Processing math: 100%

2013-09-04

Marco Mazzucchelli

Isometry invariant geodesics

(a question from Grove?)

Key idea: Marco is interested in geometrically distinct geodesic arcs. The key tool is to consider the local homology of a critical point/set. The idea is to look at the local homology of iterates.

(For instance, to find infinitely many geodesics from p to q, the variational problem doesn't have a \mathbb{Z}/m symmetry anymore, so cannot use the symmetry anymore)

Suppose I is an isometry. Want the existence of \tau so I( \gamma(t) ) = \gamma( t + \tau).

If I is the identity, then just looking for closed.

Question: when do we have infinitely many invariant geodesics?

Thm (Grove): If there exists a non-closed I invariant geodesic then the closure of \gamma(\mathbb R) contains uncountably infinitely many I-invariant geodesics.

It therefore remains to study periodic I invariant geodesics.

Variational setting:
\Lambda( M, I ) = \{ \gamma \colon \mathbb R \to M \, | \, I ( \gamma(t) ) = \gamma (t+1) \quad \forall t \in \mathbb R \}

Has \mathbb R action by time shift.

Energy E( \gamma ) = \int_0^1 | \dot \gamma |^2 dt

The critical points of E are I-invariant geodesics. The critical orbit of \gamma is \mathbb R if not closed, or S^1 if closed.

Warning: every such \gamma comes with infinitely many critical points of E.

\tau \in \mathbb R. \gamma^\tau(t) = \gamma( \tau t).
For each \gamma that is a periodic critical point of E of period p \ge 1, also have \gamma^{mp + 1} for any m \in \mathbb{N}.

Study Morse-theoretic properties of this set \{ \gamma^{mp+1} \}.

Grove-Tanaka: show that m \mapsto nul( \gamma^{mp+1}) is bounded,
m \mapsto index( \gamma^{mp+1} ) groes linearly and finally m \mapsto rank L_*( \gamma^{mp+1}) is bounded.

Theorem: Assume \pi_1(M) =0 and I is homotopic to the identity and H^*(M, \mathbb Q) has more than 1 generator.

Then, there exist infinitely many I-invariant geodesics.

(RMK: I homotopic to the identity allows us to identify the homology of \Lambda(M, I) with the loop space homology.)

(When I is the identity, this recovers Gromoll-Meyer theorem.)

Theorem (Mazzucchelli):
Assume:

  •  M = M_1 \times M_2 (as manifolds... do not require a product metric).
  •  rank H_1(M, \mathbb Z) > 0 and dim M_2 \ge 2.
  •  I is homotopic to the identity.

Then, there exist infinitely many I-invariant geodesics.


Theorem (M):

in a general setting: if exist \gamma closed I invariant and d \ge 2 such that L_*( \gamma^{mp+1}) is non-trivial for infinitely many m then there are infinitely many I invariant geodesics.


Marco remarks that the ideas are closely related to Umberto's discussion of symplectically degenerate maxima.



No comments:

Post a Comment