Exercise about almost sure convergence

187 Views Asked by At

I have the following problem:

Let $X_n$, $n \geq 1$ be independent r.v. identically distributed with the probability function:

$P(X_{i} = k) = P(X_{i} = - k) = \frac{p}{2} q^{k-1}$, $k = 1,2,...,m$, $P(X_{i} = m + 1) = P(X_{i} = -m -1) = \frac{q^{m}}{2}$

where $m>1$, $p\in (0,1)$ and $q = 1-p$. Calculate the almost surely limit of $Z_{n} = \frac{\frac{X_{1}}{X_{2}} + \frac{X_{3}}{X_{4}} +...+ \frac{X_{2n-1}}{X_{2n}}}{X_{1}^{2} + X_{2}^{2} + ... + X_{n}^{2}}$.

The exercise also adds a suggestion that is as follows: Prove that the numerator's limit is $0$ and that the denominator's limit is always greater than $0$.

So far I think I can say that the denominator's limit must be greater than $0$ because for all $x\in \mathbb{R}-\{0\}$, $x^{2} > 0$. I know that $X_{i} \neq 0$ and, therefore, $X_{1}^{2} + X_{2}^{2} + ... + X_{n}^{2} > 0$.

Let me know if there's any mistake in my reasoning and also if you find any way to proceed to find the answer to the problem.

1

There are 1 best solutions below

1
On

I'll give you another hint. Note that we may write $$Z_n = \frac{W_n}{Y_n}$$ where $W_n = \frac{1}{n}\left( \frac{X_1}{X_2} + \frac{X_3}{X_4} + \ldots + \frac{X_{2n-1}}{2n} \right)$ and $Y_n = \frac{1}{n}\left( X_1^2 + X_2^2 + \ldots +X_n^2 \right)$. Notice that both $W_n$ and $Y_n$ are the means of iid random variables with finite mean. The strong law of large numbers tells us that $$W_n \xrightarrow{a.s.} E\left[ \frac{X_1}{X_2} \right] $$ and $$Y_n \xrightarrow{a.s.} E[X_1^2]>0$$ Can you compute these expectations and take it from there?