As part of the Aerospace Engineering Graduate Seminar Series, Karthik Duraisamy, Associate Professor of Aerospace Engineering at the University of Michigan, will join us for a talk on “Physics-Constrained Data-Driven Modeling of Complex Systems” on Monday, March 19th at 2:00pm in 3151 Learned Hall.
Karthik Duraisamy is an Associate Professor of Aerospace Engineering at the University of Michigan, Ann Arbor. He obtained a doctorate in Aerospace engineering and a Master’s degree in applied mathematics from the University of Maryland, College Park. Prior to his appointment in 2013 at the University of Michigan, he spent time at Stanford University and the University of Glasgow. At the University of Michigan, he is the Director of the Center for Data-driven Computational Physics and the Airforce Center of Excellence on Rocket Combustor Dynamics. His research interests are in data-driven and reduced order modeling, turbulence modeling and simulations and numerical methods for partial differential equations.
With the proliferation of high-resolution datasets and advances in computing and algorithms over the past decade, data science has risen as a discipline in its own right. Machine learning-driven models have attained spectacular success in commercial applications such as language translation, speech and face recognition, bioinformatics, and advertising. The natural question to ask then is: Can we bypass the traditional ways of intuition/hypothesis-driven model creation and instead use data to generate predictions of physical problems? The first part of this talk will review recent work in which inference and machine learning have been used to extract operator matrices, discover dynamical systems, and derive the solution of differential equations. The second part of the talk will discuss the challenges of extending these methods and data-driven modeling in general in the prediction of complex real-world problems. For instance, in turbulence modeling, regardless of the quantity of interest, there may exist several latent variables that might not be identifiable without a knowledge of the physics; we may not have enough data in all regimes of interest; and the data may be noisy and of variable quality.