A new conjugate gradient method with exact line search
Conjugate gradient (CG) methods have been practically used to solve large-scale unconstrained optimization problems due to their simplicity and low memory storage. In this paper, we proposed a new type of CG coefficients ( ) k . The k is computed as an average between two different types of meth...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
HIKARI Ltd.
2015
|
| Subjects: | |
| Online Access: | http://eprints.unisza.edu.my/6530/1/FH02-FIK-15-03666.jpg http://eprints.unisza.edu.my/6530/ http://dx.doi.org/10.12988/ams.2015.53243 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Conjugate gradient (CG) methods have been practically used to solve large-scale unconstrained optimization problems due to their simplicity and low memory storage. In this paper, we proposed a new type of CG coefficients ( ) k . The k is computed as an average between two different types of method which are Polak and Ribiere (PR) and Norrlaili et al. (NRMI). Numerical comparisons are made with the five others k proposed by the early researches. A set of eight unconstrained optimization problems with several different variables are used in this paper. It is shown that, the new proposed k with an exact line search is possessed global convergence properties. Numerical results also show that this new k outperforms some of these CG methods. |
|---|