Hi I need some help with the next exercise: Let $Y_1,Y_2...$ be independent random variables with mean zero and finite variances and let $X_n=Y_1+Y_2+....+Y_n$

Use Doob´s inequality to show that:

$$P(\max_{1\leq k\leq n}|X_k|>\epsilon)\leq\frac{\sum_{k=1}^nVar Y_k}{\epsilon^2}\ \ \ for\ \ \epsilon>0$$

Here, I did the next:

Using the Doob's inequality, I have:

$P(\max_{1\leq k\leq n}|X_k|>\epsilon)\leq\frac{E[X_n^2]}{\epsilon^2}\ \ \ for\ \ \epsilon>0$, and since the expectation of every random variable is $0$. I have: $$Var(X_n)=E[X_n^2]-E[X_n]E[X_n]$$, note that $E[X_n]=E[Y_1]+E[Y_2]+...=0$ and also $Var(X_n)=Var[Y_1+...+Y_n]=\sum_{k=1}^nVar[Y_k]$. using these facts I get the claim. Is ok my propose?

2)Asuume further that $$\sum_{k=1}^{\infty}\frac{1}{k^2}Var(Y_k)<\infty$$

Use the previous inequality to show that $$Z_n=\sum_{k=1}^{\infty}\frac{Yk}{k}\xrightarrow{a.s}Z$$ as $n\rightarrow\infty$ for some finite random variable $Z$.

3) Deduce that $$\frac{1}{n}\sum_{k=1}^{n}Y_k\xrightarrow{a.s}0\ \ as\ \ n\rightarrow\infty$$


I think Part(i) is correct.

Part (ii) is the Kolmogorov's two series theorem, whose proof uses the Kolmogorov inequality in Part(i).

Part (iii): Kronecker’s Lemma: if $\sum_{k=1}^n \frac{Y_k}{k}$ converges, then $\frac{\sum_{k=1}^n Y_k}{n} \rightarrow 0.$


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.