udiv_qr_3by2 vs divappr

[Niels Möller]

paul zimmermann <Paul.Zimmermann at inria.fr> writes:

> your idea works. If I am correct, we save a \beta/4 factor, which is enough
> as long as \beta>=4.

I've updated the paper (same location,
https://www.lysator.liu.se/~nisse/misc/schoolbook-divappr.pdf).

The new version gets the stricter bounds right, hopefully, and it also
replaces [p_0 > 0] by 1, simplifying the algorithm.

Thanks a lot for feedback and suggestions!

Regards,
/Niels

-- 
Niels Möller. PGP-encrypted email is preferred. Keyid 368C6677.
Internet email is subject to wholesale government surveillance.