Effective results on uniformization and intrinsic GCM spheres in perturbations of Kerr
Extracted source text ยท arXiv 1912.12195
\documentclass[a4paper,12pt]{article}
\usepackage{amsthm,amsfonts,amssymb,euscript}
\usepackage{latexsym, multicol, fancybox}
\usepackage{graphicx}
\usepackage{color}
\usepackage{amsmath, amsthm, amssymb, bm}
\usepackage{epstopdf}
\usepackage{caption}
\usepackage{psfrag}
% below: definition of commend \widecheck
\DeclareFontFamily{U}{mathx}{\hyphenchar\font45}
\DeclareFontShape{U}{mathx}{m}{n}{
<5> <6> <7> <8> <9> <10>
<10.95> <12> <14.4> <17.28> <20.74> <24.88>
mathx10
}{}
\DeclareSymbolFont{mathx}{U}{mathx}{m}{n}
\DeclareFontSubstitution{U}{mathx}{m}{n}
\DeclareMathAccent{\widecheck}{0}{mathx}{"71}
% widecheck command
\def\trXc{\widecheck{\tr X}}
\def\trXbc{\widecheck{\tr\Xb}}
\def\Hc{\widecheck{H}}
\def\Zc{\widecheck{Z}}
\def\Pc{\widecheck{P}}
\def\ombc{\widecheck{\omb}}
\def\trchcheck{\widecheck{\trch}}
\def\trchbcheck{\widecheck{\trch}}
\def\ombcheck{\widecheck{\omb}}
\def\rhocheck{\widecheck{\rho}}
\def\mucheck{\widecheck{\mu}}
\def\Ombcheck{\widecheck{\Omb}}
\def\vsocheck{\widecheck{\vsi}}
%% Theorem commands
\newtheorem{theorem}{Theorem}[section]
\newtheorem{lemma}[theorem]{Lemma}
\newtheorem{proposition}[theorem]{Proposition}
\newtheorem{corollary}[theorem]{Corollary}
\newtheorem{definition}[theorem]{Definition}
\newtheorem{remark}[theorem]{Remark}
\newtheorem{conjecture}[theorem]{Conjecture}
\newtheorem*{thmmain}{Main Theorem}
\newtheorem*{thmGCM1}{Theorem GCM-S}
\newtheorem*{thmGCM2}{Theorem GCM-T}
%% Environement commands
\setlength{\textwidth}{16cm} \setlength{\oddsidemargin}{0cm}
\setlength{\evensidemargin}{0cm}
\numberwithin{equation}{section}
\parindent = 0 pt
\parskip = 12 pt
\newcommand{\bea}{\begin{eqnarray}}
\newcommand{\eea}{\end{eqnarray}}
\def\beaa{\begin{eqnarray*}}
\def\eeaa{\end{eqnarray*}}
\def\ba{\begin{array}}
\def\ea{\end{array}}
\def\be#1{\begin{equation} \label{#1}}
\def \eeq{\end{equation}}
\def\bsplit{\begin{split}}
\newcommand{\nn}{\nonumber}
%% Miscelaneous commands
\def\lab{\label}
\def\les{\lesssim}
\def\ges{\gtrsim}
\def\c{\cdot}
\def\lot{\mbox{l.o.t.}}
\def\dual{{\,^\star \mkern-2mu}}
\def\tr{\mbox{tr}}
\def\atr{\,^{(a)}\mbox{tr}}
\newcommand{\hch}{\hat{\chi}}
\newcommand{\hchb}{\underline{\hat{\chi}}\,}
\def\ov{\overline}
%% Derivatives
\renewcommand{\div}{\mbox{div}\,}
\newcommand{\curl}{\mbox{curl}\,}
\def\nab{\nabla}
\def\lap{\Delta}
\def\pr{\partial}
\def\dkb{ \, \mathfrak{d} \mkern-9mu /}
\def\dk{\mathfrak{d}}
\def\DDdS{\DDd^{\,\S}}
\def\DDsS{\DDd^{\,\S, \star}}
\def\dddS{\ddd^{\,\S}}
\def\dddSdiez{\ddd^{\,\S, \#}}
\def\ddsSdiez{\big( \ddd^{\,\S, \#}\big)^\star}
\def\muS{\mu^\S}
%% star Commands
\def\dds{ \, d \hspace{-1.5pt}\dual \mkern-14mu /\,\,}
\def\ddd{ \, d \hspace{-2.4pt} \mkern-6mu /\,}
\def\ddsS{{\ddd^{\,\S,*}}}
\def\ddsSone{{\ddd_1^{\,\S,\star}}}
\def\ddsStwo{{\ddd_2^{\,\S,\star}}}
\def\ddsSnone{{\ddd_1^{\,\S(n),\star}}}
\def\ddsSntwo{{\ddd_2^{\,\S(n),\star}}}
\def\curlSn{\curl^{\S(n)}}
\def\divSn{\div^{\S(n)}}
\def\dddSn{\ddd^{\S(n)}}
\def\rhod{ \,^\star \hspace{-2.2pt} \rho}
\def\mud{ \,^\star \hspace{-2.2pt} \mu}
\newcommand{\dec}{\delta_{dec}}
%% overset commands
\def\ovA{\overset{\circ}{ A}}
\def\ovB{\overset{\circ}{ B}}
\def\ovf{\overset{\circ}{ f}}
\def\ovu{\overset{\circ}{ u}}
\def\ovd{\overset{\circ}{ d}}
\def\ovs{\overset{\circ}{ s}}
\def\ovp{\overset{\circ}{ p}}
\def\ove{\overset{\circ}{ e}}
\def\ovr{\overset{\circ}{ r}}
\def\ovla{\protect\overset{\circ}{ \la}\,}
\def\ovb{\protect\overset{\circ}{b}\,}
\def\ovPhi{\protect\overset{\circ}{\Phi}\,}
\def\ovphi{\protect\overset{\circ}{\phi}\,}
\def\ovS{\overset{\circ}{ \S}}
\def\ovJ{\overset{\,\,\circ}{J}}
\def\ddsov{ \, \ovd\dual \hspace{-2.4pt} \mkern-14mu/ }
\def\dddov{ \, \ovd \hspace{-2.4pt} \mkern-6mu /}
\def\kabar{\ov{\ka}}
\def\kabbar{\ov{\kab}}
\def\ombar{\ov{\om}}
\def\rhobar{\ov{\rho}}
\def\epg{\protect\overset{\circ}{\ep}}
\def\ug{\overset{\circ}{u}}
\def\sg{\overset{\circ}{s}}
\def\rg{\overset{\circ}{r}}
\def\cg{\overset{\circ}{c}}
\def\mg{\overset{\circ}{m}}
\def\ovf{\overset{\circ}{ f}}
\def\ovu{\overset{\circ}{ u}}
\def\ovd{\overset{\circ}{ d}}
\def\ovs{\overset{\circ}{ s}}
\def\ovp{\overset{\circ}{ p}}
\def\ove{\overset{\circ}{ e}}
\def\ovr{{\overset{\circ}{ r}\,}}
\def\ovm{{\overset{\circ}{ m}\,}}
\def\ovS{\overset{\circ}{ S}}
\def\ovga{\overset{\circ}{\ga}\,}
\def\ovgS{\overset{\circ}{\gS}\,}
\def\ovg{\overset{\circ}{ g}}
\def\ovGa{\overset{\circ}{ \Ga}}
%% Command for GCM construction
\def\gazero{\ga^{(0)}}
\def\gatwoone{\ga^{(2,1)}}
\def\gaonetwo{\ga^{(1,2)}}
\def\dddone{\ddd^{(1)}}
\def\dddtwo{\ddd^{(2)}}
\def\ddsone{\dds^{(1)}}
\def\ddstwo{\dds^{(2)}}
\def\ddszero{\dds^{(0)}}
\def\dddzero{\ddd^{(0)}}
\def\muc{\widecheck{\mu}}
\def\Unn{U^{(n+1)}}
\def\Un{U^{(n)}}
\def\hnn{h^{(n+1)}}
\def\hbnn{\underline{h}^{(n+1)}}
\def\hn{h^{(n)}}
\def\hbn{\underline{h}^{(n)}}
\def\wn{w^{(n)}}
\def\wnn{w^{(n+1)}}
\def\ddsn{\dds^{(n)}}
\def\dddn{\ddd^{(n)}}
\def\en{e^{(n)}}
\def\enn{e^{(n+1)}}
\def\lapSn{\lap^{\S(n)}}
\def\hb{\underline{h}}
%% ^S commands
\def\dkbS{{\dkb^\S}}
\def\vthS{\vth^\S}
\def\vthbS{\vthb^\S}
\def\etaS{\eta^\S}
\def\etabS{\etab^\S}
\def\kaS{\ka^\S}
\def\kabS{\kab^\S}
\def\omS{\om^\S}
\def\ombS{\omb^\S}
\def\zeS{\ze^\S}
\def\aS{\a^\S}
\def\aaS{\aa^\S}
\def\bS{\b^\S}
\def\bbS{\bb^\S}
\def\rhoS{\rho^\S}
\def\gaS{{\ga^\S}\,}
\def\thS{{\th^\S}}
\def\vphiS{{\vphi^\S}}
\def\trchS{{\trch^\S}}
\def\atrchS{{\atrch^\S}}
\def\trchbS{{\trchb^\S}}
\def\atrchbS{{\atrch^\S}}
%% Greek Letters
\def\a{\alpha}
\def\b{\beta}
\def\ga{\gamma}
\def\Ga{\Gamma}
\def\de{\delta}
\def\De{\Delta}
\def\ep{\epsilon}
\def\la{\lambda}
\def\La{\Lambda}
\def\si{\sigma}
\def\Si{\Sigma}
\def\om{\omega}
\def\Om{\Omega}
\def\ro{\rho}
\def\rhod{\dual \rho}
\def\th{{\theta}}
\def\Th{{\Theta}}
\def\ka{\kappa}
\def\ze{\zeta}
\def\Up{\Upsilon}
\def\ks{\kappa}
\def\Th{\Theta}
\def\Thb{\underline{\Theta}}
\def\atrch{\atr\chi}
\def\atrchb{\atr\chib}
\def\ak{\atrch}
\def\akb{\atrchb}
%% Greek letters var
\def\vep{\varepsilon}
\def\vphi{{\varphi}}
\def\vth{\vartheta}
\def\varoc{\check{\varrho}}
\def\vsi{\varsigma}
%% Underline commands
\def\thb{\underline{\th}}
\def\vthb{\underline{\vth}}
\renewcommand{\aa}{\protect\underline{\a}}
\newcommand{\bb}{\protect\underline{\b}}
\def\omb{\protect\underline{\om}}
\def\Lb{{\underline{L}}}
\def\LLb{{\underline{\L}}}
\def\lb{{\underline l}}
\def\llb{{\underline\l}}
\def\unda{{\underline{a}}}
\def\Omb{\underline{\Omega}}
\def\ub{{\underline{u}} }
\newcommand{\chib}{\protect\underline{\chi}}
\newcommand{\xib}{\protect\underline{\xi}}
\newcommand{\etab}{\protect\underline{\eta}}
\def\ksb{\underline{\kappa}}
\def\kab{\protect\underline{\kappa}}
\def\Nb{\underline{N}}
%% mathcal letters come in double caps
\def\AA{{\mathcal A}}
\def\BB{{\mathcal B}}
\def\CC{{\mathcal C}}
\def\DD{{\mathcal D}}
\def\EE{{\mathcal E}}
\def\FF{{\mathcal F}}
\def\GG{{\mathcal G}}
\def\HH{{\mathcal H}}
\def\II{{\mathcal I}}
\def\JJ{{\mathcal J}}
\def\KK{{\mathcal K}}
\def\Lie{{\mathcal L}}
\def\LL{{\mathcal L}}
\def\MM{{\mathcal M}}
\def\NN{{\mathcal N}}
\def\OO{{\mathcal O}}
\def\PP{{\mathcal P}}
\def\QQ{{\mathcal Q}}
\def\RR{{\mathcal R}}
\def\SS{{\mathcal S}}
\def\TT{{\mathcal T}}
\def\UU{{\mathcal U}}
\def\YY{{\mathcal Y}}
\def\XX{{\mathcal X}}
\def\ZZ{{\mathcal Z}}
%% Bold-face letters come in capitals
\def\A{{\bf A}}
\def\B{{\bf B}}
\def\C{{\bf C}}
\def\D{{\bf D}}
\def\E{{\bf E}}
\def\F{{\bf F}}
\def\G{{\bf G}}
\def\H{{\bf H}}
\def\I{{\bf I}}
\def\J{{\bf J}}
\def\K{{\bf K}}
\def\L{{\bf L}}
\def\M{{\bf M}}
\def\N{{\bf N}}
\def\O{{\bf O}}
\def\Q{{\bf Q}}
\def\R{{\bf R}}
\def\P{{\bf P}}
\def\S{{\bf S}}
\def\T{{\bf T}}
\def\U{{\bf U}}
\def\X{{\bf X}}
\def\Y{{\bf Y}}
\def\Z{{\bf Z}}
\def\g{{\bf g}}
\def\h{{\bf h}}
\def\m{{\bf m}}
\def\p{{\bf p}}
\def\q{{\bf q}}
\def\t{{\bf t}}
\def\u{{\bf u}}
\def\l{{\bf l}}
%% Underline letters (have a b)
\def\Ab{\underline{A}}
\def\Bb{\underline{B}}
\def\Eb{\underline{E}}
\def\Fb{\underline F}
\def\Hb{\underline{H}}
\def\Rb{\underline{R}}
\def\Vb{\underline{V}\,}
\def\lb{{\underline{l}}}
\def\mub{{\underline{\mu}}}
\def\fb{\protect\underline{f}}
\def\CCb{\underline{\CC}}
\def\Cb{{\underline{C}}}
%% Overline notations
\def\Dov{\overline{\D}}
\def\squareov{\overline{\square}}
%% Blackboard bold letters come in triples
\def\AAA{{\Bbb A}}
\def\CCC{{\Bbb C}}
\def\MMM{{\Bbb M}}
\def\NNN{{\Bbb N}}
\def\QQQ{{\Bbb Q}}
\def\RRR{{\Bbb R}}
\def\SSS{{\Bbb S}}
\def\HHH{{\Bbb H}}
\def\ZZZ{{\Bbb Z}}
%% Mathfrak commands
\def\hk{\mathfrak{h}}
\def\af{\mathfrak{a}}
\def\sk{\mathfrak{s}}
%% check notations
\def\Rc{\check R}
\def\Tc{\check T}
\def\rhoc{\check \rho}
\def\Gac{\check \Gamma}
\def\chic{\check \chi}
\def\chibc{\check{\underline{\chi}}}
\def\kac{\widecheck \ka}
\def\kabc{\widecheck{\underline{\ka}}}
\def\Omc{\check{\Omega}}
\def\Oc{\check{\Omega}}
\def\Ombc{\check{\underline{\Omega}}}
\def\Ombc{\check{\underline{\Omega}}}
\def\omc{\check \omega}
\def\ombc{\underline{\check \omega}}
\def\vsic{\check{\vsi}}
%commands for spacetime quanitities
\def\chih{\widehat{\chi}}
\def\chibh{\widehat{\chib}}
\def\trch{\tr \chi}
\def\trchb{\tr \chib}
\def\hot{\widehat{\otimes}}
\def\c{\cdot}
\def \f12{\frac 1 2 }
\def\ov{\overline}
\def\err{\mbox{Err}}
\def\Lap{\,^{(p)}\La}
\def\gS{ g \mkern-8.5mu/\,}
\def\err{\mbox{Err}}
\def\errb{\underline{Err}}
\def\Lab{\underline{\La}}
\def\aka{\,^{(a)} \ka}
\def\akab{\,^{(a)} \kab}
\def\epg{\protect\overset{\circ}{\ep}}
\def\dg{\overset{\circ}{\de}}
\def\undB{\underline{B}}
\def\XXone{\XX_{(1)}}
\def\XXtwo{\XX_{(2)}}
\def\XXa{\XX_{(a)}}
\def\XXb{\XX_{(b)}}
\def\XXonec{\overset{\circ}{ \XX}_{(1)}}
\def\XXtwoc{\overset{\circ}{ \XX}_{(2)}}
\def\ovJq{\overset{\circ}{J}\mkern-0.5mu ^{(q)}}
\def\YYone{{\YY_{(1)}}}
\def\YYtwo{{\YY_{(2)}}}
\def\YYa{\YY_{(a)}}
\def\YYb{\YY_{(b)}}
\newcommand{\Red}{\textcolor{red}}
\newcommand{\Blue}{\textcolor{blue}}
\def\Psin{\Psi^{(n)} }
\def\Un{U^{(n)}}
\def\Sn{{S^{(n)}}}
\def\fn{f^{(n)}}
\def\fbn{\fb^{(n)}}
\def\fnn{f^{(n+1)}}
\def\fbnn{\fb^{(n+1)}}
\def\ovlan{\ovla^{(n)}}
\def\ovlann{\ovla^{(n+1)}}
\def\Fn{F^{(n)}}
\def\Fnn{F^{(n+1)}}
\def\en{e^{(n)}}
\def\enn{e^{(n+1)}}
\def\enS{{e^{\S(n)}}}
\def\curln{\curl^{(n)}}
\def\nabn{\nab^{(n)}}
\def\divn{\div^{(n)}}
\def\lapn{\lap^{(n)}}
\def\Sn{ S^{(n)}}
\def\SSn{\S^{(n)}}
\def\Snn{ S^{(n+1)}}
\def\YYn{\YY^{(n)}}
\def\gndiez{g^{\S, \#_n}}
\def\gn{g^{\S(n)}}
\def\DDn{\DD^{(n)}}
\def\Hnn{H^{(n+1)}}
\def\Gnn{G^{(n+1)}}
\def\Snnt{ \widetilde{S}^{(n+1)}}
\def\Unnt{ \widetilde{U}^{(n+1)}}
\def\Hinfty{H^{(\infty)}}
\def\Ginfty{G^{(\infty)}}
\def\Sinfty{ S^{(\infty)}}
\def\Uinfty{ U^{(\infty)}}
\def\finfty{f^{(\infty)}}
\def\fbinfty{\fb^{(\infty)}}
\def\ovlainfty{\ovla^\infty}
\def\DDinfty{\DD^{(\infty)}}
\def\Finfty{F^{(\infty)}}
\def\divzero{{\overset{\circ}{ \div}}}
\def\curlzero{{\overset{\circ}{ \curl}}}
\def\lapzero{{\overset{\circ}{ \lap}}}
\def\nabzero{{\overset{\circ}{ \nab}}}
\def\curlinfty{\curl^{(\infty)}}
\def\divinfty{\div^{(\infty)}}
\def\nabinfty{\nab^{(\infty)}}
\def\DDnn{\DD^{(n+1)}}
\def\jp{j^{(p)}}
\def\Jp{J^{(p)}}
\def\JpS{{J^{(p, \S)}}}
\def\JJpS{\J^{(p,\S)} }
\def\Cbp{\Cb^{(p)}}
\def\Mp{M^{(p)}}
\def\CbpS{\Cb^{(\S, p)}}
\def\MpS{M^{(\S, p)}}
\def\Uinfty{U^{(\infty)}}
\def\Sinfty{S^{(\infty)}}
\def\finfty{f^{(\infty)}}
\def\fbinfty{\fb^{(\infty)}}
\def\ovlainfty{\ovla^{(\infty)}}
\def\einfty{e^{(\infty)}}
\def\ddsSinftyone{{\ddd_1^{\,\S^{(\infty)},*}}}
\def\ddsSinftytwo{{\ddd_2^{\,\S^{(\infty)},*}}}
\def\curlSinfty{\curl^{\S^{(\infty)}}}
\def\divSinfty{\div^{\S^{(\infty)}}}
\def\kainfty{\ka^{(\infty)}}
\def\kabinfty{\kab^{(\infty)}}
\def\muinfty{\mu^{(\infty)}}
\def\lapSinfty{\lap^{\S(\infty)}}
\def\Cinfty{C^{(\infty)}}
\def\Minfty{M^{(\infty)}}
\def\Cbinfty{\Cb^{(\infty)}}
\def\Cbpinfty{\Cb^{(\infty),p}}
\def\Mpinfty{M^{(\infty),p}}
\def\Jpinfty{J^{(\infty),p}}
\def\dddSinfty{\ddd^{\S^{(\infty)}}}
\def\kadot{\dot{\ka}}
\def\kabdot{\dot{\kab}}
\def\mudot{\dot{\mu}}
\def\Cbdot{\dot{\Cb}}
\def\Mdot{\dot{M}}
\def\Cbpdot{{\dot{\Cb}^{(p)}}}
\def\Mpdot{{\dot{M}^{(p)}}}
\def\kann{\ka^{(n+1)}}
\def\kabnn{\kab^{(n+1)}}
\def\munn{\mu^{(n+1)}}
\def\kanndot{(\de\ka)^{(n+1)}}
\def\kabnndot{(\de\kab)^{(n+1)}}
\def\munndot{(\de\mu)^{(n+1)}}
\def\Cbnn{\Cb^{(n+1)}}
\def\Cbpnn{\Cb^{(n+1), p}}
\def\Mpnn{M^{(n+1), p}}
\def\Mnn{M^{(n+1)} }
\def\Cbnndot{\Cbdot^{(n+1)}}
\def\Cbpnndot{\Cbdot^{(n+1), p}}
\def\Mpnndot{\Mdot^{(n+1), p}}
\def\Mnndot{\Mdot^{(n+1)} }
\def\kan{\ka^{(n)}}
\def\kabn{\kab^{(n)}}
\def\mun{\mu^{(n)}}
\def\Cbn{\Cb^{(n)}}
\def\Cbpn{\Cb^{(n), p}}
\def\Mpn{M^{(n), p}}
\def\Mn{M^{(n)} }
\def\Cbndot{\Cbdot^{(n)}}
\def\Cbpndot{\Cbdot^{(n), p}}
\def\Mpndot{\Mdot^{(n), p}}
\def\Mndot{\Mdot^{(n)} }
\def\fnndot{(\de f)^{(n+1)}}
\def\fbnndot{ (\de\fb)^{(n+1)}}
\def\ovlanndot{ (\de\ovla)^{(n+1)}}
\def\fndot{(\de f)^{(n)}}
\def\fbndot{ (\de\fb)^{(n)}}
\def\ovlandot{ (\de\ovla)^{(n)}}
\def\Cbnnde{\, \de\Cbnn}
\def\Cbpnnde{\, \de\Cbpnn}
\def\Mnnde{\, \de \mkern-3mu\Mnn}
\def\Mpnnde{\, \de \mkern-3mu\Mpnn}
\def\ovlanndiez{\ovla^{n+1,\#}}
\def\fnndiez{f^{n+1,\#}}
\def\fbnndiez{\fb^{n+1,\#}}
\def\ovlandiez{\ovla^{n,\#}}
\def\fndiez{f^{n,\#}}
\def\fbndiez{\fb^{n,\#}}
\def\ovlanndiez{\ovla^{n+1,\#}}
\def\fnndiez{f^{n+1,\#}}
\def\fbnndiez{\fb^{n+1,\#}}
\def\kandiez{\ka^{n,\#}}
\def\kabndiez{\kab^{n,\#}}
\def\mundiez{\mu^{n,\#}}
\def\kanndiez{\ka^{n+1,\#}}
\def\kabnndiez{\kab^{n+1,\#}}
\def\munndiez{\mu^{n+1,\#}}
\def\curln{\curl^{(n)}}
\def\divn{\div^{(n)}}
\def\lapn{\lap^{(n)}}
\def\curlnmin{\curl^{(n-1)}}
\def\divnmin{\div^{(n-1)}}
\def\lapnmin{\lap^{(n-1)}}
\def\gn{g^{(n)}}
\def\Fndiez{ F^{(n), \#}}
\def\Fnndiez{ F^{(n+1), \#}}
\def\JpSn{J^{\S(n), p}}
\def\curlnmin{\curl^{(n-1)}}
\def\divnmin{\div^{(n-1)}}
\def\lapnmin{\lap^{(n-1)}}
\def\Psih{\widehat{\Psi}}
\def\kabcS{ \widecheck{\kab^\S}}
\def\ovM{\protect\overset{\circ}{ \M}}
\def\trchbc{\widecheck{\trchb} }
\newdimen\cdsep
\cdsep=3em
\def\cdstrut{\vrule height .6\cdsep width 0pt depth .4\cdsep}
\def\@cdstrut{{\advance\cdsep by 2em\cdstrut}}
\def\arrow#1#2{
\ifx d#1
\llap{$\scriptstyle#2$}\left\downarrow\cdstrut\right.\@cdstrut\fi
\ifx u#1
\llap{$\scriptstyle#2$}\left\uparrow\cdstrut\right.\@cdstrut\fi
\ifx r#1
\mathop{\hbox to \cdsep{\rightarrowfill}}\limits^{#2}\fi
\ifx l#1
\mathop{\hbox to \cdsep{\leftarrowfill}}\limits^{#2}\fi
}
\def\JpSS{\J^{(p, \S)}}
\def\Jpov{\ovJ\,^{(p)}}
\def\Psih{\widehat{\Psi}}
\def\Def{D \hspace{-1.5pt} e \hspace{-1.8pt} f}
\begin{document}
\title{Effective results on uniformization and intrinsic GCM spheres in perturbations of Kerr}
\author{Sergiu Klainerman, J\'er\'emie Szeftel}
\maketitle
{\bf Abstract.} \textit{This is a follow-up of our paper \cite{KS-Kerr1} on the construction of general covariant modulated (GCM) spheres in perturbations of Kerr, which we expect to play a central role in establishing their nonlinear stability. We reformulate the main results of that paper using a canonical definition of $\ell=1$ modes on a $2$-sphere embedded in a $1+3$ vacuum manifold. This is based on a new, effective, version of the classical uniformization theorem which allows us to define such modes and prove their stability for spheres with comparable metrics. The reformulation allows us to prove a second, intrinsic, existence theorem for GCM spheres, expressed purely in terms of geometric quantities defined on it. A natural definition of angular momentum for such GCM spheres is also introduced, which we expect to play a key role in determining the final angular momentum for general perturbations of Kerr.}
\tableofcontents
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Introduction}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
This is a follow-up of our paper \cite{KS-Kerr1} on the construction of general covariant modulated (GCM) spheres in perturbations of Kerr.
We reformulate the main results of that paper using a canonical definition of $\ell=1$ modes on a $2$-sphere embedded in a $1+3$ vacuum manifold. This is based on a new, effective, version of the classical uniformization theorem which allows us to define such modes and prove their stability for spheres with comparable metrics. The reformulation allows us
to prove a second, intrinsic, existence theorem for GCM spheres, expressed purely in terms of geometric quantities defined on it.
A natural definition of angular momentum for such GCM spheres is also introduced, which we expect to play a key role in the definition of angular momentum for general perturbations of Kerr.
We refer to the introduction in \cite{KS-Kerr1} for a short description of the role of GCM spheres in the proof of the nonlinear stability of Schwarzschild in \cite{KS-Schw} and the crucial role we expect them to play in extending that result to Kerr.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Review of the main results of \cite{KS-Kerr1}}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Background space}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
As in \cite{KS-Kerr1} we consider spacetime regions $\RR$ foliated by a geodesic foliation $S(u, s)$ induced by an outgoing optical function\footnote{As explained in Remark 1.5 of \cite{KS-Kerr1}, this is not strictly necessary. Any other foliation satisfying comparable asymptotic assumptions would also work.} $u$ with $s$ a properly normalized affine parameter along the null geodesic generators of $L=-\g^{\a\b}\pr_\b u\pr_\a $. We denote by $r=r(u, s)$ the area radius of $S(u,s)$ and let $(e_3, e_4, e_1, e_2)$ be an adapted null frame with $e_4$ proportional to $L$ and $e_1, e_2 $ tangent to spheres $S=S(u, s)$, see section \ref{sec:backgroundspacetime}.
The main assumptions made in \cite{KS-Kerr1} were that the Ricci and curvature coefficients, relative to the adapted null frame, have the same asymptotics in powers of $r$ as in Schwarzschild space. Note that these assumptions hold true in the far region of Kerr and is expected to hold true for the far region of realistic perturbations of Kerr. The actual size of the perturbation from Kerr is measured with respect to a small parameter $\epg>0$, see sections \ref{subsubsect:regionRR1} and \ref{subsubsect:regionRR2} for precise definitions.
%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Definition of GCM spheres}
%%%%%%%%%%%%%%%%%%%%%%%
These are topological spheres $\S$ embedded in $\RR$ endowed with a null frame $e_3^\S, e_4^\S, e_1^\S, e_2^\S$ adapted to $\S$ (i.e. $e_1^\S, e_2^\S$ tangent to $\S$), relative
to which the null expansions $\ka^\S=\trch^\S$, $\kab^\S=\trchb^\S$ and mass aspect function
$\mu^\S$ take Schwarzschildian values, i.e.
\bea
\lab{Introd:GCMspheres1}
\ka^\S -\frac{2}{r^\S} =0, \qquad \kab^\S+\frac{2\Up^\S}{r^\S} =0, \qquad \mu^\S - \frac{2m^\S}{(r^\S)^3} =0,
\eea
where $r^\S, m^\S$ denote the area radius and Hawking mass of $\S$ and $\Up^\S=1-\frac{2m^\S}{r^\S}$, see section \ref{sec:backgroundspacetime} for precise definitions.
As explained in the introduction to \cite{KS-Kerr1}, these conditions have to be relaxed with respect to the $\ell=0, 1$ modes on $\S$. To make sense of this statement requires us to fix a basis of $\ell=1$ modes on $\S$, which generalize the $\ell=1$ spherical harmonics of
the standard sphere\footnote{Recall that on the standard sphere $\SSS^2$, in spherical coordinates $(\th, \vphi)$, these are
$J^{(0, \SSS^2)}=\cos\th$, $J^{(+,\SSS^2)}=\sin\th\cos\vphi$, $J^{(-,\SSS^2)}=\sin\th\sin\vphi$.}.
Assuming the existence of such a basis $\JpS, p\in\big\{ -, 0, +\big\}$, we define, for a scalar function $h$,
\bea
( h)^\S_{\ell=1} &:=&\left\{\int_{\S} h \JpS, \quad p=-, 0, +\right\}.
