Karim BELABAS on Tue, 18 Jun 2002 15:41:34 +0200 (MEST)

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

Re: nffactor() regression

On Tue, 18 Jun 2002, Karim BELABAS wrote:

> ? nffactor(nfinit(polsubcyclo(17,8,y)),x^48-104*x^46+4664*x^44-122476*x^42+2137838*x^40-26567700*x^38+245144964*x^36-1725955872*x^34+9441692003*x^32-40611588 644*x^30+138356971048*x^28-374714866240*x^26+807289826646*x^24-1380693858220*x^22+1866021172640*x^20-1978766780068*x^18+1630151673857*x^16-1029505301024*x^14 +489498952012*x^12-170832297056*x^10+42133382284*x^8-6904507136*x^6+669868016*x^4-28899680*x^2+16)
>   ***   bug in LLL_cmbf [no factor], please report

OK, removing 0 columns in LLL_check_progress() was not the brightest thing to
do (even though the projected lattice remains the same, we may lose small
vectors in the original lattice).

But I notice I had forgotten to impose a minimal precision when truncating
entries in van Hoeij's lattice [ this is already enforced over Z ]. And I had
not tested this code with so few modular factors.

Imposing at least 32 bits of precision (might be overkill) fixes this
problem, as well as the original one.

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/