Instability of high dimensional Hamiltonian Systems: Multiple resonances do not impede diffusion

TL;DR

This paper demonstrates that in high-dimensional Hamiltonian systems, multiple resonances do not prevent the occurrence of diffusion in action variables, even across large gaps among invariant tori, under certain conditions.

Contribution

It introduces a new analysis of multiple resonances in high-dimensional Hamiltonian systems, showing they can be crossed and do not impede diffusion, extending previous results to more complex models.

Findings

Existence of orbits with arbitrary action excursions in large domains

Multiple resonances can be crossed due to their codimension

A simple Melnikov potential condition ensures resonance crossing

Abstract

We consider models given by Hamiltonians of the form $$H(I,\phi,p,q,t;\epsilon) = h(I) + \sum_{j = 1}^n \pm(\frac{1}{2} p_j^2 + V_j(q_j)) + \epsilon Q(I,\phi,p,q,t;\epsilon)$$ where $I,\phi$ are d-dimensional actions and angles, $p,q$ are n-dimensional real conjugated variables, and $t$ is an angle. These are higher dimensional analogues, both in the center and hyperbolic directions, of the models studied in previous papers by the athors. All these models present the large gap problem. We show that, for $\epsilon$ small enough, under regularity and explicit non-degeneracy conditions on the model, there are orbits whose action variables $I$ perform rather arbitrary excursions in a domain of size O(1). This domain includes resonance lines and, hence, large gaps among $d$-dimensional KAM tori. The method of proof follows closely the strategy of [DLS(=Delshams, Llave and Seara) 2006]. The main new phenomenon that appears when the dimension $d$ of the center directions is larger than one, is the existence of multiple resonances. We show that, since these multiple resonances happen in sets of codimension greater than one in the space of actions $I$, they can be contoured. This corresponds to the mechanism called diffusion across resonances in the Physics literature. The present paper, however, differs substantially from [DLS 2006]. On the technical details of the proofs, we have taken advantage of the theory of the scattering map [DLS 2008], not available when the above papers were written. We have analyzed the conditions imposed on the resonances in more detail. More precisely, we have found that there is a simple condition on the Melnikov potential which allows us to conclude that the resonances are crossed. In particular, this condition does not depend on the resonances. So that the results are new even when applied to the models in [DLS 2006].