In this talk, I will address dynamic scheduling problems for queueing networks, which have applications in various industries such as call centers, semiconductor manufacturing, and data centers. I will begin by discussing formal approximations of these problems within both the Halfin-Whitt and conventional heavy-traffic regimes. The Halfin-Whitt regime leads to high-dimensional drift rate control problems, while the conventional heavy-traffic regime results in singular control problems.

To develop effective solutions for these control problems, I will present a computational method that builds on the approach of Han et al. (Proceedings of the National Academy of Sciences, 2018, 8505-8510). This method will be followed by a general technique for implementing the derived solutions within the original queueing networks, testing them against the best available benchmarks.

If time permits, I will also mention other operations, logistics and supply chain management applications that can be tackled similarly.

This talk is based on papers coauthored with Mike Harrison, Ebru Kasikaralar and Nian Si.

For further information please contact: erika.somma@unibocconi.it