Nonequilibrium thermodynamics as a symplecto-contact reduction and relative information entropy

TL;DR

This paper develops a geometric framework for nonequilibrium thermodynamics using symplectic and contact geometry, deriving thermodynamic phase space and equilibrium states through reduction and Legendre submanifolds.

Contribution

It introduces a novel geometric derivation of thermodynamic phase space as a contact manifold via symplectic reduction and relates phase transitions to graph selectors in symplecto-contact geometry.

Findings

Derivation of thermodynamic phase space as a contact manifold.

Interpretation of Maxwell's equal-area law via graph selectors.

Connection between phase transitions and Aubry-Mather theory.

Abstract

Both statistical phase space (SPS), which is $\Gamma = T^*\mathbb R^{3N}$ of $N$-body particle system, and kinetic theory phase space (KTPS), which is the cotangent bundle $T^*\mathcal P(\Gamma)$ of the probability space $\mathcal P(\Gamma)$ thereon, carry canonical symplectic structures. Starting from this first principle, we provide a canonical derivation of thermodynamic phase space (TPS) of nonequilibrium thermodynamics as a contact manifold in two steps. First, regarding the collective observation of observables in SPS as a moment map defined on KTPS, we apply the Marsden-Weinstein reduction and obtain a mesoscopic phase space in between KTPS and TPS as a (infinite dimensional) symplectic fibration. Then we show that the reduced relative information entropy defines a generating function that provides a covariant construction of a thermodynamic equilibrium as a Legendrian submanifold. This Legendrian submanifold is not necessarily graph-like. We interpret the Maxwell construction of \emph{equal-area law} as the procedure of finding a continuous, not necessarily differentiable, thermodynamic potential and explain the associated phase transition by identifying the procedure with that of finding a graph selector in symplecto-contact geometry and in the Aubry-Mather theory of dynamical system.