From Deep to Physics-Informed Learning of Turbulence: Diagnostics
This work addresses the problem of automating turbulence simulations for researchers in fluid dynamics, but it is incremental as it builds on existing deep learning methods.
The paper tested three deep learning neural network schemes trained on direct numerical simulations of turbulence to accelerate hydrodynamic codes, finding that static schemes fail to reproduce small-scale intermittency and large-scale geometry, while dynamic schemes correct these issues.
We describe tests validating progress made toward acceleration and automation of hydrodynamic codes in the regime of developed turbulence by three Deep Learning (DL) Neural Network (NN) schemes trained on Direct Numerical Simulations of turbulence. Even the bare DL solutions, which do not take into account any physics of turbulence explicitly, are impressively good overall when it comes to qualitative description of important features of turbulence. However, the early tests have also uncovered some caveats of the DL approaches. We observe that the static DL scheme, implementing Convolutional GAN and trained on spatial snapshots of turbulence, fails to reproduce intermittency of turbulent fluctuations at small scales and details of the turbulence geometry at large scales. We show that the dynamic NN schemes, namely LAT-NET and Compressed Convolutional LSTM, trained on a temporal sequence of turbulence snapshots are capable to correct for the caveats of the static NN. We suggest a path forward towards improving reproducibility of the large-scale geometry of turbulence with NN.