Logic-based decomposition methods for the travelling purchaser problem

Abstract

We present novel branch-and-check and logic-based Benders decomposition techniques for the Travelling Purchaser Problem, an important optimization problem with applications in vehicle routing, logistics, and warehouse management. Our master problem determines a set of markets and directed travel arcs that satisfy product purchase constraints with relaxed travel costs. Our subproblem identifies subtours within this master assignment and produces a set of generalized subtour elimination cuts. We show that the proposed technique demonstrates strong performance on the asymmetric problem variants, finding optimal solutions to previously unsolved instances, while performing competitively on a number of symmetric problem classes. Furthermore, our model is implemented unchanged for the four problem variants whereas other state-of-the-art approaches are variant-specific.