Adaptive Computing and Multi-Fidelity Learning - IPAM at UCLA

Explore adaptive computing and multi-fidelity learning in this 49-minute conference talk presented by Juliane Mueller from the National Renewable Energy Laboratory at IPAM's workshop on Complex Scientific Workflows at Extreme Computational Scales. Delve into ongoing research that combines low- and high-fidelity simulation models for efficient optimization and uncertainty quantification. Discover optimization formulations that consider available compute resources as constraints, maximizing information gain while balancing fidelity levels. Examine application examples benefiting from this approach, particularly when addressing challenges in scaling up experiments and simulations. Gain insights into cutting-edge techniques for managing complex scientific workflows and computational resources in extreme-scale computing environments.