Chapter 1
Chaos in Statistical Physics
Rainer Klages
School of Mathematical Sciences Queen Mary University of London Mile End Road, London E1 4NS, UK
Max Planck Institute for the Physics of Complex Systems Nöthnitzer Str. 38, D-01187 Dresden, Germany [email protected] This chapter introduces to chaos in dynamical systems and how this theory can be applied to derive fundamental laws of statistical physics from first principles. We first elaborate on the concept of deterministic chaos by defining and calculating Lyapunov exponents and dynamical entropies as fundamental quantities characterising chaos. These quantities are shown to be related to each other by Pesinâs theorem. Considering open systems where particles can escape from a set asks for a generalisation of this theorem which involves fractals, whose properties we briefly describe. We then cross-link this theory to statistical physics by discussing simple random walks on the line, their characterisation in terms of diffusion, and the relation to elementary concepts of Brownian motion. This sets the scene for considering the problem of chaotic diffusion. Here we derive a formula exactly expressing diffusion in terms of the chaos quantities mentioned above.
1. Introduction
A dynamical system is a system, represented by points in abstract space, that evolves in time. Very intuitively, one may say that the path of a point particle generated by a dynamical system looks âchaoticâ if it displays ârandom-lookingâ evolution in time and space. The simplest dynamical systems that can exhibit chaotic dynamics are one-dimensional maps, as we will discuss below. Further details of such dynamics, including a mathematically rigorous definition of chaos, are given in Chapter 6.a Surprisingly, abstract low-dimensional chaotic dynamics bears similarities to the dynamics of interacting physical many-particle systems. This is the key point that we explore in this chapter.
Over the past few decades it was found that famous statistical physical laws like Ohmâs law for electric conduction, Fourierâs law for heat conduction, and Fickâs law for the diffusive spreading of particles, which a long time ago were formulated phenomenologically, can be derived from first principles in chaotic dynamical systems. This sheds new light on the rigorous mathematical foundations of Nonequilibrium Statistical Physics, which is the theory of the dynamics of many-particle systems under external gradients or fields. The external forces induce transport of physical quantities like charge, energy, or matter. The goal of non-equilibrium statistical physics is to derive macroscopic statistical laws describing such transport starting from the microscopic dynamics for the single parts of many-particle systems. While the conventional theory puts in randomness âby handâ by using probabilistic, or stochastic, equations of motion like random walks or stochastic differential equations, recent developments in the theory of dynamical systems enable to do such derivations starting from deterministic equations of motion. Determinism means that no random variables are involved, rather, randomness is generated by chaos in the underlying dynamics. This is the field of research that will be introduced by this chapter.
This theory also illustrates the emergence of complexity in systems under non-equilibrium conditions: Due to the microscopic nonlinear interaction of the single parts in a complex many-particle system novel, non-trivial dynamics, in this case exemplified by universal transport laws, may emerge on macroscopic scales. The dynamics of a complex system, as a whole, is thus different from the sum of its single parts. In the very simplest case, this idea is illustrated by the interaction of a point particle with a scatterer, where the latter is modelled by a one-dimensional map. This is our vehicle of demonstration in the following, because this simple model can be solved rigorously analytically.
Our chapter consists of two sections: In Section 2 we introduce to two important quantities...