Fig. 6. ADP solves sequences of smaller subproblems. (Left) A typical space–time network used in dynamic resource allocation. (Right) Some smaller subproblems. We need
an approximation to capture the impact of present decisions on the future (Powell, 2008).