12
$\begingroup$

Find the value of $\displaystyle \sum_{m=1}^\infty \sum^{\infty}_{n=1}\frac{m\cdot n}{(m+n)!}$.

My try: $$\sum^{\infty}_{m=1}\bigg(\frac{m}{(m+1)! }+\frac{2m}{(m+2)!}+\cdots\cdots +\frac{m+\infty}{(m+\infty)!}\bigg)$$

I did not understand how to start it.

I could use some help. Thanks.

$\endgroup$
  • 1
    $\begingroup$ Fix $m$ and then figure out what happens to $\sum\limits_{n=1}^\infty \frac{m\cdot n}{(m+n)!}$, it should be a function $m$. Then you can sum this function with the outer sum. $\endgroup$ – Tony S.F. Jan 31 '18 at 13:43
13
$\begingroup$

Note that by letting $k=m+n$, we have that (terms can be arranged because they are non-negative), $$\begin{align} \sum^{\infty}_{m=1}\sum^{\infty}_{n=1}\frac{m\cdot n}{(m+n)!} &=\sum^{\infty}_{k=2}\frac{1}{k!}\sum_{n=1}^{k-1}n(k-n)= \frac{1}{6}\sum^{\infty}_{k=2}\frac{k^3-k}{k!}\\&=\frac{1}{6}\sum^{\infty}_{k=2}\frac{k(k-1)(k-2)+3k(k-1)}{k!}\\ &= \frac{1}{6}\sum^{\infty}_{k=3}\frac{1}{(k-3)!}+\frac{1}{2}\sum^{\infty}_{k=2}\frac{1}{(k-2)!}\\ &=\frac{e}{6}+\frac{e}{2}=\frac{2e}{3}.\end{align}$$

$\endgroup$
  • $\begingroup$ Dear Robert, I did not understand how you did the change of variable for the indexes, and how you substitue $k$ instead of $m$, can you explain it please? $\endgroup$ – Nizar Jan 31 '18 at 14:07
  • $\begingroup$ @Nizar Consider $\mathbb{N}^+\times\mathbb{N}^+$ as the union of the diagonals $\{(1,1)\}\cup\{(2,1),(1,2)\}\cup\{(3,1),(2,2),(1,3)\}\cup\dots$. $\endgroup$ – Robert Z Jan 31 '18 at 14:12
  • $\begingroup$ @Nizar It's the change of indices from $(m;n)$ to $(k;n)$ where $k=m+n$. $\endgroup$ – Crostul Jan 31 '18 at 14:50
  • 2
    $\begingroup$ It is worth pointing out that it is valid to rearrange the terms of the series because its terms are non-negative (by Tonelli's theorem). $\endgroup$ – Maxim Jan 31 '18 at 14:50
  • $\begingroup$ @Maxim Of course... thanks for pointing out! $\endgroup$ – Robert Z Jan 31 '18 at 14:57
8
$\begingroup$

The slickest approaches have already been outlined by Marco and Robert, so I will show an overkill just for fun. For any $n\in\mathbb{N}$ we have $\frac{1}{n!}=[z^n]e^z = \frac{1}{2\pi i}\oint \frac{e^z}{z^{n+1}}\,dz $, hence

$$ \sum_{m,n\geq 1}\frac{mn}{(m+n)!}=\frac{1}{2\pi i}\oint\sum_{m,n\geq 1}\frac{m}{z^m}\cdot\frac{n}{z^n}\cdot\frac{e^z}{z}\,dz=\frac{1}{2\pi i}\oint\frac{z e^{z}}{(1-z)^4}\,dz $$ and $$ \sum_{m,n\geq 1}\frac{mn}{(m+n)!}=\operatorname*{Res}_{z=1}\frac{z e^z}{(1-z)^4} =\color{red}{\frac{2e}{3}}.$$

$\endgroup$
6
$\begingroup$

We have $$\sum_{m\geq1}\sum_{n\geq1}\frac{mn}{\left(n+m\right)!}=\sum_{m\geq1}\frac{1}{\left(m-1\right)!}\sum_{n\geq1}\frac{n}{\left(n-1\right)!}\int_{0}^{1}t^{m}\left(1-t\right)^{n-1}dt$$ $$=e\int_{0}^{1}t\left(2-t\right)dt=\color{red}{\frac{2}{3}e}$$ where the exchange of the series and the integral is justified by the dominated convergence theorem.

