If $X \sim \text{U}(0,1)$, what is the distribution of $Y = \max(X,1/X)$? I know for this particular problem, $Y = \max(X,1/X) = 1/X$, whose distribution can be easily attained directly. However, I've been trying another approach by starting with this

$$P(Y \leq y) = P(X \leq y,\frac{1}{X} \leq y)=P(\frac{1}{X}\leq y| X \leq y)P(X \leq y),$$

and I've been stuck. I would be appreciate if someone could show me how it could be proceeded. Thank you very much.

  • 5
    $\begingroup$ Is $X$ really Uniform(0,1)? Because in that case, max($X, 1/X$) is always $1/X$. $\endgroup$ – Cliff AB Jun 25 '18 at 3:43
  • $\begingroup$ Yes, I've mentioned that for this particular case, max(X,1/X) would be 1/X, but I'm trying a more general way to deal with this kind of problem. $\endgroup$ – diidoobiib Jun 25 '18 at 3:44
  • 1
    $\begingroup$ If you tried $X \sim U(0,2)$, you could come up with a nice solution that might generalize (depending on what you want to generalize too). $\endgroup$ – Cliff AB Jun 25 '18 at 3:53
  • $\begingroup$ @CliffAB so supposing that X~U(0,2), how can I derive the distribution of max(X,1/X)? $\endgroup$ – diidoobiib Jun 25 '18 at 4:04

As you have acknowledged in your question, this problem is trivial in the case where $0 < X < 1$. In the more general case (taking $\tfrac{1}{0} = +\infty$ for definiteness) you have:

$$Y = \max(X, \tfrac{1}{X}) = \begin{cases} \tfrac{1}{X} & & \text{if }X < -1 \\[6pt] X & & \text{if }-1 \leqslant X < 0 \\[6pt] +\infty & & \text{if } X = 0 \\[6pt] \tfrac{1}{X} & & \text{if }0 < X < 1 \\[6pt] X & & \text{if }X \geqslant 1 \\[6pt] \end{cases}$$

If you restrict this to the case where $X>0$ (which includes the last two categories), you have:

$$\begin{equation} \begin{aligned} F_y(y) \equiv \mathbb{P}(Y \leqslant y) &= \mathbb{P}(\max (X, \tfrac{1}{X}) \leqslant y) \\[6pt] &= \mathbb{P}(X \geqslant \tfrac{1}{y}, 0 < X < 1) + \mathbb{P}(X \leqslant y, X \geqslant 1) \\[6pt] &= \mathbb{P}(\tfrac{1}{y} \leqslant X < 1) + \mathbb{P}(1 \leqslant X \leqslant y). \\[6pt] \end{aligned} \end{equation}$$

And of course, in the case where $0<y<1$ this probability is zero, as expected.

| 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.