Cosets and Lagrange’s theorem

These are notes on cosets and Lagrange’s theorem some of which may already have been
lecturer. There are some questions for you included in the text. You should write the
answers in the boxes. And there is no need to stop completely if you can’t answer one
particular question—continue on to the next sections and come back to it later. If this
was not lectured, it would be about one week’s work—roughly 10 hours—so perhaps it
will be 3–4 hours given your previous work. How long you actually need will depend on
how thoroughly you want to study.

It might be useful to know the aims of this workbook. It is meant to help you to get used
to working with a balance of abstract ideas and concrete calculations. I would like you
to understand the statements of definitions and theorems a bit better from working with
them in examples, and I would like you to become more comfortable with reading and
writing abstract proofs.

There are six sections that have exercises. (Section 7 is optional.)

1. Lagrange’s theorem

2. Cosets

3. Cosets have the same size

4. Cosets partition the group

5. The proof of Lagrange’s theorem

6. Case study: subgroups of Isom(Sq)

Reminder about notation

When talking about groups in general terms, we always write the group operation as
though it is multiplication: thus we write gh ∈ G to denote the group operation applied
to g and h (in that order). And we denote the identity element in G by 1G .
However, in examples the group operation might be addition, or anything else. Our
general statements cover that case, but must be interpreted correctly. If we prove that
ab2 = 1G , then in the case where G = Z, we realise this as a+2b = 0—the group operation
is addition, and the identity element is 0 ∈ Z.

1
 Cosets and Lagrange’s theorem

1 Lagrange’s theorem
Lagrange’s theorem is about finite groups and their subgroups. It is very important in
group theory, and not just because it has a name.

Theorem 1 (Lagrange’s theorem) Let G be a finite group and H ⊂ G a subgroup of

G. Then |H| divides |G|.

We will prove this theorem later in the workbook. But first we begin to see what the
theorem means.

1.1 Understanding the statement

Remember that |G| denotes the number of elements of the group G; it makes perfect sense
because G is finite. Similarly, since H ⊂ G, certainly H is also finite and again writing
|H| makes sense.
Now Lagrange’s theorem says that whatever groups H ⊂ G we have, |H| divides |G|.
That’s an amazing thing, because it’s not easy for one number to divide another. For
example, if we had a group G1 with |G1 | = 77, then any subgroup of G1 could only have
size 1, 7, 11 or 77. So if you were working out the elements of a subgroup H1 of G1 and
you could see 12 different elements of H1 already, then in fact you would be finished: you
would know that |H1 | = 77, and so the subgroup would have to be the whole of G.
That example is a bit artificial. Nevertheless, seeing how a theorem is used in practice
helps you to understand it, so we look next at a true application of Lagrange’s theorem.

1.2 A favourite application of Lagrange’s theorem

The same counting argument as above (but easier) proves your favourite first corollary of
Lagrange’s theorem. Remember that 2 is the smallest prime—1 is not a prime.

Corollary 2 If G is a finite group with |G| prime, then G is cyclic.

Proof Step 1: Show that |G| ≥ 2 and conclude that there is some element g ∈ G which
is not equal to the identity 1G .

2
 Step 2: Using g from Step 1, show that the subgroup H = hgi ⊂ G also has |H| ≥ 2.

Step 3: Since |G| is prime, conclude from Lagrange’s theorem that |H| = |G|.

Step 4: Conclude that hgi = G.

Step 5: Write down the definition of cyclic group (from lecture notes or a textbook),
and conclude from the definition that G is cyclic as claimed.

Q.E.D.
We worked that proof out in very close detail. I write it out again much more briefly
below—it’s merely a condensed version of what you wrote above. The two versions of
the proof are equally valid since they follow the same logical course and address the same
mathematical points. You can decide which you prefer to read and which you prefer to
write for yourself.

Proof (of Corollary 2 again) Let p = |G|. Since p ≥ 2, there is an element g ∈ G

with g 6= 1G . Consider the subgroup hgi ⊂ G generated by g. We have | hgi | ≥ 2 since
both 1G , g ∈ hgi. So by Lagrange’s theorem | hgi | = p. Thus hgi = G, and so, by
definition, G is cyclic as claimed. Q.E.D.
We will prove Lagrange’s theorem over the next few sections. We start by defining
cosets, since they will be the main technical tool.

3
2 Cosets
2.1 The definition of left coset
We want to understand the following definition—it is very important in group theory (and
in abstract algebra more generally). Incidentally, this definition talks about left cosets.
Right cosets are important too, but we will consider them another time.

