2

Question:

Prove that $G=$SL$(2, \mathbb{F}_5)$ is an extension of $\mathbb{Z}_2$ by $A_5$ which is not a semidirect product.

(This is a question from Rotman's Advanced Modern Algebra which I am trying to self-learn.)

I guess that we can send $\mathbb{Z}_2$ to $$K = \{ \begin{bmatrix} 1 & 0 \\ 0 & 1 \\ \end{bmatrix} , \begin{bmatrix} -1 & 0 \\ 0 & -1 \\ \end{bmatrix} \} $$ which is normal in $G$. and then I read online that $G/K$ is isomorphic to $PSL(2, \mathbb{F}_5)$ which is isomorphic to $A_5$.

So I can create the short exact sequence $$ 0 \to \mathbb{Z}_2 \to^i G \to^p A_5 \to 1$$

Does this suffice to show that $G$ is an extension of $\mathbb{Z}_2$ by $A_5$?

Next, I want to show that this is not a semidirect product, which means I need to show that the extension is not split, which means I need to show there does not exist any homomorphism $j :A_5 \to G$ such that $pj$ is the identity on $A_5$. But how do I do this? Is it not possible to use the isomorphism $G/K \cong A_5$ to construct some homomorphism?

Shaun
  • 44,997
eatfood
  • 2,414
  • (+1, interesting question!) Don't have Rotman's book, but I've always used the phrase "$G$ is an extension of $A$ by $B$" to mean that there is an exact sequence $0 \to B \to G \to A \to 0$ instead of the other way around! Is this a quirk of the textbook or standard? – hunter Feb 07 '20 at 23:05
  • @hunter it is neither. The terminology “$A$ by $B$” and “$B$ by $A$” are each used by different authors. – KCd Feb 07 '20 at 23:23
  • @hunter I have no idea actually, this is my first time learning about extensions XD – eatfood Feb 08 '20 at 16:41

2 Answers2

2

Your work is totally fine.

Now, for your last question: if the sequence were split, then $A_5$ would be isomorphic to a subgroup of index $2$ of $G$, and you would get a surjective morphism $G\to \mathbb{Z}/2\mathbb{Z}$ with kernel isomorphic to $A_5$.

Now since $\mathbb{Z}/2\mathbb{Z}$ is abelian, this morphism sends any element of $[G,G]$ to $0$ (because any commutator is mapped to $0$. It is a well-known fact that $[SL_n(K),SL_n(K)]=SL_n(K)$ except if $n=2$ and $K=\mathbb{F}_2$ or $\mathbb{F}_3$. Using this classical fact, you see that your morphism is trivial, and you get a contradiction.

GreginGre
  • 15,028
  • Thanks! Proof for the well-known fact: https://math.stackexchange.com/q/2372936/577979 – eatfood Feb 07 '20 at 00:10
  • Note your argument is not specific to ${\rm SL}_2(\mathbf F_5)$: for every field $K$ not of characteristic $2$ other than $\mathbf F_3$, the exact sequence $1 \rightarrow {I_2} \rightarrow {\rm SL}_2(K) \rightarrow {\rm PSL}_2(K) \rightarrow 1$ does not split. – KCd Feb 07 '20 at 23:34
1

For the last part, note that by looking at Rational or Jordan Canonical forms one can easily see that $SL(2,5)$ has only one class of involutions, namely the class containing $-I$. This element is central, so there is exactly one involution.

Hence the extension of the Sylow-2-subgroup of $SL(2,5)$ (or order $8$) by the central involution can't split, so neither can the extension of the whole group.

ancient mathematician
  • 14,102
  • 2
  • 16
  • 31