Fields Rings and Modules Lecture Notes

LECTURE NOTES
MA2314: FIELDS, RINGS AND MODULES (2017)
SERGEY MOZGOVOY
Contents
1. Rings 2
1.1. Basic definitions 2
1.2. Ideals and quotient rings 4
1.3. Ring homomorphisms 7
1.4. Algebras 9
2. Integral domains 12
2.2. UFD 13
2.3. PID 15
2.4. GCD and LCM 16
2.5. Euclidean domains 17
2.6. Field of fractions 19
2.7. Factorization in polynomial rings 20
2.8. Cyclotomic polynomials 22
3. Fields 24
3.2. Field extensions 25
3.3. Splitting fields, Finite fields, Algebraically closed fields 27
3.4. Constructions with compass and straightedge 29
4. Symmetric polynomials 32
4.1. Discriminant 34
5. Modules 35
5.1. Definition and examples 35
5.2. Homomorphisms and submodules 36
5.3. Simple and indecomposable modules 38
5.4. Chinese remainder theorem 40
5.5. Modules over PID 41
5.6. Noetherian modules 43
Date: April 4, 2017.

1
2 SERGEY MOZGOVOY
1. Rings
1.1. Basic definitions. Consider the set Z of integer numbers. It has two binary operations +
(addition) and · (multiplication) compatible with each other:
a(b + c) = ab + ac, (b + c)a = ba + ca.
We will use this example as a motivation for a formal description of the above operations and
their properties.
Definition 1.1. An abelian group is a pair (A, +), where A is a set and + : A × A → A is a map
(written (a, b) 7→ a + b) such that
(1) (Associativity) (a + b) + c = a + (b + c).
(2) (Commutativity) a + b = b + a.
(3) (Existence of zero) There exists an element 0 ∈ A such that 0 + a = a + 0 = a ∀a ∈ A.
(4) (Existence of negative) For any a ∈ A there exists an element b ∈ A such that a + b = 0.
It is denoted by −a.
Example 1.2. The set of natural numbers
N = {0, 1, 2, 3, . . . }
has an obvious addition operation. But it is not a group: it does not contain negatives of nonzero
elements. For example −1 6∈ N. The following are examples of abelian groups
(1) The set Z of integer numbers.
(2) The set Q of rational numbers.
(3) The set R of real numbers.
(4) The set C of complex numbers.
♦
Definition 1.3. A ring is a triple (R, +, ·), where R is a set and + : R × R → R, · : R × R → R
are binary operations such that
(1) (R, +) is an abelian group.
(2) (Associativity of multiplication) (a · b) · c = a · (b · c).
(3) (Existence of unity) ∃1 ∈ R such that 1a = a1 = a for all a ∈ R.
(4) (Distributivity) a(b + c) = ab + ac, (b + c)a = ba + ca for all a, b, c ∈ R.
Definition 1.4. Let R be a ring. Then
(1) R is called a commutative ring if for any a, b ∈ R: ab = ba.
(2) R is called a division ring if for any nonzero a ∈ R there exists b ∈ R such that ab = ba = 1
(it is denoted by a−1 and is called the inverse of a).
(3) R is called a field if it is a commutative ring and a division ring.
Example 1.5. We met already quite a few examples of rings.
(1) The sets
Z, Q, R, C
are rings with respect to the natural operations of addition and multiplication. All of
them are commutative. The rings Q, R, C are also fields as all nonzero elements in them
are invertible. The ring Z is not a field. For example, the element 2 ∈ Z does not have
inverse in Z as 1/2 6∈ Z.
(2) The sets of polynomials Z[x], Q[x], R[x], C[x] are commutative rings. They are not fields.
(3) The set Mn (R) of n×n matrices with real coefficients is a ring. Addition and multiplication
of matrices A = (aij ), B = (bij ) is given by
A + B = (cij ), cij = aij + bij ,
n
X
AB = (dij ), dij = aik bkj .
k=1
FIELDS, RINGS AND MODULES 3
The zero element of this ring is the zero matrix. The unity element of this ring is the
identity matrix  
1 0 ··· 0
0 1 · · · 0
In =  
. . . . . . . . . . . . .
0 ...... 1
Similarly, the set Mn (C) of n × n matrices with complex coefficients is a ring. For n ≥ 2
they are not commutative. For example
( 00 10 ) ( 10 00 ) = ( 00 00 ) , ( 10 00 ) ( 00 10 ) = ( 00 10 ) .
Therefore ( 00 10 ) ( 10 00 ) 6= ( 10 00 ) ( 00 10 ). For n ≥ 2 they are also not division rings. For
example, the matrix ( 10 00 ) is not invertible.
♦
Lemma 1.6. If R is a ring then
(1) The zero element is unique.
(2) The negative of any element is unique.
(3) The unity is unique.
Proof. If 00 is another zero then 0 + 00 = 0 and 0 + 00 = 00 . Therefore 0 = 00 . Assume that an
element a has two negatives b, b0 . Then
b = b + (a + b0 ) = (b + a) + b0 = 0 + b0 = b0 .
If 10 is another unity then 1 · 10 = 1 = 10 .
Lemma 1.7. Let R be a ring. Then
(1) 0a = a0 = 0.
(2) (−a)b = a(−b) = −ab.
Proof.
0a + a = 0a + 1a = (0 + 1)a = 1a = a.
Therefore 0a = 0. Similarly a0 = 0.
(−a)b + ab = (−a + a)b = 0b = 0.
Therefore (−a)b = −ab. Similarly a(−b) = −ab.
4 SERGEY MOZGOVOY
1.2. Ideals and quotient rings. Let R be a ring.

Definition 1.8. A subset I ⊂ R is called an ideal of R if
(1) I is a subgroup of (R, +), that is
(a) 0 ∈ I.
(b) a, b ∈ I =⇒ a + b ∈ I.
(c) a ∈ I =⇒ −a ∈ I.
(2) a ∈ I, r ∈ R =⇒ ra ∈ I, ar ∈ I
Remark 1.9. For any subsets A, B ⊂ R define
A + B = { a + b | a ∈ A, b ∈ B} , AB = { ab | a ∈ A, b ∈ B} .
Then the last condition can be written as RI ⊂ I, IR ⊂ I. ♦
Remark 1.10. Note that the subsets {0} ⊂ R and R ⊂ R are ideals. An ideal I ⊂ R is called
proper if it is a proper subset of R, that is, I 6= R. ♦
Example 1.11. For any n ∈ Z the set I = Zn is an ideal in the ring Z:
(1) 0n = 0 ∈ Zn.
(2) If kn ∈ Zn, ln ∈ Zn then kn + ln = (k + l)n ∈ Zn.
(3) If kn ∈ Zn then −kn = (−k)n ∈ Zn.
(4) If kn ∈ Z and r ∈ Z then r · kn = (rk)n ∈ Zn.
♦
Lemma 1.12. All ideals of Z are of the form Zn for some n ∈ Z.
Proof. Let I ⊂ Z be an ideal. If I = {0} then I = Z0. Assume that I is nonzero. Let n be the
minimal positive element of I. We will prove that I = Zn. Inclusion Zn ⊂ I is clear. Assume
that m ∈ I\Zn. Dividing m by n with remainder we can write m = qn + r for integers q, r with
0 ≤ r < n. Actually 0 < r < n as m ∈/ Zn. As m, n ∈ I also
r = m − qn = m − n − · · · − n ∈ I.
This contradicts to the minimality of n.
1.2.1. Ideal generated by a set.
T
Lemma 1.13. Let R be a ring and let (It )t∈T be a collection of ideals in R. Then t∈T It is an
ideal in R.
T
Proof. Let I = t∈T It . Then
(1) 0 ∈ I as 0 ∈ It ∀t ∈ T .
(2) a, b ∈ I =⇒ a, b ∈ It ∀t ∈ T =⇒ a + b ∈ It ∀t ∈ T =⇒ a + b ∈ I.
(3) a ∈ I =⇒ a ∈ It ∀t ∈ T =⇒ −a ∈ It ∀t ∈ T =⇒ −a ∈ I.
(4) a ∈ I, r ∈ R =⇒ a ∈ It ∀t ∈ T =⇒ ra, ar ∈ It ∀t ∈ T =⇒ ra, ar ∈ I.

Definition 1.14. Let F ⊂ R be a subset. Denote by (F ) the smallest ideal of R that contains
F , that is, the intersection of all ideals that contain F . It is called an ideal generated by F . If
F = {f1 , . . . , fn }, then we denote (F ) also by (f1 , . . . , fn ).
Remark 1.15. An ideal (F ) can be described as a set of all finite sums
(F ) = { a1 f1 b1 + · · · + ak fk bk | k ≥ 0, fi ∈ F, ai , bi ∈ R} .
If R is commutative then
(F ) = { a1 f1 + · · · + ak fk | k ≥ 0, fi ∈ F, ai ∈ R} .
♦
Example 1.16. An ideal generated by n ∈ Z is (n) = Zn = nZ. ♦
Remark 1.17. Given a commutative ring R and two elements a, b ∈ R, we say that a divides b
(or b is a multiple of a) if there exists c ∈ R such that b = ac. We write a | b in this case. Note
that a | b if and only if b ∈ (a). ♦
1.2.2. Quotient rings. Let R be a ring and I ⊂ R be an ideal. We will construct a quotient ring
R/I as follows:
Define a binary relation ∼ on R (this is a subset of R × R) by the rule
a ∼ b ⇐⇒ a − b ∈ I
(we say that a, b are congruent modulo I and write also a ≡ b mod I). This is an equivalence
relation:
(1) Reflexivity: a ∼ a, because a − a = 0 ∈ I.
(2) Symmetry: if a ∼ b then b ∼ a, because if a − b ∈ I then b − a = −(a − b) ∈ I.
(3) Transitivity: a ∼ b, b ∼ c =⇒ a ∼ c, because if a − b ∈ I, b − c ∈ I then (a − b) + (b − c) =
a − c ∈ I.
The equivalence class [a] of an element a ∈ R is given by
[a] = a + I = { a + b | b ∈ I}
and is also called a congruence class of a modulo I. The set of all equivalence classes is denoted
by R/I.
Theorem 1.18. The set R/I with an addition and multiplication
(a + I) + (b + I) = (a + b) + I, (a + I) · (b + I) = ab + I
is a ring, called a quotient ring. Its zero element is 0 + I and its unity element is 1 + I.
Proof. First of all we have to show that addition and multiplication are well defined. This means
that we have to show that if a ∼ a0 and b ∼ b0 then
(a + I) + (b + I) = (a0 + I) + (b0 + I), (a + I) · (b + I) = (a0 + I) · (b0 + I).
To show the first equality we have to show
(a + b) + I = (a0 + b0 ) + I
that is, (a + b) − (a0 + b0 ) ∈ I. But
(a + b) − (a0 + b0 ) = (a − a0 ) + (b − b0 ) ∈ I.
To show the second equality we have to show
ab + I = a0 b0 + I
that is, ab − a0 b0 ∈ I. But
ab − a0 b0 = a(b − b0 ) + (a − a0 )b0 ∈ I
0 0
as b − b ∈ I and a − a ∈ I.
Let us prove now that R/I is a ring. We check first that (R/I, +) is an abelian group:
(1) (a + I + b + I) + c + I = (a + b + c) + I = a + I + (b + I + c + I).
(2) a + I + b + I = (a + b) + I = (b + a) + I = b + I + a + I.
(3) The element 0 + I = I ∈ R/I is zero: a + I + 0 + I = (a + 0) + I = a + I.
(4) For any (a+I) ∈ R/I there exists negative (−a+I): (a+I)+(−a+I) = (a−a)+I = 0+I.
Let us check the remaining axioms:
(1) ((a + I) · (b + I)) · (c + I) = abc + I = (a + I) · ((b + I) · (c + I)).
(2) The element 1 + I ∈ R/I is the unity element: (a + I)(1 + I) = a + I = (1 + I)(a + I).
(3) (Distibutivity)
(a + I)(b + I + c + I) = (a + I)((b + c) + I) = (ab + ac) + I
= (ab + I) + (ac + I) = (a + I)(b + I) + (a + I)(c + I).
Similarly one can prove the second distributivity property.

6 SERGEY MOZGOVOY
Example 1.19. Consider the ring Z with an ideal nZ. Then we can construct the quotient ring
Z/nZ, called the ring of congruence classes of integers modulo n. It consists of n elements which
are congruence classes of 0, 1, . . . , n − 1. For any m ∈ Z, let [m] = m + nZ be the congruence class
of m.
In the ring Z/2Z we have [1] + [1] = [0] and [1] · [1] = [1]. In Z/3Z we have [2] · [2] = [4] = [1]
(as 4 ≡ 1 mod 3). This means that [2] is invertible in Z/3Z. On the other hand, in Z/4Z we have
[2] · [2] = [4] = [0] (as 4 ≡ 0 mod 4). This means that [2] is not invertible in Z/4Z. The difference
between these two rings stems from the fact that 3 is prime and 4 is not. The general picture is
described in the following theorem. ♦
Theorem 1.20. The ring Z/nZ is a field if and only if n is a prime number.
Proof. Necessary: assume that n is not prime. Then n = km for some 1 < k, m < n. In Z/nZ we
have k, m 6= 0, but km = n = 0. This means that k, m are zero divisors and Z/nZ is not a field.
Sufficient: assume that n = p is a prime. Let 1 ≤ k < p be a number that represents some
nonzero element in R = Z/pZ. Then multiplication k : R → R (given by [m] 7→ [k] [m]) is injective:
if not, then [k] [m] = 0 for some [m] 6= 0. But this would imply that p | km and therefore p | k or
p | m (see the next result), a contradiction. As R is finite, the map k : R → R should be actually
bijective. This implies that k is invertible.
Lemma 1.21. Let p be a prime number and a, b be integers. If p | ab then p | a or p | b.
Proof. Let I = { n ∈ Z | p | na}. Then p, b ∈ I. The set I is an ideal in Z. Therefore I = dZ for
some d ≥ 1. As p ∈ I = dZ, we conclude that d | p and therefore d = 1 or d = p. If d = 1 then
p | d · a = a and we are done. If d = p then from b ∈ I = pZ we conclude that p | b.
1.3. Ring homomorphisms.

Definition 1.22. Let R be a ring. A subset S ⊂ R is called a subring if S is itself a ring when
addition and multiplication is restricted from R to S and if 1R (the unity of R) is contained in S.
Remark 1.23. Given a ring R and a subset S ⊂ R, to verify that S is a subring of R we have to
check the following axioms:
(1) a, b ∈ S =⇒ a + b ∈ S.
(2) a, b ∈ S =⇒ ab ∈ S.
(3) a ∈ S =⇒ −a ∈ S.
(4) 0, 1 ∈ S.
♦
Remark 1.24. For example, there is a chain of subrings Z ⊂ Q ⊂ R ⊂ C. But what about the
ring Z/nZ? Is it a subring of Z? The answer is not: all nonzero elements of Z/5Z are invertible,
but it would be rather difficult to find many invertible elements in Z. It turns out that it works
the other way around: there is a canonical map Z → Z/nZ that preserves the ring structures.
This leads us to the following definition. ♦
Definition 1.25. A map ϕ : R → S between two rings is called a ring homomorphism if
(1) ϕ(a + b) = ϕ(a) + ϕ(b).
(2) ϕ(ab) = ϕ(a)ϕ(b).
(3) ϕ(1R ) = 1S .
A homomorphism ϕ is called an isomorphism if it is bijective.
Example 1.26. If S ⊂ R is a subring, then the inclusion map
i : S → R, i(s) = s ∀s ∈ S
is a ring homomorphism, called a canonical embedding. ♦
Example 1.27. Let C[a, b] be the set of continuous functions f : [a, b] → R on the closed interval
[a, b]. This is a ring with addition and multiplication defined pointwise:
(f + g)(x) = f (x) + g(x), (f g)(x) = f (x)g(x), ∀x ∈ [a, b]
for any f, g ∈ C[a, b]. Let x0 ∈ [a, b]. We define a ring homomorphism ϕ : C[a, b] → R by
C[a, b] 3 f 7→ f (x0 ) ∈ R.
It is called the evaluation map at the point x0 . ♦
Lemma 1.28. Let R be a ring and I ⊂ R be an ideal. The map π : R → R/I given by
a 7→ a + I
is a ring homomorphism, called the canonical homomorphism of a quotient ring.
Proof. We have
π(a + b) = (a + b) + I = (a + I) + (b + I) = π(a) + π(b).
π(ab) = ab + I = (a + I)(b + I) = π(a)π(b).
π(1) = 1 + I.

