Expand description
Fast compensated (extended-precision) arithmetic
Extends the precision of basic numerical types such as f64 by
compensating for the error. For convenience, we rovide the Df64 type,
which implements all basic arithmetic traits, but does so in quadruple
(double-double) precision. This is usually faster than emulated quad
precision and multi-precision by several orders of magnitude.
Modules§
- arith
- Basic algorithms of compensated arithmetic.
- checks
- Floating point checks.
- gauss
- Gauss quadrature rules
Structs§
- Compensated
- Type for compensated arithmetic.
Traits§
- AddFast
- Addition under the assumption of ordered arguments.
- Compensated
Arithmetic - Arithmetic with compensated errors.
- SubFast
- Subtraction under the assumption of ordered arguments.
Type Aliases§
- Df64
- Compensated f64 (emulated quad precision) type.