31

What is the mean and variance of Squared Gaussian: $Y=X^2$ where: $X\sim\mathcal{N}(0,\sigma^2)$?

It is interesting to note that Gaussian R.V here is zero-mean and non-central Chi-square Distribution doesn't work.

Thanks.

Michael Hardy
  • 1
  • 30
  • 276
  • 565
kaka
  • 1,722
  • 3
  • 19
  • 31

2 Answers2

37

We can avoid using the fact that $X^2\sim\sigma^2\chi_1^2$, where $\chi_1^2$ is the chi-squared distribution with $1$ degree of freedom, and calculate the expected value and the variance just using the definition. We have that $$ \operatorname E X^2=\operatorname{Var}X=\sigma^2 $$ since $\operatorname EX=0$ (see here).

Also, $$ \operatorname{Var}X^2=\operatorname EX^4-(\operatorname EX^2)^2. $$ The fourth moment $\operatorname EX^4$ is equal to $3\sigma^4$ (see here). Hence, $$ \operatorname{Var}X^2=3\sigma^4-\sigma^4=2\sigma^4. $$

Cm7F7Bb
  • 16,114
  • 4
  • 31
  • 56
24

Note that $X^2 \sim \sigma^2 \chi^2_1$ where $\chi^2_1$ is the Chi-squared distribution with 1 degree of freedom. Since $E[\chi^2_1] = 1, \text{Var}[\chi^2_1] = 2$ we have $E[X^2] = \sigma^2, \text{Var}[X^2] = 2 \sigma^4$.

iballa
  • 583
  • 3
  • 6
  • Thanks. Does $X^2 \sim \sigma^2 \chi^2_1$ holds true too if $X\sim\mathcal{N}(\mu,\sigma^2)$ ? – kaka Jan 19 '14 at 02:14
  • 4
    @kaka No. If $X = \sigma Y + \mu$ where $Y$ is a standard normal, then $X^2 = \sigma^2 Y^2 + 2 \sigma \mu Y + \mu^2$. – iballa Jan 28 '14 at 20:16