Let $P_0$ be a minimal parabolic subgroup of a connected, reductive group $G$ over a $p$-adic field $k$. Let $P$ be a parabolic subgroup containing $P_0$ with Levi decomposition $P = MN$. Let $N^-$ be a group such that $P^- = MN^-$ is opposite to $P$.

Let $\pi$ be a smooth, irreducible representation of $M$, and consider a function $f$ in the space of the induced representation $\sigma = \operatorname{Ind}_{P}^G \pi $. Then $f$ is locally constant, and there is a compact set $Z$ such that if $f(x) \neq 0$, then $px \in Z$ for some $p \in P$, i.e. $f$ is compactly supported modulo $P$.

I have seen in a few talks and papers the practice of fixing a function $f \in \sigma$ which vanishes outside of the open set $PN^-$. I was wondering a couple of things:

1 . Are are there "many" such functions? How do we know that they exist at all?

2 . If $f$ vanishes outside $PN^-$, since $N^-$ is the union of its open compact subgroups, is it possible to find a compact open subgroup $N_0$ of $N^-$ such that if $f(x) \neq 0$, then $px \in N_0$ for some $p \in P$?

Question 2 seems like a natural thing to expect, but I could not prove it. I thought I might take the image in $P \backslash G$ of the support of $f$, which is compact, and cover it by the images of the open compact subgroups $N_0$. The problem is that I don't think the images of $N_0$ are open in $P \backslash G$, since they are not open in $G$.

  • $\begingroup$ Surely you don't mean $P^- = P N^-$, but $P^- = M N^-$? $\endgroup$
    – LSpice
    Mar 12, 2018 at 15:29
  • $\begingroup$ Sorry, can you explain more on why $PN^-$ being a direct product implies (1)? (2) makes sense. $\endgroup$
    – D_S
    Mar 12, 2018 at 15:34
  • $\begingroup$ This was extremely helpful, thank you very much $\endgroup$
    – D_S
    Mar 12, 2018 at 17:47
  • $\begingroup$ I have moved my comments to an answer. $\endgroup$
    – LSpice
    Mar 12, 2018 at 17:56

1 Answer 1


Answer moved from the comments.

  1. They exist because $P \times N^- \to G$ is an open embedding (even on the level of varieties). This implies that, for $f \in \mathrm C^\infty_{\mathrm c}(N^−; \sigma)$, the function $p n^− \mapsto \pi(p)f(n^−)$ has image in the appropriate function space (for $n^− \in N^−$, the image of $f$ is constant on $\operatorname{stab}_P(f(n^−))N^−_0$, where $N^−_0$ is an open subgroup of $N^−$ such that $f$ is constant on $n^− N^−_0$).

  2. Not $N_0$ (which I think it would be better to call $N^−_0$, as above) but $P N_0$ is open, because of (1), so the image of $N_0$ in $P\backslash G$ is open.


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.