Let $f$ be a Maass cusp form of $\Gamma_0(N)$ on the upper half plane with character $\chi$ mod $N$ and eigenvalue $1/4+\mu^2$.

What is the complete $L$-function of the Rankin-Selberg product $L(s,f\times \tilde f)$? I am having trouble writing down the local part at $p|N$. I can't find good references either.


See here for the answer (in particular, it's in section 1 and 2). Basically, the answer isn't pretty - it depends on the local components of the automorphic representation $\pi$ of $\mathrm{GL}_2(\mathbb{A}_{\mathbb{Q}})$ associated to the Maass form $f$.

So if $p \nmid N$, then the local component $\pi_p$ of $\pi$ is a principal series representation $\omega_{p,1} \boxplus \omega_{p,2}$, where $\omega_{p,1}, \omega_{p,2}$ are unramified characters of $\mathbb{Q}_p^{\times}$. We have that \[L_p(s,\pi_p) = \frac{1}{(1 - \omega_{p,1}(p) p^{-s})(1 - \omega_{p,2}(p) p^{-s})}\] where $\omega_{p,1}(p) + \omega_{p,2}(p) = \lambda_f(p)$, the Hecke eigenvalue of $f$ at $p$, and $\omega_{p,1}(p) \omega_{p,2}(p) = \chi(p)$. Then the local Rankin-Selberg $L$-function is \[L_p(s,\pi_p \times \widetilde{\pi_p}) = \frac{1}{(1 - |\omega_{p,1}(p)|^2 p^{-s})(1 - \omega_{p,1}(p) \overline{\omega_{p,2}(p)} p^{-s})(1 - \overline{\omega_{p,1}(p)} \omega_{p,2}(p) p^{-s})(1 - |\omega_{p,2}(p)|^2 p^{-s})}.\]

If $p \mid N$ and $\pi_p$ is a principal series representation $\omega_{p,1} \boxplus \omega_{p,2}$, where at least one of $\omega_{p,1}, \omega_{p,2}$ is ramified, then the same result holds as for the case $p \nmid N$ with the minor change that if $|\omega_{p,1}|^2$ is a ramified character of $\mathbb{Q}_p^{\times}$, we replace $|\omega_{p,1}(p)|^2$ with $0$, and similarly if $\omega_{p,1} \overline{\omega_{p,2}}$, $\overline{\omega_{p,1}} \omega_{p,2}$, or $|\omega_{p,2}|^2$ is ramified (see Jeremy's comment below). Note that if both $\omega_{p,1}$ and $\omega_{p,2}$ are ramified then $p^2 \mid N$. This is part of Proposition 1.4 of the linked paper.

If $p \mid N$ and the local component $\pi_p$ of the automorphic representation is supercuspidal (which can only happen if $p^2 \mid N$), then $\lambda_f(p) = 0$ and \[L_p(s,\pi_p) = 1,\] and either \[L_p(s,\pi_p \times \widetilde{\pi_p}) = \frac{1}{1 - p^{-s}}\] or \[L_p(s,\pi_p \times \widetilde{\pi_p}) = \frac{1}{1 - p^{-2s}}.\] The former occurs if $\pi_p \ncong \pi_p \otimes \eta_p$, where $\eta_p$ is the unramified quadratic character of $\mathbb{Q}_p^{\times}$, while the latter occurs if $\pi_p \cong \pi_p \otimes \eta_p$. This is Corollary 1.3 of the linked paper.

