1

I was trying to understand why does binomial coefficient work and finally could find a formula, but first let me explain how I came up with that: Assume we have a box containing $5$ balls such that each ball has it's own unique number, we are going to choose $2$ of $5$ balls , in how many ways it can be done?

We can illustrate the balls with numbers from $1$ to $5$ (for simplicity): $1,2,3,4,5$

The all possible chooses are:

$$\left(\color{red}{1},2\right) ,\left(1,3\right),\left(1,4\right),\left(1,5\right) \tag{4 chooses}$$ $$\left(\color{red}{2},3\right),\left(2,4\right),\left(2,5\right) \tag{3 chooses}$$ $$\left(\color{red}{3},4\right),\left(3,5\right)\tag{2 chooses}$$ $$\left(\color{red}{4},5\right)\tag{1 chooses}$$

The number of chooses is $10$

From this simple example we conclude that if the numbers where from $1$ to $n$ then we had: $$\left(\color{red}{1},2\right),\left(1,3\right),\left(1,4\right),\left(1,5\right),...,\left(1,n\right)\tag{n-1 chooses}$$ $$\left(\color{red}{2},3\right),\left(2,4\right),\left(2,5\right),...,\left(2,n\right)\tag{n-2 chooses}$$ $$\left(\color{red}{3},4\right),\left(3,5\right),...,\left(3,n\right)\tag{n-3 chooses}$$ $$\left(\color{red}{4},5\right),...,\left(3,n\right)\tag{ n-4 chooses}$$ $$\vdots$$ $$\left(\color{red}{n-1},n\right)\tag{1 chooses}$$ Summing gives:

$$\left(n-1\right)+\left(n-2\right)+...+\left(2\right)+\left(1\right)=\frac{n\left(n-1\right)}{2}=\frac{n\left(n-1\right)\left(n-2\right)!}{2!\left(n-2\right)!}=\frac{n!}{2!\left(n-2\right)!}={{n}\choose{2}}$$

For another more difficult example assume we have a box containing $6$ balls such that each ball has it's own unique number, we are going to choose $4$ of $6$ balls from the box,again for simplicity we assume that the numbers on the balls are from $1$ to $6$, then we list these balls in a line with their numbers:

$1,2,3,4,5,6$

The all possible chooses are:

1,2,3,4    1,2,4,5   1,2,5,6
1,2,3,5    1,2,4,6
1,2,3,6

1,3,4,5    1,3,5,6
1,3,4,6

1,4,5,6

2,3,4,5     2,3,5,6
2,3,4,6

2,4,5,6

3,4,5,6

I used this very nice pattern to generalize the situation when we want to choose $4$ elements from a set with $n$ elements:

$$ \color{blue}{(1,2,3,4) \; \; \; \; \; (1,2,3,5) \; \;\; \;\; (1,2,3,6) \; \;\; \;\; (1,2,3,7) \; \; \; \;\; ... \; \;\; (1,2,3,n)}$$

$$ \color{blue}{(1,2,4,5) \; \; \; \; \; (1,2,4,6) \; \; \; \; \; (1,2,4,7) \; \; \; \; \; ... \; \; \; \; \; (1,2,4,,n)} $$

$$ \color{blue}{(1,2,5,6) \; \; \; \; \; (1,2,5,7) \; \; \; \; \; ... \; \; \; \; \; (1,2,5,,n)} $$

$$ \color{blue}{(1,2,6,7) \; \; \; \; \; ... \; \; \; \; \; (1,2,5,,n)} $$

$$\color{blue}{\vdots}$$

$$\color{blue}{(1,2,n,n-1)}$$


$$\color{blue}{(1,3,4,5) \; \; \; \; \; (1,3,4,6) \; \; \; \; \; (1,3,4,7) \; \; \; \; \; ... \; \; \; \; \; (1,3,4,n)}$$ $$\color{blue}{(1,3,5,6) \; \; \; \; \; (1,3,5,7) \; \; \; \; \; ... \; \; \; \; \; (1,3,5,n)}$$ $$\color{blue}{(1,3,6,7) \; \; \; \; \; ... \; \; \; \; \; (1,3,5,n)}$$ $$\color{blue}{\vdots}$$

