It is not published yet, but I finished it last night and it is in a state that is sort of ok: http://apps.topcoder.com/wiki/display/tc/SRM+569
Remember when I said that I liked the mathy div1 1000 points problem of this match? I predicted that although difficult, it would have a solution that would not be impossible to explain, so that was good by me.
It turns out this problem is the coolest thing I have solved in months. I just really like this twist on the usual matrix exponentiation idea. This problem does not use matrix exponentiation, but it adapts that idea into something (that at least to me is) fresh.