HACCing the Universe
Abstract:
Cosmology is the science of mining the sky to understand the make up and evolution of the universe. Modern sky surveys collect so much information that solving this grandest of inverse problems is no longer limited by statistics, but by our ability to control systematic errors, including the ability to model cosmic structure formation. In this talk I will describe HACC (Hardware/Hybrid Accelerated Cosmology Code), an extreme-scale code designed for attacking these problems, the new algorithms, methods and simulation techniques developed for it, and the associated ML/AI approaches we are working on for applying forward modeling based methods in cosmology.
Biography:
Salman Habib is the Director of Argonne’s Computational Science (CPS) Division. A theoretical and computational physicist, his scientific interests cover a broad span of research, ranging from quantum field theory and quantum information to the formation of cosmological structure. For the last twenty years, he has been very interested in combining large-scale simulation techniques with advanced statistical methods to solve scientific inference problems.