What does Cofactor work do?

BigDukeSix · 2020-12-28, 10:54

Hi guys and gals.

I did a read through the forums but I´m not entirely sure what what the cofactor prp does.
My understanding right now: exponents with a factor are going to the cf-queue and the factors are checked if they are prp.
What happens if a factor is not prp? Will the tf going deeper until there is found another factor wich will be checked for again?
If I´m not misstaken, a prp-test would not find a factor?

Could someone please explain in easy words the workflow of the cofactor work? (I´m neither a native english speaker nor a mathematician

)

Thank you very much in advance.

ATH · 2020-12-28, 14:23

Cofactor PRP is one of our side projects.

The main GIMPS project is to find Mersenne Primes by trial factoring candidates and then running a PRP test on exponents without a factor to check for probable primes followed by a certificate on the PRP test (new) or a PRP double check.
After that the exponent is uninteresting for the main project, but then other people in side projects will run PRP test on the cofactor to find large PRPs and others will continue to try and find factors by trial factoring, P-1 tests and ECM curves.

PRP on cofactors is still fairly new ~2-3 years and have "only" reached ~11M exponents, but is slowly going higher and higher and whenever a new factor is found below 11M the exponent is immediately scheduled for a PRP cofactor test, so everything is kept checked below the "wavefront".

All the sideproject trial factoring, P-1 and ECM is not very organized instead people are working on their favorite ranges and method.
There are a few people organizing their efforts in this thread, but right now the thread is "hijacked" by posts concerning testing a new Prime95 version 30.4b3 for P-1:
https://mersenneforum.org/showthread.php?t=22476

BigDukeSix · 2020-12-28, 15:04

Thank you very much. This is a very nice explanation for somebody like me.

Uncwilly · 2020-12-28, 15:56

Just a little clarification.

If we have a candidate number and we find a factor, the part that remains after dividing the number by the known factor is the co-factor. If we find several factors for the number, still what ever remains is the cofactor. We are running the PRP on that number to see if it is prime and thus, we known all of the prime factors for the candidate number. There are some numbers that have a PRP test run on the cofactors, but not a double check or a certificate on the PRP. (The ability to run that is recent). So there is a group of us working to get all of those double checks done. That way, going forward we should not have to do the longer (relative to the cert) DC's any more.

lavalamp · 2020-12-28, 16:24

Quote:

Originally Posted by ATH

PRP on cofactors is still fairly new ~2-3 years and have "only" reached ~11M exponents

Where are the PRPs for this recorded? I'd be interested to see how many mersenne (pseudo) semiprimes have popped out.

Uncwilly · 2020-12-28, 16:33

The residues are part of the record for the main exponent:
https://www.mersenne.org/report_expo...9735347&full=1

lavalamp · 2020-12-28, 16:43

Is there anywhere they are recorded separately such that one might peruse a list of them, in the same way as the mersenne primes are recorded separately. Once the exponent gets large enough (>1E4? >1E6?) I assume they are relatively rare finds.

Viliam Furik · 2020-12-28, 17:21

Quote:

Originally Posted by lavalamp

Is there anywhere they are recorded separately such that one might peruse a list of them, in the same way as the mersenne primes are recorded separately. Once the exponent gets large enough (>1E4? >1E6?) I assume they are relatively rare finds.

I think this might be what you are looking for: https://www.mersenne.ca/prp.php?show=1.

There is a list of a total of 350 fully or probably fully factored Mersenne numbers, plus the 51 Mersenne primes. You can set the page to show you cofactor state of all the exponents.

firejuggler · 2020-12-28, 17:25

isn't that what the fully factored mersenne numer thread used for?

edit :

this thread https://www.mersenneforum.org/showthread.php?t=19407

lavalamp · 2020-12-28, 18:08

Quote:

Originally Posted by Viliam Furik

I think this might be what you are looking for: https://www.mersenne.ca/prp.php?show=1.

Yes, nice!

chris2be8 · 2020-12-29, 16:26

M1399 shows as PRP on that page, but it's cofactor is proven prime in factordb. Although I didn't do it.

