Hyperdynamics boost factor achievable with an ideal bias potential.

Abstract

Hyperdynamics is a powerful method to significantly extend the time scales amenable to molecular dynamics simulation of infrequent events. One outstanding challenge, however, is the development of the so-called bias potential required by the method. In this work, we design a bias potential using information about all minimum energy pathways (MEPs) out of the current state. While this approach is not suitable for use in an actual hyperdynamics simulation, because the pathways are generally not known in advance, it allows us to show that it is possible to come very close to the theoretical boost limit of hyperdynamics while maintaining high accuracy. We demonstrate this by applying this MEP-based hyperdynamics (MEP-HD) to metallic surface diffusion systems. In most cases, MEP-HD gives boost factors that are orders of magnitude larger than the best existing bias potential, indicating that further development of hyperdynamics bias potentials could have a significant payoff. Finally, we discuss potential practical uses of MEP-HD, including the possibility of developing MEP-HD into a true hyperdynamics.