Given an MDP and a particular time step t

of a task $($continuing or episodic$),$ the $-$return,