Markov-achievable payoffs for finite-horizon decision models. (Q1965904)

The paper investigates a problem of the decision making theory. The considered question is if the mean of the given payoff function evaluated under a permissible control can be achieved or even overcame using a permissible Markov control. Assuming payoff function based on recent state and on the \(n\)-th earlier decisions and states, the authors derived properties which are sufficient and necessary for the examined task.