Adam Subel - Stable a posteriori LES of 2D turbulence using convolutional neural networks: Backscattering analysis and generalization to higher Re via transfer learning

Using 2D decaying turbulence as a testbed, this work, co-authored by Adam Subel, explores the importance of accurately capturing backscattering in generating stable online LES. Additionally, it shows that transfer learning can enable generalization to flows with a higher Reynolds number, even if an increase in resolution is needed.