The intended solution is to implement the ceiling division. Using double and built-in ceiling functions isn't accurate enough due to floating-point precision errors (unless you use ~128~-bit floating point). Division with two integers using the built-in division operator is floor division (e.g. ~5/2 = 2~). To implement ceiling division with integers, you can simply use ~\lceil \frac{x}{y} \rceil = \lfloor \frac{x+y-1}{y} \rfloor~ under the given constraints.

Complexity: ~\mathcal O(T)~