Legacy Document

frexp

You can use the frexp function to find out the values of a floating-point number's fraction field and exponent field.

double_t frexp (double_t x, int *exponent);

x

Any floating-point number.

exponent

A pointer to an integer in which the value of the exponent can be returned.

DESCRIPTION

The frexp function splits its first argument into a fraction part and a base 2 exponent part. This is an ANSI standard C library function.

$frexp(x,n)\; =\; f$ such that $x\; =\; f\times 2n$

or

$frexp(x,n)\; =\; f$ such that $n\; =\; (1+logb(x))$ and $f\; =\; scalb(x,-n)$

The return value of frexp is the value of the fraction field of the argument x. The exponent field of x is stored in the address pointed to by the exponent argument.

For finite nonzero inputs, frexp returns either 0.0 or a value whose magnitude is between 0.5 and 1.0.

The ldexp and scalb functions perform the inverse operation (compute $f\times 2n$ ).

EXCEPTIONS

If x is finite and nonzero, the result of $frexp(x,n)$ is exact.

SPECIAL CASES

Table 10-13 shows the results when the input argument to the frexp function is a zero, a NaN, or an Infinity.

Special cases for the frexp function

Operation	Result	Exceptions raised
$frexp(+0,n)$	+0 (n = 0)	None
$frexp(-0,n)$	$-0$ (n = 0)	None
$frexp(NaN,n)$	NaN (n is undefined)	None[34]
$frexp(+$ ,n)	+ (n is undefined)	None
$frexp(-$ ,n)	$-$ (n is undefined)	None

EXAMPLES

z = frexp(2E300, n);    /* z  0.746611 and n = 998. In other
                           words, 2  10300  0.746611  2998. */

---

[34] If the NaN is a signaling NaN, the invalid exception is raised.