Reporting small factors to PRP doesn't work

DukeBG · 2018-03-28, 15:06

Hello!

I'll start from an example. this number is currently marked as PRP. It also has a quality of being divisible by 2699701957. However, reporting this small factor doesn't change it into CF and the number remains PRP instead.

This is not the first time I've encountered this behavior. Checking it against a base in the Primality section helps turn it into C, and then the factor is added correctly. However, this extra step shouldn't be required.

(I have few more such examples if needed).

I've tried emailing the address specified on the "Imprint" page, but didn't recieve any reply in weeks. Can anyone advise a correct way to report the bug to an appropriate person?

chris2be8 · 2018-03-28, 16:04

I reported some problems to that address last year and they were eventually fixed, after several weeks. So that's the right address, but don't hold your breath waiting for a reply.

Chris

MisterBitcoin · 2018-03-28, 16:33

Quote:

Originally Posted by DukeBG

Hello!

I'll start from an example. this number is currently marked as PRP. It also has a quality of being divisible by 2699701957. However, reporting this small factor doesn't change it into CF and the number remains PRP instead.

This is not the first time I've encountered this behavior. Checking it against a base in the Primality section helps turn it into C, and then the factor is added correctly. However, this extra step shouldn't be required.

(I have few more such examples if needed).

I've tried emailing the address specified on the "Imprint" page, but didn't recieve any reply in weeks. Can anyone advise a correct way to report the bug to an appropriate person?

Please feel free to report any problems you encounter with similiar problems. Seems like there is an bug in the code that prevents a PRP turning to CF if you add a factor.

I also did some ECM on the cofactor and found with B1=4500 a P12 factor.

DukeBG · 2018-03-28, 17:04

Quote:

Originally Posted by MisterBitcoin

Please feel free to report any problems you encounter with similiar problems. Seems like there is an bug in the code that prevents a PRP turning to CF if you add a factor.

By "reporting" do you mean writing an email to that address or posting to these forums?

MisterBitcoin · 2018-03-28, 19:26

Quote:

Originally Posted by DukeBG

By "reporting" do you mean writing an email to that address or posting to these forums?

I´ve tryed to reach him multible time via mail, no responce.

You can do booth.

It´s better for all.

I´ll soon do a Primo test for a 12dd PRP, to see how long it will take. Imagine you waste a few CPU-days only to find out that this number is composite and had an small 10 digit factor.

That would hurt.

DukeBG · 2018-04-01, 20:13

Found and "fixed" 13 such PRPs with 10-digit factors now. The smallest was 4547 digits.

MisterBitcoin · 2018-04-01, 22:21

Quote:

Originally Posted by DukeBG

Found and "fixed" 13 such PRPs with 10-digit factors now. The smallest was 4547 digits.

13 prp´s with factors? Ouch.
Did they have any special form (rep-digit; primorial/factorial) or just co-factors with no short term? Maybe a bug in pfgw.

Would be good to know, also please post them.

DukeBG · 2018-04-03, 10:31

Quote:

Originally Posted by MisterBitcoin

13 prp´s with factors? Ouch.
Did they have any special form (rep-digit; primorial/factorial) or just co-factors with no short term? Maybe a bug in pfgw.

Would be good to know, also please post them.

Not sure why you mention pfgw. Is there evidence that FactorDB is using it "behind the scenes"?

I feel like FactorDB's PRP is just a "weak" probable prime test, not strong Miller-Rabin and definitely not strong Lucas test.

MisterBitcoin · 2018-04-03, 14:28

Quote:

Originally Posted by DukeBG

Not sure why you mention pfgw. Is there evidence that FactorDB is using it "behind the scenes"?

I feel like FactorDB's PRP is just a "weak" probable prime test, not strong Miller-Rabin and definitely not strong Lucas test.

See here.
FDB is using PFGW for primality and factored N+1/N-1-tests. Notice: PFGW can only test for probable prime, if N-1/N+1 side is not deeply enough factored.
Here is an examble.

vebis · 2018-05-02, 08:11

I add factors for 10666 c109s today.

vebis · 2018-05-02, 11:16

Quote:

Originally Posted by vebis

I add factors for 10666 c109s today.

wrong thread

2018-03-28, 15:06	#1
DukeBG Mar 2018 3·43 Posts	Reporting small factors to PRP doesn't work Hello! I'll start from an example. this number is currently marked as PRP. It also has a quality of being divisible by 2699701957. However, reporting this small factor doesn't change it into CF and the number remains PRP instead. This is not the first time I've encountered this behavior. Checking it against a base in the Primality section helps turn it into C, and then the factor is added correctly. However, this extra step shouldn't be required. (I have few more such examples if needed). I've tried emailing the address specified on the "Imprint" page, but didn't recieve any reply in weeks. Can anyone advise a correct way to report the bug to an appropriate person?

2018-05-02, 08:11	#10
vebis Oct 2015 44₁₆ Posts	I add factors for 10666 c109s today. Last fiddled with by vebis on 2018-05-02 at 08:12

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
GPU72 doesn't show all available DC work	Mark Rose	GPU to 72	2	2014-01-24 00:33
mfaktc doesn't work for me	sonjohan	GPU Computing	11	2012-01-27 15:10
Reporting sequences with work done	10metreh	Aliquot Sequences	7	2009-10-30 01:08
Desktop doesn't work.	clowns789	Software	7	2003-10-19 21:33
SkipTrialFactoring=1 doesn't work?	cmokruhl	Software	1	2002-10-15 19:04

2018-03-28, 16:04	#2
chris2be8 Sep 2009 2×1,039 Posts	I reported some problems to that address last year and they were eventually fixed, after several weeks. So that's the right address, but don't hold your breath waiting for a reply. Chris

2018-04-01, 20:13	#6
DukeBG Mar 2018 201₈ Posts	Found and "fixed" 13 such PRPs with 10-digit factors now. The smallest was 4547 digits.