Variable Projection Methods for an Optimized Dynamic Mode Decomposition

The dynamic mode decomposition (DMD) has become a leading tool for data-driven modeling of dynamical systems, providing a regression framework for fitting linear dynamical models to time-series measurement data. We present a simple algorithm for computing an optimized version of the DMD for data which may be collected at unevenly spaced sample times. By making use of the variable projection method for nonlinear least squares problems, the algorithm is capable of solving the underlying nonlinear optimization problem efficiently. We explore the performance of the algorithm with some numerical examples for synthetic and real data from dynamical systems and find that the resulting decomposition displays less bias in the presence of noise than standard DMD algorithms. Because of the flexibility of the algorithm, we also present some interesting new options for DMD-based analysis.

[1]  Thomas M. Smith,et al.  Daily High-Resolution-Blended Analyses for Sea Surface Temperature , 2007 .

[2]  G. Golub,et al.  Separable nonlinear least squares: the variable projection method and its applications , 2003 .

[3]  P. Schmid,et al.  Dynamic mode decomposition of numerical and experimental data , 2008, Journal of Fluid Mechanics.

[4]  C. Loan,et al.  Nineteen Dubious Ways to Compute the Exponential of a Matrix , 1978 .

[5]  David L. Donoho,et al.  The Optimal Hard Threshold for Singular Values is 4/sqrt(3) , 2013, 1305.5870.

[6]  M. R. Osborne.  Nonlinear least squares — the Levenberg algorithm revisited , 1976, The Journal of the Australian Mathematical Society. Series B. Applied Mathematics.

[7]  James V. Burke,et al.  Algorithmic Differentiation of Implicit Functions and Optimal Values , 2008 .

[8]  Kurt Bryan,et al.  Making Do with Less: An Introduction to Compressed Sensing , 2013, SIAM Rev..

[9]  T. M. Chin,et al.  A long-term record of blended satellite and in situ sea-surface temperature for climate monitoring, modeling and environmental studies , 2016 .

[10]  Steven L. Brunton,et al.  On dynamic mode decomposition: Theory and applications , 2013, 1312.0041.

[11]  Sabine Van Huffel,et al.  Exponential Data Fitting and Its Applications , 2018 .

[12]  Clarence W. Rowley,et al.  Model Reduction for fluids, Using Balanced Proper Orthogonal Decomposition , 2005, Int. J. Bifurc. Chaos.

[13]  R. Leroux,et al.  Dynamic mode decomposition for non-uniformly sampled data , 2016 .

[14]  D. Marquardt.  An Algorithm for Least-Squares Estimation of Nonlinear Parameters , 1963 .

[15]  David R. Williams,et al.  Linear models for control of cavity flow oscillations , 2006, Journal of Fluid Mechanics.

[16]  W. Marsden.  I and J , 2012 .

[17]  R.G. Baraniuk,et al.  Compressive Sensing [Lecture Notes] , 2007, IEEE Signal Processing Magazine.

[18]  Peter J. Schmid,et al.  Sparsity-promoting dynamic mode decomposition , 2012, 1309.4165.

[19]  Nathan Halko,et al.  Finding Structure with Randomness: Probabilistic Algorithms for Constructing Approximate Matrix Decompositions , 2009, SIAM Rev..

[20]  Kenneth Levenberg.  A METHOD FOR THE SOLUTION OF CERTAIN NON – LINEAR PROBLEMS IN LEAST SQUARES , 1944 .

[21]  Clarence W. Rowley,et al.  De-biasing the dynamic mode decomposition for applied Koopman spectral analysis of noisy datasets , 2015, Theoretical and Computational Fluid Dynamics.

[22]  C. Rowley,et al.  Modeling of transitional channel flow using balanced proper orthogonal decomposition , 2007, 0707.4112.

[23]  N. Benjamin Erichson,et al.  Randomized low-rank Dynamic Mode Decomposition for motion detection , 2015, Comput. Vis. Image Underst..

[24]  Aleksandr Y. Aravkin,et al.  Estimating nuisance parameters in inverse problems , 2012, 1206.6532.

[25]  Gene H. Golub,et al.  The differentiation of pseudo-inverses and non-linear least squares problems whose variables separate , 1972, Milestones in Matrix Computation.

[26]  I. Mezić,et al.  Analysis of Fluid Flows via Spectral Properties of the Koopman Operator , 2013 .

[27]  F. Guéniat,et al.  A dynamic mode decomposition approach for large and arbitrarily sampled systems , 2015 .

[28]  Clarence W. Rowley,et al.  Characterizing and correcting for the effect of sensor noise in the dynamic mode decomposition , 2014, Experiments in Fluids.

[29]  Clarence W. Rowley,et al.  Variants of Dynamic Mode Decomposition: Boundary Condition, Koopman, and Fourier Analyses , 2012, J. Nonlinear Sci..

[30]  Steven L. Brunton,et al.  Dynamic mode decomposition - data-driven modeling of complex systems , 2016 .

[31]  I. Mezić,et al.  Spectral analysis of nonlinear flows , 2009, Journal of Fluid Mechanics.

[32]  E.J. Candes,et al.  An Introduction To Compressive Sampling , 2008, IEEE Signal Processing Magazine.

[33]  Dianne P. O'Leary,et al.  Variable projection for nonlinear least squares problems , 2012, Computational Optimization and Applications.

[34]  Gene H. Golub,et al.  Ill-conditioned eigensystems and the computation of the Jordan canonical form , 1975, Milestones in Matrix Computation.

[35]  Dmitriy Drusvyatskiy,et al.  Variable projection without smoothness , 2016 .

[36]  Victor Pereyra and Godela Scherer.  Exponential data fitting , 2010 .

[37]  Shervin Bagheri,et al.  Koopman-mode decomposition of the cylinder wake , 2013, Journal of Fluid Mechanics.

[38]  Linda Kaufman,et al.  A Variable Projection Method for Solving Separable Nonlinear Least Squares Problems , 1974 .

[39]  Axel Ruhe,et al.  Algorithms for separable nonlinear least squares problems , 1980 .

[40]  A. Gilbert,et al.  A generalization of variable elimination for separable inverse problems beyond least squares , 2013, 1302.0441.

[41]  Clarence W. Rowley,et al.  Dynamics and control of high-reynolds-number flow over open cavities , 2006 .

[42]  Clarence W. Rowley,et al.  Spectral analysis of fluid flows using sub-Nyquist-rate PIV data , 2014, Experiments in Fluids.

[43]  Katharine M. Mullen,et al.  Algorithms for separable nonlinear least squares with application to modelling time-resolved spectra , 2007, J. Glob. Optim..

[44]  Kevin Barraclough,et al.  I and i , 2001, BMJ : British Medical Journal.

[45]  D. Donoho,et al.  The Optimal Hard Threshold for Singular Values is 4 / √ 3 , 2013 .