Enhancements Self-Scaling Quasi-Newton for Unconstrained Optimization

Abstract

A self-scaling for the quasi-Newton tecnique is derive by using a second_order Taylor's expansion to achieve optimal computational performance.   Following this, new updating formulas for the quasi-Newton method are introduced based on the newly derived self-scaling equation. The numerical results confirm this derivation and suggest that the new method could potentially rival the BFGS method in terms of performance.