\documentclass[leqno]{article}

\usepackage[frenchb,english]{babel}
%\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage{amsfonts}
\usepackage{enumerate}
\usepackage{vmargin}
\usepackage[all]{xy}
\usepackage{mathrsfs}
\usepackage{mathtools}
\usepackage{lmodern}
\usepackage{slashed}
\usepackage[colorlinks=true,linkcolor=blue,pagebackref=true]{hyperref}%
%\usepackage{comment}
\setmarginsrb{3cm}{3cm}{3.5cm}{3cm}{0cm}{0cm}{1.5cm}{3cm}
%\setmarginsrb{1}  {2}   {3}   {4}  {5}  {6}  {7}   {8}
%1 est la marge gauche
%2 est la marge en haut
%3 est la marge droite
%4 est la marge en bas
%5 fixe la hauteur de l'entête
%6 fixe la distance entre l'entête et le texte
%7 fixe la hauteur du pied de page
%8 fixe la distance entre le texte et le pied de page
%\setmarginsrb{3cm}{3cm}{2.5cm}{2cm}{0cm}{0cm}{1.5cm}{4cm}
\usepackage{comment}


\usepackage{accents}
%\begin{document}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%doublehat


\newlength{\dhatheight}
\newcommand{\doublehat}[1]{%
    \settoheight{\dhatheight}{\ensuremath{\hat{#1}}}%
    \addtolength{\dhatheight}{-0.2ex}%
    \hat{\vphantom{\rule{1pt}{\dhatheight}}%
    \smash{\hat{#1}}}}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\footskip1.3cm

\newcommand{\A}{\ensuremath{\mathcal{A}}}
\newcommand{\Cbb}{\ensuremath{\mathbb{C}}}
\newcommand{\C}{\mathrm{C}}
%\newcommand{\D}{\ensuremath{\mathcal{D}}}
\newcommand{\D}{\mathrm{D}}
\newcommand{\E}{\ensuremath{\mathbb{E}}}
\newcommand{\F}{\mathcal{F}}
\newcommand{\N}{\ensuremath{\mathbb{N}}}
\newcommand{\B}{\mathrm{B}} %Bounded operators
\let\H\relax %\H ne fait plus rien !
\newcommand{\H}{\mathrm{H}}
\newcommand{\I}{\mathrm{I}} %matrice identit\'e
\newcommand{\II}{\mathrm{II}} %type II
\newcommand{\K}{\ensuremath{\mathbb{K}}}
\newcommand{\KP}{\ensuremath{\mathbb{KP}}}
\let\L\relax %\L ne fait plus rien !
\newcommand{\L}{\mathrm{L}}
\let\O\relax %\O ne fait plus rien !
\newcommand{\J}{\mathcal{J}} %involution
\newcommand{\O}{\ensuremath{\mathbb{O}}}
\let\P\relax %\P ne fait plus rien !
%\newcommand{\P}{\mathcal{P}} %algebre des poly trigo
\newcommand{\P}{\mathsf{P}} %Peirce
\newcommand{\Q}{\mathrm{Q}}
\newcommand{\IV}{\mathrm{IV}} %type IV
\newcommand{\V}{\mathrm{V}} %type V
\newcommand{\VI}{\mathrm{VI}} %type VI
\newcommand{\SG}{\mathrm{SG}} %Sierpinski gasket
\newcommand{\BMO}{\mathrm{BMO}} %Bounded mean oscillation
\newcommand{\Mult}{\mathrm{Mult}}
\newcommand{\scr}{\mathscr}
\newcommand{\SU}{\mathrm{SU}}
\newcommand{\Tron}{\mathcal{T}}
\newcommand{\SO}{\mathrm{SO}}
\newcommand{\Aut}{\mathrm{Aut}} %automorphism
\newcommand{\QG}{\ensuremath{\mathbb{G}}}
\newcommand{\M}{\mathrm{M}}
\newcommand{\CAR}{\mathrm{CAR}} %CAR algebra
\newcommand{\CB}{\mathrm{CB}} %completely bounded
\newcommand{\diag}{\mathrm{diag}} %diagonal
\newcommand{\RCD}{\mathrm{RCD}}
\newcommand{\dist}{\mathrm{d}}

\newcommand{\meas}{\mu}
\newcommand{\CC}{\mathrm{CC}}
\newcommand{\HS}{\mathrm{HS}}
\newcommand{\absco}{\mathrm{absco}} %absolute convex hull
\newcommand{\NP}{\mathrm{NP}}
\newcommand{\PPT}{\mathrm{PPT}}
\newcommand{\inv}{\diamond}
\newcommand{\Mat}{\ensuremath{\mathbb{M}}}
%
\let\div\relax
\newcommand{\div}{\mathrm{div}}
\let\cal\relax
\newcommand{\cal}{\mathcal}
\newcommand{\conv}{\mathrm{conv}}
\newcommand{\Z}{\ensuremath{\mathbb{Z}}}
\newcommand{\Zc}{\mathrm{Z}}
\newcommand{\R}{\ensuremath{\mathbb{R}}}
%\newcommand{\C}{\mathrm{C}}
\newcommand{\T}{\ensuremath{\mathbb{T}}}
\newcommand{\U}{\mathcal{U}}
\newcommand{\W}{\mathrm{W}}
\newcommand{\sgn}{\mathrm{sgn}}
\let\j\relax %\ker ne fait plus rien !
\newcommand{\j}{\frak{j}} 
\let\S\relax %\ker ne fait plus rien !
\newcommand{\S}{\mathrm{S}} %espace des états
\newcommand{\vect}{\ensuremath{\mathop{\rm Span\,}\nolimits}}
%\newcommand{\tr}{\ensuremath{\mathop{\rm Tr}\nolimits}}
\newcommand{\clos}{\ensuremath{\mathop{\rm closure\,}\nolimits}}
\newcommand{\Id}{\mathrm{Id}}
\newcommand{\PL}{\mathrm{PL}}
\newcommand{\VN}{\mathrm{VN}} %von Neumann
\newcommand{\Asym}{\mathrm{Asym}} %matrices antisymetriques
\newcommand{\Sym}{\mathrm{Sym}} %Symmetric
\newcommand{\Reg}{\mathrm{Reg}} %regular
\newcommand{\la}{\langle}
\newcommand{\ra}{\rangle}
\newcommand{\EA}{\mathrm{EA}} %entanglement assisted
\newcommand{\CPAP}{\mathrm{CPAP}}
\newcommand{\CD}{\mathrm{CD}}
\newcommand{\flip}{\mathrm{flip}}
\newcommand{\CBAP}{\mathrm{CBAP}}
\newcommand{\MK}{\mathrm{MK}} %Monge-Kantorovitch distance
\newcommand{\CCAP}{\mathrm{CBAP}}
\newcommand{\ccb}{\mathrm{ccb}} %co-completely bounded
\newcommand{\cocb}{\mathrm{cocb}}
\renewcommand{\leq}{\ensuremath{\leqslant}}
\renewcommand{\geq}{\ensuremath{\geqslant}}
\newcommand{\qed}{\hfill \vrule height6pt  width6pt depth0pt}
\newcommand{\bnorm}[1]{ \big\| #1  \big\|}
\newcommand{\Bnorm}[1]{ \Big\| #1  \Big\|}
\newcommand{\bgnorm}[1]{ \bigg\| #1  \bigg\|}
\newcommand{\Bgnorm}[1]{ \Bigg\| #1  \Bigg\|}
\newcommand{\norm}[1]{\left\Vert#1\right\Vert}
\newcommand{\Wa}{\mathcal{W}} %Walsch
\newcommand{\mk}{\medskip}
\newcommand{\xra}{\xrightarrow}
\newcommand{\co}{\colon}
\newcommand{\otp}{\widehat{\ot}}
\newcommand{\otpb}{\hat{\ot}}
\newcommand{\ot}{\otimes}
\newcommand{\ovl}{\overline}
\newcommand{\otvn}{\ovl\ot}
%\DeclareMathOperator{\sgn}{\mathrm{sgn}} %signature
\newcommand{\ul}{\mathcal{U}}
\newcommand{\dec}{\mathrm{dec}}
\newcommand{\pdec}{\mathrm{pdec}}
\newcommand{\reg}{\mathrm{reg}}
\newcommand{\Dec}{\mathrm{Dec}}
\newcommand{\cb}{\mathrm{cb}}
\newcommand{\fin}{\mathrm{fin}} %nombre de coefficients fini
\newcommand{\TRO}{\mathrm{TRO}}
\newcommand{\Ch}{\mathrm{Ch}}
\newcommand{\PDec}{\mathrm{PDec}}
\newcommand{\QWEP}{\mathrm{QWEP}}
\newcommand{\dsp}{\displaystyle}
\newcommand{\op}{\mathrm{op}} %opposed structure
\let\i\relax %\i ne fait plus rien !
\newcommand{\i}{\mathrm{i}}
\let\j\relax %\j ne fait plus rien !
\newcommand{\j}{\mathrm{j}} %j quaternions
\let\k\relax %\k ne fait plus rien !
\newcommand{\k}{\mathrm{k}} %k quaternions
\newcommand{\loc}{\mathrm{loc}}
\newcommand{\ov}{\overset}
\newcommand{\sa}{\mathrm{sa}}
\newcommand{\JW}{\mathrm{JW}}
\newcommand{\JC}{\mathrm{JC}}
\newcommand{\JB}{\mathrm{JB}}
\newcommand{\kR}{\mathfrak{R}}
\newcommand{\JBW}{\mathrm{JBW}}
\newcommand{\Vol}{\mathrm{Vol}}
\newcommand{\fix}{\mathrm{fix}}
\newcommand{\ORNP}{\mathrm{ORNP}}%Radon-Nykodym operator space
\newcommand{\MD}{\mathrm{MD}} %multiplicative domain
\newcommand{\LMD}{\mathrm{LMD}} %left multiplicative domain
\newcommand{\RMD}{\mathrm{RMD}} %left multiplicative domain
\newcommand{\w}{\mathrm{w}} %weak
\newcommand{\wo}{\mathrm{wo}} %weak operator
\newcommand{\leftsupp}{{\mathrm{leftsupp}}}
\newcommand{\epsi}{\varepsilon}
\renewcommand{\d}{\mathop{}\mathopen{}\mathrm{d}} %op\'erateur diff\'erentiel
\newcommand{\e}{\mathrm{e}} %constante e
\renewcommand{\d}{\mathop{}\mathopen{}\mathrm{d}} 
\newcommand{\bra}[1]{\langle{#1}|}
\newcommand{\ket}[1]{|{#1}\rangle}
\newcommand{\Pol}{\mathrm{Pol}} %Espce des polynomes groupes quantiques

\def\kb{ \ket \bra }
\newcommand{\proj}[1]{ | #1 \kb  #1|}
\DeclareMathOperator{\Span}{span} %sev engendre
\DeclareMathOperator{\Lip}{\mathrm{Lip}} %Lipschitz
\DeclareMathOperator{\lip}{\mathrm{lip}} %lipschitz
\DeclareMathOperator{\aut}{Aut} %automorphisme
\DeclareMathOperator{\supp}{supp} %support
\DeclareMathOperator{\Fix}{Fix} %fixed subspace
\DeclareMathOperator{\tr}{Tr} %trace
\DeclareMathOperator{\Tr}{Tr} %trace
\let\ker\relax %\ker ne fait plus rien !
\DeclareMathOperator{\ker}{Ker} %range
\DeclareMathOperator{\Ran}{Ran} %range
\DeclareMathOperator{\dom}{dom} %domaine
\DeclareMathOperator{\rank}{rank} %domaine
\let\Re\relax %\Re ne fait plus rien
\DeclareMathOperator{\Re}{Re} %partie r\'eelle
\let\Im\relax %\Im ne fait plus rien
\DeclareMathOperator{\Im}{Im} %partie imaginaire
\DeclareMathOperator{\card}{card} %cardinal
\DeclareMathOperator{\Irr}{Irr} %Irreductible
\DeclareMathOperator{\Rep}{Rep} %Irreductible
\DeclareMathOperator{\Ind}{Ind} %induced
\DeclareMathOperator*{\esssup}{esssup} 
\DeclareMathOperator*{\Lie}{Lie} 
\DeclareMathOperator{\diam}{diam}
\DeclareMathOperator{\Sp}{Sp}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% presentation
\selectlanguage{english}
\newtheorem{thm}{Theorem}[section]
\newtheorem{defi}[thm]{Definition}
\newtheorem{prop}[thm]{Proposition}
\newtheorem{conj}[thm]{Conjecture}
\newtheorem{quest}[thm]{Question}
\newtheorem{Property}[thm]{Property}
\newtheorem{cor}[thm]{Corollary}
\newtheorem{lemma}[thm]{Lemma}
\newtheorem{fact}[thm]{Fact}
\newtheorem{remark}[thm]{Remark}
\newtheorem{example}[thm]{Example}
\newtheorem{prob}[thm]{Problem}


\newenvironment{proof}[1][]{\noindent {\it Proof #1} : }{\hbox{~}\qed
\smallskip
}

%\newenvironment{proof1.1}[1][]{\noindent {\it Proof of Theorem 1.1 #1} : }{\hbox{~}\qed
%\smallskip
%}
%
%\newenvironment{proof3.2}[1][]{\noindent {\it Proof of Theorem 3.2 #1} : }{\hbox{~}\qed
%\smallskip
%}
\usepackage{tocloft}
\setlength{\cftbeforesecskip}{0pt}
		
\numberwithin{equation}{section}
\usepackage[nottoc,notlot,notlof]{tocbibind}%biblio dans sommaire
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%Biblio: ce paragraphe permet de reduire l'espacement entre les items bibliographiques

\let\OLDthebibliography\thebibliography
\renewcommand\thebibliography[1]{
  \OLDthebibliography{#1}
  \setlength{\parskip}{0pt}
  \setlength{\itemsep}{0pt plus 0.3ex}
}

%reallywidehat
%\usepackage{scalerel}

\newcommand\reallywidehat[1]{\arraycolsep=0pt\relax%
\begin{array}{c}
\stretchto{
  \scaleto{
    \scalerel*[\widthof{\ensuremath{#1}}]{\kern-.5pt\bigwedge\kern-.5pt}
    {\rule[-\textheight/2]{1ex}{\textheight}} %WIDTH-LIMITED BIG WEDGE
  }{\textheight} % 
}{0.5ex}\\           % THIS SQUEEZES THE WEDGE TO 0.5ex HEIGHT
#1\\                 % THIS STACKS THE WEDGE ATOP THE ARGUMENT
\rule{-1ex}{0ex}
\end{array}
}

\begin{document}
\selectlanguage{english}
\title{\bfseries{A modular theory of contractive projections on Banach spaces with applications to noncommutative $\L^p$-spaces}}
\date{}
\author{\bfseries{C\'edric Arhancet}}%
%Coulhon-Varopulos dimension, curvature,and functional calculus
%On the spectral dimension of spectral triples
%associated to Markov semigroups of operators
\maketitle
%Levi geometry, modular deweighting and contractive projections on noncommutative $\mathrm{L}^p$-spaces
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%


\begin{abstract}
We develop a differential-geometric approach to contractive projections on complex Banach spaces based on Levi forms of plurisubharmonic functions. We show that if a projection $P$ contracts a $\mathrm{C}^2$ plurisubharmonic function $\Phi$, then, at any point of its range, the Levi form of $\Phi$ defines a tangent Hilbertian geometry for which $P$ becomes an orthogonal projection. When the range is finite-dimensional and the relevant Levi forms are nondegenerate, comparing the Levi geometries associated with different points leads naturally to positive relative operators and associated one-parameter groups. We apply this framework to tracial noncommutative $\mathrm{L}^4$-spaces. Exploiting the cubic form of the duality mapping, we characterize contractive projections by a Jordan triple orthogonality property and prove that a projection on $\mathrm{L}^4(\mathcal{M})$ is $2$-contractive if and only if it is completely contractive. When the ambient von Neumann algebra is finite-dimensional, we identify the relative operator associated with the left and right tangent forms on the range with a restriction of an inverse square root of a modular operator on a linking von Neumann algebra. This entails that the range of any $2$-contractive projection is completely isometric to a rectangular noncommutative $\mathrm{L}^4$-space associated with a $\mathrm{W}^*$-ternary ring of operators. As a consequence, every such projection is contractively decomposable. Finally, for every $1 < p < \infty$ with $p \neq 2$, we prove that every finite-rank $2$-contractive projection on a tracial noncommutative $\mathrm{L}^p$-space is completely contractive. For $p > 2$, we obtain a weighted ternary description of its range in terms of a finite-dimensional $\mathrm{W}^*$-$\mathrm{TRO}$.
\end{abstract}


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\makeatletter
 \renewcommand{\@makefntext}[1]{#1}
 \makeatother
 \footnotetext{
 %The authors are supported by the research program ANR-18-CE40-0021 (project HASCON).\\
 2020 {\it Mathematics subject classification:}
 46L51, 17C65, 47B01, 46L07.
%  46L51  Noncommutative measure and integration
%  46M35       Abstract interpolation of topological vector spaces [See also 46B70]
% 46L07       Operator spaces and completely bounded maps [See also 47L25]
% 43A22 Homomorphisms and multipliers of function spaces on groups, semigroups, etc.
% 43A15 Lp-spaces and other function spaces on groups, semigroups, etc.
% 47D03 Groups and semigroups of linear operators For nonlinear operators.
% 58B34 	Noncommutative geometry (à la Connes)
% 81P45   	Quantum information, communication, networks (quantum-theoretic aspects)
% 94A40   	Channel models (including quantum) in information and communication theory
%46B70   	Interpolation between normed linear spaces
%46L70   	Nonassociative selfadjoint operator algebras
%17C65   	Jordan structures on Banach spaces and algebras
%46L10 	General theory of von Neumann algebra47B01   	
%47B01 Operators on Banach spaces
%46B20   	Geometry and structure of normed linear spaces
\\
{\it Key words}: contractive projections, Banach spaces, plurisubharmonic functions, noncommutative $\L^p$-spaces, Levi forms, duality mappings, ternary rings of operators, complete contractivity.}
%noncommutative $L^p$-spaces, operator spaces


{
  \hypersetup{linkcolor=blue}
 \tableofcontents
}


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Introduction}
\label{sec:Introduction}

The study of contractive projections and contractively complemented subspaces is one of the classical themes of the isometric theory of Banach spaces. Given a Banach space $X$ and a closed subspace $Y\subset X$, the basic problem is to decide whether there exists a projection $P\co X\to X$ such that
\begin{equation}
P^2=P,\qquad \Ran P=Y,\qquad \norm{P}_{X \to X}
\leq 1,
\end{equation}
and, when such a projection exists, to understand both the intrinsic geometry of $Y$ and the structure of $P$. These two questions are closely related. Indeed, on a smooth Banach space, a subspace can be the range of at most one projection of norm one, see \cite{CoS70}. Since noncommutative $\L^p$-spaces are smooth for $1<p<\infty$, the classification of their contractively complemented subspaces is, in this range, essentially equivalent to the classification of their contractive projections.

For classical $\L^p$-spaces, this problem admits a rigid answer. By the work of Ando \cite{And66}, completed in greater generality by Bernau and Lacey \cite{BeL74}, a contractively complemented subspace of a classical $\L^p$-space, for $1<p<\infty$ and $p\neq2$, is itself isometric to an $\L^p$-space. Moreover, the corresponding projections are described in terms of weighted conditional expectations. Thus, in the commutative setting, norm-one complementation essentially preserves the category of $\L^p$-spaces.

The noncommutative situation is substantially richer. The first obstruction already occurs for Schatten spaces. If $\sigma\co S^p\to S^p$ denotes the transpose map, then
\begin{equation}
P=\frac12(\Id+\sigma)
\end{equation}
is a contractive projection whose range is the space of symmetric matrices. This range is not, in general, a Schatten space. More generally, the seminal work of Arazy and Friedman \cite{ArF92} gives a complete description of contractively complemented subspaces of Schatten spaces in terms of the $\L^p$-versions of Cartan factors of types I--IV. In particular, rectangular operator spaces, symmetric and antisymmetric operator spaces, and spin factors all occur naturally. The same work explicitly raises the problem of describing the contractively complemented subspaces of general noncommutative $\L^p$-spaces.

This problem remains one of the basic structural questions concerning noncommutative $\L^p$-spaces. Its difficulty is not merely technical. The examples above indicate that the range of a contractive projection need not carry an associative structure and that Jordan and ternary structures are unavoidable. At the endpoint $p=1$, this phenomenon is reflected by the appearance of preduals of $\JBW^*$-triples, see \cite{FrB85b}. On the other hand, once the canonical operator space structure of a noncommutative $\L^p$-space is taken into account, rectangular structures become especially natural. In the Schatten setting, Le Merdy, Ricard and Roydor \cite{LRR09} showed that completely $1$-complemented subspaces are precisely, up to the appropriate direct sums, spaces of the form $S^p(H,K)$. These are the $\L^p$-counterparts of ternary rings of operators.

This suggests that the problem of contractive complementation has several levels of rigidity. A scalar norm-one condition may detect a Jordan structure, whereas matricial contractivity should remember the non-symmetrized ternary product. This principle is consistent with a number of results obtained in recent years. The range of a $2$-positive contractive projection on an arbitrary noncommutative $\L^p$-space is completely order and completely isometrically isomorphic to another noncommutative $\L^p$-space \cite{ArR24}; in the $\sigma$-finite case the projection can be viewed as a conditional expectation modulo a symmetric two-sided change of density. Positive contractive projections lead more generally to Jordan conditional expectations and to nonassociative $\L^p$-spaces associated with $\JW^*$-algebras, see \cite{Arh24a,Arh26b}. In a different direction, contractively decomposable projections are naturally related to $\W^*$-ternary rings of operators and to rectangular noncommutative $\L^p$-spaces \cite{Arh24b}. Recent results give a more intrinsic description of their ranges in terms of rectangular $\L^p$-spaces \cite{Bof26,Arh26c}.

The general contractive case, however, is much less understood. In particular, there is no general mechanism which starts only from the metric inequality
\begin{equation}
\norm{P(x)}_{\L^p(\cal{M})}
\leq\norm{x}_{\L^p(\cal{M})}
\end{equation}
and recovers the hidden Jordan or ternary structure of the range. The main purpose of the present paper is to introduce such a mechanism. Our approach is based on the differential geometry of the norm, and more precisely on the Levi forms of plurisubharmonic functions on complex Banach spaces.

Differential methods have appeared before in the study of norm-one projections. In particular, second derivatives of norms were successfully used in \cite{LRG07} to study contractive complementation in vector-valued spaces. Here we take a different point of view. Instead of considering only the real Hessian of the norm, we exploit the complex structure and the associated Levi form. Levi forms on complex Banach spaces were investigated in \cite{Lig76}. The key observation is that a contractive projection produces, at any point of its range, a canonical Hilbertian geometry.

More precisely, let $\Omega$ be an open subset of a complex Banach space $X$, let $\Phi\co\Omega\to\R$ be a $\C^2$ plurisubharmonic function and suppose that a projection $P\co X\to X$ satisfies $P(\Omega) \subset \Omega$ and
\begin{equation}
\Phi(Px)
\leq \Phi(x),\quad x \in \Omega.
\end{equation}
If $a\in\Omega\cap\Ran P$, then the Levi form $\scr L_{\Phi,a}$ satisfies
\begin{equation}
\scr L_{\Phi,a}(Pz)
\leq \scr L_{\Phi,a}(z),\quad z \in X.
\end{equation}
After quotienting by the null space of the Levi form and completing, one obtains a Hilbert space on which $P$ becomes an orthogonal projection. Consequently, although the complement $\Id-P$ need not be contractive for the original Banach norm, both $P$ and $\Id-P$ are contractive for every tangent Levi geometry based at a point of $\Ran P$. Equivalently, the involution
\begin{equation}
2P-\Id
\end{equation}
becomes a unitary involution on each of these tangent Hilbert spaces.


Thus a single Banach space projection gives rise to a whole family of Hilbertian geometries indexed by the points of its range. At first order, contractivity is governed by duality mappings and norming functionals, in accordance with the classical characterization of Calvert \cite{Cal75}. The first-order condition identifies the directions annihilated by the corresponding norming functional. Differentiating this condition along the range yields the orthogonality relations for the Levi form. The role of the second-order viewpoint is therefore not to provide an independent contractivity condition, but to package these relations into a Hilbertian geometry. When the Levi form is positive, the decomposition
\begin{equation}
X
=\Ran P \oplus \ker P
\end{equation}
becomes orthogonal for this tangent geometry, making relative positive operators and their spectral calculus available.

Comparing two such Hilbertian geometries leads naturally to relative positive operators. Suppose first that $\Ran P$ is finite-dimensional and that the restrictions to $\Ran P$ of the Levi forms based at two points $a,b \in \Omega \cap \Ran P$ are nondegenerate. If $\langle\cdot,\cdot\rangle_a$ and $\langle\cdot,\cdot\rangle_b$ denote the corresponding inner products on $\Ran P$, there exists a unique positive invertible operator $\Delta_{b|a}\co\Ran P\to\Ran P$ characterized by
\begin{equation}
\la x,y \ra_b
=\la x,\Delta_{b|a}y \ra_a,
\quad x,y \in \Ran P.
\end{equation}
It is selfadjoint for the first Hilbertian structure, and its imaginary powers form a one-parameter group of isometries for both tangent Hilbertian geometries.

A stronger phenomenon appears when the relative operator can be constructed on a common ambient Hilbert completion. More precisely, if the Levi forms at $a$ and $b$ are nondegenerate on $X$ and define equivalent Hilbertian norms, the corresponding ambient relative operator $\widehat{\Delta}_{b|a}$ commutes with the orthogonal extension of $P$. Consequently, the closures of $\Ran P$ and $\ker P$ are reducing subspaces for $\widehat{\Delta}_{b|a}$, and
\begin{equation}
P\widehat{\Delta}_{b|a}^{\i t}
=
\widehat{\Delta}_{b|a}^{\i t}P,
\quad t\in\R.
\end{equation}
Thus the relative one-parameter group dynamically preserves the decomposition determined by the projection. This is the origin of the modular terminology used in this paper. The appearance of positive relative operators and their imaginary powers is reminiscent of modular constructions associated with standard subspaces, see for instance \cite{CoL25}. The important difference is that here the modular object is not given in advance: it is extracted directly from a norm-one projection and from the local differential geometry of the ambient Banach space.

The case of noncommutative $\L^4$-spaces is particularly well suited to this approach. There are two reasons for this. First, the fourth power of the norm has an explicit Levi form. If
\begin{equation}
\Phi_4(x)=\norm{x}_{\L^4(\cal{M})}^4,
\end{equation}
then
\begin{equation}
\label{intro-Levi-L4}
\scr L_{\Phi_4,a}(z)
=
2\bigl(\norm{a^*z}_{\L^2(\cal{M})}^2+\norm{za^*}_{\L^2(\cal{M})}^2\bigr).
\end{equation}
Thus the tangent Hilbert geometry separates naturally into a left part and a right part. This is precisely the two-sided structure which is needed in order to recover rectangular operator spaces.

Second, the duality mapping of $\L^4(\cal M)$ is cubic:
\begin{equation}
\label{intro-duality-L4}
\frak j_4(x)=xx^*x.
\end{equation}
Polarization of this cubic mapping produces the Jordan triple product
\begin{equation}
\label{intro-Jordan-triple}
\{x,y,z\}
=
\frac12(xy^*z+zy^*x).
\end{equation}
The special role of the exponent $4$ in noncommutative $\L^p$-geometry also appears in the theory of $2$-isometries, where selfpolar forms can be used to recover operator-algebraic structure; see \cite{JRS05}. Our situation is different: we start with a projection rather than an isometry, and no positivity or matricial assumption is imposed initially. Nevertheless, the cubic duality mapping allows us to extract a ternary structure directly from scalar contractivity.

Indeed, let $P\co\L^4(\cal M)\to\L^4(\cal M)$ be a contractive projection and set
\begin{equation}
Y=\Ran P,\qquad
Y^\sharp=\Ran P^\dagger,
\end{equation}
where $P^\dagger$ denotes the adjoint projection under the $\L^4$--$\L^{\frac{4}{3}}$ trace duality. We prove that
\begin{equation}
\label{intro-dual-range}
Y^\sharp
=
\frak j_4(Y)
=
\{xx^*x:x\in Y\}.
\end{equation}
In particular, the nonlinear image $\frak j_4(Y)$ is a linear subspace of $\L^{\frac{4}{3}}(\cal M)$. Polarization then yields
\begin{equation}
\label{intro-Jordan-dual-range}
\{Y,Y,Y\}
\subset Y^\sharp.
\end{equation}
Thus scalar contractivity forces Jordan triple relations taking values in the dual range. At this stage, however, the product is not internal to $Y$, so these relations do not yet equip $Y$ with a Jordan triple structure. Recovering an internal ternary product requires the additional deweighting argument developed below.

The first nontrivial matrix level yields a stronger conclusion. One of the main results of this paper is that for projections on $\L^4$ the level $2$ condition already contains all matrix levels. More precisely, we prove that
\begin{equation}
\label{intro-main-equivalence}
P\text{ is $2$-contractive}
\quad\Longleftrightarrow\quad
YY^*Y\subset Y^\sharp
\quad\Longleftrightarrow\quad
P\text{ is completely contractive}.
\end{equation}
Hence, for projections on noncommutative $\L^4$-spaces, complete contractivity can be tested on $2\times2$ matrices. This collapse of an infinite family of matricial norm inequalities to a single matrix level is one of the main rigidity phenomena established here.

By trace-adjoint duality, the same equivalence holds for projections on $\L^{\frac43}(\cal M)$, see Corollary~\ref{cor-dual-two-contractive-L43}.

For Schatten spaces, the equivalence between $2$-contractive and completely contractive complementation follows from \cite[Theorem~1.1]{LRR09}. The contribution of \eqref{intro-main-equivalence} is a direct duality-mapping proof beyond the Schatten setting, together with the intrinsic criterion $YY^*Y\subset Y^\sharp$. Theorem~\ref{thm-Haagerup-even-finite-level} extends this equivalence to Haagerup $\L^4$-spaces associated with arbitrary von Neumann algebras. We refer also to \cite{ArL26}, \cite{Bof25} and \cite{You83} for related results.

The mechanism behind \eqref{intro-main-equivalence} is transparent. Scalar contractivity gives only the symmetrized expression \eqref{intro-Jordan-triple}. By applying the scalar result to the amplification $\Id_{S_2^4}\ot P$ and choosing suitable $2\times2$ block matrices, one can separate the two outer variables and isolate the non-symmetrized product $xy^*z$. Once
\begin{equation}
xy^*z\in Y^\sharp,\quad x,y,z\in Y,
\end{equation}
is known, the same property at every matrix level follows entrywise, and the cubic characterization of contractivity implies complete contractivity. In this sense, scalar contractivity detects Jordan triple geometry, whereas $2$-contractivity detects TRO geometry.

The same polynomial principle gives a finite-level criterion at every even exponent. If $p=2m$ for some integer $m \geq 2$ and
\begin{equation}
\label{intro-even-level}
N_m
\ov{\mathrm{def}}{=}
\left\lceil\frac{2m}{3}\right\rceil,
\end{equation}
we prove that, for an arbitrary von Neumann algebra $\cal M$, an $N_m$-contractive projection on its Haagerup space $\L^{2m}(\cal M)$ is completely contractive.

The proof extracts an arbitrary alternating product of length $2m-1$ as a torus Fourier coefficient of the power duality mapping at the matrix level $N_m$. Trace-adjoint duality gives the analogous result at the conjugate exponent $\frac{2m}{2m-1}$. In particular, level $2$ suffices for $p=6$ and $p=\frac{6}{5}$. We do not claim that the level in \eqref{intro-even-level} is optimal. Indeed, for Schatten spaces, level $2$ is already known to suffice for any $1 <p < \infty$, $p \neq 2$, by \cite[Theorem~1.1]{LRR09}.

The remaining difficulty is that the ternary products still belong a priori to the dual range $Y^\sharp$, not to $Y$. In the finite-dimensional setting, the unsymmetrized ternary relation yields natural left and right multiplication maps from $Y$ into $Y^\sharp$. Choose a basis $x_1,\ldots,x_k$ of $Y$ and define
\begin{equation}
\label{intro-a-b}
a
=\sum_{j=1}^k x_jx_j^*,
\qquad
b
=\sum_{j=1}^k x_j^*x_j.
\end{equation}
We prove the exact two-sided relation
\begin{equation}
\label{intro-density-dual-range}
aY=Y^\sharp=Yb.
\end{equation}
This algebraic relation has a natural tangent-Hilbert interpretation. Indeed, if $\frak H^\ell$ and $\frak H^r$ denote the sums of the left and right tangent forms associated with the points $x_1,\ldots,x_k$, then
\[
\frak H^\ell(x,y)=\Tr(y^*ax),
\qquad
\frak H^r(x,y)=\Tr(y^*xb),
\]
and
\[
\sum_{j=1}^k\scr L_{\Phi_4,x_j}(x,y)
=
2\bigl(\frak H^\ell(x,y)+\frak H^r(x,y)\bigr).
\]
The multiplication maps $y\mapsto ay$ and $y\mapsto yb$ are precisely the Riesz representatives of these two Hilbertian structures in the common dual range $Y^\sharp$. Their relative operator is therefore
\begin{equation}
\label{intro-relative-operator}
\Delta(y)=a^{-1}yb.
\end{equation}
Moreover, we have
\begin{equation}
\label{mod-intro}
\Delta^{\i t}(Y)
=Y,\quad t \in \R.
\end{equation}
In fact, these maps are complete isometries for the inherited operator space structure, and the corresponding left-right unitary action commutes with the projection. The relation \eqref{mod-intro} admits a modular interpretation. Let $e=s_\ell(P)$ and $f=s_r(P)$ and consider the linking von Neumann algebra
\begin{equation}
\cal R(e\cal Mf)
=
\begin{bmatrix}
e\cal Me&e\cal Mf\\
f\cal Me&f\cal Mf
\end{bmatrix}
\end{equation}
together with the density
\begin{equation}
D=
\begin{bmatrix}
a^2&0\\
0&b^2
\end{bmatrix}.
\end{equation}
Then the restriction of the modular automorphism group associated with $D$ to the upper-right corner is
\begin{equation}
\sigma_t^D(y)=a^{2\i t}yb^{-2\i t},
\end{equation}
and the relative group coming from the two Levi geometries satisfies
\begin{equation}
\Delta^{\i t}(y)=\sigma_{-\frac t2}^D(y).
\end{equation}
Thus the ``modular group'' obtained abstractly by comparing tangent Hilbert geometries is exactly the restriction of a genuine von Neumann algebra modular group on the appropriate linking algebra.

This identification leads to a two-sided deweighting of the range. Namely, we set
\begin{equation}
\label{intro-deweighting}
\cal V
=
a^{-\frac14}Yb^{-\frac14}.
\end{equation}
Using the invariance under the relative modular group together with the ternary relation $YY^*Y\subset Y^\sharp$, we prove that $\cal V$ is a finite-dimensional $\W^*$-$\TRO$. Hence
\begin{equation}
Y=a^{\frac14}\cal Vb^{\frac14}.
\end{equation}
The quarter powers in this formula are not an arbitrary change of variables: they are dictated simultaneously by the exponent $p=4$, by the Levi geometry and by the modular structure of the linking algebra.

We further show that this algebraic deweighting is compatible with the operator space norms. More precisely, after equipping the linking algebra of $\cal V$ with the natural faithful state determined by $D$, the preceding identification induces a surjective complete isometry
\begin{equation}
\L^4(\cal{V},\varphi)
\to \Ran P.
\end{equation}
Consequently, when the ambient von Neumann algebra is finite-dimensional, the range of any $2$-contractive projection on its noncommutative $\L^4$-space is completely isometric to a rectangular noncommutative $\L^4$-space associated with a $\W^*$-$\TRO$. Moreover, we obtain an explicit formula for the original projection in terms of a state-preserving conditional expectation on the linking algebra. More precisely, if $\E_{12}$ denotes the upper-right corner of this conditional expectation, then
\begin{equation*}
P(x)
=a^{\frac{1}{4}}\E_{12}\big(a^{-\frac{1}{4}}exf b^{-\frac{1}{4}}\big)b^{\frac{1}{4}},
\quad x \in \L^4(\cal M).
\end{equation*}
This realizes $P$ as the off-diagonal part of a completely positive contractive projection on a matrix amplification and proves directly that $P$ is contractively decomposable.



The preceding results fit naturally into the broader program of understanding which nonassociative structures are selected by norm-one projections on noncommutative $\L^p$-spaces. They also provide a new route to these structures. The existing classifications for Schatten spaces rely on the fine structure theory of Cartan factors, while the results for positive and decomposable projections use, in one form or another, order, lifting or change-of-density arguments. Our approach starts instead from the bare contractive inequality and differentiates it. The Jordan triple relations arise from the first-order geometry encoded by the duality mapping, while the TRO structure is recovered from the matricial ternary relations and the subsequent deweighting. The modular group arises from comparing the left and right tangent Hilbertian structures associated with the range.

This viewpoint suggests a useful general principle. The correct object attached to a contractively complemented subspace may not be visible directly inside the ambient $\L^p$-space. At first it can appear in a dual range or in a tangent Hilbert space. A suitable relative modular operator can then provide the change of density which transports this infinitesimal structure back to the original range. The case $p=4$ treated here gives a concrete realization of this principle and provides evidence that differential and modular methods may be useful for the general problem of classifying contractively complemented subspaces of noncommutative $\L^p$-spaces.

\paragraph{Organization of the paper}
In Section~\ref{sec-preliminaries}, we first recall the necessary background on noncommutative $\L^p$-spaces, ternary rings of operators, Jordan structures and Levi forms on complex Banach spaces. In Section~\ref{sec-Duality}, we study duality mappings associated with gauge functions and their interaction with contractive projections. This provides the first-order part of our approach. In Section~\ref{sec-tangent-Hilbert-geometry}, we introduce the tangent Hilbert geometry associated with a plurisubharmonic function contracted by a projection and establish the corresponding orthogonality, bicontractivity and intertwining properties. In finite dimensions, we compare different tangent Hilbert structures by means of relative positive operators and their imaginary powers.

In Section~\ref{sec-the-case-NC}, we specialize the general theory to noncommutative $\L^4$-spaces. We compute explicitly the Levi form of the fourth power of the norm and relate it to the Peirce decomposition associated with partial isometries. In Section~\ref{sec-L4}, we use the cubic duality mapping to characterize scalar contractivity in terms of the Jordan triple product and to prove the equivalence between $2$-contractivity and complete contractivity. We then return to the tangent Hilbert geometry. When the ambient von Neumann algebra is finite-dimensional, the left and right tangent forms associated with a basis of the range define two Hilbertian structures whose relative operator is identified with the restriction of an inverse square root of a modular operator on a linking von Neumann algebra.

This leads to the two-sided deweighting \eqref{intro-deweighting}, the $\W^*$-$\TRO$ structure of the deweighted range and the complete isometric identification with a rectangular noncommutative $\L^4$-space. In Section~\ref{sec-even-exponents}, we prove the finite matrix-level criterion \eqref{intro-even-level} for all even exponents and its trace-dual counterpart.

Finally, in Section~\ref{sec-weighted-tangent-general-p}, we return to arbitrary exponents $p>2$. We introduce weighted left and right tangent geometries and associated Gram densities for finite-rank $2$-contractive projections, identify the resulting relative operators with modular dynamics, and obtain a weighted $\W^*$-$\TRO$ description of the range. This yields, by duality, complete contractivity of every finite-rank $2$-contractive projection on a tracial noncommutative $\L^p$-space for $1<p<\infty$, $p\neq2$.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Preliminaries}
\label{sec-preliminaries}Unless explicitly stated otherwise, $\cal M$ is a semifinite von Neumann algebra equipped with a normal semifinite faithful trace $\tau$. For $1\leq p<\infty$, we denote by $\L^p(\cal M)$ the space of all $\tau$-measurable operators $x$ affiliated with $\cal M$ such that $\tau(|x|^p)<\infty$, equipped with the norm
\begin{equation}
\label{Def-norm-Lp}
\norm{x}_{\L^p(\mathcal{M})}
\ov{\mathrm{def}}{=} \bigl(\tau(\vert x\vert^p)\bigr)^{\frac{1}{p}}.
\end{equation}
Here $\vert x \vert \ov{\mathrm{def}}{=}(x^*x)^{\frac{1}{2}}$ denotes the modulus of $x$. We also set $\L^{\infty}(\mathcal{M}) \ov{\mathrm{def}}{=} \mathcal{M}$, equipped with its operator norm. The subspace $\L^p(\mathcal{M}) \cap \mathcal{M}$ is dense in $\L^p(\mathcal{M})$. Unless explicitly stated otherwise, we use the sesquilinear trace duality
\[
\la x,y\ra_{\L^p(\cal M),\L^{p^*}(\cal M)}
\ov{\mathrm{def}}{=}\tau(y^*x),
\quad \frac1p+\frac1{p^*}=1.
\]
In the remainder of the paper, we write $\tr$ for $\tau$. See \cite{Ray03}, \cite{PiX03}, \cite{Pis98} for more information on noncommutative $\L^{p}$-spaces and to the books \cite{Bla06}, \cite{KaR97a}, \cite{KaR97b} and \cite{Tak03} for the topic of von Neumann algebras.

For $n\geq1$, we equip $\M_n\otvn\cal M$ with the trace $\Tr_n\ot\tau$ and use the canonical identification $
\L^p(\M_n\otvn\cal M)
=S_n^p[\L^p(\cal M)]$. If $T\co\L^p(\cal M)\to\L^p(\cal M)$ is bounded, its $n$th amplification is the entrywise map
\[
T_n
\ov{\mathrm{def}}{=} \Id_{S_n^p} \ot T,
\quad
T_n([x_{ij}])
=[T(x_{ij})].
\]
We say that $T$ is $n$-contractive if $\norm{T_n}\leq1$, and completely contractive if it is $n$-contractive for any integer $n \geq 1$. We equip $\L^p(\cal{M})$ with its natural operator space structure \cite{Pis98}. With this structure, the canonical identification $
\L^p(\M_n\otvn\cal M)
=
S_n^p[\L^p(\cal M)]$ is completely isometric. Consequently, the completely bounded norm of a bounded map $T$ between noncommutative $\L^p$-spaces is
\[
\norm{T}_{\cb}
=
\sup_{n \geq 1}
\norm{\Id_{S_n^p}\ot T}.
\]
Likewise, a linear map is a complete isometry if and only if all its matrix amplifications are isometries. %Thus, throughout the paper, the terms \emph{completely contractive} and \emph{complete isometry} are used in their operator space sense.

We also recall the notion of decomposability used below. Let $\cal M$ and $\cal N$ be von Neumann algebras and let $T \co \L^p(\cal M) \to \L^p(\cal N)$ be a bounded map. Put
\[
T^\circ(x)\ov{\mathrm{def}}{=} T(x^*)^*,
\quad x\in\L^p(\cal M).
\]
The map $T$ is decomposable if there exist bounded linear maps $v_1,v_2 \co \L^p(\cal M) \to \L^p(\cal N)$ such that the block map $
\begin{bmatrix}
v_1 & T \\
T^\circ & v_2
\end{bmatrix}
\co \L^p(\M_2 \otvn \cal M) \to \L^p(\M_2 \otvn\cal N),
\quad
\begin{bmatrix}a&b\\c&d\end{bmatrix}
\mapsto
\begin{bmatrix}v_1(a)&T(b)\\
T^\circ(c)&v_2(d)\end{bmatrix}$, is completely positive. Its decomposable norm is $
\norm{T}_{\mathrm{dec}}
\ov{\mathrm{def}}{=}
\inf\max\{\norm{v_1},\norm{v_2}\}$, where the infimum is taken over all such pairs $(v_1,v_2)$. We say that $T$ is contractively decomposable if $\norm{T}_{\mathrm{dec}}\leq1$. See \cite[Definition~4.1]{Arh24b}.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Jordan structures}

\paragraph{Ternary rings of operators} Rectangular operator spaces are naturally modeled by off-diagonal corners of von Neumann algebras. The intrinsic objects corresponding to such corners are ternary rings of operators, which we briefly recall. A ternary ring of operators (or simply $\TRO$) is a norm closed subspace $V$ of the space $\B(H,K)$, for some complex Hilbert spaces $H$ and $K$, which is closed under the triple product $(x,y,z) \mapsto xy^*z$, see, e.g., \cite[4.4.1 p.~161]{BLM04}. A TRO $V$ is called a $\W^*$-$\TRO$ if it is weak* closed in the dual Banach space $\B(H,K)$. A sub-TRO of a TRO $V$ is a closed subspace $W$ of $V$ satisfying $W W^*W \subset W$. We refer to \cite{BLM04}, \cite{BFT12}, \cite{BuT19}, \cite{BuT13a}, \cite{BuT13b}, \cite{DoR07}, \cite{EsM23}, \cite{Ham92}, \cite{Ham99}, \cite {KaR02}, \cite{NeR03}, \cite{PlR19}, \cite{SaS13}, \cite{SaS17} and \cite{Zet83} for more information on TROs.

\begin{example} \normalfont
\label{typical-TRO}
A basic example of $\W^*$-$\TRO$ is given by $e\cal{M}f$, where $e,f$ are projections of a von Neumann algebra $\cal{M}$.
\end{example}

\paragraph{Operator space structures of TROs}
Since the results of this paper are formulated in terms of complete isometries, the canonical operator space structure of a TRO will play an essential role. Every TRO admits an operator space structure. Indeed, let us assume that $V$ is a TRO contained in $\B(H,K)$. Then for each integer $n \geq 1$, the matrix space $\M_n(V)$ can be identified with a TRO contained in $\M_n(\B(H,K)) \cong \B(H^n,K^n)$. This provides a canonical operator space matrix norm on $V$ such that each matrix space $\M_n(V)$ is a TRO. By \cite[Proposition 2.1 p.~265]{KaR02} or \cite[Proposition 2.1 p.~83]{Ham99}, the TRO-matrix norms are uniquely determined on each TRO and do not depend on the choice of the representing Hilbert spaces.

\paragraph{$\TRO$-homomorphisms} A $\TRO$-homomorphism (or triple morphism) is a linear map $T \co V \to W$ between two $\TRO$s respecting the triple product, i.e.,
$$
T(xy^*z) 
= T(x)T(y)^*T(z), \quad x,y,z \in V.
$$
If, in addition, $T$ is an injection from $V$ onto $W$, we call $T$ a $\TRO$-isomorphism from $V$ onto $W$. By \cite[Proposition 2.4 p.~498]{EOR01}, a TRO-isomorphism between $\W^*$-TROs is necessarily weak* continuous. By  \cite[Proposition 2.1 p.~495]{EOR01}, every TRO-homomorphism is completely contractive, and every injective $\TRO$-homomorphism is completely isometric. Finally, by \cite[Proposition 2.1 p.~83]{Ham99} (see also \cite[Corollary 4.4.6 p.~163]{BLM04}), if $T \co V \to W$ is a linear map between TRO's then $T$ is a surjective complete isometry if and only if $T$ is a surjective 2-isometry if and only if $T$ is a TRO-isomorphism. Thus, for TROs, the complete isometric structure already determines the triple product.

\begin{example} \normalfont
\label{E-finite-dim-TRO}
Every finite-dimensional $\TRO$ $V$ is completely isometric to a finite direct sum of rectangular matrix spaces, i.e., it has the form
$$
V 
= \M_{m_1,n_1} \oplus_\infty \cdots \oplus_\infty \M_{m_k,n_k}, 
$$
see \cite[Corollary A.2 p.~302]{Kan13}.
\end{example}

%\paragraph{Abstract characterization} There is also an abstract characterisation of $\W^*$-TROs due to Zettl \cite{Zet83}, which we recall. An abstract $\W^*$-TRO is a dual Banach space $V$ equipped with a ternary operation $
%\{\cdot, \cdot, \cdot\} \co V \times V \times V \to V$ such that the following conditions hold ($x,y,z,u,v \in V$):
%\begin{enumerate}
%\item the operation is linear in the first and the third variable and conjugate linear in the second,
%\item $\{\{x,y,z\}, u, v\} = \{x, \{u,z,y\}, v\} = \{x, y, \{z, u, v\}\}$,
%\item $\|\{x,y,z\}\| \leq \|x\| \|y\| \|z\|$,
%\item $\|\{x,x,x\}\| = \|x\|^3$.
%\end{enumerate}
 %Zettl showed that these abstractly defined objects have concrete representations as $\W^*$-TROs.

\paragraph{Linking algebra} The linking algebra allows us to pass back and forth between the rectangular setting of TROs and the usual setting of von Neumann algebras. If $\cal{M}$ is a von Neumann algebra and $e$ is an orthogonal projection in $\cal{M}$, then $e\cal{M} e^\perp$ is a $\W^*$-TRO by Example \ref{typical-TRO}. Conversely, let $V$ be a $\W^*$-TRO contained in $\B(H,K)$. Replacing $H$ and $K$, if necessary, with the essential subspaces $\ovl{V^*K}$ and $\ovl{VH}$, respectively, we may and do assume that the representation is nondegenerate. We can then consider the adjoint space $V^* \ov{\mathrm{def}}{=} \{x^* : x \in V \}$, which is a subspace of the space $\B(K,H)$ and the von Neumann subalgebras 
\begin{equation}
\label{def-MV-NV}
\mathrm{M}(V) \ov{\mathrm{def}}{=} \ovl{\Span VV^*}^{\w^*}
\quad \text{and} \quad 
\mathrm{N}(V) \ov{\mathrm{def}}{=} \ovl{\Span V^*V}^{\w^*}
\end{equation}
of the algebras $\B(K)$ and $\B(H)$. Since $\mathrm{M}(V) \cdot V \subset V$ and $V \cdot \mathrm{N}(V) \subset V$, we can introduce the von Neumann algebra
\begin{equation}
\label{Linking-algebra}
\mathrm{R}(V)
\ov{\mathrm{def}}{=} \begin{bmatrix}
  \mathrm{M}(V)   &  V \\
  V^*   &  \mathrm{N}(V) \\
\end{bmatrix}
\subset \B(K \oplus H),
\end{equation}
which is called the linking von Neumann algebra of $V$, see \cite{KaR02}, \cite[p.~846]{Rua04} and \cite{WCW24}. Then we have a TRO-isomorphism 
\begin{equation}
\label{TRO-linking}
V 
= e \mathrm{R}(V)e^\perp,
\end{equation}
where $e 
\ov{\mathrm{def}}{=} \begin{bmatrix}
    \Id_K & 0  \\
    0 & 0  \\
\end{bmatrix}
$ 
and 
$e^\perp
\ov{\mathrm{def}}{=} \begin{bmatrix}
   0  &  0 \\
   0  &  \Id_H \\
\end{bmatrix}$. So a $\W^*$-TRO $V$ can be identified with the off-diagonal corner at the (1,2) position of its linking von Neumann algebra. By \cite[Lemma 2.1 p.~847]{Rua04}, the central covers $c(e)$ and $c(e^\perp)$ of $e$ and $e^\perp$ in the linking von Neumann algebra $\mathrm{R}(V)$ are equal to $1$.

Conversely, it is worth noting that, according to \cite[Theorem 2.1]{WCW24}, a von Neumann algebra is $*$-isomorphic to the linking von Neumann algebra of a $\W^*$-TRO if and only if it contains no abelian direct summand.

%\paragraph{Completely contractive projections on TROs} 
%We will use the following result \cite[Theorem 4.4.9 p.~165]{BLM04} (it is essentialy a slight generalization of a result of Youngson \cite{You83}).
%
%\begin{thm}
%\label{Th-youngson}
%Let $V$ be a $\TRO$ and $P \co V \to V$ be a completely contractive projection. For any $x, y, z \in V$, we have  
%%\begin{equation}
%%\label{exp-2}
%%P\big(P(x)P(y)^*P(z)\big) 
%%= P(xP(y)^*P(z)) 
%%= P(P(x)yP(z)) 
%%= P(P(x)P(y)^*z).
%%\end{equation}
%$$
%P\big(P(x)P(y)^*P(z)\big) 
%= P(xP(y)^*P(z)) 
%= P(P(x)y^*P(z)) 
%= P(P(x)P(y)^*z).
%$$
%Furthermore, the range $\Ran(P)$ of $P$ is a $\TRO$ with triple product $(x,y,z) \mapsto P(xy^*z)$.
%\end{thm}

%This means that the class of TROs is closed under completely contractive projections.

%Let $W$ be a sub-TRO of a TRO $V$. A projection $P \co V \to V$ on a sub-TRO $W$ is a TRO-conditional expectation on $W$ \cite[p.~499]{EOR01} if we have
%\begin{equation}
%\label{TRO-cond-exp}
%P(zx^*y) 
%= P(z)x^*y,\quad 
%P(xz^*y) 
%= xP(z)^*y, \quad 
%P(xy^*z) 
%= xy^*P(z), \quad z \in V, x, y \in W.
%\end{equation}
%
%\begin{remark} \normalfont
%\label{Rem-TRO-cond}
%Let $P \co V \to V$ be a projection from $V$ onto a sub-TRO $W$. Then by \cite[Theorem 2.5]{EOR01} (see also \cite[Corollary 4.4.10 p.~166]{BLM04}) the following properties are equivalent:
%\begin{enumerate}
%\item $P$ is completely contractive,
%\item $P$ is contractive,
%\item $P$ is a contractive conditional expectation.
%\end{enumerate}
%\end{remark}
%
%
%\paragraph{Projections} 
%The following is \cite[Theorem 3.2.6 p.~297]{Meg98} combined with \cite[5.10 p.~148]{FHHMPZ01}  and \cite[Lemma  Exercise 3.17 p.~304]{Meg98}.
%
%\begin{prop}
%\label{Prop-Fabian}
%Let $X$ be a Banach space and consider a bounded map $P \co X \to X$. Then $P$ is a projection if and only if $P^*\co X^* \to X^*$ is a projection. In this case, we have an isometric isomorphism
%\begin{equation}
%\label{Ran-P-P*}
%(\Ran P)^* 
%\cong \Ran P^*.
%\end{equation}
%and
%\begin{equation}
%\label{}
%\Ran P^*
%= (\ker P)^\perp
%\quad \text{and} \quad
%\Ran P
%= {}^\perp(\ker P^*).
%\end{equation}
%\end{prop}
%
\paragraph{$\JB^*$-triples and $\JBW^*$-triples} 
We recall that an element $a$ in a Banach algebra $A$ is called hermitian \cite[p.~156]{Chu12} if $\norm{\exp \i ta} = 1$ for all $t \in \R$.


A $\JB^*$-triple, e.g. \cite[p.~130]{CGRP14} \cite[Definition 2.5.25 p.~164]{Chu12}, is a complex Banach space $X$ equipped with a continuous triple product $X \times X \times X \to X$, $(x,y,z) \mapsto \{x,y,z\}$ such that

\begin{enumerate}
\item $\{x,y,z\}$ is linear in $x$ and $z$, conjugate linear in $y$, and symmetric in $x$ and $z$,

\item $\{a,b,\{x,y,z\}\} = \{\{a,b,x\},y,z\} - \{x,\{b,a,y\},z\} + \{x,y,\{a,b,z\}\}$ for any $x,y,z,a,b \in X$,

\item the left multiplication $\mathrm{D}(x,x) \co X \to X$, defined by
\begin{equation}
\label{Left-mult-JBW-triple}
\mathrm{D}(x,x)y
\ov{\mathrm{def}}{=} \{x,x,y\},\quad y\in X,
\end{equation}
is Hermitian in the algebra $\B(X)$ of bounded operators on $X$ and has positive spectrum,

\item for any $x \in X$, we have $\norm{\{x,x,x\}}_X = \norm{x}_X^3$.
\end{enumerate}
The last axiom is a Jordan analogue of the Gelfand-Naimark axiom of $\mathrm{C}^*$-algebras. Moreover, if $X$ is a dual Banach space, then it is called a $\JBW^*$-triple \cite[Definition 2.5.30]{Chu12} \cite[p.~528]{CGRP14}. In this case, the predual is unique by \cite[Theorem 5.7.38 p.~233]{CGRP18}. %\cite[Theorem 3.21]{Hor87b}
We refer to the books \cite{Chu12}, \cite{CGRP14}, \cite{CGRP18}, \cite{Fri05}, \cite{Isi19} and \cite{Upm85} and to the important papers \cite{Hor87a}, \cite{Hor87b} and \cite{HoN88} for more information on $\JB^*$-triples and $\JBW^*$-triples.

\begin{example} \normalfont
\label{Ex-JBW-algebra-triple}
A $\JBW^*$-algebra $\cal{M}$ admits a structure of $\JBW^*$-triple with triple product 
\begin{equation}
\label{triple-prod-JBWstar}
(x,y,z) \mapsto \{x,y,z\} 
\ov{\mathrm{def}}{=} (x \circ y^*) \circ z + (z \circ y^*) \circ x - (x \circ z) \circ y^*.
\end{equation}
See \cite[Lemma 3.1.6 p.~174]{Chu12} and \cite[p.~224]{CGRP18}. %In particular, $\JW^*$-algebras are $\JW^*$-triples. % \cite[Definition 2.5.33 p.~168]{Chu12}.
%In this case, the triple product is given by $(x,y,z) \mapsto \{x,y,z\}=\frac{1}{2}(xy^*z+zy^*x)$.
\end{example}


\paragraph{Triple homomorphisms} A linear map $T \co X \to Y$ between two $\JB^*$-triples $X$ and $Y$ is called a triple homomorphism if it preserves the triple product:
$$
T(\{x,y,z\}) 
= \{T(x), T (y), T(z)\}, \quad x,y, z \in X.
$$
A bijective triple homomorphism is called a triple isomorphism. In $\JB^*$-triples, the metric structure and the algebraic structure determine each other. Indeed, by \cite[Theorem 3.1.7 p.~175]{Chu12} and \cite[Theorem 3.1.20 p.~183]{Chu12} a linear bijection between two $\JB^*$-triples is an isometry if and only if it is a triple-isomorphism. When both triples are $\JBW^*$-triples, such a bijection is weak* continuous with respect to their canonical preduals, see \cite[Corollary 5.7.39 p.~234]{CGRP18}.




\paragraph{$\JC^*$-triples and $\JW^*$-triples} A $\JC^*$-triple (or $\mathrm{J}^*$-algebra) \cite[Definition 2.5.34 p.~169]{Chu12} \cite[p.~440]{CGRP18} \cite[Definition 22.7.1 p.~460]{Isi19} is a norm closed subspace of $\B(H,K)$ which is also closed under the map $x \mapsto xx^*x$. By a polarization argument, any $\JC^*$-triple is closed under the triple product $(x,y,z) \mapsto \{x,y,z\}\ov{\mathrm{def}}{=}\frac{1}{2}(xy^*z+zy^*x)$. A $\JC^*$-triple is a $\JB^*$-triple. A weak* closed $\JC^*$-subtriple of $\B(H,K)$ is called a $\JW^*$-triple. We shall also use this terminology for a $\JBW^*$-triple which is isometrically triple isomorphic to such a concrete $\JW^*$-triple.



\begin{example} \normalfont
A $\W^*$-$\TRO$ is a $\JW^*$-triple for the symmetrized triple product
\[
\{x,y,z\}
\ov{\mathrm{def}}{=}\frac12(xy^*z+zy^*x).
\]
\end{example}

\begin{example} \normalfont
By \cite[Proposition 6.1.41 p.~362]{CGRP18}, the complexification $\mathrm{H}_3(\O)_\C=\mathrm{H}_3(\O_\C)$ of the $\JBW$-factor $\mathrm{H}_3(\O)$ is equipped with a structure of $\JBW^*$-factor. The underlying $\JBW^*$-triple is called a Cartan factor of type VI.
\end{example}

\begin{example} \normalfont
By \cite[p.~140]{Isi19}, the space $\Asym_n \ov{\mathrm{def}}{=} \{x \in \M_n : x^\top=-x \}$ of skew-symmetric complex matrices of $\M_n$ is a $\JBW^*$-triple (called a Cartan factor of type $\mathrm{III}_n$). If $n$ is even, it is even equipped with a structure of reversible $\JW^*$-algebra by \cite[Proposition 25.2.2 p.~513]{Isi19}.
\end{example}


Let $\cal{A}$ be a $\C^*$-algebra. The formula
\begin{equation}
\label{triple product-C-star}
\{x,y,z\}
\ov{\mathrm{def}}{=}
\frac12(xy^*z+zy^*x),
\quad
x,y,z \in \cal{A},
\end{equation}
endows $\cal{A}$ with a structure of $\JB^*$-triple. The same formula defines a $\JB^*$-triple structure on $\B(H,K)$ for any complex Hilbert spaces $H$ and $K$; see \cite[Fact~4.1.41 p.~465]{CGRP14}. Furthermore, each $\JB^*$-algebra is a $\JB^*$-triple with the same norm and triple product 
$$
\{x,y,z\}
= (x\circ y^*) \circ z + (z\circ y^*)\circ x - (x\circ z)\circ y^*
$$ 
(see \cite[Theorem 4.1.45]{CGRP14}). By a $\JB^*$-subtriple of a $\JB^*$-triple $E$ we mean a norm closed subspace of $E$ which is also closed for the triple product.

\paragraph{Tripotents and Peirce projections} An element $u$ in a $\JB^*$-triple $X$ is called a tripotent if it is a fixed point of the triple product, i.e., $\{u,u,u\} = u$, see \cite[Definition 4.2.13 p.~505]{CGRP14} and \cite[Definition 1.2.31 p.~28]{Chu12} for a larger context.

\begin{example}
\normalfont
\label{ex-trip-partial-iso}
If $X$ is a $\JC^*$-triple (in particular if $X$ is a $\JW^*$-triple), by \eqref{triple product-C-star} the equality $\{u,u,u\} = u$ is equivalent to $uu^*u=u$, for any $u \in X$. So the tripotents are precisely the partial isometries of $X$, by \cite[Exercise 15 p.~244]{Con90}.
\end{example}

With a tripotent $u$ and $0 \leq i \leq 2$, we can introduce the Peirce projections $\P_i^u \co X \to X$ with range $X_i(u)$. For any $x \in X$, we have
\begin{equation}
\label{Peirce-1}
\P_2^u(x) 
\ov{\mathrm{def}}{=} \{u, \{u, x, u\},u\}, \quad
\P_1^u \ov{\mathrm{def}}{=} 2\big(\mathrm{D}(u, u) - \P_2^u\big) 
\end{equation}
and 
\begin{equation}
\label{Peirce-2}
\P_0^u+ \P_1^u+ \P_2^u 
= \Id_{X}.
\end{equation}
These maps are contractive linear projections by \cite[Corollary 1.2]{FrB85b}. A crucial property of $\JBW^*$-triples is that for a tripotent $u$ of $X$ the Peirce-2 subspace $X_2(u)$ is a $\JBW^*$-algebra with product $(a,b) \mapsto a \circ b \ov{\mathrm{def}}{=} \{a,u,b\}$, involution $a^* \ov{\mathrm{def}}{=} \{u, a, u\}$ where $a,b \in X_2(u)$, and unit $u$.
%The Peirce-2 subspace $E_{2}(u)$ is a unital JB$^*$-algebra with unit $u$, product $a \circ_{u} b = \{ aub\}$ and involution $a^{\sharp_{u}} = \{ uau\}$ (c.f. \cite[Theorem 2.2]{BraKaUp78} and \cite[Theorem 3.7]{KaUp77}; \cite[p.\ 185]{Chu12}).
%Given a tripotent $u\in E$, the mappings $P_{i} (u) : E\to E_{i} (u)$, $(i=0,1,2)$, defined by \cite[Fact 4.2.14 p.~505]{CGRP14}
%$$
%\P_2^u 
%= L(u,u)(2 L(u,u) -\Id_{E}), \P_1(u) 
%= 4L(u,u)(\Id_{E}-L(u,u)),
%$$ 
%$$ \ 
%\hbox{ and } 
%\P_0^u 
%=(\Id_{E}-L(u,u)) (\Id_{E}-2 L(u,u)),
%$$ 
%are contractive linear projections, called the Peirce projections associated with $u$. The range of $\P_i^u$ is the eigenspace $E_i(u)$ of $L(u, u)$ corresponding to the eigenvalue $\frac{i}{2},$ and
%\begin{equation}
%\label{Peirce-decomposition}
%E
%= E_{2}(u) \oplus E_{1}(u) \oplus E_0(u)
%\end{equation}
%is the Peirce decomposition of $E$ relative to $u$. Furthermore, the following Peirce rules are satisfied, 
%\begin{equation}
%\label{eq peirce rules1} 
%\{ E_{2}(u),E_{0}(u),E\} 
%= \{ E_{0}(u),E_{2}(u),E\} 
%=\{0\},
%\end{equation} 
%\begin{equation}
%\label{eq peirce rules2} 
%\{E_{i}(u),E_{j} (u),E_{k}(u)\}
%\subseteq E_{i-j+k} (u),
%\end{equation} 
%where $E_{i-j+k} (u) = \{0\}$ whenever $i-j+k \notin \{ 0,1,2\}$ (\cite{FriRu85} or \cite[Th.\ 1.2.44]{Chu12}). For $x,y,z$ in a $\JB^*$-triple $E$ we have \cite[Cor. 3]{FriRu86} $\|\{x,y,z\}\|\leq \norm{x}\norm{y}\norm{z}$. A tripotent $u$ is called complete if $E_0(u) $ reduces to $\{0\}$.



\begin{example} \normalfont
\label{Pierce-example}
In the case where $X$ is a $\JBW^*$-algebra and $e \in X$ is a projection, i.e. a selfadjoint idempotent, the Peirce projections are given by the following expressions (essentially \cite[p.~48]{HOS84})
\begin{equation}
\label{Pierce-example-fo}
\P_2^e(x)
=\{e, x^*, e\}, \quad \P_1^e(x)= 2\{e,x^*,1-e\}, \quad \P_0^e(x)=\{1-e,x^*,1-e\},
\end{equation}
where $x \in X$.
\end{example}

A tripotent $u$ in a $\JBW^*$-triple $X$ is called complete if $X_0(u) = \{0\}$ \cite[p.~16]{HKPP20} \cite[p.~517]{CGRP14} and unitary if $X = X_2(u)$. %that is if $\{u, u, x\} = x$ for all $x \in \cal{M}$.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Banach space geometry}

Following \cite[Definition 5.1.1 p.~426]{Meg98} and \cite[Definition 1.1 p.~41]{Cio90}, we say that a Banach space $X$ is strictly convex (or rotund) if for any $t \in (0,1)$ and any $x,y \in X$ with $\norm{x}_X=\norm{y}_X=1$ and $x \not= y$ we have $\norm{tx+(1-t)y}_X < 1$. According to \cite[Proposition 1.2 p.~41]{Cio90}, this is equivalent to say that if $x \not= y$ with $\norm{x}_X=\norm{y}_X=1$ then $\norm{x+y}_X < 2$. A Banach space $X$ is called smooth \cite[Definition 3.1 p.~21]{Cio90} (see also \cite[pp.~480-481]{Meg98}) if for any $x \in X \setminus \{0\}$ there exists a unique $x^* \in X^*$ such that $\norm{x^*}_{X^*} = 1$ and $\la x,x^* \ra_{X,X^*} = \norm{x}_X$. If $X$ is a \textsl{reflexive} Banach space, it is known \cite[Definition 5.4.7 p.~481]{Meg98} \cite[Corollary 1.4 p.~43]{Cio90} that $X$ is smooth if and only if $X^*$ is strictly convex and $X$ is strictly convex if and only if $X^*$ is smooth.

The differentiability of the norm is a very useful tool in understanding the geometry of the underlying Banach space. Various notions of differentiability of the norm of Banach spaces include Fr\'echet, Gâteaux, and strong subdifferentiability.

