Exploring Equilibrium Systems with Nonequilibrium Simulations

Thumbnail Image


Publication or External Link






Equilibrium sampling is at the core of computational thermodynamics, aiding our understanding of various phenomena in the natural sciences including phase coexistence, molecular solvation, and protein folding. Despite the widespread development of novel sampling strategies over the years, efficient simulation of large complex systems remains a challenge. While the majority of current methods such as simulated tempering, replica exchange, and Monte Carlo methods rely solely on the use of equilibrium techniques, recent results in statistical physics have uncovered the possibility to sample equilibrium states through nonequilibrium simulations.

In our first study we present a new replica exchange sampling strategy, "Replica Exchange with Nonequilibrium Switches," which uses nonequilibrium simulations to enhance equilibrium sampling. In our method, trial swap configurations between replicas are generated through nonequilibrium switching simulations which act to drive the replicas towards each other in phase space. By means of these switching simulations we can increase an effective overlap between replicas, enhancing the probability that these moves are accepted and ultimately leading to more effective sampling of the underlying energy landscape. Simulations on model systems reveal that our method can be beneficial in the case of low replica overlap, able to match the efficiency of traditional replica exchange while using fewer processors. We also demonstrate how our method can be applied for the calculation of solvation free energies.

In a second, separate study, we investigate the dynamics leading to the dissociation of Na-Cl in water. Here we employ tools of rare event sampling to deduce the role of the surrounding water molecules in promoting the dissociation of the ion pair. We first study the thermodynamic forces leading to dissociation, finding it to be driven energetically and opposed entropically. In further analysis of the system dynamics, we deduce a) the spatial extent over which solvent fluctuations influence dissociation, b) the role of sterics and electrostatics, and c) the importance of inertia in enhancing the reaction probability.