Concurrent Multiscale Simulation at Finite Temperature: Coarse-Grained Molecular Dynamics

Abstract

With the advent of nanotechnology, predictive simulations of nanoscale systems have become in great demand. In some cases, nanoscale systems can be simulated directly at the level of atoms. The atomistic techniques used range from models based on a quantum mechanical treatment of the electronic bonds to those based on more empirical descriptions of the interatomic forces. In many cases, however, even nanoscale systems are too big for a purely atomistic approach, typically because the nanoscale device is coupled to its surroundings, and it is necessary to simulate the entire system comprising billions of atoms. A well-known example is the growth of nanoscale epitaxial quantum dots in which the size, shape and location of the dot is affected by the elastic strain developed in a large volume of the substrate as well as the local atomic bonding. The natural solution is to model the surroundings with a more coarse-grained (CG) description, suitable for the intrinsically longer length scale. The challenge then is to develop the computational methodology suitable for this kind of concurrent multiscale modeling, one in which the simulated length scale can be changed smoothly and seamlessly from one region of the simulation to another while maintaining the fidelity of the relevant mechanics, dynamics and thermodynamics.