Dubbelprecision

Dubbelprecisie, or double precision, is a common format for representing floating-point numbers in computing. It typically uses 64 bits per value, following the IEEE 754 standard. The bit layout consists of 1 sign bit, 11 exponent bits, and 52 significand (fraction) bits, with the exponent biased by 1023. This structure provides a wide dynamic range and about 15 to 17 decimal digits of precision.

The largest finite value is about 1.7976931348623157e308, the smallest positive normalized about 2.2250738585072014e-308, and smaller subnormals go down to about 5.0e-324. The representation can also encode zero, infinities, and NaN values for error signaling.

In practice, dubbelprecisie is standard in many programming languages. In C and C++, the type double; in Java, double; in Python, float uses double-precision under the hood. On most hardware, double-precision operations use the native floating-point unit, offering good performance but at the cost of higher memory use than single precision.

Advantages include higher precision to reduce rounding errors and a larger dynamic range. Disadvantages include higher memory usage (8 bytes per value) and potentially slower computation on hardware optimized for single precision. Floating-point arithmetic is subject to rounding errors, representation issues such as non-exact decimal values, and special values like NaN and infinities.

Subnormal numbers extend the lower end of the range but can have reduced precision and slower operations. In numerical work, care is needed with equality checks, accumulation errors, and conversion from decimal input. See also: single precision; IEEE 754; floating-point arithmetic.