Mathematica Multinomial Expansion in Mathematica

AI Thread Summary
The discussion revolves around programming the multinomial expansion in Mathematica. The original poster seeks to create a function that computes the expansion of the form [Σ x_l]^n, emphasizing the challenge posed by the summation of nonnegative integer indices. They provide an example with (x_1 + x_2 + x_3)^2 to illustrate their goal. Responses highlight the limitations of using the Expand function for their specific case, particularly when dealing with exponentials in the expression [Σ A_n e^(-γ/γ_n)]^m. Suggestions include using a loop structure to handle variable numbers of terms and exponents, akin to an odometer, or constructing an m-fold Cartesian product of a set to facilitate the iteration. The conversation emphasizes the need for a more general approach to handle variable inputs effectively.
EngWiPy
Messages
1,361
Reaction score
61
Hello,

I have the following equation, which is simply the multinomial exapnsion:

\left[\sum_{l=1}^L\,x_l\right]^n=\sum_{k_1,k_2,\ldots,k_L}{n\choose k_1,k_2,\ldots,k_L}\prod_{l=1}^L\,x_l^{k_l}

where the summation is taken over all possibilities that the summation of the nonnegative integer indices k1,k2,...,kL equal n.

How can I program this using Mathematica??

Thanks in advance
 
Physics news on Phys.org
Any suggestion please? I am really need this. Any help will be greatly appreciated.
 
It really isn't clear what you want to do.
 
Hurkyl said:
It really isn't clear what you want to do.

Ok, first of all, thank you for replying. What I want to do is to write a function using Mathematica that computes the multinomial expansion, which is the right hand side equation in the first post. The difficulty as I see it, comes from the summation. For example:

(x_1+x_2+x_3)^2=\underbrace{\sum_{k_1,k_2,k_3}{2\choose k_1,k_2,k_3}\prod_{n=1}^{3}x_n^{k_n}}_{\text{Multinomial Expansion}}

where the summation in the right hand side takes place over all possibilities that k_1+k_2+k_3=2, and
\left(\begin{smallmatrix}2\\k_1,k_2,k_2\end{smallmatrix}\right)=\frac{2!}{k_1!\,k_2!\,k_3!}

Doing that yields to the final result:

(x_1+x_2+x_3)^2=x_1^2+x_2^2+x_3^2+2x_1x_2+2x_1x_3+2x_2x_3


I hope that I explain the point.

Regards
 
What's wrong with the Expand function?
 
Hurkyl said:
What's wrong with the Expand function?

The problem is that, the multinomial I have looks like this:

\left[\sum_{n=1}^NA_n\,\tex{e}^{-\gamma/\gamma_n}\right]^m

I want to exctract the exponentials and combine them with other exponentials and integrate them all over \gamma to do my derivation for the performance metrics.
 
So far I reached to this point:

Code:
x = 0;
For[k1 = 0, k1 <= 2, k1++,
 For[k2 = 0, k2 <= 2, k2++,
  For[k3 = 0, k3 <= 2, k3++,
   If[k1 + k2 + k3 == 2,
    x = x + Multinomial[k1, k2, k3]*x1^k1*x2^k2*x3^k3];
   If[k1 == 2 && k2 == 2 && k3 == 2, Print[x]]]]]

x1^2+2 x1 x2+x2^2+2 x1 x3+2 x2 x3+x3^2

But how can I write it in general form, when the number of terms and the exponent are variables?
 
I'm still not sure why you think this will result in anything different than the expand function...


When the number of loops is variable, you essentially have to implement an odometer -- e.g. as a list whose n-th element is the value of the n-th variable. Then you have just one for loop that initializes / steps this odometer.


It might be easier, however, to construct the m-fold Cartesian product of the set {0, 1, ..., N} with itself. There is surely a builtin function to compute Cartesian products of lists, and it should be easy enough to use that to write a function that iterates it m times. But I don't have mathematica on my home computer to tell you what that function would be.

(The Cartesian product of A with B is the set AxB whose elements are all length-2 lists whose first element is from A and whose second element is from B)
 
Back
Top