\eea
A scalar function $h$ is said to be supported on $\ell\le 1$ modes, i.e. $(f)^\S_{\ell\ge 2}=0$, if there exist constants
$A_0, B_{-}, B_0, B_{+} $ such that
\bea
h= A_0+ B_{-} J^{( -, \S) }+B_0 J^{( 0, \S)} + B_{+} J^{(+, \S)}.
\eea
With this definition, \eqref{Introd:GCMspheres1} is replaced by
\bea
\lab{Introd:GCMspheres2}
\ka^\S -\frac{2}{r^\S} =0, \qquad \left( \kab^\S+\frac{2\Up^\S}{r^\S}\right)_{\ell\ge 2 }=0, \qquad \left( \mu^\S - \frac{2m^\S}{(r^\S)^3} \right)_{\ell\ge 2}=0.
\eea
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Deformations of spheres and frame transformations}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
The construction of GCM spheres
in \cite{KS-Kerr1} was obtained by deforming a given sphere $ \ovS=S(\ovu, \ovs)$ of the background foliation of $\RR$. An $O(\dg)$ deformation of $\ovS$ is defined by a
map $\Psi:\ovS\to \S $ of the form
\bea\lab{eq:definitionofthedeformationmapPsiinintroduction}
\Psi(\ovu, \ovs, y^1, y^2)=\left( \ovu+ U(y^1, y^2 ), \, \ovs+S(y^1, y^2 ), y^1, y^2 \right)
\eea
with $(U, S)$ smooth functions on $\ovS$, vanishing at a fixed point of $\ovS$, of size proportional to the small constant $\dg$.
Given such a deformation we identify, at any point on $\S$, two important null frames.
\begin{enumerate}
\item The null frame $( e_3, e_4, e_1, e_2)$ of the background foliation of $\RR$.
\item A null frame $( e^\S_3, e^\S_4, e^\S_1, e^\S_2)$ adapted to the sphere $\S$.
\end{enumerate}
In general, two null frames $(e_3, e_4, e_1, e_2)$ and $(e_3', e_4', e_1', e_2') $ are related by a frame transformation of the form, see Lemma \ref{Lemma:Generalframetransf},
\bea
\lab{eq:Generalframetransf-intro}
\bsplit
e_4'&=\la\left(e_4 + f^b e_b +\frac 1 4 |f|^2 e_3\right),\\
e_a'&= \left(\de_{ab} +\frac{1}{2}\fb_af_b\right) e_b +\frac 1 2 \fb_a e_4 +\left(\frac 1 2 f_a +\frac{1}{8}|f|^2\fb_a\right) e_3,\\
e_3'&=\la^{-1}\left( \left(1+\frac{1}{2}f\c\fb +\frac{1}{16} |f|^2 |\fb|^2\right) e_3 + \left(\fb^b+\frac 1 4 |\fb|^2f^b\right) e_b + \frac 1 4 |\fb|^2 e_4 \right),
\end{split}
\eea
where the scalar $\la$ and the 1-forms $f$ and $\fb$ are called the transition coefficients of the transformation. As explained in the introduction to \cite{KS-Kerr1}, one can relate all Ricci and curvature coefficients of the primed frame in terms of the Ricci and curvature coefficients of the un-primed one. In particular, the GCM conditions \eqref{Introd:GCMspheres2} can be expressed in terms of differential conditions for the transition coefficients $(f, \fb, \la) $. The condition that the horizontal part of the frame $(e^\S_1, e_2^\S)$ is tangent to $\S$ also leads to a relation between the gradients of the scalar functions $U, S$, introduced in \eqref{eq:definitionofthedeformationmapPsiinintroduction}, and $(f, \fb) $. Roughly, we thus expect to derive a coupled system\footnote{See the introduction in \cite{KS-Kerr1} for further explanations.} between Laplace equations for $U, S$, on $\ovS$ and an elliptic Hodge system for $F=(f, \fb, \la-1)$ on $\S$.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{GCM spheres with non canonical $\ell=1$ modes in \cite{KS-Kerr1}}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Here is a short version of our main result in \cite{KS-Kerr1}.
\begin{theorem}[Existence of GCM spheres in \cite{KS-Kerr1}]
\lab{Theorem:ExistenceGCMS1-intro}
Let $\RR$ be fixed spacetime region, endowed with an outgoing geodesic foliation $S(u, s)$, verifying
specific asymptotic assumptions\footnote{Compatible with small perturbations of Kerr.} expressed in terms of two parameters $0<\dg\leq \epg$. In particular we assume that
the GCM quantities of the background spheres in $\RR$, i.e.
\bea
\lab{Introd:GCMspheres3}
\ka-\frac 2 r , \qquad \left( \kab+\frac{2\Up}{r}\right)_{\ell\ge 2 }, \qquad \left( \mu- \frac{2m}{r^3} \right)_{\ell\ge 2},
\eea
are small with respect to the parameter $\dg$.
Let $\ovS=S(\ovu, \ovs)$ be a fixed sphere of the foliation with $\rg$ and $\mg$ denoting respectively its area radius and Hawking mass, with $\rg$ sufficiently large.
Then,
for any fixed triplets $\La, \Lab \in \RRR^3$ verifying
\bea
|\La|,\, |\Lab| &\les & \dg,
\eea
there
exists a unique GCM sphere $\S=\S^{(\La, \Lab)}$, which is a deformation of $\ovS$,
such that
\bea
\lab{Introd:GCM spheres2-again}
\ka^\S -\frac{2}{r^\S} =0, \qquad \left( \kab^\S+\frac{2\Up^\S}{r^\S}\right)_{\ell\ge 2 }=0, \qquad \left( \mu^\S - \frac{2m^\S}{(r^\S)^3} \right)_{\ell\ge 2}=0,
\eea
and
\bea
\lab{Introd:GCM spheres-LaLab}
(\div^\S f)_{\ell=1}=\La, \qquad (\div^\S\fb)_{\ell=1}=\Lab,
\eea
where $(f, \fb, \la)$ denote the transition coefficients of the transformation \eqref{eq:Generalframetransf-intro} from the background frame of $\RR$ to the frame adapted to $\S$.
\end{theorem}
\begin{remark}
\eqref{Introd:GCMspheres3}, \eqref{Introd:GCM spheres2-again} and \eqref{Introd:GCM spheres-LaLab} depend on the definition of $\ell=1$ modes
on $\ovS$ and $\S$. In \cite{KS-Kerr1} the choice $J^{(p, \S)}$ of $\ell=1$ modes for $\S$ was tied to the modes
$\Jp$ of $\ovS$ via the formula,
\bea
\JpS=\Jp\circ\Psi^{-1},
\eea
with $\Psi$ given by \eqref{eq:definitionofthedeformationmapPsiinintroduction}. The $\ell=1$ modes $\Jp$ on $\ovS$ were chosen such that they verify certain natural conditions, see \eqref{eq:Jpsphericalharmonics}, but where not uniquely defined and thus not canonical.
\end{remark}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Main new results on uniformization of spheres}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
One of the main goals of this paper is to remove the arbitrariness in the definition of the $\ell=1$ modes on the deformed GCM spheres $\S$. We do that by appealing to two new results concerning uniformization of nearly round spheres.
The first result, based on works\footnote{ We thank A. Chang for bringing this works to our attention. The precise formulation given in Theorem \ref {theorem-Eff.uniformalization-appendix} does not seem to appear in the literature we are aware of. } by Onofri \cite{Onofri} and Chang-Yang \cite{CY1}, \cite{CY2}, given in Theorem \ref {theorem-Eff.uniformalization-appendix}, provides an effective version of the classical uniformization theorem. The second result, given in Theorem \ref{Thm:unifor-twometrics}, based on Theorem \ref {theorem-Eff.uniformalization-appendix} and previous work\footnote{We thank C. De Lellis for bringing this paper to our attention.} of Frisecke James and M\"uller \cite{FJM}, allows us to formulate and prove a stability result for effective uniformization
of nearby spheres. Together with a notion of calibration, which we introduce in section \ref{subsection:calibration}, this result allows us to define a canonical definition of $\ell=1$ modes for deformed spheres $\S$.
%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Classical uniformization}
%%%%%%%%%%%%%%%%%%%%%%%%
According to the classical uniformization theorem, if $S$ is a closed, oriented and connected surface of genus 0 and $g^S$ is a Riemannian metric on $S$, then, there exists a smooth diffeomorphism $\Phi:\SSS^2\to S$ and a smooth conformal factor
$u$ on $\SSS^2$ such that
\bea
\lab{standard-unifS-intro}
\Phi^\#( g^S) = (r^S)^2 e^{2u} \ga_0
\eea
where $\ga_0$ is the canonical metric on the standard sphere $\SSS^2$ and $r^S$ the area radius of $S$. Unfortunately the conformal factor $u$ is not unique and it is thus difficult to control its size and, more importantly for our applications to GCM spheres, it does not allow one to compare the the conformal factors for two nearby spheres. We state below two results which overcome these difficulties. Both results are applicable to almost round spheres, i.e. closed $2$-surfaces $S$ with Gauss curvature $K^S$ sufficiently close to
that of a round sphere. More precisely, we have for some sufficiently small $\ep>0$
\bea
\lab{eqAlmostroound-Intro}
\left|K^S-\frac{1}{( r^S)^2} \right|\le \frac{\ep}{( r^S)^2}.
\eea
In what follows we state, in a simplified version,
our main new results on effective uniformization.
%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Effective uniformization}
%%%%%%%%%%%%%%%%%%%%%%%%
\begin{theorem}[Effective uniformization 1]
\lab{Thm:effectiveU1-Intro}
Given an almost round sphere $(S, g^S)$ as above there exists, up to isometries\footnote{i.e. all the solutions are of the form $(\Phi\circ O, u\circ O)$ for $O\in O(3)$.} of $\SSS^2$, a unique diffeomorphism $\Phi:\SSS^2\to S$ and a unique conformal factor $u$ such that
\bea
\bsplit
\Phi^\#( g^S)& = (r^S)^2e^{2u} \ga_{\SSS^2},\\
\int_{\SSS^2} e^{2u} x^i &=0, \qquad i=1,2,3.
\end{split}
\eea
Moreover the size of the conformal factor $u$
is small with respect to the parameter $\ep$, i.e.
\beaa
\| u\|_{L^\infty (\SSS^2) } \les \ep.
\eeaa
\end{theorem}
The result is restated in Theorem \ref{theorem-Eff.uniformalization-appendix}, in the case of radius 1 spheres, and in general in Corollary \ref{proposition:effective-uniformisation}.
We use this effective uniformization to define a basis of $\ell=1$ modes on $S$ as follows
\bea
\lab{intro-canmodes-intro}
J^{(p,S)}&=&J^{(p, \SSS^2)}\circ \Phi^{-1}
\eea
where $J^{(p, \SSS^2)}$ are the standard $\ell=1$ spherical harmonics of the round sphere $\SSS^2$.
We note however that this definition is still ambiguous with respect to arbitrary rotations of $\SSS^2$. We can remove this arbitrariness for deformations $\S$ of $\ovS$ by calibrating the effective uniformization of $\S$ with that of $\ovS$, see the discussion in section \ref{subsection:calibration}. This requires however our stable uniformization result which we discuss below.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Stability of the effective uniformization}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{theorem}[Effective uniformization 2]
\lab{Thm:effectiveU2-Intro}
Consider two almost round spheres $(S_1, g^{S_1} )$ and $(S_2, g^{S_2} )$ and their respective
canonical uniformizations $(\Phi_1, u_1)$, $(\Phi_2, u_2)$ provided by the effective uniformization result mentioned above, and a smooth map $\Psi:S_1\to S_2$.
Assume also that the metrics $g^{S_1} $ and $\Psi^\#( g^{S_2} )$
are close to each other in $S^1$,
\bea
\lab{eq:intro.unifor-twometricsB}
\big( r^{S_1}\big)^{-2} \| g^{S_1}- \Psi^\#( g^{S_2} )\|_{L^\infty(S^1)} \le \de
\eea
relative to a small parameter $\de>0$.
Then, there exists $O\in O(3)$ such that the map $\Psih:=(\Phi_2)^{-1}\circ \Psi\circ \Phi_1$ verifies,
\bea
\lab{eq:intro.Psih-Id}
\| \Psih-O\|_{L^\infty(\SSS^2)}&\les &\de.
\eea
Moreover the conformal factors $u_1, u_2$ verify
\bea
\big \|u_1- \Psih^\# u_2\big \|_{L^\infty(\SSS^2)} \les \de.
\eea
\end{theorem}
The result is restated in Theorem \ref{Thm:unifor-twometrics}. Its proof is based on a result due to G. Friesecke, R. James and S. M\"uller \cite{FJM} which improves on a classical result of F. John \cite{John}.
To avoid the indeterminacy with respect to $O\in O(3)$, we introduce a notion of calibration between $(\Phi_1, v_1) $ and $(\Phi_2, v_2)$ with respect to the map $\Psi$, see section \ref{subsection:calibration}. Under this notion of calibration we can replace \eqref{eq:intro.Psih-Id} by,
\bea
\| \Psih-I\|_{L^\infty(\SSS^2)}&\les &\de.
\eea
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Stability of the canonical $\ell=1$ modes}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Consider two almost round spheres $(S_1, g^{S_1} )$ and $(S_2, g^{S_2} )$ and their respect
canonical uniformizations $(\Phi_1, u_1)$, $(\Phi_2, u_2)$ provided by the effective uniformization result mentioned above, and a smooth map $\Psi:S_1\to S_2$.
Assume also that the metrics $g^{S_1} $ and $\Psi^\#( g^{S_2} )$
are close to each other in $S^1$, i.e.,
\bea
\lab{eq:intro.unifor-twometricsB:bis}
\| g^{S_1}- \Psi^\#( g^{S_2} )\|_{H^{2}(S^1)} \le \de.
\eea
Assume also that the uniformization maps of $S_1, S_2$ are calibrated, as in section \ref{subsection:calibration}
and let
\beaa
J^i=J^{S_i}= J\circ \Phi_i^{-1}, \qquad i=1,2,
\eeaa
be the $\ell=1$ canonical modes of
$S_1, S_2 $. Then
\bea
\Big| J^1- J^2\circ \Psi \Big| &\les &\de.
\eea
The result, stated in Proposition \ref{Prop:comparison.canJ}, follows from Corollary \ref{Cor:unifor-twometrics-calibrated} which is an immediate consequence
of Theorem \ref{Thm:unifor-twometrics} and calibration.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Construction of GCM spheres with canonical $\ell=1$ modes}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Canonical version of Theorem \ref{Theorem:ExistenceGCMS1-intro}}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{theorem}[GCM spheres with canonical $\ell=1$ modes]
\lab{Theorem:ExistenceGCMS1-can-intro}
Under the same assumptions as in Theorem \ref{Theorem:ExistenceGCMS1-intro}
there exists a unique\footnote{Up to a rotation of $\SSS^2$.} GCM sphere $\S=\S^{(\La, \Lab)}$, which is a deformation of $\ovS$,
such that
\bea
\lab{Introd:GCM spheres2-again'}
\ka^\S -\frac{2}{r^\S} =0, \qquad \left( \kab^\S+\frac{2\Up^\S}{r^\S}\right)_{\ell\ge 2 }=0, \qquad \left( \mu^\S - \frac{2m^\S}{(r^\S)^3} \right)_{\ell\ge 2}=0,
\eea
and
\bea
\lab{Introd:GCM spheres-LaLab'}
(\div^\S f)_{\ell=1}=\La, \qquad (\div^\S\fb)_{\ell=1}=\Lab,
\eea
with canonically defined $\ell=1$
modes, i.e. $ \J^{(p,S)}=J^{(p,\SSS^2)}\circ \Phi^{-1}$, see \eqref{intro-canmodes-intro}.
\end{theorem}
The result is restated in Theorem \ref{Theorem:ExistenceGCMS1-2}.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Intrinsic GCM spheres}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Using Theorem \ref{Theorem:ExistenceGCMS1-can-intro} we prove an intrinsic\footnote{Without reference to the background foliation.} result on the existence and uniqueness of GCM surfaces where we replace the $\ell=1$ conditions \eqref{Introd:GCM spheres-LaLab'} on $\div^\S(f)$ and $\div^\S(\fb)$ by the vanishing of the $\ell=1$ modes of $\b^\S$ and $ \widecheck{\trchbS}$.
\begin{theorem}[Intrinsic GCM spheres with canonical $\ell=1$ modes]
\lab{theorem:ExistenceGCMS2-Intro}
Under slightly stronger assumptions on the background foliation of $\RR$ there
exists a unique\footnote{Up to a rotation of $\SSS^2$.} GCM deformation of $\ovS$ verifying, in addition to \eqref{Introd:GCM spheres2-again'},
\bea
( \div^\S \b^\S)_{\ell=1}=0, \qquad \widecheck{\trchbS}_{\ell=1}=0,
\eea
relative to the canonical $\ell=1$ modes of $\S$.
\end{theorem}
The result is restated in Theorem \ref{theorem:ExistenceGCMS2}.
\begin{remark}
The assumptions on the spacetime region $\RR$ in Theorem \ref{theorem:ExistenceGCMS2-Intro} are in particular satisfied in Kerr for $r$ sufficiency large, see Lemma \ref{lemma:controlfarspacetimeregionKerrassumptionRR}. We can thus apply Theorem \ref{theorem:ExistenceGCMS2-Intro} in that context, and obtain the existence of intrinsic GCM spheres $\S_{Kerr}$ in Kerr for $r$ sufficiency large, see Corollary \ref{cor:ExistenceGCMS1inKerr}. The intrinsic GCM spheres $\S$ of Theorem \ref{theorem:ExistenceGCMS2-Intro} thus correspond to the analog of $\S_{Kerr}$ in perturbations of Kerr for $r$ sufficiency large.
\end{remark}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Definition of angular momentum}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
The result of Theorem \ref{theorem:ExistenceGCMS2-Intro} is unique up to a rotation
of $\SSS^2$ in the definition of the canonical $\ell=1$ modes on $\S$, see \eqref{intro-canmodes-intro}. We can remove this final ambiguity\footnote{The only case where the ambiguity cannot be removed is when \eqref{eq:removalofambiguityrotationthankstocurlbeta-intro} holds for any one choice of canonical modes, i.e. $(\curl\b^\S)_{\ell=1}=0$. In that case, we set $a^\S=0$ which is consistent with \eqref{eq:definitionofangularmomentumaSonthesphereS-intro}.}
by adjusting the choice of canonical modes $\J^{(p, \S)} $ on $\S$, performing a suitable rotation in $\SSS^2$, such that
\bea\lab{eq:removalofambiguityrotationthankstocurlbeta-intro}
\int_\S \curl^\S \b^\S\, \J^{(\pm, \S)}=0.
\eea
This allows us to define on $\S$ the angular parameter $a^\S$ by the formula\footnote{Note that in a Kerr space $\KK(a, m)$, relative to a geodesic foliation normalized on $\II^+$, we have $ \int_\S \curl^\S \b^\S \J^{(\pm, \S)}=0$ and $ \int_\S \curl^\S \b^\S \J^{(0, \S)} =\frac{8\pi a m}{ (r^\S)^3}+O(\frac{ma^2}{(r^\S)^4})$.}
\bea\lab{eq:definitionofangularmomentumaSonthesphereS-intro}
a^\S:=\frac{(r^\S)^3}{8\pi m^\S} \int_\S \curl^\S \b^\S \J^{(0, \S)}.
\eea
The result is stated more precisely in Corollary \ref{Corr:ExistenceGCMS2}.
\begin{remark}
Given its relation with GCM spheres, we expect that the definition given above will play a key role in determining the final angular momentum for general perturbations of Kerr.
We also note that other definitions of angular momentum have been proposed in the literature, see
\cite{Sz} for a comprehensive review, and \cite{Rizzi} and \cite{Chen} for recent interesting proposals.
\end{remark}
%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Structure of the paper}
%%%%%%%%%%%%%%%%%%%%%%%%
The structure of the paper is as follows.
\begin{itemize}
\item In section 2, we review some of the classic results on the uniformization theorem for spheres.
\item In section 3, we provide a proof of our first effective uniformization result stated in Theorem \ref{Thm:effectiveU1-Intro} and use it to define canonical basis of $\ell=1$ modes.
\item
In Section 4, we prove our main stability result for effective uniformization stated in Theorem \ref{Thm:effectiveU2-Intro}.
\item In Section 5, we review the geometric set-up we need to state our GCM results.
\item In Section 6, we review the main result of \cite{KS-Kerr1}, stated in Theorem \ref{Theorem:ExistenceGCMS1-intro}, and prove its new version, stated in Theorem \ref{Theorem:ExistenceGCMS1-can-intro}, using the notion of canonical ${\ell=1} $ basis introduced in section 3.
\item In section 7, we prove our intrinsic GCM result stated in Theorem \ref{theorem:ExistenceGCMS2-Intro} and use it to define our notion of angular momentum. We also illustrate the result in Corollary \ref{cor:ExistenceGCMS1inKerr} applied to a far region of a Kerr spacetime.
\end{itemize}
%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Acknowledgments}
%%%%%%%%%%%%%%%%%%%%%%%%
The authors are grateful to A. Chang and C. De Lellis for very helpful suggestions and references in connection to our results on effective uniformization in sections 3 and 4.
The first author is supported by the NSF grant DMS 180841 as well as by the Simons grant 10011738. He would like to thank the Laboratoire Jacques-Louis Lions of Sorbonne Universit\'e and IHES for their hospitality during his many visits. The second author is supported by ERC grant ERC-2016 CoG 725589 EPGR.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Review of uniformization results for the sphere}
\lab{sec:reviewstandarduniformization}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
In this section, we review some well-known results concerning the
uniformization theorem for the sphere that will be used in section \ref{sec:effectiveuniformization}.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Uniformization for metrics on $\SSS^2$}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
We start with the following well known calculation.
\begin{lemma}
\lab{Le:transf-Gauss}
Let $S$ a surface and let $g^S$ be a Riemannian metric on $S$. For a scalar function $u$ on $S$, the Gauss curvature of $ e^{2u} g^S$ is connected to that of $g^S$ by the formula
\bea
K( e^{2u} g^S)=e^{-2u} \big(K^S-\lap_{S} u \big)
\eea
where $K^S$ denotes the Gauss curvature of $S$ and $\lap_S $ the Laplace-Beltrami on $S$.
\end{lemma}
According to the classical uniformization theorem, if $S$ is a closed, oriented and connected surface of genus 0 and $g^S$ is a Riemannian metric on $S$, then, there exists a smooth diffeomorphism $\Phi:\SSS^2\to S$ and a smooth conformal factor
$u$ on $\SSS^2$ such that
\bea
\lab{standard-unifS}
\Phi^\#( g^S) = e^{2u} \ga_0
\eea
where $\ga_0$ is the canonical metric on the standard sphere $\SSS^2$. In view of Lemma \ref{Le:transf-Gauss} above, if we denote by $g$ the metric on $\SSS^2$, $g=\Phi^\# (g^S) = e^{2u} \ga_0$, we derive
\bea
\lab{eq:Euler-K}
\lap_0 u +K(g) e^{2u} =1
\eea
where we have used the notation
\beaa
\De_0 &:=& \De_{\ga_0}.
\eeaa
We thus have the following corollary of Lemma \ref{Le:transf-Gauss}.
\begin{corollary}
\lab{Corr:transf-Gauss}
If $\Phi^\# g^S= e^{2u} \ga_0 $, then $u $ verifies the equation
\bea
\lab{eq:Euler-K-Phi}
\lap_0 u + (K^S\circ \Phi) e^{2u} =1,
\eea
where $K^S$ is the Gauss curvature of $g^S$ on $S$.
\end{corollary}
\begin{proof}
The proof follows from \eqref{eq:Euler-K} in view of the fact that $K\big(\Phi^\# g^S\big) = K(g^S)\circ \Phi$.
\end{proof}
\begin{definition}
Let $\M$ denote the group of conformal transformations of $\SSS^2$, i.e. the set of diffeomorphisms $\Phi$ of $\SSS^2$ such that $\Phi^\# \ga_0= e^{2u} \ga_0$ for some scalar function $u$ on $\SSS^2$.
\end{definition}
\begin{remark}\lab{rem:remarkonequationEuler-K=1}
Let $\Phi\in\M$ so that $\Phi^\# \ga_0= e^{2u} \ga_0$. Then, $u$ satisfies\footnote{This follows immediately from writing $\Phi^\# \ga_0= e^{2u} \ga_0$ in matrix form, by evaluating on an orthonormal frame, and then taking the absolute value of the determinant on both sides. Also, recall that $|\det d\Phi|$ is an intrinsic scalar on $\SSS^2$, i.e. it does not depend on the particular choice of orthonormal frame.}
\beaa
u= \frac 12 \log |\det d\Phi|.
\eeaa
Also, in view of Corollary \ref{Corr:transf-Gauss}, we have
\bea
\lab{eq:Euler-K=1}
\lap_0 u+ e^{2u} =1.
\eea
\end{remark}
\begin{lemma}
All solutions of \eqref{eq:Euler-K=1} are of the form $u=\frac 1 2 \log | \det d\Phi|$ with $\Phi\in \M$.
\end{lemma}
\begin{proof}
We have already checked that for every $\Phi\in \M$, $u=\frac 12 \log |\det d\Phi| $ is a solution of \eqref{eq:Euler-K=1}, see Remark \ref{rem:remarkonequationEuler-K=1}.
On the other hand, if $u$ is a solution of that equation, then $ K(e^{2u}\ga_0) =1$ by Lemma \ref{Le:transf-Gauss} which implies that there exist a diffeomorphism
$\Phi:\SSS^2\to \SSS^2 $ such that $\Phi^\# \ga_0=e^{2u}\ga_0$, i.e. $\Phi\in \M $ and $u=\frac 1 2 \log|\det d \Phi| $.
\end{proof}
\begin{corollary}
If $u$ is a solution of \eqref{eq:Euler-K} with $K=K(g)$, and if $\Phi\in \M$, then
\beaa
u_\Phi &:=& u\circ \Phi+ \frac 1 2 \log \big|\det d\Phi \big|
\eeaa
is also a solution of \eqref{eq:Euler-K} with $K$ replaced by $K\circ \Phi$.
\end{corollary}
\begin{proof}
If $\Phi\in \M$ then, since $\Phi^\# \ga_0=|\det(d\Phi)| \ga_0$ and $g=e^{2u}\gamma_0$,
\beaa
\Phi^\# g= e^{2 u\circ \Phi} \Phi^\# \ga_0=e^{2u\circ \Phi} |\det (d\Phi)| \ga_0= e^{2 v}\ga_0, \qquad v=u\circ \Phi+ \frac 1 2 \log\big| \det d\Phi\big|.
\eeaa
Hence if $u$ is a solution of \eqref{eq:Euler-K}, with $K=K(g) $, then so is $ u_\Phi=u\circ \Phi+ \frac 1 2 \log \big|\det d\Phi \big| $ with $K$ replaced by $K\circ \Phi$, as $K\big(\Phi^\# g\big) = K(g)\circ \Phi$.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Conformal isometries of $\SSS^2$ and the M\"obius group}
\lab{sec:conformalisometrymoebiusgroup}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
We represent the standard sphere $\SSS^2$ as $\big\{ x\in\RRR^3, |x|^2=1\big\}$. Let $N=(0,0, 1)$ denote the north pole of $\SSS^2$.
Through the stereographic projection from the North pole to the equatorial plane plane $(x^1, x^2)$ we consider the complex coordinate
\bea
z=\frac{x^1+i x^2}{1-x^3}
\eea
with the inverse transformation
\bea\lab{eq:inversestereographicprojectionformula}
x^1= \frac{2}{1+|z|^2} \Re z,\quad x^2 =\frac{2}{1+|z|^2} \Im z, \quad x^3 =\frac{|z|^2-1}{|z|^2+1}
\eea
and pull-back of the standard metric $\ga_0$ on $\SSS^2$
\bea
\lab{standard sphere-z}
4 \big(1+ |z|^2\big)^{-2} |dz|^2.
\eea
The conformal isometry group $\M$ of $\SSS^2$ consists in fact of M\"obius transforms and conjugation of M\"obius transforms, see for example Theorem 18.10.4 and section 18.10.2.4 in \cite{Berger}, and can thus be identified with $SL(2,\CCC)$,
\bea
z\to \frac{az+b}{cz+d},\qquad z\to \frac{a\ov{z}+b}{c\ov{z}+d},\qquad ad-bc=1, \quad a,b, c, d\in \CCC.
\eea
The particular case of M\"obius transforms where $d=a^{-1}>0$ and $b=c=0$ will play an important role. Given $t>0$ and a point $p\in \SSS^2$, we can choose coordinates such that $p$ is at the north pole and obtain scale transformations defined by
\bea\lab{eq:defintionscaletranformations}
\Phi_{p, t} z= tz.
\eea
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Onofri functional and inequality}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Following \cite{Onofri}, as well as \cite{CY1} \cite{CY2}, we introduce the functional
\bea
\bsplit
S[u]&=\frac{1}{4\pi} \left( \int_{\SSS^2} |\nab u|^2+ 2 \int_{\SSS^2} u\right).
\end{split}
\eea
\begin{lemma}
\lab{eq:Gageforu}
The functional $S$ is invariant with respect to the transformations $ u\to u_\Phi$, i.e.
\bea
S[u_\Phi]=S[u].
\eea
\end{lemma}
\begin{proof}
See \cite{Onofri}.
\end{proof}
\begin{definition}
We define the center of mass of $e^{2u}$ to be
\bea
CM[e^{2u}]&=&\frac{\int_{\SSS^2} x e^{2u} }{\int_{\SSS^2} e^{2u}}
\eea
where $x=(x^1, x^2, x^3)$ on the sphere $\SSS^2$.
Also, we define the spaces of functions
\bea
\bsplit
\SS:=&\Big\{ u\in H^1(\SSS^2)\,\,\textrm{ such that }\,\, CM[e^{2u}]=0\Big\}, \\
\SS_0:=&\left\{ u\in \SS \,\,\textrm{ such that }\,\, \int_{\SSS^2} e^{2u}=4\pi\right\}.
