Given a finite group if the automorphism group is known is it possible to write down all the automorphisms with respective orders? For example say the group $Z_{p^{2}}$ has automorphism group isomorphic to $Z_{p(p-1)}$ and I have to find an automorphism of order $p$, $p-1$. Now I know that one automorphism of order $p$ is $y \mapsto y^{p+1}$ . But how to actually find them out for this other groups?
-
Can you generalize the automorphism of $\mathbb{Z}/p$ of order $p-1$? – Michael Burr Jul 19 '15 at 21:13
-
$Aut(Z_{p})$ is a cyclic group of order $p-1$ . So there are $\phi(p-1)$ such automorphisms . That much I can see but genaralization? – user118494 Jul 19 '15 at 21:33
-
Now, what is $p\mathbb{Z}/p^2$ isomorphic to? – Michael Burr Jul 19 '15 at 21:40
-
$pZ_{p}$ as the elements are { $p,2p,3p,......(p-1)p$} ? – user118494 Jul 19 '15 at 21:46
-
And $0$. (So, how many elements are there?) – Michael Burr Jul 19 '15 at 23:23
-
There are $p$ elements. – user118494 Jul 20 '15 at 05:29
-
But what next? @MichaelBurr A little more help to lead to the answer, please? – user118494 Jul 22 '15 at 12:48
1 Answers
Below is some background on the automorphism group of any cyclic group $\mathbb{Z}/n\mathbb{Z}$, then we give a general way to find such an element of order $p$, and leave the case of powers $p-1$ for the time being.
First, note that an automorphism $\psi:\mathbb{Z}/n\mathbb{Z}\rightarrow \mathbb{Z}/n\mathbb{Z}$ is completely determined by the image of $1$. So $\psi(1)$ must be an element relatively prime to $n$, otherwise the order collapses. That is, if $\psi(1)=d$ where $\gcd(d,n)>1$, then $ord(d)<n$.
So $\psi(1)$ is relatively prime to $n$ and there are exactly $\phi(n)$ choices for the image of $1$, each of which determine a unique automorphism of the group $\mathbb{Z}/n\mathbb{Z}$. This gives $|Aut(\mathbb{Z}/n\mathbb{Z})|=\phi(n)$ and every element of the group is a map $1\mapsto r$ where $\gcd(r,n)=1$.
Let $\psi_r$ denote the map $1\mapsto r$ in $Aut(\mathbb{Z}/n\mathbb{Z})$. To calculate $ord(\psi_r)$ note $\psi_r(1)=r$, $\psi_r(\psi_r(1))=r^2$, and so forth. In this way we can define an isomorphism $Aut(\mathbb{Z}/n\mathbb{Z})\rightarrow (\mathbb{Z}/n\mathbb{Z})^\times$ by sending $\psi_r\mapsto r$.
So the problem at hand is equivalent to finding the least power $m$ such that $r^m\equiv 1 \pmod{n}$ and to find an element $r$ such that $r^m\equiv 1\pmod{n}$ with $m$ fixed.
Let's work in $\mathbb{Z}/p^2\mathbb{Z}$ for odd primes $p$ now so that the elements of $Aut(\mathbb{Z}/p^2\mathbb{Z})$ are $\{1,...,p-1,p+1,...,2p-1,2p+1,...,p^2-1\}$. We want to choose an element $r$ of the above such that when raised to the power of $p$ we get $1$.
Pick a general element $(mp+k)$ where $k\in \{1,...,p-1\}$ and $m\in \{0,...p-1\}$. By the binomial theorem we have
$$(mp+k)^p=\sum_{i=0}^p {p\choose i}(mp)^{p-i}(k)^i$$
Taking congruences modulo $p^2$ gives the last two terms
$$(mp+k)^p\equiv {p\choose p-1}(mp)k^{p-1}+k^p\equiv k^p \pmod{p^2}$$
Where the last congruence is found after calculating ${p\choose p-1}=p$. So we can check all powers $k^p\equiv 1 \pmod{p^2}$ where $k\in \{1,...,p-1\}$ and we get all others by varying $m\in \{0,...,p-1\}$.
To find at least one solution observe $k=1$ satisfies $k^p\equiv 1 \pmod{p^2}$ so that we can take any element of the set $\{1, p+1, 2p+1,..., p^2-p+1\}$.
Now for the case to the power of $p-1$ we can follow the above so that the binomial theorem gives
$$(mp+k)^{p-1}=\sum_{i=0}^{p-1}{p-1\choose i} (mp)^{p-1-i}k^i$$
Taking congruences again
$$(mp+k)^{p-1}\equiv {p-1\choose p-2}(mp)k^{p-2}+k^{p-1}=(p-1)(mp)k^{p-2}+k^{p-1}\equiv k^{p-1}-mpk^{p-2}\pmod{p^2}$$
So we want to simplify $k^{p-1}-mpk^{p-2}=(k-pm)k^{p-2}\equiv 1\pmod{p^2}$. I'm currently stuck so I'll think about this last one for a little bit.
- 5,809
-
I'm pretty sure in general it is difficult to find elements of particular orders unless you know the group structure itself and have orders in mind. It's possible to just manually check the orders of elements of $\mathbb{Z}/n\mathbb{Z}^\times$ by hand but it's not necessarily easy to find elements of particular orders. – Eoin Jul 24 '15 at 16:34
-
-
I know that's not easy .Actually I am learning semi direct products of groups and automorphisms are necessary for that , of course the structures of the groups being specified. Examples are shown in books but I cannot understand how they found that automorphism or on what basis that particular one is chosen. – user118494 Jul 27 '15 at 16:20
-
@user118494 That sounds interesting. I would venture "inspired guess" but it's probably just the use of a lot deeper theory. I don't think I know enough group theory to properly respond. :D – Eoin Jul 29 '15 at 04:40