\paragraph{Gâteaux differentiability}
Let $\Omega$ be an open subset of a normed vector space $E$, let $f \co \Omega \to \R$ be a real-valued function, let $x \in \Omega$ and let $h$ be a nonzero element of $E$. The function $f_h$ given by $t \mapsto f(x + t h)$ is defined on an open interval containing $0$. If the limit
$$
\lim_{t \to 0} \frac{f(x + t h) - f (x)}{t}
$$
exists, then we denote this derivative by $\nabla_x f (h)$. It is called the G\^ateaux derivative (directional derivative) of $f$ at $x$ in the direction $h$. If $\nabla_x f(h)$ is defined and $\lambda \in \R \setminus \{0\}$, then $\nabla_x f (\lambda h)$ is defined and $\nabla_x f (\lambda h) = \lambda \nabla_x f(h)$. The function $f$ is G\^ateaux differentiable at $x$ if $\nabla_x f(h)$ exists for any $h \in E$ and the map $h \mapsto \nabla_x f(h)$ is a continuous real linear functional on $E$.


If $f$ is Fr\'echet differentiable at $x$, it is also G\^ateaux differentiable there, and 
\begin{equation}
\label{Gateaux-Frechet}
\mathrm{D}_\R f(x).h
=\nabla_x f (h), \quad h \in E.
\end{equation}
For a Banach space $X$ and $x\in X\setminus\{0\}$, the norm is Fr\'echet differentiable at $x$ if and only if the norming functional at $x$ is unique and is strongly exposed by $x$ in the closed unit ball of $X^*$. We refer to the book \cite{DGZ93} for more information.

According to \cite[Theorem 3.5 p.~22]{Cio90} (see also \cite[Theorem 1 p.~22]{Die75}), a Banach space $X$ is smooth if and only if its norm $\norm{\cdot}_X$ is G\^ateaux differentiable on $X \setminus \{0\}$. 


Recall the following result \cite{CoS70}.

\begin{prop}[Cohen--Sullivan]
\label{prop-Sullivan}
A subspace $Y$ of a smooth Banach space $X$ can be the range of at most one projection of norm one. 
\end{prop}

The following is \cite[Theorem 3.2.6 p.~297]{Meg98} combined with \cite[5.10 p.~148]{FHHMPZ01}.

\begin{prop}
\label{Prop-Fabian}
Let $X$ be a Banach space and consider a bounded linear operator $P \co X \to X$. Then $P$ is a projection if and only if $P^*\co X^* \to X^*$ is a projection. In this case and if $P$ is contractive, $P(X)^*$ is canonically isometrically isomorphic to $P^*(X^*)$, i.e.,
\begin{equation}
\label{Ran-P-P*}
(\Ran P)^*
\cong \Ran(P^*).
\end{equation}
\end{prop}


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Plurisubharmonic functions on open subsets of Banach spaces}
%\paragraph{Plurisubharmonic functions on open subsets of Banach spaces}

If $\Omega$ is an open set of the complex plane $\mathbb{C}$, we recall \cite[p.~246]{Muj86} \cite[Definition 3.1.2 p.~60]{ArG01} that a function $f \co \Omega \to [-\infty,\infty)$ is said to be subharmonic if $f$ is upper semicontinuous and if for each $a \in \Omega$ and each $r > 0$ such that $\ovl{B}(a,r) \subset \Omega$ we have the inequality
$$
f(a)
\leq \frac{1}{2\pi} \int_{0}^{2\pi} f(a+r\e^{\i \theta}) \d \theta
$$
and if $f$ is not identically $-\infty$ on any connected component of $\Omega$.

Let $\Omega$ be an open subset of a \textit{complex} Banach space $X$. Following \cite[Definition 34.3 p.~247]{Muj86}, we say that a function $f \co \Omega \to [-\infty, \infty)$ is plurisubharmonic if $f$ is upper semicontinuous and if for any $a \in \Omega$ and any $b \in X$ such that $%a+b\ovl{B}(0,1)=
\{a+\lambda b : |\lambda|\leq 1\} \subset \Omega$ we have
$$
f(a)
\leq \frac{1}{2\pi} \int_{0}^{2\pi} f(a+b\e^{\i \theta}) \d \theta.
$$
Since $f$ is Borel measurable and bounded above by \cite[Proposition 34.1 p.~245]{Muj86} on each disc $a + \ovl{B}(0,1)b \subset \Omega$, the integral $\int_{0}^{2\pi} f(a+b\e^{\i \theta}) \d \theta $ is well-defined, though it can be $-\infty$. 

An upper semicontinuous function $f\colon\Omega\to[-\infty,\infty)$ is plurisubharmonic if and only if for any $x,y \in X$, its restriction $z \mapsto f(x+zy)$ is subharmonic or identically $-\infty$ on each connected component of $\{z \in \mathbb{C} : x+z y \in \Omega \}$.





%Let $f \co \Omega \to  [-\infty,\infty)$ be upper semicontinuous. Then $f$ is plurisubharmonic if and only if for each finite dimensional subspace $M$ of $X$ the restriction of $f$ to $\Omega \cap M$ is plurisubharmonic.
\begin{example} \normalfont
\label{Ex-C-conv-pluri}
Let $\Omega$ be an open convex subset of a complex Banach space $X$. Any continuous convex function $f \co \Omega \to \R$ is plurisubharmonic.
\end{example}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Levi forms of functions defined on open subsets of Banach spaces}

%livre Chrstine Laurent





Levi forms in the setting of complex Banach spaces have been studied in \cite{Lig76}. 
%We refer to the book \textbf{REF} for the finite-dimensional case. 
We briefly recall the relevant terminology. Let $\Omega$ be an open subset of a complex Banach space $X$ and let $\Phi \co \Omega \to \R$ be a real function of class $\mathrm{C}^2$, where Fr\'echet differentiability is understood with respect to the underlying real Banach space $X_{\R}$. Thus, for each $a \in \Omega$, the real Hessian $
\D_{\R}^2\Phi(a) \co X_{\R} \times X_{\R} \to \R$ is a continuous symmetric $\R$-bilinear form. In general it is not complex bilinear. The complex structure nevertheless determines a canonical Hermitian form from this real Hessian, namely the Levi form of $\Phi$ at $a$. 
With the convention that Hermitian forms are complex linear in the first variable, it is given by 
\begin{equation}
\label{Levi-form-diagonal}
\scr{L}_{\Phi,a}(z)
=\frac14\left(\D_{\R}^2\Phi(a)(z,z)+\D_{\R}^2\Phi(a)(\i z,\i z)\right), \quad z \in X,
\end{equation}
explicitly introduced in \cite[p.~1]{Lig76}. In particular, we have
\begin{equation}
\label{Hermitian-Levi}
\scr{L}_{\Phi,a}(z,w)
=\ovl{\scr{L}_{\Phi,a}(w,z)}, \quad z,w \in X.
\end{equation}
Its polar form is given by polarization by the following formula.

\begin{prop}
For any $z,w \in X$, we have
\begin{equation}
\label{Levi-form-general}
\scr{L}_{\Phi,a}(z,w)
=\frac14\big(\D_{\R}^2\Phi(a)(z,w)+\D_{\R}^2\Phi(a)(\i z,\i w)
+\i\D_{\R}^2\Phi(a)(z,\i w)-\i\D_{\R}^2\Phi(a)(\i z,w)\big).
\end{equation}
\end{prop}

\begin{proof}
For any $z,w \in X$, we use the shorthand notation $\phi(z,w) \ov{\mathrm{def}}{=} \D_{\R}^2\Phi(a)(z,w)$. Then $\phi$ is a continuous symmetric $\R$-bilinear form on $X_{\R} \times X_{\R}$. The diagonal form can be written
\[
\scr{L}_{\Phi,a}(z) 
\ov{\eqref{Levi-form-diagonal}}{=} \frac{1}{4} \left( \phi(z,z) + \phi(\i z, \i z) \right)
\]
For any $z,w \in X$, the polarization identity \cite[Theorem 1.1 p.~2]{Wei80} %Monier polarization 
gives
\begin{align*}
\MoveEqLeft
\scr{L}_{\Phi,a}(z,w) 
= \frac{1}{4} \sum_{k=0}^{3} \i^k \scr{L}_{\Phi,a}(z + \i^k w) \\       
&=\frac{1}{4} \Big( \underbrace{\scr{L}_{\Phi,a}(z+w) - \scr{L}_{\Phi,a}(z-w)}_{A} + \i \big( \underbrace{\scr{L}_{\Phi,a}(z+\i w) - \scr{L}_{\Phi,a}(z-\i w)}_{B} \big) \Big).
\end{align*}
On the one hand, using the identity $\phi(x+y,x+y) - \phi(x-y,x-y) = 4\phi(x,y)$ we see that
\begin{align*}
A &= \scr{L}_{\Phi,a}(z+w) - \scr{L}_{\Phi,a}(z-w) \\
&= \frac{1}{4} \left( \phi(z+w, z+w) + \phi(\i z + \i w, \i z + \i w) \right) - \frac{1}{4} \left( \phi(z-w, z-w) + \phi(\i z - \i w, \i z - \i w) \right) \\
&= \frac{1}{4} \left( \phi(z+w, z+w) - \phi(z-w, z-w) \right) + \frac{1}{4} \left( \phi(\i z + \i w, \i z + \i w) - \phi(\i z - \i w, \i z - \i w) \right) \\
&= \frac{1}{4} \left( 4\phi(z,w) \right) + \frac{1}{4} \left( 4\phi(\i z, \i w) \right) 
= \phi(z,w) + \phi(\i z, \i w)
\end{align*}
On the other hand, we have
\begin{align*}
B &= \scr{L}_{\Phi,a}(z+\i w) - \scr{L}_{\Phi,a}(z-\i w) \\
&= \frac{1}{4} \left( \phi(z+\i w, z+\i w) + \phi(\i z - w, \i z - w) \right) - \frac{1}{4} \left( \phi(z-\i w, z-\i w) + \phi(\i z + w, \i z + w) \right) \\
&= \frac{1}{4} \left( \phi(z+\i w, z+\i w) - \phi(z-\i w, z-\i w) \right) - \frac{1}{4} \left( \phi(\i z + w, \i z + w) - \phi(\i z - w, \i z - w) \right) \\
&= \frac{1}{4} \left( 4\phi(z, \i w) \right) - \frac{1}{4} \left( 4\phi(\i z, w) \right) 
= \phi(z, \i w) - \phi(\i z, w)
\end{align*}
Combining these results, we deduce that
\[
\scr{L}_{\Phi,a}(z,w) 
= \frac{1}{4} \Big( \phi(z,w) + \phi(\i z, \i w) + \i \left( \phi(z, \i w) - \phi(\i z, w) \right) \Big)
\]
Distributing $\i$ and using the original notation $\D_{\R}^2\Phi(a)$, we obtain the equation \eqref{Levi-form-general}, i.e.,
\[
\scr{L}_{\Phi,a}(z,w) 
= \frac{1}{4} \left( \D_{\R}^2\Phi(a)(z,w) + \D_{\R}^2\Phi(a)(\i z,\i w) + \i \D_{\R}^2\Phi(a)(z,\i w) - \i \D_{\R}^2\Phi(a)(\i z,w) \right)
\]
\end{proof}

It is stated in \cite[Proposition 2.3 p.~67]{Lig76} \cite[Proposition 35.2 p.~257]{Muj86} that a real function $\Phi \co \Omega \to \R$ of class $\mathrm{C}^2$ is plurisubharmonic if and only if for any $a \in \Omega$ and any $z \in X$ we have 
\begin{equation}
\label{Levi-def-pos}
\scr{L}_{\Phi,a}(z,z) 
\geq 0.
\end{equation}

%fait dans [livre Chrstine Laurent] voir aussi\cite[p.~256-257]{Muj86}:
%There is a natural complex-geometric interpretation of the quadratic forms considered previously.
%Equivalently, if $g_{a,z}$ denotes the real-valued function of one complex variable defined in a neighborhood of $0$ by
%\[
%g_{a,z}(\lambda)
%=\Phi(a+\lambda z),
%\]
%then
%\begin{equation}
%\label{Levi-form-complex-line}
%\scr{L}_{\Phi,a}(z,z)
%=\frac{\partial^2 g_{a,z}}{\partial\lambda\,\partial\overline{\lambda}}(0)
%=\frac14\Delta g_{a,z}(0).
%\end{equation}
%Thus, the Levi form is obtained by restricting $\Phi$ to complex affine lines and taking the ordinary two-dimensional Laplacian at the base point. 


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Duality mappings associated with a gauge function}
\label{sec-Duality}


\subsection{Definitions}

Following \cite[Definition 4.1 p.~25]{Cio90} and \cite[Definition 2.13, p.~123]{Pat18}, a gauge function (or a weight function) is a continuous strictly increasing function $\varphi \co \R_+ \to \R_+$ such that $\varphi(0)=0$ and $\lim_{t \to \infty} \varphi(t) = \infty$. Let $X$ be a Banach space and let $\varphi \co \R_+ \to \R_+$ be a gauge function. For each $x \in X \setminus \{0\}$, we define, as in \cite[Definition 2.14, p.~123]{Pat18} \cite[Definition 4.1 p.~25]{Cio90}, the subset
\begin{equation}
\label{Def-JX-gauge}
\J_{X,\varphi}(x)
\ov{\mathrm{def}}{=}
\big\{x^* \in X^* : \la x,x^*\ra_{X,X^*}
=\norm{x}_X\norm{x^*}_{X^*}, 
\norm{x^*}_{X^*}=\varphi(\norm{x}_X) \big\}
\end{equation}
of the dual $X^*$, and we let $\J_{X,\varphi}(0) \ov{\mathrm{def}}{=} \{0\}$. According to \cite[Theorem 2.11, p.~124]{Pat18} or \cite[Remark 4.2 p.~25]{Cio90}, the subset $\J_{X,\varphi}(x)$ is nonempty for any $x \in X$. The multivalued map $\J_{X,\varphi} \co X \to 2^{X^*}$ is called the duality mapping associated with the gauge function $\varphi$. The duality mapping associated with the gauge function $\varphi(t)=t$ is called the normalized duality mapping, which we denote simply by $\J_X$. According to \cite[Corollary 4.5 p.~27]{Cio90}, the Banach space $X$ is smooth if and only if each duality mapping $\J_{X,\varphi}$ is single-valued. In this case, set
\[
F_\varphi(x)
\ov{\mathrm{def}}{=}
\int_0^{\norm{x}_X}\varphi(s)\d s,
\quad x\in X.
\]
The same result gives
\begin{equation}
\label{derivation-duality-mapping}
\nabla_xF_\varphi(y)
=
\Re\la y,\J_{X,\varphi}(x) \ra_{X,X^*},
\quad x\in X\setminus\{0\},\quad y\in X.
\end{equation}
The choice of the gauge only modifies the radial normalization of the duality mapping. More precisely, by \cite[Proposition 4.7(f) p.~28]{Cio90}, if $x \neq 0$, then
\begin{equation}
\label{relation-duality-gauge-normalized}
\J_{X,\varphi}(x)
=
\frac{\varphi(\norm{x}_X)}{\norm{x}_X} \J_X(x).
\end{equation}
In particular, all duality mappings associated with gauge functions determine the same norming functionals up to a positive scalar factor.

%Moreover, if we let $\Psi(t) \ov{\mathrm{def}}{=} \int_0^t \varphi(s) \d s$ for any $t \geq 0$, then Asplund's theorem gives
%\begin{equation}
%\label{duality-subdifferential}
%\frak{j}_{X,\varphi}(x)
%=
%\partial\big(\Psi\circ\norm{\cdot}_X\big)(x),
%\quad x\in X, 
%\end{equation}
%see \cite[Theorem 4.4 p.~26]{Cio90}. 


\begin{example} \normalfont
Suppose that $1 < p < \infty$. We will mainly use the gauge
\begin{equation}
\label{power-gauge}
\varphi_p(t)
\ov{\mathrm{def}}{=}t^{p-1}, \quad t \geq 0
\end{equation}
and we write
\begin{equation}
\label{Def-JpX}
\frak{j}_{p,X}
\ov{\mathrm{def}}{=}
\J_{X,\varphi_p}.
\end{equation}
Then we have
\begin{equation}
\label{relation-Jp-normalized}
\frak{j}_{p,X}(x)
\ov{\eqref{relation-duality-gauge-normalized}}{=}
\norm{x}_X^{p-2} \J_X(x),
\quad x \neq 0.
\end{equation}
If $X$ is smooth, then
\begin{equation}
\label{derivation-duality-mapping-2}
\nabla_x\left(\frac1p\norm{\cdot}_X^p\right)(y)
=
\Re\la y,\frak{j}_{p,X}(x)\ra_{X,X^*},
\quad
x\in X\setminus\{0\}, y\in X.
\end{equation}
In this case, for any $x\in X$, the element $\frak{j}_{p,X}(x)$ is characterized by the equalities
\begin{equation}
\label{properties-Jp}
\norm{\frak{j}_{p,X}(x)}_{X^*}
\ov{\eqref{Def-JX-gauge}}{=} \norm{x}_X^{p-1}
\quad\text{and}\quad
\big\la x,\frak{j}_{p,X}(x) \big\ra_{X,X^*}
\ov{\eqref{Def-JX-gauge}}{=} \norm{x}_X^p.
\end{equation}
\end{example}



%Equivalently, we have
%\begin{equation}
%\label{formula-J,p,X}
%\cal{J}_{p,X}
%=
%\partial\left(\frac{1}{p}\norm{\cdot}_X^p\right)
%\end{equation}
%by \eqref{duality-subdifferential}. In particular, for $x^*\in\cal{J}_{p,X}(x)$ we have

Now, suppose that the Banach space $X$ is reflexive, smooth and strictly convex. Then \cite[Proposition 4.7(e) p.~28]{Cio90} implies that the duality mapping associated with a gauge $\varphi$ is bijective with inverse given by the duality mapping on $X^*$ associated with the inverse gauge $\varphi^{-1}$. Since
\[
\varphi_p^{-1}(t)=t^{\frac{1}{p-1}}=t^{p^*-1}=\varphi_{p^*}(t),
\quad \frac1p+\frac1{p^*}=1,
\]
we obtain
\begin{equation}
\label{inverse-power-duality-mapping}
\frak{j}_{p^*,X^*}\frak{j}_{p,X}=\Id_X
\quad\text{and}\quad
\frak{j}_{p,X}\frak{j}_{p^*,X^*}=\Id_{X^*}.
\end{equation}


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Contractive projections and duality mappings}

The following lemma is a generalization of \cite[Lemma 1]{And66} and is fundamental for us.

\begin{lemma}
\label{Lemma-smooth}
Let $X$ be a smooth strictly convex Banach space. Suppose that $1 < p < \infty$. Let $P \co X \to X$ be a contractive projection. For any $x \in X$, we have
\begin{equation}
\label{power-duality-map-range-projection}
x \in \Ran P
\quad \Longleftrightarrow \quad
\frak{j}_{p,X}(x) \in \Ran P^*.
\end{equation}
%where $\frak{j}_{p,X}$ denotes the duality mapping associated with the gauge $\varphi_p(t)=t^{p-1}$ introduced in \eqref{power-gauge}.
\end{lemma}

\begin{proof}
Suppose first that $x \in \Ran P$. Since $P(x)=x$, using the contractivity of $P^* \co X^* \to X^*$, we obtain 
\begin{align*}
\MoveEqLeft
\norm{x}_X^p         
\ov{\eqref{properties-Jp}}{=} \big\la x,\frak{j}_{p,X}(x) \big\ra_{X,X^*} 
=\big\la P(x),\frak{j}_{p,X}(x) \big\ra_{X,X^*}  \\
&=\big\la x,P^*(\frak{j}_{p,X}(x)) \big\ra_{X,X^*}
\leq \norm{x}_X \norm{P^*(\frak{j}_{p,X}(x))}_{X^*}
\leq \norm{x}_X\norm{\frak{j}_{p,X}(x)}_{X^*}
\ov{\eqref{properties-Jp}}{=} \norm{x}_X^{p}.
\end{align*}
Consequently, if $x \neq 0$, we obtain $\big\la x,P^*(\frak{j}_{p,X}(x)) \big\ra_{X,X^*}
=
\norm{x}_X^p$ and
\[
\norm{P^*(\frak{j}_{p,X}(x))}_{X^*}
=
\norm{x}_X^{p-1}.
\]
Thus $P^*(\frak{j}_{p,X}(x))$ belongs to the duality mapping of $x$ associated with the gauge $\varphi_p$. Since this duality mapping is single-valued, we obtain $
P^*(\frak{j}_{p,X}(x))
=\frak{j}_{p,X}(x)$ and therefore $\frak{j}_{p,X}(x) \in \Ran P^*$. The case $x = 0$ is obvious.

Conversely, suppose that $\frak{j}_{p,X}(x)$ belongs to the subspace $\Ran P^*$. Then we have $
P^*(\frak{j}_{p,X}(x))
=\frak{j}_{p,X}(x)$ and hence
\begin{align*}
\MoveEqLeft
\norm{x}_X^p
\ov{\eqref{properties-Jp}}{=}
\la x,\frak{j}_{p,X}(x) \ra_{X,X^*} 
=\big\la x,P^*(\frak{j}_{p,X}(x)) \big\ra_{X,X^*}
=\big\la P(x),\frak{j}_{p,X}(x) \big\ra_{X,X^*} \\
&\leq \norm{P(x)}_X \norm{\frak{j}_{p,X}(x)}_{X^*}
\ov{\eqref{properties-Jp}}{=} \norm{P(x)}_X \norm{x}_X^{p-1}
\leq \norm{x}_X^p.         
\end{align*}
If $x \neq 0$, it follows that $\norm{P(x)}_X=\norm{x}_X$ and
\begin{equation}
\label{inter-82}
\big\la P(x),\frak{j}_{p,X}(x) \big\ra_{X,X^*}
=
\norm{x}_X^p.
\end{equation}
By \eqref{properties-Jp} and \eqref{inter-82}, the functional $
x^*\ov{\mathrm{def}}{=}\frac{\frak{j}_{p,X}(x)}{\norm{x}_X^{p-1}}$ satisfies $\norm{x^*}_{X^*}=1$ and
\[
\bigg\la \frac{x}{\norm{x}_X},x^*\bigg\ra_{X,X^*}
=
\bigg\la \frac{P(x)}{\norm{x}_X},x^*\bigg\ra_{X,X^*}
=1.
\]
Consequently, we have
\[
1
=
\bigg\la \frac{x+P(x)}{2\norm{x}_X},x^*\bigg\ra_{X,X^*}
\leq
\norm{\frac{x+P(x)}{2\norm{x}_X}}_X
\leq
\frac{\norm{x}_X+\norm{P(x)}_X}{2\norm{x}_X}
=1.
\]
Hence
\[
\norm{\frac{x+P(x)}{2\norm{x}_X}}_X=1.
\]
Since both $x/\norm{x}_X$ and $P(x)/\norm{x}_X$ belong to the unit sphere and $X$ is strictly convex, we obtain $P(x)=x$. We conclude that $x \in \Ran P$.
\end{proof}


\begin{remark} \normalfont
In Lemma~\ref{Lemma-smooth}, the implication from $\Ran P$ to $\Ran P^*$ only uses smoothness, whereas the converse implication only uses strict convexity.
\end{remark}

The preceding property actually characterizes smooth strictly convex Banach spaces for any fixed power gauge. 

\begin{prop}
\label{Prop-characterization-smooth-strict-convex}
Let $X$ be a Banach space, let $1<p<\infty$, and let $j_p \co X \to X^*$ satisfy 
\begin{equation}
\label{power-selec}
\norm{j_{p}(x)}_{X^*}
= \norm{x}_X^{p-1}
\quad\text{and}\quad
\big\la x,j_{p}(x) \big\ra_{X,X^*}
= \norm{x}_X^p, \quad x \in X.
\end{equation}
The following assertions are equivalent.
\begin{enumerate}
\item The Banach space $X$ is smooth and strictly convex, and $j_p=\frak{j}_{p,X}$.
\item For any contractive projection $P \co X \to X$ and any $x \in X$, we have
\begin{equation}
\label{characterization-ranges-power-duality-map}
x\in\Ran P
\quad\Longleftrightarrow\quad
j_p(x)\in\Ran P^*.
\end{equation}
\end{enumerate}
\end{prop}

\begin{proof}
1. $\Rightarrow 2.$: Suppose first that $X$ is smooth and strictly convex. The equivalence \eqref{characterization-ranges-power-duality-map} follows from Lemma~\ref{Lemma-smooth}.

2. $\Rightarrow 1.$: Conversely, assume \eqref{characterization-ranges-power-duality-map}. We first prove that $X$ is smooth. Let $u \in X$ satisfy $\norm{u}_X=1$, and let $\phi \in X^*$ be any norming functional for $u$, that is,
\begin{equation}
\label{inter-99}
\norm{\phi}_{X^*}=1
\quad\text{and}\quad
\phi(u)=1.
\end{equation}
Consider the rank-one operator $P_{\phi,u} \co X \to X$ defined by $P_{\phi,u}(x) \ov{\mathrm{def}}{=} \phi(x)u$. It is a contractive projection with $\Ran P_{\phi,u} = \mathbb{C}u$. Since $u \in \Ran P_{\phi,u}$, condition \eqref{characterization-ranges-power-duality-map} implies that $
j_p(u) \in \Ran P_{\phi,u}^*$. For any $g \in X^*$, we have
\[
P_{\phi,u}^*(g)
=g(u)\phi,
\]
and therefore $\Ran P_{\phi,u}^*=\mathbb{C}\phi$. Hence $j_p(u)=\lambda \phi$ for some scalar $\lambda$. Since $\norm{u}_X=1$, \eqref{power-selec} gives
\[
1=j_p(u)(u)
=\lambda \phi(u)
\ov{\eqref{inter-99}}{=} \lambda.
\]
Thus $j_p(u)=\phi$. Since $\phi$ was an arbitrary norming functional for $u$, the norming functional of $u$ is unique. Hence $X$ is smooth. By the single-valuedness of the duality mapping associated with $\varphi_p$, we also obtain $
j_p
=\frak{j}_{p,X}$. It remains to prove strict convexity. Suppose, by contradiction, that $X$ is not strictly convex. Then, by \cite[Proposition 1.2 p.~41]{Cio90}, there exist distinct elements $u,v \in X$ such that
\[
\norm{u}_X=\norm{v}_X=1
\quad\text{and}\quad
\norm{\frac{u+v}{2}}_X=1.
\]
Choose $\phi \in X^*$ with $\norm{\phi}_{X^*}=1$ and $
\phi\left(\frac{u+v}{2}\right)
=1$. Since $|\phi(u)|\leq1$ and $|\phi(v)|\leq1$, the equality $\phi(u)+\phi(v)=2$ implies
\[
\phi(u)=\phi(v)=1.
\]
By the smoothness proved previously and since $\norm{u}_X=\norm{v}_X=1$, we have
\[
j_p(u)
=\phi
=j_p(v).
\]
Consider again the contractive rank-one projection $P_{\phi,u}(x)=\phi(x)u$. Since $j_p(v)=\phi$ belongs to $\Ran P_{\phi,u}^*$, condition \eqref{characterization-ranges-power-duality-map} yields
\[
v \in \Ran P_{\phi,u}
=\mathbb{C}u.
\]
Hence $v=\lambda u$ for some scalar $\lambda$ with $|\lambda|=1$. Since $\phi(u)=\phi(v)=1$, we obtain $\lambda=1$, and therefore $u=v$, a contradiction. Thus $X$ is strictly convex.
\end{proof}

%\begin{remark} \normalfont
%\label{Remark-Calvert}
There is also a classical characterization of contractively complemented subspaces in terms of duality mappings. If $X$ is reflexive, smooth and strictly convex, Calvert showed that a closed subspace $Y$ of $X$ is the range of a contractive projection if and only if the image of $Y$ under the normalized duality mapping is a linear subspace of $X^*$, see \cite{Cal75}. This characterization is unchanged if one uses the power duality mapping $\frak{j}_{p,X}$. Indeed, by \cite[Proposition 4.7 (d),(f) pp.~27--28]{Cio90}, the normalized duality mapping and $\frak{j}_{p,X}$ have the same radial directions, and since $Y$ is a subspace, we have
\begin{equation}
\label{same-image-power-normalized}
\frak{j}_{p,X}(Y)=\J_X(Y).
\end{equation}
Consequently, we have the equivalence
\[
Y\text{ is contractively complemented}
\quad\Longleftrightarrow\quad
\frak{j}_{p,X}(Y)\text{ is a linear subspace of }X^*,
\]
under the assumptions of Calvert's theorem. Moreover, we have the following observation.

\begin{prop}
Let $X$ be a reflexive smooth strictly convex Banach space. Suppose that $1 < p < \infty$. Let $P \co X \to X$ be a contractive projection. If $Y=\Ran P$ then
\begin{equation}
\label{power-duality-map-range-equality}
\frak{j}_{p,X}(Y)=\Ran P^*.
\end{equation}
\end{prop}

\begin{proof}
Lemma~\ref{Lemma-smooth} gives the inclusion $\frak{j}_{p,X}(Y)\subset\Ran P^*$. Conversely, let $z \in \Ran P^*$ and let $p^*$ be the conjugate exponent of $p$. Since $X$ is reflexive, the space $X^*$ is smooth and strictly convex. Applying Lemma~\ref{Lemma-smooth} to the contractive projection $P^* \co X^* \to X^*$, we obtain
\[
\frak{j}_{p^*,X^*}(z) \in \Ran P^{**}.
\]
Under the canonical identification $X^{**}=X$, we have $P^{**}=P$ and therefore $\Ran P^{**}=Y$. Hence $\frak{j}_{p^*,X^*}(z)\in Y$. Consequently, we have
\[
\frak{j}_{p,X}\frak{j}_{p^*,X^*}(z)
\ov{\eqref{inverse-power-duality-mapping}}{=} z,
\]
and therefore $z \in \frak{j}_{p,X}(Y)$. This proves \eqref{power-duality-map-range-equality}.
\end{proof}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Case of noncommutative $\L^p$-spaces}

If the Banach space  $X$ is a noncommutative $\L^p$-space, the power duality mapping admits a particularly simple algebraic expression. This is the noncommutative generalization of the classical formula $\frak{j}_{p,\L^p(\Omega)}(f) = |f|^{p-1}\sgn(f)$ for any $f \in \L^p(\Omega)$, stated in \cite[Proposition 4.9 p.~72]{Cio90}, where $\Omega$ is a measure space.

\begin{prop}
\label{prop-J-lp}
Suppose that $1<p<\infty$. Let $\cal{M}$ be a von Neumann algebra equipped with a normal semifinite faithful trace. For this proposition, we identify $\L^{p^*}(\cal{M})$ with the dual of $\L^p(\cal{M})$ through the bilinear duality bracket
\begin{equation}
\label{bilinear-duality-bracket}
\la x,y \ra_{\L^p(\cal{M}),\L^{p^*}(\cal{M})}
\ov{\mathrm{def}}{=} \tr(xy).
\end{equation}
If $x \in \L^p(\cal{M})$ has polar decomposition $x=u|x|$, then
\begin{equation}
\label{comput-J}
\frak{j}_{p,\L^p(\cal{M})}(x)
= |x|^{p-1}u^*.
\end{equation}
\end{prop}

\begin{proof}
Recall that $\L^p(\cal{M})$ is strictly convex and smooth by \cite[Corollary 5.2]{PiX03}. Hence $\frak{j}_{p,\L^p(\cal{M})}$ is single-valued. Since $u^*u=s(|x|)$, we have
\[
\bnorm{|x|^{p-1}u^*}_{\L^{p^*}(\cal{M})}
=
\bnorm{|x|^{p-1}}_{\L^{p^*}(\cal{M})}
=
\norm{x}_{\L^p(\cal{M})}^{p-1}.
\]
Furthermore, using the tracial property,
\[
\big\la x,|x|^{p-1}u^*\big\ra_{\L^p(\cal{M}),\L^{p^*}(\cal{M})}
\ov{\eqref{bilinear-duality-bracket}}{=}
\tr\big(u|x|^pu^*\big)
=
\tr\big(|x|^pu^*u\big)
=
\tr(|x|^p)
=
\norm{x}_{\L^p(\cal{M})}^p.
\]
Thus $|x|^{p-1}u^*$ satisfies \eqref{properties-Jp}, and therefore it is equal to $\frak{j}_{p,\L^p(\cal{M})}(x)$.
\end{proof}

In particular, if $p=4$, we obtain $\frak{j}_{4,\L^4(\cal{M})}(x)=|x|^{3}u^*=x^*xx^*$.


\begin{remark} \normalfont
\label{Remark-sesquilinear-duality}
For $1<p<\infty$, let $p^* \ov{\mathrm{def}}{=} \frac{p}{p-1}$. In the remainder of the paper, we identify $\L^{p^*}(\cal M)$ with the dual of $\L^p(\cal M)$ through the sesquilinear pairing
\begin{equation}
\label{sesquilinear-pairing}
\la x,y\ra
=
\Tr(y^*x),
\quad
x\in\L^p(\cal M), y\in\L^{p^*}(\cal M),
\end{equation}
rather than through the bilinear pairing \eqref{bilinear-duality-bracket}. More precisely, the map
\[
\iota_p\co \L^{p^*}(\cal M)\to(\L^p(\cal M))^*,
\qquad
[\iota_p(y)](x)=\Tr(y^*x),
\]
is a conjugate linear isometric bijection. Under this identification, if $x=u|x|$ is the polar decomposition of $x$, the power duality mapping associated with the gauge $t\mapsto t^{p-1}$ is represented by
\begin{equation*}
\frak{j}_{p,\L^p(\cal M)}(x)=u|x|^{p-1},
\end{equation*}
in the sense that
\[
\J_{\L^p(\cal M),\varphi_p}
=
\iota_p\circ\frak{j}_{p,\L^p(\cal M)}.
\]
In particular, $\frak{j}_{2,\L^2(\cal M)}=\Id_{\L^2(\cal M)}$ and, when $p=4$, we have
\begin{equation}
\label{duality-map-L4}
\frak{j}_{4,\L^4(\cal M)}(x)
=
u|x|^3
=
xx^*x.
\end{equation}
Moreover, \eqref{inverse-power-duality-mapping} becomes
\begin{equation}
\label{inverse-power-duality-mapping-NC}
\frak{j}_{\frac43,\L^{\frac43}(\cal M)}
\frak{j}_{4,\L^4(\cal M)}
=
\Id_{\L^4(\cal M)},
\qquad
\frak{j}_{4,\L^4(\cal M)}
\frak{j}_{\frac43,\L^{\frac43}(\cal M)}
=
\Id_{\L^{\frac43}(\cal M)}.
\end{equation}
The formula \eqref{duality-map-L4} is especially adapted to our purposes, since polarization of the cubic map $x\mapsto xx^*x$ produces the Jordan triple product $(x,y,z)\mapsto\frac12(xy^*z+zy^*x)$. Moreover, this convention preserves rectangular supports. Indeed, if $e$ and $f$ are projections in $\cal M$ and $x\in e\L^p(\cal M)f$, then $\frak{j}_{p,\L^p(\cal M)}(x)$ belongs to $e\L^{p^*}(\cal M)f$, whereas the bilinear realization of the duality mapping belongs to the opposite corner $f\L^{p^*}(\cal M)e$.
\end{remark}





%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Tangent Hilbert geometry}
\label{sec-tangent-Hilbert-geometry}
\subsection{First order geometry}

We first explain the geometric idea underlying our approach. The basic principle is purely Banach geometric. Let $X$ be a smooth complex Banach space, considered below as a real Banach space whenever differentiation of real-valued functions is involved. Let $P\co X\to X$ be a contractive projection and fix $a\in\Ran P$ with $a\neq0$. Suppose that $1<p<\infty$ and consider the convex function
\begin{equation}
\label{def-Phi-p-tangent}
\Psi_p(x)\ov{\mathrm{def}}{=}\frac1p\norm{x}_X^p,
\quad x\in X.
\end{equation}
Since $P(a)=a$, for any $z \in X$ and any $t \in \R$, we have
\begin{equation}
\label{basic-tangent-contractivity}
\Psi_p(a+tP(z))
=
\Psi_p(P(a+tz))
\leq
\Psi_p(a+tz).
\end{equation}
Thus the geometry of the $p$-power of the norm near a point $a$ of the range is more curved in the ambient direction $z$ than in its projected direction $P(z)$. This elementary observation naturally leads to differential methods. Such methods have already proved useful in the study of contractive projections. In particular, second derivatives of norms were used in \cite[Section 3]{LRG07} to detect structural properties of their ranges. Since $X$ is smooth, the function $\Psi_p$ is Gâteaux differentiable at every nonzero point. By the identification of the power duality mapping with the Gâteaux derivative of $\frac1p\norm{\cdot}_X^p$, we have
\begin{equation}
\label{first-variation-Phi-p}
\nabla_a\Psi_p(z)
=
\Re\big\la z,\frak{j}_{p,X}(a)\big\ra_{X,X^*},
\quad z\in X.
\end{equation}
Here $z\mapsto\nabla_a\Psi_p(z)$ is a continuous real linear functional on the underlying real Banach space $X_{\R}$.
To exploit \eqref{basic-tangent-contractivity}, fix $z\in X$ and define
\[
F_z(t)
\ov{\mathrm{def}}{=}
\Psi_p(a+tz)-\Psi_p(a+tP(z)),
\quad t \in \R.
\]
By \eqref{basic-tangent-contractivity}, $F_z(t) \geq 0$ for any $t \in \R$, whereas $F_z(0)=0$. Hence $t=0$ is a minimum of $F_z$. Since $F_z$ is differentiable at $0$, we obtain $F_z'(0)=0$. Therefore
\begin{align}
\label{first-tangent-orthogonality}
0
=\nabla_a\Psi_p(z)-\nabla_a\Psi_p(P(z))
=\nabla_a\Psi_p(z-P(z)).
\end{align}
Using \eqref{first-variation-Phi-p}, this says
\[
\Re\big\la z-P(z),\frak{j}_{p,X}(a)\big\ra_{X,X^*}
=0.
\]
Applying the same identity to $\i z$ instead of $z$ gives
\[
\Im\big\la z-P(z),\frak{j}_{p,X}(a)\big\ra_{X,X^*}
=0,
\]
and hence
\begin{equation}
\label{complex-first-tangent-orthogonality}
\big\la z-P(z),\frak{j}_{p,X}(a)\big\ra_{X,X^*}=0.
\end{equation}
Thus the directions belonging to $\ker P$ are tangent, at $a$, to the level hypersurface of $\Psi_p$, or equivalently to the sphere through $a$. In particular, the first-order geometry is precisely the geometry encoded by the duality mapping.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Second order geometry}

Suppose now that $\Psi_p$ is twice Fr\'echet differentiable at $a$ as a real-valued function on $X_{\R}$. Since $F_z$ has a minimum at $0$, we have
\begin{equation}
\label{second-derivative-minimum}
F_z''(0)\geq0.
\end{equation}
Indeed, we have
\[
F_z''(0)
=
\lim_{t\to0}
\frac{F_z(t)+F_z(-t)-2F_z(0)}{t^2},
\]
and the numerator is positive. On the other hand, by the chain rule,
\[
F_z''(0)
=
\D_{\R}^2\Psi_p(a)[z,z]
-
\D_{\R}^2\Psi_p(a)[P(z),P(z)].
\]
Consequently,
\begin{equation}
\label{abstract-second-tangent-contraction}
\D_{\R}^2\Psi_p(a)[P(z),P(z)]
\leq
\D_{\R}^2\Psi_p(a)[z,z],
\quad z \in X.
\end{equation}
Since $\Psi_p$ is convex, the real symmetric bilinear form $\D_{\R}^2\Psi_p(a)$ is positive semidefinite. Indeed, for any $z\in X$, the function $ t\mapsto \Psi_p(a+tz)$ is convex on $\R$, and therefore
\[
\D_{\R}^2\Psi_p(a)[z,z]
=
\left.\frac{\d^2}{\d t^2}\right|_{t=0}\Psi_p(a+tz)
\geq 0.
\]
Consequently, the quadratic form $z\mapsto\D_{\R}^2\Psi_p(a)[z,z]$ defines a possibly degenerate local Hilbertian metric. If
\[
\cal N_a
\ov{\mathrm{def}}{=}
\{z\in X:\D_{\R}^2\Psi_p(a)[z,z]=0\},
\]
then the Cauchy--Schwarz inequality for positive semidefinite bilinear forms shows the equality $\D_{\R}^2\Psi_p(a)[z,w]=0$ for any $z\in\cal N_a$ and every $w\in X$. Hence $\D_{\R}^2\Psi_p(a)$ induces a positive definite real inner product on $X/\cal N_a$, whose completion is a real Hilbert space. The Levi form introduced below may be viewed as the Hermitian counterpart of this real second-order geometry.


The following result says that the Levi form measures exactly the second-order deviation of $\Phi$ from the mean-value property along complex lines.

\begin{prop}
Let $\Omega$ be an open subset of a complex Banach space $X$ and let $\Phi\co\Omega\to\R$ be of class $\mathrm C^2$. For any $z \in X$ and any $a \in \Omega$, we have
\begin{equation}
\label{Levi-average-limit}
\scr{L}_{\Phi,a}(z)
=\lim_{r \to 0}\frac1{r^2}
\left(
\frac1{2\pi}\int_0^{2\pi}\Phi(a+r\e^{\i t}z) \d t-\Phi(a)
\right).
\end{equation}
\end{prop}

\begin{proof}
For any $a \in \Omega$ and any $z \in X$, Taylor's formula \cite[(8.14.3) p.~190]{Die69} on the underlying real Banach space gives, as $r \to 0$,
\begin{equation}
\label{Taylor-1}
\Phi(a+r\e^{\i t}z)
=\Phi(a)+r\D_{\R}\Phi(a)(\e^{\i t}z)
+\frac{r^2}{2}\D_{\R}^2\Phi(a)(\e^{\i t}z,\e^{\i t}z)+o(r^2),
\end{equation}
where the remainder is uniform in $t \in [0,2\pi]$. Since $\e^{\i t}z = (\cos t)z+(\sin t)\i z$, we have
\[
\frac1{2\pi} \int_0^{2\pi}\D_{\R} \Phi(a)(\e^{\i t}z) \d t
=\frac1{2\pi} \int_0^{2\pi} \big[(\cos t)\D_{\R} \Phi(a)(z) + (\sin t)\D_{\R}\Phi(a)(\i z)\big] \d t
=0
\]
and
\begin{align*}
\MoveEqLeft
\frac1{2\pi} \int_0^{2\pi}
\D_{\R}^2\Phi(a)(\e^{\i t}z,\e^{\i t}z) \d t 
=\frac1{2\pi} \int_0^{2\pi}
(\cos t)^2\D_{\R}^2  \Phi(a)(z,z) + (\cos t \sin t)\D_{\R}^2\Phi(a)(z,\i z) \\
&+(\sin t \cos t)\D_{\R}^2\Phi(a)(\i z,z)+(\sin t)^2\D_{\R}^2\Phi(a)(\i z,\i z) \d t \\
&=\frac12\left(\D_{\R}^2\Phi(a)(z,z)+\D_{\R}^2\Phi(a)(\i z,\i z)\right) 
\ov{\eqref{Levi-form-diagonal}}{=} 2\scr{L}_{\Phi,a}(z).
\end{align*}
Consequently, we have
\begin{equation}
\label{circular-average-Levi}
\frac1{2\pi} \int_0^{2\pi} \Phi(a+r\e^{\i t}z) \d t
\ov{\eqref{Taylor-1}}{=} \Phi(a)+r^2\scr{L}_{\Phi,a}(z)+o(r^2).
\end{equation}
We conclude that
\begin{equation*}
\scr{L}_{\Phi,a}(z)
\ov{\eqref{circular-average-Levi}}{=} \lim_{r \to 0} \frac1{r^2}
\left(
\frac1{2\pi} \int_0^{2\pi} \Phi(a+r\e^{\i t}z) \d t - \Phi(a)
\right).
\end{equation*}
\end{proof}

There is also a useful positivity interpretation. Recall that a $\mathrm{C}^2$ real-valued function $\Phi \co \Omega \to \R$ on an open subset of a complex Banach space is plurisubharmonic precisely when its restriction to every complex affine line is subharmonic, or equivalently when $
\scr{L}_{\Phi,a}(z)
\ov{\eqref{Levi-def-pos}}{\geq} 0$ 
for any $a \in \Omega$ and any $z \in X$. Since $\Phi_p=\norm{\cdot}_X^p$ is convex for $p \geq 1$, its restriction to every complex affine line is a convex, hence subharmonic, function on $\mathbb{C}$. Consequently, wherever $\Phi_p$ is of class $C^2$,
\begin{equation}
\label{Levi-p-positive}
\scr{L}_{\Phi_p,a}(z)
\geq0.
\end{equation}




%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Hilbertian geometry associated to a projection contracting a plurisubharmonic function}

A contractive projection is automatically contractive for the infinitesimal Hermitian geometry determined by the Levi form at every point of its range.

\begin{prop}
\label{prop-Levi-contractive-projection}
Let $\Omega$ be an open subset of a complex Banach space $X$. Consider a plurisubharmonic function $\Phi \co \Omega \to \R$ of class $\mathrm{C}^2$. Let $P \co X \to X$ be a bounded projection such that $P(\Omega) \subset \Omega$  satisfying
\begin{equation}
\label{Phi-contractivity}
\Phi(P(x))
\leq \Phi(x),\quad x \in \Omega.
\end{equation}
Let $a \in \Omega \cap \Ran P$. Then
\begin{equation}
\label{Levi-projection-contraction}
\scr{L}_{\Phi,a}(P(z))
\leq \scr{L}_{\Phi,a}(z),
\quad z \in X.
\end{equation}
Let
\begin{equation}
\label{Levi-null-space}
\cal{N}_{\Phi,a}
\ov{\mathrm{def}}{=}
\{z \in X : \scr{L}_{\Phi,a}(z)=0 \}.
\end{equation}
Then $P$ leaves $\cal N_{\Phi,a}$ invariant and induces an orthogonal projection on the Hilbert space obtained by completing $X/\cal N_{\Phi,a}$ for the inner product induced by $\scr{L}_{\Phi,a}(\cdot,\cdot)$. Moreover, we have
\begin{equation}
\label{Levi-orthogonality}
\scr{L}_{\Phi,a}(P(z),y)
=\scr{L}_{\Phi,a}(z,y),
\quad z\in X,\quad y \in \Ran P.
\end{equation}
\end{prop}

\begin{proof}
Let $z \in X$. Note that $P(a)=a$. For all sufficiently small $r > 0$ and any $t \in \R$, using the contractivity \eqref{Phi-contractivity}, we see that 
\[
\Phi(a+r\e^{\i t}P(z))
=\Phi(P(a)+r\e^{\i t}P(z))
=\Phi(P(a+r\e^{\i t}z))
\ov{\eqref{Phi-contractivity}}{\leq} \Phi(a+r\e^{\i t}z).
\]
Averaging over $t\in[0,2\pi]$ and using \eqref{circular-average-Levi}, we obtain
\[
\Phi(a)+r^2\scr{L}_{\Phi,a}(Pz)+o(r^2)
\leq
\Phi(a)+r^2\scr{L}_{\Phi,a}(z)+o(r^2).
\]
After division by $r^2$ and passage to the limit $r\to0^+$, this gives \eqref{Levi-projection-contraction}.

Since $\Phi$ is plurisubharmonic, its Levi form is positive semidefinite. Therefore the Cauchy--Schwarz inequality for positive sesquilinear forms shows that
\[
\scr{L}_{\Phi,a}(z)=0
\quad\Rightarrow\quad
\scr{L}_{\Phi,a}(z,w)=0,\quad w\in X.
\]
Consequently, $\cal N_{\Phi,a}$ is the radical of $\scr{L}_{\Phi,a}(\cdot,\cdot)$ and the latter induces an inner product on $X/\cal N_{\Phi,a}$. If $z \in \cal N_{\Phi,a}$, then \eqref{Levi-projection-contraction} implies
\[
0 \leq \scr{L}_{\Phi,a}(Pz)
\ov{\eqref{Levi-projection-contraction}}{\leq} \scr{L}_{\Phi,a}(z)
\ov{\eqref{Levi-null-space}}{=} 0,
\]
and therefore $P(z)$ belongs to $\cal N_{\Phi,a}$. Hence $P$ induces a projection $\widetilde{P} \co X/\cal N_{\Phi,a} \to X/\cal N_{\Phi,a}$, $[z] \mapsto [P(z)]$ on the quotient space $X/\cal N_{\Phi,a}$. Moreover, we have
\[
\scr{L}_{\Phi,a}(\widetilde{P}[z])
=
\scr{L}_{\Phi,a}(Pz)
\ov{\eqref{Levi-projection-contraction}}{\leq}
\scr{L}_{\Phi,a}(z).
\]
So  the map $\widetilde{P}$ is contractive for the induced Hilbertian norm. It therefore extends to a contractive projection on the Hilbert space completion of $X/\cal N_{\Phi,a}$. Recall that every contractive projection on a Hilbert space is orthogonal by \cite[Proposition 3.3 p.~37]{Con90}. Since $[y]$ belongs to the range of the induced projection for any $y \in \Ran P$, we obtain
\[
\scr{L}_{\Phi,a}(z-Pz,y)
=0,
\]
which is \eqref{Levi-orthogonality}.
\end{proof}


The Levi form also defines a conjugate linear map into the dual space.
\begin{defi}
Let $\Omega$ be an open subset of a complex Banach space $X$, let $\Phi \co \Omega \to \R$ be of class $\mathrm{C}^2$ and let $a\in\Omega$. We introduce the conjugate linear operator $\Lambda_{\Phi,a} \co X \to X^*$ defined by
\begin{equation}
\label{Levi-dual-map}
\big\la z,\Lambda_{\Phi,a}(y) \big\ra_{X,X^*}
\ov{\mathrm{def}}{=}
\scr{L}_{\Phi,a}(z,y),
\quad z,y \in X.
\end{equation}
We call $\Lambda_{\Phi,a}$ the Levi dual map of $\Phi$ at $a$.
\end{defi}


\begin{prop}
Under the assumptions of Proposition~\ref{prop-Levi-contractive-projection}, let $a\in\Omega\cap\Ran P$. Then
\begin{equation}
\label{Levi-adjoint-fixed}
P^*\Lambda_{\Phi,a}(y)
=\Lambda_{\Phi,a}(y),
\quad y \in \Ran P.
\end{equation}
\end{prop}

\begin{proof}
For any $z \in X$ and $y \in \Ran P$, we have
\begin{align*}
\MoveEqLeft
\big \la z,P^*\Lambda_{\Phi,a}(y) \big\ra_{X,X^*}
=\big \la P(z),\Lambda_{\Phi,a}(y) \big\ra_{X,X^*}
\ov{\eqref{Levi-dual-map}}{=} \scr{L}_{\Phi,a}(Pz,y) \\
&\ov{\eqref{Levi-orthogonality}}{=} \scr{L}_{\Phi,a}(z,y)
\ov{\eqref{Levi-dual-map}}{=} \big\la z,\Lambda_{\Phi,a}(y) \big\ra_{X,X^*}.         
\end{align*}
We conclude that $P^*\Lambda_{\Phi,a}(y)=\Lambda_{\Phi,a}(y)$.
\end{proof}





The next result may be regarded as an infinitesimal bicontractivity principle. A contractive projection $P$ need not have a contractive complement $\Id-P$ for the original Banach norm. Nevertheless, at every point $a\in \Omega \cap \Ran P$, the projection and its complement are simultaneously contractive for the Hilbertian geometry determined by the Levi form. Equivalently, the Banach space symmetry $2P-\Id$, which need not be an isometry of $X$, becomes an involutive isometry after passing to the Levi Hilbert space at $a$.



\begin{cor}
\label{cor-Levi-Pythagoras-bicontractivity}
Let $\Omega$ be an open subset of a complex Banach space $X$. Consider a plurisubharmonic function $\Phi \co \Omega \to \R$ of class $\mathrm{C}^2$. Let $P \co X \to X$ be a bounded projection such that $P(\Omega) \subset \Omega$ satisfying $\Phi(P(x)) \leq \Phi(x)$ for any $x \in \Omega$. Let $a \in \Omega \cap \Ran P$. Then, for any $z,w \in X$, we have
\begin{equation}
\label{Levi-Pythagoras}
\scr{L}_{\Phi,a}(z,w)
=
\scr{L}_{\Phi,a}(Pz,Pw)
+
\scr{L}_{\Phi,a}((\Id-P)z,(\Id-P)w).
\end{equation}
In particular, we have
\begin{equation}
\label{Levi-Pythagoras-diagonal}
\scr{L}_{\Phi,a}(z)
=
\scr{L}_{\Phi,a}(Pz)
+
\scr{L}_{\Phi,a}((\Id-P)z),
\quad z\in X.
\end{equation}
Consequently, both $P$ and $\Id-P$ are contractive for the Levi seminorm at $a$:
\begin{equation}
\label{Levi-bicontractivity}
\scr{L}_{\Phi,a}(Pz)
\leq
\scr{L}_{\Phi,a}(z),
\qquad
\scr{L}_{\Phi,a}((\Id-P)z)
\leq
\scr{L}_{\Phi,a}(z).
\end{equation}
Moreover, the symmetry
\begin{equation}
\label{Levi-symmetry}
S\ov{\mathrm{def}}{=}2P-\Id
\end{equation}
is an isometry for the Levi form:
\begin{equation}
\label{Levi-symmetry-isometry}
\scr{L}_{\Phi,a}(Sz,Sw)
=
\scr{L}_{\Phi,a}(z,w),
\quad z,w \in X.
\end{equation}
Hence $S$ leaves $\cal{N}_{\Phi,a}$ invariant and induces an involutive unitary operator on the Hilbert space $\cal{H}_{\Phi,a}$ obtained by completing $X/\cal{N}_{\Phi,a}$ for the inner product induced by $\scr{L}_{\Phi,a}(\cdot,\cdot)$. In particular, the operators induced by $P$ and $\Id-P$ on this Hilbert space are complementary orthogonal projections.
\end{cor}

\begin{proof}
Let $z,w \in X$. Since $P(w) \in \Ran P$, applying \eqref{Levi-orthogonality} with $(\Id-P)z$ instead of $z$, we obtain
\begin{equation}
\label{inter-525}
\scr{L}_{\Phi,a}((\Id-P)z,Pw)
\ov{\eqref{Levi-orthogonality}}{=}
\scr{L}_{\Phi,a}(P(\Id-P)z,Pw)
=0,
\end{equation}
where the last equality follows from $P(\Id-P)=0$. Since the Levi form is Hermitian, \eqref{inter-525} also gives
\[
\scr{L}_{\Phi,a}(Pz,(\Id-P)w)
=\overline{\scr{L}_{\Phi,a}((\Id-P)w,Pz)}
\ov{\eqref{Levi-orthogonality}}{=}\overline{\scr{L}_{\Phi,a}(P(\Id-P)w,Pz)}
=0.
\]
Using the decompositions $z=Pz+(\Id-P)z$ and $w=Pw+(\Id-P)w$, sesquilinearity gives
\begin{align*}
\MoveEqLeft
\scr{L}_{\Phi,a}(z,w)
=\scr{L}_{\Phi,a}\big(Pz+(\Id-P)z,Pw+(\Id-P)w\big) \\
&=\scr{L}_{\Phi,a}\big(Pz,Pw\big) +\scr{L}_{\Phi,a}\big(Pz,(\Id-P)w\big) 
+\scr{L}_{\Phi,a}\big((\Id-P)z,Pw\big) \\
&+\scr{L}_{\Phi,a}\big((\Id-P)z,(\Id-P)w\big)  
\ov{\eqref{inter-525}}{=} \scr{L}_{\Phi,a}(Pz,Pw) + \scr{L}_{\Phi,a}((\Id-P)z,(\Id-P)w),
\end{align*}
which proves \eqref{Levi-Pythagoras}. Taking $w=z$ gives \eqref{Levi-Pythagoras-diagonal}. Since $\Phi$ is plurisubharmonic, its Levi form is positive semidefinite, and hence both terms on the right-hand side of \eqref{Levi-Pythagoras-diagonal} are nonnegative. This proves \eqref{Levi-bicontractivity}.

Now, we prove \eqref{Levi-symmetry-isometry}, i.e., $\scr{L}_{\Phi,a}(Sz,Sw) = \scr{L}_{\Phi,a}(z,w)$ for any $z,w \in X$. Since
\begin{equation}
\label{def-S-inter-46}
Sz
\ov{\eqref{Levi-symmetry}}{=}  Pz-(\Id-P)z
\quad\text{and}\quad
Sw
\ov{\eqref{Levi-symmetry}}{=} Pw-(\Id-P)w,
\end{equation}
we have
\begin{align*}
\MoveEqLeft
\scr{L}_{\Phi,a}(Sz,Sw)
\ov{\eqref{def-S-inter-46}}{=} \scr{L}_{\Phi,a}\big(Pz-(\Id-P)z,Pw-(\Id-P)w\big) =\scr{L}_{\Phi,a}\big(Pz,Pw\big)\\
&-\scr{L}_{\Phi,a}\big((\Id-P)z,Pw\big)-\scr{L}_{\Phi,a}\big(Pz,(\Id-P)w\big)+\scr{L}_{\Phi,a}\big((\Id-P)z,(\Id-P)w\big)\\
&=\scr{L}_{\Phi,a}(Pz,Pw)
+
\scr{L}_{\Phi,a}((\Id-P)z,(\Id-P)w) 
=
\scr{L}_{\Phi,a}(z,w),         
\end{align*}
since the mixed terms vanish by the preceding orthogonality relations \eqref{inter-525}. Moreover, $S^2=\Id$. Thus $S$ preserves the null space $\cal N_{\Phi,a}$ and induces an involutive isometry on the associated Hilbert completion. Since this induced operator is an involutive isometry on a Hilbert space, it is a selfadjoint unitary. The identities
\[
P=\frac12(\Id+S)
\quad\text{and}\quad
\Id-P=\frac12(\Id-S)
\]
then show that the operators induced by $P$ and $\Id-P$ are complementary orthogonal projections.
\end{proof}



\begin{cor}
\label{cor-Levi-intertwining}
Under the assumptions of Proposition~\ref{prop-Levi-contractive-projection}, let $a\in\Omega\cap\Ran P$. Then the Levi dual map $\Lambda_{\Phi,a} \co X \to X^*$ defined by \eqref{Levi-dual-map} satisfies the intertwining relation
\begin{equation}
\label{Levi-intertwining}
P^*\Lambda_{\Phi,a}
=
\Lambda_{\Phi,a}P.
\end{equation}
Equivalently, $P$ is selfadjoint with respect to the Levi form:
\begin{equation}
\label{Levi-selfadjointness}
\scr{L}_{\Phi,a}(Pz,y)
=
\scr{L}_{\Phi,a}(z,Py),
\quad z,y \in X.
\end{equation}
In particular, if the Levi form $\scr{L}_{\Phi,a}(\cdot,\cdot)$ is nondegenerate, then
\begin{equation}
\label{Levi-kernel-orthogonal-range}
\ker P
=
(\Ran P)^{\perp_{\Phi,a}},
\end{equation}
where
\[
(\Ran P)^{\perp_{\Phi,a}}
\ov{\mathrm{def}}{=}
\{ z \in X : \scr{L}_{\Phi,a}(z,y)=0\text{ for any }y \in \Ran P\}.
\]
Consequently, in this case $P$ is uniquely determined by its range as the orthogonal projection onto $\Ran P$ for the Hilbertian geometry induced by the Levi form at $a$.
\end{cor}

\begin{proof}
Let $z,y \in X$. We first observe that
\[
\scr{L}_{\Phi,a}(Pz,(\Id-P)y)=0.
\]
Indeed, by the Hermitian property of the Levi form and \eqref{Levi-orthogonality},
\[
\scr{L}_{\Phi,a}(Pz,(\Id-P)y)
= \ovl{\scr{L}_{\Phi,a}((\Id-P)y,Pz)}
\ov{\eqref{Levi-orthogonality}}{=}
\ovl{\scr{L}_{\Phi,a}(P(\Id-P)y,Pz)}
=0.
\]
Consequently, since $y = (\Id-P)(y)+Py$, we have
\[
\scr{L}_{\Phi,a}(Pz,y)
=\scr{L}_{\Phi,a}\big(Pz,(\Id-P)(y)+Py\big)
=
\scr{L}_{\Phi,a}(Pz,Py).
\]
Since $P(y)$ belongs to the subspace $\Ran P$, another application of \eqref{Levi-orthogonality} yields
\[
\scr{L}_{\Phi,a}(Pz,Py)
=
\scr{L}_{\Phi,a}(z,Py).
\]
This proves \eqref{Levi-selfadjointness}. Now, for any $z,y\in X$, using \eqref{Levi-dual-map}, we obtain
\begin{align*}
\MoveEqLeft
\big\la z,P^*\Lambda_{\Phi,a}(y)\big\ra_{X,X^*}
=
\big\la Pz,\Lambda_{\Phi,a}(y)\big\ra_{X,X^*}
\ov{\eqref{Levi-dual-map}}{=}
\scr{L}_{\Phi,a}(Pz,y) \\
&=
\scr{L}_{\Phi,a}(z,Py)
\ov{\eqref{Levi-dual-map}}{=}
\big\la z,\Lambda_{\Phi,a}(Py) \big\ra_{X,X^*}.         
\end{align*}
Hence
\[
P^*\Lambda_{\Phi,a}(y)
=\Lambda_{\Phi,a}(Py),
\]
which is \eqref{Levi-intertwining}. Suppose finally that the Levi form is nondegenerate. If $z \in \ker P$ and $y\in \Ran P$, then \eqref{Levi-selfadjointness} gives
\[
\scr{L}_{\Phi,a}(z,y)
=
\scr{L}_{\Phi,a}(z,Py)
\ov{\eqref{Levi-selfadjointness}}{=}
\scr{L}_{\Phi,a}(Pz,y)
=
0.
\]
Thus
\[
\ker P
\subset (\Ran P)^{\perp_{\Phi,a}}.
\]
Conversely, let $z \in (\Ran P)^{\perp_{\Phi,a}}$. Since $P(z)$ belongs to $\Ran P$, we have
\[
0
=
\scr{L}_{\Phi,a}(z,Pz)
=\scr{L}_{\Phi,a}(z,P^2z)
=\scr{L}_{\Phi,a}(Pz,Pz).
\]
The nondegeneracy of the Levi form implies $Pz=0$. Hence $z \in \ker P$, proving \eqref{Levi-kernel-orthogonal-range}.
\end{proof}

\begin{remark} \normalfont
\label{rem-Levi-intertwining-meaning}
Relation \eqref{Levi-intertwining} is stronger than the fixed-point relation \eqref{Levi-adjoint-fixed}. It says that the Levi dual map intertwines the projection $P$ with its Banach adjoint $P^*$. Thus, although $P$ need not be selfadjoint for any ambient Hilbert space structure, it becomes selfadjoint for every Levi geometry based at a point $a\in\Omega\cap\Ran P$. When the Levi form at $a$ is nondegenerate, the decomposition
\[
X
= \Ran P \oplus \ker P
\]
is therefore orthogonal for this Hilbertian geometry.
\end{remark}

\begin{remark} \normalfont
\label{rem-Levi-orthogonality-first-order}
The Levi orthogonality relation \eqref{Levi-orthogonality} can also be obtained by differentiating the first-order condition along the range. Indeed, let $\Phi\co\Omega\to\R$ be of class $\C^2$, let $P$ be a bounded projection such that $P(\Omega)\subset\Omega$ and $\Phi\circ P\leq\Phi$, and fix $a\in \Omega \cap \Ran P$. For any $a'\in\Omega\cap\Ran P$ and any $k\in\ker P$, the function
\[
t\mapsto \Phi(a'+tk)-\Phi(a')
\]
has a minimum at $0$. Hence
\[
\D_{\R}\Phi(a')[k]=0.
\]
Differentiating this identity at $a'=a$ in a direction $y\in\Ran P$ gives
\[
\D_{\R}^2\Phi(a)[y,k]=0.
\]
Since both $\Ran P$ and $\ker P$ are complex subspaces, applying the same argument to $\i y$ and $\i k$ and using \eqref{Levi-form-general} yields
\[
\scr L_{\Phi,a}(k,y)=0,
\quad
k \in \ker P, y \in \Ran P.
\]
Thus the orthogonality itself is already encoded in the variation of the first-order condition. Plurisubharmonicity supplies the positivity which turns the Levi form into a possibly degenerate Hilbertian geometry; this Hilbertian realization is what allows us to introduce relative positive operators and their imaginary powers.
\end{remark}


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Relative metric operators and modular groups}
\label{sec-relative-modular}

\begin{prop}
\label{prop-relative-Riesz-operator}
Let $Y$ be a complex vector space equipped with two positive definite Hermitian forms $\langle\cdot,\cdot\rangle_0$ and $\langle\cdot,\cdot\rangle_1$, linear in the first variable. Let $Y'$ denote the algebraic dual of $Y$. For $j\in\{0,1\}$, consider the conjugate linear Riesz map $\cal R_j\co Y\to Y'$ defined by
\begin{equation}
\label{relative-metric-Riesz-maps}
\cal R_j(y)(x)
\ov{\mathrm{def}}{=}
\langle x,y\rangle_j,
\quad x,y\in Y.
\end{equation}
Suppose that
\begin{equation}
\label{common-Riesz-range}
\cal R_0(Y)=\cal R_1(Y).
\end{equation}
Then the formula
\begin{equation}
\label{relative-metric-operator-definition}
\Delta
\ov{\mathrm{def}}{=}
\cal R_0^{-1}\cal R_1
\end{equation}
defines a linear bijection $\Delta\co Y\to Y$. It is the unique linear operator satisfying
\begin{equation}
\label{relative-metric-comparison}
\langle x,y\rangle_1
=
\langle x,\Delta y\rangle_0,
\quad
x,y \in Y.
\end{equation}
Moreover, $\Delta$ is positive and symmetric with respect to $\langle\cdot,\cdot\rangle_0$, in the sense that
\begin{equation}
\label{relative-metric-selfadjoint-positive}
\langle\Delta x,y\rangle_0
=
\langle x,\Delta y\rangle_0
\quad\text{and}\quad
\langle x,\Delta x\rangle_0
=
\langle x,x\rangle_1>0
\end{equation}
for any $x,y\in Y$ with $x\neq0$.
If $Y$ is finite-dimensional, condition \eqref{common-Riesz-range} is automatic, $\Delta$ has spectrum contained in $(0,\infty)$ and
\begin{equation}
\label{relative-metric-square-root}
\langle x,y\rangle_1
=
\big\la \Delta^{\frac12}x,\Delta^{\frac12}y\big\ra_0,
\quad
x,y \in Y.
\end{equation}
\end{prop}



\begin{proof}
For $j\in\{0,1\}$, the map $\cal R_j$ is conjugate linear. Indeed, for $\lambda\in\mathbb C$ and $x,y\in Y$, we have
\[
\cal R_j(\lambda y)(x)
=
\langle x,\lambda y\rangle_j
=
\overline{\lambda}\langle x,y\rangle_j
=
\overline{\lambda}\cal R_j(y)(x).
\]
Moreover, $\cal R_j$ is injective. Indeed, if $\cal R_j(y)=0$, then
\[
\langle y,y\rangle_j
=
\cal R_j(y)(y)
=
0,
\]
and the positive definiteness of $\langle\cdot,\cdot\rangle_j$ gives $y=0$.
By \eqref{common-Riesz-range}, the map $\cal R_0^{-1}\cal R_1$ is well-defined on $Y$. Since both $\cal R_0^{-1}$ and $\cal R_1$ are conjugate linear, their composition $\Delta$ is linear. It is bijective, with inverse $
\Delta^{-1}
=
\cal R_1^{-1}\cal R_0$. For $x,y\in Y$, the definition of $\Delta$ gives
\[
\cal R_0(\Delta y)
=
\cal R_1(y).
\]
Evaluating both sides at $x$, we obtain
\[
\langle x,\Delta y\rangle_0
=
\cal R_0(\Delta y)(x)
=
\cal R_1(y)(x)
=
\langle x,y\rangle_1,
\]
which proves \eqref{relative-metric-comparison}.
Suppose that $T\co Y\to Y$ is another linear operator satisfying
\[
\langle x,y\rangle_1
=
\langle x,Ty\rangle_0,
\quad x,y\in Y.
\]
Then $\cal R_0(Ty)=\cal R_1(y)=\cal R_0(\Delta y)$. Since $\cal R_0$ is injective, we obtain $Ty=\Delta y$ for any $y \in Y$. This proves uniqueness.
Using \eqref{relative-metric-comparison} and the Hermitian property of the two forms, we have
\begin{align*}
\langle\Delta x,y\rangle_0
&=
\overline{\langle y,\Delta x\rangle_0}
=
\overline{\langle y,x\rangle_1}
=
\langle x,y\rangle_1
=
\langle x,\Delta y\rangle_0.
\end{align*}
Thus $\Delta$ is symmetric with respect to $\langle\cdot,\cdot\rangle_0$. Furthermore, we have
\[
\langle x,\Delta x\rangle_0
\ov{\eqref{relative-metric-comparison}}{=}
\langle x,x\rangle_1>0
\]
for any nonzero $x\in Y$. Hence $\Delta$ is positive. If $Y$ is finite-dimensional, the injective conjugate linear maps $\cal R_0,\cal R_1 \co Y \to Y'$ are automatically surjective. Thus \eqref{common-Riesz-range} is automatic. The spectral theorem applied to the positive selfadjoint operator $\Delta$ gives $\sigma(\Delta)\subset(0,\infty)$ and a positive square root $\Delta^{1/2}$. Hence
\[
\big\langle\Delta^{\frac12}x,\Delta^{\frac12}y\big\rangle_0
=
\langle x,\Delta y\rangle_0
\ov{\eqref{relative-metric-comparison}}{=}
\langle x,y\rangle_1,
\]
which proves \eqref{relative-metric-square-root}.
\end{proof}

\begin{remark} \normalfont
\label{rem-finite}
In finite dimension, $\Delta$ is a positive invertible selfadjoint operator.
\end{remark}

\begin{remark} \normalfont
\label{rem-relative-metric-Riesz-range}
The equality of the Riesz ranges in \eqref{common-Riesz-range}, rather than the surjectivity of the Riesz maps onto the whole algebraic dual, is the essential assumption. This formulation is useful when both Hermitian forms are represented in a common dual subspace, as happens below for the Levi forms and for the left and right tangent forms taking their Riesz representatives in $\Ran P^\dagger$.
\end{remark}

The family of Levi forms associated with points of the range can be compared by positive operators. This gives an intrinsic relative modular structure. Such an operator is obtained by raising one index of the second metric with the first one.

\begin{prop}
\label{prop-relative-operator-Hilbert-forms}
Let $Y$ be a finite-dimensional complex vector space equipped with two positive definite Hermitian forms $h_0$ and $h_1$, which we assume to be linear in the first variable. There exists a unique linear operator $\Delta_{1|0}\co Y\to Y$ such that
\begin{equation}
\label{relative-Hilbert-forms}
h_1(x,y)
=
h_0(x,\Delta_{1|0}y),
\quad
x,y\in Y.
\end{equation}
The operator $\Delta_{1|0}$ is invertible, positive and selfadjoint with respect to $h_0$. More precisely,
\begin{equation}
\label{relative-operator-selfadjoint}
h_0(\Delta_{1|0}x,y)
=
h_0(x,\Delta_{1|0}y),
\quad
x,y\in Y,
\end{equation}
and
\begin{equation}
\label{relative-operator-positive}
h_0(x,\Delta_{1|0}x)
=
h_1(x,x)>0,
\quad
x\in Y\setminus\{0\}.
\end{equation}
Consequently,
\begin{equation}
\label{relative-square-root-Hilbert-forms}
h_1(x,y)
=
h_0\left(\Delta_{1|0}^{\frac12}x,\Delta_{1|0}^{\frac12}y\right),
\quad
x,y \in Y.
\end{equation}
If $G_0$ and $G_1$ denote the Gram matrices of $h_0$ and $h_1$ in the same basis, so that
\[
h_j(x,y)=[y]^*G_j[x],
\quad
j \in \{0,1\},
\]
then
\begin{equation}
\label{relative-operator-Gram-matrices}
[\Delta_{1|0}]
=
G_0^{-1}G_1.
\end{equation}
We call $\Delta_{1|0}$ the \emph{relative operator} of $h_1$ with respect to $h_0$.
\end{prop}

\begin{proof}
For $j \in \{0,1\}$, consider the Riesz map $\cal R_j\co Y\to Y^*$ defined by
$$
\cal R_j(y)(x)
\ov{\mathrm{def}}{=}
h_j(x,y).
$$
Since $h_j$ is positive definite, $\cal R_j$ is injective. Since $Y$ is finite-dimensional, $\dim Y=\dim Y^*$, and therefore $\cal R_j$ is bijective. Notice that $\cal R_j$ is conjugate linear. Hence the composition
\begin{equation}
\label{abstract-relative-Riesz-operator}
\Delta_{1|0}
\ov{\mathrm{def}}{=}
\cal R_0^{-1}\cal R_1
\end{equation}
is linear. By definition,
\[
\cal R_0(\Delta_{1|0}y)=\cal R_1(y),
\]
and therefore, for any $x,y \in Y$, we have
\[
h_0(x,\Delta_{1|0}y)
=
\cal R_0(\Delta_{1|0}y)(x)
=
\cal R_1(y)(x)
=
h_1(x,y).
\]
This proves \eqref{relative-Hilbert-forms}. It also proves uniqueness, since the nondegeneracy of $h_0$ implies that an operator satisfying \eqref{relative-Hilbert-forms} is uniquely determined.

