Solution to Algebra : Chapter 0 by Paolo Aluffi

mactonya1

Last updated at December 29, 2021

v0.5.3b

1 https://github.com/mactonya/
Copyright (C) 2020-2022 mactonya @ https://github.com/mactonya/.
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.3 or
any later version published by the Free Software Foundation; with the
Invariant Sections being Prologue. A copy of the license is included in
the section entitled ”GNU Free Documentatin License”.

Special thanks to the following contributers:

Prologue

Over a few months I want to improve my skills in solving algebra problems. I tried to find a
textbook that can serves me good and is good enough to use in self-study.
Eventually, this is what I felt the most “comfortable” book in my opinion. It doesn’t contain
that much unlike Dummit & Foote, but the writing style, the explanation, and the exercises really
served me well.
So here is the solution to Algebra : Chapter 0. There are a few important points to note here:

• The solution is only hosted on my GitHub page

https://github.com/mactonya/algebra-chapter-0-solutions.

If you find this document outside this page, you might have an outdated version of the solution
which might have errors, so please be aware.

• I will update the solution irregularly.

• I’ll try to write this beginner-friendly (as I am also a beginner), so the answer might be way
too detailed/verbose. Sorry if you find this annoying.

• If you found an error in the solutions, typos, bad grammar or want to give an advise on LaTeX
formatting, etc., don’t hesitate to open an issue or a pull request on my repo.

• PRs to new solutions are now open. You are free to write solutions for this document
now. When writing a solution, be sure to make it very beginner friendly: you should make all
logics, the use of theorems and the idea very clear. You are also very welcome to refine the
proofs if they are unclear.

Best,

mactonya @ https://github.com/mactonya/
Department of Mathematics, National Taiwan University
Updated December 29th, 2021

ii
Contents

Prologue ii

I Preliminaries: Set theory and categories 1

I.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
I.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
I.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
I.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
I.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

II Groups, first encounter 6

II.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
II.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
II.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
II.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
II.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
II.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
II.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
II.8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
II.9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

III Rings and modules 19

III.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
III.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
III.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
III.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
III.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
III.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
III.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

IV Groups, second encounter 38

IV.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
IV.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
IV.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
IV.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
IV.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
IV.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

V Irreducibility and factorization in integral domain 52

V.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
V.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
V.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

VI Linear Algebra 60

iii
CONTENTS iv

61
Chapter I

Preliminaries: Set theory and categories

Throughout this solution manual, we will use the same notation (and convention) as in the book,
with probably a little to none changes.
For your convenience, it is recommended to search your question via whatever your browser
provides (e.g. F3). The format of questions are Chapter (in roman).Section.Question.
In the following, categories are denoted using the Sans-serif font, e.g. Set.

I.1
Problem I.1.1. Locate a discussion of Russel’s paradox, and understand it.

Problem I.1.2. Prove that if ∼ is an equivalence relation on a set S, then the corresponding family
P∼ defined in §1.5 is indeed a partition of S.

Proof. The union of such class must contain S by definition, as at worse the elements can be in the
equivalence class formed by themselves. It suffices to check disjointness: If a ∈ [x], a ∈ [y] but x  y,
then transitivity implies x ∼ a, a ∼ y ⇒ x ∼ y, a contradiction. 

I.2
Problem I.2.1. How many different bijection are there between a set S with n elements and itself?

Solution. The first number has n choices; to make the map a bijection, the next number has only
(n − 1) choices remaining. By continuing choosing, we have n! different bijections. 

Problem I.2.5. Formulate a notion of epimorphism, in the style of the notion of monomorphism,
and prove a result analogous to Proposition 2.3, for epimorphisms and surjections.

Solution. Epimorphism are right-cancelable; that is,

A function f : A → B is a epimorphism if for all sets Z and all functions β, β 0 : Z → A,

β ◦ f = β 0 ◦ f =⇒ β = β 0 .

We shall prove the following:

Proposition. A function is surjective if and only if it is an epimorphism.
Proof.
(⇒) Let f be surjective. By Proposition I.2.1, a surjective function has a right-inverse, which we
call it g. Then if β, β 0 : B → Z are arbitrary function such that β ◦ f = β 0 ◦ f , then by composition
with g we obtain

(β ◦ f ) ◦ g = (β 0 ◦ f ) ◦ g ⇒ β ◦ (f ◦ g) = β 0 ◦ (f ◦ g) ⇒ β ◦ idA = β 0 ◦ idA ⇒ β = β 0

1
CHAPTER I. PRELIMINARIES: SET THEORY AND CATEGORIES 2

as desired.
(⇐) Let f be an epimorphism. We need to consider some special β : B → Z so we can prove the
assertion. We done this by ”labeling”: define
(
1, b ∈ im f
β(b) = , β 0 (b) = 1
0, b ∈
/ im f

Then since
β ◦ f = β0 ◦ f ⇒ β = β0
this implies that beta receives only values in im f , so im f ⊇ B. Since we have im f ⊆ B clearly for
any function f , we conclude that im f = B, which is the definition of surjectivity. 

I.3
Problem I.3.1. Let C be a category. Consider a structure Cop with

• Obj(Cop ) = Obj(C);
• for A, B objects of Cop , HomCop (A, B) := HomC (B, A).

Show how to make this into a category.

Solution. For f ∈ HomCop (A, B), g ∈ HomCop (B, C), define the composite of morphisms by

g ◦ f := f g

where f g is defined in the sense of the category C. Now we check the definition of category:

• 1A exists as HomCop (A, A) := HomC (A, A) 3 1A ;

• The composition works as intended: the map on the right is a morphism from C to A;
• The composite law is checked as

(h ◦ g) ◦ f = gh ◦ f = f (gh) = (f g)h = h ◦ f g = h ◦ (g ◦ f );

• Identity morphism work as intended:

1A ◦ f = f 1A = f, f ◦ 1A = 1A f = f.

Problem I.3.11. Draw the relevant diagrams and define composition and identities for the category
CA,B mentioned in Example 3.9. Do the same for the category Cα,β mentioned in Example 3.10.

Solution. By reversing the arrow of CA,B , we obtain:

• Objects of this category are diagrams

A
f

Z
g
B
CHAPTER I. PRELIMINARIES: SET THEORY AND CATEGORIES 3

• morphisms are
A f1
A f2

Z1 −→ Z2
g1 g2
B B
which are commutative diagrams
f1
A f2

Z1 σ
Z2 .
g2
B g1

For the case Cα,β :

• Objects are diagrams
A
α f

C Z
β g
B
• morphisms are
A f1
A f2
α α

C Z1 −→ C Z2
β g1 β g2
B B
which are commutative diagrams
f2
A f1
α

C Z1 σ
Z2 .
β g1
B g2

composition and identity are defined analogously as in Example 3.5. 

I.4
Problem I.4.3. Let A, B be objects of a category C, and let f ∈ HomC (A, B) be a morphism.
• Prove that if f has a right-inverse, then f is an epimorphism.
• Show that the converse does not hold, by giving an explicit example of a category and an
epimorphism without a right-inverse.
Proof. Let g be the right inverse of f , i.e. f g = 1. Then for any morphism h, h0 ∈ HomC (B, Z),
h ◦ f = h0 ◦ f ⇒ h ◦ f ◦ g = h0 ◦ f ◦ g ⇒ h ◦ 1 = h0 ◦ 1 ⇒ h = h0
showing that f is an epimorphism. For a counterexample in which the converse does not hold,
consider C = Z, objects are integers, and morphisms are the relation ≤ (c.f. p.p.27). Then
f :1→2
is an epimorphism, but there are no right inverse for f , since there are no morphisms in HomC (2, 1).

CHAPTER I. PRELIMINARIES: SET THEORY AND CATEGORIES 4

I.5
Problem I.5.1. Prove that a final object in a category C is initial in the opposite category Cop
(I.3.1).

Proof. Let F be a final object in C, which means that the set HomC (A, F ) is a singleton for all
A ∈ Obj(C). Since
HomC (A, F ) = HomCop (F, A)
we have that F is initial in Cop . 

Problem I.5.2. Prove that ∅ is the unique initial object in Set.

Proof. Let A be another initial object. By definition, there is only one function that maps ∅ to A:
it’s the empty function. Note that empty function 0 : ∅ → A does make sense: recall the definition
of function from X to Y is that for all x ∈ X, there exists a unique y ∈ Y such that f (x) = y. In
order to let 0 fails to be a function, we want to find x ∈ ∅ such that there exist y, y 0 ∈ A such that
y 0 = 0(x) = y. But as there is no element in ∅, you can’t even find x to begin with, so 0 is indeed a
function.
Now to continue, since A by assumption is also initial, there exists a unique function that sends
elements of A to ∅. But such function cannot exist unless A = ∅: for all x ∈ A, there does not
exist a unique y ∈ ∅ such that f (x) = y: there is nothing to begin with finding a y as we can’t find
anything in an empty set! 

Problem I.5.12. Define the notions of fibered products and fibered coproducts, as terminal objects
of the categories Cα,β , Cα,β considered in Example 3.10 (cf. also I.3.11), by stating carefully the
corresponding universal properties.
As it happens, Set has both fibered products and fibered coproducts. Define these objects
’concretely’, in terms of naive set theory.

Solution. Fibered product is final in Cα,β ; that is, there are only one morphism in
 
 fa A A
ia
α α

 
Hom  Z C , F
 
 C 

 
fb β ib β
B B

for any choice of the triple (Z, fa , fb ). Expand this to a diagram leads to the following universal
property:
The triple (F, ia : F → A, ib : F → B) is universal in the sense that for every triple (Z, fa : Z →
A, fb : Z → B), there exists a unique morphism ϕ : Z → F such that the diagram

fa
Z
∃!ϕ

F ia
A
fb
ib α

B β
C

commutes. Fibered product are also called pullback.

Fibered coproduct is initial in Cα,β . Following the same argument as above, we have the following
universal property:
CHAPTER I. PRELIMINARIES: SET THEORY AND CATEGORIES 5

The triple (I, iA : A → I, iB : B → I) is universal in the sense that for every triple (Z, fA : A →
Z, fB : B → Z), there exists a unique morphism ϕ : I → Z such that the diagram
α
C A
β iA
fA
iB
B I
∃!ϕ

fB
Z
commutes. Fibered coproduct are also called pushout.
Set has fibered products: Let us define
A ×C B := I = {(a, b) : a ∈ A, b ∈ B, α(a) = β(b)}
with projections ia , ib . We check that this satisfy the universal property: define
ϕ(z) := (fa (z), fb (z))
we check:
• ib ϕ = fb (resp. ia ϕ = fa ):
ib ϕ(z) = ib (fa (z), fb (z)) = fb (z)
• αia = βib :
!
αia (a, b) = α(a) = β(b) = βib (a, b).
note that ! is true since I guarantees the existence of b.
Set also has fibered coproducts, but it’s more complicated. We first define an equivalence relation:
define
R = {(α(x), 0) ∼ (β(x), 1) : x ∈ C}
This gives an equivalence relation on A q B, which gives a new structure I = (A q B)/ ∼. Let
iA (a) = (a, 0), iB (b) = (b, 1), then it is direct that iB β = iA α. Now we define
(
fA (x) if c = 0
ϕ[i = (x, c)] =
fB (x) if c = 1
We need to check that it is well-defined, then it is direct that ϕβ = fB (resp. ϕα = fA ), proving the
universal property. There are two cases to consider:
• Case [(a, 0)] = [(a0 , 0)] (resp. [(b, 1)] = [(b0 , 1)]): If there are relations
a = α(x) ∼ β(x) = β(x0 ) ∼ α(x0 ) = a0
then they evaluated to the same value since
ϕ[(a, 0)] = ϕiA (a) = ϕiA (α(x)) = ϕiB (β(x)) = ϕiB (β(x0 )) = ϕiA (α(x0 )) = ϕiA (a0 ) = ϕ[(a0 , 0)]
• Case [(a, 0)] = [(b, 1)]: If there are relations
a = α(x) ∼ β(x) = b
then
ϕ[(a, 0)] = ϕiA (a) = ϕiA (α(x)) = ϕiB (β(x)) = ϕiB (b) = ϕ[(b, 1)]
as desired.
By the above analysis, as all elements in the same equivalence class connects to the other by some
chain
a = α(x1 ) ∼ β(x1 ) = β(x2 ) ∼ α(x2 ) = α(x3 ) · · · = b,
and since every ∼ preserves the result, ϕ is well-defined. 
Chapter II

Groups, first encounter

Unless otherwise specified, in the following G denotes a group, e denotes the identity of G. Some
description and hints are omitted for simplicity.

II.1
Problem II.1.3. Prove that (gh)−1 = h−1 g −1 for all elements g, h of a group G.

Proof. We knew that (gh)−1 (gh) = e, so multiply both sides by h−1 g −1 gives the desired equation. 

Problem II.1.4. Suppose that g 2 = e for all elements g of a group G; prove that G is commutative.

Proof. For any g, h ∈ G we can form a new element gh; this element is still in G, so it must
hold that (gh)2 = ghgh = e. As g 2 = h2 = e, multiply both sides by hg gives the commutative
requirement. 

Problem II.1.7. Prove Corollary 1.11: Let g be an element of finite order, and let N ∈ Z. Then
g N = e if and only if N is a multiple of |g|.

Proof. If g N = e, by Lemma 1.10 this implies |g| | N , which is equivalent to N is a multiple of |g|.
Conversely if N = k|g| for some k ∈ N, then g N = g k|g| = ek = e. 

Problem
Q II.1.8. Let G be a finite abelian group with exactly one element f of order 2. Prove that
g∈G g = f .

Proof. For all elements

Q that is not of order 2, they have an inverse that is not itself, so they canceled
out in the product g∈G g, leaving only elements that is of order 2, i.e. f . 

Problem II.1.10. If the order of g is odd, what can you say about the order of g 2 ?

Solution. The order of g 2 is |g| since the only number that divides |g| and in {2, 4, ..., 2|g|} is 2|g| if
|g| is odd. 

Problem II.1.11. Prove that for all g, h in a group G, |gh| = |hg|.

Proof. Simply observe that e = (gh)|gh| = g(hg)(|gh|−1) h, therefore

g −1 h−1 = (hg)−1 = (hg)|gh|−1

hence (hg)|gh| = e. The other case ((gh)|hg| = e) is similar. 

Problem II.1.13. Give an example showing that |gh| =

6 lcm(|g|, |h|) even if g and h commute.

Solution. In C4 , |1 + 3| = |0| = 1 but lcm(|1|, |3|) = 4. Clearly C4 is abelian. 

6
CHAPTER II. GROUPS, FIRST ENCOUNTER 7

Problem II.1.14. As a counterpoint of II.1.13, prove that if g and h commute and gcd(|g|, |h|) = 1,
then |gh| = |g||h|.
Proof. One has |gh| divides lcm(|g|, |h|) = |g||h| by Proposition II.1.14, so it suffices to prove that
|g||h| divides |gh|. Let N = |gh|. By noting that (gh)N = g N hN since g and h commutes, we have
(gh)N |h| = e|h| = g N |h| hN |h| = g N |h|
so |g| divides N |h|, which implies |g| divides N since gcd(|g|, |h|) = 1. Similarly |h| divides N ,
therefore |g||h| divides N = |gh|, as desired. 
Problem II.1.15. Let G be a commutative group, and let g ∈ G be an element of maximal finite
order. Prove that if h has finite order in G, then |h| divides |g|.
Proof. Suppose that |h| does not divide |g|, then we can assume that |g| = pm r, |h| = pn s, where p
is a prime, r, s relatively prime to p and m < n. Since |h| does not divide |g|, gcd(h, g) = 1. Then
m
by II.1.14 we can calculate the order of g p hs , which is pn r. But this element has order bigger than
g, which contradicts to the maximality of g. Hence |h| must divide |g|. 

II.2
Problem II.2.5. Describe generators and relations for all dihedral groups D2n .
Solution. There are two elements: r, which is rotation, and s is the reflection. We would except that
• the n-gon preserves vertices after n rotations, i.e. rn = e;
• after two consecutive reflections, the vertices will return to before reflections, i.e. s2 = e;
• doing (reflection→rotation) twice would return to the original position, i.e. (sr)2 = e, or
srs = r−1 . Note that this also says that srk s = r−k .
Then we can write every element in D2n as sa rb where 0 ≤ s ≤ 1, 0 ≤ r < n. To show that they
really determine D2n , let a, b ∈ D2n , and consider their product. If a = sa1 ra2 , b = sb1 rb2 , then
ab = sa1 ra2 sb1 rb2
if we assume that b1 = 1 (or it would just be sa1 ra2 +b2 ),then
sa1 ra2 srb2 = sa1 s(sra2 s)rb2 = sa1 sr−a2 rb2 = sa1 +1 r−a2 +b2
so ab is an element of D2n . This extends to (in)finite products, so this relations determines D2n . 
Problem II.2.10. Prove that Z/nZ consists of precisely n elements.
Proof. The elements are [0]n , [1]n , . . . , [n − 1]n , and notice that [n]n = [0]n , [n + 1]n = [1]n , and so on.
No two elements listed above are the same by a simple check. 
Problem II.2.14. Show that the multiplication in Z/nZ is a well-defined action.
Proof. If a ≡ a0 mod n and b ≡ b0 mod n, then a = a0 + kn, b = b0 + ln for k, l ∈ Z, therefore
(ab) − (a0 b0 ) = (a0 + kn)(b0 + ln) − a0 b0 = a0 ln + b0 kn + kln2 ≡ 0 mod n
as desired. 
Problem II.2.16. Find the last digit of 123823718238456 .
Solution. 123823718238456 ≡ 718238456 = 499119228 = 24014559614 ≡ 14559614 = 1 mod 10. 
Problem II.2.17. Show that if m ≡ m0 mod n, then gcd(m, n) = 1 if and only if gcd(m0 , n) = 1.
Proof. We can write m = nk + m0 for n ∈ Z and use Euclidean Algorithm to conclude. 
CHAPTER II. GROUPS, FIRST ENCOUNTER 8

II.3
Problem II.3.1. Let ϕ : G → H be a morphism in a category C with products. Explain why there
is a unique morphism (ϕ × ϕ) : G × G → H × H compatible in the evident way with the natural
projections.

Solution. The compatibility of (ϕ × ϕ) comes from the commutative diagram

ϕ
G H
π2 ρ2
∃!(ϕ×ϕ)
G×G H ×H
π1 ρ1
ϕ
G H

which is easy to check. The uniqueness follows from the universal property of products that there is
a unique homomorphism such that the diagram

∃!(ϕ×ϕ)
G×G H ×H

commutes. 

Problem II.3.3. Show that if G, H are abelian groups, then G × H satisfies the universal property
for coproducts in Ab.

Proof. Let A be an arbitrary abelian group, fG , fH be homomorphisms, iG , iH be inclusions. We are

required to prove the commutativity of the diagram

G fG

iG
∃!ϕ
G×H A
iH

H fH

To check the universal property, define ϕ(g, h) := fG (g)fH (h). It is direct that the diagram com-
mutes. Finally, ϕ is a homomorphism since for g1 , g2 ∈ G, h1 , h2 ∈ H,

ϕ((g1 , h1 )(g2 , h2 )) = ϕ(g1 g2 , h1 h2 ) = fG (g1 g2 )fH (h1 h2 ) = fG (g1 )fG (g2 )fH (h1 )fH (h2 )
abelian
====== fG (g1 )fH (h1 )fG (g2 )fH (h2 ) = ϕ(g1 , h1 )ϕ(g2 , h2 )

as desired. 

