Two-Stage Optimization Problems with Multivariate Stochastic Order Constraints

Darinka Dentcheva, Eli Wolfhagen

Research output: Contribution to journalArticlepeer-review

9 Scopus citations

Abstract

We propose a two-stage risk-averse stochastic optimization problem with a stochastic-order constraint on a vector-valued function of the second-stage decisions. This model is motivated by a multiobjective second-stage problem. We formulate optimality conditions for the problem and analyse the Lagrangian relaxation of the order constraint. We propose two decomposition methods to solve the problems and prove their convergence. The methods are based on Lagrangian relaxation of the order constraints and on a construction of successive risk-neutral two-stage problems. Additionally, we propose a new combinatorial method for verification of the multivariate order relation, which is a key part of both methods. We analyse a supply chain problem using our model and we apply our methods to solve the optimization problem. Numerical results confirm the efficiency of the proposed methods.

Original languageEnglish
Pages (from-to)1-22
Number of pages22
JournalMathematics of Operations Research
Volume41
Issue number1
DOIs
StatePublished - Feb 2016

Keywords

  • Bundle method
  • Increasing convex order
  • Inverse cover inequality
  • Multiobjective optimization
  • Stochastic dominance
  • Trust-region method

Fingerprint

Dive into the research topics of 'Two-Stage Optimization Problems with Multivariate Stochastic Order Constraints'. Together they form a unique fingerprint.

Cite this