What is the distribution of $Z=X_1X_2$?

71 Views Asked by At

So my question went like this:

If $X_1\sim\text{Unif(0,a)}$ and $X_2\sim\text{Unif(0,a)}$ independently, then find the distribution of $Z=X_1X_2$.

I tried the following:

Let $Z_1=\frac{X_1}{a}\sim\text{Unif(0,1)}$ and $Z_2=\frac{X_2}{a}\sim\text{Unif(0,1)}$. Thus $Z_1$ and $Z_2$ are independent.

Also, we know that, $-2\log_eZ_1\sim\chi_2^2$ and $-2\log_eZ_2\sim\chi_2^2$.

This implies,

$-2\log_e{Z_1Z_2}\sim\chi_4^2\implies-2\log_e{\frac{X_1X_2}{a^2}}\sim\chi_4^2\implies-2\log_e{\frac{Z}{a^2}}\sim\chi_4^2$.

Let $Z^*=-2\log_e{\frac{Z}{a^2}}\sim\chi_4^2$.

Thus, applying the Jacobian Transformation, I found the pdf of $Z$ as: $$f_Z(z)=\begin{cases}\frac{\left(-\log_e{z/a^2}\right)e^{-\log_e{z/a^2}}}{2z} & 0\leq z\leq a^2\\ 0 & \text{otherwise}\end{cases}$$ Is what I did, right?

Actually, before proceeding like this, I tried using the direct method using the Jacobian transformation but I was getting something like this:

$$\begin{align}f_Z(z)&=\int_0^a{f_{X_1}\left(\frac{z}{x_2}\right)f_{X_2}\left(x_2\right)\frac{1}{x_2}}.dx_2\\ &=\frac{1}{a^2}\int_0^a\frac{1}{x_2}.dx_2\end{align}$$ But this integral is not defined.

So does this mean that the direct formula of the Jacobian Transformation is invalid in this case?

1

There are 1 best solutions below

0
On

Just to simplify the notation I set $Z=XY$.

To use the jacobian Method set

$$\begin{cases} z=xy \\ v=x \end{cases}$$

The jacobian is $|J|=\frac{1}{v}$ thus

$$f_{UZ}(u,z)=\frac{1}{a^2v}\mathbb{1}_{(0;a^2)}(z)\cdot\mathbb{1}_{(\frac{z}{a};a)}(v)$$

Thus

$$f_Z(z)=\frac{1}{a^2}\int_{\frac{z}{a}}^a \frac{1}{v}dv=\frac{2\log a-\log z}{a^2}$$