Reduced Scaling Hilbert Space Variational Monte Carlo

TL;DR

This paper demonstrates that the computational cost of Hilbert space variational Monte Carlo methods can be reduced from fifth to fourth order in system size by developing a semi-stochastic approach that exploits tensor sparsity.

Contribution

The authors introduce a semi-stochastic method that reduces the formal cost scaling of Hilbert space VMC, matching real space methods, and show its effectiveness on large molecules.

Findings

Cost scaling reduced from fifth to fourth order.

Accuracy and uncertainty are maintained with the new approach.

Cost crossover achieved at around 50 electrons.

Abstract

We show that for both single-Slater-Jastrow and Jastrow geminal power wave functions, the formal cost scaling of Hilbert space variational Monte Carlo can be reduced from fifth to fourth order in the system size, thus bringing it in line with the long-standing scaling of its real space counterpart. While traditional quantum chemistry methods can reduce costs related to the two-electron integral tensor through resolution of the identity and Cholesky decomposition approaches, we show that such approaches are ineffective in the presence of Hilbert space Jastrow factors. Instead, we develop a simple semi-stochastic approach that can take similar advantage of the near-sparsity of this four-index tensor. Through demonstrations on alkanes of increasing length, we show that accuracy and overall statistical uncertainty are not meaningfully affected and that a total cost crossover is reached as early as 50 electrons.