Towards the ground state of molecules via diffusion Monte Carlo on neural networks

TL;DR

This paper introduces a neural network-based trial wavefunction for diffusion Monte Carlo, significantly improving the accuracy and efficiency of calculating ground state energies of molecules and materials, advancing electronic structure methods.

Contribution

The work presents a novel neural network approach integrated with fixed-node DMC, surpassing existing neural network methods in accuracy and efficiency for electronic structure calculations.

Findings

Achieved higher accuracy in ground state energy calculations.

Demonstrated efficiency improvements over previous neural network methods.

Provided a new benchmark for correlated electronic wavefunction solutions.

Abstract

Diffusion Monte Carlo (DMC) based on fixed-node approximation has enjoyed significant developments in the past decades and become one of the go-to methods when accurate ground state energy of molecules and materials is needed. The remaining bottleneck is the limitations of the inaccurate nodal structure, prohibiting more challenging electron correlation problems to be tackled with DMC. In this work, we apply the neural-network based trial wavefunction in fixed-node DMC, which allows accurate calculation of a broad range of atomic and molecular systems of different electronic characteristics. Our method is superior in both accuracy and efficiency compared to state-of-the-art neural network methods using variational Monte Carlo. Overall, this computational framework provides a new benchmark for accurate solution of correlated electronic wavefunction and also shed light on the chemical understanding of molecules.