A list-based algorithm for evaluation of large deviation functions

As the analog of the free energy for dynamical trajectories, the large deviation function plays a central role in the statistical mechanics of systems far from equilibrium. Here, we identify numerical issues that can arise when the model of interest evolves according to a continuous-time dynamics. This analysis motivates the introduction of an algorithm in which a list of previously visited states is used to resample the distribution of interest. We discuss the convergence properties of our algorithm in detail and demonstrate its application to the single-site zero-range process and the many-site totally asymmetric exclusion process.