A new class of self-scaling for quasi-newton method based on the quadratic model

Basim A. Hassan (University of Mosul)
Ranen M. Sulaiman (University of Mosul)

Indonesian Journal of Electrical Engineering and Computer Science
Vol 21, No 3: March 2021

Quasi-Newton method is an efficient method for solving unconstrained optimization problems. Self-scaling is one of the common approaches in the modification of the quasi-Newton method. A large variety of self-scaling of quasi-Newton methods is very well known. In this paper, based on quadratic function we derive the new self-scaling of quasi-Newton method and study the convergence property. Numerical results on the collection of problems showed the self-scaling of quasi-Newton methods which improves overall numerical performance for BFGS method.