Discussion related to cado-nfs

[Pierrick Gaudry <pierrick.gaudry@loria.fr>]

Hi again,

Here is a follow-up on this issue of computing dlp in GF(p^2), and how to
pass a target that should, indeed, be a degree 1 polynomial.

Since this is very experimental, automatic tests are not fully in place,
and at some point, a commit broke something that was working (but not
documented). This is commit 26ab2734b.

This won't be fixed soon, I'm afraid, because it requires some time. But
a workaround exists, because, really, this is just the glueing between
the various scripts that is broken.

Here is an example:

Assume you want to compute the dlp of the element "5,7" (these are the
coefficients of the polynomial 5+7x) in GF(p^2), where
p=100000000000000000039, modulo 164354743277891 that divides p+1.

Then, you can type the following:

  ./cado-nfs.py -dlp -gfpext 2 -ell 164354743277891 target=5,7 
100000000000000000039

This will fail, because right now the main script recognizes this syntax
as a list of targets (5, then 7), which is ok for prime fields, but
contradicts the syntax for extension fields.

However, it gives the command line that fails. This is the one starting
with something like:
  
  /path_to_build/scripts/descent.py --target 5 --gfpext 2 --prefix p2dd20 
[...]

As you can see, only "5" is passed as a target to the descent script,
instead of "5,7". It is then possible to copy-paste the full
command-line, and replace "5" by "5,7", and this should finish the work.

Regards,
Pierrick