Coarse grain to atomistic mapping algorithm : a tool for multiscale simulations

A key component to the success of multiscale modeling approaches is the ability to switch between molecular representations involving different levels of detail. In particular, the transition from a coarse to a fine representation is challenging because it requires the creation of information. To address this challenge, an algorithm is presented to fill in the detail when a coarse grain representation of a molecular system is replaced by an atomistic representation. The algorithm consists of minimization on the Lie group SO(3) for every coarse grain site, using the components of the atomistic force field that operate between atoms belonging to different coarse grain sites. This method is maximally efficient because the optimization is done at the coarse level using frozen atomistic library structures corresponding to each kind of coarse grain unit. The algorithm can be applied to any system since its input requirements are simply the force fields and molecular structures of both levels. The role of the algorithm in dual-resolution multiscale simulation methodology is discussed, and the efficacy of the algorithm is demonstrated through its implementation to liquid dodecane.