Application of neural network to sign problem via path optimization method

TL;DR

This paper proposes a neural network-based path optimization method to mitigate the sign problem in quantum field theory simulations, demonstrating improved phase factors and enabling safer Monte Carlo computations.

Contribution

It introduces a neural network approach to optimize integration paths in complexified space, addressing the sign problem in multi-dimensional quantum field theories.

Findings

Enhanced average phase factor after path optimization

Successful application to two-dimensional $^4$ theory at finite chemical potential

Enables safer hybrid Monte-Carlo simulations

Abstract

We introduce the feedforward neural network to attack the sign problem via the path optimization method. The variables of integration is complexified and the integration path is optimized in the complexified space by minimizing the cost function which reflects the seriousness of the sign problem. For the preparation and optimization of the integral path in multi-dimensional systems, we utilize the feedforward neural network. We examine the validity and usefulness of the method in the two-dimensional complex $\lambda \phi^4$ theory at finite chemical potential as an example of the quantum field theory having the sign problem. We show that the average phase factor is significantly enhanced after the optimization and then we can safely perform the hybrid Monte-Carlo method.