# Questions about the subgroups of $SU(2)$ and relevant problems?

This question is based on the invariant gauge groups in condensed matter physics( https://physics.stackexchange.com/questions/76644/different-invariant-gauge-groups-igg-on-different-lattices-with-the-same-form ).

As we know, the group $SU(2)$ can be viewed as the set of all the $2\times2$ unitary matrices with determinant $1$. And let the groups $U(1)=\left \{\begin{pmatrix} e^{i\theta}&0 \\ 0 & e^{-i\theta} \end{pmatrix} \mid 0\leqslant\theta<2\pi \right \}$, $\mathbb{Z}_2=\left \{I,-I \right \}$, where $I$ is the $2\times2$ identity matrix. Then $U(1)$ and $\mathbb{Z}_2$ are both the subgroups of $SU(2)$. My questions are as follows:

(1) In addition to $U(1)$ and $\mathbb{Z}_2$(of course $SU(2)$ itself and its subgroup { $I$ }), are there any other subgroup in $SU(2)$?

(2) Does $SU(2)$ have two subgroups called $A$ and $B$ such that: $U(1)$ is a subgroup of $A$ , $B$ is a subgroup of $U(1)$, and $\mathbb{Z}_2$ is a subgroup of $B$ ?

(3) Also as we know, $SO(3)\cong SU(2)/\mathbb{Z}_2$, and does $SU(2)$ have a subgroup called $C$ such that $C\cong SO(3)$? More generally, if $K$ is a normal subgroup of group $G$, then does $G$ have a subgroup called $H$ such that $H\cong G/K$?

• Perhaps you mean proper subgroups in (2). – user1551 Sep 9 '13 at 10:44
• To (1), at least, what about all of $SU(2)$ itself? – Neal Sep 9 '13 at 10:44
• @ user1551, by ' proper subgroups', what do you mean? Thanks. – Kai Li Sep 9 '13 at 15:31

1. There are plenty of interesting subgroups of $SU(2)$. For instance the rotational symmetry group of the icosahedron embeds (in many ways) into $SO(3)$, and the inverse image of that subgroup under the projection $SU(2)\to SO(3)$ is a subgroup of $SU(2)$ of order $120$. More prosaically the indicated copy of $U(1)$ has many conjugate subgroups in $SU(2)$.
2. The existence of $A$ and $B$ are independent. An interesting choice for $A$ is its normaliser in $SU(2)$, which apart from the matrices $D$ of $U(1)$ itself also contains matrices $\binom{0~-1}{1~\phantom+0}D$. For $B$ you could take the subset of matrices $D$ with $\theta$ a multiple of $\frac\pi{n}$ for some integer $n>1$. Explicitly one can take $$A = \left \{\begin{pmatrix}e^{i\theta}&0 \\ 0 & e^{-i\theta} \end{pmatrix} \mid 0\leq\theta<2\pi \right \} \cup \left \{\begin{pmatrix} 0 & -e^{-i\theta}\\ e^{i\theta}&0\end{pmatrix} \mid 0\leq\theta<2\pi \right \}$$ and $$B= \left \{\begin{pmatrix}e^{\frac{k\pi}7i}&0 \\ 0 & e^{-\frac{ k\pi}7i} \end{pmatrix} \mid k=0,1,\ldots,13 \right \}$$
3. No, $SO(3,\Bbb R)$ does not embed into $SU(2)$ (which is an example of the fact that in general quotient subgroups do not always embed as subgroups). It would be somewhat technical to show that there is no embedding as abstract subgroup (where one would have to exclude highly discontinuous embeddings as well), so I'll just show that such an embedding as a Lie subgroup (so compatible with the differential structure of the groups) is impossible. In such an embedding there would be some Lie algebra element $X$ such that te one parameter subgroup $\{\,\exp tX\mid t\in\Bbb R\,\}$ lies in the subgroup. But that set always contains the element $-I$ of $SU(2)$, whereas $SO(3,\Bbb R)$ contains no central elements of order$~2$.
• @ Marc van Leeuwen, thank you very much for your excellent answer. But I'm still unclear about: In "2", can you write the explicit form of the groups $A$ and $B$? And for "3", is there any proof for your answer 'No and no'? Thanks a lot. – Kai Li Sep 9 '13 at 15:30
• @ Marc van Leeuwen, thanks for your update. – Kai Li Sep 10 '13 at 9:54

1) What about any discrete subgroup in $U(1)$?

2) $B=A=U(1)$ and $Z_2≅±1⊂U(1)$

3) The representation theory of $SO(3)$ is wellknown, so you only have to look up whether there are two-dim'l complex representations. For the general question, no! Consider $G=\mathbb{Z}$.

Concerning my question (3). $SO(3)$ can not be viewed as a subgroup of $SU(2)$. Since there is only one element $-1$ of order 2 in $SU(2)$, while there are more than one elements of order 2 in $SO(3)$.

Referring to my second question, let $\mathbb{Z}_N=\left \{ \begin{pmatrix} e^{i\theta}&0 \\ 0& e^{-i\theta} \end{pmatrix} \mid \theta=\frac{2\pi}{N},2\cdot\frac{2\pi}{N},...,2\pi\right \}$ with positive integer $N$, then $\mathbb{Z}_N$ is a subgroup of $U(1)$. And $\mathbb{Z}_N=B$ for even $N$.