Cube of summation of independent (uncorrelated) random variables

1k Views Asked by At

Basically, I am trying to prove an equality involving skewness of summation of random indepdent variables. However, I am stuck with the part after expressing sum of cube to $E[X^3] - 3E[X^2]\mu +2\mu^3$. I tried to solve using similar techniques for solving variances but now I am lost. And I could not find any helpful clue or proof.

Given setting is as the following:

When $X = x_1 + x_2 + ... + x_n$, how to compute $E[X^3]$?

Also, how to express $E[X^3]$ using triple summations with $i, j, k = 1, 2, ... , n$ when $i, j, \& k$ may (or may not - possibly in both cases) equivalent?

Many thanks in advance.

1

There are 1 best solutions below

1
On BEST ANSWER

Also, how to express $E[X^3]$ using triple summations with $i, j, k = 1, 2, ... , n$ when $i, j, \& k$ may (or may not - possibly in both cases) equivalent?

$$\begin{align}\mathsf E(X^3) ~=~& \mathsf E({\sum}_{i,j,k} X_iX_jX_k) \\[1ex]~=~& {\sum}_i\mathsf E(X_i^3)+3{\sum}_{(i,j,k): i=j , j\neq k}\mathsf E(X_i^2X_k)+{\sum}_{(i,j,k):i\neq j,j\neq k,k\neq i}\mathsf E(X_iX_jX_k)\\[1ex]~=~& n\,\mathsf E(X_i^3)+3\,n\,(n-1)\,\mathsf E(X_i^2X_k)+n\,(n-1)\,(n-2)\,\mathsf E(X_iX_jX_k)\end{align}$$

Assuming these are independent and identically distributed random variables.

$$\begin{align}\mathsf E(X^3) ~=~& n\,\mathsf E(X_i^3)+3\,n\,(n-1)\,\mathsf E(X_i^2)\,\mathsf E(X_i)+n\,(n-1)\,(n-2)\,\mathsf E(X_i)^3\end{align}$$