Enhanced sampling simulations with the Colvars module

Giacomo Fiorin

Temple University

Molecular dynamics simulations of soft-matter and biological systems are often severely limited in the capability to achieve statistical convergence of computed thermodynamic properties. Among the standard approaches to overcome this limitation are biasing forces and time-dependent potentials applied on collective variables. Using these methods to obtain unbiased probability distributions is challenging, but generally highly reproducible. The collective variables (Colvars) module included in LAMMPS offers a common interface to multiple types of biasing methods and free energy estimators. Recent developments in Colvars allow the study of new types of systems, and the automation of enhanced sampling approaches to aid the development of computational models.