Karim BELABAS on Wed, 5 Feb 2003 11:52:27 +0100 (MET)

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

Re: polredabs() observation

On Tue, 4 Feb 2003, Igor Schein wrote:
> It's still pretty much broken.  Consider the following trivial cases:
> polredabs(x^4+10^80*x^2+1,16)
> precision issue, takes years
> polredabs(x^8+10^80*x^4+1,16)
> too many vectors, also takes years
> Both used to finish in no time.

OK, my fix from yesterday was too optimistic. It assumed that overestimating
the bound by at most 1.e-4% was not a big deal, and we could operate on
machine floats. I was wrong.

In your examples above, you quickly get 1e40 more vectors in this way [and
much larger bounds than expected due to catastrophic cancellation...]

Better now?

Karim Belabas                    Tel: (+33) (0)1 69 15 57 48
Dép. de Mathématiques, Bât. 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/