Nonequilibrium Stationary States and Phase Transitions in Directed Ising Models

TL;DR

This paper investigates the nonequilibrium stationary states and phase transitions in directed Ising models across various lattice structures, identifying conditions for Gibbsian states and exploring phase transition phenomena.

Contribution

It characterizes when directed Ising models have Gibbsian stationary states and examines phase transition behavior, especially on Cayley trees.

Findings

Gibbsian stationary states exist only in 1D and 2D models with specific transition rates.

Directed models on high-coordination lattices lack Gibbsian stationary states.

Phase transition on Cayley trees occurs for branching ratio q ≥ 3.

Abstract

We study the nonequilibrium properties of directed Ising models with non conserved dynamics, in which each spin is influenced by only a subset of its nearest neighbours. We treat the following models: (i) the one-dimensional chain; (ii) the two-dimensional square lattice; (iii) the two-dimensional triangular lattice; (iv) the three-dimensional cubic lattice. We raise and answer the question: (a) Under what conditions is the stationary state described by the equilibrium Boltzmann-Gibbs distribution? We show that for models (i), (ii), and (iii), in which each spin "sees" only half of its neighbours, there is a unique set of transition rates, namely with exponential dependence in the local field, for which this is the case. For model (iv), we find that any rates satisfying the constraints required for the stationary measure to be Gibbsian should satisfy detailed balance, ruling out the possibility of directed dynamics. We finally show that directed models on lattices of coordination number $z\ge8$ with exponential rates cannot accommodate a Gibbsian stationary state. We conjecture that this property extends to any form of the rates. We are thus led to the conclusion that directed models with Gibbsian stationary states only exist in dimension one and two. We then raise the question: (b) Do directed Ising models, augmented by Glauber dynamics, exhibit a phase transition to a ferromagnetic state? For the models considered above, the answers are open problems, to the exception of the simple cases (i) and (ii). For Cayley trees, where each spin sees only the spins further from the root, we show that there is a phase transition provided the branching ratio, $q$, satisfies $q \ge 3$.