Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Suppose that we have an iid sample $X_1,\dots,X_n$ with a distribution function $F$. Denote $\bar X_n:=\frac{1}{n}\sum_{i=1}^n X_i$ and $\bar X_n^*:=\frac{1}{n}\sum_{i=1}^n X_i^*,$ where $X_1^*,\dots,X_n^*$ are iid from the empirical distribution function $\hat F$ given the sample $X_1,\dots,X_n$. Does it then follow that $$\sup_x|\mathbb P(\sqrt{n}\bar X_n \leq x) - \mathbb P(\sqrt{n}\bar X_n^* \leq x)| \rightarrow 0 \ \ \mathrm{a.s.}$$

share|improve this question
1  
Shouldn't there be a limit $n\to \infty$ added? –  Raskolnikov Mar 6 '12 at 17:48
add comment

1 Answer 1

Suppose the variance of the distribution $F$ is $\sigma^2<\infty$, and denote $\mu$ the mean of $X_i$. By Lindeberg–Lévy CLT, we have $$\mathbb{P}\left(\sqrt{n}\bar{X}_n\leq x\right)\rightarrow_d N\left(\mu,\sigma^2\right).$$ Denote $\Phi_{\mu,\sigma}(x)$ the CDF of a normal distributed random variable with mean $\mu$ and variance $\sigma^2$ evaluated at $x$. Because $\Phi_{\mu,\sigma}(x)$ is continuous throughout $\mathbb{R}$, we further have $$\sup_x\left|\mathbb{{P}}\left(\sqrt{n}\bar{X}_{n}\leq x\right)-\Phi_{\mu,\sigma}\left(x\right)\right|\rightarrow0,\,n\rightarrow\infty.$$ Next, we can show that $$\mathbb{P}\left(\sqrt{n}\bar{X}_n^*\leq x\right)\rightarrow_d N\left(\mu,\sigma^2\right).$$ Note that $X_1^*,\ldots,X_n^*$ are iid, and $$E\left(X_{i}^{*}\right)=E\left\{ E\left(X_{i}^{*}\mid X_{1},\ldots,X_{n}\right)\right\} =E\left(\frac{1}{n}\sum_{i=1}^{n}X_{i}\right)=\mu.$$ Similaryly, we have $Var(X_i^*)=\sigma^2$. Then the above convergence follows from the CLT. And the convergence is also uniform. In the end, your claims holds because \begin{aligned} & \sup_{x}\left|\mathbb{P}\left(\sqrt{n}\bar{X}_{n}\leq x\right)-\mathbb{P}\left(\sqrt{n}\bar{X}_{n}^{*}\leq x\right)\right|\\ = & \sup_{x}\left|\mathbb{P}\left(\sqrt{n}\bar{X}_{n}\leq x\right)-\Phi_{\mu,\sigma}\left(x\right)+\Phi_{\mu,\sigma}\left(x\right)-\mathbb{P}\left(\sqrt{n}\bar{X}_{n}^{*}\leq x\right)\right|\\ \leq & \sup_{x}\left|\mathbb{P}\left(\sqrt{n}\bar{X}_{n}\leq x\right)-\Phi_{\mu,\sigma}\left(x\right)\right|+\sup_{x}\left|\mathbb{P}\left(\sqrt{n}\bar{X}_{n}^{*}\leq x\right)-\Phi_{\mu,\sigma}\left(x\right)\right|\\ = & 0. \end{aligned}

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.