double is intended to be a native real floating-point type of a natural size, capable of representing a practical range of real numbers. (long double is usually a larger and more capable type, while float is usually less capable but more compact.) A floating-point constant of type double is not suffixed.

Various macros in <float.h> describe the capabilities of double. It supports DBL_DIG decimal digits of precision, or DBL_MANT_DIG digits in the native base FLT_RADIX.

double has a range of ±DBL_MAX. The smallest positive non-zero value it can take is DBL_TRUE_MIN, which could be subnormal. The smallest positive normalized value it can take is DBL_MIN.

From C11, if subnormal values are possible, DBL_HAS_SUBNORM expands to 1. If not, it expends to 0. If it is undeterminable, it expands to -1.

DBL_MIN_10_EXP is the lowest integer x that you can give to pow(10.0, x) and still get a normalized value. DBL_MAX_10_EXP is the highest integer x that you can give to pow(10.0, x) and still get a normalized value.

DBL_MIN_EXP is the lowest integer x that you can give to pow(FLT_RADIX, x - 1) and still get a normalized value. DBL_MAX_EXP is the highest integer x that you can give to pow(FLT_RADIX, x - 1) and still get a normalized value. DBL_EPSILON is the smallest value you can add to 1.0 to yield something other than 1.0.

DBL_DECIMAL_DIG is the number of decimal digits that double can retain without rounding error.

From C99, if __STDC_NO_COMPLEX__ is defined, double _Complex is a complex floating-point type that has the same representation as an array of two doubles, the first being the real part, and the second the imaginary part, of the complex number it represents. If imaginary is defined in <complex.h>, double _Imaginary is an imaginary floating-point type that has the same representation as a double. double imaginary and double complex are alternative names for these types, available in <complex.h>.

#include <complex.h>
double complex CMPLX(double re, double im);

The macro CMPLX expands to a constant expression suitable to initialize a double complex with static or thread-local storage, with a value re + I * im, a complex number with a real component re and an imaginary component im, provided re and im are similarly suitable.

If CMPLX is unavailable, a portable implementation could be:

#define CMPLX(RE, IM) \
  (*(double _Complex *) (double[2]) { (RE), (IM) })

This information derives from a comp.std.c thread in May 2009.

A value of type double complex can be split into cartesian co-ordinates with creal and cimag, or into polar co-ordinates with carg and cabs.