Endomorphisms of \(\mathcal{B(H)}\). II: Finitely correlated states on \(\mathcal{O}_n\) (Q1355074)

Fix a separable infinite-dimensional Hilbert space \({\mathcal H}\). Let \({\mathcal B}({\mathcal H})\) denote the von Neumann algebra of all bounded linear operators on \({\mathcal H}\). By an endomorphism of \({\mathcal B}({\mathcal H})\) we mean a unital *-endomorphism. The Powers index of such an endomorphism \(\alpha\) is a number \(n\in\{1,2,\dots, \infty\}\) such that the commutant of \(\alpha({\mathcal B}({\mathcal H}))\) is of type \(I_n\), see [\textit{R. T. Powers}, Can. J. Math. 40, No. 1, 86-114 (1988; Zbl 0632.46058)]. If \(n\) is the index of \(\alpha\), then there is a related non-degenerate *-representation of the Cuntz algebra \({\mathcal O}_n\). For \(n\) finite, the relation is as follows. Let \(S_1,S_2,\dots, S_n\) be isometries on \({\mathcal H}\) such that \(\sum^n_{i= 1} S_iS^*_i= 1\). Then we have a representation of the Cuntz algebra \({\mathcal O}_n\) and the associated endomorphism \(\alpha\) of \({\mathcal B}({\mathcal H})\) is given by \[ \alpha(x)= \sum S_ixS^*_i. \] Two endomorphisms \(\alpha\), \(\beta\) of \({\mathcal B}({\mathcal H})\) are called conjugate if there is a *-automorphism \(\gamma\) of \({\mathcal B}({\mathcal H})\) such that \(\alpha\circ\gamma= \gamma\circ\beta\). This will imply that \(\alpha\) and \(\beta\) have the same index \(n\). Moreover, the associated representations of the Cuntz algebra \({\mathcal O}_n\) are unitarily equivalent up to the canonical action of the unitary group \(U(n)\), see [\textit{M. Laca}, J. Oper. Theory 30, No. 1, 85-108 (1993)]. In this paper, two special classes of endomorphisms are considered. The first are the ergodic endomorphisms, i.e., those having only scalar multiples of the identity as fixed points. The other class, which is strictly smaller, consists of endomorphisms \(\alpha\) such that \[ \bigcap^\infty_{n= 1}\alpha^n({\mathcal B}({\mathcal H}))= \mathbb{C} 1. \] An ergodic endomorphism gives rise to an irreducible representation of the Cuntz algebra. Classifying the conjugacy classes of ergodic endomorphisms amounts to the classification of the pure states of \({\mathcal O}_n\), up to the action of \(U(n)\) and unitary equivalence. Because \({\mathcal O}_n\) is an antiliminal \(C^*\)-algebra, this classification is non-smooth. Therefore, the authors consider the subset of finitely correlated states. These states still yield a large class of ergodic endomorphisms. And on the other hand, their classifications turns out to be more tractable. For part I see the authors and \textit{G. L. Price}, Proc. Symp. Pure Math. 59, 93-138 (1996; Zbl 0860.46042).