Theoretical chemistry has always attempted to perform accurate quantum mechanics (QM) calculations on larger and larger systems. While it is usually the case that these goals are difficult to reconcile, many attempts have been made to develop highly scalable methods that are also highly accurate. A number of fragment-based methods have been proposed to improve the computational efficiency of methods such as Møller-Plesset second order perturbation theory (MP2) and couple cluster theory (CC) for large systems. Most of these methods involve calculations of agglomerates of fragments, in one way or another, to include important many-body interactions. The fragment molecular orbital (FMO) method is a well established ab initio method capable of taking advantage of modern supercomputers, such as the Blue Gene Q system Mira at the Argonne Leadership Computing Facility (ALCF). The FMO method is based on dividing molecules into fragments and performing ab initio calculations on fragments, their dimers and, optionally, trimers. The decomposition of the system in this way makes it possible to perform QM calculations on large biological molecules. In contrast to many other fragment-based methods, the effect of the environment is rigorously accounted for by the electrostatic potential (ESP) due to all remaining fragments that are not explicitly included in a given n-mer calculation. The use of highly accurate levels of theory, such as coupled cluster (CC), in conjunction with FMO allows for the goal of scalable and highly accurate QM calculations to be reached. However, while CC methods have become the de facto standard of accurate quantum chemistry, the scaling of the method (O(n7)) with the number of electrons limits the size of the fragments. With the advent of supercomputers, Diffusion Monte Carlo (DMC-QMC) has proven to provide similar accuracy compared to CC, with a scaling of only O(n3) with the number of particles. This paper describes the merger of FMO with QMC methods to develop a combine FMO-QMC method for highly accurate and scalable all electron calculations. Two freely available programs (GAMESS and QMCPack) have been interfaced, and test calculations show a high level of accuracy when compared to full DMC and CC calculations. Scalability of both FMO and QMCPack has been demonstrated beyond 200K cores. Future applications of FMO-QMC include: drug-receptor binding, materials science, and the development of new force fields for biological applications.

Highly accurate fragment molecular orbital/quantum Monte Carlo method for large molecular systems

251st National Meeting (2016)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.