6
$\begingroup$

Let $G$ be a connected, reductive group over a number field $k$. Let $v$ be a place of $k$.

If $v$ is finite, an admissible representation of $G(k_v)$ is defined to be an abstract representation of $G(k_v)$ which is smooth (the stabilizer of any vector is open in $G(k_v)$) and such that the $K$-fixed vectors of any open compact subgroup $K$ of $G(k_v)$ is finite dimensional.

If $v$ is infinite (real or complex), then $G(k_v)$ has the structure of a real Lie group. An admissible representation of $G(k_v)$ is defined here (https://en.wikipedia.org/wiki/Admissible_representation). It is define to be a representation $\pi: G(k_v) \rightarrow \textrm{GL}(\mathscr H)$ on a Hilbert space $\mathscr H$, together with a maximal compact subgroup $K$ of $G(k_v)$, with the following properties:

1 . $G(k_v) \times \mathscr H \rightarrow \mathscr H$ is continuous.

2 . The restriction of $\pi$ to $K$ is unitary.

3 . Each irreducible unitary representation of $K$ occurs with finite multiplicity.

I have a few questions about this:

1 . When $v$ is infinite, does the definition depend on the choice of maximal compact subgroup $K$ of $G(k_v)$? I don't know if they are all conjugate as in the case where $G$ is semisimple.

2 . When $v$ is infinite, does the notion of admissible representation require a Hilbert space? Is there such a thing as, say, an admissible Banach representation? (dropping the unitary assumptions)

3 . What is the definition of an admissible representation of $G(\mathbb{A}_k)$, where $\mathbb{A}_k$ is the ring of adeles of $k$?

Thank you.

$\endgroup$
3
  • 3
    $\begingroup$ If nothing else, the Wiki entry on this is incorrect: the whole repn space need not be a Hilbert space, and quite often it is convenient that it merely be Frechet, e.g., non-unitary principal series repns. $\endgroup$ Commented May 11, 2017 at 16:53
  • $\begingroup$ Do you know a reference which defines these things properly? $\endgroup$
    – D_S
    Commented May 11, 2017 at 18:38
  • 3
    $\begingroup$ AMS Proc Symp 33 (a.k.a. "Corvallis conference"), Bump's book on automorphic forms, or Goldfeld-Hundley's book, or Bernstein-Gelbart "Langlands program" book, or almost any other source (including on-line). It surprises me a bit that Wiki erred on this. "Admissible" basically requires that the multiplicities of $K$-types are finite. This can make sense for quite general topological vector spaces (e.g., quasi-complete, locally convex), and at least Frechet or LF-spaces arise in practice. $\endgroup$ Commented May 11, 2017 at 18:47

0

You must log in to answer this question.