3 ms·
I have a problem with the mathematical proof document of mpfr accuracy. The mathematical setup work is not rigourous enough. I allow myself to say that, becaus
by sylware 4y ago
I have a problem with the mathematical proof document of mpfr accuracy.
The mathematical setup work is not rigourous enough. I allow myself to say that, because I know my maths teachers would have rejected such work.
Solution, cleanup and redo, this time very rigourously, the maths setup work of this document. Yes, it will probably add a few pages.
I had a look at SLEEF, but I could not find the mathematical proof document of the accuracy of its approximants.
- reikonomusha 4y agoWhich document are you talking about? This one [1]? If so, "because my maths teachers" is not a very good rebuttal to what is obviously a form of program documentation and not peer reviewed publication. The document could use some editorial cleanup, but so don't see anything systematically wrong with the math. It's understandable by people versed in the language of numerical analysis, which is certainly the target audience. [1] https://www.mpfr.org/algorithms.pdf https://www.mpfr.org/algorithms.pdf
- sylware 4y agoYes, this document has not a rigourous math setup: for instance rule 9. I don't think there is an accuracy proof document of SLEEF approximants.