0
$\begingroup$

Let the random variables $Y_1,\ldots,Y_n$ be independent and identically distributed (i.i.d.) (standard) Inverse Gaussian random variables with parameters $\mu$ and $\lambda$.

Then, let the random variables $\tilde{A}$ and $A$ be given, which are defined as follows: $A = \sqrt n \cdot \tilde{A} = \sqrt n \left(\frac{1}{n} \cdot \sum_{i=1}^n \left[e^{b_i Y_i}\right] - \theta\right)$ ($b_i \in \mathbb{R}$ ($i=1,\ldots,n$) are constants (with $b_i \neq b_j$ for any $i \neq j$) and $\theta$ is a parameter).

Goal: Determine what ultimately follows if one uses the (Lindeberg-Lévy) Central Limit Theorem on $e^{b_i Y_i}$.

The CLT informs us that $\sqrt n (\overline{Y} - E(Y_i)) \xrightarrow[]{D} \mathcal{N}(0,Var(Y_i))$, i.e. $\sqrt n \left(\frac{1}{n} \cdot \sum_{i=1}^n [Y_i] - \mu \right) \xrightarrow[]{D} \mathcal{N}(0,\frac{\mu^3}{\lambda})$.

It seems to me that it follows that $\sqrt n \left(\frac{1}{n} \cdot \sum_{i=1}^n \left[e^{b_i Y_i}\right]- e^{b_i \mu}\right) \xrightarrow[]{D} \ldots$.

Question: How to proceed from here (i.e. how to determine the limit distribution of $\sqrt n \cdot \tilde{A}$)?

$\endgroup$
0
$\begingroup$

You may consider the random variable \begin{align} \frac{1}{s_n} \sum_{i=1}^n (e^{b_i Y_i} - e^{b_i \mu}) \end{align} where $s_n^2 = \sum_{i=1}^n \sigma_i^2$ and $\sigma_i$ is the variance of $e^{b_iY_i}$. Then, you can check whether the Lindeberg's condition is satisfied. If it is satisfied, using the Lindeberg CLT \begin{align} \frac{1}{s_n} \sum_{i=1}^n (e^{b_i Y_i} - e^{b_i \mu}) \xrightarrow[]{D} \mathcal{N}(0,1) \end{align} as $n\to\infty$.

$\endgroup$

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.