The Float
type is an arbitrary precision float. It supports
correctly rounded computation of arithmetic (+
, -
, *
, /
),
square roots and precision conversions. It makes no claims to high
performance, but does try to keep it in mind.
Built on ramp
, and implemented with
a lot of reference to [1].
[1]: Muller, Jean-Michel; Brisebarre, Nicolas; de Dinechin, Florent; Jeannerod, Claude-Pierre; Lefèvre, Vincent; Melquiond, Guillaume; Revol, Nathalie; Stehlé, Damien; Torres, Serge (2010). Handbook of Floating-Point Arithmetic. Birkhäuser. ISBN 978-0-8176-4705-6.