14.

LAGRANGE'S THEOREM
14.1. Subgroups
If G is a group and H is a subset of G then H is a subgroup if
(i) xy G for all x, y G
(ii) 1 G
(iii) x1 G for all x G.
A subgroup H is a group in its own right. Every group is a subgroup of itself. All other subgroups are called proper subgroups. The set {1} is a subgroup in any group and is called the
trivial subgroup.
We write H G if H is a subgroup of G and H < G if H is a proper subgroup of G,
that is, if H is a subgroup but is not G itself.
Example 1: 2Z (the group of even integers) is a subgroup of Z (under +).
WARNING: G and H must have the same binary operation before one can say H G. For
example R# (the group of non-zero real numbers under multiplication) is not a subgroup of R
(the group of all real numbers under addition) even though it is a subset.
Theorem 1: For all g G, g is a subgroup of G.
Proof: (1) For all r, s grgs = gr+s g . (2) 1 = g0 g. (3) For all r, (gr)1 = gr g.

14.2. Cosets
Suppose H G. We define a relation on G by defining x y if x = yh for some
h H.
Theorem 2: is an equivalence relation.
Proof:
Reflexive: Let a G. Then a = a1. Since 1 H, a a.
Symmetric: Suppose a b. Then a = bh for some h H and so ah1 = b. Since h1 H, b a.
Transitive: Suppose a b and b c. Then a = bh for some h H and b = ck for some k H.
Thus a = (ck)h = c(kh). Since kh H, a c.
NOTE: Each of the three properties of an equivalence relation comes from one of the three
closure properties of a subgroup.
The equivalence classes under are called the right cosets of H in G. The coset
containing g is denoted by gH. We define left cosets similarly, writing them as Hg.
The right coset containing g is
gH = {gh | h H}
and the left coset containing g is
Hg = {hg | h H}.

171

NOTE: (1) Many books define left and right cosets in the opposite manner. However the
notation gH and Hg always mean what we have stated above.
(2) Just because aH = bH does not mean that a = b. It merely means that they are in the same
right coset, or equivalently that b1a H.
Example 2: (1) G = R#, H = {1} under the operation of multiplication. The cosets are all of
the form {x}.
(2) G = C#, the group of non-zero complex numbers under multiplication and let
H = {z C | |z| = 1}.
The right (or left) cosets of H in G are the concentric circles with the origin as centre.
(3) G = the Symmetry Group of a square:
1
r
r2
r3
a
1
r
r2
r3
a
1
2
3
r
r
r
1
b
r
2
3
2
r
r
1
r
c
r
3
2
3
r
1
r
r
d
r
a
d
c
b
1
a
b
c
d
a
r2
b
c
b
a
d
r3
c
d
a
b
c
r
d
Let H = {1, a}. The right and left cosets of H in G are:
H1 = {11, a1} = {1, a}
1H = {11, 1a} = {1, a}
Hb = {1b, ab} = {b, r3}
bH = {b1, ba} = {b,r2}
Hc = {1c, ac} = {c, r2}
cH = {c1,ca} = {c,r3}
Hd = {1d, ad} = {d, r}
dH = {d1, da} = {d,r}

b
b
c
d
a
r3
1
r
r2

c
c
d
a
b
r2
r
1
r3

d
d
a
b
c
r
r3
r2
1

Theorem 3:
(1) The subgroup H is itself one of the cosets of H in G.
(2) Two elements a, b belong to the same coset (of H in G) if and only if b1a H.
Proof:
(1) H = 1H.
(2) a, b belong to the same coset aH = bH
a = bh for some h H
b1a = h for some h H
b1a H.
Generally, equivalence classes can be of different sizes. But in the case of cosets, they
all have the same size.
Theorem 4: If H G then every coset of H in G has H elements.

172

Proof: There is a natural 1-1 correspondence between any coset aH and H itself viz. ah h.
Hence the number of elements in each is the same.

