Suppose a property $\phi$ holds for a strictly positive proportion of the prime numbers, i.e. such that $$\liminf_{x\to\infty}{\pi_\phi(x)\over \pi(x)}\gt 0,$$ where $\pi_\phi(x)$ counts the number of primes less than or equal to $x$ having the property $\phi$, and $\pi$ is the usual prime-counting function.

According to [here] and [here], cases of this include the following:

  1. $\phi_1(n)$="changing any single decimal digit of $n$ (not including leading zeros) produces a composite number"
  2. $\phi_2(n)$="changing any single decimal digit of $n$ (including any leading zero) produces a composite number"

E.g., $n=294001$ has property $\phi_1$, but not property $\phi_2$ (because $10294001$ is prime). In fact:

Infinitely many primes (indeed a positive proportion) have property $\phi_2$, yet there is no known example.

Q1: What are some other number-theoretic properties proven to hold for a strictly positive proportion of the primes, yet there is no specific prime for which the property has been proven to hold?

Q2: Are there similar cases with respect to $\mathbb{N}?$ I.e., some property holds for a strictly positive proportion of $\mathbb{N},$ yet there is no specific $n\in\mathbb{N}$ for which the property has been proven to hold? In this case, I suppose the proportion of $\mathbb{N}$ satisfying property $\phi$ would be defined as $$\liminf_{x\to\infty}{C_\phi(x)\over x},\ \text{ where }\ C_\phi(x)=\#\{n\in\mathbb{N}: n\le x,\ \phi(n) \}.$$

  • $\begingroup$ "n is prime" has density 0 so both conditions you list have density 0 as well. Am I missing something? $\endgroup$ May 21 at 21:25
  • 1
    $\begingroup$ @OlivierBégassat Thank you. I was wanting to generalize to $\mathbb{N}$ but made a misstep -- hopefully this makes sense now. $\endgroup$
    – r.e.s.
    May 21 at 21:44
  • $\begingroup$ A tempting answer would be the property "$n > 10^{10^{10^{10}}}$" - but I suppose that's not what you're looking for. $\endgroup$ May 21 at 21:57
  • $\begingroup$ @MiloBrandt I was going to say "positive and less than 1" to avoid cases like that, but it seems that would throw out other interesting cases too. $\endgroup$
    – r.e.s.
    May 21 at 22:14
  • $\begingroup$ The distinction between density of primes and density of $\mathbb N$ is kind of irrelevant, because you can take $\phi$ about primes and construct $\rho(n)=\phi_(p_n),$ as I did in my answer before you corrected the question to be about density in primes. $\endgroup$ May 21 at 22:38

The current state of Artin’s conjecture can be put this way:

$\phi(p):$ $p$ is a multiplicative generator modulo $q$ for infinitely many primes $q.$

where $p$ is prime.

As of now, it is only known that $\phi(p)$ is true for all but at most two primes $p,$ but the proof gives no hint at what the two values are, and we don’t know $\phi(p)$ is true for any single prime $p.$

Still, the density of $\phi$ amongst primes is $1,$ and we could only do one better - if we have a theorem with one potential but unknown counterexample.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.