Symbolic Computations Approach in Controlled Enumeration Methods Applied to Discrete Optimization

Abstract

A coupling symbolic and numeric computation approach in the discrete optimization of problems with linear objective function is presented. An enumeration method according to the non decreasing values of the objective function is supplied with an additional symbolic module containing the domain oriented knowledge. This module eliminates the constraints verification for “non- promising” propositions and enables a significant reduction in number of design variables variants that must be checked for feasibility to find the global minimum. An optimization example for a simple truss structure and a few effective heuristic rules are presented.