I'm trying to show that the Lie group $G=SO(n+1) \times SO(2)$ acts multiplicity-free on the cotangentbundle $T^* (SO(n+1)/SO(n-1))$.

That means:

1) There exists an $\operatorname{Ad}^*_G$-equivariant momentum map $$\Phi \colon T^* (SO(n+1)/SO(n-1)) \to \mathfrak{g}^*$$ 2) For $\alpha \in \mathfrak{g}^*$ the isotropygroup $G_\alpha=K$ acts transitively on the connected components of $\Phi^{-1}(\alpha)$.

If $\mathfrak{k}$ is the Lie algebra for $K$, $\mathfrak{k}^0$ the annihilator of $\mathfrak{k}$ in $\mathfrak{g}^*$ and $\mathcal{O}$ a $\operatorname{Ad}^*_G$-orbit in $\mathfrak{g}^*$, then this can be rephrased as:
$\mathcal{O} \cap \mathfrak{k}^0$ is a finite union of $\operatorname{Ad}^*_K$-orbits.

Let $SO(n-1)$ be identified as the subgroup of $SO(n+1)$, by the injection $A \mapsto \begin{pmatrix} E_2 & 0 \\ 0 & A \end{pmatrix}$ and $SO(2)$ as a subgroup through $R \mapsto \begin{pmatrix} R & 0 \\ 0 & E_{n-1} \end{pmatrix}$.

If $SO(2)$ induces a left action on $SO(n+1)$ by right-multiplication and $SO(n+1)$ by left-multiplication, then these actions commute and so $G =SO(n+1) \times SO(2)$ acts on $M=SO(n+1)/SO(n-1)$.

Let $q_0 = \begin{pmatrix} E_2 & 0\\ 0 & SO(n-1) \end{pmatrix} \in G$ and denote by $K=G_{q_0}$ the isotropy group for $q_0$. Now I computed the corresponding Lie algebra $$\mathfrak{k} = \left\{ \middle( \begin{pmatrix} X & 0 \\ 0 & Y \end{pmatrix}, \begin{pmatrix} X & 0 \\ 0 & 0 \end{pmatrix}\middle) \middle| X \in so(2), \ Y \in so(n-1) \right\}$$ and the annihilator $\mathfrak{k}^0 \subset \mathfrak{g}^*$ using the identification of $so(n+1)$ with $so(n+1)^*$ by $${so(n+1) \ni Z \mapsto \alpha \in so(n+1)^*}$$ for $\alpha(X) = \operatorname{tr}(ZA), \ \forall A \in so(n+1)$.

So the corresponding set for $\mathfrak{k}^0$ is $$\tilde{\mathfrak{k}} = \left\{\left( \begin{pmatrix} Z_1 & Z_2 \\ -Z_2^T & 0 \end{pmatrix}, -Z_1 \right) \middle| Z_1 \in SO(2), \ Z_2 \in \mathbb{R}^{2 \times (n-1)} \right\} .$$

Since our identification of $so(n+1)$ with $so(n+1)^*$ is $Ad_G$-invariant, we identified the adjoint-action of $G$ on $\mathfrak{g}$ with the coadjoint-action of $G$ on $\mathfrak{g}^*$. With all this in mind we now have to prove that for $\operatorname{Ad}_G$-orbits $\mathcal{O}' \subset \mathfrak{g}$ we have that $\mathcal{O}' \cap \tilde{\mathfrak{k}}$ is the finite union of $K$-orbits.

Now take an element $Z = \left( \begin{pmatrix} Z_1 & Z_2 \\ -Z_2^T & 0 \end{pmatrix}, -Z_1 \right) \in \tilde{\mathfrak{k}}$ and let $\mathcal{O}'$ be the $\operatorname{Ad}_G$ orbit through $Z$.
Since $SO(2)$ is an abelian Lie group, we have that each element $X \in \mathcal{O}' \cap \tilde{\mathfrak{k}}$ is of the form $X = \left( \begin{pmatrix} Z_1 & X_2 \\ -X_2^T & 0 \end{pmatrix} , -Z_1 \right)$.

But from here on, I don't know what to do. One idea was to calculate the Eigenvalues of $\begin{pmatrix} Z_1 & X_2 \\ -X_2^T & 0 \end{pmatrix}$ so that I get conditions on the entries of $X_2$, then using this conditions to show that $K$ acts locally transitive on $\mathcal{O}' \cap \tilde{\mathfrak{k}}$ and finally showing that $\mathcal{O}' \cap \tilde{\mathfrak{k}}$ has only finitely many connected components. Would that be the right way, or is there some easier one?


1 Answer 1


Let $G_\mathbb{C}$ be the complexification of a compact group $G$. Let $H\subseteq G$ be a closed subgroup. Then there is the following criterion: $M=T^*(G/H)$ is multiplicity-free (as a Hamiltinian manifold) if and only if $X=G_\mathbb{C}/H_\mathbb{C}$ is spherical (as a $G_\mathbb{C}$-variety, i.e. a Borel subgroup $B$ has an open dense orbit). This is a consequence of Brion's criterion.

In your case, $G_\mathbb{C}=SO(n+1,\mathbb{C})\times SO(2,\mathbb{C})$ and $H_\mathbb{C}=SO(n-1,\mathbb{C})\times SO(2,\mathbb{C})$ where the second factor is embedded diagonally. Now we take the quotient of $X$ by the central $SO(2,\mathbb{C})$ and obtain a morphism of $X$ to $\overline X:=SO(n+1,\mathbb{C})/H_\mathbb{C}$. So $B$ has a dense orbit on $X$ if and only if it has one on $\overline X$. But $\overline X$ is spherical since it is a symmetric variety.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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