Vandermonde's convolution

Introduction

Vandermonde's convolution is an identity that combines binomial coefficients and is mainly used in deriving formulas in combinatorics.

Vandermonde's convolution formula

∑i=0k(ni)(mk−i)=(n+mk)\sum_{i=0}^k\binom{n}{i}\binom{m}{k-i}=\binom{n+m}{k}

Proof

Consider a proof using the binomial theorem:

∑k=0n+m(n+mk)xk=(x+1)n+m=(x+1)n(x+1)m=∑r=0n(nr)xr∑s=0m(ms)xs=∑k=0n+m∑r=0k(nr)(mk−r)xk\begin{aligned} \sum_{k=0}^{n+m}\binom{n+m}{k}x^k&=(x+1)^{n+m}\\ &=(x+1)^n(x+1)^m\\ &=\sum_{r=0}^n\binom{n}{r}x^r\sum_{s=0}^m\binom{m}{s}x^s\\ &=\sum_{k=0}^{n+m}\sum_{r=0}^k\binom{n}{r}\binom{m}{k-r}x^k\\ \end{aligned}

Therefore:

(n+mk)=∑r=0k(nr)(mk−r)\binom{n+m}{k}=\sum_{r=0}^k\binom{n}{r}\binom{m}{k-r}

For a proof based on its combinatorial meaning:

Choosing n+mn+m elements from a set of size kk is equivalent to splitting the set of size n+mn+m into two sets of sizes nn and mm, and then choosing nn elements from the first set ii and mm elements from the second set k−ik-i. Since we enumerate ii, it is enough to consider one such split, because different splits are equivalent.

Corollaries

Corollary 1 and proof

∑i=−rs(nr+i)(ms−i)=(n+mr+s)\sum_{i=-r}^{s}\binom{n}{r+i}\binom{m}{s-i}=\binom{n+m}{r+s}

The proof is similar to that of the original formula.

Corollary 2 and proof

∑i=1n(ni)(ni−1)=(2nn−1)\sum_{i=1}^n\binom{n}{i}\binom{n}{i-1}=\binom{2n}{n-1}

By basic combinatorics, we derive:

∑i=1n(ni)(ni−1)=∑i=0n−1(ni+1)(ni)=∑i=0n−1(nn−1−i)(ni)=(2nn−1)\sum_{i=1}^n\binom{n}{i}\binom{n}{i-1}=\sum_{i=0}^{n-1}\binom{n}{i+1}\binom{n}{i}=\sum_{i=0}^{n-1}\binom{n}{n-1-i}\binom{n}{i}=\binom{2n}{n-1}

Corollary 3 and proof

∑i=0n(ni)2=(2nn)\sum_{i=0}^n\binom{n}{i}^2=\binom{2n}{n}

By basic combinatorics, we derive:

∑i=0n(ni)2=∑i=0n(ni)(nn−i)=(2nn)\sum_{i=0}^n\binom{n}{i}^2=\sum_{i=0}^n\binom{n}{i}\binom{n}{n-i}=\binom{2n}{n}

Corollary 4 and proof

∑i=0m(ni)(mi)=(n+mm)\sum_{i=0}^m\binom{n}{i}\binom{m}{i}=\binom{n+m}{m}

By basic combinatorics, we derive:

∑i=0m(ni)(mi)=∑i=0m(ni)(mm−i)=(n+mm)\sum_{i=0}^m\binom{n}{i}\binom{m}{i}=\sum_{i=0}^m\binom{n}{i}\binom{m}{m-i}=\binom{n+m}{m}

(n+mm)\binom{n+m}{m} is the familiar number of paths in a grid, so we can consider a proof based on its combinatorial meaning.

In a grid, going from (0,0)(0,0) to (n,m)(n,m) takes n+mn+m steps. Let (0,0)(0,0) be the upper-left corner; taking nn steps downward and mm steps to the right gives (n+mm)\binom{n+m}{m} possibilities.

From another viewpoint, split the n+mn+m steps into two parts, first taking nn steps and then mm steps. If the first nn steps contain ii rightward steps, then the last mm steps contain m−im-i rightward steps, proving the identity.

Exercises

References and notes

  1. Vandermonde's Convolution Formula

This content comes from OI Wiki: Vandermonde's convolution. See the original for authors and revision history. AIOJ has adapted the formatting, in-site links, and resource paths.

Content is shared under CC BY-SA 4.0 and SATA license; additional terms are governed by the original notice. Thanks to the OI Wiki authors and contributors; per the SATA terms, please visit the OI Wiki repository and star it to show your support.