Decision Programming: A Stochastic Optimization Framework for Multi-Stage Decision Problems

The project will further develop the decision programming framework as a methodology for modelling and solving multi-stage decision problems under uncertainty. These developments will make it possible to address a range of general decision problems in which existing methods face issues in terms of problem representation and computational tractability. In specific, we will develop modelling approaches and solution strategies, investigating how to incorporate multiple objectives, risk and ambiguity considerations as well as continuous decisions; and how to develop parallelizable solution methods based on decomposition techniques. The outputs of this project offers researchers and practitioners a general modelling approach for addressing challenging decision problems, including those encoungered in diagnostic testing in healthcare, selection of risk mitigation actions for safety-critical systems and cost-benefit analyses for climate change mitigation strategies, among others.