2
$\begingroup$

Suppose that $|X_j| \leq C <\infty$ is a uniformly bounded sequence for $j=1,2,\ldots$ with $X_j \sim [0, \sigma_j^2]$ and that $X_j$ are independent. I would like to show the UAN condition for central limit theorems, in that:

$$ \dfrac{\max_{1 \leq j\leq n} \sigma_j}{\sqrt{Var\left(\sum_{j=1}^n X_j\right)}} \to 0 $$

I am not sure how to bound this. The best I came up with still involves $n$ in the numerator. Does anyone have any ideas? Thanks.

$\endgroup$
1
$\begingroup$

Using independence and the fact that all the random variables are centered, to question reduces to the following: does the convergence $$\tag{*}\lim_{n\to +\infty} \frac{\max_{1\leqslant j\leqslant n}\sigma_j^2}{\sum_{i=1}^n\sigma_i^2 }  =0 $$ hold? If the series $\sum_{i=1}^{ +\infty} \sigma_i^2$ diverges, then we are done, since $\sigma_j^2\leqslant C^2$ for each $j$. Otherwise, (*) does not hold if the series $\sum_{i=1}^{ +\infty} \sigma_i^2$ converges because there exists $N$ such that $\sigma_N\neq 0$ (otherwise we would divide by zero) and for $n\geqslant N$, $$\frac{\max_{1\leqslant j\leqslant n}\sigma_j^2}{\sum_{i=1}^n\sigma_i^2 } \geqslant \frac{ \sigma_N^2}{\sum_{i=1}^n\sigma_i^2 }$$ an the right hand side converges to a positive constant.

$\endgroup$

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.