5
$\begingroup$

When $X,Y$ are iid random variables, uniformly distributed on $[0,1]$, then $Z=X*Y$ has the density $-\log(z)$, as it is shown here:

product distribution of two uniform distribution, what about 3 or more

The question is: For which density of $X,Y$ is the product $Z$ uniformly distributed on $[0,1]$?

$\endgroup$
4
$\begingroup$

Let $W_1,W_2$ be independent Gamma random variables with $\alpha=1/2$ and $\beta=1$. Then $e^{-W_1}\,e^{-W_2}=e^{-(W_1+W_2)}$ where $W_1+W_2$ is exponential with mean 1. An easy transformation argument shows that $e^{-(W_1+W_2)}$ has a uniform(0,1) distribution.

$e^{-W_1}$ and $e^{-W_2}$ are i.i.d. random variables whose product is uniform(0,1).

What kind of density does $e^{-W}$ have? The density of $W$ is $f(w)={1\over\sqrt{\pi w}}\, e^{-w}\, {\bf 1}_{(0,\infty)}(w).$ Setting $X=e^{-W}$, we have $|dw/dx|= 1/x$ so that the density of $X$ is $$g(x)= {1\over\sqrt{\pi \log(1/x)}} e^{-(-\log(x))} (1/x) {\bf 1}_{(0,\infty)}(-\log(x)) = {1\over\sqrt{\pi \,\log(1/x)}}\, {\bf 1}_{(0,1)}(x).$$

Here is a graph of this function:

enter image description here

$\endgroup$
1
  • $\begingroup$ Impressive! Thank you so much! $\endgroup$ – Daniel Siebert Jul 30 '15 at 16:36

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.