Compute the difference two numbers and compute the numerical round-off error. See Shewchuk 1997 Adaptive Precision Floating-Point Arithmetic and Fast Robust Geometric Predicates This should only be used if you know that the magnitude of a is greater than or equal to b, otherwise, you should use the slower twoDiff routine
Interfaced with fastTwoDiff()
Type | Intent | Optional | Attributes | Name | ||
---|---|---|---|---|---|---|
real(kind=r32), | intent(in) | :: | a | First number |
||
real(kind=r32), | intent(in) | :: | b | Second number |
Result and its error
Interfaced with fastTwoDiff()
Type | Intent | Optional | Attributes | Name | ||
---|---|---|---|---|---|---|
real(kind=r64), | intent(in) | :: | a | First number |
||
real(kind=r64), | intent(in) | :: | b | Second number |
Result and its error