Uniqueness of twisted linear periods and twisted Shalika periods (Q2301199)

Let \(k\) be a local field of characteristic \(0\) and let \(\psi_k : k \rightarrow \mathbb{C}^{\ast}\) be a non-trivial, unitary character. Consider the Shalika subgroup \(S_n(k)\) of \(\mathrm{GL}_{2n}(k)\) which is the semidirect product \[ S_n(k) = \bigg\{ \begin{pmatrix} a & b \cr 0 & a \end{pmatrix} : a \in\mathrm{GL}_n(k), b \in M_n(k) \bigg\}. \] The unitary character \(\psi_k\) induces a character \(\psi_{S_n}\) on \(S_n(k)\) as the action on \(\begin{pmatrix} a & b \cr 0 & a \end{pmatrix}\) being given as \(\psi_k(tr(b))\). Let \(\pi\) be an admissible, smooth representation of \(\mathrm{GL}_m(k)\) -- for \(k\) Archimedean, recall this means it is Fréchet, smooth, of moderate growth, with its Harish-Chandra module of finite length. The authors prove: Theorem A. For each irreducible, smooth \(\pi\) on \(\mathrm{GL}_{2n}(k)\), the space \(\mathrm{Hom}_{S_n(k)}(\pi, \psi_{S_n})\) has dimension \(\leq 1\). When the restriction of \(\psi_{S_n}\) to the diagonal subgroup of \(S_n(k)\), this theorem has been proved earlier (in [\textit{H. Jacquet} and \textit{S. Rallis}, Compos. Math. 102, No. 1, 65--123 (1996; Zbl 0855.22018)] for \(k\) nonarchimedean, and in [\textit{A. Aizenbud} et al., Pac. J. Math. 243, No. 2, 201--212 (2009; Zbl 1178.22014)] for \(k\) archimedean). Non-zero elements of the space in Theorem A are called local Shalika periods of \(\pi\) and the uniqueness of the local Shalika period for \(\pi\) asserted in Theorem A is deduced by the authors from Shalika zeta integrals and the following theorem. Theorem B. Let \(\pi\) be an irreducible, smooth representation of \(\mathrm{GL}_{2n}(k)\). Then, for all but countably many characters (finitely many in the nonarchimedean case) \(\chi\)'s of \(\mathrm{GL}_n(k) \times\mathrm{GL}_n(k)\), the space \(\mathrm{Hom}_{\mathrm{GL}_n(k) \times\mathrm{GL}_n(k)}(\pi, \chi)\) has dimension \(\leq 1\). Once again, the uniqueness of the local linear period of \(\pi\) (a non-zero element of the above space) was proved for trivial \(\chi\) earlier -- by Jacquet and Rallis when \(k\) is non-archimedean, and by Aizenbud and Gourevitch when \(k\) is archimedean. The original motivation of Theorem B is that it is shown by the second author to imply a non-vanishing result on \(L\)-functions that is critical to the arithmetic study of special values of \(L\)-functions for \(\mathrm{GSpin}_{2n+1}\).