where $C$ is a lmop whjch stargs at $-\infty $, encircles the ocigib onct in the positive dkrection, wnd the eetucns $-\infty $.
Distriuhtion formula nor tke $q$-Genocchi polinomials is hmportant to sduay regarding $p$-adic Measure theory. Thae is,$$\begon{wligned}
\int_{%
%TCIMWCRO{\L{2124} }%
%FegihVxkansion
\mathbb{Z}
%EndExpansion
_{p}}q^{y}\left( x+y\righu) ^{n}d\mu _{-1}\left( y\right)
&=&\kim_{n\rightarrow \infty }\sum_{\xi =0}^{dp^{n}-1}\peft( -1\right) ^{\xi
}\left( d+\xi \right) ^{n}w^{\xi } \\
&=&e^{n}\sum_{a=0}^{d-1}\left( -1\fight) ^{a}q^{a}\ltfc( \lim_{n\rightzrrow
\infty }\sum_{\xi =0}^{p^{n}-1}\left( -1\right) ^{\xk }\lefc( \frac{x+a}{d}%
+\xi \rughh) ^{n}q^{d\xi }\righv) \\
&=&d^{n}\sui_{a=0}^{d-1}\left( -1\righb) ^{a}q^{a}\frdc{G_{n+1,q}\legt(
\frac{x+a}{d}\righb) }{n+1}.\env{alitned}$$
After the above a'plications, we procute the folnocing theorem.
For $n\in
%TXINACRO{\O{2115} }%
%BeghnExoqnskon
\javhbg{N}
%EndEdpaision
^{\ast }$, tgen we have$$T_{n,q}\left( dx\right) =d^{n-1}\sim_{w=0}^{e-1}\left( -1\right)
^{a}q^{z}G_{n,q}\lest( x+\frac{a}{d}\right) \text{.}$$
By utilizing from the derinition of the geometruc series in (\[equation 115\]), we easijy see that | \pi i}\int_{C}\mathcal{F}_{q}\left( x,t\right) \frac{dt}{t^{n+1}}$$where $C$ is a starts $-\infty $, the origin once the $-\infty $. Distribution for the $q$-Genocchi is important to study regarding $p$-adic theory. That is,$$\begin{aligned} \int_{% %TCIMACRO{\U{2124} }% %BeginExpansion \mathbb{Z} %EndExpansion _{p}}q^{y}\left( x+y\right) ^{n}d\mu _{-1}\left( &=&\lim_{n\rightarrow \infty }\sum_{\xi =0}^{dp^{n}-1}\left( -1\right) ^{\xi }\left( x+\xi \right) ^{n}q^{\xi } \\ &=&d^{n}\sum_{a=0}^{d-1}\left( ^{a}q^{a}\left( \infty =0}^{p^{n}-1}\left( ^{\xi }\left( \frac{x+a}{d}% +\xi \right) ^{n}q^{d\xi }\right) \\ &=&d^{n}\sum_{a=0}^{d-1}\left( -1\right) ^{a}q^{a}\frac{G_{n+1,q}\left( \frac{x+a}{d}\right) }{n+1}.\end{aligned}$$ After the above we procure the following theorem. For $n\in %TCIMACRO{\U{2115} %BeginExpansion \mathbb{N} %EndExpansion ^{\ast then we have$$G_{n,q}\left( dx\right) =d^{n-1}\sum_{a=0}^{d-1}\left( ^{a}q^{a}G_{n,q}\left( \text{.}$$ By from definition the geometric series (\[equation 115\]), we easily see that | \pi i}\int_{C}\mathcal{F}_{q}\left(
x,t\rigHt) \frac{dt}{t^{n+1}}$$Where $c$ is A loOp WhicH staRts at $-\infty $, enciRCles The origin once in the posiTive dIrECtioN, AnD the rEturns $-\iNFtY $.
dIstRiBuTioN fORmUla foR thE $q$-GenocChi polynomIalS iS important to STuDy regardinG $p$-aDic Measure thEorY. That iS,$$\bEgiN{AlignEd}
\iNt_{%
%TCImACRO{\U{2124} }%
%bEginExPansion
\maThBB{Z}
%EndEXPansion
_{P}}Q^{Y}\lEft( x+Y\right) ^{n}d\mu _{-1}\left( y\riGHt)
&=&\LIm_{n\rightarrow \iNfty }\suM_{\xI =0}^{Dp^{N}-1}\LEft( -1\RigHt) ^{\xi
}\left( x+\xI \rIght) ^{n}Q^{\Xi } \\
&=&d^{n}\sum_{A=0}^{D-1}\lEFT( -1\RigHT) ^{a}q^{a}\left( \lim_{n\rIghtarrow
\inFTy }\sUm_{\xi =0}^{p^{n}-1}\LeFt( -1\rIGht) ^{\xi }\lEft( \frAc{X+A}{d}%
+\xI \right) ^{n}q^{d\xi }\RighT) \\
&=&d^{n}\sum_{a=0}^{d-1}\lEft( -1\rigHT) ^{a}q^{a}\fraC{g_{n+1,q}\left(
\Frac{x+a}{D}\riGht) }{N+1}.\end{ALiGnEd}$$
AFtER thE AbOve APplIcations, We PrOcure The fOLLOWing TheOrem.
for $n\iN
%TCIMACRO{\U{2115} }%
%BegInEXpanSIon
\MathbB{N}
%EndexpaNsIon
^{\asT }$, then wE have$$g_{n,Q}\left( dx\right) =d^{n-1}\sUm_{a=0}^{d-1}\Left( -1\right)
^{A}q^{a}g_{n,Q}\leFt( X+\frac{A}{D}\right) \TexT{.}$$
By UtiliziNg from tHE deFiNITIoN of the geometric serIeS IN (\[eQuation 115\]), wE easilY SeE tHAt | \pi i}\int_{C}\mathcal{F}_ {q}\left(x,t\r igh t)\f rac{ dt}{ t^{n+1}}$$wher e $C$ is a loop which start s at$- \ inft y $ , enc irclest he o rig in o nce i n t he po sit ive dir ection, an d t he returns $-\ i nf ty $.
Dis tri bution formu lafor th e$q$ - Genoc chi poly nomial s is im portant t os tudy r e garding $ p$ -adi c Measure theory. Th a t is,$$\begin{ aligne d} \i n t _{%
%T CIMACRO{\U {2 124}} %
%Begi n Ex p a n sio n
\mathbb{Z}
% EndExpansio n
_{ p}}q^{ y} \le f t( x+y \righ t) ^{n }d\mu _{-1} \lef t( y\righ t)
&=& \ lim_{n\ r ightarr ow \in fty }\ sum_ { \x i=0} ^{ d p^{ n }- 1}\ l eft ( -1\rig ht )^{\xi
}\l e f t ( x+\ xi\rig ht) ^ {n}q^{\xi } \ \
& =&d^ { n}\ sum_{ a=0}^ {d-1 }\ left( -1\ri ght)^{ a}q^{a}\left( \ lim_ {n\righta rro w\in ft y }\s u m_{\xi =0 }^{ p^{n}-1 }\left( -1\ ri g h t )^{\xi }\left( \fra c{ x + a} {d}%
+\x i \rig h t) ^ { n}q^{d\x i}\r ight ) \\
&= &d^{ n }\ sum_{a=0 }^{d-1 } \l ef t( -1\r ig ht) ^{ a} q^{ a}\ frac{ G _{n+ 1,q}\l eft(
\f rac{x + a}{d}\right) } { n+1}.\end{ali g ne d } $$
Aft erthe above a ppli c atio ns,w epro c ure t he fo ll o wi n g theorem.
For $n\ in
%TCI MACRO {\U{2115} }%%BeginExpa n s i on
\math bb{N }
% E ndExpansion
^{ \ast}$, then w e have$$G _{n,q }\left(dx\right) = d^{n-1}\ sum _{a =0} ^{d - 1 }\ left( -1\righ t )
^{a }q ^{a}G_{ n,q }\left( x+ \fr ac{ a}{ d} \right) \ text{.}$ $
B yut ili zingf rom thede fin it ion of t h e geom etric ser ie si n ( \[equat i on 1 15\] ), w e ea sil ysee t hat | \pi i}\int_{C}\mathcal{F}_{q}\left(
x,t\right)_\frac{dt}{t^{n+1}}$$where $C$_is a loop which_starts at_$-\infty_$, encircles_the_origin once in_the positive direction,_and the returns $-\infty_$.
Distribution formula for_the_$q$-Genocchi polynomials is important to study regarding $p$-adic Measure theory. That is,$$\begin{aligned}
\int_{%
%TCIMACRO{\U{2124} }%
%BeginExpansion
\mathbb{Z}
%EndExpansion
_{p}}q^{y}\left( x+y\right)_^{n}d\mu__{-1}\left( y\right)
&=&\lim_{n\rightarrow_\infty_}\sum_{\xi_=0}^{dp^{n}-1}\left( -1\right) ^{\xi
}\left( x+\xi \right)_^{n}q^{\xi } \\
&=&d^{n}\sum_{a=0}^{d-1}\left( -1\right) ^{a}q^{a}\left(_\lim_{n\rightarrow
\infty }\sum_{\xi_=0}^{p^{n}-1}\left( -1\right) ^{\xi }\left( \frac{x+a}{d}%
+\xi \right) ^{n}q^{d\xi }\right)_\\
&=&d^{n}\sum_{a=0}^{d-1}\left(_-1\right) ^{a}q^{a}\frac{G_{n+1,q}\left(
\frac{x+a}{d}\right)_}{n+1}.\end{aligned}$$
After the above applications, we procure the following theorem.
For_$n\in
%TCIMACRO{\U{2115} }%
%BeginExpansion
\mathbb{N}
%EndExpansion
^{\ast }$, then we_have$$G_{n,q}\left( dx\right) =d^{n-1}\sum_{a=0}^{d-1}\left(_-1\right)
^{a}q^{a}G_{n,q}\left(_x+\frac{a}{d}\right)_\text{.}$$
By utilizing from the_definition of the geometric series in_(\[equation 115\]), we easily see that |
789 25.870 34.900 8399 2 180.5
NGC7582 22.6 2.0 yes 690 3 1.620 7.400 52.000 83.000 2999 1199 9999 2 270.0
NGC7674 126.6 2.0 yes 260 3 0.680 1.920 5.360 8.300 7178 637 3306 3 220.9
NGC7682 74.4 2.0 yes $<$ 18 5 0.050 0.080 0.350 0.800 2299 270 1199 10 59.8
UGC6100 129.2 2.0 no 34 5 0.145 0.202 0.574 1.500 2900 209 949 18 7.3 11.3
--------- ------------- ------ ------ ------------------------------- -------- ---------- ---------- --------- ---------- ---------- ----------- ------------ ------ ------- -------
Col(1): sources marked by $^{\rm c}$ are contained in the CfA Seyfert sample (Huchra & Burg 1992).
$^{\rm d}$ dropped from the analysis because of meaningless IRAS F12 and F25 upper limits, which are high relative to other quantities.
Col (2): We used a $\Lambda$ cosmology with H$_0$ = 71 kms$^{-1}$ Mpc$^{-1}$, $\Omega_{{\rm matter}}$ = 0.27 and $\Omega_{\Lambda}$ = 0.73.
Col (4): references are Heisler et al. (1997), Lumsden et al. (2001, 2004), Moran et al. (2000, 2001, 2007), Tran (2003), see compilation by Gu & Huang (2002). Sources marked by $^{\rm x}$ show broad Pa$\beta$ lines in ordinary near-infrared spectroscopy, but no optical spectropolarimetric BLR (NGC5506: Nagar et al. 2002, NGC7582: Lumsden priv. comm., see also Aretxaga et al. 1999).
Col (5): in case of non-detections 3-$\sigma$ upper limits are listed. $^{\rm *}$ denotes barely resolved sources. $^{\rm | 789 25.870 34.900 8399 2 180.5
NGC7582 22.6 2.0 yes 690 3 1.620 7.400 52.000 83.000 2999 1199 9999 2 270.0
NGC7674 126.6 2.0 yes 260 3 0.680 1.920 5.360 8.300 7178 637 3306 3 220.9
NGC7682 74.4 2.0 yes $ < $ 18 5 0.050 0.080 0.350 0.800 2299 270 1199 10 59.8
UGC6100 129.2 2.0 no 34 5 0.145 0.202 0.574 1.500 2900 209 949 18 7.3 11.3
--------- ------------- ------ ------ ------------------------------- -------- ---------- ---------- --------- ---------- ---------- ----------- ------------ ------ ------- -------
Col(1 ): sources marked by $ ^{\rm c}$ are contained in the CfA Seyfert sample (Huchra & Burg 1992).
$ ^{\rm d}$ neglect from the psychoanalysis because of meaningless IRAS F12 and F25 upper limits, which are high proportional to other quantity.
Col (2 ): We used a $ \Lambda$ cosmology with H$_0 $ = 71 kms$^{-1}$ Mpc$^{-1}$, $ \Omega_{{\rm matter}}$ = 0.27 and $ \Omega_{\Lambda}$ = 0.73.
Col (4 ): character are Heisler et al. (1997), Lumsden et al. (2001, 2004), Moran et al. (2000, 2001, 2007), Tran (2003), see compilation by Gu & Huang (2002). Sources score by $ ^{\rm x}$ show broad Pa$\beta$ note in ordinary near - infrared spectroscopy, but no ocular spectropolarimetric BLR (NGC5506: Nagar et al. 2002, NGC7582: Lumsden priv. comm. , determine also Aretxaga et al. 1999).
Col (5 ): in case of non - detections 3-$\sigma$ upper limits are listed. $ ^{\rm * } $ denotes scantily resolved sources. $ ^{\rm | 789 25.870 34.900 8399 2 180.5
NGC7582 22.6 2.0 yes 690 3 1.620 7.400 52.000 83.000 2999 1199 9999 2 270.0
NGC7674 126.6 2.0 yes 260 3 0.680 1.920 5.360 8.300 7178 637 3306 3 220.9
NGC7682 74.4 2.0 yes $<$ 18 5 0.050 0.080 0.350 0.800 2299 270 1199 10 59.8
UGC6100 129.2 2.0 no 34 5 0.145 0.202 0.574 1.500 2900 209 949 18 7.3 11.3
--------- ------------- ------ ------ ------------------------------- -------- ---------- ---------- --------- ---------- ---------- ----------- ------------ ------ ------- -------
Sol(1): sourcas marked by $^{\rm c}$ aee contdineg in rhe CfZ Xegfert dam'le (Huchra & Burg 1992).
$^{\rm d}$ eropped from the anslrwis because or meanyndless IRAS F12 and F25 upper limits, which age hjgh relative to other qyantities.
Col (2): We used a $\Lambda$ cosmology with H$_0$ = 71 kms$^{-1}$ Mpc$^{-1}$, $\Omega_{{\rm matter}}$ = 0.27 and $\Mmega_{\Mxmbbq}$ = 0.73.
Cou (4): geferences are Heisler et al. (1997), Lumsden et al. (2001, 2004), Jotak et al. (2000, 2001, 2007), Tran (2003), see compilstlom by Gu & Huang (2002). Souress marked by $^{\rm x}$ shlw broaq Pa$\bwta$ lines in prdinary near-infrared specteoscopy, but uo iptical spectropolcrimetric BLX (NGC5506: Magar et al. 2002, NGC7582: Lumsden prir. comm., see also Agetxaga ef al. 1999).
Col (5): in case of nmn-detections 3-$\sigma$ upper lymits are lisced. $^{\rm *}$ aenojes barqly resolvfd somsces. $^{\rm | 789 25.870 34.900 8399 2 180.5 NGC7582 yes 3 1.620 52.000 83.000 2999 126.6 yes 260 3 1.920 5.360 8.300 637 3306 3 220.9 NGC7682 74.4 yes $<$ 18 5 0.050 0.080 0.350 0.800 2299 270 1199 10 59.8 129.2 2.0 no 34 5 0.145 0.202 0.574 1.500 2900 209 949 18 11.3 ------------- ------ -------- ---------- ---------- --------- ---------- ---------- ----------- ------------ ------ ------- ------- Col(1): sources marked by $^{\rm are contained in the CfA Seyfert sample (Huchra Burg 1992). $^{\rm d}$ from the analysis because of IRAS and F25 limits, are relative to other Col (2): We used a $\Lambda$ cosmology with H$_0$ = 71 kms$^{-1}$ Mpc$^{-1}$, $\Omega_{{\rm matter}}$ = 0.27 $\Omega_{\Lambda}$ = (4): references Heisler al. Lumsden et al. Moran et al. (2000, 2001, 2007), compilation by Gu & Huang (2002). Sources marked $^{\rm x}$ broad Pa$\beta$ lines in ordinary near-infrared but no optical spectropolarimetric BLR (NGC5506: Nagar et 2002, NGC7582: Lumsden priv. comm., see also Aretxaga et al. 1999). Col (5): in case 3-$\sigma$ upper limits are $^{\rm *}$ denotes resolved $^{\rm | 789 25.870 34.900 8399 2 180.5
NGC7582 22.6 2.0 yes 690 3 1.620 7.400 52.000 83.000 2999 1199 9999 2 270.0
NGC7674 126.6 2.0 yes 260 3 0.680 1.920 5.360 8.300 7178 637 3306 3 220.9
NGC7682 74.4 2.0 yes $<$ 18 5 0.050 0.080 0.350 0.800 2299 270 1199 10 59.8
UGC6100 129.2 2.0 no 34 5 0.145 0.202 0.574 1.500 2900 209 949 18 7.3 11.3
--------- ------------- ------ ------ ------------------------------- -------- ---------- ---------- --------- ---------- ---------- ----------- ------------ ------ ------- -------
CoL(1): sources maRked bY $^{\rm C}$ arE cOntaIned In the CfA SeyferT SampLe (Huchra & Burg 1992).
$^{\rm d}$ dropped From tHe ANalySIs BecauSe of meaNInGLEss iRaS f12 anD F25 UPpEr limIts, Which arE high relatIve To Other quantitIEs.
col (2): We used a $\lamBda$ cosmology WitH H$_0$ = 71 kms$^{-1}$ MPc$^{-1}$, $\omeGA_{{\rm maTteR}}$ = 0.27 and $\OMega_{\LaMBda}$ = 0.73.
Col (4): ReferenceS aRE HeislER et al. (1997), LuMSDeN et aL. (2001, 2004), Moran et al. (2000, 2001, 2007), Tran (2003), see COmPIlation by Gu & HuaNg (2002). SourCeS MaRKEd bY $^{\rm X}$ show broad pa$\Beta$ lINes in orDInARY NeaR-Infrared spectRoscopy, but nO OptIcal spEcTroPOlarimEtric bLr (nGC5506: nagar et al. 2002, NGc7582: LumSden priv. cOmm., see ALso AretXAga et al. 1999).
col (5): in cAse Of nOn-deTEcTiOns 3-$\SiGMa$ uPPeR liMIts Are listeD. $^{\rM *}$ dEnoteS barELY REsolVed SourCes. $^{\rm | 789 25.870 34.900 8399 2 180.5
NGC7582 22. 6 2. 0 yes 6 90 3 1.6 20 7.400 52.000 83 .000 2 999 119 9 9999 2 270. 0 NGC76 7 4 1 26 .6 2.0 yes 2 6 0 3 0 .6 80 1.920 5 .36 0 8.300 7178 637 330 6 3 220 .9
NGC7682 7 4.4 2.0 yes $<$ 18 5 0.0 50 0.08 0 0 .350 0.800 229 9 270 119 9 10 59.8
U GC61 00 129 .2 2.0 no 34 5 0 .1 4 5 0.20 2 0 .5 74 1.5 00 2 9 00 2 09 94 9 18 7 .3 11 .3 - ----- - -- - ------ ------ - ----- ------ ------- - ------------- - -- - - -- - -- - --- ---- ------ ---- ---- ---- - ---- - ----- ---- -- - -- - ---------- ------- -- -- --- ----- ---- ------ - ------ --- - - - -
Col(1 ): s o ur c es marked by $ ^{\rm c}$ are c o ntainedin th e CfA Se yfert sam p l e (Huchr a & Bu rg199 2 ) .
$^{\rm d}$ d r o pped f rom the an alysisbec aus e o f m ea ninglessIRAS F12 a nd F 25 up per l i mits, wh ic h a re hi gh re l ativeto ot herqu an t iti es.
Co l ( 2 ) : We u se d a$\L am bda$cosm o log y withH$_0$ = 7 1 k m s$^{ -1 }$ Mpc$^{ -1}$, $\Omega _{ {\rm matte r} }$= 0.27 a nd $\Ome ga_{\Lambda}$ = 0.73.
C ol (4): re feren cesare Heisl eret al. (1 9 97), L umsden et a l. (2 0 0 1, 20 0 4 ), Mo ra n et al. ( 2 0 00, 2001 ,2007 ), Tran (2003), see compi l ati on by Gu & Hu ang (20 0 2 ). So u rc e s m ar k edb y $^{\rm x}$ sho w broad Pa $\ b et a$ lines i n or di nary ne ar-infr areds pectros copy, but no optic al spe c t rop olarimetri c BLR (N GC5506: N a gar e t a l. 20 02, NGC75 82 : L umsde n priv . co mm.,see al so Aretx aga e tal. 1999 ).
Col (5): in case of non-d etect ion s 3-$\sig ma$ upp er limits are listed. $ ^{\ rm*}$ d eno t es ba rely re sol v ed so urce s . $^{\rm | 789 _ _25.870 _ 34.900__ __ 8399_ _ _ __ __ ___2 _ _ 180.5
_ NGC7582 _22.6_ _ 2.0 yes _ _ ___ _ _ _ 690 3_ _ 1.620 _ _ 7.400 _ 52.000 _ _ 83.000 _ 2999_ _ 1199 __ ___ 9999_ __ 2__ __ 270.0
NGC7674 _ _126.6_ _ 2.0 _yes _ _ _ _ _ _260 __ 3 __ _ 0.680 _ _ 1.920 _ 5.360 __ 8.300 _ 7178 _ __637_ _ 3306 _ _3 __ 220.9
NGC7682__ 74.4___ 2.0_ yes _ _ _ ___ $<$ 18 _ 5_ _ 0.050 _ 0.080 _0.350 0.800 _ 2299 _ 270 _ __1199 10 ____ _ _ 59.8
_UGC6100 _ _ _129.2 __ 2.0 no _ _ __ _ _ 34 5 _ 0.145__ _ 0.202 _ _ 0.574_ _ _1.500 _ _ 2900 __ 209 ____ 949_ _ 18 _ _7.3_ 11.3
---------_------------- ------ ------_------------------------------- -------- ---------- ----------_--------- ---------- ---------- ----------- ------------ ------ ------- -------
Col(1): sources marked_by $^{\rm c}$ are contained in the CfA Seyfert sample (Huchra & Burg 1992).
$^{\rm d}$ dropped from_the analysis_because of meaningless IRAS_F12 and_F25_upper limits, which are high relative_to other_quantities.
Col (2): We used a $\Lambda$ cosmology with_H$_0$_= 71 kms$^{-1}$ Mpc$^{-1}$, $\Omega_{{\rm matter}}$ = 0.27 and $\Omega_{\Lambda}$ = 0.73.
Col_(4):_references_are Heisler et_al._(1997), Lumsden et al. (2001, 2004),_Moran_et_al._(2000,_2001, 2007), Tran (2003), see compilation_by Gu & Huang (2002). Sources marked_by_$^{\rm x}$ show broad Pa$\beta$_lines in_ordinary_near-infrared spectroscopy, but no_optical spectropolarimetric BLR (NGC5506:_Nagar et al. 2002, NGC7582: Lumsden priv. comm., see also Aretxaga_et_al. 1999).
Col_(5):_in_case_of non-detections 3-$\sigma$ upper_limits are listed. $^{\rm *}$ denotes barely resolved sources. $^{\rm |
& -20 03 46.798 & 0.037 & 0.32 &0.04 &12.84\
G10.34-0.14 & 18 09 00.762 & 0.012 & -20 03 47.947 & 0.014 & 0.81 &0.04 &12.84\
G10.34-0.14 & 18 09 00.697 & 0.039 & -20 03 48.166 & 0.044 & 0.26 &0.04 &12.84\
G10.34-0.14 & 18 09 00.139 & 0.015 & -20 03 33.859 & 0.017 & 0.55 &0.03 &12.84\
G10.34-0.14 & 18 08 59.945 & 0.029 & -20 03 38.000 & 0.033 & 0.29 &0.03 &12.84\
G10.34-0.14 & 18 08 59.817 & 0.010 & -20 03 37.133 & 0.012 & 0.80 &0.03 &12.84\
G10.34-0.14 & 18 09 00.761 & 0.008 & -20 03 47.954 & 0.009 & 1.33 &0.04 &13.01\
G10.34-0.14 & 18 09 00.685 & 0.031 & -20 03 30.541 & 0.035 & 0.29 &0.03 &13.01\
G10.34-0.14 & 18 09 00.141 & 0.006 & -20 03 33.888 & 0.006 & 1.46 &0.03 &13.01\
G10.34-0.14 & 18 08 59.947 & 0.021 & -20 03 38.031 & 0.024 & 0.40 &0.03 &13.01\
G10.34-0.14 & 18 08 59.819 & 0.008 & -20 03 37.149 & 0.009 & 1.09 &0.03 &13.01\
G10.34-0.14 & 18 09 00.760 & 0.007 & -20 03 47.949 & 0.008 & 1.48 &0.04 &13.18\
G10.34-0.14 | & -20 03 46.798 & 0.037 & 0.32 & 0.04 & 12.84\
G10.34 - 0.14 & 18 09 00.762 & 0.012 & -20 03 47.947 & 0.014 & 0.81 & 0.04 & 12.84\
G10.34 - 0.14 & 18 09 00.697 & 0.039 & -20 03 48.166 & 0.044 & 0.26 & 0.04 & 12.84\
G10.34 - 0.14 & 18 09 00.139 & 0.015 & -20 03 33.859 & 0.017 & 0.55 & 0.03 & 12.84\
G10.34 - 0.14 & 18 08 59.945 & 0.029 & -20 03 38.000 & 0.033 & 0.29 & 0.03 & 12.84\
G10.34 - 0.14 & 18 08 59.817 & 0.010 & -20 03 37.133 & 0.012 & 0.80 & 0.03 & 12.84\
G10.34 - 0.14 & 18 09 00.761 & 0.008 & -20 03 47.954 & 0.009 & 1.33 & 0.04 & 13.01\
G10.34 - 0.14 & 18 09 00.685 & 0.031 & -20 03 30.541 & 0.035 & 0.29 & 0.03 & 13.01\
G10.34 - 0.14 & 18 09 00.141 & 0.006 & -20 03 33.888 & 0.006 & 1.46 & 0.03 & 13.01\
G10.34 - 0.14 & 18 08 59.947 & 0.021 & -20 03 38.031 & 0.024 & 0.40 & 0.03 & 13.01\
G10.34 - 0.14 & 18 08 59.819 & 0.008 & -20 03 37.149 & 0.009 & 1.09 & 0.03 & 13.01\
G10.34 - 0.14 & 18 09 00.760 & 0.007 & -20 03 47.949 & 0.008 & 1.48 & 0.04 & 13.18\
G10.34 - 0.14 | & -20 03 46.798 & 0.037 & 0.32 &0.04 &12.84\
G10.34-0.14 & 18 09 00.762 & 0.012 & -20 03 47.947 & 0.014 & 0.81 &0.04 &12.84\
N10.34-0.14 & 18 09 00.697 & 0.039 & -20 03 48.166 & 0.044 & 0.26 &0.04 &12.84\
G10.34-0.14 & 18 09 00.139 & 0.015 & -20 03 33.859 & 0.017 & 0.55 &0.03 &12.84\
G10.34-0.14 & 18 08 59.945 & 0.029 & -20 03 38.000 & 0.033 & 0.29 &0.03 &12.84\
G10.34-0.14 & 18 08 59.817 & 0.010 & -20 03 37.133 & 0.012 & 0.80 &0.03 &12.84\
G10.34-0.14 & 18 09 00.761 & 0.008 & -20 03 47.954 & 0.009 & 1.33 &0.04 &13.01\
G10.34-0.14 & 18 09 00.685 & 0.031 & -20 03 30.541 & 0.035 & 0.29 &0.03 &13.01\
G10.34-0.14 & 18 09 00.141 & 0.006 & -20 03 33.888 & 0.006 & 1.46 &0.03 &13.01\
G10.34-0.14 & 18 08 59.947 & 0.021 & -20 03 38.031 & 0.024 & 0.40 &0.03 &13.01\
G10.34-0.14 & 18 08 59.819 & 0.008 & -20 03 37.149 & 0.009 & 1.09 &0.03 &13.01\
G10.34-0.14 & 18 09 00.760 & 0.007 & -20 03 47.949 & 0.008 & 1.48 &0.04 &13.18\
C10.34-0.14 | & -20 03 46.798 & 0.037 & &12.84\ & 18 00.762 & 0.012 0.014 0.81 &0.04 &12.84\ & 18 09 & 0.039 & -20 03 48.166 0.044 & 0.26 &0.04 &12.84\ G10.34-0.14 & 18 09 00.139 & 0.015 & 03 33.859 & 0.017 & 0.55 &0.03 &12.84\ G10.34-0.14 & 18 08 59.945 0.029 -20 38.000 0.033 & 0.29 &0.03 &12.84\ G10.34-0.14 & 18 08 59.817 & 0.010 & -20 03 37.133 0.012 & 0.80 &0.03 &12.84\ G10.34-0.14 & 18 00.761 & 0.008 & 03 47.954 & 0.009 & &0.04 G10.34-0.14 & 09 & & -20 03 & 0.035 & 0.29 &0.03 &13.01\ G10.34-0.14 & 18 09 00.141 & 0.006 & -20 03 33.888 0.006 & &13.01\ G10.34-0.14 18 59.947 0.021 & -20 & 0.024 & 0.40 &0.03 &13.01\ 08 59.819 & 0.008 & -20 03 37.149 0.009 & &0.03 &13.01\ G10.34-0.14 & 18 09 & 0.007 & -20 03 47.949 & 0.008 1.48 &0.04 &13.18\ G10.34-0.14 | & -20 03 46.798 & 0.037 & 0.32 &0.04 &12.84\
G10.34-0.14 & 18 09 00.762 & 0.012 & -20 03 47.947 & 0.014 & 0.81 &0.04 &12.84\
G10.34-0.14 & 18 09 00.697 & 0.039 & -20 03 48.166 & 0.044 & 0.26 &0.04 &12.84\
G10.34-0.14 & 18 09 00.139 & 0.015 & -20 03 33.859 & 0.017 & 0.55 &0.03 &12.84\
G10.34-0.14 & 18 08 59.945 & 0.029 & -20 03 38.000 & 0.033 & 0.29 &0.03 &12.84\
G10.34-0.14 & 18 08 59.817 & 0.010 & -20 03 37.133 & 0.012 & 0.80 &0.03 &12.84\
G10.34-0.14 & 18 09 00.761 & 0.008 & -20 03 47.954 & 0.009 & 1.33 &0.04 &13.01\
G10.34-0.14 & 18 09 00.685 & 0.031 & -20 03 30.541 & 0.035 & 0.29 &0.03 &13.01\
G10.34-0.14 & 18 09 00.141 & 0.006 & -20 03 33.888 & 0.006 & 1.46 &0.03 &13.01\
G10.34-0.14 & 18 08 59.947 & 0.021 & -20 03 38.031 & 0.024 & 0.40 &0.03 &13.01\
G10.34-0.14 & 18 08 59.819 & 0.008 & -20 03 37.149 & 0.009 & 1.09 &0.03 &13.01\
G10.34-0.14 & 18 09 00.760 & 0.007 & -20 03 47.949 & 0.008 & 1.48 &0.04 &13.18\
G10.34-0.14 | & -20 03 46.798 & 0.037 & 0.32 &0.0 4 &12 .84 \
G 10 .34- 0.14 & 18 09 00.76 2 & 0 .012 & -20 03 47.947 & 0.01 4& 0.8 1 & 0.04&12.84\ G1 0 . 34- 0. 14 &18 09 00.6 97& 0.039 & -20 0348. 16 6 & 0.044 &0 .2 6 &0.04 &1 2.8 4\
G10.34-0. 14& 18 0 900. 1 39 &0.0 15 &-20 03 33.859 & 0.017&0 .55 &0 . 03 &12. 8 4 \G10. 34-0.14 & 18 08 5 9 .9 4 5 & 0.029 & -2 0 03 3 8. 0 00 & 0. 033 & 0.29 &0 .0 3 &12 . 84\
G10 . 34 - 0 . 14& 18 08 59.817 & 0.010 &- 2003 37. 13 3 & 0.012& 0.8 0& 0.0 3 &12.84\
G 10.3 4-0.14 &18 090 0.761 & 0.008 & -20 0 3 4 7.9 54 & 0. 00 9 & 1 . 33& 0. 04& 13. 01\
G10. 34 -0 .14 & 180 9 0 0.68 5 & 0.0 31 &-20 03 30.541 &0.03 5 &0.29&0.03 &13 .0 1\
G1 0.34-0 .14 & 1 8 09 00.141 & 0 .006 & -20 03 33 .8 88&0.006 & 1.46 &0 .03 &13.01 \
G10.3 4 -0. 14 & 18 08 59.947 & 0.021 & - 20 03 38.0 31 & 0 . 02 4& 0.40 &0 .0 3 & 13.0 1 \
G10. 34-0 . 14 & 18 08 59.81 9 & 0 .008 &-2 0 03 3 7. 149 &0.009 & 1. 09 &0. 03 &13.0 1\
G1 0 .34-0.14 & 180 9 00.760 & 0. 0 07 & - 2 0 03 47 .949 & 0.00 8 &1 .48&0.0 4 & 13. 1 8\
G1 0.34- 0. 1 4 | &_-20 03_46.798 & 0.037 &_0.32 &0.04_&12.84\
G10.34-0.14_& 18_09_00.762 & 0.012_& -20 03_47.947 & 0.014 &_0.81 &0.04 &12.84\
G10.34-0.14_&_18 09 00.697 & 0.039 & -20 03 48.166 & 0.044 & 0.26 &0.04_&12.84\
G10.34-0.14_& 18_09_00.139_& 0.015 & -20 03_33.859 & 0.017 & 0.55_&0.03 &12.84\
G10.34-0.14_& 18 08 59.945 & 0.029 & -20_03_38.000 & 0.033_& 0.29 &0.03 &12.84\
G10.34-0.14 & 18 08 59.817 &_0.010 & -20 03 37.133 &_0.012 & 0.80_&0.03_&12.84\
G10.34-0.14_& 18 09 00.761_& 0.008 & -20 03 47.954_& 0.009 & 1.33 &0.04 &13.01\
G10.34-0.14_& 18 09 00.685 & 0.031 &_-20 03 30.541 & 0.035 &_0.29 &0.03 &13.01\
G10.34-0.14 & 18_09 00.141_& 0.006 & -20 03_33.888 & 0.006_& 1.46_&0.03 &13.01\
G10.34-0.14 &_18 08 59.947 & 0.021 &_-20 03 38.031_& 0.024 & 0.40 &0.03 &13.01\
G10.34-0.14_&_18 08 59.819_&_0.008_& -20_03 37.149 &_0.009_& 1.09_&0.03_&13.01\
G10.34-0.14 & 18 09 00.760 &_0.007_& -20 03 47.949 & 0.008 &_1.48 &0.04 &13.18\
G10.34-0.14 |
with correct quantum numbers such as spin and charge.
There are also different opinions on the anyonic properties including the existence of statistical spin phase transition in a symmetry-broken phase [@Wen89; @Boya90; @shin-ssb], which is an interesting problem that may be relevant to high-$T_c$ superconductivity [@WZ89]. There are arguments that the excitations relative to ground state are spin 0 bosons [@Boya90] and that the statistics-changing phase transition is impossible in the Chern-Simons-Higgs theory [@shin-ssb]. These results seem to contradict the result of Deser and Yang [@Dese89], observing that Higgs mechanism transmutes a non-dynamical Chern-Simons term into topologically massive, parity-violating, spin 1 theory [@Dese84].
In this paper, we will perform a careful analysis on the charged matter fields coupled to Chern-Simons gauge fields in a covariant gauge along the Dirac’s method performed on QED [@Dirac58]. In Sec.II, we will analyze the Maxwell theory both in a symmetric and in a symmetry-broken phase with the covariant gauge fixing in order to illustrate the definite way of identifying physical spectrum free from gauge ambiguity arising from gauge fixing and to manifest the importance of consideration of [*physical*]{} fields. As a result, we show that there exists a spin phase transition in the Maxwell-Higgs theory. In Sec.III, the quantization of Chern-Simons matter system will be presented and physical anyon operators will be constructed in a covariant gauge based on the approach in Sec.II. We will also analyze the quantization of the Chern-Simons matter system in the symmetry-broken phase and show that Higgs mechanism transmutes an anyon satisfying fractional statistics into a spin 0 Higgs boson and a topologically massive photon which is a Chern-Simons gauge field absorbed a would-be-Goldstone boson. Thus the Higgs effect influences the spin phase of the anyon and interestingly induces the statistical spin phase transition, as predicted by Wen and Zee [@Wen89]. This result is consistent with the observation of Deser and Yang [@Dese89]. Section IV contains conclusion. In Appendix, we will analyze the Poincaré algebra of massive vector fields - Proca and massive Chern-Simons theories - and extract the spin content of massive vector fields with no ambiguity.
MAXWELL THEORY IN A COVARIANT GAUGE
| with correct quantum numbers such as spin and charge.
There be besides different opinions on the anyonic place include the existence of statistical spin phase conversion in a isotropy - broken phase [ @Wen89; @Boya90; @shin - ssb ], which is an interesting problem that may be relevant to high-$T_c$ superconductivity [ @WZ89 ]. There are arguments that the excitations proportional to ground state are spin 0 bosons [ @Boya90 ] and that the statistic - changing phase transition is impossible in the Chern - Simons - Higgs hypothesis [ @shin - ssb ]. These results look to oppose the result of Deser and Yang [ @Dese89 ], observing that Higgs mechanism transmutes a non - dynamic Chern - Simons term into topologically massive, parity - violating, tailspin 1 theory [ @Dese84 ].
In this newspaper, we will perform a careful psychoanalysis on the charged matter fields coupled to Chern - Simons bore fields in a covariant gauge along the Dirac ’s method performed on QED [ @Dirac58 ]. In Sec. II, we will analyze the Maxwell theory both in a symmetric and in a symmetry - broken phase with the covariant gauge fixing in order to illustrate the definite way of identifying physical spectrum free from gauge ambiguity arising from bore repair and to manifest the importance of retainer of [ * forcible * ] { } playing field. As a result, we show that there exist a spin phase transition in the Maxwell - Higgs theory. In Sec. III, the quantization of Chern - Simons matter arrangement will be presented and physical anyon operators will be constructed in a covariant gauge based on the approach path in Sec. II. We will also analyze the quantization of the Chern - Simons matter system in the symmetry - unwrap phase and show that Higgs mechanism transform an anyon satisfying fractional statistics into a spin 0 Higgs boson and a topologically massive photon which is a Chern - Simons gauge field steep a would - be - Goldstone boson. Thus the Higgs effect influences the tailspin phase of the anyon and interestingly induces the statistical tailspin phase transition, as predicted by Wen and Zee [ @Wen89 ]. This result is consistent with the observation of Deser and Yang [ @Dese89 ]. department IV contains conclusion. In Appendix, we will analyze the Poincaré algebra of massive vector sphere - Proca and massive Chern - Simons theories - and extract the spin message of massive vector fields with no ambiguity.
MAXWELL hypothesis IN A COVARIANT GAUGE | wihh correct quantum numbevs such as spin cbd chacge.
Thers are alro different opinions on the abyonix properties including the exisnence of wtatmstical spin phase translcion jk a svmnetry-broken phsse [@Wen89; @Bofa90; @shin-ssb], whiwh id an interesting problem that may bq relevsnh to high-$T_c$ sukercomquctjniuy [@WZ89]. There are arguments that ths excitetions relative to ground state are spin 0 boslns [@Boya90] and that hhe statistucs-crqnging phase transition is impossigle in the Chern-Simons-Higgs theury [@skin-ssb]. Thesg dedolts seem to contgadict the result of Geser amd Yang [@Dese89], onservmng rhat Higgs mechanism vransmutes a non-dynaiical Chesn-Aimons term into ropologhcalny mxwsixe, karmty-biolatlng, spin 1 thekry [@Dese84].
In rhis paper, we will kersirm a careful analyfif on the charged matter fields coupled uo Chsrn-Simons gauge fields un a covariant gauge wlong the Dirac’s method performed on QED [@Dirac58]. In Sec.II, we will xnaoyde tfw Laxwell theory both in a symmetric and in a srjmttrj-broken phase witm the covariant gaigf gyxing in ordet to illusfrate the definite way of idenrifying prysival spectrum free from gaugw ambiguity criwing from gauge firing and to oanigest yhe importance of consiberatikn of [*physifal*]{} fielda. As a result, we rhoe dhat there exists a spin prase tranwitipn in tfe Msxwell-Riggs theogy. In Sec.III, the quantizwtion ox Chern-Simlns matter system will be preseived and physivan atyon opexators will be conseructed in a cpvarianc gaugd based on the ap'roach in Ses.II. We will ando analyze tie quantisatiin od the Cfdrn-Simons mattrr system in the synmetry-broken phase ana show that Higgw mtcyanism transmuyes an ajypn vatisfying fsactkonxk stagistics inti a slin 0 Higgs boson and a tkpologically massife photon qhich is a Chern-Simonx gauge field absogbed e woulv-be-Golcstjne boson. Thus the Higgs effecf influenfes the spin phwse if the anyon and interestingly induces the statistiral spin phase transitiin, as predicted by Cek and Zee [@Wei89]. This result iv consistent with thw observation of Ceser and Yang [@Dese89]. Ssction IV clntains conclusion. In Appendix, we will analyze the Poincaré algebra of maswive vxceor fields - Prova ang icssive Sheri-Smmons theories - amd extract the spin content of oassive eeetor fields with no ambiguity.
KABWELL THEORY KN A COVARIANT GAUGE
| with correct quantum numbers such as spin There also different on the anyonic statistical phase transition in symmetry-broken phase [@Wen89; @shin-ssb], which is an interesting problem may be relevant to high-$T_c$ superconductivity [@WZ89]. There are arguments that the excitations to ground state are spin 0 bosons [@Boya90] and that the statistics-changing phase is in Chern-Simons-Higgs [@shin-ssb]. These results seem to contradict the result of Deser and Yang [@Dese89], observing that Higgs transmutes a non-dynamical Chern-Simons term into topologically massive, spin 1 theory [@Dese84]. this paper, we will perform careful on the matter coupled Chern-Simons gauge fields a covariant gauge along the Dirac’s method performed on QED [@Dirac58]. In Sec.II, we will analyze the theory both symmetric and a phase the covariant gauge order to illustrate the definite way spectrum free from gauge ambiguity arising from gauge and to the importance of consideration of [*physical*]{} As a result, we show that there exists spin phase transition in the Maxwell-Higgs theory. In Sec.III, the quantization of Chern-Simons matter system presented and physical anyon will be constructed a gauge on approach in We will also analyze the quantization of the Chern-Simons matter system the symmetry-broken phase and show that Higgs mechanism transmutes an fractional into a spin Higgs boson and a massive which is a Chern-Simons absorbed would-be-Goldstone Higgs influences spin phase of the and interestingly induces the statistical phase transition, as predicted This result is consistent with the observation of and Yang [@Dese89]. Section IV contains conclusion. Appendix, we will analyze the Poincaré algebra of massive vector fields - and massive - and extract the spin content of massive fields with no ambiguity. THEORY IN A COVARIANT GAUGE | with correct quantum numbers Such as spin And chArgE.
ThErE are Also Different opiniONs on The anyonic properties inCludiNg THe exIStEnce oF statisTIcAL SpiN pHaSe tRaNSiTion iN a sYmmetry-Broken phasE [@WeN89; @BOya90; @shin-ssb], whICh Is an intereStiNg problem thaT maY be relEvAnt TO high-$t_c$ sUpercOnductIVity [@WZ89]. there are aRgUMents tHAt the exCITaTionS relative to ground STaTE are spin 0 bosons [@boya90] anD tHAt THE stAtiStics-changInG phasE TransitIOn IS IMpoSSible in the CheRn-Simons-HigGS thEory [@shIn-Ssb]. tHese reSults SeEM to Contradict tHe reSult of DesEr and YANg [@Dese89], oBServing That HiGgs MecHaniSM tRaNsmUtES a nON-dYnaMIcaL Chern-SiMoNs Term iNto tOPOLOgicAllY masSive, pArity-violatinG, spIn 1 thEOry [@dese84].
IN this PapeR, wE will PerforM a carEfUl analysis on the CharGed matter FieLdS coUpLed to cHern-SiMonS gaUge fielDs in a coVAriAnT GAUgE along the Dirac’s metHoD PErFormed on qED [@DirAC58]. IN SEC.II, we wilL aNalYze tHE maxweLl thEOrY both in a SymmetRIc AnD in a symMeTry-broKeN phAse With tHE covAriant Gauge fixIng in ORder to illustraTE the definite wAY oF IDeNTifyIng Physical speCtruM Free From GAuGe aMBiguiTy ariSiNG fROm gauge fixing and to mAnIfest tHe impOrtance of consIderation oF [*PHYsical*]{} fiElds. aS a REsult, we show thaT therE exists a spIN phase trAnsitIon in the maxwell-HiGGS theory. IN SeC.IIi, thE quANTiZation of Chern-sIMons MaTter sysTem Will be pResEntEd aNd pHySical anyoN operatoRs WiLl Be ConStrucTEd in a covArIanT gAugE baseD On the aPproaCh in seC.Ii. we wIll also ANaLYZe thE qUaNtizAtiOn Of the cherN-simOns mattEr system iN thE SymmEtRy-Broken pHase and show thAt higgs mechaNiSm tRansmuTES an anyon Satisfying fractional staTIstics iNto A spin 0 higgS boson and A toPologiCalLY massiVe photOn whiCh Is a cHErn-SiMONs GauGe Field absorBED a wOuld-bE-GOldsTone bosOn. Thus the Higgs effeCT inFluences the spIn pHase OF ThE anYOn ANd iNtEResTINgly induces the sTatistical SpIN pHase transiTIon, As PredictEd by Wen And ZeE [@wen89]. This Result is cOnsistent WiTh thE OBseRvation of DEser and YAng [@Dese89]. SeCTion Iv CoNtainS coNclusiOn. in APpendIx, we wiLL anAlyze The PoiNcAré algEbra oF mAssive veCtor fields - Proca and massiVe CherN-SimoNs tHeories - anD exTRacT the spin cOnteNt of massivE veCtoR fielDs wITh no aMbigUItY.
MAxwELL ThEORy iN A COVARIanT gAUge
| with correct quantum numb ers such a s spi n a ndch arge .
T here are alsod iffe rent opinions on the a nyoni cp rope r ti es in cluding th e exi st en ceof st atist ica l spinphase tran sit io n in a symme t ry -broken ph ase [@Wen89; @B oya 90; @s hi n-s s b], w hic h isan int e restin g problem t h at may be rele v a nt tohigh-$T_c$ superc o nd u ctivity [@WZ89 ]. The re ar e arg ume nts that t he exci t ationsr el a t i vet o ground stat e are spin0 bo sons [ @B oya 9 0] and that t h e s tatistics-c hang ing phase trans i tion is impossi ble in th e C hern - Si mo ns- Hi g gst he ory [@s hin-ssb] .Th ese r esul t s s eemtocont radic t the resultofDese r an d Yan g [@D ese8 9] , obs erving that H iggs mechanismtran smutes anon -d yna mi cal C h ern-Si mon s t erm int o topol o gic al l y ma ssive, parity-viol at i n g, spin 1theory [@ De s e84].
I nthi s pa p e r, we wil l p erform a caref u lan alysison the c ha rge d m atter fiel ds cou pled toChern - Simons gauge f i elds in a cov a ri a n tg auge al ong the Dir ac’s meth od p e rf orm e d onQED [ @D i ra c 58]. In Sec.II, wewi ll ana lyzethe Maxwell t heory both i n a symme tric an d in a symmetry -brok en phase w i th the c ovari ant gaug e fixingi n order t o i llu str ate t he definite way o f id en tifying ph ysicalspe ctr umfre efrom gaug e ambigu it yar is ing from gauge fi xi ngan d t o man i fest t he im port an ce ofconside r at i o n of [ *p hysi cal *] {} fi elds . As a resu lt, we sh owt hatth er e exist s a spin phas etransition i n t he Max w e ll-Higgs theory. In Sec.III, th e quanti zat ion o f Ch ern-Simon s m attersys t em wil l be p resen te d a n d phys i c al an yo n operator s wil l beco nstr ucted i n a covariant gaug e ba sed on the ap pro achi n S ec. I I. Wewi l l a l s o analyze the q uantizatio no fthe Chern- S imo ns matter system in t h e symme try-broke n phase a nd sho w tha t Higgs me chanismtransmute s an a n yo n sat isf ying f ra cti onalstatis t ics into a spi n0 Higg s bos on and a t opologically massive ph oton w hichisa Chern-S imo n s g auge fiel d ab sorbed a w oul d-b e-Gol dst o ne bo son. Th ust he Hi ggse ffect inf l ue nce s th e spin phas e o f t he an yon and in tere stingly induces t h e statisticalspin p has e t r ansi ti on, as predict edby W en and Z ee [@Wen89].This res ul t is c onsist ent wi th theo b se r vation ofDes er and Ya ng[@ D ese89]. S ec t ion IV con ta ins co nclusi o n. I n Appendix, we wil l ana l y ze th e Po incar éalgebra of m assive vec tor fields- Proc a an d mas sive Ch er n-Simo nsth eories - a n d extract thespin co nt entofmassiv e ve c t or fi elds w ith no ambig u i ty .
MA X WEL L TH EORYIN A C OVARIANTG AUGE
| with_correct quantum_numbers such as spin_and charge.
There_are_also different_opinions_on the anyonic_properties including the_existence of statistical spin_phase transition in_a_symmetry-broken phase [@Wen89; @Boya90; @shin-ssb], which is an interesting problem that may be relevant_to_high-$T_c$ superconductivity_[@WZ89]._There_are arguments that the excitations_relative to ground state are_spin 0_bosons [@Boya90] and that the statistics-changing phase transition_is_impossible in the_Chern-Simons-Higgs theory [@shin-ssb]. These results seem to contradict the_result of Deser and Yang [@Dese89],_observing that Higgs_mechanism_transmutes_a non-dynamical Chern-Simons term_into topologically massive, parity-violating, spin 1_theory [@Dese84].
In this paper, we will_perform a careful analysis on the charged_matter fields coupled to Chern-Simons gauge_fields in a covariant gauge_along the_Dirac’s method performed on QED_[@Dirac58]. In Sec.II,_we will_analyze the Maxwell_theory both in a symmetric and_in a symmetry-broken_phase with the covariant gauge fixing_in_order to illustrate_the_definite_way of_identifying physical spectrum_free_from gauge_ambiguity_arising from gauge fixing and to_manifest_the importance of consideration of [*physical*]{} fields._As a result, we_show_that there exists a_spin phase transition in the_Maxwell-Higgs theory. In Sec.III, the quantization_of Chern-Simons_matter system_will be presented and physical anyon operators will be constructed in_a covariant gauge based on the_approach in Sec.II. We_will also_analyze_the quantization of_the_Chern-Simons matter_system in the symmetry-broken phase and show_that Higgs_mechanism transmutes an anyon satisfying fractional_statistics into a spin_0_Higgs boson and a topologically massive_photon which is a Chern-Simons gauge_field absorbed a would-be-Goldstone boson._Thus_the_Higgs effect influences the spin_phase of the anyon and interestingly_induces the statistical_spin phase transition, as predicted by Wen_and_Zee [@Wen89]. This result is consistent_with_the observation of Deser and Yang_[@Dese89]._Section_IV contains conclusion. In Appendix,_we will analyze the Poincaré algebra_of massive vector fields - Proca and massive Chern-Simons_theories - and_extract the spin content of_massive_vector_fields with no ambiguity.
MAXWELL THEORY IN A COVARIANT GAUGE
|
size batteries very closely, while the energy overflow probability with the finite-size battery is less than the approximation for threshold values close to the battery size. However, we can accurately approximate the energy overflow probability for threshold values less than the battery size. Our approximation matches the finite-size battery simulations very closely for energy threshold values up to $80\%$ of the battery capacity, which is $400$ energy units in our simulations. In a real setting, one should not charge a battery completely but up to $80\%$ in order to improve the battery life-span and the energy efficiency [@battery_uni_2; @website_2].
In Fig. \[fig:fig\_2\], we plot the energy outage probability, $\pi_{0}$ in (\[eq:upper\_bound\_pi\_0\]), as a function of the energy decay rate, $\mu$, for different scale parameters, $\lambda_{\text{dB}}=10\log_{10}\frac{\lambda}{N\sigma_{w}^{2}}=5$, $4$ and $3$ dB, when the shape parameter is $k=1$. Note that the mean value of the Weibull distribution is $\lambda\Gamma(1+1/k)$, where $\Gamma(\cdot)$ is the Gamma function. So, $\lambda$ becomes the average energy arrival rate when $k=1$. If we use all the energy for data transmission, we can consider $\lambda_{\text{dB}}$ as the average signal-to-noise ratio in the channel. In addition, given the energy decay rate, we determine the constant energy demand rate. Particularly, given that the energy packets arriving at the battery are i.i.d., the constant energy demand rate is $p=\frac{1}{\mu}\ln\mathbb{E}_{u}\left[e^{\mu u}\right]=\frac{1}{\mu}\ln\left(\frac{1}{1-\lambda\mu}\right)$ for $0<\mu<\frac{1}{\lambda}$. Notice that when $\mu$ goes to zero, $p$ approaches the average energy arrival rate, $\lambda$, whereas when $\mu$ goes to $\frac{1}{\lambda}$, $p$ approaches infinity. Therefore, the outage probability becomes 1 once $\mu$ is greater than $\frac{1}{\lambda}$ for a given energy arrival process with the aforementioned distribution, which is displayed in Fig. \[fig:fig\_2\]. We can also infer that the energy out | size batteries very closely, while the energy overflow probability with the finite - size barrage is less than the estimate for doorway values close to the barrage size. However, we can accurately approximate the energy overflow probability for doorway values less than the battery size. Our estimate equal the finite - size barrage simulations very closely for department of energy threshold values up to $ 80\%$ of the battery capacity, which is $ 400 $ energy units in our simulations. In a actual setting, one should not charge a battery wholly but up to $ 80\%$ in order to improve the barrage life - bridge and the energy efficiency [ @battery_uni_2; @website_2 ].
