Learning phase transitions from dynamics

TL;DR

This paper introduces a method using recurrent neural networks to classify phases of matter from dynamical data, successfully identifying phase diagrams for many-body localization and time-crystalline phases.

Contribution

The paper demonstrates a novel application of recurrent neural networks to classify phases of matter directly from dynamical observables, extending phase diagram analysis to dynamical regimes.

Findings

Accurately reproduces known phase diagrams for many-body localization.

Identifies dynamical phases in a driven system with a time-crystalline phase.

Shows neural networks can classify phases from experimental dynamical data.

Abstract

We propose the use of recurrent neural networks for classifying phases of matter based on the dynamics of experimentally accessible observables. We demonstrate this approach by training recurrent networks on the magnetization traces of two distinct models of one-dimensional disordered and interacting spin chains. The obtained phase diagram for a well-studied model of the many-body localization transition shows excellent agreement with previously known results obtained from time-independent entanglement spectra. For a periodically-driven model featuring an inherently dynamical time-crystalline phase, the phase diagram that our network traces in a previously-unexplored regime coincides with an order parameter for its expected phases.