Information Relaxations and Duality in Stochastic Dynamic Programs

We describe a dual approach to stochastic dynamic programming: we relax the constraint that the chosen policy must be temporally feasible and impose a penalty that punishes violations of temporal feasibility. We describe the theory underlying this dual approach and demonstrate its use in dynamic programming models related to inventory control, option pricing, and oil exploration.


Working paper, Fuqua School of Business, Duke University.