How can we find all the normal subgroups?

  • Context: MHB 
  • Thread starter Thread starter mathmari
  • Start date Start date
  • Tags Tags
    Normal
Click For Summary

Discussion Overview

The discussion revolves around finding all normal subgroups of the dihedral group $D_n$. Participants explore various approaches, including enumeration of subgroups, analysis of cyclic subgroups, and the implications of elements' forms within the group. The conversation includes theoretical considerations and specific subgroup structures.

Discussion Character

  • Exploratory
  • Technical explanation
  • Debate/contested
  • Mathematical reasoning

Main Points Raised

  • Some participants suggest enumerating all subgroups to identify normal ones, considering elements like $as^i$ and $s^j$.
  • It is proposed that finding all cyclic subgroups is a useful first step in the analysis.
  • Participants note that the structure of dihedral groups involves reflections and rotations, which may lead to different types of subgroups.
  • There is a discussion about the normality of subgroups generated by elements of the form $s^k$ and the implications of having elements like $s^ka$ in a normal subgroup.
  • Some participants argue that if a subgroup contains an element of the form $s^ka$, it must also contain $s^2$ under certain conditions.
  • Questions arise regarding the conditions under which certain elements must belong to a subgroup and how to conclude the structure of these subgroups.
  • There is mention of the importance of whether $n$ is even or odd, as this affects the number and type of normal subgroups.
  • Participants explore the implications of the factor group $D_n/K$ and how it relates to the normal subgroups of $D_n$.

Areas of Agreement / Disagreement

Participants express various viewpoints on how to approach the problem, and there is no clear consensus on the methods or conclusions regarding the normal subgroups of $D_n$. The discussion remains unresolved with multiple competing views and hypotheses presented.

Contextual Notes

Some limitations include the dependence on the parity of $n$ and the need for careful consideration of subgroup structures based on the elements chosen. The discussion also highlights the complexity of determining normality in the context of dihedral groups.

mathmari
Gold Member
MHB
Messages
4,984
Reaction score
7
Hey! :o

I want to find all the normal subgroups of $D_n$.

We have that $$D_n=\langle a, s \mid a^2=1=s^n, asa=s^{-1}\rangle$$

We have that $K$ is a normal subgroup of $D_n$ iff $$gkg^{-1}\in K, \forall g\in D_n \text{ and } \forall k\in K$$

Could you give me some hints how we could find all these subgroups? (Wondering)
 
Physics news on Phys.org
mathmari said:
Hey! :o

I want to find all the normal subgroups of $D_n$.

We have that $$D_n=\langle a, s \mid a^2=1=s^n, asa=s^{-1}\rangle$$

We have that $K$ is a normal subgroup of $D_n$ iff $$gkg^{-1}\in K, \forall g\in D_n \text{ and } \forall k\in K$$

Could you give me some hints how we could find all these subgroups? (Wondering)

Hey mathmari! (Smile)

How about enumerating all subgroups and see where that brings us? (Wondering)
To do so, we need to pick 1 or 2 (or more) elements and see which group it generates, and then see what it takes to make it normal.
 
I like Serena said:
How about enumerating all subgroups and see where that brings us? (Wondering)
To do so, we need to pick 1 or 2 (or more) elements and see which group it generates, and then see what it takes to make it normal.

Do we maybe pick the elements $as^i$ and $s^j$ for $0\leq i,j <n$ ? (Wondering)
 
Try finding all the cyclic subgroups, first.

With dihedral groups it is helpful to remember there are two types of elements: reflections, and rotations. The reflections have an "$a$" in them.

For subgroups generated by two elements, we have three main kinds of subgroups:

1. Generated by two rotations. This type should be trivial to analyze.
2. Generated by a reflection, and a rotation. This should be easy once you see what happens:

Hint: $s^ka = as^{-k}$, for all $k \in \Bbb N$. Use induction to prove this.

3. Generated by two distinct reflections. This one is a bit tricky.

General warning: it matters whether or not $n$ is even or odd (the even $n$'s produce more subgroups). The prime factorization of $n$ also matters.
 