Definition 3 Let G be a group and H ⊂ G a subgroup. A left coset of H in G is a subset

of G of the form gH for some g ∈ G.
For a left coset gH, the element g ∈ G is referred to as a representative of the coset.

2.2 Multiplying elements and sets

Of course, the expression gH does not make immediate sense from the group axioms.
What it means, by definition, is

gH = {gh | h ∈ H} .

To put this another way, the golden rule is this: if you know that f ∈ gH, then you
can conclude that there is some h ∈ H so that f = gh.
Here is an example of how the golden rule works.

Applying the golden rule

Consider G = S4 and H = {id, (1, 2)}. If g = (2, 3, 4), then gH =

{(2, 3, 4), (2, 3, 4)(1, 2)} = {(2, 3, 4), (1, 3, 4, 2)}. Now let f = (3, 4, 2, 1)—this
is an element of gH. Which h ∈ H satisfies f = gh?

Or if g = (1, 3)(2, 4), then gH = {(1, 3)(2, 4), (1, 4, 2, 3)}. If you let f =
(1, 4, 2, 3), which h ∈ H satisfies f = gh this time?

In the box below, compute the two cosets g1 H ⊂ S4 and g2 H ⊂ S4 for

H = {id, (1, 2, 3, 4), (1, 3)(2, 4), (1, 4, 3, 2)} and g1 = (1, 3, 2), g2 = (1, 2, 3, 4).

4
 Computing cosets

One of these two cosets is equal to H itself; the other is disjoint from H. We consider
why this happens in section 4 below.

2.3 One coset can have many representatives

Compute another example—in this case the group operation is addition, so we write a + b
rather than ab, and similarly cosets are written a + H rather than aH.

Cosets in Z

Let G = Z and H = 5Z = {5n | n ∈ Z} = {. . . , −5, 0, 5, 10, . . . }. The coset

2 + H is the set {2 + 5n | n ∈ Z} = {. . . , −8, −3, 2, 7, 12, . . . }, which is of
course just the set of numbers congruent to 2 mod 5.

We say that 2 is a representative of the coset 2 + H.

Which of the numbers 17, 152, 21, −18, −2 lie in the set 2 + H?

Now calculate 7 + H =

You should see that 7 + H is exactly the same subset of Z as 2 + H. Therefore we can
also say that 7 is a representative of 2 + H. The point is that 7 and 2 are the same mod
5. In fact, so are −8, 12, 152, or indeed any other element of the set 2 + H. We can refer
to any of them as a representive of this coset.

3 Cosets have the same size

The precise statement of the proposition below is what this section title really means.
Proposition 4 Let G be a group and H ⊂ G a subgroup. If H is a finite group, then
every left coset of H in G is finite, and moreover |gH| = |H| for any g ∈ G.

5
 In other words, any (left) coset of H in G has exactly the same number of elements
as H does. Your first thought is to see whether this is true in examples. In section 4.1
you computed all cosets of a particular subgroup of S3 , so you should look back there to
see that indeed in that case every coset has exactly 2 elements.

3.1 Why do cosets all have the same size?

Proposition 4 follows from an even better result—it’s better because it doesn’t need to
talk about finiteness at all.

Proposition 5 Let G be a group and H ⊂ G a subgroup. Fix g ∈ G. Then the map

ϕg : H −→ gH defined by ϕg (h) = gh

(which is a map of sets) is a bijection.

This proof is easy: as you’ll see below, the map is surjective by the golden rule and
injective by left cancellation, and we use nothing more complicated than that.

Proof Step 1: Show that if f ∈ gH, then there is an h ∈ H with ϕg (h) = f .

Step 2: Show that if h1 , h2 ∈ H satisfy ϕg (h1 ) = ϕg (h2 ), then h1 = h2 .

Step 3: Triumphant declaration! (I’ll do this—I have to do everything round here.)

So ϕg is bijective because it is surjective by Step 1 and injective by Step 2.

Q.E.D.
Finally, observe that Proposition 4 follows from Proposition 5 because if there is a
bijection between H and gH then these two sets have the same number of elements.

6
4 Cosets partition the group
The title of this section is an important slogan that we want to understand properly and
then prove—at the moment it may mean very little to us.

