Selfless W ∗ -probability spaces and Connes’​ bicentralizer problem

Cyril Houdayer, Amine Marrakchi · 2025

Read Selfless W ∗ -probability spaces and Connes’​ bicentralizer problem by Cyril Houdayer, Amine Marrakchi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-11409.txt.

Introduction

Cyril Houdayer and Amine Marrakchi introduce selfless W-star probability spaces and study how the notion connects with Connes' bicentralizer problem. A W-star probability space pairs a von Neumann algebra, a large algebra of operators, with a faithful normal state that plays the role of an expectation. The bicentralizer problem asks about operators that commute with everything the state fixes, a subtle structural question for such algebras.

Their main result gives a clean sufficient case. If an algebra is a separable factor of type three-one with trivial bicentralizer, then the paired space is selfless for every faithful normal state. That link ties the new notion to a central problem in operator algebra theory.

Catalog record

CREATORCyril Houdayer, Amine Marrakchi
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE13 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-11409

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Selfless W ∗ -probability spaces and Connes’​ bicentralizer problem Thanks: CH is supported by ERC Advanced Grant NET 101141693 Cyril Houdayer, Amine Marrakchi Abstract . We introduce the notion of selfless W ∗ -probability space and study its connection with Connes’​ bicentralizer problem. In particular, we show that if M M is a separable type III 1 {\rm III_{1}} factor with trivial bicentralizer, then ( M , φ ) (M,\varphi) is selfless for every faithful normal state φ ∈ M ∗ \varphi\in M_{\ast} . 2020 Mathematics Subject Classification 46L10, 46L36, 46L54 1. Introduction In [ Ro23 ] , Robert introduced a new class of C ∗ -probability spaces, which he called selfless , characterized by the existence of a copy of themselves in their ultrapower that is freely independent from the diagonal copy (thus being “free from themselves”). This property quickly attracted the attention of numerous researchers as it implies many important regularity properties and is satisfied by a large class of examples (see [ AGKEP24 , HKER25 , RTV25 , Vi25 , Oz25 ] ). In this short note, we introduce a parallel notion of selfless W ∗ -probability space and we relate this notion to Connes’​ bicentralizer problem. A W ∗ - probability space is a pair (M,\varphi) that consists of a von Neumann algebra M endowed with a faithful normal state \varphi\in M_{\ast} . For W ∗ -probability spaces (M,\varphi) and (N,\psi) , we say that (M,\varphi)\subset(N,\psi) is an inclusion of W ∗ -probability spaces if M\subset N and if there exists a faithful normal conditional expectation \operatorname{E}:N\to M such that \varphi\circ\operatorname{E}=\psi . In that case, \operatorname{E}:N\to M is the unique faithful normal conditional expectation such that \varphi\circ\operatorname{E}=\psi . Following [ GH21 ] , we say that an inclusion of W ∗ -probability spaces (M,\varphi)\subset(N,\psi) is existentially closed if there exists a nonprincipal ultrafilter \mathcal{U} on some directed set I such that (M,\varphi)\subset(N,\psi)\subset(M,\varphi)^{\mathcal{U}} , where (M,\varphi)\subset(M,\varphi)^{\mathcal{U}} is the diagonal inclusion. Note that if N is separable (i.e.