Normal subgroup

Revision as of 20:31, 7 May 2008 by Boy Soprano II (talk | contribs) (start of an article; I'll be back to finish it)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

A normal subgroup ${\rm H}$ of a group ${\rm G}$ is a subgroup of ${\rm G}$ for which the relation "$xy^{-1} \in {\rm H}$" of $x$ and $y$ is compatible with the law of composition on ${\rm G}$, which in this article is written multiplicatively. The quotient group of ${\rm G}$ under this relation is often denoted ${\rm G/H}$ (said, "${\rm G}$ mod ${\rm H}$"). (Hence the notation $\mathbb{Z}/n\mathbb{Z}$ for the integers mod $n$.)

Description

Note that the relation $xy^{-1} \in {\rm H}$ is compatible with right multiplication for any subgroup ${\rm H}$: for any $a \in {\rm G}$, \[(xa)(ya)^{-1} = (xa)(a^{-1}y^{-1}) = xy^{-1} \in {\rm H}.\] On the other hand, if ${\rm H}$ is normal, then the relation must be compatible with left multiplication by any $a\in {\rm G}$. This is true if and only $xy^{-1} \in {\rm H}$ implies \[axy^{-1}a^{-1} = (ax)(ay)^{-1} \in {\rm H} .\] Since any element of ${\rm H}$ can be expressed as $xy^{-1}$, the statement "${\rm H}$ is normal in ${\rm G}$" is equivalent to the following statement:

  • For all $a\in {\rm G}$ and $g\in {\rm H}$, $aga^{-1} \in G$,

which is equivalent to both of the following statements:

  • For all $a \in {\rm G}$, $a{\rm G}a^{-1} \subseteq {\rm G}$;
  • For all $a \in {\rm G}$, $a {\rm G \subseteq G}a$.

By symmetry, the last condition can be rewritten thus:

  • For all $a \in {\rm G}$, $a {\rm G = G} a$.

Examples

In an Abelian group, every subgroup is a normal subgroup.

Every group is a normal subgroup of itself. Similarly, the trivial group is a subgroup of every group.

Consider the smallest nonabelian group, $S_3$ (the symmetric group on three elements); call its generators $x$ and $y$, with $x^3 = y^2 = (xy)^2 =e$, the identity. It has two nontrivial subgroups, the one generated by $x$ (isomorphic to $\mathbb{Z}/3\mathbb{Z}$ and the one generated by $y$ (isomorphic to $\mathbb{Z}/2\mathbb{Z}$). Of these, the second is normal but the first is not.

If ${\rm G}$ and ${\rm G'}$ are groups, and $f: {\rm G \to G'}$ is a homomorphism of groups, then the inverse image of the identity of ${\rm G'}$ under $f$, called the kernel of $f$ and denoted $\text{Ker}(f)$, is a normal subgroup of ${\rm G}$ (see the proof of theorem 1 below). In fact, this is a characterization of normal subgroups, for if ${\rm H}$ is a normal subgroup of ${\rm G}$, the kernel of the canonical homomorphism $f:{\rm G \to G/H}$ is ${\rm H}$.

Note that if ${\rm H'}$ is a normal subgroup of ${\rm H}$ and ${\rm H}$ is a normal subgroup of ${\rm G}$, ${\rm H'}$ is not necessarily a normal subgroup of ${\rm G}$.

Group homomorphism theorems

Theorem 1. An equivalence relation $\mathcal{R}(x,y)$ on elements of a group ${\rm G}$ is compatible with the group law on ${\rm G}$ if and only if it is equivalent to a relation of the form $xy^{-1} \in {\rm H}$, for some normal subgroup ${\rm H}$ of ${\rm G}$.

Proof. One direction of the theorem follows from our definition, so we prove the other, namely, that any relation $\mathcal{R}(x,y)$ compatible with the group law on ${\rm G}$ is of the form $xy^{-1} \in {\rm H}$, for a normal subgroup ${\rm H}$.

To this end, let ${\rm H}$ be the set of elements equivalent to the identity, $e$, under $\mathcal{R}$. Evidently, if $x \equiv y \pmod{\mathcal{R}}$, then $xy^{-1} \equiv e \pmod{\mathcal{R}}$, so $xy^{-1} \in {\rm H}$; the converse holds as well, so $\mathcal{R}(x,y)$ is equivalent to the statement "$xy^{-1} \in {\rm H}$". Also, for any $x,y \in {\rm H}$, \[xy \equiv ee \equiv e \pmod{\mathcal{R}},\] so $xy \in {\rm H}$. Thus ${\rm H}$ is closed under the group law on ${\rm G}$, so ${\rm H}$ is a subgroup of ${\rm G}$. Then by definition, ${\rm H}$ is a normal subgroup of ${\rm G}$. $\blacksquare$

See also