Molecular simulations of liquids, interfaces, materials, and general condensed phase environments can be complicated by the need to efficiently and accurately model multiple components and all of their corresponding physical processes. The key challenges to overcome for the successful modeling of these systems is the development of accurate representations of molecular interactions, inclusion of chemical reactivity and correct physics in the model, and general and efficient algorithms that permit sampling the required length and time scales at a reasonable computational cost. For the case of modeling chemical reactions, ab initio condensed phase simulations, which explicitly include electronic degrees of freedom and allow for the making and breaking of chemical bonds, are extremely useful, but their computational cost can be too prohibitive to obtain sufficient statistics. The multistate simulation framework provides one class of methods that can be used to combine the computational efficiency of popular molecular mechanics force fields and the chemical reactivity of electronic structure calculations. The work presented here will focus on recently developed manager/worker parallelization strategies to efficiently compute multistate simulations for systems containing many reactive species. Recent work on computational strategies to search for optimal zeolite structures for separation and chemical conversions will also be discussed along with the modeling of x-ray damage mechanisms in nanomaterials. Central to all of these modeling problems is the requirement for efficient computational strategies to access required time and length scales in order to properly address the challenging scientific questions being addressed today and those far-reaching questions being considered for tomorrow.

This work was supported by the U.S. Dept. of Energy under Contract DE-AC02-06CH11357.