We propose a different way to satisfy both gradient expansion limiting behavior and the Lieb-Oxford bound in a generalized gradient approximation exchange functional by extension of the Perdew-Burke-Ernzerhof (PBE) form. Motivation includes early and recent exploration of modified values for the gradient expansion coefficient in the PBE exchange-correlation functional (cf. the PBEsol functional) and earlier experience with a numerical cutoff for large-s (s proportional to absolute value(vector differential n)/n(4/3)) in a version of the deMon molecular code. For either the original PBE or the PBEsol choice of the gradient coefficient, we find improved performance from using an s-dependent (spatially varying) satisfaction of the Lieb-Oxford bound which quenches to uniform electron gas behavior at large s. The mean absolute deviations (MADs) in atomization energies for a widely used test set of 20 small molecules are reduced by about 22% relative to PBE and PBEsol. For these small molecules, the bond length MADs are essentially unchanged.
No takes yet. Share an insight, caveat, or question.
Vela et al. (2009) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: