Data-Driven Closure Modeling Using Derivative-free Kalman Methods

Explore data-driven closure modeling for complex dynamical systems in this hour-long talk by Dr. Jinlong Wu from the University of Wisconsin–Madison. Delve into the use of derivative-free Kalman methods for learning closure models from indirect and limited data. Discover examples of sparse identification of dynamical systems and the learning of stochastic closures. Gain insights into improving predictions for turbulence, cloud dynamics, and other complex systems where resolving all degrees of freedom remains challenging. Learn about the potential of machine learning techniques in advancing closure models beyond traditional approaches, addressing limitations in representation power and empirical calibration processes.