isnan

Determines if the given floating point number arg is a not-a-number (NaN) value. The macro returns an integral value.

FLT_EVAL_METHOD is ignored: even if the argument is evaluated with more range and precision than its type, it is first converted to its semantic type, and the classification is based on that (this matters if the evaluation type supports NaNs, while the semantic type does not).

Parameters

Return value

Nonzero integral value if arg is a NaN, ​0​ otherwise.

Notes

There are many different NaN values with different sign bits and payloads, see nan.

NaN values never compare equal to themselves or to other NaN values. Copying a NaN may change its bit pattern.

Another way to test if a floating-point value is NaN is to compare it with itself: bool is_nan(double x) { return x != x; }

Example

#include <stdio.h>
#include <math.h>
#include <float.h>
 
int main(void)
{
    printf("isnan(NAN)         = %d\n", isnan(NAN));
    printf("isnan(INFINITY)    = %d\n", isnan(INFINITY));
    printf("isnan(0.0)         = %d\n", isnan(0.0));
    printf("isnan(DBL_MIN/2.0) = %d\n", isnan(DBL_MIN/2.0));
    printf("isnan(0.0 / 0.0)   = %d\n", isnan(0.0/0.0));
    printf("isnan(Inf - Inf)   = %d\n", isnan(INFINITY - INFINITY));
}

isnan(NAN)         = 1
isnan(INFINITY)    = 0
isnan(0.0)         = 0
isnan(DBL_MIN/2.0) = 0
isnan(0.0 / 0.0)   = 1
isnan(Inf - Inf)   = 1

References