Test problems for constrained nonlinear mathematical programming algorithms

Publisher Website
Google Scholar
Abstract
A collection of constrained nonlinear programing problems for use in testing optimization algorithms is presented. The problems vary in size from two variables to one hundred variables with various combinations of linear/nonlinear constraints and objective functions. IBM Fortran IV programs were written to provide function values and gradients for the objective function and constraints. Each coded problem was checked at several points against published results, and a validation process was used to check the values of the objective function, constraints, and gradients. The problems were collected from various sources, and many of them have been used by other authors in published results of their algorithm testing. This report should also be useful in an educational setting to provide students with experience in nontrivial problems. Listings of the IBM Fortran code are included in this report. 10 tables.

This publication has 0 references indexed in Scilit: