Let $G_1 \subset G$ be the rational points of $p$-adic reductive groups sharing the same derived group. There are some well known results relating representations of $G_1$ to representations of $G$, for example in Marko Tadić's paper Notes on representations of non-Archimedean $\operatorname{SL}(n)$. Consider the following theorem:

Theorem: Let $\pi_1$ be an irreducible admissible representation of $G_1$. There exists an irreducible admissible representation $\pi$ of $G$ such that $\pi_1$ is isomorphic to a subrepresentation of $\pi\rvert_{G_1}$. If $\pi_1$ is generic and supercuspidal, then $\pi$ can be chosen to be generic and supercuspidal.

The proof involves finding a finite rank free group $S_1$ inside $Z(G)$ such that $S_1 \cap G_1 = 1$, and $G/S_1G_1$ is compact. One extends $\pi$ to $S_1G_1$ by making it trivial on $S_1$, and then induces to $G$.

I was wondering how much control we may have for the central character. If $\pi_1$, $\pi_1'$ are irreducible admissible representations of $G_1$ having the same central character, is it always possible to choose the $\pi$, $\pi'$ in the theorem to also have the same central character? If anyone knows a reference that deals with this sort of question, it would be greatly appreciated.

  • $\begingroup$ Note: Tadić's last name is spelled as so, not with a 'k'. I have edited accordingly, since this is on the front page anyway. $\endgroup$
    – LSpice
    May 3 at 13:37

1 Answer 1


I think I have something that works, modulo two mild assumptions:

1 . $Z(G_1) \subseteq Z(G)$.

2 . Given a smooth character $\omega_1$ of $Z(G_1)$, there exists an extension $\omega$ of $\omega_1$ to a smooth character of $Z(G)$.

(Sketch): Given a representation $\pi_1$ of $G_1$ with central character $\omega_1$, extend $\omega_1$ to a character $\omega$ of $Z(G)$. Define a representation $\pi^{\vee}$ of $Z(G)G_1$ by $\pi^{\vee}(zg_1) = \omega(z)\pi(g_1)$ for $z \in Z(G)$ and $g_1 \in G_1$. This is well defined. Now, $G/Z(G)G_1$ is a finite abelian group. Induce to get $\operatorname{Ind}_{Z(G)G_1}^G \pi^{\vee}$.

The map $f \mapsto f(1)$ should given an isomorphism of $Z(G)G_1$-representations for an irreducible $G$-subrepresentation $\pi$ of $\operatorname{Ind}_{Z(G)G_1}^G \pi^{\vee}$. Let $\omega_{\pi}$ be the central character of $\pi$. For $f$ in the space of $\pi$ with $f(1) \neq 0$, we have

$$\omega_{\pi}(z)f(1) = f(z) = \omega(z)f(1)$$

and so the central character of $\pi$ is just the character $\omega$. This if we start with $\pi_1, \pi_1'$ with the same central character $\omega_1 = \omega_{\pi_1} = \omega_{\pi_2}$, the central characters of the corresponding $\pi_1$ and $\pi_2$ are the same, equal to $\omega$.

  • $\begingroup$ Smooth characters $\omega_1$ always extend. Choose an open subgroup $U$ of $Z(G)$ whose intersection with $Z(G_1)$ is contained in the kernel of $\omega_1$. Now use the fact that $\mathbb C^\times$ is injective as an Abelian group to define a homomorphism $\omega : Z(G)/U \to \mathbb C^\times$ extending $\omega_1 : Z(G_1)/(U \cap Z(G_1)) \to \mathbb C^\times$. In general, even if $Z(G_1)$ is not contained in $Z(G)$, why not just extend to $Z(G)$ the restriction to $Z(G) \cap Z(G_1)$ of $\omega_1$, then proceed as you do here? $\endgroup$
    – LSpice
    May 3 at 13:40
  • $\begingroup$ But I am confused; you use $\pi^\vee$ and $\pi$, which I would normally take to mean contragredient representations, but they aren't even defined on the same group. $\endgroup$
    – LSpice
    May 3 at 13:42

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.