Model-Independent Quantum Phases Classifier

TL;DR

This paper introduces a model-independent quantum phases classifier using k-Nearest Neighbors, capable of identifying phases in unseen models, advancing toward a universal quantum phase detection method.

Contribution

It develops a k-NN based classifier that can identify quantum phases across different models without prior training on those specific models.

Findings

High probability of correctly classifying phases across models

Demonstrates potential for universal quantum phase classification

First step toward a model-agnostic quantum phase detector

Abstract

Machine learning has revolutionized many fields of science and technology. Through the $k$-Nearest Neighbors algorithm, we develop a model-independent classifier, where the algorithm can classify phases of a model to which it has never had access. For this, we study three distinct spin-$1$ models with some common phases: the XXZ chains with uniaxial single-ion-type anisotropy, the bound alternating XXZ chains, and the bilinear biquadratic chain. We show that, with high probability, algorithms trained with two of these models can determine common phases with the third. It is the first step toward a universal classifier, where an algorithm is able to detect any phase with no knowledge about the Hamiltonian, only knowing partial information about the quantum state.