5
$\begingroup$

I'm reading Casselman's notes "Introduction to the theory of admissible representations of p-adic reductive groups". In chapter 4 "The asymptotic behavior of matrix coefficients", the main result of this chapter states that for some torus elements $a$ (in a sense, those who are "close to zero"), the matrix coefficients $\left\langle\pi(a)v,\tilde{v}\right\rangle$ equlas to $\left\langle\pi_N(a)u,\tilde{u}\right\rangle_N$, where $\pi$ is an admissible representation, $\pi_N$ is its Jacquet module with respect to the unipotent radical $N$, $v\in V_\pi$, $\tilde{v}\in \widetilde{V}_\pi$, and $u$ (respectively, $\tilde{u}$) is the image of $v$ (respectively, $\tilde{v}$) in $V_{\pi,N}$ (respectively, $\widetilde{V}_{\pi,N}$).

In his notes "Remarks on Macdonald’s book on p-adic spherical functions", Casselman refers to this result (Theorem 6.7) and then makes the following remark: "This result says that any matrix coefficient is asymptotically equal to an $A$-finite expression".

Apparently, Casselman means that $\left\langle\pi_N(a)u,\tilde{u}\right\rangle_N$ is a sum of $A$-finite functions. I can't figure out why this is true.

This question is related to the old question Reference request - Jacquet module and asymptotic of matrix coefficients. I would very much appreciate a rather detailed explanation or a good reference for both the p-adic case and the Archimedean case.

$\endgroup$

1 Answer 1

2
$\begingroup$

For the $p$-adic case, the idea is as follows (thanks to Elad for pointing out the direction). Recall (from the unpublished notes by Casselman, Introduction to the theory of admissible representations of p-adic reductive groups) that $k$ is a non-Archimedean locally compact field, $G$ is a group of $k$-rational points of a reductive algebraic group defined over $k$, and $P$ is a parabolic subgroup of $G$ with Levi decomposition $P=MN$.

We note that it is sufficient to show that the function $x\mapsto\left\langle \pi_{N}(x)u,\tilde{u}\right\rangle _{N}$ is an $A_M$-finite function ($A_M$ is the center of the Levi part $M$). Indeed, From Jacquet and Langlands (Lemma 8.1 in Automorphic Forms on $\operatorname{GL}(2)$: Part I, volume 114. Springer, 2006), one can deduce that the space of continuous finite functions on the locally compact abelian group $A_M$ is spanned by functions of the form \begin{equation} \prod_{i=1}^{r}\chi_i(a_i)\left\lvert a_i\right\rvert^{p'_i}\log_q ^{p_i}\left\lvert a_i\right\rvert, \end{equation} where $r$ is such that $A_M\cong k^r$, $\left(p'_{1},\dotsc,p'_{r}\right)\in \mathbb{R}^r$, $\left(p_{1},\dotsc,p_{r}\right)\in \mathbb{Z}^r_{\ge 0}$, and for all $1\leq i\leq r$, $\chi_i:k^\times \to \mathbb{C}^\times$ are unitary characters.

Now, in order to prove that the function $x\mapsto\left\langle \pi_{N}(x)u,\tilde{u}\right\rangle _{N}$ is an $A_M$-finite function, we use the following technical lemma.

  • Lemma. Let $R$ be a group with center $Z\left(R\right)\cong K\times\mathbb{Z}^{r}$, where $K$ is a compact group. Let $\left(H,\sigma\right)$ be a (complex) smooth $R$-module of finite length and let $v\in H$. Then, the $Z\left(R\right)$-module generated by $v$ is finite dimensional.

The Jacquet module is a smooth $G$-module of finite length (See Theorems 3.3.1 and 6.3.10 in the abovementioned unpublished notes by Casselman). Hence, we apply the lemma with $R=M$ (with $Z(R)=A_M$), $H=V_N$, $\sigma=\pi_N$, and $v=u\in V_{N}$. This gives that $U:=\left\{ \pi_{N}\left(a\right)u|\ a\in A_{M}\right\} $ is of finite dimension. Let $\left\{ \pi_{N}\left(b_{1}\right)u,\dotsc, \pi_{N}\left(b_{\ell}\right)u\right\}$ be a basis of $U$. Then, \begin{equation} \pi_{N}\left(a\right)u=\sum_{i=1}^{\ell}c_{i}(a)\pi_{N}\left(b_i\right)u. \end{equation} Therefore, \begin{equation} \left\langle \pi_{N}\left(ma\right)u,\tilde{u}\right\rangle _{N}=\sum_{i=1}^{\ell}c_{i}\left(a\right)\left\langle \pi_{N}\left(m b_i\right)u,\tilde{u}\right\rangle _{N}. \end{equation} For more details, as well as the proof of the lemma, see Hazan - A Note on the Asymptotic Expansion of Matrix Coefficients over $p$-adic Fields.

For the Archimedean case, one can find an asymptotic expansion of the matrix coefficients, as a finite sum of finite functions, in Casselman's paper Jacquet modules for real reductive groups (see the Lemma in Section 5, Proceedings of the International Congress of Mathematicians (Helsinki, 1978)).

$\endgroup$
4
  • $\begingroup$ I would disagree for links to a note on the arXiv ! SE is more reliable than a personal webpage, but the arXiv is meant to be a repository of research in a way that SE isn't. $\endgroup$ Dec 2, 2022 at 19:56
  • $\begingroup$ You are so right! Sorry and thank you. Let me edit my answer and write the details here asap. $\endgroup$
    – Zahi Hazan
    Dec 3, 2022 at 21:38
  • $\begingroup$ If this answers your question, then don't forget to accept it! $\endgroup$
    – LSpice
    Dec 4, 2022 at 15:16
  • 1
    $\begingroup$ Good point @LSpice. Thanks for your edits as well! $\endgroup$
    – Zahi Hazan
    Dec 4, 2022 at 15:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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