How do I prove the sum of the total number of permutation inversions

805 Views Asked by At

If $p = p_1 . . . p_n$ is a permutation and an inversion in $p$ is a pair $1 ≤ i < j ≤ n$ such that $j$ appears to the left of $i$ in $p$, how do I prove that $\sum_{p∈S_n}x^{inv(p)} = (1 + x)(1 + x + x^2)· · ·(1 + x + · · · + x^{n−1})$, where $inv(p)$ is the total number of inversions in $p$?

What I got is for each pair $i < j$ the probability of inversion is 1/2 and as there are n(n−1)/2 such pairs, the expected number of inversions is $n(n-1)/4$, but this isn't a statistics problem and doesn't help me in proving the sum, so I'm not sure what to do here

1

There are 1 best solutions below

0
On

This boils down to induction and picking the last element in the permutation. Suppose this is true for $S_n$, where the series is $P_n(x) $. In $S_{n+1}$, the permutations where $\sigma(n+1)=n+1$ are represented by $P_n(x) $. If $\sigma(n+1)=n$, the first $n$ elements have inversions distributed like $P_n(x) $, but we have one additional inversion, giving us $xP_n(x) $.

In general, if $\sigma(n+1)=k$, the inversions are distributed like $x^{n+1-k}P_n(x)$.