Fundamental \(C\)-strata for classical groups (Q1887575): Difference between revisions

Suppose \(F\) is a nonarchimedean local field, and \(G\) is the group of \(F\)-points of a connected reductive \(F\)-group. To study the irreducible admissible representations of \(G\), \textit{R. E. Howe} [Pac. J. Math. 73, 479-538 (1977; Zbl 0385.22009)] initiated a method of studying ``minimal \(K\)-types'' (also called ``fundamental strata''). Such objects should be defined roughly as follows. For each parahoric subgroup \(P\) of \(G\), one defines suitable filtrations \(\{P_n\}\). A ``stratum'' is an irreducible representation of \(P_n\) that is trivial on \(P_{n+1}\). Given a representation \(\pi\) of \(G\), one can ask what strata \(\pi\) contains upon restriction to different groups of the form \(P_n\). Roughly speaking, one can then consider those strata for which the parameter \(n\) is minimal. (A precise statement depends on the normalizations of the various filtrations.) One expects all such strata to have a special form (conjectured by Moy), and then defines a ``fundamental stratum'' or ``minimal \(K\)-type'' to be any representation of a quotient \(P_n/P_{n+1}\) that has this form. One would then like to prove that every representation \(\pi\) contains a fundamental stratum. Such existence theorems were first proved for general linear groups, then for several groups of low rank. \textit{L. Morris} [Duke Math. J. 64, No. 3, 501-553 (1991; Zbl 0799.22009)] addressed the case where \(G\) is a classical group and \(F\) has odd residual characteristic. His construction is explicit, involving self-dual lattice chains. \textit{A. Moy} and \textit{G. Prasad} [Invent. Math. 116, 393-408 (1994; Zbl 0804.22008)] produced a construction that is valid for all simply connected \(G\), and subsequently [Comment. Math. Helv. 71, 98-121 (1996; Zbl 0860.22006)] generalized it for almost all reductive \(G\). In the present paper, the authors return to the case of classical groups, excluding ramified unitary groups. Their aim is to modify the methods of Morris in order to avoid a gap in one of his proofs. Instead of working with the ``fundamental \(G\)-strata'' defined by Morris, they introduce ``fundamental \(C\)-strata''. The latter notion is more rigid than the definitions of Morris or Moy-Prasad. Nevertheless, the authors show that every representation of a group of the kind they consider contains a fundamental \(C\)-stratum. A possible benefit of working with classical groups is that the constructions can be made explicit enough to give one a handle on the set of all fundamental strata that must be considered in order to account for all representations of \(G\). Such results exist for other groups (see the work of \textit{J.-L. Kim} and \textit{F. Murnaghan} [Am. J. Math. 125, No. 6, 1199--1234 (2003; Zbl 1037.22035)]), but only under additional hypotheses.