1
$\begingroup$

Let $X_i$ and $Y_i$ $(i\in\mathbb{N})$ be two strictly stationary sequences of real valued random variables with finite variance. We have two empirical distribution functions $F_{1,n}:=\frac{1}{n} \sum_{i=1}^n 1\{X_i\leq u\}$ and $F_{2,n}:=\frac{1}{n} \sum_{i=1}^n 1\{Y_i\leq u\}$. Assume that \begin{align*} \sup_{u\in\mathbb{R}} |F_{1,n}(u)-F_{2,n}(u)|\xrightarrow{p} 0 \end{align*}in probability and that $F_{1,n}(u)\xrightarrow{n\rightarrow\infty}c(u) Z$ in distribution where $Z$ is a real valued random variable with continuous distribution and $c: \mathbb{R}\rightarrow \mathbb{R_{>0}}$ continuous and increasing. Follows now that $F_{2,n}(u)\xrightarrow{n\rightarrow\infty}c(u) Z$?

$\endgroup$
1
$\begingroup$

Yes. To see that, show and use the fact that if $(X_n,n\geqslant 1)$ is a sequence of random variables converging in distribution to $X$, and $(Y_n,n\geqslant 1)$ converges in probability to $0$, then $X_n+Y_n\to X$ in distribution.

We have to prove that $P(X_n+Y_n\leqslant t)\to P(X\leqslant t)$ at each point where $s\mapsto P(X\leqslant s)$ is continuous. Fix $t_0$ such a point. Then use the fact that the term $Y_n$ does not perturb too much.

$\endgroup$
  • $\begingroup$ Thanks for the quit answer! Can you give me a tip how to proof your statement? thanks! $\endgroup$ – stroem Jul 5 '13 at 7:46
  • $\begingroup$ See edit.${}{}$ $\endgroup$ – Davide Giraudo Jul 5 '13 at 8:09
  • $\begingroup$ Thanks. You were really helpful! $\endgroup$ – stroem Jul 5 '13 at 11:17
  • $\begingroup$ Ok, we have now \begin{align*}F_{2,n}(u)\xrightarrow{n\rightarrow\infty}c(u) Z.\end{align*} for all $u\in\mathbb{R}$. Can we now argue that $F_{2,n}(u)\xrightarrow{n\rightarrow\infty}c(u) Z$ uniform in $u$? $\endgroup$ – stroem Jul 10 '13 at 8:59

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.