Computational Enhancements to the Method of Multipliers

Abstract

Transformation or penalty function techniques have enjoyed wide popularity for the solution of nonlinear programming problems in recent years. Many methods have been proposed in this class, including the Method of Multipliers. In this paper we present two enhancements which may be used with any transformation technique. The first is a technique for direct handling of variable bounds, and the second is a problem scaling algorithm. Both techniques have been implemented in connection with a Method of Multipliers code, BIAS, and applied to 23 test problems. The numerical results indicate a significant increase in the robustness and overall efficiency of the transformation method employed.