Measurement of Algorithm Performance in Minimization of Constrained Problems

Abstract

In this paper, constrained non-linear programming problems are solved by using the Sequential Unconstrained Minimization Technique ( SUMT). The most popular formulas (DFP& BFGS ) are used with SUMT in minimization of constrained problems. Numerical comparison shows that the Number of Constrained evaluations (NOC) must be used instead of Number of Function evaluations (NOF) as a main factor in the measurement of algorithm performance.