Thanks for letting me know; I realized I had a bug in the test which caused it to terminate early. The shortest solution is indeed unique, and the tester should now reject other solutions as being too long (or not all-resistant).

Done. The function now takes in the number of ranks as an additional argument.

Thanks, an oversight. I've now translated the local clojure tests into python.

Oh, good catch --- fixed.

Sure, I've been considering generalizations that would avoid this problem. One possibility would be to ask for a function of two variables, k=the number of triples, and n=the number of copies of each rank in the deck (or m=the number of available ranks). I want to make sure I don't make it too obscure, though.

When I submit the default boilerplate code in Clojure, it throws an error:
Exception in thread "main" java.io.FileNotFoundException: Could not locate math_parser__init.class or math_parser.clj on classpath.