Problem II.3.6. Consider the product C2 × C3 , which is a coproduct in Ab. Show that it is not a
coproduct of C2 and C3 in Grp.
CHAPTER II. GROUPS, FIRST ENCOUNTER 9

Proof. If C2 × C3 is a coproduct, then take A = S3 . Although there are injective homomorphisms

ϕ1 : C2 → S3 by ϕ1 (1) = (12) or other two cycle

ϕ2 : C3 → S3 by ϕ2 (1) = (123) or other three cycle

but there are no homomorphisms ϕ : C2 ×C3 → S3 that satisfies the universal property of coproducts:
Observe that any choice of cycles in ϕ1 and ϕ2 will exhaust all possible element of S3 , hence forces
ϕ to be an isomorphism. But the element ϕ(1, 1) must be either a 2(or 3)-cycle (i.e. ϕ2 (1, 1) (or
ϕ3 (1, 1)) is zero), and neither (1, 1)2 nor (1, 1)3 are (0, 0), and ϕ will map a non-identity element
to the identity, a contradiction (since ϕ is an isomorphism and must map only (0, 0) to the trivial
cycle). 
Problem II.3.8. Define a group G with two generators x, y, subject to the relations x2 = e, y 3 = e.
Prove that G is a coproduct of C2 and C3 in Grp.
Proof.
C2 f

i1
∃!ϕ
G A
i2

g
C3

To define ϕ properly so that the diagram commutes, we have no choice but to define

ϕ(x) = f (x), ϕ(y) = g(y)

Commutativity and uniqueness is immediate, so it remains to show that ϕ is a group homomorphism,

which is immediate if you just explicitly list two items g, h ∈ G and evaluate ϕ(g)ϕ(h) and ϕ(gh). 

II.4
Problem II.4.3. Prove that a group of order n is isomorphic to Z/nZ if and only if it contains an
element of order n.
Proof. Let G be such group.
(⇒) Trivial: 1̄ has order n.
(⇐) Let g be an element of order n. Then consider a homomorphism ϕ : G → Z/nZ with ϕ(g) = 1̄.
This map is injective: if it weren’t, then there are some integers 0 ≤ i < j < n such that ϕ(g i ) =
ϕ(g j ). But then ϕ(g j−i ) = 0̄, which says g j−i = eG , so g has order less than n, a contradiction. 
Problem II.4.7. Let G be a group. Prove that the function G → G defined by g 7→ g −1 is a
homomorphism if and only if G is abelian. Prove that g 7→ g 2 is a homomorphism if and only if G
is abelian.
Proof. Let g, h ∈ G. If ϕ : g 7→ g −1 is a homomorphism, then it must satisfy

g −1 h−1 = ϕ(g)ϕ(h) = ϕ(gh) = (gh)−1 = h−1 g −1

take inverse to both sides shows that G is abelian. If ϕ : g 7→ g 2 is a homomorphism, then it must
satisfy
gghh = g 2 h2 = ϕ(g)ϕ(h) = ϕ(gh) = (gh)2 = ghgh
multiply by g −1 on the left and h−1 on the right shows that G is abelian. The converse to both
statements are easy: we can simply rearrange the terms on the two sides of above equations to
”achive homomorphism” by the commutativity of G. 
CHAPTER II. GROUPS, FIRST ENCOUNTER 10

Problem II.4.8. Let g ∈ G. Prove that the function γg : G → G defined by γg (a) = gag −1 is an
automorphism of G. Prove that the function G → Aut(G) defined by g → γg is a homomorphism,
and show that this homomorphism is trivial if and only if G is abelian.

Proof. γg is injective since if gag −1 = gbg −1 then a = b; it is surjective since for k ∈ G we can let
g −1 kg so that γg (g −1 kg) = k; it is a homomorphism since

γg (ab) = gabg −1 = gag −1 gbg −1 = γg (a)γg (b).

If G is abelian then the automorphism is simply γg (a) = a; conversely if gag −1 = a then ga = ag

for all a, g ∈ G, hence abelian. 

Problem II.4.9. Prove that if m, n are positive integers such that gcd(m, n) = 1, then Cmn ∼
=
Cm × Cn .

Proof.
ϕ : Cmn → Cm × Cn , ϕ(a) = (a mod m, a mod n)
is a homomorphism and a bijection, as one can check directly. 

Problem II.4.11. Assuming the fact that the equation xd = 1 can have at most d solutions in Z/pZ
for a prime p, prove that (Z/pZ)∗ is cyclic.

Proof. Let g be an element of maximal order, and by II.1.15, all elements have degree that divides
|g|, i.e. |h||g| = 1 for all h ∈ G. Using the fact, we have |G| ≤ |d|, since only at most |g| elements can
be the solution to h|g| = 1. Clearly we also have |G| ≥ |d|, so |G| = |d|. Thus the proof is complete
by II.4.3. 

Problem II.4.13. Prove that AutGrp (Z/2Z × Z/2Z) ∼

= S3 .
Proof. To make an automorphism ϕ, ϕ must fix (0, 0), leaving 6 possible permutations for ele-
ments (0, 1), (1, 0), (1, 1). It suffices to check that all permutations of these elements are homomor-
phisms(hence isomorphisms). 

Problem II.4.14. Prove that the order of the group of automorphisms of a cyclic group Cn is the
number of positive integers r ≤ n that are relatively prime to n (cf. II.6.14).

Proof. We shall first show that every endomorphism of cyclic group C is of form ϕn (x) = xn for
some n. Indeed, if σ is a endomomorphism that σ(x) = xa = ϕa (x), then for every xb ∈ C we have

σ(xb ) = σ(x)b = (xa )b = (xb )a = ϕ(xb )

so every endomorphism is of form ϕn : x 7→ xn for some n. Now to make this into an automorphism,
if k is not relatively prime to n, say gcd(n, k) = r > 1, then for a generator x ∈ Cn , we have

ϕk (xn/r ) = xn/r·k = xn·k/r = (xn )k/r = ek/r = e

and since n/r is not n, ϕk maps a non-identity element to e, in which it is already mapped by e ∈ Cn ,
so ϕk fails to be a bijection. Therefore the order of Aut(Cn ) is the number of positive integers that
is relatively prime to n. 

Problem II.4.16. Prove the Wilson’s theorem: for p ∈ N>1 , p is a prime if and only if

(p − 1)! ≡ −1 mod p
CHAPTER II. GROUPS, FIRST ENCOUNTER 11

Proof. (⇒) Assuming that the result of II.1.8 and II.4.11 is true, consider G = (Z/nZ)∗ . It is cyclic,
and has exactly one element of order 2 since for 0 ≤ k ≤ p − 2,
(p − 1 − k)2 ≡ 1 + 2k + k 2 ≡ 1 mod p ⇐⇒ k(k + 2) ≡ 0 mod p
and such solution can only be k = 0 or p − 2 since p is a prime, which correspond to p − 1 and 1
(identity). Therefore by II.1.8
Y
g = (p − 1)! ≡ (p − 1) ≡ −1 mod p
g∈G

as desired.
(⇐) If p is not a prime, then there exists 1 < k < p such that k|p. Since k < p we have k|(p − 1)!,
i.e.
(p − 1)! ≡ rk mod p for some r ∈ Z
and clearly no choice of r will make rk ≡ −1 mod p by the fact that k|p. Therefore p must be a
prime. 

II.5
Problem II.5.3. Use the universal property of free groups to prove that the map j : A → F (A) is
injective.
Proof. If there is a, b ∈ A such that j(a) = j(b) but a 6= b, then let f be a set function such that
f (a) 6= f (b); in particular, let G = Z and let f (a) = 1, f (b) = 2. Then there are no homomorphisms
that will make the diagram commute, therefore j must be injective. 
Problem II.5.6. Prove that the group F ({x, y}) is a coproduct Z ∗ Z of Z by itself in the category
Grp.
Proof. We are given the universal property of free group: for j : {x, y} → F ({x, y}), ∃G, f such that
the diagram
∃!ϕ
F ({x, y}) G
j
f

{x, y}
commutes. To check that it is a coproduct, consider the coproduct diagram composed with above.
Let i(0) = x, j be the inclusion, then we have the following diagram:

Z f
γ i

j ∃!ϕ
{x, y} F ({x, y}) G

γ i
Z g

Note that the arrows j, h, ϕ comes from the free group diagram. From this, we have f ◦ γ = ϕ ◦ j. To
check the coproduct diagram commutes, it suffices to check f = ϕ ◦ i (the case g = ϕ ◦ i is identical).
To do this, define γ(x) = 0, γ(y) = 1. Then
f ◦ γ(x) = f (0) = ϕ(x) = ϕ ◦ j(x), f ◦ γ(y) = f (1) = ϕ(y) = ϕ ◦ j(y)
Since f (1) = ϕ ◦ i(1) = ϕ(y), the homomorphisms agree on the generator, hence are the same. 
CHAPTER II. GROUPS, FIRST ENCOUNTER 12

II.6
Problem II.6.5. Let G be a commutative group, and let n > 0 be an integer. Prove that {g n : g ∈
G} is a subgroup of G. Prove that this is not necessarily the case if G is not commutative.
Proof. For any two elements a, b in the set, they can be represented as g n and hn respectively. Now

ab−1 = g n h−n = (gh−1 )n

which shows that ab−1 is also in the set, proving the set is a subgroup. A counterexample would be
D6 , the dihedral group with 6 elements, with the choice n = 3. Let s denote the reflection, r denotes
the rotation, we then have

{g 3 : g ∈ D3 } = {1, r3 , r2·3 , s3 , (sr)3 , (sr2 )3 } = {1, 1, 1, s, sr, sr2 }

this set is not a subgroup, as s−1 sr = r is not an element of this set. 

Problem II.6.7. Show that inner automorphisms (the collection of γg in II.4.8) form a subgroup
Inn(G) of Aut(G), and show that Inn(G) is cyclic if and only if Inn(G) is trivial if and only if G is
abelian. Deduce that if Aut(G) is cyclic, then G is abelian.
Proof. Inn(G) is a subgroup since

γg ◦ γh−1 = gh−1 ahg −1 = (gh−1 )a(gh−1 )−1 ∈ Inn(G).

If Inn(G) is cyclic, then let γg (a) = gag −1 be a generator of order n. Then for any b ∈ G,
we have γb (x) = γgn (x), for some integer n. Then by plug in b into the homomorphism, we have
gbg −1 = bn bb−n . This gives gb = bg ∀b ∈ G, so γg is in fact trivial. Since the generator is trivial,
we conclude that Inn(G) is trivial. If Inn(G) is trivial, then the function given in II.4.8 can only be
the trivial map, so G is abelian by II.4.8. Finally, if G is abelian, then all inner automorphisms are
trivial, and clearly trivial group is cyclic.
The last statement follows from Proposition II.6.11 that every subgroup of cyclic group is cyclic.

Problem II.6.8. Prove that an abelian group G is finitely generated if and only if there is a surjective
homomorphism
Z ⊕ ··· ⊕ Z  G
| {z }
n times

for some n.
Proof. (⇒) As the group is abelian, for G = ha1 , · · · an i, we can represent an element g uniquely as

g = ap11 · · · apnn

where pi ∈ Z, i = 1, · · · n. Therefore we can explicitly write down the surjective homomorphism

ϕ : Z ⊕ ··· ⊕ Z  G by ϕ(p1 , · · · , pn ) = ap11 · · · apnn = g

| {z }
n times

as desired.
(⇐) By the universal property of Z⊕n we have the following diagram that commutes:
∃!ϕ
Z⊕n G
j (*)
f

{1, · · · , n}
CHAPTER II. GROUPS, FIRST ENCOUNTER 13

To prove, it suffices to ”replace” the set {1, · · · , n} by a subset of G.

∃!ϕ
Z⊕n G
j
j̃
i
f
{1, · · · , n} A

By the diagram (∗ ), we have i ◦ f = ϕ ◦ j. It is a fast check that the diagram formed by j̃, i and
ϕ commutes. Finally since A is a finite set and im ϕ = G, it follows by definition that G is finitely
generated. 
Problem II.6.14. Let φ be the Euler’s φ-function. Prove that for n ∈ N,
X
φ(m) = n.
m>0,m|n

Proof. Let hxi = Cn . We have the trivial equation

X
1=n
g∈Cn

Now note that every element in Cn generates a cyclic subgroup. To establish the result, we show
that for every d > 0 that is a divisor of n, the subgroup of order d is unique, i.e. the unique subgroup
is given by
hxn/d i = {g ∈ G : g d = 1}
Indeed, if g = xkn/d for some positive integer k, then g d = xkn = 1. Conversely, if g d = 1, then we
have g = xm for some m since x is a generator. But this means that xmd = 1, and this implies n|md.
Hence we have
g = xm = xn/d·dm/n = xn/d ∈ hxn/d i
as desired.
Now we count the generators of each subgroup of Cn , which is φ(d) for every d that is a divisor
of n. Since every element in Cn generates a cyclic subgroup Cd , the sum of generator along each
subgroup is exactly n, namely X X
1= φ(m) = n
g∈Cn m:m|n

which proved the assertion. 

Problem II.6.15. Prove that if ϕ : G → G0 has a left inverse, then ϕ is a monomorphism.
Proof. If a, b ∈ G are distinct elements that satisfies ϕ(a) = ϕ(b), then having left inverse means
there exists a homomorphism ψ such that ψ ◦ ϕ = idG . Then we would have ψ ◦ ϕ(a) = ψ ◦ ϕ(b),
which means a = b, a contradiction. 

II.7
Problem II.7.2. Is the image of a group homomorphism necessarily a normal subgroup of the
target?
Solution. Well no: the image of ϕ : C2 → S3 defined by ϕ(1) = (1 2) is im ϕ = {e, (1 2)}, which is
not normal. 
Problem II.7.3. Verify that the equivalent conditions for normality given in §7.1 are indeed equiv-
alent.
CHAPTER II. GROUPS, FIRST ENCOUNTER 14

Proof. Let g ∈ G be fixed.

• (gng −1 ∈ N ⇒ gN g −1 ⊆ N ) is clear.
• (gN g −1 ⊆ N ⇒ gN g −1 = N ): For n ∈ N , there is an element g −1 ng ∈ N by normality, so
g(g −1 ng)g −1 = n, showing that gN g −1 ⊇ N .
• (gN g −1 = N ⇒ gN ⊆ N g): For h ∈ gN , there is h = gn for some n ∈ N . By normality of N ,
there is some n0 ∈ N such that gng −1 = n0 , or gn = n0 g. Hence h = n0 g, therefore h ∈ N g.
• (gN ⊆ N g ⇒ gN = N g): If gN ⊆ N g, then we also have g −1 N ⊆ N g −1 , which is N g ⊆ gN .
• (gN = N g ⇒ gng −1 ∈ N ): If gn = n0 g, then gng −1 = n0 . Since N is a subgroup, gng −1 ∈ N .

Problem II.7.7. Let n be a positive integer. Let H ⊂ G be the subgroup generated by all elements
of order n in G. Prove that H is normal.

Proof. For a ∈ H, g ∈ G, since an = e,

(gag −1 )n = gan g −1 = e

we have gag −1 ∈ H, hence normal. 

Problem II.7.8. Prove Proposition 7.6: If H is any subgroup of a group G, the relation ∼L defined
by a ∼L b ⇔ a−1 b ∈ H is an equivalence relation.

Proof.

• It is reflexive: for any g ∈ G, g −1 g = e ∈ H, so it is always true that g ∼L g;

• It is symmetric: for g, h ∈ G in which g ∼L h, we have g −1 h ∈ H, and its inverse, which is

h−1 g must also be in G, which implies h ∼L g;

• It is transitive: for a, b, c ∈ G in which a ∼L b, b ∼L c, we have a−1 b ∈ H and b−1 c ∈ H. This

implies a−1 c ∈ H, so a ∼L c.

Problem II.7.10. Let G be a group, and H ⊆ G be a subgroup. Show that H is normal in G if

and only if for all γ ∈ Inn(G) (II.6.7), γ(H) ⊆ H.

Proof. One side is trivial: if H is normal then for all g ∈ G, γg (h) = ghg −1 ∈ H for all h ∈ H. The
other side is also trivial because gHg −1 ⊆ H implies normality for H. 

Problem II.7.11. Prove that the commutator subgroup [G, G] is normal, and the quotient G/[G, G]
is commutative.

Proof. Observe

gaba−1 b−1 g −1 = (gag −1 )(gbg −1 )(ga−1 g −1 )(gb−1 g −1 ) = xyx−1 y −1 ∈ [G, G]

for x = gag −1 , y = gbg −1 . The quotient is commutative since aba−1 b−1 [G, G] = [G, G] implies
ab[G, G] = ba[G, G]. 

Problem II.7.12. Let F = F (A) be a free group, and let f : A → G be a set-function from the set
A to a commutative group G. Prove that f induces a unique homomorphism F/[F, F ] → G, where
[F, F ] is the commutator subgroup of F defined in Exercise 7.11. Conclude that F/[F, F ] ∼
= F ab (A).
CHAPTER II. GROUPS, FIRST ENCOUNTER 15

Proof. We need to define a proper homomorphism f˜ : F/[F, F ] → G. By the universal property

of free group, we have a unique homomorphism ϕ : F → G induced from f . Now observe that for
g, h ∈ A,
ϕ(g)ϕ(h)ϕ(g)−1 ϕ(h)−1 = ϕ(ghg −1 h−1 ) = e
as G is commutative, we know that ϕ vanish on [F, F ]. Now we just define

f˜ : F/[F, F ] → G by f˜(x[F, F ]) = ϕ(x).

It is a fast check that f˜ is the required homomorphism. This gives the following diagram.

F/[F, F ]
π ∃!f˜

∃!ϕ
F ab (A) G
f
j

Since both triangles commutes, the ”triangle” formed by the edges π ◦ j, f and f˜ also commutes. By
general nonsense (Proposition I.5.4), we conclude that F/[F, F ] ∼
= F ab (A). 

II.8
Problem II.8.2. Extend Example 8.6 as follows. Suppose G is a group and H ⊆ G is a subgroup
of index 2, that is, such that there are precisely two (say, left-) cosets of H in G. Prove that H is
normal in G.

Proof. Let x ∈ H, and we need to prove that gxg −1 ∈ H for all g ∈ G. If g ∈ H then there is
nothing to prove, so assume that g ∈ aH, another coset of H in G. We can write g = ah for some
h, so it remains to study ahxh−1 a−1 . By noting that ahxh−1 ∈ aH, we know that ahxh−1 does not
belong to H, and in the sense of right cosets, ahxh−1 must belong to Ha, so there exists h0 ∈ H
such that ahxh−1 = h0 a. Finally

gxg −1 = ahxh−1 a−1 = h0 aa−1 = h0 ∈ H

which shows that H is normal. 

Problem II.8.7. Let (A|R), (A0 |R 0 ), be the presentation for groups G, G0 , respectively, and assume
that A and A0 are disjoint. Prove that

G ∗ G0 := (A ∪ A0 | R ∪ R 0 )

satisfies the universal property for the coproduct of G and G0 in Grp.

Proof. Write H = R ∪ R 0 . Let us construct a homomorphism from G to G ∗ G0 . As G = F (A)/R,

by the universal property of quotient we have a commutative diagram
f
F (A) G ∗ G0
π
∃!ϕ1

F (A)/R

In particular, we let f be the map that maps to the cosets of H, i.e. f (w) = wH. Then naturally
we have ϕ1 (wR) = wH. Similarly, for G0 we have another homomorphism ϕ2 (vR 0 ) = vH.
CHAPTER II. GROUPS, FIRST ENCOUNTER 16

Now it suffices to check the universal property. For every homomorphism that maps G and G0
to a group K, which we call them f1 and f2 , we can define φ : G ∗ G0 → K by
|w|
Y
f1 (wi R)χF (A) (wi ) + f2 (wi R 0 )χF (A0 ) (wi )


φ(wH) =
i=1

where w = w1 · · · wn , χ is the indicator function. The commutative of the coproduct diagram is

clear, and φ is clearly a homomorphism since we can clearly combine two finite product to one. 
Problem II.8.13. Let G be a finite group, and assume |G| is odd. Prove that every element of G
is a square.
Proof. Let |G| = 2n − 1, n ∈ N. For every g ∈ G, we have

g = g · g 2n−1 = g 2n = (g n )2

which implies that every element in G is a square. 

