TRodingerCurrOpinStructBio2005a

Groups free energy methods into three classes: FEP and other non-equilibrium work methods that use exponential averaging; TI; and replica exchange methods. Points out the challenge of dealing with rugged landscapes, and reviews a number of different approaches. "Many of the existing methods may perform better along specific pathways or for specific types of problems... The application domain of various methods remains poorly defined, which underlines the need for extensive comparisons using appropriate model systems." Summarizes some problems with comparisons --- everybody either works on toy models, or very different systems, so it is very difficult to compare different methods.