An extension of compact operators by compact operators with no nontrivial multipliers (Q1731736): Difference between revisions

The main theme of the article under review is the translation of an interesting topological example on compactification to the setting of noncommutative C\(^*\)-algebras. There are locally compact topological spaces with only one compactification in the sense that their one-point and their Čech-Stone compactifications coincide. While there are simple examples for this phenomenon, \textit{S. Mrówka} has constructed one by using almost disjoint subsets of \(\mathbb{N}\), more precisely by using families of uncountably many infinite subsets of \(\mathbb{N}\) every pair of which has finite intersection [Fundam. Math. 94, 83--92 (1977; Zbl 0348.54017)]. Mrówka's space \(X\) is scattered (each nonempty subset has a relatively isolated point) and in the language of spaces of continuous functions some properties of \(X\) can be translated into the validity of the short exact sequence \(0\rightarrow c_0\rightarrow C_0(X)\rightarrow c_0(\mathfrak{c})\rightarrow0\) (with \(\mathfrak{c}\) the cardinality of the continuum), i.e., \(C_0(X)\) is the extension of \(c_0\) by \(c_0(\mathfrak{c})\). Moreover, that \(\mathbb{N}\) is open and dense in \(X\) means that the image of \(c_0\) in \(C_0(X)\) is an essential ideal (a closed ideal whose intersection with any other ideal is nontrivial). The situation above motivated the authors to translate Mrówka's example into the noncommutative context. More precisely, they adopt the canonical dictionary which makes \(c_0\) correspond to the compact operators \(\mathcal{K}(\ell^2)\), \(c_0(\mathfrak{c})\) to \(\mathcal{K}(\ell^2(\mathfrak{c}))\), \(C_0(X)\) to a (noncommutative) C\(^*\)-algebra \(\mathcal{A}\), the one-point compactification of \(X\) to the unitization of \(\mathcal{A}\), and the Čech-Stone compactification of \(X\) to the multiplier algebra of \(\mathcal{A}\). The paper culminates in the main result (in the article's last section) assuring the existence of a type I C\(^*\)-algebra \(\mathcal{A}\) whose unitization equals its multiplier algebra and which satisfies the short exact sequence \[ 0\rightarrow\mathcal{K}(\ell^2)\rightarrow\mathcal{A}\rightarrow\mathcal{K}(\ell^2(\mathfrak{c}))\rightarrow0.\tag{*} \] The authors build on Mrówka's construction, but in order to adapt it to the noncommutative setting, they introduce some technically involved modifications by making use of what they call systems of almost matrix units which, very roughly speaking, are doubly indexed almost disjoint families of compact operators which, moreover, are ``paired by doubling'' and then generate \(\mathcal{A}\) as a C\(^*\)-subalgebra of \(\mathcal{B}(\ell^2)\). In this way, \(\mathcal{A}\) has more properties than explicitly stated in the main result. They are summarized in the article's introduction, to wit, \begin{itemize} \item[(1)] in (*), the image of \(\mathcal{K}(\ell^2)\) in \(\mathcal{A}\) is essential; \item[(2)] as often, there is a difference between separable and nonseparable situations: \(\mathcal{A}\) is not stable (\(^*\)-isomorphic to its tensor product with \(\mathcal{K}(\ell^2)\)), while \(\mathcal{K}(\ell^2)\) and \(\mathcal{K}(\ell^2(\mathfrak{c}))\) are, whereas, if \(\ell^2(\mathfrak{c})\) in (*) were replaced with the separable \(\ell^2\), then the extension \(\mathcal{A}\) would have to be stable, too; \item[(3)] \(\mathcal{A}\) is a scattered C\(^*\)-algebra (each nonzero subalgebra has a minimal projection -- the coupling link with scattered sets being the (nontrivial) fact that a compact space is scattered iff all Radon measures on it are atomic). As a consequence, \(\mathcal{A}\) and its subalgebras are approximately finite dimensional C\(^*\)-algebras. \end{itemize} The paper touches several different intertwining areas, but is written in a reader-friendly way which enables readers to cope with the technical difficulties. Partly, it relies on a companion paper of the same authors [Topology Appl. 240, 183--209 (2018; Zbl 1395.46046)].