http://arxiv.org/abs/2205.08663
Turbulence plays an integral role in astrophysical phenomena, including core-collapse supernovae (CCSN). Unfortunately, current simulations must resort to using subgrid models for turbulence treatment, as direct numerical simulations (DNS) are too expensive to run. However, subgrid models used in CCSN simulations lack accuracy compared to DNS results. Recently, Machine Learning (ML) has shown impressive prediction capability for turbulence closure. We have developed a physics-informed, deep convolutional neural network (CNN) to preserve the realizability condition of Reynolds stress that is necessary for accurate turbulent pressure prediction. The applicability of the ML model was tested for magnetohydrodynamic (MHD) turbulence subgrid modeling in both stationary and dynamic regimes. Our future goal is to utilize our ML methodology within the MHD CCSN framework to investigate the effects of accurately-modeled turbulence on the explosion rate of these events.
P. Karpov, C. Huang, I. Sitdikov, et. al.
Thu, 19 May 22
53/61
Comments: N/A