[Cado-nfs-discuss] dlp

Pierrick Gaudry pierrick.gaudry at loria.fr
Wed Oct 26 08:04:42 CEST 2016


On Tue, Oct 25, 2016 at 09:02:35PM +0300, Христофор Бобров wrote:
> Hi,
> We are trying to find dlp for 100 digits number.
> We chose a polynomial with parameters
> 
> tasks.polyselect.degree = 5
> tasks.polyselect.P = 5000000
> tasks.polyselect.admax = 2e8
> tasks.polyselect.adrange = 1e4
> tasks.polyselect.incr = 60
> tasks.polyselect.nq = 1000
> tasks.polyselect.nrkeep = 2000
> 
> What are the recommendations for selection of the parameters for the other stages?
> 
> Thanks a lot!
> 
> Bobrov.

Hi,

Degree 5 is probably too large for a 100 digits dlp.
I recommend degree 4.

For the other stages, I can't say just now what is the best choice.
I will try a few things and commit them to the development version (maybe
today or tomorrow). Now that dlp is fully operational, it make sense to
have more parameter sets. Certainly not optimal ones, but things that
work...

Regards,
Pierrick

PS: could you avoid html formatting in your messages on the mailing-list?


More information about the Cado-nfs-discuss mailing list