Multi-heap constraint handling in gray box evolutionary algorithms

Many optimization problems provide access to the partial or total explicit algebraic representation of the problem, including subfunctions and variable interaction graphs. This extra information allows the development of efficient solvers through new appropriate operators. Besides distinctive reproduction operators for a variety of problem categories, Gray Box algorithms have been proposed as a form to explore this additional information during the search. Considering recent evolutionary operators in the literature, we propose adaptations to Gray Box evolutionary reproduction operators and local search algorithms to deal with constrained problems, a field still little explored in gray box evolutionary optimization. The results show that the proposed methods achieve better solutions than traditional algorithms in a set of constrained binary and integer problems and reach optimal solutions in the literature for some instances.