Let the Markov decision process of Figure $1,$ whose transitions are labeled by the names of the actions

and the probabilities of transitions; the states are labeled by the corresponding rewards $($expressing a utility

function$)$