Abstract:
The metaGGA class of functionals for describing the exchange-correlation (XC) energy
in density functional theory (DFT) is conventionally constructed as a functional dependent on the
density, density gradient, and kinetic energy density (KED). The addition of the KED makes
metaGGAs more accurate functionals than ones that use the density and its gradient alone but
also more computationally expensive for some applications such as ab initio molecular dynamics
simulations. The calculation of the XC energy in metaGGAs can be made less expensive by
replacing the explicit orbital dependence in the KED with expressions involving only the particle
density and its gradient and Laplacian. We test the validity of recent deorbitalization strategies in
the literature by visualizing their predictions for the KED and related quantities, and comparing
these to exact calculations. For an effective test, we perform these calculations on semiconductor
solids with varying ionicity and atomic number. We explore how well the KED can be
represented by a single metaGGA model in terms of the scaled gradient, Laplacian of the
density, and scaled density. The results show a near-universal linear correlation with Laplacian
and gradient for regions outside of the atomic bond, which can be fit to a straight line when
viewed at the optimal rotation. The calculations of exact KED and electron density are done with
the ABINIT DFT plane-wave pseudopotential code.
