Turbulent Flow data as PyTorch tensors for ML: Kolmogorov Flow at Re=222, and Kelvin-Helmholtz instability

This dataset contains three files, listed below. The Kolmogorov flow is generated using a spectral solver, available at: https://github.com/google/jax-cfd. The Kelvin-Helmholtz Instability is generated using an in-house code.<br><br><b>Case 1: Kolmogorov Flow</b><br><code>nu_0p0045_2500_8f_uv_128.pt</code> -- a PyTorch tensor containing 2500 eight-frame videos of a 2D Re=222 forced turbulent flow (Kolmogorov flow), with only velocity vectors provided. The first 2000 samples are used as training data, the next 450 are used for validation and the final 50 are used to test the model, after training.<br><br><b>Case 2: Kelvin Helmholtz Instability</b><i>Training and Validation</i><code>:</code><code>kh_8f_72_208_r34568.pt</code> -- a PyTorch tensor containing 1000 eight-frame videos of a Kelvin-Helmholtz instability flow from 5 realisations of the flow (i.e. initialised from different random seeds). Each two hundred videos are from one simulation - the last two hundred may be used as a validation set.<br><i>Testing:</i><code> </code><code>kh_8f_72_208_r9.pt</code> -- a PyTorch tensor containing 200 eight-frame videos of a Kelvin-Helmholtz instability flow from a realisation of the flow different to the above. This is used as the test set for a model trained on <code>kh_8f_72_208_r34568.pt</code>.<br><br><br>