On 03/01/06, Cale Gibbard
I managed to do better with the following program which gets the following time report on my machine real 0m8.175s user 0m7.742s sys 0m0.186s as opposed to real 0m23.232s user 0m21.115s sys 0m0.077s for the shootout code.
I didn't try too hard to optimise it heavily, but it does use a tree-based permutation generator I stole from some code which was in an n-queens solution I had laying around (pretty sure it's not mine), and an obvious memoisation hack when handling the flips.
Hmm, do the permutations have to be in their specific order? This permutation generator seems to go through them in a somewhat different order. It seems irrelevant to the problem, but since they want the permutations as part of the output, it's a good question. :) In that case, I wonder if it would be best to use some other generator to print the first 30, then switch to some faster generator for the actual computation. :) - Cale