Classes
struct	FloatingULP
	Compare floating point or complex values within a given ULP tolerance. More...

struct	FloatingULP< T >
	ULP comparator for complex numbers over floating point. More...

Concepts
concept	IeeeBinaryReal
	Real scalar stored in an IEEE binary32, binary64, or binary128 interchange representation.

concept	UlpOrderedReal
	Real scalar with an implemented ordering of adjacent representable values.

concept	UlpComparable
	Concept satisfied if `T` can be compared in ULPs.

Functions
template<UlpOrderedReal T>
long long	float_abs_distance (T a, T b)
	Return the absolute diagnostic ULP distance between two real scalars.

template<uni20::Complex T> requires UlpOrderedReal<uni20::make_real_t<T>>
long long	float_abs_distance (T a, T b)

template<UlpOrderedReal T>
long long	float_distance (T a, T b)
	Return the signed distance in ULPs between two supported real scalars.

Function Documentation

◆ float_abs_distance() [1/2]

template<UlpOrderedReal T>

long long uni20::check::float_abs_distance	(	T	a,
		T	b
	)

inline

Return the absolute diagnostic ULP distance between two real scalars.

template<uni20::Complex T>
requires UlpOrderedReal<uni20::make_real_t<T>>

long long uni20::check::float_abs_distance	(	T	a,
		T	b
	)

inline

template<UlpOrderedReal T>

long long uni20::check::float_distance	(	T	a,
		T	b
	)

inline

Return the signed distance in ULPs between two supported real scalars.

Positive if b > a, negative if a > b.
Returns 0 if a == b (including +0 vs -0).
Returns max<long long> if either value is NaN or if infinities differ.
Saturates when the finite distance exceeds the signed diagnostic range.
Template Parameters

T IEEE binary interchange or native extended long double scalar type.