Fractional Tikhonov regularization for linear discrete ill-posed problems (Q533711): Difference between revisions
From MaRDI portal
Created a new Item |
ReferenceBot (talk | contribs) Changed an Item |
||
(7 intermediate revisions by 6 users not shown) | |||
Property / author | |||
Property / author: Lothar Reichel / rank | |||
Property / author | |||
Property / author: Lothar Reichel / rank | |||
Normal rank | |||
Property / review text | |||
For a given discrete ill-posed problem \(Ax \approx b\), the following regularized least-squares problem is proposed: \[ \min_{x \in \mathbb R^n}\{ \|Ax - b\|^2_W + \mu \|x\|^2 \} \] with the weighted norm \(\|y\|_W = (y^T W y)^{1/2}\) for \(W = (AA^T)^{(\alpha-1)/2}\). This formulation requires the specification of two parameters \(\alpha,\mu>0\) and much of the paper is devoted to studying the influence of these parameters on the solution. The corresponding normal equations \[ ( (A^TA)^{(\alpha+1)/2} + \mu I)x = (A^TA)^{(\alpha-1)/2} A^T b \] characterize the solution, implying that \(x\) admits an expansion \[ x = \varphi(\sigma_1) ( u_1^T b)v_1 + \varphi(\sigma_2) ( u_2^T b)v_2 + \cdots, \] where \(\sigma_j,u_j,v_j\) denote the singular values/vectors of \(A\). The so-called filter function \(\varphi\) takes the form \(\varphi(\sigma) = \frac{\sigma^\alpha}{\sigma^{\alpha + 1} + \mu}\). This expansion not only allows for the computation of \(x\) for moderately sized problems (for large-scale problems a Krylov subspace approach is proposed) but also allows for the derivation of various theoretical results. It is shown how the discrepancy principle can be applied to determine a suitable value of \(\mu\). The sensitivity of the resulting choice for \(\mu\) is investigated. Concerning the choice of \(\alpha\), the authors conclude -- based on numerical experiments -- that ``\dots letting \(\alpha\) be smaller than, but close to unity, such as \(\alpha = 0.8\), gives better results than \(\alpha = 1\) for almost all examples. Smaller values of \(\alpha\), such as \(\alpha = 0.5\), may work even better for some examples, at the cost of rendering worse results for others.'' | |||
Property / review text: For a given discrete ill-posed problem \(Ax \approx b\), the following regularized least-squares problem is proposed: \[ \min_{x \in \mathbb R^n}\{ \|Ax - b\|^2_W + \mu \|x\|^2 \} \] with the weighted norm \(\|y\|_W = (y^T W y)^{1/2}\) for \(W = (AA^T)^{(\alpha-1)/2}\). This formulation requires the specification of two parameters \(\alpha,\mu>0\) and much of the paper is devoted to studying the influence of these parameters on the solution. The corresponding normal equations \[ ( (A^TA)^{(\alpha+1)/2} + \mu I)x = (A^TA)^{(\alpha-1)/2} A^T b \] characterize the solution, implying that \(x\) admits an expansion \[ x = \varphi(\sigma_1) ( u_1^T b)v_1 + \varphi(\sigma_2) ( u_2^T b)v_2 + \cdots, \] where \(\sigma_j,u_j,v_j\) denote the singular values/vectors of \(A\). The so-called filter function \(\varphi\) takes the form \(\varphi(\sigma) = \frac{\sigma^\alpha}{\sigma^{\alpha + 1} + \mu}\). This expansion not only allows for the computation of \(x\) for moderately sized problems (for large-scale problems a Krylov subspace approach is proposed) but also allows for the derivation of various theoretical results. It is shown how the discrepancy principle can be applied to determine a suitable value of \(\mu\). The sensitivity of the resulting choice for \(\mu\) is investigated. Concerning the choice of \(\alpha\), the authors conclude -- based on numerical experiments -- that ``\dots letting \(\alpha\) be smaller than, but close to unity, such as \(\alpha = 0.8\), gives better results than \(\alpha = 1\) for almost all examples. Smaller values of \(\alpha\), such as \(\alpha = 0.5\), may work even better for some examples, at the cost of rendering worse results for others.'' / rank | |||
Normal rank | |||
Property / reviewed by | |||
Property / reviewed by: Daniel Kressner / rank | |||
Normal rank | |||
Property / Mathematics Subject Classification ID | |||
Property / Mathematics Subject Classification ID: 65F22 / rank | |||
Normal rank | |||
Property / zbMATH DE Number | |||
Property / zbMATH DE Number: 5883651 / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
ill-posed problem | |||
Property / zbMATH Keywords: ill-posed problem / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
fractional Tikhonov regularization | |||
Property / zbMATH Keywords: fractional Tikhonov regularization / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
weighted residual norm | |||
Property / zbMATH Keywords: weighted residual norm / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
