4 6810 12 142 131 3 579 11

All the answers that are not of the form 6k+2 or 6k+3 can be very simple with property of rotational symmetry (Nick's output for 6 and 12), all the answers in the sample output of the problem in the problemset archive have rotational symmetry, even the output for 8, which is of the form 6k+2.subbu wrote:I am convinced that there is some general pattern for the case

n = 6K+2 or 6K+3 , seeing the very low runtimes.

I can't see it yet.

Try to generate all the different answers for 8, 14 and 20 that are rotational symmetric (this is relatively small subset of all the answers for 8, 14 and 20), look for simple answers (similar to these which are not of the form 6k+2 or 6k+3) and then you should find out some really simple pattern. Answers for 9, 15, 21, ... can be easily made from 8, 14, 20, ...