Discrete discontinuous basis projection method for large-scale electronic structure calculations

TL;DR

This paper introduces a discontinuous basis projection method that significantly accelerates large-scale electronic structure calculations by reducing eigenproblem dimensions without sacrificing accuracy.

Contribution

The method constructs an efficient, systematically improvable basis to project the Hamiltonian, enabling large reductions in eigenproblem size for electronic structure computations.

Findings

Achieves accurate energies and forces with 8-25 basis functions per atom.

Reduces eigenproblem dimension by 1-3 orders of magnitude.

Applicable across various systems with maintained accuracy.

Abstract

We present an approach to accelerate real-space electronic structure methods several fold, without loss of accuracy, by reducing the dimension of the discrete eigenproblem that must be solved. To accomplish this, we construct an efficient, systematically improvable, discontinuous basis spanning the occupied subspace and project the real-space Hamiltonian onto the span. In calculations on a range of systems, we find that accurate energies and forces are obtained with 8--25 basis functions per atom, reducing the dimension of the associated real-space eigenproblems by 1--3 orders of magnitude.