An implementation of Deep Kooman method idea from Bethany Lusch' paper using Keras/Tensorflow. At current stage, the code only considers discrete spectrum, but can be modified to extend to continuous specturm cases.
The following figure shows the two basins of attraction from unforced Duffing equation learned by visualizing the eigenfunction
Architecture.py
: contains model architectures.
Utils.py
: some utility functions.
train.py
: training script.
inference.py
: making inferences on data with learned model.