How can a dynamic programming solution for the unbounded knapsack decision problem be used to come up with a dynamic programming solution to the subset-sum decision problem? This limitation seems to render the unbounded knapsack problem useless.
In the unbounded knapsack, we simply store true or false for if some subset of integers sum up to our target value. However, if we have a limit on the frequency of the use of these integers, the optimal substructure at least appears to fail. How can this be done?