This proposal seeks to improve the performance of density functional theory (DFT) as implemented in the commercial quantum chemistry software package Q-Chem. DFT strikes the right balance between accuracy and computational cost, and is used to model molecular processes in a wide variety of disciplines, including biology, chemistry, and materials science. In the work of our Phase I SBIR, we developed a novel method, called IncDFT, short for incremental DFT, that reduced the computational cost of the most time consuming step in DFT energy calculations by a factor of two. IncDFT, takes advantage of the iterative nature of DFT calculations by reusing, rather than recomputing, computationally expensive quantities that differ insignificantly between iteration cycles. In this Phase II proposal, IncDFT will be developed for all major aspects of DFT calculations in Q-Chem, including computation of the energy with gradient corrected functionals, the evaluation of critical points on the potential energy surface, the computation of the analytical Hessian, and calculation of excited electronic states. Care will also be taken to maintain Q-Chem's excellent parallel efficiency. The successful completion of this project will allow Q-Chem to provide its end-users with a software package that will significantly reduce the execution time for calculations employing DFT, thereby greatly increasing their productivity accordingly.
DFT is the preferred computational model in many areas of chemical and biological research due to its accuracy. Still, its application can be time-consuming. The research in the Phase II proposal will produce a faster version of Q-Chem for DFT calculations, which will enable users to finish their calculations in less time, or run calculations on larger molecular systems with the same cost. Thus, their productivity will be significantly enhanced.
