Sampling Rare Events in Dissipative Nonequilibrium Systems
Stephen Williams (Research School of Chemistry, Australian National University)

We consider methods of sampling rare events in driven nonequilibrium systems. To do this we appeal to nonequilibrium statistical mechanics, allowing us to perform importance sampling on the nonequilibrium distribution function. In contrast to existing methods, the one introduced here is suitable for systems governed by inertial equations of motion.
Unlike non-inertial systems, here the dominant contribution for rare barrier crossings comes from a very small subset of configuration space, which was occupied at an earlier time. Existing stochastic methods for nonequilibrium dynamics do not sample this subset with adequate bias. To address this we make use of recent advances in non-linear response theory, which allow for the distribution function of a system (at a later time) that is initially in some specified nonequilibrium state to be expressed as a relatively simple expression involving the initial state and a path integral. The effectiveness of the method will be demonstrated using results from a simple computational model, and extended to transport across a liquid-liquid interface.