0
$\begingroup$

I'm struggling with this exercise:

For each $k\in \mathbb{N}$, let $(X^{(k)}_n)_{n\in\mathbb{N}}$ be a sequence of real random variables converging to $0$ in probabilty as $n\to\infty$. Define for $k,n\in\mathbb{N}$, $$Y_n^{(k)}:=\sum\limits_{i=1}^k X_n^{(i)}$$ and for $\epsilon>0$ arbitrary $$f_n(k):=P(|Y_n^{(k)}|>\epsilon)$$ Let $k\in \mathbb{N}$. Show that $f_n(k)\to 0\quad (Y_n^{(k)} \xrightarrow{P}0)$, as $n\to \infty$

So the $k$ is the amount of sequences and the $n$ is at which position the sequence begins. So we have to show $\lim\limits_{n\to\infty}P(|Y_n^{(k)}-0|>\epsilon)=0$ This is the same as $$\lim\limits_{n\to\infty}P(|Y_n^{(1)}-0|>\epsilon,\cdots,|Y_n^{(k)}-0|>\epsilon)=0$$ $$=\lim\limits_{n\to\infty}P(|\sum\limits_{i=1}^1 X_n^{(i)}-0|>\epsilon,\cdots,|\sum\limits_{i=1}^k X_n^{(i)}-0|>\epsilon)=0 $$

So as far as I understand, this means that we have from left to right an increasing sequence of random variables where each random variable converges to $0$.

$$=\lim\limits_{n\to\infty}P(|\underbrace{X_n^{(1)}-0}_{= converges}|>\epsilon,\cdots,|\sum\limits_{i=1}^k X_n^{(i)}-0|>\epsilon)=0 $$

So can I some how argue, that each therefore the whole expression must also converge to $0$ in probabilty? Also we do not know whether $(X_n)_{n\in \mathbb{N}}$ are independent, but are the $X^{(i)}_k$ independent? Since I not know anything about independence I can't use something like the law of large numbers.

Thanks for help

$\endgroup$
  • $\begingroup$ "So we have to show $\lim\limits_{n\to\infty}P(|Y_n^{(k)}-0|>\epsilon)=0$ This is the same as $\lim\limits_{n\to\infty}P(|Y_n^{(1)}-0|>\epsilon,\cdots,|Y_n^{(k)}-0|>\epsilon)=0$" Well, no this is not the same. What makes you think this is? $\endgroup$ – Did May 5 '16 at 20:56
1
$\begingroup$

If $Y_n^k=\sum_{i=1}^k X_n^i>\epsilon$, one of those summands must be greater than $\epsilon/k$. So $P\{Y_n^k>\epsilon\}\le P\{\cup_{i=1}^k \{X_n^i>\epsilon/k\}\}\le \sum_{i=1}^k P\{X_n^i>\epsilon/k\} \to 0$ as $n\to\infty$.

$\endgroup$

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.