4
$\begingroup$

I need to find a group homomorphism from $SL_2 (\mathbb Z / 5 \mathbb Z)$ to $S_5$. There is obviously the trivial homomorphism but I am then asked to deduce that $SL_2 (\mathbb Z / 5 \mathbb Z)/\{I,-I\} \cong A_5$, so ideally I want a homomorphism with kernel $\{I,-I\}$, and image $A_5$ so I can use the isomorphism theorem. Any hints?

$\endgroup$
5
  • 1
    $\begingroup$ You could show that ${\rm PSL}(2,5)$ is a simple group of order $60$, and use thus that $A_5$ is the unique simple group of order $60$. You can show more generally that ${\rm PSL}(2,K)$ is simple provided $K$ has more than three elements. $\endgroup$
    – Pedro Tamaroff
    Jan 12 '15 at 23:14
  • $\begingroup$ As a general tactic, morphisms $G\to S_n$ arise whenever $G$ has a subgroup of index $n$, see this. If you could show that ${\rm SL}(2,5)$ has a subgroup of index $5$ that contains the center of ${\rm SL}(2,5)$ you might be in good path. $\endgroup$
    – Pedro Tamaroff
    Jan 12 '15 at 23:41
  • $\begingroup$ (It is interesting to note that ${\rm SL}(2,5)$ is not isomorphic to $S_5$, however!) $\endgroup$
    – Pedro Tamaroff
    Jan 12 '15 at 23:48
  • $\begingroup$ This is related to math.stackexchange.com/questions/1096446 $\endgroup$
    – Derek Holt
    Jan 13 '15 at 9:19
  • 2
    $\begingroup$ @PedroTamaroff Yes, there are three non-solvable groups of order $120$, $S_5$ with a $2$ at the top, ${\rm SL}(2,5)$ with a $2$ at the bottom, and $A_5 \times C_2$ with both. $\endgroup$
    – Derek Holt
    Jan 13 '15 at 9:21
3
$\begingroup$

I couldn't claim that this is an easiest way; but I have explained once this solutionway in my classroom for undergraduates.

The group $A_5$ has presentation $\langle x,y\colon x^2=1, y^3=1, (xy)^5=1\rangle$.

We can obtain a homomorphism you expect by this presentation. We find a pair of elements $X,Y$ in $SL(2,5)$ such that $X$ has order $2$, $Y$ has order $3$ and $(XY)$ has order $5$.

Each element of $SL(2,5)$ gives a Mobius transformation $z\mapsto \frac{az+b}{cz+d}$ where $a,b,c,d$ are in the domain=codomain=$\{0,1,2,3, 4,\infty\}$ of the transformation (convention $1/0=\infty$ and $1/\infty=0$). So we may try to find two Mobius transformation, one of order $2$, one of order $3$ so that their product has order $5$.

One choice for Mobius map of order $2$ is $z\mapsto \frac{-1}{z}$. For order $3$ Mobius map, one possibility is to find a map $z\mapsto \frac{az+b}{cz+d}$, such that $0\mapsto 1\mapsto \infty\mapsto 0$ (so that its order will be $3$), and it is in fact $z\mapsto \frac{1}{1-z}$.

Now it is easy to check that their product turns out to be order $5$.

What are the corresponding element in $SL(2,5)$? Well. These are $$X= \begin{bmatrix} 0 & 1\\ -1 & 0\end{bmatrix} \mbox{ and } Y= \begin{bmatrix} 1 & 0\\ -1 & 1 \end{bmatrix}$$ which can be written just by considering the coefficients involved in the Mobius transformations. Now, the map $X\mapsto x$ and $Y\mapsto y$ can be uniquely extended to a homomorphism $f\colon SL(2,5)$ to $A_5$ (or $S_5$) (from the definition of the free group; I will not make it too technical.) And, thats it. .

$\endgroup$
2
  • $\begingroup$ You said that you were looking for $X$ and $Y$ in ${\rm SL}(2,5)$ such that $X$, $Y$, and $XY$ have orders $2,3,5$, but in your final choice of $X$ and $Y$, they have orders $4,5,3$. Note also that the only element of ${\rm SL}(2,5)$ of order $2$ is $-I_2$, which is in its centre. $\endgroup$
    – Derek Holt
    Jan 13 '15 at 13:17
  • $\begingroup$ Yaa! If we go modulo $\{\pm I\}$, then we get a homomorphism. So here $SL(2,5)\rightarrow PSL(2,5)\rightarrow A_5$ is the required homomorphism, where first one is quotient and second is obtained by $\bar{X}\mapsto x$, $\bar{Y}\mapsto y$. Thanks for notification. $\endgroup$
    – Groups
    Jan 13 '15 at 15:23
0
$\begingroup$

Just wanted to add some extra commentary.


Let $I$ be the rotational symmetry group of the (regular) icosahedron. By orbit-stabilizer, it has order $60$. It is a subgroup of $\mathrm{SO}(3)$, the full group of $3\times3$ rotation matrices. Every edge of the icosahedron uniquely determines a compound of three inscribed semiperpendicular golden rectangles:

$\hskip 0.5in$ icosa

There are $6$ icosahedron edges per compound, so there are $30/6=5$ compounds. Note $I$ acts on this set $\Omega$ of $5$ compounds, indeed we may identify $I$ with an index $2$ subgroup of $\mathrm{Sym}\,\Omega$, forcing it to be the alternating subgroup by uniqueness. Thus, $I\cong A_5$ by labelling the compounds $1$ thru $5$.

On the other hand, there is a double cover $2I$ of the icosahedral group (called the binary icosahedral group), which is the preimage of $I$ under the $2$-to-$1$ group homomorphism $S^3\to\mathrm{SO}(3)$, where $S^3$ is the group of versors (unit quaternions). (There are many resources discussing how the antipodal versors $\pm \exp(\theta\mathbf{u})$ represent the $3$D rotation around the $\mathbf{u}$-axis by $2\theta$, where $\mathbf{u}$ is a unit vector.) By considering quaternions as a $2$D complex vector space, we may identify quaternions with $2\times2$ complex matrices, in which case $S^3$ may be identified with $\mathrm{SU}(2)$. Then the elements of $2I$ may be reduced mod $5$ to get $\overline{2I}=\mathrm{SL}_2(\mathbb{F}_5)$. Since the kernel of $2I\to I$ corresponds to the kernel of $\mathrm{SL}_2(\mathbb{F}_5)\to\mathrm{PSL}_2(\mathbb{F}_5)$, we conclude $I\cong\mathrm{PSL}_2(\mathbb{F}_5)$.

Thus, $\mathrm{PSL}_2(\mathbb{F}_5)\cong A_5$. I explained this in my answer here.


Symmetric groups have trivial outer automorphism group, except for the exceptional case of $\mathrm{Out}\,S_6=\mathbb{Z}_2$. An outer automorphism of $S_6$ swaps cycle types $(\cdot\cdot)$ and $(\cdot\cdot)(\cdot\cdot)(\cdot\cdot)$ as well as $(\cdot\cdot\cdot)$ with $(\cdot\cdot\cdot)(\cdot\cdot\cdot)$. While the subgroup $A_5$ stabilizes a point, its image under an outer automorphism is an "exotic" copy of $A_5$ which is actually transitive. This matches (up to relabelling) the action of $\mathrm{PSL}_2(\mathbb{F}_5)$ on $\mathbb{F}_5\mathbb{P}^1$ and of $I$ on the set $\Phi$ of antipodal pairs of vertices of the icosahedron (note $|\mathbb{F}_5\mathbb{P}^1|=|\Phi|=6$ in contrast to $|\Omega|=5$).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.