Residual-Guided Dictionary Learning for Spectrally Accurate Koopman Approximation

DOI10.48550/arXiv.2606.29083MaRDI QIDQ7270386arXivFDO

Authors Matthew J. Colbrook, George Coote

Publication date 27 June 2026

Published in arXiv

Tldr We train neural-network dictionaries to minimize Residual Dynamic Mode Decomposition (DMD) residuals, a measure of operator-level a posteriori error that tests whether computed eigenvalues and modes are genuine Koopman spectral objects, while penalizing condition number for numerical stability.

Full work available at URL https://doi.org/10.48550/arXiv.2606.29083

arXiv classification

cs.LGmath.DSmath.NA

Cites work

• Analysis of Fluid Flows via Spectral Properties of the Koopman Operator
• Another look at residual dynamic mode decomposition in the regime of fewer snapshots than dictionary size
• Beyond expectations: Residual Dynamic Mode Decomposition and Variance for Stochastic Dynamical Systems
• Deep learning for universal linear embeddings of nonlinear dynamics
• Dynamical systems of continuous spectra
• Extended dynamic mode decomposition with dictionary learning: a data-driven adaptive spectral decomposition of the Koopman operator
• Forecasting Sequential Data using Consistent Koopman Autoencoders
• Hamiltonian systems and transformations in Hilbert space.
• Koopman invariant subspaces and finite linear representations of nonlinear dynamical systems for control
• Learning Deep Neural Network Representations for Koopman Operators of Nonlinear Dynamical Systems
• Learning Koopman Invariant Subspaces for Dynamic Mode Decomposition
• Linearly recurrent autoencoder networks for learning dynamics
• Machine learning for fluid mechanics
• Modern Koopman theory for dynamical systems
• On Data-Driven Unbiased Predictors using the Koopman Operator
• Physics-Informed Probabilistic Learning of Linear Embeddings of Nonlinear Dynamics with Guaranteed Stability
• Residual Dynamic Mode Decomposition: Robust and verified Koopmanism
• Residual Pseudospectra Reveal a Physics-Informed Koopman Backbone for Tropical Pacific Variability and ENSO Prediction
• Rigorous data‐driven computation of spectral properties of Koopman operators for dynamical systems
• Spectral analysis of climate dynamics with operator-theoretic approaches
• Spectral analysis of nonlinear flows
• Spectral properties of dynamical systems, model reduction and decompositions

This page was built for publication: Residual-Guided Dictionary Learning for Spectrally Accurate Koopman Approximation

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7270386)