Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I would like to know all finite subgroups of $\operatorname{Aut}(\mathbb{P}^1)$.

I am aware that any automorphism of $\mathbb{P}^1$ is given by a Möbius transformation $$ z\mapsto\frac{az+b}{cz+d} $$ and thus there is an identification $$ \operatorname{Aut}(\mathbb{P}^1)\cong \operatorname{PSL}(2,\mathbb{C})\cong \operatorname{SO}(3, \mathbb{C}). $$ I thought this solved the question, but what I know is the classification of finite subgroups of real orthogonal group $\operatorname{SO}(3, \mathbb{R})$.

share|cite|improve this question
I think the third groupe of your line of isomorphism is in fact $SO(3,\mathbb{R})$, not $SO(3,\mathbb{C})$ and hence your problem is solved (you can check that $SO(3,\mathbb{C})$ doesn't have the good dimension as a real manifold if you're not covinced...). – Simon Henry Oct 13 '12 at 11:22

So pointed out by Simon Henry, $\operatorname{PSL}(2,\mathbb{C})\cong \operatorname{SO}(3, \mathbb{R})$, not $\operatorname{SO}(3, \mathbb{C})$. Recall that $SO(3, \mathbb{R})$ is the set of orientation-preserving isometries of $\mathbb{R}^3$, and hence $S^2$ (which is homeomorphic to $\mathbb{P}^1$).

The finite subgroups of $SO(3, \mathbb{R})$ have been classified, see the groupprops wikipage for example. I will briefly outline them below.

Note that $SO(2, \mathbb{R})$ naturally includes in $SO(3, \mathbb{R})$. More precisely, the map

\begin{align*} SO(2, \mathbb{R}) &\to SO(3, \mathbb{R})\\ A &\mapsto \begin{bmatrix} A & 0\\ 0 & 1\end{bmatrix} \end{align*}

gives an isomorphism between $SO(2, \mathbb{R})$ and a subgroup of $SO(3, \mathbb{R})$. Therefore, any subgroup of $SO(2, \mathbb{R})$ can be realised as a subgroup of $SO(3, \mathbb{R})$; in particular, every finite subgroup of $SO(2, \mathbb{R})$ will give a finite subgroup $SO(3, \mathbb{R})$. Note that the elements of $SO(2, \mathbb{R})$ are simply the rotation matrices and so we have $SO(2, \mathbb{R}) \cong S^1$. The only finite subgroups of $S^1$ are finite cyclic subgroups, so $SO(3, \mathbb{R})$ has such subgroups.

Another infinite family of finite subgroups of $SO(3, \mathbb{R})$ are the dihedral groups $D_{2n}$, which are the symmetries of a regular $n$-gon in a plane. Although these symmetries leave the plane fixed, they do not come from elements of $SO(2, \mathbb{R})$ as above because reflections swap the sides of the plane.

The remaining finite subgroups of $SO(3, \mathbb{R})$ rely on the classification of platonic solids. There is a subgroup isomorphic to $A_4$ given by the symmetries of a tetrahedron, a subgroup isomorphic to $S_4$ given by the symmetries of a cube or octahedron, and a subgroup isomorphic to $A_5$ given by the symmetries of an icosahedron or dodecahedron. Note, the platonic solids which appear in pairs are dual polyhedra (a tetrahedron is self-dual).

In addition to all of the finite subgroups discussed above, we must also consider their subgroups. In doing so, we see that all of them have already been mentioned${}^1$. Therefore, a finite subgroup of $SO(3, \mathbb{R})$ is isomorphic to one of the following: $\mathbb{Z}/n\mathbb{Z}$, $D_{2n}$, $S_3$, $A_4$, $S_4$, $A_5$.

${}^1$ I initially thought $S_3$ (a subgroup of both $S_4$ and $A_5$) was missing, but as pointed out by pjs36, $S_3 \cong D_6$.

share|cite|improve this answer
An emphatic +1! But isn't $S_3 \cong D_6$? – pjs36 May 25 '15 at 6:14
@pjs36: Yes it is! I totally overlooked this. I will edit accordingly, thanks. – Michael Albanese May 25 '15 at 6:15

(I am not allowed to add any comments, so I add an answer.) In contrast to what has been said here, the groups $\mathrm{PSL}(2,\mathbf C)$ and $\mathrm{SO}(3,\mathbf R)$ are not naturally isomorphic. The topological group $\mathrm{PSL}(2,\mathbf C)$ is not compact. Indeed, the sequence of matrices $$ \left\{ \begin{pmatrix} 1&n\\0&1 \end{pmatrix} \right\}_{n\in\mathbf N}$$ in $\mathrm{PSL}(2,\mathbf C)$ does not have a convergent subsequence. The group $\mathrm{SO}(3,\mathbf R)$, however, is compact, being a topological subspace of the cube of the unit sphere in $\mathbf R^3$. Therefore, the groups $\mathrm{PSL}(2,\mathbf C)$ and $\mathrm{SO}(3,\mathbf R)$ are not isomorphic as topological groups, and definitely not naturally isomorphic. A more involved argument probably shows that they are not isomorphic as abstract groups, but that is not to the point here.

What is relevant, is that $\mathrm{PSL}(2,\mathbf C)$ contains the compact subgroup $\mathrm{PSU}(2,\mathbf C)$. Moreover, any finite, or compact for that matter, subgroup of $\mathrm{PSL}(2,\mathbf C)$ is conjugate to a subgroup of $\mathrm{PSU}(2,\mathbf C)$ (cf. this question). Now, the latter group is isomorphic to $\mathrm{SO}(3,\mathbf R)$ (cf. wikipedia), and the answer of Michael Albanese applies.

share|cite|improve this answer

There's an easy proof of the classification of finite subgroups of ${\rm PGL}(2, \mathbb{C})$ in Dolgachev's notes: McKay Correspondence

You can also search for Blichfeldt's book: Finite Collineation Groups.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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