Large Deviation Limit for Discrete-Time, Totally Observed Stochastic Control Problems with Multiplicative Cost

摘要

We show that a large class of discrete-time dynamic games can be obtained as a limit of stochastic control problems with multiplicative cost. Our approach consists in analyzing the large deviation properties of the Markov kernels associated with the stochastic dynamics, and allows us to give a unitary treatment of several nonlinear models.