Sampling Enhancement and Free Energy Prediction by the Flying Gaussian Method

We present a novel sampling enhancement and free energy prediction technique based on parallel simulation of the studied system with a shared bias potential. This history-independent bias potential is defined using selected degrees of freedom (collective variables). Each parallel walker of the system bears a single Gaussian shaped bias potential centered in current values of collective variables. Sampling enhancement is achieved by concentration of multiple walkers in certain free energy minimum. The method was successfully demonstrated on selected molecular systems, and presumed advantages over methods based on a history-dependent bias potential are discussed.