This state of the art means that practitioners often have to develop their
own DP software. This could be a very rewarding but not necessarily easy
endeavour. In fact, there are anecdotal evidence that this could be quite a
tricky task. One reason for this is that the performance of some DP algorithms
can be sped-up significantly using suitable data structures. Dijkstra’s Algorithm
is a good example (e.g. Denardo, 2003).