5
$\begingroup$

How does one calculate the distribution of a random variable $Z = X + Y$, where $X$ is Poisson distributed and $Y$ is normally distributed and independent of $X$? How to do that in general case? Do you know any textbook about this topic? It's really hard for me to seach on this topic, because I don't know what to call this problem.

$\endgroup$
3
  • 2
    $\begingroup$ See any probability textbook. The distribution function of sum of iid variables is convolution of individual distribution functions. $\endgroup$
    – mpiktas
    Apr 6, 2012 at 8:06
  • $\begingroup$ @mpiktas But X and Y are not identically distributed: X is poissonian and Y is gaussian. $\endgroup$
    – Tomas
    Apr 6, 2012 at 19:08
  • $\begingroup$ @tomas, I meant to write id. Independence is the key here. $\endgroup$
    – mpiktas
    Apr 7, 2012 at 8:26

1 Answer 1

5
$\begingroup$

Here I would calculate the distribution of $Z=X+Y$ using conditional decomposition. If $X \sim {\cal P}(\lambda)$ and $Y \sim {\cal N}(\mu, \sigma^2)$ then the conditional distribution of $Z$ given $X=x$ is the distribution of $x+Y$ which is the normal distribution ${\cal N}(\mu+x, \sigma^2)$. Finally, denoting by $f_{\mu, \sigma}$ the density function of ${\cal N}(\mu, \sigma^2)$, the density of $Z$ is the function $$z \mapsto \sum_{x \geq 0}p_x f_{\mu+x, \sigma}(z)$$ where $p_x=\Pr(X=x)=\exp(-\lambda)\frac{\lambda^x}{x!}$. Of course it remains to do the calculation of this sum.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.