Coursera Spring Sale
40% Off Coursera Plus Annual!
Grab it
Learn how to develop energy-stable machine-learning models for non-Newtonian hydrodynamics that maintain molecular fidelity in this 43-minute conference presentation. Discover a novel ML-based approach that reduces high-dimensional multi-scale systems to reliable macro-scale models with low-dimensional variational structures while preserving canonical degeneracies and symmetry constraints. Explore how this methodology addresses the fundamental challenge of creating reliable closures for computational modeling of multi-scale systems, particularly when clear scale separation doesn't exist. Examine the application to non-Newtonian hydrodynamics of polymeric fluids as a practical example, understanding how this approach differs from conventional ML modeling by directly learning energy variational structures from micro-models through end-to-end processes. Master the concept of joint learning using micro-macro encoder functions that result in physically interpretable multi-scale models. Understand how this framework enables the use of pre-existing energy stable numerical schemes, ensuring computational efficiency and numerical robustness for real-world applications in electrochemical systems and beyond.
