Dynamical large deviations of two-dimensional kinetically constrained models using a neural-network state ansatz

TL;DR

This paper introduces a neural network approach to analyze dynamical large deviations in classical systems, specifically applying it to the Fredrickson-Andersen model in one and two dimensions, revealing new insights into its dynamical activity.

Contribution

It demonstrates the application of a neural network ansatz, originally for quantum systems, to classical dynamical large deviations, including the first size-scaling analysis in two dimensions.

Findings

Successfully computed the scaled cumulant-generating function for the model.

Performed the first size-scaling analysis of dynamical activity in two dimensions.

Showed broad applicability of neural-network state ansatz across physics domains.

Abstract

We use a neural network ansatz originally designed for the variational optimization of quantum systems to study dynamical large deviations in classical ones. We obtain the scaled cumulant-generating function for the dynamical activity of the Fredrickson-Andersen model, a prototypical kinetically constrained model, in one and two dimensions, and present the first size-scaling analysis of the dynamical activity in two dimensions. These results provide a new route to the study of dynamical large-deviation functions, and highlight the broad applicability of the neural-network state ansatz across domains in physics.