Branch-and-price for solving integer programs with a huge number of variables: Methods and applications

Abstract

Many interesting discrete optimization problems, including airline crew scheduling, vehicle routing and cutting stock, have “good” integer programming formulations that require a huge number of variables. Good means that the linear programming relaxations give tight bounds which implies small search trees. We will discuss classes of problems for which this type of formulation is desirable, and special (non-standard) methodology that is needed to solve these integer programs.