In the paper "Symplectic manifolds and their Lagrangian submanifolds", Weinstein showed that locally all the Lagrangian foliations are symplectomorhic to the fiber foliation of a cotangent bundle.

I have a tiny problem in one step in his argument. Let $N$ be a smooth manifold and $U$ a small neighborhood of the zero section of $T^*N$ with the canonical symplectic structure on $T^*N$. Let $F$ be a Lagrangian foliation on $U$ such that it is transversal to $N$ at each point of $N$. Weinstein asserts that there exist a diffeomorphism $h$ from a neighborhood of $N$ to some other neighborhood of $N$ such that $h$ is the identity map on $N$ and $h$ maps leaves of $F$ onto cotangent fibers. I am wondering how this $h$ could be constructed. Probably it has something to do with the Whitney extension theorem. Maybe this is something really standard, I would like to know the details anyway.

Thank you for your attention.

  • 1
    $\begingroup$ See 31.20 of mat.univie.ac.at/~michor/dgbook.pdf $\endgroup$ Oct 20, 2014 at 5:17
  • $\begingroup$ @PeterMichor: Thanks for the reference! I agree that we can choose the symplectomorphism carefully such that the given Lagrangian foliation agrees with the cotangent fiber foliation on the zero section. However it seems that what Weinstein says is that this agreement holds even for points near zero section. I am not sure how this can be achieved. $\endgroup$
    – Piojo
    Oct 21, 2014 at 14:23

1 Answer 1


From the assumption that the intersections are transverse, together with the fact that the foliation is smooth, it is possible to find a smooth map $\phi \colon T^*N \to TU$ which maps each fibre $T^*_pN$ by a linear isomorphism onto the plane inside $T_pU$ that is tangent to the foliation.

Fix a Riemannian metric $g$ on $U$. In a neighbourhood of the zero-section of $T^*N$ we can compose $\phi$ with the exponential map defined on each leaf that is induced by the Riemannian metric given by the restriction of $g$ to the leaf. This is the required diffeomorphism.

The crucial thing is of course to check that the above constructions indeed are smooth. To see this, it is advisable to use smooth local coordinates $(x_1,...,x_{2n})$ in which the foliation is of the form $\{x_1=...=x_n=0\}$ (such coordinates exist by the definition of a smooth foliation).


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.