2006, Vol.9, No.3, pp.209-239
In this paper I discuss some aspects of a problem with a
long history, which continues to be of current interest due to its
great importance to many applications: The stability of motion in
N-degree of freedom Hamiltonian systems. I will start with N
small and proceed to the case of N arbitrarily large, in an
attempt to understand the thermodynamic limit, where
and statistical mechanics is expected to take over from
classical mechanics. Our domain is the Euclidian phase space
of the qk,
pk, k=1,2,...,N, position and
momentum coordinates. The dynamics is governed by Hamilton's
equations of motion and the solutions (or orbits) lie on a
(2N-1)-dimensional compact manifold (the so-called "energy
surface"), defined by H(q1, ..., qN, p1, ..., pN)=E,
where H is the Hamiltonian and E is the (constant) energy of the
system. Of primary importance in our discussion is the connection
between local and global dynamics, i.e. the seemingly paradoxical
relevance of events occurring in small scale regions of the energy
surface, to the stability of motion in large domains, affecting the
properties of the system as a whole. This link is provided by a
detailed study of what I call Simple Periodic Orbits (SPOs), i.e
periodic solutions where all variables oscillate with equal
frequencies,
, returning to the same
values after a single maximum (and minimum) in their evolution over
one period T.
We will start by recalling some fundamental
concepts concerning the stability of dynamical systems, as
introduced by one of the forefathers of this field, the great
Russian Mathematician A. M. Lyapunov, more than 110 years ago.
First, we shall review his two main methods for studying the
solutions in the vicinity of equilibrium points that led to his
proof of the existence of periodic solutions, as continuations of
the corresponding oscillations of the linearized system of
equations. We will then apply his theory of the continuation of
normal modes of N-degree of freedom Hamiltonian systems to the
famous Fermi-Pasta-Ulam lattice to explain how these SPOs can help
resolve the paradox of the FPU recurrences. I will then discuss how
the study of other SPOs, corresponding to in-phase and
out-of-phase oscillations in the FPU and other Hamiltonians, help
us understand the transition to large scale chaotic behavior,
characterized by invariant spectra of Lyapunov exponents, as well as
identify domains where the motion is still quasiperiodic, lying on
invariant N-dimensional tori (on which the
are
rationally independent). Finally, I will report on a recent
discovery of a very efficient spectrum of indices distinguishing
ordered from chaotic motions in conservative dynamical systems,
called the GALIk, k=1,2,...,2N.
These represent an important
generalization of the SALI, used by many researchers to identify
domains of chaos and order not only in N-degree-of-freedom
Hamiltonian systems, but also 2N-dimensional symplectic maps.
Full text: Acrobat PDF (766KB)
Copyright © Nonlinear Phenomena in Complex Systems. Last updated: September 6, 2006