3

Let $G$ be a group s.t. every non-identity element has order 2. If $G$ is finite, prove $|G| = 2^n$ and $G \simeq C_2 \times C_2 \times\cdots\times C_2$

I know G is abelian since $ab = (ab)^{-1} = b^{-1} a^{-1} = ba$ for all non-trivial $a,b \in G$ so I have several questions remaining:

  1. How do I prove $|G| = 2^n$? I'd like to say we use induction to prove this but I'm at a loss as to where I would start.
  2. Why is $G \simeq C_2 \times C_2 \times\cdots\times C_2$?

I've read several answers to question similar to this yet unfortunately most of them involve Galois Fields and vector spaces, both concepts I'm unfamiliar with. I'd greatly appreciate an intuitive proof.

2 Answers2

2

Suppose $G$ is finite, then we have $|G|= \prod_{i=1}p_{i}^{r_{i}}$. By the hypothesis and Cauchy's theorem, $p_{i} = 2$. So that $|G| = 2^{n}$.

Let $x, y \in G$, then $xy = 1$ or $(xy)^{2} = 1$ by the hypothesis and so we have $xy = yx$ since $g^{2} = 1 \implies g = g^{-1}$. Hence $G$ is an abelian group.

Write $G = \{g_{1}, g_{2}, ..., g_{n}\}$ where $|g_{i}| = d_{i}$. Let $H = \prod_{i=1}^{n}\mathbb{Z} \backslash d_{i}\mathbb{Z}$.

Then $\phi: H \rightarrow G: \phi (g_{1}, ..., g_{n}) = \prod_{i=1}^{n}g_{i}^{k_{i}}$ is a well-defined surjective map since $G$ is abelian but by order consideration, $\ker \phi = \{1 \}$. So we have the desired isomorphism since $d_{i} = 2$.

  • I'm sorry but could you be more specific about how we can conclude that $|G| = 2^{n}$ by Lagrange's theorem and the hypothesis? – Jonathan Lee Dec 15 '15 at 04:39
  • Also, is there a way we can prove that G≃C2×C2…C2 without using the fundamental theorem for finitely generated abelian group? In other words, could we accomplish this proof using isomorphisms? – Jonathan Lee Dec 15 '15 at 04:42
  • I think in place of Lagrange's I should have said Cauchy's Theorem. If there is $p_{j}$ different from $2$, then $G$ has a non-identity element of order different from 2 which contradicts the hypothesis. So all the $p_{i}$ are just 2. Then the product is just $2^{\sum_{i=1}^{k}r_{i}} = 2^{n}$ – Daniel Akech Thiong Dec 15 '15 at 04:47
  • Okay, I understand that much. Now then, why do we get that $G \simeq C_{2} \times C_{2} \dots C_{2}$ as the primary decomposition of G and not anything else? We have not gone over the fundamental theorem for finitely generated abelian groups yet in class so I'm particularly curious about this one – Jonathan Lee Dec 15 '15 at 04:52
  • I have just written a completely new answer along the lines of what you have suggest: using isomorphism theorems and I have just posted it together with my earlier answer. – Daniel Akech Thiong Dec 15 '15 at 04:58
  • If anyone ever asks you to prove Cauchy's theorem for finite abelian groups, then this construction given in the answer is actually how one constructs an element of order prime p for every prime dividing the order of an abelian group. I thought I mention this since I have invoked Cauchy's theorem in the proof. – Daniel Akech Thiong Dec 15 '15 at 05:05
  • Can you please describe what the notations $p_i , r_i $ describe here? I can't realy see how you get the first conclusion. – Ron Abramovich Nov 12 '20 at 18:23
1

(1) $G$ is Abelian because $x y=x(x y)^2y=x x y x y y=x^2(y x)y^2=y x.$.. (2) For non-negative integer $n$, let $H_n$ denote a subgroup of $G$ with $|H_n|=2^n. $ It is easily seen that if $H_n\ne G$ and $x\in G\backslash H_n$ then $H_{n+1}=H_n\cup x H_n$ is a subgroup of $G$ with $|H_{n+1}|=2^{n+1}$....(3) Now $H_0=\{1\}$ exists. Take a sequence $x_0,...,x_n$ in $G$ and a sequence $H_0,...,H_n$, with $x_0=1$, $H_0=\{1\}$, such that for $0\leq i< n$ we have $x_{i+1}\not\in H_i$ and $H_{i+1}=H_i\cup x_{i+1}H_i,$ and where $n$ is as large as possible. By part (2) we must have $H_n=G$. This gives us $|G|=2^n$....(4) In the non-trivial case $n\geq 1$, observe that $H_n$ (which is $G$) is isomorphic to $\prod_{j=0}^{j=n-1}(H_{n-j}/H_{n-j-1})$ and each member of this product is a two-element group.