0
$\begingroup$

Suppose $X_0,X_1,\cdots$ are iid $Poisson(\theta)$ r.v.

Define $Y_k = X_k I_{\{ X_{k-1} = 0 \}}$ for $k=1,2,3,\cdots$

Find the limit of $Var(\sqrt{n}\overline{Y_n})$ and asymptotic distribution of $\sqrt{n}\left(\bar{Y}_{n}-\mathrm{E} \bar{Y}_{n}\right)$.


It's not difficult to find $E(Y_k) = E(X_k)P(X_{k-1}=0) = \theta exp(-\theta)$. But when it comes to variance, I know that $Y_k$ is independent to $Y_{k+i}$ for $i>1$. However, I dont know how to calculate the covariance of $Y_k$ and $Y_{k+1}$. Besides, since $Y_k$ are not i.i.d. random variables, how can I use CLT to get the asymptotic distribution?

$\endgroup$

1 Answer 1

0
$\begingroup$

The covariance of $Y_k$, $Y_{k+1}$ can be found directly: $$ \mathbb E(Y_kY_{k+1})=\mathbb E(X_kX_{k+1}I_{\{X_{k-1}=0, X_k=0\}}) = \mathbb E(0) = 0. $$ Indeed, if $X_k\neq 0$ then $I_{\{X_{k-1}=0, X_k=0\}}=0$. If $X_k=0$, then $X_kX_{k+1}=0$. So the product $Y_kY_{k+1}$ is zero for every elementary event $\omega$. Then $$ \text{Cov}(Y_k,Y_{k+1}) = 0- \mathbb E(Y_k)\mathbb E(Y_{k+1}) = -\theta^2 e^{-2\theta}. $$ So using variance of a sum we get $$ \text{Var}\left(\sqrt{n}\overline{Y_n}\right) = \frac1n \text{Var}\left(\sum_{i=1}^n Y_i\right) = \frac1n\bigl(n\text{Var}(Y_1) + 2(n-1)\cdot\text{Cov}(Y_1,Y_2) \bigr) = \ldots \tag{1}\label{1} $$ Here $\mathbb E(Y_1^2)=\mathbb E(X_1^2)\mathbb P(X_0=0)=(\theta^2+\theta)e^{-\theta}$, and $$ \text{Var}(Y_1)=(\theta^2+\theta)e^{-\theta} - \theta^2e^{-2\theta}. $$

Concerning the limiting distribution, I’m not sure that one can use the well-known fact that a sequence of one-dependent random variables satisfies the CLT. Look at this paper for example.

In our case $m=1$ and limiting distribution is normal with zero mean and variance $$ \text{Var}(Y_1) + 2\text{Cov}(Y_1,Y_2). $$ This variance is exactly the limit of \eqref{1} as $n\to\infty$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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