Test vectors for Rankin-Selberg \(L\)-functions (Q2288302)

Let \(F\) be a non-Archimedean local field and let \(m<n\). This article considers explicit test vectors for local Rankin-Selberg \(L\)-functions \(L(s,\pi\boxtimes \tau)\) where \(\pi\) and \(\tau\) are irreducible admissible representations of \(\mathrm{GL}_n(F)\) and \(\mathrm{GL}_m(F)\). More precisely, if \(\pi\) and \(\tau\) possess non-vanishing Whittaker models associated to an additive character \(\psi\), one considers the local zeta integral \[ \Psi(s;W_\pi,W_\tau')=\int_{U_m(F)\backslash \mathrm{GL}_m(F)}W_\pi\left(\begin{array}{cc} g & \\ & I_{n-m} \end{array}\right)W'_\tau(g)||\det(g)||^{s-\frac{n-m}{2}}dg, \] where \(W_\pi\) (resp. \(W'_\tau\)) are Whittaker functions associated to \(\pi\) (resp. \(\tau\)). It is well known that this integral converges for \(\mathrm{Re}(s)>> 1\), and that the image spans a fractional ideal \(\mathbb{C}[q^s,q^{-s}]L(s,\pi\boxtimes\tau)\) of the ring \(\mathbb{C}[q^s,q^{-s}]\). Here, \(L(s,\pi\boxtimes\tau)\) is the local Rankin-Selberg \(L\)-factor of Jacquet-Piatetski-Shapiro-Shalika. A test vector is a pair \((W_\pi,W_\tau')\) such that \[\Psi(s;W_\pi,W_\tau')=L(s,\pi\boxtimes\tau).\] One defines the ``naive'' Rankin-Selberg \(L\)-factor \(L(s,\pi\times \tau)\) in terms of the Langlands parameters of \(\pi\) and \(\tau\), and in general there is a relation between the two \(L\)-factors \[L(s,\pi\times \tau)=P_{\pi,\tau}(q^{-s})L(s,\pi\boxtimes \tau)\] for some non-zero polynomial \(P_{\pi,\tau}(X)\in \mathbb{C}[X]\) depending on \(\pi\) and \(\tau\). The main result of this article is to provide an explicit pair of Whittaker functions \((W_\pi,W_\tau')\) satisfying \[ \Psi(s;W_\pi,W_\tau')=L(s,\pi\times \tau) \] for any irreducible admissible \(\pi\) and \(\tau\). Thus, the authors produce test vectors whenever \(L(s,\pi\times \tau)=L(s,\pi\boxtimes \tau)\). The method relies on the theory of essential vectors for irreducible admissible representations of \(\mathrm{GL}_n(F)\) and \(\mathrm{GL}_m(F)\). When \(\tau\) is unramified, it was previously known that essential vectors give test vectors and that \(L(s,\pi\times \tau)=L(s,\pi\boxtimes \tau)\), so there is nothing to prove. When \(\tau\) is ramified, it is known that the zeta integral of the essential vectors vanishes. To remedy this, the authors average the essential vector of \(\pi\) over a finite number of unipotent elements controlled by the conductor of \(\tau\) and the natural embedding of \(\mathrm{GL}_m(F)\) into \(\mathrm{GL}_n(F)\) as an upper left block. Pairing this averaged vector with the essential vector of \(\tau\) gives the \(L\)-factor up to an explicit scalar depending on \(\tau\), so rescaling gives the result.