I think this is really cool. Thanks for posting this.

11/6 Finally got it. Sort of. It took about 80min to generate the answers for n=6, which I then hardcoded, so I'm wondering what the intended solution was.

============

EDIT: Well, I decided to test more for J, and I see that I clearly get things wrong, time to fix... (4 2 != 2 4)

I seem to be quite hasty in my conclusions.

============

I'm suspicious of the judge for problem J. I get WA on test case 10, and through testing the judge, this is the first case which relies on the modulus. I've changed the modulus and still get AC up to test case 10, so I suspect that the judge code used a different modulus than the problem statement.

Alternatively, it could be some small case that I get wrong regardless, but I've verified that my program correctly solves n=1, for any m, and perhaps also for n=2, any m.

EDIT: Turns out that I must have added some inefficiency with the way I purchased fuel. I reimplemented right now, worked fine.

Original: Looks like the time limit is too strict for Java on F. While not 100% certain my algorithm is optimal, based off of the problem set analysis posted by Xelios, I have the intended algorithm...