عنوان المقالة: A new hybrid WC-FR conjugate gradient-algorithm with modified secant condition for unconstrained optimization
رنا زيدان يونس الكواز | Rana Z. Al-Kawaz | 4531
نوع النشر
مجلة علمية
المؤلفون بالعربي
المؤلفون بالإنجليزي
Rana Z. Al-Kawaz & Abbas Y. Al-Bayati
الملخص الانجليزي
An accelerated hybrid Conjugate Gradient (CG) algorithm represents the subject of this paper. The parameter is computed as a convex combination of Fletcher and Reeves,[22] and Wu-Chen,[3], ie. The parameter in the convex combination is computed in such a way that the direction corresponding to the CG algorithm is the best direction we know, ie the Newton direction, while the pair (,) satisfies the classical secant condition, where, and. It is shown that both for uniformly convex functions and for general nonlinear functions the new proposed algorithm with strong Wolfe line search is globally convergent. This algorithm uses an acceleration scheme modifying the step-length for improving the reduction of the function values along the iterations. The technique was given by (Andrei [15]). Numerical comparisons with some similar CG algorithms show that the new proposed hybrid computational scheme outperforms the CG algorithms given by Wu-Chen and FR. A set of 35 unconstrained optimization problems with several different dimensions are used in this paper.
تاريخ النشر
01/01/2012
الناشر
Science & Knowledge Publishing Corporation Limited (SCIK)
رقم المجلد
2
رقم العدد
4
ISSN/ISBN
1927-5307
الصفحات
937
رابط خارجي
http://scik.org/index.php/jmcs/article/download/193/113
الكلمات المفتاحية
.conjugate gradient; convex combination; uniformly convex function; globally convergent; unconstrained optimization problem
رجوع