Problem II.8.14. Generalize the result of II.8.13: if G is a group of order n and k is an integer
relatively prime to n, then the function G → G, g → g k is surjective.
Proof. By the prime condition, we can apply Bezout’s identity, namely there exists integers a, b such
that an + bk = 1. Then for every g ∈ G, we have

g = g · g −an = g 1−an = g bk = (g b )k

which implies that every element in G is a k-power of some element in G. 

Problem II.8.17. Assume that G is a finite abelian group, and let p be a prime divisor of |G|.
Prove that there exists an element in G of order p.
Proof. We proceed by induction. Clearly if |G| = 1 then the statement is true. Now suppose for all
abelian group with order less than n, we can find a element whose order is a prime and a divisor of
G. Then for any group G that has order n, consider an element g ∈ G, and consider the subgroup
generated by g, H = hgi.
Clearly H is cyclic, so we can find a element g |g|/q of order q where q is a prime since

1 = g |g| = (g |g|/q )q

provided that q | |g|. Now if q = p, then we are done; otherwise, we replace G with G/hhi, where
h = g |g|/q (note that all subgroups are normal since G is abelian). Now this quotient has order less
than n, and by induction, we can find an element of order p in it, which we call it mhhi. Finally the
element mhq has order p, since
(mhq )p = mp g p|g| = 1
Note that the commutativity is used here. 
Problem II.8.20. Assume that G is a finite abelian group, and let d be a divisor of |G|. Prove that
there exists a subgroup H ⊆ G of order d.
Proof. We proceed by induction. Clearly if |G| = 1 then the statement is true. Now suppose for all
abelian group with order less than n, we can find a subgroup whose order is a divisor of |G|. Then
if |G| = n, then by II.8.17, we have an element in G that is of order p, where p is a prime and a
divisor of d. If p = d, then we are done. Otherwise, we consider the quotient G/hpi. This group has
order |G|/p, and by induction hypothesis, we can find a subgroup H in the quotient that is of order
d/p. Now we claim that the set

H 0 = {gpn : n ∈ {0, · · · , p − 1}, ghpi ∈ H}

CHAPTER II. GROUPS, FIRST ENCOUNTER 17

is a subgroup of order d. It is indeed a subgroup since for g, h ∈ H 0 ,

gh−1 = apk b−1 p−l = ab−1 pk−l ∈ H 0

for some a, b that is a coset representative (ab−1 hpi ∈ H since H is a subgroup). As the cosets are
disjoint, there are precisely p · d/p = d elements in H 0 , proving the assertion. 
Problem II.8.21. Let H, K be subgroups of a group G. Construct a bijection between the set of
cosets hK with h ∈ H and the set of left-cosets of H ∩ K in H. If H and K are finite, prove that
|H| · |K|
|HK| = .
|H ∩ K|

Proof. The map hK ↔ h(K ∩H), h ∈ H is a bijection: it is well-defined since for g, h ∈ H, gK = hK

implies gh−1 ∈ K, and since g, h ∈ H, gh−1 ∈ H ∩ K and hence g(H ∩ K) = h(H ∩ K). It is injective
by reversing the above argument, and surjective by construction.

{hK : h ∈ H} ←→ {h(H ∩ K) : h ∈ H}

Now the set on the left has |HK|/|H| elements in total, and the set on the right has |H|/|H ∩ K|.
A simple rearrangement gives the result. 
Problem II.8.22. Let ϕ : G → G0 be a group homomorphism, and let N be the smallest normal
subgroup containing im ϕ. Prove that G0 /N satisfies the universal property of coker ϕ in Grp.
Proof. By universal property of quotient, for every homomorphism α : G0 → L, the homomorphism
ᾱ : G0 /N → L exists and is unique. Now it suffices to check the universal property of cokernel. For
any α : G0 → L such that α ◦ ϕ = 0, define ᾱ(gN ) = α(g). We need to check that this is well defined.
If ᾱ(gN ) = ᾱ(hN ) but α(g) 6= α(h), then gh−1 ∈
/ ker α. However since α ◦ ϕ = 0, im ϕ ⊆ ker α. By
noting that N is normal and minimal, we have

ker α ⊇ N 3 gh−1

since gN = hN . This is a contradiction, therefore α(g) = α(h), showing the well-definedness of ᾱ.
Then
ᾱ(π(ϕ(g)) = ᾱ(N ) = α(e) = eL
for all g ∈ G. This shows ᾱ ◦ π ◦ ϕ = 0, and the assertion is proved. 
Problem II.8.24. Show that epimorphisms in Grp do not necessarily have right-inverses.
Proof. Let
ϕ : Z → Z2 , ϕ(x) = x mod 2
this map has no right inverses as any homomorphism from Z2 to Z can only be the identity map. 

II.9
Problem II.9.7. Prove that stabilizers are indeed subgroups.
Proof. Assume G acts on A, and pick a ∈ A. For g, h ∈ StabG (a), we have

gh−1 a = g(h−1 (ha)) = ga = a

as required. 
Problem II.9.11. Let G be a finite group, and let H be a subgroup of index p, where p is the
smallest prime dividing |G|. Prove that H is normal in G.
CHAPTER II. GROUPS, FIRST ENCOUNTER 18

Proof. We consider the left-multiplication action of G on the left cosets of H, which is g · hH = ghH.
This induces a homomorphism ϕ : G → Sp , whose kernel includes H since

if g ∈ ker ϕ, then aH = gaH ∀a ∈ G ⇒ g = gH ⇒ g ∈ H.

Then G/ ker ϕ ∼= im ϕ, so G/ ker ϕ is a subgroup of Sp , therefore it has order dividing p!. However
by Lagrange, such order also divides |G|, and hence must be divisible by p, so |G/ ker ϕ| = p. Finally

p = [G : H] = [G : ker ϕ][ker ϕ : H] = p[ker ϕ : H]

which leads to [ker ϕ : H] = 1. Since ker ϕ ⊆ H, ker ϕ = H by index consideration, proving the
assertion. 

Problem II.9.12. Let G be a group, and let H ⊆ G be a subgroup of index n. Prove that H
contains a subgroup K that is normal in G and such that [G : K] divides the gcd of |G| and n!. (In
particular, [G : K] ≤ n!.)

Proof. Following the same pattern from II.9.11, consider the left-multiplication action of G on the
left cosets of H, which is g · hH = ghH. This induces a homomorphism ϕ : G → Sn (as there are n
left cosets), whose kernel includes H since

if g ∈ ker ϕ, then aH = gaH ∀a ∈ G ⇒ g = gH ⇒ g ∈ H.

Define K = ker ϕ. Then G/K ∼ = im ϕ, so G/K is a subgroup of Sn , therefore it has order dividing
n!. By Lagrange, such order also divides |G|, so we’ve found the required K. 

Problem II.9.13. Prove ’by hand’ that that for all subgroups H of a group G and ∀g ∈ G, G/H
and G/(gHg −1 ) (endowed with the action of G by left-multiplication) are isomorphic in G-Set.

Proof. We want to find a bijection function ϕ : G/H → G/gHg −1 such that the diagram
idG ×ϕ
G × G/H G × G/gHg −1
ρ ρ0
ϕ
G/H G/gHg −1

commutes. Indeed the most natural map would be ϕ(xH) = (gxg −1 )gHg −1 . We check that this is
well-defined; if aH = bH, then gaHg −1 = gbHg −1 clearly. We now check that this is a bijection, by
explicitly give the inverse

φ : G/gHg −1 → G/H, φ(xgHg −1 ) = (g −1 xg)H

so ϕ ◦ φ = id. Therefore G/H and G/(gHg −1 ) are isomorphic in G-Set. Note that if we assume
ϕ(xH) = xgHg −1 , then H would need to be normal in order to be well-defined. 

Problem II.9.17. Consider G as a G-set, by acting with left-multiplication. Prove that

AutG−Set(G) ∼
= G.
Proof. The set of automorphisms on G − Set(G) are bijections that satisfies gϕ(h) = ϕ(gh). In
particular we can define
ϕg (h) = g −1 h
this is clearly a bijection and forms a group structure by ϕg ϕh = ϕgh . We now consider the map
ψ : AutG−Set(G) → G by ψ(ϕg ) = g. We claim that this is an isomorphism. Indeed, its kernel
is precisely ϕe , which is the identity of AutG−Set(G) . The map is clearly surjective, and it is an
homomorphism by construction. Therefore AutG−Set(G) ∼ = G. 
Chapter III

Rings and modules

Unless otherwise specified, in the following R = (R, +, ·) denotes an arbitrary ring with identity
(the book assumes this throughout this book), 0, 1 denotes the additive and multiplicative identity
of R, respectively. In the case of possible confusion, I will use 0R , 1R instead.
Some description and hints are omitted for simplicity.

III.1
Problem III.1.1. Prove that if 0 = 1 in a ring R, then R is a zero ring.
Proof. If r is any element in R, then

r =r·1=r·0=0

showing that R = 0. 
Problem III.1.6. Prove that if a and b are nilpotent in R and ab = ba, then so is a + b.
Proof. If an = 0, bm = 0, then
 
n+m n+m n + m − 1 n+m−1
(a + b) =a + a b + . . . + bn+m
1
and all terms are zeros since every term either have an or bm . If we do not assume that ab = ba,
then the statement would be false, for example, in Mn (Z),
   
1 0 0 1
and
1 0 0 1
     
1 0 0 1 1 1
are nilpotent of degree 3, but + = is not nilpotent. 
1 0 0 1 1 1
Problem III.1.7. Prove that [m] is nilpotent in Z/nZ if and only if m is divisible by all prime
factors of n.
Proof. (⇒) If [m]k = [0] for some integer k, then this implies mk = dn for some integer d. Now we
write n = pa11 · · · pann , where pi are primes, and ai are positive integers. Then

mk = dpa11 · · · pann

and it is clear to see that m must contain each pi at least once.

(⇐) If n = pa11 · · · pann where pi are primes, and ai are positive integers, then we can write

m = pb11 · · · pbnn d

19
CHAPTER III. RINGS AND MODULES 20

where bi , d are positive integers, and pi - d for all i. Define

  
a1 an
f = floor max ,···
b1 bn

then let r = mf /n, which is an integer larger than 0 by the choice of f . Finally

mf = nr = 0 mod n

showing that m is nilpotent in Z/nZ. 

Problem III.1.9. Prove Proposition 1.12, that is:

• The inverse of a two-sided unit is unique;

• two-sided units form a group under multiplication.

Proof. For a two-sided unit v, we have uv = 1 and vw = 1 for some u, w ∈ R. Then

w = 1 · w = uvw = u · 1 = u

showing that w = u, so the inverse can be uniquely defined as v −1 = u. Now as the inverse is unique,
we can define a group structure, using the multiplication from the ring R. We check that

• 1 is a unit as 1 · 1 = 1;
• for a unit u, u−1 is also a unit;
• associativity is clear.

Problem III.1.15. Prove that R[x] is a domain if and only if R is a domain.

Proof.
(⇒) Trivial since R ⊂ R[x].
(⇐) Assume the contrary that R[x] is not a domain. Then we can find f = ni=0 ai xi , g = m j
P P
j=0 bj x ,
f 6= 0, g 6= 0 such that f g = 0. Then we would have an bm = 0, and since R is a domain, either an
or bm is zero. Without loss of generality, we can reduce the case to f = a0 6= 0. Then by the same
argument, we would arrive at a0 b0 = 0, since all higher terms must be zero. But this contradict to
the assumption that R is a domain, since f = a0 and g = b0 are nonzero. Hence R[x] must be a
domain. 

III.2
Problem III.2.1. Prove that if there is a homomorphism from a zero ring to a ring R, then R is a
zero ring.

Proof. If 1R is the multiplicative identity of R, then for any homomorphism ϕ : 0 → R,

0R = ϕ(0) = ϕ(1) = 1R

and by III.1.1, R is a zero-ring. 

Problem III.2.6. Verify the ’extension property’ of polynomial ring:

Let α : R → S be a fixed ring homomorphism, and let s ∈ S be an element commuting with α(r)
for all r ∈ R. Then there is a unique ring homomorphism ᾱ : R[x] → S extending α and sending x
to s.
CHAPTER III. RINGS AND MODULES 21

ai xi ∈ R[x], we have no choice but to define

P
Proof. Indeed, for i≥0
!
X X
ᾱ ai x i = α(ai )si (1)
i≥0 i≥0

so that ᾱ(r) = α(r) and x sends to s in this map. It is clearly a homomorphism (note that the
commutativity of s is used in the proof of ᾱ(f g) = ᾱ(f )ᾱ(g)), so it suffices to check that ᾱ is unique.
But it is clear by the fact that any map that extends α and send x to s must have the same value
evaluated as in (1). 
Problem III.2.9. Prove that the center of R is a subring. Moreover, prove that the center of a
division ring is a field.
Proof. A subset of a ring S is a subring if it is a subgroup of (R, +), closed under multiplication,
and 1 is in it. So we check that:
• it is a subgroup of (R, +): for a, b ∈ C, for all r ∈ R,

(a − b)r = ar − br = ra − rb = r(a − b)

showing that a − b ∈ C, hence a subgroup;

• closed under multiplication: for a, b ∈ C, for all r ∈ R,

abr = a(br) = a(rb) = (ar)b = (ra)b = rab

showing that ab ∈ C;
• finally, 1 is in C since 1r = r1 for all r ∈ R.
Clearly the center forms a commutative ring since for a, b ∈ C, ab = ba. Then it follows by
definition that a commutative division ring is a field. 
Problem III.2.10. Prove that the centralizer of a is a subring for every a ∈ R. Prove that the
center is the intersection of all its centralizers, and prove that every centralizer of a division ring is
a division ring.
Proof. We use the same test as above. Let Cx denotes the centralizer of x.
• It is a subgroup of (R, +): for a, b ∈ Cx ,

(a − b)x = ax − bx = xa − xb = x(a − b)

showing that a − b ∈ Cx , hence a subgroup;

• closed under multiplication: for a, b ∈ Cx ,

abx = a(bx) = a(xb) = (ax)b = (xa)b = xab

showing that ab ∈ Cx ;
• finally, 1 is in Cx since 1x = x1.
It is easy that the center is the intersection of all its centralizers, since such elemet in the intersection
must commute with the whole ring R. Finally, if R is a division ring, then for every element a ∈ Cx ,
we can show that a−1 ∈ Cx :

ax = xa ⇒ axa−1 = x ⇒ xa−1 = a−1 x

Therefore every element in Cx has a inverse, and by definition, Cx is a division ring. 

CHAPTER III. RINGS AND MODULES 22

Problem III.2.11. Prove that a division ring R which consists of p2 elements where p is a prime,
is commutative.
Proof. Suppose the contrary that R is not commutative. Then the center C must be a proper
subring, which can only consist of p elements by Lagrange. Now let r ∈ R\C. Then the centralizer
of r will contain at least r and C by III.2.10, therefore the centralizer of r must be R itself (again
by Lagrange), for every r ∈ R\C. But then the intersection of all centralizer are now R (element of
center has centralizer R clearly), which is a contradiction to that C is proper. Therefore R must be
commutative, i.e. a field. 
→ Q. Describe the cokernel of ι in Ab and its
Problem III.2.12. Consider the inclusion map ι : Z ,−
cokernel in Ring.
Solution. In Ab, this is easy: it is just Q/ im ι = Q/Z. However in Ring, we notice that for any map
α : Q → F that satisfy α ◦ ι = 0, we have

0F = α(1) = α ◦ ι(1) = α(1) = 1F

which shows that F must be the zero ring by III.1.1. Now the unique homomorphism ᾱ : coker ι → F
must also be the zero map, and by the requirement ᾱ ◦ π ◦ ι = 0, we finally have π ◦ ι = 0, and by
the same argument as above, we have that the codomain of π is the zero ring, i.e. coker ι = 0. 

III.3
Problem III.3.2. Let ϕ : R → S be a ring homomorphism, and let J be an ideal of S. Prove that
ϕ−1 (J) is an ideal.
Proof. The ideal is clearly nonempty, so it suffices to check that ϕ−1 (J) is a additive subgroup and
satisfies the absorption property. For x, y ∈ ϕ−1 (J), we have ϕ(x), ϕ(y) ∈ J, so ϕ(x) − ϕ(y) =
ϕ(x − y) ∈ J, therefore x − y ∈ ϕ−1 (J), showing that it is a subgroup of (R, +).
Now for any r ∈ R, a ∈ ϕ−1 (J), we have ϕ(a) ∈ J, so ϕ(r)ϕ(a) = ϕ(ra) ∈ J, and hence
ra ∈ ϕ−1 (J), showing the left-absorption property. The right case is the same. 
Problem III.3.3. Let ϕ : R → S be a ring homomorphism, and let J be an ideal of R.

• Show that ϕ(J) need not be an ideal of S.

• Assume that ϕ is surjective; then prove that ϕ(J) is an ideal of S.
• Assume that ϕ is surjective, and let I = ker ϕ. Let J¯ = ϕ(J). Prove that

R/I ∼ R
= .
J¯ I +J

Proof. Let ϕ : Z ,−→ R be inclusion (and clearly a homomorphism). Then every ideal of Z will be
directly transformed into R. But since R is a field, by III.3.8 (which will be proved later) the possible
ideal of R are only {0} and R itself, so the image of a homomorphism need not to be an ideal.
However, If ϕ is surjective, Then ϕ(J) is indeed an ideal: if ϕ(x), ϕ(y) ∈ ϕ(J), then so is
ϕ(x) − ϕ(y) = ϕ(x − y) ∈ ϕ(J). The absorption property is also true since ϕ(r)ϕ(x) = ϕ(rx) ∈ ϕ(J).
Finally, we consider the homomorphism

φ : R/I → R/(I + J), φ(a + I) = a + I + J

φ is clearly a surjective homomorphism, and by first isomorphism theorem

R/I ∼ R
=
ker φ I +J
CHAPTER III. RINGS AND MODULES 23

so it remains to solve ker φ, which is

ker φ = {a + I : a + I + J = I + J}
= {a + b + I : a ∈ I, b ∈ J}
= {b + I : b ∈ J}
= {ϕ(b) ∈ S : b ∈ J} (regarding R/I as S)
= ϕ(J) = J¯

therefore
R/I ∼ R
=
J¯ I +J
as required. 

Problem III.3.7. Let R be a ring, and let a ∈ R. Prove that Ra is a left-ideal of R and aR is a
right-ideal of R. Prove that a is a left-, resp. right-, unit if and only if R = aR, resp. R = Ra.

Proof. We prove only the left-ideal case since the same argument holds for right-ideal case. Ra is a
subgroup of (R, +) since for ra, sa ∈ Ra, ra − sa = (r − s)a ∈ Ra. The absorption property follows
easily since rsa = (rs)a ∈ Ra.
If a is a right unit, then there exists u such that ua = 1. Then 1 is contained in Ra, and since
for all r ∈ R, r · 1 ∈ Ra, we conclude that R = Ra. 

Problem III.3.8. Prove that R is a division ring if and only if its only left-ideals and right-ideals
are {0} and R.
In particular, a commutative ring R is a field if and only if the only ideals of R are {0} and R.

Proof. (⇒) If a nonzero element a is in the left-ideal I, then so is 1 since

1 = a−1 a ∈ I by definition

Therefore any nonzero left-ideals are automatically R itself. The right-ideal case is the same.
(⇐) If a nonzero element a does not have a left inverse, then aR would be a proper right-ideal by
III.3.7. Therefore all elements must have left(and hence right) inverse. 

Problem III.3.10. Let ϕ : k → R be a ring homomorphism, where k is a field and R is a nonzero

ring. Prove that ϕ is injective.

Proof. ϕ is injective if and only if ker ϕ = {0} by Proposition III.2.4. Also, the ideals of k are only
{0} and k by III.3.8. If ker ϕ = {0} then there is nothing to prove, so let ker ϕ = k. But this means
that ϕ = 0, so we have
1R = ϕ(1) = 0 = ϕ(0) = 0R
and by III.1.1, R is a zero ring, a contradiction to the hypothesis. Therefore ker ϕ = {0}, showing
that ϕ is injective. 

