If $X_i, i=1,...,$ are independent, identically distributed $\operatorname{N}(0,1)$ random variables, and $Y_i = X_i^2$ are independent $\operatorname{Gamma}(\frac{1}{2},\frac{1}{2})$ RVs, use the Central Limit Theorem to calculate the approximate values of the probabilities:

$\sum_{1}^{50} {Y_i \gt 40}$ and $\sum_{1}^{50} {Y_i \lt 60}$

I'm not really sure how to go about this.

  • 1
    $\begingroup$ Hi Mathlete, this appears to be standard bookwork and looks like it's probably related to some course. Would you mind adding the self-study tag please? [You might like to read through the faq item relating to homework (a synonym for self-study) and the self study tag wiki for a description of how those questions are dealt with. In particular, there are some expectations of both askers and answerers that differ slightly from the other kinds of questions.] $\endgroup$ – Glen_b Apr 23 '13 at 0:06
  • $\begingroup$ Because $E(Y_i)=E(X_i^2)=E((X_i-0)^2)$ is the variance of $X_i$, clearly the mean of $Y_i$ is $1$. Because $E((Y_i-1)^2)=E(Y_i^2)-2E(Y_i)+E(1)=E(X_i^4)-2+1=2$, the variance of $Y_i$ is $2$. That's all you need to solve this problem. $\endgroup$ – whuber Apr 25 '13 at 13:23

It's been a couple of semesters since the delta method, so bear with me here.

What does the CLT tell us?

$\frac{\sqrt{n}(\bar{x}-\mu)}{\sigma} \rightarrow N(0,1)$

You can use the delta method to determine the asymptotic distribution of $X^2$.

$\sqrt{n}(g(X_n) - g(\mu)) \rightarrow N(0,\sigma^2(g'(\mu))^2)$

So you need to set your $g(X)$, in this case $X^2$. $g(\mu) = \mu^2$. you know the distribution of $X^2$ so you know what the $\sigma^2$ is. From there you should be able to turn this into a N(0,1) and use z-tables to figure out what the probability of $\Sigma Y_i > 40$ and $\Sigma Y_i < 60$. Hope this helps.

  • $\begingroup$ So I would get $\sqrt{n}$$(X_n^2)$ -> $N(0,(g'(0))^2)$ How would I follow on from there? $\endgroup$ – Mathlete Apr 22 '13 at 23:02
  • $\begingroup$ Considering I get $g'(\mu)=0$ have I done something wrong? $\endgroup$ – Mathlete Apr 22 '13 at 23:07
  • $\begingroup$ no. $g(\mu) = \mu^2$ so $g'(\mu) = 2\mu$, $g'(\mu)^2$ = $4\mu^2$ $\endgroup$ – Eric Peterson Apr 22 '13 at 23:08
  • 1
    $\begingroup$ bmbolstad.com/teaching/Stat215b/Lab4/delta_method.pdf Here are a few examples on how to carry out the delta method and get a normal distribution asymptotically. Hope this helps. Once you've seen a few carried out, maybe you can solve this one. Also, look at the wiki page for delta method: en.wikipedia.org/wiki/Delta_method $\endgroup$ – Eric Peterson Apr 22 '13 at 23:24
  • 1
    $\begingroup$ @ClarkW.Griswold my guess is they've probably just done the CLT, and haven't even mentioned the Delta method. The CLT is sufficient for this question. $\endgroup$ – Glen_b Apr 23 '13 at 0:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.