\end{split}
\eea
\end{definition}
\begin{remark}\lab{rem:ifuiscenteredthensoisucricO}
If $u$ belongs to $\SS$ (resp. $\SS_0$), then $u\circ O$ also belongs to $\SS$ (resp. $\SS_0$) for any $O\in O(3)$.
\end{remark}
\begin{lemma}
\lab{le:renormalization}
Given $u$ a smooth function on $\SSS^2$ there exists a conformal transformation $\Phi$ such that $u_\Phi\in \SS$, i.e. $\int_{\SSS^2} e^{2 u_\Phi} x =0$.
\end{lemma}
\begin{proof}
See \cite{Onofri}, as well as Proposition 2.2. in \cite{CY1} and Lemma 2 in \cite{Chang}.
\end{proof}
\begin{remark}
In fact, $\Phi$ in Lemma \ref{le:renormalization} is unique up to isometries of $\SSS^2$, see Lemma \ref{lemma:improved-renormalization:bis}.
\end{remark}
\begin{proposition}[Onofri]
\lab{Thm:Onofri}
Given $f\in H^1(\SSS^2) $ we have,
\bea
\lab{eq:Thm.Onofri}
\log\left(\int_{\SSS^2} e^{2u}\right) &\le S[u]
\eea
with equality iff $u=\frac 1 2 \log |\det \Phi| $ for some conformal $\Phi\in \M$.
\end{proposition}
\begin{proof}
See \cite{Onofri}.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Improved Onofri inequality and applications}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
The following is Proposition B in \cite{CY1} which improves the Onofri inequality for $u\in \SS$.
\begin{proposition}
\lab{PropChang-:propB}
There exists an $a<1$ such that for all $u\in \SS$
\bea
\frac{1}{4\pi}\int e^{2u}\le \exp{\left(a\frac{1}{4\pi}\int |\nab u|^2+ 2 \frac{1}{4\pi}\int u \right)}.
\eea
\end{proposition}
\begin{corollary}
\lab{cor:Chang1}
Let $a<1$ as in Proposition \ref{PropChang-:propB}. If $u\in \SS_0$ then,
\bea
\frac{1}{4\pi} \int|\nab u|^2 \le (1-a)^{-1} S[u].
\eea
\end{corollary}
\begin{proof}
Since $u\in \SS_0$, we have in view of Proposition \ref{PropChang-:propB}
\bea
1=\frac{1}{4\pi}\int e^{2u}\le \exp{\left(a\frac{1}{4\pi}\int |\nab u|^2+ 2 \frac{1}{4\pi}\int u \right)}
\eea
from which we deduce
\beaa
a\frac{1}{4\pi}\int|\nab u|^2+ 2 \frac{1}{4\pi}\int u &\ge& 0.
\eeaa
By definition of $S[u]$ we infer
\beaa
(1-a)\frac{1}{4\pi}\int |\nab u|^2 =S[u]-\left( a \frac{1}{4\pi}\int |\nab u|^2 +2 \frac{1}{4\pi}\int u\right)\le S[u]
\eeaa
as desired.
\end{proof}
The following is Proposition 4.1 in \cite{CY2}.
\begin{proposition}
\lab{prop:Chang}
Let $K>0$ be the Gauss curvature of the metric $e^{2w} \ga_0$ on $\SSS^2$, so that $w$ verifies \eqref{eq:Euler-K}, i.e. $\lap w +K e^{2 w} =1$.
Then $S[w] \leq C(K)$ for some constant $C$ depending only on $\max_{\SSS^2} K$ and $\min_{\SSS^2} K$.
\end{proposition}
\begin{proof}
In view of Lemma \ref{le:renormalization} and Lemma \ref{eq:Gageforu}, we may assume that $w$ is centered, i.e. $w\in \SS$. The proof then follows by exploiting that $w$ satisfies $\lap w +K e^{2 w} =1$, and relying on an estimate satisfied by functions in $\SS$. See \cite{CY2} for the details.
\end{proof}
The following is Corollary 4.3. in \cite{CY2}.
\begin{corollary}
\lab{corr:Alice}
Given any $\de>0$, there exists $\ep>0$ sufficiently small such that if $\|K-1\|_{\infty} \leq \ep$ then $S[w] \le \de$ for all $w, K$ which verify equation \eqref{eq:Euler-K}.
\end{corollary}
\begin{proof}
The proof, based on a contradiction argument, relies on Lemma \ref{eq:Gageforu}, Proposition \ref{prop:Chang} and the fact that any centered solution of $\lap_0 w +e^{2w}=1$ must vanish identically. See \cite{CY2} for the details.
\end{proof}
We restate the result in the following.
\begin{corollary}
\lab{corr:Alice2}
Let $g$ a metric on $\SSS^2$ such that $ g= e^{2u} \ga_0 $ with $u\in \SS_0$.
Then, given any $\de>0$, there exists $\ep>0$ sufficiently small such that if
$ \| K(g)-1\|_{L^\infty} \leq \ep$, then
\bea
\| u\|_{H^1} &\leq& \de.
\eea
\end{corollary}
\begin{proof}
According to Corollary \ref{corr:Alice}, given $\de >0$ sufficiently small,
we can find $\ep>0$ such that if $ \| K(g)-1\|_{L^\infty} \leq \ep$, then $S[u]\le \de^2$. In view of Corollary \ref{cor:Chang1}, we deduce
\beaa
\frac{1}{4\pi}\int |\nab u|^2 &\leq& (1-a)^{-1}\de^2.
\eeaa
Together with the bound $S[u]\leq \de^2$, we infer
\beaa
\|\nabla u\|_{L^2}+\left|\int u\right| &\les& \de.
\eeaa
Together with the Poincar\'e inequality for scalars on $\SSS^2$, we infer
\beaa
\|u\|_{H^1} &\les & \de
\eeaa
as desired.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Effective uniformization for nearly round spheres}
\lab{sec:effectiveuniformization}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Let $(S, g^S)$ be a fixed sphere of area radius $1$, i.e. $|S| =4\pi$. The goal of this section is to obtain the following improvement of the results reviewed in section \ref{sec:reviewstandarduniformization}.
\begin{theorem}[Effective uniformization]
\lab{theorem-Eff.uniformalization-appendix}
Let $(S, g^S)$ be a fixed sphere with $|S| =4\pi$. There exists, up to isometries\footnote{i.e. all the solutions are of the form $(\Phi\circ O, u\circ O)$ for $O\in O(3)$. In particular, recall that if $u$ is centered, then so is $u\circ O$ for $O\in O(3)$, see Remark \ref{rem:ifuiscenteredthensoisucricO}.} of $\SSS^2$, a unique diffeomorphism $\Phi:\SSS^2\to S$ and a unique centered conformal factor $u$, i.e. $u\in\SS$, such that $ \Phi^\#( g^S) = e^{2u} \ga_0$. Moreover, under the almost round condition
\bea
\lab{eq:Almostround}
\|K^S-1\|_{L^\infty} \le \ep
\eea
where $K^S=K(g^S)$, the following properties are verified for sufficiently small $\ep>0$.
\begin{enumerate}
\item We have
\bea
\lab{eq:Thm.-Eff.uniformalization1-appendix}
\| u\circ\Phi^{-1} \|_{L^\infty (S) } \les \ep.
\eea
\item If in addition $\|K^S -1\|_{H^s(S)} \leq \ep$ for some $s\geq 0$, then
\bea
\lab{eq:Thm.-Eff.uniformalization1-appendix:bis}
\| u\circ \Phi^{-1} \|_{H^{2+s}(S) } \les \ep.
\eea
\end{enumerate}
\end{theorem}
\begin{remark}
To the best of our knowledge the estimates \eqref{eq:Thm.-Eff.uniformalization1-appendix}, \eqref{eq:Thm.-Eff.uniformalization1-appendix:bis} have not been stated as such in the literature. The uniqueness statement also appears to be new.
\end{remark}
\begin{remark}
One can easily adapt the statement of Theorem \ref{theorem-Eff.uniformalization-appendix} to the case $|S|\neq 4\pi$, see Corollary \ref{proposition:effective-uniformisation}.
\end{remark}
Theorem \ref{theorem-Eff.uniformalization-appendix} will be proved in section \ref{sec:proofof-theorem-Eff.uniformalization-appendix}. We first provide improvements of Lemma \ref{le:renormalization} and Corollary \ref{corr:Alice2}.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Uniqueness for Lemma \ref{le:renormalization}}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
In this section, we provide a uniqueness statement for Lemma \ref{le:renormalization}, see Lemma \ref{lemma:improved-renormalization:bis}. We also strengthen the conclusions of Lemma \ref{le:renormalization} for nearly round spheres, see Lemma \ref{lemma:improved-renormalization}. To this end, we start with the following decomposition of conformal isometries of $\SSS^2$.
\begin{lemma}\lab{lemma:decompositionofhomographies}
Any $\Phi\in \M$ admits the following decomposition
\beaa
\Phi=O_1\circ\Phi_{N, t}\circ O_2
\eeaa
where $O_1, O_2\in O(3)$, $N=(0,0,1)$, $t>0$, and the scale transformation $\Phi_{N, t}$ has been introduced in \eqref{eq:defintionscaletranformations}.
\end{lemma}
\begin{proof}
Upon multiplying $\Phi$ by a reflexion, which belongs to $O(3)$, we may assume that $\Phi$ corresponds, in view of section \ref{sec:conformalisometrymoebiusgroup}, to a M\"obius transformation, and hence to a matrix in $SL(2,\CCC)$. Then, using that, see for example \cite{JoSi},
\begin{enumerate}
\item composition of M\"obius transformations corresponds to matrix multiplication in $SL(2,\CCC)$,
\item $SO(3)$ corresponds to M\"obius transformations with matrices in $SU(2)$,
\item scale transformations $\Phi_{N, t}$ with $t>0$ correspond to the following matrix in $SL(2,\CCC)$
\beaa
\left(\ba{cc}
\sqrt{t} & 0\\
0 & \frac{1}{\sqrt{t}}
\ea\right),
\eeaa
\end{enumerate}
the lemma reduces to proving the following decomposition for a matrix $A\in SL(2,\CCC)$
\bea\lab{eq:reductionofthedecompositionMtoSL2C}
A=U_1DU_2, \qquad U_1, U_2\in SU(2), \qquad D=\left(\ba{cc}
\sqrt{t} & 0\\
0 & \frac{1}{\sqrt{t}}
\ea\right).
\eea
To prove the claim \eqref{eq:reductionofthedecompositionMtoSL2C}, we apply the polar decomposition to $A$
\beaa
A=RU_3, \qquad R=R^*, \qquad R>0, \qquad \det(R)=1, \qquad U_3\in SU(2),
\eeaa
where $\det(R)=1$ and $U_3\in SU(2)$ since $\det(A)=1$. Also, $R$ being hermitian semidefinite positive, we may diagonalize it as follows
\beaa
R=U_1\left(\ba{cc}
\la & 0\\
0 & \la^{-1}
\ea\right)U_1^*, \qquad U_1\in SU(2), \qquad \la>0,
\eeaa
where we have used the fact that $\det(R)=1$. Finally, we have
\beaa
A=U_1\left(\ba{cc}
\sqrt{t} & 0\\
0 & \frac{1}{\sqrt{t}}
\ea\right)U_2, \textrm{ where }U_2=U_1^*U_3, \qquad t=\la^2>0,
\eeaa
which is \eqref{eq:reductionofthedecompositionMtoSL2C} as desired.
\end{proof}
\begin{lemma}
\lab{lemma:improved-renormalization:bis}
Let $u$ be a smooth function on $\SSS^2$ and let $\Phi$ be a conformal transformation. Assume that both $u$ and $u_\Phi$ belong to $\SS$. Then, we have\footnote{Recall from Remark \ref{rem:ifuiscenteredthensoisucricO} that if $u$ belongs to $\SS$, then $u\circ O$ also belongs to $\SS$ for any $O\in O(3)$.} $\Phi\in O(3)$.
\end{lemma}
\begin{proof}
Decomposing $\Phi$ using Lemma \ref{lemma:decompositionofhomographies}, it suffices in fact to prove that if $u$ and $u_{\Phi_{N,t}}$ are in $\SS$, where $t>0$, then $\Phi_{N,t}=I$, i.e. $t=1$. Thus, from now on, we assume that $u$ and $u_{\Phi_{N,t}}$ are in $\SS$ with $t>0$.
Since $u$ and $u_{\Phi_{N,t}}$ are in $\SS$, we have
\bea\lab{eq:consequenceofuanduPhitzarebothinSS}
\nn 0 &=& \int_{\SSS^2}e^{2u_{\Phi_{N,t}}}x -\int_{\SSS^2}e^{2u}x =\int_{\SSS^2}e^{2u\circ\Phi_{N,t}}|\det(d\Phi_{N,t})|x -\int_{\SSS^2}e^{2u}x\\
& =& \int_{\SSS^2}e^{2u}x\circ\Phi_{N,t}^{-1} -\int_{\SSS^2}e^{2u}x = \int_{\SSS^2}e^{2u}\Big(x\circ\Phi_{N,\frac{1}{t}} -x\Big)
\eea
where we used the definition of $u_{\Phi_{N,t}}$, the change of variable formula, and the fact that $\Phi_{N,t}^{-1}=\Phi_{N,t^{-1}}$. Now, in view of the formula for $\Phi_{N,t}$, we have, using the formula for stereographic coordinates, see \eqref{eq:inversestereographicprojectionformula},
\beaa
x^1\circ\Phi_{N,t^{-1}}=\frac{2t^{-1}}{1+t^{-2}|z|^2} \Re z,\quad x^2\circ\Phi_{N,t^{-1}} =\frac{2t^{-1}}{1+t^{-2}|z|^2} \Im z, \quad x^3\circ\Phi_{N,t^{-1}} =\frac{t^{-2}|z|^2-1}{t^{-2}|z|^2+1}.
\eeaa
Since
\beaa
|z|^2=\frac{1+x^3}{1-x^3},
\eeaa
we infer, using \eqref{eq:inversestereographicprojectionformula} again,
\bea\lab{eq:xcomposedwithscaletransformation}
\bsplit
x^1\circ\Phi_{N,-t} &= \frac{2t^{-1}}{t^{-2}(1+x^3)+(1-x^3)}x^1,\\
x^2\circ\Phi_{N,-t} &= \frac{2t^{-1}}{t^{-2}(1+x^3)+(1-x^3)}x^2, \\
x^3\circ\Phi_{N,-t} &= \frac{t^{-2}(1+x^3)-(1-x^3)}{t^{-2}(1+x^3)+(1-x^3)}.
\end{split}
\eea
In particular, the last identity of \eqref{eq:xcomposedwithscaletransformation} yields
\beaa
x^3\circ\Phi_{N,-t} -x^3 &=& (t^{-2}-1)\frac{1-(x^3)^2}{t^{-2}(1+x^3)+(1-x^3)}
\eeaa
and thus,
\beaa
\int_{\SSS^2}e^{2u}\Big(x^3\circ\Phi_{N,\frac{1}{t}} -x^3\Big) &=& (t^{-2}-1)\int_{\SSS^2}e^{2u}\frac{1-(x^3)^2}{t^{-2}(1+x^3)+(1-x^3)}
\eeaa
so that we have, in view of \eqref{eq:consequenceofuanduPhitzarebothinSS},
\beaa
(t^{-2}-1)\int_{\SSS^2}e^{2u}\frac{1-(x^3)^2}{t^{-2}(1+x^3)+(1-x^3)} &=& 0.
\eeaa
Since the integral is strictly positive and $t>0$, we infer $t=1$ as desired.
\end{proof}
\begin{lemma}
\lab{lemma:improved-renormalization}
Let $\ep$ and $\de$ two constants such that $0<\de\leq\ep$. Given $u$ a smooth function on $\SSS^2$ such that
\beaa
\left|\int_{\SSS^2}e^{2u}x\right|\leq\de, \qquad \|u\|_{L^\infty}\leq \ep.
\eeaa
Then, for $\ep>0$ sufficiently small, there exists a conformal transformation $\Phi$ such that $u_\Phi\in \SS$ and satisfying in addition $\big| \Phi-I \big|\les \de$.
\end{lemma}
\begin{proof}
Let $p_1=(1,0,0)$, $p_2=(0,1,0)$ and $p_3=(0,0,1)$, and let us consider the map
\beaa
\Theta: \RRR^3\to\RRR^3, \qquad \Theta(t)=\int_{\SSS^2}e^{2u_{\Phi_{p_1, t_1}\circ\Phi_{p_2, t_2}\circ\Phi_{p_3, t_3}}}x.
\eeaa
To prove the lemma, it suffices to exhibit $t\in\RRR^3$ such that $\Theta(t)=0$ and $|t-(1,1,1)|\les\de$.
Using the definition of $u_{\Phi_{p_i,t_i}}$, the change of variable formula, and the fact that $\Phi_{p_i,t_i}^{-1}=\Phi_{p_i,t_i^{-1}}$, we have
\beaa
\Theta(t) &=& \int_{\SSS^2}e^{2u}x\circ\Phi_{p_3, t_3^{-1}}\circ\Phi_{p_2, t_2^{-1}}\circ\Phi_{p_1, t_1^{-1}}.
\eeaa
Since $|\int_{\SSS^2}e^{2u}x|\leq\de$, and since $\Phi_{p_i,1}$ is the identity, we have
\bea\lab{eq:estimateofThetaforapplyingtheBanachfixedpointtheoremforzerosofTheta}
\Theta(1,1,1) &=& \int_{\SSS^2}e^{2u}x=O(\de).
\eea
Next, we compute the differential of $\Theta$ at $t=(1,1,1)$. To this end, note that we have in view of \eqref{eq:xcomposedwithscaletransformation}
\beaa
\Theta(1,1,t_3) &=& \int_{\SSS^2}e^{2u}x\circ \Phi_{p_3, \frac{1}{t_3}}=\Theta(1,1,1)+\int_{\SSS^2}e^{2u}\Big(x\circ \Phi_{N, \frac{1}{t_3}}-x\Big)\\
&=& \Theta(1,1,1)+\int_{\SSS^2}e^{2u}
\left(\ba{c}
\frac{2t^{-1}}{t^{-2}(1+x^3)+(1-x^3)}x^1\\
\frac{2t^{-1}}{t^{-2}(1+x^3)+(1-x^3)}x^2\\
\frac{t^{-2}(1+x^3)-(1-x^3)}{t^{-2}(1+x^3)+(1-x^3)}
\ea\right).
\eeaa
We infer, using also $\|u\|_{L^\infty}\leq \ep$,
\beaa
\pr_{t_3}\Theta(1,1,1) &=& \int_{\SSS^2}e^{2u}\left(\ba{c}
2x^1x^3\\
2x^2x^3\\
-(1-(x^3)^2)
\ea\right)=O(\ep)+\int_{\SSS^2}\left(\ba{c}
2x^1x^3\\
2x^2x^3\\
-(1-(x^3)^2)
\ea\right)
\eeaa
and hence
\beaa
\pr_{t_3}\Theta(1,1,1)=-\frac{8\pi}{3}\left(\ba{c}
0\\
0\\
1
\ea\right)+O(\ep).
\eeaa
One easily derives corresponding identities for $\pr_{t_1}\Theta(1,1,1)$ and $\pr_{t_2}\Theta(1,1,1)$ which yields
\bea\lab{eq:estimateofThetaforapplyingtheBanachfixedpointtheoremforzerosofTheta:bis}
d\Theta_{|_{(1,1,1)}}=-\frac{8\pi}{3}I+O(\ep).
\eea
Thus, finding $t$ such that $\Theta(t)=0$ amounts to solving the following fixed point
\beaa
t-(1,1,1) &=& -(d\Theta_{|_{(1,1,1)}})^{-1}\left[\Theta(1,1,1) +\left(\Theta(t) -\Theta(1,1,1)-d\Theta_{|_{(1,1,1)}}\big(t-(1,1,1)\big)\right)\right]
\eeaa
whose existence, together with the desired estimate $|t-(1,1,1)|\les\de$, follows, in view of \eqref{eq:estimateofThetaforapplyingtheBanachfixedpointtheoremforzerosofTheta} and \eqref{eq:estimateofThetaforapplyingtheBanachfixedpointtheoremforzerosofTheta:bis}, from the Banach fixed point theorem.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Improvement of Corollary \ref{corr:Alice2} for nearly round spheres}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
In the proposition below, we improve the results of Corollary \ref{corr:Alice2} for nearly round spheres.
\begin{proposition}
\lab{proposition:effective-uniformisation-Appendix}
Assume that the metric $g= e^{2u} \ga_0$ on $\SSS^2$, with $u\in \SS_0$, is such that $ \| K(g)-1\|_{L^\infty} \leq \ep$.
Then, for $\ep>0$ sufficiently small, we have
\bea\lab{eq:boundin:proposition:effective-uniformisation-Appendix}
\| u\|_{L^\infty(\SSS^2)} \les \ep.
\eea
\end{proposition}
\begin{proof}
Let $\de>0$ a sufficiently small universal constant which will be chosen later. Then, according to Corollary \ref{corr:Alice2}, if $\ep$, such that $\|K-1\|_{L^\infty}\le \ep$ holds, is sufficiently small, i.e. $0<\ep\leq \ep(\de)\ll 1$, we have
\beaa
\| u\|_{H^1} &\le& \de.
\eeaa
Our goal is to improve this estimate by showing that the stronger bound \eqref{eq:boundin:proposition:effective-uniformisation-Appendix} holds.
Since $g= e^{2u} \ga_0$, the scalar function $u$ verifies equation \eqref{eq:Euler-K}, i.e.
\beaa
\lap_0 u+ K e^{2u} =1,
\eeaa
where $K=K(g)$, which we rewrite in the form,
\beaa
\lap_0 u+ (e^{2u}-1)= -(K-1 ) e^{2u}
\eeaa
or,
\beaa
\lap_0 u+ 2 u &=& - (K-1 ) e^{2u} - f(u), \quad\textrm{ where }\quad f(u):= e^{2u}-1-2u=\sum_{n\ge 2 } \frac{2^n}{n!} u^n.
\eeaa
Also, since $u\in\SS_0$, we have $\int e^{2u} x =0$. Together with $\int x=0$, we deduce,
\beaa
0 &=& \int e^{2u}x=\int \big( 1+2u+ f(u) \big)x=2 \int u x +\int f(u) x
\eeaa
and hence
\beaa
\left|\int u\, x \right| \les \int |f(u)|\les \|f(u)\|_{L^2}.
\eeaa
Note that $f(u)\le 4 u^2 e^{ 2u} $.
By standard elliptic estimates we have\footnote{Note that the kernel of $\lap_0 + 2$ is given by the $\ell=1$ spherical harmonics, and that $\int u x$ corresponds to the projection on these spherical harmonics.}
\beaa
\| u\|_{H^2}&\les& \| (\lap_0 + 2)u \|_{L^2}+ \left|\int u\, x \right|
\eeaa
and hence
\beaa
\| u\|_{H^2}&\les& \|(K-1) e^{2u} \|_{L^2} +\| f(u) \|_{L^2} + \left|\int u\, x \right| \\
&\les&\ep \| e^{2u} \|_{L^2}+ \| f(u) \|_{L^2}\\
&\les& \ep \| e^{2u} \|_{L^2} + 4 \| u\|_{L^\infty}^2 \| e^{2u} \|_{L^2}.
\eeaa
In view of Onofri inequality \eqref{eq:Thm.Onofri}, we have
\beaa
\log \int e^{4 u} &\le& S[2u] \les \| u\|_{H^1}^2 +\| u\|_{H^1} \les \de \le 1
\eeaa
where we have used $\| u\|_{H^1} \le\de$ and chosen $\de>0$ sufficiently small. We deduce,
\beaa
\| u\|_{H^2}&\les& \ep + \|u\|_{L^\infty}^2.
\eeaa
By a calculus inequality
\beaa
\| u\|_{L^\infty}\les \|u\|_{L^2}^{\frac{1}{2}} \| u\|_{H^2}^{\frac{1}{2}}.
\eeaa
Hence,
\beaa
\| u\|_{H^2}&\les& \ep+ \|u\|_{L^2} \| u\|_{H^2}.
\eeaa
Thus, since $\|u\|_{H^1}\le \de$, we infer that, in fact, for $\de>0$ small
\beaa
\| u\|_{H^2}&\les& \ep.
\eeaa
In other words, if $\|K-1\|_{L^\infty} \le \ep $, for $\ep>0$ sufficiently small, then $\|u\|_{H^2}\les \ep$. In particular, by Sobolev inequality, we have $\| u\|_{L^\infty}\les \ep$ as stated.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Proof of Theorem \ref{theorem-Eff.uniformalization-appendix}}
\lab{sec:proofof-theorem-Eff.uniformalization-appendix}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Existence part}
%%%%%%%%%%%%%%%%%%%%%%%%%
According to the standard uniformization theorem, there exists a scalar function $\tilde{u}$ and a map $\widetilde{\Phi}:\SSS^2\to S$ such that we have $\widetilde{\Phi}^\# (g^S)= e^{2\tilde{u}} \ga_0$ with $ \int_{\SSS^2} e^{2\tilde{u}}=4\pi$. In view of Lemma \ref{le:renormalization}, we can find $\Psi\in \M$ such that $u=\tilde{u}_\Psi\in \SS_0$, and then, $g=\Phi^\# (g^S)=e^{2u} \ga_0$ with $\Phi=\widetilde{\Phi}\circ\Psi$.
%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Uniqueness part}
%%%%%%%%%%%%%%%%%%%%%%%%%
We address below the issue of uniqueness, up to isometries of $\SSS^2$, of $(\Phi, u)$. Assume that we have
\bea
( \Phi_i)^\# g^S= e^{2u_i} \ga_0, \qquad i=1,2,
\eea
with $\Phi_i:\SSS^2\to S$ diffeomorphisms and $u_i$ centered conformal factors. Let $\Psi=\Phi_2^{-1} \circ \Phi_1$ so that $\Psi:\SSS^2\to \SSS^2$. In view of the above, we have
\beaa
\Psi^\# \ga_0&=& \Phi_1^\#(\Phi_2^{-1})^\#\ga_0=\Phi_1^\#(e^{-2u_2\circ\Phi_2^{-1}}g^S) =e^{-2u_2\circ\Phi_2^{-1}\circ\Phi_1}e^{2u_1}\ga_0
\eeaa
and hence
\beaa
\Psi^\# \ga_0&=& e^{2(u_1- u_2\circ \Psi)} \ga_0.
\eeaa
Therefore $\Psi$ is a conformal isometry of $\SSS^2$, and we deduce in view of Remark \ref{rem:remarkonequationEuler-K=1}
\beaa
u_1- u_2\circ \Psi=\frac 1 2 \log |\det\Psi|,
\eeaa
i.e.,
\beaa
u_1=( u_2)_\Psi.
\eeaa
Since both $u_1$ and $u_2$ are centered we deduce, in view of Lemma \ref{lemma:improved-renormalization:bis}, $\Psi=O$ with $O\in O(3)$ and therefore $\Phi_1=\Phi_2\circ O$ and $u_1=u_2\circ O$ as stated.
%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Estimates}
%%%%%%%%%%%%%%%%%%%%%%%%%
From now on, we assume in addition the almost round condition \eqref{eq:Almostround}. In view of \eqref{eq:Almostround}, and the fact that $g=e^{2u} \ga_0$ with $u\in \SS_0$, we deduce by Proposition \ref{proposition:effective-uniformisation-Appendix} that $\| u\|_{L^\infty(\SSS^2) } \les \ep $. This implies $\|u\circ\Phi^{-1} \|_{L^\infty(S)} \les \ep $ as stated in \eqref{eq:Thm.-Eff.uniformalization1-appendix}. To prove the higher derivative estimates we observe that
\bea
(\Phi^{-1})^\# \ga_0 &=& e^{-2 v}g^S, \qquad v:= u\circ\Phi^{-1},
\eea
which together with Lemma \ref{Le:transf-Gauss} implies
\beaa
1=K((\Phi^{-1})^\# \ga_0)=K( e^{-2v} g^S)=e^{2v} \big(K^S+\lap_{S} v \big)
\eeaa
so that $v$ verifies
\beaa
\lap_S v = \big(e^{-2v}-1\big) -\big(K^S-1\big), \qquad \|v\|_{L^\infty(S)} \les \ep.
\eeaa
In view of the almost round condition \eqref{eq:Almostround}, we easily deduce that in fact $\| v\|_{H^2(S)} \les \ep $.
Moreover, if we have in addition $\|K^S-1\|_{H^s} \le \ep$, then, by standard elliptic regularity, we obtain $\| v\|_{H^{2+s} (S)} \les \ep$ as stated.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Effective uniformization for nearly round spheres of arbitrary area}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Let $(S, g^S)$ be a fixed sphere, and let $r^S$ denote its area radius, i.e. $r^S$ satisfies
\beaa
|S|=4\pi (r^S)^2.
\eeaa
Given a positive integer $s$, we introduce the following norm on $S$
\bea
\lab{definition:spaceH^k(Sbutnotbold)}
\| f\|_{\hk_s(S)}:&=&\sum_{i=0}^s \|( r^S \nab^S )^i f\|_{L^2(S)}.
\eea
The goal of the following corollary is to extend Theorem \ref{theorem-Eff.uniformalization-appendix} to the case $r^S\neq 1$.
\begin{corollary}
\lab{proposition:effective-uniformisation}
Let $(S, g^S)$ be a fixed sphere. There exists, up to isometries\footnote{i.e. all the solutions are of the form $(\Phi\circ O, u\circ O)$ for $O\in O(3)$.} of $\SSS^2$, a unique diffeomorphism $\Phi:\SSS^2\to S$ and a unique centered conformal factor $u$, i.e. $u\in\SS$, such that
\beaa
\Phi^\#( g^S) = (r^S)^2e^{2u} \ga_0.
\eeaa
Moreover, under the almost round condition
\bea\lab{eq:Almostround-S}
\left\| K^S-\frac{1}{ (r^S) ^2} \right\|_{L^\infty(S)} &\leq& \frac{\ep}{(r^S)^2},
\eea
the following properties are verified for sufficiently small $\ep>0$.
\begin{enumerate}
\item We have
\bea
\lab{eq:Thm.-Eff.uniformalization1}
\| u\circ\Phi^{-1} \|_{L^\infty (S) } &\les& \ep.
