Earliest - finish - time first algorithm

Dynamic programming can be very confusing until you’ve used it a bunch of times, so the best way to learn it is to simply do a whole bunch of examples. One way of viewing it is as a much more complicated version of divide-and-conquer a la mergesort or quicksort. In those cases, we could divide the problem into two subproblems, solve it optimally on each subproblem, and then combine the solutions (in the case of mergesort by a merge, in the case of quicksort trivially). This is great when it works, but in some cases it’s not so simple. Dynamic programming is a way of rescuing the divide-and-conquer ideas from cases where it seems like it shouldn’t work – for example, if the subproblems overlap.