Indeed, it very much depends on the quality of the implementation of the runtime libraries whether the signed zero is respected or not. Anton’s article illustrates that nicely. (Tangential to this: see the post by Beliavsky - Fortran code snippets - #22 by Beliavsky - where he illustrates the use of complex numbers to determine the derivative of a real (!) function with just one evaluation.)
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Inversion of complex numbers with a zero imaginary part | 2 | 191 | February 7, 2025 | |
| Complex arithmetic significantly slower? - not really | 27 | 1373 | October 20, 2022 | |
| Error Handling for negative real numbers | 34 | 3516 | July 1, 2020 | |
| Wrong (strange outputs) of real numbers | 27 | 1560 | April 26, 2023 | |
|
`nint(-0.1) = ?`
|
18 | 1752 | December 8, 2022 |