Machine-learning-assisted Monte Carlo fails at sampling computationally hard problems
Disordered Systems and Neural Networks
2023-03-15 v3
Abstract
Several strategies have been recently proposed in order to improve Monte Carlo sampling efficiency using machine learning tools. Here, we challenge these methods by considering a class of problems that are known to be exponentially hard to sample using conventional local Monte Carlo at low enough temperatures. In particular, we study the antiferromagnetic Potts model on a random graph, which reduces to the coloring of random graphs at zero temperature. We test several machine-learning-assisted Monte Carlo approaches, and we find that they all fail. Our work thus provides good benchmarks for future proposals for smart sampling algorithms.
Cite
@article{arxiv.2210.11145,
title = {Machine-learning-assisted Monte Carlo fails at sampling computationally hard problems},
author = {Simone Ciarella and Jeanne Trinquier and Martin Weigt and Francesco Zamponi},
journal= {arXiv preprint arXiv:2210.11145},
year = {2023}
}