Error analysis for finite element operator learning methods for solving parametric second-order elliptic PDEs
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909183723438080 |
|---|---|
| author | Hong, Youngjoon Ko, Seungchan Lee, Jaeyong |
| author_facet | Hong, Youngjoon Ko, Seungchan Lee, Jaeyong |
| contents | In this paper, we provide a theoretical analysis of a type of operator learning method without data reliance based on the classical finite element approximation, which is called the finite element operator network (FEONet). We first establish the convergence of this method for general second-order linear elliptic PDEs with respect to the parameters for neural network approximation. In this regard, we address the role of the condition number of the finite element matrix in the convergence of the method. Secondly, we derive an explicit error estimate for the self-adjoint case. For this, we investigate some regularity properties of the solution in certain function classes for a neural network approximation, verifying the sufficient condition for the solution to have the desired regularity. Finally, we will also conduct some numerical experiments that support the theoretical findings, confirming the role of the condition number of the finite element matrix in the overall convergence. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_17868 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Error analysis for finite element operator learning methods for solving parametric second-order elliptic PDEs Hong, Youngjoon Ko, Seungchan Lee, Jaeyong Numerical Analysis Machine Learning In this paper, we provide a theoretical analysis of a type of operator learning method without data reliance based on the classical finite element approximation, which is called the finite element operator network (FEONet). We first establish the convergence of this method for general second-order linear elliptic PDEs with respect to the parameters for neural network approximation. In this regard, we address the role of the condition number of the finite element matrix in the convergence of the method. Secondly, we derive an explicit error estimate for the self-adjoint case. For this, we investigate some regularity properties of the solution in certain function classes for a neural network approximation, verifying the sufficient condition for the solution to have the desired regularity. Finally, we will also conduct some numerical experiments that support the theoretical findings, confirming the role of the condition number of the finite element matrix in the overall convergence. |
| title | Error analysis for finite element operator learning methods for solving parametric second-order elliptic PDEs |
| topic | Numerical Analysis Machine Learning |
| url | https://arxiv.org/abs/2404.17868 |