filter function | |||
Property / zbMATH Keywords: filter function / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
discrepancy principle | |||
Property / zbMATH Keywords: discrepancy principle / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
solution norm constraint | |||
Property / zbMATH Keywords: solution norm constraint / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
Krylov subspace method | |||
Property / zbMATH Keywords: Krylov subspace method / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
least-squares problem | |||
Property / zbMATH Keywords: least-squares problem / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
normal equations | |||
Property / zbMATH Keywords: normal equations / rank | |||
Normal rank | |||
Property / zbMATH Keywords | |||
singular values | |||
Property / zbMATH Keywords: singular values / rank | |||
Normal rank | |||
Property / describes a project that uses | |||
Property / describes a project that uses: Regularization tools / rank | |||
Normal rank | |||
Property / MaRDI profile type | |||
Property / MaRDI profile type: MaRDI publication profile / rank | |||
Normal rank | |||
Property / full work available at URL | |||
Property / full work available at URL: https://doi.org/10.1007/s10543-011-0313-9 / rank | |||
Normal rank | |||
Property / OpenAlex ID | |||
Property / OpenAlex ID: W2168246956 / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Q4868585 / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Lanczos-based exponential filtering for discrete ill-posed problems / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Tikhonov Regularization with a Solution Constraint / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Q5689624 / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Regularization tools version \(4.0\) for matlab \(7.3\) / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Regularization by fractional filter methods and data smoothing / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: Accelerating the LSTRS Algorithm / rank | |||
Normal rank | |||
Property / cites work | |||
Property / cites work: A Trust-Region Approach to the Regularization of Large-Scale Discrete Forms of Ill-Posed Problems / rank | |||
Normal rank | |||
links / mardi / name | links / mardi / name | ||
Latest revision as of 01:05, 4 July 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Fractional Tikhonov regularization for linear discrete ill-posed problems |
scientific article |
Statements
Fractional Tikhonov regularization for linear discrete ill-posed problems (English)
0 references
4 May 2011
0 references
For a given discrete ill-posed problem \(Ax \approx b\), the following regularized least-squares problem is proposed: \[ \min_{x \in \mathbb R^n}\{ \|Ax - b\|^2_W + \mu \|x\|^2 \} \] with the weighted norm \(\|y\|_W = (y^T W y)^{1/2}\) for \(W = (AA^T)^{(\alpha-1)/2}\). This formulation requires the specification of two parameters \(\alpha,\mu>0\) and much of the paper is devoted to studying the influence of these parameters on the solution. The corresponding normal equations \[ ( (A^TA)^{(\alpha+1)/2} + \mu I)x = (A^TA)^{(\alpha-1)/2} A^T b \] characterize the solution, implying that \(x\) admits an expansion \[ x = \varphi(\sigma_1) ( u_1^T b)v_1 + \varphi(\sigma_2) ( u_2^T b)v_2 + \cdots, \] where \(\sigma_j,u_j,v_j\) denote the singular values/vectors of \(A\). The so-called filter function \(\varphi\) takes the form \(\varphi(\sigma) = \frac{\sigma^\alpha}{\sigma^{\alpha + 1} + \mu}\). This expansion not only allows for the computation of \(x\) for moderately sized problems (for large-scale problems a Krylov subspace approach is proposed) but also allows for the derivation of various theoretical results. It is shown how the discrepancy principle can be applied to determine a suitable value of \(\mu\). The sensitivity of the resulting choice for \(\mu\) is investigated. Concerning the choice of \(\alpha\), the authors conclude -- based on numerical experiments -- that ``\dots letting \(\alpha\) be smaller than, but close to unity, such as \(\alpha = 0.8\), gives better results than \(\alpha = 1\) for almost all examples. Smaller values of \(\alpha\), such as \(\alpha = 0.5\), may work even better for some examples, at the cost of rendering worse results for others.''
0 references
ill-posed problem
0 references
fractional Tikhonov regularization
0 references
weighted residual norm
0 references
filter function
0 references
discrepancy principle
0 references
solution norm constraint
0 references
Krylov subspace method
0 references
least-squares problem
0 references
normal equations
0 references
singular values
0 references