Discussion related to cado-nfs

[eric.jeancolas@free.fr]

Well, I understand better the problem.

In a preprocessing process, some "*" disappeared. So at the beginning, it's 
of course a mistake from my side.
The number 
181570647363941951922859674104963771103015312977807439204421953197794821673476296854978179746096739603508598016332617111965534228905323443567644055007324027034168949
 should have been indeed
181*5706473639419519*22859674104963771103015312977807439204421953197794821673476296854978179746096739603508598016332617111965534228905323443567644055007324027034168949
 and I should have submitted only the last one.
Of course, all 3 numbers divides the polynomial.
The erroneous number is 
181*10^162+5706473639419519*10^146+22859674104963771103015312977807439204421953197794821673476296854978179746096739603508598016332617111965534228905323443567644055007324027034168949
What is strange is that this number doesn't appear as erroneous in the first 
check process. When a number is not the right one, I usually get something 
like:
Error:Generate Factor Base:     Error, N does not divide resultant of given 
polynomials

Not here...

So it seems that the sqrt check process is more accurate. And of course, the 
sqrt check process is right. Maybe it's worth reinforce the first check 
process.

Best regards.

Eric


----- Mail original -----
De: "Pierrick Gaudry" <pierrick.gaudry@loria.fr>
À: "eric jeancolas" <eric.jeancolas@free.fr>
Cc: cado-nfs@inria.fr
Envoyé: Lundi 26 Avril 2021 07:44:36
Objet: Re: [cado-nfs] sqrt error

Morning,

In this file, n (a.k.a N) should be the number you want to factor (or a
small multiple of it). It seems that this is not the case in the example
below. I think that (for once) the current error message is not so bad:
what is really needed for things to work is precisely that the given N
divides the resultant of the given polynomials.

Changing the line with
n: 
169999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999920
should work.

Regards,
Pierrick

On Mon, Apr 26, 2021 at 07:24:08AM +0200, eric.jeancolas@free.fr wrote:
> Hello
> 
> I got the following error in the last step of a process:
> Error:Square Root: Error, N does not divide resultant of given polynomials, 
> or there are multiplicities.
> Error:Square Root: Can not compute m.
> I'm in a snfs mode.
> The polynomial is as follows:
> n: 
> 181570647363941951922859674104963771103015312977807439204421953197794821673476296854978179746096739603508598016332617111965534228905323443567644055007324027034168949
> skew: 1.36
> type: snfs
> c0: -80
> c5: 17
> Y0: 100000000000000000000000000000000
> Y1: -1
> 
> Which parameter can I tune to prevent this error?
> If such a parameter is easy to tune, could you add a line on the Error 
> message to give such an advice? With something like:
> Error:Square Root: This is due to blah blah, try a larger tasks.blah.blah 
> parameter...
> 
> Best regards.
> Éric