Séminaire en format hybride au local 4488 du GERAD ou Zoom.

Personalized shift scheduling problems, such as nurse scheduling, are considered challenging because they typically require a personalized schedule for each employee. This talk will explore different aspects of this problem. It will show how to model several generic (soft and hard) constraints, such as limiting the number of consecutive assignments or prohibiting a particular pattern. Then, we will look at two different decompositions that are embedded in a branch-and-price algorithm, as well as at many speed-up techniques. The performance of the proposed approach is demonstrated by computational results on several nurse scheduling benchmarks. Finally, a short demonstration of the developed open-source software will be carried out.