Consider a discounted MDP with a finite state space and a bounded reward function. Let and & be two stationa (z) (z), otherwise. Show that r(r,o) max(v(z, ), v(z, t)) for each z e x