A Novel Strategy of Combining Variable Ordering Heuristics for Constraint Satisfaction Problems

Variable ordering heuristic plays a central role in solving constraint satisfaction problems. Many heuristics have been proposed and well-studied. In order to take advantage of the fact that many generic variable ordering heuristics work well for different problems, we propose a novel method in this paper, namely ParetoHeu, to combine variable ordering heuristics. At each node of the search tree, a set of candidate variables is generated by a new strategy based on Pareto optimality and a variable is selected from the set randomly. The method is easy to be implemented in constraint solvers. The experiments on various benchmark problems show that ParetoHeu is more efficient than both the participant heuristics which are popular in constraint solvers. It is also more robust than some classical strategies which have been used to combine variable ordering heuristics.