On the length and area spectrum of analytic convex domains

TL;DR

This paper establishes exponentially small bounds for differences in periodic actions of analytic convex domains, with applications to billiard trajectories, advancing classical smooth case results.

Contribution

It introduces a new exponential upper bound for action differences near resonant curves in analytic twist maps, with applications to billiard problems.

Findings

Exponential bounds for action differences near resonant RICs.

Lengths and areas of periodic billiard trajectories are exponentially close.

Improves classical results for smooth convex billiard domains.

Abstract

Area-preserving twist maps have at least two different $(p,q)$-periodic orbits and every $(p,q)$-periodic orbit has its $(p,q)$-periodic action for suitable couples $(p,q)$. We establish an exponentially small upper bound for the differences of $(p,q)$-periodic actions when the map is analytic on a $(m,n)$-resonant rotational invariant curve (resonant RIC) and $p/q$ is "sufficiently close" to $m/n$. The exponent in this upper bound is closely related to the analyticity strip width of a suitable angular variable. The result is obtained in two steps. First, we prove a Neishtadt-like theorem, in which the $n$-th power of the twist map is written as an integrable twist map plus an exponentially small remainder on the distance to the RIC. Second, we apply the MacKay-Meiss-Percival action principle. We apply our exponentially small upper bound to several billiard problems. The resonant RIC is a boundary of the phase space in almost all of them. For instance, we show that the lengths (respectively, areas) of all the $(1,q)$-periodic billiard (respectively, dual billiard) trajectories inside (respectively, outside) analytic strictly convex domains are exponentially close in the period $q$. This improves some classical results of Marvizi, Melrose, Colin de Verdi\`ere, Tabachnikov, and others about the smooth case.