Example 1.29. Applying this lemma to R = Z and I = nZ, we obtain a canonical homomorphism
π : Z → Z/nZ. ♦
Lemma 1.30. If ϕ : R → S is a ring homomorphism then
(1) ϕ(0) = 0.
(2) ϕ(−a) = −ϕ(a).
Proof. 1. ϕ(0) = ϕ(0 + 0) = ϕ(0) + ϕ(0). Therefore ϕ(0) = 0.
2. ϕ(−a) + ϕ(a) = ϕ(−a + a) = ϕ(0) = 0. Therefore ϕ(−a) = −ϕ(a).
8 SERGEY MOZGOVOY
Definition 1.31. Let ϕ : R → S be a ring homomorphism. Define

(1) the kernel of ϕ by
ker ϕ = { a ∈ R | ϕ(a) = 0} ⊂ R.
(2) the image of ϕ by
im ϕ = { ϕ(a) | a ∈ R} ⊂ S.
Lemma 1.32. Let ϕ : R → S be a ring homomorphism. Then
(1) ker ϕ ⊂ R is an ideal.
(2) im ϕ ⊂ S is a subring.
Proof. 1. We have
(1) ϕ(0) = 0. Therefore 0 ∈ ker ϕ.
(2) For any a, b ∈ ker ϕ: ϕ(a + b) = ϕ(a) + ϕ(b) = 0. Therefore a + b ∈ ker ϕ.
(3) For any a ∈ ker ϕ: ϕ(−a) = −ϕ(a) = 0. Therefore −a ∈ ker ϕ.
(4) For any a ∈ ker ϕ, r ∈ R: ϕ(ra) = ϕ(r)ϕ(a) = ϕ(r)0 = 0 and ϕ(ar) = ϕ(a)ϕ(r) = 0.
Therefore ra, ar ∈ ker ϕ.
This proves that ker ϕ is an ideal in R.
2. We have 1S ∈ im ϕ as ϕ(1R ) = 1S . Given two elements ϕ(a), ϕ(b) ∈ im ϕ, we have
(1) ϕ(a) + ϕ(b) = ϕ(a + b) ∈ im ϕ.
(2) −ϕ(a) = ϕ(−a) ∈ im ϕ.
(3) ϕ(a)ϕ(b) = ϕ(ab) ∈ im ϕ.
This proves that im ϕ is a subring of S.
Lemma 1.33. A ring homomorphism ϕ : R → S is injective if and only if ker ϕ = 0 (we denote
the zero ideal {0} by 0).
Proof. Assume that ϕ is injective. If ϕ(a) = 0 then ϕ(a) = ϕ(0) =⇒ a = 0. Therefore ker ϕ = 0.
Assume that ker ϕ = 0. If ϕ(a) = ϕ(b), then ϕ(a − b) = 0 =⇒ a − b ∈ ker ϕ =⇒ a − b = 0
=⇒ a = b. Therefore ϕ is injective.
Theorem 1.34 (Homomorphism Theorem). For any ring homomorphism ϕ : R → S, there is a
unique homomorphism ϕ : R/ ker ϕ → S that makes the following diagram commute (ϕ = ϕ ◦ π)
π
R R/ ker ϕ
ϕ
ϕ
S
It induces an isomorphism ϕ : R/ ker ϕ → im ϕ.
Proof. Let I = ker ϕ.
Uniqueness. From the requirement ϕ = ϕπ we obtain ϕ(a) = ϕπ(a) = ϕ(a + I) ∀a ∈ R. This
means that for any equivalence class a+I we require ϕ(a+I) = ϕ(a) and ϕ is uniquely determined.
Existence. For any equivalence class a + I, we define ϕ(a + I) = ϕ(a). This map is well-defined:
if a ∼ b then a − b ∈ I = ker ϕ =⇒ ϕ(a − b) = 0 =⇒ ϕ(a) = ϕ(b). This map is a ring
homomorphism:
(1) ϕ(a + I + b + I) = ϕ(a + b) = ϕ(a) + ϕ(b) = ϕ(a + I) + ϕ(b + I).
(2) ϕ((a + I)(b + I)) = ϕ(ab + I) = ϕ(ab) = ϕ(a)ϕ(b) = ϕ(a + I)ϕ(b + I).
(3) ϕ(1R + I) = ϕ(1R ) = 1S .
For any a ∈ R we have ϕ(a) = ϕ(a + I) = ϕπ(a). Therefore ϕ = ϕπ and the diagram commutes.
Consider the map ϕ : R/I → im ϕ. It is surjective as for any ϕ(a) ∈ im ϕ we have ϕ(a + I) =
ϕ(a). It is also injective: if ϕ(a + I) = 0, then ϕ(a) = ϕ(a + I) = 0 =⇒ a ∈ I, therefore
a + I = I = 0 + I. This means that ϕ : R/I → im ϕ is bijective and therefore an isomorphism.
1.4. Algebras.
Definition 1.35. Let R be a commutative ring. A ring S is called an algebra over R if R is a
subring of S and for any r ∈ R, s ∈ S: rs = sr.
Remark 1.36. For any ring S, define its center by
Z(S) = { a ∈ S | ab = ba ∀b ∈ S} .
If S is an algebra over R, then R ⊂ Z(S). ♦
Remark 1.37. If R is a field and S is an algebra over R then S is a vector space over R. PnAssume
that (e1 , . . . , en ) is a basis of S over R. Then all elements of S are of the form x = i=1 xi ei ,
where xi ∈ R. To define the P multiplication
P on S, it is enough to describe the products ei ej ∈ S
for all i, j. Indeed, if x = xi ei , y = yi ei with xi , yi ∈ R, then
X X X
xy = xi ei yj ej = xi yj · ei ej .
i,j
♦
1.4.1. The algebra of matrices. Let R be a commutative ring. Let Mn (R) be the set of n × n
matrices with coefficients in R. It is a ring with respect to the usual addition and multiplication:
given matrices A = (aij ), B = (bij ) in Mn (R), we define
n
X
A + B = (cij ), cij = aij + bij , AB = (dij ), dij = aik bkj .
k=1
The ring Mn (R) is an algebra over R. Indeed, R can be embedded into Mn (R) by the rule
r 7→ rIn , r ∈ R,
where In is an identity matrix in Mn (R). Then R is a subring of Mn (R) and its elements commute
with all matrices
(rIn )A = A(rIn ) = rA.
The algebra Mn (R) over R is called the matrix algebra (or the matrix ring).
1.4.2. The algebra of quaternions. The algebra of quaternions H is an algebra over R with a basis
1, i, j, k. The multiplication law is given on the basis by requiring that 1 is the identity and
i2 = j 2 = k 2 = −1, ij = k, jk = i, ki = j, ji = −k, kj = −i, ik = −j.
Remark 1.38. This algebra was invented by Hamilton on October 16, 1843 while walking near
the Broome Bridge, Dublin. This event is commemorated by a stone plaque near the bridge. For
a long time quaternions were a mandatory exam topic in Dublin. ♦
Actually it is enough to require just
i2 = j 2 = k 2 = ijk = −1.
Indeed, i, j, k are invertible and therefore ijk = k 2 implies ij = k =⇒ kj = ij 2 = −i and so on.
Given an element x = a + bi + cj + dk ∈ H, we define the absolute value and the conjugate of x by
p
|x| = a2 + b2 + c3 + d2 , x = a − bi − cj − dk.
Then
2
xx = (a + bi + cj + dk)(a − bi − ci − dk) = a2 + b2 + c2 + d2 = |x|
2
and similarly xx = |x| . This implies that if x 6= 0 then
x x
x· 2 = 2 ·x=1
|x| |x|
and the element |x|x 2 is inverse to x. This means that all nonzero elements of H are invertible, that
is, H is a division ring. It is non-commutative (for example ij = k and ji = −k). The ring H is
10 SERGEY MOZGOVOY
an algebra over R if we embed R ⊂ H by the rule a 7→ a1 (note that the elements of R commute
with the elements of H). We can also embed C ⊂ H by
a + bi 7→ a1 + bi ∈ H.
This makes C a subring of H. But H is not an algebra over C: ij 6= ji, that is, the element i ∈ C
does not commute with elements of H.
1.4.3. The algebra of polynomials. Let R be a commutative ring. Define the algebra R[x] of
polynomials in one variable x with coefficients in R to be the set of sequences
f = (f0 , f1 , f2 , . . . ), fi ∈ R ∀i ≥ 0
such that all but a finite number of elements fi are zero. We will write elements f ∈ R[x] in a
more customary form
X
f = f0 + f1 x + f2 x2 + · · · = fk xk .
k≥0
Given two polynomials f, g ∈ R[x], we define their sum f + g ∈ R[x] by
X
f +g = (fk + gk )xk
k≥0
and define their product f g ∈ R[x] by

k
!
X X
fg = fi gk−i xk .
k≥0 i=0
These operations define a structure of a ring on R[x].

Remark 1.39. Note that
(1) The zero element of R[x] is a polynomial 0 = 0 + 0x + 0x2 + . . . .
(2) The unity of R[x] is a polynomial 1 = 1 + 0x + 0x2 + . . . .
(3) The ring R[x] is an algebra over R if we embed R ⊂ R[x] by the rule r 7→ r +0x+0x2 +. . . .
♦
Definition 1.40. Let f = f0 + f1 x + · · · ∈ R[x] be a polynomial. Then
(1) The element f0 is called the constant term of f .
(2) The number max { k ≥ 0 | fk 6= 0} is called the degree of f and is denoted by deg f . If
f = 0 then we define deg f = −∞.
(3) If n = deg f , then the element fn is called the leading coefficient of f . If fn = 1 then f is
called a monic polynomial.
Remark 1.41. Let S be an algebra over R and let f = i≥0 fi xi ∈ R[x] be a polynomial. We
P
define the evaluation of f at s ∈ S (or the substitution of s into f ) to be

X
f (s) = fi si ∈ S.
i≥0
♦
Theorem 1.42 (Evaluation of polynomials). Let S be an algebra over a commutative ring R.
Given an element s ∈ S, there exists a unique ring homomorphism ϕs : R[x] → S such that
ϕ(a) = a ∀a ∈ R, ϕs (x) = s.
For any polynomial f ∈ R[x], we have ϕs (f ) = f (s).
Proof. Let us prove the uniqueness. For any f ∈ R[x], we have
 
X X X X
ϕs (f ) = ϕs  fk xk  = ϕs (fk xk ) = ϕs (fk )ϕs (x)k = fk sk
k≥0 k≥0 k≥0 k≥0
and this proves that ϕs is uniquely determined. Conversely, if we define ϕs using this rule then
ϕs (a) = a ∀a ∈ R and ϕs (x) = s. Let us show that ϕs is a ring homomorphism. It is clear that it
preserves the additive structures. Concerning the product, we have
 !  !
X X k X X k
k
ϕs (f g) = ϕs  fi gk−i x  = fi gk−i sk
k≥0 i=0 k≥0 i=0
  
