Unit in the Last Place (ULP)

Library for performing comparisons and dynamically calculating floating point limits including the Unit in the Last Place (ULP). More...

Collaboration diagram for Unit in the Last Place (ULP):

Files
file	ulp.h

Functions
double	bn_ulp (double val)
	Calculate unit in the last place (ULP) for a given value.
float	bn_ulpf (float val)
	same as bn_ulp() but for single precision floating point.

Machine epsilons, smallest step after 1.0 @ {
double	bn_dbl_epsilon (void)
float	bn_flt_epsilon (void)

Numeric positive normalized range limits @ {
double	bn_dbl_min (void)
double	bn_dbl_max (void)
float	bn_flt_min (void)
float	bn_flt_max (void)

Square-rooted limits @ {
float	bn_flt_min_sqrt (void)
float	bn_flt_max_sqrt (void)
double	bn_dbl_min_sqrt (void)
double	bn_dbl_max_sqrt (void)

Next possible value up/down from a given value @ {
double	bn_nextafter_up (double val)
double	bn_nextafter_dn (double val)
float	bn_nextafterf_up (float val)
float	bn_nextafterf_dn (float val)

Detailed Description

Library for performing comparisons and dynamically calculating floating point limits including the Unit in the Last Place (ULP).

Function Documentation

bn_dbl_epsilon()

double bn_dbl_epsilon ( void  )

extern

bn_flt_epsilon()

float bn_flt_epsilon ( void  )

extern

bn_dbl_min()

double bn_dbl_min ( void  )

extern

bn_dbl_max()

double bn_dbl_max ( void  )

extern

bn_flt_min()

float bn_flt_min ( void  )

extern

bn_flt_max()

float bn_flt_max ( void  )

extern

bn_flt_min_sqrt()

float bn_flt_min_sqrt ( void  )

extern

bn_flt_max_sqrt()

float bn_flt_max_sqrt ( void  )

extern

bn_dbl_min_sqrt()

double bn_dbl_min_sqrt ( void  )

extern

bn_dbl_max_sqrt()

double bn_dbl_max_sqrt ( void  )

extern

bn_nextafter_up()

double bn_nextafter_up ( double  val )

extern

bn_nextafter_dn()

double bn_nextafter_dn ( double  val )

extern

bn_nextafterf_up()

float bn_nextafterf_up ( float  val )

extern

bn_nextafterf_dn()

float bn_nextafterf_dn ( float  val )

extern

bn_ulp()

double bn_ulp ( double  val )

extern

Calculate unit in the last place (ULP) for a given value.

If provided value is positive, it's the positive distance to the next double. If provided value is negative, it's the negative distance to the next negative double.

Parameters

val	the input double value.

Returns

distance to the next representable double precision value

bn_ulpf()

float bn_ulpf ( float  val )

extern

same as bn_ulp() but for single precision floating point.