Simulating the Schwinger Model with a Regularized Variational Quantum Imaginary Time Evolution

TL;DR

This paper introduces a regularized variational quantum imaginary time evolution (rVQITE) method to improve the simulation of the Schwinger model, overcoming stability issues and achieving accurate phase diagram results.

Contribution

We propose a regularized version of VQITE, called rVQITE, which enhances stability and convergence in simulating the Schwinger model using quantum algorithms.

Findings

rVQITE outperforms standard VQITE in convergence speed

Simulated phase diagrams match exact computational results

Method demonstrates potential for non-perturbative quantum simulations

Abstract

The Schwinger model serves as a benchmark for testing non-perturbative algorithms in quantum chromodynamics (QCD), emphasizing its similarities to QCD in strong coupling regimes, primarily due to the phenomena such as confinement and charge screening. However, classical algorithms encounter challenges when simulating the Schwinger model, such as the "sign problem" and the difficulty in handling large-scale systems. These limitations motivate the exploration of alternative simulation approaches, including quantum computing techniques, to overcome the obstacles. While existing variational quantum algorithms (VQAs) methods for simulating the Schwinger model primarily rely on mathematical gradient-based optimization, which sometimes fail to provide intuitive and physically-guided optimization pathways. In contrast, the Variational Quantum Imaginary Time Evolution (VQITE) method offers a physically-inspired optimization approach. Therefore, we introduce that VQITE holds promise as a potent tool for simulating the Schwinger model. However, the standard VQITE method is not sufficiently stable, as it encounters difficulties with the non-invertible matrix problem. To address this issue, we have proposed a regularized version of the VQITE, which we have named the Regularized-VQITE (rVQITE) method, as it incorporates a truncation-based approach. Through numerical simulations, we demonstrate that our proposed rVQITE approach achieves better performance and exhibits faster convergence compared to other related techniques. We employ the rVQITE method to simulate the phase diagrams of various physical observables in the Schwinger model, and the resulting phase boundaries are in agreement with those obtained from an exact computational approach.