Selected Solutions To Dummit and Foote's Abstract Algebra Third Edition

Selected Solutions to Dummit and Foote’s
Abstract Algebra Third Edition
Greg Kikola
July 16, 2020

ii
Copyright © 2019–2020 Greg Kikola.

License CC BY-SA 4.0:
Creative Commons Attribution-ShareAlike 4.0 International.
This document lives at:

https://www.gregkikola.com/projects/guides/
You can find the LATEX source code on GitHub at:
https://github.com/gkikola/sol-dummit-foote
Contents
Preface v
0 Preliminaries 1
0.1 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
0.2 Properties of the Integers . . . . . . . . . . . . . . . . . . . . . . 4
0.3 Z/nZ: The Integers Modulo n . . . . . . . . . . . . . . . . . . . . 11
1 Introduction to Groups 17
1.1 Basic Axioms and Examples . . . . . . . . . . . . . . . . . . . . . 17
1.2 Dihedral Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.3 Symmetric Groups . . . . . . . . . . . . . . . . . . . . . . . . . . 38
1.4 Matrix Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
1.5 The Quaternion Group . . . . . . . . . . . . . . . . . . . . . . . . 52
1.6 Homomorphisms and Isomorphisms . . . . . . . . . . . . . . . . . 54
1.7 Group Actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
2 Subgroups 75
2.1 Definition and Examples . . . . . . . . . . . . . . . . . . . . . . . 75
2.2 Centralizers and Normalizers . . . . . . . . . . . . . . . . . . . . 83
2.3 Cyclic Groups and Cyclic Subgroups . . . . . . . . . . . . . . . . 91
2.4 Subgroups Generated by Subsets of a Group . . . . . . . . . . . . 102
2.5 The Lattice of Subgroups of a Group . . . . . . . . . . . . . . . . 110
3 Quotient Groups and Homomorphisms 123

3.1 Definitions and Examples . . . . . . . . . . . . . . . . . . . . . . 123
3.2 More on Cosets and Lagrange’s Theorem . . . . . . . . . . . . . 150
3.3 The Isomorphism Theorems . . . . . . . . . . . . . . . . . . . . . 161
3.4 Composition Series and the Hölder Program . . . . . . . . . . . . 168
3.5 Transpositions and the Alternating Group . . . . . . . . . . . . . 175
A Cartesian Products and Zorn’s Lemma 183

A.1 Cartesian Products . . . . . . . . . . . . . . . . . . . . . . . . . . 183
A.2 Partially Ordered Sets and Zorn’s Lemma . . . . . . . . . . . . . 184
iii
iv CONTENTS
Preface
This is an unofficial solution guide to the book Abstract Algebra, Third Edition,
by David S. Dummit and Richard M. Foote. It is intended for students who are
studying algebra with Dummit and Foote’s text. I encourage students who use
this guide to first attempt each exercise for themselves before looking up the
solution, as doing exercises is an essential part of learning mathematics.
In writing this guide, I have avoided using techniques or results before the
point at which they are introduced in the text. My solutions should therefore
be accessible to someone who is reading through Dummit and Foote for the first
time.
Given the large number of exercises, errors are unavoidable in a work such
as this. I have done my best to proofread each solution, but mistakes will make
it through nonetheless. If you find one, please feel free to tell me about it in an
email: gkikola@gmail.com. I appreciate any corrections or feedback.
Please know that this guide is currently unfinished. I am slowly working on
adding the remaining chapters, but this will be done at my own pace. If you
need a solution to an exercise that I have not included, try typing the problem
statement into a web search engine such as Google; it is likely that someone else
has already posted a solution.
This guide is licensed under the Creative Commons Attribution-ShareAlike
4.0 International License. To view a copy of this license, visit
http://creativecommons.org/licenses/by-sa/4.0/
I am deeply grateful to the authors, David S. Dummit and Richard M. Foote,
for producing a wonderfully comprehensive and thoroughly well-written work. I
also owe particular thanks to those readers who have taken the time to contact
me in order to point out mistakes or to give general feedback.
Greg Kikola
www.gregkikola.com
gkikola@gmail.com
v
vi PREFACE
Chapter 0
Preliminaries
0.1 Basics
Let A be the set of 2 × 2 matrices over R, let

1 1
M= ,
0 1
and let
B = {X ∈ A | M X = XM }.
0.1.1 Exercise 1
Determine which of the following elements of A lie in B:

1 1 1 1 0 0 1 1 1 0 0 1
, , , , , .
0 1 1 1 0 0 1 0 0 1 1 0
Solution. It is easy to verify that

1 1 0 0 1 0
, , and
0 1 0 0 0 1
all commute with M : the first matrix is M itself, and the latter two are the
zero matrix and the identity matrix, all of which will commute. So each of these
matrices is in B.
We can check the remaining matrices individually: Let

1 1 1 1 0 1
P = , Q= , and R = .
1 1 1 0 1 0
Direct computation shows that

2 2 1 2
MP = 6= = P M,
1 1 1 2

2 1 1 2
MQ = 6= = QM,
1 0 1 1
and
1 1 0 1
MR = 6 = = RM.
1 0 1 1
So P, Q, R 6∈ B.
1
2 CHAPTER 0. PRELIMINARIES
0.1.2 Exercise 2
Prove that if P, Q ∈ B, then P + Q ∈ B.
Proof. Let
a b e f
P = and Q =
c d g h
be matrices in the set B, so that M P = P M and M Q = QM . Then we have

1 1 a+e b+f
M (P + Q) =
0 1 c+g d+h

a+e+c+g b+f +d+h
=
c+g d+h

a+c b+d e+g f +h
= +
c d g h
= MP + MQ
= P M + QM

a a+b e e+f
= +
c c+d g g+h

a+e a+b+e+f
=
c+g c+d+g+h

a+e b+f 1 1
=
c+g d+h 0 1
= (P + Q)M.
Therefore P + Q ∈ B.
0.1.3 Exercise 3
Prove that if P, Q ∈ B, then P Q ∈ B.
Proof. A similar argument to the one in Exercise 2 above will show that P Q ∈ B
for any P, Q ∈ B.
0.1.4 Exercise 4

p q
Find conditions on p, q, r, s which determine precisely when ∈ B.
r s
Solution. Let
p q
P = .
r s
Then
1 1 p q p+r q+s
MP = =
0 1 r s r s
while
p q 1 1 p p+q
PM = = .
r s 0 1 r r+s
0.1. BASICS 3
Therefore, M P = P M if and only if r = 0 and p = s. Hence

p p + q
B= p, q ∈ R .
0 p
0.1.5 Exercise 5
Determine whether the following functions f are well defined:
(a) f : Q → Z defined by f (a/b) = a.
(b) f : Q → Q defined by f (a/b) = a2 /b2 .
Solution. (a) f is not well defined since, for example,
1 2
f (1/2) = 1, f (2/4) = 2, but = .
2 4
(b) Suppose a, b, c, d ∈ Z with b, d 6= 0 are such that
a c
= .
b d
Then
a2 a 2 c 2 c2
f (a/b) = 2
= = = 2 = f (c/d).
b b d d
Therefore f is well defined.
0.1.6 Exercise 6
Determine whether the function f : R+ → Z defined by mapping a real number
r to the first digit to the right of the decimal point in a decimal expansion of r
is well defined.
Solution. f is not well defined since decimal expansions are not unique. For
example, 1 = 1.0 = 0.999 . . . but f (1.0) = 0 and f (0.999 . . .) = 9.
0.1.7 Exercise 7
Let f : A → B be a surjective map of sets. Prove that the relation
a ∼ b if and only if f (a) = f (b)
is an equivalence relation whose equivalence classes are the fibers of f .
Proof. That ∼ is an equivalence relation on A follows directly from the fact that
= is an equivalence relation on the set B.
Now let b ∈ B be arbitrary. Since f is surjective, there is an a in A such
that f (a) = b. Then the equivalence class of a is the set
{x ∈ A | x ∼ a}.
But by definition of ∼, this set is equal to
{x ∈ A | f (x) = f (a) = b}.
Therefore the equivalence class of a is precisely the fiber of f over b.
0.2 Properties of the Integers

0.2.1 Exercise 1
For each of the following pairs of integers a and b, determine their greatest
common divisor, their least common multiple, and write their greatest common
divisor in the form ax + by for some integers x and y.
(a) a = 20, b = 13.
Solution. Applying the Division Algorithm repeatedly, we get
20 = 1(13) + 7
13 = 1(7) + 6
7 = 1(6) + 1
6 = 6(1) + 0.
The first nonzero remainder is 1, so (20, 13) = 1. That is, the two numbers
are relatively prime.
The least common multiple, [20, 13], is given by
20 · 13
= 260.
(20, 13)
To write 1 as a linear combination of 20 and 13, we work backwards and

substitute:
1 = 7 − 1(6)
= 7 − 1(13 − 1(7)) (Substituting 6 = 13 − 7)
= 2(7) − 1(13)
= 2(20 − 1(13)) − 1(13) (Substituting 7 = 20 − 13)
= 2(20) − 3(13).
(b) a = 69, b = 372.
Solution. As above, we have
372 = 5(69) + 27
69 = 2(27) + 15
27 = 1(15) + 12
15 = 1(12) + 3
12 = 4(3) + 0,
so (69, 372) = 3, which gives [69, 372] = 8556. And again, as before, we
0.2. PROPERTIES OF THE INTEGERS 5
write
3 = 15 − 1(12)
= 15 − 1(27 − 1(15)) (Substituting 12 = 27 − 15)
= 2(15) − 1(27)
= 2(69 − 2(27)) − 1(27) (Substituting 15 = 69 − 2(27))
= 2(69) − 5(27)
= 2(69) − 5(372 − 5(69)) (Substituting 27 = 372 − 5(69))
= 27(69) − 5(372).
(c) a = 792, b = 275.
Solution.
792 = 2(275) + 242
275 = 1(242) + 33
242 = 7(33) + 11
33 = 3(11) + 0.
Hence (792, 275) = 11. Calculating the least common multiple gives
[792, 275] = 19 800. Then
11 = 242 − 7(33)
= 242 − 7(275 − 242)
= 8(242) − 7(275)
= 8(792 − 2(275)) − 7(275)
= 8(792) − 23(275).
(d) a = 11 391, b = 5673.
Solution. Using the methods above, we get

(11 391, 5673) = 3,
[11 391, 5673] = 21 540 381
and
−126(11 391) + 253(5673) = 3.
(e) a = 1761, b = 1567.
Solution.
(1761, 1567) = 1,
[1761, 1567] = 2 759 487,
and
−105(1761) + 118(1567) = 1.
(f) a = 507885, b = 60808.
Solution.
(507885, 60808) = 691,

[507885, 60808] = 44 693 880,
and
−17(507885) + 142(60808) = 691.
0.2.2 Exercise 2
Prove that if the integer k divides the integers a and b then k divides as + bt
for every pair of integers s and t.
Proof. Suppose a and b are such that k | a and k | b. By definition, this means
that there exists integers m and n such that a = mk and b = nk. Therefore, for
any integers s and t,
as + bt = (mk)s + (nk)t
= (ms + nt)k.
Since ms + nt must be an integer (due to closure of integer addition and mul-

tiplication), this shows that k | (as + bt).
0.2.3 Exercise 3
Prove that if n is composite then there are integers a and b such that n divides
ab but n does not divide either a or b.
Proof. The Fundamental Theorem of Arithmetic guarantees that n is the prod-
uct of two or more (possibly equal) prime factors. Let a be one of the prime
factors, and let b be n/a. Note that b must be an integer since a | n. Note also
that a, b > 1.
Now n = ab, so clearly n | ab. However, n - a since a is prime and n is
composite.
Finally, suppose for contradiction that n | b. Then there is an integer k > 1
such that b = kn. Multiplying by a on both sides gives ab = akn or n = akn.
Dividing by n then gives ak = 1. But this is absurd because a and k are both
integers greater than 1. This contradiction shows that n - b, so the proof is
complete.
0.2.4 Exercise 4
Let a, b, and N be fixed integers with a and b nonzero and let d = (a, b) be the
greatest common divisor of a and b. Suppose x0 and y0 are particular solutions
to ax + by = N (i.e., ax0 + by0 = N ). Prove for any integer t that the integers
b a
x = x0 + t and y = y0 − t (1)
d d
are also solutions to ax + by = N .
Proof. Substituting for x and y in ax + by gives

b a ab ab
a x0 + t x + b y0 − t = (ax0 + by0 ) + t − t
d d d d
= ax0 + by0
= N.
This holds regardless of the value of t, so (1) is always a valid solution.
0.2.5 Exercise 5
Determine the value ϕ(n) for each integer n ≤ 30 where ϕ denotes the Euler
ϕ-function.
Solution. For each n, the value of ϕ(n) can be determined by first finding the
prime factorization of n,
αk
n = pα1 α2
1 p2 · · · pk , where each pi is prime,
and then by applying the formula given in the text:
ϕ(n) = pα
1
1 −1
(p1 − 1)p2α2 −1 (p2 − 1) · · · pkαk −1 (pk − 1).
For example, to find ϕ(18), we factor 18 = 2 · 32 . Applying the formula then

gives
ϕ(18) = 21−1 (2 − 1) · 32−1 (3 − 1)

=1·1·3·2
= 6.
Applying this process to each n ≤ 30 produces the following table:
n 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
ϕ(n) 1 1 2 2 4 2 6 4 6 4 10 4 12 6 8
n 16 17 18 19 20 21 22 23 24 25 26 27 28
ϕ(n) 8 16 6 18 8 12 10 22 8 20 12 18 12
n 29 30
ϕ(n) 28 8
This process can be used to easily find ϕ(n) for any n whose prime factor-
ization is known.
0.2.6 Exercise 6
Prove the Well Ordering Property of Z by induction and prove the minimal
element is unique:
Theorem. If A is any nonempty subset of Z+ , there is some element m ∈ A

such that m ≤ a, for all a ∈ A.
Proof. Suppose for contradiction that A has no minimal element. We will prove
by (strong) induction on n that for each n ∈ Z+ , n 6∈ A. This will show that A
is the empty set, which would contradict the requirement that A be nonempty.
Clearly 1 6∈ A, for otherwise 1 would be a least element (since 1 ≤ a for all
a ∈ Z+ ). Now suppose that 1, 2, . . . , k 6∈ A for some positive integer k. Then
k + 1 cannot be a member of A since otherwise k + 1 would be the minimal
element. This completes the inductive step, which shows that A is the empty
set, giving the needed contradiction to show that A has a minimal element.
Finally, to show that the minimal element is unique, suppose A has two
minimal elements, a and b. Since a is minimal, a ≤ b. But b is minimal, so
b ≤ a. So a ≤ b and a ≥ b and therefore a = b.
0.2.7 Exercise 7
If p is a prime prove that there do not exist nonzero integers a and b such that
√
a2 = pb2 (i.e., p is not a rational number).
Proof. Suppose for contradiction that a and b are nonzero integers with
a2 = pb2 .
Without loss of generality we may also assume that a and b have no factors in
common (if they do have factors in common, just divide the factors from both
sides of the equation).
Now p | a2 . And since p is prime, we must also have p | a (this uses the
“important property” mentioned in item (8) on page 6 of the text). Then there
is an integer m such that a = pm and hence (pm)2 = pb2 , or p2 m2 = pb2 . This
implies that pm2 = b2 so that p | b2 , which implies p | b. But a and b were
chosen to have no factors in common, yet p is a common factor. This gives the
needed contradiction.
0.2.8 Exercise 8
Let p be a prime, n ∈ Z+ . Find a formula for the largest power of p which
divides n! = n(n − 1)(n − 2) · · · 2 · 1.
Solution. The only integers less than n that are divisible by p are the multiples
of p, of which there are

n
p
of them, where bxc denotes the floor of x (i.e., the greatest integer less than or
equal to x).
However, multiples of p2 each contribute a second factor of p. Multiples of
3
p contribute a third additional factor of p, and so on. Therefore the highest
power of p that divides n! is given by
blogp nc
n n n X n
+ 2 + 3 + ··· = .
p p p pk
k=1
0.2.10 Exercise 10
Prove for any given positive integer N there exist only finitely many integers
n with ϕ(n) = N where ϕ denotes Euler’s ϕ-function. Conclude in particular
that ϕ(n) tends to infinity as n tends to infinity.
Solution. Fix a value of N > 0, and let A be the set of all solutions n to the
equation ϕ(n) = N . We must show that A is a finite set.
First we will show that for any n ∈ A, there cannot be a prime factor of n
larger than N + 1. For if there are prime factors larger than N + 1, then we
may choose the smallest such prime p. Then if q is any prime factor of n with
q ≥ p, we may write n = q k r, where r is some positive integer relatively prime
to q. Therefore we have
ϕ(n) = ϕ(q k )ϕ(r)

= q k−1 (q − 1)ϕ(r)
≥ q − 1 > N.
But ϕ(n) = N , so this is a contradiction. This shows that all prime factors of
n must be at most N + 1.
Now let p1 , p2 , . . . , pm be all the prime factors less than or equal to N + 1
(note that this set of primes is finite). Then every n ∈ A can be written in the
form
n = p1α1 pα αm
2 · · · pm ,
2
where each αi ≥ 0 and αj > 0 for at least one index j. Now observe that each
αi can be one of only finitely many possible values, since ϕ(psi ) = psi (pi −1) > N
for sufficiently large values of s, and N is the product of each ϕ(pα i ). So the
i
distinct values of n in A must be finite in number, because there are only finitely
many possible primes in their prime factorizations and their exponents can take
only finitely many possible values.
Finally, let M be any positive integer. Since there are only finitely many
values of n such that ϕ(n) ≤ M , we may choose the largest such n. Then
ϕ(m) > M for all m > n, which shows that ϕ(n) tends to infinity as n tends to
infinity.
0.2.11 Exercise 11
Prove that if d divides n then ϕ(d) divides ϕ(n) where ϕ denotes Euler’s ϕ-
function.
Solution. First consider the case where n = pk for some prime number p. Then
if d | n we must have d = p` for some integer ` with 0 ≤ ` ≤ k. So
ϕ(n) = ϕ(pk ) = pk−1 (p − 1) and ϕ(d) = ϕ(p` ) = p`−1 (p − 1).
Now let a = pk−` . Then aϕ(d) = ϕ(n), so ϕ(d) | ϕ(n).

The more general case will follow from the fact that ϕ is a multiplicative
function: Let n be a positive integer having prime factorization
αk
n = p1α1 pα
2 · · · pk ,
2
and suppose d is an integer that divides n. Then d can be written as a prod-

uct of these same prime factors p1 , . . . , pk , provided that we allow some of the
exponents to be zero. That is, we may write
d = pβ1 1 pβ2 2 · · · pβkk with 0 ≤ βi ≤ αi for each i.
Then
αk
ϕ(n) = ϕ(pα α2
1 )ϕ(p2 ) · · · ϕ(pk )
1
(2)
and
ϕ(d) = ϕ(pβ1 1 )ϕ(pβ2 2 ) · · · ϕ(pβkk ). (3)
Now each pβi i divides pα
i ,
i
so from the argument in the first paragraph, we know
that ϕ(pβi i ) | ϕ(pαi
i
) for each i. Therefore we may find an integer ai such that
βi
ϕ(pα
i
i
) = a i ϕ(pi ). Therefore, equations (2) and (3) imply that
ϕ(n) = a1 ϕ(pβ1 1 ) · a2 ϕ(pβ2 2 ) · · · ak ϕ(pβkk )

= (a1 a2 · · · ak )ϕ(d),
so ϕ(d) | ϕ(n).
0.3. Z/N Z: THE INTEGERS MODULO N 11
0.3 Z/nZ: The Integers Modulo n

0.3.1 Exercise 1
Write down explicitly all the elements in the residue classes of Z/18Z.
Solution. The residue classes are
0̄ = {0, 18, −18, 36, −36, . . .},

1̄ = {1, 19, −17, 37, −35, . . .},
2̄ = {2, 20, −16, 38, −34, . . .},
3̄ = {3, 21, −15, 39, −33, . . .},
4̄ = {4, 22, −14, 40, −32, . . .},
5̄ = {5, 23, −13, 41, −31, . . .},
6̄ = {6, 24, −12, 42, −30, . . .},
7̄ = {7, 25, −11, 43, −29, . . .},
8̄ = {8, 26, −10, 44, −28, . . .},
9̄ = {9, 27, −9, 45, −27, . . .},
10 = {10, 28, −8, 46, −26, . . .},
11 = {11, 29, −7, 47, −25, . . .},
12 = {12, 30, −6, 48, −24, . . .},
13 = {13, 31, −5, 49, −23, . . .},
14 = {14, 32, −4, 50, −22, . . .},
15 = {15, 33, −3, 51, −21, . . .},
16 = {16, 34, −2, 52, −20, . . .},
and
17 = {17, 35, −1, 53, −19, . . .}.
0.3.2 Exercise 2
Prove that the distinct equivalence classes in Z/nZ are precisely 0̄, 1̄, 2̄, . . . , n − 1
(use the Division Algorithm).
Proof. Consider the equivalence class k̄. Using the Division Algorithm, we may
find an integer q and an integer r such that
k = qn + r, with 0 ≤ r < n.
Now k ≡ r (mod n) and r is an integer between 0 and n − 1, so this shows that

k̄ = r̄. Thus the equivalence classes in Z/nZ are a subset of {0̄, 1̄, . . . , n −¯ 1}.
Finally, we note that the equivalence classes 0̄, . . . , n − 1 are actually distinct
from each other. For, if not, suppose ā = b̄ where 0 ≤ b ≤ a ≤ n − 1. Then
n | (a − b), and since 0 ≤ a − b ≤ n − 1, we must have a − b = 0 so that a = b.
Therefore the distinct equivalence classes are precisely 0̄, . . . , n − 1.
0.3.3 Exercise 3
Prove that if a = an 10n + an−1 10n−1 + · · · + a1 10 + a0 is any positive integer
then a ≡ an +an−1 +· · ·+a1 +a0 (mod 9) (note that this is the usual arithmetic
rule that the remainder after division by 9 is the same as the sum of the decimal
digits mod 9 – in particular an integer is divisible by 9 if and only if the sum of
its digits is divisible by 9).
Solution. Let a be as stated. Since 10 ≡ 1 (mod 9) we may apply Theorem 3
to write
a ≡ an 1n + an−1 1n−1 + · · · + a1 + a0 (mod 9)
≡ an + an−1 + · · · + a1 + a0 (mod 9).
0.3.4 Exercise 4
Compute the remainder when 37100 is divided by 29.
Solution. 372 = 1369 ≡ 6 (mod 29). Successive squaring then yields
374 ≡ 62 = 36 ≡ 7 (mod 29)
8 2
37 ≡ 7 = 49 ≡ 20 (mod 29)
16 2
37 ≡ 20 = 400 ≡ 23 (mod 29)
32 2
37 ≡ 23 = 529 ≡ 7 (mod 29)
64 2
37 ≡ 7 = 49 ≡ 20 (mod 29).
So
37100 = 3764 3732 374 ≡ 20 · 7 · 7 ≡ 23 (mod 29).
100
Therefore 37 has a remainder of 23 when divided by 29.
0.3.5 Exercise 5
Compute the last two digits of 91500 .
Solution. 91500 = 33000 = 271000 . Now 272 = 729 ≡ 29 (mod 100), and succes-
sive squaring then gives
274 ≡ 292 = 841 ≡ 41 (mod 100),
8 2
27 ≡ 41 = 1681 ≡ 81 (mod 100),
16 2
27 ≡ 81 = 6561 ≡ 61 (mod 100),
32 2
27 ≡ 61 = 3721 ≡ 21 (mod 100),
64 2
27 ≡ 21 = 441 ≡ 41 (mod 100).
At this point the numbers start to repeat, so that 27128 ≡ 81 (mod 100), 27256 ≡
61 (mod 100), and 27512 ≡ 21 (mod 100). Therefore
91500 = 271000 = 27512 27256 27128 2764 2732 278
≡ 21 · 61 · 81 · 41 · 21 · 81 = (1281)(3321)(1701)
≡ 81 · 21 · 1 ≡ 1 (mod 100).
Therefore, the last two digits of 91500 are 01.
0.3.6 Exercise 6
Prove that the squares of the elements in Z/4Z are just 0̄ and 1̄.
Proof.
02 = 0 ≡ 0 (mod 4),
2
1 =1≡1 (mod 4),
22 = 4 ≡ 0 (mod 4),
2
3 =9≡1 (mod 4).
0.3.7 Exercise 7
Prove for any integers a and b that a2 + b2 never leaves a remainder of 3 when
divided by 4 (use the previous exercise).
Proof. a2 and b2 are each either congruent to 0 or to 1, modulo 4. Adding

a2 + b2 then gives four cases:
0+0≡0 (mod 4),

0+1≡1 (mod 4),
1+0≡1 (mod 4),
1+1≡2 (mod 4).
In every case, a2 + b2 never has a remainder of 3 when divided by 4.
0.3.8 Exercise 8
Prove that the equation
a2 + b2 = 3c2 (4)
has no solutions in nonzero integers a, b, and c.
Proof. Consider the equation modulo 4. From the previous exercise, the left-
hand side cannot be congruent to 3. However, the right-hand side is congruent
to either 0 or 3, so therefore both sides must be congruent to 0. That is,
a2 + b2 ≡ c2 ≡ 0 (mod 4).
This immediately implies that c is even. Now, if a is even, then b must be

even, since b2 = c2 − a2 is even. On the other hand, if a is odd, then b must be
odd for the same reason. But if a and b are both odd, then we may find integers
m and n such that
a2 + b2 = (2m + 1)2 + (2n + 1)2

= 4m2 + 4m + 1 + 4n2 + 4n + 1
≡2 (mod 4).
This is impossible, so a, b, and c must all be even.

Now, if possible, suppose that a, b, and c are three positive integers which
satisfy the equation (4). Since all three integers must be even, their squares
each contain a factor of 4. Divide both sides by 4 to get a new equation,
α2 + β 2 = γ 2 ,
where α < a, β < b, and γ < c.

But by the same argument as before, α, β, and γ must be even, so their
squares are divisible by 4 and we can again find an even smaller set of solutions.
This process could be repeated indefinitely, to get smaller and smaller positive
integer solutions. Clearly this is not possible, so there are no solutions in the
nonzero integers.
0.3.9 Exercise 9
Prove that the square of any odd integer always leaves a remainder of 1 when
divided by 8.
Proof. If a is an odd integer, then a can be written as 2k + 1 for some integer

k, and
a2 = (2k + 1)2 = 4k 2 + 4k + 1 = 4k(k + 1) + 1.
Now k(k + 1) must be even, since it is the product of consecutive integers.

Therefore 4k(k + 1) is divisible by 8. Therefore a2 ≡ 1 (mod 8).
0.3.10 Exercise 10
Prove that the number of elements of (Z/nZ)× is ϕ(n) where ϕ denotes the
Euler ϕ-function.
Proof. We will show that the elements in (Z/nZ)× are precisely those residue
classes whose representatives are relatively prime to n.
First suppose that a ∈ (Z/nZ)× and let b be the multiplicative inverse of a
modulo n, so that ab ≡ 1 (mod n). Then n | (ab − 1) so we may find an integer
m such that mn = ab − 1. Rearranging, we get ab − mn = 1. But this shows
that the greatest common divisor of a and n is 1 (if not, we could factor the
left-hand side to get a product of two integers, not both 1, that equals 1, which
is impossible). Therefore any number in (Z/nZ)× must be relatively prime to
n.
Now, for the other direction, suppose that a is any integer relatively prime
to n. Then we can use the Euclidean algorithm to write the common divisor 1
as a linear combination of a and n, that is,
ax + ny = 1, x, y ∈ Z.
But then ax ≡ 1 (mod n), so x is the multiplicative inverse of a modulo n, i.e.,

a ∈ (Z/nZ)× .
Since there are exactly ϕ(n) least residues which are coprime to n, the set
(Z/nZ)× has exactly ϕ(n) elements.
0.3.11 Exercise 11
Prove that if ā, b̄ ∈ (Z/nZ)× , then ā · b̄ ∈ (Z/nZ)× .
Proof. Let ā and b̄ be in (Z/nZ)× as stated. Then ā has a multiplicative inverse

x̄ and b̄ has an inverse ȳ. Then
(āx̄)(b̄ȳ) ≡ 1 · 1 ≡ 1 (mod n).
Rearranging the left-hand side, we see that x̄ȳ is the multiplicative inverse of
āb̄, so that āb̄ ∈ (Z/nZ)× .
0.3.12 Exercise 12
Let n ∈ Z, n > 1, and let a ∈ Z with 1 ≤ a ≤ n. Prove if a and n are
not relatively prime, there exists an integer b with 1 ≤ b < n such that ab ≡ 0
(mod n) and deduce that there cannot be an integer c such that ac ≡ 1 (mod n).
Proof. Let d = (a, n) and let b = n/d. Then b is an integer with 1 ≤ b < n
(since d > 1). Similarly, a/d is also an integer. So we have
n a
ab = a =n ≡0 (mod n).
d d
Now suppose c is such that ac ≡ 1 (mod n). Then abc ≡ b (mod n). But
this is clearly impossible, since abc ≡ 0 (mod n) and b 6≡ 0 (mod n). Therefore
such a c cannot exist.
0.3.13 Exercise 13
Let n ∈ Z, n > 1, and let a ∈ Z with 1 ≤ a ≤ n. Prove that if a and n are
relatively prime then there is an integer c such that ac ≡ 1 (mod n).
Proof. Since (a, n) = 1, we may find integers c and d such that ac + nd = 1.

This implies that ac ≡ 1 (mod n).
0.3.14 Exercise 14
Conclude from the previous two exercises that (Z/nZ)× is the set of elements
ā of Z/nZ with (a, n) = 1 and hence prove Proposition 4. Verify this directly
in the case n = 12.
Solution. From the previous two exercises we know that a and n are relatively
prime if and only if there is an integer c such that ac ≡ 1 (mod n), i.e., if and
only if a has a multiplicative inverse modulo n.
For n = 12, we have the following multiplication table:
0 1 2 3 4 5 6 7 8 9 10 11
0 0 0 0 0 0 0 0 0 0 0 0 0
1 0 1 2 3 4 5 6 7 8 9 10 11
2 0 2 4 6 8 10 0 2 4 6 8 10
3 0 3 6 9 0 3 6 9 0 3 6 9
4 0 4 8 0 4 8 0 4 8 0 4 8
5 0 5 10 3 8 1 6 11 4 9 2 7
6 0 6 0 6 0 6 0 6 0 6 0 6
7 0 7 2 9 4 11 6 1 8 3 10 5
8 0 8 4 0 8 4 0 8 4 0 8 4
9 0 9 6 3 0 6 9 3 0 6 9 3
10 0 10 8 6 4 2 0 10 8 6 4 2
11 0 11 10 9 8 7 6 5 4 3 2 1
The only values which have a multiplicative inverse are 1, 5, 7, and 11, which
are precisely those values which are coprime to 12.
0.3.15 Exercise 15
For each of the following pairs of integers a and n, show that a is relatively
prime to n and determine the multiplicative inverse of ā in Z/nZ.
(a) a = 13, n = 20
Solution. Applying the Euclidean algorithm gives

20 = 1(13) + 7
13 = 1(7) + 6
7 = 1(6) + 1,
so (20, 13) = 1. And we can write
1=7−6
= 7 − (13 − 7)
= 2(7) − 13
= 2(20 − 13) − 13
= 2(20) − 3(13).
So (−3) = 17 is the multiplicative inverse of 13 in Z/20Z.
(b) a = 69, n = 89
Solution. The same procedure will show that (69, 89) = 1 and that ā has
an inverse of 40.
(c) a = 1891, n = 3797
Solution. ā has an inverse of 253.
(d) a = 6 003 722 857, n = 77 695 236 973
Solution. ā has an inverse of 77 695 236 753.

Chapter 1
Introduction to Groups
1.1 Basic Axioms and Examples

1.1.1 Exercise 1
Determine which of the following binary operations are associative:
(a) the operation ? on Z defined by a ? b = a − b
Solution. (1 ? 2) ? 3 = −4 while 1 ? (2 ? 3) = 2, so ? is not associative.
(b) the operation ? on R defined by a ? b = a + b + ab
Solution. ? is associative: let a, b, c be real numbers. Then
(a ? b) ? c = (a + b + ab) ? c
= (a + b + ab) + c + (a + b + ab)c
= a + b + c + ab + ac + bc + abc
= a + (b + c + bc) + a(b + c + bc)
= a ? (b + c + bc)
= a ? (b ? c).
(c) the operation ? on Q defined by a ? b = (a + b)/5
Solution. (5 ? 20) ? 15 = 4 while 5 ? (20 ? 15) = 12/5. Therefore ? is not

associative.
(d) the operation ? on Z × Z defined by (a, b) ? (c, d) = (ad + bc, bd)
Solution. ? is associative: let (a, b), (c, d), (e, f ) be members of Z × Z.
17
18 CHAPTER 1. INTRODUCTION TO GROUPS
Then

(a, b) ? (c, d) ? (e, f ) = (ad + bc, bd) ? (e, f )

= (ad + bc)f + bde, bdf
= (adf + bcf + bde, bdf )

= adf + b(cf + de), bdf
= (a, b) ? (cf + de, df )

= (a, b) ? (c, d) ? (e, f ) .
(e) the operation ? on Q − {0} defined by a ? b = a/b
Solution. (125 ? 25) ? 5 = 1 while 125 ? (25 ? 5) = 25, so ? is not associative.
1.1.2 Exercise 2
Decide which of the binary operations in the preceding exercises are commuta-
tive.
(a) the operation ? on Z defined by a ? b = a − b
Solution. ? is not commutative since, for example, 1 ? 2 = −1 while 2 ? 1 =

1.
(b) the operation ? on R defined by a ? b = a + b + ab
Solution. ? is commutative since, for any a, b ∈ R,
a ? b = a + b + ab
= b + a + ba
= b ? a.
(c) the operation ? on Q defined by a ? b = (a + b)/5
Solution. ? is commutative since + is commutative in Q.
(d) the operation ? on Z × Z defined by (a, b) ? (c, d) = (ad + bc, bd)
Solution. ? is commutative: Let (a, b) and (c, d) be elements of Z × Z.

Then
(a, b) ? (c, d) = (ad + bc, bd)

= (cb + da, db)
= (c, d) ? (a, b).
(e) the operation ? on Q − {0} defined by a ? b = a/b
Solution. ? is not commutative since 1 ? 2 = 1/2 but 2 ? 1 = 2.

1.1. BASIC AXIOMS AND EXAMPLES 19
1.1.3 Exercise 3
Prove that addition of residue classes in Z/nZ is associative (you may assume
it is well defined).
Proof. Let ā, b̄, c̄ be residue classes in Z/nZ. Then by Theorem 3 in Section 0.3
along with the associativity of + in Z, we may write
(ā + b̄) + c̄ = (a + b) + c
= a + (b + c)
= ā + (b̄ + c̄).
So addition of residue classes is associative.
1.1.4 Exercise 4
Prove that multiplication of residue classes in Z/nZ is associative (you may
assume it is well defined).
Proof. As in the previous exercise, this follows from Theorem 3 in Section 0.3
together with the associativity of · in Z.
1.1.5 Exercise 5
Prove for all n > 1 that Z/nZ is not a group under multiplication of residue
classes.
Proof. Let n > 1. Then there is a residue class in Z/nZ which does not contain
0. Call this nonzero residue class ā. Then 0̄ cannot be the identity element in
Z/nZ since ā · 0̄ = 0̄ 6= ā. So suppose the identity element is ē. Then, 0̄ also has
no inverse in Z/nZ, since b̄ · 0̄ = 0̄ 6= ē for any b̄ in Z/nZ. Since the element 0̄
does not have an inverse, Z/nZ is not a group under multiplication.
1.1.6 Exercise 6
Determine which of the following sets are groups under addition:
(a) the set of rational numbers (including 0 = 0/1) in lowest terms whose
denominators are odd
Solution. Let the set be denoted A. Then A is a group having identity 0

and, for each a ∈ A, an inverse −a. To prove this, we need only show that
A is closed under addition.
Suppose a and b are any elements in A. Then we can find integers p, q, r, s
with
p r
a= and b =
q s
in lowest terms with q, s odd. Then we have
ps + rq u
a+b= = ,
qs v
where u and v are integers with u/v in lowest terms. Now, since u/v was
obtained by eliminating common factors, we have u | (ps + rq) and v | qs.
But if 2 | v, then necessarily 2 | qs. But this cannot be, since qs is odd,
being the product of odd integers. Hence A is closed under addition and
is therefore a group.
(b) the set of rational numbers in lowest terms whose denominators are even,
together with 0
Solution. Let A denote the set. Then A is not a group since 3/2 ∈ A but
3 3 6 3
+ = = 6∈ A.
2 2 2 1
(c) the set of rational numbers of absolute value < 1
Solution. Again, this set is not closed under addition since, for example,
3 3
+ > 1.
4 4
Therefore it is not a group.
(d) the set of rational numbers of absolute value ≥ 1 together with 0
Solution. This set is not closed under addition since, for example,
12 8 4
− = 6≥ 1.
5 5 5
Therefore it is not a group.
(e) the set of rational numbers with denominators equal to 1 or 2
Solution. Denote the set by A. Let a and b be arbitrary integers. Then

a, b, a/2, b/2 ∈ A. There are several cases. First,
a b a+b
+ = ∈ A.
1 1 1
Now consider
a b a+b
+ = .
2 2 2
If this fraction is in lowest terms, then it is in A. If not, then there must be
a common factor of 2 and the fraction can be written with a denominator
of 1 and thus is in A.
Finally, consider
a b b a 2a + b
+ = + = .
1 2 2 1 2
As before, this is either in lowest terms, or can be reduced to lowest terms
by dividing the numerator and denominator by 2. In either case, this
number is in A.
Since A is closed under addition, it is easily seen to be a group: the identity
is 1 = 1/1 and the inverse of a/b ∈ A is −a/b.
(f) the set of rational numbers with denominators equal to 1, 2, or 3
Solution. This set is not closed under addition, since

1 1 5
+ = .
2 3 6
Hence this is not a group.
1.1.7 Exercise 7
Let G = {x ∈ R | 0 ≤ x < 1} and for x, y ∈ G let x ? y be the fractional part of
x + y (i.e., x ? y = x + y − [x + y] where [a] is the greatest integer less than or
equal to a). Prove that ? is a well defined binary operation on G and that G is
an abelian group under ? (called the real numbers mod 1).
Proof. Let x, y ∈ G be arbitrary. Then 0 ≤ x + y < 2. There are two cases: if
x + y < 1 then [x + y] = 0 and x ? y ∈ G. On the other hand, if 1 ≤ x + y < 2
then [x + y] = 1 and x ? y = x + y − 1 ∈ G. Therefore ? is a well defined binary
operation on G.
Let x, y, z ∈ G. If x + y < 1 and y + z < 1, then
(x ? y) ? z = (x + y − 0) ? z
= x + y + z − [x + y + z]
= x ? (y + z − 0)
= x ? (y ? z).
On the other hand, if 1 ≤ x + y < 2 and 1 ≤ y + z < 2, then
(x ? y) ? z = (x + y − 1) ? z
= x + y + z − 1 − [x + y + z − 1]
= x ? (y + z − 1)
= x ? (y ? z).
Finally, if 1 ≤ x + y < 2 and 0 ≤ y + z < 1, then [x + y] = 1, [y + z] = 0, and
[x + y + z − 1] = [x + y + z] − 1, so
(x ? y) ? z = (x + y − 1) ? z
= x + y + z − 1 − [x + y + z − 1]
= x + y + z − 1 − [x + y + z] + 1
= x + y + z − [x + y + z]
= x ? (y + z − 0)
= x ? (y ? z).
And the case where x+y < 1 and y+z ≥ 1 is similar. Therefore, ? is associative.
Since 0 ∈ G, G has an identity (x ? 0 = 0 ? x = x for each x in G). And every
element has an inverse: the inverse of 0 is 0, and for nonzero x ∈ G, 1 − x ∈ G
is an inverse since
x ? (1 − x) = x + (1 − x) − [x + (1 − x)] = 1 − 1 = 0.
Therefore G is a group under ?.
1.1.8 Exercise 8
Let G = {z ∈ C | z n = 1 for some n ∈ Z+ }.
(a) Prove that G is a group under multiplication (called the group of roots of
unity in C).
Proof. Let z, w ∈ G so that z n = 1 and wm = 1.

Note that 1 ∈ G since 11 = 1, so G has an identity. And every element
of G is nonzero, so for each z ∈ G we may let z −1 = 1/z so that every
element in G has an inverse (since (1/z)n = 1/z n = 1 so 1/z ∈ G).
By the commutativity of multiplication in C, we have
(zw)nm = z nm wmn = (z n )m (wm )n = 1m 1n = 1
for each m, n ∈ Z+ . Therefore, G is closed under multiplication. And

associativity follows from associativity of multiplication in C.
Therefore G is a group.
(b) Prove that G is not a group under addition.
Proof. G is a not a group under addition since it is not closed: 1 ∈ G but

1 + 1 = 2 6∈ G since there is no n ∈ Z+ with 2n = 1.
1.1.9 Exercise 9
√
Let G = {a + b 2 ∈ R | a, b ∈ Q}.
(a) Prove that G is a group under addition.
Proof. Associativity of
√ + in G follows from associativity of + in R. G has
an identity 0 = 0 + 0 2 and for every p in G we may take q = −p as its
additive inverse. So we need only show that G is closed under addition.
√ √
Let p = a + b 2 and q = c + d 2 with a, b, c, d ∈ Q. Then
√
p + q = a + c + (b + d) 2, where a + c ∈ Q and b + d ∈ Q,
so p + q ∈ G and G is a group.
(b) Prove that the nonzero elements of G are a group under multiplication.
Proof. Again, associativity

√ follows from associativity in R. This time the
identity√is 1 = 1 + 0 2. And for any rational numbers a and b not both
0, a + b 2 ∈ G − {0} and
√
1 a−b 2
√ = √ √
a+b 2 (a + b 2)(a − b 2)
√
a−b 2
= 2
a − 2b2
√

a b
= 2 2
− 2 2
2 ∈ G − {0},
a − 2b a − 2b
so every element in G − {0} has an inverse (note that the denominator

a2 − 2b2 is nonzero since a, b ∈ Q and there is no rational square root of
2).
The set is also closed under multiplication since for any a, b, c, d ∈ Q with
a, b not both 0 and c, d not both 0,
√ √ √
a + b 2 c + d 2 = ac + 2bd + (ad + bc) 2 ∈ G − {0}.
This shows that G − {0} is a group under multiplication.
1.1.10 Exercise 10
Prove that a finite group is abelian if and only if its group table is a symmetric
matrix.
Proof. List the elements of the group in a fixed order along the top row and
first column of the group table. Then the group is abelian if and only if the
i, jth entry in its group table is equal to the j, ith entry, which is true if and
only if the table forms a symmetric matrix.
1.1.11 Exercise 11
Find the orders of each element of the additive group Z/12Z.
Solution. 0̄ has order 1. 1̄ has order 12 since 1 · 1̄, 2 · 1̄, . . ., 11 · 1̄ are nonzero
while 12 · 1̄ = 0̄. Similarly, we find the following orders for the elements:
x 0̄ 1̄ 2̄ 3̄ 4̄ 5̄ 6̄ 7̄ 8̄ 9̄ 10 11
|x| 1 12 6 4 3 12 2 12 3 4 6 12
1.1.12 Exercise 12
Find the orders of the following elements of the multiplicative group (Z/12Z)× :
1̄, −1, 5̄, 7̄, −7, 13.
Solution. We get the following table:
x 1̄ −1 5̄ 7̄ −7 13
|x| 1 2 2 2 2 1
1.1.13 Exercise 13
Find the orders of the following elements of the additive group Z/36Z: 1̄, 2̄, 6̄,
9̄, 10, 12, −1, −10, −18.
Solution. We get the following table:
x 1̄ 2̄ 6̄ 9̄ 10 12 −1 −10 −18
|x| 36 18 6 4 18 3 36 18 2
1.1.14 Exercise 14
Find the orders of the following elements of the multiplicative group (Z/36Z)× :
1̄, −1, 5̄, 13, −13, 17.
Solution. We have the following table:
x 1̄ −1 5̄ 13 −13 17
|x| 1 2 6 3 6 2
1.1.15 Exercise 15
Let G be a group. Prove that
−1 −1
(a1 a2 . . . an )−1 = a−1
n an−1 . . . a1
for all a1 , a2 , . . . , an ∈ G.
Proof. We use induction on n. If n = 1, the result is obvious. Suppose it holds

for n = k, where k ≥ 1. Then for any a1 , . . . , ak+1 in G, we have
(a1 . . . ak ak+1 )(a−1 −1 −1 −1 −1 −1

k+1 ak . . . a1 ) = (a1 . . . ak )(ak+1 ak+1 )(ak . . . a1 )
= (a1 . . . ak )(a−1 −1
k . . . a1 ),
and this is equal to 1 by the induction hypothesis. Therefore (a1 . . . ak+1 )−1 =
a−1 −1
k+1 . . . a1 and the statement holds for all positive integers n.
1.1.16 Exercise 16
Let x be an element of a group G. Prove that x2 = 1 if and only if |x| is either
1 or 2.
Proof. First, if |x| = 1 then x = 1 so x2 = 12 = 1. If |x| = 2, then x2 = 1 by

definition.
For the other direction, suppose x2 = 1. Then |x| ≤ 2. But the order of an
element must be at least 1, so |x| = 1 or |x| = 2.
1.1.17 Exercise 17
Let x be an element of a group G. Prove that if |x| = n for some positive integer
n then x−1 = xn−1 .
Proof. Since |x| = n, we have xn = 1. But xn = xn−1 x = xxn−1 , so xn−1 x = 1

which shows that x−1 = xn−1 .
1.1.18 Exercise 18
Let x and y be elements of a group G. Prove that xy = yx if and only if
y −1 xy = x if and only if x−1 y −1 xy = 1.
Proof. If xy = yx, then y −1 xy = y −1 yx = 1x = x. Multiplying by x−1 then
gives x−1 y −1 xy = 1.
On the other hand, if x−1 y −1 xy = 1, then we may multiply on the left
by x to get y −1 xy = x. Then multiplying on the left by y gives xy = yx as
desired.
1.1.19 Exercise 19
Let x ∈ G for G a group and let a, b ∈ Z+ .
(a) Prove that xa+b = xa xb and (xa )b = xab .
Proof. xa xb consists of a a factors of x, multiplied by b factors of x, for a

total of a + b factors of x. Therefore xa+b = xa xb by definition. Similarly,
(xa )b = xab by the same reasoning.
(b) Prove that (xa )−1 = x−a .
Proof. Since x−a = (x−1 )a , we need to show that (xa )−1 = (x−1 )a . We
use induction on a. For a = 1, the result is trivial. Suppose it holds for
a = k, k ≥ 0. Then
(xk+1 )(x−1 )k+1 = xk (xx−1 )(x−1 )k = xk (x−1 )k ,
which by the induction hypothesis must be 1. Therefore the result holds

for all positive integers a.
(c) Establish part (a) for arbitrary integers a and b (positive, negative, or
zero).
Proof. For any integer a, xa x0 = xa = xa+0 and similarly x0 xa = x0+a .

