Phase diagram study of a two-dimensional frustrated antiferromagnet via unsupervised machine learning

TL;DR

This paper employs unsupervised machine learning, specifically convolutional autoencoders, to map the phase diagram of a 2D frustrated antiferromagnetic Ising model, detecting phase transitions and ground state degeneracy without prior phase knowledge.

Contribution

It introduces a novel application of autoencoders for phase classification and transition detection in a frustrated magnetic system using anomaly detection techniques.

Findings

Autoencoders successfully identify phase boundaries.

High-temperature training data can still detect phases.

Autoencoders reveal ground state degeneracy through reconstruction error.

Abstract

We apply unsupervised learning techniques to classify the different phases of the $J_1-J_2$ antiferromagnetic Ising model on the honeycomb lattice. We construct the phase diagram of the system using convolutional autoencoders. These neural networks can detect phase transitions in the system via `anomaly detection', without the need for any label or a priori knowledge of the phases. We present different ways of training these autoencoders and we evaluate them to discriminate between distinct magnetic phases. In this process, we highlight the case of high temperature or even random training data. Finally, we analyze the capability of the autoencoder to detect the ground state degeneracy through the reconstruction error.