X X
= fi si   gj sj  = ϕs (f )ϕs (g).
i≥0 j≥0

Remark 1.43. This proposition implies, that for any polynomial f ∈ R[x] and for any element
r ∈ R, we can evaluate f (r) = ϕr (f ) ∈ R. Moreover, for any matrix A ∈ Mn (R), we can evaluate
f (A) = ϕA (f ) ∈ Mn (R) (recall that Mn (R) is an algebra over R). ♦
Definition 1.44. Let S be an algebra over R and f ∈ R[x]. An element s ∈ S is called a root
(or a zero) of a polynomial f if f (s) = 0.
Remark 1.45. Define an algebra of polynomials in several variables inductively by the rule
R[x1 , . . . , xn ] = (R[x1 , . . . , xn−1 ])[xn ].
Such polynomials can be written in the form
X
f= fi1 ...in xi11 . . . xinn ,
i1 ,...,in ≥0
where fi1 ...in ∈ R and all but a finite number of these elements are zero. ♦
Example 1.46. Consider a matrix A = ( 10 11 ) ∈ M2 (R). Then A − I = ( 00 11 ) and (A − I)2 = 0.
This means that A is a root of a polynomial p = (x − 1)2 ∈ R[x]. Consider a ring homomorphism
ϕA : R[x] → M2 (R), f 7→ f (A).
We proved that p(A) = 0, hence p ∈ ker ϕA . As ker ϕA is an ideal, the entire ideal (p) = R[x]p
generated by p = (x − 1)2 is contained in ker ϕA . One can show that actually ker ϕA = (p). ♦
12 SERGEY MOZGOVOY
2. Integral domains
2.1. Basic definitions.
Remark 2.1. There is the following chain of commutative ring classes
Fields ⊂ Euclidean domains ⊂ Principal ideal domains
⊂ Unique factorization domains ⊂ Integral domains ⊂ Commutative rings
We introduced already commutative rings and fields. Our goal will be to fill the gap. From now
on all rings are assumed to be commutative. ♦
Definition 2.2. Let R be a commutative ring.
(1) R is called an integral domain if ab = 0 implies a = 0 or b = 0 for arbitrary a, b ∈ R.
(2) An element a ∈ R is called a zero divisor if there exists nonzero b ∈ R such that ab = 0.
An element that is not a zero divisor is called a non-zero-divisor.
(3) An element a ∈ R is called an invertible element (or a unit) if there exists b ∈ R such that
ab = 1. The set of all units of R is denoted by R× .
(4) Two elements a, b ∈ R are called associates (we write a ∼ b) if a = ub for some u ∈ R× .
Remark 2.3. A commutative ring R is an integral domain ⇐⇒ nonzero elements of R are
non-zero-divisors. ♦
Example 2.4. The ring Z is an integral domain. The only invertible elements of Z are ±1. ♦
Example 2.5. Any field is an integral domain. For any field k the ring k[x] is an integral domain.
♦
Remark 2.6 (Cancellation). Assume that R is an integral domain and a ∈ R is nonzero. If
ab = ac for some b, c ∈ R then b = c. Indeed, a(b − c) = 0 and as a is not a zero divisor, we
conclude that b − c = 0. Therefore b = c. ♦
Lemma 2.7. Let R be an integral domain. Then two elements a, b ∈ R are associates if and only
if (a) = (b) (if and only if a | b and b | a).
Proof. If b = ua with u ∈ R× , then b ∈ (a) and (b) ⊂ (a). Similarly, from a = u−1 b we conclude
(a) ⊂ (b) and therefore (a) = (b).
Conversely, if (a) = (b) then b = ua, a = vb for some u, v ∈ R. Therefore a = vb = uva =⇒
a(1 − uv) = 0 =⇒ a = 0 or uv = 1. If a = 0, then b = ua = 0 = a. If uv = 1 then u is invertible
and a, b are associates.
2.2. UFD.
Definition 2.8. Let R be an integral domain.
/ R× and it can be written as a product
(1) An element p ∈ R is called irreducible if p 6= 0, p ∈
only in a trivial way, that is, if p = ab then a ∈ R× or b ∈ R× .
(2) An element p ∈ R is called prime if p 6= 0, p ∈ / R× and if p | ab implies p | a or p | b for all
a, b ∈ R.
Example 2.9. A positive n ∈ Z is irreducible if and only if it is prime (in the usual sense). ♦
Remark 2.10. Any prime element is irreducible. Indeed, if p = ab then p | ab, hence p | a or
p | b. Without loss of generality p | a, hence a = cp for some c ∈ R. This implies p = ba = bcp and
bc = 1, hence b ∈ R× . We will see later that there are rings with irreducible elements that are not
prime. ♦
Definition 2.11. An integral domain R is called a unique factorization domain if
(1) Any a ∈ R\{0} can be written as a product of a unit and irreducible elements, that is,
a = up1 , . . . , pk
×
where u ∈ R and pi are irreducible.
(2) This representation is unique up to units and a permutation of factors. That is, if
a = vq1 , . . . , ql
is another decomposition into a product of a unit and irreducibles, then l = k and there
exist a permutation σ ∈ Sk and elements ui ∈ R× such that qi = ui pσi for all i.
An integral domain satisfying just the first axiom is called a factorization domain.
Example 2.12.
(1) The ring Z is a UFD.
(2) Any field is a UFD for trivial reasons (all nonzero elements are invertible).
(3) If K is a field, then K[x] is a UFD as we will see later.
♦
Example 2.13. Let us consider a ring which is not a unique factorization domain. Let
√ n √ o
R = Z[ −5] = a + b 5i a, b ∈ Z ⊂ C.
We claim that the element 6 has two different factorizations into irreducible factors
√ √
6 = 2 · 3 = (1 + 5i)(1 − 5i).
√
For any element z = a + b 5i ∈ R we have
2
|z| = a2 + 5b2 ∈ Z
2 2 2 2
and if z = z1 z2 then |z| = |z1 | |z2 | . Therefore if z ∈ R is invertible, then |z| ∈ Z is invertible.
√
This implies |z| = 1 and z = ±1. Assuming that we can write some of the elements z = 2, 3, 1± −5
2 2 2
as z = z1 z2 with z1 , z2 6∈ R× , we obtain that |z| = 4, 9, 6 is equal to |z1 | · |z2 | and therefore
2 √
|zi | are equal to 2 or 3. But there are no such elements in R. This implies that 2, 3, 1 ± −5
are irreducible. They are not associates of each other as R× = {±1}. This implies that the above
factorizations are not equivalent.
Equality
√ √
2 · 3 = 6 = (1 + −5)(1 − −5)
√ √ √
implies that 2 | (1 +√ −5)(1 − −5). But 2 does not divide 1 ± −5. This means that 2 is not
prime in the ring Z[ −5], although we have seen that 2 is irreducible. ♦
Theorem 2.14. Let R be a factorization domain (satisfies just the first axiom of a UFD). Then
R is a UFD if and only if every irreducible element of R is prime.
14 SERGEY MOZGOVOY
Proof. ⇒. Let R be a UFD, p ∈ R be irreducible and p | ab, that is, ab = pc for some a, b, c ∈ R.
We want to show that p | a or p | b. Let
Y Y Y
a= ai , b= bi , c= ci
be factorizations into irreducible elements. Then the element ab = pc has two factorizations
Y Y Y
ai bi = p ci .
From the uniqueness of factorizations we conclude that p equals (up to a unit) to one of ai or bj .
This means that p divides a or b.
⇐. Assume that an element a has two decompositions into irreducibles
m
Y Yn
a= pi = qi .
i=1 i=1
By assumption all elements pi are prime. Let p = pm . Then p should divide one of qi (we can
assume that it is qn ). But qn is irreducible, therefore qn = up for some u ∈ R× . Dividing both
Qm−1 Qn−1
sides by p we obtain i=1 pi = u i=1 qi . By induction on m, these factorizations are the same
up to the permutation of factors and multiplications by units.
2.3. PID.
Definition 2.15. Let R be a commutative ring.
(1) An ideal I ⊂ R is called principal if it can be generated by one element, that is ∃a ∈ R
such that I = (a) = Ra.
(2) R is called a principal ideal domain if it is an integral domain and any ideal of R is a
principal ideal.
Example 2.16. We have seen that any ideal in Z has a form nZ for some n ∈ Z. This implies
that Z is a principal ideal domain. ♦
Lemma 2.17. Let R be a principal ideal domain. Then any irreducible element in R is prime.
Proof. Let p ∈ R be irreducible and let p | ab. Assume that p - a. The ideal (p, a) is a principal
ideal, therefore (p, a) = (d) for some d ∈ R. In particular p ∈ (d) and p = cd for some c ∈ R. As p
is irreducible, we have c ∈ R× or d ∈ R× . If c ∈ R× , then (p) = (d) 3 a and p | a, a contradiction.
If d ∈ R× , then (a, p) = (d) = R. Therefore 1 = ax + py for some x, y ∈ R =⇒ b = abx + bpy
=⇒ p | b.
Theorem 2.18. Any PID is a UFD.
Proof. We have to show that a PID R is a factorization domain. The fact that R is a UFD will
follow then from Theorem 2.14 and the fact that every irreducible element in R is prime according
to the previous lemma.
Assume that a can not be represented as a product of irreducible elements (up to a unit). Then
/ R× .
a is not a unit and not irreducible. Therefore we can decompose it as a = a1 b1 with a1 , b1 ∈
Without loss of generality we can assume that a1 does not have a factorization into a product.
Continuing this argument, we obtain a sequence of elements satisfying ai = ai+1 bi+1 and therefore
(strict inclusions here)
(a) ⊂ (a1 ) ⊂ (a2 ) ⊂ . . .
The union of these ideals is again an ideal. Moreover, it is a principal ideal (c) as R is a PID.
We have c ∈ (an ) for some n ≥ 1. Therefore (an ) = (c). This implies that (an ) = (an+1 ), a
contradiction. This proves the existence of a decomposition.
16 SERGEY MOZGOVOY
2.4. GCD and LCM. Let R be an integral domain.

Definition 2.19. The greatest common divisor of two elements a, b ∈ R is an element d ∈ R
(denoted by gcd(a, b)) such that
(1) d | a, d | b.
(2) If c | a, c | b =⇒ c | d.
The elements a, b are called coprime if gcd(a, b) = 1.
Remark 2.20. A GCD of two elements does not exist in all rings, but we will see later that it
exists in UFD. If d is a GCD of a, b then any its associate (that is, an element of the form ud for
some u ∈ R× ) is also a GCD of a, b. Conversely, if c, d are two GCD of a, b then d | c and c | d.
This means that c, d are associates. We denote by gcd(a, b) any of these associates. ♦
√
Example 2.21. Consider the ring R = Z[ −5] and an equation
√ √
(1 + −5)(1 − −5) = 2 · 3 = 6
in R. Consider the elements √
a = (1 + −5) · 2, b = 6.
√
Then d = 1 + −5 is a common divisor of a and b and also c = 2 is a common divisor of a and b.
We also know that they are irreducible. However, neither of them is a greatest common divisor of
a, b as d 6| c and c 6| d. ♦
Definition 2.22. The least common multiple of two elements a, b ∈ R is an element m ∈ R
(denoted by lcm(a, b)) such that
(1) a | m, b | m.
(2) If a | c, b | c then m | c.
Remark 2.23. Again we can show that a LCM of two elements a, b is determined only up to a
unit. Any of them is denoted by lcm(a, b). ♦
Lemma 2.24. Let R be a principal ideal domain and a, b ∈ R. Then
(1) d = gcd(a, b) if and only if (a, b) = (d).
(2) m = lcm(a, b) if and only if (a) ∩ (b) = (m).
In particular GCD and LCM exist in principal ideal domains.
Proof. Let us prove just the first statement. Assume that (d) = (a, b). Then a, b ∈ (d) =⇒ d | a
and d | b. If c | a and c | b then a, b ∈ (c) and this implies (d) = (a, b) ⊂ (c) and therefore c | d.
This means that d = gcd(a, b). Conversely, if d0 is a GCD of a, b then d0 and d are associates and
therefore (d0 ) = (d) = (a, b).
Lemma 2.25. Let R be a principal ideal domain. Then a, b ∈ R are coprime if and only if there
exist x, y ∈ R such that ax + by = 1.
Proof. Condition that a, b are coprime means that gcd(a, b) = 1. Condition that there ∃x, y ∈ R:
ax + by = 1 means that (a, b) = R = (1). Now we apply the previous lemma.
Theorem 2.26. GCD and LCM exist in unique factorization domains.
Proof. Any two elements a, b can be written in the form
Yn n
Y
a=u pki i , b=v plii ,
i=1 i=1
×
where u, v ∈ R , p1 , . . . , pn are distinct prime elements (not associate to each other) and ki , li ≥ 0.
Define
Y n Yn
d= psi i , si = min{ki , li }, m= ptii , ti = max{ki , li }.
i=1 i=1
We claim that d = gcd(a, b) and m = lcm(a, b). Let is show this justQfor d. It is clear that d | a
n
and d | b. Assume that c | a and c | b. Up to a unit, we can write c = i=1 pri i , where ri ≤ ki and
ri ≤ li . This implies ri ≤ si and c | d.
2.5. Euclidean domains. Euclidean domains are rings where an analogue of the Euclidean di-
vision of integers (that is, division with a remainder) is possible.
Definition 2.27. An integral domain R is called a Euclidean domain if there exists a function
δ : R\{0} → N = {0, 1, 2, . . . }
(called a Euclidean function or a degree function) such that for any a, b ∈ R\{0} there exist
elements q, r ∈ R (quotient and remainder) such that a = bq + r and either r = 0 or δ(r) < δ(b).
Example 2.28. The ring Z with the degree function δ(n) = |n| is a Euclidean domain. ♦
Example 2.29. For any field k, the algebra of polynomials k[x] with the degree function δ(f ) =
deg(f ) is a Euclidean domain. This follows from the following result. ♦
Theorem 2.30. Let R be a commutative ring and let f, g ∈ R[x] with g a monic polynomial of
degree d, that is, g(x) = xd + gd−1 xd−1 + · · · + g0 . Then there exist unique polynomials q, r ∈ R[x]
such that f = gq + r and deg r < d.
Proof. Let us choose a polynomial q ∈ R[x] such that the polynomial r := f − gq has a minimal
possible degree. Assume that r = re xe + · · · + r0 has degree e ≥ d. Then
e−1
X d−1
X
f − g(q + re xe−d ) = r − re gxe−d = ri xi − re gi xe−d+i
i=0 i=0
has degree < e, a contradiction.
To prove the uniqueness, assume that f = gq 0 + r0 with deg r0 < d. Then g(q − q 0 ) = r − r0 .
If q 6= q 0 then deg(g(q − q 0 )) ≥ deg g = d and deg(r − r0 ) < d, a contradiction. Therefore q = q 0 ,
hence also r = r0 .
Corollary 2.31. An element a ∈ R is a root of a polynomial f ∈ R[x] (that is, f (a) = 0) if and
only if (x − a) divides f .
Proof. We can divide with a remainder f = (x − a)q + r, where deg r < deg(x − a) = 1. This
means that r ∈ R. If a is a root of f then r = f (a) = 0. Therefore (x − a) divides f . Conversely,
if (x − a) divides f then clearly f (a) = 0.
Remark 2.32. If R is an integral domain, then a polynomial f ∈ R[x] has at most deg f roots.
Indeed, if a ∈ R is a root of f , then we can write f = (x − a)q with deg q = deg f − 1. Any root
of f different from a is a root of q. By induction on degree, q has at most deg q = deg f − 1 roots.
Therefore f has at most deg f roots.
This statement is false if R is not an integral domain. For example, the polynomial x2 − 1 in
Z8 [x] has four roots. ♦
Example 2.33. The ring Z[i] = { x + yi | x, y ∈ Z} ⊂ C is called the ring of Gaussian integers. It
2
is a Euclidean domain with a degree function δ(x + yi) = |x + yi| = x2 + y 2 . Such degrees are
precisely those natural numbers that can be written as a sum of two squares. ♦
Lemma 2.34. The ring Z[i] of Gaussian integers is Euclidean.
Proof. Let a, b ∈ Z[i]\{0}. Consider the element ab−1 = x + yi ∈ C. There exist m, n ∈ Z such
that |x − m| ≤ 21 and |y − n| ≤ 12 . Let q = m + ni and
r = a − bq = b(ab−1 − q) = b((x − m) + (y − n)i).
We have
2 2 2 2
δ(r) = |r| = |b| · ((x − m)2 + (y − n)2 ) ≤ |b| · (1/4 + 1/4) < |b| = δ(b).

18 SERGEY MOZGOVOY
Theorem 2.35. A Euclidean domain is a principal ideal domain and therefore also a unique
factorization domain.
Proof. Let R be a Euclidean domain with a degree function δ : R\{0} → N and let I ⊂ R be a
nonzero ideal. Let b ∈ I\{0} have a minimal possible value δ(b). Then (b) ⊂ I and we claim that
I = (b). If a ∈ I\(b), then we can write a = bq + r with r = 0 or δ(r) < δ(b). If r = 0, then a = bq
and a ∈ (b), a contradiction. If r 6= 0 then δ(r) < δ(b) and r = a − bq ∈ I, contradicting to the
minimality of δ(b). This implies that I = (b).
Example 2.36. This theorem implies in particular, that any polynomial ring k[x] over a field k is
a principal ideal domain, hence a unique factorization domain. Therefore one has GCD and LCM
in this ring. For example, the polynomials x2 + 1, x + 1 over Q have gcd = 1. On the other hand,
the same polynomials over Z2 have gcd = x + 1 (note that x2 + 1 = x2 − 1 = (x − 1)(x + 1)). ♦
Remark 2.37 (Euclidean algorithm). As we have seen, any Euclidean domain R is a UFD
and therefore its elements have the greatest common divisors. There is an algorithm, called the
Euclidean algorithm, to find gcd(a, b) for any a, b ∈ R. This a straightforward generalization
of a similar algorithm for integers. Namely, we apply the following sequence of division with
remainders, until we obtain the zero remainder:
a = q1 b + r1 , δ(r1 ) < δ(b),
b = q2 r1 + r2 , δ(r2 ) < δ(r1 ),
r1 = q3 r2 + r3 , δ(r3 ) < δ(r2 ),
...................
rk−2 = qk rk−1 + rk , δ(rk ) < δ(rk−1 ),
rk−1 = qk+1 rk , δ(rk+1 ) = 0.
♦
Then
gcd(a, b) = gcd(b, r1 ) = gcd(r1 , r2 ) = · · · = gcd(rk−1 , rk ) = rk .
This algorithm can also be used in order to find x, y ∈ R such that xa + yb = gcd(a, b) = rk .
Indeed, first we can write
r1 = a − q1 b.
Then
r2 = b − q2 r1 = b − q2 (a − q1 b) = −q2 a + (1 + q1 q2 )b.
Continuing this process, we obtain an expression for rk as a linear combination of a and b.
2.6. Field of fractions.

