# Contact structure on a circle bundle over a symplectic manifold.

Consider a Symplectic manifold D (with $H^1(D)=0$) with symplectic form $w$. Let V be the total space of a circle bundle over D with non-trivial Euler class $e\in H^2(D)$. You may think of V as the set of unit vectors in a complex line L bundle over D with chern class e. Then we can construct a symplectic form ,denote it by same symbol $w$, on total space of L whose restriction to D is the original w. The question is whether V has a contact structure with a contact form $\alpha$ such that $d\alpha=w\mid_V$.

Looking the the Gysinn sequence : $0\rightarrow H^0(D) \rightarrow H^2(D) \rightarrow H^2(V) \rightarrow 0$ it seems that the answer is:

Yes iff $w$ is multiple of e.

I Just wanted to make sure that my conclusion is correct!

• How do you construct the symplectic form on $L$? – Tim Perutz Apr 30 '10 at 1:36
• Lets fix a hermitian metric on L and let r be the distance function. for $\pi:L\rightarrow D$ consider the two form : $\tilde{w} = π^*w+ r \pi^*e + dr ∧ \beta$ Where $\beta$ is the connection one-form on L\{zerosection} satisfying $\beta{\partial{\theta}}=1$. $\tilde{w}$ extends across D and its restriction to D is $w$. See page 15 of annals paper:" The symplectic sum formula for Gromov-Witten invariants" – Mohammad Farajzadeh-Tehrani Apr 30 '10 at 5:41
• This, presumably, is symplectic near the zero-section only, in general? Anyway, I concur that if $e$ isn't a multiple of $[w]$, this form is not exact on circle-bundles in $L$, and that conversely, a multiple of a connection form with curvature $const. w$ is contact. – Tim Perutz Apr 30 '10 at 12:06
• My question is local, so I am interested just in a small neighborhood of D in L. – Mohammad Farajzadeh-Tehrani Apr 30 '10 at 19:19

Let's $w$ be the symplectic structure on D and $\tilde{w}$ be the one on total space of line bundle . Then we have Gysin exact sequence above which gives $H^2(V)=H^2(D)/ \left\langle e\right\rangle$ and so for $\tilde{w}$ to be of the form $[d\alpha]= \in H^2(V)$ it has to be a multiple of e, when restricted to D.