ABSTRACT
A vast array of problems currently addressed by computer simulations, including biological systems, involve the analysis of properties on long time and length scales derived from simulations on relatively short time and length scales [Katsoulakis, Majda, and Vlachos 2003]. Although these techniques can provide a great deal of insight into the processes under study, traditional simulations of this type are limited in scope by their computational costs, which impose an upper limit on the time scale that can be studied (currently in the nanosecond range, for biological systems [Sastry et al. 2005]). This limitation has lead to the development of a wide variety of techniques attempting to provide longer time and length scale information than traditional (usually atomistic) simulations, many of which fall into the category of coarse-graining. In the broadest possible sense, the term “coarse-graining” (CG) can be used to refer to any simulation technique in which a simulated
system is simpli ed by clustering several subcomponents of it into one component, thus effectively reducing the computational complexity by removing both degrees of freedom and interactions from the system. The fundamental assumption behind such techniques is that by eliminating insigni cant degrees of freedom, one will be able to obtain physically correct data on the properties of a system over longer time scales than would otherwise be achievable [Schütte et al. 1999].
