Contractibility of the automorphism group of a von Neumann algebra
Narutaka Ozawa · 2024
Read Contractibility of the automorphism group of a von Neumann algebra by Narutaka Ozawa (2024) 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-2412-17564.txt.
Introduction
Narutaka Ozawa wrote a 2024 mathematics paper on the shape of a symmetry group in operator algebra. The objects are von Neumann algebras, large algebras of operators used in quantum theory. Their automorphism group collects all the ways to relabel the algebra while keeping its structure intact.
Ozawa proves that the approximately inner part of that group, for a separable and strongly stable algebra, is contractible in the u-topology. Contractible means the group's shape can be shrunk to a point without tearing, a strong form of topological simplicity. As a direct result, the full automorphism group of the hyperfinite type III factor is also contractible. That factor is a central object in the field. The proof settles a basic question about its symmetry space.
Catalog record
| CREATOR | Narutaka Ozawa |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 36 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-17564
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Contractibility of the automorphism group of a von Neumann algebra Thanks: The author was partially supported by JSPS KAKENHI Grant Numbers 24K00527 and 20H00114 Narutaka Ozawa Abstract . We prove that the approximately inner automorphism group of a separable strongly stable von Neumann algebra is contractible in the u-topology. Thus the automorphism group of the hyperfinite type III 1 \mathrm{III}_{1} factor is contractible. 1991 Mathematics Subject Classification Primary 58B05; Secondary 46L10 1. Introduction In this paper, we continue Popa and Takesaki’s study ( [ PT ] ) of contractibility of the unitary group and the automorphism group of a von Neumann algebra. Let M be a von Neumann algebra and \Aut(M) denote the group of * -automorphisms on M , equipped with the u-topology, i.e., the pointwise convergence topology on the predual M_{*} . The u-topology makes \Aut(M) a topological group, which is Polish if M has separable predual. We are interested in the closed normal subgroup \Int(M)\subset\Aut(M) of approximately inner automorphisms, which is the closure of the subgroup \mathrm{Int}(M) of inner automorphisms. The von Neumann algebra M is said to be strongly stable (or McDuff ) if M\cong M\otimes\mathcal{R} , where \mathcal{R} denotes the hyperfinite (or AFD) factor of type \mathrm{II}_{1} with separable predual. The following theorem generalizes Popa and Takesaki’s theorem ( [ PT ] ). Theorem A . Let M be a strongly stable von Neumann algebra with separable predual. Then \Int(M) is contractible. The case for the hyperfinite factor \mathcal{R}_{\mathrm{III}_{1}} of type \mathrm{III}_{1} with separable predual has attracted considerable attention because of applications to geometric topology and mathematical physics (see e.g., [ ST ] ). The factor \mathcal{R}_{\mathrm{III}_{1}} is strongly stable (and so are all separable hyperfinite type \mathrm{II} and type \mathrm{III} von Neumann algebras, see [ Ta ] ). That \Aut(\mathcal{R}_{\mathrm{III}_{1}})=\Int(\mathcal{R}_{\mathrm{III}_{1}}) is proved by Kawahigashi, Sutherland, and Takesaki ( [ KST ] ). Corollary . The automorphism group \Aut(\mathcal{R}_{\mathrm{III}_{1}}) of \mathcal{R}_{\mathrm{III}_{1}} is contractible. We denote by \mathcal{U}(M) the unitary group of M equipped with the ultrastrong topology. Let N\subset M be an inclusion of von Neumann algebras (which is always assumed to be unital). The unitary group \mathcal{U}(N) of N will be identified with its image in \mathcal{U}(M) , as they are canonically homeomorphically isomorphic. The following is asserted in [ PT ] , but the proof contains a gap, as explained in Section 5 . We bridge the gap on this occasion. Theorem B . Let N\subset M be an inclusion of \sigma -finite von Neumann algebras and assume that N is strongly stable. Then the quotient map from \mathcal{U}(M) onto \mathcal{U}(M)/\mathcal{U}(N) admits a continuous cross section. Equivalently, there is a continuous equivariant retraction from \mathcal{U}(M) onto \mathcal{U}(N) . 