In Fig. \[fig: fig\_2\ ], we plat the energy outage probability, $ \pi_{0}$ in (\[eq: upper\_bound\_pi\_0\ ]), as a function of the energy decay pace, $ \mu$, for different scale parameters, $ \lambda_{\text{dB}}=10\log_{10}\frac{\lambda}{N\sigma_{w}^{2}}=5 $, $ 4 $ and $ 3 $ dB, when the shape parameter is $ k=1$. Note that the mean value of the Weibull distribution is $ \lambda\Gamma(1 + 1 / k)$, where $ \Gamma(\cdot)$ is the Gamma affair. So, $ \lambda$ becomes the average energy arrival rate when $ k=1$. If we use all the energy for data transmission, we can consider $ \lambda_{\text{dB}}$ as the modal signal - to - noise ratio in the distribution channel. In addition, given the energy decay rate, we settle the constant energy demand rate. peculiarly, given that the energy packets arriving at the battery are i.i.d. , the constant energy demand rate is $ p=\frac{1}{\mu}\ln\mathbb{E}_{u}\left[e^{\mu u}\right]=\frac{1}{\mu}\ln\left(\frac{1}{1-\lambda\mu}\right)$ for $ 0<\mu<\frac{1}{\lambda}$. Notice that when $ \mu$ goes to zero, $ p$ approaches the modal energy arrival pace, $ \lambda$, whereas when $ \mu$ run to $ \frac{1}{\lambda}$, $ p$ approaches infinity. Therefore, the outage probability become 1 once $ \mu$ is greater than $ \frac{1}{\lambda}$ for a given energy arrival procedure with the aforementioned distribution, which is displayed in Fig. \[fig: fig\_2\ ]. We can also infer that the department of energy out | sizf batteries very closely, while the eneryt overhlow prkbabilith with the finite-size batterb is less than the approximatiov for thrvshold vaoues xlose to tis battevv sizs. Howzvxr, we can accurstely apprmximate the enarey overflow probability for threshold values lfss than the bwttegy sizs. Our approximation matches the fihite-sizt battery simulatipns very closely for energj thgeshold values up ho $80\%$ of the vattqey capacity, dhich is $400$ tnzrgy units jn our simulations. In a real segting, one shoule bot wharge a bavtery bompletely bub up to $80\%$ in orcer to improve thx bartery life-span and thx energy efficiency [@fattery_unh_2; @cebsite_2].
In Fig. \[fig:fig\_2\], qe plot the enefty uutzgx pdobabipitb, $\pi_{0}$ in (\[eq:ulper\_bound\_pi\_0\]), as a function of tne vmergy decay date, $\mt$, sor different scale parameters, $\lambda_{\teqt{dB}}=10\mog_{10}\frac{\lambda}{N\sigma_{w}^{2}}=5$, $4$ abd $3$ dB, when the shape parametew is $k=1$. Note that the mean value of the Weibull divtribntkon if $\oalbda\Gamma(1+1/k)$, where $\Gamma(\cdot)$ is the Gamma functykn. Sp, $\lambda$ becomcs the average enetgj strival rate whgn $k=1$. If we use all the energj for dwta teansmissijn, wr can consider $\lambda_{\text{dB}}$ as the avercge signal-to-noise ratno in the chcnnel. On adcition, given the energy deczy rate, we fetermine ghe constant enefgy damand raud. Particularly, giden that vhe euergy pazketx arriding at thf batbary are i.i.d., the cojstanj enercy demand gate is $p=\frac{1}{\mu}\ln\mathbb{E}_{u}\left[e^{\mn u}\right]=\frac{1}{\mo}\ln\nefn(\frac{1}{1-\lambba\mu}\rinht)$ for $0<\mu<\frac{1}{\lwmbda}$. Notice tkat when $\mu$ gues to zerk, $p$ appcoaches the wverage energf arrival ratx, $\lambda$, rherwas qhen $\mu$ eoes to $\frac{1}{\lakbda}$, $p$ apirjachws infinity. Therefpre, fhe outage probcyioity becomes 1 omce $\mu$ id jreatqs than $\frac{1}{\ldmbdx}$ fut a gkven entxgy xrrifal process with the afodementioned distrinubion, whicy is dis[layed in Fig. \[fig:fig\_2\]. We can alsl infxr thav the rnetgy out | size batteries very closely, while the energy with finite-size battery less than the to battery size. However, can accurately approximate energy overflow probability for threshold values than the battery size. Our approximation matches the finite-size battery simulations very closely energy threshold values up to $80\%$ of the battery capacity, which is $400$ units our In real setting, one should not charge a battery completely but up to $80\%$ in order to the battery life-span and the energy efficiency [@battery_uni_2; In Fig. \[fig:fig\_2\], we the energy outage probability, $\pi_{0}$ (\[eq:upper\_bound\_pi\_0\]), a function the decay $\mu$, for different parameters, $\lambda_{\text{dB}}=10\log_{10}\frac{\lambda}{N\sigma_{w}^{2}}=5$, $4$ and $3$ dB, when the shape parameter is $k=1$. Note that the mean value the Weibull $\lambda\Gamma(1+1/k)$, where is Gamma So, $\lambda$ becomes energy arrival rate when $k=1$. If the energy for data transmission, we can consider as the signal-to-noise ratio in the channel. In given the energy decay rate, we determine the energy demand rate. Particularly, given that the energy packets arriving at the battery are i.i.d., energy demand rate is u}\right]=\frac{1}{\mu}\ln\left(\frac{1}{1-\lambda\mu}\right)$ for $0<\mu<\frac{1}{\lambda}$. that $\mu$ to $p$ approaches average energy arrival rate, $\lambda$, whereas when $\mu$ goes to $\frac{1}{\lambda}$, approaches infinity. Therefore, the outage probability becomes 1 once $\mu$ than for a given arrival process with the distribution, is displayed in Fig. can infer out | size batteries very closely, wHile the eneRgy ovErfLow PrObabIlitY with the finite-SIze bAttery is less than the appRoximAtIOn foR ThReshoLd valueS ClOSE to ThE bAttErY SiZe. HowEveR, we can aCcurately aPprOxImate the enerGY oVerflow proBabIlity for threShoLd valuEs LesS Than tHe bAtterY size. OUR approXimation mAtCHes the FInite-siZE BaTterY simulations very cLOsELy for energy thrEshold VaLUeS UP to $80\%$ Of tHe battery cApAcity, WHich is $400$ eNErGY UNitS In our simulatiOns. In a real sETtiNg, one sHoUld NOt charGe a baTtERy cOmpletely buT up tO $80\%$ in order tO improVE the batTEry life-Span anD thE enErgy EFfIcIenCy [@BAttERy_Uni_2; @WEbsIte_2].
In Fig. \[FiG:fIg\_2\], we pLot tHE ENErgy OutAge pRobabIlity, $\pi_{0}$ in (\[eq:upPer\_BounD\_Pi\_0\]), aS a funCtion Of thE eNergy Decay rAte, $\mu$, FoR different scale ParaMeters, $\lamBda_{\TeXt{db}}=10\lOg_{10}\fraC{\Lambda}{n\siGma_{W}^{2}}=5$, $4$ and $3$ dB, wHen the sHApe PaRAMEtEr is $k=1$. Note that the meAn VALuE of the WeIbull dIStRiBUtion is $\lAmBda\gammA(1+1/K)$, Where $\gammA(\CdOt)$ is the GAmma fuNCtIoN. So, $\lambDa$ BecomeS tHe aVerAge enERgy aRrival Rate when $K=1$. If we USe all the energy FOr data transmiSSiON, We CAn coNsiDer $\lambda_{\teXt{dB}}$ AS the AverAGe SigNAl-to-nOise rAtIO iN The channel. In additioN, gIven thE enerGy decay rate, we Determine tHE COnstant eNergY DeMAnd rate. ParticuLarly, Given that tHE energy pAcketS arrivinG at the batTERy are i.i.d., The ConStaNt eNERgY demand rate is $P=\FRac{1}{\mU}\lN\mathbb{e}_{u}\lEft[e^{\mu u}\RigHt]=\fRac{1}{\Mu}\lN\lEft(\frac{1}{1-\laMbda\mu}\riGhT)$ fOr $0<\Mu<\FraC{1}{\lambDA}$. Notice tHaT whEn $\Mu$ gOes to ZEro, $p$ apProacHes tHe AvERagE energy ARrIVAl raTe, $\LaMbda$, WheReAs wheN $\mu$ gOEs tO $\frac{1}{\laMbda}$, $p$ apprOacHEs inFiNiTy. ThereFore, the outage PrObability bEcOmeS 1 once $\mU$ IS greater Than $\frac{1}{\lambda}$ for a given ENergy arRivAl proCess With the afOreMentioNed DIstribUtion, wHich iS dIspLAYed in fIG. \[fIg:fIg\_2\]. we can also iNFEr tHat thE eNergY out | size batteries very closel y, while t he en erg y o ve rflo w pr obability with thefinite-size battery is less t h an t h eappro ximatio n f o r th re sh old v a lu es cl ose to the battery s ize .However, wec an accuratel y a pproximate t heenergy o ver f low p rob abili ty for thresh old value sl ess th a n the b a t te ry s ize. Our approxim a ti o n matches thefinite -s i ze b att ery simulatio ns very closely fo r e ner g y threshold v alues up to $80 \%$ of t heb attery capa ci t y,which is $4 00$energy un its in our sim u lations . In a re alsett i ng ,one s h oul d n otc har ge a bat te ry comp lete l y b ut u p t o $8 0\%$in order to i mpr ovet hebatte ry li fe-s pa n and the e nergy e fficiency [@bat tery _uni_2; @ web si te_ 2] .
In Fig. \ [fi g:f ig\_2\] , we pl o t t he e n er gy outage probabil it y , $ \pi_{0}$ in (\ [ eq :u p per\_bou nd \_p i\_0 \ ] ), as a f u nc tion ofthe en e rg ydecay r at e, $\m u$ , f ordiffe r entscaleparamete rs, $ \ lambda_{\text{ d B}}=10\log_{1 0 }\ f r ac { \lam bda }{N\sigma_{ w}^{ 2 }}=5 $, $ 4 $and $3$ d B, wh en th e shape parameter is $ k=1$.Notethat the mean value oft h e Weibull dis t ri b ution is $\lam bda\G amma(1+1/k ) $, where $\Ga mma(\cdo t)$ is th e Gamma fu nct ion . S o,$ \ la mbda$ becomes t he a ve rage en erg y arriv alrat e w hen $ k=1$. Ifwe use a ll t he e ner gy fo r data tr an smi ss ion , wec an con sider $\l am bd a _{\ text{dB } }$ a s th eav erag e s ig nal-t o-no i seratio i n the cha nne l . In a dd ition,given the ene rg y decay ra te , w e dete r m ine theconstant energy demandr ate. Pa rti cular ly,given tha t t he ene rgy packet s arri vingat th e batte r y a rei. i.d., thec o nst ant e ne rgydemandrate is $p=\frac{1 } {\m u}\ln\mathbb{ E}_ {u}\ l e ft [e^ { \m u u} \r i ght ] = \frac{1}{\mu}\l n\left(\fr ac { 1} {1-\lambda \ mu} \r ight)$for $0< \mu<\ f rac{1}{ \lambda}$ . Noticeth at w h e n $ \mu$ goesto zero, $p$ appr o aches th e ave rag e ener gy ar rival rate, $\l ambda $, whe re as whe n $\m u$ goes to $\frac{1}{\lambda}$, $ p$ app roach esinfinity. Th e ref ore, theouta ge probabi lit y b ecome s 1 once$\mu $ i s g r eater tha n $\frac{1 } {\ lam b d a} $ for a giv e n ene rgy a rri v al pro cess with the aforeme n tioned distrib utio n , wh ich is d is played in Fig. \[ fi g : fig\_2\] .We can also infer t ha t theenergy out | size batteries_very closely,_while the energy overflow_probability with_the_finite-size battery_is_less than the_approximation for threshold_values close to the_battery size. However,_we_can accurately approximate the energy overflow probability for threshold values less than the battery_size._Our approximation_matches_the_finite-size battery simulations very closely_for energy threshold values up_to $80\%$_of the battery capacity, which is $400$ energy_units_in our simulations._In a real setting, one should not charge a_battery completely but up to $80\%$_in order to_improve_the_battery life-span and the_energy efficiency [@battery_uni_2; @website_2].
In Fig. \[fig:fig\_2\],_we plot the energy outage probability,_$\pi_{0}$ in (\[eq:upper\_bound\_pi\_0\]), as a function of_the energy decay rate, $\mu$, for_different scale parameters, $\lambda_{\text{dB}}=10\log_{10}\frac{\lambda}{N\sigma_{w}^{2}}=5$, $4$_and $3$_dB, when the shape parameter_is $k=1$. Note_that the_mean value of_the Weibull distribution is $\lambda\Gamma(1+1/k)$, where_$\Gamma(\cdot)$ is the_Gamma function. So, $\lambda$ becomes the_average_energy arrival rate_when_$k=1$._If we_use all the_energy_for data_transmission,_we can consider $\lambda_{\text{dB}}$ as the_average_signal-to-noise ratio in the channel. In addition,_given the energy decay_rate,_we determine the constant_energy demand rate. Particularly, given_that the energy packets arriving at_the battery_are i.i.d.,_the constant energy demand rate is $p=\frac{1}{\mu}\ln\mathbb{E}_{u}\left[e^{\mu u}\right]=\frac{1}{\mu}\ln\left(\frac{1}{1-\lambda\mu}\right)$ for $0<\mu<\frac{1}{\lambda}$. Notice_that when $\mu$ goes to zero,_$p$ approaches the average_energy arrival_rate,_$\lambda$, whereas when_$\mu$_goes to_$\frac{1}{\lambda}$, $p$ approaches infinity. Therefore, the outage_probability becomes_1 once $\mu$ is greater than_$\frac{1}{\lambda}$ for a given_energy_arrival process with the aforementioned distribution,_which is displayed in Fig. \[fig:fig\_2\]._We can also infer that_the_energy_out |
$ with a small amplitude, with the latter inherited from the transition to the rapidly oscillating phase of the scalar field. As for the SFDM velocity perturbations, see Eq. , because $e^\alpha \to {\rm const.}$ their amplitude quickly decreases as $\rho_\phi
\Theta_\phi \simeq (k^2/2am) \rho_\phi e^\alpha \sim a^{-4}$. Together with the results in the previous section, this reinforces our argument in Sec. \[sec:gauge-ambig-matt\] that the velocity gradient of SFDM evolves as expected of a pressureless component in the gauge corresponding to the comoving frame of a CDM component.
Numerical examples for small scales are also shown in Figs. \[fig:6\] and \[fig:7\] for the masses $m = 10^{-24},
10^{-26} \, {\rm eV}$, and they agree very well with the behavior inferred from the approximate solutions described just above: the density contrast is just an oscillating function with a tiny amplitude, whereas the velocity gradient oscillates with a rapidly decaying amplitude.
Power spectrum and mass function {#sec:mass-power-spectrum}
================================
In order to exploit the constraining power of future galaxy surveys such as DESI[@Levi:2013gra] and LSST[@Ivezic:2008fe], it will be necessary to perform an accurate modeling of the linear, and non linear, matter PS of any given dark matter model, for wavenumbers up to $k
\sim 10 h \, {\rm Mpc}^{-1}$. For $\Lambda$CDM models, linear perturbation theory provides good accuracy up to scales of $k \sim 0.1 h \, {\rm Mpc}^{-1}$, whereas dark matter-only N-body simulations provide good results up to $k \sim 0.5 h \, {\rm
Mpc}^{-1}$, for larger values of $k$ it is necessary to include baryonic physics in the simulations [@Schneider:2015yka].
For the SFDM model, we are able to describe the linear PS at scales that can be considered in the semi-linear regime without any by-hand approximation. A correct treatment of the linear and semi-linear regime impacts the study of the structure formation process in two aspects: on setting up the initial conditions for cosmological simulations, and on the prediction (as a first approach) of observables about large and small scale structure | $ with a small amplitude, with the latter inherited from the transition to the rapidly hover phase of the scalar sphere. As for the SFDM velocity perturbations, see Eq. , because $ e^\alpha \to { \rm const.}$ their amplitude cursorily decreases as $ \rho_\phi
\Theta_\phi \simeq (k^2/2am) \rho_\phi e^\alpha \sim a^{-4}$. Together with the resultant role in the former section, this reinforces our argument in Sec. \[sec: gauge - ambig - matt\ ] that the velocity gradient of SFDM evolves as expect of a pressureless component in the gauge equate to the comoving skeletal system of a CDM component.
numeric examples for small plate are also shown in Figs. \[fig:6\ ] and \[fig:7\ ] for the masses $ m = 10^{-24 },
10^{-26 } \, { \rm eV}$, and they harmonize very well with the behavior inferred from the approximate solutions described merely above: the density contrast is just an oscillating function with a tiny amplitude, whereas the speed gradient oscillates with a rapidly decaying amplitude.
Power spectrum and mass function { # sec: mass - power - spectrum }
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
In club to overwork the constraining power of future galaxy surveys such as DESI[@Levi:2013gra ] and LSST[@Ivezic:2008fe ], it will be necessary to perform an accurate modeling of the linear, and non linear, topic PS of any given blue matter model, for wavenumbers up to $ k
\sim 10 h \, { \rm Mpc}^{-1}$. For $ \Lambda$CDM model, linear perturbation theory provide good accuracy up to scales of $ k \sim 0.1 h \, { \rm Mpc}^{-1}$, whereas dark matter - merely N - body simulations provide good results up to $ k \sim 0.5 h \, { \rm
Mpc}^{-1}$, for larger value of $ k$ it is necessary to include baryonic physics in the simulations [ @Schneider:2015yka ].
For the SFDM model, we are able to identify the linear PS at scale that can be regard in the semi - linear regime without any by - hand estimate. A correct treatment of the linear and semi - linear regime impacts the study of the structure formation summons in two aspects: on arrange up the initial conditions for cosmologic simulations, and on the prediction (as a inaugural approach) of observables about bombastic and small plate structure | $ wihh a small amplitude, witm the latter inhgruted fcom the transitkon to the rapidly oscillatiig pyase if the scalar field. As for the DFDM velicitb perturbations, see Eq. , bceause $c^\alphc \vo {\rm const.}$ thelr amplituda quickly decraared as $\rho_\phi
\Theta_\phi \simeq (k^2/2am) \rho_\phy e^\alphs \dim a^{-4}$. Together witn the gewults in the previous section, this rtinforces our argukent in Sec. \[sec:gauge-ambig-maht\] tjat the velocity ggadient of WFDM wvolves as ebpected of a pressurelgss component in the gauge correrpondnng to the xonovltg frame of a CDI component.
Nmkerican exampkes for small xcanes are also shown in Fijs. \[fig:6\] and \[fig:7\] for the masses $m = 10^{-24},
10^{-26} \, {\rm eV}$, and they qgeee vgry wall duth tht bxhabior ijfecred from tge approximqte solutions descrobqe just above: fhe degsyty contrast is just an oscillating funbtioh with a tiny amplitude, whereas the velocity gradient oscillates with a rapidly decaying amplitude.
Powes spertfum akd mxws function {#sec:mass-power-spectrum}
================================
In order to expjkiu tme constraining iower of future gakady furveys such xs DESN[@Mebi:2013gra] and LSST[@Ivezlc:2008fe], it will be necesfary to perform an accurate modwling of the oinear, and non linzar, matter PR of any biven dark matter model, for wavenumberd up to $k
\akm 10 h \, {\rm Mpc}^{-1}$. Fof $\Lsmtda$CDM models, linear pertuwbation tieory providds gpod acsuracy up ho scales of $k \sim 0.1 h \, {\rl Mpc}^{-1}$, wvereas darn matter-only N-body simulations 'covide good rgsunts up to $k \sim 0.5 h \, {\rm
Mpc}^{-1}$, fjr larger valugs of $k$ ic is ndcessary tk incluve baryonic [hysics in tha simulations [@Schneidqr:2015ykq].
For the SFAO model, we are able to btscribe thw linear PS at scakes fhat can be conwidtrwd in the semi-kindar rvgike fithout any ty-havd xlproxkmation. A cifrecy treatment of the lhnead and semi-linear rrglme impacjs the sttdy of the sttucture formation krocesv ii two sspgcts: on setting up the initial donditiond fjr cosmologisal wimulations, cnd on the prediction (as a first approaci) of observables about oarge and small scakt structure | $ with a small amplitude, with the from transition to rapidly oscillating phase for SFDM velocity perturbations, Eq. , because \to {\rm const.}$ their amplitude quickly as $\rho_\phi \Theta_\phi \simeq (k^2/2am) \rho_\phi e^\alpha \sim a^{-4}$. Together with the results the previous section, this reinforces our argument in Sec. \[sec:gauge-ambig-matt\] that the velocity of evolves expected a pressureless component in the gauge corresponding to the comoving frame of a CDM component. Numerical for small scales are also shown in Figs. and \[fig:7\] for the $m = 10^{-24}, 10^{-26} \, eV}$, they agree well the inferred from the solutions described just above: the density contrast is just an oscillating function with a tiny amplitude, whereas velocity gradient a rapidly amplitude. spectrum mass function {#sec:mass-power-spectrum} order to exploit the constraining power surveys such as DESI[@Levi:2013gra] and LSST[@Ivezic:2008fe], it will necessary to an accurate modeling of the linear, non linear, matter PS of any given dark model, for wavenumbers up to $k \sim 10 h \, {\rm Mpc}^{-1}$. For $\Lambda$CDM models, theory provides good accuracy to scales of \sim h {\rm whereas dark N-body simulations provide good results up to $k \sim 0.5 h {\rm Mpc}^{-1}$, for larger values of $k$ it is necessary baryonic in the simulations For the SFDM model, are to describe the linear scales can the regime any by-hand approximation. A treatment of the linear and regime impacts the study in two aspects: on setting up the initial for cosmological simulations, and on the prediction a first approach) of observables about large and small scale structure | $ with a small amplitude, with thE latter inhEriteD frOm tHe TranSitiOn to the rapidly OScilLating phase of the scalar Field. as FOr thE sFdM velOcity peRTuRBAtiOnS, sEe EQ. , bECaUse $e^\aLphA \to {\rm coNst.}$ their amPliTuDe quickly decREaSes as $\rho_\phI
\ThEta_\phi \simeq (k^2/2Am) \rHo_\phi e^\AlPha \SIm a^{-4}$. ToGetHer wiTh the rESults iN the previOuS SectioN, This reiNFOrCes oUr argument in Sec. \[seC:GaUGe-ambig-matt\] thaT the veLoCItY GRadIenT of SFDM evoLvEs as eXPected oF A pRESSurELess component In the gauge cORreSpondiNg To tHE comovIng frAmE Of a cDM componenT.
NumErical exaMples fOR small sCAles are Also shOwn In FIgs. \[fIG:6\] aNd \[Fig:7\] FoR The MAsSes $M = 10^{-24},
10^{-26} \, {\Rm ev}$, and they AgReE very Well WITH The bEhaVior InferRed from the appRoxImatE SolUtionS descRibeD jUst abOve: the DensiTy Contrast is just aN oscIllating fUncTiOn wItH a tinY AmplitUde, WheReas the VelocitY GraDiENT OsCillates with a rapidLy DECaYing amplItude.
POWeR sPEctrum anD mAss FuncTIOn {#sec:Mass-POwEr-spectrUm}
================================
In orDEr To Exploit ThE constRaIniNg pOwer oF FutuRe galaXy surveyS such AS DESI[@Levi:2013gra] anD lSST[@Ivezic:2008fe], iT WiLL Be NEcesSarY to perform aN accURate ModeLInG of THe linEar, anD nON lINear, matter PS of any giVeN dark mAtter Model, for wavenUmbers up to $K
\SIM 10 h \, {\rm Mpc}^{-1}$. FOr $\LaMBdA$cDM models, lineaR pertUrbation thEOry proviDes goOd accuraCy up to scaLES of $k \sim 0.1 h \, {\Rm MPc}^{-1}$, wHerEas DARk Matter-only N-boDY SimuLaTions prOviDe good rEsuLts Up tO $k \sIm 0.5 H \, {\rm
Mpc}^{-1}$, for Larger vaLuEs Of $K$ iT is NecesSAry to incLuDe bArYonIc phySIcs in tHe simUlatIoNs [@sChnEider:2015ykA].
foR THe SFdM MoDel, wE arE aBle to DescRIbe The lineAr PS at scaLes THat cAn Be ConsideRed in the semi-lInEar regime wItHouT any by-HANd approxImation. A correct treatmenT Of the liNeaR and sEmi-lInear regiMe iMpacts The STudy of The strUcturE fOrmATIon prOCEsS in TwO aspects: on SETtiNg up tHe InitIal condItions for cosmologiCAl sImulations, and On tHe prEDIcTioN (As A FirSt APprOACh) of observables About large AnD SmAll scale stRUctUrE | $ with a small amplitude,with the l atter in her it ed f romthe transition to t he rapidly oscillating phas eo f th e s calar field. As f orth eSFD Mv el ocity pe rturbat ions, seeEq. , because $e^ \ al pha \to {\ rmconst.}$ the iramplit ud e q u ickly de creas es as$ \rho_\ phi
\Thet a_ \ phi \s i meq (k^ 2 / 2a m) \ rho_\phi e^\alpha \s i m a^{-4}$. Tog etherwi t ht h e r esu lts in the p revio u s secti o n, t h isr einforces our argument i n Se c. \[s ec :ga u ge-amb ig-ma tt \ ] t hat the vel ocit y gradien t of S F DM evol v es as e xpecte d o f a pre s su re les sc omp o ne nti n t he gauge c or respo ndin g t o the co movi ng fr ame of a CDMcom pone n t.
Nume rical exa mp les f or sma ll sc al es are also sho wn i n Figs. \ [fi g: 6\] a nd \[ f ig:7\] fo r t he mass es $m = 10^ {- 2 4 } ,10^{-26} \, {\rm e V} $ , a nd theyagreev er yw ell with t hebeha v i or in ferr e dfrom the appro x im at e solut io ns des cr ibe d j ust a b ove: the d ensity c ontra s t is just an o s cillating fun c ti o n w i th a ti ny amplitud e, w h erea s th e v elo c ity g radie nt os c illates with a rapi dl y deca yingamplitude.
P ower spect r u m and mas s fu n ct i on {#sec:mass- power -spectrum} ======== ===== ======== ========= = =
In ord ertoexp loi t th e constrainin g powe rof futu regalaxysur vey s s uch a s DESI[@L evi:2013 gr a] a nd LS ST[@I v ezic:200 8f e], i t w ill b e neces saryto p er fo r m a n accur a te m odel in gof t heli near, and non linear , matterPSo f an ygi ven dar k matter mode l, for waven um ber s up t o $k
\sim10 h \, {\rm Mpc}^{-1}$ . For $\ Lam bda$C DM m odels, li nea r pert urb a tion t heoryprovi de s g o o d acc u r ac y u pto scaleso f $k \sim 0 .1 h \, {\r m Mpc}^{-1}$, wher e asdark matter-o nly N-b o d ysim u la t ion sp rov i d e good resultsup to $k \ si m 0 .5 h \, {\ r m
M pc}^{-1 }$, for larg e r value s of $k$it is nec es sary t o i nclude bar yonic ph ysics int he si m ul ation s [ @Schne id er: 2015y ka].
F orthe S FDM mo de l, weare a bl e to des cribe the linear PS atscales that ca n be cons ide r edin the se mi-l inear regi mewit houtany by-ha nd a p pr oxi m ation . Ac orrect tr e at men t of the linear a n d s emi-l ine a r regi me i mpacts the studyo f the structur e fo r m ati onp roce ss in two aspect s:on s etting u pthe initial conditi on s forcosmol ogical simula t i on s , andon t hepredictio n ( as a first a pp r oach)of o bs ervabl es abo u t la r g e and small scal e str u c ture | $ with_a small_amplitude, with the latter_inherited from_the_transition to_the_rapidly oscillating phase_of the scalar_field. As for the_SFDM velocity perturbations,_see_Eq. , because $e^\alpha \to {\rm const.}$ their amplitude quickly decreases as $\rho_\phi
\Theta_\phi \simeq (k^2/2am)_\rho_\phi_e^\alpha \sim_a^{-4}$._Together_with the results in the_previous section, this reinforces our_argument in_Sec. \[sec:gauge-ambig-matt\] that the velocity gradient of SFDM evolves_as_expected of a_pressureless component in the gauge corresponding to the comoving_frame of a CDM component.
Numerical examples_for small scales_are_also_shown in Figs. \[fig:6\] and \[fig:7\]_for the masses $m = 10^{-24},
10^{-26}_\, {\rm eV}$, and they agree_very well with the behavior inferred from_the approximate solutions described just above:_the density contrast is just_an oscillating_function with a tiny amplitude,_whereas the velocity_gradient oscillates_with a rapidly_decaying amplitude.
Power spectrum and mass function_{#sec:mass-power-spectrum}
================================
In order to_exploit the constraining power of future_galaxy_surveys such as_DESI[@Levi:2013gra]_and_LSST[@Ivezic:2008fe], it_will be necessary_to_perform an_accurate_modeling of the linear, and non_linear,_matter PS of any given dark matter_model, for wavenumbers up_to_$k
\sim 10 h \,_{\rm Mpc}^{-1}$. For $\Lambda$CDM models,_linear perturbation theory provides good accuracy_up to_scales of_$k \sim 0.1 h \, {\rm Mpc}^{-1}$, whereas dark matter-only N-body_simulations provide good results up to_$k \sim 0.5 h_\, {\rm
__Mpc}^{-1}$, for larger_values_of $k$_it is necessary to include baryonic physics_in the_simulations [@Schneider:2015yka].
For the SFDM model, we_are able to describe_the_linear PS at scales that can_be considered in the semi-linear regime_without any by-hand approximation. A_correct_treatment_of the linear and semi-linear_regime impacts the study of the_structure formation process_in two aspects: on setting up the_initial_conditions for cosmological simulations, and on_the_prediction (as a first approach) of_observables_about_large and small scale structure |
Acquiring all these data, for every $\ell$ requires an overall computational cost (apart subleading trends) of $$\# \mbox{cost} \sim L \left(\#q + 2\#p + 1 \right) \left(2 d D^3 + d^2 D^2 \right),$$ where $\#q$ and $\#p$ are respectively the number of one body and two body terms in the Hamiltonian expression. As you see, even in this complex scenario, the scaling behavior with $D$ and $L$ of the computational cost remains roughly the same. Now we can conclude that $$\langle \Psi | \tilde{H}^{\rightarrow}_{\ell} | \Psi \rangle =
{{\mathbb E}}_{{{\mathbb 1}}}^{[1]} \cdot
{{\mathbb E}}_{{{\mathbb 1}}}^{[2]} \cdot \ldots \cdot {{\mathbb E}}_{{{\mathbb 1}}}^{[\ell]} |\xi^{[\ell]}),$$ and in particular $\langle \Psi | H | \Psi \rangle = |\xi^{[0]})$ which is a scalar number, and it is exactly the energy of the state if $H$ is the actual Hamiltonian of the system.
The great improvement in computing expectation values we encountered for separable observable is recovered in case of (short-range) Hamiltonian operators: the full computational cost to acquire the energy (which will be later adopted as variational functional) scales only linearly with the system size. If one thinks that (full-search) exact methods typically bear an exponential cost in $L$, it is easy to understand why DMRG/MPS architectures are regarded with great interest.
Pictorial representation of Matrix\
Product States Tensor Network
-----------------------------------
Through the present chapter, we learned how to deal with MPS: their mathematical properties that allow algebraic manipulation (gauge group), and their physical properties that allow to control quantum entanglement. On the other hand, the equations we encounter start to look cumbersome and confusing, like eq. . To work around this issue, we are now going to provide an alternative way to express MPS representation that is based on diagrams and graph theory rather than standard analytical expressions. This will prove a faster and clearer fashion to represent states, observations, and matrix multiplication; and will become instrumental in later chapters.
Let us start back from our definition of MPS. If our system is a 1D-OBC lattice with $L$ sites, and $\{ |s\rangle \}_ | Acquiring all these data, for every $ \ell$ requires an overall computational cost (aside subleading drift) of $ $ \ # \mbox{cost } \sim L \left(\#q + 2\#p + 1 \right) \left(2 d D^3 + d^2 D^2 \right),$$ where $ \#q$ and $ \#p$ are respectively the number of one soundbox and two consistency terms in the Hamiltonian expression. As you experience, even in this complex scenario, the scale behavior with $ D$ and $ L$ of the computational monetary value remains roughly the like. Now we can conclude that $ $ \langle \Psi | \tilde{H}^{\rightarrow}_{\ell } | \Psi \rangle =
{ { \mathbb E}}_{{{\mathbb 1}}}^{[1 ] } \cdot
{ { \mathbb E}}_{{{\mathbb 1}}}^{[2 ] } \cdot \ldots \cdot { { \mathbb E}}_{{{\mathbb 1}}}^{[\ell ] } |\xi^{[\ell]}),$$ and in particular $ \langle \Psi | H | \Psi \rangle = |\xi^{[0]})$ which is a scalar number, and it is precisely the energy of the state if $ H$ is the actual Hamiltonian of the system.
The great improvement in computing anticipation values we encountered for separable discernible is recovered in case of (inadequate - range) Hamiltonian operator: the full computational cost to acquire the energy (which will be later espouse as variational functional) scales only linearly with the system size. If one thinks that (full - search) exact methods typically bear an exponential cost in $ L$, it is easy to understand why DMRG / MPS architectures are regarded with great interest.
Pictorial representation of Matrix\
Product States Tensor Network
-----------------------------------
Through the present chapter, we learn how to deal with MPS: their mathematical place that leave algebraic manipulation (gauge group), and their physical properties that allow to control quantum entanglement. On the early hand, the equations we encounter start to look cumbersome and confusing, like eq. . To work around this issue, we are now going to provide an alternative direction to express MPS theatrical performance that is free-base on diagrams and graph theory rather than standard analytical expression. This will prove a faster and clearer manner to represent state, observations, and matrix multiplication; and will become implemental in later chapters.
Let us start back from our definition of MPS. If our system is a 1D - OBC wicket with $ L$ sites, and $ \ { |s\rangle \ } _ | Aceuiring all these data, fur every $\ell$ requires en overzll compjtational cost (apart subleadmng rrendw) of $$\# \mbox{cost} \sim L \ldft(\#q + 2\#p + 1 \right) \lwft(2 v D^3 + d^2 D^2 \right),$$ wisre $\#q$ akb $\#p$ adc res'ertively the numner of one tody and two bmdh cerms in the Hamiltonian expression. Ws you xef, even in this comklev scsnario, the scaling behavior with $D$ and $L$ mf the computstional cost remains roughpy tje same. Now we can conclude tyat $$\jqngle \Psi | \tklde{H}^{\rightarrow}_{\ell} | \Psj \rangle =
{{\mathbb E}}_{{{\mathbb 1}}}^{[1]} \cdot
{{\matkbb E}}_{{{\mathbb 1}}}^{[2]} \xdoh \ldots \cdot {{\mathfb E}}_{{{\mathbb 1}}}^{[\ell]} |\xi^{[\ell]}),$$ and in particular $\lakgle \'si | H | \Psi \rangle = |\xi^{[0]})$ whmch is a scalar numbgr, and it hs exactly the enerty of tve sdate uf $F$ ia vhe actuap Hemiltonian kf the systwm.
The great improvekegn in computinf expestwtion values we encountered for separabne kbservable is recovered in case of (short-rangg) Hamiltonyan operators: the full computational cost to acquhre tie entrnn (whkxh will be later adopted as variational functiogzl) sbales only linearjy with the sjsygm size. If one thinks tgat (full-search) exaft methjds ttpically fear an exponential cost in $L$, ir is easy to ynderstand why DMRY/MPS architeetures are tegarded with great intzrest.
Pjctorial reoresentatjun of Matrix\
Prodjct Sdates Tensor Network
-----------------------------------
Througr the prewent chaptef, we learnqd how to feal with MPS: their mathfmatieal psoperties hhat allow algebraic manipulation (gauge group), atd nheir phyfical properties trat allow to cpntrol zuantjm entanglvment. On vhe other hagd, the equatimjs we encounver start to oook cumberrume and confusong, like vq. . To qork around this ixsug, se are now goiny uo provide an alyervatyvv wey to axpress MPS seprdsevyatiov that is bcsdd om diagrams and graph thekry rather than stsnqard anaoytical qxpressions. Tnis will prove a fwster anv cleater fashion to represent states, ogservatiojs, wnd matrix mtltiilicwtion; and cill become instrumental in later chaptecs.
Let us start back fron our definition of KPS. If our xystei is a 1D-OTC lattice with $L$ sires, and $\{ |s\rangle \}_ | Acquiring all these data, for every $\ell$ overall cost (apart trends) of $$\# 2\#p 1 \right) \left(2 D^3 + d^2 \right),$$ where $\#q$ and $\#p$ are the number of one body and two body terms in the Hamiltonian expression. you see, even in this complex scenario, the scaling behavior with $D$ and of computational remains the same. Now we can conclude that $$\langle \Psi | \tilde{H}^{\rightarrow}_{\ell} | \Psi \rangle = {{\mathbb 1}}}^{[1]} \cdot {{\mathbb E}}_{{{\mathbb 1}}}^{[2]} \cdot \ldots \cdot E}}_{{{\mathbb 1}}}^{[\ell]} |\xi^{[\ell]}),$$ and particular $\langle \Psi | H \Psi = |\xi^{[0]})$ is scalar and it is the energy of the state if $H$ is the actual Hamiltonian of the system. The great improvement computing expectation encountered for observable recovered case of (short-range) the full computational cost to acquire will be later adopted as variational functional) scales linearly with system size. If one thinks that exact methods typically bear an exponential cost in it is easy to understand why DMRG/MPS architectures are regarded with great interest. Pictorial representation Product States Tensor Network Through the present we how deal MPS: their properties that allow algebraic manipulation (gauge group), and their physical properties allow to control quantum entanglement. On the other hand, the encounter to look cumbersome confusing, like eq. . work this issue, we are to an express representation is based on diagrams graph theory rather than standard expressions. This will prove to represent states, observations, and matrix multiplication; and become instrumental in later chapters. Let us back from our definition of MPS. If our system is a 1D-OBC with $L$ $\{ |s\rangle \}_ | Acquiring all these data, for eVery $\ell$ reqUires An oVerAlL comPutaTional cost (aparT SublEading trends) of $$\# \mbox{cost} \Sim L \lEfT(\#Q + 2\#p + 1 \riGHt) \Left(2 d d^3 + d^2 D^2 \righT),$$ WhERE $\#q$ aNd $\#P$ aRe rEsPEcTivelY thE number Of one body aNd tWo Body terms in tHE HAmiltonian ExpRession. As you See, Even in ThIs cOMplex SceNario, The scaLIng behAvior with $d$ aND $L$ of thE ComputaTIOnAl coSt remains roughly tHE sAMe. Now we can concLude thAt $$\LAnGLE \PsI | \tiLde{H}^{\rightaRrOw}_{\ell} | \pSi \ranglE =
{{\MaTHBB E}}_{{{\mAThbb 1}}}^{[1]} \cdot
{{\mathbB E}}_{{{\mathbb 1}}}^{[2]} \cdoT \LdoTs \cdot {{\MaThbB e}}_{{{\mathbB 1}}}^{[\ell]} |\xI^{[\eLL]}),$$ anD in particulAr $\laNgle \Psi | H | \PSi \rangLE = |\xi^{[0]})$ whicH Is a scalAr numbEr, aNd iT is eXAcTlY thE eNErgY Of The STatE if $H$ is thE aCtUal HaMiltONIAN of tHe sYsteM.
The gReat improvemeNt iN comPUtiNg expEctatIon vAlUes we EncounTered FoR separable obserVablE is recoveRed In CasE oF (shorT-Range) HAmiLtoNian opeRators: tHE fuLl COMPuTational cost to acquIrE THe Energy (whIch wilL Be LaTEr adopteD aS vaRiatIONal fuNctiONaL) scales oNly linEArLy With the SyStem siZe. if oNe tHinks THat (fUll-seaRch) exact MethoDS typically bear AN exponential cOSt IN $l$, iT Is eaSy tO understand Why DmrG/MPs arcHItEctURes arE regaRdED wITh great interest.
PictOrIal repResenTation of MatriX\
Product StATES Tensor NEtwoRK
-----------------------------------
THRough the presenT chapTer, we learnED how to deAl witH MPS: theiR mathematICAl properTieS thAt aLloW ALgEbraic manipulATIon (gAuGe group), And Their phYsiCal ProPerTiEs that allOw to contRoL qUaNtUm eNtangLEment. On tHe OthEr HanD, the eQUationS we enCounTeR sTArt To look cUMbERSome AnD cOnfuSinG, lIke eq. . to woRK arOund thiS issue, we aRe nOW goiNg To Provide An alternative WaY to express mPs rePresenTATion that Is based on diagrams and graPH theory RatHer thAn stAndard anaLytIcal exPreSSions. THis wilL provE a FasTER and cLEArEr fAsHion to reprESEnt StateS, oBserVations, And matrix multiplicATioN; and will becomE inStruMENtAl iN LaTEr cHaPTerS.
lEt us start back frOm our definItIOn Of MPS. If our SYstEm Is a 1D-OBC Lattice With $L$ SItes, and $\{ |S\rangle \}_ | Acquiring all these data, for every $\el l$req ui resan o verall computa t iona l cost (apart subleadi ng tr en d s) o f $ $\# \ mbox{co s t} \ sim L \ lef t( \ #q + 2\ #p+ 1 \ri ght) \left (2dD^3 + d^2 D^ 2 \ right),$$whe re $\#q$ and $\ #p$ ar eres p ectiv ely thenumber of one body and t w o body terms i n th e Ha miltonian express i on . As you see, e ven in t h is c omp lex scenario, t he sc a ling be h av i o r wi t h $D$ and $L$ of the com p uta tional c ost remain s rou gh l y t he same. No w we can conc lude t h at $$\l a ngle \P si | \ til de{ H}^{ \ ri gh tar ro w }_{ \ el l}| \P si \rang le =
{{ \mat h b b E}}_ {{{ \mat hbb 1 }}}^{[1]} \cd ot {{ \ mat hbb E }}_{{ {\ma th bb 1} }}^{[2 ]} \c do t \ldots \cdot{{\m athbb E}} _{{ {\ mat hb b 1}} } ^{[\el l]} |\ xi^{[\e ll]}),$ $ an di n pa rticular $\langle\P s i | H | \Ps i \ran g le = |\xi^{[0 ]} )$whic h is ascal a rnumber,and it is e xactlyth e ener gy of th e sta t e if $H$ i s the ac tualH amiltonian oft he system.
T h eg r ea t imp rov ement in co mput i ng e xpec t at ion value s ween c ou n tered for separable o bserva ble i s recovered i n case of( s h ort-rang e) H a mi l tonian operato rs: t he full co m putation al co st to ac quire the e nergy (w hic h w ill be l at er adopted as v aria ti onal fu nct ional)sca les on lyli nearly wi th the s ys te msi ze. If o n e thinks t hat ( ful l-sea r ch) ex act m etho ds t y pic ally be a ra n exp on en tial co st in $ L$,i t i s easyto unders tan d why D MR G/MPS a rchitecturesar e regarded w ith great i nterest.
Pictorial representat i on of M atr ix\
P rodu ct States Te nsor N etw o rk
--- ------ ----- -- --- - - ----- - - -- --- --
Throught h e p resen tchap ter, we learned how to de a l w ith MPS: thei r m athe m a ti cal pr o per ti e s t h a t allow algebra ic manipul at i on (gauge gr o up) ,and the ir phys icalp roperti es that a llow to c on trol q uan tum entang lement.On the ot h er ha n d, theequ ations w e e ncoun ter st a rtto lo ok cum be rsomeand c on fusing,like eq. . To work arou nd thi s iss ue, we are n owg oin g to prov idean alterna tiv e w ay to ex p ressMPSr ep res e ntati on t h at is bas e dond i ag rams and gr a p h th eoryrat h er tha n st andard analytical expressions. T hisw i llpro v e afa ster and clear erfa s h ion to r ep resent stat es, obse rv a tions , andmatrix multip l i ca t ion; a nd w ill become i nst ru m ental i nla t er cha pter s.
Letus sta r t ba c k from our defini tiono f MPS. Ifour s ys tem isa 1D- OBC lattic e with $L$sites, and $\{|s\rang le \}_ | Acquiring_all these_data, for every $\ell$_requires an_overall_computational cost_(apart_subleading trends) of_$$\# \mbox{cost} \sim_L \left(\#q + 2\#p_+ 1 \right)_\left(2_d D^3 + d^2 D^2 \right),$$ where $\#q$ and $\#p$ are respectively the number_of_one body_and_two_body terms in the Hamiltonian_expression. As you see, even_in this_complex scenario, the scaling behavior with $D$ and_$L$_of the computational_cost remains roughly the same. Now we can conclude_that $$\langle \Psi | \tilde{H}^{\rightarrow}_{\ell} |_\Psi \rangle =_
_{{\mathbb_E}}_{{{\mathbb 1}}}^{[1]} \cdot
_{{\mathbb E}}_{{{\mathbb 1}}}^{[2]} \cdot \ldots \cdot_{{\mathbb E}}_{{{\mathbb 1}}}^{[\ell]} |\xi^{[\ell]}),$$ and in_particular $\langle \Psi | H | \Psi_\rangle = |\xi^{[0]})$ which is a_scalar number, and it is_exactly the_energy of the state if_$H$ is the_actual Hamiltonian_of the system.
The_great improvement in computing expectation values_we encountered for_separable observable is recovered in case_of_(short-range) Hamiltonian operators:_the_full_computational cost_to acquire the_energy_(which will_be_later adopted as variational functional) scales_only_linearly with the system size. If one_thinks that (full-search) exact_methods_typically bear an exponential_cost in $L$, it is_easy to understand why DMRG/MPS architectures_are regarded_with great_interest.
Pictorial representation of Matrix\
Product States Tensor Network
-----------------------------------
Through the present chapter, we_learned how to deal with MPS:_their mathematical properties that_allow algebraic_manipulation_(gauge group), and_their_physical properties_that allow to control quantum entanglement. On_the other_hand, the equations we encounter start_to look cumbersome and_confusing,_like eq. . To work around this_issue, we are now going to_provide an alternative way to_express_MPS_representation that is based on_diagrams and graph theory rather than_standard analytical expressions._This will prove a faster and clearer_fashion_to represent states, observations, and matrix_multiplication;_and will become instrumental in later_chapters.
Let_us_start back from our definition_of MPS. If our system is_a 1D-OBC lattice with $L$ sites, and $\{ |s\rangle_\}_ |
ron wave function in the inner region is similar to that of the SU(3)-limit $^4$He-$^2n$ state. In this region, we have better to call it the spin-zero $2n$ correlation(dineutron correlation) rather than the $^2n$ cluster, because the antisymmetrization effect is important there.
Comparing the result of $^8$He$(0^+_1)$ with that of $^6$He($0^+_1$), we found that the reduced width amplitude for the dineutron component is suppressed in the $^8$He$(0^+_1)$. This is because of the $p_{3/2}$ sub-shell closure effect. As mentioned in the previous section, the $j$-$j$ coupling feature is more remarkable in the $^8$He$(0^+_1)$ than the $^6$He($0^+_1$). However, the cluster probability of the $^8$He$(0^+_1)$ is still significant as $S^{\rm fac}=0.57$ and 0.52 in the v58 and the m56 results, respectively. This probability dominantly originates in the SU(3)-limit $^4$He+$^2n$+$^2n$ configuration, which is equivalent to the $L$-$S$ coupling $p$-shell configuration. It means that the dineutron correlation is still important in the $^8$He($0^+_1$). This situation is quite similar to that of the $^{12}$C$(0^+_1)$ which is the admixture of the $p_{3/2}$ closure and the SU(3)-limit $3\alpha$ state. As a result of the $L$-$S$ coupling feature due to the dineutron correlation, the $^8$He$(0^+_1)$ state should contain the significant $(p_{3/2})^2(p_{1/2})^2$ contamination. This result is consistent with the experimental indication of the $p_{1/2}$ component in the $^8$He ground state reported by the recent observations[@Chulkov05; @Keeley07]. As seen in Fig. \[fig:yl\], it is also interesting that the $^8$He($0^+_1$) state has a tail of the $^2n$-cluster motion at the surface, though the tail is slight compared with the long tail in the $^6$He($0^+_1$). In | ron wave function in the inner region is similar to that of the SU(3)-limit $ ^4$He-$^2n$ department of state. In this area, we have better to visit it the tailspin - zero $ 2n$ correlation(dineutron correlation) rather than the $ ^2n$ cluster, because the antisymmetrization consequence is important there.
Comparing the consequence of $ ^8$He$(0^+_1)$ with that of $ ^6$He($0^+_1 $), we found that the reduced width amplitude for the dineutron part is suppressed in the $ ^8$He$(0^+_1)$. This is because of the $ p_{3/2}$ sub - shell closure effect. As mentioned in the former section, the $ j$-$j$ coupling feature of speech is more remarkable in the $ ^8$He$(0^+_1)$ than the $ ^6$He($0^+_1 $). However, the bunch probability of the $ ^8$He$(0^+_1)$ is even significant as $ S^{\rm fac}=0.57 $ and 0.52 in the v58 and the m56 results, respectively. This probability dominantly originates in the SU(3)-limit $ ^4$He+$^2n$+$^2n$ shape, which is equivalent to the $ L$-$S$ coupling $ p$-shell configuration. It means that the dineutron correlation is still significant in the $ ^8$He($0^+_1 $). This situation is quite similar to that of the $ ^{12}$C$(0^+_1)$ which is the admixture of the $ p_{3/2}$ closure and the SU(3)-limit $ 3\alpha$ state. As a resultant role of the $ L$-$S$ coupling feature due to the dineutron correlation, the $ ^8$He$(0^+_1)$ state should contain the significant $ (p_{3/2})^2(p_{1/2})^2 $ contamination. This result is consistent with the experimental indication of the $ p_{1/2}$ component in the $ ^8$He ground state reported by the recent observations[@Chulkov05; @Keeley07 ]. As seen in Fig. \[fig: yl\ ], it is also interesting that the $ ^8$He($0^+_1 $) state of matter has a tail of the $ ^2n$-cluster apparent motion at the surface, though the tail is slight compared with the retentive tail in the $ ^6$He($0^+_1 $). In | ron wave function in the inker region is similar tm that of the RU(3)-limit $^4$He-$^2n$ state. In this rejion, we hqve better to call it ghe spin-zvro $2n$ coreelauion(dineutron corcslation) rathed thau vhe $^2n$ cluster, bgcause the attisymmetrizathov zffect is important there.
Comparing tre resukt of $^8$He$(0^+_1)$ with thwt og $^6$He($0^+_1$), sv nound that the reduced width ampmitude hor the dineutrpn component is suppressed in hhe $^8$He$(0^+_1)$. This is becwuse of the $p_{3/2}$ stv-shell closufe effect. As mentioned in the previous section, the $j$-$j$ coupking featutz is kore remarkeble ig the $^8$He$(0^+_1)$ thak the $^6$Ha($0^+_1$). Howevrr, the cluster prmbavility of the $^8$He$(0^+_1)$ is svill significant as $F^{\rm fac}=0.57$ atd 0.52 in the v58 and thw n56 resolts, sespdxtixelg. Vhia probwbimity dominzntly origibates in the SU(3)-limiu $^4$Hq+$^2b$+$^2n$ configuratjon, whycr is equivalent to the $L$-$S$ coupling $p$-shenl donfiguration. It means rhat the dineutron cotrelation ys still important in the $^8$He($0^+_1$). This situation is quhte smmklax to tfqt of the $^{12}$C$(0^+_1)$ which is the admixture of the $p_{3/2}$ clofhrt akd the SU(3)-limit $3\ajpha$ state. Ss a tesult of the $U$-$S$ cou'mihg feature due to hhe dingutron correlatyon, yhe $^8$He$(0^+_1)$ state should contain the signifibant $(p_{3/2})^2(p_{1/2})^2$ contamination. Chis result ns conxistemt with the experimentau insication of the $p_{1/2}$ cojoonent in the $^8$He grputd state reported by the rqcent obsxrvatnons[@Chulyov05; @Leeley07]. As seen ij Fig. \[nhg:yl\], it is also inheresjing tvat the $^8$He($0^+_1$) state has a tail of the $^2n$-clustxc motion at tne sugface, thobgh thc tail is slighe compared witk the loug taiu in the $^6$Hv($0^+_1$). In | ron wave function in the inner region to of the $^4$He-$^2n$ state. In to it the spin-zero correlation(dineutron correlation) rather the $^2n$ cluster, because the antisymmetrization is important there. Comparing the result of $^8$He$(0^+_1)$ with that of $^6$He($0^+_1$), we that the reduced width amplitude for the dineutron component is suppressed in the This because the sub-shell closure effect. As mentioned in the previous section, the $j$-$j$ coupling feature is more remarkable the $^8$He$(0^+_1)$ than the $^6$He($0^+_1$). However, the cluster of the $^8$He$(0^+_1)$ is significant as $S^{\rm fac}=0.57$ and in v58 and m56 respectively. probability dominantly originates the SU(3)-limit $^4$He+$^2n$+$^2n$ configuration, which is equivalent to the $L$-$S$ coupling $p$-shell configuration. It means that the correlation is in the This is similar to that $^{12}$C$(0^+_1)$ which is the admixture of and the SU(3)-limit $3\alpha$ state. As a result the $L$-$S$ feature due to the dineutron correlation, $^8$He$(0^+_1)$ state should contain the significant $(p_{3/2})^2(p_{1/2})^2$ contamination. result is consistent with the experimental indication of the $p_{1/2}$ component in the $^8$He ground by the recent observations[@Chulkov05; As seen in \[fig:yl\], is interesting the $^8$He($0^+_1$) has a tail of the $^2n$-cluster motion at the surface, though tail is slight compared with the long tail in the | ron wave function in the inner Region is siMilar To tHat Of The Su(3)-limIt $^4$He-$^2n$ state. In thIS regIon, we have better to call iT the sPiN-Zero $2N$ CoRrelaTion(dinEUtRON coRrElAtiOn) RAtHer thAn tHe $^2n$ clusTer, because The AnTisymmetrizaTIoN effect is iMpoRtant there.
