Generic classes for floating point types. The interface is loosely based off of the C math library.

- data FPClassification
- = FPInfinite
- | FPNaN
- | FPNormal
- | FPSubNormal
- | FPZero

- class (Fractional a, Poset a) => Roundable a where
- class Fractional a => Floating a where
- class Floating a => RealFloat a where
- fma :: a -> a -> a -> a
- copysign :: a -> a -> a
- nextafter :: a -> a -> a
- atan2 :: a -> a -> a
- fmod :: a -> a -> a
- frem :: a -> a -> a
- hypot :: a -> a -> a
- cbrt :: a -> a
- exp2 :: a -> a
- expm1 :: a -> a
- log10 :: a -> a
- log1p :: a -> a
- log2 :: a -> a
- logb :: a -> a
- erf :: a -> a
- erfc :: a -> a
- gamma :: a -> a
- lgamma :: a -> a
- rint :: a -> a
- nearbyint :: a -> a
- infinity :: a
- nan :: a
- pi :: a

- class (Roundable a, RealFloat a) => PrimFloat a where
- classify :: a -> FPClassification

# Documentation

data FPClassification Source

Classification of floating point values.

class (Fractional a, Poset a) => Roundable a whereSource

Class for types which can be rounded to integers. The rounding functions in the Prelude are inadequate for floating point because they shoehorn their results into an integral type.

Minimal complete definition: `toIntegral`

and `round`

.

toIntegral :: Integral b => a -> Maybe bSource

Discards the fractional component from a value. Results in `Nothing`

if the result cannot be represented as an integer, such as if the input
is infinite or NaN.

class Fractional a => Floating a whereSource

Class for floating point types (real or complex-valued).

Minimal complete definition: everything.

class Floating a => RealFloat a whereSource

Class for real-valued floating point types.

Minimal complete definition: all except `pi`

, `infinity`

and `nan`

.

Fused multiply-add.

`copysign x y`

computes a value with the magnitude of `x`

but the sign
of `y`

.

nextafter :: a -> a -> aSource

`nextafter x y`

computes the next representable value after `x`

in the
direction of `y`

.

`atan2 y x`

computes the principal value of the arctangent of `y/x`

.
The signs of the input determine the quadrant of the result.

`fmod x y`

computes `x - n*y`

, where `n`

is the integral quotient of
`x/y`

, rounded towards zero.

`frem x y`

computes `x - n*y`

, where `n`

is the integral quotient of
`x/y`

, rounded to the nearest integer, with halfway values rounded to
even.

Euclidean distance function without undue overflow.

Cube root.

Base-2 exponential function.

Computes `exp x - 1`

without undue cancellation.

Base-10 logarithm function.

Computes `log (x + 1)`

without undue cancellation.

Base-2 logarithm function.

Extracts the exponent of a floating point value. If the value is subnormal, the result is as if the value were normalized.

Error function.

Complementary error function.

Gamma function.

Log gamma function.

Round to the nearest integer according to the current rounding
direction. The default rounding direction is towards the nearest
integer with halfway values rounded to even. If the resulting value
differs from the input, the `Inexact`

exception is raised.

Same as `rint`

, except that the `Inexact`

exception is not raised.