4.1 What does ‘cosets partition the group’ mean?

Consider G = S3 and let H = {id, (1, 2)} ⊂ G. Compute all left cosets of H in G in the
boxes below; the first couple are already done.

id H = H (1, 2)H =

(1, 3)H = {(1, 3), (1, 2, 3)} (1, 2, 3)H =

(2, 3)H = (1, 3, 2)H =

What you should have found is that if you look at any two of the boxes, then the results
are either exactly the same or disjoint. (Check your answers if that’s not true.)
The subsets of S3 computed above form a partition of S3 . In words: a partition is a
division of the whole of a set into mutually disjoint subsets. The mathematical definition
is more precise.

Definition 6 If X is a set, then a partition of X is a collection of subsets Yα ⊂ X, for

some indexing set α ∈ A, for which

(i) X = ∪Yα , where the union is taken over all α ∈ A, and

(ii) if α1 , α2 ∈ A, then either Yα1 = Yα2 or Yα1 ∩ Yα2 = ∅.

Another example. Consider the set X = {1, 2, . . . 8} of the first 8 positive integers.
Here are several randomly-chosen subsets of X:

Y1 = {1, 2, 3}, Y2 = {2, 4, 6, 8}, Y3 = {4, 6, 8},

Y4 = {1, 3, 5, 7}, Y5 = {5, 7}, Y6 = {8}, Y7 = {4, 6}.

The collection of three subsets Y1 , Y3 , Y5 form a partition of X since

(i) X = Y1 ∪ Y3 ∪ Y5 , and
(ii) each of Y1 ∩ Y3 , Y1 ∩ Y5 and Y3 ∩ Y5 is empty.

7
Find two other collections of subsets from among the Yi listed above that also form
partitions of X.

4.2 Why do cosets partition the group?

Proposition 7 Let G be a group and H ⊂ G a subgroup. The set of all left cosets of H
in G is a partition of G.

For the proof, we must check conditions (i) and (ii) of Definition 6.

Show that if g ∈ G then there is some coset that contains g. [Hint: 1 ∈ H.]

The box above confirms condition (i)—it says that G is a subset of the union of all cosets,
and so it must be equal to the union of all cosets.
The following lemma checks condition (ii).

Lemma 8 (Notation as in Proposition 7.) Let g1 , g2 ∈ G. Then either g1 H = g2 H or

g1 H ∩ g2 H = ∅.

This statement says that in the given circumstances one (or both) of two events, P
and Q say, must happen. A typical method of proof for such statements is to see what
happens if you assume that Q does not happen. If you can show that P must happen,
then you are done. That will be our strategy: we will imagine that g1 H ∩ g2 H 6= ∅ and
then we will prove that g1 H = g2 H.
Before we start the proof, let’s think through an important point: if there is an element
f ∈ g1 H ∩ g2 H, then by the golden rule there are elements h1 , h2 ∈ H so that

f = g1 h1 and f = g2 h2 .

In particular, this shows that g1 h1 = g2 h2 , and we can rearrange this equation to prove
that g1 ∈ g2 H. With those thoughts in mind, we are ready for the proof.

8
Proof (of Lemma 8)
Step 1: Show that if g1 H ∩ g2 H 6= ∅, then there are h1 , h2 ∈ H so that g1 h1 = g2 h2 .

Step 2: Show that g1 h1 = g2 h2 implies that g1 ∈ g2 H.

Step 3: Show that g1 ∈ g2 H implies that g1 H ⊂ g2 H. [Hint: golden rule again.]

Step 4: Explain why you are finished. (I’ll do this.)

Repeating Steps 2 and 3 with g1 and g2 exchanged shows that g2 H ⊂ g1 H.

Together with Step 3, this implies that g1 H = g2 H. The proof is complete.

Q.E.D.

9
5 The proof of Lagrange’s theorem
The idea is that the cosets of H divide G up into equal-sized pieces; and since the size of
each piece is |H|, the result follows. But that’s just the executive summary—we’ll take it
more slowly.
First recall the statement of Lagrange’s theorem.

Now use Proposition 7 to conclude that G is the disjoint union of finitely many cosets
g1 H,. . . , gk H, for some g1 , . . . , gk ∈ G.

Finally, using Proposition 4, deduce that |G| = k|H| and observe that this proves the
theorem.

