I was trying to show that every finite subgroup of $\mathbb C^{\times}$ is equal to $G_n$ (the nth roots of unity) for some $n \in \mathbb N$ without invoking Lagrange's theorem, I got stuck at one point, I would appreciate some help to complete my solution.

Let $H \neq \{1\}$ be a finite subgroup of $\mathbb C^{\times}$. For any $x \in H \setminus \{1\}$ it is easy to show that there is a minimum $a$ natural such that $x^a=1$ and for all $k$, $x^k \in \{1,x,...,x^{a-1}\}$ with all these elements distinct. Since $H$ is finite, then $H=\{x_1,...,x_m\}$, consider $H'=\{x_{n_1},...,x_{n_r}\}$ all the elements in $H$ such that $x_{n_k} \neq x_{n_j}^m$ for all $k \neq j$ and all $m \in \mathbb Z$. For each $n_j$, $\exists$ $\alpha_{n_j}$ with ${x_{n_j}}^{\alpha_{n_j}}=1$ and $\alpha_{n_j}$ the minimum natural number that satisfies this.

I want to show $H=G_{lcm(\alpha_{n_1},...,\alpha_{n_r})}$

Let $h \in H$, then $h=x_{n_1}^{\beta_{n_1}}...x_{n_r}^{\beta_{n_r}}$, $0\leq \beta_{n_i} <\alpha_{n_i}$

If $t={lcm(\alpha_{n_1},...,\alpha_{n_r})}$, it is easy to verify that $h^t=1$, so $H \subset G_t$. I got stuck trying to show the other inclusion (or that $|H|=t$). I would appreciate any corections or/and suggestions. Thanks in advance.

  • $\begingroup$ What are you defining $G_n$ to be here? $\endgroup$ – Ramified_Minds May 1 '15 at 23:36
  • $\begingroup$ The nth roots of unity $\endgroup$ – user156441 May 1 '15 at 23:54

Here's one way you could go: Since you have already found that $H \subset G_t$, and you presumably know that $G_t$ is cyclic, generated by $e^{2\pi i/t}$, use the fact that every subgroup of $G_t$ is cyclic, equal to $G_{t'}$ for some $t' \mid t$. So if $H \subsetneq G_t$, then $H = G_{t'}$ and $x_{n_j}^{t'} = 1$ for every $j$.

Derive a contradiction with the fact that the order of each $x_{n_j}$ is $\alpha_j$ and $t = lcm(\alpha_{n_1}, \ldots, \alpha_{n_r})$; namely, $\alpha_{n_j} \mid t'$ for every $j$. You don't need Lagrange's theorem for this fact; you simply need the fact that the powers of $x_{n_j}$ cycle with period $\alpha_{n_j}$.


Think about the structure theorem for finite abelian groups:

Every finite abelian group is isomorphic to the product of cyclic groups with prime power degree, ie

$$H\cong\prod_{p\: \text{prime}}\prod_{i=0} \mathbb{Z}/p^{i}\mathbb{Z}$$

Where the $i$ are $0$ for all but finitely many $i$.

Then we can see that working under $\mathbb{C}^{\times}$, the group $G_{p^{i}}$ is isomorphic to $\mathbb{Z}/p^{i}\mathbb{Z}$.

So we can think about finite subgroups as the 'product' of the sets $\mu_{p^{i}}$, the $p^{i}$-th roots of unity. Since each $\mu_{p^{i}}$ contains $\mu_{p^{j}}$, for $j<i$, we need only consider the highest power of $p$. So take the product of sets of the $\mu_{p^i}$, for different primes $p$. Since each $p$ is coprime, we see that $\zeta_{p_1}\not\in\mu_p$ for any $p\neq p_1$. Thus, the product of all the sets must contain $\{1,\zeta_{p_1^i},\zeta_{p_1^i}^2,\ldots,\zeta_{p_1^i}^{p_1-1}\}$ and $\{\zeta_{p_2^j},\zeta_{p_2^j}^2,\ldots,\zeta_{p_2^j}^{p_2-1}\}$, and so on. This means that the product of these sets must be the $l$-th roots of unity, where $l=$lcm$(p_1^i,p_2^j,\ldots,p_n^k)$. But since these are all primes, this is simply $p_1^{a_1}p_2^{a^2}\ldots p_n^{a_n}$.

For an example, let's think about the possible subgroups of size $8$. An abelian group of size $8$ is either $C_8$, $C_4\times C_2$ or $C_2\times C_2\times C_2$. The second group does not have an analogue as a subgroup of $\mathbb{C}^{\times}$, because $G_2\subset G_4$, so when we take the 'set product' we simply get $G_4$ again, and similarly with $C_2^3$. Thus we are left with $G_8$.


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.