Hard time factoring Normal Distribution based on transformation problem.

62 Views Asked by At

My professor gave problems out to practice for our final on Wednesday. This problem is based on the transformation of two random variables. It a 5 part problem, so I will list the necessary portions for this part.

$W=X$
$Z=X+Y$

This leads to part (c), $f_{XY}(w,z-w)=f_X(w)f_Y(z-w)$, independence implied.
We are then given for part (d) $$f_U(u)=\frac{1}{\sqrt{2\pi}}\text{exp}{\bigg(-\bigg(u-\frac{z^2}{2}\bigg)\bigg)}$$
with mean $\frac{z}{2}$ and variance of $\frac{1}{2}$. Based on (c) and (d) he asks us to find $f_{X+Y}(z)$, where $X$ and $Y$ follow a standard normal distribution. Here is his answer $$\begin{align}f_{X+Y}(z)=&~ \int_{-\infty}^{\infty}\frac{1}{\sqrt{2\pi}}e^{-w^2/2}\frac{1}{\sqrt{2\pi}}e^{-(w-z)^2/2}\operatorname d w \\[1ex] =&~ \frac{1}{2\sqrt{\pi}}e^{-z^2/4}\int_{-\infty}^{\infty}\frac{1}{\sqrt{\pi}}\text{exp}\bigg(-\bigg(w-\frac{z^2}{2}\bigg)\bigg)\operatorname d w\end{align}$$

I understand that he is setting the problem up to make a substitution using what we have from part (d), but I do not understand how he factored $e^{-z^2/4}$ from the equation. I have played around with it in so many different ways, have checked laws of exponents, and I am seriously stumped. Any help would be much appreciated.

1

There are 1 best solutions below

0
On BEST ANSWER

You are right; you cannot get that end result.   It looks like a typo is the sauce of confusion.   Taking it slow we have:

$$\begin{align}f_{X+Y}(z)=&~ \int_{-\infty}^{\infty}\frac{1}{\sqrt{2\pi}}e^{-w^2/2}\frac{1}{\sqrt{2\pi}}e^{-(w-z)^2/2}\operatorname d w \\[1ex] =&~ \frac{1}{2\sqrt{\pi}}\int_{-\infty}^{\infty}\frac{1}{\sqrt{\pi}}e^{-((w-z)^2+w^2)/2}\operatorname d w \\[1ex] =&~ \frac{1}{2\sqrt{\pi}}\int_{-\infty}^{\infty}\frac{1}{\sqrt{\pi}}e^{-(w^2-wz+z^2/2)}\operatorname d w \\[1ex] =&~ \frac{1}{2\sqrt{\pi}}\int_{-\infty}^{\infty}\frac{1}{\sqrt{\pi}}e^{-(w^2-wz+z^2/4)-z^2/4}\operatorname d w \\[1ex] =&~ \frac{1}{2\sqrt{\pi}}e^{-z^2/4}\int_{-\infty}^{\infty}\frac{1}{\sqrt{\pi}}e^{-(w-\frac{z}{2})^2}\operatorname d w \\[1ex] =&~ \frac{1}{2\sqrt{\pi}}e^{-z^2/4}\int_{-\infty}^{\infty}\frac{1}{\sqrt{\pi}}\exp\left(-\left(w-\frac{z}{2}\right)^2\right)\operatorname d w \end{align}$$

Notice the placement of the square in the exponential function.   It is outside the inner brackets, rather than just over the $z$.

$$f_U(u) = \frac{1}{\sqrt{2\pi}}\exp{\left(-\left(u-\frac{z}{2}\right)^2\right)}$$