Short-Term Scientific Mission

Advancing ROA Education and Benchmarking

Main theme: Problem Modelling and User Experience
Grantee: Marco Chiarandini, University of Southern Denmark, Odense, Denmark
Host: Carlos Fonseca, University of Coimbra, Coimbra, Portugal

Start date: 2024-09-23
End date: 2024-09-28
Awarded: 2024-06-19
Report approved: 2024-11-06

Description

Research on randomized heuristic algorithms for optimization lacks standards to enhance development and benchmarking. This STSM focuses on developing a ROAR-NET API specification to create a language-agnostic interface for optimization problems and solvers, ensuring alignment, reusability, and extensibility. Another goal is to exchange teaching experiences and design curricula and materials for a one-week training school on randomized heuristics within ROAR-NET. Discussions will also address infrastructure for a real-time automated assessment system to support learning activities. Lastly, the STSM aims to acquire knowledge on preference handling and multiobjective optimization for applications such as course timetabling.

Achievements

During the STSM, Marco Chiarandini collaborated with Carlos Fonseca, Alexandre Jesus, and Luís Paquete on the planned tasks. Progress was made on the ROAR-NET API specification, resulting in a preliminary agreement on a format for documentation, the inclusion of facilities for multiple neighborhoods, and a tentative type organization including a measure of solution quality. Centered around the new specification, a plan for the First ROAR-NET Training School was outlined, including lectures, practical problem-solving sessions, and coding activities. Discussions on multiobjective optimization led to new insights on handling preferences and comparators. Overall, the STSM laid the groundwork for publishing the API specification and for the development of teaching materials. These directions have been pursued since the end of the STSM and are ongoing.