Now suppose a > 0, b < 0. If a + b > 0, then xa+b x−b = x(a+b)+(−b) = xa
by part (a). Multiplying both sides of this equation on the right by xb
gives xa+b = xa xb as desired. On the other hand, if a + b < 0, then
x−(a+b) xa = x−(a+b)+a = x−b . Multiplying both sides of this equation on
the right by x−a gives x−(a+b) = x−b x−a , so
(xa+b )−1 = x−(a+b) = x−b x−a = (xb )−1 (xa )−1 = (xa xb )−1 .
The last equality follows from part (4) of Proposition 1 in the text. Since
inverses are unique (by the same proposition) we have xa+b = xa xb .
The case where a < 0, b > 0 is entirely similar to the argument above.
Finally, if a and b are both negative, then
xa+b = (x−a−b )−1 = (x−b−a )−1 = (x−b x−a )−1 = xa xb .
This completes the proof.

1.1.20 Exercise 20
For x an element in G a group show that x and x−1 have the same order.
Proof. Suppose |x| = n for finite n. Then xn = 1 so
(x−1 )n = x−n = (xn )−1 = 1−1 = 1,
which shows x−1 has finite order and |x−1 | ≤ |x|. On the other hand, if |x−1 | = k
then −1
xk = (x−1 )−k = (x−1 )k = 1−1 = 1,
so x has finite order and |x| ≤ |x−1 |. This shows that |x| = |x−1 | when either
x or x−1 is of finite order. The only alternative is that x and x−1 are both of
infinite order.
1.1.21 Exercise 21
Let G be a finite group and let x be an element of G of order n. Prove that if
n is odd, then x = (x2 )k for some integer k ≥ 1.
Proof. If n is odd, then we may write n = 2k − 1 for some k ∈ Z+ . Then we

have
xn = x2k−1 = 1.
Multiplying both sides by x then gives
x2k−1 x = x,
so
x = x2k−1+1 = x2k = (x2 )k .
1.1.22 Exercise 22
If x and g are elements of the group G, prove that |x| = |g −1 xg|. Deduce that
|ab| = |ba| for all a, b ∈ G.
Proof. A simple induction argument will show that (g −1 xg)k = g −1 xk g for any
k ∈ Z+ . So if |x| = n, then xn = 1 and we have
(g −1 xg)n = g −1 xn g = g −1 1g = 1,
which shows that g −1 xg is of finite order and |g −1 xg| ≤ |x|. However, if

|g −1 xg| = k, then (g −1 xg)k = 1 so
xk = gg −1 xk gg −1 = g(g −1 xg)k g −1 = g1g −1 = 1,
which shows that x is of finite order and |x| ≤ |g −1 xg|. Therefore |x| = |g −1 xg|.
This also shows that if x is of infinite order, then g −1 xg is of infinite order
and vice versa.
Finally, for any a, b ∈ G,
|ab| = |b(ab)b−1 | = |ba|.

1.1.23 Exercise 23
Suppose x ∈ G for G a group and |x| = n < ∞. If n = st for some positive
integers s and t, prove that |xs | = t.
Proof. Let |x| = n where n = st. Then
1 = xn = xst = (xs )t ,
so |xs | ≤ t. Now suppose |xs | = r. Then (xs )r = xsr = 1. But |x| = st, so we
have sr ≥ st or r ≥ t, which gives |xs | ≥ t. Therefore |xs | = t.
1.1.24 Exercise 24
If a and b are commuting elements of the group G, prove that (ab)n = an bn for
all n ∈ Z.
Lemma. If a and b are commuting elements of a group G, then an b = ban for

all positive integers n.
Proof. We use induction on n. The base case is trivial, so suppose an b = ban

for some positive integer n. Then
an+1 b = aan b = aban = baan = ban+1 ,
which completes the inductive step. Hence an b = ban for all positive n.
Proof of main result. First we will use induction on n to show that (ab)n = an bn
in the case where n is positive. For n = 1, the result is obvious. Suppose the
result is true for n = k, for some positive integer k. Then
(ab)k+1 = (ab)(ab)k = abak bk = aak bbk = ak+1 bk+1 ,
where the second-to-last equality makes use of the above lemma. This shows
that the result holds for all positive integers n.
Next, in the case where n = 0, we get (ab)0 = 1 = a0 b0 .
Finally, using the result from Exercise 1.1.19, we have for any n < 0,
−1 −1
(ab)n = (ba)n = (ba)−n = b−n a−n = an bn .
Therefore the result holds for all integers n.
1.1.25 Exercise 25
Let G be a group. Prove that if x2 = 1 for all x ∈ G then G is abelian.
Proof. For any x ∈ G, we have x = x−1 . Let a, b ∈ G be arbitrary. Then we

have
ab = (ab)−1 = b−1 a−1 = ba.
Here we have made use of property (4) from Proposition 1. This shows that G
is abelian.
1.1.26 Exercise 26
Assume H is a nonempty subset of the group (G, ?) which is closed under the
binary operation on G and is closed under inverses, i.e., for all h and k ∈ H, hk
and h−1 ∈ H. Prove that H is a group under the operation ? restricted to H
(such a subset H is called a subgroup of G).
Proof. (a) Associativity of ? in H follows from associativity of ? in G.
(b) Since H is nonempty, it must have an element a. Then by hypothesis
a−1 ∈ H and therefore aa−1 = e ∈ H, where e denotes the identity of G.
Therefore H has an identity.
(c) For each a ∈ H, a−1 ∈ H by hypothesis so every element of H has an
inverse in H.
This shows that (H, ?) is a group.
1.1.27 Exercise 27
Prove that if x is an element of the group G then {xn | n ∈ Z} is a subgroup of
G (called the cyclic subgroup of G generated by x).
Proof. Let H be the subset stated above. We know H is nonempty since x0 = e
is a member of H. If a = xm and b = xn are any two elements in H, then
ab = xm xn = xm+n by Exercise 1.1.19. So ab ∈ H which shows that H is
closed under the binary operation of G. H is also closed under inverses, since
a−1 = (xm )−1 = x−m ∈ H. Therefore, by the previous exercise, H is a subgroup
of G.
1.1.28 Exercise 28
Let (A, ?) and (B, ) be groups and let A × B be their direct product. Verify
all the group axioms for A × B:
(a) prove that the associative law holds
(b) prove that (1, 1) is the identity of A × B, and
(c) prove that the inverse of (a, b) is (a−1 , b−1 ).
Proof. (a) For all (ai , bi ) ∈ A × B with i = 1, 2, 3 we have
(a1 , b1 )[(a2 , b2 )(a3 , b3 )] = (a1 , b1 )(a2 a3 , b2 b3 )

= a1 (a2 a3 ), b1 (b2 b3 )

= (a1 a2 )a3 , (b1 b2 )b3
= (a1 a2 , b1 b2 )(a3 , b3 )
= [(a1 , b1 )(a2 b2 )](a3 , b3 ).
This shows associativity.
(b) For any (a, b) ∈ A × B we have
(a, b)(1, 1) = (a ? 1, b 1) = (a, b).
Therefore (1, 1) is the identity of A × B.
(c) For any (a, b) ∈ A × B,
(a, b)(a−1 , b−1 ) = (a ? a−1 , b b−1 ) = (1, 1),
so (a, b)−1 = (a−1 , b−1 ).
1.1.29 Exercise 29
Prove that A × B is an abelian group if and only if both A and B are abelian.
Proof. First, if A and B are abelian and if (a, b) and (c, d) are any members of
A × B, then
(a, b)(c, d) = (ac, bd) = (ca, db) = (c, d)(a, b),
so A × B is abelian.
For the other direction, suppose A × B is abelian. Let a1 , a2 ∈ A and
b1 , b2 ∈ B. Then since A × B is abelian, we have
(a1 a2 , b1 b2 ) = (a1 , b1 )(a2 , b2 ) = (a2 , b2 )(a1 , b1 ) = (a2 a1 , b2 b1 ).
Equating components shows that a1 a2 = a2 a1 and b1 b2 = b2 b1 . Therefore A

and B are both abelian.
1.1.30 Exercise 30
Prove that the elements (a, 1) and (1, b) of A × B commute and deduce that the
order of (a, b) is the least common multiple of |a| and |b|.
Proof. If a ∈ A and b ∈ B, then (a, 1), (1, b) ∈ A × B and
(a, 1)(1, b) = (a1, 1b) = (a, b) = (1a, b1) = (1, b)(a, 1).
Now, we will show by induction on n that (a, b)n = (an , bn ) for any positive
integer n. The base case is obvious. Suppose (a, b)k = (ak , bk ) for some k > 0.
Then
(a, b)k+1 = (a, b)k (a, b) = (ak , bk )(a, b) = (ak+1 , bk+1 )
so the statement holds for all n ∈ Z+ . This implies that |(a, 1)| = |a| and
|(1, b)| = |b|.
Let the least common multiple of |a| and |b| be ` and suppose |(a, b)| = k.
Then m|a| = n|b| = ` for some integers m and n. Since (a, 1) and (1, b) commute,
we have
(a, b)` = ((a, 1)(1, b))`

= (a, 1)` (1, b)`
= (a, 1)m|a| (1, b)n|b|
= (1, 1)(1, 1)
= (1, 1).
So k ≤ `. Now since (a, b)k = (1, 1), we have ak = 1 and bk = 1. This implies |a|
divides k and |b| divides k. So k is a common multiple of |a| and |b|. Therefore
` ≤ k. This shows that ` = k, which completes the proof.
1.1.31 Exercise 31
Prove that any finite group G of even order contains an element of order 2.
Proof. Define t(G) to be the set {g ∈ G | g 6= g −1 }. Then t(G) must have an

even number of elements because g ∈ t(G) if and only if g −1 ∈ t(G) and any
such g, g −1 must be distinct. Since G also has an even number of elements, the
set G − t(G) has an even number of elements.
Now G − t(G) is nonempty since the identity e 6∈ t(G). Therefore there is a
nonidentity element a ∈ G − t(G). But since a 6∈ t(G), we have a = a−1 so that
a2 = e but a 6= e. Thus a is an element of order 2, completing the proof.
1.1.32 Exercise 32
If x is an element of finite order n in a group G, prove that the elements
1, x, x2 , . . . , xn−1 are all distinct. Deduce that |x| ≤ |G|.
Proof. Suppose the contrary, so that xs = xt for 1 ≤ s < t < n. Then xt x−s =
xt−s = 1. But 1 ≤ t − s < n, so |x| < n, a contradiction. This shows that each
of 1, x, . . . , xn−1 are distinct so that |G| ≥ |x|.
1.1.33 Exercise 33
Let x be an element of finite order n in the group G.
(a) Prove that if n is odd then xi 6= x−i for all i = 1, 2, . . . , n − 1.
Proof. Fix a positive integer i < n. Then xi xn−i = 1 so x−i = xn−i . By

the previous exercise, if i 6= n − i, then xi 6= xn−i . Since inverses are
unique, we have in this case that xi 6= x−i .
Now, if n is odd, then necessarily i 6= n − i, so xi 6= x−i .
(b) Prove that if n = 2k and 1 ≤ i < n then xi = x−i if and only if i = k.
Proof. For any 1 ≤ i < n such that i 6= k, we have i 6= n − i so xi 6= x−i

by the argument in the first part of the problem. And if i = k, then
xi xi = x2k = xn = 1, so xi = x−i in this case (and only this case).
1.1.34 Exercise 34
If x is an element of infinite order in the group G, prove that the elements xn ,
n ∈ Z are all distinct.
Proof. Let x have infinite order and suppose xm = xn with n ≤ m. Then

xm−n = 1. If m − n > 0 then x has finite order, which is a contradiction.
Therefore m = n. This shows that each xm is distinct.
1.1.35 Exercise 35
If x is an element of finite order n in a group G, use the Division Algorithm
to show that any integral power of x equals one of the elements in the set
{1, x, x2 , . . . , xn−1 }.
Proof. Let x have order n and suppose k is any integer.
Since n must be greater than 0, we may use the Division Algorithm to find
integers q and r such that k = qn + r, where 0 ≤ r < n. Then
xk = xqn+r = (xn )q xr = 1xr = xr , where 0 ≤ r < n,
which completes the proof.
1.1.36 Exercise 36
Assume G = {1, a, b, c} is a group of order 4 with identity 1. Assume also that
G has no elements of order 4. Use the cancellation laws to show that there is a
unique group table for G. Deduce that G is abelian.
Proof. From the previous exercises, we know that each element in G besides 1
either has order equal to 2 or 3. By Exercise 1.1.31 there is an element in G
with order 2. Without loss of generality, we may suppose that this element is a.
Then a2 = 1. Now ab 6= 1 since that would imply b = a−1 = a. Next,
ab 6= a since otherwise the cancellation law would give b = 1. Similarly, ab 6= b
since otherwise a = 1. So we must have ab = c. Using the same reasoning,
we must have ba = c and ac = ca = b. Using this information, we have
b2 = (ca)(ac) = c(a2 )c = c2 .
Now, if b2 6= 1 then we must have |b| = 3 so that b3 = 1. Then
a = ab3 = (ab)b2 = c3 .
But since c3 = a 6= 1, we have |c| = 2 so 1 = c2 = b2 and |b| = 2, a contradiction.

This shows that b2 = c2 = 1. Finally,
bc = (ac)c = ac2 = a,
and similarly cb = a.
Combining all of this information gives the following group table:
1 a b c
1 1 a b c
a a 1 c b
b b c 1 a
c c b a 1
And we can readily see that G is abelian.
1.2 Dihedral Groups

In these exercises, D2n has the usual presentation
D2n = hr, s | rn = s2 = 1, rs = sr−1 i.
1.2.1 Exercise 1
Compute the order of each of the elements in the following groups:
(a) D6
Solution. The elements of D6 are 1, r, r2 , s, sr, and sr2 . We have |1| = 1,

|r| = 3, |r2 | = 3, and |s| = 2. Since (sr)2 = srsr = s2 r−1 r = 1 and
(sr2 )2 = sr2 sr2 = s2 r−2 r2 = 1, we have |sr| = |sr2 | = 2.
(b) D8
Solution. Again we have |1| = 1, |r| = 4, |r2 | = 2, and |r3 | = 4. For k

with 0 ≤ k ≤ 3, we have (srk )2 = srk srk = s2 r−k rk = 1 so |srk | = 2.
(c) D10
Solution. Since 5 is prime, we have for each k with 0 ≤ k ≤ 4,
|rk | = 5 and |srk | = 2.
1.2.2 Exercise 2
Use the generators and relations above to show that if x is any element of D2n
which is not a power of r, then rx = xr−1 .
Proof. Since any element of D2n that is not a power of r has the form srk for
some integer k, we have
rx = rsrk = sr−1 rk = sr−k = srk r−1 = xr−1 .
1.2.3 Exercise 3
Use the generators and relations above to show that every element of D2n which
is not a power of r has order 2. Deduce that D2n is generated by the two elements
s and sr, both of which have order 2.
Proof. As in the previous exercise, such elements have the form srk . srk is
distinct from the identity, and
(srk )2 = srk srk = s2 r−k rk = 1,
so |srk | = 2.
Now, the elements of D2n are 1, r, r2 , . . . , rn , and s, sr, . . . , srn . Each rk
can be written as (s(sr))k , and each srk can be written as s(s(sr))k , so D2n is
generated by {s, sr}, each element of which has order 2.
1.2. DIHEDRAL GROUPS 33
1.2.4 Exercise 4
If n = 2k is even and n ≥ 4, show that z = rk is an element of order 2 which
commutes with all elements of D2n . Show also that z is the only nonidentity
element of D2n which commutes with all elements of D2n .
Proof. r has order n, so by Exercise 1.1.33, we know that z = z −1 , that is,

rk = r−k . Let x ∈ D2n be arbitrary. Then x can be written either r` or sr` for
some integer `. In the first case,
zx = rk r` = rk+` = r` rk = xz,
and in the second case,
zx = rk sr` = sr−k r` = srk r` = sr` rk = xz.
This shows that z commutes with each element of D2n .

Now suppose z 0 is any nonidentity element in D2n which commutes with
every element in D2n . Then in particular z 0 commutes with s. So if z 0 = rt for
some integer t, then z 0 s = sz 0 , and
z 0 s = rt s = sr−t .
Therefore rt = r−t . By Exercise 1.1.33, we must have t = k. On the other

hand, if z 0 = srt , then
z 0 s = srt s = s2 r−t = r−t .
So srt = r−t , but this is impossible, since a reflection cannot also be a rotation.
Therefore z 0 = z and z is the only nonidentity element which commutes with
all elements in the group.
1.2.5 Exercise 5
If n is odd and n ≥ 3, show that the identity is the only element of D2n which
commutes with all elements of D2n .
Proof. The proof is essentially the same as in the previous exercise, except the
odd case from Exercise 1.1.33 is used instead of the even one.
1.2.6 Exercise 6
Let x and y be elements of order 2 in any group G. Prove that if t = xy then
tx = xt−1 (so that if n = |xy| < ∞ then x, t satisfy the same relations in G as
s, r do in D2n ).
Proof. Note that x = x−1 and y = y −1 . If t = xy then
tx = xyx = xy −1 x−1 = x(xy)−1 = xt−1 .

1.2.7 Exercise 7
Show that ha, b | a2 = b2 = (ab)n = 1i gives a presentation in D2n in terms
of the two generators a = s and b = sr of order 2 computed in Exercise 1.2.3
above.
Proof. Suppose a2 = b2 = (ab)n = 1. Then s2 = a2 = 1 and rn = (s2 r)n =

(ab)n = 1. Since b2 = 1, we have srsr = 1. Multiplying each side of this
equation on the right by r−1 and then on the left by s gives s2 (rs)1 = sr−1 or
rs = sr−1 . This shows that the relations s2 = rn = 1 and rs = sr−1 follow
from the relations for a and b.
Now suppose s2 = rn = 1 and rs = sr−1 . Then a2 = s2 = 1, b2 = srsr =
2 −1
s r r = 1, and (ab)n = (s(sr))n = (s2 r)n = rn = 1. Therefore the relations
for a and b follow from those for r and s, so that the above is a presentation for
D2n in terms of a and b.
1.2.8 Exercise 8
Find the order of the cyclic subgroup of D2n generated by r.
Solution. Let G = hri be the cyclic subgroup of D2n generated by r. Then

each element of G can be written rk for some integer k. If k > 0 then rk is a
clockwise rotation about the origin by 2kπ/n radians. If k < 0, then rk is a
rotation counterclockwise by −2kπ/n radians. If |k| ≥ n, then the rotation is
equivalent to a rotation r` where 0 ≤ ` < n. And 1, r, r2 , . . . , rn−1 are distinct,
so G is given by
G = {1, r, r2 , . . . , rn−1 },
and we have |G| = n.
1.2.9 Exercise 9
Let G be the group of rigid motions in R3 of a tetrahedron. Show that |G| = 12.
Proof. A tetrahedron has 4 vertices. Label them from 1 to 4. Then a rigid

motion in G can send vertex 1 to 4 possible places. Once the new position of
vertex 1 has been chosen, there are three adjacent vertices at which to place
vertex 2. The positions of the remaining two vertices will then be completely
determined by the positions of the first two. Therefore there are 4(3) = 12
possible symmetries, so |G| = 12.
1.2.10 Exercise 10
Let G be the group of rigid motions in R3 of a cube. Show that |G| = 24.
Proof. A cube has 8 vertices, and each vertex has 3 adjacent vertices. So there
are 8 possibilities for the position of the first vertex, followed by 3 possibilities for
the position of the second, resulting in 8(3) = 24 symmetries. So |G| = 24.
1.2.11 Exercise 11
Let G be the group of rigid motions in R3 of an octahedron. Show that |G| = 24.
Proof. An octahedron has 6 vertices and each vertex has 4 adjacent vertices.
So, using the same reasoning as in the previous two exercises, we get |G| =
6(4) = 24.
1.2.12 Exercise 12
Let G be the group of rigid motions in R3 of a dodecahedron. Show that
|G| = 60.
Proof. We have 20 vertices, and each vertex has 3 neighboring vertices. So

|G| = 20(3) = 60.
1.2.13 Exercise 13
Let G be the group of rigid motions in R3 of an icosahedron. Show that |G| = 60.
Proof. We have 12 vertices, with each vertex adjacent to 5 vertices, giving |G| =
12(5) = 60.
1.2.14 Exercise 14
Find a set of generators for Z.
Solution. Z is generated by {1} since each n ∈ Z can be written as n1 = n.
1.2.15 Exercise 15
Find a set of generators and relations for Z/nZ.
Proof. Similar to the previous exercise, {1̄} can generate Z/nZ since every el-
ement can be expressed as a repeated addition of 1̄. 1̄ satisfies the relation
n1̄ = 0̄.
1.2.16 Exercise 16
Show that the group hx1 , y1 | x21 = y12 = (x1 y1 )2 = 1i is the dihedral group D4
(where x1 may be replaced by the letter r and y1 by s).
Proof. D4 has the usual presentation hr, s | r2 = s2 = 1, rs = sr−1 i. Since

r = r−1 , that last relation become rs = sr. Let x1 = r and y1 = s. Then
rs = sr implies (x1 y1 )2 = (rs)2 = rsrs = r2 s2 = 1. On the other hand, if
(rs)2 = 1, then rsrs = 1 and, multiplying on the left by r and on the right by s,
this becomes sr = rs. So the relations rs = sr and (x1 y1 )2 = 1 are equivalent.
Therefore the above group is D4 .
1.2.17 Exercise 17
Let X2n be the group with presentation
X2n = hx, y | xn = y 2 = 1, xy = yx2 i.
(a) Show that if n = 3k then X2n has order 6, and it has the same generators
and relations as D6 when x is replaced by r and y by s.
Proof. Let n = 3k. Suppose xn = y 2 = 1 and that xy = yx2 . Note that,

as shown in the text,
x = xy 2 = yx2 y = yxyx2 = y 2 x4 = x4
and the cancellation law implies x3 = 1. Letting x = r and y = s, we then

have r3 = s2 = 1 and
rs = xy = yx2 = sr2 = sr−1 .
Now suppose that r3 = s2 = 1 and rs = sr−1 . Then
xn = x3k = (r3 )k = 1k = 1,
y 2 = s2 = 1, and
xy = rs = sr−1 = sr2 = yx2 .
Since the generators and relations are the same, X2n is D6 and has order
6.
(b) Show that if (3, n) = 1, then x satisfies the additional relation: x = 1. In

this case deduce that X2n has order 2.
Proof. Using the same argument as in the previous part, we must have
x3 = 1. If (3, n) = 1 then either n = 3k + 1 or n = 3k + 2 for some integer
k. If n = 3k + 1 then
x = 1k x = (x3 )k x = x3k x = x3k+1 = xn = 1,
and if n = 3k + 2 then
x−1 = 1k+1 x−1 = (x3 )k+1 x−1 = x3k+3 x−1 = x3k+2 = xn = 1.
But if x−1 = 1 then x = 1. In either case, the relation x = 1 holds so X2n

is the set {1, y} with the relation y 2 = 1, so |X2n | = 2.
1.2.18 Exercise 18
Let Y be the group with presentation
Y = hu, v | u4 = v 3 = 1, uv = v 2 u2 i.
(a) Show that v 2 = v −1 .

Proof. v(v 2 ) = v 3 = 1 which implies v 2 = v −1 .
(b) Show that v commutes with u3 .
Proof. Since
v 2 u3 v = (v 2 u2 )(uv) = (uv)(v 2 u2 ) = uv 3 u2 = u3 ,
we have
vu3 = v(v 2 u3 v) = v 3 u3 v = u3 v,
so v commutes with u3 .
(c) Show that v commutes with u.
Proof. Since u4 = 1, we have u9 = u4 u4 u = u. And since v commutes

with u3 we have
uv = u9 v = u6 vu3 = u3 vu6 = vu9 = vu.
Therefore v commutes with u.
(d) Show that uv = 1.
Proof. Since u and v commute, we get
uv = (uv)(u4 v 3 ) = u5 v 4 = (v 2 u2 )(u3 v 2 ) = (uv)(u3 v 2 ) = u4 v 3 = 1.
(e) Show that u = 1, deduce that v = 1, and conclude that Y = 1.
Proof. Since u4 v 3 = 1 we have
1 = u4 v 3 = u3 (uv)v 2 = u3 v 2 = u2 (uv)v = u2 v = u(uv) = u.
Then v = uv = 1 so that 1 is the only element of Y . Y is therefore the

trivial group of order 1.
1.3 Symmetric Groups

1.3.1 Exercise 1
Let σ be the permutation
1 7→ 3 2 7→ 4 3 7→ 5 4 7→ 2 5 7→ 1
and let τ be the permutation
1 7→ 5 2 7→ 3 3 7→ 2 4 7→ 4 5 7→ 1.
Find the cycle decompositions of each of the following permutations: σ, τ , σ 2 ,
στ , τ σ, and τ 2 σ.
Solution. Applying the permutations from right to left, we get
σ = (1 3 5)(2 4)
τ = (1 5)(2 3)
σ 2 = (1 5 3)
στ = (2 5 3 4)
τ σ = (1 2 4 3)
and
τ 2 σ = (1 3 5)(2 4).
1.3.2 Exercise 2
Let σ be the permutation
1 7→ 13 2 7→ 2 3 7→ 15 4 7→ 14 5 7→ 10
6 7→ 6 7 7→ 12 8 7→ 3 9 7→ 4 10 7→ 1
11 7→ 7 12 7→ 9 13 7→ 5 14 7→ 11 15 7→ 8
and let τ be the permutation
1 7→ 14 2 7→ 9 3 7→ 10 4 7→ 2 5 7→ 12
6 7→ 6 7 7→ 5 8 7→ 11 9 7→ 15 10 7→ 3
11 7→ 8 12 7→ 7 13 7→ 4 14 7→ 1 15 7→ 13.
Find the cycle decomposition of the following permutations: σ, τ , σ 2 , στ , τ σ,
and τ 2 σ.
Solution. We find
σ = (1 13 5 10)(3 15 8)(4 14 11 7 12 9)
τ = (1 14)(2 9 15 13 4)(3 10)(5 12 7)(8 11)
σ 2 = (1 5)(3 8 15)(4 11 12)(7 9 14)(10 13)
στ = (1 11 3)(2 4)(5 9 8 7 10 15)(13 14)
τ σ = (1 4)(2 9)(3 13 12 15 11 5)(8 10 14)
and
τ 2 σ = (1 2 15 8 3 4 14 11 12 13 7 5 10).
1.3. SYMMETRIC GROUPS 39
1.3.3 Exercise 3
For each of the permutations whose cycle decompositions were computed in the
preceding two exercises compute its order.
Solution. For the first exercise, σ = (1 3 5)(2 4), σ 2 = (1 5 3), σ 3 = (2 4), σ 4 =

(1 3 5), σ 5 = (1 5 3)(2 4) and σ 6 = 1. Therefore |σ| = 6. Similarly, τ = (1 5)(2 3)
and τ 2 = 1, so |τ | = 2. σ 2 is a 3-cycle and so has order 3, and στ and τ σ are
both 4-cycles and so have order 4. Lastly, τ 2 σ = σ so |τ 2 σ| = 6.
For the second exercise, we could proceed in the same way. Or we could
observe that, since a t-cycle has order t, the order of a product of disjoint cycles
will be the least common multiple of the lengths of each cycle. This gives
|σ| = [3, 4, 6] = 12,

|τ | = [2, 3, 5] = 30,
|σ 2 | = [2, 3] = 6,
|στ | = [2, 3, 6] = 6,
|τ σ| = [2, 3, 6] = 6,
and
|τ 2 σ| = 13.
1.3.4 Exercise 4
Compute the order of each of the elements in the following groups:
(a) S3
Solution. All elements in S3 can be written as a single t-cycle, with t being

the order of the element:
Permutation Order in S3
1 1
(1 2) 2
(1 3) 2
(2 3) 2
(1 2 3) 3
(1 3 2) 3
(b) S4
Solution. The order of each element in S4 is simply the least common

multiple of the lengths of each cycle in its cycle decomposition:
Permutation Order Permutation Order Permutation Order

1 1 (1 2 4) 3 (1 3)(2 4) 2
(1 2) 2 (1 3 4) 3 (1 4)(2 3) 2
(1 3) 2 (2 3 4) 3 (1 2 3 4) 4
(1 4) 2 (1 3 2) 3 (1 2 4 3) 4
(2 3) 2 (1 4 2) 3 (1 3 2 4) 4
(2 4) 2 (1 4 3) 3 (1 3 4 2) 4
(3 4) 2 (2 4 3) 3 (1 4 2 3) 4
(1 2 3) 3 (1 2)(3 4) 2 (1 4 3 2) 4
1.3.5 Exercise 5
Find the order of (1 12 8 10 4)(2 13)(5 11 7)(6 9).
Solution. Since the cycles are disjoint, the order of this element in S13 is the
least common multiple of the cycle lengths: [2, 3, 5] = 30.
1.3.6 Exercise 6
Write out the cycle decomposition of each element of order 4 in S4 .
Solution. See Exercise 1.3.4.
1.3.7 Exercise 7
Write out the cycle decomposition of each element of order 2 in S4 .
Solution. See Exercise 1.3.4.
1.3.8 Exercise 8
Prove that if Ω = {1, 2, 3, . . . } then SΩ is an infinite group.
Proof. Let n be any positive integer and consider the permutation σn which
sends 2n − 1 to 2n and sends 2n to 2n − 1, while fixing all other elements in Ω.
Clearly σn ∈ SΩ .
Now, if i and j are distinct positive integers, then the numbers 2i − 1,
2i, 2j − 1, 2j are distinct from one another, so that σi and σj have cycle
decompositions that are disjoint. Thus σ1 , σ2 , . . . , σn , . . . are distinct elements
in SΩ , and therefore SΩ is infinite.
1.3.9 Exercise 9
(a) Let σ be the 12-cycle (1 2 3 4 5 6 7 8 9 10 11 12). For which positive
integers i is σ i also a 12-cycle?
Solution. By applying σ twice we can determine that
σ 2 = (1 3 5 7 9 11)(2 4 6 8 10 12).
So σ 2 is not a 12-cycle. In this way we can also determine that σ 3 and σ 4

are also not 12-cycles. However,
σ 5 = (1 6 11 4 9 2 7 12 5 10 3 8)
so σ 5 is a 12-cycle.
Continuing in this way, we can see that σ 6 consists of a product of 2-cycles,
σ 7 is a 12-cycle, σ 8 is a product of 3-cycles, σ 9 is a product of 4-cycles,
σ 10 is a product of 6-cycles, and σ 11 is a 12-cycle. And higher powers will
simply repeat the pattern.
Therefore, σ i is a 12-cycle for i = 1, 5, 7, 11 as well as any integers which
have a remainder of 1, 5, 7, or 11 when divided by 12. We can also
characterize these values as being precisely those values of i for which
(12, i) = 1.
(b) Let τ be the 8-cycle (1 2 3 4 5 6 7 8). For which positive integers i is τ i also
an 8-cycle?
Solution. As in the previous part, 8-cycles will be formed from any ex-
ponent i which is coprime to 8, that is, any i such that (8, i) = 1. This
means that i = 1, 3, 5, 7 or any congruent values modulo 8.
(c) Let ω be the 14-cycle (1 2 3 4 5 6 7 8 9 10 11 12 13 14). For which positive

integers i is ω i also a 14-cycle.
Solution. Again, it is easy to verify that values of i for which (14, i) = 1

will produce 14-cycles. So i = 1, 3, 5, 9, 11, 13 or congruent values modulo
14.
1.3.10 Exercise 10
Prove that if σ is the m-cycle (a1 a2 . . . am ), then for all i ∈ {1, 2, . . . , m},
σ i (ak ) = ak+i , where k + i is replaced by its least positive residue mod m.
Deduce that |σ| = m.
Proof. Fix a positive integer m. We will use induction on i to show that σ i (ak ) =
ak+i for each positive integer i. Since σ cyclically permutes a1 , . . . , am , we have
σ(ak ) = ak+1 for each k (taking am+1 = a1 ), so the base case is satisfied.
Now suppose σ i (ak ) = ak+i for some positive integer i. Then
σ i+1 (ak ) = σ(σ i (ak ))

= σ(ak+i )
= ak+i+1 ,
again replacing k + i and k + i + 1 with their least positive residues mod m.

This completes the inductive step, so σ i (ak ) = ak+i for each integer i > 0.
Finally, if 1 ≤ i < m then σ i sends a1 to a1+i 6= a1 so that σ i is not the
identity. But σ m sends ak to ak+m = ak for each k. Therefore σ m = 1, which
shows that |σ| = m.
1.3.11 Exercise 11
Let σ be the m-cycle (1 2 . . . m). Show that σ i is also an m-cycle if and only if
i is relatively prime to m.
Proof. Fix a value for i. For the remainder of the proof, given a variable k,
let k ∗ denote the least positive residue of k modulo m. That is, let k ∗ be the
smallest positive integer such that k ∗ ≡ k (mod m).
Now, if (i, m) = 1, then the residues i∗ , (2i)∗ , . . . , ((m − 1)i)∗ must be
distinct. To see this, note that i has a multiplicative inverse (by Proposition 4
of Section 0.3), so if s and t are integers with si ≡ ti (mod m), it follows that
s ≡ t (mod m). Now, observe that σ i (m) = i∗ , σ i (i∗ ) = (2i)∗ , and in general,
σ i ((ki)∗ ) = ((k + 1)i)∗ . So σ i is the m-cycle
σ i = (m i∗ (2i)∗ (3i)∗ . . . ((m − 1)i)∗ ).
To prove the other direction, suppose σ i is an m-cycle and let d = (i, m).
Then there are integers x and y such that dx = i and dy = m. Then
(σ i )y = (σ dx )y = (σ dy )x = (σ m )x = 1x = 1.
Therefore |σ i | ≤ y. But σ i is an m-cycle, so its order is m. Therefore y = m

and d = 1. Hence i is relatively prime to m.
1.3.12 Exercise 12
(a) If τ = (1 2)(3 4)(5 6)(7 8)(9 10) determine whether there is an n-cycle σ
(n ≥ 10) with τ = σ k for some integer k.
Solution. Consider the n-cycle
σ = (1 3 5 7 9 2 4 6 8 10).
Then σ 5 = τ .
(b) If τ = (1 2)(3 4 5) determine whether there is an n-cycle (n ≥ 5) with

τ = σ k for some integer k.
Solution. Suppose that it is possible, and let σ be an n-cycle such that

σk = τ .
If n > 5 then σ k must fix 6, 7, . . . . But if σ is an n-cycle, then the only
way σ k can fix any of these values is if it fixes every value, that is, if
σ k = 1 6= τ . Therefore we can suppose that n = 5.
Now since σ k is not an n-cycle, we know by the previous exercise that k
is not relatively prime to n. But n = 5 is prime, so 5 | k and there is an
integer ` such that k = 5`. Then σ k = (σ 5 )` = 1` = 1 6= τ . This is a
contradiction, so our assumption that σ exists was invalid.
1.3.13 Exercise 13
Show that an element has order 2 in Sn if and only if its cycle decomposition is
a product of commuting 2-cycles.
Proof. Let n be a positive integer and suppose σ ∈ Sn has order 2. Let i, j
be distinct integers in {1, 2, . . . , n} such that σ(i) = j. Then since σ 2 = 1, we
must have σ(j) = i. Thus (i j) is a cycle in the cycle decomposition of σ. And
this is true for any such integers, so that no cycle in the decomposition of σ has
length more than 2. Thus we can write σ as a product of disjoint (and hence
commuting) 2-cycles.
Now suppose that σ is a member of Sn such that its cycle decomposition is
a product of commuting 2-cycles, so that
σ = (a1 b1 )(a2 b2 )(a3 b3 ) · · · (ak bk ).
Since each cycle commutes, we have
σ 2 = (a1 b1 )2 (a2 b2 )2 · · · (ak bk )2 = 1k = 1.
Since σ is not the identity, |σ| = 2.
1.3.14 Exercise 14
Let p be a prime. Show that an element has order p in Sn if and only if its
cycle decomposition is a product of commuting p-cycles. Show by an explicit
example that this need not be the case if p is not prime.
Proof. This is a generalization of the previous exercise, and the proof will be
similar. Fix a positive integer n.
Suppose that σ ∈ Sn has order p. Write down the cycle decomposition of σ,
σ = τ1 τ2 · · · τk ,
where each τi is a cycle and τi is disjoint from τj when i 6= j. Since these cycles
are disjoint, they commute with each other and we can write
1 = σ p = τ1p τ2p · · · τkp .
Since the original cycles were disjoint, it follows that τip and τjp are disjoint for
i 6= j, and we must have τip = 1 for each i. This implies that the length of
the cycle τi divides p. But p is prime, so τi is either a p-cycle or the identity.
Therefore σ is the product of commuting p-cycles.
To prove the other direction, suppose that σ ∈ Sn can be written as a
product of commuting p-cycles for p a prime, so that
σ = τ1 τ2 · · · τk
with each τi a p-cycle. Since the cycles commute, we have
σ p = τ1p τ2p · · · τkp = 1k = 1.
So |σ| ≤ p. On the other hand, since τ is a p-cycle, τ t 6= 1 for any positive
integer t less than p. So σ t cannot be the identity permutation. Therefore
|σ| = p.
Lastly, suppose p is not prime. For example, take p = 6 and n = 6. Then
σ = (1 2)(3 4 5) has order 6 but it cannot be written as a product of commuting
6-cycles.
1.3.15 Exercise 15
Prove that the order of an element in Sn equals the least common multiple of
the lengths of the cycles in its cycle decomposition.
Proof. Let σ ∈ Sn have the cycle decomposition
σ = τ1 τ2 τ3 · · · τk ,
where each τi is a cycle and the cycles are pairwise disjoint (and therefore
commute). Suppose |σ| = n. Then
1 = σ n = τ1n τ2n · · · τkn ,
which implies that τin = 1 for each i (the τi ’s are disjoint, so if any τin 6= 1 then
σ n 6= 1). So if τi is a t-cycle, it follows that t | n. Therefore n is a common
multiple of the lengths of each cycle in the cycle decomposition of σ.
On the other hand, if m is any common multiple of these lengths, then
σ m = τ1m · · · τkm = 1, so we must have n ≤ m which shows that n is the least
common multiple of the cycle lengths.
1.3.16 Exercise 16
Show that if n ≥ m then the number of m-cycles in Sn is given by
n(n − 1)(n − 2) · · · (n − m + 1)
. (1.1)
m
Proof. We count the number of ways to form an m-cycle. There are n choices for
the value in the first position, n − 1 choices for the value in the second position,
. . . , and (n − m + 1) choices for the mth position. However, each cycle can be
represented in m different ways, depending on the choice of starting value. So
the actual number of distinct m-cycles is given by the expression (1.1).
1.3.17 Exercise 17
Show that if n ≥ 4 then the number of permutations in Sn which are the product
of two disjoint 2-cycles is n(n − 1)(n − 2)(n − 3)/8.
Proof. Using the same reasoning as in the previous exercise, there are n(n−1)/2
ways to choose the first 2-cycle, and there are (n − 2)(n − 3)/2 ways to choose
the second 2-cycle. However, the order of the two 2-cycles doesn’t matter, so
we divide the product by 2 to get n(n − 1)(n − 2)(n − 3)/8 possibilities.
1.3.18 Exercise 18
Find all numbers n such that S5 contains an element of order n.
Solution. For each n with 1 ≤ n ≤ 5, we can find an n-cycle in S5 . n = 6 is

also possible, for example, with (1 2)(3 4 5). But a combination of longer cycles
of different lengths is not possible since the underlying set {1, 2, 3, 4, 5} only has
five elements. Therefore the only possibilities for n are 1, 2, 3, 4, 5, and 6.
1.3.19 Exercise 19
Find all numbers n such that S7 contains an element of order n.
Solution. Clearly n = 1, 2, . . . , 7 are all valid. If σ ∈ S7 contains a 2-cycle, then

the only other cycles of different lengths that can be in the cycle decomposition
of σ is a 3-cycle, a 4-cycle, or a 5-cycle. The 2, 3 combination would have an order
of 6, the 2, 4 combination would have an order of 4, and the 2, 5 combination
would have an order of 10. We could also have a 3-cycle together with a 4-
cycle, resulting in a permutation with order 12. So the only possible orders are
n = 1, 2, 3, 4, 5, 6, 7, 10, and 12.
1.3.20 Exercise 20
Find a set of generators and relations for S3 .
Solution. The set S3 contains the six permutations 1, (1 2), (1 3), (2 3), (1 2 3),
and (1 3 2). By taking powers of each element we can see that S3 is not cyclic, so
we need at least two generators. Let α = (1 2) and β = (1 2 3). Then (1 3) = βα,
(2 3) = αβ, and (1 3 2) = β 2 . We have the relation α2 = β 3 = 1. But this is
not enough information to deduce that αβ has order 2, for example. So we may
include (αβ)2 = 1, which is enough to determine the orders of the remaining
elements. So
S3 = hα, β | α2 = β 3 = (αβ)2 = 1i.
1.4 Matrix Groups

Let F be a field and let n ∈ Z+ .
1.4.1 Exercise 1
Prove that |GL2 (F2 )| = 6.
Proof. Consider the matrix

a b
, a, b, c, d ∈ F2 .
c d
The determinant of this matrix is ad − bc. To be nonzero, either a and b are

nonzero, or b and c are nonzero, but not both. So the members of GL2 (F2 ) are

1 0 1 1 1 0 0 1 1 1 0 1
, , , , , and .
0 1 0 1 1 1 1 0 1 0 1 1
1.4.2 Exercise 2
Write out all the elements of GL2 (F2 ) and compute the order of each element.
Solution. Direct computation produces the following orders:

1 0 1 1 1 0 0 1 1 1 0 1
A
0 1 0 1 1 1 1 0 1 0 1 1 .
|A| 1 2 2 2 3 3
1.4.3 Exercise 3
Show that GL2 (F2 ) is non-abelian.
Proof. We have
1 1 1 1 1 0
=
1 0 0 1 1 1
but
1 1 1 1 0 1
= .
0 1 1 0 1 0
Therefore GL2 (F2 ) is non-abelian.
1.4.4 Exercise 4
Show that if n is not prime then Z/nZ is not a field.
Proof. Let n be a composite number, so that n = ab with a, b > 1. Then (a, n) =

a > 1, so by Proposition 4 of Section 0.3, a does not have a multiplicative inverse.
And a is nonzero, so Z/nZ is not a field.
1.4. MATRIX GROUPS 47
1.4.5 Exercise 5
Show that GLn (F ) is a finite group if and only if F has a finite number of
elements.
Proof. First, if F is finite, then GLn (F ) must be finite since there are only
finitely many n × n matrices with entries from F .
On the other hand, suppose F is not finite. Then for every α ∈ F with α 6= 0,
the matrix αI has nonzero determinant. Therefore GLn (F ) is infinite.
1.4.6 Exercise 6
2
If |F | = q is finite, prove that |GLn (F )| < q n .
2
Proof. Since F has q elements, there are only q n possible n × n matrices over
F that can be formed. Since at least one of these matrices has zero determinant
2
(take for example the zero matrix), it follows that |GLn (F )| < q n .
1.4.7 Exercise 7
Let p be a prime. Prove that the order of GL2 (Fp ) is p4 − p3 − p2 + p.
Proof. Let A be a 2 × 2 matrix over Fp that is not in GL2 (Fp ). Write

a b
A= ,
c d
and note that ad − bc = 0. We have two cases: a = 0 or a 6= 0.