14.3. Lagranges Theorem

Theorem 5 (Lagrange): The order of a subgroup (of a finite group) divides the order of the
group.
Proof: Suppose there are m cosets of H in G. Since G is the disjoint union of them and each
coset has |H| elements, it follows that |G| = m.|H| and so |H| divides |G|.
This is a very powerful result. It shows that the number of elements in a group greatly
affects its structure.
Example 3: If |G| = 14 , the only possible orders for a subgroup are 1, 2, 7 and 14.
Theorem 6: Groups of prime order are cyclic.
Proof: Suppose |G| = p where p is prime.
Since p 2 we may choose g G such that g 1. Let H = g.
Let |H| = n. Now n divides p and n > 1 so n = p. Hence G = H and so is cyclic.
Since the order of an element is the order of the cyclic subgroup it generates, we have:
Theorem 7: The order of an element of a finite group divides the order of the group.

14.4. Eulers Theorem

Recall that if m is any positive integer Zm# denotes the group of all numbers from 1 to
m 1 which are coprime with m, under the operation of multiplication modulo m. [The
coprimeness is what ensures the existence of inverses]
For example Z7## = {1, 2, 3, 4, 5, 6};
Z8 = {1, 3, 5, 7};
Z10# = {1, 3, 7, 9}.
The order of Zm# is denoted by (m). The function is called the Euler function
[N.B. you pronounce Euler as Oiler].
For example, (7) = 6; (8) = 4; (10) = 4.
It's all very well to find (m) for small values of m by counting the elements of Zm#
what do we do for large m? Is there a formula?
If a and b are coprime integers (i.e. have no common factors) it can be shown that:
(ab) = (a) (b). For example, (70) = (7) (10) = 6.4 = 24. This means that (m) can be
computed provided we have a formula for prime powers.

173

Theorem 8: If p is prime, (pn) = pn1(p 1).

Proof: Of the pn numbers from 0 to pn 1 there are precisely pn1 multiples of p. The
remaining pn pn1 = pn1(p 1) numbers will be precisely the ones with no factor in common
with pn. Hence (pn) = pn1(p 1).
Example 4: (200) = (23.5)
= 2(2 1) . 51(5 1)
= 4.1.5.4
= 80.
Theorem 9(Euler): If a is coprime with m then a(m) = 1 (mod m), or in other words,
a(m) leaves a remainder of 1 when divided by m.
Proof: Suppose a is coprime with m. Then a Zm#. Suppose it has order n. By Lagrange's
Theorem n divides (m).
Thus (m) = nq for some q Z.
(m)
Now a
= (an)q = 1q = 1.
Corollary (Fermat): If p is prime and does not divide a, then ap1 = 1 (mod p).
Euler's theorem can be used to calculate the remainders of certain very large numbers.
Example 5: What is the remainder on dividing 21000 by 42?
Solution: (42) = (2.3.7) = 12 so 212 = 1 (mod 42).
Dividing 1000 by 12 we get a remainder of 4. [1000 = 12.83 + 4]
So 21000 = (212)83.24 = 183.24 = 16.
Hence 21000 leaves a remainder of 16 when divided by 42.
NOTE: To work this out directly by calculating 21000 first would need vast amounts of
computing power.

14.5. Generators and Relations

A finite group can be described by means of a multiplication table but its structure is
hidden in a mass of details. A more compact way of representing a group structure is in terms
of generators and relations.
A set of generators for a group is a set of elements such that every element can be
expressed in terms of them. A relation is an equation that holds between the generators. The
notation
g1, g2, ... , gm | R1, R2, ... , Rn
represents a group generated by the elements g1, g2, ... , gm where the relations R1, R2, ... , Rn
hold.
A typical example is:
a, b, c | a7 = 1, b4 = 1, c5 = 1, ba = a1b, ca = ac, cb = bc
where the elements are strings of a's, b's and c's and where the six relations can be used to
simplify such strings. To multiply two strings you just write one following the other and
simplify where possible using the relations.
For example if x = a4bc3 and y = a2b3c4 in the above group then xy = a4bc3a2b3c4.