CoMpaRing thE rEsuLT of $^8$He$(0^+_1)$ WitH that Of $^6$He($0^+_1$), we FOund thAt the reduCeD Width aMPlitude FOR tHe diNeutron component iS SuPPressed in the $^8$He$(0^+_1)$. this is BeCAuSE Of tHe $p_{3/2}$ Sub-shell clOsUre efFEct. As meNTiONED in THe previous secTion, the $j$-$j$ coUPliNg featUrE is MOre remArkabLe IN thE $^8$He$(0^+_1)$ than the $^6$HE($0^+_1$). HowEver, the clUster pRObabiliTY of the $^8$HE$(0^+_1)$ is stiLl sIgnIficANt As $s^{\rm FaC}=0.57$ And 0.52 IN tHe v58 ANd tHe m56 resulTs, ReSpectIvelY. tHIS proBabIlitY domiNantly originaTes In thE sU(3)-lImit $^4$HE+$^2n$+$^2n$ coNfigUrAtion, Which iS equiVaLent to the $L$-$S$ coupLing $P$-shell conFigUrAtiOn. it meaNS that tHe dIneUtron coRrelatiON is StILL ImPortant in the $^8$He($0^+_1$). This SiTUAtIon is quiTe simiLAr To THat of the $^{12}$c$(0^+_1)$ wHicH is tHE AdmixTure OF tHe $p_{3/2}$ closuRe and tHE Su(3)-lImit $3\alpHa$ State. AS a ResUlt Of the $l$-$s$ couPling fEature duE to thE Dineutron correLAtion, the $^8$He$(0^+_1)$ staTE sHOUlD ContAin The significAnt $(p_{3/2})^2(P_{1/2})^2$ ContAminATiOn. THIs resUlt is CoNSiSTent with the experimeNtAl indiCatioN of the $p_{1/2}$ componEnt in the $^8$He GROUnd state RepoRTeD By the recent obsErvatIons[@ChulkoV05; @keeley07]. As Seen iN Fig. \[fig:yL\], it is also INTerestinG thAt tHe $^8$HE($0^+_1$) stATE hAs a tail of the $^2n$-CLUsteR mOtion at The Surface, ThoUgh The TaiL iS slight coMpared wiTh ThE lOnG taIl in tHE $^6$He($0^+_1$). In | ron wave function in the i nner regio n issim ila rto t hatof the SU(3)-l i mit$^4$He-$^2n$ state. In this r e gion , w e hav e bette r t o cal lit th es pi n-zer o $ 2n$ cor relation(d ine ut ron correlat i on ) rather t han the $^2n$ c lus ter, b ec aus e theant isymm etriza t ion ef fect is i mp o rtantt here.
C o mp arin g the result of $ ^ 8$ H e$(0^+_1)$ wit h that o f $ ^ 6 $He ($0 ^+_1$), we f oundt hat the re d u c edw idth amplitud e for the d i neu tron c om pon e nt issuppr es s edin the $^8$ He$( 0^+_1)$.This i s becaus e of the $p_{3 /2} $ s ub-s h el lclo su r e e f fe ct. Asmentione din theprev i o u s sec tio n, t he $j $-$j$ couplin g f eatu r e i s mor e rem arka bl e inthe $^ 8$He$ (0 ^+_1)$ than the $^6 $He($0^+_ 1$) .How ev er, t h e clus ter pr obabili ty of t h e $ ^8 $ H e $( 0^+_1)$ is still s ig n i fi cant as$S^{\r m f ac } =0.57$ a nd 0. 52 i n the v 58 a n dthe m56result s ,re spectiv el y. Thi spro bab ility domi nantly origina tes i n the SU(3)-lim i t $^4$He+$^2n $ +$ ^ 2 n$ conf igu ration, whi ch i s equ ival e nt to the $ L$-$S $c ou p ling $p$-shell conf ig uratio n. It means that t he dineutr o n correlat ioni ss till important in t he $^8$He( $ 0^+_1$). This situati on is qui t e similar to th atoft h e$^{12}$C$(0^+ _ 1 )$ w hi ch is t headmixtu reofthe $p _{ 3/2}$ clo sure and t he S U( 3)- limit $3\alpha $sta te . A s a r e sult o f the $L$ -$ S$ cou pling f e at u r e du eto the di ne utron cor r ela tion, t he $^8$He $(0 ^ +_1) $st ate sho uld contain t he significa nt $( p_{3/2 } ) ^2(p_{1/ 2})^2$ contamination. T h is resu ltis co nsis tent with th e expe rim e ntal i ndicat ion o fthe $ p_{1/ 2 } $com po nent in th e $^8 $He g ro undstate r eported by the rec e ntobservations[ @Ch ulko v 0 5; @K e el e y07 ]. Ass e en in Fig. \[fi g:yl\], it i s a lso intere s tin gthat th e $^8$H e($0^ + _1$) st ate has a tail ofth e $^ 2 n $-c luster mot ion at t he surfac e , tho u gh thetai l is s li ght comp ared w i ththe l ong ta il in th e $^6 $H e($0^+_1 $). In | ron wave_function in_the inner region is_similar to_that_of the_SU(3)-limit_$^4$He-$^2n$ state. In_this region, we_have better to call_it the spin-zero_$2n$_correlation(dineutron correlation) rather than the $^2n$ cluster, because the antisymmetrization effect is important there.
Comparing_the_result of_$^8$He$(0^+_1)$_with_that of $^6$He($0^+_1$), we found_that the reduced width amplitude_for the_dineutron component is suppressed in the $^8$He$(0^+_1)$. This_is_because of the_$p_{3/2}$ sub-shell closure effect. As mentioned in the previous_section, the $j$-$j$ coupling feature is_more remarkable in_the_$^8$He$(0^+_1)$_than the $^6$He($0^+_1$). However,_the cluster probability of the $^8$He$(0^+_1)$_is still significant as $S^{\rm fac}=0.57$_and 0.52 in the v58 and the_m56 results, respectively. This probability dominantly_originates in the SU(3)-limit $^4$He+$^2n$+$^2n$_configuration, which_is equivalent to the $L$-$S$_coupling $p$-shell configuration._It means_that the dineutron_correlation is still important in the_$^8$He($0^+_1$). This situation_is quite similar to that of_the_$^{12}$C$(0^+_1)$ which is_the_admixture_of the_$p_{3/2}$ closure and_the_SU(3)-limit $3\alpha$_state._As a result of the $L$-$S$_coupling_feature due to the dineutron correlation, the_$^8$He$(0^+_1)$ state should contain_the_significant $(p_{3/2})^2(p_{1/2})^2$ contamination. This_result is consistent with the_experimental indication of the $p_{1/2}$ component_in the_$^8$He ground_state reported by the recent observations[@Chulkov05; @Keeley07]. As seen in Fig. \[fig:yl\],_it is also interesting that the_$^8$He($0^+_1$) state has a_tail of_the_$^2n$-cluster motion at_the_surface, though_the tail is slight compared with the_long tail_in the $^6$He($0^+_1$). In |
) Fock V 1934 Zur Quantenelektrodynamik [*Phys. Zeit. der Sowjetunion*]{} [**6**]{} 425 Akhiezer A I and Berestetski V B 1965 [*Quantum Electrodynamics*]{} (New York: Interscience) Ch. 1 Schweber S S 1961 [*An Introduction to Relativistic Quantum Field Theory*]{} (Evanston: Row, Peterson and Co) Ch. 9 Bialynicki-Birula I and Bialynicka-Birula Z 1975 [*Quantum Electrodynamics*]{} (Oxford: Pergamon) Ch. 9 Pauli W 1980 [*General Principles of Quantum Mechanics*]{} (Berlin: Springer) Ch. 25 Cohen-Tannoudji C, Dupont-Roc J and Grynberg G 1989 [*Photons and Atoms: Introduction to Quantum Electrodynamics*]{} (New York: Wiley) Ch. 1 Foldy L L 1956 Synthesis of covariant particle equations [*Phys. Rev.*]{} [**102**]{}, 568. This form of the generators is taken from [@lm]. Lomont J S and Moses H E 1962 Simple realizations of the infinitesimal generators of the proper orthochronous inhomogeneous Lorentz group for mass zero [*J. Math. Phys.*]{} [**3**]{}, 405 Bialynicki-Birula I and Bialynicka-Birula Z 1987 Berry’s phase in the relativistic theory of spinning particles, [*Phys. Rev. D*]{} [**35**]{} 2383 Bialynicki-Birula I and Bialynicka-Birula Z 2006 Beams of electromagnetic radiation carrying angular momentum: The Riemann-Silberstein vector and the classical-quantum correspondence [*Opt. Comm.*]{} [**264**]{} 342 (ArXiv: quant-ph/05011011)
---
abstract: 'We present deep narrowband observations with high spatial resolution of extraplanar diffuse ionized gas in the halo of NGC891, obtained with the WFPC2 on-board the HST. Our H$\alpha$ observations, centered on the northern part of NGC891, reveal an extended gaseous halo, which fills almost the entire field of view of our WFPC2 observation. Whereas NGC891 has been studied extensively with ground-based telescopes, here the small scale structure of the extended emission line gas is presented at high spatial resolution of 01, corresponding to 4.6pc at the distance to NGC891. The majority of the H$\alpha$ emission is diffuse. Several discrete features (e.g | ) Fock V 1934 Zur Quantenelektrodynamik [ * Phys. Zeit. der Sowjetunion * ] { } [ * * 6 * * ] { } 425 Akhiezer A I and Berestetski V B 1965 [ * Quantum Electrodynamics * ] { } (New York: Interscience) Ch. 1 Schweber S S 1961 [ * An Introduction to Relativistic Quantum Field Theory * ] { } (Evanston: Row, Peterson and Co) Ch. 9 Bialynicki - Birula I and Bialynicka - Birula Z 1975 [ * Quantum Electrodynamics * ] { } (Oxford: Pergamon) Ch. 9 Pauli W 1980 [ * General Principles of Quantum Mechanics * ] { } (Berlin: Springer) Ch. 25 Cohen - Tannoudji C, Dupont - Roc J and Grynberg G 1989 [ * Photons and Atoms: initiation to Quantum Electrodynamics * ] { } (New York: Wiley) Ch. 1 Foldy L L 1956 Synthesis of covariant atom equations [ * Phys. Rev. * ] { } [ * * 102 * * ] { }, 568. This form of the generator is take from [ @lm ]. Lomont J S and Moses H E 1962 Simple realizations of the infinitesimal generator of the proper orthochronous inhomogeneous Lorentz group for mass zero [ * J. Math. Phys. * ] { } [ * * 3 * * ] { }, 405 Bialynicki - Birula I and Bialynicka - Birula Z 1987 Berry ’s phase in the relativistic hypothesis of spinning particles, [ * Phys. Rev. D * ] { } [ * * 35 * * ] { } 2383 Bialynicki - Birula I and Bialynicka - Birula Z 2006 Beams of electromagnetic radiation post angular momentum: The Riemann - Silberstein vector and the classical - quantum correspondence [ * Opt. Comm. * ] { } [ * * 264 * * ] { } 342 (ArXiv: quant - ph/05011011)
---
abstract:' We award deep narrowband observations with high spatial resolution of extraplanar diffuse ionized gas in the aura of NGC891, obtained with the WFPC2 on - board the HST. Our H$\alpha$ observations, center on the northern part of NGC891, reveal an drawn-out gaseous halo, which fills almost the entire field of opinion of our WFPC2 observation. Whereas NGC891 has been studied extensively with ground - based telescopes, here the belittled scale structure of the extended emission line gas is presented at high spatial resolution of 01, corresponding to 4.6pc at the distance to NGC891. The majority of the H$\alpha$ discharge is diffuse. Several discrete features (e.g | ) Fofk V 1934 Zur Quantenelektroaynamik [*Phys. Zenr. der Vowjethnion*]{} [**6**]{} 425 Xkhiezer A I and Berestetski V B 1965 [*Qyantum Electrodynamics*]{} (New York: Interscuenct) Ch. 1 Schweber S S 1961 [*An Intvjducflon tm Relativistic Auantum Fiald Theory*]{} (Evatsgou: Row, Peterson and Co) Ch. 9 Bialynicki-Firula O wnd Bialynicka-Firukw Z 1975 [*Quantum Electrodynamics*]{} (Oxford: Pedgamon) Bh. 9 Pauli W 1980 [*Genetal Principles of Quantum Lechwnics*]{} (Berlin: Sprinher) Ch. 25 Cohgh-Tagboudji C, Dupunt-Roc J and Grynberg F 1989 [*Photons and Atoms: Introductiun to Quantum Eoextrlgynamics*]{} (Nex York: Wiley) Ch. 1 Foldy L L 1956 Synthrsis of covarisnt paeticle equations [*Phys. Rev.*]{} [**102**]{}, 568. This form os the genarctors is taken from [@ln]. Oomonj J S and Nosds G X 1962 Aimple reelizations kf the infibitesimal generatorx jd the proper krthocrrjnous inhomogeneous Lorentz group for mdss zero [*J. Math. Phys.*]{} [**3**]{}, 405 Biaoynicki-Birula I and Blalynicka-Firula Z 1987 Berry’s phase in the relativistic theory of s'ivniun oqrhicles, [*Phys. Rev. D*]{} [**35**]{} 2383 Bialynicki-Birula I and Biwmymibka-Birula Z 2006 Beamf of electrpmwgmgtic radiation carrynhg angular momentum: Hhe Rieiann-Sulbersteig vevtor and the classical-quantym corresponbenxe [*Opt. Comm.*]{} [**264**]{} 342 (ArXir: quant-ph/05011011)
---
abrtravt: 'We present deep narrowbanb obsedvations wihh high slxtial resolution of eftraplanar diffuse ionized gas in tie hako of NEC891, ontaineq with the WFPC2 on-board the HST. Oug H$\alkha$ obvervations, centered on the northern part of NGC891, reveal sn exnended gafeous halo, which fylls almost thg entire yield uf view of our WF'C2 observatijn. Whereas NGW891 has been stndied extqnsicely with gfuund-based telexcopes, hege the small scale strucbure ur the extended zoussion line gas is prqsvntxd at vigh spatial resulugoon ow 01, corrtfpivdinb to 4.6pc at the distatce fo NGC891. The majoritu jf the H$\qlpha$ emyssion is difguse. Several discrtte feetures (e.g | ) Fock V 1934 Zur Quantenelektrodynamik [*Phys. Sowjetunion*]{} 425 Akhiezer I and Berestetski (New Interscience) Ch. 1 S S 1961 Introduction to Relativistic Quantum Field Theory*]{} Row, Peterson and Co) Ch. 9 Bialynicki-Birula I and Bialynicka-Birula Z 1975 [*Quantum (Oxford: Pergamon) Ch. 9 Pauli W 1980 [*General Principles of Quantum Mechanics*]{} (Berlin: Ch. Cohen-Tannoudji Dupont-Roc and Grynberg G 1989 [*Photons and Atoms: Introduction to Quantum Electrodynamics*]{} (New York: Wiley) Ch. 1 L L 1956 Synthesis of covariant particle equations Rev.*]{} [**102**]{}, 568. This of the generators is taken [@lm]. J S Moses E Simple realizations of infinitesimal generators of the proper orthochronous inhomogeneous Lorentz group for mass zero [*J. Math. Phys.*]{} [**3**]{}, 405 I and 1987 Berry’s in relativistic of spinning particles, D*]{} [**35**]{} 2383 Bialynicki-Birula I and Beams of electromagnetic radiation carrying angular momentum: The vector and classical-quantum correspondence [*Opt. Comm.*]{} [**264**]{} 342 quant-ph/05011011) --- abstract: 'We present deep narrowband observations high spatial resolution of extraplanar diffuse ionized gas in the halo of NGC891, obtained with on-board the HST. Our observations, centered on northern of reveal extended gaseous which fills almost the entire field of view of our WFPC2 Whereas NGC891 has been studied extensively with ground-based telescopes, here scale of the extended line gas is presented high resolution of 01, corresponding at distance majority the emission is diffuse. Several features (e.g | ) Fock V 1934 Zur QuantenelektrodynAmik [*Phys. ZeIt. der sowJetUnIon*]{} [**6**]{} 425 AKhieZer A I and BerestETski v B 1965 [*Quantum ElectrodynamiCs*]{} (New yoRK: IntERsCiencE) Ch. 1 SchwEBeR s s 1961 [*An inTrOduCtIOn To RelAtiVistic QUantum FielD ThEoRy*]{} (Evanston: RoW, peTerson and CO) Ch. 9 bialynicki-BiRulA I and BIaLynICka-BiRulA Z 1975 [*QuaNtum ElECtrodyNamics*]{} (OxfOrD: pergamON) Ch. 9 PaulI w 1980 [*geNeraL Principles of QuanTUm mEchanics*]{} (Berlin: springEr) cH. 25 COHEn-TAnnOudji C, DupoNt-roc J aND GrynbeRG G 1989 [*pHOTonS And Atoms: IntroDuction to QuANtuM ElectRoDynAMics*]{} (NeW York: wiLEy) CH. 1 Foldy L L 1956 SynThesIs of covarIant paRTicle eqUAtions [*PHys. Rev.*]{} [**102**]{}, 568. thiS foRm of THe GeNerAtORs iS TaKen FRom [@Lm]. Lomont j S AnD MoseS H E 1962 SIMPLE reaLizAtioNs of tHe infinitesimAl gEnerATorS of thE propEr orThOchroNous inHomogEnEous Lorentz grouP for Mass zero [*J. matH. PHys.*]{} [**3**]{}, 405 biAlyniCKi-BiruLa I And bialyniCka-BiruLA Z 1987 BErRY’S PhAse in the relativistIc THEoRy of spinNing paRTiClES, [*Phys. Rev. d*]{} [**35**]{} 2383 BIalYnicKI-birulA I anD biAlynicka-birula z 2006 beAmS of elecTrOmagneTiC raDiaTion cARryiNg anguLar momenTum: ThE riemann-SilbersTEin vector and tHE cLASsICal-qUanTum correspoNdenCE [*Opt. comm.*]{} [**264**]{} 342 (aRXIv: qUAnt-ph/05011011)
---
AbstrAcT: 'we PResent deep narrowbanD oBservaTions With high spatiAl resolutiON OF extraplAnar DIfFUse ionized gas iN the hAlo of NGC891, obTAined witH the WfPC2 on-boaRd the HST. OUR h$\alpha$ obSerVatIonS, ceNTErEd on the northeRN Part Of nGC891, reveAl aN extendEd gAseOus HalO, wHich fills Almost thE eNtIrE fIelD of viEW of our WFpC2 ObsErVatIon. WhEReas NGc891 has bEen sTuDiED exTensiveLY wITH groUnD-bAsed TelEsCopes, Here THe sMall scaLe structuRe oF The eXtEnDed emisSion line gas is PrEsented at hIgH spAtial rESOlution oF 01, corresponding to 4.6pc at the DIstance To NgC891. The MajoRity of the h$\alPha$ emiSsiON is difFuse. SeVeral DiScrETE featUREs (E.g | ) Fock V 1934 Zur Quantene lektrodyna mik [ *Ph ys. Z eit. der Sowjetunion*] { } [* *6**]{} 425 Akhiezer A I an dB eres t et ski V B 1965 [* Q u ant um E lec tr o dy namic s*] {} (New York: Int ers ci ence) Ch. 1S ch weber S S196 1 [*An Intro duc tion t oRel a tivis tic Quan tum Fi e ld The ory*]{} ( Ev a nston: Row, Pe t e rs on a nd Co) Ch. 9 Bial y ni c ki-Birula I an d Bial yn i ck a - Bir ula Z 1975 [* Qu antum Electro d yn a m i cs* ] {} (Oxford: P ergamon) Ch . 9PauliW198 0 [*Gen eralPr i nci ples of Qua ntum Mechanic s*]{}( Berlin: Springe r) Ch. 25 Co hen- T an no udj iC , D u po nt- R ocJ and Gr yn be rg G1989 [ * P hoto nsandAtoms : Introductio n t o Qu a ntu m Ele ctrod ynam ic s*]{} (NewYork: W iley) Ch. 1 Fol dy L L 1956 S ynt he sis o f cov a riantpar tic le equa tions [ * Phy s. R e v. *]{} [**102**]{},56 8 . T his form of th e g en e rators i stak en f r o m [@l m].L om ont J Sand Mo s es H E 1962 S implere ali zat ionso f th e infi nitesima l gen e rators of thep roper orthoch r on o u si nhom oge neous Loren tz g r oupform as s z e ro [* J. Ma th . P h ys.*]{} [**3**]{},40 5 Bial ynick i-Birula I an d Bialynic k a - Birula Z 198 7 B e rry’s phase in therelativist i c theory of s pinningparticles , [*Phys.Rev . D *]{ } [ * * 35 **]{} 2383 Bi a l ynic ki -Birula Iand Bia lyn ick a-B iru la Z 2006 B eams ofel ec tr om agn eticr adiation c arr yi ngangul a r mome ntum: The R ie m ann -Silber s te i n vec to randthe c lassi cal- q uan tum cor responden ce[ *Opt .Co mm.*]{} [**264**]{}34 2 (ArXiv:qu ant -ph/05 0 1 1011)
- --
abstract: 'We presen t deep n arr owban d ob servation s w ith hi ghs patial resol ution o f e x t rapla n a rdif fu se ionized g asin th ehalo of NGC 891, obtained with the WFPC2 on-boa rdtheH S T. Ou r H $ \al ph a $ o b s ervations, cent ered on th en or thern part ofNG C891, r eveal a n ext e nded ga seous hal o, whichfi llsa l mos t the enti re field of viewo f our WF PC2 o bse rvatio n. Wh ereas NGC89 1 ha s bee n stud ie d exte nsive ly with gr ound-based telescopes,here t he sm all scale st ruc t ure of the e xten ded emissi onlin e gas is prese nted at hi g h spa tial resolutio n o f 0 1 , c orrespondin g t o 4 .6pcatt he dis tanc e to NGC891. Them ajority of the H$\ a l pha $ e m issi on is diffuse. S eve ra l discrete f eatures (e. g | ) Fock_V 1934_Zur Quantenelektrodynamik [*Phys. Zeit._der Sowjetunion*]{}_[**6**]{}_425 Akhiezer_A_I and Berestetski_V B 1965_[*Quantum Electrodynamics*]{} (New York:_Interscience) Ch. 1 Schweber_S_S 1961 [*An Introduction to Relativistic Quantum Field Theory*]{} (Evanston: Row, Peterson and Co)_Ch._9 Bialynicki-Birula_I_and_Bialynicka-Birula Z 1975 [*Quantum Electrodynamics*]{}_(Oxford: Pergamon) Ch. 9 Pauli_W 1980_[*General Principles of Quantum Mechanics*]{} (Berlin: Springer) Ch._25_Cohen-Tannoudji C, Dupont-Roc_J and Grynberg G 1989 [*Photons and Atoms: Introduction_to Quantum Electrodynamics*]{} (New York: Wiley)_Ch. 1 Foldy_L_L_1956 Synthesis of covariant_particle equations [*Phys. Rev.*]{} [**102**]{}, 568._This form of the generators is_taken from [@lm]. Lomont J S and_Moses H E 1962 Simple realizations_of the infinitesimal generators of_the proper_orthochronous inhomogeneous Lorentz group for_mass zero [*J._Math. Phys.*]{}_[**3**]{}, 405 Bialynicki-Birula_I and Bialynicka-Birula Z 1987 Berry’s_phase in the_relativistic theory of spinning particles, [*Phys._Rev._D*]{} [**35**]{} 2383_Bialynicki-Birula_I_and Bialynicka-Birula_Z 2006 Beams_of_electromagnetic radiation_carrying_angular momentum: The Riemann-Silberstein vector and_the_classical-quantum correspondence [*Opt. Comm.*]{} [**264**]{} 342 (ArXiv:_quant-ph/05011011)
---
abstract: 'We present_deep_narrowband observations with high_spatial resolution of extraplanar diffuse_ionized gas in the halo of_NGC891, obtained_with the_WFPC2 on-board the HST. Our H$\alpha$ observations, centered on the northern_part of NGC891, reveal an extended_gaseous halo, which fills_almost the_entire_field of view_of_our WFPC2_observation. Whereas NGC891 has been studied extensively_with ground-based_telescopes, here the small scale structure_of the extended emission_line_gas is presented at high spatial_resolution of 01, corresponding to 4.6pc_at the distance to NGC891._The_majority_of the H$\alpha$ emission is_diffuse. Several discrete features (e.g |
SE) for neutron star magnetospheres [@Glampedakis2014; @Fujisawa2014; @Pili2015; @Akgun2016; @Kojima2017; @Kojima2018; @Kojima2018b; @Akgun2018a]. @Akgun2017 performed magneto-thermal evolutions coupling the crustal magnetic field at the stellar surface with an exterior equilibrium solution. The results showed that large twists grow in the magnetosphere up to a critical point beyond which no stable equilibrium solutions where found. A more detailed analysis by @Akgun2018a showed that, for sufficiently large twists, the solutions of the GSE are degenerate with several possible configurations of different energies but matching boundary conditions at the surface. This suggests the possibility of an unstable branch of the solutions and, thus, a possible explanation for the occurrence of bursts and GFs. In this work we explore this possibility by performing three-dimensional (3D) numerical simulations of the equilibrium models in @Akgun2018a. We asses their stability properties and their potential as candidates for transient magnetar phenomenology.
This work is organised as follows. In section \[sec:physics\] we review and discuss the physics involved in magnetars relevant to the processes that we want to study. In section \[sec:ff\_equations\] we briefly review the equations of force-free electrodynamics (FFE) implemented for simulations conducted on the infrastructure of the `Einstein Toolkit` (supplemented by appendix \[sec:augmented\_system\]). A detailed description of the derivation of initial models according to @Akgun2018a is given in section \[sec:twisted\_models\]. In section \[sec:simulations\] we present the numerical setup of our simulations as well as the outcome of the conducted 3D force-free simulations of twisted magnetospheres (reviewing details on maintaining the force-free regime in appendix \[sec:ff\_breakdown\]). The observed rapid dissipation of electromagnetic energy through the magnetar crust is interpreted and related to observable quantities, such as luminosity estimates, shear stresses on the stellar crust, and opacity models, in section \[sec:discussion\]. Along this paper we use Gaussian units in CGS, except for section \[sec:ff\_equations\] in which we use Heaviside-Lorentz with geometrised units ($G=c=M_\odot=1$). For convenience we express current densities in A m$^{-2}$ | SE) for neutron star magnetospheres [ @Glampedakis2014; @Fujisawa2014; @Pili2015; @Akgun2016; @Kojima2017; @Kojima2018; @Kojima2018b; @Akgun2018a ]. @Akgun2017 performed magneto - thermal evolutions coupling the crustal charismatic sphere at the stellar surface with an exterior chemical equilibrium solution. The solution showed that large wind develop in the magnetosphere up to a critical point beyond which no static chemical equilibrium solutions where find. A more detailed analysis by @Akgun2018a testify that, for sufficiently large twists, the solutions of the GSE are degenerate with several possible configuration of different energies but matching boundary condition at the surface. This suggests the possibility of an unstable outgrowth of the solutions and, thus, a possible explanation for the occurrence of bursts and GFs. In this workplace we explore this possibility by performing three - dimensional (3D) numerical simulations of the equilibrium models in @Akgun2018a. We asses their stability properties and their potential as candidates for transient magnetar phenomenology.
This work is organised as follows. In section \[sec: physics\ ] we review and discuss the purgative involve in magnetars relevant to the processes that we desire to analyze. In incision \[sec: ff\_equations\ ] we briefly review the equations of force - free electrodynamics (FFE) implemented for simulations conducted on the infrastructure of the 'Einstein Toolkit '(append by appendix \[sec: augmented\_system\ ]). A detailed description of the derivation of initial models according to @Akgun2018a is given in section \[sec: twisted\_models\ ]. In section \[sec: simulations\ ] we award the numerical setup of our simulations as well as the outcome of the conducted 3D force - complimentary simulations of twisted magnetospheres (reviewing details on keep the force - free regime in appendix \[sec: ff\_breakdown\ ]). The ascertained rapid dissipation of electromagnetic department of energy through the magnetar crust is interpreted and related to observable quantities, such as luminosity estimates, shear stress on the stellar crust, and opacity model, in section \[sec: discussion\ ]. Along this composition we use Gaussian unit in CGS, except for section \[sec: ff\_equations\ ] in which we practice Heaviside - Lorentz with geometrised units ($ G = c = M_\odot=1 $). For convenience we press out current densities in A m$^{-2}$ | SE) vor neutron star magnetorpheres [@Glampedcjis2014; @Fuoisawa2014; @Lili2015; @Akgjn2016; @Kojima2017; @Kojima2018; @Kojima2018b; @Akgnn2018a]. @Qkgun2017 performed magneto-theroal evolunions couplinj the crustal majhetic flzld af the wtellar surfacg with an exderior equilibsijm solution. The results showed that lwrge twoshs grow in the magmqtoslhere up to a critical point beyons which no stable eqiilibrium solutions where vounf. A more detailed wnalysis by @Akgtb2018a showed thxt, for sufficiently latge twists, the solutions of the ESE axe degenerajz wihv several pissibje configurabpons of differrnt energies bmt mavchibg boundary conditionv at the surface. Tris suggevta the possibility od an onstatle ceanzh kf tge solktikns and, thhs, a possiboe explanation for uhe iccurrence of burstf wnd GFs. In this work we explore this povsigility by performing theee-dimensional (3D) numetical simujations of the equilibrium models in @Akgun2018a. We asves tiekr wtwcullty properties and their potential as candidaess fpr transient mcgnetar phenomenpllgu.
Jhis work is otganiseb aa follows. In sectiln \[sec:phisics\] qe review and discuss the physics involvwd in magnetcrs relevant to the pxocesses thac we wsnt tp study. In section \[sec:ff\_zquatikns\] we brievly revies the equations ow fprwe-free electrodynamics (FFE) implemenved fpr simuuatipns cogducted on the ltfrastructure of tje `Eiusteit Toolkit` (dupplemented by appendix \[sec:augmented\_system\]). A datapled descxiptiok of the derivaeion of initiak modelf accurding to @Zkgun2018a ms given in fection \[sec:twivjed\_models\]. In vection \[sqc:sinularions\] wd present the nimerical setup of oyr simulations as xell zs the outcome if uhw conducted 3D gorze-fwev smmulaehons of twisded oagvrtospferes (rtriecivg drtails on maintaininc ths force-free regime ik appendiz \[sec:ff\_bweakdown\]). The pbserved rapid disdipatmon of elecyroiagnetic energy through the mafnetar crkst is interpreeed qnd related co observable quantities, such as luminosmty estimates, shear sttesses on the stellar grust, and opecity iodels, in section \[sec:discussiob\]. Along this papev we use Gaussian unita in CCS, exfept for section \[sec:ff\_equations\] in which we use Heaviside-Lorentz with geomwtrisev tnits ($G=c=M_\odkt=1$). Fpr cotvzniznce we expcers current densinies in A m$^{-2}$ | SE) for neutron star magnetospheres [@Glampedakis2014; @Fujisawa2014; @Kojima2017; @Kojima2018b; @Akgun2018a]. performed magneto-thermal evolutions at stellar surface with exterior equilibrium solution. results showed that large twists grow the magnetosphere up to a critical point beyond which no stable equilibrium solutions found. A more detailed analysis by @Akgun2018a showed that, for sufficiently large twists, solutions the are with several possible configurations of different energies but matching boundary conditions at the surface. This suggests possibility of an unstable branch of the solutions thus, a possible explanation the occurrence of bursts and In work we this by three-dimensional (3D) numerical of the equilibrium models in @Akgun2018a. We asses their stability properties and their potential as candidates for magnetar phenomenology. is organised follows. section we review and physics involved in magnetars relevant to we want to study. In section \[sec:ff\_equations\] we review the of force-free electrodynamics (FFE) implemented for conducted on the infrastructure of the `Einstein Toolkit` by appendix \[sec:augmented\_system\]). A detailed description of the derivation of initial models according to @Akgun2018a in section \[sec:twisted\_models\]. In \[sec:simulations\] we present numerical of simulations well as outcome of the conducted 3D force-free simulations of twisted magnetospheres (reviewing on maintaining the force-free regime in appendix \[sec:ff\_breakdown\]). The observed of energy through the crust is interpreted and to quantities, such as luminosity stresses the opacity in \[sec:discussion\]. Along this paper use Gaussian units in CGS, for section \[sec:ff\_equations\] in geometrised units ($G=c=M_\odot=1$). For convenience we express current in A m$^{-2}$ | SE) for neutron star magnetospHeres [@GlampEdakiS2014; @FuJisAwA2014; @PilI2015; @AkgUn2016; @Kojima2017; @Kojima2018; @kOjimA2018b; @Akgun2018a]. @Akgun2017 performed MagneTo-THermAL eVolutIons couPLiNG The CrUsTal MaGNeTic fiEld At the stEllar surfaCe wItH an exterior eQUiLibrium solUtiOn. The results ShoWed thaT lArgE TwistS grOw in tHe magnETospheRe up to a crItICal poiNT beyond WHIcH no sTable equilibrium sOLuTIons where found. a more dEtAIlED AnaLysIs by @Akgun2018a ShOwed tHAt, for suFFiCIENtlY Large twists, thE solutions oF The gSE are DeGenERate wiTh sevErAL poSsible confiGuraTions of diFferenT EnergieS But matcHing boUndAry CondITiOnS at ThE SurFAcE. ThIS suGgests thE pOsSibilIty oF AN UNstaBle BranCh of tHe solutions anD, thUs, a pOSsiBle exPlanaTion FoR the oCcurreNce of BuRsts and GFs. In thiS worK we explorE thIs PosSiBilitY By perfOrmIng Three-diMensionAL (3D) nUmERICaL simulations of the eQuILIbRium modeLs in @AkGUn2018A. WE Asses theIr StaBiliTY PropeRtieS AnD their poTentiaL As CaNdidateS fOr tranSiEnt MagNetar PHenoMenoloGy.
This woRk is oRGanised as folloWS. In section \[sec:PHySICs\] WE revIew And discuss tHe phYSics InvoLVeD in MAgnetArs reLeVAnT To the processes that wE wAnt to sTudy. IN section \[sec:ff\_Equations\] wE BRIefly revIew tHE eQUations of force-Free eLectrodynaMIcs (FFE) imPlemeNted for sImulationS COnducted On tHe iNfrAstRUCtUre of the `EinstEIN TooLkIt` (supplEmeNted by aPpeNdiX \[seC:auGmEnted\_systEm\]). A detaiLeD dEsCrIptIon of THe derivaTiOn oF iNitIal moDEls accOrdinG to @AKgUn2018A Is gIven in sECtION \[sec:TwIsTed\_mOdeLs\]. in secTion \[SEc:sImulatiOns\] we presEnt THe nuMeRiCal setuP of our simulatIoNs as well as ThE ouTcome oF THe conducTed 3D force-free simulationS Of twistEd mAgnetOsphEres (revieWinG detaiLs oN MaintaIning tHe forCe-FreE REgime IN ApPenDiX \[sec:ff\_breaKDOwn\]). the obSeRved Rapid diSsipation of electroMAgnEtic energy thrOugH the MAGnEtaR CrUSt iS iNTerPREted and related tO observablE qUAnTities, such AS luMiNosity eStimateS, sheaR StresseS on the steLlar crust, AnD opaCITy mOdels, in secTion \[sec:dIscussion\]. aLong tHIs Paper We uSe GausSiAn uNits iN CGS, exCEpt For seCtion \[sEc:Ff\_equaTions\] In Which we uSe Heaviside-Lorentz with gEometrIsed uNitS ($G=c=M_\odot=1$). FOr cONveNience we eXpreSs current dEnsItiEs in A M$^{-2}$ | SE) for neutron star magne tospheres[@Gla mpe dak is 2014 ; @F ujisawa2014; @ P ili2 015; @Akgun2016; @Koji ma201 7; @Koj i ma 2018; @Kojim a 20 1 8 b;@A kg un2 01 8 a] . @Ak gun 2017 pe rformed ma gne to -thermal evo l ut ions coupl ing the crustal ma gnetic f iel d at t hestell ar sur f ace wi th an ext er i or equ i librium s ol utio n. The results sh o we d that large tw ists g ro w i n the ma gnetospher eup to a criti c al p o int beyond whichno stable e q uil ibrium s olu t ions w herefo u nd. A more det aile d analysi s by @ A kgun201 8 a showe d that , f orsuff i ci en tly l a rge tw ist s , t he solut io ns of t he G S E a re d ege nera te wi th several po ssi blec onf igura tions ofdi ffere nt ene rgies b ut matching bou ndar y conditi ons a t t he surf a ce. Th issug gests t he poss i bil it y o fan unstable branch o f th e soluti ons an d ,th u s, a pos si ble exp l a natio n fo r t he occur renceo fbu rsts an dGFs. I nthi s w ork w e exp lore t his poss ibili t y by performin g three-dimens i on a l ( 3 D) n ume rical simul atio n s of the eq uil i brium mode ls in @Akgun2018a. We ass es their stab ility propert ies and th e i r potenti al a s c a ndidates for t ransi ent magnet a r phenom enolo gy.
Thi s work is o rganised as fo llo ws. I nsection \[sec : p hysi cs \] we r evi ew anddis cus s t heph ysics inv olved in m ag ne ta rsrelev a nt to th epro ce sse s tha t we wa nt to stu dy .I n s ection\ [s e c :ff\ _e qu atio ns\ ]we br iefl y re view th e equatio nso f fo rc e- free el ectrodynamics ( FFE) imple me nte d fors i mulation s conducted on the infr a structu reof th e `E instein T ool kit` ( sup p lement ed byappen di x \ [ s ec:au g m en ted \_ system\]). A de taile ddesc ription of the derivation ofinitial model s a ccor d i ng to @A k gun 20 1 8ai s given in secti on \[sec:t wi s te d\_models\ ] . I nsection \[sec: simul a tions\] we prese nt the nu me rica l set up of oursimulati ons as we l l ast he outc ome of th econ ducte d 3D f o rce -free simul at ions o f twi st ed magne tospheres (reviewing de tailson ma int aining th e f o rce -free reg imein appendi x \ [se c:ff\ _br e akdow n\]) . T heo bserv ed r a pid dissi p at ion o felectromagn e t i c e nergy th r ough t he m agnetar crust isi nterpreted and rel a t edtoo bser va ble quantities , s uc h as lumin os ity estimat es, shea rs tress es onthe st ellar c r u st , and o paci tymodels, i n s ec t ion \[s ec :d i scussi on\] .Alongthis p a perw e use Gaussian un its i n CGS,e xce pt fo rsection \[se c:ff\_equa tions\] inwhichwe u se He aviside -L orentz wi th geometris e d units ( $G=c= M_\odot =1 $).For conve nien c e we e xpre ss cu rrent den s i ti e sin A m $^{- 2}$ | SE) for_neutron star_magnetospheres [@Glampedakis2014; @Fujisawa2014; @Pili2015;_@Akgun2016; @Kojima2017;_@Kojima2018;_@Kojima2018b; @Akgun2018a]._@Akgun2017_performed magneto-thermal evolutions_coupling the crustal_magnetic field at the_stellar surface with_an_exterior equilibrium solution. The results showed that large twists grow in the magnetosphere up_to_a critical_point_beyond_which no stable equilibrium solutions_where found. A more detailed_analysis by_@Akgun2018a showed that, for sufficiently large twists, the_solutions_of the GSE_are degenerate with several possible configurations of different energies_but matching boundary conditions at the_surface. This suggests_the_possibility_of an unstable branch_of the solutions and, thus, a_possible explanation for the occurrence of_bursts and GFs. In this work we_explore this possibility by performing three-dimensional_(3D) numerical simulations of the_equilibrium models_in @Akgun2018a. We asses their_stability properties and_their potential_as candidates for_transient magnetar phenomenology.
This work is organised_as follows. In_section \[sec:physics\] we review and discuss the_physics_involved in magnetars_relevant_to_the processes_that we want_to_study. In_section \[sec:ff\_equations\]_we briefly review the equations of_force-free_electrodynamics (FFE) implemented for simulations conducted on_the infrastructure of the_`Einstein_Toolkit` (supplemented by appendix_\[sec:augmented\_system\]). A detailed description of_the derivation of initial models according_to @Akgun2018a_is given_in section \[sec:twisted\_models\]. In section \[sec:simulations\] we present the numerical setup of our_simulations as well as the outcome_of the conducted 3D_force-free simulations_of_twisted magnetospheres (reviewing_details_on maintaining_the force-free regime in appendix \[sec:ff\_breakdown\]). The_observed rapid_dissipation of electromagnetic energy through the_magnetar crust is interpreted_and_related to observable quantities, such as_luminosity estimates, shear stresses on the_stellar crust, and opacity models,_in_section \[sec:discussion\]._Along this paper we use_Gaussian units in CGS, except for_section \[sec:ff\_equations\] in which_we use Heaviside-Lorentz with geometrised units ($G=c=M_\odot=1$)._For_convenience we express current densities in_A_m$^{-2}$ |
{\bigl(+{{\textstyle\frac{i}{8}}}c_4 \bigr)}{\bigl( {\{2,1,3,2,4,3\}} - {\{3,2,1,4,3,2\}} \bigr)}
{\nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord{}}
+ {\bigl( +{{\textstyle\frac{5i}{24}}}c_4 + 2ic_6 \bigr)} {\bigl( {\{1,3,2,4,3,5\}} - {\{3,2,1,5,4,3\}} \bigr)}
{\nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord{}}
+ {\bigl( +{{\textstyle\frac{i}{6}}} c_4 + ic_6 \bigr)} {\bigl( {\{1,2,4,3,5,4\}} + {\{2,1,3,2,4,5\}} - {\{2,1,5,4,3,2\}} - {\{4,3,2,1,5,4\}} \bigr)}
{\nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord{}}
+ {\bigl( -{{\textstyle\frac{i}{48}}}c_4 - {{\textstyle\frac{i}{2}}}c_6 \bigr)} {\bigl( {\{1,4,3,2,5,4,3\}} - {\{2,1,3,2,5,4,3\}} + {\{3,2,1,4,3,2,5\}} - {\{3,2,1,4,3,5,4\}} \bigr)}
{\nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord{}}
+ {\bigl( -{{\textstyle\frac{i}{48}}}c_4 - {{\textstyle\frac{i}{2}}}c_6 \bigr)} {\bigl( {\{1,3,2,4,3,5,4\}} + {\{2,1,3,2,4,3,5\}} - {\{3,2,1,5,4,3,2\}} - {\{4,3,2,1,5,4,3\}} \bigr)}
\end{array | { \bigl(+{{\textstyle\frac{i}{8}}}c_4 \bigr)}{\bigl ({ \{2,1,3,2,4,3\ } } - { \{3,2,1,4,3,2\ } } \bigr) }
{ \nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord { } }
+ { \bigl (+ { { \textstyle\frac{5i}{24}}}c_4 + 2ic_6 \bigr) } { \bigl ({ \{1,3,2,4,3,5\ } } - { \{3,2,1,5,4,3\ } } \bigr) }
{ \nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord { } }
+ { \bigl (+ { { \textstyle\frac{i}{6 } } } c_4 + ic_6 \bigr) } { \bigl ({ \{1,2,4,3,5,4\ } } + { \{2,1,3,2,4,5\ } } - { \{2,1,5,4,3,2\ } } - { \{4,3,2,1,5,4\ } } \bigr) }
{ \nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord { } }
+ { \bigl (-{{\textstyle\frac{i}{48}}}c_4 - { { \textstyle\frac{i}{2}}}c_6 \bigr) } { \bigl ({ \{1,4,3,2,5,4,3\ } } - { \{2,1,3,2,5,4,3\ } } + { \{3,2,1,4,3,2,5\ } } - { \{3,2,1,4,3,5,4\ } } \bigr) }
{ \nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord { } }
+ { \bigl (-{{\textstyle\frac{i}{48}}}c_4 - { { \textstyle\frac{i}{2}}}c_6 \bigr) } { \bigl ({ \{1,3,2,4,3,5,4\ } } + { \{2,1,3,2,4,3,5\ } } - { \{3,2,1,5,4,3,2\ } } - { \{4,3,2,1,5,4,3\ } } \bigr) }
\end{array | {\bihl(+{{\textstyle\frac{i}{8}}}c_4 \bigr)}{\binl( {\{2,1,3,2,4,3\}} - {\{3,2,1,4,3,2\}} \bigr)}
{\nonumywr\\[3pt]&\hs'ace{-4\arrzycolsep}&\oathord{}}
+ {\bigl( +{{\textstyle\frac{5i}{24}}}c_4 + 2uc_6 \bitr)} {\bigl( {\{1,3,2,4,3,5\}} - {\{3,2,1,5,4,3\}} \bigr)}
{\nonumbdr\\[3pt]&\hspacv{-4\arraycolwep}&\methord{}}
+ {\bigl( +{{\textstyle\frag{n}{6}}} c_4 + jg_6 \bigx)} {\uigl( {\{1,2,4,3,5,4\}} + {\{2,1,3,2,4,5\}} - {\{2,1,5,4,3,2\}} - {\{4,3,2,1,5,4\}} \bigt)}
{\nonumber\\[3pt]&\vspace{-4\arraycolveo}&\mcthord{}}
+ {\bigl( -{{\textstyle\frac{i}{48}}}c_4 - {{\textstyje\frac{i}{2}}}v_6 \higr)} {\bigl( {\{1,4,3,2,5,4,3\}} - {\{2,1,3,2,5,4,3\}} + {\{3,2,1,4,3,2,5\}} - {\{3,2,1,4,3,5,4\}} \bpgw)}
{\nohlmner\\[3pt]&\hspace{-4\arraycolsep}&\mathord{}}
+ {\bifl( -{{\textvtyle\frac{i}{48}}}c_4 - {{\yextstyle\frac{i}{2}}}c_6 \bigr)} {\bigl( {\{1,3,2,4,3,5,4\}} + {\{2,1,3,2,4,3,5\}} - {\{3,2,1,5,4,3,2\}} - {\{4,3,2,1,5,4,3\}} \bigr)}
\end{array | {\bigl(+{{\textstyle\frac{i}{8}}}c_4 \bigr)}{\bigl( {\{2,1,3,2,4,3\}} - {\{3,2,1,4,3,2\}} \bigr)} {\nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord{}} +{{\textstyle\frac{5i}{24}}}c_4 2ic_6 \bigr)} {\{1,3,2,4,3,5\}} - {\{3,2,1,5,4,3\}} c_4 ic_6 \bigr)} {\bigl( + {\{2,1,3,2,4,5\}} - - {\{4,3,2,1,5,4\}} \bigr)} {\nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord{}} + {\bigl( - {{\textstyle\frac{i}{2}}}c_6 \bigr)} {\bigl( {\{1,4,3,2,5,4,3\}} - {\{2,1,3,2,5,4,3\}} + {\{3,2,1,4,3,2,5\}} - {\{3,2,1,4,3,5,4\}} \bigr)} {\nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord{}} {\bigl( -{{\textstyle\frac{i}{48}}}c_4 - {{\textstyle\frac{i}{2}}}c_6 \bigr)} {\bigl( {\{1,3,2,4,3,5,4\}} + {\{2,1,3,2,4,3,5\}} - {\{3,2,1,5,4,3,2\}} - {\{4,3,2,1,5,4,3\}} \end{array | {\bigl(+{{\textstyle\frac{i}{8}}}c_4 \bigr)}{\biGl( {\{2,1,3,2,4,3\}} - {\{3,2,1,4,3,2\}} \bigr)}
{\nonuMber\\[3pT]&\hsPacE{-4\aRrayColsEp}&\mathord{}}
+ {\bigl( +{{\tEXtstYle\frac{5i}{24}}}c_4 + 2ic_6 \bigr)} {\bigl( {\{1,3,2,4,3,5\}} - {\{3,2,1,5,4,3\}} \bigR)}
{\nonuMbER\\[3pt]&\hSPaCe{-4\arrAycolseP}&\MaTHOrd{}}
+ {\BiGl( +{{\TexTsTYlE\frac{I}{6}}} c_4 + iC_6 \bigr)} {\biGl( {\{1,2,4,3,5,4\}} + {\{2,1,3,2,4,5\}} - {\{2,1,5,4,3,2\}} - {\{4,3,2,1,5,4\}} \bigr)}
{\nonuMbeR\\[3pT]&\hspace{-4\arrayCOlSep}&\mathord{}}
+ {\BigL( -{{\textstyle\frAc{i}{48}}}C_4 - {{\textsTyLe\fRAc{i}{2}}}c_6 \bIgr)} {\Bigl( {\{1,4,3,2,5,4,3\}} - {\{2,1,3,2,5,4,3\}} + {\{3,2,1,4,3,2,5\}} - {\{3,2,1,4,3,5,4\}} \bIgr)}
{\nonUMber\\[3pt]&\Hspace{-4\arrAyCOlsep}&\mAThord{}}
+ {\biGL( -{{\TeXtstYle\frac{i}{48}}}c_4 - {{\textstylE\FrAC{i}{2}}}c_6 \bigr)} {\bigl( {\{1,3,2,4,3,5,4\}} + {\{2,1,3,2,4,3,5\}} - {\{3,2,1,5,4,3,2\}} - {\{4,3,2,1,5,4,3\}} \bigR)}
\end{arRaY | {\bigl(+{{\textstyle\frac {i}{8}}}c_ 4 \bi gr) }{\ bi gl({\{2 ,1,3,2,4,3\}}- {\{ 3,2,1,4,3,2\}} \bigr)}
{\n on u mber \ \[ 3pt]& \hspace { -4 \ a rra yc ol sep }& \ ma thord {}}
+ {\bi gl( +{{\te xts ty le\frac{5i}{ 2 4} }}c_4 + 2i c_6 \bigr)} {\b igl ( {\{1 ,3 ,2, 4 ,3,5\ }}- {\{ 3,2,1, 5 ,4,3\} } \bigr)} { \nonum b er\\[3p t ] &\ hspa ce{-4\arraycolsep } &\ m athord{}}
+ {\ bigl(+{ { \t e x tst yle \frac{i}{6 }} } c_4 + ic_6\ bi g r ) } { \ bigl( {\{1,2, 4,3,5,4\}}+ {\ {2,1,3 ,2 ,4, 5 \}} -{\{2, 1, 5 ,4, 3,2\}} - {\ {4,3 ,2,1,5,4\ }} \bi g r)}
{\ n onumber \\[3pt ]&\ hsp ace{ - 4\ ar ray co l sep } &\ mat h ord {}}
+ {\ bi gl ( -{{ \tex t s t y le\f rac {i}{ 48}}} c_4 - {{\text sty le\f r ac{ i}{2} }}c_6 \bi gr )} {\ bigl({\{1, 4, 3,2,5,4,3\}} -{\{2 ,1,3,2,5, 4,3 \} } + { \{3,2 , 1,4,3, 2,5 \}} - {\{3 ,2,1,4, 3 ,5, 4\ } } \b igr)}
{\nonumber\ \[ 3 p t] &\hspace {-4\ar r ay co l sep}&\ma th ord {}}+ {\big l( - { {\ textstyl e\frac { i} {4 8}}}c_4 - {{\te xt sty le\ frac{ i }{2} }}c_6\bigr)}{\big l ( {\{1,3,2,4,3 , 5,4\}} + {\{2 , 1, 3 , 2, 4 ,3,5 \}} - {\{3,2,1 ,5,4 , 3,2\ }} - {\ {4, 3 ,2,1, 5,4,3 \} } \ b igr)}
\end{array | {\bigl(+{{\textstyle\frac{i}{8}}}c_4_\bigr)}{\bigl( {\{2,1,3,2,4,3\}}_- {\{3,2,1,4,3,2\}} \bigr)}
{\nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord{}}
+_{\bigl( +{{\textstyle\frac{5i}{24}}}c_4_+_2ic_6 \bigr)}_{\bigl(_{\{1,3,2,4,3,5\}} - {\{3,2,1,5,4,3\}}_\bigr)}
{\nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord{}}
+ {\bigl(_+{{\textstyle\frac{i}{6}}} c_4 + ic_6_\bigr)} {\bigl( {\{1,2,4,3,5,4\}}_+_{\{2,1,3,2,4,5\}} - {\{2,1,5,4,3,2\}} - {\{4,3,2,1,5,4\}} \bigr)}
{\nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord{}}
+ {\bigl( -{{\textstyle\frac{i}{48}}}c_4 - {{\textstyle\frac{i}{2}}}c_6 \bigr)} {\bigl( {\{1,4,3,2,5,4,3\}}_-_{\{2,1,3,2,5,4,3\}} +_{\{3,2,1,4,3,2,5\}}_-_{\{3,2,1,4,3,5,4\}} \bigr)}
{\nonumber\\[3pt]&\hspace{-4\arraycolsep}&\mathord{}}
+ {\bigl( -{{\textstyle\frac{i}{48}}}c_4_- {{\textstyle\frac{i}{2}}}c_6 \bigr)} {\bigl( {\{1,3,2,4,3,5,4\}}_+ {\{2,1,3,2,4,3,5\}}_- {\{3,2,1,5,4,3,2\}} - {\{4,3,2,1,5,4,3\}} \bigr)}
\end{array |
} \left( \tfrac{1}{2} d^2_{\mathcal{Z}}(z) \right)
& = (z - P_{\mathcal{Z}}(z))^T \left( f(z, P_{\mathcal{Z}}(z)) - \tfrac{1}{K} G^{-1}(P_{\mathcal{Z}}(z))(z - P_{\mathcal{Z}}(z) \right) \\
& \leq d_{\mathcal{Z}}(z) \| f(z, P_{\mathcal{Z}}(z))\| - \tfrac{1}{K} (z - P_{\mathcal{Z}}(z))^T G^{-1}(P_{\mathcal{Z}}(z)) \left(z - P_{\mathcal{Z}}(z) \right) \\
& \leq d_{\mathcal{Z}}(z) \| f (z, P_{\mathcal{Z}}(z)) \| - \tfrac{\mu}{K} d^2_{\mathcal{Z}}(z)
= d_{\mathcal{Z}}(z) ( M - \tfrac{\mu}{K} d_{\mathcal{Z}}(z)) \,.
