A multi-lab research team led by Paul Kent of Oak Ridge National Laboratory is developing a computer application called QMCPACK to enable precise and reliable predictions of the fundamental properties of materials critical in energy research.
The application uses quantum Monte Carlo simulations to compute material properties down at the level of atoms and electrons. Simulation data will complement experimental data in discovering new materials to improve energy technology from batteries to generation methods.
QMCPACK is a subproject of the Department of Energy’s Exascale Computing Project. Kent said in an ECP podcast episode that QMCPACK had achieved proof of principle with a new application design that targets exascale GPU architectures. Researchers will use exascale computing to explore scientific problems once considered prohibitively complex.
The QMCPACK team is working on achieving a 50-fold performance increase and ensuring application robustness and readiness for scientific research on ORNL’s Frontier exascale supercomputer. — Scott Gibson