Don Leclair writes:
There is the risk that more than one factor (or all factors!) might
be extracted in between GCD's. This is easy to work around: After
each unsuccessful GCD, save the current iteration number. If the
next GCD(N,accumulator) is equal to 'N', back track to the
iteration you saved and start again, executing the GCD more
frequently.
I believe that Factor98, at least, does things as Don describes,
except that it does not "back up". Certainly, it does sometimes print
"this factor is composite" (or similar) and I end up factoring those
composite factors, when they're sent to me, using the ecmfactor
program of the mers package. In fact, the ecmfactor program was
written by me based on ecm3 precisely because I needed to factor some
composite factors found by Factor98.
As always, feel free to send me new code, bug reports, fixes, etc.,
for the package. I haven't seen many lately, but I also doubt it's
perfect.:) As for new features, PrimeNet for the LL testers is in the
early, pre-compilation, stage of implementation, so you don't need to
mention that one to me, but go ahead and send me other ideas.
Will
http://www.garlic.com/~wedgingt/mers.tar.gz
mers.tgz (same contents)
beta.tgz (beta version)
README.html (nearest thing to a FAQ)
