Re: dumb question about rationals

[Vincent Torri <Vincent.Torri@iecn.u-nancy.fr>]

On Wed, 8 Jun 2005, Bill Allombert wrote:

> On Wed, Jun 08, 2005 at 07:20:23AM +0200, Vincent Torri wrote:
> > Hello,
> >
> > i've just started to use PARI (that is, the library), and I want to use
> > the rational objects. But I have some problems.
> > My program is quite simple :
> >
> >    GEN r1;
> >    GEN r2;
> >
> >    pari_init (1000000, 2);
> >
> >    r1 = cgetg (3, t_FRAC);
> >    r2 = cgetg (3, t_FRAC);
> >
> >    r1[1] = (long)2;
> >    r1[2] = (long)2;
> >
> >    r2[1] = (long)2;
> >    r2[2] = (long)2;
> >
> >    r2 = gadd (r2, r1);
> >
> > but, when i launch the program, i get:
> >
> >    ***   segmentation fault: bug in PARI or calling program.
> >    ***   Error in the PARI system. End of program.
> >
> > Could you please tell me why there is this problem, and where my error(s)
> > is (are)
>
> 'r1[1] = (long)2;' is not correct, you must convert the C long integer 2 to a
> PARI GEN integer using stoi().
> 'r1[1] = (long)stoi(2);' should work.
>
> Secondly, the numerator and the denominator must be coprime else you
> build a incorrect PARI object, so 2/2 is not allowed.
>
> Usually it is simpler to use gdiv() to build rational numbers.
>
> I suggest you try out GP2C. This program convert GP code to C code, so
> you can use it to generate example code you can reuse.
>
> For example, what you are doing is equivalent to the GP code
> r1=2/2;
> r2=2/2;
> r2=r1+r2;
>
> GP2C will translate it to the following C code:
>
>  r1 = gdivgs(gdeux, 2);
>  r2 = gdivgs(gdeux, 2);
>  r2 = gadd(r1, r2);

Ok, thank you for your quick answer.

My aim is to construct rationals from any long. It seems that
 r = gdiv (stoi (l1), stoi (l2));
is working (l1 and l2 are long int).

In that case, should I construct r, that is, should I call cgetg ?
If no, what is the precision of the rational ? May i change it after this 
operation ?

Also, I've not well undestood the use of cgetg.

cgetg (N, t_FRAC) allocate a rational whose components (numerator and 
denominator) are coded on N*32 bits, right ?

In case that there is an overflow, does PARI increase the precision of the 
rational itself, or is there an error message (or something else) ?

I ask that because the computations I do lead to intermediate rationals 
with large coprime numerator and denominator (I don't know the limit of 
these numbers).

Thank you

Vincent