Consider permutations $\pi$ of the set $\{1,\dots,n\}$ having the symmetry property $\pi \pi^* \pi = \pi^*$, where $\pi^*$ is the "reflection" $k \mapsto n+1-k$. Are there references or other information about such a $\pi$?

  • You really do mean $\pi\pi^*\pi$, not $\pi\pi^*\pi^{-1}$? – LSpice Sep 12 at 19:09
  • 2
    Also, is $\pi^*$ meant to depend on $\pi$, say as $k \mapsto n + 1 - \pi(k)$, or is it really a fixed permutation independent of $\pi$? If the latter, then you are just looking for permutations conjugate by a fixed involution to their inverses. – LSpice Sep 12 at 19:10
  • 4
    In other words, $\pi \pi^{*} $ is involution. It should reduce the questions about your permutations to the questions about involutions. – Fedor Petrov Sep 12 at 19:11
  • @FedorPetrov's re-phrasing is better than mine. – LSpice Sep 12 at 19:11
  • 1
    Why don't you use a notation without the letter $\pi$ to denote the reflection? – YCor Sep 15 at 16:59
up vote 1 down vote accepted

The problem boils down to finding permutations $\pi$ such that $\pi(n+1-x)=y\iff\pi(n+1-y)=x$. Let $X\subset\{1,...,n\}$ be such that $n-\lvert X\rvert$ is even and let $P$ be a partition of $\{1,...,n\}\backslash X$ into pairs and denote, for every element $k$ of $\{1,...,n\}\backslash X$, its companion by $P(k)$. Then, the permutation defined by $\pi(n+1-k)=P(k)$ for every $k\in\{1,...,n\}\backslash X $ and $\pi(n+1-x)=x$ for every $x\in X$ satisfies the required condition. It is clear that appropriate permutations are in a one-to-one correspondence with $(X,P)$ couples.

  • Isn't this just a complicated way of saying "$\pi$ is of the form $\sigma\pi^*$, where $\sigma$ is an involution" (which is what @FedorPetrov's comment's that you quote says)? I guess it depends on the poster's needs which form of the statement is more convenient …. – LSpice Sep 15 at 12:49
  • I thought a constructive/descriptive approach might be helpful depending on the OP's needs indeed. I just lightened my response a bit. – Nicéphore Bayekula Sep 15 at 16:05
  • By the way, as a TeX matter, \mid is only intended as a binary operator (like 2 \mid 6: $2 \mid 6$). For delimiters, you want \lvert and \rvert (like \lvert2\rvert: $\lvert2\rvert$). I made the edit here. – LSpice Sep 15 at 17:41

Your Answer


By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.