GWNUM library

bearnol · 2006-02-22, 19:10

I'm excited by the GWNUM library.
(I'm using GMP-ECM to try and factor numbers of the form 2^p+1)
Is it freely available? Where can I get hold of it, please?
J

Prime95 · 2006-02-22, 19:34

There are two answers:

1) If you want to link gwnum into gmp-ecm (using the as-yet-unreleased next version of GMP-ECM), then you need a GPL version of gwnum. Look at ftp://mersenne.org/gimps. There are 3 GPL'ed libraries there. You are an early adopter so be prepared for problems.

2) If you want to write your own software, the gwnum library is part of the Mersenne sources. Go to http://mersenne.org/source.htm

If you are ECMing, the safest solution right now is to poke around in the factoring threads for instructions on how to let prime95 do stage 1 and GMP-ECM do stage 2.

bearnol · 2006-02-22, 19:40

Thanks very much, for what looks like some very good and helpful advice. I'll check it out...
J

bearnol · 2006-02-23, 16:25

Here are my timings. Perhaps someone could please confirm this is as expected (note the overall time for the curve is about the same)?
[I'm sticking to trying to use GMP-ECM alone for the moment, since I want to use ECMNet - and I'm not sure one can use mprime in conjunction with ECMNet?]
Perhaps also there is some hidden benefit (in the new version) I'm missing?
Or is my input format correct - I'm just giving the program the raw integer, and hoping/assuming it'll be able to recognise its form (2^p+1) and hence apply GWNUM, but maybe this is incorrect?
[please feel free to move this post/thread to the GMP-ECM section if seems more appropriate...?]
Thanks,
J

[james@tosh gmp-ecm]$ cp ecm6.0.1 ecm
[james@tosh gmp-ecm]$ ./ecmloop.jgw 250000 1 P19937.txt
GMP-ECM 6.0.1 [powered by GMP 4.1.2] [ECM]
Input number has 6002 digits
Using B1=250000, B2=116469998, polynomial Dickson(3), sigma=486772376
Step 1 took 2731326ms
Step 2 took 1133302ms
[james@tosh gmp-ecm]$ cp ecm6.1gwnum ecm
[james@tosh gmp-ecm]$ ./ecmloop.jgw 250000 1 P19937.txt
GMP-ECM 6.1 [powered by GMP 4.1.2 and GWNUM 24.11] [ECM]
Input number has 6002 digits
Using B1=250000, B2=258157000, polynomial Dickson(3), sigma=2344265460
Step 1 took 106655ms
Step 2 took 3897080ms

Prime95 · 2006-02-23, 16:48

I think there are some special command line arguments that tell gmp-ecm it can use the gwnum library for stage 1 - or maybe you have to use the fomula 2^19937+1 rather than the decimal expansion.

Contact akruppa for help

akruppa · 2006-02-24, 09:45

Base 2 numbers should be recognized by GMP-ECM. Due to exams, I can't look into this further atm, I'll get back to it later.

Alex

bearnol · 2006-02-24, 11:19

Hi Alex,
Thanks very much for your response, and offer of further help. I very much appreciate it - please don't rush... and "Good Luck!" in your exams
James

2006-02-22, 19:10	#1
bearnol Sep 2005 127 Posts	GWNUM library I'm excited by the GWNUM library. (I'm using GMP-ECM to try and factor numbers of the form 2^p+1) Is it freely available? Where can I get hold of it, please? J

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Prime numbers library	Martin	Lounge	12	2017-12-02 17:18
Using YAFU as a C-library	LangerJan	YAFU	2	2013-02-17 06:18
factoring with MIRACL library	prss	Factoring	22	2011-06-15 20:25
Quickest fft library	nuggetprime	Software	3	2011-01-09 01:24
GWNUM library and llr	leizhoucn	Programming	2	2007-11-05 09:34

2006-02-22, 19:34	#2
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 1FDE₁₆ Posts	There are two answers: 1) If you want to link gwnum into gmp-ecm (using the as-yet-unreleased next version of GMP-ECM), then you need a GPL version of gwnum. Look at ftp://mersenne.org/gimps. There are 3 GPL'ed libraries there. You are an early adopter so be prepared for problems. 2) If you want to write your own software, the gwnum library is part of the Mersenne sources. Go to http://mersenne.org/source.htm If you are ECMing, the safest solution right now is to poke around in the factoring threads for instructions on how to let prime95 do stage 1 and GMP-ECM do stage 2.

2006-02-22, 19:40	#3
bearnol Sep 2005 127 Posts	Thanks very much, for what looks like some very good and helpful advice. I'll check it out... J

2006-02-23, 16:25	#4
bearnol Sep 2005 7F₁₆ Posts	Here are my timings. Perhaps someone could please confirm this is as expected (note the overall time for the curve is about the same)? [I'm sticking to trying to use GMP-ECM alone for the moment, since I want to use ECMNet - and I'm not sure one can use mprime in conjunction with ECMNet?] Perhaps also there is some hidden benefit (in the new version) I'm missing? Or is my input format correct - I'm just giving the program the raw integer, and hoping/assuming it'll be able to recognise its form (2^p+1) and hence apply GWNUM, but maybe this is incorrect? [please feel free to move this post/thread to the GMP-ECM section if seems more appropriate...?] Thanks, J [james@tosh gmp-ecm]$ cp ecm6.0.1 ecm [james@tosh gmp-ecm]$ ./ecmloop.jgw 250000 1 P19937.txt GMP-ECM 6.0.1 [powered by GMP 4.1.2] [ECM] Input number has 6002 digits Using B1=250000, B2=116469998, polynomial Dickson(3), sigma=486772376 Step 1 took 2731326ms Step 2 took 1133302ms [james@tosh gmp-ecm]$ cp ecm6.1gwnum ecm [james@tosh gmp-ecm]$ ./ecmloop.jgw 250000 1 P19937.txt GMP-ECM 6.1 [powered by GMP 4.1.2 and GWNUM 24.11] [ECM] Input number has 6002 digits Using B1=250000, B2=258157000, polynomial Dickson(3), sigma=2344265460 Step 1 took 106655ms Step 2 took 3897080ms

2006-02-23, 16:48	#5
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 17736₈ Posts	I think there are some special command line arguments that tell gmp-ecm it can use the gwnum library for stage 1 - or maybe you have to use the fomula 2^19937+1 rather than the decimal expansion. Contact akruppa for help

2006-02-24, 09:45	#6
akruppa "Nancy" Aug 2002 Alexandria 2,467 Posts	Base 2 numbers should be recognized by GMP-ECM. Due to exams, I can't look into this further atm, I'll get back to it later. Alex

2006-02-24, 11:19	#7
bearnol Sep 2005 1111111₂ Posts	Hi Alex, Thanks very much for your response, and offer of further help. I very much appreciate it - please don't rush... and "Good Luck!" in your exams James