Hi, Is it true that all irreducible unitary representations of a residually finite group are finite dimensional?

Actually I suspect that it is not, but cannot find any example.

  • 2
    $\begingroup$ How about the free group with two generators? Every representation of every 2-generated group is a representation of $F_2$. $\endgroup$ – Mark Sapir May 14 '11 at 12:02
  • 3
    $\begingroup$ Mark is right. Here is for example a nice result by Michael Cowling and Tim Steger: Let $G$ be a non-compact simple Lie group, and let $\pi$ be a unitary irreducible representation of $G$; assume that $\pi$ is not in the discrete series of $G$ (you exclude countably many rep's in a continuum). Then the restriction of $\pi$ to any lattice $\Gamma$ in $G$, is an irreducible representation of $\Gamma$. $\endgroup$ – Alain Valette May 14 '11 at 13:24
  • $\begingroup$ To clarify: when you talk about an irreducible representation $\pi:G \to {\rm End}(V)$, do you mean that there is no proper $G$-invariant subspace other than zero, or do you mean that there is no proper, closed $G$-invariant subspace other than zero $\endgroup$ – Yemon Choi May 15 '11 at 3:57
  • $\begingroup$ I believe the standard definition is no closed $G$-invariant subspaces. $\endgroup$ – Benjamin Hayes May 23 '11 at 7:31

As Mark Sapir pointed out, this is of course false. There are tons of infinite dimensional irreducible representations of residually finite groups. In fact, I think that any finitely generated group for which the answer is yes, is virtually abelian.

However, you could also ask whether all irreducible representations of a given residually finite group are weakly contained in finite dimensional representations. This is true for $\mathbb F_2$ but for example not for $SL(3,\mathbb Z)$. For the group $\mathbb F_2 \times \mathbb F_2$, this problem is open and equivalent to Connes Embedding problem.

  • 2
    $\begingroup$ @ Andreas: Correct, the assumption in the OP ("every unitary irrep is finite dimension") implies that the group is type I, so by Thoma's theorem the group is virtually abelian. Your sentence "You could ask...", ending with a (in)famous open question, is a bit tough (:-). $\endgroup$ – Alain Valette May 21 '11 at 17:06
  • 1
    $\begingroup$ Rats! I was going to say that! $\endgroup$ – Simon Thomas May 21 '11 at 17:09

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.