3
$\begingroup$

Is it true, that $\forall x \in \mathbb N, \pi(x+200)-\pi(x) \leq 50 $ ?

$$\pi(x)=\text{card}(\{n \in [0,x] \cap \mathbb N, n\text{ is prime}\})$$

$\endgroup$
0

1 Answer 1

12
$\begingroup$

Yes.

Up to $207$ there are $46$ primes. Hence, the inequality is true for $x \le 7$.

Let $$\pi_{210}(x) = \textrm{card}(\{n \in [0,x] \cap \mathbb{N}, \, \gcd(n,210)=1\}).$$ For $x>7$, $\pi(x+200)-\pi(x) \le \pi_{210}(x+200)-\pi_{210}(x)$. Since $\pi_{210}$ is $210$-periodic, it is enough to verify that $\pi_{210}(x+200)-\pi_{210}(x) \le 50$ for $x \le 210$, which can be done by hand or by computer. Here is SageMath code:

L = [int(gcd(i,210)==1) for i in range(420)]

max(sum([numpy.array(L[i:][:210]) for i in range(200)]))

and the last line outputs $47$. So the bound can be improved to this number.

The number $210=2 \times 3 \times 5 \times 7$ was chosen because $\prod_{p \mid 210}(1-1/p) < 50/200$.

$\endgroup$
4
  • $\begingroup$ We have $\pi(x+210)-\pi(x) \leq 48$ $\endgroup$
    – Dattier
    Sep 2, 2020 at 11:30
  • 2
    $\begingroup$ Indeed. By considering coprimality with $2310 = 2 \times 3 \times 5 \times 7 \times 11$, the same method produces $\pi(x+210)-\pi(x) \le 46$ for $x>11$; small $x$ can be checked by hand, and in fact there are cases of equality. The second Hardy-Littlewood Conjecture predicts $\pi(x+210)-\pi(x) \le \pi(210)=46$ also (for $x \ge 2$), but the general form of the conjecture is believed to be false. $\endgroup$ Sep 2, 2020 at 11:36
  • 1
    $\begingroup$ for x=1, we have $\pi(211)-\pi(1)=47$ $\endgroup$
    – Dattier
    Sep 2, 2020 at 11:41
  • 1
    $\begingroup$ Yes, sorry. That's the only counterexample though. $\endgroup$ Sep 2, 2020 at 11:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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