Printing floating-point numbers: a faster, always correct method.

Marc Andrysco, Ranjit Jhala, Sorin Lerner

Browse the full POPL paper archive.