Problem III.3.12. Let R be a commutative ring. Prove that the set of nilpotent elements forms
an ideal of R. This ideal is called the nilradical of R.

Proof. From III.1.6 we already know that it forms a subgroup of (R, +) by relpacing b with −b, so
it remains to check that it is an ideal. Let I be such ideal. If a ∈ R, r ∈ I and rn = 0, then since
!
(ar)n = an rn = 0

in which ! is where commutativity is used. Therefore ar ∈ I, proving the absorption property.

For an counter-example where R is not commutative, simply consider the example of III.1.6: it
is not even a subgroup of (R, +). 
CHAPTER III. RINGS AND MODULES 24

Problem III.3.13. Let R be a commutative ring, and let N be its nilradical. Prove that R/N
contains no nonzero nilpotent elements. Such a ring is said to be reduced.
Proof. Pick an element a ∈ R\N . Then for every integer n > 0,
 
n n n n−1
(a + N ) = a + a N + · · · + N n = an + N
1
Since a is not nilpotent, an 6= 0 for every n, showing that a + N is not nilpotent for a ∈ R\N . 

III.4
Problem III.4.1. Let R be a ring, and let {Iα }α∈A be a family of ideals of R. We let
( )
X X
Iα := rα such that rα ∈ Iα and rα = 0 for all but finitely many α .
α∈A α∈A

Prove that {Iα }α∈A is an ideal of R and that it is the smallest ideal containing all of the ideals Iα .
Proof. We only consider the case when A = {1, 2}: Any other A follows the same exact argument.
Let I = I1 + I2 . I is a subgroup of (R, +) : the two elements in I can be represented as r1 + r2
and r10 + r20 , and clearly (r1 − r10 ) + (r2 − r20 ) is in I. The absorption property is also clear, since
r(r1 + r2 ) = (rr1 + rr2 ) ∈ I.
Now it suffice to show that I is minimal. For every ideal that contains I1 and I2 , they must also
contain r1 + r2 for r1 ∈ I1 and r2 ∈ I2 , since ideal is a subgroup of (R, +). Therefore every such ideal
must also contain I, proving the minimality of I. 
Problem III.4.2. Prove that the homomorphic image of a Noetherian ring is Noetherian.
Proof. Let R be Noetherian, S be any ring, ϕ : R → S be a surjective ring homomorphism. Let
J be an ideal of S. By III.3.2, the preimage is an ideal, which we call I = ha1 , . . . , an i. We claim
that J = hϕ(a1 ), . . . , ϕ(an )i, so every finitely generated ideal will map to a finitely generated ideal,
proving that S is Noetherian.
Indeed, since ai ∈ ϕ−1 (J), ϕ(ai ) ∈ J for i = 1, . . . , n, so hϕ(a1 ), . . . ϕ(an )i ⊆ J. On the other
hand, for an element j ∈ J, there exists i ∈ R such that ϕ(i) = j by surjectivity, therefore i ∈ I, so
i is generated by elements a1 , . . . , an , i.e. i = r1 a1 + · · · + rn an . Then since ϕ is a homomorphism,
ϕ(i) = j = ϕ(r1 a1 + · · · + rn an ) = s1 ϕ(a1 ) + · · · + sn ϕ(an )
so J ⊆ hϕ(a1 ), . . . ϕ(an )i, and the claim is proved. 
Problem III.4.3. Prove that the ideal (2, x) of Z[x] is not principal.
Proof. Assume that (f ) = (2, x). Then there is some q ∈ Z[x] such that f q = 2. Then f, q are
constant and f must be 2 since 1 is not in it. But we also have f g = x for some g ∈ Z[x], and there
are no possible choice of g such that 2g = x. Hence (2, x) is not principal. 
Problem III.4.4. Prove that if k is a field, then k[x] is a PID.
Proof. Let I be any ideal of k[x]. If I = (0), then there is nothing to prove. Otherwise, there is some
polynomial f ∈ I that has minimal degree in I and is monic (since you can do scalar division). We
claim that I = (f ). Indeed, for g ∈ I, we can use division algorithm to write
g(x) = f (x)q(x) + r(x)
where deg r(x) < deg f (x). Since k[x] is a subgroup, r = g − f q ∈ I, and by the minimality of f ,
r(x) = 0, so every element of I can be written as g(x)f (x) for some g ∈ k[x], showing that k[x] is a
PID. 
CHAPTER III. RINGS AND MODULES 25

Problem III.4.5. Let I, J be ideals in a commutative ring R, such that I + J = (1). Prove that
IJ = I ∩ J.

Proof. If x ∈ IJ, then it can be represented as ij for some i ∈ I, j ∈ J, and by the property of ideal,
ji ∈ I, ij ∈ J, so ij ∈ I ∩ J. Conversely, we have

I ∩ J = (I ∩ J)(1) = (I ∩ J)(I + J) = (I ∩ J)I + (I ∩ J)J ⊆ IJ + IJ = IJ

showing the identity. 

Problem III.4.7. Let R = k be a field. Prove that every nonzero (principle) ideal in k[x] is
generated by a unique monic polynomial.

Proof. From III.4.4 we already know that every ideal is generated by a single polynomial f . Since k
is a field, we can do division, so there is a monic polynomial f (x)/a where a is the coefficient of the
largest degree in f . Then it’s trivial that (f ) = (f /a). 

Problem III.4.10. Let d be an integer that is not the square of an integer, and consider the subset
of C defined by √ √
Q( d) := {a + b d | a, b ∈ Q}
√
• Prove that Q( d) is a subring of C.
√ √
• Define a function N : Q( d)√→ Z by N (a + b d) := a2 − b2 d. Prove that N (zw) = N (z)N (w)
and that N (z) 6= 0 if z ∈ Q( d), z 6= 0. N is called a norm.
√ √
• Prove that Q( d) is a field and in fact the smallest subfield of C containing both Q and d.
√
• Prove that Q( d) ∼ = Q[t]/(t2 − d).

Proof.
√ √ √ √
• Subring property is clear by a + b d − (c + d d) = (a − c) + (b − d) d ∈ Q( d).
√
• If N (a + b d) = 0, then a2 = b2 d, and since d is not a square, a cannot be a rational number,
so a = 0 = b. The multiplicative property is easily checked by
√ √
N ((a + b d)(m + n d)) = (am + bnd)2 − (an + bm)2 d
= (am)2 − (an)2 d − (bm)2 d + (bnd)2 + 2ambnd − 2ambnd
√ √
= (a2 − b2 d)(m2 − n2 d) = N (a + b d)N (m + n d)
√
• An inverse of a + b d is
1 1  √ 
√ = 2 a − b d .
a+b d a − b2 d
• The homomorphism √ √
ϕ : Q[t] → Q( d), ϕ(f (x)) = f ( d)
has kernel (t2 − d), and the result is immediate by first isomorphism theorem.

Problem III.4.11. Let R be a commutative ring, a ∈ R, and f1 (x), . . . , fr (x) ∈ R[x].

• Prove the equality of ideals

(f1 (x), . . . , fr (x), x − a) = (f1 (a), . . . , fr (a), x − a).

CHAPTER III. RINGS AND MODULES 26

• Prove the useful substitution trick

R[x] ∼ R
=
(f1 (x), . . . , fr (x), x − a) (f1 (a), . . . , fr (a))

Proof. We consider only the case k = 1; the other cases are just extending the same argument. We
are required to prove that
(f (x), x − a) = (f (a), x − a)
For f (x), we can apply division algorithm to get

f (x) = q(x)(x − a) + r

where q(x) ∈ R[x], r ∈ R. By plug in x = a, we obtain r = f (a). Therefore f (x) is generated by

f (a) and (x − a), showing f (x) ∈ (f (a), x − a). On the other hand, note the division algorithm also
implies
f (a) = f (x) − q(x)(x − a) ∈ (f (x), x − a)
therefore f (a) ∈ (f (x), x − a), so (f (x), x − a) = (f (a), x − a). Now since R[x]/(x − a) ∼= R, by
III.3.3
R ∼ R[x]
=
ϕ(J) ker ϕ + J
for an ideal J ∈ R[x], ϕ : R[x] → R a surjective homomorphism. It is clear that how should we
choose these: by taking
J = (f1 (x), . . . , fr (x)), ϕ(f (x)) = f (a)
we have
R ∼ R[x]
=
(f1 (a), . . . , fr (a)) (f1 (x), . . . , fr (x), x − a)
as desired (note that ϕ is surjective). 
Problem III.4.13. Let R be an integral domain. For all k = 1, . . . , n, prove that (x1 , . . . , xk ) is
prime in R[x1 , . . . , xn ].
Proof. We proceed by induction. For the case k = 1, we have
R[x] ∼
= R (p.p.151)
(x)

and since R is a domain, it follows by definition that (x) is a prime ideal. Suppose that for k < n,
the argument holds. Then for k = n, choose

J = (x1 , . . . , xn−1 ), → R[x1 , . . . , xn−1 ]

ϕ : R[x1 , . . . , xn ] ,−

where ϕ is the inclusion map and ker ϕ = (xn ). Then by III.3.3

R[x1 , . . . , xn ]/(xn ) ∼ R[x1 , . . . , xn ]

=
(x1 , . . . , xn−1 ) (x1 , . . . , xn−1 ) + (xn )
which simplifies to
R[x1 , . . . , xn−1 ] ∼ R[x1 , . . . , xn ]
=
(x1 , . . . , xn−1 ) (x1 , . . . , xn )
By induction hypothesis, the quotient on the left is a domain since (x1 , . . . , xn−1 ) is a prime ideal,
therefore by definition, (x1 , . . . , xn ) is a prime ideal. 
Problem III.4.16. Let R be a commutative ring, and let P be a prime ideal of R. Suppose 0 is
the only zero-divisor of R contained in P . Prove that R is an integral domain.
CHAPTER III. RINGS AND MODULES 27

Proof. Let a, b ∈ R such that ab = 0. Then since 0 ∈ P , ab ∈ P , so either a ∈ P or b ∈ P . Without

loss of generality, let a ∈ P . If a = 0, then we are done; otherwise, a 6= 0, and since ab = 0, we must
have b = 0 as a is not a zero divisor (0 is the only zero-divisor in P ). In both cases, we show that
ab = 0 implies a = 0 or b = 0, showing that R is a domain. 

Problem III.4.18. Let R be a commutative ring, and let N be its nilradical (III.3.12). Prove that
N is contained in every prime ideal of R.

Proof. Let xn = 0 for some positive integer n, and P a prime ideal. Then since 0 ∈ P , we have

P 3 0 = xn = x · xn−1

By the property of prime ideal, either x ∈ P or xn−1 in P . If the former case is true, then we are
done; else, we can reduce to the case where either x ∈ P or xn−2 ∈ P . By continuing this process,
we will arrive at either x ∈ P or x ∈ P , showing that in any cases, x ∈ P . Therefore all nilpotent
elements are in P , proving the statement. 

Problem III.4.21. Let k be an algebraic closed field, and let I ⊆ k[x] be an ideal. Prove that I is
maximal if and only if I = (x − c) for some c ∈ k.

Proof. (⇐) We have

k[x] ∼
=k (p.p.151)
(x − c)
and since k is a field, it follows by definition that (x − c) is maximal.
(⇒) Let J be a maximal ideal. By III.4.4, k[x] is a PID, hence every ideal is being generated by
a single monic polynomial f (x) ∈ k[x] (III.4.7). Since k is algebraic closed, we can write f (x) =
q(x)(x − c) for some q(x) ∈ k[x], c ∈ k. Then

J = (f (x)) = (q(x)(x − c)) ⊆ (x − c)

and by Proposition III.4.11, either J = (x − c) or J = k[x]. The latter case could not happen since
the maximal can not be k[x] itself, therefore J = (x − c), as desired. 
Unless otherwise specified, in the following M denotes a (left-)module over R.

III.5
Problem III.5.2. Prove claim 5.1.

Proof. Let σ : R → EndAb (M ) be a ring homomorphism and ρ : R × M → M a function. We verify

the following properties:

• ρ(r, m + n) = ρ(r, m) + ρ(r, n).

Note that σ(r) is a endomorphism on M . Then

ρ(r, m + n) = σ(r)(m + n) = σ(r)(m) + σ(r)(n) = ρ(r, m) + ρ(r, n)

• ρ(r + s, m) = ρ(r, m) + ρ(s, m).

ρ(r + s, m) = σ(r + s)(m) = σ(r)(m) + σ(s)(m) = ρ(r, m) + ρ(s, m)

• ρ(rs, m) = ρ(r, ρ(s, m)).

ρ(rs, m) = σ(rs)(m) = σ(r)σ(s)(m) = σ(r)ρ(s, m) = ρ(r, ρ(s, m))

CHAPTER III. RINGS AND MODULES 28

• ρ(1, m) = m.
ρ(1, m) = σ(1)(m) = 1(m) = m

Problem III.5.3. Prove that 0 · m = 0 and that (−1) · m = −m for all m ∈ M .

Proof. 0m = (0 + 0)m = 0m + 0m, 0m = 0. Since 0 = 0m = (−1 + 1)m = (−1)m + m, (−1)m =

−m. 

Problem III.5.4. (Schur’s Lemma) Let R be a ring. A nonzero R-module M is simple (or irre-
ducible) if its only submodules are {0} and M . Let M, N be simple modules, and let ϕ : M → N
be a homomorphism of R-modules. Prove that either ϕ = 0 or ϕ is an isomorphism.

Proof. The kernel of a R-module homomorphism is a submodule of M , which can only be {0} or
M . If ker ϕ = M then ϕ = 0, and if ker ϕ = {0} then ϕ is injective. The image of a R-module
homomorphism is a submodule of N , which again can only be {0} or M . if im ϕ = {0} then ϕ = 0,
and if im ϕ = N then ϕ is surjective.
So there are four different combination of images and kernels:

• ker ϕ = M, im ϕ = {0} ⇒ ϕ = 0;
• ker ϕ = M, im ϕ = N ⇒ ϕ = 0, N = 0, which can’t be by hypothesis;
• ker ϕ = {0}, im ϕ = {0} ⇒ ϕ = 0, M = 0, which can’t be by hypothesis;
• ker ϕ = {0}, im ϕ = N ⇒ ϕ is an isomorphism.

so either ϕ = 0 or ϕ is an isomorphism. 

Problem III.5.5. Let R be a commutative ring, viewed as an R-module over itself, and let M be
an R-module. Prove that HomR-Mod (R, M ) ∼
= M as R-modules.
Proof. Every module homomorphism f : R → M is uniquely determined by the result of f (1) (which
is a unit). Therefore

HomR-Mod (R, M ) → M
f 7→ f (1)

is an isomorphism of R-modules. 

Problem III.5.11. Let R be commutative, and let M be an R-module. Prove that there is a natural
bijection between the set of R[x]-module structures on M (extending the given R-module structure)
and EndR−Mod (M ).

Proof. If f ∈ EndR−Mod (M ), then we have to show that there are some suitable maps

R[x] × M → M
(f (x), m) → ?

ai xi , then
P
that makes M into a R[x]-module. We consider (g(x), m) → g(f )(m), where if g(x) = i
X
σ(f, m) = ai f i (m) where f i = f ◦ · · · ◦ f
| {z }
i i times

We can easily check by definition that M is a R[x]-module. Conversely, if M is a R[x]-module, then

define f (m) = xm. Then f is indeed an endomorphism (note that the commutativity of R ensures
that rxm = xrm for r ∈ R, so f is an endomorphism), proving the statement. 
CHAPTER III. RINGS AND MODULES 29

Problem III.5.12. Let M, N be R-modules, and let ϕ : M → N be a homomorphism of R-modules

which has a inverse (therefore a bijection). Prove that ϕ−1 is also a homomorphism of R-modules.
Conclude that a bijective R-module homomorphism is a R-module isomorphism.

Proof. Since
ϕ(ϕ−1 (m) + ϕ−1 (n)) = m + n = ϕ(ϕ−1 (m + n))
we have ϕ−1 (m) + ϕ−1 (n) = ϕ−1 (m + n). And

ϕ(rϕ−1 (m)) = rϕ(ϕ−1 (m)) = rm = ϕ(ϕ−1 (rm))

so rϕ−1 (m) = ϕ−1 (rm) indeed. 

Problem III.5.14. Prove Proposition 5.18, that is:

Let N, P be submodules of an R-module M . Then

• N + P is a submodule of M ;
• N ∩ P is a submodule of P , and
N +P ∼ P
= .
N N ∩P
Proof. Every element of N + P can be written as n + p where n ∈ N, p ∈ P . Then it is clear
that r(n + p) = rn + rp ∈ N + P for r ∈ M . For the intersection N ∩ P , it is also clear that for
p ∈ P, n ∈ N ∩ P , pr ∈ N since r ∈ N , and pr ∈ P since p ∈ P .
The proof for the second isomorphism theorem follows exactly the same as in groups (Proposition
II.8.11). Consider the homomorphism
N +P
ϕ:P → , ϕ(p) = pN
N
it is surjective since for every (n + p)N , there is a corresponding p. Then

ker ϕ = {p ∈ P : p ∈ N } = P ∩ N

finally it follows by first isomorphism theorem that

N +P ∼ P
= .
N N ∩P


III.6
Problem III.6.1. Prove Claim 6.3, that is, F R (A) ∼
= R⊕A .
Proof. Observe that every element in R⊕A can be uniquely written as
X
ra χ(a)
a∈A

where χ(a) = χa (x), the indicator function of a, and ra ∈ R for a ∈ A. Then it suffices to check the
universal property of free modules: given a function f : A → M where M is a module, we show that
the following diagram
∃!ϕ
R⊕A M
χ
f
A
CHAPTER III. RINGS AND MODULES 30

commutes. Indeed, we define !

X X
ϕ ra χ(a) = ra f (a)
a∈A a∈A

then the diagram clearly commutes (and is unique). Finally, ϕ is a R−Mod homomorphism since
! !
X
X X X X X
ϕ ra χ(a) + ϕ ra0 χ(a) = ra f (a) + ra0 f (a) = (ra + ra0 )f (a)
a∈A a∈A a∈A a∈A a∈A
! !
X X X
=ϕ (ra + ra0 )χ(a) =ϕ ra χ(a) + ra0 χ(a)
a∈A a∈A a∈A

Note that R-module’s definition gurantees the commutativity of X (scalar multiplication is direct).


Problem III.6.3. Let R be a ring, M an R-module, and p : M → M an R-module homomorphism

such that p2 = p. Prove that M ∼
= ker p ⊕ im p.
Proof. We are required to prove that the diagram
fk
ker p
ik

∃!ϕ
M N
im
im p
fm

commutes. Notice that for x ∈ ker p, p(x) = 0, and

for x ∈ im p, x − p(x) = p(y) − p(p(y)) = p(y) − p(y) = 0

where p(y) = x. This suggest that we define ϕ as

ϕ(x) = fk (x − p(x)) + fm (p(x))

Indeed, if x ∈ ker p, then ϕ(x) = fk (x); if x ∈ im p, then ϕ(x) = fm (p(x)) = fm (x) since for x ∈ im p,

p(y) = x, p(p(y)) = p(y) ⇒ p(x) = x.

But what about x ∈ ker p ∩ im p? In fact, the only element in the intersection is 0, as such x must
have
x = p(y) = p(p(y)) = p(x) = 0
so ϕ is well-defined. Now it suffices to check that ϕ is a homomorphism, which is direct since p, fk
and fm are both R-homomorphisms, so it preserves the action on M (check yourself if you’re not
convinced). Therefore by the universal property of coproduct, ker p ⊕ im p ∼
= M. 

Problem III.6.4. Let R be a ring, and let n > 1. View R⊕(n−1) as a submodule of R⊕n , via the
injective homomorphism R⊕(n−1) ,−
→ R⊕n defined by

→ (r1 , . . . , rn−1 , 0).

