[Foxtrot] APR-CL primality test

[Matej Pfajfar]

OK how about this -
Al can you (quickly?) test Miller-Rabin? I suspect that isn't difficult
cause it's an easier algorithm.
And Shahrum and Matt can test APR-CL. However if you can't identify the
bugs, I don't want you sacrificing other bugs because of it.

I'll update javadocs so matt and shahrum can have a look.

Mat

On Mon, 19 Feb 2001, Alan Lawrence wrote:

> 
> Compiles, but doesn't work :-(
> 
> For "smallish" numbers (anything up to ~~100 digits), it hits upon a
> definite answer (prime/composite) in the early stages of the
> algorithm; I'm assuming this to be correct, though there is that suspect
> number that Maple says is prime and my code says is composite....
> 
> But for larger numbers (150ish digits; 512 bits), it reports that every
> number is composite.....which is obviously wrong! Most likely cause is
> bugs in the Jacobi Sum arithmetic code
> (foxtrot/src/foxtrot/primetest/JacobiSum.java; javadoc it and then look in
> the supplement to that paper). Question is, is it worth our time testing
> it and getting it working, or should we just stick with Miller-Rabin?
> Seems a bit of a waste of my hours if we leave it, but then, everyone else
> has limited hours too....any thoughts?
> 
> Cheers,
> Alan
> 
> 
> 
> _______________________________________________
> Foxtrot mailing list
> Foxtrot@srcf.ucam.org
> http://www.srcf.ucam.org/mailman/listinfo/foxtrot
> 

Matej Pfajfar
St John's College, University of Cambridge, UK

PGP Public Key @ www.telesat.si/~badbytes
Most people are good people, the rest of us are going to 
run the world. -- badbytes