A General Framework for Dynamic Programming Equation and Time Consistency
Résumé
Time consistency is an important property for a sequence of stochastic optimization problem, especially with regard to risk attitudes. It is known that a sequence of problem is time-consistent if a Dynamic Programming equation holds. In this paper we present a general framework for Dynamic Programming where general aggregators in time and uncertainties are considered instead of the expectation and sum. The key assumptions are montonicity, commutation property, and decomposability of aggregators.
Origine : Fichiers produits par l'(les) auteur(s)