First, if a = 0 then d can take any of p possible values, while bc = 0. Again
there are two cases: if b = 0 then there are p possible values that c can take. If
b 6= 0 (this can happen in p − 1 ways), then c = b−1 . So there are p possibilities
for d, multiplied by p + (p − 1) = 2p − 1 possibilities for b and c, which gives a
total of 2p2 − p choices for the case where a = 0.
Next, if a 6= 0, this can happen in p − 1 ways. Then d = bca−1 . Now b and
c can take any value and then d is determined by the other variables, so there
are p2 (p − 1) = p3 − p2 possibilities for this case.
Totaling the two cases, we find that there are
(p3 − p2 ) + (2p2 − p) = p3 + p2 − p
possible matrices that A can be. Since there are p4 total 2 × 2 matrices over Fp ,
it follows that
|GL2 (Fp )| = p4 − p3 − p2 + p.
1.4.8 Exercise 8
Show that GLn (F ) is non-abelian for any n ≥ 2 and any F .
Proof. Note that every field has an additive identity 0 and a distinct multiplica-
tive identity 1, so by restricting our proof to using only these two values from
F , the result will hold for any F .
We will use induction on n. The base case n = 2 was proved in Exercise 1.4.3
(the proof works for any F as noted above). Now assume that GLn−1 (F ) is
non-abelian for some n ≥ 3, and let A and B be non-commuting members of

GLn−1 (F ). Then, using block matrices, we get

A 0 B 0 AB 0 BA 0 B 0 A 0
= 6= = .
0 1 0 1 0 1 0 1 0 1 0 1
Therefore GLn (F ) is non-abelian, and this completes the proof.
1.4.9 Exercise 9
Prove that the binary operation of matrix multiplication of 2 × 2 matrices with
real number entries is associative.
Proof. Direct computation gives

a b e f i j ae + bg af + bh i j
=
c d g h k l ce + dg cf + dh k l

(ae + bg)i + (af + bh)k (ae + bg)j + (af + bh)l
= ,
(ce + dg)i + (cf + dh)k (ce + dg)j + (cf + dh)l
while

a b e f i j a b ei + f k ej + f l
=
c d g h k l c d gi + hk gj + hl

a(ei + f k) + b(gi + hk) a(ej + f l) + b(gj + hl)
= .
c(ei + f k) + d(gi + hk) c(ej + f l) + d(gj + hl)
Now, by comparing these two matrices using the associative and commutative
properties of the real numbers, the result will follow.
1.4.10 Exercise 10
Let

a
b
G= a, b, c ∈ R, a 6= 0, c 6= 0 .
c 0

a1 b1 a2 b2
(a) Compute the product of and to show that G is closed
0 c1 0 c2
under matrix multiplication.
Solution. We have

a1 b1 a2 b2 a1 a2 a1 b2 + b1 c2
= ∈ G,
0 c1 0 c2 0 c1 c2
so G is closed under multiplication.

a b
(b) Find the matrix inverse of and deduce that G is closed under
0 c
inverses.
Solution. Since ac 6= 0 the matrix is invertible and we get

−1 !
1 b
− ac

a b 1 c −b a
= = 1
∈ G.
0 c ac 0 a 0 c
So, G is closed under inverses.
(c) Deduce that G is a subgroup of GL2 (R).
Solution. This follows from Exercise 1.1.26.
(d) Prove that the set of elements of G whose two diagonal entries are equal
(i.e., a = c) is also a subgroup of GL2 (R).
Solution. Call this set H. We have

a1 b1 a2 b2 a1 a2 a1 b2 + b1 a2
= ∈H
0 a1 0 a2 0 a1 a2
and −1 !
1
− ab2

a b a
= 1
∈ H.
0 a 0 a
H is closed under matrix multiplication and inversion, so H is a subgroup

of GL2 (R).
1.4.11 Exercise 11
Let   
 1 a b 
H(F ) = 0 1 c  a, b, c ∈ F .
0 0 1
 
Let    
1 a b 1 d e
X = 0 1 c and Y = 0 1 f
0 0 1 0 0 1
be elements of H(F ).
(a) Compute the matrix XY and deduce that H(F ) is closed under matrix
multiplication. Exhibit explicit matrices such that XY 6= Y X (so that
H(F ) is always non-abelian).
Solution. We have
    
1 a b 1 d e 1 a+d af + b + e
XY = 0 1 c  0 1 f  = 0 1 c + f  ∈ H(F ),
0 0 1 0 0 1 0 0 1
so H(F ) is closed under multiplication. Moreover,

    
1 1 0 1 0 0 1 1 1
0 1 0 0 1 1 = 0 1 1
0 0 1 0 0 1 0 0 1
while     
1 0 0 1 1 0 1 1 0
0 1 1 0 1 0 = 0 1 1 ,
0 0 1 0 0 1 0 0 1
so H(F ) is always non-abelian.
(b) Find an explicit formula for the matrix inverse X −1 and deduce that H(F )
is closed under inverses.
Solution. Let  
1 −a ca − b
Z = 0 1 −c  .
0 0 1
By performing the multiplication, it is easily seen that XZ = ZX = I,
where I is the 3 × 3 identity matrix. It follows that Z = X −1 and since
Z ∈ H(F ), we see that H(F ) is closed under inverses.
(c) Prove the associative law for H(F ) and deduce that H(F ) is a group of
order |F |3 .
Solution. We have
    
1 a b 1 d e 1 g h
0 1 c  0 1 f  0 1 i 
0 0 1 0 0 1 0 0 1
  
1 a + d af + b + e 1 g h
= 0 1 c + f  0 1 i 
0 0 1 0 0 1
 
1 a + d + g af + ai + b + di + e + h
= 0 1 c+f +i 
0 0 1
and
    
1 a b 1 d e 1 g h
0 1 c  0 1 f  0 1 i 
0 0 1 0 0 1 0 0 1
  
1 a b 1 d + g di + e + h
= 0 1 c  0 1 f +i 
0 0 1 0 0 1
 
1 a + d + g af + ai + b + di + e + h
= 0 1 c+f +i ,
0 0 1
so multiplication in H(F ) is associative. This shows that H(F ) is a sub-

group of GL3 (F ).
Now, consider the matrix X above. If |F | = n < ∞ then each of a, b, c
can take any of n values each. So |H(F )| = n3 .
(d) Find the order of each element of the finite group H(Z/2Z).
Solution. Obviously |I| = 1. For the rest, we find

 2  2  2
1 0 0 1 1 0 1 0 1
0 1 1 = 0 1 0 = 0 1 0 = I,
0 0 1 0 0 1 0 0 1
so these have order 2,

 2  2
1 1 1 1 0 1
0 1 0 = 0 1 1 = I,
0 0 1 0 0 1
so these also have order 2, and

 4  4
1 1 0 1 1 1
0 1 1 = 0 1 1 = I,
0 0 1 0 0 1
so these have order 4. |H(Z/2Z)| = 23 = 8, so these are all the elements

in the group.
(e) Prove that every nonidentity element of the group H(R) has infinite order.
Solution. We will show by induction on n that

 n  
1 a b 1 na nb + n(n − 1)ac/2
0 1 c  = 0 1 nc . (1.2)
0 0 1 0 0 1
Since any nonidentity element has one of a, b, or c nonzero, this will be

enough to show that the element has infinite order.
The base case n = 1 is evident. Suppose (1.2) holds for some n ≥ 1. Then
 n+1   
1 a b 1 a b 1 na nb + n(n − 1)ac/2
0 1 c = 0 1 c  0 1 nc 
0 0 1 0 0 1 0 0 1
 
1 (n + 1)a (n + 1)b + n(n + 1)ac/2
= 0 1 (n + 1)c ,
0 0 1
so (1.2) holds for all positive integers n and the result follows.
1.5 The Quaternion Group

1.5.1 Exercise 1
Compute the order of each of the elements in Q8 .
Solution. 1 has order 1 and −1 has order 2. Since i2 = j 2 = k 2 = −1, we see

that i, j, k each have order 4. And since (−i)2 = (−j)2 = (−k)2 = −1, we know
that −i, −j, and −k have order 4 also.
1.5.2 Exercise 2
Write out the group tables for S3 , D8 and Q8 .
Solution. S3 :
1 (1 2) (1 3) (2 3) (1 2 3) (1 3 2)
1 1 (1 2) (1 3) (2 3) (1 2 3) (1 3 2)
(1 2) (1 2) 1 (1 3 2) (1 2 3) (2 3) (1 3)
(1 3) (1 3) (1 2 3) 1 (1 3 2) (1 2) (2 3)
(2 3) (2 3) (1 3 2) (1 2 3) 1 (1 3) (1 2)
(1 2 3) (1 2 3) (1 3) (2 3) (1 2) (1 3 2) 1
(1 3 2) (1 3 2) (2 3) (1 2) (1 3) 1 (1 2 3)
D8 :
1 r r2 r3 s sr sr2 sr3
1 1 r r2 r3 s sr sr2 sr3
r r r2 r3 1 sr3 s sr sr2
r2 r2 r3 1 r sr2 sr3 s sr
r3 r3 1 r r2 sr sr2 sr3 s
s s sr sr2 sr3 1 r r2 r3
sr sr sr2 sr3 s r3 1 r r2
sr2 sr2 sr3 s sr r2 r3 1 r
sr3 sr3 s sr sr2 r r2 r3 1
Q8 :
1 −1 i −i j −j k −k
1 1 −1 i −i j −j k −k
−1 −1 1 −i i −j j −k k
i i −i −1 1 k −k −j j
−i −i i 1 −1 −k k j −j
j j −j −k k −1 1 i −i
−j −j j k −k 1 −1 −i i
k k −k j −j −i i −1 1
−k −k k −j j i −i 1 −1
1.5. THE QUATERNION GROUP 53
1.5.3 Exercise 3
Find a set of generators and relations for Q8 .
Solution. One presentation is
Q8 = h−1, i, j, k | i2 = j 2 = k 2 = ijk = −1i,
where −1 commutes with the other elements of Q8 .

1.6 Homomorphisms and Isomorphisms

Let G and H be groups.
1.6.1 Exercise 1
Let ϕ : G → H be a homomorphism.
(a) Prove that ϕ(xn ) = ϕ(x)n for all n ∈ Z+ .
Proof. We use induction on n. The case for n = 1 is clear. Suppose

ϕ(xn ) = ϕ(x)n for some particular n ∈ Z+ . Then
ϕ(xn+1 ) = ϕ(xxn ) = ϕ(x)ϕ(xn ) = ϕ(x)ϕ(x)n = ϕ(x)n+1 ,
so the result holds for all n ∈ Z+ .
(b) Do part (a) for n = −1 and deduce that ϕ(xn ) = ϕ(x)n for all n ∈ Z.
Solution. Let 1G and 1H denote the identities of G and H, respectively.

Then
ϕ(1G )ϕ(1G ) = ϕ(1G ) = ϕ(1G )1H ,
and it follows from the cancellation law that ϕ(1G ) = 1H . Since the
identity is preserved, we will simply use 1 to denote the identity of both
groups from this point forward.
Now, for any x ∈ G,
ϕ(x)ϕ(x−1 ) = ϕ(xx−1 ) = ϕ(1) = 1,
which shows that ϕ(x−1 ) = ϕ(x)−1 . Then for n ∈ Z+ ,

−1
ϕ(x−n ) = ϕ((xn )−1 ) = ϕ(xn )−1 = ϕ(x)n = ϕ(x)−n .
Therefore ϕ(xn ) = ϕ(x)n holds for all n ∈ Z.
1.6.2 Exercise 2
If ϕ : G → H is an isomorphism, prove that |ϕ(x)| = |x| for all x ∈ G. Deduce
that any two isomorphic groups have the same number of elements of order n
for each n ∈ Z+ . Is the result true if ϕ is only assumed to be a homomorphism?
Proof. First suppose |x| = n < ∞. By the previous exercise, we have
ϕ(x)n = ϕ(xn ) = ϕ(1) = 1.
So |ϕ(x)| ≤ n. On the other hand, if |ϕ(x)| = k, then
ϕ(xk ) = ϕ(x)k = 1.
But 1 is the only element in G which gets sent to 1 in H, since ϕ is a bijection.

This shows that xk = 1, so that k ≥ n. Hence |ϕ(x)| = n.
1.6. HOMOMORPHISMS AND ISOMORPHISMS 55
Now suppose x has infinite order. If |ϕ(x)| = n < ∞, then ϕ(xn ) = ϕ(x)n =
1, and since ϕ is a bijection we must have xn = 1, a contradiction. Therefore
ϕ(x) must also have infinite order.
From the above we know that |x| = |ϕ(x)| for each x, and since ϕ is a
bijection this shows that G and H have the same number of elements of each
order.
Finally, this result does not necessarily hold for homomorphisms. For ex-
ample, let H be the trivial group {1} and take the function θ : G → H defined
by θ(x) = 1 for all x ∈ G. Then θ(x)θ(y) = θ(xy), so this is a homomorphism,
but every element in H has order 1, which is not true of G (unless G is also
trivial).
1.6.3 Exercise 3
If ϕ : G → H is an isomorphism, prove that G is abelian if and only if H is
abelian. If ϕ : G → H is a homomorphism, what additional conditions on ϕ (if
any) are sufficient to ensure that if G is abelian, then so is H?
Solution. Since ϕ must be invertible (it is a bijection) and since ϕ−1 must be
an isomorphism from H to G, the proof only needs to work in one direction. So
let x, y ∈ H be arbitrary, and let a = ϕ−1 (x) and b = ϕ−1 (y). If G is abelian,
then
xy = ϕ(a)ϕ(b) = ϕ(ab) = ϕ(ba) = ϕ(b)ϕ(a) = yx,
so H is also abelian, and the proof is complete.
Note that the same result does not hold for homomorphisms. For instance,
let ϕ : Z/2Z → D6 be given by ϕ(0̄) = 1 and ϕ(1̄) = s. Then ϕ is a homomor-
phism and Z/2Z is abelian, but D6 is not abelian.
However, if we add the constraint that ϕ is surjective, then the result does
hold: Suppose G is abelian, let x, y ∈ H be arbitrary, and pick a ∈ ϕ−1 (x) and
b ∈ ϕ−1 (y) (that is, a and b are chosen from the fibers of ϕ over x and y). Then,
as before,
xy = ϕ(a)ϕ(b) = ϕ(ab) = ϕ(ba) = ϕ(b)ϕ(a) = yx,
so H is abelian.
1.6.4 Exercise 4
Prove that the multiplicative groups R − {0} and C − {0} are not isomorphic.
Proof. Every element in R − {0} has infinite order, aside from 1 and −1 which
have orders 1 and 2, respectively. However, C − {0} has elements of order 4,
namely i and −i. Therefore these groups are not isomorphic.
1.6.5 Exercise 5
Prove that the additive groups R and Q are not isomorphic.
Proof. There is no bijection between R and Q, since the former is uncountable

and the latter is countable. Therefore the groups (R, +) and (Q, +) are not
isomorphic.
1.6.6 Exercise 6
Prove that the additive groups Z and Q are not isomorphic.
Proof. Suppose the contrary, and let ϕ : Q → Z be an isomorphism. Let
a = ϕ(1).
Then
1 1 1
a=ϕ + = 2ϕ .
2 2 2
Therefore 2 divides a. For the same reason, we also have

1
a = 3ϕ .
3
So 3 divides a. Using the same argument we see that the integer a is actually
divisible by every positive integer. The only way this is possible is if a = 0. But
then, for any n ∈ Z, we would have ϕ(n) = nϕ(1) = na = 0. So ϕ is clearly not
an injection, and this gives the necessary contradiction. Therefore the additive
groups Z and Q are not isomorphic.
1.6.7 Exercise 7
Prove that D8 and Q8 are not isomorphic.
Proof. We may simply look at the orders of the elements in each group. For
example, D8 has 4 elements with order 2 (namely, s, sr, sr2 , and sr3 ), while
Q8 only has one element with order 2 (namely −1). Therefore D8 ∼ 6 Q8 .
=
1.6.8 Exercise 8
Prove that if n 6= m, Sn and Sm are not isomorphic.
Proof. Since Sn has order n! and Sm has order m!, there is no bijection from Sn
to Sm unless n = m. Therefore Sn and Sm are not isomorphic when n 6= m.
1.6.9 Exercise 9
Prove that D24 and S4 are not isomorphic.
Proof. D24 has elements of order 12, namely r, r5 , r7 , and r11 . However, S4
has no elements of order 12, since every permutation in S4 is either a 2-cycle or
product of 2-cycles (which have order 2), a 3-cycle (which has order 3), or a 4-
cycle (which has order 4). Since isomorphisms must preserve orders of elements,
D24 and S4 cannot be isomorphic.
1.6.10 Exercise 10
Fill in the details of the proof that the symmetric group S∆ and SΩ are isomor-
phic if |∆| = |Ω| as follows: let θ : ∆ → Ω be a bijection. Define
ϕ : S∆ → SΩ by ϕ(σ) = θ ◦ σ ◦ θ−1 for all σ ∈ S∆
and prove the following:
(a) ϕ is well defined, that is, if σ is a permutation of ∆ then θ ◦ σ ◦ θ−1 is a

permutation of Ω.
Proof. For any permutation σ of ∆, it is clear that ϕ(σ) = θ ◦ σ ◦ θ−1 is

a function from Ω to itself. We want to show that it is a bijection.
Suppose a, b ∈ Ω are such that ϕ(σ)(a) = ϕ(σ)(b). Since θ is an injection,
this implies that (σ ◦ θ−1 )(a) = (σ ◦ θ−1 )(b). But σ is also an injection,
so θ−1 (a) = θ−1 (b) and, similarly, we have a = b. This shows that ϕ(σ) is
an injection.
Now let y ∈ Ω be arbitrary. Then we may take
x = ϕ(σ −1 )(y) = (θ ◦ σ −1 ◦ θ−1 )(y)
so that ϕ(σ)(x) = y. This shows that ϕ(σ) is a surjection. Hence ϕ(σ) is

a bijection from Ω to itself, that is, ϕ(σ) is a permutation of Ω.
(b) ϕ is a bijection from S∆ to SΩ .
Proof. Define ψ : SΩ → S∆ by
ψ(τ ) = θ−1 ◦ τ ◦ θ for any τ ∈ SΩ .
By the same argument as in part (a), ψ is well-defined. Moreover, for any

σ ∈ S∆ ,
(ψ ◦ ϕ)(σ) = ψ(θ ◦ σ ◦ θ−1 ) = θ−1 ◦ θ ◦ σ ◦ θ−1 ◦ θ = σ,
and for any τ ∈ SΩ ,
(ϕ ◦ ψ)(τ ) = ϕ(θ−1 ◦ τ ◦ θ) = θ ◦ θ−1 ◦ τ ◦ θ ◦ θ−1 = τ.
Therefore ψ is a two-sided inverse of ϕ, so that ϕ is a bijection.
(c) ϕ is a homomorphism, that is, ϕ(σ ◦ τ ) = ϕ(σ) ◦ ϕ(τ ).
Proof. Let σ, τ ∈ S∆ . Then
ϕ(σ) ◦ ϕ(τ ) = θ ◦ σ ◦ θ−1 ◦ θ ◦ τ ◦ θ−1 = θ ◦ σ ◦ τ ◦ θ−1 = ϕ(σ ◦ τ ).
Therefore ϕ is a homomorphism, and hence an isomorphism.

1.6.11 Exercise 11
Let A and B be groups. Prove that A × B ∼
= B × A.
Proof. Define the function ϕ : A × B → B × A by
ϕ(a, b) = (b, a) for any (a, b) ∈ A × B.
This is a homomorphism, since
ϕ((a, b)(c, d)) = ϕ(ac, bd) = (bd, ac) = (b, a)(d, c) = ϕ(a, b)ϕ(c, d).
It is also a surjection, since for any (b, a) ∈ B × A we can take (a, b) ∈ A × B

so that ϕ(a, b) = (b, a). Finally, if (a, b), (c, d) ∈ A × B are such that ϕ(a, b) =
ϕ(c, d) then (b, a) = (d, c). Then b = d and a = c, so (a, b) = (c, d) and ϕ is an
injection. This shows that ϕ is a bijection and hence an isomorphism.
1.6.12 Exercise 12
Let A, B, and C be groups and let G = A × B and H = B × C. Prove that
G×C ∼ = A × H.
Proof. Define ϕ : G × C → A × H as follows. For any ((a, b), c) ∈ G × C by
ϕ((a, b), c) = (a, (b, c)).
It is very straightforward to verify that ϕ is a bijection and a homomorphism,

and hence G × C ∼ = A × H.
1.6.13 Exercise 13
Let G and H be groups and let ϕ : G → H be a homomorphism. Prove that
the image of ϕ, ϕ(G), is a subgroup of H. Prove that if ϕ is injective then
G∼= ϕ(G).
Proof. We know that ϕ(G) is nonempty, since in particular ϕ(1) is mapped to

some element in H (we know from earlier exercises that the identity is preserved
so ϕ(1) = 1, but we do not strictly need that information here). Let a, b ∈ ϕ(G)
be arbitrary. Then there exist α, β ∈ G such that ϕ(α) = a, and ϕ(β) = b.
Then
ab = ϕ(α)ϕ(β) = ϕ(αβ),
so ab ∈ ϕ(G) and ϕ(G) is closed under the binary operation of H. Moreover,
by Exercise 1.6.1,
a−1 = ϕ(α)−1 = ϕ(α−1 ),
so ϕ(G) is closed under inverses. Hence ϕ(G) is a subgroup of H.
Now, if we define ϕ∗ : G → ϕ(G) by ϕ∗ (γ) = ϕ(γ) for each γ ∈ G, then ϕ∗
is surjective by definition. If, in addition, ϕ is injective, then ϕ∗ is a bijection
and G ∼ = ϕ(G).
1.6.14 Exercise 14
Let G and H be groups and let ϕ : G → H be a homomorphism. Define the
kernel of ϕ to be {g ∈ G | ϕ(g) = 1H } (so the kernel is the set of elements in G
which map to the identity of H, i.e., is the fiber over the identity of H). Prove
that the kernel of ϕ is a subgroup of G. Prove that ϕ is injective if and only if
the kernel of ϕ is the identity subgroup of G.
Proof. From Exercise 1.6.1 we know that ϕ(1G ) = 1H so the kernel of ϕ is

nonempty. Suppose a, b ∈ ker ϕ. Then
ϕ(ab) = ϕ(a)ϕ(b) = 1H 1H = 1H ,
and ab ∈ ker ϕ. Additionally, if a ∈ ker ϕ then
ϕ(a−1 ) = ϕ(a)−1 = 1−1

H = 1H
and a−1 ∈ ker ϕ. Therefore ker ϕ is a subgroup of G.
1.6.15 Exercise 15
Define a map π : R2 → R by π((x, y)) = x. Prove that π is a homomorphism
and find the kernel of π.
Proof. For any (x1 , y1 ), (x2 , y2 ) ∈ R2 , we have
π((x1 , y1 ) + (x2 , y2 )) = π(x1 + x2 , y1 + y2 ) = x1 + x2 = π(x1 , y1 ) + π(x2 , y2 ),
so π is a homomorphism. Also,
ker π = {(0, y) ∈ R2 | y ∈ R}.
1.6.16 Exercise 16
Let A and B be groups and let G be their direct product, A × B. Prove that the
maps π1 : G → A and π2 : G → B defined by π1 ((a, b)) = a and π2 ((a, b)) = b
are homomorphisms and find their kernels.
Proof. For any (a, b) and (c, d) ∈ A × B, we have
π1 ((a, b)(c, d)) = π1 (ac, bd) = ac = π1 (a, b)π1 (c, d)
and
π2 ((a, b)(c, d)) = π2 (ac, bd) = bd = π2 (a, b)π2 (c, d),
so π1 and π2 are homomorphisms. Their kernels are
ker π1 = {(1, b) ∈ A × B | b ∈ B}
and
ker π2 = {(a, 1) ∈ A × B | a ∈ A}.
1.6.17 Exercise 17
Let G be any group. Prove that the map from G to itself defined by g 7→ g −1
is a homomorphism if and only if G is abelian.
Proof. Suppose G is abelian. Then for any a, b ∈ G,
(ab)−1 = b−1 a−1 = a−1 b−1 ,
so g 7→ g −1 is a homomorphism. Conversely, suppose g 7→ g −1 is a homomor-
phism and let a, b ∈ G be arbitrary. Then b−1 a−1 = (ba)−1 and we have
ab = (a−1 )−1 (b−1 )−1 = (b−1 a−1 )−1 = [(ba)−1 ]−1 = ba,
so G is abelian.
1.6.18 Exercise 18
Let G be any group. Prove that the map from G to itself defined by g 7→ g 2 is
a homomorphism if and only if G is abelian.
Proof. Suppose G is abelian. Then for any a, b ∈ G,
(ab)2 = abab = a2 b2 ,
and g 7→ g 2 is a homomorphism. Now suppose g 7→ g 2 is a homomorphism.
Then for any a, b ∈ G,
a2 b2 = (ab)2 = abab,
and multiplying both sides of the equation a2 b2 = abab on the left by a and on
the right by b gives ab = ba, so that G is abelian.
1.6.20 Exercise 20
Let G be a group and let Aut(G) be the set of all isomorphisms from G onto
G. Prove that Aut(G) is a group under function composition (called the auto-
morphism group of G and the elements of Aut(G) are called automorphisms of
G).
Proof. Let ϕ, ψ ∈ Aut(G). Then ϕ ◦ ψ is a bijection from G to itself. It is also
a homomorphism, since for any a, b ∈ G,
(ϕ ◦ ψ)(ab) = ϕ(ψ(a)ψ(b)) = (ϕ ◦ ψ)(a)(ϕ ◦ ψ)(b).
This shows that ϕ ◦ ψ ∈ Aut(G) so Aut(G) is closed under composition. And
function composition is always associative.
Clearly the identity map 1 : G → G is an isomorphism, so Aut(G) has an
identity. And for any ϕ ∈ Aut(G), ϕ−1 must exist since ϕ is a bijection. Now,
for any a, b ∈ G let a∗ = ϕ−1 (a) and b∗ = ϕ−1 (b). Since ϕ is a homomorphism,
we have
ϕ(a∗ b∗ ) = ϕ(a∗ )ϕ(b∗ ) = ab,
which implies that a∗ b∗ = ϕ−1 (ab). Then
ϕ−1 (a)ϕ−1 (b) = a∗ b∗ = ϕ−1 (ab),
and we see that ϕ−1 is an isomorphism and hence ϕ−1 ∈ Aut(G). So elements in
Aut(G) have inverses. Therefore Aut(G) is a group under function composition.
1.6.21 Exercise 21
Prove that for each fixed nonzero k ∈ Q the map from Q to itself defined by
q 7→ kq is an automorphism of Q.
Proof. Fix a nonzero k ∈ Q and let ϕ : Q → Q be given by ϕ(r) = kr. Then for
any a, b ∈ Q,
ϕ(a + b) = k(a + b) = ka + kb = ϕ(a) + ϕ(b),
so ϕ is a homomorphism. To show that it is a bijection, note that it must

be surjective since for any a ∈ Q, we may take b = a/k so that ϕ(b) = a.
And ϕ must be injective since for any a, b ∈ Q, ϕ(a) = ϕ(b) implies ka = kb
which implies a = b since k is nonzero. Therefore ϕ is a bijection and hence an
automorphism of Q.
1.6.22 Exercise 22
Let A be an abelian group and fix some k ∈ Z. Prove that the map a 7→ ak is
a homomorphism from A to itself. If k = −1, prove that this homomorphism is
an isomorphism (i.e., is an automorphism of A).
Proof. Fix k ∈ Z and let ϕ : A → A be the mapping a 7→ ak . Then for any

a, b ∈ A, we have
ϕ(ab) = (ab)k = ak bk = ϕ(a)ϕ(b),
where the second equality follows from the fact that A is abelian. So ϕ is a
homomorphism.
In the case where k = −1, ϕ must be a bijection since it is its own inverse
function. Hence a 7→ a−1 is an automorphism of A.
1.6.23 Exercise 23
Let G be a finite group which possesses an automorphism σ such that σ(g) = g
if and only if g = 1. If σ 2 is the identity map from G → G, prove that G is
abelian (such an automorphism σ is called fixed point free of order 2).
Proof. Consider the map ϕ : G → G given by ϕ(x) = x−1 σ(x). For any x, y ∈ G,
if ϕ(x) = ϕ(y) then
x−1 σ(x) = y −1 σ(y)
or, rearranging,
σ(y) = yx−1 σ(x).
This gives
y = σ(σ(y)) = σ(yx−1 σ(x)) = σ(yx−1 )x
and multiplying on the right by x−1 gives
yx−1 = σ(yx−1 ). (1.3)
Since σ is fixed point free, (1.3) then implies that yx−1 = 1 or x = y. Therefore
ϕ is an injection, and hence a bijection since it maps the finite set G to itself.
Therefore every x ∈ G can be written in the form x = y −1 σ(y) for some y ∈ G.
Now let x ∈ G be arbitrary. Then, for some y ∈ G,
σ(x) = σ(y −1 σ(y)) = σ(y)−1 y.
However, since (ab)−1 = b−1 a−1 for a, b in any group, we also have
−1
σ(y)−1 y = σ(y)−1 (y −1 )−1 = y −1 σ(y) = x−1 .
Hence σ(x) = x−1 for all x ∈ G.

Finally, let a, b ∈ G be arbitrary. Then
σ(ab) = (ab)−1 = b−1 a−1 = σ(b)σ(a) = σ(ba).
But σ is an injection, so ab = ba. This shows that G is abelian.
1.6.24 Exercise 24
Let G be a finite group and let x and y be distinct elements of order 2 in G that
generate G. Prove that G ∼ = D2n , where n = |xy|.
Proof. Let t = xy. By Exercise 1.2.6, we have tx = xt−1 . Note also that x and
t generate G, since y can be written as y = xt. So by repeated application of
the relation tx = xt−1 , we may express any member of G uniquely in the form
xi tj for some integers i, j with 0 ≤ i ≤ 1 and 0 ≤ j < n (the representation
is unique since t has order n, which implies that t, t2 , . . . , tn−1 are all distinct).
Therefore |G| = 2n.
Now let ϕ : D2n → G be given by
ϕ(si rj ) = xi tj , for i, j ∈ Z with 0 ≤ i ≤ 1 and 0 ≤ j ≤ n − 1.
Since every element in D2n can be written uniquely as si rj with the above
restrictions on i and j, the function ϕ is well defined. And since x and t satisfy
the same relations in G that s and r satisfy in D2n , ϕ must be a homomorphism.
We will now show that ϕ is a bijection. For any b ∈ G, write b = xi tj for
i ∈ {0, 1} and j ∈ {0, 1, . . . , n − 1}. Then if a = si rj , we have ϕ(a) = b, which
shows that ϕ is surjective. Since |G| = |D2n |, this is enough to show that ϕ is
a bijection.
The function ϕ is a bijective homomorphism, hence it is an isomorphism and
D2n ∼= G.
1.6.25 Exercise 25
Let n ∈ Z+ , let r and s be the usual generators of D2n and let θ = 2π/n.

cos θ − sin θ
(a) Prove that the matrix is the matrix of the linear transfor-
sin θ cos θ
mation which rotates the x, y plane about the origin in a counterclockwise
direction by θ radians.
Proof. For a vector (x, y) in R2 we have

cos θ − sin θ x x cos θ − y sin θ
= .
sin θ cos θ y x sin θ + y cos θ
The distance d of this point from the origin is

p
d = (x cos θ − y sin θ)2 + (x sin θ + y cos θ)2
q
= x2 cos2 θ + y 2 sin2 θ + x2 sin2 θ + y 2 cos2 θ
p
= x2 + y 2 ,
which is the same distance as (x, y) is from the origin. Moreover, the angle
α between these two vectors is given by
x(x cos θ − y sin θ) + y(x sin θ + y cos θ)

cos α =
x2 + y 2
x2 cos θ − xy sin θ + xy sin θ + y 2 cos θ
=
x2 + y 2
2π
= cos θ = cos .
n
So we see that α = 2π/n. This shows that the image of the point (x, y)
under this transformation is the same point rotated about the origin by
an angle of θ.
(b) Prove that the map ϕ : D2n → GL2 (R) defined on generators by

cos θ − sin θ 0 1
ϕ(r) = and ϕ(s) =
sin θ cos θ 1 0
extends to a homomorphism of D2n into GL2 (R).

0 1
Proof. Since is easily seen to be a reflection across the line y = x,
1 0
it is evident that ϕ(r) and ϕ(s) satisfy the same relations as do r and s.
Namely, if I is the 2 × 2 identity matrix, we have
ϕ(r)n = ϕ(s)2 = I and ϕ(r)ϕ(s) = ϕ(s)ϕ(r)−1 .
The latter relation comes from the fact that reflecting across the line y = x
and then rotating by θ is the same as first rotating by 2π − θ and then
reflecting across the line.
Since ϕ(r) and ϕ(s) satisfy the same relations as the corresponding gen-
erators of D2n , we see that ϕ extends to a homomorphism.
(c) Prove that the homomorphism ϕ in part (b) is injective.
Proof. Let H denote the subgroup of GL2 (R) generated by ϕ(r) and ϕ(s).
Then the function ψ : D2n → H defined by restricting the codomain of ϕ
is surjective. But it is not difficult to see that |H| = 2n = |D2n |, so the
map ψ and hence ϕ must also be injective.
1.6.26 Exercise 26
Let i and j be the generators of Q8 described in Section 5. Prove that the map
ϕ from Q8 to GL2 (C) defined on generators by
√
−1 0
√ 0 −1
ϕ(i) = and ϕ(j) =
0 − −1 1 0
extends to a homomorphism. Prove that ϕ is injective.

Proof. First, we have
√ 2
−1 √0 −1 0
ϕ(i)2 = = = −I,
0 − −1 0 −1
2
0 −1 −1 0
ϕ(j)2 = = = −I,
1 0 0 −1
so we may take ϕ(−1) = −I. And −I commutes with all members of GL2 (C).
Also,
√ √
−1 √0 0 −1 √0 − −1
ϕ(i)ϕ(j) = = .
0 − −1 1 0 − −1 0
So we may let √
√0 − −1
ϕ(k) = .
− −1 0
Note that ϕ(k)2 = −I as expected.
To summarize, we have
ϕ(i)2 = ϕ(j)2 = ϕ(k)2 = ϕ(i)ϕ(j)ϕ(k) = ϕ(−1),
so ϕ(i), ϕ(j), ϕ(k), and ϕ(−1) satisfy all the same relations as given in our
presentation for Q8 in Exercise 1.5.3. Therefore ϕ extends to a homomorphism.
Lastly, consider the subgroup of GL2 (C) generated by ϕ(i), ϕ(j), ϕ(k), and
ϕ(−1). It is not difficult to see that this subgroup contains exactly eight elements
(those named plus their inverses and the identity). So the function obtained
from ϕ by restricting its codomain to this subgroup must be surjective. Since
its domain and codomain share the same cardinality, it must also be injective.
Hence ϕ is injective.
1.7. GROUP ACTIONS 65
1.7 Group Actions

1.7.1 Exercise 1
Let F be a field. Show that the multiplicative group of nonzero elements of F
(denoted by F × ) acts on the set F by g · a = ga, where g ∈ F × , a ∈ F and ga
is the usual product in F of the two field elements.
Proof. Let g1 , g2 ∈ F × . Then for any a ∈ F ,
g1 · (g2 · a) = g1 · g2 a = g1 (g2 a) = (g1 g2 )a = (g1 g2 ) · a,
where the second-to-last equality follows from the associativity of multiplication

in F . Also, for any a ∈ F × ,
1 · a = 1a = a,
since 1 is the identity of the group F × . And 1(0) = 0 (which follows from
distributivity), so we can say that 1 · a = a for all a ∈ F . Therefore the mapping
(g, a) 7→ ga of F × × F → F is a group action.
1.7.2 Exercise 2
Show that the additive group Z acts on itself by z · a = z + a for all z, a ∈ Z.
Proof. For all z1 , z2 , a ∈ Z, we have
z1 · (z2 · a) = z1 + (z2 + a) = (z1 + z2 ) + a = (z1 + z2 ) · a
and
0 · a = 0 + a = a.
Therefore Z acts on itself as stated.
1.7.3 Exercise 3
Show that the additive group R acts on the x, y plane R × R by r · (x, y) =
(x + ry, y).
Proof. For any r1 , r2 ∈ R and any (x, y) ∈ R2 , we have
r1 · (r2 · (x, y)) = r1 · (x + r2 y, y)

= (x + r2 y + r1 y, y)
= (x + (r1 + r2 )y, y)
= (r1 + r2 ) · (x, y)
and
0 · (x, y) = (x + 0y, y) = (x, y).
Therefore R acts on R2 in the manner stated above.

1.7.4 Exercise 4
Let G be a group acting on a set A and fix some a ∈ A. Show that the following
sets are subgroups of G:
(a) the kernel of the action
Proof. Suppose g, h are in the kernel of the action. Then for any b ∈ A,
(gh) · b = g · (h · b) = g · b = b,
so gh is in the kernel, and the kernel is closed under the group operation.
Moreover, if g is in the kernel then
b = 1 · b = (g −1 g) · b = g −1 · (g · b) = g −1 · b,
so g −1 is in the kernel.
Therefore the kernel of the group action is a nonempty subset of G which is
closed under the binary operation of G and which is closed under inverses,
so the kernel is a subgroup of G.
(b) {g ∈ G | ga = a} (called the stabilizer of a in G)
Proof. The stabilizer of a is nonempty since 1 is a member. Now let g, h

be any members of the stabilizer. Then
(gh) · a = g · (h · a) = g · a = a,
so the stabilizer is closed under the group operation. It is also closed under
inverses, since
a = 1 · a = (g −1 g) · a = g −1 · (g · a) = g −1 · a.
Therefore the stabilizer is a subgroup of G.
1.7.5 Exercise 5
Prove that the kernel of an action of the group G on the set A is the same as
the kernel of the corresponding permutation representation G → SA .
Proof. Let ϕ : G → SA be the permutation representation of the group action

on A, so that for g ∈ G and a ∈ A, ϕ(g)(a) = g · a.
If g ∈ ker ϕ, then ϕ(g) = 1, where 1 is the identity permutation on A. Then
g.a = a for all a ∈ A, and g is in the kernel of the action. Conversely, if g
is in the kernel of the action, then g.a = a for all a ∈ A, so that ϕ(g) = 1
and g ∈ ker ϕ. Therefore the kernel of the group action and the kernel of the
corresponding permutation representation are the same.
1.7.6 Exercise 6
Prove that a group G acts faithfully on a set A if and only if the kernel of the
action is the set consisting only of the identity.
Proof. First, suppose that G acts faithfully on A and let g be an element in the
kernel of the action. Then g · a = a for all a ∈ A. However, 1 · a = a for all
a ∈ A, so the elements 1 and g induce the same permutation on A. Since G acts
faithfully, this must mean that g = 1, so that the kernel of the action is the set
{1}.
For the converse, suppose that the kernel of the action is the set {1}. Pick two
elements g and h in G and suppose that g and h induce the same permutation
on A. Then for any a ∈ A, g · a = h · a. But then
a = (g −1 g) · a = g −1 · (g · a) = g −1 · (h · a) = (g −1 h) · a.
Therefore g −1 h is in the kernel of the action, so g −1 h = 1. This implies that

g = h, so that distinct elements in G must induce distinct permutations on A.
This shows that G acts faithfully on A.
1.7.7 Exercise 7
Prove that in Example 2 in this section the action is faithful.
Proof. If V is a vector space over a field F , then the multiplicative group F ×
acts on the set V via the mapping a · v = av for a ∈ F × and v ∈ V . We want
to show that this action is faithful.
Let a, b ∈ F × be such that a · v = b · v for all v ∈ V . Then
0 = a · v + −(a · v)
= a · v + −(b · v)
= av − bv
= (a − b)v.
Since (a − b)v is 0 even when v is nonzero, this implies that a − b = 0 or a = b.

Therefore distinct elements in F × must induce distinct permutations on V and
the action is faithful.
1.7.8 Exercise 8
Let A be a nonempty set and let k be a positive integer with k ≤ |A|. The
symmetric group SA acts on the set B consisting of all subsets of A of cardinality
k by σ · {a1 , . . . , ak } = {σ(a1 ), . . . , σ(ak )}.
(a) Prove that this is a group action.
Proof. Suppose σ1 , σ2 ∈ SA . Then for any subset {a1 , . . . , ak } of A,
σ1 · (σ2 · {a1 , . . . , ak }) = σ1 · {σ2 (a1 ), . . . , σ2 (ak )}

= {σ1 (σ2 (a1 )), . . . , σ1 (σ2 (ak ))}
= (σ1 ◦ σ2 ) · {a1 , . . . , ak }
and
1 · {a1 , . . . , ak } = {1(a1 ), . . . , 1(ak )} = {a1 , . . . , ak }.
Therefore the specified mapping is a group action.
(b) Describe explicitly how the elements (1 2) and (1 2 3) act on the six 2-
element subsets of {1, 2, 3, 4}.
Solution. We have
(1 2) · {1, 2} = {2, 1},

(1 2) · {1, 3} = {2, 3},
(1 2) · {1, 4} = {2, 4},
(1 2) · {2, 3} = {1, 3},
(1 2) · {2, 4} = {1, 4},
(1 2) · {3, 4} = {3, 4},
and
(1 2 3) · {1, 2} = {2, 3},

(1 2 3) · {1, 3} = {2, 1},
(1 2 3) · {1, 4} = {2, 4},
(1 2 3) · {2, 3} = {3, 1},
(1 2 3) · {2, 4} = {3, 4},
(1 2 3) · {3, 4} = {1, 4}.
1.7.9 Exercise 9
Do both parts of the preceding exercise with “ordered k-tuples” in place of “k-
element subsets,” where the action on k-tuples is defined as above but with set
braces replaced by parentheses.
Solution. The work is essentially the same, but with k-tuples replacing the
k-element subsets, so we omit it. Note that in part (b) there are twice as
many different 2-tuples as there are 2-element subsets, since the ordering of the
elements is significant.
1.7.10 Exercise 10
With reference to the preceding two exercises determine:
(a) for which values of k the action of Sn on k-element subsets is faithful
Solution. The action of SA on k-element subsets of a set A is faithful

for all integers k with 1 ≤ k < |A|, which we will now show. Suppose
σ1 and σ2 are distinct permutations in SA . Label the elements of A
as {a1 , a2 , . . . , an }, where n = |A|. Without loss of generality, we may
suppose that σ1 (a1 ) 6= σ2 (a1 ) (if not, relabel the elements of A so that
this is true).
Now, take any k-element subset B of A which contains a1 but which does
not contain (σ1−1 ◦ σ2 )(a1 ) (this is possible since 1 ≤ k < |A|). Then
σ1 · B does not contain σ2 (a1 ), however σ2 · B does. Therefore distinct
permutations in SA induce distinct permutations on the k-element subsets
of A, so the action is faithful (again, assuming 1 ≤ k < |A|).
(b) for which values of k the action of Sn on ordered k-tuples is faithful
Solution. The action of SA on ordered k-tuples of elements of A is faithful

for all integers k with 1 ≤ k ≤ |A|. To see this, suppose that σ1 , σ2 are
distinct permutations in SA . Suppose for example that σ1 (a1 ) 6= σ2 (a1 )
and consider the k-tuple B = (a1 , a2 , . . . , ak ). Then the first coordinate
in σ1 · B is distinct from the first coordinate of σ2 · B. Therefore distinct
permutations in SA induce distinct permutations on the set of k-tuples,
so the action is faithful.
1.7.11 Exercise 11
Write out the cycle decomposition of the eight permutations in S4 corresponding
to the elements in D8 given by the action of D8 on the vertices of a square (where
the vertices of the square are labelled as in Section 2).
Solution. Let ϕ : D8 → S4 be the permutation representation associated to the

action of D8 on the vertices {1, 2, 3, 4} of a square. Then
ϕ(1) = 1,
ϕ(r) = (1 2 3 4),
ϕ(r2 ) = (1 3)(2 4),
ϕ(r3 ) = (1 4 3 2),
ϕ(s) = (2 4),
ϕ(sr) = (1 4)(2 3),
ϕ(sr2 ) = (1 3),
and
ϕ(sr3 ) = (1 2)(3 4).
1.7.12 Exercise 12
Assume n is an even positive integer and show that D2n acts on the set consisting
of pairs of opposite vertices of a regular n-gon. Find the kernel of this action
(label vertices as usual).
Solution. Fix an even positive integer n. The set of pairs of opposite vertices
of a regular n-gon, labeled in the usual way, is the set P = {P1 , P2 , . . . , Pn/2 }
where n n o
Pk = k, + k .
2
D2n acts on P by x · Pk = P` where P` is the set of images of the vertices in

Pk under the symmetry x. For example, in D8 , sr · {2, 6} = {7, 3} because sr
maps vertex 2 to vertex 7 and vertex 6 to vertex 3.
Let x, y ∈ D2n . It is clear by definition of the action that x·(y·Pk ) = (xy)·Pk
and that 1 · Pk = Pk . So this is a group action. The only symmetry in D2n
which fixes all vertices is the identity 1. However, since the order of vertices
in each pair does not matter, any symmetry which only sends vertices to their
opposites will also fix pairs of vertices. The only symmetry which does this is
rn/2 . There is no symmetry which fixes only some vertices and which sends
all others to their opposite vertices, so the kernel of the action is just the set
{1, rn/2 }.
1.7.13 Exercise 13
Find the kernel of the left regular action.
Solution. Let G be a group. The kernel of the left regular action is the set
{g ∈ G | gh = h for all h ∈ G}.
By uniqueness of the identity, it is clear that this set is simply {1}. Therefore
the left regular action is always faithful.
1.7.14 Exercise 14
Let G be a group and let A = G. Show that if G is non-abelian then the maps
defined by g · a = ag for all g, a ∈ G do not satisfy the axioms of a (left) group
action of G on itself.
Proof. Since G is non-abelian, there exist g1 , g2 ∈ G such that g1 g2 6= g2 g1 .

Then g1 · (g2 · a) = ag2 g1 but (g1 g2 ) · a = ag1 g2 . If ag2 g1 = ag1 g2 then the
cancellation law gives g2 g1 = g1 g2 , a contradiction. Therefore this map does
not define a left group action.
1.7.15 Exercise 15
Let G be any group and let A = G. Show that the maps defined by g · a = ag −1
for all g, a ∈ G do satisfy the axioms of a (left) group action of G on itself.
Proof. Let g1 , g2 , a ∈ G be arbitrary. Then
g1 · (g2 · a) = g1 · (ag2−1 ) = ag2−1 g1−1 = a(g1 g2 )−1 = (g1 g2 ) · a
and
1 · a = a1−1 = a1 = a.
Therefore this map does define a group action.

