Very newbish questions

[drf]

Hello all,
I've only just started trying to use Msieve and possibly GGNFS to factor a number of integres roughly in the <116 digit range. There are not RSA type numbers so quite often some small factors will be found reducing the size to ~105-110 digits.

I have downloaded both Msieve and GGNFS as well as the factmsieve.py script. I have built everything from source using the latest stable releases (1.52 msieve, 0.77.1 GGNFS) without any issues on Ubuntu 14.04 LTS. So far so good.

My issue comes when I try to get msieve to perform factorization using gnfs. I haven't quite figured out how to do that using the msieve aplication built form demo.c. Reading the factmsieve.py script I think I might have figured out how to perform gnfs using msieve's flags to perform one stage at a time. Unfortunately when I try that, after successfully generating polynomials, trying to perform sieving (-ns) gives me an assert error in malloc and the whole process dies.

Trying to use the factmsieve.py script I again managed to get it to generate polynomials, but when the ggnfs sieving part gets called it ends up exiting with an error code of -11 and the script terminates.

I have sadly very little knowledge about the gnfs algorithm so it is proving quite difficult to figure out what might be going wrong. I would be very grateful for any hints or pointers.

Thanks,
Drf.

[R.D. Silverman]

Originally Posted by drf

I have sadly very little knowledge about the gnfs algorithm so it is proving quite difficult to figure out what might be going wrong. I would be very grateful for any hints or pointers.

Thanks,
Drf.

You gave yourself the hint. Don't be lazy. Take the time to learn how the algorithm works.
There are plenty of sources.

[henryzz]

Originally Posted by drf

Hello all,
I've only just started trying to use Msieve and possibly GGNFS to factor a number of integres roughly in the <116 digit range. There are not RSA type numbers so quite often some small factors will be found reducing the size to ~105-110 digits.

I have downloaded both Msieve and GGNFS as well as the factmsieve.py script. I have built everything from source using the latest stable releases (1.52 msieve, 0.77.1 GGNFS) without any issues on Ubuntu 14.04 LTS. So far so good.

My issue comes when I try to get msieve to perform factorization using gnfs. I haven't quite figured out how to do that using the msieve aplication built form demo.c. Reading the factmsieve.py script I think I might have figured out how to perform gnfs using msieve's flags to perform one stage at a time. Unfortunately when I try that, after successfully generating polynomials, trying to perform sieving (-ns) gives me an assert error in malloc and the whole process dies.

Trying to use the factmsieve.py script I again managed to get it to generate polynomials, but when the ggnfs sieving part gets called it ends up exiting with an error code of -11 and the script terminates.

I have sadly very little knowledge about the gnfs algorithm so it is proving quite difficult to figure out what might be going wrong. I would be very grateful for any hints or pointers.

Thanks,
Drf.

For starters that source package for ggnfs is ancient. Use the svn source instead. factmsieve.py is much newer than 0.77.1 and might not work with binaries that old.

[EdH]

(I see Henryzz beat me to some of this.)

I wonder if you might have installed ggnfs from sourceforge. I (and another) have had trouble with that package. Instead I use the git-hub version (referenced on my page below).

I have a step-by-step page to install several packages (including ggnfs and msieve) on an Ubuntu system at the following link, in case you might like to review it:

Steps to install and set up Aliqueit on an Ubuntu* (or antiX) computer

There is also a thread here that has a self-install script:

Build Instructions for Ubuntu 14.04

You should review the entire thread.

[VBCurtis]

Or you could take the lazy (software-wise) way out and grab the binaries for GGNFS sievers from the top post of the M991 team factoring thread in the factoring forum. For a beginner, understanding what msieve's flags do is a nice start and usually sufficient grasp to get going on factorizations. There are many parameters in the factmsieve script for improved efficiency, but those don't matter until yo uhave working software!

[R.D. Silverman]

Originally Posted by VBCurtis

Or you could take the lazy

<snip>

There are many parameters in the factmsieve script for improved efficiency, but those don't matter until yo uhave working software!

Parameters don't matter unless you understand the algorithm.

[xilman]

Originally Posted by R.D. Silverman

Parameters don't matter unless you understand the algorithm.

I'm with Bob's approach on this question, except that I would point out that frobbing the knobs on a black box can make it work differently well. Sometimes it gets better and sometimes it don't.

You've gotta ask yourself one question: "Do I feel lucky?" Well, do ya, punk?

[drf]

Thank you for the responses. The problem ended up being with the wrong version of ggnfs. I still haven't figured out how to convince msieve to actually perform a gnfs factorization on it's own though.
Drf

[R.D. Silverman]

Originally Posted by drf

Thank you for the responses. The problem ended up being with the wrong version of ggnfs. I still haven't figured out how to convince msieve to actually perform a gnfs factorization on it's own though.
Drf

I thought that msieve only had a QS siever. Am I wrong?

[jyb]

Originally Posted by R.D. Silverman

I thought that msieve only had a QS siever. Am I wrong?

Yes. It has an NFS line siever, but Jason has long recommended that it not be used anymore since, well, it's a line siever.

[henryzz]

Originally Posted by jyb

Yes. It has an NFS line siever, but Jason has long recommended that it not be used anymore since, well, it's a line siever.

I think it has been broken for a while as well.