8

I want to show that the following holds:

$$\sum_{k=-m}^{n} \binom{m+k}{r} \binom{n-k}{s} =\binom{m+n+1}{r+s+1}.$$

I have an idea of what is going on here. On the RHS we are selecting $r+s+1$ elements from the set $[m+n+1]$. Another way of doing this is to partition the set $[m+n+1]$ into two sets in several ways by shifting $k$ elements from one set to the other. For example, we can partition $[m+n+1]$ into a set with $m+k$ elements and a set with $n+1-k$ elements. From the set with $m+k$ elements we can select $r$ elements and from the set with $n+1-k$ elements we can select $s+1$. However, this is not matching the LHS. I believe the problem is that we don't know which $k$ elements are being shifted around.

I thought rewriting the sum as follows would help:

$$\sum_{k=0}^{m+n} \binom{k}{r} \binom{n+m-k}{s}.$$

I just don't understand why the "$+1$" on the top and bottom of the RHS disappear. I have spent several hours on this problem, and I have run out of ideas. Please help, I truly appreciate it.

Martin Sleziak
  • 56,060
Cicero
  • 81

2 Answers2

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{equation} \sum_{k = -m}^{n}{m + k \choose r}{n - k \choose s} = {m + n + 1 \choose r + s + 1}:\ {\large ?} \label{1}\tag{1} \end{equation} LHS variables must satisfy $\left\{\begin{array}{rcl} \ds{m + k \geq r \geq 0} & \ds{\implies} & \left\{\begin{array}{l} \ds{r \geq 0 } \\[2mm] \ds{k \geq r - m \geq -m} \end{array}\right. \\[1mm] \ds{n - k \geq s \geq 0} & \ds{\implies} & \left\{\begin{array}{l} \ds{s \geq 0 } \\[2mm] \ds{k \leq n - s \leq n} \end{array}\right. \end{array}\right.$

Then, \begin{align} &\color{#f00}{\sum_{k = -m}^{n}{m + k \choose r}{n - k \choose s}} = \sum_{k = r - m}^{n - s}{m + k \choose r}{n - k \choose s} = \sum_{k = r - m}^{\infty}{m + k \choose r}{n - k \choose s} \\[5mm] = &\ \sum_{k = 0}^{\infty}{m + \pars{k + r - m} \choose r} {n - \pars{k + r - m} \choose s} = \sum_{k = 0}^{\infty}{k + r \choose k} {m + n - r - k \choose m + n - r - k - s} \\[5mm] = &\ \sum_{k = 0}^{\infty}{-r - 1 \choose k}\pars{-1}^{k} {-s - 1 \choose m + n - r - s - k}\pars{-1}^{m + n - r - s - k} \\[5mm] = &\ \pars{-1}^{m + n + r + s}\sum_{k = 0}^{\infty}{-r - 1 \choose k} \oint_{\verts{z} = 1^{-}}{\pars{1 + z}^{-s - 1} \over z^{m + n - r - s - k + 1}} \,{\dd z \over 2\pi\ic} \\[5mm] = &\ \pars{-1}^{m + n + r + s} \oint_{\verts{z} = 1^{-}}{\pars{1 + z}^{-s - 1} \over z^{m + n - r - s + 1}} \sum_{k = 0}^{\infty}{-r - 1 \choose k}z^{k}\,{\dd z \over 2\pi\ic} \\[5mm] = &\ \pars{-1}^{m + n + r + s} \oint_{\verts{z} = 1^{-}}{\pars{1 + z}^{-s - 1} \over z^{m + n - r - s + 1}} \pars{1 + z}^{-r - 1}\,{\dd z \over 2\pi\ic} \\[5mm] = &\ \pars{-1}^{m + n + r + s} \oint_{\verts{z} = 1^{-}}{\pars{1 + z}^{-s - 1 - r - 1} \over z^{m + n - r - s + 1}}\,{\dd z \over 2\pi\ic} = \pars{-1}^{m + n + r + s}{-s - r - 2 \choose m + n - r - s} \\[5mm] = &\ \pars{-1}^{m + n + r + s}{s + r + 2 + m + n - r - s - 1 \choose m + n - r - s} \pars{-1}^{m + n - r - s} \\[5mm] = &\ {m + n + 1 \choose m + n - r - s} = {m + n + 1 \choose \bracks{m + n + 1} - \bracks{m + n - r - s}} = \color{#f00}{m + n + 1 \choose r + s + 1} \end{align}
Felix Marin
  • 94,079
0

Suppose we seek to evaluate

$$\sum_{k=-m}^n {m+k\choose r} {n-k\choose s} = \sum_{k=0}^{m+n} {k\choose r} {m+n-k\choose s}.$$

Here we may assume that $r$ and $s$ are non-negative. We introduce

$${m+n-k\choose s} = \frac{1}{2\pi i} \int_{|z|=\epsilon} \frac{1}{z^{m+n-k-s+1}} \frac{1}{(1-z)^{s+1}} \; dz.$$

This integral controls the range, being zero when $k\gt m+n$ and we may extend the range of $k$ to infinity. We get for the sum

$$\frac{1}{2\pi i} \int_{|z|=\epsilon} \frac{1}{z^{m+n-s+1}} \frac{1}{(1-z)^{s+1}} \sum_{k\ge 0} {k\choose r} z^k \; dz \\ = \frac{1}{2\pi i} \int_{|z|=\epsilon} \frac{1}{z^{m+n-s+1}} \frac{1}{(1-z)^{s+1}} \sum_{k\ge r} {k\choose r} z^k \; dz \\ = \frac{1}{2\pi i} \int_{|z|=\epsilon} \frac{1}{z^{m+n-s+1}} \frac{z^r}{(1-z)^{s+1}} \sum_{k\ge 0} {k+r\choose r} z^k \; dz \\ = \frac{1}{2\pi i} \int_{|z|=\epsilon} \frac{1}{z^{m+n-s+1}} \frac{z^r}{(1-z)^{s+1}} \frac{1}{(1-z)^{r+1}} \; dz \\ = \frac{1}{2\pi i} \int_{|z|=\epsilon} \frac{1}{z^{m+n-r-s+1}} \frac{1}{(1-z)^{r+s+2}} \; dz.$$

We get for the answer

$${m+n-r-s+r+s+1\choose r+s+1} = {m+n+1\choose r+s+1}.$$

Marko Riedel
  • 64,728