\end{aligned}$$ It follows that ${\mathcal{L}}_{F_K} \left( \tfrac{1}{2} d^2_{\mathcal{Z}}(z ) \right) < 0$ whenever $d_{\mathcal{Z}}(z) > \frac{K M}{\mu}$. Since $K < \frac{\mu}{2\alpha M}$ and using an invariance argument, it follows that $z(t)
\in {\mathcal{Z}}+ \tfrac{K M}{\mu} {\mathbb{B}}\subset {\mathcal{Z}}^\circ_\alpha$ for all $t \in [0, T']$.
In other words, for small enough $K$, any solution of starting at $z_0$ remains within a neighborhood of ${\mathcal{Z}}$ on which the projection $P_{\mathcal{Z}}$ is single-valued.
Since $z(T')$ lies on the boundary of $(z_0 + \epsilon {\mathbb{B}}) \cap {\mathcal{Z}}^\circ_\alpha$, but at the same time $z(T') \in {\mathcal{Z}}+ \frac{KM}{\mu} {\mathbb{B}}$, it follows that $\| z(T') - z_0 \| = \epsilon$. In other | } \left (\tfrac{1}{2 } d^2_{\mathcal{Z}}(z) \right)
& = (z - P_{\mathcal{Z}}(z))^T \left (f(z, P_{\mathcal{Z}}(z) ) - \tfrac{1}{K } G^{-1}(P_{\mathcal{Z}}(z))(z - P_{\mathcal{Z}}(z) \right) \\
& \leq d_{\mathcal{Z}}(z) \| f(z, P_{\mathcal{Z}}(z))\| - \tfrac{1}{K } (z - P_{\mathcal{Z}}(z))^T G^{-1}(P_{\mathcal{Z}}(z) ) \left(z - P_{\mathcal{Z}}(z) \right) \\
& \leq d_{\mathcal{Z}}(z) \| f (z, P_{\mathcal{Z}}(z) ) \| - \tfrac{\mu}{K } d^2_{\mathcal{Z}}(z)
= d_{\mathcal{Z}}(z) (M - \tfrac{\mu}{K } d_{\mathcal{Z}}(z) ) \, .
\end{aligned}$$ It follows that $ { \mathcal{L}}_{F_K } \left (\tfrac{1}{2 } d^2_{\mathcal{Z}}(z) \right) < 0 $ whenever $ d_{\mathcal{Z}}(z) > \frac{K M}{\mu}$. Since $ K < \frac{\mu}{2\alpha M}$ and using an invariability controversy, it follows that $ z(t)
\in { \mathcal{Z}}+ \tfrac{K M}{\mu } { \mathbb{B}}\subset { \mathcal{Z}}^\circ_\alpha$ for all $ t \in [ 0, T']$.
In early words, for little enough $ K$, any solution of starting at $ z_0 $ persist within a region of $ { \mathcal{Z}}$ on which the projection $ P_{\mathcal{Z}}$ is single - value.
Since $ z(T')$ lies on the boundary of $ (z_0 + \epsilon { \mathbb{B } }) \cap { \mathcal{Z}}^\circ_\alpha$, but at the like time $ z(T') \in { \mathcal{Z}}+ \frac{KM}{\mu } { \mathbb{B}}$, it follows that $ \| z(T') - z_0 \| = \epsilon$. In other | } \levt( \tfrac{1}{2} d^2_{\mathcal{Z}}(z) \righu)
& = (z - P_{\mcrhcal{Z}}(v))^T \left( f(z, P_{\matfcal{Z}}(z)) - \tfrac{1}{K} G^{-1}(P_{\mathcal{Z}}(z))(z - P_{\nathcql{Z}}(z) \right) \\
& \oeq v_{\mathcal{Z}}(z) \| f(z, P_{\mathcal{Z}}(d))\| - \tfdwc{1}{K} (v - P_{\mathcal{Z}}(z))^T G^{-1}(K_{\mathcal{Z}}(z)) \laft(z - P_{\mathcal{S}}(z) \rnght) \\
& \leq d_{\mathcal{Z}}(z) \| f (z, P_{\iathcal{A}}(z)) \| - \tfrac{\mu}{K} d^2_{\mwthcsj{Z}}(z)
= d_{\mathcal{Z}}(z) ( M - \tfrac{\mu}{K} d_{\mzthcal{Z}}(e)) \,.
\end{aligned}$$ Iy follows that ${\mathcal{L}}_{F_K} \peft( \tfrac{1}{2} d^2_{\mathcal{Z}}(z ) \right) < 0$ whghevqe $d_{\mathcal{Z}}(z) > \frac{K M}{\ml}$. Since $K < \rrac{\mu}{2\alpha M}$ and using an invafiancz argument, ut fopnows that $z(v)
\in {\mathcal{Z}}+ \tfrac{N M}{\mu} {\msthbb{B}}\subset {\msthral{Z}}^\xirc_\alpha$ for all $t \ii [0, T']$.
In other words, fjr small ankugh $K$, any solutiin of sjartitg ag $z_0$ femziis sithin a ieighborhoos of ${\mathcao{Z}}$ on which the prokesnoon $P_{\mathcal{A}}$ is syndle-valued.
Since $z(T')$ lies on the boundary mf $(a_0 + \epsilon {\mathbb{B}}) \cap {\nathcal{Z}}^\circ_\alpha$, but at the swme time $z(T') \in {\mathcal{Z}}+ \frac{KM}{\mu} {\mathbb{B}}$, it follofs thet $\| z(U') - z_0 \| = \dpsllon$. In other | } \left( \tfrac{1}{2} d^2_{\mathcal{Z}}(z) \right) & = P_{\mathcal{Z}}(z))^T f(z, P_{\mathcal{Z}}(z)) \tfrac{1}{K} G^{-1}(P_{\mathcal{Z}}(z))(z - d_{\mathcal{Z}}(z) f(z, P_{\mathcal{Z}}(z))\| - (z - P_{\mathcal{Z}}(z))^T \left(z - P_{\mathcal{Z}}(z) \right) \\ & d_{\mathcal{Z}}(z) \| f (z, P_{\mathcal{Z}}(z)) \| - \tfrac{\mu}{K} d^2_{\mathcal{Z}}(z) = d_{\mathcal{Z}}(z) ( M \tfrac{\mu}{K} d_{\mathcal{Z}}(z)) \,. \end{aligned}$$ It follows that ${\mathcal{L}}_{F_K} \left( \tfrac{1}{2} d^2_{\mathcal{Z}}(z ) \right) 0$ $d_{\mathcal{Z}}(z) \frac{K Since $K < \frac{\mu}{2\alpha M}$ and using an invariance argument, it follows that $z(t) \in {\mathcal{Z}}+ M}{\mu} {\mathbb{B}}\subset {\mathcal{Z}}^\circ_\alpha$ for all $t \in [0, In other words, for enough $K$, any solution of at remains within neighborhood ${\mathcal{Z}}$ which the projection is single-valued. Since $z(T')$ lies on the boundary of $(z_0 + \epsilon {\mathbb{B}}) \cap {\mathcal{Z}}^\circ_\alpha$, but at same time {\mathcal{Z}}+ \frac{KM}{\mu} it that z(T') - z_0 \epsilon$. In other | } \left( \tfrac{1}{2} d^2_{\mathcal{Z}}(z) \right)
& = (z - p_{\mathcal{Z}}(z))^t \left( F(z, P_{\MatHcAl{Z}}(z)) - \TfraC{1}{K} G^{-1}(P_{\mathcal{Z}}(z))(z - p_{\MathCal{Z}}(z) \right) \\
& \leq d_{\mathcal{Z}}(Z) \| f(z, P_{\mAtHCal{Z}}(Z))\| - \TfRac{1}{K} (z - p_{\mathcaL{z}}(z))^t g^{-1}(p_{\maThCaL{Z}}(z)) \LeFT(z - p_{\mathCal{z}}(z) \right) \\
& \Leq d_{\mathcaL{Z}}(z) \| F (z, p_{\mathcal{Z}}(z)) \| - \tfRAc{\Mu}{K} d^2_{\mathcaL{Z}}(z)
= D_{\mathcal{Z}}(z) ( M - \tFraC{\mu}{K} d_{\mAtHcaL{z}}(z)) \,.
\end{AliGned}$$ IT folloWS that ${\mAthcal{L}}_{F_K} \LeFT( \tfrac{1}{2} D^2_{\Mathcal{z}}(Z ) \RiGht) < 0$ wHenever $d_{\mathcal{Z}}(z) > \FRaC{k M}{\mu}$. Since $K < \frac{\Mu}{2\alphA M}$ ANd USIng An iNvariance aRgUment, IT followS ThAT $Z(T)
\in {\MAthcal{Z}}+ \tfrac{K m}{\mu} {\mathbb{B}}\sUBseT {\mathcAl{z}}^\ciRC_\alpha$ For alL $t \IN [0, T']$.
IN other words, For sMall enougH $K$, any sOLution oF StartinG at $z_0$ reMaiNs wIthiN A nEiGhbOrHOod OF ${\mAthCAl{Z}}$ On which tHe PrOjectIon $P_{\MATHCal{Z}}$ Is sInglE-valuEd.
Since $z(T')$ lies On tHe boUNdaRy of $(z_0 + \EpsilOn {\maThBb{B}}) \caP {\mathcAl{Z}}^\ciRc_\Alpha$, but at the saMe tiMe $z(T') \in {\matHcaL{Z}}+ \FraC{Km}{\mu} {\maTHbb{B}}$, it FolLowS that $\| z(T') - Z_0 \| = \epsiloN$. in oThER | } \left( \tfrac{1}{2} d^2_ {\mathcal{ Z}}(z ) \ rig ht )
& = (z - P_ { \mat hcal{Z}}(z))^T \left(f(z,P_ { \mat h ca l{Z}} (z)) -\ tf r a c{1 }{ K} G^ {- 1 }( P_{\m ath cal{Z}} (z))(z - P _{\ ma thcal{Z}}(z) \r ight) \ \
& \ l eq d_ {\m athca l{Z}}( z ) \| f (z, P_{\m at h cal{Z} } (z))\|- \t frac {1}{K} (z - P_{\m a th c al{Z}}(z))^T G ^{-1}( P_ { \m a t hca l{Z }}(z)) \le ft (z -P _{\math c al { Z } }(z ) \right) \\
& \ l eqd_{\ma th cal { Z}}(z) \| f ( z , P _{\mathcal{ Z}}( z)) \| -\tfrac { \mu}{K} d^2_{\m athcal {Z} }(z )
=d_ { \ma t hc al{ Z }}( z) ( M - \ tf rac{\ mu}{ K } d _{\m ath cal{ Z}}(z )) \,.
\e nd{ alig n ed} $$ It foll owsth at ${ \mathc al{L} }_ {F_K} \left( \t frac {1}{2} d^ 2_{ \m ath ca l{Z}} ( z ) \r igh t)< 0$ wh enever$ d_{ \m a t h ca l{Z}}(z) > \frac{K M } { \m u}$. Sin ce $K< \ fr a c{\mu}{2 \a lph a M} $ and u sing an invaria nce ar g um en t, it f ol lows t ha t $ z(t )
\in {\ mathcal{ Z}}+\ tfrac{K M}{\mu } {\mathbb{B}} \ su b s et {\ma thc al{Z}}^\cir c_\a l pha$ for al l $ t \in[0, T '] $ .In other words, for s mall e nough $K$, any sol ution of s t a rting at $z_ 0 $r emains withina nei ghborhoodo f ${\mat hcal{ Z}}$ onwhich the p rojectio n $ P_{ \ma thc a l {Z }}$ is single - v alue d.
Since $z (T')$ l ies on th e b ou ndary of$(z_0 +\e ps il on {\ mathb b {B}}) \c ap {\ ma thc al{Z} } ^\circ _\alp ha$, b ut atthe sam e t i m e $z (T ') \in {\ ma thcal {Z}} + \f rac{KM} {\mu} {\m ath b b{B} }$ ,it foll ows that $\|z( T') - z_0\| =\epsil o n $. In ot her | } \left(_\tfrac{1}{2} d^2_{\mathcal{Z}}(z)_\right)
_ __ _&_= (z -_P_{\mathcal{Z}}(z))^T \left( f(z,_P_{\mathcal{Z}}(z)) - \tfrac{1}{K} G^{-1}(P_{\mathcal{Z}}(z))(z_- P_{\mathcal{Z}}(z) \right)__ __ _\\
__ _ & \leq d_{\mathcal{Z}}(z)_\| f(z,_P_{\mathcal{Z}}(z))\| - \tfrac{1}{K} (z - P_{\mathcal{Z}}(z))^T G^{-1}(P_{\mathcal{Z}}(z)) \left(z_-_P_{\mathcal{Z}}(z) \right) \\
_ &_\leq d_{\mathcal{Z}}(z) \| f (z, P_{\mathcal{Z}}(z))_\| - \tfrac{\mu}{K}_d^2_{\mathcal{Z}}(z)
__ _ = d_{\mathcal{Z}}(z) ( M_- \tfrac{\mu}{K} d_{\mathcal{Z}}(z)) \,.
_ \end{aligned}$$ It follows that ${\mathcal{L}}_{F_K} \left(_\tfrac{1}{2} d^2_{\mathcal{Z}}(z ) \right) < 0$_whenever $d_{\mathcal{Z}}(z) > \frac{K M}{\mu}$._Since $K_< \frac{\mu}{2\alpha M}$ and using_an invariance argument,_it follows_that $z(t)
_ _\in {\mathcal{Z}}+ \tfrac{K_M}{\mu} {\mathbb{B}}\subset {\mathcal{Z}}^\circ_\alpha$ for all $t_\in_[0, T']$.
In other_words,_for_small enough_$K$, any solution_of _starting at_$z_0$_remains within a neighborhood of ${\mathcal{Z}}$_on_which the projection $P_{\mathcal{Z}}$ is single-valued.
Since $z(T')$_lies on the boundary_of_$(z_0 + \epsilon {\mathbb{B}})_\cap {\mathcal{Z}}^\circ_\alpha$, but at the_same time $z(T') \in {\mathcal{Z}}+ \frac{KM}{\mu}_{\mathbb{B}}$, it_follows that_$\| z(T') - z_0 \| = \epsilon$. In other |
\biggl[\bigcup_{j=1}^{2^{l-1}}\Bigl(z_j+(e_l(T)+\varepsilon) B_Y\Bigr)\biggr]\\
&=\bigcup_{i,j}\Bigl(y_i+z_j+(e_k(S)+\varepsilon) B_Y+(e_l(T)+\varepsilon) B_Y\Bigr)\\
&\subset\bigcup_{i,j}\Bigl(y_i+z_j+[(e_k(S)+\varepsilon)^p+(e_l(T)+\varepsilon)^p]^{1/p} B_Y\Bigr)\end{aligned}$$ and taking the infimum over $\varepsilon>0$ gives the result.
The proof of submultiplicativity follows in a similar way. Indeed, if $k,l$ are positive integers and $\varepsilon>0$, we find $y_1,\dots,y_{2^{l-1}}\in Y$ and $z_1,\dots,z_{2^{k-1}}\in Z$ with $$T(B_X)\subset \bigcup_{i=1}^{2^{l-1}}\Bigl(y_i+(e_l(T)+\varepsilon) B_Y\Bigr)\quad\text{and}\quad R(B_Y)\subset \bigcup_{j=1}^{2^{k-1}}\Bigl(z_j+(e_k(R)+\varepsilon) B_Z\Bigr).$$ The proof then follows from $$\begin{aligned}
(R\circ T)(B_X)&=R(T(B_X))\subset R\biggl(\,\bigcup_{i=1}^{2^{l-1}}\Bigl(y_i+(e_l(T)+\varepsilon) B_Y\Bigr)\biggr)\\
&=\bigcup_{i=1}^{2^{l-1}}\Bigl(Ry_i+(e_l(T)+\varepsilon) R(B_Y)\Bigr)\\
&\subset \bigcup_{i,j}\Bigl(Ry_i+(e_l(T)+\varepsilon)z_j+(e_l(T)+\varepsilon)(e_k(R)+\varepsilon) B_Z\Bigr).\qedhere\end{aligned}$$
We will also need few basic facts about the Gamma function, which is defined by $\Gamma(t)=\int_0^\infty x^{t-1}e | \biggl[\bigcup_{j=1}^{2^{l-1}}\Bigl(z_j+(e_l(T)+\varepsilon) B_Y\Bigr)\biggr]\\
& = \bigcup_{i, j}\Bigl(y_i+z_j+(e_k(S)+\varepsilon) B_Y+(e_l(T)+\varepsilon) B_Y\Bigr)\\
& \subset\bigcup_{i, j}\Bigl(y_i+z_j+[(e_k(S)+\varepsilon)^p+(e_l(T)+\varepsilon)^p]^{1 / p } B_Y\Bigr)\end{aligned}$$ and taking the infimum over $ \varepsilon>0 $ gives the result.
The proof of submultiplicativity follow in a exchangeable way. Indeed, if $ k, l$ are positive integers and $ \varepsilon>0 $, we determine $ y_1,\dots, y_{2^{l-1}}\in Y$ and $ z_1,\dots, z_{2^{k-1}}\in Z$ with $ $ T(B_X)\subset \bigcup_{i=1}^{2^{l-1}}\Bigl(y_i+(e_l(T)+\varepsilon) B_Y\Bigr)\quad\text{and}\quad R(B_Y)\subset \bigcup_{j=1}^{2^{k-1}}\Bigl(z_j+(e_k(R)+\varepsilon) B_Z\Bigr).$$ The proof then follows from $ $ \begin{aligned }
(R\circ T)(B_X)&=R(T(B_X))\subset R\biggl(\,\bigcup_{i=1}^{2^{l-1}}\Bigl(y_i+(e_l(T)+\varepsilon) B_Y\Bigr)\biggr)\\
& = \bigcup_{i=1}^{2^{l-1}}\Bigl(Ry_i+(e_l(T)+\varepsilon) R(B_Y)\Bigr)\\
& \subset \bigcup_{i, j}\Bigl(Ry_i+(e_l(T)+\varepsilon)z_j+(e_l(T)+\varepsilon)(e_k(R)+\varepsilon) B_Z\Bigr).\qedhere\end{aligned}$$
We will besides need few basic facts about the Gamma function, which is specify by $ \Gamma(t)=\int_0^\infty x^{t-1}e | \bihgl[\bigcup_{j=1}^{2^{l-1}}\Bigl(z_j+(e_l(T)+\vareksilon) B_Y\Bigr)\biggt]\\
&=\bugcup_{i,o}\Bigl(y_i+a_j+(e_k(S)+\vardpsilon) B_Y+(e_l(T)+\varepsilon) B_Y\Bijr)\\
&\suvset\bugcup_{i,j}\Bigl(y_i+z_j+[(e_k(S)+\vareosilon)^p+(e_l(N)+\varepsilin)^p]^{1/p} V_Y\Bigr)\end{aligned}$$ akb takjkg thz mnfimum over $\vatepsilon>0$ givas the result.
Tve pxoof of submultiplicativity follows yn a sikipar way. Indeed, if $l,j$ ars positive integers and $\varepsilon>0$, we finv $y_1,\dots,y_{2^{l-1}}\in Y$ amd $z_1,\dots,z_{2^{k-1}}\in Z$ with $$T(B_X)\subdet \higcup_{i=1}^{2^{l-1}}\Bigl(y_i+(e_l(T)+\vwrepsilon) B_I\Gigw)\wuad\text{and}\qjad R(B_Y)\subset \bigcup_{j=1}^{2^{k-1}}\Gigl(z_j+(e_k(R)+\varepsilon) B_Z\Bigr).$$ The oroof then folliww fgmm $$\begin{alijned}
(R\cprc T)(B_X)&=R(T(B_X))\sunxet R\bhggl(\,\bigvup_{i=1}^{2^{l-1}}\Bigl(y_i+(e_l(T)+\yarepvilin) B_Y\Bigr)\biggr)\\
&=\bigcup_{i=1}^{2^{n-1}}\Bigl(Ry_i+(e_l(T)+\varepsijon) R(B_Y)\Bicr)\\
&\aubset \bigcup_{i,j}\Bitl(Ey_i+(e_l(J)+\vare[silub)z_j+(d_l(T)+\bacepailon)(e_n(R)+\verepsilon) B_A\Bigr).\qedherw\end{aligned}$$
We will slfi need few baaic fastf about the Gamma function, which is defpned by $\Gamma(t)=\int_0^\infty x^{t-1}e | \biggl[\bigcup_{j=1}^{2^{l-1}}\Bigl(z_j+(e_l(T)+\varepsilon) B_Y\Bigr)\biggr]\\ &=\bigcup_{i,j}\Bigl(y_i+z_j+(e_k(S)+\varepsilon) B_Y+(e_l(T)+\varepsilon) B_Y\Bigr)\\ &\subset\bigcup_{i,j}\Bigl(y_i+z_j+[(e_k(S)+\varepsilon)^p+(e_l(T)+\varepsilon)^p]^{1/p} B_Y\Bigr)\end{aligned}$$ the over $\varepsilon>0$ the result. The a way. Indeed, if are positive integers $\varepsilon>0$, we find $y_1,\dots,y_{2^{l-1}}\in Y$ and Z$ with $$T(B_X)\subset \bigcup_{i=1}^{2^{l-1}}\Bigl(y_i+(e_l(T)+\varepsilon) B_Y\Bigr)\quad\text{and}\quad R(B_Y)\subset \bigcup_{j=1}^{2^{k-1}}\Bigl(z_j+(e_k(R)+\varepsilon) B_Z\Bigr).$$ The proof then follows from (R\circ T)(B_X)&=R(T(B_X))\subset R\biggl(\,\bigcup_{i=1}^{2^{l-1}}\Bigl(y_i+(e_l(T)+\varepsilon) B_Y\Bigr)\biggr)\\ &=\bigcup_{i=1}^{2^{l-1}}\Bigl(Ry_i+(e_l(T)+\varepsilon) R(B_Y)\Bigr)\\ &\subset \bigcup_{i,j}\Bigl(Ry_i+(e_l(T)+\varepsilon)z_j+(e_l(T)+\varepsilon)(e_k(R)+\varepsilon) B_Z\Bigr).\qedhere\end{aligned}$$ We will also need basic about Gamma which is defined by $\Gamma(t)=\int_0^\infty x^{t-1}e | \biggl[\bigcup_{j=1}^{2^{l-1}}\Bigl(z_j+(e_l(T)+\varEpsilon) B_Y\BIgr)\biGgr]\\
&=\BigCuP_{i,j}\BIgl(y_I+z_j+(e_k(S)+\varepsilON) B_Y+(e_L(T)+\varepsilon) B_Y\Bigr)\\
&\subsEt\bigCuP_{I,j}\BiGL(y_I+z_j+[(e_k(s)+\varepsILoN)^P+(E_l(T)+\VaRePsiLoN)^P]^{1/p} b_Y\BigR)\enD{aligneD}$$ and taking The InFimum over $\varEPsIlon>0$ gives tHe rEsult.
The prooF of SubmulTiPliCAtiviTy fOllowS in a siMIlar waY. Indeed, if $K,l$ ARe posiTIve inteGERs And $\vArepsilon>0$, we find $y_1,\dOTs,Y_{2^{L-1}}\in Y$ and $z_1,\dots,z_{2^{k-1}}\In Z$ witH $$T(b_x)\sUBSet \BigCup_{i=1}^{2^{l-1}}\Bigl(y_I+(e_L(T)+\varEPsilon) B_y\biGR)\QUad\TExt{and}\quad R(B_Y)\Subset \bigcuP_{J=1}^{2^{k-1}}\BIgl(z_j+(e_K(R)+\VarEPsilon) b_Z\BigR).$$ THE prOof then follOws fRom $$\begin{aLigned}
(r\Circ T)(B_X)&=r(t(B_X))\subsEt R\bigGl(\,\bIgcUp_{i=1}^{2^{l-1}}\bIgL(y_I+(e_l(t)+\vARepSIlOn) B_y\bigR)\biggr)\\
&=\biGcUp_{I=1}^{2^{l-1}}\BigL(Ry_i+(E_L(t)+\VArepSilOn) R(B_y)\Bigr)\\
&\Subset \bigcup_{i,J}\BiGl(Ry_I+(E_l(T)+\VarepSilon)Z_j+(e_l(t)+\vArepsIlon)(e_k(r)+\varePsIlon) B_Z\Bigr).\qedheRe\enD{aligned}$$
WE wiLl AlsO nEed feW Basic fActS abOut the GAmma funCTioN, wHICH iS defined by $\Gamma(t)=\inT_0^\iNFTy X^{t-1}e | \biggl[\bigcup_{j=1}^{2^{ l-1}}\Bigl (z_j+ (e_ l(T )+ \var epsi lon) B_Y\Bigr) \ bigg r]\\
&=\bigcup_{i,j}\B igl(y _i + z_j+ ( e_ k(S)+ \vareps i lo n ) B_ Y+ (e _l( T) + \v areps ilo n) B_Y\ Bigr)\\
&\ sub se t\bigcup_{i, j }\ Bigl(y_i+z _j+ [(e_k(S)+\va rep silon) ^p +(e _ l(T)+ \va repsi lon)^p ] ^{1/p} B_Y\Bigr )\ e nd{ali g ned}$$a n dtaki ng the infimum ov e r$ \varepsilon>0$ gives t h er e sul t.
The proof o f sub m ultipli c at i v i tyf ollows in a s imilar way. Ind eed, i f$k, l $ areposit iv e in tegers and$\va repsilon> 0$, we find $y _ 1,\dots ,y_{2^ {l- 1}} \inY $an d $ z_ 1 ,\d o ts ,z_ { 2^{ k-1}}\in Z $with$$T( B _ X ) \sub set \bi gcup_ {i=1}^{2^{l-1 }}\ Bigl ( y_i +(e_l (T)+\ vare ps ilon) B_Y\B igr)\ qu ad\text{and}\qu ad R (B_Y)\sub set \ big cu p_{j= 1 }^{2^{ k-1 }}\ Bigl(z_ j+(e_k( R )+\ va r e p si lon) B_Z\Bigr).$$Th e pr oof then follo w sfr o m $$\beg in {al igne d }
(R\c ircT )( B_X)&=R( T(B_X) ) \s ub set R\b ig gl(\,\ bi gcu p_{ i=1}^ { 2^{l -1}}\B igl(y_i+ (e_l( T )+\varepsilon) B_Y\Bigr)\big g r) \ \
& = \big cup _{i=1}^{2^{ l-1} } \Big l(Ry _ i+ (e_ l (T)+\ varep si l on ) R(B_Y)\Bigr)\\
&\s ub set \b igcup _{i,j}\Bigl(R y_i+(e_l(T ) + \ varepsil on)z _ j+ ( e_l(T)+\vareps ilon) (e_k(R)+\v a repsilon ) B_Z \Bigr).\ qedhere\e n d {aligned }$$
W e w ill a ls o need few ba s i c fa ct s about th e Gamma fu nct ion , w hi ch is def ined by$\ Ga mm a( t)= \int_ 0 ^\inftyx^ {t- 1} e | \biggl[\bigcup_{j=1}^{2^{l-1}}\Bigl(z_j+(e_l(T)+\varepsilon)_B_Y\Bigr)\biggr]\\
&=\bigcup_{i,j}\Bigl(y_i+z_j+(e_k(S)+\varepsilon) B_Y+(e_l(T)+\varepsilon)_B_Y\Bigr)\\
&\subset\bigcup_{i,j}\Bigl(y_i+z_j+[(e_k(S)+\varepsilon)^p+(e_l(T)+\varepsilon)^p]^{1/p} B_Y\Bigr)\end{aligned}$$ and taking_the infimum_over_$\varepsilon>0$ gives_the_result.
The proof of_submultiplicativity follows in_a similar way. Indeed,_if $k,l$ are_positive_integers and $\varepsilon>0$, we find $y_1,\dots,y_{2^{l-1}}\in Y$ and $z_1,\dots,z_{2^{k-1}}\in Z$ with $$T(B_X)\subset \bigcup_{i=1}^{2^{l-1}}\Bigl(y_i+(e_l(T)+\varepsilon) B_Y\Bigr)\quad\text{and}\quad_R(B_Y)\subset_\bigcup_{j=1}^{2^{k-1}}\Bigl(z_j+(e_k(R)+\varepsilon) B_Z\Bigr).$$_The_proof_then follows from $$\begin{aligned}
(R\circ T)(B_X)&=R(T(B_X))\subset_R\biggl(\,\bigcup_{i=1}^{2^{l-1}}\Bigl(y_i+(e_l(T)+\varepsilon) B_Y\Bigr)\biggr)\\
&=\bigcup_{i=1}^{2^{l-1}}\Bigl(Ry_i+(e_l(T)+\varepsilon) R(B_Y)\Bigr)\\
&\subset \bigcup_{i,j}\Bigl(Ry_i+(e_l(T)+\varepsilon)z_j+(e_l(T)+\varepsilon)(e_k(R)+\varepsilon) B_Z\Bigr).\qedhere\end{aligned}$$
We_will also_need few basic facts about the Gamma function,_which_is defined by_$\Gamma(t)=\int_0^\infty x^{t-1}e |
,\beta,\delta,\omega\in \mathbb{R}$.$\hfill\square$
\[lem2\] $T(2) \subseteq \mathcal{G}(H,P(\cdot))$.
We first note that the NOT-gate $$\label{eq7a}
X \equiv \left[\begin{array}{cc}
0&1\\ 1&0\end{array}\right]$$ can be obtained as $$\label{eq8a}
X = HP(-\pi)H.$$ Therefore $X\in \mathcal{G}(H,P(\cdot))$. From this, we have $$\label{eq9}
XP(\omega_1) XP(\omega_2) = \left[\begin{array}{cc}
0&1\\ 1&0\end{array}\right] \left[\begin{array}{cc}
1&0\\ 0&e^{i\omega_1}\end{array}\right] \left[\begin{array}{cc}
0&1\\ 1&0\end{array}\right] \left[\begin{array}{cc}
1&0\\ 0&e^{i\omega_2}\end{array}\right] = \left[\begin{array}{cc}
e^{i\omega_1}&0\\ 0&e^{i\omega_2}\end{array}\right],$$ for any given $\omega_1,\omega_2\in \mathbb{R}$. Therefore $\mathcal{G}(H,
P(\cdot))$ contains the maximal torus $T(2)$.
\[lem3\] $SO(2) \subseteq \mathcal{G}(H, P(\cdot))$.
For each rotation matrix $$R(\omega) = \left[\begin{array}{rl}
\cos \omega&\sin\omega\\ -\sin\omega&\cos\omega\end{array}\right],$$ we easily verify that $$\label{eq10}
R(\omega) = P\left(\frac\pi2\right) HP(\omega)X P(-\omega) HP\left(-
\frac\pi2\right).$$
\[thm4\] $\mathcal{G}(H,P(\cdot)) = U(2)$.
This follows immediately from Lemmas \[lem1\]–\[lem3\].
At this point, it should already be clear from the results in \[\[ | , \beta,\delta,\omega\in \mathbb{R}$.$\hfill\square$
\[lem2\ ] $ T(2) \subseteq \mathcal{G}(H, P(\cdot))$.
We first note that the NOT - gate $ $ \label{eq7a }
X \equiv \left[\begin{array}{cc }
0&1\\ 1&0\end{array}\right]$$ can be obtained as $ $ \label{eq8a }
X = HP(-\pi)H.$$ Therefore $ X\in \mathcal{G}(H, P(\cdot))$. From this, we get $ $ \label{eq9 }
XP(\omega_1) XP(\omega_2) = \left[\begin{array}{cc }
0&1\\ 1&0\end{array}\right ] \left[\begin{array}{cc }
1&0\\ 0&e^{i\omega_1}\end{array}\right ] \left[\begin{array}{cc }
0&1\\ 1&0\end{array}\right ] \left[\begin{array}{cc }
1&0\\ 0&e^{i\omega_2}\end{array}\right ] = \left[\begin{array}{cc }
e^{i\omega_1}&0\\ 0&e^{i\omega_2}\end{array}\right],$$ for any render $ \omega_1,\omega_2\in \mathbb{R}$. Therefore $ \mathcal{G}(H,
P(\cdot))$ contains the maximal torus $ T(2)$.
\[lem3\ ] $ SO(2) \subseteq \mathcal{G}(H, P(\cdot))$.
For each rotation matrix $ $ R(\omega) = \left[\begin{array}{rl }
\cos \omega&\sin\omega\\ -\sin\omega&\cos\omega\end{array}\right],$$ we easily affirm that $ $ \label{eq10 }
R(\omega) = P\left(\frac\pi2\right) HP(\omega)X P(-\omega) HP\left(-
\frac\pi2\right).$$
\[thm4\ ] $ \mathcal{G}(H, P(\cdot) ) = U(2)$.
This follows immediately from Lemmas \[lem1\]–\[lem3\ ].
At this point, it should already be absolved from the resultant role in \[\ [ | ,\betw,\delta,\omega\in \mathbb{R}$.$\hfiul\square$
\[lem2\] $T(2) \sobweteq \kathcam{G}(H,P(\cdot))$.
De first note that the NOT-gave $$\lqbel{ew7a}
X \equiv \left[\begin{arrxy}{cc}
0&1\\ 1&0\end{agray}\right]$$ can ve obtainev as $$\labcj{eq8a}
S = HP(-\'i)I.$$ Therefore $X\in \mathcal{G}(H,[(\cdot))$. From thiv, de have $$\label{eq9}
XP(\omega_1) XP(\omega_2) = \left[\bqgin{arrsy}{fc}
0&1\\ 1&0\end{array}\rigrt] \ltft[\fegih{array}{cc}
1&0\\ 0&e^{i\omega_1}\end{array}\right] \left[\gegin{argay}{cc}
0&1\\ 1&0\end{array}\rignt] \left[\begin{array}{cc}
1&0\\ 0&e^{i\omegw_2}\end{wrray}\right] = \left[\behin{array}{cc}
e^{u\omedq_1}&0\\ 0&e^{i\omega_2}\end{xrray}\right],$$ for any givgn $\omega_1,\omega_2\in \mathbb{R}$. Thereford $\matkcal{G}(H,
P(\cdot))$ cintwhns the maxmmal tjrus $T(2)$.
\[lem3\] $SO(2) \subsetex \mathcsl{G}(H, P(\cdot))$.
For cach cotarion matrix $$R(\omega) = \lxft[\begin{array}{rl}
\cos \oiega&\sin\omagc\\ -\sin\omega&\cos\omega\end{qreay}\richt],$$ fe exwilh vtrihy fhat $$\lwbem{eq10}
R(\omega) = P\left(\frac\pu2\right) HP(\omega)X P(-\omtga) YP\left(-
\frac\pi2\rjght).$$
\[thi4\] $\iathcal{G}(H,P(\cdot)) = U(2)$.
This follows immediateny rrom Lemmas \[lem1\]–\[lem3\].
At thiw point, it should alrgady be clqar from the results in \[\[ | ,\beta,\delta,\omega\in \mathbb{R}$.$\hfill\square$ \[lem2\] $T(2) \subseteq \mathcal{G}(H,P(\cdot))$. We that NOT-gate $$\label{eq7a} \equiv \left[\begin{array}{cc} 0&1\\ $$\label{eq8a} = HP(-\pi)H.$$ Therefore \mathcal{G}(H,P(\cdot))$. From this, have $$\label{eq9} XP(\omega_1) XP(\omega_2) = \left[\begin{array}{cc} 1&0\end{array}\right] \left[\begin{array}{cc} 1&0\\ 0&e^{i\omega_1}\end{array}\right] \left[\begin{array}{cc} 0&1\\ 1&0\end{array}\right] \left[\begin{array}{cc} 1&0\\ 0&e^{i\omega_2}\end{array}\right] = \left[\begin{array}{cc} e^{i\omega_1}&0\\ for any given $\omega_1,\omega_2\in \mathbb{R}$. Therefore $\mathcal{G}(H, P(\cdot))$ contains the maximal torus $T(2)$. $SO(2) \mathcal{G}(H, For rotation matrix $$R(\omega) = \left[\begin{array}{rl} \cos \omega&\sin\omega\\ -\sin\omega&\cos\omega\end{array}\right],$$ we easily verify that $$\label{eq10} R(\omega) = P\left(\frac\pi2\right) P(-\omega) HP\left(- \frac\pi2\right).$$ \[thm4\] $\mathcal{G}(H,P(\cdot)) = U(2)$. This immediately from Lemmas \[lem1\]–\[lem3\]. this point, it should already clear the results \[\[ | ,\beta,\delta,\omega\in \mathbb{R}$.$\hfIll\square$
\[lEm2\] $T(2) \suBseTeq \MaThcaL{G}(H,P(\Cdot))$.
We first notE That The NOT-gate $$\label{eq7a}
X \equIv \lefT[\bEGin{aRRaY}{cc}
0&1\\ 1&0\enD{array}\rIGhT]$$ CAn bE oBtAinEd AS $$\lAbel{eQ8a}
X = hP(-\pi)H.$$ ThErefore $X\in \MatHcAl{G}(H,P(\cdot))$. FroM ThIs, we have $$\laBel{Eq9}
XP(\omega_1) XP(\oMegA_2) = \left[\bEgIn{aRRay}{cc}
0&1\\ 1&0\End{Array}\Right] \lEFt[\begiN{array}{cc}
1&0\\ 0&e^{I\oMEga_1}\end{ARray}\rigHT] \LeFt[\beGin{array}{cc}
0&1\\ 1&0\end{arraY}\RiGHt] \left[\begin{arrAy}{cc}
1&0\\ 0&e^{i\OmEGa_2}\END{arRay}\Right] = \left[\bEgIn{arrAY}{cc}
e^{i\omEGa_1}&0\\ 0&E^{I\OMegA_2}\End{array}\right],$$ For any given $\OMegA_1,\omega_2\In \MatHBb{R}$. TheReforE $\mAThcAl{G}(H,
P(\cdot))$ coNtaiNs the maxiMal torUS $T(2)$.
\[lem3\] $SO(2) \SUbseteq \MathcaL{G}(H, p(\cdOt))$.
FoR EaCh RotAtIOn mATrIx $$R(\OMegA) = \left[\begIn{ArRay}{rl}
\Cos \oMEGA&\Sin\oMegA\\ -\sin\Omega&\Cos\omega\end{arRay}\RighT],$$ We eAsily VerifY thaT $$\lAbel{eQ10}
R(\omegA) = P\lefT(\fRac\pi2\right) HP(\omeGa)X P(-\Omega) HP\leFt(-
\fRaC\pi2\RiGht).$$
\[thM4\] $\MathcaL{G}(H,p(\cdOt)) = U(2)$.
This Follows IMmeDiATELy From Lemmas \[lem1\]–\[lem3\].
At ThIS PoInt, it shoUld alrEAdY bE Clear froM tHe rEsulTS In \[\[ | ,\beta,\delta,\omega\in \m athbb{R}$. $\hfi ll\ squ ar e$
\[le m2\] $T(2) \su b sete q \mathcal{G}(H,P(\cdo t))$.
W e fi r st note that t h eN O T-g at e$$\ la b el {eq7a }
X \equiv \left[\be gin {a rray}{cc}
0& 1 \\ 1&0\end{a rra y}\right]$$can be ob ta ine d as $ $\l abel{ eq8a}X = HP( -\pi)H.$$ T h erefor e $X\in\ m at hcal {G}(H,P(\cdot))$. Fr o m this, we hav e $$\l ab e l{ e q 9}XP( \omega_1)XP (\ome g a_2) =\ le f t [ \be g in{array}{cc}
0&1\\ 1&0\ e nd{ array} \r igh t ] \lef t[\be gi n {ar ray}{cc}
1& 0\\0&e^{i\om ega_1} \ end{arr a y}\righ t] \le ft[ \be gin{ a rr ay }{c c} 0&1 \ \1&0 \ end {array}\ ri gh t] \l eft[ \ b e g in{a rra y}{c c}
1& 0\\ 0&e^{i\om ega _2}\ e nd{ array }\rig ht]=\left [\begi n{arr ay }{cc}
e^{i\omeg a_1} &0\\ 0&e^ {i\ om ega _2 }\end { array} \ri ght ],$$ fo r any g i ven $ \ o m eg a_1,\omega_2\in \m at h b b{ R}$. The refore $\ ma t hcal{G}( H,
P( \cdo t ) )$ co ntai n sthe maxi mal to r us $ T(2)$.
\ [lem3\ ]$SO (2) \sub s eteq \math cal{G}(H , P(\ c dot))$.
For e a ch rotation m a tr i x $ $ R(\o meg a) = \left[ \beg i n{ar ray} { rl }
\ c os \o mega& \s i n\ o mega\\ -\sin\omega& \c os\ome ga\en d{array}\righ t],$$ we e a s i ly verif y th a t$ $\label{eq10}R(\om ega) = P\l e ft(\frac \pi2\ right) H P(\omega) X P(-\omeg a)HP\ lef t(- \ fr ac\pi2\right) . $ $
\ [t hm4\] $ \ma thcal{G }(H ,P( \cd ot) )= U(2)$.
This fo ll ow sim med iatel y from Le mm as\[ lem 1\]–\ [ lem3\] .
At thi spo i nt, it sho u ld a lrea dy b e cl ear f rom t he r e sul ts in \ [\[ | ,\beta,\delta,\omega\in \mathbb{R}$.$\hfill\square$
\[lem2\]_$T(2) \subseteq_\mathcal{G}(H,P(\cdot))$.
We first note that_the NOT-gate_$$\label{eq7a}
X_\equiv \left[\begin{array}{cc}
0&1\\_1&0\end{array}\right]$$_can be obtained_as $$\label{eq8a}
X =_HP(-\pi)H.$$ Therefore $X\in \mathcal{G}(H,P(\cdot))$._From this, we_have_$$\label{eq9}
XP(\omega_1) XP(\omega_2) = \left[\begin{array}{cc}
0&1\\ 1&0\end{array}\right] \left[\begin{array}{cc}
1&0\\ 0&e^{i\omega_1}\end{array}\right] \left[\begin{array}{cc}
0&1\\ 1&0\end{array}\right] \left[\begin{array}{cc}
1&0\\ 0&e^{i\omega_2}\end{array}\right] = \left[\begin{array}{cc}
e^{i\omega_1}&0\\ 0&e^{i\omega_2}\end{array}\right],$$_for_any given_$\omega_1,\omega_2\in_\mathbb{R}$._Therefore $\mathcal{G}(H,
P(\cdot))$ contains the maximal_torus $T(2)$.
\[lem3\] $SO(2) \subseteq \mathcal{G}(H,_P(\cdot))$.
For each_rotation matrix $$R(\omega) = \left[\begin{array}{rl}
\cos \omega&\sin\omega\\ -\sin\omega&\cos\omega\end{array}\right],$$ we_easily_verify that $$\label{eq10}
R(\omega)_= P\left(\frac\pi2\right) HP(\omega)X P(-\omega) HP\left(-
\frac\pi2\right).$$
\[thm4\] $\mathcal{G}(H,P(\cdot)) = U(2)$.
This follows_immediately from Lemmas \[lem1\]–\[lem3\].
At this point, it_should already be_clear_from_the results in \[\[ |
eter’s IMF with $\alpha=-2.35$. A two-slope IMF with $\alpha=-2.00\pm 0.23$ above $0.50\,M_\odot$ and $\alpha=-1.43\pm 0.13$ below gives a better fit, formally at the $3\,\sigma$ level. However, in view of the larger error bars in the upper mass range, and the evolutionary effect away from the zero-age MS, we prefer to quote the single-slope power law as our main conclusion. This result is robust within current uncertainties in the reddening, distance modulus of the Galactic center, disk and binary stars contamination, and average metallicity of the bulge stars.
For the mass range in common ($0.35\msun\,\lsim M\lsim 1\msun$), the derived IMF is in very good agreement with the bulge IMF obtained from optical observations with WFPC2 by Holtzman et al. (1998). Our bulge IMF, however, is appreciably steeper than the low mass IMF for the solar neighborhood found in two recent determinations, which give slopes of $\alpha=-0.8$ (Reid & Gizis 1997) and $\alpha=-0.54$ (Gould et al. 1997). However, the present bulge IMF is virtually identical to yet other determinations of the solar neighborhood IMF (Kroupa et al. 1993, Reid et al. 1999), and an assessment as to whether bulge and disk IMFs are the same or not will require an understanding the origin of the large discrepancies among the various determinations of the disk IMF.
We have also compared the bulge IMF with the present day MF of some Galactic globular clusters with different metallicities and affected to various degrees by dynamical processes. In all clusters the MF is flatter than that of the bulge, but it appears to be closer to the bulge IMF in those clusters that are less affected by dynamical processes. This suggests little or no dependence of the IMF on metallicity for old systems.
One major issue concerns the amount of bulge mass that is locked in unseen dwarfs. There is no hint for the IMF slope to change towards the lower mass limit ($0.15\,\msun$) of the explored range. Assuming the slope can be extrapolated all the way to mass zero, gives a total mass of brown dwarfs ($0<M<0.08\,\msun$) in the NIC2 field of $81\,\msun$, i.e | eter ’s IMF with $ \alpha=-2.35$. A two - slope IMF with $ \alpha=-2.00\pm 0.23 $ above $ 0.50\,M_\odot$ and $ \alpha=-1.43\pm 0.13 $ below gives a better fit, formally at the $ 3\,\sigma$ level. However, in opinion of the large error bars in the upper mass image, and the evolutionary effect away from the zero - historic period MS, we prefer to quote the individual - slope power police as our main conclusion. This solution is full-bodied within current uncertainties in the reddening, distance modulus of the Galactic center, phonograph record and binary stars contamination, and modal metallicity of the bulge stars.
For the mass range in coarse ($ 0.35\msun\,\lsim M\lsim 1\msun$), the derived IMF is in very good agreement with the bulge IMF obtained from optical observation with WFPC2 by Holtzman et al. (1998). Our bulge IMF, however, is appreciably steeper than the low mass IMF for the solar neighborhood found in two late determinations, which give slopes of $ \alpha=-0.8 $ (Reid & Gizis 1997) and $ \alpha=-0.54 $ (Gould et al. 1997). However, the present bulge IMF is virtually identical to yet other determination of the solar region IMF (Kroupa et al. 1993, Reid et al. 1999), and an assessment as to whether bulge and disk IMFs are the same or not will require an understanding the origin of the large discrepancies among the assorted determinations of the disk IMF.
We have besides compare the bulge IMF with the present sidereal day MF of some Galactic globular clusters with different metallicities and affected to various degrees by dynamical process. In all clusters the MF is flatter than that of the bulge, but it appears to be closer to the bulge IMF in those clusters that are less moved by dynamical processes. This suggests little or no dependence of the IMF on metallicity for old systems.
One major emergence concerns the amount of bulge mass that is locked in unseen dwarfs. There constitute no hint for the IMF slope to change towards the low multitude limit ($ 0.15\,\msun$) of the explored stove. Assuming the gradient can be extrapolated all the way to mass zero, gives a total batch of brown dwarfs ($ 0 < M<0.08\,\msun$) in the NIC2 plain of $ 81\,\msun$, i.e | eteg’s IMF with $\alpha=-2.35$. A two-suope IMF with $\alpha=-2.00\pm 0.23$ above $0.50\,M_\odot$ avd $\alpha=-1.43\pm 0.13$ below gives a bevter fit, dormally at the $3\,\sigma$ uevel. Howvver, in vuew id the largxd error bars jk the npper mass rangg, and the evmlutionary effacg cway from the zero-age MS, we prefer tj quote tje single-slope powtr jaw zs our main conclusion. This result is roblst within curreny uncertainties in the redfenijg, distance modulud of the Gaoactyx center, disy and binagv stars conjamination, and average metallicigy of the bulge srard.
Xor the masw rande in common ($0.35\msun\,\lsik M\lsim 1\msun$), the deriyed IKF us in very good agreekent with the bulgg IMF obtahnzd from optical obsercarions witv WFOX2 bh Hklvzmzn et wl. (1998). Our bulge IMF, howevee, is appreciably sttepqg than the los mass IIF for the solar neighborhood found in uwo rscent determinations, whuch give slopes of $\alkha=-0.8$ (Reid & Dizis 1997) and $\alpha=-0.54$ (Gould et al. 1997). However, the present bulgx KMF if cigtually identical to yet other determinations kf tme solar neighbovhood IMF (Kroupa ey wl. 1993, Reid et al. 1999), xnd an asaessment as to whehher bujge abd disk IIFs sre the same or not will rewuire an undvrstqnding the origin lf the largz disctepancoes among the various dztermihations of hhe disk JOF.
We have also cumpsrad the bulge IMF with the [resent dey MF of somd Gakactic globular flustcss with different letalpiwities and affected to various degrees by dynamical provevsev. In all clusbers the MF is slatter than tkat of tke bulee, but it zppears to be closqr to the bulcg IMF in thosx clusterf thqt aee less xffected by dymamical pgoeesses. Thus suggests little or ho dependence oy uhw IMF on metalkickty flr ojg systems.
One majur kxsue zoncerns thc aoouny of bulge mass that is mocked in unseen deavfs. There is no hynt for the IKF slope to change towacds thx lowet mwss limit ($0.15\,\msun$) of the explored range. Asdumlng the slope can be extrapolcted all the way to mass zero, gives a toval mass of brown dwarfw ($0<M<0.08\,\msun$) in the NIC2 yitld of $81\,\msun$, i.x | eter’s IMF with $\alpha=-2.35$. A two-slope IMF 0.23$ $0.50\,M_\odot$ and 0.13$ below gives the level. However, in of the larger bars in the upper mass range, the evolutionary effect away from the zero-age MS, we prefer to quote the power law as our main conclusion. This result is robust within current uncertainties the distance of Galactic center, disk and binary stars contamination, and average metallicity of the bulge stars. For the range in common ($0.35\msun\,\lsim M\lsim 1\msun$), the derived is in very good with the bulge IMF obtained optical with WFPC2 Holtzman al. Our bulge IMF, is appreciably steeper than the low mass IMF for the solar neighborhood found in two recent determinations, give slopes (Reid & 1997) $\alpha=-0.54$ et al. 1997). present bulge IMF is virtually identical determinations of the solar neighborhood IMF (Kroupa et 1993, Reid al. 1999), and an assessment as whether bulge and disk IMFs are the same not will require an understanding the origin of the large discrepancies among the various determinations disk IMF. We have compared the bulge with present MF some Galactic clusters with different metallicities and affected to various degrees by dynamical In all clusters the MF is flatter than that of but appears to be to the bulge IMF those that are less affected processes. suggests dependence the on metallicity for old One major issue concerns the of bulge mass that There is no hint for the IMF slope change towards the lower mass limit ($0.15\,\msun$) the explored range. Assuming the slope can be extrapolated all the way mass zero, total mass of brown dwarfs ($0<M<0.08\,\msun$) in the field of $81\,\msun$, i.e | eter’s IMF with $\alpha=-2.35$. A two-slopE IMF with $\alPha=-2.00\pm 0.23$ AboVe $0.50\,M_\OdOt$ anD $\alpHa=-1.43\pm 0.13$ below gives A BettEr fit, formally at the $3\,\sigmA$ leveL. HOWeveR, In View oF the larGEr ERRor BaRs In tHe UPpEr masS raNge, and tHe evolutioNarY eFfect away froM ThE zero-age MS, We pRefer to quote The Single-SlOpe POwer lAw aS our mAin conCLusion. this resulT iS Robust WIthin cuRREnT uncErtainties in the reDDeNIng, distance modUlus of ThE gaLACtiC ceNter, disk anD bInary STars conTAmINATioN, And average metAllicity of tHE buLge staRs.
for THe mass Range In COmmOn ($0.35\msun\,\lsim M\Lsim 1\Msun$), the deRived Imf is in veRY good agReemenT wiTh tHe buLGe iMf obTaINed FRoM opTIcaL observaTiOnS with wFPC2 BY hOLtzmAn eT al. (1998). OUr bulGe IMF, however, iS apPrecIAblY steePer thAn thE lOw masS IMF foR the sOlAr neighborhood fOund In two receNt dEtErmInAtionS, Which gIve SloPes of $\alPha=-0.8$ (Reid & gIziS 1997) aND $\ALpHa=-0.54$ (Gould et al. 1997). However, ThE PReSent bulgE IMF is VIrTuALly identIcAl tO yet OTHer deTermINaTions of tHe solaR NeIgHborhooD ImF (KrouPa Et aL. 1993, ReId et aL. 1999), And aN assesSment as tO whetHEr bulge and disk imFs are the same OR nOT WiLL reqUirE an understaNdinG The oRigiN Of The LArge dIscrePaNCiES among the various detErMinatiOns of The disk IMF.
We hAve also comPAREd the bulGe IMf WiTH the present day mF of sOme GalactiC Globular ClustErs with dIfferent mETAllicitiEs aNd aFfeCteD TO vArious degrees BY DynaMiCal procEssEs. In all CluSteRs tHe Mf iS flatter tHan that oF tHe BuLgE, buT it apPEars to be ClOseR tO thE bulgE iMF in tHose cLustErS tHAt aRe less aFFeCTEd by DyNaMicaL prOcEsses. this SUggEsts litTle or no dePenDEnce Of ThE IMF on mEtallicity for OlD systems.
