1
$\begingroup$

Never used multi-indexed summations in my life, neither has anyone else I know.

https://en.wikipedia.org/wiki/Multinomial_theorem

does not define an upper index for the multi-indexed sum, which gives the sum no meaning whatsoever.

However, it does for some reason have an ordered set of numbers as an index for a multi-indexed sum and then completely fails to explain the procedure for each of those indices in that set. Is it a nested sum? A product of sums? A sum of products? Is each sum a coefficient of some polynomial? Is each polynomial a coefficient of some sum? And to what end-index? How do you use any part of this theorem? The world may never know.

$\endgroup$
1
  • $\begingroup$ Comments are not for extended discussion; this conversation has been moved to chat. $\endgroup$
    – Pedro
    Commented Mar 23, 2019 at 13:00

1 Answer 1

1
$\begingroup$

Here we look at the connection between binomial expansion and multinomial expansion for the cases $n=2$ and $n=3$ which might give a better idea what's going on.

Case $n=2$:

\begin{align*} \color{blue}{(x_1+x_2)^n}&\color{blue}{=\sum_{k_1=0}^n\binom{n}{k_1}x_1^{k_1}x_2^{n-k_1}}\\ &=\sum_{k_1=0}^n\frac{n!}{k_1!(n-k_1)!}x_1^{x_1}x_2^{n-k_1}\\ &=\sum_{{k_1+k_2=n}\atop{k_1,k_2\geq 0}}\frac{n!}{k_1!k_2!}x_1^{k_1}x_2^{k_2}\tag{1}\\ &\,\,\color{blue}{=\sum_{{k_1+k_2=n}\atop{k_1,k_2\geq 0}}\binom{n}{k_1,k_2}x_1^{k_1}x_2^{k_2}} \end{align*}

Comment:

  • In (1) we introduce a new index variable $k_2=n-k_1$. Note we also state in the index region $k_1,k_2\geq 0$ which is sometimes silently assumed.

Case $n=3$:

\begin{align*} \color{blue}{(x_1+x_2+x_3)^n}&=(x_1+(x_2+x_3))^n\\ &=\sum_{k_1=0}^n\binom{n}{k_1}x_1^{k_1}(x_2+x_3)^{n-k_1}\\ &\,\,\color{blue}{=\sum_{k_1=0}^n\binom{n}{k_1}x_1^{k_1}\sum_{k_2=0}^{n-k_1}\binom{n-k_1}{k_2}x_2^{k_2}x_3^{n-k_1-k_2}}\\ &=\sum_{k_1=0}^n\sum_{k_2=0}^{n-k_1}\frac{n!}{k_1!(n-k_1)!}\cdot\frac{(n-k_1)!}{k_2!(n-k_1-k_2)!}x_1^{k_1}x_2^{k_2}x_3^{n-k_1-k_2}\\ &=\sum_{k_1=0}^n\sum_{k_2=0}^{n-k_1}\frac{n!}{k_1!k_2!(n-k_1-k_2)!}x_1^{k_1}x_2^{k_2}x_3^{n-k_1-k_2}\\ &=\sum_{k_1=0}^n\sum_{{k_2+k_3=n-k_1}\atop{k_2,k_3\geq 0}}\frac{n!}{k_1!k_2!k_3!}x_1^{k_1}x_2^{k_2}x_3^{k_3}\tag{2}\\ &=\sum_{{k_1+k_2+k_3=n}\atop{k_1,k_2,k_3\geq 0}}\frac{n!}{k_1!k_2!k_3!}x_1^{k_1}x_2^{k_2}x_3^{k_3}\\ &\,\,\color{blue}{=\sum_{{k_1+k_2+k_3=n}\atop{k_1,k_2,k_3\geq 0}}\binom{n}{k_1,k_2,k_3}x_1^{k_1}x_2^{k_2}x_3^{k_3}}\\ \end{align*}

  • In (2) we introduce a new index variable $k_3=n-k_1-k_2$ similarly as we did in (1).

Hint: You might find chapter 2: Sums in Concrete Mathematics by R.L. Graham, D.E. Knuth and O. Patashnik helpful. It provides a thorough introduction in the usage of sums.

$\endgroup$
8
  • $\begingroup$ How do you "ensure" $k_1$ and $k_2$ always add up to $n$? They could be any random integer. It looks like there are $|K|-1$ nested sums. Frankly I would prefer nested sums to over-specified notation. Would the next nested sum for 4 terms to the power of $n$ be from $k_3$ to $k_1-k_2$ or to $n-k_1-k_2$? $\endgroup$
    – Vane Voe
    Commented Mar 23, 2019 at 21:29
  • $\begingroup$ @VaneVoe: The expression $k_1+k_2=n$ ensures that $k_1$ and $k_2$ always add up to $n$. Since we also have $k_1,k_2\geq 0$ we always have valid tupel $(k_1,k_2)=(k_1,n-k_1)$ as in the line above. $\endgroup$ Commented Mar 23, 2019 at 21:50
  • $\begingroup$ That expressing gives no information on what $k_1$ and $k_2$ actually are. Is $k_1$ 1? Is it 2? Is it $n-1$? Is it $n-2$? Is it $3!$? $\endgroup$
    – Vane Voe
    Commented Mar 23, 2019 at 21:55
  • $\begingroup$ But is the nested sum principal right? That would make it easier to understand. $\endgroup$
    – Vane Voe
    Commented Mar 23, 2019 at 22:14
  • 1
    $\begingroup$ @VaneVoe: There is no difference at all. Recall we have an equality chain. It's just a matter of convenience which kind of representation we choose. In my answer an interesting aspect could be to go from the more familiar binomial representation step by step to the multinomial representation and try to grasp why these representations all mean the same. $\endgroup$ Commented Mar 23, 2019 at 22:20

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .