Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

If $\lambda,\mu$ are two dominant weights for a Lie group $G$, then there is a canonical (up to scale, perhaps) surjection $V_\lambda \otimes V_\mu \to V_{\lambda+\mu}$ of finite-dimensional representations, which I have occasionally heard called the "Cartan projection".

To each irreducible (or standard) Harish-Chandra module for $({\mathfrak g},K)$, one can associate a $K$-orbit on $G/B$. For finite-dimensional representations, this orbit is the open $K$-orbit.

I want to know what analogues exist of the Cartan projection out of $V\otimes W$ if $V,W$ are two Harish-Chandra modules with the same associated $K$-orbit (other than the open orbit case above).

The answer may be something like "every H-C quotient of $V\otimes W$ has the wrong Gel$'$fand-Kirillov dimension for it to again have that associated $K$-orbit," in which case I'd appreciate references that make that most clear. (I will be sad, but not overly surprised, if that is the case.)

EDIT: For consistency of notation, let's take $G,K$ to be the complexifications of $G_0,K_0$, where $K_0$ is a maximal compact in $G_0$.

share|improve this question
I'm not sure I understand, so can we go through an example? What should be the answer for $G=SL_2(\mathbb R)$? Here, the tensor product of two holomorphic discrete series, $D_k$ and $D_l$, is the direct sum of all holomorphic discrete series with weight at least $k+l$, that is: $D_k\otimes D_l\simeq \bigoplus_{i=0}^\infty D_{k+l+2i}$. (On the other hand, the tensor product of unitary principal series is a direct integral of principal series plus holomorphic discrete series.) –  B R Oct 24 '11 at 16:13
(Note added, to make $SL_2({\mathbb R})$ to be $G_0$ not $G$.) Here $K=SO(2)$, and the $K$-orbit is the north pole. Then I would want the map $D_k \otimes D_l \to D_{k+l}$. –  Allen Knutson Oct 24 '11 at 16:29
...which is to say, yes! this is exactly the sort of thing I want. –  Allen Knutson Oct 24 '11 at 16:30
Does your "direct integral" comment say that I can't get principal $\otimes$ principal onto principal? –  Allen Knutson Oct 24 '11 at 16:47
Well, I wish I knew more, then :) The $SL_2(\mathbb R)$ case is in Repka's early work (the papers with "Tensor products" in the title). Section 7 of Kobayashi's paper (ms.u-tokyo.ac.jp/~toshi/texpdf/jfa98-full.pdf) may point you to some references (and Theorem 7.4 is useful). And Section 4 of Weissman's paper (people.ucsc.edu/~weissman/MultMod.pdf) may do the same (in fact, Prop 4.2 may be exactly what you want). –  B R Oct 24 '11 at 17:02
show 2 more comments

1 Answer

If $G$ is simply connected semisimple and admits holomorphic discrete series, then the tensor product of the holomorphic discrete series of two weights decomposes as a direct sum of other holomorphic discrete series, each appearing with finitely mulitplicity. (In short, holomorphic discrete series comport themselves in general just as in the case of $SL_2(\mathbb R)$.)

I forget the reference now; sorry! [Added: Actually, section 7 of the Kobyashi paper linked to by BR above gives the result, I think.]

I think that for other Harish-Chandra modules, tensor products have infinite multiplicities (or in more analytic terms, involve direct integrals rather than direct sums). Even the generic discrete series for $Sp_4(\mathbb R)$ should demonstrate this behaviour, if I remember correctly. (And tensoring a holomorphic discrete series by an anti-holomorphic discrete series would also give infinite multiplicities.)

share|improve this answer
add comment

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.