1.7.16 Exercise 16
Let G be any group and let A = G. Show that the maps defined by g ·a = gag −1
for all g, a ∈ G do satisfy the axioms of a (left) group action (this action of G
on itself is called conjugation).
Proof. For any g1 , g2 , a ∈ G we have
g1 · (g2 · a) = g1 · (g2 ag2−1 ) = g1 g2 ag2−1 g1−1 = (g1 g2 )a(g1 g2 )−1 = (g1 g2 ) · a
and
1 · a = 1a1 = a,
so this mapping does define a group action.
1.7.17 Exercise 17
Let G be a group and let G act on itself by left conjugation, so each g ∈ G maps
G to G by
x 7→ gxg −1 .
For fixed g ∈ G, prove that conjugation by g is an isomorphism from G onto
itself (i.e., is an automorphism of G). Deduce that x and gxg −1 have the same
order for all x in G and that for any subset A of G, |A| = |gAg −1 | (here
gAg −1 = {gag −1 | a ∈ A}).
Proof. Fix a g ∈ G and let ϕ : G → G denote the map x 7→ gxg −1 . Then for
any x1 , x2 ∈ G we have
ϕ(x1 x2 ) = gx1 x2 g −1 = (gx1 g −1 )(gx2 g −1 ) = ϕ(x1 )ϕ(x2 )
so ϕ is a homomorphism.
Next, suppose ϕ(x1 ) = ϕ(x2 ). Then gx1 g −1 = gx2 g −1 and multiplying both
sides of this equation on the left by g −1 and on the right by g gives x1 = x2 , so
that ϕ is injective.
Now let y ∈ G be arbitrary. Then x = g −1 yg is such that ϕ(x) = y, so ϕ is
surjective. Therefore ϕ is an automorphism.
Since isomorphisms preserve order, we see that each element x in G has the
same order as its conjugate gxg −1 . Moreover, if A ⊆ G then the restriction of
ϕ to A, ϕ|A : A → gAg −1 , is still a bijection, so |A| = |gAg −1 |.
1.7.18 Exercise 18
Let H be a group acting on a set A. Prove that the relation ∼ on A defined by
a ∼ b if and only if a = hb for some h ∈ H
is an equivalence relation. (For each x ∈ A the equivalence classes of x under ∼

is called the orbit of x under the action of H. The orbits under the action of H
partition the set A.)
Proof. Since a = 1a we have a ∼ a. And if a = hb for h ∈ H then
h−1 a = h−1 (hb) = (h−1 h)b = b,
so a ∼ b implies b ∼ a.
Lastly, suppose a ∼ b and b ∼ c and let h1 , h2 ∈ H be such that a = h1 b and
b = h2 c. Then a = h1 (h2 c) = (h1 h2 )c and a ∼ c. Hence ∼ is an equivalence
relation.
1.7.19 Exercise 19
Let H be a subgroup of the finite group G and let H act on G (here A = G) by
left multiplication. Let x ∈ G and let O be the orbit of x under the action of
H. Prove that the map
H→O defined by h 7→ hx
is a bijection (hence all orbits have cardinality |H|). From this and the preceding
exercises deduce Lagrange’s Theorem:
if G is a finite group and H is a subgroup of G then |H| divides |G|.
Proof. Let ϕ : H → O denote the map h 7→ hx. Suppose ϕ(h) = ϕ(k) for
h, k ∈ H. Then hx = kx and right cancellation implies that h = k, so that ϕ is
injective. And ϕ is surjective by definition (y ∈ O means that there is h ∈ H
such that hx = y). Therefore ϕ is a bijection and |H| = |O|.
From the previous exercise, we know that the orbits under the action of H
partition G. Each equivalence class O has cardinality |H|, so |G| = n|H| where
n is the number of orbits. Hence |H| divides |G|.
1.7.20 Exercise 20
Show that the group of rigid motions of a tetrahedron is isomorphic to a sub-
group of S4 .
Proof. Call the group of rigid motions of the tetrahedron G. Number each
vertex and let A denote the set {1, 2, 3, 4}. Then each rigid motion α ∈ G
induces a permutation σα ∈ S4 of A. G acts on A via the map αi = σα (i).
Since each distinct α ∈ G permutes the vertices in a different way, we get
an injective homomorphism
ϕ : G → S4 given by ϕ(α) = σα .
Then ϕ(G) is a subgroup of S4 , and by simply restricting the codomain of ϕ we

have an isomorphism from G to this subgroup of S4 .
1.7.21 Exercise 21
Show that the group of rigid motions of a cube is isomorphic to S4 .
Proof. Again let G denote the group of rigid motions and let A = {1, 2, 3, 4},
where each i ∈ A corresponds to a pair of opposing vertices on a cube. Each
α ∈ G sends each pair of opposing vertices to a new pair of opposing vertices.
Therefore G acts on A.
Consider the homomorphism ϕ : G → S4 given by
ϕ(α)(i) = αi.
Then ϕ is injective since each distinct rigid motion α ∈ G gives rise to a different
permutation of A. From Exercise 1.2.10 we know that |G| = 24 = |S4 |, so ϕ is
in fact an isomorphism.
1.7.22 Exercise 22
Show that the group of rigid motions of an octahedron is isomorphic to S4 .
Deduce that the groups of rigid motions of a cube and an octahedron are iso-
morphic.
Proof. Number each pair of opposing faces of the octahedron 1, 2, 3, 4. Let G
be the group of rigid motions of the octahedron and let A = {1, 2, 3, 4}. Each
α ∈ G sends each pair of opposing faces to a new pair of opposing faces, so G
acts on A.
As in the previous exercise, we see that the homomorphism
ϕ : G → S4 given by ϕ(α)(i) = αi.
is injective. By Exercise 1.2.11 we have |G| = 24 = |S4 |, so ϕ is an isomorphism.

From this and the previous exercise, we see that the groups of rigid motions
of the cube and the octahedron are isomorphic.
1.7.23 Exercise 23
Explain why the action of the group of rigid motions of a cube on the set of
three pairs of opposite faces is not faithful. Find the kernel of this action.
Solution. The group of rigid motions of a cube has order 24 but the permutations
on the set of pairs of opposite faces has order |S3 | = 6. Therefore the action
cannot be faithful.
Construct a line through the center of each pair of opposite faces. Then
a 180◦ rotation about each of these lines will send each pair of opposite faces
to itself. These are the only rotations that fix pairs of opposing faces, so the
kernel of the action consists of these three 180◦ rotations along with the identity
transformation.
Chapter 2
Subgroups
2.1 Definition and Examples

Let G be a group.
2.1.1 Exercise 1
In each of (a)–(e) prove that the specified subset is a subgroup of the given
group.
(a) the set of complex numbers of the form a + ai, a ∈ R (under addition)
Proof. Call the set H. H is obviously nonempty. For any a + ai and b + bi

in H, we have
(a + ai) − (b + bi) = (a − b) + (a − b)i ∈ H,
so by Proposition 1, H ≤ C.
(b) the set of complex numbers of absolute value 1, i.e., the unit circle in the
complex plane (under multiplication)
Proof. Let H denote the complex numbers of absolute value 1, and let z
denote the conjugate of z. Then H is nonempty and for any z, w ∈ H, we
have
|w|
|zw−1 | = |z||w−1 | = |z| 2 = 1,
|w|
so zw−1 ∈ H. Therefore H ≤ C× .
(c) for fixed n ∈ Z+ the set of rational numbers whose denominators divide
n (under addition)
Proof. Let H denote the subset in question. H is clearly not empty. Let
a/b ∈ H and c/d ∈ H be in lowest terms, where n = bx = dy for some
x, y ∈ Z+ . Then
a c ax cy ax − cy
− = − = .
b d n n n
75
76 CHAPTER 2. SUBGROUPS
After writing this fraction in lowest terms, its denominator will be some
factor of n, so (a/b − c/d) ∈ H as required. Therefore H ≤ Q.
(d) for fixed n ∈ Z+ the set of rational numbers whose denominators are
relatively prime to n (under addition)
Proof. Again, let H denote the subset, which is clearly nonempty. Take
a/b and c/d in H, so that (b, n) = (d, n) = 1. Then
a c ad − bc
− = .
b d bd
Let k = (bd, n). If k > 1, then there is a prime number m which divides
k. Then m | bd which implies m | b or m | d, which is impossible since
m | n. Therefore k = 1 and a/b − c/d ∈ H. So H ≤ Q.
(e) the set of nonzero real numbers whose square is a rational number (under
multiplication)
Proof. Let H be the set in question, which is clearly nonempty. If a, b ∈ H,

then a2 , b2 ∈ Q. Then
a 2 a2
= 2 ∈ Q,
b b
so a/b ∈ H. Hence H ≤ R× .
2.1.2 Exercise 2
In each of (a)–(e) prove that the specified subset is not a subgroup of the given
group:
(a) the set of 2-cycles in Sn for n ≥ 3
Proof. For any n ≥ 3, the 2-cycles (1 2) and (2 3) are members of Sn , yet

(2 3)(1 2) = (1 3 2) which is not a 2-cycle. So this set is not a subgroup.
(b) the set of reflections in D2n for n ≥ 3
Proof. Since s and sr are reflections, but s(sr) = r is not, this set is not
closed under the group operation so it is not a subgroup.
(c) for n a composite integer > 1 and G a group containing an element of

order n, the set {x ∈ G | |x| = n} ∪ {1}
Proof. Let p | n for p a prime. Then (xn/p )p = xn = 1, so |xn/p | < n and

the set is not closed under the group operation.
(d) the set of (positive and negative) odd integers in Z together with 0
Proof. Since 1+1 = 2, this set is not closed under addition and is therefore
not a subgroup.
(e) the set of real numbers whose square is a rational number (under addition)
√ √ √ √
Proof. 2 and 3 are in this subset, but 2 + 3 is not, so this cannot
be a subgroup.
2.1. DEFINITION AND EXAMPLES 77
2.1.3 Exercise 3
Show that the following subsets of the dihedral group D8 are actually subgroups:
(a) {1, r2 , s, sr2 }
Proof. This is a finite group so it suffices to show that it is closed under

the group operation of composition. We have
r2 (r2 ) = 1,
r2 (s) = sr2 ,
r2 (sr2 ) = s,
s(r2 ) = sr2 ,
s2 = 1,
s(sr2 ) = r2 ,
sr2 (r2 ) = s,
sr2 (s) = r2 ,
sr2 (sr2 ) = 1.
Therefore this subset is a subgroup.
(b) {1, r2 , sr, sr3 }
Proof. Again, we can simply enumerate the possibilities. We find that

r2 (r2 ) = sr(sr) = sr3 (sr3 ) = 1,
r2 (sr) = sr(r2 ) = sr3 ,
r2 (sr3 ) = sr3 (r2 ) = sr,
and
sr(sr3 ) = sr3 (sr) = r2 .
Therefore this is a subgroup.
2.1.4 Exercise 4
Give an explicit example of a group G and an infinite subset H of G that is
closed under the group operation but is not a subgroup of G.
Solution. Let G = R× with the operation of multiplication. Then if H is the
nonzero integers, H is closed under multiplication but is not a subgroup since
it is not closed under inverses (for example, 2 has no inverse in H).
2.1.5 Exercise 5
Prove that G cannot have a subgroup H with |H| = n − 1, where n = |G| > 2.
Proof. If such a subgroup H does exist, then it must exclude exactly one element
g from G. Since |H| ≥ 2, we can take a nonidentity element h ∈ H.
Consider the element gh. If gh 6∈ H, then gh = g and cancellation implies
that h is the identity, which is a contradiction. On the other hand, if gh ∈ H,
then (gh)h−1 = g ∈ H, a contradiction. So the subgroup H does not exist.
2.1.6 Exercise 6
Let G be an abelian group. Prove that {g ∈ G | |g| < ∞} is a subgroup of G
(called the torsion subgroup of G). Give an explicit example where this set is
not a subgroup when G is non-abelian.
Solution. Let G be abelian and let H be the elements of G having finite order.
H is nonempty since 1 ∈ H. Suppose a, b ∈ H. Then |a| = m and |b| = n for
some finite m and n. Since G is abelian we have
(ab−1 )mn = amn (bmn )−1 = 1.
Therefore ab−1 ∈ H and H is a subgroup of G.

Now, for a non-abelian counterexample, consider the group of invertible
functions from R → R under function composition. Let f : R → R and g : R → R
be given by
f (x) = −x and g(x) = 1 − x.
Then f and g have order 2 but f ◦ g, given by x 7→ x − 1, has infinite order.
2.1.7 Exercise 7
Fix some n ∈ Z with n > 1. Find the torsion subgroup of Z × (Z/nZ). Show
that the set of elements of infinite order together with the identity is not a
subgroup of this direct product.
Solution. Let G = Z × (Z/nZ) (with componentwise addition) and let H be the

torsion subgroup. Since every nonzero integer has infinite order, members of H
must have the form (0, k) for k ∈ Z/nZ. But Z/nZ is a finite group, so all of its
members have finite order. Therefore H = {(0, k) | k ∈ Z/nZ}. And we know
that this is a subgroup by the previous exercise.
Now let K be the set of elements of G having infinite order together with
the identity. Then (1, 1) ∈ K and (−1, 0) ∈ K, but (1, 1) + (−1, 0) = (0, 1) 6∈ K.
Therefore K is not a subgroup of G.
2.1.8 Exercise 8
Let H and K be subgroups of G. Prove that H ∪ K is a subgroup if and only
if either H ⊆ K or K ⊆ H.
Proof. Suppose H ∪ K is a subgroup of G. If H ⊆ K then we are done.

Suppose H 6⊆ K so that there is h ∈ H such that h 6∈ K. Let k be any element
in K. Since h, k ∈ H ∪ K, we must have hk ∈ H ∪ K. But if hk ∈ K, then
hk(k −1 ) = h ∈ K, which contradicts the choice of h. So hk ∈ H. And h−1 ∈ H,
so h−1 (hk) = k ∈ H. Hence every element of K is in H so that K ⊆ H.
Conversely, suppose H ⊆ K. Then H ∪ K = K is a subgroup. Similarly, if
K ⊆ H, then H ∪ K = H is a subgroup. This completes the proof.
2.1.9 Exercise 9
Let G = GLn (F ), where F is any field. Define
SLn (F ) = {A ∈ GLn (F ) | det(A) = 1}
(called the special linear group). Prove that SLn (F ) ≤ GLn (F ).

Proof. First, note that SLn (F ) is nonempty since the identity matrix I has
determinant 1. Now let A, B ∈ SLn (F ). We know from linear algebra that
det(AB −1 ) = det(A) det(B −1 ) = det(A) det(B)−1 = 1,
so AB −1 ∈ SLn (F ), which shows that SLn (F ) is a subgroup.
2.1.10 Exercise 10
(a) Prove that if H and K are subgroups of G then so is their intersection
H ∩ K.
Proof. Since 1 ∈ H and 1 ∈ K, 1 ∈ H ∩ K and the intersection is

nonempty. For any a, b ∈ H ∩ K, we must have ab−1 ∈ H since a, b ∈ H
and H is a subgroup. Similarly we must have ab−1 ∈ K, so ab−1 ∈ H ∩ K
and H ∩ K ≤ G.
(b) Prove that the intersection of an arbitrary nonempty collection of sub-

groups of G is again a subgroup of G (do not assume the collection is
countable).
Proof. Let Hα be a subgroup of G for all α belonging to some set of indices

A. Let \
H= Hα .
α∈A
Then 1 ∈ H so H is nonempty. If a, b ∈ H, then for any α we have

a, b ∈ Hα , so ab−1 ∈ Hα and we see that ab−1 ∈ H as well. Hence
H ≤ G.
2.1.11 Exercise 11
Let A and B be groups. Prove that the following sets are subgroups of the
direct product A × B:
(a) {(a, 1) | a ∈ A}
Proof. Call the set H. Then (1, 1) ∈ H so H is nonempty. For any

a1 , a2 ∈ A we have a1 a−1
2 ∈ A, so (a1 , 1)(a2 , 1)−1 = (a1 a−1
2 , 1) ∈ H.
Therefore H ≤ A × B.
(b) {(1, b) | b ∈ B}
Proof. The proof is almost the same as in part (a): H is nonempty, and for
any b1 , b2 ∈ B we have (1, b1 )(1, b2 )−1 = (1, b1 b−1
2 ) ∈ H, so H ≤ A×B.
(c) {(a, a) | a ∈ A}, where here we assume B = A (called the diagonal

subgroup)
Proof. Again, call the subset H. (1, 1) ∈ H so H is nonempty. For any

a1 , a2 ∈ A, we have a1 a−1
2 ∈ A so (a1 , a1 )(a2 , a2 )
−1
= (a1 a−1 −1
2 , a1 a2 ) ∈ H.
2
Therefore H is a subgroup of A .
2.1.12 Exercise 12
Let A be an abelian group and fix some n ∈ Z. Prove that the following sets
are subgroups of A:
(a) {an | a ∈ A}
Proof. Call the subset H. Then 1n = 1 ∈ H so H is nonempty. If

an , bn ∈ H then, since A is abelian,
an (bn )−1 = an (b−1 )n = (ab−1 )n .
Therefore an (bn )−1 ∈ H and H ≤ A.
(b) {a ∈ A | an = 1}
Proof. Again, call the set H. Then 1 ∈ H so H is nonempty. Suppose

a, b ∈ H. Then an = 1 and (b−1 )n = (bn )−1 = 1−1 = 1. Since A is abelian,
we have (ab−1 )n = an (b−1 )n = 1 so ab−1 ∈ H. Therefore H ≤ A.
2.1.13 Exercise 13
Let H be a subgroup of the additive group of rational numbers with the property
that 1/x ∈ H for every nonzero element x of H. Prove that H = 0 or Q.
Proof. Suppose H is a subgroup of Q with the given property. Certainly 0 ∈ H.

If H = 0 then there is nothing left to prove, so suppose H 6= 0. Then x ∈ H
for some nonzero x ∈ Q. And we may take x to be positive, since if x < 0 then
−x > 0 and −x ∈ H since H is closed under additive inverses.
Write x = a/b for positive integers a and b. Since H is closed under addition,
we have
b terms
z }| {
a a a
bx = + + · · · + = a ∈ H.
b b b
Also, a is nonzero, so by hypothesis 1/a ∈ H. By the same reasoning as above,
we have a(1/a) = 1 ∈ H. And since H is closed under addition and inverses,
this shows that Z ⊆ H.
Now, let r ∈ Q be arbitrary and write r = p/q for integers p and q (with q
nonzero). Since q ∈ H, we have 1/q ∈ H and so p(1/q) = p/q ∈ H. This shows
that Q ⊆ H. But H ⊆ Q, so H = Q and the proof is complete.
2.1.14 Exercise 14
Show that {x ∈ D2n | x2 = 1} is not a subgroup of D2n (here n ≥ 3).
Proof. In D2n , s2 = 1 and (sr)2 = srsr = s2 r−1 r = 1, so these elements are in
the subset. However, their product s(sr) = r has order n > 2. So this set is not
closed under the group operation and thus is not a subgroup.
2.1.15 Exercise 15
Let H1 ≤ H2 ≤ · · · be an ascending chain of subgroups of G. Prove that
S ∞
i=1 Hi is a subgroup of G.
S∞
Proof. Let H = i=1 Hi . Obviously 1 ∈ H, so H is nonempty. Let a, b ∈ H.
Then a ∈ Hi for some i and b ∈ Hj for some j. If k = max(i, j), then a and b
both belong to Hk , so ab−1 belongs also to Hk . Therefore ab−1 ∈ H as required.
Hence H ≤ G.
2.1.16 Exercise 16
Let n ∈ Z+ and let F be a field. Prove that the set
{(aij ) ∈ GLn (F ) | aij = 0 for all i > j}
is a subgroup of GLn (F ) (called the group of upper triangular matrices).

Proof. Fix an n ∈ Z+ and call the set of n × n upper triangular matrices H. H
is nonempty, since the identity matrix is in H.
Let A, B ∈ H, where the ijth entry of A is aij and the corresponding entry
of B is bij . If AB = C with C = (cij ) then
n
X
cij = aik bkj ,
k=0
and if i > j then this sum must be 0 since aik = 0 for k < i and bkj = 0 for
k ≥ i > j. Therefore H is closed under multiplication.
Lastly, we need to show that H is closed under inverses. Consider the matrix
A. Since A ∈ GLn (F ) we know that A is invertible. And since the determinant
of an upper triangular matrix is the product of the diagonal entries, we must
have aii 6= 0 for each i.
Let D = A−1 , so that DA = I for D = (dij ). Suppose that D is not upper
triangular, and let dij be nonzero for some i > j. Suppose also that dij is the
first nonzero entry in row i. Then
n
X
dik akj = 0
k=1
since DA = I. But dik = 0 for each k < j since dij is the first nonzero entry
in the row. And akj = 0 for each k > j since A is upper triangular. Therefore
the only term which survives is dij ajj , which is nonzero. But then the sum is
nonzero, which gives a contradiction. Therefore D is upper triangular and H is
closed under inverses. This shows that H ≤ GLn (F ).
2.1.17 Exercise 17
Let n ∈ Z+ and let F be a field. Prove that the set
{(aij ) ∈ GLn (F ) | aij = 0 for all i > j, and aii = 1 for all i}
is a subgroup of GLn (F ).
Proof. Again, call the set H. We know H is nonempty since I ∈ H.
Let A, B ∈ H. By the previous exercise we know that the product AB must
be upper triangular. So we need only check that the diagonal entries are each
1. For each i, we have
Xn
aik bki = aii bii = 1,
k=1
since all nondiagonal terms are 0 (because aik = 0 for k < i and bki = 0 for
k > i). Therefore H is closed under products.
Now let D = (dij ) be such that DA = I. Again, by the previous exercise we
know that D must be an upper triangular matrix, so we need only ensure that
the diagonal entries are each 1. For each i, we have
n
X
1= dik aki = dii aii = dii ,
k=1
since nondiagonal terms are 0. Therefore H is closed under inverses, and H ≤

GLn (F ).
2.2. CENTRALIZERS AND NORMALIZERS 83
2.2 Centralizers and Normalizers,

Stabilizers and Kernels
2.2.1 Exercise 1
Prove that
CG (A) = {g ∈ G | g −1 ag = a for all a ∈ A}.
Proof. By multiplying on the left by g and on the right by g −1 , we see that

g −1 ag = a if and only if gag −1 = a. Therefore the above set is a valid alternative
way to define the centralizer of A.
2.2.2 Exercise 2
Prove that CG (Z(G)) = G and deduce that NG (Z(G)) = G.
Proof. Let g ∈ G be arbitrary. If a ∈ Z(G), then in particular, ga = ag

which shows that gag −1 = a. Therefore g ∈ CG (Z(G)) for any g ∈ G, so
G ≤ CG (Z(G)). But we know CG (Z(G)) ≤ G, so this establishes equality.
Since CG (A) ≤ NG (A) for any A ⊆ G, we must have NG (Z(G)) = G.
2.2.3 Exercise 3
Prove that if A and B are subsets of G with A ⊆ B then CG (B) is a subgroup
of CG (A).
Proof. Suppose A and B are as stated. Let g ∈ CG (B). Then gbg −1 = b for any
b ∈ B. But A ⊆ B, so gag −1 = a for any a ∈ A as well. Therefore g ∈ CG (A).
This shows that CG (B) ⊆ CG (A), and since both are subgroups of G, we have
CG (B) ≤ CG (A).
2.2.4 Exercise 4
For each of S3 , D8 , and Q8 compute the centralizers of each element and find
the center of each group. Does Lagrange’s Theorem simplify your work?
Solution. The centralizer of 1 (for any group) is the entire group. The central-
izers of the other elements can be computed directly. For example, CS3 ((1 2))
must at minimum include 1 and (1 2) itself. We can test the other elements
directly (note (1 2)−1 = (1 2)):
(1 2)(1 3)(1 2) = (2 3),

(1 2)(2 3)(1 2) = (1 3),
(1 2)(1 2 3)(1 2) = (1 3 2),
(1 2)(1 3 2)(1 2) = (1 2 3).
So, CS3 ((1 2)) = {1, (1 2)}.

We can use Lagrange’s Theorem to reduce some of the checking. For ex-
ample, let a = (1 3). Then CS3 (a) must include the subgroup {1, (1 3)}, so 2
divides |CS3 (a)|. On the other hand, |CS3 (a)| divides |S3 | = 6. Therefore there
are only two possibilities, either |CS3 (a)| = 2 or 6. Since (1 3) does not commute
with (1 2), we know that the order must be 2. So CS3 (a) = {1, (1 3)}. Similarly,
we find CS3 ((2 3)) = {1, (2 3)}.
Now let a = (1 2 3). We have a−1 = (1 3 2) = a2 so CS3 (a) must contain the
cyclic subgroup {1, a, a2 } and we see that 3 divides |CS3 (a)|. So the order is
either 3 or 6. But it must be 3, since (1 2 3) does not commute with (1 2), for
example. So CS3 (a) = {1, a, a2 }. Similarly, CS3 ((1 3 2)) is this same set.
From the above results, we see that the center of S3 is Z(S3 ) = {1}, since
no non-identity element commutes with every element of S3 .
Similarly, we may find the centralizers of D8 :
CD8 (r) = {1, r, r2 , r3 },

CD8 (r2 ) = D8 ,
CD8 (r3 ) = {1, r, r2 , r3 },
CD8 (s) = {1, r2 , s, sr2 },
CD8 (sr) = {1, r2 , sr, sr3 },
CD8 (sr2 ) = {1, r2 , s, sr2 },
CD8 (sr3 ) = {1, r2 , sr, sr3 }.
And we see that Z(D8 ) = {1, r2 }.

Finally, for Q8 , we have:
CQ8 (−1) = Q8 ,
CQ8 (i) = {1, −1, i, −i},
CQ8 (−i) = {1, −1, i, −i},
CQ8 (j) = {1, −1, j, −j},
CQ8 (−j) = {1, −1, j, −j},
CQ8 (k) = {1, −1, k, −k},
CQ8 (−k) = {1, −1, k, −k}.
And Z(Q8 ) = {1, −1}.
2.2.5 Exercise 5
In each of parts (a) to (c) show that for the specified group G and subgroup A
of G, CG (A) = A and NG (A) = G.
(a) G = S3 and A = {1, (1 2 3), (1 3 2)}.
Solution. A is a cyclic subgroup generated by (1 2 3), so A ≤ CG (A). By

Lagrange’s Theorem, 3 divides |CG (A)| which divides |S3 | = 6, so either
|CG (A)| = 3 or it is 6. But it can’t be 6 since, for example, (1 2) does not
commute with (1 2 3). Therefore |CG (A)| = 3 and we see that CG (A) = A.
Since CG (A) ≤ NG (A), we again have either |NG (A)| = 3 or 6. However,
(1 2)A(1 2) = {1, (1 3 2), (1 2 3)} = A,
so |NG (A)| > 3. Therefore NG (A) = G.

(b) G = D8 and A = {1, s, r2 , sr2 }.
Solution. The elements of A all commute with one another and in fact
form a subgroup of D8 . By Lagrange, |CG (A)| = 4 or 8. But r does not
commute with s, for example, so |CG (A)| = 4 and we have CG (A) = A.
Since CG (A) ≤ NG (A), we must have either NG (A) = A or NG (A) = G.
Since
rAr−1 = {1, sr2 , r2 , s} = A,
we must have NG (A) = G.
(c) G = D10 and A = {1, r, r2 , r3 , r4 }.
Solution. A is a cyclic subgroup. Again, by Lagrange, we must have

|CG (A)| = 5 or 10. But s and r do not commute, so it must be the
former. Hence CG (A) = A.
For the normalizer, we simply note that
sAs = {1, r4 , r3 , r2 , r} = A,
so NG (A) = G.
2.2.6 Exercise 6
Let H be a subgroup of the group G.
(a) Show that H ≤ NG (H). Give an example to show that this is not neces-
sarily true if H is not a subgroup.
Solution. Let g ∈ H. First, take x ∈ gHg −1 . Then x = ghg −1 for some

h ∈ H. Since H is a subgroup and thus closed under the group operation,
we have x ∈ H. Conversely, if x ∈ H then by definition x ∈ gHg −1 .
Therefore gHg −1 = H and we see that g ∈ NG (H). Since this is true for
all g ∈ H, it follows that H ≤ NG (H).
As a counterexample for the case where H is not a subgroup, consider
G = D4 and H = {1, r, s}. Then sHs = {1, r3 , s} =
6 H, so s 6∈ NG (H)
and H 6⊆ NG (H).
(b) Show that H ≤ CG (H) if and only if H is abelian.
Proof. Suppose H ≤ CG (H) and let a, b ∈ H. Then a ∈ CG (H) so in

particular, aba−1 = b, or equivalently, ab = ba. Therefore H is abelian.
Conversely, suppose H is abelian. If a ∈ H, then ah = ha for each
h ∈ H. Equivalently, aha−1 = h, so that a ∈ CG (H). This shows that
H ≤ CG (H).
2.2.7 Exercise 7
Let n ∈ Z with n ≥ 3. Prove the following:
(a) Z(D2n ) = 1 if n is odd
Proof. Suppose rk ∈ Z(D2n ) for k ∈ Z+ . Since srk = r−k s, we must

have rk = r−k . Therefore r2k = 1 and we see that 2 | n, which is a
contradiction.
s can’t be in the center since it doesn’t commute with r. Now suppose
srk ∈ Z(D2n ), with k ∈ Z+ . In order to commute with r, we must have
(srk )r = r(srk ). This implies srk+1 = srk−1 , so rk+1 = rk−1 and we see
that r2 = 1, which means n ≤ 2, another contradiction.
Therefore the only element in Z(D2n ) is 1.
(b) Z(D2n ) = {1, rk } if n = 2k
Proof. From the previous proof we know that the only possible candidates
are 1 and rk where n = 2k. And since r2k = 1 we see that rk = r−k . Any
element x in D2n can be written as x = si rj for i ∈ {0, 1} and j ∈ Z,
j ≥ 0, so,
rk (si rj ) = si r−k rj = si rk rj = si rk+j = (si rj )rk .
Hence Z(D2n ) = {1, rk }.
2.2.8 Exercise 8
Let G = Sn , fix an i ∈ {1, 2, . . . , n} and let Gi = {σ ∈ G | σ(i) = i} (the
stabilizer of i in G). Use group actions to prove that Gi is a subgroup of G.
Find |Gi |.
Proof. Let A = {1, 2, . . . , n}. We know that Sn acts on A by σ · j = σ(j). Then
Gi = {σ ∈ G | σ · i = i}.
Now 1 ∈ Gi by definition of a group action, so Gi is nonempty. Suppose

σ, τ ∈ Gi . Again, by definition of an action,
στ · i = σ · (τ · i) = σ · i = i,
so Gi is closed under composition. And, since
i = (b−1 b) · i = b−1 · (b · i) = b−1 · i,
we see that Gi is closed under inverses. Therefore Gi ≤ G.

Lastly, since every member of Gi fixes i, we have that Gi ∼
= Sn−1 so that
|Gi | = (n − 1)!.
2.2.9 Exercise 9
For any subgroup H of G and any nonempty subset A of G define NH (A) to
be the set {h ∈ H | hAh−1 = A}. Show that NH (A) = NG (A) ∩ H and deduce
that NH (A) is a subgroup of H (note that A need not be a subset of H).
Proof. Certainly NH (A) ⊆ NG (A), since every member h of NH (A) is a member

of G for which hAh−1 = A. And NH (A) ⊆ H, so NH (A) ⊆ NG (A) ∩ H.
Now pick h ∈ NG (A) ∩ H. Since h ∈ NG (A), we have hAh−1 = A. And
since h ∈ H, we see that h ∈ NH (A). This shows that NG (A) ∩ H ⊆ NH (A).
Therefore NH (A) = NG (A)∩H and NH (A) must be a subgroup of G (and hence
H) since it is the intersection of two subgroups of G (see Exercise 2.1.10).
2.2.10 Exercise 10
Let H be a subgroup of order 2 in G. Show that NG (H) = CG (H). Deduce
that if NG (H) = G then H ≤ Z(G).
Proof. Let H = {1, a} and suppose g ∈ NG (H). Then gHg −1 = H. Since

g1g −1 = 1, we must have gag −1 = a. Therefore g ∈ CG (H) and we see that
NG (H) ≤ CG (H).
Now suppose g ∈ CG (H). Then g1g −1 = 1 and gag −1 = a, so gHg −1 = H
and g ∈ NG (H). Hence CG (H) ≤ NG (H) and in fact CG (H) = NG (H).
Finally, if NG (H) = G, then CG (H) = G so that H ≤ Z(G).
2.2.11 Exercise 11
Prove that Z(G) ≤ NG (A) for any subset A of G.
Proof. Fix a subset A of G. Let g ∈ Z(G). Then gag −1 = a for all a ∈ A, so

this shows that gAg −1 = A. Therefore Z(G) ≤ NG (A).
2.2.12 Exercise 12
Let R be the set of all polynomials with integer coefficients in the independent
variables x1 , x2 , x3 , x4 i.e., the members of R are finite sums of elements of
the form axr11 xr22 xr33 xr44 , where a is any integer and r1 , . . . , r4 are nonnegative
integers. For example,
12x51 x72 x4 − 18x32 x3 + 11x61 x2 x33 x23

4 (2.1)
is a typical element of R. Each σ ∈ S4 gives a permutation of {x1 , . . . , x4 } by

defining σ · xi = xσ(i) . This may be extended to a map from R to R by defining
σ · p(x1 , x2 , x3 , x4 ) = p(xσ(1) , xσ(2) , xσ(3) , xσ(4) )
for all p(x1 , x2 , x3 , x4 ) ∈ R (i.e., σ simply permutes the indices of the variables).
For example, if σ = (1 2)(3 4) and p(x1 , . . . , x4 ) is the polynomial in (2.1) above,
then
σ · p(x1 , x2 , x3 , x4 ) = 12x52 x71 x3 − 18x31 x4 + 11x62 x1 x34 x23

3
= 12x71 x52 x3 − 18x31 x4 + 11x1 x62 x23 3
3 x4 .
(a) Let p = p(x1 , . . . , x4 ) be the polynomial in (2.1) above, let σ = (1 2 3 4)

and let τ = (1 2 3). Compute σ · p, τ · (σ · p), (τ ◦ σ) · p, and (σ ◦ τ ) · p.
Solution. τ ◦ σ = (1 3 4 2) and σ ◦ τ = (1 3 2 4). So

σ · p = 12x1 x52 x73 − 18x33 x4 + 11x23 6 3
1 x2 x3 x4 ,
τ · (σ · p) = 12x71 x2 x53 − 18x31 x4 + 11x1 x23 6 3
2 x3 x4 ,
(τ ◦ σ) · p = 12x71 x2 x53 − 18x31 x4 + 11x1 x23 6 3
2 x3 x4 ,
and
(σ ◦ τ ) · p = 12x1 x53 x74 − 18x2 x34 + 11x23 3 6

1 x2 x3 x4 .
(b) Prove that these definitions give a (left) group action of S4 on R.
Proof. Clearly 1 · p = p for any p ∈ R. For any σ, τ ∈ S4 we have

σ · (τ · p(x1 , x2 , x3 , x4 )) = σ · p(xτ (1) , xτ (2) , xτ (3) , xτ (4) )
= p(xσ(τ (1)) , xσ(τ (2)) , xσ(τ (3)) , xσ(τ (4)) )
= (σ ◦ τ ) · p(x1 , x2 , x3 , x4 ).
Therefore the mapping is a group action.
(c) Exhibit all permutations in S4 that stabilize x4 and prove that they form
a subgroup isomorphic to S3 .
Solution. The stabilizer of x4 consists of all permutations which fix x4 :

1, (1 2), (1 3), (2 3), (1 2 3), (1 3 2). But these correspond precisely to the
elements of S3 , so these permutations form a subgroup of S4 which is
isomorphic to S3 .
(d) Exhibit all permutations in S4 that stabilize the element x1 +x2 and prove
that they form an abelian subgroup of order 4.
Solution. If σ is a member of the stabilizer Rx1 +x2 then σ · (x1 + x2 ) =

x1 + x2 . There are two ways this can happen: σ can fix 1 and 2, or σ can
send 1 to 2 and vice versa. So the permutations in the stabilizer are 1,
(1 2), (3 4), and (1 2)(3 4). All of these elements commute with each other
(since they consist of disjoint cycles), so Rx1 +x2 is an abelian subgroup of
order 4.
(e) Exhibit all permutations in S4 that stabilize the element x1 x2 + x3 x4 and

prove that they form a subgroup isomorphic to the dihedral group of order
8.
Solution. The permutations in S4 which stabilize x1 x2 + x3 x4 are 1, (1 2),

(3 4), (1 2)(3 4), (1 3)(2 4), (1 4)(2 3), (1 3 2 4), and (1 4 2 3).
Let ϕ : Rx1 x2 +x3 x4 → D8 be the map for which
ϕ(1 2) = s and ϕ(1 3 2 4) = r.
Then ϕ extends to an isomorphism since (1 2)2 = 1, (1 3 2 4)4 = 1 and
(1 2)(1 3 2 4) = (1 3 2 4)−1 (1 2).
(f) Show that the permutations in S4 that stabilize the element (x1 +x2 )(x3 +
x4 ) are exactly the same as those found in part (e).
Solution. Checking each possibility in turn will show that the permuta-
tions in this stabilizer are exactly the same as those in the previous part
of the problem.
2.2.13 Exercise 13
Let n be a positive integer and let R be the set of all polynomials with integer
coefficients in the independent variables x1 , x2 , . . . , xn , i.e., the members of R
are finite sums of elements of the form axr11 xr22 · · · xrnn , where a is any integer
and r1 , . . . , rn are nonnegative integers. For each σ ∈ Sn define a map
σ: R → R by σ · p(x1 , x2 , . . . , xn ) = p(xσ(1) , xσ(2) , . . . , xσ(n) ).
Prove that this defines a (left) group action of Sn on R.
Proof. Clearly 1 · p(x1 , . . . , xn ) = p(x1 , . . . , xn ). And for σ, τ ∈ Sn we have
σ · (τ · p(x1 , x2 , . . . , xn )) = σ · p(xτ (1) , xτ (2) , . . . , xτ (n) )
= p(xσ(τ (1)) , xσ(τ (2)) , . . . , xσ(τ (n)) )
= (σ ◦ τ ) · p(x1 , x2 , . . . , xn ).
Therefore this mapping does define a group action on R.
2.2.14 Exercise 14
Let H(F ) be the Heisenberg group over the field F introduced in Exercise 1.4.11.
Determine which matrices lie in the center of H(F ) and prove that Z(H(F )) is
isomorphic to the additive group F .
Solution. Let    
1 a b 1 d e
X = 0 1 c and Y = 0 1 f
0 0 1 0 0 1
be elements of H(F ). If X ∈ Z(H(F )) then XY = Y X for any Y ∈ H(F ).
Computing XY and Y X for the matrices above gives
    
1 a b 1 d e 1 a + d af + b + e
XY = 0 1 c  0 1 f  = 0 1 c+f 
0 0 1 0 0 1 0 0 1
and     
1 d e 1 a b 1 a+d b + cd + e
Y X = 0 1 f  0 1 c  = 0 1 c + f .
0 0 1 0 0 1 0 0 1
So af + b + e = b + cd + e or af = cd. Since Y can be arbitrary, the only way
to guarantee this is for a = c = 0. If a and c are both nonzero, then any Y with
f = 0 and d = 1 will not commute with X. Therefore,
  
 1 0 a

Z(H(F )) = 0 1 0 ∈ H(F ) a ∈ F .
0 0 1
 
We can see that Z(H(F )) ∼

= F since the map ϕ : F → Z(H(F )) given by
 
1 0 a
ϕ(a) = 0 1 0
0 0 1
is an isomorphism.
2.3. CYCLIC GROUPS AND CYCLIC SUBGROUPS 91
2.3 Cyclic Groups and Cyclic Subgroups

2.3.1 Exercise 1
Find all subgroups of Z45 = hxi, giving a generator for each. Describe the
containments between these subgroups.
Solution. The subgroups are generated by xd where d divides 45. And we have
hxa i ≤ hxb i if (b, 45) | (a, 45). This gives the following subgroup relationships:
Z45 = hxi > hx3 i, hx5 i, hx9 i, hx15 i, 1,

hx3 i > hx9 i, hx15 i, 1,
hx5 i > hx15 i, 1,
hx9 i > 1,
hx15 i > 1,
1 = hx0 i.
2.3.2 Exercise 2
If x is an element of the finite group G and |x| = |G|, prove that G = hxi. Give
an explicit example to show that this result need not be true if G is an infinite
group.
Proof. Let x ∈ G where |x| = |G| = n < ∞. By Proposition 2, we know that

1, x, x2 , . . . , xn−1 are all distinct elements in G. But G contains only n elements,
so this must be the entirety of G. Therefore G = hxi.
This is not always true if |x| = |G| = ∞. For example, in the additive group
Z, |2| = |Z| = ∞ but clearly Z is not generated by 2.
2.3.3 Exercise 3
Find all generators for Z/48Z.
Solution. The generators are those residue classes whose representatives are
relatively prime to 48. Therefore the generators are 1̄, 5̄, 7̄, 11, 13, 17, 19, 23,
25, 29, 31, 35, 37, 41, 43, and 47.
2.3.4 Exercise 4
Find all generators for Z/202Z.
Solution. 202 = 2 · 101, so the generators are all residue classes having odd
representatives excluding 101.
2.3.5 Exercise 5
Find the number of generators for Z/49000Z.
Solution. If ϕ denotes the Euler ϕ-function, then the number of generators is

given by
ϕ(49000) = ϕ(23 )ϕ(53 )ϕ(72 )
= 22 (2 − 1)52 (5 − 1)7(7 − 1)
= 4 · 100 · 42
= 16800.
2.3.6 Exercise 6
In Z/48Z write out all elements of hāi for every ā. Find all inclusions between
subgroups in Z/48Z.
Solution. The elements of each subgroup are
Z/48Z = h1̄i = {0̄, 1̄, 2̄, 3̄, . . . , 46, 47},
h2̄i = {0̄, 2̄, 4̄, 6̄, . . . , 44, 46},
h3̄i = {0̄, 3̄, 6̄, 9̄, . . . , 42, 45},
h4̄i = {0̄, 4̄, 8̄, 12, . . . , 40, 44},
h6̄i = {0̄, 6̄, 12, 18, 24, 30, 36, 42},
h8̄i = {0̄, 8̄, 16, 24, 32, 40},
h12i = {0̄, 12, 24, 36},
h16i = {0̄, 16, 32},
h24i = {0̄, 24},
h0̄i = {0̄}.
And we have the following inclusions:
h0̄i, h1̄i, h2̄i, h3̄i, h4̄i, h6̄i, h8̄i, h12i, h16i, h24i ≤ h1̄i,
h0̄i, h2̄i, h4̄i, h6̄i, h8̄i, h12i, h16i, h24i ≤ h2̄i,
h0̄i, h3̄i, h6̄i, h12i, h16i, h24i ≤ h3̄i,
h0̄i, h4̄i, h8̄i, h12i, h16i, h24i ≤ h4̄i,
h0̄i, h6̄i, h12i, h24i ≤ h6̄i,
h0̄i, h8̄i, h16i, h24i ≤ h8̄i,
h0̄i, h12i, h24i ≤ h12i,
h0̄i, h16i ≤ h16i,
h0̄i, h24i ≤ h24i,
h0̄i ≤ h0̄i.
2.3.7 Exercise 7
Let Z48 = hxi and use the isomorphism Z/48Z ∼= Z48 given by 1̄ 7→ x to list all
subgroups of Z48 as computed in the preceding exercise.
Solution. The subgroups are hxi, hx2 i, hx3 i, hx4 i, hx6 i, hx8 i, hx12 i, hx16 i, hx24 i,
and 1.
2.3.8 Exercise 8
Let Z48 = hxi. For which integers a does the map ϕa defined by ϕa : 1̄ 7→ xa
extend to an isomorphism from Z/48Z onto Z48 .
Solution. Choose an a with (a, 48) = d > 1 and set b = 48/d. If ϕa is a

homomorphism, then
ϕa (b̄) = ϕa (b · 1̄) = ϕa (1̄)b = xab = (x48 )a/d = 1 = ϕa (0̄).
Therefore, in this case, ϕa is not an injection and thus not an isomorphism.

This suggests that ϕa extends to an isomorphism if and only if (a, 48) = 1,
which we will now prove. First we show that the function b̄ 7→ xab is well defined,
i.e., that the value of the function is not affected by the choice of representative
for b̄. Suppose b̄ = c̄. Then 48k = b − c for some integer k, and we have
ϕa (b̄) = xab = xa(48k+c) = (x48 )ak xac = 1ak xac = xac = ϕa (c̄).
Now, ϕa is certainly a homomorphism, since
ϕa (b̄ + c̄) = xa(b+c) = xab xac = ϕa (b̄)ϕa (c̄).
To show injectivity, suppose ϕa (b̄) = ϕa (c̄). Then xab = xac or xa(b−c) = 1.

Hence 48 | a(b − c) and since (a, 48) = 1 we have 48 | (b − c). This shows that
b̄ = c̄.
Finally, since |Z/48Z| = |Z48 | < ∞, we know that injectivity of ϕa implies
surjectivity, so that ϕa is an isomorphism.
2.3.9 Exercise 9
Let Z36 = hxi. For which integers a does the map ψa defined by ψa : 1̄ 7→ xa
extend to a well defined homomorphism from Z/48Z into Z36 . Can ψa ever be
a surjective homomorphism?
Solution. Suppose b̄ = c̄ for integers b and c. If ψa is well defined then ψa (b̄) =

ψa (c̄), that is, xab = xac . Then xa(b−c) = 1 so we must have 36 | a(b − c).
But 48 | (b − c), so there is an integer k for which 48k = b − c and we have
that 36 | 48ak. If we choose b̄ and c̄ so that k = 1, then we must have 3 | a
as a necessary condition for 36 | 48ak. It is also sufficient that 3 | a, since
36 | 144mk.
Since
ψa (b̄ + c̄) = xa(b+c) = xab xac = ψa (b̄)ψa (c̄),
we see that ψa is a well defined homomorphism if and only if 3 | a.
Lastly, suppose ψa (b̄) = x. Since a = 3k for some integer k, we have
x = ψa (b̄) = xab = x3kb .
Therefore x3kb−1 = 1 and we see that 36 divides 3kb − 1. But this is impossible
since if 36m = 3kb − 1 then 1 = 3kb − 36m = 3(kb − 12m) and 3 | 1, a
contradiction. So the homomorphism ψa can never be surjective.
2.3.10 Exercise 10
What is the order of 30 in Z/54Z? Write out all the elements and their orders
in h30i.
Solution. Since |1̄| = 54, by Proposition 5 (2) we have
54 54
|30| = |30 · 1̄| = = = 9.
(30, 54) 6
Then
h30i = {0̄, 6̄, 12, 18, 24, 30, 36, 42, 48}
where
|0̄| = 1,
|6̄| = 9,
|12| = 9,
|18| = 3,
|24| = 9,
|30| = 9,
|36| = 3,
|42| = 9,
|48| = 9.
2.3.11 Exercise 11
Find all cyclic subgroups of D8 . Find a proper subgroup of D8 which is not
cyclic.
Solution. The cyclic subgroups are
h1i = 1,
hri = hr3 i = {1, r, r2 , r3 },
hr2 i = {1, r2 },
hsi = {1, s},
hsri = {1, sr},
hsr2 i = {1, sr2 },
hsr3 i = {1, sr3 }.
A proper subgroup that is not cyclic is hs, r2 i = {1, r2 , s, sr2 }.
2.3.12 Exercise 12
Prove that the following groups are not cyclic:
(a) Z2 × Z2
Proof. Let Z2 = hxi. Checking each element of Z2 × Z2 , we see that none

generate the whole group:
h(1, 1)i = {(1, 1)},

h(1, x)i = {(1, 1), (1, x)},
h(x, 1)i = {(1, 1), (x, 1)},
and
h(x, x)i = {(1, 1), (x, x)}.
Therefore Z2 × Z2 is not cyclic.
(b) Z2 × Z
Proof. If Z2 × Z is cyclic, then it must have a generator of the form (1, n)

or (x, n) for some n ∈ Z. But (1, n) cannot be a generator since it only
generates elements whose first component is 1.
So the generator must have the form (x, n). Now n can only be 1 or −1,
since otherwise we could not get all the integers in the second component.
But neither of these is a generator since, for example, (1, 1) is not in either
cyclic subgroup. Therefore Z2 × Z is not cyclic.
(c) Z × Z
Proof. Any generator for Z × Z must have the form (±1, ±1) since there
is no other way to generate all of the integers in each component. But
every element in a subgroup generated by (±1, ±1) must have components
which differ only in sign. For example, none of these elements will generate
(1, 2). Therefore Z × Z is not cyclic.
2.3.13 Exercise 13
Prove that the following pairs of groups are not isomorphic:
(a) Z × Z2 and Z
Proof. Isomorphisms preserve order of elements, so Z × Z2 cannot be iso-

morphic to Z since the element (0, x) in Z × Z2 has order 2 but no element
in Z has order 2.
(b) Q × Z2 and Q
Proof. Again, Q has no elements of order 2, but |(0, x)| = 2 in Q × Z2 .

