Abstract
Taking a subset of the constraints of a general mixed integer linear program up into the objective function in a Lagrangean fashion (with fixed multipliers) yields what the author calls a Lagrangean relaxation of the original program. The paper gives a reasonably comprehensive development of the use of this simple device in the context of branch- and-bound. The selective application of these ideas can yield significant improvements in performance for special classes of problems.

This publication has 0 references indexed in Scilit: