The eternal quest for floating-point precision – I wonder if anyone's been there done that already. Guess not, a paper on it. https://www.reddit.com/user/self