We prove selfadjointness. Using \eqref{relative-Hilbert-forms} and the Hermitian property of $h_0$ and $h_1$, we have
\begin{align*}
h_0(\Delta_{1|0}x,y)
=
\overline{h_0(y,\Delta_{1|0}x)}
=
\overline{h_1(y,x)}
=
h_1(x,y)
=
h_0(x,\Delta_{1|0}y).
\end{align*}
Thus $\Delta_{1|0}$ is selfadjoint with respect to $h_0$. Moreover, for any nonzero $x \in Y$, we have
\[
h_0(x,\Delta_{1|0}x)
\ov{\eqref{relative-Hilbert-forms}}{=}
h_1(x,x)>0
\]
Hence $\Delta_{1|0}$ is positive and, in particular, injective. Since $Y$ is finite-dimensional, it is invertible.

The positive square root $\Delta_{1|0}^{\frac12}$ is therefore well-defined. Using its selfadjointness with respect to $h_0$, we obtain
\[
h_0\left(\Delta_{1|0}^{\frac12}x,\Delta_{1|0}^{\frac12}y\right)
=
h_0(x,\Delta_{1|0}y)
\ov{\eqref{relative-Hilbert-forms}}{=}
h_1(x,y),
\]
which proves \eqref{relative-square-root-Hilbert-forms}.

Finally, fix a basis of $Y$ and let $D$ be the matrix of $\Delta_{1|0}$. By assumption,
\[
h_0(x,\Delta_{1|0}y)
=
[D y]^*G_0[x]
=
[y]^*D^*G_0[x].
\]
Comparing this expression with
\[
h_1(x,y)
=
[y]^*G_1[x],
\]
we obtain $
D^*G_0=G_1$. Taking adjoints and using $G_0^*=G_0$ and $G_1^*=G_1$ gives $G_0D=G_1$. Hence
\[
D=G_0^{-1}G_1,
\]
which proves \eqref{relative-operator-Gram-matrices}.
\end{proof}

\begin{remark} \normalfont
\label{rem-relative-operator-positivity}
The positivity and selfadjointness of $\Delta_{1|0}$ in Proposition~\ref{prop-relative-operator-Hilbert-forms} are relative to the Hilbert structure $h_0$. In particular, the matrix $G_0^{-1}G_1$ need not be selfadjoint for the standard Euclidean inner product, although it is selfadjoint and positive for the inner product determined by $G_0$.
\end{remark}


\begin{prop}
\label{prop-closed-relative-metric-operator}
Let $\cal{H}$ be a complex Hilbert space equipped with a Hermitian inner product $\langle\cdot,\cdot\rangle_0$, linear in the first variable. Let $\cal D$ be a dense subspace of $\cal H$ equipped with a positive definite Hermitian form $\langle\cdot,\cdot\rangle_1$. Suppose that this form is closed with respect to $\langle\cdot,\cdot\rangle_0$, that is, $\cal D$ is complete for the norm
\[
\norm{x}_{\cal D}
\ov{\mathrm{def}}{=}
\left(\norm{x}_{\cal H}^2+\langle x,x\rangle_1\right)^{\frac12}.
\]
Then there exists a unique positive selfadjoint operator $\Delta$ on $\cal H$ such that
\begin{equation}
\label{relative-metric-form-domain}
\dom\Delta^{\frac12}=\cal D
\end{equation}
and
\begin{equation}
\label{relative-metric-form}
\langle x,y\rangle_1
=
\left\langle \Delta^{\frac12}x,\Delta^{\frac12}y\right\rangle_0,
\quad
x,y \in \cal D.
\end{equation}
In particular, for $x\in\cal D$ and $y\in\dom\Delta$, we have
\begin{equation}
\label{relative-metric-operator-form}
\langle x,y\rangle_1
=
\langle x,\Delta y\rangle_0.
\end{equation}
Moreover, $\ker\Delta=\{0\}$. We call $\Delta$ the \emph{relative metric operator} of $\langle\cdot,\cdot\rangle_1$ with respect to $\langle\cdot,\cdot\rangle_0$.

If $\cal D=\cal H$ and there exists $C>0$ such that
\begin{equation}
\label{relative-form-upper-bound}
\langle x,x\rangle_1
\leq
C\langle x,x\rangle_0,
\quad x \in \cal H,
\end{equation}
then $\Delta$ is bounded and \eqref{relative-metric-operator-form} holds for all $x,y\in\cal H$. If, in addition, there exists $c>0$ such that
\begin{equation}
\label{relative-form-lower-bound}
c\langle x,x\rangle_0
\leq
\langle x,x\rangle_1,
\quad x \in \cal H,
\end{equation}
then $\Delta$ is boundedly invertible and
\begin{equation}
\label{relative-operator-bounds}
c\Id\leq\Delta\leq C\Id.
\end{equation}
Consequently, if the two Hermitian forms define equivalent Hilbert norms on the same vector space, then $\Delta$ is a bounded positive invertible operator characterized by
\begin{equation}
\label{relative-operator-equivalent-Hilbert-norms}
\langle x,y\rangle_1
=
\langle x,\Delta y\rangle_0,
\quad x,y\in\cal H.
\end{equation}
\end{prop}

\begin{proof}
The existence of $\Delta$ follows from the representation theorem for densely defined closed positive sesquilinear forms. Indeed, the form
\[
q(x,y)
\ov{\mathrm{def}}{=}
\langle x,y\rangle_1,
\quad x,y\in\cal D,
\]
is densely defined, positive and closed on the Hilbert space $(\cal H,\langle\cdot,\cdot\rangle_0)$. Hence there exists a unique positive selfadjoint operator $\Delta$ on $\cal H$ such that $\dom\Delta^{\frac12}=\cal D$ and
\[
q(x,y)
=
\big\langle\Delta^{\frac12}x,\Delta^{\frac12}y\big\rangle_0,
\quad x,y \in \cal D.
\]
This proves \eqref{relative-metric-form-domain} and \eqref{relative-metric-form}.

Let now $x\in\cal D$ and $y\in\dom\Delta$. Since $y\in\dom\Delta$ if and only if $y\in\dom\Delta^{\frac12}$ and $\Delta^{\frac12}y\in\dom\Delta^{\frac12}$, the selfadjointness of $\Delta^{\frac12}$ gives
\[
\langle x,y\rangle_1
\ov{\eqref{relative-metric-form}}{=}
\big\langle\Delta^{\frac12}x,\Delta^{\frac12}y\big\rangle_0
=
\langle x,\Delta y\rangle_0,
\]
which proves \eqref{relative-metric-operator-form}.

Suppose that $x\in\ker\Delta$. Then $x\in\dom\Delta\subset\cal D$ and
\[
\la x,x \ra_1
\ov{\eqref{relative-metric-form}}{=}
\bnorm{\Delta^{\frac12}x}_{\cal H}^2
=0.
\]
Since $\langle\cdot,\cdot\rangle_1$ is positive definite, we obtain $x=0$. Hence $\ker\Delta = \{0\}$. Notice that this does not imply in general that $\Delta^{-1}$ is bounded, since $0$ may belong to the continuous spectrum of $\Delta$.

Suppose now that $\cal D=\cal H$ and that \eqref{relative-form-upper-bound} holds. For any $y \in \cal H$, the map $x\mapsto\langle x,y\rangle_1$ is a continuous linear functional on $(\cal H,\langle\cdot,\cdot\rangle_0)$, since by the Cauchy--Schwarz inequality for $\langle\cdot,\cdot\rangle_1$,
\[
|\langle x,y\rangle_1|
\leq
\langle x,x\rangle_1^{\frac12}\langle y,y\rangle_1^{\frac12}
\leq
C\norm{x}_{\cal H}\norm{y}_{\cal H}.
\]
By the Riesz representation theorem, there exists a unique element $\Delta y\in\cal H$ such that
\[
\langle x,y\rangle_1
=
\langle x,\Delta y\rangle_0,
\quad x \in \cal H.
\]
The uniqueness of the representing vector shows that the map $y\mapsto\Delta y$ is linear. Moreover,
\[
|\langle x,\Delta y\rangle_0|
=
|\langle x,y\rangle_1|
\leq
C\norm{x}_{\cal H}\norm{y}_{\cal H},
\]
and taking the supremum over $\norm{x}_{\cal H}\leq1$ yields $\norm{\Delta y}_{\cal H}\leq C\norm{y}_{\cal H}$. Thus $\Delta$ is bounded and $\norm{\Delta}\leq C$. This bounded operator coincides with the positive selfadjoint operator associated above with the closed form, by uniqueness in the representation theorem.

The Hermitian property of $\langle\cdot,\cdot\rangle_1$ gives, for $x,y\in\cal H$,
\[
\langle\Delta x,y\rangle_0
=
\overline{\langle y,\Delta x\rangle_0}
=
\overline{\langle y,x\rangle_1}
=
\langle x,y\rangle_1
=
\langle x,\Delta y\rangle_0,
\]
so $\Delta$ is selfadjoint. Furthermore, we have
\[
\la x,\Delta x\ra_0
=
\la x,x \ra_1
\geq 0,
\]
so $\Delta$ is positive.

Finally, if \eqref{relative-form-lower-bound} also holds, then
\[
c\norm{x}_{\cal H}^2
\leq
\langle x,\Delta x\rangle_0
\leq
C\norm{x}_{\cal H}^2,
\quad x \in \cal H.
\]
Therefore $c\Id\leq\Delta\leq C\Id$. In particular, $0\notin\sigma(\Delta)$ and $\Delta$ is boundedly invertible, with $\norm{\Delta^{-1}}\leq c^{-1}$. This proves \eqref{relative-operator-bounds} and the final assertion.
\end{proof}

\begin{remark} \normalfont
\label{rem-relative-metric-operator-unbounded}
In finite dimension, every positive definite Hermitian form is continuous and any two Hilbert norms are equivalent. Hence the relative metric operator of Proposition~\ref{prop-closed-relative-metric-operator} is automatically bounded and invertible. In infinite dimension, this need no longer be the case. The natural object associated with two Hilbertian geometries is in general an unbounded positive selfadjoint operator, and the fundamental identity is then the quadratic-form identity
\[
\langle x,y\rangle_1
=
\big\langle\Delta^{\frac12}x,\Delta^{\frac12}y\big\rangle_0,
\quad
x,y \in \dom \Delta^{\frac12},
\]
rather than the operator identity $\langle x,y\rangle_1=\langle x,\Delta y\rangle_0$, which requires $y\in\dom\Delta$.
\end{remark}






\begin{prop}
\label{prop-relative-Levi-operator}
Let $\Omega$ be an open subset of a complex Banach space $X$, let $\Phi \co \Omega \to \R$ be a plurisubharmonic function of class $C^2$, and let $P\co X\to X$ be a bounded projection satisfying the assumptions of Proposition~\ref{prop-Levi-contractive-projection}.  Assume that $\Ran P$ is finite-dimensional. Let $a,b \in \Omega \cap \Ran P$ and suppose that the restrictions of $\scr{L}_{\Phi,a}$ and $\scr{L}_{\Phi,b}$ to the subspace $\Ran P$ are nondegenerate. Then the conjugate linear maps
\[
\Lambda_a
\ov{\mathrm{def}}{=}
\Lambda_{\Phi,a}|_{\Ran P}
\quad\text{and}\quad
\Lambda_b
\ov{\mathrm{def}}{=}
\Lambda_{\Phi,b}|_{\Ran P}
\]
are bijections from $\Ran P$ onto $\Ran P^*$. Consequently, the formula
\begin{equation}
\label{def-relative-Levi-operator}
\Delta_{\Phi,b|a}
\ov{\mathrm{def}}{=}
\Lambda_a^{-1}\Lambda_b
\end{equation}
defines a linear bijection $\Delta_{\Phi,b|a} \co \Ran P \to \Ran P$. It is uniquely characterized by
\begin{equation}
\label{relative-Levi-representation}
\scr{L}_{\Phi,b}(x,y)
=\scr{L}_{\Phi,a}(x,\Delta_{\Phi,b|a}y),
\quad
x,y \in \Ran P.
\end{equation}
Moreover, $\Delta_{\Phi,b|a}$ is positive and selfadjoint for the Hilbert space structure defined by $\scr{L}_{\Phi,a}$. Finally, for $a,b,c \in \Omega \cap \Ran P$ for which the corresponding Levi forms are nondegenerate on $\Ran P$, we have
\begin{equation}
\label{relative-Levi-inverse}
%\label{relative-Levi-cocycle}
\Delta_{\Phi,a|b}
=\Delta_{\Phi,b|a}^{-1}
\quad \text{and} \quad
\Delta_{\Phi,c|a}
=\Delta_{\Phi,b|a}\Delta_{\Phi,c|b}.
\end{equation}
\end{prop}


\begin{proof}
By \eqref{Levi-adjoint-fixed}, we have
\[
\Lambda_{\Phi,a}(\Ran P)\subset\Ran P^*
\quad\text{and}\quad
\Lambda_{\Phi,b}(\Ran P)\subset\Ran P^*.
\]
Suppose that $\Lambda_a(y)=0$ for some $y\in\Ran P$. Then
\[
0
=
\big\langle y,\Lambda_a(y)\big\rangle_{X,X^*}
\ov{\eqref{Levi-dual-map}}{=}
\scr L_{\Phi,a}(y,y).
\]
Since the restriction of $\scr L_{\Phi,a}$ to $\Ran P$ is nondegenerate, we obtain $y=0$. Thus $\Lambda_a$ is injective. Similarly, $\Lambda_b$ is injective. Since $
\dim\Ran P^*
=
\dim\Ran P$, both maps are bijections from $\Ran P$ onto $\Ran P^*$.
The restriction map $
\rho\co\Ran P^*\to(\Ran P)^*$ defined by
$$
\rho(x^*)\ov{\mathrm{def}}{=}x^*|_{\Ran P},
$$
is an isomorphism. Indeed, if $x^*\in\Ran P^*$ and $\rho(x^*)=0$, then $P^*x^*=x^*$ and, for any $z\in X$, we have
\[
\langle z,x^*\rangle_{X,X^*}
=
\langle z,P^*x^*\rangle_{X,X^*}
=
\langle Pz,x^*\rangle_{X,X^*}
=
0.
\]
Hence $x^*=0$, so $\rho$ is injective, and its surjectivity follows from the equality of dimensions.
For $x,y\in\Ran P$, we have
\[
\big[\rho\Lambda_a(y)\big](x)
=
\big\langle x,\Lambda_a(y)\big\rangle_{X,X^*}
\ov{\eqref{Levi-dual-map}}{=}
\scr L_{\Phi,a}(x,y),
\]
and similarly
\[
\big[\rho\Lambda_b(y)\big](x)
=
\scr L_{\Phi,b}(x,y).
\]
Thus $\rho\Lambda_a$ and $\rho\Lambda_b$ are precisely the Riesz maps of the two Hermitian forms $\scr L_{\Phi,a}|_{\Ran P}$ and $\scr L_{\Phi,b}|_{\Ran P}$. Proposition~\ref{prop-relative-Riesz-operator} therefore gives the relative metric operator
\[
\Delta_{\Phi,b|a}
=
(\rho\Lambda_a)^{-1}(\rho\Lambda_b)
=
\Lambda_a^{-1}\Lambda_b.
\]
It is a positive selfadjoint linear bijection on the Hilbert space determined by $\scr L_{\Phi,a}$ and satisfies
\[
\scr L_{\Phi,b}(x,y)
=
\scr L_{\Phi,a}(x,\Delta_{\Phi,b|a}y),
\quad
x,y\in\Ran P.
\]
Finally, we have
\[
\Delta_{\Phi,a|b}
=
\Lambda_b^{-1}\Lambda_a
=
(\Lambda_a^{-1}\Lambda_b)^{-1}
=
\Delta_{\Phi,b|a}^{-1},
\]
and
\[
\Delta_{\Phi,b|a}\Delta_{\Phi,c|b}
=
\Lambda_a^{-1}\Lambda_b\Lambda_b^{-1}\Lambda_c
=
\Lambda_a^{-1}\Lambda_c
=
\Delta_{\Phi,c|a}.
\]
This proves \eqref{relative-Levi-inverse}.
\end{proof}

 %In particular,
%\begin{equation}
%\label{relative-Levi-square-root}
%\scr{L}_{\Phi,b}(x,y)
%=\scr{L}_{\Phi,a}\left(\Delta_{\Phi;b|a}^{\frac12}x, \Delta_{\Phi;b|a}^{\frac12}y\right),
%\quad x,y \in \Ran P.
%\end{equation}

 %Selfadjointness gives by polarization
%\[
%\scr{L}_{\Phi,a}\left(\Delta_{\Phi,b|a}^{\frac12}x,
%\Delta_{\Phi,b|a}^{\frac12}y\right)
%=\scr{L}_{\Phi,a}(x,\Delta_{\Phi,b|a}y)
%\ov{\eqref{inter-500}}{=} \scr{L}_{\Phi,b}(x,y),
%\]
%which proves \eqref{relative-Levi-square-root}.


\begin{cor}
\label{cor-relative-Levi-group}
Under the assumptions of Proposition~\ref{prop-relative-Levi-operator}, the family
\begin{equation}
\label{relative-Levi-unitary-group}
U_{b|a,t}
\ov{\mathrm{def}}{=}
\Delta_{\Phi,b|a}^{\i t},
\quad
t \in \R,
\end{equation}
is a one-parameter group of surjective isometries for the two Hilbert space structures induced on $\Ran P$ by $\scr L_{\Phi,a}$ and $\scr L_{\Phi,b}$. More precisely,
\begin{equation}
\label{relative-Levi-group-unitary-a}
\scr{L}_{\Phi,a}(U_{b|a,t}x,U_{b|a,t}y)
=\scr{L}_{\Phi,a}(x,y)
\end{equation}
and
\begin{equation}
\label{relative-Levi-group-unitary-b}
\scr{L}_{\Phi,b}(U_{b|a,t}x,U_{b|a,t}y)
=
\scr{L}_{\Phi,b}(x,y)
\end{equation}
for any $x,y \in \Ran P$ and any $t \in \R$. For any $s \in [0,1]$, the formula
\begin{equation}
\label{interpolated-Levi-forms}
\scr{L}_{a,b}^{(s)}(x,y)
\ov{\mathrm{def}}{=}
\scr{L}_{\Phi,a}(x,\Delta_{\Phi,b|a}^s y)
\end{equation}
defines a positive definite Hermitian form on the subspace $\Ran P$, with
\[
\scr{L}_{a,b}^{(0)}
=\scr{L}_{\Phi,a}
\quad\text{and}\quad
\scr{L}_{a,b}^{(1)}
=\scr{L}_{\Phi,b}.
\]
\end{cor}

\begin{proof}
Since $\Delta_{\Phi,b|a}$ is positive and selfadjoint for $\scr{L}_{\Phi,a}$, its imaginary powers are unitary for $\scr{L}_{\Phi,a}$. This proves \eqref{relative-Levi-group-unitary-a}. Using \eqref{relative-Levi-representation} and the fact that the operator $\Delta_{\Phi,b|a}$ commutes with its imaginary powers, we obtain
\begin{align*}
\MoveEqLeft
\scr{L}_{\Phi,b}(U_{b|a,t}x,U_{b|a,t}y)
\ov{\eqref{relative-Levi-representation}}{=} \scr{L}_{\Phi,a}(U_{b|a,t}x,\Delta_{\Phi,b|a}U_{b|a,t}y) \\
&=\scr{L}_{\Phi,a}(U_{b|a,t}x,U_{b|a,t}\Delta_{\Phi,b|a}y)
=\scr{L}_{\Phi,a}(x,\Delta_{\Phi,b|a}y)
\ov{\eqref{relative-Levi-representation}}{=} \scr{L}_{\Phi,b}(x,y).
\end{align*}
This proves \eqref{relative-Levi-group-unitary-b}. For $0 \leq s \leq 1$, the operator $\Delta_{\Phi,b|a}^s$ is positive and invertible. Hence
\[
\scr{L}_{a,b}^{(s)}(x,y)
=\scr{L}_{\Phi,a}\left(\Delta_{\Phi,b|a}^{\frac s2}x,
\Delta_{\Phi,b|a}^{\frac s2}y\right),
\]
which shows that $\scr{L}_{a,b}^{(s)}$ is a positive definite Hermitian form. The endpoint identities follow immediately from the definition and \eqref{relative-Levi-representation}.
\end{proof}

The preceding construction is more informative when the relative operator is first defined on an ambient tangent Hilbert space and the range of the projection is subsequently shown to be reducing. 

\begin{prop}
\label{prop-ambient-relative-Levi-reduction}
Under the assumptions of Proposition~\ref{prop-Levi-contractive-projection}, let $a,b\in\Omega\cap\Ran P$. Suppose that the Levi forms $\scr{L}_{\Phi,a}$ and $\scr{L}_{\Phi,b}$ are nondegenerate on $X$ and that their Hilbertian norms are equivalent. Let $\cal H_a$ denote the Hilbert space completion of $X$ for the norm $x\mapsto\scr L_{\Phi,a}(x,x)^{\frac12}$. There exists a unique bounded positive invertible operator $\widehat\Delta_{\Phi,b|a} \co \cal H_a \to \cal H_a$ such that
\begin{equation}
\label{ambient-relative-Levi-representation}
\scr{L}_{\Phi,b}(x,y)
=\scr{L}_{\Phi,a}(x,\widehat\Delta_{\Phi,b|a}y),
\quad
x,y \in X.
\end{equation}
Moreover, the orthogonal extension of $P$ to $\cal H_a$ commutes with $\widehat\Delta_{\Phi,b|a}$:
\begin{equation}
\label{ambient-relative-Levi-commutation}
P\widehat\Delta_{\Phi,b|a}
=
\widehat\Delta_{\Phi,b|a}P.
\end{equation}
Consequently, the subspaces $\ovl{\Ran P}^{\cal H_a}$ and $\ovl{\ker P}^{\cal H_a}$ reduce the operator $\widehat\Delta_{\Phi,b|a}$. In particular, we have
\begin{equation}
\label{ambient-relative-Levi-group-commutation}
P\widehat\Delta_{\Phi,b|a}^{\i t}
=
\widehat\Delta_{\Phi,b|a}^{\i t}P,
\quad
t \in \R.
\end{equation}
\end{prop}

Here, we use the same notation for the continuous extensions of the two Levi forms to $\mathcal H_a$, and for the orthogonal extension of $P$ to $\mathcal H_a$. In particular, expressions involving $\widehat\Delta_{\Phi,b|a}y\in\mathcal H_a$ are evaluated using these extended forms.


\begin{proof}
The equivalence of the two Hilbertian norms implies, by the Riesz representation theorem, that there exists a unique bounded positive invertible operator $\widehat\Delta_{\Phi,b|a}$ satisfying \eqref{ambient-relative-Levi-representation}. By Corollary~\ref{cor-Levi-intertwining}, the projection $P$ is selfadjoint for both $\scr{L}_{\Phi,a}$ and $\scr{L}_{\Phi,b}$. Let $x,y \in X$. Using successively \eqref{ambient-relative-Levi-representation} and Levi selfadjointness, we obtain
\begin{align*}
\scr{L}_{\Phi,a}(x,\widehat\Delta_{\Phi,b|a}Py)
&=\scr{L}_{\Phi,b}(x,Py)
\ov{\eqref{Levi-selfadjointness}}{=} \scr{L}_{\Phi,b}(Px,y)
=\scr{L}_{\Phi,a}(Px,\widehat\Delta_{\Phi,b|a}y)
=\scr{L}_{\Phi,a}(x,P\widehat\Delta_{\Phi,b|a}y).
\end{align*}
The nondegeneracy of $\scr{L}_{\Phi,a}$ gives
\[
\widehat\Delta_{\Phi,b|a}Py
=
P\widehat\Delta_{\Phi,b|a}y, \quad
y \in X.
\]
By continuity, this identity extends to $\cal H_a$, proving \eqref{ambient-relative-Levi-commutation}. Since $P$ is an orthogonal projection on $\cal H_a$, commutation with the selfadjoint operator $\widehat\Delta_{\Phi,b|a}$ means that both its range and its kernel are reducing. Commutation with the Borel functional calculus, and in particular with the imaginary powers, follows.
\end{proof}



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Power functions}

A quasi-norm on a complex vector space $X$ is a map $\norm{\cdot}_X \co X \to [0,\infty)$ satisfying the following conditions:
\begin{enumerate}
\item $\norm{x}_X=0$ if and only if $x=0$;

\item $\norm{\lambda x}_X=|\lambda|\norm{x}_X$ for any $x\in X$ and any $\lambda\in\mathbb C$;

\item there exists a constant $C>0$ such that
\[
\norm{x+y}_X
\leq
C\bigl(\norm{x}_X+\norm{y}_X\bigr),
\quad x,y \in X.
\]
\end{enumerate}
A quasi-normed space is a vector space equipped with a quasi-norm. We refer to the survey \cite{Kal03} for more information on quasi-normed spaces. If $X$ is a topological vector space whose topology is defined by a quasi-norm which is uniformly continuous on bounded subsets of $X$, then $X$ is called a continuously quasi-normed space.

In the quasi-normed setting, plurisubharmonicity is understood through upper semicontinuity and the submean inequality on each complex affine line. Following \cite[p.~115]{DGT84} and \cite{Edg85}, a continuously quasi-normed space is said to be locally $\PL$-convex if, for any $x,y\in X$ and any sufficiently small $r>0$, we have
\[
\norm{x}_X
\leq
\frac1{2\pi}\int_0^{2\pi}
\norm{x+r\e^{\i t}y}_X
\d t.
\]
By \cite[Theorem~1.1 p.~40]{Ale81} and \cite[Proposition~2.2 p.~115]{DGT84}, this condition is equivalent to each of the following properties:
\begin{enumerate}
\item the function $
X\to[-\infty,\infty)$, $x\mapsto\log\norm{x}_X$ is plurisubharmonic, where $\log 0\ov{\mathrm{def}}{=}-\infty$;

\item for any, equivalently for some, $0 < p < \infty$, the function $X \to \R_+$, $x \mapsto \norm{x}_X^p$ is plurisubharmonic. Equivalently, we have
\[
\norm{x}_X
\leq
\left(
\frac1{2\pi}\int_0^{2\pi}
\norm{x+r\e^{\i t}y}_X^p
\d t
\right)^{\frac1p}
\]
for any $x,y\in X$ and any sufficiently small $r>0$.
\end{enumerate}
We refer also to \cite{Pav91}.

\begin{example}
\normalfont
\label{Banach-spaces}
As observed in \cite[Remark p.~116]{DGT84}, any complex Banach space is locally $\PL$-convex.
\end{example}

We now return to the Banach space setting, where the differential results of the preceding subsections apply. Let $X$ be a complex Banach space and let $1\leq p<\infty$. Consider the function
\begin{equation}
\label{def-de-Phi-p}
\Phi_p\co X \to \R_+,
\qquad
\Phi_p(x)
\ov{\mathrm{def}}{=}
\norm{x}_X^p.
\end{equation}
The function $\Phi_p$ is continuous and convex, hence plurisubharmonic. Let $P\co X\to X$ be a contractive projection. Then
\[
\Phi_p(Px)
\leq
\Phi_p(x),
\quad x\in X.
\]

Suppose that $a\in\Ran P$ and that $\Phi_p$ is of class $\C^2$ on an open neighborhood $U$ of $a$. Set
\[
\Omega
\ov{\mathrm{def}}{=}
U\cap P^{-1}(U).
\]
Then $\Omega$ is an open neighborhood of $a$. Moreover, $P(\Omega)\subset\Omega$. Indeed, if $x\in\Omega$, then $Px\in U$ and
\[
P(Px)=Px\in U,
\]
so that $Px\in P^{-1}(U)$. Hence $Px\in\Omega$.

Therefore Proposition~\ref{prop-Levi-contractive-projection} applies to the restriction of $\Phi_p$ to $\Omega$ and gives
\begin{equation}
\label{Levi-p-contractive}
\scr L_{\Phi_p,a}(Pz)
\leq
\scr L_{\Phi_p,a}(z),
\quad
z \in X.
\end{equation}
Thus every point of the range at which the power function $\Phi_p$ is locally of class $\C^2$ gives rise to a tangent Hilbertian geometry for which $P$ is contractive, and hence orthogonal after passing to the associated Hilbert completion.


%A quasi-norm on a complex vector space $X$ is a  map $\norm{\cdot}_X \co X\rightarrow [0,\infty)$  satisfying the following conditions:
%\begin{enumerate}
%\item $\norm{x}_X=0$ if and only if $x=0$,
%
%\item $\norm{\lambda x}_X=|\lambda|\cdot \norm{x}_X$, for any $x \in X$ and any $\lambda \in \mathbb{C}$,
%
%\item there exists a constant $C$ such that $\norm{x+y}_X \leq C\left[\norm{x}_X+\norm{y}_X\right]$ for any $x,y \in X$.
%\end{enumerate}
%A quasi-normed space is a vector space equipped with a quasi-norm. We refer to the survey \cite{Kal03} for more information on quasi-normed spaces. If $X$ is a topological vector space whose topology is defined by a
%quasi-norm that is uniformly continuous on bounded sets of $X$, then $X$ is called a continuously quasi-normed space.
%
%
%
%Following \cite[p.~115]{DGT84} \cite{Edg85}, a continuously quasi-normed space is said to be locally $\PL$-convex 
%%\cite{Pav91}
%if for any $x,y \in X$ and any small enough $r$, we have
%$$
%\norm{x}_X
%\leq \frac{1}{2\pi} \int_{0}^{2\pi} \norm{x+r\e^{\i t}y}_X \d t.
%$$
%It turns out that, by \cite[Theorem 1.1 p.~40]{Ale81} \cite[Proposition 2.2 p.~115]{DGT84}, this condition is equivalent to one of the following properties.
%\begin{enumerate}
	%\item the function $X \to [-\infty,\infty)$, $x \mapsto \log \norm{x}_X$ is plurisubharmonic, where $\log 0 \ov{\mathrm{def}}{=} -\infty$
	%\item for any (or some) $0 < p < \infty$, the function $X \to \R^+$, $x \mapsto \norm{x}_X^p$ is plurisubharmonic, i.e.,
%$$
%\norm{x}_X
%\leq \bigg(\frac{1}{2\pi} \int_{0}^{2\pi} \norm{x+r\e^{\i t}y}_X^p \d t\bigg)^{\frac{1}{p}}.
%$$
%\end{enumerate}
%We refer also to \cite{Pav91}.
%
%\begin{example}
%\normalfont
%\label{Banach-spaces}
%As observed in \cite[Remark p.~116]{DGT84}, a Banach space is locally $\PL$-convex.
%\end{example}
%
%Fix a locally $\PL$-convex quasi-normed space $X$. Let $0 < p < \infty$. Consider the function $\Phi_p \co X \to \R_+$ defined by
%\begin{equation}
%\label{def-de-Phi-p-bis}
%\Phi_p(x)
%\ov{\mathrm{def}}{=} \norm{x}_X^p, \quad x \in X.
%\end{equation}
%%The most relevant instance of Proposition~\ref{prop-Levi-contractive-projection} for us is obtained from powers of the norm.  and suppose that the function
%%\[
%%\Phi_p(x)
%%\ov{\mathrm{def}}{=} \frac1p\norm{x}_X^p
%%\]
%%is of class $\mathrm{C}^2$ at the point under consideration. Note that $\Phi_p$ is convex: For any $x,y \in X$ and any $\theta \in [0,1]$, the triangle inequality in $X$ and the scalar convexity of the function $t \mapsto t^p$ give
%%\begin{align*}
%%\MoveEqLeft
%%\Phi_p(\theta x+(1-\theta)y)
%%\ov{\eqref{Iochum-introduction}}{=} \norm{\theta x+(1-\theta)y}_{X}^p 
%%\leq \big(\theta\norm{x}_{X}+(1-\theta)\norm{y}_{X}\big)^p \\
%%&\leq \theta\norm{x}_{X}^p+(1-\theta)\norm{y}_{X}^p
%%\ov{\eqref{Iochum-introduction}}{=} \theta\Phi_p(x)+(1-\theta)\Phi_p(y).
%%\end{align*}
%It is plurisubharmonic.  
%%\textbf{Plus court \cite[Proposition 34.12 p.~253]{Muj86} ?}. 
%If $P \co X \to X$ is a contractive projection, then $\Phi_p(P(x)) \leq \Phi_p(x)$ for any $x \in X$. Consequently, for any $a \in \Ran P$, we have
%\begin{equation}
%\label{Levi-p-contractive-bis}
%\scr{L}_{\Phi_p,a}(P(z))
%\ov{\eqref{Levi-projection-contraction}}{\leq} 
%\scr{L}_{\Phi_p,a}(z),
%\quad z \in X.
%\end{equation}
%
%For the differential conclusions below, assume that $X$ is a complex Banach space. Let $p\geq1$ and put $\Phi_p(x)=\|x\|_X^p$. Let $P\colon X\to X$ be a contractive projection. At every $a\in\operatorname{Ran}P$ at which $\Phi_p$ is of class $\mathrm C^2$ on a neighborhood, the Levi form is defined and satisfies $\mathscr L_{\Phi_p,a}(Pz)\leq\mathscr L_{\Phi_p,a}(z)$ for every $z\in X$. Indeed, if $U$ is such a neighborhood, then $\Omega=U\cap P^{-1}(U)$ is an open neighborhood of $a$ satisfying $P(\Omega)\subset\Omega$, and Proposition~\ref{prop-Levi-contractive-projection} applies.
%
%
%\begin{example} \normalfont
%Let $X$ be a complex Banach space and let $1\leq p<\infty$. Consider the function $\Phi_p\co X\to\R_+$ defined by
%\begin{equation}
%\label{def-de-Phi-p}
%\Phi_p(x)
%\ov{\mathrm{def}}{=}\norm{x}_X^p,
%\quad x \in X.
%\end{equation}
%The function $\Phi_p$ is continuous and convex, hence plurisubharmonic. If $P \co X \to X$ is a contractive projection, then $\Phi_p(Px)\leq\Phi_p(x)$ for any $x \in X$. Consequently, whenever $\Phi_p$ is of class $\mathrm{C}^2$ at a point $a \in \Ran P$, Proposition~\ref{prop-Levi-contractive-projection} gives
%\begin{equation}
%\label{Levi-p-contractive}
%\scr L_{\Phi_p,a}(Pz)
%\leq
%\scr L_{\Phi_p,a}(z),
%\quad z \in X.
%\end{equation}
%\end{example}

%\cite{Pav91}
%The most relevant instance of Proposition~\ref{prop-Levi-contractive-projection} for us is obtained from powers of the norm.  and suppose that the function
%\[
%\Phi_p(x)
%\ov{\mathrm{def}}{=} \frac1p\norm{x}_X^p
%\]
%is of class $\mathrm{C}^2$ at the point under consideration. Note that $\Phi_p$ is convex: For any $x,y \in X$ and any $\theta \in [0,1]$, the triangle inequality in $X$ and the scalar convexity of the function $t \mapsto t^p$ give
%\begin{align*}
%\MoveEqLeft
%\Phi_p(\theta x+(1-\theta)y)
%\ov{\eqref{Iochum-introduction}}{=} \norm{\theta x+(1-\theta)y}_{X}^p 
%\leq \big(\theta\norm{x}_{X}+(1-\theta)\norm{y}_{X}\big)^p \\
%&\leq \theta\norm{x}_{X}^p+(1-\theta)\norm{y}_{X}^p
%\ov{\eqref{Iochum-introduction}}{=} \theta\Phi_p(x)+(1-\theta)\Phi_p(y).
%\end{align*}
%\textbf{Plus court \cite[Proposition 34.12 p.~253]{Muj86} ?}. 

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{The case of noncommutative $\L^4$-spaces}
\label{sec-the-case-NC}

\subsection{The Levi form of $\norm{\cdot}_{\L^4(\cal{M})}^4$}

We now apply the preceding general principle to noncommutative $\L^4$-spaces. In this case, the Levi form admits an explicit algebraic expression. This is the point where the general Banach-space geometry begins to recover the Jordan triple structure of the ambient von Neumann algebra. Recall that $\Phi_4(x)
\ov{\mathrm{def}}{=} \norm{x}_{\L^4(\cal{M})}^4$ for any $x \in \L^4(\cal{M})$. The following exact fourth-moment formula is the main reason why the case $p=4$ is algebraically accessible.

\begin{lemma}
\label{lem-circular-fourth-moment}
Let $a,z \in \L^4(\cal{M})$ and let $r \geq 0$. Then
\begin{equation}
\label{circular-fourth-moment}
\frac1{2\pi}\int_0^{2\pi}
\norm{a+r\e^{\i t}z}_{\L^4(\cal{M})}^4 \d t
= \norm{a}_{\L^4(\cal{M})}^4
+ 2r^2\left(
\norm{a^*z}_{\L^2(\cal{M})}^2+\norm{za^*}_{\L^2(\cal{M})}^2
\right)
+ r^4\norm{z}_{\L^4(\cal{M})}^4.
\end{equation}
\end{lemma}

\begin{proof}
Put $A\ov{\mathrm{def}}{=}a^*a$, $B\ov{\mathrm{def}}{=}a^*z$ and $C\ov{\mathrm{def}}{=}z^*z$. Then
\[
(a+r\e^{\i t}z)^*(a+r\e^{\i t}z)
=A+r\e^{\i t}B+r\e^{-\i t}B^*+r^2C.
\]
By H\"older's inequality, $A,B,C$ belong to $\L^2(\cal M)$, and every product occurring after the square is expanded belongs to $\L^1(\cal M)$. Thus the expansion is an identity in $\L^1(\cal M)$, and we may apply the trace and integrate its finitely many terms separately. Integration over $t$ annihilates every term with a nonzero Fourier degree. Hence
\begin{align*}
\MoveEqLeft
\frac1{2\pi}\int_0^{2\pi} \norm{a+r \e^{\i t}z}_{\L^4(\cal{M})}^4 \d t 
\ov{\eqref{Def-norm-Lp}}{=}\frac1{2\pi}\int_0^{2\pi} \tr\big[(a+r \e^{\i t}z)^*(a+r\e^{\i t}z)\big]^2 \d t \\
&=
\tr(A^2)
+r^2\tr(AC+CA+BB^*+B^*B)
+r^4\tr(C^2).
\end{align*}
By the tracial property, we have
\[
\tr(AC)=\tr(CA)=\norm{za^*}_{\L^2(\cal M)}^2
\quad\text{and}\quad
\tr(BB^*)=\tr(B^*B)=\norm{a^*z}_{\L^2(\cal M)}^2.
\]
Together with $\tr(A^2)=\norm{a}_4^4$ and $\tr(C^2)=\norm{z}_4^4$, this proves \eqref{circular-fourth-moment}.
\end{proof}

By H\"older's inequality, the real four-linear form $\L^4(\cal{M})^4 \to \mathbb{R}$, $(x_1,x_2,x_3,x_4) \mapsto \Re \Tr(x_1^*x_2x_3^*x_4)$ is continuous on $\L^4(\cal{M})^4$. Consequently, $\Phi_4(x)=\Tr(x^*xx^*x)$ is a continuous real polynomial of degree four, and in particular is of class $\C^\infty$.

\begin{cor}
\label{Levi-form-noncom-L4}
For any $a \in \L^4(\cal{M})$, we have
\begin{equation}
\label{Levi-L4-diagonal}
\scr{L}_{\Phi_4,a}(z)
=2\left(\norm{a^*z}_{\L^2(\cal{M})}^2+\norm{za^*}_{\L^2(\cal{M})}^2\right), \quad z \in \L^4(\cal{M}),
\end{equation}
and
\begin{equation}
\label{Levi-L4-polar}
\scr{L}_{\Phi_4,a}(y,z)
=2\left(\tr(z^*aa^*y)+\tr(az^*ya^*)\right), \quad y,z \in \L^4(\cal{M}).
\end{equation}
\end{cor}


\begin{proof}
We have
\begin{align*}
\MoveEqLeft
\scr{L}_{\Phi_4,a}(z)
\ov{\eqref{Levi-average-limit}}{=} \lim_{r \to 0}\frac1{r^2}
\left(
\frac1{2\pi}\int_0^{2\pi}\Phi_4(a+r\e^{\i t}z) \d t-\Phi_4(a)
\right) \\        
&\ov{\eqref{circular-fourth-moment}}{=} \lim_{r \to 0}\frac1{r^2}
\left(
2r^2(\norm{a^*z}_2^2+\norm{za^*}_2^2)
+ r^4\norm{z}_4^4
\right) 
=2\left(\norm{a^*z}_{\L^2(\cal{M})}^2+\norm{za^*}_{\L^2(\cal{M})}^2\right).
\end{align*}
The second formula follows by polarization.
\end{proof}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Levi form and Peirce decomposition of a partial isometry}
\label{sec-tangent-Peirce}

There is a particularly transparent interpretation of the tangent Hilbert form at a tripotent. If $w \in \cal{M}$ is a tripotent, then, by Example~\ref{ex-trip-partial-iso}, $w$ is a partial isometry. Set $e \ov{\mathrm{def}}{=} ww^*$ and $f \ov{\mathrm{def}}{=} w^*w$. The multiplication operator $\mathrm{D}(w,w) \co \cal{M}\to \cal{M}$ associated with $w$ is
\begin{equation}
\label{Peirce-operator-partial-isometry}
\mathrm{D}(w,w)z
\ov{\eqref{Left-mult-JBW-triple}}{=} \{w,w,z\}
\ov{\eqref{triple product-C-star}}{=} \frac12(ww^*z+ zw^*w)
=\frac12(ez+zf), \quad z \in \cal M.
\end{equation}
The corresponding Peirce decomposition \eqref{Peirce-2} is especially explicit. If $e^\perp=1-e$ and $f^\perp=1-f$, the Peirce projections are given by
\begin{align*}
\MoveEqLeft
\P_2^w(z)         
=\mathrm{D}(w,w)(2 \mathrm{D}(w,w) -\Id_{\cal{M}})(z) 
\ov{\eqref{Peirce-operator-partial-isometry}}{=} \mathrm{D}(w,w)(ez+zf-z)\\
&\ov{\eqref{Peirce-operator-partial-isometry}}{=}\frac12\big(e(ez+zf-z)+(ez+zf-z)f\big)
=ezf.
\end{align*}
Similarly, we have
\begin{equation}
\label{Peirce-projections-partial-isometry}
\P_1^w(z)
=ezf^\perp+e^\perp zf, 
\quad \text{and} \quad
\P_0^w(z)
= e^\perp zf^\perp.
\end{equation}
We introduce the corresponding $\L^2$-Peirce subspaces 
\[
\cal{H}_2(w) \ov{\mathrm{def}}{=} e\L^2(\cal M)f,\qquad
\cal{H}_1(w) \ov{\mathrm{def}}{=} e\L^2(\cal M)f^\perp\oplus_2e^\perp\L^2(\cal M)f,\qquad
\cal{H}_0(w) \ov{\mathrm{def}}{=} e^\perp\L^2(\cal M)f^\perp.
\]
These subspaces are mutually orthogonal in the Hilbert space $\L^2(\cal{M})$. In the sequel, we consider the operators defined on $\L^2(\cal{M})$ instead of $\cal{M}$. From \eqref{Peirce-operator-partial-isometry}, we obtain
\begin{equation}
\label{Peirce-operator-spectral-decomposition}
\mathrm{D}(w,w)
=\P_2^w+\frac12 \P_1^w.
\end{equation}
In particular, the operator $\mathrm{D}(w,w)$ acts as multiplication by $1$, $\frac12$ and $0$ on $\cal{H}_2(w)$, $\cal{H}_1(w)$ and $\cal{H}_0(w)$, respectively; a value is an eigenvalue precisely when the corresponding subspace is nonzero.

Now, we show that the tangent Hilbert form at a tripotent is exactly the quadratic form associated with four times the Peirce operator. 

\begin{prop}
Suppose that $w \in \L^4(\cal{M}) \cap \cal{M}$ is a tripotent. The restriction of the Levi form $\scr L_{\Phi_4,w}$ to $\L^4(\cal M)\cap\L^2(\cal M)$ extends uniquely to a continuous Hermitian form on $\L^2(\cal M)$, and
\begin{equation}
\label{tangent-form-Peirce-operator}
\scr{L}_{\Phi_4,w}(z,y)
=4\big\la \mathrm{D}(w,w)z,y \big\ra_{\L^2(\cal{M})},
\quad y,z \in \L^2(\cal{M}).
\end{equation}
Moreover, we have
\begin{align}
\label{tangent-form-Peirce-decomposition}
\MoveEqLeft
\scr{L}_{\Phi_4,w}(z,z)
= 4\norm{\P_2^w(z)}_{\L^2(\cal{M})}^2+2\norm{\P_1^w(z)}_{\L^2(\cal{M})}^2,\quad z \in \L^2(\cal{M}).
\end{align}
\end{prop}

\begin{proof}
For $y,z\in\L^4(\cal M)\cap\L^2(\cal M)$, the equalities $ww^*=e$ and $w^*w=f$, the tracial property and \eqref{Levi-L4-polar} give
\begin{align*}
\MoveEqLeft
\scr{L}_{\Phi_4,w}(z,y)
\ov{\eqref{Levi-L4-polar}}{=}2\tr(y^*ww^*z)+2\tr(wy^*zw^*) 
=2\tr(y^*ez)+2\tr(y^*zf) \\
&=2\tr(y^*(ez+zf)) 
\ov{\eqref{Peirce-operator-partial-isometry}}{=}4\big\la \mathrm{D}(w,w)z,y \big\ra_{\L^2(\cal{M})}. 
\end{align*}
The last expression defines a continuous Hermitian form on $\L^2(\cal M)$. Since $\L^4(\cal M)\cap\L^2(\cal M)$ is dense in $\L^2(\cal M)$, it is the unique continuous extension of the Levi form and proves \eqref{tangent-form-Peirce-operator}.

Write $
z=z_2+z_1^\ell+z_1^r+z_0$, where
\[
z_2 \ov{\mathrm{def}}{=} ezf,\qquad
z_1^\ell \ov{\mathrm{def}}{=}  ezf^\perp,\qquad
z_1^r \ov{\mathrm{def}}{=}  e^\perp zf,\qquad
z_0 \ov{\mathrm{def}}{=}  e^\perp zf^\perp.
\]
Since these four vectors are mutually orthogonal in the Hilbert space $\L^2(\cal{M})$, we have $ez=z_2+z_1^\ell$ and $zf=z_2+z_1^r$. Consequently, we have
\begin{align*}
\MoveEqLeft
\scr{L}_{\Phi_4,w}(z,z)
=2\norm{ez}_2^2+2\norm{zf}_2^2
=4\norm{z_2}_2^2+2\norm{z_1^\ell}_2^2+2\norm{z_1^r}_2^2
=4\norm{\P_2^w(z)}_2^2+2\norm{\P_1^w(z)}_2^2.
\end{align*}
This proves \eqref{tangent-form-Peirce-decomposition}.
\end{proof}

In particular, we have
\begin{equation}
\label{kernel-tangent-Peirce-zero}
\{z \in \L^2(\cal{M}) : \scr{L}_{\Phi_4,w}(z,z)=0\}
=\cal{H}_0(w).
\end{equation}
Hence the $\L^2$-Peirce-$0$ component is precisely the part which is invisible to the tangent Hilbert geometry at $w$, whereas the $\L^2$-Peirce-$2$ and $\L^2$-Peirce-$1$ components are detected with respective weights $4$ and $2$. Equivalently, $4\mathrm{D}(w,w)$ acts with the weights $4$, $2$ and $0$ on the three Peirce components, omitting any value whose component is zero. This gives a concrete link between the infinitesimal geometry of the $\L^4$-norm and the Peirce geometry of the ambient $\JB^*$-triple. In particular, any $z \in e\L^2(\cal M)f$ belongs to the $\L^2$-Peirce-$2$ subspace and
\begin{equation}
\label{full-tripotent-tangent-form}
\scr{L}_{\Phi_4,w}(z,y)
=4\la z,y \ra_{\L^2(\cal{M})},
\quad
z,y \in e\L^2(\cal{M})f.
\end{equation}
This explains why the existence of a full partial isometry in the range turns the tangent Hilbert geometry into the ordinary $\L^2$-geometry.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Contractive projections on noncommutative $\L^4$-spaces}
\label{sec-L4}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{First order geometry and triple product}




Let $\cal{M}$ be a von Neumann algebra and let $P\co \L^4(\cal{M}) \to \L^4(\cal{M})$ be a bounded projection. We set $Y \ov{\mathrm{def}}{=} \Ran P$ and $Z \ov{\mathrm{def}}{=} \ker P$. We use the sesquilinear duality
\[
\la x,\eta\ra
\ov{\mathrm{def}}{=}
\tr(\eta^*x),
\quad
x \in \L^4(\cal{M}),
\eta \in \L^{\frac43}(\cal{M}).
\]
We denote by $P^\dagger \co \L^{\frac43}(\cal{M}) \to \L^{\frac43}(\cal{M})$ the adjoint map determined by
\begin{equation}
\label{def-adjoint-P-L4}
\tr\big((P^\dagger y)^*x\big)
=
\tr(y^*P(x)),
\quad
x \in \L^4(\cal{M}), y\in\L^{\frac43}(\cal{M}).
\end{equation}
Thus $P^\dagger$ is a projection, $\Ran P^\dagger=(\ker P)^\perp$, and we also set
\begin{equation}
\label{range-adjoint-annihilator}
Y^\sharp
\ov{\mathrm{def}}{=}\Ran P^\dagger=(\ker P)^\perp.
\end{equation}
For any element $x \in \L^4(\cal{M})$ with polar decomposition $x=u|x|$, define 
\begin{equation}
\label{duality-map-L4-recalled}
\frak{j}_4(x)
\ov{\mathrm{def}}{=}\frak{j}_{4,\L^{4}(\cal{M})}(x)  
=u|x|^3
=x x^* x \in \L^{\frac43}(\cal{M}).
\end{equation}
We have
\begin{equation}
\label{duality-map-norm-L4}
\norm{\mathfrak j_4(x)}_{\frac43}
=
\norm{x}_4^3
\quad\text{and}\quad
\tr(\mathfrak j_4(x)^*x)
=
\norm{x}_4^4.
\end{equation}
Likewise, if $y \in \L^{\frac43}(\cal{M})$ with polar decomposition $y=v|y|$, we set
\[
\mathfrak j_{\frac43}(y)
\ov{\mathrm{def}}{=} v|y|^{\frac13} \in \L^4(\cal{M}).
\]
Then
\begin{equation}
\label{inverse-duality-maps}
\mathfrak j_{\frac43}\mathfrak j_4(x)
\ov{\eqref{inverse-power-duality-mapping-NC}}{=} x
\quad \text{and} \quad
\mathfrak j_4\mathfrak j_{\frac43}(y)
\ov{\eqref{inverse-power-duality-mapping-NC}}{=}y.
\end{equation}

Recall that by \cite[(4.1.25) p.~465]{CGRP14} a $\C^*$-algebra $\cal{A}$ becomes a $\JB^*$-triple under its own norm and the triple product $\{x,y,z\} \ov{\mathrm{def}}{=} \frac12(xy^*z+zy^*x)$, where $x,y,z \in \cal{A}$. Inspired by this result, we introduce the $\L^4$-Jordan triple product $\{\cdot,\cdot,\cdot\} \co \L^4(\cal{M}) \times \L^4(\cal{M}) \times \L^4(\cal{M}) \to \L^{\frac43}(\cal{M})$ defined by the same formula
\begin{equation}
\label{ambient-Jordan-triple-product}
\{x,y,z\}
\ov{\mathrm{def}}{=}
\frac12(xy^*z+zy^*x),
\quad
x,y,z \in \L^4(\cal{M}).
\end{equation}
Note this map is well-defined by H\"older's inequality.

\begin{prop}
\label{prop-contractivity-cubic-characterization}
Let $P \co \L^4(\cal{M}) \to \L^4(\cal{M})$ be a bounded projection. The following assertions are equivalent.
\begin{enumerate}
\item The projection $P$ is contractive.
\item For any $x \in \Ran P$, we have $\frak{j}_4(x) \in \Ran P^\dagger$, i.e.,
\begin{equation}
\label{duality-vector-fixed-adjoint}
P^\dagger(\frak{j}_4(x))=\frak{j}_4(x).
\end{equation}
\item For any $x \in \Ran P$ and any $w \in \ker P$, we have
\begin{equation}
\label{cubic-orthogonality-range-kernel}
\tr\big((x x^* x)^*w\big)
=0.
\end{equation}
\item For any $x,y,z \in \Ran P$ and any $w \in \ker P$, we have
\begin{equation}
\label{triple-orthogonality-range-kernel}
\tr\big(\{x,y,z\}^* w\big)
=0.
\end{equation}
\end{enumerate}
If these conditions hold, then
\begin{equation}
\label{range-adjoint-duality-image}
\frak{j}_4(\Ran P)
=\Ran P^\dagger.
\end{equation}
In particular, although $\frak{j}_4$ is nonlinear, its range $\frak{j}_4(\Ran P)$ is a linear subspace of the Banach space $\L^{\frac43}(\cal{M})$.
\end{prop}

\begin{proof}
1 $\Rightarrow$ 2. Under the conjugate linear isometric identification of $\L^{\frac43}(\cal M)$ with $(\L^4(\cal M))^*$ induced by the trace pairing, the Banach adjoint $P^*$ corresponds to $P^\dagger$. Hence the implication follows from Lemma~\ref{Lemma-smooth}.
%Suppose first that $P$ is contractive. Then $P^\dagger$ is also contractive. Let $y \in \Ran P$. Since $P(y)=y$, we have
%\[
%\tr\big((P^\dagger\mathfrak j_4(y))^*y\big)
%=
%\tr\big(\mathfrak j_4(y)^*P(y)\big)
%=
%\norm{y}_4^4.
%\]
%Moreover,
%\[
%\norm{P^\dagger\mathfrak j_4(y)}_{\frac43}
%\leq
%\norm{\mathfrak j_4(y)}_{\frac43}
%=
%\norm{y}_4^3.
%\]
%H\"older's inequality therefore gives
%\[
%\norm{y}_4^4
%\leq
%\norm{P^\dagger\mathfrak j_4(y)}_{\frac43}\norm{y}_4
%\leq
%\norm{y}_4^4.
%\]
%Thus $P^\dagger\mathfrak j_4(y)$ is a norming functional for $y$. Since $\L^4(\cal{M})$ is smooth, this norming functional is unique. Hence
%\[
%P^\dagger\mathfrak j_4(y)
%=\mathfrak j_4(y),
%\]
%which proves that the first assertion implies the second one.

2 $\iff$ 3. By \eqref{range-adjoint-annihilator}, condition \eqref{duality-vector-fixed-adjoint} is equivalent to
$\frak{j}_4(x) \in (\ker P)^\perp$, i.e., $x x^* x \in (\ker P)^\perp$, which is precisely \eqref{cubic-orthogonality-range-kernel}. Thus the second and third assertions are equivalent.

3 $\Rightarrow$ 1. We next show that the third assertion implies the first one. Let $x \in \L^4(\cal{M})$ and write
\[
x=y+z,
\quad
y=P(x)\in \Ran P,\quad z=x-P(x) \in \ker P.
\]
If $y\neq0$, then \eqref{cubic-orthogonality-range-kernel} and H\"older's inequality yield
\[
\norm{y}_4^4
=
\tr(\frak{j}_4(y)^*y)
=
\tr(\mathfrak j_4(y)^*x)
\leq
\norm{\mathfrak j_4(y)}_{\frac43}\norm{x}_4
=
\norm{y}_4^3\norm{x}_4.
\]
Consequently, $\norm{P(x)}_4=\norm{y}_4 \leq \norm{x}_4$. The same conclusion is trivial if $y=0$. Hence $P$ is contractive.

3 $\iff$ 4. It remains to compare the third and fourth assertions. Direct expansion gives the polarization identity
\begin{equation}
\label{polarization-cubic-map}
\{x,y,z\}
=
\frac1{32}
\sum_{\alpha,\beta \in \{1,-1,\i,-\i \}}
\alpha\overline{\beta}\,
\mathfrak j_4(x+\alpha y+\beta z).
\end{equation}

Indeed, expanding the right-hand side and summing over the fourth roots of unity eliminates every term except
\[
\frac12(xy^*z+zy^*x).
\]
Since $\Ran P$ is a subspace, condition \eqref{cubic-orthogonality-range-kernel} and \eqref{polarization-cubic-map} imply \eqref{triple-orthogonality-range-kernel}. Conversely, taking $x=y=z$ in \eqref{triple-orthogonality-range-kernel} gives \eqref{cubic-orthogonality-range-kernel}. This proves the equivalence of the four assertions.

Finally, we already know that $\mathfrak j_4( \Ran P) \subset \Ran P^\dagger$. Let $y \in \Ran P^\dagger$. Since $\norm{P^\dagger}=\norm{P}\leq1$, Lemma~\ref{Lemma-smooth}, applied to the contractive projection $P^\dagger \co \L^{\frac43}(\cal{M}) \to \L^{\frac43}(\cal{M})$, gives
\[
\frak{j}_{\frac43}(y)
\in
\Ran (P^\dagger)^\dagger
=
\Ran P.
\]
Here $(P^\dagger)^\dagger=P$ under the canonical reflexive identification coming from the trace duality.
Using \eqref{inverse-duality-maps}, we obtain
\[
y
\ov{\eqref{inverse-duality-maps}}{=}
\frak{j}_4\big(\frak{j}_{\frac43}(y)\big)
\in\frak{j}_4(\Ran P).
\]
This proves \eqref{range-adjoint-duality-image}.
\end{proof}



\begin{cor}
\label{cor-dual-triple-closure}
Let $P \co \L^4(\cal{M}) \to \L^4(\cal{M})$ be a contractive projection. Then
\begin{equation}
\label{dual-triple-closure}
\{\Ran P,\Ran P,\Ran P\}
= \Ran P^\dagger
=\frak {j}_4(\Ran P).
\end{equation}
Equivalently, for any $x,y,z \in \Ran P$, we have
\begin{equation}
\label{triple-fixed-by-adjoint}
P^\dagger\left(\frac12(xy^*z+zy^*x)\right)
=\frac12(xy^*z+zy^*x).
\end{equation}
\end{cor}

\begin{proof}
Proposition~\ref{prop-contractivity-cubic-characterization} gives the inclusion from left to right. Conversely, any element of $\Ran P^\dagger$ has the form $\frak j_4(y)=\{y,y,y\}$ for some $y\in\Ran P$, by \eqref{range-adjoint-duality-image}.
\end{proof}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{First order geometry and $2$-contractive projections on $\L^4$-spaces}

For $n \geq 1$, let $P_n \ov{\mathrm{def}}{=} \Id_{S_n^4} \ot P \co \L^4(\M_n(\cal M)) \to \L^4(\M_n(\cal{M}))$. Under the canonical trace duality, we have $
P_n^\dagger
=
\Id_{S_n^{\frac43}} \ot P^\dagger$ and hence
\begin{equation}
\label{range-adjoint-amplification}
\Ran P_n^\dagger
=
\M_n(\Ran P^\dagger).
\end{equation}

\begin{thm}
\label{thm-two-contractive-completely-contractive-L4}
Let $\cal{M}$ be a von Neumann algebra and let $P \co \L^4(\cal{M}) \to \L^4(\cal{M})$ be a bounded projection. The following assertions are equivalent.
\begin{enumerate}
\item $P$ is $2$-contractive.
\item For any $x,y,z \in \Ran P$, we have
\begin{equation}
\label{unsymmetrized-triple-dual-range}
xy^*z \in \Ran P^\dagger.
\end{equation}
Equivalently,
\begin{equation}
\label{unsymmetrized-triple-fixed}
P^\dagger(xy^*z)=xy^*z,
\quad
x,y,z \in \Ran P.
\end{equation}
\item $P$ is completely contractive.
\item For any integer $n \geq 1$, we have
\begin{equation}
\label{matrix-duality-map-range}
xx^*x\in \M_n(\Ran P^\dagger),
\quad
x \in \M_n(\Ran P).
\end{equation}
\end{enumerate}
In this case, for any $n \geq 1$,
\begin{equation}
\label{matrix-duality-map-equality}
\{xx^*x : x \in \M_n(\Ran P)\}
=
\M_n(\Ran P^\dagger).
\end{equation}
\end{thm}

\begin{proof}
 $3 \Rightarrow 1$. It is immediate.

$1 \Rightarrow 2$. Suppose that $P$ is $2$-contractive. Then $P_2 \co \L^4(\M_2(\cal{M})) \to \L^4(\M_2(\cal{M}))$ is a contractive projection. By Proposition~\ref{prop-contractivity-cubic-characterization}, we have
\[
aa^*a \in \Ran P_2^\dagger
=
\M_2(\Ran P^\dagger),
\quad
a \in \M_2(\Ran P).
\]
Since $\M_2(\Ran P^\dagger)$ is linear, the polarization formula \eqref{polarization-cubic-map}, applied in $\M_2(\cal M)$, gives
\begin{equation}
\label{matrix-Jordan-triple-dual-range}
\frac12(ab^*c+cb^*a) \in \M_2(\Ran P^\dagger),
\quad
a,b,c \in \M_2(\Ran P).
\end{equation}

Let $x,y,z \in \Ran P$ and consider the elements $
a\ov{\mathrm{def}}{=}\begin{bmatrix}
 0    &  x \\
  0   &  0 \\
\end{bmatrix}$, $b \ov{\mathrm{def}}{=} \begin{bmatrix}
   0  &  0 \\
   0  &  y \\
\end{bmatrix}$ and $c \ov{\mathrm{def}}{=}
\begin{bmatrix}
    0& 0  \\
   z  & 0  \\
\end{bmatrix}$.
Then
$$
ab^*c
=\begin{bmatrix}
 0    &  x \\
  0   &  0 \\
\end{bmatrix}\begin{bmatrix}
   0  &  0 \\
   0  &  y \\
\end{bmatrix}^*\begin{bmatrix}
    0& 0  \\
   z  & 0  \\
\end{bmatrix}
=\begin{bmatrix}
  0  & xy^*  \\
  0   & 0  \\
\end{bmatrix}\begin{bmatrix}
    0& 0  \\
   z  & 0  \\
\end{bmatrix}
=\begin{bmatrix}
  xy^*z   &  0 \\
   0  &  0 \\
\end{bmatrix}.
$$
and
$$
cb^*a
=\begin{bmatrix}
    0& 0  \\
   z  & 0  \\
\end{bmatrix}
\begin{bmatrix}
   0  &  0 \\
   0  &  y \\
\end{bmatrix}^*
\begin{bmatrix}
 0    &  x \\
  0   &  0 \\
\end{bmatrix}
=\begin{bmatrix}
  0   &  0 \\
  0   &  0 \\
\end{bmatrix}\begin{bmatrix}
 0    &  x \\
  0   &  0 \\
\end{bmatrix}
=0.
$$
It follows from \eqref{matrix-Jordan-triple-dual-range} that
\[
\frac12\begin{bmatrix}
xy^*z&0\\
0&0
\end{bmatrix}
=\frac12(ab^*c+cb^*a)
\in \M_2(\Ran P^\dagger).
\]
Since $\M_2(\Ran P^\dagger)$ is a linear space, it follows that $xy^*z \in \Ran P^\dagger$. 

