How Well Does Kohn-Sham Regularizer Work for Weakly Correlated Systems?

TL;DR

This paper evaluates the effectiveness of Kohn-Sham regularizer (KSR) in weakly correlated systems, introducing spin-adapted KSR and demonstrating its superior performance in predicting ground-state energies.

Contribution

The paper introduces spin-adapted KSR (sKSR) with trainable local, semilocal, and nonlocal approximations, and assesses its generalizability and accuracy in weakly correlated systems.

Findings

Nonlocal functional outperforms existing machine learning functionals.

Semilocal approximation has comparable generalization error to other approaches.

Mean absolute error of 2.7 milli-Hartrees on test systems.

Abstract

Kohn-Sham regularizer (KSR) is a differentiable machine learning approach to finding the exchange-correlation functional in Kohn-Sham density functional theory (DFT) that works for strongly correlated systems. Here we test KSR for weak correlation. We propose spin-adapted KSR (sKSR) with trainable local, semilocal, and nonlocal approximations found by minimizing density and total energy loss. We assess the atoms-to-molecules generalizability by training on one-dimensional (1D) H, He, Li, Be, Be$^{++}$ and testing on 1D hydrogen chains, LiH, BeH$_2$, and helium hydride complexes. The generalization error from our semilocal approximation is comparable to other differentiable approaches, but our nonlocal functional outperforms any existing machine learning functionals, predicting ground-state energies of test systems with a mean absolute error of 2.7 milli-Hartrees.