Latent Diffeomorphic Dynamic Mode Decomposition

TL;DR

Latent Diffeomorphic Dynamic Mode Decomposition (LDDMD) is a novel method that combines DMD interpretability with RNN predictive capabilities to analyze and forecast complex non-linear systems, demonstrated through streamflow prediction.

Contribution

LDDMD introduces a new data reduction approach that effectively models non-linear systems with memory, balancing interpretability and predictive power.

Findings

Successfully applied to streamflow prediction

Effectively models complex non-linear systems

Maintains simplicity for interpretability

Abstract

We present Latent Diffeomorphic Dynamic Mode Decomposition (LDDMD), a new data reduction approach for the analysis of non-linear systems that combines the interpretability of Dynamic Mode Decomposition (DMD) with the predictive power of Recurrent Neural Networks (RNNs). Notably, LDDMD maintains simplicity, which enhances interpretability, while effectively modeling and learning complex non-linear systems with memory, enabling accurate predictions. This is exemplified by its successful application in streamflow prediction.