Some new results on the ergodic control of partially observed Markov chains

We derive the dynamic programming equations for the ergodic control problem of partially observed Markov chains with finite state space and compact action space under a rather weak condition imposed on the optimal discounted policies.