Abstract Nonsense

Crushing one theorem at a time

Review of Group Theory: Normal Subgroups and Quotient Groups (Pt. I)

Point of post: In this post we discuss several different formulations of normal subgroups and discuss how one can define a group structure on the set of cosets of such a normal subgroup.


One sees the idea of a “quotient structure” pop up frequently in mathematics. There are quotient spaces in both topology and linear algebra for example. One mods out by the set-theoretic kernel of semi-metric to turn it into a metric etc. Of course, for groups the same holds. It turns out though that like algebra there are some pretty serious theorems to do with quotient groups (the isomorphism theorems) which we shall see in subsequent posts. The problem is that unlike both linear algebra (where all the underlying group structure of the vector spaces was assumed abelian) and topology the way we “want to” define a quotient group doesn’t always work; there is a necessary and sufficient condition on a subgroup such that the quotient group structure “makes sense”. This condition is normality which, in a nutshell, says that the subgroup is invariant under conjugation (or that it’s invariant under all inner automorphisms).

Inner Automorphisms

Let G be a group, and fix g_0\in G. We define the inner automorphism induced by g_0, denoted i_{g_0}, to be the map

i_{g_0}:G\to G:g\mapsto g_0^{-1}g g_0

It’s not difficult to see that i_{g_0} is deserving of it’s name in that it really is an automorphism. Moreover, consider that for any fixed g_1,g_2\in G and variable g\in G we have that


so that the map

\Phi:G\to \text{Aut}(G):g\mapsto i_g

is a homomorphism. We call \Phi\left(G\right)\leqslant \text{Aut}(G) the inner automorphism group of G and denote it \text{Inn}(G).

Normal Subgroups

Let G be a group and N\leqslant G. We say that N is a normal subgroup of G, and denote this N\trianglelefteq G, if for every g\in G the following holds:


There are several equivalent definitions of normality. Namely:

Theorem: Let G be a group and N\leqslant G. Then, the following are equivalent

\displaystyle \begin{aligned}&\mathbf{(1)}\quad N\trianglelefteq G\\ &\mathbf{(2)}\quad gN=Ng\textit{ for all }g\in G\\ &\mathbf{(3)}\quad \textit{The set of left cosets of }N\textit{ coincides with the set of right cosets}\\ &\mathbf{(4)}\quad i_g\left(N\right)\subseteq N\textit{ for all }g\in G\\ &\mathbf{(5)}\quad \textit{There exists a group }G'\textit{ and a homomorphism }\theta:G\to G'\textit{ such that }N=\ker\theta\end{aligned}


\mathbf{(1)}\implies \mathbf{(2)}: Let g\in G be arbitrary. We know that


so, let h\in gN then h=gn for some n\in N and so g^{-1}h\in N=gNg^{-1}. Thus, there exists some n'\in N such that g^{-1}h=g^{-1}n'g and so h=n'g so that h\in Ng. The reverse inclusion is done identically.

\mathbf{(2)}\Longleftrightarrow\mathbf{(3)}: The forward implication is trivial. To do the reverse implication note that since gN is a right coset we have that gN=Nh for some h\in G. But, since g\in gN it follows that g\in Nh. But, recall that the right cosets of a subgroup partition G and since g\in Ng it follows that Nh=Ng

\mathbf{(2)}\implies\mathbf{(4)}: Let h\in g^{-1}Ng, then h=g^{-1}ng for some n\in G. In particular we have that gh=ng. Thus, gh\in Ng=gN and thus there exists some n'\in N such that gh=gn' and thus h=n'\in N.

\mathbf{(4)}\implies\mathbf{(1)}: Note that since i_g\left(N\right)\subseteq N for all g\in G we need only prove the reverse inclusion for all g\in G. It clearly suffices to do this for a fixed but arbitrary g_0\in G. To do this let n\in N. By assumption we have that i_{g_0^{-1}}\left(N\right)\subseteq N and so there exists n'\in N such that g_0 n g_0^{-1}=n' or equivalently n=g_0^{-1}n'g_0 and thus n\in i_{g_0}\left(N\right) as desired.

\mathbf{(1)}\Longleftrightarrow \mathbf{(5)}: To prove the reverse implication we note that if \theta:G\to G' is a homomorphism with N=\ker\theta then for every g\in G we have that


so that g^{-1}Ng\subseteq N. The conclusion follows by appealing to our last implication.

