THE PERFORMANCE OF A MODIFIED ARMIJO LINE SEARCH RULE IN BFGS OPTIMIZATION METHOD

Abstract

The performance of a modified Armijo line search rule related to BFGS gradient type method with the results from other well-known line search rules are compared as well as analyzed. Although the modified Armijo rule does require as much computational cost as the other rules, it shows more efficient in finding local minima of unconstrained optimization problems. The sensitivity of the parameters used in the line search rules is also analyzed. The results obtained by implementing algorithms in Matlab for the test problems in [3] are presented.

This article was migrated from the previous system via automation. The abstract may not be written correctly. Please view the PDF file.