Cunningham Table Composites in FactorDB

[Batalov]

This is a test of the visualization for the remaining composites at a glance, using FactorDB*. Comfortably, this post will hardly need any maintenance (after initial set-up). Everyone is free to fix omissions in FactorDB!

It is important to remember that the content may be out of date with the real Cunningham Tables and that all work reservations an reports go to their maintainer.

Table 2- | Table 2+ | Table 2LM
Table 3- | Table 3+ | Table 3LM (extension to 1800)
Table 5- | Table 5+ | Table 5LM
Table 6- | Table 6+ | Table 6LM
Table 7- | Table 7+ | Table 7LM
Table 10-| Table 10+| Table 10LM
Table 11-| Table 11+| Table 11LM
Table 12-| Table 12+| Table 12LM

Well, this is only a test. Please discuss below.

*FactorDB is a brainchild of Markus Tervooren (known as Syd). There is a separate thread for reference.

[Batalov]

Alternative split view for 2+
Table 2+ (odd) | Table 2+ (even)

P.S. Need to think how to make extra LMs (3s and 6s are done, but there are some left in others).

[kar_bon]

that's what i'm thought of:

one click and you can see, which numbers are not factored yet.
and the Cxx without searching some other tables/pages!

BTW: i never had in mind to use the FactorDB in first and omit the Cunningham-pages!

[Batalov]

FactorDB has a problem by not knowing Aurifeuillian factorizations internally.

For example for 2LM tables, the attempt to separate 2L and 2M (in order to see what is left to be done) fails miserably.

Example: FactorDB knows all factors of 2^1614+1, but it cannot produce factors of 2^807+2^404+1 and 2^807-2^404+1.
But ask it to factorize
(2^807+2^404+1)*(2^807-2^404+1)
and it will! (It will do the math and will look up the resulting number 2^1614+1 in its numeric value.)

So, it's not immediately ready for this trick in full. For now, I've simply sent lookups to 2^(4*n+2)+1, where L and M parts are mixed together.

This is inconvenient.

[Mini-Geek]

So why don't you post and request that it learn Aurifeuillian factorizations?
http://www.mersenneforum.org/showthr...=11119&page=10
I wouldn't expect it to happen very soon since it's transitioning to a major update, and I expect Syd will have more urgent issues than that, but might as well request it.

[bdodson]

Originally Posted by Batalov

Well, this is only a test. Please discuss below.

So once the database finshes reloading, is there a
recommended pointer on submitting prime factors?

My first report depends upon the factoring method;
(Cunningham) gmp-ecm factors go to the ECMNET
quick-start cgi, and PaulZ sends an email to the
ecmnet list, which includes Sam. (Cunningham) sieving
factors are usually reported directly to Sam, by whichever
co-worker ran the matrix (Greg, Tom and Serge for large
matrices, or myself for smaller ones). My second report
has been going either here or to one of the other forum
threads (factoring, ecmnet, nfsnet). Sounds like I ought
to be submitting to the DB second, and noting that if/when
I post a notice and/or info on the relevant forum thread.

So once the hungry DB has been fed, I'd hope people take
the time to look over Cunningham ecm progress on PaulZ's
page (top10; or quick-start then by_date) and over-all
progress from Sam's pages. Keeping the DB current should
most likely be done by reading in entires from Sam's pages,
especially "the most recent page" link. Then we can continue
tracking history and specific content in the individual tables
here; depending upon interest(s).

-bd

[kar_bon]

Originally Posted by bdodson

So once the database finshes reloading, is there a
recommended pointer on submitting prime factors?

there's a link/page where you can copy/paste factorizations or upload a file.

only give the factorization as "Composite = Factor1 * Factor2 * ..."
will insert it in the DB. no need to specify a special b^n+/-1 for instance.

[Batalov]

I've now reused the 2LM simple (but not precise) URLs for other LMs.
Note the c400 and c373 in Table 5LM - these are fairly easily algebraically split into two composites each. Left as an exercise to the reader! (The solution is easy if a bit cumbersome, just start from L and M forms conveniently found in the main table after each section.)

P.S. yes, I did make a suggestion in the other subforum, but I know how busy Syd may be, especially now. So it is formulated in such a way that can take away a burden from him and place it on a submitter (who can submit them as a | b).

[Batalov]

P.S. Split the c400 and c373.

[bdodson]

Originally Posted by kar_bon

that's what i'm thought of:

one click and you can see, which numbers are not factored yet.
and the Cxx without searching some other tables/pages!

Uhm; it's now several days since the initial Aug 12th post (over
on the 6+ thread), which I incorrectly read as a reply. Sounds
like the DB is partly intended as the (n+1)st factorization function
(cf. gmp-ecm posts about "complete" factorizations, rather than
"find a factor and stop"). Especially useful for locating the hard
part of an un-tested input. I'm usually mostly muterring about
what to do with given known hard composites.

On reporting factors found, and locating someone else's recently
found factor ... Just curious, suppose I can wait until I have a
factor to report; does filing a report require registration/login? I
see that there's provision for pasting in gmp-ecm and msieve reports,
which settles that, as far as I'm concerned.

If you had been replying to either my or Andi47's post, and I'd
submitted my new factor to the DB, is there a DB provision for
browsing recently submitted Cunningham factors, resp., factors
recently submitted to one of the 16 tables, such as 6+? The
tables you've suggested, as Serge is working on above, immediately
drop a number such as the C216 cofactor of 6, 304+ upon submitting
the report, as someone did for my p78. This happened to be a
"first hole" (the trad. Cunningham term, along with "first five holes"),
so upon submitting the p78, the 6+ table changes by dropping the
first entry (for 304+), and starting with ... oops, the link you posted
has already expired ... presumably due to the recent updating? Well,
anyway, clicking on Serge's link, I get ... (starting with) 6, 314+.

And on my previous point, doesn't my submitting the p78 mean that
DB users have the factor and/or updated list of remaining composite
cofactors, without being bothered with (and/or "informed by") info
on the factoring method --- i.e., either the Step 1 limit and sigma
of the elliptic curve, for ecm, or snfs -vs- gnfs for a sieving report?
Much less mention of Lehigh's contribution of the cputime; and not
to mention my own (and my co-factorer's) effort. I haven't gotten
to looking yet, but unless the DB intention is to reduce the info that
had been available (not as fluently, perhaps, but available) by looking
on PaulZ's ECMNET page (quick-start ...) or Sam's page (most recent
page ...), there ought to be some sort of flag or link as to where to
look for further info (method, factorer, _DATE_, history &etc).

BTW: i never had in mind to use the FactorDB in first and omit the Cunningham-pages!

Yes; I see. Yet another of your "BTW" transitions.
Thanks & Regards, Bruce

[kar_bon]

the Factoring Database is only for collecting factorizations of a wide range of numbers.

there's (by now) no option to display the name of the submitter/calculater nor data like
B1-values of an ECM-found, sigmas or dates.

although this new verison of the DB user-logins are possible, this is only for
collecting 'points' to using the workers (PC's from the DB or other users) for factoring.

Markus (the programmer) has done a great work the last months with this database, which
was the first one ever collecting factors. there're many (and quite very special) sites with
showing factorizations of numbers. but nobody ever tried to combine all in one.

if you wish such data should be also found in this DB, ask him!
perhaps more data could be shown then here.

as i mentioned, that DB should not replace any other tables with quite more and specialised
data. it's only to have a quick look for a certain factorization without first surching that corresponding page in the www!

BTW: -