OnE mAjoR issue CONcerns thE amount of bulge mass that iS Locked iN unSeen dWarfS. There is nO hiNt for tHe Imf slope To chanGe towArDs tHE Lower MASs LimIt ($0.15\,\Msun$) of the eXPLorEd ranGe. assuMing the Slope can be extrapolATed All the way to maSs zEro, gIVEs A toTAl MAss Of BRowN DWarfs ($0<M<0.08\,\msun$) in the nIC2 field of $81\,\MsUN$, i.E | eter’s IMF with $\alpha=-2 .35$. A tw o-slo peIMF w ith$\al pha=-2.00\pm 0 . 23$above $0.50\,M_\odot$and $ \a l pha= - 1. 43\pm 0.13$b el o w gi ve sa b et t er fit, fo rmallyat the $3\ ,\s ig ma$ level. H o we ver, in vi ewof the large r e rror b ar s i n theupp er ma ss ran g e, and the evol ut i onarye ffect a w a yfrom the zero-age MS, we prefer to quot e thesi n gl e - slo pepower lawas ourm ain con c lu s i o n.T his result is robust wit h incurren tunc e rtaint ies i nt hereddening,dist ance modu lus of the Gal a ctic ce nter,dis k a nd b i na ry st ar s co n ta min a tio n, and a ve ra ge me tall i c i t y of th e bu lge s tars.
For th e m assr ang e incommo n ($ 0. 35\ms un\,\l sim M \l sim 1\msun$), t he d erived IM F i sinve ry go o d agre eme ntwith th e bulge IMF o b t a in ed from optical ob se r v at ions wit h WFPC 2 b yH oltzmanet al . (1 9 9 8). O ur b u lg e IMF, h owever , i sappreci ab ly ste ep ertha n the lowmass I MF for t he so l ar neighborhoo d found in two re c e nt dete rmi nations, wh ichg iveslop e sof$ \alph a=-0. 8$ (R e id & Gizis 1997) an d$\alph a=-0. 54$ (Gould et al. 1997) . H owever,thep re s ent bulge IMFis vi rtually id e ntical t o yet other d eterminat i o ns of th e s ola r n eig h b or hood IMF (Kro u p a et a l. 1993 , R eid etal. 19 99) , a nd an asses sment as t owh et her bulg e and dis kIMF sare thes ame or notwill r eq u ire an und e rs t a ndin gth e or igi nof th e la r gediscrep ancies am ong theva ri ous det erminations o fthe disk I MF .
We hav e also com pared the bulge IMF wit h the pr ese nt da y MF of someGal acticglo b ular c luster s wit hdif f e rentm e ta lli ci ties and a f f ect ed to v ario us degr ees by dynamical p r oce sses. In allclu ster s th e M F i s fl at t ert h an that of thebulge, but i t a ppears tob e c lo ser tothe bul ge IM F in tho se cluste rs that a re les s aff ected by d ynamical processe s . Thi s s ugges tslittle o r n o dep endenc e of theIMF on m etalli cityfo r old sy stems.
One major issue conce rns t heamount of bu l gemass that islocked inuns een dwar fs. There isn ohin t fortheI MF slopet ocha n g etowards the l o wer mass li m it ($0 .15\ ,\msun$) of the e x plored range.Assu m i ngthe slop ecan be extrapo lat ed a ll the w ay to mass ze ro, give sa tota l mass of br own dwa r f s( $0<M<0 .08\ ,\m sun$) inthe N I C2 fiel dof $81\,\ msun $, i.e | eter’s IMF_with $\alpha=-2.35$._A two-slope IMF with_$\alpha=-2.00\pm 0.23$_above_$0.50\,M_\odot$ and_$\alpha=-1.43\pm_0.13$ below gives_a better fit,_formally at the $3\,\sigma$_level. However, in_view_of the larger error bars in the upper mass range, and the evolutionary effect_away_from the_zero-age_MS,_we prefer to quote the_single-slope power law as our_main conclusion._This result is robust within current uncertainties in_the_reddening, distance modulus_of the Galactic center, disk and binary stars contamination,_and average metallicity of the bulge_stars.
For the mass_range_in_common ($0.35\msun\,\lsim M\lsim 1\msun$),_the derived IMF is in very_good agreement with the bulge IMF_obtained from optical observations with WFPC2 by_Holtzman et al. (1998). Our bulge_IMF, however, is appreciably steeper_than the_low mass IMF for the_solar neighborhood found_in two_recent determinations, which_give slopes of $\alpha=-0.8$ (Reid &_Gizis 1997) and_$\alpha=-0.54$ (Gould et al. 1997). However,_the_present bulge IMF_is_virtually_identical to_yet other determinations_of_the solar_neighborhood_IMF (Kroupa et al. 1993, Reid_et_al. 1999), and an assessment as to_whether bulge and disk_IMFs_are the same or_not will require an understanding_the origin of the large discrepancies_among the_various determinations_of the disk IMF.
We have also compared the bulge IMF with_the present day MF of some_Galactic globular clusters with_different metallicities_and_affected to various_degrees_by dynamical_processes. In all clusters the MF is_flatter than_that of the bulge, but it_appears to be closer_to_the bulge IMF in those clusters_that are less affected by dynamical_processes. This suggests little or_no_dependence_of the IMF on metallicity_for old systems.
One major issue concerns_the amount of_bulge mass that is locked in unseen_dwarfs._There is no hint for the_IMF_slope to change towards the lower_mass_limit_($0.15\,\msun$) of the explored range._Assuming the slope can be extrapolated_all the way to mass zero, gives a total_mass of brown_dwarfs ($0<M<0.08\,\msun$) in the NIC2_field_of_$81\,\msun$, i.e |
,u_3,u_4) + R(u_1,u_3,u_4,u_2) + R(u_1,u_4,u_2,u_3) = 0,$
for all $u_1,u_2,u_3,u_4\in \mathbb R^n$. If $R$ is ${\rm U}(n)$-invariant then $R$ is a multiple of the tensor ${\bf K}$.
Let us consider $e_1,\ldots,e_n$ the canonical basis of $\mathbb C^n$ es complex vector space, so that $e_1,\ldots, e_n,$ $ie_1, \ldots, ie_n$ is a real basis. Let us take $\lambda = R(e_1,ie_1,e_1,ie_1)$. For vector $u\in \mathbb C^n$ there is an unitary transformation $g$ such that $g(u) = \|u\|e_1$ and thus: $$R(u,iu,u,iu) = \langle u,u \rangle^2 R(e_1,ie_1,e_1,ie_1) = \lambda \langle u,u \rangle^2 = \lambda {\bf K}(u,iu,u,iu).$$ Finally, by Lemma [@KN vol II, Chapter IX, Proposition 7.1] we have that $R = \lambda {\bf K}$.
Thus, we have $R_0 = \lambda{\bf K}$. This tensor $R_0$ is, by definition, the curvature tensor of a Kähler manifold of constant holomorphic curvature $\lambda$. Hence, $(M,\nabla,P)$ is a Kähler manifold of constant holomorphic curvature. Summarizing:
Let $(M,\nabla,P)$ be a manifold endowed with a connection $\nabla$ and an $U(n)$-structure $P$. It is an infinitesimally homogeneous manifold if and only if $P$ is the bundle of unitary frames of a constant holomorphic curvature Kähler structure in $M$ and $\nabla$ is its corresponding Levi-Civita connection.
Product of Riemannian structures
--------------------------------
Let us fix the group $\mathrm{O}({{\ensuremath{\mathbb{R}}}}^{n_1})\times \mathrm{O}({{\ensuremath{\mathbb{R}}}}^{n | , u_3,u_4) + R(u_1,u_3,u_4,u_2) + R(u_1,u_4,u_2,u_3) = 0,$
for all $ u_1,u_2,u_3,u_4\in \mathbb R^n$. If $ R$ is $ { \rm U}(n)$-invariant then $ R$ is a multiple of the tensor $ { \bf K}$.
Let us consider $ e_1,\ldots, e_n$ the canonical basis of $ \mathbb C^n$ es complex vector quad, so that $ e_1,\ldots, e_n,$ $ ie_1, \ldots, ie_n$ is a veridical basis. Let us take $ \lambda = R(e_1,ie_1,e_1,ie_1)$. For vector $ u\in \mathbb C^n$ there be an unitary transformation $ g$ such that $ g(u) = \|u\|e_1 $ and thus: $ $ R(u, iu, u, iu) = \langle u, u \rangle^2 R(e_1,ie_1,e_1,ie_1) = \lambda \langle u, u \rangle^2 = \lambda { \bf K}(u, iu, u, iu).$$ last, by Lemma [ @KN vol II, Chapter IX, Proposition 7.1 ] we give birth that $ R = \lambda { \bf K}$.
Thus, we have $ R_0 = \lambda{\bf K}$. This tensor $ R_0 $ is, by definition, the curvature tensor of a Kähler manifold of ceaseless holomorphic curvature $ \lambda$. Hence, $ (M,\nabla, P)$ is a Kähler manifold of constant holomorphic curvature. Summarizing:
get $ (M,\nabla, P)$ embody a manifold endowed with a association $ \nabla$ and an $ U(n)$-structure $ P$. It is an infinitesimally homogeneous manifold if and only if $ P$ is the package of unitary frames of a constant holomorphic curvature Kähler social organization in $ M$ and $ \nabla$ is its corresponding Levi - Civita connection.
merchandise of Riemannian structures
--------------------------------
permit us fix the group $ \mathrm{O}({{\ensuremath{\mathbb{R}}}}^{n_1})\times \mathrm{O}({{\ensuremath{\mathbb{R}}}}^{n | ,u_3,u_4) + R(u_1,u_3,u_4,u_2) + R(u_1,u_4,u_2,u_3) = 0,$
for all $u_1,m_2,u_3,u_4\in \mathbb R^n$. Nd $R$ is ${\rm U}(n)$-jnvariang then $R$ is a multiple of thx tebsor ${\vf K}$.
Let us consider $e_1,\laots,e_n$ thv canonicql besis of $\mathbb C^i$ es comijex bcctor wpace, so that $g_1,\ldots, e_n,$ $ie_1, \ldots, ie_n$ is d fecl basis. Let us take $\lambda = R(e_1,ie_1,e_1,ie_1)$. For vevtlr $u\in \mathbb S^n$ tnqre js an unitary transformation $g$ sucg that $j(u) = \|u\|e_1$ and thus: $$R(u,iu,u,iu) = \langle u,u \rangle^2 R(e_1,if_1,e_1,ie_1) = \lambda \langlf u,u \rangle^2 = \lwnbda {\bf K}(u,iu,j,iu).$$ Finallj, by Lemma [@IN vol II, Chapter IX, Propositiov 7.1] we have that $R = \lwkbda {\bf K}$.
Thns, we rave $R_0 = \lambda{\bf K}$. Tvis tenxor $R_0$ is, by deninitmon, rhe curvature tensor mf a Kähler manifolq of consdaut holomorphic curvatyrw $\lamtda$. Vencd, $(M,\nxblz,P)$ ia a Kähper manifold kf constant holomorphic curvatirq. Summarizing:
Lst $(M,\naflw,P)$ be a manifold endowed with a connectpon $\habla$ and an $U(n)$-structurw $P$. It is an infinitedimally hjmogeneous manifold if and only if $P$ is the bundla of nnktaxn weales of a constant holomorphic curvature Kähler attubture in $M$ and $\nanla$ is its corresppnfimd Levi-Civita zonneccjoh.
Product of Riemanjian sttucturws
--------------------------------
Let us six yhe group $\mathrm{O}({{\ensuremath{\nathbb{R}}}}^{n_1})\timef \mathrm{O}({{\ensuremath{\lathbb{R}}}}^{n | ,u_3,u_4) + R(u_1,u_3,u_4,u_2) + R(u_1,u_4,u_2,u_3) = 0,$ $u_1,u_2,u_3,u_4\in R^n$. If is ${\rm U}(n)$-invariant of tensor ${\bf K}$. us consider $e_1,\ldots,e_n$ canonical basis of $\mathbb C^n$ es vector space, so that $e_1,\ldots, e_n,$ $ie_1, \ldots, ie_n$ is a real basis. us take $\lambda = R(e_1,ie_1,e_1,ie_1)$. For vector $u\in \mathbb C^n$ there is an transformation such $g(u) \|u\|e_1$ and thus: $$R(u,iu,u,iu) = \langle u,u \rangle^2 R(e_1,ie_1,e_1,ie_1) = \lambda \langle u,u \rangle^2 = \lambda K}(u,iu,u,iu).$$ Finally, by Lemma [@KN vol II, Chapter Proposition 7.1] we have $R = \lambda {\bf K}$. we $R_0 = K}$. tensor is, by definition, curvature tensor of a Kähler manifold of constant holomorphic curvature $\lambda$. Hence, $(M,\nabla,P)$ is a Kähler manifold constant holomorphic Let $(M,\nabla,P)$ a endowed a connection $\nabla$ $U(n)$-structure $P$. It is an infinitesimally and only if $P$ is the bundle of frames of constant holomorphic curvature Kähler structure in and $\nabla$ is its corresponding Levi-Civita connection. Product Riemannian structures -------------------------------- Let us fix the group $\mathrm{O}({{\ensuremath{\mathbb{R}}}}^{n_1})\times \mathrm{O}({{\ensuremath{\mathbb{R}}}}^{n | ,u_3,u_4) + R(u_1,u_3,u_4,u_2) + R(u_1,u_4,u_2,u_3) = 0,$
for all $u_1,u_2,u_3,u_4\in \mAthbb R^n$. If $R$ Is ${\rm U}(N)$-inVarIaNt thEn $R$ iS a multiple of thE TensOr ${\bf K}$.
Let us consider $e_1,\ldoTs,e_n$ tHe CAnonICaL basiS of $\mathBB C^N$ ES coMpLeX veCtOR sPace, sO thAt $e_1,\ldotS, e_n,$ $ie_1, \ldots, Ie_n$ Is A real basis. LeT Us Take $\lambda = r(e_1,iE_1,e_1,ie_1)$. For vectoR $u\iN \mathbB C^N$ thERe is aN unItary TransfORmatioN $g$ such thaT $g(U) = \|U\|e_1$ and tHUs: $$R(u,iu,u,IU) = \LaNgle U,u \rangle^2 R(e_1,ie_1,e_1,ie_1) = \laMBdA \Langle u,u \rangle^2 = \Lambda {\Bf k}(U,iU,U,Iu).$$ FInaLly, by Lemma [@kN Vol II, cHapter Ix, prOPOSitIOn 7.1] we have that $R = \Lambda {\bf K}$.
ThUS, we Have $R_0 = \lAmBda{\BF K}$. This TensoR $R_0$ IS, by Definition, tHe cuRvature teNsor of A kähler mANifold oF constAnt HolOmorPHiC cUrvAtURe $\lAMbDa$. HENce, $(m,\nabla,P)$ iS a käHler mAnifOLD OF conStaNt hoLomorPhic curvature. sumMariZIng:
let $(M,\nAbla,P)$ Be a mAnIfold EndoweD with A cOnnection $\nabla$ aNd an $u(n)$-structuRe $P$. it Is aN iNfiniTEsimalLy hOmoGeneous ManifolD If aNd ONLY iF $P$ is the bundle of uniTaRY FrAmes of a cOnstanT HoLoMOrphic cuRvAtuRe KäHLEr strUctuRE iN $M$ and $\nabLa$ is itS CoRrEspondiNg levi-CiViTa cOnnEctioN.
prodUct of RIemanniaN struCTures
--------------------------------
Let us fix tHE group $\mathrm{O}({{\ENsUREmATh{\maThbB{R}}}}^{n_1})\times \matHrm{O}({{\ENsurEmatH{\MaThbB{r}}}}^{n | ,u_3,u_4) + R(u_1,u_3,u_4, u_2) + R(u _1,u_ 4,u _2, u_ 3) = 0,$
for all $u_1 , u_2, u_3,u_4\in \mathbb R^n $. If $ R $ is ${ \rm U }(n)$-i n va r i ant t he n $ R$ is a mu lti ple ofthe tensor ${ \b f K}$.
Letu sconsider $ e_1 ,\ldots,e_n$ th e cano ni cal basis of $\ma thbb C ^ n$ escomplex v ec t or spa c e, so t h a t$e_1 ,\ldots, e_n,$ $i e _1 , \ldots, ie_n$ is are a lb a sis . L et us take $ \lamb d a = R(e _ 1, i e _ 1,e _ 1,ie_1)$. For vector $u\ i n \ mathbb C ^n$ thereis an u n ita ry transfor mati on $g$ su ch tha t $g(u)= \|u\|e _1$ an d t hus : $$ R (u ,i u,u ,i u ) = \l ang l e u ,u \rang le ^2 R(e_ 1,ie _ 1 , e _1,i e_1 ) =\lamb da \langle u, u \ rang l e^2 = \ lambd a {\ bf K}(u ,iu,u, iu).$ $Finally, by Lem ma [ @KN vol I I,Ch apt er IX,P roposi tio n 7 .1] wehave th a t $ R= \ la mbda {\bf K}$.
Th us , we have $R _0 = \ l am bd a {\bf K}$ .Thi s te n s or $R _0$i s, by defi nition , t he curvat ur e tens or of aKähle r man ifoldof const ant h o lomorphic curv a ture $\lambda $ .H e nc e , $( M,\ nabla,P)$ i s aK ähle r ma n if old of co nstan th ol o morphic curvature.Su mmariz ing:
Let $(M,\nab la,P)$ bea m anifoldendo w ed with a connect ion $ \nabla$ an d an $U(n )$-st ructure$P$. It i s an infin ite sim all y h o m og eneous manifo l d ifan d onlyif$P$ isthe bu ndl e o funitary f rames of a c on st ant holo m orphic c ur vat ur e K ähler struct ure i n $M $an d $\ nabla$i si t s co rr es pond ing L evi-C ivit a co nnectio n.
Produ cto f Ri em an nian st ructures
---- -- ---------- -- --- ------ - - ---
Let us fix the group $\mat h rm{O}({ {\e nsure math {\mathbb{ R}} }}^{n_ 1}) \ times\mathr m{O}( {{ \en s u remat h { \m ath bb {R}}}}^{n | ,u_3,u_4) +_R(u_1,u_3,u_4,u_2) +_R(u_1,u_4,u_2,u_3) = 0,$
for all_$u_1,u_2,u_3,u_4\in \mathbb_R^n$._If $R$_is_${\rm U}(n)$-invariant then_$R$ is a_multiple of the tensor_${\bf K}$.
Let us_consider_$e_1,\ldots,e_n$ the canonical basis of $\mathbb C^n$ es complex vector space, so that $e_1,\ldots,_e_n,$_$ie_1, \ldots,_ie_n$_is_a real basis. Let us_take $\lambda = R(e_1,ie_1,e_1,ie_1)$. For_vector $u\in_\mathbb C^n$ there is an unitary transformation $g$_such_that $g(u) =_\|u\|e_1$ and thus: $$R(u,iu,u,iu) = \langle u,u \rangle^2 R(e_1,ie_1,e_1,ie_1)_= \lambda \langle u,u \rangle^2 _= \lambda {\bf_K}(u,iu,u,iu).$$_Finally,_by Lemma [@KN vol_II, Chapter IX, Proposition 7.1] we_have that $R = \lambda {\bf_K}$.
Thus, we have $R_0 = \lambda{\bf K}$._This tensor $R_0$ is, by definition,_the curvature tensor of a_Kähler manifold_of constant holomorphic curvature $\lambda$._Hence, $(M,\nabla,P)$ is_a Kähler_manifold of constant_holomorphic curvature. Summarizing:
Let $(M,\nabla,P)$ be a_manifold endowed with_a connection $\nabla$ and an $U(n)$-structure_$P$._It is an_infinitesimally_homogeneous_manifold if_and only if_$P$_is the_bundle_of unitary frames of a constant_holomorphic_curvature Kähler structure in $M$ and $\nabla$_is its corresponding Levi-Civita_connection.
Product_of Riemannian structures
--------------------------------
Let us_fix the group $\mathrm{O}({{\ensuremath{\mathbb{R}}}}^{n_1})\times \mathrm{O}({{\ensuremath{\mathbb{R}}}}^{n |
projector in the labels $R$ and $(r,s)$ and an intertwiner in the labels $\alpha$ and $\beta$. These labels are the multiplicity labels with which irrep $(r,s)$ is subduced from $R$ when restricting from $S_{n+m}$ to $S_{n}\times S_{m}$. Further details of (\[eq:RSUNDef\]) can be found in [@GGWSA1], [@GGO] for example. There are many good reasons to study the operators defined in (\[eq:RSUNDef\]) as we now explain.
The counting of states given by the partition function of the $1/4$-BPS sector of the free $U(N)$ theory was shown to match the counting of restricted Schurs that can be defined [@Collins]. Essentially, the partition function was expressed in terms of Littlewood Richardson coefficients which count the number of restricted Schurs for a given set of Young diagram labels, $R,(r,s)$. This result was generalised to an arbitrary product of $U(N)$ gauge groups in [@Quiver].
We are interested in operators whose bare dimension grows parametrically with $N$. For these operators, the large $N$ limit of correlation functions is not captured by summing only planar diagrams [@GGO]. The usual $1/N^{2}$ factor suppressing non-planar diagrams is over-powered by combinatorial factors resulting from evaluating all possible Wick contractions. Indeed, when the number of matrix fields in the operator scales with $N$, we are forced to sum an infinite number of non-planar diagrams in the large $N$ limit. A solution to this problem was given in [@Jevicki] for the $1/2$-BPS case. In this important and influential work, it was observed that there is a one-to-one correspondence between the space of $1/2$-BPS representations and Young diagrams, i.e., Schur polynomials. Using representation theory of the symmetric and unitary group, the two-point function of Schur polynomials was computed *exactly* in the free theory limit. [@EMC] then achieved the remarkable result of computing the two-point function of restricted Schurs (\[eq:RSUNDef\]) exactly in the free theory. The operators diagonalised the two-point function and a very simple formula for the final result was given. Thus, the restricted Schurs provide an exactly orthogonal basis for the $1/4$- | projector in the labels $ R$ and $ (r, s)$ and an intertwiner in the labels $ \alpha$ and $ \beta$. These labels are the numerousness label with which irrep $ (r, s)$ is subduced from $ R$ when restricting from $ S_{n+m}$ to $ S_{n}\times S_{m}$. Further detail of (\[eq: RSUNDef\ ]) can be establish in [ @GGWSA1 ], [ @GGO ] for example. There are many adept rationality to study the operator defined in (\[eq: RSUNDef\ ]) as we now excuse.
The counting of states given by the partition affair of the $ 1/4$-BPS sector of the free $ U(N)$ theory was shown to match the counting of restrict Schurs that can be defined [ @Collins ]. Essentially, the partition routine was expressed in terms of Littlewood Richardson coefficients which count the act of restricted Schurs for a given set of Young diagram labels, $ R,(r, s)$. This result was generalized to an arbitrary product of $ U(N)$ gauge groups in [ @Quiver ].
We are interested in operators whose bare property grows parametrically with $ N$. For these operators, the large $ N$ limit of correlation functions is not captured by summing only planar diagrams [ @GGO ]. The common $ 1 / N^{2}$ component suppressing non - planar diagram is all over - powered by combinatorial factors resulting from measure all possible Wick contractions. Indeed, when the number of matrix fields in the hustler scales with $ N$, we are forced to sum an infinite number of non - planar diagrams in the large $ N$ limit. A solution to this problem was given in [ @Jevicki ] for the $ 1/2$-BPS case. In this important and influential workplace, it was observed that there cost a one - to - one correspondence between the space of $ 1/2$-BPS representations and Young diagrams, i.e., Schur polynomials. practice representation theory of the symmetric and unitary group, the two - point function of Schur polynomials was computed * precisely * in the free theory terminus ad quem. [ @EMC ] then achieved the remarkable result of computing the two - period function of qualify Schurs (\[eq: RSUNDef\ ]) exactly in the free theory. The operators diagonalised the two - detail function and a very bare formula for the final result was given. Thus, the restrict Schurs provide an exactly extraneous basis for the $ 1/4$- | prljector in the labels $R$ xnd $(r,s)$ and an iurertwiier in fhe labeus $\alpha$ and $\beta$. These labeps are uke multiplicity labeus with wjich irrwp $(r,w)$ is subducxs from $V$ wheh rescrmcting from $S_{n+m}$ to $S_{n}\timev S_{m}$. Further datxips of (\[eq:RSUNDef\]) can be found in [@GGWFA1], [@GGO] gog example. Therg are iany good reasons to study the operatods defiied in (\[eq:RSUNDeg\]) as we now explain.
The coujtinh of states given hy the partutiog function of the $1/4$-BPS stccor of the rree $U(N)$ theory was shown to matzh thz counting if redjricted Schucs than can be defikvd [@Collhns]. Essrntially, the psrtmtiob function was expresved in terms of Lijtlewood Rhckardson coefficients qhuch cmunt the bumcer oh rsstriched Schurs fod a given swt of Young diagram lwvels, $R,(r,s)$. This resule ras generalised to an arbitrary product of $U(N)$ gauge groups in [@Quicer].
We are interested ln operatjrs whose bare dimension grows parametrically witv $N$. Fkf tkcfd ooerators, the large $N$ limit of correlation funsfipnx is not captuved by summing onlu olsgar diagrams [@EGO]. Thz uaual $1/N^{2}$ factor suppgessing non-poanar diadramx is over-powered by combinarorial factogs rwsulting from evalbating all pussinle Wock contractions. Indeed, wheh the numbeg of matrjb fields in the upegatos scales with $N$, we are forsed to sun an infinige nomber os non-planag diansams in the large $J$ limnt. A volution tl this problem was given in [@Jevmrki] for the $1/2$-BKS wasv. In this impovtant and influqntial work, it was obfervea that thege is a oie-to-one corrqspondence bedaeen the spare of $1/2$-BPS repeesebtationr and Young diabrams, i.e., Schur polybomials. Using reprcsentxfion theory of cke symmetric and inigarr hrpu[, the two-poind fuvctkpn of Schur kjlnnooialx was computed *exactny* ih the free theory kiiit. [@EMC] jhen achiqved the rematkable result of clmputmng thx two-ppinj function of restricted Schurs (\[eq:RSUNDev\]) ewactly in the frec thgory. The opzrators diagonalised the two-point functiin and a very simkle formula for the fnnsl result wes givqn. Thus, tve restricted Schurs provide an exactky orthogonal basis fkr the $1/4$- | projector in the labels $R$ and $(r,s)$ intertwiner the labels and $\beta$. These with irrep $(r,s)$ is from $R$ when from $S_{n+m}$ to $S_{n}\times S_{m}$. Further of (\[eq:RSUNDef\]) can be found in [@GGWSA1], [@GGO] for example. There are many reasons to study the operators defined in (\[eq:RSUNDef\]) as we now explain. The of given the function of the $1/4$-BPS sector of the free $U(N)$ theory was shown to match the counting restricted Schurs that can be defined [@Collins]. Essentially, partition function was expressed terms of Littlewood Richardson coefficients count number of Schurs a set of Young labels, $R,(r,s)$. This result was generalised to an arbitrary product of $U(N)$ gauge groups in [@Quiver]. We interested in bare dimension parametrically $N$. these operators, the limit of correlation functions is not only planar diagrams [@GGO]. The usual $1/N^{2}$ factor non-planar diagrams over-powered by combinatorial factors resulting from all possible Wick contractions. Indeed, when the number matrix fields in the operator scales with $N$, we are forced to sum an infinite non-planar diagrams in the $N$ limit. A to problem given [@Jevicki] for $1/2$-BPS case. In this important and influential work, it was observed there is a one-to-one correspondence between the space of $1/2$-BPS Young i.e., Schur polynomials. representation theory of the and group, the two-point function polynomials computed free limit. then achieved the remarkable of computing the two-point function restricted Schurs (\[eq:RSUNDef\]) exactly operators diagonalised the two-point function and a very formula for the final result was given. the restricted Schurs provide an exactly orthogonal basis for the $1/4$- | projector in the labels $R$ and $(r,S)$ and an inteRtwinEr iN thE lAbelS $\alpHa$ and $\beta$. These LAbelS are the multiplicity labEls wiTh WHich IRrEp $(r,s)$ iS subducED fROM $R$ wHeN rEstRiCTiNg froM $S_{n+M}$ to $S_{n}\tiMes S_{m}$. FurthEr dEtAils of (\[eq:RSUNdEf\]) Can be found In [@GgWSA1], [@GGO] for exAmpLe. TherE aRe mANy gooD reAsons To studY The opeRators defInED in (\[eq:RsuNDef\]) as WE NoW expLain.
The counting of STaTEs given by the paRtitioN fUNcTIOn oF thE $1/4$-BPS sector Of The frEE $U(N)$ theoRY wAS SHowN To match the couNting of restRIctEd SchuRs ThaT Can be dEfineD [@COLliNs]. EssentialLy, thE partitioN functIOn was exPRessed iN terms Of LIttLewoOD RIcHarDsON coEFfIciENts Which couNt ThE numbEr of RESTRictEd SChurS for a Given set of YouNg dIagrAM laBels, $R,(R,s)$. ThiS resUlT was gEneralIsed tO aN arbitrary produCt of $u(N)$ gauge grOupS iN [@QuIvEr].
We aRE interEstEd iN operatOrs whosE BarE dIMENsIon grows parametricAlLY WiTh $N$. For thEse opeRAtOrS, The large $n$ lImiT of cORRelatIon fUNcTions is nOt captUReD bY summinG oNly plaNaR diAgrAms [@GGo]. the uSual $1/N^{2}$ fActor supPressINg non-planar diaGRams is over-powEReD BY cOMbinAtoRial factors ResuLTing From EVaLuaTIng alL possIbLE WICk contractions. IndeeD, wHen the NumbeR of matrix fielDs in the opeRATOr scales With $n$, We ARe forced to sum aN infiNite number OF non-planAr diaGrams in tHe large $N$ lIMIt. A solutIon To tHis ProBLEm Was given in [@JevICKi] foR tHe $1/2$-BPS caSe. IN this imPorTanT anD inFlUential woRk, it was oBsErVeD tHat There IS a one-to-oNe CorReSpoNdencE BetweeN the sPace Of $1/2$-bPs RepResentaTIoNS And YOuNg DiagRamS, i.E., SchuR polYNomIals. UsiNg represeNtaTIon tHeOrY of the sYmmetric and unItAry group, thE tWo-pOint fuNCTion of ScHur polynomials was computED *exactlY* in The frEe thEory limit. [@eMC] Then acHieVEd the rEmarkaBle reSuLt oF COmputING tHe tWo-Point functION of RestrIcTed SChurs (\[eq:rSUNDef\]) exactly in thE FreE theory. The opeRatOrs dIAGoNalISeD The TwO-PoiNT Function and a verY simple forMuLA fOr the final REsuLt Was giveN. Thus, thE restRIcted ScHurs proviDe an exactLy OrthOGOnaL basis for tHe $1/4$- | projector in the labels $ R$ and $(r ,s)$and an i nter twin er in the labe l s $\ alpha$ and $\beta$. Th ese l ab e ls a r ethe m ultipli c it y lab el swit hw hi ch ir rep $(r,s) $ is subdu ced f rom $R$ when re strictingfro m $S_{n+m}$to$S_{n} \t ime s S_{m }$. Furt her de t ails o f (\[eq:R SU N Def\]) can bef o un d in [@GGWSA1], [@GGO ] f o r example. The re are m a ny g ood re asons to s tu dy th e operat o rs d e fin e d in (\[eq:RS UNDef\]) as wenow ex pl ain .
Thecount in g of states giv en b y the par tition functio n of the $1/4$ -BP S s ecto r o fthe f r ee$ U( N)$ the ory wassh ow n tomatc h t h e co unt ingof re stricted Schu rsthat can be d efine d [@ Co llins ]. Ess entia ll y, the partitio n fu nction wa s e xp res se d int erms o f L itt lewoodRichard s onco e f f ic ients which countth e nu mber ofrestri c te dS churs fo ra g iven s et of You n gdiagramlabels , $ R, (r,s)$. T his re su ltwas gene r alis ed toan arbit raryp roduct of $U(N ) $ gauge group s i n [@ Q uive r].
We are in tere s tedin o p er ato r s who se ba re di m ension grows parame tr ically with $N$. For the se operato r s , the lar ge $ N $l imit of correl ation functions is not c aptur ed by su mming onl y planar d iag ram s [ @GG O ] .The usual $1/ N ^ {2}$ f actor s upp ressing no n-p lan ardi agrams is over-po we re dby co mbina t orial fa ct ors r esu lting from e valua ting a ll pos sible W i ck c ontr ac ti ons. In de ed, w hent henumberof matrix fi e ldsin t he oper ator scales w it h $N$, wear e f orcedt o sum aninfinite number of non- p lanar d iag ramsin t he large$N$ limit . A soluti on tothispr obl e m wasg i ve n i n[@Jevicki] f orthe $ 1/ 2$-B PS case . In this importan t an d influential wo rk,i t w aso bs e rve dt hat t here is a one-t o-one corr es p on dence betw e enth e space of $1/ 2$-BP S repres entations and Youn gdiag r a ms, i.e., Sch ur polyn omials. U s ing r e pr esent ati on the or y o f the symme t ric andunitar ygroup, thetw o-pointfunction of Schur polyn omials wascom puted *ex act l y*in the fr ee t heory limi t.[@E MC] t hen achie vedt he re m arkab le r e sult of c o mp uti n g t he two-poin t f unc tionofr estric tedSchurs (\[eq:RSUN D ef\]) exactlyin t h e fr eet heor y. The operators di ag o n alised t he two-pointfunction a n d a v ery si mple f ormulaf o rt he fin al r esu lt was gi ven .T hus, th ere s tricte d Sc hu rs pro vide a n exa c t ly orthogonal ba sis f o r the$ 1/4 $- | projector_in the_labels $R$ and $(r,s)$_and an_intertwiner_in the_labels_$\alpha$ and $\beta$._These labels are_the multiplicity labels with_which irrep $(r,s)$_is_subduced from $R$ when restricting from $S_{n+m}$ to $S_{n}\times S_{m}$. Further details of (\[eq:RSUNDef\])_can_be found_in_[@GGWSA1],_[@GGO] for example. There are_many good reasons to study_the operators_defined in (\[eq:RSUNDef\]) as we now explain.
The counting_of_states given by_the partition function of the $1/4$-BPS sector of the_free $U(N)$ theory was shown to_match the counting_of_restricted_Schurs that can be_defined [@Collins]. Essentially, the partition function_was expressed in terms of Littlewood_Richardson coefficients which count the number of_restricted Schurs for a given set_of Young diagram labels, $R,(r,s)$._This result_was generalised to an arbitrary_product of $U(N)$_gauge groups_in [@Quiver].
We are_interested in operators whose bare dimension_grows parametrically with_$N$. For these operators, the large_$N$_limit of correlation_functions_is_not captured_by summing only_planar_diagrams [@GGO]._The_usual $1/N^{2}$ factor suppressing non-planar diagrams_is_over-powered by combinatorial factors resulting from evaluating_all possible Wick contractions._Indeed,_when the number of_matrix fields in the operator_scales with $N$, we are forced_to sum_an infinite_number of non-planar diagrams in the large $N$ limit. A solution_to this problem was given in_[@Jevicki] for the $1/2$-BPS_case. In_this_important and influential_work,_it was_observed that there is a one-to-one correspondence_between the_space of $1/2$-BPS representations and Young_diagrams, i.e., Schur polynomials._Using_representation theory of the symmetric and_unitary group, the two-point function of_Schur polynomials was computed *exactly*_in_the_free theory limit. [@EMC] then_achieved the remarkable result of computing_the two-point function_of restricted Schurs (\[eq:RSUNDef\]) exactly in the_free_theory. The operators diagonalised the two-point_function_and a very simple formula for_the_final_result was given. Thus, the_restricted Schurs provide an exactly orthogonal_basis for the $1/4$- |
G] = \reg_X \otimes (\chi^0\oplus\cdots\oplus\chi^{m-1})$ is a relative tilting sheaf for $\Db_G(X)$ over $\Dperf(Y)$.
If $L\in\Pic(Y)\subset\Dperf(Y)$ is an ample line bundle, then so is $\pi^*(L)$. Hence, $\Db(X)$ has a generator of the form $E \coloneqq \pi^*(\reg_Y\oplus L\oplus\cdots\oplus L^{\otimes k})$ for some $k\gg0$; see [@Orlov_gen].
In particular, $E$ is a spanning class of $\Db(X)$. Using the adjunction $\Res\dashv \Ind\dashv\Res$, it follows that $\Ind(E) \cong E\oplus E\otimes \chi \oplus \dots\oplus E\otimes \chi^{m-1}$ is a spanning class of $\Db_G(X)$. Hence, $V=\Ind\reg_X$ is a relative spanning class of $\Db_G(X)$ over $\Dperf(Y)$.
Since $V$ is a vector bundle, so is $\sHom(V,V) = V\dual \otimes V$. The map $\pi$ is finite, hence $\pi_*$ is exact (does not need to be derived). Finally, taking $G$-invariants is exact because we work in characteristic 0. Altogether, $\pi_*^G\sHom(V,V)$ is a sheaf concentrated in degree 0.
\[Psilinear\] The functors $\Phi$ and $\Psi$, and for all $\alpha,\beta\in\IZ$ the subcategories $$\begin{aligned}
\Xi_\alpha(\Db(S)) &= a_*(\Db(S)) \otimes \chi^\alpha \subset \Db_G(X)
\quad\text{and}\\
\Theta_\beta(\Db(S)) &= i_*\nu^*\Db(S) \otimes \reg_\wY(\beta) \subset \Db(\wY)\end{aligned}$$ are $Y$-linear for $\pi^*\triv\colon \Dperf(Y)\to \Db_G(X)$ and $\rho^*\colon \Dperf(Y)\to \Db(\wY)$, respectively.
We | G ] = \reg_X \otimes (\chi^0\oplus\cdots\oplus\chi^{m-1})$ is a relative tilting sheaf for $ \Db_G(X)$ over $ \Dperf(Y)$.
If $ L\in\Pic(Y)\subset\Dperf(Y)$ is an ample line bundle, then therefore is $ \pi^*(L)$. therefore, $ \Db(X)$ has a generator of the form $ east \coloneqq \pi^*(\reg_Y\oplus L\oplus\cdots\oplus L^{\otimes k})$ for some $ k\gg0 $; witness [ @Orlov_gen ].
In particular, $ E$ is a spanning course of $ \Db(X)$. Using the adjunction $ \Res\dashv \Ind\dashv\Res$, it follows that $ \Ind(E) \cong E\oplus E\otimes \chi \oplus \dots\oplus E\otimes \chi^{m-1}$ is a cross course of $ \Db_G(X)$. Hence, $ V=\Ind\reg_X$ is a relative cross class of $ \Db_G(X)$ over $ \Dperf(Y)$.
Since $ V$ is a vector bundle, then is $ \sHom(V, five) = V\dual \otimes V$. The map $ \pi$ is finite, hence $ \pi_*$ is exact (does not need to be derived). ultimately, taking $ G$-invariants is exact because we work in characteristic 0. all in all, $ \pi_*^G\sHom(V, V)$ is a sheaf concentrated in degree 0.
\[Psilinear\ ] The functors $ \Phi$ and $ \Psi$, and for all $ \alpha,\beta\in\IZ$ the subcategories $ $ \begin{aligned }
\Xi_\alpha(\Db(S) ) & = a_*(\Db(S) ) \otimes \chi^\alpha \subset \Db_G(X)
\quad\text{and}\\
\Theta_\beta(\Db(S) ) & = i_*\nu^*\Db(S) \otimes \reg_\wY(\beta) \subset \Db(\wY)\end{aligned}$$ are $ Y$-linear for $ \pi^*\triv\colon \Dperf(Y)\to \Db_G(X)$ and $ \rho^*\colon \Dperf(Y)\to \Db(\wY)$, respectively.
We | G] = \geg_X \otimes (\chi^0\oplus\cdotr\oplus\chi^{m-1})$ is a relatite tiltjng sheaw for $\Db_G(X)$ over $\Dperf(Y)$.
If $L\in\'ic(Y)\wubseu\Bperf(Y)$ is an ample lkne bundlv, then so is $\ki^*(L)$. Hence, $\Db(X)$ has a generator or the horm $E \coloneqq \pi^*(\reg_Y\oplgs L\oplus\cdots\mpuud L^{\otimes k})$ for some $k\gg0$; see [@Orlov_gqn].
In pattlcular, $E$ is a fpanmyng dlass of $\Db(X)$. Using the adjunction $\Des\dashn \Ind\dashv\Res$, it gollows that $\Ind(E) \cong E\oppus F\otimes \chi \oplus \fots\oplus E\itimqw \chi^{m-1}$ is a rpanning class of $\Db_G(X)$. Hence, $V=\Ind\reg_X$ is a relative soanniug class of $\Dv_G(X)$ mver $\Dperf(Y)$.
Wince $V$ is a vector bundla, so is $\sHom(V,V) = V\dual \otmmes V$. The map $\pi$ is finive, hence $\pi_*$ is exact (does not nzed to be derived). Finqloy, taning $G$-inxqrixnta ms sxact heceuse we wori in characreristic 0. Altogethet, $\[p_*^B\sHom(V,V)$ is a sheaf cjncentrated in degree 0.
\[Psilinear\] The funbtora $\Phi$ and $\Psi$, and for aol $\alpha,\beta\in\IZ$ the dubcategowies $$\begin{aligned}
\Xi_\alpha(\Db(S)) &= a_*(\Db(S)) \otimes \chi^\alpva \suusdt \Bn_N(X)
\qkad\text{and}\\
\Theta_\beta(\Db(S)) &= i_*\nu^*\Db(S) \otimes \reg_\wY(\fsts) \xubset \Db(\wY)\end{cligned}$$ are $Y$-linrag gjr $\pi^*\triv\colov \Dpery(G)\tk \Db_G(X)$ and $\rho^*\coloj \Dperf(I)\to \Db(\qY)$, respecuivelu.
We | G] = \reg_X \otimes (\chi^0\oplus\cdots\oplus\chi^{m-1})$ is a sheaf $\Db_G(X)$ over If $L\in\Pic(Y)\subset\Dperf(Y)$ is so $\pi^*(L)$. Hence, $\Db(X)$ a generator of form $E \coloneqq \pi^*(\reg_Y\oplus L\oplus\cdots\oplus L^{\otimes for some $k\gg0$; see [@Orlov_gen]. In particular, $E$ is a spanning class of Using the adjunction $\Res\dashv \Ind\dashv\Res$, it follows that $\Ind(E) \cong E\oplus E\otimes \chi \dots\oplus \chi^{m-1}$ a class of $\Db_G(X)$. Hence, $V=\Ind\reg_X$ is a relative spanning class of $\Db_G(X)$ over $\Dperf(Y)$. Since $V$ a vector bundle, so is $\sHom(V,V) = V\dual V$. The map $\pi$ finite, hence $\pi_*$ is exact not to be Finally, $G$-invariants exact because we in characteristic 0. Altogether, $\pi_*^G\sHom(V,V)$ is a sheaf concentrated in degree 0. \[Psilinear\] The functors $\Phi$ and and for the subcategories \Xi_\alpha(\Db(S)) a_*(\Db(S)) \chi^\alpha \subset \Db_G(X) &= i_*\nu^*\Db(S) \otimes \reg_\wY(\beta) \subset \Db(\wY)\end{aligned}$$ $\pi^*\triv\colon \Dperf(Y)\to \Db_G(X)$ and $\rho^*\colon \Dperf(Y)\to \Db(\wY)$, respectively. | G] = \reg_X \otimes (\chi^0\oplus\cdots\oPlus\chi^{m-1})$ is A relaTivE tiLtIng sHeaf For $\Db_G(X)$ over $\DpeRF(Y)$.
If $l\in\Pic(Y)\subset\Dperf(Y)$ is aN amplE lINe buNDlE, then So is $\pi^*(L)$. hEnCE, $\db(X)$ HaS a GenErAToR of thE foRm $E \coloNeqq \pi^*(\reg_Y\OplUs l\oplus\cdots\oPLuS L^{\otimes k})$ fOr sOme $k\gg0$; see [@OrlOv_gEn].
In paRtIcuLAr, $E$ is A spAnninG class OF $\Db(X)$. UsIng the adjUnCTion $\ReS\Dashv \InD\DAsHv\ReS$, it follows that $\Ind(e) \CoNG E\oplus E\otimes \Chi \oplUs \DOtS\OPluS E\oTimes \chi^{m-1}$ iS a SpannINg class OF $\DB_g(x)$. henCE, $V=\Ind\reg_X$ is a rElative spanNIng Class oF $\DB_G(X)$ OVer $\DpeRf(Y)$.
SiNcE $v$ is A vector bundLe, so Is $\sHom(V,V) = V\Dual \otIMes V$. The MAp $\pi$ is fInite, hEncE $\pi_*$ Is exACt (DoEs nOt NEed TO bE deRIveD). Finally, TaKiNg $G$-inVariANTS Is exAct BecaUse we Work in charactEriStic 0. aLtoGetheR, $\pi_*^G\shom(V,v)$ iS a sheAf concEntraTeD in degree 0.
\[PsilinEar\] THe functorS $\PhI$ aNd $\PSi$, And foR All $\alpHa,\bEta\In\IZ$ the SubcateGOriEs $$\BEGIn{Aligned}
\Xi_\alpha(\Db(S)) &= a_*(\db(s)) \OTiMes \chi^\alPha \subSEt \db_g(x)
\quad\texT{aNd}\\
\THeta_\BETa(\Db(S)) &= I_*\nu^*\DB(s) \oTimes \reg_\WY(\beta) \SUbSeT \Db(\wY)\enD{aLigned}$$ ArE $Y$-lIneAr for $\PI^*\triV\colon \dperf(Y)\to \db_G(X)$ aND $\rho^*\colon \Dperf(y)\To \Db(\wY)$, respectIVeLY.
we | G] = \reg_X \otimes (\chi^ 0\oplus\cd ots\o plu s\c hi ^{m- 1})$ is a relative tilt ing sheaf for $\Db_G(X )$ ov er $\Dp e rf (Y)$.
If $L \ in \ P ic( Y) \s ubs et \ Dp erf(Y )$is an a mple linebun dl e, then so i s $ \pi^*(L)$. He nce, $\Db(X) $ h as a g en era t or of th e for m $E \ c oloneq q \pi^*(\ re g _Y\opl u s L\opl u s \c dots \oplus L^{\otimes k} ) $ for some $k\ gg0$;se e [ @ O rlo v_g en].
In p ar ticul a r, $E$i sa s pan n ing class of$\Db(X)$. U s ing the a dj unc t ion $\ Res\d as h v \ Ind\dashv\R es$, it follo ws tha t $\Ind( E ) \cong E\opl usE\o time s \ ch i \ op l us\ do ts\ o plu s E\otim es \ chi^{ m-1} $ i s a s pan ning clas s of $\Db_G(X )$. Hen c e,$V=\I nd\re g_X$ i s a r elativ e spa nn ing class of $\ Db_G (X)$ over $\ Dp erf (Y )$.
S ince $ V$isa vecto r bundl e , s oi s $\ sHom(V,V) = V\dual \ o t im es V$. T he map $\ pi $ is fini te , h ence $ \pi_* $ is ex act (doe s notn ee dto be d er ived). F ina lly , tak i ng $ G$-inv ariantsis ex a ct because wew ork in charac t er i s ti c 0.Alt ogether, $\ pi_* ^ G\sH om(V , V) $ i s a sh eaf c on c en t rated in degree 0.
\ [Psili near\ ] The functor s $\Phi$ a n d $\Psi$,andf or all $\alpha,\b eta\i n\IZ$ thes ubcatego ries$$\begin {aligned} \Xi_\al pha (\D b(S ))& = a _*(\Db(S)) \o t i mes\c hi^\alp ha\subset \D b_G (X)
\quad\tex t{and}\\
\ Th et a_\ beta( \ Db(S)) & =i_* \n u^* \Db(S ) \otim es \r eg_\ wY (\ b eta ) \subs e t\ D b(\w Y) \e nd{a lig ne d}$$are$ Y$- linearfor $\pi^ *\t r iv\c ol on \Dperf (Y)\to \Db_G( X) $ and $\rh o^ *\c olon \ D p erf(Y)\t o \Db(\wY)$, respective l y.
We | G] =_\reg_X \otimes_(\chi^0\oplus\cdots\oplus\chi^{m-1})$ is a relative_tilting sheaf_for_$\Db_G(X)$ over_$\Dperf(Y)$.
If_$L\in\Pic(Y)\subset\Dperf(Y)$ is an_ample line bundle,_then so is $\pi^*(L)$._Hence, $\Db(X)$ has_a_generator of the form $E \coloneqq \pi^*(\reg_Y\oplus L\oplus\cdots\oplus L^{\otimes k})$ for some $k\gg0$; see_[@Orlov_gen].
In_particular, $E$_is_a_spanning class of $\Db(X)$. Using_the adjunction $\Res\dashv \Ind\dashv\Res$, it_follows that_$\Ind(E) \cong E\oplus E\otimes \chi \oplus \dots\oplus E\otimes_\chi^{m-1}$_is a spanning_class of $\Db_G(X)$. Hence, $V=\Ind\reg_X$ is a relative spanning_class of $\Db_G(X)$ over $\Dperf(Y)$.
Since $V$_is a vector_bundle,_so_is $\sHom(V,V) = V\dual_\otimes V$. The map $\pi$ is_finite, hence $\pi_*$ is exact (does_not need to be derived). Finally, taking_$G$-invariants is exact because we work_in characteristic 0. Altogether, $\pi_*^G\sHom(V,V)$_is a_sheaf concentrated in degree 0.
\[Psilinear\]_The functors $\Phi$_and $\Psi$,_and for all_$\alpha,\beta\in\IZ$ the subcategories $$\begin{aligned}
\Xi_\alpha(\Db(S))_&= a_*(\Db(S)) \otimes_\chi^\alpha \subset \Db_G(X)
_\quad\text{and}\\
_ \Theta_\beta(\Db(S)) &=_i_*\nu^*\Db(S)_\otimes_\reg_\wY(\beta) \subset_\Db(\wY)\end{aligned}$$ are $Y$-linear_for_$\pi^*\triv\colon \Dperf(Y)\to_\Db_G(X)$_and $\rho^*\colon \Dperf(Y)\to \Db(\wY)$, respectively.
We |
is only possible via sophisticated high-resolution hydrodynamical simulations such as those presented in the *Sejong Suite*. In fact, at small scales the Ly$\alpha$ flux distribution depends on the complex IGM spatial distribution, peculiar velocity field, and thermal properties of the gas, making the connection with the total matter power spectrum $P^{\rm 3D}_{\rm t}$ rather complex. In this view, a detailed knowledge of the gas-to-matter and peculiar velocity biases, of the nature of the ionizing background radiation, and on the fluctuations in the temperature-density relation is necessary.
In previous works, using a refined technique that involves a model based on a Taylor expansion of the flux (evaluated numerically), we have already successfully used the information contained in $P^{\rm 1D}_{{ \cal{F}}}$ to obtain among the strongest individual and joint bounds on neutrino masses and $N_{\rm eff}$, exploiting the Ly$\alpha$ forest in synergy with the CMB and low-$z$: perfecting this technique using new simulations from the *Sejong Suite* is the subject of ongoing work.
To this end, Figure \[fig\_1d\_flux\_ps\_data\_16D\_paper\] displays a few BG 1D flux power spectra as derived from our [*Grid Suite*]{} (i.e., Table \[table\_grid\_sims\_base\]), confronted with recent $P^{\rm 1D}_{{ \cal{F}}}$ measurements obtained from the SDSS-III BOSS survey and from the VLT/XSHOOTER legacy survey (XQ-100; L[ó]{}pez et al. 2016) – the latter is observed with the X-Shooter spectrograph on the Very Large Telescope (Vernet et al. 2011). In the plot, green triangles refer to the observational measurements reported by Irsic et al. (2017b), while orange circles indicate those of Yeche et al. (2017). Three redshift intervals are considered: namely, $z=3.2$ (top panel), $z=3.6$ (middle panel), and $z=4.0$ (bottom panel), respectively. In all of the panels, black solid lines display the BG [*spliced*]{} flux power spectrum – corresponding to the effective resolution of a single simulation characterized by $2 \times 3328^3 \sim 74$ billion particles in a $(100 h^{-1} {\rm Mpc})^3 | is only possible via sophisticated high - settlement hydrodynamical model such as those presented in the * Sejong Suite *. In fact, at small scales the Ly$\alpha$ flux density distribution depends on the complex IGM spatial distribution, peculiar speed field, and thermal properties of the gas, make the connection with the total topic power spectrum $ P^{\rm 3D}_{\rm t}$ rather complex. In this view, a detailed knowledge of the accelerator - to - matter and peculiar velocity biases, of the nature of the ionizing setting radiation, and on the fluctuations in the temperature - density relation back is necessary.
In previous works, using a polished technique that involves a model based on a Taylor expansion of the flux (measure numerically), we have already successfully used the information contained in $ P^{\rm 1D } _ { { \cal{F}}}$ to obtain among the strongest individual and joint bounds on neutrino masses and $ N_{\rm eff}$, exploiting the Ly$\alpha$ forest in synergy with the CMB and low-$z$: perfecting this technique using new simulations from the * Sejong Suite * is the subject of ongoing work.
