Gelfand pairs \((\text{Sp}_{4n}(F),\text{Sp}_{2n}(E))\) (Q710482): Difference between revisions

The author proves that the symmetric pairs \((\mathrm{Sp}_{4n}(F),\mathrm{Sp}_{2n}(E))\) and \((\mathrm{GSp}_{4n}(F),\mathrm{GSp}_{2n}(E)^{\circ})\) are Gelfand pairs where \(E\) is a two-dimensional semisimple algebra over \(F\) (a nonarchimedean field of characteristic zero) and \(\mathrm{GSp}_{2n}(E)^{\circ}\) denotes the subgroup of \(\mathrm{GSp}_{2n}(E)\) having similitude in \(F\). In many previous articles that prove pairs of \(p\)-adic groups \((G,H)\) are Gelfand pairs, the general argument involves showing that distributions on \(G\) that are invariant under the adjoint action of \(H\) are invariant under some anti-involution \(\sigma\) of \(G\) for which \(\sigma(H)=H\). Perhaps the most heralded recent example of a paper utilizing these methods is the work of \textit{A. Aizenbud, D. Gourevitch, S. Rallis} and \textit{G. Schiffmann} [Ann. Math. (2) 172, No. 2, 1407--1434 (2010; Zbl 1202.22012)]. Similar techniques are also used to prove the uniqueness of various models for representations of various reductive groups as in \textit{M. J. Heumos} and \textit{S. Rallis} [Pac. J. Math. 146, No. 2, 247--279 (1990; Zbl 0752.22012)], \textit{E. M. Baruch} and \textit{S. Rallis} [J. Lond. Math. Soc., II. Ser. 62, No. 1, 183--197 (2000; Zbl 0953.22019)], and \textit{J. D. Adler} and \textit{D. Prasad} [Isr. J. Math. 153, 221--245 (2006; Zbl 1137.22009)]. The paper under review offers a separate method of proof that follows from the work of \textit{A. Aizenbud, D. Gourevitch} and \textit{E. Sayag} [Duke Math. J. 149, No. 3, 509--567 (2009; Zbl 1221.22018)]. In this case, the author recalls the notion of descendants of the pair \((G,H,\theta)\), where \(\theta: G\rightarrow G\) is given by \(\theta(g)=\sigma(g^{-1})\) for the involution \(\sigma\) mentioned above. These descendants \((G_{x}, H_{x}, \theta|_{G_x})\) are indexed by a certain set of semi-simple elements \(x\) in \(G\). On must check that these descendants satisfy a regularity condition (recalled in Definition 2.3 of the paper under review) that is slightly technical and regards \(H_x\) equivariant distributions on a certain subset of the Lie algebra of \(G_x\). Section 2 then determines a complete list of descendants for the symmetric pairs \((\mathrm{Sp}_{4n}(F),\mathrm{Sp}_{2n}(E))\) and \((\mathrm{GSp}_{4n}(F),\mathrm{GSp}_{2n}(E)^{\circ})\) where \(E\) and \(F\) are as in the first paragraph. Section 3 demonstrates the regularity of most of these descendants. In the case that \(E\) is a quadratic extension, the regularity condition is shown for all descendants. Thus the generalized Harish-Chandra descent of Aizenbud, Gourevitch, and Sayag (see above) establishes that these symmetric pairs are Gelfand pairs. In the case of \(E\) being the split quadratic algebra, the author establishes the regularity of the various descendants (thus proving the symmetric pair is a Gelfand pair) when \(n\leq 2\). When \(n\geq 3\), the author then indicates that the only obstruction to proving that these symmetric pairs are Gelfand pairs is establishing the regularity of symmetric pairs of the form \((\mathrm{Sp}_{2(m_1+m_2)}(F),\mathrm{Sp}_{2m_1}(F)\times\mathrm{Sp}_{2m_2}(F))\) which appear as descendants in these unresolved cases.