So I'm trying to do this annoying proof and without going into further details I think after quite a while of thinking I found it. Now I get stuck with an annoying sum (of sums..) where I don't quite know if there exists a closed form and if so how to find it.

So as I already said I try to find a closed form to the following series $\sum\limits_{i=2}^{n} \frac{H_i}{i+1}$ where $H_n$ is the harmonic series ($\sum\limits_{i=1}^{n} \frac{1}{i}$).

So yeah any hint for a closed form of the above series is more than welcome! Thanks in advance for any help.

  • $\begingroup$ The authors in the following wiki try to give a proof for the closed form of $H_n$ which I would mistrust due to the slightly dodgy way of representing an integral representation of an alternating sum. Check it out though, may get you somewhere en.wikipedia.org/wiki/Harmonic_number#Calculation $\endgroup$ – Kevin Oct 12 '16 at 13:10
  • $\begingroup$ Yeah well besides the dodgy integral this doesn't really help me in this case. Since it's not only the harmonic series but there's also a sum and a division.. Thanks though! $\endgroup$ – Desperate Oct 12 '16 at 13:19
  • $\begingroup$ I know, but I thought that with a little algebra you could import the division by $i+1$ and manipulate. $\endgroup$ – Kevin Oct 12 '16 at 13:22
  • $\begingroup$ Well I kind of have a closed form if it weren't for the division .. $\sum\limits_{i=1}^n H_i = (n+1) \cdot (H_{n+1} - 1)$ .. what I want to say is that it doesn't matter if I have the harmonic series still in there as long as I can get the other sum away. $\endgroup$ – Desperate Oct 12 '16 at 13:49

A preliminary manipulation:

$$\sum_{i=2}^{n}\frac{H_i}{i+1}=-\frac{1}{2}+\sum_{i=1}^{n}\frac{H_i}{i+1}=-\frac{1}{2}+\sum_{i=1}^{n}\frac{H_{i+1}}{i+1}-\sum_{i=1}^{n}\frac{1}{(i+1)^2}$$ gives: $$\sum_{i=2}^{n}\frac{H_i}{i+1}=\sum_{k=1}^{n+1}\frac{H_k}{k}-\left(H_{n+1}^{(2)}+\frac{1}{2}\right)\tag{1} $$ and now: $$ \sum_{k=1}^{n+1}\frac{H_k}{k}=\sum_{k=1}^{n+1}\frac{1}{k}\sum_{j=1}^{k}\frac{1}{j}=\sum_{1\leq j\leq k\leq n+1}\frac{1}{j\cdot k}=\frac{1}{2}\left[\left(\sum_{i=1}^{n+1}\frac{1}{i}\right)^2+\sum_{i=1}^{n+1}\frac{1}{i^2}\right] \tag{2}$$ leads to: $$ \sum_{i=2}^{n}\frac{H_i}{i+1}=\color{red}{\frac{H_{n+1}^2-H_{n+1}^{(2)}-1}{2}}.\tag{3}$$ Here $H_m^{(2)}$ stands for $\sum_{k=1}^{m}\frac{1}{k^2}$, as usual. We exploited $H_{m}=H_{m+1}-\frac{1}{m+1}$.

Non-believers may just take $(3)$ as a claim and prove it through induction on $n$.
Anyway, the crucial part $(2)$ is just an instance of the following identity:

$$ \sum_{k=1}^{n}\sum_{j=1}^{k}f(j)\cdot f(k)=\sum_{1\leq j\leq k\leq n}f(j)\cdot f(k) = \frac{1}{2}\left[\left(\sum_{k=1}^{n}f(k)\right)^2+\sum_{k=1}^{n}f(k)^2\right].$$

  • $\begingroup$ +1 IIRC, there is also an expression involving Stirling numbers of the second kind. Probably not what is expected here though. $\endgroup$ – Jean-Claude Arbaut Oct 12 '16 at 17:10
  • $\begingroup$ Stirling numbers of the first kind in fact. :) $\endgroup$ – Jean-Claude Arbaut Oct 12 '16 at 20:00
  • $\begingroup$ @Jean-ClaudeArbaut: clearly so, just written in a explicit way, depending on generalized harmonic numbers. $\endgroup$ – Jack D'Aurizio Oct 12 '16 at 20:01
  • $\begingroup$ Well first off thanks a bunch! I agree that Stirling numbers are way beyond what I need thanks for that aswell though! $\endgroup$ – Desperate Oct 12 '16 at 20:12
  • 1
    $\begingroup$ aaaaaah I see!!! sorry for the inconvenience I was thinking it would be way less obvious $\endgroup$ – Desperate Oct 12 '16 at 20:29

