A 64-bit or 128-bit representation of a number with a precision of 16 or 34 decimal digits and an exponent. Decimal floating-point numbers include normal numbers, subnormal numbers, and the special values of infinity, NaN, and sNaN. In IBM SQL, decimal floating-point numbers are not considered to be floating-point numbers. See also not-a-number, infinity, signalling NaN, subnormal number, normal number, floating-point number.