Average optimal stationary policies: convexity and convergence conditions in linear stochastic control systems

This paper provides a set of conditions for the existence of an optimal stationary policy in the long-run average cost control problem of linear stochastic systems. The main conditions are based on convexity of the cost by stage and convergence of trajectories. The discrete-time system is assumed to be linear with respect to the state but the controls take an abstract state-feedback structure, possibly a nonlinear one. An application is considered to illus- trate the derived theory.