\eea
\item If in addition
\bea\lab{eq:Almostround-S-k}
\left \|K^S -\frac{1}{(r^S)^2}\right\|_{\hk_s(S)} &\leq& \frac{\ep}{r^S}
\eea
for some $s\geq 0$, then
\bea
\left \|u\circ\Phi^{-1}\right\|_{\hk_{s+2}(S)} &\les& \ep r^\S.
\eea
\end{enumerate}
\end{corollary}
\begin{proof}
Consider the metric $\tilde{g}^S$ on $S$ given by
\beaa
\tilde{g}^S &:=& \frac{1}{(r^S)^2}g^S.
\eeaa
Then, $(S, \tilde{g}^S)$ has area radius 1, and $K(\tilde{g}^S)=(r^S)^{2}K(g^S)$ so that, in view of \eqref{eq:Almostround-S},
the almost round condition \eqref{eq:Almostround} holds for $\tilde{g}^S$. Thus, we may apply Theorem \ref{theorem-Eff.uniformalization-appendix} to $(S, \tilde{g}^S)$ which then implies the first two conclusions of Corollary \ref{proposition:effective-uniformisation} for $(S, g^S)$.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Canonical basis of $\ell=1$ modes on $S$}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Let $S$ be an almost round sphere, i.e. verifying \eqref{eq:Almostround-S}. The goal of this section is to define on $S$ a canonical generalization of the $\ell=1$ spherical harmonics.
Recall that the $\ell=1$ spherical harmonics $J^{\SSS^2}=(J^{(-, \SSS^2)}, J^{(0, \SSS^2)}, J^{(+, \SSS^2)} ) $ are given by the restriction of $x^1, x^2, x^3$ to $\SSS^2$. More precisely, in polar coordinates,
\bea\lab{eq:defofstandardsphericalharmonics}
J^{(0, \SSS^2)} =x^3= \cos\th, \qquad J^{(+, \SSS^2)} = x^1=\sin\th \cos \vphi, \qquad J^{(-, \SSS^2)} = x^2=\sin\th \sin \vphi.
\eea
\begin{lemma}
We have, for $p, q\in\{-, 0, +\}$,
\bea
\bsplit
\lap_0 J^{(p, \SSS^2)} &= - 2 J^{(p, \SSS^2)},\\
\int_{\SSS^2 } J^{(p, \SSS^2)} J^{(q, \SSS^2)} da_{\ga_0} &= \frac{4\pi}{3} \de_{pq}, \\
\int_{\SSS^2} J^{(p, \SSS^2)} da_{\ga_0} &=0.
\end{split}
\eea
\end{lemma}
\begin{proof}
Straightforward verification.
\end{proof}
\begin{definition}[Basis of canonical $\ell=1$ modes on $S$]
\lab{definition:ell=1mpdesonS}
Let $(S,g^S)$ be an almost round sphere, i.e. verifying \eqref{eq:Almostround-S}. Let $(\Phi, u)$ the unique, up to isometries of $\SSS^2$, uniformization pair given by Corollary \ref{proposition:effective-uniformisation},
i.e.,
\beaa
\Phi:\SSS^2\longrightarrow S, \qquad \Phi^\# (g^S) = (r^S)^2e^{2u}\ga_0,\qquad u\in\SS.
\eeaa
We define the basis of canonical $\ell=1$ modes on $S$ by
\bea
J^S &:=& J^{\SSS^2}\circ \Phi^{-1},
\eea
where $J^{\SSS^2}$ denotes the $\ell=1$ spherical harmonics, see \eqref{eq:defofstandardsphericalharmonics}.
\end{definition}
\begin{remark}
Note that the canonical basis is unique up to a rotation on $\SSS^2$.
\end{remark}
\begin{lemma}
\lab{lemma:basicpropertiesofJforcanonicalell=1basis}
Consider $(S,g^S)$ a sphere of area radius $r^S$ verifying the almost round condition \eqref{eq:Almostround-S}. Let $(\Phi, u)$ the unique, up to isometries of $\SSS^2$, uniformization pair given by Corollary \ref{proposition:effective-uniformisation}.
Let $J^S$ denote the basis of canonical $\ell=1$ modes on $S$ of Definition \ref{definition:ell=1mpdesonS}. Then, we have
\bea
\bsplit
\lap_S J^{(p,S)} &=-\frac{2}{(r^S)^2} J^{(p,S)} +\frac{2}{(r^S)^2}\big(1- e^{-2v}\big)J^{(p,S)}, \\
\int_S J^{(p,S)} J^{(q,S)} da_g &= \frac{4\pi}{3}(r^S)^2 \de_{pq}+ \int_S J^{(p,S)} J^{(q,S)} \big(1- e^{-2v} \big) da_{g^S},\\
\int_S J^{(p,S)} da_g &=0,
\end{split}
\eea
with $\lap^S $ the Laplace-Beltrami of the metric $g^S$ and with $v:= u\circ\Phi^{-1}$.
Moreover we have,
\bea
\lab{eq:PropertiesofJ^S}
\bsplit
\lap_S J^{(p,S)} &=\left(-\frac{2}{(r^S)^2} +O\left(\frac{\ep}{(r^S)^2}\right) \right)J^{(p,S)}, \\
\int_S J^{(p,S)}J^{(q,S)} da_g &= \frac{4\pi}{3}(r^S)^2\de_{pq}+ O(\ep(r^S)^2),
\end{split}
\eea
where $\ep>0$ is the smallness constant appearing in the almost round condition \eqref{eq:Almostround-S}.
\end{lemma}
\begin{proof}
Since $(\Phi, u)$ is the unique, up to isometries of $\SSS^2$, uniformization pair given by Corollary \ref{proposition:effective-uniformisation}, we have
\beaa
\Phi:\SSS^2\longrightarrow S, \qquad \Phi^\# (g^S) = (r^S)^2e^{2u}\ga_0,\qquad u\in\SS.
\eeaa
We rewrite this as follows
\bea
g^S=(r^S)^2e^{2v} g_0, \qquad g_0:=\big(\Phi^{-1}\big)^\# \ga_0, \qquad v:= u\circ\Phi^{-1}.
\eea
Since $\lap_0J^{\SSS^2}=-2J^{\SSS^2}$ and $g_0= \big(\Phi^{-1}\big)^\# \ga_0$, we have, in view of the definition of $J^S$, i.e. $J^S=J^{\SSS^2}\circ \Phi^{-1}=\big(\Phi^{-1}\big)^\# J^{\SSS^2}$,
\beaa
\lap_{g_0} J^S =-2 J^S.
\eeaa
On the other hand, since $g^S=(r^2)^2e^{2v} g_0$, we have in view of the conformal invariance of the Laplacian, $\lap_S = (r^S)^{-2}e^{-2v} \lap_{g_0} $. We deduce
\beaa
\lap_S J^S= (r^S)^{-2}e^{-2v} \lap_{g_0} J^S=-2(r^S)^{-2} e^{-2v} J^S=-\frac{2}{(r^S)^2} J^S+\frac{2}{(r^S)^2}\Big(- e^{-2v}+1 \Big)J^S.
\eeaa
Also,
\beaa
\int_S J^{(p,S)} J^{(q,S)} da_{g^S}&=&\int_S J^{(p,S)} J^{(q,S)} (r^S)^2 e^{2v} da_{g_0}\\
&=& (r^S)^2\int_S\big(\Phi^{-1}\big)^\#J^{(p,\SSS^2)}\big(\Phi^{-1}\big)^\#J^{(q,\SSS^2)}da_{\big(\Phi^{-1}\big)^\#\ga_0}\\
&&+(r^S)^2 \int_S J^{(p,S)} J^{(q,S)} \big( e^{2v}-1\big) da_{g_0}\\
&=& (r^S)^2\int_{\SSS^2}J^{(p,\SSS^2)}J^{(q,\SSS^2)}da_{\ga_0}+ \int_S J^{(p,S)} J^{(q,S)} \big( e^{2v}-1\big) e^{-2v} da_{g^S} \\
&=& \frac{4\pi}{3}(r^S)^2 \de_{pq}+ \int_SJ^{(p,S)} J^{(q,S)} \big(1- e^{-2v} \big) da_{g^S}.
\eeaa
Also, in view of the centeredness of $u$,
\beaa
\int_S J^S da_{g^S}&=&\int_S J^{\SSS^2}\circ \Phi^{-1} (r^S)^2e^{2v} da_{g_0}=(r^S)^2\int_{S}\big(\Phi^{-1}\big)^\#(e^{2u}J^{\SSS^2}) da_{\big(\Phi^{-1}\big)^\#\ga_0}\\
&=& (r^S)^2\int_{\SSS^2} e^{2u}x da_{\ga_0} =0
\eeaa
as stated.
The proof of \eqref{eq:PropertiesofJ^S} follows immediately in view of the fact that $v=u\circ\Phi$ satisfies $\| v\|_{L^\infty}\les \ep$, see \eqref{eq:Thm.-Eff.uniformalization1}.
\end{proof}
\begin{corollary}
Let $(S,g^S)$ verifying \eqref{eq:Almostround-S}. Let $(\Phi, u)$ the unique, up to isometries of $\SSS^2$, uniformization pair given by Corollary \ref{proposition:effective-uniformisation}.
Let $J^S$ denote the basis of canonical $\ell=1$ modes on $S$ of Definition \ref{definition:ell=1mpdesonS}. Then, for sufficiently small $\ep>0$, the following holds\footnote{Note that, a priori, one would expect the right-hand side of \eqref{eq:specialcancellationell=1modeofGausscurvature} to be $O(\ep)$. The fact that it is actually $O(\ep^2)$ is an application of Corollary \ref{proposition:effective-uniformisation} and \eqref{eq:PropertiesofJ^S}, see the proof below.}
\bea\lab{eq:specialcancellationell=1modeofGausscurvature}
\int_S\left(K^S-\frac{1}{(r^S)^2}\right)J^{(p,S)} &=& O(\ep^2),\qquad p=0,+,-,
\eea
where $K^S$ and $r^S$ denote respectively the Gauss curvature and the area radius of $S$.
\end{corollary}
\begin{proof}
Since $\Phi^*(g^S)=(r^S)^2e^{2u}\ga_0$, we have $K((r^S)^{-2}e^{-2u}\Phi^*(g^S))=1$ and hence
\beaa
1 &=& K((r^S)^{-2}e^{-2u}\Phi^*(g^S))=K((r^S)^{-2}\Phi^*(e^{-2v}g^S))\\
&=&K((r^S)^{-2}e^{-2v}g^S)\circ\Phi, \qquad\quad v:=u\circ\Phi^{-1},
\eeaa
which together with Lemma \ref{Le:transf-Gauss} yields
\beaa
1= K( (r^S)^{-2}e^{-2v} g^S)=(r^S)^2 e^{2v} \big(K^S+\lap_{S} v \big).
\eeaa
We infer
\bea
\lab{eq:K^S}
K^S -\frac{1}{(r^S)^2} &=& -\left(\Delta_S+\frac{2}{(r^S)^2}\right)v+\frac{f(-v)}{(r^S)^2}, \quad\qquad f(v):=e^{2v}-1-2v.
\eea
Multiplying by $J^{(p,S)}$, integrating on $S$ and using integration by parts, we deduce
\beaa
\int_S\left(K^S-\frac{1}{(r^S)^2}\right)J^{(p,S)} &=& -\int_S J^{(p,S)}\left(\Delta_S+\frac{2}{(r^S)^2}\right)v+\int_S\frac{f(-v)}{(r^S)^2}J^{(p,S)}\\
&=& -\int_Sv\left(\Delta_S+\frac{2}{(r^S)^2}\right)J^{(p,S)}+\int_S\frac{f(-v)}{(r^S)^2}J^{(p,S)}.
\eeaa
Together with \eqref{eq:PropertiesofJ^S}, the control $|v|\les\ep$ provided by \eqref{eq:Thm.-Eff.uniformalization1}, and the definition of $f$, we infer
\beaa
\left|\int_S\left(K^S-\frac{1}{(r^S)^2}\right)J^{(p,S)}\right| &\les& \int_S\frac{\ep}{(r^S)^2}|v|+\int_S\frac{|v|^2}{(r^S)^2}\les \ep^2
\eeaa
as desired.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Stability of uniformization for nearby spheres}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Consider two almost round spheres spheres $(S_1, g^{S_1} )$ and $(S_2, g^{S_2} )$, i.e. verifying \eqref{eq:Almostround-S}, and their respective uniformization pairs $(\Phi_1, u_1)$, $(\Phi_2, u_2)$, i.e.
\bea
\lab{eq:unifor-twometricsA}
\bsplit
& \Phi_1:\SSS^2 \longrightarrow S_1, \qquad g_1:=\Phi_1^\# (g^{S_1})= (r^{S_1})^2e^{2u_1} \ga_0,\\
& \Phi_2:\SSS^2 \longrightarrow S_2, \qquad g_2:=\Phi_2^\#( g^{S_2})= (r^{S_2})^2e^{2u_2} \ga_0,
\end{split}
\eea
and $u_1, u_2$ defined on $\SSS^2$ verifying the conclusions of Corollary \ref{proposition:effective-uniformisation}.
We assume in addition given a smooth diffeomorphism $\Psi:S_1\to S_2$ such that the metrics $g^{S_1} $ and $\Psi^\#( g^{S_2} )$ are close to each other in $S^1$ with respect to the coordinate chart provided by $\Phi_1$, i.e. for some $0<\de\le \ep$,
\bea
\lab{eq:unifor-twometricsB}
\left\| g^{S_1}- \Psi^\#( g^{S_2}) \right\|_{L^\infty(S_1)} + \frac{1}{(r^{S_1})} \left\| g^{S_1}- \Psi^\#( g^{S_2}) \right\|_{\hk_{4}(S^1)} \le (r^{S_1})^2 \de.
\eea
The goal of this section is to show the existence of a canonical diffeomorphism $\Psih:\SSS^2\to \SSS^2$ which relates the two uniformization maps.
More precisely we prove the following.
\begin{theorem}
\lab{Thm:unifor-twometrics}
Under the assumptions above, let $\Psih:\SSS^2\to \SSS^2$ be the unique smooth diffeomorphism such that $\Psi\circ \Phi_1=\Phi_2\circ \Psih$. Then, the following holds true.
\begin{enumerate}
\item The diffeomorphism $\Psih$ is smooth and there exists $O\in O(3)$ such that
\bea
\lab{eq:unifor-twometrics1}
\| \Psih-O\|_{L^\infty(\SSS^2)}+ \| \Psih-O\|_{H^1(\SSS^2)}&\les &\de.
\eea
\item The conformal factors $u_1, u_2$ verify
\bea
\lab{eq:unifor-twometrics2}
\big \|u_1- \Psih^\# u_2\big \|_{L^\infty(\SSS^2)} \les \de.
\eea
\end{enumerate}
\end{theorem}
\begin{remark}
Let us note the following concerning assumption \eqref{eq:unifor-twometricsB}.
\begin{itemize}
\item It is clearly not sharp in terms of regularity. Sharper results could be obtained by working in H\"older spaces. On the other hand, in view of our applications, $\hk_s(S)$ are the natural spaces.
\item It is coordinate dependent. Though it is sufficient for our applications, it would be nice find a coordinate independent condition sufficient to recover the conclusions of Theorem \ref{Thm:unifor-twometrics}.
\end{itemize}
\end{remark}
The proof of Theorem \ref{Thm:unifor-twometrics} is postponed to section \ref{sec:proofofThm:unifor-twometrics}.
It will rely in particular on the control of almost isometries of $\SSS^2$ discussed in the next section.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Almost isometries of $\SSS^2$}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{proposition}
\lab{prop:diffeomorphismTh}
Let $\Th :\SSS^2\to \SSS^2$ be a $C^2$ diffeomorphism such that,
\beaa
\| \Th^\#\ga_0 -\ga_0\|_{L^\infty(\SSS^2)} \le \de.
\eeaa
Then, there exists $O\in O(3)$ such that
\bea
\lab{eq:unifor-twometrics6}
\|\Theta - O\|_{L^\infty(\SSS^2)}+\|\Theta - O\|_{H^1(\SSS^2)} &\les& \de.
\eea
\end{proposition}
\begin{proof}
We first extend the map $\Th$ to a map $\widetilde{\Th} :\RRR^3\setminus \{0\} \to \RRR^3\setminus \{0\} $ such that for every $ x=r\om, \om\in \SSS^2$, $r>0$,
\bea\lab{eq:definitionofwidetildeThetafortheproofofdiffeomorphismTh}
\widetilde{\Th} (r\om)=r \Th(\om).
\eea
It is easy to check that, denoting by $e$ the euclidian metric,
\bea\lab{eq:themapwidetildeThetaisanalmostisometryoftheEuclideanspace}
\| \widetilde{\Th} ^\# e - e\|_{L^\infty(D) } \les \de, \qquad D:=\left\{ x\in \RRR^3\,\,\Big/\,\, \frac 1 2 \le |x|\le 2\right\}.
\eea
We deduce
\beaa
\|\nab \widetilde{\Th} \big(\nab \widetilde{\Th} \big)^t -I \|_{L^\infty(D)} \les \de,
\eeaa
i.e.
\beaa
\|A A^t -I \|_{L^\infty(D)} \les \de, \qquad A:= \nab \widetilde{\Th},
\eeaa
where $A$ is a function on $D$ taking values in 3 by 3 matrices.
By the polar decomposition of $A$, there exists, at every point $p\in D$, a unique orthogonal matrix $O\in O(3)$ such that
\beaa
A= (AA^t)^{\frac{1}{2}} O.
\eeaa
We deduce, at every $p\in D$,
\beaa
|A-O |&=& \big|(AA^t)^{\frac{1}{2}} -I\big|\les \de.
\eeaa
Therefore,
\beaa
\sup_{D}\min_{ O\in O(3) } \Big| \nab \widetilde{\Th} -O\Big|&\les \de
\eeaa
i.e.,
\beaa
\| \textrm{dist}(\nab \widetilde{\Th}, O(3))\|_{L^\infty(D)}\les \de.
\eeaa
In particular, the determinant of $\nab \widetilde{\Th}$ cannot vanish on $D$, and hence, since $D$ is connected, is either strictly positive or strictly negative everywhere on $D$. From now on, we assume for simplicity that the determinant of $\nab \widetilde{\Th}$ is strictly positive everywhere on $D$, the other case being similar. We infer
\bea\lab{eq:disctanceofnabwidetildeThtoOof3issmall}
\| \textrm{dist}(\nab \widetilde{\Th}, SO(3))\|_{L^\infty(D)}\les \de.
\eea
We rely on the following result\footnote{The result in \cite{FJM} actually covers only $p=2$. The case $1<p<+\infty$ is in section 2.4 of S. Conti and B. Schweizer \cite{CoSc}.} of G. Friesecke, R. James and S. M\"uller in \cite{FJM} pointed to us by Camillo De Lellis.
\begin{theorem}
\lab{Thm:FJM1}
Let $ D\subset \RRR^n $ a bounded Lipschitz domain, $n\ge 2$ and $1<p<+\infty$. There exists a constant $C>0$ depending only on $D$ and $p$ with the property:
For every $\psi \in W^{1,p} (D, \RRR^n)$ there exists a rotation $O_0\in SO(n) $ such that,
\bea
\|\nab \psi - O_0\|_{L^p(D)} &\les& C\| \textrm{dist}(\nab \psi, SO(n))\|_{L^p(D)}.
\eea
\end{theorem}
\begin{remark}\lab{rem:FirtzJohnBMOestimate}
The estimate of Theorem \ref{Thm:FJM1} fails in $L^\infty$ but still holds in BMO from a well known estimate of F. John \cite{John}.
\end{remark}
We infer from \eqref{eq:disctanceofnabwidetildeThtoOof3issmall} and Theorem \ref{Thm:FJM1} the existence of a rotation
$O_0\in SO(3) $ such that
\beaa
\|\nab \widetilde{\Th} - O_0\|_{L^3(D)} + \|\nab \widetilde{\Th} - O_0\|_{L^2(D)} &\les& \de.
\eeaa
Restricting to the sphere, we infer,
\beaa
\|\nab \Th- O_0\|_{L^3(\SSS^2)} +\|\nab \Th- O_0\|_{L^2(\SSS^2)} &\les& \de.
\eeaa
Together with Poincar\'e inequality and the Sobolev embedding, this yields
\beaa
\|\Theta - O_0\|_{L^\infty(\SSS^2)}+\|\nab (\Theta - O_0)\|_{L^2(\SSS^2)} &\les& \de
\eeaa
as stated. This concludes the proof of Proposition \ref{prop:diffeomorphismTh}.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Proof of Theorem \ref{Thm:unifor-twometrics}}
\lab{sec:proofofThm:unifor-twometrics}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
We note that the following is an immediate consequence of
\eqref{eq:unifor-twometricsB}
\bea
\lab{comp:rS1-to-rS2}
\left|\frac{r^{S_1}}{ r^{S_2} }-1\right| &\les& \de.
\eea
Denote
\beaa
g_1= \Phi^\#_1( g^{S_1}), \qquad g_2= \big(\Psi\circ\Phi_1\big)^\# ( g^{S_2}).
\eeaa
In view of our assumptions, we have
\bea
\lab{eq:unifor-twometrics3}
g_1= (r^{S_1})^2e^{2u_1} \ga_0, \qquad \int_{\SSS^2} e^{2u_1} x =0, \qquad \|u_1\|_{H^4(\SSS^2)} \le \ep,
\eea
and
\bea
\lab{eq:unifor-twometrics4}
\|g_1-g_2\|_{L^\infty(\SSS^2)} + \|g_1-g_2\|_{H^{4}(\SSS^2)} \les (r^{S_1})^2 \de.
\eea
We deduce,
\beaa
\left\| e^{2 u_1} \ga_0 - \frac{1}{(r^{S_1})^2 } g_2 \right\|_{L^\infty(\SSS^2)} + \left\| e^{2 u_1} \ga_0 - \frac{1}{(r^{S_1})^2 } g_2 \right\|_{H^{4}(\SSS^2)} \les \de,
\eeaa
or, in view of the control of $u_1$ as well as \eqref{comp:rS1-to-rS2}
\bea
\| \ga_0 - e^{-2 u_1} \widetilde{g}_2 \|_{L^\infty(\SSS^2)}+ \| \ga_0 - e^{-2 u_1} \widetilde{g}_2 \|_{H^{4}(\SSS^2)} \les \de, \qquad \widetilde{g}_2:= \frac{1}{(r^{S_2})^2 } g_2.
\eea
Consequently
\beaa
\| K\big( e^{-2u_1} \widetilde{g}_2\big)-1\|_{H^2(\SSS^2)} \les \de.
\eeaa
We can thus apply Corollary \ref{proposition:effective-uniformisation} to the metric $e^{-2u_1} \widetilde{g}_2 $ and deduce that there exists, up to an isometry of $\SSS^2$, a unique centered conformal $v$
and a unique smooth diffeomorphism $\Th:\SSS^2\to \SSS^2$ such that
\bea
\lab{eq:unifor-twometrics5}
\Th^\#\Big( e^{-2u_1} \widetilde{g}_2 \Big)=(r_{1, 2})^2e^{2v} \ga_0, \qquad \|v \|_{H^{4}(\SSS^2)} \les \de.
\eea
Hence,
\beaa
e^{-2 u_1} \widetilde{g}_2 =(r_{1, 2})^2 e^{2 v\circ \Th^{-1} } (\Th^{-1}) ^\# \ga_0.
\eeaa
Since $ \| \ga_0 - e^{-2 u_1} \widetilde{g}_2 \|_{H^{4}} \les \de$ it follows that,
\beaa
\| \ga_0 - (r_{1, 2})^2e^{2 v\circ \Th^{-1} } (\Th^{-1}) ^\# \ga_0\|_{H^{4}(\SSS^2)} \les \de, \qquad |r_{1, 2}-1|\les\de,
\eeaa
where we also used $\| v\|_{L^\infty(\SSS^2)} \les \de$ for the second inequality.
Since $\| v\|_{H^{4}(\SSS^2)} \les \de$ and $\Th$ is smooth we deduce,
\bea
\lab{eq:unifor-twometrics5'}
\| \ga_0 - \Th ^\# \ga_0\|_{H^{4}(\SSS^2)} \les \de.
\eea
We now appeal to Proposition \ref{prop:diffeomorphismTh} which yields the existence of $O\in O(3)$ such that
\bea\lab{eq:consequenceof:unifor-twometrics5'}
\|\Theta - O\|_{L^\infty(\SSS^2)}+\|\Theta - O\|_{H^1(\SSS^2)} &\les& \de.
\eea
We rewrite \eqref{eq:unifor-twometrics5} in the form
\beaa
\Th^\# \widetilde{g}_2= (r_{1, 2})^2e^{2v+2u_1\circ \Th} \ga_0= e^{2\tilde{u}_2 }\ga_0, \qquad \tilde{u}_2:= v+ u_1\circ \Th+\log(r_{1, 2}).
\eeaa
Note that $\tilde{u}_2$ is a priori not centered. On the other hand, we have
\bea\lab{eq:Linftyboundfortildeu2onSSS2}
\|\tilde{u}_2\|_{L^\infty} &\leq& \|v\|_{L^\infty}+\|u_1\|_{L^\infty}+|\log(r_{1, 2})|\les\ep.
\eea
Also, since $|r_{1, 2}-1|\les\de$, $\| v\|_{H^4}\les \de$ and $\|\Th - O\|_{L^\infty}\les \de$,
\beaa
\int_{\SSS^2} e^{\tilde{u}_2} x da_{\ga_0} &=& \int_{\SSS^2} e^{2v+2u_1\circ \Th+2\log(r_{1, 2})} x da_{\ga_0}\\
&=& \int_{\SSS^2} e^{2u_1\circ \Th} x da_{\ga_0} + \int_{\SSS^2} e^{2u_1\circ \Th} \big( (r_{1, 2})^2e^{2v}-1\big) x da_{\ga_0} \\
&= &\int_{\SSS^2} e^{2u_1} (x\circ \Th^{-1} ) \frac{1}{|\det d \Th|} d a_{\ga_0}+O(\de)\\
&= &\int_{\SSS^2} e^{2u_1} x\circ O^{-1} d a_{\ga_0} +O(\de).
\eeaa
We deduce, remembering that $O$ is a matrix in $O(3)$ and $x\in \mathbb{R}^3$,
\beaa
\int_{\SSS^2} e^{\tilde{u}_2} x da_{\ga_0} &= &\int_{\SSS^2} e^{2u_1} O^{-1} x d a_{\ga_0}+O(\de)\\
&=& O^{-1}\left(\int_{\SSS^2} e^{2u_1} x d a_{\ga_0}\right)+O(\de).
\eeaa
Since $u_1$ is centered, we infer
\bea
\lab{eq:unifor-twometrics7}
\int_{\SSS^2} e^{2\tilde{u}_2} x &= & O(\de).
\eea
In view of \eqref{eq:Linftyboundfortildeu2onSSS2} and \eqref{eq:unifor-twometrics7}, Lemma \ref{lemma:improved-renormalization} applies so that there exists $\psi\in \M$ such that\footnote{The statement of Lemma \ref{lemma:improved-renormalization} actually yields $|\psi-I|\les\de$. Here we use the fact that $\SS$ is invariant by $O(3)$, see Remark \ref{rem:ifuiscenteredthensoisucricO}, so that we may indeed assume, by composing with $O^{-1}\in O(3)$, that $|\psi-O^{-1}|\les \de$.}
\bea
u_2 ':= (\widetilde{u}_2)_\psi = \tilde{u}_2\circ \psi+\frac 1 2 \log|\det \psi| \in \SS, \qquad \big\|\psi-O^{-1} \big\|_{L^\infty(\SSS^2)}\les \de.
\eea
We deduce
\beaa
\psi^\# \Th^\# \widetilde{g}_2 = e^{\tilde{u}_2 \circ \psi} \psi^\# \ga_0= e^{2\tilde{u}_2 \circ \psi} | \det d\psi |\ga_0 =e^{2\tilde{u}_2 \circ \psi +\log|\det d\psi| } \ga_0= e^{2 u'_2} \ga_0.
\eeaa
Also, in view of the control on $\Th$ and $\psi$, we have
\beaa
\| \Th \circ \psi -I\|_{L^\infty(\SSS^2)}+ \| \Th \circ \psi -I\|_{H^1(\SSS^2)} \les \de.
\eeaa
Therefore we have found a diffeomorphism
\bea
\Psih' :=(\Th\circ \psi)^{-1}, \qquad \Psih':\SSS^2\longrightarrow\SSS^2,
\eea
with the properties
\bea
(\Psih'^{-1})^\# \widetilde{g_2} &=& e^{2 u'_2} \ga_0, \qquad u_2' \in \SS,
\eea
and
\bea\lab{eq:almosttherightestimateetoconcludestabilityunifromizationproofupmissingrotation:0}
\| \Psih' -I\|_{L^\infty(\SSS^2)}+ \| \Psih' -I\|_{H^1(\SSS^2)} \les \de.
\eea
Also, recall that
\beaa
u_2' &=& \tilde{u}_2\circ \psi+\frac 1 2 \log|\det \psi| = \big( v+ u_1\circ \Th +2\log(r_{1, 2})\big)\circ\psi +\frac 1 2 \log|\det \psi|\\
&=& u_1\circ \Psih'^{-1} + v_\psi +2\log(r_{1, 2}).
\eeaa
We infer
\beaa
\Psih'^\# u'_2 - u_1 &=& \Psih'^\#\big( v_\psi\big) +2\log(r_{1, 2}),
\eeaa
and thus, since $|r_{1, 2}-1|\les\de$, $\| v\|_{H^{4}} \le \de$, $ \| \Psih' -I\|_{L^\infty}\les\de$ and $\|\psi-O^{-1}\|_{L^\infty} \les \de$, we deduce,
\bea\lab{eq:almosttherightestimateetoconcludestabilityunifromizationproofupmissingrotation:1}
\|u_1 - \Psih'^\#(u'_2) \|_{L^\infty(\SSS^2)} \les \de.
