Convergence and optimality of adaptive regularization for ill-posed deconvolution problems in infinite spaces
The paper is concerned with approximate methods for solving the linear operator equation \(Wr=d_n\), where \(r\) belongs to an input Hilbert space \(R\) and \(d_n=d_{true}+n\) belongs to an output Hilbert space \(D\), and \(n\) is a random vector containing a noise or an error. Regularization methods of Tikhonov type consists in replacing the considered equation by an equation family \((\Phi+\alpha H)r=\widetilde{r}\), where \(\Phi=W^*W\), \(\widetilde{r}=W^*d_n\). The authors choose \(H=\Phi^{-1}\) (under the assumption that \(\Phi^{-1}\) exists) and call the corresponding method the adaptive regularization. They remark that this method was first proposed in the paper of \textit{G. A. Ryzhikov} and \textit{M. S. Biryulina}, [``Sharp deconvolution in elimination of multiples, 68th Ann. Internat. Mtg. Soc. Expl. Geophys. (1988)], for finite dimensional spaces. Now, this paper gives a theoretical analysis of the method. The cases of a priori choice as well as that of a posteriori choice of the regularization parameter \(\alpha\) are investigated. Results on convergence and asymptotical optimality are proved. The authors remark that the method can be useful in deconvolution problems which arise frequently in geophysical sciences.
- Iterated adaptive regularization for the operator equations of the first kind
- Solution of Ill-Posed Problems via Adaptive Grid Regularization: Convergence Analysis
- On the Adaptive Selection of the Parameter in Regularization of Ill-Posed Problems
- On optimal solutions of the deconvolution problem
- Tikhonov Regularization for Finitely and Infinitely Smoothing Operators
This page was built for publication: Convergence and optimality of adaptive regularization for ill-posed deconvolution problems in infinite spaces
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2431964)
