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 trying to get a better understanding of multiplicities in geometric quantization, and so I've been concentrating on a specific simple case: let $\mathcal{O}\subset\mathfrak{g}^*$ be an integral coadjoint orbit of a compact semisimple Lie group $G$ with KKS symplectic form $\omega$, defined by $\omega_\mu(-\mathrm{ad}_\xi^*\mu,-\mathrm{ad}_\zeta^\*\mu)=\mu([\xi,\zeta])$, and compatible line bundle-connection pair $(L,\nabla)$. Let $(L^\*,\nabla^*)$ be the dual bundle, with base $(\mathcal{O}^-,-\omega)$. Put a totally complex polarization $F$ on $\mathcal{O}$, with corresponding complex structure $\mathcal{J}_F$, and the complex conjugate polarization $F^\*$ on $\mathcal{O}^-$, with corresponding complex structure $\mathcal{J} _{F^*} = -\mathcal{J}_F$. The momentum map corresponding to the diagonal coadjoint $G$-action on $\mathcal{O}\times\mathcal{O}^-$ is $\mathrm{J}(\mu,\nu)=\mu-\nu$. The covariantly constant sections of $L$ form an irrep, and those of $L^\*$ the dual irrep, so the $G$-invariant sections of $L\otimes L^*$ should then be one-dimensional, and I'm trying to verify this explicitly. Following some ideas in Guillemin and Sternberg's 1982 paper "Geometric quantization and multiplicities of group representations", I using the complex structure to to extend the $G$-action on $\mathcal{O}\times\mathcal{O}^-$ to a $G^\mathbb{C}$-action, by defining the infinitesimal generator corresponding to $i\xi\in i\mathfrak{g}$ to be $(i\xi)_{\mathcal{O}\times\mathcal{O^-}}=\mathcal{J}\xi\_{\mathcal{O}\times\mathcal{O^-}}$, (where $\mathcal{J}$ is the combined complex structure) and exponentiating.

I want to show that the so-defined $G^\mathbb{C}$-action is transitive on the whole of $\mathcal{O}\times\mathcal{O}^-$, but I'm having trouble doing so (and maybe it's not?). The problem is that the action isn't free everywhere, and I don't really understand how it spans the tangent space at a general point $(\mu,\nu)$. I think it would help if I understood what $\ker T_{(\mu,\nu)}\mathrm{J}$ was everywhere. For example, if I can show that $\ker T_{(\mu,\nu)}\mathrm{J}\subset \mathfrak{g}\cdot(\mu,\nu)$, then since $\left(\mathfrak{g}\cdot(\mu,\nu)\right)^\omega = \ker T_{(\mu,\nu)}\mathrm{J}$, $\;\mathfrak{g}\cdot(\mu,\nu)$ will be coisotropic, and so $(i\mathfrak{g})\cdot(\mu,\nu)$ will be transversal to it (by general properties of complex structures). Then its just comes down to counting dimensions. But maybe this is the wrong track.

Any help appreciated. Thanks.

share|cite|improve this question
The $G^{\mathbb C}$ action isn't transitive, it just has a dense orbit, which I think is enough for your argument. You can identify each coadjoint orbit with $G^{\mathbb C}/B$, and then the orbits of $G^{\mathbb C}$ on that are $B\backslash G^{\mathbb C}/B \cong W$. Finitely many orbits, hence one of them is open dense. One $T$-fixed point in this open dense orbit is $(1, w_0)$, where you should be able calculate. (Here $w_0$ is the long element in $W$.) – Allen Knutson Sep 21 '12 at 12:20
Thanks a lot for the clarification. I guess I was pretty naive in my thinking on this, but your comment makes it clear what I need to read next. – user26687 Sep 22 '12 at 6:51
Also, anyone with sufficiently high reputation, feel free to close this question. – user26687 Sep 22 '12 at 6:53

Your Answer


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

Browse other questions tagged or ask your own question.