Abstract: Kinetic Monte Carlo (KMC) is a method for simulating atomistic processes of interest to physicists, chemists and material scientists. It is capable of simulating systems with time and length scales that are orders of magnitude larger than molecular dynamics(MD) techniques. It does so by approximating the evolution of a system (which are governed by Newton’s laws in MD) by explicit transitions between states in a discrete state space (the set of thermodynamically metastable states), where transition rates are given by transition state theory. While KMC can be applied to a wide range of systems, we examine its application to crystallographic ones where atoms occupy positions on a fixed lattice. In this case, it is possible to take advantage of local structure for performance gains, leading to computationally efficient KMC variants and implementations. In this talk, I describe several ways to do so, in addition to reviewing general algorithmic techniques used in simulations. The talk will be fairly self-contained and should be relevant to those interested in simulations and scientific computing.
Speaker: Kris Reyes