Short-Term Scientific Mission

Development of Open-Source Libraries for Extending Existing Optimization Problem Modelling Frameworks Using University Timetabling as a Case Study

Main theme: Problem Modelling and User Experience
Grantee: Ahmad Mahir Sadaldin Othman, University of Southern Denmark, Odense, Denmark
Host: Luca Di Gaspero, University of Udine, Udine, Italy

Start date: 2024-06-24
End date: 2024-07-11
Awarded: 2024-06-11
Report approved: 2024-08-29

Description

The aim of the STSM is to evolve the Python black-box optimization API used in the SIGEVO Summer School of 2023. The API will be extended to other metaheuristics, particularly those that use multiple neighborhoods, which will require formalizing the concept of a neighborhood. Additionally, support for multi-objective optimization, such as lexicographic or Pareto, is planned to be added, facilitating the modeling of more complex problems. These features will be added as new classes and methods external to the API. For a practical application of these changes, a real-life multi-objective problem arising in University Timetabling at the University of Southern Denmark will be used as a case study.

Achievements

The STSM enhanced the API to provide better support for multi-neighborhood approaches and multi-objective optimization. A Python module was developed that adds various new classes and methods to support these features, complementing the original API and facilitating its use on more complex problems. The resulting work is accompanied by a tutorial available at https://roarnet-ext.github.io/, which showcases an example of using the features to solve a simplified version of the Educational Timetabling problem at the University of Southern Denmark. This work will be further developed as part of Ahmad’s Master Thesis, which is due by the end of December 2024.