Speeding-up Ab Initio Molecular Dynamics with Hybrid Functionals using Adaptively Compressed Exchange Operator based Multiple Timestepping

TL;DR

This paper introduces a method combining adaptively compressed exchange operators with multiple time step integration to significantly accelerate hybrid functional ab initio molecular dynamics simulations, making them more computationally feasible.

Contribution

The paper presents a novel approach that reduces the computational cost of hybrid functional AIMD by integrating adaptively compressed exchange operators with multiple time stepping.

Findings

Achieved significant speed-up in AIMD simulations using the proposed method.

Validated the approach on a realistic condensed matter system.

Demonstrated the method's efficiency and potential for routine use.

Abstract

Ab initio molecular dynamics (AIMD) simulations using hybrid density functionals and plane waves are of great interest owing to the accuracy of this approach in treating condensed matter systems. On the other hand, such AIMD calculations are not routinely carried out since the computational cost involved in applying the Hartree Fock exchange operator is very high. In this work, we make use of a strategy that combines adaptively compressed exchange operator formulation and multiple time step integration to significantly reduce the computational cost of these simulations. We demonstrate the efficiency of this approach for a realistic condensed matter system.