We define a class of machine-learned flow-based sampling algorithms for lattice gauge theories that are gauge-invariant by construction. We demonstrate the application of this framework to U(1) gauge theory in two spacetime dimensions, and find that near critical points in parameter space the approach is orders of magnitude more efficient at sampling topological quantities than more traditional sampling procedures such as Hybrid Monte Carlo and Heat Bath.
@article{arxiv.2003.06413,
title = {Equivariant flow-based sampling for lattice gauge theory},
author = {Gurtej Kanwar and Michael S. Albergo and Denis Boyda and Kyle Cranmer and Daniel C. Hackett and Sébastien Racanière and Danilo Jimenez Rezende and Phiala E. Shanahan},
journal= {arXiv preprint arXiv:2003.06413},
year = {2020}
}