Multicanonical Monte Carlo simulations

Multicanonical simulations are a Markov Chain Monte Carlo method for the simulation of physical models, which yields canonical expectation values over a range of temperatures. Instead of the temperature also other parameters of physical interest may be used, leading to variants of the method. The weight factors of the multicanonical calculation can be calculated recursively and in this paper we shall address in some details the performance of such a recursion. Subsequently, two recent applications are selected and briefly discussed: multi-overlap spin glass simulations and the helix formation of poly-alanine.