Martina Gancarova and Michael Todd,
School of Operations Research and Information Engineering, Cornell University
We describe a general result indicating that the loss in objective value is not great when optimizing a linear function over a compact convex set, when the solution evaluated instead optimizes a closely related linear objective.