\(\mathcal{Z}\) is universal (Q2264068): Difference between revisions

The Jiang-Su algebra \(\mathcal{Z}\) is an infinite-dimensional, simple, unital, separable, nuclear \({C}^*\)-algebra with the same Elliott invariant as the complex numbers. It plays a very important role in Elliott's program to classify simple, nuclear \({C}^*\)-algebras by \(K\)-theoretical data, as it may be regarded as a stably finite analogue of the Cuntz algebra \(\mathcal{O}_\infty\). The algebra \(\mathcal{Z}\) was first constructed by \textit{X. Jiang} and \textit{H. Su} [Am. J. Math. 121, No. 2, 359--413 (1999; Zbl 0923.46069)] as a certain direct limit of dimension drop algebras, where they also showed that any simple, monotracial direct limit of prime dimension drop algebras is isomorphic to \(\mathcal{Z}\). Equivalently, the Jiang-Su algebra can be described as the initial object in the category of strongly self-absorbing \({C}^*\)-algebras [\textit{W. Winter}, J. Noncommut. Geom. 5, No. 2, 253--264 (2011; Zbl 1227.46041)]. However, these presentations are not very convenient when trying to define a homomorphism from \(\mathcal{Z}\), and one would like to have an alternative description of \(\mathcal{Z}\) which would make this task easier. The paper under review addresses this issue by exhibiting an explicit presentation of \(\mathcal{Z}\) as a universal \({C}^*\)-algebra on countably many generators and relations. The presentation makes intensive use of \({C}^*\)-algebras ``generated by completely positive contractive maps of order zero'' (a notion that is made precise in Section 2). Though technical, this presentation, obtained in Section 3, is very explicit and, as the authors point out, involves only *-polynomials and order relations. Though not explicitly contained in this article, the presentation of \(\mathcal{Z}\) as a universal \({C}^*\)-algebra allows one to characterize those separable \({C}^*\)-algebras \(A\) satisfying \(A\otimes\mathcal{Z}\cong A\) without making reference to \(\mathcal{Z}\) itself (and hence without having to review its technical construction). Denote by \(\mathcal{W}\) the simple, monotracial, stably projectionless, nuclear \({C}^*\)-algebra constructed by \textit{A. Kishimoto} and \textit{A. Kumjian} [Can. J. Math. 48, No. 5, 980--996 (1996; Zbl 0865.46054)], and also studied by \textit{S. Razak} [Can. J. Math. 54, No. 1, 138--224 (2002; Zbl 1038.46051)] and \textit{B. Jacelon} [J. Lond. Math. Soc., II. Ser. 87, No. 2, 365--383 (2013; Zbl 1275.46047)]. Having trivial \(K\)-theory, this algebra can be thought of as a stably finite analogue of \(\mathcal{O}_2\) (and hence it is nonunital), and it is moreover believed to be an example of what a nonunital strongly self-absorbing \({C}^*\)-algebra should be (even though it is not even known whether \(\mathcal{W}\otimes\mathcal{W}\) is isomorphic to \(\mathcal{W}\)). In Section 4, the authors show how the relations used to describe \(\mathcal{Z}\) as a universal \({C}^*\)-algebra can be ``deformed'' to obtain an explicit presentation of \(\mathcal{W}\) as a universal \({C}^*\)-algebra on countably many generators and relations. The universal presentations of \(\mathcal{Z}\) and \(\mathcal{W}\) are used to give a concrete construction of a trace preserving embedding \(\mathcal{W}\to \mathcal{Z}\) which is canonical at the level of the Cuntz semigroup (uniqueness, up to approximate unitary equivalence, of such a map follows from results of \textit{L. Robert} [Adv. Math. 231, No. 5, 2802--2836 (2012; Zbl 1268.46041)]. As is pointed out in Section 5, one motivation for this work is to study (strong) self-absorption properties of \(\mathcal{Z}\) and \(\mathcal{W}\). Evidence that this could be possible is given by the fact that UHF-algebras of infinite type admit presentations as universal \({C}^*\)-algebras generated by order zero maps for which it is straightforward to show that these algebras are strongly self-absorbing. The situation for \(\mathcal{Z}\) and \(\mathcal{W}\) is less fortunate. The given presentation of \(\mathcal{Z}\) allows one to show, again using results of Robert [loc.\,cit.], that the infinite tensor product \(\bigotimes_{n=1}^\infty \mathcal{Z}\) is strongly self-absorbing (although it is not clear from the picture given that \(\mathcal{Z}\) is isomorphic to \(\bigotimes_{n=1}^\infty \mathcal{Z}\)). On the other hand, and as mentioned above, it is still not known whether \(\mathcal{W}\) is self-absorbing.