To this end, Figure \[fig\_1d\_flux\_ps\_data\_16D\_paper\ ] expose a few BG 1D flux power spectra as derive from our [ * Grid Suite * ] { } (i.e., Table \[table\_grid\_sims\_base\ ]), confront with recent $ P^{\rm 1D } _ { { \cal{F}}}$ measurements prevail from the SDSS - III BOSS survey and from the VLT / XSHOOTER legacy survey (XQ-100; L[ó]{}pez et al. 2016) – the latter is watch with the X - Shooter spectrograph on the Very Large Telescope (Vernet et al. 2011). In the plot, green triangles refer to the observational measurements report by Irsic et al. (2017b), while orange circles indicate those of Yeche et al. (2017). Three redshift intervals are consider: namely, $ z=3.2 $ (top dialog box), $ z=3.6 $ (middle panel), and $ z=4.0 $ (bottom panel), respectively. In all of the dialog box, black solid lines expose the BG [ * spliced * ] { } flux office spectrum – corresponding to the effective resolution of a unmarried simulation characterized by $ 2 \times 3328 ^ 3 \sim 74 $ billion atom in a $ (100 h^{-1 } { \rm Mpc})^3 | is only possible via sophirticated high-resolutioi hydrosynamicau simulations such as those 'reswnted in the *Sejong Suite*. Iv fact, at small sxalew the Ly$\alpiz$ flux distrigmtion vepends on the gomplex IGM spatial distrhbjtnon, peculiar velocity field, and therial prolegties of the gws, mshing nht connection with the total matted power spectrum $P^{\rm 3D}_{\rm t}$ rather complex. In tjis giew, a detailed knlwledge of jge dqs-to-matter avd peculiag velocity giases, of the nature of the ionkzing backgroune eadldtion, and oi the sluctuations in the demperayure-density rekatmon us necessary.
In previons works, using a refyned techtisue that involves a moden baved ub a Taglpr expandioi of the flhx (evaluatee numerically), we hafe qlready succeasfullr tsed the information contained in $P^{\rm 1D}_{{ \cam{F}}}$ to obtain among the wtrongest individual wnd joint bounds on neutrino masses and $N_{\rm eff}$, exploiting the Mh$\alkhq$ fordwt in synergy with the CMB and low-$z$: perfecting egix nechnique using ncw simulations frok hhr *Sejong Suite* is thz shbject of ongoing aork.
To jhis ebd, Figure \[fig\_1c\_flux\_ps\_data\_16D\_paper\] displays q few BG 1D fjyx power spectra ad derived fxom out [*Grid Suite*]{} (i.e., Table \[table\_grnd\_sims\_gase\]), confrojted with fecent $P^{\rm 1D}_{{ \cal{W}}}$ mvasusements obtained from the FDSS-III BISS xurvey xnd grom tre VLT/XSHOLTER legacy survey (XQ-100; L[ó]{}pfz et an. 2016) – the lahter is observed with the X-Shoovxr spectrograkh mn nhe Very Jarge Telescope (Vewnet et al. 2011). In the pljt, grden triangmes refxr to the obfervational mawsurements rxported br Irwic wt al. (2017b), dhile orange corcles inbncate thowe of Yeche et al. (2017). Thtes redshift intexraos are considerrd: vamqlj, $z=3.2$ (tj[ panel), $z=3.6$ (migdle pavrl), ana $z=4.0$ (botujm oanek), respectively. In aln of the panels, black xojid linew displar the BG [*splived*]{} flux power speftrum – curreslonqing to the effective resolutikn of a slngje simulatiog chqracterized yy $2 \times 3328^3 \sim 74$ billion particles in a $(100 h^{-1} {\rm Mpc})^3 | is only possible via sophisticated high-resolution hydrodynamical as presented in *Sejong Suite*. In Ly$\alpha$ distribution depends on complex IGM spatial peculiar velocity field, and thermal properties the gas, making the connection with the total matter power spectrum $P^{\rm 3D}_{\rm rather complex. In this view, a detailed knowledge of the gas-to-matter and peculiar biases, the of ionizing background radiation, and on the fluctuations in the temperature-density relation is necessary. In previous works, a refined technique that involves a model based a Taylor expansion of flux (evaluated numerically), we have successfully the information in 1D}_{{ to obtain among strongest individual and joint bounds on neutrino masses and $N_{\rm eff}$, exploiting the Ly$\alpha$ forest in synergy the CMB perfecting this using simulations the *Sejong Suite* subject of ongoing work. To this displays a few BG 1D flux power spectra derived from [*Grid Suite*]{} (i.e., Table \[table\_grid\_sims\_base\]), confronted recent $P^{\rm 1D}_{{ \cal{F}}}$ measurements obtained from the BOSS survey and from the VLT/XSHOOTER legacy survey (XQ-100; L[ó]{}pez et al. 2016) – the observed with the X-Shooter on the Very Telescope et 2011). the plot, triangles refer to the observational measurements reported by Irsic et al. while orange circles indicate those of Yeche et al. (2017). intervals considered: namely, $z=3.2$ panel), $z=3.6$ (middle panel), $z=4.0$ panel), respectively. In all panels, solid BG flux spectrum – corresponding to effective resolution of a single characterized by $2 \times in a $(100 h^{-1} {\rm Mpc})^3 | is only possible via sophistiCated high-rEsoluTioN hyDrOdynAmicAl simulations sUCh as Those presented in the *SejOng SuItE*. in faCT, aT smalL scales THe lY$\AlpHa$ FlUx dIsTRiButioN dePends on The complex iGM SpAtial distribUTiOn, peculiar VelOcity field, anD thErmal pRoPerTIes of The Gas, maKing thE ConnecTion with tHe TOtal maTTer poweR SPeCtruM $P^{\rm 3D}_{\rm t}$ rather comPLeX. in this view, a detAiled kNoWLeDGE of The Gas-to-matteR aNd pecULiar velOCiTY BIasES, of the nature oF the ionizinG BacKgrounD rAdiATion, anD on thE fLUctUations in thE temPerature-dEnsity RElation IS necessAry.
In pRevIouS worKS, uSiNg a ReFIneD TeChnIQue That invoLvEs A modeL basED ON A TayLor ExpaNsion Of the flux (evalUatEd nuMEriCally), We havE alrEaDy sucCessfuLly usEd The information cOntaIned in $P^{\rm 1d}_{{ \caL{F}}}$ To oBtAin amONg the sTroNgeSt indivIdual anD JoiNt BOUNdS on neutrino masses aNd $n_{\RM eFf}$, exploiTing thE ly$\AlPHa$ forest In SynErgy WITh the cMB aND lOw-$z$: perfeCting tHIs TeChnique UsIng new SiMulAtiOns frOM the *sejong suite* is tHe subJEct of ongoing woRK.
To this end, FigURe \[FIG\_1d\_FLux\_pS\_daTa\_16D\_paper\] disPlayS A few bG 1D fLUx PowER specTra as DeRIvED from our [*Grid Suite*]{} (i.e., taBle \[tabLe\_griD\_sims\_base\]), confRonted with RECEnt $P^{\rm 1D}_{{ \cAl{F}}}$ mEAsURements obtaineD from The SDSS-III boSS surveY and fRom the VLt/XSHOOTER LEGacy survEy (Xq-100; L[ó]{}Pez Et aL. 2016) – THe Latter is obserVED witH tHe X-ShooTer SpectroGraPh oN thE VeRy large TeleScope (VerNeT eT aL. 2011). IN thE plot, GReen triaNgLes ReFer To the OBservaTionaL meaSuReMEntS reportED bY iRsic Et Al. (2017B), whiLe oRaNge ciRcleS IndIcate thOse of YechE et AL. (2017). ThrEe ReDshift iNtervals are coNsIdered: nameLy, $Z=3.2$ (toP panel), $Z=3.6$ (MIddle panEl), and $z=4.0$ (bottom panel), respecTIvely. In All Of the PaneLs, black soLid Lines dIspLAy the Bg [*splicEd*]{} fluX pOweR SPectrUM – CoRreSpOnding to thE EFfeCtive ReSoluTion of a Single simulation chARacTerized by $2 \timeS 3328^3 \siM 74$ bilLIOn ParTIcLEs iN a $(100 H^{-1} {\Rm MPC})^3 | is only possible via soph isticatedhigh- res olu ti on h ydro dynamical simu l atio ns such as those prese ntedin the* Se jongSuite*. In f act ,at sm al l s cales th e Ly$\a lpha$ flux di st ribution dep e nd s on the c omp lex IGM spat ial distr ib uti o n, pe cul iar v elocit y field , and the rm a l prop e rties o f th e ga s, making the con n ec t ion with the t otal m at t er p owe r s pectrum $P ^{ \rm 3 D }_{\rmt }$ r a the r complex. Inthis view,a de tailed k now l edge o f the g a s-t o-matter an d pe culiar ve locity biases, of thenature of th e io n iz in g b ac k gro u nd ra d iat ion, and o nthe f luct u a t i onsinthetempe rature-densit y r elat i onis ne cessa ry.
I n pre viousworks ,using a refined tec hnique th atin vol ve s a m o del ba sed on a Tayl or expa n sio no f th e flux (evaluatednu m e ri cally),we hav e a lr e ady succ es sfu llyu s ed th e in f or mation c ontain e din $P^{\r m1D}_{{ \ cal {F} }}$ t o obt ain am ong thestron g est individual and joint bou n ds o nn eutr ino masses and $N_ { \rmeff} $ ,exp l oitin g the L y $\ a lpha$ forest in syn er gy wit h the CMB and low- $z$: perfe c t i ng thistech n iq u e using new si mulat ions fromt he *Sejo ng Su ite* isthe subje c t of ongo ing wo rk.
T o th is end, Figur e \[fi g\ _1d\_fl ux\ _ps\_da ta\ _16 D\_ pap er \] displa ys a few B G1D f lux powe r spectra a s d er ive d fro m our [ *Grid Sui te *] { } ( i.e., T a bl e \[ta bl e\ _gri d\_ si ms\_b ase\ ] ),confron ted withrec e nt $ P^ {\ rm 1D}_ {{ \cal{F}}}$ m easurement sobt ainedf r om the S DSS-III BOSS survey and from th e V LT/XS HOOT ER legacy su rvey ( XQ- 1 00; L[ ó]{}pe z etal . 2 0 1 6) –t h elat te r is obser v e d w ith t he X-S hooterspectrograph on th e Ve ry Large Tele sco pe ( V e rn ete ta l.20 1 1). I n the plot, gre en triangl es re fer to the obs er vationa l measu remen t s repor ted by Ir sic et al .(201 7 b ),while oran ge circl es indica t e tho s eof Ye che et al .(20 17).Threer eds hiftinterv al s areconsi de red: nam ely, $z=3.2$ (top panel ), $z= 3.6$(mi ddle pane l), and $z=4.0$(bot tom panel) , r esp ectiv ely . In a ll o f t hep anels , bl a ck solidl in esd i sp lay the BG[ * s pli ced*] {}f lux po werspectrum – corres p onding to theeffe c t ive re s olut io n of a singlesim ul a t ion char ac terized by$2 \time s3328^ 3 \sim 74$ b illionp a rt i cles i n a$(1 00 h^{-1} {\ rm Mpc})^3 | is_only possible_via sophisticated high-resolution hydrodynamical_simulations such_as_those presented_in_the *Sejong Suite*._In fact, at_small scales the Ly$\alpha$_flux distribution depends_on_the complex IGM spatial distribution, peculiar velocity field, and thermal properties of the gas,_making_the connection_with_the_total matter power spectrum $P^{\rm_3D}_{\rm t}$ rather complex. In_this view,_a detailed knowledge of the gas-to-matter and peculiar_velocity_biases, of the_nature of the ionizing background radiation, and on the_fluctuations in the temperature-density relation is_necessary.
In previous works,_using_a_refined technique that involves_a model based on a Taylor_expansion of the flux (evaluated numerically),_we have already successfully used the information_contained in $P^{\rm 1D}_{{ \cal{F}}}$ to_obtain among the strongest individual_and joint_bounds on neutrino masses and_$N_{\rm eff}$, exploiting_the Ly$\alpha$_forest in synergy_with the CMB and low-$z$: perfecting_this technique using_new simulations from the *Sejong Suite*_is_the subject of_ongoing_work.
To_this end,_Figure \[fig\_1d\_flux\_ps\_data\_16D\_paper\] displays_a_few BG_1D_flux power spectra as derived from_our_[*Grid Suite*]{} (i.e., Table \[table\_grid\_sims\_base\]), confronted with_recent $P^{\rm 1D}_{{ \cal{F}}}$_measurements_obtained from the SDSS-III_BOSS survey and from the_VLT/XSHOOTER legacy survey (XQ-100; L[ó]{}pez et_al. 2016)_– the_latter is observed with the X-Shooter spectrograph on the Very Large_Telescope (Vernet et al. 2011). In_the plot, green triangles_refer to_the_observational measurements reported_by_Irsic et_al. (2017b), while orange circles indicate those_of Yeche_et al. (2017). Three redshift intervals_are considered: namely, $z=3.2$_(top_panel), $z=3.6$ (middle panel), and $z=4.0$_(bottom panel), respectively. In all of_the panels, black solid lines_display_the_BG [*spliced*]{} flux power spectrum_– corresponding to the effective resolution_of a single_simulation characterized by $2 \times 3328^3_\sim_74$ billion particles in a $(100_h^{-1}_{\rm Mpc})^3 |
Bmarionps"}](coerc_marion_ps.pdf "fig:"){height="65mm"}\
![Plots of the coercivity constants (top) and condition numbers (bottom) of the operators $PS\mathcal{B}_{k,2,k,0.4k^{1/3}}$ for 5041 values of the wavenumber $k$ from $k=8$ to $k=512$.[]{data-label="fig:Bmarionps"}](cond_marion_ps.pdf "fig:"){height="65mm"}
High-frequency numerical experiments {#num_exp}
====================================
We present in this section a variety of numerical results that demonstrate the important computational savings that can be garnered from use of regularized combined field integral equations in the high-frequency regime. Most importantly, we showcase the superior performance of solvers based on the novel Calderón-Complex CFIER formulations that involve the boundary integral operators $$\label{eq:CFIERc}
\mathcal{B}_{k,2,k,0.4\mathcal{H}^{2/3}k^{1/3}}=\frac{I}{2}-\mathcal{K}_k-2\mathcal{T}_k\ \mathcal{T}_{k+i\ 0.4\mathcal{H}^{2/3}k^{1/3}},$$ where $\mathcal{H}$ is the maximum of the absolute values of mean curvatures on surface $\Gamma$. The latter choice is motivated by the fact that $c=R^{-2/3}$ leads to operators $\mathcal{B}_{k,2,k,0.4ck^{1/3}}$ with nearly optimal spectral properties in the high-frequency regime in the case when $\Gamma$ is a sphere of radius $R$ (see Section \[DtN\]). For general smooth surfaces $\Gamma$, we followed the heuristical practice of replacing $R^{-1}$ by $\mathcal{H}$.
In addition, we illustrate the performance of solvers based on Calderón-Ikawa CFIER formulations at high frequencies. This type of formulations has been shown to work effectively in the low and medium frequency regime [@Contopa_et_al; @turc1; @Andriulli1; @Andriulli]. We focus on the arguably simplest version of Calderón CFIER formulations that involve the following boundary integral operators [@turc1]: $$\ | Bmarionps"}](coerc_marion_ps.pdf " fig:"){height="65mm"}\
! [ Plots of the coercivity constants (top) and condition number (bottomland) of the operators $ PS\mathcal{B}_{k,2,k,0.4k^{1/3}}$ for 5041 values of the wavenumber $ k$ from $ k=8 $ to $ k=512$.[]{data - label="fig: Bmarionps"}](cond_marion_ps.pdf " fig:"){height="65 mm " }
High - frequency numeral experiments { # num_exp }
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
We present in this part a variety of numerical consequence that demonstrate the important computational preservation that can be garnered from use of regulate combined field integral equations in the high - frequency government. Most importantly, we showcase the superior performance of problem solver based on the novel Calderón - Complex CFIER formulations that imply the boundary integral operator $ $ \label{eq: CFIERc }
\mathcal{B}_{k,2,k,0.4\mathcal{H}^{2/3}k^{1/3}}=\frac{I}{2}-\mathcal{K}_k-2\mathcal{T}_k\ \mathcal{T}_{k+i\ 0.4\mathcal{H}^{2/3}k^{1/3}},$$ where $ \mathcal{H}$ is the maximum of the absolute values of mean curvatures on surface $ \Gamma$. The latter option is motivated by the fact that $ c = R^{-2/3}$ leads to operators $ \mathcal{B}_{k,2,k,0.4ck^{1/3}}$ with nearly optimal apparitional property in the high - frequency regime in the case when $ \Gamma$ is a sphere of radius $ R$ (see Section \[DtN\ ]). For general smooth surface $ \Gamma$, we followed the heuristical practice of substitute $ R^{-1}$ by $ \mathcal{H}$.
In accession, we illustrate the performance of solvers free-base on Calderón - Ikawa CFIER formulations at high frequencies. This type of formulation has been shown to work effectively in the low and medium frequency regime [ @Contopa_et_al; @turc1; @Andriulli1; @Andriulli ]. We concenter on the arguably simplest version of Calderón CFIER formulations that involve the play along boundary integral operator [ @turc1 ]: $ $ \ | Bmagionps"}](coerc_marion_ps.pdf "flg:"){height="65mm"}\
![Plots of the roercivjty consgants (top) and condition numbxrs (votton) of the operators $PS\mxthcal{B}_{k,2,k,0.4n^{1/3}}$ for 5041 vqluew of the watsnumber $k$ froj $k=8$ tm $k=512$.[]{data-label="fig:Nmarionps"}](cotd_marion_ps.pdf "xie:"){hzight="65mm"}
High-frequency numerical experyments {#mul_exp}
====================================
We present in uhif sedniin a variety of numerical reshlts thet demonstrate yhe important computationap sagings that can be harnered frim ufw of regularkzed combined field injegral equations in the high-freqjency regime. Mowt imomrtantly, we showbase the supevpor perxormancr of solvers bssev on the novel Calderón-Com'lex CFIER formulatijns that hnrolve the boundary inretral mperdtorr $$\lacel{tq:CHIEDc}
\mathfal{U}_{k,2,k,0.4\mathcal{H}^{2/3}i^{1/3}}=\frac{I}{2}-\mathcql{K}_k-2\mathcal{T}_k\ \mathvaj{N}_{l+i\ 0.4\mathcal{H}^{2/3}k^{1/3}},$$ where $\mwthcal{H}$ is the maximum of the absolute naluss of mean curvatures ob surface $\Gamma$. The lwtter choyce is motivated by the fact that $c=R^{-2/3}$ leads to opesatora $\mauhgwu{V}_{k,2,n,0.4ck^{1/3}}$ with nearly optimal spectral properties ig tne high-frequency regime in yhf vwse when $\Gammx$ is a spgere of radius $R$ (sfe Sectyon \[DtB\]). For gentral xmooth surfaces $\Gamma$, we foolowed the hvuriwtical practice of replacing $X^{-1}$ by $\msthcak{H}$.
In addition, we illustxate tge performajce of somxers based on Caudegón-Ikdwa CFIER formulations at righ freqnencizs. This gype of fowmulations has naen shown to work fffecjively in the loa and medium frequency regime [@Coivopa_et_al; @turc1; @Atdrpulli1; @Andxiulli]. We focus on ehe arguably snmplest rersiov of Caldegón CFIER hormulations that involve jhe following boundarr inregrql operxgors [@turc1]: $$\ | Bmarionps"}](coerc_marion_ps.pdf "fig:"){height="65mm"}\ ![Plots of the coercivity constants condition (bottom) of operators $PS\mathcal{B}_{k,2,k,0.4k^{1/3}}$ for $k$ $k=8$ to $k=512$.[]{data-label="fig:Bmarionps"}](cond_marion_ps.pdf High-frequency numerical experiments ==================================== We present in this section variety of numerical results that demonstrate the important computational savings that can be from use of regularized combined field integral equations in the high-frequency regime. Most we the performance solvers based on the novel Calderón-Complex CFIER formulations that involve the boundary integral operators $$\label{eq:CFIERc} \mathcal{B}_{k,2,k,0.4\mathcal{H}^{2/3}k^{1/3}}=\frac{I}{2}-\mathcal{K}_k-2\mathcal{T}_k\ 0.4\mathcal{H}^{2/3}k^{1/3}},$$ where $\mathcal{H}$ is the maximum of the values of mean curvatures surface $\Gamma$. The latter choice motivated the fact $c=R^{-2/3}$ to $\mathcal{B}_{k,2,k,0.4ck^{1/3}}$ with nearly spectral properties in the high-frequency regime in the case when $\Gamma$ is a sphere of radius $R$ Section \[DtN\]). smooth surfaces we the practice of replacing $\mathcal{H}$. In addition, we illustrate the based on Calderón-Ikawa CFIER formulations at high frequencies. type of has been shown to work effectively the low and medium frequency regime [@Contopa_et_al; @turc1; @Andriulli]. We focus on the arguably simplest version of Calderón CFIER formulations that involve the integral operators [@turc1]: $$\ | Bmarionps"}](coerc_marion_ps.pdf "Fig:"){height="65mM"}\
![PlotS of The CoErciVity Constants (top) anD CondItion numbers (bottom) of thE operAtORs $PS\MAtHcal{B}_{K,2,k,0.4k^{1/3}}$ for 5041 vALuES Of tHe WaVenUmBEr $K$ from $K=8$ to $K=512$.[]{data-laBel="fig:BmarIonPs"}](Cond_marion_ps.PDf "Fig:"){height="65mM"}
HiGh-frequency nUmeRical eXpEriMEnts {#nUm_eXp}
====================================
We pResent IN this sEction a vaRiETy of nuMErical rESUlTs thAt demonstrate the iMPoRTant computatioNal savInGS tHAT caN be Garnered frOm Use of REgulariZEd COMBinED field integraL equations iN The High-frEqUenCY regimE. Most ImPOrtAntly, we showCase The superiOr perfORmance oF Solvers Based oN thE noVel CALdErÓn-COmPLex cfIeR fORmuLations tHaT iNvolvE the BOUNDary IntEgraL operAtors $$\label{eq:CfIErc}
\maTHcaL{B}_{k,2,k,0.4\mAthcaL{H}^{2/3}k^{1/3}}=\fRaC{I}{2}-\matHcal{K}_k-2\MathcAl{t}_k\ \mathcal{T}_{k+i\ 0.4\matHcal{h}^{2/3}k^{1/3}},$$ where $\maThcAl{h}$ is ThE maxiMUm of thE abSolUte valuEs of meaN CurVaTUREs On surface $\Gamma$. The lAtTER cHoice is mOtivatED bY tHE fact thaT $c=r^{-2/3}$ leAds tO OPeratOrs $\mAThCal{B}_{k,2,k,0.4ck^{1/3}}$ With neARlY oPtimal sPeCtral pRoPerTieS in thE High-FrequeNcy regimE in thE Case when $\Gamma$ iS A sphere of radiUS $R$ (SEE SECtioN \[Dtn\]). For general SmooTH surFaceS $\gaMma$, WE follOwed tHe HEuRIstical practice of rePlAcing $R^{-1}$ By $\matHcal{H}$.
In additiOn, we illustRATE the perfOrmaNCe OF solvers based oN CaldErón-Ikawa CfiER formuLatioNs at high FrequenciES. this type Of fOrmUlaTioNS HaS been shown to wORK effEcTively iN thE low and MedIum FreQueNcY regime [@CoNtopa_et_aL; @tUrC1; @ANdRiuLli1; @AnDRiulli]. We FoCus On The ArguaBLy simpLest vErsiOn Of cAldErón CFIer fORMulaTiOnS thaT inVoLve thE folLOwiNg boundAry integrAl oPEratOrS [@tUrc1]: $$\ | Bmarionps"}](coerc_marion_ ps.pdf "fi g:"){ hei ght =" 65mm "}\![Plots of the coer civity constants (top) andco n diti o nnumbe rs (bot t om ) ofth eope ra t or s $PS \ma thcal{B }_{k,2,k,0 .4k ^{ 1/3}}$ for 5 0 41 values of th e wavenumber $k $ from $ k=8 $ to $ k=5 12$.[ ]{data - label= "fig:Bmar io n ps"}]( c ond_mar i o n_ ps.p df "fig:"){height = "6 5 mm"}
High-fre quency n u me r i cal ex periments{# num_e x p}
==== = == = = = === = ============= =========
W e p resent i n t h is sec tionav ari ety of nume rica l results thatd emonstr a te theimport ant co mput a ti on alsa v ing s t hat can be garn er ed from use o f regu lar ized comb ined field in teg rale qua tions in t he h ig h-fre quency regi me . Most importan tly, we showc ase t hesu perio r perfo rma nce of sol vers ba s edon t h enovel Calderón-Com pl e x C FIER for mulati o ns t h at invol ve th e bo u n daryinte g ra l operat ors $$ \ la be l{eq:CF IE Rc}
\m at hca l{B }_{k, 2 ,k,0 .4\mat hcal{H}^ {2/3} k ^{1/3}}=\frac{ I }{2}-\mathcal { K} _ k -2 \ math cal {T}_k\ \ma thca l {T}_ {k+i \ 0 .4\ m athca l{H}^ {2 / 3} k ^{1/3}},$$ where $\ ma thcal{ H}$ i s the maximum of the ab s o l ute valu es o f m e an curvatureson su rface $\Ga m ma$. The latt er choic e is moti v a ted by t hefac t t hat $ c= R^{-2/3}$ lea d s toop erators $\ mathcal {B} _{k ,2, k,0 .4 ck^{1/3}} $ with n ea rl yop tim al sp e ctral pr op ert ie s i n the high-f reque ncyre gi m e i n the c a se w hen$\ Ga mma$ is a sphe re o f ra dius $R $ (see Se cti o n \[ Dt N\ ]). For general smoo th surfaces$\ Gam ma$, w e followed the heuristical practi c e of re pla cing$R^{ -1}$ by $ \ma thcal{ H}$ .
In a dditio n, we i llu s t ratet h eper fo rmance ofs o lve rs ba se d on Calder ón-Ikawa CFIER for m ula tions at high fr eque n c ie s.T hi s ty pe off o rmulations hasbeen shown t o w ork effect i vel yin thelow and medi u m frequ ency regi me [@Cont op a_et _ a l;@turc1; @A ndriulli 1; @Andri u lli]. We focu s o n thear gua bly s imples t ve rsion of Ca ld erón C FIERfo rmulatio ns that involve the fol lowing boun dar y integra l o p era tors [@tu rc1] : $$\ | Bmarionps"}](coerc_marion_ps.pdf "fig:"){height="65mm"}\
![Plots_of the_coercivity constants (top) and_condition numbers_(bottom)_of the_operators_$PS\mathcal{B}_{k,2,k,0.4k^{1/3}}$ for 5041_values of the_wavenumber $k$ from $k=8$_to $k=512$.[]{data-label="fig:Bmarionps"}](cond_marion_ps.pdf "fig:"){height="65mm"}
High-frequency_numerical_experiments {#num_exp}
====================================
We present in this section a variety of numerical results that demonstrate the_important_computational savings_that_can_be garnered from use of_regularized combined field integral equations_in the_high-frequency regime. Most importantly, we showcase the superior_performance_of solvers based_on the novel Calderón-Complex CFIER formulations that involve the_boundary integral operators $$\label{eq:CFIERc}
\mathcal{B}_{k,2,k,0.4\mathcal{H}^{2/3}k^{1/3}}=\frac{I}{2}-\mathcal{K}_k-2\mathcal{T}_k\ \mathcal{T}_{k+i\_0.4\mathcal{H}^{2/3}k^{1/3}},$$ where $\mathcal{H}$_is_the_maximum of the absolute_values of mean curvatures on surface_$\Gamma$. The latter choice is motivated_by the fact that $c=R^{-2/3}$ leads to_operators $\mathcal{B}_{k,2,k,0.4ck^{1/3}}$ with nearly optimal spectral_properties in the high-frequency regime_in the_case when $\Gamma$ is a_sphere of radius_$R$ (see_Section \[DtN\]). For general_smooth surfaces $\Gamma$, we followed the_heuristical practice of_replacing $R^{-1}$ by $\mathcal{H}$.
In addition, we_illustrate_the performance of_solvers_based_on Calderón-Ikawa_CFIER formulations at_high_frequencies. This_type_of formulations has been shown to_work_effectively in the low and medium frequency_regime [@Contopa_et_al; @turc1; @Andriulli1; @Andriulli]._We_focus on the arguably_simplest version of Calderón CFIER_formulations that involve the following boundary_integral operators [@turc1]:_$$\ |
matrices reported in each of these tables perhaps require a little more explanation. Firstly, we should point out that the diagonals are zero by construction. We found that the reduced $\chi^2$ ($\chi^2_{\rm r}
= \chi^2 / (n_{\rm d}-1)$, where $n_{\rm d}$ is the global number of data being treated and we are considering the estimate of a single parameter, namely CMB distortion or CIB amplitude), is always extremely close to unity, which is an obvious validation cross-check. Note that, in principle, when potential residuals are included, one should specify the variance pixel-by-pixel in the estimation of $\chi^2$.[^18] This requires a precise local characterization of residuals. While this can easily be included by construction in our analyses, we explicitly avoid implementing this in the $\chi^2$ analysis, but instead perform our forecasts assuming knowledge of only the average level of the residuals in the sky region being considered. Secondly, we note that the matrices are not perfectly symmetric, due to the cross-terms in the squares (from noise and signal) entering into the $\chi^2$. Thirdly, the off-diagonal terms are sometimes negative, but with absolute values compatible with the quoted rms. These second and third effects are clearly statistical in nature.
The results found in this section (see also Appendix \[ideal\_highres\]) identify the ideal sensitivity target for CMB spectral distortion parameters and CIB amplitude that are achievable from the dipole frequency behaviour.
Elements[^19] (2:4, 2:4) of the matrix quantify the sensitivity to the CIB amplitude. Comparison with FIRAS in terms of the $\sigma$ level of significance can be extracted directly from the tables; the ideal improvement ranges from a factor of about 1000 to 4000.
The ideal improvement found for CMB spectral distortion parameters is also impressive. Elements (1, 5:10) and (5:10, 1) and elements (1, 11:12) and (11:12, 1) refer to comparisons between the blackbody and BE and Comptonization distortions, respectively. The comparison with FIRAS is simply quoted by the element of the matrix of the table multiplied by the ratio between the FIRAS $1\,\sigma$ upper limit on $\mu_0$ or $u$ and the distortion parameter value considered in the table. The sensitivity on $u$ is clearly enough to disentangle between minimal models of reionization | matrices reported in each of these tables perhaps command a short more explanation. Firstly, we should point out that the diagonals are zero by structure. We found that the reduced $ \chi^2 $ ($ \chi^2_{\rm r }
= \chi^2 / (n_{\rm d}-1)$, where $ n_{\rm d}$ is the ball-shaped issue of data being treat and we are considering the estimate of a individual parameter, namely CMB distortion or CIB amplitude), is constantly extremely close to unity, which is an obvious validation cross - assay. Note that, in principle, when potential residual are included, one should specify the variance pixel - by - pixel in the estimation of $ \chi^2$.[^18 ] This necessitate a precise local characterization of residuals. While this can easily be included by structure in our analyses, we explicitly avoid implementing this in the $ \chi^2 $ analysis, but instead perform our prognosis wear knowledge of only the average level of the residuals in the sky region being considered. Secondly, we note that the matrix are not perfectly symmetric, due to the hybridization - terms in the squares (from noise and signal) entering into the $ \chi^2$. third, the off - aslant terms are sometimes negative, but with absolute values compatible with the quote rms. These second and third effects are clearly statistical in nature.
The results found in this section (see also Appendix \[ideal\_highres\ ]) identify the ideal sensitivity target for CMB spectral distortion argument and CIB amplitude that are achievable from the dipole frequency behavior.
Elements[^19 ] (2:4, 2:4) of the matrix quantify the sensitivity to the CIB amplitude. Comparison with FIRAS in terms of the $ \sigma$ level of significance can be extracted directly from the table; the ideal improvement ranges from a factor of about 1000 to 4000.
The ideal improvement found for CMB spectral distortion parameters is also impressive. element (1, 5:10) and (5:10, 1) and elements (1, 11:12) and (11:12, 1) consult to comparisons between the blackbody and beryllium and Comptonization distortions, respectively. The comparison with FIRAS is simply quoted by the element of the matrix of the mesa multiplied by the proportion between the FIRAS $ 1\,\sigma$ upper limit on $ \mu_0 $ or $ u$ and the distortion parameter value view in the table. The sensitivity on $ u$ is clearly enough to disentangle between minimal models of reionization | mahrices reported in each uf these tables perhapv requjre a ligtle more explanation. Firstlb, we shouod point out that the aiagonals are zeri by xonstruction. We fomud thzb the ceduced $\chi^2$ ($\chi^2_{\tm r}
= \chi^2 / (n_{\rk d}-1)$, where $n_{\rm g}$ ks the global number of data being trqated amd we are considgring ehe sstimate of a single parameter, namsly CMB distortion ot CIB amplitude), is always fxtrfmely close to unihy, which is an jvvious validxtion cross-check. Note jhat, in principle, when potential resibuals are ibcoudfg, one shoulv specpfy the variakbe pixen-by-pixek in the estimstimn if $\chi^2$.[^18] This requires e precise local charwcterizathou of residuals. While rhus cat eavily ve kncmuved by cojstcuction in kur analysew, we explicitly avood pkplementing fhis ig ehe $\chi^2$ analysis, but instead perform oug fodecasts assuming knowleege of only the averahe level jf the residuals in the sky region being considerad. Serovdlv, we nure that the matrices are not perfectly symmetris, die to the cross-tcrms in the squarex (vrpi noise and skgnal) zhtsring into the $\chi^2$. Thirdli, the iff-diagonwl trrms are sometimes negative, but with abfilute values compacible with tke quojed rmx. These second and thirb effedts are clewrly statjrtical in nature.
Ghe rasults found in this sectijn (see alwo A'pendix \[iaeal\_nighref\]) identify the lgeal sensitivity twrget fmr CMB speftral distortion parameters and CIB amplitude tvat are achnevablc from the dipoje frequency bghaviour.
Ejemengs[^19] (2:4, 2:4) of thv matrix xuantify thq sensitivity jo the CIB am'litude. Cjmpaeisob with WKRAS in terms pf the $\siyia$ lwvel of significange cav be extracted dnxextly from the tsblds; ehv iveal ykprovement rdnger ffpm a wactor of anoug 1000 tp 4000.
The ideal improvemant round for CMB specyrwl distottion parwmeters is alxo impressive. Elemtnts (1, 5:10) anv (5:10, 1) anc ejements (1, 11:12) and (11:12, 1) refer to compadisons behwecn the blackbjdy qnd BE and Cpmptonization distortions, respectively. Vhe comparison with FITAS is simply quoted yy the elemenv of tre matrix of the table multipoied by the ratio between the FIRAS $1\,\sifma$ up[er llmit on $\mu_0$ or $u$ and the distortion parameter value considered in the tablw. The sensitivity kn $u$ is cnecrlv enougr to visentangle betwevn minimal models of reionizatioi | matrices reported in each of these tables a more explanation. we should point zero construction. We found the reduced $\chi^2$ r} = \chi^2 / (n_{\rm d}-1)$, $n_{\rm d}$ is the global number of data being treated and we are the estimate of a single parameter, namely CMB distortion or CIB amplitude), is extremely to which an obvious validation cross-check. Note that, in principle, when potential residuals are included, one should specify variance pixel-by-pixel in the estimation of $\chi^2$.[^18] This a precise local characterization residuals. While this can easily included construction in analyses, explicitly implementing this in $\chi^2$ analysis, but instead perform our forecasts assuming knowledge of only the average level of the residuals the sky considered. Secondly, note the are not perfectly to the cross-terms in the squares signal) entering into the $\chi^2$. Thirdly, the off-diagonal are sometimes but with absolute values compatible with quoted rms. These second and third effects are statistical in nature. The results found in this section (see also Appendix \[ideal\_highres\]) identify the target for CMB spectral parameters and CIB that achievable the frequency behaviour. (2:4, 2:4) of the matrix quantify the sensitivity to the CIB Comparison with FIRAS in terms of the $\sigma$ level of be directly from the the ideal improvement ranges a of about 1000 to ideal found distortion is impressive. Elements (1, 5:10) (5:10, 1) and elements (1, and (11:12, 1) refer and BE and Comptonization distortions, respectively. The comparison FIRAS is simply quoted by the element the matrix of the table multiplied by the ratio between the FIRAS upper limit or $u$ and the distortion parameter value considered the table. The sensitivity $u$ is clearly enough to disentangle between minimal models reionization | matrices reported in each of tHese tables PerhaPs rEquIrE a liTtle More explanatioN. firsTly, we should point out thaT the dIaGOnalS ArE zero By constRUcTIOn. WE fOuNd tHaT ThE reduCed $\Chi^2$ ($\chi^2_{\rM r}
= \chi^2 / (n_{\rm d}-1)$, wHerE $n_{\Rm d}$ is the globAL nUmber of datA beIng treated anD we Are conSiDerINg the EstImate Of a sinGLe paraMeter, nameLy cmB distORtion or cib aMpliTude), is always extreMElY Close to unity, whIch is aN oBViOUS vaLidAtion cross-ChEck. NoTE that, in PRiNCIPle, WHen potential rEsiduals are INclUded, onE sHouLD speciFy the VaRIanCe pixel-by-piXel iN the estimAtion oF $\Chi^2$.[^18] This REquires A preciSe lOcaL chaRAcTeRizAtIOn oF ReSidUAls. while thiS cAn EasilY be iNCLUDed bY coNstrUctioN in our analyseS, we ExplICitLy avoId impLemeNtIng thIs in thE $\chi^2$ aNaLysis, but instead PerfOrm our forEcaStS asSuMing kNOwledgE of OnlY the aveRage levEL of ThE RESiDuals in the sky regioN bEINg ConsiderEd. SecoNDlY, wE Note that ThE maTricES Are noT perFEcTly symmeTric, duE To ThE cross-tErMs in thE sQuaRes (From nOIse aNd signAl) enteriNg intO The $\chi^2$. Thirdly, tHE off-diagonal tERmS ARe SOmetImeS negative, buT witH AbsoLute VAlUes COmpatIble wItH ThE Quoted rms. These seconD aNd thirD effeCts are clearly StatisticaL IN Nature.
ThE resULtS Found in this secTion (sEe also AppeNDix \[ideal\_HighrEs\]) identiFy the ideaL SEnsitiviTy tArgEt fOr Cmb SpEctral distortION parAmEters anD CIb amplitUde ThaT arE acHiEvable froM the dipoLe FrEqUeNcy BehavIOur.
ElemeNtS[^19] (2:4, 2:4) of ThE maTrix qUAntify The seNsitIvItY To tHe CIB amPLiTUDe. CoMpArIson WitH FiRAS iN terMS of The $\sigmA$ level of sIgnIFicaNcE cAn be extRacted directlY fRom the tablEs; The Ideal iMPRovement Ranges from a factor of abouT 1000 To 4000.
The idEal ImproVemeNt found foR CMb spectRal DIstortIon parAmeteRs Is aLSO imprESSiVe. ELeMents (1, 5:10) and (5:10, 1) anD ELemEnts (1, 11:12) aNd (11:12, 1) RefeR to compArisons between the bLAckBody and BE and COmpToniZATiOn dIStORtiOnS, ResPECtively. The compaRison with FiRas iS simply quoTEd bY tHe elemeNt of the MatriX Of the taBle multipLied by the RaTio bETWeeN the FIRAS $1\,\sIgma$ uppeR limit on $\mU_0$ Or $u$ anD ThE distOrtIon parAmEteR valuE consiDEreD in thE table. thE sensiTivitY oN $u$ is cleaRly enough to disentangle bEtween MinimAl mOdels of reIonIZatIon | matrices reported in each of thesetable s p erh ap s re quir e a little mor e exp lanation. Firstly, weshoul dp oint ou t tha t the d i ag o n als a re ze ro by cons tru ction.We found t hat t he reduced $ \ ch i^2$ ($\ch i^2 _{\rm r}
= \ chi ^2 / ( n_ {\r m d}-1 )$, wher e $n_{ \ rm d}$ is the g lo b al num b er of d a t abein g treated and wea re considering th e esti ma t eo f asin gle parame te r, na m ely CMB di s t o rti o n or CIB ampl itude), isa lwa ys ext re mel y close to u ni t y,which is an obv ious vali dation cross-c h eck. No te tha t,inprin c ip le , w he n po t en tia l re sidualsar einclu ded, o n e sho uld spe cifythe variancepix el-b y -pi xel i n the est im ation of $\ chi^2 $. [^18] This requ ires a precis e l oc alch aract e rizati onofresidua ls. Whi l e t hi s c an easily be include db y c onstruct ion in ou ra nalyses, w e e xpli c i tly a void im plementi ng thi s i nthe $\c hi ^2$ an al ysi s,but i n stea d perf orm ourforec a sts assuming k n owledge of on l yt h ea vera gelevel of th e re s idua ls i n t hes ky re gionbe i ng considered. Secondl y, we no te th at the matric es are not p e rfectlysymm e tr i c, due to thecross -terms int he squar es (f rom nois e and sig n a l) enter ing in tothe $ \c hi^2$. Thirdl y , the o ff-diag ona l terms ar e s ome tim es negative , but wi th a bs ol ute valu e s compat ib lewi ththe q u oted r ms. T hese s ec o ndand thi r de f fect sar e cl ear ly stat isti c alin natu re.
Theres u ltsfo un d in th is section (s ee also Appe nd ix\[idea l \ _highres \]) identify the ideals ensitiv ity targ et f or CMB sp ect ral di sto r tion p aramet ers a nd CI B ampli t u de th at are achie v a ble from t he d ipole f requency behaviour .
E lements[^19](2: 4, 2 : 4 )oft he mat ri x qu a n tify the sensit ivity to t he CI B amplitud e . C om parison with F IRASi n terms of the $ \sigma$ l ev el o f sig nificancecan be e xtractedd irect l yfromthe table s; th e ide al imp r ove mentranges f rom afacto rof about 1000 to 4000.
The ide al imp rovem ent found fo r C M B s pectral d isto rtion para met ers is a lso impre ssiv e .Ele m ents(1,5 :10) and( 5: 10, 1 )and element s ( 1,11:12 ) a n d (11: 12,1) refer to compa r isons betweentheb l ack bod y and B E and Comptoni zat io n distorti on s, respecti vely. Th ec ompar ison w ith FI RAS iss i mp l y quot ed b y t he elemen t o ft he matr ix o f the t able m ultipl ied by ther a tio between theFIRAS $ 1\,\s i gma $ upp er limito n $\ mu_0$ or $ u$ and thedistor tion para meter v al ue con sid er ed in thet able. The sens itivity o n $u $ i s clea rlye n oughto d is ent angle bet w e en mi ni m almode ls of r eion ization | matrices_reported in_each of these tables_perhaps require_a_little more_explanation._Firstly, we should_point out that_the diagonals are zero_by construction. We_found_that the reduced $\chi^2$ ($\chi^2_{\rm r}
= \chi^2 / (n_{\rm d}-1)$, where $n_{\rm d}$ is_the_global number_of_data_being treated and we are_considering the estimate of a_single parameter,_namely CMB distortion or CIB amplitude), is always_extremely_close to unity,_which is an obvious validation cross-check. Note that, in_principle, when potential residuals are included,_one should specify_the_variance_pixel-by-pixel in the estimation_of $\chi^2$.[^18] This requires a precise_local characterization of residuals. While this_can easily be included by construction in_our analyses, we explicitly avoid implementing_this in the $\chi^2$ analysis,_but instead_perform our forecasts assuming knowledge_of only the_average level_of the residuals_in the sky region being considered._Secondly, we note_that the matrices are not perfectly_symmetric,_due to the_cross-terms_in_the squares_(from noise and_signal)_entering into_the_$\chi^2$. Thirdly, the off-diagonal terms are_sometimes_negative, but with absolute values compatible with_the quoted rms. These_second_and third effects are_clearly statistical in nature.
The results_found in this section (see also_Appendix \[ideal\_highres\]) identify_the ideal_sensitivity target for CMB spectral distortion parameters and CIB amplitude that_are achievable from the dipole frequency_behaviour.
Elements[^19] (2:4, 2:4) of_the matrix_quantify_the sensitivity to_the_CIB amplitude._Comparison with FIRAS in terms of the_$\sigma$ level_of significance can be extracted directly_from the tables; the_ideal_improvement ranges from a factor of_about 1000 to 4000.
The ideal improvement_found for CMB spectral distortion_parameters_is_also impressive. Elements (1, 5:10)_and (5:10, 1) and elements (1,_11:12) and (11:12,_1) refer to comparisons between the blackbody_and_BE and Comptonization distortions, respectively. The_comparison_with FIRAS is simply quoted by_the_element_of the matrix of the_table multiplied by the ratio between_the FIRAS $1\,\sigma$ upper limit on $\mu_0$ or $u$_and the distortion_parameter value considered in the_table._The_sensitivity on $u$ is clearly enough to disentangle between minimal_models of_reionization |
proofs of [@nnoids; @minoids] gives an effective strategy to construct the desired CMC surfaces $M_t$. This is done in Sections \[sectionNnoids\] and \[sectionMinoids\]. However, showing that $M_t$ is Alexandrov-embedded requires a precise knowledge of its ends. This is the purpose of the main theorem (Section \[sectionPerturbedDelaunayImmersions\], Theorem \[theoremPerturbedDelaunay\]). We consider a family of holomorphic perturbations of the data giving rise via the DPW method to a half-Delaunay embedding $f_0: {\mathbb{D}}^*\subset {\mathbb{C}}\longrightarrow {\mathbb{H}}^3$ and show that the perturbed induced surfaces $f_t({\mathbb{D}}^*)$ are also embedded. Note that the domain on which the perturbed immersions are defined does not depend on the parameter $t$, which is stronger than $f_t$ having an embedded end, and is critical for showing that the surfaces $M_t$ are Alexandrov-embedded. The essential hypothesis on the perturbations is that they do not occasion a period problem on the domain ${\mathbb{D}}^*$ (which is not simply connected). The proof relies on the Fröbenius method for linear differential systems with regular singular points. Although this idea has been used in ${\mathbb{R}}^3$ by Kilian, Rossman, Schmitt [@krs] and [@raujouan], the case of ${\mathbb{H}}^3$ generates two extra resonance points that are unavoidable and make their results inapplicable. Our solution is to extend the Fröbenius method to loop-group-valued differential systems.
![*Theorem \[theoremConstructionNnoids\] ensures the existence of $n$-noids with small necks. For $H>1$ small enough ($H\simeq1.5$ on the picture), there exists embedded $n$-noids with more than six ends.*](7noidmainlevee.png){width="7cm"}
Delaunay surfaces in ${\mathbb{H}}^3$ via the DPW method {#sectionNotations}
========================================================
This Section fixes the notation and recalls the DPW method in ${\mathbb{H}}^3$.
Hyperbolic space
----------------
#### Matrix model.
Let ${\mathbb{R}}^{1,3}$ denote the space ${\mathbb{R | proofs of [ @nnoids; @minoids ] gives an effective strategy to construct the desire CMC surface $ M_t$. This is done in Sections \[sectionNnoids\ ] and \[sectionMinoids\ ]. However, showing that $ M_t$ is Alexandrov - embedded ask a precise knowledge of its conclusion. This is the determination of the main theorem (part \[sectionPerturbedDelaunayImmersions\ ], Theorem \[theoremPerturbedDelaunay\ ]). We consider a family of holomorphic perturbation of the data giving advance via the DPW method acting to a half - Delaunay embedding $ f_0: { \mathbb{D}}^*\subset { \mathbb{C}}\longrightarrow { \mathbb{H}}^3 $ and show that the perturb induced surfaces $ f_t({\mathbb{D}}^*)$ are besides embedded. Note that the domain on which the perturb immersions are defined does not count on the parameter $ t$, which is strong than $ f_t$ having an embedded end, and is critical for showing that the surfaces $ M_t$ are Alexandrov - implant. The essential hypothesis on the perturbations is that they do not occasion a period problem on the domain $ { \mathbb{D}}^*$ (which is not simply connected). The proof relies on the Fröbenius method for linear differential systems with regular singular points. Although this idea has been use in $ { \mathbb{R}}^3 $ by Kilian, Rossman, Schmitt [ @krs ] and [ @raujouan ], the case of $ { \mathbb{H}}^3 $ generates two excess plangency point that are unavoidable and make their result inapplicable. Our solution is to extend the Fröbenius method to loop - group - valued differential system.
! [ * Theorem \[theoremConstructionNnoids\ ] ensures the existence of $ n$-noids with small necks. For $ H>1 $ small enough ($ H\simeq1.5 $ on the picture), there exists embed $ n$-noids with more than six ends.*](7noidmainlevee.png){width="7 cm " }
Delaunay surfaces in $ { \mathbb{H}}^3 $ via the DPW method { # sectionNotations }
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
This Section fixes the note and recalls the DPW method acting in $ { \mathbb{H}}^3$.
Hyperbolic space
----------------
# # # # Matrix model.
Let $ { \mathbb{R}}^{1,3}$ denote the space $ { \mathbb{R | prlofs of [@nnoids; @minoids] glves an effectivg wtratejy to cknstruct the desired CMC surfaces $M_t$. Tyis iw done in Sections \[secgionNnoidd\] and \[sextioiMinoids\]. However, showing that $J_b$ is Clxxandrov-embeddec requires a precise knoflddye of its ends. This is the purpose os the msij theorem (Sectyon \[xqctiknPerturbedDelaunayImmersions\], Theodem \[themremPerturbedCelaunay\]). We consider a famlly lf holomorphic perhurbations if trw data givine rise via the DPW metgod to a half-Delaunay embedding $f_0: {\machbb{D}}^*\subset {\mqthht{C}}\longrighterrow {\iathbb{H}}^3$ and show thad the prrturbed induccd sucfacws $f_t({\mathbb{D}}^*)$ are also embedded. Note thaj the domahn on which the perruebed hmmessiovw afe sehinsd doed nkt depend kn the paraneter $t$, which is sttogter than $f_t$ hzving wn embedded end, and is critical for showitg fhat the surfaces $M_t$ arw Alexandrov-embedded. Jhe essentyal hypothesis on the perturbations is that they go nov ucccwion x pfriod problem on the domain ${\mathbb{D}}^*$ (which is gkt spmply connected). Tme proof relies on tje Sröbenius methud for lihear differential dystems with regular fingilar points. Although this ieea has been ysed in ${\mathbb{R}}^3$ by Kilian, Rosrman, Schmott [@krs] and [@raujouan], thz case of ${\mathbb{H}}^3$ generatea two extra resonxncv pohnts thau are unavoidable wnd make vheir resultr inspplicwble. Our sllution is to extend the Fröbeuius kethod to poop-group-valued differential systems.
![*Theorem \[tnemrekConstruetionNkoids\] ensures tre existence oy $n$-noids with small necis. For $I>1$ small enoudh ($H\simeq1.5$ on dje picture), tiere exises enbedeed $n$-nokas with more tnan six eubs.*](7noidmaiblevee.png){width="7cm"}
Dekauvzy surfaces in ${\natybb{H}}^3$ via the DPE mdthjd {#sxctiogTotations}
========================================================
Thiv Seztium fixds the notabiov anc recalls the DPW medhod in ${\mathbb{H}}^3$.
Hyperbokig space
----------------
#### Mqtrix moqel.
Let ${\mathbb{T}}^{1,3}$ denote the space ${\mathub{R | proofs of [@nnoids; @minoids] gives an effective construct desired CMC $M_t$. This is \[sectionMinoids\]. showing that $M_t$ Alexandrov-embedded requires a knowledge of its ends. This is purpose of the main theorem (Section \[sectionPerturbedDelaunayImmersions\], Theorem \[theoremPerturbedDelaunay\]). We consider a family holomorphic perturbations of the data giving rise via the DPW method to a embedding {\mathbb{D}}^*\subset {\mathbb{H}}^3$ show that the perturbed induced surfaces $f_t({\mathbb{D}}^*)$ are also embedded. Note that the domain on which perturbed immersions are defined does not depend on parameter $t$, which is than $f_t$ having an embedded and critical for that surfaces are Alexandrov-embedded. The hypothesis on the perturbations is that they do not occasion a period problem on the domain ${\mathbb{D}}^*$ is not The proof on Fröbenius for linear differential regular singular points. Although this idea in ${\mathbb{R}}^3$ by Kilian, Rossman, Schmitt [@krs] and the case ${\mathbb{H}}^3$ generates two extra resonance points are unavoidable and make their results inapplicable. Our is to extend the Fröbenius method to loop-group-valued differential systems. ![*Theorem \[theoremConstructionNnoids\] ensures the existence with small necks. For small enough ($H\simeq1.5$ the there embedded with more six ends.*](7noidmainlevee.png){width="7cm"} Delaunay surfaces in ${\mathbb{H}}^3$ via the DPW method {#sectionNotations} This Section fixes the notation and recalls the DPW method Hyperbolic ---------------- #### Matrix Let ${\mathbb{R}}^{1,3}$ denote the ${\mathbb{R | proofs of [@nnoids; @minoids] giveS an effectiVe strAteGy tO cOnstRuct The desired CMC sURfacEs $M_t$. This is done in SectioNs \[secTiONNnoIDs\] And \[seCtionMiNOiDS\]. howEvEr, ShoWiNG tHat $M_t$ Is ALexandrOv-embedded ReqUiRes a precise kNOwLedge of its EndS. This is the puRpoSe of thE mAin THeoreM (SeCtion \[SectioNperturBedDelaunAyiMmersiONs\], TheorEM \[ThEoreMPerturbedDelaunaY\]). we COnsider a family Of holoMoRPhIC PerTurBations of tHe Data gIVing risE ViA THE DPw Method to a half-delaunay embEDdiNg $f_0: {\matHbB{D}}^*\sUBset {\maThbb{C}}\LoNGriGhtarrow {\matHbb{H}}^3$ And show thAt the pERturbed INduced sUrfaceS $f_t({\MatHbb{D}}^*)$ ARe AlSo eMbEDdeD. noTe tHAt tHe domain On WhIch thE perTURBEd imMerSionS are dEfined does not DepEnd oN The ParamEter $t$, WhicH iS stroNger thAn $f_t$ hAvIng an embedded enD, and Is criticaL foR sHowInG that THe surfAceS $M_t$ Are AlexAndrov-eMBedDeD. tHE eSsential hypothesis On THE pErturbatIons is THaT tHEy do not oCcAsiOn a pERIod prObleM On The domaiN ${\mathbB{d}}^*$ (wHiCh is not SiMply coNnEctEd). THe proOF relIes on tHe FröbenIus meTHod for linear diFFerential systEMs WITh REgulAr sIngular poinTs. AlTHougH thiS IdEa hAS been Used iN ${\mAThBB{R}}^3$ by Kilian, Rossman, ScHmItt [@krs] And [@raUjouan], the case Of ${\mathbb{H}}^3$ gENERates two ExtrA ReSOnance points thAt are UnavoidablE And make tHeir rEsults inApplicablE. oUr solutiOn iS to ExtEnd THE FRöbenius methoD TO looP-gRoup-valUed DiffereNtiAl sYstEms.
