3
$\begingroup$

Let $x_1, \ x_2$ be independent random variables.

The probability density function $f(x_i)= 2x \ \ \ \ for \ \ \ \ 0<x_i < 1 $ and $f(x_i)=0 \ \ \ \text{otherwise}$

Then, the density function $f(x_1,x_2)=4x_1x_2$ for $0< x_1, x_2 < 1$

And Y is defined as $Y=X_1+X_2$

What is the distribution function $G(Y)$? Solve by using distribution function method(graph method)


What I did is as follows:

$G(Y)=P(Y \le y)= P( X_1+X_2 \le y)=\int \int f(x_1,x_2) dx_2 dx_1$

And I draw the plot of $Y=X_1+X_2$. But I'm stacking here. Please show me.

$\endgroup$
0
$\begingroup$

Comment: Notice that $X_i \sim Beta(2,1).$ The following R code simulates 100,000 realizations of $Y$, and makes their 'empirical distribution function'. The CDF of $Y$ obtained from the Answers by @BCLC or @probablyme should very closely match the ECDF.

 m = 10^5;  x1 = rbeta(m, 2, 1);  x2 = rbeta(m, 2, 1);  y = x1 + x2
 plot.ecdf(y, pch=20)

enter image description here

$\endgroup$
1
$\begingroup$

Wolfram Alpha gives us bounds. Try changing $y=1$ with different numbers here.

So we have

$$F_Y(y) = P(X_1 + X_2 \le y) = \int_0^{y} \int_0^{y - x_1} 4x_1x_2 dx_2dx_1$$

$\endgroup$
1
$\begingroup$

Keep going, you just need the bounds. Draw a picture for yourself to the see regions of interest. Then, if you want to use a double integral, then when $0\leq y\leq 1$, \begin{align*} P(X_1+X_2\leq y)&= P(X_2 \leq -X_1 +y)\\ &= \int_0^y \int_0^{-x_2+y} f(x_1,x_2)\,dx_1dx_2 \end{align*}

When $1<y\leq 2$, it is easier to do \begin{align*} P(X_1+X_2\leq y)&= 1 - P(X_2 > -X_1 +y)\\ &= 1 - \int_y^2 \int_{-x_2+y}^2 f(x_1,x_2)\,dx_1dx_2 \end{align*}

$\endgroup$

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.