FastAI for complex systems/non-linear dynamical systems modeling

Deep Learning
1

Hello FastAI community,

I recently have been interested in the intersection of machine learning and complex systems/non-linear dynamical systems, especially in the case of chaotic dynamics.

Here is some relevant literature/news that piqued my interest:

  1. https://www.quantamagazine.org/machine-learnings-amazing-ability-to-predict-chaos-20180418/
  2. https://aip.scitation.org/doi/abs/10.1063/5.0016505
  3. https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.120.024102

Two questions:

  1. Does anyone have databunch/dataloader and model.py code for resevoir computing in FastAI they would be willing to share? I would also be interested in LSTM code as well. The cases I have seen for time-series forecasting with LSTM in FastAI involve a 1D time series. Here, I am more interested in multi-variate time series.

  2. Would anyone be interested in a study group to learn and implement techniques in FastAI at the intersection of machine learning and complex/non-linear dynamical systems? This could be an informative article to reproduce:
    https://npg.copernicus.org/articles/27/373/2020/