Forces from stochastic density functional theory under nonorthogonal atom-centered basis sets

TL;DR

This paper introduces a formalism for calculating forces in stochastic density functional theory using nonorthogonal atom-centered basis sets, demonstrating its accuracy and applicability to a solvated peptide system.

Contribution

It develops a new force calculation method within stochastic DFT for nonorthogonal basis sets and applies an embedded-fragment approach to reduce statistical errors.

Findings

Systematic bias in forces is about 0.065 eV/Å with 120 stochastic orbitals.

Bias magnitude is system-size independent.

Bond length deviations are less than 1% compared to deterministic DFT.

Abstract

We develop a formalism for calculating forces on the nuclei within the linear-scaling stochastic density functional theory (sDFT) in a nonorthogonal atom-centered basis-set representation (Fabian et al. WIREs Comput Mol Sci. 2019;e1412. https://doi.org/10.1002/wcms.1412) and apply it to Tryptophan Zipper 2 (Trp-zip2) peptide solvated in water. We use an embedded-fragment approach to reduce the statistical errors (fluctuation and systematic bias), where the entire peptide is the main fragment and the remaining 425 water molecules are grouped into small fragments. We analyze the magnitude of the statistical errors in the forces and find that the systematic bias is of the order of $0.065\,eV/\r{A}$ ($\sim1.2\times10^{-3}E_{h}/a_{0}$) when 120 stochastic orbitals are used, independently of systems size. This magnitude of bias is sufficiently small to ensure that the bond lengths estimated by stochastic DFT (within a Langevin molecular dynamics simulation) will deviate by less than 1% from those predicted by a deterministic calculation.