AN EFFICIENT ITERATIVE FRAMEWORK FOR UNCONSTRAINED NONLINEAR OPTIMIZATION

Abstract

In multi-step Quasi-Newton methods, to update Hessian at each step, interpolatory polynomial is used to derived information from more than one previous steps. To calculate interpolating points Ford and Moghrabi developed some techniques, Fixed and Accumulative method. In this paper, we use three step Accumulative technique with new scheming to update Hessian. The numerical results indicates successive performance with comparison to existing BFGS method.