Об уточнении лагранжевых двойственных оценок в бинарных и булевых квадратичных задачах

Abstract

The approach for improvement of dual lagrangian bounds in quadratic optimization problems with binary (±1) and boolean (0 −1) variables is considered. It is based on use of families superfluous constraints in form of equality, which for these problems can be constructed as a result of introduction new variable in the form of products already existing variable. Is shown, that the introduction of these constraints improves accuracy of lagrangian dual bounds problem.