Can someone give me an easy (two or three line) proof for the fact that $\dbinom{n}{k}$ is divisible by $n$ for $k\not=0, n$ and $\gcd(n,k)=1.$
Here $\dbinom{n}{k}$ denotes usual binomial coefficients.

- 53,687

- 18,220
- 5
- 47
- 87
-
1A combinatorial proof would be cool! – MathematicsStudent1122 Sep 05 '16 at 00:43
-
1There is a combinatorial proof that $k \dbinom{n}{k}=n \dbinom{n-1}{k-1} $, but you will not avoid an arithmetic argument since we are using a gcd. – WNG Sep 05 '16 at 00:53
-
Some related older posts: The divisibility of a particular case of $\binom{x}{y}$ (when $(x,y) = 1$) and ${\gcd(n,m)\over n}{n\choose m}$ is an integer – Martin Sleziak Sep 12 '19 at 10:35
2 Answers
$k>0$, thus $\dbinom{n}{k}=\frac n k\dbinom{n-1}{k-1} $ Thus if $\gcd(n,k)=1$, n divides $\frac n k\dbinom{n-1}{k-1} $
Bonus : Combinatorial proof that $\dbinom{n}{k}=\frac n k\dbinom{n-1}{k-1} $ :
If you want to choose a team of k people amongst n with one team leader, you can :
choose a leader first (n possibilities) then choose the remaining k-1 members amongst the n-1 remaining possiblities
choose the k people first, then choose a leader amongst the k people team.
Thus $k\dbinom{n}{k}=n \dbinom{n-1}{k-1}$

- 594
-
Suppose $\gcd(n,k)=d.$ Is it true that$\dbinom{n}{k}$ is divisible by $\dfrac{n}{d}$ ? – Bumblebee Sep 05 '16 at 01:31
-
2
The slightly more general result that $\frac{n}{\gcd (n,k)}$ divides $\binom{n}{k}$ (with $n\geq 1$ and $1\leq k \leq n$) can be proven quite easily using a couple results from group theory, as well as the identity $$k\binom{n}{k}=n\binom{n-1}{k-1}$$ Consider the (additive) group $\mathbb{Z}/n\mathbb{Z}$ of integers modulo $n$. The element $\bar{1}_n$ has order $n$, and for $k\in \lbrace 1,...,n \rbrace$, we have that $\binom{n}{k}(k{\bar 1}_{n})=\binom{n}{k}k{\bar 1}_{n}=\binom{n-1}{k-1}n{\bar 1}_{n}\underset{n \textrm{ divides } \binom{n-1}{k-1}n}{=} {\bar 0}_{n}$. Thus $\binom{n}{k}$ is divisible by the order of $k{\bar 1}_{n}$, which is equal to $\frac{n}{\gcd (n,k)}$. QED

- 186
-
-
You don't have to define $n/k$ mod $n$ for this proof to be valid. At one point I multiplied an element of $\mathbb Z /n \mathbb Z$ not by $n/k$ but by $\frac{n}{k}\binom{n-1}{k-1}$, which is an integer, so everything is well-defined.
That being said, you're bringing up a good point since that step isn't needed for the proof. I'll edit my comment to clear up the confusion, thanks for the reply.
– KCJV Sep 23 '20 at 08:58