Condition (K) for inverse semigroups and the ideal structure of their \(C^{\ast}\)-algebras (Q1714857)

The theory of graph \(C^*\)-algebras has inspired a number of generalizations, and consequently it has provided motivation for large body of research in recent years. Groupoids play a central role in successful approaches to these algebras, specially when considering the \(C^*\)-algebra of the graph groupoid [\textit{A. Kumjian} et al., Pac. J. Math. 184, No. 1, 161--174 (1998; Zbl 0917.46056)]. After stated the basic properties of graph \(C^*\)-algebras, the natural next step is determine its ideal structure. When considering the combinatorial approach, there are two properties to keep in mind: \begin{itemize} \item[1.] Condition (L): every cycle of the graph has an entrance, which turns out to be equivalent to the graph groupoid being effective. \item[2.] Condition (K): every vertex that is basis of a cycle has at least two return paths, which is equivalent to the restriction groupoid associated to any closed invariant subset of the unit space being effective [\textit{L. Orloff Clark} et al., Trans. Am. Math. Soc. 371, No. 8, 5461--5486 (2019; Zbl 1451.16025)]; equivalently, the path groupoid is strongly effective. \end{itemize} There is a particularly interesting picture of the graph groupoid as the tight groupoid of its natural underlying inverse semigroup [\textit{R. Exel}, Bull. Braz. Math. Soc. (N.S.) 39, No. 2, 191--313 (2008; Zbl 1173.46035)], which provides a way of extending the results about graph \(C^*\)-algebras to some generalizations (see, e.g. [\textit{R. Exel} and \textit{E. Pardo}, Adv. Math. 306, 1046--1129 (2017; Zbl 1390.46050); \textit{D. Milan} and \textit{B. Steinberg}, Groups Geom. Dyn. 8, No. 2, 485--512 (2014; Zbl 1308.46062)], among others). In the paper under review, the authors develop appropriate notions of Conditions (L) and (K) for inverse semigroups and their associated groupoids. After collecting the basic background about inverse semigroups, topological groupoids and directed graphs needed in the sequel in Section 2. They devote Section 3 to state the right notion of Condition (L) for inverse semigroups. To this end, they recover a congruence relation, denoted \(\leftrightarrow\), introduced in [\textit{D. H. Lenz}, Proc. Edinb. Math. Soc., II. Ser. 51, No. 2, 387--406 (2008; Zbl 1153.22003)], which identifies idempotent elements which coincide in any tight representation of the inverse semigroup. They proof that the quotient map by \(\leftrightarrow\) define homomorphisms between filter spaces, which restricts to tight filters, whence they state Theorem 3.12: there is a canonical isomorphism of topological groupoids from \(\mathcal{G}_{\text{tight}}(S)\) to \(\mathcal{G}_{\text{tight}}(S/\leftrightarrow)\). Thus, using results of [\textit{R. Exel} and \textit{E. Pardo}, Semigroup Forum 92, No. 1, 274--303 (2016; Zbl 1353.20040)], they conclude Theorem 3.14: if \(S\) is an inverse semigroup with \(0\), then \(\mathcal{G}_{\text{tight}}(S)\) is effective if and only if \(S/\leftrightarrow\) is fundamental. The converse being true when \(S\) is Hausdorff [\textit{B. Steinberg}, J. Pure Appl. Algebra 220, No. 3, 1035--1054 (2016; Zbl 1383.20038)], the define \(S\) to satisfy Condition (L) when \(S/\leftrightarrow\) is fundamental. In Section 4 they develop connection between ideals of \(S\) and closed invariant subsets of the spaces of filters \(\widehat{E}_0\), and tight filters \(\widehat{E}_{\text{tight}}\), of \(S\), which correspond to the unit spaces of the universal groupoid \(\mathcal{G}(S)\) and the tight groupoid \(\mathcal{G}_{\text{tight}}(S)\) respectively. To this end, they define maps \(k\) and \(h\) from closed subsets of \(\widehat{E}_0\) to ideals of \(E(S)\) (and back). The map \(h\) is always injective, and given an ideal \(I\) of \(S\), it allows to define isomorphisms of topological groupoids (Theorem 4.10) \[\Phi_I:\mathcal{G}(S)\vert_{h(I\cap E)^c} \rightarrow \mathcal{G}(I) \quad \text{ and }\quad \Phi^I:\mathcal{G}(S)\vert_{h(I\cap E)} \rightarrow \mathcal{G}(S/I).\] Using the notion of outer cover [Exel and Pardo, 2016, loc. cit.] (see also [\textit{D. H. Lenz}, Proc. Edinb. Math. Soc., II. Ser. 51, No. 2, 387--406 (2008; Zbl 1153.22003)]), they state the notion of saturated ideal of \(S\), and show that Theorem 4.10 restrict to tight groupoids when we restrict to saturated ideals of \(S\) (Theorem 4.19). With all these tools the authors state a suitable notion of Condition (K) in Section 5. In analogy with the case of graphs, they define Condition (K) for inverse semigroups appealing to the satisfaction of Condition (L) in suitable quotients. To be concrete, \(S\) satisfies Condition (K) exactly when \(/I\) satisfies Condition (L) for every saturated ideal \(I\) of \(S\). Using the results in Section 4, they conclude that, if \(S\) is Hausdorff and \(E(S)\) satisfies the trapping condition (Definition 4.11), then \(\mathcal{G}_{\text{tight}}(S)\) is strongly effective if and only if \(S\) satisfies Condition (K) (Theorem 5.2). If, additionally, \(E(S)\) admits finite covers (Definition 4.12), then \(S\) satisfies Condition (K) if and only if there is a canonical lattice isomorphism from saturated ideals of \(S\) to: \begin{itemize} \item[1.] Basic ideals of the Steinberg algebra \(A_R(\mathcal{G}_{\text{tight}}(S))\) (Theorem 5.4). \item[2.] Ideals of \(C^*(\mathcal{G}_{\text{tight}}(S))\) (Theorem 5.6). \end{itemize} Since Condition (K) is difficult to prove, the authors provide some sufficient conditions: every congruence on \(S\) is Rees (Corollary 5.10). In Section 6, they apply the above results to self-similar graph actions [Exel and Pardo, 2016, loc. cit.]. Concretely, given a graph \(\Lambda\), they show that the property required in Corollary 5.10 is equivalent to Condition (M) in [\textit{Z. Mesyan} and \textit{J. D. Mitchell}, Semigroup Forum 93, No. 1, 111--130 (2016; Zbl 1354.20034)] for graph semigroups. Also, they characterize Condition (M) in terms of \(E(S_\Lambda/I)\) being \(0\)-disjunctive for every ideal \(I\) of \(S_\Lambda\) (Corollary 6.3). This result extends to self-similar graphs (Proposition 6.11): if \((G, \Lambda, \phi)\) is a self-similar graph, they show that \(S_{G,\Lambda}\) is fundamental if and only if \((G, \Lambda, \phi)\) is faithful (Proposition 6.13), thus concluding Theorem 6.14: every congruence in \(S_{G,\Lambda}\) is Rees if and only if \((G, \Lambda, \phi)\) is strongly faithful and satisfies Condition (M). The authors close the paper by doing some remarks about the uniqueness theorems appearing in [\textit{S. M. LaLonde} and \textit{D. Milan}, Semigroup Forum 95, No. 2, 321--344 (2017; Zbl 1394.46041)].