3 ms·
It is easy to track the real result of the computation with floating-point intervals: round down when computing the lower bound, and up when computing the upper
by pascal_cuoq 13y ago
It is easy to track the real result of the computation with floating-point intervals: round down when computing the lower bound, and up when computing the upper bound of each interval.