A trust branching path heuristic for zero–one programming

For 0–1 problems, we propose an exact Branch and Bound procedure where branching strategy is based on empirical distribution of each variable within three intervals [0,ϵ],[ϵ,1-ϵ],[1-ϵ,1][0,ϵ],[ϵ,1-ϵ],[1-ϵ,1] under the linear relaxation model. We compare the strategy on multiknapsack and maximum clique problems with other heuristics.