Speaker
Description
Super-resolution tools have been originally invented for image super-resolution but are also increasingly used for improving scientific simulations or data-storage. Examples range from cosmology to urban prediction. One particular network framework, physics-informed enhanced super-resolution generative adversarial networks (PIESRGANs), has been shown to be a powerful tool for subfilter modeling. It is the basis for JuLES (JUelich Large-Eddy Simulation) which has been recently developed to generate AI super-resolution models at scale and accelerate large-scale simulations significantly. This talk highlights important modeling aspects employing PIESRGAN with applications to HPC simulations. The examples range from simple homogeneous isotropic turbulence to finite-rate-chemistry premixed flame kernels. A priori and a posteriori results are presented.
