Exponentially small splitting of separatrices beyond Melnikov analysis: Rigorous results

Abstract

We study the problem of exponentially small splitting of separatrices of one degree of freedom classical Hamiltonian systems with a non-autonomous perturbation fast and periodic in time. We provide a result valid for general systems which are algebraic or trigonometric polynomials in the state variables. It consists on obtaining a rigorous proof of the asymptotic formula for the measure of the splitting. We obtain that the splitting has the asymptotic behavior K ε β e −a/ε , identifying the constants K , β, a in terms of the system features. We consider several cases. In some cases, assuming the perturbation is small enough, the values of K , β coincide with the classical Melnikov approach. We identify the limit size of the perturbation for which this theory holds true. However for the limit cases, which appear naturally both in averaging and bifurcation theories, we encounter that, generically, K and β are not well predicted by Melnikov theory.

A General Mechanism of Diffusion in Hamiltonian Systems: Qualitative Results

“…p exp. q" r // for some p; q; r 2 Q, as in [2]. Nevertheless, to obtain the behavior of the error function g.…”

Transversality of homoclinic orbits to hyperbolic equilibria in a Hamiltonian system, via the Hamilton–Jacobi equation

“…In a neighborhood of γ, the form p = ∇S u,s (q) for the invariant manifolds W u,s is valid as far as the solution of the Riccati equation is bounded, since this allows us to use q 2 as the second parameter of the invariant manifolds. Otherwise, if the condition η(t) = 0 is not satisfied in the whole orbit γ, or the solution of the Riccati equation is not bounded, other parameters should be used (see [BFGS11] for the case of a 1 d.o.f with a periodic perturbation). We recall that, if the configuration manifold Q is compact and the potential V (ξ) has a unique nondegenerate maximum, and γ is a homoclinic orbit asymptotic to this maximum, then the condition η(t) = 0 is always satisfied (see the last paragraph of Section 1.2).…”

Critical velocity in kink-defect interaction models: Rigorous results

“…In this paper we prove that it is indeed a first order of ∆E. Note that this is not always the case: in general problems presenting exponentially small phenomena, often the Melnikov integral is not the dominant part of the total loss of energy over a separatrix of a Hamiltonian system (see [2]).…”

“…The model. The sine-Gordon equation is a nonlinear hyperbolic partial differential equation given by (1) ∂ 2 t u − ∂ 2 x u + sin(u) = 0, which presents a family of kinks u k (x, t) given by (2) u k (x, t) = 4 arctan exp…”