Does sequence convergence implies convergence in probability?

312 Views Asked by At

I am currently learning for my statistic final exam and I want to understand one part of the proof for the Delta method. I am going to skip some details of the proof and only ask one important question:

Let $(\Omega,\mathcal{A},P)$ be a propability space. Let $(c_n)_{n\in\mathbb{N}}$ be a sequence in $\mathbb{R}$ with $c_n\longrightarrow c$ for a $c\in\mathbb{R}$. Define the real random variable $Y_n:= c_n$. Then $Y_n$ converges in probability against the constant random variable $Y:=c$, i.e. \begin{equation} \lim_{n\rightarrow \infty}P(|Y_n-Y|>\varepsilon)=0 \end{equation} for all $\varepsilon >0$.

I want to know if that is true or the implication is true in the other direction, or is maybe an equivalence. Thanks for your answers.

1

There are 1 best solutions below

0
On BEST ANSWER

This is true, because in fact $Y_n\to Y$ almost surely, which implies convergence in probability.

The reverse direction is also true. Suppose $Y_n\to Y$ in probability. Then note that since $Y_n$ and $Y$ are a.s. constant, $$P(|Y_n-Y|>\epsilon) = \begin{cases} 1 & \text{ if } |c_n-c|>\epsilon\\ 0 & \text{ if } |c_n-c|\leq \epsilon. \end{cases}$$

Now let $\epsilon>0$ be given. Then since the above expression converges to $0$ as $n\to \infty$, we have that for $n$ large enough $P(|Y_n-Y|>\epsilon)=0$ (try to prove this yourself). By the above expression, this implies that $|c_n-c|\leq\epsilon$ if $n$ is large enough.