\eea
Next, let
\beaa
\Phi_2':=\Psi\circ \Phi_1\circ \Psih'^{-1}.
\eeaa
Then, by the definition of $g_2$ and $\widetilde{g}_2$,
\beaa
\widetilde{g}_2=\frac{1}{(r^{S_2})^2 } g_2= \frac{1}{(r^{S_2})^2 }(\Psi\circ \Phi_1)^\# g^{S_2}, \eeaa
and the construction of $\Psih'$ we have
\beaa
( \Phi_2')^\# g^{S_2}&=& (\Psih'^{-1})^\# \big( \Psi\circ \Phi_1\big)^\# g^{S_2} = (r^{S_2})^2 (\Psih'^{-1})^\# \widetilde{g}_2= (r^{S_2})^2 e^{2u_2'} \ga_0
\eeaa
and by the definition of $\Phi_2, u_2$ we also have,
\beaa
( \Phi_2)^\# g^{S_2}= (r^{S_2})^2 e^{2u_2} \ga_0.
\eeaa
Therefore, since both $u_2$ and $u_2'$ are centered, from the uniqueness statement
of Corollary \ref{proposition:effective-uniformisation}, there existe $O\in O(3)$ such that,
\bea\lab{eq:almosttherightestimateetoconcludestabilityunifromizationproofupmissingrotation:2}
u_2'=u_2\circ O, \qquad \Psi\circ \Phi_1=\Phi_2\circ O\circ \Psih',
\eea
where the last identity corresponds to $\Phi_2'=\Phi_2\circ O$. We now define
\beaa
\Psih &:=& O\circ \Psih'.
\eeaa
Then, together with \eqref{eq:almosttherightestimateetoconcludestabilityunifromizationproofupmissingrotation:0}, \eqref{eq:almosttherightestimateetoconcludestabilityunifromizationproofupmissingrotation:1} and \eqref{eq:almosttherightestimateetoconcludestabilityunifromizationproofupmissingrotation:2}, we infer
\beaa
\Psi\circ \Phi_1=\Phi_2\circ \Psih, \qquad \| \Psih -O\|_{L^\infty}+ \| \Psih -O\|_{H^1} \les \de, \qquad \|u_1 - \Psih^\#(u_2) \|_{L^\infty(\SSS^2)} \les \de,
\eeaa
as stated. This concludes the proof of Theorem \ref{Thm:unifor-twometrics}.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Higher regularity estimates for Theorem \ref{Thm:unifor-twometrics}}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Though \eqref{eq:unifor-twometrics1} \eqref{eq:unifor-twometrics2} suffice for our main application to GCM spheres, we provide here, for the sake of completeness, a discussion of corresponding higher regularity estimates. This relies on a recent personal communication of Camillo De Lellis and Stefan M\"uller. More precisely they can prove the following.
\begin{theorem}[C. De Lellis, S. M\"uller \cite{DeLellis}]
\lab{th:DeLellisMullerhigherregularity}
Let $n\geq 2$. Consider the balls $B_1\subset B_2 \subset \RRR^n$ and $e$ the euclidian metric on $\RRR^n$.
Consider a map $u\in H^1( B_2, \RRR^n)$ verifying $\det \nab u>0$ and,
\bea
\big \| u^\# e - e\big \|_{L^\infty(B_1)} \le \frac 1 2.
\eea
Then, there exists a rotation $O\in SO(n)$ such that, relative to the H\"older norms $C^\a$, $0<\a<1$,
\beaa
\big\| \nab u- O\big \|_{C^\a( B_1)} \les \big \| u^\# e - e\big \|_{C^\a(B_2)}.
\eeaa
\end{theorem}
The goal of this section is to prove the following proposition that extends the estimates \eqref{eq:unifor-twometrics1}, \eqref{eq:unifor-twometrics2} of Theorem \ref{Thm:unifor-twometrics} to corresponding higher regularity estimates.
\begin{proposition}
\lab{Thm:unifor-twometrics:higherregularity}
Assume, in addition to the assumptions of Theorem \ref{Thm:unifor-twometrics}, that
\bea
\lab{eq:unifor-twometricsB:higherregularityassumption}
\left\| g^{S_1}- \Psi^\#( g^{S_2}) \right\|_{\hk_{4+s}(S^1)} \le \de
\eea
for some $s\geq 0$. Then, the following higher regularity analogs of \eqref{eq:unifor-twometrics1}, \eqref{eq:unifor-twometrics2} hold true.
\begin{enumerate}
\item The diffeomorphism $\Psih$ is smooth and there exists $O\in O(3)$ such that
\bea
\lab{eq:unifor-twometrics1:higherregularity}
\| \Psih-O\|_{H^{5+s}(\SSS^2)}&\les &\de.
\eea
\item The conformal factors $u_1, u_2$ verify
\bea
\lab{eq:unifor-twometrics2:higherregularity}
\big \|u_1- \Psih^\# u_2\big \|_{H^{4+s}(\SSS^2)} \les \de.
\eea
\end{enumerate}
\end{proposition}
The only part of the proof of Theorem \ref{Thm:unifor-twometrics} which does not immediately extend to higher regularity is when applying Proposition \ref{prop:diffeomorphismTh} to infer \eqref{eq:consequenceof:unifor-twometrics5'} from \eqref{eq:unifor-twometrics5'}. Thus, to complete the proof of Proposition \ref{Thm:unifor-twometrics:higherregularity}, it suffices to prove the following extension of Proposition \ref{prop:diffeomorphismTh} to higher regularity.
\begin{proposition}
\lab{prop:diffeomorphismTh:higherregularity}
Let $\Th :\SSS^2\to \SSS^2$ be a $C^2$ diffeomorphism such that,
\bea\lab{eq:themapThetaisanalmostisometryofthestandardspherehigherreg}
\| \Th^\#\ga_0 -\ga_0\|_{H^{s+2}(\SSS^2)} \le \de
\eea
for some $s\geq 0$. Then, there exists $O\in O(3)$ such that
\bea
\lab{eq:unifor-twometrics6:higherregularityesitmate}
\|\Theta - O\|_{H^{s+3}(\SSS^2)} &\les& \de.
\eea
\end{proposition}
\begin{proof}
Recall that the proof of Proposition \ref{prop:diffeomorphismTh} proceeds by extending the map $\Theta$ to a map $\widetilde{\Th} :\RRR^3\setminus \{0\} \to \RRR^3\setminus \{0\} $ such that for every $ x=r\om, \om\in \SSS^2$, $r>0$,
\beaa
\widetilde{\Th} (r\om)=r \Th(\om),
\eeaa
see \eqref{eq:definitionofwidetildeThetafortheproofofdiffeomorphismTh}. Under the assumption \eqref{eq:themapThetaisanalmostisometryofthestandardspherehigherreg}, we obtain the following higher regularity analog of the estimate \eqref{eq:themapwidetildeThetaisanalmostisometryoftheEuclideanspace}
\bea\lab{eq:esaimteofwidetildeThetainviewofthelinearizationofitseequation}
\| \widetilde{\Th}^\# e - e\|_{H^{s+2}(D) } \les \de, \qquad D=\left\{ x\in \RRR^3\,\,\Big/\,\, \frac 1 2 \le |x|\le 2\right\}.
\eea
In view of Theorem \ref{th:DeLellisMullerhigherregularity}, covering $\SSS^2$ by finitely many euclidean balls\footnote{More precisely, $B_1$ and $B_2$ of Theorem \ref{th:DeLellisMullerhigherregularity} are chosen to be euclidean balls centered on points of $\SSS^2$ of radius given respectively by $1/4$ and $1/2$ so that both $B_1$ and $B_2$ are included in $D$ and the union of the $B_1$'s includes $D_1$.}, and using that $H^{2}(D)$ embeds in $C^{1/2}(D)$ by Sobolev, we infer the existence of $O\in O(3)$ such that
\beaa
\|\nabla\widetilde{\Th} - O\|_{C^{\frac{1}{2}}(D_1)} \les \de, \qquad D_1:=\left\{ x\in \RRR^3\,\,\Big/\,\, \frac{3}{4} \le |x|\le \frac{5}{4}\right\}.
\eeaa
We deduce the following decomposition for $\widetilde{\Th}$
\bea\lab{eq:decompostionofwidetildeThetainviewofthelinearizationofitseequation}
\widetilde{\Th}=(I+\widetilde{\Th}')O, \qquad \|\widetilde{\Th}'\|_{L^\infty(D_1)}+\|\nabla\widetilde{\Th}'\|_{L^\infty(D_1)}\les\de.
\eea
Next, we rewrite \eqref{eq:esaimteofwidetildeThetainviewofthelinearizationofitseequation} in matrix form
\bea\lab{eq:linearizationofitseequationforwidetildeTheta:notlinearizedyet}
\nabla\widetilde{\Th}(\nabla\widetilde{\Th})^t =I+ E, \qquad \|E\|_{H^{s+2}(D) } \les \de.
\eea
Linearizing this equation based on the decomposition of $\widetilde{\Th}$ provided by \eqref{eq:decompostionofwidetildeThetainviewofthelinearizationofitseequation}, we infer on $D_1$
\bea\lab{eq:linearizationofitseequationforwidetildeTheta}
S(\widetilde{\Th}') &=& E-\nabla\widetilde{\Th}'(\nabla\widetilde{\Th}')^t, \qquad S(\widetilde{\Th}'):=\nabla\widetilde{\Th}'+(\nabla\widetilde{\Th}')^t.
\eea
Note that $S$ satisfies the well-known Korn inequality, see for example \cite{Cia},
\bea\lab{eq:KorninequalityonD1}
\|f\|_{H^1(D_1)}\leq C_{D_1}\Big(\|f\|_{L^2(D_1)}+\|S(f)\|_{L^2(D_1)}\Big), \qquad f:D_1\to \RRR^3.
\eea
where $S(f)$ denotes the symmetric part of the gradient of $f$.
Differentiating \eqref{eq:linearizationofitseequationforwidetildeTheta}, relying on the estimate \eqref{eq:linearizationofitseequationforwidetildeTheta:notlinearizedyet} for $E$ and the a priori bound \eqref{eq:decompostionofwidetildeThetainviewofthelinearizationofitseequation} for $\widetilde{\Th}'$ on $D_1$, and making use of Korn inequality \eqref{eq:KorninequalityonD1}, we infer
\beaa
\|\widetilde{\Th}'\|_{H^{s+3}(D_1)}\les\de
\eeaa
and hence
\beaa
\|\widetilde{\Th}-O\|_{H^{s+3}(D_1)}\les\de.
\eeaa
Restricting to the sphere, we obtain
\beaa
\|\Theta - O\|_{H^{s+3}(\SSS^2)} &\les& \de
\eeaa
as desired.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Calibration of uniformization maps between spheres}
\lab{subsection:calibration}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
In order to eliminate the arbitrariness with respect to isometries of $\SSS^2$ in Theorem \ref{Thm:unifor-twometrics}, see \eqref{eq:unifor-twometrics1}, we calibrate
the effective uniformization maps\footnote{Given by Corollary \ref{proposition:effective-uniformisation}.} $\Phi_1:\SSS^2 \to S_1 $, $\Phi_2:\SSS_2 \to S_2 $, for given diffeomorphism $\Psi :S_1 \to S_2 $, as follows.
\begin{definition}
\lab{def:calibration}
On $\SSS^2$ we fix\footnote{In particular, one can choose $N=(0,0,1)$ and $v=(1,0,0)$.} a point $N$ and a unit vector $v$ in the tangent space $T_N\SSS^2$. Given $\Psi:S_1\to S_2$, we say that the effective uniformization maps $\Phi_1:\SSS^2 \to S_1 $, $\Phi_2:\SSS^2 \to S_2 $ are calibrated relative to $\Psi$ if the map $\Psih:=(\Phi_2)^{-1}\circ \Psi\circ \Phi_1:\SSS^2\to \SSS^2$ is such that
\begin{enumerate}
\item The map $\Psih$ fixes the point $N$, i.e. $\Psih(N)=N$.
\item The tangent map $\Psih_\#$ fixes the direction of $v$, i.e. $\Psih_\#(v)=a_{1,2}v$ where $a_{1,2}>0$.
\item The tangent map $\Psih_\#$ preserves the orientation of $T_N\SSS^2$.
\end{enumerate}
\end{definition}
\begin{lemma}\lab{lemma:existenceofcalibrationforagivenmapPhi1}
Given $\Psi:S_1\to S_2$ and a fixed effective uniformization map $\Phi_1:\SSS^2 \to S_1 $ for $S_1$. Then, there exists a unique, effective, uniformization for $S_2$ calibrated with that of $S_1$ relative to $\Psi$.
\end{lemma}
\begin{proof}
Let $\Phi_2'$ be an effective uniformization map for $S_2$ and let $\Psih':=(\Phi'_2)^{-1}\circ \Psi\circ \Phi_1$ so that $\Psih':\SSS^2\to \SSS^2$. Recall that for any $O\in O(3)$, $\Phi_2$ given by $\Phi_2=\Phi_2'\circ O^{-1}$ is also an effective uniformization map for $S_2$. We have in this case $\Psih=O\Psih'$, and $\Phi_1$ and $\Phi_2$ are calibrated relative to $\Psi$ if and only if $O$ satisfies the following.
\begin{enumerate}
\item $O(\Psih'(N))=N$.
\item $\Psih'_\#(Ov)=av$ where $a>0$.
\item $\Psih'_\#\circ O$ preserves the orientation of $T_N\SSS^2$.
\end{enumerate}
Since $\Psih'$ is $O(\de)$ close to an element of $O(3)$, see \eqref{eq:unifor-twometrics1}, there exists a unique $O\in O(3)$ satisfying the conditions above.
\end{proof}
We now state the following corollary to Theorem \ref{Thm:unifor-twometrics}.
\begin{corollary}
\lab{Cor:unifor-twometrics-calibrated}
In addition to the assumptions of Theorem \ref{Thm:unifor-twometrics}, assume that the
maps $\Phi_1, \Phi_2$ are calibrated relative to $\Psi$ according to Definition \ref{def:calibration}. Then $\Psih$ verifies
\bea
\lab{eq:unifor-twometrics1-calibr}
\| \Psih-I\|_{L^\infty(\SSS^2)}+ \| \Psih-I\|_{H^1(\SSS^2)}&\les &\de.
\eea
The conformal factors $u_1, u_2$ verify
\bea
\lab{eq:unifor-twometrics2-calibr}
\big \|u_1- \Psih^\# u_2\big \|_{L^\infty(\SSS^2)} \les \de.
\eea
\end{corollary}
\begin{proof}
In view of Theorem \ref{Thm:unifor-twometrics}, there exists $O\in O(3)$ such that $\Psih$ satisfies
\bea
\lab{eq:unifor-twometrics1:bisfortheproof}
\| \Psih-O\|_{L^\infty(\SSS^2)}+ \| \Psih-O\|_{H^1(\SSS^2)}&\les &\de
\eea
and \eqref{eq:unifor-twometrics2-calibr} holds.
It remains to check \eqref{eq:unifor-twometrics1-calibr}. In view of \eqref{eq:unifor-twometrics1:bisfortheproof} and the fact that $\Phi_1, \Phi_2$ are calibrated relative to $\Psi$ according to Definition \ref{def:calibration}, we infer that
\begin{enumerate}
\item $O(N)=N+O(\de)$,
\item $O(v)=av+O(\de)$ with $a>0$.
\item $O$ preserves the orientation of $T_N\SSS^2$.
\end{enumerate}
These conditions imply immediately that $|O-I|\les\de$ which together with \eqref{eq:unifor-twometrics1:bisfortheproof} yields \eqref{eq:unifor-twometrics1-calibr} as desired.
\end{proof}
\begin{lemma}[Transitivity of calibrations]
\lab{Le:Transitivitycalibrations}
Let $\Phi_1:\SSS^2 \to S_1 $, $\Phi_2:\SSS^2 \to S_2 $ and $\Phi_3:\SSS^2 \to S_3 $ three effective uniformization maps. Let $\Psi_{12}: S_1\to S_2$ and $\Psi_{13}:S_1\to S_3$ satisfying \eqref{eq:unifor-twometricsB}
and assume that $\Phi_1$, $\Phi_2$ are calibrated relative to $\Psi_{12}$, while $\Phi_1$, $\Phi_3$ are calibrated relative to $\Psi_{13}$. Then, $\Phi_2$, $\Phi_3$ are calibrated relative to $\Psi_{23}:=\Psi_{13}\circ\Psi_{12}^{-1}$.
\end{lemma}
\begin{proof}
Since $\Phi_1$, $\Phi_2$ are calibrated relative to $\Psi_{12}$, and $\Phi_1$, $\Phi_3$ are calibrated relative to $\Psi_{13}$, $\Psih_{12}:=(\Phi_2)^{-1}\circ \Psi_{12}\circ \Phi_1$ and $\Psih_{13}:=(\Phi_3)^{-1}\circ \Psi_{13}\circ \Phi_1$ satisfy the three properties of Definition \ref{def:calibration}. Then, introducing
\beaa
\Psi_{23}=\Psi_{13}\circ\Psi_{12}^{-1}, \qquad \Psih_{23}:=(\Phi_3)^{-1}\circ \Psi_{23}\circ \Phi_2,
\eeaa
we have $\Psih_{23}=\Psih_{13}\circ\Psih_{12}^{-1}$ so that $\Psih_{23}$ also satisfy the three properties of Definition \ref{def:calibration}. Hence, $\Phi_2$, $\Phi_3$ are calibrated relative to $\Psi_{23}$ as desired.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Comparison of $\ell=1$ modes between two spheres}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Consider, as in Theorem \ref{Thm:unifor-twometrics}, two almost round spheres $(S_1, g^{S_1} )$ and $(S_2, g^{S_2} )$ and
a smooth map $\Psi:S_1\to S_2$ such that, as in \eqref{eq:unifor-twometricsB}, that the metrics $g^{S_1} $ and $\Psi^\#( g^{S_2} )$ are close to each other in $S^1$. Assume that $(\Phi_1, u_1)$ , $(\Phi_2, u_2)$ are effective uniformization
maps of $S_1$ and $S_2$, calibrated as in definition \ref{def:calibration}.
We define
\beaa
J^i=J^{S_i}= J^{\SSS^2}\circ \Phi_i^{-1}, \qquad i=1,2,
\eeaa
to be the $\ell=1$ canonical modes of
$S_1, S_2 $ according to Definition \ref{definition:ell=1mpdesonS}.
We want to compare $J^1$ with $ J^2\circ \Psi^{-1} $. We prove the following result.
\begin{proposition}
\lab{Prop:comparison.canJ}
Under the assumptions of Corollary \ref{Cor:unifor-twometrics-calibrated} we have
\bea
\Big| J^1- J^2\circ \Psi \Big| &\les &\de.
\eea
\end{proposition}
\begin{proof}
Indeed, using that $\Psi\circ\Phi_1=\Phi_2\circ\Psih$,
\beaa
J^2\circ \Psi= J^{\SSS^2}\circ \Phi_2^{-1} \circ \Psi =J^{\SSS^2}\circ\Psih\circ \Phi_1^{-1}.
\eeaa
Hence,
\beaa
J^1- J^2\circ \Psi= J^{\SSS^2}\circ \Phi_1^{-1} -J^{\SSS^2}\circ\Psih\circ \Phi_1^{-1}.
\eeaa
This implies, together with \eqref{eq:unifor-twometrics1-calibr},
\beaa
\Big| J^1- J^2\circ \Psi\Big| \les \sup_{\SSS^2} \Big| I -\Psih\Big| \les \de
\eeaa
as stated.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Review of the geometric set-up in \cite{KS-Kerr1}}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Background spacetime}
\lab{sec:backgroundspacetime}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
As in \cite{KS-Kerr1}, we consider given a vacuum spacetime $\RR$ with metric $\g$
endowed with an outgoing geodesic foliation by spheres $S(u, s)$ of fixed $(u, s) $, where $u$ is an outgoing optical function\footnote{That is $\g^{\a\b}\pr_\a u\pr_\b u =0$.} with $L=-\g^{\a\b} \pr_\b u \pr_\a $ its null geodesic generator and $s$ chosen such that
\beaa
L(s) =\frac{1}{\vsi}, \qquad L(\vsi)=0.
\eeaa
Let $e_4=\vsi L $ and $e_3$ the unique null vectorfield orthogonal to $S(u,s)$ and such that $\g(e_3, e_4)=-2$. We then let $(e_1, e_2)$ an orthogonal basis of the tangent space of $S(u, s)$.
The corresponding connection coefficients relative to the null frame $(e_3, e_4, e_1, e_3)$ are denoted by $\chi, \chib, \xi, \xib, \om, \omb, \eta, \etab, \ze$ and the null components of the curvature tensor by $\a, \aa, \b, \bb, \rho, \rhod$. For the convenience of the reader we recall their definition below.
\beaa
\begin{split}
\chib_{ab}&=g(\D_ae_3, e_b),\qquad \,\,\chi_{ab}=g(\D_ae_4, e_b),\\
\xib_a&=\frac 1 2 \g(\D_3 e_3 , e_a),\qquad \xi_a=\frac 1 2 \ g(\D_4 e_4, e_a),\\
\omb&=\frac 1 4 g(\D_3e_3 , e_4),\qquad\,\, \om=\frac 1 4 g(\D_4 e_4, e_3),\qquad \\
\etab_a&=\frac 1 2 (\D_4 e_3, e_a),\qquad \quad \eta_a=\frac 1 2 g(\D_3 e_4, e_a),\qquad\\
\ze_a&=\frac 1 2 g(\D_{e_a}e_4, e_3),
\end{split}
\eeaa
and
\beaa
\a_{ab} &=&\R(e_a, e_4, e_b, e_4) , \qquad \aa_{ab} =\R(e_a, e_3, e_b, e_3), \\
\b_{a} &=&\frac 1 2 \R_(e_a, e_4, e_3, e_4), \qquad \bb_{a} =\frac 1 2 \R_(e_a, e_3, e_3, e_4),\\
\rho&=& \frac 1 4 \R_(e_3, e_4, e_3, e_4), \qquad \rhod =\dual R( e_3, e_4, e_3, e_4) .
\eeaa
The null second fundamental forms $\chi, \chib$ are further decomposed in their traces $\ka=\trch$ and $\kab=\trchb$, and traceless parts $\chih$ and $\chibh$. We recall, see Lemma 2.4 in \cite{KS-Kerr1}, that the geodesic nature of the foliation implies
\beaa
\om=\xi=0, \qquad \etab = -\ze, \qquad \vsi=\frac{2}{e_3(u)}.
\eeaa
The volume radius $ r=r(u, s)$ is defined such that the volume of $S$ is given by $4\pi r^2$. The Hawking mass $m=m(u, s)$ of $S=S(u, s) $ is given by the formula
\bea
\frac{2m}{r}=1+\frac{1}{16\pi}\int_{S_{}}\trch \trchb.
\eea
The Gauss curvature of $S$ is denoted by $K$. It verifies the Gauss equation
\bea\lab{eq:Gaussequation}
K=-\rho -\frac{1}{4} \trch \trchb +\frac{1}{2}\chih\c\chibh.
\eea
The mass aspect function $\mu$ is defined by
\bea
\label{def:massaspectfunctions.general}
\mu &:=& - \div \ze -\rho+\frac 1 2 \chih\c \chibh.
\eea
As in \cite{KS-Kerr1}, we define the renormalized quantities
\beaa
&&\widecheck{\trch} := \trch -\frac{2}{r}, \qquad
\widecheck{\trchb} := \trchb +\frac{2\Up}{r},\qquad
\widecheck{\omb} := \omb -\frac{m}{r^2},\\
&&\widecheck{K} := K -\frac{1}{r^2},\qquad\,\,\,\,\widecheck{\rho} := \rho +\frac{2m}{r^3},\qquad \quad\,\,\,
\widecheck{\mu} := \mu -\frac{2m}{r^3}, \\
&&\widecheck{\Omb} :=\Omb+\Up, \qquad\quad
\widecheck{\varsigma} := \varsigma-1,
\eeaa
where
\beaa
\Up :=1-\frac{2m}{r},
\eeaa
and the sets
\bea
\lab{definition:Ga_gGa_b}
\bsplit
\Ga_g &:= \Bigg\{\widecheck{\trch},\,\, \chih, \,\, \ze, \,\, \widecheck{\trchb},\,\, r\widecheck{\mu} ,\,\, r\widecheck{\rho}, \,\, r\dual\rho, \,\, r\b, \,\, r\a, \,\, r\widecheck{K}, \,\, r^{-1} \big(e_4(r)-1\big),\,\, r^{-1}e_4(m)\Bigg\},\\
\Ga_b &:= \Bigg\{\eta, \,\,\chibh, \,\, \ombc, \,\, \xib,\,\, r\bb, \,\, \aa, \,\, r^{-1}\widecheck{\Omb}, \,\,r^{-1}\widecheck{\varsigma}, \,\, r^{-1}(e_3(r)+\Up\big), \,\, r^{-1}e_3(m) \Bigg\}.
\end{split}
\eea
%%%%%%%%%%%%%%%%%%%%
\subsubsection{Adapted coordinates}
%%%%%%%%%%%%%%%%%%%%%
Recall, see \cite{KS-Kerr1}, that a coordinate system $(u, s, y^1, y^2)$ is said to be adapted to an outgoing geodesic foliation as above if $e_4(y^1)= e_4(y^2)=0$. In that case the spacetime metric can be written in the form
\bea
\lab{spacetimemetric-y-coordinates}
\g &=& - 2\vsi du ds + \vsi^2\Omb du^2 +g_{ab}\big( dy^a- \vsi \undB^a du\big) \big( dy^b-\vsi \undB^b du\big),
\eea
where
\bea
\Omb=e_3(s), \qquad \undB^a =\frac{1}{2} e_3(y^a), \qquad g_{ab}=\g(\pr_{y^a}, \pr_{y^b}).
\eea
Relative to these coordinates
\beaa
e_4=\pr_s, \qquad \pr_u = \vsi\left(\frac{1}{2}e_3-\frac{1}{2}\Omb e_4-\undB^a\pr_{y^a}\right), \qquad \pr_{y^a}= \sum_{c=1,2} Y_{(a)}^c e_c, \qquad a=1,2,
\eeaa
with coefficients $ Y_{(a)}^b $ verifying
\beaa
g_{ab}=\sum_{c=1, 2} Y_{(a)}^c Y_{(b)}^c.
\eeaa
As in \cite{KS-Kerr1} we assume that $\RR $ is covered by two coordinate systems , i.e. $\RR=\RR_N\cup \RR_S$,
such that
\begin{enumerate}
\item The North coordinate chart $\RR_N$ is given by the coordinates
$(u, s, y_{N}^1, y_{N}^2)$ with $(y^1_{N})^2+(y^2_{N})^2<2$.
\item The South coordinate chart $\RR_S$ is given by the coordinates
$(u, s, y_{S}^1, y_{S}^2)$ with $(y^1_{S})^2+(y^2_{S})^2<2$.
\item The two coordinate charts intersect in the open equatorial region
$\RR_{Eq}:=\RR_N\cap \RR_S$ in which both coordinate systems are defined.
\item In $\RR_{Eq} $ the transition functions between the two coordinate systems are given by the smooth functions $ \varphi_{SN}$ and $\varphi_{NS}= \varphi_{SN}^{-1} $.
\end{enumerate}
The metric coefficients for the two coordinate systems are given by
\beaa
\g &=& - 2\vsi du ds + \vsi^2\Omb du^2 +g^{N}_{ab}\big( dy_N^a- \vsi \undB_{N}^a du\big) \big( dy_N^b-\vsi \undB_N^b du\big),\\
\g &=& - 2\vsi du ds + \vsi^2\Omb du^2 +g^{S}_{ab}\big( dy_S^a- \vsi \undB_{S}^a du\big) \big( dy_S^b-\vsi \undB_S^b du\big),
\eeaa
where
\beaa
\Omb=e_3(s), \qquad \undB_N^a =\frac{1}{2} e_3(y_N^a), \qquad \undB_S^a =\frac{1}{2} e_3(y_S^a).
\eeaa
On a 2-sphere $S(u,s)$ and $f\in \SS_p(S)$, $p=0,1,2$, we consider the following norms,
\bea
\label{Norms-spacetimefoliation-GSMS}
\bsplit
\| f\|_{\infty} :&=\| f\|_{L^\infty(S)}, \qquad \| f\|_{2} :=\| f\|_{L^2(S)}, \\
\|f\|_{\infty,k} &= \sum_{i=0}^k \|\dk^i f\|_{\infty }, \qquad
\|f\|_{2,k}=\sum_{i=0}^k \|\dk^i f\|_{2},
\end{split}
\eea
where $\dk^i$ stands for any combination of length $i$ of operators of the form
$e_3, r e_4, r\nab $.
%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Reduced region $\RR$}
\lab{subsubsect:regionRR1}
%%%%%%%%%%%%%%%%%%%%%%
In what follows we restrict our attention to smaller region of the original $\RR$, still denoted $\RR$ defined as follows.
\begin{definition}
\label{defintion:regionRRovr}
Let $m_0>0$ a constant. Let $\epg >0$ two sufficiently small constants, and let $(\ug, \sg, \rg)$ three real numbers with $\rg$ sufficiently large so that
\bea\lab{eq:rangeofrgandepsilon}
\epg\ll m_0, \qquad\qquad \rg\gg m_0.
\eea
We define $\RR=\RR$ to be the region
\bea
\lab{definition:RR(dg,epg)}
\RR:=\left\{|u-\ug|\leq\epg,\quad |s-\sg|\leq \epg \right\},
\eea
such that assumptions {\bf A1-A4} below with constant $\epg$ on the background foliation of $\RR$, are verified.
\end{definition}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Main assumptions for $\RR$}
\lab{subsubsect:regionRR2}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Given an integer $s_{max}\geq 3$, we assume\footnote{In view of \eqref{eq:assumtioninRRforGagandGabofbackgroundfoliation}, we will often replace $\Ga_g$ by $r^{-1} \Ga_b$.} the following.
\begin{enumerate}
\item[\bf A1.]