174

This can be simplified as follows:

a4bc3a2b3c4 = a4ba2c3b3c4 (since ca = ac)
= a4ba2b3c7 (since cb = bc)
= a4ba2b3c7 (since c5 = 1)
= a4(ba)ab3c7
= a4(a1b)ab3c7 (since ba = a1b)
= a3(ba)b3c7
= a3(a1b)b3c7 (again, since ba = a1b)
= a2b4c7
= a2c2 (since b4 = c5 = 1).
In this case we could have done this simplification by observing that c an move freely
across as and bs and that every time the letter b crosses the letter a the a gets inverted. If
we bring all the bs to the front then the a4 gets inverted 4 times, so stays as a4 and the a2 gets
inverted 3 times so becomes a2. All the cs can slip down to the end with no change.
Hence a4bc3a2b3c4 = b4a4a2c7 = b4a2c7 = a2c2.
This example is rather more complicated than those we shall encounter. A much
simpler example is a | an = 1 which denotes the cyclic group of order n (there is essentially
only one since all cyclic groups of the same order are isomorphic).
The infinite cyclic group, an example of which is (Z,+), is denoted by a | , with an
empty set of relations.
Not every relation that holds in a group needs to be listed. Only enough so that any
other relation can be deduced from those that are given.
For example in the group a | a4 = 1 the relation a12 = 1 must hold but it is just a
consequence of the one which is listed.

14.6. Dihedral Groups

The family of cyclic groups contain those groups with the simplest possible group
structure. A closely related family is the family of dihedral groups. The dihedral group of
order 2n is the group:
D2n = a, b | an = 1, b2 = 1, ba = a1b.
Dihedral groups occur naturally in many different guises. D2n is, for example, the group
structure of the Symmetry Group of a regular n-sided polygon. The symmetry operations
consist of the rotation R through 2/n, and its powers plus the 180 rotations about the n axes of
k
symmetry. But if Q denotes any one of these, the others can be expressed in the form RkQR .
If, for example, there is a vertical axis of symmetry and Q denotes the 180 degree
rotation about it, any other 180 symmetry operation can be achieved by rotating the axis until it
becomes vertical (Rk), carrying out Q about the vertical axis, and then rotating the axis back to
its original position (Rk)
Now Rn = 1 (n successive rotations through 2/n);
Q2 = 1 (two successive 180 rotations) and
QR = R1Q (try it!).

175

So this symmetry group is R, Q | Rn = 1, Q2 = 1, QR = R1Q, that is, it is the dihedral

group of order 2n. In particular D8 is the symmetry group of a square.

14.7. Dihedral Arithmetic

The dihedral group D2n = a, b | an = 1, b2 = 1, ba = a1b has its relations built up from
the basic three relations in the list. Let us examine their implications.
an = 1: This means that any expression involving a's and b's need not have any string of
successive a's longer than n-1, because any block of n successive a's is an which, because it is
equal to the identity, can be removed.
For example in D8 = a, b | a4 = 1, b2 = 1, ba = a1b, an element such as aabaaaaaaaba
can be simplified to aabaaaba = a2ba3a by removing an a4 from the middle.
b2 = 1: This means that it is never necessary to have two successive b's. For example in D8 an
expression such as abbbaaabbaa can be simplified to abaaaaa by using b2 = 1, twice. This can
be further reduced to aba by use of the relation a4 = 1.
Another consequence of b2 = 1 is b = b1 (just multiply both sides on the left by b1).
This means that there is never any need to have b1 in any expression.
ba = a1b: It is this third relation that makes dihedral groups non-abelian (except for the trivial
cases of D4 and D2 where a1 = a). Expressing this relation in words, we can say that every time
each b passes across each a it inverts it converts it to a1.
Consequently if we have an expression involving a mixture of a's and b's we can move
all the a's up to the front and all the b's down to the back just as we would if the commutative
law was in force. The difference is that the a's get inverted every time a b crosses over. This is
the dihedral "twist".
Example 6: The expression aba3ba2bab can be written as:
aa3bba2ab = a2bba2ab = a2a2ab = ab.
Theorem 10: The elements of D2n are:
1
a
a2
a3
...
an1
2
3
b
ab
ab
ab
...
an1b
Proof: Because of the relation ba = a1b we can express every element in the form arbs. But
because an = 1 and b2 = 1, we may assume that r = 0, 1, 2, ... , n 1 and j = 0 or 1.
Notice that the first row consists of the cyclic subgroup, H, generated by a, and the
second row is the left coset Hb.
Theorem 11: The elements of the form akb all have order 2.
Proof: (akb)2 = akbakb = akakbb = bb = 1.

