Here $n$ is some natural number. For example, among the primes $< 1000$ I found four such combinations:

\begin{array}( 11 & 13 & 17 & 19 \\ 101 & 103 & 107 & 109 \\ 191 & 193 & 197 & 199 \\ 821 & 823 & 827 & 829 \end{array}

Using Mathematica I was able to move further, so the sequence $n_k$ starts with:


A question already exists on this topic, however there is not a lot of information there.

I would like to know if the sequence $n_k$ was studied before, and what can we tell about the distribution of $n_k$ among the natural numbers?

Distances $n_{k+1}-n_k$ seem to grow on average, but 'close' quadruples still exist even for large $n_k$, for example:


The plot of all the distances for $n_k<10^6$ is provided below (there are $898$ of them):

enter image description here

As the author of the linked question stated, every $n_k$ has the form $3m+1$, so the distances are all divisible by $3$.

So, the main thing I ask is some reference on the topic, or additional information about this sequence.

Found OEIS A007811 with some information


It is conjectured that the number of such $n_k$ up to $X$ is of the size (ignoring leading constants) $$ \frac{X}{\log^4 X},$$ and similarly that the number of $k$-tuples up to $X$ in fixed, admissible configurations is of the size $$ \frac{X}{\log^k X}.$$

Your tuple is $(10n+1, 10n+3, 10n+7, 10n+9)$. If you were to consider the $8$-tuple $(10n+1, 10n+3, 10n+7, 10n+9, 10n+91, 10n+93, 10n+97, 10n+99)$, (which I think is admissible but I didn't actually check), then it is conjectured that the number of such $8$-tuples up to $X$ is of the size $$ \frac{X}{\log^8 X}.$$ Notice that this is actually two of your $4$-tuples separated by $90$. So conjecturally we believe there should be infinitely "smallish" gaps between $4$-tuples of your shape.

More distribution-style statements can be made along these lines. You'll get very far by looking up the prime $k$-tuple conjecture and studying its progress and results.

  • $\begingroup$ What does 'fixed, admissible configurations' mean? What about something like $1000n+111,1000n+333,1000n+777,1000n+999$? $\endgroup$ – Yuriy S Mar 30 '16 at 19:01
  • $\begingroup$ Right! The word "admissible" means that it doesn't trivially not have infinitely many primes. For instance, $n, n+1$ is not "admissible." $\endgroup$ – davidlowryduda Mar 30 '16 at 19:15
  • $\begingroup$ How come the known result $\frac{X}{\log^k(X)}$ is such a poor estimate for $k>1$? For example, let $R_k = \frac{\text{estimate}}{\text{actual}}$ and $X=10^6$. Then, rounded off, we have, $$R_1 = \frac{72400}{78500} = 0.92\\ R_4 = \frac{27}{898} = 0.03$$ Only the first is a respectable prediction. Are there formulas for $k>1$ with better estimates? $\endgroup$ – Tito Piezas III Mar 30 '16 at 20:12
  • $\begingroup$ @TitoPiezasIII, you are confusing numbers here. $10^6$ is the boundary for $n$, while $X$ would be around $10^7$ $\endgroup$ – Yuriy S Mar 30 '16 at 20:20
  • 1
    $\begingroup$ @TitoPiezasIII, this function doesn't give approximate values for a number of quadruples, it describes how this number grows on average with $X$. And it seems to be doing a good enough job. $\endgroup$ – Yuriy S Mar 30 '16 at 21:13

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.