Chapter 12 Generalized Ensemble Molecular Dynamics Methods Generalized ensemble molecular dynamics simulation methods can be used to improve the sampling of lower energy configurations. In this class of methods the following approaches have been widely used in simulations of macromolecular sys- tems (Hansmann and Okamoto 1999): multicanonical sampling (Berg and Neuhaus 1991, 1992), the broad histogram method (de Oliveira et al. 1996; de Oliveira 1998), Wang-Landau algorithm (Wang and Landau 2001a), Tsallis weights meth- ods (Tsallis 1988), and parallel tempering or replica exchange method (Penna 1995; Hukushima and Nemoto 1996; Geyer 1992). All of the above mentioned generalized-ensemble approaches have the same starting point, that is, the replacement of canonical Boltzmann-like weights at temperature T with non-Boltzmann weights, which allows the system is escaping from the local minimum states. In this chapter, we will discuss the choice of different weights for those methods that are most often used in molecular dynamics simulations according to Kamberaj (2019). 12.1 Multicanonical Sampling Method In the multicanonical ensemble (MUCA), the states are multiplied by a non- Boltzmann multicanonical factor, W mu (E), generating in this way a uniform probability distribution of the energy, P mu (E) (Berg and Neuhaus 1991, 1992): P mu (E) ∝ Ω(E)W mu (E) ≡ constant (12.1) Therefore, the multicanonical ensemble represents a free random walks in the potential energy space, since it is characterized by a flat distribution, and hence © Springer Nature Switzerland AG 2020 H. Kamberaj, Molecular Dynamics Simulations in Statistical Physics: Theory and Applications, Scientific Computation, https://doi.org/10.1007/978-3-030-35702-3_12 423