3
$\begingroup$

Problem:
The join pdf of $(X,Y)$ is given by
$ f(x,y) = \begin{cases} \Big( \frac{5}{32} \Big) x^2(4 - y) & \text{for } x < y < 2x, 0 < x < 2 \\ 0 & \text{otherwise} \\ \end{cases} $
where $k = \frac{5}{32}$. Find the marginal pdf of $Y$.
Answer: \begin{eqnarray*} f_y(y) &=& \int_{-\infty}^{\infty} f(x,y) \,\, dx \\ f_y(y) &=& \int_{0}^{2} \Big( \frac{5}{32} \Big) x^2(4 - y) \,\, dx \\ f_y(y) &=& \Big( \frac{5}{32} \Big) \frac{x^3}{3}(4 - y) \Big|_{x = 0}^{x = 2} \\ f_y(y) &=& \Big( \frac{5}{32} \Big) \frac{2^3}{3}(4 - y) \\ f_y(y) &=& \Big( \frac{5}{12} \Big) (4 - y) \\ \end{eqnarray*} My answer is: \begin{eqnarray*} f_y(y) &=& \begin{cases} \Big( \frac{5}{12} \Big)(4x^3 - \frac{3x^4}{2}) & \text{for } 0 < y < 4 \\ 0 & \text{otherwise} \\ \end{cases} \\ \end{eqnarray*} but the books answer is:
\begin{eqnarray*} f_y(y) &=& \begin{cases} \Big( \frac{5}{12} \Big) (4 - y) & \text{for } 0 < y < 2 \\ \Big( \frac{5}{32} \Big)\Big( \frac{1}{3} \Big)(4-y)(8 - \frac{y^3}{8}) & \text{for } 2 < y < 4 \\ 0 & \text{otherwise} \\ \end{cases} \\ \end{eqnarray*} I would like to know what I did wrong.
Thanks
Bob

$\endgroup$
  • 2
    $\begingroup$ You obeyed the bound $0<x< 2$ but not the bound $y/2<x<y$. $\endgroup$ – kimchi lover Sep 11 '17 at 23:12
1
$\begingroup$

The support for the joint distribution is $\{(x,y): 0<x<2, x<y<2x\}$

  • That is the triangle $\triangle(0,0)(2,2)(2,4)$

This is the same region as $\{(x,y): 0<y<4, y/2< x< \min\{2,y\}\}$

Which is the union: $\{(x,y): 0<y<2, y/2<x<y\}\cup\{(x,y): 2\leqslant y< 4, y/2<x<2\}$

  • That is $\triangle(0,0)(1,2)(2,2)\cup\triangle(1,2)(2,2)(2,4)$

Plot the points and you see how the triangle is divided into two parts at the $y=2$ horizon.   Draw horizonal likes through the upper and lower triangles and see why the bounds for the integral producing the $Y$-marginal pdf are different for each part.

Hence why:

$$f_Y(y) =\frac{5}{32} \begin{cases}\displaystyle \int_{y/2}^y x^2(4-y)\;\mathrm d\, x &:& 0<y<2 \\[1ex]\displaystyle \int_{y/2}^2 x^2(4-y)\;\mathrm d\, x &:& 2\leqslant y< 4 \\[1ex] 0 &:& \text{otherwise}\end{cases}$$

$\endgroup$
0
$\begingroup$

It seems like you have a good idea of what to do but just made a tiny mistake, I'll clear that up and leave you the work of finishing up.

The problem is that the set of possible values of $y$ depends on the value of $x$. You are not taking this into account. To do so you can write the joint pdf in terms of characteristic functions.

So, rewrite the pdf as follows:

$$ f(x,y) = \left( \frac{5}{32} \right) x^2 (4-y) \mathbf{1}_{ [x,2x]} (y) \cdot \mathbf{1}_{[0,2]}(x) $$

Now aply the same method you did before, integrating with respect to all possible values of $x$.

Take into account that if $ y$ must be between $x$ and $2x$, this is equivalent to $x$ being smaller than $y$ and $2x$ being bigger than $y$. (Draw the region for easier integration)

Hope this helps!

$\endgroup$
  • 1
    $\begingroup$ I now understand why my answer is wrong and my guess is that $f_y(y) = \int_{\frac{y}{2}}^{y} \Big( \frac{5}{32} \Big) x^2(4 - y) \,\, dx $ is the right integral. Is it? $\endgroup$ – Bob Sep 11 '17 at 23:54
  • 1
    $\begingroup$ @Bob not exactly. That one is not considering the bound on $x$. When you do integration on two variables it is a very good practice to draw the region (set of $(x,y)$ values). Whenever yo draw this region you'll find out exactly where the solution of the book comes from. It is just from integrating. $\endgroup$ – Joaquin San Sep 12 '17 at 5:21

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.