MathUtils Namespace Reference

Math utility class. Note that the test() routine should return true for all implementations. More...

Functions
int	round_int (double x)
	Round to nearest integer. This routine does fast rounding to the nearest integer. In the case (k + 0.5 for any integer k) we round up to k+1, and in all other instances we should return the nearest integer. Thus, { -1.5, -0.5, 0.5, 1.5 } is rounded to { -1, 0, 1, 2 }. It preserves the property that round(k) - round(k-1) = 1 for all doubles k.
int	round_int (float x)
	Round to nearest integer.
int	truncate_int (double x)
	Truncate to nearest integer. This routine does fast truncation to an integer. It should simply drop the fractional portion of the floating point number.
int64_t	abs (int64_t a)
unsigned	bitcount (unsigned v)
void	hack ()
template<typename FloatT >
bool	FloatEquals (FloatT f1, FloatT f2, FloatT maxDelta)
template<typename T , std::enable_if_t< std::is_floating_point< T >::value, bool > = true>
T	RoundF (const T value, const T multiple)
	Round a floating point number to nearest multiple.

Detailed Description

Math utility class. Note that the test() routine should return true for all implementations.

See http://ldesoras.free.fr/doc/articles/rounding_en.pdf for an explanation of the technique used on x86.

Function Documentation

abs()

int64_t MathUtils::abs ( int64_t  a )

inline

bitcount()

unsigned MathUtils::bitcount ( unsigned  v )

inline

FloatEquals()

template<typename FloatT >

bool MathUtils::FloatEquals ( FloatT  f1, FloatT  f2, FloatT  maxDelta  )

inline

Compare two floating-point numbers for equality and regard them as equal if their difference is below a given threshold.

It is usually not useful to compare float numbers for equality with the standard operator== since very close numbers might have different representations.

hack()

void MathUtils::hack ( )

inline

round_int() [1/2]

int MathUtils::round_int ( double  x )

inline

Round to nearest integer. This routine does fast rounding to the nearest integer. In the case (k + 0.5 for any integer k) we round up to k+1, and in all other instances we should return the nearest integer. Thus, { -1.5, -0.5, 0.5, 1.5 } is rounded to { -1, 0, 1, 2 }. It preserves the property that round(k) - round(k-1) = 1 for all doubles k.

Make sure MathUtils::test() returns true for each implementation.

See also

truncate_int, test

round_int() [2/2]

int MathUtils::round_int ( float  x )

inline

Round to nearest integer.

See also

truncate_int, test

RoundF()

template<typename T , std::enable_if_t< std::is_floating_point< T >::value, bool > = true>

T MathUtils::RoundF ( const T  value, const T  multiple  )

inline

Round a floating point number to nearest multiple.

Parameters

value	The value to round
multiple	The multiple

Returns

The rounded value

truncate_int()

int MathUtils::truncate_int ( double  x )

inline

Truncate to nearest integer. This routine does fast truncation to an integer. It should simply drop the fractional portion of the floating point number.

Make sure MathUtils::test() returns true for each implementation.

See also

round_int, test