Re: Bug#40: bnrinit() bug

[Karim Belabas <Karim.Belabas@math.u-psud.fr>]

* Igor Schein [2004-05-31 07:48]:
> Package: pari
> Version: 2.2.8 (development CHANGES-1.944)
> Severity: normal
> 
> \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
>    realprecision = 28 significant digits
>    seriesprecision = 16 significant terms
>    format = g0.28
> 
> parisize = 8000000, primelimit = 500000
> ? bnrinit(bnfinit(y^16-232*y^14+17564*y^12-592696*y^10+10090294*y^8-87937112*y^6+356253116*y^4-482477960*y^2+143400625),60,1);
>   *** bnrinit: impossible inverse modulo: Mod(5, 300).
> \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\

Fixed in CVS. A typo in recent primedec() code [ some computation mod p,
should have been mod p^2 ]. I have added some missing garbage collection as
well.

Note to myself: bnrinit(,,1) insists on finding generators as ideals with
the right "signature", which is a very bad idea. It would be (much) better
to represent an element of Cl_f(K) by a collection of signs at the infinite
places of the divisor f, and an ideal coprime to the finite part of f [ in
compact representation, e.g a partial factorisation into principal ideals
arising from successive products and LLL reductions ]. And the code would
be simpler.

This would break backward compatibility in a serious way though. In the
same way as using compact representations for isprincipal() output or
fundamental units would.

Cheers,

    Karim.

> P.S.
> 
> How come the first 3 default lines are indented, then a blank line,
> then an unindented line with 2 defaults on it?  A bit of a visual
> dissonance. 

History [ there were no "defaults" up to PARI 1.39.16 (released as 1.900 !) ].

These 3 lines don't serve any real purpose:
-- one can query them individually, or ask for the values of all defaults
-- among the defaults, these 3 are not especially useful on startup, and
they can't be set via the command line (contrary to the other 2).

I have removed them to reduce screen clutter.
-- 
Karim Belabas                     Tel: (+33) (0)1 69 15 57 48
Dep. de Mathematiques, Bat. 425   Fax: (+33) (0)1 69 15 60 19
Universite Paris-Sud              http://www.math.u-psud.fr/~belabas/
F-91405 Orsay (France)            http://pari.math.u-bordeaux.fr/  [PARI/GP]