Phase space structure of multi-dimensional systems by means of the mean exponential growth factor of nearby orbits (Q1397779)

The ``mean exponential growth factor of nearby orbits'' (MEGNO) was introduced by two of the authors in [Astron. Astrophys. Sup. 147 (2000)], as a new numerical computational tool for phase space analysis of Hamiltonian systems. If \(\gamma\) is an orbit of a Hamiltonian system on a compact energy surface, and \(\delta_\gamma\) is the norm of the vector which satisfies the linear variational equation around \(\gamma\), the MEGNO \(Y(\gamma)\) is defined as the function \(Y(\gamma)(t)= 2t^{-1}\int_0^t t'\dot{\delta}_\gamma(t')/\delta_\gamma(t') dt'\). The present paper is meant to demonstrate the relevance of this concept for getting detailed information about the global behaviour of the system. In particular, it turns out to be efficient for investigating the phase space structure both in regular and stochastic regions; it further produces an estimate of the largest Lyapunov characteristic number and can give information about the location of stable and unstable periodic orbits. A number of these features are illustrated on the 2-dimensional Hénon-Heiles model. Next, a perturbed 3-dimensional quartic oscillator is studied, which enables illustrating that the MEGNO can also give a clear view of resonance structures. There are ample illustrations of numerical computations; among other things, these reveal the existence of certain stability zones at resonance intersections. In Section 4, a generalization \(Y_{m,n}(\gamma)\) of the MEGNO is introduced, which reduces to \(Y(\gamma)\) for \(m=1\), \(n=-1\). It appears that also the case \(m=2\), \(n=0\) can give relevant information to distinguish regular from chaotic behaviour. In the final section, the idea of \(Y_{m,n}\) is extended also to area-preserving maps.