Fletcher reeves like cg formula approach on broyden family update

• Main Authors: Mustafa, Mamat, Sofi, A.Z.M., Mohd, I., Ibrahim, M.A.H
• Format: Conference or Workshop Item
• Language: English
• Published: 2014
• Subjects: QA75 Electronic computers. Computer science; T Technology (General)
• Online Access: http://eprints.unisza.edu.my/364/1/FH03-FIK-15-02452.jpg; http://eprints.unisza.edu.my/364/

The Broyden family update in quasi-Newton method is known as one of the most efficient update method in solving unconstrained optimization. However, by using the standard search direction, sometimes the algorithm may lead to the failure for some complicated problems and this scenario worsen when the cases of the initial points selected are far away from the minimizer. To overcome this scenario, we proposed a new search direction by using the Fletcher Reeves formula in the conjugate gradient method to be fit with the new search direction. We proved that this new search direction globally converge and positively affect the Broyden family update in quasi-Newton method.