2. Preliminary for Proof of Theorem A Let M be a von Neumann algebra with a faithful normal state \omega . The corresponding 2 -norm on M is given by \|x\|_{2}\coloneqq\omega(x^{*}x)^{1/2} . The topology induced by the 2 -norm coincides with the ultrastrong topology on bounded subsets of M . It also coincides with the ultraweak topology on the unitary group \mathcal{U}(M) . For a\in M , we define a\omega\in M_{*} by (a\omega)(x)\coloneqq\omega(xa) . One has \|a\omega\|\leq\|a\|_{2} by the Cauchy–Schwarz inequality. Let \UCP_{\sigma}(M,N) denote the set of ultraweakly continuous unital completely positive maps from a von Neumann algebra M into N . Note that \Aut(M)\subset\UCP_{\sigma}(M,M) . Let \mathop{\mathcal{S}_{\mathrm{n}}}(N) denote the set of normal states on N . For \alpha\in\UCP_{\sigma}(M,N) , we define its pre-conjugate \alpha^{*}\colon N_{*}\to M_{*} by \alpha^{*}(\varphi)=\varphi\circ\alpha . Note that (\alpha\beta)^{*}=\beta^{*}\alpha^{*} . We extend the definition of the u-topology on \Aut(M) and define the u-topology on \UCP_{\sigma}(M,N) to be the pointwise norm convergence topology of the pre-conjugates. Namely, the u-topology is induced by the pseudo-metrics \{\dist_{\varphi}:\varphi\in\mathop{\mathcal{S}_{\mathrm{n}}}(N)\} , where \dist_{\varphi}(\alpha,\beta)\coloneqq\|(\alpha^{*}-\beta^{*})(\varphi)\|=\|\varphi\circ\alpha-\varphi\circ\beta\|. We also write \dist_{F}\coloneqq\max_{\varphi\in F}\dist_{\varphi} for a finite subset F\subset\mathop{\mathcal{S}_{\mathrm{n}}}(N) . We note that the composition is jointly u-continuous (because all pre-conjugates are norm contractive) and that the u-topology is compatible with the tensor product of von Neumann algebras (because the algebraic tensor product (M_{1})_{*}\otimes(M_{2})_{*} is dense in (M_{1}\otimes M_{2})_{*} ). Moreover, the pseudo-metric \dist_{\varphi} is right invariant on \Aut(M) , i.e., \dist_{\varphi}(\alpha,\beta)=\dist_{\varphi}(\mathrm{id},\beta\alpha^{-1}) for every \alpha,\beta\in\Aut(M) , and the homomorphism \mathcal{U}(M)\ni u\mapsto\Ad u\in\Aut(M) is continuous. Here \Ad u is the inner automorphism defined by (\Ad u)(x)=uxu^{*} . We consider a type \mathrm{II}_{1} factor N and denote by \End(N) the set of unital * -endomorphisms on N . The unique tracial state is denoted by \tau , or \tau_{N} to emphasize N . We recall that the p-topology on \End(N) is the pointwise ultraweak convergence topology. It coincides with the pointwise 2-norm convergence topology, where the 2-norm is taken w.r.t. the trace \tau . Every \psi\in\End(N) is trace-preserving and thus isometric w.r.t. the 2-norm. The p-topology is weaker than the u-topology, but they coincide on \Aut(N) . Lemma 1 . Let N be a type \mathrm{II}_{1} factor. Suppose that \psi_{n}\in\Aut(N) converge to \psi\in\End(N) in the p-topology, and denote by E the unique trace-preserving conditional expectation from N onto \psi(N) . Then, E\psi_{n}\to\psi and \psi_{n}^{-1}\to\psi^{-1}E in the u-topology. Proof. Let a\in N be such that a\geq 0 and \tau(a)=1 . Then since \|(\psi^{-1}E)(a)-(\psi_{n}^{-1}E)(a)\|_{2}=\|\psi_{n}\psi^{-1}(E(a))-E(a)\|_{2}\to 0, one has (E\psi_{n})^{*}(a\tau)=(\psi_{n}^{-1}E)(a)\tau\to(\psi^{-1}E)(a)\tau=(E\psi)^{*}(a\tau) in norm. Since the states of the form a\tau are norm dense in \mathop{\mathcal{S}_{\mathrm{n}}}(N) , the