|
|
|
#include <math.h> #include <stdio.h>
enum fp_class_type fp_class(x) double x;
int finite(x) double x;
int ilogb(x) double x;
int isinf(x) double x;
int isnan(x) double x;
int isnormal(x) double x;
int issubnormal(x) double x;
int iszero(x) double x;
int signbit(x) double x;
void ieee_retrospective(f) FILE *f;
void nonstandard_arithmetic()
void standard_arithmetic()
double copysign(x,y) double x, y;
double fabs(x) double x;
double fmod(x,y) double x, y;
double nextafter(x,y) double x, y;
double remainder(x,y) double x, y;
double scalbn(x,n) double x; int n;
Most of these functions provide capabilities required by ANSI/IEEE Std 754-1985 or suggested in its appendix.
fp_class(x) corresponds to the IEEE's class() and classifies x as zero, subnormal, normal, \(if, or quiet or signaling NaN. <floatingpoint.h> defines enum fp_class_type. The following functions return 0 if the indicated condition is not satisfied:
ilogb(x) returns the unbiased exponent of x in integer format. ilogb(±\(if) = +MAXINT and ilogb(0) = -MAXINT; <values.h> defines MAXINT as the largest int. ilogb(x) never generates an exception. When x is subnormal, ilogb(x) returns an exponent computed as if x were first normalized.
ieee_retrospective(f) prints a message to the FILE f listing all IEEE accrued exception-occurred bits currently on, unless no such bits are on or the only one on is "inexact". It's intended to be used at the end of a program to indicate whether some IEEE floating-point exceptions occurred that might have affected the result.
standard_arithmetic() and nonstandard_arithmetic() are meaningful on systems that provide an alternative faster mode of floating-point arithmetic that does not conform to the default IEEE Standard. Nonstandard modes vary among implementations; nonstandard mode may, for instance, result in setting subnormal results to zero or in treating subnormal operands as zero, or both, or something else. standard_arithmetic() reverts to the default standard mode. On systems that provide only one mode, these functions have no effect.
copysign(x,y) returns x with y's sign bit.
fabs(x) returns the absolute value of x.
nextafter(x,y) returns the next machine representable number from x in the direction y.
remainder(x, y) and fmod(x, y) return a remainder of x with respect to y; that is, the result r is one of the numbers that differ from x by an integral multiple of y. Thus (x - r)/y is an integral value, even though it might exceed MAXINT if it were explicitly computed as an int. Both functions return one of the two such r smallest in magnitude. remainder(x, y) is the operation specified in ANSI/IEEE Std 754-1985; the result of fmod(x, y) may differ from remainder()'s result by ±y. The magnitude of remainder's result can not exceed half that of y; its sign might not agree with either x or y. The magnitude of fmod()'s result is less than that of y; its sign agrees with that of x. Neither function can generate an exception as long as both arguments are normal or subnormal. remainder(x, 0), fmod(x, 0), remainder(\(if, y), and fmod(\(if, y) are invalid operations that produce a NaN.
scalbn(x, n) returns x* 2**n computed by exponent manipulation rather than by actually performing an exponentiation or a multiplication. Thus
1 <= scalbn(fabs(x),-ilogb(x)) < 2
for every x except 0, \(if, and NaN.
|
|
|
Created by unroff & hp-tools. © somebody (See intro for details). All Rights Reserved. Last modified 11/5/97