Two modified conjugate gradient methods and their application to image restoration problems

Mehamdia Abd Elhamid^a, Yacine Chaib^b, Hisham M. Khudhur^c, Raouf Ziadi<sup>d

a</sup> Université of M’hamed Bougara
^b Mohamed-Cherif Messaadia University - Souk Ahras
^c College of Computer Science and Mathematics, University of Mosul
^d Université Ferhat Abbas de Sétif

Аннотация: Conjugate gradient methods have significantly contributed to the discovery of minimizes of large-scale unconstrained optimization problems. In this paper, based on the Liu–Storey conjugate gradient method, two modified conjugate gradient methods (named MC1 and MC2 methods) are presented for unconstrained optimization. Under usual assumptions, the two presented methods are proven to be sufficient descent at each iteration. The global convergence results of our methods is established using the strong Wolfe line search (SWLS). Numerical tests demonstrate the effectiveness of the MC1 and MC2 methods when compared to certain existing methods in view of the Dolan and Moré performance profile. Furthermore, the practical applications of these methods in image restoration problems is also considered.

Ключевые слова: Unconstrained optimization, Conjugate gradient method, Analyse convergence, Numerical comparisons

Язык публикации: английский

DOI: 10.15826/umj.2025.2.013

Реферативные базы данных:

•