Turbulence modeling in the age of machine learning

Turbulence is one of the most complex and chaotic phenomena in nature, and accurately modeling it remains one of the grand challenges of science and engineering. Traditional turbulence models are built upon decades of physical insight, distilled into simplified but practical representations of the underlying physics. With the rise of machine learning, new opportunities have emerged to improve turbulence modeling, but the problem is no less difficult: turbulence remains high-dimensional, nonlinear, and unpredictable. In this talk, I will present recent work on leveraging machine learning, with an emphasis on reinforcement learning and numerically consistent approaches, to develop turbulence models that respect both data and physics. I will argue that while machine learning provides powerful tools, success in this domain requires embedding scientific knowledge at the core of model design. Only by combining physical understanding with modern AI can we build models that are both accurate and reliable for turbulent flows.

Bio

Professor Bae's research focuses on the physical understanding and modeling of Jane Bae is an Assistant Professor of Aerospace at the Graduate Aerospace Laboratories at Caltech. She received her Ph.D. in Computational and Mathematical Engineering from Stanford University in 2018. She was a postdoctoral fellow in the Graduate Aerospace Laboratories at Caltech and the Institute for Applied Computational Science at Harvard University before joining the Caltech faculty. Her main research focuses on computational fluid mechanics, in particular on modeling and control of unsteady and nonequilibrium wall-bounded turbulence.