I'm given a joint pdf with the following bounds

f(x,y)= \begin{cases} x^2 & \text{ for } 0 \leq y \leq 1-x^2 \\ 0 & \text{ otherwise} \end{cases}

I am trying to find the Marginal pdf of y. I thought the integral would be from -1 to 1 since that's what it seems to be bound by on the graph, but that's wrong. What am I doing wrong?


Draw the downward-facing parabola $y=1-x^2$. To find the (marginal) density of the random variable $y$, we "integrate out" $x$.

But first we need to make a fix. The "density" is not a density, since the double integral of our "density" over the region below the parabola and above the $x$-axis is $\frac{4}{15}$. To make it a density we need to use $\frac{15}{4}x^2$, not $x^2$.

It is clear that the density is $0$ outside the interval $[0,1]$. Inside the interval, we go from $x=-\sqrt{1-y}$ (the left branch of the parabola) to $x=\sqrt{1-y}$ (the right branch). So inside the interval we want $$\int_{x=-\sqrt{1-y}}^{\sqrt{1-y}} \frac{15}{4} x^2\,dx.$$ Integrate. We get $\frac{5}{2}(1-y)^{3/2}$ on the interval $[0,1]$.


$$f_Y(y)=\int_{-\sqrt{1-y}}^{\sqrt{1-y}}x^2dx=\dfrac{2(1-y)^{3/2}}{3}$$ for $0<y<1$.

  • $\begingroup$ how did you get the limits of integration? $\endgroup$
    – nodel
    Jul 16 '15 at 3:28
  • $\begingroup$ @risdel from the support: $0< y < 1- x^2 < 1$. This means that $1>1-y>x^2>0$, which in turn means that ${-}\sqrt{1-y}<x<\sqrt{1-y}$ and $0<y<1$. $\endgroup$ Jul 16 '15 at 3:38
  • $\begingroup$ would it be safe to assume an absolute value relationship where x^2 ≤(1−y) => −√(1 − y) ≤ x ≤√(1−y) $\endgroup$
    – nodel
    Jul 16 '15 at 3:56
  • 1
    $\begingroup$ @risdel Yes. It is algebraically reasonable to do so. Plot $\,y= 1-x^2\,$. The area below this curve and above the x-axis is the support of the joint distribution. Draw a series of horizontal line segments across this area (particularly at $y=0.36$ and $y=0.64$). From this we can see from the plot we have $\,x\in\big({-}\sqrt{1-y\,}..{+}\sqrt{1-y\,}\big)\,$ for each $\,y\in\big(0..1\big)\,$. $\endgroup$ Jul 16 '15 at 4:20

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.