The dynamics of magnetic flows for energies above Mañé's critical value (Q1424044)

Consider a Riemannian manifold \(M\), endowed with a magnetic field, which is supposed to have an exact pullback \(d\theta\) on \(\widetilde M\), the universal cover of \(M\). The 1-form \(\theta\) has to be bounded in the following: The associated Lagrangian is \(L(x,v):={1\over 2}| v|^2_x- \theta_x(v)\). Denote by \(c(L)\) its so-called Mañé's critical value, infimum of \(k\)'s such that closed loops in \(\widetilde M\) have positive \((L+k)\)-action. The authors show first that minimal magnetic geodesics (that are Euler-Lagrange trajectories relative to \(L\)) of energy \(k> c(L)\) are Riemannian \((A,0)\)-quasi-geodesics, with \(\lim_{k\to\infty}\, A=1\), thereby improving analoguous results by Grogret, and Boyland-Golé. Then, in the negatively curved case, they get a control on the distance between a quasi-geodesic and its Morse geodesic, which implies in turn the convergence of magnetic geodesics to geodesics, as the magnetic potential \(\theta\) goes to \(0\). Finally, the authors establish (without curvature assumption) that the (magnetic) Mañé's metric associated with the \((L+k)\)-action (for \(k> c(L)\)) differs from the Riemannian metric \(d\) by \(O(d^2)\), so that Mañé's length equals \(d\) (through Mañé's metric generally does not).