176

14.8. Groups of Order 2p

Recall that as a consequence of Lagrange's Theorem we showed that groups of prime
order are cyclic. This is an example of a classification theorem. A classification theorem says,
in effect, "tell me a little about a group and I will tell you everything". Here, just by knowing
that the order of the group is prime, we essentially know everything about the structure of the
group.
Constructing classification theorems is one of the goals of a group-theorist. Many of
them are around. There is a classification theorem for finite abelian groups which says, in
effect, "tell me the orders of the elements of a finite abelian group and I will tell you all you
want to know about the group's structure". This is a big theorem but not too big to present in a
full course on group theory. The proof might take two or three pages to write out but that is not
for us here.
The biggest classification theorem in the whole subject is the famous Classification of
Finite Simple Groups. It doesn't take too much to explain what a simple group is. A simple
group is one where there is no subgroup H (other than 1 and the group itself) for which the left
and right cosets are the same.
The statement of the theorem, essentially a list of all finite simple groups, would take a
couple of pages to write out but the proof has never been assembled in one place for it has been
estimated that it would occupy about 20,000 pages! Thousands of group-theorists have been
working on the problem for over a hundred years and they have gradually dealt with more and
more cases until finally, a few years ago, the last piece was fitted into the jig-saw. It is an
achievement that is surely worthy of a place in the Guiness Book Of Records.
The next big classification theorem may be the classification of finite p-groups, groups
whose order are a power of a prime.
But to conclude our brief excursion into group theory we present a much more modest
classification theorem, the classification of groups of order 2p (where p is prime). We shall
show that all such groups are either cyclic or dihedral.
Modest though it may be in comparison with the Finite Simple Group Classification
Theorem, it still probably represents one of the longest and most sustained pieces of reasoning
you will have encountered so far in mathematics.
The result is not one you are likely to use. Nor will you ever be likely to need to make
use of a knowledge of the proof. But then this could be said of the vast amount of mathematics
you learn. It's the mode of reasoning, the way it shows how a problem can be clearly and
logically attacked by a sustained piece of reasoning, that will be of benefit to you.
So take a packed lunch, for this will be a long journey. You may have heard of the
thriller called The Thirty-nine Steps. Well here they are!

177

14.9. The Thirty Nine Steps

Theorem 12: If |G| = 2p for some prime p, G is cyclic or dihedral.
Proof:
(1) Suppose that |G| = 2p where p is prime
(2) Suppose that p is odd.
(2) By Lagrange's theorem the order of every element is 1, 2, p or 2p.
(3) Suppose G contains an element of order 2p.
(4) Hence G is cyclic.
(5) Since |G| is even, G must contain an element, b, of order 2.
(6) Suppose that all the elements of G, except 1, have order 2.
(7) Then G is abelian.
(8) Choose a, b G of order 2 with a b.
(9) Hence H = {1, a, b, ab} is a subgroup of G of order 4.
(10) Since 4 does not divide 2p we get a contradiction.
(11) So G must have an element, a, of order p.
(12) Let H = a.
(13) Since 2 does not divide p, we must have b H.
(14) The right cosets of H in G must be H and bH.
(15) Similarly the left cosets are H and Hb.
(16) Since bH and Hb consist of all the elements outside H, they must be equal, i.e. Hb = bH.
(17) Now ba bH so ba Hb.
(18) Hence ba = arb for some integer r = 0, 1, ... , p 1.
(19) Hence b2a = barb
(20)
= baa ... ab (where there are r factors of a).
(21)
= arar ... arb2 (where there are r factors of ar).
(22)
= ar (since b2 = 1).
(23) So ar = a,
(24) Hence p divides r2 1.
(25) Since p is prime and r2 1 = (r 1)(r + 1), p divides r 1 or r + 1.
(26) Thus ba = ab or a1b.
(27) Suppose ba = ab.
(28) Now the order of ab must be 1, 2 or p.
(29) Suppose ab = 1.
(30) The a = b1 = b, a contradiction.
(31) Suppose ab has order 2.
(32) Then 1 = (ab)2 = a2b2 = a2, a contradiction.
(33) Suppose ab has order p.
(34) Then 1 = (ab)p = apbp = bp, a contradiction.
1
(35) So ba = a b and so G = a, b | ap = b2=1, ba = a1b = D2p.
(36) Suppose that p = 2.
(37) Then every element of G (except 1) has order 2.
(38) Hence G is abelian.
(39) Hence G = a, b | a2 = b2 =1, ba = ab> = a, b | a2 = b2 =1, ba = a1b = D4.

