Denumerable state semi-Markov decision processes with unbounded costs, average cost criterion

摘要: Abstract This paper establishes a rather complete optimality theory for the average cost semi-Markov decision model with denumerable state space, compact metric action sets and unbounded one-step costs case where underlying Markov chains have single ergotic set. Under condition which, roughly speaking, requires existence of finite set such that supremum over all stationary policies expected time total absolute incurred until first return to this are any starting state, we shall verify solution equation an optimal policy.