Architecting The Future Of Scientific Discovery: Unleashing Frontier Artificial Intelligence And Advanced Simulation For The Doe Genesis Mission
IRB 2207
Initiated by the U.S. Department of Energy, the Genesis Mission is a national framework designed to accelerate scientific discovery and technological innovation by deeply integrating artificial intelligence with exascale supercomputing and advanced experimental infrastructure. In this talk, I will give an overview of the works done at OSU that led to a funded Genesis mission phase I project. I will first talk about our recent work on deploying 3D Gaussian Splatting and Stable Diffusion to construct scientific ensemble simulation surrogates. Then I will describe our effort on employing Implicit Neural Networks for compact scientific data representations, query, and bi-directional reasoning. By bridging the gap between frontier AI and complex physical simulation, these foundational advancements ultimately pave the way for next-generation scientific discovery.