Rank-two graphs whose \(C^*\)-algebras are direct limits of circle algebras (Q852609): Difference between revisions

We can view \(\mathbb N^{2}\) as the morphisms in a category with one object identified with the identity morphism \(\mathbf 0\), and with composition given by addition. A \textit{2-graph} is given by a functor \(d:\Lambda\to{\mathbb N}^{2}\) such that if \(d(\lambda)={\mathbf {n+m}}\), then \(\lambda=\mu\nu\) where \(d(\mu)={\mathbf n}\) and \(d(\nu)={\mathbf m}\). The idea is to see \(\lambda\) as a ``path of length \(\mathbf n\)'' in a rank-2 generalization of a directed graph. The corresponding \(C^{*}\)-algebra \(C^{*}(\Lambda)\) is a universal \(C^{*}\)-algebra for families of partial isometries which satisfy relations determined by \(\Lambda\). A nice treatment of 2- (and higher rank) graphs can be found in chapter 10 of [\textit{I.\,Raeburn}, ``Graph algebras'' (CBMS Regional Conference Series in Mathematics 103, American Mathematical Society (AMS), Providence) (2005; Zbl 1079.46002)]. In the paper under review, the authors make an extensive study of a class of 2-graphs \(\Lambda\) whose corresponding \(C^{*}\)-algebras are \(A\mathbb T\)-algebras. Associated to each 2-graph \(\Lambda\) are two directed graphs -- called the \textit{blue graph} and the \textit{red graph} determined by \(d^{-1}((1,0))\) and \(d^{-1}((0,1))\). The first main result is that if the blue graph contains no cycles and the red graph is a union of cycles, then \(C^{*}(\Lambda)\) is \(A\mathbb T\). The paper then specializes its focus to a certain subclass of 2-graphs \(\Lambda\) such that the blue graph is a Bratteli diagram and such that the red graph respects the inductive structure. The authors call such 2-graphs rank-2 Bratteli diagrams. They give an independent calculation of the \(K\)-theory, \(K_{*}(C^{*}(\Lambda))\), of the \(C^{*}\)-algebra of rank-2 Bratteli diagrams. With the \(K\)-theory in hand, this leads naturally to questions about simplicity and criteria for the corresponding \(C^{*}\)-algebra to have real-rank zero. Specifically, assuming that \(\Lambda\) is a rank-2 Bratteli diagram, the authors are able to give necessary and sufficient conditions for \(C^{*}(\Lambda)\) to be simple. They also show that when \(\Lambda\) satisfies what they term to be the large-permutation factorizations property, then \(C^{*}(\Lambda)\) is simple and has real-rank zero. The authors also study which pairs of groups \((K_{0},K_{1})\) can arise as \(K_{*}(C^{*}(\Lambda))\) for a rank-2 Bratteli diagram. Additional results include showing that all irrational rotation \(C^{*}\)-algebras and some Bunce--Deddens \(C^{*}\)-algebras can arise as full corners of \(C^{*}(\Lambda)\) for certain \(\Lambda\). Thus they produce a large family of examples of \(C^{*}\)-algebras of higher rank graphs which are neither purely infinite nor AF. Thus the dichotomy which holds for the \(C^{*}\)-algebras of ordinary directed graphs (that is, 1-graphs) does not hold for higher rank graphs.