​ N has separable predual), then \mathcal{U} can be chosen to be a nonprincipal ultrafilter on \mathbb{N} . Adapting [ Ro23 , Definition 2.1] to the von Neumann algebraic realm, we say that a W ∗ -probability space (M,\varphi) is selfless if the first factor inclusion (M,\varphi)\subset(M,\varphi)\ast(M,\varphi) is existentially closed. Popa’s seminal work [ Po95 ] shows that (M,\tau) is selfless for any separable type {\rm II_{1}} factor M endowed with its canonical trace \tau . Houdayer–Isono [ HI14 ] extended Popa’s result by showing that (M,\varphi) is selfless for any separable factor M endowed with a faithful normal state \varphi\in M_{\ast} for which (M_{\varphi})^{\prime}\cap M=\mathbb{C}1 . In this note, we show that a diffuse separable W ∗ -probability space (M,\varphi) is selfless if and only if it has a trivial bicentralizer . Recall from [ Co80 , Ha85 ] , that the bicentralizer \operatorname{B}(M,\varphi) of a W ∗ -probability space (M,\varphi) is the set of all elements x\in M that satisfy the following condition: For every \varepsilon>0 , there exists \delta>0 such that for every u\in\mathscr{U}(M) , if \|u\varphi-\varphi u\|<\delta , then \|ux-xu\|_{\varphi}<\varepsilon . We then obtain the following characterization. Theorem A . Let (M,\varphi) be a diffuse separable \mathrm{W}^{*} -probability space. The following assertions are equivalent: (i) (M,\varphi) is selfless. (ii) The first factor inclusion (M,\varphi)\subset(M,\varphi)\ast(N,\psi) is existentially closed for some nontrivial \mathrm{W}^{*} -probability space (N,\psi) . (iii) The first factor inclusion (M,\varphi)\subset(M,\varphi)\ast(\operatorname{L}(\mathbb{Z}),\tau_{\mathbb{Z}}) is existentially closed, where \tau_{\mathbb{Z}} is the canonical trace on \operatorname{L}(\mathbb{Z}) . (iv) \operatorname{B}(M,\varphi)=\mathbb{C}1 . We point out that the first three conditions above are analogous to the ones appearing in [ Ro23 , Theorem 2.6] for C ∗ -probability spaces. The bicentralizer \operatorname{B}(M,\varphi)\subset M is a von Neumann subalgebra such that \operatorname{B}(M,\varphi)\subset(M_{\varphi})^{\prime}\cap M (see [ Ha85 , Proposition 1.3] ). Thus Theorem A strengthens Houdayer–Isono’s result [ HI14 , Theorem A] . In fact, the proof of Theorem A relies on [ HI14 , Theorem A] combined with a diagonal argument. Observe that if \operatorname{B}(M,\varphi)=\mathbb{C}1 , then M must be a factor. Moreover, according to [ Ok21 ] , one and exactly one of the following assertions hold: • (M,\varphi) is a tracial factor of type {\rm I}_{n} for n\in\mathbb{N}^{*} or of type {\rm II_{1}} . • There exists \lambda\in(0,1) such that (M,\varphi) is a type {\rm III_{\lambda}} factor endowed with its \frac{2\pi}{|\log(\lambda)|} -periodic faithful normal state. In that case, we have (M_{\varphi})^{\prime}\cap M=\mathbb{C}1 . • M is a type {\rm III_{1}} factor. In that case, using [ Ha85 , Corollary 1.5] , we further have \operatorname{B}(M,\psi)=\mathbb{C}1 for every faithful normal state \psi\in M_{\ast} . In particular, Theorem A implies that no W ∗ -probability space of type {\rm II_{\infty}} or type {\rm III_{0}} can be selfless (this also follows by combining [ GH21 , Theorem 3.5] and [ Ue11 , Theorem 4.1] ). On the other hand, a famous conjecture of Connes, known as Connes’​ bicentralizer problem, claims that \operatorname{B}(M,\varphi)=\mathbb{C}1 for every type {\rm III_{1}} factor M and every faithful normal state \varphi\in M_{*} . This conjecture has been verified for several families of type {\rm III_{1}} factors such as amenable factors [ Ha85 ] , factors with a Cartan subalgebra, free products [ HU15 ] , semisolid factors [ HI15 ] , q -deformed Araki–Woods factors [ HI20 , Bi24 ] and tensor products of type {\rm III_{1}} factors [ Ma25 ] . Corollary B . Let M be a separable type {\rm III_{1}} factor satisfying Connes’​ bicentralizer conjecture. Then for every faithful normal state \varphi\in M_{\ast} , the \mathrm{W}^{*} -probability space (M,\varphi) is selfless. 