Nonlinear Least pth Approximation and Nonlinear Programming with Applications in the Design of Networks and Systems

Abstract

<p>The purpose of this thesis is to present a unified treatment of nonlinear least pth and nonlinear minimax approximation problems, and a new method for nonlinear programming. Least pth approximation with values of p in the range 1,000 to 1,000,000 in conjunction with the Fletcher-Powell and Fletcher algorithms have been successfully applied to a variety of network and system optimization problems. A comparison is made between two new ~lgorithrns for nonlinear minimax approximation and some of the existing ones. Also, the new approach to nonlinear programming is compared with the well-known SUMT. A critical review of unconstrained optimization is also included.</p>