There is a need for high performance parallel software for first-principles computer modeling of large molecules to address a wide range of problems across biology, chemistry and materials science. This Phase II proposal will address this opportunity by synergistically combining the latest developments in algorithms for quantum chemical calculations on large molecules with parallel computing technology. Parallel algorithrns will be developed for large-scale density functional theory calculations based on linear scaling algorithms. The components needed to evaluate energies, analytical gradients (forces), and analytical hessians (force constants) will be addressed, as needed for real-world end-user applications. Excited states will also be available via time-dependent density functional theory. The resulting code will be scalable both with respect to number of processors, through efficient parallelization, and with respect to the size of the molecule, through the adoption of linear scaling methods wherever possible. In Phase III, this work will be distributed commercially as a release of the Q-Chem program. Due to the use of the well-established Messaging Passing Interface (MPI) standard in the Phases I and II development work, this package will be capable of running on both current and future parallel computers from major vendors such as SGI, IBM, HP, Sun and Compaq/DEC.
The research in the Phase II will combine the linear scaling methods developed recently in quantum chemistry field and the parallel computing technologies. It will bring parallel computing capabilities of quantum chemistry software to a whole new level and will enable scientists in academia and industry to run their calculations much faster and to study much larger systems than currently possible.