178

SUMMARY

Subgroup: H G means H is a subset of G which is closed under , 1 and inverse.

Includes cyclic subgroups g consisting of all powers of a single generator, g.
Cosets: Right cosets are gH = {ghhH}. They include H itself and all have the same size viz.
|H|. (Left cosets, Hg, similarly.)
Lagrange's Theorem: If H G, |H|divides |G|.
Euler's Theorem: (m) = |Zm#| = the number of integers from 1 to m 1, coprime with m.
If a,b are coprime (ab) = (a)(b) and (pn) = pn1(p 1).
If a,m are coprime a(m) = 1 (mod m).
Fermat's Theorem: If the prime p doesn't divide a, ap1 = 1 (mod p).
Generators and Relations: a, b, ... | R1, ... is the group generated by a,b,... subject to the
relations R1, ...
Cyclic Groups: Cn = a | an = 1 is the cyclic group of order n.
Dihedral Groups: D2n = a, b | an = 1, b2 = 1, ba = a1b is the dihedral group of order 2n.
If n > 2, D2n is non-abelian. Moving each b past an a inverts the a.
The elements are: 1, a, a2 , ... , an1 (a subgroup of order n) and
b, ab, a2b, ... , an1b
(each one having order 2)
Groups of Order p: are cyclic.
Groups of Order 2p: are cyclic or dihedral.

EXERCISES FOR CHAPTER 14

Exercise 1: If G = (Z, +), which of the following subsets are subgroups of G?
A = the set of even integers;
B = the set of odd integers;
C = the set of non-negative integers;
D = {0}
E = the set of integers which are expressible as 42m + 1023n for integers m,n.
Exercise 2: If G = (C#, ), the set of non-zero complex numbers under multiplication, which of
the following subsets are subgroups of G?
A = the set of non-zero rational numbers;
B = the set of positive integers;
C = {1, 1, i, i};
D = {1, , 2};
E = {a + bia > 0};
F = {1, , 2, 3, ... }.

179

Exercise 3: If G is the group whose table is given below, show that H = {1, c, d} and
K = {1, b} are both subgroups of G. Find all the left and right cosets of each subgroup.
1 a b c d e
1 1 a b c d e
a a 1 c b e d
b b d 1 e a c
c c e a d 1 b
d d b e 1 c a
e e c d a b 1
Exercise 4:
If G = a,b | a4 = 1, b3 = 1, ab = ba> and H is the cyclic subgroup generated by b, find the left
and right cosets of H in G.
Exercise 5: In the dihedral group D10 = a, b | a5 = 1, b2 = 1, ab = ba1 , simplify
a7b3a2baba3ba2a7b2a .
Exercise 6: Simplify a13b5a2b7a2ba in the above group D10.
Exercise 7: Find the order of H given the following clues:
(a) H is a proper subgroup of a group of order 68
(b) H is non-cyclic
Exercise 8: Find the order of H given the following clues.
(a) H is a subgroup of some group of order 100.
(b) H contains no element of order 2.
(c) H is not cyclic.
Exercise 9: Find the order of H given the following clues:
(a) H is a subgroup of some group G of order 168.
(b) H is a subgroup of another group K of order 112.
(c) H is not cyclic or dihedral.
(d) H contains an element of order 7.
(e) H has more than two left cosets in K.
Exercise 10: If G is a group, the centre (zentrum in German) of G is defined to be
Z(G) = {gG | gx = xg for all x G}. In other words, it is the set of all elements which
commute with everything.
(a) Prove that Z(G) is a subgroup of G.
(b) Find Z(D2n). [HINT: You will need to consider odd and even values of n separately.]

