The other day while discussing math, and primes specifically, the following question came to mind, and I figured I'd ask it here to see what people's opinions on it might be.

Main Question:Suppose that tomorrow someone proves that some function always generates (concrete) primes for any input. How should this affect lists such as the Largest Known Primes?

Let me give a little more detail to demonstrate why I feel this question is not entirely trivial or fanciful.

Firstly, the requirement that the function be able to concretely generate primes is meant to avoid 'stupid' examples such as Nextprime(n) which, given a 'largest known prime' P, yields a larger prime Nextprime(P). Note however that the definition of Nextprime does not actually explicitly state what this prime is, any implementation of it (in Maple or Mathematica for example) simply loops through the integers bigger than the input, testing each for primality in some fashion.

On the other hand, one candidate for such a function might be the Catalan sequence defined by:

$C(0) = 2$, $C(n+1) = 2^{C(n)}-1$

Although $C(5) = 2^{170141183460469231731687303715884105727}-1$ is far too large to test by current methods (with rougly $10^{30}$ times as many digits as the current largest known prime), and although the current consensus is that $C(5)$ is likely composite, it does not seem entirely out of the realm of possibility that someone might eventually find some very clever way of showing $C(5)$ is prime, or even that $C(n)$ is always prime, or perhaps some other concretely defined sequence.

The point is this: once one knows that every element of a sequence is prime, does this entirely negate things like the list of largest known primes? Or does the fact that $C(n)$ for $n\geq 5$ has too many digits to ever calculate all of them (instead only being able to calculate the first few or last few digits) mean that even if they were somehow proven prime it would not technically be 'known'?

Note also that in the realm of finite simple groups the analogous question is already tough to decide since there are infinite families of such groups known, but concrete descriptions (such as generators and relations or character tables) are not always available or even computable within reasonable time constraints. Likewise one could pose analogous questions in other branches (largest volume manifolds with certain constraints, etc.)

Anyhow, it seems like a reasonable question for serious mathematicians to consider, so I just want to hear what other's opinions are on the subject (and if anyone can think of a better title, feel free to suggest).

9more comments