Detecting quantum phase transitions in a frustrated spin chain via transfer learning of a quantum classifier algorithm

TL;DR

This paper introduces a machine learning framework, including transfer learning and quantum classifiers, to detect quantum phase transitions in a frustrated spin chain model, offering an alternative to traditional methods.

Contribution

It demonstrates the use of transfer learning with classical and quantum machine learning to identify multiple phases in a quantum spin model, even with limited training data.

Findings

Machine learning can successfully classify phases in the ANNNI model.

Transfer learning enables detection of new phases with limited training.

Quantum nearest neighbors outperform classical methods in phase detection.

Abstract

The classification of phases and the detection of phase transitions are central and challenging tasks in diverse fields. Within physics, it relies on the identification of order parameters and the analysis of singularities in the free energy and its derivatives. Here, we propose an alternative framework to identify quantum phase transitions. Using the axial next-nearest neighbor Ising (ANNNI) model as a benchmark, we show how machine learning can detect three phases (ferromagnetic, paramagnetic, and a cluster of the antiphase with the floating phase). Employing supervised learning, we demonstrate the feasibility of transfer learning. Specifically, a machine trained only with nearest-neighbor interactions can learn to identify a new type of phase occurring when next-nearest-neighbor interactions are introduced. We also compare the performance of common classical machine learning methods with a version of the quantum nearest neighbors (QNN) algorithm.