The paper “Dynamically Significant Latent Representations of Dynamical Programs“, co-authored by Imran Nasim (IBM, and Visiting Lecturer at Surrey) and Michael Henderson (IBM, New York), has been revealed within the MPDI journal “Arithmetic” (open entry hyperlink right here). The paper presents a data-driven hybrid modeling method to sort out the issue of discount by combining numerically derived representations and latent representations obtained from an autoencoder. The latent representations are validated and they’re proven to be dynamically interpretable. Moreover, the paper probes the topological preservation of the latent illustration with respect to the uncooked dynamical knowledge utilizing strategies from persistent homology. Lastly, the paper reveals that the framework is generalizable, having been efficiently utilized to each integrable and non-integrable programs that seize a wealthy and numerous array of resolution sorts. The methodology doesn’t require any prior dynamical data of the system and can be utilized to find the intrinsic dynamical habits in a purely data-driven manner. The picture under reveals Determine 7 from the paper.