2

Let $n$ and $k$ be integers with $n \geq k \geq 0$. Find a combinatorial proof for

$$\binom{n+1}{k} = \binom{n}{k} + \binom{n-1}{k-1} + \cdots + \binom{n-k}{0} .$$

My approach: I was thinking to use the binomial formula as in $$2^n = \sum{\binom{n}{k}1^k1^{n-k}} .$$ I also tried to use Pascal's Identity $\binom{n}{r}=\binom{n-1}{r-1}+\binom{n-1}{r}$.

  • Using the binomial theorem (although I don't understand how you plan to do that) would be an "algebraic" proof, not a "combinatorial" proof. They're asking for a proof by counting things. – saulspatz Jan 29 '18 at 05:07
  • The problem is to find a combinatorial proof. I double checked – merry-go-round-one Jan 29 '18 at 05:10
  • MathJax works in the titles too, @JohnBaek. – Shaun Jan 29 '18 at 05:10
  • In case it is not clear why the post I linked is the same identity, remember that $\binom{n}{k}=\binom{n}{n-k}$. If you wish not to look at the post I linked and just want a hint, recognize that any subset of ${1,2,3,\dots,n+1}$ falls into exactly one of the following cases: doesn't have a $1$, has a $1$ but not a $2$, has a $1$ and $2$ but not a $3$, ..., has the first $i$ numbers but not the $i+1$'st number, etc... – JMoravitz Jan 29 '18 at 05:11
  • If you just need the result, it is $\binom{2n-r}{n-r+1}$, assuming that the top is (1,1). – Thern Mar 18 '18 at 12:49
  • @Thern Are you sure of your result? for n=6 r=2 we should obtain 21 – user Mar 18 '18 at 13:10
  • 1
    @gimusi You are right, I made a mess here. But note that I use the top as (1,1), so what is n=6, r=2 is the same as n=7, r=3 for me. The formula is actually simpler: $\binom{n}{n-r+1}$. If you use top as (0,0), it is $\binom{n+1}{n-r+1}$. For n=6, r=2, this is $\binom{7}{5}=\frac{7\cdot 6}{2\cdot 1}=21$. – Thern Mar 19 '18 at 08:57
  • @Thern You are welcome! Thae latter matches with my derivation. – user Mar 19 '18 at 09:07
  • @ShivamKumar Please remember that you can choose an answer among the given if the OP is solved, more details here https://meta.stackexchange.com/questions/5234/how-does-accepting-an-answer-work – user Mar 19 '18 at 22:57
  • Since someone wants to delete this topic, I think we need to open this. – Michael Rozenberg Jan 12 '20 at 11:55

4 Answers4

3

I think you can try to interpret this identity using the following real-life problem:

Imagine you have a plate of $n+1$ treats: $k$ of them are chocolate pieces, and the rest are Brussels sprouts. In how many ways can you eat all of them, one by one? Assume the chocolate pieces are indistinguishable from each other, and so are Brussels sprouts.

The answer is, of course, $\binom{n+1}{k}$. However, let's count differently, depending on what you eat first:

  • You bravely go straight on to a Brussels sprout, before eating any chocolate. There are $\binom{n}{k}$ ways to do that, as there will be $n$ treats left and $k$ of them are chocolate pieces;
  • You first eat one piece of chocolate, and then go on to eating a Brussels sprout: there are $\binom{n-1}{k-1}$ ways to do that, as there will be $n-1$ treats left, $k-1$ of them chocolate;
  • You first eat two pieces of chocolate, then go on to a Brussels sprout: similarly, it can be done in $\binom{n-2}{k-2}$ ways;
  • You first eat three pieces of chocolate...

etc. until:

  • You eat $k-1$ chocolate pieces first, then one Brussels sprout. Then, only one of the remaining $n-(k-1)$ treats is a chocolate, and the number of ways to do that is $\binom{n-(k-1)}{1}$;
  • You eat all the chocolate pieces first, and then you eat all of the Brussels sprouts: it can be done in only one way, which can also be written as $1=\binom{n-k}{0}$.

Generally, if you eat first $m$ pieces of chocolate ($0\le m \le k$) before getting on with your first Brussels sprout, there are $\binom{n-m}{k-m}$ ways to do it: after eating the initial chocolate pieces and the Brussels sprout, there will be $n-m$ treats left on the plate, $k-m$ of them being chocolate, so you can eat them in $\binom{n-m}{k-m}$ ways.

