Karim BELABAS on Tue, 22 Oct 2002 14:44:59 +0200 (MEST)


[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: polredabs() regression


On Fri, 18 Oct 2002, Igor Schein wrote:
> the following command keeps doubling the stack:
>
> polredabs(x^24-120*x^22+5689*x^20-144636*x^18+2206322*x^16-21100008*x^14+127153357*x^12-469833036*x^10+999219758*x^8-1108003080*x^6+610467037*x^4-157039836*x^2+14737921);
>
> This regression was first introduced in
>
> GP/PARI CALCULATOR Version 2.2.4 (development CHANGES-1.489)

A typo in LLL_cmbf (my variant of van Hoeij's factorizer): we have two
lattices called CM_L and M_L, supposedly related by CM_L = C * M_L; I was
updating CM_L, but M_L was updated just after it was needed...

In very rare cases this causes an infinite loop: we see no factors when the
right lattice is determined, because the previous M_L is used for the check,
and we never check again with a correct M_L since the lattice never changes
from this point on.

This is fixed in CVS.

    Karim.
-- 
Karim Belabas                    Tel: (+33) (0)1 69 15 57 48
Dép. de Mathematiques, Bat. 425  Fax: (+33) (0)1 69 15 60 19
Université Paris-Sud             Email: Karim.Belabas@math.u-psud.fr
F-91405 Orsay (France)           http://www.math.u-psud.fr/~belabas/
--
PARI/GP Home Page: http://www.parigp-home.de/