MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $G$ be a real reductive Lie group, $P$ its parabolic subgroup with Levi decomposition $P=MN$, let $\mathfrak{n}$ be the nilpotent Lie algebra of $N$. Suppose given a smooth representation $(\pi,V)$ of $G$ with $V$ some Frechet space. We can form both $V/(\mathfrak{n}V)$ and $V/(N.V),$ where
$$ N.V=\operatorname{span}\{n.v-v|v\in V, n\in N\} $$

(here we take the closure of $\mathfrak{n}V$ and $N.V$ in the two quotients). Now question is there any difference between these two quotients? or equivalently, what's the relation between the closure of $\mathfrak{n}V$ and that of $N.V$? On one direction we have $\mathfrak{n}V$ is contained in $N.V$, so the question amounts to ask whether there is some $V$, so that the latter contains the former properly.

A related phenomenon is that, when considering the minimal parabolic subgroup $B=TU$, if $\chi$ is a generic character of $U$, $\eta$ the derivative of $\chi$, so $\eta$ is a nondegenerate complex linear form on the Lie algebra $\mathfrak{u}$ of $U$. There are two versions of Whittaker functional in literature, defined either in terms of pair $(U,\chi)$ or $\mathfrak{u},\eta$. And I'm wondering if they are equivalent.

Edit (Victor Protsak): definition of N.V has been made explicit.

share|cite|improve this question
By $N.V$, do you mean the space generated by $n\cdot v-v$ (a la the p-adic Jacquet module)? – B R Mar 29 '11 at 17:22
yes, I should be careful. – user1832 Mar 29 '11 at 17:25
If the Jacquet module is finite-dimensional then it follows from the Lie correspondence that the two spaces are the same; this is certainly the case if $V$ is irreducible. Is this sufficient for your purposes or are you interested in a more general situation? – Victor Protsak Mar 29 '11 at 19:41
Thanks Victor. May I know that's the Lie correspondence you mentioned? To my knowledge, the Jacquet module is finite dimensional in case of maximal nilpotent Lie subalgebra (i.e.,\mathfrak{u}), even if $V$ is not irreducible. But does this still hold for general nilpotent $\mathfrak{n}$? – user1832 Mar 29 '11 at 21:29

Maybe I'm misreading the set-up, but the answer to the first question seems to be obviously no. Since the group $N$ contains the identity element, the quotient by the closure of $N.V$ must be zero. On the other hand, when $V$ is finite dimensional and irreducible for $SL_2(\mathbb{R})$, the subspace $\mathfrak{n}.V$ is the sum of all but the lowest weight space and the quotient is nonzero.

ADDED: Assuming your interest is in the real groups and their infinite dimensional representations (with some added structure), as opposed to $p$-adic analogues, there are a lot of papers to consult. Being a non-specialist in all of this, the only source that comes to mind is a long paper by Kostant, On Whittaker vectors and representation theory, lnvent. Math. 48, 101- 184 (1978). That might be relevant to your comparison between the group and Lie algebra situations.

share|cite|improve this answer
Thanks for your post, Jim. It should be $N.V-V$ in stead of $N.V$. When $V$ is a nice smooth representation, the subrepresentation theorem says that the quotient of $V$ by $\mathfrak{n}V$ is never zero. – user1832 Mar 29 '11 at 17:30

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.