Solving the market split problem via branch-and-cut (Q2655892)

Summary: In 1999 some researchers put forth some small but extremely difficult 0/1 problems derived from the market split (market sharing) problem originally proposed by a researcher in 1978. These problems, offered as a challenge to the optimisation community, motivated a series of research efforts trying several competing methodologies. As part of their testing, they employed both branch-and-bound and branch-and-cut algorithms and reported that branch-and-bound performed poorly and branch-and-cut did even worse. Based on this result they concluded that `standard approaches to integer programming are not well suited for this class of problems'. Other researchers offered similar assessments. In this note we demonstrate that current state of the art branch-and-cut methods, unavailable earlier, is able to solve these problems to optimality. Then we also solve 40 numerical problems to optimality and also find that it works quite nicely on even larger problem instances.