$$\color{blue}{(1,3,n,n-1)}$$


$$\vdots$$$$\vdots$$ $$\vdots$$$$\vdots$$


$$\color{blue}{(1,n-3,n-2,n-1) \; \; \; \; \; (1,n-3,n-2,n)} $$

$$\color{blue}{(1,n-3,n-1,n)} $$


$$\color{blue}{(1,n-2,n-1,n)}$$ $$\underbrace{\text{Number of chooses}}_\textrm{$\sum_{k=1}^{n-3}k\left(n-\left(k+2\right)\right)$}$$

$$\color{red}{ (2,3,4,5) \; \; \; \; \; (2,3,4,6) \; \; \; \; \; (2,3,4,7) ... \; \; \; \; \; (2,3,4,n)}$$ $$\color{red}{ (2,3,5,6) \; \; \; \; \; (2,3,5,7) \; \; \; \; \; ... \; \; \; \; \; (2,3,5,n)}$$ $$\color{red}{ (2,3,6,7) \; \; \; \; \;... \; \; \; \; \; (2,3,6,n)}$$ $$\color{red}{ \vdots}$$ $$\color{red}{ (2,3,n,n-1)}$$


$$ \color{red}{ (2,4,5,6) \; \; \; \; \; (2,4,5,7) \; \; \; \; \; ... \; \; \; \; \;(2,4,n-1,n)}$$ $$\color{red}{ (2,4,6,7) \; \; \; \; \; ... \; \; \; \; \;(2,4,n-1,n)}$$$$\color{red}{ \vdots}$$ $$\color{red}{ (2,4,n-1,n)}$$


$$\vdots$$$$\vdots$$ $$\vdots$$$$\vdots$$


$$\color{red}{(2,n-3,n-2,n-1) \; \; \; \; \; (2,n-3,n-2,n)}$$

$$\color{red}{(2,n-3,n-1,n)}$$


$$\color{red}{(2,n-2,n-1,n)}$$

$$\underbrace{\text{Number of chooses}}_\textrm{$\sum_{k=1}^{n-4}k\left(n-\left(k+3\right)\right)$}$$

$$\color{green}{(n-4,n-3,n-2,n-1) \; \; \; \; \; (n-4,n-3,n-2,n)}$$


$$\color{green}{(n-4,n-2,n-1,n)}$$ $$\underbrace{\text{Number of chooses}}_\textrm{$\sum_{k=1}^{2}k\left(n-\left(k+\left(n-3\right)\right)\right)$}$$ $$(n-3,n-2,n-1,n)$$ $$\underbrace{\text{Number of chooses}}_\textrm{$\sum_{k=1}^{1}k\left(n-\left(k+\left(n-2\right)\right)\right)$}$$

Summing the elements with a same color gives us the number of chooses::

(n-3)+(n-4)+(n-5)+(n-6)...+1
(n-4)+(n-5)+(n-6)...+1
(n-5)+(n-6)+...+1
...
1

(n-3)+2(n-4)+3(n-5)+4(n-6)+...+(n-3)
(n-4)+2(n-5)+(n-6)+...+(n-4)
(n-5)+2(n-6)+...+(n-5)
...
1

$\vdots$


1

For example the first box tells us that in how many ways we can choose $4$ balls such that the ball with number $1$ is in all of the choices.(the balls are chosen based on their value, for example if we chosen a balls with number $1$ then our next choose should be a ball with a number greater than $1$, this explains why in the last step we just can fix the ball with number $n-4$)

