0

Let $G$ be a finite abelian group with exactly one element of order $2$ denoted by $\alpha$. Prove that $\prod_{g \in G} g = \alpha$

Okay so I was given two hints for this problem

Hint 1 : If $g \in G$ doesn't have order $2$, then $g \neq g^{-1}$

Hint 2: If you write out $0+1+2+3+4+50+1+2+3+4+5 \mod 6$, why does it come out $3$, other than computing brute force?

I'm not sure at all how to use hint 1.

The answer for Hint 2 is that you can pair up $5+1=6$ and $4+2 = 6$ and be left with $3$. In that hint we are looking at the additive group $\mathbb{Z}/6\mathbb{Z}$, and we have $|5| = |1| = 6$ and $|4| = |2| = 3$, so this hint would suggest looking at group operating each two elements of $G$ which have the same order with each other.

But the problem is, is that even though $G$ may be a finite group, some $g \in G$ may have $|g| = \infty$, and furthermore every $g \in G$ may have a unique order, (please correct me if I'm wrong). So I'm not sure how I could use it to prove $\prod_{g \in G} g = \alpha$

At this point all I can say about $G$, is that by Lagrange's Theorem, $|G| = 2k$ for some $k \in \mathbb{Z}^+$, and that (trivially) letting $\alpha \in G$ be the element of order $2$, we have $\alpha = \alpha^{-1}$.

I'm assuming there's some algebraic trick that I'm missing here, because I'm sure all I'd have to do is just write out $$\prod_{g \in G} g = g_1 \cdot g_2 \cdot \ ... \ \cdot g_{2k}$$ and find some way to simplify this expression.

EDIT: Thanks to the comments and answers below I realized that since $G$ is abelian, and for any $g \in G$, we have $g^{-1} \in G$, so $\prod_{g \in G} g = g_1 \cdot (g_1^{-1}) \cdot ... \cdot g_i \cdot(g_i^{-1}) \cdot ... \cdot (g_k) \cdot (g_k^{-1})$ where for some $i \in \{1, ..., k\}$ we have $g_i = \alpha$, but now I don't see why $\prod_{g \in G} g \neq e$ where $e$ is the identity of $G$?

Perturbative
  • 12,972
  • 1
    How can a finite group have an element of infinite order? By Lagrange's theorem, that's not possible. The proof should work by pairing all elements not of order two in the product, and then we'd be left with only $\alpha$. – Sarvesh Ravichandran Iyer Dec 04 '17 at 04:26
  • This is the answer to your edit: the element $\alpha$ appears only one time in the product. Since $\alpha$ has order two, its inverse is still $\alpha$ and $\alpha$ appears only one time in the product. – InsideOut Dec 04 '17 at 07:39

2 Answers2

3

Note that in your answer for hint 2, you are pairing $1$ with $5$ and $4$ with $2$. In the group $\mathbb{Z}/6\mathbb{Z}$, this exactly corresponds to pairing up inverse elements. Try to do this same pairing method for $G$. How many of the elements can be paired up in this fashion?

TomGrubb
  • 12,909
  • All elements can be paired up with their inverses, correct? Because we're working in an abelian group – Perturbative Dec 04 '17 at 04:46
  • but now I don't see why $\prod_{g \in G} g \neq e$?, where $e$ is the identity of $G$ – Perturbative Dec 04 '17 at 04:52
  • @Perturbative What happens to elements of order 2? – TomGrubb Dec 04 '17 at 04:53
  • If $\alpha \in G$ is al element of order $2$, then $\alpha = \alpha^{-1}$, so $\alpha^2 = e$ – Perturbative Dec 04 '17 at 05:09
  • @Perturbative You are correct there. But if the product is $\prod_{g\in G} g$ then how would we ever get $\alpha^2$ if $\alpha$ only comes around once? (; – Andrew Tawfeek Dec 04 '17 at 05:44
  • @AndrewTawfeek I thought $\alpha$ would cancel out with $\alpha^{-1}$, so we'd only be left with $\prod_{g \in G} g = e$ – Perturbative Dec 04 '17 at 05:47
  • @Perturbative Note that $\prod_{g \in G} g$ means we're taking the product of all the elements in $G$ but not writing down the same element twice. So somewhere in that big product, we have $\alpha$. Now how would $\alpha$ ever find it's partner $\alpha^{-1}$ to cancel with if it's partner is itself? – Andrew Tawfeek Dec 04 '17 at 05:52
  • @AndrewTawfeek Ahh I see, okay so now, since $\alpha = \alpha^{-1} \in G$ we have that $G \setminus {\alpha}$ contains $2k-1$ elements. Since inverses are unique, how could we pair up inverses then? Because there'd always be one element which isn't paired up with its inverse (we can pair up the rest of the $2k-2$ elements in $G$) – Perturbative Dec 04 '17 at 06:10
1

