2

I read a statement in my textbook, which was taken as a premise, that for some element $g$ of a finite group $G$, that the order of the element $g$ is the same as the cyclic subgroup, $<g>$.

I think I might be confusing the order of an element with the order of a group. If I'm not mistaken, $<g>$ is equal to the powers of $g$. Since $G$ is finite, we can write these powers as $g^0, g_1, \ldots, g_m$ where $g$ has order $m$. (More commonly, I think we'd just write the identity element as $1$.) The reason it has order $m$, I believe, is because $g^m$ is equal to the identity, i.e., if we apply $g$ to itself $m$ times we get back $1$, so we need not continue to write higher, or lower, negative powers of $g$ because we'll simply be repeating ourselves.

Is the argument here, then, that $<g>$ has $m$ elements by definition, provided $m$ is the order of $g$? I think this question might be a bit trivial, but it would be great to get these concepts down.

Thanks in advance.

2 Answers2

5

It's a question of how you define the order of an element.

Definition 1 The order of the element $g$ in the finite group $G$ is the least positive integer $m$ such that $g^m=1$.

Such a least integer exists, because there surely are $p$ and $q$ distinct positive integers such that $g^p=g^q$, by the pigeonhole principle. We can assume $p>q$, so that $g^{p-q}=1$. Then a positive integer $n$ with the property $g^n=1$ exists and therefore also the least one exists.

Now we can show that the elements $g^0,g^1,\dots,g^{m-1}$ are pairwise distinct (easy). Moreover if $n$ is any integer, we can write $n=mq+r$, with $0\le r<m$ and $g^n=g^r$ is one of the previously listed elements. Hence $\langle g\rangle=\{g^0=1,g^1=g,\dots,g^{m-1}\}$ has indeed $m$ elements.

Definition 2 The order of the element $g$ in the finite group $G$ is the cardinality of $\langle g\rangle=\{g^n:n\in\mathbb{Z}\}$.

Consider the homomorphism $\varphi_g\colon\mathbb{Z}\to G$ defined by $\varphi_g(n)=g^n$. The image of $\varphi_g$ is, by definition, $\langle g\rangle$. By the homomorphism theorem, $$ \langle g\rangle\cong \mathbb{Z}/\ker\varphi_g $$ Since $\ker\varphi_g$ is a subgroup of $\mathbb{Z}$, it is of the form $\ker\varphi_g=m\mathbb{Z}$ for a unique $m>0$ (it cannot be $m=0$, because otherwise $\mathbb{Z}/\ker\varphi_g$ would be infinite. Then $$ \langle g\rangle\cong \mathbb{Z}/\ker\varphi_g=\mathbb{Z}/m\mathbb{Z} $$ and so $m$ equals the order of $g$ (as defined here). Since $m$ is the least positive integer in $m\mathbb{Z}=\ker\varphi_g$, it is the least positive integer such that $g^m=1$.

Bonus for definition 2: the fact that the order of $g$ divides $|G|$ is a consequence of Lagrange’s theorem.

egreg
  • 238,574
0

By definition the order of $<g>$ is the cardinality of this subgroup. This subgroup consists of all elements of the form $g^k$, for $k\in \mathbb{Z}$.

Now, because $g$ is of order $m$, we see that if $k=l \, \operatorname{mod}(m)$, then $g^k=g^l$. It follows that every element of the form $g^k$ equals an element of the form $g^l$ for some $l\in \{ 0,1,\ldots ,m-1 \}$.

Moreover, it is also true that for any two integers $l, l' \in \{ 0,1,\ldots ,m-1 \}$, we have $g^l=g^{l'}$ if and only if $l=l'$. Indeed, otherwise, assuming that $l< l'$ without loss of generality, we would have $g^{l'-l}=1$ with $0< l-l' < m$, which comes as a contradiction of the definition of the order of an element (as the least positive integer $m$ such that $g^m=1$).

Hence, we deduce that the cardinality (order) of $<g>$ really is $m$, the order of the element $g$.

NB: To justify it in a little more elaborate way, we can argue that we have a group morphism $\mathbb{Z}\rightarrow <g>$ sending an integer $k$ to $g^k$, which is surjective and has kernel $m\mathbb{Z}$. Hence, we have an isomorphism $<g>\simeq \mathbb{Z}/m\mathbb{Z}$, proving that this group has cardinality (order) $m$.

Suzet
  • 5,482
  • 12
  • 35
  • This was very helpful, thank you. One more question, if you wouldn't mind: are there any additional conditions we require to posit the existence of a cyclic subgroup? Or does this hold for any element, $g \in G$ for an arbitrary group, $G$? –  Jul 11 '18 at 14:23
  • By definition, a group is cyclic if and only if it is generated by one element. Hence, for any $g\in G$,$$ is cyclic. If you are interested in conditions about the existence of a finite cyclic subgroup (other than the trivial one ${ e }$ where $e$ is the unit element), well I can not say much. There are groups, like $\mathbb{Z}$ with the addition, where there is no non trivial finite cyclic subgroups (the non trivial cyclic subgroup have the form $n\mathbb{Z}$ for some nonzero $n$. You may be interested, in the case of finitely generated abelian groups, in the classification theorem [...] – Suzet Jul 11 '18 at 14:30
  • that you can see here https://en.wikipedia.org/wiki/Finitely_generated_abelian_group . Namely, we are able to classify every finitely generated abelian groups up to isomorphisms. With such a classification in mind, one may be able to understand more about the existence of finite cyclic subgroups. (Also, note that if you are only considering finite groups, then you are not concerned at all with what I just wrote here). – Suzet Jul 11 '18 at 14:31