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.
Description
[2003.06413] Equivariant flow-based sampling for lattice gauge theory
%0 Journal Article
%1 kanwar2020equivariant
%A Kanwar, Gurtej
%A Albergo, Michael S.
%A Boyda, Denis
%A Cranmer, Kyle
%A Hackett, Daniel C.
%A Racanière, Sébastien
%A Rezende, Danilo Jimenez
%A Shanahan, Phiala E.
%D 2020
%K deep-learning equivariance graphs physics sampling
%T Equivariant flow-based sampling for lattice gauge theory
%U http://arxiv.org/abs/2003.06413
%X 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{kanwar2020equivariant,
abstract = {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.},
added-at = {2020-03-16T14:43:00.000+0100},
author = {Kanwar, Gurtej and Albergo, Michael S. and Boyda, Denis and Cranmer, Kyle and Hackett, Daniel C. and Racanière, Sébastien and Rezende, Danilo Jimenez and Shanahan, Phiala E.},
biburl = {https://www.bibsonomy.org/bibtex/2c4b948f1b49e8dbffbe32f9a715da72e/kirk86},
description = {[2003.06413] Equivariant flow-based sampling for lattice gauge theory},
interhash = {ab76b5d88f9bcc281b4c8301c3aa57bf},
intrahash = {c4b948f1b49e8dbffbe32f9a715da72e},
keywords = {deep-learning equivariance graphs physics sampling},
note = {cite arxiv:2003.06413Comment: 6 pages, 4 figures},
timestamp = {2020-03-16T14:43:00.000+0100},
title = {Equivariant flow-based sampling for lattice gauge theory},
url = {http://arxiv.org/abs/2003.06413},
year = 2020
}