The random variable Z is equal to X + Y. Where X and Y are independent and they both are exponentially distributed with parameter $\lambda = 1$;

How do i find the probability density function of Z? Firstly i searched about this and I've found about convolution yet I don't manage to apply it properly?

$$F_Z(z) = \int_{-\infty}^{\infty}F_X(x)F_Y(z-x) dx$$ And from this i go to

$$\int_{0}^{\infty}e^{-x}e^{-(z-x)}dx = \int_{0}^{\infty}e^{-z}e^{-x+x}dx = \int_{0}^{\infty}e^zdx = xe^{-z}|_0^{\infty}$$ which seems clearly wrong??

How can i solve it? Can this be solved in the following way : find the common distribution function and use the derivates of it to get to the density function?

Could really use some references to consult as well!

Thank you!

  • $\begingroup$ Two mistakes: first, $F_X=F_Y$ is not $F_X(x)=e^{-x}$ but $$F_X(x)=1-e^{-x}$$ second, the convolution formula for CDFs is $$F_Z(z)=\int_\mathbb RF_X(x)f_Y(z-x)dx$$ where the integral involves a CDF and a PDF. $\endgroup$ – Did Dec 31 '17 at 0:46
  • $\begingroup$ @Did they do explicitly say probability density function even if they used the wrong notation for it. $\endgroup$ – spaceisdarkgreen Dec 31 '17 at 1:14

You are provided that the distributions are exponential: $f_X(x)= \mathsf e^{-x}\mathbf 1_{x\geqslant 0}\\ f_Y(y)= \mathsf e^{-y}\mathbf 1_{y\geqslant 0}$

Where the support is indicated using an indicator : $~\mathbf 1_{Q}=\begin{cases}1 & \textsf{if}& Q\textsf{ is true}\\0 &:&\textsf{otherwise}\end{cases}$

The convolution is thus $$\begin{split}f_Z(z) &= \int_{-\infty}^{\infty} f_X(x)\cdot f_Y(z-x)~\mathsf d x \\ &= \int_{-\infty}^{\infty} \mathsf e^{-x}\mathbf 1_{x\geqslant 0}\cdot \mathsf e^{-(z-x)}\mathbf 1_{(z-x)\geqslant 0}~\mathsf d x\\ &= \int_{-\infty}^{\infty} \mathsf e^{-z+x-x}\mathbf 1_{0\leqslant x}\mathbf 1_{x\leqslant z}~\mathsf d x \\ & = \mathsf e^{-z}\mathbf 1_{0\leqslant z}\cdot\int_0^z\mathsf d x \\ & = z\,\mathsf e^{-z}\mathbf 1_{z\leqslant 0} \end{split}$$

In short: always be careful about the support.

  • $\begingroup$ Thank your for your help! $\endgroup$ – Eduard6421 Dec 31 '17 at 14:19

First, the support of the exponential distribution is $x>0$ so you need to account for when $z-x$ goes negative and the second factor goes to zero by only integrating from $0$ to $z$ rather than from $0$ to $\infty.$

Second, you made a sign error and the remaining exponential factor should be $e^{-z}.$ So you should get $ze^{-z}$ for the final answer.

  • $\begingroup$ I corrected the mistake of $e^{-z}$, but i still do not see how to get $ze^{-z}$ as final answer? $\endgroup$ – Eduard6421 Dec 30 '17 at 23:51
  • $\begingroup$ That's what my first paragraph was about. You integrate from $x=0$ to $z.$ You're integrating over $x,$ not $z$ so the $e^{-z} $ just comes out of the integral, if that's what is confusing you. $\endgroup$ – spaceisdarkgreen Dec 31 '17 at 1:16
  • $\begingroup$ Thank you for your help! $\endgroup$ – Eduard6421 Dec 31 '17 at 14:19

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.