180

Exercise 11: If G = (R, +), which of the following subsets are subgroups of G?
A = the set of integers;
B = the set {x10 < x < 10};
C = the set of numbers of the form a + b2 where a, b are integers;
D = the set of real numbers whose decimal expansions are finite.
Exercise 12: If G = (C#, ), which of the following subsets are subgroups of G?
A = the set of complex numbers whose modulus is 1;
B = the set of complex numbers whose modulus is an integer;
C = the set of complex numbers whose modulus is a rational number;
D = the set of solutions to the equation z9 + z8 z 1 = 0.
Exercise 13: G is the group whose table is given below. Show that H = {1, a, d, f} and
K = {1, d} are both subgroups of G. Find all the left and right cosets of each subgroup.
1 a b c
d e
f
g
b c
d e
f
g
1 1 a
d e
g f
c
1 b
a a
a
e
f
b b g d 1 c
c
e
1 d b f
g a
c
c
b 1 g a
e
d d f
e
b f
a
g d c
1
e
f
1
g
e
a
b
d
c
f
a
f
e
1 b d
g g c
Exercise 14:
Find the left and right cosets of {1, b} in the dihedral group D12 = a, b | a6 = b2 = 1, ba = a1 b.
Exercise 15: In the dihedral group
D14 = a, b | a7 = 1, b2 = 1, ba = a1b, simplify a9b7a4baba2ba3a12ba .
Exercise 16: Find the order of H given the following clues:
(a) H is a proper subgroup of a group of order 52.
(b) H is non-cyclic
Exercise 17: Find the order of H given the following clues:
(a) H is a subgroup of some group G of order 100.
(b) H is a subgroup of another group K of order 40.
(c) H is not cyclic or dihedral.
Exercise 18: Find the order of H given the following clues:
(a) H is a subgroup of some group G of order 20.
(b) H is non-abelian.
(c) G contains an element g of order 2 and an element h of order 5.
(d) H contains h but not g.

181

Exercise 19: If G is a group, the derived subgroup of G is defined to be the subgroup

generated by all the elements of the form x1y1xy as x, y range over the elements of the group.
Find the derived subgroup of D2n. [HINT: You will need to consider odd and even values of n
separately.]

SOLUTIONS FOR CHAPTER 14

Exercise 1:
A is a subgroup since the sum of two even integers is even, 0 (the identity of G) is even, and the
additive inverse, n of any even integer is even.
B is not a subgroup. In fact the sum of two odd integers is never odd. As well as not being
closed under addition, B does not contain the identity of G.
C is not a subgroup. It is closed under addition and it does contain the identity, 0. But it is not
closed under inverses. If n 0 it is not true that n 0.
D is a subgroup as is easily verified. It is always the case for any group that the set consisting
of just the identity is a subgroup.
E is a subgroup. Closure under addition comes from checking that:
(42a + 1023b) + (42c + 1023d) = 42(a+c) + 1023(b+d). Since a + c and b + d will be integers if
a, b, c, d are this sum has the required form. Zero can be expressed as 420 + 10230 and so is
in the subset, and (42m + 1023n) = 42(a) + 1023(b) which has the required.
Exercise 2:
a
c
ac
A is a subgroup since the product of two rational numbers b and d is the rational number bd ,
a
b
the integer 1 is rational and the inverse of every rational number b is the rational number a .
B is not a subgroup because although it is closed under multiplication and contains the identity,
it is not closed under inverses. For example, 2 is an integer but not its multiplicative inverse
.
C is a subgroup.
D is not a subgroup. It contains the identity and is closed under inverses, but it is not closed
under multiplication. For example 2 is in the subset but not 22.
E is likewise not a subgroup, for much the same reasons as D. The set E consists of all
complex numbers which lie to the right of the imaginary axis, and that set certainly contains 1
and is closed under inverses. But it fails to satisfy the most fundamental property of subgroups