(r1 , . . . , rn−1 ) ,−

Give a one-line proof that

R⊕n ∼
= R.
R⊕(n−1)
CHAPTER III. RINGS AND MODULES 31

Proof. The surjective map

(r1 , . . . , rn−1 , rn )  rn .
has kernel precisely R⊕(n−1) , therefore by first isomorphism theorem

R⊕n ∼
= R.
R⊕(n−1)


Problem III.6.5. For any ring R and any two sets A1 , A2 , prove that (R⊕A1 )⊕A2 ∼
= R⊕(A1 ×A2 ) .

Proof. By III.6.1, it is equivalent to prove the following diagram commutes:

∃!ϕ
(R⊕A1 )⊕A2 M
j
f

A1 × A2

To do this, note that an element in (R⊕A1 )⊕A2 is a function g : A2 → R⊕A1 , in which we send an
element a2 ∈ A2 to (
1 if x = a1
ja1 ,a2 (x) := (p.p.168)
6 a1
0 if x =
this suggests us to define

j(a1 , a2 ) 7→ (ja1 ,a2 (b2 ))(b1 ) = χa1 (b1 )χa2 (b2 )

function. Then it follows the same pattern as in III.6.1: for f : A1 ×A2 → M

where χ is the indicatorP
given and any element a1 ∈A1 ,a2 ∈A2 ra1 ,a2 (ja1 ,a2 (b2 ))(b1 ) ∈ (R⊕A1 )⊕A2 , define
!
X X
ϕ ra1 ,a2 (ja1 ,a2 (b2 ))(b1 ) = ra1 ,a2 f (a1 , a2 )
a1 ∈A1 ,a2 ∈A2 a1 ∈A1 ,a2 ∈A2

The commutativity of the diagram is direct. Finally, the check for ϕ is a R − Mod homomorphism
is the same as in III.6.1. 

Problem III.6.7. Let A be any set, and for any module M over a ring R, define
Y M
M A := M, M ⊕A := M.
a∈A a∈A

Prove that ZN  Z⊕N .

Proof. Note that ZN can be regarded as the collection of functions

f :Z→N

which is the collection of all infinite sequences in Z. This set has uncountably many elements (as
one can argue using Cantor’s diagonal argument). On the other hand, Z⊕N is also the collection of
these function, but with the additional criterion that

f (n) = 0 for all but finitely many n ∈ Z

which says that this set collects all finite sequence in Z, and as we know (i.e. can construct a bijection
to Z), this set is countable. As the cardinality does not match, ZN  Z⊕N , as required. 
CHAPTER III. RINGS AND MODULES 32

Problem III.6.9. Let R be a ring, F a nonzero free R-module, and let ϕ : M → N be a homomor-
phism of R-modules. Prove that ϕ is onto if and only if for all R-module homomorphisms α : F → N
there exists an R-module homomorphism β : F → M such that α = ϕ ◦ β.

Proof. As M is free, it is generated by a set X = {xi } (not necessarily finite).

(⇒) Let {ni } ∈ N be such that ϕ(xi ) = ni . If ϕ is onto, then each ni corresponds to a mi ∈ M such
that ϕ(mi ) = ni . We then just define β(xi ) = mi , and the commutativity is clear (note that β might
not be unique, but that’s fine).
(⇐) If ϕ is not onto, i.e. there exists n ∈ N such that n ∈ / im ϕ, then this also means that
n∈/ im(ϕ ◦ β) for any β. Now we choose a suitable α so α = ϕ ◦ β does not hold. Indeed, we can
define
α(xi ) = n
for all i. Then the commutativity does not hold for any choice of β, a contradiction. Therefore ϕ
must be surjective. 

Problem III.6.10. Let M, N, and Z be R-modules, and let µ : M → Z, ν : N → Z be homomor-

phism of R-modules. Prove that R − Mod has ’fibered products’(cf. Exercise I.5.12).

Proof. As in the case Set(I.5.12), we define fibered coproduct to be the set of elements that agrees
on Z after being pushed by µ and ν:

M ×Z N := {(m, n) ∈ M ⊕ N : m ∈ M, n ∈ N, µ(m) = ν(n)}

By the universal property of fibered product on Set, the diagram with the set-function ϕ(z) :=
(fM (z), fN (z)) makes the following diagram

fN
P
∃!ϕ

M ×Z N πN N
fM πM ν

M µ Z

commutes, regarding in Set. Now we check that M ×Z N indeed is a submodule of M ⊕ N : for

(m, n) ∈ M ×Z N , r(m, n) = (rm, rn), and since µ(m) = ν(n), rµ(m) = µ(rm) = ν(rn) = rν(n), so
(rm, rn) ∈ M ×Z N as required.
Now it remains to check ϕ is a R-module homomorphism, which is direct. 

Problem III.6.11. Define a notion of fibered coproduct of two R-modules M, N , along an R-module
A, in the style of Exercise III.6.10 (and cf. I.5.12).
Prove that fibered coproducts exist in R-Mod. The fibered coproduct M ⊕A N is called the
push-out of M along ν (or of N along µ).

Proof. The universal property is as the same stated in I.5.12, but by replacing every set with R-
modules and every morphism with R-Mod homomorphisms. We now show that the fibered coproduct
is almost the same in Set: define an equivalence relation

S = {(µ(x), ν(x)) ∈ M ⊕ N : x ∈ A}

on M ⊕ N , and let M ⊕A N := (M ⊕ N )/S. We show that R is a submodule, so the quotient make

sense. For (m, n) ∈ S,

r(m, n) = r(µ(x), ν(x)) = (rµ(x), rν(x)) = (µ(rx), ν(rx)) ∈ S

CHAPTER III. RINGS AND MODULES 33

which shows that S is indeed an R-module. Now define

ϕ((m, n) + R) = fM (m) + fN (n)

It is a simple check that ϕ is a R-module homomorphism, and ϕ is well-defined, using the same
argument as in Set(I.5.12). This makes the following diagram
ν
A N
µ iN
fN

M iM
M ⊕A N
∃!ϕ

fM
Z

commutes, as we check:
• iN ν = iM µ:
iN ν(x) = (0, ν(x)) + S = (µ(x), 0) + S = iM µ(x)

• fM = ϕiM (resp. fN = ϕiN ):

ϕiM (m) = ϕ((m, 0) + S) = fM (m).


Problem III.6.14. Prove that the ideal (x1 , x2 , . . . ) of the ring R = Z[x1 , x2 , . . . ] is not finitely
generated (as an ideal, i.e. as an R-module).
Proof. If it were, then there exists a surjective R-Mod homomorphism

ϕ : R⊕n  (x1 , x2 , . . . ).

Then we collect the polynomials

{ϕ(0, . . . , 1 , . . . , 0)}ni=1
i-th place

Since each polynomials can only contain finitely many indeterminates, and there are only finite poly-
nomials, there must be some indeterminates xj that is not in the domain of ϕ (as there are countably
many indeterminates in the ideal), contradicting to the surjectivity of ϕ. Therefore (x1 , x2 , . . . ) is
not finitely generated. 
Problem III.6.16. Let R be a ring. A (left-)R-module M is cyclic if M = hmi for some m ∈ M .
Prove that simple modules (cf. Exercise III.5.4) are cyclic. Prove that an R-module M is cyclic if
and only if M ∼
= R/I for some (left-)ideal I. Prove that every quotient of a cyclic module is cyclic.
Proof. By the universal property of free module there is a unique homomorphism of R-modules

ϕ : R{m} → M

Since M is simple, we can only have ϕ(R{m} ) = 0 or ϕ(R{m} ) = M . We definitely can’t have ϕ = 0
unless m = 0, so ϕ(R{m} ) = M = hmi for m 6= 0.
If M = hmi, then we define a R-module homomorphism ϕ : R → M by ϕ(r) = rm. It is
surjective by construction, and we have M ∼ = R/ ker ϕ. Conversely if M ∼ = R/I, then there is a
surjective R-module homomorphism ϕ : R → M such that its kernel is I. By identifing R with
R{m} , the result is now clear.
The last statement follows from that you can restrict a surjective map ϕ : R → M to another
surjective map ϕ0 : R → M/N . 
CHAPTER III. RINGS AND MODULES 34

Problem III.6.17. Let M be a cyclic R-module, so that M ∼

= R/I for a (left-)ideal I, and let N
be another R-module.

• Prove that HomR-Mod (M, N ) ∼

= {n ∈ N : (∀a ∈ I), an = 0}.
• For a, b ∈ Z, prove that HomR-Mod (Z/aZ, Z/bZ) ∼
= Z/ gcd(a, b)Z.
Proof. Every homomorphism ϕ : M → N is begin fixed by the generator of M , so we only need to
investigate ϕ(m) where M = hmi. To make ϕ into an R-module homomorphism, we must have

ϕ(a + I + b + I) = ϕ(a + I) + ϕ(b + I) and rϕ(a + I) = ϕ(ra + rI)

In particular, let ϕ(m + I) = n, where m + I is the element identified by the generator m. Clearly
we must have ϕ(I) = 0, and for r ∈ R we have

rϕ(m + I) = ϕ(rm + I) = rn

so if rm ∈ I, i.e. r ∈ I, then rn = 0. So the set of all possibe ϕ(m) coincide with the set on the
right, showing the isomorphism. Now

HomR-Mod (Z/aZ, Z/bZ) ∼

= {n ∈ Z/bZ : (∀a ∈ aZ)an = 0}

which is precisely Z/ gcd(a, b)Z. 

Problem III.6.18. Let M be an R-module, and let N be a submodule of M . Prove that if N and
M/N are both finitely generated, then M is finitely generated.

Proof. Let {ai + N }m n

i=1 be generators of M/N , and {bi }i=1 be generators of N . Then for every
m ∈ M , we consider
X m m
X
m+N = ri (ai + N ) = r i ai + N
i=1 i=1

this says that m − i=1 ri ai ∈ N , and therefore we can again write m − m

Pm P Pn
i=1 ri ai = j=1 si bi . To
this point we showed that every element in M can be generated by {ai , bj }1≤i≤m,1≤j≤n , showing that
M is finitely generated. 

III.7
Problem III.7.1. Assume that the complex

··· 0 M 0 ···

is exact. Prove that M ∼

= 0.
Proof.
0 = im(0 −→ M ) = ker(M −→ 0) = M.


Problem III.7.2. Assume that the complex

··· 0 M M0 0 ···

is exact. Prove that M ∼

= M 0.
Proof. The map (M −→ M 0 ) is both a monomorphism and an epimorphism by Example III.7.1 and
Example III.7.2. By definition, the map is an isomorphism. 
CHAPTER III. RINGS AND MODULES 35

Problem III.7.3. Assume that the complex

ϕ
··· 0 L M M0 N 0 ···

is exact. Show that, up to natural identifications, L = ker ϕ and N = coker ϕ.

Proof. The map (L −→ M ) is a monomorphism, so by canonical decomposition
L ∼
L= = im(L −→ M ) = ker(M −→ M 0 ) = ker ϕ.
ker(L −→ M )

The map (M 0 −→ N ) is an epimorphism, so it follows by first isomorphism theorem that

M0 M0 M0 ∼
coker ϕ = = = = N.
im ϕ im(M −→ M 0 ) ker(M 0 −→ N )

Problem III.7.6. Prove the ’split epimorphism’ part of Proposition 7.5, that is,ϕ has a right-inverse
if and only if the sequence
ϕ
0 ker ϕ M N 0

splits.
Proof.
(⇐) If the sequence splits, then by identifying ϕ with the projection map from ker ϕ ⊕ N to N , we
can let ψ : N → ker ϕ ⊕ N to be the inclusion, and it gives a right-inverse.
(⇒) Assume that ϕ has a right inverse, which says that
ψ
N M
ϕ
id
N

To prove the statement, we claim that M ∼

= ker ϕ ⊕ N . This isomorphism is given by

(k, n) 7→ k + ψ(n)

it has inverse
m 7→ (m − ψϕ(m), ϕ(m))
Indeed, we check
m 7→ (m − ψϕ(m), ϕ(m)) 7→ m − ψϕ(m) + ψϕ(m) = m
and m − ψϕ(m) is in ker ϕ since

ϕ(m − ψϕ(m)) = ϕ(m) − ϕψϕ(m) = 0

and the claim is proved. 

Problem III.7.7. Let
0 M N P 0
be a short exact sequence of R-modules, and let L be an R-module.
(i) Prove that there is an exact sequence

0 HomR-Mod (P, L) HomR-Mod (N, L) HomR-Mod (M, L).

CHAPTER III. RINGS AND MODULES 36

(ii) Redo Exercise III.6.17.

(iii) Construct an example showing that the rightmost homomorphism in (i) need not to be onto.
(iv) Show that if the original sequence splits, then the rightmost homomorphism in (i) is onto.
Proof.
(i)
β α
0 M N P 0
a(x)=x◦α b(y)=y◦β
0 HomR-Mod (P, L) HomR-Mod (N, L) HomR-Mod (M, L)
With definition as above, we show that
ker a = 0 and im a = ker b.
Clearly a is injective: if a(x) = a(y), then x ◦ α = y ◦ α, and since α is an epimorphism by
exactness, x = y. For the second part, note by exactness
P ∼
= N/ ker α = N/ im β = coker β
which leads us to consider the universal property of cokernels
0

β y
M N L
α
∃!ϕ

P ∼
= coker β

so for y ∈ ker b, i.e. y ◦ β = 0, by universal property of cokernel, there is a unique ϕ : P → L

such that ϕ ◦ α = y, which shows ker b ⊆ im a. Since the other inclusion is clear (by definition
of chain complex), we conclude that im a = ker b.
(ii) The exact sequence

0 I i
R π
R/I ∼
=M 0

with the target R-module N yields another exact sequence

0 HomR-Mod (M, N ) HomR-Mod (R, N ) HomR-Mod (I, N )

by canonical decomposition
G ∼
G= = im a = ker b
ker a
and ker b is
ker ϕ = {φ(x) = nx ∈ HomR-Mod (R, N ) : φ ◦ i = 0HomR-Mod (I,N ) } ∼
= {n ∈ R : (∀a ∈ I) an = 0}
as required.
(iii) The example
·2 π
0 Z Z Z/2Z 0
yields the exact sequence (with target Z)
·2
0 HomR-Mod (Z/2Z, Z) HomR-Mod (Z, Z) HomR-Mod (Z, Z)

but the last morphism is not surjective as all morphism that is of form ϕ(x) = nx where n is
odd is missing in the first HomR-Mod (Z, Z).
CHAPTER III. RINGS AND MODULES 37

(iv) The map

φ◦i
HomR-Mod (M ⊕ P, L) −→ HomR-Mod (M, L)
is clearly surjective: for each ϕ : M → L, we can find ϕ0 : M ⊕ P → L defined by ϕ0 ((m, p)) =
ϕ(m), so that the restriction of ϕ0 on M is precisely ϕ.

Problem III.7.8. Prove that every exact sequence

0 M N F 0

of R-modules, with F free, splits.

Proof. By exactness, ϕ : N −→ F is surjective. Therefore by III.6.9, for every α : F → F , there is

β : F → N such that α = ϕ ◦ β. In particular, let α = idF , then ϕ ◦ β = idF .

F
β
1
i ϕ
0 M N F 0

With this, we now show that M ⊕ F ∼

= N . Define

h : M ⊕ F → N, h(m, f ) = i(m) + β(f )

h is clearly an R-module homomorphism, so it remains to show that it is an isomorphism. h is

injective: if h(m, f ) = 0, then

i(m) + β(f ) = 0 ⇒ ϕi(m) + ϕβ(f ) = 0 ⇒ 0 (definition of chain complex) + f = 0

showing that f = 0. Then i(m) = 0, so we must have m = 0. h is surjective: we want to find m, f

such that i(m) + β(f ) = n for n ∈ N . By applying ϕ we have

ϕi(m) + ϕβ(f ) = 0 + f = ϕ(n)

so we have the candidate of f . Now it remains to decide m in which i(m) = n − β(ϕ(n)): notice
that by exactness, im i = ker ϕ, so we check that ϕ(n − β(ϕ(n))) = 0 to guarantee the existence of
m:
ϕ(n − β(ϕ(n)) = ϕ(n) − ϕ ◦ β ◦ ϕ(n) = ϕ(n) − ϕ(n) = 0
Hence h is an isomorphism, and by definition, the sequence splits. 
Chapter IV

Groups, second encounter

Unless otherwise specified, in the following G denotes a group, e denotes the identity of G.
The conjugacy class of an element g is denoted by [g]. Some description and hints are omitted for
simplicity.
Unless otherwise specified, all groups in this chapter are finite.

IV.1
Problem IV.1.1. Let p be a prime integer, let G be a p-group, and let S be a set such that
|S| =
6 0 mod p. If G acts on S, prove that the action must have fixed points.
Proof. This is direct by Corollary IV.1.3: since |S| 6= 0 mod p, the set of fixed points Z satisfies
|S| ≡ |Z| =
6 0. 
Problem IV.1.4. Let G be a group, and let N be a subgroup of Z(G). Prove that N is normal in
G.
Proof. For g ∈ G, n ∈ N ,
gng −1 = gg −1 n = n ∈ N.
One should note that normal is not transitive: if G E H and H E I, it is in general not true that
G E I. 
Problem IV.1.5. Let G be a group. Prove that G/Z(G) is isomorphic to the group Inn(G) (II.6.7).
Then prove Lemma 1.5 again.
Proof. Let ϕ : G → Inn(G), ϕ(g) = γg (a) := gag −1 be a homomorphism (II.4.8). By construction it
is clearly surjective, and the kernel is

ker ϕ = {g : gag −1 = a} ⇒ {g : ga = ag} = Z(G)

therefore by first isomorphism theorem, G/Z(G) ∼

= Inn(G). If G/Z(G) is cyclic, then by II.6.7 G is
commutative. 
Problem IV.1.6. Let p, q be prime integers, and let G be a group of order pq. Prove that either G
is commutative or the center of G is trivial. Conclude that every group of order p2 , for a prime p, is
commutative.
Proof. The subgroups can only be of order 1, p, q or pq by Lagrange, and |Z(G)| can be one of these
four. If |Z(G)| = 1, then there is nothing to prove; if |Z(G)| = p(or q), then the quotient is cyclic, so
it follows by Lemma IV.1.5 that G is commutative; if |Z(G)| = pq, then G is clearly commutative.
By Corollary IV.1.9, the center of a nontrivial p-group is nontrivial, so the order of the center for
|G| = p2 can not be 1. Then by above, all the remaining cases will conclude that G is commutative.


38
CHAPTER IV. GROUPS, SECOND ENCOUNTER 39

Problem IV.1.8. Let p be a prime number, and let G be a p-group: |G| = pr . Prove that G
contains a normal subgroup of order pk for every nonnegative k ≤ r.

Proof. We proceed by induction. If r = 1 then there is nothing to prove, so we assume that for
n < r, the p-group with order pn has a normal subgroup of order pk for k ≤ n.
Now consider the center of G: it is abelian and is a nontrivial p-group by Corollary IV.1.9, so by
II.8.20, there exists a (normal) subgroup N that is of order p in Z(G). By IV.1.4, N is normal in G,
so we can consider the quotient G/N . The quotient is a p-group and has order pr−1 , so by induction
hypothesis, G/N has normal subgroups of order pk for k ≤ r − 1, which we name them Hk for each
k. By noting that Hk contains N , we can identify each Hk by Hk /N via Proposition II.8.9. Finally,
since |Hk /N | = pk , |Hk | = pk+1 , so we’ve found normal subgroup of order pk for k ≤ r, proving the
statement. 

Problem IV.1.9. Let p be a prime number, G a p-group, and H a nontrivial normal subgroup of
G. Prove that H ∩ Z(G) 6= {e}.

