Composite version of B&B algorithm: experimental verification of the efficiency

Abstract

Abstract The aim of this paper is to verify the efficiency for a composite version of branch and bound algorithm combining B&B branching strategy and bounds computing with the pruning rules used in dynamic programming for selecting of direction of movement by a search tree. As criteria of efficiency we used random access memory (RAM) and time gains of finding integer programming problem solution by a composite version of branch and bound algorithm in comparison with its traditional implementation. The number of variables in the problems and restrictions did not exceed 25 and 15, respectively. Experimental verification have shown that for solving time used as a criterion of the efficiency, the advantages of composite version are beyond doubt compared to the used RAM value as criterion often taking the side of the classical B&B algorithm.