I am given that $X$ has c.f. $\phi(t)$, I need to find the random variable whose c.f. is equal to $\frac{1}{2-\phi(t)}$ in terms of $X$.

My idea is that express $\frac{1}{2-\phi(t)}$ as a series, since $|\phi(t)| \leq 1$ so we have $$ \frac{1}{2-\phi(t)} = \sum_{n = 0}^{\infty}\frac{\phi(t)^n}{2^{n+1}} $$ From the question asked here, I am guessing that this c.f. corresponds to the random variable (I may be wrong):

$$ Z = \sum_{n = 0}^{\infty}I(A = n)Z_n $$ where $P(A = n) = \frac{1}{2^{n+1}}, \ n = 0,1,2,...$ and $Z_n = \sum_{i = 1}^{n}Y_i$, $Z_0 = 0$, $Y_i$ are iid r.v.'s s.t. $Y_i \sim X$.

But I don't know how to prove this, can anyone point out a general direction? Thanks so much!


You already got the answer, you are just not completely writing that out. In probability context we usually write

$$\displaystyle Z = \sum_{i=1}^A Y_i$$

as a random sum, with the convention that $Z = 0$ when $A = 0$. We assumed $A$ is independent of $Y_i$ also, and they are defined as what you written. Then by law of total expectation, the characteristic function of $Z$ is

$$ \begin{align} \phi_Z(t) &= E\left[\exp\left\{itZ\right\}\right] \\ &= E\left[\exp\left\{it\sum_{i=1}^A Y_i\right\}\right] \\ &= \sum_{n=0}^{\infty} E\left[\exp\left\{it\sum_{i=1}^A Y_i\right\} \Bigg| A=n \right] \Pr\{A = n\} \\ &= \Pr\{A = 0\} + \sum_{n=1}^{\infty} E\left[\exp\left\{it\sum_{i=1}^n Y_i\right\}\right] \Pr\{A = n\} \\ &= \frac {1} {2} + \sum_{n=1}^{\infty} E\left[\prod_{i=1}^n\exp\left\{it Y_i\right\}\right] \frac {1} {2^{n+1}} \\ &= \frac {1} {2} + \sum_{n=1}^{\infty} \prod_{i=1}^n E\left[\exp\left\{it Y_i\right\}\right] \frac {1} {2^{n+1}} \\ &= \frac {1} {2} + \sum_{n=1}^{\infty} \phi_X(t)^n \frac {1} {2^{n+1}} \\ &= \sum_{n=0}^{\infty} \frac {\phi_X(t)^n} {2^{n+1}} \\ &= \frac {1/2} {1 - \phi_X(t)/2} \\ &= \frac {1} {2 - \phi_X(t)} \end{align}$$

where line $1$ using the definition of characteristic function, line $2$ using the definition of $Z$, line $3$ using the law of total expectation, line $4$ using the independence of $A$ and $Y_i$ and the convention of the random sum, line $5$ using the basic property of exponential function, line $6$ using the independence of $Y_i$, line $7$ using the definition of characteristic function of $Y_i$ and they are identically distributed with the identical CF $\phi_X$, and the remaining lines are just some algebra to simplify the expression.

I have not study those complex analysis so if there is any holes in the above steps in regarding complex number, please help to fill them. I guess this should work.


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.