Update THESHOLD name in jacobi.c

Seth Troisi braintwo at gmail.com
Wed Apr 17 01:34:51 UTC 2019


Thank you for the link, it's nice to see it matches my understanding of
what's going on.
Did your draft paper ever get published anywhere such that I can link to it
as a reference?

I in reading the 2010 comment chains I particularly enjoyed this comment
about you,
https://gmplib.org/list-archives/gmp-devel/2010-January/001458.html

I did a first pass at a new high level overview of the algorithm.
https://github.com/sethtroisi/libgmp/pull/5/files
(I can't currently verify the html docs but I'll do so when I get home)

I'd appreciate some guidance around gmp documentation.
1. Is it enough to make incremental improvements to this section (my
preference)? or should I strive for a single, larger, completely fleshed
out patch?
2. Should I add descriptions of other considered algorithms and their
downsides (e.g. EGCD, BGCD)?
3. Who do you consider to be the audience for these documents? Should there
be additional introductory links to these concepts?

Thanks for all the help,
Seth


On Tue, Apr 16, 2019 at 1:13 PM Niels Möller <nisse at lysator.liu.se> wrote:

> Seth Troisi <braintwo at gmail.com> writes:
>
> > Per Niels' suggestion I'm taking a look at the Jacobi documentation.
>
> Algorithm description was most recently posted here:
> https://gmplib.org/list-archives/gmp-bugs/2017-August/004190.html
>
> > I noticed that jacobi.c aliases GCD_DC_THRESHOLD then uses both names
> once.
> > https://gmplib.org/repo/gmp/file/tip/mpn/generic/jacobi.c#l177
> > https://gmplib.org/repo/gmp/file/tip/mpn/generic/jacobi.c#l196
>
> I agree it should be JACOBI_DC_THRESHOLD at both places. This makes it
> possible (but optional) to use a separaately tuned jacobi threshold.
> Which is unlikely to make much difference. The use of mpn_hgcd2_jacobi
> rather than mpn_hgcd2 implies a *sligthly* larger constant for the linear
> part of the work.
>
Agreed.

Silly question, what does _DC_ refer to in the threshold name?


> Regards,
> /Niels
>
> --
> Niels Möller. PGP-encrypted email is preferred. Keyid 368C6677.
> Internet email is subject to wholesale government surveillance.
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: docs_jacobi.patch
Type: text/x-patch
Size: 2431 bytes
Desc: not available
URL: <https://gmplib.org/list-archives/gmp-devel/attachments/20190416/aba028a8/attachment-0001.bin>


More information about the gmp-devel mailing list