I recently learned about parallel tempering, or replica exchange MCMC sampling. Mu understanding is that multiple simulations are performed at different temperatures in parallel, then based on the Metropolis criterion configurations are exchanged between the simulations. By making high temperature configurations available to simulations at low temperatures, and vice versa, this allows for sampling both high and low energy configurations. This may be a valuable addition to the MC algorithm.
I recently learned about parallel tempering, or replica exchange MCMC sampling. Mu understanding is that multiple simulations are performed at different temperatures in parallel, then based on the Metropolis criterion configurations are exchanged between the simulations. By making high temperature configurations available to simulations at low temperatures, and vice versa, this allows for sampling both high and low energy configurations. This may be a valuable addition to the MC algorithm.