Have a look at the partition numbers

[fivemack]

If you want the opportunity of running ECM 24/7 and getting a freshly-baked factor per CPU in your inbox most mornings, can I commend to you the partition numbers from

http://www.asahi-net.or.jp/~KC2H-MSM/mathland/part/

I am currently clearing up the 13000-13500 range and running ECM on the 20000-21000 range, but in other ranges there are literally thousands of numbers as small as C101 that have seen only ECM with B1=50k, and others that have been well-ECMed and are just sitting there begging to be GNFSed. Come on in, fill your boots!

There *are* interesting divisibility properties to be found among the partition numbers: the famous 5|p(5n+4), 7|p(7n+5), 11|p(11n+6), and infinitely many (see http://www.sciencenews.org/articles/20050618/bob9.asp ) weirder ones (for example, part(17303n+237) is divisible by 13), but I'm fairly sure they don't apply for factors this big; on the other hand, there being no strong mathematical use for the factors is scarcely a disincentive for other factorisation projects.

Trying to sound like an advocate for the Oregon Trail,

Tom

[R.D. Silverman]

Originally Posted by fivemack

If you want the opportunity of running ECM 24/7 and getting a freshly-baked factor per CPU in your inbox most mornings, can I commend to you the partition numbers from

http://www.asahi-net.or.jp/~KC2H-MSM/mathland/part/

I am currently clearing up the 13000-13500 range and running ECM on the 20000-21000 range, but in other ranges there are literally thousands of numbers as small as C100 that have seen only ECM with B1=50k, and even some small enough that QS is the right method. Come on in, fill your boots!

There *are* interesting divisibility properties to be found among the partition numbers: the famous 5|p(5n+4), 7|p(7n+5), 11|p(11n+6), and infinitely many (see http://www.sciencenews.org/articles/20050618/bob9.asp ) weirder ones (for example, part(17303n+237) is divisible by 13), but I'm fairly sure they don't apply for factors this big; on the other hand, there being no strong mathematical use for the factors is scarcely a disincentive for other factorisation projects.

Trying to sound like an advocate for the Oregon Trail,

Tom

You should be aware that the partition numbers were part of the RSA
Challenge contest back in the 90's. They were restricted to those
with prime indices, but a VERY LARGE NUMBER of them were fully factored.
Check the RSA website to see if the data is still available.

I will check to see if I still have a copy of the data.

Bob

[fivemack]

Originally Posted by R.D. Silverman

You should be aware that the partition numbers were part of the RSA
Challenge contest back in the 90's. They were restricted to those
with prime indices, but a VERY LARGE NUMBER of them were fully factored.
Check the RSA website to see if the data is still available.

I will check to see if I still have a copy of the data.

Bob

That data has already been merged into Mishima's tables; at least, I looked at http://www.ontko.com/~rayo/primes/ to get the RSA-challenge data, then grepped the Mishima tables for some of the largish factors from http://www.ontko.com/~rayo/primes/hr_p135.txt and found them all.

[R.D. Silverman]

Originally Posted by fivemack

That data has already been merged into Mishima's tables; at least, I looked at http://www.ontko.com/~rayo/primes/ to get the RSA-challenge data, then grepped the Mishima tables for some of the largish factors from http://www.ontko.com/~rayo/primes/hr_p135.txt and found them all.

That is good news. I am glad that the data did not get lost.

[Andi47]

Reserving 14984 c104 for msieve (QS)

[ET_]

Originally Posted by Andi47

Reserving 14984 c104 for msieve (QS)

Should we tell Mishima about our reservation?

Luigi

[fivemack]

I don't think there are all that many people working on these numbers ... contributions this year have, with one exception, been by me or by Hisanori Mishima himself, which is really why I posted here to drum up more interest. It would be nice to tell him about the reservations, but he manages the site manually so we don't want to overload him entirely.

[fivemack]

Out of curiosity, I collected all the cofactors together, and did 35 rounds of GCD(product of half the cofactors selected at random, product of the other half), all of which returned 1. So I think it's quite unlikely (IE there are about 2^24 pairs, and for any pair the probability is 2^-34 that both were on the same side all 35 times) that any number will divide more than one of the cofactors.

[Andi47]

Reserving 29328 c101 and 28819 c102 (as an attempt to compare msieve's QS and GNFS with numbers in similar size)

[smh]

Originally Posted by fivemack

I don't think there are all that many people working on these numbers ... contributions this year have, with one exception, been by me or by Hisanori Mishima himself, which is really why I posted here to drum up more interest. It would be nice to tell him about the reservations, but he manages the site manually so we don't want to overload him entirely.

I've done quite a few last year before cleaning up the small composites on the cyclotomic tables. Most < 110 digits are done now (I did 7-800 between 100 and 110 digits in the last few months) so i might return to the partition tables. Most composites above 20K haven't had enough ecm though. Last time i found quite a few factors with GNFS in the 30 digit range.

I also has been a long time i did any numbers of the (Generalized) Cullen and Woodall numbers so i might return there first.

[bdodson]

Originally Posted by R.D. Silverman

You should be aware that the partition numbers were part of the RSA
Challenge contest back in the 90's. They were restricted to those
with prime indices, but a VERY LARGE NUMBER of them were fully factored.
Check the RSA website to see if the data is still available.

I will check to see if I still have a copy of the data.

Bob

As I recall, Paul Zimmermann and Arjen are still waiting for the
last quarter's prize. We quit when the prizes were no longer
being given. But my recollection is that there was another round
of factors waiting for the quarter to finish (to be eligible for the next
quarter's prize), and that some of the later ones that we did
submit never got included.

I can check, but it has been quite some time. Again, we were only
doing the ones of prime index (the thought being that there were
too many of each given size, otherwise). Also, aside from the wish
that there might be instances of new congruences/divisibilities found,
these are (iirc) only gnfs candidates, with no chance of snfs --- that
was one of the appeals, in our case. Most of our early gnfs trials were
taken from the partition list, the first ones for gnfs above c110,
including c112, c119, c116 from our Crypto'95 paper on the cycle explosion
(gnfs with four large primes). -Bruce

PS -- sentance fragment patched. There are also some of our
factors on one of Brent's early pages, from back when he was taking all
ecm factors above p40 (especially the year or two when we were
working toward p50, finished with Curry's p53).