Remark 2.38. The ring Z of integers has non-invertible elements. If we allow to invert nonzero
elements then we will end up with a field Q of rational numbers. Our goal is to formalize this
construction for arbitrary integral domains. ♦
Let R be an integral domain and let S = R\{0} (it is not a subring, but it is closed with respect
to multiplication). Define a binary relation on R × S by the rule
(a, s) ∼ (b, t) ⇐⇒ at = bs.
(We think of a pair (a, s) as a fraction a/s). This is an equivalence relation:
(1) Reflexivity: (a, s) ∼ (a, s).
(2) Symmetry: (a, s) ∼ (b, t) =⇒ (b, t) ∼ (a, s).
(3) Transitivity: (a, s) ∼ (b, t) (b, t) ∼ (c, v) =⇒ (a, s) ∼ (c, v). Indeed, we have at = bs and
bv = ct. This implies atv = bsv = cts and therefore av = cs, that is, (a, s) ∼ (c, v).
We denote the set of equivalence classes of such pairs by Q(R) and denote the equivalence class
of a pair (a, s) by as . Define addition and multiplication on Q(R) by the rules
a b at + bs a b ab
+ = , · = .
s t st s t st
One can verify that these operations are well-defined and they equip Q(R) with a ring structure.
The zero element of Q(R) is 10 . The unity of Q(R) is 11 .
Theorem 2.39. Let R be an integral domain. Then the ring Q(R) is a field, called the field of
fractions of R. The canonical map i : R → Q(R), r 7→ 1r , is an injective ring homomorphism.
If f : R → K is an injective ring homomorphism to a field K, then there exists a unique ring
homomorphism f¯ : Q(R) → K that makes the following diagram commute (f = f¯i)
i
R Q(R)
f f¯
K
Proof. Let as ∈ Q(R) be a nonzero element. Then as 6= 01 , that is, a 6= 0. This implies that
s a s 1 a
a ∈ Q(R). We have s · a = 1 and this means that s is invertible. Therefore Q(R) is a field.
The map i : R → Q(R) is obviously a ring homomorphism. For example, for a, b ∈ R
a b a·1+b·1 a+b
i(a) + i(b) = + = = = i(a + b).
1 1 1·1 1
To show that it is injective, assume that a ∈ ker i. Then a1 = 01 . Then a = 0 and therefore
ker i = 0.
Uniqueness of f¯. For any a ∈ R we have f¯( a1 ) = f¯(i(a)) = f (a). If s ∈ R\{0} then f¯( 1s )f¯( 1s ) =
f¯( 1 ) = f (1) = 1 and f¯( 1s ) = f (s). Therefore f¯( 1s ) = 1/f (s). This implies f¯( as ) = f¯( a1 )f¯( 1s ) =
1
f (a)/f (s) and uniqueness follows.

Existence of f¯. We define f¯( as ) = f (a)/f (s). One verifies easily that this map is well defined
and is a ring homomorphism. The composition (f¯i)(a) = f¯( a1 ) = f (a)/f (1) = f (a). Therefore
f = f¯i.
Example 2.40. Consider an algebra of polynomials k[x] over a field k. The field of quotients
Q(k[x]) is denoted by k(x) and is called the field of rational functions over k. Its elements are
fractions fg(x)
(x)
, where f, g are polynomials over k and g 6= 0. ♦
20 SERGEY MOZGOVOY
2.7. Factorization in polynomial rings. We know that if k is a field then k[x] is a UFD. This
is not enough to show that k[x1 , . . . , xn ] is a UFD. Our goal will be to show that if R is a UFD
then also R[x] is a UFD. This will imply that R[x1 , . . . , xn ] is a UFD.
Let R be a UFD. We know that any two elements a, b ∈ R have a GCD. Similarly we can define
the greatest common divisor of several elements a1 , . . . , an which we denote by gcd(a1 , . . . , an ).
Pn
Definition 2.41. Given a polynomial f = i=0 fi xi in R[x], we define its content to be d(f ) =
gcd(f0 , . . . , fn ). We say that f is primitive if d(f ) = 1.
Lemma 2.42 (Gauss’s lemma). Let R be a UFD and f, g ∈ R[x]. Then
(1) If f, g are primitive then f g is primitive.
(2) d(f g) = d(f )d(g) (up to a unit).
Proof. 1. Assume that f g is not primitive. Then there exists some prime p ∈ R such that all
coefficients of f g are divisible by p. For any polynomial h ∈ R[x], we will denote by h its image
in R/(p)[x]. Then f g = 0 in (R/pR)[x].
The ring R/pR is an integral domain. Indeed, if a, b ∈ R are such that [a][b] = 0 in R/pR, then
ab ∈ pR =⇒ p | ab =⇒ p | a or p | b =⇒ [a] = 0 or [b] = 0 in R/pR.
This implies that (R/pR)[x] is also an integral domain. From f · g = f g = 0 we obtain that
f = 0 or g = 0. This means that p divides all the coefficients of f or all the coefficients of g, that
is, either f or g is not primitive, a contradiction.
2. Indeed, let a = d(f ) and b = d(g). Then f = af ∗ and g = bg ∗ for primitive f ∗ , g ∗ ∈ R[x].
Therefore
d(f g) = d(abf ∗ g ∗ ) = abd(f ∗ g ∗ ) = ab = d(f )d(g),
where d(f ∗ g ∗ ) = 1 as f ∗ g ∗ is primitive.
Corollary 2.43. Let R be a UFD and f ∈ R[x] be a non-constant irreducible polynomial. Then
f is irreducible in Q(R)[x].
Proof. Assume that f = gh, where g, h ∈ Q(R)[x] have positive degrees. We can write
a c
g = g∗ , h = h∗ ,
b d
where a, b, c, d ∈ R and g ∗ , h∗ ∈ R[x] are primitive. Then f = ac ∗ ∗
bd g h and
bd · d(f ) = ac · d(g ∗ h∗ ) = ac.
ac
This implies that bd = d(f ) ∈ R and
a ∗ c ∗
f= g · h = d(f )g ∗ h∗ .
b d
This contradicts to the irreducibility of f in R[x].
Corollary 2.44. Let f, g, h ∈ Q[x] be monic and f = gh. If f ∈ Z[x], then g, h ∈ Z[x].
Proof. Let g = ab g ∗ , h = dc h∗ , where a, b, c, d ∈ Z and g ∗ , h∗ ∈ Z[x] are primitive. Then f = ac ∗ ∗
bd g h
ac
and, comparing contents, we get bd = 1 (f is monic, hence primitive). Comparing the leading
coefficients in g ∗ = ab g, we obtain ab ∈ Z and similarly dc ∈ Z. From ac bd
= 1 we conclude ab = dc = 1
∗ ∗
(up to a sign). This implies g = g ∈ Z[x] and h = h ∈ Z[x].
Theorem 2.45. If R is a UFD then R[x] is a UFD. The irreducible elements of R[x] are
(1) Irreducible elements of R.
(2) Primitive polynomials in R[x] that are irreducible in Q(R)[x].
Proof. We can embed R[x] into Q(R)[x], where Q(R) is a field of fractions of R. We know that
Q(R)[x] is a UFD. Let us show that elements described in the statement of the theorem are
irreducible in R[x]. All irreducible elements in R are also irreducible in R[x]. Also any primitive
p ∈ R[x] that is irreducible in Q(R)[x] is irreducible in R[x]: it can not be written as product
of two polynomials having positive degree, and if it is equal to cf for c ∈ R and f ∈ R[x] then
d(p) = 1 = d(c)d(f ) = cd(f ) =⇒ c ∈ R× =⇒ p is irreducible.
Let us show that every polynomial in R[x] can be factorizedQinto a product of irreducibles
in R[x]. Given f ∈ R[x], we can factorize it as a product f = qi of irreducible polynomials
in Q(R)[x]. We can write qi = ci pi , where ci ∈ Q(R) and pi ∈ R[x] is a primitive polynomial
irreducible in Q(R)[x]. Then Y
f =c pi ,
Q Q
where c = ci ∈ Q(R). The product Q pi is a primitive polynomial in R[x] by the Gauss lemma.
If c = ab then we can write bf = a pi and comparing the GCD of coefficients on both sides we
obtain b | a, that is c = ab ∈ R. Now we decompose c into a product of irreducible elements. We
have seen that polynomials pi are irreducible in R[x]. This proves the existence of a factorization
into irreducibles. This shows also that every irreducible element in R[x] is of the form described
in the statement of the theorem.
To show the uniqueness, we have to prove that any irreducible polynomial p ∈ R[x] is prime.
This is clear if p ∈ R. If deg p > 0 then p is a primitive polynomial, irreducible in Q(R)[x]. Assume
that p | f g in R[x]. As p is irreducible in Q(R)[x], it is also prime in Q(R)[x]. Therefore p | f
or p | g in Q(R)[x]. Without loss of generality p | f in Q(R)[x]. Then there exist a, b ∈ R\0 and
primitive h ∈ R[x] such that
a
f = hp.
b
Then bf = ahp and d(bf ) = bd(f ) = ad(hp) = a (as h and p are primitive). This means that
a a
b = d(f ) ∈ R, b h ∈ R[x], and p | f in R[x]. Therefore p is prime in R[x] and this finishes the
proof of the uniqueness of factorizations.
Pn n
Theorem 2.46 (Eisenstein’s criterion). Let f = i=0 fn x ∈ Z[x] be a degree n polynomial and
p be a prime number such that
(1) p | f0 , . . . , fn−1 ,
(2) p - fn ,
(3) p2 - f0 .
Then f is irreducible in Q[x].
Proof. We can assume that f is primitive (divide f by its content, if necessary). We will show that
f is irreducible in Z[x] (by the previous theorem this will imply that f is also irreducible in Q[x]).
Assuming the converse, we can write f = gh, for some non-constant g, h ∈ Z[x]. Then modulo
p, we obtain f = gh in Zp [x]. By our assumption f i = 0 for 0 ≤ i < n and therefore f = f n xn ,
where f n 6= 0 in Zp . This implies g = bxk , h = cxl , where k = deg g > 0, l = deg h > 0 and
b, c ∈ Zp are nonzero. This implies that the constant terms of g, h are divisible by p and therefore
f0 is divisible by p2 . A contradiction.
22 SERGEY MOZGOVOY
2.8. Cyclotomic polynomials.

Definition 2.47.
(1) An element ξ ∈ C is called a root of unity if ξ n = 1 for some n ≥ 1. In this case ξ = e2πik/n
for some 0 ≤ k < n.
(2) Given n ≥ 1, an element ξ ∈ C is called an n-th root of unity if ξ n = 1.
(3) For any root of unity ξ, define ord ξ = min { n ≥ 1 | ξ n = 1}.
(4) An element ξ is called a primitive n-th root of unity if ξ n = 1 and ξ k 6= 1 for 1 ≤ k < n.
Equivalently, ord ξ = n.
Remark 2.48. An element ξ = e2πik/n is an n-th primitive root of unity if and only if gcd(k, n) =
1. Indeed, if d = gcd(k, n) > 1, then ξ n/d = e2πik/d = 1 as k/d ∈ Z. This implies that ord ξ < n.
Conversely, if gcd(k, n) = 1 and ξ m = 1 for some 1 ≤ m < n, then n | km =⇒ n | m, a
contradiction. ♦
Remark 2.49. Let ξ be a d-th primitive root of unity. Then ξ n = 1 if and only if d | n. Indeed,
ξ = e2πik/d with gcd(k, d) = 1. Therefore 1 = ξ n = e2πikn/d =⇒ d | kn =⇒ d | n. Every n-th
root of unity is a primitive root for a unique d | n. ♦
Let Y
Φd (x) = (x − ξ).
ord ξ=d
By the previous discussion
Y Y Y Y
xn − 1 = (x − ξ) = (x − ξ) = Φd (x).
ξ n =1 d|n ord ξ=d d|n
We obtain by induction (and Gauss lemma) that Φd (x) ∈ Z[x]. This polynomial is called a d-th
cyclotomic polynomial.
Example 2.50. We have Φ1 (x) = x − 1. For any prime p, we have d | p if and only if d = 1 or
d = p. This implies
xp − 1 = (x − 1)Φp (x),
hence
Φp (x) = xp−1 + · · · + x + 1.
We have
x4 − 1 = Φ1 (x)Φ2 (x)Φ4 (x),
2
hence Φ4 (x) = x + 1.
x6 − 1 = Φ1 (x)Φ2 (x)Φ3 (x)Φ6 (x),
hence
(x3 − 1)(x3 + 1) x3 + 1
Φ6 (x) = = = x2 − x + 1.
(x3 − 1)Φ2 (x) x+1
♦
Theorem 2.51 (Kronecker). The polynomial Φd (x) is irreducible for every d ≥ 1.
Let us prove first a simpler version.
Lemma 2.52 (Gauss). For every prime p, the polynomial
Φp (x) = xp−1 + · · · + x + 1
is irreducible.
Proof (Eisenstein). Applying the substitution x = y + 1, we obtain
(y + 1)p − 1 (y + 1)p − 1
Φp (y + 1) = = .
y+1−1 y
The coefficient of y p−1 is one. The coefficient of y k−1 is kp for 1 ≤ k < p and is divisible by p.

The coefficient of y 0 is p1 = p is not divisible by p2 . By Eisenstein criterium our polynomial is

irreducible.
Proof of Theorem 2.51 (Dedekind). Let f (x) be in an irreducible factor of Φn (x). It has integer
coefficients by Gauss’s lemma. We will prove that if ξ is a root of f and p is prime, coprime to
n, then f (ξ p ) = 0. This implies by induction that if (m, n) = 1, then f (ξ m ) = 0. As ξ is a root
of f , it is a root of Φn and is a primitive n-th root of 1. If f (ξ m ) = 0 for all (m, n) = 1, then all
primitive n-th roots of 1 are roots of f , hence Φn = f and Φn is irreducible.
For any two polynomials f, g ∈ Z[x], we have
p
p
X p i p−i
(f + g) = f g ≡ f p + g p (mod p).
i=0
i
Pd
Applying this to the sum of monomials f = i=0 fi xi ∈ Z[x], we obtain
d
X d
X
f (x)p ≡ fip xip ≡ fi xip = f (xp ) (mod p),
i=0 i=0
where we used ap ≡ a (mod p) for a ∈ Z (Fermat’s little theorem).
Let g be an irreducible polynomial such that g(ξ p ) = 0. If f = g, then we are done, hence we
assume that f 6= g. As ξ is a root of g(xp ), we obtain that f divides g(xp ). Therefore, modulo p,
f (x) divides g(xp ) = g(x)p . Hence f (x), g(x) have a non-trivial common factor h(x). As f (ξ) = 0
2
and g(ξ p ) = 0, both polynomials divide xn − 1, hence also f g divides xn − 1. This implies that h
n n n−1
divides x − 1 in Fp [x]. But x − 1 ∈ Fp [x] does not have multiple factors (the derivative nx
is non-zero and coprime to xn − 1). A contradiction.
24 SERGEY MOZGOVOY
3. Fields
3.1. Basic definitions. Recall that a field is a commutative ring such that every its nonzero
element is invertible. We know the fields Q, R, C. Also we know that for any prime integer p, the
quotient ring Fp = Z/pZ is a field. The next lemma gives an important source of new fields.
Lemma 3.1. Let K be a field and p ∈ K[x] be irreducible. Then K[x]/(p) is a field.
Proof. Let f ∈ K[x] be such that [f ] 6= 0 in K[x]/(p). If d = gcd(f, p), then d | p, hence d = p or
d = 1 (up to a unit). If d = p then p | f =⇒ f ∈ (p) =⇒ [f ] = 0, a contradiction. If d = 1, then
there exist u, v ∈ K[x] such that f u + pv = 1. This implies that [f ] [u] = 1 in K[x]/(p). Therefore
[f ] is invertible and K[x]/(p) is a field.
Definition 3.2. Let L be a field. A subring K ⊂ L is called a subfield if K is a field (when
equipped with an induced ring structure). The field L is called a field extension of K. We write
L/K in this case.
Lemma 3.3. Let L be a field. An intersection of a collection of subfields of L is a subfield.
Proof. Let (Ki )i∈I be a collection of subfield of L and let K = ∩i∈I Ki . It is clear that 0, 1 ∈ K.
For any a, b ∈ K we have a, b ∈ Ki ∀i ∈ I. Therefore a + b, ab ∈ Ki ∀i ∈ I. This implies
a + b, ab ∈ K. Therefore K ⊂ L is a subring. Finally, if a ∈ K is nonzero, then a−1 ∈ Ki ∀i ∈ I.
Therefore a−1 ∈ K. This means that K ⊂ L is a subfield.
Remark 3.4. Given a field K and a subset S ⊂ K, the intersection of all subfields of K that
contain S is a subfield called a subfield generated by S. It is the minimal subfield that contains S.
The subfield of K generated by ∅ (or by {0, 1}) is called the prime subfield of K. It is the
smallest subfield contained in K. ♦
Definition 3.5. Let K be a field. We define its characteristic char K to be the minimal integer
p ≥ 2 such that
p · 1 := 1 + · · · + 1 = 0
| {z }
p summands
if such p exists and zero otherwise.
Theorem 3.6. The characteristic of a field K is either zero or a prime number. The prime
subfield of K is
(1) Q if char K = 0.
(2) Fp if p = char K is prime.
Proof. Consider a map f : Z → K defined for n ≥ 0 by
f (n) = n · 1 := 1 + · · · + 1
| {z }
n summands
and f (−n) = −f (n). This is a ring homomorphism. Indeed, for any m, n ≥ 0, we have
f (m + n) = 1 + · · · + 1 = (1 + · · · + 1) + (1 + · · · + 1) = f (m) + f (n),
| {z } | {z } | {z }
m+n m n
f (mn) = 1 + · · · + 1 = (1 + · · · + 1) · (1 + · · · + 1) = f (m) · f (n)

