1
$\begingroup$

This question already has an answer here:

Can some one help me point in the right direction or point to some resources that will help me prove that sum of two jointly distributed Gaussian r.v. with a given correlation coefficient is also a Gaussian r.v. Thanks

$\endgroup$

marked as duplicate by whuber Apr 13 '15 at 15:59

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

  • 2
    $\begingroup$ Many people take the definition of jointly Gaussian random variables as a collection of random variables such that $\sum_i a_iX_i$ is a Gaussian random variable for all choices of real numbers $a_i$. Thus, what you are asking for is a tautology for these people. Since $$\begin{align}E[X+Y]&= E[X]+E[Y]\\\operatorname{var}(X+Y)&=\operatorname{var}(X)+\operatorname{var}(Y)+2{\rho}\sqrt{\operatorname{var}(X)\operatorname{var}(Y)}\end{align}$$ apply even for nonGaussian random variables, there is little left to do... $\endgroup$ – Dilip Sarwate Apr 12 '15 at 22:55
  • $\begingroup$ Thank you. But is the proof this simple? Don't I have to show something more? $\endgroup$ – user100503 Apr 12 '15 at 23:04
  • $\begingroup$ If I recall correctly, it was shown in Kendall's The Advanced Theory of Statistics. Vol. 1: Distribution Theory. $\endgroup$ – corey979 Apr 13 '15 at 0:02
  • 1
    $\begingroup$ What is your definition of jointly Gaussian random variables? If you use the one stated in my previous comment, then you don't need to "show" very much more than what I wrote. If you start from the joint density, then there can be a lot of algebra involved. A proof starting from characteristic functions, instead of joint densities, is given in this answer written by @probabilityislogic $\endgroup$ – Dilip Sarwate Apr 13 '15 at 13:14