Stable and Efficient Linear Scaling First-Principles Molecular Dynamics for 10,000+ atoms

TL;DR

This paper demonstrates that combining the extended Lagrangian Born-Oppenheimer molecular dynamics method with the density matrix approach enables efficient, accurate, and reliable first-principles molecular dynamics simulations of systems exceeding 30,000 atoms using linear-scaling DFT techniques.

Contribution

The paper introduces a robust method for large-scale O(N) FPMD simulations by integrating advanced techniques, enabling accurate simulations of very large systems.

Findings

Successful simulation of 32,768 atoms with reliable accuracy

Identification of conditions for accurate O(N) FPMD calculations

Validation of the combined method's efficiency and robustness

Abstract

The recent progress of linear-scaling or O(N) methods in the density functional theory (DFT) is remarkable. We expect that first-principles molecular dynamics (FPMD) simulations based on DFT can now treat more realistic and complex systems using the O(N) technique. However, very few examples of O(N) FPMD simulations exist so far and the information for the accuracy or reliability of the simulations is very limited. In this paper, we show that efficient and robust O(N) FPMD simulations are now possible by the combination of the extended Lagrangian Born-Oppenheimer molecular dynamics method, which was recently proposed by Niklasson et al (Phys. Rev. Lett. 100, 123004 (2008)), and the density matrix method as an O(N) technique. Using our linear-scaling DFT code Conquest, we investigate the reliable calculation conditions for the accurate O(N) FPMD and demonstrate that we are now able to do actual and reliable self-consistent FPMD simulation of a very large system containing 32,768 atoms.