On enveloping \(C^\ast\)-algebras of Hecke algebras (Q393057)

A folklore idea in operator theory is to build \(C^*\)-algebras associated to objects from other fields of mathematics such as groups, semigroups, graphs, dynamical systems, subshifts, etc. This can be achieved in many ways which may result in non-isomorphic \(C^*\)-algebras. Most commonly one considers a Hausdorff completion of a Banach *-algebra \(\mathcal{A}\) with respect to a family of contractive *-representations and produces a universal enveloping \(C^*\)-algebra. However, when the ambient *-algebra is not normed (or a norm is not taken into account) and the morphisms are just pre-*-representations on inner product spaces, then an enveloping \(C^*\)-algebra need not exist. A reason is that the image of an element \(a\in \mathcal{A}\) under a pre-*-representation need not be bounded. In the exact opposite case, we say that \(a \in \mathcal{A}\) is automatically bounded, and let \(\mathcal{A}_b\) be the *-subalgebra of the automatically bounded elements in \(\mathcal{A}\). When \(\mathcal{A} = \mathcal{A}_b\), then \(\mathcal{A}\) is called a BG\(^*\)-algebra, and it is known that BG\(^*\)-algebras have enveloping \(C^*\)-algebras (see [\textit{T. W. Palmer}, Banach algebras and the general theory of \(^*\)-algebras. II: \(^*\)-algebras. Encyclopedia of Mathematics and Its Applications. 79. Cambridge: Cambridge University Press (2001; Zbl 0983.46040)]). In the paper under review, the problem of the existence of enveloping \(C^*\)-algebras for Hecke pairs is examined and in fact, the author achieves a little bit more. One of the first results concerns a sufficient condition, given by a graph \(\mathcal{G}\) associated to the structure of a *-algebra \(\mathcal{A}\) with a vector basis \(\mathcal{B}\). It is shown that, if \(\mathcal{A}\) is generated as a *-algebra by the elements of the co-hereditary sets of \(\mathcal{G}\), then \(\mathcal{A}\) is a BG\(^*\)-algebra. A Hecke pair \((G, \Gamma)\) consists of a group \(G\) and a subgroup \(\Gamma\) such that every double coset \(\Gamma g \Gamma\) is the union of finitely many right (and left) cosets. Following \textit{A. Krieg} [Mem. Am. Math. Soc. 435, 158 p. (1990; Zbl 0706.11029)] and \textit{S. Kaliszewski}, \textit{M. B. Landstad} and \textit{J. Quigg} [Proc. Edinb. Math. Soc., II. Ser. 51, No. 3, 657--695 (2008; Zbl 1165.46037)], given a Hecke pair one can construct a *-algebra \(\mathcal{H}(G,\Gamma)\), called the Hecke algebra. The completion of \(\mathcal{H}(G,\Gamma)\) with respect to an \(L^1\)-norm is denoted by \(L^1(G,\Gamma)\). Even though an enveloping \(C^*\)-algebra of \(L^1(G,\Gamma)\) always exists, the same is not true for \(\mathcal{H}(G,\Gamma)\). A counterexample is constructed by \textit{R. W. Hall} in [Hecke \(C^*\)-algebras, PhD thesis, The Pennsylvania State University (1999)] for the pair \((SL_2(\mathbb{Q}_p), SL(\mathbb{Z}_p))\), where \(p\) is a prime number. From the work of Hall [loc.\,cit.] and that of Kaliszewski, Landstad and Quigg [loc.\,cit.], it is shown that a correspondence between the unitary representations of \(G\) that are generated by their \(\Gamma\)-fixed vectors and the representations of \(\mathcal{H}(G,\Gamma)\) holds precisely when an enveloping \(C^*\)-algebra exists. In the paper under review, it is shown that, when all double cosets generate finite co-hereditary sets, then the enveloping \(C^*\)-algebra of \(\mathcal{H}(G,\Gamma)\) exists. Note that this is stronger than the previous condition given for general *-algebras to be BG\(^*\)-algebras, and consequently gives a stronger result (see below). Furthermore, a method for checking the validity of the condition is given based on iterated commutators. The author applies his results to deduce the existence of the enveloping \(C^*\)-algebra of \(\mathcal{H}(G,\Gamma)\) in many cases: (i) when \(\Gamma\) has finite index in \(G\); (ii) when \((G,\Gamma)\) is directed; (iii) for Iwahori Hecke algebras; (iv) when \(\Gamma\) is a protonormal subgroup of \(G\); (v) when \(\Gamma\) is subnormal in \(G\); (vi) when \(\Gamma\) is ascendant in \(G\); (vii) when \(\Gamma\) has finitely many conjugates in \(G\); (viii) when \(G\) is finite-by-nilpotent; (ix) when \(G\) is hypercentral; (x) when \(G\) is an FC-group and \(\Gamma\) is finite; (xi) when \(G\) is locally nilpotent and \(\Gamma\) is finite; (xii) when \(G\) is locally finite and \(\Gamma\) is finite. Most importantly, with the exception of (ii) where the answer is not known, in all cases the enveloping \(C^*\)-algebra is shown to coincide with the one of \(L^1(G,\Gamma)\). The author also provides a technique that reduces the problem of the existence of the enveloping \(C^*\)-algebra of \(\mathcal{H}(G,\Gamma)\) to that of the smaller \(\mathcal{H}(N,\Gamma)\), where \(\Gamma \subseteq N \trianglelefteq G\). A list of questions is given at the end of the paper.