Let $\mathbb{Z}/n\mathbb{Z}$, $\left(\mathbb{Z}/n\mathbb{Z}\right)^{\times}$, and $\mathcal{D}_n$ be the ring of integers modulo $n$, its multiplicative group of order $\varphi(n)$, and the set of divisors of $n$, respectively. The map $\phi: \left(\mathbb{Z}/n\mathbb{Z}\right)^{\times}\times\mathbb{Z}/n\mathbb{Z}\rightarrow\mathbb{Z}/n\mathbb{Z}$, given by $\phi(g,x):=gx$ is a left group action on $\mathbb{Z}/n\mathbb{Z}$. I am having a hard time in showing that $\phi$ decomposes $\mathbb{Z}/n\mathbb{Z}$ into $|\mathcal{D}_n|$-many distinct orbits given by $\left(\mathbb{Z}/n\mathbb{Z}\right)^{\times}\cdot\bar{\delta}$, where $\delta$ is a divisor of $n$. Let me give an example. If $n=12$, then $\mathbb{Z}/12\mathbb{Z}=\{\bar{0},\dots,\bar{11}\}$, $\left(\mathbb{Z}/12\mathbb{Z}\right)^{\times}=\{\bar{1}, \bar{5},\bar{7}, \bar{11}\}$, and $\mathcal{D}_{12}=\{1,2,3,4,6,12\}$ and we have: $$\left(\mathbb{Z}/12\mathbb{Z}\right)^{\times}\cdot\bar{1}=\{\bar{1}, \bar{5},\bar{7}, \bar{11}\},$$ $$\left(\mathbb{Z}/12\mathbb{Z}\right)^{\times}\cdot\bar{2}=\{\bar{2}, \bar{10}\},$$ $$\left(\mathbb{Z}/12\mathbb{Z}\right)^{\times}\cdot\bar{3}=\{\bar{3}, \bar{9}\},$$ $$\left(\mathbb{Z}/12\mathbb{Z}\right)^{\times}\cdot\bar{4}=\{\bar{4}, \bar{8}\},$$ $$\left(\mathbb{Z}/12\mathbb{Z}\right)^{\times}\cdot\bar{6}=\{\bar{6}\},$$ $$\left(\mathbb{Z}/12\mathbb{Z}\right)^{\times}\cdot\bar{0}=\{\bar{0}\},$$ and $$\bigcup_{\delta|12}\left(\mathbb{Z}/12\mathbb{Z}\right)^{\times}\cdot\bar{\delta}=\mathbb{Z}/12\mathbb{Z}.$$ I would be very grateful for any help or insights.
- 282,220
- 86
-
crossposted : https://mathoverflow.net/questions/395022/orbits-of-mathbbz-n-mathbbz-under-the-action-of-its-multiplicative-group – verret Jun 10 '21 at 18:36
-
"... its multiplicative group of order $n$..." The multiplicative group is of order $\varphi(n)$ (Euler's phi), not $n$. – Arturo Magidin Jun 10 '21 at 18:54
-
@ArturoMagidin Yes, of course. Thank you. – L. Cardoso Jun 10 '21 at 18:57
1 Answers
Let $k\ge l$ be two positive integers and let $p$ be a prime number, then all integers $a\in\{1,\ldots,p^l-1\}$ coprime to $p^l$ are clearly coprime to $p^k$ as well, so the homomorphism $$(\mathbb{Z}/p^k\mathbb{Z})^\times\to\left(\mathbb{Z}/p^l\mathbb{Z}\right)^\times$$ is surjective. Note that for all pairs of coprime positive integers $m_1,m_2$, we have $$\left(\mathbb{Z}/m_1m_2\mathbb{Z}\right)^\times\cong \left(\mathbb{Z}/m_1\mathbb{Z}\right)^\times\times\left(\mathbb{Z}/m_2\mathbb{Z}\right)^\times.$$ It follows that for all positive integers $n$ and $d\mid n$, the natural map $$\left(\mathbb{Z}/n\mathbb{Z}\right)^\times \to \left(\mathbb{Z}/d\mathbb{Z}\right)^\times$$ is surjective. In particular, the kernel has cardinality $\varphi(n)/\varphi(d)$.
