On the effect of adding nodes to TSP instances: An empirical analysis

Abstract

Our human society is experiencing complex problems nowadays, which require large amounts of computing resources, fast algorithms and efficient implementations. These real-world problems generate new instances for the classical, academic problems as well as new data collections that can be used for assessing the available solving packages. This paper focuses on the Traveling Salesman Problem, which is one of the most studied combinatorial optimization problems, with many variants and broad applications. In order to allow a smooth integration with the current Geographic Information Systems (GIS) technologies, the instances described in this work are specified by geographic coordinates, and they use the orthodromic distance. A sequence of similar instances is defined, and the characteristics of the state-of-the-art exact solver results on these instances are presented and discussed.