A new method for solving linear ill-posed problems

Zhang J., MƏMMƏDOV M.

Applied Mathematics and Computation, vol.218, no.20, pp.10180-10187, 2012 (SCI-Expanded, Scopus) identifier

  • Nəşrin Növü: Article / Article
  • Cild: 218 Say: 20
  • Nəşr tarixi: 2012
  • Doi nömrəsi: 10.1016/j.amc.2012.03.098
  • jurnalın adı: Applied Mathematics and Computation
  • Jurnalın baxıldığı indekslər: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Səhifə sayı: pp.10180-10187
  • Açar sözlər: Generalized cross validation, Ill-posed problems, L-curve, Newton's method, Tikhonov regularization
  • Açıq Arxiv Kolleksiyası: Məqalə
  • Adres: Bəli

Qısa məlumat

In this paper, we propose a new method for solving large-scale ill-posed problems. This method is based on the Karush-Kuhn-Tucker conditions, Fisher-Burmeister function and the discrepancy principle. The main difference from the majority of existing methods for solving ill-posed problems is that, we do not need to choose a regularization parameter in advance. Experimental results show that the proposed method is effective and promising for many practical problems. © 2012 Elsevier Inc. All rights reserved.