182

 it is not closed under multiplication. For example, ei/4 lies to the right of the imaginary axis
but its square is i and lies on that axis, and its cube, e3i/4 lies to the left.
F is a subgroup because it is the cyclic subgroup generated by .
Exercise 3:
The fact that H and K are subgroups of G
tables from the main table:
H
1
c
d
K
1
b

can be most easily seen by extracting their group

1
1
c
d

c
c
d
1
1
1
b

d
d
1
c
b
b
1

Since every entry in each table belongs to the subset in each case each subset is closed under
multiplication. Clearly each contains the identity and, since 1 appears in each row and column,
every element has an inverse within the respective subset.
The right cosets of H in G are H = {1, c, d} and aH = (a1, ac, ad} = {a, b, e}.
The left cosets of H in G are H = {1, c, d} and Ha = {1a, ca, da} = {a, e, b}.
NOTE that in this example the left and right cosets are the same, even though the group is nonabelian.
The right cosets of K in G are K = {1, b}, aK = {a1, ab} = {a, c} and dK = {d1, db} = {d, e}.
NOTE that we didn't waste our time with bK or cK because those elements were already
included and we would have simply repeated the first two cosets. For example bK = {b1, bb} =
{b, 1} = {1, b}. So always use as a representative for a new coset, an element which has not yet
been included.
The left cosets of K in G are K = {1, b}, Ka = {1a, ba} = {a, d} and Kc = {1c, bc} = {c, e}
NOTE that in this case the left cosets and the right cosets give two different subdivisions of the
group. Thus K is not one of these normal subgroups.
Exercise 4:
The elements are: 1, a, a2, a3,
b, ab, a2b, a3b,
b2, ab2, a2b2, a3b2.
The right cosets are H = {1, b, b2}
aH = {a, ab, ab2}

183

a2H = {a2, a2b, a2b2}

and a3H = {a3, a3b, a3b2}
Of course since this group is abelian the left cosets are the same as the right cosets.
Exercise 5:
Using b2 = 1 this becomes a7ba2baba3ba2ba7a, and combining powers of a we get
a7ba2baba3ba10. Using a5 = 1 we get a2ba3baba3b. Now we need to make use of the relation
ba = a1b. Moving each b past an a, inverts it. Moving the second last b to the back we get
a2ba3baa3b2 = a2ba3ba3 . Moving the next b down gives a2ba3a3b = a2bb = a2.
Exercise 6:
The expression can be successively simplified as follows:
a13b5a2b7a2ba a3ba2ba2ba a3ba2ba2a1b a3ba2bab a3ba2a1 a3ba a2b .
Exercise 7:
By Lagrange's Theorem divides 68. Since H is not cyclic and groups of prime order are cyclic,
|H| is not prime. Similarly it is not 1 (groups of order 1 contain just the identity and are cyclic).
So we are looking for a divisor of 68 which is neither 1 nor prime. Since the prime
factorisation of 68 is 22.17 , the only possibilities are 4 and 68. However since H is a proper
subgroup, this means it is not the whole of G and so 68 is ruled out. Thus, by a process of
elimination, |H| = 4.
Exercise 8:
By Lagrange's Theorem |H| divides 100. H being non-cyclic rules out 1 and the primes 2 and 5,
leaving 4, 5, 10, 20, 25, 50 and 100. Now groups of even order must contain an element of
order 2. Since H does not, it must have odd order, leaving 25 as the only possibility.
Exercise 9:
By Lagrange, |H| divides both 168 = 837 and 112 = 167 and therefore must divide their
greatest common divisor, which is 56. Since H contains an element of order 7, |H| must be
divisible by 7. This limits the possibilities to 7, 14, 28 and 56. Now since H is neither cyclic
nor dihedral, |H| cannot be prime or twice a prime [groups of order p are cyclic; groups of order
2p are cyclic or dihedral]. This narrows down the possibilities to 28 and 56. Now if |H| was
56 there would be exactly 2 left cosets in K which has order 112. By clue (e) this is not so, and
hence 56 is ruled out. Therefore |H| must be 28.
Exercise 10:
(a) This is easily verified. Note that gx = xg implies that xg1 = g1x .
(b) D2n = a, b | an =1, b2=1, ba= a1b. If ar Z(G) then arb = bar = arb, so a2r = 1 which
means that n divides 2r. If n is odd this means n divides r and so ar = 1. If n is even ar is 1 or
an/2. Similarly one can check that no element of the form arb commutes with a.
So Z(D2n) = {1} if n is odd and {1, an/2} if n is even.

184