You can read it here: http://apps.topcoder.com/wiki/display/tc/TCO+2013+Round+2A#ThePowers
The problem is easy to define. Count the total number of results for X^Y (X raised to the Y-th) where (1 <= X <= A) and (1 <= X <= B), both A and B are at most 1000000000.
I found it very difficult, (and also the other problems turned out to be very difficult to me) hence why I took that long (I also had some anxiety issues related to health (again) but nothing big). The explained editorial is the intended solution. It appears that there are solutions out there that some people might find "easier". I am not sure, they look hard to read to me. I think that although the solutions might be simpler, they might be harder to explain. Although this one explanation turned out to be much longer than I expected.
The rest of the editorial is coming today. I must first participate in SRM 575. Someone managed to beat me by a week and put an alternative (to nothing) explanation for the "easy" problem, so you can read that. I am going to add my explanation for the greedy approach (lovable one). The medium problem is definitely an anti-vexorian mine. I tend to have amnesia in relation to linear equations problems.