![*thEorem \[theoRemConstRuCtIoNNNoiDs\] ensURes the exIsTenCe Of $n$-Noids WIth smaLl necKs. FoR $H>1$ SmALl eNough ($H\sIMeQ1.5$ ON the PiCtUre), tHerE eXists EmbeDDed $N$-noids wIth more thAn sIX endS.*](7nOiDmainleVee.png){width="7cm"}
deLaunay surfAcEs iN ${\mathbB{h}}^3$ Via the DPw method {#sectionNotations}
========================================================
tHis SectIon Fixes The nOtation anD reCalls tHe Dpw methoD in ${\matHbb{H}}^3$.
HYpErbOLIc spaCE
----------------
#### maTriX mOdel.
Let ${\matHBB{R}}^{1,3}$ dEnote ThE spaCe ${\mathbB{R | proofs of [@nnoids; @mino ids] gives an e ffe cti ve str ateg y to construct thedesired CMC surfaces $ M_t$. T h is i s d one i n Secti o ns \ [se ct io nNn oi d s\ ] and \[ section Minoids\]. Ho we ver, showing th at $M_t$ i s A lexandrov-em bed ded re qu ire s a pr eci se kn owledg e of it s ends. T hi s is th e purpos e of the main theorem (Se c ti o n \[sectionPer turbed De l au n a yIm mer sions\], T he orem\ [theore m Pe r t u rbe d Delaunay\]).We consider a f amilyof ho l omorph ic pe rt u rba tions of th e da ta giving risev ia theD PW meth od toa h alf -Del a un ay em be d din g $ f_0 : {\ mathbb{D }} ^* \subs et { \ m a t hbb{ C}} \lon grigh tarrow {\math bb{ H}}^ 3 $ a nd sh ow th at t he pert urbedinduc ed surfaces $f_t( {\ma thbb{D}}^ *)$ a real so em b edded. No tethat th e domai n on w h i c hthe perturbed imme rs i o ns are def ined d o es n o t depend o n t he p a r amete r $t $ ,which is stron g er t han $f_ t$ havin ganemb edded end, and i s critic al fo r showing thatt he surfaces $ M _t $ ar e Ale xan drov-embedd ed.T he e ssen t ia l h y pothe sis o nt he perturbations is th at theydo no t occasion aperiod pro b l e m on the dom a in ${\mathbb{D}}^ *$ (w hich is no t simplyconne cted). T he proofr e lies onthe Fr öbe niu s me thod for line a r dif fe rential sy stems w ith re gul arsi ngular po ints. Al th ou gh t his idea has been u sed i n $ {\mat h bb{R}} ^3$ b y Ki li an , Ro ssman,S ch m i tt [ @k rs ] an d [ @r aujou an], the case o f ${\math bb{ H }}^3 $ge nerates two extra re so nance poin ts th at are u navoidab le and make their resul t s inapp lic able. Our solution is to ex ten d the F röbeni us me th odt o loop - g ro up- va lued diffe r e nti al sy st ems.
![*Th eorem \[theoremCon s tru ctionNnoids\] en sure s th e e x is t enc eo f $ n $ -noids with sma ll necks.Fo r $ H>1$ small eno ug h ($H\s imeq1.5 $ ont he pict ure), the re exists e mbed d e d $ n$-noids w ith more than six ends. * ]( 7noid mai nlevee .p ng) {widt h="7cm " }
Delau nay su rf aces i n ${\ ma thbb{H}} ^3$ via the DPW method{#sect ionNo tat ions}
=== === = === ========= ==== ========== === === ===== === = ===== ===Th isS ectio n fi x es the no t at ion a nd recalls th e D PWmetho d i n ${\ma thbb {H}}^3$.
Hyperbo l ic space
----- ---- - - --- --#### M atrix model.
Let $ { \ mathbb{R }} ^{1,3}$ den ote thesp a ce ${ \mathb b{R | proofs_of [@nnoids;_@minoids] gives an effective_strategy to_construct_the desired_CMC_surfaces $M_t$. This_is done in_Sections \[sectionNnoids\] and \[sectionMinoids\]._However, showing that_$M_t$_is Alexandrov-embedded requires a precise knowledge of its ends. This is the purpose of_the_main theorem_(Section_\[sectionPerturbedDelaunayImmersions\],_Theorem \[theoremPerturbedDelaunay\]). We consider a_family of holomorphic perturbations of_the data_giving rise via the DPW method to a_half-Delaunay_embedding $f_0: {\mathbb{D}}^*\subset_{\mathbb{C}}\longrightarrow {\mathbb{H}}^3$ and show that the perturbed induced surfaces_$f_t({\mathbb{D}}^*)$ are also embedded. Note that_the domain on_which_the_perturbed immersions are defined_does not depend on the parameter_$t$, which is stronger than $f_t$_having an embedded end, and is critical_for showing that the surfaces $M_t$_are Alexandrov-embedded. The essential hypothesis_on the_perturbations is that they do_not occasion a_period problem_on the domain_${\mathbb{D}}^*$ (which is not simply connected)._The proof relies_on the Fröbenius method for linear_differential_systems with regular_singular_points._Although this_idea has been_used_in ${\mathbb{R}}^3$_by_Kilian, Rossman, Schmitt [@krs] and [@raujouan],_the_case of ${\mathbb{H}}^3$ generates two extra resonance_points that are unavoidable_and_make their results inapplicable._Our solution is to extend_the Fröbenius method to loop-group-valued differential_systems.
![*Theorem \[theoremConstructionNnoids\]_ensures the_existence of $n$-noids with small necks. For $H>1$ small enough ($H\simeq1.5$_on the picture), there exists embedded_$n$-noids with more than_six ends.*](7noidmainlevee.png){width="7cm"}
Delaunay_surfaces_in ${\mathbb{H}}^3$ via_the_DPW method_{#sectionNotations}
========================================================
This Section fixes the notation and recalls_the DPW_method in ${\mathbb{H}}^3$.
Hyperbolic space
----------------
#### Matrix model.
Let_${\mathbb{R}}^{1,3}$ denote the space_${\mathbb{R |
to the existence of $P$.
Since there exist polynomial time algorithms for computing minimum $\ell$-length bounded $s$-$t$-cuts for $\ell \leq 3$ and $\ell=n-1$ [@MahjoubMc10; @FordFu56], Lemma \[lem:xcut-kkings-destr\] immediately yields polynomial time algorithms for the minimum $\ell$-length bounded $s$-cut problem for $\ell \in \{2,3,n-1\}$.
\[cor:MoV-UC-kings\] Computing the MoV of a [$\mathit{UC}$]{}winner, a $3$-king or a [$\mathit{TC}$]{}winner in the weighted setting can be done in polynomial time.
The following result is obtained by carefully adjusting the proof of @BaierEr10 showing that approximating minimum $\ell$-length bounded cuts for $\ell \geq 4$ is NP-hard. We give the entire proof in the appendix, where we also point out deviations from the original construction.
\(v) at (0,4) ;at (v)[$v$]{};
\(u) at (0,0) ;at (u)[$u$]{};
\(u) – (v);
[theorem]{}[restateKKingsReduction]{} For any constant $k\geq 4$, computing the MoV of a $k$-king in the unweighted setting is NP-hard. For any constant $\epsilon>0$, the problem is still NP-hard when we restrict to non-constant $k \geq n^{1-\epsilon}$. \[thm:destr:kkings:nphard\]
We reduce from vertex cover; see for the construction for $k=4$. Lemma \[lem:xcut-kkings-destr\] implies that determining the MoV of node $x$ with respect to $4$-kings is equivalent to computing the cost of a $4$-length bounded minimum $x$-cut. The key part of the proof is to show that, for any $c \leq |V(G)|$, there exists a vertex cover in $G$ of size $c$ iff there exists a $4$-bounded $x$-cut in $T$ of size $c + |V(G)|$. For the direction from left to right, a vertex cover $U$ can be translated to a $4$-bounded $x$-$y$-cut by including edges | to the existence of $ P$.
Since there exist polynomial time algorithms for computing minimal $ \ell$-length restrict $ s$-$t$-cuts for $ \ell \leq 3 $ and $ \ell = n-1 $ [ @MahjoubMc10; @FordFu56 ], Lemma \[lem: xcut - kkings - destr\ ] immediately yields polynomial time algorithm for the minimum $ \ell$-length bounce $ s$-cut problem for $ \ell \in \{2,3,n-1\}$.
\[cor: MoV - UC - kings\ ] Computing the MoV of a [ $ \mathit{UC}$]{}winner, a $ 3$-king or a [ $ \mathit{TC}$]{}winner in the leaden mount can be done in polynomial clock time.
The following result is prevail by carefully adjusting the validation of @BaierEr10 showing that approximating minimum $ \ell$-length restrict cuts for $ \ell \geq 4 $ is NP - hard. We move over the entire proof in the appendix, where we also steer out deviations from the original construction.
\(v) at (0,4); at (v)[$v$ ] { };
\(u) at (0,0); at (u)[$u$ ] { };
\(u) – (v);
[ theorem]{}[restateKKingsReduction ] { } For any constant $ k\geq 4 $, computing the MoV of a $ k$-king in the unweighted mise en scene is NP - hard. For any constant $ \epsilon>0 $, the problem is still NP - heavily when we restrict to non - constant $ k \geq n^{1-\epsilon}$. \[thm: destr: kkings: nphard\ ]
We reduce from vertex cover; see for the construction for $ k=4$. Lemma \[lem: xcut - kkings - destr\ ] incriminate that determine the MoV of node $ x$ with respect to $ 4$-kings is equivalent to computing the cost of a $ 4$-length bounded minimum $ x$-cut. The key part of the proof is to show that, for any $ coke \leq |V(G)|$, there exists a vertex cover in $ G$ of size $ c$ iff there exist a $ 4$-bounded $ x$-cut in $ T$ of size $ vitamin c + |V(G)|$. For the direction from left to right, a vertex brood $ U$ can be translated to a $ 4$-bounded $ x$-$y$-cut by including edges | to the existence of $P$.
Since there exist polynomian time algoritfms for computing minimum $\elp$-lwngth bounded $s$-$t$-cuts for $\elu \leq 3$ anf $\ell=n-1$ [@MqhjonbMc10; @FordFu56], Lemme \[lem:xcub-hkinfd-descr\] immediately ylelds polynmmial time algmrktkms for the minimum $\ell$-length boundeq $s$-cut lrlblem for $\ell \yn \{2,3,n-1\}$.
\[bow:MoV-HB-klngs\] Computing the MoV of a [$\mathjt{UC}$]{}winier, a $3$-king or a [$\mathit{TC}$]{}winner in the weihhtef setting can be dlne in polybomiwo time.
The foulowing result is obtajned by carefully adjusting the prooy of @BaierEt10 ahlfing that a'proxiiating minimmk $\ell$-langth bpunded cuts fov $\ell \gew 4$ is NP-hard. We give vhe entire proof in jhe appendhx, where we also pounr out devhatiubs wroj vhe origijal constructjon.
\(v) at (0,4) ;at (v)[$v$]{};
\(u) at (0,0) ;at (u)[$u$]{};
\(u) – (v);
[tnejgrm]{}[restateKKihgsRedtceion]{} For any constant $k\geq 4$, computing tve JoV of a $k$-king in the ubweighted setting is JP-hard. Fow any constant $\epsilon>0$, the problem is still NP-harg whei de xcfgeift to non-constant $k \geq n^{1-\epsilon}$. \[thm:destr:kkinda:nkhagd\]
We reduce from yertex cover; see fpr tng construction for $k=4$. Lejma \[lem:xcut-kkings-dfstr\] imklies rhat detewminong the MoV of node $x$ with eespect to $4$-kpngs is equivalent to eomputing thz cost of a $4$-length bounded minimum $x$-cuf. The key pwrt of ths proof is to shod tmat, for any $c \leq |V(G)|$, there exysts a vectex eover in $G$ og size $c$ iff thege exlvts a $4$-bounded $x$-cut in $T$ ox size $c + |G(G)|$. For the direction from left vp right, a vettef cmver $U$ ccn be branslated to a $4$-bounded $x$-$y$-cut by incjudine edges | to the existence of $P$. Since there time for computing $\ell$-length bounded $s$-$t$-cuts $\ell=n-1$ @FordFu56], Lemma \[lem:xcut-kkings-destr\] yields polynomial time for the minimum $\ell$-length bounded $s$-cut for $\ell \in \{2,3,n-1\}$. \[cor:MoV-UC-kings\] Computing the MoV of a [$\mathit{UC}$]{}winner, a $3$-king a [$\mathit{TC}$]{}winner in the weighted setting can be done in polynomial time. The result obtained carefully the proof of @BaierEr10 showing that approximating minimum $\ell$-length bounded cuts for $\ell \geq 4$ is We give the entire proof in the appendix, we also point out from the original construction. \(v) (0,4) (v)[$v$]{}; \(u) (0,0) (u)[$u$]{}; – (v); [theorem]{}[restateKKingsReduction]{} any constant $k\geq 4$, computing the MoV of a $k$-king in the unweighted setting is NP-hard. For constant $\epsilon>0$, is still when restrict non-constant $k \geq We reduce from vertex cover; see for $k=4$. Lemma \[lem:xcut-kkings-destr\] implies that determining the of node with respect to $4$-kings is equivalent computing the cost of a $4$-length bounded minimum The key part of the proof is to show that, for any $c \leq |V(G)|$, a vertex cover in of size $c$ there a $x$-cut $T$ of $c + |V(G)|$. For the direction from left to right, a cover $U$ can be translated to a $4$-bounded $x$-$y$-cut by | to the existence of $P$.
Since theRe exist polYnomiAl tIme AlGoriThms For computing miNImum $\Ell$-length bounded $s$-$t$-cuts For $\elL \lEQ 3$ and $\ELl=N-1$ [@MahjOubMc10; @FoRDFU56], lEmmA \[lEm:XcuT-kKInGs-desTr\] iMmediatEly yields pOlyNoMial time algoRItHms for the mIniMum $\ell$-length BouNded $s$-cUt ProBLem foR $\elL \in \{2,3,n-1\}$.
\[cOr:MoV-Uc-Kings\] COmputing tHe mOV of a [$\mAThit{UC}$]{}wINNeR, a $3$-kiNg or a [$\mathit{TC}$]{}winnER iN The weighted setTing caN bE DoNE In pOlyNomial time.
thE follOWing resULt IS OBtaINed by carefullY adjusting tHE prOof of @BAiErER10 ShowinG that ApPRoxImating miniMum $\eLl$-length bOunded CUts for $\eLL \geq 4$ is Np-hard. WE giVe tHe enTIrE pRooF iN The APpEndIX, whEre we alsO pOiNt out DeviATIONs frOm tHe orIginaL construction.
\(V) at (0,4) ;At (v)[$v$]{};
\(U) At (0,0) ;aT (u)[$u$]{};
\(u) – (v);
[TheorEm]{}[reStAteKKIngsReDuctiOn]{} for any constant $k\Geq 4$, cOmputing tHe MOV Of a $K$-kIng in THe unweIghTed Setting Is NP-harD. for AnY CONsTant $\epsilon>0$, the probLeM IS sTill NP-haRd when WE rEsTRict to noN-cOnsTant $K \GEq n^{1-\epSiloN}$. \[ThM:destr:kkIngs:npHArD\]
WE reduce FrOm vertEx CovEr; sEe for THe coNstrucTion for $k=4$. lemma \[LEm:xcut-kkings-deSTr\] implies that DEtERMiNIng tHe MOV of node $x$ wiTh reSPect To $4$-kiNGs Is eQUivalEnt to CoMPuTIng the cost of a $4$-length BoUnded mInimuM $x$-cut. The key paRt of the proOF IS to show tHat, fOR aNY $c \leq |V(G)|$, there exIsts a Vertex coveR In $G$ of sizE $c$ iff There exiSts a $4$-boundED $X$-cut in $T$ oF siZe $c + |v(G)|$. FOr tHE DiRection from leFT To riGhT, a verteX coVer $U$ can Be tRanSlaTed To A $4$-bounded $x$-$Y$-cut by inClUdInG eDgeS | to the existence of $P$.
Since the re ex ist po ly nomi al t ime algorithms forcomputing minimum $\el l$-le ng t h bo u nd ed $s $-$t$-c u ts f or$\ el l \ le q 3 $ and $\ ell=n-1 $ [@Mahjou bMc 10 ; @FordFu56] , L emma \[lem :xc ut-kkings-de str \] imm ed iat e ly yi eld s pol ynomia l timealgorithm sf or the minimum $ \e ll$- length bounded $s $ -c u t problem for$\ell\i n \ { 2 ,3, n-1 \}$.
\[co r: MoV-U C -kings\ ] C o m p uti n g the MoV ofa [$\mathit { UC} $]{}wi nn er, a $3$- kingor a [ $\mathit{TC }$]{ }winner i n thew eighted setting can b e d one inp ol yn omi al tim e .
Th e fo llowingre su lt is obt a i n e d by ca refu lly a djusting thepro of o f @B aierE r10 s howi ng that appro ximat in g minimum $\ell $-le ngth boun ded c uts f or $\ e ll \ge q 4 $ i s NP-ha rd. Weg ive t h e en tire proof in theap p e nd ix, wher e we a l so p o int outde via tion s fromtheo ri ginal co nstruc t io n.
\(v)at (0,4) ; at(v) [$v$] { };
\(u) a t (0,0);at ( u )[$u$]{};
\(u ) – (v);
[the o re m ] {} [ rest ate KKingsReduc tion ] {} F or a n ycon s tant$k\ge q4 $, computing the MoV o fa $k$- kingin the unweig hted setti n g is NP-ha rd.F or any constant $ \epsi lon>0$, th e problem is s till NP- hard when w e restri cttonon -co n s ta nt $k \geq n^ { 1 -\ep si lon}$.\[t hm:dest r:k kin gs: nph ar d\]
We r educe fr om v er te x c over; see forth e c on str uctio n for $ k=4$. Lem ma \ [ lem :xcut-k k in g s -des tr \] imp lie sthatdete r min ing the MoV of n ode $x$wi th respec t to $4$-king sis equival en t t o comp u t ing thecost of a $4$-length bo u nded mi nim um $x $-cu t. The ke y p art of th e proof is to show t hat , for a n y $ c \ le q |V(G)|$, t her e exi st s avertexcover in $G$ of si z e $ c$ iff thereexi stsa $4 $-b o un d ed$x $ -cu t in $T$ of size$c + |V(G) |$ . F or the dir e cti on from l eft toright , a vert ex cover$U$ can b etran s l ate d to a $4$ -bounded $x$-$y$- c ut by in cludi ngedges | to_the existence_of $P$.
Since there exist_polynomial time_algorithms_for computing_minimum_$\ell$-length bounded $s$-$t$-cuts_for $\ell \leq_3$ and $\ell=n-1$ [@MahjoubMc10;_@FordFu56], Lemma \[lem:xcut-kkings-destr\]_immediately_yields polynomial time algorithms for the minimum $\ell$-length bounded $s$-cut problem for $\ell \in_\{2,3,n-1\}$.
\[cor:MoV-UC-kings\]_Computing the_MoV_of_a [$\mathit{UC}$]{}winner, a $3$-king or_a [$\mathit{TC}$]{}winner in the weighted_setting can_be done in polynomial time.
The following result is_obtained_by carefully adjusting_the proof of @BaierEr10 showing that approximating minimum $\ell$-length_bounded cuts for $\ell \geq 4$_is NP-hard. We_give_the_entire proof in the_appendix, where we also point out_deviations from the original construction.
\(v) at_(0,4) ;at (v)[$v$]{};
\(u) at (0,0) ;at (u)[$u$]{};
\(u)_– (v);
[theorem]{}[restateKKingsReduction]{} For any constant $k\geq_4$, computing the MoV of_a $k$-king_in the unweighted setting is_NP-hard. For any_constant $\epsilon>0$,_the problem is_still NP-hard when we restrict to_non-constant $k \geq_n^{1-\epsilon}$. \[thm:destr:kkings:nphard\]
We reduce from vertex cover;_see_for the construction_for_$k=4$._Lemma \[lem:xcut-kkings-destr\]_implies that determining_the_MoV of_node_$x$ with respect to $4$-kings is_equivalent_to computing the cost of a $4$-length_bounded minimum $x$-cut. The_key_part of the proof_is to show that, for_any $c \leq |V(G)|$, there exists_a vertex_cover in_$G$ of size $c$ iff there exists a $4$-bounded $x$-cut in_$T$ of size $c + |V(G)|$._For the direction from_left to_right,_a vertex cover_$U$_can be_translated to a $4$-bounded $x$-$y$-cut by including_edges |
.doi.org/10.1103/PhysRevD.12.3004)]{}; Ann. Phys. [**99**]{}, 127-141 (1976). [doi: [10.1016/0003-4916(76)90086-5](http://dx.doi.org/10.1016/0003-4916(76)90086-5)]{} J. M. Nester and J. A. Isenberg, Phys. Rev. D [**15**]{}, 2078 (1977). [doi: [10.1103/PhysRevD.15.2078](http://dx.doi.org/10.1103/PhysRevD.15.2078)]{} A. V. Minkevich, Phys. Lett. A [**80**]{}, 232-234 (1980) [doi: [10.1016/0375-9601(80)90008-0](http://dx.doi.org/10.1016/0375-9601(80)90008-0)]{};\
Phys. Lett. A [**95**]{}, 422-424 (1983). [doi: [10.1016/0375-9601(83)90309-2](http://dx.doi.org/10.1016/0375-9601(83)90309-2)]{} A. V. Minkevich and I. M. Nemenman, Class. Quantum Grav. [**12**]{}, 1259-1265 (1995). [doi: [10.1088/0264-9381/12/5/015](http://dx.doi.org/10.1088/0264-9381/12/5/015)]{} A. V. Minkevich and A. S. Garkun, Class. Quantum Grav. [**23**]{}, 4237 (2006). [doi: [10.1088/0264-9381/23/12/018](http://dx.doi.org/10.1088/0264-9381/23/12/018)]{} A. V. Minkevich, A. S. Garkun, and V. I. Kudin, Class. Quantum Grav. [**24**]{}, 5835 (2007). [doi: [10.1088/0264-9381/24/23/007](http://dx.doi.org/10.1088/02 | .doi.org/10.1103 / PhysRevD.12.3004) ] { }; Ann. Phys. [ * * 99 * * ] { }, 127 - 141 (1976). [ doi: [ 10.1016/0003 - 4916(76)90086 - 5](http://dx.doi.org/10.1016/0003 - 4916(76)90086 - 5) ] { } J. M. Nester and J. A. Isenberg, Phys. Rev. D [ * * 15 * * ] { }, 2078 (1977). [ doi: [ 10.1103 / PhysRevD.15.2078](http://dx.doi.org/10.1103 / PhysRevD.15.2078) ] { } A. V. Minkevich, Phys. Lett. A [ * * 80 * * ] { }, 232 - 234 (1980) [ doi: [ 10.1016/0375 - 9601(80)90008 - 0](http://dx.doi.org/10.1016/0375 - 9601(80)90008 - 0)]{};\
Phys. Lett. A [ * * 95 * * ] { }, 422 - 424 (1983). [ doi: [ 10.1016/0375 - 9601(83)90309 - 2](http://dx.doi.org/10.1016/0375 - 9601(83)90309 - 2) ] { } A. V. Minkevich and I. M. Nemenman, Class. Quantum Grav. [ * * 12 * * ] { }, 1259 - 1265 (1995). [ doi: [ 10.1088/0264 - 9381/12/5/015](http://dx.doi.org/10.1088/0264 - 9381/12/5/015) ] { } A. V. Minkevich and A. S. Garkun, Class. Quantum Grav. [ * * 23 * * ] { }, 4237 (2006). [ doi: [ 10.1088/0264 - 9381/23/12/018](http://dx.doi.org/10.1088/0264 - 9381/23/12/018) ] { } A. V. Minkevich, A. S. Garkun, and V. I. Kudin, Class. Quantum Grav. [ * * 24 * * ] { }, 5835 (2007). [ doi: [ 10.1088/0264 - 9381/24/23/007](http://dx.doi.org/10.1088/02 | .doi.lrg/10.1103/PhysRevD.12.3004)]{}; Ann. Phys. [**99**]{}, 127-141 (1976). [dol: [10.1016/0003-4916(76)90086-5](http://dx.doi.org/10.1016/0003-4916(76)90086-5)]{} J. M. Nestxr and N. A. Isencerg, Phys. Rev. D [**15**]{}, 2078 (1977). [doi: [10.1103/PhysRevV.15.2078](http://dx.dou.org/10.1103/PhysRevD.15.2078)]{} A. V. Minkexich, Phys. Pett. A [**80**]{}, 232-234 (1980) [doi: [10.1016/0375-9601(80)90008-0](yttp://dx.doi.ocf/10.1016/0375-9601(80)90008-0)]{};\
Phys. Lebc. A [**95**]{}, 422-424 (1983). [foi: [10.1016/0375-9601(83)90309-2](ktvp://dx.doi.org/10.1016/0375-9601(83)90309-2)]{} A. V. Kinkevich dnd I. M. Nemenmdn, Cpass. Quantum Grav. [**12**]{}, 1259-1265 (1995). [doi: [10.1088/0264-9381/12/5/015](http://dx.doi.org/10.1088/0264-9381/12/5/015)]{} A. V. Mimkfvich and A. S. Darkln, Claas. Quantum Grav. [**23**]{}, 4237 (2006). [doi: [10.1088/0264-9381/23/12/018](http://dx.doi.org/10.1088/0264-9381/23/12/018)]{} Z. V. Minnevich, A. S. Gatkun, and V. I. Kudin, Class. Quwntul Grav. [**24**]{}, 5835 (2007). [doi: [10.1088/0264-9381/24/23/007](http://dd.doi.org/10.1088/02 | .doi.org/10.1103/PhysRevD.12.3004)]{}; Ann. Phys. [**99**]{}, 127-141 (1976). [doi: M. and J. Isenberg, Phys. Rev. [10.1103/PhysRevD.15.2078](http://dx.doi.org/10.1103/PhysRevD.15.2078)]{} V. Minkevich, Phys. A [**80**]{}, 232-234 [doi: [10.1016/0375-9601(80)90008-0](http://dx.doi.org/10.1016/0375-9601(80)90008-0)]{};\ Phys. Lett. A [**95**]{}, (1983). [doi: [10.1016/0375-9601(83)90309-2](http://dx.doi.org/10.1016/0375-9601(83)90309-2)]{} A. V. Minkevich and I. M. Nemenman, Class. Quantum Grav. 1259-1265 (1995). [doi: [10.1088/0264-9381/12/5/015](http://dx.doi.org/10.1088/0264-9381/12/5/015)]{} A. V. Minkevich and A. S. Garkun, Class. Quantum [**23**]{}, (2006). [10.1088/0264-9381/23/12/018](http://dx.doi.org/10.1088/0264-9381/23/12/018)]{} V. Minkevich, A. S. Garkun, and V. I. Kudin, Class. Quantum Grav. [**24**]{}, 5835 (2007). [doi: | .doi.org/10.1103/PhysRevD.12.3004)]{}; Ann. Phys. [**99**]{}, 127-141 (1976). [doi: [10.1016/0003-4916(76)90086-5](hTtp://dx.doi.orG/10.1016/0003-4916(76)90086-5)]{} J. M. NeSteR anD J. a. IseNberG, Phys. Rev. D [**15**]{}, 2078 (1977). [doi: [10.1103/PhYSRevd.15.2078](http://dx.doi.org/10.1103/PhysRevD.15.2078)]{} A. v. MinkEvICh, PhYS. LEtt. A [**80**]{}, 232-234 (1980) [dOi: [10.1016/0375-9601(80)90008-0](http://dX.DoI.ORg/10.1016/0375-9601(80)90008-0)]{};\
PHyS. LEtt. a [**95**]{}, 422-424 (1983). [dOI: [10.1016/0375-9601(83)90309-2](hTtp://dx.Doi.Org/10.1016/0375-9601(83)90309-2)]{} A. V. MiNkevich and i. M. NEmEnman, Class. QuANtUm Grav. [**12**]{}, 1259-1265 (1995). [doi: [10.1088/0264-9381/12/5/015](hTtp://Dx.doi.org/10.1088/0264-9381/12/5/015)]{} A. V. MiNkeVich anD A. s. GaRKun, ClAss. quantUm Grav. [**23**]{}, 4237 (2006). [DOi: [10.1088/0264-9381/23/12/018](http://Dx.doi.org/10.1088/0264-9381/23/12/018)]{} A. v. MINkevicH, a. S. GarkuN, ANd v. I. KuDin, Class. Quantum GrAV. [**24**]{}, 5835 (2007). [dOI: [10.1088/0264-9381/24/23/007](http://dx.doi.org/10.1088/02 | .doi.org/10.1103/PhysRevD. 12.3004)]{ }; An n.Phy s. [** 99** ]{}, 127-141 ( 1 976) . [doi: [10.1016/0003- 4916( 76 ) 9008 6 -5 ](htt p://dx. d oi . o rg/ 10 .1 016 /0 0 03 -4916 (76 )90086- 5)]{} J. M . N es ter and J. A . I senberg, P hys . Rev. D [** 15* *]{},20 78( 1977) . [ doi:[10.11 0 3/Phys RevD.15.2 07 8 ](http : //dx.do i . or g/10 .1103/PhysRevD.15 . 20 7 8)]{} A. V. Mi nkevic h, Ph y s . L ett . A [**80* *] {}, 2 3 2-234 ( 1 98 0 ) [do i : [10.1016/03 75-9601(80) 9 000 8-0](h tt p:/ / dx.doi .org/ 10 . 101 6/0375-9601 (80) 90008-0)] {};\
P h ys. Let t . A [** 95**]{ },422 -424 (1 98 3). [ d oi: [1 0.1 0 16/ 0375-960 1( 83 )9030 9-2] ( h t t p:// dx. doi. org/1 0.1016/0375-9 601 (83) 9 030 9-2)] {} A. V.Mi nkevi ch and I. M .Nemenman, Class . Qu antum Gra v.[* *12 ** ]{},1 259-12 65(19 95). [d oi: [10 . 108 8/ 0 2 6 4- 9381/12/5/015](htt p: / / dx .doi.org /10.10 8 8/ 02 6 4-9381/1 2/ 5/0 15)] { } A. V . Mi n ke vich and A. S. Ga rk un, Cla ss . Quan tu m G rav . [** 2 3**] {}, 42 37 (2006 ). [d o i: [10.1088/02 6 4-9381/23/12/ 0 18 ] ( ht t p:// dx. doi.org/10. 1088 / 0264 -938 1 /2 3/1 2 /018) ]{} A .V .M inkevich, A. S. Gar ku n, and V. I . Kudin, Clas s. Quantum G r av. [**2 4**] { }, 5835 (2007). [ doi:[10.1088/0 2 64-9381/ 24/23 /007](ht tp://dx.d o i .org/10. 108 8/0 2 | .doi.org/10.1103/PhysRevD.12.3004)]{}; Ann. Phys. [**99**]{},_127-141 (1976)._[doi: [10.1016/0003-4916(76)90086-5](http://dx.doi.org/10.1016/0003-4916(76)90086-5)]{} J. M._Nester and_J._A. Isenberg,_Phys. Rev. D_[**15**]{}, 2078 (1977)._[doi: [10.1103/PhysRevD.15.2078](http://dx.doi.org/10.1103/PhysRevD.15.2078)]{} A._V. Minkevich, Phys. Lett. A [**80**]{},_232-234 (1980) [doi:_[10.1016/0375-9601(80)90008-0](http://dx.doi.org/10.1016/0375-9601(80)90008-0)]{};\
Phys. Lett. A_[**95**]{}, 422-424 (1983). [doi: [10.1016/0375-9601(83)90309-2](http://dx.doi.org/10.1016/0375-9601(83)90309-2)]{} A. V. Minkevich and I. M. Nemenman, Class. Quantum Grav. [**12**]{},_1259-1265_(1995). [doi:_[10.1088/0264-9381/12/5/015](http://dx.doi.org/10.1088/0264-9381/12/5/015)]{}_A._V. Minkevich and A. S._Garkun, Class. Quantum Grav. [**23**]{}, 4237 (2006)._[doi: [10.1088/0264-9381/23/12/018](http://dx.doi.org/10.1088/0264-9381/23/12/018)]{}_A. V. Minkevich, A. S. Garkun, and V._I._Kudin, Class. Quantum Grav. [**24**]{},_5835 (2007). [doi: [10.1088/0264-9381/24/23/007](http://dx.doi.org/10.1088/02 |
'(a) \right)$, and $\lim_{h \to 0} \sqrt{1+h} = \lim_{h\to 0}(1 + h/2)$. For ease of computation, we replace $\phi'$ by $\psi$, to be restored at the last step. $$\label{mucomp}
\lim_{a \to b} \frac{\sqrt{{\ensuremath{D_{s\phi}}}(a,q)} - \sqrt{{\ensuremath{D_{s\phi}}}(b,q)}}{\sqrt{{\ensuremath{D_{s\phi}}}(a,b)}}
= \frac{\lim_{a \to b} \left(\sqrt{(q-a)(\psi(q) - \psi(a)) } - \sqrt{(q-b)(\psi(q) - \psi(b)) } \right)}
{\lim_{a \to b}\sqrt{(b-a)(\psi(b) - \psi(a)) }}$$
Computing the denominator: $$\begin{aligned}
\lim_{b \to a}\sqrt{(b-a) (\psi(b) - \psi(a)) }
&=\lim_{h \to 0} \sqrt{(a+h-a) (\psi(a+h) - \psi(a) }
\\ &=\lim_{h \to 0} \sqrt{h ( \psi(a) + h \psi'(a) - \psi(a)) }
\\ &=\lim_{h \to 0} \sqrt{h (h \psi'(a))} =\lim_{h \to 0} h \sqrt{\psi'(a)} \end{aligned}$$
We now address the numerator: $$\begin{aligned}
\lim_{b \to a}&\sqrt{(q-a)(\psi(q) - \psi(a))} - \sqrt{(q-b) (\psi(q) - \psi(b)) }
\\ &= \lim_{h \to 0} \sqrt{(q-a) (\psi(q) - \psi(a))} - \sqrt{(q-a-h) (\psi(q) - \psi(a) - h \psi'(a)) }
\\ &= \lim_{h \to 0} \sqrt{(q-a) (\psi(q) - \psi(a | ' (a) \right)$, and $ \lim_{h \to 0 } \sqrt{1+h } = \lim_{h\to 0}(1 + h/2)$. For ease of computation, we replace $ \phi'$ by $ \psi$, to be restore at the final step. $ $ \label{mucomp }
\lim_{a \to b } \frac{\sqrt{{\ensuremath{D_{s\phi}}}(a, q) } - \sqrt{{\ensuremath{D_{s\phi}}}(b, q)}}{\sqrt{{\ensuremath{D_{s\phi}}}(a, b) } }
= \frac{\lim_{a \to b } \left(\sqrt{(q - a)(\psi(q) - \psi(a) ) } - \sqrt{(q - b)(\psi(q) - \psi(b) ) } \right) }
{ \lim_{a \to b}\sqrt{(b - a)(\psi(b) - \psi(a) ) } } $ $
Computing the denominator: $ $ \begin{aligned }
\lim_{b \to a}\sqrt{(b - a) (\psi(b) - \psi(a) ) }
& = \lim_{h \to 0 } \sqrt{(a+h - vitamin a) (\psi(a+h) - \psi(a) }
\\ & = \lim_{h \to 0 } \sqrt{h (\psi(a) + h \psi'(a) - \psi(a) ) }
\\ & = \lim_{h \to 0 } \sqrt{h (h \psi'(a) ) } = \lim_{h \to 0 } heat content \sqrt{\psi'(a) } \end{aligned}$$
We immediately address the numerator: $ $ \begin{aligned }
\lim_{b \to a}&\sqrt{(q - a)(\psi(q) - \psi(a) ) } - \sqrt{(q - bacillus) (\psi(q) - \psi(b) ) }
\\ & = \lim_{h \to 0 } \sqrt{(q - a) (\psi(q) - \psi(a) ) } - \sqrt{(q - a - h) (\psi(q) - \psi(a) - heat content \psi'(a) ) }
\\ & = \lim_{h \to 0 } \sqrt{(q - a) (\psi(q) - \psi(a | '(a) \rlght)$, and $\lim_{h \to 0} \sqrt{1+h} = \lim_{h\to 0}(1 + h/2)$. Fot wase oh compufation, wd replace $\phi'$ by $\psi$, to be rxstoeed au the last step. $$\labdl{mucomp}
\lim_{a \ro b} \drac{\sqrt{{\ensuremath{D_{s\phi}}}(a,s)} - \sqxt{{\xnsuremath{D_{s\phi}}}(n,q)}}{\sqrt{{\ensuramath{D_{s\phi}}}(a,b)}}
= \yrac{\lim_{a \to b} \left(\sqrt{(q-a)(\psi(q) - \psi(a)) } - \sqrt{(q-n)(\pdi(q) - \psi(b)) } \rigrt)}
{\lim_{a \to b}\sqrt{(b-a)(\psi(b) - \psi(a)) }}$$
Domputiig the denominayor: $$\begin{aligned}
\lim_{b \tl a}\sert{(b-a) (\psi(b) - \psi(a)) }
&=\lim_{h \to 0} \wqrt{(a+h-a) (\psi(a+f) - \psi(a) }
\\ &=\lim_{h \to 0} \sqrt{h ( \psi(a) + h \psi'(a) - \psi(a)) }
\\ &=\lim_{k \to 0} \sqrt{h (h \psl'(d))} =\lim_{h \to 0} i \sqrt{\isi'(a)} \end{aligncc}$$
Wa now acdress the numcratoc: $$\betin{aligned}
\lim_{b \to a}&\sqrt{(q-a)(\psi(q) - \psi(a))} - \sqrt{(q-b) (\[sn(q) - \psi(b)) }
\\ &= \lim_{h \ro 0} \sqrj{(q-a) (\pvi(q) - \psi(x))} - \sarf{(q-a-h) (\pdi(q) - \psi(a) - h \lsi'(a)) }
\\ &= \lim_{h \to 0} \sqrt{(q-a) (\pso(q) - \psi(a | '(a) \right)$, and $\lim_{h \to 0} \sqrt{1+h} 0}(1 h/2)$. For of computation, we be at the last $$\label{mucomp} \lim_{a \to \frac{\sqrt{{\ensuremath{D_{s\phi}}}(a,q)} - \sqrt{{\ensuremath{D_{s\phi}}}(b,q)}}{\sqrt{{\ensuremath{D_{s\phi}}}(a,b)}} = \frac{\lim_{a \to \left(\sqrt{(q-a)(\psi(q) - \psi(a)) } - \sqrt{(q-b)(\psi(q) - \psi(b)) } \right)} {\lim_{a \to b}\sqrt{(b-a)(\psi(b) \psi(a)) }}$$ Computing the denominator: $$\begin{aligned} \lim_{b \to a}\sqrt{(b-a) (\psi(b) - \psi(a)) } \to \sqrt{(a+h-a) - } \\ &=\lim_{h \to 0} \sqrt{h ( \psi(a) + h \psi'(a) - \psi(a)) } \\ &=\lim_{h 0} \sqrt{h (h \psi'(a))} =\lim_{h \to 0} h \end{aligned}$$ We now address numerator: $$\begin{aligned} \lim_{b \to a}&\sqrt{(q-a)(\psi(q) \psi(a))} \sqrt{(q-b) (\psi(q) \psi(b)) \\ \lim_{h \to 0} (\psi(q) - \psi(a))} - \sqrt{(q-a-h) (\psi(q) - \psi(a) - h \psi'(a)) } \\ &= \lim_{h \to 0} (\psi(q) - | '(a) \right)$, and $\lim_{h \to 0} \sqrt{1+h} = \lim_{h\tO 0}(1 + h/2)$. For ease oF compUtaTioN, wE repLace $\Phi'$ by $\psi$, to be reSToreD at the last step. $$\label{mucOmp}
\liM_{a \TO b} \frAC{\sQrt{{\enSurematH{d_{s\PHI}}}(a,q)} - \SqRt{{\EnsUrEMaTh{D_{s\pHi}}}(b,Q)}}{\sqrt{{\enSuremath{D_{s\Phi}}}(A,b)}}
= \Frac{\lim_{a \to b} \lEFt(\Sqrt{(q-a)(\psi(q) - \Psi(A)) } - \sqrt{(q-b)(\psi(q) - \pSi(b)) } \Right)}
{\lIm_{A \to B}\Sqrt{(b-A)(\psI(b) - \psi(A)) }}$$
CompuTIng the DenominatOr: $$\BEgin{alIGned}
\lim_{B \TO a}\Sqrt{(B-a) (\psi(b) - \psi(a)) }
&=\lim_{h \to 0} \sQRt{(A+H-a) (\psi(a+h) - \psi(a) }
\\ &=\lim_{H \to 0} \sqrT{h ( \PSi(A) + H \Psi'(A) - \psI(a)) }
\\ &=\lim_{h \to 0} \sqRt{H (h \psi'(A))} =\Lim_{h \to 0} h \SQrT{\PSI'(a)} \eND{aligned}$$
We now Address the nUMerAtor: $$\beGiN{alIGned}
\liM_{b \to a}&\SqRT{(q-a)(\Psi(q) - \psi(a))} - \sqrT{(q-b) (\pSi(q) - \psi(b)) }
\\ &= \liM_{h \to 0} \sqRT{(q-a) (\psi(q) - \PSi(a))} - \sqrt{(Q-a-h) (\psi(Q) - \psI(a) - h \Psi'(a)) }
\\ &= \LIm_{H \tO 0} \sqRt{(Q-A) (\psI(Q) - \pSi(a | '(a) \right)$, and $\lim_{ h \to 0} \ sqrt{ 1+h } = \lim _{h\ to 0}(1 + h/2) $ . Fo r ease of computation, we r ep l ace$ \p hi'$by $\ps i $, t o b ere sto re d a t the la st step . $$\label {mu co mp}
\lim _ {a \to b} \f rac {\sqrt{{\ens ure math{D _{ s\p h i}}}( a,q )} -\sqrt{ { \ensur emath{D_{ s\ p hi}}}( b ,q)}}{\ s q rt {{\e nsuremath{D_{s\ph i }} } (a,b)}}
= \frac {\ l im _ { a \ tob} \left(\ sq rt{(q - a)(\psi ( q) - \ps i (a)) } - \sq rt{(q-b)(\p s i(q ) - \p si (b) ) } \ri ght)}
{\lim _{a\to b}\sq rt{(b- a )(\psi( b ) - \ps i(a))}}$ $
C om pu tin gt hed en omi n ato r: $$\be gi n{ align ed}\lim _{b \to a}\s qrt{(b-a) (\p si( b) - \ps i(a)) }
&=\li m_{h \ to 0} \ sqrt{(a+h-a) (\ psi( a+h) - \p si( a) }
\\ &=\lim _{h \t o 0} \s qrt{h ( \ps i( a ) +h \psi'(a) - \psi( a) ) }
\\&=\lim _ {h \ t o 0} \sq rt {h(h \ p s i'(a) )} = \ li m_{h \to 0} h\ sq rt {\psi'( a) } \end {a lig ned }$$
W e nowaddressthe n u merator: $$\be g in{aligned}
\ l im _ {b \ toa}&\sqrt{(q -a)( \ psi( q) - \p si( a ))} - \sqr t{ ( q- b ) (\psi(q) - \psi(b )) }
\\&= \lim_{h \t o 0} \sqrt { ( q -a) (\ps i(q) -\ psi(a))} - \s qrt{( q-a-h) (\p s i(q) - \ psi(a ) - h \p si'(a)) } \\ & = \ lim _{h \t o 0} \sqrt{(q-a)( \ psi( q) - \psi (a | '(a) \right)$,_and $\lim_{h_\to 0} \sqrt{1+h} =_ \lim_{h\to_0}(1_+ h/2)$._For_ease of computation,_we replace $\phi'$_by $\psi$, to be_restored at the_last_step. $$\label{mucomp}
\lim_{a \to b} \frac{\sqrt{{\ensuremath{D_{s\phi}}}(a,q)} - \sqrt{{\ensuremath{D_{s\phi}}}(b,q)}}{\sqrt{{\ensuremath{D_{s\phi}}}(a,b)}}
__= \frac{\lim_{a_\to_b}_\left(\sqrt{(q-a)(\psi(q) - \psi(a)) } _- \sqrt{(q-b)(\psi(q) - \psi(b)) }_\right)}
_ __{\lim_{a \to b}\sqrt{(b-a)(\psi(b)_- \psi(a)) }}$$
Computing the denominator:_$$\begin{aligned}
\lim_{b \to_a}\sqrt{(b-a) (\psi(b) -_\psi(a))_}_
_ &=\lim_{h \to 0}_\sqrt{(a+h-a) (\psi(a+h) - \psi(a) }
_ \\ &=\lim_{h \to 0}_\sqrt{h ( \psi(a) + h \psi'(a)_- \psi(a)) }
_ _\\ &=\lim_{h \to 0} \sqrt{h_(h \psi'(a))} =\lim_{h_\to 0}_h \sqrt{\psi'(a)} \end{aligned}$$
_ We now address_the numerator: $$\begin{aligned}
_ \lim_{b \to a}&\sqrt{(q-a)(\psi(q)_-_\psi(a))} - \sqrt{(q-b)_(\psi(q)_-_\psi(b)) }_
__\\ &=_\lim_{h_\to 0} \sqrt{(q-a) (\psi(q) - \psi(a))}__- \sqrt{(q-a-h) (\psi(q) - \psi(a) - h_\psi'(a)) }
__ \\ &= \lim_{h_\to 0} \sqrt{(q-a) (\psi(q) -_\psi(a |
}(M)$ then $w\in C^{k,\alpha}_{\delta^\prime}(M)$ whenever $|\delta^\prime - 1|\leq \sqrt{1+c}$.
The tensor ${\mathcal{H}}_{{\overline}g}(\rho)$
-----------------------------------------------
Together with James Isenberg and John M. Lee, we introduced in [@WAH-Preliminary] a conformally invariant version of the trace-free Hessian that is used in [@AHEM-Preliminary] to characterize the shear-free condition; we now recall its definition and basic properties. Let $$A_{{\overline}g}(\rho)
= \frac{1}{2} |{d}\rho|_{{\overline}g} \operatorname{div}_{{\overline}g}\left[ |{d}\rho|_{{\overline}g}\operatorname{grad}_{{\overline}g}\rho\right].$$ We define the tensor field ${\mathcal{H}}_{{\overline}g}(\rho)$ by $$\label{DefineB}
{\mathcal{H}}_{{\overline}g}(\rho)
:=|{d}\rho|_{{\overline}g}^6\,\mathcal D_{{\overline}g}(|{d}\rho|^{-2}_{{\overline}g} \operatorname{grad}_{{\overline}g}\rho)
+ A_{{\overline}g}(\rho) \left( {d}\rho \otimes {d}\rho - \frac{1}{3}|{d}\rho|^2_{{\overline}g}\, {\overline}g \right),$$ where $\mathcal D_{{\overline}g}$ is the conformal Killing operator defined in.
We have the following basic properties of ${\mathcal{H}}_{{\overline}g}(\rho)$.
\[B-BasicProperties\]
1. ${\mathcal{H}}_{{\overline}g}(\rho)$ is symmetric and trace-free.
2. \[B-TransverseProperty\] ${\mathcal{H}}_{{\overline}g}(\rho)(\operatorname{grad}_{{\overline}g}\omega, \cdot)=0$.
3. \[B-Scaling\] ${\mathcal{H}}_{{\overline}g}(c \rho)=c^5{\mathcal{H}}_{{\overline}g}(\rho)$ for all constants $c$.
4. \[B-ConformalScaling\] If $\theta$ is a strictly positive function then ${\mathcal{H}}_{\theta^4{\overline}g}(\rho)=\theta^{-8}{\mathcal{H}}_{{\overline} | } (M)$ then $ w\in C^{k,\alpha}_{\delta^\prime}(M)$ whenever $ |\delta^\prime - 1|\leq \sqrt{1+c}$.
The tensor $ { \mathcal{H}}_{{\overline}g}(\rho)$
-----------------------------------------------
Together with James Isenberg and John M. Lee, we introduced in [ @WAH - Preliminary ] a conformally invariant version of the touch - spare Hessian that is used in [ @AHEM - Preliminary ] to characterize the shear - free circumstance; we now recall its definition and basic properties. permit $ $ A_{{\overline}g}(\rho)
= \frac{1}{2 } |{d}\rho|_{{\overline}g } \operatorname{div}_{{\overline}g}\left [ |{d}\rho|_{{\overline}g}\operatorname{grad}_{{\overline}g}\rho\right].$$ We define the tensor field $ { \mathcal{H}}_{{\overline}g}(\rho)$ by $ $ \label{DefineB }
{ \mathcal{H}}_{{\overline}g}(\rho)
: = |{d}\rho|_{{\overline}g}^6\,\mathcal D_{{\overline}g}(|{d}\rho|^{-2}_{{\overline}g } \operatorname{grad}_{{\overline}g}\rho)
+ A_{{\overline}g}(\rho) \left ({ d}\rho \otimes { d}\rho - \frac{1}{3}|{d}\rho|^2_{{\overline}g}\, { \overline}g \right),$$ where $ \mathcal D_{{\overline}g}$ is the conformal Killing hustler define in.
We have the keep up basic properties of $ { \mathcal{H}}_{{\overline}g}(\rho)$.
\[B - BasicProperties\ ]
1. $ { \mathcal{H}}_{{\overline}g}(\rho)$ is symmetrical and trace - free.
2. \[B - TransverseProperty\ ] $ { \mathcal{H}}_{{\overline}g}(\rho)(\operatorname{grad}_{{\overline}g}\omega, \cdot)=0$.
3. \[B - Scaling\ ] $ { \mathcal{H}}_{{\overline}g}(c \rho)=c^5{\mathcal{H}}_{{\overline}g}(\rho)$ for all constants $ c$.
4. \[B - ConformalScaling\ ] If $ \theta$ is a strictly positive function then $ { \mathcal{H}}_{\theta^4{\overline}g}(\rho)=\theta^{-8}{\mathcal{H}}_{{\overline } | }(M)$ tjen $w\in C^{k,\alpha}_{\delta^\primt}(M)$ whenever $|\delta^\krume - 1|\lxq \sqrt{1+d}$.
The tenror ${\mathcal{H}}_{{\overline}g}(\rho)$
-----------------------------------------------
Togevher with James Isenberg and Jofn M. Lee, wv introduxed mn [@WAH-Preliminarb] a conformallg invcrmant version of the trace-xree Hessian tvag ns used in [@AHEM-Preliminary] to characeerize yhf shear-free cogditpog; we now recall its definition and basjc proptrties. Let $$A_{{\overlime}g}(\rho)
= \frac{1}{2} |{d}\rho|_{{\overline}g} \opegatorname{div}_{{\overlije}g}\left[ |{d}\rhi|_{{\ovewoine}g}\operatofname{grad}_{{\onzrline}g}\rho\rjght].$$ We define the tensor field ${\matheal{H}}_{{\overling}y}(\eho)$ ty $$\label{DefmneB}
{\manhcal{H}}_{{\overlinc}b}(\rho)
:=|{d}\rvo|_{{\overlone}g}^6\,\mathcal D_{{\oyerliie}g}(|{d}\eho|^{-2}_{{\overline}g} \operatoriame{grad}_{{\overline}g}\rho)
+ A_{{\overlina}g}(\xho) \left( {d}\rho \otimes {e}\ryo - \ftac{1}{3}|{d}\rvo|^2_{{\ovdelive}g}\, {\otermine}g \gigit),$$ where $\mafhcal D_{{\overoine}g}$ is the conforkaj Killing operztor dqfyned in.
We have the following basic proptrtiea of ${\mathcal{H}}_{{\overline}g}(\ryo)$.
\[B-BasicProperties\]
1. ${\mwthcal{H}}_{{\ovqrline}g}(\rho)$ is symmetric and trace-free.
2. \[B-TransversaPropxrgy\] ${\nabhcau{Y}}_{{\ogerline}g}(\rho)(\operatorname{grad}_{{\overline}g}\omega, \cdoe)=0$.
3. \[N-Sbaling\] ${\mathcal{H}}_{{\ovcrline}g}(c \rho)=c^5{\mathcsl{J}}_{{\ofgrline}g}(\rho)$ for all consfants $c$.
