1. Let $T$ be a linear algebraic torus over $\mathbb C$ and $X$ be a smooth quasi-projective symplectic $T$-variety. Also, assume that the action of $T $ is free and $X/T$ exists as a smooth variety. Is this action Hamiltonian?

  2. If the above action is Hamiltonian let us denote the algebraic moment map $X\rightarrow t^{\vee}$ by $\mu$. Does there exist a one-to-one correspondence, via $\mu$, between the symplectic leaves of $X/T$ and the coadjoint orbits of $T$ (which are just points of $t^{\vee}$ in this case)?


This is an affirmative answer to $1$ (assuming by Hamiltonian you mean in the sense of real symplectic geometry).

Yes. Due to Sumihiro's theorem (+ standard facts about Hamiltonians). In fact the assumptions smooth quotient and symplectic are not necessary, just smooth and quasi-projective is enough. Throughtout when I mean symplectic I mean in the real sense.

Sumihiro's Theorem ("Equivariant completion", J. Math. Kyoto Univ., 14: 1–28) implies that: For any normal, quasi projective variety with an algebraic $T$-action there exists $N>0$ and an $T$-equivariant embedding $$\phi : X \rightarrow \mathbb{CP}^N,$$ for some algebraic $T$-action on $\mathbb{CP}^N$.

Next, it is known that every algebraic torus action on projective space is linear, so in a particular homogeneous coordinate system of $\mathbb{CP}^N$ the action is given by $$z.[z_{0}: \ldots : z_{n}] = [z^{a_{0}}z_0 : \ldots : z^{a_{n}}z_n] , $$ for some integers $a_{0}, \ldots a_n$.

Then, this action is Hamiltonian (with respect to the Fubini-Study Kahler form on $\mathbb{CP}^n$) with Hamiltonian $$ H([z_{0}: \ldots : z_{n}] ) = \frac{\sum a_i |z_i|^2}{\sum|z_i|^2}.$$

Finally, there is a general fact about Hamiltonian torus actions, the proof is linear algebra.

Fact: Let $(M,\omega)$ be a symplectic manifold with a Hamiltonian $T$-action and moment map $F$. Then if $N \subset M$ is an invariant, symplectic submanifold then the restricted action on $N$ is Hamiltonian with moment map $F|_{N}$.

So applying this result to the symplectic submanifold $\phi(X)$ to obtain that the action is Hamiltonian. I could probably find references for these statements if you'd like.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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