Re: Frobenius test and GMP library

From: Cristiano (cristiano.pi_at_NSquipo.it)
Date: 12/24/04


Date: Fri, 24 Dec 2004 11:42:42 GMT

Sisyphus wrote:
> [rob@localhost frobenius]$ ./frobenius.exe 10
> 212697255460959188811228955285461874921
> Performing 10 iterations of the frobenius probable prime test on
> 212697255460959188811228955285461874921....
>
> Probable prime with parameters
> 68874042992856834480547917601092539187,
> 53020279949960851054630937398442978906
[...]

I get the same!

I don't know whether you have the time, but it would be very helpful to try
this:
you should change the lines
    mpz_urandomm(a, rand_state, N);
    mpz_urandomm(b, rand_state, N);
with: a=a+1 and b=b+1 which should be
    mpz_add_ui(a, a, 1);
    mpz_add_ui(b, b, 1);

Using that code, the primes 107, 32507 and 74471 are declared composites
(which seems very strange to me). Do you get the same?

Thank you
Cristiano