If $p \mid N$ and the local component $\pi_p$ is a Steinberg representation $\omega_p \mathrm{St}_p$, where $\omega_p$ is a unitary character of $\mathbb{Q}_p^{\times}$, then $\widetilde{\pi_p} = \omega_p^{-1} \mathrm{St}_p$. We have that \[L_p(s,\pi_p) = L_p\left(s + \frac{1}{2}, \omega_p\right)\] and \[L_p(s,\pi_p \times \widetilde{\pi_p}) = \zeta_p(s + 1) \zeta_p(s) = \frac{1}{(1 - p^{-s-1})(1 - p^{-s})}.\] This is part of Proposition 1.4. Note that if $\omega_p$ is unramified, we must have that $p \parallel N$, $\lambda_f(p) = \omega_p(p) p^{-1/2}$, and so \[L_p(s,\pi_p) = \frac{1}{1 - \omega_p(p) p^{-s - 1/2}},\] If $\omega_p$ is ramified, then $p^2 \mid N$, $\lambda_f(p) = 0$, and \[L_p(s,\pi_p) = 1.\] In particular, if $N$ is squarefree and $\chi$ is the principal character, then for $p \mid N$, $\pi_p$ must be $\omega_p \mathrm{St}_p$ with $\omega_p$ the trivial character, so $\lambda_f(p) = p^{-1/2}$.

Finally, the archimedean $L$-function is \[L_{\infty}(s,\pi_{\infty}) = \pi^{-s - \kappa} \Gamma\left(\frac{s + i\mu + \kappa}{2}\right) \Gamma\left(\frac{s - i\mu + \kappa}{2}\right),\] where $\kappa = 0$ if $f$ is even and $1$ if $f$ is odd, and \[L_{\infty}(s,\pi_{\infty} \times \widetilde{\pi_{\infty}}) = \pi^{-2s} \Gamma\left(\frac{s + 2i\mu}{2}\right) \Gamma\left(\frac{s}{2}\right)^2 \Gamma\left(\frac{s - 2i\mu}{2}\right).\]

The completed $L$-function $\Lambda(s,f \times \widetilde{f})$ is then defined to be \[\Lambda(s,f \times \widetilde{f}) = L_{\infty}(s,\pi_{\infty} \times \widetilde{\pi_{\infty}}) \prod_p L_p(s,\pi_p \times \widetilde{\pi_p}).\] Theorem 2.2 of the linked paper states that this extends meromorphically to the entire complex plane with only simple poles at $s = 0$ and $s = 1$.

| cite | improve this answer | |
  • $\begingroup$ A minor clarification: on the subject of how we should interpret $|\omega_{1,p}(p)|^{2}$, I think we should let $\chi = \omega_{1,p} \overline{\omega_{1,p}}$ and then if interpret $|\omega_{1,p}(p)|^{2}$ as being equal to $\chi(p)$, if $\chi$ is unramified. (Note that we can have $\chi$ be unramified even if $\omega_{1,p}$ is.) $\endgroup$ – Jeremy Rouse Mar 12 '15 at 12:21
  • $\begingroup$ @JeremyRouse, yes, this is completely correct! $\endgroup$ – Peter Humphries Mar 12 '15 at 13:06
  • $\begingroup$ Very good summary. Note that for $\pi$ coming from a Maass form of odd weight, the archimedean factors of $L(s,\pi_\infty)$ are slightly different. Good references are Proposition 8.1 in Duke-Friedlander-Iwaniec: The subconvexity problem for Artin L-functions (Inventiones, 2002) and Section 11.14 in Goldfeld-Hundley: Automorphic Representations and L-Functions for the General Linear Group, Volume 1 (Cambridge University Press, 2011). $\endgroup$ – GH from MO Mar 16 '15 at 14:38
  • $\begingroup$ @GHfromMO, thanks, I think it should be all correct now (there was a missing $\pi^{-\kappa}$ term). Also I'm not sure why this was just downvoted... $\endgroup$ – Peter Humphries Mar 17 '15 at 18:11
  • 1
    $\begingroup$ To add to this 3 year old comment chain, the Proposition 8.1 of Duke-Friedlander-Iwaniec has a slight error which affects if the $\kappa = 1$ shift in the gamma factor occurs with $\mu$ or $-\mu$. I corrected it in Section 12 of arxiv.org/pdf/1710.03624.pdf $\endgroup$ – Matt Young Mar 19 '18 at 21:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.