No label defined (Q4237982)

The C*-algebra \( {\mathcal{O}}_{H}\), generalising the Cuntz C*-algebra \({\mathcal{O}}_{d}\) generated by \(d\) isometries [see \textit{J. Cuntz}, Commun. Math. Phys. 57, 173--185 (1977; Zbl 0399.46045)], is the \(C^*\) completion of the algebra \(^{o}{\mathcal{O}}_{H}\) generated by subspaces consisting of all bounded linear operators between tensor powers of \(H\) [as done in \textit{T. Ceccerini, S. Dophicher, C. Pinzeri} and \textit{J. E. Roberts}, J. Funct. Anal. 125, No. 2, 416--437 (1994; Zbl 0816.46068)]. For \( G\) a subgroup of \( {\mathcal{U}}(H)\), \({\mathcal{O}}_G\) denotes the fixed point subalgebra of \({\mathcal{O}}_{H}\) under automorphisms coming from the canonical action of G on \({\mathcal{O}}_{H}\). Let \(M\) be a separable von Neumann factor (the author's use of \(M\) in standard form is unnecessary), \(\theta\) a faithful normal state of \(M\) realised on a separable Hilbert space \(H\). These define the corresponding ground-state vector \(\psi\). The first theorem states that if \(M\) is injective then \({\mathcal{O}}_{M}\) is the C*-inductive limit, as \(n\to +\infty\) of von Neumann tensor products of n copies of the dual \(M'\), crossed with the action of the symmetric group \({\mathcal{S}}(n)\) permuting the factors. The proof implicitly uses amenability of the symmetric group. He uses a result in Ceccherini et al.(loc.cit.) that \({\mathcal{O}}_{G} \) is the limit of fixed point algebras of the generating subspaces for \({\mathcal{O}}_H\). He quotes, without any reference, ``The (sic)'' Theorem of Weyl. There is a theorem in Chapter 4 of \textit{H. Weyl}'s book [``The Classical Groups, their invariants and representations'', Princeton (1939)] implying that \(\{ u_{1} \otimes...\otimes u_{n}; u_{i} = u \in {\mathcal{U}}(M_m)\} '\) \( m,n \in {\mathbb{N}}\), is made up of the fixed points \(M_{m} \otimes\dots\otimes M_{m}\), (\(n\) copies), under the action of \({\mathcal{U}}(n)\). However the author's theorem is concerned with \({\mathcal{S}}\) not \({\mathcal{U}}\); since \({\mathcal{U}}(d)\) and \({\mathcal{S}}(d)\) can be considered each as the centralizer of the other [the `double centralizer theorem' of \textit{I. Schur}, Dissertation, Berlin (1901)] the result stays valid when one replaces \({\mathcal{U}}\) by \({\mathcal{S}}\) [cf. also \textit{R. W. Carter} and \textit{G. Lusztig}, Math. Z. 136, 193--242 (1974; Zbl 0298.20009)]. In the other theorem the author claims that the restriction of the infinite product \( \bigotimes^{\psi}{\theta_{i}}\), \(\theta_{i} = \theta\) for all \(i\), [see \textit{A. Guichardet}, Ann. Sci. École Norm. Super., III. Ser. 83, 1--52 (1966; Zbl 0154.38905)], to \({\mathcal{O}}_{M}\) is a pure state. The author identifies the von Neumann algebra generated by \({\mathcal{O}}_{M}\) on \(H\) with the asymptotically abelian dynamical system \(({\mathcal{A}}, G, \alpha)\) where \({\mathcal{A}} = {\bigotimes_{1}^{\infty}{M'}}\) (see Guichardet, loc. cit.), \(G = {\mathcal{S}}(\infty)\) and \(\alpha\) denotes the action of permuting the factors. As in [\textit{E. Størmer}, J. Funct. anal. 3, 48--68 (1969; Zbl 0167.43403)], the state is unique \(G\)-invariant and ergodic, i.e., extremal amongst the invariant states. However, no justification is given that the state is pure. The article is hard to read; for instance, besides leaving out crucial points much is left undefined and the notation used varies considerably within the article.