Chris

2020-12-28, 10:54	#1
BigDukeSix Dec 2020 7 Posts	What does Cofactor work do? Hi guys and gals. I did a read through the forums but I´m not entirely sure what what the cofactor prp does. My understanding right now: exponents with a factor are going to the cf-queue and the factors are checked if they are prp. What happens if a factor is not prp? Will the tf going deeper until there is found another factor wich will be checked for again? If I´m not misstaken, a prp-test would not find a factor? Could someone please explain in easy words the workflow of the cofactor work? (I´m neither a native english speaker nor a mathematician ) Thank you very much in advance.

2020-12-28, 14:23	#2
ATH Einyen Dec 2003 Denmark 101111011101₂ Posts	Cofactor PRP is one of our side projects. The main GIMPS project is to find Mersenne Primes by trial factoring candidates and then running a PRP test on exponents without a factor to check for probable primes followed by a certificate on the PRP test (new) or a PRP double check. After that the exponent is uninteresting for the main project, but then other people in side projects will run PRP test on the cofactor to find large PRPs and others will continue to try and find factors by trial factoring, P-1 tests and ECM curves. PRP on cofactors is still fairly new ~2-3 years and have "only" reached ~11M exponents, but is slowly going higher and higher and whenever a new factor is found below 11M the exponent is immediately scheduled for a PRP cofactor test, so everything is kept checked below the "wavefront". All the sideproject trial factoring, P-1 and ECM is not very organized instead people are working on their favorite ranges and method. There are a few people organizing their efforts in this thread, but right now the thread is "hijacked" by posts concerning testing a new Prime95 version 30.4b3 for P-1: https://mersenneforum.org/showthread.php?t=22476 Last fiddled with by ATH on 2020-12-28 at 14:24

2020-12-28, 17:25	#9
firejuggler Apr 2010 Over the rainbow 2·5·11·23 Posts	isn't that what the fully factored mersenne numer thread used for? edit : this thread https://www.mersenneforum.org/showthread.php?t=19407 Last fiddled with by firejuggler on 2020-12-28 at 17:32

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Same residue for cofactor with different number of known factors?	matzetoni	PrimeNet	3	2019-04-07 14:00
SNFS of C112 cofactor of 608^60+480^76	sean	YAFU	22	2017-05-20 17:05
Cofactor Search	carpetpool	Information & Answers	4	2016-12-12 04:35
Formula for cofactor for Fermat numbers.	literka	Factoring	7	2012-04-05 09:51
Certification of a 10064 bit prp cofactor	MatWur-S530113	PrimeNet	9	2010-02-16 12:52

2020-12-28, 15:04	#3
BigDukeSix Dec 2020 7 Posts	Thank you very much. This is a very nice explanation for somebody like me.

2020-12-28, 15:56	#4
Uncwilly 6809 > 6502 """"""""""""""""""" Aug 2003 101×103 Posts 2·3²·521 Posts	Just a little clarification. If we have a candidate number and we find a factor, the part that remains after dividing the number by the known factor is the co-factor. If we find several factors for the number, still what ever remains is the cofactor. We are running the PRP on that number to see if it is prime and thus, we known all of the prime factors for the candidate number. There are some numbers that have a PRP test run on the cofactors, but not a double check or a certificate on the PRP. (The ability to run that is recent). So there is a group of us working to get all of those double checks done. That way, going forward we should not have to do the longer (relative to the cert) DC's any more.

2020-12-28, 16:33	#6
Uncwilly 6809 > 6502 """"""""""""""""""" Aug 2003 101×103 Posts 2×3²×521 Posts	The residues are part of the record for the main exponent: https://www.mersenne.org/report_expo...9735347&full=1

2020-12-28, 16:43	#7
lavalamp Oct 2007 Manchester, UK 3²×149 Posts	Is there anywhere they are recorded separately such that one might peruse a list of them, in the same way as the mersenne primes are recorded separately. Once the exponent gets large enough (>1E4? >1E6?) I assume they are relatively rare finds.

2020-12-29, 16:26	#11
chris2be8 Sep 2009 5×401 Posts	M1399 shows as PRP on that page, but it's cofactor is proven prime in factordb. Although I didn't do it. Chris