"Brian J. Beesley" wrote:
> On 25 May 00, at 23:39, Ken Kriesel wrote:
>
> > I saw something similar once, out of hundreds of ECM curves
> > that I've run.
> > It happened on the curve being run while I adjusted memory limits
> > on the program (Version 20.4.1, April 25 file date), if I recall correctly.
>
> There was a problem with some versions which resulted in this error
> message appearing constantly. I think the problem was related to a
> factor which was already known being found again - this happens
> constantly but should be resolved by reference to the low?.txt file.
I got the same error in a F14 curve (I jumped out of my seat). George told me
that it
was a bug in v20.1 where the GCD routine sometimes returned bogus factors, which
were then trapped in a sanity check. That error was fixed in 20.3 . I dont know
about
errors by faulty low[mp].txt files, but that sure was not the reason in my case,
as F14 has
no known factors and the error occurred only once out of hundreds of curves.
> > >[Thu May 25 15:00:24 2000]
> > >ECM found a factor in curve #263, stage #2
> > >Sigma=1155032128984, B1=250000, B2=25000000.
> > >ERROR: Factor doesn't divide N!
>
> Suggest you ensure that you're running v20.4 & that your lowm.txt
> file is up-to-date & not corrupt. Sometimes you can get extra
>
> BTW from George's ECM status pages it seems that either factors have
> become _very_ hard to find recently. Anyone have any views on this?
I think the 2^p+1 list still offers a few good candidates. There have been much
fewer curves
run on them than on their "2 less" counterparts.
I'm focusing on F14 right now, while Richard Crandall's challenge still is on,
but predictably,
nothing much happened so far.
Ciao,
Alex.
_________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