$2 \Rightarrow 3$. Now, assume that \eqref{unsymmetrized-triple-dual-range} holds. Let $n \geq 1$ and let $x=[x_{ij}] \in \M_n(\Ran P)$. For any integers $1 \leq i,l \leq n$, we have
\begin{equation}
\label{entries-cubic-matrix}
(xx^*x)_{il}
=
\sum_{j,k=1}^n x_{ij}x_{kj}^*x_{kl}.
\end{equation}
Each term on the right-hand side belongs to $\Ran P^\dagger$ by assumption. Hence
\[
xx^*x \in \M_n(\Ran P^\dagger)
=
\Ran P_n^\dagger.
\]
Proposition~\ref{prop-contractivity-cubic-characterization}, applied to the projection $P_n$, shows that $P_n$ is contractive. Since this holds for any $n \geq 1$, the projection $P$ is completely contractive. Thus $(2) \Rightarrow(3)$.

The equivalence with the fourth assertion follows from the same argument. Finally, once $P_n$ is contractive, Proposition~\ref{prop-contractivity-cubic-characterization} gives
\[
\Ran P_n^\dagger
=\{xx^*x : x \in \Ran P_n\}.
\]
Using \eqref{range-adjoint-amplification} and $\Ran P_n = \M_n(\Ran P)$, we obtain \eqref{matrix-duality-map-equality}.
\end{proof}

\begin{remark} \normalfont
The matricial hypothesis in Theorem~\ref{thm-two-contractive-completely-contractive-L4} cannot be replaced by scalar contractivity. Indeed, let $\sigma$ be the transpose on $S_2^4$ and let
\[
P
\ov{\mathrm{def}}{=}
\frac12(\Id+\sigma).
\]
The projection $P$ is contractive and its range is the space of symmetric $2\times2$ matrices. Moreover, $P^\dagger$ is the same symmetrization on $S_2^{\frac43}$. If $
x=e_{11}$, $y=e_{12}+e_{21}$, $z=e_{22}$ then $x,y,z\in\Ran P$, whereas
\[
xy^*z=e_{12}\notin\Ran P^\dagger.
\]
Consequently, Theorem~\ref{thm-two-contractive-completely-contractive-L4} shows that $P$ is not $2$-contractive.
\end{remark}

\begin{cor}
\label{cor-dual-TRO-product-L4}
Under the assumptions of Theorem~\ref{thm-two-contractive-completely-contractive-L4}, suppose that one of its equivalent conditions holds. Then
\begin{equation}
\label{dual-range-generated-by-triples}
\Ran P^\dagger
=
\{xy^*z : x,y,z \in \Ran P\}
=
\{yy^*y : y \in \Ran P\}.
\end{equation}
Consequently,
\begin{equation}
\label{kernel-ternary-annihilator}
\ker P
=\{w \in \L^4(\cal M) : \tr((xy^*z)^*w)=0 \text{ for all }x,y,z \in \Ran P\}.
\end{equation}
\end{cor}

\begin{proof}
By Theorem~\ref{thm-two-contractive-completely-contractive-L4}, any product $xy^*z$ belongs to $\Ran P^\dagger$. Conversely, Proposition~\ref{prop-contractivity-cubic-characterization} gives
\[
\Ran P^\dagger
=\{yy^*y : y \in \Ran P\}.
\]
Since each diagonal product $yy^*y$ is a product of the required form, this proves \eqref{dual-range-generated-by-triples}. Equality \eqref{kernel-ternary-annihilator} follows from $\Ran P^\dagger=(\ker P)^\perp$.
\end{proof}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{The conjugate exponent $\frac43$}

The preceding result has the following dual consequence.

\begin{cor}
\label{cor-dual-two-contractive-L43}
Let $Q\co\L^{\frac43}(\cal M)\to\L^{\frac43}(\cal M)$ be a bounded projection. Then $Q$ is $2$-contractive if and only if it is completely contractive.

Suppose, in addition, that $Q$ is contractive, and set $Z\ov{\mathrm{def}}{=}\Ran Q$. Then
\begin{equation}
\label{dual-range-L43}
\Ran Q^\dagger
=
\frak j_{\frac43}(Z).
\end{equation}
In particular, $\frak j_{\frac43}(Z)$ is a linear subspace of $\L^4(\cal M)$ and
\begin{equation}
\label{dual-Jordan-condition-L43}
\big\{\frak j_{\frac43}(Z),\frak j_{\frac43}(Z),\frak j_{\frac43}(Z)\big\}
=Z.
\end{equation}
Under this scalar contractivity assumption, the following assertions are equivalent:
\begin{enumerate}
\item $Q$ is $2$-contractive;
\item $Q$ is completely contractive;
\item for any $x,y,z\in Z$, we have
\begin{equation}
\label{dual-TRO-condition-L43}
\frak j_{\frac43}(x)\frak j_{\frac43}(y)^*\frak j_{\frac43}(z)
\in Z.
\end{equation}
\end{enumerate}
If these conditions hold, then
\begin{equation}
\label{dual-exact-TRO-image-L43}
Z
=
\big\{\frak j_{\frac43}(x)\frak j_{\frac43}(y)^*\frak j_{\frac43}(z):x,y,z\in Z\big\}.
\end{equation}
\end{cor}

\begin{proof}
Let
\[
P\ov{\mathrm{def}}{=}Q^\dagger
\co\L^4(\cal M)\to\L^4(\cal M)
\]
be the adjoint of $Q$ under the trace duality. Thus
\begin{equation}
\label{adjoint-Q-L43}
\tr\big(P(a)^*x\big)
=
\tr\big(a^*Q(x)\big),
\quad
x \in \L^{\frac43}(\cal M), a \in\L^4(\cal M).
\end{equation}
The map $P$ is a bounded projection and, by reflexivity, $P^\dagger=Q$. For any integer $n \geq 1$, set
\[
Q_n\ov{\mathrm{def}}{=}\Id_{S_n^{\frac43}}\ot Q
\quad\text{and}\quad
P_n\ov{\mathrm{def}}{=}\Id_{S_n^4}\ot P.
\]
Under the duality associated with $\Tr_n\ot\tr$, we have
\begin{equation}
\label{adjoint-amplification-L43}
Q_n^\dagger=P_n.
\end{equation}
Indeed, if $X=[x_{ij}]\in\L^{\frac43}(\M_n\otvn\cal M)$ and $A=[a_{ij}]\in\L^4(\M_n\otvn\cal M)$, then
\begin{align*}
(\Tr_n\ot\tr)\big(A^*Q_n(X)\big)
=
\sum_{i,j=1}^n\tr\big(a_{ij}^*Q(x_{ij})\big)
\ov{\eqref{adjoint-Q-L43}}{=}
\sum_{i,j=1}^n\tr\big(P(a_{ij})^*x_{ij}\big)
=
(\Tr_n\ot\tr)\big(P_n(A)^*X\big).
\end{align*}
Consequently, we have
\begin{equation}
\label{amplification-norm-duality-L43}
\norm{Q_n}
=
\norm{P_n},
\quad n \geq 1.
\end{equation}
It follows that $Q$ is $2$-contractive if and only if $P$ is $2$-contractive, and that $Q$ is completely contractive if and only if $P$ is completely contractive. Theorem~\ref{thm-two-contractive-completely-contractive-L4}, applied to $P$, proves the first assertion.

Suppose now that $Q$ is contractive. Then $P$ is contractive by \eqref{amplification-norm-duality-L43}. Set $Y\ov{\mathrm{def}}{=}\Ran P$. Since $P^\dagger=Q$, Proposition~\ref{prop-contractivity-cubic-characterization} gives
\[
\frak j_4(Y)
=
\Ran P^\dagger
=Z.
\]
Applying $\frak j_{\frac43}$ and using \eqref{inverse-duality-maps}, we obtain
\[
Y
=
\frak j_{\frac43}(Z).
\]
Since $Y=\Ran Q^\dagger$, this proves \eqref{dual-range-L43}. Corollary~\ref{cor-dual-triple-closure}, applied to $P$, yields
\[
\{Y,Y,Y\}
=
\Ran P^\dagger
=Z,
\]
which is precisely \eqref{dual-Jordan-condition-L43}.

Finally, Theorem~\ref{thm-two-contractive-completely-contractive-L4} gives
\[
P\text{ is $2$-contractive}
\quad\Longleftrightarrow\quad
YY^*Y\subset\Ran P^\dagger.
\]
Using $Y=\frak j_{\frac43}(Z)$, $\Ran P^\dagger=Z$ and \eqref{amplification-norm-duality-L43}, this is exactly the equivalence of the three assertions. Under these conditions, Corollary~\ref{cor-dual-TRO-product-L4}, applied to $P$, also gives \eqref{dual-exact-TRO-image-L43}.
\end{proof}


\subsection{Left and right tangent Hilbert spaces and 2-contractive projections}
\label{sec-left-right-contractivity}


By \eqref{Levi-L4-polar}, we have
\begin{equation}
\scr{L}_{\Phi_4,a}(y,z)
=2\left(\tr(z^*aa^*y)+\tr(az^*ya^*)\right), \quad y,z \in \L^4(\cal{M}).
\end{equation}
In particular, we have
\begin{equation}
\label{norm-Levi-2}
\frac12\scr L_{\Phi_4,a}(z,z)
=\norm{a^*z}_{\L^2(\cal{M})}^2+\norm{za^*}_{\L^2(\cal{M})}^2.
\end{equation}
We introduce the following positive semidefinite Hermitian forms
\begin{equation}
\label{left-right-inner}
\la z,w \ra_{\ell,a}
\ov{\mathrm{def}}{=}
\tr(w^*aa^*z)
\quad \text{and} \quad
\la z,w \ra_{r,a}
\ov{\mathrm{def}}{=}
\tr(aw^*za^*).
\end{equation}
By quotienting and completing, we obtain two Hilbert spaces $\cal{H}_{\ell,a}$ and $\cal{H}_{r,a}$ with norms
\begin{equation}
\norm{z}_{\ell,a}
=\norm{a^*z}_{\L^2(\cal{M})},
\quad \text{and} \quad
\norm{z}_{r,a}
=\norm{za^*}_{\L^2(\cal{M})}.
\end{equation}

The next result gives separate left and right tangent estimates. This is stronger than the estimate obtained from scalar contractivity, which only controls the sum of the two forms. The additional $2$-contractivity assumption separates the two sides.

\begin{prop}
\label{prop-left-right-tangent-estimates-two-contractive}
Let $P \co \L^4(\cal{M}) \to \L^4(\cal{M})$ be a $2$-contractive projection. For any $a \in \Ran P$ and any $z \in \L^4(\cal{M})$, we have
\begin{equation}
\label{separate-left-right-tangent-estimates}
\norm{P(z)}_{\ell,a}
\leq
\norm{z}_{\ell,a}
\quad \text{and} \quad
\norm{P(z)}_{r,a}
\leq
\norm{z}_{r,a}.
\end{equation}
Moreover, for any $a,y \in \Ran P$, we have
\begin{equation}
\label{left-right-Riesz-vectors-dual-range}
P^\dagger(aa^*y)
=aa^*y
\quad \text{and} \quad
P^\dagger(ya^*a)
=ya^*a.
\end{equation}
\end{prop}

\begin{proof}
Let $a \in \Ran P$ and let $z \in \L^4(\cal{M})$. For any $t > 0$, consider the element $X_t \ov{\mathrm{def}}{=}
\begin{bmatrix}
a & tz\\
0 & 0
\end{bmatrix}$ of the space $\L^4(\M_2(\cal{M}))$. Since the projection $P_2 \co \L^4(\M_2(\cal M)) \to \L^4(\M_2(\cal{M}))$ is contractive and $P(a)=a$, we have
\[
\norm{
\begin{bmatrix}
a & tP(z)\\
0 & 0
\end{bmatrix}}_{\L^4(\M_2(\cal M))}^4
=\norm{P_2(X_t)}_{\L^4(\M_2(\cal M))}^4
\leq \norm{X_t}_{\L^4(\M_2(\cal M))}^4
=\norm{
\begin{bmatrix}
a & tz\\
0 & 0
\end{bmatrix}}_{\L^4(\M_2(\cal M))}^4.
\]
Since $
X_tX_t^*
=
\begin{bmatrix}
aa^*+t^2zz^*&0\\
0&0
\end{bmatrix}$, we obtain
\begin{align*}
\MoveEqLeft
\norm{X_t}_{\L^4(\M_2(\cal M))}^4
=\tr\big[(aa^*+t^2zz^*)^2\big] 
= \norm{a}_{\L^4(\cal{M})}^4
+2t^2\norm{a^*z}_{\L^2(\cal{M})}^2
+t^4\norm{z}_{\L^4(\cal{M})}^4.
\end{align*}
Consequently, we have
\[
\norm{a}_{\L^4(\cal{M})}^4
+
2t^2\norm{a^*P(z)}_{\L^2(\cal{M})}^2
+
t^4\norm{P(z)}_{\L^4(\cal{M})}^4
\leq
\norm{a}_{\L^4(\cal{M})}^4
+
2t^2\norm{a^*z}_{\L^2(\cal{M})}^2
+
t^4\norm{z}_{\L^4(\cal{M})}^4.
\]
After division by $2t^2$ and passage to the limit $t\to 0^+$, we obtain the inequality
\[
\norm{a^*P(z)}_{\L^2(\cal{M})}
\leq
\norm{a^*z}_{\L^2(\cal{M})}.
\]
Using instead the column matrix $Y_t
\ov{\mathrm{def}}{=}
\begin{bmatrix}
a & 0\\
tz & 0
\end{bmatrix}$ and the identity $
Y_t^*Y_t
=
\begin{bmatrix}
a^*a+t^2z^*z&0\\
0&0
\end{bmatrix}$, we similarly obtain the inequality
\[
\norm{P(z)a^*}_{\L^2(\cal{M})}
\leq
\norm{za^*}_{\L^2(\cal{M})}.
\]
This proves \eqref{separate-left-right-tangent-estimates}. Finally, for $a,y \in \Ran P$, the two products $aa^*y$ and $ya^*a$ are fixed by $P^\dagger$ by \eqref{unsymmetrized-triple-fixed}. This proves \eqref{left-right-Riesz-vectors-dual-range}.
\end{proof}


Conversely, either one of the two families of tangent estimates already characterizes complete contractivity.

\begin{prop}
\label{prop-one-sided-tangent-criterion}
Let $P \co \L^4(\mathcal M) \to \L^4(\mathcal M)$ be a bounded projection. The following assertions are equivalent:
\begin{enumerate}
\item $P$ is completely contractive.
\item $\|u^*P(x)\|_{\L^2(\cal{M})}\leq\|u^*x\|_{\L^2(\cal{M})}$ for any $u\in \Ran P$ and any $x\in\L^4(\mathcal M)$.
\item $\|P(x)u^*\|_{\L^2(\cal{M})}\leq\|xu^*\|_{\L^2(\cal{M})}$ for any $u\in \Ran P$ and any $x\in\L^4(\mathcal M)$.
\end{enumerate}
\end{prop}

\begin{proof}
The first assertion implies the other two by Proposition~\ref{prop-left-right-tangent-estimates-two-contractive}. Suppose that the second assertion holds. For fixed $u\in \Ran P$, the projection $P$ induces a contractive, hence orthogonal, projection on the Hilbert completion associated with the positive semidefinite form $h_u(x,z)=\Tr(z^*uu^*x)$. Therefore $h_u(x-Px,z)=0$ for any $z \in \Ran P$ and every $x \in \L^4(\cal{M})$. Equivalently, $uu^*z \in \Ran P^\dagger$. Polarization gives
\begin{equation*}
xy^*z
=\frac14\sum_{k=0}^3 \i^k(x+\i^ky)(x+\i^ky)^*z \in \Ran P^\dagger,
\quad x,y,z \in \Ran P.
\end{equation*}
Theorem~\ref{thm-two-contractive-completely-contractive-L4} implies complete contractivity. The third assertion is treated in the same way, using the forms $k_u(x,z)=\Tr(u z^*x u^*)$ and the polarization formula $xy^*z
=
\frac14\sum_{k=0}^3(-\i)^k
x(y+\i^kz)^*(y+\i^kz)$.
\end{proof}




%\begin{remark} \normalfont
%\label{rem-geometric-meaning-separate-tangent-estimates}
%%Proposition~\ref{prop-left-right-tangent-estimates-two-contractive} admits a useful geometric interpretation. For $a\in\Ran P$, split the tangent form \eqref{Levi-gamma-identification} into its left and right parts
%%\begin{equation}
%%\label{left-right-tangent-forms}
%%\la z,w \ra_{\ell,a}
%%\ov{\mathrm{def}}{=}
%%\tr(w^*aa^*z),
%%\quad
%%\la z,w \ra_{r,a}
%%\ov{\mathrm{def}}{=}
%%\tr(aw^*za^*).
%%\end{equation}
%%These are positive sesquilinear forms and
%%\[
%%\gamma_a^\ell(z,z)=\norm{a^*z}_2^2,
%%\qquad
%%\gamma_a^r(z,z)=\norm{za^*}_2^2,
%%\qquad
%%\gamma_a=\gamma_a^\ell+\gamma_a^r.
%%\]
%%Scalar contractivity only implies that $P$ contracts the sum $\gamma_a^\ell+\gamma_a^r$. The additional $2$-contractivity assumption separates the two sides and Proposition~\ref{prop-left-right-tangent-estimates-two-contractive} says precisely that
%%\begin{equation}
%%\label{separate-tangent-form-contractions}
%%\gamma_a^\ell(Pz,Pz)\leq\gamma_a^\ell(z,z)
%%\quad\text{and}\quad
%%\gamma_a^r(Pz,Pz)\leq\gamma_a^r(z,z).
%%\end{equation}
%%Thus $P$ is contractive not only for the symmetrized tangent Hilbert geometry, but independently for the left and right tangent geometries.
%
%As before, after quotienting by the null spaces of $\gamma_a^\ell$ and $\gamma_a^r$ and completing, $P$ induces contractive projections on Hilbert spaces, hence orthogonal projections. Therefore, for every $y\in\Ran P$,
%\begin{equation}
%\label{separate-tangent-orthogonality}
%\la Pz,y\ra_{\ell,a}=\la z,y\ra_{\ell,a}
%\quad\text{and}\quad
%\la Pz,y\ra_{r,a}=\la z,y\ra_{r,a}.
%\end{equation}
%The corresponding Riesz representatives with respect to the $\L^4$--$\L^{4/3}$ duality are $aa^*y$ and $ya^*a$, respectively:
%\[
%\la z,y \ra_{\ell,a}
%=\tr((aa^*y)^*z),
%\qquad
%\la z,y \ra_{r,a}
%=\tr((ya^*a)^*z).
%\]
%Consequently, \eqref{separate-tangent-orthogonality} is \textbf{exactly equivalent to}
%\[
%P^\dagger(aa^*y)=aa^*y,
%\qquad
%P^\dagger(ya^*a)=ya^*a.
%\]
%This explains the second assertion of Proposition~\ref{prop-left-right-tangent-estimates-two-contractive}: the two fixed vectors in the dual range are the Riesz representatives of the two separate tangent Hilbert structures.

%The distinction between scalar and $2$-contractivity is important here. At the scalar level, the fourth-power norm sees the two quantities $\norm{a^*z}_2^2$ and $\norm{za^*}_2^2$ only through their sum, and therefore naturally produces the symmetrized expression
%\[
%aa^*y+ya^*a=2\{a,a,y\}.
%\]
%Passing to $2\times2$ row and column matrices separates the two orientations. The row computation detects one of the forms in \eqref{left-right-tangent-forms}, while the column computation detects the other one. Thus matricial contractivity remembers an orientation which is invisible to scalar Banach-space geometry.

%This phenomenon is closely related to the expected TRO structure of the range. For a TRO $V$, the two products $VV^*$ and $V^*V$ generate the two diagonal algebras of its linking algebra, while $V$ itself is the off-diagonal corner. The forms $\gamma_a^\ell$ and $\gamma_a^r$ should therefore be regarded as infinitesimal versions of these two diagonal structures. The relations
%\[
%aa^*y\in\Ran P^\dagger
%\quad\text{and}\quad
%ya^*a\in\Ran P^\dagger
%\]
%are the first indication that the left and right coefficient structures can be recovered separately. After matricial amplification, the corresponding finite-family estimates lead to orthogonal projections on the $\L^2$-closures of $\operatorname{span}(YY^*)$ and $\operatorname{span}(Y^*Y)$, which are the natural candidates for the two diagonal $\L^2$-corners of the linking algebra associated with the sought TRO.
%
%In this sense, Proposition~\ref{prop-left-right-tangent-estimates-two-contractive} is not merely a strengthening of a norm estimate. It is the point where the local Banach geometry begins to distinguish the left and right operator-algebraic structures. Scalar contractivity leads naturally to Jordan triple information, whereas $2$-contractivity separates the two sides and points toward the ordered TRO product $xy^*z$.
%\end{remark}



%\begin{cor}
%\label{cor-one-sided-orthogonality-two-contractive-L4}
%Let $P \co \L^4(\cal{M}) \to \L^4(\cal{M})$ be a $2$-contractive projection. If $y \in \Ran P$ and $z\in\L^4(\cal{M})$, then
%\[
%y^*z=0
%\quad \to \quad
%y^*P(z) = 0
%\]
%and
%\[
%yz^* = 0
%\quad\to\quad
%yP(z)^* = 0.
%\]
%Consequently, if
%\[
%e=\bigvee_{y \in Y}s_\ell(y)
%\qquad\text{and}\qquad
%f=\bigvee_{y \in Y}s_r(y),
%\]
%then
%\[
%P(x)=P(exf),
%\qquad
%x \in \L^4(\cal{M}).
%\]
%\end{cor}
%
%\begin{proof}
%If $y^*z=0$, then $\norm{y^*z}_2=0$, and the first inequality in \eqref{separate-left-right-tangent-estimates} gives $y^*P(z)=0$. The second assertion follows similarly. The corner reduction is proved as in Proposition~\ref{prop-support-reduction-L4-contractivity}.
%\end{proof}

%\begin{remark} \normalfont
%\label{rem-Jordan-versus-TRO-dual-range}
%For a merely contractive projection on $\L^4(\cal{M})$, scalar polarization only gives
%\[
%xy^*z+zy^*x \in \Ran P^\dagger.
%\]
%The $2$-contractivity assumption allows one to place the variables in suitable matrix entries and eliminate the second summand. This gives the ordered ternary product
%\[
%xy^*z\in \Ran P^\dagger.
%\]
%Thus scalar contractivity detects a Jordan triple structure, whereas $2$-contractivity detects a TRO structure at the level of the dual range.
%\end{remark}

%\begin{remark} \normalfont
%\label{rem-remaining-density-problem-L4}
%Theorem~\ref{thm-two-contractive-completely-contractive-L4} does not yet imply that
%\[
%YY^*Y \subset Y.
%\]
%What it gives is the exact relation
%\[
%YY^*Y \subset \Ran P^\dagger.
%\]
%The remaining problem is therefore to identify the dual range $\Ran P^\dagger$ with a two-sided change of density of $Y$. More precisely, one expects to construct positive operators $h$ and $k$ and a W$^*$-TRO $V$ such that $Y$ is completely isometric to the corresponding rectangular $\L^4$-space. Once such an identification is obtained, the relation $YY^*Y\subset \Ran P^\dagger$ should become the ordinary TRO relation $VV^*V\subset V$.
%\end{remark}



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Recall that, if $x=u|x|$ is the polar decomposition of a measurable operator, its left and right support projections are
\[
s_\ell(x)\ov{\mathrm{def}}{=}uu^*
\quad\text{and}\quad
s_r(x)\ov{\mathrm{def}}{=}u^*u.
\]
If $h$ is positive, we write
\[
s(h)\ov{\mathrm{def}}{=}s_\ell(h)=s_r(h).
\]
For a subspace $Y\subset\L^p(\cal M)$, we set
\[
s_\ell(Y)\ov{\mathrm{def}}{=}\bigvee_{x\in Y}s_\ell(x)
\quad\text{and}\quad
s_r(Y)\ov{\mathrm{def}}{=}\bigvee_{x\in Y}s_r(x).
\]
If $P$ is a projection, we write $s_\ell(P)\ov{\mathrm{def}}{=}s_\ell(\Ran P)$ and $s_r(P)\ov{\mathrm{def}}{=}s_r(\Ran P)$. Thus
\begin{equation}
\label{range-support-corner}
\Ran P
\subset
s_\ell(P)\L^p(\cal M)s_r(P).
\end{equation}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Left and right parts of the Levi form}
\label{sec-modular-finite}
\begin{prop}
\label{prop-global-left-right-Levi-finite-dimensional}
Let $\cal M$ be a semifinite von Neumann algebra equipped with a normal semifinite faithful trace. Let
$
P\co\L^4(\cal M)\to\L^4(\cal M)$ be a finite-rank $2$-contractive projection with nonzero range $Y$. Put $
e\ov{\mathrm{def}}{=}s_\ell(Y)$ and $f\ov{\mathrm{def}}{=}s_r(Y)$. Let $(x_1,\ldots,x_k)$ be a linear basis of the space $Y$ and define
\begin{equation}
\label{global-Gram-densities}
a
\ov{\mathrm{def}}{=}
\sum_{j=1}^k x_jx_j^*,
\quad \text{and} \quad
b
\ov{\mathrm{def}}{=}
\sum_{j=1}^k x_j^*x_j.
\end{equation}
Then $a,b \in \L^2(\cal M)$,
\begin{equation}
\label{a-b-supports}
s(a)=s_\ell(P) 
\quad \text{and} \quad s(b)=s_r(P)
\end{equation}
and
\begin{equation}
\label{global-exact-density-ranges}
a(\Ran P)
=\Ran P^\dagger
=(\Ran P)b.
\end{equation}
If, in addition, $\cal M$ is finite-dimensional, then $a$ and $b$ are invertible in the reduced algebras $e\cal Me$ and $f\cal Mf$, respectively.
\end{prop}



%\begin{proof}
%Put $Y^\sharp=\Ran P^\dagger$. H\"older's inequality gives $a,b\in\L^2(\cal M)$. The support identities follow from
%\begin{equation*}
%s(a)=\bigvee_{j=1}^m s_\ell(x_j)=e,
%\qquad
%s(b)=\bigvee_{j=1}^m s_r(x_j)=f.
%\end{equation*}
%By Theorem~\ref{thm-two-contractive-completely-contractive-L4}, the products $x_jx_j^*y$ and $yx_j^*x_j$ belong to $Y^\sharp$ whenever $y\in Y$. Hence $aY\subset Y^\sharp$ and $Yb\subset Y^\sharp$.
%
%
%\end{proof}

\begin{proof}
The support identities follow from
\[
s(a)
\ov{\eqref{global-Gram-densities}}{=}
\bigvee_{j=1}^k s_\ell(x_j)
= s_\ell(P)
\quad\text{and}\quad
s(b)
\ov{\eqref{global-Gram-densities}}{=}
\bigvee_{j=1}^k s_r(x_j)
= s_r(P).
\]
By Theorem~\ref{thm-two-contractive-completely-contractive-L4}, for $x,y,z \in \Ran P$ we have $xy^*z \ov{\eqref{unsymmetrized-triple-dual-range}}{\in} \Ran P^\dagger$. Consequently, for any $y \in \Ran P$, we have
\[
ay
\ov{\eqref{global-Gram-densities}}{=}
\sum_{j=1}^k x_jx_j^*y
\in \Ran P^\dagger
\quad \text{and} \quad
yb
\ov{\eqref{global-Gram-densities}}{=}
\sum_{j=1}^k yx_j^*x_j
\in \Ran P^\dagger.
\]
Thus the left and right multiplication maps $L_a \co \Ran P \to \Ran P^\dagger$, $y \mapsto ay$ and $R_b \co \Ran P \to \Ran P^\dagger$, $y \mapsto yb$ are well-defined. 

We prove the injectivity without assuming that the densities have bounded inverses. Suppose that $ay=0$ for some $y\in Y$. For $n\geq1$, put $e_n=\mathbf1_{[\frac{1}{n},n]}(a)$ and let $g_n(t)=t^{-1}\mathbf1_{[\frac{1}{n},n]}(t)$, with $g_n(0)=0$. Since $g_n(a) \in \cal M$ and $g_n(a)a=e_n$, we obtain $e_ny=g_n(a)(ay)=0$. The projections $e_n$ increase strongly to $e$, so $e_ny\to ey=y$ in $\L^4(\cal M)$. Thus $y=0$. The same argument on the right shows that $yb=0$ implies $y=0$.

The two multiplication maps from $Y$ into $Y^\sharp$ are therefore injective. Since $P$ has finite rank and $\dim Y^\sharp=\dim Y$, both maps are surjective. This proves \eqref{global-exact-density-ranges}. Finally, if $\cal M$ is finite-dimensional, the support identities imply that $a$ and $b$ are invertible in $e\cal Me$ and $f\cal Mf$, respectively.
%Since the elements $a$ and $b$ have full supports, both maps are injective. Since $\dim \Ran P=\dim \Ran P^\dagger$, they are bijective, proving \eqref{global-exact-density-ranges}. 
\end{proof}



\begin{prop}
\label{prop-global-left-right-Levi-finite-dimensional-bis}
Assume, in addition to the assumptions and notation of Proposition~\ref{prop-global-left-right-Levi-finite-dimensional}, that $\cal M$ is finite-dimensional. Define the forms
\begin{align}
\label{global-left-Levi-form}
\frak{H}^\ell(x,y)
&\ov{\mathrm{def}}{=}
\sum_{j=1}^k \tr(y^*x_jx_j^*x)
\ov{\eqref{global-Gram-densities}}{=} \tr(y^*ax), \quad x,y \in \Ran P, \\
\label{global-right-Levi-form}
\frak{H}^r(x,y)
&\ov{\mathrm{def}}{=}
\sum_{j=1}^k \tr(y^*xx_j^*x_j)
\ov{\eqref{global-Gram-densities}}{=} \tr(y^*xb), \quad x,y \in \Ran P.
\end{align}
These Hermitian forms are positive definite on the subspace $\Ran P$. The relative metric endomorphism defined in Proposition \ref{prop-relative-Riesz-operator} by
\begin{equation}
\label{global-Levi-comparison}
\frak{H}^r(x,y)
=\frak{H}^\ell(x,\Delta y), \quad x,y \in \Ran P,
\end{equation}
is given by
\begin{equation}
\label{global-relative-modular-operator}
\Delta(y)
\ov{\mathrm{def}}{=}
a^{-1}yb,
\quad
y\in \Ran P. 
\end{equation}
Furthermore, we have
\begin{equation}
\label{global-relative-complex-powers}
\Delta^z(y)
=a^{-z} y b^z,
\quad y \in \Ran P, z \in \mathbb{C}.
\end{equation}
\end{prop}

Here $a^{-1}$ denotes the inverse of $a$ in the reduced algebra $e\cal{M}e$. More generally, the powers $a^{-z}$ and $b^z$ in \eqref{global-relative-complex-powers} are computed in the reduced algebras $e\cal{M}e$ and $f\cal{M}f$, respectively.

\begin{proof}
Set $Y \ov{\mathrm{def}}{=} \Ran P$ and $Y^\sharp\ov{\mathrm{def}}{=} \Ran P^\dagger$. Since $a$ and $b$ have full supports, the two forms are positive definite. Indeed, we have
\[
\frak H^\ell(x,x)
=
\tr(x^*ax)
=
\bnorm{a^{\frac12}x}_2^2
\quad\text{and}\quad
\frak H^r(x,x)
=
\tr(x^*xb)
=
\bnorm{xb^{\frac12}}_2^2.
\]
Consider the trace-duality map $\iota \co Y^\sharp \to Y^*$ defined by
$$
[\iota(\eta)](x)
\ov{\mathrm{def}}{=}
\tr(\eta^*x),
\quad x \in Y,\eta \in Y^\sharp.
$$
This map is conjugate linear and injective. Indeed, if $\iota(\eta)=0$ for some $\eta\in Y^\sharp$, then $P^\dagger(\eta)=\eta$ and, for any $z \in \L^4(\cal M)$,
\[
\tr(\eta^*z)
=
\tr((P^\dagger\eta)^*z)
\ov{\eqref{def-adjoint-P-L4}}{=}
\tr(\eta^*Pz)
=
0.
\]
The nondegeneracy of the $\L^4$--$\L^{4/3}$ duality gives $\eta=0$. Since $\dim Y^\sharp=\dim Y$, the map $\iota$ is a conjugate linear bijection from $Y^\sharp$ onto $Y^*$. For $y\in Y$, define
\[
L_a(y)\ov{\mathrm{def}}{=}ay
\quad\text{and}\quad
R_b(y)\ov{\mathrm{def}}{=}yb.
\]
By \eqref{global-exact-density-ranges}, both maps are linear bijections from $Y$ onto $Y^\sharp$. Moreover, for $x,y\in Y$,
\[
[\iota L_a(y)](x)
=
\tr((ay)^*x)
=
\tr(y^*ax)
=
\frak H^\ell(x,y),
\]
whereas
\[
[\iota R_b(y)](x)
=
\tr((yb)^*x)
=
\tr(by^*x)
=
\tr(y^*xb)
=
\frak H^r(x,y).
\]
Thus $\iota L_a$ and $\iota R_b$ are the Riesz maps of $\frak H^\ell$ and $\frak H^r$, respectively. Proposition~\ref{prop-relative-Riesz-operator} gives
\[
\Delta
=
(\iota L_a)^{-1}(\iota R_b)
=
L_a^{-1}R_b.
\]
Consequently, we have
\[
\Delta(y)
=
a^{-1}yb,
\quad
y\in Y,
\]
which proves \eqref{global-relative-modular-operator}. Moreover,
\[
\frak H^r(x,y)
=
\frak H^\ell(x,\Delta y),
\quad
x,y\in Y,
\]
and $\Delta$ is positive and selfadjoint for the Hilbertian structure defined by $\frak H^\ell$. It remains to prove \eqref{global-relative-complex-powers}. On the full corner $e\cal Mf$, define
\[
T(\xi)
\ov{\mathrm{def}}{=}
a^{-1}\xi b.
\]
Equip $e\cal Mf$ with the Hermitian inner product
\[
\langle\xi,\eta\rangle_a
\ov{\mathrm{def}}{=}
\tr(\eta^*a\xi).
\]
For $\xi,\eta \in e\cal Mf$, we have
\begin{align*}
\langle T\xi,\eta\rangle_a
=
\tr(\eta^*a a^{-1}\xi b)
=
\tr(\eta^*\xi b)
\quad \text{and} \quad \langle\xi,T\eta\rangle_a
=
\tr((a^{-1}\eta b)^*a\xi)
=
\tr(b\eta^*\xi)
=
\tr(\eta^*\xi b).
\end{align*}
Hence $T$ is selfadjoint. Furthermore, we have
\[
\la T\xi,\xi \ra_a
=
\tr(\xi^*\xi b)
= \bnorm{\xi b^{\frac12}}_2^2
\geq0,
\]
so $T$ is positive. Equality \eqref{global-exact-density-ranges} gives
\[
T(Y)
=
a^{-1}Yb
=
Y.
\]
Thus $Y$ is an invariant subspace of the selfadjoint operator $T$ and is therefore reducing.
More explicitly, $L_{a^{-1}}$ and $R_b$ are commuting positive selfadjoint operators for $\langle\cdot,\cdot\rangle_a$. Indeed,
\[
\langle L_{a^{-1}}\xi,\eta\rangle_a
=
\tr(\eta^*\xi)
=
\langle\xi,L_{a^{-1}}\eta\rangle_a
\]
and
\[
\langle R_b\xi,\eta\rangle_a
=
\tr(\eta^*a\xi b)
=
\langle\xi,R_b\eta\rangle_a,
\]
while their quadratic forms are respectively $\norm{\xi}_2^2$ and $\bnorm{a^{\frac12}\xi b^{\frac12}}_2^2$. Consequently, the joint functional calculus gives
\[
T^z(\xi)
=
a^{-z}\xi b^z,
\quad
\xi\in e\cal Mf, z\in\mathbb C.
\]
Since $\Delta=T|_Y$, restricting this identity to $Y$ proves \eqref{global-relative-complex-powers}.
\end{proof}

\begin{remark} \normalfont
\label{rem-basis-dependence}
The Gram densities $a,b$ depend on the chosen basis of $\Ran P$ only through the Hermitian inner product on $\Ran P$ for which this basis is orthonormal. In particular, they are unchanged by unitary changes of basis, but not by arbitrary invertible changes of basis. Thus the construction attaches modular data to a choice of Hilbertian structure on $\Ran P$. It does not produce a single relative operator determined by $P$ alone. All the results below hold for any such choice.
\end{remark}

\begin{cor}
\label{cor-modular-complete-isometries}
Under the assumptions and notation of Proposition~\ref{prop-global-left-right-Levi-finite-dimensional-bis}, the linear maps $\Delta^{\i t}\co\Ran P\to\Ran P$, $t\in\R$, are surjective complete isometries for the inherited $\L^4$ operator space structure. Put $u_t \ov{\mathrm{def}}{=} a^{-\i t}+(1-e)$ and $v_t \ov{\mathrm{def}}{=} b^{\i t}+(1-f)$. Then
\[
P(u_txv_t)=u_tP(x)v_t,
\quad x \in \L^4(\cal M),\ t\in\R.
\]
In particular, we have
\[
P(a^{-\i t}xb^{\i t})
=a^{-\i t}P(x)b^{\i t},
\quad x \in e\L^4(\cal M)f.
\]
\end{cor}

\begin{proof}
The identity $\Delta^{\i t}(y)=a^{-\i t}yb^{\i t}$ shows that these maps are restrictions of left and right multiplication by unitaries in the reduced algebras, so they are complete isometries. To obtain the covariance relation on the whole corner, put $u_t=a^{-\i t}+(1-e)$ and $v_t=b^{\i t}+(1-f)$. The map $\mathcal U_t(x)=u_txv_t$ is a surjective complete isometry on $\L^4(\cal{M})$ and preserves $Y$. Thus $\mathcal U_t^{-1}P\cal{U}_t$ is another contractive projection onto $Y$. Uniqueness gives $P \cal{U}_t= \cal{U}_tP$, which restricts to the displayed identity.
\end{proof}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{2-contractive projections on $\L^4$-spaces and ternary rings of operators}
\label{sec-2-contract-TRO}

\begin{prop}
\label{prop-linking-modular-interpretation}
Under the assumptions and notation of Proposition~\ref{prop-global-left-right-Levi-finite-dimensional-bis}, in particular with $\cal M$ finite-dimensional, consider the corner von Neumann algebra
\[
\cal R
\ov{\mathrm{def}}{=}
\begin{bmatrix}
e\cal{M}e&e\cal{M}f\\
f\cal{M}e&f\cal{M}f
\end{bmatrix}.
\]
Here we identify $\cal R$ with the corner $s(\M_2\otvn\cal M)s$ where $s \ov{\mathrm{def}}{=}
\begin{bmatrix}
e&0\\
0&f
\end{bmatrix}$ and equip it with the faithful trace $
\tr_{\cal R}
\ov{\mathrm{def}}{=}
(\Tr_2\ot\tau)|_{\cal R}$. The corner $e \cal{M} f$ has left support $e$ and right support $f$, so $\cal R$ is its linking von Neumann algebra. Define the positive invertible density
\begin{equation}
\label{linking-density}
D
\ov{\mathrm{def}}{=}
\begin{bmatrix}
a^2&0\\
0&b^2
\end{bmatrix}.
\end{equation}
Let $(\sigma_t^{D})_{t \in \R}$ be the modular automorphism group of the faithful positive functional $\cal{R} \to \mathbb{C}$, $X \mapsto \tr_{\cal R}(DX)$. On the upper-right corner, we have
\begin{equation}
\label{linking-modular-group-corner}
\sigma_t^{D}(y)
=a^{2\i t}yb^{-2\i t},
\quad
y\in e\cal Mf.
\end{equation}
Consequently, the relative operator $\Delta$ of Proposition~\ref{prop-global-left-right-Levi-finite-dimensional-bis} satisfies
\begin{equation}
\label{relative-group-as-linking-modular-group}
\Delta^{\i t}(y)
=
\sigma_{-\frac t2}^{D}(y), 
\quad y \in \Ran P.
\end{equation}
If $\Delta_D$ denotes the standard modular operator on $\L^2(\cal R)$ associated with $D$, then on the upper-right corner
\begin{equation}
\label{standard-linking-modular-operator}
\Delta_{D}(y)
=a^2yb^{-2},
\end{equation}
and hence
\begin{equation}
\label{relative-operator-half-modular}
\Delta
=
\Delta_{D}^{-\frac12}
\end{equation}
as operators on the invariant subspace $\Ran P$.
\end{prop}

\begin{proof}
Since $\Ran P\subset e\cal Mf$ and $e$ is the join of the left supports of the elements of $\Ran P$, the weakly closed ideal generated by $(e\cal Mf)(e\cal Mf)^*$ in $e\cal Me$ has support $e$ and is therefore equal to $e\cal Me$. Similarly, the weakly closed ideal generated by $(e\cal Mf)^*(e\cal Mf)$ is $f\cal Mf$. This proves the linking-algebra assertion.

In the tracial realization of a finite-dimensional von Neumann algebra, the modular automorphism group associated with an invertible density $D$ is
\[
\sigma_t^{D}(X)
=D^{\i t}X D^{-\i t}.
\]
Since $D^{\i t}
\ov{\eqref{linking-density}}{=}
\begin{bmatrix}
a^{2\i t}&0\\
0&b^{2\i t}
\end{bmatrix}$, we obtain
\[
\sigma_t^{D}\left(\begin{bmatrix}
0&y\\
0&0
\end{bmatrix}\right)
=D^{\i t}
\begin{bmatrix}
0&y\\
0&0
\end{bmatrix}
D^{-\i t}
=\begin{bmatrix}
a^{2\i t}&0\\
0&b^{2\i t}
\end{bmatrix}\begin{bmatrix}
0&y\\
0&0
\end{bmatrix}
\begin{bmatrix}
a^{-2\i t}&0\\
0&b^{-2\i t}
\end{bmatrix}
=
\begin{bmatrix}
0&a^{2\i t}yb^{-2\i t}\\
0&0
\end{bmatrix},
\]
which proves \eqref{linking-modular-group-corner}. Taking the parameter $-\frac t2$ gives
\[
\sigma_{-\frac t2}^{D}(y)
=
a^{-\i t}yb^{\i t}
=
\Delta^{\i t}(y),
\]
where we used \eqref{global-relative-complex-powers}. This proves \eqref{relative-group-as-linking-modular-group}. The standard modular operator on the tracial Hilbert space $\L^2(\cal R)$ is
\[
\Delta_{D}
=
L_{D}R_{D^{-1}}.
\]
Its restriction to the upper-right corner is therefore
\[
y\mapsto a^2yb^{-2},
\]
which proves \eqref{standard-linking-modular-operator}. Its inverse square root is $
y\mapsto a^{-1}yb$, which is precisely $\Delta$. This proves \eqref{relative-operator-half-modular}.
\end{proof}

Following \cite{Arh24b} (see also \cite[p.~869]{KaR02}), we recall the rectangular noncommutative $\L^p$-space of a $\W^*$-$\TRO$. Let $\cal V$ be a $\W^*$-$\TRO$, let $\cal R(\cal V)$ be its linking von Neumann algebra, and assume that $\cal R(\cal V)$ is $\sigma$-finite and equipped with a normal faithful state $\varphi$. If $q\in\cal R(\cal V)$ is the projection defining the upper-right corner $\cal V=q\cal R(\cal V)(1-q)$, then
\[
\L^p(\cal V,\varphi)
\ov{\mathrm{def}}{=}
q\L^p(\cal R(\cal V),\varphi)(1-q),
\quad
1 \leq p < \infty,
\]
where the ambient spaces are Kosaki's noncommutative $\L^p$-spaces introduced in \cite{Kos84}. We also set $
\L^\infty(\cal V,\varphi)
\ov{\mathrm{def}}{=}
\cal V$. If, in addition, $q$ belongs to the centralizer of $\varphi$, then the symmetric Kosaki realization preserves the four corners determined by $q$ and $1-q$. In this case, $\L^p(\cal V,\varphi)$ identifies with the closure of the algebraic corner $\cal V=q\cal R(\cal V)(1-q)$ in the symmetric realization of $\L^p(\cal R(\cal V),\varphi)$.


We shall use the following finite-dimensional realization. Let $\cal R$ be a finite-dimensional von Neumann algebra with faithful trace $\tr_{\cal R}$, and let $\psi(X)=\tr_{\cal R}(dX)$ for a positive invertible density $d\in\cal R$. In the symmetric realization of Kosaki's space, an element $X\in\cal R$ represents the vector
\begin{equation}
\label{Kosaki-symmetric-realization}
d^{\frac1{2p}}X d^{\frac1{2p}}
\in\L^p(\cal R,\tr_{\cal R}),
\end{equation}
and its $\L^p(\cal R,\psi)$-norm is the tracial $\L^p$-norm of this vector. We use the same realization at every matrix level, with density $\Id_{\M_n}\ot d$. If $q\in\cal R$ is a projection commuting with $d$, then the symmetric realization preserves the corners determined by $q$. Indeed,
\[
d^{\frac1{2p}}qX(1-q)d^{\frac1{2p}}
=
q\,d^{\frac1{2p}}Xd^{\frac1{2p}}(1-q),
\quad
X \in \cal R.
\]
Equivalently, $q$ belongs to the centralizer of the state $\psi$, and the algebraic upper-right corner is represented by the corresponding upper-right corner of the tracial $\L^p$-space.
%Here $p^*$ is the conjugate index of $p$, so that $D^{\frac{1-\theta}{p^*}} y D^{\frac{\theta}{p^*}}$ belongs to $L^1(\cal{M},\varphi)$ provided that $y$ belongs to $\L^p(\cal{M},\varphi)$.



The preceding interpretation also explains the quarter-powers appearing in the deweighting procedure. We shall use the following consequence of Takesaki's conditional expectation theorem and the $\L^p$-extension of conditional expectations from \cite[Lemma~2.2]{JuX03}.

\begin{lemma}
\label{lem-modular-invariant-Kosaki-subspace}
Let $\cal N$ be a von Neumann subalgebra of a von Neumann algebra $\cal R$ with the same unit. Let $\psi$ be a normal faithful state on $\cal R$ and suppose that
\begin{equation}
\label{modular-invariance-subalgebra}
\sigma_t^\psi(\cal N)=\cal N,
\quad t\in\R.
\end{equation}
Let $\varphi\ov{\mathrm{def}}{=}\psi|_{\cal N}$. Suppose that  $1 \leq p \leq \infty$. The canonical inclusion $\cal N\subset\cal R$ induces a complete isometry
\begin{equation}
\label{Kosaki-complete-isometric-inclusion}
J_p\co \L^p(\cal N,\varphi) \to \L^p(\cal R,\psi).
\end{equation}
Moreover, the range of $J_p$ is the range of a completely contractive projection. If $\cal R$ is finite-dimensional, let $\tr_{\cal R}$ be a faithful trace, and let $d_{\cal R}\in\cal R$ and $d_{\cal N}\in\cal N$ be the densities of $\psi$ and $\varphi$ with respect to $\tr_{\cal R}$ and $\tr_{\cal R}|_{\cal N}$, respectively. For $1\leq p<\infty$, the maps above are characterized in the symmetric tracial realizations by
\begin{align*}
J_p\big(d_{\cal N}^{\frac1{2p}}x d_{\cal N}^{\frac1{2p}}\big)
&=d_{\cal R}^{\frac1{2p}}x d_{\cal R}^{\frac1{2p}},
&&x\in\cal N,\\
\E_p\big(d_{\cal R}^{\frac1{2p}}x d_{\cal R}^{\frac1{2p}}\big)
&=d_{\cal N}^{\frac1{2p}}\E(x)d_{\cal N}^{\frac1{2p}},
&&x\in\cal R.
\end{align*}
These identities also hold at every matrix level, with the corresponding amplified densities. In particular, the projection $J_p\E_p$ is represented on $\L^p(\cal R,\tr_{\cal R})$ by
\begin{equation*}
\xi\mapsto d_{\cal R}^{\frac1{2p}}
\E\big(d_{\cal R}^{-\frac1{2p}}\xi d_{\cal R}^{-\frac1{2p}}\big)
d_{\cal R}^{\frac1{2p}}.
\end{equation*}
%If $\cal R$ is finite-dimensional, $\tr_{\cal R}$ is a faithful trace on $\cal R$, and
%\[
%\psi(x)=\tr_{\cal R}(d x),
%\quad x\in\cal R,
%\]
%for some positive invertible density $d\in\cal R$, then, under the symmetric realization of Kosaki's spaces, the map $J_p$ is determined on $\cal N$ by
%\begin{equation}
%\label{Kosaki-symmetric-inclusion-density}
%J_p(x)
%=
%d^{\frac1{2p}}x d^{\frac1{2p}},
%\quad x\in\cal N.
%\end{equation}
\end{lemma}

\begin{proof}
By Takesaki's theorem \cite[Theorem~IX.4.2]{Tak03}, the modular invariance assumption \eqref{modular-invariance-subalgebra} yields a normal faithful conditional expectation $\E\co\cal R\to\cal N$ such that $\psi\circ E=\psi$. At $p=\infty$, the inclusion and $\E$ are completely contractive and their composition on $\cal N$ is the identity. At $p=1$, the corresponding statement follows by passing to the preadjoint and the restriction map. Since these endpoint maps are compatible with the Kosaki interpolation couples, complex interpolation \cite{BeL76}, or equivalently \cite[Lemma~2.2]{JuX03}, gives completely contractive maps $
J_p\co\L^p(\cal N,\varphi)\to\L^p(\cal R,\psi)$, $
\E_p\co\L^p(\cal R,\psi)\to\L^p(\cal N,\varphi)$ with $\E_pJ_p=\Id_{\L^p(\cal N,\varphi)}$. Indeed, the contractive extensions are the state-preserving extensions of $E$ to noncommutative $\L^p$-spaces, and applying the same construction to $\Id_{\M_n}\ot E$ gives complete contractivity at every matrix level. Consequently, we have
\[
\norm{x}_p
=\norm{\E_pJ_p(x)}_p
\leq\norm{J_p(x)}_p
\leq\norm{x}_p,
\quad x \in \L^p(\cal N,\varphi),
\]
and the same inequalities hold at every matrix level. Thus $J_p$ is a complete isometry. Finally, the map $J_p\E_p$ is a completely contractive projection whose range is $J_p(\L^p(\cal N,\varphi))$.

The formulas in the symmetric realizations follow from the common-core construction in \cite[Section~2, Lemma~2.1 and Lemma~2.2]{JuX03}, taking the symmetric parameter $\theta = \frac{1}{2}$. Applying the same construction to the matrix amplifications gives the corresponding formulas at every matrix level.
\end{proof}

%\begin{proof}
%%By Takesaki's theorem \cite[Theorem IX.4.2]{Tak03}, the modular invariance assumption \eqref{modular-invariance-subalgebra} yields a unique normal faithful conditional expectation $\E \co \cal{R} \to \cal{N}$ such that $\psi \circ \E=\psi$. At the level $p=\infty$, the inclusion $J_\infty\co\cal N \to\cal R$ and the conditional expectation $\E \co \cal R \to \cal N$ are completely contractive and satisfy
%%\[
%%\E J_\infty=\Id_{\cal N}.
%%\]
%%At the predual level, the preadjoint $E_*\co\cal N_*\to\cal R_*$ and the restriction map $(J_\infty)_*\co\cal R_*\to\cal N_*$ are completely contractive and satisfy
%%\[
%%(J_\infty)_*\E_*=\Id_{\cal N_*}.
%%\]
%%Consequently, $\E_*$ is a complete isometry.
%%
%%Since $\varphi=\psi|_{\cal N}$ and $\psi \circ \E=\psi$, these maps are compatible with the embeddings defining the Kosaki interpolation couples $(\cal N,\cal N_*)$ and $(\cal R,\cal R_*)$. Complex interpolation therefore yields completely contractive maps
%%\[
%%J_p\co\L^p(\cal N,\varphi) \to \L^p(\cal R,\psi)
%%\quad\text{and}\quad
%%\E_p\co\L^p(\cal R,\psi) \to \L^p(\cal N,\varphi)
%%\]
%%such that $\E_pJ_p=\Id_{\L^p(\cal N,\varphi)}$. Hence, for every $x \in \L^p(\cal N,\varphi)$,
%%\[
%%\norm{x}_{\L^p(\cal N,\varphi)}
%%=
%%\norm{\E_pJ_p(x)}_{\L^p(\cal N,\varphi)}
%%\leq
%%\norm{J_p(x)}_{\L^p(\cal R,\psi)}
%%\leq
%%\norm{x}_{\L^p(\cal N,\varphi)}.
%%\]
%%Thus $J_p$ is an isometry. The same interpolation argument at every matrix level shows that $J_p$ is a complete isometry. Moreover, $J_pE_p$ is a completely contractive projection onto $J_p(\L^p(\cal N,\varphi))$.
%Suppose now that $\cal R$ is finite-dimensional and that $\psi(x)=\Tr_{\cal R}(dx)$. Under the standard symmetric realization of the Kosaki space, see \cite{Kos84}, the element $x\in\cal R$ is represented in $\L^p(\cal R,\psi)$ by
%\[
%d^{\frac1{2p}}x d^{\frac1{2p}}.
%\]
%Since the inclusion of $\cal N$ into $\cal R$ is state preserving and the preceding interpolation construction is the canonical one, the image of $x\in\cal N$ under $J_p$ is represented by the same element $d^{\frac1{2p}}x d^{\frac1{2p}}$ in $\L^p(\cal R,\psi)$. This proves \eqref{Kosaki-symmetric-inclusion-density}.
%\end{proof}



%\begin{cor}
%\label{cor-finite-dimensional-modular-deweighting}
%Assume that the von Neumann algebra $\cal{M}$ is finite-dimensional. Let $P \co \L^4(\cal M) \to \L^4(\cal{M})$ be a $2$-contractive projection. Under the assumptions of Theorem~\ref{thm-global-left-right-Levi-finite-dimensional} with and $e \ov{\mathrm{def}}{=} s_\ell(P)$ and $f \ov{\mathrm{def}}{=} s_r(P)$, define the subspace
%\begin{equation}
%\label{finite-dimensional-deweighted-space}
%\cal{V}
%\ov{\mathrm{def}}{=}
%a^{-\frac14}(\Ran P)b^{-\frac14}.
%\end{equation}
%of $e\cal{M}f$. Then $\cal{V}$ is a finite-dimensional $\TRO$ and
%\begin{equation}
%\label{finite-dimensional-weighted-range}
%\Ran P
%=a^{\frac14}\cal V b^{\frac14}.
%\end{equation}

%Thus $\Ran P$ is precisely the symmetric $\L^4$-realization of the upper-right $\TRO$ corner $\cal V$ for the linking density $\mathsf D$. \textbf{un truc comme ca doit etre vrai}: Finally, $\cal{V} \to \Ran P$, $v \mapsto a^{\frac14}\cal V b^{\frac14}$ induces an isometry if $\cal{V}$ is equipped with the norm of $\L^p(e\cal{M}f)$
%\end{cor}

\begin{cor}
\label{cor-finite-dimensional-modular-deweighting}
Let $\cal{M}$ be a finite-dimensional von Neumann algebra. Let $P \co \L^4(\cal M) \to \L^4(\cal M)$ be a $2$-contractive projection with $\Ran P\neq\{0\}$. Under the assumptions and notation of Section~\ref{sec-modular-finite}, let $e \ov{\mathrm{def}}{=} s_\ell(P)$, $f \ov{\mathrm{def}}{=} s_r(P)$ and define the subspace
\begin{equation}
\label{finite-dimensional-deweighted-space}
\cal V
\ov{\mathrm{def}}{=}
a^{-\frac14}(\Ran P)b^{-\frac14}
\end{equation}
of $e\cal M f$. Here and below, all negative powers of $a$ and $b$ are taken in the reduced algebras $e\cal Me$ and $f\cal Mf$, respectively. Then $\cal V$ is a finite-dimensional $\W^*$-$\TRO$ and
\begin{equation}
\label{finite-dimensional-weighted-range}
\Ran P=a^{\frac14}\cal Vb^{\frac14}.
\end{equation}
Consider the von Neumann algebra $
\cal R
\ov{\mathrm{def}}{=}
\begin{bmatrix}
e\cal Me&e\cal Mf\\
f\cal Me&f\cal Mf
\end{bmatrix}$ and let
$D
\ov{\mathrm{def}}{=}
\begin{bmatrix}
a^2&0\\
0&b^2
\end{bmatrix}$. Then the linking von Neumann algebra $\cal R(\cal V)$ is a unital von Neumann subalgebra of $\cal R$ and is invariant under the modular automorphism group associated with $D$:
\begin{equation}
\label{linking-algebra-V-modular-invariance}
\sigma_t^{D}(\cal R(\cal V))
=
\cal R(\cal V),
\quad t \in \R.
\end{equation}
Set
\begin{equation}
\label{normalization-linking-state}
c
\ov{\mathrm{def}}{=}
\tr_{\cal R}(D)
\quad \text{and} \quad
\psi(X)
\ov{\mathrm{def}}{=}
c^{-1}\tr_{\cal R}(D X),
\quad X \in \cal R,
\end{equation}
and let $\varphi\ov{\mathrm{def}}{=}\psi|_{\cal R(\cal V)}$. Then $\varphi$ is a normal faithful state on $\cal R(\cal V)$, and the map defined on the finite-dimensional space $\L^4(\cal{V},\varphi)=\cal V$ by
\begin{equation}
\label{complete-isometry-deweighted-space}
U(v)
\ov{\mathrm{def}}{=}
c^{-\frac14}a^{\frac14}vb^{\frac14},
\quad v \in \cal V,
\end{equation}
is a surjective complete isometry $U \co \L^4(\cal V,\varphi) \to \Ran P$. In particular, the space $\Ran P$ is completely isometric to a rectangular noncommutative $\L^4$-space.
\end{cor}

\begin{proof}
Set $Y\ov{\mathrm{def}}{=}\Ran P$. The identity \eqref{finite-dimensional-weighted-range} follows directly from the definition of $\cal V$ and from $Y=eYf$. We first prove that $\cal V$ is a $\TRO$. For $z \in \mathbb{C}$, define
\begin{equation}
\label{def-Theta-z}
\Theta^z(v)
\ov{\mathrm{def}}{=}
a^{-z}vb^z,
\quad v\in\cal V.
\end{equation}
If $v=a^{-\frac14}yb^{-\frac14}$ with $y\in Y$, then
\[
\Theta^z(v)
=\Theta^z(a^{-\frac{1}{4}}yb^{-\frac{1}{4}})
\ov{\eqref{def-Theta-z}}{=}
a^{-\frac14}a^{-z}yb^zb^{-\frac14}
\ov{\eqref{global-relative-complex-powers}}{=}
a^{-\frac14}\Delta^z(y)b^{-\frac14}.
\]
Since $\Delta^z(\Ran P)=\Ran P$ by \eqref{global-relative-complex-powers}, we obtain $\Theta^z(\cal V)=\cal V$, and $\Theta^z$ is a bijection of $\cal V$ with inverse $\Theta^{-z}$. Let $v_1,v_2,v_3 \in \cal V$ and consider the elements
\begin{equation}
\label{def-des-yi}
y_1
\ov{\mathrm{def}}{=}
a^{\frac14}v_1b^{\frac14}, \quad 
y_2
\ov{\mathrm{def}}{=} a^{\frac14}v_2b^{\frac14}
\quad \text{and} \quad 
y_3 \ov{\mathrm{def}}{=} a^{\frac14}v_3b^{\frac14}
\end{equation}
of the subspace $\Ran P$. By Theorem~\ref{thm-two-contractive-completely-contractive-L4}, the element $
r
\ov{\mathrm{def}}{=}
y_1y_2^*y_3$ belongs to the subspace $\Ran P^\dagger$. Since $\Ran P^\dagger \ov{\eqref{global-exact-density-ranges}}{=} a(\Ran P)$, there exists $x \in \Ran P$ such that 
\begin{equation}
\label{r=ax}
r 
=a x.
\end{equation}
By definition \eqref{finite-dimensional-deweighted-space} of the subspace $\cal{V}$, we have $a^{-\frac14}x b^{-\frac14} \in \cal{V}$. A direct computation gives
\begin{align*}
a^{-\frac14} x b^{-\frac14}
\ov{\eqref{r=ax}}{=} a^{-\frac54}rb^{-\frac14}
=a^{-\frac54} y_1y_2^*y_3  b^{-\frac14}
\ov{\eqref{def-des-yi}}{=} a^{-1}v_1b^{\frac12}v_2^* a^{\frac12}v_3
\ov{\eqref{def-Theta-z}}{=} \Theta^1(v_1)\Theta^{-\frac12}(v_2)^*v_3.
\end{align*}
Thus $
\Theta^1(v_1)\Theta^{-\frac12}(v_2)^*v_3
\in \cal{V}$. Since $\Theta^1$ and $\Theta^{-\frac12}$ are bijections of $\cal{V}$, their values range independently over $\cal V$. Hence
\[
u_1u_2^*u_3 \in \cal{V},
\quad
u_1,u_2,u_3 \in \cal{V}.
\]
Consequently, we have the inclusion $\cal{V} \cal{V}^*\cal{V} \subset \cal{V}$. Since $\cal V$ is finite-dimensional, it is weak* closed, and therefore it is a $\W^*$-$\TRO$. Moreover, if we consider the map $J \co \cal{M} \to\M_2(\cal{M})$, $v \mapsto
\begin{bmatrix}
0&v\\
0&0
\end{bmatrix}$, then
\begin{equation}
\label{symmetric-L4-linking-realization}
J(a^{\frac14}vb^{\frac14})
=D^{\frac18} J(v) D^{\frac18},
\quad
v \in \cal V.
\end{equation}
Indeed, we have
\begin{align*}
\MoveEqLeft
D^{\frac18}
J(v)
D^{\frac18}
=D^{\frac18}
\begin{bmatrix}
0&v\\
0&0
\end{bmatrix}
D^{\frac18}
\ov{\eqref{linking-density}}{=}
\begin{bmatrix}
a^{\frac14}&0\\
0&b^{\frac14}
\end{bmatrix}
\begin{bmatrix}
0&v\\
0&0
\end{bmatrix}
\begin{bmatrix}
a^{\frac14}&0\\
0&b^{\frac14}
\end{bmatrix} \\
&=\begin{bmatrix}
0 & a^{\frac14}v \\
0 & 0
\end{bmatrix}\begin{bmatrix}
a^{\frac14}&0\\
0&b^{\frac14}
\end{bmatrix}
=
\begin{bmatrix}
0&a^{\frac14}vb^{\frac14}\\
0&0
\end{bmatrix}
=J(a^{\frac14}vb^{\frac14}),
\end{align*}
which proves \eqref{symmetric-L4-linking-realization}. We next check that the linking algebra of $\cal V$ has the same unit as $\cal R$. Let $(x_1,\ldots,x_k)$ be the basis used in \eqref{global-Gram-densities} and put
\[
v_j
\ov{\mathrm{def}}{=}
a^{-\frac14}x_jb^{-\frac14}
\in\cal V,
\quad
j\in\{1,\ldots,k\}.
\]
Since $b$ is positive and invertible in the reduced finite-dimensional algebra $f\cal Mf$, the functional calculus gives
\begin{equation}
\label{lower-bound-b-inverse}
b^{-\frac12}
\geq
\norm{b}_{\L^\infty(\cal M)}^{-\frac12}f.
\end{equation}
Consequently, we deduce that
\begin{align}
\label{inter-765}
\sum_{j=1}^k v_jv_j^*
= a^{-\frac14}
\Bigg(\sum_{j=1}^k x_jb^{-\frac12}x_j^*\Bigg)
a^{-\frac14}
\ov{\eqref{lower-bound-b-inverse}}{\geq}
\norm{b}_{\L^\infty(\cal M)}^{-\frac12}
a^{-\frac14}
\Bigg(\sum_{j=1}^k x_jx_j^*\Bigg)
a^{-\frac14}
\ov{\eqref{global-Gram-densities}}{=}
\norm{b}_{\L^\infty(\cal M)}^{-\frac12}a^{\frac12}.
\end{align}
The operator on the left-hand side of \eqref{inter-765} belongs to $e\cal Me$, so its support is dominated by $e$. On the other hand, \eqref{inter-765} and $s(a^{\frac{1}{2}})=e$ imply that its support dominates $e$. Hence
\[
s\bigg(\sum_{j=1}^k v_jv_j^*\bigg)
= e,
\]
and therefore $s_\ell(\cal V)=e$. Similarly, we have $
a^{-\frac12}
\geq
\norm{a}_{\L^\infty(\cal M)}^{-\frac12}e$, and hence
\begin{align*}
\sum_{j=1}^k v_j^*v_j
=
b^{-\frac14}
\Bigg(\sum_{j=1}^k x_j^*a^{-\frac12}x_j\Bigg)
b^{-\frac14}
\geq
\norm{a}_{\L^\infty(\cal M)}^{-\frac12}
b^{-\frac14}
\Bigg(\sum_{j=1}^k x_j^*x_j\Bigg)
b^{-\frac14}
=
\norm{a}_{\L^\infty(\cal M)}^{-\frac12}b^{\frac12}.
\end{align*}
It follows that
\[
s\left(\sum_{j=1}^k v_j^*v_j\right)
=
f,
\]
and therefore $s_r(\cal V) = f$. Consequently, we have $1_{\cal R(\cal V)}
=
\begin{bmatrix}
e&0\\
0&f
\end{bmatrix}
=
1_{\cal R}$. Thus $\cal R(\cal V)$ is a unital von Neumann subalgebra of $\cal R$. Now, we use Proposition~\ref{prop-linking-modular-interpretation}. If $v \in \cal V$ and $t \in \R$, then
\begin{align*}
\sigma_t^{D}(J(v))
&\ov{\eqref{linking-modular-group-corner}}{=}
J(a^{2\i t}vb^{-2\i t})
 \ov{\eqref{def-Theta-z}}{=} 
J(\Theta^{-2\i t}(v)).
\end{align*}
Since $\Theta^{-2\i t}(\cal V)=\cal V$, we obtain
\[
\sigma_t^{D}(J(\cal V))=J(\cal V),
\quad t \in \R.
\]
The map $\sigma_t^{D} \co \cal R \to \cal R$ is a $*$-automorphism. It therefore also leaves $J(\cal V)^*$ invariant, as well as the von Neumann algebras generated by $J(\cal V)J(\cal V)^*$ and $J(\cal V)^*J(\cal V)$. Since $
\cal R(\cal V)
 \ov{\eqref{Linking-algebra}}{=} 
\begin{bmatrix}
\ovl{\Span\cal V\cal V^*}^{\w^*}&\cal V\\
\cal V^*&\ovl{\Span\cal V^*\cal V}^{\w^*}
\end{bmatrix}$, we conclude that $
\sigma_t^{D}(\cal R(\cal V))
=
\cal R(\cal V)$ for any $t \in \R$. This proves \eqref{linking-algebra-V-modular-invariance}. Let $
q
\ov{\mathrm{def}}{=}
\begin{bmatrix}
e&0\\
0&0
\end{bmatrix}
\in\cal R(\cal V)$. Then
\[
1_{\cal R(\cal V)}-q
=
\begin{bmatrix}
0&0\\
0&f
\end{bmatrix}
\quad \text{and} \quad
\cal V
=
q\cal R(\cal V)(1_{\cal R(\cal V)}-q).
\]
Since the density $D$ is block diagonal, we have $
qD=Dq$. Hence $q$ is fixed by $(\sigma_t^D)_{t\in\R}$. Since the modular group of $\psi$ is $(\sigma_t^D)_{t\in\R}$ and $\cal R(\cal V)$ is globally invariant under this group, Takesaki's theorem gives
\[
\sigma_t^\varphi
=
\sigma_t^\psi|_{\cal R(\cal V)}
=
\sigma_t^D|_{\cal R(\cal V)},
\quad
t \in \R.
\]
Consequently, we have
\[
\sigma_t^\varphi(q)=q,
\quad
t \in \R,
\]
so that $q$ belongs to the centralizer of $\varphi$. In particular, the symmetric Kosaki realization of $\L^p(\cal R(\cal V),\varphi)$ preserves the upper-right corner defining $\L^p(\cal V,\varphi)$. Applying Lemma~\ref{lem-modular-invariant-Kosaki-subspace}, the inclusion $\cal R(\cal V)\subset\cal R$ induces a complete isometry 
$$
J_4\co
\L^4(\cal R(\cal V),\varphi)
\to
\L^4(\cal R,\psi).
$$ 
Since $q$ belongs to the centralizer of $\varphi$ and commutes with the density $d \ov{\mathrm{def}}{=} c^{-1}D$ of $\psi$, the symmetric realizations on both sides preserve the corners determined by $q$. Hence $J_4$ restricts to a complete isometry
\[
J_4 \co \L^4(\cal{V},\varphi) \to q\L^4(\cal{R},\psi)(1-q).
\]
Recall that $d = c^{-1}D$, the density of $\psi$ with respect to $\tr_{\cal R}$. By the definition of the canonical inclusion $J_4$ in Lemma~\ref{lem-modular-invariant-Kosaki-subspace}, the image of $v\in\cal V$ is represented in the ambient symmetric realization by $d^{\frac18} J(v) d^{\frac18}$. Hence, for any $v \in \cal V$, we have
\begin{align*}
\MoveEqLeft
%\label{rectangular-L4-norm-identification}
\norm{v}_{\L^4(\cal V,\varphi)}
=\bnorm{d^{\frac18}J(v)d^{\frac18}}_{\L^4(\cal R,\tr_{\cal{R}})}
=c^{-\frac14}\bnorm{D^{\frac18}J(v)D^{\frac18}}_{\L^4(\cal{R},\tr_{\cal R})}\\
&\ov{\eqref{symmetric-L4-linking-realization}}{=}
c^{-\frac14}\bnorm{J(a^{\frac14}vb^{\frac14})}_{\L^4(\cal{R},\tr_{\cal R})}
=c^{-\frac14}\bnorm{a^{\frac14}vb^{\frac14}}_{\L^4(\cal M)}
\ov{\eqref{complete-isometry-deweighted-space}}{=}
\norm{U(v)}_{\L^4(\cal{M})}.
\end{align*}
The same calculation at the $n$th matrix level uses the density $\Id_{\M_n}\ot d$ and the map $\Id_{\M_n}\ot J$. Hence $U$ is a complete isometry. Moreover, we have
\[
U(\cal V)
\ov{\eqref{complete-isometry-deweighted-space}}{=} 
c^{-\frac14}a^{\frac14}\cal Vb^{\frac14}
\ov{\eqref{finite-dimensional-weighted-range}}{=}
c^{-\frac14}\Ran P
=\Ran P,
\]
since $\Ran P$ is a linear subspace. Hence $U$ is surjective. This completes the proof.
\end{proof}

\begin{prop}
\label{prop-explicit-weighted-TRO-expectation}
Retain the assumptions and notation of Corollary~\ref{cor-finite-dimensional-modular-deweighting}. Put $\cal N\ov{\mathrm{def}}{=}\cal R(\cal V)$, and let $\E\co\cal R\to\cal N$ be the $\psi$-preserving conditional expectation provided by Takesaki's theorem. Define $\E_{12}\co e\cal M f\to\cal V$ by
\begin{equation*}
\E\left(\begin{bmatrix}
0&z\\0&0\end{bmatrix}\right)
=\begin{bmatrix}0&\E_{12}(z)\\0&0
\end{bmatrix}.
\end{equation*}
Then the original projection satisfies
\begin{equation}
\label{explicit-original-projection-formula}
P(x)
=a^{\frac{1}{4}}\E_{12}\big(a^{-\frac{1}{4}}e x f b^{-\frac{1}{4}}\big)b^{\frac{1}{4}},
\quad x \in \L^4(\mathcal M).
\end{equation}
In particular, $P$ is contractively decomposable.
\end{prop}

\begin{proof}
The diagonal corner projections belong to $\cal{N}$. By $\cal{N}$-bimodularity, $\E$ preserves the four corners, so $\E_{12}$ is well-defined. The symmetric $\mathrm L^4$ realization of the state-preserving conditional expectation is
\begin{equation*}
\cal{T}(X)
=D^{\frac{1}{8}}\E\big(D^{-\frac{1}{8}}XD^{-\frac{1}{8}}\big)D^{\frac{1}{8}},
\quad X \in \L^4(\mathcal R,\Tr_{\mathcal R}).
\end{equation*}
Normalization of the density by the scalar $c$ cancels from this formula. The map $\mathcal T$ is completely positive and completely contractive: complete positivity follows directly from the displayed formula, and complete contractivity follows from the state-preserving $\L^4$ extension of $\E$. It is a projection because $\E^2=\E$.

