Johan Van Leeuwaarden We develop an exact solution framework for a broad class of Distributionally Robust Optimization (DRO) problems with general uncertainty structure. Within the class of moment- and confidence-set-based ambiguity sets, existing exact methods are largely limited to max-of-affine functions under ambiguity sets with strictly nested confidence sets. To enlarge this scope while preserving tractability, we introduce … Read more
The Appointment Scheduling Problem (ASP) involves scheduling a finite number of customers with uncertain service times, served consecutively by a single server with the goal of minimizing the weighted costs of waiting time, idle time, and overtime, which can be written as a sum-of-max linear functions. We introduce a novel robust optimization approach that considers … Read more
Chebyshev’s inequality provides an upper bound on the tail probability of a random variable based on its mean and variance. While tight, the inequality has been criticized for only being attained by pathological distributions that abuse the unboundedness of the underlying support and are not considered realistic in many applications. We provide alternative tight lower … Read more