Improved heuristics for the traveling purchaser problem

For the Traveling Purchaser Problem we describe improvement subprocedures that can be incorporated in known heuristics. Computational results show the effectiveness of these subprocedures in general savings, commodity adding and tour reduction heuristics for test instances up to size 200 and four different cost types, including both symmetric and non symmetric costs as well as coordinate based costs. The average improvement of the solutions is substantial, i.e., from three to more than seven times smaller gaps for the best heuristic, based on savings.