Say $Y$ is a random variable with support on $(1, 2, 3, \ldots ) $ and has the PMF:
$$\frac{C}{2^x}$$
How would you find the MGF from here and how would you arrange your answer so that it is not in sum notation?
Say $Y$ is a random variable with support on $(1, 2, 3, \ldots ) $ and has the PMF:
$$\frac{C}{2^x}$$
How would you find the MGF from here and how would you arrange your answer so that it is not in sum notation?
On
By equating the total probability to unity, it can be shown that $C=1$.
Then, by definition, \begin{eqnarray*} M_{X}(t)&=& E(e^{tX})=\sum_{x=1}^{\infty}e^{tX}P(X=x)=\sum_{x=1}^{\infty}\left(\dfrac{e^t}{2}\right)^x\\ &=&\left(\dfrac{e^t}{2} + \left(\dfrac{e^t}{2}\right)^{2} + \left(\dfrac{e^t}{2}\right)^{3} + \cdots \right)\\ &=& \dfrac{e^t}{2} \left( 1 + \left(\dfrac{e^t}{2}\right) + \left(\dfrac{e^t}{2}\right)^{2} + \cdots \right)\\ &=&\dfrac{e^t}{2}\cdot \dfrac{1}{1-e^t/2}\\ M_{X}(t)&=&\dfrac{e^t}{2-e^t} \end{eqnarray*}
First we consider that PMF must sum up to unity so $(P_y(Y) = \frac{1}{2^y})$ for $y \in \mathbb N$: $$\sum_{y=1}^{\infty}\frac{C}{2^y}=1 \Rightarrow C \frac{\frac{1}{2}}{1-\frac{1}{2}}=1 \Rightarrow C = 1$$
Then by definition of MGF we see that ($i=\sqrt{-1}$): $$\Phi_Y(\omega) = \mathbb{E}\{ e^{i\omega y} \}= \sum_{y=1}^{\infty} \frac{e^{i\omega y}}{2^y} = \frac{\frac{e^{i\omega}}{2}}{1-\frac{e^{i\omega}}{2}}=\frac{e^{i\omega}}{2-e^{i\omega}}$$
Also we have the geometric series formula: for a complex number $z$ with $|z|<1$ we have: $$\sum_{i=a}^{N}z^n = S = z^a+z^{a+1} +...z^N$$ $$\sum_{i=a}^{N}z^{n+1} = zS = z^{a+1}+z^{a+2} +...z^{N+1}$$ $$zS-S = (z-1)S = z^{N+1}-z^a \Rightarrow S = \frac{z^{N+1}-z^a}{z-1}= \frac{z^a-z^{N+1}}{1-z}$$ When $N\rightarrow\infty$ we have $|z|^{N+1}\rightarrow 0$ and the formula becomes $$S=\frac{z^a}{1-z}$$ For above consider the case $a=1$. In first one $z=\frac{1}{2}$ and in the second $z = \frac{e^{i\omega}}{2}$
Also it can be interpreted as the discrete Fourier's transform of the sequence of the PMF which can be obtained from z-transform, also.