Multiplicity one theorems for the generalized doubling method (with an appendix by Avraham Aizenbud and Dmitry Gourevitch) (Q6172695): Difference between revisions

Let \(G\) be one of the split reductive groups \(\mathrm{GL}_c\), \(\mathrm{SO}_c\), \(\mathrm{GSpin}_c\) and \(\mathrm{Sp}_{c}\), and \(F\) be a global field with ring of adèles \(\mathbb{A}\). Let \(\pi\) be a cuspidal automorphic representation of \(G(\mathbb{A})\) and \(\tau\) be a cuspidal automorphic representation of \(\mathrm{GL}_k(\mathbb{A})\) for some \(k\). Then \textit{Y. Cai} et al. [Invent. Math. 217, No. 3, 985--1068 (2019; Zbl 1480.11060)] (see also [\textit{Y. Cai} et al., ``Doubling constructions: local and global theory, with an application to global functoriality for non-generic cuspidal representations'', Preprint, \url{arXiv:1802.02637}]) showed that there is a Rankin-Selberg integral, the generalized doubling integral, that represents the tensor product \(L\)-function \(L(s,\pi\times \tau)\). The goal of this important paper is to prove the local multiplicity one theorem arising from this integral representation. This is the statement that a natural Hom space that arises after unfolding is at most one dimensional. The Hom space depends on a complex parameter \(s\) and this is established outside a discrete subset of \(s\). The proof requires the analysis of distributions supported on certain double cosets, and mirrors (using root exchange and the theory of Bernstein-Zelevinsky derivatives) the global unfolding, but with additional complications. In the non-archimedean local setting one must consider non-supercuspidal representations of \(G(F_\nu)\). For archimedean fields, the number of double cosets is uncountable and there are more distributions due to transverse derivatives. The authors' proof for archimedean fields makes use of a criteria for the vanishing of vector-valued distributions on smooth manifolds that is established in an appendix by Aizenbud and Gourevitch using tools from functional analysis and generalizing work of \textit{J. A. C. Kolk} and \textit{V. S. Varadarajan} [Indag. Math., New Ser. 7, No. 1, 67--96 (1996; Zbl 0892.22010)]. The authors explain how their main result allows one to define local \(\gamma\) and \(\varepsilon\) factors and a local \(L\)-function for pairs \((\pi_\nu,\tau_\nu)\). (See also [\textit{Y. Cai} et al., Geom. Funct. Anal. 32, No. 6, 1233--1333 (2022; Zbl 1511.11050)] where these factors are analyzed and shown to satisfy a list of properties first established in the generic case by \textit{F. Shahidi} [Ann. Math. (2) 132, No. 2, 273--330 (1990; Zbl 0780.22005)].) Under certain hypotheses, the authors also develop analogous results for metaplectic covering groups with suitably matching covers. In addition, they explain an application of their main result to analytic aspects of the unfolding of the generalized doubling integral. \par In the non-archimedean case the multiplicity one result for supercuspidal representations was independently obtained by \textit{Y. Cai} [Math. Z. 297, No. 3--4, 1075--1104 (2021; Zbl 1478.11077)]. The twisted doubling integral may be combined with the Converse Theorem of \textit{J. W. Cogdell} and \textit{I. I. Piatetski-Shapiro} [J. Reine Angew. Math. 507, 165--188 (1999; Zbl 0912.11022)] to give a new proof of the functorial lifting from \(G\) to a suitable \(\mathrm{GL}_N\) (see [\textit{Y. Cai} et al., loc. cit.]). The results of this paper play a significant role in that work.