A Harris-type van der Waals density functional scheme

TL;DR

This paper introduces a Harris-type scheme to accelerate van der Waals density functional calculations for large biomolecular systems, enabling more efficient non-selfconsistent studies of dispersion interactions.

Contribution

It adapts the Harris scheme to vdW-DF, linking dispersion effects with self-consistency in DFT for large sparse systems.

Findings

Significant speed-up in vdW-DF calculations for large systems

Formal connection between dispersion effects and self-consistency

Validation of the Harris scheme's effectiveness in this context

Abstract

Large biomolecular systems, whose function may involve thousands of atoms, cannot easily be addressed with parameter-free density functional theory (DFT) calculations. Until recently a central problem was that such systems possess an inherent sparseness, that is, they are formed from components that are mutually separated by low-electron-density regions where dispersive forces contribute significantly to the cohesion and behavior. The introduction of, for example, the van der Waals density functional (vdW-DF) method [PRL 92, 246401 (2004)] has addressed part of this sparse-matter system challenge. However, while a vdW-DF study is often as computationally efficient as a study performed in the generalized gradient approximation, the scope of large-sparse-matter DFT is still limited by computer time and memory. It is costly to self-consistently determine the electron wavefunctions and hence the kinetic-energy repulsion. In this paper we propose and evaluate an adaption of the Harris scheme [PRB 31, 1770 (1985)]. This is done to speed up non-selfconsistent vdW-DF studies of molecular-system interaction energies. Also, the Harris-type analysis establishes a formal link between dispersion-interaction effects on the effective potential for electron dynamics and the impact of including selfconsistency in vdW-DF calculations [PRB 76, 125112 (2007)].