2
$\begingroup$

Given a real vector bundle, say the tangent bundle of a manifold, some obstructions to this being the underlying real bundle of a complex bundle come from characteristic classes. These can prove the lack of complex structure on the underlying manifold. I was wondering if perhaps there was a similar set of classes for symplectic manifolds?

$\endgroup$
  • $\begingroup$ Given a symplectic manifold, there exists a contractible set of almost complex structures compatible with the symplectic structure. These almost complex structures give a complex vector bundle structure on the tangent bundle. You can define characteristic classes to be the chern classes of this complex vector bundle. $\endgroup$ – Thomas Rot Jun 13 '16 at 8:41
3
$\begingroup$

You have classifying spaces $BSp(n)$ and $BGl(2n)$ and from the inclusion $Sp(n)$ into $Gl(2n)$ you get a map $\iota \colon BSp(n) \to BGl(n)$.

A characteristic class for a real vector bundle is nothing else than a cohomology class of $BGl(n)$ (This follows from naturality). And you can define characteristic classes analogously for symplectic bundles by taking cohomology classes for $BSp(n)$. A symplectic structure of your vector bundle now corresponds to a lift of the classifying map to $BGl(n)$ into $BSp(n)$.

But now if you understand $\iota^*$ you can get that certain characteristic classes can't occure if you have such a lift. For example if you were looking at real bundles and asking for a complex structure you would get that every odd Stiefel-Whitney-Class would have to vanish since the complex grassmanian has no cohomology in odd dimensions.

I'm not aware of how the cohomology ring of $BSp(n)$ looks like, but with that information and a little computation of $\iota^*$ you would get explicit classes, that obstruct such a lift.

$\endgroup$
  • 1
    $\begingroup$ I think $BSP(n)=G_n(\mathbb{H}^\infty)=\mathbb{Z}[a_1,\ldots, a_n]$ with degree $a_i=4$, so all stiefel whitney classes in degree $0,1,2,3 \pmod 4$ have to vanish, and probably $c_n=0$ for $n$ odd as well. $\endgroup$ – Thomas Rot Jun 13 '16 at 8:43

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.