first assertion follows. The proof of the second is similar. ∎ We collect a few well-known facts about the hyperfinite type \mathrm{II}_{1} factor \mathcal{R} . See [ Ta ] for general information. We set \mathcal{R}_{\infty}\coloneqq\bigotimes_{n=1}^{\infty}\mathcal{R} and write \mathcal{R}_{n}\coloneqq\bigotimes_{m=1}^{n}\mathcal{R} the first n tensor product. Hence for \mathcal{R}_{n}^{\mathrm{c}}\coloneqq\bigotimes_{m=n+1}^{\infty}\mathcal{R} , one has \mathcal{R}_{\infty}=\mathcal{R}_{n}\otimes\mathcal{R}_{n}^{\mathrm{c}} . Note that \mathcal{R}_{n}\cong\mathcal{R} for every n\in\mathbb{N}\cup\{\infty\} . The factor \mathcal{R} is strongly self-absorbing in the following sense. Lemma 2 . Let’s write \mathcal{R}=\mathcal{Q}\otimes\mathcal{S} , where \mathcal{Q} and \mathcal{S} are copies of \mathcal{R} . Then there are * -isomorphisms \rho_{\mathcal{Q}}\colon\mathcal{Q}\to\mathcal{R} and \rho_{\mathcal{S}}\colon\mathcal{S}\to\mathcal{R} and a p-continuous map \psi\colon[0,1]\to\UCP_{\sigma}(\mathcal{R},\mathcal{R}\otimes\mathcal{R}) such that \psi_{0}(x)=x\otimes 1 for x\in\mathcal{R} , \psi_{t} are surjective * -isomorphisms for t>0 , and \psi_{1}(x\otimes y)=\rho_{\mathcal{Q}}(x)\otimes\rho_{\mathcal{S}}(y) for x\in\mathcal{Q} and y\in\mathcal{S} . The map \psi satisfies (\mathrm{id}_{\mathcal{R}}\times\tau_{\mathcal{R}})\psi_{t}\to\mathrm{id}_{\mathcal{R}} and \psi_{t}^{-1}\to\mathrm{id}_{\mathcal{R}}\times\tau_{\mathcal{R}} as t\to 0 in the u-topology, where \mathrm{id}_{\mathcal{R}}\times\tau_{\mathcal{R}} is the slice map given by (\mathrm{id}_{\mathcal{R}}\times\tau_{\mathcal{R}})(x\otimes y)=\tau(y)x . Proof. For notational convenience, we replace \mathcal{R}=\mathcal{Q}\otimes\mathcal{S} with \mathcal{R}_{\infty}=\mathcal{R}_{1}\otimes\mathcal{R}_{1}^{\mathrm{c}} . Moreover we work with \UCP_{\sigma}(\mathcal{R}_{\infty},\mathcal{R}\otimes\mathcal{R}_{\infty}) and swap left with right and t\colon 1\mapsto 0 with s\colon 1\mapsto\infty . There is a u-continuous map \sigma\colon[0,1]\to\Aut(\mathcal{R}\otimes\mathcal{R}) that connects \sigma_{0}\coloneqq\mathrm{id} to the flip automorphism \sigma_{1} , given by x\otimes y\mapsto y\otimes x . The map \sigma can be constructed from that for the 2 -by- 2 matrix algebra \mathbb{M}_{2} via the isomorphism \mathcal{R}\cong\bigotimes_{\mathbb{N}}\mathbb{M}_{2} . We write \sigma^{(n)}_{t}\in\Aut(\mathcal{R}_{\infty}) the copy of \sigma_{t} applied at the \{n,n+1\} -th tensor product component. For n\in\mathbb{N} and t\in[0,1] , we define the * -isomorphism \psi_{n+t}\colon\mathcal{R}_{\infty}\to\mathcal{R}\otimes\mathcal{R}_{\infty} by \psi_{n+t}=\psi_{n}\sigma^{(n)}_{t} , where \psi_{n}(x_{1}\otimes x_{2}\otimes\cdots)\coloneqq x_{n}\otimes(x_{1}\otimes\cdots\otimes x_{n-1}\otimes x_{n+1}\otimes\cdots). One has \psi_{1}(x\otimes y)=\rho(x)\otimes\rho^{\prime}(y) for x\in\mathcal{R}_{1} and y\in\mathcal{R}_{1}^{\mathrm{c}} , where \rho\colon\mathcal{R}_{1}\to\mathcal{R} is the canonical identification and \rho^{\prime}\colon\mathcal{R}_{1}^{\mathrm{c}}\to\mathcal{R}_{\infty} is the shift isomorphism. The map \psi is u-continuous on [1,\infty) , and since \bigcup_{n}\mathcal{R}_{n}\otimes\mathbb{C}1_{\mathcal{R}_{n}^{\mathrm{c}}}\subset\mathcal{R}_{\infty} is ultrastrongly dense, one has \psi_{s}\to 1\otimes\mathrm{id}_{\mathcal{R}_{\infty}} as s\to\infty in the p-topology (NB: but not in the u-topology). That \psi satisfies the last statement follows from Lemma 1 , as the conditional expectation E onto \ran\psi_{0}=\mathcal{R}\otimes\mathbb{C}1 is \mathrm{id}\otimes\tau and hence \psi_{0}^{-1}E=\mathrm{id}\times\tau . ∎ Lemma 3 . The trace-preserving conditional expectations E_{n}\coloneqq\mathrm{id}_{\mathcal{R}_{n}}\otimes\tau_{\mathcal{R}_{n}^{\mathrm{c}}}