Machine learning nonequilibrium phase transitions in charge-density wave insulators

TL;DR

This paper introduces a machine learning framework that predicts electronic forces in nonequilibrium charge-density wave insulators, enabling efficient and accurate simulations of phase transitions driven by electronic forces.

Contribution

The authors develop a neural network-based force field that bypasses expensive Green's function calculations, significantly improving simulation efficiency for nonequilibrium lattice dynamics.

Findings

Neural network accurately predicts local electronic forces.

Force field reproduces domain wall motion and phase transition dynamics.

Achieves orders of magnitude faster simulations than traditional methods.

Abstract

Nonequilibrium electronic forces play a central role in voltage-driven phase transitions but are notoriously expensive to evaluate in dynamical simulations. Here we develop a machine learning framework for adiabatic lattice dynamics coupled to nonequilibrium electrons, and demonstrate it for a gating induced insulator to metal transition out of a charge density wave state in the Holstein model. Although exact electronic forces can be obtained from nonequilibrium Green's function (NEGF) calculations, their high computational cost renders long time dynamical simulations prohibitively expensive. By exploiting the locality of the electronic response, we train a neural network to directly predict instantaneous local electronic forces from the lattice configuration, thereby bypassing repeated NEGF calculations during time evolution. When combined with Brownian dynamics, the resulting machine learning force field quantitatively reproduces domain wall motion and nonequilibrium phase transition dynamics obtained from full NEGF simulations, while achieving orders of magnitude gains in computational efficiency. Our results establish direct force learning as an efficient and accurate approach for simulating nonequilibrium lattice dynamics in driven quantum materials.