Trace spaces of counterexamples to Naimark's problem

From MaRDI portal
(Redirected from Publication:1785773)



Abstract: A counterexample to Naimark's problem is a Cast-algebra that is not isomorphic to the algebra of compact operators on some Hilbert space, yet still has only one irreducible representation up to unitary equivalence. It is well-known that such algebras must be nonseparable, and in 2004 Akemann and Weaver used the diamond principle (a set theoretic axiom independent from ZFC) to give the first counterexamples. For any such counterexample A, the unitary group U(A) acts transitively on the pure states, which are the extreme points of the state space S(A). It is conceivable that this implies (as happens for finite-dimensional simplexes) that the action of U(A) on S(A) has at most one fixed point, i.e. A has at most one trace. We give a strong negative answer here assuming diamond. In particular, we adapt the Akemann-Weaver construction to show that the trace space of a counterexample to Naimark's problem can be affinely homeomorphic to any metrizable Choquet simplex, and can also be nonseparable.


\textit{M. A. Naĭmark} proved that any two irreducible representations of $K(\mathcal{H})$ are unitarily equivalent [Usp. Mat. Nauk 3, No. 5(27), 52--145 (1948; Zbl 0036.07701)], and later asked whether this property characterizes $K(\mathcal{H})$ up to an isomorphism [\textit{M. A. Naĭmark}, Usp. Mat. Nauk 6, No. 6(46), 160--164 (1951; Zbl 0045.21402)]. To be precise, Naĭmark's problem is stated as follows: ``If $A$ is a $C^*$-algebra with only one irreducible representation up to unitary equivalence, is $A$ isomorphic to $K(\mathcal{H})$ for some (non necessarily separable) Hilbert space $\mathcal{H}$? \par On the one hand, from 1951 on various partial solutions to Naĭmark's problem were obtained, due to \textit{I. Kaplansky} [Trans. Am. Math. Soc. 70, 219--255 (1951; Zbl 0042.34901)], \textit{A. Rosenberg} [Am. J. Math. 75, 523--530 (1953; Zbl 0053.25903)], as well as \textit{N. Suri} and \textit{M. Tomforde} [Ill. J. Math. 61, No. 3--4, 479--495 (2017; Zbl 1414.46043)]. On the other hand, \textit{C. Akemann} and \textit{N. Weaver}, using Jensen's diamond axiom, constructed a counterexample to Naĭmark's problem generated by $\aleph_1$ elements [Proc. Natl. Acad. Sci. USA 101, No. 20, 7522--7525 (2004; Zbl 1064.46034)]; moreover, they showed that the existence of such a counterexample is independent of ZFC (the independence of Naĭmark's problem from ZFC remains unknown). \par In the paper under review, the author studies Akemann and Weaver's counterexamples in order to understand which kind of properties they satisfy, concentrating on trace spaces of $A$ and, more concretely, on the fixed points by the action of $\mathcal{U}(A)$ on the state space $S(A)$ by conjugacy. Apparently, it should be plausible that such counterexamples have at most one trace. Nevertheless, he shows that it is far to hold. Concretely, he shows: \par Theorem 1. Assume Jensen's diamond axiom. Then the following holds: \begin{itemize} \item[(1)] For every metrizable Choquet simplex $X$ there is a counterexample to Naĭmark's problem whose trace space is affinely homeomorphic to $X$. \item[(2)] There is a counterexample to Naĭmark's problem whose trace space is nonseparable. \end{itemize} Theorem 2. Assume Jensen's diamond axiom. For every metrizable Choquet simplex $X$ and $1\leq n\leq \aleph_0$, there is a $C^*$-algebra $A$ such that \begin{itemize} \item[(1)] $A$ is simple, unital, nuclear and of density character $\aleph_1$, \item[(2)] $A$ is not isomorphic to its opposite algebra, \item[(3)] $A$ has exactly $n$ equivalence classes of pure states, \item[(4)] all automorphisms of $A$ are inner, \item[(5)] either one of the following conditions can be obtained: \begin{itemize} \item[(a)] $T(A)$ is affinely homeomorphic to $X$. \item[(b)] $T(A)$ is nonseparable. \end{itemize} \end{itemize} A basic tool for obtaining these results is to refine Kishimoto-Ozawa-Sakai's theorem [\textit{A. Kishimoto} et al., Can. Math. Bull. 46, No. 3, 365--372 (2003; Zbl 1066.46052)].











This page was built for publication: Trace spaces of counterexamples to Naimark's problem

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1785773)