| {z } | {z } | {z }
mn m n
and similarly for any m, n ∈ Z. Ideal ker f ⊂ Z can be written as ker f = (p) for some p ≥ 0.
If p = 0 then f is injective and therefore n · 1 6= 0 for any n > 0. This means that char K = 0.
Injective map f : Z → K can be extended to f : Q → K. We obtain a subfield Q ⊂ K generated
by 0, 1. This implies that Q is a prime subfield of K.
If p > 0 then p is the minimal positive integer such that p · 1 = 0 in K. Therefore char K = p.
Moreover, there is an injective ring homomorphism Z/(p) = Z/ ker f → K. Therefore Z/(p) is
an integral domain. If p is not prime, then there exist 1 < a, b < p such that ab = p. Then the
corresponding congruence classes a, b ∈ Z/(p) are nonzero and ab = 0. This contradicts to the
fact that Z/(p) is an integral domain. We conclude that p is prime and therefore Z/(p) is a field.
It is a subfield of K generated by 0, 1. Therefore it is a prime subfield of K.
3.2. Field extensions.

Definition 3.7. Let L/K be a field extension and a1 , . . . , an ∈ L. We denote by K(a1 , . . . , an ) ⊂ L
the subfield generated by K and a1 , . . . , an and denote by K[a1 , . . . , an ] ⊂ L the subring generated
by K and a1 , . . . , an : nX o
bi1 ,...,in ai11 . . . ainn bi1 ,...,in ∈ K .

Extension L/K is called simple if there exists an element a ∈ L such that L = K(a).
Example 3.8.
(1) Consider the field extension C/R. Then R(i) = R[i] = C.
(2) Let K be a field and L = K(x) be the field of rational functions over K. Then the ring
generated by K and x is the ring of polynomials K[x]. The field generated by K and x is
K(x) = L. √ √ √ √
(3) Let us show that √Q[ 2] is a field and therefore Q( 2) = Q[ 2]. Any element of Q[ 2] is
of the form a + b 2, where a, b ∈ Q. It’s inverse is
√
1 a−b 2 a b √
√ = 2 = − 2 ∈ Q[2].
a+b 2 a + 2b2 a2 + 2b2 a2 + 2b2
√ √
Therefore all nonzero elements of Q[ 2] are invertible in Q[ 2] and it is a field.
♦
Definition 3.9. Let L/K be a field extension.
(1) L can be considered in a natural way as a vector space over K: it is an abelian group and
the scalar multiplication K × L → L can be defined using multiplication in L. Define the
degree of the field extension
[L : K] = dimK L ∈ N ∪ {∞}.
(2) L/K is called a finite field extension if [L : K] < ∞.
√
Example 3.10. We have [C : R] = 2, [Q[ 2] : Q] = 2, K(x) : K = ∞, R : Q = ∞. ♦
Remark 3.11. Let K be a finite field and let p = char K. We know that Fp ⊂ K. Let n = [K : Fp ]
and (e1 , . . . , en ) be a basis of K over Fp . Then every element in K can be uniquely written in the
form
b1 e 1 + · · · + bn e n
for some b1 , . . . , bn ∈ Fp . Therefore the number of elements of K equals pn . This means that the
number of elements of a finite field is always a power of a prime number. One can show that for
any n ≥ 1 there exists a unique (up to an isomorphism) field having pn elements. It is denoted by
Fpn . ♦
Definition 3.12. Let L/K be a field extension. An element a ∈ L is called algebraic over K if
there exists a nonzero polynomial f ∈ K[x] such that f (a) = 0. Otherwise a is called transendental.
Example 3.13.
algebraic over R. It is a root of the polynomial x2 + 1 ∈ R[x].
i ∈ C is √
(1) The element √
(2) The element 2 ∈ Q[ 2] is algebraic over Q. It is a root of the polynomial x2 − 2 ∈ Q[x].
(3) The element x ∈ K(x) is transcendental over K.
(4) The element π ∈ R is transcendental over Q (Lindemann, 1882).
♦
Theorem 3.14. Let L/K be a field extension and a ∈ L be algebraic over K. Then
(1) Among all monic polynomials in K[x] that have root a there exists a unique polynomial p
having minimal degree. It is called the minimal polynomial of a over K.
(2) The minimal polynomial p ∈ K[x] of a is irreducible. If a is a root of f ∈ K[x], then p | f .
(3) There is an isomorphism of fields
K[x]/(p) → K[a] = K(a), f 7→ f (a).
26 SERGEY MOZGOVOY
(4) We have [K(a) : K] = deg p. If d = deg p, then (1, a, . . . , ad−1 ) is a basis of K(a) over K.
Proof. 1. Consider the ring homomorphism ϕa : K[x] → L given by f 7→ f (a). As a is algebraic,
the kernel ker ϕa ⊂ K[x] is a non-trivial prinicipal ideal. Let p ∈ K[x] be the monic polynomial
such that ker ϕa = (p). If f ∈ K[x] is another monic polynomial that has root a, then f ∈ (p) and
p | f . This implies that deg p ≤ deg f and if deg p = deg f then p = f .
2. We have K[x]/(p) ⊂ L and therefore K[x]/(p) is an integral domain. This implies that p is
irreducible. Indeed, if p = f g then f g = 0 in K[x]/(p) =⇒ f = 0 or g = 0 =⇒ p | f or p | g
and this means that p = f or p = g up to a unit. We have seen already that if f ∈ K[x] has root
a then p | f .
3. The kernel of the map ϕa : K[x] → L is (p) and its image is K[a]. This implies that
K[x]/(p) ' K[a]. We know that K[x]/(p) is a field as p is irreducible. Therefore K[a] is a field
and K(a) = K[a].
4. The basis of K[x]/(p) is given by (1, x, . . . , xd−1 ). Therefore the basis of K[a] = K(a) is
given by (1, a, . . . , ad−1 ).
Lemma 3.15. Let L/K be a finite field extension. Then any element a ∈ L is algebraic over K.
Proof. Let n = [L : K] = dimK L. Then the n + 1 elements 1, a, . . . , an are linearly dependent
over K. Therefore there exist elements f0 , . . . , fn ∈ K not all of which are zero such that
f0 + f1 a + · · · + fn an = 0.
Pn
Let f = i=0 fi xi ∈ K[x]. Then f 6= 0 and f (a) = 0.
Example 3.16. This lemma implies that any complex number a ∈ C is algebraic over R. More-
over, the elements 1, a, a2 are linearly dependent over R, hence there exists a nonzero polynomial
f = f2 x2 + f1 x + f0 ∈ R[x]
such that f (a) = 0. This means that every complex number a is a root of a quadratic polynomial
with real coefficients. ♦
Lemma 3.17. Let L/K be a field extension and a ∈ L be transcendental. Then K(a) ' K(x).
Proof. Consider a ring homomorphism ϕ : K[x] → L, f 7→ f (a). This map is injective as a as
otherwise a would be algebraic. All nonzero elements in K[x] are mapped to the invertible elements
in L, therefore we can extend ϕ to a ring homomorphism K(x) = Q(K[x]) → L. This map is
injective and its image is a field generated by K and a, that is, K(a). Therefore K(x) ' K(a).
Theorem 3.18. Consider finite field extensions K ⊂ L ⊂ M . Then
[M : K] = [M : L] · [L : K].
Proof. Let r = [L : K], s = [M : L], let l1 , . . . , lr be a basis of L/K and let m1 , . . . , ms be a basis
of M/L. We claim that the products li mj form a basis of M/K. Given x ∈ M we can write
X
x= yj mj , yj ∈ L
j=1
and we can write X

yj = xij li , xij ∈ K.
Therefore X X
x= yj mj = xij li mj
i,j
and this means that li mj generate M/K. P To prove that they are linearly independent assume that
we have elements xij ∈ K such that i,j xij li mj = 0. As m1 , . . . , ms are linearly independent
P
over L, we conclude that ij xij li = 0 for every j. As li are linearly independent over K, we
conclude that xij = 0 for all i, j. This implies that
{ li mj | 1 ≤ i ≤ r, 1 ≤ j ≤ s}
is a basis of M/K and therefore [M : K] = rs.
3.3. Splitting fields, Finite fields, Algebraically closed fields.

Definition 3.19. A field K is called algebraically closed if every non-constant polynomial in K[x]
has at least one root over K. Equivalently, any polynomial in K[x] splits over K, that is, it is a
product of linear polynomials.
Theorem 3.20 (Fundamental Theorem of Algebra). The field C of complex numbers is alge-
braically closed.
Remark 3.21. If K is an algebraically closed field and L/K is a finite field extension then L = K.
Indeed, we know that every a ∈ L is algebraic over K. Let p ∈ K[x] be the minimal polynomial
of a. Then p is a product of linear polynomials with coefficients in K, hence a ∈ K. We conclude
that L = K. ♦
Definition 3.22. Let K be a field.
(1) We say that a nonconstant polynomial f ∈ K[x] splits over a field extension L/K if there
exist a ∈ K and c1 , . . . , cn ∈ L such that
f (x) = a(x − c1 ) . . . (x − cn ).
(2) A field extension L/K is called a splitting field of f ∈ K[x] if f splits over L and L =
K(c1 , . . . , cn ), where c1 , . . . , cn ∈ L are the roots of f .
Theorem 3.23. Let K be a field and f ∈ K[x] be a polynomial of degree n ≥ 1. Then there exists
a (unique up to an isomorphism) splitting field L of f over K. Moreover, [L : K] ≤ n!.
Pd i
Proof. We will only prove existence. Let p = i=0 pi x be an irreducible divisor of f . Let
0 0
L = K[x]/(p), I = (p), and a = x + I ∈ L . Then
p(a) = p0 + p1 (x + I) + · · · + pd (x + I)d = (p0 + p1 x + · · · + pd xd ) + I = p(x) + I = I.
Therefore p(a) = 0 in L0 . This implies that f (a) = 0 and we can write
f (x) = (x − a)g(x)
0
for some polynomial g ∈ L [x] of degree n − 1. By induction on n, there exists a splitting field L
of g over L0 of degree [L : L0 ] ≤ (n − 1)!. This is also a splitting field of f over K. Note that
[L0 : K] = deg p ≤ deg f = n.
Therefore
[L : K] = [L : L0 ] · [L0 : K] ≤ (n − 1)! · n = n!.

Theorem 3.24. Every finite field has pn elements, for some prime p and integer n ≥ 1. For every
prime p and integer n ≥ 1, there exists a (unique up to an isomorphism) field with pn elements.
It is denoted by Fpn .
Proof. Let K be a finite field of characteristic p. Then p is prime and K contains a prime subfield
Fp = Z/pZ. If n = [K : Fp ], then #K = pn .
Let K be a field with q = pn elements. Then the set K × = K\{0} of invertible elements is a
group with respect to multiplication. It has q − 1 elements, hence aq−1 = 1 for all a ∈ K × . This
implies that aq = q
Qa for all a ∈ K. Therefore all elements of K are roots of qthe polynomial x − x,
q
hence x − x = a∈K (x − a). This implies that K is a splitting field of x − x over Fp , hence is
unique.
Let us prove existence of a field with q = pn elements. Let K be the splitting field of xq − x
over Fp . We claim that K is the set of q distinct roots of xq − x, hence contains q elements. Let
L ⊂ K be the set of all roots of f (x) = xq − x. Then L is a subfield of K:
(1) 0, 1 ∈ L as 0q = 0 and 1q = 1.
(2) If a, b ∈ L, then aq = a, bq = b =⇒ (ab)q = aq bq = ab =⇒ ab ∈ L.
(3) If 0 6= a ∈ L, then aq = a =⇒ (a−1 )q = a−q = a−1 =⇒ a−1 ∈ L.
28 SERGEY MOZGOVOY
(4) Let a, b ∈ L. We have (a + b)q = aq + bq (binomial coefficients kq are divisible by p if

0 < k < q). This implies (a + b)q = aq + bq = a + b =⇒ a + b ∈ L.

