0
$\begingroup$

Gaussian is a continuous p.d.f and Poisson is discrete p.d.f. Let x be a random variable with Gaussian distribution and y be r.v with Poisson. Can a variable z=xy exist, if so then with what p.d.f? If not is there a way to make normal distribution discrete and do so?

$\endgroup$

1 Answer 1

0
$\begingroup$

Yes $Z$ exists, and it is continuous except for the fact that it has a point mass at 0. In fact you can show that its CDF satisfies $\mathbb{P}[Z\leq z]=\sum_{n=0}^\infty\mathbb{P}[nX\leq z]\mathbb{P}[Y=n]$, which is differentiable away from $z=0$. If you want to make a Normal distribution discrete, the easiest way to do this is by using the floor or ceiling function, as they round the Normal to make it an integer.

$\endgroup$
2
  • $\begingroup$ Can you please also tell me the resource(Books, e-material) where I can study this theory in detail? I need to model a process involving such P.D.F. I do not think that normal undergraduate Probability books include this, but not sure. $\endgroup$
    – Userhanu
    Feb 22, 2017 at 4:20
  • $\begingroup$ There's probably nothing with this exact process, but this is just another example of what you study under the umbrella of joint distributions. Questions like "What is the distribution of $X+Y$, $XY$, $\frac{X}{Y}$" given the distribution of $X$ and $Y$ are elementary problems found in whatever undergraduate textbook you can get your hands on. $\endgroup$ Feb 22, 2017 at 7:06

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .