Introduction To Robust Optimization

Robust optimization models are used to address optimization problems with uncertain data. These models enforce the condition that a feasible solution should satisfy all constraints under any possible realization of constraint data. Furthermore, the optimization process should provide a worst-case guarantee on the quality of the solution, for any possible realization of objective data. Thus, for a minimization problem, the objective minimizes the maximum objective that could result from uncertain data. This article provides an overview of robust optimization models and properties, with an emphasis on robust linear optimization. Keywords: linear optimization; robust linear programming; budgeted uncertainty; Stackelberg game; polyhedral uncertainty