Fast construction of the exchange operator in an atom-centered basis with concentric atomic density fitting

TL;DR

This paper introduces a linear-scaling algorithm for efficiently constructing the Hartree-Fock exchange matrix using concentric atomic density fitting, enabling large-scale computations with improved speed and scalability.

Contribution

The paper presents a novel linear-scaling method leveraging distance dependence and density matrix decay for efficient exchange matrix construction in large molecules.

Findings

Successfully computed systems with up to 1536 atoms and 15585 basis functions.

Achieved the largest quadruple-zeta Hartree-Fock computation to date.

Efficient screening of high angular momentum contributions enhances scalability.

Abstract

A linear-scaling algorithm is presented for computing the Hartree-Fock (HF) exchange matrix using concentric atomic density fitting. The algorithm utilizes the stronger distance dependence of the three-center electron repulsion integrals along with the rapid decay of the density matrix to accelerate the construction of the exchange matrix. The new algorithm is tested with computations on systems with up to 1536 atoms and 15585 basis functions, the latter of which represents, to our knowledge, the largest quadruple-zeta HF computation ever performed. Our method handles screening of high angular momentum contributions in a particularly efficient manner, allowing the use of larger basis sets for large molecules without a prohibitive increase in cost.