2

Yesterday I saw this question: A question about divisibility of sum of two consecutive primes (you should read the OP to understand the full problem), it just asks to prove that for all $k\in \mathbb Z^+$, there exist infinitely many consecutive primes such that : $$k\mid p_{n+1}+p_n.$$ the guy who asked this took care of the cases where $k=1,2,3,4,6$. The general case where $k$ is any positive integer is beyond me, but I attempted to prove the case where $k=12$ and I wonder it the proof is correct:

Assume that the twin prime conjecture is true, which says that there exist infinitely many consecutive primes such that $$p_{n+1}-p_n=2$$

Since any prime $\ge 5$ is on the form $6k\pm 1$ and every pair of twin primes is on the form $(6k-1,6k+1)$, Hence $$p_{n+1}+p_n=6k+1+6k-1=12k$$ $$12\mid p_{n+1}+p_n$$

for infinitely many consecutive primes?

Note that the twin prime conjecture also implies the case where $k=1,2,3,4,6$, because $1,2,3,4,6\mid 12$

PNT
  • 4,295
  • As you mentioned , this proof requires the truth of the twin prime conjecture, but it is a valid proof. – Peter May 26 '21 at 12:02
  • if we assume the truth of polignac's conjecture, can we do better than $k=12$? @Peter – PNT May 26 '21 at 13:20
  • I think we can extend this result , but I have no good approach yet. Intuitively , I would guess, that we can find infinite many pairs for every $k$. – Peter May 26 '21 at 13:22
  • that's exactly what I thought in the beginning but it turns out that you can't do any better than $12$, at least that's what I've found. @Peter – PNT May 26 '21 at 13:27
  • Your same argument works for $8$ too. – arbashn Jul 13 '21 at 15:33

1 Answers1

1

Assuming Schinzel's hypothesis , for every positive integer $k$ , there are infinite many positive integers $\ n\ $ , such that $\ kn-1\ $ and $\ kn+1\ $ are both prime. Those primes are obviously consecutive because they have difference $\ 2\ $ (in fact they are twin primes). The sum is $\ 2kn\ $ which is divisible by $\ k\ $.

Of course, Schinzel's hypothesis is much stronger than the twin prime conjecture, but at least this is some evidence that we can find infinite many pairs for every $\ k\ $.

Peter
  • 86,576
  • There is a gap in your proof, note that $k$ is a variable, in other words, it's not fixed, so it changes with respect to the primes $kn\pm 1$. The OP asked given a fixed $k\in \mathbb Z^+$ are there infinitely many consecutive primes s.t $k\mid p_{n+1}+p_n$ @Peter – PNT May 26 '21 at 13:42
  • 1
    You have not read the answer carefully. For every $k$ , we can find infinite many $n$ giving a suitable twin prime pair , if Schinzel's hypotehsis is true. – Peter May 26 '21 at 13:43
  • 1
    yeah... sorry you're right. @Peter – PNT May 26 '21 at 13:45
  • 1
    You've said that Schinzel's hypothesis is weaker than the twin prime conjecture, but it looks like a generalization of it? @Peter – PNT May 26 '21 at 13:49
  • Oops , this time you are right :) I will fix it. I surely mixed it with "weaker evidence" – Peter May 26 '21 at 13:51
  • 1
    Schinzel's hypothesis is far too strong. Dickson's conjecture suffices. – arbashn Jul 13 '21 at 15:31
  • @arbashn Thank you for pointing that out. I did not remember this weaker conjecture. – Peter Jul 13 '21 at 15:40