CPU-time and RAM memory optimization for solving dynamic inverse problems using gradient-based approach
•Gradient-based approach is a common scheme for solving the inverse problems.•It is crucial to efficiently compute the gradient of the functional.•The proposed algorithm for the gradient calculation is 10 percent faster.•Low memory requirements are important for large grids. Numerical solution of in...
Gespeichert in:
Veröffentlicht in: | Journal of computational physics 2021-08, Vol.439, p.110374, Article 110374 |
---|---|
Hauptverfasser: | , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | •Gradient-based approach is a common scheme for solving the inverse problems.•It is crucial to efficiently compute the gradient of the functional.•The proposed algorithm for the gradient calculation is 10 percent faster.•Low memory requirements are important for large grids.
Numerical solution of inverse problem for 2D acoustic system of conservation laws by gradient type method requires storage of O(N3) elements which is crucial on large grids with O(N) points in single dimension. In this article we present an approach to save twice memory on the stage of adjoint problem and gradient calculation and compare it with usual approach in memory and CPU time cost. Numerical comparison for CPU time and memory of one step of iteration process which consists of direct problem solution, adjoint problem solution and calculation of the gradient are presented. |
---|---|
ISSN: | 0021-9991 1090-2716 |
DOI: | 10.1016/j.jcp.2021.110374 |