The LDA-1/2 method applied to atoms and molecules

TL;DR

This paper evaluates the LDA-1/2 method for atoms and molecules, comparing its accuracy to CCSD(T) and G0W0, and analyzes its assumptions to understand its benefits and limitations.

Contribution

The study extends the application of LDA-1/2 to finite systems and compares its performance with established methods, providing insights into its accuracy and theoretical foundations.

Findings

LDA-1/2 achieves similar accuracy to G0W0 for molecular orbital energies.

LDA-1/2 is computationally less demanding than G0W0.

Analysis reveals the strengths and limitations of the LDA-1/2 assumptions.

Abstract

The LDA-1/2 method has proven to be a viable approach for calculating band gaps of semiconductors. To address its accuracy for finite systems, we apply LDA-1/2 to atoms and the molecules of the $GW100$ test set. The obtained energies of the highest-occupied molecular orbitals are validated against CCSD(T) data and the $G_0W_0$ approach of many-body perturbation theory. The accuracy of LDA-1/2 and $G_0W_0$ is found to be the same, where the latter is computationally much more involved. To get insight into the benefits and limitations of the LDA-1/2 method, we analyze the impact of each assumption made in deriving the methodology.