NONLINEAR-PROGRAMMING BY CUMULATIVE APPROXIMATION REFINEMENT

A mathematical programming method particularly suited for structural a nd multidisciplinary optimization problems is presented. It is based o n the idea that all available information about the problem should be used in the attempt to obtain rapid convergence. Thus, it uses informa tion from previous iterations to produce steadily improving approximat ions of the implicit objective and constraint functions of the problem . In some cases, this creates a response surface type of approximation . The ability of the method to store known information about the behav iour of the problem makes it well-suited for practical multidisciplina ry optimization.