pdf of product of standard normal and uniform

667 Views Asked by At

Even though this question has been answered somewhat here, here, here, I'd like to ask what the probability density function of $z = x y$ is, where $x \sim \mathcal{U}[-1,1]$, $y \sim \mathcal{N}(0,1)$, and $z$ are all univariate. Furthermore, what are the mean and variance of $z$? (By symmetry, it is easy to see $\mathbb{E}[z] = 0$, but I want a more rigorous proof.)

Edit: What if $x \sim \mathcal{U}[-\theta,\theta]$?

1

There are 1 best solutions below

3
On BEST ANSWER

Since $|X|,\,|Y|$ have respective PDFs $1_{[0,\,1]}(x),\,\sqrt{\tfrac{2}{\pi}}e^{-y^2/2}1_{[0,\,\infty)}(y)$, $|Z|=|X||Y|$ has on-$[0,\,\infty)$ PDF $\int_0^1x^{-1}\sqrt{\tfrac{2}{\pi}}e^{-z^2/(2x^2)}dx=\tfrac{1}{\sqrt{2\pi}}\Gamma\left(0,\,\tfrac12z^2\right)$ in terms of the upper incomplete Gamma function. Since $X,\,Y$ have symmetric distributions, $Z$'s PDF on $\Bbb R$ is $\tfrac{1}{\sqrt{8\pi}}\Gamma\left(0,\,\tfrac12z^2\right)$.