Zlatko Drmač
Abstract
The Dynamic Mode Decomposition (DMD) is a versatile and increasingly popular method for data driven analysis of dynamical systems that arise in a variety of applications in, e.g., computational fluid dynamics, robotics or machine learning. In the framework of numerical linear algebra, it is a data driven Rayleigh-Ritz procedure applied to a DMD matrix that is derived from the supplied data. In some applications, the physics of the underlying problem implies hermiticity of the DMD matrix, so the general DMD procedure is not computationally optimal. Furthermore, it does not guarantee important structural properties of the Hermitian eigenvalue problem and may return non-physical solutions. This paper proposes a software solution to the Hermitian (including the real symmetric) DMD matrices, accompanied with a numerical analysis that contains several fine and instructive numerical details. The eigenpairs are computed together with their residuals, and perturbation theory provides error bounds for the eigenvalues and eigenvectors. The software is developed and tested using the
- [1] . 2017. Roadmap for the Development of a Linear Algebra Library for Exascale Computing: SLATE: Software for Linear Algebra Targeting Exascale.
SLATE Working Notes 01, ICL-UT-17-02.Google Scholar
- [2] . 2021. Dynamic Mode Decomposition and Its Application in Various Domains: An Overview. Springer International Publishing, Cham, 121–132.
DOI: Google ScholarCross Ref
- [3] . 1999. LAPACK Users’ Guide (Third Ed.). Society for Industrial and Applied Mathematics, Philadelphia, PA, USA. Google ScholarDigital Library
- [4] . 2021. Physics-informed dynamic mode decomposition (piDMD). arXiv e-prints, Article
arXiv:2112.04307 (Dec. 2021), arXiv:2112.04307 pages.arxiv:2112.04307 [math.DS]Google Scholar - [5] . 2022. Modern Koopman theory for dynamical systems. SIAM Rev. 64, 2 (2022), 229–340.
DOI: arXiv:https://doi.org/10.1137/21M1401243 Google ScholarDigital Library - [6] . 2012. Applied Koopmanism. Chaos: An Interdisciplinary Journal of Nonlinear Science 22, 4 (2012), 047510.
DOI: arXiv:https://doi.org/10.1063/1.4772195 Google ScholarCross Ref - [7] . 2020. Modes of homogeneous gradient flows. arXiv e-prints, Article
arXiv:2007.01534 (July 2020), arXiv:2007.01534 pages.DOI: arxiv:2007.01534 [math.DS]Google ScholarCross Ref - [8] . 2022. The mpEDMD algorithm for data-driven computations of measure-preserving dynamical systems. arXiv e-prints, Article
arXiv:2209.02244 (Sept. 2022), arXiv:2209.02244 pages.DOI: arxiv:2209.02244 [math.NA]Google ScholarCross Ref - [9] . 2023. Residual dynamic mode decomposition: Robust and verified Koopmanism. Journal of Fluid Mechanics 955 (2023), A21.
DOI: Google ScholarCross Ref - [10] . 2024. Rigorous data–driven computation of spectral properties of Koopman operators for dynamical systems. Communications on Pure and Applied Mathematics 77, 1 (2024), 221–283.
DOI: arXiv:https://onlinelibrary.wiley.com/doi/pdf/10.1002/cpa.22125 Google ScholarCross Ref - [11] . 2012. Communication-optimal parallel and sequential QR and LU factorizations. SIAM Journal on Scientific Computing 34, 1 (2012), A206–A239.
DOI: arXiv:http://dx.doi.org/10.1137/080731992 Google ScholarDigital Library - [12] . 1997. Applied Numerical Linear Algebra. SIAM.
DOI: Google ScholarCross Ref - [13] . 2014. Accelerating Numerical Dense Linear Algebra Calculations with GPUs. Springer International Publishing, Cham, 3–28.
DOI: Google ScholarCross Ref - [14] . 2020. Dynamic Mode Decomposition-A Numerical Linear Algebra Perspective. Springer International Publishing, Cham, 161–194.
DOI: Google ScholarCross Ref - . [n. d.]. A LAPACK implementation of the dynamic mode decomposition. ACM Trans. Math. Softw. (to appear) ([n. d.]).Google Scholar
- [16] . 2022. A LAPACK Implementation of the Dynamic Mode Decomposition I.
Technical Report . Department of Mathematics, University of Zagreb, Croatia, and AIMdyn Inc. Santa Barbara, CA.LAPACK Working Note 298 .Google Scholar - [17] . 2022. A LAPACK Implementation of the Dynamic Mode Decomposition II.
Technical Report . Department of Mathematics, University of Zagreb, Croatia, and AIMdyn Inc. Santa Barbara, CA.LAPACK Working Note 300 .Google Scholar - [18] . 2023. A data driven Koopman-Schur decomposition for computational analysis of nonlinear dynamics. arXiv e-prints, Article
arXiv:2312.15837 (Dec. 2023), arXiv:2312.15837 pages.DOI: arxiv:2312.15837 [math.NA]Google ScholarCross Ref - [19] . 2018. Data driven modal decompositions: Analysis and enhancements. SIAM Journal on Scientific Computing 40, 4 (2018), A2253–A2285.
