Abstract:We study a two-stage natural disaster management problem modeled as a stochastic program, where the first stage consists of a facility location problem, deciding where to open facilities and pre-allocate resources such as medical and food kits, and the second stage is a fixed-charge transportation problem, routing resources to affected areas after observing a disaster. Our model has binary variables present in both stages. Due to the lack of data, classical stochastic programming approaches may be ill-suited, and we propose a two-stage distributionally robust formulation with a Wasserstein ambiguity set, where we consider distributions consistent with historical data and a tunable parameter to control the level of risk aversion. We develop a tailored column-and-constraint generation (CCG) algorithm to solve an extensive reformulation, where scenarios are iteratively generated. We handle the presence of binary variables in the second stage by leveraging the structure of our support set and second-stage problem, and provide conditions under which the optimal value of the latter is concave with respect to the intensity of the disaster, leading to an efficient scenario generation procedure. We also show that our results extend to the case where the second stage is a fixed-charge network flow problem. We perform extensive computational experiments demonstrating the computational advantage of our method over classical CCG implementations on synthetic instances, and illustrate the benefits of our approach on a popular case study from the literature of hurricane threats on the Gulf of Mexico in the United States.

Data-Driven Distributionally Robust Risk-Averse Two-Stage Stochastic Linear Programming over Wasserstein Ball

Solving Wasserstein Robust Two-stage Stochastic Linear Programs Via Second-order Conic Programming

Second-Order Conic Programming Approach for Wasserstein Distributionally Robust Two-Stage Linear Programs

A study of distributionally robust mixed-integer programming with Wasserstein metric: on the value of incomplete data

Data-driven Multistage Distributionally Robust Linear Optimization with Nested Distance

A Wasserstein distributionally robust chance constrained programming approach for emergency medical system planning problem

Wasserstein Distributionally Robust Control Barrier Function using Conditional Value-at-Risk with Differentiable Convex Programming

Wasserstein distributionally robust optimization and its tractable regularization formulations

A Bayesian Risk Approach to Data-driven Stochastic Optimization: Formulations and Asymptotics

CVaR-Based Approximations of Wasserstein Distributionally Robust Chance Constraints with Application to Process Scheduling

A two-step approach to Wasserstein distributionally robust chance- and security-constrained dispatch

An inexact mixed risk-aversion two-stage stochastic programming model for water resources management under uncertainty

Wasserstein Distributionally Robust Inverse Multiobjective Optimization

Integer Programming Approaches for Distributionally Robust Chance Constraints with Adjustable Risks

Robust Wasserstein Optimization and its Application in Mean-CVaR

Distributionally robust disaster relief planning under the Wasserstein set

Distributionally Robust Optimization Under Moment Uncertainty with Application to Data-Driven Problems

Data-Driven Bayesian Nonparametric Wasserstein Distributionally Robust Optimization

Distributionally Robust Optimization with Correlated Data from Vector Autoregressive Processes

On Wasserstein Distributionally Robust Mean Semi-Absolute Deviation Portfolio Model: Robust Selection and Efficient Computation

Non-concave distributionally robust stochastic control in a discrete time finite horizon setting