Time Series Source Separation Using Dynamic Mode Decomposition
From MaRDI portal
Publication:5114419
Time series, auto-correlation, regression, etc. in statistics (GARCH) (62M10) Signal detection and filtering (aspects of stochastic processes) (60G35) Time series analysis of dynamical systems (37M10) Perturbation theory of linear operators (47A55) Eigenvalue problems for linear operators (47A75) Dynamical systems in numerical analysis (37N30)
Abstract: The Dynamic Mode Decomposition (DMD) extracted dynamic modes are the non-orthogonal eigenvectors of the matrix that best approximates the one-step temporal evolution of the multivariate samples. In the context of dynamical system analysis, the extracted dynamic modes are a generalization of global stability modes. We apply DMD to a data matrix whose rows are linearly independent, additive mixtures of latent time series. We show that when the latent time series are uncorrelated at a lag of one time-step then, in the large sample limit, the recovered dynamic modes will approximate, up to a column-wise normalization, the columns of the mixing matrix. Thus, DMD is a time series blind source separation algorithm in disguise, but is different from closely related second order algorithms such as the Second-Order Blind Identification (SOBI) method and the Algorithm for Multiple Unknown Signals Extraction (AMUSE). All can unmix mixed stationary, ergodic Gaussian time series in a way that kurtosis-based Independent Components Analysis (ICA) fundamentally cannot. We use our insights on single lag DMD to develop a higher-lag extension, analyze the finite sample performance with and without randomly missing data, and identify settings where the higher lag variant can outperform the conventional single lag variant. We validate our results with numerical simulations, and highlight how DMD can be used in change point detection.
Recommendations
- Time series decomposition into oscillation components and phase estimation
- Multiresolution mode decomposition for adaptive time series analysis
- Blind source separation for compositional time series
- Time-dependent wave splitting and source separation
- Multivariate time series decomposition into oscillation components
- Blind source separation using time-frequency analysis
- Dynamic mode decomposition and its variants
- Data-driven spatiotemporal modal decomposition for time frequency analysis
- Detecting regime transitions in time series using dynamic mode decomposition
Cites work
- scientific article; zbMATH DE number 1233949 (Why is no real title available?)
- scientific article; zbMATH DE number 1049347 (Why is no real title available?)
- scientific article; zbMATH DE number 6026126 (Why is no real title available?)
- scientific article; zbMATH DE number 5223072 (Why is no real title available?)
- scientific article; zbMATH DE number 3408799 (Why is no real title available?)
- 10.1162/jmlr.2003.4.7-8.1297
- A Probabilistic and RIPless Theory of Compressed Sensing
- A data-driven approximation of the koopman operator: extending dynamic mode decomposition
- Analysis of Fluid Flows via Spectral Properties of the Koopman Operator
- Autocorrelation, autoregression and autoregressive approximation
- Derivatives and Perturbations of Eigenvectors
- Dynamic mode decomposition for financial trading strategies
- Dynamic mode decomposition of numerical and experimental data
- Dynamic mode decomposition with control
- Efficient independent component analysis
- High-dimensional Ising model selection using \(\ell _{1}\)-regularized logistic regression
- Koopman Operator Family Spectrum for Nonautonomous Systems
- Koopman-mode decomposition of the cylinder wake
- Matrix Analysis
- Multidimensional multitaper spectral estimation
- On Consistent Estimates of the Spectrum of a Stationary Time Series
- On consistency and sparsity for principal components analysis in high dimensions
- On dynamic mode decomposition: theory and applications
- Online Dynamic Mode Decomposition for Time-Varying Systems
- OptShrink: An Algorithm for Improved Low-Rank Signal Matrix Denoising by Optimal, Data-Driven Singular Value Shrinkage
- Random perturbation of low rank matrices: improving classical bounds
- Separation of uncorrelated stationary time series using autocovariance matrices
- Some estimates of norms of random matrices
- Spectral analysis of nonlinear flows
- The method of proper orthogonal decomposition for dynamical characterization and order reduction of mechanical systems: an overview
- The uniform convergence of autocovariances
- Variants of dynamic mode decomposition: boundary condition, Koopman, and Fourier analyses
Cited in
(5)- Data-driven spatiotemporal modal decomposition for time frequency analysis
- Blind source separation for compositional time series
- Graph embedded dynamic mode decomposition for stock price prediction
- Decomposing signals from dynamical systems using shadow manifold interpolation
- Dynamic deconvolution and identification of independent autoregressive sources
This page was built for publication: Time Series Source Separation Using Dynamic Mode Decomposition
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5114419)
