0
$\begingroup$

Suppose $x_i, i=1, ..., n$ are independently distributed with mean $0$ and variance $\sigma^2$. Then $\frac{1}{\sqrt n}\sum_{i=1}^n a_ix_i$ converges to a normal distribution with mean 0 and variance $\sigma^2 p$ if $\frac{1}{n}\sum_{i=1}^n a_i^2 \to p > 0$.

How to prove? What I learned was that central limit theorem was for independently identically distributed random variables. But this is not the case here.

| cite | improve this question | |
$\endgroup$
  • $\begingroup$ What are $a_i$, and $p$? Where did this question arise? $\endgroup$ – Carl Apr 18 at 17:50
  • $\begingroup$ $a_i, p$ are constants. $\endgroup$ – John Smith Apr 18 at 18:02
0
$\begingroup$

What you have is what is called the weighted sum CLT. As far as I know, the exact conditions for it to hold is that $\sup_{1\leq i \leq n} \frac{a_i^2}{\sum_{j=1}^n a_j^2} \xrightarrow{n\to \infty} 0$.

This theorem states that for $\{W_n\}$ an i.i.d sequence of real random variables wih common mean 0 and variance 1,

$$ \frac{1}{\sqrt{\sum_{j=1}^n a_j^2}} \sum_{i=1}^n a_i W_i \xrightarrow{d} \mathcal{N}(0,1) \tag{1}$$

You can prove it using Lyapunov CLT which is for independent but not identically distributed variables. You should check the answer to this question in this same forum).

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ So, if $\frac{1}{n}\sum_{i=1}^n a_i^2 \to p > 0$ implies $\frac{\max a_i^2}{\sum_{i=1}^n a_i^2} \to 0$, then a CLT can be applied to prove the original assertion. But I am stuck with this now. $\endgroup$ – John Smith Apr 18 at 19:57
  • $\begingroup$ I was just stating the exact minimum conditions. In your case this of course holds. Note that any $a_i$ must be finite. Say $\sup_i a_i=c<\infty$. Then of course $\frac{c}{np}\to 0$ as $n\to \infty$. $\endgroup$ – Oriol B Apr 18 at 20:27
  • $\begingroup$ Could you please elaborate more on the derivations? I don't see why $\sup_i a_i=c$ is a constant, how do you get $\frac{c}{np}$. $\endgroup$ – John Smith Apr 18 at 20:35
  • $\begingroup$ What I am saying is that the series must be convergent and hence each $a_i$ must be finite. Since you are dividing by something that goes to infinity (since you have a $n$ term) this limit goes to 0. I only defined the supremum as $c$ to further clarify (you can obviate this). Also if your sequence $\{a_i\}$ is bounded (probably the case) $\sup \equiv \max$ (I just used supremum to be more precise). Was the confusion about this? $\endgroup$ – Oriol B Apr 18 at 21:07
  • $\begingroup$ Thanks although I wish I could understand your arguments better. "the series" means what here? what limit is "this limit"? Where did you get the result that the sequence $\{a_i\}$ is bounded? $\endgroup$ – John Smith Apr 18 at 21:24

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.