We stall the proof of the converse to a later point in this post. \blacksquare


We next wish to know how normal subgroups interact with homomorphisms. In particular we note that:


Theorem: Let G and G' be groups and N\trianglelefteq G and N'\trianglelefteq G'. Then, if \theta\in\text{Hom}\left(G,G'\right); \theta^{-1}\left(N'\right)\trianglelefteq G. If \theta is surjective then \theta\left(N\right)\trianglelefteq G'.

Proof: To prove the first part we let g\in G and note that

\begin{aligned}\theta\left(g^{-1}\theta^{-1}\left(N'\right)g\right) &= \theta\left(g^{-1}\right)\theta\left(\theta^{-1}\left(N'\right)\right)\\ &= \theta(g)^{-1}\theta\left(\theta^{-1}\left(N'\right)\right)\theta(g)\\ & \subseteq \theta(g)^{-1}N'\theta(g)\\ &= N\end{aligned}

and thus i_g\left(\theta^{-1}\left(N'\right)\right)\subseteq \theta^{-1}\left(N'\right) for every g\in G. It follows from our earlier characterization of normality that \theta^{-1}\left(N'\right)\trianglelefteq G.


Suppose now that \theta is surjective. We note then that for any g'\in G' we have that g'=\theta(g) for some g\in G. Thus,


from where normality follows. \blacksquare



1.  Lang, Serge. Undergraduate Algebra. 3rd. ed. Springer, 2010. Print.

2. Dummit, David Steven., and Richard M. Foote. Abstract Algebra. Hoboken, NJ: Wiley, 200


December 31, 2010 - Posted by | Algebra, Group Theory, Uncategorized | , , ,


  1. […] Point of post: This post is a continuation of this one. […]

    Pingback by Review of Group Theory: Normal Subgroups and Quotient Groups (Pt. II) « Abstract Nonsense | January 1, 2011 | Reply

  2. […] We merely recall that for any the inner automorphism is an automorphism and so in particular and . It follows from our assumption that […]

    Pingback by Review of Group Theory: Interesting Consequence of the First Isomorphism Theorem « Abstract Nonsense | January 2, 2011 | Reply

  3. […] Proof: Merely note that the canonical projection […]

    Pingback by Review of Group Theory: The Third Isomorphism Theorem « Abstract Nonsense | January 2, 2011 | Reply

  4. […] the theorem we merely note that if then (since is an epimorphism and we’ve already proven this). Conversely, suppose that . Then, since using the same techniques above we know that  and thus […]

    Pingback by Review of Group Theory: The Fourth Isomorphism Theorem (The Correspondence Theorem) « Abstract Nonsense | January 4, 2011 | Reply

  5. […] by the set is called the commutator subgroup of and is denoted . Our first claim is that is a normal subgroup of . Indeed it clearly suffices to check that for every . Recall the inner automorphism and the […]

    Pingback by Review of Group Theory: The Commutator Subgroup and the Abelianization of a Group « Abstract Nonsense | February 27, 2011 | Reply

  6. […] lies in the codomain) since if then . Evidently then is injective (being the restriction of the inner automorphism ) and thus it suffices to prove that is surjective. To see this suppose that is an element of the […]

    Pingback by Representation Theory: The Square Root Function and its Relation to Irreducible Characters « Abstract Nonsense | March 24, 2011 | Reply

  7. […] said, there is an interesting phenomenon concerning subrings. Namely, while it turned out that normal subgroups were nicer than general subgroups they weren’t the only thing that was important. For ring […]

    Pingback by Subrings « Abstract Nonsense | June 15, 2011 | Reply

  8. […] out by any old subset, or even any old subring, of a ring. It turns out, much the same as the case for groups that what defines these special subrings is that they are the kernels of ring homomorphisms. But, […]

    Pingback by Definition and Basics of Ideals « Abstract Nonsense | June 21, 2011 | Reply

  9. […] to play the role for rings that normal subgroups played for groups it seems natural to look at the facts that held there. Things such as the preimage of an ideal is an ideal, etc. We also explore some of the ramifcations […]

    Pingback by Ideals and Homomorphisms « Abstract Nonsense | June 24, 2011 | Reply

  10. […] generalization of  the idea of a quotient space for vector spaces. Namely, we shall just take the quotient group and multiply scalars representative-wise. As in all algebraic theories we shall occupy a central […]

    Pingback by Quotient Modules « Abstract Nonsense | November 7, 2011 | Reply

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: