I have two independent continuous random variables $X$ and $Y$ with

  • pdf's : $f(x)$ and $f(y)$
  • cdf's : $F(x)$ and $F(y)$
  • a constant $a$

I am trying to express using the given pdf/cdf functions the following expressions:

  1. $Pr(X\cdot Y > a)$
  2. $E(X\cdot Y | (X\cdot Y < a))$

The question becomes trivial if I find the pdf/cdf of a variable $Z = X\cdot Y$

This corresponds to the product distribution, however I am not sure how to use it to derive the cdf of $Z$ in order to find expression 1. and 2.

  • 2
    $\begingroup$ $$f_Z(z)=\int f_X(x)f_Y\left(\frac{z}x\right)\frac1{|x|}dx$$ $\endgroup$ – Did Jul 14 '16 at 19:58

In general if we let $X=g(U,V), Y=h(U, V)$ (and assuming this is a one-to-one coordinate transfer) we can use the Jacobian formula:

$$\begin{align}f_{U,V}(u,v)=&~ \Big\lvert\dfrac{\partial (x,y)}{\partial (u,v)}\Big\rvert\,f_{X,Y}(x,y) \\[1ex] =&~ \Big\lvert\dfrac{\partial (g(u,v), h(u,v))}{\partial (u,v)}\Big\rvert\,f_{X,Y}(g(u,v), h(u,v))\end{align}$$

Where $\dfrac{\partial (x,y)}{\partial (u,v)}$ is shorthand for the determinant of the Jacobian matrix $$\Big\lvert\dfrac{\partial (x,y)}{\partial (u,v)}\Big\rvert=\begin{Vmatrix}\partial x/\partial u & \partial x/\partial v \\ \partial y/\partial u & \partial y/\partial v \end{Vmatrix}$$

  In this specific case we take $U=X, V= XY$ (or $Z$), so $X=U, Y=V/U$ (or $Z/X$) ... and also $f_{X,Y}(x,y) = f(x)\,f(y)$ due to being independent and identically distributed

$$\begin{align}f_{X,Z}(x,z) =&~ \Big\lvert\dfrac{\partial (x,z/x)}{\partial (x,z)}\Big\rvert\,f(x)\,f(z/x)\\[1ex] =&~ \begin{Vmatrix}\partial x/\partial x & \partial x/\partial z \\ \partial (z/x)/\partial x & \partial (z/x)/\partial z \end{Vmatrix}\,f(x)\,f(z/x)\\[1ex] =&~ \begin{Vmatrix}1 & 0 \\ -z/x^2 & 1/x \end{Vmatrix}\,f(x)\,f(z/x)\\[1ex] =&~ \lvert \tfrac 1 x \rvert\,f(x)\,f(z/x)\end{align}$$ And hence the marginal is $\displaystyle f_{XY}(z) = \int_\Bbb R \lvert \tfrac 1 x \rvert\,f(x)\,f(z/x)\operatorname d x$


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.