Learning Theory for Dynamical Systems

DOI10.1137/22M1516865zbMATH Open1526.37089arXiv2208.05349OpenAlexW4385658405MaRDI QIDQ6132792FDOQ6132792

Authors: T. Berry, Suddhasattwa Das 

Publication date: 17 August 2023

Published in: SIAM Journal on Applied Dynamical Systems (Search for Journal in Brave)

Abstract: The task of modelling and forecasting a dynamical system is one of the oldest problems, and it remains challenging. Broadly, this task has two subtasks - extracting the full dynamical information from a partial observation; and then explicitly learning the dynamics from this information. We present a mathematical framework in which the dynamical information is represented in the form of an embedding. The framework combines the two subtasks using the language of spaces, maps, and commutations. The framework also unifies two of the most common learning paradigms - delay-coordinates and reservoir computing. We use this framework as a platform for two other investigations of the reconstructed system - its dynamical stability; and the growth of error under iterations. We show that these questions are deeply tied to more fundamental properties of the underlying system - the behavior of matrix cocycles over the base dynamics, its non-uniform hyperbolic behavior, and its decay of correlations. Thus, our framework bridges the gap between universally observed behavior of dynamics modelling; and the spectral, differential and ergodic properties intrinsic to the dynamics.

Full work available at URL: https://arxiv.org/abs/2208.05349

Recommendations

• Learning latent dynamics for partially observed chaotic systems
• Learning strange attractors with reservoir systems
• Learning dynamics by reservoir computing (In Memory of Prof. Pavol Brunovský)
• Koopman operator, geometry, and learning of dynamical systems

zbMATH Keywords

mixingLyapunov exponentreservoir computingmatrix cocycledirect forecastdelay-coordinatesiterative forecast

Mathematics Subject Classification ID

Time series analysis of dynamical systems (37M10) Uniformly hyperbolic systems (expanding, Anosov, Axiom A, etc.) (37D20) Nonuniformly hyperbolic systems (Lyapunov exponents, Pesin theory, etc.) (37D25) Simulation of dynamical systems (37M05) Computational methods for ergodic theory (approximation of invariant measures, computation of Lyapunov exponents, entropy, etc.) (37M25) Dynamical systems in numerical analysis (37N30)

Cites Work

• Diffusion maps
• Ergodic theory of differentiable dynamical systems
• Lyapunov exponents, entropy and periodic orbits for diffeomorphisms
• Statistical properties of dynamical systems with some hyperbolicity
• Embedology
• Recurrence times and rates of mixing
• The metric entropy of diffeomorphisms. II: Relations between entropy, exponents and dimension
• Markov structures and decay of correlations for non-uniformly expanding dynamical systems
• FAMILIES OF INVARIANT MANIFOLDS CORRESPONDING TO NONZERO CHARACTERISTIC EXPONENTS
• Time-scale separation from diffusion-mapped delay coordinates
• Stable manifolds for nonautonomous equations without exponential dichotomy
• Time series analysis. Forecasting and control
• Nonlinear prediction of chaotic time series
• Products of Random Matrices
• SRB measures as zero-noise limits
• Random matrix products and measures on projective spaces
• The metric entropy of diffeomorphisms. I: Characterization of measures satisfying Pesin's entropy formula
• The Lyapunov exponents of generic volume-preserving and symplectic maps
• Numerical orbits of chaotic processes represent true orbits
• Title not available (Why is that?)
• Title not available (Why is that?)
• Coherent structures and isolated spectrum for Perron–Frobenius cocycles
• Lyapunov exponents of hyperbolic measures and hyperbolic periodic orbits
• Regularity of invariant graphs for forced systems
• Reservoir computing approaches to recurrent neural network training
• The topological invariance of Lyapunov exponents in embedded dynamics
• Lyapunov exponents and rates of mixing for one-dimensional maps
• Lyapunov exponents of linear cocycles. Continuity via large deviations
• Smoothness of invariant manifolds for nonautonomous equations
• Regularized local linear prediction of chaotic time series
• Model reduction with memory and the machine learning of dynamical systems
• A variational approach to the consistency of spectral clustering
• Nonlinear Regression
• Delay-coordinate maps and the spectra of Koopman operators
• Super convergence of ergodic averages for quasiperiodic orbits
• Koopman spectra in reproducing kernel Hilbert spaces
• Time-series learning of latent-space dynamics for reduced-order model closure
• Using machine learning to replicate chaotic attractors and calculate Lyapunov exponents from data
• Identification and prediction of low dimensional dynamics
• A concept of homeomorphic defect for defining mostly conjugate dynamical systems
• Quantitative Pesin theory for Anosov diffeomorphisms and flows
• Liapunov multipliers and decay of correlations in dynamical systems
• (Dis)continuity of Lyapunov exponents
• Reproducing kernel Hilbert space compactification of unitary evolution groups
• Is the largest Lyapunov exponent preserved in embedded dynamics?
• Analog forecasting with dynamics-adapted kernels
• Learning strange attractors with reservoir systems
• Embedding and approximation theorems for echo state networks
• Operator-theoretic framework for forecasting nonlinear time series with kernel analog techniques
• Optimal Construction of Koopman Eigenfunctions for Prediction and Control
• Equivalence of physical and SRB measures in random dynamical systems
• Deep learning of conjugate mappings
• Echo state networks trained by Tikhonov least squares are \(L^2(\mu)\) approximators of ergodic dynamical systems
• Machine learning for prediction with missing dynamics
• Stochastic nonlinear time series forecasting using time-delay reservoir computers: performance and universality
• Fading memory echo state networks are universal
• Approximation of Bernoulli measures for non-uniformly hyperbolic systems
• Erratum: “On explaining the surprising success of reservoir computing forecaster of chaos? The universal machine learning dynamical system with contrasts to VAR and DMD” [Chaos 31(1), 013108 (2021)]

Cited In (8)

• Learning strange attractors with reservoir systems
• Fluctuation-Dissipation Theorem and Models of Learning
• Riddled basins of the optimal states in learning dynamical systems
• A proposal on machine learning via dynamical systems
• Data-driven cold starting of good reservoirs
• Learning about dynamical systems via unfalsification of hypotheses
• Title not available (Why is that?)
• LEARNING TRAJECTORIES OF DYNAMICAL SYSTEMS

This page was built for publication: Learning Theory for Dynamical Systems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6132792)