An Efficient Step Size For BFGS

eng Ansari Education and Research Society Journal of Ultra Scientist of Physical Sciences December 2008 20 3 841 847 jusps-A 1453 article An Efficient Step Size For BFGS Azfi Zaidi Mohammad Sofi (azfizaidi@yahoo.com) 1 Mustafa Mamat (mus@umt.edu.my) 1 Ismail Mohd (ismail@umt.edu.my) 1 Yosza Dasril (yosza@utem.edu.my) 2 Department of Mathematics, Facutly of Science and Technology, University Malaysia Terengganu (UMT) (Malaysia) Faculty of Electronics and Computer Engineering University Teknikal Melaka (UteM) Melaka, Malaysia In this paper we introduced what we are calling a combination of step size of Broyden-Fletcher-Goldfarb-Shanno (CSS-BFGS) to solve unconstrained optimization problem. Analytically, we have proved that the algorithm is superlinearly convergence. The numerical result of CSS-BFGS was shown at the end of this paper. The result shown that in term of iteration number ni and function evaluation nf, our alternative step size is more effective than the original BFGS algorithm.    https://www.ultrascientist.org/paper/1453/ u00a0exact line search step size BFGS differebtiable superlinearly convergence