2
$\begingroup$

Given a set $S$ containing all possible permutations of a vector $v = (1, 2, 3, ..., n-1, n)$, find the size of the set $P$, where $P$ is defined as the set of possible component-wise sums obtained by adding $k$ elements of $S$ together. Each element in $P$ is a sum of $k$ vectors from $S$, i.e. $P=\sum_{i=1}^kS$. I'm wondering if there is a nice looking formula for $|P|$ as a function of $n$ and $k$.

Edit: the main reason why I'm considering this is the following. Imagine $n$ players play a game. Each round, they get ranked (without ties). The best player gets $n$ points, the second best $n-1$, etc... and the last one only gets $1$. These points gets added to the scoreboard, which starts at $0$ for each player. I want to know how many possible scoreboards there are after $k$ rounds.

$\endgroup$
8
  • $\begingroup$ I was pretty tired, wanted to say component-wise $\endgroup$
    – Talesseed
    Jun 16 at 8:58
  • 2
    $\begingroup$ I can't find a table in OEIS, but for $k=2$ there's A175176 with the suggestion that $n=11, k=2$ is unknown. $\endgroup$ Jun 16 at 10:32
  • 3
    $\begingroup$ As far as I know, counting/describing vectors representable as the sum of two permutations is an open problem, which I have heard Alex Postnikov discuss. You can see him discuss it starting around the 57 minute of youtube.com/watch?v=thn16ncB4Js. $\endgroup$ Jun 16 at 19:15
  • $\begingroup$ Thanks a lot for the references! $\endgroup$
    – Talesseed
    Jun 17 at 15:00
  • 1
    $\begingroup$ @Talesseed: It's a brute force although a bit optimized. Eg. it starts with computing sequences corresponding to oeis.org/A019589 and then sums up their rearrangements. I don't think it's feasible going above $n=14$ this way. $\endgroup$ Jun 19 at 17:17

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.