Fig. 3: Schematics of SSMLearn.
From: Data-driven modeling and prediction of non-linearizable dynamics via spectral submanifolds

First, he data-driven, SSM-based model reduction algorithm implemented in SSMLearn diagnoses and approximates the dominant SSM from the input data. Next, it constructs a data-driven reduced-order model as an extended normal form on the SSM. Finally, the algorithm uses this model to predict individual unforced trajectories and the response of the system under additional forcing.