On Friday 04 September 2009 08:33:01 Cactus wrote:
> On Sep 4, 4:58 am, Bill Hart <[email protected]> wrote:
> > I don't personally like the name mpz_practical_prime_p(). The reason is
> > that it doesn't mean anything particular to anyone but us.
>
> I also doubt that it will help to have two differenet functions that
> appear to do the same job to anyone who is unaware of the subtleties
> involved.
>

If the user has an algorithm which requires one function over the other , but 
is not aware of the difference between them , then should they really should 
not be writing anything that sophisticated. Either of the new functions would 
be better than the old one in any case.

> I would personally prefer to call this a test:
>
>   mpz_prime_test(mpz_t n, enum alg, uint trials) + rand_state?
>
> to avoid potentially confusing names.
>
> I am also uneasy about returning true for something that might be
> false even if we do represent this uncertainty in the name.
>
> How about this:
>
>   mpz_is_composite(mpz_t n) that returns 1 if composite or 0 if the
> test fails (compositeness undetermined).

This may well be how I will implement it , with mpz_practical_prime being a 
macro on top of it. I just did it this way as that is how it is usually 
presented.Some books do point out that it would be better to call probable 
prime tests , composite tests instead. If we do the composite test instead 
then we can return more than just true or false , we could return a "why" , 
ie a trial divisor or a witness.


>
> for Jason's 'practical prime' proposal?
>
> We can then explain that this removes most composites rapidly and
> allows the more costly but more precise prime tests for numers that
> remain.
>
>     Brian
>
> 


--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"mpir-devel" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at 
http://groups.google.com/group/mpir-devel?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to