$\endgroup$
3
$\begingroup$

Using generating functions: let's start off with $$F(x, y) = \sum_{m=0}^\infty \sum_{n=0}^\infty \frac{x^m y^n} {(m+n)!}.$$ If we group the terms with $m+n = k$, we get $\frac{1}{k!} (x^k + x^{k-1} y + \cdots + y^k) = \frac{1}{k!} \cdot \frac{x^{k+1} - y^{k+1}}{x-y}$. Therefore, $$F(x,y) = \sum_{k=0}^\infty \frac{1}{k!} \cdot \frac{x^{k+1} - y^{k+1}}{x-y} = \frac{x e^x - y e^y}{x-y}.$$ Now, $$F_{xy}(x,y) = \sum_{m=1}^\infty \sum_{n=1}^\infty \frac{mn}{(m+n)!} x^{m-1} y^{n-1}.$$ Thus, the original sum is equal to $F_{xy}(1, 1)$. From here, it would be a straightforward but tedious calculation to find a closed-form formula for $F_{xy}$. That will give a fraction with $(x-y)^3$ in the denominator; however, it is not hard to see that $F_{xy}$ should be continuous at $(1,1)$, so you can calculate the value at $(1,1)$ as a limit of this quotient as $(x,y) \to (1,1)$. (For example, first substituting $x=1$ since $\{ (1,y) \mid y \ne 1 \}$ has cluster point at $(1,1)$, and then either using l'Hopital's rule three times or expanding a Taylor series about $y=1$ should work.) This will give the final answer $\frac{2}{3} e$.

For instance, in a Maxima session, I get (with a little hand editing of the transcript):

(%i1) diff(diff((x*exp(x)-y*exp(y))/(x-y),x),y);

(%i2) limit(limit(%o1,x,1),y,1);
          2 %e
(%o2)     ----
           3
$\endgroup$
1
$\begingroup$

$\newcommand{\bbx}[1]{\,\bbox[15px,border:1px groove navy]{\displaystyle{#1}}\,} \newcommand{\braces}[1]{\left\lbrace\,{#1}\,\right\rbrace} \newcommand{\bracks}[1]{\left\lbrack\,{#1}\,\right\rbrack} \newcommand{\dd}{\mathrm{d}} \newcommand{\ds}[1]{\displaystyle{#1}} \newcommand{\expo}[1]{\,\mathrm{e}^{#1}\,} \newcommand{\ic}{\mathrm{i}} \newcommand{\mc}[1]{\mathcal{#1}} \newcommand{\mrm}[1]{\mathrm{#1}} \newcommand{\pars}[1]{\left(\,{#1}\,\right)} \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\root}[2][]{\,\sqrt[#1]{\,{#2}\,}\,} \newcommand{\totald}[3][]{\frac{\mathrm{d}^{#1} #2}{\mathrm{d} #3^{#1}}} \newcommand{\verts}[1]{\left\vert\,{#1}\,\right\vert}$ \begin{align} \sum_{m = 1}^{\infty}\sum_{n = 1}^{\infty}{mn \over \pars{m + n}!} & = \sum_{m = 1}^{\infty}\sum_{n = 1}^{\infty}{mn \over \pars{m - 1}!\, n!}\,{\Gamma\pars{m}\Gamma\pars{n + 1} \over \Gamma\pars{m + n + 1}!} \\[5mm] & = \sum_{m = 1}^{\infty}\sum_{n = 1}^{\infty} {m \over \pars{m - 1}!\pars{n - 1}!} \int_{0}^{1}t^{m - 1}\pars{1 - t}^{n}\,\dd t \\[5mm] &= \int_{0}^{1}\sum_{m = 1}^{\infty}{m\, t^{m - 1} \over \pars{m - 1}!} \sum_{n = 1}^{\infty}{\pars{1 - t}^{n} \over \pars{n - 1}!}\,\dd t \\[5mm] & = \int_{0}^{1}\underbrace{\sum_{m = 0}^{\infty} {\pars{m + 1}t^{m} \over m!}}_{\ds{=\ \expo{t}\pars{1 + t}}}\ \underbrace{\sum_{n = 0}^{\infty}{\pars{1 - t}^{n + 1} \over n!}\,\dd t} _{\ds{=\ \expo{1 -t}\pars{1 - t}}} \\[5mm] & = \expo{}\int_{0}^{1}\pars{1 - t^{2}}\,\dd t = \bbx{{2 \over 3}\,\expo{}} \approx 1.8122 \end{align}

$\endgroup$

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.