Short range DFT combined with long-range local RPA within a range-separated hybrid DFT framework

TL;DR

This paper introduces a method combining short-range DFT with long-range local RPA within a range-separated hybrid framework, reducing computational effort for long-range correlation calculations in molecular dimers.

Contribution

It presents a novel approach that integrates short-range DFT with long-range RPA, utilizing excitation selection and dispersion-only approximation to improve efficiency.

Findings

Effective in reducing computational cost

Applicable to various molecular dimers

Maintains accuracy in long-range correlation calculations

Abstract

Selecting excitations in localized orbitals to calculate long-range correlation contributions to range-separated density-functional theory can reduce the overall computational effort significantly. Beyond simple selection schemes of excited determinants, the dispersion-only approximation, which avoids counterpoise-corrected monomer calculations, is shown to be particularly interesting in this context, which we apply to the random-phase approximation. The approach has been tested on dimers of formamide, water, methane and benzene.