For $k\le s_{max}$
\bea\lab{eq:assumtioninRRforGagandGabofbackgroundfoliation}
\bsplit
\| \Ga_g\|_{k, \infty}&\leq \epg r^{-2},\\
\| \Ga_b\|_{k, \infty}&\leq \epg r^{-1}.
\end{split}
\eea
\item[\bf A2.] The Hawking mass $m=m(u,s)$ of $S(u, s)$ verifies
\bea\lab{eq:assumtionsonthegivenusfoliationforGCMprocedure:Hawkingmass}
\sup_{\RR}\left|\frac{m}{m_0}-1\right| &\leq& \epg.
\eea
\item[\bf A3.]
In the region of their respective validity\footnote{That is the quantities on the left verify the same estimates as those for $\Ga_b$, respectively $\Ga_g$.} we have
\bea
\undB_N^a,\,\, \undB_S^a \in r^{-1}\Ga_b, \qquad Z_N^a,\,\, Z_S^a \in \Ga_b,\qquad r^{-2} \widecheck{g}^{N}_{ab}, \,\, r^{-2} \widecheck{g}^{S}_{ab} \in r\Ga_g
\eea
where
\beaa
\widecheck{g} ^{N}\!_{ab} &=& g^N_{ab} - \frac{4r^2}{1+(y^1_{N})^2+(y^2_{N})^2) }\de_{ab},\\
\widecheck{g}^{S}\!_{ab} &=& g^S_{ab} - \frac{4r^2}{(1+(y^1_{S})^2+(y^2_{S})^2) } \de_{ab}.
\eeaa
\item[\bf A4.] We assume the existence of a smooth family of scalar functions\footnote{The property of the scalar functions $\Jp$ above is motivated by the fact that the $\ell=1$ spherical harmonics on the standard sphere $\SSS^2$, given by $J^{(0, \SSS^2)}=\cos\th, \, J^{(+, \SSS^2)}=\sin\th\cos\vphi, \, J^{(-, \SSS^2)}=\sin\th\sin\vphi$,
satisfy \eqref{eq:Jpsphericalharmonics} with $\epg=0$. Note also that on $\SSS^2$,
\beaa
\int_{\mathbb{S}^2}(\cos\th)^2=\int_{\mathbb{S}^2}(\sin\th\cos\vphi)^2=\int_{\mathbb{S}^2}(\sin\th\sin\vphi)^2=\frac{4\pi}{3}, \qquad |\SSS^2|=4\pi.
\eeaa,}
$\Jp:\RR\to\RRR$, for $p=0,+,-$, verifying the following properties, for all surfaces $S$ of the background foliation
\bea
\lab{eq:Jpsphericalharmonics}
\bsplit
\Big(r^2\lap+2\Big) \Jp &= O(\epg),\qquad p=0,+,-,\\
\frac{1}{|S|} \int_{S} \Jp J^{(q)} &= \frac{1}{3}\de_{pq} +O(\epg),\qquad p,q=0,+,-,\\
\frac{1}{|S|} \int_{S} \Jp &=O(\epg),\qquad p=0,+,-.
\end{split}
\eea
\end{enumerate}
\begin{remark}
We note that the assumptions {\bf A1}, {\bf A2}, {\bf A3}, {\bf A4}, are expected to be valid in the far regions, i.e. $r$ large, of a perturbed Kerr.
\end{remark}
Given a scalar function $f$ defined on a sphere $S$ we define the $\ell=1$ modes of $f$ by the triplet
\beaa
(f)_{\ell=1} :=\left\{ \int_S f \Jp, \qquad p=0,+,-\right\}.
\eeaa
%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{$O(\epg)$-spheres}
%%%%%%%%%%%%%%%%%%%%%%%%%
Given a compact $2$-surface $\S\subset \RR$, not necessarily a leaf $S(u,s)$ of the background geodesic foliation of $\RR$, we denote by $\chi^\S$, $\chib^\S$, $\ze^\S$,..., the corresponding Ricci coefficients and by $\a^\S$, $\b^\S$, $\rho^\S$, ..., the corresponding curvature coefficients. We also denote by $r^\S$, $m^\S$, $K^\S$ and $\mu^\S$ respectively the corresponding area radius, Hawking mass, Gauss curvature and mass aspect function.
Finally, we denote by $\nab^\S$ the corresponding covariant derivative on $\S$.
Also, we introduce the following norm on $\S$
\bea
\lab{definition:spaceH^k(boldS)}
\| f\|_{\hk_s(\S)}:&=&\sum_{i=0}^s \|( \dkb^\S )^i f\|_{L^2(\S)}, \qquad \dkb^\S =r^\S \nab^\S.
\eea
%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{General frame transformations}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{lemma}
\lab{Lemma:Generalframetransf}
Given a null frame $(e_3, e_4, e_1, e_2)$, a general null transformation from the null frame $(e_3, e_4, e_1, e_2)$ to another null frame $(e_3', e_4', e_1', e_2')$ can be written in the form,
\bea
\lab{eq:Generalframetransf}
\bsplit
e_4'&=\la\left(e_4 + f^b e_b +\frac 1 4 |f|^2 e_3\right),\\
e_a'&= \left(\de_{ab} +\frac{1}{2}\fb_af_b\right) e_b +\frac 1 2 \fb_a e_4 +\left(\frac 1 2 f_a +\frac{1}{8}|f|^2\fb_a\right) e_3,\qquad a=1,2,\\
e_3'&=\la^{-1}\left( \left(1+\frac{1}{2}f\c\fb +\frac{1}{16} |f|^2 |\fb|^2\right) e_3 + \left(\fb^b+\frac 1 4 |\fb|^2f^b\right) e_b + \frac 1 4 |\fb|^2 e_4 \right),
\end{split}
\eea
where $\la$ is a scalar, $f$ and $\fb$ are horizontal 1-forms. The dot product and magnitude $|\c |$ are taken with respect to the standard euclidian norm of $\RRR^2$. We call $(f, \fb, \la)$ the transition coefficients of the change of frame. We denote $F:=(f, \fb, \ovla)$ where $\ovla=\la-1$.
\end{lemma}
Relative to the primed frame the connection coefficients $\Ga' $ can be related to the connection coefficients
$\Ga$ of the background frame by specific formulas involving the transition coefficients and their derivatives.
Similarly the curvature coefficients $R'$ of the primed frame are connected to $R$ by formulas involving
the transition coefficients. The precise formulas are given in Proposition 3.3 in \cite{KS-Kerr1}. In this paper we only make use of the transformation formulas for $\ka=\trch$, $\kab =\trchb$ and $\b$. The transformation formulas for the first two have the form, see Proposition 3.4 in \cite{KS-Kerr1},
\bea
\lab{eq:transfforbandkab1}
\bsplit
\ka' &= \ka+ \ka \ovla+\div'f +F\c \Ga_b+ F\c\nab' F+r^{-1} F^2,\\
\kab' &= \kab- \kab \ovla+\div'\fb +F\c \Ga_b+ F\c\nab' F+r^{-1} F^2.
\end{split}
\eea
The transformation formula for $\b$ has the form, see Proposition 3.3. in \cite{KS-Kerr1},
\bea
\lab{eq:transfforbandkab2}
\b'&=& \b +\frac 3 2\big( f \rho+\dual f \rhod\big)+ r^{-1} \Ga_g \c F.
\eea
%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Deformation of surfaces in $\RR$}
%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{definition}
\label{definition:Deformations}
We say that $\S$ is an $O(\dg)$\, deformation of $ \ovS$ if there exist smooth scalar functions $U, S$ defined on $\ovS$ and a map
a map $\Psi:\ovS\to \S $ verifying, on any coordinate chart $(y^1, y^2) $ of $\ovS$,
\bea
\Psi(\ovu, \ovs, y^1, y^2)=\left( \ovu+ U(y^1, y^2 ), \, \ovs+S(y^1, y^2 ), y^1, y^2 \right)
\eea
with $(U, S)$ smooth functions on $\ovS$ of size $\dg$.
\end{definition}
\begin{definition}
Given a deformation $\Psi:\ovS\to \S$ we say that
a new frame $(e_3', e_4', e_1', e_2')$ on $\S$, obtained from the standard frame $(e_3, e_4, e_1, e_2)$ via the transformation \eqref{eq:Generalframetransf}, is $\S$-adapted if the horizontal vectorfields $e'_1, e'_2$ are tangent to $\S$ or, equivalently $e_3', e_4' $ are perpendicular to $S$.
\end{definition}
We will need the following lemmas.
\begin{lemma}
\lab{Le:Transportcomparison}
Let $\Psi:\ovS\longrightarrow \S $ be a deformation in $\RR$ as in Definition \ref{definition:Deformations}, $F$ a scalar function on $\RR$ and $F^\# =F\circ \Psi $ its pull back to $\ovS$ by $\Psi$. We have
\bea
\big\| F^\#-F\big\|_{L^\infty(\ovS)}&\les& \big \| ( U, S) \big\|_{L^\infty(\ovS)} \sup_{\RR}\Big(\big| e_3 F \big|+ |e_4 F|+| \nab F |\Big).
\eea
\end{lemma}
\begin{proof}
See Lemma 5.7 in \cite{KS-Kerr1}.
\end{proof}
\begin{lemma}\lab{lemma:comparison-gaS-ga}
Let $\ovS \subset \RR$. Let $\Psi:\ovS\longrightarrow \S $ be a deformation generated by the functions $(U, S)$ as in Definition \ref{definition:Deformations} and denote by $g^{\S,\#}$ the pull back of the metric $g^\S$ to $\ovS$. Assume the bound, for $s\le s_{max}+1$.
\bea
\label{assumption-UV-dg}
\| (U, S)\|_{L^\infty(\ovS)} +r ^{-1} \big\|(U, S)\big\|_{\hk_{s}(\ovS)} &\les& \dg.
\eea
Then
\bea
\frac{r^\S}{\ovr}= 1 + O(r ^{-1} \dg )
\eea
where $r^\S$ is the area radius of $\S$ and $\ovr$ that of $\ovS$.
Also,
\bea
\big\| g^{\S, \#} -\ovg\big\|_{L^\infty} +r^{-1} \big\| g^{\S, \#} -\ovg\big\|_{\hk_{s}(\ovS)}\les \dg r.
\eea
\end{lemma}
\begin{proof}
See Lemma 5.8 in \cite{KS-Kerr1}.
\end{proof}
As a corollary, we deduce the following.
\begin{lemma}
\lab{Lemma:coparison-forintegrals}
Under the same assumptions as in Lemma \ref{lemma:comparison-gaS-ga}, the following estimate holds for a scalar function $F$ defined on $\RR$.
\beaa
\left|\int_\S F -\int_{\ovS} F\right| &\les& \dg r \sup_{\RR} \Big( |F| + r \big(|e_3 F|+ |e_4 F|+|\nab F| \big)\Big).
\eeaa
\end{lemma}
\begin{proof}
We have,
\beaa
\int_\S F -\int_{\ovS} F&=&\int_{\ovS} F\circ \Psi \frac{\sqrt{\det g^{\S,\#}}}{\sqrt{ \det \ovg}} da_{\ovg}-\int_{\ovS} F da_{\ovg} \\
&=&
\int_{\ovS} F\circ\Psi \left(\frac{\sqrt{\det g^{\S,\#}}}{\sqrt{\det \ovg}}-1\right) + \int_{\ovS} \big|F\circ\Psi-F\big|
\eeaa
where $g^{\S,\#}$ is the pull back of the metric $g^\S$ to $\ovS$ by the map $\Psi$.
In view of Lemma \ref{lemma:comparison-gaS-ga}
\beaa
\big\| g^{\S, \#} -\ovg \big\|_{L^\infty(\ovS)} &\les r\dg.
\eeaa
Hence, using also Lemma \ref{Le:Transportcomparison},
\beaa
\left|\int_\S F -\int_{\ovS} F\right|&\les&\dg \rg \sup_{\ovS} | F|+
\big(\ovr\big)^2 \sup_{\ovS} \big |F\circ\Psi-h\big|\\
&\les& \dg \rg \sup_{\ovS} | F|+ \dg ( \ovr )^2 \Big(|e_3 F|+ |e_4 F|+|\nab F| \Big)
\eeaa
as stated.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Adapted non canonical $\ell=1$ modes}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Consider a deformation $\Psi:\ovS \to \S$ and recall the existence of the family of scalar functions $\Jp$, $p\in\big\{0, +, -\big\}$, on $\RR$ introduced in assumption {\bf A4}, see \eqref{eq:Jpsphericalharmonics}, which form a basis of the $\ell=1$ modes on the spheres $S(u,s)$ of $\RR$, and hence in particular on $\ovS$. We denote by
$\Jpov $ the restriction of the family $\Jp$ to $\ovS$.
\begin{definition}
\lab{def:ell=1sphharmonicsonS}
We define the basis of adapted $\ell=1$ modes $\JpS$ on $\S$ by
\beaa
\JpS = \Jpov\circ\Psi^{-1}, \qquad p\in\big\{ -, 0, +\big\}.
\eeaa
\end{definition}
\begin{proposition}
Assume that the parameters $(U, S)$ of the deformation verifies the bounds
\bea\lab{eq:boundonUSin h_snorms}
\|( U, S)\|_{\hk_{s_{max}+1}(\ovS)} &\les& r\dg.
\eea
Assume $\Jp$ is an admissible triplet of $\ell=1$ modes on $\RR$ (and hence on $\ovS$), i.e. satisfying \eqref{eq:Jpsphericalharmonics}.
Then $\JpS$ is an admissible triplet of $\ell=1$ modes on $\S$, i.e.
\bea
\lab{eq:admissibleJpS}
\bsplit
\Big((r^\S)^2\lap^\S +2\Big) \JpS &= O(\epg),\\
\frac{1}{|\S|} \int_{\S} \JpS J^{( q, \S)} &= \frac{1}{3}\de_{pq} +O(\epg),\\
\frac{1}{|\S|} \int_{\S} \JpS &=O(\epg).
\end{split}
\eea
Moreover at all point of $\S$ we have
\bea
\lab{eq:admissibleJpS2}
\Big|\JpS-\Jp\Big|&\les \dg.
\eea
\end{proposition}
\begin{proof}
See Proposition 5.12 in \cite{KS-Kerr1}.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{GCM spheres with canonical $\ell=1$ modes}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{GCM spheres with non-canonical $\ell=1$ modes in \cite{KS-Kerr1}}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
We review below the result on existence and uniqueness of GCM spheres proved in \cite{KS-Kerr1}. The result was proved in the context of an arbitrary choice of a $\ell=1$ modes on $\ovS$, denoted $\Jp$ which verify
the assumptions {\bf A4}.
\begin{theorem}[GCM spheres with non-canonical $\ell=1$ modes \cite{KS-Kerr1}]
\lab{Theorem:ExistenceGCMS1}
Let $\ovS=S(\ovu, \ovs)$ be a fixed sphere of the background geodesic foliation\footnote{ Verifying the assumptions ${\bf A1-A4}$. Note in particular that assumption {\bf A3} was made w.r.t. a general version of the ${\ell=1} $ modes of $\S$, see Definition \ref{def:ell=1sphharmonicsonS}.} of $\RR$ with $\rg$ and $\mg$ denoting its area radius and Hawking mass.
Assume that the GCM quantities $\ka, \kab, \mu$ of the background foliation verify the following:
\bea
\bsplit
\ka&=\frac{2}{r}+\dot{\ka},\\
\kab&=-\frac{2\Up}{r} + \Cb_0+\sum_p \Cbp \Jp+\dot{\kab},\\
\mu&= \frac{2m}{r^3} + M_0+\sum _p\Mp \Jp+\dot{\mu},
\end{split}
\eea
where
\bea
\lab{eq:GCM-improved estimate1-again}
|\Cb_0, \Cbp| &\les& r^{-2} \epg, \qquad |M_0, \Mp| \les r^{-3} \epg,
\eea
and
\bea
\lab{eq:GCM-improved estimate2-again}
\big\| \kadot, \kabdot\|_{\hk_{s_{max} }(\S) }&\les& r^{-1}\dg,\qquad
\big\|\mudot\| _{\hk_{s_{max} }(\S) }\les r^{-2}\dg.
\eea
Then
for any fixed triplets $\La, \Lab \in \RRR^3$ verifying
\bea\lab{eq:assumptionsonLambdaabdLambdabforGCMexistence}
|\La|,\, |\Lab| &\les & \dg,
\eea
there
exists a unique GCM sphere $\S=\S^{(\La, \Lab)}$, which is a deformation of $\ovS$,
such that there exist constants $\Cb^\S_0$, $\CbpS$, $ M^\S_0$, $\MpS$, $p\in\{-,0, +\}$ for which the following GCM conditions are verified
\bea
\lab{def:GCMC}
\bsplit
\ka^\S&=\frac{2}{r^\S},\\
\kab^\S &=-\frac{2}{r^\S}\Up^\S+ \Cb^\S_0+\sum_p \CbpS \JpS,\\
\mu^\S&= \frac{2m^\S}{(r^\S)^3} + M^\S_0+\sum _p\MpS \JpS,
\end{split}
\eea
where
\bea
\lab{eq:non-canonicalJpS}
\JpS=\Jpov\circ\Psi^{-1}.
\eea
Relative to these modes we also have
\bea
\lab{GCMS:l=1modesforffb}
(\div^\S f)_{\ell=1}&=\La, \qquad (\div^\S\fb)_{\ell=1}=\Lab.
\eea
The resulting deformation has the following additional properties:
\begin{enumerate}
\item The triplet $F=(f,\fb,\ovla)$ verifies, for $s\le s_{max}+1$,
\bea
\lab{eq:ThmGCMS1}
\|\ovla\|_{\hk_{s+1}(\S)}+\|(f,\fb)\|_{\hk_{s }(\S)} &\les &\dg.
\eea
\item The GCM constants $\Cb^\S_0,\, \CbpS$, \, $ M^\S_0$, \, $\MpS, \, p\in\{-,0, +\}$ verify
\bea
\lab{eq:ThmGCMS2}
\bsplit
\big| \Cb^\S_0-\Cb_0\big|+\big| \CbpS-\Cbp\big|&\les r^{-2}\dg,\\
\big| M^\S_0-M_0\big|+\big| \MpS-\Mp\big|&\les r^{-3}\dg.
\end{split}
\eea
\item The volume radius $r^\S$ verifies
\bea
\lab{eq:ThmGCMS3}
\left|\frac{r^\S}{\rg}-1\right|\les r^{-1} \dg.
\eea
\item The parameter functions $U, S$ of the deformation verify, for $s\le s_{max}+1$,
\bea
\lab{eq:ThmGCMS4}
\|( U, S)\|_{\hk_{s}(\ovS)} &\les& r \dg.
\eea
\item The Hawking mass $m^\S$ of $\S$ verifies the estimate
\bea
\lab{eq:ThmGCMS5}
\big|m^\S-\ovm\big|&\les &\dg.
\eea
\item The well defined\footnote{Note that while the Ricci coefficients $\ka^\S, \kab^\S, \chih^\S, \chibh^\S, \ze^\S$ as well as all curvature components and mass aspect function $\mu^\S$ are well defined on $\S$, this in not the case of $\eta^\S, \etab^\S, \xi^\S, \xib^\S, \om^\S, \omb^\S$ which require the derivatives of the frame in the $e_3^\S$ and $e_4^\S$ directions.}
Ricci and curvature coefficients of $\S$ verify,
\bea
\lab{eq:ThmGCMS6}
\bsplit
\| \Ga^\S_g\|_{\hk_{s_{max} }(\S) }&\les \epg r^{-1},\\
\| \Ga^\S_b\|_{\hk_{s_{max} }(\S) }&\les \epg.
\end{split}
\eea
\item The transition parameters $f, \fb, \ovla$ are continuously differentiable with respect to $\La, \Lab $ and
\bea\lab{eq:property7oftheoldGCMtheoremnoncanonicalell=1modes}
\bsplit
\frac{\pr f }{\pr \La}&=O\big( r^{-1}\big), \quad \frac{ \pr f }{\pr \Lab}=O\big(\dg r^{-1} \big), \quad
\frac{\pr\fb }{\pr \La}&=O\big(\dg r^{-1}\big), \quad \frac{\pr\fb}{\pr \Lab}=O\big( r^{-1} \big),\\
\frac{\pr\ovla }{\pr \La}&=O\big(\dg r^{-1}\big), \quad \frac{\pr\ovla }{\pr \Lab}=O\big(\dg r^{-1}\big).
\end{split}
\eea
\item The parameter functions $(U, S)$ of the deformation are are continuously differentiable with respect to $\La, \Lab $
and
\bea
\frac{\pr U}{\pr \La}=O(1), \quad \frac{\pr U}{\pr \Lab}=O(1),\quad \frac{\pr S}{\pr \La}=O(1) , \quad \frac{\pr S}{\pr \Lab}=O(1).
\eea
\item Relative to the coordinate system induced by $\Psi$ the metric $g^{\La, \Lab}$ of $\S=\S^{(\La, \Lab)}$ is continuous with respect to the parameters $\La, \Lab$ and verifies
\bea\lab{eq:property9oftheoldGCMtheoremnoncanonicalell=1modes}
\big\| \pr_\La g^\S, \, \pr_{\Lab} g^\S\|_{L^\infty(\S)} &\les O( r^2).
\eea
\item The Deformations $\S^{(\La, \Lab)}$ also depend continuously on the choice the $\ell=1$ basis of $\ovS$ verifying condition {\bf A4.}
\end{enumerate}
\end{theorem}
Note that the $\ell=1$ modes on $\S$ given by \eqref{eq:non-canonicalJpS} are not canonical on $\S$, i.e. they do not correspond to the modes defined using the effective uniformization map for $\S$. In the next section we show that we can modify the result of Theorem \ref{Theorem:ExistenceGCMS1} so that the same statements hold when we replace $\JpS$ by canonical modes. In the proof of this new theorem, we make use of Theorem \ref{Theorem:ExistenceGCMS1} for the following class of $\ell=1$ modes $\Jp$ for $\ovS$.
\begin{definition}
\lab{definition:J[S]modes}
Given an arbitrary $O(\dg)$ deformation of $\ovS$, $\Psi: \ovS\to S $ and $ J^S=\big\{J^{(p,S)}, \,\, p=-,0,+\big\} $ a canonical $\ell=1$
basis for $S$ we denote by
\beaa
\ovJ[S]=\big\{\ovJ^{(p)}[S], \quad p=-,0,+\big\}
\eeaa
the $\ell=1$ basis of $\ovS$ given by
\bea
\ovJ^{(p)}[S]= J^{(p,S)}\circ\Psi, \qquad \text{ or simply }\quad \ovJ[S]= J^S\circ\Psi.
\eea
\end{definition}
Note that $\ovJ[S]$ defined above satisfies the assumptions ${\bf A4}$.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Construction of GCM spheres with canonical $\ell=1$ modes}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
We state and prove a new version of Theorem \ref{Theorem:ExistenceGCMS1} stated relative to a canonical
definition of ${\ell=1} $ modes of $\S$. In other words we replace the adapted $\JpS $ modes on $\S$ defined by \eqref{eq:non-canonicalJpS}
\beaa
\JpS&=&\Jpov \circ \Psi^{-1}
\eeaa
with the following definition of canonical modes for deformed surfaces $\S$.
\begin{definition}
\lab{definition:canmodesS}
Given a deformation map $\Psi:\ovS\to \S$ and a fixed effective uniformization map
$(\ovPhi, \ovphi)$ for $\ovS$ we let $(\Phi, \phi) $ be the unique effective uniformization map of $\S$ calibrated
with $(\ovPhi, \ovphi)$ relative to the map $\Psi$, in the sense of Definition \ref{def:calibration}. With this choice we define the canonical $\ell=1$ modes of $\S$ by the formula
\bea
\lab{def:JpSS}
\JpSS&=&J^{(p, \SSS^2)} \circ \Phi^{-1}
\eea
with $J^{(p, \SSS^2)}$ the $\ell=1 $ basis on $\SSS^2$ given by Definition \ref{definition:ell=1mpdesonS}.
\end{definition}
\begin{remark}
\lab{remark:calibration-of-deformations}
Assume given two deformations $\Psi_i: \ovS\to \S_i$, $ i=1,2$ with canonical effective uniformization maps $(\Phi_i, u_i) $, $\Phi_i:\SSS^2 \to \S_i$ $ i=1,2$, both calibrated through the respective maps with the fixed effective uniformization map $(\ovPhi, \ovphi)$ for $\ovS$. Then, in view of Lemma \ref{Le:Transitivitycalibrations}, $(\Phi_1, u_1) $, $(\Phi_2, u_2) $ are also calibrated with respect to the map $\Psi_{12}: \S_1 \to \S_2$, with $\Psi_{12}=\Psi_2\circ\Psi_1^{-1}$.
\end{remark}
\begin{theorem}[GCM spheres with canonical $\ell=1$ modes]
\lab{Theorem:ExistenceGCMS1-2}
Under the same assumptions as in Theorem \ref{Theorem:ExistenceGCMS1},
for any fixed triplets $\La, \Lab \in \RRR^3$ verifying
\bea\lab{eq:assumptionsonLambdaabdLambdabforGCMexistence:bis}
|\La|,\, |\Lab| &\les & \dg,
\eea
there
exists a unique GCM sphere $\S=\S^{(\La, \Lab)}$, which is a deformation of $\ovS$,
such that there exist constants $\Cb^\S_0$,\, \, $\CbpS$, \, $ M^\S_0$, \, $\MpS, \, p\in\{-,0, +\}$ for which the following GCM conditions are verified
\bea
\lab{def:GCMCversion2}
\bsplit
\ka^\S&=\frac{2}{r^\S},\\
\kab^\S &=-\frac{2}{r^\S}\Up^\S+ \Cb^\S_0+\sum_p \CbpS \J^{(p, \S)} ,\\
\mu^\S&= \frac{2m^\S}{(r^\S)^3} + M^\S_0+\sum _p\MpS \J^{(p, \S)}.
\end{split}
\eea
Moreover,
\bea
\lab{GCMS:l=1modesforffb.version2}
(\div^\S f)_{\ell=1}&=\La, \qquad (\div^\S\fb)_{\ell=1}=\Lab.
\eea
where $\J^{(p, \S)}$ form the basis of canonical $\ell=1$ modes of $\, \S$ in the sense of Definition \ref{definition:canmodesS} and $(\div^\S f)_{\ell=1}$, $(\div^\S \fb)_{\ell=1}$ are defined
with respect to them. Moreover
\begin{enumerate}
\item The resulting deformation verifies all the properties 1-9 as in Theorem \ref{Theorem:ExistenceGCMS1}.
\item The canonical $\ell=1$ modes $\J^{(p, \S)}$ verify,
\bea
\lab{eq:PropertiesofJ^S-strong2}
\bsplit
\lap_S \J^{(p,\S)} &=\left(-\frac{2}{(r^\S)^2} +O\left(\frac{\epg}{(r^\S)^3}\right) \right)\J^{(p,\S)}, \\
\int_S \J^{(p,\S)}\J^{(q,\S)} da_g &= \frac{4\pi}{3}(r^\S)^2\de_{pq}+ O(\epg r^\S),\\
\int_S \J^{(p,\S)} da_g&=0.
\end{split}
\eea
\end{enumerate}
\end{theorem}
\begin{remark}
Note that \eqref{eq:PropertiesofJ^S-strong2} follows from the fact that $K^\S-\frac{1}{(r^{\S})^2}\in r^{-1} \Ga_g$, i.e.
\beaa
\left| K^\S-\frac{1}{(r^{\S})^2}\right| &\les& \frac{\epg}{(r^\S)^{3}}
\eeaa
and the result of Lemma \ref{lemma:basicpropertiesofJforcanonicalell=1basis}
applied to the case when $\ep=r^{-1} \epg$.
\end{remark}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Proof of Theorem \ref{Theorem:ExistenceGCMS1-2}}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
We proceed by an iteration argument in which, in order to make comparison between the
$\ell =1$ modes of different deformation spheres, we shall need the following lemma.
\begin{lemma}
\lab{lemma:comparisonJ2}
Assume given two almost round spheres\footnote{Such that \eqref{eq:Almostround-S-k} is verified.} $(\S_1, g^{\S_1} )$ and $(\S_2, g^{\S_2} )$, their respective canonical uniformization maps
$(\Phi_1, \phi_1)$, $(\Phi_2, \phi_2)$, and $\J^{(p, \S_1)}, \, \J^{(p, \S_2)} $ be the corresponding canonical basis of $\ell=1$ modes of $\S_1$ and $\S_2$ calibrated according to Definition \ref{def:calibration}.
Consider also a given smooth map\, $ \Psi :\S_1\to \S_2$ and assume that the two metric are comparable relative to a small parameter $\de_{12 }$, i.e.
\bea
\lab{eq:unifor-twometricsB1}
\| g^{\S_1}- \Psi^\#( g^{\S_2} )\|_{\hk^{2}(\S^1)} \le \de_{12} (r^{\S_1})^2.
\eea
Then the following estimate holds true.
\bea
\max_p \sup_{\S_1}\Big|\J^{(p, \S_1)}- \J^{(p, \S_2)} \circ \Psi \Big| \les (r^{\S_1}) ^{-1} \de_{12}.
\eea
\end{lemma}
\begin{proof}
This follows immediately from the proof of Proposition \ref{Prop:comparison.canJ}
with $\de = (r^{\S_1}) ^{-1} \de_{12} $.
\end{proof}
We rephrase the statement of the lemma above by introducing the following notation.
\begin{definition}
Given two neighboring, almost round spheres, $(\S_1, g^{\S_1} )$ and $(\S_2, g^{\S_2} )$
and a map $ \Psi: \S_1\longrightarrow \S_2$ we define their distance relative to $\Psi$, measured in $\S_1$, to be
\bea
\lab{distance:g_1g_2}
d_{\S_1}( g^{\S_1}, g^{\S_2}) &:=& (r^{\S_1})^{-2} \| g^{\S_1}- \Psi^\#( g^{\S_2} )\|_{\hk^{2}(\S^1)}.
\eea
\end{definition}
\begin{corollary}\lab{lemma:comparisonJ2:corollary}
Under the same situation as in Lemma \ref{lemma:comparisonJ2} we have,
\bea
\sup_{\S_1} \Big|\J^{\S_1}- \J^{\S_2} \circ \Psi \Big| \les (r^{\S_1}) ^{-1} d_{\S_1}( g^{\S_1}, g^{\S_2}).