Hence the two groups are not isomorphic.
2.3.14 Exercise 14
Let σ = (1 2 3 4 5 6 7 8 9 10 11 12). For each of the following integers a
compute σ a : a = 13, 65, 626, 1195, −6, −81, −570 and −1211.
Solution. Since |σ| = 12, the powers of σ consist of exactly 12 distinct elements.
We can use the Division Algorithm to reduce arbitrary powers to their least
residues. For example,
626 = 52(12) + 2,
so σ 626 = (σ 12 )52 σ 2 = σ 2 . Applying this process for each of the given values
produces the following permutations:
σ 13 = σ 1(12)+1 = σ = (1 2 3 4 5 6 7 8 9 10 11 12)
σ 65 = σ 5(12)+5 = σ 5 = (1 6 11 4 9 2 7 12 5 10 3 8)
σ 626 = σ 52(12)+2 = σ 2 = (1 3 5 7 9 11)(2 4 6 8 10 12)
σ 1195 = σ 99(12)+7 = σ 7 = (1 8 3 10 5 12 7 2 9 4 11 6)
σ −6 = σ −1(12)+6 = σ 6 = (1 7)(2 8)(3 9)(4 10)(5 11)(6 12)
σ −81 = σ −7(12)+3 = σ 3 = (1 4 7 10)(2 5 8 11)(3 6 9 12)
σ −570 = σ −48(12)+6 = σ 6 = (1 7)(2 8)(3 9)(4 10)(5 11)(6 12)
σ −1211 = σ −101(12)+1 = σ = (1 2 3 4 5 6 7 8 9 10 11 12).
2.3.15 Exercise 15
Prove that Q × Q is not cyclic.
Proof. In Exercise 2.3.12 we showed that Z × Z is not cyclic. But Z × Z is
a subgroup of Q × Q, and a cyclic group cannot have a non-cyclic subgroup.
Therefore Q × Q is not cyclic.
2.3.16 Exercise 16
Assume |x| = n and |y| = m. Suppose that x and y commute: xy = yx. Prove
that |xy| divides the least common multiple of m and n. Need this be true if x
and y do not commute? Give an example of commuting elements x, y such that
the order of xy is not equal to the least common multiple of |x| and |y|.
Solution. Let ` be the least common multiple of m and n. Then there are
integers a and b such that am = ` and bn = `. So if |x| = m and |y| = n for
commuting elements x and y, then
(xy)` = x` y ` = (xm )a (y n )b = 1.
Therefore |xy| must divide ` by Proposition 3, which completes the proof.
We note that this need not be true if x and y do not commute. For example,
in the symmetric group S3 , |(1 2)| = |(2 3)| = 2 but (1 2)(2 3) = (1 2 3) which
has order 3. Clearly 3 - 2.
Finally, for an example where x and y commute but the order of xy does
not equal the least common multiple of |x| and |y|, consider the cyclic group
Z10 . This group is abelian so all elements commute, and we have |x2 | = 5 and
|x3 | = 10, but |x5 | = 2 6= 10.
2.3.17 Exercise 17
Find a presentation for Zn with one generator.
Solution. We know that Zn can be generated by a single element x which satis-

fies the one relation xn = 1. Moreover, any group generated by a single element
and satisfying only this relation must be isomorphic to Zn , since all cyclic groups
of the same order are isomorphic. So one possible presentation is
Zn = hx | xn = 1i.
2.3.18 Exercise 18
Show that if H is any group and h is an element of H with hn = 1, then there
is a unique homomorphism from Zn = hxi to H such that x 7→ h.
Proof. Define the function ϕ : Zn → H by
ϕ(xn ) = hn .
First we need to show that ϕ is well defined. Suppose xa = xb . Then xa−b = 1

so n | (a − b) (Proposition 3). So there is an integer c such that cn = a − b. We
may then write a = cn + b so that
ϕ(xa ) = ha = hcn+b = (hn )c hb = hb = ϕ(xb )
as required.
To show that ϕ is a homomorphism, consider two arbitrary elements y = xk
and z = x` in Zn . By the exponent rules established in Exercise 1.1.19, we have
ϕ(yz) = ϕ(xk x` ) = ϕ(xk+` ) = hk+` = hk h` = ϕ(xk )ϕ(x` ) = ϕ(y)ϕ(z),
so ϕ is indeed a homomorphism.
Lastly, to show uniqueness, suppose ψ : Zn → H is any homomorphism such
that ψ(x) = h. Then for any integer k, we wish to show that we must have
ψ(xk ) = hk .
Note that we only need to consider 0 ≤ k ≤ n − 1 since any other power is equal
to one of these. We now proceed by induction on k. ψ(x) = h by assumption,
so the base case is satisfied. Suppose ψ(xk ) = hk for some nonnegative integer
k. Then by the definition of a homomorphism and by the inductive hypothesis,
ψ(xk+1 ) = ψ(xk x) = ψ(xk )ψ(x) = hk h = hk+1 ,
which establishes that ϕ = ψ and thus completes the proof.
2.3.19 Exercise 19
Show that if H is any group and h is an element of H, then there is a unique
homomorphism from Z to H such that 1 7→ h.
Proof. Let ϕ : Z → H be given by
ϕ(n) = hn .
Then ϕ is a function which maps 1 to h. It is also a homomorphism, since for

any integers m and n,
ϕ(m + n) = hm+n = hm hn = ϕ(m)ϕ(n).
Finally, this homomorphism is uniquely determined because any homomorphism

ψ : Z → H such that ψ(1) = h must satisfy ψ(n) = ψ(n1) = ψ(1)n = hn .
2.3.20 Exercise 20
Let p be a prime and let n be a positive integer. Show that if x is an element
n
of the group G such that xp = 1 then |x| = pm for some m ≤ n.
n
Proof. If xp = 1, then by Proposition 3 we have |x| divides pn . But the only
integers that divide a prime power pn are smaller prime powers pm (including
p0 = 1). Therefore |x| = pm for some nonnegative integer m with m ≤ n.
2.3.23 Exercise 23
Show that (Z/2n Z)× is not cyclic for any n ≥ 3.
Proof. By Theorem 7 we know that if (Z/2n Z)× is cyclic, then it must have at
most one subgroup with order 2. Therefore the proof will be complete if we can
show that (Z/2n Z)× has more than one distinct subgroup of order 2. This is
equivalent to showing that the group has more than one element with order 2.
For n ≥ 3, we have
(2n − 1)2 ≡ (−1)2 ≡ 1 (mod 2n )
and
(2n−1 − 1)2 = 22n−2 − 2n + 1

= 2n−2 2n − 2n + 1
≡1 (mod 2n ).
As long as n ≥ 3, both of the elements 2n − 1 and 2n−1 − 1 are distinct from

1̄ and hence have order 2. And the two elements are distinct, since 2n−1 6≡ 2n
(mod 2n ). Hence (Z/2n Z)× cannot be cyclic.
2.3.24 Exercise 24
Let G be a finite group and let x ∈ G.
(a) Prove that if g ∈ NG (hxi) then gxg −1 = xa for some a ∈ Z.
Proof. If g ∈ NG (hxi) then ghxig −1 = hxi. Since gxg −1 ∈ ghxig −1 , we

must have gxg −1 ∈ hxi or gxg −1 = xa for some integer a.
(b) Prove conversely that if gxg −1 = xa for some a ∈ Z then g ∈ NG (hxi).
Proof. Suppose gxg −1 = xa for an integer a.

First we will show that
gxk g −1 = (gxg −1 )k for all k ∈ Z. (2.2)
This obviously holds for k = 0, so suppose k 6= 0. Since gx−1 g −1 =

(gxg −1 )−1 , it is sufficient to show that (2.2) is true for positive integers k.
We proceed by induction on k. The base case is trivial. Suppose gxk g −1 =
(gxg −1 )k for some positive integer k. Then
gxk+1 g −1 = gxk xg −1 = gxk (g −1 g)xg −1

= (gxk g −1 )(gxg −1 ) = (gxg −1 )k+1 ,
so (2.2) holds for all integers k.

Now suppose y ∈ ghxig −1 . Then there is k ∈ Z such that y = gxk g −1 .
From the preceding paragraph, we then have y = (gxg −1 )k = xak . There-
fore y ∈ hxi so that ghxig −1 ≤ hxi.
But we know that |ghxig −1 | = |hxi| by Exercise 1.7.17. Since x has finite
order (G is finite), it follows that ghxig −1 = hxi and g ∈ NG (hxi).
2.3.25 Exercise 25
Let G be a cyclic group of order n and let k be an integer relatively prime to
n. Prove that the map x 7→ xk is surjective. Use Lagrange’s Theorem to prove
the same is true for any finite group of order n.
Proof. We will prove the general result directly. Fix an integer k relatively
prime to n and let ϕ denote the map x 7→ xk for the group G, where |G| = n.
Since (n, k) = 1, we may find a, b ∈ Z such that
ak + bn = 1.
Now let g ∈ G be arbitrary and consider the image of g a under ϕ. We have
ϕ(g a ) = g ak = g 1−bn = g(g n )−b .
We know by Langrange’s Theorem that |g| divides n (since hgi is a cyclic sub-
group of order |g|), so g n = 1. We then have
ϕ(g a ) = g,
which completes the proof that ϕ is surjective.
2.3.26 Exercise 26
Let Zn be a cyclic group of order n and for each integer a let
σa : Zn → Zn by σa (x) = xa for all x ∈ Zn .

(a) Prove that σa is an automorphism of Zn if and only if a and n are relatively

prime.
Proof. σa is a homomorphism since for any x, y ∈ Zn ,
σa (xy) = (xy)a = xa y a = σa (x)σa (y),
where the second equality holds due to the fact that Zn is abelian. There-
fore σa is an automorphism if and only if it is bijective. But Zn is finite,
so σa is bijective if and only if it is surjective.
If a and n are relatively prime, then we know by the previous exercise that
σa is surjective and hence an automorphism.
Conversely, suppose σa is a bijection, let d = (n, a), and find integers b, c
such that n = bd and a = cd. If g is a generator for Zn then
σa (g b ) = g ab = g cdb = g cn = (g n )c = 1 = σa (1).
If g b is distinct from 1, then the map σa is not injective and we have a

contradiction. Therefore g b = 1, and since |g| = n it follows that n = b so
that d = (n, a) = 1.
(b) Prove that σa = σb if and only if a ≡ b (mod n).
Proof. Let Zn = hgi.

First suppose that σa = σb . Then in particular
g a = σa (g) = σb (g) = g b
so that g a−b = 1. Thus by Proposition 3, n | (a − b). That is, a ≡ b

(mod n).
Conversely, if a ≡ b (mod n) then nc = a − b for some integer c and we
have for any g k ∈ Zn ,
σa (g k ) = g ak = g (nc+b)k = g bk (g n )ck = g bk = σb (g k ),
hence σa = σb as required.
(c) Prove that every automorphism of Zn is equal to σa for some integer a.
Proof. Let ϕ ∈ Aut(Zn ) and suppose ϕ(g) = g k , where g is a generator

for Zn . Then for any g i ∈ Zn , we have
ϕ(g i ) = ϕ(g)i = (g k )i = g ik = σk (g i ),
which shows that ϕ = σk .
(d) Prove that σa ◦ σb = σab . Deduce that the map ā 7→ σa is an isomorphism

of (Z/nZ)× onto the automorphism group of Zn (so Aut(Zn ) is an abelian
group of order ϕ(n)).
Proof. It is immediate from the definition of σa and σb that
(σa ◦ σb )(g i ) = σa (g bi ) = g abi = σab (g i )
for any g i ∈ Zn .
By part (a) above, we know that ā ∈ (Z/nZ)× if and only if σa ∈ Aut(Zn ).
So we may define
ψ : (Z/nZ)× → Aut(Zn ) by ψ(ā) = σa .
We know by part (b) that ψ is well defined. And ψ is a homomorphism

since
ψ(āb̄) = ψ(ab) = σab = σa ◦ σb = ψ(ā) ◦ ψ(b̄).
Now, ψ is injective by part (b), and it is surjective by part (c). Therefore
ψ is an isomorphism and (Z/nZ)× ∼ = Aut(Zn ). It follows that Aut(Zn ) is
an abelian group of order ϕ(n) where ϕ is the Euler ϕ-function.
2.4 Subgroups Generated by Subsets of a Group

2.4.1 Exercise 1
Prove that if H is a subgroup of G then hHi = H.
Proof. Let H be a subgroup of G. Certainly H ≤ hHi. Now suppose h ∈ hHi.

Then
\
h∈ K.
H⊆K
K≤G
But H itself is a subgroup of G containing itself as a subset, so by definition

h ∈ H. This shows that hHi ≤ H so that hHi = H as required.
2.4.2 Exercise 2
Prove that if A is a subset of B then hAi ≤ hBi. Give an example where A ⊆ B
with A 6= B but hAi = hBi.
Proof. Let a ∈ hAi. Since A ⊆ B we have a ∈ B so that a ∈ hBi. This shows

that hAi ≤ hBi.
For the requested example, simply consider Z4 with A = {x} and B =
{x, x3 }. Certainly A ⊂ B but hAi = hBi.
2.4.3 Exercise 3
Prove that if H is an abelian subgroup of a group G then hH, Z(G)i is abelian.
Give an explicit example of an abelian subgroup H of a group G such that
hH, CG (H)i is not abelian.
Proof. Let H ≤ G be an abelian subgroup and let g, h ∈ hH, Z(G)i be arbitrary.

By Proposition 9, g and h can be written as a finite product
g = g11 , . . . , gm
m
and h = hδ11 , . . . , hδnn ,
where each gi and hi (not necessarily distinct) is in H ∪ Z(G). Now, members

of H commute with each other and with members of Z(G), and members of
Z(G) commute with each other and with members of H. Therefore all of the
elements in H ∪ Z(G) commute with one another, so we may write
gh = g11 , . . . , gm
m δ1
h1 , . . . , hδnn = hδ11 , . . . , hδnn g11 , . . . , gm
m
= hg.
Since g, h ∈ hH, Z(G)i were arbitrary, this shows that hH, Z(G)i is abelian.
To show that hH, CG (H)i is not necessarily abelian, consider the dihedral
group D8 with H = {1, r2 }. Since 1 and r2 are in Z(D8 ), we have CG (H) = D8 .
Therefore hH, CG (H)i = D8 is not abelian, even though H is abelian.
2.4. SUBGROUPS GENERATED BY SUBSETS OF A GROUP 103
2.4.4 Exercise 4
Prove that if H is a subgroup of G then H is generated by the set H − {1}.
Proof. If H = {1} then H − {1} is the empty set which indeed generates the
trivial subgroup H. So suppose |H| > 1 and pick a nonidentity element h ∈ H.
Since 1 = hh−1 ∈ hH − {1}i (Proposition 9), we see that H ≤ hH − {1}i. By
minimality of hH − {1}i, the reverse inclusion also holds so that hH − {1}i =
H.
2.4.5 Exercise 5
Prove that the subgroup generated by any two distinct elements of order 2 in
S3 is all of S3 .
Proof. There are three elements of order 2 in S3 , namely (1 2), (1 3), and (2 3).
For h(1 2), (1 3)i we have
(2 3) = (1 3)(1 2)(1 3), (1 2 3) = (1 3)(1 2), and (1 3 2) = (1 2)(1 3),
so h(1 2), (1 3)i = S3 . By symmetry, we also have
h(1 2), (2 3)i = h(1 3), (2 3)i = S3 .
Therefore the desired result holds.
2.4.6 Exercise 6
Prove that the subgroup of S4 generated by (1 2) and (1 2)(3 4) is a noncyclic
group of order 4.
Proof. Let a = (1 2) and b = (1 2)(3 4). Note that a2 = b2 = 1, and ab = ba =

(3 4). We see that the set A = {1, a, b, (3 4)} is closed under composition and
inverses and hence is a subgroup of order 4. Therefore ha, bi = A and we see
that A is noncyclic since in particular it has two distinct elements with order
2.
2.4.7 Exercise 7
Prove that the subgroup of S4 generated by (1 2) and (1 3)(2 4) is isomorphic to
the dihedral group of order 8.
Proof. Let a = (1 2), b = (1 3)(2 4) and c = (1 3 2 4). It is easy to check that

ab = c so c ∈ ha, bi.
Clearly a2 = c4 = 1. Since b is a product of disjoint 2-cycles, it follows that
−1
b = b so that
ca = aba = a(ab)−1 = ac−1 .
Since a and c satisfy all the same relations as do s and r in D8 , it follows that
there is a homomorphism ϕ : D8 → ha, bi defined by
ϕ(si rj ) = ai (ab)j , where i ∈ {0, 1} and j ∈ {0, 1, 2, 3}.

Since c, c2 , and c3 are all distinct, it follows that ϕ is injective. And ϕ is

surjective since every finite product of powers of a and b can be reduced to the
form ai (ab)j in the same way that elements of D8 can be expressed in the form
si rj . ϕ is bijective, so it is an isomorphism and we have D8 ∼
= ha, bi.
2.4.8 Exercise 8
Prove that S4 = h(1 2 3 4), (1 2 4 3)i.
Proof. Let A = h(1 2 3 4), (1 2 4 3)i. By inspection, we find that
(1 4 2) = (1 2 4 3)(1 2 3 4).
Therefore A contains an element of order 3 as well as elements of order 4. Thus
3 and 4 both divide |A|. But |A| also divides 24, so the only possibilities for |A|
are 12 and 24.
To eliminate 12 as a possible order, note that
(1 2) = (1 2 3 4)(1 2 4 3)3 (1 2 3 4)
and
(1 3 2 4) = (1 3)(2 4) = (1 2 3 4)(1 2 4 3)(1 2 3 4)(1 2 4 3)2 .
So by the previous exercise, we know that A contains a subgroup isomorphic to
D8 . Therefore 8 divides |A| so we must have A = S4 .
2.4.9 Exercise 9

1 1
Prove that SL2 (F3 ) is the subgroup of GL2 (F3 ) generated by and
0 1
1 0
.
1 1
Proof. Let
1 1 1 0
A= and B = .
0 1 1 1
Note that A, B ∈ SL2 (F3 ). We are told that we may assume that the subgroup
SL2 (F3 ) has order 24, so we can show that hA, Bi = SL2 (F3 ) if we can show
that it has more than 12 elements (since the order of hA, Bi must divide 24).
The matrices I, A, and B, make three elements, so we need to find ten more:

2 1 2 2 1 0
A = , B = ,
0 1 2 1

2 1 2 0
AB = , (AB)2 = ,
1 1 0 2

1 2 1 1
(AB)3 = , BA = ,
2 2 1 2

2 2 2 0
A2 B 2 = , ABA = ,
2 1 1 2

2 1 0 2
BAB = , A2 B = .
0 2 1 1
So |hA, Bi| = 24 and hA, Bi = SL2 (F3 ).
2.4.10 Exercise 10

0 −1 1 1
Prove that the subgroup of SL2 (F3 ) generated by and is
1 0 1 −1
isomorphic to the quaternion group of order 8.
Proof. Let

0 −1 1 1 2 1
A= ,B = , and C = AB = .
1 0 1 −1 1 1
Now note that A2 = B 2 = C 2 = ABC = −I and that −I commutes with A, B,

and C. Therefore there is a surjective homomorphism ϕ : Q8 → hA, Bi given by
ϕ(i) = A and ϕ(j) = B.
This shows that |hA, Bi| ≤ 8. But I, −I, A, B, C are five distinct elements of
hA, Bi, so |hA, Bi| = 8 and ϕ is an isomorphism.
2.4.11 Exercise 11
Show that SL2 (F3 ) and S4 are two nonisomorphic groups of order 24.
Proof. In the previous exercise we saw that SL2 (F3 ) has a subgroup isomorphic
to Q8 . Q8 has six elements of order 4 (Exercise 1.5.1) while S4 also has six
elements of order 4 (Exercise 1.3.4). But earlier we showed that S4 was gen-
erated by two of these elements (Exercise 2.4.8). Therefore S4 cannot contain
a subgroup isomorphic to Q8 , and this is enough to show that SL2 (F3 ) is not
isomorphic to S4 .
2.4.12 Exercise 12
Prove that the subgroup of upper triangular matrices in GL3 (F2 ) is isomorphic
to the dihedral group of order 8.
Proof. Let T denote the upper triangular matrices of GL3 (F2 ). There are 6
entries in a 3 × 3 matrix that are on or above the diagonal. However, none of
the diagonal entries can be zero since such matrices would have a determinant
of zero. Therefore T has only 23 = 8 elements.
Now, let
   
1 1 1 1 1 1
A = 0 1 1 and B = 0 1 0 .
0 0 1 0 0 1
By direct computation we can see that A4 = B 2 = I and that AB = BA−1 .

Thus the mapping ϕ : D8 → T for which ϕ(r) = A and ϕ(s) = B extends to a
homomorphism. Since A generates a cyclic subgroup of T with order 4 and B is
not in this subgroup, it follows that hA, Bi = T which shows that ϕ is actually
surjective. Since |T | = |D8 | it must be injective as well. Hence T ∼
= D8 .
2.4.13 Exercise 13
Prove that the multiplicative group of positive rational numbers is generated by
the set
1
p is a prime .
p
Proof. Call the set A and let r = s/t be a positive rational number, where s
and t are relatively prime integers. s and t can each be factored into a finite
(possibly empty) product of powers of distinct prime factors. But every prime
p is a member of hAi, since 1/p ∈ A and hAi is closed under inverses. Therefore
r is a finite product of members of hAi, so r ∈ hAi by Proposition 9. This shows
that hAi is the multiplicative group Q+ .
2.4.14 Exercise 14
A group H is called finitely generated if there is a finite set A such that H = hAi.
(a) Prove that every finite group is finitely generated.
Proof. If G is a finite group, then simply note that G = hGi.
(b) Prove that Z is finitely generated.
Proof. Z is finitely generated since it is cyclic: it is generated by {1}.
(c) Prove that every finitely generated subgroup of the additive group Q is
cyclic.
Proof. Let H be a finitely generated subgroup of Q, so that H = hAi for

some finite set
a1 a2 an
A= , ,..., ,
b1 b2 bn
where each bi 6= 0 and (ai , bi ) = 1 for each i. Set K = b1 b2 · · · bn . We will
show that H is a subgroup of the cyclic subgroup generated by 1/K.
Now, every element h of H can be written
ai1 ai ai
h = c1 + c2 2 + · · · + ck k , cj ∈ Z.
bi1 bi 2 bik
By multiplying the numerator and denominator of each fraction by the

necessary quantity, we can make each fraction have K in its denominator.
Then the fractions can be added together to get a single fraction of the
form
J 1
h= =J , J ∈ Z.
K K
This shows that h ∈ h1/Ki, so H is a subgroup of a cyclic group and is
hence cyclic.
(d) Prove that Q is not finitely generated.

Proof. Suppose Q is finitely generated. Then it must be cyclic by the

previous part of the exercise. Therefore Q = hp/qi for some relatively
prime integers p and q with q nonzero. Let r be any prime number that
is not a factor of q. Since p/q is a generator for Q we must have
p 1
k = for some k ∈ Z.
q r
But then q = pkr, so that r is a factor of q. This contradiction shows that
Q cannot be finitely generated.
2.4.15 Exercise 15
Exhibit a proper subgroup of Q which is not cyclic.
Solution. By the previous exercise, we know that such a subgroup cannot be
finitely generated. Consider the set A given by

1
A= k = 0, 1, 2, . . . .
2k
Let H = hAi. Then H is a proper subgroup of Q since, for example, 1/3 is not
a member. If H is cyclic, let it be generated by p/q where p, q ∈ Z. Now q must
be a power of 2, say q = 2n . Then 1/2n+1 is not an integer multiple of p/q, but
it is in H. This shows that H is not cyclic.
2.4.16 Exercise 16
A subgroup M of a group G is called a maximal subgroup if M 6= G and the
only subgroups of G which contain M are M and G.
(a) Prove that if H is a proper subgroup of the finite group G then there is a
maximal subgroup of G containing H.
Proof. If H is maximal, then we are done. If H is not maximal, then

there is a subgroup K1 of G such that H < K1 < G. If K1 is maximal,
we are done. But if K1 is not maximal, there is a subgroup K2 with
H < K1 < K2 < G. If K2 is maximal, we are done, and if not, keep
repeating the procedure. Since G is finite, this process must eventually
come to an end, so that Kn is maximal for some positive integer n. Then
Kn is a maximal subgroup containing H.
(b) Show that the subgroup of all rotations in a dihedral group is a maximal
subgroup.
Proof. Fix a positive integer n > 1 and let H ≤ D2n consist of the ro-
tations of D2n . That is, H = hri. Now, this subgroup is proper since it
does not contain s. If H is not maximal, then by the previous proof we
know there is a maximal subset K containing H. Then K must contain a
reflection srk for k ∈ {0, 1, . . . , n − 1}. Then since srk ∈ K and rn−k ∈ K,
it follows by closure that
s = (srk )(rn−k ) ∈ K.
But D2n = hr, si, so this shows that K = D2n , which is a contradiction.
Therefore H must be maximal.
(c) Show that if G = hxi is a cyclic group of order n ≥ 1 then a subgroup H

is maximal if and only if H = hxp i for some prime p dividing n.
Proof. Suppose H is a maximal subgroup of G. Then H is cyclic, and we

may write H = hxk i for some integer k, with k > 1. Let d = (n, k). Since
H is a proper subgroup, we know by Proposition 6 that d > 1. Choose a
prime factor p of d. If k = p = d then k | n as required.
If, however, k is not prime, then consider the subgroup K = hxp i. Since p
is a proper divisor of k, it follows that H < K. But H is maximal, so we
must have K = G. Again by Proposition 6, we must then have (p, n) = 1.
However, p divides d which divides n, so p | n and (p, n) = p > 1, a
contradiction. Therefore k = p and the left-to-right implication holds.
Now, for the converse, suppose H = hxp i for p a prime dividing n. If H
is not maximal then the first part of this exercise shows that there is a
maximal subgroup K containing H. Then K = hxq i. So xp ∈ hxq i, which
implies q | p. But the only divisors of p are 1 and p. If q = 1 then K = G
and K cannot be a proper subgroup, and if q = p then H = K and H
cannot be a proper subgroup of K. This contradiction shows that H is
maximal.
2.4.17 Exercise 17
This is an exercise involving Zorn’s Lemma to prove that every nontrivial finitely
generated group possesses maximal subgroups. Let G be a finitely generated
group, say G = hg1 , g2 , . . . , gn i, and let S be the set of all proper subgroups of
G. Then S is partially ordered by inclusion. Let C be a chain in S.
(a) Prove that the union, H, of all the subgroups in C is a subgroup of G.
Proof. We assume that C is nonempty. Set

[
H= K.
K∈C
Since a subgroup cannot be empty, H is nonempty. Suppose a and b are

any members of H. Then a ∈ K1 and b ∈ K2 for some K1 , K2 ∈ C. Since
C is a chain, we must have either K1 ≤ K2 or K2 ≤ K1 (or both). Without
loss of generality, we may assume K1 ≤ K2 . Then a and b both belong to
K2 . K2 is a subgroup, so ab−1 also belongs to K2 and hence to H. This
shows that H is a subgroup of G.
(b) Prove that H is a proper subgroup.
Proof. Assume the contrary, so that in particular H contains each gen-

erator g1 , . . . , gn of G. Then there are subgroups K1 , K2 , . . . , Kn (not
necessarily distinct) such that gi ∈ Ki for each i with 1 ≤ i ≤ n. Since C

is a chain, we can order the Ki so that
K1 ≤ K2 ≤ K3 ≤ · · · ≤ Kn .
Then every generator of G belongs to the subgroup Kn . This shows that
every element of G must belong to Kn , so that Kn is not proper. But
every member of C is a proper subgroup, so this gives a contradiction.
(c) Use Zorn’s Lemma to show that S has a maximal element (which is, by
definition, a maximal subgroup).
Proof. Since G is nontrivial, the set S is nonempty because the trivial

subgroup is a proper subgroup. The set H constructed above is also a
proper subgroup and so belongs to S. And each Ki in C is a subgroup of
H, so H is an upper bound for C. Since C was chosen arbitrarily, we have
shown that every chain in the nonempty partially ordered set S has an
upper bound. By Zorn’s Lemma, S must have a maximal element.
2.4.19 Exercise 19
A nontrivial abelian group A (written multiplicatively) is called divisible if for
each element a ∈ A and each nonzero integer k there is an element x ∈ A such
that xk = a, i.e., each element has a k th root in A (in additive notation, each
element is the k th multiple of some element of A).
(a) Prove that the additive group of rational numbers, Q, is divisible.
Proof. Let r ∈ Q and k ∈ Z be arbitrary, with k nonzero. Then x = r/k

is such that kx = r. Hence Q is divisible.
(b) Prove that no finite abelian group is divisible.
Proof. Let G be any nontrivial finite abelian group, and suppose |G| = n.
Since G is nontrivial, we may choose a nonidentity element x ∈ G. Then
there is no element y ∈ G such that y n = x, for the simple reason that
we must have y n = 1 (since |y| has to divide n). Therefore G is not
divisible.
2.4.20 Exercise 20
Prove that if A and B are nontrivial abelian groups, then A × B is divisible if
and only if both A and B are divisible groups.
Proof. Suppose A × B is divisible. Let a ∈ A and b ∈ B be arbitrary, and let k
be any nonzero integer. Then there is (c, d) ∈ A × B such that (c, d)k = (a, b).
But (c, d)k = (ck , dk ), so ck = a and dk = b, which shows that A and B are
both divisible.
Conversely, let A and B be divisible, let (a, b) ∈ A × B be arbitrary, and
let k be any nonzero integer. Since A is divisible, there is c ∈ A with ck = a,
and since B is divisible there is d ∈ B with dk = b. Then (c, d)k = (a, b), which
shows that A × B is divisible.
2.5 The Lattice of Subgroups of a Group

2.5.1 Exercise 1
Let H and K be subgroups of G. Exhibit all possible sublattices which show
only G, 1, H, K and their joins and intersections. What distinguishes the
different drawings?
Solution. In general, when H and K are distinct, with a nontrivial intersection
and a join that is a proper subgroup, the sublattice might look something like
the following:
hH, Ki
H K
H ∩K
In other cases, we may have H = K, or H ≥ K:
G G
H
H=K
K
1 1
There are several other possibilities, for example one of H or K could be trivial,
or we could have hH, Ki = G, and various other options. The drawings are
distinguished by the relationships between the various subgroups.
2.5.2 Exercise 2
In each of (a) to (d) list all subgroups of D16 that satisfy the given condition.
(a) Subgroups that are contained in hsr2 , r4 i
Solution. From the lattice given in the text, we see that the subgroups
contained in hsr2 , r4 i are hsr2 , r4 i, hsr6 i, hsr2 i, hr4 i, and 1.
(b) Subgroups that are contained in hsr7 , r4 i

2.5. THE LATTICE OF SUBGROUPS OF A GROUP 111
Solution. hsr7 , r4 i = hsr3 , r4 i, so the subgroups contained in this sub-

group are hsr3 , r4 i, hr4 i, hsr3 i, hsr7 i, and 1.
(c) Subgroups that contain hr4 i
Solution. The subgroups containing hr4 i are hr4 i, hsr2 , r4 i, hs, r4 i, hr2 i,
hsr3 , r4 i, hsr5 , r4 i, hs, r2 i, hri, hsr, r2 i, and D16 itself.
(d) Subgroups that contain hsi
Solution. The subgroups containing hsi are hsi, hs, r4 i, hs, r2 i, and D16 .
2.5.3 Exercise 3
Show that the subgroup hs, r2 i of D8 is isomorphic to V4 .
Proof. The subgroup hs, r2 i consists of the elements {1, s, r2 , sr2 }, and V4 =
{1, a, b, c}. Note that both groups are abelian and of order 4.
Define the mapping ϕ : hs, r2 i → V4 by
ϕ(1) = 1, ϕ(s) = a, ϕ(r2 ) = b, and ϕ(sr2 ) = c.
We now directly verify that ϕ is a homomorphism:
ϕ(s2 ) = ϕ(1) = 1 = a2 = ϕ(s)2 ,

ϕ(sr2 ) = c = ab = ϕ(s)ϕ(r2 ),
ϕ(ssr2 ) = ϕ(r2 ) = b = ac = ϕ(s)ϕ(sr2 ),
ϕ(r4 ) = ϕ(1) = 1 = b2 = ϕ(r2 )2 ,
ϕ(r2 sr2 ) = ϕ(s) = a = bc = ϕ(r2 )ϕ(sr2 ),
ϕ((sr2 )2 ) = ϕ(1) = 1 = c2 = ϕ(sr2 )2 .
Since both groups are abelian, this is enough to show that ϕ is a homomorphism.
But ϕ is clearly also a bijection, so ϕ is an isomorphism and hs, r2 i ∼
= V4 .
2.5.4 Exercise 4
Use the given lattice to find all pairs of elements that generate D8 (there are 12
pairs).
Solution. First, we know that D8 = hs, ri. Now, looking at the cyclic subgroups
in the lattice, we see that the only subgroup containing both hsi and hrsi is D8
itself. Hence hs, rsi = D8 . Similarly, the only subgroup containing hsi and hr3 si
is D8 , so hs, r3 si = D8 . Continuing in this way, we can find all the pairs that
generate D8 (noting that hri = hr3 i):
hs, ri, hs, r3 i, hs, rsi, hs, r3 si, hr2 s, ri, hr2 s, r3 i,
hr2 s, rsi, hr2 s, r3 si, hr, rsi, hr3 , rsi, hr3 , r3 si, hr, r3 si.
No other pairing can generate all of D8 .

2.5.5 Exercise 5
Use the given lattice to find all elements x ∈ D16 such that D16 = hx, si (there
are 8 such elements x).
Solution. Note that hri = hr3 i = hr5 i = hr7 i. We now proceed as in the
previous problem, pairing hsi with other cyclic subgroups such that all of D16 is
the smallest group containing both subgroups. We find the following generating
pairs:
hs, ri, hs, r3 i, hs, r5 i, hs, r7 i, hs, sr3 i, hs, sr7 i, hs, sr5 i, hs, sri.
2.5.6 Exercise 6
Use the given lattices to help find the centralizers of every element in the fol-
lowing groups:
(a) D8
Solution. Since s commutes with r2 , we see from the lattice that CD8 (s) =
hs, r2 i (this centralizer cannot be all of D8 since s does not commute
with r). r2 commutes with everything (it is in the center of D8 ), so
CD8 (r2 ) = D8 . By similar reasoning, we find the following centralizers:
CD8 (1) = D8 ,
CD8 (r) = hri,
CD8 (r2 ) = D8 ,
CD8 (r3 ) = hri,
CD8 (s) = hs, r2 i,
CD8 (rs) = hrs, r2 i,
CD8 (r2 s) = hs, r2 i,
CD8 (r3 s) = hrs, r2 i.
(b) Q8
Solution. We know that −1 commutes with every element, but i, j, and

k do not commute with each other. Therefore
CQ8 (1) = Q8 ,
CQ8 (−1) = Q8 ,
CQ8 (i) = CQ8 (−i) = hii,
CQ8 (j) = CQ8 (−j) = hji,
CQ8 (k) = CQ8 (−k) = hki.
(c) S3
Solution. From the lattice we see that every nontrivial subgroup is max-
imal, so the centralizer of each cycle is either the subgroup generated by
that cycle, or else all of S3 . But none of (1 2), (1 3), (2 3), and (1 2 3) com-
mute with each other, so none of the centralizers can be all of S3 , aside
from CS3 (1). This gives
CS3 (1) = S3 ,
CS3 (1 2) = h(1 2)i,
CS3 (1 3) = h(1 3)i,
CS3 (2 3) = h(2 3)i,
CS3 (1 2 3) = CS3 (1 3 2) = h(1 2 3)i.
(d) D16
Solution. We use similar reasoning as we did for D8 .

CD16 (1) = D16 ,
CD16 (r) = CD16 (r ) = CD16 (r3 ) = hri,
2
CD16 (r5 ) = CD16 (r6 ) = CD16 (r7 ) = hri,

CD16 (r4 ) = D16 ,
CD16 (s) = CD16 (sr4 ) = hs, r4 i,
CD16 (sr) = CD16 (sr5 ) = hsr5 , r4 i,
CD16 (sr2 ) = CD16 (sr6 ) = hsr2 , r4 i,
CD16 (sr3 ) = CD16 (sr7 ) = hsr3 , r4 i.
2.5.7 Exercise 7
Find the center of D16 .
Solution. We already found in Exercise 2.2.7 that Z(D2n ) = 1 if n is odd and
Z(D2n ) = {1, rk } if n = 2k. Therefore Z(D16 ) = {1, r4 }. Alternatively, we
could use the results from the previous problem, where we saw that 1 and r4
were the only elements with centralizers equal to all of D16 .
2.5.8 Exercise 8
In each of the following groups find the normalizer of each subgroup:
(a) S3
Solution. S3 has six subgroups. From the lattice, we see that each nontriv-
ial proper subgroup H of S3 is maximal, so we either have NS3 (H) = H
or NS3 (H) = S3 .
For H = h(1 2)i, since
(1 3)(1 2)(1 3) = (2 3) 6∈ h(1 2)i,
we see that (1 3)H(1 3)−1 6= H, so NS3 (H) = H. The same is true for the
other subgroups generated by 2-cycles. So,
NS3 (h(1 2)i) = h(1 2)i, NS3 (h(1 3)i) = h(1 3)i,
and NS3 (h(2 3)i) = h(2 3)i.
For H = h(1 2 3)i = {1, (1 2 3), (1 3 2)}, we have
(1 2)(1 2 3)(1 2) = (1 3 2) ∈ H,
(1 2)(1 3 2)(1 2) = (1 2 3) ∈ H,
so (1 2) is in the normalizer of H. Therefore
NS3 (h1 2 3i) = NS3 (1) = NS3 (S3 ) = S3 .
(b) Q8
Solution. Since 1 and −1 are in the center of Q8 , we have
NQ8 (1) = NQ8 (−1) = NQ8 (Q8 ) = Q8 .
The other subgroups are maximal, so each normalizer is either the sub-
group itself or else all of Q8 . Since
j(i)j −1 = (−k)(−j) = kj = −i ∈ hii,
and
j(−i)j −1 = k(−j) = −kj = i ∈ hii,
we see that j ∈ NQ8 (hii). Therefore NQ8 (hii) = Q8 . By an entirely similar

argument, we see that the same is true for hji and hki. So
NQ8 (hii) = NQ8 (hji) = NQ8 (hki) = Q8 .
2.5.9 Exercise 9
Draw the lattices of subgroups of the following groups:
(a) Z/16Z
Solution.
Z/16Z
h2i
h4i
h8i
h16i = h0i
(b) Z/24Z
Solution.
Z/24Z
h2i
h3i
h4i
h6i
h8i
h12i
h24i = h0i
(c) Z/48Z
Solution.
Z/48Z
h2i
h3i
h4i
h6i
h8i
h12i
h16i
h24i
h48i = h0i
2.5.10 Exercise 10
= Z4 or G ∼
Classify groups of order 4 by proving that if |G| = 4 then G ∼ = V4 .
Proof. Let G = {1, a, b, c}. If G is cyclic, then certainly G ∼
= Z4 since all cyclic
groups of the same order are isomorphic. So assume that G is not cyclic, so
that no element has order 4. Since the order of each element must divide the
order of the group, it follows that a, b, c each have order 2.
Now consider the product ab. If ab = 1, then multiplying by a on the left
gives b = a, so a and b are not distinct, which is a contradiction. If ab = a,
then multiplying by a gives b = 1, another contradiction. For the same reason
we cannot have ab = b. So the only possibility is ab = c.
Using exactly the same argument, we can see that ba = c, ac = ca = b,

and bc = cb = a. Since G has the same multiplication table as V4 , it follows
that G ∼
= V4 . Indeed, any identity-preserving bijection between them is an
isomorphism.
2.5.11 Exercise 11
Consider the group of order 16 with the following presentation:
QD16 = hσ, τ | σ 8 = τ 2 = 1, στ = τ σ 3 i
(called the quasidihedral or semidihedral group of order 16). This group has
three subgroups of order 8: hτ, σ 2 i ∼
= D8 , hσi ∼= Z8 and hσ 2 , στ i ∼
= Q8 and
every proper subgroup is contained in one of these three subgroups. Fill in the
missing subgroups in the provided lattice of all subgroups of the quasidihedral
group, exhibiting each subgroup with at most two generators.
Solution. Certainly hσ 2 i is between hσi and hσ 4 i. By taking powers of the other
elements, we see that the other missing cyclic subgroups are
hτ σi = {1, τ σ, σ 4 , τ σ 5 },
hτ σ 3 i = {1, τ σ 3 , σ 4 , τ σ 7 },
hτ σ 4 i = {1, τ σ 4 }.
and
hτ σ 6 i = {1, τ σ 6 }.
hσ 4 , τ i must contain hτ σ 4 i, and we see that hτ σ 6 i must be the sibling of hτ σ 2 i,
whose containing subgroup would then be hσ 4 , τ σ 2 i. The remaining cyclic sub-
groups are contained in hσ 2 , τ σi. This gives the following lattice.
QD16
hσ 2 , τ i hσi hσ 2 , τ σi
hσ 4 , τ σ 2 i hσ 4 , τ i hσ 2 i hτ σi hτ σ 3 i
hτ σ 2 i hτ σ 6 i hτ σ 4 i hτ i hσ 4 i
1
2.5.12 Exercise 12
The group
A = Z2 × Z4 = ha, b | a2 = b4 = 1, ab = bai
has order 8 and has three subgroups of order 4: ha, b2 i ∼ = V4 , hbi ∼
= Z4 and
∼
habi = Z4 and every proper subgroup is contained in one of these three. Draw
the lattice of all subgroups of A giving each subgroup in terms of at most two
generators.
Solution. Writing out the elements of A (in terms of a and b) gives
A = {1, a, b, b2 , b3 , ab, ab2 , ab3 }.
There are three elements with order 2, namely a, b2 , and ab2 . Since ha, b2 i ∼
= V4 ,
we see that hai, hb2 i, and hab2 i must be directly contained in ha, b2 i. From this
and the other given information, we form the following lattice.
A
ha, b2 i habi hbi
hai hab2 i hb2 i
2.5.13 Exercise 13
The group
G = Z2 × Z8 = hx, y | x2 = y 8 = 1, xy = yxi
has order 16 and has three subgroups of order 8: hx, y 2 i ∼
= Z2 × Z4 , hyi ∼= Z8
∼
and hxyi = Z8 and every proper subgroup is contained in one of these three.
Draw the lattice of all subgroups of G, giving each subgroup in terms of at most
two generators.
Solution. Since hx, y 2 i ∼
= Z2 × Z4 , we see that the lattice of G contains the
lattice from the previous exercise within its structure, with a replaced by x and
b replaced by y 2 . Adding in the maximal subgroups hyi and hxyi produces the
following lattice.
hx, y 2 i hxyi hyi
hx, y 4 i hxy 2 i hy 2 i
hxi hxy 4 i hy 4 i
2.5.14 Exercise 14
Let M be the group of order 16 with the following presentation:
hu, v | u2 = v 8 = 1, vu = uv 5 i
(sometimes called the modular group of order 16). It has three subgroups of
order 8: hu, v 2 i, hvi and huvi and every proper subgroup is contained in one of
these three. Prove that hu, v 2 i ∼
= Z2 × Z4 , hvi ∼
= Z8 and huvi ∼
= Z8 . Show that
the lattice of subgroups of M is the same as the lattice of subgroups of Z2 × Z8
but that these two groups are not isomorphic.
Solution. We will use the presentation for Z2 ×Z4 given in Exercise 2.5.12. Since
u2 = (v 2 )4 = 1 and u(v 2 ) = (v 2 )u, it follows that the mapping ϕ : Z2 × Z4 →
hu, v 2 i defined by
ϕ(a) = u and ϕ(b) = v 2
extends to a homomorphism. ϕ is surjective by construction and hence bijective

since we know that both groups Z2 × Z4 and hu, v 2 i have order 8. Therefore
hu, v 2 i ∼
= Z2 × Z4 .
Since we know |hvi| = |huvi| = 8, we automatically know that both sub-
groups are isomorphic to Z8 , since all cyclic groups of the same order are iso-
morphic. We see that the subgroups of M share the same relationships as the
subgroups of Z2 × Z8 , so they have the same lattice, with x replaced by u and
y replaced by v:
hu, v 2 i huvi hvi
hu, v 4 i huv 2 i hv 2 i
hui huv 4 i hv 4 i
Finally, we note that, despite having the same lattice, M is not isomorphic
to Z2 × Z8 since the latter is abelian and M is not (uv 6= vu).
2.5.15 Exercise 15
Describe the isomorphism type of each of the three subgroups of D16 of order
8.
Solution. Since |r| = 8, we see that hri ∼

= Z8 .
Next, consider the subgroup H = hs, r2 i and observe that
(r2 )4 = s2 = 1 and (r2 )s = sr6 = s(r2 )−1 .
Since s and r2 in D16 satisfy the same relations as s and r do in D8 , the map
ϕ : D8 → H given by
ϕ(r) = r2 and ϕ(s) = s
extends to a surjective homomorphism. And it is easy to see that H consists
of only eight elements, namely elements of the form si r2j where i ∈ {0, 1} and
j ∈ {0, 1, 2, 3}. Therefore ϕ must be a bijection, and we have hs, r2 i ∼
= D8 .
Now consider the subgroup K = hsr, r2 i. We have
(r2 )4 = (sr)2 = 1 and (r2 )(sr) = sr7 = (sr)(r2 )−1 .
So the map ψ : D8 → K given by
ϕ(r) = r2 and ϕ(s) = sr
extends to a surjective homomorphism, which must be injective as well since K

has 8 elements. Therefore hsr, r2 i ∼
= D8 .
2.5.16 Exercise 16
Use the lattice of subgroups of the quasidihedral group of order 16 to show that
every element of order 2 is contained in the proper subgroup hτ, σ 2 i.
Proof. Every element of order 2 generates a cyclic subgroup having order 2.