Proof. Let G act on itself by conjugation. Since H is normal, it is the union of some conjugacy class
and some element of Z(G), with each conjugacy class of order pn for some n by Corollary II.9.10. If
H ∩ Z(G) = {e}, then this means that H only take e from Z(G), and since the order of all conjugacy
classes in H are divisible by p, we would arrive at |H| ≡ 1 mod p, a contradiction since |H| must
be a multiple of p. 

Problem IV.1.10. Prove that if G is a group of odd order and g ∈ G is conjugate to g −1 , then
g = e.

Proof. Suppose g 6= e. Since [g] contains g −1 , there are two cases:

• If g = g −1 , then g 2 = 1, so |g| = 2. But this is impossible since |g| does not divide |G|, a
contradiction.

• If g 6= g −1 , then since [g] must be odd order, there is some y ∈ [g] such that g = xyx−1 . But
this implies g −1 = xy −1 x−1 , so y −1 ∈ [g], and y 6= y −1 by above. So this says that [g] must
contain even number of elements(so must have even order), which again is impossible.

By above, we must have g = e, proving the assertion. 

Problem IV.1.14. Let G be a group, and assume [G : Z(G)] = n is finite. Let A ⊆ G be any
subset. Prove that the number of conjugates of A is at most n.

Proof. We claim that there is a surjective set function from G/Z(G) to {gAg −1 }g∈G . Define

ϕ : G/Z(G) → {gAg −1 }g∈G , ϕ(gZ) = gAg −1

We check that it is well defined: If gZ = hZ, then gh−1 ∈ Z. Now for any element α = gAg −1
we have α = gag −1 for some a ∈ A, so we have g −1 αg = a, and hg −1 αgh−1 = hah−1 . Since
gh−1 ∈ Z, hg −1 αgh−1 = hg −1 gh−1 α = α, so α ∈ hAh−1 , hence gAg −1 = hAh−1 , which showed the
well-definedness. Clearly the map is surjective by construction, and by above, there can be only at
most [G : Z(G)] = n distinct conjugates of A, which proved the assertion. 

Problem IV.1.17. Let H be a proper subgroup of a finite group G. Prove that G is not the union
of the conjugates of H.

Proof. By Lemma IV.1.13, the numbers of conjugates of H is [G : NG (H)]. Since H ⊆ NG (H),

[G : NG (H)]|H| ≤ [G : H]|H| = |G|. Even if the equality might hold, by noting that every
conjugate is a subgroup and e is a common element for all subgroup, there are in fact at most
([G : NG (H)]|H| − |H| + 1) distinct elements in the union of all conjugates of H. Since this number
is strictly less than G, G will never be the union of conjugates of H. 
CHAPTER IV. GROUPS, SECOND ENCOUNTER 40

Problem IV.1.18. Let S be a set endowed with a transitive action of finite group G, and assume
|S| ≥ 2. Prove that there exists a g ∈ G without fixed points in S, that is, such that gs 6= s for all
s ∈ S.
Proof. In the sense of Proposition II.9.9, we can assume that S = G/H (left cosets, not quotient! )
where H = StabG (s) for some s ∈ S, with H proper in G (as |S| ≥ 2). Suppose the contrary, i.e.
every g satisfies gkH = kH for some k. This means k −1 gk ∈ H, or equivalently, g ∈ kHk −1 . So
every element in G is in some conjugacy class of H, which is a contradiction to IV.1.17 that G cannot
be exhausted by conjugates of H. Hence G must have some elements that has no fixed points on S,
as desired. 
Problem IV.1.21. Let H, K be subgroups of a group G, with H ⊆ NG (K). Verify that the function
γ : H → AutGrp (K) defined by conjugation is a homomorphism of group and that ker γ = H ∩ZG (K),
where ZG (K) is the centralizer of K.
Proof. Let γ maps h to an automorphism ϕh (k) = hkh−1 . It is a group homomorphism since

γ(g)γ(h) 7→ ϕg ϕh (k) = ghkh−1 g −1 = ϕ(gh) 7→ γ(gh).

The kernel of this map is

ker γ = {h ∈ H : hkh−1 = k ∀k ∈ K} = {h ∈ H : hk = kh ∀k ∈ K} = H ∩ ZG (K).


Problem IV.1.22. Let G be a finite group, and let H be a cyclic subgroup og G of order p. Assume
that p is the smallest prime dividing the order of G and that H is normal in G. Prove that H is
contained in the center of G.
Proof. In the sense of IV.1.21, we have a homomorphism γ : G → AutGrp (H) since H ⊆ NG (G) = G.
By II.4.14, AutGrp (H) has order φ(p) = p − 1. But since G does not contain an element of order
p − 1 by the minimality of p, γ can only be the trivial homomorphism, so it has kernel equal to G.
But by IV.1.21, ker γ = G ∩ ZG (H) = ZG (H), so we must have ZG (H) = G, which means that the
element that commutes with h are the whole G, i.e. H ⊆ Z(G), as desired. 

IV.2
Problem IV.2.1. Prove Claim 2.2: Let G be a finite group, let p be a prime divisor of |G|, and let
N be the number of cyclic subgroups of G of order p. Then N ≡ 1 mod p.
Proof. We proceed with the same argument as in Theorem IV.2.1. Let S be a set that collects the
p-tuple
(a1 , . . . , ap )
such that a1 · · · ap = 1. It is clear that |S| = |G|p−1 , and since a2 · · · ap a1 = 1, we can consider the
action of Z/pZ on S, by

αm : (a1 , . . . , an ) 7→ (am+1 , . . . , ap , a1 , . . . , am )

By Corollary IV.1.3, |Z| ≡ |S| mod 0, where Z is the fixed points under Z/pZ. The fixed points
are of form (a, . . . , a) for a ∈ G, and since (e, . . . , e) ∈ Z and p divides |Z|, |Z| > 1. Now notice that
for each a ∈ G such that (a, . . . , a) ∈ Z, a is a generator for some cyclic group of order p, so there
are N (p − 1) + 1(identity) elements in Z. But since |Z| ≡ 0 mod p, we have

Np − N + 1 ≡ 0 mod p =⇒ N ≡ 1 mod p

as desired. 
CHAPTER IV. GROUPS, SECOND ENCOUNTER 41

Problem IV.2.2. Let G be a group. A subgroup H of G is characteristic if ϕ(H) ⊆ H for every

automorphism ϕ of G.
• Prove that every characteristic subgroups are normal.
• Let H ⊆ K ⊆ G, with H characteristic in K and K normal in G. Prove that H is normal in
G.
• Let G, K be groups, and assume that G contains a single subgroup H isomorphic to K. Prove
that H is normal in G.
• Let K be a normal subgroup of a finite group G, and assume that |K| and |G/K| are relatively
prime. Prove that K is characteristic in G.
Proof.
• Consider γg (h) := ghg −1 for all g ∈ G. Then gHg −1 ⊆ H by characteristic property of H, so
H is normal.
• By normalness of K, we have gKg −1 = K, so γg is an automorphism on K. Then since
γg (H) ⊆ H, gHg −1 ⊆ H, so H is normal.
• Let ϕ be any automorphism of G. Then ϕ(H) ∼ = H ∼= K since ϕ is an isomorphism. But
since H is the only subgroup that is isomorphic to K, ϕ(H) = H, so H is characteristic, hence
normal.
• Let ϕ be any automorphism of G, and let π : G → G/K be the quotient homomorphism.
Let K 0 = ϕ(K). Then π(K 0 ) is a subgroup of G/K, so |π(K 0 )| divides |G/K|. Also, by first
isomorphism theorem, K 0 / ker π ∼
= im π = π(K 0 ), so |π(K 0 )| divides |K 0 | = |K|. Since |K| and
|G/K| are relatively prime, we can only have |π(K 0 )| = 1, i.e. π(K 0 ) = eG/H . Combining with
ker π = K, we have
ϕ(K) = K 0 ⊆ ker π = K
as desired.

Problem IV.2.4. Prove that a nontrivial group G is simple if and only if its only homomorphic
image are the trivial group and G itself (up to isomorphism).
Proof. (⇒) Let ϕ : G → G0 be a surjective homomorphism. By first isomorphism theorem,
G/ ker ϕ ∼
= G0 . But since kernel is a normal subgroup, the only possibility of G0 are G/{e} = G or
G/G = {e}.
(⇐) If G is not simple, i.e. there are some nontrivial normal subgroup of G, which we call it H,
then ϕ : G → G/H, g 7→ gH is a surjective homomorphism, and G/H is neither {e} nor G(up to
isomorphism), a contradiction. 
Problem IV.2.5. Let G be a simple group, and assume ϕ : G → G0 is a nontrivial group homo-
morphism. Prove that ϕ is injective.
Proof. ker ϕ can only be {0} or G by simpleness. If ker ϕ = {0} the we are done; if ker ϕ = G then
ϕ = 0, which can’t be by hypothesis. 
Problem IV.2.6. Prove that there are no simple groups of order 4, 8, 9, 16, 25, 27, 32 or 49. In fact,
prove that no p-group of order ≥ p2 is simple.
Proof. The center of p-group, by Corollary IV.1.9, is nontrivial. Since center is a normal subgroup,
no group of order pn for n ≥ 2 is simple. 
Problem IV.2.8. Let G be a finite group, p a prime integer, and let N be the intersection of the p-
Sylow subgroups of G. Prove that N is a normal p-subgroup of G and that every normal p-subgroup
of G is contained in N .
CHAPTER IV. GROUPS, SECOND ENCOUNTER 42

Proof. Let P be a p-Sylow, then we can let N = g∈G gP g −1 . The conjugate of N is pN p−1 =
T
−1 0
T
g∈G pgP (pg) , which is again N , so N is normal. Now if N is a normal p-subgroup, then by Sylow
II we can assume that N ⊆ P . Then for all g ∈ G, N = gN g ⊆ gP g , so N ⊆ g∈G gP g −1 = N ,
0 0 −1 −1 0
T
and N 0 is in N , as required. 

Problem IV.2.9. Let P be a p-Sylow subgroup of a finite group G, and let H ⊆ G be a p-subgroup.
Assume H ⊆ NG (P ). Prove that H ⊆ P .

Proof. By noting that P is normal in NG (P ) (Remark IV.1.12), we consider P H, which is a subgroup

of NG (P ) by Proposition II.8.11. Then by second isomorphism theorem

PH ∼ H
=
P P ∩H
|P ||H|
Now |P H| = |P ∩H|
by II.8.21, and since either |P ∩ H| = 1 or |H| by Sylow II, P H is a p-
group, and it must be P since P is the maximal p-subgroup of G. Then we have H ⊆ P since
PH = P ⇔ H ⊆ P. 

Problem IV.2.10. Let P be a p-Sylow subgroup of a finite group G, and act with P by conjugation
on the set of p-Sylow subgroups of G. Show that P is the unique fixed point of this action.

Proof. Let S be the collection of p-Sylow subgroups of G, and let P act on S by conjugation. If H
is any p-Sylow that is fixed by P , then we have H ⊆ NG (P ) (P HP −1 = H ⇒ HP H −1 = P ), so we
can apply IV.2.9 and obtain H ⊆ P . But by Sylow II, H must be P , proving the statement. 

Problem IV.2.12. Let P be a p-Sylow subgroup of a finite group G, and let H ⊆ G be a subgroup
containing the normalizer NG (P ). Prove that [G : H] ≡ 1 mod p.

Proof. By Sylow III, [G : NG (P )] ≡ 1 mod p. Since H contains P , P is also a p-Sylow of H.

Since H ⊇ NG (P ), the normalizer of P in H is also NG (P ), so NH (P ) = NG (P ). Then clearly
[G : NG (P )] = [G : NH (P )] ≡ 1 mod p. Finally

[G : NG (P )] [G : NG (P )]
[G : H] = =
[H : NG (P )] [H : NH (P )]

and since both numerator and the denominator are both congruent to 1 mod p, [G : H] ≡ 1
mod p. 

Problem IV.2.13. Let P be a p-Sylow subgroup of a finite group G.

• Prove that if P is normal in G, then it is in fact characteristic in G.

• Let H ⊆ G be a subgroup containing the Sylow subgroup P . Assume P is normal in H and
H is normal in G. Prove that P is normal in G.
• Prove that NG (NG (P )) = NG (P ).

Proof.

• Since gcd(|P |, |G/P |) = 1 as P is Sylow, by the 4th point of IV.2.2, P is characteristic in G.

• By above, P is characteristic in H, so by 2nd point of IV.2.2, P is normal in G.
• We have the normal chain
P E NG (P ) E NG (NG (P ))
and by above, P is normal in NG (NG (P )), so for any g ∈ NG (NG (P )), gP g −1 = P , i.e. g ∈
NG (P ). Since the other inclusion is clear, we conclude that NG (NG (P )) = NG (P ).


CHAPTER IV. GROUPS, SECOND ENCOUNTER 43

Problem IV.2.16. Let G be a noncommutative group of order 8. Prove that G ∼

= D8 or G ∼
= Q8 .
Proof.

• G does not contain an element of order 8 as it would be C8 by II.4.3; if all nonidentity element
are of order 2, then
gh = (gh)−1 = h−1 g −1 = hg
so G is commutative.
• Let y be an element of order 4: note that hyi must be normal (II.8.2). So pick an element
x∈/ hyi, and since hyi ( hx, yi, we must have hx, yi = G by order consideration. Also G/hyi
has order 2, so x2 ∈ hyi, and x can be order 4 (which implies x2 = y 2 ) or order 2 (x2 = 1).
• Let G = {e, y, y 2 , y 3 , x, yx, y 2 x, y 3 x}. We can determine the multiplication table of G by
directly computing all possible products of any two elements.
• xyx−1 is an element of order 4, and by normalness

xyx−1 ∈ {e, y, y 2 , y 3 }

so we must have xyx−1 = y 3 by order consideration (xyx−1 = y would imply that xy = yx,
which clearly isn’t), and rearrangement gives xy = y 3 x.
• If we set x2 = e, then there is isomorphism

ϕ : G → D8 , ϕ(y a xb ) = ra sb

if we set x2 = y 2 , then there is isomorphism

ϕ : G → Q8 , ϕ(y a xb ) = ia j b

It is a routine check that both are indeed homomorphisms and bijections.

Problem IV.2.17. Let R be a division ring, and assume that |R| = 64. Prove that R is necessarily
commutative (hence, a field).

Proof.

• By excluding 0, the units of R form a group of order 63, and it contains a 3-Sylow which is of
order 9. Note that such subgroup is commutative since p2 groups are commutative (IV.1.6).
Call this group G.
• The possible order of subrings of R are 1, 2, 4, 8, 16, 32, 64. Since it contains G, such sub-division
ring must contain G as its subgroup. But by excluding 0, the possible order of subgroups formed
by units are
1, 3, 7, 15, 31, 63
and only 63 divides 9. So the only subring that contains G is R.
• However, every centralizer of a division ring forms a division ring (III.2.10), so all elements
that commutes with G must form a sub-division ring.
• Since the only sub-division ring of R is R itself, G commutes with all elements of R, hence
contained in the center of R. But again since center forms a subring (III.2.9), center must also
be R itself since it contains G. This shows that R is commutative, proving the assertion.


CHAPTER IV. GROUPS, SECOND ENCOUNTER 44

IV.3
Problem IV.3.1. Prove that Z has normal series of arbitrary length.
Proof. If Z has a normal series Z ) · · · ) nZ, then we have an extended normal series Z ) · · · )
nZ ) 2nZ, which can be further extended by infinitely times. 
Problem IV.3.3. Prove that every finite group has a composition series. Prove that Z does not
have a composition series.
Proof. Proceed by induction, since groups of order 1 has a composition series, assume that for a
given positive integer n, all groups that has order less than n admits a composition series. Then for
|G| = n, it suffices to show that there exists some normal subgroup H such that G/H is simple, and
the rest follows from induction.
Indeed, we can let H be the largest normal subgroup (in the sense that if H is normal in H 0 ,
then H 0 = H, assuming H 0 proper). Then in view of Proposition II.8.9, G/H is simple if and only
if H is the largest normal subgroup of G, as required.
Z does not have a composition series: If there were one, then all decomposition factors are of the
form dZ/dpZ ∼ = Z/pZ where p is a prime. Then we can clearly write
n
Y
{0} = pi Z ( · · · ( p1 p2 p3 Z ( p1 p2 Z ( p1 Z ( 1Z = Z
i=1

for pi being primes. But this is absurd since product of primes will never be zero. 
Problem IV.3.4. Find an example of two nonisomorphic groups with the same decomposition
factors.
Solution. The groups D8 and C8 both has C2 and C4 as their normal subgroups, so there are series

D8 . C4 . C2 . {e}
C8 . C4 . C2 . {e}


Problem IV.3.5. Show that if H, K are normal subgroups of a group G, then HK is a normal
subgroup of G.
Proof. For hk ∈ HK, g ∈ G,

ghkg −1 = ghg −1 gkg −1 ∈ (gHg −1 )(gKg −1 ) = HK

so gHKg −1 = HK, hence normal. 

Problem IV.3.8. Prove Lemma 3.7: Let ϕ : G1 → G2 be a group homomorphism. Then ∀g, h ∈ G1
we have
ϕ([g, h]) = [ϕ(g), ϕ(h)]
and ϕ(G01 ) ⊆ G02 .
Proof.
ϕ([g, h]) = ϕ(ghg −1 h−1 ) = ϕ(g)ϕ(h)ϕ(g −1 )ϕ(h−1 ) = [ϕ(g), ϕ(h)].
The subgroup inclusion is immediate. 
Problem IV.3.10. Let G be a group. Define inductively an increasing sequence Z0 = {e} ⊆ Z1 ⊆
Z2 ⊆ · · · of subgroups of G as follows: for i ≥ 1, Zi is the subgroup of G corresponding (as in
Proposition II.8.9) to the center of G/Zi−1 .
CHAPTER IV. GROUPS, SECOND ENCOUNTER 45

• Prove that each Zi is normal in G, so that this definition make sense.

A group is nilpotent if Zm = G for some m.
• Prove that G is nilpotent if and only if G/Z(G) is nilpotent.
• Prove that p-groups are nilpotent.
• Prove that nilpotent groups are solvable.
• Find a solvable group that is not nilpotent.
Proof.
• Z1 = Z(G) which is clearly normal in G; if Zi−1 is normal in G, then Zi /Zi−1 ∼
= Z(G/Zi−1 ) E
G/Zi−1 , so by Proposition II.8.10, Zi is normal in G.
• If G/Z(G) is nilpotent, then it admits a series

{e} = Z0 ⊆ Z1 · · · ⊆ Zk = G/Z(G)

”multiplies the sequence” by Z(G) gives a series

0
{e} = Z−1 ⊆ Z00 = Z(G) ⊆ · · · ⊆ Zk0 = G

Since it is clear that

G/Z(G) ∼ G Zi ∼ Zi0
= 0 and = 0
Zi−1 Zi−1 Zi−1 Zi−1
0
we have G/Zi−1 ∼ 0
= Zi0 /Zi−1 , so the series satisfies the definition given above, and hence G is
nilpotent. The converse is the same as above but in reverse.
• A group of order p where p is a prime is clearly nilpotent as it is cyclic. If all groups of order
pk−1 is nilpotent, since p-groups have nontrivial center, Z/G(Z) has order less than or equal
to pk−1 , so it is nilpotent, and by above, all groups that is of order pk is nilpotent.
• If G has a series
{e} = Z0 ⊆ Z1 · · · ⊆ Zk = G
then by noting that Zi /Zi−1 ∼
= Z(G/Zi−1 ), all quotient of this series is abelian, hence solvable.
• S3 is solvable, but it has trivial center (cf. Example IV.1.10), so it can’t be nilpotent.

