This grant proposes to explore the use of approximate dynamic programming (ADP) and simulation optimization (SO) methods, both separately and in conjunction with each other, to assist in a number of applications, primarily in the emergency services. The fields of ADP and SO have developed separately to address closely related problems, and so there is great potential to strengthen the links between the two areas. The proposed work would develop those links, tackle the large-scale computing issues that arise, and work on challenging new modeling questions that have come up as a consequence of the PIs? collaborations with emergency service providers.
The PIs have already had some success using ADP to devise a relocation scheme for stylized models of ambulance operations, whereby ambulances are moved in real time in anticipation of future demand. A short-term goal is to move closer to full implementation through existing close relationships with a provider of software for emergency-service planning. In the longer term, any links between ADP and SO will likely lead to effective computational methods for making real-time decisions in complex systems that are subject to significant uncertainty. These methods will find applications not only in the emergency services, but also in transportation logistics, supply chain management, revenue management and potentially many other fields.
