Computational Logistics

The course provides a Python-based introduction into computational methods and tool-boxes in logistics. It will cover analytics concepts from optimization, simulation, machine learning, and metaheuristics for analyzing logistics problems and providing state-of-the-art decision support. After a brief introduction into the programming language Python, mathematical programming problems in logistics such as the transportation, the traveling salesman, vehicle routing and lot- sizing and scheduling problems are modeled and solved with Python-Gurobi. Advanced techniques to solve large-scale problems complement the basics. For the simulation of logistics systems, simple queuing and inventory systems will be addressed using simpy. Advanced simulation optimization and design of experiments concepts will be illustrated using logistics problems. For data-driven logistics methods, supervised and unsupervised machine learning algorithms using the toolbox sci-kit will be covered. When large and complex problems cannot be solved using exact methods, metaheuristics have gained large importance. The course will cover and implement basic concepts of local search- and population-based methods.

After participating in this module, students are able to understand the different concepts of heuristics and metaheuristics. They are able to implement these methods in a computer and analyze the results. Students further comprehend the weaknesses and strengths of the methods. They are able to apply the method to a practical problem in the area of logistics in the context of a project. Through the project report and the presentations, students further improve their skills of writing academic reports and carrying out discussions within a research environment. Students are able to integrate involved persons into the various tasks considering the group situation. Furthermore the students conduct solution processes through their constructive and conceptual acting in a team. The course will prepare the students for their master thesis.

Grading is based on homework assignments (25%) and an individually graded final group term project (75%). Students work in groups on the project. In addition to implementing the model, students are required to conduct numerical analyses, which will be submitted in digital form. At the end of the module, students present their work in a final presentation and participate in the discussion of the projects of their fellow participants. The grade is determined by a final presentation, including a discussion of the results on individual level.

The module requires a solid knowledge in linear optimization. Basic knowledge of a programming language and the course “Modelling, Optimization, and Simulation” are required.

In lectures, students learn to understand the theory of heuristics, metaheuristics, and matheuristics. In the integrated exercise sessions, students learn the use and the implementation of these algorithms. For solving the exercises, they are provided with the necessary software, such as Java or Xpress. In addition, student groups will work on a group project throughout the semester, which they will present in a final presentation and a report.