Expected value of random ordered pairs

1.1k Views Asked by At

Question

Suppose we receive numbers $1, 2, 3, 4, 5, 6$ in a random order. Define an inversion as an ordered pair of numbers $(i, j)$ such that $i < j$ but $j$ comes before $i$ in the random order.

For example, if we receive the numbers $3, 5, 1, 4, 2, 6$, then the inversions would be: $(1,3), (2,5), (2,3), (2,4), (2,5), (4,5)$.

(a) How many different ways can we order the numbers $1 $ to $6$?

(b) Fix two numbers, $i$ and $j$. What is the probability that $i$ comes before $j$?

(c) What is the expected number of inversions in a random order?

I am having particular trouble with (c), but I've included how I approached the other components as well.

My solutions:

(a) Given that order matters here, the number of ways to order 1 through 6 is simply $6!$.

(b) I actually wrote out six boxes, $1...6$ and wrote out the ways that $i$ could come before $j$. For example, there are $5$ boxes that $j$ could fill when we have the following: $i,O,O,O,O,O$, where $O$ represents an empty spot that $j$ may occupy. Doing so I find 15 ways that $i$ can come before $j$. There are an equal number of ways that $j$ can come before $i$. So the probability is $\frac{15}{30} = \frac{1}{2}$.

(c) My intuition here is to find the expected value, $E[X_k]$, that any given $i$ is less than $j$, but such that $j$ comes before $i$. Given the linearity of expectations, I could then extrapolate for all six integers, giving us $6*E[X_k]$, or written more formally, $\sum_{k=1}^6 E[X_k]$.

In part (b) I found that the probability that $i$ comes before $j$ is $\frac{1}{2}$, which would also equal the probability that $j$ comes before $i$. This probability would need to be multiplied by the number of integers that come before a number $i$, but that number changes depending on where $i$ sits in the list of $6$. I'm not exactly sure where to go from here (I also attempted this problem with an indicator function, again to no avail).

1

There are 1 best solutions below

6
On BEST ANSWER

You need to look at pairs of numbers, not individual numbers. For each pair $\langle i,j\rangle$ with $1\le i<j\le 6$ let $X_{i,j}$ be the random variable that is $1$ if $j$ comes before $i$ and $0$ otherwise. From (b) you know that $\Bbb E[X_{i,j}]=\frac12$. The random variable that you want is $X=\sum_{1\le i<j\le 6}X_{i,j}$, and by linearity of expectation we have

$$\Bbb E[X]=\sum_{1\le i<j\le 6}\Bbb E[X_{i,j}]=\sum_{1\le i<j\le 6}\frac12\;,$$

which is ... ?