Let $\Pi$ be a Poisson structure on a manifold $M$. Then we can define a differential $d$ on the complex $\Lambda^{\bullet}M$ $$ C^{\infty}M \to TM \to...\Lambda^kTM \to... $$ in the following way: $$ d = [\Pi,\cdot], $$ where $[\cdot,\cdot]$ is the Schouten-Nijenhuis bracket. The cohomology of this complex is called Poisson cohomology.

I have several question about this construction:

  1. Suppose $M=\mathbb R^{2d+s}$ with a Poisson structure $\sum\limits_{i=1}^{d}\frac{\partial}{\partial q_i}\wedge \frac{\partial}{\partial p_i}$. Is it true that Poisson cohomology of $M$ are zero?

  2. Is this construction functorial? I mean, let $f\colon M\to N$ be a poisson map, then do we have a map of complexes $f_*\colon \Lambda^{\bullet}M \to \Lambda^{\bullet}N$?

  3. Let $F\colon M\times I \to N$ be a homotopy such that for each $t$ $F_t$ is a Poisson map. Is it true that $F_0$ and $F_1$ induce the same map from Poisson cohomology of $M$ to Poisson cohomology of $N$?


First of all, one can think of Poisson cohomology (or, rather, Poisson cochains) as functions on the $QP$-manifold $T^*[1]M$ with the differential as you describe. In particular, it has a natural odd Poisson ($\mathbb{P}_2$) structure given by the Schouten bracket.

  1. You can split $T^*[1](\mathbb{R}^{2d+s}) = T^*[1](\mathbb{R}^{2d})\times T^*[1](\mathbb{R}^s)$ and so the same decomposition applies to Poisson cohomology. Since $\mathbb{R}^{2d}$ is symplectic, its Poisson cohomology coincides with the de Rham cohomology which is quasi-isomorphic to $\mathbb{R}$ in degree 0. Therefore, Poisson cohomology of $\mathbb{R}^{2d+s}$ coincides with the Poisson cohomology of $\mathbb{R}^s$ with the zero Poisson structure, i.e. functions on $T^*[1](\mathbb{R}^s)$.

  2. Given a map of manifolds $M\rightarrow N$, there is no natural map on their cotangent bundles. Instead, there is a Lagrangian correspondence $$T^* M \leftarrow T^*N\times_N M \rightarrow T^* N.$$ The same construction works with a shift and so instead of a map one has a dg bimodule between Poisson cohomologies of $M$ and $N$. The underlying graded bimodule is $C^\infty(T^*[1] N\times_N M)$. The differential is constructed as follows. Recall that for any coisotropic submanifold $C\hookrightarrow X$ the graded commutative algebra $\bigwedge^\bullet N_{C/X}$ carries a natural differential (a version of the differential you wrote). The map $M\rightarrow N$ is Poisson iff its graph $M\rightarrow M \times N$ is coisotropic and taking the exterior algebra on the normal bundle of this coisotropic you recover $C^\infty(T^*[1] N\times_N M)$ and that's the required differential.

  3. Given such a homotopy, you get a homotopy of (dg) bimodules from point 2, which seems to be the best you can hope for.

| cite | improve this answer | |

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.