Problem IV.3.15. Let p, q be prime integers, and let G be a group of order p2 q. Prove that G is
solvable.
Proof. If p = q then there is nothing to prove (G is a p-group), so assume two other cases:
• p > q. By Sylow III, G can only have 1 p-Sylow: if there are 1 + p p-Sylows, then we would
have too much elements as (p2 − 1)(p + 1) = p3 + p2 − p > p3 > p2 q. So there is a normal
subgroup H that has order p2 , and [G : H] = q. Since H and G/H are solvable, G is solvable
by Corollary IV.3.13.
• p < q. By Sylow III, the numbers of q-Sylows nq satisfies nq | p2 and nq ≡ 1 mod q. Since p is
a prime, nq can be one of 1, p and p2 . If nq = 1, then we can find a normal subgroup H such
that |H| = q, and [G : H] = p2 , so G is solvable; if nq = p, then we would have p ≡ 1 mod q,
but this can’t happen since p < q; if nq = p2 , then there are

p2 q − p2 (q − 1) = p2

elements outside the union of q-Sylows (including e), which can precisely fit in a p-Sylow, so
by the case p > q, G is again solvable.
CHAPTER IV. GROUPS, SECOND ENCOUNTER 46

Therefore for |G| = p2 q, G is solvable. One should note that this is still true for all groups G such
that |G| = pn q where n is a positive integer. This can be proved by induction on n, and it follows
the same pattern as above. 
Problem IV.3.16. Prove that every group of order < 120 and 6= 60 is solvable.
Proof. There are several tests to check that a group is solvable:
(i) p-groups (Example IV.3.12);
(ii) pq groups (Corollary IV.3.13);
(iii) pn q groups (IV.3.15);
(iv) pqr groups: we will give a proof later;
(v) ”Exceptions”: 36, 72, 84, 90, 100, 108.
This gives the following fancy chart (60 has A5 as an exception, and 120 has S5 ).
1 2 3 4 5 6 7 8 9 10 11 12
13 14 15 16 17 18 19 20 21 22 23 24
25 26 27 28 29 30 31 32 33 34 35 36
37 38 39 40 41 42 43 44 45 46 47 48
49 50 51 52 53 54 55 56 57 58 59 60 !
61 62 63 64 65 66 67 68 69 70 71 72
73 74 75 76 77 78 79 80 81 82 83 84
85 86 87 88 89 90 91 92 93 94 95 96
97 98 99 100 101 102 103 104 105 106 107 108
109 110 111 112 113 114 115 116 117 118 119 120 !
Now we handle the special cases. Let np denotes the numbers of p-Sylow subgroup:
• 36 = 22 · 32 : We can have n3 = 1 or 4. The former would give a decomposition 9 × 4, and for
the latter we consider the action by conjugation on 3-Sylows; this induces a homomorphism
ϕ : G → S4 , and hence a homomorphism G/ ker ϕ ,→ S4 . Finally ker ϕ is not trivial since
36 > 24, so | ker ϕ| > 1, and |G/ ker ϕ| ≤ 18. It then follows that the quotient (and the kernel)
is solvable by the table.
• 72 = 23 · 32 : We can have n3 = 1 or 4. The former would give a decomposition 18 × 4, and the
latter case is the same as in the case 36.
• 84 = 22 · 3 · 7: n7 must be 1 since (1 + 7) - 12.
• 90 = 2 · 32 · 5: We only consider the cases where n3 , n5 are not 1. By simple calculation, we
have n5 = 6, n3 = 10. But then if all 3-Sylow intersects trivially, then sum of elements that
has order 3 or 5 is 10(9 − 1) + 6(5 − 1) = 104 > 90, which is too much.
So there is some H, K: 3-Sylows such that |H ∩ K| = 3 (can’t be 9: then H = K). Now
|H||K|
= |HK| = 27
|H ∩ K|
and also
[H : H ∩ K] = [K : H ∩ K] = 3
so H ∩ K is normal in H and K by II.9.11. We ”claim” that H ∩ K is normal in G, by evaluate
the normalizer N = NG (H ∩ K) (cf. Remark IV.1.12). Note that this subgroup includes HK
by normalness (HK(H ∩ K)K −1 H −1 = H(H ∩ K)H −1 = H ∩ K), so the order of N satisfies
|N | ≥ 27, |N | | 90, 9 | |N | (Lagrange on H)
and candidates of |N | are 45 and 90. In the former case we have [G : N ] = 2 so N is normal by
II.8.2, and the latter case implies H ∩ K is normal (Remark IV.1.12). Either way, the quotient
with respect to normal subgroups has order < 45, and by the chart, it is solvable.
CHAPTER IV. GROUPS, SECOND ENCOUNTER 47

• 100 = 22 · 52 : We can only have n5 = 1 since (1 + 5) > 4.

• 108 = 22 · 33 : We can have n3 = 1 or 4. The former would give a decomposition 27 × 4, and
the latter case is the same as in the case 36.
Finally it suffices to prove the following lemma:
Lemma. Let p, q, r be primes such that p > q > r. Then a group that has order pqr is solvable.
Proof. Let us investigate the possibility of different conbination of Sylow subgroups.
• For np , there is nothing to prove if np = 1, and since p is the largest we cannot have np = q or
r, so we must have np = qr.
• For nq , there is nothing to prove if nq = 1, and we cannot have nq = r, so at worse we have
nq ≥ p.
• For nr , there is nothing to prove if nr = 1, so at worse we have nr ≥ q.
Now at worse, G would contain way too much elements as

qr(p − 1) + p(q − 1) + q(r − 1) = pqr − qr + pq − p + qr − q > pqr.

Therefore nk = 1 for some k ∈ {p, q, r}, and the lemma is proved. 

All above finishes the proof. 

IV.4
Problem IV.4.1. Compute the number of elements in the conjugacy class of
 
1 2 3 4 5 6 7 8
8 1 2 7 5 3 4 6

in S8 .
Solution. This permutation is of type (5, 2, 1), so all permutation that is of type (5, 2, 1) is in the
conjugacy class. There are
8·7·6·5·4 3·2
· = 3360
5 2
elements in the conjugacy class of this permutation. 
Problem IV.4.5. Find the class formula for Sn , n ≤ 6.
Solution. The case n = 1, 2, 3, 5 has been done in the book, and the case n = 4 will be done in the
next problem, so we only do n = 6:
Cycle type Counts Cycle type Counts
6! 6!
(6) = 120 (3,1,1,1) = 40
6 3! · 3
6! 6! 4! 2! 1
(5,1) = 144 (2,2,2) · · · = 15
1·5 4! · 2 2! · 2 2 3!
6! 2! 6! 4! 1
(4,2) · = 90 (2,2,1,1) · · = 45
2! · 4 2 4! · 2 2! · 2 2!
6! 6!
(4,1,1) = 90 (2,1,1,1,1) = 15
2! · 4 4! · 2
6! 3! 1
(3,3) · · = 40 (1,1,1,1,1,1) 1
3! · 3 3 2
6! 3!
(3,2,1) · = 120 Sum 720
3! · 3 2

CHAPTER IV. GROUPS, SECOND ENCOUNTER 48

Problem IV.4.6. Let N be a normal subgroup of S4 . Prove that |N | = 1, 4, 12, or 24.

Proof. We only need to prove the case |N | = 4 (12 follows from II.8.2). Note that normal subgroups
are the union of conjugates, so by noting that the class formula

24 = |{z}
1 + |{z}
6 + |{z}
8 + |{z}
3 + |{z}
6
e (ab) (abc) (ab)(cd) (abcd)

we can pick
N = {e, (12)(34), (13)(24), (14)(23)}
and this is indeed normal and of order 4. 
Problem IV.4.7. Prove that Sn is generated by (12) and (12 . . . n).
Proof. It suffices to get all transpositions. Denote τ = (12 . . . n), and note τ −1 = (n n − 1 . . . 1).
First we observe that
τ (12)τ −1 = τ −1 (12) = (n1)
Then we replace (12) with (n1), we obtain (n; n − 1). Continuing this process, we obtain all trans-
positions that is of type (k k + 1) for 1 ≤ k < n and (n1). Now we form all transpositions of type
(1n), by observing
(13) = (23)(12)(32)
and replace (12) by (13) obtains (14), so we have all transpositions of type (1n). Finally we can
form any transpositions via
(mn) = (1m)(1n)(m1)
therefore Sn is generated by (12) and (12 . . . n). 
Problem IV.4.10.
• Prove that there are exactly (n − 1)! n-cycles in Sn .
• More generally, find a formula for the size of the conjugacy class of a permutation of given
type in Sn .
Proof. There are n! way to arrange n elements in a line, but since cycles are invariant under ”rota-
tion”, i.e.
(12 · · · n) = (n12 · · · n − 1) = · · · = (23 · · · n1)
and there are n repeated cycles (including itself) for each distinct cycle in Sn , so there are (n − 1)!
n-cycles.
Now given a type (t1 , · · · , tk ) where t1 ≤ · · · ≤ tk , the first term has n!/(n − t1 )! choices on
elements, and the second term has (n − t1 )!/(n − t1 − t2 )! choices on elements, etc. Then each ti -cycle
counts its repeated cycle, which is precisely ti , and divide them. So the size is
n! (n − t1 )! tk ! n!
· ··· =
(n − t1 )! t1 (n − t1 − t2 )! t2 tk t1 t2 · · · tk
Finally for repeated choice of cycles (i.e. ti = ti+1 = · · · = ti+m ), we need to divide them by (m + 1)!.
Let ci denotes the count of the number i appearing in (t1 , · · · , tk ), then the final size is
n! 1
· .
t1 t2 · · · tk c1 !c2 ! · · · cn !

Problem IV.4.11. Let p be a prime integer. Compute the number of p-Sylow subgroups of Sp . Use
this result and Sylow’s third theorem to prove again the ’only if’ implication in Wilson’s theorem
(cf. Exercise II.4.16.)
CHAPTER IV. GROUPS, SECOND ENCOUNTER 49

Proof. There are (p − 1)! p-cycles in Sp by IV.4.10, and any of them belongs to some p-Sylow. Since
p-Sylows only intersects at e as p is a prime, if there are m p-Sylow subgroups, then there is m(p − 1)
p-cycles. A simple comparsion gives m = (p − 2)!.
By Sylow III we must have
(p − 2)! ≡ 1 mod p
so
(p − 1)! ≡ p − 1 ≡ −1 mod p
which is the result. 
Problem IV.4.21. Prove that A6 is simple, by using its class formula.
Proof. By excluding all odd cycles from the class formula in IV.4.5, we have

360 = 144 + 90 + 40 + 40 + 45 + 1

the divisors of 360 excluding 1 and 360 are

2, 3, 4, 5, 6, 8, 9, 10, 12, 15, 18, 20, 24, 30, 36, 40, 45, 60, 72, 90, 120, 180

and they must be the sum of the numbers appearing in the class equation, with 1 being a must. But
this is not possible by a simple calculation. 

IV.5
Problem IV.5.1. Let G be a finite group, and let P1 , . . . , Pr be its Sylow subgroups. Assume all
Pi are normal in G.
• Prove that G ∼
= P 1 × · · · × Pr .
• Prove that G is nilpotent.
Proof. A p-Sylow is normal if and only if it is the only p-Sylow in the group. Also, for p 6= q, p, q
being primes, the p-Sylow and q-Sylow have trivial intersection (i.e. {e}), by order consideration.
Therefore by Proposition IV.5.3 and a simple induction, P1 P2 · · · Pr ∼ = P1 × · · · × Pr . Finally, since
|G| = |P1 ||P2 | · · · |Pr |, we conclude that G ∼ P
= 1 2 P · · · P ∼ P
r = 1 × · · · × Pr .
Now let |G| = p1 · · · pr , with Pi being the only pi -Sylow in G (hence normal). Then G ∼ =
P1 × · · · × Pr , and clearly it is nilpotent (cyclic) since it is the product of cyclic groups. Now assume
that for fixed k1 , . . . , kr , all groups that is of order pl11 · · · plrr where li < ki , i = 1, . . . r, and has
only one pi Sylow for each i, is nilpotent. Now assume |G| = pk11 · · · pkr r . Then by observing that
Z(G) ∼ = Z(P1 ) × · · · Z(Pr ), we have
G ∼ P1 Pr
= × ··· ×
Z(G) Z(P1 ) Z(Pr )

This group has order pl11 · · · plrr where li < ki , i = 1, . . . r (p-groups has nontrivial center, so each
quotient has less order), and by noting for groups H and K, H(K) E H × K, we have that Pi /Z(Pi )
is normal for each i. By order consideration, they are the only pi -Sylow for each i. Therefore by
induction hypothesis, G/Z(G) is nilpotent, and by IV.3.10, G is nilpotent, proving the assertion.

Problem IV.5.4. Prove that the sequence
·2
0 Z Z Z/2Z 0

is exact but does not split.

CHAPTER IV. GROUPS, SECOND ENCOUNTER 50

Proof. Exactness is a simple routine check; all subgroups of Z are kZ for integers k, and Z/2Z can’t
be any of them. 

Problem IV.5.5. In Proposition III.7.5 we have seen that if an exact sequence

ϕ
0 M N N/(ϕ(M )) 0

of abelian groups splits, then ϕ has a left-inverse. Is it necessarily the case for split sequence of
groups?

Solution. It is not the case for groups: consider

ϕ
0 C3 S3 C2 0

There are nontrivial maps ϕ by sending 1 to a given 3-cycle. However there are no nontrivial maps
from S3 to C3 : by order consideration, all 2-cycles must map to 0 in C3 , and by noting

(12)(13) = (123) (12)(23) = (132)

we have that 3-cycles must also map to 0. Therefore no nontrivial maps from S3 to C3 exists, and
every nontrivial ϕ can’t have a left-inverse. 

Problem IV.5.6. Prove Lemma 5.8:

The structure (N × H, •θ ) is a group, with identity element (eN , eH ).

Proof. The existence of inverse is already proven in Lemma 5.8; (eN , eH ) is indeed the identity since
for (n, h) ∈ (N × H, •θ ),

(n, h) •θ (eN , eH ) = (nθh (eN ), heH ) = (n, h)

the associativity holds: for (ni , hi ) ∈ (N × H, •θ ), i = 1, 2, 3, by noting that θab (x) = θa (θb (x))
(θ : H → AutGrp (N ) is a homomorphism),

((n1 , h1 ) •θ (n2 , h2 )) •θ (n3 , h3 ) = (n1 θh1 (n2 ), n2 h2 ) •θ (n3 , h3 )

= (n1 θh1 (n2 )θh1 h2 (n3 ), h1 h2 h3 )
= (n1 θh1 (n2 )θh1 (θh2 (n3 )), h1 h2 h3 )
= (n1 θh1 ((n2 )θh2 (n3 )), h1 h2 h3 )
= (n1 , h1 ) •θ (n2 θh2 (n3 ), h2 h3 )
= (n1 , h1 ) •θ ((n2 , h2 ) •θ (n3 , h3 ))

Problem IV.5.8. Prove that if G = N o H is commutative, then G ∼

= N × H.
Proof. It suffices to show that the action θh is trivial for each h, i.e. θh (n) = n for all n ∈ N, h ∈ H.
This can be shown by the identity
!
(n, h) = (nθeH (eN ), h) = (n, eH )(eN , h) = (eN , h)(n, eH ) = (eN θh (n), h) = (θh (n), h)

Abelian property is used in !. 

Problem IV.5.15. Let G be a group of order 28. Prove that there are four groups of order 28 up
to isomorphism.

Proof.
CHAPTER IV. GROUPS, SECOND ENCOUNTER 51

• Notice that there can be only one 7-Sylow as (1 + 2) - 4 and 1 + 4 > 4. Such group (call it N )
must be normal.
• Note AutGrp (N ) ∼
= C6 . There are two possibility for homomorphisms ϕ : C4 → AutGrp (N ):
One being the trivial map, and another being ϕ(1) = (1 → 4) or ϕ(3) = (1 → 4) (|ϕ(1)| =
|(1 → 4)| = 2 must divide |1| = 4). For the case ϕ : C2 × C2 → AutGrp (N ), there is trivial
map, and another map ϕ((1, 0)) = (1 → 4) or ϕ((0, 1)) = (1 → 4).
• Each automorphism determines a groups structure up to isomorphism. In the case where the
automorphism is trivial, we have C4 × C7 and C2 × C2 × C7 .
• Let
N = hr2 i, H = {e, r, s, sr}
with automorphism ϕ : H → Aut(N ) given by the nontrivial automorphism as above, so we
indeed have D28 ∼
= (C2 × C2 ) oϕ C7 . Note that

ϕ : D28 → C2 × D14 , ϕ(r) = (1, r), ϕ(s) = (0, s)

defines an isomorphism, so D28 ∼ = C2 × D14 . The other noncommutative group of order 28 is

called the generalized quaternion group.

IV.6
Problem IV.6.2. Complete the classification of groups of order 8 (cf. Exercise IV.2.16).

Solution. Abelian cases are C8 , C4 × C2 , C2 × C2 × C2 as factorization of 8 tells us; nonabelian cases

are D8 and Q8 as in IV.2.16. 

Problem IV.6.6. How many abelian groups of order 1024 are there, up to isomorphism?

Solution. As 1024 = 210 , we are asking how many ways are there to distribute 10 into unordered
integer partitions. Denote (k, n) to be the number of ways to partition k into unordered integer
partitions, with each partition no less than n and contains at least one partition of size n. We then
see that X
(k, 0) = (i, k − i) + 1
i∈N,i>k/2

For example, we solve n = 6 (cf. IV.4.5; note that this is exactly the count of different conjugacy
classes of S6 ):
(6, 0) = (5, 1) + (4, 2) + (3, 3) + 1
In this case, (5, 1) is the number of cycles that has 1 in it, and every partition must be larger than 1.
So (5, 1) consist of cycles (5, 1), (4, 1, 1), (3, 2, 1), (3, 1, 1, 1), (2, 2, 1, 1), (2, 1, 1, 1, 1) and (1, 1, 1, 1, 1, 1);
(4, 2) has (4, 2), (2, 2, 2), and (3, 3) has (3, 3); finally, there is (6), which is the +1 at the end of the
formula. This counts to a total of 11 elements.
So we need to solve (10, 0). Assuming that (9, 1) = (9, 0) = 30 (which can be shown inductively),
we can calculate the remaining
(8,2) (8,2), (6,2,2), (5,3,2), (4,4,2), (4,2,2,2), (3,3,2,2), (2,2,2,2,2) 7
(7,3) (7,3), (4,3,3) 2
(6,4) (6,4) 1
(5,5) (5,5) 1
Therefore there are 30 + 7 + 2 + 1 + 1 + 1 = 42 ways to partition, so there are 42 different abelian
groups of order 1024, up to isomorphism. 
Chapter V

Irreducibility and factorization in integral

domain

Unless otherwise stated, all rings in this chapter are commutative.

V.1
Problem V.1.1. Let R be an Notherian ring, and let I be an ideal of R. Prove that R/I is a
Notherian ring.

Proof. The projection ϕ : R → R/I is clearly an surjective homomorphism, and by III.4.2 R/I is
Notherian. 

Problem V.1.2. Prove that if R[x] is Notherian, then so is R.

Proof.
π : R[x] → R[x]/(x) ∼
=R
is surjective, and by V.1.1 R is Notherian. 

Problem V.1.4. Let R be the ring of real-valued continuous functions on the interval [0, 1]. Prove
that R is not Noetherian.

Proof. Let {fn }∞

n=1 be continuous functions so that fn has support on [0, 1 − 2
−n
] (i.e. fn = 0 on
−n
(1 − 2 , 1]). Then
(f1 ) ⊆ (f2 ) ⊆ · · · (fn ) ⊆ · · ·
is an increasing sequence of ideals that does not terminate. By Proposition V.1.1, R is not Noetherian.


Problem V.1.6. Let I be an ideal of R[x], and let A ⊆ R be the set defined in the proof of Theorem
1.2. Prove that A is an ideal of R.

Proof. A is a subgroup of (R, +): For a, b ∈ A, there is some f, g ∈ I so that the leading coefficient
of f (resp. g) is a (resp. b). Assume that deg(f ) ≥ deg(g). Then f − xdeg(f )−deg(g) g is an element of
I, and it has leading coefficient a − b, which is in A, so A is a subgroup.
A satisfies absorption property: If a ∈ R, then there is some f ∈ I such that a is the leading
coefficient of f . Then rf ∈ I has leading coefficient ra, which is in A, so ra ∈ A for all r ∈ R.
Therefore A is an ideal. 

