This is the code repo for the paper: "Discovering New Interpretable Conservation Laws as Sparse Invariants". We propose Sparse Invariant Detector, SID, which is able to discover all conserved quantities, given the knowledge of differential equations and the basis functions. Each example is made self-consistent in jupyter notebooks.

If you want to try your own differential equations $\dot{x}=f(x)$, the only two things you need to modify to toy.ipynb are (1) specify $f(x)$; (2) specify basis functions, by creating basis.txt. If you're dealing with polynomial bases, "create_ploy_basis_file" can help you create the file. If you're working with other functions, you will need to create your own basis.txt.