Discrete optimisation with polynomially detectable boundaries and restricted level sets

Abstract

The paper describes an optimization procedure for a class of discrete optimization problems which is defined by certain properties of the boundary of the feasible region and level sets of the objective function. It is shown that these properties are possessed, for example, by various scheduling problems, including a number of well-known NP-hard problems which play an important role in scheduling theory. For an important particular case the presented optimization procedure is compared with a version of the branch-and-bound algorithm by means of computational experiments