Moment-Based Distributionally Robust Optimization
摘要
The idea of distributionally robust optimization (DRO) originated in the middle of the 20th century, where the ambiguity set is built with first- and second-order moments constraints. Moment-based DRO not only has the longest history among different DRO techniques but also is one of the most popular and mature ways to model the inexactness of empirical probability distributionProbability distributionempirical probability distribution for uncertainty-integrated optimization problems. This chapter focuses on the moment-based DRO, typical moment-based ambiguity sets are summarized, the reformulation methods are explained, and finally, the moment-based DRO incorporating unimodality is also introduced.