Code to accompany the paper Discovery of Physics from Data: Universal Laws and Discrepancies.

The figures from the paper were all generated using the Jupyter notebook generate_figures.ipynb. Likewise, the figures from the supplementary material were generated with supplementary_material.ipynb.

To run the Jupyter notebooks you will need the following Python packages. We give the versions of each package used when generating the figures for the paper.

Matplotlib (3.1.2)
Numpy (1.18.1)
Pandas (0.25.3)
Python (3.7.5)
Scikit-learn (0.22.1)
Scipy (1.4.1)
Seaborn (0.9.0)

You can install these packages with pip via

pip install -r requirements.txt

Some examples in the supplementary material additionally use the following package.

PySINDy (0.12.0)

PySINDy can be installed with

pip install pysindy

The paper can be found here. A preprint is also available on the arXiv.

@article{desilva2020discovery,
 author = {de Silva, Brian M. and Higdon, David M. and Brunton, Steven L. and Kutz, J. Nathan},
 doi = {10.3389/frai.2020.00025},
 issn = {2624-8212},
 journal = {Frontiers in Artificial Intelligence},
 pages = {25},
 title = {Discovery of Physics From Data: Universal Laws and Discrepancies},
 url = {https://www.frontiersin.org/article/10.3389/frai.2020.00025},
 volume = {3},
 year = {2020}
}