From the lattice of QD16 (Exercise 2.5.11) we see that every cyclic subgroup of
QD16 is contained in hσ 2 , τ i except for hτ σi and hτ σ 3 i. But neither τ σ nor τ σ 3
has order 2 (they both have order 4), so we see that every element of order 2 is
indeed contained in hσ 2 , τ i.
2.5.17 Exercise 17
Use the lattice of subgroups of the modular group M of order 16 to show that
the set {x ∈ M | x2 = 1} is a subgroup of M isomorphic to the Klein 4-group.
Proof. We know from Exercise 2.5.10 that every group of order 4 is isomorphic
to either Z4 or V4 . From the lattice we constructed in Exercise 2.5.14, we see
that hu, v 4 i contains only the four elements 1, u, v 4 , and uv 4 . Since each of these
(aside from 1) has order 2, we know that hu, v 4 i is not cyclic and so cannot be
isomorphic to Z4 . Therefore hu, v 4 i ∼ = V4 .
2.5.18 Exercise 18
Use the lattice to help find the centralizer of every element of QD16 .
Solution. Since σ 4 τ = τ σ 12 = τ σ 4 , we see that σ 4 commutes with τ and hence

belongs to the center of the group (since it commutes with each generator).
From the lattice, we see that the centralizer of τ is either equal to hσ 4 , τ i,
or to hσ 2 , τ i, or to all of QD16 . But we can rule out the latter two cases, since
τ does not commute with σ 2 :
σ2 τ = τ σ6 .
A similar case can be made for the centralizer of τ σ 4 .
For τ σ 2 and τ σ 6 we see that both commute with σ 4 and τ σ 2 , but neither
commutes with σ 2 , so their centralizer must be hσ 4 , τ σ 2 i.
τ σ and τ σ 3 do not commute with σ 2 , so their respective centralizers are just
the cyclic subgroups that they generate.
Lastly, the powers of σ all commute with each other. Putting all this infor-
mation together, we get
CQD16 (1) = CQD16 (σ 4 ) = QD16 ,

CQD16 (σ) = CQD16 (σ 2 ) = CQD16 (σ 3 ) = hσi,
CQD16 (σ 5 ) = CQD16 (σ 6 ) = CQD16 (σ 7 ) = hσi,
CQD16 (τ ) = CQD16 (τ σ 4 ) = hσ 4 , τ i,
CQD16 (τ σ 2 ) = CQD16 (τ σ 6 ) = hσ 4 , τ σ 2 i,
CQD16 (τ σ) = CQD16 (τ σ 5 ) = hτ σi,
CQD16 (τ σ 3 ) = CQD16 (τ σ 7 ) = hτ σ 3 i.
2.5.19 Exercise 19
Use the lattice to help find ND16 (hs, r4 i).
Solution. Let H = hs, r4 i. From the lattice, we see that the normalizer of H
must be either H itself, or hs, r2 i, or D16 . Since
r2 (r4 )(r2 )−1 = r2 r4 r6 = r4 ∈ H,

r2 s(r2 )−1 = r2 sr6 = sr4 ∈ H,
and
r2 (sr4 )(r2 )−1 = r2 sr2 = s ∈ H,
we see that r2 H(r2 )−1 = H so hs, r2 i ≤ ND16 (H). However, since
rsr7 = sr7 r7 = sr6 6∈ H,
we see that rHr−1 6= H so the normalizer of H cannot be all of D16 . Therefore

ND16 (H) = hs, r2 i.
2.5.20 Exercise 20
Use the lattice of subgroups of QD16 to help find the normalizers
(a) NQD16 (hτ σi)
Solution. Let H = hτ σi. From the lattice (Exercise 2.5.11) we see that
there are only three possibilities for the normalizer of H: it is either H
itself, or hσ 2 , τ σi or else all of QD16 . Since H = {1, τ σ, σ 4 , τ σ 5 }, we can
compute the elements of σ 2 H(σ 2 )−1 as follows:
σ 2 (τ σ)σ 6 = τ σ 6 σ 7 = τ σ 5 ,
σ2 σ4 σ6 = σ4 ,
σ 2 (τ σ 5 )σ 6 = σ 2 τ σ 3 = τ σ.
So σ 2 H(σ 2 )−1 = H and therefore σ 2 is in the normalizer of H. But σ is

not in the normalizer, since
σ(τ σ)σ 7 = στ = τ σ 3 6∈ H.
Thus the only possibility is NQD16 (H) = hσ 2 , τ σi.
(b) NQD16 (hτ, σ 4 i)
Solution. Let K = hτ, σ 4 i. Again there are three possibilities. K =

{1, τ, σ 4 , τ σ 4 }, so as before, we compute the elements of σ 2 K(σ 2 )−1 :
σ 2 τ σ 6 = τ σ 12 = τ σ 4 ,
σ2 σ4 σ6 = σ4 ,
σ 2 (τ σ 4 )σ 6 = σ 2 τ σ 2 = τ.
So σ 2 K(σ 2 )−1 = K and we must have hτ, σ 2 i ≤ NQD16 (K). And since
στ σ 7 = τ σ 2 6∈ K,
we see that the normalizer cannot be all of QD16 . Therefore NQD16 (K) =
hτ, σ 2 i.
Chapter 3
Quotient Groups and

Homomorphisms
3.1 Definitions and Examples

Let G and H be groups.
3.1.1 Exercise 1
Let ϕ : G → H be a homomorphism and let E be a subgroup of H. Prove
that ϕ−1 (E) ≤ G (i.e., the preimage or pullback of a subgroup under a homo-
morphism is a subgroup). If E E H prove that ϕ−1 (E) E G. Deduce that
ker ϕ E G.
Proof. Note that ϕ(1) = 1 ∈ E so ϕ−1 (E) is nonempty. Suppose a, b ∈ ϕ−1 (E),
so that ϕ(a) = x and ϕ(b) = y for some x, y ∈ E. Then, since ϕ is a homomor-
phism, we have
ϕ(ab−1 ) = ϕ(a)ϕ(b)−1 = xy −1 ∈ E,
which shows that ab−1 ∈ ϕ−1 (E). By the subgroup criterion, this shows that
ϕ−1 (E) ≤ G.
Now suppose that E is a normal subgroup of H. Let g ∈ G and n ∈ ϕ−1 (E).
Then ϕ(g) = h for some h ∈ H and ϕ(n) = x for some x ∈ E. We have
ϕ(gng −1 ) = ϕ(g)ϕ(n)ϕ(g)−1
= hxh−1 .
But hxh−1 ∈ E since E E H, so gng −1 ∈ ϕ−1 (E). The choice of g and n were
arbitrary, so this shows that ϕ−1 (E) E G.
Lastly, if we let E be the trivial subgroup of H, then the above shows that
ker ϕ = ϕ−1 (E) E G since the trivial subgroup is always normal.
3.1.2 Exercise 2
Let ϕ : G → H be a homomorphism of groups with kernel K and let a, b ∈
ϕ(G). Let X ∈ G/K be the fiber above a and let Y be the fiber above b, i.e.,
123
124 CHAPTER 3. QUOTIENT GROUPS AND HOMOMORPHISMS
X = ϕ−1 (a), Y = ϕ−1 (b). Fix an element u of X (so ϕ(u) = a). Prove that if
XY = Z in the quotient group G/K and w is any member of Z, then there is
some v ∈ Y such that uv = w.
Proof. Let v = u−1 w. We want to show that v ∈ Y , or ϕ(v) = b. Since ϕ is a

homomorphism and since Z = ϕ−1 (ab), we have
ϕ(v) = ϕ(u−1 w)
= ϕ(u)−1 ϕ(w)
= a−1 (ab)
= (a−1 a)b
= b.
So v ∈ Y as required.
3.1.3 Exercise 3
Let A be an abelian group and let B be a subgroup of A. Prove that A/B is
abelian. Give an example of a non-abelian group G containing a proper normal
subgroup N such that G/N is abelian.
Solution. Let a1 B, a2 B ∈ A/B, where a1 , a2 ∈ A. Since A is abelian, we have
a1 Ba2 B = (a1 a2 )B = (a2 a1 )B = a2 Ba1 B.
Therefore A/B is abelian.

For the second part of the problem, let G be the non-abelian dihedral group
D8 and let N the proper normal subgroup hr2 i. In the text it was shown that
G/N ∼ = V4 , the Klein four-group, which is abelian. Therefore G/N is abelian
even though G is not.
3.1.4 Exercise 4
Prove that in the quotient group G/N , (gN )α = g α N for all α ∈ Z.
Proof. First, if α = 0, then (gN )0 = 1N = g 0 N , so the statement is true in this

case. We also know by Proposition 5 that (gN )−1 = g −1 N . So it will suffice to
prove the statement only for positive α, which we will do by induction on α.
The case where α = 1 is trivial. For the inductive step, suppose that the
statement (gN )k = g k N holds for some particular k ≥ 1. Then
(gN )k+1 = (gN )k gN

= g k N gN
= (g k g)N
= g k+1 N.
By induction, we conclude that (gN )α = g α N for all α ≥ 1, so the proof is

complete.
3.1. DEFINITIONS AND EXAMPLES 125
3.1.5 Exercise 5
Use the preceding exercise to prove that the order of the element gN in G/N
is n, where n is the smallest positive integer such that g n ∈ N (and gN has
infinite order if no such positive integer exists). Give an example to show that
the order of gN in G/N may be strictly smaller than the order of g in G.
Solution. Fix an element gN in G/N . First, if possible, let n be the smallest
positive integer such that g n ∈ N . Then g n N = 1N . So, by the previous
exercise, we know that (gN )n = 1N . This shows that |gN | ≤ n. On the other
hand, if m is any positive integer with (gN )m = 1N then, using the previous
exercise again, g m N = 1N so that g m ∈ N . Since n is the smallest positive
integer with g n ∈ N , this shows that |gN | ≥ n, which completes the proof for
the case of finite order.
Next, suppose that there is no such n. Then for each positive integer k,
g k 6∈ N . If gN were to have finite order, say (gN )m = 1N , then the previous
exercise would show that g m ∈ N , giving a contradiction. This shows that gN
has infinite order, which completes the proof.
Lastly, for the example, consider G = Z4 , the cyclic group of order 4. Let
x be a generator for G and take N = hx2 i = {1, x2 }. Now the element x2 has
order 2 in G, but x2 N = 1N has order 1 in G/N .
3.1.6 Exercise 6
Define ϕ : R× → {±1} by letting ϕ(x) be x divided by the absolute value of x.
Describe the fibers of ϕ and prove that ϕ is a homomorphism.
Solution. The fiber above 1 is the positive reals, and the fiber above −1 is the
negative reals.
Let x, y ∈ R× be arbitrary. Then
xy x y
ϕ(xy) = = · = ϕ(x)ϕ(y),
|xy| |x| |y|
3.1.7 Exercise 7
Define π : R2 → R by π((x, y)) = x + y. Prove that π is a surjective homomor-
phism and describe the kernel and fibers of π geometrically.
Solution. For any (x1 , y1 ), (x2 , y2 ) ∈ R2 , we have
π((x1 , y1 ) + (x2 , y2 )) = π((x1 + x2 , y1 + y2 ))
= (x1 + x2 ) + (y1 + y2 )
= (x1 + y1 ) + (x2 + y2 )
= π((x1 , y1 )) + π((x2 , y2 )),
so π is a homomorphism. And for any x ∈ R, we have π((x, 0)) = x + 0 = x, so
π is also surjective.
ker π is simply the diagonal line whose equation is x + y = 0. And for a ∈ R,
the fiber over a is the line with equation x + y = a, which is just a translate of
the kernel.
3.1.8 Exercise 8
Let ϕ : R× → R× be the map sending x to the absolute value of x. Prove that ϕ
is a homomorphism and find the image of ϕ. Describe the kernel and the fibers
of ϕ.
Solution. For any x, y ∈ R× we have
ϕ(xy) = |xy| = |x||y| = ϕ(x)ϕ(y)
and ϕ is a homomorphism. Its image is R+ , the positive reals.

The kernel of ϕ is the set {−1, 1}, since ϕ(±1) = 1 and no other real number
has an absolute value of 1. Likewise, the fiber over the real number a is {−a, a}.
3.1.9 Exercise 9
Define ϕ : C× → R× by ϕ(a + bi) = a2 + b2 . Prove that ϕ is a homomorphism
and find the image of ϕ. Describe the kernel and the fibers of ϕ geometrically
(as subsets of the plane).
Solution. Let a + bi and c + di be any members of C× . Then
ϕ((a + bi)(c + di)) = ϕ((ac − bd) + (ad + bc)i)

= (ac − bd)2 + (ad + bc)2
= a2 c2 − 2abcd + b2 d2 + a2 d2 + 2abcd + b2 c2
= a2 (c2 + d2 ) + b2 (c2 + d2 )
= (a2 + b2 )(c2 + d2 )
= ϕ(a + bi)ϕ(c + di),
and ϕ is a homomorphism. Let a + bi ∈ C× . Since a and b cannot

√ both be zero,
a2 + b2 > 0. So im ϕ ⊆ R+ . But for any a ∈ R+ , we have ϕ( a + 0i) = a and
we see that im ϕ = R+ .
The kernel of ϕ is the set
ker ϕ = {a + bi ∈ C× | a2 + b2 = 1}.
This is a circle of radius 1 centered at the origin

√ in the complex plane. For
a ∈ R+ , the fiber over a is the circle of radius a.
3.1.10 Exercise 10
Let ϕ : Z/8Z → Z/4Z by ϕ(ā) = ā. Show that this is a well defined, surjective
homomorphism and describe its fibers and kernel explicitly (showing that ϕ is
well defined involves the fact that ā has a different meaning in the domain and
range of ϕ).
Solution. To show that ϕ is well defined we need to show that any choice of
representative for a particular congruence class in Z/8Z will produce the same
congruence class in Z/4Z under ϕ. Suppose then that ā = b̄, where ā, b̄ ∈ Z/8Z.
Then for some integer k, a = b + 8k and we have
ϕ(ā) = ϕ(b + 8k)
= b + 8k
= b + 4(2k)
= b̄
= ϕ(b̄)
and ϕ is well defined. It is also a homomorphism since
ϕ(ā + b̄) = ϕ(a + b)
=a+b
= ā + b̄
= ϕ(ā) + ϕ(b̄).
And it is clearly surjective.
The fibers of ϕ are
ϕ−1 (0̄) = {0̄, 4̄},
ϕ−1 (1̄) = {1̄, 5̄},
ϕ−1 (2̄) = {2̄, 6̄},
ϕ−1 (3̄) = {3̄, 7̄},
and ker ϕ = ϕ−1 (0̄) = {0̄, 4̄}.
3.1.11 Exercise 11
Let F be a field and let

a b
a, b, c ∈ F, ac 6= 0 ≤ GL2 (F ).
G=
0 c
(a) Prove that the map

a b
ϕ: 7→ a
0 c
is a surjective homomorphism from G onto F × . Describe the fibers and
kernel of ϕ.
Solution. First, for any a ∈ F × , we have

a 1
ϕ = a,
0 1
so ϕ is surjective. And for any a, b, c, d, e, f ∈ F with ac 6= 0 and df 6= 0,
we have

a b d e ad ae + bf
ϕ =ϕ
0 c 0 f 0 cf
= ad

a b d e
=ϕ ϕ ,
0 c 0 f
For each a ∈ F × , the fiber over a is given by

−1 a x
ϕ (a) = x, y ∈ F, y 6= 0 ,
0 y
with the kernel being the fiber above 1.
(b) Prove that the map

a b
ψ: 7→ (a, c)
0 c
is a surjective homomorphism from G onto F × × F × . Describe the fibers

and kernel of ψ.
Solution. The proof is entirely similar to the proof for ϕ in the previous
part of the problem and is omitted here. The fiber over (a, c) is

−1 a x
ψ ((a, c)) = x∈F ,
0 c
with ker ψ = ψ −1 ((1, 1)).
(c) Let

1 b
b∈F .
H=
0 1
Prove that H is isomorphic to the additive group F .
Proof. Define ρ : H → F by

1 b
ρ = b.
0 1
Then ρ is a bijection since it has the obvious two-sided inverse ρ−1 : F → H

given by

−1 1 b
ρ (b) = ,
0 1
and it is a homomorphism since

1 b1 1 b2 1 b1 + b2
ρ =ρ
0 1 0 1 0 1
= b1 + b2

1 b1 1 b2
=ρ +ρ .
0 1 0 1
Therefore ρ is an isomorphism and H ∼

= F.
3.1.12 Exercise 12
Let G be the additive group of real numbers, let H be the multiplicative group
of complex numbers of absolute value 1 (the unit circle S 1 in the complex plane)
and let ϕ : G → H be the homomorphism ϕ : r 7→ e2πir . Draw the points on
a real line which lie in the kernel of ϕ. Describe similarly the elements in the
fibers of ϕ above the points −1, i, and e4πi/3 of H.
Solution. The kernel of ϕ is simply Z, since e2πir = 1 if and only if r is an
integer:
−5 −4 −3 −2 −1 0 1 2 3 4 5
The specified fibers are

−1 1 1
k∈Z ,
ϕ (−1) = + Z = k +
2 2

1 1
ϕ−1 (i) = + Z = k +

k∈Z ,
4 4

2 2
ϕ−1 (e4πi/3 ) = + Z = k +

k∈Z .
3 3
3.1.13 Exercise 13
Repeat the preceding exercise with the map ϕ replaced by the map ϕ : r 7→ e4πir .
Solution. In this case the kernel is 21 Z:
−5 −4 −3 −2 −1 0 1 2 3 4 5
The fibers are

−1 1 1 2k + 1
k∈Z ,
ϕ (−1) = + Z =
4 2 4

1 1 4k +1
ϕ−1 (i) = + Z =

k∈Z ,
8 2 8

1 1 3k +2
ϕ−1 (e4πi/3 ) = + Z =

k∈Z .
3 2 6
3.1.14 Exercise 14
Consider the additive quotient group Q/Z.
(a) Show that every coset of Z in Q contains exactly one representative q ∈ Q
in the range 0 ≤ q < 1.
Proof. Let t ∈ Q be arbitrary. Write t = m/n in lowest terms. We may

use the division algorithm to find unique integers q and r with 0 ≤ r < n
such that m = nq + r. Then t = q + r/n, where 0 ≤ r/n < 1. Then we
have
nr o r
t + Z = {t + k | k ∈ Z} = + (k + q) k ∈ Z = + Z.

n n
From the uniqueness of q and r, it follows that r/n is the only represen-
tative of t + Z that is in the range [0, 1).
(b) Show that every element of Q/Z has finite order but that there are ele-
ments of arbitrarily large order.
Proof. Again, let t ∈ Q be arbitrary, with t = m/n for integers m and n,

with n > 0. Then
n(t + Z) = nt + Z = m + Z = 0 + Z,
so t + Z has finite order (note that the first equality follows from Exer-
cise 3.1.4).
Given any positive integer k, the coset 1/k + Z has order k. Since k can be
made arbitrarily large, we see that Q/Z contains elements of arbitrarily
large order.
(c) Show that Q/Z is the torsion subgroup of R/Z.
Proof. We need to show that the only elements in R/Z having finite order
belong to Q/Z. So suppose r is an irrational representative of a coset
having finite order n. Then
nr + Z = n(r + Z) = 0 + Z.
This implies that nr ∈ Z, say m = nr. Then r = m/n and r ∈ Q,

contradicting our choice of r. Therefore r cannot have finite order. So
Q/Z is indeed the torsion subgroup of R/Z.
(d) Prove that Q/Z is isomorphic to the multiplicative group of roots of unity
in C× .
Proof. Let S 1 be the unit circle in the complex plane and let G be the
multiplicative group of the roots of unity. From Exercise 3.1.12 we know
that R/Z ∼= S 1 , as we can exhibit the explicit isomorphism ϕ by
ϕ(r + Z) = e2πir .
Now consider the torsion subgroup of S 1 . By definition, z ∈ S 1 has finite

order, say n, if and only if z n = 1, i.e. if and only if z ∈ G. So we see that
G is actually the torsion subgroup of S 1 .
Since R/Z is isomorphic to S 1 , it follows that their torsion subgroups are
also isomorphic. Hence Q/Z ∼= G.
3.1.15 Exercise 15
Prove that a quotient of a divisible abelian group by any proper subgroup is
also divisible. Deduce that Q/Z is divisible.
Proof. Let G be a divisible abelian group and let H < G be a proper subgroup.
Let g ∈ G be arbitrary and let k 6= 0 be an integer. Since G is divisible, there
is an element x ∈ G such that xk = g. Then, by Exercise 3.1.4, we know that
(xH)k = xk H = gH,
which shows that G/H is also divisible.
We have shown in Exercise 2.4.19 that Q is divisible. Since Z is a proper
subgroup, we must have that Q/Z is divisible as well.
3.1.16 Exercise 16
Let G be a group, let N be a normal subgroup of G and let G = G/N . Prove
that if G = hx, yi then G = hx̄, ȳi. Prove more generally that if G = hSi for any
subset S of G, then G = hSi.
Proof. We will prove the general case, since the arguments are similar. Suppose
G = hSi. If S is empty then G and N are trivial and we vacuously have G = hSi.
So let S be nonempty. Then for any g ∈ G, we may write
g = s1 s2 · · · sk , where si ∈ S for 1 ≤ i ≤ k.
Let S be cosets of the form sN , where s ∈ S. Then gN may be written as
gN = (s1 s2 · · · sk )N
= (s1 N )(s2 N ) · · · (sk N ).
This shows that every coset in G can be written as a product of cosets in S.

Therefore G = hSi.
3.1.17 Exercise 17
Let G be the dihedral group of order 16:
G = hr, s | r8 = s2 = 1, rs = sr−1 i
and let G = G/hr4 i be the quotient of G by the subgroup generated by r4 (this
subgroup is the center of G, hence is normal).
(a) Show that the order of G is 8.
Solution. hr4 i = {1, r4 }. So for each g ∈ G, we can write ghr4 i = {g, gr4 },
and we see that the cosets in G partition G into sets of two elements.
|G| = 16, so there must be 8 distinct cosets in G.
(b) Exhibit each element of G in the form s̄a r̄b , for some integers a and b.
Solution. The elements of G are

1̄ = {1, r4 }, s̄ = {s, sr4 },
r̄ = {r, r5 }, s̄r̄ = {sr, sr5 },
r̄2 = {r2 , r6 }, s̄r̄2 = {sr2 , sr6 },
r̄3 = {r3 , r7 }, s̄r̄3 = {sr3 , sr7 }.
(c) Find the order of each of the elements of G exhibited in (b).
Solution. The orders are
|1̄| = 1, |s̄| = 2,
|r̄| = 4, |sr| = 2,
|r2 | = 2, |sr2 | = 2,
|r3 | = 4, |sr3 | = 2.
(d) Write each of the following elements of G in the form s̄a r̄b , for some
integers a and b as in (b): rs, sr−2 s, s−1 r−1 sr.
Solution. Since rs = sr−1 = sr7 , and since sr7 = sr3 , we have rs = sr3 .
Likewise,
sr−2 s = s2 r2 = r2
and
s−1 r−1 sr = s−1 sr2 = r2 .
(e) Prove that H = hs̄, r̄2 i is a normal subgroup of G and H is isomorphic

to the Klein 4-group. Describe the isomorphism type of the complete
preimage of H in G.
Solution. By definition, H is a subgroup of G since it is generated by mem-

bers of G. To show that H E G, we need only check that the conjugates
of the generators of H lie in H.
Take any element ḡ ∈ G. If ḡ = rk for some integer k with 0 ≤ k ≤ 3,
then
gsg −1 = rk sr−k = sr−2k = s(r2 )−k ∈ H
and
gr2 g −1 = rk r2 r−k = r2 ∈ H.
The other possibility is that ḡ = srk . Then
gsg −1 = srk s(srk )−1 = srk ssrk = sr2k = s(r2 )k ∈ H
and
gr2 g −1 = srk r2 (srk )−1 = srk+2 srk = r−2 = (r2 )−1 ∈ H.
This shows that H is normal in G.
Note that H = {1, r2 , s̄, sr2 } has exactly four elements. It has already
been shown that every group having four elements is isomorphic to either
the Klein 4-group V4 or to the cyclic group Z4 (see Exercise 2.5.10). Z4 has
only one element of order 2, but it is easy to check that every nonidentity
element in H has order 2, so H ∼ = V4 .
Lastly, the complete preimage of H is π −1 (H), where π : G → G is the
natural projection of G onto G. From the cosets found earlier, we find
that
π −1 (H) = {1, r2 , r4 , r6 , s, sr2 , sr4 , sr6 }.
Call this preimage A. If we notice that
(r2 )4 = s2 = 1 and r2 s = sr−2 = s(r2 )−1 ,
we see that this subgroup A behaves like D8 , and indeed, the mapping
ϕ : A → D8 determined by ϕ(r2 ) = r and ϕ(s) = s extends to a bijective
homomorphism. Therefore A ∼ = D8 .
(f) Find the center of G and describe the isomorphism type of G/Z(G).
Solution. From Exercise 3.1.16, we know that G = hr̄, s̄i. Since r2 com-
mutes with both of these generators, r2 ∈ Z(G). However, the elements r̄
and r3 do not commute with s̄, and the elements sr, sr2 , and sr3 do not
commute with r̄. So
Z(G) = {1̄, r2 }.
The elements of G/Z(G) are
1 = {1̄, r2 }, s = {s̄, sr2 },

r = {r̄, r3 }, sr = {sr, sr3 }.
Notice that each nonidentity element has order 2. Therefore G/Z(G) is

isomorphic to the Klein 4-group, V4 .
3.1.18 Exercise 18
Let G be the quasidihedral group of order 16:
G = hσ, τ | σ 8 = τ 2 = 1, στ = τ σ 3 i
and let G = G/hσ 4 i be the quotient of G by the subgroup generated by σ 4 (this

subgroup is the center of G, hence is normal).
Solution. Since hσ 4 i = {1, σ 4 }, its cosets partition G into pairs of ele-

ments, so that there must be 8 distinct cosets.
(b) Exhibit each element of G in the form τ̄ a σ̄ b , for some integers a and b.
Solution. The elements of G are as follows.
1̄ = {1, σ 4 }, τ̄ = {τ, τ σ 4 },
σ̄ = {σ, σ 5 }, τ σ = {τ σ, τ σ 5 },
σ 2 = {σ 2 , σ 6 }, τ σ 2 = {τ σ 2 , τ σ 6 },
σ 3 = {σ 3 , σ 7 }, τ σ 3 = {τ σ 3 , τ σ 7 }.

Solution. Computing the orders of the elements, we find
|1̄| = 1, |τ̄ | = 2,
|σ̄| = 4, |τ σ| = 2,
|σ 2 | = 2, |τ σ 2 | = 2,
|σ 3 | = 4, |τ σ 3 | = 2.
(d) Write each of the following elements of G in the form τ̄ a σ̄ b , for some
integers a and b as in (b): στ , τ σ −2 τ , τ −1 σ −1 τ σ.
Solution. We have
στ = τ σ 3 ,
τ σ −2 τ = σ 2 ,
and
τ −1 σ −1 τ σ = σ 6 = σ 2 .
(e) Prove that G ∼

= D8 .
Proof. In G, we know |τ̄ | = 2 and |σ̄| = 4, and we also know that
στ = τ σ 3 = τ σ 7 = τ σ −1 .
So G satisfies the same relations as those given in the presentation of D8

and has the same number of elements as D8 . Therefore G ∼ = D8 .
3.1.19 Exercise 19
Let G be the modular group of order 16:
G = hu, v | u2 = v 8 = 1, vu = uv 5 i
and let G = G/hv 4 i be the quotient of G by the subgroup generated by v 4 (this

subgroup is contained in the center of G, hence is normal).
Solution. As in the previous two exercises, the cosets of hv 4 i = {1, v 4 }

partition G into pairs of elements, so that G = 8.
(b) Exhibit each element of G in the form ūa v̄ b , for some integers a and b.
Solution. The elements are
1̄ = {1, v 4 }, ū = {u, uv 4 },
v̄ = {v, v 5 }, ūv̄ = {uv, uv 5 },
v̄ 2 = {v 2 , v 6 }, ūv̄ 2 = {uv 2 , uv 6 },
v̄ 3 = {v 3 , v 7 }, ūv̄ 3 = {uv 3 , uv 7 }.
Solution. The orders of the elements are
|1̄| = 1, |ū| = 2,
|v̄| = 4, |uv| = 4,
|v 2 | = 2, |uv 2 | = 2,
|v 3 | = 4, |uv 3 | = 4.
(d) Write each of the following elements of G in the form ūa v̄ b , for some
integers a and b as in (b): vu, uv −2 u, u−1 v −1 uv.
Solution. We have vu = uv 5 = uv,
uv −2 u = uv 6 u = vuvu = vuuv 5 = v 6 = v 2 ,
and
u−1 v −1 uv = uv 7 uv = uv 5 (v 2 uv) = vuuv 10 v = v 12 = 1̄.
(e) Prove that G is abelian and is isomorphic to Z2 × Z4 .
Proof. First, note that since uv = uv 5 = vu, the generators of G commute.

Therefore G is abelian.
Let x be a generator for Z2 and y a generator for Z4 . Define the map
ϕ : G → Z2 × Z4 by
ϕ(ūa v̄ b ) = (xa , y b ), a, b ∈ Z.
It is not hard to see that ϕ is well defined, since in both groups the
exponents a and b can be reduced modulo 2 or 4, respectively, in the same
way.
Now, let ḡ = ūa v̄ b and h̄ = ūc v̄ d be two elements of G. Then
ϕ(ḡ h̄) = ϕ(ua v b uc v d )

= ϕ(ua+c v b+d )
= (xa+c , y b+d )
= (xa , y b )(xc , y d )
= ϕ(ḡ)ϕ(h̄),
and we see that ϕ is a homomorphism.

ϕ is also surjective since each element of Z 2 × Z 4 can be obtained with
the appropriate exponents on ū and v̄. Being a surjective map between
two sets of the same size, ϕ must be a bijection, and thus an isomorphism.
Hence G ∼ = Z2 × Z4 .
3.1.20 Exercise 20
Let G = Z/24Z and let G e = G/h12i, where for each integer a we simplify
notation by writing a as ã.
e
(a) Show that G e = {0̃, 1̃, . . . , f
11}.
Proof. Since h12i = {0̄, 12}, each coset will consist of the pair
{n̄, 12 + n} = ñ for 0 ≤ n ≤ 11,
and each of these is distinct.
(b) Find the order of each element of G.

e

|0̃| = 1, |6̃| = 2,
|1̃| = 12, |7̃| = 12,
|2̃| = 6, |8̃| = 3,
|3̃| = 4, |9̃| = 4,
|4̃| = 3, |f
10| = 6,
|5̃| = 12, |f
11| = 12.
e∼
(c) Prove that G = Z/12Z.
e → Z/12Z by
Proof. Define the function ϕ : G
ϕ(ñ) = n̄.
This is clearly a bijection, and it is a homomorphism since
e + ñ) = m + n = m + n̄ = ϕ(m)
ϕ(m e + ϕ(ñ)
e ñ ∈ G.
for any m, e∼
e Therefore G = Z/12Z.
3.1.21 Exercise 21
Let G = Z4 × Z4 be given in terms of the following generators and relations:
G = hx, y | x4 = y 4 = 1, xy = yxi.
Let G = G/hx2 y 2 i (note that every subgroup of the abelian group G is normal).
Proof. Since x and y commute, (x2 y 2 )2 = 1 and so hx2 y 2 i = {1, x2 y 2 }.

The cosets of hx2 y 2 i in G are as follows:
1̄ = {1, x2 y 2 }, ȳ = {y, x2 y 3 },
x̄ = {x, x3 y 2 }, xy = {xy, x3 y 3 },
x2 = {x2 , y 2 }, x2 y = {x2 y, y 3 },
x3 = {x3 , xy 2 }, x3 y = {x3 y, xy 3 }.
Each of these is distinct, and all elements in G are accounted for. Therefore
|G| = 8.
(b) Exhibit each element of G in the form x̄a ȳ b , for some integers a and b.
Solution. As given above, the elements are 1̄, x̄, x2 , x3 , ȳ, xy, x2 y, and
x3 y.

|1̄| = 1, |ȳ| = 4,
|x̄| = 4, |xy| = 2,
|x2 | = 2, |x2 y| = 4,
|x3 | = 4, |x3 y| = 2.
(d) Prove that G ∼

= Z4 × Z2 .
Proof. Let a be a generator for Z4 and b a generator for Z2 . Also, in G,

let ū = x̄ and v̄ = xy, and note that every element in G can be written in
the form ūm v̄ n for integers m and n. Define a function ϕ : G → Z4 × Z2
by
ϕ(um v n ) = (am , bn ).
First we show that ϕ is well defined. Suppose um v n = up v q . Then
xm+n y n = xp+q y q and we must have n ≡ q (mod 4) and m + n ≡ p + q
(mod 4), which together imply that m ≡ p (mod 4) and n ≡ q (mod 2).
This means that in Z4 ×Z2 , the elements (am , bn ) and (ap , bq ) are actually
the same, so ϕ is a well defined function.
Next, ϕ is easily seen to be surjective. And since both groups have the
same order, it is also a bijection.
Finally, ϕ is a homomorphism since
ϕ((am , bn )(ap , bq )) = ϕ((am+p , bn+q ))
= xm+p y n+q
= xm y n · xp y q
= ϕ((am , bn ))ϕ((ap , bq )).
Therefore G ∼
= Z4 × Z2 .
3.1.22 Exercise 22
(a) Prove that if H and K are normal subgroups of a group G then their
intersection H ∩ K is also a normal subgroup of G.
Proof. Suppose H and K are normal subgroups of G. We already know

that H ∩ K is a subgroup of G, so we need to show that it is normal.
Choose any g ∈ G and any x ∈ H ∩ K. Since x ∈ H and H E G, we
know gxg −1 ∈ H. Likewise, since x ∈ K and K E G, we have gxg −1 ∈ K.
Therefore gxg −1 ∈ H ∩ K. This shows that g(H ∩ K)g −1 ⊆ H ∩ K, and
this is true for all g ∈ G. By Theorem 6 (5) (which we will prove in
Exercise 3.1.25), this is enough to show that H ∩ K E G.
(b) Prove that the intersection of an arbitrary nonempty collection of normal

subgroups of a group is a normal subgroup (do not assume the collection
is countable).
Proof. Let G be a group and let I be a nonempty set of indices, not

necessarily countable. Consider the collection of subgroups {Nα | α ∈ I},
where Nα E G for each α ∈ I. Let
\
N= Nα .
α∈I
We know N is a subgroup of G by Exercise 2.1.10.

For any g ∈ G and any n ∈ N , we must have n ∈ Nα for each α. And
since Nα E G, we have gng −1 ∈ Nα for each α. Therefore gng −1 ∈ N ,
which shows that gN g −1 ⊆ N for each g ∈ G. As before, this is enough
to complete the proof.
3.1.23 Exercise 23
Prove that the join of any nonempty collection of normal subgroups of a group
is a normal subgroup.
Proof. Let G be a group and I a nonempty set of indices, and let {Nα | α ∈ I}
be a collection of subgroups of G, with Nα E G for each α ∈ I. Let N be the
join of all the subgroups in the collection:
N = hNα | α ∈ Ii.
Fix an element g ∈ G and n ∈ N . Since n belongs to the join of {Nα }, we

can write n as the product of finitely many elements, each belonging to one of
the sets in the collection:
n = x1 x2 · · · xk , where for each i, xi ∈ Nαi for some αi ∈ I.
Since Nαi E G for each i, we have gxi g −1 ∈ Nαi ⊆ N . Now notice that we can
write
gng −1 = gx1 x2 x3 · · · xk−1 xk g −1

= gx1 (g −1 g)x2 (g −1 g)x3 · · · xk−1 (g −1 g)xk g −1
= (gx1 g −1 )(gx2 g −1 ) · · · (gxk g −1 ).
Therefore gng −1 can be written as a product of elements in N , and so must

be in N by closure. Since g and n were chosen arbitrarily, this shows that
gN g −1 ⊆ N for each g ∈ G so that N E G.
3.1.24 Exercise 24
Prove that if N E G and H is any subgroup of G then N ∩ H E H.
Proof. We know that the intersection of two subgroups is a subgroup, so N ∩ H
is a subgroup of H. Let h ∈ H and x ∈ N ∩ H. Then since x ∈ N and N E G,
hxh−1 ∈ N . But x ∈ H so hxh−1 ∈ H. Therefore hxh−1 ∈ N ∩ H. Since this
is true for any h ∈ H and x ∈ N ∩ H, we have N ∩ H E H.
3.1.25 Exercise 25
(a) Prove that a subgroup N of G is normal if and only if gN g −1 ⊆ N for all
g ∈ G.
Proof. The left-to-right implication is immediate from the definition. For

the other direction, suppose that N is a subgroup of a group G, with
gN g −1 ⊆ N for each g ∈ G. (3.1)
Fix an element g0 ∈ G and let x ∈ N . Consider the element y = g0−1 xg0 .

Then y ∈ N by (3.1) (simply take g = g0−1 ). Therefore
g0 yg0−1 = g0 (g0−1 xg0 )g0−1 = x.
This shows that x ∈ g0 N g0−1 . Since g0 ∈ G was arbitrary, x ∈ gN g −1 for

each g so that N ⊆ gN g −1 . Together with (3.1), we have gN g −1 = N so
that N E G.
(b) Let G = GL2 (Q), let N be the subgroup of upper triangular matrices with
integer entries and 1’s on the diagonal, and let g be the diagonal matrix
with entries 2, 1. Show that gN g −1 ⊆ N but g does not normalize N .
Proof. We have

1 b 2 0
N= b∈Z and g = .
0 1 0 1
For any b ∈ Z,
1
1 b −1 2 0 1 b 0 1 2b
g g = 2 = ∈ N,
0 1 0 1 0 1 0 1 0 1
and we see that gN g −1 ⊆ N .

But notice that elements in gN g −1 must have an even integer in the upper-
right entry. This will always be the case, regardless of which n ∈ N we
use. Now consider the element

1 1
x= ∈ N.
0 1
Then there is no matrix n in N with the property that gng −1 = x, be-

cause as we have observed, the upper-right entry of gng −1 is always even.
Consequently, g does not normalize N .
3.1.26 Exercise 26
Let a, b ∈ G.
(a) Prove that the conjugate of the product of a and b is the product of the
conjugate of a and the conjugate of b. Prove that the order of a and the
order of any conjugate of a are the same.
Proof. For any g ∈ G, we have
g(ab)g −1 = ga(g −1 g)bg −1 = (gag −1 )(gbg −1 ),
so the conjugate of a product is the product of the conjugates. Note that a

simple induction argument will allow us to extend this result to products
of any finite number of elements of G.
Next, suppose |a| = n < ∞. From the above result, we have
(gag −1 )n = gan g −1 = gg −1 = 1,
so |gag −1 | ≤ n. On the other hand, if (gag −1 )k = 1, then gak g −1 = 1, and

multiplying on the left by g −1 and on the right by g gives ak = g −1 g = 1,
so that |gag −1 | ≥ n. Therefore
|gag −1 | = n = |a|, for all g ∈ G.
Lastly, if |a| = ∞, then there is no positive integer n such that an = 1.

But if (gag −1 )n = 1 for some positive n, then the same argument as before
shows that an = 1, which is a contradiction. Therefore |gag −1 | = ∞ in
this case.
(b) Prove that the conjugate of a−1 is the inverse of the conjugate of a.
Proof. For any g ∈ G,
(ga−1 g −1 )(gag −1 ) = ga−1 (g −1 g)ag −1 = ga−1 ag −1 = gg −1 = 1,
so ga−1 g −1 = (gag −1 )−1 .
(c) Let N = hSi for some subset S of G. Prove that N E G if gSg −1 ⊆ N for
all g ∈ G.
Proof. First note that if S is the empty set, then N is the trivial subgroup
and is therefore normal in G. So assume that S is nonempty.
Now suppose gSg −1 ⊆ N for all g ∈ G, and pick any x ∈ N . Since
N = hSi, we may write
x = s1 s2 · · · sk , where si ∈ S for each i = 1, 2, . . . , k.
Since we have already proven above that the conjugate of a product is the
product of the conjugates, we have for all g ∈ G that
gxg −1 = (gs1 g −1 )(gs2 g −1 ) · · · (gsk g −1 ) ∈ N.
Therefore gN g −1 ⊆ N for all g ∈ G and we can conclude that N E G.
(d) Deduce that if N is the cyclic group hxi, then N is normal in G if and
only if for each g ∈ G, gxg −1 = xk for some k ∈ Z.
Proof. Since a ∈ N if and only if a = xk for some integer k, the left-to-

right implication is immediate from the definition of a normal subgroup,
and the other direction is just a special case of the previous result, with
S = {x}.
(e) Let n be a positive integer. Prove that the subgroup N of G generated by

all the elements of G of order n is a normal subgroup of G.
Proof. Let S = {g ∈ G | |g| = n} and let N = hSi. If S is empty then

N is the trivial subgroup and is normal in G, so assume S is nonempty.
Then for any s ∈ S and g ∈ G, we have
|gsg −1 | = |s| = n,
so gsg −1 ∈ S ⊆ N . Then gSg −1 ⊆ N for each g ∈ G, and we can apply

our earlier result to conclude that N E G.
3.1.27 Exercise 27
Let N be a finite subgroup of a group G. Show that gN g −1 ⊆ N if and only if
gN g −1 = N . Deduce that NG (N ) = {g ∈ G | gN g −1 ⊆ N }.
Proof. Fix an element g ∈ G. If gN g −1 = N then certainly gN g −1 ⊆ N .
Conversely, suppose that gN g −1 ⊆ N . Define the function ϕ : N → gN g −1
by
ϕ(x) = gxg −1 .
We will show that ϕ is a bijection. First, if gag −1 = gbg −1 for a, b ∈ N , then
cancellation shows that a = b. Therefore ϕ is injective. And if y ∈ gN g −1 , then
by definition there is some x ∈ N with y = gxg −1 = ϕ(x), so ϕ is surjective. We
conclude that ϕ is a bijection, and therefore |gN g −1 | = |N |. But gN g −1 ⊆ N
and N is finite, so we must have equality:
gN g −1 = N.
Finally, we know by definition that NG (N ) = {g ∈ G | gN g −1 = N }. As

was just proved, gN g −1 = N if and only if gN g −1 ⊆ N (since N is finite), so
the normalizer can also be written as
NG (N ) = {g ∈ G | gN g −1 ⊆ N }.
3.1.28 Exercise 28
Let N be a finite subgroup of a group G and assume N = hSi for some subset
S of G. Prove that an element g ∈ G normalizes N if and only if gSg −1 ⊆ N .
Proof. Fix an element g ∈ G. First, if gN g −1 = N then it must be true that
gSg −1 ⊆ N , since gSg −1 is a subset of gN g −1 .
Conversely, suppose gSg −1 ⊆ N . If S is empty then N is trivial and must
be normal, so suppose S is nonempty. Choose any x ∈ N . Since S generates
N , we have
x = s1 s2 · · · sk , where si ∈ S for each i = 1, 2, . . . , k.

As the conjugate of a product is the product of the conjugates, we know that

gxg −1 = gs1 g −1 · · · gsk g −1 . Since gsi g −1 ∈ gSg −1 ⊆ N , we have gxg −1 ∈ N .
And our choice of x ∈ N was arbitrary, so gN g −1 ⊆ N . Since N is finite, we
may use the result from Exercise 3.1.27, to conclude that g normalizes N .
3.1.29 Exercise 29
Let N be a finite subgroup of G and suppose G = hT i and N = hSi for some
subsets S and T of G. Prove that N is normal in G if and only if tSt−1 ⊆ N
for all t ∈ T .
Proof. If N is normal in G then gN g −1 = N for all g ∈ G, so clearly tSt−1 ⊆ N
for any t ∈ T .
For the other direction, suppose that tSt−1 ⊆ N for all t ∈ T . If S or T is
the empty set, then the result is obvious, so suppose S and T are nonempty.
Choose any g ∈ G. We can write
g = t1 t2 · · · tk with ti ∈ T for each i = 1, 2, . . . , k.
We will use induction on k to prove that gSg −1 ⊆ N . If k = 1, then
gSg −1 = t1 St−1
1 ⊆ N , so the base case is satisfied. Now assume that gSg
−1
⊆N
whenever g can be written as the product of k elements from t, and consider
g = t1 t2 · · · tk tk+1 , ti ∈ T for each i.
Set h = t1 t2 · · · tk , so that g = htk+1 . By the induction assumption, hSh−1 ⊆ N .
So for any s ∈ S, we have
gsg −1 = htk+1 s(htk+1 )−1 = htk+1 st−1
k+1 h
−1
= hxh−1 ,
where x = tk+1 st−1
k+1 ∈ N . So x can be written as
x = s1 s2 · · · s` , si ∈ S for i = 1, 2, . . . , `
and
hxh−1 = (hs1 h−1 )(hs2 h−1 ) · · · (hs` h−1 ) ∈ N.
So gsg −1 ∈ N , which gives gSg −1 ⊆ N . By induction, this statement is true
for any g ∈ G. And since N is finite, our result from Exercise 3.1.28 finishes the
proof.
3.1.30 Exercise 30
Let N ≤ G and let g ∈ G. Prove that gN = N g if and only if g ∈ NG (N ).
Proof. Suppose gN = N g. Then for any x ∈ N , there is a y ∈ N such that
gx = yg. Multiplying on the right by g −1 gives gxg −1 = y ∈ N . This is true
for any x ∈ N , so gN g −1 ⊆ N . On the other hand, if x ∈ N , then there
is a y ∈ N such that xg = gy, and multiplying on the right by g −1 gives
x = gyg −1 ∈ gN g −1 . So N ⊆ gN g −1 and we conclude that the two sets are
equal. Therefore g ∈ NG (N ).
Conversely, suppose g normalizes N . Let x ∈ N be arbitrary. Then we
have x ∈ gN g −1 so that x = gyg −1 for some y ∈ N . Multiplying on the right
by g gives xg = gy. Therefore xg ∈ gN for all x ∈ N , so N g ⊆ gN . By a
symmetric argument, we also have gN ⊆ N g. Therefore the two sets are equal:
gN = N g.
3.1.31 Exercise 31
Prove that if H ≤ G and N is a normal subgroup of H then H ≤ NG (N ).
Deduce that NG (N ) is the largest subgroup of G in which N is normal (i.e., is
the join of all subgroups H for which N E H).
Proof. We already know that NG (N ) is a subgroup, so we only need to establish

that H ⊆ NG (N ). But this is easy: Let h ∈ H. Since N E H we have
hN h−1 = N . Therefore h ∈ NG (N ). This shows that H ≤ NG (N ).
Clearly N E NG (N ), and we have shown that every subgroup of G in which
N is normal must be a subgroup of NG (N ). This implies that NG (N ) is the
largest subgroup of G in which N is normal.
3.1.32 Exercise 32
Prove that every subgroup of Q8 is normal. For each subgroup find the isomor-
phism type of its corresponding quotient.
Solution. The subgroups of Q8 are 1, h−1i, hii, hji, hki, and Q8 . From the
lattice for Q8 , we know that hii, hji, and hki are maximal subgroups, so their
normalizers are either themselves or Q8 . But it is easy to check that, for ex-
ample, jhii(−j) = hii, so NQ8 (hii) = Q8 . By a similar argument, we conclude
that
NQ8 (hii) = NQ8 (hji) = NQ8 (hki) = Q8 ,
so each of these subgroups is normal. And h−1i is certainly normal since it is

in the center of Q8 . Therefore every subgroup of Q8 is normal.
Now Q8 /h−1i = {1̄, ī, j̄, k̄} has order 4. And since ī · ī = −1 = 1̄, we see that
ī has order 2. Similarly j̄ and k̄ have order 2. From the classification of groups
of order 4, we know that Q8 /h−1i ∼ = V4 , where V4 is the Klein four-group.
Similarly, Q8 /hii = {1̄, j̄} has order 2, so it must be isomorphic to Z2 . By
symmetry, Q8 /hji ∼ = Q8 /hki ∼ = Z2 as well.
3.1.33 Exercise 33
Find all normal subgroups of D8 and for each of these find the isomorphism
type of its corresponding quotient.
Solution. Certainly the trivial subgroup h1i is normal in D8 , as is D8 itself.

