2

I want to solve the following problem from Dummit & Foote's Abstract Algebra text (p. 184, Exercise 5):

Let $G=\text{Hol}(Z_2 \times Z_2)$

(a) Prove that $G=H \rtimes K$ where $H=Z_2 \times Z_2$ and $K \cong S_3$. Deduce that $|G|=24$.

(b) Prove that $G$ is isomorphic to $S_4$. [Obtain a homomorphism from $G$ into $S_4$ by letting $G$ act on the left cosets of $K$. Use Exercise 1 to show that this representation is faithful.]

Exercise 1 basically states that if $G=H \rtimes_\varphi K$ then $C_K(H)= \ker \varphi$.

My attempt:

(a) By definition $\text{Hol}(Z_2 \times Z_2)=(Z_2 \times Z_2) \rtimes_{\text{id}} \text{Aut}(Z_2 \times Z_2)$. Since the automorphism group of the Klein 4-group is isomorphic to $S_3$ (a previous result) we are done.

(b) Let $G$ act on $G/K$ by left multiplication, and let $\pi_K:G \to S_{G/K} \cong S_4$ be the induced permutation representation. We have that $$\ker \pi_K=\bigcap_{g \in G} gKg^{-1}$$ is the normal core of $K$...

This is where I'm stuck, since the homomorphism $\varphi$ in the semidirect product $G$ is the identity, Exercise 1 gives $C_K(H)=1$. I was thinking of showing that the action is faithful by proving $\ker \pi_K \leq C_K(H)(=1)$, but I couldn't do it.

Can anyone please help me solve part (b) using Exercise 1 as hinted by the authors?

Thank you!

Chain Markov
  • 16,012
user1337
  • 24,859
  • If by "the homomorphism $;\phi;$ in the semidirect product..." you mean the homomorphism that realizes the semidirect product, then it can't be trivial (i.e., the identity automorphism) as then we'd get a direct product. And if you insist in using D&F's hint, then you have to tell us what that exercise $;1;$ is ! – Timbuc Jan 05 '15 at 18:24
  • @Timbuc I mean that $\varphi:K=\text{Aut}(Z_2 \times Z_2) \to \text{Aut}(H)=\text{Aut}(Z_2 \times Z_2)$ is the identity map.

    P.S. Exercise 1 appears below the quote.

    – user1337 Jan 05 '15 at 18:26
  • But you also wrote $;H\rtimes_\varphi K;$ , so it should be $;\varphi: K\to,\text{Aut},(\Bbb Z_2\times\Bbb Z_2);$ , and thus it is not (it can't be) what you say it is! – Timbuc Jan 05 '15 at 18:28
  • @Timbuc I can't see why not, sorry. Could you please elaborate on that? – user1337 Jan 05 '15 at 18:29
  • the semidirect product $;A\rtimes_\varphi B;$ is realized by means of a homomorphism $;\varphi: B\to \text{Aut},A;$ , and this homom. is trivial iff we have a direct product . – Timbuc Jan 05 '15 at 18:31
  • @Timbuc I believe that the trivial homomorphism is defined by $\varphi(a)=1_B$ for all $a \in A$, while the identity homomorphism is defined by $\varphi(a)=a$ (where $A=B$ of course). – user1337 Jan 05 '15 at 18:35
  • When you say that "the hom. is the identity", I understand that $;\varphi(b):=Id_A;;\forall,b\in A;$ , with $;Id_A(a)=a;;\forall,a\in A;$ . You're making $;\varphi;$ act on $;A;$ , but it actually acts on $;B;l$ and one gets an automorphism of $;A;$ . Am I misunderstanding something? – Timbuc Jan 05 '15 at 21:30

2 Answers2

7

Here is my attempt at a solution. Glaring, painfully obvious errors are possible.

Suppose $g \in$ ker $\pi_H$. Then we would have $gxK=xK$ for every $x \in G$. In particular, note that we must have $g \in K$ (setting $x = 1$). $$gxK=xK \enspace\forall x\in G\Leftrightarrow \\x^{-1}gx \in K \enspace \forall x\in G \Leftrightarrow \\x^{-1}gxg^{-1} \in K \enspace \forall x\in G$$

Where the final line is because $g$ (and therefore $g^{-1}$) $\in K$. Restricting $x$ to only $H$ instead of all of $G$, we must have that $$x^{-1}gxg^{-1} \in K \enspace \forall x \in H$$ By normality of $H$, we must also have that $x^{-1}gxg^{-1} \in H \enspace \forall x \in H$, but since $H \cap K = 1$, $x^{-1}gxg^{-1} =1 \enspace \forall x \in H$, or $gx=xg \enspace \forall x \in H$. This means that $g \in C_K(H)$, but since $C_K(H)=1$, $g$ was the identity this whole time.

DJG
  • 173
2

Ok, perhaps now I can be of some help: since in the holomorph the homomorphism is the identity one, we have that $\;\ker\varphi=1\;$ , from which

$$\;C_{\text{Aut}(\Bbb Z_2\times\Bbb Z_2)}(\Bbb Z_2\times\Bbb Z_2)=1\iff \text{the only automorphism centralizing}\;\;\Bbb Z_2\times\Bbb Z_2\;$$

is the identity one (all this happens in the holomorph, of course, but it follows at once from the fact that the only permutation leaving fixed all the elements is the identity one) .

Thus, in the action of $\;G\;$ on the left cosets of $\;K\;$ (and thus in the induced homomorphism $\;G\to S_4\;$), we get that $\;x\cdot(gK):=(xg)K=gK\iff g^{-1}xg\in K\;$ and the last expression is true as inner automorphisms are automorphisms (this is painfully tautological).

Thus, the homomorphism is a monomorphism (i.e., the action is faithful) and we're done.

Timbuc
  • 34,795
  • I'm sorry but I don't understand this. I'm looking for a statement of the form, for any $x \ker \pi_K$, $x=1$ or something of this form. – user1337 Jan 06 '15 at 14:10