a.) Let U, V be uniformly distributed over the set $\{(u,v): $$0<u<v<1$}.

Let $X$ = $-$$log(U)$, $Y$ = $-$$log(V)$, $Z$ = $max$($X$,$Y$).

a.) Draw the support of the joint distribution ($U$, $V$) and the joint pdf $U$ and $V$.

Here, I use the identity that the joint pdf equals the conditional pdf multiplied by the marginal pdf.

$f_{u,v}$($u$,$v$) = $f_{u|v}$($u|v$)$*f_v(v)$ = $\frac{1}{1-u}$*$I(0<u<v)$, where $I$ is the identity function.

The $\frac{1}{1-u}$ part comes from the PDF of the uniform distribution of $V$ over $(u, 1)$: $f_v(v) = \frac{1}{1-u}$

b.) Find the joint PDF of (X,Y). What is its support?

Really not sure on this one. What are my first steps? I'm guessing they come from part a.) but I think my work for part a.) is wrong.

c.) Find the conditional expectation $E$($Z$|$Y$).

My work so far:

$E$($Z$|$Y$) = $E$($max(X,Y)$|$Y$) = $max(E(X|Y),Y)$, and

$E(X|Y) = \int_0^v f_{X|Y}(x|y)*xdx = \int_0^v \frac{1}{(vu)^2}*xdx = \frac{v^2}{(vu)^2} = \frac{1}{v^2}$.

I calculated $f_{X|Y}(x|y)$ in part b.) (not shown) using the derivative formula for deriving PDFs but I'm almost entirely sure it's wrong.


In part a, it says uniformly distributed over the set $\{(u,v):0<u<v<1\}$, which means the region between lines $u=0, v=u,v=1$ ($v$ is in y-axis and $u$ is in x-axis). So, he joint PDF is 1/Area of this region.

In part b, you can apply Jacobian technique. Another method is to calculate $F_{XY}(x,y)$ and differentiate wrt $u$ and $v$. It's a good exercise, but I'd highly advise the former. Also, the region of support is very important. Hint: it'll extend to infinity but be under y=x line.

In part c, it seems $\max(X,Y)=X$ because $U<V\rightarrow Y<X$.

  • $\begingroup$ I'm still having issues on part a.). Drawing the support, I get that the joint PDF is 1/(2*v), because the area between u = 0, v = u and v = 1 is a triangle with height 1 and length v. But this leads to a joint PDF that does not integrate to 1, which violates the definition of the PDF. Do you have any other insights for me? $\endgroup$
    – Bo Jack
    Jun 23 '20 at 21:06
  • $\begingroup$ No, the area is $1/2$. $\endgroup$
    – gunes
    Jun 23 '20 at 21:42
  • $\begingroup$ Thank you. Then I get that the PDF of (U, V) = 2 (a valid PDF) and with the Jacobian technique, I get that the PDF of (X, Y) = 2e^{-x-y} (also a valid PDF). Then, I try to get E[Z|Y] = E[X|Y] by getting the conditional PDF of (X | Y), which I calculate to be -e^{y-x} / {-e^y + 1}. This doesn't seem right to me because my expected value depends on Y even though its conditional on Y. Im not sure where I went wrong here, so if I could get one last piece of advice, that'd be great. $\endgroup$
    – Bo Jack
    Jun 24 '20 at 18:43
  • $\begingroup$ I'm not sure about your algebraical calculations but that is normal, E[X|Y] is a function of $Y$. $\endgroup$
    – gunes
    Jun 24 '20 at 19:09
  • $\begingroup$ Sounds great, thank you! $\endgroup$
    – Bo Jack
    Jun 24 '20 at 19:17

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.