Let $s=\operatorname{diag}(e,f) \in \M_2(\cal{M})$. Identifying $\cal{R}$ with $s\mathrm M_2(\cal{M})s$, extend $\mathcal T$ to $\L^4(\M_2\overline\otimes\mathcal M)$ by $\widetilde{\mathcal T}(X)=\mathcal T(sXs)$. This remains completely positive and completely contractive. Since $D$ is block diagonal and $\E$ preserves corners, the extension has block form
\begin{equation*}
\widetilde{\mathcal T}\left(\begin{bmatrix}
x_{11}&x_{12}\\
x_{21}&x_{22}
\end{bmatrix}\right)
=\begin{bmatrix}
v_1(x_{11})&Q(x_{12})\\
Q^\circ(x_{21})&v_2(x_{22})
\end{bmatrix},
\end{equation*}
where $\|v_1\|,\|v_2\|\leq1$ and $Q$ is the right-hand side of \eqref{explicit-original-projection-formula}. Thus $\|Q\|_{\dec} \leq 1$. Its range is $a^{\frac{1}{4}}\mathcal Vb^{\frac{1}{4}}=\Ran P$, and it fixes this range because $\E_{12}|_{\mathcal V}=\Id_{\mathcal V}$. Therefore $Q$ is a contractive projection onto $\Ran P$. Uniqueness of contractive projections onto a fixed nonzero subspace of the smooth space $\L^4(\mathcal M)$ gives $Q=P$.
\end{proof}


\begin{remark} \normalfont
\label{rem-Levi-modular-summary}
By Example \ref{E-finite-dim-TRO}, the $\W^*$-TRO $\cal{V}$ is completely isometric to a finite $\ell^\infty$-direct sum of rectangular matrix spaces.
\end{remark}

%\begin{remark} \normalfont
%\label{rem-Levi-modular-summary}
%The preceding results give a unified interpretation of the finite-dimensional construction. The operators $a$ and $b$ are the Gram densities of the global left and right Levi geometries, the operator $\Delta=a^{-1}(\cdot)b$ is their relative operator, and the same operator is the inverse square root of the standard modular operator of the diagonal linking density $\mathsf D=\operatorname{diag}(a^2,b^2)$. Finally, the factors $a^{\frac{1}{4}}$ and $b^{\frac{1}{4}}$ arise because $\mathsf D^{\frac{1}{8}}$ is the symmetric density used to realize the upper-right corner in $\L^4(\cal R)$.
%\end{remark}

The preceding explicit representation immediately yields the following consequence.

\begin{cor}
\label{cor-contractively-decomposable}
Assume that the von Neumann algebra $\cal{M}$ is finite-dimensional. A $2$-contractive projection $P \co \L^4(\cal M) \to \L^4(\cal{M})$ is contractively decomposable.
\end{cor}

%\begin{proof}
%If $\Ran P=\{0\}$, then $P=0$ and the assertion is immediate. Suppose that $\Ran P \neq \{0\}$. By Corollary~\ref{cor-finite-dimensional-modular-deweighting}, there exist a finite-dimensional $\W^*$-$\TRO$ $\cal V$, a normal faithful state $\varphi$ on its linking von Neumann algebra $\cal R(\cal V)$, and a surjective complete isometry $
%U \co \L^4(\cal V,\varphi) \to \Ran P$. Since $\cal M$ is finite-dimensional, it is $\sigma$-finite and has separable predual. Hence \cite[Theorem~4.1]{Arh26c} yields a contractively decomposable projection $
%Q \co \L^4(\cal M) \to \L^4(\cal M)$ such that $\Ran Q=\Ran P$.
%
%In particular, $Q$ is contractive. The original projection $P$ is also contractive since it is $2$-contractive. Since $\Ran P\neq\{0\}$, both $P$ and $Q$ have norm one. Moreover, the noncommutative space $\L^4(\cal M)$ is smooth. Proposition~\ref{prop-Sullivan} therefore implies that there exists at most one norm-one projection onto the subspace $\Ran P$. Consequently, we have $P=Q$. Since $Q$ is contractively decomposable, the same is true for $P$.
%\end{proof}

\begin{proof}
If $\Ran P=\{0\}$, then $P=0$ and the assertion is immediate. Otherwise, the result follows from Proposition~\ref{prop-explicit-weighted-TRO-expectation}.
\end{proof}


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Finite matrix-level criteria at even exponents}
\label{sec-even-exponents}

The polynomial argument used for the exponent $4$ extends to every even integral exponent. The matrix level obtained below is sufficient. We do not claim that it is optimal.

\begin{thm}
\label{thm-even-exponent-finite-level}
Let $m$ be an integer with $m \geq 2$. Set $p \ov{\mathrm{def}}{=}2m$ and $N_m \ov{\mathrm{def}}{=}
\left\lceil\frac{2m}{3}\right\rceil$. Let $P \co \L^p(\cal M) \to \L^p(\cal M)$ be a bounded projection. We denote by $P^\dagger$ its adjoint under the sesquilinear trace duality and set $
Y\ov{\mathrm{def}}{=}\Ran P$ and $Y^\sharp\ov{\mathrm{def}}{=}\Ran P^\dagger$. The following assertions are equivalent.
\begin{enumerate}
\item The projection $P$ is $N_m$-contractive.
\item For any $y_1,\ldots,y_{2m-1}\in Y$, we have
\begin{equation}
\label{alternating-word-dual-range-even}
y_1y_2^*y_3\cdots y_{2m-2}^*y_{2m-1}
\in Y^\sharp.
\end{equation}
\item The projection $P$ is completely contractive.
\end{enumerate}
If these conditions hold, then
\begin{equation}
\label{exact-alternating-word-dual-range-even}
\begin{split}
Y^\sharp
&=
\left\{
y_1y_2^*y_3\cdots y_{2m-2}^*y_{2m-1}
:
y_1,\ldots,y_{2m-1}\in Y
\right\}
=
\left\{
y(y^*y)^{m-1}:y\in Y
\right\}.
\end{split}
\end{equation}
Moreover, for any integer $n \geq 1$, we have
\begin{equation}
\label{matrix-duality-map-equality-even}
\Ran P_n^\dagger
=
\left\{
X(X^*X)^{m-1}:X\in\M_n(Y)
\right\}
=
\M_n(Y^\sharp),
\end{equation}
where $P_n\ov{\mathrm{def}}{=}\Id_{S_n^p}\ot P$.
\end{thm}

\begin{proof}
The implication $3\Rightarrow1$ is immediate. We first prove $1\Rightarrow2$. Put $
r\ov{\mathrm{def}}{=}2m-1$. Suppose that $P_{N_m}$ is contractive. At matrix level $N_m$, the power duality mapping is
\[
\frak j_p(X)
=
X(X^*X)^{m-1},
\quad
X \in \L^p(\M_{N_m} \otvn \cal M).
\]
Moreover, we have $
P_{N_m}^\dagger
=
\Id_{S_{N_m}^{p^*}}\ot P^\dagger$ and $\Ran P_{N_m}^\dagger
=
\M_{N_m}(Y^\sharp)$. Consequently, Lemma~\ref{Lemma-smooth} gives
\begin{equation}
\label{matrix-duality-inclusion-even}
X(X^*X)^{m-1}
\in
\M_{N_m}(Y^\sharp),
\quad
X \in \M_{N_m}(Y).
\end{equation}
Fix $y_1,\ldots,y_r \in Y$. For $0 \leq k \leq r$, set $
v_k
\ov{\mathrm{def}}{=}
1+\left\lfloor\frac{k}{3}\right\rfloor$. Then $v_0=1$ and $v_r=N_m$. Let $(E_{ij})_{1\leq i,j\leq N_m}$ be the canonical matrix units of $\M_{N_m}$ and put
\[
F_k
\ov{\mathrm{def}}{=}
E_{v_{k-1},v_k},
\quad
A_k
\ov{\mathrm{def}}{=}
\begin{cases}
F_k\ot y_k,&k\text{ odd},\\
F_k^*\ot y_k,&k\text{ even},
\end{cases}
\quad 1\leq k\leq r.
\]
Thus $A_k \in \M_{N_m}(Y)$ for any integer $k$. For any $\zeta=(\zeta_1,\ldots,\zeta_r) \in \T^r$, define
\[
X(\zeta)
\ov{\mathrm{def}}{=}
\sum_{k=1}^r\zeta_kA_k.
\]
By \eqref{matrix-duality-inclusion-even}, we have $
\frak j_p(X(\zeta))
\in
\M_{N_m}(Y^\sharp)$ for any $\zeta \in \T^r$. Let $\mu$ be the normalized Haar measure on $\T^r$. Consider the Fourier coefficient
\begin{equation}
\label{Fourier-coefficient-even}
C
\ov{\mathrm{def}}{=}
\int_{\T^r}
\left(
\prod_{\substack{1\leq k\leq r\\ k\text{ odd}}}
\overline{\zeta_k}
\right)
\left(
\prod_{\substack{1\leq k\leq r\\ k\text{ even}}}
\zeta_k
\right)
\frak j_p(X(\zeta))
\d\mu(\zeta).
\end{equation}
Repeated applications of H\"older's inequality show that the map $\zeta \mapsto \frak j_p(X(\zeta))$ is continuous from $\T^r$ into the space $\L^{p^*}(\M_{N_m}\otvn\cal M)$. Hence the integral in \eqref{Fourier-coefficient-even} is a Bochner integral. Since $\M_{N_m}(Y^\sharp)$ is a closed linear subspace, we obtain
\begin{equation}
\label{Fourier-coefficient-in-dual-range}
C\in\M_{N_m}(Y^\sharp).
\end{equation}

We determine the $(1,N_m)$-entry of $C$. Expanding the duality mapping gives
\[
\frak j_p(X(\zeta))
=
\sum_{k_1,\ldots,k_r=1}^r
\zeta_{k_1}\overline{\zeta_{k_2}}
\zeta_{k_3}\cdots
\overline{\zeta_{k_{r-1}}}\zeta_{k_r}
A_{k_1}A_{k_2}^*A_{k_3}\cdots A_{k_{r-1}}^*A_{k_r}.
\]
For a fixed term, let $a_k$ be the number of occurrences of the label $k$ in an unstarred position and let $b_k$ be its number of occurrences in a starred position. Orthogonality of the characters of $\T^r$ shows that this term survives in \eqref{Fourier-coefficient-even} only if
\begin{equation}
\label{Fourier-exponent-conditions-even}
a_k-b_k
=
\begin{cases}
1,&k\text{ odd},\\
-1,&k\text{ even}.
\end{cases}
\end{equation}
There are $m$ unstarred positions and $m-1$ starred positions. Since there are $m$ odd labels, \eqref{Fourier-exponent-conditions-even} gives
\[
m
=
\sum_{k=1}^r a_k
\geq
\sum_{\substack{1\leq k\leq r\\k\text{ odd}}}a_k
\geq m.
\]
Consequently, equality holds throughout. We obtain
\[
a_k=1,\quad b_k=0
\quad(k\text{ odd}),
\quad
a_k=0,\quad b_k=1
\quad(k\text{ even}).
\]
Thus the surviving terms are indexed by permutations $\pi$ of $\{1,\ldots,r\}$ satisfying
\[
\pi(j)\equiv j\pmod 2,
\quad 1\leq j\leq r.
\]
For such a permutation, the corresponding product is
\[
F_{\pi(1)}F_{\pi(2)} \cdots F_{\pi(r)}
\ot
y_{\pi(1)}y_{\pi(2)}^* \cdots y_{\pi(r-1)}^*y_{\pi(r)}.
\]
We claim that the identity permutation is the only such permutation for which the matrix product has a nonzero $(1,N_m)$-entry. Indeed, the matrix unit $F_k$ is the directed edge
\[
v_{k-1} \to v_k.
\]
For any vertex $1 \leq s<N_m$, the effective edge list contains, in this order, two loops at $s$ and the forward edge from $s$ to $s+1$. The remaining zero, one or two edges are loops at $N_m$. A composable path from $1$ to $N_m$ which uses every edge must use the forward edges in their natural order, and every loop at a vertex must occur after arrival at that vertex and before departure from it. Hence the order is forced, except possibly for interchanging the two loops at a fixed vertex. The labels of these two loops are consecutive and therefore have opposite parity. Such an interchange is excluded by $\pi(j) \equiv j\pmod2$. The same argument applies to the two possible terminal loops. This proves the claim.

Consequently, we have
\[
C_{1,N_m}
=
y_1y_2^*y_3\cdots y_{r-1}^*y_r.
\]
It follows from \eqref{Fourier-coefficient-in-dual-range} that this alternating word belongs to $Y^\sharp$. This proves \eqref{alternating-word-dual-range-even}.

We next prove $2\Rightarrow3$. Let $n\geq1$ and let $X=[x_{ij}]\in\M_n(Y)$. For $1\leq i_0,i_{2m-1}\leq n$, we have
\begin{align}
\label{entries-even-duality-map}
\MoveEqLeft
\bigl[X(X^*X)^{m-1}\bigr]_{i_0,i_{2m-1}}\\
&=
\sum_{i_1,\ldots,i_{2m-2}=1}^n
\left[
\prod_{\ell=0}^{m-2}
\left(
x_{i_{2\ell},i_{2\ell+1}}
x_{i_{2\ell+2},i_{2\ell+1}}^*
\right)
\right]
x_{i_{2m-2},i_{2m-1}}. \nonumber
\end{align}
Every summand on the right-hand side is an alternating product of $2m-1$ elements of $Y$. Hence the second assertion gives
\[
\frak j_p(X)
=
X(X^*X)^{m-1}
\in
\M_n(Y^\sharp)
=
\Ran P_n^\dagger.
\]

Let $W\in\L^p(\M_n\otvn\cal M)$ and put $X\ov{\mathrm{def}}{=}P_n(W)$. Since $
W-X\in\ker P_n$ and $
\frak j_p(X)\in\Ran P_n^\dagger=(\ker P_n)^\perp$, 
we have
\[
(\Tr_n\ot\tr)\bigl(\frak j_p(X)^*(W-X)\bigr)=0.
\]
Consequently, we have
\begin{align*}
\MoveEqLeft
\norm{X}_p^p
=(\Tr_n\ot\tr)\bigl(\frak j_p(X)^*X\bigr)
=(\Tr_n\ot\tr)\bigl(\frak j_p(X)^*W\bigr)
\leq
\norm{\frak j_p(X)}_{p^*}\norm{W}_p
=
\norm{X}_p^{p-1}\norm{W}_p.
\end{align*}
If $X\neq0$, division by $\norm{X}_p^{p-1}$ gives
\[
\norm{P_n(W)}_p\leq\norm{W}_p.
\]
The same conclusion is immediate when $X=0$. Hence $P_n$ is contractive. Since $n\geq1$ was arbitrary, $P$ is completely contractive.

It remains to prove the exact range descriptions. Applying \eqref{power-duality-map-range-equality}, under the trace-duality identification, to each contractive amplification $P_n$, we obtain
\[
\Ran P_n^\dagger
=
\frak j_p(\Ran P_n)
=
\left\{
X(X^*X)^{m-1}:X\in\M_n(Y)
\right\}.
\]
Together with $\Ran P_n^\dagger=\M_n(Y^\sharp)$, this proves \eqref{matrix-duality-map-equality-even}. Taking $n=1$ gives
\[
Y^\sharp
=
\left\{
y(y^*y)^{m-1} : y \in \Ran P
\right\}.
\]
Each element on the right-hand side is an alternating product of $2m-1$ elements of $\Ran P$, whereas the second assertion gives the reverse inclusion for arbitrary alternating products. This proves \eqref{exact-alternating-word-dual-range-even}.
\end{proof}

The preceding theorem has an immediate dual counterpart.

\begin{cor}
\label{cor-conjugate-even-exponent-finite-level}
Let $m$ be an integer with $m \geq 2$. Set $p \ov{\mathrm{def}}{=}2m$, $q\ov{\mathrm{def}}{=}p^*
=\frac{2m}{2m-1}$, $N_m\ov{\mathrm{def}}{=} \left\lceil\frac{2m}{3}\right\rceil$. Let $Q\co\L^q(\cal M) \to \L^q(\cal M)$ be a bounded projection and put $Z\ov{\mathrm{def}}{=}\Ran Q$. The following assertions are equivalent.
\begin{enumerate}
\item The projection $Q$ is $N_m$-contractive.
\item The projection $Q$ is completely contractive.
\item The projection $Q$ is contractive and, for any $z_1,\ldots,z_{2m-1} \in Z$, we have
\begin{equation}
\label{dual-alternating-word-conjugate-even}
\frak j_q(z_1)
\frak j_q(z_2)^*
\frak j_q(z_3)\cdots
\frak j_q(z_{2m-2})^*
\frak j_q(z_{2m-1})
\in Z.
\end{equation}
\end{enumerate}
If these conditions hold, then
\begin{equation}
\label{exact-dual-alternating-word-conjugate-even}
Z
=\big\{
\frak j_q(z_1)
\frak j_q(z_2)^*\cdots
\frak j_q(z_{2m-2})^*
\frak j_q(z_{2m-1})
:z_1,\ldots,z_{2m-1}\in Z
\big\}.
\end{equation}
Moreover, for any integer $n \geq 1$, we have
\begin{equation}
\label{matrix-duality-map-equality-conjugate-even}
\Ran Q_n^\dagger
=
\frak j_{q,\L^q(\M_n\otvn\cal M)}\bigl(\M_n(Z)\bigr)
=
\M_n(\Ran Q^\dagger),
\end{equation}
where $Q_n\ov{\mathrm{def}}{=}\Id_{S_n^q}\ot Q$.
\end{cor}

\begin{proof}
Set $
P\ov{\mathrm{def}}{=}Q^\dagger
\co
\L^p(\cal M)\to\L^p(\cal M)$. By reflexivity, $P^\dagger=Q$. Moreover, under the matrix trace duality,
\[
(Q_n)^\dagger=P_n,
\quad n \geq 1.
\]
Consequently, we have
\begin{equation}
\label{adjoint-amplification-norm-equality-even}
\norm{Q_n}
=
\norm{(Q_n)^\dagger}
=
\norm{P_n},
\quad n \geq 1.
\end{equation}
It follows from Theorem~\ref{thm-even-exponent-finite-level} that $Q$ is $N_m$-contractive if and only if $P$ is $N_m$-contractive, if and only if $P$ is completely contractive, if and only if $Q$ is completely contractive. This proves the equivalence of the first two assertions.

Suppose that $Q$ is $N_m$-contractive. Then $Q$ is contractive. Equation \eqref{power-duality-map-range-equality}, under the trace-duality identification, gives
\[
\frak j_q(Z)
=
\Ran Q^\dagger
=
\Ran P,
\quad
\Ran P^\dagger
=
\Ran Q
=Z.
\]
Applying \eqref{alternating-word-dual-range-even} to $P$ gives \eqref{dual-alternating-word-conjugate-even}.

Conversely, suppose that $Q$ is contractive and that \eqref{dual-alternating-word-conjugate-even} holds. Again, we have
\[
\Ran P
=
\Ran Q^\dagger
=
\frak j_q(Z),
\quad
\Ran P^\dagger
=Z.
\]
Thus \eqref{dual-alternating-word-conjugate-even} is precisely the second assertion of Theorem~\ref{thm-even-exponent-finite-level} for $P$. Hence $P$ is completely contractive. Equation \eqref{adjoint-amplification-norm-equality-even} then shows that $Q$ is completely contractive.

Finally, \eqref{exact-dual-alternating-word-conjugate-even} follows from \eqref{exact-alternating-word-dual-range-even} applied to $P$. Since any amplification $Q_n$ is contractive, applying \eqref{power-duality-map-range-equality} to $Q_n$ gives
\[
\frak j_{q,\L^q(\M_n\otvn\cal M)}\bigl(\M_n(Z)\bigr)
=
\Ran Q_n^\dagger
=
\M_n(\Ran Q^\dagger),
\]
which proves \eqref{matrix-duality-map-equality-conjugate-even}.
\end{proof}

The preceding finite-level equivalences are not restricted to the semifinite tracial setting. The same argument extends to Haagerup noncommutative $\L^p$-spaces associated with arbitrary von Neumann algebras.

\begin{thm}
\label{thm-Haagerup-even-finite-level}
Let $\cal{M}$ be an arbitrary von Neumann algebra, and let $\L^p(\mathcal M)$ denote its Haagerup noncommutative $\L^p$-space. Let $m \in \mathbb N$ with $m\geq2$, and put $p=2m$ and $N_m=\lceil \frac{2m}{3}\rceil$. Let $P \co \L^p(\cal{M}) \to \L^p(\mathcal M)$ be a bounded projection, and write $Y=\Ran P$ and $Y^\sharp = \Ran P^\dagger$. The following conditions are equivalent: 

\begin{enumerate}
\item $P$ is $N_m$-contractive,
	
\item any alternating product $y_1y_2^*y_3\cdots y_{2m-2}^*y_{2m-1}$ with all $y_j \in Y$, belongs to $Y^\sharp$,

\item $P$ is completely contractive.
\end{enumerate}
The corresponding equivalence between $N_m$-contractivity and complete contractivity also holds at the conjugate exponent $p^*=\frac{2m}{2m-1}$.
\end{thm}

\begin{proof}
In this proof, $\Tr$ denotes the canonical trace functional on Haagerup $\L^1$-spaces, including their matrix amplifications, rather than the semifinite trace of the crossed product. We use the standard properties of Haagerup noncommutative $\L^p$-spaces. For $1<p<\infty$, these spaces are smooth and strictly convex, the H\"older product $
\L^{p_1}(\cal M)\times\cdots\times\L^{p_k}(\cal M)
\longrightarrow
\L^r(\cal M)$ where $
\frac1r=\sum_{j=1}^k\frac1{p_j}$, is continuous whenever the exponents are admissible, and the canonical trace satisfies the usual cyclicity relation for H\"older products. In particular, an alternating product of $2m-1$ elements of $\L^{2m}(\cal M)$ belongs to $\L^{\frac{2m}{2m-1}}(\cal M)$. If $x=u|x|$, the power duality mapping for the sesquilinear trace pairing is represented by
\[
\frak j_p(x)=u|x|^{p-1}.
\]
For $p=2m$, this becomes $
\frak j_{2m}(x)
=
x(x^*x)^{m-1}$. Finally, we have $
\L^p(\M_n\otvn\cal M)
=
S_n^p[\L^p(\cal M)]$ completely isometrically, and this identification is compatible with the H\"older products above and with trace-adjoint duality.

If $P_{N_m}$ is contractive, Lemma~\ref{Lemma-smooth} yields $\mathfrak j_p(X) \in \mathrm M_{N_m}(Y^\sharp)$ for any $X \in \M_{N_m}(Y)$. The torus Fourier coefficient and matrix-unit construction in the proof of Theorem~\ref{thm-even-exponent-finite-level} use only this inclusion, continuity of the polynomial product, and multiplication of matrix units. They therefore apply without change and show that any alternating word of length $2m-1$ in $Y$ belongs to $Y^\sharp$.

Conversely, suppose that all such words belong to $Y^\sharp$. Expanding each entry of $X(X^*X)^{m-1}$ gives $\mathfrak j_p(X) \in \M_n(Y^\sharp)=\Ran P_n^\dagger$ for any $X \in \M_n(Y)$. If $W \in \L^p(\mathrm M_n\overline\ot \mathcal M)$ and $X=P_n(W)$, trace-adjoint duality and H\"older's inequality give
\begin{equation*}
\|X\|_p^p
=\Tr(\mathfrak j_p(X)^*W)
\leq\|X\|_p^{p-1}\|W\|_p.
\end{equation*}
Thus every $P_n$ is contractive. The reverse implication from complete contractivity is immediate. Finally, $(P_n)^\dagger=(P^\dagger)_n$ and $\|P_n\|=\|(P^\dagger)_n\|$ give the assertion at the conjugate exponent.
\end{proof}


\begin{remark} \normalfont
For Schatten spaces, \cite[Theorem~1.1]{LRR09}, combined with uniqueness of norm-one projections in smooth Banach spaces, implies that $2$-contractivity of a projection is equivalent to complete contractivity for any $1<p<\infty$, $p\neq2$. Theorem~\ref{thm-Haagerup-even-finite-level} shows that, on the Haagerup $\L^{2m}$-space of an arbitrary von Neumann algebra, the finite level $
N_m=\left\lceil\frac{2m}{3}\right\rceil$ is sufficient for complete contractivity. We do not claim that this level is optimal. In particular, $N_m=2$ for $p=4$ and $p=6$.
\end{remark}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Weighted tangent geometry for $p>2$}
\label{sec-weighted-tangent-general-p}

Throughout this section, let $\cal M$ be a semifinite von Neumann algebra equipped with a normal semifinite faithful trace $\Tr$. Fix $2<p<\infty$, and put $p^*=\frac{p}{p-1}$, $r=\frac{p}{p-2}$ and $\alpha = \frac{p-2}{2}$. Thus $\frac{1}{p^*}=\frac{1}{p}+\frac{1}{r}$ and $\frac{1}{2}=\frac{1}{p}+\frac{1}{2r}$. Let $P \co \L^p(\cal M) \to \L^p(\cal M)$ be a $2$-contractive projection. Its trace adjoint $P^\dagger\co\L^{p^*}(\cal M) \to \L^{p^*}(\cal M)$ is determined by
\begin{equation}
\label{gp-trace-adjoint}
\Tr\big((P^\dagger\eta)^*x\big)
=\Tr\big(\eta^*P(x)\big),
\quad x \in \L^p(\cal M), \eta \in \L^{p^*}(\cal M).
\end{equation}
We write $Y=\Ran P$ and $Y^\sharp=\Ran P^\dagger$. No positivity assumption on $P$ is imposed.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Separate weighted tangent estimates}

\begin{prop}
\label{prop-gp-weighted-tangent-estimates}
For $u \in \Ran P$, put $a_u=(uu^*)^\alpha$ and $b_u=(u^*u)^\alpha$. Then $a_u,b_u \in \L^r(\cal M)_+$, and for any $z \in \L^p(\cal M)$, we have
\begin{equation}
\label{gp-separate-weighted-estimates}
\begin{split}
\big\|(uu^*)^{\frac{p-2}{4}}P(z)\big\|_2
&\leq\big\|(uu^*)^{\frac{p-2}{4}}z\big\|_2,\\
\big\|P(z)(u^*u)^{\frac{p-2}{4}}\big\|_2
&\leq\big\|z(u^*u)^{\frac{p-2}{4}}\big\|_2.
\end{split}
\end{equation}
Moreover, for any $x \in \L^p(\cal M)$, the following intertwining identities hold :
\begin{equation}
\label{gp-weighted-intertwining}
P^\dagger(a_ux)=a_uP(x),
\quad
P^\dagger(xb_u)=P(x)b_u.
\end{equation}
In particular, we have $a_uY\subset Y^\sharp$ and $Yb_u\subset Y^\sharp$.
\end{prop}

\begin{proof}
The identities $2\alpha r=p$ and $\|u^*\|_p=\|u\|_p$ give
\begin{equation*}
\|a_u\|_r=\|b_u\|_r=\|u\|_p^{p-2}.
\end{equation*}
H\"older's inequality therefore shows that $a_u^{1/2}z,zb_u^{1/2}\in\L^2(\cal M)$ and $a_uz,zb_u\in\L^{p^*}(\cal M)$ whenever $z\in\L^p(\cal M)$.

Put $q=p/2>1$. For $H,K\in\L^q(\cal M)_+$, the norm-derivative formula \eqref{derivation-duality-mapping-2}, together with the represented duality mapping $\frak j_q(H)=H^{q-1}$, yields
\begin{equation}
\label{gp-trace-power-derivative}
\lim_{s\downarrow0}\frac{\Tr((H+sK)^q)-\Tr(H^q)}{s}
=q\Tr(H^{q-1}K).
\end{equation}
When $H=0$, this follows directly from $q>1$. Thus \eqref{gp-trace-power-derivative} does not require $H$ to be invertible. Only first-order differentiability on $\L^q(\cal M)$ is used.

Fix $u\in Y$ and $z\in\L^p(\cal M)$. For $s\geq0$, consider
\begin{equation*}
R_s=\begin{bmatrix}u&s^{1/2}z\\0&0\end{bmatrix}
\in\L^p(\M_2\overline\otimes\cal M).
\end{equation*}
Since $P(u)=u$ and $P_2$ is contractive, the identity
\begin{equation*}
R_sR_s^*=\begin{bmatrix}uu^*+szz^*&0\\0&0\end{bmatrix}
\end{equation*}
gives
\begin{equation*}
\Tr\big((uu^*+sP(z)P(z)^*)^q\big)
\leq\Tr\big((uu^*+szz^*)^q\big).
\end{equation*}
Both sides agree at $s=0$. Subtracting their common value, dividing by $s>0$, and applying \eqref{gp-trace-power-derivative}, we obtain
\begin{equation*}
\Tr\big(a_uP(z)P(z)^*\big)\leq\Tr(a_uzz^*).
\end{equation*}
By H\"older's inequality and trace cyclicity, these terms are $\|a_u^{1/2}P(z)\|_2^2$ and $\|a_u^{1/2}z\|_2^2$, respectively. This proves the first estimate.

Using instead $C_s=\begin{bmatrix}u&0\\s^{1/2}z&0\end{bmatrix}$ and computing $C_s^*C_s$, we obtain
\begin{equation*}
\Tr\big((u^*u+sP(z)^*P(z))^q\big)
\leq\Tr\big((u^*u+sz^*z)^q\big).
\end{equation*}
The same differentiation gives $\Tr(b_uP(z)^*P(z))\leq\Tr(b_uz^*z)$, which is the second estimate in \eqref{gp-separate-weighted-estimates}.

Define continuous positive semidefinite Hermitian forms on $\L^p(\cal M)$ by
\begin{equation}
\label{gp-local-weighted-forms}
h_u^\ell(x,y)=\Tr(y^*a_ux),
\qquad
h_u^r(x,y)=\Tr(y^*xb_u).
\end{equation}
They are linear in the first variable. By \eqref{gp-separate-weighted-estimates}, $P$ is contractive for each associated seminorm. After quotienting by its null space and completing, the induced projection is therefore orthogonal. Consequently,
\begin{equation*}
h_u^\ell(Px,y)=h_u^\ell(x,Py),
\qquad
h_u^r(Px,y)=h_u^r(x,Py),
\quad x,y \in \L^p(\cal M).
\end{equation*}
Using \eqref{gp-trace-adjoint} and the nondegeneracy of trace duality gives \eqref{gp-weighted-intertwining}. Taking $x\in Y$ proves the two range inclusions.
\end{proof}

\subsection{Gram densities for finite-rank projections}

\begin{prop}
\label{prop-gp-Gram-densities}
Suppose that $P$ has finite rank and $Y\neq\{0\}$. Choose a basis $(x_1,\ldots,x_d)$ of $Y$, and put
\begin{equation}
\label{gp-Gram-densities}
A_p
=\sum_{j=1}^d(x_jx_j^*)^{\frac{p-2}{2}},
\qquad
B_p
=\sum_{j=1}^d(x_j^*x_j)^{\frac{p-2}{2}}.
\end{equation}
Let $e=s_\ell(Y)$ and $f=s_r(Y)$. Then $A_p,B_p\in\L^r(\cal M)_+$, $s(A_p)=e$, $s(B_p)=f$, and
\begin{equation}
\label{gp-exact-density-ranges}
A_pY
=Y^\sharp
=YB_p.
\end{equation}
The multiplication maps $L_{A_p}\co x\mapsto A_px$ and $R_{B_p}\co x\mapsto xB_p$ are bounded from $\L^p(\cal M)$ into $\L^{p^*}(\cal M)$ and satisfy
\begin{equation}
\label{gp-global-intertwining}
P^\dagger L_{A_p}=L_{A_p}P,
\qquad
P^\dagger R_{B_p}=R_{B_p}P.
\end{equation}
Their restrictions to $Y$ are linear bijections onto $Y^\sharp$.
\end{prop}

\begin{proof}
Membership in $\L^r(\cal M)$ and the boundedness of the multiplication maps follow from Proposition~\ref{prop-gp-weighted-tangent-estimates} and H\"older's inequality. Taking a strictly positive power preserves supports, and the support of a finite sum of positive measurable operators is the join of their supports. Hence
\begin{equation*}
s(A_p)=\bigvee_{j=1}^d s_\ell(x_j)
=e,
\qquad
s(B_p)=\bigvee_{j=1}^d s_r(x_j)
=f.
\end{equation*}
Summing \eqref{gp-weighted-intertwining} over the basis gives \eqref{gp-global-intertwining}, and therefore $A_pY\subset Y^\sharp$ and $YB_p\subset Y^\sharp$.

We verify injectivity without using bounded inverses of the densities. If $A_py=0$ for $y\in Y$, put $e_n=\mathbf1_{[1/n,n]}(A_p)$ and $g_n(t)=t^{-1}\mathbf1_{[1/n,n]}(t)$, with $g_n(0)=0$. Then $g_n(A_p)\in\cal M$ and
\begin{equation*}
e_ny=g_n(A_p)(A_py)=0.
\end{equation*}
Since $e_n\uparrow e$ strongly, $e_ny\to ey=y$ in $\L^p(\cal M)$, so $y=0$. Spectral truncation of $B_p$ gives the analogous conclusion for right multiplication. Finally, $\dim Y^\sharp=\dim Y=d$. The two injective maps from $Y$ into $Y^\sharp$ are therefore surjective, proving \eqref{gp-exact-density-ranges}.
\end{proof}

\subsection{Relative operators and modular realization}

\begin{thm}
\label{thm-gp-relative-modular-structure}
Retain the assumptions and notation of Proposition~\ref{prop-gp-Gram-densities}. The formulas
\begin{equation}
\label{gp-global-Hilbert-forms}
H_p^\ell(x,y)=\Tr(y^*A_px),
\qquad
H_p^r(x,y)=\Tr(y^*xB_p),
\quad x,y\in Y,
\end{equation}
define positive definite Hermitian forms on $Y$. Define
\begin{equation}
\label{gp-relative-operator}
\Delta_p=(L_{A_p}|_Y)^{-1}(R_{B_p}|_Y),
\end{equation}
where both restricted multiplication maps have codomain $Y^\sharp$. Then $\Delta_p$ is a positive invertible selfadjoint operator on $(Y,H_p^\ell)$ and
\begin{equation}
\label{gp-relative-identities}
A_p\Delta_p(y)=yB_p,
\qquad
H_p^r(x,y)=H_p^\ell(x,\Delta_py),
\quad x,y\in Y.
\end{equation}
Its imaginary powers satisfy
\begin{equation}
\label{gp-imaginary-powers}
\Delta_p^{\i t}(y)=A_p^{-\i t}yB_p^{\i t},
\quad y \in Y, t \in \R.
\end{equation}
Here imaginary powers of $A_p$ and $B_p$ are taken on their supports, and are unitaries of $e\cal Me$ and $f\cal Mf$, respectively. In particular, these maps form a one-parameter group of surjective complete isometries of $Y$ for its inherited $\L^p$ operator space structure. They are also unitary for both forms in \eqref{gp-global-Hilbert-forms}.

Consider the corner von Neumann algebra
\begin{equation}
\label{gp-linking-density}
\cal R=
\begin{bmatrix}e\cal Me&e\cal Mf\\f\cal Me&f\cal Mf\end{bmatrix},
\qquad
D_p
= \begin{bmatrix}
A_p^r&0\\
0&B_p^r
\end{bmatrix}.
\end{equation}
Equip $\cal R$ with the restriction $\Tr_{\cal R}$ of $\Tr_2\ot\Tr$. Then $D_p\in\L^1(\cal R)_+$ has support $1_{\cal R}$. Consequently,
\begin{equation}
\label{gp-linking-state}
c_p
=\Tr_{\cal R}(D_p),
\qquad
\psi_p(X)
=c_p^{-1}\Tr_{\cal R}(D_pX),
\quad X \in \cal R,
\end{equation}
define a finite positive constant $c_p$ and a normal faithful state $\psi_p$.

For $y\in e\L^p(\cal M)f$, write $\jmath(y)=\begin{bmatrix}0&y\\
0&0\end{bmatrix}$. Let $\sigma_{t,p}^{\psi_p}$ denote the canonical extension of the modular automorphism $\sigma_t^{\psi_p}$ to the tracial space $\L^p(\cal R)$. Then
\begin{equation}
\label{gp-relative-group-modular}
\jmath(\Delta_p^{\i t}y)
=\sigma_{-\frac{t}{r},p}^{\psi_p}(\jmath(y)),
\quad y \in Y, t\in \R.
\end{equation}
Thus the time rescaling is $-\frac{t}{r}=-(p-2)\frac{t}{p}$.

Finally, let $\Delta_{\psi_p}$ be the standard modular operator on $\L^2(\cal R,\Tr_{\cal R})$ associated with the vector $c_p^{-\frac{1}{2}}D_p^{\frac{1}{2}}$. The map
\begin{equation}
\label{gp-Hilbert-embedding}
W_p \co (Y,H_p^\ell) \to \L^2(\cal R,\Tr_{\cal R}),
\qquad
W_p(y)
=\jmath(A_p^{1/2}y),
\end{equation}
is a linear isometry. Its finite-dimensional range $\cal K_p=W_p(Y)$ reduces $\Delta_{\psi_p}$, and
\begin{equation}
\label{gp-standard-modular-identification}
W_p\Delta_pW_p^{-1}
=\big(\Delta_{\psi_p}|_{\cal K_p}\big)^{-\frac{1}{r}}.
\end{equation}
No finite-dimensionality assumption on the ambient algebra $\cal M$ is needed.
\end{thm}

\begin{proof}
H\"older's inequality makes both forms in \eqref{gp-global-Hilbert-forms} well-defined, and
\begin{equation*}
H_p^\ell(y,y)=\|A_p^{\frac{1}{2}}y\|_2^2,
\qquad
H_p^r(y,y)=\|yB_p^{\frac{1}{2}}\|_2^2.
\end{equation*}
Since $Y\subset e\L^p(\cal M)f$, $s(A_p)=e$ and $s(B_p)=f$, spectral truncation as in the preceding proof shows that these quantities vanish only when $y=0$. Thus both forms are positive definite.

By its definition, $\Delta_p$ is a linear bijection and satisfies the first identity in \eqref{gp-relative-identities}. For $x,y \in Y$, we obtain
\begin{equation*}
H_p^\ell(x,\Delta_py)
=\Tr\big((A_p\Delta_py)^*x\big)
=\Tr\big((yB_p)^*x\big)
=\Tr(y^*xB_p)
=H_p^r(x,y).
\end{equation*}
The Hermitian property and positive definiteness of $H_p^r$ imply that $\Delta_p$ is selfadjoint and positive for $H_p^\ell$. Since $Y$ is finite-dimensional and $\Delta_p$ is invertible, it has a basis of eigenvectors with strictly positive eigenvalues.

We next prove \eqref{gp-imaginary-powers}, taking particular care with possibly unbounded densities. Let $y\in Y$ be an eigenvector, with $\Delta_py=\lambda y$ and $\lambda>0$. Then
\begin{equation}
\label{gp-eigenvector-intertwining}
\lambda A_py=yB_p.
\end{equation}
All these products belong to $\L^{p^*}(\cal M)$. Put $R=(\lambda A_p+e)^{-1}\in e\cal Me$ and $S=(B_p+f)^{-1}\in f\cal Mf$. Multiplying \eqref{gp-eigenvector-intertwining} after adding $y$ to both sides by these bounded resolvents gives $Ry=yS$. Polynomial approximation therefore yields
\begin{equation}
\label{gp-continuous-functional-intertwining}
g(R)y=yg(S),
\quad g \in \C([0,1]).
\end{equation}
Functional calculi here are taken in the reduced algebras. Neither $R$ nor $S$ has a spectral atom at $0$ or $1$: the resolvents have trivial kernels on their supports, and $A_p$ and $B_p$ also have trivial kernels there.

Fix $t\in\R$. Choose continuous functions $\chi_n\co[0,1]\to[0,1]$ with compact support in $(0,1)$ and $\chi_n(s)\to1$ for any $0<s<1$. Define $g_n(s)=\chi_n(s)((1-s)/s)^{\i t}$ on $(0,1)$ and $g_n(0)=g_n(1)=0$. The spectral theorem gives bounded strong* convergence
\begin{equation*}
g_n(R)\longrightarrow(\lambda A_p)^{\i t},
\qquad
g_n(S)\longrightarrow B_p^{\i t}.
\end{equation*}
Bounded strong* convergence of multipliers implies convergence in $\L^p$ after multiplication by a fixed $\L^p$-vector. For completeness, this follows by approximating that vector in $\L^p$ by bounded operators $z$ with finite trace support and using
\begin{equation*}
\|T_nz\|_p^p\leq\|T_nz\|_\infty^{p-2}\|T_nz\|_2^2
\end{equation*}
when $T_n\to0$ strongly and is uniformly bounded; the right multiplication statement follows by taking adjoints. Passing to the limit in \eqref{gp-continuous-functional-intertwining} gives $(\lambda A_p)^{\i t}y=yB_p^{\i t}$, and hence
\begin{equation*}
A_p^{-\i t}yB_p^{\i t}=\lambda^{\i t}y=\Delta_p^{\i t}y.
\end{equation*}
Applying this to an eigenbasis proves \eqref{gp-imaginary-powers} on all of $Y$.

The maps in \eqref{gp-imaginary-powers} and their inverses are restrictions of left and right unitary multiplications on the rectangular corner. They are therefore complete isometries for the inherited operator space structure. Unitarity for $H_p^\ell$ follows from the spectral theorem. The relation $H_p^r(x,y)=H_p^\ell(x,\Delta_py)$ and commutation of $\Delta_p$ with its imaginary powers imply unitarity for $H_p^r$.

The polar partial isometries of elements of $Y$ belong to $e\cal Mf$, and the joins of their left and right supports are $e$ and $f$. Thus $e\cal Mf$ has full left and right supports in the reduced algebras, so $\cal R$ is its linking von Neumann algebra. Since $A_p,B_p\in\L^r(\cal M)$, we have $D_p\in\L^1(\cal R)_+$. Its support is $\operatorname{diag}(e,f)=1_{\cal R}$, so \eqref{gp-linking-state} defines a normal faithful state. By the tracial modular formula, see \cite{Tak03}, its modular automorphism group is
\begin{equation}
\label{gp-modular-action}
\sigma_t^{\psi_p}(X)=D_p^{\i t}XD_p^{-\i t},
\quad X\in\cal R, t\in\R.
\end{equation}
The scalar normalization $c_p$ cancels from this formula. The same unitary conjugation defines its isometric extension to any tracial $\L^q(\cal R)$, $1\leq q<\infty$. On the upper-right corner, it gives
\begin{equation*}
\sigma_{t,p}^{\psi_p}(\jmath(y))
=\jmath\big(A_p^{\i rt}yB_p^{-\i rt}\big),
\quad y \in e\L^p(\cal M)f.
\end{equation*}
Combining this with \eqref{gp-imaginary-powers} proves \eqref{gp-relative-group-modular}.

The map $W_p$ is well-defined by $1/2=1/p+1/(2r)$ and H\"older's inequality. Moreover,
\begin{equation*}
\langle W_px,W_py\rangle_{\L^2(\cal R)}
=\Tr(y^*A_px)=H_p^\ell(x,y),
\quad x,y \in Y,
\end{equation*}
so it is an isometry. The standard modular group on the tracial Hilbert space is $\Delta_{\psi_p}^{\i t}\xi=D_p^{\i t}\xi D_p^{-\i t}$. Consequently,
\begin{equation}
\label{gp-Hilbert-modular-intertwining}
\Delta_{\psi_p}^{\i t}W_py
=\jmath\big(A_p^{1/2}A_p^{\i rt}yB_p^{-\i rt}\big)
=W_p\Delta_p^{-\i rt}y.
\end{equation}
Thus $\cal K_p$ is invariant under the modular unitaries and their inverses, and hence reduces their selfadjoint generator and $\Delta_{\psi_p}$. Restricting \eqref{gp-Hilbert-modular-intertwining} to this finite-dimensional subspace and applying the spectral theorem gives
\begin{equation*}
\Delta_{\psi_p}|_{\cal K_p}=W_p\Delta_p^{-r}W_p^{-1},
\end{equation*}
which is equivalent to \eqref{gp-standard-modular-identification}.
\end{proof}

\begin{cor}
\label{cor-gp-modular-covariance}
Under the assumptions of Theorem~\ref{thm-gp-relative-modular-structure}, put $U_t=A_p^{-\i t}+(1-e)$ and $V_t=B_p^{\i t}+(1-f)$. Then
\begin{equation}
\label{gp-modular-covariance}
P(U_txV_t)
=U_tP(x)V_t,
\quad x \in \L^p(\cal M), t \in\R.
\end{equation}
\end{cor}

\begin{proof}
The map $\cal U_t\co x\mapsto U_txV_t$ is a surjective complete isometry of $\L^p(\cal M)$, and \eqref{gp-imaginary-powers} gives $\cal U_t(Y)=Y$. Thus $\cal U_t^{-1}P\cal U_t$ and $P$ are contractive projections onto the same nonzero subspace. Since $\L^p(\cal M)$ is smooth, Proposition~\ref{prop-Sullivan} gives $\cal U_t^{-1}P\cal U_t=P$, proving the assertion.
\end{proof}

\begin{remark} \normalfont
If $\cal M$ is finite-dimensional, then $A_p$ and $B_p$ are invertible in their reduced algebras, and \eqref{gp-relative-operator} becomes $\Delta_p(y)=A_p^{-1}yB_p$. For a general semifinite algebra, the definition through the restricted multiplication maps avoids any assumption that these inverses are bounded or even $\Tr$-measurable. Likewise, the embedding $W_p$ is used because $Y$ need not be contained in the original tracial $\L^2$-space.

For $p=4$, we have $r=2$, $A_4=\sum_jx_jx_j^*$ and $B_4=\sum_jx_j^*x_j$, so the modular time rescaling is $-\frac{t}{2}$. The preceding results establish weighted tangent and modular structures for any $p > 2$. The next subsection shows that, in the finite-rank case, these weighted identities lead to a ternary description of the range and to complete contractivity.
\end{remark}



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Ternary structure and complete contractivity in the finite-rank case}
\label{sec-fr-arbitrary-exponents}

We now complete the weighted construction for finite-rank projections. The appropriate ternary structure is obtained after a change of density: for $p\neq4$, one cannot in general expect the unweighted inclusion $YY^*Y\subset Y^\sharp$.

We shall use Propositions~\ref{prop-gp-weighted-tangent-estimates} and \ref{prop-gp-Gram-densities}, together with Theorem~\ref{thm-gp-relative-modular-structure}, for finite spanning families rather than just bases. Their proofs remain valid with this modification: the support identities only require that the family span $Y$, and the dimension argument uses $\dim Y^\sharp =\dim Y$, not the length of the family.

\begin{lemma}
\label{lem-fr-bounded-normalization}
Let $1<r<\infty$, let $A\in\L^r(\cal M)_+$ and put $e=s(A)$. If $a\in\L^r(\cal M)_+$ satisfies $a\leq A$, there exists a unique $C\in e\cal Me$ with $0\leq C\leq e$ such that $a= A^{\frac{1}{2}}CA^{\frac{1}{2}}$.
\end{lemma}

\begin{proof}
Put $e_n=\mathbf1_{[\frac{1}{n},n]}(A)$ and let $R_n=A^{-\frac{1}{2}}e_n$ be defined by bounded Borel functional calculus. The measurable operator $C_n=R_naR_n$ satisfies $0\leq C_n\leq e_n$, so $C_n$ belongs to $e\cal Me$. If $m\leq n$, then $e_mC_ne_m=C_m$. A weak* cluster point $C$ of the bounded sequence $(C_n)$ therefore satisfies $0 \leq C\leq e$ and $e_nCe_n=C_n$ for any $n$. It follows that
\begin{equation*}
e_nA^{\frac{1}{2}}CA^{\frac{1}{2}}e_n=e_nae_n.
\end{equation*}
Both $A^{\frac{1}{2}}CA^{\frac{1}{2}}$ and $a$ belong to $\L^r(\cal M)$, and compression by $e_n\uparrow e$ converges to the identity in $\L^r$ on operators supported by $e$. Hence $a=A^{\frac{1}{2}}CA^{\frac{1}{2}}$. The same compressions recover $e_nCe_n$ from $a$, proving uniqueness.
\end{proof}

\begin{lemma}
\label{lem-fr-invariant-algebra-factorization}
Let $\cal A\subset e\cal Me$ be a finite-dimensional von Neumann algebra with unit $e$. Suppose that $A\in\L^r(\cal M)_+$ has support $e$ and satisfies $A^{\i t}\cal A A^{-\i t}=\cal A$ for any $t\in\R$. Then there exist a positive invertible element $d\in\cal A$ and a positive operator $c\in\L^r(\cal M)$ with support $e$, affiliated with $\cal A'\cap e\cal Me$, such that $A=cd=dc$.
\end{lemma}

\begin{proof}
The restrictions of $\operatorname{Ad}(A^{\i t})$ form a norm-continuous one-parameter automorphism group of $\cal A$. They fix its center pointwise. On each full matrix summand, the generator is an inner $*$-derivation. Consequently, there exists $k=k^*\in\cal A$ such that
\begin{equation*}
A^{\i t}xA^{-\i t}
=\e^{\i tk}x\e^{-\i tk},
\quad x \in \cal A, t \in \R.
\end{equation*}
Set $d=\e^k$. Since the displayed automorphisms fix $d$, this element commutes with the spectral projections of $A$. Thus $c=Ad^{-1}$ is positive, belongs to $\L^r(\cal M)$, and has support $e$. Moreover, $c^{\i t}=A^{\i t}d^{-\i t}$ commutes with $\cal A$ for any $t$. The spectral theorem shows that $c$ is affiliated with $\cal A'\cap e\cal Me$.
\end{proof}

\begin{thm}
\label{thm-fr-general-p-TRO}
Let $2<p<\infty$ and let $P\co\L^p(\cal M) \to \L^p(\cal M)$ be a finite-rank $2$-contractive projection with nonzero range $Y$. Put $Y^\sharp=\Ran P^\dagger$, $e=s_\ell(Y)$ and $f=s_r(Y)$. There exist a finite-dimensional $\W^*$-$\TRO$ $\cal V\subset e\cal Mf$ with left and right supports $e,f$, and an operator $h\in\L^p(\cal M)_+$ with $s(h)=e$, such that
\begin{equation}
\label{fr-weighted-range-structure}
Y
=h\cal V,
\qquad
Y^\sharp
=h^{p-1}\cal V.
\end{equation}
The operator $h$ commutes with the left von Neumann algebra $\mathrm M(\cal V)=\operatorname{span}(\cal V\cal V^*)$. Moreover, $P$ is completely contractive.
\end{thm}

\begin{proof}
Put $\alpha=(p-2)/2$ and $r=p/(p-2)$. Choose a basis $u_1,\ldots,u_m$ of $Y$, and let $\cal F$ be the finite family consisting of these vectors and the vectors $u_j+\i^k u_l$, where $j<l$ and $k\in\{0,1,2,3\}$. Define
\begin{equation}
\label{fr-enlarged-Gram-densities}
A=\sum_{u\in\cal F}(uu^*)^\alpha,
\qquad
B=\sum_{u\in\cal F}(u^*u)^\alpha.
\end{equation}
The finite-spanning-family versions of Proposition~\ref{prop-gp-Gram-densities} and Theorem~\ref{thm-gp-relative-modular-structure} give
\begin{equation}
\label{fr-weighted-invariance}
A,B\in\L^r(\cal M)_+,
\quad s(A)=e,
\quad s(B)=f,
\quad A^{-\i t}YB^{\i t}=Y
\quad(t\in\R).
\end{equation}

\smallskip
\noindent\textit{The weighted Hilbert projection.}
Let $\cal H=e\L^2(\cal M)f$ and define $J(x)=A^{1/2}x$ on $e\L^p(\cal M)f$. H\"older's inequality makes this map well-defined. It is injective and has dense range in $\cal H$. To check density, let $z\in e\cal Mf\cap\L^2(\cal M)$ and put $e_n=\mathbf1_{[1/n,n]}(A)$. Since $p>2$, the bounded $\L^2$-vector $z$ belongs to $\L^p$. Thus $x_n=A^{-1/2}e_nz$ belongs to $e\L^p(\cal M)f$, and $J(x_n)=e_nz\to z$ in $\L^2$. Such vectors $z$ are dense in $\cal H$.

Summing the left inequalities in \eqref{gp-separate-weighted-estimates} over $\cal F$ gives $\|J(Px)\|_2\leq\|Jx\|_2$. Consequently,
\begin{equation}
\label{fr-Hilbert-projection}
QJx=JPx,
\quad x\in e\L^p(\cal M)f,
\end{equation}
defines a contractive projection on a dense subspace of $\cal H$. Its extension is an orthogonal projection with finite-dimensional range $\cal K=J(Y)$.

The left support of $\cal K$ is $e$. Indeed, suppose that $b\in e\cal Me$ satisfies $bA^{\frac{1}{2}}u_j=0$ for any $j$. Put $T=bA^{\frac{1}{2}} \in \L^{2r}(\cal M)$. Spectral truncation of $|u_j|$ in the equality $Tu_j=0$ gives $T s_\ell(u_j)=0$. Since $\bigvee_j s_\ell(u_j)=e$ and $T=Te$, it follows that $T=0$. Spectral truncation of $A$ then gives $b=0$.

\smallskip
\noindent\textit{A finite-dimensional algebra of normalized weights.}
For $u\in\cal F$, put $a_u=(uu^*)^\alpha$. Since $0\leq a_u\leq A$, Lemma~\ref{lem-fr-bounded-normalization} gives a positive contraction $C_u\in e\cal Me$ such that
\begin{equation}
\label{fr-normalized-weights}
a_u=A^{1/2}C_uA^{1/2}.
\end{equation}
For $t\in\R$, define $u_t=A^{-\i t}uB^{\i t}\in Y$ and $C_u(t)=A^{-\i t}C_uA^{\i t}$. Unitary covariance of functional calculus gives
\begin{equation*}
(u_tu_t^*)^\alpha
=A^{-\i t}a_uA^{\i t}
=A^{\frac{1}{2}}C_u(t)A^{\frac{1}{2}}.
\end{equation*}
The left weighted form associated with $u_t$ is therefore
\begin{equation*}
h_{u_t}^\ell(x,z)
=\langle C_u(t)Jx,Jz\rangle_{\cal H}.
\end{equation*}
By \eqref{gp-weighted-intertwining}, $P$ is selfadjoint for this form. Combining this with \eqref{fr-Hilbert-projection} and the selfadjointness of $Q$ shows that
\begin{equation}
\label{fr-normalized-commutation}
QL_{C_u(t)}
=L_{C_u(t)}Q
\quad\text{on }\cal H.
\end{equation}
Let $\cal A$ be the von Neumann subalgebra of $e\cal Me$ generated by $e$ and all the operators $C_u(t)$, for $u\in\cal F$ and $t\in\R$. Equation~\eqref{fr-normalized-commutation} shows that left multiplication by $\cal A$ leaves $\cal K$ reducing. Its restriction to $\cal K$ is a unital $*$-representation. This representation is faithful because $\cal K$ has left support $e$. Hence $\cal A$ embeds into $\B(\cal K)$ and is finite-dimensional. Its definition also gives $A^{\i t}\cal A A^{-\i t}=\cal A$ for any $t \in \R$.

Apply Lemma~\ref{lem-fr-invariant-algebra-factorization} to write $A=cd$, where $d \in \cal A$ is positive and invertible and $c\in\L^r(\cal M)_+$ has support $e$ and commutes with $\cal A$. Put
\begin{equation}
\label{fr-commuting-density}
h
=c^{\frac{1}{p-2}} \in \L^p(\cal M)_+.
\end{equation}
For $u \in \cal F$, let $k_u=d^{\frac{1}{2}}C_ud^{\frac{1}{2}} \in \cal A_+$. Then \eqref{fr-normalized-weights} becomes $a_u=ck_u$. Since $c$ commutes with $k_u$, joint functional calculus gives
\begin{equation}
\label{fr-recovering-bounded-vectors}
|u^*|
=(a_u){\frac{1}{p-2}}
=h k_u^{\frac{1}{p-2}}.
\end{equation}
Write the left polar decomposition of $u$ as $u=|u^*|w_u$, and put $v_u=k_u^{\frac{1}{p-2}}w_u\in e\cal Mf$. Then
\begin{equation}
\label{fr-bounded-vector-products}
u=hv_u,
\qquad
v_uv_u^*
=k_u^{\frac{2}{p-2}} \in \cal A.
\end{equation}
Here $s(|u^*|) = s(k_u)$, since $s(h)=e$ and $h$ commutes with $k_u$.

\smallskip
\noindent\textit{The deweighted range is a TRO.}
Let $\cal V=\operatorname{span}\{v_{u_1},\ldots,v_{u_m}\}$. Multiplication by $h$ is injective on $e\cal Mf$, so $Y=h\cal V$ and $\dim\cal V=m$. The same injectivity shows that
\begin{equation*}
v_{u_j+\i^k u_l}=v_{u_j}+\i^k v_{u_l}.
\end{equation*}
In particular, every $v_u$ occurring in \eqref{fr-bounded-vector-products} belongs to $\cal V$. Since $A=cd$ and $\frac{r}{2}=\frac{1}{2}+\frac{1}{p-2}$, we have
\begin{equation*}
\cal K=A^{\frac{1}{2}}Y
=c^{r/2}d^{\frac{1}{2}}\cal V.
\end{equation*}
For $b\in\cal A$ and $v\in\cal V$, invariance of $\cal K$ gives some $v'\in\cal V$ with
\begin{equation*}
bc^{\frac{r}{2}}d^{\frac{1}{2}}v
=c^{\frac{r}{2}}d^{\frac{1}{2}}v'.
\end{equation*}
The operator $c^{\frac{r}{2}}$ commutes with $\cal A$ and has support $e$, so cancellation gives $v'=d^{-\frac{1}{2}}bd^{\frac{1}{2}}v$. Since $d^{-1/2}\cal A d^{1/2}=\cal A$, this proves $\cal A\cal V\subset\cal V$.

The diagonal products $v_{u_j}v_{u_j}^*$ belong to $\cal A$ by \eqref{fr-bounded-vector-products}. For $j<l$, polarization gives
\begin{equation*}
v_{u_j}v_{u_l}^*
=\frac14\sum_{k=0}^3\i^k
\big(v_{u_j}+\i^k v_{u_l}\big)
\big(v_{u_j}+\i^k v_{u_l}\big)^*
\in\cal A.
\end{equation*}
Taking adjoints gives the remaining products. Thus $\cal V\cal V^*\subset\cal A$, and therefore
\begin{equation*}
\cal V\cal V^*\cal V
\subset\cal A\cal V
\subset\cal V.
\end{equation*}
The finite-dimensional space $\cal V$ is weak* closed, so it is a $\W^*$-$\TRO$. The polar decompositions in \eqref{fr-bounded-vector-products} give the same left and right supports for $u_j$ and $v_{u_j}$, hence $s_\ell(\cal V)=e$ and $s_r(\cal V)=f$. Finally, $h$ commutes with $\cal A$, and therefore with $\mathrm M(\cal V)\subset\cal A$.

\smallskip
\noindent\textit{Power duality at every matrix level.}
For a bounded operator $v=w|v|$ and $1<s<\infty$, write $F_s(v)=w|v|^{s-1}$. This notation does not require $v$ to belong to the ambient tracial $\L^s$-space. For $v\in\cal V$, the commutation just obtained gives
\begin{equation}
\label{fr-power-duality-factorization}
\frak j_p(hv)
=\big((hv)(hv)^*\big)^{\frac{p-2}{2}}hv
=h^{p-1}(vv^*)^{\frac{p-2}{2}}v
=h^{p-1}F_p(v).
\end{equation}
Functional calculus in the finite-dimensional linking algebra of $\cal V$ shows that $F_s(\cal V)\subset\cal V$ for any $1 < s < \infty$. The identities $F_{p^*}F_p=\Id$ and $F_pF_{p^*}=\Id$ consequently give $F_p(\cal V)=\cal V$. Since $P$ is contractive, \eqref{power-duality-map-range-equality}, interpreted under trace duality, and \eqref{fr-power-duality-factorization} yield $Y^\sharp=\frak j_p(Y)=h^{p-1}\cal V$.

