Efficient Algorithms for Many-Body Hard Particle Molecular Dynamics

Many-body simulations are very CPU-time consuming, making the problem of having efficient algorithms specially relevant. In this paper we propose a strategy--for the simulation of hard particle systems--that is efficient, memory saving, and easy to understand and to program. The time intervals by which the simulation proceeds are the increments between collisions (events), and these are dictated by the system itself. Hence these are event-driven simulations. Our strategy is devised to (a) minimize the number of coordinate updates per event, (b) predict new events only between nearby particles, and (c) efficiently manage the events predicted during the simulation. Empirical results are given to show the performance of our strategy in different computers as well as to compare with other approaches. It is seen that our proposed algorithm is efficient for a wide density range. We also include an analysis of the performance of the strategy proposed.