[Math] Legendrian Tubular Neighborhood Theorem

contact-geometrydg.differential-geometrylegendrian-submanifoldssg.symplectic-geometry

Given a Lagrangian submanifold $L\subset(M,\omega)$ of a symplectic manifold, we have Alan Weinstein's celebrated Lagrangian tubular neighborhood theorem. I now look for the analog on Legendrian submanifolds $K\subset(Y,\lambda)$ of contact manifolds. As the Lagrangian neighborhood construction essentially relies on the Moser method, I don't think it'd be too hard to build a Legendrian neighborhood by using the analogous Gray stability.

However, I am specifically questioning whether I can get a Legendrian neighborhood theorem directly from the Lagrangian neighborhood theorem: By passing to the symplectization $(\mathbb{R}\times Y,d(e^t\lambda)$), a Legendrian submanifold $K$ becomes a Lagrangian submanifold $\mathbb{R}\times K$. I would love to project some Lagrangian tubular neighborhood down into $Y$ to get a desired neighborhood of $K$, but can this actually be done?

(I spoke with Alan, and he said this might be achieved somehow by viewing all of our constructions equivariantly using the translation $\mathbb{R}$-action on our bundle $\mathbb{R}\times Y\to Y$.)

Best Answer

Basically, Weinstein's theorem says that you can embed $T^*L$ into $M$ like that: $$ T^*L\cong NL\cong \mathcal{T}_L\subseteq M, $$ where $NL$ is the normal bundle and $\mathcal{T}_L$ is a tubular neighborhood, in such a way that the canonical symplectic form on $T^*L$ is the pull-back of $\omega$. So, the "contact counterpart" of above chain of identifications should read $$ J^1K\cong NK\cong \mathcal{T}_K\subseteq Y $$ where now $J^1K$ is the first-order jet bundle of (smooth) functions on $K$. Notice that all manifolds appearing above are $(2n+1)$-dimensional, and all bundles are over $K$ with $(n+1)$-dimensional fibers. I'm sure this observation can be found in the literature about jet spaces and/or contact manifolds.

My guess is that the above embedding, at least locally, can be always found. Concerning the real question, i.e., whether the canonical contact form on $J^1K$ is the pull-back of $\lambda$, I'm not able to answer, but I feel it is not true!

Related Question