Let $n\geq1$, put $h_n=\Id_{\M_n}\ot h$, and write $X=h_nV$ for $X\in\M_n(Y)$ and $V\in\M_n(\cal V)$. The same computation, now in the matrix amplification, gives
\begin{equation}
\label{fr-all-matrix-duality}
\frak j_p(X)=h_n^{p-1}F_p(V)
\in\M_n(h^{p-1}\cal V)
=\M_n(Y^\sharp)
=\Ran P_n^\dagger.
\end{equation}
For arbitrary $Z\in\L^p(\M_n\overline\otimes\cal M)$, set $X=P_nZ$. Since $Z-X \in \ker P_n$, equation~\eqref{fr-all-matrix-duality} and H\"older's inequality imply
\begin{equation*}
\|X\|_p^p
=\Tr\big(\frak j_p(X)^*Z\big)
\leq\|X\|_p^{p-1}\|Z\|_p.
\end{equation*}
The trace here is the amplified trace. Hence $\|P_nZ\|_p\leq\|Z\|_p$, including the trivial case $X=0$. Since $n$ is arbitrary, $P$ is completely contractive.
\end{proof}

\begin{cor}
\label{cor-fr-symmetric-Gram-deweighting}
Under the assumptions of Theorem~\ref{thm-fr-general-p-TRO}, choose any basis $x_1,\ldots,x_m$ of $Y$ and define $A_p,B_p$ by \eqref{gp-Gram-densities}. Put $\gamma=1/(2(p-2))$. Then the $\TRO$ constructed above satisfies
\begin{equation}
\label{fr-symmetric-Gram-deweighting}
Y=A_p^\gamma\cal V B_p^\gamma.
\end{equation}
More precisely, any $y \in Y$ has a unique representation $y=A_p^\gamma vB_p^\gamma$ with $v \in \cal V$. Thus $A_p^{-\gamma}YB_p^{-\gamma}$ can be interpreted as a space of bounded solutions, without assuming that the negative powers are measurable operators. This space is independent of the chosen basis.
\end{cor}

\begin{proof}
Let $Y=h\cal V$ be as in Theorem~\ref{thm-fr-general-p-TRO}. The right algebra $\mathrm N(\cal V)=\operatorname{span}(\cal V^*\cal V)$ is finite-dimensional with unit $f$. Choose a basis $v_1,\ldots,v_m$ of $\cal V$, put $S=\sum_jv_j^*v_j$, and define $w_j=v_jS^{-1/2}\in\cal V$, taking the inverse in $\mathrm N(\cal V)$. Then $\sum_jw_j^*w_j=f$. Set
\begin{equation*}
k=\sum_jw_j^*h w_j\in\L^p(\cal M)_+.
\end{equation*}
For $v\in\cal V$, commutation of $h$ with $\mathrm M(\cal V)$ gives
\begin{equation*}
vk=\sum_jvw_j^*h w_j
=h\sum_jvw_j^*w_j=hv.
\end{equation*}
The positivity of $h$ and $s(h)=e$, together with $\sum_jw_j^*w_j=f$, imply $s(k)=f$. Taking adjoints in $hv=vk$ and multiplying by elements of $\cal V$ shows that $k$ commutes with $\mathrm N(\cal V)$. Resolvent intertwining, followed by spectral functional calculus, also gives $h^{1/2}v=vk^{1/2}$.

Write the chosen basis of $Y$ as $x_j=hz_j=z_jk$, with $z_1,\ldots,z_m$ a basis of $\cal V$. Then
\begin{equation*}
A_p
=h^{p-2}S_\ell,
\qquad
B_p
=k^{p-2}S_r,
\end{equation*}
where $S_\ell=\sum_j(z_jz_j^*)^{(p-2)/2}\in\mathrm M(\cal V)$ and $S_r=\sum_j(z_j^*z_j)^{(p-2)/2}\in\mathrm N(\cal V)$. These positive elements have supports $e,f$ and are invertible in their respective finite-dimensional algebras. Therefore
\begin{equation*}
A_p^\gamma\cal V B_p^\gamma
=h^{\frac{1}{2}}S_\ell^\gamma\cal V S_r^\gamma k^{\frac{1}{2}}
=h^{\frac{1}{2}}\cal V k^{\frac{1}{2}}
=h\cal V=Y.
\end{equation*}
Here $\mathrm M(\cal V)\cal V\subset\cal V$ and $\cal V\mathrm N(\cal V)\subset\cal V$ imply $S_\ell^\gamma\cal V S_r^\gamma=\cal V$. Since $s(A_p)=e$ and $s(B_p)=f$, the map $v \mapsto A_p^\gamma vB_p^\gamma$ is injective on $e\cal Mf$, by spectral truncation. This proves uniqueness and the asserted interpretation. The same space $\cal V$ works for every basis, proving the final assertion.
\end{proof}

\begin{cor}
\label{cor-fr-all-exponents-complete-contractivity}
Let $1 < p < \infty$ with $p \neq 2$. Any finite-rank $2$-contractive projection on the tracial noncommutative space $\L^p(\cal M)$ is completely contractive.
\end{cor}

\begin{proof}
The zero projection is immediate, and Theorem~\ref{thm-fr-general-p-TRO} handles $p > 2$. If $1<p<2$, then $p^* > 2$ and $P^\dagger$ is a finite-rank $2$-contractive projection on $\L^{p^*}(\cal M)$. The theorem makes $P^\dagger$ completely contractive. Since $(P_n)^\dagger=(P^\dagger)_n$ and $\|P_n\|=\|(P^\dagger)_n\|$ for any integer $n$, the same follows for $P$.
\end{proof}

\begin{remark} \normalfont
\label{rem-fr-scope-and-unweighted-products}
The finite-rank hypothesis is used when the algebra of normalized weights is represented faithfully on the finite-dimensional space $\cal K$. It makes that algebra finite-dimensional, so its modular action can be implemented internally. The argument does not prove complete contractivity of an arbitrary infinite-rank $2$-contractive projection.

The unweighted inclusion $YY^*Y \subset Y^\sharp$ is false for $p\neq4$, even in rank one. Indeed, on $\L^p(\mathbb C^2)=\ell_p^2$, the projection
\begin{equation*}
P(x_1,x_2)
=\frac{x_1+2^{p-1}x_2}{1+2^p}(1,2)
\end{equation*}
is completely contractive. At matrix level $n$, this follows from the triangle inequality in $S_n^p$ and the scalar H\"older inequality applied to the two matrix norms. Its range is $Y=\mathbb C(1,2)$, whereas $Y^\sharp=\mathbb C(1,2^{p-1})$ and $YY^*Y=\mathbb C(1,8)$. These last two lines coincide only when $p=4$. Formula~\eqref{fr-weighted-range-structure}, rather than an unweighted cubic inclusion, is the appropriate replacement for general $p$.
\end{remark}

%\paragraph{AI statement.} 
%The author acknowledges the use of AI tools, for language polishing, LaTeX editing, exploratory mathematical discussions and feedback during the development and preparation of this manuscript. 
%Suggestions from these AI interactions were subsequently examined, reformulated, incorporated into the manuscript, and independently verified by the authors. The authors take full responsibility for all mathematical content in the final manuscript.
%The author acknowledges the use of AI tools, including ChatGPT, for language polishing, LaTeX editing, and exploratory mathematical discussions during the development and preparation of this manuscript. Some ideas used in the proof-development process in Section \ref{sec:splitting} arose during interactions with GPT-5.6 Sol. Suggestions from these AI interactions were subsequently examined, reformulated, incorporated into the manuscript, and independently verified by the authors. The authors take full responsibility for all mathematical content in the final manuscript.


\paragraph{Competing interests} The author declares that he has no competing interests.

\paragraph{Data availability} No data sets were generated during this study.
%s and Grants

%\paragraph{Acknowledgment} 
%Part of this work was carried out during a visit of the first author to Nankai University (China) in 2025, supported by the National Natural Science Foundation of China (Grant No.~12571143). The first author gratefully acknowledges the hospitality of Nankai University.


%I would like to thank Li Gao for a very brief yet enlightening conversation regarding the <<proof>> of \cite[Lemma 3.11 p.~3434]{GJL20}. The proof of Proposition \ref{prop-Gao} is inspired by this discussion. The author wishes to express profound gratitude to the referee whose meticulous and comprehensive review significantly contributed to the improvement of this paper. The detailed comments and constructive suggestions provided were invaluable in enhancing both the accuracy and clarity of the mathematical arguments presented herein. This thorough engagement has undoubtedly enriched the quality of the work.

%The author gratefully acknowledges the support from the French National Research Agency grant ANR-18-CE40-0021 (project HASCON). We are thankful to Shouhei Honda, Naotaka Kajino, Jun Kigami, Bogdan Nica, Arup Kumar Pal, Jiayin Pan, El Maati Ouhabaz, Adrián González-Pérez, Sang-Gyun Youn, and Melchior Wirth for their valuable insights during brief discussions. Our appreciation extends to Li Gao and Bogdan Nica for their feedback and corrections, and to Christopher Sogge for pointing us to the reference \cite{BSS21}. Finally, a special note of gratitude goes to Bruno Iochum for the numerous discussions we had between the inception of the first version of this paper and the release of its preprint \cite{IoZ23}.


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\small
\begin{thebibliography}{79}%79

%%\bibitem[AbA02]{AbA02}
%%Y. A. Abramovich and C. D. Aliprantis.
%%\newblock An invitation to operator theory.
%%\newblock Graduate Studies in Mathematics, 50. American Mathematical Society, Providence, RI, 2002.
%
%%\bibitem[AVDR81]{AVDR81}
%%G. N. Agaev, N. Y. Vilenkin, G. M. Dzhafarli, A. I. Rubinshtein.
%%\newblock Multiplicative systems of functions and harmonic analysis on zero-dimensional groups (Russian).
%%\newblock Elm, Baku, 1981.
%%
%%
%%%\bibitem[AMTW1]{AMTW1}
%%%A. Ambainis, M. Mosca, A. Tapp and R. de Wolf.
%%%\newblock Private quantum channels. 
%%%\newblock 41st Annual Symposium on Foundations of Computer Science (Redondo Beach, CA, 2000), 547--553, IEEE Comput. Soc. Press, Los Alamitos, CA, 2000.
%%
%%%\bibitem[ACN00]{ACN00}
%%%G. G. Amosov, A. S. Holevo and R. F. Werner.
%%%\newblock On some additivity problems in quantum information theory. 
%%%\newblock math-ph/0003002, 2000.
%%
%%%\bibitem[ACN20]{ACN20}
%%%M. Alaghmandan, J. Crann and M. Neufang.
%%%\newblock Mapping ideals of quantum group multipliers. 
%%%\newblock Adv. Math. 374 (2020), 107353, 52 pp.
%%
%%
%%\bibitem[AlC18]{AlC18}
%%F. Alajaji and P.-N. Chen.
%%\newblock An introduction to single-user information theory.
%%\newblock Springer Undergrad. Texts Math. Technol., Springer, Singapore, 2018.
%%
%%\bibitem[ACP20]{ACP20}
%%V. V. Albert, J. P. Covey, and J. Preskill.
%%\newblock Robust Encoding of a Qubit in a Molecule.
%%\newblock Phys. Rev. X 10, 031050.
%%
%%\bibitem[AHK80]{AHK80}
%%S. Albeverio and R. Hoegh-Krohn.
%%\newblock Ergodic actions by compact groups on $C^*$-algebras. 
%%\newblock Math. Z. 174 (1980), no. 1, 1--17. 
%%
%%%\bibitem{AcC82}
%%%L. Accardi and C. Cecchini.
%%%\newblock Conditional expectations in von Neumann algebras and a theorem of Takesaki.
%%%\newblock J. Funct. Anal. 45 (1982), 245--273.
%%
%%%\bibitem[AcL]{AcL}
%%%L. Accardi and R. Longo.
%%%\newblock Martingale convergence of generalized conditional expectations.
%%%\newblock J. Funct. Anal. 118 (1993), no. 1, 119--130.
%%

%\bibitem[Alb34]{Alb34}
%A. A. Albert.
%\newblock On a certain algebra of quantum mechanics.
%\newblock Ann. of Math. (2) 35 (1934), no. 1, 65--73.

\bibitem[Ale81]{Ale81}
A. B. Aleksandrov.
\newblock Essays on non locally convex Hardy classes.
\newblock In: Complex Analysis and Spectral Theory (V. P. Havin and N. K. Nikolski, eds.), Lecture Notes in Mathematics, vol. 864, pp. 1--89. Springer, Berlin, 1981.

%\bibitem[Alf71]{Alf71}%
%E. M. Alfsen.
%\newblock Compact Convex Sets and Boundary Integrals.
%\newblock Ergebnisse der Mathematik und ihrer Grenzgebiete. 2. Folge, vol. 57. Springer-Verlag, Berlin Heidelberg, 1971.

%\bibitem[AHOS80]{AHOS80}
%E. M. Alfsen, H. Hanche-Olsen and F. W. Schultz.
%\newblock State spaces of $C^*$-algebras. 
%\newblock Acta Math. 144 (1980), no. 3-4, 267--305. 

%\bibitem[AlS78]{AlS78}
%E. M. Alfsen and F. W. Shultz.
%\newblock State spaces of Jordan algebras. 
%\newblock Acta Math. 140 (1978), no. 3-4, 155--190. 

%\bibitem[AlS01]{AlS01}%
%E. M. Alfsen and F. W. Shultz.
%\newblock State Spaces of Operator Algebras: Basic Theory, Orientations, and $C^*$-products.
%\newblock Mathematics: Theory \& Applications. Birkh\"auser Boston, MA, 2001.

%\bibitem[AlS03]{AlS03}
%E. M. Alfsen and F. W. Shultz.
%\newblock Geometry of state spaces of operator algebras. 
%\newblock Mathematics: Theory \& Applications. Birkhauser Boston, Inc., Boston, MA, 2003. 

%\bibitem[AlS98]{AlS98}
%E. M. Alfsen and F. W. Shultz.
%\newblock Orientation in operator algebras. 
%\newblock Proc. Natl. Acad. Sci. USA95 (1998), no.12, 6596--6601. 

%\bibitem[Alf79]{Alf79}
%E. M. Alfsen.
%\newblock On the state spaces of Jordan and $C^*$-algebras. 
%\newblock Alg\`ebres d'op\'erateurs et leurs applications en physique math\'ematique (Proc. Colloq., Marseille, 1977), pp. 15--40, Colloq. Internat. CNRS, 274, CNRS, Paris, 1979.   

%\bibitem[ASS78]{ASS78}
%E. M. Alfsen, F. W. Shultz, E. St\o rmer.
%\newblock A Gelfand-Neumark theorem for Jordan algebras. 
%\newblock Advances in Math. 28 (1978), no. 1, 11--56. 


%\bibitem[AlT07]{AlT07}%
%C. D. Aliprantis and R. Tourky.
%\newblock Cones and Duality.
%\newblock Graduate Studies in Mathematics, vol. 84. American Mathematical Society, Providence, RI, 2007.

%%
%%%\bibitem[And]{And}
%%%T. Ando.
%%%\newblock Contractive projections in $L_{p}$ spaces.
%%%\newblock Pacific J. Math. 17 (1966), 391--405.
%%
%%%\bibitem[AnD]{AnD}
%%%C. Anantharaman-Delaroche.
%%%\newblock On ergodic theorems for free group actions on noncommutative spaces.
%%%\newblock Probab. Theory Related Fields 135 (2006), no. 4, 520--546.

\bibitem[And66]{And66}
T. Ando.
\newblock Contractive projections in $L_{p}$ spaces.
\newblock Pacific J. Math. 17 (1966), 391--405.


%%%\bibitem[AHW]{AHW}
%%%H. Ando, U. Haagerup, C. Winslow.
%%%\newblock Ultraproducts, QWEP von Neumann Algebras, and the Effros-Mar\'echal Topology.
%%%\newblock Arxiv.
%%
%%%\bibitem[AlS]{AlS}
%%%Alfsen-Shultz.
%%%\newblock Geometry of state spaces of operator algebras.
%%%\newblock Book.
%%
%%%\bibitem[AnH]{AnH}
%%%H. Ando and U. Haagerup.
%%%\newblock Ultraproducts of von Neumann algebras.
%%%\newblock J. Funct. Anal. 266 (2014), no. 12, 6842--6913.  
%%
%%%\bibitem[Ara68]{Ara68}
%%%H. Araki.
%%%\newblock On the diagonalization of a bilinear Hamiltonian by a Bogoliubov transformation.
%%%\newblock Publ. Res. Inst. Math. Sci. Ser. A 4 (1968), 387--412.
%%
%%
%%%\bibitem[ArF1]{ArF1}
%%%J. Arazy and Y. Friedman.
%%%\newblock Contractive projections in $C_1$ and $C_\infty$.
%%%\newblock Mem. Amer. Math. Soc. 13 (1978), no. 200, 1--165.

\bibitem[ArF92]{ArF92}
J. Arazy and Y. Friedman.
\newblock Contractive projections in $C_p$.
\newblock Mem. Amer. Math. Soc. 459 (1992), 1--109.
%%
%%%\bibitem[Amo22]{Amo22}
%%%G. G. Amosov.
%%%\newblock On capacity of quantum channels generated by irreducible projective unitary representations of finite groups.
%%%\newblock Quantum Inf. Process. 21 (2022), no. 2, Paper No. 81, 15 pp.
%%
%%%\bibitem[Are04]{Are04}
%%%W. Arendt.
%%%\newblock Semigroups and evolution equations: functional calculus, regularity and kernel estimates. 
%%%\newblock Evolutionary equations. Vol. I, 1--85, Handb. Differ. Equ., North-Holland, Amsterdam, 2004. 
%%
%%%\bibitem[ABHN11]{ABHN11}
%%%W. Arendt, C. J. K. Batty, M. Hieber and F. Neubrander.
%%%\newblock Vector-valued Laplace transforms and Cauchy problems. Second edition.
%%%\newblock Monographs in Mathematics, 96. Birkhäuser/Springer Basel AG, Basel, 2011.
%%
%%%\bibitem[Arh11]{Arh11}
%%%C. Arhancet.
%%%\newblock Noncommutative Fig\`a-Talamanca-Herz algebras for Schur multipliers.
%%%\newblock Integral Equations Operator Theory 70 (2011), no. 4, 485--510.

%\bibitem[Arh19]{Arh19}
%C. Arhancet.
%\newblock Dilations of semigroups on von Neumann algebras and noncommutative $\L^p$-spaces.
%\newblock J. Funct. Anal. 276 (2019), no. 7, 2279--2314.

\bibitem[Arh24a]{Arh24a}
C. Arhancet.
\newblock Nonassociative $\L^p$-spaces and embeddings in noncommutative $\L^p$-spaces.
\newblock J. Math. Anal. Appl. 537 (2024), no. 1, Paper No. 128307, 24 pp.

\bibitem[Arh24b]{Arh24b}
C. Arhancet.
\newblock Contractively decomposable projections on noncommutative $\L^p$-spaces.
\newblock J. Math. Anal. Appl. 533 (2024), no. 2, Paper No. 128017, 36 pp.

\bibitem[ArR24]{ArR24}
C. Arhancet and Y. Raynaud.
\newblock 2-positive contractive projections on noncommutative $\mathrm{L}^p$-spaces.
\newblock J. Funct. Anal. 286 (2024), no. 11, Paper No. 110424, 34 pp..

%%
%%%\bibitem[Arh13]{Arh13}
%%%C. Arhancet.
%%%\newblock On Matsaev's conjecture for contractions on noncommutative $L^p$-spaces. 
%%%\newblock J. Operator Theory 69 (2013), no. 2, 387--421.

%\bibitem[Arh25]{Arh25}
%C. Arhancet.
%\newblock Entanglement-assisted classical capacities of some channels acting as radial multipliers on fermion algebras. 
%\newblock J. Funct. Anal. 288 (2025), no. 5, Paper No. 110790, 47 pp.

%\bibitem[Arh26]{Arh26}
%C. Arhancet.
%\newblock Quantum information theory via Fourier multipliers on quantum groups.
%\newblock Preprint, arXiv:2008.12019.

\bibitem[Arh26b]{Arh26b}
C. Arhancet.
\newblock Positive contractive projections on noncommutative $\L^p$-spaces and nonassociative $\L^p$-spaces.
\newblock Preprint.

 
\bibitem[Arh26c]{Arh26c}
C. Arhancet.
\newblock Completely isometric subspaces of noncommutative $\mathrm{L}^p$-spaces and contractive projections.
\newblock Preprint, arXiv:2608.20082.


\bibitem[ArL26]{ArL26}
C. Arhancet and L. Li.
\newblock Spectral versus interpolation norms in tracial nonassociative $\L^p$-spaces.
\newblock Preprint, arXiv:2604.23232.

%%
%%\bibitem[Arh25a]{Arh25a}
%%C. Arhancet.
%%\newblock Entanglement-assisted classical capacities of some channels acting as radial multipliers on fermion algebras.
%%\newblock J. Funct. Anal. 288 (2025), no. 5, Paper No. 110790.
%%
%%\bibitem[Arh25b]{Arh25b}
%%C. Arhancet.
%%\newblock Quantum information theory via Fourier multipliers on quantum groups.
%%\newblock Preprint, arXiv: arXiv:2008.12019.
%%
%%
%%%\bibitem[Arh23]{Arh23}
%%%C. Arhancet.
%%%\newblock Sobolev algebras on Lie groups and noncommutative geometry.
%%%\newblock Preprint, arXiv:2203.06603.
%
%\bibitem[ArK23]{ArK23}
%C. Arhancet and C. Kriegler.
%\newblock Projections, multipliers and decomposable maps on noncommutative $\L^p$-spaces.
%\newblock M\'emoires de la Soci\'et\'e Math\'ematique de France (N.S.) (2023), no. 177.
%
%\bibitem[ArK25]{ArK25}
%C. Arhancet and C. Kriegler.
%\newblock Decomposable Fourier Multipliers and an Operator-Algebraic Characterization of Amenability.
%\newblock Preprint.
%
%
%%\bibitem[ArK22]{ArK22}
%%C. Arhancet and C. Kriegler.
%%\newblock Riesz transforms, Hodge-Dirac operators and functional calculus for multipliers I.
%%\newblock Preprint, arXiv:1903.10151.
%%
%%\bibitem[ArK24]{ArK24}
%%C. Arhancet and C. Kriegler.
%%\newblock Optimal angles in functional calculus.
%%\newblock Preprint.
%%
%%\bibitem[ArK25]{ArK25}
%%C. Arhancet and C. Kriegler.
%%\newblock Sharp functional calculus for the Taibleson operator on non-archimedean local fields.
%%\newblock Preprint.
%%
%%
%%
%%%\bibitem[ArK4]{ArK4}
%%%C. Arhancet and C. Kriegler.
%%%\newblock Projections, multipliers and decomposable maps on noncommutative $\L^p$-spaces II.
%%%\newblock Preprint.


\bibitem[ArG01]{ArG01}%
D. H. Armitage and S. J. Gardiner.
\newblock Classical Potential Theory.
\newblock Springer Monographs in Mathematics. Springer, London, 2001.

\bibitem[BeL74]{BeL74}
S. J. Bernau and H. E. Lacey.
\newblock The range of a contractive projection on an $L^p$-space.
\newblock Pacific J. Math. 53 (1974), 21--41.

%\bibitem[BL77]{BL77}%
%S. J. Bernau and H. E. Lacey.
%\newblock Bicontractive projections and reordering of $L_p$-spaces.
%\newblock Pacific J. Math. 69 (1977), no. 2, 291--302.



\bibitem[Bla06]{Bla06}
B. Blackadar.
\newblock Operator algebras. Theory of $C^*$-algebras and von Neumann algebras.
\newblock Encyclopaedia of Mathematical Sciences, 122. Operator Algebras and Non-commutative Geometry, III. Springer-Verlag, Berlin, 2006.
%%
%%
%%%\bibitem[Aud09]{Aud09}
%%%K M. R. Audenaert.
%%%\newblock A note on the $p \to q$ norms of 2-positive maps. 
%%%\newblock Linear Algebra Appl. 430 (2009), no. 4, 1436--1440. 
%%
%%%\bibitem[ABW1]{ABW1}
%%%S. Avsec, M. Brannan and M. Wasilewski.
%%%\newblock Complete metric approximation property for $q$-Araki-Woods algebras.
%%%\newblock J. Funct. Anal. 274 (2018), no. 2, 544--572.

%\bibitem[Ayu82]{Ayu82}
%S. A. Ayupov.
%\newblock Extension of traces and type criterions for Jordan algebras of selfadjoint operators.
%\newblock Math. Z. 181 (1982), no. 2, 253--268.

%\bibitem[Ayu84]{Ayu84}%
%S. A. Ayupov.
%\newblock Integration on Jordan algebras (english version).
%\newblock Math. USSR-Izv. 22 (1984), no. 1, 1--21.

%\bibitem[AyA85]{AyA85}
%S. A. Ayupov and R. Z. Abdullaev.
%\newblock The Radon-Nikodym theorem for weights on semifinite JBW-algebras.
%\newblock Math. Z. 188 (1985), no. 4, 475--484.

%\bibitem[Ayu87]{Ayu87}
%S. A. Ayupov.
%\newblock Jordan operator algebras.
%\newblock Journal of Soviet Mathematics, vol. 37 (1987), 1422--1448.

%\bibitem[Ayu92]{Ayu92}
%S. A. Ayupov.
%\newblock Center-valued traces on real operator algebras.
%\newblock Funktsional. Anal. i Prilozhen. 26 (1992), no. 2, 1--9, 96; translation in Funct. Anal. Appl. 26 (1992), no. 2, 77--83.

%\bibitem[ARU97]{ARU97}
%S. A. Ayupov, A. Rakhimov and S. Usmanov.
%\newblock Jordan, real and Lie structures in operator algebras.
%\newblock Mathematics and its Applications, 418. Kluwer Academic Publishers Group, Dordrecht, 1997.
%%
%%\bibitem[BaB70]{Bab70}
%%N. B. Backhouse and C. J. Bradley.
%%\newblock Projective representations of space groups. I. Translation groups.
%%\newblock Quart. J. Math. Oxford Ser. (2) 21 (1970), 203--222.
%%
%%\bibitem[Bac70]{Bac70}
%%N. B. Backhouse.
%%\newblock Projective representations of space groups. II. Factor systems.
%%\newblock Quart. J. Math. Oxford Ser. (2) 21 (1970), 277--295.

%\bibitem[Bae02]{Bae02}
%J. Baez.
%\newblock The octonions.
%\newblock Bull. Amer. Math. Soc 39 (2002), 145--205.


%%\bibitem[BaK73]{BaK73}
%%L. Baggett and A. Kleppner.
%%\newblock Multiplier representations of abelian groups.
%%\newblock J. Funct. Anal. 14 (1973), 299--324.
%%
%%%\bibitem[BJLRF]{BJLRF}
%%%I. Bardet, M. Junge, N. LaRacuente, C. Rouz\'e, D. S. França.
%%%\newblock Group transference techniques for the estimation of the decoherence times and capacities of quantum Markov semigroups. 
%%%\newblock Preprint, arXiv:1904.11043. 
%%
%%\bibitem[Bar54]{Bar54}
%%V. Bargmann.
%%\newblock On unitary ray representations of continuous groups.
%%\newblock Ann. of Math. (2) 59 (1954), 1--46.
%%
%%
%%%\bibitem[BeH1]{BeH1}
%%%B. Bekka and P. de la Harpe.
%%%\newblock Unitary representations of groups, duals, and characters.
%%%\newblock Preprint, arXiv:1912.07262.
%%
%%\bibitem[Ban17]{Ban17}
%%T. Banica.
%%\newblock Liberation theory for noncommutative homogeneous spaces. 
%%\newblock Ann. Fac. Sci. Toulouse Math. (6) 26 (2017), no. 1, 127--156. 
%%
 %%
%%\bibitem[Ban20]{Ban20}
%%T. Banica.
%%\newblock Quantum isometries and noncommutative geometry. 
%%\newblock Preprint, 2020 \href{https://banica.u-cergy.fr/}{https://banica.u-cergy.fr/}. 
%%
%%\bibitem[BaG10]{BaG10}
%%T. Banica and D. Goswami.
%%\newblock Quantum isometries and noncommutative spheres. 
%%\newblock Comm. Math. Phys. 298 (2010), no. 2, 343--356. 
%%
%%\bibitem[BSS12]{BSS12}
%%T. Banica, A. Skalski and P. Soltan.
%%\newblock Noncommutative homogeneous spaces: the matrix case. 
%%\newblock J. Geom. Phys. 62 (2012), no. 6, 1451--1466. 
 %%
%%%\bibitem[BaC17]{BaC17}
%%%T. Banica and A. Chirvasitu.
%%%\newblock Thoma type results for discrete quantum groups. 
%%%\newblock Internat. J. Math. 28 (2017), no. 14, 1750103, 23 pp. 

%\bibitem[Bar07]{Bar07}
%J. Barrett.
%\newblock Information processing in generalized probabilistic theories.
%\newblock Phys. Rev. A 75 (2007), 032304.

%\bibitem[BGW15]{BGW15}
%H. Barnum, M. A. Graydon and A. Wilce.
%\newblock Some Nearly Quantum Theories.
%\newblock Proceedings QPL 2015.

%\bibitem[BGW20]{BGW20}
%H. Barnum, M. A. Graydon and A. Wilce.
%\newblock Composites and Categories of Euclidean Jordan Algebras.
%\newblock Quantum 4, 359 (2020).

%\bibitem[BUW23]{BUW23}
%H. Barnum, C. Ududec and J. van de Wetering.
%\newblock Self-duality and Jordan structure of quantum theory follow from homogeneity and pure transitivity.
%\newblock Preprint, arXiv:2306.00362.

%\bibitem[BaW11]{BaW11}
%H. Barnum and A. Wilce
%\newblock Information Processing in Convex Operational Theories.
%\newblock Electronic Notes in Theoretical Computer Science 270 (2011), no. 1, 3--15.

%\bibitem[BaW14]{BaW14}
%H. Barnum and A. Wilce.
%\newblock Local Tomography and the Jordan Structure of Quantum Theory.
%\newblock Found Phys 44, 192--212 (2014).

%%%\bibitem[Bat21]{Bat21}
%%%B.-O. Battseren.
%%%\newblock Groups with tame cuts. 
%%%\newblock PhD Thesis, 2021. 
%%
%%\bibitem[Bau10]{Bau10}
%%S. B\"auml.
%%\newblock On Bound Key And The Use Of Bound Entanglement.
%%\newblock Diploma thesis, Ludwig Maxi- milians Universit\"at M\"unchen, (2010). 
%%
%%\bibitem[BCHW15]{BCHW15}
%%S. B\"auml, M. Christandl, K. Horodecki and  A. Winter.
%%\newblock Limitations on quantum key repeaters.
%%\newblock Nat. Comm. 6, 6908 (2015).
%%
%%
%%%\bibitem[BeM73]{BeM73}
%%%B. Beauzamy and B. Maurey.
%%%\newblock Op\'erateurs de convolution $r$-sommants sur un groupe compact ab\'elien. 
%%%\newblock C. R. Acad. Sci. Paris S\'er. A-B 277 (1973), A113--A115.
%%
%%\bibitem[BeC09]{BeC09}
%%E. B\'edos and R. Conti.
%%\newblock On twisted Fourier analysis and convergence of Fourier series on discrete groups.
%%\newblock J. Fourier Anal. Appl. 15 (2009), no. 3, 336--365.
%%
%%
%%\bibitem[BeS08]{BeS08}
%%S. Beigi and P. W. Shor.
%%\newblock On the Complexity of Computing Zero-Error and Holevo Capacity of Quantum Channels.
%%\newblock Preprint, arXiv:0709.2090.
%%
%%
%%%\bibitem[BeT03]{BeT03}
%%%E. B\'edos and L. Tuset.
%%%\newblock Amenability and co-amenability for locally compact quantum groups. 
%%%\newblock Internat. J. Math. 14 (2003), no. 8, 865--884.
%%
%%%\bibitem[BeH20]{BeH20}
%%%B. Bekka and P. de la Harpe.
%%%\newblock Unitary representations of groups, duals, and characters.
%%%\newblock Mathematical Surveys and Monographs, 250. American Mathematical Society, Providence, RI, 2020.
%%
%%%\bibitem[BWZ13]{BWZ13}
%%%I. Bengtsson, S. Weis, K. Zyczkowski.
%%%\newblock Geometry of the set of mixed quantum states: an apophatic approach.
%%%\newblock Geometric methods in physics, 175--197, Trends Math., Birkh\"auser/Springer, Basel, 2013. 
%%
%%\bibitem[BRV06]{BRV06}
%%J. Bichon, A. De Rijdt and S. Vaes.
%%\newblock Ergodic coactions with large multiplicity and monoidal equivalence of quantum groups. 
%%\newblock Comm. Math. Phys. 262 (2006), no. 3, 703--728. 
%%
%%%\bibitem[BiJ00]{BiJ00}
%%%D. Bisch and V. Jones.
%%%\newblock Singly generated planar algebras of small dimension. 
%%%\newblock Duke Math. J. 101 (2000), no. 1, 41--75. 
%%
%%%\bibitem[BDG21]{BDG21}
%%%M. Bischoff, S. Del Vecchio and L. Giorgetti.
%%%\newblock Compact hypergroups from discrete subfactors. 
%%%\newblock J. Funct. Anal. 281 (2021), no. 1, Paper No. 109004, 78 pp. 
%%
%%%\bibitem[BGR1]{BGR1}
%%%L. G. Brown, P. Green and M. A. Rieffel.
%%%\newblock Stable isomorphism and strong Morita equivalence of $C^*$-algebras.
%%%\newblock Pacific J. Math. 71 (1977), no. 2, 349--363.

%\bibitem[BeI78]{BeI78}
%J. B\'ellissard and B. Iochum.
%\newblock Homogeneous self-dual cones, versus Jordan algebras. The theory revisited. 
%\newblock Ann. Inst. Fourier (Grenoble) 28 (1978), no. 1, v, 27--67.

%\bibitem[BeI79]{BeI79}
%J. B\'ellissard and B. Iochum.
%\newblock L'alg\`ebre de Jordan d'un c\^one autopolaire facialement homog\`ene. 
%\newblock C. R. Acad. Sci. Paris S\'er. A-B 288 (1979), no. 4, A229--A232.

%\bibitem[BeI80]{BeI80}
%J. B\'ellissard and B. Iochum.
%\newblock Homogeneous self-dual cones and Jordan algebras. Quantum fields--algebras, processes. 
%\newblock Proc. Sympos., Univ. Bielefeld, Bielefeld, 1978, 153--165, Springer, Vienna, 1980.


%%%\bibitem[BKK1]{BKK1}
%%%C. B\'eny, A. Kempf and D. W. Kribs.
%%%\newblock Quantum error correction on infinite-dimensional Hilbert spaces. 
%%%\newblock J. Math. Phys. 50 (2009), no. 6, 062108, 24 pp. 
%%
%%\bibitem[BSST02]{BSST02}
%%C. H. Bennett, P. W. Shor, J. A. Smolin and A. V. Thapliyal.
%%\newblock Entanglement-assisted capacity of a quantum channel and the reverse Shannon theorem.
%%\newblock IEEE Trans. Inform. Theory 48 (2002), no. 10, 2637--2655.
%
\bibitem[BeL76]{BeL76}
J. Bergh and J. L\"ofstr\"om.
\newblock Interpolation spaces. An Introduction.
\newblock Springer-Verlag, Berlin, Heidelberg, New York, 1976.


%\bibitem[Bha97]{Bha97}
%R. Bhatia.
%\newblock Matrix analysis.
%\newblock Grad. Texts in Math., 169. Springer-Verlag, New York, 1997.

%%%\bibitem[Bia97]{Bia97}
%%%P. Biane.
%%%\newblock Free hypercontractivity.
%%%\newblock Comm. Math. Phys. 184 (1997), no. 2, 457--474.
%
\bibitem[BLM04]{BLM04}
D. Blecher and C. Le Merdy.
\newblock Operator algebras and their modules-an operator space approach.
\newblock London Mathematical Society Monographs. New Series, 30. Oxford Science Publications. The Clarendon Press, Oxford University Press, Oxford, 2004.
%%
%%%\bibitem[BlH95]{BlH95}
%%%W. R. Bloom and H. Heyer.
%%%\newblock Harmonic analysis of probability measures on hypergroups. 
%%%\newblock De Gruyter Studies in Mathematics, 20. Walter de Gruyter \& Co., Berlin, 1995. 
%%
%%\bibitem[Boc95]{Boc95}
%%F. P. Boca.
%%\newblock Ergodic actions of compact matrix pseudogroups on $C^*$-algebras. 
%%\newblock Recent advances in operator algebras (Orl\'eans, 1992). Ast\'erisque No. 232 (1995), 93--109. 

\bibitem[Bof25]{Bof25}
E. Boffy.
\newblock Positive contractive projections in Schatten Space.
\newblock Preprint, arXiv:2509.23701.

\bibitem[Bof26]{Bof26}%
E. Boffy.
\newblock On the structure of contractively decomposable projections on noncommutative $L^p$-spaces and Schatten spaces.
\newblock Preprint, arXiv:2607.11443.


%\bibitem[BHK17]{BHK17}
%M. Bohata, J. Hamhalter, O. F. K. Kalenda.
%\newblock Decompositions of preduals of $\JBW$ and $\JBW^*$ algebras. 
%\newblock J. Math. Anal. Appl. 446 (2017), no. 1, 18--37.


%%%\bibitem[BoD06]{BoD06}
%%%A. T. B\"ol\"ukbasi and T. Dereli.
%%%\newblock On the SU(3) Parametrization of Qutrits.
%%%\newblock J. Phys.: Conf. Ser. 36 (2006), 28--32

%\bibitem[Bos76]{Bos76}
%W. B\"os.
%\newblock Direct integrals of selfdual cones ans standard forms of von Neumann algebras.
%\newblock .


%\bibitem[Bou03]{Bou03}
%N. Bourbaki.
%\newblock Elements of mathematics. Topological vector spaces. Chapters 1–5. Transl. from
%the French by H. G. Eggleston and S. Madan.
%\newblock Berlin. Springer, softcover printing of the 1st. English edition of 1987 edition, 2003.

%%\bibitem[Bou19]{Bou19}%
%%N. Bourbaki.
%%\newblock \'El\'ements de math\'ematique. Th\'eories spectrales. Chapitres 1 et 2. (French) [Elements of mathematics. Spectral theories. Chapters 1 and 2]. Second edition.
%%\newblock Springer, Cham, 2019.
%%
%%%\bibitem[Bou04]{Bou04}
%%%N. Bourbaki.
%%%\newblock Integration. II. Chapters 7--9. Translated from the 1963 and 1969 French originals by Sterling K. Berberian. Elements of Mathematics (Berlin).
%%%\newblock Springer-Verlag, Berlin, 2004.
%%
%%%\bibitem[BCLY20]{BCLY20}
%%%M. Brannan, B. Collins, H. H. Lee and S.-G. Youn.
%%%\newblock Temperley-Lieb quantum channels. 
%%%\newblock Comm. Math. Phys. 376 (2020), no. 2, 795--839. 
%%
%%%\bibitem[BCV1]{BCV1}
%%%M. Brannan, A. Chirvasitu and A. Viselter.
%%%\newblock Actions, quotients and lattices of locally compact quantum groups. 
%%%\newblock Doc. Math. 25 (2020), 2553--2582.
%%
%%%\bibitem[Bra17]{Bra17}
%%%M. Brannan.
%%%\newblock Approximation properties for locally compact quantum groups. 
%%%\newblock Topological quantum groups, 185--232, Banach Center Publ., 111, Polish Acad. Sci. Inst. Math., Warsaw, 2017. 
%%
%%%\bibitem[BrR97]{BrR97}
%%%O. Bratteli and D. W. Robinson.
%%%\newblock Operator algebras and quantum statistical mechanics. 2.
%%%\newblock Texts Monogr. Phys., Springer-Verlag, Berlin, 1997.

%\bibitem[BKU78]{BKU78}
%R. Braun, W. Kaup and H. Upmeier.
%\newblock A holomorphic characterization of Jordan $C^*$-algebras.
%\newblock Math. Z. 161 (1978), no. 3, 277--290.



%%%\bibitem[BrN06]{BrN06}
%%%J. Brodzki and G. A. Niblo.
%%%\newblock Approximation properties for discrete groups. 
%%%\newblock C*-algebras and elliptic theory, 23--35, Trends Math., Birkh\"auser, Basel, 2006. 
%%
%%%\bibitem[BrO08]{BrO08}
%%%N. P. Brown and N. Ozawa.
%%%\newblock $C^*$-algebras and finite-dimensional approximations.
%%%\newblock Graduate Studies in Mathematics, 88. American Mathematical Society, Providence, RI, 2008.

%\bibitem[BuC92]{BuC92}
%L. J. Bunce and C.-H. Chu.
%\newblock Compact operations, multipliers and Radon-Nikod\'ym property in $\JB^*$-triples.
%\newblock Pacific J. Math. 153 (1992), no. 2, 249--265.




\bibitem[BFT12]{BFT12}
L. J. Bunce, B. Feely and R. M. Timoney.
\newblock Operator space structure of $JC^*$-triples and TROs, I. 
\newblock Math. Z. 270 (2012), no. 3-4, 961--982. 

\bibitem[BuT13a]{BuT13a}
L. J. Bunce and R. M. Timoney.
\newblock Universally reversible $JC^*$-triples and operator spaces. 
\newblock J. Lond. Math. Soc. (2) 88 (2013), no. 1, 271--293. 

\bibitem[BuT13b]{BuT13b}
L. Bunce 	and R. M. Timoney.
\newblock On the universal TRO of a $JC^*$-triple, ideals and tensor products. 
\newblock Q. J. Math. 64 (2013), no. 2, 327--340. 

\bibitem[BuT19]{BuT19}
L. J. Bunce and R. M. Timoney.
\newblock On decompositions of ternary rings of operators and their morphisms. 
\newblock Math. Proc. R. Ir. Acad. 119A (2019), no. 2, 173--186. 

\bibitem[CGRP14]{CGRP14}
M. Cabrera Garcia and A. Rodriguez Palacios.
\newblock Non-associative normed algebras. Vol. 1. The Vidav-Palmer and Gelfand-Naimark theorems.
\newblock Encyclopedia of Mathematics and its Applications, 154. Cambridge University Press, Cambridge, 2014.

\bibitem[CGRP18]{CGRP18}
M. Cabrera Garcia and A. Rodriguez Palacios.
\newblock Non-associative normed algebras. Vol. 2. Representation theory and the Zel'manov approach.
\newblock Encyclopedia of Mathematics and its Applications, 167. Cambridge University Press, Cambridge, 2018.

%%\bibitem[CaL93]{CaL93}
%%E. A. Carlen and E. H. Lieb.
%%\newblock Optimal hypercontractivity for Fermi fields and related noncommutative integration inequalities.
%%\newblock Comm. Math. Phys. 155 (1993), no. 1, 27--46.
%%
%%\bibitem[CaM14]{CaM14}
%%E. A. Carlen and J. Maas.
%%\newblock An analog of the 2-Wasserstein metric in non-commutative probability under which the fermionic Fokker-Planck equation is gradient flow for the entropy.
%%\newblock Comm. Math. Phys. 331 (2014), no. 3, 887--926.
%%
%%
%%%\bibitem[CL]{CL}
 %%%Valerio Capraro, Martino Lupini.
%%%\newblock Introduction to Sofic and Hyperlinear groups and Connes' embedding conjecture.
%%%\newblock .

\bibitem[Cal75]{Cal75}
B. Calvert.
\newblock Convergence sets in reflexive Banach Spaces.
\newblock Proc. Amer. Math. Soc. 47 no. 2 (1975), 423--428.
%%
%%%\bibitem[ChE]{ChE}
%%%M. D. Choi and E. G. Effros.
%%%\newblock Injectivity and operator spaces.
%%%\newblock J. Funct. Anal. 24 (1977), no. 2, 156--209.
%%
%%%\bibitem[Chd]{Chd}
%%%M. Choda.
%%%\newblock A Remark on the Normal Expectations.
%%%\newblock Proc. Japan Acad. 44 (1968), 462--466.
%%
%%%\bibitem[CKM21]{CKM21}
%%%P.-E. Caprace, M. Kalantar and N. Monod.
%%%\newblock A type I conjecture and boundary representations of hyperbolic groups.
%%%\newblock Preprint, arXiv:2110.00190.
%%
%%%\bibitem[Cas17]{Cas17}
%%%M. Caspers.
%%%\newblock Locally compact quantum groups. 
%%%\newblock Topological quantum groups, 153--184, Banach Center Publ., 111, Polish Acad. Sci. Inst. Math., Warsaw, 2017. 
%%
%%
%%%\bibitem[CaV22]{CaV22}
%%%M. Caspers and G. Vos.
%%%\newblock BMO spaces of $\sigma$-finite von Neumann algebras and Fourier-Schur multipliers on $\mathrm{SU}_q(2)$. 
%%%\newblock Studia Math. 262 (2022), no. 1, 45--91. 

%\bibitem[CFR01]{CFR01}
%C. M. Caves, C. A. Fuchs and P. Rungta.
%\newblock Entanglement of Formation of an Arbitrary State of Two Rebits.
%\newblock Found Phys Lett 14, 199--212 (2001).

%%%\bibitem[Cha1]{Cha1}
%%%J. A. Chávez-Domínguez.
%%%\newblock Completely $(q,p)$-mixing maps. 
%%%\newblock Illinois J. Math. 56 (2012), no. 4, 1169--1183.
%%
%%\bibitem[CYT19]{CYT19}
%%L. Chen, Y. Yang and W.-S. Tang.
%%\newblock Positive-partial-transpose square conjecture for $n = 3$.
%%\newblock Physical review A 99, 012337 (2019).
%%
%%\bibitem[CDPM23]{CDPM23}
%%G. Chiribella, K. R. Davidson, V. I. Paulsen and M. Rahaman.
%%\newblock Positive maps and entanglement in real Hilbert spaces.
%%\newblock Ann. Henri Poincar\'e 24 (2023), no. 12, 4139--4168.
%%
%%\bibitem[CJK09]{CJK09}
%%M-D. Choi, N. Johnston and D. W. Kribs.
%%\newblock The multiplicative domain in quantum error correction. 
%%\newblock J. Phys. A 42 (2009), no. 24, 245303, 15 pp. 
%%
%%%\bibitem[CST08]{CST08}
%%%T. Ceccherini-Silberstein, F. Scarabotti and F. Tolli.
%%%\newblock Harmonic analysis on finite groups. Representation theory, Gelfand pairs and Markov chains. 
%%%\newblock Cambridge Studies in Advanced Mathematics, 108. Cambridge University Press, Cambridge, 2008. 
%%
%%%\bibitem[ChE]{ChE}
%%%M. D. Choi and E. G. Effros.
%%%\newblock Injectivity and operator spaces.
%%%\newblock J. Funct. Anal. 24 (1977), no. 2, 156--209.
%%
%%%\bibitem[ChV99]{ChV99}
%%%Y. A. Chapovsky and L. I. Vainerman.
%%%\newblock Compact quantum hypergroups. 
%%%\newblock J. Operator Theory 41 (1999), no. 2, 261--289. 
%%
%%%\bibitem[Cha17]{Cha17}
%%%I. Chatterji.
%%%\newblock Introduction to the rapid decay property. 
%%%\newblock Around Langlands correspondences, 53--72, Contemp. Math., 691, Amer. Math. Soc., Providence, RI, 2017. 

%\bibitem[CDP11a]{CDP11a}%
%G. Chiribella, G. M. D'Ariano, and P. Perinotti.
%\newblock Informational derivation of quantum theory.
%\newblock Phys. Rev. A 84 (2011), 012311.




%%\bibitem[ChH1]{ChH1}
%%A. Chirvasitu and S. O. Hoche.
%%\newblock Ergodic actions of the compact quantum group $O_{-1}(2)$. 
%%\newblock Preprint, arXiv:1708.06457.
%%
%%%\bibitem[ChL23]{ChL23}
%%%M.-D. Choi and C.-K. Li.
%%%\newblock On unital qubit channels. 
%%%\newblock Preprint, arXiv:2301.01358.
%%
%%
%%\bibitem[CPSW00]{CPSW00}
%%P. Clement, B. de Pagter, F. A. Sukochev and H. Witvliet.
%%\newblock Schauder decomposition and multiplier theorems.
%%\newblock Studia Math. 138 (2000), no. 2, 135--163.

\bibitem[CoS70]{CoS70}
H. B. Cohen and F. E. Sullivan.
\newblock Projecting onto cycles in smooth, reflexive Banach spaces. 
\newblock Pacific J. Math. 34 (1970), 355--364. 


\bibitem[Con90]{Con90}
J. B. Conway.
\newblock A course in functional analysis. Second edition.
\newblock Graduate Texts in Mathematics, 96. Springer-Verlag, New York, 1990.
%%
%%%\bibitem[Coo10]{Coo10}
%%%T. Cooney.
%%%\newblock A Hausdorff-Young inequality for locally compact quantum groups.
%%%\newblock Internat. J. Math. 21 (2010), no. 12, 1619--1632.
%%
%%\bibitem[CYZ18]{CYZ18}
%%B. Collins, Z. Yin and P. Zhong.
%%\newblock The PPT square conjecture holds generically for some classes of independent states.
%%\newblock J. Phys. A 51 (2018), no. 42, 425301, 19 pp.
%%
%%
%%%\bibitem[CoT06]{CoT06}
%%%T. M. Cover and J. A. Thomas.
%%%\newblock Elements of information theory. Second edition.
%%%\newblock Wiley-Interscience [John Wiley \& Sons], Hoboken, NJ, 2006.
%%
%%\bibitem[CrN13]{CrN13}
%%J. Crann and M. Neufang.
%%\newblock Quantum channels arising from abstract harmonic analysis. 
%%\newblock J. Phys. A 46 (2013), no. 4, 045308, 22 pp. 
%%
%%%\bibitem[CrN22]{CrN22}
%%%J. Crann and M. Neufang.
%%%\newblock A non-commutative Fej\'er theorem for crossed products, the approximation property, and applications. 
%%%\newblock Int. Math. Res. Not. IMRN 2022, no. 5, 3571--3601. 
%%
%%
%%%\bibitem[Cra17]{Cra17}
%%%J. Crann.
%%%\newblock Amenability and covariant injectivity of locally compact quantum groups II. 
%%%\newblock Canad. J. Math. 69 (2017), no. 5, 1064--1086.
%%
%%%\bibitem[CKLT1]{CKLT1}
%%%J. Crann, D. W. Kribs, R. H. Levene and I. G. Todorov.
%%%\newblock State convertibility in the von Neumann algebra framework. 
%%%\newblock Comm. Math. Phys. 378 (2020), no. 2, 1123--1156. 
%%
%%%\bibitem[Con]{Con1}
%%%A. Connes.
%%%\newblock G\'eom\'etrie non commutative.
%%%\newblock .
%%%
%%
%%%\bibitem[Con]{Con2}
%%%A. Connes.
%%%\newblock Une classification des facteurs de type III.
%%%\newblock Ann. Sci. Ecole Norm. Sup. (4) 6 (1973), 133-252.
%%
%%%\bibitem[CPPR]{CPPR}
%%%M. Caspers, J. Parcet, M. Perrin and \'E. Ricard.
%%%\newblock Noncommutative de Leeuw theorems.
%%%\newblock Forum Math. Sigma 3 (2015), e21.

%\bibitem[Cho19]{Cho19}%
%K. Cho.
%\newblock Effectuses in Categorical Quantum Foundations.
%\newblock PhD thesis, Radboud University, 2019.

\bibitem[Chu12]{Chu12}
C.-H. Chu.
\newblock Jordan structures in geometry and analysis. 
\newblock Cambridge Tracts in Mathematics, 190. Cambridge University Press, Cambridge, 2012.


%%%\bibitem[Dae1]{Dae1}
%%%A. van Daele.
%%%\newblock Continuous crossed products and type III von Neumann algebras.
%%%\newblock London Mathematical Society Lecture Note Series, 31. Cambridge University Press, Cambridge-New York, 1978. 
%%
%%%\bibitem[DeJ1]{DeJ1}
%%%A. Defant and M. Junge.
%%%\newblock Marius Maximal theorems of Menchoff-Rademacher type in non-commutative $L_q$-spaces.
%%%\newblock J. Funct. Anal. 206 (2004), no. 2, 322--355. 
%%
%%
%%
%%
%%
%%%
%%%\bibitem[DP]{DP}
%%%B. de Pagter.
%%%\newblock Non-commutative Banach function spaces.
%%%\newblock Positivity, 197--227, Trends Math., Birkhäuser, Basel, 2007.
%%
%%%\bibitem[DP]{DP}
%%%De Pagter.
%%%\newblock Course von Neumann algebras pdf.
%%%\newblock .
%%
%%
%%%\bibitem[DLG2]{DLG2}
%%%K. de Leeuw and I. Glicksberg.
%%%\newblock The decomposition of certain group representations.
%%%\newblock J. Analyse Math. 15 (1965), 135--192.
%%
%%
%%%\bibitem[BrL1]{BrL1}
%%%Bruss.
%%%\newblock Quantum Information: From Foundations to Quantum Technology Applications.
%%%\newblock  Dagmar Bruss (Sous la direction de), Gerd Leuchs.
%%
%%\bibitem[CXY13]{CXY13}
%%Z. Chen, Q. Xu and Z. Yin.
%%\newblock Harmonic analysis on quantum tori.
%%\newblock Comm. Math. Phys. 322 (2013), no. 3, 755--805.
%%
%%%\bibitem[CJK09]{CJK09}
%%%M.-D. Choi, N. Johnston and D. W. Kribs.
%%%\newblock The multiplicative domain in quantum error correction. 
%%%\newblock J. Phys. A 42 (2009), no. 24, 245303, 15 pp.
 %%
%%\bibitem[Chr05]{Chr05}
%%M. Christandl.
%%\newblock Bipartite entanglement: A cryptographic point of view.
%%\newblock University of Cambridge, 2005.
%%
%%\bibitem[Chr12]{Chr12}
%%M. Christandl.
%%\newblock PPT square conjecture.
%%\newblock Banff International Research Station Workshop: Operator Structures in Quantum Information Theory, 2012.
%%
%%\bibitem[CMW19]{CMW19}
%%M. Christandl, A. M\"uller-Hermes, M. M. Wolf.
%%\newblock When do composed maps become entanglement breaking?
%%\newblock Ann. Henri Poincaré 20 (2019), no. 7, 2295--2322.
%%
%%%\bibitem[ChL02]{ChL02}
%%%C.-H. Chu and A. T.-M. Lau.
%%%\newblock Harmonic functions on groups and Fourier algebras. 
%%%\newblock Lecture Notes in Mathematics, 1782. Springer-Verlag, Berlin, 2002. 
%%
%%%\bibitem[CKPS1]{CKPS1}
%%%A. Church, D. W. Kribs, R. Pereira and S. Plosker.
%%%\newblock Private quantum channels, conditional expectations, and trace vectors. 
%%%\newblock Quantum Inf. Comput. 11 (2011), no. 9-10, 774--783. 

\bibitem[Cio90]{Cio90}%
I. Cioranescu.
\newblock Geometry of Banach Spaces, Duality Mappings and Nonlinear Problems.
\newblock Mathematics and Its Applications, vol. 62. Kluwer Academic Publishers, Dordrecht, 1990.


%\bibitem[CoS98]{CoS98}
%F. Cobos and T. Schonbek.
%\newblock On a theorem by Lions and Peetre about interpolation between a Banach space and its dual. 
%\newblock Houston J. Math. 24 (1998), no. 2, 325--344. 
 
%\bibitem[CoS03]{CoS03}
%J. H. Conway and D. Smith.
%\newblock On quaternions and octonions.
%\newblock A K Peters, Ltd, 2003.


\bibitem[CoL25]{CoL25}%
R. Correa da Silva and G. Lechner.
\newblock Inclusions of Standard Subspaces.
\newblock Comm. Math. Phys. 406 (2025), no. 11, art. 283.

%%%\bibitem[CDHPRRS20]{CDHPRRS20}
%%%S. X. Cui, D. Ding, X. Han, G. Penington, D. Ranard, B. C. Rayhaun, Z. Shangnan.
%%%\newblock Kitaev's quantum double model as an error correcting code. 
%%%\newblock Quantum 4, 331 (2020). 
%%
%%\bibitem[Cur99]{Cur99}
%%C. W. Curtis.
%%\newblock Pioneers of representation theory: Frobenius, Burnside, Schur, and Brauer.
%%\newblock Hist. Math., 15. American Mathematical Society, Providence, RI; London Mathematical Society, London, 1999.
%%
%%
%%\bibitem[CwJ84]{CwJ84}
%%M. Cwikel and S. Janson.
%%\newblock Interpolation of analytic families of operators.
%%\newblock Studia Math. 79 (1984), no. 1, 61--71.
%%
%%%\bibitem[DFW21]{DFW21}
%%%B. Das, U. Franz and X. Wang.
%%%\newblock Invariant Markov semigroups on quantum homogeneous spaces. 
%%%\newblock J. Noncommut. Geom. 15 (2021), no. 2, 531--580. 
%%
%%%\bibitem[DHS04]{DHS04}
%%%N. Datta, A. S. Holevo and Y. Suhov.
%%%\newblock A quantum channel with additive minimum output entropy.
%%%\newblock arXiv:quant-ph/0403072.
%%
%%%\bibitem[DHS06]{DHS06}
%%%N. Datta, A. S. Holevo and Y. Suhov.
%%%\newblock Additivity for transpose depolarizing channels.
%%%\newblock International Journal of Quantum Information 04 (2006), No. 01, pp. 85--98.
%%

\bibitem[DGT84]{DGT84}%
W. J. Davis, D. J. H. Garling, and N. Tomczak-Jaegermann.
\newblock The complex convexity of quasi-normed linear spaces.
\newblock J. Funct. Anal. 55 (1984), 110--150.

%%
%%%\bibitem[DaR1]{DaR1}
%%%M. Daws and V. Runde.
%%%\newblock Reiter's properties $(P_1)$ and $(P_2)$ for locally compact quantum groups.
%%%\newblock J. Math. Anal. Appl. 364 (2010), no. 2, 352--365.

%\bibitem[Daw21]{Daw21}%
%M. Daws.
%\newblock One-parameter isometry groups and inclusions between operator algebras.
%\newblock New York J. Math. 27 (2021), 164--204.


%%%\bibitem[Daw10]{Daw10}
%%%M. Daws.
%%%\newblock Multipliers, self-induced and dual Banach algebras.
%%%\newblock Dissertationes Math. 470 (2010), 62 pp.
%%
%%
%%
%%%\bibitem[Daw2]{Daw2}
%%%M. Daws.
%%%\newblock Representing multipliers of the Fourier algebra on non-commutative $L^p$ spaces.
%%%\newblock Canad. J. Math. 63 (2011), no. 4, 798--825.
%%
%%%\bibitem[Daw11]{Daw11}
%%%M. Daws.
%%%\newblock Multipliers of locally compact quantum groups via Hilbert $C^*$-modules.
%%%\newblock J. Lond. Math. Soc. (2) 84 (2011), no. 2, 385--407.
%%
%%%\bibitem[Daw12]{Daw12}
%%%M. Daws.
%%%\newblock Completely positive multipliers of quantum groups.
%%%\newblock Internat. J. Math. 23 (2012), no. 12, 1250132, 23 pp.
%%
%%%\bibitem[DFSW16]{DFSW16}
%%%M. Daws, P. Fima, A. Skalski and S. White.
%%%\newblock The Haagerup property for locally compact quantum groups.
%%%\newblock J. Reine Angew. Math. 711 (2016), 189--229. 
%%
%%%\bibitem[DSV1]{DSV1}
%%%M. Daws, A. Skalski and A. Viselter.
%%%\newblock Around property (T) for quantum groups. 
%%%\newblock Comm. Math. Phys. 353 (2017), no. 1, 69--118.
%%
%%%\bibitem[DCH]{DCH}
%%%J. de Canni\`ere and U. Haagerup.
%%%\newblock Multipliers of the Fourier algebras of some simple Lie groups and their discrete subgroups.
%%%\newblock Amer. J. Math. 107 (1985), no. 2, 455--500. 
%%
%%%\bibitem[Dav96]{Dav96}
%%%K. R. Davidson.
%%%\newblock $C^*$-algebras by example.
%%%\newblock Fields Inst. Monogr. 6. American Mathematical Society, Providence, RI, 1996.
%%
%%\bibitem[DeC17]{DeC17}
%%K. De Commer.
%%\newblock Actions of compact quantum groups. 
%%\newblock Topological quantum groups, 33--100, Banach Center Publ., 111, Polish Acad. Sci. Inst. Math., Warsaw, 2017. 
%%
%%\bibitem[DeC11]{DeC11}
%%K. De Commer.
%%\newblock On projective representations for compact quantum groups. 
%%\newblock J. Funct. Anal. 260 (2011), no. 12, 3596--3644. 
%%
%%\bibitem[DMN22]{DMN22}
%%K. De Commer, R. Martos and R. Nest.
%%\newblock Projective representation theory for compact quantum groups and the quantum Baum-Connes assembly map. 
%%\newblock Preprint, arXiv:2112.04365. 
%%
%%%\bibitem[DKSS1]{DKSS1}
%%%K. De Commer, P. Kasprzak, A. Skalski and P. M. Soltan.
%%%\newblock Quantum actions on discrete quantum spaces and a generalization of Clifford's theory of representations. 
%%%\newblock Israel J. Math. 226 (2018), no. 1, 475--503. 
%%
%%\bibitem[DeF93]{DeF93}
%%A. Defant and K. Floret.
%%\newblock Tensor norms and operator ideals.
%%\newblock North-Holland Mathematics Studies, 176. North-Holland Publishing Co., Amsterdam, 1993.
%%
%%%\bibitem[dJR1]{dJR1}
%%%M. de Jeu and J. Rozendaal.
%%%\newblock Disintegration of positive isometric group representations on $L^p$-spaces.
%%%\newblock Positivity 21 (2017), no. 2, 673--710.
%%
%%%\bibitem[DVD11a]{DVD11a}
%%%L. Delvaux and A. Van Daele.
%%%\newblock Algebraic quantum hypergroups. 
%%%\newblock Adv. Math. 226 (2011), no. 2, 1134--1167. 
 %%
%%%\bibitem[DVD11b]{DVD11b}
%%%L. Delvaux and A. Van Daele.
%%%\newblock Algebraic quantum hypergroups II. Constructions and examples. 
%%%\newblock Internat. J. Math. 22 (2011), no. 3, 407--434. 

\bibitem[DGZ93]{DGZ93}%
R. Deville, G. Godefroy, and V. Zizler.
\newblock Smoothness and Renormings in Banach Spaces.
\newblock Pitman Monographs and Surveys in Pure and Applied Mathematics, vol. 64. Longman Scientific \& Technical, Harlow, 1993.



%\bibitem[DeP07]{DeP07}
%B. de Pagter.
%\newblock Non-commutative Banach function spaces.
%\newblock Positivity, 197--227. Trends Math. Birkh\"auser Verlag, Basel, 2007.

%\bibitem[DL70]{DL70}%
%E. B. Davies and J. T. Lewis.
%\newblock An operational approach to quantum probability.
%\newblock Comm. Math. Phys. 17 (1970), 239--260.

%\bibitem[DeG13]{DeG13}
%J. Derezi\'nski and C. G\'erard.
%\newblock Mathematics of quantization and quantum fields.
%\newblock Cambridge Monogr. Math. Phys. Cambridge University Press, Cambridge, 2013.
%%
%%%\bibitem[Der85]{Der85}
%%%Y. Derriennic.
%%%\newblock Entropie, th\'eor\`emes limite et marches al\'eatoires.
%%%\newblock Publ. Inst. Rech. Math. Rennes [Publications of the Rennes Institute of Mathematical Research] Universit\'e de Rennes I, Institut de Recherche Math\'ematique de Rennes, Rennes, 1985.

\bibitem[Die75]{Die75}%
J. Diestel.
\newblock Geometry of Banach Spaces---Selected Topics.
\newblock Lecture Notes in Mathematics, vol. 485. Springer-Verlag, Berlin, 1975.

\bibitem[Die69]{Die69}%
J. Dieudonn\'e.
\newblock Foundations of Modern Analysis, enlarged ed.
\newblock Pure and Applied Mathematics, vol. 10-I. Academic Press, New York, 1969.


%\bibitem[DS04]{DS04}
%B. de Pagter and F. A. Sukochev.
%\newblock Differentiation of operator functions in non-commutative $L^p$-spaces. 
%\newblock J. Funct. Anal. 212 (2004), no. 1, 28--75. 

%\bibitem[DWS02]{DWS02}
%B. de Pagter, H. Witvliet, F. A. Sukochev.
%\newblock Double operator integrals.
%\newblock J. Funct. Anal. 192 (2002), no. 1, 52--111.




%%\bibitem[DRVV10]{DRVV10}
%%A. De Rijdt and N. Vander Vennet.
%%\newblock Actions of monoidally equivalent compact quantum groups and applications to probabilistic boundaries.
%%\newblock Ann. Inst. Fourier (Grenoble) 60 (2010), no. 1, 169--216.
%%
%%\bibitem[DJKRB06]{DJKRB06}
%%I. Devetak, M. Junge, C. King and M. B. Ruskai.
%%\newblock Multiplicativity of completely bounded $p$-norms implies a new additivity result. 
%%\newblock Comm. Math. Phys. 266 (2006), no. 1, 37--63. 
%%
%%%\bibitem[Dieu68]{Dieu2}
%%%J. Dieudonn\'e.
%%%\newblock \'El\'ements d'analyse. Tome II: Chapitres XII \`a XV. (French)
%%%\newblock Cahiers Scientifiques, Fasc. XXXI Gauthier-Villars, \'Editeur, Paris 1968 x+408 pp.
%%
%%\bibitem[DiV04]{DiV04}
%%T. Digernes and V. S. Varadarajan.
%%\newblock Models for the irreducible representation of a Heisenberg group.
%%\newblock Infin. Dimens. Anal. Quantum Probab. Relat. Top. 7 (2004), no. 4, 527--546.
%%

%\bibitem[Dix53]{Dix53}
%J. Dixmier.
%\newblock Formes lin\'eaires sur un anneau d'op\'erateurs. (French).
%\newblock Bull. Soc. Math. France 81 (1953), 9--39. 


%%
%%%\bibitem[Dix63]{Dix3}
%%%J. Dixmier.
%%%\newblock Traces sur les $C^*$-alg\`ebres. (French)
%%%\newblock Ann. Inst. Fourier (Grenoble) 13 (1963), no. fasc., fasc. 1, 219--262. 
%%
%%%\bibitem[Dix77]{Dix77}
%%%J. Dixmier.
%%%\newblock $C\sp*$-algebras. Translated from the French by Francis Jellett.
%%%\newblock North-Holland Mathematical Library, Vol. 15. North-Holland Publishing Co., Amsterdam-New York-Oxford, 1977. 

%\bibitem[Dix81]{Dix81}
%J. Dixmier.
%\newblock Von Neumann algebras. With a preface by E. C. Lance. Translated from the second French edition by F. Jellett
%\newblock North-Holland Mathematical Library, 27. North-Holland Publishing Co., Amsterdam-New York, 1981. 
%%
%%
%%
%%\bibitem[DJT95]{DJT95}
%%J. Diestel, H. Jarchow and A. Tonge.
%%\newblock Absolutely summing operators.
%%\newblock Cambridge Studies in Advanced Mathematics, 43. Cambridge University Press, 1995.
%%
%%\bibitem[DoS00]{DoS00}
%%P. G. Dodds and F. A. Sukochev.
%%\newblock Non-commutative bounded Vilenkin systems.
%%\newblock Math. Scand. 87 (2000), no. 1, 73--92.




%\bibitem[DoS24]{DoS24}
%P. G. Dodds, B. de Pagter and F. A. Sukochev.
%\newblock Noncommutative Integration and Operator Theory.
%\newblock Birkh\"auser Cham, Progress in Mathematics, 2024.

%\bibitem[DoS09]{DoS09}
%P. G. Dodds and F. A. Sukochev.
%\newblock Submajorisation inequalities for convex and concave functions of sums of measurable operators.
%\newblock Positivity 13 (2009), no. 1, 107--124.

%%\bibitem[DoS01]{DoS01}
%%P. G. Dodds and F. A. Sukochev.
%%\newblock Vilenkin bases in non-commutative $L^p$-spaces.
%%\newblock Geometric analysis and applications (Canberra, 2000), 30--41. Proc. Centre Math. Appl. Austral. Nat. Univ., 39
%%Australian National University, Centre for Mathematics and its Applications, Canberra, 2001.
%%
%%\bibitem[DFPS01]{DFPS01}
%%P. G. Dodds and S. V. Ferleger, B. de Pagter and F. A. Sukochev.
%%\newblock Vilenkin systems and generalized triangular truncation operator.
%%\newblock Integral Equations Operator Theory 40 (2001), no. 4, 403--435.

