Weighted iterated local branching for mathematical programming problems with binary variables

Abstract Local search algorithms are frequently used to handle complex optimization problems involving binary decision variables. One way of implementing a local procedure is by using mixed-integer programming solver explore neighborhood defined through constraint that limits the number variables whose values allowed change in given iteration. Recognizing not all equally promising when searching for better neighboring solutions, we propose weighted iterated branching heuristic. This new differs from similar existing methods since it considers groups and associates with each group limit on can change. The weights indicate expected contribution flipping trying identify improving solutions current neighborhood. When fails an solution iteration, proposed heuristic may force into regions space utilizing least flip. tested benchmark instances optimum satisfiability problem, computational results show method superior alternative without weights.