MiniGAP: A Proxy App for ML Prediction of Molecular Properties
DescriptionOne could argue that AI/ML has disrupted chemistry and materials science, as well as other disciplines, in part, due to the synergy of accessible databases, accelerated computing resources, and community-supported and open codes. Three common elements found in AI/ML in chemistry research are data collection, fingerprint selection, and regression models. This presentation will discuss these three elements in the context of HPC environments, particularly exascale technologies. It will also include the development of a proxy application for predicting chemical properties and developing machine learning interatomic potentials that reproduce out-of-equilibrium properties with accuracy within experimental errors. Although training for these properties and potentials requires low-scale simulations, they can be used later to infer interactions between millions of atoms in petascale computers. Acknowledgement: This research used resources of the Argonne Leadership Computing Facility, which is a DOE Office of Science User Facility supported under Contract DE-AC02-06CH11357. This research was supported by the Exascale Computing Project (17-SC-20-SC), a collaborative effort of the U.S. Department of Energy Office of Science and the National Nuclear Security Administration.
TimeMonday, June 2618:00 - 18:30 CEST
Computer Science, Machine Learning, and Applied Mathematics