Discussion related to cado-nfs

[Pierrick Gaudry <pierrick.gaudry@loria.fr>]

On Sat, Apr 08, 2017 at 10:13:45PM -0400, Earl Fechter wrote:
> No, I don't. But I'm planning on using a bunch of jailbroken iphones as
> slaves to my desktop computer. I know I'll realistically have a very
> slim chance of splitting the number with my home computer, but I'm
> doing it like the lottery: you know you probably wont win, but that
> doesn't keep you from trying.

An iphone (or a bunch of them) is not appropriate for this kind of
computation. Furthermore, it might be quite hard to get CADO-NFS work on
those (we didn't try and won't).

As for the lottery aspect, the number field sieve algorithm that is used
in CADO-NFS does not have this kind of "I might get lucky" property (ok,
if you want to go into the details of the mathematics, this statement
could be discussed, but the general idea is still valid).
There is another algorithm, called ECM, that has this "lottery" property
you want, in the sense that it is possible to estimate the probability of
finding a factor for given parameters and running time.

If your goal is just to dream that you might factor a large number, then
you'd better dream with ECM than with CADO-NFS. GMP-ECM is the most
popular implementation (not for iphone, of course).

> About that parameter: what IS a parameter, and how do you find one?
> H.S freshman, not the world's best mathematician, lol.

parameterS (plural). CADO-NFS has tons of parameters that must be tuned
for different input sizes. These are stored in the parameters/factor/
subdirectory. If don't know the algorithm, you have no chance to come up
with decent parameters.

I recommend that you abandon this project to use CADO-NFS for huge
numbers with iphones.

Regards.
Pierrick