The fine spectral expansion of the Rankin-Selberg period
Paul Boisseau
TL;DR
This work establishes the fine spectral expansion for the Rankin--Selberg period attached to the Rankin--Selberg spherical variety $X=G/H$ with $G= ext{GL}_n imes ext{GL}_{n+1}$ and $H= ext{GL}_n$. The authors develop a robust framework of regularized Rankin--Selberg periods, extending beyond Arthur-type representations to include non-tempered and residual contributions, and prove a spectral expansion via careful contour-shifting arguments à la Langlands, together with precise control of Eisenstein series and their residues. A key innovation is the regularization of periods for general inducing data $(I,P, ho)$, the parabolic-descent description of these periods, and the analysis of their poles, residues, and functional equations, culminating in an explicit expansion of $J^H(g,f)$ as an integral over unitary contours of relative characters. The results provide a crucial ingredient toward the Jacquet--Rallis trace formula’s fine spectral expansion, with implications for non-tempered Gan–Gross–Prasad-type conjectures and related L-value phenomena. The paper also furnishes detailed bounds for discrete Eisenstein series on $ ext{GL}_n$ and develops the machinery to extend Langlands’ spectral decomposition to broader spaces of automorphic functions, enabling precise Euler-product expressions for regularized periods.
Abstract
We state and prove the spectral expansion of the theta series attached to the Rankin-Selberg spherical variety $(\mathrm{GL}_{n+1} \times \mathrm{GL}_n)/\mathrm{GL}_n$. This is a key result towards the fine spectral expansion of the Jacquet-Rallis trace formula. Our expansion is written in terms of regularized Rankin--Selberg periods for non-tempered automorphic representations, which we show compute special values of $L$-functions. The proof relies on shifts of contours of integration à la Langlands. We also establish two technical but crucial results on bounds and singularities for discrete Eisenstein series of $\mathrm{GL}_n$ in the positive Weyl chamber.