Alright, so as the problem says, lets let $G$ be a finite abelian group with one element, which we'll call $\alpha$, of order $2$.

Firstly, importatly recall the realization you made in the comments: $$\alpha^2=e \implies \alpha=\alpha^{-1}.$$

So now lets look at our product. We could go ahead and say $|G|$ is some postivie integer $n$ so we have something to refer to. Then $$\prod_{g\in G} g = g_1g_2g_3 \cdots \alpha \cdots g_{(n-1)}g_n$$ where each $g_i$ is a distinct element in $G$ and we know that $\alpha$, since it is also in $G$, is somewhere in that product.

As you realized thanks to the other answer, because this group is abelian we can shift things around. So lets go ahead and put every element next to it's inverse in the product -- but wait! This is where another thing we know comes into play: $\alpha$ is the only element of order $2$, that is, it's the only element whose inverse is itself.

Note that only elements of order $2$ have the property that their inverse is themself.

So now, rearranging our product, $\alpha$ is the only one doomed to live its life in solidarity without an inverse, because it's partner $\alpha^{-1}$ is nowhere to be seen, because it would require $\alpha$ to occur twice in the product: $$g_1g_1^{-1}g_2g_2^{-1}\cdots \alpha \cdots g_{\left(\frac{n-2}{2}-1\right)}g_{\left(\frac{n-2}{2}-1\right)}^{-1} g_{\frac{n-2}{2}}g_{\frac{n-2}{2}}^{-1}$$

(The reason the subscripts got wonky and went up to $\frac{n-2}{2}$ since two elements have the same subscript, but there's two odd one out, $e$ and $\alpha$. Don't worry too much about the subscripts here, just focus on everything having an inverse present except $\alpha$ (and $e$, but thats less important).)

So after canceling galore, we have:

$$\prod_{g\in G} g = ee \cdots \alpha \cdots ee = \alpha.$$

  • Thanks for the answer, I understand everything except the cancellation part (and the subscript part). This may sound really silly, but I'm having trouble accepting how the pairs of inverses + $\alpha$ gives you an even number of elements of $G$ (which we need by Lagranges theorem). Because you're going to have $k$-pairs of elements with their inverses, so $2k$ elements, plus $\alpha$ it gives you $2k + 1$ elements in $G$, contradicting Lagranges theorem – Perturbative Dec 04 '17 at 08:18
  • @Perturbative Whoops, you and I made the same mistake! Don't count $e$ because it's it's own inverse as well. So $|G|=n$ and ignoring $\alpha$ and $e$ leaves us with $n-2$ elements. Each of these elements has an inverse somewhere else in the set, thus giving us $\frac{n-2}{2}$ pairs. It follows that $n$ must be even for ${n-2}$ to be divisble by $2$. – Andrew Tawfeek Dec 04 '17 at 19:25
  • @Perturbative Could you clarify what about the canceling step you find odd? – Andrew Tawfeek Dec 04 '17 at 19:27
  • 1
    Yep I forgot about $e$, which was what was causing my problem, so the contradiction to Langranges theorem that I stated above wouldn't hold :). Thanks for all your help! I really appreciate it! – Perturbative Dec 05 '17 at 12:47
  • @Perturbative No worries, you and I both momentarily forgot about $e$! Glad I could help (: – Andrew Tawfeek Dec 05 '17 at 13:02