Let $I(n, k)$ be the number of $n$-permutations that have $k$ inversions.
Here is a post that explains how to find $I(n,2)$: Find number of permutations with 2 inversions
I am interested in $I(n,3)$.
There are three ways to end up with exactly three inversions.
- There must have been three steps which created one inversion each. Hence, we simply need to choose which three steps created the three single inversions which can be chosen in $\binom{n-1}{3}$ ways.
- There must have been a step which created three inversions. Hence, we simply need to choose which step created the three inversions which can be chosen in $\binom{n-3}{1}$ ways.
- There must have been a step which created one inversion and a step which created two inversions.
I am stuck on (3). At first glance, I thought we can simply choose a step to create one inversion and a step to create two inversions in $\binom{n-1}{1}\binom{n-2}{1}$ ways. But this over-counts and I am not fully sure of the reason why. Is it because after we pick a step to create a single inversion in $\binom{n-1}{1}$ ways we don't actually have $\binom{n-2}{1}$ ways to pick a step to create two inversions? We should have less steps, correct?
Can someone help me finish this last case?
For reference $I(4,3)=6$ and $I(5,3)=15$ I believe which we can use as sanity checks.
Also, I was just playing around with numbers and I found this formula: $I(n,3)=\binom{n-1}{3}+\binom{n-3}{1}+(n-2)^2$ which works satisfies $I(4,3)=6$ and $I(5,3)=15$ so it might be the answer. But I have no clue how to interpret $(n-2)^2$.
(Maybe first pick a step that creates two inversions in $\binom{n-2}{1}$ ways and then for each of those choices, I'm guessing, we have $\binom{n-2}{1}$ ways to pick a step that creates one inversion. Hence we get $(n-2)(n-2)$? But then shouldn't we also be able to pick first the step that creates one inversion in $\binom{n-1}{1}$ ways and then for each of those choices, we have $\binom{n-3}{1}$ ways to pick a step that creates two inversions. Hence $(n-1)(n-3)$?)