14

I'm looking for a simple proof that up to isomorphism every group of order $2p$ ($p$ prime greater than two) is either $\mathbb{Z}_{2p}$ or $D_{p}$ (the Dihedral group of order $2p$).

I should note that by simple I mean short and elegant and not necessarily elementary. So feel free to use tools like Sylow Theorems, Cauchy Theorem and similar stuff.

Thanks a lot!

Shaun
  • 44,997
Serpahimz
  • 3,781

9 Answers9

29

Since we are allowed to use Sylow, we can assume $G$ is generated by $x,y$ with $x^p=y^2=1$, where $\langle x \rangle \lhd G$, so $y^{-1}xy = x^t$ for some $t$ with $1 \le t \le p-1$. Then $x = y^{-2}xy^2 = x^{t^2}$, so $p$ divides $t^2-1 = (t-1)(t+1)$, hence $p$ divides $t-1$ or $t+1$ and the only possibilities are $t=1$ or $p-1$, giving the cyclic and dihedral groups.

Derek Holt
  • 90,008
  • That is indeed a nice and short proof, thanks :) – Serpahimz Jan 31 '13 at 14:45
  • More generally if $y^{-1}xy = x^t$ and $y^j = 1$, then $x = y^{-j}xy^j = x^{t^j}$, so $t^j \equiv 1 \mod{\operatorname{ord}(x)}$. This trick is sometimes useful for classifying finite groups of given order – Mikko Korhonen Jan 31 '13 at 15:03
  • 1
    Hi Derek, how is "$x = y^{-2}xy^2 = x^{t^2}$" true? – jstnchng Dec 07 '14 at 07:40
  • 6
    @jstn $x=e^{-1}xe=(y^2)^{-1}xy^2=y^{-1}(y^{-1}xy)y=y^{-1}x^t y=\underbrace{(y^{-1}xy)\cdot (y^{-1}xy)\cdots(y^{-1}xy)}_{\text{tea time}}=(x^t)^t=x^{t^2}$. – bfhaha Mar 21 '16 at 09:34
  • 1
    Hello Derek!! I am reading your answer and I have some questions... We have that $|G|=2p$, so there are $2$-Sylow and $p$-Sylow in $G$, right? $$P\in \text{Syl}_2(G) , \ |P|=2 \ Q\in \text{Syl}_p(G) , \ |Q|=p$$ Why do we have that $G$ is generated by $x,y$ ? – Mary Star Mar 24 '16 at 15:46
  • 2
    Because $x$ has order $p$ and $y$ has order $2$ so the subgroup they generate has order divisible by $G$, and so it must equal $G$. – Derek Holt Mar 24 '16 at 16:54
  • So, with my notations, we consider that $x\in Q$ and $y\in P$, right? @DerekHolt – Mary Star Mar 24 '16 at 16:56
  • 1
    Yes you can assume that. – Derek Holt Mar 24 '16 at 16:58
  • How exactly do we get that $x = y^{-2}xy^2 = x^{t^2}$ ? @DerekHolt – Mary Star Mar 24 '16 at 18:20
  • 1
    That is explained in the comment by bfhaha above. – Derek Holt Mar 24 '16 at 18:24
  • We have that $1\leq t\leq p-1$ and $p\mid t-1 \Rightarrow p\leq t-1 \Rightarrow p+1\leq t$. Therefore, $p+1\leq t\leq p-1$. But how do we conclude that $t=1$ ? @DerekHolt – Mary Star Mar 24 '16 at 19:21
  • 1
    $p|t-1$ does not imply $p \le t-1$ it implies $t-1=0$. – Derek Holt Mar 24 '16 at 19:25
  • When $t=1$ we have that $y^{-1}xy=x$. Why does it hold then that $G$ is isomorphic to $\mathbb{Z}_{2p}$ ? @DerekHolt – Mary Star Mar 24 '16 at 20:19
  • In that case we have that $xy=yx$. That means that $G$ is abelian, right? Is the only abelian group of order $2p$ the $\mathbb{Z}_{2p}$ ? @DerekHolt – Mary Star Mar 25 '16 at 00:13
  • 2
    Note that $xy$ has order $2p$ so the group has to be cyclic. – Derek Holt Mar 25 '16 at 07:04
