Representation theory and numerical AF-invariants. The representations and centralizers of certain states on 𝒪_{𝒹} (Q4458913)

Let \({\mathcal{O}}_d\) denote the Cuntz \(C^*\)-algebra on generators \(s_1,\dots,s_d\), \(2\leq d<\infty\). Not only are these algebras important in the study of infinite \(C^*\)-algebras and type III factors, but they turn out also to be useful in other areas, such as dynamical systems, wavelets, or several-variable operator theory. Each element \(L=(L_1,\dots,L_d)\in {\mathbb{R}}^d\) defines a one-parameter group \(\sigma^{(L)}=(\sigma_t^{(L)})_{t\in {\mathbb{R}}}\) of automorphisms of \({\mathcal{O}}_d\) defined by \(\sigma_t^{(L)}(s_k)=e^{itL_k} s_k\), \(k=1,\dots,d\). If, say, \(L_1,\dots,L_d>0\), there exists a unique \(\beta>0\) for which \({\mathcal{O}}_d\) admits a KMS\(_\beta\) state \(\omega_L\). Moreover, the value of \(\beta\) is the unique solution of \(e^{-\beta L_1}+\cdots+e^{-\beta L_d}=1\), and \(\omega_L\) is unique and defined for any multi-indices \(\alpha=(\alpha_1,\dots,\alpha_m)\in \{ 1,\dots,d\}^m\), \(\gamma=(\gamma_1,\dots,\gamma_n)\in \{ 1,\dots,d\}^n\), by \(\omega_L (s_\alpha s_\gamma^*)=\delta_{\alpha \gamma} e^{-\beta L(\alpha)}\), where \(L(\alpha)=L_{\alpha_1}+\cdots+L_{\alpha_m}\). Assume that any pair \((L_j,L_k)\) is rationally dependent. In this case, the group \(\sigma^{(L)}\) is periodic and by plain renormalization one may assume that \(L_1,\dots,L_d\) are positive integers which are relatively prime. Moreover, the fixed-point algebra \({\mathcal{U}}_L\) of \(\sigma^{(L)}\) is an AF-algebra which can be explicitly described as the closure of the linear span of the monomials \(s_\alpha s_\gamma^*\) with multi-indices \(\alpha\) and \(\gamma\) such that \(L(\alpha)=L(\gamma)\), and which is simple and has \(\tau=\omega |_{{\mathcal{U}}_L}\) as its unique trace state. The work under review focuses on two topics. The first one is related with the representations of \({\mathcal{O}}_d\): general spatial representations are given an ergodic interpretation using cocycle formulations and the GNS representation associated with the state \(\omega_L\) is analysed using the free semigroup on \(d\) generators. The one-parameter automorphism group \(\sigma^{(L)}\) and the KMS\(_\beta\) state \(\omega_L\) are then studied in detail. The structure of the fixed-point AF-algebra \({\mathcal{U}}_L\) is described explicitly and its Bratteli diagram is shown to be stationary in Section 4. The second part of the memoir is concerned with the concrete analysis of isomorphism classes of these AF-algebras. Work done in the 60's by Glimm and in the 70's by Bratteli, Elliott, and by Effros, Handelman and Shen, shows that isomorphism classes of AF-algebras are in canonical one-to-one correspondence with equivalence classes of Bratteli diagrams (which are combinatorial objects), and with isomorphism classes of dimension groups (which are abelian ordered groups with additional properties called Riesz interpolation property and unperforation). The problem, however, is that while dimension groups and Bratteli diagrams are not hard to compute for the algebras \({\mathcal{U}}_L\), it is very hard to distinguish between their isomorphism classes. To explicitly decide isomorphism/non-isomorphism between these AF-algebras, several numerical invariants associated with the dimension groups \(D({\mathcal{U}}_L)\) are displayed and investigated, leading to a number of interesting partial answers. The problem of \(C^*\)-equivalence and some of these invariants are connected with a new concept of equivalence which is weaker than strong shift equivalence and shift equivalence, previously studied in dynamics. For results in this direction, see the joint work of the first two authors with \textit{K.H. Kim} and \textit{F. Roush} [Ergodic Theory Dyn. Syst. 20, No. 6, 1639--1656 (2000; Zbl 0982.46042)].