Small parameter limit for ergodic, discrete-time, partially observed, risk-sensitive control problems (Q5936216)

The paper studies the small noise limit for discrete-time, partially observed, risk-sensitive control problems over an infinite horizon. Using large deviations techniques it is shown that the limit of the value functions of such problems is given by the value function of a deterministic dynamic game and that the convergence is uniform on compact subsets. A key role in the developement of the results is played by the definition of the value functions of partially observed problems as mappings acting on a space of sequences (information vectors) rather than on a space of measures (information states).