Across manufacturing and service industries, organizations face a common and costly challenge: how to allocate limited resources in complex, ever-changing conditions. Classic mathematical tools for solving allocation problems have long run into a wall known as the “curse of dimensionality,” where the sheer number of scenarios and courses of action makes identifying efficient resource allocation guidelines practically impossible at scale. As a result, many organizations rely on models that fail to capture real-world complexity, hindering opportunities for meaningful efficiency improvements and raising operational costs.

This is why Professor Antoine Sauré has been awarded a Natural Sciences and Engineering Research Council Discovery Grant for his research program titled “Advanced Methods for Identifying Enhanced Dynamic Resource Allocation Policies.”
Recent advances in approximate dynamic programming (ADP) have begun to chip away at these limitations by using approximation techniques to make otherwise intractable problems solvable. However, three critical challenges remain: Approximation models must be rich enough to faithfully represent the true costs and consequences of different decisions; They must not overcomplicate the problem or the solution; They must scale to the large, complex settings where the need is greatest. Closing these gaps requires a new generation of methods, just what this research sets out to build.
Bridging the gap between theory and real-world complexity
Professor Sauré’s research program will seek to develop methods combining data-driven prediction tools with mathematical programming models and computer simulation, to understand how different decisions affect overall system performance. His research will use approximation architectures inspired by machine learning, including neural networks. These architectures can capture the complexity of real-world problems while allowing researchers to solve them efficiently. For settings where even these methods strain under the computational load, the research program will deploy decomposition strategies, closed-form solutions for simplified cases and parallel computing to reduce solution times.
The research is being developed in collaboration with healthcare organizations, with immediate applications including combined inter-day and intra-day appointment scheduling, clinic and operating room planning, patient no-show and cancellation management, and ambulance relocation.
Why this research matters
This project addresses a long-standing challenge: identifying decision-making guidelines that are both theoretically sound and practical to use on a large scale. It will explore how predictive analytics, mathematical programming and computer simulation can work together. This includes situations involving multiple process stages, limited or incompatible resources, uncertainty in resource consumption and availability, and different geographical locations. This research program could also help researchers develop better approaches to managing resources and making decisions in a wide range of fields.
The payoff is concrete and far-reaching. In healthcare, better resource allocation translates into shorter patient wait times, fewer wasted appointments, and more responsive emergency services. The methods developed will also be adaptable to other sectors, including warehouse management, budget planning and production planning. At a time when public and private organizations are under pressure to do more with less, this research offers a rigorous, practical path toward smarter resource use and cost savings for Canadians.

