3

This is Exercise 3.2.10 of Robinson's "A Course in the Theory of Groups (Second Edition)". According to this search and Approach0, it is new to MSE.

The Details:

On page 8, ibid.,

Let $R$ be a commutative ring with identity. Define ${\rm SL}(n, R)$ to be the set of all $n\times n$ matrices over $R$ with determinant equal to $1$. [It is a group under matrix multiplication, called the special linear group of degree $n$ over $R$.]

On page 74, ibid.,

In case $R={\rm GF}(q)$, [we denote ${\rm SL}(n, R)$ by] $${\rm SL}(n, q).$$

On page 28, ibid., the definition of a derived subgroup $G'$ of a group $G$ is the subgroup generated by all commutators $[x,y]=x^{-1}y^{-1}xy$ of $G$.

To explain the title, I use the

Definition: A group $G$ is perfect if $G=G'.$

The Question:

Let $q$ be an odd prime power greater than $3$. Prove that ${\rm SL}(2,q)$ equals its derived subgroup.

Thoughts:

Let $q=p^\ell>3$ for an odd prime $p$.

My first thought was to try out small values of $q$, like $q=7$; here is a GAP experiment:

gap> G:=SL(2,7);
SL(2,7)
gap> Size(G);
336
gap> DG:=DerivedSubgroup(G);
<group of 2x2 matrices over GF(7)>
gap> Size(DG);
336

But nothing came of such considerations, since the groups are seemingly too big for a naïve approach. The next $q$ value is $3^2=9$: the order of ${\rm SL}(2, 9)$ is $720$ according to GAP. The next $q$ value after that is $11$: the order of ${\rm SL}(2, 11)$ is $1320$ according to GAP.

I don't know what I was hoping to see with using GAP.


My linear algebra is a bit rusty.


I guess I should move on like so . . .

Let $S={\rm SL}(2,q)$. Then

$$S=\left\{\begin{pmatrix} a & b \\ c & d\end{pmatrix}\,\middle|\, a,b,c,d\in GF(q), ad-bc=1\right\}.$$

Let $A=\begin{pmatrix} a & b \\ c & d\end{pmatrix}, B=\begin{pmatrix} x & y \\ z & t\end{pmatrix}\in S$. Then $\det(A)=\det(B)=1$. Consider $[A, B]$, a generator of $S'$. We have $[A,B]\in S$ by definition, so, like I used in the GAP code above, we may conclude that $S'\subseteq S.$

What I need to do is show that an arbitrary $\Xi=\begin{pmatrix} \alpha & \beta \\ \gamma & \delta\end{pmatrix}\in S$ is an element of $S'$. This is where I'm stuck.


It is well-known that

$$S'=\displaystyle\bigcap_{T\le S,\\ X\subseteq T}T,$$

where $X=\{[U,V]\in S\mid U,V\in S\}.$ I'm not sure how this helps.


Please help :)

Shaun
  • 44,997
  • Rotman has a neat proof of this in his introductory Springer book. – Pedro Jun 08 '21 at 21:46
  • I didn't find anything in my copy, @PedroTamaroff. I have the fourth edition. Please may I have a page number or some approximation of where it is in some other edition? – Shaun Jun 08 '21 at 21:50
  • 1
    In page 358 (same edition), after defining what a perfect group is, Rotman refers to Theorems 8.13 and 8.23. These involve claims about the simplicity of the projective special linear groups, but do imply that $\mathrm{SL}(n,q)$ is simple except for $(n,q)=(2,2)$ or $(2,3)$. I think you can also show this by hand using the Steinberg relations. – Pedro Jun 08 '21 at 22:37
  • 1
    The problem in low degrees is that you don't have enough wiggle room to solve for the unknowns because the field is too small, or because the matrices are not large enough. I recall reading this computation (or doing it myself) a while ago, but it is not difficult! Edit: looking at it, I am thinking about the proof of Theorem 8.13. – Pedro Jun 08 '21 at 22:38
  • 1
    (For reference, it took me a bit to find it, but I just went to the glossary and searched for simple group and perfect group, and the second one lead to the answer. ;)) – Pedro Jun 08 '21 at 22:46
  • Thank you, @PedroTamaroff :) – Shaun Jun 08 '21 at 22:55

2 Answers2

3

Let $q>3$. Then there exists a nonzero element $a\in \Bbb F_q$ with $a^2-1\neq 0$. Let $A=\begin{pmatrix} 1 & -x \cr 0 & 1 \end{pmatrix}$ and $B=\begin{pmatrix} a & 0 \cr 0 & a^{-1} \end{pmatrix}$ in $SL(2,q)$. Then \begin{align*} [A,B] & = ABA^{-1}B^{-1} \\ & =\begin{pmatrix} 1 & (a^2-1)x \cr 0 & 1 \end{pmatrix} \end{align*} So every element of the corresponding transvection group is a commutator. Since the transvections generate $SL(2,q)$, this is enough to show that $SL(2,q)$ is perfect.

Dietrich Burde
  • 130,978
  • Something seems wrong. Where have you used that $q$ is an odd prime power? Is that part unnecessary? – Shaun Jun 09 '21 at 22:15
  • 1
    @Shaun: $SL(n,q)$ is perfect for $n\ge 3$ or for $n=2$ and $q\ge 4$. A very nice source for this approach is chapter 4 of the book "The Geometry of the Classical Groups" by Donald E. Taylor. – j.p. Jun 10 '21 at 07:00
  • 3
    @Shaun, I have used it to find an $a$ different from $0,1,-1$. This is only possible for $q\ge 4$. – Dietrich Burde Jun 10 '21 at 08:48
1

I might approach it via Bruhat decomposition. Let $B$ be the group of upper triangular matrices in $G$. It is the semidirect product of the group $T$ of diagonal matrices in $G$ and the group $N$ of upper triangular matrices with $1$s on the diagonal. First, show that every element of $B$ is a product of commutators in $G$.

Next, use the fact that $G$ is the disjoint union of the double cosets $BwB$, where $w$ ranges through a set of permutation matrices in $G$. This is called the Bruhat decomposition. For example, $G = \operatorname{SL}_2$ is the disjoint union of $B = B1B$ and $BwB$, where

$$w = \begin{pmatrix} & 1 \\ -1 \end{pmatrix}.$$

Thus you just have to show that $B$ and every such $w$ is a product of commutators.

D_S
  • 33,891