Expanded Functionality\nand Portability for the Colvars\nLibrary

Colvars is an open-source C++ library that provides a\nmodular toolkit\nfor collective-variable-based molecular simulations. It allows practitioners\nto easily create and implement descriptors that best fit a process\nof interest and to apply a wide range of biasing algorithms in collective\nvariable space. This paper reviews several features and improvements\nto Colvars that were added since its original introduction. Special\nattention is given to contributions that significantly expanded the\ncapabilities of this software or its distribution with major MD simulation\npackages. Collective variables can now be optimized either manually\nor by machine-learning methods, and the space of descriptors can be\nexplored interactively using the graphical interface included in VMD.\nBeyond the spatial coordinates of individual molecules, Colvars can\nnow apply biasing forces to mesoscale structures and alchemical degrees\nof freedom and perform simulations guided by experimental data within\nensemble averages or probability distributions. It also features advanced\ncomputational schemes to boost the accuracy, robustness, and general\napplicability of simulation methods, including extended-system and\nmultiple-walker adaptive biasing force, boundary conditions for metadynamics,\nreplica exchange with biasing potentials, and adiabatic bias molecular\ndynamics. The library is made available directly within the main distributions\nof the academic software GROMACS, LAMMPS, NAMD, Tinker-HP, and VMD.\nThe robustness of the software and the reliability of the results\nare ensured through the use of continuous integration with a test\nsuite within the source repository.