No label defined (Q4032154)

One basic idea of Green's theory of vertices and sources is to subdivide the class of indecomposable representations of a finite group \(G\) into subclasses of representations ``coming'' from a given subgroup, that is which are direct summands of an indecomposable representation of that subgroup induced to \(G\). One obtains so a partition of this class into subclasses parametrized by conjugacy classes of \(p\)-subgroups (vertices), where \(p\) is the characteristic of the field over which the representations are taken. A second parameter (depending on the first one) is the source of a representation \(M\), which is an \(N_ G(P)\)-conjugacy class of indecomposable representations of the vertex \(P\) of \(M\). The main idea of this paper is to refine the notion of vertices and sources in two ways: First by allowing not only subgroups, but also subfactors of \(G\) as possible vertices. Here we have to replace the functor induction by Harish-Chandra induction, which takes modules for subfactors of \(G\) into \(G\)-modules. Secondly we take up an idea of Grabmeier: He defined vertices and sources with respect to subsystems of subgroups which are closed under intersection and conjugation (Mackey system). We generalize this idea as well to Mackey systems of subfactors. We fix such a Mackey system. We show that Higman's lemma holds. As a consequence, we show that Harish-Chandra vertices are always associated, where the notion of associated subfactors replaces conjugation of subgroups. We then introduce a notion of semisimple vertices and simple sources. This is a vertex theory for the Grothendieck group of the category of representations of \(G\). Then we apply this to the representation theory of finite groups of Lie type. We show that Harish-Chandra induction from conjugate Levi subgroups (through given parabolic subgroups) are isomorphic functors, if the characteristic of the field over which the representations are taken is different from the characteristic of the field of definition of \(G\). As a consequence, the (semisimple) Harish-Chandra vertices with respect to the Mackey system of subfactors consisting of Levi subgroups are always full associativity classes. This theory applies in characteristic 0 as well, and in this case one gets back the classical subdivision of irreducible complex characters of \(G\) into Harish-Chandra series. In non-describing positive characteristic one gets Hiss' Harish-Chandra series of irreducible Brauer characters (using semisimple Harish-Chandra vertices) and a subdivision of all indecomposable representations into Harish-Chandra series.