4
$\begingroup$

Take $k$ consecutive composite integers from a prime gap. What is known about the largest prime divisor of their product?

It seems to me that except for the triplet $(8,9,10)$ and the pair $(8,9)$ , this largest prime divisor is always larger than $2k$, but I could not find an elementary (my level) proof.

As $k$ grows it seems that the sharpness of $2k$ as lower bound is lost, therefore I expect that there should exist a more suitable lower bound, from which an "easy" argument for the minoration by $2k$ would follow.

The question is essentially the same as this one from MSE but it did not get any answer. I could only find messy $k$-specific arguments for $k$ up to $4$ which I hope are correct.

$\endgroup$
5
  • $\begingroup$ Do you have any reference to a "nonelementary" proof? $\endgroup$
    – Wojowu
    Oct 31, 2015 at 13:09
  • $\begingroup$ @Wojowu. No, but I am not professional mathematician, so my knowledge of (and access to) math literature is very limited. If you know of relevant papers, I 'd be interested. $\endgroup$
    – René Gy
    Oct 31, 2015 at 13:21
  • $\begingroup$ I was asking because when you said "...I could not find an elementary (my level) proof.", I thought you might have found a hard proof and you were just looking for a simpler one. $\endgroup$
    – Wojowu
    Oct 31, 2015 at 13:22
  • $\begingroup$ You might be interested in a paper of Filip Najman at arxiv.org/abs/1108.3710 . The largest prime divisor of the product n+1 to n +f(k) is looked at and shown to be larger than k when n is larger than k. f(k) is pretty small and conjectured to be O(log(k)^2). So you don't need as many as k consecutive integers. Gerhard "Sees This As Smooth Intervals" Paseman, 2015.11.02 $\endgroup$ Nov 2, 2015 at 18:18
  • $\begingroup$ @Gerhard That is very interesting, thanks. Probably above my current level though. $\endgroup$
    – René Gy
    Nov 3, 2015 at 17:47

1 Answer 1

8
$\begingroup$

There have been several investigations into the largest prime factor of a product of consecutive integers; the Sylvester--Schur theorem is an early example. Here is a survey by Shorey and Tijdeman: https://www.math.leidenuniv.nl/~tijdeman/shoretij.pdf

In

Laishram, Shanta(6-TIFR-SM); Shorey, T. N.(6-TIFR-SM) The greatest prime divisor of a product of consecutive integers. Acta Arith. 120 (2005), no. 3, 299–306

it is shown that $$ P(n(n+1) \cdots (n+k-1)) > 2k $$ as long as $n > \max\{k+13, \frac{279}{262}k\}$.

Note that for large $k$, the lower bound condition on $n$ is implied by your hypothesis that all of $n, n+1, \dots, n+k-1$ are composite; so this answers your question for large $k$. (And for bounded $k$, it reduces checking your conjecture to a finite computation.)

There is more good news: the MathSciNet review records that "the proof is elementary".

$\endgroup$
1
  • $\begingroup$ If the numbers are composite, can we give a constant lower bound on $n+k-1$, as long as $n>2k$? $\endgroup$
    – Daniel W.
    Jun 6, 2020 at 11:45

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.