Macroscopic Determinism in Interacting Systems Using Large Deviation Theory

Abstract

We consider the quasi-deterministic behavior of systems with a large number, $n$, of deterministically interacting constituents. This work extends the results of a previous paper [J. Stat. Phys. 99:1225-1249 (2000)] to include vector-valued observables on interacting systems. The approach used here, however, differs markedly in that a level-1 large deviation principle (LDP) on joint observables, rather than a level-2 LDP on empirical distributions, is employed. As before, we seek a mapping $\psi_{t}$ on the set of (possibly vector-valued) macrostates such that, when the macrostate is given to be $a_0$ at time zero, the macrostate at time $t$ is $\psi_{t}(a_0)$ with a probability approaching one as $n$ tends to infinity. We show that such a map exists and derives from a generalized dynamic free energy function, provided the latter is everywhere well defined, finite, and differentiable. We discuss some general properties of $\psi_{t}$ relevant to issues of irreversibility and end with an example of a simple interacting lattice, for which an exact macroscopic solution is obtained.