Rigorous Order Statistics with Indicator Functions

219 Views Asked by At

If three people are randomly placed along a 1 mile road, the probability that no two of them are less than $m$ miles apart for $m \leq \frac{1}{2}$ could be solved by using the density for the order statistics: $f_{X_{(1)},X_{(2)},X_{(3)}}=3!$ for $0 < x_1 < x_2 < x_3 < 1$, and reasoning that:

$0 < x_1 < 1-2m$,

$x_1+m < x_2 < 1-m$

and $x_2+m < x_3 < 1$, and writing outL

$3!\int_0^{1-2m} \int_{x_1+d}^{1-m} \int_{x_2+m}^1 dx_3dx_2dx_1=(1-2m)^3$

However, I sometimes struggle with these problems so I'm trying a more general approach with indicator functions, but I'm having trouble:

First, I set my density to all of $\mathbb{R}^3$ $f_{X_{(1)},X_{(2)},X_{(3)}}=3!I_{0 < x_1 < 1}I_{0 < x_2 < 1}I_{0 < x_3 < 1}$

and then my desired probability is:

${\int \int \int}_{x_2>x_1+m,x_3>x_2+m} 3!I_{0 < x_1 < 1}I_{0 < x_2 < 1}I_{0 < x_3 < 1} dx_1dx_2dx_3$

But I'm not sure how to convert the set being integrated over into Indicator functions in a way that will let me proceed. For instance, if I try:

${\int \int \int} I_{x_2>x_1+m}I_{x_3>x_2+m}3!I_{0 < x_1 < 1}I_{0 < x_2 < 1}I_{0 < x_3 < 1} dx_1dx_2dx_3$, I wouldn't know where to go from here. Is what I'm trying to do possible...that is get the right integration limits for each integral, without having to "reason" what they should be?

1

There are 1 best solutions below

2
On BEST ANSWER

Starting with

$$\displaystyle {\int \int \int} I_{x_2>x_1+m}I_{x_3>x_2+m}3!I_{0 < x_1 < 1}I_{0 < x_2 < 1}I_{0 < x_3 < 1} dx_1dx_2dx_3$$

it s a simple reordering to get

$$\displaystyle 3! \int_{x_1} I_{0 < x_1 < 1} \left(\int_{x_2} I_{0 < x_2 < 1} I_{x_2>x_1+m}\left(\int_{x_3} I_{x_3>x_2+m}I_{0 < x_3 < 1} dx_3\right) dx_2\right) dx_1$$

with the inner bracket becoming $$\int_{x_3} I_{x_3>x_2+m}I_{0 < x_3 < 1} dx_3 = (1-(x_2+m))I_{x_2+m\le 1}$$ which is similar to the $\displaystyle \int_{x_2+m}^1 dx_3$ you calculated earlier.

Just continue in this way with the other two integrals and you will be repeating what you did previously.