Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Here I have the following conjecture -
Let $$S_1(n)= \frac{(n-1)! +1}{n}$$ then there exist infinite prime numbers $p$ for which $S_1(p)$ is prime. And I don't know how to prove it.

Let $C_1(n)=[S_1(n)]$ where $[n]$ denotes the integral part. Then I have another conjecture that $[C_1(1) ,C_1(2),C_1(3) ,C_1(4),C_1(5) ,C_1(6)...]$ contains infinite many primes . Are there any ways to prove it?

share|cite|improve this question
Is there any motivation for this conjecture, i.e. why do you think it is true? I couldn't find any other solution else then 5,7,11,29 below 500. – benh Dec 28 '13 at 18:11
@benh ... and $S_1(29)=10513391193507374500051862069$ is already quite a beast when talking about everyday primes. – Hagen von Eitzen Dec 28 '13 at 18:15
These kind of questions (where the numbers are very sparse) are essentially always intractable, so good luck. – Igor Rivin Dec 28 '13 at 18:23
Next after 29 seems to be 773, but my isprime is still running (though ispseudoprime was quite confident) – Hagen von Eitzen Dec 28 '13 at 18:25
@RaymondManzoni lets say we have $S_2(n)= \frac{(n-2)! +1}{n}$ then can similar arguments be made and infinite primes proved in this case too? – Shivam Patel Dec 29 '13 at 14:07
up vote 11 down vote accepted

The first primes solution are at OEIS as A050299 : $$1, 5, 7, 11, 29, 773, 1321, 2621$$ with the comment "There are no other terms up to $6550$ and the corresponding next prime has more than $22150$ digits. No more terms below $30941$."

Mike Oakes added a comment (and its correction since an important factor was missing) :
"Asymptotically, the probability that $p$ is prime is $\dfrac 1{\log(p)}$, and that $N =\dfrac {1+(p-1)!}p$ is prime is $\displaystyle \frac 1{\log(N)} \approx \frac 1{\log((p-2)!)} \approx\frac 1{p\log(p)}$ by Stirling.

CORRECTION: since no prime $< p$ divides $N$, the probability that $N$ is prime must be multiplied by the factor $\;e^{\gamma}\log(p)$ (using Mertens' theorem), where $\gamma$ is Euler's constant, with $e^{\gamma}=1.7810724\cdots$

So the expected number of such primes between $p_1$ and $p_2$ is of order : $$ e^{\gamma}\;\int_{p_1}^{p_2} \frac{\log(x)\;dx}{x\log(x)^2} = e^{\gamma}(\log(\log(p_2))-\log(\log(p_1)))$$ which is unbounded as $\,p_2\to\infty$.

Putting in the numbers, this predicts:

  1. that for $\;2 \le p \le 30941$ there should be $4.8$ such primes, which is in reasonable agreement with the actual number found ($7$); and
  2. that asymptotically there is on average one such prime between $p_1$ and $p_2$ whenever $p_2 = (p_1)^{1.753....}$

Mike Oakes (with error pointed out by David Farmer, Dean Hickerson, Paul Jobling, Carl Pomerance, Bjorn Poonen, Noam D. Elkies)"

Let's consider with more care the new problem for $n$ integer : $$C_1(n)= \left\lfloor\frac{(n-2)! +1}{n}\right\rfloor$$ I obtained the first solutions : $$7,29,61,139,383(?)$$

I'll distinguish two cases :

  • $n$ not prime which is true with probability $\approx 1-\dfrac 1{\log(n)}$ multiplied by the $\dfrac 1{n\,\log(n)}$ term from Mike Oakes' analysis. The partial contribution for $n\in(n_1,n_2)$ will then be of order : $$\tag{1}\int_{n_1}^{n_2} \left(1-\frac 1{\log(x)}\right)\frac{dx}{x\log(x)} = \left.\log(\log(x))+\frac 1{\log(x)}\right|_{n_1}^{n_2}\approx\left.\log(\log(x))\right|_{n_1}^{n_2}$$
  • for $n=p$ prime and using Wilson's theorem I got $\;(p-1)(p-2)!\equiv (p-1)\pmod{p}\;$ and (since $(p,p-1)=1$) $\;(p-2)!\equiv 1\pmod{p}\;$ so that $\;\displaystyle \left\lfloor \frac{(p-2)!+1}p\right\rfloor=\frac{(p-2)!-1}p$ but $(p-2)!-1$ can't have divisors smaller than $p-1$ and thus we still can't have divisors $<p$ when $p$ is prime $>3$ (another result $\;(p-1)\left\lfloor \frac{(p-2)!+1}p\right\rfloor=\frac{(p-1)!+1}p-1\,$).
    The repartition for $n=p$ prime will thus be asymptotically the same as in your first problem : $$\tag{2}e^{\gamma}\;\int_{n_1}^{n_2} \frac 1{\log(x)}\frac{\log(x)\;dx}{x\log(x)} = \left.e^{\gamma}\;\log(\log(x))\right|_{n_1}^{n_2}$$ Combining $(1)$ and $(2)$ we get an average number of solutions in $(n_1,n_2)$ of order : $$\left(1+e^{\gamma}\right)(\log(\log(n_2))-\log(\log(n_1)))$$ and may hope for one solution going from $n_1$ to $n_2=(n_1)^c$ with $c=e^{1/(1+e^{\gamma})}\approx 1.433$ (instead of $e^{1/e^{\gamma}}\approx 1.753$ previously).
share|cite|improve this answer
Manzori suppose we have $S_2(n)= \frac{(n-2)! +1}{n}$ then what is the order of expected number of primes between $p_1$ and $p_2$? – Shivam Patel Dec 30 '13 at 9:03
@ShivamPatel: I updated my answer and gave a solution for primes and for integers. – Raymond Manzoni Dec 30 '13 at 15:06

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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