An efficient time-stepping scheme for ab initio molecular dynamics simulations

Abstract

In ab initio molecular dynamics simulations of real-world problems, the simple Verlet method is still widely used for integrating the equations of motion, while more efficient algorithms are routinely used in classical molecular dynamics. We show that if the Verlet method is used in conjunction with pre- and postprocessing, the accuracy of the time integration is significantly improved with only a small computational overhead. The validity of the processed Verlet method is demonstrated in several examples including ab initio molecular dynamics simulations of liquid water. The structural properties obtained from the processed Verlet method are found to be sufficiently accurate even for large time steps close to the stability limit. This approach results in a 2x performance gain over the standard Verlet method for a given accuracy.

32 pages, 11 figures

References in corpus (8)

• An Efficient and Accurate Car-Parrinello-like Approach to Born-Oppenheimer Molecular Dynamics
• Time-reversible Born-Oppenheimer molecular dynamics
• Efficient formalism for large scale ab initio molecular dynamics based on time-dependent density functional theory
• Generalized Extended Lagrangian Born-Oppenheimer Molecular Dynamics
• Multiple Time Step Integrators in Ab Initio Molecular Dynamics
• Discretization errors in molecular dynamics simulations with deterministic and stochastic thermostats
• Single molecule pulling with large time steps
• Stabilization of Ab Initio Molecular Dynamics Simulations at Large Time Steps