Let $a_n>0$ and $S_n=\sum_{k=1}^{n}a_n$. If $\lim_{n \rightarrow \infty}S_n = +\infty$, then $\sum_{n=1}^{\infty}\frac{a_n}{S_n}=+\infty$. I think this is an important example because it tells us that there exists no series which diverge slowest. So I want to verify this fact in different aspect.

I know a method which uses Cauchy's Theorem. For any $n \in \bf N$ , choose a sufficiently large $p \in \bf{N}$. we have $$\sum_{k=n+1}^{n+p}\frac{a_k}{S_k}\geq \frac{S_{n+p}-S_{n}}{S_{n+p}}\geq \frac{1}{2}$$

Is there any other approach to it? Thanks very much.

  • 4
    $\begingroup$ That's probably the simplist way. See this also (I assume this is the argument you're using). $\endgroup$ Jul 15, 2013 at 12:27
  • 5
    $\begingroup$ If you just want to prove that there is no slowest diverging series, you can see that you can actually choose the increasing sequence $S_n$, and define $a_n=S_n-S_{n-1}$. Then, for any sequence $S_n$ you can get the slower diverging sequence $\sqrt{S_n}$. $\endgroup$
    – Rodrigo
    Apr 30, 2014 at 4:57
  • $\begingroup$ Is that answer your question? $\endgroup$
    – user146010
    Sep 11, 2014 at 3:25

1 Answer 1


First : $a_n= S_n - S_{n-1}$

Let $U_n$ be : $U_n = \frac{S_n- S_{n-1}}{S_n} = 1 - \frac{S_{n-1}}{S_n}$

Then : $\ln(S_n) -\ln(S_{n-1}) = -\ln(1-U_n)$

Suppose the series of general term $(U_n)$ converges:

$\implies (U_n) \rightarrow 0$

$\implies$ $\ln(S_n) - \ln(S_{n-1}) = U_n + o(U_n) $

It means that $( \ln(S_n) - \ln(S_{n-1}) )$ is the term of a convergent series, thus implying that $(S_n)$ is a converging sequence, which is absurd.

Hence the series of general term $(U_n)$ diverges, always.

Now one interesting question is: what c does make $U_n(c) = \frac{a_n}{S_n^c}$ converge? You know that $c \geq 1$ , but is it the inf?

  • 1
    $\begingroup$ Downvotes and no comments, classic math.exchange. Deontology still not a thing here, shameful really . The guy who pointed to the proof of op's very result got 4 upvotes for his comment, I give another solution and a generalization of his example and I get 2 downvotes. You guys actually achieve to turn maths into mediocrity, I have to say I'm baffled but not positively $\endgroup$
    – mvggz
    Aug 11, 2016 at 9:14
  • 2
    $\begingroup$ Insulting everyone for an opinion of one person is neither productive, nor helpful, nor it stimulates upvotes. If you suspect that foul play is involved, you can always flag your answer for moderator attention. $\endgroup$ Aug 13, 2016 at 18:14
  • $\begingroup$ Note that upvotes on comments are very different from upvotes on answers. Also, I think the reason that comment was upvoted was to indicate agreement with the claim "that's probably the simplest way". $\endgroup$ Aug 13, 2016 at 18:40
  • 1
    $\begingroup$ And what's the point of downvoting an answer if there is not one single information about why. Just what's the point.. And you don't even seem to find it the most troubling thing altogether, that comforts me about the seriousness of this community. If this site is not about pedagogy and emulation, then it's useless just like downvoting an answer with not a single explanation. I did not believe it was possible to disgust me of mathematical material, but this community actually achieved that. That's not how I 'exchange' about math with someone, I leave that to all of you $\endgroup$
    – mvggz
    Aug 15, 2016 at 15:16

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .