Skip to main content

Section 3.1 Cosets and Lagrange’s Theorem

Subsection Cosets

“If I had inherited a fortune I should probably not have cast my lot with mathematics.”
―Joseph-Louis Lagrange

Definition 3.1. Cosets.

For any \(H\leq G\) and any \(g\in G\) let
\begin{equation*} gH=\{gh|h\in H\}\;\text{and}\; Hg=\{hg|h\in H\} \end{equation*}
be defined as a left coset and right coset of \(H\) in \(G\text{,}\) respectively.

Exercise 3.2. Abelian Cosets.

If \(G\) is an abelian then \(gH=Hg\) for all \(g\in G\text{.}\)

Proof.

\((1) \Rightarrow (2):\) if \(x\) and \(y\) belong to the same left coset \(gH\) of \(H\) in \(G\) then \(x=gh'\) and \(y=gh''\) for some \(h',h''\in H\text{,}\) so \(g=y(h'')^{-1}\) and therefore \(x=y(h'')^{-1}h'=yh\) where \(h=(h'')^{-1}h'\in H\text{.}\)
\((2) \iff (3):\) \(x=yh\) for some \(h \in H\) \(\iff\) \(y = xh^{-1}\) and \(h^{-1} \in H\text{.}\)
\((2) \iff (4):\) \(x=yh\) for some \(h \in H\) \(\iff\) \(y^{-1} x=h\in H\text{.}\)
\((4) \iff (5):\) \(y^{-1} x\in H \iff (y^{-1} x)^{-1}\in H \iff x^{-1}y\in H\text{.}\)
\((2) \Rightarrow (6):\) Suppose \(x = yh\) for some \(h \in H\text{,}\) then by \(2. \Rightarrow 3.\) we also have \(y = xh''\) for some \(h'' \in H\text{.}\) Then we have
\begin{equation*} xH=\{xh'\mid h'\in H\}=\{yhh'\mid h'\in H\}\subset yH \text{ and} \end{equation*}
\begin{equation*} yH=\{yh'\mid h'\in H\}=\{xh''h'\mid h'\in H\}\subset xH, \end{equation*}
thus \(xH=yH\text{.}\)
\((6) \Rightarrow (1):\) Since \(e_G=e_H\in H\text{,}\) we have \(x=xe_G\in xH\) and \(y=ye_G\in yH\text{.}\) If \(xH=yH\) then, \(x\) and \(y\) belong to the same left coset.

Proof.

Let me prove the assertions for right cosets. Every element \(g\) of \(G\) belongs to at least one right coset, since \(g \in Hg\) (since \(e \in H\)). We need to show any two cosets are either identical or disjoint: if \(Hx\) and \(Hy\) share an element, then it follows from \((1)\Rightarrow (6)\) of Lemma 3.4 that \(Hx=Hy\text{.}\) This proves that the right cosets partition \(G\text{.}\) To see that all right cosets have the same cardinality as \(H\text{,}\) define a function
\begin{equation*} \rho: H \to Hg \end{equation*}
by \(\rho(h) = hg\text{.}\) We see \(\rho\) is onto and if \(\rho(h) = \rho(h')\) then \(hg = h'g\) and hence \(h = h'\text{,}\) so that \(\rho\) is also one-to-one.

Definition 3.9. Index.

In finite groups, the common number of left or right cosets of a subgroup \(H\) in a group \(G\) is denoted as \([G:H]\) and is called the index of \(H\) in \(G\text{.}\)

Example 3.10. Cosets in \(D_{2n}\).

For \(G =D_{2n}\) and \(H = \langle s \rangle = \{e,s\}\text{,}\) the left cosets of \(H\) in \(G\) are
\begin{equation*} \{e, s\}, \{r, rs\}, \{r^2, r^2s\}, \cdots , \{r^{n-1}, r^{n-1}s\} \end{equation*}
and the right cosets are
\begin{equation*} \{e, s\}, \{r, r^{-1}s\}, \{r^2, r^{-2}s\}, \cdots , \{r^{n-1}, r^{-n+1}s\}. \end{equation*}
Note that these lists are not the same, but they do have the same length. We have \(|G| = 2n\text{,}\) \(|H| = 2\) and \([G:H] = n\text{.}\)
Now that we have proved Lagrange’s Theorem, a host of other results now become available to us.
The notion of index is most useful when it is finite, but note that this does not require that \(G\) to be finite.

Example 3.12. Finite Index in Infinite Group.

Let \(G = \mathbb{Z}\) and \(H = 7\mathbb{Z}\text{.}\) Then \([G:H] = 7\) since the cosets are \(H, 1 + H, 2 + H, \dots, 6 + H\text{.}\) (Since \(G\) is abelian, left and right cosets are automatically the same by Exercise 3.2)

Exercise 3.13. Number of Left and Right Cosets (Ininite Version).

Show that even if \(G\) is not finite the number of left and right cosets of a subgroup \(H\leq G\) is still the same.
Hint.
Consider the map \(gH\mapsto Hg^{-1}\) and show it’s a bijection. Why is the inverse needed?
This seems like it might be important.
Here are some more corollaries to Lagrange’s Theorem.

Exercise 3.15.

Let \(G\) be a group of composite order. Then \(G\) contains a non-trivial proper subgroup.

Exercise 3.16. Fermat’s Little Theorem.

Prove Fermat’s Little Theorem: for every integer \(a\) and every prime \(p\text{,}\) \(a^p\mod p=a\mod p\text{.}\)
If you can believe it, there’s actually another important corollary to Lagrange’s Theorem, but we’ll cover it when its more relevant. Wouldn’t want to spoil all the fun at once, right?
 1 
To spoil all the fun at once, see: Theorem 6.3, part (1).