Spectrum-splitting approach for Fermi-operator expansion in all-electron Kohn-Sham DFT calculations

TL;DR

This paper introduces a spectrum-splitting method for all-electron Kohn-Sham DFT calculations that significantly improves computational efficiency and accuracy by dividing the eigenspace into core and valence parts, reducing the polynomial degree needed.

Contribution

The paper presents a novel spectrum-splitting approach that enhances Fermi-operator expansion efficiency in all-electron DFT calculations by dividing the eigenspace and employing Chebyshev filtering and localization.

Findings

Achieved five-fold reduction in polynomial degree for silicon nano-clusters.

Demonstrated high accuracy in ground-state energies (~10^{-4} Ha/atom).

Showed the method's necessity for accurate gold atom and nano-cluster calculations.

Abstract

We present a spectrum-splitting approach to conduct all-electron Kohn-Sham density functional theory (DFT) calculations by employing Fermi-operator expansion of the Kohn-Sham Hamiltonian. The proposed approach splits the subspace containing the occupied eigenspace into a core-subspace, spanned by the core eigenfunctions, and its complement, the valence-subspace, and thereby enables an efficient computation of the Fermi-operator expansion by reducing the expansion to the valence-subspace projected Kohn-Sham Hamiltonian. The key ideas used in our approach are: (i) employ Chebyshev filtering to compute a subspace containing the occupied states followed by a localization procedure to generate non-orthogonal localized functions spanning the Chebyshev-filtered subspace; (ii) compute the Kohn-Sham Hamiltonian projected onto the valence-subspace; (iii) employ Fermi-operator expansion in terms of the valence-subspace projected Hamiltonian to compute the density matrix, electron-density and band energy. We demonstrate the accuracy and performance of the method on benchmark materials systems involving silicon nano-clusters up to 1330 electrons, a single gold atom and a six-atom gold nano-cluster. The benchmark studies on silicon nano-clusters revealed a staggering five-fold reduction in the Fermi-operator expansion polynomial degree by using the spectrum-splitting approach for accuracies in the ground-state energies of $\sim 10^{-4} Ha/atom$ with respect to reference calculations. Further, numerical investigations on gold suggest that spectrum splitting is indispensable to achieve meaningful accuracies, while employing Fermi-operator expansion.