  • So, we have that since the order of $xy$ is $2p$ and $|G|=2p$, the group $G$ is generated by $xy$, right? So, $G$ is cyclic. Does this only hold when $t=1$, so $xy=yx$ ? In the case when $t=p-1$, we have that $x^p=1=y^2$ and $y^{-1}xy=x^{p-1} \Rightarrow yxy=x^{-1}$. So, $x$ and $y$ generate the dihedral group $D_p$, right? @DerekHolt – Mary Star Mar 25 '16 at 09:48
  • 1
    Yes that's all correct! – Derek Holt Mar 25 '16 at 15:06
  • Isn't $G$ in both cases cyclic? Or only at the case $t=1$ ? I got stuck right now... Also, is $Z_{2p}$ abelian? Or how do we use the fact that in the case $t=1$ we have that $xy=yx$ ? @DerekHolt – Mary Star Mar 25 '16 at 22:21
  • I thought about it again... Do we use the fact that $xy=yx$ to conclude that the order of $xy$ is $2p$ ? @DerekHolt – Mary Star Mar 26 '16 at 10:11
  • 1
    Yes that's right. – Derek Holt Mar 26 '16 at 11:19
  • We have that $(xy)^{2p} \ \overset{xy=yx}{=} \ x^{2p}y^{2p}=(x^p)^2(y^2)^p=e$. But how do we know that $2p$ is the smallest power such that $(xy)^{2p}=e$ so that $2p$ is the order of $xy$ ? @DerekHolt – Mary Star Mar 26 '16 at 11:52
  • 1
    Please stop asking questions and try and solve the problem yourself. – Derek Holt Mar 26 '16 at 11:57
  • Suppose that $2p$ is not the smallest power, then say $n$ is the smallest power such that $(xy)^n=e$. That means that $$n\mid 2p \Rightarrow n\mid 2 \text{ or } n\mid p \Rightarrow n\in {1,2,p}$$ @DerekHolt – Mary Star Mar 26 '16 at 12:31
  • If $n=1$, then $xy=e \Rightarrow x=y^{-1}=y$. Since the order of $y$ is $2$, we would have that the order $x$ is also $2$. That is not true, since the order $x$ is $p$. Or can $p$ take the value $2$ ? $$$$ If $n=2$ we have that $(xy)^2=e \Rightarrow x^2y^2=e \Rightarrow x^2=e$. That is only true when the order of $x$ is $2$, so when $p=2$. $$$$ If $n=p$ we have that $(xy)^p=e \Rightarrow x^py^p=e \Rightarrow y^p=e$. That is only true when the order of $y$ is $p$, so when $p=2$. $$$$ Is that correct? Can it be that $p=2$ ? @DerekHolt – Mary Star Mar 26 '16 at 12:31
  • Or do we not show that $2p$ is the order of $xy$ in that way? @DerekHolt – Mary Star Mar 26 '16 at 13:45
  • A question out of time, @DerekHolt: "Since we are allowed to use Sylow, we can assume $G$ is generated by $x,y$ with $x^p=y^2=1$...": doesn't this follow just from Cauchy and $\langle x\rangle\cap\langle y\rangle={1}$? –  Jan 28 '22 at 10:53
  • 1
    Yes it follows from Cauchy or from Sylow. – Derek Holt Jan 28 '22 at 11:02
4

Since the $2$-Sylow subgroup is cyclic, the group has a normal $2$-complement (corollary to Burnside's transfer theorem), which means that the $p$-Sylow subgroup is normal (or just use that any subgroup of index $2$ is normal). Thus, the group is a semidirect product of a cyclic group of order $p$ and one of order $2$. Since the Automorphism group of the cyclic group of order $p$ has a unique subgroup of order $2$, this means that there can only be one non-trivial such semidirect product, and since $D_p$ is such a semidirect product, it must be it. If the semidirect product is trivial, we of course get the cyclic group of order $2p$.

