| Parallelization of kinetic Mont Carlo (kMC) is a difficult task due to the inherent asynchronicity of discrete event simulations. Recently, we have developed a scheme based on the introduction of null events that allows for a trivial synchronization of multiple processors and enables massive scale parallelism of kMC simulations. The method eliminates the need for cumbersome iterative synchronization techniques commonly found in asynchronous kMC algorithms. Our method has enabled KMC simulations of molecular systems up to one billion in number where scale effects are critical. Here we apply the algorithm to continuum diffusion problems and discrete Ising systems near the critical temperature. Our method could provide an avenue for parallel kMC calculations of irradiated microstructures and other highly heterogeneous problems. |
|