--- In primenumbers@y..., David Cleaver <wraithx@m...> wrote:
> Hello all,
> I'm not sure who to address this question to, so I'll just ask the
There is a little bit of a FAQ located at:
is the main page for the ECM
distributed search. If you dig around for a while on that site, you
will find information, along with links (many dead :( however) which
tell you about "optimal" B1 selections, and reference papers to the
ECM algorithm along with the many tweaks which gmp_ecm has had.
> how, exactly, do you use this ecm factoring program? (you know,
> the ecm4c-r4-... thingy) I'm not sure I understand what the "loop"
> variable is for,
ECM is a probabalistic algorithm. You simply throw "enough" curves
at a problem, and there is a very good chance a factor will drop out.
Originally, the ecm4c only performed one loop test. .BAT files (or
unix script files) had to be written to run the thing over and over
again. The loops simply keeps ecm4c from exiting. It runs more than
one loop (with different "random" sigma's).
> or why I would want to increment B1 or even how often
> B1 is incremented.
The B1 increment was added by request. If you do not want to use the
program in this manner, here is a link to a table of "suggested" B1
> And, why is there the "-k" option when you can set
> B2 explicitly already?
Don't know. This was in the ecm4c code. You might send an email to
Paul Zimmerman (ecm4c was written by him). The -k allows you to
adjust the constant which B1 is multiplied by to create B2.
> Is there already a FAQ out there that I'm missing?
> If not, could this be the start of it? And if there is, could
> someone upload it to the files area under the factoring/gmp-ecm
> folder? I'd really like to help contribute some factorizations,
Then simply start using ecm4c on composites which have not yet been
factored, and which have not been exaustively searched with ECM.
I have two recommendations:
1. Do not work with B1 values which are too small for the task
2. Do not work with B1 values which are too large for the task.
Also if factorizing expressions which have not had any factorization
done on them at all, I highly recommend some -t and -r trail factoring
and Pollard/Brent Rho work to remove any tiny factors, before the
ECM work starts.
> but need to
> understand the tools before I start. Thanks in advance for any
> help you can provide.
> -David C.