Karim Belabas on Sat, 15 Feb 2014 01:51:01 +0100


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

Re: Large errors for eint()


* Jeroen Demeyer [2014-02-13 16:12]:
> On 2014-02-11 12:25, Jeroen Demeyer wrote:
> >On 2013-11-09 23:36, Karim Belabas wrote:
> >>1) For eint1, it was an oversight (1-word cancellation just before
> >>returning the result, plus a few other places where we computed with one
> >>word less than intended). This one is now fixed in commit f0ed0cd2
> >Unfortunately, there has been a regression again since f0ed0cd2.
> 
> I have bisected this to
> 
> commit d0f0585dd7ecbff44b8083215af007275421bde7
> Author: Karim Belabas <Karim.Belabas@math.u-bordeaux1.fr>
> Date:   Thu Dec 19 18:30:04 2013 +0100
> 
>     57- eint1 misbehaved at large precision [#1499]

Fixed in master now. Again catastrophic cancellation: this commit let
the expansion around 0 be used in a larger region, where continued
fraction was much slower; unfortunately, in that new region, the number of
cancelled digits (when using expansion around 0) is no longer bounded,
as it used to be. So we need to take that into account...

Thanks for your report !

    K.B.
--
Karim Belabas, IMB (UMR 5251)  Tel: (+33) (0)5 40 00 26 17
Universite Bordeaux 1          Fax: (+33) (0)5 40 00 69 50
351, cours de la Liberation    http://www.math.u-bordeaux1.fr/~kbelabas/
F-33405 Talence (France)       http://pari.math.u-bordeaux1.fr/  [PARI/GP]
`