Just need some hints on finding the distribution of $Z =\frac{min(X,Y)}{max(X,Y)}$

Where X and Y are iid ~ Unif(0,1).

$P(Z \gt z) = P(\frac{min(X,Y)}{max(X,Y)} \gt z) = P(min(X,Y) \gt z*max(X,Y))$

$= P(X \gt Y, min(X,Y) \gt z*max(X,Y)) + P(Y \gt X, min(X,Y) \gt z*max(X,Y))$

$= P(X \gt Y \gt z*X) + P(Y \gt X \gt z*Y)$

$= \int^x_{zx} 1 dy + \int^y_{zy} 1 dx$

$= x - zx + y - zy = x(1-z) + y(1-z) = (1-z)(x+y)$

Not sure if this is correct, especially the integral part. If anyone could help me out it would be appreciated. Thanks


$X$ and $Y$ are i.i.d. $U(0,1)$ and $Z = \frac{\min\{X,Y\}}{\max\{X,Y\}}$. The support of $Z$ is $[0,1]$

$Pr\left(Z=\frac{\min\{X,Y\}}{\max\{X,Y\}} > z\right), z \in [0,1]$

$= Pr(\{Y>zX\} \cap \{X>Y\}) + Pr(\{X>zY \cap \{X\le Y\})$

$= Pr(X>Y>zX) + Pr(Y\ge X>zY)$

Be noted that you are calculating $Pr(X>Y>zX)$ not $Pr(x>Y>zx)$, you have to add up the probability for all possible $x$. For $z \in [0,1]$,


$=\int_0^1 \int_{zx}^x dy dx = \int_0^1 (x-zx) dx$ (Drawing a graph of $y$ against $x$ will help you a lot)

$=(1-z)\int_0^1 xdx$


and similarly,

$Pr(Y\ge X>zY) = \int_0^1 \int_{zy}^y dx dy = \frac{1}{2}(1-z)$


$Pr(Z>z) = \left\{\begin{array}{l l}1 & z<0 \\ 1-z & 0 \le z \le 1 \\ 0 & z>1\end{array}\right.$

| cite | improve this answer | |

I know this has already been marked answered, but there's a much nicer way to do this! You don't need to do any integrals at all.

First, this problem is symmetrical in $X$ and $Y$, so we'll try to use that to simplify it a bit.

In fact, we can use the symmetry to show that the distribution of $Z$ does not change if we condition on $X > Y$. This is because $$P(Z < z) = P(Z < z \mid X < Y) P(X < Y) + P(Z < z \mid X > Y) P(X > Y) = \frac12 \left(P(Z < z \mid X < Y) + P(Z < z \mid X > Y)\right),$$ and $P(Z < z | X < Y) = P(Z < z | X > Y)$ by symmetry of $X$ and $Y$, so the above simplifies to $$P(Z < z) = P(Z < z \mid X > Y).$$

But under these conditions, $max(X, Y) = X$ and $min(X, Y) = Y$, so we have simply $Z = \frac{Y}{X}$. The easiest next step is probably to find the distribution of $Z$ conditional on what $X$ is, so that we can get rid of that nasty random division.

So now we'll condition on $X = a$ (in addition to $X > Y$). Under this condition, $Y \sim U(0, a)$, and so $Z = Y / X = Y/ a \sim U(0, 1)$. But this is the same for every $a$! So it doesn't matter that we conditioned on $a$, and we have $(Z | X = a, X > Y) = (Z | X > Y) = Z$ and hence $Z \sim U(0, 1)$ as well.

| cite | improve this answer | |

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.