DOI: arXiv:https://doi.org/10.1137/17M1144155 Google ScholarDigital Library - [20] . 2021. Identification of nonlinear systems using the infinitesimal generator of the Koopman semigroup - A numerical implementation of the Mauroy-Goncalves method. Mathematics 9, 17 (2021).
DOI: Google ScholarCross Ref - [21] . 1955. Some metric inequalities in the space of matrices. Proc. Amer. Math. Soc. 6, 1 (1955), 111–116. http://www.jstor.org/stable/2032662Google ScholarCross Ref
- [22] . 2020. Extraction and prediction of coherent patterns in incompressible flows through space-time Koopman analysis. Physica D: Nonlinear Phenomena 402 (2020), 132211.
DOI: Google ScholarCross Ref - [23] . 1988. The symmetric procrustes problem. BIT 28 (1988), 133–143.
DOI: Google ScholarDigital Library - [24] . 2012. Low-rank and sparse dynamic mode decomposition. Center for Turbulence Research, Annual Research Briefs (2012), 139–152.Google Scholar
- [25] . 2014. Sparsity-promoting dynamic mode decomposition. Physics of Fluids 26, 2 (
Feb. 2014), 024103.Google ScholarCross Ref - [26] . 2011. The Nearest Orthogonal or Unitary Matrix.
Lecture Notes . University of California, Berkeley.Google Scholar - [27] . 2013. Analysis of fluid flows via spectral properties of the Koopman operator. Annual Reviews of Fluid Mechanics 45 (2013), 357–378.Google ScholarCross Ref
- [28] . 2022. Randomized Numerical Linear Algebra - A Perspective on the Field with an Eye to Software.
Technical Report 299. LAPACK Working Note. http://www.netlib.org/lapack/lawnspdf/lawn299.pdfGoogle Scholar - [29] . 2015. Reproducible tall-skinny QR. In 2015 IEEE 22nd Symposium on Computer Arithmetic. 152–159.
DOI: Google ScholarDigital Library - [30] . 1998. The Symmetric Eigenvalue Problem, Classics in Applied Mathematics 20. SIAM, Philadelphia, PA.Google ScholarDigital Library
- [31] . 2009. Spectral analysis of nonlinear flows. Journal of Fluid Mechanics 641 (2009), 115–127.Google ScholarCross Ref
- [32] . 2010. Dynamic mode decomposition of numerical and experimental data. Journal of Fluid Mechanics 656 (
10 August 2010), 5–28.DOI: Google ScholarCross Ref - [33] . 2011. Application of the dynamic mode decomposition to experimental data. Experiments in Fluids 50, 4 (2011), 1123–1130.
DOI: Google ScholarCross Ref - [34] . 2021. Data-driven and operator-based tools for the analysis of turbulent flows. In Advanced Approaches in Turbulence, (Ed.). Elsevier, 243–305.
DOI: Google ScholarCross Ref - [35] . 2022. Dynamic mode decomposition and its variants. Annual Review of Fluid Mechanics 54, 1 (2022), 225–254.
DOI: arXiv:https://doi.org/10.1146/annurev-fluid-030121-015835 Google ScholarCross Ref - [36] . 2011. Applications of the dynamic mode decomposition. Theoretical and Computational Fluid Dynamics 25, 1 (2011), 249–259.
DOI: Google ScholarCross Ref - [37] . 1990. Matrix Perturbation Theory. Academic Press.Google Scholar
- [38] . 2010. Towards dense linear algebra for hybrid GPU accelerated manycore systems. Parallel Comput. 36, 5-6 (
June 2010), 232–240.DOI: Google ScholarDigital Library - [39] . 2010. Dense linear algebra solvers for multicore with GPU accelerators. In Proc. of the IEEE IPDPS’10. IEEE Computer Society, Atlanta, GA, 1–8.
DOI: Google ScholarCross Ref - [40] . 2014. On dynamic mode decomposition: Theory and applications. Journal of Computational Dynamics 1, 2 (2014), 391–421.Google ScholarCross Ref
- [41] . 1969. Condition numbers and equilibration of matrices. Numer. Math. 14 (1969), 14–23.Google ScholarDigital Library
- [42] . 2015. A data–driven approximation of the Koopman operator: Extending dynamic mode decomposition. Journal of Nonlinear Science 25, 6 (
June 2015), 1307–1346.Google ScholarCross Ref
Index Terms
- Hermitian Dynamic Mode Decomposition - Numerical Analysis and Software Solution
Recommendations
A LAPACK Implementation of the Dynamic Mode Decomposition
The Dynamic Mode Decomposition (DMD) is a method for computational analysis of nonlinear dynamical systems in data driven scenarios. Based on high fidelity numerical simulations or experimental data, the DMD can be used to reveal the latent structures in ...
Bidiagonal SVD Computation via an Associated Tridiagonal Eigenproblem
The Singular Value Decomposition (SVD) is widely used in numerical analysis and scientific computing applications, including dimensionality reduction, data compression and clustering, and computation of pseudo-inverses. In many cases, a crucial part of ...
A Useful Form of Unitary Matrix Obtained from Any Sequence of Unit 2-Norm $n$-Vectors
Charles Sheffield pointed out that the modified Gram-Schmidt (MGS) orthogonalization algorithm for the QR factorization of $B\!\in\!\R^{n\times k}$ is mathematically equivalent to the QR factorization applied to the matrix $B$ augmented with a $k\times ...
Comments