Peer review

[giogenna16]

Hi, Overall, I think that your algorithm works well and seems to be able to reach the optimum in a reasonable time for small N: the Dijkstra's goal have been achieved. The readme file is very concise, but, if you analyze the code at the same time, it is useful to understand better; however, doing it slightly more detailed can be even more helpful. I think that the definition of the cost is clever: analyzing the performances, the algorithm seems to cross less nodes/states with these costs with respect to, for example, a cost that considers only the repetitions (without normalizing and setting the repetitions equal to sys.float_info.min if they are equal to zero). I believe that there is a small contradiction in the definition of the nodes and states between the code and the readme file: for instance, for n=5, you print: "Found a solution for N=5 with 3 processed states, with 50 nodes added in the frontier and with 0 repetitions." and, in the readme file, you say that it passes through 3 nodes I like a lot the formulation of the state as a subset of the indeces of the list of lists.

[FlavioPatti]

Thanks a lot, I'll keep in mind for next time:)