\eea
\end{corollary}
%%%%%%%%%%%%%%%%%%%%
\subsubsection{Iteration procedure}
%%%%%%%%%%%%%%%%%%%%
We define iteratively a sequence of maps $\Psi(n) :\ovS\longrightarrow \S(n)$, for fixed $\La, \Lab$, as follows.
\begin{enumerate}
\item We start with the deformation map $\Psi(1) :\ovS\longrightarrow \S(1)$ constructed by Theorem \ref{Theorem:ExistenceGCMS1} with respect to $\ovJ(1):=\ovJ_{can} $, the canonical $\ell=1$ basis on $\ovS$. We define $J^1 $ to be the adapted $\ell=1$ basis on $S(1)$ i.e.
\beaa
J^1=J(1)=\ovJ_{can} \circ \Psi(1)^{-1}.
\eeaa
We also denote by $\J^1_{can}=\J(1)_{can} $ the canonical $\ell=1$ basis of $\S(1)$.
\item We define $\Psi(2) :\ovS\longrightarrow \S(2) $ to be the deformation map
constructed by Theorem \ref{Theorem:ExistenceGCMS1} relative to the $\ell=1$ basis on $\ovS$ given by
\bea
\ovJ(2) := \J_{can}(1)\circ\Psi(1).
\eea
We note that $ \ovJ(2)$ verifies assumption {\bf A4} and thus the assumptions of Theorem
\ref{Theorem:ExistenceGCMS1} apply.
The adapted $\ell=1$ basis of $\S(2)$ is given by
\bea
J^2=J(2):=\ovJ(2)\circ \Psi(2)^{-1}= \J^1_{can}\circ\Psi(1) \circ \Psi(2)^{-1}.
\eea
\item Assume that the sequence of deformation maps
$\Psi(n) :\ovS\longrightarrow \S(n)$ has already been constructed by Theorem \ref{Theorem:ExistenceGCMS1} with the help of the sequence $\ovJ(n) $ of $\ell =1$ bases
on $\ovS$ given recursively by
\bea
\ovJ(n)&=& \J_{can}^{n-1} \circ \Psi(n-1)
\eea
where $\J_{can}^{n-1}=\J_{can}(n-1)$ is the canonical $\ell =1$ basis of $\S(n-1)$.
The adapted sequence of $\ell =1$ bases of $S(n)$ is given by
\bea
J^n=J(n) := \ovJ(n)\circ\Psi(n)^{-1}.
\eea
We then define $\Psi(n+1) :\ovS\longrightarrow \S(n+1)$ to be the new deformation map
constructed by Theorem \ref{Theorem:ExistenceGCMS1} relative to the $\ell=1$ basis on $\ovS$ given by
\bea
\ovJ(n+1)&=& \J^n_{can} \circ \Psi(n),
\eea
where $\J_{can} ^n=\J_{can}(n)$ is the canonical $\ell =1$ basis of $\S(n)$.
We define the adapted $\ell=1$ basis on $\S(n+1)$ by
\beaa
J^{n+1}= J(n+1)= \ovJ(n+1)\circ\Psi(n+1)^{-1} = \J_{can}^n \circ \Psi(n)\circ\Psi(n+1)^{-1}.
\eeaa
Introducing the maps
\bea
\bsplit
& \Psi_{n+1, n}:\S(n+1) \longrightarrow \S(n), \qquad \Psi_{n+1, n}:= \Psi(n)\circ \Psi(n+1)^{-1}, \\
&\Psi_{n, n+1}:\S(n) \longrightarrow \S(n+1), \qquad \Psi_{n, n+1}:=\Psi(n+1)\circ \Psi(n) ^{-1},
\end{split}
\eea
we deduce the formula
\bea
\lab{eq:identityJ^nJ^{n+1}}
J^{n+1}&= \J^n_{can} \circ \Psi_{n+1, n}.
\eea
\end{enumerate}
%%%%%%%%%%%%%%%%%%%%%%%%%
\subsubsection{Boundedness and contraction}
%%%%%%%%%%%%%%%%%%%%%%%%%%
Consider the quintets
\beaa
\QQ^{(n)}= \Big(U^{(n)}, S^{(n)}, \ovla^{(n)}, f^{(n)}, \fb^{(n)}\Big)=
\Big(U^{(n)}, S^{(n)}, F^{(n)}\Big)
\eeaa
where $ U^{(n)}, S^{(n)} $ are the defining functions of the map $\Psi(n)$ and $F^{(n)}$
the corresponding transition function defined in Theorem \ref{Theorem:ExistenceGCMS1}. Consider also the corresponding ninetets
\beaa
\NN^{(n)}:=\Big(U^{(n)}, S^{(n)}, \ovla^{(n)}, f^{(n)}, \fb^{(n)}; \, \Cbn_0, \Mn_0, \Cbpn, \Mpn\Big).
\eeaa
We define the norms,
\bea
\label{quintet-norm}
\bsplit
\| Q^{(n)}\|_k: &= r^{-1} \Big\| \big(U^{(n)}, S^{(n)}\big)\Big \|_{\hk_{k}(\ovS)} +\Big \|\big( f^{(n)}, \fb^{(n)}, \ovla^{(n)}\big) \Big\|_{\hk_{k}(\S)}
\end{split}
\eea
and,
\beaa
\big\| \NN^{(n)} \big\|_k= \| \QQ^{(n)}\|_k+ r^2 \left(\big| \Cbndot_0\big|+\sum_p\big| \Cbpndot\big| +r \big| \Mndot_0\big|+
r \sum_p\big| \Mpndot\big|\right)
\eeaa
where,
\beaa
\Cbndot_0= \Cbn_0-\Cb_0, \,\,\,\, \Cbpndot= \Cbpn-\Cbp, \,\,\,\, \Mndot_0= \Mn_0-M_0, \,\,\,\, \Mpndot=\Mpn-\Mp.
\eeaa
The equations \eqref{def:GCMC} verified by each ninetet $\NN^{(n)}$ are given by
\bea
\lab{def:GCMC-n}
\bsplit
\ka^{\S(n)}&=\frac{2}{r^{\S(n)}},\\
\kab^{\S(n)} &=-\frac{2\Up^{\S(n)}}{r^{\S(n)}}+ \Cb^{(n)}_0+\sum_p\Cb^{(n),p} J^{n,p},\\
\mu^{\S(n)}&= \frac{2m^{\S(n)}}{(r^{\S(n)})^3} + M^{(n)}_0+\sum_p M^{(n),p} J^{n,p}.
\end{split}
\eea
Relative to the modes $J^n $ we also have
\bea
\lab{GCMS:l=1modesforffb-n}
\int_{\S(n)} (\div^{\S(n)} \fn) J^n &=\La, \qquad \int_{\S(n)} (\div^{\S(n)}\fbn) J^n =\Lab.
\eea
The maps $\Psi(n)=\big(\Un, \Sn \big) $ and the triplet $\Fn =\big( \fn, \fbn, \ovlan\big)$ verify a coupled system of equations which we write schematically as follows\footnote{See the system (6.16)-(6.19) in \cite{KS-Kerr1} for a full description.} .
\bea
\bsplit
\lapzero \Sn&= \divzero\Big( \SS(\fn, \fbn, \Ga)_b Y^b_{(a)}(n) \Big)^{\#(n)}, \\
\lapzero \Un&= \divzero\Big( \UU(\fn, \fbn, \Ga)_b Y^b_{(a)}(n) \Big)^{\#(n)},\\
\DDn \Fn&=\GG^{(n)} + \HH(\Fn, \Ga).
\end{split}
\eea
Here $\#_n$ denotes the pull-back by the map $\Psi(n)$ and $Y_{(a)}(n)$ the push-forward by $\Psi(n)$ of the coordinate vectorfields $\pr_{y^a} $, $a=1,2$. The expressions
$(\UU, \SS)$
are linear plus higher order terms in $(f, \fb) $ and $\HH$ is linear plus higher order in $F$. The term\footnote{It also depends on the quantities appearing in \eqref{eq:GCM-improved estimate1-again}, \eqref{eq:GCM-improved estimate2-again}. } $ \GG^{(n)}$ is linear with respect to the constants $\Cbndot_0, \, \Cbpndot ,\, \Mndot_0, \,\,\Mpndot$.
{\bf Step 1.} \textit{Boundedness of the iterates.} The ninetets $\NN^{(n)}$ are uniformly bounded
\bea
\lab{eq:nintetbound}
\big\| \NN^{(n)} \big\|_{s_{max}+1} &\les&\dg,
\eea
where \eqref{eq:nintetbound} is an immediate consequence of the estimates \eqref{eq:ThmGCMS1}, \eqref{eq:ThmGCMS2} and \eqref{eq:ThmGCMS4} in Theorem \ref{Theorem:ExistenceGCMS1}.
{\bf Step 2.} \textit{Contraction of the iterates.} Since we cannot compare directly
the ninetets $\NN^{(n)}$ we compare instead their pull-back by $\Psi(n)$ to $\ovS$ i.e.
\bea
\NN^{n,\#}:=\Big(U^{(n)}, S^{(n)}, \ovla^{n,\#}, f^{n,\#}, \fb^{n,\#}; \, \Cbn_0, \Mn_0, \Cbpn, \Mpn\Big)
\eea
where $\ovla^{n,\#}, f^{n,\#}, \fb^{n,\#}$ are the pull-backs by the map $\Psi(n)$ of the triplet $\ovla^{(n)}, f^{(n)}, \fb^{(n)} $
defined on the sphere $\S(n)$.
We also introduce the modified norms
\bea
\bsplit
\big\| \NN^{n,\#} \big\|_{k, \ovS} :&= r^{-1} \Big\| \big(U^{(n)}, S^{(n)}\big)\Big \|_{\hk_{k}(\ovS)} +
\Big\|\big( f^{n,\#}, \fb^{n,\#}, \ovla^{n,\#}\big)\Big\| _{\hk_{k}(\ovS)}\\
&+ r^{2} \left(\big| \Cbndot_0\big|+\sum_p\big| \Cbpndot\big| +r \big| \Mndot_0\big|+
r \sum_p\big| \Mpndot\big|\right).
\end{split}
\eea
In view of the Sobolev norm comparison of Proposition 5.9 in \cite{KS-Kerr1} we deduce from
\eqref{eq:nintetbound}
\bea
\lab{eq:nintetboundmodified}
\big\| \NN^{n,\#} \big\|_{s_{max}+1, \ovS} &\les&\dg.
\eea
Contraction in this modified norms is established in the following.
\begin{proposition}
\lab{Prop:contractionforNN}
The following estimate holds true.
\bea
\|\NN^{n+1,\#}-\NN^{n,\#}\|_{3,\ovS} &\les& \epg \|\NN^{n,\#}-\NN^{n-1,\#}\|_{3,\ovS}.
\eea
\end{proposition}
The proof of Proposition \ref{Prop:contractionforNN} is very similar to that of Proposition 6.5 in \cite{KS-Kerr1}. The difference is in the way our basis of adapted $\ell=1$ modes $J^n$ were defined above, which in turn requires to compare $J^{n+1}$ and $J^n$, see \eqref{eq:intermadiaryestimateforproofcontractionschemeintrinsicGCM} below. To deal with this issue, we rely on the following lemma.
\begin{lemma}\lab{lemma:comparisionofJnplus1andJnforcontractionscheme}
We have,
\beaa
\Big| J^{n+1} \circ \Psi_{n,n+1} - J^n\Big|&\les r^{-1} d_n \big( \gn, g^{(n-1)} \big)
\eeaa
with $\gn$ the metric of $\S(n)$.
\end{lemma}
\begin{proof}
We write,
\beaa
J^{n+1} \circ \Psi_{n, n+1} - J^n&=& \J_{can}^n \circ \Psi_{n+1, n} \circ \Psi_{n, n+1} - J^n= \J_{can}(n)-J(n)\\
&=& \J_{can}^n - \J_{can}^{n-1} \circ \Psi_{n, n-1}.
\eeaa
Thus, in view of Corollary \ref{lemma:comparisonJ2:corollary} applied to the pairs $\big(S(n), \gn\big)$,
$\big(S(n-1), g^{(n-1)} \big)$ and map $\Psi=\Psi_{n, n-1}$,
\beaa
\Big| J^{n+1} \circ \Psi_{n, n+1} - J^n\Big| = \Big| \J_{can}^n - \J_{can}^{n-1} \circ \Psi_{n, n-1} \Big| &\les& r^{-1} d_n \big( \gn, g^{(n-1)} \big)
\eeaa
as stated.
\end{proof}
\begin{proof}[Proof of Proposition \ref{Prop:contractionforNN}]
Proposition 6.5 in \cite{KS-Kerr1} yields
\bea\lab{eq:intermadiaryestimateforproofcontractionschemeintrinsicGCM}
\|\NN^{n+1,\#}-\NN^{n,\#}\|_{3,\ovS} &\les& \epg \|\NN^{n,\#}-\NN^{n-1,\#}\|_{3,\ovS} +\dg\Big| J^{n+1} \circ \Psi_{n, n+1} - J^n\Big|
\eea
where the last term in the RHS comes from comparing the $\ell=1$ modes of $\fnn$, $\fbnn$ with the ones of $\fn$, $\fbn$ using \eqref{GCMS:l=1modesforffb-n}. Using Lemma \ref{lemma:comparisionofJnplus1andJnforcontractionscheme}, we infer
\beaa
\|\NN^{n+1,\#}-\NN^{n,\#}\|_{3,\ovS} &\les& \epg \|\NN^{n,\#}-\NN^{n-1,\#}\|_{3,\ovS} +\dg d_n \big( \gn, g^{(n-1)} \big).
\eeaa
Making use of the expressions of the metrics $\gn$, $g^{(n-1)}$ with respect to the maps $\Psi(n)$, $\Psi(n-1)$, see Lemma 5.2 in \cite{KS-Kerr1}, and using the definition of $\Psi(n)$, $\Psi(n-1)$ respectively in terms of the scalar functions $(U^{(n)}, S^{(n)})$ and $(U^{(n-1)}, S^{(n-1)})$, see Definition \ref{definition:Deformations}, we can then estimate
$d_n \big( \gn, g^{(n-1)} \big)$, similarly to the Proposition 6.5 in \cite{KS-Kerr1}, as follows
\beaa
d_n \big( \gn, g^{(n-1)} \big) &\les& r^{-1} \Big\| \big(U^{(n)}-U^{(n-1)}, S^{(n)}- S^{(n-1)}\big)\Big \|_{\hk_{3}(\ovS)}.
\eeaa
We deduce, using also $\dg\leq\epg$ and the definition of $\NN^{n,\#}$,
\beaa
\|\NN^{n+1,\#}-\NN^{n,\#}\|_{3,\ovS} &\les& \epg \|\NN^{n,\#}-\NN^{n-1,\#}\|_{3,\ovS} +\dg d_n \big( \gn, g^{(n-1)} \big)\\
&\les& \epg \Bigg[\|\NN^{n,\#}-\NN^{n-1,\#}\|_{3,\ovS}\\
&&+r^{-1} \Big\| \big(U^{(n)}-U^{(n-1)}, S^{(n)}- S^{(n-1)}\big)\Big \|_{\hk_{3}(\ovS)}\Bigg]\\
&\les& \epg \|\NN^{n,\#}-\NN^{n-1,\#}\|_{3,\ovS}
\eeaa
as desired.
\end{proof}
%%%%%%%%%%%%%%%%%%%
\subsubsection{Convergence}
%%%%%%%%%%%%%%%%%%%%
In view of \eqref{eq:nintetbound} and Proposition \ref{Prop:contractionforNN}, it follows that the sequence $\Psi(n)$ converges to a deformation map
$\Psi:\ovS \longrightarrow \S$,
$\S(n)\to \S$, $\Psi_{n+1, n}\to I_\S$, $ \J^n_{can} \to \J^\S_{can} $ and $J^n\to J$. In view of \eqref{eq:identityJ^nJ^{n+1}},
\beaa J^{n+1}&= \J^n_{can} \circ \Psi_{n+1, n}.
\eeaa
We deduce $J =\J^\S_{can}$.
The limiting map $\Psi$ thus verifies the desired properties of Theorem
\ref{Theorem:ExistenceGCMS1-2} with respect to a canonical $\ell=1$ basis of $\S$ as desired. This concludes the proof of Theorem \ref{Theorem:ExistenceGCMS1-2}.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Construction of intrinsic GCM spheres}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
In what follows we prove an important corollary of Theorem \ref{theorem:ExistenceGCMS2} which makes use of the arbitrariness of $\La, \Lab$ to ensure the vanishing of the $\ell=1$ modes of $\b$ and $\trchbc$. The result requires stronger assumptions than those made in {\bf A1}, see \eqref{eq:assumtioninRRforGagandGabofbackgroundfoliation}, namely we assume that $\Ga_b$ has the same behavior as $\Ga_g$, i.e.
{\bf A1-Strong.} For $k\le s_{max}$,
\bea
\lab{eq:assumtioninRRforGagandGabofbackgroundfoliation'}
\Big\|(\Ga_g, \Ga_b) \Big\|_{k, \infty} \les \epg r^{-2}.
\eea
\begin{remark}\lab{rmk:remarkonimprovedrdecaybytradingudecay}
In applications to our envisioned proof of the nonlinear stability of Kerr, we plan to apply the result below in spacetime regions where we have $r\sim u$. Such regions
appear naturally when we extend a given GCM admissible spacetime $\MM$, see Theorem M7 in \cite{KS-Schw}, in a neighborhood of its top GCM sphere. The improvement in $r$ for $\Ga_b$ is a consequence of the fact that we can exchange, in such regions, the decay factors in $u$ for decay in $r$.
\end{remark}
We also need to make stronger assumptions concerning the basis of $\ell=1$ modes for the background foliation. These assumptions are also consistent with the what we expect in applications to the nonlinear stability of Kerr.
{\bf A4-Strong.} We assume the existence of a smooth family of $\ell=1$ modes $\Jp:\RR\to\RRR$ verifying the verifying the following properties.
\begin{enumerate}
\item On each sphere $S$ of the background foliation $\RR$, $\Jp$ is $O(r^{-1}\dg)$ close to a canonical
basis of $\ell=1$ modes $\Jp_{can}$, i.e.
\bea
\Big| \Jp - \Jp_{can}\Big| \les \dg r^{-1}.
\eea
\item The $\Jp$ family verifies
\bea
\lab{eq:Jpsphericalharmonics-strong}
\bsplit
\Big(r^2\lap+2\Big) \Jp &= O(\epg r^{-1} ),\qquad p=0,+,-,\\
\frac{1}{|S|} \int_{S} \Jp J^{(q)} &= \frac{1}{3}\de_{pq} +O(\epg r^{-1} ),\qquad p,q=0,+,-,\\
\frac{1}{|S|} \int_{S} \Jp &=O(\epg),\qquad p=0,+,-.
\end{split}
\eea
\item The $\Jp$ family also verifies
\bea
\lab{assumptionA4-strong}
\max_p\sup_{\RR} \Big(|e_3(\Jp)|+|e_4(\Jp)|+|\nab \Jp|\Big) \les \epg r^{-1}.
\eea
\end{enumerate}
\begin{lemma}
\lab{Lemma:ComparisonJ-strong}
Under the assumptions {\bf A1-Strong}, {\bf A2, A3} and {\bf A4-Strong},
consider a $O(\dg)$-deformation $\Psi: \ovS\longrightarrow \S$, with $(U,S)$ as in Definition \ref{definition:Deformations},
verifying the assumption, see \eqref{eq:ThmGCMS4},
\bea
\lab{eq::ComparisonJ-strong1}
\| (U, S)\|_{L^\infty(\ovS)} + r^{-1} \big\|(U, S)\big\|_{\hk_{s_{max}+1}(\ovS)} &\les \dg.
\eea
Assume also that the canonical $\ell=1$ modes of $\ovS$ and $\S$ are calibrated.
Then:
\begin{enumerate}
\item The following comparison estimate between $\JJpS$ and $\JpS$ holds true,
\bea
\max_p\sup_\S \Big|\JJpS-\JpS\Big| \les \dg (r^\S)^{-1}.
\eea
\item Also
\bea
\max_p\sup_\S\Big| \JpS-\Jp \Big| \les \epg\dg (r^\S)^{-1}.
\eea
\end{enumerate}
\end{lemma}
\begin{proof}
Recall that, see \eqref{eq:non-canonicalJpS}, $ \JpS=\Jpov \circ \Psi^{-1}$.
We write,
\beaa
\JJpS-\JpS&=& \JJpS- \Jpov \circ \Psi^{-1} = \JJpS- \Jpov_{can} \circ \Psi^{-1} +\big( \Jpov_{can} -\Jpov\big) \circ \Psi^{-1}.
\eeaa
Hence, in view of the assumption {\bf A4-strong} on $\ovS$,
\beaa
\Big|\JJpS-\JpS\Big|&\les& \Big| \JJpS- \Jpov_{can} \circ \Psi^{-1} \Big|+ \Big| \big( \Jpov_{can} -\Jpov\big) \circ \Psi^{-1}\Big|\\
&\les& \Big| \JJpS- \Jpov_{can} \circ \Psi^{-1} \Big|+\dg r^{-1}.
\eeaa
It remains to estimate $| \JJpS- \Jpov_{can} \circ \Psi^{-1}|$. Making use of the assumption \eqref{eq::ComparisonJ-strong1} and Lemma \ref{lemma:comparison-gaS-ga} we have
\beaa
\big\| \ovg- \Psi^\# (g^\S)\big\|_{L^\infty(\ovS)} +( \ovr)^{-1} \big\| \ovg- \Psi^\# (g^\S)\big\|_{\hk^2(\ovS)}\les \dg \ovr.
\eeaa
We are thus in a position to apply the comparison Lemma \ref{lemma:comparisonJ2} to deduce,
\beaa
\max_p \sup_{\ovS} \Big|\JJpS\circ \Psi- \Jpov_{can}\Big|&\les \dg( \ovr)^{-1}
\eeaa
or, equivalently,
\beaa
\max_p \sup_{\S}\Big| \JJpS- \Jpov_{can} \circ \Psi^{-1} \Big|&\les&\dg (r^\S)^{-1}.
\eeaa
Hence,
\beaa
\Big|\JJpS-\JpS\Big| &\les& \Big| \JJpS- \Jpov_{can} \circ \Psi^{-1} \Big|+\dg r^{-1}\les\dg (r^\S)^{-1}
\eeaa
as stated in the first part of the lemma.
To prove the second part, we appeal to Lemma \ref{Le:Transportcomparison}
applied to the scalar function $F=\Jp$ which yields
\beaa
\|\Jp\circ\Psi -\Jp\|_{L^\infty(\ovS)} &\les& \big \| ( U, S) \big\|_{L^\infty(\ovS)} \sup_{\RR}\Big(|e_3(\Jp)|+|e_4(\Jp)|+|\nab \Jp|\Big)\\
&\les& r^{-1}\epg\dg
\eeaa
where we used \eqref{eq::ComparisonJ-strong1} and \eqref{assumptionA4-strong} in the last inequality. Recalling that $r^\S$, $\ovr$ and $r|_\S$ are comparable, we deduce
$$
\max_p\sup_\S\Big| \JpS-\Jp \Big| = \max_p\sup_\S\Big| \Jp\circ\Psi^{-1}-\Jp \Big|=\max_p\sup_{\ovS}\Big| \Jp\circ\Psi -\Jp \Big|\les (r^\S)^{-1}\epg\dg
$$
as stated.
\end{proof}
We are ready to state and prove our intrinsic GCM Theorem
\ref{theorem:ExistenceGCMS2}.
\begin{theorem}[Existence of intrinsic GCM spheres]
\lab{theorem:ExistenceGCMS2}
Assume that the spacetime region $\RR$ verifies the assumptions {\bf A1-Strong, A2, A3}, {\bf A4-Strong}.
We further assume that, relative to the $\ell=1$ modes of the background foliation,
\bea
\lab{Assumptions:theorem-ExistenceGCMS2}
(\div \b )_{\ell=1}=O( \dg r^{-3}), \qquad (\widecheck{\trch})_{\ell=1}=O(\dg r^{-1}), \qquad (\widecheck{\trchb})_{\ell=1}=O(\dg r^{-1}).
\eea
Then, there exist unique constants $ M^\S_0, \, \MpS$, \, $p\in\{-,0, +\}$ such that
\bea
\lab{def:GCMC2}
\bsplit
\ka^\S&=\frac{2}{r^\S},\\
\kab^\S &=-\frac{2}{r^\S}\Up^\S,\\
\mu^\S&= \frac{2m^\S}{(r^\S)^3} + M^\S_0+\sum _p\MpS \J^{(p, \S)},
\end{split}
\eea
and
\bea
\lab{def:GCMC2-b}
\int_\S \div^\S \b^\S \J^{(p, \S)} =0,
\eea
where $ \J^{(p, \S)} $ is a canonical $\ell=1$ basis for $\S$ calibrated, relative by $\Psi $, with the canonical $\ell=1$ basis of $\ovS$. Moreover the deformation verifies the properties \eqref{eq:ThmGCMS3}, \eqref{eq:ThmGCMS4}, \eqref{eq:ThmGCMS5}, \eqref{eq:ThmGCMS6} stated in Theorem \ref{Theorem:ExistenceGCMS1}.
\end{theorem}
\begin{remark}
\lab{Remark:theorem-ExistenceGCMS2}
The assumptions \eqref{Assumptions:theorem-ExistenceGCMS2} for $(\div \b)_{\ell=1}$ and $(\widecheck{\trch})_{\ell=1}$ holds true in general under the assumptions {\bf A1-A3}. The corresponding assumption for $(\trchbc)_{\ell=1}$
holds true in regions, discussed in Remark \ref{rmk:remarkonimprovedrdecaybytradingudecay}, where $r\sim u$, see the proof of Theorem M4 in \cite{KS-Schw}.
\end{remark}
The proof of Theorem \ref{theorem:ExistenceGCMS2} relies on the following lemma.
\begin{lemma}
\lab{Lemma:GCMS2}
Assume that the spacetime region $\RR$ verifies the assumptions {\bf A1-Strong, A2, A3}, {\bf A4-Strong}. Let $\S=\S^{(\La, \Lab)} $ be a deformation of $\ovS$ as constructed by Theorem \ref{Theorem:ExistenceGCMS1-2} with
\beaa
\int_\S \div^\S f\, \J^{(p, \S)}=\La, \qquad \int_\S \div^\S \fb\, \J^{(p, \S)}=\Lab.
\eeaa
The following identities hold true.
\bea
\lab{eq:ExistenceGCMS2.1}
\bsplit
\La&=\frac{(r^\S)^3}{3 m^\S}\Big[- (\div^\S \b^\S)_{\ell=1} + (\div \b)_{\ell=1} \Big] + F_1(\La, \Lab),\\
\Lab &= \Up^\S \La +\frac{r^\S}{ 3m^\S} \Big[( \kabc^\S)_{\ell=1} + \Up^\S ( \kac)_{\ell=1} - (\kabc)_{\ell=1} \Big]+ F_2(\La, \Lab),
\end{split}
\eea
where $F_1, F_2$ are continuously differentiable functions of $\La, \Lab$ verifying
\bea
\big|F_1, F_2\big| \les \epg \dg, \qquad \big|\pr_\La (F_1, F_2) , \pr_{\Lab} (F_1, F_2)\big|\les \epg.
\eea
\end{lemma}
We postpone the proof of Lemma \ref{Lemma:GCMS2} to section \ref{sec:proofofLemma:GCMS2}.
\begin{proof}[Proof of Theorem \ref{theorem:ExistenceGCMS2}]
We note that under the assumptions of the theorem the system
\beaa
\bsplit
\La&=\frac{(r^\S)^3}{3 m^\S} (\div \b)_{\ell=1} + F_1(\La, \Lab),\\
\Lab &= \Up^\S \La +\frac{r^\S }{3 m^\S} \Big( \Up^\S ( \kac)_{\ell=1} - (\kabc)_{\ell=1} \Big)+ F_2(\La, \Lab),
\end{split}
\eeaa
has a unique solution $\La_0, \Lab_0$ verifying the estimate
\beaa
\big|\La_0\big|+\big| \Lab_0\big| \les \dg.
\eeaa
Therefore, taking $\La=\La_0, \Lab=\Lab_0$ in \eqref{eq:ExistenceGCMS2.1},
we deduce
\beaa
(\div^\S \b^\S)_{\ell=1}=0, \qquad (\kabcS)_{\ell=1}=0.
\eeaa
It remains to check \eqref{def:GCMC2}. According to Theorem
\ref{Theorem:ExistenceGCMS1-2},
there exist constants $\Cb^\S_0$,\, \, $\CbpS$, \, $ M^\S_0$, \, $\MpS, \, p\in\{-,0, +\}$ such that
\beaa
\bsplit
\ka^\S&=\frac{2}{r^\S},\\
\kab^\S &=-\frac{2}{r^\S}\Up^\S+ \Cb^\S_0+\sum_p \CbpS \J^{(p, \S)} ,\\
\mu^\S&= \frac{2m^\S}{(r^\S)^3} + M^\S_0+\sum _p\MpS \J^{(p, \S)}.
\end{split}
\eeaa
Using the two first equations above, the definition of the Hawking mass, and using the fact that $\int_\S \J^{(p, \S)}=0$, we infer
\beaa
\Up^\S&=& 1 -\frac{2m^\S}{r^\S} =-\frac{1}{16 \pi} \int_{\S } \ka^\S \kab^\S=-\frac{1}{8 \pi r^\S} \int_{\S }\left(-\frac{2}{r^\S}\Up^\S+ \Cb^\S_0+\sum_p \CbpS \J^{(p, \S)}\right)\\
&=& \Up^\S -\frac{r^\S}{2}\Cb^\S_0
\eeaa
and hence,
\beaa
\Cb^\S_0&=& 0.
\eeaa
Therefore,
\beaa
\kabc^\S&=& \sum_p \CbpS \J^{(p, \S)}.
