The most frequent approach to test the performance of a new algorithm for a robust optimization problem is to generate random instances. Even if real-world instances are available, it is benefcial to use random instances in addition, so that the high number of test instances allow for a statistically more meaningful analysis. The usual approach here is to choose parameter values uniformly at random from a given interval. Therefore, randomly uniform (RU) instances for the well-known Traveling Salesman Problem (TSP) under discrete uncertainty set can be found on this page.
To better understand the instance les, n is the number of edges in a complete graph, and hence √n is the number of nodes, also the number of scenarios is set to N = √n. In all instance files, there exists N + 1 lines. The rst line consists of two numbers, n and N, respectively. The following N lines which contain n number, represent dierent discrete scenarios. Hence, each number of these N lines is a cost of the correspondent edge in the related scenario. 100 random instances for each edge size equal to 100, 144 and 196 are generated.
The information on this page has been created based on the paper “Benchmarking Problems for Robust Discrete Optimization” by Dr. Marc Goerigk (Network and Data Science Management, University of Siegen, Germany) and Mohammad Khosravi (Network and Data Science Management, University of Siegen, Germany).