Now let $n$ and $d\mid n$ be positive integers and let $x\in \mathbb{Z}/n\mathbb{Z}$ of order $d$. Consider the action of $\left(\mathbb{Z}/n\mathbb{Z}\right)^\times$ on $\mathbb{Z}/n\mathbb{Z}$ and in particular the stabilizer of $x$. This stabilizer may be identified with all elements $g\in \{1,\ldots, n-1\}$ coprime to $n$ such that $gx=x$, which may be rewritten as $n\mid (g-1)x$, or $d\mid g-1$. These $g$ make up precisely the kernel of the aforementioned homomorphism $\left(\mathbb{Z}/n\mathbb{Z}\right)^\times\to\left(\mathbb{Z}/d\mathbb{Z}\right)^\times$.
We conclude that the cardinality of the stabilizer is $\varphi(n)/\varphi(d)$, and with the orbit-stabilizer theorem it follows that the orbit of $x$ consists of precisely $$\frac{\varphi(n)}{\varphi(n)/\varphi(d)}=\varphi(d)$$ elements. Finally, note that automorphisms preserve order, so the orbit of $x$ can only contain elements of order $d$, and there exist precisely $\varphi(d)$ elements of order $d$ in $\mathbb{Z}/n\mathbb{Z}$. We are done.
- 8,789
-
The relationship between $x$ and $d$ seems unclear to me; did you mean to choose $d$ to be the smallest integer with $dx \equiv 0 (mod n)$ (which is not hard to prove must be a divisor of $n$)? – user44191 Jun 10 '21 at 20:55
-
@user44191 Yes, $d$ is the order of $x$ in $\mathbb{Z}/n\mathbb{Z}$. Thanks for pointing that out, I've edited my post. – Mastrem Jun 10 '21 at 21:01
-
@Mastrem First of all, thank you for your contribution. I have a question though, for the natural homomorphism from $\left(\mathbb{Z}/n\mathbb{Z}\right)^{\times}$ to $\left(\mathbb{Z}/d\mathbb{Z}\right)^{\times}$, are you assuming $n$ and $d|n$ to be coprime? – L. Cardoso Jun 11 '21 at 17:19
-
@L.Cardoso No, I am not. If that were the case, the proof of surjectivity would be much easier, since then $(\mathbb{Z}/n\mathbb{Z})^\times \cong \left(\mathbb{Z}/d\mathbb{Z}\right)^\times\times \left(\mathbb{Z}/(n/d)\mathbb{Z}\right)^\times$ – Mastrem Jun 11 '21 at 17:29
-
@Mastrem OK. Running the risk of a stupid question, what is the natural homomorphism you used then? – L. Cardoso Jun 11 '21 at 17:31
-
@L.Cardoso Reducing modulo $d$. If you have an integer coprime to $n$, then it is coprime to $d$ as well. So we can just send the class $(a\mod n)$ to $(a\mod d)$. – Mastrem Jun 11 '21 at 17:32
-
-
@Mastrem do you have an e-mail or a social media through which I could reach you? – L. Cardoso Jun 13 '21 at 00:59
-
@L.Cardoso I'm not really comfortable sharing personal contact information, but if you have any further questions I'm happy to take a look at them. – Mastrem Jun 13 '21 at 08:46
-
@Mastrem That's ok. I just wanted to give you due credit for your answer in the paper and I didn't know how to do it properly. I am sorry for the inconvenience and once again thanks for your contribution. – L. Cardoso Jun 13 '21 at 18:27
-
@Mastrem My paper is coming out soon and I have used a part of your idea in a proof. I think it is at least ethical to appropriately thank you in the acknowledgment section. In order to do that I need your author's name. If you don't want to be mentioned, that's ok, but in that case, in order to avoid plagiarizing issues, I ask you to please say so in a comment. Thank you! – L. Cardoso Oct 04 '21 at 22:40
-
@L.Cardoso Congratulations on your paper! If you want to acknowledge me, you can refer to me as 'Math.Stackexchange user Mastrem' or someting like that, and if it is appropriate (I don't know how much of the ideas in this answer you used), add a reference to this post. – Mastrem Oct 05 '21 at 07:25
-