joint work with Shanshan Wang, Jinlin Li
We will present a distributionally robust chance constraint optimization model, where the probability distribution specifying the chance constraint is robustified using a Wasserstein ambiguity set. For this problem, we will present novel techniques for strengthening big-M type formulations, and techniques for generating inequalities for use within a branch-and-cut framework. Results based on experiments using real data from an operating room will be presented, where the problem under consideration is to assign surgeries to the operating room.
joint work with Hamed Rahimian, Guzin Bayraksan
Recent research has studied the notion of effective scenarios for static distributionally robust stochastic programs. Roughly speaking, a scenario is deemed effective if its removal changes the optimal value of the problem. In this presentation we discuss the extension of these ideas to the case of multistage stochastic programs. Such extension requires proper ways of defining the meaning of removing a scenario in a dynamic context. Computational and analytical results show that identifying such effective scenarios may provide useful insight on the underlying uncertainties of the problem.
joint work with Hedieh Ashrafi
We consider the management of projects under uncertainty in a firm's market entry, competitors' entry and product adoption dynamics. The company can adjust manpower staffing on a portfolio of projects (new products), which diversifies its portfolio but delays market entry for each product line, making it more likely a rival will launch a competing product first. We present a robust optimization approach to balance those concerns and extend our framework to dynamic policies based on information revealed over time about competitors' products entering the market.