3
$\begingroup$

I need help figuring out the following.

Let $X$ and $Y$ have the bivariate normal distribution

$$ f_{XY}(x,y) = \frac{1}{2\pi \sqrt{1-\rho^2}} \exp \left( -\frac{x^2 +y^2 - 2\rho xy}{2(1-\rho^2)} \right) $$

Show that $X$ and $Z=\frac{Y−ρX}{\sqrt{1−ρ2}}$ are independent standard normal random variables.

For $X$, is it asking to show that $f(x)$ is equal to the density function of the standard normal? If yes, how can I do that with $Z$? And how could I prove they are independent given the information provided? I would really appreciate any help here.

I also have to proof $P(X>0,Y>0) = \frac 14 + \frac{1}{2\pi}\arcsin \rho$. How can I do this using the answer to the above? I'm kind of new trying to understand this concept.

$\endgroup$
  • $\begingroup$ I don't think the result is correct unless $X$ and $Y$ are standard normal to start with $\endgroup$ – Dilip Sarwate Mar 7 '16 at 4:05
  • $\begingroup$ "(since $X$ and $Y$ are independent)" No, they are not. They are correlated random variables. $\endgroup$ – Dilip Sarwate Mar 7 '16 at 14:22
1
$\begingroup$

I assume your initial distribution is $$ f_{XY}(x,y) = \frac{1}{2\pi \sqrt{1-\rho^2}} \exp \left( -\frac{x^2 +y^2 - 2\rho xy}{2(1-\rho^2)} \right) $$

Apply the transformation $Y = Z\sqrt{1-\rho^2} + \rho X$

Including the Jacobian determinant $\sqrt{1-\rho^2}$, the pdf for $X,Z$

$$ f_{XZ}(x,z) = \frac{1}{2\pi } \exp \left( -\frac{x^2 +z^2(1-\rho^2) +2\rho zx\sqrt{1-\rho^2} +\rho^2x^2 - 2\rho (zx\sqrt{1-\rho^2} + \rho x^2)}{2(1-\rho^2)} \right) $$

You can show that this is equivalent to the pdf for independent standard normals.

$$ f_{XZ}(x,z) = \frac{1}{2\pi } \exp \left( -\frac{x^2 +z^2}{2} \right) $$

$\endgroup$
  • $\begingroup$ thank you very much for your answer. I'm kind of new to this particular topic. How could I show your function is equivalent to the independent standard normals? $\endgroup$ – user320559 Mar 7 '16 at 10:27
  • $\begingroup$ Just cancel terms and you'll get the expression at the end :). Are you unsure how the expression at the end is the pdf for independent standard normal variables? If so, $f_{XZ} = f_X(x)f_Z(z)$ for independent $X$ and $Z$, and $f_X(x) = (2\pi)^{-1/2} \exp(-x^2/2)$ is the PDF for the standard normal. $\endgroup$ – D.A.N. Mar 7 '16 at 10:36
  • $\begingroup$ thanks! I also have to proof P(X>0,Y>0)= (1/4) + (1/2pi)(arcsin p). I thought I would have to do the integral from 0 to infinity of density function of the standard normal and multiply it by itseld (since X and Y are independent), but the answer I get is 1/2. Should I be using a different approach for this problem? I will also add this to my original question. $\endgroup$ – user320559 Mar 7 '16 at 12:04

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.