On unitarizability in the case of classical \(p\)-adic groups (Q2314004)

The author discusses some approaches to the problem of the unitarizability of irreducible (complex) representations of \(p\)-adic classical groups. In the excellent introduction he discusses the (solved) unitarizability problem for general linear groups, and compares it to the (unsolved) analogous problem for the classical groups. The main idea is to relate the unitarizability of an irreducible representation \(\pi\) of a \(p\)-adic classical group with the unitarizability of all of its Jantzen's components. Roughly, the Jantzen component has its cuspidal support only on the copies of one (twisted) cuspidal representation of a general linear group and a cuspidal representation of a smaller classical group. Thus, the author formulates two conjectures. The first one, the so called preservation of unitarizability question, claims that \(\pi\) is unitarizable if and only if all the Jantzen components are unitarizable. The other, the independence of unitarizability, examines the unitarizable representations supported only on one Jantzen line. Let say we have two such lines. One is ``spanned'' by an unitarizable irreducible cuspidal representation \(\rho\) of a general linear group, and by a cuspidal irreducible representation \(\sigma\) of a classical group. Say that the induced representation (we use the Zelevinsky notation) \(\nu^{s}\rho \rtimes \sigma \) reduces for (necessarily unique) \(\alpha_{\rho,\sigma} \in \frac{1}{2}\mathbb{Z}_{\ge 0}\). If we look at the set of all the unitarizable representations supported on the \(\alpha_{\rho,\sigma}+\mathbb{Z}\) twists of \(\rho\) (and \(\sigma\)), is there a bijection from that set to the analogous set obtained from some other \(\rho'\) and \(\sigma'\) but with the property \(\alpha_{\rho,\sigma}=\alpha_{\rho',\sigma'}?\) The author gives some support for these conjectures in the case of generic unitarizable representations of the classical groups, as well as for the unramified unitarizable representations of such groups. He also proves part of the preservation of the unitarizability conjecture in the case in which one of the Jantzen components has the same infinitesimal character as the generalized Steinberg representation. For the entire collection see [Zbl 1401.11005].