I've changed mpz_likely_prime_p to do 10 M-R tests with random bases. I also added a test to the test suite which uses mpz_probab_prime_p with 100 tests to determine if the number is very likely prime.
The test code does not permit mpz_likely_prime_p to claim any primes are composite. But it will allow a single composite to be declared prime before aborting, i.e. it will have to happen twice before there is an abort. I think the chances of this happening at random are so small that even if every GMP and MPIR user in the world ran this test code as many times as there are GMP and MPIR users in the world then maybe it might happen once. The chances of it being a machine failure or compiler bug, etc, are much greater. If there is a protest, I'll remove this test. But it's test code, not code used in a mathematical theorem, so I personally am very comfortable doing this. It is designed to give the user an indication of whether it is likely their library is ok to trust or not. Bill. -- You received this message because you are subscribed to the Google Groups "mpir-devel" group. To post to this group, send email to mpir-de...@googlegroups.com. To unsubscribe from this group, send email to mpir-devel+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/mpir-devel?hl=en.
