Recording: www.youtube.com… Title: Cognitive Simulation: Combining simulation and experiment with artificial intelligence Speaker: Dr. Brian K. Spears, Director, Cognitive Simulation Initiative, Lawrence Livermore National Laboratory Abstract:
Large-scale scientific endeavors often focus on improving predictive capabilities by challenging theory-driven simulations with experimental data. Yet, both simulation and experiment have become overwhelmingly rich, with a complex of observables including scalars, vector-valued data, and various images. At Lawrence Livermore National Laboratory (LLNL), we are using modern artificial intelligence (AI) technologies to combine predictive simulation models with rich experimental data. We call this set of methods cognitive simulation. We will describe a strategic LLNL research effort aimed at using recent advances in deep learning, computational workflows, and computer architectures to develop improved predictive models. We will present work from a wide range of applications, including using inertial confinement fusion research and experiments at the world’s largest laser, the National Ignition Facility (NIF), as a testbed. We will describe advances in machine learning architectures and methods necessary to handle strong nonlinearities and multimodal data in ICF science and other applications. We will also cover state-of-the-art tools that we developed to steer physics simulation, uncertainty quantification, and model training at enormous scale using the GPU-rich architecture of the Sierra supercomputer. We finally describe our ongoing efforts to accelerate deep learning for high-performance computing using new hardware accelerator technologies. Bio:
Brian Spears is a physicist at Lawrence Livermore National Laboratory. He is a principal architect of cognitive simulation methods – artificial intelligence (AI) methods that combine high-performance simulation and precision experiments with the goal of improving model predictions. He applies cognitive simulation techniques to stockpile stewardship missions with emphasis on quantifying uncertainty in inertial confinement fusion (ICF) experiments and advancing certification methods for the US nuclear weapons stockpile. He also uses cognitive simulation research applications to guide development of next-generation supercomputers. His responsibilities include setting vision for AI development and deployment at the Laboratory while driving LLNL leadership in AI for science. He has designed ICF experiments for 15 years, including the first cryogenic layered experiments at the National Ignition Facility. He developed new ICF ignition metrics using the first large-scale ensembles of 2D ICF simulations. He received the LLNL 2019 Mid-Career Recognition for career achievements in research and the Hyperion HPC Innovation Award. Brian completed his PhD at the University of California, Berkeley where he studied topological methods for high-dimensional dynamical systems. He also holds a BS in mechanical engineering and a BA in liberal arts from the University of Texas at Austin. When not doing science, he can be found racing his bike or chauffeuring his two daughters to swim and gymnastics. * This seminar is part of the ME Graduate Seminar Series (ME520). All talks are free and open to public. ME students and alumni are encouraged to attend. |