\eeaa
Projecting on the canonical $\ell=1$ basis and using the condition $(\kabc^\S)_{\ell=1}=0$
and the property $ \int_S \J^{(p,\S)}\J^{(q,\S)} da_g = \frac{4\pi}{3}(r^\S)^2\de_{pq}+ O(\epg r^\S)$ in \eqref{eq:PropertiesofJ^S-strong2}
we then conclude that $ \CbpS=0$. Hence, we have finally obtained
\beaa
\bsplit
\ka^\S&=\frac{2}{r^\S},\\
\kab^\S &=-\frac{2}{r^\S}\Up^\S,\\
\mu^\S&= \frac{2m^\S}{(r^\S)^3} + M^\S_0+\sum _p\MpS \J^{(p, \S)},
\end{split}
\eeaa
and
\beaa
\int_\S \div^\S \b^\S \J^{(p, \S)} =0
\eeaa
as desired.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Proof of Lemma \ref{Lemma:GCMS2}}
\lab{sec:proofofLemma:GCMS2}
%%%%%%%%%%%%%%%%%%%%%%%%%%
Note first that we are in position to apply the comparison Lemma \ref{Lemma:ComparisonJ-strong} according to which $\max_p\sup_\S \Big|\JJpS-\JpS\Big| \les \dg (r^\S)^{-1}$ and $\max_p\sup_\S\Big| \JpS-\Jp \Big| \les \epg\dg (r^\S)^{-1}$,
and therefore
\bea
\lab{eq:comparsonsJs}
\max_p\sup_\S \Big|\JJpS-\Jp\Big| \les \dg(r^\S)^{-1}.
\eea
Let $(f, \fb, \la)$ denote the transition coefficients between the background frame of $\RR$ and
the frame $(e_3^\S, e_4^\S, e_1^\S, e_2^\S)$ adapted to the deformation $\S$.
{\bf Step 1.} Consider a GCM sphere $\S^{(\La, \Lab)}$ as in Theorem \ref{Theorem:ExistenceGCMS1-2}.
Since $\rhod\in r^{-1}\Ga_g$ we rewrite the transformation \eqref{eq:transfforbandkab2} for $\b$ in the form
\beaa
\b^\S&=&\b +\frac 3 2 f \rho + r^{-1} \Ga_g \c F.
\eeaa
\begin{remark}
Note that all quantities defined on $\S$, such as $r^\S$, $m^\S$, $(f, \fb, \ovla)$ and $\J^{(p, \S)}$ depend in fact continuously on the parameters $\La, \Lab$ even though this dependence is not made explicit.
\end{remark}
Since $|m^\S-m|\les \dg$, $|r^\S-r|\les \dg$, and
\beaa
\rho&=&-\frac{2m}{r^3} +\rhoc= -\frac{2m^\S}{(r^\S)^3} +\rhoc + \left( \frac{2m^\S}{(r^\S)^3} - \frac{2m}{r^3}\right) \\
&=& -\frac{2m^\S}{(r^\S)^3}+ r^{-1}\Ga_g + O(r^{-3} \dg),
\eeaa
we deduce
\beaa
\b^\S+\frac{3m^\S}{(r^\S)^3} f &=&\b + r^{-1}\Ga_g\c F.
\eeaa
Taking the divergence, we infer
\beaa
\div^\S \b^\S+\frac{3m^\S}{(r^\S)^3} \div^\S f &=&\div^\S \b+ \div^\S \big(r^{-1}\Ga_g\c F\big)\\
&=&\div \b + r^{-2} (\dk \Ga_g) \c F + \div^\S \big(r^{-1}\Ga_g\c F\big)\\
&=&\div\b+E_1(\La, \Lab),
\eeaa
with error term of the form
\bea
E_1(\La, \Lab)&:=& r^{-2} (\dk \Ga_g) \c F + \div^\S \big(r^{-1}\Ga_g\c F\big).
\eea
We deduce,
\bea
\lab{identity:badmode-b}
\frac{3m^\S}{(r^\S)^3}\int_\S ( \div^\S f) \J^{(p, \S)}&=&-\int_\S (\div^\S \b^\S) \J^{(p, \S)}+ \int_\S (\div \b ) \J^{(p, \S)}
+\int_\S E_1\J^{(p, \S)}.
\eea
In view of \eqref{eq:comparsonsJs} we have $\big|\J^{(p, \S)} - \Jp\big| \les \dg r^{-1}$.
Therefore, since $\b\in r^{-1}\Ga_g$,
\beaa
\int_\S (\div \b ) \J^{(p, \S)}&=& \int_\S (\div \b ) \Jp + \int_\S (\div \b )\big(\J^{(p, \S)}- \Jp\big)\\
&=& \int_\S (\div \b ) \Jp +O( r^{-3} \epg\dg)\\
&=& \int_{\ovS } (\div \b ) \Jp+\left( \int_\S (\div \b ) \Jp - \int_{\ovS} (\div \b ) \Jp\right) +O( r^{-3} \epg\dg).
\eeaa
To estimate the term
$\int_\S (\div \b ) \Jp - \int_{\ovS} (\div \b ) \Jp$ we appeal to Lemma \ref {Lemma:coparison-forintegrals} applied to $F=(\div \b ) \Jp$. Thus
\beaa
\left|\int_\S F -\int_{\ovS} F\right| &\les& \dg r \sup_{\RR} \Big( |F| + r \big(|e_3 F|+ |e_4 F|+|\nab F| \big)\Big).
\eeaa
Note that in view of the fact that $\div \b = O(\epg r^{-4} )$, $e_3(\div \b) = O(\epg r^{-5})$ and $e_3\Jp =O( \epg r^{-1}) $, we have
\beaa
\big|e_3 F\big|&\les & \big|e_3(\div \b) \big|+ \big|\div \b\big| \big|e_3 \Jp \big|\les \epg r^{-5}.
\eeaa
The other terms are treated similarly, and we infer that
\beaa
\left|\int_\S F -\int_{\ovS} F\right| &\les& \epg \dg r^{-3}.
\eeaa
We deduce
\beaa
\int_\S (\div \b ) \J^{(p, \S)}&=& \int_{\ovS} (\div \b ) \Jp +O( \epg\dg r^{-3} )=(\div \b)_{\ell=1} +O( \epg\dg r^{-3} ).
\eeaa
Since $\int_\S ( \div^\S f) \J^{(p, \S)}=\La$, after gathering all error terms in $F_1=F_1(\La, \Lab)$, we deduce from \ref{identity:badmode-b}
\bea
\La&=& \frac{(r^\S)^3}{3 m^\S}\Big[- (\div^\S \b^\S)_{\ell=1} + (\div \b)_{\ell=1} \Big] + F_1(\La, \Lab)
\eea
where
\beaa
F_1(\La, \Lab)&=& \frac{(r^\S)^3}{3 m^\S} \int_{\S^{\La, \Lab} } E_1(\La, \Lab)\JpS+\frac{(r^\S)^3}{3 m^\S}\left(\int_\S (\div \b ) \J^{(p, \S)} -(\div\b)_{\ell=1} \right).
\eeaa
In view of the above, we easily check that
\beaa
\Big| F_1(\La, \Lab)\Big|\les \epg \dg
\eeaa
as stated.
{\bf Step 2.} We next consider the equations \eqref{eq:transfforbandkab1}
\beaa
\bsplit
\ka^\S &= \ka+ \ka \ovla+\div^\S f +F\c \Ga_b+ F\c\nab^\S F+r^{-1} F^2,\\
\kab^\S &= \kab- \kab \ovla+\div^\S\fb +F\c \Ga_b+ F\c\nab^\S F+r^{-1} F^2.
\end{split}
\eeaa
Since $| F| \les r^{-1} \dg $, and since we have $\big|\Ga_b\big| \les r^{-2}\epg$ in view of {\bf A1-strong}, and using the GCM condition $\ka^\S=2/r^\S$, we deduce
\beaa
\bsplit
\div^\S f+ \ka \ovla &= \frac{2}{r^\S} - \ka +O\big( \epg \dg r^{-3} \big),\\
\div^\S\fb - \kab \ovla &= \kab^\S - \kab +O\big( \epg \dg r^{-3} \big).
\end{split}
\eeaa
Differentiating w.r.t. $\Delta^\S$, and using {\bf A1-strong} for $\ka$ and $\kab$, we infer
\beaa
\bsplit
\Delta^\S\div^\S f+ \ka \Delta^\S\ovla &= - \Delta^\S\ka +O\big( \epg \dg r^{-5} \big),\\
\Delta^\S\div^\S\fb - \kab \Delta^\S\ovla &= \Delta^\S\kab^\S - \Delta^\S\kab +O\big( \epg \dg r^{-5} \big).
\end{split}
\eeaa
This yields
\beaa
\kab\Delta^\S\div^\S f+\ka\Delta^\S\div^\S\fb &=& \ka\Delta^\S\kab^\S -\kab\Delta^\S\ka -\ka\Delta^\S\kab +O\big( \epg \dg r^{-6} \big).
\eeaa
Hence, using the control of $\ka$ and $\kab$ provided by {\bf A1-Strong}, $|m-m^\S|\les\dg$, and $|r-r^\S|\les \dg$, we deduce
\beaa
-\frac{2\Up^\S}{r^\S}\Delta^\S\div^\S f+\frac{2}{r^\S}\Delta^\S\div^\S\fb &=& \frac{2}{r^\S}\Delta^\S\kab^\S +\frac{2\Up^\S}{r^\S}\Delta^\S\ka -\frac{2}{r^\S}\Delta^\S\kab +O\big( \epg \dg r^{-6} \big),
\eeaa
or,
\beaa
\Delta^\S\big(\div^\S\fb -\Up^\S\div^\S f\big) &=& \Delta^\S\kab^\S +\Up^\S\Delta^\S\ka -\Delta^\S\kab +O\big( \epg \dg r^{-5} \big).
\eeaa
Next, we focus on $\Delta^\S\ka$ and $\Delta^\S\kab$. Recall from \eqref{eq:Generalframetransf} that
\beaa
e_a^\S &= \left(\de_{ab} +\frac{1}{2}\fb_af_b\right) e_b +\frac 1 2 \fb_a e_4 +\left(\frac 1 2 f_a +\frac{1}{8}|f|^2\fb_a\right) e_3,\qquad a=1,2.
\eeaa
Recalling the definition of $\Ga_g$ and $\Ga_b$ in \eqref{definition:Ga_gGa_b}, we have $\ka-\frac 2 r =\kac \in \Ga_g$, $\kab+\frac{2\Up}{r}=\kabc \in \Ga_g$ and
\beaa
e_4(r)-1 \in r\Ga_g,\quad e_3(r)+\Up \in r\Ga_b, \quad
e_4(m)\in r\Ga_g, \quad e_3(m)\in r\Ga_b.
\eeaa
Thus using assumption {\bf A1-Strong} and $| F| \les r^{-1} \dg $, we infer that
\beaa
\nab^\S\ka &=& \nab\ka+ \frac{\Up}{r^2}f -\frac{1}{r^2}\fb +O(r^{-4}\epg\de),\\
\nab^\S\kab &=& \nab\kab - \frac{\Up\left(1-\frac{4m}{r}\right)}{r^2}f +\frac{1-\frac{4m}{r}}{r^2}\fb +O(r^{-4}\epg\dg).
\eeaa
Taking the divergence, and using {\bf A1-strong} for $\ka$ and $\kab$, $|m-m^\S|\les\dg$ and $|r-r^\S|\les \dg$, we deduce
\beaa
\Delta^\S\ka &=& \Delta\ka+ \frac{\Up^\S}{(r^\S)^2}\div^\S f -\frac{1}{(r^\S)^2}\div^\S\fb +O(r^{-5}\epg\dg),\\
\Delta^\S\kab &=& \Delta\kab - \frac{\Up^\S\left(1-\frac{4m^\S}{r^\S}\right)}{(r^\S)^2}\div^\S f +\frac{1-\frac{4m^\S}{r^\S}}{(r^\S)^2}\div^\S\fb +O(r^{-5}\epg\dg).
\eeaa
We introduce the notations
\beaa
h &:=& \fb-\Up^\S f.
\eeaa
and rewrite the above equations in the form
\beaa
\Delta^\S\big(\div^\S h\big) &=& \Delta^\S\kab^\S +\Up^\S\Delta^\S\ka -\Delta^\S\kab +O\big( \epg \dg r^{-5} \big),
\eeaa
and
\beaa
\Delta^\S\ka &=& \Delta\ka -\frac{1}{(r^\S)^2}\div^\S h +O(r^{-5}\epg\dg),\\
\Delta^\S\kab &=& \Delta\kab +\frac{1-\frac{4m^\S}{r^\S}}{(r^\S)^2}\div^\S h +O(r^{-5}\epg\dg).
\eeaa
We infer that
\beaa
\left(\Delta^\S + \frac{2}{(r^\S)^2}\right)\div^\S h - \frac{6m^\S}{(r^\S)^3}\div^\S h &=& \Delta^\S\kab^\S +\Up^\S\Delta\ka -\Delta\kab +O\big( \epg \dg r^{-5} \big).
\eeaa
Projecting over the basis of canonical $\ell=1$ modes $\J^{(p, \S)} $, integrating by parts and using, see \eqref{eq:PropertiesofJ^S-strong2},
\beaa
\left(\lap^\S +\frac{2}{(r^\S)^2 } \right) \J^{(p, \S)}= O\left(\frac{\epg}{(r^\S)^3 } \right)\J^{(p, \S)},
\eeaa
we deduce,
\beaa
-\frac{ 6 m^\S}{(r^\S)^3} \int_\S \div^\S h \J^{(p, \S)}+ \frac{O(\epg)}{(r^\S)^3 }\int_\S \div^\S h \J^{(p, \S)} = \int_\S \Big(\Delta^\S\kab^\S +\Up^\S\Delta\ka -\Delta\kab\Big) \J^{(p, \S)} + O( \epg \dg r^{-3})
\eeaa
i.e.
\beaa
\frac{ 6 m^\S }{(r^\S)^3} \int_\S \div^\S h \J^{(p, \S)}&=& -\int_\S \Big(\Delta^\S\kab^\S +\Up^\S\Delta\ka -\Delta\kab\Big) \J^{(p, \S)} + O( \epg \dg r^{-3})
\eeaa
or,
\bea
\lab{eq:l-1mode-ofdivh}
\int_\S \div^\S h \J^{(p, \S)}= -\frac{(r^\S)^3 }{ 6m^\S} \int_\S \Big(\Delta^\S\kab^\S +\Up^\S\Delta\ka -\Delta\kab\Big) \J^{(p, \S)}+O( \epg \dg).
\eea
Next, we focus on the RHS of \eqref{eq:l-1mode-ofdivh}. We have
\beaa
\int_\S\Delta\ka \J^{(p, \S)} &=& \int_\S\Delta\ka \Jp +\int_\S\Delta\ka (\J^{(p, \S)} -\Jp)\\
&=& \int_{\ovS}\Delta\ka \Jp +\left(\int_\S \Delta\ka \Jp - \int_{\ovS}\Delta\ka \Jp\right) +\int_\S\Delta\ka (\J^{(p, \S)} -\Jp).
\eeaa
Thus, using \eqref{eq:comparsonsJs}, {\bf A1-Strong} for $\ka$, and Lemma \ref {Lemma:coparison-forintegrals} applied to $F=(\Delta\ka ) \Jp$, we infer, proceeding as in Step 1,
\beaa
\int_\S\Delta\ka \J^{(p, \S)} &=& \int_{\ovS}\Delta\ka \Jp +O(r^{-3}\epg\dg) = -\frac{2}{r^2}(\kac)_{\ell=1} +O(r^{-3}\epg\dg).
\eeaa
Similarly, we obtain
\beaa
\int_\S\Delta\kab \J^{(p, \S)} &=& -\frac{2}{r^2}(\kabc)_{\ell=1} +O(r^{-3}\epg\dg).
\eeaa
This yields
\beaa
\int_\S \Big(\Delta^\S\kab^\S +\Up^\S\Delta\ka -\Delta\kab\Big) \J^{(p, \S)} &=& \int_\S \Delta^\S\kab^\S \J^{(p, \S)} +\Up^\S\int_\S \Delta\ka \J^{(p, \S)} - \int_\S\Delta\kab \J^{(p, \S)}\\
&=& -\frac{2}{(r^\S)^2}\int_\S \kabc^\S \J^{(p, \S)} -\frac{2\Up^\S}{r^2}(\kac)_{\ell=1} +\frac{2}{r^2}(\kabc)_{\ell=1} +O(r^{-3}\epg\dg).
\eeaa
Since $|r-r^\S|\les \dg$, we infer
\beaa
\int_\S \Big(\Delta^\S\kab^\S +\Up^\S\Delta\ka -\Delta\kab\Big) \J^{(p, \S)} &=& -\frac{2}{(r^\S)^2}(\kabc^\S)_{\ell=1} -\frac{2\Up^\S}{(r^\S)^2}(\kac)_{\ell=1} +\frac{2}{(r^\S)^2}(\kabc)_{\ell=1} +O(r^{-3}\epg\dg).
\eeaa
Together with \eqref{eq:l-1mode-ofdivh}, we deduce
\beaa
\int_\S \div^\S h \J^{(p, \S)}= \frac{r^\S }{ 3m^\S}\Big((\kabc^\S)_{\ell=1} +\Up^\S(\kac)_{\ell=1} -(\kabc)_{\ell=1} \Big)+O( \epg \dg).
\eeaa
Recalling $h= \fb-\Up^\S f$ and the definition of $\La, \Lab$, we deduce,
\beaa
\Lab-\Up^\S \La &=&\frac{ r^\S }{3m^\S} \Big( ( \kabc^\S)_{\ell=1} + \Up^\S ( \kac)_{\ell=1} - (\kabc)_{\ell=1} \Big)+ F_2(\La, \Lab)
\eeaa
with
\beaa
|F_2(\La, \Lab)| &\les& \epg \dg
\eeaa
as stated. This concludes the proof of Lemma \ref{Lemma:GCMS2}.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Definition of angular momentum}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
The result of Theorem \ref{theorem:ExistenceGCMS2} is unique up to a rotation
of $\SSS^2$ in the definition of the canonical $\ell=1$ modes on $\S$, see Definition \ref{definition:canmodesS}. We can remove this final ambiguity
by choosing that rotation such that the $p=\pm$ component of $(\curl^\S \b^\S)_{\ell=1}$ vanish. We state this result in the following corollary
of Theorem \ref{theorem:ExistenceGCMS2}.
\begin{corollary}
\lab{Corr:ExistenceGCMS2}
Under the same assumptions as in Theorem \ref{theorem:ExistenceGCMS2} we have, in addition to \eqref{def:GCMC2} and \eqref{def:GCMC2-b},
\begin{itemize}
\item either, for any choice of a canonical ${\ell=1}$ basis of $\S$,
\beaa
(\curl^\S \b^\S)_{\ell=1}=0,
\eeaa
\item or there exists a unique $\ell=1$ basis of canonical modes of $\S$ such that
\bea
\lab{angular-momentum}
\int_\S \curl^\S \b^\S\, \J^{(\pm, \S)}=0, \qquad \int_\S \curl^\S \b^\S\, \J^{(0, \S)}\neq 0.
\eea
\end{itemize}
We then define the angular parameter $a^\S$ on $\S$ by the formula\footnote{Note that in a Kerr space $\KK(a, m)$, relative to a geodesic foliation normalized on $\II^+$, we have $ \int_\S \curl^\S \b^\S \J^{\pm, S}=0$ and $ \int_\S \curl^\S \b^\S \J^{0, S} =\frac{8\pi a m}{ (r^\S)^3}+O(\frac{ma^2}{ (r^\S)^4}) $. }
\bea
a^\S:=\frac{(r^\S)^3}{8\pi m^\S} \int_\S \curl^\S \b^\S \J^{(0, \S)}.
\eea
With this definition, we have $a^\S=0$ in the first case, while $a^\S\neq 0$ in the second case.
\end{corollary}
\begin{proof}
Recall that the basis of canonical $\ell=1$ modes on $\S$ are given by the formula
\beaa
\JpSS&=&J^{(p, \SSS^2)} \circ \Phi^{-1}
\eeaa
with $J^{(p, \SSS^2)}$ the $\ell=1 $ basis on $\SSS^2$ given by Definition \ref{definition:ell=1mpdesonS}, and $(\Phi, \phi) $ an effective uniformization map of $\S$. In particular, we have by change of variable
\beaa
\int_\S \curl^\S \b^\S\, \J^{(p, \S)} &=& \int_{\SSS^2}e^{2\phi}\,\curl^\S \b^\S\circ\Phi\, J^{(p, \SSS^2)}
\eeaa
and hence
\beaa
(\curl^\S \b^\S)_{\ell=1} &=& \int_{\SSS^2}e^{2\phi}\,\curl^\S \b^\S\circ\Phi\, x
\eeaa
where $x$ is the position vector on $\SSS^2$. Changing the definition of $\ell=1$ modes by a rotation of $\SSS^2$ consist in changing $(\Phi, \phi) $ to $(\Phi\circ O, \phi\circ O)$ with $O\in O(3)$ which yields
\beaa
\int_{\SSS^2}e^{2\phi\circ O}\,\curl^\S \b^\S\circ\Phi\circ O\, x &=& \int_{\SSS^2}e^{2\phi}\,\curl^\S \b^\S\circ\Phi\, O^{-1}x\\
&=& O^{-1}\left(\int_{\SSS^2}e^{2\phi}\,\curl^\S \b^\S\circ\Phi\, x\right)\\
&=& O^{-1}(\curl^\S \b^\S)_{\ell=1}.
\eeaa
In particular, $(\curl^\S \b^\S)_{\ell=1}$ is identified with a vector $v$ in $\RRR^3$, and changing the definition of $\ell=1$ modes by a rotation of $\SSS^2$ amounts to apply a rotation to $v$. The proof follows then from the fact that, given a vector $v$ in $\RRR^3$ with $v\neq 0$, there exists a unique rotation of the $(x^1, x^2, x^3)$ coordinates of $\RRR^3$ such that $v$ points in the direction of the $x^3$ axis.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Intrinsic GCM spheres in Kerr}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Consider $\KK(a_0, m_0)$, $|a_0|< m_0$, a sub-extremal Kerr spacetime
endowed with an outgoing optical function $u$ normalized at $\II^+$. We denote by
$S(u, s) $ the spheres of the induced geodesic foliation, with $s$ the affine parameter, and $r$ the area radius, normalized on $ \II^+$ such that $\lim_{r\to \infty} \frac{s}{r} =1$. Let $(e_4, e_3, e_2, e_2)$ be the associated null frame with $e_4 =-\g^{\a\b}\pr_\b u\pr_\a$. Define also the corresponding angular coordinates $(\th, \vphi)$, properly normalized at infinity, with $e_4(\th)=e_4(\vphi)=0$, and the corresponding $\Jp$ defined by them, i.e.
\bea
J^{(0)}=\cos \th, \qquad J^{(+)}=\sin\th \cos\vphi,\qquad J^{(-)}= \sin\th \sin\vphi.
\eea
Finally we consider the spacetime region
\beaa
\RR(r_0) = \Big\{ r\ge r_0\Big\} \subset \KK(a_0, m_0).
\eeaa
\begin{lemma}
\lab{lemma:controlfarspacetimeregionKerrassumptionRR}
If $r_0=r_0(m_0)$ is sufficiently large, the region $\RR(r_0)$, endowed with the geodesic foliation described above, verifies the assumptions {\bf A1-Strong}, {\bf A2, A3} and {\bf A4-Strong}, as well as \eqref{Assumptions:theorem-ExistenceGCMS2}, with the smallness constants
\beaa
\epg= \dg= \frac{a_0m_0}{r_0}.
\eeaa
\end{lemma}
\begin{proof}
{\bf A1-Strong}, {\bf A2, A3} and {\bf A4-Strong}, as well as the estimate for $(\div\b)_{\ell=1}$ in \eqref{Assumptions:theorem-ExistenceGCMS2}, follow immediately from Lemma 2.10 in \cite{KS-Kerr1}. It then remains to prove \eqref{Assumptions:theorem-ExistenceGCMS2} for $(\kac)_{\ell=1}$ and $(\kabc)_{\ell=1}$. Lemma 2.10 in \cite{KS-Kerr1} yields $\kac=O(a_0^2r^{-3})$ and $\kabc=O(a_0^2r^{-3})$. One can easily push the asymptotic to the next order to obtain
\beaa
\kac = \frac{a_0^2}{r^3}\Big(c_1+c_2(\cos\th)^2\Big)+O\left(\frac{m_0a_0^2}{r^4}\right), \qquad \kabc = \frac{a_0^2}{r^3}\Big(\underline{c}_1+\underline{c}_2(\cos\th)^2\Big)+O\left(\frac{m_0a_0^2}{r^4}\right),
\eeaa
for some universal constants $c_1$, $c_2$, $\underline{c}_1$ and $\underline{c}_2$. Since
\beaa
\Big(c_1+c_2(\cos\th)^2\Big)_{\ell=1}=0, \qquad \Big(\underline{c}_1+\underline{c}_2(\cos\th)^2\Big)_{\ell=1}=0,
\eeaa
we infer
\beaa
(\kac)_{\ell=1}=O\left(\frac{m_0a_0^2}{r^2}\right), \qquad (\kabc)_{\ell=1}=O\left(\frac{m_0a_0^2}{r^2}\right),
\eeaa
which concludes the proof of \eqref{Assumptions:theorem-ExistenceGCMS2}.
\end{proof}
\begin{corollary}[Existence of intrinsic GCM spheres in Kerr]
\lab{cor:ExistenceGCMS1inKerr}
If $r_0\gg m_0$ is sufficiently large, then any sphere $\ovS\subset \RR(r_0)$ admits a unique deformation $\Psi:\ovS\longrightarrow \S$ verifying \eqref{def:GCMC2} and \eqref{def:GCMC2-b}. Moreover, the deformation verifies the properties \eqref{eq:ThmGCMS3}, \eqref{eq:ThmGCMS4}, \eqref{eq:ThmGCMS5} and \eqref{eq:ThmGCMS6} stated in Theorem \ref{Theorem:ExistenceGCMS1}.
\end{corollary}
\begin{proof}
In view of Lemma \ref{lemma:controlfarspacetimeregionKerrassumptionRR}, the assumptions {\bf A1-Strong}, {\bf A2, A3} and {\bf A4-Strong}, as well as \eqref{Assumptions:theorem-ExistenceGCMS2}, are satisfied by the spacetime region $\RR(r_0)$ provided that $r_0=r_0(m_0)$ is sufficiently large, with the smallness constants $\epg$ and $\dg$ given by
\beaa
\epg= \dg= \frac{a_0m_0}{r_0}.
\eeaa
Thus, Theorem \ref{theorem:ExistenceGCMS2} applies which concludes the proof of the corollary.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{thebibliography}{99}
\bibitem{Berger} M. Berger, \textit{Geometry II}, Universitext, Springer, 1987.
\bibitem{CY1} S.-Y. A. Chang, P. Yang, \textit{Prescribing Gaussian curvature on $S^2$}, Acta Math., 159 (1987), 215-259.
\bibitem{CY2} S.-Y. A. Chang, P. Yang, \textit{A perturbation result on prescribing scalar curvature on $S^n$}, Duke Math. J. 64 (1991), 27-69.
\bibitem{Chang} S.-Y. A. Chang, \textit{The Moser-Trudinger inequality and applications to some problems in conformal geometry}, IAS/ Park City Mathematical series,
1996, 65-125.
\bibitem{Chen} P.-N. Chen, M.-T. Wang, S.-T. Yau, \textit{Quasilocal angular momentum and center of mass in general relativity}, Adv. Theor. Math. Phys. 20, (2016), 671-682.
\bibitem{Ch-Kl} D. Christodoulou, S. Klainerman, \textit{The global nonlinear stability of the Minkowski space}, Princeton University Press (1993).
\bibitem{Cia} P. G. Ciarlet, \textit{On Korn's inequality}, Chinese Ann. Math., Ser B 31 (2010), pp. 607-618.
\bibitem{CoSc} S. Conti, B. Schweizer, \textit{Rigidity and Gamma convergence for solid-solid phase transition with $SO(2)$ invariance}, Comm. Pure Appl. Math. 59, 830-868, 2006.
\bibitem{FJM} G. Friesecke, R. James, S. M\"uller, \textit{A Theorem on Geometric Rigidity and the Derivation of Nonlinear Plate Theory from Three-Dimensional Elasticity}, Comm. Pure Appl. Math. 55, 1461-1506, 2002.
\bibitem{GeKSz} E. Giorgi, S. Klainerman, J. Szeftel, \textit{A general formalism for the stability of Kerr}, in preparation.
\bibitem{HuYau} G. Huisken, S.-T. Yau, \textit{Definition of center of mass for isolated physical systems and unique foliations by stable spheres with constant mean curvature.} Invent. Math. 124, 281-311 (1996).
\bibitem{John} F. John. \textit{Rotation and Strain} Comm. Pure Appl. Math, 44, 391-413 (1961).
\bibitem{JoSi} G. Jones, D. Singerman, \textit{Complex functions - An algebraic and geometric viewpoint}, Cambridge University Press, Cambridge, 1987.
\bibitem{KS-Schw} S. Klainerman, J. Szeftel, \textit{Global nonlinear stability of Schwarzschild spacetime under polarized perturbations}, arXiv:1711.07597. Submitted to Annals of Math. Studies.
\bibitem{KS-Kerr1} S. Klainerman, J. Szeftel, \textit{Construction of GCM spheres in perturbations of Kerr}, arXiv:1911.00697.
\bibitem{DeLellis} Camillo De Lellis, \textit{Personal communication}.
\bibitem{Onofri} E. Onofri, \textit{On the positivity of the effective action in a theory of random surfaces}, Comm. Math. Phys. 86 (1982), 321-326.
\bibitem{Rizzi} A. Rizzi, \textit{Angular Momentum in General Relativity: A new Definition}, Phys. Rev. Letters, vol 81, no6, 1150-1153, 1998.
\bibitem{Sz} L. B. Szabados, \textit{Quasi-Local Energy-Momentum and Angular Momentum in General Relativity}, Living Rev. Relativity, 12, (2009), 4.
\end{thebibliography}
\end{document}