Keywords: ODE Discovery, Symbolic Regression
TL;DR: A new generalized variational formulation that can discover most forms of ODEs.
Abstract: Understanding the mathematical connections between variables in a physical system, such as Ordinary Differential Equations (ODEs) is an essential part of the scientific method. This is where symbolic regression plays a key role in looking for closed-form functions given a dataset. We extend the results of the original Discovering Closed-Form ODEs from Observed Trajectories (D-CODE) by considering a generalized variational formulation that can work with most forms of ODEs. We conclude the paper with numerical results and applications.