1
$\begingroup$

I have a few questions regarding this problem that I do not quite understand.

For a random variable U distributed uniformly on the interval [0,1] its CDF is given by

$$\ F_U(u) = \begin{Bmatrix} 0, & u < 0 \\ u, & 0 \le u \le 1, \\ 1, & u > 1 \\ \end{Bmatrix} $$

  1. I don't understand why for values of u > 1 the CDF of U is 1. I would think that $\ F_U $ would be undefined outside of the interval that the random variable U is defined on.

====The solution to the problem in post title and explanation of my lack of understanding====

Let Z = max{X,Y}. For z$\ \in $ (0,1)

$\ F_Z(z) = P(Z \le z ) $

= P(X$\le$z and Y$\le$z) = $\ F_X(z)F_Y(z)$ = $\ z^2 $

  1. I understand that the CDF of a u.r.v. is $\ \frac{x-a}{b-a}$ for X on [a,b] so that $\ F_X(z)$ and $\ F_Y(z) $ are equal to z, but I do not understand why P(X$\le$z and Y$\le$z) = $\ F_Z$
$\endgroup$
  • $\begingroup$ $X$ and $Y$ has to be smaller or equal to $z$ if $max(x,y)=z$. Try out some values for $x$ and $y$. // $x,y$ cannot be greater than $z$. $\endgroup$ – callculus Mar 24 '17 at 23:32
  • $\begingroup$ The PDF of $Z=\max(X,Y)$ is the derivative of the CDF, which is asked here. An answer is given here. $\endgroup$ – Harry49 Sep 14 '17 at 15:25
2
$\begingroup$

I don't understand why for values of $u > 1$ the CDF of $U$ is $1$. I would think that $F_U$ would be undefined outside of the interval that the random variable $U$ is defined on.

The probability density function is defined for values above the support's supremum; it equals zero there.   (Likewise for values below the infimum.)

$U$ has a support of $(0;1)$, so $\mathsf P(U\leqslant 1)=1$ by the Law of total probability.   A random variable is certainly not-more than the supremum of its support.   Likewise it will also certainly not be greater than any higher value.   $\therefore~\forall u\geqslant 1 : \mathsf P(U\leqslant u)= 1~$.

I do not understand why $P(X\leqslant z\text{ and }Y\leqslant z) = F_Z(z)$

Because $F_Z(z)=\mathsf P(Z\leqslant z)$, by definition, and $Z=\max\{X,Y\}$.

Now when $Z$ is not greater than $z$, neither $X$ nor $Y$ is greater than $z$, and vice versa.   Thus the event $\{Z\leq z\}$ is the join of the events $\{X\leqslant z\}$ and $\{Y\leqslant z\}$.

$$\begin{align}F_Z(z) &= \mathsf P(Z\leqslant z) \\ & =\mathsf P(\min\{X,Y\}\leqslant z) \\ &= \mathsf P(X\leqslant z~\cap~ Y\leqslant z) \\ &= \mathsf P(X\leqslant z)~\mathsf P(Y\leqslant z) \\ &= F_X(z)~F_Y(z)\end{align}$$

$\endgroup$

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.