Reducing autocorrelation times in lattice simulations with generative adversarial networks
Authors:
- Jan M PawlowskiJPJan M Pawlowski
- Julian M UrbanJUJulian M Urban
Open Access
- 8 October 2020
- journal article
- research article
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 1 (4) , 045011
Abstract
Short autocorrelation times are essential for a reliable error assessment in Monte Carlo simulations of lattice systems. In many interesting scenarios, the decay of autocorrelations in the Markov chain is prohibitively slow. Generative samplers can provide statistically independent field configurations, thereby potentially ameliorating these issues. In this work, the applicability of neural samplers to this problem is investigated. Specifically, we work with a generative adversarial network (GAN). We propose to address difficulties regarding its statistical exactness through the implementation of an overrelaxation step, by searching the latent space of the trained generator network. This procedure can be incorporated into a standard Monte Carlo algorithm, which then permits a sensible assessment of ergodicity and balance based on consistency checks. Numerical results for real, scalar φ 4-theory in two dimensions are presented. We achieve a significant reduction of autocorrelations while accurately reproducing the correct statistics. We discuss possible improvements to the approach as well as potential solutions to persisting issues.Keywords
All Related Versions
Funding Information
- Deutsche Forschungsgemeinschaft (EXC 2181/1 - 390900948)
- EMMI
- Bundesministerium für Forschung, Technologie und Raumfahrt (05P18VHFCA)
This publication has 34 references indexed in Scilit:
- Discovering phases, phase transitions, and crossovers through unsupervised machine learning: A critical examinationPhysical Review E, 2017
- Learning phase transitions by confusionNature Physics, 2017
- Machine learning phases of matterNature Physics, 2017
- Solving the quantum many-body problem with artificial neural networksScience, 2017
- Self-learning Monte Carlo methodPhysical Review B, 2017
- Accelerated Monte Carlo simulations with restricted Boltzmann machinesPhysical Review B, 2017
- Restricted Boltzmann machines for the long range Ising modelsModern Physics Letters B, 2016
- Discovering phase transitions with unsupervised learningPhysical Review B, 2016
- Learning thermodynamics with Boltzmann machinesPhysical Review B, 2016
- Overrelaxation and Monte Carlo simulationPhysical Review D, 1987