This project continues work done under IRI-9322400, which approaches instances of the general nonlinear programming problem through the use of hybrid systems based on genetic algorithms. The systems, in which genetic operators may vary with the type of problem, is designed to handle both continuous and discrete objective functions and both linear and nonlinear constraints on continuous, integer, and Boolean variables. Some interesting clues to the difficulty in obtaining global solutions to these problems have been found in the previous work and are now being investigated. The results of this research will provide a better understanding of the principles of genetic algorithms and their applicability to constrained problems of the sort found in intelligent systems and other computational applications.