Continuing this way finally the last box tells us in how many ways we can choose $4$ balls such that the ball with number $n-4$ is in all of the choices. Summing the all terms contained in each gray box we get:

$$\sum_{k=1}^{n-3}k\left(n-\left(k+2\right)\right)+\sum_{k=1}^{n-4}k\left(n-\left(k+3\right)\right)+...+\sum_{k=1}^{1}k\left(n-\left(k+\left(n-2\right)\right)\right)$$$$=\sum_{m=3}^{n-1}\sum_{k=1}^{n-m}k\left(n-\left(k+\left(m-1\right)\right)\right)$$$$=\sum_{m=3}^{n-1}\left[n\sum_{k=1}^{n-m}k-\sum_{k=1}^{n-m}k^{2}-\left(m-1\right)\sum_{k=1}^{n-m}k\right]$$$$=\sum_{m=3}^{n-1}n\frac{\left(n-m\right)\left(n-m+1\right)}{2}$$$$-\sum_{m=3}^{n-1}\left(2\left(n-m\right)+1\right)\frac{\left(n-m\right)\left(n-m+1\right)}{6}$$$$-\sum_{m=3}^{n-1}\left(m-1\right)\frac{\left(n-m\right)\left(n-m+1\right)}{2}$$$$=\sum_{m=3}^{n-1}\left[\frac{\left(n-m\right)\left(n-m+1\right)}{3!}\left(n-m+2\right)\right]$$$$=\sum_{m=3}^{n-1}\frac{\left(n-m+2\right)\left(n-m+1\right)\left(n-m\right)}{3!}\frac{\left(n-m-1\right)!}{\left(n-m-1\right)!}$$$$=\sum_{m=3}^{n-1}\frac{\left(n-m+2\right)!}{3!\left(n-m-1\right)!}=\sum_{m=0}^{n-4}{{n-m-1}\choose{3}}$$

This is equal to ${{n}\choose{4}}$ , but I don't know how to show that.

After deriving this formula I generalized that for the case when we want to choose $m$ objects from a set with cardinality $n$:

$$\color{red}{\sum_{m=0}^{n-k}{{n-m-1}\choose{k-1}}={{n}\choose{k}}}$$

Can someone prove this? (It would be nice if someone use index shifting such that the formula is defined when both $n,k=0$)

Also the formula is defined when both $n$ and $k$ are $\ge1$

PS...The pattern is more beautiful than I expressed and they actually makes a pine tree shaped pattern such that is every step starting with choosing a new ball fixing in all of the chooses the steps of the shape decreases till we finally just have one choose.

  • 1
    It seems to me you already gave a combinatorial proof of the red displayed equation? You showed that the two sides represent two ways of counting the same thing. What else do you want? Are you looking for an algebraic proof of what you proved combinatorially? – joriki Feb 11 '20 at 18:52
  • @joriki,well first I just used some very simple counting without using any kind of any combinatorial concepts, then I derived a formula that works and it's validity has been checked , but I just want a proof for the last formula to complete my way ( even algebraic ways I've ever seen never explain where does the idea come from but finally I find it and a proof is needed ) –  Feb 11 '20 at 18:57

1 Answers1

2

We have

$$\sum_{m=0}^{n-k} {n-m-1\choose k-1} = \sum_{m=0}^{n-k} {n-m-1\choose n-k-m} \\ = [z^{n-k}] (1+z)^{n-1} \sum_{m=0}^{n-k} z^m (1+z)^{-m} $$

Note that with the second form we have for $m$ in the range $[0,n-k]$ that $n-k-m$ is non-negative even when $n=k$ or $k=0.$ The coefficient extractor enforces the range and we may continue with

$$[z^{n-k}] (1+z)^{n-1} \sum_{m\ge 0} z^m (1+z)^{-m} \\ = [z^{n-k}] (1+z)^{n-1} \frac{1}{1-z/(1+z)} \\ = [z^{n-k}] (1+z)^{n-1} \frac{1+z}{1+z-z} \\ = [z^{n-k}] (1+z)^{n} = {n\choose n-k} = {n\choose k}$$

as claimed.

Remark. When $k=0$ we get $\sum_{m=0}^n \frac{(n-m-1)^\underline{n-m}}{(n-m)!}$, the terms of which are all zero except for $m=n$ which is $\frac{(-1)^{\underline{0}}}{0!} = 1$ for a total of one, which is in turn ${n\choose 0}$.

Marko Riedel
  • 61,317