If $\sum_{n}E(|X_{n}-X|^{r})<\infty$ for some $r>0$. Show $X_{n}\rightarrow X$ a.s

65 Views Asked by At

I am trying to show the following: Let $X_{n},X$ r.v so that $\sum_{n}E(|X_{n}-X|^{r})<\infty$ for some $r>0$. Show $X_{n}\rightarrow X$ a.s My Proof:

For hypothesis, exist some $r>0$ with $r\in \mathbb{N}$ so that $\sum_{n}E(|X_{n}-X|^{r})<\infty$ then $\sum_{n}E(|X_{n}-X|^{r})$ converge. Now to convergence criterion we know that $$E(|X_{n}-X|^{r})\rightarrow 0$$ i.e. $$X_{n}\xrightarrow{L_{r}} X.$$ This follows readily from Chebyshev's inequality for any $\epsilon>0$ then $$P[|X_{n}-X|\geq \epsilon]\leq \frac{E(|X_{n}-X|^{r})}{\epsilon^{r}}$$ Now we take series in both ways then $$\sum_{n=1}^{\infty} P[|X_{n}-X|\geq \epsilon]\leq \sum_{n=1}^{\infty} \frac{E(|X_{n}-X|^{r})}{\epsilon^{r}} < \sum_{n=1}^{\infty} E(|X_{n}-X|^{r}) < \infty $$ therefore
$$\sum_{n=1}^{\infty} P[|X_{n}-X|\geq \epsilon]< \infty $$ by Borel-Cantelli lemma, we have $$P(\limsup_{n\rightarrow \infty} |X_{n}-X|\geq \epsilon )=0$$ Let $w\in [\lim_{n\rightarrow \infty} X_{n}=X]$. Those $M\geq 1$ exist $N \geq 1$ so that $n\geq N$ $$|X_{n}(w)-X(w)|<\frac{1}{M}$$ That means that $$w\in \bigcap_{M=1}^{\infty} \bigcup_{N=1}^{\infty} \bigcap_{n=N}^{\infty}\left( |X_{n}-X|\leq \frac{1}{M}\right)$$ therefore $$P[\lim_{n\rightarrow \infty} X_{n}=X]=1-P\left[ \lim_{n\rightarrow \infty} X_{n}\neq X \right]=1-P\left[ \bigcup_{M=1}^{\infty} \bigcap_{N=1}^{\infty} \bigcup_{n=N}^{\infty} |X_{n}-X|\geq \frac{1}{M}\right]\geq 1-P \sum_{M=1}^{\infty} \left[ \bigcap_{N=1}^{\infty} \bigcup_{n=N}^{\infty} |X_{n}-X|\geq \frac{1}{M}\right] $$ Now $$P[\lim_{n\rightarrow \infty} X_{n}=X]\geq 1-P \sum_{M=1}^{\infty} \left[ \bigcap_{N=1}^{\infty} \bigcup_{n=N}^{\infty} |X_{n}-X|\geq \frac{1}{M}\right]= 1-P \sum_{M=1}^{\infty} \left[\limsup_{n\rightarrow \infty}|X_{n}-X|\geq \frac{1}{M}\right]=1-0.$$

Then $$P[\lim_{n\rightarrow \infty} X_{n}=X ]=1$$

Therefore, $X_{n}\rightarrow X$ a.s .

Did I do anything wrong or is there something to improve?

1

There are 1 best solutions below

2
On BEST ANSWER

Looks correct. A simpler but equivalent proof is to note that $E(\sum_{n}|X_n - X|^r) = \sum_n E(|X_n - X|^r) < \infty$ implies $\sum_{n}|X_n - X|^r < \infty$ a.s., so $|X_n - X|^r \to 0$ a.s.