Let $X_1 \sim Uniform(0,1)$, and $X_2 \sim Uniform(0, x_1)$, where $x_1$ is the realized value of $X_1$. Find $P(X_1 + X_2 \geq 1)$.

I know that I need the joint distribution of $X_1$ and $X_2$.

$f_1(x_1) = 1, \ 0 < x_1 < 1$

$f_{2|1}(x_2|x_1) = \frac{1}{x_1}, \ 0 < x_2 < x_1$

$f_{1,2}(x_1, x_2) = f_{2|1}(x_2|x_1)f_1(x_1) = \frac{1}{x_1}, \ 0 < x_2 < x_1 < 1$.

What are the limits of integration when integrating this joint PDF to get the desired probability?

  • $\begingroup$ What relation do $X_1$ and $X_2$ have? Independent? $\endgroup$ – user158565 Nov 30 '18 at 1:51
  • $\begingroup$ @user158565 They can't be independent; the form of dependence is given in the question. $\endgroup$ – Glen_b Nov 30 '18 at 5:01
  • $\begingroup$ @Glen_b Answer was modified. $\endgroup$ – user158565 Nov 30 '18 at 5:19

enter image description here

Here is some hint:

You already get the joint pdf, and the range of $X_1$, and $X_2$ as showed in graph black and red area.

The probability of the event = integral of pdf on the area that event defined. In your case, $X_1+X_2>1$ is the red area in the graph above. So $$\Pr(X_1+X_2>1) = \int_?^?\int_?^? \frac 1{x_1}dx_1dx_2$$ So you need find the limits of the integrals, then you get the answer.

  • $\begingroup$ Thanks, that's really helpful. I think that it's easier to integrate with respect to $x_2$ first, from $1- x_1$ to $x_1$. Then, I integrate with respect to $x_1$, from $0.5$ to $1$. $\endgroup$ – MSE Dec 2 '18 at 14:07
  • $\begingroup$ You are right. Graph is very helpful for this kind of questions. $\endgroup$ – user158565 Dec 2 '18 at 16:40

My advice with figuring out limits is draw a picture (by hand will generally suffice, though):

plot of X2 vs X1 with joint density indicated by dot density, X1 + X2 = 1 line marked in

You should be able to figure out the upper and lower limits from that. Consider which variable to integrate over first; one may be a tad easier than the other.

There's also a symmetry you can exploit which simplifies the slightly trickier of the two integrals, which may save a line.


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.