The Weak Form Is Stronger Than You Think - Data-Driven Physical Systems

Learn about groundbreaking advances in data-driven modeling through this technical talk by Professor David Bortz from the University of Colorado Boulder. Explore how weak form approaches are revolutionizing scientific machine learning, particularly in equation learning and parameter estimation. Discover the WSINDy and WENDy algorithms that demonstrate superior performance in handling noise-corrupted data, offering up to ten times more accuracy and significantly enhanced robustness compared to traditional methods. Examine practical applications across ordinary, partial, and stochastic differential equations, as well as coarse-graining and reduced order modeling. Gain insights into how these innovative approaches are transforming conventional model development, discretization, parameter inference, and model refinement processes, presented by an expert whose research has been supported by NSF, NIH, DOE, and DOD.