2
$\begingroup$

Let $X$ and $Y$ be independent and identically distributed random variables with moment generating function $M(t)=E(e^{tX});\ \ -\infty<t<\infty$ then $E(\dfrac{e^{tX}}{e^{tY}})$ equals ?

$(A)=M(t)M(-t)$

$(B)=1$

$(C)=(M(t))^2$

$(D)=\frac{M(t)}{M(-t)}$

My input: Since its given that random variables are i.i.d so $E(\dfrac{\require{\cancel}\cancel{e^{tX}}}{\cancel{e^{tY}}})=E(1)=1 $

can I do that?

$\endgroup$
  • 1
    $\begingroup$ What happens e.g. for $X:=1$ (constant random variable) and $Y:=2$ (also constant random variable)? Does the expression equal $1$? $\endgroup$ – saz Feb 4 '18 at 16:38
  • $\begingroup$ You assumed $X=Y$ with probability $1$ when you did the cancellation. That is not necessarily true. $\endgroup$ – StubbornAtom Feb 4 '18 at 16:42
  • $\begingroup$ @saz Oh yes i was thinking totally wrong. I took a case when they both take same probability. Silly me. $\endgroup$ – Daman Feb 4 '18 at 16:42
  • $\begingroup$ @StubbornAtom i am showing new work wait. $\endgroup$ – Daman Feb 4 '18 at 16:42
  • $\begingroup$ Another example: $X,Y \sim {\cal N}(0,1)$ iid. $\endgroup$ – GNUSupporter 8964民主女神 地下教會 Feb 4 '18 at 17:28
4
$\begingroup$

$X,Y$ are independent of each other, so $e^{tX}$ and $e^{tY}$ are independent of each other. This gives $M(t) = E[e^{tX}] = E[e^{tY}]$

Moreover, $M(t) = E[e^{tX}]$ and $M(-t) = E[e^{-tY}]$ are independent of each other, this gives $$E\left[\frac{e^{tX}}{e^{tY}}\right] = E[e^{tX} \cdot e^{-tY}] = E[e^{tX}] \cdot E[e^{-tY}] = M(t) M(-t). $$

$\endgroup$
  • 1
    $\begingroup$ I was showing same work. Anyways thanks :) $\endgroup$ – Daman Feb 4 '18 at 16:48
3
$\begingroup$

No you cannot do that. What you can do: $$ E\left(\frac{e^{t X}}{e^{t Y}}\right)=E\left(e^{t X} e^{-t Y}\right) = E(e^{t X}) E(e^{-t Y}) = M(t) M(-t)$$ where the second step is due to independence. As the random variables are identically distributed $M_X(t) = M_Y(t) =M(t)$.

$\endgroup$
3
$\begingroup$

Observe that since $X$ and $Y$ are independent, it follows that $e^{tX}$ and $e^{-tY}$ are independent for $t\in\mathbb{R}$. Thus $$ Ee^{tX}e^{-tY}=Ee^{tX}E{e^{-tY}}=M_{X}(t)M_{Y}(-t)=M_{X}(t)M_X(-t) $$ since $X, Y$ are equal in distribution.

$\endgroup$

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.