Evaluation of gradient corrections in grid-free density functional theory

Abstract

The Almlöf–Zheng approach to grid-free density functional theory (DFT) uses the resolution of the identity (RI) instead of a finite grid to evaluate the integrals. Application of the RI can lead to stability problems, particularly when gradients are involved. The focus of the current work is on choosing a stable method of evaluating the gradient correction using the RI. A stable method is compared to several unstable methods.