It can be proven also using summation by parts $$S=\sum_{i=2}^{n}\frac{H_{i}}{i+1}=\sum_{i=2}^{n}\frac{1}{i+1}\cdot H_{i}=H_{n}\left(H_{n+1}-\frac{3}{2}\right)-\sum_{i=2}^{n-1}\frac{\left(H_{i+1}-\frac{3}{2}\right)}{i+1} $$ hence $$\sum_{i=2}^{n}\frac{H_{i}}{i+1}=H_{n}\left(H_{n+1}-\frac{3}{2}\right)-\sum_{i=2}^{n-1}\frac{H_{i+1}}{i+1}+\frac{3}{2}\sum_{i=2}^{n-1}\frac{1}{i+1} $$ $$=H_{n}H_{n+1}-\sum_{i=2}^{n}\frac{H_{i}}{i+1}+\frac{H_{n}}{n+1}-\sum_{i=2}^{n-1}\frac{1}{\left(i+1\right)^{2}}-\frac{9}{4} $$ so $$\sum_{i=2}^{n}\frac{H_{i}}{i+1}=\color{red}{\frac{1}{2}\left(H_{n}H_{n+1}+\frac{H_{n}}{n+1}-H_{n}^{\left(2\right)}-1\right)}.$$ Note that this is the same result of the other answers, since $$H_{n}=H_{n+1}-\frac{1}{n+1}.$$

  • 1
    $\begingroup$ (+1) I was going to add this approach to my previous answer, but there is no need now. Well done, as always. $\endgroup$ – Jack D'Aurizio Oct 13 '16 at 13:45

$\newcommand{\braces}[1]{\left\lbrace\,{#1}\,\right\rbrace} \newcommand{\bracks}[1]{\left\lbrack\,{#1}\,\right\rbrack} \newcommand{\dd}{\mathrm{d}} \newcommand{\ds}[1]{\displaystyle{#1}} \newcommand{\expo}[1]{\,\mathrm{e}^{#1}\,} \newcommand{\ic}{\mathrm{i}} \newcommand{\mc}[1]{\mathcal{#1}} \newcommand{\mrm}[1]{\mathrm{#1}} \newcommand{\pars}[1]{\left(\,{#1}\,\right)} \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\root}[2][]{\,\sqrt[#1]{\,{#2}\,}\,} \newcommand{\totald}[3][]{\frac{\mathrm{d}^{#1} #2}{\mathrm{d} #3^{#1}}} \newcommand{\verts}[1]{\left\vert\,{#1}\,\right\vert}$ \begin{align} \sum_{i = 2}^{n}{H_{i} \over i + 1} & = \sum_{i = 2}^{n}\braces{{1 \over 2}\bracks{\pars{H_{i} + {1 \over i + 1}}^{2} - H_{i}^{2} - \pars{1 \over i + 1}^{2}}} \\[5mm] & = {1 \over 2}\sum_{i = 2}^{n}H_{i + 1}^{2} - {1 \over 2}\sum_{i = 2}^{n}H_{i}^{2} - {1 \over 2}\sum_{i = 2}^{n}{1 \over \pars{i + 1}^{2}} \\[5mm] & = {1 \over 2}\sum_{i = 3}^{n + 1}H_{i}^{2} - \bracks{{1 \over 2}\,H_{2}^{2} + {1 \over 2}\sum_{i = 3}^{n + 1}H_{i}^{2} - {1 \over 2}\,H_{n + 1}^{2}} - {1 \over 2}\sum_{i = 3}^{n + 1}{1 \over i^{2}}\quad \pars{~\mbox{note that}\ {1 \over 2}\,H_{2}^{2} = {9 \over 8}~} \\[5mm] & = {1 \over 2}\,H_{n + 1}^{2} - {9 \over 8} - \pars{-{1 \over 2} - {1 \over 8} + {1 \over 2}\sum_{i = 1}^{n + 1}{1 \over i^{2}}} = \bbox[8px,border:1px groove navy]{{1 \over 2}\,H_{n + 1}^{2} - {1 \over 2} - {1 \over 2}\,H_{n + 1}^{\pars{2}}} \end{align}


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.