Monoids and cuspidal group characters. (Q2464345)

Let \(M\) be a finite monoid with unit group \(G\). It is well known that the complex irreducible representations of \(M\) can be classified according to the \(\mathcal J\)-class (apex) of \(M\) they come from. An irreducible representation of \(G\) is called cuspidal if it is not a component of the restriction to \(G\) of an irreducible representation of \(M\) with apex \(J\neq G\). In case the semigroup algebra \(\mathbb{C} M\) is semisimple, it was shown by the author and the reviewer [Int. J. Algebra Comput. 1, No. 1, 33-47 (1991; Zbl 0752.20034)] that an irreducible character \(\chi\) of \(G\) is cuspidal if and only if for every idempotent \(e\in M\), \(e\neq 1\), we have \(\sum_{x\in U(e)}\chi(x)=0\), where \(U(e)=\{x\in G\mid xe=e\}\). If \(G\) is a group of Lie type and \(M\) is its canonical compactification defined by the author and \textit{L. E. Renner} [Trans. Am. Math. Soc. 337, No. 1, 305-319 (1993; Zbl 0787.20039)] then the algebra \(\mathbb{C} M\) is semisimple by a result of the author and the reviewer [Trans. Am. Math. Soc. 323, No. 2, 563-581 (1991; Zbl 0745.20057)] and this characterization of cuspidal representations agrees with the classical Harish-Chandra theory of cuspidal representations of \(G\). In the present paper, the author proves his earlier conjecture, stating that in the general case (when the algebra \(\mathbb{C} M\) is not necessarily semisimple) an irreducible character \(\chi\) of \(G\) is cuspidal if and only if for every \(e\in M\), \(e\neq 1\), one has \(\sum_{x\in V(e)}\chi(x)=0\), where \(V(e)=\{x\in G\mid exe=e\}\). Certain consequences of this result are derived and some examples are given. More information is obtained in the important special case of monoids in which every two \(\mathcal J\)-related idempotents are \(G\)-conjugate, called IC-monoids.