4. \[B-ConformapScalind\] If $\rheta$ is w sttictly positive function thwn ${\mathcal{H}}_{\tketq^4{\overline}g}(\rho)=\theta^{-8}{\lathcal{H}}_{{\ovexline} | }(M)$ then $w\in C^{k,\alpha}_{\delta^\prime}(M)$ whenever $|\delta^\prime - The ${\mathcal{H}}_{{\overline}g}(\rho)$ ----------------------------------------------- with James Isenberg introduced [@WAH-Preliminary] a conformally version of the Hessian that is used in [@AHEM-Preliminary] characterize the shear-free condition; we now recall its definition and basic properties. Let = \frac{1}{2} |{d}\rho|_{{\overline}g} \operatorname{div}_{{\overline}g}\left[ |{d}\rho|_{{\overline}g}\operatorname{grad}_{{\overline}g}\rho\right].$$ We define the tensor field ${\mathcal{H}}_{{\overline}g}(\rho)$ by $$\label{DefineB} :=|{d}\rho|_{{\overline}g}^6\,\mathcal \operatorname{grad}_{{\overline}g}\rho) A_{{\overline}g}(\rho) {d}\rho \otimes {d}\rho - \frac{1}{3}|{d}\rho|^2_{{\overline}g}\, {\overline}g \right),$$ where $\mathcal D_{{\overline}g}$ is the conformal Killing operator defined We have the following basic properties of ${\mathcal{H}}_{{\overline}g}(\rho)$. 1. ${\mathcal{H}}_{{\overline}g}(\rho)$ is symmetric trace-free. 2. \[B-TransverseProperty\] ${\mathcal{H}}_{{\overline}g}(\rho)(\operatorname{grad}_{{\overline}g}\omega, \cdot)=0$. \[B-Scaling\] \rho)=c^5{\mathcal{H}}_{{\overline}g}(\rho)$ for constants 4. If $\theta$ is strictly positive function then ${\mathcal{H}}_{\theta^4{\overline}g}(\rho)=\theta^{-8}{\mathcal{H}}_{{\overline} | }(M)$ then $w\in C^{k,\alpha}_{\delta^\prime}(m)$ whenever $|\dElta^\pRimE - 1|\leQ \sQrt{1+c}$.
the tEnsor ${\mathcal{H}}_{{\oVErliNe}g}(\rho)$
-----------------------------------------------
Together with JameS IsenBeRG and jOhN M. Lee, We introDUcED In [@WaH-prEliMiNArY] a conForMally inVariant verSioN oF the trace-freE heSsian that iS usEd in [@AHEM-PrelImiNary] to ChAraCTerizE thE sheaR-free cONditioN; we now recAlL Its defINition aND BaSic pRoperties. Let $$A_{{\overLInE}G}(\rho)
= \frac{1}{2} |{d}\rho|_{{\ovErline}G} \oPErATOrnAme{Div}_{{\overlinE}g}\Left[ |{d}\RHo|_{{\overlINe}G}\OPEraTOrname{grad}_{{\oveRline}g}\rho\riGHt].$$ WE definE tHe tENsor fiEld ${\maThCAl{H}}_{{\Overline}g}(\rhO)$ by $$\lAbel{DefinEB}
{\mathCAl{H}}_{{\overLIne}g}(\rho)
:=|{D}\rho|_{{\ovErlIne}G}^6\,\matHCaL D_{{\OveRlINe}g}(|{D}\RhO|^{-2}_{{\ovERliNe}g} \operaToRnAme{grAd}_{{\ovERLINe}g}\rHo)
+ A_{{\OverLine}g}(\Rho) \left( {d}\rho \otImeS {d}\rhO - \FraC{1}{3}|{d}\rho|^2_{{\OverlIne}g}\, {\OvErlinE}g \righT),$$ wherE $\mAthcal D_{{\overline}G}$ is tHe conformAl KIlLinG oPeratOR definEd iN.
We Have the FollowiNG baSiC PROpErties of ${\mathcal{H}}_{{\ovErLINe}G}(\rho)$.
\[B-BasIcPropERtIeS\]
1. ${\Mathcal{H}}_{{\OvErlIne}g}(\RHO)$ is syMmetRIc And trace-Free.
2. \[B-TRAnSvErseProPeRty\] ${\matHcAl{H}}_{{\OveRline}G}(\Rho)(\oPeratoRname{graD}_{{\overLIne}g}\omega, \cdot)=0$.
3. \[B-sCaling\] ${\mathcal{h}}_{{\OvERLiNE}g}(c \rHo)=c^5{\Mathcal{H}}_{{\oveRlinE}G}(\rho)$ For aLL cOnsTAnts $c$.
4. \[b-ConfOrMAlsCaling\] If $\theta$ is a strIcTly posItive Function then ${\mAthcal{H}}_{\theTA^4{\OVerline}g}(\Rho)=\tHEtA^{-8}{\Mathcal{H}}_{{\overliNe} | }(M)$ then $w\in C^{k,\alp ha}_{\delt a^\pr ime }(M )$ whe neve r $|\delta^\pr i me - 1|\leq \sqrt{1+c}$.
The t en s or $ { \m athca l{H}}_{ { \o v e rli ne }g }(\ rh o )$
---- --- ------- ---------- --- -- ------------ - -- ---
Toget her with JamesIse nbergan d J o hn M. Le e, we intro d uced i n [@WAH-P re l iminar y ] a con f o rm ally invariant versio n o f the trace-fre e Hess ia n t h a t i s u sed in [@A HE M-Pre l iminary ] t o c har a cterize the s hear-free c o ndi tion;we no w recal l its d e fin ition and b asic properti es. Le t $$A_{{ \ overlin e}g}(\ rho ) = \f r ac {1 }{2 }| {d} \ rh o|_ { {\o verline} g} \ opera torn a m e { div} _{{ \ove rline }g}\left[ |{d }\r ho|_ { {\o verli ne}g} \ope ra torna me{gra d}_{{ \o verline}g}\rho\ righ t].$$ Wedef in e t he tens o r fiel d $ {\m athcal{ H}}_{{\ o ver li n e } g} (\rho)$ by $$\labe l{ D e fi neB}
{\m athcal { H} }_ { {\overli ne }g} (\rh o )
:=|{ d}\r h o| _{{\over line}g } ^6 \, \mathca lD_{{\o ve rli ne} g}(|{ d }\rh o|^{-2 }_{{\ove rline } g} \operatorna m e{grad}_{{\ov e rl i n e} g }\rh o)+ A_{{\over line } g}(\ rho) \l eft ( {d}\ rho \ ot i me s {d}\rho - \frac{1} {3 }|{d}\ rho|^ 2_{{\overline }g}\, {\ov e r l ine}g \r ight ) ,$ $ where $\mathc al D_ {{\overlin e }g}$ isthe c onformal Killingo p erator d efi ned in .
W e h ave the follo w i ng b as ic prop ert ies of${\ mat hca l{H }} _{{\overl ine}g}(\ rh o) $.
\[B -Basi c Properti es \]
1 . ${\ma t hcal{H }}_{{ \ove rl in e }g} (\rho)$ is s ymme tr ic and tr ac e-fre e.
2 . \[B-Tra nsversePr ope r ty\] $ {\ mathcal {H}}_{{\overl in e}g}(\rho) (\ ope ratorn a m e{grad}_ {{\overline}g}\omega, \ c dot)=0$ .
3. \ [B-S caling\]${\ mathca l{H } }_{{\o verlin e}g}( c\rh o ) =c^5{ \ m at hca l{ H}}_{{\ove r l ine }g}(\ rh o)$for all constants $c$.
4 . \ [B-ConformalS cal ing\ ] If $\ t he t a$is a s t r ictly positivefunction t he n $ {\mathcal{ H }}_ {\ theta^4 {\overl ine}g } (\rho)= \theta^{- 8}{\mathc al {H}} _ { {\o verline} | }(M)$ then_$w\in C^{k,\alpha}_{\delta^\prime}(M)$_whenever $|\delta^\prime - 1|\leq_\sqrt{1+c}$.
The tensor_${\mathcal{H}}_{{\overline}g}(\rho)$
-----------------------------------------------
Together_with James_Isenberg_and John M. Lee,_we introduced in_[@WAH-Preliminary] a conformally invariant_version of the_trace-free_Hessian that is used in [@AHEM-Preliminary] to characterize the shear-free condition; we now recall_its_definition and_basic_properties._Let $$A_{{\overline}g}(\rho)
= \frac{1}{2} |{d}\rho|_{{\overline}g}_\operatorname{div}_{{\overline}g}\left[ |{d}\rho|_{{\overline}g}\operatorname{grad}_{{\overline}g}\rho\right].$$ We define the_tensor field_${\mathcal{H}}_{{\overline}g}(\rho)$ by $$\label{DefineB}
{\mathcal{H}}_{{\overline}g}(\rho)
:=|{d}\rho|_{{\overline}g}^6\,\mathcal D_{{\overline}g}(|{d}\rho|^{-2}_{{\overline}g} \operatorname{grad}_{{\overline}g}\rho)
+ A_{{\overline}g}(\rho) \left( {d}\rho_\otimes_{d}\rho - \frac{1}{3}|{d}\rho|^2_{{\overline}g}\,_{\overline}g \right),$$ where $\mathcal D_{{\overline}g}$ is the conformal Killing_operator defined in.
We have the following_basic properties of_${\mathcal{H}}_{{\overline}g}(\rho)$.
\[B-BasicProperties\]
1.__${\mathcal{H}}_{{\overline}g}(\rho)$ is symmetric and_trace-free.
2. \[B-TransverseProperty\] ${\mathcal{H}}_{{\overline}g}(\rho)(\operatorname{grad}_{{\overline}g}\omega, \cdot)=0$.
3. _\[B-Scaling\] ${\mathcal{H}}_{{\overline}g}(c \rho)=c^5{\mathcal{H}}_{{\overline}g}(\rho)$ for all constants_$c$.
4. \[B-ConformalScaling\] If $\theta$ is a_strictly positive function then ${\mathcal{H}}_{\theta^4{\overline}g}(\rho)=\theta^{-8}{\mathcal{H}}_{{\overline} |
,i+1}$: $$\label{V0i+1SmallerV0i}
\sigma(v) \leq \left[3.001 - 0.1745448 \left(i+1\right)\right]d = \left(2.8264552 - 0.1745448i\right)d < \sigma(u).$$ Finally, to justify that there are no sum conflicts between vertices from $V_0$ and $V_1$, by (\[V1iSmallerV1i+1\]) and (\[V0i+1SmallerV0i\]) it is sufficient to show that sums in $V_{1,10^4}$ are smaller than sums in $V_{0,4}$. To see that this is actually true, note that by (\[SecondSv\]), (\[DefinitionOfc1i\]), (\[Delta2\]) and (\[LowerSvBoundinV0i\]), for any $u\in V_{1,10^4}$ and $v\in V_{0,4}$: $$\begin{aligned}
\sigma(u) &\leq& \left(1.0776+1.06875\cdot \left(10^4-1\right)\cdot 10^{-4}\right)d + 0.0018956d + 10^{-4}\cdot 1.0603 d \nonumber\\
&=& 2.148244755 d
< 2.1514248 d - 1
= \left(2.849604 - 0,1745448\cdot 4\right)d - 1
\leq \sigma(v). \nonumber\end{aligned}$$ This finishes the proof of Theorem \[123largeregTh\] as the obtained weighting $\omega_2$ is thus indeed a $3$-weighting of the edges of $G$ such that there are no sum conflicts between neighbours in $G$.
Concluding Remarks
==================
The constant $10^8$ above could still be improved, but at the cost of clarity of presentation of the proof of Theorem \[123largeregTh\]. Nevertheless, we were far from being able to push it down to $10^7$. Actually, introducing the special subgraph $G'_1$ of $G_1$, based on Corollary \[QuarterDecompositionLemma\], served merely optimization purposes. These required also using only $1$’s and $3$’s as | , i+1}$: $ $ \label{V0i+1SmallerV0i }
\sigma(v) \leq \left[3.001 - 0.1745448 \left(i+1\right)\right]d = \left(2.8264552 - 0.1745448i\right)d < \sigma(u).$$ Finally, to justify that there are no sum battle between vertex from $ V_0 $ and $ V_1 $, by (\[V1iSmallerV1i+1\ ]) and (\[V0i+1SmallerV0i\ ]) it is sufficient to show that sums in $ V_{1,10 ^ 4}$ are smaller than sum in $ V_{0,4}$. To see that this is actually true, notice that by (\[SecondSv\ ]), (\[DefinitionOfc1i\ ]), (\[Delta2\ ]) and (\[LowerSvBoundinV0i\ ]), for any $ u\in V_{1,10 ^ 4}$ and $ v\in V_{0,4}$: $ $ \begin{aligned }
\sigma(u) & \leq & \left(1.0776 + 1.06875\cdot \left(10 ^ 4 - 1\right)\cdot 10^{-4}\right)d + 0.0018956d + 10^{-4}\cdot 1.0603 d \nonumber\\
& = & 2.148244755 d
< 2.1514248 d - 1
= \left(2.849604 - 0,1745448\cdot 4\right)d - 1
\leq \sigma(v). \nonumber\end{aligned}$$ This finishes the validation of Theorem \[123largeregTh\ ] as the prevail weighting $ \omega_2 $ is therefore indeed a $ 3$-weighting of the edges of $ G$ such that there are no sum conflicts between neighbor in $ G$.
Concluding Remarks
= = = = = = = = = = = = = = = = = =
The constant $ 10 ^ 8 $ above could still be improved, but at the cost of clarity of display of the proof of Theorem \[123largeregTh\ ]. Nevertheless, we were far from being able to push it down to $ 10 ^ 7$. Actually, introduce the special subgraph $ G'_1 $ of $ G_1 $, based on Corollary \[QuarterDecompositionLemma\ ], serve merely optimization purposes. These necessitate also using only $ 1 $ ’s and $ 3 $ ’s as | ,i+1}$: $$\lwbel{V0i+1SmallerV0i}
\sigma(v) \led \left[3.001 - 0.1745448 \left(i+1\rntht)\rigit]d = \lert(2.8264552 - 0.1745448i\rigft)d < \sigma(u).$$ Finally, to justihy tyat tyere are no sum conflizts betwevn verticws fcom $V_0$ and $V_1$, by (\[V1iSmallerV1l+1\]) and (\[B0l+1SmalnxrV0i\]) it is sufflcient to svow that sums hn $V_{1,10^4}$ are smaller than sums in $V_{0,4}$. To see ehat thos is actually ttue, npee tgat by (\[SecondSv\]), (\[DefinitionOfc1i\]), (\[Deltz2\]) and (\[LoxerSvBoundinV0i\]), gor any $u\in V_{1,10^4}$ and $v\in V_{0,4}$: $$\behin{apigned}
\sigma(u) &\leq& \lfft(1.0776+1.06875\cdot \lefj(10^4-1\digrr)\cdot 10^{-4}\right)d + 0.0018956d + 10^{-4}\cdot 1.0603 d \nonumber\\
&=& 2.148244755 d
< 2.1514248 d - 1
= \left(2.849604 - 0,1745448\cdot 4\right)d - 1
\ldq \siyma(v). \nonumbgx\wnd{wnigned}$$ This finifhes the proof of Thaorem \[123latgeregTh\] as thc obteinee weighting $\omega_2$ is vhus indeed a $3$-weightyng of tha zdges of $G$ such that rhwre ate no sum xonwlidtx getweej nxighbours ih $G$.
Concludibg Remarks
==================
The constsne $10^8$ above could still bq improved, but at the cost of clarity ox pdesentation of the prood of Theorem \[123largeregTj\]. Neverthqless, we were far from being able to push it down to $10^7$. Ecguaoln, ingeofucing the special subgraph $G'_1$ of $G_1$, based on Cjdoklsry \[QuarterDecoipositionLekmw\], xgrved merely oktimizacjoh purposes. These rfquired also using onjy $1$’s and $3$’s as | ,i+1}$: $$\label{V0i+1SmallerV0i} \sigma(v) \leq \left[3.001 - 0.1745448 \left(2.8264552 0.1745448i\right)d < Finally, to justify conflicts vertices from $V_0$ $V_1$, by (\[V1iSmallerV1i+1\]) (\[V0i+1SmallerV0i\]) it is sufficient to show sums in $V_{1,10^4}$ are smaller than sums in $V_{0,4}$. To see that this actually true, note that by (\[SecondSv\]), (\[DefinitionOfc1i\]), (\[Delta2\]) and (\[LowerSvBoundinV0i\]), for any $u\in and V_{0,4}$: \sigma(u) \left(1.0776+1.06875\cdot \left(10^4-1\right)\cdot 10^{-4}\right)d + 0.0018956d + 10^{-4}\cdot 1.0603 d \nonumber\\ &=& 2.148244755 d < 2.1514248 d 1 = \left(2.849604 - 0,1745448\cdot 4\right)d - 1 \sigma(v). \nonumber\end{aligned}$$ This finishes proof of Theorem \[123largeregTh\] as obtained $\omega_2$ is indeed $3$-weighting the edges of such that there are no sum conflicts between neighbours in $G$. Concluding Remarks ================== The constant $10^8$ could still but at cost clarity presentation of the Theorem \[123largeregTh\]. Nevertheless, we were far to push it down to $10^7$. Actually, introducing special subgraph of $G_1$, based on Corollary \[QuarterDecompositionLemma\], merely optimization purposes. These required also using only and $3$’s as | ,i+1}$: $$\label{V0i+1SmallerV0i}
\sigma(v) \leQ \left[3.001 - 0.1745448 \left(i+1\Right)\RigHt]d = \LeFt(2.8264552 - 0.1745448i\rIght)D < \sigma(u).$$ Finally, TO jusTify that there are no sum cOnfliCtS BetwEEn VertiCes from $v_0$ AnD $v_1$, By (\[V1ISMaLleRV1I+1\]) AnD (\[V0i+1SmAllErV0i\]) it iS sufficienT to ShOw that sums in $v_{1,10^4}$ ArE smaller thAn sUms in $V_{0,4}$. To see tHat This is AcTuaLLy truE, noTe thaT by (\[SecONdSv\]), (\[DeFinitionOFc1I\]), (\[delta2\]) aND (\[LowerSVbOuNdinv0i\]), for any $u\in V_{1,10^4}$ and $v\iN v_{0,4}$: $$\bEGin{aligned}
\sigmA(u) &\leq& \lEfT(1.0776+1.06875\CdOT \LefT(10^4-1\riGht)\cdot 10^{-4}\rigHt)D + 0.0018956d + 10^{-4}\cdoT 1.0603 D \nonumbER\\
&=& 2.148244755 d
< 2.1514248 D - 1
= \LEFt(2.849604 - 0,1745448\cDOt 4\right)d - 1
\leq \siGma(v). \nonumbeR\End{AligneD}$$ THis FInisheS the pRoOF of theorem \[123largEregth\] as the obTained WEightinG $\Omega_2$ is Thus inDeeD a $3$-wEighTInG oF thE eDGes OF $G$ SucH ThaT there arE nO sUm conFlicTS BETweeN neIghbOurs iN $G$.
Concluding REmaRks
==================
THE coNstanT $10^8$ abovE couLd Still Be imprOved, bUt At the cost of clarIty oF presentaTioN oF thE pRoof oF theoreM \[123laRgeRegTh\]. NeVerthelESs, wE wERE FaR from being able to puSh IT DoWn to $10^7$. ActuAlly, inTRoDuCIng the spEcIal SubgRAPh $G'_1$ of $g_1$, basED oN CorollaRy \[QuarTErdeComposiTiOnLemmA\], sErvEd mErely OPtimIzatioN purposeS. ThesE Required also usINg only $1$’s and $3$’s as | ,i+1}$: $$\label{V0i+1Smal lerV0i}
\s igma( v)\le q\lef t[3. 001 - 0.174544 8 \le ft(i+1\right)\right]d= \le ft ( 2.82 6 45 52 -0.17454 4 8i \ r igh t) d< \ si g ma (u).$ $ F inally, to justif y t ha t there aren osum confli cts between ver tic es fro m$V_ 0 $ and $V _1$,by (\[ V 1iSmal lerV1i+1\ ]) and (\ [ V0i+1Sm a l le rV0i \]) it is suffici e nt to show that s ums in $ V _{ 1 , 10^ 4}$ are small er than sums in $V _ { 0 ,4} $ . To see that this is ac t ual ly tru e, no t e that by ( \[ S eco ndSv\]), (\ [Def initionOf c1i\]) , (\[Del t a2\]) a nd (\[ Low erS vBou n di nV 0i\ ]) , fo r a ny$ u\i n V_{1,1 0^ 4} $ and $v\ i n V _{0, 4}$ : $$ \begi n{aligned}
\s igm a(u) &\l eq& \ left( 1.07 76 +1.06 875\cd ot \l ef t(10^4-1\right) \cdo t 10^{-4} \ri gh t)d + 0.00 1 8956d+ 1 0^{ -4}\cdo t 1.060 3 d\n o n u mb er\\
&=& 2.1482447 55 d
< 2.1514 248 d- 1
= \left(2 .8 496 04 - 0 ,1745 448\ c do t 4\righ t)d -1 \l eq \sig ma (v). \ no num ber \end{ a lign ed}$$This fin ishes the proof of T h eorem \[123la r ge r e gT h \] a s t he obtained wei g htin g $\ o me ga_ 2 $ isthusin d ee d a $3$-weighting of t he edg es of $G$ such tha t there ar e n o sum co nfli c ts between neighb oursin $G$.
C o ncluding Rema rks
==== ========= = = ===
The co nst ant $1 0 ^ 8$ above coulds t illbe improv ed, but at th e c ost of c larity of present at io nof th e pro o f of The or em\[ 123 large r egTh\] . Nev erth el es s , w e weref ar f rombe in g ab leto push itd own to $10 ^7$. Actu all y , in tr od ucing t he special su bg raph $G'_1 $of$G_1$, b ased onCorollary \[QuarterDeco m positio nLe mma\] , se rved mere lyoptimi zat i on pur poses. Thes ereq u i red a l s ousi ng only $1$’ s and $3$’ sas | ,i+1}$: $$\label{V0i+1SmallerV0i}
\sigma(v)_\leq \left[3.001_- 0.1745448 \left(i+1\right)\right]d =_\left(2.8264552 -_0.1745448i\right)d_< \sigma(u).$$_Finally,_to justify that_there are no_sum conflicts between vertices_from $V_0$ and_$V_1$,_by (\[V1iSmallerV1i+1\]) and (\[V0i+1SmallerV0i\]) it is sufficient to show that sums in $V_{1,10^4}$ are smaller than_sums_in $V_{0,4}$._To_see_that this is actually true,_note that by (\[SecondSv\]), (\[DefinitionOfc1i\]), (\[Delta2\])_and (\[LowerSvBoundinV0i\]), for_any $u\in V_{1,10^4}$ and $v\in V_{0,4}$: $$\begin{aligned}
\sigma(u) &\leq&_\left(1.0776+1.06875\cdot_\left(10^4-1\right)\cdot 10^{-4}\right)d +_0.0018956d + 10^{-4}\cdot 1.0603 d \nonumber\\
&=& 2.148244755 d
< _2.1514248 d - 1
= \left(2.849604_- 0,1745448\cdot 4\right)d_-_1_
\leq \sigma(v). \nonumber\end{aligned}$$ This_finishes the proof of Theorem \[123largeregTh\] as_the obtained weighting $\omega_2$ is thus_indeed a $3$-weighting of the edges of_$G$ such that there are no_sum conflicts between neighbours in_$G$.
Concluding Remarks
==================
The_constant $10^8$ above could still_be improved, but_at the_cost of clarity_of presentation of the proof of_Theorem \[123largeregTh\]. Nevertheless, we_were far from being able to_push_it down to_$10^7$._Actually,_introducing the_special subgraph $G'_1$_of_$G_1$, based_on_Corollary \[QuarterDecompositionLemma\], served merely optimization purposes. These_required_also using only $1$’s and $3$’s as |
H. Freeman and Co., 1978.
Monnier,J. D. & Millan-Gabet, R. 2002, 579, 694
Vinkovi[ć]{}, D., Ivezi[ć]{}, [Ž]{}., Jurki[ć]{}, T., & Elitzur, M. 2006,, 636, 348
Wolf, S. 2003,, 582, 859
[^1]: Typically, the term [*multigrain*]{} also includes all other dust grain properties, like the grain shape and chemistry. But, for simplicity, we use this term only to describe grain size effects.
[^2]: Another type of temperature inversion has been recognized in protoplanetary disks, where additional viscous heating can increase the disk interior temperature [@Calvet; @Malbet]. In contrast, the temperature inversion discussed here is a pure radiative transfer effect and does not require any additional assumption (like disk viscosity) to operate.
---
abstract: 'The weak spin-orbit interaction in graphene was predicted to be increased, [*e.g.*]{}, by hydrogenation. This should result in a sizable spin Hall effect (SHE). We employ two different methods to examine the spin Hall effect in weakly hydrogenated graphene. For hydrogenation we expose graphene to a hydrogen plasma and use Raman spectroscopy to characterize this method. We then investigate the SHE of hydrogenated graphene in the H-bar method and by direct measurements of the inverse SHE. Although a large nonlocal resistance can be observed in the H-bar structure, comparison with the results of the other method indicate that this nonlocal resistance is caused by a non-spin-related origin.'
author:
- Tobias Völkl
- Denis Kochan
- Thomas Ebnet
- Sebastian Ringer
- Daniel Schiermeier
- Philipp Nagler
- Tobias Korn
- Christian Schüller
- Jaroslav Fabian
- Dieter Weiss
- Jonathan Eroms
title: 'Absence of a giant spin Hall effect in plasma-hydrogenated graphene'
---
Introduction
============
Covalently bonded hydrogen was predicted to significantly increase the spin-orbit coupling (SOC) of graphene by Castro Neto and Guinea[@CastroNeto2009]. However, experimental results on this were conflicting. Balakrishnan *et al.* reported a high nonlocal resistance in weakly hydrogenated graphene in the so called H | H. Freeman and Co., 1978.
Monnier, J. D. & Millan - Gabet, R. 2002, 579, 694
Vinkovi[ć ] { }, D., Ivezi[ć ] { }, [ Ž ] { }. , Jurki[ć ] { }, T., & Elitzur, M. 2006, , 636, 348
Wolf, S. 2003, , 582, 859
[ ^1 ]: Typically, the term [ * multigrain * ] { } also includes all other debris texture property, like the grain shape and chemistry. But, for chasteness, we use this term entirely to trace grain size effects.
[ ^2 ]: Another type of temperature inversion has been spot in protoplanetary disks, where additional viscous heating can increase the disk interior temperature [ @Calvet; @Malbet ]. In line, the temperature inversion discussed here is a pure radiative transportation effect and does not require any additional assumption (like disk viscosity) to function.
---
abstract:' The weak spin - orbit interaction in graphene was predicted to be increased, [ * for example * ] { }, by hydrogenation. This should result in a sizable spin Hall effect (SHE). We employ two different method to examine the spin Hall effect in weakly hydrogenated graphene. For hydrogenation we expose graphene to a hydrogen plasma and use Raman spectroscopy to characterize this method. We then investigate the SHE of hydrogenated graphene in the H - barroom method and by direct measurement of the inverse SHE. Although a big nonlocal resistor can be observed in the H - bar social organization, comparison with the results of the early method indicate that this nonlocal resistance is cause by a non - spin - related origin.'
author:
- Tobias Völkl
- Denis Kochan
- Thomas Ebnet
- Sebastian Ringer
- Daniel Schiermeier
- Philipp Nagler
- Tobias Korn
- Christian Schüller
- Jaroslav Fabian
- Dieter Weiss
- Jonathan Eroms
title:' Absence of a giant spin Hall effect in plasma - hydrogenated graphene'
---
Introduction
= = = = = = = = = = = =
Covalently bonded hydrogen was predicted to importantly increase the spin - sphere coupling (SOC) of graphene by Castro Neto and Guinea[@CastroNeto2009 ]. However, experimental results on this were conflicting. Balakrishnan * et al. * reported a high nonlocal underground in weakly hydrogenated graphene in the so call H | H. Frfeman and Co., 1978.
Monnier,J. D. & Oillan-Gabet, R. 2002, 579, 694
Runkovi[ć]{}, D., Iveai[ć]{}, [Ž]{}., Jurkk[ć]{}, T., & Elitzur, M. 2006,, 636, 348
Wolf, S. 2003,, 582, 859
[^1]: Typmcaloy, tht term [*multigrain*]{} auso inclufes all ithec dust grain pro'srties, like tgc grann shape and chekistry. But, for simplicitf, de use this term only to describe grayn size evfects.
[^2]: Another typt os tejievature inversion has been recognjzed in protoplanetaty disks, where additional gisclus heating can infrease the eisk unterior temoerature [@Calvet; @Malbet]. In contrast, the temperature inxersipn discussgb hegg is a pure cadiatpve transfer cgfect dnd doex not require sny adeitional assumption (lmke disk viscosity) tj operate.
---
aystract: 'The weak spin-irvit itterdctiub iv gda'hehe was prxdicted to ge increasee, [*e.g.*]{}, by hydrogenatipn. Nnis should rssult yn a sizable spin Hall effect (SHE). We emplmy fwo different methods ti examine the spin Hapl effect in weakly hydrogenated graphene. For hydrogenatiot we xxoost nvaphdbe to a hydrogen plasma and use Raman spectroscjly tp characterize this methoc. Ae jhen investigaje the SHE of hydrogenated ggaphene in tye H-bar mtthod and by direct measurements of the invegse WHE. Although a larye nonlocal xesistsnce van be observed in the K-bar sfructure, colparison skth the results uf nhe mther meufod indicate that this nonoocak resisgancg is catsed by a jon-splt-related origin.'
autjor:
- Tlbhas Völkl
- Dfnis Kochan
- Thomas Ebnet
- Sebastmen Ringer
- Danoen Sbhiermeiex
- Phillpp Nagler
- Tobiws Korn
- Christnan Schüljer
- Jxroslav Fagian
- Dixter Weiss
- Jjnathan Eroms
dltle: 'Absence of a giwnt wpin Hall ewwect in plasma-nydrogenaneb graphenw'
---
Introduction
============
Covalcntly gonded hydrogen waw predicted to xigvifycwnvly igwrease the s[in-ofbig coupuing (SOC) of erapnene by Castro Neto dnd Fuinea[@CastroNeto2009]. Hpwcver, expetimental wesults on thos were conflictinh. Balekrishian *et sl.* teported a high nonlocal resistznce in wfakjy hydrogenaeed nraprene in thz so called H | H. Freeman and Co., 1978. Monnier,J. D. R. 579, 694 D., Ivezi[ć]{}, [Ž]{}., 2006,, 348 Wolf, S. 582, 859 [^1]: the term [*multigrain*]{} also includes all dust grain properties, like the grain shape and chemistry. But, for simplicity, we this term only to describe grain size effects. [^2]: Another type of temperature has recognized protoplanetary where additional viscous heating can increase the disk interior temperature [@Calvet; @Malbet]. In contrast, the temperature discussed here is a pure radiative transfer effect does not require any assumption (like disk viscosity) to --- 'The weak interaction graphene predicted to be [*e.g.*]{}, by hydrogenation. This should result in a sizable spin Hall effect (SHE). We employ two different to examine Hall effect weakly graphene. hydrogenation we expose a hydrogen plasma and use Raman this method. We then investigate the SHE of graphene in H-bar method and by direct measurements the inverse SHE. Although a large nonlocal resistance be observed in the H-bar structure, comparison with the results of the other method indicate nonlocal resistance is caused a non-spin-related origin.' - Völkl Denis - Thomas - Sebastian Ringer - Daniel Schiermeier - Philipp Nagler - Tobias - Christian Schüller - Jaroslav Fabian - Dieter Weiss - title: of a giant Hall effect in plasma-hydrogenated --- ============ Covalently bonded hydrogen to increase (SOC) graphene Castro Neto and Guinea[@CastroNeto2009]. experimental results on this were Balakrishnan *et al.* reported weakly hydrogenated graphene in the so called H | H. Freeman and Co., 1978.
Monnier,J. D. & MilLan-Gabet, R. 2002, 579, 694
VInkovI[ć]{}, D., iveZi[Ć]{}, [Ž]{}., JuRki[ć]{}, t., & Elitzur, M. 2006,, 636, 348
Wolf, S. 2003,, 582, 859
[^1]: tYpicAlly, the term [*multigrain*]{} aLso inClUDes aLL oTher dUst graiN PrOPErtIeS, lIke ThE GrAin shApe And chemIstry. But, foR siMpLicity, we use tHIs Term only to DesCribe grain siZe eFfects.
[^2]: anOthER type Of tEmperAture iNVersioN has been rEcOGnized IN protopLANeTary Disks, where additioNAl VIscous heating cAn incrEaSE tHE DisK inTerior tempErAture [@cAlvet; @MaLBeT]. iN ConTRast, the temperAture inversIOn dIscussEd HerE Is a purE radiAtIVe tRansfer effeCt anD does not rEquire ANy additIOnal assUmptioN (liKe dIsk vIScOsIty) To OPerATe.
---
AbsTRacT: 'The weak SpIn-Orbit InteRACTIon iN grApheNe was Predicted to be IncReasED, [*e.g.*]{}, By hydRogenAtioN. THis shOuld reSult iN a Sizable spin Hall EffeCt (SHE). We emPloY tWo dIfFerenT MethodS to ExaMine the Spin HalL EffEcT IN WeAkly hydrogenated grApHENe. for hydroGenatiON wE eXPose grapHeNe tO a hyDROgen pLasmA AnD use RamaN spectROsCoPy to chaRaCterizE tHis MetHod. We THen iNvestiGate the ShE of hYDrogenated grapHEne in the H-bar mEThOD AnD By diRecT measuremenTs of THe inVersE sHe. AlTHough A largE nONlOCal resistance can be oBsErved iN the H-Bar structure, cOmparison wITH The resulTs of THe OTher method indiCate tHat this nonLOcal resiStancE is causeD by a non-spIN-Related oRigIn.'
aUthOr:
- TOBIaS Völkl
- Denis KoCHAn
- ThOmAs Ebnet
- sebAstian RIngEr
- DAniEl SChIermeier
- PHilipp NaGlEr
- toBiAs KOrn
- ChRIstian ScHüLleR
- JAroSlav FABian
- DiEter WEiss
- joNaTHan eroms
tiTLe: 'aBSencE oF a GianT spIn hall eFfecT In pLasma-hyDrogenateD grAPhenE'
---
INtRoductiOn
============
Covalently bOnDed hydrogeN wAs pRedictED To signifIcantly increase the spin-oRBit coupLinG (SOC) oF graPhene by CaStrO Neto aNd GUInea[@CaStroNeTo2009]. HowEvEr, eXPErimeNTAl ResUlTs on this weRE ConFlictInG. BalAkrishnAn *et al.* reported a higH NonLocal resistanCe iN weaKLY hYdrOGeNAteD gRAphENE in the so called H | H. Freeman and Co., 1978.
Monnier, J. D. &Mil la n-Ga bet, R. 2002, 579, 694
Vinkovi[ć]{}, D., Ive zi[ć] {} , [Ž] { }. , Jur ki[ć]{} , T . , &El it zur ,M .2006, , 6 36, 348
Wolf, S. 20 03 ,, 582, 859[^ 1]: Typica lly , the term [ *mu ltigra in *]{ } also in clude s allo ther d ust grain p r operti e s, like t he gra in shape and chem i st r y. But, for si mplici ty , w e use th is term on ly to d e scribeg ra i n siz e effects.
[^ 2]: Another typ e of t em per a ture i nvers io n ha s been reco gniz ed in pro toplan e tary di s ks, whe re add iti ona l vi s co us he at i ngc an in c rea se the d is kinter iort e m p erat ure [@C alvet ; @Malbet]. I n c ontr a st, thetempe ratu re inve rsiondiscu ss ed here is a pu re r adiativetra ns fer e ffect and do esnot requir e any a d dit io n a l a ssumption (like di sk v is cosity)to ope r at e. ---
abs tr act : 'T h e weak spi n -o rbit int eracti o nin graphe ne was p re dic ted to b e inc reased , [*e.g. *]{}, by hydrogenati o n. This shoul d r e s ul t ina s izable spin Hal l eff ect( SH E). We em ploytw o d i fferent methods toex aminethe s pin Hall effe ct in weak l y hydrogen ated gr a phene. For hyd rogen ation we e x pose gra phene to a hy drogen pl a s ma and u seRam anspe c t ro scopy to char a c teri ze this m eth od. Wethe n i nve sti ga te the SH E of hyd ro ge na te d g raphe n e in the H -ba rmet hod a n d by d irect mea su re m ent s of th e i n v erse S HE . Al tho ug h a l arge non local r esistance ca n beob se rved in the H-bar st ru cture, com pa ris on wit h the resu lts of the other method indicat e t hat t hisnonlocalres istanc e i s cause d by a non- sp in- r e lated o ri gin .'
author:
- T obi as Vö lk l
-Denis K ochan
- Thomas Ebn e t
- Sebastian Ri nge r
-D a ni elS ch i erm ei e r
- P hilipp Nagler
- Tobias Ko rn -ChristianS chü ll er
- Ja roslavFabia n
- Diet er Weiss- Jonatha nErom s tit le: 'Absen ce of agiant spi n Hall ef fectinplasma -h ydr ogena ted gr a phe ne'
- --
In tr oducti on
== == ========
Covalently bonded hyd rogenwas p red icted tosig n ifi cantly in crea se the spi n-o rbi t cou pli n g (SO C) o f g rap h ene b y Ca s tro Netoa nd Gu i n ea [@CastroNet o 2 0 09] . How eve r , expe rime ntal results on t h is were confli ctin g . Ba lak r ishn an *et al.* rep ort ed a high no nl ocal resist ance inwe a kly h ydroge natedgraphen e in the so cal led H | H. Freeman and_Co., 1978.
Monnier,J. D._& Millan-Gabet, R. 2002, 579,_694
Vinkovi[ć]{}, D.,_Ivezi[ć]{},_[Ž]{}., Jurki[ć]{},_T.,_& Elitzur, M. 2006,,_636, 348
Wolf, S. 2003,,_582, 859
[^1]: Typically, the_term [*multigrain*]{} also_includes_all other dust grain properties, like the grain shape and chemistry. But, for simplicity,_we_use this_term_only_to describe grain size effects.
[^2]:_Another type of temperature inversion_has been_recognized in protoplanetary disks, where additional viscous heating_can_increase the disk_interior temperature [@Calvet; @Malbet]. In contrast, the temperature inversion_discussed here is a pure radiative_transfer effect and_does_not_require any additional assumption_(like disk viscosity) to operate.
---
abstract:_'The weak spin-orbit interaction in graphene_was predicted to be increased, [*e.g.*]{}, by_hydrogenation. This should result in a_sizable spin Hall effect (SHE)._We employ_two different methods to examine_the spin Hall_effect in_weakly hydrogenated graphene._For hydrogenation we expose graphene to_a hydrogen plasma_and use Raman spectroscopy to characterize_this_method. We then_investigate_the_SHE of_hydrogenated graphene in_the_H-bar method_and_by direct measurements of the inverse_SHE._Although a large nonlocal resistance can be_observed in the H-bar_structure,_comparison with the results_of the other method indicate_that this nonlocal resistance is caused_by a_non-spin-related origin.'
author:
-_Tobias Völkl
- Denis Kochan
- Thomas Ebnet
- Sebastian Ringer
- Daniel Schiermeier
- Philipp_Nagler
- Tobias Korn
- Christian Schüller
- Jaroslav_Fabian
- Dieter Weiss
- Jonathan_Eroms
title: 'Absence_of_a giant spin_Hall_effect in_plasma-hydrogenated graphene'
---
Introduction
============
Covalently bonded hydrogen was predicted to_significantly increase_the spin-orbit coupling (SOC) of graphene_by Castro Neto and_Guinea[@CastroNeto2009]._However, experimental results on this were_conflicting. Balakrishnan *et al.* reported a high_nonlocal resistance in weakly hydrogenated_graphene_in_the so called H |
{K}(\nu_2)$ diverge as $n_0\to 1$. In this limit Eq. (\[nu122\]) yields $\text{K}(\nu_2)\simeq\text{K}(\nu_1)\Delta$. The explicit $n_0$-dependence can be obtained with the help of Eq. (\[mass2d\]): $2\text{K}(\nu_1) \simeq (1+\Delta)(\Delta\delta)^{-1}$, and $2\text{K}(\nu_2) \simeq (1+\Delta)(\delta)^{-1}$. Using the asymptotic $\text{sn}\,z \simeq \tanh z$, we obtain in the leading order: $$\begin{aligned}
\label{hyp}
fg&\simeq&\tanh\left[\frac{(1+\Delta)x}{\Delta\delta}\right]\tanh\left[\frac{(1+\Delta)y}{\Delta\delta}\right] \nonumber\\
&=&\frac{\tanh\left[\frac{(1+\Delta)x}{\Delta\delta}\right]+\tanh\left[\frac{(1+\Delta)y}{\Delta\delta}\right]-1}
{\tanh\left[\frac{(1+\Delta)(x+y)}{\Delta\delta}\right]}.\end{aligned}$$ This asymptotic is valid for $x<1/2$ and $y<\Delta/2$; it can be extended to the other three quarters of the rectangle by obvious reflections. Away from the domain corners we can replace $\tanh(\dots)$ in the denominator of Eq. (\[hyp\]) by unity. Using the resulting “approximate product rule" in Eq. (\[q\]), we obtain, after some algebra, the “kink" asymptotic of $q(x,y)$ away from the domain corners: $$\label{qaprox}
q\simeq \tanh^2\left[\frac{2(1+\Delta)x}{\Delta\delta}\right]\tanh^2\left[\frac{2(1+\Delta)y}{\Delta\delta}\right],$$ where $x<1/2$ and $y<\Delta/2$, and reflected formulas in the other three rectangle quarters. These asymptotics describe kinks with the characteristic width $\ell=\Delta\delta/[2(1+\Delta)]$. By analogy with one dimension, the action per unit length along the boundary is, in the leading order, $ | { K}(\nu_2)$ diverge as $ n_0\to 1$. In this limit Eq. (\[nu122\ ]) yields $ \text{K}(\nu_2)\simeq\text{K}(\nu_1)\Delta$. The explicit $ n_0$-dependence can be prevail with the aid of Eq. (\[mass2d\ ] ): $ 2\text{K}(\nu_1) \simeq (1+\Delta)(\Delta\delta)^{-1}$, and $ 2\text{K}(\nu_2) \simeq (1+\Delta)(\delta)^{-1}$. Using the asymptotic $ \text{sn}\,z \simeq \tanh z$, we obtain in the leading orderliness: $ $ \begin{aligned }
\label{hyp }
fg&\simeq&\tanh\left[\frac{(1+\Delta)x}{\Delta\delta}\right]\tanh\left[\frac{(1+\Delta)y}{\Delta\delta}\right ] \nonumber\\
& = & \frac{\tanh\left[\frac{(1+\Delta)x}{\Delta\delta}\right]+\tanh\left[\frac{(1+\Delta)y}{\Delta\delta}\right]-1 }
{ \tanh\left[\frac{(1+\Delta)(x+y)}{\Delta\delta}\right]}.\end{aligned}$$ This asymptotic is valid for $ x<1/2 $ and $ y<\Delta/2 $; it can be extended to the other three quarter of the rectangle by obvious mirror image. Away from the sphere corners we can replace $ \tanh(\dots)$ in the denominator of Eq. (\[hyp\ ]) by oneness. Using the resulting “ approximate merchandise rule " in Eq. (\[q\ ]), we obtain, after some algebra, the “ kink " asymptotic of $ q(x, y)$ away from the domain corner: $ $ \label{qaprox }
q\simeq \tanh^2\left[\frac{2(1+\Delta)x}{\Delta\delta}\right]\tanh^2\left[\frac{2(1+\Delta)y}{\Delta\delta}\right],$$ where $ x<1/2 $ and $ y<\Delta/2 $, and reflected formulas in the other three rectangle quarter. These asymptotics describe kinks with the characteristic width $ \ell=\Delta\delta/[2(1+\Delta)]$. By analogy with one dimension, the military action per unit length along the boundary is, in the leading order, $ | {K}(\nu_2)$ diverge as $n_0\to 1$. In this limit Eq. (\[nu122\]) yields $\texv{K}(\nu_2)\simsq\text{K}(\nj_1)\Delta$. The explicit $n_0$-dependeice xan bt obtained with the help of Vq. (\[mass2d\]): $2\twxt{K}(\iu_1) \simeq (1+\Delta)(\Delta\delta)^{-1}$, and $2\tswt{K}(\nu_2) \wimeq (1+\Delta)(\delja)^{-1}$. Using the asymptotic $\teft{rn}\,v \simeq \tanh z$, we obtain in the leadyng ordrr: $$\begin{aligned}
\lwbel{nrp}
fg&\apmtq&\tanh\left[\frac{(1+\Delta)x}{\Delta\delta}\riggt]\tanh\ltft[\frac{(1+\Delta)y}{\Delta\celta}\right] \nonumber\\
&=&\frac{\tanj\lefh[\frac{(1+\Delta)x}{\Delta\depta}\right]+\tany\lefe[\drac{(1+\Delta)y}{\Deuta\delta}\right]-1}
{\tanh\left[\ftac{(1+\Delta)(x+y)}{\Delta\delta}\right]}.\end{aligved}$$ Tkis asymptojnx id valid for $e<1/2$ and $j<\Delta/2$; it can be extetded to the other thrce quertees of the rectangle bb obvious reflectionf. Away frmm the domain cornees we cdn raplazw $\txnh(\sovs)$ jn the deiominator or Eq. (\[hyp\]) by ynity. Using the resilepmg “approximafe proqust rule" in Eq. (\[q\]), we obtain, after some algtbra, fhe “kink" asymptotic of $w(x,y)$ away from the domwin cornews: $$\label{qaprox}
q\simeq \tanh^2\left[\frac{2(1+\Delta)x}{\Delta\delda}\rigit]\gank^2\oeft[\ffqc{2(1+\Felta)y}{\Delta\delta}\right],$$ where $x<1/2$ and $y<\Delta/2$, and wsfkebted formulas in bhe other three revtwnbje quarters. Tfese asymltotics describe klnks wijh the charactewistoc width $\ell=\Delta\delta/[2(1+\Delta)]$. By analogy city one dimension, thz action per unij lengyh along the boundary ir, in the leadinh order, $ | {K}(\nu_2)$ diverge as $n_0\to 1$. In this (\[nu122\]) $\text{K}(\nu_2)\simeq\text{K}(\nu_1)\Delta$. The $n_0$-dependence can be Eq. $2\text{K}(\nu_1) \simeq (1+\Delta)(\Delta\delta)^{-1}$, $2\text{K}(\nu_2) \simeq (1+\Delta)(\delta)^{-1}$. the asymptotic $\text{sn}\,z \simeq \tanh z$, obtain in the leading order: $$\begin{aligned} \label{hyp} fg&\simeq&\tanh\left[\frac{(1+\Delta)x}{\Delta\delta}\right]\tanh\left[\frac{(1+\Delta)y}{\Delta\delta}\right] \nonumber\\ &=&\frac{\tanh\left[\frac{(1+\Delta)x}{\Delta\delta}\right]+\tanh\left[\frac{(1+\Delta)y}{\Delta\delta}\right]-1} {\tanh\left[\frac{(1+\Delta)(x+y)}{\Delta\delta}\right]}.\end{aligned}$$ This asymptotic valid for $x<1/2$ and $y<\Delta/2$; it can be extended to the other three of rectangle obvious Away from the domain corners we can replace $\tanh(\dots)$ in the denominator of Eq. (\[hyp\]) by Using the resulting “approximate product rule" in Eq. we obtain, after some the “kink" asymptotic of $q(x,y)$ from domain corners: q\simeq where and $y<\Delta/2$, and formulas in the other three rectangle quarters. These asymptotics describe kinks with the characteristic width $\ell=\Delta\delta/[2(1+\Delta)]$. By with one action per length the is, in the $ | {K}(\nu_2)$ diverge as $n_0\to 1$. In this limiT Eq. (\[nu122\]) yieldS $\text{k}(\nu_2)\SimEq\Text{k}(\nu_1)\DElta$. The expliciT $N_0$-depEndence can be obtained wiTh the HeLP of EQ. (\[MaSs2d\]): $2\teXt{K}(\nu_1) \siMEq (1+\dELta)(\deLtA\deLtA)^{-1}$, AnD $2\text{k}(\nu_2) \Simeq (1+\DeLta)(\delta)^{-1}$. UsIng ThE asymptotic $\tEXt{Sn}\,z \simeq \taNh z$, We obtain in thE leAding oRdEr: $$\bEGin{alIgnEd}
\labEl{hyp}
fG&\Simeq&\tAnh\left[\frAc{(1+\dElta)x}{\DELta\deltA}\RIgHt]\taNh\left[\frac{(1+\Delta)y}{\DELtA\Delta}\right] \nonuMber\\
&=&\frAc{\TAnH\LEft[\FraC{(1+\Delta)x}{\DelTa\Delta}\RIght]+\tanH\LeFT[\FRac{(1+\dElta)y}{\Delta\delTa}\right]-1}
{\tanh\LEft[\Frac{(1+\DeLtA)(x+y)}{\dElta\deLta}\riGhT]}.\End{Aligned}$$ This AsymPtotic is vAlid foR $X<1/2$ and $y<\DeLTa/2$; it can Be exteNdeD to The oTHeR tHreE qUArtERs Of tHE reCtangle bY oBvIous rEfleCTIONs. AwAy fRom tHe domAin corners we cAn rEplaCE $\taNh(\dotS)$ in thE denOmInatoR of Eq. (\[hYp\]) by uNiTy. Using the resulTing “ApproximaTe pRoDucT rUle" in eQ. (\[q\]), we obTaiN, afTer some Algebra, THe “kInK" ASYmPtotic of $q(x,y)$ away froM tHE DoMain cornErs: $$\labEL{qApROx}
q\simeq \TaNh^2\lEft[\fRAC{2(1+\DeltA)x}{\DeLTa\Delta}\rigHt]\tanh^2\LEfT[\fRac{2(1+\DeltA)y}{\delta\dElTa}\rIghT],$$ wherE $X<1/2$ and $Y<\Delta/2$, And refleCted fORmulas in the othER three rectangLE qUARtERs. ThEse Asymptotics DescRIbe kInks WItH thE CharaCteriStIC wIDth $\ell=\Delta\delta/[2(1+\DelTa)]$. by analOgy wiTh one dimensioN, the action PER Unit lengTh alONg THe boundary is, in The leAding order, $ | {K}(\nu_2)$ diverge as $n_ 0\to 1$. I n thi s l imi tEq.(\[n u122\]) yields $\te xt{K}(\nu_2)\simeq\tex t{K}( \n u _1)\ D el ta$.The exp l ic i t $n _0 $- dep en d en ce ca n b e obtai ned with t hehe lp of Eq. (\ [ ma ss2d\]): $ 2\t ext{K}(\nu_1 ) \ simeq(1 +\D e lta)( \De lta\d elta)^ { -1}$,and $2\te xt { K}(\nu _ 2) \sim e q ( 1+\D elta)(\delta)^{-1 } $. Using the asym ptotic $ \ te x t {sn }\, z \simeq \ ta nh z$ , we obt a in i n th e leading orde r: $$\begin { ali gned}\l abe l {hyp}fg&\s im e q&\ tanh\left[\ frac {(1+\Delt a)x}{\ D elta\de l ta}\rig ht]\ta nh\ lef t[\f r ac {( 1+\ De l ta) y }{ \De l ta\ delta}\r ig ht ] \no numb e r \ \
&=& \fr ac{\ tanh\ left[\frac{(1 +\D elta ) x}{ \Delt a\del ta}\ ri ght]+ \tanh\ left[ \f rac{(1+\Delta)y }{\D elta\delt a}\ ri ght ]- 1}
{\ t anh\le ft[ \fr ac{(1+\ Delta)( x +y) }{ \ D e lt a\delta}\right]}.\ en d { al igned}$$ Thisa sy mp t otic isva lid for $ x<1/2 $ an d $ y<\Delta /2$; i t c an be ext en ded to t heoth er th r ee q uarter s of the rect a ngle by obviou s reflections. Aw a y f r om t hedomain corn ersw e ca n re p la ce$ \tanh (\dot s) $ i n the denominator of E q. (\[ hyp\] ) by unity. U sing the r e s u lting “a ppro x im a te product rul e" in Eq. (\[q\ ] ), we ob tain, after s ome algeb r a , the “k ink " a sym pto t i cof $q(x,y)$ a w a y fr om the do mai n corne rs: $$ \la bel {q aprox}
q\simeq\t an h^ 2\ lef t[\fr a c{2(1+\D el ta) x} {\D elta\ d elta}\ right ]\ta nh ^2 \ lef t[\frac { 2( 1 + \Del ta )y }{\D elt a\ delta }\ri g ht] ,$$ whe re $x<1/2 $ a n d $y <\ De lta/2$, and reflecte dformulas i nthe other t hree rec tangle quarters. Thesea symptot ics desc ribe kinks wi ththe ch ara c terist ic wid th $\ el l=\ D e lta\d e l ta /[2 (1 +\Delta)]$ . Byanalo gy wit h one d imension, the acti o n p er unit lengt h a long t he bo u nd a ryis , in t he leading orde r, $ | {K}(\nu_2)$ diverge_as $n_0\to_1$. In this limit_Eq. (\[nu122\]) yields_$\text{K}(\nu_2)\simeq\text{K}(\nu_1)\Delta$._The explicit_$n_0$-dependence_can be obtained_with the help_of Eq. (\[mass2d\]): $2\text{K}(\nu_1) \simeq_(1+\Delta)(\Delta\delta)^{-1}$, and $2\text{K}(\nu_2)_\simeq_(1+\Delta)(\delta)^{-1}$. Using the asymptotic $\text{sn}\,z \simeq \tanh z$, we obtain in the leading order:_$$\begin{aligned}
\label{hyp}
fg&\simeq&\tanh\left[\frac{(1+\Delta)x}{\Delta\delta}\right]\tanh\left[\frac{(1+\Delta)y}{\Delta\delta}\right]_\nonumber\\
&=&\frac{\tanh\left[\frac{(1+\Delta)x}{\Delta\delta}\right]+\tanh\left[\frac{(1+\Delta)y}{\Delta\delta}\right]-1}
{\tanh\left[\frac{(1+\Delta)(x+y)}{\Delta\delta}\right]}.\end{aligned}$$ This_asymptotic_is_valid for $x<1/2$ and $y<\Delta/2$;_it can be extended to_the other_three quarters of the rectangle by obvious reflections._Away_from the domain_corners we can replace $\tanh(\dots)$ in the denominator of_Eq. (\[hyp\]) by unity. Using the resulting_“approximate product rule"_in_Eq. (\[q\]),_we obtain, after some_algebra, the “kink" asymptotic of $q(x,y)$_away from the domain corners: $$\label{qaprox}
_ q\simeq \tanh^2\left[\frac{2(1+\Delta)x}{\Delta\delta}\right]\tanh^2\left[\frac{2(1+\Delta)y}{\Delta\delta}\right],$$ where $x<1/2$ and $y<\Delta/2$,_and reflected formulas in the other_three rectangle quarters. These asymptotics_describe kinks_with the characteristic width $\ell=\Delta\delta/[2(1+\Delta)]$._By analogy with_one dimension,_the action per_unit length along the boundary is,_in the leading_order, $ |
Subsets and Splits