\bibitem[DoR07]{DoR07}
Z. Dong and Z.-J. Ruan.
\newblock Weak* exactness for dual operator spaces. 
\newblock J. Funct. Anal. 253 (2007), no. 1, 373--397. 



%%%\bibitem[Dou]{Dou}
%%%R. G. Douglas.
%%%\newblock Contractive projections on an $\mathcal{L}_{1}$ space.
%%%\newblock Pacific J. Math. 15 1965 443--462.


\bibitem[Edg85]{Edg85}%
G. A. Edgar.
\newblock Complex martingale convergence.
\newblock In: Banach Spaces (N. J. Kalton and E. Saab, eds.), Lecture Notes in Mathematics, vol. 1166, pp. 38--59. Springer-Verlag, Berlin, 1985.


%%%\bibitem[EFHN]{EFHN}
%%%T. Eisner, B. Farkas, M. Haase and R. Nagel.
%%%\newblock Operator theoretic aspects of ergodic theory.
%%%\newblock Graduate Texts in Mathematics, 272. Springer, Cham, 2015.

%\bibitem[EPV25]{EPV25}
%G. M. Escolanoa, A. M. Peralta and A. R. Villenaa.
%\newblock Preservers of operator commutativity.
%\newblock J. Math. Anal. Appl. 55 (2025), no. 2, 129796, 64pp.

%%
%%%\bibitem[EKR1]{EKR1}
%%%E. G. Effros, J. Kraus and Z.-J. Ruan.
%%%\newblock On two quantized tensor products. 
%%%\newblock Operator algebras, mathematical physics, and low-dimensional topology (Istanbul, 1991), 125--145, Res. Notes Math., 5, A K Peters, Wellesley, MA, 1993.
%%
%%\bibitem[Edw55]{Edw55}
%%R. E Edwards.
%%\newblock On factor functions. 
%%\newblock Pacific J. Math. 5 (1955), 367--378.

%\bibitem[Edw71]{Edw71}%
%C. M. Edwards.
%\newblock Classes of operations in quantum theory.
%\newblock Comm. Math. Phys. 20 (1971), 26--56.
 
%\bibitem[Edw72]{Edw72}%
%C. M. Edwards.
%\newblock The theory of pure operations.
%\newblock Comm. Math. Phys. 24 (1972), 260--288.

%\bibitem[Edw80]{Edw80}
%C. M. Edwards.
%\newblock On Jordan $W^*$-algebras.
%\newblock Bull. Sci. Math. (2) 104 (1980), no. 4, 393--403.

%%\bibitem[EdW82]{EdW82}
%%R. E. Edwards.
%%\newblock Fourier series. Vol. 2.
%%\newblock Grad. Texts in Math. 85, Springer-Verlag, New York-Berlin, 1982.

%%%\bibitem[EKR93]{EKR93}
%%%E. G. Effros, J. Kraus and Z.-J. Ruan.
%%%\newblock On two quantized tensor products. 
%%%\newblock Operator algebras, mathematical physics, and low-dimensional topology (Istanbul, 1991), 125--145, Res. Notes Math., 5, A K Peters, Wellesley, MA, 1993.

\bibitem[EOR01]{EOR01}
E. Effros, N. Ozawa and Z.-J. Ruan.
\newblock On injectivity and nuclearity for operator spaces.
\newblock Duke Math. J. 110 (2001), no. 3, 489--521.

%\bibitem[EfP03]{EfP03}
%E. Effros and M. Popa,
%\newblock Feynman diagrams and Wick products associated with $q$-Fock space.
%\newblock Proc. Natl. Acad. Sci. USA 100 (2003), no. 15, 8629--8633.


%%%\bibitem[EfR03]{EfR03}
%%%E. G. Effros and Z.-J. Ruan.
%%%\newblock Operator space tensor products and Hopf convolution algebras. 
%%%\newblock J. Operator Theory 50 (2003), no. 1, 131--156.
%%
%%%\bibitem[EfR90]{EfR90}
%%%E. G. Effros and Z.-J. Ruan.
%%%\newblock Multivariable multipliers for groups and their operator algebras. 
%%%\newblock Operator theory: operator algebras and applications, Part 1 (Durham, NH, 1988), 197--218,
%%%Proc. Sympos. Pure Math., 51, Part 1, Amer. Math. Soc., Providence, RI, 1990.
%%
%%\bibitem[EfR00]{EfR00}
%%E. Effros and Z.-J. Ruan.
%%\newblock Operator spaces.
%%\newblock Oxford University Press (2000).
%%
%%%\bibitem[EiW07]{EiW07}
%%%J. Eisert and M. Wolf.
%%%\newblock Gaussian quantum channels. 
%%%\newblock Quantum information with continuous variables of atoms and light, 23--42, Imp. Coll. Press, London, 2007. 
%%
%%%\bibitem[EHMS21]{EHMS21}
%%%C. Eltschka, M. Huber, Simon Morelli, and Jens Siewert.
%%%\newblock The shape of higher-dimensional state space: Bloch-ball analog for a qutrit.
%%%\newblock Quantum 5, 485 (2021).
%%
%%\bibitem[Eno77]{Eno77}
%%M. Enock.
%%\newblock Produit crois\'e d'une alg\`ebre de von Neumann par une alg\`ebre de Kac. (French) 
%%\newblock J. Funct. Anal. 26 (1977), no. 1, 16--47. 
%%
%%%\bibitem[EnS92]{EnS92}
%%%M. Enock and J.-M. Schwartz.
%%%\newblock Kac algebras and duality of locally compact groups. With a preface by Alain Connes. With a postface by Adrian Ocneanu.
%%%\newblock Springer-Verlag, Berlin, 1992. 
%%
%%%\bibitem[EnV96]{EnV96}
%%%M. Enock and L. Vainerman.
%%%\newblock Deformation of a Kac algebra by an abelian subgroup. 
%%%\newblock Comm. Math. Phys. 178 (1996), no. 3, 571--596. 
%%
%%%\bibitem[EfS]{EfS}
%%%E. Effros and E. St\o rmer.
%%%\newblock Positive projections and Jordan structure in operator algebras.
%%%\newblock Math. Scand. 45 (1979), no. 1, 127--138.
%%
%%%\bibitem[EOR1]{EOR1}
%%%E. Effros, N. Ozawa and Z.-J. Ruan.
%%%\newblock On injectivity and nuclearity for operator spaces.
%%%\newblock Duke Math. J. 110 (2001), no. 3, 489--521.

 
%\bibitem[EPV25b]{EPV25b}
%G. M. Escolano, A. M. Peralta and A. R. Villena.
%\newblock The Mackey-Gleason-Bunce-Wright problem for vector-valued measures on projections in a $\JBW^*$-algebra.
%\newblock Preprint, arXiv:2509.03213.

\bibitem[EsM23]{EsM23}%
Y. Estaremi and M. Mathieu.
\newblock On the range of TRO-conditional expectations.
\newblock Linear Algebra Appl. 670 (2023), 78--103.


%\bibitem[Ell66]{Ell66}%
%A. J. Ellis.
%\newblock Linear operators in partially ordered normed vector spaces.
%\newblock J. London Math. Soc. 41 (1966), 323--332.

%\bibitem[EvK98]{EvK98}
%D. E. Evans and Y. Kawahigashi.
%\newblock Quantum Symmetries on Operator Algebras.
%\newblock Oxford Mathematical Monographs. Oxford Science Publications. The Clarendon Press, Oxford University Press, New York, 1998.


%%%\bibitem[Eym]{Eym}
%%%P. Eymard.
%%%\newblock L'alg\`ebre de Fourier d'un groupe localement compact (French).
%%%\newblock  Bull. Soc. Math. France 92 (1964), 181--236.

\bibitem[FHHMPZ01]{FHHMPZ01}
M. Fabian, P. Habala, P. Hajek, V. Montesinos Santalucia, J. Pelant and V. Zizler.
\newblock Functional analysis and infinite-dimensional geometry.
\newblock CMS Books in Mathematics/Ouvrages de Math\'ematiques de la SMC, 8. Springer-Verlag, New York, 2001.

%\bibitem[FaK86]{FaK86}
%T. Fack and H. Kosaki.
%\newblock Generalized $s$-numbers of $\tau$-measurable operators.
%\newblock Pacific J. Math. 123 (1986), no. 2, 269--300.

%%%\bibitem[FHMV04]{FHMV04}
%%%M. Fannes, B. Haegeman, M. Mosnyi, D. Vanpeteghem.
%%%\newblock Additivity of minimal entropy output for a class of covariant channels.
%%%\newblock arXiv:quant-ph/0410195.


%\bibitem[FaK94]{FaK94}
%J. Faraut and A. Kor\'anyi.
%\newblock Analysis on symmetric cones.
%\newblock Oxford Math. Monogr. Oxford Sci. Publ. The Clarendon Press, Oxford University Press, New York, 1994.

%\bibitem[FeP20]{FeP20}
%F. J. Fernández-Polo.
%\newblock The minimax principle and related topics in the Jordan setting.
%\newblock Linear Algebra Appl. 604 (2020), 293--323.


%%\bibitem[Fol16]{Fol16}
%%G. B. Folland.
%%\newblock A course in abstract harmonic analysis. Second edition. 
%%\newblock Textbooks in Mathematics. CRC Press, Boca Raton, FL, 2016. 
%%
%%%\bibitem[FHS11]{FHS11}
%%%B. E. Forrest, H. H. Lee and E. Samei.
%%%\newblock Projectivity of modules over Fourier algebras. 
%%%\newblock Proc. Lond. Math. Soc. (3) 102 (2011), no. 4, 697--730.
%%
%%%\bibitem[FSS17]{FSS17}
%%%U. Franz, A. Skalski and P. M. Soltan.
%%%\newblock Introduction to compact and discrete quantum groups. 
%%%\newblock Topological quantum groups, 9--31, Banach Center Publ., 111, Polish Acad. Sci. Inst. Math., Warsaw, 2017. 
%%
%%%\bibitem[FrG06]{FrG06}
%%%U. Franz and R. Gohm.
%%%\newblock Random walks on finite quantum groups. 
%%%\newblock Quantum independent increment processes. II, 1--32, Lecture Notes in Math., 1866, Springer, Berlin, 2006. 

%%%\bibitem[FrS09]{FrS09}
%%%U. Franz and A. Skalski.
%%%\newblock On idempotent states on quantum groups. 
%%%\newblock J. Algebra 322 (2009), no. 5, 1774--1802. 
%%
%%\bibitem[Fre13]{Fre13}%[Fre13]
%%D. H. Fremlin.
%%\newblock Measure theory. Vol. 4. Topological measure spaces. Part I. Corrected second printing of the 2003 original.
%%\newblock Torres Fremlin, Colchester, 2006.
%%
%%
%%\bibitem[FrTS21]{FrTS21}
%%A. Freslon, F. Taipe and S. Wang.
%%\newblock Tannaka-Krein reconstruction and ergodic actions of easy quantum groups. 
%%\newblock Preprint, arXiv:2112.07506. 

\bibitem[Fri05]{Fri05}
Y. Friedman.
\newblock Physical applications of homogeneous balls. With the assistance of Tzvi Scarr.
\newblock Prog. Math. Phys., 40. Birkh\"auser Boston, Inc., Boston, MA, 2005.

%\bibitem[FrP22]{FrP22}
%Y. Friedman and A. M. Peralta.
%\newblock Representation of symmetry transformations on the sets of tripotents of spin and Cartan factors.
%\newblock Anal. Math. Phys. 12 (2022), no. 1, Paper No. 37, 52 pp.


\bibitem[FrB85b]{FrB85b}
Y. Friedman and B. Russo.
\newblock Structure of the predual of a $\JBW^*$-triple.
\newblock J. Reine Angew. Math. 356 (1985), 67--89.

%\bibitem[FrR92]{FrR92}
%Y. Friedman and B. Russo.
%\newblock Geometry of the dual ball of the spin factor. 
%\newblock Proc. London Math. Soc.  (3)65(1992), 142--174. 

%%%\bibitem[FuW07]{FuW07}
%%%M. Fukuda and M. M. Wolf.
%%%\newblock Simplifying additivity problems using direct sum constructions. 
%%%\newblock J. Math. Phys. 48 (2007), no. 7, 072101, 7 pp. 
%%

%\bibitem[Fur16]{Fur16}%
%R. Furber.
%\newblock Categorical Duality in Probability and Quantum Foundations.
%\newblock PhD thesis, Radboud University, 2016.

%\bibitem[GKR24]{GKR24}
%O. van Gaans, A. Kalauch and M. Roelands.
%\newblock Order theoretical structures in atomic JBW-algebras: disjointness, bands, and centres.
%\newblock Positivity 28 (2024), no. 1, Paper No. 10, 54 pp..





%%\bibitem[GPLS09]{GPLS09}
%%R. Garcia-Patr\'{o}n, S. Pirandola, S. Lloyd and J. H. Shapiro.
%%\newblock Reverse coherent information. 
%%\newblock Physical review letters, 102(21):210501, 2009. 
%%
%%%\bibitem[EfS2]{EfS2}
%%%E. Effros and E. St\o rmer.
%%%\newblock Jordan algebras of self-adjoint operators.
%%%\newblock Trans. Amer. Math. Soc. 127 1967 313--316.
%%
%%%\bibitem[FrB]{FrB}
%%%Y. Friedman and B. Russo.
%%%\newblock Solution of the contractive projection problem.
%%%\newblock J. Funct. Anal. 60 (1985), no. 1, 56--79.
%%
%%%\bibitem[Gol]{Gol}
%%%S. Goldstein.
%%%\newblock Conditional expectations in $L^p$-spaces over von Neumann algebras.
%%%\newblock  Quantum probability and applications, II (Heidelberg, 1984), 233--239, Lecture Notes in Math., 1136, Springer, Berlin, 1985.   
%%
%%
%%
%%
%%\bibitem[GJL18a]{GJL18a}
%%L. Gao, M. Junge and N. LaRacuente.
%%\newblock Capacity estimates via comparison with TRO channels.
%%\newblock Comm. Math. Phys. 364 (2018), no. 1, 83--121.
%%
%%\bibitem[GJL18b]{GJL18b}
%%L. Gao, M. Junge, Nicholas LaRacuente.
%%\newblock Capacity bounds via operator space methods.
%%\newblock J. Math. Phys. 59 (2018), no. 12, 122202, 17 pp.
%%
%%\bibitem[GJL20]{GJL20}
%%L. Gao, M. Junge and N. LaRacuente.
%%\newblock Fisher Information and Logarithmic Sobolev Inequality for Matrix Valued Functions.
%%\newblock Ann. Henri Poincar\'e 21 (2020), no. 11, 3409--3478.
%%
%%\bibitem[GJL20b]{GJL20b}
%%L. Gao, M. Junge and N. LaRacuente.
%%\newblock Relative entropy for von Neumann subalgebras. 
%%\newblock Internat. J. Math. 31 (2020), no. 6, 2050046, 35 pp.
%%
%%%\bibitem[GiL1]{GiL1}
%%%M. Girard and J. Levick.
%%%\newblock Twirling channels have minimal mixed-unitary rank. 
%%%\newblock Preprint, arXiv:2005.07056. 
%%
%%%\bibitem[GLLLPPW]{GLLLPPW}
%%%M. Girard, D. Leung, J. Levick, C.-K. Li, V. Paulsen, Y. T. Poon, J. Watrous.
%%%\newblock On the mixed-unitary rank of quantum channels. 
%%%\newblock Preprint, arXiv:2003.14405. 
%%
%%\bibitem[Gir15]{Gir15}
%%S. M. Girvin.
%%\newblock In Quantum machines: measurement and control of engineered quantum systems,
%%edited by M. H. Devoret, B. Huard, R. J. Schoelkopf, and L. F. Cugliandolo (Oxford University
%%Press, Oxford, 2015) Chap. 3.
%%\newblock 
%%
%%%\bibitem[GJP17]{GJP17}
%%%A. Gonz\'alez-P\'erez, M. Junge and J. Parcet.
%%%\newblock Smooth Fourier multipliers in group algebras via Sobolev dimension. 
%%%\newblock Ann. Sci. \'Ec. Norm. Sup\'er. (4) 50 (2017), no. 4, 879--925.
%%
%%\bibitem[Gos73]{Gos73}
%%J. Gosselin.
%%\newblock Almost everywhere convergence of Vilenkin-Fourier series.
%%\newblock Trans. Amer. Math. Soc. 185 (1973), 345--370 (1974).
%%
%%
%%%\bibitem[GSSS16]{GSSS16}
%%%S. K. Goyal, B. N. Simon, R. Singh and S. Simon.
%%%\newblock Geometry of the generalized Bloch sphere for qutrits.
%%%\newblock J. Phys. A 49 (2016), no. 16, 165203, 29 pp.
%%
%%%\bibitem[GMS15]{GMS15}
%%%V. P. Gupta, P. Mandayam and V. S. Sunder.
%%%\newblock The Functional Analysis of Quantum Information Theory. A collection of notes based on lectures by Gilles Pisier, K. R. Parthasarathy, Vern Paulsen and Andreas Winter. 
%%%\newblock Lecture Notes in Physics, 902. Springer, Cham, 2015.
		%%
%%%\bibitem[Gro93]{Gro93}
%%%L. Gross.
%%%\newblock Uniqueness of ground states for Schr\"odinger operators over loop groups.
%%%\newblock J. Funct. Anal. 112 (1993), no. 2, 373--441.		
		%%
%%\bibitem[Gro72]{Gro72}
%%L. Gross.
%%\newblock Existence and uniqueness of physical ground states.
%%\newblock J. Funct. Anal. 10 (1972), 52--109.		
		%%
%%\bibitem[GIN18]{GIN18}
%%L. Gyongyosi, S. Imre and H. V. Nguyen.
%%\newblock A Survey on Quantum Channel Capacities. 
%%\newblock IEEE Communications Surveys and Tutorials 20 (2018), no. 2, 1149--1205. 	
	%%
%%\bibitem[GuW15]{GuW15}
%%M. K. Gupta and M. M. Wilde.
%%\newblock Multiplicativity of completely bounded $p$-norms implies a strong converse for entanglement-assisted capacity. 
%%\newblock Comm. Math. Phys. 334 (2015), no. 2, 867--887. 	
	%%
%%%\bibitem[GoL1]{GoL1}
%%%S. Goldstein and L. E. Labuschagne.
%%%\newblock Composition operators on Haagerup $L^p$-spaces.
%%%\newblock Infin. Dimens. Anal. Quantum Probab. Relat. Top. 12 (2009), no. 3, 439--468.	
%%
%%%\bibitem[GuR]{GuR}
%%%S. Guerre. and Y. Raynaud.
%%%\newblock Sur les isom\'etries de $L^p(X)$ et le th\'eorème ergodique vectoriel. (French) [The isometries of $L^p(X)$ and the vector-valued ergodic theorem].
%%%\newblock Canad. J. Math. 40 (1988), no. 2, 360--391.
%%
%%%\bibitem[Haa]{Haa}
%%%U. Haagerup.
%%%\newblock Injectivity and decomposition of completely bounded maps.
%%%\newblock .

%\bibitem[HaS95]{HaS95}
%U. Haagerup and E. St\o rmer.
%\newblock Positive projections of von Neumann algebras onto JW-algebras.
%\newblock Proceedings of the XXVII Symposium on Mathematical Physics (Torun, 1994). Rep. Math. Phys. 36 (1995), no. 2-3, 317--330.

%%%\bibitem[Han]{Han}
%%%Kyung Hoon Han.
%%%\newblock Noncommutative $L_p$-space and operator system.
%%%\newblock .
 %%
%%%\bibitem[Hal13]{Hal13}
%%%B. Hall.
%%%\newblock Quantum Theory for Mathematicians.
%%%\newblock Springer New York, 2013. 
%%
%%\bibitem[HRS20]{HRS20}
%%E. P. Hanson, C. Rouz\'e and D. Stilck França.
%%\newblock Eventually entanglement breaking Markovian dynamics: structure and characteristic times.
%%\newblock Ann. Henri Poincar\'e 21 (2020), no. 5, 1517--1571.
%%
%%\bibitem[HaS21]{HaS21}
%%S. Hatui and P. Singla.
%%\newblock On Schur multiplier and projective representations of Heisenberg groups.
%%\newblock J. Pure Appl. Algebra   225 (2021), no. 11, Paper No. 106742, 16 pp.
%%
%%\bibitem[Hig01]{Hig01}
%%R. J. Higgs.
%%\newblock Projective representations of abelian groups.
%%\newblock J. Algebra 242 (2001), no. 2, 769--781.
%%
%%\bibitem[HLS81]{HLS81}
%%R. Hoegh-Krohn, M. B. Landstad and E. Stormer.
%%\newblock Compact ergodic groups of automorphisms. 
%%\newblock Ann. of Math. (2) 114 (1981), no. 1, 75--86. 
%%
%%
%%%\bibitem[HeR]{HeR}
 %%%C. W. Henson and Y. Raynaud.
%%%\newblock On the theory of $L_p(L_q)$-Banach lattices.
%%%\newblock  Positivity 11 (2007), no. 2, 201--230.
%%
%%%\bibitem[HJX]{HJX}
%%%U. Haagerup, M. Junge and Q. Xu.
%%%\newblock A reduction method for noncommutative $L_p$-spaces and applications.
%%%\newblock Trans. Amer. Math. Soc. 362 (2010), no. 4, 2125--2165.
%%
%%%\bibitem[Izu]{Izu}
%%%H. Izumi.
%%%\newblock Constructions of non-commutative $L^p$-spaces with a complex parameter arising from modular actions.
%%%\newblock Internat. J. Math. 8 (1997), no. 8, 1029--1066.
%%
%%
%%%\bibitem[Jun]{Jun}
%%%M. Junge.
%%%\newblock Fubini's theorem for ultraproducts of noncommmutative $L_p$-spaces.
%%%\newblock Canad. J. Math. 56 (2004), no. 5, 983--1021.
%%
%%%\bibitem[Jun2]{Jun2}
%%%M. Junge.
%%%\newblock Fubini's theorem for ultraproducts of noncommmutative $L_p$-spaces II.
%%%\newblock Preprint.
%%%
%%%\bibitem[JR1]{JR1}
%%%M. Junge and Z.-J. Ruan.
%%%\newblock Approximation properties for noncommutative $L_p$-spaces associated with discrete groups.
%%%\newblock Duke Math. J. 117 (2003), no. 2, 313--341.
%%
%%
%%%\bibitem[JoL]{JoL}
%%%W. B. Johnson and J. Lindenstrauss.
%%%\newblock Basic concepts in the geometry of {B}anach spaces, Handbook of the geometry of {B}anach spaces, {V}ol. {I}.
%%%\newblock  North-Holland, Amsterdam, 2001, pp.~1--84.
%%
%%%\bibitem[GL1]{GL1}
%%%S. Goldstein and J. M. Lindsay.
%%%\newblock KMS-symmetric Markov semigroups.
%%%\newblock Math. Z. 219 (1995), no. 4, 591--608.
%%
%%%\bibitem[GL2]{GL2}
%%%S. Goldstein and J. M. Lindsay.
%%%\newblock Markov semigroups KMS-symmetric for a weight.
%%%\newblock Math. Ann. 313 (1999), no. 1, 39--67.
%%
%%%\bibitem[Gra14]{Gra14}
%%%L. Grafakos.
%%%\newblock Classical Fourier analysis. Third edition. 
%%%\newblock Graduate Texts in Mathematics, 249. Springer, New York, 2014.
%%
%%\bibitem[HJX10]{HJX10}
%%U. Haagerup, M. Junge and Q. Xu.
%%\newblock A reduction method for noncommutative $L_p$-spaces and applications.
%%\newblock  Trans. Amer. Math. Soc. 362 (2010), no. 4, 2125--2165.
%%
%%%\bibitem[Haa78]{Haa78}
%%%U. Haagerup.
%%%\newblock An Example of a nonnuclear C*-Algebra, which has the metric approximation property.
%%%\newblock Invent. Math. 50 (1978/79), no. 3, 279--293.
%%
%%%\bibitem[Haa1]{Haa1}
%%%U. Haagerup.
%%%\newblock On the dual weights for crossed products of von Neumann algebras. II. Removing separability conditions.
%%%\newblock Math. Scand. 43 (1978/79), no. 1, 119--140.
%%
%%%\bibitem[Haa2]{Haa2}
%%%U. Haagerup.
%%%\newblock Operator-valued weights in von Neumann algebras. II.
%%%\newblock J. Funct. Anal. 33 (1979), no. 3, 339--361.
%
%\bibitem[Haa85]{Haa85}
%U. Haagerup.
%\newblock Injectivity and decomposition of completely bounded maps.
%\newblock Operator algebras and their connections with topology and ergodic theory (Bu\c{s}teni, 1983), 170--222, Lecture Notes in Math., 1132, Springer, Berlin, 1985.
%
%
%
 %
%%%\bibitem[Haa4]{Haa4}
%%%U. Haagerup.
%%%\newblock On the dual weights for crossed products of von Neumann algebras. I. Application of operator valued weights.
%%%\newblock Math. Scand. 43 (1978/79), no. 1, 99--118.
%%
%%\bibitem[HaM11]{HaM11}
%%U. Haagerup and M. Musat.
%%\newblock Factorization and dilation problems for completely positive maps on von Neumann algebras.
%%\newblock Comm. Math. Phys. 303 (2011), no. 2, 555--594.
%%
%%%\bibitem[Ham1]{Ham1}
%%%M. Hamana.
%%%\newblock Injective envelopes of dynamical systems.
%%%\newblock Operator algebras and operator theory (Craiova, 1989), 69--77, Pitman Res. Notes Math. Ser., 271, Longman Sci. Tech., Harlow, 1992.
%%
%%%\bibitem[Ham2]{Ham2}
%%%M. Hamana.
%%%\newblock Triple envelopes and Silov boundaries of operator spaces.
%%%\newblock Math. J. Toyama Univ. 22 (1999), 77--93.
%
%\bibitem[Han13]{Han13}
%K. H. Han.
%\newblock The predual of the space of decomposable maps from a $C^*$-algebra into a von Neumann algebra.
%\newblock J. Math. Anal. Appl. 402 (2013), no. 2, 463--476.
%
%
%%%\bibitem[Har99]{Har99}
%%%A. Harcharras.
%%%\newblock Fourier analysis, Schur multipliers on $S^p$ and non-commutative $\Lambda(p)$-sets.
%%%\newblock Studia Math. 137 (1999), no. 3, 203--260.
%%
%%\bibitem[Has99]{Has09}
%%M. B. Hastings.
%%\newblock A Counterexample to Additivity of Minimum Output Entropy. 
%%\newblock Nature Physics 5, 255 (2009).
%%
%%%\bibitem[HaS1]{HaS1}
%%%U. Haagerup and E. St\o rmer.
%%%\newblock Positive projections of von Neumann algebras onto JW-algebras.
%%%\newblock Proceedings of the XXVII Symposium on Mathematical Physics (Torun, 1994). Rep. Math. Phys. 36 (1995), no. 2-3, 317--330.


%\bibitem[Hay25]{Hay25}
%B. Hayes.
%\newblock Direct integrals of Hilbert spaces and von Neumann algebras.
%\newblock Internet.


%%%\bibitem{HeR79}%[HeR79]
%%%E. Hewitt and K. A. Ross.
%%%\newblock Abstract harmonic analysis. Vol. I.
%%%Structure of topological groups, integration theory, group
%%%representations. Second edition.
%%%\newblock Grundlehren der Mathematischen Wissenschaften, 115. Springer-Verlag, Berlin-New York, 1979.
%%
%%\bibitem[HeR70]{HeR70}
%%E. Hewitt and K. A. Ross.
%%\newblock Abstract harmonic analysis. Vol. II: Structure and analysis for compact groups. Analysis on locally compact Abelian groups.
%%\newblock Die Grundlehren der mathematischen Wissenschaften, Band 152, Springer-Verlag, New York-Berlin, 1970.
%%

%\bibitem[HaP89]{HaP89}
%U. Haagerup and G. Pisier.
%\newblock Factorization of analytic functions with values in noncommutative $L^1$-spaces and applications. 
%\newblock Canad. J. Math. 41 (1989), no. 5, 882--906. 

%%%\bibitem[HRS]{HRS}
%%%U. Haagerup, H.P. Rosenthal and F. A. Sukochev.
%%%\newblock Banach embedding properties of non-commutative $L^p$-spaces.
%%%\newblock Mem. Amer. Math. Soc. 163 (2003), no. 776.

\bibitem[Ham92]{Ham92}
M. Hamana.
\newblock Injective envelopes of dynamical systems.
\newblock Operator algebras and operator theory (Craiova, 1989), 69--77, Pitman Res. Notes Math. Ser., 271, Longman Sci. Tech., Harlow, 1992.

\bibitem[Ham99]{Ham99}
M. Hamana.
\newblock Triple envelopes and Silov boundaries of operator spaces.
\newblock Math. J. Toyama Univ. 22 (1999), 77--93.


%\bibitem[HKP20]{HKP20}
%J. Hamhalter, O. Kalenda and A. M. Peralta.
%\newblock Finite tripotents and finite $\JBW^*$-triples.
%\newblock J. Math. Anal. Appl. 490 (2020), no. 1, 124217, 65 pp..

%\bibitem[HKP23]{HKP23}
%J. Hamhalter, O. Kalenda and A. M. Peralta.
%\newblock Determinants in Jordan matrix algebras.
%\newblock Linear Multilinear Algebra 71 (2020), no. 6, 961--1002.

\bibitem[HKPP20]{HKPP20}
J. Hamhalter, O. F. K. Kalenda, A. M. Peralta and H. Pfitzner.
\newblock Measures of weak non-compactness in preduals of von Neumann algebras and $\JBW^*$-triples. 
\newblock J. Funct. Anal. 278 (2020), no. 1, 108300, 69 pp. 

%\bibitem[Han83]{Han83}
%H. Hanche-Olsen.
%\newblock On the structure and tensor products of JC-algebras.
%\newblock Can. J. Math. 35 (1983), 1059--1074.

%\bibitem[Han85]{Han85}
%H. Hanche-Olsen.
%\newblock JB algebras with tensor products are $C^*$ algebras.
%\newblock in H. Araki et al. (eds.), Operator Algebras and their Connections with Topology and Ergodic Theory, Lecture Notes in Mathematics 1132 (1985), 223--229.

\bibitem[HOS84]{HOS84}
H. Hanche-Olsen and E. St\o rmer.
\newblock Jordan operator algebras.
\newblock Monographs and Studies in Mathematics, 21. Pitman (Advanced Publishing Program), Boston, MA, 1984.

%\bibitem[HaW12]{HaW12}
%L. Hardy and W. K. Wootters.
%\newblock Limited Holism and Real-Vector-Space Quantum Theory.
%\newblock Found Phys 42 (2012), 454--473.


%\bibitem[Har01]{Har01}%
%L. Hardy.
%\newblock Quantum Theory From Five Reasonable Axioms.
%\newblock Preprint, arXiv:quant-ph/0101012.

%\bibitem[Har74]{Har74}
%L. A. Harris
%\newblock Bounded symmetric homogeneous domains in infinite dimensional spaces.
%\newblock Proceedings on Infinite Dimensional Holomorphy. Lecture Notes in Mathematics, vol 364. Springer, Berlin, Heidelberg, 1974.

%\bibitem[Har81]{Har81}
%L. A. Harris.
%\newblock A generalization of $C^*$-algebras. 
%\newblock Proc. London Math. Soc. (3) 42 (1981), no. 2, 331--361.




%%
%%%\bibitem[HKL1]{HKL1}
%%%J. A. Holbrook, D. W. Kribs and R. Laflamme.
%%%\newblock Noiseless subsystems and the structure of the commutant in quantum error correction. 
%%%\newblock Quantum Inf. Process. 2 (2003), no. 5, 381--419 (2004).


\bibitem[Hor87a]{Hor87a}
G. Horn.
\newblock Classification of $\JBW^*$-triples of type I. 
\newblock Math. Z. 196 (1987), no. 2, 271--291. 

\bibitem[HoN88]{HoN88}
G. Horn and E. Neher.
\newblock  Classification of continuous $\JBW^*$-triples. 
\newblock Trans. Amer. Math. Soc. 306 (1988), no. 2, 553--578. 

\bibitem[Hor87b]{Hor87b}
G. Horn.
\newblock Characterization of the predual and ideal structure of a $\JBW^*$-triple. 
\newblock Math. Scand. 61 (1987), no. 1, 117--133. 



%%%\bibitem[HLR1]{HLR1}
%%%R. Haydon, M. Levy and Y. Raynaud.
%%%\newblock Randomly normed spaces.
%%%\newblock Travaux en Cours [Works in Progress], 41. Hermann, Paris, 1991.

%\bibitem[HeI92]{HeI92}
%F. J. Herves and J.M. Isidro.
%\newblock Isometries and automorphisms of the spaces of spinors.
%\newblock Rev. Mat. Univ. Complutense Madrid 5 (1992), no. 2-3, 193--200.



%%%\bibitem[Hol02]{Hol02}
%%%A. S. Holevo.
%%%\newblock On entanglement-assisted classical capacity.
%%%\newblock J. Math. Phys. 43 (2002), no. 9, 4326--4333.
%%
%%\bibitem[Hol19]{Hol19}
%%A. S. Holevo.
%%\newblock Quantum Systems, Channels, Information: A Mathematical Introduction. 2nd edition
%%\newblock Gruyter Studies in Mathematical Physics, 16. De Gruyter, Berlin, 2019.
%%
%%\bibitem[Hol05]{Hol05}
%%A. S. Holevo.
%%\newblock Additivity conjecture and covariant channels.
%%\newblock International Journal of Quantum Information, Vol. 3, No. 1 (2005) 41--47.
%%
%%%\bibitem[Hol16]{Hol16}
%%%A. S. Holevo.
%%%\newblock On the constrained classical capacity of infinite-dimensional covariant quantum channels.
%%%\newblock Journal of Mathematical Physics 57, 015203 (2016).
%%
%%\bibitem[HoW01]{HoW01}
%%A. S. Holevo and R.F. Werner.
%%\newblock Evaluating capacities of bosonic gaussian
%%channels.
%%\newblock Phys. Rev. A 63 (2001), 032312.
%%
%%%\bibitem[Hor60]{Hor60}
%%%L. H\"ormander.
%%%\newblock Estimates for translation invariant operators in $\L^p$ spaces. 
%%%\newblock Acta Math. 104 (1960), 93--140. 
%%
%%\bibitem[HSR03]{HSR03}
%%M. Horodecki, P. W. Shor and M. B. Ruskai.
%%\newblock Entanglement breaking channels. 
%%\newblock Rev. Math. Phys. 15 (2003), no. 6, 629--641. 
%%
%%%\bibitem[HoR11]{HoR11}
%%%C. Houdayer and \'E. Ricard.
%%%\newblock Approximation properties and absence of Cartan subalgebra for free Araki-Woods factors.
%%%\newblock Adv. Math. 228 (2011), no. 2, 764--802.
%%
%%
%%%\bibitem[HNR10]{HNR10}
%%%Z. Hu, M. Neufang and Z.-J. Ruan.
%%%\newblock Multipliers on a new class of Banach algebras, locally compact quantum groups, and topological centres. 
%%%\newblock Proc. Lond. Math. Soc. (3) 100 (2010), no. 2, 429--458. 
%%
%%%\bibitem[HNR11]{HNR11}
%%%Z. Hu, M. Neufang and Z.-J. Ruan.
%%%\newblock Completely bounded multipliers over locally compact quantum groups. 
%%%\newblock Proc. Lond. Math. Soc. (3) 103 (2011), no. 1, 1--39. 
%%
%%%\bibitem[HuT22]{HuT22}
%%%L. Huang, Z. Liu and J. Wu.
%%%\newblock Quantum convolution inequalities on Frobenius von Neumann algebras. 
%%%\newblock Preprint, arXiv:2204.04401. 
%%
%%%\bibitem[HuT83]{HuT83}
%%%T. Huruya and J. Tomiyama.
%%%\newblock Completely bounded maps of $C^*$-algebras. 
%%%\newblock J. Operator Theory 10 (1983), no. 1, 141--152. 

%\bibitem[HvNVW16]{HvNVW16}
%T. Hyt\"onen, J. van Neerven, M. Veraar and L. Weis.
%\newblock Analysis in Banach spaces, Volume~I: Martingales and Littlewood-Paley theory.
%\newblock Springer, 2016. 
%%
%%%\bibitem[HvNVW2]{HvNVW2}
%%%T. Hyt\"onen, J. van Neerven, M. Veraar and L. Weis.
%%%\newblock Analysis in Banach spaces, Volume~II: Probabilistic Methods and Operator Theory. 
%%%\newblock Springer, 2018. 
%%
%%
%%%\bibitem[HoT]{HoT}
%%%T. Honda and W. Takahashi.
%%%\newblock Norm one linear projections and generalized conditional expectations in Banach spaces.
%%%\newblock Sci. Math. Jpn. 69 (2009), no. 3, 303--313.
%%
%%\bibitem[Hua99]{Hua99}
%%S.-Z. Huang.
%%\newblock Completeness of eigenvectors of group representations of operators whose Arveson spectrum is scattered.
%%\newblock Proc. Amer. Math. Soc. 127 (1999), no. 5, 1473--1482.
%%
%%\bibitem[Hu93]{Hu93}
%%Y. Z. Hu.
%%\newblock Hypercontractivit\'e pour les fermions, d'apr\`es Carlen-Lieb. (French) [Hypercontractivity for fermions, after Carlen and Lieb].
%%\newblock S\'eminaire de Probabilit\'es, XXVII, 86–96. Lecture Notes in Math., 1557. Springer-Verlag, Berlin, 1993.


%\bibitem[Ioc84]{Ioc84}
%B. Iochum.
%\newblock C\^ones autopolaires et alg\`ebres de Jordan. (French) [Self-dual cones and Jordan algebras].
%\newblock Lecture Notes in Mathematics, 1049. Springer-Verlag, Berlin, 1984.

%\bibitem[Ioc86]{Ioc86}
%B. Iochum.
%\newblock Nonassociative $L^p$-spaces.
%\newblock Pacific J. Math. 122 (1986), no. 2, 417--433.

\bibitem[Isi19]{Isi19}
J. M. Isidro.
\newblock Jordan triple systems in complex and functional analysis. 
\newblock Mathematical Surveys and Monographs, 243. American Mathematical Society, Providence, RI, 2019. 


%%\bibitem[Izu02]{Izu02}
%%M. Izumi.
%%\newblock Non-commutative Poisson boundaries and compact quantum group actions. 
%%\newblock Adv. Math. 169 (2002), no. 1, 1--57. 
%%
%%\bibitem[Izu04]{Izu04}
%%M. Izumi.
%%\newblock Non-commutative Poisson boundaries. 
%%\newblock Discrete geometric analysis, 69--81, Contemp. Math., 347, Amer. Math. Soc., Providence, RI, 2004. 
%%
%%%\bibitem[Jac1]{Jac1}
%%%A. Jacobs.
%%%\newblock The quantum $E(2)$ group. 
%%%\newblock PhD thesis, KU Leuven 2005. 

%\bibitem[JaH14]{JaH14}
%P. Janotta and H. Hinrichsen.
%\newblock Generalized probability theories: what determines the structure of quantum theory?.
%\newblock J. Phys. A: Math. Theor. 47 (2014), no. 32, 323001.

%%%\bibitem[Jen06]{Jen06}
%%%A. Jencova.
%%%\newblock A relation between completely bounded norms and conjugate channels. 
%%%\newblock Comm. Math. Phys. 266 (2006), no. 1, 65--70. 
%%
%%%\bibitem[Jew75]{Jew75}
%%%R. I. Jewett.
%%%\newblock Spaces with an abstract convolution of measures. 
%%%\newblock Advances in Math. 18 (1975), no. 1, 1--101.
%%
%%%\bibitem[JLJ16]{JLJ16}
%%%C. Jiang, Z. Liu and J. Wu.
%%%\newblock Noncommutative uncertainty principles. 
%%%\newblock J. Funct. Anal. 270 (2016), no. 1, 264--311. 
%%
%%\bibitem[JiY20]{JiY20}
%%R. Jin and Y. Yang.
%%\newblock Investigation of the PPT squared conjecture for high dimensions.
%%\newblock Preprint, arXiv:2010.15554, 2020.
%%
%%%\bibitem[JKLP1]{JKLP1}
%%%T. Jochym-O'Connor, D.W. Kribs, R. Laflamme, S. Plosker.
%%%\newblock Private quantum subsystems. 
%%%\newblock Physical Review Letters, 111, 030502 (2013). 
%%
%%%\bibitem[JKT1]{JKT1}
%%%N. Johnston, D. W. Kribs and C.-W. Teng.
%%%\newblock Operator algebraic formulation of the stabilizer formalism for quantum error correction. 
%%%\newblock Acta Appl. Math. 108 (2009), no. 3, 687--696. 
%%
%%%\bibitem[JoK1]{JoK1}
%%%N. Johnston and D. W. Kribs.
%%%\newblock Generalized multiplicative domains and quantum error correction. 
%%%\newblock Proc. Amer. Math. Soc. 139 (2011), no. 2, 627--639. 
%%
%%%\bibitem[Jon83]{Jon83}
%%%V. F. R. Jones.
%%%\newblock Index for subfactors.
%%%\newblock Invent. Math. 72 (1983), no. 1, 1--25.
%%
%%%\bibitem[Jon21]{Jon21}
%%%V. F. R. Jones.
%%%\newblock Planar algebras, I.
%%%\newblock New Zealand J. Math. 52 (2021 [2021--2022]), 1--107.
%%
%%%\bibitem[Jon00]{Jon00}
%%%V. F. R. Jones.
%%%\newblock The planar algebra of a bipartite graph.
%%%\newblock Knots in Hellas '98 (Delphi), 94--117, Ser. Knots Everything, 24, World Sci. Publ., River Edge, NJ, 2000.
%%

%\bibitem[Jor33]{Jor33}
%P. Jordan.
%\newblock  \"Uber die Multiplikation quantenmechanischer Gr\"o ßen.
%\newblock Z. Physik 80 (1933), 285--291.


%\bibitem[JNW34]{JNW34}
%P. Jordan, J. von Neumann and E. Wigner.
%\newblock On an algebraic generalization of the quantum mechanical formalism.
%\newblock Ann. of Math. (2) 35 (1934), no. 1, 29--64.



%%%\bibitem[Jun]{Jun}
%%%M. Junge.
%%%\newblock Fubini's theorem for ultraproducts of noncommmutative $L_p$-spaces.
%%%\newblock Canad. J. Math. 56 (2004), no. 5, 983--1021.
%%
%%%\bibitem[Jun1]{Jun1}
%%%M. Junge.
%%%\newblock Fubini's theorem for ultraproducts of noncommmutative $L_p$-spaces II.
%%%\newblock Preprint.
%%
%%%\bibitem[Jun99]{Jun99}
%%%M. Junge.
%%%\newblock Factorization theory for spaces of operators. 
%%%\newblock Institut for Matematik og Datalogi, Odense Universitet, 1999.

\bibitem[JRS05]{JRS05}
M. Junge, Z.-J. Ruan and D. Sherman.
\newblock A classification for 2-isometries of noncommutative $L_p$-spaces.
\newblock Israel J. Math. 150 (2005), 285--314.


%%
%%%\bibitem[JoW28]{JoW28}
%%%P. Jordan and E. Wigner.
%%%\newblock \"Uber das Paulische \"Aquivalenzverbot.
%%%\newblock Z. Physik 47 (1928), 631--651.
%%
%
%\bibitem[JMX06]{JMX06}
%M. Junge, C. Le Merdy and Q. Xu.
%\newblock $H^\infty$ functional calculus and square functions on noncommutative $L^p$-spaces.
%\newblock Ast\'erisque No. 305 (2006).
%%
%%%\bibitem[JNR09]{JNR09}
%%%M. Junge, M. Neufang and Z.-J. Ruan.
%%%\newblock A representation theorem for locally compact quantum groups. 
%%%\newblock Internat. J. Math. 20 (2009), no. 3, 377--400. 
%
%\bibitem[JuR04]{JuR04}
%M. Junge and Z.-J. Ruan.
%\newblock Decomposable maps on non-commutative $L_p$-spaces.
%\newblock Operator algebras, quantization, and noncommutative geometry, 355--381, Contemp. Math., 365, Amer. Math. Soc., Providence, RI, 2004.
%
%%\bibitem[JuP15]{JuP15}
%%M. Junge and C. Palazuelos.
%%\newblock Channel capacities via $p$-summing norms.
%%\newblock Adv. Math. 272 (2015), 350--398.
%%
%%%\bibitem[JuP16]{JuP16}
%%%M. Junge and C. Palazuelos.
%%%\newblock CB-norm estimates for maps between noncommutative Lp-spaces and quantum channel theory.
%%%\newblock Int. Math. Res. Not. IMRN 2016, no. 3, 875--925.
%%
%%\bibitem[JuP10]{JuP10}
%%M. Junge and J. Parcet.
%%\newblock Mixed-norm inequalities and operator space $L_p$ embedding theory.
%%\newblock Mem. Amer. Math. Soc. 203 (2010), no. 953.
%%
%%%\bibitem[JP2]{JP2}
%%%M. Junge and J. Parcet.
%%%\newblock Maurey's factorization theory for operator spaces.
%%%\newblock Math. Ann. 347 (2010), no. 2, 299--338.
%%
%%%\bibitem[JR1]{JR1}
%%%M. Junge and Z.-J. Ruan.
%%%\newblock Approximation properties for noncommutative $L_p$-spaces associated with discrete groups.
%%%\newblock Duke Math. J. 117 (2003), no. 2, 313--341.

\bibitem[JuX03]{JuX03}
M. Junge  and Q. Xu.
\newblock Noncommutative Burkholder/Rosenthal inequalities.
\newblock Ann. Probab. 31 (2003), no. 2, 948--995.

%%%\bibitem{JuX07}
%%%M. Junge and Q. Xu.
%%%\newblock Noncommutative maximal ergodic theorems.
%%%\newblock J. Amer. Math. Soc. 20 (2007), no. 2, 385--439.
%%
%%%\bibitem[KaP66]{KaP66}
%%%G. I. Kac and V. G. Paljutkin.
%%%\newblock Finite ring groups. (Russian)
%%%\newblock Trudy Moskov. Mat. Obsc. 15 (1966), 224--261. 
%%
%%%\bibitem[KaV83]{KaV83}
%%%V. A. Kaimanovich and A. M. Vershik.
%%%\newblock Random walks on discrete groups: boundary and entropy.
%%%\newblock Ann. Probab. 11 (1983), no. 3, 457--490.
%%
%%%\bibitem[Kan69]{Kan69}
%%%E. Kaniuth.
%%%\newblock Der Typ der regul\"aren Darstellung diskreter Gruppen. (German).
%%%\newblock Math. Ann. 182 (1969), 334--339.
%%
%%%\bibitem[Kan2]{Kan2}
%%%E. Kaniuth.
%%%\newblock Die Struktur der regul\"aren Darstellung lokalkompakter Gruppen mit invarianter Umgebungsbasis der Eins. (German).
%%%\newblock Math. Ann. 194 (1971), 225--248.
%%
%%\bibitem[Kad04]{Kad04}%[Kad04]
%%R. V. Kadison.
%%\newblock Non-commutative conditional expectations and their applications.
%%\newblock Operator algebras, quantization, and noncommutative geometry, 143--79. Contemp. Math., 365. American Mathematical Society, Providence, RI, 2004.
%%


%\bibitem[Kad51]{Kad51}%
%R. V. Kadison.
%\newblock A Representation Theory for Commutative Topological Algebra.
%\newblock Mem. Amer. Math. Soc. (1951), no. 7, 39 pp.
 

\bibitem[KaR97a]{KaR97a}
R. V. Kadison and J. R. Ringrose.
\newblock Fundamentals of the theory of operator algebras. Vol. I. Elementary theory. Reprint of the 1983 original. 
\newblock Graduate Studies in Mathematics, 15. American Mathematical Society, Providence, RI, 1997. 

\bibitem[KaR97b]{KaR97b}
R. V. Kadison and J. R. Ringrose.
\newblock Fundamentals of the theory of operator algebras. Vol. II. Advanced theory. Corrected reprint of the 1986 original.
\newblock Graduate Studies in Mathematics, 16. American Mathematical Society, Providence, RI, 1997.

\bibitem[Kan13]{Kan13}
M. Kaneda.
\newblock Ideal decompositions of a ternary ring of operators with predual. 
\newblock Pacific J. Math. 266 (2013), no. 2, 297--303. 

%\bibitem[Kau96]{Kau96}
%W. Kaup.
%\newblock On spectral and singular values in JB*-triples.
%\newblock Proc. Roy. Irish Acad. Sect. A 96 (1996), no. 1, 95--103.

%\bibitem[KLW24]{KLW24}
%S. Khatri, L. Lami and M. M. Wilde.
%\newblock Principles of Quantum Communication Theory: A Modern Approach. 
%\newblock Preprint, arXiv:2011.04672. %arXiv:2011.04672


%%%\bibitem[KNR14]{KNR14}
%%%M. Kalantar, M. Neufang and Z.-J. Ruan.
%%%\newblock Realization of quantum group Poisson boundaries as crossed products. 
%%%\newblock Bull. Lond. Math. Soc. 46 (2014), no. 6, 1267--1275. 

%\bibitem[KP21]{KP21}
%O. F. K. Kalenda and A. M. Peralta.
%\newblock Extension of isometries from the unit sphere of a rank-2 Cartan factor.
%\newblock Anal. Math. Phys. 11(2021), no. 15, 25 pp.

\bibitem[Kal03]{Kal03}
N. Kalton.
\newblock Quasi-Banach spaces.
\newblock Handbook of the geometry of Banach spaces, vol. 2, 1099--1130. North-Holland, Amsterdam, 2003.

%%%\bibitem[Kal01]{Kal01}
%%%A. A. Kalyuzhnyi.
%%%\newblock Conditional expectations on compact quantum groups and new examples of quantum hypergroups. 
%%%\newblock Methods Funct. Anal. Topology 7 (2001), no. 4, 49--68. 
%%
%%
%%\bibitem[KaT13]{KaT13}
%%E. Kaniuth and K. F. Taylor.
%%\newblock Induced representations of locally compact groups. 
%%\newblock Cambridge Tracts in Mathematics, 197. Cambridge University Press, Cambridge, 2013. 
%%
%%%\bibitem[KaL18]{KaL18}
%%%E. Kaniuth and A. T.-M. Lau.
%%%\newblock Fourier and Fourier-Stieltjes algebras on locally compact groups. 
%%%\newblock Mathematical Surveys and Monographs, 231. American Mathematical Society, Providence, RI, 2018. 
%%
%%\bibitem[Kar85]{Kar85}
%%G. Karpilovsky.
%%\newblock Projective representations of finite groups. 
%%\newblock Monographs and Textbooks in Pure and Applied Mathematics, 94. Marcel Dekker, Inc., New York, 1985.
%%
%%\bibitem[Kar87]{Karp87}
%%G. Karpilovsky.
%%\newblock The Schur multiplier.
%%\newblock London Math. Soc. Monogr. (N.S.), 2. The Clarendon Press, Oxford University Press, New York, 1987.
%%
%%\bibitem[Kar94]{Kar94}
%%G. Karpilovsky.
%%\newblock Group representations. Vol. 3.
%%\newblock North-Holland Math. Stud., 180. North-Holland Publishing Co., Amsterdam, 1994. 
%%
%
\bibitem[KaR02]{KaR02}
M. Kaur and Z.-J. Ruan.
\newblock Local properties of ternary rings of operators and their linking $C^*$-algebras.
\newblock J. Funct. Anal. 195 (2002), no. 2, 262--305.
%
%%%\bibitem[KMP1]{KMP1}
%%%M. Kennedy, N. A. Manor and V. I. Paulsen, Vern I.
%%%\newblock Composition of PPT maps. 
%%%\newblock Quantum Inf. Comput. 18 (2018), no. 5-6, 472--480. 
%%
%%\bibitem[KaS14]{KaS14}
%%P. Kasprzak, P. M. Soltan.
%%\newblock Embeddable quantum homogeneous spaces. 
%%\newblock J. Math. Anal. Appl. 411 (2014), no. 2, 574--591. 
%%
%%%\bibitem[Key02]{Key02}
%%%M. Keyl.
%%%\newblock Fundamentals of Quantum Information Theory. 
%%%\newblock Phys. Rep. 369, no. 5, 431--548 (2002).
%%
%%%\bibitem[KhW20]{KhW20}
%%%S. Khatri, M. M. Wilde.
%%%\newblock Principles of Quantum Communication Theory: A Modern Approach. 
%%%\newblock Preprint, arXiv:2011.04672. 
%%
%%\bibitem[KaS82]{KaS82}
%%Y. Katayama and G. Song.
%%\newblock Ergodic co-actions of discrete groups.
%%\newblock Math. Japon. 27 (1982), no. 2, 159--175.
%%
%%\bibitem[KMP18]{KMP18}
%%M. Kennedy, N. A. Manor, V. I. Paulsen.
%%\newblock Composition of PPT maps.
%%\newblock Quantum Inf. Comput. 18 (2018), no. 5-6, 472--480.
%%
%%\bibitem[Kin02]{Kin02}
%%C. King.
%%\newblock Additivity for unital qubit channels. 
%%\newblock J. Math. Phys. 43 (2002), no. 10, 4641--4653. 
%%
%%\bibitem[Kin03]{Kin03}
%%C. King.
%%\newblock Maximal $p$-norms of entanglement breaking channels. 
%%\newblock Quantum Inf. Comput. 3 (2003), no. 2, 186--190. 
 %%
%%%\bibitem[KiR01]{KiR01}
%%%C. King and M. B. Ruskai.
%%%\newblock Minimal entropy of states emerging from noisy quantum channels.
%%%\newblock IEEE Trans. Inform. Theory 47 (2001), no. 1, 192--209.

%\bibitem[Kin83]{Kin83}
%W. P. C. King.
%\newblock Semifinite traces on $\JBW$-algebras.
%\newblock Math. Proc. Cambridge Philos. Soc. 93 (1983), no. 3, 503--509.

%%%\bibitem[Kit02]{Kit02}
%%%A. Y. Kitaev.
%%%\newblock Topological quantum codes and anyons. 
%%%\newblock Quantum computation: a grand mathematical challenge for the twenty-first century and the millennium (Washington, DC, 2000), 267--272, Proc. Sympos. Appl. Math., 58, AMS Short Course Lecture Notes, Amer. Math. Soc., Providence, RI, 2002.
%%
%%%\bibitem[Kit03]{Kit03}
%%%A. Y. Kitaev.
%%%\newblock Fault-tolerant quantum computation by anyons. 
%%%\newblock Ann. Physics 303 (2003), no. 1, 2--30. 
%%
%%%\bibitem[Kit06]{Kit06}
%%%A. Y. Kitaev.
%%%\newblock Anyons in an exactly solved model and beyond. 
%%%\newblock Ann. Physics 321 (2006), no. 1, 2--111.
%%
%%%\bibitem[Kit10]{Kit10}
%%%A. Y. Kitaev.
%%%\newblock Topological phases and quantum computation. 
%%%\newblock Notes written by C. Laumann. Exact methods in low-dimensional statistical physics and quantum computing, 101--125, Oxford Univ. Press, Oxford, 2010.
%%
%%%\bibitem[KlR78]{KlR78}
%%%A. Klein and B. Russo.
%%%\newblock Sharp inequalities for Weyl operators and Heisenberg groups. 
%%%\newblock Math. Ann. 235 (1978), no. 2, 175--194. 
%%
%%\bibitem[Kle62]{Kle62}
%%A. Kleppner.
%%\newblock The structure of some induced representations.
%%\newblock Duke Math. J. 29 (1962), 555--572.
%%
%%\bibitem[Kle65]{Kle65}
%%A. Kleppner.
%%\newblock Multipliers on abelian groups.
%%\newblock Math. Ann. 158 (1965), 11--34.
%%
%%\bibitem[Kle74]{Kle74}%
%%A. Kleppner.
%%\newblock Continuity and measurability of multiplier and projective representations.
%%\newblock J. Funct. Anal. 17 (1974), 214--226.
%%
%%%\bibitem[KSSS19]{KSSS19}
%%%V. Kodiyalam, S. Sruthymurali, L. Sohan and V. S. Sunder.
%%%\newblock On a presentation of the spin planar algebra.
%%%\newblock Proc. Indian Acad. Sci. Math. Sci. 129 (2019), no. 2, Paper No. 27, 11 pp.
%%
%%
%%
%%\bibitem[KMS20]{KMS20}
%%P. Kopszak, M. Mozrzymas, M. Studzinski.
%%\newblock Positive maps from irreducibly covariant operators. 
%%\newblock J. Phys. A 53 (2020), no. 39, 395306, 33 pp.
%%
%%%\bibitem[KPC10]{KPC10}
%%%A. A. Kalyuzhnyi, G. B. Podkolzin, Y. A. Chapovsky.
%%%\newblock Harmonic analysis on a locally compact hypergroup. 
%%%\newblock Methods Funct. Anal. Topology 16 (2010), no. 4, 304--332. 
%%
%%%\bibitem[KrS16]{KrS16}
%%%J. Krajczok and P. M. Soltan.
%%%\newblock Center of the algebra of functions on the quantum group $\mathrm{SU}_q(2)$ and related topics. 
%%%\newblock Comment. Math. 56 (2016), no. 2, 251--272.
%%
%%%\bibitem[Kos1]{Kos1}
%%%R. P. Kostecki.
%%%\newblock $\mathrm{W}^*$-algebras and noncommutative integration.
%%%\newblock Preprint, 2014.
%%
%%
%%%\bibitem[KhS]{KhS}
%%%M. A. Khamsi and B. Sims.
%%%\newblock Ultra-methods in metric fixed point theory, Handbook of metric fixed point theory.
%%%\newblock Kluwer Acad. Publ., Dordrecht, 2001, pp.~177--199.
%%
%%%\bibitem[Kir]{Kir}
%%%E. Kirchberg.
%%%\newblock On nonsemisplit extensions, tensor products and exactness of group C*-algebras.
%%%\newblock {\em Invent. Math.} 112, no. 3: 449--489, 1993.
%%
%%%\bibitem[Kos2]{Kos2}
%%%H. Kosaki.
%%%\newblock Applications of the complex interpolation method to a von Neumann algebra: noncommutative $L^p$-spaces. 
%%%\newblock J. Funct. Anal. 56 (1984), no. 1, 29--78. 
%%
%%%\bibitem[Kos3]{Kos3}
%%%H. Kosaki.
%%%\newblock Positive cones and $L^p$-spaces associated with a von Neumann algebra. 
%%%\newblock J. Operator Theory 6 (1981), no. 1, 13--23. 
 %%
%%%\bibitem[Kir1]{Kir1}
%%%E. Kirchberg.
%%%\newblock On restricted perturbations in inverse images and a description of normalizer algebras in $C^*$-algebras. 
%%%\newblock J. Funct. Anal. 129 (1995), no. 1, 1--34. 
%%
%%
%%%\bibitem[KiK1]{KiK1}
%%%C. King and N. Koldan.
%%%\newblock Comparison of matrix norms on bipartite spaces. 
%%%\newblock J. Operator Theory 67 (2012), no. 2, 561--580. 
%%
%%%\bibitem[KrS18]{KrS18}
%%%J. Krajczok and P. M. Soltan.
%%%\newblock Compact quantum groups with representations of bounded degree. 
%%%\newblock J. Operator Theory 80 (2018), no. 2, 415--428.

%\bibitem[KPS82]{KPS82}
%S. G. Krein, Y. I. Petunn and E. M. Semenov.
%\newblock Interpolation of linear operators. Translated from the Russian by J. Szucs. 
%\newblock Translations of Mathematical Monographs, 54. American Mathematical Society, Providence, R.I., 1982. 


%%%\bibitem[Kre1]{Kre1}
%%%Kretschmann.
%%%\newblock Information Transfer through Quantum Channels.
%%%\newblock .

%\bibitem[Lab99]{Lab99}
%L. E. Labuschagne.
%\newblock Composition operators on non-commutative $L^p$-spaces.
%\newblock Exposition. Math. 17 (1999), no. 5, 429--467.

%%%\bibitem[Lab2]{Lab2}
%%%L. E. Labuschagne.
%%%\newblock A crossed product approach to Orlicz spaces.
%%%\newblock Proc. Lond. Math. Soc. (3) 107 (2013), no. 5, 965--1003.
%%
%%%\bibitem[Lab3]{Lab3}
%%%L. E. Labuschagne.
%%%\newblock Multipliers on noncommutative Orlicz spaces.
%%%\newblock Quaest. Math. 37 (2014), no. 4, 531--546.
%%
%%%\bibitem[Lac1]{Lac1}
%%%H. E. Lacey.
%%%\newblock The isometric theory of classical Banach spaces.
%%%\newblock Die Grundlehren der mathematischen Wissenschaften, Band 208. Springer-Verlag, New York-Heidelberg, 1974.
%%
%%%\bibitem[LaM]{LaM}
%%%L. E. Labuschagne and W. A. Majewski.
%%%\newblock Maps on noncommutative Orlicz spaces.
%%%\newblock Illinois J. Math. 55 (2011), no. 3, 1053--1081 (2013).
%%
%%%\bibitem[LaU1]{LaU1}
%%%A. Lau and A. Ulger.
%%%\newblock Some geometric properties on the Fourier and Fourier-Stieltjes algebras of locally compact groups, Arens regularity and related problems. 
%%%\newblock Trans. Amer. Math. Soc. 337 (1993), no. 1, 321--359. 
%%
%%\bibitem[LaP18]{LaP18}
%%F. Latr\'emoli\`ere and J. Packer.
%%\newblock Noncommutative solenoids.
%%\newblock New York J. Math. 24A (2018), 155--191.


\bibitem[Kos84]{Kos84}
H. Kosaki.
\newblock Applications of the complex interpolation method to a von Neumann algebra: noncommutative $L^p$-spaces. 
\newblock J. Funct. Anal. 56 (1984), no. 1, 29--78. 


%%%\bibitem[Kra1]{Kra1}
%%%J. Krajczok.
%%%\newblock Coamenability of type I locally compact quantum groups.
%%%\newblock Preprint, arXiv:2001.06740.
%%
%%\bibitem[Kur18]{Kur18}
%%Y. Kuramochi
%%\newblock Entanglement-breaking channels with general outcome operator algebras. 
%%\newblock J. Math. Phys. 59 (2018), no. 10, 102206, 15 pp.
%%
%%\bibitem[Kus05]{Kus05}
%%J. Kustermans.
%%\newblock Locally compact quantum groups. 
%%\newblock Lecture Notes in Math., 1865, Springer-Verlag, Berlin, 2005, 99--180. 
%%
%%\bibitem[KuV03]{KuV03}
%%J. Kustermans and S. Vaes.
%%\newblock Locally compact quantum groups in the von Neumann algebraic setting. 
%%\newblock Math. Scand. 92 (2003), no. 1, 68--92. 
%%
%%%\bibitem[KuV00]{KuV00}
%%%J. Kustermans and S. Vaes.
%%%\newblock The operator algebra approach to quantum groups. 
%%%\newblock Proc. Natl. Acad. Sci. USA 97 (2000), no. 2, 547--552. 
%%
%%\bibitem[LaG15]{LaG15}
%%L. Lami and V. Giovannetti.
%%\newblock Entanglement-breaking indices.
%%\newblock J. Math. Phys. 56 (2015), no. 9, 092201, 26 pp.
%%
%%\bibitem[LaG16]{LaG16}
%%L. Lami and V. Giovannetti.
%%\newblock Entanglement-saving channels.
%%\newblock J. Math. Phys. 57 (2016), no. 3, 032201, 34 pp.
%%
%%\bibitem[Lar71]{Lar71}
%%R. Larsen.
%%\newblock An introduction to the theory of multipliers.
%%\newblock Springer-Verlag, 1971.
%%
%%%\bibitem[Lev1]{Lev1}
%%%J. Levick.
%%%\newblock An Uncertainty Principle For Completely Positive Maps. 
%%%\newblock arXiv:1611.06352. 
%%
%%%\bibitem[LPW21]{LPW21}
%%%Z. Liu, S. Palcoux and J. Wu.
%%%\newblock Fusion bialgebras and Fourier analysis: analytic obstructions for unitary categorification. 
%%%\newblock Adv. Math. 390 (2021), Paper No. 107905, 63 pp. 
%%
%%%\bibitem[LeY22]{LeY22}
%%%H. H. Lee and S.-G. Youn.
%%%\newblock Quantum channels with quantum group symmetry. 
%%%\newblock  Comm. Math. Phys. 389 (2022), no. 3, 1303--1329.
%%
%%\bibitem[LeP95]{LeP95}
%%S. T. Lee and J. A. Packer.
%%\newblock Twisted group $C^*$-algebras for two-step nilpotent and generalized discrete Heisenberg groups.
%%\newblock J. Operator Theory 34 (1995), no. 1, 91--124.





%\bibitem[LMZ21]{LMZ21}
%C. Le Merdy and S. Zadeh.
%\newblock $\ell^1$-contractive maps on noncommutative $\L^p$-spaces.
%\newblock J. Operator Theory 85 (2021), no.2, 417--442.
%
%\bibitem[LMZ22a]{LMZ2a}
%C. Le Merdy and S. Zadeh.
%\newblock Surjective separating maps on noncommutative $L^p$-spaces.
%\newblock Math. Nachr. 295 (2022), no. 1, 175--188.
%
%\bibitem[LMZ22b]{LMZ2b}
%C. Le Merdy and S. Zadeh.
%\newblock On factorization of separating maps on noncommutative $L^p$-spaces.
%\newblock Indiana Univ. Math. J. 71 (2022), no. 5, 1967–-2000.
%
%
%
%
%%%\bibitem[LKP1]{LKP1}
%%%J. Levick, D. W. Kribs and R. Pereira.
%%%\newblock Quantum privacy and Schur product channels. 
%%%\newblock Rep. Math. Phys. 80 (2017), no. 3, 333--347. 
%%
%%%\bibitem[LJKLP1]{LJKLP1}
%%%J. Levick, T. Jochym-O'Connor, D. W. Kribs, R. Laflamme and R. Pereira.
%%%\newblock Private quantum subsystems and quasiorthogonal operator algebras. 
%%%\newblock J. Phys. A 49 (2016), no. 12, 125302, 14 pp. 
%%
%%%\bibitem[LPW21]{LPW21}
%%%Z. Liu, S. Palcoux and J. Wu.
%%%\newblock Fusion bialgebras and Fourier analysis: analytic obstructions for unitary categorification. 
%%%\newblock Adv. Math. 390 (2021), Paper No. 107905, 63 pp. 
%%
%%%\bibitem[Kri1]{Kri1}
%%%D. W. Kribs.
%%%\newblock A brief introduction to operator quantum error correction. 
%%%\newblock Operator theory, operator algebras, and applications, 27--34, Contemp. Math., 414, Amer. Math. Soc., Providence, RI, 2006. 
%%
%%%\bibitem[Kri2]{Kri2}
%%%D. W. Kribs.
%%%\newblock A quantum computing primer for operator theorists. 
%%%\newblock Linear Algebra Appl. 400 (2005), 147--167. 
%%
%%%\bibitem[KPLRN1]{KPLRN1}
%%%D. W. Kribs, R. Pereira, J. Levick, M. Rahaman and M. I. Nelson.
%%%\newblock Quantum complementarity and operator structures. 
%%%\newblock Quantum Inf. Comput. 19 (2019), no. 1-2, 67--83. 
%%
%%%\bibitem[KLPL1]{KLPL1}
%%%D. W. Kribs, R. Laflamme, D. Poulin, M. Lesosky.
%%%\newblock Operator quantum error correction. 
%%%\newblock Quantum Inf. Comput. 6 (2006), no. 4-5, 382--398. 