As L is a field containing all roots of xq − x (and nothing else), it is the splitting field of this
polynomial, hence K = L. We only need to show that f (x) = xq − x does not have multiple
roots. Indeed, assume that f (x) = (x − a)2 · g(x) for some g ∈ K[x] and a ∈ K. We can formally
define a derivative of polynomials. Then f 0 (x) = 2(x − a)g(x) + (x − a)2 g 0 (x) and f 0 (a) = 0. But
f 0 (x) = qxq−1 − 1 = −1 as q = pn = 0 in K. A contradiction. We conclude that all roots of xq − x
are distinct, hence #K = #L = q.
Lemma 3.25. Let K be a field. The following conditions are equivalent.
(1) Any non-constant polynomial in K[x] has at least one root over K.
(2) Any polynomial in K[x] splits over K, that is, it is a product of linear polynomials.
(3) Any irreducible polynomial over K is linear.
(4) Any algebraic element over K is contained in K.
(5) If L/K is a finite field extension, then L = K.
A field K satisfying these properties is called an algebraically closed field.
Proof. 1 =⇒ 2. Let f ∈ K[x] have degree n and c ∈ K be its root. Then f (x) = (x − c)g(x)
for some polynomial g ∈ K[x] of degree n − 1. By induction on n, we can write g(x) = a(x −
c1 ) . . . (x − cn−1 ). Therefore
f (x) = a(x − c)(x − c1 ) . . . (x − cn−1 ).
2 =⇒ 3. Let p ∈ K[x] be an irreducible polynomial. By assumption it is a product of linear
polynomials. Therefore p is linear.
3 =⇒ 4. Let L/K be a field extension and a ∈ L be algebraic over K. Then the minimal
polynomial of a over K is irreducible. By assumption it is linear. This implies that a ∈ K.
4 =⇒ 5. Let L/K be a finite field extension. Then any element a ∈ L is algebraic. By assumption
a ∈ K. This implies that L = K.
5 =⇒ 1. We can assume that f ∈ K[x] is irreducible. Then the field K[x]/(f ) is finite over K.
By assumption L = K[x]/(f ) = K and therefore
deg f = [L : K] = 1,
that is, f is linear.
Theorem 3.26 (Fundamental Theorem of Algebra). The field C of complex numbers is alge-
braically closed.
Proof. Let
f (z) = z n + fn−1 z n−1 + · · · + f0
be a non-constant polynomial over C that does not have roots over C. There exist r > 0 such that
for all z ∈ C with |z| > r, we have
|f (z)| > |f (0)| .
Let Dr = { z ∈ C | |z| ≤ r}. Then
inf |f (z)| = inf |f (z)| = |f (z0 )| > 0
z∈C z∈Dr
for some z0 ∈ Dr (as Dr is compact). If f (z) 6= 0 for all z ∈ C, then 1/f (z) is holomorphic and
bounded over C:
1 1
|f (z)| ≥ |f (z0 )| =⇒ ≤ < +∞.
|f (z)| |f (z0 )|
By the Liouville’s Theorem such function should be constant. Therefore f (z) is constant. A
contradiction.
3.4. Constructions with compass and straightedge. In this section we discuss constructions
on the plane R2 using a compass and a straightedge. We will usually identify R2 with C. Our
goal is to understand the set K ⊂ C of points that can be constructed using a compass and a
straightedge starting from the points 0 and 1. The following problems were studied already in
Ancient Greece:
(1) Duplication of a cube (construct a > 0 such that a3 = 2 · 13 = 2).
(2) Trisection of an arbitrary angle (given an angle ϕ ∈ [0, 2π] construct an angle ϕ/3, that
is, given a point eiϕ ∈ C, construct a point eiϕ/3 ).
(3) Quadrature of a circle (construct a > 0 such that a square with sides of length a has an
area of a radius one circle, that is, a2 = π)
We will show that these constructions are not possible with a compass and a straighedge.
Definition 3.27. Let K ⊂ C be a subset.
(1) A line through two distinct points in K is called an elementary K-constructible line.
(2) A circle that has some point in K and a center in K is called an elementary K-constructible
circle.
(3) All elementary K-constructible lines and circles are called elementary K-constructible
objects.
(4) A point in the intersection of two different elementary K-constructible objects is called
an elementary K-constructible point.
Lemma 3.28. Let L/K be a field extension and char K 6= 2. Then the following conditions are
equivalent
(1) [L : K] = 2.
(2) L = K[a] for some a ∈ L\K such that a2 ∈ K.
Such fields extension is called quadratic.
Proof. 1 =⇒ 2. Let b ∈ L\K. Its minimal polynomial has degree 2 and can be written in the
form
p(x) = x2 + cx + d = (x + c/2)2 + (d − c2 /4), c, d ∈ K.
From p(b) = 0 we obtain
(b + c/2)2 = c2 /4 − d.
Therefore the element a = b + c/2 ∈ L\K satisfies
a2 = c2 /4 − d ∈ K
and we have L = K[a].
2 =⇒ 1. The minimal polynomial of a over K is p(x) = x2 − a2 . Therefore
[L : K] = [K[a] : K] = deg p = 2.

Theorem 3.29. The following subsets of C coincide
(1) The smallest subset K ⊂ C that contains 0, 1 and all its elementary constructible points.
(2) The minimal subfield L ⊂ C that contains all its square roots.
Proof. L ⊂ K: To prove this we need to show that K is a field closed under taking square roots.
Then L ⊂ K as L is the minimal field with this property. Let us show first that K is a field.
Given z, w ∈ C, one can construct a parallelogram with vertices 0, z, w, z + w. Therefore one can
construct z + w. It is easy to construct −z. To construct zw or z/w, we represent z = aeiϕ and
w = aeiψ , where a, b ∈ R are the lengths of z, w and ϕ, ψ ∈ [0, 2π] are the angles between z, w and
the x-axis. Then zw = abei(ϕ+ψ) and similarly for z/w. One can easily add and subtract angles.
Therefore we just have to multiply and divide real positive numbers. In the following picture the
lines AC and BD are parallel and therefore a1 = cb , that is, c = ab. This implies that if we know
real a, b > 0, then we can construct c = ab. And if we know real b, c > 0, then we can construct
a = cb . This implies that K is a field.
30 SERGEY MOZGOVOY
C
b
O 1 A a B
is stable under taking square roots. We have to prove that given z = aeiϕ ,
Let us show that K √
we can also construct aeiϕ/2 . It is easy
√ to construct a bisector of an angle. Therefore we can
construct an angle ϕ/2. To construct a, consider the following picture, where we start with
intervals OA and AB, construct a circle with the diameter OB, and raise a perpendicular to OB
at the point A. The angle OCB is right.
O 1 A a B
We have
(12 + h2 ) + (a2 + h2 ) = OC 2 + BC 2 = OB 2 = (1 + a)2
√
which implies h2 = a, that is, h = a.
K ⊂ L: To prove this we will show that L 3 0, 1 is closed under elementary constructions.
Then K ⊂ L as K is the minimal set with this property. First, we claim that z ∈ L if and only if
Re z, Im z ∈ L ∩ R. If L ⊂ C is closed under taking square roots, then so is its conjugate L and
the intersection L ∩ L (if x2 = a for a ∈ L ∩ L then x ∈ L ∩ L). From the minimality of L, we
obtain L = L ∩ L =⇒ L = L. This implies that if z = x + iy ∈ L =⇒ z = x − iy ∈ L =⇒
x = 12 (z + z) ∈ L and iy = 21 (z − z) ∈ L. Note that i ∈ L as i2 = −1 ∈ L. Therefore y ∈ L.
Conversely, if x, y ∈ L ∩ R then also x + iy ∈ L.
An elementary L-constructible circle consists of points x + iy satisfying
(x − a)2 + (y − b)2 = c2
for some a, b, c ∈ L ∩ R. An elementary L-constructible line through the points x1 + iy1 ∈ L and
x2 + iy2 ∈ L (with x1 6= x2 , y1 6= y2 ) has an equation
x − x1 y − y1
= .
x2 − x1 y2 − y1
It can be written in the form
ax + by + c = 0,
where a = y2 − y1 ∈ L ∩ R, b = x1 − x2 ∈ L ∩ R and c = ax1 + by1 ∈ L ∩ R. Intersection points
of L-constructible circles and lines have coordinates (x, y), where x, y are solutions of linear and
quadratic equations with coefficients in L ∩ R. By assumption, L is closed under taking square
roots and therefore under taking quadratic extensions or taking roots of quadratic equations.
This implies that x, y ∈ L and therefore x + iy ∈ L. Therefore L is closed under elementary
constructions and K ⊂ L.
Remark 3.30. The points of K = L from the previous theorem are called constructible points.
The angle ϕ ∈ [0, 2π] is called constructible if the point eiϕ on the unit circle is constructible. ♦
Corollary 3.31. If a ∈ C is constructible, then it is algebraic over Q and its minimal polynomial
has degree 2k for some k ≥ 0.
Proof. By the previous theorem, there exists a chain of field extensions
Q = K0 ⊂ K1 ⊂ . . . ⊂ Kn
such that a ∈ Kn and [Ki : Ki−1 ] = 2 for 1 ≤ i ≤ n. Then
Yn
[Kn : Q] = [Ki : Ki−1 ] = 2n .
i=1
This implies that Kn is a finite field extension of Q. Therefore a is algebraic over Q. If p ∈ Q[x]
is the minimal polynomial of a, then
deg p = [Q[a] : Q]
is a divisor of [Kn : Q] = 2n as
[Kn : Q] = [Kn : Q[a]] · [Q[a] : Q].
Therefore deg p is a power of 2.
Corollary 3.32. The following constructions are impossible with a compass and a straightedge
(1) Duplication of a cube.
(2) Trisection of an arbitrary angle.
(3) Quadrature of a circle.
Proof. 1. Assume that we can construct a > 0 such that a3 = 2. The minimal polynomial of
a = 21/3 over Q is x3 −2. Its degree is 3 which is not a power of 2. Therefore a is not constructible.
2. Assume that we can trisect an arbitrary angle. One can construct angles π/3 and ϕ =
2π/3. Therefore the point eiϕ = e2πi/3 is constructible. We will show that the angle ϕ/3 is not
constructible, that is, the point z = eiϕ/3 = e2πi/9 is not constructible. We have
z 9 = e9·2πi/9 = e2πi = 1.
Therefore z is a root of the polynomial
x9 − 1 = (x3 − 1)(x6 + x3 + 1).
It is clear that z 3 6= 1. Therefore z is a root of the polynomial p(x) = x6 + x3 + 1. This polynomial
is irreducible. To see this we apply the Eisenstein’s criterion to the polynomial
p(x + 1) = (x + 1)6 + (x + 1)3 + 1 = x6 + 6x5 + 15x4 + 21x3 + 18x2 + 9x + 3
with prime 3. The irreducibility of p implies that p is a minimal polynomial of z. Its degree is 6
which is not a power of 2. Therefore z is not constructible.
3. If the quadrature of the circle is possible, then the value a > 0 satisfying a2 = π is
algebraic. This implies that a2 is also algebraic. But by the theorem of Lindemann (1882) π
is not algebraic.
32 SERGEY MOZGOVOY
4. Symmetric polynomials
Let k be a field. For every permutation σ ∈ Sn , define a map
σ : k[x1 , . . . , xn ] → k[x1 , . . . , xn ], f 7→ σf = f (xσ1 , . . . , xσn ).
This defines an action of the group Sn on the algebra k[x1 , . . . , xn ]. A polynomial f ∈ k[x1 , . . . , xn ]
is called symmetric if σf = f for all σ ∈ Sn . The set
Λn = k[x1 , . . . , xn ]Sn
of all symmetric polynomials is a subalgebra of k[x1 , . . . , xn ].
Example 4.1.
(1) For every 1 ≤ k ≤ n, the polynomial
X
ek = xi1 . . . xik
1≤i1 <···<ik ≤n
is a symmetric polynomial (of degree k) in Λn , called an elementary symmetric polynomial.

For example,
e1 = x1 + x2 + · · · + xn ,
e2 = x1 x2 + x1 x3 + x2 x3 + · · · + xn−1 xn ,
en = x1 . . . xn .
We can write
Yn
(x − xi ) = xn − e1 xn−1 + e2 xn−2 + · · · + (−1n )en .
i=1
(2) For every k ≥ 1, the polynomial

X
hk = xi1 . . . xik
1≤i1 ≤···≤ik ≤n
is a symmetric polynomial (of degree k) in Λn , called a complete symmetric polynomial.

For example,
h1 = x1 + x2 + · · · + xn ,
h2 = x21 + x1 x2 + x22 + x1 x3 + · · · + xn−1 xn + x2n .
(3) For every k ≥ 1, the polynomial
pk = xk1 + · · · + xkn
is a symmetric polynomial (of degree k) in Λn , called a power sum.
♦
Theorem 4.2. The elements e1 , . . . , en generate Λn over k and are algebraically independent.
This means that every element in Λn is a polynomial in e1 , . . . en and if
X
f (e1 , . . . , en ) = fi1 ,...,in ei11 . . . einn = 0, fi1 ,...,in ∈ k,
i1 ,...,in ≥0
then f = 0.
Proof. Consider the lexicographic order on Nn :
(i1 , . . . , in ) > (j1 , . . . , jn ) ⇐⇒ ∃k ≥ 1 : il = jl ∀l < k, ik > jk .
Let f be a symmetric polynomial and let xi11
. . . xinn be the highest monomial in f (with a non-zero
coefficient) wrt lexicographic order. Then i1 ≥ i2 ≥ · · · ≥ in (as otherwise we can exchange some
of ik and get a higher monomial). The highest monomial in
(1) e1i1 −i2 ei22 −i3 . . . einn
is
x1i1 −i2 (x1 x2 )i2 −i3 . . . (x1 . . . xn )in = xi11 xi22 . . . xinn .
We can subtract (1) from f (with an appropriate coefficient) in order to kill the highest monomial.
The highest monomial of the new polynomial has a smaller degree (with respect to the lexicographic
order) and we can assume by induction that the new polynomial can be expressed as a polynomial
of e1 , . . . , en .
Let us show that e1 , . . . , en are algebraically independent. The highest monomial of ek11 . . . eknn
is P P
ki ki
xk11 (x1 x2 )k2 . . . (x1 . . . xn )kn = x1 i≥1 x2 i≥2 . . .
This implies that if ek11 . . . eknn 6= el11 . . . elnn , then the corresponding highest monomials are differ-
ent. Given a non-trivial linear combination of products ek11 . . . eknn over k, consider the highest
monomials of all ek11 . . . eknn (having non-zero coefficients). The highest of them appears only
once, hence has a non-zero coefficient and the whole linear combination is nonzero. This implies
that the products ek11 . . . eknn are linearly independent and the elements e1 , . . . , en are algebraically
independent.
Remark 4.3. In the same way we can show that h1 , . . . , hn generate Λn and are algebraically
independent. Similarly (if k has characteristic zero) p1 , . . . , pn generate Λn and are algebraically
independent. ♦
Remark 4.4. For example in Λ2 , we have e1 = h1 = p1 = x1 + x2 and
e2 = x1 x2 , h2 = x21 + x1 x2 + x22 , p2 = x21 + x22 .
This implies
1 2
h2 = e21 − e2 = (p + p2 ), p2 = e21 − 2e2 .
2 1
♦
Remark 4.5. Consider generating functions
X n
Y
k
E(t) = ek t = (1 + xi t),
k≥0 i=1
n n
X
k
Y Y 1
H(t) = hk t = (1 + xi t + x2i t2 + ...) = .
i=1 i=1
1 − xi t
k≥0
Then
H(t)E(−t) = 1.
n n X k k
X 1 X xi t X pk
log H(t) = log = = tk .
i=1
1 − x i t i=1
k k
k≥1 k≥1
♦
34 SERGEY MOZGOVOY
Qn
4.1. Discriminant. Let f ∈ k[x] and L/k be its splitting field, so that f (x) = a i=1 (x − xi ),
where xi ∈ L are the roots of f . Define the discriminant of f
Y
∆ = a2n−2 (xi − xj )2
i<j
which is symmetric in x1 , . . . , xn . Note that ∆ is zero if and only if f has a multiple root. Let us
show that ∆ ∈ k. By the previous result ∆ is a polynomial in e1 , . . . , en , where
f (x) = xn − e1 xn−1 + e2 xn−2 + · · · + (−1)n en .
This means that ∆ is a polynomial in the coefficients of f , hence ∆ ∈ k.
Let n = 2 and f = x2 − e1 x + e2 = x2 + bx + c. Then we get the usual discriminant
∆ = (x1 − x2 )2 = e21 − 4e2 = b2 − 4c.
Let n = 3 and assume for simplicity that
f = x3 − e1 x2 + e2 x − e3 = x3 + bx + c,
that is, e1 = x1 + x2 + x3 = 0. As ∆ has degree 6, we have
∆ = ue23 + ve32 ,
for some u, v ∈ Z. If x3 = 0, then x1 + x2 = 0 and
x21 x22 (x1 − x2 )2 = vx31 x32 ,
hence v = −4. If x1 = x2 = 1, x3 = −2, then
0 = 4u − 4(1 − 2 − 2)3 ,
hence u = −27. We obtain
∆ = −27e23 − 4e32 = −27c2 − 4b3 .
Example 4.6. Let us compute the discriminant of xn − 1. Let ξ = e2πi/n . Then
n−1
n n Y Y
(ξ i − ξ j )2 = (−1)( 2 ) (ξ i − ξ j ) = (−1)( 2 )
Y Y
∆= ξ i (1 − ξ j−i )
0≤i<j<n i6=j i=0 j6=i
n−1
n n n n
= (−1)( 2 ) (1 − ξ k ) = (−1)( 2 ) ξ ( 2 ) nn = (−1)( 2 ) eπi(n−1) nn = ±nn .
Y Y
ξi
i=0 k6=0
n
−1
where we used the fact that k6=0 (x − ξ k ) = xx−1 = 1 + x + · · · + xn−1 and substituted x = 1.
Q
A discriminant of f is a polynomial in the coefficients of f . This implies that the discriminant of
xn − 1 is still ±nn if we consider xn − 1 as a polynomial in Fp [x]. In particular, if p | n, then ∆ = 0
and xn − 1 has multiple roots over Fp . If p 6| n, then ∆ 6= 0 and xn − 1 does not have multiple roots
over Fp . The last statement also follows from the fact that xn − 1 and its derivative nxn−1 are
are non-zero and coprime, hence xn − 1 can not have multiple factors (and in particular multiple
roots). ♦
5. Modules
5.1. Definition and examples. Modules over rings generalize the notion of a vector space over
a field.
Definition 5.1. Let R be a ring. A module over R (or an R-module) is an abelian group (M, +)
together with a map
R × M → M, (a, m) 7→ a · m = am,
called a multiplication, such that for all a, b ∈ R and m, n ∈ M
(1) a(bm) = (ab)m,
(2) 1m = m,
(3) a(m + n) = am + an,
(4) (a + b)m = am + bm.
Remark 5.2. If R is a field, then an R-module is usually called an R-vector space or a vector
space over R. ♦
Example 5.3. Any ring R is a module over itself. More generally, for any n ≥ 1, the product Rn
is an R-module with multiplication defined by a(x1 , . . . , xn ) = (ax1 , . . . , axn ). ♦
Example 5.4. Let I ⊂ R be an ideal. Then I is a module over R with a natural multiplication
R × I 3 (a, b) 7→ ab ∈ I. ♦
Example 5.5. Let f : R → S be a ring homomorphism and M be an S-module. Then M can be
considered also as an R-module. Multiplication is given by
a · m = f (a) · m, a ∈ R, m ∈ M.
We say that the R-module structure on M is obtained by restriction of scalars. In particular, S
is a module over R with multiplication given by
a · s = f (a) · s, a ∈ R, s ∈ S.
If I ⊂ R is an ideal, then there is a canonical ring homomorphism π : R → R/I. Therefore R/I
is an R-module, with multiplication given by a(b + I) = ab + I. ♦
Remark 5.6. Let M be an R-module, 0R be the zero element of R and 0M be the zero element
of M . Then
(1) 0R m = 0M for any m ∈ M .
(2) (−1)m = −m for any m ∈ M .
(3) a0M = 0M for any a ∈ R.
Indeed,
0R m + 0R m = (0R + 0R )m = 0R m.
Therefore 0R m = 0M . Similarly,
(−1)m + m = (−1 + 1)m = 0R m = 0M .
Therefore (−1)m = −m. Finally
a0M + a0M = a(0M + 0M ) = a0M .
Therefore a0M = 0M . ♦
Example 5.7. Let M be an abelian group. Then M is automatically a Z-module. Indeed, for
any k ≥ 0 and m ∈ M , define
k · m = m + ··· + m
| {z }
k summands
and (−k)m = −km. Note that this is a unique possible structure of a Z-module on M because
we should have 1 · m = m, 2 · m = (1 + 1)m = m + m and generally (k + 1)m = km + m. ♦
36 SERGEY MOZGOVOY
5.2. Homomorphisms and submodules.

