Journal of the Operations Research Society of China ›› 2019, Vol. 7 ›› Issue (1): 69-106.doi: 10.1007/s40305-018-00238-5

所属专题: Continuous Optimization

• • 上一篇    下一篇

  

  • 收稿日期:2018-05-16 修回日期:2018-11-26 出版日期:2019-03-30 发布日期:2019-03-30

Distributions with Maximum Spread Subject to Wasserstein Distance Constraints

John Gunnar Carlsson, Ye Wang   

  1. University of Southern California, Los Angeles, CA 90089-4017, USA
  • Received:2018-05-16 Revised:2018-11-26 Online:2019-03-30 Published:2019-03-30
  • Contact: John Gunnar Carlsson, Ye Wang E-mail:jcarlsso@usc.edu;wang141@usc.edu

Abstract: Recent research on formulating and solving distributionally robust optimization problems has seen many different approaches for describing one's ambiguity set, such as constraints on first and second moments or quantiles. In this paper, we use the Wasserstein distance to characterize the ambiguity set of distributions, which allows us to circumvent common overestimation that arises when other procedures are used, such as fixing the center of mass and the covariance matrix of the distribution. In particular, we derive closed-form expressions for distributions that are as "spread out" as possible, and apply our result to a problem in multi-vehicle coordination.

Key words: Distributionally robust optimization, Surveillance, Districting