Linearly recurrent autoencoder networks for learning dynamics
From MaRDI portal
data-driven analysishigh-dimensional systemsKoopman operatorneural networksnonlinear systemsreduced-order modeling
Time series analysis of dynamical systems (37M10) Computational methods for ergodic theory (approximation of invariant measures, computation of Lyapunov exponents, entropy, etc.) (37M25) Linear composition operators (47B33) Numerical problems in dynamical systems (65P99) Neural networks for/in biological studies, artificial life and related topics (92B20)
Abstract: This paper describes a method for learning low-dimensional approximations of nonlinear dynamical systems, based on neural-network approximations of the underlying Koopman operator. Extended Dynamic Mode Decomposition (EDMD) provides a useful data-driven approximation of the Koopman operator for analyzing dynamical systems. This paper addresses a fundamental problem associated with EDMD: a trade-off between representational capacity of the dictionary and over-fitting due to insufficient data. A new neural network architecture combining an autoencoder with linear recurrent dynamics in the encoded state is used to learn a low-dimensional and highly informative Koopman-invariant subspace of observables. A method is also presented for balanced model reduction of over-specified EDMD systems in feature space. Nonlinear reconstruction using partially linear multi-kernel regression aims to improve reconstruction accuracy from the low-dimensional state when the data has complex but intrinsically low-dimensional structure. The techniques demonstrate the ability to identify Koopman eigenfunctions of the unforced Duffing equation, create accurate low-dimensional models of an unstable cylinder wake flow, and make short-time predictions of the chaotic Kuramoto-Sivashinsky equation.
Recommendations
- Physics-Informed Probabilistic Learning of Linear Embeddings of Nonlinear Dynamics with Guaranteed Stability
- Generalizing dynamic mode decomposition: balancing accuracy and expressiveness in Koopman approximations
- Learning latent dynamics for partially observed chaotic systems
- Learning proper orthogonal decomposition of complex dynamics using heavy-ball neural ODEs
- Koopman Operator Inspired Nonlinear System Identification
Cites work
- A data-driven approximation of the koopman operator: extending dynamic mode decomposition
- A kernel-based method for data-driven Koopman spectral analysis
- Applied Koopmanism
- Balanced realization and model reduction for unstable systems
- Deep learning
- Discovering governing equations from data by sparse identification of nonlinear dynamical systems
- Dynamic mode decomposition of numerical and experimental data
- Extended dynamic mode decomposition with dictionary learning: A data-driven adaptive spectral decomposition of the Koopman operator
- scientific article; zbMATH DE number 1418964 (Why is no real title available?)
- Kernel based partially linear models and nonlinear identification
- Model Reduction for Flow Analysis and Control
- MODEL REDUCTION FOR FLUIDS, USING BALANCED PROPER ORTHOGONAL DECOMPOSITION
- Optimal mode decomposition for unsteady flows
- Partially-Linear Least-Squares Regularized Regression for System Identification
- Principal component analysis in linear systems: Controllability, observability, and model reduction
- Spectral analysis of nonlinear flows
- Spectral properties of dynamical systems, model reduction and decompositions
- Stability and transition in shear flows
- Turbulence, coherent structures, dynamical systems and symmetry.
- Why does deep and cheap learning work so well?
Cited in
(94)- Deep learning for model order reduction of multibody systems to minimal coordinates
- Inverse multiobjective optimization: inferring decision criteria from data
- Data-driven operator theoretic methods for phase space learning and analysis
- Lift \& learn: physics-informed machine learning for large-scale nonlinear dynamical systems
- A long short-term memory embedding for hybrid uplifted reduced order models
- Machine learning for prediction with missing dynamics
- A data-driven, physics-informed framework for forecasting the spatiotemporal evolution of chaotic dynamics with nonlinearities modeled as exogenous forcings
- State estimation with limited sensors -- a deep learning based approach
- tgEDMD: approximation of the Kolmogorov operator in tensor train format
- Assessment of end-to-end and sequential data-driven learning for non-intrusive modeling of fluid flows
- Dynamics reconstruction and classification via Koopman features
- Recovering missing CFD data for high-order discretizations using deep neural networks and dynamics learning
- Model reduction of dynamical systems on nonlinear manifolds using deep convolutional autoencoders
- Variational approach for learning Markov processes from time series data
- A hybrid neural network model for the dynamics of the Kuramoto-Sivashinsky equation
- Learning nonlinear state-space models using autoencoders
- Parametric dynamic mode decomposition for reduced order modeling
- Model reduction for the material point method via an implicit neural representation of the deformation map
- Deterministic learning of completely resonant nonlinear wave systems with Dirichlet boundary conditions
- Data-driven model predictive control using interpolated Koopman generators
- Sparsity-promoting algorithms for the discovery of informative Koopman-invariant subspaces
- On Koopman mode decomposition and tensor component analysis
- scientific article; zbMATH DE number 7370558 (Why is no real title available?)
