Let $X_1$ and $X_2$ be two random variables uniformly distributed on $(0, 1)$. It is easy to calculate the distribution of minimum and maximum of these two numbers:

$$ P[\max(X_1, X_2)<x] = x^2 $$

$$ P[\min(X_1, X_2)<x] = 1 - P[\min(X_1, X_2)>x] \\ = 1 - (1-x)^2 = 2x-x^2 $$

But what would be the distribution of the minimum, given that the maximum is given and equal to some number $z$, i.e.

$$ P[\min(X_1, X_2)<x | \max(X_1, X_2)=z] = ? $$

  • $\begingroup$ why is the first result $2x - x^2$? I keep getting $x(1-x)+x(1-x) = 2x - 2x^2$ $\endgroup$ – Alex Jun 10 '14 at 15:20
  • 1
    $\begingroup$ @Alex, I added the derivation that I think is correct. In your case, didn't you miss the case that both of the random variables are smaller than $x$? $\endgroup$ – Grzenio Jun 10 '14 at 15:39
  • $\begingroup$ OK thanks I got it $\endgroup$ – Alex Jun 10 '14 at 15:47

Good question, this had me stumped at first. The minimum is actually the uniform distribution on [0,z].

Without loss of generality, let $X_1$ be at the maximum value z.

Then $P(X_2 > z) = 0$ since z is the max, and $P(X_2 \le z) = 1$.

Let Y be the event that $X_2 \le z$. Then by Bayes rule. $$ f_{X_2}(x|Y) = P(Y|X_2=x)f_{X_2}(x)/P(Y) \\ = 0 \quad for \quad x > z \\ = 1(1)/z \quad for \quad x \le z \\ $$ Therefore the minimum is uniform on [0,z]

| cite | improve this answer | |
  • $\begingroup$ Thanks for that! Does it hold in the more general case where we have $N$ random variables $X_1$, ..., $X_N$, and we condition the distribution of one of them (which is not maximal) on the maximum of all these random variables being $z$? $\endgroup$ – Grzenio Jun 12 '14 at 7:50
  • $\begingroup$ New problem $\implies$ New question. $\endgroup$ – Did Jul 14 '14 at 18:20

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.