Processing math: 1%

2013-09-05

Pedro Salamão

(joint work with Umberto Hryniewicz and Joan Licata)

Question: determine the topology of a co-oriented contact manifold from the Reeb flow of defining contact form.

Motivating example: characterisation of the tight 3-sphere. If \alpha is dynamically convex & non-degenerate and admits a closed Reeb orbit P so that P is unknotted, has self-linking number -1 and the CZ index is 3. Then M is the tight 3-sphere. (Converse also true.)

(If, P as above, then is binding of an open book decomposition of M with disk-like pages. This theorem is proved by constructing this open book.)

L(p,q) with standard contact structure \xi_{\mathrm{std}}.

Let K be a knot. Let p \ge 1. Then a p-disk for K is a disk whose interior embeds in M \setminus K and u|_{\partial D} is a p-cover of K.

If K admits a p-disk, it is p-unknotted and is an order p rational unknot.

The image of the Hopf fibres project to p-unknots K_0 and K_1 inside L(p,q) with self-linking number -p.

If you look at the p-disk with boundary on K_0, and go around the boundary annulus once, you come back to a sheet that is shifted over by -q. (Call this the monodromy.)

For K_1 the monodromy is q' where qq' = 1 (mod p).

Theorem:  (Hryniewicz, Licata, Salamão)
 (M, \xi) with c_1(\xi) = 0 on \pi_2(M). Let p \in \mathbb N^*

Then, contactomorphic to L(p,q), \xi_{\mathrm{std}} if and only if \xi is the kernel of a non-degen contact form \alpha admitting a prime closed Reeb orbit K such that

  • K is p-unknotted, CZ( K^p ) \ge 3, self-linking K^p = -p and monodromy is -q
  • every contractible Reeb orbit P \subset M \setminus K with CZ index 2 is not contractible in M \setminus K.


Strategy of proof: make p disk for K become a fast pseudoholomorphic plane with \pi du \ne 0. Then, foliate complement with these planes. Technical condition on index 2 orbits gives rational open book and then result.

Great idea: C^0 perturb the disk we start with, to obtain the following picture:

  • characteristic foliation has only one nicely elliptic singular point e and all leaves reach the boundary transversely (recall the boundary is a transverse knot)
  • transverse to R near the boundary (this uses the fact the CZ index is at least 3 for the p-fold cover of the knot)
  • the disk does not contain any periodic orbits of the Reeb flow (i.e. only one is the boundary)
  • Local model near e with nice Darboux coordinates.
Ah, idea now is to consider the Bishop disks near singularity. One boundary component is the constant disk in e. Other one is a building. Chasing through what happens, must break on orbits with index at least 2. This index 2 orbit condition allows us to exclude most of what can happen. From this, we conclude that the other boundary component of the building is the plane we want.

Additional applications:

Geodesic flows and restricted planar 3 body problem: both live on \mathbb RP^3 = L(2,1).

Surface of revolution: double cover of meridians have CZ index 3 or 4. Linking argument means hypotheses satisfied for other orbits.

Finsler metrics with pinched flag curvature (pinching based on reversibility). Paternain says is dynamically convex and Rademacher says there is a simple closed geodesic. Some work says this satisfies hypotheses of the theorem.

planar restricted 3 body problem with circular earth-moon orbit.
Albers, Fish, Frauenfelder, Hofer, Paternain, von Koert: subsets of these in various papers show that low energy gives two components each of which regularizes to a Reeb flow on L(2,1). Obtain disk like surfaces of section for component near the moon (by working on S^3). Also show dynamical convexity but not geometric convexity near the first critical value for rotating Kepler problem. Retrograde orbit has index 3 but the direct orbit has higher index.

Possible application of this rational open book stuff: work with prescribed orbit with high index, or for perturbations that destroy dynamical convexity.

No comments:

Post a Comment