Definition 5.8. Let R be a ring and M, N be R-modules.
(1) A map f : M → N is called a homomorphism of R-modules (or R-linear) if
(a) f (m + m0 ) = f (m) + f (m0 ) for any m, m0 ∈ M .
(b) f (am) = af (m) for any a ∈ R, m ∈ M .
(2) A bijective homomorphism f : M → N is called an isomorphism.
(3) A homomorphism f : M → M is called an endomorphism. A bijective endomorphism
f : M → M is called an automorphism.
(4) The set of all R-module homomorphism f : M → N is denoted by HomR (M, N ).
(5) The set of all R-module endomorphisms f : M → M is denoted by EndR (M ) = HomR (M, M ).
Lemma 5.9. Let M, N be two R-modules. Then
(1) HomR (M, N ) is an abelian group, with an addition defined by
(f + g)m = f (m) + g(m), f, g ∈ HomR (M, N ), m ∈ M.
(2) EndR (M ) = Hom(M, M ) is a ring, with an addition defined as above and a multiplication
defined by composition
(f g)m = f (g(m)), f, g ∈ HomR (M, N ), m ∈ M.
Example 5.10. Let K be a field and V = K n be a vector space of dimension n. Given a matrix
A = (aij ) ∈ Mn (K), we can associate with it a linear map
 t
X X
fA : V → V, x = (x1 , . . . , xn )t 7→ Ax =  a1j xj , . . . , anj xj  .
j j
The map
Mn (K) → EndK (V ), A 7→ fA
is an isomorphism of rings. GivenPa linear map f : V → V , one reconstructs the matrix A =
(aij ) ∈ Mn (K) by the rule f ej = i aij ei , where (e1 , . . . , en ) is the standard basis of V = K n .
We will often identify A ∈ Mn (K) and fA ∈ EndK (V ). ♦
Remark 5.11. If M is an abelian group, then the ring EndZ (M ) is denoted by End(M ). If M is
an R-module, then there is a ring homomorphism
ϕ : R → End(M ), ϕ(a)(m) = am, a ∈ R, m ∈ M.
For example
ϕ(ab)m = (ab)m = a(bm) = ϕ(a)(bm) = ϕ(a)(ϕ(b)m)
and therefore ϕ(ab) = ϕ(a)◦ϕ(b). Conversely, given an abelian group M and a ring homomorphism
ϕ : R → End(M ), we can equip M with an R-module structure
a · m = ϕ(a)(m) ∈ M, a ∈ R, m ∈ M.
♦
Example 5.12. Let V be a vector space over a field K. Given a K-linear map A ∈ EndK (V ),
the evaluation map
K[x] → EndK (V ), f 7→ f (A)
is a ring homomorphism, hence V gets a structure of a K[x]-module
f · v = f (A)(v), f ∈ K[x], v ∈ V.
Conversely, if V is a K[x]-module (extending the K-vector space structure on V ), then multipli-
cation by x induces a K-linear map A : V → V . ♦
Definition 5.13. Let M be an R-module. A subset N ⊂ M is called a submodule if N is a
subgroup and for any a ∈ R, m ∈ N we have am ∈ N (that is, RN ⊂ N ).
Example 5.14. Let R be a commutative ring. Then the submodules of R are precisely the ideals
of R. ♦
Lemma 5.15. Let f : M → N be a homomorphism of R-modules. Then

(1) The set ker f = { m ∈ M | f (m) = 0} is a submodule of M .
(2) The set im f = { f (m) | m ∈ M } is a submodule of N .
(3) f is injective if and only if ker f = 0.
Remark 5.16. Given an R-module M and a submodule N ⊂ M , one can define the quotient
M/N of abelian groups (similar to the construction of the quotient ring R/I for an ideal I ⊂ R).
Define an equivalence relation on M by
m ∼ m0 ⇐⇒ m − m0 ∈ N.
The equivalence class of m ∈ M is
[m] = m + N = { m + n | n ∈ N } .
The set of equivalence classes is denoted by M/N . It has a structure of an abelian group defined
by
[m] + [m0 ] = [m + m0 ].
♦
Theorem 5.17. Let M be an R-module and N ⊂ M be a submodule. Then
(1) The quotient group M/N has a structure of an R-module defined by
a · [m] = [am], a ∈ R, m ∈ M.
(2) The natural map π : M → M/N , m 7→ [m], is a homomorphism of R-modules.
(3) If f : M → M 0 is a homomorphism of R-modules, then there exists a unique homomor-
phism f¯ : M/ ker f → M 0 that makes the following diagram commute (f¯ ◦ π = f )
π
M M/ ker f
f f¯
M0
∼
The map f¯ induces an isomorphism f¯ : M/ ker f −
→ im f .
38 SERGEY MOZGOVOY
5.3. Simple and indecomposable modules.

Definition 5.18. Let R be a ring and M be a nonzero R-module.
(1) A module M is called simple (or irreducible) if it does not contain any submodules except
zero and itself.
(2) A module M is called indecomposable if it can not be written as an (internal) direct sum
of nonzero submodules, that is, there are no nonzero submodules N, N 0 ⊂ M such that
N ∩ N 0 = 0 and N + N 0 = M .
(3) A submodule N ( M is called maximal, if there are no submodules N ( L ( M .
Remark 5.19. Any simple module is indecomposable. ♦
Remark 5.20. Given two R-modules N, N 0 , define a new R-module N ⊕ N 0 , called an (external)
direct sum of N and N 0 , to be the product of abelian groups N ×N 0 equipped with a multiplication
a(n, n0 ) = (an, an0 ), a ∈ R, n ∈ N, n0 ∈ N 0 .
A module M is indecomposable if and only if it is not isomorphic to a direct sum N ⊕ N 0 for some
nonzero modules N, N 0 . ♦
Example 5.21. If K is a field, then K is a simple module over itself. ♦
Example 5.22. Let R be a PID and p ∈ R be irreducible. Then
(1) The module (p) ⊂ R is maximal: if (p) ⊂ (a) ⊂ R =⇒ a | p =⇒ a = 1 or a = p (up to
a unit) =⇒ (a) = R or (a) = (p). The quotient R/(p) is a field and is a simple module
over R/(p) and over R.
(2) The module M = R/(p2 ) is not simple. It has a submodule N = (p)/(p2 ). The quotient
M/N is isomorphic to R/(p). Note that N = (p)/(p2 ) is also isomorphic to R/(p):
R/(p) → (p)/(p2 ), a 7→ ap.
2
(3) The module M = R/(p ) is indecomposable. Indeed, for any submodule N ⊂ M , there
exists an ideal (p2 ) ⊂ (a) ⊂ R such that N = (a)/(p2 ). We obtain from a | p2 that
a = 1, p or p2 (up to a unit). Therefore the only submodules of M = R/(p2 ) are 0 =
(p2 )/(p2 ) ⊂ (p)/(p2 ) ⊂ (1)/(p2 ) = M . If N1 ⊕ N2 = M then N1 ⊂ N2 or N2 ⊂ N1 .
Assuming the former without loss of generality, we obtain N1 = N1 ∩ N2 = 0, hence M is
indecomposable.
(4) Let p, q ∈ R be coprime. Then M = R/(pq) is an (internal) direct sum of N1 = (p)/(pq) ⊂
M and N2 = (q)/(pq): let a ∈ R be such that [a] ∈ N1 ∩ N2 . Then p | a and q | a =⇒
pq | a =⇒ [a] = 0 in R/(pq). Therefore N1 ∩ N2 = 0. There exist u, v ∈ R such that
1 = up + vq =⇒ 1 ∈ N1 + N2 =⇒ N1 + N2 = R/(pq) = M .
This proves that M = R/(pq) is not indecomposable. Note that R/(q) ' (p)/(pq) = N1 ,
[a]q 7→ [ap]pq and similarly R/(p) ' (q)/(pq) = N2 . This implies that R/(pq) ' R/(p) ⊕
R/(q) (external direct sum).
♦
Lemma 5.23 (Schur’s lemma). For any simple R-module M , the endomorphism ring EndR (M )
is a division ring, that is, any endomorphism M → M is either zero or invertible.
Proof. Assume that f : M → M is nonzero. Then ker f ⊂ M is a submodule not equal to M .
Therefore ker f = 0 and f is injective. On the other hand im f ⊂ M is a submodule no equal
to zero. Therefore im f = M and f is surjective. This implies that f is bijective and therefore
invertible.
Corollary 5.24. Let R be an algebra over an algebraically-closed field K. If M is a simple
R-module, finite-dimensional over K, then EndR (M ) = K.
Proof. We know that S = EndR (M ) is a division ring. As M is f.d., EndK (M ) is also f.d. For
every a ∈ S, the elements 1, a, . . . , an ∈ S ⊂ EndK (M ) are linearly dependent over K for large n.
Therefore there exists monic f ∈ K[x] with f (a) = 0. As K is algebraically closed, we can write
Q Q
f = i (x − ci ) with ci ∈ K. Then f (a) = i (a − ci ) = 0. The ring S is an integral domain and
a − ci ∈ S for all i. Therefore a − ci = 0 for some i, hence a = ci ∈ K.
Alternative proof: Let f : M → M be an R-homomorphism. As a linear operator on a K-vector
space, f admits an eigenvector v ∈ M with an eigenvalue λ ∈ K (as K is algebraically closed).
Then f 0 = f − λ Id ∈ EndR (M ), f 0 (v) = f (v) − λv = 0 =⇒ ker f 0 6= 0. But ker f 0 ⊂ M is a
submodule and M is simple, hence ker f 0 = M =⇒ f 0 = 0. This implies f = λ Id.
Lemma 5.25. Let M be an R-module and N ⊂ M be a submodule. Then N ⊂ M is maximal if
and only if M/N is simple.
Proof. Consider a canonical homomorphism π : M → M/N . There is a bijection between the
submodules N ⊂ L ⊂ M and the submodules of M/N given by L 7→ π(L) = L/N ⊂ M/N . The
inverse is given by
M/N ⊃ L0 7→ π −1 (L0 ) ⊂ M.
The submodule N ⊂ M is maximal ⇐⇒ N and M are the only submodules N ⊂ L ⊂ M
⇐⇒ 0 and M/N are the only submodules of M/N ⇐⇒ M/N is simple.
Lemma 5.26. Let R be a commutative ring and M be a simple R-module. Then there exists a
maximal ideal I ⊂ R such that M ' R/I.
Proof. Let m ∈ M be a nonzero element. Consider a homomorphism of R-modules
f : R → M, f (r) = rm.
Then im f ⊂ M is a nonzero submodule and therefore im f = M . Let I = ker f . Then I is a
submodule and therefore an ideal of R. There is an isomorphism
R/I ' im f = M.
As M is simple, we obtain by the previous lemma that I ⊂ R is a maximal submodule and a
maximal ideal.
Definition 5.27. Let R be a ring and M be an R-module. A composition series of a module M
is a chain of submodules
0 = M0 ⊂ M1 ⊂ . . . ⊂ Mn = M
such that the modules Mi /Mi−1 are simple for 1 ≤ i ≤ n. The modules Mi /Mi−1 are called
the subquotients of the composition series. The number n is called the length of the composition
series.
Theorem 5.28 (Jordan-Hölder Theorem). Let M be an R-module that has at least one compo-
sition series. Then any two composition series of M have the same length and have isomorphic
subquotients (up to a permutation). The length of a composition series is called the length of M .
Example 5.29. Let R be a PID and p, q ∈ R be two irreducible elements, not associate to each
other. Then M = R/(pq) ' R/(p) ⊕ R/(q) has the following filtrations:
M0 = 0, M1 = R/(p), M2 = M,
M00= 0, M10 = R/(q), M20 = M.
The corresponding subquotients are
M1 /M0 ' R/(p), M2 /M1 ' R/(q),
M10 /M00' R/(q), M20 /M10 ' R/(p).
The tuples of subquotients are the same up to a permutation. ♦
40 SERGEY MOZGOVOY
5.4. Chinese remainder theorem. We have seen that given a PID R and coprime elements
p, q ∈ R, we have
R/(pq) ' R/(p) ⊕ R/(q).
We will generalize this statement by proving the Chinese Remainder Theorem. In its classical
form, formulated in the 3rd century AD by Sunzi (not to be confused with Sun Tzu, the author
of “The art of war”) it is
Theorem 5.30. Let n1 , . . . , nk be positive, pairwise coprime integers. Then for any integers
a1 , . . . , ak , there exists an integer a such that
a ≡ ai (mod ni ), ∀i = 1, . . . , k.
There exists a unique such integer with 0 ≤ a < n = n1 . . . nk .
This theorem can be formulated as a statement that the map
Z/nZ → Z/n1 Z × . . . × Z/nk Z, a + nZ 7→ (a + n1 Z, . . . , a + nk Z)
is an isomorphism (of rings or of abelian groups).
Proof. TheQabove map is injective. Indeed, if [a] = a + nZ is mapped to zero, then ni | a for all i,
hence n = i ni | a as the elements ni are pairwise coprime. This implies that [a] = 0. Injectivity
of the map implies bijectivity as the groups on both sides have the same number of elements.
We can generalize the above result to arbitrary PID.
Theorem 5.31. Let R be a PID and n1 , . . . , nk be pairwise coprime elements (that is, gcd(ni , nj ) =
1 for i 6= j). Then the map
R/(n) → R/(n1 ) × . . . × R/(nk ), a + (n) 7→ (a + (n1 ), . . . , a + (nk )),
is an isomorphism of rings (or R-modules), where n = n1 . . . nk .
Remark 5.32. In particular, we see that for any coprime p, q ∈ R, we have an isomorphism
R/(pq) ' R/(p) ⊕ R/(q) of R-modules. ♦
This statement, in its own right, can be generalized as follows
Theorem 5.33. Let R be a commutative ring and I1 , . . . , Ik be ideals of R that are pairwise
coprime: Ii + Ij = R, i 6= j. Then the map
ϕ : R/I → R/I1 × . . . × R/Ik , a + I 7→ (a + I1 , . . . , a + Ik ),
is an isomorphism of rings (or R-modules), where I = ∩i Ii . Moreover I = I1 I2 . . . Ik .
Proof. The kernel of the map
R → R/I1 × . . . × R/Ik
is ∩i Ii = I. This implies injectivity of ϕ. Let us prove surjectivity. For any i 6= j, we can find
eij ∈ Ii and eji ∈ Ij such that 1 = eij + eji . Then, for any i, we have
Y Y
1= (eij + eji ) ∈ Ii + eji .
j6=i j6=i
Q Q
Let ei ∈ Ii and fi = j6=i eji ∈ j6=i Ij be such that 1 = ei + fi . Given elements [ai ] ∈ R/Ii for
P
all i, we claim that a = j fj aj satisfies a ≡ ai (mod Ii ) for all i. Indeed,
X
a − ai = a − (ei + fi )ai = fj aj − ei ai ∈ Ii
j6=i
as ei ∈ Ii and fj ∈ Ii for j 6= i.T Therefore

