Is my approach wrong?

#	User	Rating
1	tourist	4009
2	jiangly	3831
3	Radewoosh	3646
4	jqdai0815	3620
4	Benq	3620
6	orzdevinwang	3529
7	ecnerwala	3446
8	Um_nik	3396
9	gamegame	3386
10	ksun48	3373

#	User	Contrib.
1	cry	164
1	maomao90	164
3	Um_nik	163
4	atcoder_official	160
5	-is-this-fft-	158
6	awoo	157
7	adamant	156
8	TheScrasse	154
8	nor	154
10	Dominater069	153

http://www.usaco.org/index.php?page=viewproblem2&cpid=597

I wrote multiple solutions for this problem:

O(n+nlogn^2) Solution - I wrote this slower solution to compare with my faster one

O(n+logn^3) Solution - My original solution

My approach: - Store values, multiply them by two. This avoid errors in dealing with floating-point values - Precompute prefix and suffix: at every point store the value of radius necessary to destroy all haybales to the left (prefix) and the right (suffix). - Binary Search on Size of initial Radius (of explosion) - Binary Search of location - upper_bound/lower_bound inside to check for location, and then check if possible location using prefix/suffix values

I have tried to debug this for about a week, I asked about 4 people, and I changed my method and I still get WA with binary search.

Some things that I tried: - Wrote slower code for less chance of error; this code produces same output - Changing upper and lower values of binary search to the only possible locations of haybales (I changed it back because this had no effect)

Some things I haven't tried: - Throw away prefix/suffix and check it in O(n) <-- the person who suggested this also acknowledged my prefix/suffix were correct

Should I completely changed my approach? I think this approach can go. If it is just another bug (hopefully), does anyone have an idea what the bug is?