Convergence in probability of order statistic random variables with uniform distribution.

777 Views Asked by At

Please verify if my solution to this exercise has any mistake (Specially in the second part):

Let $(X_n)$ be a sequence of iid random variables with distribuction uniform $U[0,1]$. Show that $X_{(n)} \to 1$ in probability.

Let $0<\epsilon\leq1$. Then $$P(|X_{(n)} - 1|> \epsilon) = P(|1 - X_{(n)}| > \epsilon) = P(1 - X_{(n)} > \epsilon) = P(X_{(n)} < 1 - \epsilon) = F_{X_{(n)}}(1-\epsilon) = {(F_{X_n}(1-\epsilon))}^n = (1-\epsilon)^n \to 0$$

If $\epsilon>1$, then $P(|X_{(n)}-1|>\epsilon)=0$, since $X_{n}$ is never greater than $1$ nor less than $0$, then neither is $X_{(n)}$, so the distance from $X_{(n)}$ to $1$ can't be greater than $1$.

Thanks.

1

There are 1 best solutions below

0
On BEST ANSWER

Your answer is definitely right.

Actually we only care the case when $\epsilon$ is very small rather than large $\epsilon$, because you want to prove when $n \rightarrow \infty$, there is no difference between $X_{(n)}$ and $1$ in almost every point of domain (except the subset of domain whose measure is 0), so you will try to use smaller $\epsilon$ as possible.