10
6 Case study: subgroups of Isom(Sq)
We know that Isom(Sq) is a group with exactly 8 elements. By Lagrange’s theorem, any
subgroup H ⊂ Isom(Sq) must have |H| dividing 8—that is, H must have 1, 2, 4 or 8
elements. The only subgroup of order 1 is the trivial subgroup {id}. And if |H| = 8, then
H is the whole group. So we only have to think about subgroups of order 2 or 4.
Before we start, it’s worth noting the order of each element of Isom(Sq)—remember,
the order of g ∈ G is the least integer n > 0 for which g n = 1G . (You may need to remind
yourself of the notation used in lectures for elements of Isom(Sq).)

element id R R2 R3 ρx ρy α β

order

6.1 Subgroups of order 2

If {id, σ} is an order 2 subgroup of Isom(Sq), show that σ 2 = id.

List all elements of Isom(Sq) of order 2.

Finally list all subgroups of Isom(Sq) of order 2.

11
6.2 Subgroups of order 4
Let H ⊂ G = Isom(Sq) be a subgroup of order 4. There are two cases to distinguish. If
H contains an element g ∈ H of order 4, then H = hgi ⊂ G. Using the table of element
orders above, list all such subgroups of order 4.

The second case is when H does not contain any element of order 4. In that case—looking
at the table of element orders again—it must consist of the identity and three elements
of order 2.
Use Lagrange’s theorem to show that if g1 , g2 ∈ G are distinct elements of order 2,
then hg1 , g2 i either has order 4 or actually equals G.

By considering pairs of elements of order 2, list the remaining subgroups of order 4.

12
6.3 The subgroup lattice of Isom(Sq)
We can draw all the subgroups on one very beautiful picture. Here, the bigger the group,
the higher it is in the diagram. The arrows are inclusions from a smaller subgroup to a
larger one. It’s a bit of a jigsaw puzzle to get the subgroups in the right boxes, but see
if you can fit your lists of subgroups into the picture. To make it easier, I’ve drawn it
so that subgroups of the same order lie at the same height on the page: from the top,
therefore, you see rows of subgroups of size 8, 4, 2 and 1.

Isom(Sq)

% ↑ -

% ↑ - ↑ % ↑ -

{id, R2 }

- - ↑ % %

{id}

This is a very well-known picture, although I stole it from David Mond at Warwick; you
can find it on page 12 of his lecture notes

www.warwick.ac.uk/∼masbm/Lectures/groups.pdf

if you would like to see a more geometric derivation of these subgroups (and, happily, he
too has left the boxes blank so it doesn’t spoil the exercise).

13
7 When is a coset also a subgroup? (Optional)
Here’s an extra thought about cosets if you’ve got time for more work.
We stick to the notation H ⊂ G. Obviously a left coset of H is a subset of G—by the
closure axiom, if you like. However a typical left coset is not a subgroup of G: just look
at the examples above—most of the cosets do not even contain the identity. In fact,

Proposition 9 Let G be a group, H ⊂ G a subgroup and g ∈ G. The coset gH is a

subgroup of G if and only if g ∈ H.

I want to prove this. The statement is an ‘if and only if’, which often—although not
always—means that the proof is best done in two stages. I’m even going to split the
statement into two halves to make this clearer. Here’s the first half; it’s the ‘only if’ part
of Proposition 9.

Lemma 10 Let G be a group, H ⊂ G a subgroup and g ∈ G. If the coset gH is a

subgroup of G, then g ∈ H.

Proof Since gH is a group in its own right, gH must contain the identity element 1.
That is, 1 ∈ {gh | h ∈ H}. So there is some element h ∈ H for which gh = 1. This
implies that g = h−1 , and h−1 is certainly an element of H. Q.E.D.
The second half is the ‘if’ part of Proposition 9.

Lemma 11 Let G be a group, H ⊂ G a subgroup. If g ∈ H, then the coset gH is a

subgroup of G.

The proof below is slightly sneaky. We could simply run through the group axioms and
check that they hold for gH—that’s fine and will work, but it’s slightly awkward. Instead,
we find a short-cut by proving that gH = H—for that is indeed a subgroup of G.

Proof First observe that gH ⊂ H; this follows from the closure axiom for H because
g ∈ H. To complete the proof, we show that H ⊂ gH, for then gH = H is a subgroup as
claimed.
Pick h ∈ H. We can write h = g(g −1 h). Since g ∈ H, also g −1 ∈ H and so also
g h ∈ H. So h = g(g −1 h) ∈ gH. We have shown that every h ∈ H is also in gH; in
−1

other words H ⊂ gH, as required. Q.E.D.

Gavin Brown, Clare Dunning

Kent, January 2006

14