1

Problem:

Let $G$ be a finitely generated group, and $H \subset G$ a subgroup of finite index. Show that $H$ is finitely generated.

My work:

I wrote this proof:

Consider any finitely generated group $G = \langle x_1, x_2, \ldots, x_n \rangle$ and a subgroup $H$ with a finite index. For some $g_i \in G - H$, $G/H = \{H, g_1 H, \ldots, g_m H\}$ so consider the set $S = \{1, g_1, \ldots, g_m\}$, since $G/H$ is a partition of $G$, for all $x \in G$ there is some $h \in H$ and some $s \in S$ such that $x = sh$. Since $G$ is finitely generated any element of $H$ can be expressed as $x_{i_1}^{\alpha_1} \ldots x_{i_k}^{\alpha_k} = (s_{i_1}h_{i_1})^{\alpha_1} \ldots (s_{i_k}h_{i_k})^{\alpha_k}$

But, as you can see, it is incomplete since I was wondering if those were the generators of $H$ so I came up with a concrete example to see what happened.

I thought of the group $D_4$, consider the $90^{\circ}$ rotation $r$ and any reflection $h$, $D_4 = \langle r, h \rangle$ and the subgroup $G = \{1, h, r^2 h, r^2\} \subset D_4$. We can check that $D_4/G = \{G, rG\}$ so consider the set $A = \{1, r\}$. Since $D_4/G$ is a partition of $D_4$ we know that for all elements $d \in D_4$, for some $a \in A$ and some $g \in G$, $d = ag$ and thus $D_4 = \langle r, h \rangle = \langle a_1 g_1, a_2 g_2 \rangle = \langle r 1, 1h \rangle$. After finishing this example I can't see why it is useful knowing all of the above because I can't see a way to conclude that $G = \langle r^2h, h\rangle$ is indeed finitely generated.

In my proof I took that approach because I was given the following hint:

Choose a finite subset $S \subset G$ containing one representative of each coset of $H$, so every element of $G$ is the product of an element of $S$ and an element of $H$. Given a word in the generators of $G$ and their inverses, how do you rewrite it as the product of an element of $S$ and an element of $H$?

I would appreaciate any hints on how to continue my proof.

I already saw subgroups of finitely generated groups with a finite index but I don't understand how to conclude that $h_{i_1} \ldots h_{i_k}$ are the generators of $H$.

Thanks in advance.

Torrente
  • 931
  • 1
  • 3
  • 14
  • 1
    As your example is a finite group, "finitely generated" is a bit silly: every finite group is finitely generated, since it is "generated" by all its elements. If you want to try to see what is going on, you should really start with an infinite, but finitely generated, group. – Arturo Magidin Aug 15 '22 at 16:56
  • @ArturoMagidin I already saw that post before asking however I wasn't able to follow their advice so I wanted to know how to continue my proof. – Torrente Aug 15 '22 at 17:07
  • Context that would have been helpful, as in "I saw the duplicate here but I am stuck at this point of the proof..." Or were we supposed to guess that? – Arturo Magidin Aug 15 '22 at 17:14
  • What don't you understand? The answer I linked to gives the algorithm." Don't just say "I don't understand." What do you not understand about the argument presented? – Arturo Magidin Aug 15 '22 at 17:37
  • @ArturoMagidin The first hint given is the same as the one I was given, but after applying it I get stuck at the same point where I stop my proof attempt. In your comment you say that any element $h \in H$ can be written as a product of $h_{i_1} t_{i_1} \ldots$ and $h_{i_k} t_{i_k}$ so that makes me think that $H$ is generated by $h_{i_1} t_{i_1} \ldots$ and $h_{i_k} t_{i_k}$ but Nuno states that it is generated by $h_{i_1} \ldots$ and $h_{i_k}$ instead so I don't know how to arrive to Nuno's conclusion. – Torrente Aug 15 '22 at 17:45
  • 1
    Write $h=x_{i_1}\cdots x_{i_m}$. Replace $x_{i_1}$ by $h_{i_1}t_{i_1}$, so now you have $h=h_{i_1}(t_{i_1}x_{i_2})\cdots x_{i_m}$. Now write $t_{i_1}x_{i_2}$ as $h_{i_2}t_{i_2}$, so you have $h=h_{i_1}h_{i_2}(t_{i_2}x_{i_3})\cdots x_{i_m}$. Continue until you end up with $h=h_{i_1}\cdots h_{i_m}t_{i_m}$. But since the left hand side is in $H$, and the right hand side is in $Ht_{i_m}$, then we must have that $t_{i_m}=e$. – Arturo Magidin Aug 15 '22 at 17:51
  • @ArturoMagidin that solves my question, I appreciate the help. – Torrente Aug 15 '22 at 17:53

0 Answers0