Deveno said:
Try finding all the cyclic subgroups, first.

Is $\langle s^i\rangle$ for $i\mid n$ a subgroup of $D_n$ ? (Wondering)

If it is, do we have to check if an element of the form $s^ka$ belongs to this subgroup or if it belongs to or generate an other subgroup? (Wondering)
Deveno said:
For subgroups generated by two elements, we have three main kinds of subgroups:

1. Generated by two rotations. This type should be trivial to analyze.
2. Generated by a reflection, and a rotation. This should be easy once you see what happens:
3. Generated by two distinct reflections. This one is a bit tricky.

1. Is it generated by a "$s$" ?
2. Is this subgroup of the form $\langle a, s\rangle$ ?
3. Is it generated by an "$a$" ?

(Wondering)
 
I think you're working too hard. As you showed previously, any subgroup of $<s>$ is normal since $<s>$ is normal. So the normal subgroups are the identity, $D_n$, any subgroup of $<s>$ and maybe some more. So suppose $H$ is a proper normal subgroup that is not a subgroup of $<s>$. Then $H$ contains an element of the form $s^ka$ for some $k$. Compute the conjugate $$s^{-1}s^kas=s^ks^{-1}as=s^ka(as^{-1}a)s=s^kas^2\in H$$
So $s^2\in H$. If $n$ is odd, $K=<s^2>=<s>\subset H$ and so $H=D_n$. Suppose $n$ is even. Consider the factor group $D_n/K$. Easily then $H/K$ is one of the 3 proper normal subgroups of $D_n/K$. Thus $H=<s^2,a>$ or $H=<s^2,sa>$, each of these is isomorphic to $D_m$ where $m=n/2$.
 
Last edited:
johng said:
So suppose $H$ is a proper normal subgroup that is not a subgroup of $<s>$. Then $H$ contains an element of the form $s^ka$ for some $k$. Compute the conjugate $$s^{-1}s^kas=s^ks^{-1}as=s^ka(as^{-1}a)s=s^kas^2\in H$$
So $s^2\in H$.

Why does this stand? Why does $H$ contain an element of the form $s^ka$ ? (Wondering),

How do we conclude that $s^kas^2\in H$ ? (Wondering)
 
$D_n=\{s^k\,:\,0\leq k<n\}\cup\{s^ka\,:\,0\leq k<n\}$. So if $H$ is not a subgroup (subset) of $<s>$, H contains an element of the form $s^ka$.

By definition of normality of $H$, any conjugate of an element of $H$ is also an element of $H$. So $s^kas^2\in H$ implies $s^2=(s^ka)^{-1}s^kas^2\in H$
 
Let $H$ be a normal subgroup of $D_n$.

  1. If all the elements of $H$ are of the form $s^k$ then $H=\langle s^d\rangle$ for some $d$ that divides $n$.

    $\langle s^d\rangle$ is indeed a normal subgroup of $D_n$:
    It suffices to show that $gs^dg^{-1}\in \langle s^d\rangle, \forall g\in D_n$.
    • if $g=s^k$, then $s^ks^ds^{-k}=s^{k+d-k}=s^d\in \langle s^d\rangle$
    • if $g=as^k$, then $as^ks^d(as^k)^{-1}=as^ks^ds^{-k}a^{-1}=as^ks^ds^{-k}a=as^{k+d-k}a=as^da=a\underset{d \text{ times } "s"}{\underbrace{saasaas\dots as}a}=(asa)^d=(s^{-1})^d=s^{-d}\in \langle s^d\rangle$
  2. If $H$ contains an element of the form $s^ka$, for some $k$, then since it is a normal subgroup of $D_n$, it must hold that $$s^{-1}s^kas\in H \Rightarrow s^{-1+k}as\in \Rightarrow s^ks^{-1}as\in H \Rightarrow s^kas^2\in H$$

    So $s^2=(s^ka)^{-1}s^kas^2\in H$.

