$$\sum_{m=0}^∞ \frac{1}{\sum_{n=0}^m p_n} = \frac{1}{2} + \frac{1}{5} + \frac{1}{10} + \frac{1}{17} ... $$

Where $p_n$ is the $n$th prime number, does $\sum_{m=0}^∞ \frac{1}{\sum_{n=0}^m p_n}$ converge?

  • 10
    $\begingroup$ The sum of the first $n$ primes is asymptotic to $\frac{1}{2}n^2 \ln n$, so yes. $\endgroup$ – Slade Feb 26 '16 at 4:01
  • 6
    $\begingroup$ The value of the sum is OEIS A122989, namely $1.02347632392012\ldots$. $\endgroup$ – Jeppe Stig Nielsen Feb 26 '16 at 12:43

All you need is $p_n \ge n$, so $\sum_{n=1}^m p_n \ge \sum_{n=1}^m n = m(m+1)/2$, and $\sum_{m=1}^\infty \dfrac{2}{m(m+1)}$ converges.

  • 4
    $\begingroup$ That is by far the simplest answer. +1 $\endgroup$ – Patrick Da Silva Feb 26 '16 at 16:42

The answer is yes.

For large $m$, the denominators are known to be approximately $\frac{1}{2}m^2\ln m$; in particular, they are larger than $m^2$. Hence for $m$ sufficiently large, the terms are bounded above by $\frac{1}{m^2}$.


The n-th prime is quite obviously ≥ 2n - 1. Therefore the sum of the first n primes is ≥ $n^2$. The sum in question converges and is at most the sum over $1 / n^2$ which converges to $π^2 / 6$. Actually, it must be less than $π^2 / 6 - 1/2$ because the first element of the sum is 1/2, not 1.

That sum is about 1.1493, not worlds apart from the actual result.

I'm just curious why an answer saying "all you need is $p_n ≥ n$" gets voted up while an answer saying "$p_n ≥ 2n - 1$" and giving a reasonable upper bound for the limit doesn't.

  • 5
    $\begingroup$ Whining about votes comes off as very unbecoming... $\endgroup$ – Zach466920 Feb 28 '16 at 1:35

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.