SINDy Experiments

pysindy experiments. These experiments are meant to be run via mitosis, but mitosis is not a requirement to run them. It partially annotated with call signatures, and I'm happy to add more as people point them out.

It's not yet on pypi, so install it with pip install git+https://github.com/Jacob-Stevens-Haas/gen-experiments or clone and install it locally.

Experiment Steps

There are three experiment steps made available:

• gen_experiments.data.gen_data()
• gen_experiments.odes.run()
• gen_experiments.gridsearch.run()

In addition, tentative PDE functionality is provided, though may be unstable:

• gen_experiments.data.gen_pde_data()
• gen_experiments.pdes.run()

Plotting and diagnostics

Perhaps of most significance are the SINDy diagnostic plotting. gen_experiments.plotting has a variety of functions for creating diagnostics of fitted SINDy models:

• gen_experiments.plotting.compare_coefficient_plots() and its cousin, gen_experiments.utils.unionize_coeff_matrices(), which is used to align coefficient matrices from models with different features.
• gen_experiments.utils.coeff_metrics() and gen_experiments.utils.pred_metrics()]
• gen_experiments.plotting.plot_training_data()
• gen_experiments.plotting.plot_test_trajectories()

Names

This package is distributed as pysindy-experiments, while some names still refer to gen_experiments. The latter is due to the origins in a PhD general exam.