A note on Tikhonov regularization of large linear problems

DOI10.1023/A:1026091804076MaRDI QIDQ1415410zbMATH OpenOpenAlexFDO

Authors Martin Hanke

Publication date 4 December 2003

Published in BIT (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1023/a:1026091804076

zbMATH Keywords

Gauss and Gauss-Radau quadraturesnorm of residualStieltjes integralTikhonov regularization

Mathematics Subject Classification ID

Numerical quadrature and cubature formulas (65D32) Ill-posedness and regularization problems in numerical linear algebra (65F22)

Review text

This short note completes the analysis of bounds of Tikhonov residual \(\varphi(\alpha)=\| b-Ax_\alpha\| _2^2\) given in the paper of \textit{D. Calvetti} and \textit{L. Reichel} [ibid. 43, No. 2, 263--283 (2003; Zbl 1038.65048)], where \(x_\alpha\) is a solution of Tikhonov regularization of the large linear problem \(Ax=b\). The lower and upper bounds for \(\varphi(\alpha)\) are given by the \(l\)-point Gauss and Gaus-Radau quadratures, respectively. The author proves that these bounds depend monotonically on \(l\), i.e. the upper bound is decreasing monotonically, while the lower bound is increasing monotonically with respect to \(l\). The last statement was proved by Calvetti and Reichel [loc. cit.] on the another way.

Recommendations

• Tikhonov regularization of large linear problems
• Arnoldi-Tikhonov regularization methods
• scientific article; zbMATH DE number 1304508
• GCV for Tikhonov regularization via global Golub-Kahan decomposition.
• Tikhonov Regularization with a Solution Constraint

Cited in

(4)

• An iterative Lavrentiev regularization method
• Some matrix nearness problems suggested by Tikhonov regularization
• Greedy Tikhonov regularization for large linear ill-posed problems
• Matrices, moments, and rational quadrature

Describes a project that uses

Uses Software

• Regularization tools

This page was built for publication: A note on Tikhonov regularization of large linear problems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1415410)