  • 3
    You really don't need Burnside's transfer Theorem! A Sylow $p$-subgroup has index 2 and so must be normal. – Derek Holt Jan 31 '13 at 13:53
  • @DerekHolt I know (I also mentioned that simpler version). But as the proof can actually be done about as elegantly by elementary means, but the OP explicitly allowed advanced results, I could not resist using this approach (which can of course be used in much more general settings). – Tobias Kildetoft Jan 31 '13 at 13:55
  • For $p=2$ the trivial semidirect product is not the cyclic group, but it is $D_2$ in that case. But otherwise a nice proof. – Marc van Leeuwen Jan 31 '13 at 14:27
  • @MarcvanLeeuwen right, I assume $p\neq 2$ as otherwise the result is trivial. – Tobias Kildetoft Jan 31 '13 at 14:36
3

If the group is Abelian then by the Classification of Finitely generated abelian groups we know that $G=\mathbb{Z}_{2p}$ is the only possibility (if $p$ is an odd prime).

If it is non Abelian:

By Cauchy's theorem one gets that there exits an element of order $p$ and order $2$. Call them $x$ and $y$ respectively, then all possible elements of the group are as follows:

$$\{1,x,x^2,...,x^{p-1},y,yx,yx^2,...,yx^{p-1},xy,x^2y,...x^{p-1}y\}$$ since the order of the group is $2p$, we get that $yx=x^{j}y$ for some $2\leq j\leq p-1$. ($j=1$ will force the group to be Abelian, so it not possible).

Then $yx^2=(yx)x=x^j(yx)=(x^{2j\bmod{p}})y\implies yx^k=(x^{kj\bmod{p}}y)$ by induction.

Then $y(yx)=x=y(x^jy)=(yx^j)y=x^{j^2 \bmod{p}}yy=x^{j^2 \bmod{p}}\implies x^{j^2 -1\bmod p}=1\implies j=\pm1\bmod p$. Since the group is not Abelian we get that $yx=x^{p-1}y$ is the only reasonable relation possible, now consider the homorphism induced by $y\mapsto s$ and $r=x\mapsto r $. This is clearly an isomorphism to $D_{p}$.

Therefore only groups of order $2p$ are $\mathbb{Z_{2p}}$ or $D_{p}$.

1

By Sylow-1 we can say that $\mathbb{Z}/(2)\simeq H<G$, $\mathbb{Z}/(p)\simeq K<G$, by Lagrange $H\cap K=\{e\}$ and it is easy ro see that $G=HK$.

And since $(G:K)=2$, $K\triangleleft G$.

So $G=K \rtimes_f H$ for some $f:H\to \operatorname{Aut}(K)$. It is well-known that $\operatorname{Aut}(\mathbb{Z}/(p))\simeq \mathbb{Z}/(p-1)$. So there are 2 such morphisms since $\gcd(2;p-1)=2$.

So, there are 1 or 2 groups of order $2p$, and we can name 2 of them: $\mathbb{Z}/(2p)$ and symmetries of a regular $n$-gon.

0

For $p=2$ our group has 4 elements, so it is either cyclic or the Klein-4-group. So let $p > 2$.

By Cauchy's theorem you know there is a cyclic subgroup of order $p$ and $2$, call this $H \leq G$. Let $a$ be a generator of $H$ and let $b$ be an element of order $2$. Then $a,b$ generates $G$, and thus we only need to consider how $a$ acts on $b$.

