Computational Implementation and Tests of a Sequential Linearization Algorithm for Mixed-Discrete Nonlinear Design Optimization

Abstract

A previous article gave the theoretical background and motivation for a new sequential linearization approach to the solution of mixed-discrete nonlinear optimal design problems. The present sequel article gives the implementation details of program MINSLIP based on this approach. Illustrative examples, modeling issues, and program parameter selection are discussed. A report on extensive computational results with test problems, as well as comparisons with other methods, shows advantages in both robustness and efficiency. Sample design applications are included.