Discussion related to cado-nfs

[Emmanuel Thomé <Emmanuel.Thome@inria.fr>]

I just added a consistency check. It's there:

    https://gitlab.inria.fr/cado-nfs/cado-nfs/-/merge_requests/36

The problem was that previously, the code was skipping all sorts of
checks if it found a small factor of N while trying to compute the common
mapping.

Here's now what I get when I follow the procedure that you describe:

localhost ~/NFS/cado $ ./cado-nfs.py 
181570647363941951922859674104963771103015312977807439204421953197794821673476296854978179746096739603508598016332617111965534228905323443567644055007324027034168949
 tasks.polyselect.import=tests/utils/buggy.poly
[...]
Error:Generate Factor Base: Program run on server failed with exit code -6
Error:Generate Factor Base: Command line was: 
/home/thome/NFS/cado/build/houblon/sieve/makefb -poly 
/tmp/cado.fepdhxwq/c165.poly -lim 53279688 -maxbits 14 -out 
/tmp/cado.fepdhxwq/c165.roots1.gz -t 4 > 
/tmp/cado.fepdhxwq/c165.makefb.stdout.1 2> 
/tmp/cado.fepdhxwq/c165.makefb.stderr.1
Error:Generate Factor Base: Stderr output (last 10 lines only) follow (stored 
in file /tmp/cado.fepdhxwq/c165.makefb.stderr.1):
Error:Generate Factor Base:     Error, non-trivial factors (3) of N were 
found while checking the poly file. This should not happen.
Error:Generate Factor Base:     Error, we found no compatible mapping that 
factors through the given number fields. Your poly file is most probably 
wrong.
Error:Generate Factor Base:     code BUG() : condition 
cado_poly_check_mapping(((void *)0), poly, poly->n) failed in 
cado_poly_set_plist at /home/thome/NFS/cado/utils/cado_poly.c:128 -- Abort
Error:Generate Factor Base:
Traceback (most recent call last):
  File "/home/thome/NFS/cado/./cado-nfs.py", line 122, in <module>
    factors = factorjob.run()
  File "./scripts/cadofactor/cadotask.py", line 6131, in run
    last_status = task.run()
  File "./scripts/cadofactor/cadotask.py", line 3009, in run
    raise Exception("Program failed")
Exception: Program failed


Yes, there are two distinct errors. In my opinion, both are a good reason
to abort the computation.

E.



On Tue, Apr 27, 2021 at 06:41:36AM +0200, eric.jeancolas@free.fr wrote:
> Dear Paul
> 
> I explain here what I did to test, to be sure I did what you expect.
> I downloaded a new version of cado-nfs from 
> https://gitlab.inria.fr/cado-nfs/cado-nfs/-/commits/7694b91283f11131ec54fe19c7d3ff8cdf8b97ad/
> On my conputer, I erased previous cado directories, and I made a tar on 
> cado-nfs-7694b91283f11131ec54fe19c7d3ff8cdf8b97ad.tar.gz and a make.
> I issued the following command: ./cado-nfs.py 
> 181570647363941951922859674104963771103015312977807439204421953197794821673476296854978179746096739603508598016332617111965534228905323443567644055007324027034168949
>  tasks.polyselect.import=18888_164.poly
> 
> The contents of 18888_164.poly is:
> n: 
> 181570647363941951922859674104963771103015312977807439204421953197794821673476296854978179746096739603508598016332617111965534228905323443567644055007324027034168949
> skew: 1.36
> type: snfs
> c0: -80
> c5: 17
> Y0: 1000000000000000000000000000000000
> Y1: -1
> # f(x) = 17*x^5-80
> # g(x) = -x+1000000000000000000000000000000000
>  
> I reached the following steps:
> Info:Generate Factor Base: Starting
> Info:Generate Factor Base: Finished
> Info:Generate Factor Base: Total cpu/real time for makefb: 22.1/19.6028
> Info:Generate Free Relations: Starting
> Info:Generate Free Relations: Found 1410662 free relations
> Info:Generate Free Relations: Finished
> Info:Generate Free Relations: Total cpu/real time for freerel: 
> 462.44/319.124
> Info:Lattice Sieving: Starting
> Info:Lattice Sieving: We want 63458945 relation(s)
> Info:Lattice Sieving: Adding workunit c165_sieving_30592187-30595000 to 
> database
> Info:Lattice Sieving: Adding workunit c165_sieving_30595000-30600000 to 
> database
> 
> When I have a mismatch on a number, I don't reach Lattice Sieving.
> 
> So I stopped the process (Ctrl+C) just after that.
> 
> Just to recall here: the number 
> 181570647363941951922859674104963771103015312977807439204421953197794821673476296854978179746096739603508598016332617111965534228905323443567644055007324027034168949
>  has been erroneously generated after some star removals.
> The number to be factorized is 18888_164 = (17*10^164-8)/9 = 
> (17*(10^33)^5-80)/90 = 
> 181*5706473639419519*22859674104963771103015312977807439204421953197794821673476296854978179746096739603508598016332617111965534228905323443567644055007324027034168949
> ( and eventually = 
> 181*5706473639419519*7159052365561448148227529365476870909708139900919265573*3193114526571981969989829734950032818678374603944682479503444963111309977900602192542410513
>  )
> The snfs polynomial has been generated accordingly.
> 
> Can you reproduce the problem, or is it related to my installation?
> 
> Best regards.
> 
> Eric
> 
> 
> ----- Mail original -----
> De: "Paul Zimmermann" <Paul.Zimmermann@inria.fr>
> À: "eric jeancolas" <eric.jeancolas@free.fr>
> Cc: "pierrick gaudry" <pierrick.gaudry@loria.fr>, cado-nfs@inria.fr
> Envoyé: Lundi 26 Avril 2021 14:51:07
> Objet: Re: [cado-nfs] sqrt error
> 
>        Dear Eric,
> 
> are you sure you used the exact revision I mention below (git co 7694b91)?
> I had to revert my commit later since it broke some tests.
> 
> Paul
> 
> > Date: Mon, 26 Apr 2021 13:18:12 +0200 (CEST)
> > From: eric.jeancolas@free.fr
> > Cc: pierrick gaudry <pierrick.gaudry@loria.fr>, cado-nfs@inria.fr
> > 
> > Dear Paul,
> > I reinstalled cado and fiel timestamps are 26/04/2021 11:19.
> > The problem is still the same, the number is not viewed as an error 
> > before sieving is starting.
> > Best regards.
> > Eric
> > 
> > ----- Mail original -----
> > De: "Paul Zimmermann" <Paul.Zimmermann@inria.fr>
> > À: "eric jeancolas" <eric.jeancolas@free.fr>
> > Cc: "pierrick gaudry" <pierrick.gaudry@loria.fr>, cado-nfs@inria.fr
> > Envoyé: Lundi 26 Avril 2021 09:04:19
> > Objet: Re: [cado-nfs] sqrt error
> > 
> >        Dear Eric,
> > 
> > please try the latest commit (7694b91) and please confirm it gives an 
> > error
> > during the first check process.
> > 
> > Best regards,
> > Paul