%\bibitem[Lam18]{Lam18}%
%L. Lami.
%\newblock Non-classical correlations in quantum mechanics and beyond.
%\newblock PhD thesis, Universitat Aut\`{o}noma de Barcelona, 2017.
 
%\bibitem[LPW18]{LPW18}%
%L. Lami, C. Palazuelos, and A. Winter.
%\newblock Ultimate data hiding in quantum mechanics and beyond.
%\newblock Comm. Math. Phys. 361 (2018), 661--708.

%\bibitem[LRTF21]{LRTF21}%
%L. Lami, B. Regula, R. Takagi, and G. Ferrari.
%\newblock Framework for resource quantification in infinite-dimensional general probabilistic theories.
%\newblock Phys. Rev. A 103 (2021), Paper No. 032424.

\bibitem[LRG07]{LRG07}%
B. Lemmens, B. Randrianantoanina, and O. van Gaans.
\newblock Second derivatives of norms and contractive complementation in vector-valued spaces.
\newblock Studia Math. 179 (2007), no. 2, 149--166.

%\bibitem[LeZ22]{LeZ22}
%C. Le Merdy and S. Zadeh.
%\newblock On factorization of separating maps on noncommutative $L^p$-spaces. 
%\newblock Indiana Univ. Math. J. 71 (2022), no. 5, 1967--2000. 

%\bibitem[LeZ25]{LeZ25}%
%C. Le Merdy and S. Zadeh.
%\newblock On the extension of positive maps to Haagerup non-commutative $L^p$-spaces.
%\newblock Forum Math. Sigma 13 (2025), Paper No. e65.


\bibitem[LRR09]{LRR09}
C. Le Merdy, \'E. Ricard and J. Roydor.
\newblock Completely 1-complemented subspaces of Schatten spaces.
\newblock Trans. Amer. Math. Soc. 361 (2009), no. 2, 849--887.

\bibitem[Lig76]{Lig76}%
E. Ligocka.
\newblock Levi forms, differential forms of type $(0,1)$ and pseudoconvexity in Banach spaces.
\newblock Ann. Polon. Math. 33 (1976), no. 1, 63--69.


%\bibitem[Li92]{Li92}
%B.-R. Li.
%\newblock Introduction to operator algebras.
%\newblock World Scientific Publishing Co., Inc., River Edge, NJ, 1992.

%\bibitem[LiP64]{LiP64}
%J. L. Lions and J. Peetre.
%\newblock Sur une classe d'espaces d'interpolation. (French). 
%\newblock Inst. Hautes \'Etudes Sci. Publ. Math. No. 19 (1964), 5--68. 

%%\bibitem[LoW22]{LoW22}
%%R. Longo and E. Witten.
%%\newblock A note on continuous entropy.
%%\newblock Preprint, arXiv:2202.03357.
%%
%%%\bibitem[Los84]{Los84}
%%%V. Losert.
%%%\newblock On tensor products of Fourier algebras.
%%%\newblock Arch. Math. (Basel) 43 (1984), no. 4, 370--372.
%%
%%\bibitem[LuP17]{LuP17}
%%A. Luczak and H. Podsedkowska.
%%\newblock Properties of Segal's entropy for quantum systems.
%%\newblock Internat. J. Theoret. Phys. 56 (2017), no. 12, 3783--3793.
%%
%%\bibitem[LuP19]{LuP19}
%%A. Luczak and H. Podsedkowska.
%%\newblock Mappings preserving Segal's entropy in von Neumann algebras. 
%%\newblock Ann. Acad. Sci. Fenn. Math. 44 (2019), no. 2, 769--789. 


%\bibitem[LPW23]{LPW23}
%A. Luczak, H. Podsedkowska and R. Wieczorek.
%\newblock Mappings preserving quantum Renyi's entropies in von Neumann algebras.
%\newblock arXiv:2302.02282.

%\bibitem[Lud83]{Lud83}%
%G. Ludwig.
%\newblock Foundations of Quantum Mechanics I.
%\newblock Theoretical and Mathematical Physics. Springer-Verlag, Berlin Heidelberg, 1983.


%\bibitem[Lud85]{Lud85}%
%G. Ludwig.
%\newblock An Axiomatic Basis for Quantum Mechanics: Volume~1, Derivation of {H}ilbert Space Structure.
%\newblock Springer-Verlag, Berlin Heidelberg, 1985.


%\bibitem[Lun18]{Lun18}
%A. Lunardi.
%\newblock Interpolation theory. Third edition. 
%\newblock Appunti. Scuola Normale Superiore di Pisa (Nuova Serie) [Lecture Notes. Scuola Normale Superiore di Pisa (New Series)], 16. Edizioni della Normale, Pisa, 2018. 
%%
%%%\bibitem[LWW17]{LWW17}
%%%Z. Liu, S. Wang and J. Wu.
%%%\newblock Young's inequality for locally compact quantum groups.
%%%\newblock J. Operator Theory 77 (2017), no. 1, 109--131.
%%
%%%\bibitem[LRR]{LRR}
%%%C. Le Merdy, \'E. Ricard and J. Roydor.
%%%\newblock Completely 1-complemented subspaces of Schatten spaces.
%%%\newblock Trans. Amer. Math. Soc. 361 (2009), no. 2, 849--887.
%%
%%\bibitem[Lus98]{Lus98}
%%F. Lust-Piquard.
%%\newblock Riesz transforms associated with the number operator on the Walsh system and the fermions.
%%\newblock J. Funct. Anal. 155 (1998), no. 1, 263--285.
%%
%%\bibitem[Mac58]{Mac58}%
%%G. W. Mackey.
%%\newblock Unitary representations of group extensions. I.
%%\newblock Acta Math. 99 (1958), 265--311.
%%
%%
%%%\bibitem[MaD98]{MaD98}
%%%A. Maes, Ann and A. Van Daele.
%%%\newblock Notes on compact quantum groups. 
%%%\newblock Nieuw Arch. Wisk. (4) 16 (1998), no. 1-2, 73--112. 
%%
%%%\bibitem[Maj06]{Maj06}
%%%S. Majid.
%%%\newblock What is a quantum group? 
%%%\newblock Notices Amer. Math. Soc. 53 (2006), no. 1, 30--31. 
%%
%%\bibitem[Mat21]{Mat21}
%%J. Matsuda.
%%\newblock Classification of Quantum Graphs on $M_2$ and their Quantum Automorphism Groups. 
%%\newblock Preprint, arXiv:2110.09085v2. 

%\bibitem[McC67]{McC67}
%C. A. McCarthy.
%\newblock $c^p$.
%\newblock Israel J. Math. 5 (1967), 249--271.

%\bibitem[McC78]{McC78}
%K. McCrimmon.
%\newblock Jordan algebras and their applications.
%\newblock Bull. Am. Math. Soc. 844 (1978), 612--627.

%\bibitem[McC04]{McC04}
%K. McCrimmon.
%\newblock A taste of Jordan algebras.
%\newblock Universitext. Springer-Verlag, New York, 2004.

\bibitem[Meg98]{Meg98}
R. E. Megginson.
\newblock An introduction to Banach space theory.
\newblock Graduate Texts in Mathematics, 183. Springer-Verlag, New York, 1998.

%\bibitem[Mor17]{Mor17}
%V. Moretti.
%\newblock Spectral theory and quantum mechanics. Mathematical foundations of quantum theories, symmetries and introduction to the algebraic formulation. Second edition.
%\newblock Unitext, 110. La Matematica per il 3+2. Springer, Cham, 2017.
%%
%%%\bibitem[Moo72]{Moo72}
%%%C. C. Moore.
%%%\newblock Groups with finite dimensional irreducible representations. 
%%%\newblock Trans. Amer. Math. Soc. 166 (1972), 401--410. 
%%
%%\bibitem[MSD17]{MSD17}
%%M. Mozrzymas, M. Studzinski and N. Datta.
%%\newblock Structure of irreducibly covariant quantum channels for finite groups. 
%%\newblock J. Math. Phys. 58 (2017), no. 5, 052204, 34 pp.
%%
%%%\bibitem[Naa15]{Naa15}
%%%P. Naaijkens.
%%%\newblock Kitaev's quantum double model from a local quantum physics point of view. 
%%%\newblock Advances in algebraic quantum field theory, 365--395, Math. Phys. Stud., Springer, Cham, 2015. 
%%
%%\bibitem[Mey95]{Mey95}
%%P.A. Meyer.
%%\newblock Quantum probability for probabilists. Second edition
%%\newblock Lecture Notes in Mathematics. Vol. 1538, Berlin–Heidelberg-New York: Springer, 1995.
%%
%%\bibitem[MiO14]{MiO14}
%%M. Miller and R. Olkiewicz.
%%\newblock Horodeckis criterion of separability of mixed states in von Neumann and $C^*$-algebras.
%%\newblock Infin. Dimens. Anal. Quantum Probab. Relat. Top. 17 (2014), no. 4, 1450028, 15 pp.
%%
%%\bibitem[Moo64]{Moo64}
%%C. C. Moore.
%%\newblock Extensions and low dimensional cohomology theory of locally compact groups. I.
%%\newblock Trans. Amer. Math. Soc. 113 (1964), 40--63.
%%
%%\bibitem[Moo76]{Moo76}
%%C. C. Moore.
%%\newblock Group extensions and cohomology for locally compact groups. III.
%%\newblock Trans. Amer. Math. Soc. 221 (1976), no. 1, 1--33.
%%
%%\bibitem[Mor62]{Mor62}
%%A. O. Morris.
%%\newblock The spin representation of the symmetric group.
%%\newblock Proc. London Math. Soc. (3) 12 (1962), 55--76.
%%
%%\bibitem[Mor73]{Mor73}
%%A. O. Morris.
%%\newblock Projective representations of Abelian groups.
%%\newblock J. London Math. Soc. (2) 7 (1973), 235--238.
%%
%%\bibitem[MST87]{MST87}
%%A. O. Morris, M. Saeed-Ul-Islam, E. Thomas.
%%\newblock Some projective representations of finite abelian groups.
%%\newblock Glasgow Math. J. 29 (1987), no. 2, 197–203.

\bibitem[Muj86]{Muj86}%
J. Mujica.
\newblock Complex Analysis in Banach Spaces: Holomorphic Functions and Domains of Holomorphy in Finite and Infinite Dimensions.
\newblock North-Holland Mathematics Studies, vol. 120. North-Holland, Amsterdam, 1986.

%\bibitem[Mul21]{Mul21}
%M. P. M\"uller.
%\newblock Probabilistic theories and reconstructions of quantum theory.
%\newblock SciPost Phys. Lect. Notes 28 (2021), 1--41.

%%\bibitem[NaT79]{NaT79}
%%Y. Nakagami ad M. Takesaki.
%%\newblock Duality for crossed products of von Neumann algebras.
%%\newblock Lecture Notes in Math., 731. Springer, Berlin, 1979.
%%
%%\bibitem[NaU61]{NaU61}
%%M. Nakamura and H. Umegaki.
%%\newblock A note on the entropy for operator algebras. 
%%\newblock Proc. Japan Acad. 37 (1961), 149--154. 
%%
%%%\bibitem[NSSFS08]{NSSFS08}
%%%C. Nayak, S. H. Simon, A. Stern, M. Freedman, S. Das Sarma.
%%%\newblock Nonabelian anyons and topological quantum computation. 
%%%\newblock Rev. Modern Phys. 80, 1083--1159 (2008). 
 %%
%%\bibitem[NaT79]{Nak79}
%%Y. Nakagami and M. Takesaki.
%%\newblock Duality for crossed products of von Neumann algebras. 
%%\newblock Lecture Notes in Mathematics, 731. Springer, Berlin, 1979. 

\bibitem[NeR03]{NeR03}
M. Neal and B. Russo.
\newblock Operator space characterizations of $C^*$-algebras and ternary rings.
\newblock Pacific J. Math. 209 (2003), no. 2, 339--364.


%%%\bibitem[NeT13]{NeT13}
%%%S. Neshveyev and L. Tuset.
%%%\newblock Compact quantum groups and their representation categories. 
%%%\newblock Cours Sp\'ecialis\'es [Specialized Courses], 20. Soci\'et\'e Math\'ematique de France, Paris, 2013. 
%%
%%%\bibitem[NSSS21]{NSSS21}
%%%M. Neufang, P. Salmi, A. Skalski and N. Spronk.
%%%\newblock Fixed points and limits of convolution powers of contractive quantum measures.
%%%\newblock Indiana Univ. Math. J. 70 (2021), no. 5, 1971--2009.
%%
%%%\bibitem[NeR11]{NeR11}
%%%S. Neuwirth and \'E. Ricard.
%%%\newblock Transfer of Fourier multipliers into Schur multipliers and sumsets in a discrete group.
%%%\newblock Canad. J. Math. 63 (2011), no. 5, 1161--1187.  %1001.5332
%%
%%
%%%\bibitem[NiC10]{NiC10}
%%%M. A. Nielsen and I. L. Chuang.
%%%\newblock Quantum computation and quantum information: 10th Anniversary Edition. 
%%%\newblock Cambridge University Press, Cambridge, 2010. 

%\bibitem[Nie20]{Nie20}
%G. Niestegge.
%\newblock Local tomography and the role of the complex numbers in quantum mechanics.
%\newblock Proc. A (2020) 476 (2238): 20200063.



%%%\bibitem[Mos1]{Mos1}
%%%Moslehian.
%%%\newblock A survey of the complemented subspaces problem.
%%%\newblock arXiv:math/0501048, 2005.
%%
%%%\bibitem[Ter]{Ter}
%%%M. Terp.
%%%\newblock Interpolation spaces between a von Neumann algebra and its predual. .
%%%\newblock J. Operator Theory 8 (1982), no. 2, 327--360. .
%%
%%%\bibitem[NeR]{NeR}
%%%M. Neal and B. Russo.
%%%\newblock Existence of contractive projections on preduals of JBW*-triples.
%%%\newblock Israel J. Math. 182 (2011), 293--331.
%%
%%%\bibitem[NO]{NO}
%%%P.W. Ng, and N. Ozawa.
%%%\newblock A characterization of completely 1-complemented subspaces of noncommutative $L^1$-spaces.
%%%\newblock Pacific J. Math. 205 (2002), 171--195.
 %%
%%\bibitem[Naz88]{Naz88}
%%M. L Nazarov.
%%\newblock An orthogonal basis in irreducible projective representations of the symmetric group.
%%\newblock Funktsional. Anal. i Prilozhen. 22 (1988), no. 1, 77--78 (Russian); translation in Funct. Anal. Appl. 22 (1988), no. 1, 66--68.
%%
%%\bibitem[Naz90]{Naz90}
%%M. L Nazarov.
%%\newblock Young's orthogonal form of irreducible projective representations of the symmetric group.
%%\newblock J. London Math. Soc. (2) 42 (1990), no. 3, 437--451.
%%
%%\bibitem[NeP24]{NeP24}
%%I. Nechita and S.-J. Park.
%%\newblock Random covariant quantum channels.
%%\newblock Preprint, arXiv:2403.03667.
 %%
%%\bibitem[OcS78]{OcS78}
%%W. Ochs and H. Spohn.
%%\newblock A characterization of the Segal entropy. 
%%\newblock Rep. Math. Phys. 14 (1978), no. 1, 75--87.
%%
%%%\bibitem[Oik1]{Oik1}
%%%T. Oikhberg. 
%%%\newblock Completely bounded and ideal norms of multiplication operators and Schur multipliers.
%%%\newblock Integral Equations Operator Theory 66 (2010), no. 3, 425--440.
%%
%%%\bibitem[Osa91]{Osa91}
%%%H. Osaka.
%%%\newblock Completely bounded maps between the preduals of von Neumann algebras. 
%%%\newblock Proc. Amer. Math. Soc. 111 (1991), no. 4, 961--965.
%%
%%
%%
%%%\bibitem[Oza]{Oza}
%%%N. Ozawa.
%%%\newblock About the QWEP conjecture.
%%%\newblock Internat. J. Math.15, no. 5: 501--530, 2004.
%%
%%%\bibitem[OhP93]{OhP93}
%%%M. Ohya and D. Petz.
%%%\newblock Quantum entropy and its use. 
%%%\newblock Texts and Monographs in Physics. Springer-Verlag, Berlin, 1993. 
%%
%%
%%%\bibitem[OlZ99]{OlZ99}
%%%R. Olkiewicz and B. Zegarlinski.
%%%\newblock Hypercontractivity in noncommutative $L_p$ spaces. 
%%%\newblock J. Funct. Anal. 161 (1999), no. 1, 246--285.
 %%
%%
%%\bibitem[OPT80]{OPT80}
%%D. Olesen, G. K. Pedersen and M. Takesaki.
%%\newblock Ergodic actions of compact abelian groups. 
%%\newblock J. Operator Theory 3 (1980), no. 2, 237--269. 
%%
%%\bibitem[Oml14]{Oml14}
%%T. A. Omland.
%%\newblock Primeness and primitivity conditions for twisted group $C^*$-algebras.
%%\newblock Math. Scand. 114 (2014), no. 2, 299--319.
%%
%%\bibitem[Oml15]{Oml15}
%%T. A. Omland.
%%\newblock $C^*$-algebras generated by projective representations of free nilpotent group.
%%\newblock J. Operator Theory 73 (2015), no. 1, 3--25.

%\bibitem[Pal01]{Pal01}
%T. W. Palmer.
%\newblock Banach algebras and the general theory of $*$-algebras. Vol. 2. *-algebras.
%\newblock Encyclopedia of Mathematics and its Applications, 79. Cambridge University Press, Cambridge, 2001.
%%
%%\bibitem[Pac87]{Pac87}
%%J. A. Packer.
%%\newblock $C^*$-algebras generated by projective representations of the discrete Heisenberg group.
%%\newblock J. Operator Theory 18 (1987), no. 1, 41--66.
%%
%%\bibitem[Pac08]{Pac08}
%%J. A. Packer.
%%\newblock Projective representations and the Mackey obstruction--a survey.
%%\newblock Group representations, ergodic theory, and mathematical physics: a tribute to George W. Mackey, 345–378. Contemp. Math., 449. American Mathematical Society, Providence, RI, 2008.
%%
%%\bibitem[Par69]{Par69}
%%K. R. Parthasarathy.
%%\newblock Multipliers on locally compact groups.
%%\newblock Lecture Notes in Math., Vol. 93. Springer-Verlag, Berlin-New York, 1969.

\bibitem[Pat18]{Pat18}
H. K. Pathak.
\newblock An Introduction to Nonlinear Analysis and Fixed Point Theory.
\newblock Springer, Singapore, 2018.


%%\bibitem[Pau02]{Pau02}
%%V. Paulsen.
%%\newblock Completely bounded maps and operator algebras.
%%\newblock Cambridge Univ. Press (2002).
%%
%%\bibitem[Ped18]{Ped18}
%%G. K. Pedersen.
%%\newblock $C^*$-algebras and their automorphism groups. Second edition. Edited and with a preface by S\o ren Eilers and Dorte Olesen.
%%\newblock Pure Appl. Math. (Amst.) Academic Press, London, 2018.

\bibitem[Pav91]{Pav91}%
M. Pavlovi\'c.
\newblock On the complex uniform convexity of quasi-normed spaces.
\newblock Math. Balkanica (N.S.) 5 (1991), no. 2, 92--98.


%\bibitem[PeS82]{PeS82}
%G. K. Pedersen and E. St\o rmer.
%\newblock Traces on Jordan algebras. 
%\newblock Canadian J. Math. 34 (1982), no. 2, 370--373.

%%\bibitem[Per96]{Per96}
%%A. Peres.
%%\newblock Separability criterion for density matrices.
%%\newblock Phys. Rev. Lett. 77 (1996), no. 8, 1413--1415.
%%
%%\bibitem[PSTW22]{PSTW22}
%%L.-E. Persson, F. Schipp, G. Tephnadze and F. Weisz.
%%\newblock An analogy of the Carleson-Hunt theorem with respect to Vilenkin systems.
%%\newblock J. Fourier Anal. Appl. 28 (2022), no. 3, Paper No. 48, 29 pp.
%%
%%\bibitem[Pet08]{Pet08}
%%D. Petz.
%%\newblock Quantum information theory and quantum statistics. 
%%\newblock Theoretical and Mathematical Physics. Springer-Verlag, Berlin, 2008.
%%
%%\bibitem[Pet01]{Pet01}
%%D. Petz.
%%\newblock Entropy, von Neumann and the von Neumann entropy. 
%%\newblock John von Neumann and the foundations of quantum physics (Budapest, 1999), 83--96, Vienna Circ. Inst. Yearb., 8, Kluwer Acad. Publ., Dordrecht, 2001. 
%%
%%%\bibitem[PeT1]{PeT1}
%%%G. K. Pedersen and M. Takesaki.
%%%\newblock The operator equation $THT=K$.
%%%\newblock Proc. Amer. Math. Soc. 36 (1972), 311--312.


%\bibitem[PeS25]{PeS25}
%A. M. Peralta and P. Saavedra.
%\newblock Identifying $\JBW^*$-algebras through their spheres of positive elements.
%\newblock Preprint, arXiv:2505.03287.


%%%\bibitem[Pic1]{Pic1}
%%%M. A. Picardello.
%%%\newblock Locally compact unimodular groups with atomic duals. (Italian summary).
%%%\newblock Rend. Sem. Mat. Fis. Milano 48 (1978), 197--216 (1980).
%%
%%%\bibitem[Pie]{Pie}
%%%J.-P. Pier.
%%%\newblock Amenable locally compact groups.
%%%\newblock Pure and Applied Mathematics (New York). A Wiley-Interscience Publication. John Wiley \& Sons, Inc., New York, 1984.
%%
%%\bibitem[Pie66]{Pie66}
%%A. Pietsch.
%%\newblock Absolut $p$-summierende Abbildungen in normierten R\"aumen.
%%\newblock Studia Math. 28 (1966), 333--353.
%
\bibitem[Pis98]{Pis98}
G. Pisier.
\newblock Non-commutative vector valued $L_p$-spaces and completely $p$-summing maps.
\newblock Ast\'erisque, 247, 1998.

\bibitem[Pis03]{Pis03}
G. Pisier.
\newblock Introduction to operator space theory.
\newblock Cambridge University Press, Cambridge, 2003.
%%
%%%\bibitem[Pis3]{Pis3}
%%%G. Pisier.
%%%\newblock Regular operators between non-commutative $L_p$-spaces.
%%%\newblock Bull. Sci. Math. 119 (1995), no. 2, 95--118.
%%
%%%\bibitem[Pis7]{Pis7}
%%%G. Pisier.
%%%\newblock Complex interpolation between Hilbert, Banach and operator spaces.
%%%\newblock Mem. Amer. Math. Soc. 208 (2010), no. 978.
%%%
%%%\bibitem[Pis-IHP]{Pis-IHP}
%%%G. Pisier.
%%%\newblock Martingales in {B}anach spaces (in connection with {T}ype and
  %%%{C}otype).
%%%\newblock 2011, Course IHP.
%%
%%\bibitem[Pis12]{Pis12}
%%G. Pisier.
%%\newblock Completely co-bounded Schur multipliers.
%%\newblock Oper. Matrices 6 (2012), no. 2, 263--270.
%
%\bibitem[Pis20]{Pis20}
%G. Pisier.
%\newblock Tensor Products of $C^*$-Algebras and Operator Spaces.Tensor products of $C^*$-algebras and operator spaces--the Connes-Kirchberg problem.
%\newblock London Math. Soc. Stud. Texts, 96. Cambridge University Press, Cambridge, 2020.
%

\bibitem[PiX03]{PiX03}
G. Pisier and Q. Xu.
\newblock Non-commutative $L^p$-spaces.
\newblock 1459--1517 in Handbook of the Geometry of Banach Spaces, Vol. II, edited by W.B. Johnson and J. Lindenstrauss, Elsevier (2003).

%\bibitem[Pla23]{Pla23}
%M. Pl\'avala.
%\newblock General probabilistic theories: An introduction.
%\newblock Physics Reports, Volume 1033, 7 September 2023, Pages 1--64.

\bibitem[PlR19]{PlR19}
R. Pluta and B. Russo.
\newblock On the structure of weakly closed ternary rings of operators. 
\newblock Math. Proc. R. Ir. Acad. 119A (2019), no. 2, 77--91. 


%\bibitem[PlR94]{PlR94}
%R. J. Plymen and P. L. Robinson.
%\newblock Spinors in Hilbert space. 
%\newblock Cambridge Tracts in Mathematics, 114. Cambridge University Press, Cambridge, 1994. 
%%
%%%\bibitem[Pod21]{Pod21}
%%%H. Podsedkowska.
%%%\newblock Strong subadditivity of quantum mechanical entropy for semifinite von Neumann algebras.
%%%\newblock Studia Math. 257 (2021), no. 1, 71--85.
%%
%%\bibitem[Pra11]{Pra11}
%%A. Prasad.
%%\newblock An easy proof of the Stone-von Neumann-Mackey theorem.
%%\newblock Expo. Math. 29 (2011), no. 1, 110--118.
%%
%%\bibitem[PSV10]{PSV10}
%%A. Prasad, I. Shapiro and M. K. Vemuri.
%%\newblock Locally compact abelian groups with symplectic self-duality.
%%\newblock Adv. Math. 225 (2010), no. 5, 2429--2454.
%%
%%
%%%\bibitem[QuS1]{QuS1}
%%%J. Quaegebeur and M. Sabbe.
%%%\newblock Isometric coactions of compact quantum groups on compact quantum metric spaces. 
%%%\newblock Proc. Indian Acad. Sci. Math. Sci. 122 (2012), no. 3, 351--373. 
%%
%%%\bibitem[PoS1]{PoS1}
%%%D. Potapov and F. Sukochev.
%%%\newblock The Haar system in the preduals of hyperfinite factors.
%%%\newblock Canad. Math. Bull. 54 (2011), no. 2, 347--363.
%%
%%%\bibitem[Pru]{Pru}
%%%Bebe Prunaru.
%%%\newblock On the Choi-Effros Multiplication.
%%%\newblock Arxiv.
%%
%%%\bibitem[Rand1]{Rand1}
%%%B. Randrianantoanina.
%%%\newblock On isometric stability of complemented subspaces of $L_p$.
%%%\newblock Israel J. Math. 113 (1999), 45--60.
%%
%%%\bibitem[Rand]{Rand2}
%%%B. Randrianantoanina.
%%%\newblock Norm-one projections in Banach spaces.
%%%\newblock International Conference on Mathematical Analysis and its Applications (Kaohsiung, 2000). Taiwanese J. Math. 5 (2001), no. 1, 35--95.
%%
%%%\bibitem[Rand3]{Rand3}
%%%B. Randrianantoanina.
%%%\newblock Contractive projections in Orlicz sequence spaces.
%%%\newblock Abstr. Appl. Anal. 2004, no. 2, 133--145.
%%
%%%\bibitem[Ray1]{Ray1}
%%%Raynaud, Yves.
%%%\newblock On ultrapowers of non commutative $L_p$ spaces.
%%%\newblock J. Operator Theory 48 (2002), no. 1, 41--68.
%%
%%
%%
%%\bibitem[RJV18]{RJV18}
%%M. Rahaman, S. Jaques and V. I. Paulsen.
%%\newblock Eventually entanglement breaking maps. 
%%\newblock J. Math. Phys. 59 (2018), no. 6, 062201, 11 pp. 
%%
%%\bibitem[Rah17]{Rah17}
%%M. Rahaman.
%%\newblock Multiplicative properties of quantum channels. 
%%\newblock J. Phys. A 50 (2017), no. 34, 345302, 26 pp. 
%%
%%\bibitem[RKSE10]{RKSE10}
%%P. Raynal, A. Kalev, J. Suzuki, and B.-G. Englert.
%%\newblock Encoding many qubits in a rotor.
%%\newblock Phys. Rev. A 81, 052327 (2010).
%%

\bibitem[Ray03]{Ray03}
Y. Raynaud.
\newblock $L_p$-spaces associated with a von Neumann algebra without trace: a gentle introduction via complex interpolation.
\newblock  Trends in Banach spaces and operator theory (Memphis, TN, 2001), 245--273, Contemp. Math., 321, Amer. Math. Soc., Providence, RI, 2003. 
%%
%%%\bibitem[Ray2]{Ray3}
%%%Y. Raynaud.
%%%\newblock The range of a contractive projection in $L_p(H)$.
%%%\newblock Rev. Mat. Complut. 17 (2004), no. 2, 485--512. 
%%
%%%\bibitem[Ray4]{Ray4}
%%%Y. Raynaud.
%%%\newblock Sur les sous-espaces de $L^p(L^q)$. (French) [On the subspaces of $L^p(L^q)$].
%%%\newblock S\'eminaire d'Analyse Fonctionelle 1984/1985, 49--71, Publ. Math. Univ. Paris VII, 26, Univ. Paris VII, Paris, 1986.
%%
%%%\bibitem[Ray5]{Ray5}
%%%Y. Raynaud.
%%%\newblock Sous-espaces $l^r$ et g\'eom\'etrie des espaces $L^p(L^q)$ et $L^\phi$. (French) [$l^r$ subspaces and geometry of $L^p(L^q)$ and $L^\phi$ spaces].
%%%\newblock C. R. Acad. Sci. Paris S\'er. I Math. 301 (1985), no. 6, 299--302.
%%
%%%\bibitem[Ric]{Ric}
%%%\'E. Ricard.
%%%\newblock A Markov dilation for self-adjoint Schur multipliers.
%%%\newblock Proc. Amer. Math. Soc. 136 (2008), no. 12, 4365--4372.
%
%\bibitem[Ric60]{Ric60}
%C. E. Rickart.
%\newblock General Theory of Banach algebras.
%\newblock Van Nostrand, Princeton, 1960.


%\bibitem[RSS24]{RSS24}
%B. V. Rajarama Bhat, Soumyashant Nayak, P. Shankar.
%\newblock On products of symmetries in von Neumann algebras.
%\newblock Journal of Operator Theory 92 (2024), no. 2, 579--596.

%\bibitem[RTWL${}^+$21]{RTWL21}
%M.-O. Renou, D. Trillo, M. Weilenmann, T. P. Le, A. Tavakoli, N. Gisin, A. Acin and M. Navascues.
%\newblock Quantum theory based on real numbers can be experimentally falsified.
%\newblock Nature 600 (2021), 625--629.

%\bibitem[Rot67]{Rot67}
%S. Ju. Rotfel'd.
%\newblock Remarks on the singular values of a sum of completely continuous operators. (Russian).
%\newblock Funkcional. Anal. i Prilozen. 1 (1967), no. 3, 95--96.

\bibitem[Rua04]{Rua04}
Z.-J. Ruan.
\newblock Type decomposition and the rectangular AFD property for $\W^*$-TRO's.
\newblock Canad. J. Math. 56 (2004), no. 4, 843--870.
%%
%%%\bibitem[Sc]{Sc}
%%%E. Schechter.
%%%\newblock Handbook of analysis and its foundations.
%%%\newblock  Academic Press, Inc., San Diego, CA, 1999.
%%
%%
%%
%%%\bibitem[RaX]{RaX}
%%%Y. Raynaud, Q. Xu.
%%%\newblock On subspaces of non-commutative $L_p$-spaces.
%%%\newblock J. Funct. Anal. 203 (2003), no. 1, 149-196.
%%
%%%\bibitem[Rob91]{Rob91}
%%%D. W. Robinson.
%%%\newblock Elliptic operators and Lie groups. 
%%%\newblock Oxford Mathematical Monographs. Oxford Science Publications. The Clarendon Press, Oxford University Press, New York, 1991. 
%%
%%%\bibitem[RoS1]{RoS1}
%%%D. W. Robinson and E. St\o rmer.
%%%\newblock Lie and Jordan structure in operator algebras.
%%%\newblock J. Austral. Math. Soc. Ser. A 29 (1980), no. 2, 129--142.
%%
%%%\bibitem[Sch]{Sch}
%%%L. M. Schmitt.
%%%\newblock The Radon-Nikodym theorem for $L^p$-spaces of W*-algebras.
%%%\newblock Publ. Res. Inst. Math. Sci. 22 (1986), no. 6, 1025--1034.

%\bibitem[Smi00]{Smi00}
%R. R. Smith.
%\newblock Finite dimensional injective operator spaces. 
%\newblock Proc. Amer. Math. Soc. 128 (2000), no. 11, 3461--3462. 


%%%\bibitem[RoV1]{RoV1}
%%%J. Rozendaal and M. Veraar.
%%%\newblock Fourier multiplier theorems involving type and cotype.
%%%\newblock J. Fourier Anal. Appl. 24 (2018), no. 2, 583--619.
%%
%%
%%%\bibitem[Rua96]{Rua96}
%%%Z.-J. Ruan.
%%%\newblock Amenability of Hopf von Neumann algebras and Kac algebras. 
%%%\newblock J. Funct. Anal. 139 (1996), no. 2, 466--499.
%%
%%
%%%\bibitem[Rud1]{Rud1}
%%%W. Rudin.
%%%\newblock Principles of mathematical analysis. Third edition. 
%%%\newblock International Series in Pure and Applied Mathematics. McGraw-Hill Book Co., New York-Auckland-Düsseldorf, 1976. 
%%
%%%\bibitem[Run08]{Run08}
%%%V. Runde.
%%%\newblock Characterizations of compact and discrete quantum groups through second duals. 
%%%\newblock J. Operator Theory 60 (2008), no. 2, 415--428. 
%%
%%\bibitem[Rus73]{Rus73}
%%M. B. Ruskai.
%%\newblock A generalization of entropy using traces on von Neumann algebras. 
%%\newblock Ann. Inst. H. Poincar\'e Sect. A (N.S.) 19 (1973), 357--373.
%%
%%\bibitem[Rus03]{Rus03}
%%M. B. Ruskai.
%%\newblock Qubit entanglement breaking channels. 
%%\newblock Rev. Math. Phys. 15 (2003), no. 6, 643--662.
%%
%%%\bibitem[RSW02]{RSW02}
%%%M. B. Ruskai, S. Szarek and E. Werner.
%%%\newblock An analysis of completely positive trace-preserving maps on $M_2$.
%%%\newblock Linear Algebra Appl. 347 (2002), 159--187.
%%
%%%\bibitem[Rya02]{Rya02}
%%%R. A. Ryan.
%%%\newblock Introduction to tensor products of Banach spaces.
%%%\newblock Springer Monographs in Mathematics. Springer-Verlag London, Ltd., London, 2002.
%%
%%
%%%\bibitem[SaS1]{SaS1}
%%%P. Salmi and A. Skalski.
%%%\newblock Actions of locally compact (quantum) groups on ternary rings of operators, their crossed products, and generalized Poisson boundaries.
%%%\newblock Kyoto J. Math. 57 (2017), no. 3, 667--691.
 %%
%%
%%%\bibitem[Sch1]{Sch1}
%%%G. Schlichting.
%%%\newblock Groups with representations of bounded degree. 
%%%\newblock Probability measures on groups (Proc. Fifth Conf., Oberwolfach, 1978), pp. 344--348. Lecture Notes in Math., 706, Springer, Berlin, 1979. 
%%
%%%\bibitem[Schr1]{Schr1}
%%%W. J. Schreiner.
%%%\newblock Matrix regular operator spaces.
%%%\newblock J. Funct. Anal. 152 (1998), no. 1, 136--175.
%%
%%\bibitem[ScS18]{ScS18}
%%T. T. Scheckter and F. Sukochev.
%%\newblock Weak type estimates for the noncommutative Vilenkin-Fourier series.
%%\newblock Integral Equations Operator Theory 90 (2018), no. 6, Paper No. 64, 19 pp.
%%
%%
%%%\bibitem[Schr2]{Schr2}
%%%W. J. Schreiner.
%%%\newblock Matrix regular orders on operator spaces.
%%%\newblock PhD Thesis. University of Illinois at Urbana-Champaign, 1995.
%%
%%%\bibitem[JaR1]{JaR1}
%%%S. Jaques and M. Rahaman.
%%%\newblock Spectral properties of tensor products of channels. 
%%%\newblock J. Math. Anal. Appl. 465 (2018), no. 2, 1134--1158. 

%\bibitem[Sak98]{Sak98}
%S. Sakai.
%\newblock $C^*$-algebras and $W^*$-algebras.
%\newblock Reprint of the 1971 edition. Classics in Mathematics. Springer-Verlag, Berlin, 1998.

\bibitem[SaS13]{SaS13}
P. Salmi and A. Skalski.
\newblock Inclusions of ternary rings of operators and conditional expectations.
\newblock  Math. Proc. Cambridge Philos. Soc. 155 (2013), no. 3, 475--482.


\bibitem[SaS17]{SaS17}
P. Salmi and A. Skalski.
\newblock Actions of locally compact (quantum) groups on ternary rings of operators, their crossed products, and generalized Poisson boundaries. 
\newblock Kyoto J. Math. 57 (2017), no. 3, 667--691. 
%%
%%%\bibitem[Sal1]{Sal1}
%%%P. Salmi.
%%%\newblock Subgroups and strictly closed invariant $C^*$-subalgebras. 
%%%\newblock Preprint, arXiv. 
%%
%%%\bibitem[Sal2]{Sal2}
%%%P. Salmi.
%%%\newblock Compact quantum subgroups and left invariant $C^*$-subalgebras of locally compact quantum groups. 
%%%\newblock J. Funct. Anal. 261 (2011), no. 1, 1--24. 
%%
%%\bibitem[SWS90]{SWS90}
%%F. Schipp, W. R. Wade and P. Simon.
%%\newblock Walsh series. An introduction to dyadic harmonic analysis. 
%%\newblock Adam Hilger, Ltd., Bristol, 1990. 
%
%
%\bibitem[Sch97]{Sch97}
%W. J. Schreiner.
%\newblock Matrix Regular Operator Spaces.
%\newblock .
%


%\bibitem[ShW64]{ShW64}
%D. Shale and W. F. Stinespring.
%\newblock States of the Clifford Algebra.
%\newblock Ann. of Math. 80 (1964), no. 2, 365--381.

%\bibitem[Sha21]{Sha21}%
%F. Shahandeh.
%\newblock Contextuality of general probabilistic theories.
%\newblock PRX Quantum 2 (2021), 010330.


%\bibitem[Shu79]{Shu79}
%F. W. Shultz.
%\newblock On normed Jordan algebras which are Banach dual spaces.
%\newblock J. Funct. Anal. 31 (1979), no. 3, 360--376.

%
%%\bibitem[Sch04]{Sch04}
%%J. Schur.
%%\newblock \"{U}ber die Darstellung der endlichen Gruppen durch gebrochene lineare Substitutionen.
%%\newblock J. Reine Angew. Math. 127 (1904), 20--50.
%%
%%\bibitem[Sch07]{Sch07}
%%J. Schur.
%%\newblock Untersuchungen \"uber die Darstellung der endlichen Gruppen durch gebrochene lineare Substitutionen.
%%\newblock J. Reine Angew. Math. 132 (1907), 85--137.
%%
%%\bibitem[Sch11]{Sch11}
%%J. Schur.
%%\newblock \"{U}ber die Darstellung der symmetrischen und der alternierenden Gruppe durch gebrochene lineare Substitutionen.
%%\newblock J. Reine Angew. Math. 139 (1911), 155--250.
%%
%%\bibitem[Seg60]{Seg60}
%%I. E. Segal.
%%\newblock A note on the concept of entropy.
%%\newblock J. Math. Mech. 9 (1960), 623--629.
%%
%%%\bibitem[Sek96]{Sek96}
%%%Y. Sekine.
%%%\newblock An example of finite-dimensional Kac algebras of Kac-Paljutkin type. 
%%%\newblock Proc. Amer. Math. Soc. 124 (1996), no. 4, 1139--1147. 
%%


%\bibitem[She06]{She06}
%D. Sherman.
%\newblock On the structure of isometries between noncommutative $L^p$ spaces. 
%\newblock  Publ. Res. Inst. Math. Sci. 42 (2006), no. 1, 45--82.

%%
%%%\bibitem[Sho95]{Sho95}
%%%P. W. Shor.
%%%\newblock Scheme for reducing decoherence in quantum computer memory. 
%%%\newblock Physical Review A 52(4), R2493-R2496. 
%%
%%\bibitem[ShU19]{ShU19}
%%T. Shulman and O. Uuye.
%%\newblock Approximations of subhomogeneous algebras. 
%%\newblock Bull. Aust. Math. Soc. 100 (2019), no. 2, 328--337. 
%%
%%\bibitem[SiN22]{SiN22}
%%S. Singh and I. Nechita.
%%\newblock The PPT$^2$ conjecture holds for all Choi-type maps.
%%\newblock Ann. Henri Poincar\'e 23 (2022), no. 9, 3311--3329.
%%
%%%\bibitem[SkV19]{SkV19}
%%%A. Skalski and A. Viselter.
%%%\newblock Convolution semigroups on locally compact quantum groups and noncommutative Dirichlet forms.
%%%\newblock J. Math. Pures Appl. (9) 124 (2019), 59--105.
%%
%%%\bibitem[Smi72]{Smi1}
%%%M. Smith.
%%%\newblock Regular representations of discrete groups.
%%%\newblock J. Funct. Anal. 11 (1972), 401--406.
%%
%%%\bibitem[Smi83]{Smi83}
%%%R. R. Smith.
%%%\newblock Completely bounded maps between $C^*$-algebras. 
%%%\newblock J. London Math. Soc. (2) 27 (1983), no. 1, 157--166. 
%%
%%\bibitem[SmS05]{SmS05}
%%R. R. Smith and N. Spronk.
%%\newblock Representations of group algebras in spaces of completely bounded maps. 
%%\newblock Indiana Univ. Math. J. 54 (2005), no. 3, 873--896. 
%%
%%\bibitem[Sol10]{Sol10}
%%P. M. Soltan.
%%\newblock Quantum $\SO(3)$ groups and quantum group actions on $M_2$. 
%%\newblock J. Noncommut. Geom. 4 (2010), no. 1, 1--28.


%\bibitem[SAH25]{SAH25}
%K. Sonoda, H. Arai and M. Hayashi.
%\newblock Hypothesis testing and Stein's lemma in general probability theories with Euclidean Jordan algebra and its quantum realization.
%\newblock Preprint, arXiv:2505.02487.

%%\bibitem[Spe92]{Spe92}
%%R. Speicher.
%%\newblock A noncommutative central limit theorem.
%%\newblock Math. Z. 209 (1992), no. 1, 55--66.

%\bibitem[Sta82]{Sta82}
%P. J. Stacey.
%\newblock Type $\I_2$ $\JBW$-algebras. 
%\newblock Quart. J. Math. Oxford Ser. (2) 33 (1982), no. 129, 115--127.


%%%\bibitem[Sto1]{Sto1}
%%%E. St\o rmer.
%%%\newblock Conditional expectations and projection maps of von Neumann algebras.
%%%\newblock  Operator algebras and applications (Samos, 1996), 449--461,
%%%NATO Adv. Sci. Inst. Ser. C Math. Phys. Sci., 495, Kluwer Acad. Publ., Dordrecht, 1997.
%%
%%\bibitem[SBSW15a]{SBSW15a}
%%R. S. Stankovic, P. L. Butzer, F. Schipp and W. R. Wade.
%%\newblock Dyadic Walsh analysis from 1924 onwards--Walsh-Gibbs-Butzer dyadic differentiation in science. Vol. 1. Foundations. A monograph based on articles of the founding authors, reproduced in full. In collaboration with the co-authors: Weiyi Su, Yasushi Endow, Sandor Fridli, Boris I. Golubov, Franz Pichler and Kees Onneweer.
%%\newblock Atlantis Studies in Mathematics for Engineering and Science, 12. Atlantis Press, Paris, 2015.
%%
%%\bibitem[SBSW15b]{SBSW15b}
%%R. S. Stankovic, P. L. Butzer, F. Schipp and W. R. Wade.
%%\newblock Dyadic Walsh analysis from 1924 onwards--Walsh-Gibbs-Butzer dyadic differentiation in science. Vol. 2. Extensions and generalizations. A monograph based on articles of the founding authors, reproduced in full. In collaboration with the co-authors: Weiyi Su, Yasushi Endow, Sandor Fridli, Boris I. Golubov, Franz Pichler and Kees Onneweer.
%%\newblock Atlantis Studies in Mathematics for Engineering and Science, 13. Atlantis Press, Paris, 2015.
%%
%%%\bibitem[Sto13]{Sto13}
%%%E. St\o rmer.
%%%\newblock Positive linear maps of operator algebras.
%%%\newblock Springer Monographs in Mathematics. Springer, Heidelberg, 2013.
 %%
%%\bibitem[Sto80]{Sto80}
%%E. St\o rmer.
%%\newblock Regular abelian Banach algebras of linear maps of operator algebras.
%%\newblock J. Funct. Anal. 37 (1980), no. 3, 331--373.
%%
%%%\bibitem[Sto07]{Sto07}
%%%E. St\o rmer.
%%%\newblock A reduction theorem for capacity of positive maps.
%%%\newblock Positivity 11 (2007), no. 1, 69--75.
%%
%%
%%%\bibitem[Sto4]{Sto4}
%%%E. St\o rmer.
%%%\newblock Decomposable positive maps on C*-algebras.
%%%\newblock Proc. Amer. Math. Soc. 86 (1982), no. 3, 402--404.
%%

%\bibitem[Sto66]{Sto66}
%E. St\o rmer.
%\newblock Jordan algebras of type I.
%\newblock Acta Math. 115 (1966), 165--184.

%\bibitem[Sto68]{Sto68}
%E. St\o rmer.
%\newblock Irreducible Jordan algebras of self-adjoint operators.
%\newblock Trans. Amer. Math. Soc. 130 (1968), 153--166.


%\bibitem[Sto82]{Sto82}%
%E. St\o rmer.
%\newblock Positive projections with contractive complements on $C^*$-algebras.
%\newblock J. Lond. Math. Soc. (2) 26 (1982), 132--142.

%%%
%%%\bibitem[Sto6]{Sto6}
%%%E. St\o rmer.
%%%\newblock A completely positive map associated with a positive map.
%%%\newblock Pacific J. Math. 252 (2011), no. 2, 487--492.
%%%
%%%\bibitem[Sto7]{Sto7}
%%%E. St\o rmer.
%%%\newblock A decomposition theorem for positive maps, and the projection onto a spin factor.
%%%\newblock Math. Scand. 118 (2016), no. 1, 106--118.
%%
%%%\bibitem[Sto8]{Sto8}
%%%E. St\o rmer.
%%%\newblock Decomposition of positive projections on C*-algebras.
%%%\newblock Math. Ann. 247 (1980), no. 1, 21--41.
%%
%%%\bibitem[Sto9]{Sto9}
%%%E. St\o rmer.
%%%\newblock Positive projections onto Jordan algebras and their enveloping von Neumann algebras.
%%%\newblock Ideas and methods in mathematical analysis, stochastics, and applications (Oslo, 1988), 389--393, Cambridge Univ. Press, Cambridge, 1992.
%%
%%%\bibitem[Sto10]{Sto10}
%%%E. St\o rmer.
%%%\newblock Positive projections with contractive complements on C*-algebras.
%%%\newblock J. London Math. Soc. (2) 26 (1982), no. 1, 132--142.
%%
%%%\bibitem[Sto11]{Sto11}
%%%E. St\o rmer.
%%%\newblock On the Jordan structure of C*-algebras.
%%%\newblock Trans. Amer. Math. Soc. 120 (1965), 438--447.
%%
%%%\bibitem[Sto12]{Sto12}
%%%E. St\o rmer.
%%%\newblock On projection maps of von Neumann algebras.
%%%\newblock Math. Scand. 30 (1972), 46--50.
%%%
%%%\bibitem[Sto13]{Sto13}
%%%E. St\o rmer.
%%%\newblock Multiplicative properties of positive maps.
%%%\newblock Math. Scand. 100 (2007), no. 1, 184--192.
%%
%%%\bibitem[Sto14]{Sto14}
%%%E. St\o rmer.
%%%\newblock Positive linear maps of C*-algebras. Foundations of quantum mechanics and ordered linear spaces.
%%%\newblock Advanced Study Inst., Marburg, 1973. Lecture Notes in Phys., Vol. 29, Springer, Berlin, 1974, 85--106
%%
%%
%%
%%
%%\bibitem[SiS08]{SiS08}
%%A. Sinclair and R. Smith.
%%\newblock Finite von Neumann algebras and masas.
%%\newblock London Mathematical Society Lecture Note Series, 351. Cambridge University Press, Cambridge, 2008.
%%
%%%\bibitem[Ste12]{Ste12}
%%%B. Steinberg.
%%%\newblock Representation theory of finite groups. An introductory approach.
%%%\newblock Universitext. Springer, New York, 2012.
%%
%%\bibitem[Sto74]{Sto74}
%%E. St\o rmer.
%%\newblock Spectra of ergodic transformations
%%\newblock J. Funct. Anal. 15 (1974), 202--215.
%%
%
%\bibitem[Sto13]{Sto13}
%E. St\o rmer.
%\newblock Positive linear maps of operator algebras.
%\newblock Springer Monographs in Mathematics. Springer, Heidelberg, 2013.
%%
%%
%%%\bibitem[StZ79]{StZ79}
%%%S. Stratila and L. Zsido.
%%%\newblock Lectures on von Neumann algebras.
%%%\newblock Editura Academiei, Bucharest; Abacus Press, Tunbridge Wells, 1979.
%%
%%\bibitem[Stra81]{Str81}
%%S. Stratila.
%%\newblock Modular theory in operator algebras.
%%\newblock Taylor and Francis, 1981.
%%
%%%\bibitem[Sza09]{Sza09}
%%%S. J. Szarek.
%%%\newblock Still more on norms of completely positive maps. 
%%%\newblock Proceedings of IWOTA 2008, Operator Theory: Advances and Applications, Vol. 202, Birkhauser 2009, 535--538.
%%
%%%\bibitem[StZ1]{StZ1}
%%%S. Stratila and L. Zsido.
%%%\newblock Lectures on von Neumann algebras. Revision of the 1975 original. Translated from the Romanian by Silviu Teleman.
%%%\newblock Editura Academiei, Bucharest; Abacus Press, Tunbridge Wells, 1979.
%%
%
%\bibitem[Tak02]{Tak02}%[Tak02]
%M. Takesaki.
%\newblock Theory of operator algebras. I. Reprint of the first (1979) edition.
%\newblock Encyclopaedia of Mathematical Sciences, 124. Operator Algebras and Non-commutative Geometry, 5. Springer-Verlag, Berlin, 2002.

\bibitem[Tak03]{Tak03}
M. Takesaki.
\newblock Theory of operator algebras. II.
\newblock Encyclopaedia of Mathematical Sciences, 125. Operator Algebras and Non-commutative Geometry, 6. Springer-Verlag, Berlin, 2003.
%%
%%%\bibitem[TaT1]{TaT1}
%%%M. Takesaki and N. Tatsuuma.
%%%\newblock Duality and subgroups.
%%%\newblock Ann. of Math. (2) 93 (1971), 344--364.
%%
%%%\bibitem[Tak1]{Tak1}
%%%M. Takesaki.
%%%\newblock Duality for crossed products and the structure of von Neumann algebras of type III.
%%%\newblock Acta Math. 131 (1973), 249--310.
%%
%%\bibitem[Tse24]{Tse24}
%%A. Tselishchev.
%%\newblock Littlewood–Paley–Rubio de Francia inequality for unbounded Vilenkin systems.
%%\newblock J. Approx. Theory 298 (2024), Paper No. 106006, 31 pp.
%
%\bibitem[Ter81]{Ter81}
%M. Terp.
%\newblock $L^p$ spaces associated with von Neumann algebras.
%\newblock Notes, Math. Institute, Copenhagen Univ., 1981.
%%
%%
%%%\bibitem[Ter82]{Ter82}
%%%M. Terp.
%%%\newblock Interpolation spaces between a von Neumann algebra and its predual.
%%%\newblock J. Operator Theory 8 (1982), no. 2, 327--360.
%%
%%%\bibitem[Tro16]{Tro16}
%%%S. Trotter.
%%%\newblock Involutive Algebras and Locally Compact Quantum Groups.
%%%\newblock PhD dissertation, University of Leeds, 2016.
%%
%%%\bibitem[Tus22]{Tus22}
%%%L. Tuset.
%%%\newblock Analysis and quantum groups.
%%%\newblock Springer, Cham, 2022.
%%
%%
%%%\bibitem[Tak3]{Tak3}
%%%M. Takesaki.
%%%\newblock Conditional expectations in von Neumann algebras. 
%%%\newblock J. Functional Analysis {\bf 9} (1972), 306-321.
%%
%%%\bibitem[Terp]{Terp}
%%%M. Terp.
%%%\newblock $L^p$ spaces associated with von Neumann algebras.
%%%\newblock Notes, Math. Institute, Copenhagen Univ., 1981.
%%
%%\bibitem[TiD24]{TiD24}
%%C. Tian and D. Zhou.
%%\newblock Weak type estimate for the partial sums of noncommutative Vilenkin-Fourier series.
%%\newblock Proc. Amer. Math. Soc. 152 (2024), no. 3, 1153--1167.
%%
%%
%%%\bibitem[Tom1]{Tom1}
%%%J. Tomiyama.
%%%\newblock On the product projection of norm one in the direct product of operator algebras.
%%%\newblock Tôhoku Math. J. (2) 11 1959 305--313.
%%

%\bibitem[Top65]{Top65}
%D. M. Topping.
%\newblock Jordan algebras of self-adjoint operators.
%\newblock Mem. Amer. Math. Soc. 53 (1965).

%\bibitem[Top66]{Top66}
%D. M. Topping.
%\newblock An isomorphism invariant for spin factors.
%\newblock J. Math. Mech. 15 (1966), 1055--1063.

%\bibitem[Tri95]{Tri95}
%H. Triebel.
%\newblock Interpolation theory, function spaces, differential operators. Second edition. 
%\newblock Johann Ambrosius Barth, Heidelberg, 1995.

%%%\bibitem[Tza]{Tza}
%%%L. Tzafriri.
%%%\newblock Remarks on contractive projections in $L_{p}$-space .
%%%\newblock Israel J. Math. 7 1969 9--15.

\bibitem[Upm85]{Upm85}
H. Upmeier.
\newblock Symmetric Banach manifolds and Jordan $C^*$-algebras.
\newblock North Holland, Amsterdam, 1985.

%\bibitem[Upm87]{Upm87}
%H. Upmeier.
%\newblock Jordan Algebras in Analysis, Operator Theory, and Quantum Mechanics.
%\newblock Regional Conference Series in Mathematics No. 67, Amer. Math. Soc., Providence, 1987.



%%%\bibitem[Voi]{Voi}
%%%J. Voigt.
%%%\newblock Abstract Stein interpolation.
%%%\newblock Math. Nachr. 157 (1992), 197--199.
%%
%%%\bibitem[Wat1]{Wat1}
%%%K. Watanabe.
%%%\newblock Dual of noncommutative $L^p$-spaces with $0<p<1$.
%%%\newblock Math. Proc. Cambridge Philos. Soc. 103 (1988), no. 3, 503--509.
%%
%%
%%%\bibitem[Wat2]{Wat2}
%%%K. Watanabe.
%%%\newblock On isometries between noncommutative $L^p$-spaces associated with arbitrary von Neumann algebras.
%%%\newblock J. Operator Theory 28 (1992), no. 2, 267--279.
%%
%%
%%%\bibitem[Wei]{Wei}
%%%M. Weigt.
%%%\newblock Jordan homomorphisms between algebras of measurable operators.
%%%\newblock Quaest. Math. 32 (2009), no. 2, 203--214.
%%
%%
%%\bibitem[Vae01]{Vae01}
%%S. Vaes.
%%\newblock The unitary implementation of a locally compact quantum group action. 
%%\newblock J. Funct. Anal. 180 (2001), no. 2, 426--480.
%%
%%\bibitem[Var85]{Var85}%
%%V. S. Varadarajan.
%%\newblock Geometry of quantum theory. Second edition.
%%\newblock Springer-Verlag, New York, 1985.
%%
%%
%%\bibitem[Vae02]{Vae02}
%%S. Vaes.
%%\newblock Locally compact quantum groups.
%%\newblock Ph. D. Thesis, available at \href{http://www.wis.kuleuven.ac.be/analyse/,
  %%2002}{http://www.wis.kuleuven.ac.be/analyse/,
  %%2002}.
%%
%%%\bibitem[Vae02]{Vae02}
%%%S. Vaes.
%%%\newblock Lecture notes on Locally compact quantum groups.
%%%\newblock.
%%
%%\bibitem[Vil63]{Vil63}
%%N. J. Vilenkin.
%%\newblock On a class of complete orthonormal systems.
%%\newblock Amer. Math. Soc. Transl. (2) 28 (1963) 1--35.
%%
%%\bibitem[Wal79]{Wal79}
%%D. B. Wales.
%%\newblock Some projective representations of $S_n$.
%%\newblock J. Algebra 61 (1979), no. 1, 37--57.
%%
%%\bibitem[Wat58]{Wat58}
%%C. Watari.
%%\newblock On generalized Walsh Fourier series.
%%\newblock Tohoku Math. J. (2) 10 (1958), 211--241.
%%
%%%\bibitem[VaV03]{VaV03}
%%%S. Vaes and L. Vainerman.
%%%\newblock Extensions of locally compact quantum groups and the bicrossed product construction. 
%%%\newblock Adv. Math. 175 (2003), no. 1, 1--101.
%%
%%%\bibitem[VDa14]{VDa14}
%%%A. Van Daele.
%%%\newblock Locally compact quantum groups. A von Neumann algebra approach. 
%%%\newblock SIGMA Symmetry Integrability Geom. Methods Appl. 10 (2014), Paper 082, 41 pp.
%%
%%%\bibitem[VDa97]{VDa97}
%%%A. Van Daele.
%%%\newblock The Haar measure on finite quantum groups. 
%%%\newblock Proc. Amer. Math. Soc. 125 (1997), no. 12, 3489--3500.
%%
%%\bibitem[Voi92]{Voi92}
%%J. Voigt.
%%\newblock Abstract Stein interpolation. 
%%\newblock Math. Nachr. 157 (1992), 197--199.
%%
%%%\bibitem[Wan1]{Wan1}
%%%S. Wang.
%%%\newblock Lacunary Fourier series for compact quantum groups. 
%%%\newblock Comm. Math. Phys. 349 (2017), no. 3, 895--945.
%%
%%%\bibitem[Wan2]{Wan2}
%%%S. Wang.
%%%\newblock Quantum symmetry groups of finite spaces. 
%%%\newblock Comm. Math. Phys. 195 (1998), no. 1, 195--211.
 %%
%%\bibitem[Wan99]{Wan99}
%%S. Wang.
%%\newblock Ergodic actions of universal quantum groups on operator algebras. 
%%\newblock Comm. Math. Phys. 203 (1999), no. 2, 481--498. 
%%
%%%\bibitem[Wan13]{Wan13}
%%%S. Wang.
%%%\newblock More examples of algebraic quantum hypergroups. 
%%%\newblock Algebr. Represent. Theory 16 (2013), no. 1, 205--228. 
%%
%%
 %%
%%\bibitem[Was88a]{Was88a}
%%A. Wassermann.
%%\newblock Ergodic actions of compact groups on operator algebras. II. Classification of full multiplicity ergodic actions. 
%%\newblock Canad. J. Math. 40 (1988), no. 6, 1482--1527. 
%%
%%\bibitem[Was88b]{Was88b}
%%A. Wassermann.
%%\newblock Ergodic actions of compact groups on operator algebras. III. Classification for SU(2). 
%%\newblock Invent. Math. 93 (1988), no. 2, 309--354. 
%%
%%\bibitem[Was89]{Was89}
%%A. Wassermann.
%%\newblock Ergodic actions of compact groups on operator algebras. I. General theory. 
%%\newblock Ann. of Math. (2) 130 (1989), no. 2, 273--319.
%%
%%\bibitem[Wat88]{Wat88}
%%K. Watanabe.
%%\newblock Dual of noncommutative $L^p$-spaces with $0<p<1$.
%%\newblock Math. Proc. Cambridge Philos. Soc. 103 (1988), no. 3, 503--509.

%\bibitem[Wat95]{Wat95}
%F. Watbled.
%\newblock Interpolation complexe d'un espace de Banach et de son antidual. (French. English, French summary) [Complex interpolation of a Banach space and its antidual]. 
%\newblock C. R. Acad. Sci. Paris S\'er. I Math. 321 (1995), no. 11, 1437--1440. 

\bibitem[WCW24]{WCW24}%
L. Wang, H. Chen, and N.-C. Wong.
\newblock The linking von Neumann algebras of $W^*$-TROs.
\newblock Preprint, arXiv:2407.10154.


%\bibitem[Wat00]{Wat00}
%F. Watbled.
%\newblock Complex interpolation of a Banach space with its dual. 
%\newblock Math. Scand. 87 (2000), no. 2, 200--210. 
%%\bibitem[Wat18]{Wat18}
%%J. Watrous.
%%\newblock The Theory of Quantum Information. 
%%\newblock Cambridge university press, 2018.

\bibitem[Wei80]{Wei80}
J. Weidmann.
\newblock Linear operators in Hilbert spaces. Translated from the German by Joseph Sz\"{u}cs.
\newblock Graduate Texts in Mathematics, 68. Springer-Verlag, New York-Berlin, 1980.


%%%\bibitem[WeH02]{WeH02}
%%%R. F. Werner and A. S. Holevo.
%%%\newblock Counterexample to an additivity conjecture for output purity of quantum channels.
%%%\newblock J. Math. Phys. 43 (2002), no. 9, 4353--4357.
%%
%%\bibitem[Wer16]{Wer16}
%%R. F. Werner.
%%\newblock Uncertainty relations for general phase spaces.
%%\newblock Front. Phys. 11 (2016), no. 5, 1--10.
%%
%%\bibitem[Wer84]{Wer84}
%%R. Werner.
%%\newblock Quantum harmonic analysis on phase space.
%%\newblock J. Math. Phys. 25 (1984), no. 5, 1404--1411.
%\bibitem[Wer18]{Wer18}
%K. Werner.
%\newblock Complex interpolation and non-commutative integration. 
%\newblock Adv. Oper. Theory 3 (2018), no. 1, 1--16. 

%%\bibitem[Wey50]{Wey50}
%%H. Weyl.
%%\newblock The theory of groups and quantum mechanics. Translated from the second (revised) German edition by H. P. Robertson.
%%\newblock Reprint of the 1931 English translation. Dover Publications, Inc., New York, 1950.
%%
%
%\bibitem[Wic05]{Wic05}
%A. W. Wickstead.
%\newblock Order and algebra isomorphisms of spaces of regular operators.
%\newblock Math. Ann. 332 (2005), 767--774.
%
%
%%\bibitem[Wig39]{Wig39}
%%E. Wigner.
%%\newblock On unitary representations of the inhomogeneous Lorentz group.
%%\newblock Ann. of Math. (2) 40 (1939), no. 1, 149--204.

%\bibitem[Wig59]{Wig59}
%E. P. Wigner.
%\newblock Group Theory: and its Application to the Quantum Mechanics of Atomic Spectra.
%\newblock Academic Press, New York-London, 1959.

%\bibitem[Wil19]{Wil19}
%A. Wilce.
%\newblock Conjugates, Filters and Quantum Mechanics.
%\newblock Quantum 3, 158 (2019).

%\bibitem[Wil25]{Wil25}
%A. Wilce.
%\newblock Generalized Probability Theory: notes for a short course.
%\newblock Preprint, arXiv:2501.00718.

%\bibitem[WW21]{WW21}%
%V. J. Wright and S. Weigert.
%\newblock General probabilistic theories with a Gleason-type theorem.
%\newblock Quantum 5 (2021), 588.

 

 

 

 


 
%%\bibitem[Wil17]{Wil17}
%%M. M. Wilde.
%%\newblock Quantum information theory. Second edition.
%%\newblock Cambridge University Press, Cambridge, 2017.
%%
%%%\bibitem[WiY16]{WiY16}
%%%A. Winter and D. Yang.
%%%\newblock Potential capacities of quantum channels. 
%%%\newblock IEEE Trans. Inform. Theory 62 (2016), no. 3, 1415--1424.
%%
%%%\bibitem[Wul]{Wul}
%%%D. E. Wulbert.
%%%\newblock A note on the characterization of conditional expectation operators.
%%%\newblock Pacific J. Math. 34 (1970) 285--288.
%%
%%%\bibitem[Woj1]{Woj1}
%%%P. Wojtaszczyk.
%%%\newblock Banach spaces for analysts.
%%%\newblock Cambridge Studies in Advanced Mathematics, 25. Cambridge University Press, Cambridge, 1991.
%
%\bibitem[Won]{Won}
%Papers of Wong
%\newblock
%\newblock
%
%%%\bibitem[Xio17]{Xio17}
%%%X. Xiong.
%%%\newblock Noncommutative harmonic analysis on semigroups and ultracontractivity. 
%%%\newblock Indiana Univ. Math. J. 66 (2017), no. 6, 1921--1947.
%%
%%\bibitem[YHW19]{YHW19}
%%D. Yang, K. Horodecki and A. Winter.
%%\newblock Distributed Private Randomness Distillation. 
%%\newblock Phys. Rev. Lett. 123 (2019), 170501.
%%
%%

\bibitem[You83]{You83}
M. A. Youngson.
\newblock Completely contractive projections on $C^*$-algebras.
\newblock Quart. J. Math. Oxford Ser. (2) 34 (1983), no. 136, 507--511.


%%\bibitem[You76]{You76}
%%W. S. Young.
%%\newblock Mean convergence of generalized Walsh-Fourier series.
%%\newblock Trans. Amer. Math. Soc. 218 (1976), 311--320.
%%
%%%\bibitem[Zet1]{Zet1}
%%%H. Zettl.
%%%\newblock A characterization of ternary rings of operators.
%%%\newblock Adv. in Math. 48 (1983), no. 2, 117--143. 
%%
%%%\bibitem[Zha20]{Zha20}
%%%H. Zhang.
%%%\newblock Infinitely divisible states on finite quantum groups. 
%%%\newblock Math. Z. 294 (2020), no. 1-2, 571--592.
%%
%%\bibitem[ZM68]{ZM68}
%%G. Zeller-Meier.  
%%\newblock Produits crois\'es d'une $C^*$-alg\`ebre par un groupe d'automorphismes. (French).
%%\newblock J. Math. Pures Appl. (9) 47 (1968), 101--239.

\bibitem[Zet83]{Zet83}
H. Zettl.
\newblock A characterization of ternary rings of operators.
\newblock Adv. in Math. 48 (1983), no. 2, 117--143. 


%\bibitem[Zha02]{Zha02}
%X. Zhan.
%\newblock Matrix inequalities.
%\newblock Lecture Notes in Mathematics, 1790. Springer-Verlag, Berlin, 2002.

%%%\bibitem[ZhL21]{ZhL21}
%%%Y. Zhang and S. Luo.
%%%\newblock Quantifying decoherence of Gaussian noise channels. 
%%%\newblock J. Stat. Phys. 183 (2021), no. 2, Paper No. 19, 18 pp..

\end{thebibliography}

{\footnotesize

\vspace{0.2cm}

\noindent C\'edric Arhancet\\ 
\noindent 6 rue Didier Daurat, 81000 Albi, France\\
URL: \href{http://sites.google.com/site/cedricarhancet}{https://sites.google.com/site/cedricarhancet}\\
cedric.arhancet@protonmail.com\\%cedric.arhancet@gmail.com 
ORCID: 0000-0002-5179-6972 

}








\end{document}