Show that the rate of convergence is infinity

72 Views Asked by At

Given $X_1, X_2, \ldots , X_n\sim \operatorname{i.i.d. DU}(K)$. Prove that the rate of convergence of $\widehat{K}=X_{(1)}+X_{(n)}-1$ is $\infty$.

I think I want to start by showing that $X_{(1)} \xrightarrow{P} 1$ so $\widehat{K}\xrightarrow{P}1+K-1=K$, but I don't know how to begin rigorously or how to show the rate of convergence is infinite.

1

There are 1 best solutions below

2
On BEST ANSWER

The words "the rate of convergence is infinite" can be interpreted in different ways. For example, they can be treated like this: for any $\alpha>0$ $$ n^\alpha(\hat K-K) \xrightarrow{P} 0. $$ Use the definition of convergence in probability to show that for any $\alpha>0$ $$\tag{1}\label{1} n^\alpha(X_{(1)}-1) \xrightarrow{P} 0 \ \text{ and } \ n^\alpha(X_{(n)}-K) \xrightarrow{P} 0. $$ Say, for any $\varepsilon>0$ $$ \mathbb P\left(n^\alpha|X_{(1)}-1|\geq \varepsilon\right)=\mathbb P\left(X_{(1)}\geq 1+\frac{\varepsilon}{n^\alpha}\right)\leq \mathbb P\left(X_{(1)}\geq 2\right). $$ Prove that the last probability tends to zero, and you will get the first statement in (\ref{1}) proved. The same for the second statement.