Endo-parameters for \(p\)-adic classical groups (Q2225237)

We can probe representations of a reductive \(p\)-adic group using its compact-open subgroups. Given an irreducible representation \(\rho\) of the subgroup \(K\), one simply asks if the restriction of the representation to \(K\) has nontrivial \(\rho\)-isotypic component, in which case we say that the representation \textit{contains} \((K,\rho)\). Surprisingly, this simple-minded procedure completely controls the irreducible representations of the ambient group in many cases. In their seminal study of the general linear group \textit{C. J. Bushnell} and \textit{P. C. Kutzko} [The admissible dual of \(\mathrm{GL}(N)\) via compact open subgroups. Princeton, NJ: Princeton University Press (1993; Zbl 0787.22016)] isolated a collection of such pairs \((K,\rho)\) which they called \textit{simple types}. They showed that every irreducible representation of \(\mathrm{GL}_n\) contains a simple type and moreover that the irreducible representations containing a given simple type could be classified by irreducible representations of related Hecke algebras. Simple types are constructed from an arithmetic input called a \textit{stratum}. From this object one forms a chain \(H^1 \trianglelefteq J^1 \trianglelefteq J\) of compact-open subgroups of \(G\). There is a special class of \textit{simple characters} \(\theta\) of \(H^1\). A simple character \(\theta\) canonically produces an irreducible representation \(\eta\) of \(J^1\), distinguished by the requirement that \(\eta|_{H^1}\) contain \(\theta\). One carefully extends \(\eta\) to a representation \(\kappa\) of \(J\), a so-called \textit{\(\beta\)-extension} whose precise definition is subtle. Finally, one inflates to \(J\) a cuspidal representation of the quotient \(J/J^1\), a finite reductive group. Let \(\sigma\) denote this inflation. The simple types are then the pairs \((J,\sigma\otimes\kappa)\). These constructions depend on many choices, but many of the choices are illusory. Generally speaking, one would like to impose natural equivalence relations on the choices, to the effect that equivalent choices result in the same representation theory, and to then understand the equivalence classes. When the choice is a representation, the equivalence relation is intertwining or some weakening of it and one would like the equivalence classes to be conjugacy classes, what is called an ``intertwining implies conjugacy'' theorem. Indeed, this hope entails that the relevant notion of equivalence form an equivalence relation at all, since the relation of intertwining is not a priori transitive. For both simple types and simple characters, the relevant notion of equivalence is intertwining and intertwining implies conjugacy. In their study of base change, \textit{C. J. Bushnell} and \textit{G. Henniart} [Publ. Math., Inst. Hautes Étud. Sci. 83, 105--233 (1996; Zbl 0878.11042)] introduced a weaker notion of equivalence for simple characters called \textit{endo-equivalence}, in which one allows variation of both the stratum and the ambient general group. Properly this an equivalence relation not on simple characters, but on the so-called \textit{potentially simple} (ps) characters, which are objects cooked up to allow such variation. From the point of view of the local Langlands correspondence, endo-equivalence has the following interpretation. In the construction of simple types there is a dichotomy between positive depth, where the simple character \(\kappa\) lives, and depth zero, where the cuspidal representation \(\sigma\) lives. On the Galois side, \(\sigma\) has something to do with \(L\)-parameters trivial on the wild inertia subgroup, and \(\kappa\) has something to do with \textit{wild \(L\)-parameters}, the restrictions of \(L\)-parameters to wild inertia. It is expected that wild \(L\)-parameters correspond to \textit{endo-parameters}, formal linear combinations of endo-equivalence classes. The paper under review fits into the long program of generalizing this work to the classical groups. By definition, the classical groups are the fixed points of an (outer) involution of \(\mathrm{GL}_n\). Very roughly, one can study such groups by the same methods as for the general linear group at the cost of incorporating the involution into the development of the theory: vector spaces become (\(\varepsilon\)-)hermitian spaces, simple characters become self-dual simple characters, and so on. The truth, of course, is much more complicated. To give just one example, in the group \(\mathrm{GL}_n\), the maximal compact (modulo center) tori are the units in degree-\(n\) field extensions, which are indecomposable as algebraic groups. But the maximal compact tori of classical groups can decompose as direct products of smaller tori. This fact of life led the third author to introduce in his thesis work the class of \textit{semisimple characters} for both the general linear and classical groups, generalizing Bushnell and Kutzko's simple characters. In the authors' own words, the core technical accomplishment of the paper under review is to ``generalize Bushnell and Henniart's notions of ps-character and endo-equivalence to self-dual pss-characters and endo-equivalence for \(p\)-adic classical groups, and along the way to the semisimple setting of pss-characters and endo-equivalence for \(p\)-adic general linear groups.'' Here ``pss'' abbreviates \textit{potentially semisimple}. Their generalization is much more than the simple-minded imitation of earlier definitions; it requires a full development of the theory. This work is then leveraged to prove that intertwining implies conjugacy for cuspidal types, generalizing Bushnell and Kutzko's work. Unlike the earlier work of Bushnell and Kutzko mentioned above, here the algebraically-closed coefficient field may be taken to have any characteristic prime to \(p\), a satisfyingly general hypothesis. I hope that future work will relax the assumption that \(p\neq2\), where there are surely many interesting new phenomena to be discovered. I appreciated the authors' effort to motivate the definitions and technical developments. In light of the notational complexity, failure to do so would have rendered the paper unreadable. Concrete examples and an index of notation would have made the paper more readable still.