Condition numbers and perturbation analysis for the Tikhonov regularization of discrete ill-posed problems. (Q2889371)
From MaRDI portal
| This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use this page instead for the normal view: Condition numbers and perturbation analysis for the Tikhonov regularization of discrete ill-posed problems. |
scientific article; zbMATH DE number 6043425
| Language | Label | Description | Also known as |
|---|---|---|---|
| English | Condition numbers and perturbation analysis for the Tikhonov regularization of discrete ill-posed problems. |
scientific article; zbMATH DE number 6043425 |
Statements
7 June 2012
0 references
Tikhonov regularization
0 references
perturbation analysis
0 references
condition number
0 references
discrete ill-posed problems
0 references
numerical examples
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0.9537423
0 references
0.90527076
0 references
0.89942884
0 references
0.8986991
0 references
0.8959657
0 references
0.89551026
0 references
0.8931813
0 references
0.8924022
0 references
0.89123535
0 references
Condition numbers and perturbation analysis for the Tikhonov regularization of discrete ill-posed problems. (English)
0 references
The authors consider Tikhonov regularization for discrete ill-posed problems of the form NEWLINE\[NEWLINE \min _{x}\{\| Ax-b\| _2^2+\lambda ^2\| Lx \| _2^{2}\}, \quad A\in \mathbb {R}^{m\times n},\quad L\in \mathbb {R}^{p\times n}, NEWLINE\]NEWLINE where the regularization parameter \(\lambda \) controls the weight given to the minimization of \(\| Lx\| _2\) relative to the minimization of the residual \(\| Ax-b\| _2\). The general case, when both the coefficient matrix \(A\) and the right-hand side \(b\) are perturbed, is studied. The authors define the normwise, mixed and componentwise condition numbers of Tikhonov regularization, and present formulas for computing these numbers. New perturbation results corresponding to the componentwise errors in \(A\) and \(b\) are derived. Numerical examples compare the new perturbation bounds with some well-known results by \textit{P. C. Hansen} and \textit{A. N. Malyshev}.
0 references