rounded

Correctly-rounded arbitrary-precision floating-point arithmetic.

This package provides numeric instances for MPFR that use "Implicit Configurations" from http://www.cs.rutgers.edu/~ccshan/prepose/prepose.pdf to choose a Rounding and Precision. For those that do not want to use reflection, explicit instances are provided for common precisions and for the built-in rounding modes.

This package should work correctly with GHC 8.4.4 or later.

>>> import Numeric.Rounded
>>> :set -XDataKinds
>>> exp pi :: Rounded TowardZero 512
23.140692632779269005729086367948547380266106242600211993445046409524342350690452783516971997067549219675952704801087773144428044414693835844717445879609842

rounded version 1.x is for MPFR version 4.0 and above.