Recasting the Elliott conjecture (Q2373388): Difference between revisions

The creed of the Elliott classification programme is that separable nuclear \(C^*\)-algebras should be classifiable (up to \(\ast\)-isomorphism) by \(K\)-theoretical invariants. The invariant proposed by Elliott consists of the ordered \(K\)-theory of the \(C^*\)-algebra \(A\), the Choquet simplex of states \(T(A)\) (when nonempty), and the canonical pairing between \(K_0 (A)\) and \(T(A)\). The corresponding conjecture (EC) generated a considerable amount of research and was proved for large classes of algebras, including: simple purely infinite separable nuclear \(C^*\)-algebras satisfying the Universal Coefficient Theorem [\textit{E.\,Kirchberg}, ``The classification of purely infinite \(C^*\)-algebras using Kasparov's theory'', preprint (1994)], simple unital AH algebras of slow dimension growth [\textit{M.\,Dadarlat} and \textit{G.\,Gong}, Geom.\ Funct.\ Anal.\ 7, No.\,4, 646--711 (1997; Zbl 0905.46042); \textit{G.\,A.\thinspace Elliott, G.\,Gong}, Ann.\ Math.\ (2) 144, No.\,3, 497--610 (1996; Zbl 0867.46041); \textit{G.\,A.\thinspace Elliott, G.\,Gong} and \textit{L.\,Li}, Invent.\ Math.\ 168, No.\,2, 249--320 (2007; Zbl 1129.46051)], and tracially AF algebras [\textit{H.\,Lin}, Can.\ J.\ Math.\ 53, No.\,1, 161--194 (2001; Zbl 0968.46043)]. In [Acta Math.\ 191, No.\,1, 109--142 (2003; Zbl 1072.46036)], \textit{M.\,Rørdam} produced a counterexample to (EC) in the class of simple nuclear stably infinite \(C^*\)-algebras. Shortly after, \textit{A.\,Toms} constructed counterexamples within the class of stably finite \(C^*\)-algebras [J.~Reine Angew.\ Math.\ 578, 185--199 (2005; Zbl 1073.46045)]. Using the Cuntz semigroup to distinguish among isomorphism classes of \(C^*\)-algebras, Toms eventually proved in [\textit{A. S. Toms}, Ann. Math. (2) 167, No. 3, 1029--1044 (2008; Zbl 1181.46047)] that (EC) cannot hold, even if one takes into account the real rank and the continuous homotopy invariant functors. The constructions in both the stably infinite and the stably finite case were influenced by the seminal work of \textit{J.\,Villadsen} [J.~Funct.\ Anal.\ 154, No.\,1, 110--116 (1998; Zbl 0915.46047)]. The paper under review is a mitigation attempt, pointing out to a way of reconciling the original form of (EC) and recently emerged counterexamples. One could boldly try to enlarge Elliott's invariant \(I(A)\) to the triple \((W(A),\langle I_A\rangle,I(A))\), which also takes into account the Cuntz semigroup \(W(A)\) and the order unit \(\langle I_A\rangle\), replacing (EC) by (WEC), which would assert that two simple, separable, unital, nuclear \(C^*\)-algebras with the same invariant are \(\ast\)-isomorphic. However, as previously shown by the second author, the problem of computing the Cuntz semigroup for general Abelian \(C^*\)-algebras looks untractable. As a result, the authors investigate the relation between (EC) and (WEC) in the case of the largest class of \(C^*\)-algebras for which (EC) can hold. Its elements are the so-called \({\mathcal Z}\)-stable algebras \(A\), for which \(A\simeq A\otimes {\mathcal Z}\), where \({\mathcal Z}\) denotes the Jiang--Su algebra [\textit{X.\,Jiang} and \textit{H.\,Su}, Am.\ J.\ Math.\ 121, No.\,2, 359--413 (1999; Zbl 0923.46069)]. One of the main results of this paper is that (EC) implies (WEC) within the class of \({\mathcal Z}\)-stable \(C^*\)-algebras. Furthermore, one proves that (EC) and (WEC) are equivalent (even in a functorial way) within the class of simple, unital, separable, nuclear, \({\mathcal Z}\)-stable \(C^*\)-algebras, and which are either of real rank zero or have finitely many pure tracial states. The proof gives an explicit description of the enveloping group \(K_0^*(A)\) of the Cuntz semigroup, extending significantly earlier results of \textit{B.\,Blackadar} and \textit{D.\,Handelman} [J.~Funct.\ Anal.\ 45, 297--340 (1982; Zbl 0513.46047)]. Goodearl algebras are also shown to satisfy the second theorem, giving evidence that the last condition might be removable.