A Second-Order CASSCF Algorithm with the Cholesky Decomposition of the Two-Electron Integrals

TL;DR

This paper introduces a second-order CASSCF algorithm combined with Cholesky decomposition to reduce computational costs, enabling routine calculations on larger molecular systems with improved efficiency.

Contribution

The paper presents a novel implementation of a second-order CASSCF algorithm using Cholesky decomposition, enhancing scalability and performance for large molecules.

Findings

Significant reduction in computational cost and memory usage.

Successful benchmark calculations on molecules with up to 3000 basis functions.

Improved parallel performance of the CASSCF algorithm.

Abstract

In this contribution, we present the implementation of a second-order CASSCF algorithm in conjunction with the Cholesky decomposition of the two-electron repulsion integrals. The algorithm, called Norm-Extended Optimization, guarantees convergence of the optimization, but it involves the full Hessian of the wavefunction and is therefore computationally expensive. Coupling the second-order procedure with the Cholesky decomposition leads to a significant reduction in the computational cost, reduced memory requirements, and an improved parallel performance. As a result, CASSCF calculations of larger molecular systems become possible as a routine task. The performance of the new implementation is illustrated by means of benchmark calculations on molecules of increasing size, with up to about 3000 basis functions and 14 active orbitals.