MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

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

I'm looking for references of the decomposition of $L^2(\Gamma\backslash G)$, where $G$ is a connected Lie group, and $\Gamma\subset G$ a discrete lattice; for simplicity one may assume that $G$ is the real point associated to a linear algebraic group defined over $\mathbb{Q}$, without characters defined over $\mathbb{Q}$, and $\Gamma$ is an arithmetic/congruence lattice in $G$. Write $\Omega=\Gamma\backslash G$. Then $\Omega$ has a canonical probability measure induced by the left Haar measure of $G$, and the right translation gives a unitary representation on $L^2(\Omega)$. When I consider the decomposition of $L^2(\Omega)$ into irreducible unitary representaions, I heard about the notion of continuous spectra and discrete spectra, but why are they called spectra and where may I find explicit descriptions for the classical groups?

Also what if one considers the double quotient $M=\Gamma\backslash G/K$, where $K$ is a maximal compact subgroup of $G$? Is it also described via representations of $G$? I don't see an explicit action of $G$ on it. Also if I conjugate $K$ to a second maximal compact subgroup, is there any invariant description of the decomposition of $L^2(M)$, namely independent of the choice of $K$? I heard about the notion of Shimura varieties, so maybe here I should restrict to the case where $M$ is a locally symmetric hermitian manifold.

Thanks a lot!

share|cite|improve this question
The questions here cover a huge amount of territory, so it's important to make clear what your own background in the literature is: what have you already read? – Jim Humphreys Nov 7 '10 at 22:23
I've started with two books of Iwaniec on modular forms, mainly the chapters on spectral decomposition. I haven't plunged into the volumes of Helgason yet, and I don't know if I should find detailed presentations of the locally symmetric manifolds there. The second example I'm considering is a non-reductive group: $\mathbb{R}^2\rtimes GL_2(\mathbb{R})$. This group can produce univrsal elliptic curves, also called elliptic surfaces. I want to now check the spectral decomposition for the universal elliptic curve. I'm quite lost for literatures in this topic, or maybe I'd try some Jacobi forms. – turtle Nov 8 '10 at 14:23

The discrete spectrum is a bit hard to come by, this is one of the central problems in the theory of automorphic forms. The continuous spectrum is given by Eisenstein-series, as is due to Langlands. A good survey on Langlands work is in Arthur's article in the Corvallis proceedings, Symosia in Pure Maths No 33. As for the K-invariants, well they are just that, K-invariant vectors in the representation. The group doesn't act on this space, but a certain subalgebra of the universal envolope of the Lie-algebra does, providing a host of interesting differential operators, see Helgason's books about this.

share|cite|improve this answer

The following works in the generality of a locally compact unimodular group $G$ and $\Gamma$ being a closed unimodular subgroup and $K$ being compact.

$C_c^\infty(G)$ acts on $L^2(\Gamma \backslash G)$ by convolution operators: $$ \phi \in C_c^\infty(G): f \in L^2 \longmapsto \left( x \mapsto \int\limits_{G} \phi(g) f(xg) d g\right).$$ Call this algebra representation $\pi$.

Restrict this algebra representation to $C_c^\infty(K\backslash G/K)$ and call it $\sigma$. It acts non-trivial only on the subspace $L^2(\Gamma \backslash G /K)$ by the Schur orthogonality relations.

Let $\hat{G}$ denote the unitary dual of $G$. Note that both algebra representations have a direct integral decomposition into irreducible representations $$ \pi = \int\limits_{\hat{G}} \tau \; d \mu_\pi(\tau), \qquad \pi = \int\limits_{\hat{G}} \tau \; d \mu_\sigma(\tau).$$

The support of $\mu_\sigma$ is then contained in the set of the spherical representations of $G$, i.e., those with a $K$-invariant vector. On its support, $\mu_\sigma$ coincides with $\mu_\pi$, whose support may be larger.

share|cite|improve this answer

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.