$H$ is normal since $|G : H| = 2$, so $$bab^{-1} \in H \implies bab^{-1} = a^j,\ 0 \leq j < p$$ And $b$ has order $2$, so $$a = b^2 ab^{-2} = b (bab^{-1})b^{-1} = b a^j b^{-1} = \overbrace{(b a b^{-1}) \cdots (bab^{-1})}^{j \text{ times}} = a^{j^2}$$

Thus $j^2 \equiv 1 \mod p \implies j \in \{1, p-1\}$.

If $j = 1$, then $G$ is Abelian, so $G \simeq C_{2p}$ (The cyclic group of order $2p$)

If $j = p-1$, it is clear that the mapping of $a$ to a rotation and $b$ to a mirroring is an isomorphism to $D_p$. (Since rotating then mirroring is the same as mirroring then rotating the other direction)

Eknoma
  • 39
0

By Cauchy's theorem, $G$ has an element $x$ of order $p$ and an element $y$ of order $2$. If $\text{o}(xy)=2p$ then the group is cyclic, and if $\text{o}(xy)=1,2$ then the group is dihedral because we would have the presentation $x^p=y^2=(xy)^2=1$. So suppose that $\text{o}(xy)=p$. Clearly, $xy\not\in \langle x\rangle$, since that would imply $y\in \langle x \rangle $. Thus, $G$ has at least $p$ elements of order $p$, namely $xy,x,x^2,\ldots,x^{p-1}$. However, $$p\leq |\{x\in G:\text{o}(x)=p\}|\leq 2p-2=|G\backslash \langle y\rangle|,$$ but none of the numbers $p,p+1,\ldots,2p-2$ are $\equiv -1\pmod{p}$. This contradicts Cauchy's theorem on the number of elements of order $p$.

0

You can say we have an element of order p or 2p because if all of the elements has order at most two we can say the group is abelian. Then if you have an element with order 2p it's solved. If you have an element with order p. Then you can say we have an element with order 2 because the number of elements with order two is odd in groups with even order. After finding this two elements the remaining part is the same as "Derek"s solution. The point is you can prove it easily without using Cauchy or Sylow theorem.

0

Let $|G| = 2p$, and it can be written as $G = \{e, a, \ldots, a^{p-1}, ba, \ldots, ba^{p-1}, ab, \ldots, a^{p-1}b\}$. However, this appears to be of order $3p$, so there must be an equivalence $ba = a^{n}b$ for some $n \leq p-1$.

We observe that if we have an element $b \in G$ of order 2, there exists an automorphism of $H \cong \mathbb{Z}_{p}$ defined by $\alpha(a) = bab^{-1}$ such that $\alpha^2 = \text{id}$. To understand why this is the case, consider this geometrically where $a$ and $b$ represent either rotations or reflections of a regular $2p$-sided polygon. This leads to $\alpha^2(a) = a^{n^2} = \text{id} = a \implies n^2 = 1$, which has two solutions in the ring $\mathbb{Z}_{p}$, namely $n = \pm 1$. If $n = 1$, $G$ is abelian and isomorphic to $\mathbb{Z}_p \times \mathbb{Z}_2$. Otherwise, we obtain precisely $bab^{-1} = a^{-1}$ (which also has geometric significance), making it isomorphic to $D_{2p}$.

-1

Use Cauchy Theorem

Cauchy's theorem — Let $G$ be a finite group and $p$ be a prime. If $p$ divides the order of $G$, then $G$ has an element of order $p$.

then you have an element $x\in G$ of order $2$ and another element $y\in G$ of order $p$. Now you have to show that $xy$ is of order $2p$
using commutativity we get:

$(xy)^2 = y^2$, and hence $ord(xy) \not| 2$

$(xy)^p = x$ , therefore $ord(xy) \not| p$

and $(xy)^{2p} = y^{2p} = e$, then $ord(xy) | 2p$

hence $1 <2<p<ord(xy) | 2p$
an then $ord(xy) = 2p$ because it doesn't equal to any divisor different of $2p$.