Problem V.1.8. Prove that every ideal in a Noetherian ring R contains a finite product of prime
ideals.

52
CHAPTER V. IRREDUCIBILITY AND FACTORIZATION IN INTEGRAL DOMAIN 53

Proof. Suppose there are some ideals that does not contain a finite product of prime ideals. Let us
collect these ideals and form a family F , which clearly is nonempty. Since R is Noetherian, there is
an maximal ideal with respect to inclusion in F , which we call it M . Since M is not prime, there
exists a, b ∈
/ M such that ab ∈ M . Now consider two ideals that are larger than M (so they contain
a finite product of prime ideals):
M + (a), M + (b)
Note that both of them are proper : If M + aR = R, then bM + baR = bR, and since bM + baR ⊆ M
we would have bR ⊆ M , i.e. b ∈ M , a contradiction. Then since

(M + (a))(M + (b)) ⊆ M

and since the product on the left contains a finite product of prime ideals, M contains a finite product
of prime ideals, a contradiction. Therefore F = ∅, and the assertion is proved. 

Problem V.1.12. Let R be an integral domain. Prove that a nonzero a is irreducible if and only if
(a) is maximal among proper principle ideal of R.

Proof.
(⇒) If a is irreducible but there is some b ∈ R such that (a) ⊆ (b), then we can write a = bc for
some c ∈ R. Then either b is a unit, or c is a unit. The former would lead to that (b) = R, and the
latter says that there is also c−1 such that ac−1 = b, so (a) ⊇ (b), so (a) = (b). Either way, (a) is the
maximal amongst all principle ideals.
(⇐) If a = bc, then (a) ⊆ (b). Since (a) is maximal amongst all principle ideal, we must have (b) = R
or (b) = (a). In the former we have that b is a unit, and the latter implies that c is a unit. In both
cases at least one of b and c is a unit, so a is irreducible. 

Problem V.1.17. Consider the subring of C:

√ √
Z[ −5] := {a + bi 5 | a, b ∈ Z}
√
Prove that Z[ −5] is not a UFD.

Proof.
√
• By the same argument as in the 4th point of III.4.10, Z[ −5] ∼ = Z[t]/(t2 − 5).
√
• Z[t] is Noetherian (Z is Noetherian and Hilbert Basis), so Z[t]/(t2 − 5) ∼
= Z[ −5] is Noetherian
√
by V.1.1. Since (t2 + 5) is maximal (hence prime), the quotient Z[t]/(t2 − 5) ∼ = Z[ −5] is a
domain.
√
• The norm N (a + bi 5) = a2 + 5b2 satisfies the multiplicative property by the same argument
as in the 2nd point of III.4.10.
• If an element u is a unit, then we must have N (u)N (u−1 ) = N (1) = 1, and this forces N (u) = 1
as the definition of norm guarantees N (a) ≥ 1 for all nonzero a, so u = ±1.
√ √ √
• If a, b ∈ Z[ −5] satisfies ab = 2 (resp. 3, 1 + i 5, 1 − i 5), then we have N (a)N√(b) = 4 (resp.
9, 6, 6). If N (a) ≥ N (b), then we must have N (a) = 4 (resp. 9, 6, 6) since Z[ −5] does not
contain elements such that N (a) = 2 or 3.
√ √
• 6 = 2 · 3 = (1 + i 5)(1 − i 5).
√
• Since the factorization of 6 is not unique, Z[ −5] is not a UFD.


CHAPTER V. IRREDUCIBILITY AND FACTORIZATION IN INTEGRAL DOMAIN 54

V.2
Problem V.2.1. Prove Lemma 2.1:
Let R be a UFD, and let a,b,c be nonzero elements of R. Then

• (a) ⊆ (b) ⇐⇒ the multiset of irreducible factors of b is contained in the multiset of irreducible
factors of a;
• a and b are associates ⇐⇒ the two multiset coincide;
• the irreducible factors of a product bc are the collection of all irreducible factors of b and c.

Proof.

• The inclusion on the right implies a = bp for some p, so clearly the multiset of a contains the
multiset of b. Conversely, we can let p be the product of difference of the multiset of a and b.
Then a = pb (up to associates), so (a) ⊆ (b).
• A unit u is not a product of irreducibles by definition. Therefore if (a) = (b), then a = bn
for some unit n, and since n does not contain irreducibles, the multiset must coincide. The
converse is just the reverse of this argument.
• Direct by expanding b and c.

Problem V.2.5. Let R be the subring of Z[t] consisting of polynomials with no term of degree 1.

• Prove that R is indeed a subring of Z[t], and conclude that R is an integral domain.
• List all common divisor of t5 and t6 in R.
• Prove that t5 and t6 have no gcd in R.

Proof.

• Clearly R is a subring since the difference of two polynomials in R has no term of degree 1. It
is a domain since you still can’t have two nonzero polynomials that has product 0.
• If (t5 , t6 ) ⊆ (p), then p can be 1, t, t2 , t3 , t4 or t5 .
• gcd did not exist since for any k ∈ {0, 1, 2, 3, 4, 5}, tk−1 | t5 , tk−1 | t6 , but tk−1 - tk since R does
not contain t.

Problem V.2.7. Let R be a Notherian domain, and assume that for all nonzero a, b in R, the
greatest common divisors of a and b are linear combinations of a and b. Prove that R is a PID.

Proof. We have the clear inclusion

(a, b) ⊆ (gcd(a, b))
and since gcd(a, b) is the linear combination of a and b, we also have (a, b) 3 gcd(a, b), hence
(a, b) = (gcd(a, b)). It follows that by a simple induction, all finitely generated ideals (p1 , . . . , pn ) is
equal to the principle ideal

(gcd(p1 , gcd(p2 , gcd(· · · , gcd(pn−1 , pn )))))

Since R is Notherian, all ideals are finitely generated, so R is a PID. 

Problem V.2.9. The height of a prime ideal P in a ring R is (if finite) the maximum length h of
a chain of prime ideals P0 ( P1 ( · · · ( Ph = P in R. Prove that if R is a UFD, then every prime
ideal of height 1 in R is principle.
CHAPTER V. IRREDUCIBILITY AND FACTORIZATION IN INTEGRAL DOMAIN 55

Proof. Let P be a prime ideal that is of height 1. Since P is prime and R is a UFD, there is some
irreducible element p ∈ P . Since irreducible implies prime, the ideal (p) is prime. Then we would
have a chain of prime ideals
0 ⊆ (p) ⊆ P
but since this chain must be of height 1, we must have (p) = P , showing that P is principle. 
Problem V.2.10. Assuming that every nonzero, nonunit element in a Notherian domain is contained
in a prime ideal of height 1. Prove a converse of Exercise 2.9, and conclude that a Notherian domain
R is a UFD if and only if every prime ideal of height 1 in R is principle.
Proof. Let R be a Notherian domain, and assuming that every prime ideal of height 1 is principle
in R. By Theorem V.2.5, we need to show that
• the a.c.c for principle ideals holds in R: this is clear since R is Notherian;
• every irreducible element of R is prime: let x be irreducible, and by assumption, it is contained
in a prime ideal of height 1 (hence principle), say x ∈ P = (p). This says that x = pa for some
a ∈ R, and since x is irreducible, either p is a unit (then P = R, which can’t be), or a is a unit.
Then we can write p = xa−1 , and since p is prime, x must be prime (a unit can’t be prime).
Hence R is a UFD. 
Problem V.2.12. Prove that if R[x] is a PID, then R is a field.
Proof. Recall that (x − c) is a prime ideal of R[x] (cf. Example III.4.7), and PID implies prime ⇒
maximal (Proposition III.4.13). Therefore the quotient
R[x] ∼
=R
(x − c)
is a field, by definition. 
Problem V.2.15. Prove that if R is a Euclidean domain, then R admits a Euclidean valuation v̄
such that v̄(ab) ≥ v̄(b) for all nonzero a, b ∈ R.
Proof. Let v be a valuation on R. Define

v̄(a) = min{v(ab) : b ∈ R}

then it follows that v(ab) ≥ v̄(b) for all a ∈ R, so v̄(ab) ≥ v̄(b). It suffices to check that it is a
valuation with respect to R. Let a, b ∈ R be such that b - a, and choose q, r such that a = bq + r.
We claim that we must have v̄(r) < v̄(b). Suppose not, that is, v̄(r) ≥ v̄(b). Let the minimum of
v̄(b) be achieved at, say, v̄(b) = v(bc). Since we have ac = bqc + rc,

v̄(b) = v(bc) > v(rc) ≥ v̄(r)

a contradiction to our assumption. Therefore v̄(r) < v̄(b), and v̄ is indeed a Euclidean valuation. 
Problem V.2.16. Let R be a Euclidean domain with Euclidean valuation v; assume that v(ab) ≥
v(b) for all nonzero a, b ∈ R (cf. Exercise 2.15). Prove that associate elements have the same
valuation and that units have minimum valuation.
Proof. Let a = ub with u being a unit. Then v(a) = v(ub) ≥ v(b). We also have v(b) = v(u−1 a) ≥
v(a), so v(a) = v(b). If u is a unit, then v(au) ≥ v(u) for all nonzero a, and since au exhaust all
elements of R (Ru = R), u must be minimal. 
Problem V.2.17. Let R be a Euclidean domain that is not a field. Prove that there exists a
nonzero, nonunit element c ∈ R such that ∀a ∈ R, ∃q, r ∈ R with a = qc + r and either r = 0 or r is
a unit.
CHAPTER V. IRREDUCIBILITY AND FACTORIZATION IN INTEGRAL DOMAIN 56

Proof. Let c be irreducible in R that is minimal in the norm sense. Since R is a ED, there is a
valuation v such that v(ab) ≥ v(b) for all nonzero a, b ∈ R (V.2.15). Now for every a ∈ R, there
exists q, r ∈ R such that a = qc + r and v(r) < v(c). But then r must be a unit, as all nonunit
element admits a factorization p1 · · · pn p0 (up to associates), and we have

v(p1 · · · pn p0 ) ≥ v(p2 · · · pn p0 ) ≥ · · · v(pn p0 ) ≥ v(p0 ) ≥ v(c)

Therefore r must be a unit (V.2.16) or 0. 

√ √
Problem V.2.18. Prove that the subring of Q( −19), Z[(1 + −19)/2], is not a Euclidean domain
(cf. V.1.17).
√
Proof. Denote δ = (1 + i 19)/2.

• We have N (a + bδ) = (a + b/2)2 + 19b2 /4. There is no z = a + bδ such that N (z) = 2 or 3 as

19b2 /4 > 4 if b ≥ 1, and there are no integers a such that a2 = 2 or 3. On the other hand,
N (0) = 1, N (1) = 1, N (2) = 4, N (δ) = 5. Also N (a + bδ) ≥ N (δ) = 5 if b 6= 0.
• Units in this ring are ±1 by the same argument as in 4th point of V.1.17.
• If c satisfies the condition in V.2.17, then the remainder of the expression a = qc + r can
only be ±1 or 0; in particular, c divides 2 and 3 since 2 = qc + r leads to 2 = qc, 3 = qc
and 1 = qc(can’t be as c is not a unit). By multiplicative property of norm (cf. 5th point of
V.1.17.) we can conclude that c = ±2 or ±3.
• However, there is no q such that δ = qc + r with c = ±2, ±3 and r = 0, ±1: This would give
combinations
δ = ±2q(±3q), −1 + δ = ±2q(±3q), 1 + δ = ±2q(±3q)
Note that N (δ) = 5, N (−1 + δ) = 5, N (1 + δ) = 7, and all of them are not divisible by 2 or 3,
so such q does not exist.

Since we can’t find q, r such that δ = qc + r, the ’division with remainder’ does not work in R, so
we conclude that R is not a Euclidean domain. 

V.3
In the following, ZL is the abbreviation for Zorn’s Lemma.

Problem V.3.2. Prove that a totally ordered set (Z, ) is a woset if and only if every descending
chain
z1  z2  z3  · · ·
in Z stabilizes.

Proof.
(⇒) If a sequence {zi }i∈N does not stabilize, then the set {zi }i∈N will not have an least element,
which can’t be as Z is a woset.
(⇐) If Z is not a woset, then pick A ⊆ Z such that A does not have a least element. Then we can
find a descending chain
z1  z2  z3  · · ·
such that it does not stabilize (as A does not have a least element). This is a contradiction. 

Problem V.3.6. Assuming the truth of Zorn’s lemma and the conventional set-theoretic construc-
tion, prove the well-ordering theorem (Theorem V.3.3).
CHAPTER V. IRREDUCIBILITY AND FACTORIZATION IN INTEGRAL DOMAIN 57

Proof. Let Z be a nonempty set, and let L be the set collecting the set of pairs (S, ≤) where S is a
subset of Z, and ≤ is a well-ordering on S. Define

(S, ≤)  (T, ≤0 )

if and only if S ⊆ T and ≤ is the restriction of ≤0 on S, and every element of S precedes every
element of T \S with respect to ≤0 .

• This is an order relation: clearly it is reflexive and antisymmetric. It is also transitive: if

(S, ≤)  (T, ≤0 ) and (T, ≤0 )  (U, ≤00 ), then clearly S ⊆ T ⊆ U , and every element of S
preeceds every element of T \S. Since every element of T also preeceds every element of U \T ,
every element of S preeceds (T \S) ∪ (U \T ) = U \S.
• The upper bound exists for every chain (S1 , ≤)  (S2 , ≤0 )  (S3 , ≤00 )  · · · : it is the union of
all sets ∪i Si .
• Therefore by ZL, there is a maximal element in L ; call it (M, ≤). We claim that M = Z.
Suppose not, then there is some a ∈ Z\M . Now notice that we clearly have

(M, ≤)  (M ∪ {a}, ≤0 )

by define a ≤0 m for all m ∈ M . This is a contradiction, so we must have M = Z.

Therefore every set admits a well-ordering, as required. 

Problem V.3.8. Prove that every nontrivial finitely generated group has a maximal proper sub-
group. Prove that (Q, +) has no maximal proper subgroup.

Proof. Let G = hg1 , . . . , gn i be finitely generated. By ZL, if every chain of subgroups of G has an
upper bound, then a maximal element exists in G. In order to apply ZL, let G1 ⊆ G2 ⊆ G3 ⊆ · · ·
be a chain of proper subgroups of G. Consider
[
H= Gi
i

we claim that this is a subgroup: if a, b ∈ H, then a must be in some Gia , and b (and also b−1 ) is in
some Gib . Then
ab−1 ∈ Gmax{ia ,ib } ⊆ H
so H is a subgroup of G. We now claim that H is proper. Suppose not, then the generators of G is
contained in some subgroup. Suppose that gi ∈ Ki for i = 1, . . . , n. Since they are a part of chain,
if K1 ⊆ K2 , then we have g1 , g2 ∈ K2 (w/o loss of generality). Similarly, we have that some Ki , say
Kn , must contain all generators of G, which implies Kn = G, a contradiction to the assumption that
the elements on the chain must be proper. Therefore by ZL, a maximal proper subgroup exists in
G, which proved the assertion.
To prove the second statement, let M be a subgroup of Q. We show that it can’t be maximal,
that is, there exists x such that
M ( M + hxi ( Q
Suppose M is maximal, that is, for all x ∈
/ M, M + hxi = Q. Let y ∈ H, and consider x/y: it can be
expressed as a new fraction a/b. We claim that x/a ∈
/ M + hxi. Suppose not, then there is equation

x/a = m + cx

where m ∈ M, c ∈ Q. This yields x = am + acx, and by noting x/y = a/b implies ax = by, we
have x = am + byc, and since both terms are in M (y ∈ M ), we conclude x ∈ M , a contradiction.
Therefore no subgroups of Q are maximal. 
CHAPTER V. IRREDUCIBILITY AND FACTORIZATION IN INTEGRAL DOMAIN 58

Problem V.3.9. Consider the rng (= ring without 1) consisting of the abelian group (Q, +) endowed
with the trivial multiplication qr = 0 for all q, r ∈ Q. Prove that this rng has no maximal ideals.
Proof. One notes that the absorption property of ideal is meaningless under the endowed multipli-
cation, so we can reduce the problem to
the group (Q, +) has no maximal subgroup
cf. V.3.8.

Problem V.3.13. Let R be a commutative ring, and let N be its nilradical (III.3.12). Prove that
the nilradical of R equals the intersection of all prime ideals of R.
Proof. Let r ∈
/ N.
• Let
F := {I ⊆ R : rk ∈
/ I ∀k > 0}
i.e. the set of ideals that does not contain any power of r. By ordering with inclusion, every
chain of F has a upper bound: it is the union of all ideals of that chain. The union is clearly
an ideal, and it is proper since r is still not in the union of ideals. Therefore by ZL, there is a
maximal element in F .
• Call the maximal element I; we now claim that I is prime. Suppose not, that is, there are
a, b ∈ R\I such that ab ∈ I. Then the ideals
I + (a), I + (b)
both properly contains I and are proper (cf. V.1.8), and since they are outside F , I +Ra, I +Rb
contains ri , rj , respectively. But then
ri rj ∈ (I + (a))(I + (b)) ⊆ I
so we would deduce that some power of r appears in I, a contradiction. Therefore I must be
prime.
• So if r ∈
/ N , then there are some prime ideal I that does not contain r, so it can’t be in the
intersection of all prime ideals.
With III.4.18, this shows that the nilradical of R equals the intersection of all prime ideals of R. 
Problem V.3.15. Recall that a (commutative) ring R is Notherian if every ideal of R if finitely
generated. Prove the weaker condition that if every prime ideal of R is finitely generated, then R is
Notherian.
Proof. Let F be the collection of all ideals of R that are not finitely generated: we will prove that
F = ∅.
• Suppose not, so F is nonempty. If there is a chain in F , then the upper bound of this chain is
the union of all ideals, which is clearly an ideal and is not finitely generated. So by ZL, there is
an maximal element in F ; call it I. It is important to note that by hypothesis, I is not prime.
• R/I is indeed Notherian: if it weren’t, then there are some ideals J ⊆ R/I that is not finitely
generated. But then JI would be a non-finitely generated ideal bigger than I, which can’t be
as I is maximal amongst all non-finitely generated ideals.
• Now since I is not prime, there are a, b ∈ R\I such that ab ∈ I. Then we let
J1 = I + (a), J2 = I + (b)
then J1 and J2 contains I and is not I, and their product is in I. (If you notice, this is the
third time we use this trick; cf. V.1.8.)
CHAPTER V. IRREDUCIBILITY AND FACTORIZATION IN INTEGRAL DOMAIN 59

• We can define a scalar multiplication

(x + I)(y + J1 J2 ) = xy + J1 J2

By absorption property, xy ∈ J1 , so this gives a R/I module structure on I/J1 J2 . Simliarly,

we can give a R/I module structure to J1 /J1 J2 . Notice that I/J1 J2 is a submodule of J1 /J1 J2 .
• Now since I ( J1 , J1 is finitely generated in R, so the submodule J1 /J1 J2 is finitely generated
as a R-module, hence as a finitely generated R/I-module. Since R/I is Notherian, by Corollary
III.6.8 J1 /J1 J2 is Notherian, so its submodule, I/J1 J2 , is finitely generated as an R/I-module.
• Since J1 J2 is finitely generated as it is the product of two finitely generated ideals (note that this
is not true if R is not commutative!) and I/J1 J2 is finitely generated, I is finitely generated.
This is a contradiction.

Therefore F is empty, which implies that R is Notherian. 

Chapter VI

Linear Algebra

60
 This is the end of the solution manual as of December 29, 2021.
Please revisit https://github.com/macyayaya/algebra-chapter-0-solutions/releases
for possible new releases.
Thanks for your reading.