Frobenius Probable Prime Test

Jason Moxham
Tue, 13 May 2003 14:39:41 +0100

On Tuesday 13 May 2003 04:17, Torbjorn Granlund wrote:
> Jason Moxham <> writes:
>   ...
>   I kind of in the process of mpn'ifing it , but it will take some time
>   because all the underlying functions need to mpn'ified
> I think that would be a big mistake.  Only basic building blocks
> should be mpn.  A prime test doesn't count as a basic building
> block.  Worse, subtle bugs might make an mpn level prime test
> useless.

It's intended for testing the possibility of using it for fast small prime 
test , hence it needs to mpn'ified and not intended for inclusion in GMP . 
Its really not that difficult because it splits into a number of independant 
functions(which are useful elsewhere) , which can be tested much easier than 
probable prime tests.

It should be noted that the rqf test is not as useful as it may appear , it 
has no use (ie slower) for generating large crypto primes for example.