Calculus of variations in mean and convex Lagrangians (Q1109326)

This paper is motivated by the search for periodic solutions to Hamiltonian systems with trajectories (x(t),p(t)) in \({\mathbb{R}}^{2n}\). It begins with a derivation of the Euler-Lagrange equation and Legendre condition for the following variational problem: On the space of almost periodic functions x( \(\cdot )\), minimize the functional \[ \Phi (x):=\lim_{T\to \infty}\int^{T}_{-T}L(x(t),\dot x(t)) dt. \] (Techniques from distribution theory are used heavily.) Under the additional assumption that L is jointly convex, the author shows the following: 1. Every periodic or almost periodic solution x(\(\cdot)\) of the Euler equation is such that the set \(\overline{co}\{x(t): t\in {\mathbb{R}}\}\) contains a constant solution of the Euler equation; and 2. If the Hamiltonian system generated by L has s non-constant periodic solutions (with periods linearly independent over \({\mathbb{Z}})\), then \(s\leq 2n-1\).