colvars

Colvars are collective variables, which are functions of the atomic coordinates that quantify the progress of a molecular system along a chosen process or reaction coordinate. They are used to describe and control the evolution of systems in molecular dynamics simulations, especially to enable enhanced sampling by applying biasing potentials or restraints on these variables.

A colvar can be a simple geometric measure, such as a distance between two atoms, an bond angle, or a dihedral angle, and it can also be a more complex descriptor derived from groups of atoms. Colvars can be combined into multi-dimensional or path-based descriptors, including coordination numbers, RMSD to a reference structure, or progress along a predefined path in configuration space. In practice, colvars are defined in a specification that lists the variables, their mathematical form, any transformations, and the biasing terms to be applied. The simulation software then computes the current value of each colvar and integrates the influence of the associated bias into the system dynamics.

Biasing methods commonly used with colvars include harmonic restraints for umbrella sampling, metadynamics for exploring free-energy landscapes, and time-dependent or adaptive biases to steer transitions. Colvar-based biasing can be combined with steered molecular dynamics, bias-exchange schemes, or path-based sampling techniques to study conformational changes such as folding, ligand binding, or diffusion along a reaction coordinate.

Colvars are implemented as a modular framework in several molecular dynamics packages and are central to many enhanced-sampling workflows. They provide a flexible, interpretable way to define the degrees of freedom that control a process and to connect physical coordinates to biased simulations and subsequent analysis.