- A tailored convolutional neural network for nonlinear manifold learning of computational physics data using unstructured spatial discretizations
- Deep learning models for global coordinate transformations that linearise PDEs
- Learning brain dynamics with coupled low-dimensional nonlinear oscillators and deep recurrent networks
- Koopman analysis of quantum systems*
- Modern Koopman theory for dynamical systems
- Generative stochastic modeling of strongly nonlinear flows with non-Gaussian statistics
- Time-delay observables for Koopman: theory and applications
- Physics-Informed Probabilistic Learning of Linear Embeddings of Nonlinear Dynamics with Guaranteed Stability
- Leveraging reduced-order models for state estimation using deep learning
- The structure of reconstructed flows in latent spaces
- Learning latent dynamics for partially observed chaotic systems
- Shallow neural networks for fluid flow reconstruction with limited sensors
- Kernel methods for detecting coherent structures in dynamical data
- Scalable extended dynamic mode decomposition using random kernel approximation
- Data-driven identification of parametric partial differential equations
- Adaptive path-integral autoencoder: representation learning and planning for dynamical systems
- Nonintrusive Reduced Order Modelling of Convective Boussinesq Flows
- Modelling the dynamics of nonlinear partial differential equations using neural networks
- Koopman analysis of nonlinear systems with a neural network representation
- Transformers for modeling physical systems
- Predicting turbulent dynamics with the convolutional autoencoder echo state network
- Hybrid analysis and modeling, eclecticism, and multifidelity computing toward digital twin revolution
- Deep Koopman model predictive control for enhancing transient stability in power grids
- A study on data-driven identification and representation of nonlinear dynamical systems with a physics-integrated deep learning approach: Koopman operators and nonlinear normal modes
- Generalizing dynamic mode decomposition: balancing accuracy and expressiveness in Koopman approximations
- Parsimony as the ultimate regularizer for physics-informed machine learning
- Koopman operator learning using invertible neural networks
- Data-driven feedback stabilisation of nonlinear systems: Koopman-based model predictive control
- Neural dynamic mode decomposition for end-to-end modeling of nonlinear dynamics
- Rigorous data‐driven computation of spectral properties of Koopman operators for dynamical systems
- Active-learning-driven surrogate modeling for efficient simulation of parametric nonlinear systems
- Non-intrusive model combination for learning dynamical systems
- \textit{dynoNet}: a neural network architecture for learning dynamical systems
- Deep Koopman learning of nonlinear time-varying systems
- Machine learning enhanced Hankel dynamic-mode decomposition
- Enhancing predictive capabilities in data-driven dynamical modeling with automatic differentiation: Koopman and neural ODE approaches
- Generalized quadratic embeddings for nonlinear dynamics using deep learning
- Propagating uncertainty through system dynamics in reproducing kernel Hilbert space
- Hausdorff metric based training of kernels to learn attractors with application to 133 chaotic dynamical systems
- Data-driven identification of dynamical models using adaptive parameter sets
- Koopman neural operator as a mesh-free solver of non-linear partial differential equations
- A dynamical system-based framework for dimension reduction
- Koopman dynamic-oriented deep learning for invariant subspace identification and full-state prediction of complex systems
- Phase autoencoder for limit-cycle oscillators
- CGNSDE: conditional Gaussian neural stochastic differential equation for modeling complex systems and data assimilation
- Deep learning in computational mechanics: a review
- Approximation of translation invariant Koopman operators for coupled non-linear systems
- \(\Phi\)-DVAE: physics-informed dynamical variational autoencoders for unstructured data assimilation
- Generalization error guaranteed auto-encoder-based nonlinear model reduction for operator learning
- Deep networks for system identification: a survey
- Robust control design and analysis based on lifting linearization of nonlinear systems under uncertain initial conditions
- Using signatures and Koopman operator to learn non-linear dynamics
- Modeling nonlinear control systems via Koopman Control Family: universal forms and subspace invariance proximity
- Model predictive control of fluid-structure interaction via Koopman-based reduced-order model
- Intrinsic dimensionality of Fermi-Pasta-Ulam-Tsingou high-dimensional trajectories through manifold learning: a linear approach
- A density-based manifold learning framework for reconstructing high-dimensional dynamical systems with outliers
- Data-driven sensor fault detection and isolation of nonlinear systems: deep neural-network Koopman operator
- Solving time-dependent parametric PDEs by multiclass classification-based reduced order model
- On the relationship between Koopman operator approximations and neural ordinary differential equations for data-driven time-evolution predictions
- Learning parametric Koopman decompositions for prediction and control
- Tracking and forecasting oscillatory data streams using Koopman autoencoders and Kalman filtering
- Systematic construction of continuous-time neural networks for linear dynamical systems
- Heterogeneous mixtures of dictionary functions to approximate subspace invariance in Koopman operators: why deep Koopman operators works
- Parameterized neural ordinary differential equations: applications to computational physics problems
- Extracting Koopman operators for prediction and control of nonlinear dynamics using two-stage learning and oblique projections
- Data-driven predictor of control-affine nonlinear dynamics: finite discrete-time bilinear approximation of Koopman operator
- Multi-fidelity reduced-order surrogate modelling
- Equivariance and partial observations in Koopman operator theory for partial differential equations
- Model predictive control of autonomous vehicles based on data-driven Koopman- f model with extended dynamic mode decomposition
- Residual-Guided Dictionary Learning for Spectrally Accurate Koopman Approximation
- Learning dynamically inspired bases for Koopman and transfer operator approximation
This page was built for publication: Linearly recurrent autoencoder networks for learning dynamics
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5383200)
