A quantitative analysis of Koopman operator methods for system identification and predictions
We give convergence and cost estimates for a data-driven system identification method: given an unknown dynamical system, the aim is to recover its vector field and its flow from trajectory data. It is based on the so-called Koopman operator, which uses the well-known link between differential equat...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Académie des sciences
2022-12-01
|
| Series: | Comptes Rendus. Mécanique |
| Subjects: | |
| Online Access: | https://comptes-rendus.academie-sciences.fr/mecanique/articles/10.5802/crmeca.138/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1825206011878703104 |
|---|---|
| author | Zhang, Christophe Zuazua, Enrique |
| author_facet | Zhang, Christophe Zuazua, Enrique |
| author_sort | Zhang, Christophe |
| collection | DOAJ |
| description | We give convergence and cost estimates for a data-driven system identification method: given an unknown dynamical system, the aim is to recover its vector field and its flow from trajectory data. It is based on the so-called Koopman operator, which uses the well-known link between differential equations and linear transport equations. Data-driven methods recover specific finite-dimensional approximations of the Koopman operator, which can be understood as a transport operator. We focus on such approximations given by classical finite element spaces, which allow us to give estimates on the approximation of the Koopman operator as well as the solutions of the associated linear transport equation. These approximations are thus relevant objects to solve the system identification problem.We then analyze the convergence of a variant of the generator Extended Dynamic Mode Decomposition (gEDMD) algorithm, one of the main algorithms developed to compute approximations of the Koopman operator from data. We find however that, when combining this algorithm with classical finite element spaces, the results are not satisfactory numerically, as the convergence of the data-driven approximation is too slow for the method to benefit from the accuracy of finite element spaces. In particular, for problems in dimension 1 it is less efficient than direct interpolation methods to recover the vector field. We provide some numerical examples to illustrate this last point. |
| format | Article |
| id | doaj-art-7966c0684c9349e3ae05779336968546 |
| institution | Kabale University |
| issn | 1873-7234 |
| language | English |
| publishDate | 2022-12-01 |
| publisher | Académie des sciences |
| record_format | Article |
| series | Comptes Rendus. Mécanique |
| spelling | doaj-art-7966c0684c9349e3ae057793369685462025-02-07T13:46:21ZengAcadémie des sciencesComptes Rendus. Mécanique1873-72342022-12-01351S172175110.5802/crmeca.13810.5802/crmeca.138A quantitative analysis of Koopman operator methods for system identification and predictionsZhang, Christophe0https://orcid.org/0000-0002-9978-4997Zuazua, Enrique1Chair in Dynamics, Control and Numerics, (Alexander von Humboldt Professorship), Department of Data Science, Friedrich Alexander Universität Erlangen-Nürnberg, 91058 Erlangen, GermanyChair in Dynamics, Control and Numerics, (Alexander von Humboldt Professorship), Department of Data Science, Friedrich Alexander Universität Erlangen-Nürnberg, 91058 Erlangen, Germany; Chair of Computational Mathematics, Fundación Deusto, Avenida de las Universidades 24, 48007 Bilbao, Basque Country, Spain; Departamento de Mateáticas, Universidad Autónoma de Madrid, 28049 Madrid, SpainWe give convergence and cost estimates for a data-driven system identification method: given an unknown dynamical system, the aim is to recover its vector field and its flow from trajectory data. It is based on the so-called Koopman operator, which uses the well-known link between differential equations and linear transport equations. Data-driven methods recover specific finite-dimensional approximations of the Koopman operator, which can be understood as a transport operator. We focus on such approximations given by classical finite element spaces, which allow us to give estimates on the approximation of the Koopman operator as well as the solutions of the associated linear transport equation. These approximations are thus relevant objects to solve the system identification problem.We then analyze the convergence of a variant of the generator Extended Dynamic Mode Decomposition (gEDMD) algorithm, one of the main algorithms developed to compute approximations of the Koopman operator from data. We find however that, when combining this algorithm with classical finite element spaces, the results are not satisfactory numerically, as the convergence of the data-driven approximation is too slow for the method to benefit from the accuracy of finite element spaces. In particular, for problems in dimension 1 it is less efficient than direct interpolation methods to recover the vector field. We provide some numerical examples to illustrate this last point.https://comptes-rendus.academie-sciences.fr/mecanique/articles/10.5802/crmeca.138/Koopman operatorSystem identificationFinite element spacesData-driven approximationExtended dynamic mode decomposition |
| spellingShingle | Zhang, Christophe Zuazua, Enrique A quantitative analysis of Koopman operator methods for system identification and predictions Comptes Rendus. Mécanique Koopman operator System identification Finite element spaces Data-driven approximation Extended dynamic mode decomposition |
| title | A quantitative analysis of Koopman operator methods for system identification and predictions |
| title_full | A quantitative analysis of Koopman operator methods for system identification and predictions |
| title_fullStr | A quantitative analysis of Koopman operator methods for system identification and predictions |
| title_full_unstemmed | A quantitative analysis of Koopman operator methods for system identification and predictions |
| title_short | A quantitative analysis of Koopman operator methods for system identification and predictions |
| title_sort | quantitative analysis of koopman operator methods for system identification and predictions |
| topic | Koopman operator System identification Finite element spaces Data-driven approximation Extended dynamic mode decomposition |
| url | https://comptes-rendus.academie-sciences.fr/mecanique/articles/10.5802/crmeca.138/ |
| work_keys_str_mv | AT zhangchristophe aquantitativeanalysisofkoopmanoperatormethodsforsystemidentificationandpredictions AT zuazuaenrique aquantitativeanalysisofkoopmanoperatormethodsforsystemidentificationandpredictions AT zhangchristophe quantitativeanalysisofkoopmanoperatormethodsforsystemidentificationandpredictions AT zuazuaenrique quantitativeanalysisofkoopmanoperatormethodsforsystemidentificationandpredictions |