D8 /h1i ∼
= D8 and D8 /D8 ∼= h1i.
Now consider the subgroups of order 2, aside from hr2 i. For these we have
rsr−1 = rsr3 = sr2 6∈ hsi,

r(sr)r−1 = rs = sr3 6∈ hsri,
r(sr2 )r−1 = rsr = s 6∈ hsr2 i,
r(sr3 )r−1 = rsr2 = sr 6∈ hsr3 i,
so none of hsi, hsri, hsr2 i, and hsr3 i are normal in D8 .

Next, since hr2 i = Z(D8 ), we know that hr2 i E D8 . The cosets in D8 /hr2 i
are
1̄ = {1, r2 },
r̄ = {r, r3 },
s̄ = {s, sr2 },
sr = {sr, sr3 }.
Since |r̄| = |s̄| = |sr| = 2, we see that D8 /hr2 i ∼

= V4 .
From the lattice for D8 , we know that the remaining subgroups are maximal.
So their normalizers are either the subgroups themselves or all of D8 . Since
shris−1 = {1, r3 , r2 , r} = hri,

rhs, r2 ir−1 = {1, sr2 , r2 , s} = hs, r2 i,
and
rhsr, r2 ir−1 = {1, sr3 , r2 , sr} = hsr, r2 i,
we see that ND8 (hri) = ND8 (hs, r2 i) = ND8 (hsr, r2 i) = D8 . Therefore hri,
hs, r2 i, and hsr, r2 i are normal in D8 . The cosets of hri are
1̄ = {1, r, r2 , r3 } and s̄ = {s, sr, sr2 , sr3 },
the cosets of hs, r2 i are
1̄ = {1, s, r2 , sr2 } and r̄ = {r, r3 , sr3 , sr},
and the cosets of hsr, r2 i are
1̄ = {1, sr, r2 , sr3 } and r̄ = {r, s, r3 , sr2 }.
Since there are only two distinct cosets in each case, we have
D8 /hri ∼
= D8 /hs, r2 i ∼
= D8 /hsr, r2 i ∼
= Z2 .
3.1.34 Exercise 34
Let D2n = hr, s | rn = s2 = 1, rs = sr−1 i be the usual presentation of the dihe-
dral group of order 2n and let k be a positive integer dividing n.
(a) Prove that hrk i is a normal subgroup of D2n .
Proof. First, since rk commutes with r, we know that r ∈ ND2n (hrk i).
Also,
srk s−1 = srk s = s2 r−k = (rk )−1 ∈ hrk i.
Therefore s ∈ ND2n (hrk i). Since both s and r (the generators of D2n )
normalize hrk i, we must have ND2n (hrk i) = D2n . Therefore hrk i E D2n .
(b) Prove that D2n /hrk i ∼

= D2k .
Proof. Since hrk i = {1, rk , r2k , . . . , rn−k }, we see that the order of hrk i is
n/k. Therefore, each coset will consist of n/k elements, so the cosets will
partition D2n into 2n/(n/k) = 2k distinct sets.
Consider the two cosets
r̄ = {r, rk+1 , r2k+1 , . . . , rn−k+1 }
and
s̄ = {s, srk , sr2k , . . . , srn−k }.
These are clearly distinct. Observe that
(r̄)k = rk = 1̄ and (s̄)2 = s2 = 1̄,
so |s̄| = 2 and |r̄| ≤ k. But if 0 < i < k, then ri 6= 1̄, so |r̄| = k. Moreover,
(r̄)(s̄) = rs = sr−1 = (s̄)(r̄)−1 .
Since we have shown that D2n /hrk i has the same number of elements as
D2k , and since r̄ and s̄ satisfy the same relations as r and s do in the
group presentation for D2k , it follows that D2n /hrk i ∼
= D2k .
3.1.35 Exercise 35
Prove that SLn (F ) E GLn (F ) and describe the isomorphism type of the quo-
tient group.
Solution. We have shown in Exercise 2.1.9 that SLn (F ) ≤ GLn (F ), so we only
need to show that it is normal.
Take any matrix A ∈ GLn (F ) and B ∈ SLn (F ). Then
det(ABA−1 ) = det(A) det(B) det(A)−1 = det(A) · 1 · det(A)−1 = 1.
This shows that ABA−1 ∈ SLn (F ), and we conclude that SLn (F ) E GLn (F ).
Now let G = GLn (F )/SLn (F ) and consider the map ϕ : G → F × defined
by
ϕ(A) = det(A),
where A is the coset whose representative is A.
First we must show that this mapping is well defined. Suppose the matrices
A, B ∈ GLn (F ) are such that A = B. Each element in the coset A is of the
form AS, for some S ∈ SLn (F ). Then det(AS) = det(A) det(S) = det(A). But
A = B so we also have AS = BS0 for some S0 ∈ SLn (F ). Then
det(A) = det(BS0 ) = det(B) det(S0 ) = det(B).
So ϕ is a well defined function from G to F × .

Next we show that ϕ is injective. Suppose det(A) = det(B) for the matrices
A, B ∈ GLn (F ). We need to show that A = B. Take any X ∈ A, so that
X = AS for some S ∈ SLn (F ). Then the matrix T = B −1 AS belongs to
SLn (F ) since
det(T ) = det(B −1 AS) = det(B)−1 det(A) det(S) = det(A)−1 det(A) = 1.

Now X = (BB −1 )(AS) = BT ∈ B, so that A ⊆ B. By symmetry, the reverse

inclusion holds and we see that A = B so that ϕ is injective.
For surjectivity, let f ∈ F × be arbitrary. Let A be the diagonal matrix with
upper-left entry f and all other diagonal entries 1. Then det(A) = f so that
ϕ(A) = f and ϕ is surjective.
Lastly, we show that ϕ is a homomorphism. For any A, B ∈ G, we have
ϕ(AB) = det(AB) = det(A) det(B) = ϕ(A)ϕ(B).
We have now established that ϕ is a bijective homomorphism from G to F × .

Therefore G ∼
= F ×.
3.1.36 Exercise 36
Prove that if G/Z(G) is cyclic then G is abelian.
Proof. Let G be a group such that G/Z(G) is cyclic. Then G/Z(G) = hxZ(G)i
for some x ∈ G. In particular, any coset yZ(G) can be written in the form
yZ(G) = (xZ(G))k = xk Z(G), k ∈ Z. (3.2)
Now let a, b ∈ G. Since every element of G belongs to some coset of Z(G),

(3.2) allows us to write
a = xi z1 and b = xj z2 ,
for some i, j ∈ Z and z1 , z2 ∈ Z(G). Since z1 and z2 commute with x, and since
x commutes with itself, we have
ab = xi z1 xj z2 = xj z2 xi z1 = ba.
This shows that G is abelian.
3.1.37 Exercise 37
Let A and B be groups. Show that {(a, 1) | a ∈ A} is a normal subgroup of
A × B and the quotient of A × B by this subgroup is isomorphic to B.
Proof. Let G = A × B and H = {(a, 1) | a ∈ A}. Then for any (a, 1) ∈ H and
(x, y) ∈ G, we have
(x, y)(a, 1)(x, y)−1 = (x, y)(a, 1)(x−1 , y −1 )

= (xax−1 , yy −1 )
= (xax−1 , 1) ∈ H.
This shows that H E G.

Define the mapping ϕ : G/H → B by
ϕ((a, b)H) = b.
Suppose (a1 , b1 ) and (a2 , b2 ) are representatives of the same coset of H. Then
(a1 , b1 )−1 (a2 , b2 ) = (a−1 −1

1 a2 , b1 b2 ) ∈ H.
So b−1
1 b2 = 1, which implies that b1 = b2 , so ϕ is well defined.
Next we show that ϕ is a bijection. Let (a1 , b)H and (a2 , b)H be cosets in
G/H, so that both cosets have the same image under ϕ. Then
(a1 , b)−1 (a2 , b) = (a1−1 a2 , b−1 b) = (a−1

1 a2 , 1) ∈ H.
This shows that (a1 , b)H = (a2 , b)H so that ϕ is injective. And for any b ∈ B,
ϕ((1, b)H) = b, so ϕ is surjective also.
Finally, let (a1 , b1 )H and (a2 , b2 )H be cosets of H. Then
ϕ((a1 , b1 )H(a2 , b2 )H) = ϕ((a1 a2 , b1 b2 )H)

= b1 b2
= ϕ((a1 , b1 )H)ϕ((a2 , b2 )H),
and ϕ is a bijective homomorphism. Therefore G/H ∼

= B.
3.1.38 Exercise 38
Let A be an abelian group and let D be the (diagonal) subgroup {(a, a) | a ∈ A}
of A × A. Prove that D is a normal subgroup of A × A and (A × A)/D ∼ = A.
Proof. Since A is abelian, A × A is abelian. Every subgroup of an abelian group

is normal, so D E A × A.
Two cosets (a1 , b1 )D and (a2 , b2 )D are equal if and only if
(a1 , b1 )−1 (a2 , b2 ) ∈ D,
that is if and only if a−1 −1 −1 −1

1 a2 = b1 b2 or a1 b1 = a2 b2 . Therefore the mapping
ϕ : (A × A)/D → A given by
ϕ((a, b)D) = ab−1
is a well defined injection. It is also a surjection, since for any element a ∈ A,

ϕ((a, 1)D) = a. And ϕ is a homomorphism since
ϕ((a1 , b1 )D(a2 , b2 )D) = ϕ((a1 a2 , b1 b2 )D)

= a1 a2 (b1 b2 )−1
= (a1 b−1 −1
1 )(a2 b2 )
= ϕ((a1 , b1 )D)ϕ((a2 , b2 )D),
where the second-to-last equality follows from the fact that A is abelian. We
have shown that ϕ is an isomorphism of groups, so (A × A)/D ∼ = A.
3.1.39 Exercise 39
Suppose A is the non-abelian group S3 and D is the diagonal subgroup
{(a, a) | a ∈ A}
of A × A. Prove that D is not normal in A × A.

Proof. Let a = (1 3) and b = (1 2 3) be members of A = S3 . Then a−1 = a and

b−1 = (3 2 1). Let
x = (a, b) ∈ A × A and y = (a, a) ∈ D.
Then
xyx−1 = (a3 , bab−1 ) = (a, bab−1 ).
If D E A × A, then we must have xyx−1 ∈ D, so that a = bab−1 . But
bab−1 = (1 2 3)(1 3)(3 2 1) = (2 3) 6= (1 3) = a.
We conclude that D is not a normal subgroup of A × A.
3.1.40 Exercise 40
Let G be a group, let N be a normal subgroup of G and let G = G/N . Prove that
x̄ and ȳ commute in G if and only if x−1 y −1 xy ∈ N . (The element x−1 y −1 xy
is called the commutator of x and y and is denoted [x, y].)
Proof. First suppose that xy = yx. Then xyN = yxN so that xyz1 = yxz2 for
some z1 , z2 ∈ N , which implies that
x−1 y −1 xy = z2 z1−1 ∈ N.
Conversely, assume that x−1 y −1 xy ∈ N , so that x−1 y −1 xy = z for some

z ∈ N . Then
xy = yxz.
Now a ∈ xyN if and only if a = xyz0 for some z0 ∈ N , if and only if a = yxzz0 ,
if and only if a ∈ yxN . Therefore xy = yx.
3.1.41 Exercise 41
Let G be a group. Prove that N = hx−1 y −1 xy | x, y ∈ Gi is a normal subgroup
of G and G/N is abelian (N is called the commutator subgroup of G).
Proof. For any g ∈ G and x−1 y −1 xy ∈ N we have
g(x−1 y −1 xy)g −1 = gx−1 (g −1 g)y −1 (g −1 g)x(g −1 g)yg −1

= (gxg −1 )−1 (gyg −1 )−1 (gxg −1 )(gyg −1 ) ∈ N.
Since the conjugates of the generators of N are themselves in N , this shows that
gN g −1 ⊆ N for all g ∈ G. That is, N E G.
We know that G/N is abelian by Exercise 3.1.40, since [x, y] ∈ N for any
x̄, ȳ ∈ G/N .
3.1.42 Exercise 42
Assume both H and K are normal subgroups of G with H ∩ K = 1. Prove that
xy = yx for all x ∈ H and y ∈ K.
Proof. Fix x ∈ H and y ∈ K. Since H E G, we know that y −1 xy ∈ H.
So x−1 y −1 xy ∈ H also. And since K E G, we have x−1 y −1 x ∈ K so that
x−1 y −1 xy ∈ K also. Therefore x−1 y −1 xy ∈ H ∩ K. But H ∩ K = 1. This
shows that
x−1 y −1 xy = 1, or xy = yx.
Since this is true for any x ∈ H and y ∈ K, the proof is complete.
3.1.43 Exercise 43
Assume P = {Ai | i ∈ I} is any partition of G with the property that P
is a group under the “quotient operation” defined as follows: to compute the
product of Ai with Aj take any element ai of Ai and any element aj of Aj and
let Ai Aj be the element of P containing ai aj (this operation is assumed to be
well defined). Prove that the element of P that contains the identity of G is a
normal subgroup of G and the elements of P are the cosets of this subgroup (so
P is just a quotient group of G in the usual sense).
Proof. Let P be as stated. For any x ∈ G, let x̄ denote the element of P which
contains x.
First we show that 1̄ ≤ G. 1 ∈ 1̄ so the set is nonempty. Take any x, y ∈ 1̄,
so that x̄ = ȳ = 1̄. By definition of the operation on P, we have
xy = x̄ · ȳ = 1̄ · 1̄ = 1 · 1 = 1̄,
so xy ∈ 1̄. Moreover,
x−1 = 1̄ · x−1 = x̄ · x−1 = xx−1 = 1̄,
and x−1 ∈ 1̄. This shows that 1̄ is a subgroup of G.

Next, take any g ∈ G and x ∈ 1̄. Then
gxg −1 = ḡ · x̄ · g −1 = ḡ · 1̄ · g −1 = gg −1 = 1̄.
Therefore gxg −1 ∈ 1̄ and 1̄ E G.

Finally, we show that the coset g 1̄ is a member of P. In particular, we show
that g 1̄ = ḡ. If x ∈ g 1̄, then x = gy for some y ∈ 1̄. Then
x̄ = gy = ḡ · ȳ = ḡ · 1̄ = ḡ,
so x ∈ ḡ and g 1̄ ⊆ ḡ.
Conversely, if x ∈ ḡ, then x̄ = ḡ. Multiplying on the left by ḡ −1 gives
g −1 x = g −1 g = 1̄. Therefore g −1 x ∈ 1̄. So
x = gg −1 x = g(g −1 x) ∈ g 1̄,
and we see that g 1̄ ⊇ ḡ. This shows that g 1̄ = ḡ as required.

3.2 More on Cosets and Lagrange’s Theorem

Let G be a group.
3.2.1 Exercise 1
Which of the following are permissible orders for subgroups of a group of order
120: 1, 2, 5, 7, 9, 15, 60, 240? For each permissible order give the corresponding
index.
Solution. The permissible orders are 1, 2, 5, 15, and 60. The other orders do
not divide 120 and so by Lagrange’s Theorem are not possible.
The subgroup of order 1 would have index 120, the subgroup of order 2 would
have index 60, the subgroup of order 5 would have index 24, the subgroup with
order 15 would have index 8, and the subgroup of order 60 would have index
2.
3.2.2 Exercise 2
Prove that the lattice of subgroups of S3 in Section 2.5 is correct (i.e., prove that
it contains all subgroups of S3 and that their pairwise joins and intersections
are correctly drawn).
Proof. The subgroups contained in the lattice are 1, h(1 2)i, h(1 3)i, h(2 3)i,
h(1 2 3)i, and S3 itself. Since |S3 | = 6, any nontrivial subgroups must have
order 2 or 3. Since h(1 2 3)i = h(1 3 2)i, all cyclic subgroups are accounted for.
Now suppose S3 has a non-cyclic proper subgroup H. Say H is generated by
σ and τ . Then |H| = 3 and H = {1, σ, τ }. But |σ| must divide |H|, so |σ| = 3.
Then σ and σ 2 are distinct, and we must have τ = σ 2 . Hence H is cyclic, which
gives a contradiction. This shows that all proper subgroups of S3 are cyclic.
Therefore all subgroups are present in the lattice.
Note that the subgroups of order 2 cannot themselves be subgroups of
h(1 2 3)i, since 2 - 3. Therefore every nontrivial subgroup is maximal, and the
lattice is correct.
3.2.3 Exercise 3
Prove that the lattice of subgroups of Q8 in Section 2.5 is correct.
Proof. By Lagrange’s Theorem the possible subgroups of Q8 have orders 1, 2,

4, and 8. So every nontrivial subgroup has order 2 or 4. The only element of
Q8 having order 2 is −1, so h−1i is the only possible subgroup with that order.
Every other nonidentity element has order 4. The only subgroups of order 4 are
hii, hji, and hki, since all elements of Q8 belong to one of these subgroups.
h−1i is contained in each of hii, hji, and hki, and the latter three are maximal
by Lagrange. Therefore the lattice is correct.
3.2. MORE ON COSETS AND LAGRANGE’S THEOREM 151
3.2.4 Exercise 4
Show that if |G| = pq for some primes p and q (not necessarily distinct) then
either G is abelian or Z(G) = 1.
Proof. Let G be as stated. If G is abelian, there is nothing left to prove, so

suppose G is not abelian. Then Z(G) is proper. By Lagrange, there are only
three possibilities for the order of Z(G): the order is either 1, p, or q.
Now assume that Z(G) is not trivial. Then without loss of generality we
may suppose that |Z(G)| = p. Since the center of a group is always normal, we
may consider the quotient group G/Z(G). Again by Lagrange, we have
|G| pq
|G/Z(G)| = = = q.
|Z(G)| p
Now, q is prime, so we may apply Corollary 10 to conclude that G/Z(G) is cyclic.

Then by Exercise 3.1.36, it follows that G is abelian, which is a contradiction.
Therefore Z(G) is the trivial subgroup, and the proof is complete.
3.2.5 Exercise 5
Let H be a subgroup of G and fix some element g ∈ G.
(a) Prove that gHg −1 is a subgroup of G of the same order as H.
Proof. 1 ∈ gHg −1 , so gHg −1 is nonempty. Let x, y ∈ gHg −1 . Then

x = gh1 g −1 and y = gh2 g −1 for some h1 , h2 ∈ H. We have
xy −1 = (gh1 g −1 )(gh2 g −1 )−1

= (gh1 g −1 )(gh−1
2 g
−1
)
= gh1 (g −1 g)h−1
2 g
−1
= gh1 h−1
2 g
−1
∈ gHg −1 .
By the subgroup criterion, gHg −1 ≤ G.

Now define the map ϕ : H → gHg −1 by ϕ(h) = ghg −1 . Then ϕ is clearly
surjective. It is also injective, since gh1 g −1 = gh2 g −1 implies that h1 = h2
by the left and right cancellation laws. This shows that |H| = |gHg −1 |
(in fact they are isomorphic).
(b) Deduce that if n ∈ Z+ and H is the unique subgroup of G of order n then

H E G.
Proof. For any g ∈ G, we know from the above that gHg −1 ≤ G and that
|gHg −1 | = |H| = n. Since H is the only subgroup of G with order n, it
follows that H = gHg −1 . Since this is true for every g ∈ G, H E G by
definition.
3.2.6 Exercise 6
Let H ≤ G and let g ∈ G. Prove that if the right coset Hg equals some left
coset of H in G then it equals the left coset gH and g must be in NG (H).
Proof. Suppose Hg = aH for some a ∈ G. Since g ∈ Hg, we have g ∈ aH.

Since the (right) cosets of H form a partition of G (Proposition 4), this implies
that aH = gH. Therefore Hg = gH and g ∈ NG (H).
3.2.7 Exercise 7
Let H ≤ G and define a relation ∼ on G by
a ∼ b if and only if b−1 a ∈ H.
Prove that ∼ is an equivalence relation and describe the equivalence class of

each a ∈ G. Use this to prove Proposition 4.
Proof. For any a ∈ G, we certainly have a−1 a = 1 ∈ H, so a ∼ a and ∼ is

reflexive. If a ∼ b for a, b ∈ G then, since H must be closed under inverses, we
have a−1 b = (b−1 a)−1 ∈ H so that b ∼ a, and ∼ is symmetric. Lastly, if a ∼ b
and b ∼ c for a, b, c ∈ G, then
c−1 a = c−1 (bb−1 )a = (c−1 b)(b−1 a) ∈ H,
so a ∼ c and we see that ∼ is transitive. This shows that ∼ is an equivalence

relation.
Note that the equivalence classes of ∼ form a partition of G (see Proposition 2
of Section 0.1). It is not difficult to see that a ∼ b if and only if a and b belong
to the same left coset of H. Therefore the left cosets of H form a partition of
G, providing an alternative proof for Proposition 4.
3.2.8 Exercise 8
Prove that if H and K are finite subgroups of G whose orders are relatively
prime then H ∩ K = 1.
Proof. Let H and K be as stated, and let n = |H ∩ K|. We know that the
intersection of two subgroups is a subgroup, so H ∩ K ≤ H. By Lagrange’s
Theorem, n must divide |H|. But H ∩ K ≤ K also, so n must divide |K|. And
since |H| and |K| have no common divisor other than 1, we must have n = 1.
Therefore H ∩ K is the trivial subgroup.
3.2.9 Exercise 9
Let G be a finite group and let p be a prime dividing |G|. Let S denote the set
of p-tuples of elements of G the product of whose coordinates is 1:
S = {(x1 , x2 , . . . , xp ) | xi ∈ G and x1 x2 · · · xp = 1}.
(a) Show that S has |G|p−1 elements, hence has order divisible by p.
Proof. Let T be the set of all (p − 1)-tuples of elements of G. Then T has

|G|p−1 elements. We show that S and T share the same cardinality.
Define ϕ : T → S by
ϕ(x1 , x2 , . . . , xp−1 ) = (x1 , x2 , . . . , xp−1 , (x1 x2 · · · xp−1 )−1 ).
Given two elements α, β ∈ S, if α = β then clearly their first p − 1

coordinates must be equal, so T is injective. Now suppose α ∈ S. Then
the image of the first p − 1 coordinates of α under T must be α, since the
last coordinate is completely determined by the others (group inverses are
unique). Therefore ϕ is a bijection, so |S| = |T | = |G|p−1 .
Define the relation ∼ on S by letting α ∼ β if β is a cyclic permutation of α.
(b) Show that a cyclic permutation of an element of S is again an element of

S.
Proof. Let α = (x1 , x2 , . . . , xp ) ∈ S. Then x1 x2 · · · xp = 1. Multiplying

on the left by xp and then on the right by x−1 p gives
xp x1 x2 · · · xp−1 = xp x−1
p = 1.
Therefore (xp , x1 , x2 , . . . , xp−1 ) ∈ S. We may repeat this process p − 1

times to see that all cyclic permutations of α are in S.
(c) Prove that ∼ is an equivalence relation on S.
Proof. Let α, β, γ ∈ S be arbitrary. α is a cyclic permutation of itself

(namely the identity permutation), so α ∼ α and ∼ is reflexive.
If α ∼ β then β is a cyclic permutation of α, so by taking the inverse of
this permutation we see that α is a cyclic permutation of β. Therefore
β ∼ α, and ∼ is symmetric.
Lastly, if α ∼ β and β ∼ γ then γ is a cyclic permutation of β and β
is a cyclic permutation of α, and by taking the composition of these two
permutations we see that γ is a cyclic permutation of α, so that α ∼ γ.
Thus ∼ is transitive and the proof is complete.
(d) Prove that an equivalence class contains a single element if and only if it
is of the form (x, x, . . . , x) with xp = 1.
Proof. Take any α ∈ S and let [α] denote the equivalence class containing
α.
First, if α is the only element in [α], then all cyclic permutations of α
must be the same. This is not possible unless all coordinates of α are the
same. So α has the form (x, x, . . . , x), where xp = 1.
Conversely, if α has the form (x, x, . . . , x) with all coordinates the same,
then every cyclic permutation of α will leave α unchanged. Therefore [α]
contains only the one element.
(e) Prove that every equivalence class has order 1 or p (this uses the fact that
p is a prime). Deduce that |G|p−1 = k + pd, where k is the number of
classes of size 1 and d is the number of classes of size p.
Proof. Again, let α ∈ S with [α] denoting the corresponding equivalence

class, and let
α = (x1 , x2 , . . . , xp ),
with the xi ’s not necessarily distinct. Suppose [α] contains exactly n

members. Then 1 ≤ n ≤ p. For all k ∈ Z, we must have
xi = xj whenever i + kn ≡ j (mod p), (3.3)
since α can only be cycled n times before arriving back at itself.

Now there are two cases, either n = p or 1 ≤ n < p. In the first case
there is nothing left to prove, so assume 1 ≤ n < p. Then (n, p) = 1 since
p is prime. So by Exercise 0.3.14, we know that n has a multiplicative
inverse, n−1 , modulo p. Then, taking k = n−1 , (3.3) tells us that xi = xj
whenever i + 1 ≡ j (mod p). Therefore xi+1 = xi for all i with 1 ≤ i < p,
and x1 = xp . But then every coordinate of α is the same, so [α] has only
one member. Hence n = 1 in this case.
We have shown that if [α] has exactly n elements, then n = 1 or n = p.
Since the equivalence classes partition S, we see that
|G|p−1 = |S| = k + pd,
where k is the number of classes of size 1 and d is the number of classes

of size p.
(f) Since {(1, 1, . . . , 1)} is an equivalence class of size 1, conclude from (e) that
there must be a nonidentity element x in G with xp = 1, i.e., G contains
an element of order p.
Proof. Since p divides |G|, it certainly divides |G|p−1 . So p | (k + pd). But

this implies that p | k. Therefore k > 1, so there must be more than one
equivalence class of ∼ having only one element. One of these must be of
the form (x, x, . . . , x) where x is not the identity of G. Therefore this x
is such that xp = 1, so |x| divides p. We know x is not the identity, so
|x| = p.
3.2.10 Exercise 10
Suppose H and K are subgroups of finite index in the (possibly infinite) group
G with |G : H| = m and |G : K| = n. Prove that
l.c.m.(m, n) ≤ |G : H ∩ K| ≤ mn.
Deduce that if m and n are relatively prime then |G : H ∩ K| = |G : H|·|G : K|.

Proof. For any g ∈ G, consider the cosets gH, gK, and g(H ∩ K). First, if
x ∈ g(H ∩ K), then x ∈ gH and x ∈ gK so g(H ∩ K) ⊆ (gH ∩ gK). On the
other hand, if x ∈ gH and x ∈ gK, then g −1 x ∈ H ∩ K, so x ∈ g(H ∩ K) and
we have (gH ∩ gK) ⊆ g(H ∩ K). Therefore
gH ∩ gK = g(H ∩ K) for all g ∈ G.
Now each coset of H ∩ K is the intersection of one coset of H and one coset of
K. There are exactly mn such intersections, so |G : H ∩ K| is finite and is at
most mn.
As we will show in Exercise 3.2.11 below, we must have
|G : H ∩ K| = |G : H||H : H ∩ K|.
So |G : H| divides |G : H ∩ K|. That is, if |G : H ∩ K| = s, then m | s. By the
same argument, we know that n | s also. Therefore s ≥ [m, n], where [m, n]
denotes the least common multiple of m and n. This completes the proof of the
inequality.
Finally, if (m, n) = 1, then [m, n] = mn, and we see that |G : H ∩ K| must
equal mn.
3.2.11 Exercise 11
Let H ≤ K ≤ G. Prove that |G : H| = |G : K| · |K : H| (do not assume G is
finite).
Proof. Since cosets of H are contained within cosets of K, if H has infinite index
in K or if K has infinite index in G then H has infinite index in G also. So we
will assume that |K : H| and |G : K| are finite.
Let m, n ∈ Z+ where
m = |G : K| and n = |K : H|.
Let g1 , g2 , . . . , gm be representatives of the distinct cosets of K in G, and let
k1 , k2 , . . . , kn be representatives of the distinct cosets of H in K. Take any
element a ∈ G. Since the cosets of K partition G, a belongs to exactly one
coset gi K, so a = gi b for some b ∈ K. And since the cosets of H partition K, b
belongs to exactly one coset kj H, so b = kj c for some c ∈ H. Then a = gi kj c,
where i and j are uniquely determined.
Note that, within each coset gi K, we cannot have gi kj1 H = gi kj2 H with
j1 6= j2 . For, if this is possible, then let x belong to this common coset. Then
x = gi kj1 h1 and x = gi kj2 h2 for some h1 , h2 ∈ H. Multiplying on the left by
gi−1 then gives kj1 h1 = kj2 h2 , and we see that kj1 and kj2 are representatives
of the same coset of H in K, which is a contradiction.
Therefore the cosets of H partition G into mn disjoint subsets, so
|G : H| = mn = |G : K| · |K : H|.
3.2.12 Exercise 12
Let H ≤ G. Prove that the map x 7→ x−1 sends each left coset of H in G onto
a right coset of H and gives a bijection between the set of left cosets and the
set of right cosets of H in G (hence the number of left cosets of H in G equals
the number of right cosets).
Proof. Let ϕ be a mapping between the set of left cosets of H in G to the set
of right cosets, given by
ϕ(xH) = Hx−1 .
First we show that ϕ is well defined. Suppose x and y are representatives of
the same left coset gH. Then x = gh1 and y = gh2 for some h1 , h2 ∈ H. So
x−1 = h−11 g
−1
∈ Hg −1 and y −1 = h−1 2 g
−1
∈ Hg −1 , and we see that ϕ sends
both xH and yH to the same right coset Hg −1 and is therefore well defined.
To show that ϕ is a bijection, we simply exhibit a two-sided inverse function.
Let ψ map the set of right cosets of H in G onto the set of left cosets via the
map Hx 7→ x−1 H. By the same argument as before, ψ is well defined. ψ ◦ ϕ and
ϕ ◦ ψ are obviously the identity, so ψ = ϕ−1 and ϕ is a bijection. It follows that
the number of left cosets of H in G is equal to the number of right cosets.
3.2.13 Exercise 13
Fix any labelling of the vertices of a square and use this to identify D8 as a
subgroup of S4 . Prove that the elements of D8 and h(1 2 3)i do not commute in
S4 .
Proof. Let the vertices of a square be labelled 1, 2, 3, 4 in a clockwise fashion.
Then every element in D8 induces a distinct permutation of these vertices. It is
easy to see that these permutations form a subgroup of S4 . r is identified with
(1 2 3 4) and s is identified with (2 4).
We have
(1 2 3 4)(1 2 3) = (1 3 2 4) 6= (1 3 4 2) = (1 2 3)(1 2 3 4)
and
(2 4)(1 2 3) = (1 4 2 3) 6= (1 2 4 3) = (1 2 3)(2 4).
Since the generators of D8 and the generator of h(1 2 3)i do not commute, we
see that the elements in the two subgroups do not in general commute with one
another.
3.2.14 Exercise 14
Prove that S4 does not have a normal subgroup of order 8 or a normal subgroup
of order 3.
Proof. Suppose S4 has a normal subgroup H of order 8. Now, consider that the
elements (1 2 3 4) and (1 2 4 3) cannot both be in H since they generate all of
S4 , as we showed in Exercise 2.4.8. But
(1 2 3 4) = (1 4)(3 2)(1 3) and (1 2 4 3) = (3 4)(3 2)(1 3),
so H cannot contain all of the 2-cycles (1 4), (3 2), (3 4), and (1 3). We see
then that S4 contains an element σ of order 2 which does not belong to H.
Therefore H ∩ hσi = 1 and we have by Corollary 15 that Hhσi ≤ S4 . And by
Proposition 13 we see that
|H||hσi|
|Hhσi| = = 8 · 2 = 16.
|H ∩ hσi|
Now S4 , a group of order 24, has a subgroup of order 16, which contradicts
Lagrange’s Theorem. Therefore H does not exist: there is no normal subgroup
of order 8 in S4 .
Next, suppose that S4 has a normal subgroup K of order 3. We know by
Corollary 10 that any subgroup of order 3 is cyclic. Now, S4 has more than one
subgroup of order 3, for example
h(1 2 3)i = {1, (1 2 3), (1 3 2)} and h(2 3 4)i = {1, (2 3 4), (2 4 3)}.
So K has a trivial intersection with some subgroup hτ i of order 3. Then since

K is normal we know Khτ i ≤ S4 having order 3 · 3 = 9, but this is impossible.
Therefore S4 has no normal subgroup of order 3.
3.2.15 Exercise 15
Let G = Sn and for fixed i ∈ {1, 2, . . . , n} let Gi be the stabilizer of i. Prove
that Gi ∼
= Sn−1 .
Proof. Let G act on {1, 2, . . . , n} and fix some i from this latter set. Now suppose
σ ∈ Gi . We can always write σ as a product of disjoint cycles using the Cycle
Decomposition Algorithm presented in Section 1.3. Each of the cycles in the
cycle decomposition of sigma must not contain i, since i needs to be stabilized.
Therefore Gi consists of all permutations of the set {1, 2, . . . , n} − {i}, that is it
is the permutations of a set with n − 1 elements. And it has been shown that,
for finite sets A and B, SA ∼ = SB when |A| = |B|. Therefore Gi ∼ = Sn−1 .
3.2.16 Exercise 16
Use Lagrange’s Theorem in the multiplicative group (Z/pZ)× to prove Fermat’s
Little Theorem: if p is a prime then ap ≡ a (mod p) for all a ∈ Z.
Proof. Let p be a prime. Then
(Z/pZ)× = {1̄, 2̄, 3̄, . . . , p − 1}.
Now either a is a multiple of p or not. If not, then ā ∈ (Z/pZ)× and by La-

grange’s Theorem and Corollary 9, we know that |ā| must divide p−1. Therefore
āp−1 = 1̄, so
ap−1 ≡ 1 (mod p),
and multiplying both sides by a gives the desired result.
The other possibility is that a and p are not relatively prime. In this case,
we have
ap ≡ 0 ≡ a (mod p),
and the result still holds.
3.2.17 Exercise 17
Let p be a prime and let n be a positive integer. Find the order of p̄ in
(Z/(pn − 1)Z)×
and deduce that n | ϕ(pn − 1) (here ϕ is Euler’s function).

Solution. Note that pn ≡ 1 (mod pn − 1) so |p̄| ≤ n. On the other hand, if

pk ≡ 1 (mod pn − 1), then (pn − 1) | (pk − 1) and we see that k ≥ n. Therefore
|p̄| = n.
Since (Z/(pn − 1)Z)× has order ϕ(pn − 1), we know by Corollary 9 that
n | ϕ(pn − 1).
3.2.18 Exercise 18
Let G be a finite group, let H be a subgroup of G and let N E G. Prove that
if |H| and |G : N | are relatively prime then H ≤ N .
Proof. By Proposition 13 we know that
|H||N |
|HN | = .
|H ∩ N |
Now H ∩ N ≤ H, so |H ∩ N | divides |H| and we can write
|HN | = k|N |,
where k = |H|/|H ∩ N |. But HN ≤ G since N is normal in G (Corollary 15),

so k|N | divides |G|. That is, there is an integer ` such that
|G|
k`|N | = |G|, or k` = = |G : N |.
|N |
We see that k divides |G : N |. But k also divides |H|. Since k is a common

divisor of two relatively prime numbers, it follows that k = 1. Therefore
|H| = |H ∩ N |
and we must have H = H ∩ N so that H ≤ N .
3.2.19 Exercise 19
Prove that if N is a normal subgroup of the finite group G and (|N |, |G : N |) = 1
then N is the unique subgroup of G of order |N |.
Proof. Suppose H is a subgroup that also has order |N |. Then |H| and |G : N |
are relatively prime, so we may apply the result from the previous exercise to
conclude that H ≤ N . And since they have the same order, H = N . Therefore
N is the only subgroup with order |N |.
3.2.20 Exercise 20
If A is an abelian group with A E G and B is any subgroup of G prove that
A ∩ B E AB.
Proof. We know that AB is a subgroup of G by Corollary 15. We want to show

that A ∩ B E AB. Take any member ab ∈ AB, where a ∈ A and b ∈ B, and let
c ∈ A ∩ B. Then
(ab)c(ab)−1 = abcb−1 a−1 .
Now, since A E G, it follows that bcb−1 ∈ A, and since a−1 must commute with
other members of A, we have
(ab)c(ab)−1 = a(bcb−1 )a−1 = aa−1 (bcb−1 ) = bcb−1 .
We know already that bcb−1 ∈ A. But bcb−1 is also a product of elements from
B, so bcb−1 ∈ A ∩ B. What we have shown is that for all ab ∈ AB,
(ab)(A ∩ B)(ab)−1 ⊆ A ∩ B.
This proves that A ∩ B E AB.
3.2.21 Exercise 21
Prove that Q has no proper subgroups of finite index. Deduce that Q/Z has no
proper subgroups of finite index.
Proof. Suppose Q does have a proper subgroup of finite index, call it H. Since
Q is abelian, H E Q. In Exercise 3.1.15 of the previous section, we showed
that the quotient of a divisible abelian group by a proper subgroup must also
be divisible. And we know Q is divisible by Exercise 2.4.19, so Q/H is divisible.
But then Q/H is a finite abelian group that is divisible, and we showed that this
was impossible in that same exercise. Therefore H cannot be a proper subgroup
of finite index.
We know that Q/Z cannot have proper subgroups of finite index for exactly
the same reason.
3.2.22 Exercise 22
Use Lagrange’s Theorem in the multiplicative group (Z/nZ)× to prove Euler’s
Theorem: aϕ(n) ≡ 1 (mod n) for every integer a relatively prime to n, where ϕ
denotes Euler’s ϕ-function.
Proof. We know that (Z/nZ)× has order equal to ϕ(n). If a is relatively prime to
n, then ā ∈ (Z/nZ)× so |ā| must divide ϕ(n). Therefore aϕ(n) ≡ 1 (mod n).
3.2.23 Exercise 23
100
Determine the last two digits of 33 .
Solution. Since 100 = 22 · 52 , we compute
ϕ(100) = ϕ(22 )ϕ(52 ) = (22 − 2)(52 − 5) = 40.
Also, ϕ(40) = ϕ(23 )ϕ(5) = 16.

Now (3, 16) = 1, so by Euler’s Theorem we know that
3ϕ(40) = 316 ≡ 1 (mod 40).

Therefore
3100 = 36·16+4
= (316 )6 · 34
≡ 16 · 34 (mod 40)
≡ 81 (mod 40)
≡1 (mod 40).
This shows that 3100 = 1 + 40k for some k ∈ Z.

Since (3, 100) = 1, we again have by Euler’s Theorem that
3ϕ(100) = 340 ≡ 1 (mod 100).
Therefore
100
33 = 31+40k
= 3 · (340 )k
≡ 3 · 1k (mod 100)
≡3 (mod 100).
100
So the last two digits of 33 in decimal notation are 03.
3.3. THE ISOMORPHISM THEOREMS 161
3.3 The Isomorphism Theorems

Let G be a group.
3.3.1 Exercise 1
Let F be a finite field of order q and let n ∈ Z+ . Prove that
|GLn (F ) : SLn (F )| = q − 1.
Proof. In Exercise 3.1.35 we saw that GLn (F )/SLn (F ) ∼

= F × . Therefore
|GLn (F ) : SLn (F )| = |GLn (F )/SLn (F )| = |F × | = q − 1,
since F × consists of all members of F excluding the 0 element.
3.3.2 Exercise 2
Prove all parts of the Lattice Isomorphism Theorem.
Proof. First we show that there is a bijection from the set A of subgroups A of G
containing N onto the set A of subgroups A = A/N of G/N . Let π : G → G/N
be the natural projection of G onto G/N . Then define the map Φ : A → A by
Φ(A) = π(A) = {aN | a ∈ A}.
That Φ(A) ≤ G/N for any A ≤ G is easy to check: Φ(A) is nonempty since it
includes 1N , and if a, b ∈ A, then
(aN )(bN )−1 = (ab−1 )N ∈ Φ(A).
To show Φ is injective, suppose Φ(A) = Φ(B). Let a ∈ A. Then π(a) = π(b)

for some b ∈ B, so b−1 a ∈ N and a ∈ bN . Since N ≤ B, this shows that a ∈ B
so that A ⊆ B. A similar argument will show that A ⊇ B and so A = B.
To see that Φ is surjective, let A = A/N be a subgroup of G/N . We saw in
Exercise 3.1.1 that the complete preimage of a subgroup in G/N is a subgroup
of G, so there is A ∈ A such that Φ(A) = A.
We have shown that Φ is a bijection. Now suppose A, B ≤ G with N ≤ A
and N ≤ B.
(a) A ≤ B if and only if A ≤ B.

If A ≤ B, then every coset of N in A is clearly also a coset of N in B, so
that A ≤ B. On the other hand, if A ≤ B then for any a ∈ A, we have
aN ∈ B so that b−1 a ∈ N for some b ∈ B, which implies a ∈ bN ⊆ B, so
A ≤ B.
(b) If A ≤ B, then |B : A| = |B : A|.

Every coset bA of A in B corresponds to a coset b̄A of A in B. And every
coset b̄A of A corresponds to a coset bA of A. It is then easy to check that
bA 7→ b̄A is a bijection from the cosets of A in B onto the cosets of A in
B. Therefore
|B : A| = |B : A|.
(c) hA, Bi = hA, Bi.

xN ∈ hA, Bi if and only if x ∈ hA, Bi, if and only if
x = x1 x2 · · · xn , where xi ∈ A ∪ B for each i.
But this is true if and only if
xN = (x1 N )(x2 N ) · · · (xn N ), xi N ∈ A ∪ B for each i,
if and only if xN ∈ hA, Bi. Therefore hA, Bi = hA, Bi.
(d) A ∩ B = A ∩ B.
xN ∈ A ∩ B if and only if x ∈ A ∩ B if and only if x ∈ A and x ∈ B, and
this is true if and only if xN ∈ A and xN ∈ B, that is, xN ∈ A ∩ B.
(e) A E G if and only if A E G.

