2-Period Balanced Travelling Salesman Problem: a polynomially solvable case and heuristics

TL;DR

This paper investigates the 2-period balanced TSP, identifying a polynomially solvable case with Kalmanson matrices and proposing effective heuristics for the general case, validated through computational experiments.

Contribution

It introduces a polynomial-time solution for the 2-period balanced TSP with Kalmanson matrices and develops new heuristics for the general problem, demonstrating strong computational performance.

Findings

Polynomial solution for Kalmanson matrix case

Two new heuristics outperform existing methods

Heuristics show strong results on benchmark instances

Abstract

We consider the NP-hard 2-period balanced travelling salesman problem. In this problem the salesman needs to visit a set of customers in two time periods. A given subset of the customers has to be visited in both periods while the rest of the customers need to be visited only once, in any of the two periods. Moreover, it is required that the number of customers visited in the first period does not differ from the respective number in the second period by more than p, where p is a given small constant. The objective is to find tours for visiting the customers in the two periods with minimal total length. We show that in the case when the underlying distance matrix is a Kalmanson matrix, the problem can be solved in polynomial time. For the general case, we propose two new heuristics. A combination of these two heuristics has shown very favourable results in computational experiments on benchmark instances from the literature.