2. Proof of Theorem A Proof. (\rm i)\Rightarrow(\rm ii) This is obvious by taking (N,\psi)=(M,\varphi) . (\rm ii)\Rightarrow(\rm iii) Let (N,\psi) be a nontrivial W ∗ -probability space such that the first factor inclusion (M,\varphi)\subset(M,\varphi)\ast(N,\psi) is existentially closed. We may assume that N is separable. Choose a nonprincipal ultrafilter \mathcal{U} on \mathbb{N} such that we have (M,\varphi)\subset(M,\varphi)\ast(N,\psi)\subset(M,\varphi)^{\mathcal{U}} , where (M,\varphi)\subset(M,\varphi)^{\mathcal{U}} is the diagonal inclusion. There are two cases to consider. Firstly, assume that N_{\psi}=\mathbb{C}1 . Then N is a type {\rm III_{1}} factor (see e.g.​ [ AH12 , Lemma 5.3] ). Consider the same nonprincipal ultrafilter \mathcal{V}=\mathcal{U} on \mathbb{N} and define \mathcal{W}=\mathcal{V}\otimes\mathcal{U}=\mathcal{U}\otimes\mathcal{U} , which is a nonprincipal ultrafilter on \mathbb{N} . By [ AHHM18 , Proposition 2.5] , we naturally have (M^{\mathcal{U}},\varphi^{\mathcal{U}})^{\mathcal{V}}=(M^{\mathcal{U}},\varphi^{\mathcal{U}})^{\mathcal{U}}=(M,\varphi)^{\mathcal{W}} . Then we have (M,\varphi)\subset(M,\varphi)\ast(N,\psi)^{\mathcal{U}}\subset(M,\varphi)^{\mathcal{U}}\ast(N,\psi)^{\mathcal{U}}\subset(M^{\mathcal{U}},\varphi^{\mathcal{U}})^{\mathcal{U}}=(M,\varphi)^{\mathcal{W}}, where (M,\varphi)\subset(M,\varphi)^{\mathcal{W}} is the diagonal inclusion. By [ AH12 , Proposition 4.24] , the centralizer (N^{\mathcal{U}})_{\psi^{\mathcal{U}}} is a type {\rm II_{1}} factor and so (\operatorname{L}(\mathbb{Z}),\tau_{\mathbb{Z}})\subset((N^{\mathcal{U}})_{\psi^{\mathcal{U}}},\psi^{\mathcal{U}}) . This futher implies that (M,\varphi)\subset(M,\varphi)\ast(\operatorname{L}(\mathbb{Z}),\tau_{\mathbb{Z}})\subset(M,\varphi)\ast((N^{\mathcal{U}})_{\psi^{\mathcal{U}}},\psi^{\mathcal{U}})\subset(M,\varphi)^{\mathcal{W}}, where (M,\varphi)\subset(M,\varphi)^{\mathcal{W}} is the diagonal inclusion. Therefore, the first factor inclusion (M,\varphi)\subset(M,\varphi)\ast(\operatorname{L}(\mathbb{Z}),\tau_{\mathbb{Z}}) is existentially closed. Secondly, assume that N_{\psi}\neq\mathbb{C}1 . Upon replacing (N,\psi) by (N_{\psi},\psi) , we may assume that (N,\psi) is tracial. Reasoning as is the first case, we obtain \displaystyle(M,\varphi) \displaystyle\subset(M,\varphi)\ast(N,\psi)^{\ast 2} \displaystyle=\left((M,\varphi)\ast(N,\psi)\right)\ast(N,\psi) \displaystyle\subset(M,\varphi)^{\mathcal{U}}\ast(N,\psi) \displaystyle\subset(M,\varphi)^{\mathcal{U}}\ast(N,\psi)^{\mathcal{U}} \displaystyle\subset(M^{\mathcal{U}},\varphi^{\mathcal{U}})^{\mathcal{U}}=(M,\varphi)^{\mathcal{W}}. By [ Ro23 , Lemma 2.5] , there exists n\in\mathbb{N} large enough so that the iterated free product (N,\psi)^{\ast n} is diffuse and so (\operatorname{L}(\mathbb{Z}),\tau_{\mathbb{Z}})\subset(N,\psi)^{\ast n} . Upon iterating n times the ultraproduct construction and replacing \mathcal{U} by the appropriate ultrafilter \mathcal{W}=\mathcal{U}^{\otimes n} , we obtain (M,\varphi)\subset(M,\varphi)\ast(\operatorname{L}(\mathbb{Z}),\tau_{\mathbb{Z}})\subset(M,\varphi)\ast(N,\psi)^{\ast n}\subset(M,\varphi)^{\mathcal{W}}, where (M,\varphi)\subset(M,\varphi)^{\mathcal{W}} is the diagonal inclusion. Therefore, the first factor inclusion (M,\varphi)\subset(M,\varphi)\ast(\operatorname{L}(\mathbb{Z}),\tau_{\mathbb{Z}}) is existentially closed. (\rm iii)\Rightarrow(\rm iv) Choose a nonprincipal ultrafilter \mathcal{U} on

Read the complete text