Displacement energy of unit disk cotangent bundles (Q2454421)

In symplectic geometry, the displacement energy of a compact subset is by definition the minimal Hofer norm needed to displace this set from itself. More precisely, the Hofer norm of a time-dependent Hamiltonian \(H_t\) is the quantity \[ \|H_t\|=\int_0^1(\max H_t-\min H_t)\,dt. \] The displacement energy of a compact set \(K\) is then \[ e(K)=\inf\{\|H_t\|\,|\,\phi_H^1(K)\cap K=\emptyset\}, \] where \(\phi_H^1\) is the time-one map of the Hamiltonian isotopy generated by \(H_t\). The displacement energy of a general subset is then defined by taking the supremum of the displacement energy of all its compact subsets. The displacement energy is known to be a symplectic capacity, i.e., a symplectic invariant that measures the size of subsets in symplectic manifolds. The paper under review studies the displacement energy of unit cotangent bundles. It is now a classical fact that the zero section of the cotangent bundle of a closed manifold is non displaceable. Therefore, unit cotangent bundles of closed manifolds have infinite displacement energy. The main result of the paper is that for non-closed Riemannian manifolds, the unit-cotangent bundle has finite displacement energy which is bounded above by the product of a constant (that only depends on the dimension) with the inner radius of the Riemannian manifold. For example, the inner radius of an open subset of the Euclidean space \(\mathbb R^n\) is the radius of the largest ball it contains; for a compact Riemannian manifold with boundary, it is the largest distance of a point to the boundary. The author gives many consequences of this result mainly related to periodic billiard trajectories and closed geodesic. The proof of the main result consists in introducing a new invariant of Riemannian manifolds called ``width'' which is, in the case of a compact Riemannian manifold, the maximal oscillation of a function whose norm of the differential is everywhere larger than one. An elementary Lemma shows that the width is larger than the displacement energy of the unit cotangent bundle. This reduces the main theorem of the paper to a purely Riemannian lemma: The width is bounded above by the product of a dimensional constant and the inner radius. The proof of the latter is fairly technical and based on the existence of certain good triangulations.