Apisak Vittayaprapakorn1 and Peerayuth Charnsethikul2
Abstract
The research presents a new mathematical model for solving the classical traveling salesman problem with possible extensions to case of multiple salesmen. The resulting linear integer programming model consists of 2n2 decision variables and n2+n constraints which has a smaller size as compared to other model previously proposed. This proposed model is applied to solve a set of large TSP problems including the case of 76 cities in Thailand using the state of the art Excel-Solver-Gurobi. Our preliminary results indicate efficiency advantages as the problem size grows.