Q a ≡ ai (modQIi ) for T all i and the map ϕ is surjective.
We also
T have to show that I
i i = I
i i . Inclusion I
i i ⊂ i Ii is trivial. On the other hand,
let a ∈ i Ii . Then by induction a ∈ J = I1 . . . Ik−1 . Therefore
Y
a = a(ek + fk ) ∈ JIk + Ik J = Ii
i
Q T Q
as ek ∈ Ik , fk ∈ j6=k Ij = J and a ∈ Ik . This proves that i Ii ⊂ i Ii .
5.5. Modules over PID. Consider the following two fundamental results of linear algebra and
the theory of finite abelian groups. In linear algebra one proves that every square matrix over C
is conjugate to its Jordan canonical form which is a direct sum of Jordan blocks
 
λ 1 0 ...... 0
0 λ 1 . . . . . . 0
 
0 0 λ . . . . . . 0
Jn,λ =   λ ∈ C, n ≥ 1.
. . . . . . . . . . . . . . . . . . . .

0 0 0 . . . λ 1
0 0 0 ... 0 λ
The fundamental theorem of finitely generated abelian groups states that every such group is
isomorphic to a direct sum
Zk ⊕ Z/(pn1 1 ) ⊕ · · · ⊕ Z/(pnr r ),
where k ≥ 0, pi ∈ Z are prime numbers and ni ≥ 1. In particular, every finite abelian group is
isomorphic to
Z/(pn1 1 ) ⊕ · · · ⊕ Z/(pnr r ).
We will see that these two results are essentially equivalent, if seen from an appropriate point of
view, and then we will give a general unified proof.
Given a vector space V = K n over a field K and a linear operator A ∈ EndK (V ) = Mn (K),
we can equip V with a structure of a K[x]-module (we denote it by VA )
f · v = f (A)(v), f ∈ K[x], v ∈ V.
Conversely, if V is a K[x]-module, then it is a K-vector space and we can define
A ∈ EndK (V ), A(v) = x · v, v ∈ V.
n
Example 5.34. Consider the K[x]-module V = K[x]/(x − λ) , where λ ∈ K and n ≥ 1. Choose
the basis (x − λ)n−1 , . . . , (x − λ)2 , (x − λ), 1 of V . In this basis we have
A(x − λ)k = x(x − λ)k = (x − λ)k+1 + λ(x − λ)k
for k < n − 1 and A(x − λ)n−1 ≡ λ(x − λ)n−1 (mod (x − λ)n ). Therefore the matrix of A in this
basis is exactly the Jordan block Jn,λ . ♦
Let W be another K-vector space, B ∈ EndK (W ) and WB be the corresponding K[x]-module.
Then an isomorphism g : VA → WB of K[x]-modules can be identified with a K-linear isomorphism
g : V → W such that gA = Bg:
gA(v) = g(x · v) = x · g(v) = Bg(v), v ∈ V.
−1
This means that B = gAg and the corresponding matrices are conjugate if V = W = K n . The
statement that a matrix A ∈ Mn (K) (with K = C) is conjugate to a direct sum of Jordan blocks
Jn1 ,λ1 , . . . , Jnr ,λr can be translated now to the statement that the K[x]-module VA is isomorphic
to a direct sum of K[x]-modules
K[x]/(x − λ1 )n1 ⊕ · · · ⊕ K[x]/(x − λr )nr .
The polynomials (x − λ)n are powers of irreducible polynomials x − λ ∈ K[x] and these are the
only irreducible (or prime) elements in K[x] (up to a unit) if K = C or K is algebraically closed.
In the same way for abelian groups we had summands Z/(pn ), where p is prime. This shows that
both statements are essentially equivalent, with the first statement being about modules over K[x]
and the second statement about modules over Z (that is, abelian groups). Both of them follow
from the general result we will prove next.
Definition 5.35. Let M be an R-module.
(1) A family of elements (mi )i∈I in M generates M if every element of M can be written in
the form X
ai mi , ai ∈ R
i∈I
with all but finitely many ai equal zero.
42 SERGEY MOZGOVOY
(2) M is called finitely generated over R if there exists a finite family (m1 , . . . , mk ) of elements
in M that generates M .
Theorem 5.36. Let R be a PID. A finitely generated R-module is isomorphic to a direct sum
Rk ⊕ R/(pn1 1 ) ⊕ · · · ⊕ R/(pnr r ),
where k ≥ 0, pi ∈ R are prime and ni ≥ 1. The modules R and R/(pn ) are indecomposable.
Proof. Let M be a finitely generated R-module with generators x1 , . . . , xm . Then there is a
surjective homomorphism ψ : Rm → M (where Rm 3 ei 7→ xi ). The kernel ker ψ ⊂ Rm is finitely
generated (see below). Therefore there is a surjective homomorphism φ : Rn → ker ψ and we
obtain a sequence of maps
ϕ ψ
Rn −→ Rm − → M,
where im ϕ = ker ψ so that
coker ϕ := Rm / im ϕ = Rm / ker ψ ' M.
We can represent ϕ as an m × n matrix with coefficients in R. By changing the bases in Rm and
Rn we can put this matrix to the form, called a Smith normal form,
 
a1 0 0 ..... 0
 0 a2 0 . . . . . 0
 
 0 0 . . . . . . . . 0
 
 
 . . . . . . . . . . . ar . . . . 
 
 0 0 . . . . . . . 0 0
0 0 ....... 0 0
with nonzero a1 | a2 | · · · | ar and r ≤ m, n. The elements ai are unique (up to a unit). The
algorithm is somewhat involved for general PID, but in the case of Euclidean domains one can use
the usual Gaussian elimination on rows and columns to get the required form. I omit the details.
We obtain then
M ' Rm / im ϕ = R/(a1 ) ⊕ · · · ⊕ R/(ar ) ⊕ Rm−r .
If a ∈ R is nonzero and a = i pki i is a factorization into irreducible (distinct) factors, then
Q
R/(a) ' R/(pk11 ) ⊕ R/(pk22 ) ⊕ . . .

by the Chinese Remainder theorem. We obtain the required decomposition of the module M .
Let us see that R is indecomposable. If N1 = (a) ⊂ R, N2 = (b) ⊂ R and N1 , N2 6= 0, then
0 6= ab ∈ N1 ∩ N2 , hence N1 ∩ N2 6= 0 and R 6= N1 ⊕ N2 . Similarly one can show that R/(pn ) is
indecomposable.
5.6. Noetherian modules. In the proof of the theorem we used the fact that a submodule of
Rn is finitely generated if R is a PID. Let us discuss this in more detail.
Definition 5.37. Let R be a ring.
(1) An R-module M is called Noetherian if every submodule of M is finitely generated.
(2) The ring R is called Noetherian if it is Noetherian as a (left) R-module over itself.
Example 5.38. A PID R is Noetherian. Indeed, every submodule of R is an ideal, hence a
principal ideal, generated by one element. ♦
Lemma 5.39. Let R be a ring and M be an R-module. The following conditions are equivalent
(1) Every submodule of M is finitely generated.
(2) Every increasing chain of submodules
M1 ⊂ M2 ⊂ . . . ⊂ M
stabilizes, that is, Mn = Mn+1 = . . . for n 0.
Proof. 1 =⇒ 2. Consider an increasing chain
M1 ⊂ M2 ⊂ . . . ⊂ M
and let N = ∪n≥1 Mn ⊂ M . Then N is a submodule of M and by assumption it is finitely generates.
Let x1 , . . . , xk be generators of N . Then xi ∈ Mni for some ni ≥ 1. Taking n = maxi ni , we obtain
xi ∈ Mn for all i, hence N ⊂ Mn and Mn = Mn+1 = . . . .
2 =⇒ 1. Let N ⊂ M be a submodule and let x0 = 0 ∈ N . Assuming that elements x0 , . . . , xk
in N are constructed, let Mk ⊂ N be the module generated by them. If Mk = N then N is finitely
generated and we are done. If Mk 6= N , let xk+1 ∈ N \Mk and continue the procedure. In this
way we obtain a chain of modules
M1 ⊂ M2 ⊂ . . . ⊂ N ⊂ M
with Mk 6= Mk+1 for all k ≥ 1. A contradiction.
Lemma 5.40. Let M be an R-module and N ⊂ M be a submodule. Then M is Noetherian ⇐⇒
N and M/N are Noetherian.
Proof. Let M be Noetherian. If L ⊂ N is a submodule, then L ⊂ M , hence L is finitely generated
and N is Noetherian. Let L ⊂ M/N be a submodule and let π : M → M/N be the projection.
The module L0 = π −1 (L) ⊂ M is finitely generated, hence also L = π(L0 ) is finitely generated and
M/N is Noetherian.
Assume that N and M/N are Noetherian and let L ⊂ M . Then L ∩ N ⊂ N is finitely generated
and L/(L ∩ N ) ' (L + N )/N ⊂ M/N is finitely generated. This implies that L is also finitely
generated.
Corollary 5.41. If M, N are Noetherian R-modules, then M ⊕ N is also Noetherian.
Proof. Let M 0 = M ⊕ N . Then N ⊂ M 0 and M 0 /N ' M are Noetherian. We conclude that M 0
is Noetherian.
Corollary 5.42. If R is a Noetherian ring, then the module Rn is Noetherian for all n ≥ 1.
Proof. The module Rn is a direct sum of copies of R.
n
Corollary 5.43. If R is a PID, then every submodule of R is finitely generated.
Proof. We have seen that R is Noetherian. Therefore Rn is also Noetherian. This means that
every submodule of Rn is finitely generated.
Lemma 5.44. Let M be a Noetherian module over a ring R and f : M → M be a surjective
homomorphism. Then f is an isomorphism.
Proof. Consider the chain of modules ker f ⊂ ker f 2 ⊂ . . . ⊂ M . As M is Noetherian, this

chain stabilizes =⇒ ∃n ≥ 1, ker f n = ker f 2n =⇒ im f n ∩ ker f n = 0. Indeed, if f n (x) ∈
ker f n =⇒ f 2n (x) = 0 =⇒ x ∈ ker f 2n = ker f n =⇒ f n (x) = 0. As f is surjective:
im f n = M =⇒ ker f n = im f n ∩ ker f n = 0 =⇒ ker f = 0 =⇒ f is injective.

Fields Rings and Modules Lecture Notes

Uploaded by

Copyright:

Available Formats

Fields Rings and Modules Lecture Notes

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fields Rings and Modules Lecture Notes

Uploaded by

Copyright:

Available Formats

LECTURE NOTES

MA2314: FIELDS, RINGS AND MODULES (2017)

Date: April 4, 2017.

1.2. Ideals and quotient rings. Let R be a ring.

1.3. Ring homomorphisms.

Definition 1.31. Let ϕ : R → S be a ring homomorphism. Define

and define their product f g ∈ R[x] by

These operations define a structure of a ring on R[x].

define the evaluation of f at s ∈ S (or the substitution of s into f ) to be

2.4. GCD and LCM. Let R be an integral domain.

2.6. Field of fractions.

f (a)/f (s) and uniqueness follows.

2.8. Cyclotomic polynomials.

The coefficient of y 0 is p1 = p is not divisible by p2 . By Eisenstein criterium our polynomial is

f (mn) = 1 + · · · + 1 = (1 + · · · + 1) · (1 + · · · + 1) = f (m) · f (n)

3.2. Field extensions.

and we can write X

3.3. Splitting fields, Finite fields, Algebraically closed fields.

(4) Let a, b ∈ L. We have (a + b)q = aq + bq (binomial coefficients kq are divisible by p if

0 < k < q). This implies (a + b)q = aq + bq = a + b =⇒ a + b ∈ L.

is a symmetric polynomial (of degree k) in Λn , called an elementary symmetric polynomial.

(2) For every k ≥ 1, the polynomial

is a symmetric polynomial (of degree k) in Λn , called a complete symmetric polynomial.

5.2. Homomorphisms and submodules.

Lemma 5.15. Let f : M → N be a homomorphism of R-modules. Then

5.3. Simple and indecomposable modules.

as ei ∈ Ii and fj ∈ Ii for j 6= i.T Therefore

R/(a) ' R/(pk11 ) ⊕ R/(pk22 ) ⊕ . . .

Proof. Consider the chain of modules ker f ⊂ ker f 2 ⊂ . . . ⊂ M . As M is Noetherian, this

You might also like