Free Energy Methods and Software in Computer Simulations

Understanding phenomena such as solid-phase stability, nucleation, interfacial physics, activity coefficients, solubilities, phase diagrams, reaction paths and rates, and drug design often involves calculating free-energy differences between two or more states. To calculate these free energies directly from molecular dynamics (MD) or Monte Carlo (MC) simulations often requires the use of specialized algorithms that ensure all the states of importance are sampled and their contributions are combined optimally. In recent years, numerous new techniques for calculating free energies have been developed and refined for this purpose. Furthermore, major simulation packages are increasingly incorporating these free energy methodologies. They can thus be used in calculations where the intermolecular interactions are described by empirical force fields or where potentials are derived from ab initio methods, including modern machine-learning approaches such as neural-network potentials.