F1049

LaurV · 2012-02-09, 02:10

Just got a PM today, telling me that my C185 was factored by S.Irvine. Well, I was on the linear algebra phase, with about 56 hours left to finish, after about 4 months of single-core sieving (and over 67 millions relations).

The matrix is over 5.8 million dimensions and it takes 3.8 Gigs of memory.

That is the largest factorization attempted by myself, in fact not all by myself, because I got the poly for free from fivemach.

Well, c'est la guerrrrrre!

I will let it finish to see if I get the same results

as Mr. Irvine, and I would be curious how much it took for him (from the list I see he used SNFS, I was using yafu/ggnfs, sieving on rational side, as bsquared advised me last year in october).

Batalov · 2012-02-09, 02:55

SNFS poly shouldn't be hard to make.

People were factoring Fibonacci's for decades. Did you check with them before starting the factorization? How does one make a number "mine" or "his", I'd like to learn.

LaurV · 2012-02-09, 04:46

Quote:

Originally Posted by Batalov

SNFS poly shouldn't be hard to make.

Yeah, if you know how... That time, I did not know, and I did not know what job I am committing to... As explained in the yafu thread, that time I did not know what such a factorisation means, and after a couple of successful C130-140 i said "why not to try higher". Be that a lesson for everyone

, and I committed to finish it after fivemack posted the poly just from pride. If I was stupid, at least let's try to finish what I was starting... hehe.

Quote:

People were factoring Fibonacci's for decades. Did you check with them before starting the factorization?

Why should I? I did not seek for fame or glory :P, if someone else could be faster, ok, that's life...

Additionally, they are all reading this forum, they are all more clever and more educated then me on this subject, and they did not check with me either

, none of them were telling me NOT to factor it when I started 5-6 months ago (see the date of the post which I linked before!), nor when I said I have 30M relations, later in that thread, nor when I said I have 35 or 40, or 45M relations, and moreover, noone jumped when I said (at around 45M relations) that it become hard and I was willing to give up and transfer the relations to someone else... There is all the discussion in the yafu thread, one guy from that list even commented there, and anyhow, if they would tell, I would still do the factorisation anyhow, from pride and as a lesson for myself, as I said. I never pretended or tried to be first to finish, that is why I only allocated one core to it.

Starting of this thread was not a complain, just a report of the facts. I am perfectly aware that there are people there outside who really KNOW what they are doing, opposite of me who I am just playing around, without a real goal or understanding...

Quote:

How does one make a number "mine" or "his", I'd like to learn.

You just say it, and it is yours. Do you want any number? I give you any one for free, except 23, which is mine, because is my lucky number

.

Batalov · 2012-02-09, 05:07

I don't think Sean frequently reads the forum recently, but as you can see he's been factoring Fibs for a long time. And xyyx's, and factorials, and near-repunits, and many more things.

If you want to contact him for coordination of a next factorization - you can use a few magic words for google: his location and "number theory".

LaurV · 2012-02-09, 05:25

Thanks for the advice, but I would be done with this story in about 40 to 50 hours, I have already 8 aliquot sequences waiting in line and one additional free core on a 4.2GHz CPU would be good to get rid of some C112-C114 there too...

I am big fan of Fibo numbers and golden section stuff (and Forex trader for 7-8 years now) but never played with a fibo bigger then 3 digits (233 is the biggest one I know by heart, and of course all the smaller), and "too big is too big"

, so I would end this story, and stay into more reasonable size aliquots, like C100-C120 for the time being...

Batalov · 2012-02-09, 05:48

It later occurred to me that you can also simply PM sean. (literally.)

Don't feel bad: even with reservations, double factorizations do happen -- in Cunnighams for one example, or in repunits for another (but not in near-repunits). And primality proofs collide, too. One would need both hands to count such cases! Re: ximan's potential retort from the future: "Yes, in binary, too" :-)

2012-02-09, 02:10	#1
LaurV Romulan Interpreter Jun 2011 Thailand 22614₈ Posts	F1049 Just got a PM today, telling me that my C185 was factored by S.Irvine. Well, I was on the linear algebra phase, with about 56 hours left to finish, after about 4 months of single-core sieving (and over 67 millions relations). The matrix is over 5.8 million dimensions and it takes 3.8 Gigs of memory. That is the largest factorization attempted by myself, in fact not all by myself, because I got the poly for free from fivemach. Well, c'est la guerrrrrre! I will let it finish to see if I get the same results as Mr. Irvine, and I would be curious how much it took for him (from the list I see he used SNFS, I was using yafu/ggnfs, sieving on rational side, as bsquared advised me last year in october).

2012-02-09, 02:55	#2
Batalov "Serge" Mar 2008 Phi(4,2^7658614+1)/2 2506₁₆ Posts	fibonacci(1049) cofactor SNFS poly shouldn't be hard to make. People were factoring Fibonacci's for decades. Did you check with them before starting the factorization? How does one make a number "mine" or "his", I'd like to learn.

2012-02-09, 05:07	#4
Batalov "Serge" Mar 2008 Phi(4,2^7658614+1)/2 2×7×677 Posts	I don't think Sean frequently reads the forum recently, but as you can see he's been factoring Fibs for a long time. And xyyx's, and factorials, and near-repunits, and many more things. If you want to contact him for coordination of a next factorization - you can use a few magic words for google: his location and "number theory".

2012-02-09, 05:25	#5
LaurV Romulan Interpreter Jun 2011 Thailand 2²·3³·89 Posts	Thanks for the advice, but I would be done with this story in about 40 to 50 hours, I have already 8 aliquot sequences waiting in line and one additional free core on a 4.2GHz CPU would be good to get rid of some C112-C114 there too... I am big fan of Fibo numbers and golden section stuff (and Forex trader for 7-8 years now) but never played with a fibo bigger then 3 digits (233 is the biggest one I know by heart, and of course all the smaller), and "too big is too big" , so I would end this story, and stay into more reasonable size aliquots, like C100-C120 for the time being...

2012-02-09, 05:48	#6
Batalov "Serge" Mar 2008 Phi(4,2^7658614+1)/2 2·7·677 Posts	It later occurred to me that you can also simply PM sean. (literally.) Don't feel bad: even with reservations, double factorizations do happen -- in Cunnighams for one example, or in repunits for another (but not in near-repunits). And primality proofs collide, too. One would need both hands to count such cases! Re: ximan's potential retort from the future: "Yes, in binary, too" :-)