This repository contains a Julia implementation of the generative approach to phase-classification tasks introduced in our paper.
One of the central tasks in many-body physics is the determination of phase diagrams, which can be cast as a classification problem. Typically, classification problems are tackled using discriminative classifiers that explicitly model the conditional probability of labels given a sample. Here we show that phase-classification problems are naturally suitable to be solved using generative classifiers that are based on probabilistic models of the measurement statistics underlying the physical system. Such a generative approach benefits from modeling concepts native to the realm of statistical and quantum physics, as well as recent advances in machine learning. This yields a powerful framework for mapping out phase diagrams of classical and quantum systems in an automated fashion capable of leveraging prior system knowledge.
contains code to map out phase diagrams given generative models. The source files can be found in source folder. We provide exemplary code for
-
the equilibrium phase diagram of the two-dimensional anisotropic Ising model (L=20), see this folder,
-
and the ground-state phase diagram of the cluster-Ising model (L=7), see this folder.
The corresponding data can be found in the data folder (large files need to be unzipped). Additional data is available upon request. Other physical systems can be analyzed in the same fashion by replacing the corresponding generative models.
- install julia
- download,
activate, andinstantiate[Pkg.instantiate()] our package - individual files can then be executed by calling, e.g.,
julia run_main_2D.jl - uncomment
savefig()functions to save plots
@article{arnold:2023,
title={Mapping out phase diagrams with generative classifiers},
author={Arnold, Julian and Schäfer, Frank and Edelman, Alan and Bruder, Christoph},
journal={arXiv:2306.14894},
year={2023},
url = {https://arxiv.org/abs/2306.14894}
}