    But how can we find exactly which this $H$ is? (Wondering)
 
  • #10
I'm not exactly sure what your question is. Maybe this is what you're asking:
Assume $s^ka\in H$. If $k$ is even, say $k=2j$, then since $<s^2>\subseteq H$, $a=s^{-2j}s^{2j}a\in H$ and so $H=<s^2,a>$. Similarly, if $k$ is odd, $H=<s^2,sa>$.
 
  • #11
johng said:
Assume $s^ka\in H$. If $k$ is even, say $k=2j$, then since $<s^2>\subseteq H$, $a=s^{-2j}s^{2j}a\in H$ and so $H=<s^2,a>$.

Why does it stand that $<s^2>\subseteq H$ ? (Wondering)

Having shown that $s^ka\in H$ and $a\in H$, why does it imply that $H=<s^2,a>$ ? (Wondering)

How do we know that these two elements generate a subgroup? (Wondering)

johng said:
Similarly, if $k$ is odd, $H=<s^2,sa>$.

Assume $s^ka\in H$. When $k$ is odd, say $k=2j+1$, does it stand that $<s^2>\subseteq H$ ? (Wondering)
If it holds, then $sa=ss^{-2j-1}s^{2j+1}a=s^{-2j}s^{2j+1}a\in H$ and so $H=<s^2,sa>$, right? (Wondering)
 
  • #12
$$D_n=\langle a,s\mid s^n=1=a^2, asa=s^{-1}\rangle$$

Let $N=\langle s\rangle=\{s, s^2, \dots, s^{n-1}, s^n=1\}$.
We have that $[D_n:N]=\frac{|D_n|}{|N|}=\frac{2n}{n}=2$.
So $N$ is a normal subgroup of $D_n$. Therefore, each subgroup of $N$ is a normal subgroup of $D_n$. (As shown in the thread http://mathhelpboards.com/linear-abstract-algebra-14/normal-subgroup-g-17852.html )

Let $T\neq N$, be a normal subgroup of $D_n$.

So, $\exists \ as^k$, for some $k=1, 2, \dots , n$ with $$s^{-i}(as^k)s^i\in T , \ \forall i=1, 2, \dots , n$$

Since $asa=s^{-1} \Rightarrow as=s^{-1}a \Rightarrow s^{-i}a=s^{-i+1}(s^{-1}a)=s^{-i+1}as=s^{-i+2}(s^{-1}a)s=s^{-i+2}as^2=\dots =as^i$, we have that $$s^{-i}as^ks^i\in T , \ \forall i=1, 2, \dots , n \\ \Rightarrow as^is^ks^i\in T , \ \forall i=1, 2, \dots , n \\ \Rightarrow as^{k+2i}\in T , \ \forall i=1, 2, \dots , n$$

We also have that $$a(as^k)a\in T \Rightarrow s^ka \in T\Rightarrow as^{-k}\in T$$

Then the product of these is also an element of $T$, $as^{-k}as^{k+2i}\in T \Rightarrow aas^ks^{k+2i}=s^{2k+2i}=(s^2)^{k+i}\in T$.

Now we have to take cases if $n$ is odd or even, or not? (Wondering)

If $k$ is even, say $k=2j$, we have that $as^{k+2i}\in T \Rightarrow as^{2j+2i}\in T\Rightarrow a(s^2)^{j+i}\in T\Rightarrow a\in T$.
If $k$ is odd, say $k=2j+1$, we have that $as^{k+2i}\in T \Rightarrow as^{2j+1+2i}\in T\Rightarrow as(s^2)^{j+i}\in T\Rightarrow as\in T$.
Does these two cases hold in both cases, when $n$ is odd and even? (Wondering)
 
Last edited by a moderator:

Similar threads

Replies
7
Views
2K
  • · Replies 12 ·
Replies
12
Views
4K
  • · Replies 2 ·
Replies
2
Views
2K
  • · Replies 8 ·
Replies
8
Views
2K
Replies
1
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 5 ·
Replies
5
Views
2K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 3 ·
Replies
3
Views
3K
  • · Replies 9 ·
Replies
9
Views
12K