Beyond the random phase approximation: Improved description of short range correlation by a renormalized adiabatic local density approximation

TL;DR

This paper introduces a renormalized adiabatic local density approximation (rALDA) that enhances the accuracy of short-range correlation energies in electronic structure calculations, outperforming traditional RPA methods.

Contribution

The paper presents a parameter-free rALDA kernel that improves short-range correlation descriptions in solids and molecules, extending the RPA framework with significant accuracy gains.

Findings

rALDA improves molecular atomization energy predictions by a factor of 7 compared to RPA(PBE).

rALDA enhances cohesive energy calculations of solids by a factor of 3 over RPA(PBE).

Including full shell semi-core states is crucial for accurate RPA and rALDA calculations of transition metals.

Abstract

We assess the performance of a recently proposed renormalized adiabatic local density approximation (rALDA) for \textit{ab initio} calculations of electronic correlation energies in solids and molecules. The method is an extension of the random phase approximation (RPA) derived from time-dependent density functional theory and the adiabatic connection fluctuation-dissipation theorem and contains no fitted parameters. The new kernel is shown to preserve the accurate description of dispersive interactions from RPA while significantly improving the description of short range correlation in molecules, insulators, and metals. For molecular atomization energies the rALDA is a factor of 7(4) better than RPA(PBE) when compared to experiments, and a factor of 3(1.5) better than RPA(PBE) for cohesive energies of solids. For transition metals the inclusion of full shell semi-core states is found to be crucial for both RPA and rALDA calculations and can improve the cohesive energies by up to 0.4 eV. Finally we discuss straightforward generalizations of the method, which might improve results even further.