We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Numerical Analysis

Title: A numerical comparison of some heuristic stopping rules for nonlinear Landweber iteration

Abstract: The choice of a suitable regularization parameter is an important part of most regularization methods for inverse problems. In the absence of reliable estimates of the noise level, heuristic parameter choice rules can be used to accomplish this task. While they are already fairly well-understood and tested in the case of linear problems, not much is known about their behaviour for nonlinear problems and even less in the respective case of iterative regularization. Hence, in this paper, we numerically study the performance of some of these rules when used to determine a stopping index for Landweber iteration for various nonlinear inverse problems. These are chosen from different practically relevant fields such as integral equations, parameter estimation, and tomography.
Comments: 28 pages, 12 figures
Subjects: Numerical Analysis (math.NA)
Cite as: arXiv:2205.09831 [math.NA]
  (or arXiv:2205.09831v1 [math.NA] for this version)

Submission history

From: Simon Hubmer [view email]
[v1] Thu, 19 May 2022 20:10:30 GMT (669kb,D)

Link back to: arXiv, form interface, contact.