I was so fascinated the first time I saw Eulers Number $e$ as the sum of the infinite series over $\frac{1}{n!}$ that I began playing on variants of it, learning a lot on the way.

However when I started playing around with primes things became tricky, I found: $$\sum_{n=1}^{\infty} \frac{p_n}{n!}=\frac{2}{1!}+\frac{3}{2!}+\frac{5}{3!}+\dots = 4.73863870268..$$

It intrigued me when I noticed that it appeared to converge as I know about the divergence of the sums of reciprocal primes. I couldn't find this in any literature and couldn't find a closed form.

I have ran simulations and computed this for fun to hundreds of decimal places however I have ploted a graph for $\frac{p_{n+1}}{(n+1)!} - \frac{p_n}{n!}$, $n$ ranging from $1$ to $9$

Plot enter image description here

I have checked with a program if every next difference is smaller than the previous and this holds up to $n=1000$ (with the idea that if this holds on forever it will eventually become $0$ thus the series will converge to a constant value)

Question #1: Does it converge?

Question #2: Is this something known (probably), and if yes, where could I read up on it?

  • 6
    $\begingroup$ To prove convergence it is enough to exploit the ridiculous bound $1<p_n<e^n$. $\endgroup$ – Jack D'Aurizio Jan 28 '17 at 20:17
  • 1
    $\begingroup$ By the way, although you mention that $$\sum_{n=1}^\infty \frac{1}{p_n} \to \infty,$$ the following infinite series is actually convergent: $$\sum_{n=1}^\infty\frac{(-1)^n}{p_n}\approx -0.2696063519.$$ This was proven in $1971$, and I thought that you might be interested. It is also unknown whether or not $$\sum_{n=1}^\infty \frac{n(-1)^n}{p_n}$$ is convergent or divergent. $\endgroup$ – Mr Pie Feb 16 '18 at 11:29
  • $\begingroup$ @user477343 what was proven in 1971 ? Convergence of $\sum_{n=1}^\infty\frac{(-1)^n}{p_n}$ is trivial by the alternating series theorem. $\endgroup$ – Gabriel Romon Apr 22 '18 at 11:23

Yes, it converges. As a hint, the prime number theorem gives us an asymptotic estimate $$p_n \sim n \log n. $$

See if you now can apply any of the convergence tests you know.

  • $\begingroup$ With what you have said, it seems that I could apply the ratio test to prove convergence in this case, however a closed form or approximation on the actual number would be impossible or very hard, am I right? $\endgroup$ – Ilhan Jan 28 '17 at 20:21
  • 3
    $\begingroup$ @Ilhan Instead of the prime number theorem (complicated) you can show that $\sum_n \frac{1}{p_n}$ diverges (using that $\log \zeta(s)= -\sum_n \log(1-p_n^{-s})$), which means that $\lim_{n \to \infty} \frac{p_n}{n^2} = 0$, so that $\sum_n \frac{p_n}{n!}$ converges $\endgroup$ – reuns Jan 28 '17 at 20:26
  • 2
    $\begingroup$ Or this fact math.stackexchange.com/questions/144539/… and ratio test. $\endgroup$ – rtybase Jan 28 '17 at 20:27
  • $\begingroup$ the two suggestions are indeed easier and I can now prove the convergence thus the question is answered, out of curiosity, would it be difficult to approximate the actual number? @rtybase $\endgroup$ – Ilhan Jan 28 '17 at 20:31
  • 2
    $\begingroup$ Given $p_n < (1+\varepsilon)n\log{n}$, for $\forall n > N$, you can play around $$\sum_{n}\frac{p_n}{n!}<\sum_{n=1}^{N}\frac{p_n}{n!}+(1+\varepsilon)\sum_{n>N}\frac{\log{n}}{(n-1)!}$$ and reduce this further given that $$\lim_{n\rightarrow \infty} \frac{\log{n}}{n-1}=0$$ $\endgroup$ – rtybase Jan 28 '17 at 20:39

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.