# Is $G/pG$ is a $p$-group?

Jack is trying to prove:

Let $G$ be an abelian group, and $n\in\Bbb Z$. Denote $nG = \{ng | g\in G\}$.

(1) Show that $nG$ is a subgroup in $G$.

(2) Show that if $G$ is a finitely generated abelian group, and $p$ is prime, then $G/pG$ is a $p$-group (a group whose order is a power of $p$).

I think $G/pG$ is a $p$-group because it is a direct sum of cyclic groups of order $p$. But I cannot give a detailed proof.

-
How is the operation of $n \in Z$ on $g \in G$ defined? –  Herp Derpington Nov 8 '12 at 0:00
$$\forall\,g\in G\;\;,\;pg\in pG\Longrightarrow p(g+pG)=pG\Longrightarrow$$ the element $\,p(g+pG)\,$ is zero in the quotient $\,G/pG\,$ and from here that all the elements in this quotient have order a power of p, which is precisely the definition of p-group, no matter if it is finitely generated or not. –  DonAntonio Nov 8 '12 at 2:32
@HerpDerpington: I suspect $G$ is taken to be an additive group, so that $ng$ is simply adding up $n$ terms $g$ for $n>0$ and adding up $n$ terms $-g$ for $n<0$. –  Cameron Buie Nov 8 '12 at 8:37
@DonAntonio: Why not make that an answer? –  Cameron Buie Nov 9 '12 at 6:53
@CameronBuie, I will. It's just that there were already several answers... –  DonAntonio Nov 9 '12 at 9:07

Following my comment:

$$∀g∈G,pg∈pG⟹p(g+pG)=pG⟹$$ the element $\,p(g+pG)\,$ is zero in the quotient $\,G/pG\,$ and from here that all the elements in this quotient have order a power of $\,p\,$ , which is precisely the definition of $\,p$-group, no matter if it is finitely generated or not.

-

$G/pG$ is a direct sum of a finite number of cyclic groups by the fundamental theorem of finitely generated abelian groups. Since every non-zero element of $G/pG$ is of order $p$. It is a direct sum of a finite number of cyclic groups of order $p$.

-
Three answers, Makoto? Why don't you just combine them into one, and just give them as three ways to see it? –  Cameron Buie Nov 8 '12 at 0:22
It looks like you might be trying to get more reputation by posting several answers. It's off-putting. I personally liked both your first and last answer, but I don't want to reinforce behavior that I know isn't acceptable here, so I didn't upvote either of them. If you merge your answers, I'll gladly upvote, and I suspect that others will do the same. I recommend merging this answer and the vector space answer into the last one, so that you don't lose the comments there. –  Cameron Buie Nov 8 '12 at 8:14
@CameronBuie Since I'm not rep hungry, I decline your proposal. Please answer what's wrong with posting several answers. –  Makoto Kato Nov 8 '12 at 8:26
@Makoto: I strongly disagree in this case. You gave three two-liner answers. You can easily say: "I will give three different ways to look at the problem." And divide your answer using the convenient formatting afforded to you by MarkDown. –  Willie Wong Nov 8 '12 at 9:01
@All There is now an associated meta question here. Please take any further meta discussion there. –  Bill Dubuque Nov 8 '12 at 16:51

$G/pG$ can be regarded as a finite dimensional vector space over $\mathbb{Z}/p\mathbb{Z}$. Suppose its dimension is $n$. Then $|G/pG| = p^n$.

-
Since $G/pG$ is a finitely generated torsion group, it is finite. Let $q$ be a prime number which divides $|G/pG|$. Then it has an element of order $q$ by the theorem of Cauchy. Hence $q = p$. Hence $G/pG$ is a $p$-group.
Because $G$ is finitely generated and $G/pG$ is a torsion group(i.e. every element has finite order). –  Makoto Kato Nov 8 '12 at 0:21
Consider generators $x_1, \dots, x_n$ of $G/pG$. –  Makoto Kato Nov 8 '12 at 0:27