Reducing Data Resolution for Better Reconstruction: Super-Resolution of Navier-Stokes Flows

This talk from the DDPS (Data-Driven Physical Simulation) series features Dr. Kyongmin Yeo from IBM T.J. Watson Research Center discussing super-resolution techniques for reconstructing high-resolution information from low-resolution data in physical systems. Explore the theoretical analysis of super-resolution for noisy observations in 2D Navier-Stokes flows, where Dr. Yeo demonstrates how averaging observation data over a coarse grid reduces noise before employing a dynamic observer to reconstruct the flow field. Learn about the counter-intuitive finding that increasing the spatial averaging length scale (making resolution coarser) can actually reduce deviation from ground truth, with numerical experiments confirming the existence of a critical length scale below which coarser resolution improves reconstruction. Dr. Yeo, who received his Ph.D. in Applied Mathematics from Brown University and previously worked at Lawrence Berkeley National Lab, shares insights from his research spanning soft matter physics, atmospheric sciences, hybrid physics-statistics models, and deep learning for random dynamical systems. Presented on February 28th, 2025, as part of Lawrence Livermore National Laboratory's webinar series.