Method for fast and stable first-principles molecular dynamics simulations

What is it about?

First principles quantum-based Born-Oppenheimer molecular dynamics simulations are often painfully slow or exhibit instabilities with a systematic drift in the total energy. This paper presents a framework that gets around some of those shortcomings. It combines the best of direct Born-Oppenheimer and Car-Parrinello molecular dynamics. The method avoids the non-linear eigenvalue problem and the iterative self-consistent-field optimization required prior to each force evaluation in direct Born-Oppenheimer molecular dynamics. It also avoids the tuning of material-dependent fictitious electron mass parameters and the limitations of a shorter integration time steps in Car-Parrinello molecular dynamics.

Featured Image

Photo by Raphaël Biscaldi on Unsplash

Photo by Raphaël Biscaldi on Unsplash

Why is it important?

In chemical systems with electronic instabilities, for example, where the electronic gap can be opening and closing, it is often very hard and costly to perform stable Born-Oppenheimer or Car-Parrinello molecular dynamics simulations. This paper presents a theory that provides a new and easier way to perform such simulations, which works even for highly unstable reactive chemical systems. This is of particular importance when we try to understand more complex systems at larger time and length scales.

Perspectives