3
$\begingroup$

Let $G$ be a finite group. Consider the set $X_G:=\cup_{H\le G} G/H$, where the disjoint union is taken over all left cosets of the subgroups $H \le G$. Let $S \subset G$ be a generating set for $G$ and $|g| :=$ word length with respect to $S$. For a left coset $gH$ define $|gH| := \min_{ h\in H} |gh|$. Define $\pi(G,t) = \sum_{x \in X_G} t^{|x|}$. Then $\pi(G,1) = |X_G|$. For $G =\mathbb{Z}/(n) \equiv C_n$ the cyclic group and $S=\{+1\}$, it seems that

$\pi(G,-1) = $ number of odd divisors of $n$ (OEIS: A001227). (1)

For $G =\mathbb{Z}/(n) \equiv C_n$ the cyclic group and $S=\{\pm1\}$, it seems that

$\pi(G,-1) = $ #(of divisors of $n$ of the form $4m+1$)$-$#(of divisors of $n$ of the form $4m+3$), (OEIS: A002654).

So this $\pi(G,t)$ is a polynomial which when inserting the $2$-th roots of unity, counts something. My conjecture is, that $(X_G, C_2, \pi(G,t))$ exhibits the cyclic sieving phenomenon, but for this I need a group action from the cyclic group $C_2$ to $X_G$ such that (1) is fullfilled:

$\pi(C_n,-1) = |X_{C_n}^{-1}| = $ number of odd divisors of $n$.

However, I tried $c*gH = g^{-1}H$ for $C_2 = \{1,c\}$ but this action, although natural, does not give the desired property, and I do not have a polynomial for this action to get a cyclic sieving phenomenon.

So my question is: 1) For the cyclic group $G:=C_n$, how does one define an action from $C_2$ to $X_{C_n}$ such that (1) is fullfilled? You do not have to prove that (1) is fullfilled, numerical coincidences should be ok for the first. 2) If there is such an action in 1), is it possible to define this for an arbitrary finite group $G$, hence $C_2$ acts on $X_G$?

Thanks for your help!

Edit: If (1) can be proven to be true, then applying Burnsides Lemma one would get:

$$a(n) = |X_{C_n}/C_2| = \frac{1}{2}(\sigma(n)+\tau^*(n))$$ where $\tau^*(n)$ counts the odd divisors of $n$ and $a(n) = $ number of orbits, which is conjectured to be equal to = OEIS, A079247, Number of pairs $(p,q), 0 \le p < q,$ such that $p+q$ divides $n$. Related: https://mathoverflow.net/questions/331347/a-question-concerning-some-arithmetic-function/

$\endgroup$
3
  • $\begingroup$ for (1) have you tried the action of $C_2$ on $X_G$ defined by $xH\mapsto x^{-1}H$? I haven't got time to check, but it is the obvious one to try $\endgroup$ Commented May 12, 2019 at 12:30
  • $\begingroup$ yes i have tried as mentioned in my question $\endgroup$
    – user276611
    Commented May 12, 2019 at 12:41
  • $\begingroup$ I don't think $xH\mapsto x^{-1}H$ is even well-defined on $G/H$ unless $H$ is a normal subgroup of $G$. $\endgroup$
    – anon
    Commented May 17, 2019 at 2:25

0

You must log in to answer this question.