Suppose A E G. Then if g ∈ G and a ∈ A, we have gag −1 ∈ A, and
(gN )(aN )(g −1 N ) = (gag −1 )N ∈ A.
Therefore A E G.
Conversely, suppose A E G. Then if ḡ ∈ G and ā ∈ A, we have
ḡāḡ −1 = (gag −1 )N ∈ A,
so gag −1 ∈ A. Hence A E G.
3.3.3 Exercise 3
Prove that if H is a normal subgroup of G of prime index p then for all K ≤ G
either
(a) K ≤ H or
(b) G = HK and |K : K ∩ H| = p.
Proof. Let H have prime index p as stated. Since K ≤ NG (H) = G, we may

apply the Second Isomorphism Theorem to see that KH ≤ G and H E KH.
And KH = HK by Proposition 14. Now consider the index of HK in G.
We know by Exercise 3.2.11 that
|G : H| = |G : HK| · |HK : H|.
But |G : H| is prime, so there are only two possibilities for |G : HK|: Either
HK has index 1, in which case HK = G, or |G : HK| = p. In the latter case,
|HK : H| = 1 so H = HK which implies that K ≤ H.
So either K ≤ H or G = HK. And if G = HK, then the Second Isomor-
phism Theorem tells us that K/(H ∩ K) ∼ = HK/H, so
|K : H ∩ K| = |HK : H| = |G : H| = p.
3.3.4 Exercise 4
Let C be a normal subgroup of the group A and let D be a normal subgroup of
the group B. Prove that
(C × D) E (A × B) and (A × B)/(C × D) ∼
= (A/C) × (B/D).
Proof. Define the map ϕ : A × B → (A/C) × (B/D) by
ϕ((a, b)) = (aC, bD).
This is a homomorphism since
ϕ((a1 , b1 )(a2 , b2 )) = ϕ((a1 a2 , b1 b2 ))

= (a1 a2 C, b1 b2 D)
= (a1 C, b1 D)(a2 C, b2 D)
= ϕ((a1 , b1 ))ϕ((a2 , b2 )).
Moreover, we can show that ker ϕ = C × D. For, if ϕ((a, b)) = (1C, 1D), then
a ∈ C and b ∈ D so that (a, b) ∈ C × D and ker ϕ ≤ C × D. On the other
hand, if (c, d) ∈ C × D then ϕ((c, d)) = (cC, dD) = (1C, 1D) so ker ϕ ≥ C × D.
Therefore ker ϕ = C × D.
We now have, by the First Isomorphism Theorem, that C × D E A × B and
(A × B)/(C × D) ∼ = ϕ(A × B) = (A/C) × (B/D).
3.3.5 Exercise 5
Let QD16 = hσ, τ i be the quasidihedral group described in Exercise 2.5.11.
Prove that hσ 4 i is normal in QD16 and use the Lattice Isomorphism Theorem
to draw the lattice of subgroups of QD16 /hσ 4 i. Which group of order 8 has the
same lattice as this quotient? Use generators and relations for QD16 /hσ 4 i to
decide the isomorphism type of this group.
Solution. Note that

σ 4 τ = τ σ 12 = τ σ 4 ,
so σ 4 commutes with every element of QD16 . This is enough to show that

hσ 4 i E QD16 .
Using the Lattice Isomorphism Theorem, we draw the lattice for QD16 /hσ 4 i
below. We can see that the dihedral group D8 has the same lattice structure as
QD16 /hσ 4 i.
Since
σ̄ 4 = σ 4 = 1̄, τ̄ 2 = τ 2 = 1̄,
and
τ̄ σ̄ = τ σ = σ 3 τ = σ̄ −1 τ̄ ,
we see that the generators σ̄ and τ̄ satisfy the same relations as r and s do in
D8 . Therefore QD16 /hσ 4 i ∼
= D8 .
QD16 /hσ 4 i
hσ 2 , τ i hσi hσ 2 , τ σi
hσ 4 , τ σ 2 i hσ 4 , τ i hσ 2 i hτ σi hτ σ 3 i
1̄
3.3.6 Exercise 6
Let M = hv, ui be the modular group of order 16 described in Exercise 2.5.14.
Prove that hv 4 i is normal in M and use the Lattice Isomorphism Theorem to
draw the lattice of subgroups of M/hv 4 i. Which group of order 8 has the same
lattice as this quotient? Use generators and relations for M/hv 4 i to decide the
isomorphism type of this group.
Solution. Since
uv 4 = uv 20 = v 4 u,
we see that v 4 commutes with every element of M so that hv 4 i E M . We get
the following lattice.
M/hv 4 i
hu, v 2 i huvi hvi
hu, v 4 i huv 2 i hv 2 i
1̄
Notice that the lattice looks similar to the one we constructed for Z2 × Z4
in Exercise 2.5.12. We have
ū2 = u2 = 1̄, v̄ 4 = v 4 = 1̄,
and
ūv̄ = uv = uv 25 = v 5 u = v̄ū.
Since the generators ū and v̄ satisfy the same relations as do a and b in the
presentation for Z2 × Z4 (given in Exercise 2.5.12), we conclude that M/hv 4 i is
isomorphic to Z2 × Z4 .
3.3.7 Exercise 7
Let M and N be normal subgroups of G such that G = M N . Prove that
G/(M ∩ N ) ∼
= (G/M ) × (G/N ).
Solution. We draw the lattice for G, with the double-lines representing the
quotient group G/(M ∩ N ).
G = MN
M N
M ∩N
Define ϕ : G → (G/M ) × (G/N ) by
ϕ(g) = (gM, gN ).
This is a homomorphism since
ϕ(ab) = (abM, abN ) = (aM, aN )(bM, bN ) = ϕ(a)ϕ(b).
We also have ker ϕ = M ∩ N .

Next, we will show that ϕ is a surjection. Let (aM, bN ) ∈ (G/M ) × (G/N ).
Since G = M N , we can write a = m1 n1 and b = m2 n2 for some m1 , m2 ∈ M
and n1 , n2 ∈ N . Then
aM = M a = M m1 n1 = M n1 = n1 M and bN = m2 n2 N = m2 N.
Now
ϕ(m2 n1 ) = (M m2 n1 , m2 n1 N ) = (n1 M, m2 N ) = (aM, bN ).
So, by the First Isomorphism Theorem, we have
G/(M ∩ N ) ∼
= (G/M ) × (G/N )
as required.
3.3.9 Exercise 9
Let p be a prime and let G be a group of order pa m, where p does not divide
m. Assume P is a subgroup of G of order pa and N is a normal subgroup
of G of order pb n, where p does not divide n. Prove that |P ∩ N | = pb and
|P N/N | = pa−b . (The subgroup P of G is called a Sylow p-subgroup of G. This
exercise shows that the intersection of any Sylow p-subgroup of G with a normal
subgroup N is a Sylow p-subgroup of N .)
Proof. We know by Lagrange’s Theorem that |P N | divides |G|, so that
|P N | = pk t where 0 ≤ k ≤ a and t | m.
Since P ≤ P N , we also know that pa | pk t, so that k = a. And since N ≤ P N ,

we know that n | t. Therefore
|P N | pa t
|P N/N | = = b = pa−b s, for some s ∈ Z with t = ns.
|N | p n
Since N is normal, we may apply the Second Isomorphism Theorem to de-
termine that P N/N ∼
= P/P ∩ N . Then
|P/P ∩ N | = |P N/N | = pa−b s.
But |P | must divide |P/P ∩ N |, so
|P/P ∩ N | = pa−b .
By Lagrange, this implies that
|P ∩ N | = |P |/pa−b = pb .
So |P N/N | = pa−b and |P ∩ N | = pb .
3.3.10 Exercise 10
Generalize the preceding exercise as follows. A subgroup H of a finite group G
is called a Hall subgroup of G if its index in G is relatively prime to its order:
(|G : H|, |H|) = 1. Prove that if H is a Hall subgroup of G and N E G, then
H ∩ N is a Hall subgroup of N and HN/N is a Hall subgroup of G/N .
Proof. Since |H| divides |G|, we can write |G| = k|H| for some integer k, with
(k, |H|) = 1.
By the formula from Proposition 13, we have
|H||N | |N |
|HN | = = |H| · .
|H ∩ N | |H ∩ N |
Since |HN | must divide |G|, we have
|N |
divides k.
|H ∩ N |
This tells us that (|N : H ∩ N |, |H|) = 1. And since |H ∩ N | divides |H|, we
also have
(|N : H ∩ N |, |H ∩ N |) = 1,
so that H ∩ N is a Hall subgroup of N .
Next, observe that
|G|/|N | |G|
|G/N : HN/N | = =
|HN |/|N | |HN |
|G|/|H| |G : H|
= = .
|HN |/|H| |HN : H|
This implies that |G/N : HN/N | divides |G : H|. Also,
|H||N | |H|
|HN/N | = = ,
|N ||H ∩ N | |H ∩ N |
which shows that |HN/N | divides |H|. But |G : H| and |H| are relatively prime,
so
(|G/N : HN/N |, |HN/N |) = 1.
Therefore HN/N is also a Hall subgroup of G/N .
3.4 Composition Series and the Hölder Program

3.4.1 Exercise 1
Prove that if G is an abelian simple group then G ∼
= Zp for some prime p (do
not assume G is a finite group).
Proof. Let G be an abelian simple group. Then |G| > 1 and we may take some
nonidentity element x of G. Now, either hxi = 6 G or hxi = G.
In the first case, hxi is a nontrivial proper subgroup of G. But G is abelian,
so hxi must be a normal proper subgroup, which contradicts the fact that G is
simple. Hence hxi = G.
Then G is cyclic. If G has infinite order, then by Theorem 7 of Chapter 2,
hx2 i is a nontrivial proper subgroup, again a contradiction. Therefore G has
finite order. But then, again by Theorem 7 of Chapter 2, hxn i is a proper sub-
group for any proper divisor n of |G|. Therefore |G| is prime, and by Theorem 4
of Chapter 2, G ∼ = Zp .
3.4.2 Exercise 2
Exhibit all 3 composition series for Q8 and all 7 composition series for D8 . List
the composition factors in each case.
Solution. For Q8 , the composition series are
1 E h−1i E hii E Q8 ,
1 E h−1i E hji E Q8 ,
and
1 E h−1i E hki E Q8 .
Note that each subgroup has index 2 in its containing subgroup, and so must
be normal. Each of the composition factors is isomorphic to Z2 .
For D8 , we get the following composition series:
1 E hsi E hs, r2 i E D8 ,
1 E hr2 si E hs, r2 i E D8 ,
1 E hr2 i E hs, r2 i E D8 ,
1 E hr2 i E hri E D8 ,
1 E hr2 i E hrs, r2 i E D8 ,
1 E hrsi E hrs, r2 i E D8 ,
1 E hr3 si E hrs, r2 i E D8 .
Again, each subgroup is normal within its containing subgroup since they each
have index 2. Each composition factor is isomorphic to Z2 .
3.4. COMPOSITION SERIES AND THE HÖLDER PROGRAM 169
3.4.3 Exercise 3
Find a composition series for the quasidihedral group of order 16. Deduce that
QD16 is solvable.
Solution. Since any subgroup of index 2 is normal, we see from the lattice
(Exercise 2.5.11) that
1 E hσ 4 i E hσ 2 i E hσi E QD16
is a composition series. Since each composition factor has order 2 and is thus
isomorphic to the abelian group Z2 , we see that QD16 is solvable.
3.4.4 Exercise 4
Use Cauchy’s Theorem and induction to show that a finite abelian group has a
subgroup of order n for each positive divisor n of its order.
Proof. Let G be a finite abelian group. We use induction on |G|. Certainly the
result holds for the trivial group. And if |G| = p for some prime p, then the
positive divisors of |G| are 1 and p and the result is again trivial.
Now assume that the statement is true for all groups of order strictly smaller
than |G|, and let n be a positive divisor of |G| with n > 1. First, if n is prime
then Cauchy’s Theorem allows us to find an element x ∈ G having order n.
Then hxi is the desired subgroup. On the other hand, if n is not prime, then
n has a prime divisor p, so that n = kp for some integer k. Cauchy’s Theorem
allows us to find an element x having order p. Set N = hxi. By Lagrange’s
Theorem,
|G|
|G/N | = < |G|.
|N |
Now, by the inductive hypothesis, the group G/N must have a subgroup of
order k. And by the Lattice Isomorphism Theorem, this subgroup has the form
H/N for some subgroup H of G. Then |H| = k|N | = kp = n, so that H has
order n. This completes the inductive step.
3.4.5 Exercise 5
Prove that subgroups and quotient groups of a solvable group are solvable.
Proof. Let G be a solvable group and let H ≤ G. Since G is solvable, we may

find a chain of subgroups
1 = G0 E G1 E G2 E · · · E Gn = G
so that each quotient Gi+1 /Gi is abelian. For each i, define
Hi = Gi ∩ H, 0 ≤ i ≤ n.
Then Hi ≤ Hi+1 for each i. Moreover, if g ∈ Hi+1 and x ∈ Hi , then in particular

g ∈ Gi+1 and x ∈ Gi , so that
gxg −1 ∈ Gi
because Gi E Gi+1 . But g and x also belong to H, so

gxg −1 ∈ Hi ,
which shows that Hi E Hi+1 for each i.
Next, note that
Hi = Gi ∩ H = (Gi ∩ Gi+1 ) ∩ H = Gi ∩ Hi+1 .
By the Second Isomorphism Theorem, we then have
Hi+1 /Hi = Hi+1 /(Hi+1 ∩ Gi ) ∼
= Hi+1 Gi /Gi ≤ Gi+1 /Gi .
Since Hi+1 /Hi is isomorphic to a subgroup of the abelian group Gi+1 /Gi , it
follows that Hi+1 /Hi is also abelian. This completes the proof that H is solvable.
Next, let N E G. For each i, define
Ni = Gi N, 0 ≤ i ≤ n.
Now let g ∈ Ni+1 , where g = g0 n0 with g0 ∈ Gi+1 and n0 ∈ N . Also let x ∈ Ni ,
where x = g1 n1 with g1 ∈ Gi and n1 ∈ N . Then
gxg −1 = g0 n0 g1 n1 n−1 −1
0 g0 .
Now, since N is normal in G, N g = gN , so n0 g1 = g1 n2 for some n2 ∈ N . Then

gxg −1 = g0 g1 (n2 n1 n−1 −1 −1
0 )g0 = g0 g1 n3 g0
for some n3 ∈ N . Then n3 g0−1 = g0−1 n4 for some n4 ∈ N . And g0 g1 g0−1 ∈ Gi

since Gi E Gi+1 , so
gxg −1 = g0 g1 g0−1 n4 ∈ Ni .
This shows that Ni E Ni+1 . So by the Lattice Isomorphism Theorem, we have
Ni+1 /N E Ni /N . This shows that
1 = N0 /N E N1 /N E N2 /N E · · · E Nn /N = G/N.
Moreover, the Third Isomorphism Theorem says that
(Ni+1 /N )/(Ni /N ) ∼
= Ni+1 /Ni ,
so the proof will be complete if we can show that Ni+1 /Ni is abelian.
Let x, y ∈ Ni+1 /Ni . Then
x = (g0 n0 )Ni and y = (g1 n1 )Ni
for some g0 , g1 ∈ Gi+1 and n0 , n1 ∈ N . We have
xyx−1 y −1 = (g0 n0 )(g1 n1 )(g0 n0 )−1 (g1 n1 )−1 Ni
= g0 n0 g1 n1 n−1 −1 −1 −1
0 g0 n1 g1 Ni .
Since N E G, gN = N g for any g ∈ G, so we can find n2 ∈ N such that

xyx−1 y −1 = g0 g1 g0−1 g −1 n2 Ni .
Now Ni = Gi N = N Gi since N E G (see Proposition 14 and its corollary).
Therefore
n2 Ni = n2 N Gi = N Gi = Gi N
and we get
xyx−1 y −1 = g0 g1 g0−1 g −1 Gi N = Gi N.
So xyx−1 y −1 = 1Ni or xy = yx. This completes the proof that G/N is solvable.
3.4.6 Exercise 6
Prove part (1) of the Jordan–Hölder Theorem by induction on |G|.
Proof. Let G be a finite group with G 6= 1. We want to show that G has a
composition series. We will use induction on |G|.
First, if |G| = 2, then G ∼
= Z2 and G has the composition series 1 E G.
Now suppose |G| > 2, and assume that all nontrivial groups with order
strictly less than G have a composition series.
If G is simple, then it has the composition series 1 E G and we are done.
So assume G is not simple. Let N be a normal subgroup of G, with N 6= 1
and N 6= G, and choose N so that no other proper normal subgroup has larger
order.
Since N is a proper subgroup, it has by the induction hypothesis a compo-
sition series
1 = G0 E G1 E G2 E · · · E Gr = N.
Consider the quotient group G/N . If G/N is not simple, then it has a
nontrivial proper normal subgroup M . By the Lattice Isomorphism Theorem,
there is a subgroup M E G with N < M , contradicting the fact that N is
maximal. So G/N is simple and G has the composition series
1 = G0 E G1 E G2 E · · · E Gr = N E G.
By induction, the proof is complete.
3.4.7 Exercise 7
If G is a finite group and H E G, prove that there is a composition series of G,
one of whose terms is H.
Proof. Note that if H = G then any composition series for G must contain H
as its final term, so it will suffice to prove the statement for proper subgroups
H / G.
If |G| = 1, then the result is clear. So suppose |G| > 1 and assume that the
statement is true for all groups with order strictly smaller than G. Fix a proper
normal subgroup H / G.
Now let N be a maximal proper normal subgroup of G containing H. Then
G/N is simple (by the same argument we used in the previous exercise) and
|N | < |G|. Apply the induction hypothesis to N in order to find a composition
series
1 = G0 E G1 E G2 E · · · E Gr = N E G,
where H = Gi for some i with 0 ≤ i ≤ r. Then this is a composition series
for G, one of whose terms is H. By induction, the result holds for all finite
groups.
3.4.9 Exercise 9
Prove the following special case of part (2) of the Jordan–Hölder Theorem:
assume the finite group G has two composition series
1 = N0 E N1 E · · · E Nr = G and 1 = M0 E M1 E M2 = G.
Show that r = 2 and that the list of composition factors is the same.
Proof. First note that r > 1 since the existence of M1 shows that G is not
simple.
Let H = M1 ∩ Nr−1 . By Exercise 3.1.24 we know that H E M1 . But
M1 ∼ = M1 /1 is a simple group, so we must have H = 1 or H = M1 .
If H = M1 then Nr−1 E M1 . Since M1 is simple, Nr−1 = M1 and we must
have r = 2. In this case, both composition series are exactly the same. So we
will suppose that H = 1.
Since M1 and Nr−1 are normal subgroups of G, their product M1 Nr−1 is
also normal since for m ∈ M1 , n ∈ Nr−1 , and g ∈ G we have
gmng −1 = gm(g −1 g)ng −1 = (gmg −1 )(gng −1 ) ∈ M1 Nr−1 .
Now if M1 Nr−1 6= G then M1 / M1 Nr−1 / G. By the Lattice Isomorphism
Theorem we can see that 1 = M1 /M1 / M1 Nr−1 /M1 / G/M1 . But G/M1 is
simple, so this is impossible. Therefore M1 Nr−1 = G.
By the Second Isomorphism Theorem, we have
G/M1 = M1 Nr−1 /M1 ∼
= Nr−1 /(M1 ∩ Nr−1 ) = Nr−1 /H = Nr−1 /1 ∼
= Nr−1 .
Since Nr−1 ∼= G/M1 , we have that Nr−1 is simple and r = 2.
Moreover, the composition factors in both series are isomorphic, but in the
reverse order:
G/Nr−1 ∼
= (G/1)/(G/M1 ) ∼
= M1 /1 and Nr−1 /1 ∼
= Nr−1 ∼
= G/M1 .
This completes the proof.
3.4.10 Exercise 10
Prove part (2) of the Jordan–Hölder Theorem by induction on min{r, s}.
Proof. Suppose G has the two composition series
1 = N0 E N1 E N2 E · · · E Nr = G (3.4)
and
1 = M0 E M1 E M2 E · · · E Ms = G. (3.5)
We want to show that r = s and that the composition factors (possibly taken
in a different order) are the same up to isomorphism.
As instructed, we will use induction on min{r, s}. In Exercise 3.4.9, we have
already taken care of the cases where min{r, s} ≤ 2. Now suppose the statement
is true whenever min{r, s} ≤ k for some k ≥ 2.
Let H = Nr−1 ∩ Ms−1 . Note that Nr−1 cannot be a proper subgroup of
Ms−1 since it would then be a normal subgroup and, after taking quotients, we
would have
1 / Ms−1 /Nr−1 / G/Nr−1 ,
contradicting the fact that G/Nr−1 is simple. For the same reason, Ms−1 cannot
be a proper subgroup of Nr−1 .
Now consider the case where Ms−1 = Nr−1 . Then we can apply the induc-
tion hypothesis to Ms−1 to show that r = s and that all composition factors in
the two series
1 = N0 E · · · E Nr−1 and 1 = M0 E · · · E Ms−1
are the same up to isomorphism and reordering. Since G/Ms−1 = G/Nr−1 , this
would complete the inductive step of the proof.
So we will assume that Ms−1 and Nr−1 are distinct and that neither is
contained in the other. This implies that both subgroups are proper normal
subgroups of Ms−1 Nr−1 . If Ms−1 Nr−1 6= G, then we can apply the Lattice
Isomorphism Theorem to determine that
1 / Ms−1 Nr−1 /Ms−1 / G/Ms−1 ,
which contradicts the fact that G/Ms−1 is simple. Therefore G = Ms−1 Nr−1 .
Now we may apply the Second Isomorphism Theorem to get
G/Ms−1 = Ms−1 Nr−1 /Ms−1 ∼

= Nr−1 /H (3.6)
and
G/Nr−1 = Ms−1 Nr−1 /Nr−1 ∼
= Ms−1 /H, (3.7)
showing that both Nr−1 /H and Ms−1 /H are simple.
Now let H have the composition series
1 = H0 E H1 E H2 E · · · E Ht = H. (3.8)
Then
1 = H0 E H1 E · · · E Ht = H E Nr−1
is a composition series for Nr−1 . By the induction hypothesis, t = r − 2 and the
composition factors for this series are the same as in the series
1 = N0 E N1 E · · · E Nr−1
in some order.
Similarly,
1 = H0 E H1 E · · · E Ht = H E Ms−1
is a composition series for Ms−1 , and by hypothesis, t = s − 2, and the compo-
sition factors are the same as those in the series
1 = M0 E M1 E · · · E Ms−1
in some order. This shows that s = t + 2 = r.

Moreover, the composition factors in (3.4) and (3.5) are isomorphic in some
order: r − 2 of the factors are isomorphic to the factors in (3.8), and the re-
maining two factors in each series are also isomorphic but in the reverse order,
as shown in (3.6) and (3.7).
By induction, we can conclude that part (2) of the Jordan–Hölder Theorem
holds in all cases.
3.4.12 Exercise 12
Prove (without using the Feit–Thompson Theorem) that the following are equiv-
alent:
(a) every group of odd order is solvable

(b) the only simple groups of odd order are those of prime order.
Proof. First, suppose that every group of odd order is solvable. Let G be a
simple group of odd order. Then G is solvable, so G has a chain of subgroups
1 = G0 E G1 E · · · E Gs = G,
where Gi+1 /Gi is abelian. But G is simple, so s = 1 and G itself is abelian.

Since an abelian group of finite order n has a normal subgroup of order d for
every d dividing n, it follows that G, which is simple, must have prime order.
This completes the proof of the left-to-right implication.
Next, assume that the only simple groups of odd order are those of prime
order. Let G be a group of odd order. Now let
1 = N0 E N1 E N2 E · · · E Nr = G
be a composition series for G. Fix an i with 0 ≤ i ≤ r − 1, and consider the

quotient group Ni+1 /Ni . By Lagrange’s Theorem, this quotient group must
have odd order (if G has a subgroup of even order then G itself would have even
order). But then Ni+1 /Ni is a simple group of odd order, so by hypothesis it
has prime order. But any group of prime order is abelian (Corollary 10), so all
the composition factors of G are abelian. Thus G is solvable, completing the
proof.
3.5. TRANSPOSITIONS AND THE ALTERNATING GROUP 175
3.5 Transpositions and the Alternating Group

3.5.1 Exercise 1
In Exercises 1.3.1 and 1.3.2 of Section 1.3 you were asked to find the cycle
decomposition of some permutations. Write each of these permutations as a
product of transpositions. Determine which of these is an even permutation
and which is an odd permutation.
Solution. From Exercise 1.3.1, we have
σ = (1 3 5)(2 4) = (1 5)(1 3)(2 4)

τ = (1 5)(2 3)
σ 2 = (1 5 3) = (1 3)(1 5)
στ = (2 5 3 4) = (2 4)(2 3)(2 5)
τ σ = (1 2 4 3) = (1 3)(1 4)(1 2)
and
τ 2 σ = (1 3 5)(2 4) = (1 5)(1 3)(2 4).
Of these, we can see that τ and σ 2 are even, and the rest are odd.
From Exercise 1.3.2, we get
σ = (1 13 5 10)(3 15 8)(4 14 11 7 12 9)
= (1 10)(1 5)(1 13)(3 8)(3 15)(4 9)(4 12)(4 7)(4 11)(4 14)
τ = (1 14)(2 9 15 13 4)(3 10)(5 12 7)(8 11)
= (1 14)(2 4)(2 13)(2 15)(2 9)(3 10)(5 7)(5 12)(8 11)
2
σ = (1 5)(3 8 15)(4 11 12)(7 9 14)(10 13)
= (1 5)(3 15)(3 8)(4 12)(4 11)(7 14)(7 9)(10 13)
στ = (1 11 3)(2 4)(5 9 8 7 10 15)(13 14)
= (1 3)(1 11)(2 4)(5 15)(5 10)(5 7)(5 8)(5 9)(13 14)
τ σ = (1 4)(2 9)(3 13 12 15 11 5)(8 10 14)
= (1 4)(2 9)(3 5)(3 11)(3 15)(3 12)(3 13)(8 14)(8 10)
and
τ 2 σ = (1 2 15 8 3 4 14 11 12 13 7 5 10)
= (1 10)(1 5)(1 7)(1 13)(1 12)(1 11)(1 14)(1 4)(1 3)(1 8)(1 15)(1 2).
From these, we see that only σ, σ 2 and τ 2 σ are even. The rest are odd.
3.5.2 Exercise 2
Prove that σ 2 is an even permutation for every permutation σ.
Proof. We know that σ can be written as a product of transpositions. Let it be

written as a product of n transpositions,
σ = σ1 σ2 · · · σn ,
where each σi is a transposition. Then we can write
σ 2 = (σ1 · · · σn )(σ1 · · · σn ).
Since σ 2 has been written as a product of 2n transpositions (i.e., an even number

of transpositions), σ 2 is even.
3.5.3 Exercise 3
Prove that Sn is generated by {(i i + 1) | 1 ≤ i ≤ n − 1}.
Proof. We use induction on n. For n = 1, Sn is trivial and is generated by the

empty set. Now suppose the statement is true for Sn , where n is some positive
integer, and consider Sn+1 .
For each n, let
Tn = {(i i + 1) | 1 ≤ i ≤ n − 1}.
We want to show that every permutation in Sn+1 can be written as a product
of members of Tn+1 . Let σ ∈ Sn+1 , and write σ as a product of transpositions.
For each transposition (j k) with j < k, there are two cases. First, if
1 ≤ j < k ≤ n, then the transposition belongs to Sn and by the induction
hypothesis can be written as a product of members of Tn . In this case, we are
done.
The remaining case is where k = n + 1. If j = n, then (j k) is already in
Tn+1 and we are done, so suppose j < n. Let
τ = (k − 1 k)(j k − 1)(k − 1 k).
Notice that τ (j) = k and τ (k) = j, with all other values fixed. Then (j k − 1)
can, by the inductive hypothesis, be written as a product of members of Tn . As
(k − 1 k) belongs to Tn+1 , this completes the inductive step. By induction, the
result holds for all positive integers n.
3.5.4 Exercise 4
Show that Sn = h(1 2), (1 2 3 . . . n)i for all n ≥ 2.
Proof. Note that, for i with 1 ≤ i ≤ n − 2,
(1 2 3 . . . n)(i i + 1)(1 2 3 . . . n)−1 = (i + 1 i + 2).
It follows that
{(i i + 1) | 1 ≤ i ≤ n − 1} ⊆ h(1 2), (1 2 3 . . . n)i,
so by Exercise 3.5.3, we are done.

3.5.5 Exercise 5
Show that if p is prime, Sp = hσ, τ i where σ is any transposition and τ is any
p-cycle.
Proof. Let G = hσ, τ i. For convenience, we will consider Sp to be the group of
permutations of the set {0, 1, . . . , p − 1} rather than {1, 2, . . . , p}. We can also
relabel the elements of Sp so that
σ = (0 k) and τ = (0 1 2 . . . p − 1).
Now observe that
τ στ −1 = (1 k + 1),
where k + 1 is understood to be the least residue of k + 1, modulo p. Similarly,
τ ◦ (1 k + 1) ◦ τ −1 = (2 k + 2)
and, in general,
τ ◦ (n − 1 k + n − 1) ◦ τ −1 = (n k + n),
where again all terms are reduced modulo p. This shows that (n k + n) ∈ G
for all positive integers n.
Taking n = k, we see that (k 2k) ∈ G. With n = 2k, we have (2k 3k) ∈ G,
and in general, ((m − 1)k mk) ∈ G for all positive integers m.
But
(0 k)(k 2k)(0 k) = (0 2k),
so (0 2k) ∈ G. And
(0 2k)(2k 3k)(0 2k) = (0 3k),
so (0 3k) ∈ G. Continuing in this way, we see that (0 nk) ∈ G for all positive
integers n.
Since p is prime, k is relatively prime to p and therefore has a multiplicative
inverse mod p. Let a be this inverse, so that ak ≡ 1 (mod p). Then we see that
(0 ak) = (0 1) ∈ G.
Both (0 1) and (0 1 2 . . . p−1) belong to G, so by Exercise 3.5.4, G = Sp .
3.5.6 Exercise 6
Show that h(1 3), (1 2 3 4)i is a proper subgroup of S4 . What is the isomorphism
type of this subgroup?
Solution. Let σ = (1 2 3 4) and τ = (1 3). Note that σ 4 = τ 2 = 1. Moreover,
στ = (1 4)(2 3)
and
τ σ −1 = (1 3)(1 4 3 2) = (1 4)(2 3),
so στ = τ σ −1 and we see that σ and τ satisfy exactly the same relations in
S4 as r and s do in D8 . We can therefore define a surjective homomorphism
ϕ : D8 → hσ, τ i with
ϕ(r) = σ and ϕ(s) = τ.
This shows that hσ, τ i has at most 8 elements and is thus a proper subgroup of
S4 . However, it is easy to directly verify that ϕ maps distinct elements in D8
to distinct elements in hσ, τ i, so ϕ is an isomorphism and hσ, τ i ∼
= D8 .
3.5.7 Exercise 7
Prove that the group of rigid motions of a tetrahedron is isomorphic to A4 .
Proof. In Exercise 1.7.20, we showed that the group G of rigid motions of a

tetrahedron is isomorphic to a subgroup of S4 . We will show that this subgroup
has 12 elements and consists only of even permutations, so that it must be A4 .
We have shown in Exercise 1.2.9 that there are 12 elements in G. One of
these is the identity, eight more are rotations that fix one vertex and permute the
remaining three in a 3-cycle. A 3-cycle is an even permutation by Proposition 25,
so each of these elements in G correspond to elements in A4 .
Finally, the remaining three rotations in G are 180◦ rotations leaving no fixed
points, each transposing two pairs of vertices. A product of two transpositions
is an even permutation, so these elements of G also correspond to elements in
A4 . Therefore G ∼ = A4 .
3.5.8 Exercise 8
Prove the lattice of subgroups of A4 given in the text is correct.
Proof. We know from the previous exercise, Exercise 3.5.7, that A4 is isomorphic
to the group of rigid motions of a tetrahedron. By the discussion in Section 3.2
of the text, this group has no subgroup of order 6.
By Exercise 2.5.10, any subgroup of order 4 must be isomorphic to either
Z4 or V4 , where the latter is the Klein 4-group. But we know from the previous
exercise that the rotational symmetries of the tetrahedron each have order 1,
2, or 3, so A4 cannot have a cyclic subgroup of order 4. On the other hand,
it is not difficult to verify that h(1 2)(3 4), (1 3)(2 4)i ∼
= V4 , and this is the only
subgroup of order 4 since A4 only has three elements of order 2.
By Lagrange, the only remaining possibilities for nontrivial proper subgroups
are orders 2 and 3. These must be cyclic subgroups, and since A4 has three ele-
ments of order 2 and eight elements of order 3, we see that the cyclic subgroups
shown in the lattice are the only possibilities.
The indices shown on the lattice are also easy to verify using Lagrange’s
Theorem. Therefore the lattice in the text is correct.
3.5.9 Exercise 9
Prove that the (unique) subgroup of order 4 in A4 is normal and is isomorphic
to V4 .
Proof. In the solution to the previous exercise, Exercise 3.5.8, we showed that
A4 has exactly one subgroup of order 4, which is isomorphic to V4 . So we need
only show that this subgroup is normal.
In Exercise 3.1.26, we proved that for any positive integer n, a subgroup of
G generated by all elements of order n must be normal in G. The subgroup of
A4 having order 4 is generated by all elements of order 2 in A4 , namely
(1 2)(3 4), (1 3)(2 4), and (1 4)(2 3).
Therefore h(1 2)(3 4), (1 3)(2 4)i E A4 .

3.5.10 Exercise 10
Find a composition series for A4 . Deduce that A4 is solvable.
Solution. From the previous exercise, Exercise 3.5.9, the subgroup
h(1 2)(3 4), (1 3)(2 4)i
is normal in A4 and is isomorphic to V4 . As V4 is abelian, all of its subgroups

are normal. Therefore we may take
1 E h(1 2)(3 4)i E h(1 2)(3 4), (1 3)(2 4)i E A4
as a composition series for A4 . We know that the quotient groups are simple
since they each have prime order. They are also abelian for the same reason, so
A4 is solvable.
3.5.11 Exercise 11
Prove that S4 has no subgroup isomorphic to Q8 .
Proof. If there is such a subgroup, let it be H. Note that S4 has exactly 6
elements of order 4, namely all the 4-cycles. And Q8 has 6 elements of order 4,
so H must contain all 4-cycles in S4 . Then H must also contain both
(1 2 3 4)2 = (1 3)(2 4)
and
(1 2 4 3)2 = (1 4)(2 3),
two elements of order 2. Along with the identity, this implies that H has more
than 8 elements, which is impossible. Therefore no such subgroup exists.
3.5.12 Exercise 12
Prove that An contains a subgroup isomorphic to Sn−2 for each n ≥ 3.
Proof. Consider the following map ϕ : Sn−2 → An :
(
σ, if σ is even
ϕ(σ) =
σ ◦ (n − 1 n) if σ is odd.
Now let σ and τ be members of Sn−2 . We want to show first that
ϕ(σ)ϕ(τ ) = ϕ(στ ), (3.9)
so that ϕ is a homomorphism. Certainly (3.9) holds if σ and τ are both even.

If both are odd, then
σ ◦ (n − 1 n) ◦ τ ◦ (n − 1 n) = σ ◦ τ ◦ (n − 1 n)2 = σ ◦ τ,
and (3.9) holds since στ is even. Lastly, if exactly one of σ and τ is odd, then
στ is odd and
ϕ(σ)ϕ(τ ) = σ ◦ τ ◦ (n − 1 n) = ϕ(στ ).
In all cases, (3.9) holds and ϕ is a homomorphism.

Notice also that ϕ is injective, for if ϕ(σ) = ϕ(τ ), then either both σ and τ
are even, in which case σ = τ , or both are odd in which case multiplication by
(n − 1 n) again gives σ = τ .
Now ϕ(Sn−2 ) is the image of a homomorphism into An and is thus a sub-
group of An . Restricting the codomain of ϕ to this subgroup then gives an
isomorphism. This completes the proof.
3.5.13 Exercise 13
Prove that every element of order 2 in An is the square of an element of order
4 in Sn .
Proof. Let σ ∈ An with |σ| = 2. Any element of order 2 in An is a product of

an even number of disjoint transpositions. For each pair (a b) and (c d) of these
transpositions, we can write
(a b)(c d) = (a c b d)2 .
Therefore σ can be written as a product of squares of disjoint 4-cycles. Since

disjoint cycles commute, we can write σ as the square of an element of order 4
in Sn .
3.5.14 Exercise 14
Prove that the subgroup of A4 generated by any element of order 2 and any
element of order 3 is all of A4 .
Proof. This follows from the lattice for A4 which was proven correct in Exer-
cise 3.5.8. Let σ be any element of order 3 and τ any element of order 2 in A4 .
Then hσi is a proper subgroup of hσ, τ i. But we see from the lattice that hσi is
maximal. Therefore hσ, τ i = A4 .
3.5.15 Exercise 15
Prove that if x and y are distinct 3-cycles in S4 with x 6= y −1 , then the subgroup
of S4 generated by x and y is A4 .
Proof. As with the previous exercise, this follows from the lattice for A4 . Since
every 3-cycle is even, x and y both belong to A4 . Since x and y are distinct with
x 6= y −1 , they do not both belong to the same cyclic subgroup of A4 . Since hxi
is maximal in A4 (from the lattice), it follows that hx, yi = A4 .
3.5.16 Exercise 16
Let x and y be distinct 3-cycles in S5 with x 6= y −1 .
(a) Prove that if x and y fix a common element of {1, . . . , 5}, then hx, yi ∼
= A4 .
Proof. Suppose x and y fix the element i and let H be the subgroup of
S5 consisting of permutations which fix this same element i. Define the
bijection ϕ : {1, 2, 3, 4, 5} − {i} → {1, 2, 3, 4} by
(
j, if j < i,
ϕ(j) =
j − 1, if j > i.
Then H ∼
= S4 via the isomorphism ψ : H → S4 defined by
ψ(z) = ϕ ◦ z ◦ ϕ−1 .
Now let σ = ψ(x) and τ = ψ(y). Then σ and τ are distinct 3-cycles in S4
with σ 6= τ −1 . By Exercise 3.5.15, we have
hx, yi ∼
= hσ, τ i = A4 .
(b) Prove that if x and y do not fix a common element of {1, . . . , 5}, then
hx, yi = A5 .
Proof. Note that hx, yi ≤ A5 . Since x and y do not fix a common element,
they can be written in the form
x = (a b c) and y = (a d e).
Then
xy = (a d e b c),
so hx, yi contains 5-cycles. Hence hx, yi contains a subgroup of order 5.
But
xyx−1 = (a b c)(a d e)(a c b) = (b d e)
and
yxy −1 = (a d e)(a b c)(a e d) = (b c d),
so hx, yi contains the two 3-cycles (b d e) and (b c d). Neither of these is
the inverse of the other, so by the first part of this exercise, hx, yi contains
a subgroup that is isomorphic to A4 .
Since |A4 | = 12, we have shown that hx, yi contains subgroups of order 5
and order 12. By Lagrange’s Theorem, it follows that hx, yi has an order
of at least 60. Therefore hx, yi = A5 .
3.5.17 Exercise 17
If x and y are 3-cycles in Sn , prove that hx, yi is isomorphic to Z3 , A4 , A5 , or
Z3 × Z3 .
Proof. Let T be the set of elements in {1, 2, . . . , n} that are not fixed by x or
by y. Then 3 ≤ |T | ≤ 6. We consider each case in turn.
First, if |T | = 3 then either x = y or x = y −1 . Either way, hx, yi = hxi ∼
= Z3 .
Next, if |T | = 4 or 5 then, by relabeling, we can see that hx, yi is isomorphic
to a subgroup of S5 . In both cases, Exercise 3.5.16 shows that hx, yi ∼ = A4 or
hx, yi ∼
= A5 .
The last remaining case is |T | = 6. In this case x and y are disjoint and
therefore commute. Every element in hx, yi can then be written uniquely in the
form
xa y b , where 0 ≤ a ≤ 2 and 0 ≤ b ≤ 2.
Let z be a generator for Z3 and define the map ϕ : hx, yi → Z3 × Z3 by
ϕ(xa y b ) = (z a , z b ).
It is not difficult to check that ϕ is an isomorphism, so that hx, yi ∼

= Z3 × Z3 in
this case.
In each case, the subgroup hx, yi is isomorphic to one of Z3 , A4 , A5 , or
Z3 × Z3 .
Appendix A
Cartesian Products and

Zorn’s Lemma
A.1 Cartesian Products

A.1.1 Exercise 1
Let I and J be two indexing sets and let A be an arbitrary set. For any function
ϕ : J → I define
Y Y Y
ϕ∗ : A→ A by ϕ∗ (f ) = f ◦ ϕ for all choice functions f ∈ A.
i∈I j∈J i∈I
(a) Let I = {1, 2}, let J = {1, 2, 3} and let ϕ : J → I be defined by ϕ(1) = 2,
ϕ(2) = 2, and ϕ(3) = 1. Describe explicitly how an ordered pair in A × A
maps to a 3-tuple in A × A × A under this ϕ∗ .
Solution. Suppose f corresponds to the ordered pair (a, b) in A2 . Since ϕ

sends indices 1 and 2 in J to index 2 in I and index 3 in J to index 1 in
I, we see that ϕ∗ (f ) corresponds to the 3-tuple (b, b, a) in A3 .
(b) Let I = J = {1, 2, . . . , n} and assume ϕ is a permutation of I. Describe

in terms of n-tuples in A × A × · · · × A the function ϕ∗ .
Solution. ϕ∗ sends the n-tuple (a1 , a2 , . . . , an ) in An to the n-tuple
(aϕ(1) , aϕ(2) , . . . , aϕ(n) )
in An .
183
184 APPENDIX A. CARTESIAN PRODUCTS AND ZORN’S LEMMA
A.2 Partially Ordered Sets and Zorn’s Lemma

A.2.1 Exercise 1
Let A be the collection of all finite subsets of R ordered by inclusion. Discuss
the existence (or nonexistence) of upper bounds, minimal and maximal elements
(where minimal elements are defined analogously to maximal elements). Explain
why this is not a well ordering.
Solution. Subsets of A may or may not have upper bounds. For example, take
the set of all singleton sets containing integers,
X = {{1}, {2}, {3}, . . . }.
X is a subset of A but it does not have an upper bound since Z+ 6∈ A. Any

finite subset of A, however, will have an upper bound, namely the union of the
sets in the subset. So for example {∅, {1, 2}, {1, 3, 5}} ⊂ A has the upper bound
{1, 2, 3, 5} ∈ A.
A does not have any maximal elements, since given any Y in A, we can
simply append to the set Y any real number not already in Y , in order to
obtain a new finite subset of R containing Y . A does have a minimal element,
however, namely the empty set.
Set inclusion (⊆) is not a well ordering on A since it is not a total ordering.
That is, there exist elements X and Y of A such that X 6⊆ Y and Y 6⊆ X (for
an example, take X = {0} and Y = {1}).
A.2.2 Exercise 2
Let A be the collection of all infinite subsets of R ordered by inclusion. Discuss
the existence (or nonexistence) of upper bounds, minimal and maximal elements.
Explain why this is not a well ordering.
Solution. In this case, every subset of A has an upper bound since the union of
the sets in the subset is a member of A. A has one maximal element, R itself,
but no minimal elements since, given any set X ∈ A, we may pick some element
x in X, so that X − {x} is an infinite subset of R which is contained in X.
This ordering is not a well ordering for the same reason as in the previous
exercise: set inclusion ⊆ is not a total ordering on A. For example, take X to be
the set of even integers and take Y to be the set of odd integers. Then X 6⊆ Y
and Y 6⊆ X.
A.2.3 Exercise 3
Show that the following partial orderings on the given sets are not well orderings:
(a) R under the usual relation ≤.
(b) R+ under the usual relation ≤.
(c) R+ ∪ {0} under the usual relation ≤.
(d) Z under the usual relation ≤.

A.2. PARTIALLY ORDERED SETS AND ZORN’S LEMMA 185
Proof. In each case, ≤ is a total ordering. However, it is not a well ordering

since in each case there exist nonempty subsets which have no smallest element.
For R, R+ , and R+ ∪ {0} the interval (0, 1) is a subset of each but has no
smallest member. For Z, the entire set Z itself is a subset with no smallest
member.
A.2.4 Exercise 4
Show that Z+ is well ordered under the usual relation ≤.
Proof. Given any two positive integers m and n, we must have either m ≤ n,
n ≤ m, or both (if m = n). Therefore ≤ is a total ordering.
Let A be an arbitrary nonempty subset of Z+ . Pick an integer a ∈ A. Then
the set {1, 2, . . . , a} ∩ A is nonempty and finite. Being finite, it must have a
smallest member b. If c ∈ A is such that c ≤ b, then certainly c ≤ a (by
transitivity), so c ∈ {1, 2, . . . , a} ∩ A. Since b is the smallest member of this set,
we must have c = b.
We have shown that ≤ is a total ordering such that any nonempty subset
of Z+ has a smallest member. Therefore Z+ is well ordered under the relation
≤.

Selected Solutions To Dummit and Foote's Abstract Algebra Third Edition

Uploaded by

Copyright:

Available Formats

Selected Solutions To Dummit and Foote's Abstract Algebra Third Edition

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Selected Solutions To Dummit and Foote's Abstract Algebra Third Edition

Uploaded by

Copyright:

Available Formats

Selected Solutions to Dummit and Foote’s

Abstract Algebra Third Edition

July 16, 2020

Copyright © 2019–2020 Greg Kikola.

This document lives at:

3 Quotient Groups and Homomorphisms 123

A Cartesian Products and Zorn’s Lemma 183

Therefore, M P = P M if and only if r = 0 and p = s. Hence

0.2 Properties of the Integers

(a) a = 20, b = 13.

Solution. Applying the Division Algorithm repeatedly, we get

To write 1 as a linear combination of 20 and 13, we work backwards and

(b) a = 69, b = 372.

Solution. As above, we have

(c) a = 792, b = 275.

(d) a = 11 391, b = 5673.

Solution. Using the methods above, we get

−126(11 391) + 253(5673) = 3.

(e) a = 1761, b = 1567.

(f) a = 507885, b = 60808.

(507885, 60808) = 691,

−17(507885) + 142(60808) = 691.

Since ms + nt must be an integer (due to closure of integer addition and mul-

Proof. Substituting for x and y in ax + by gives

This holds regardless of the value of t, so (1) is always a valid solution.

and then by applying the formula given in the text:

For example, to find ϕ(18), we factor 18 = 2 · 32 . Applying the formula then

ϕ(18) = 21−1 (2 − 1) · 32−1 (3 − 1)

Applying this process to each n ≤ 30 produces the following table:

Theorem. If A is any nonempty subset of Z+ , there is some element m ∈ A

ϕ(n) = ϕ(q k )ϕ(r)

ϕ(n) = ϕ(pk ) = pk−1 (p − 1) and ϕ(d) = ϕ(p` ) = p`−1 (p − 1).

Now let a = pk−` . Then aϕ(d) = ϕ(n), so ϕ(d) | ϕ(n).

and suppose d is an integer that divides n. Then d can be written as a prod-

d = pβ1 1 pβ2 2 · · · pβkk with 0 ≤ βi ≤ αi for each i.

ϕ(n) = a1 ϕ(pβ1 1 ) · a2 ϕ(pβ2 2 ) · · · ak ϕ(pβkk )

0.3 Z/nZ: The Integers Modulo n

Solution. The residue classes are

0̄ = {0, 18, −18, 36, −36, . . .},

17 = {17, 35, −1, 53, −19, . . .}.

Now k ≡ r (mod n) and r is an integer between 0 and n − 1, so this shows that

Proof. a2 and b2 are each either congruent to 0 or to 1, modulo 4. Adding

0+0≡0 (mod 4),

In every case, a2 + b2 never has a remainder of 3 when divided by 4.

This immediately implies that c is even. Now, if a is even, then b must be

a2 + b2 = (2m + 1)2 + (2n + 1)2

This is impossible, so a, b, and c must all be even.

where α < a, β < b, and γ < c.

Proof. If a is an odd integer, then a can be written as 2k + 1 for some integer

Now k(k + 1) must be even, since it is the product of consecutive integers.

But then ax ≡ 1 (mod n), so x is the multiplicative inverse of a modulo n, i.e.,

Proof. Let ā and b̄ be in (Z/nZ)× as stated. Then ā has a multiplicative inverse

(āx̄)(b̄ȳ) ≡ 1 · 1 ≡ 1 (mod n).

Proof. Since (a, n) = 1, we may find integers c and d such that ac + nd = 1.

Solution. Applying the Euclidean algorithm gives

So (−3) = 17 is the multiplicative inverse of 13 in Z/20Z.

(c) a = 1891, n = 3797

Solution. ā has an inverse of 253.

(d) a = 6 003 722 857, n = 77 695 236 973

(a, b)(a−1 , b−1 ) = (a ? a−1 , b b−1 ) = (1, 1),