6
$\begingroup$

Let $G$ be a compact group, $H$ a normal open subgroup, and $K$ a $p$-adic field (so that not all $G$-reps with coefficients in $K$ are semisimple). Let $V$ be a finite-dimensional topological $K$-vector space with continuous action of $H$, such that $V$ is irreducible as an $H$-representation. Is the induced representation $\text{Ind}_H^G(V)$ semisimple?

In this specific context, I'm thinking of $G$ as the Galois group of a field, if that's helpful.

Thanks in advance!

$\endgroup$
6
  • $\begingroup$ Is $V$ is a topological vector space over $K$? Finite-dimensional? Is the $G$-action continuous? $\endgroup$ Commented Jul 23, 2014 at 19:25
  • $\begingroup$ @Keenan: Yes, yes, and yes. Edited to reflect. $\endgroup$ Commented Jul 23, 2014 at 20:07
  • $\begingroup$ The continuity of the action does not imply that V is finite diml. $\endgroup$ Commented Jul 24, 2014 at 6:22
  • $\begingroup$ @Laurent: thanks, you're right. Somehow after my first edit, the 'irreducible' got lost in the shuffle. $\endgroup$ Commented Jul 24, 2014 at 12:32
  • $\begingroup$ I don't see why irreducible implies finite dimensional either. This is only true for complex irreps of compact groups. $\endgroup$ Commented Jul 24, 2014 at 12:43

1 Answer 1

6
$\begingroup$

The answer is yes. I think it should be an exercise in any book on representation theory. Since $H$ has finite index in $G$, and $V$ is finite dimensional, so is the representation $W=Ind _H^G (V)$ induced to $G$. In characteristic zero, this is equivalent to proving that the Zariski closure of $G$ in $GL(W)$ is reductive.

The connected component ${\mathcal G}^0$ of the Zariski closure ${\mathcal G}$ of $G$ in $GL(W)$ is unchanged if we replace $G$ by any finite index subgroup $K$. We take $K$ to be the intersection $\cap gHg^{-1}$ as $g$ varies over $G$; this is a finite intersection since $G/H$ is finite.

Restricted to $K$, the representation is semi-simple since $W$ restricted to $K$ is the span of the restriction to $K$ of the semi-simple $gV$ as $g$ varies over $G/H$. Therefore, the unipotent radical of ${\mathcal G}^0$ acts trivially on $gV$ for every $g$ and hence on $W$. Thus, ${\mathcal G}^0$ has no unipotent radical.

[Edit] I should have said that I am assuming $Char (K)=0$.

$\endgroup$

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.