Altogether all those numbers must add to our original result $\binom{n+1}{k}$, i.e.

$$\binom{n+1}{k}=\binom{n}{k}+\binom{n-1}{k-1}+\binom{n-2}{k-2}+\cdots+\binom{n-(k-1)}{1}+\binom{n-k}{0}=\sum_{m=0}^k\binom{n-m}{k-m}$$

  • Beat me to it by 30 seconds, +1. I'm a big fan of combinatoric proofs, and this is probably one of the nicest out there. – DreamConspiracy Jan 29 '18 at 05:24
  • Does this make sense to everyone??? picking i consecutive elements... are you talking about the index? – merry-go-round-one Jan 29 '18 at 16:56
  • @JohnBaek I am picking elements from the set ${1,2,3,\ldots,n+1}$. –  Jan 29 '18 at 19:47
  • You are dividing the counts to 'if i+1 is counted' and ' if i+1 is NOT counted'. Then shouldnt be (n,k)*2???? – merry-go-round-one Jan 29 '18 at 19:52
  • @JohnBaek Please read the whole answer once again. It is about: 'is $1$ not counted', 'is $1$ counted but not $2$', 'are $1$ and $2$ counted but not $3$' etc. up to 'are $1,2,3,\ldots,k$ counted but not $k+1$'. Also, is it a problem that I used the letter $i$ as the index for the summation? –  Jan 29 '18 at 19:54
  • Can we write it in better way?? FOr example, picking a president and committees from the members. – merry-go-round-one Jan 30 '18 at 03:33
  • @JohnBaek I don't know... At this time of the day all I can think of is: you have a plate of $n+1$ items, $k$ of them being chocolate, and the rest are Brussel sprouts; in how many ways can you eat them (one by one)? This is because you may decide to devour a Brussel sprout straight away, or you may decide to first take $1, 2, 3,\ldots,k$ pieces of chocolate before getting on with the inevitable. ;) –  Jan 30 '18 at 06:22
  • Yeah we can use chocolate and brussel. Can you update ur answer?? – merry-go-round-one Jan 30 '18 at 06:32
  • @JohnBaek Will do later today when I have more time. –  Jan 30 '18 at 06:52
  • Thanks...!!! Im looking forward to it – merry-go-round-one Jan 30 '18 at 12:45
2

It follows from Pascal's identity that\begin{align}\binom{n+1}k&=\binom nk+\binom n{k-1}\\&=\binom nk+\binom{n-1}{k-1}+\binom{n-1}{k-2}\\&=\binom nk+\binom{n-1}{k-1}+\binom{n-2}{k-2}+\binom{n-2}{k-3}\\&=\cdots\end{align}Can you take it from here?

1

A combinatorial proof based upon lattice paths.

The number of lattice paths of length $n+1$ from $(0,0)$ to $Q_k=(n-k+1,k)$ consisting of $(1,0)$-steps and $(0,1)$-steps is $$\binom{n+1}{n-k+1}=\binom{n+1}{k}$$ since we have to choose precisely $n-k+1$ $(1,0)$-steps from a total of $n+1$ steps.

                                enter image description here

We observe that each path which goes from $(0,0)$ to $Q_k$ crosses the vertical line $x=n-k$. We can therefore partition all $\binom{n+1}{k}$ paths according to points $P_j=(n-k,j), 0\leq j\leq k$ which are followed by a horizontal step to $(n-k+1,j)$ and the vertical steps which lead to $Q_k$.

Since there are precisely $\binom{n-k+j}{j}$ paths from $(0,0)$ to $P_j$ the number of all paths from $(0,0)$ to $Q_r=(n-k+1,k)$ is \begin{align*} \color{blue}{\binom{n+1}{k}=\binom{n}{k}+\binom{n-1}{k-1}+\cdots+\binom{n-k}{0}} \end{align*}

Markus Scheuer
  • 108,315
0

From Pascal's triangle we see that the sum is equivalent to

$$\sum_{i=n-k}^n\binom{i}{n-k}$$

then refer to Hockey-stick identity

$$\sum^m_{i=s}{i\choose s}={m+1\choose s+1} \qquad \text{ for } m,s\in\mathbb{N}, \quad m>s$$

user
  • 154,566