Minimal, Primitive, and Irreducible Polynomials

Wayne Edward Aitken

June 2021 Edition∗

This document covers the concepts of minimal and primitive polynomials. It

also discusses ways to show that a polynomial is irreducible. Aside from the first
section, proofs are provided with minor details left to the reader, sometimes in the
form of exercises.
This document was originally written as a handout for my Math 520 course
in Fall 2010. Based on the aims of the course, these notes emphasize polynomials
in Q[X] and Z[X]. However, much of the material can be generalized. For example,
much of the material on primitive polynomials extends to R[X] where R is a UFD.

1 Background
We begin by stating prerequisites and reviewing some basic materials concerning
polynomial rings.

1.1 Prerequisites
I will assume that the reader has had a good one-semester introduction to ab-
stract algebra, and so is comfortable with groups, rings, fields, integral domains,
and ideals. I will assumes a basic familiarity with standard rings and fields such
as Z, R, C, Fp . The basics of polynomial rings R[X] are also assumed, but most of
what is needed in reviewed in Section 1.2 below.
The basics of the theory of factorization in UFDs is assumed. So ideas such
as irreducible, units, associates are assumed. Some basics concerning Euclidean
domains and PIDs are also assumed. Actually not much is needed about PIDs
and UFDs in general, but only as they related to polynomial rings in a single
variable, and Z of course. Section 1.2 mentions Noetherian rings and the Hilbert
basis theorem. This is not essential for what follows, but was mentioned to put
polynomial rings in context.
Basic facts about quotient rings are assumed. Understanding of prime ideals
and maximal ideals are required together the following:
Theorem 1. Let R be a commutative ring with unity and let I be an ideal of R.
Then R/I is an integral domain if and only if I is a prime ideal. Similarly, R/I is
a field if and only if I is a maximal ideal.
∗ Version of June 8, 2021. Copyright c 2010–2021 by Wayne Edward Aitken. This work

is made available under a Creative Commons Attribution 4.0 License. Readers may copy and
redistribute this work under the terms of this license.

1
 The last section on cyclotomic polynomials assumes knowledge of roots of unit
in C using exponential notation. The proof of the main theorem in that section
assumes that reader knows, or can prove, that (X −1)p ≡ X p −1 modulo a prime p.

1.2 Polynomial Rings

We review some basics concerning polynomial rings. If R is a commutative ring
with unity and X is a symbolic variable, then R[X] is the ring of polynomials
with coefficients in R and variable X. There is a formal way to construct R[X] in
terms of finite sequences of elements in R using suitable definitions for addition and
multiplication. In practice the usual, somewhat informal, way of thinking about
polynomials is reliable enough. One can add and multiply polynomials in the usual
way, and these binary operations satisfy all the usual laws. In fact, R[X] is a
commutative ring. If R is an integral domain, then so is R[X]. If F is a field,
then F [X] is not a field, but is a rather special type of integral domain: F [X] is a
Euclidean domain. Since F [X] is a Euclidean domain, is also a PID and a UFD.

Definition 1. Suppose S is a ring with subring R. Then we call S an extension

of R, or we say that S is a ring extension of R. If S happens to be a field we call S
a field extension of R.

Remark. The terms subring and extension are often used in a situation that goes
a bit beyond the above definition. Suppose that we have an injective ring homo-
morphism R → S which we fix once and for all, and treat as canonical. Then the
image of R → S is a subring of S which is isomorphic to R. We identify R with
its image in S, and we consider R to be a subring of S. We consider S to be an
extension of R.
This is what happens with the standard map R → R[X] sending any given
element c ∈ R to the corresponding constant polynomial in R[X]. We consider
this injective homomorphism as canonical, we view R as a subring of R[X], and we
consider R[X] as an extension of R. The subring R of R[X] is called the coefficient
ring of R[X].
Degrees of nonzero polynomials are defined in the usual way. If the coefficient
ring R is an integral domain then the degree of a product will be the sum of the
degrees of the factors. (If R is not an integral domain, we get an inequality at
least).
There is a division algorithm and a corresponding quotient-remainder theorem
for F [X] when F is a field. This is why F [X] is a Euclidean domain. All Euclidean
domains have the property that every ideal is principal, which means that F [X] is
a PID (Principal Ideal Domain). All PIDs have unique factorization: every nonzero
element factors uniquely into a unit times a finite number of irreducible elements
(perhaps zero irreducible factors, and perhaps with repetitions of irreducible fac-
tors). In other words, F [X] is a UFD. The division algorithm and a form of the
quotient-remainder theorem are valid more generally in R[X] where R is a com-
mutative ring (with unity) but only when we divide g ∈ R[X] by f ∈ R[X] where
the leading coefficient of f is a unit in R. (Uniqueness in the quotient-remainder
theorem holds if R is an integral domain).

2
 One can define polynomials in several variable such as R[X, Y ] or R[X1 , . . . , Xn ].
If R = F is a field, these polynomial rings in several variables are UFDs, but
not PIDs in general.
Note that R[X, Y ] is isomorphic to R0 [Y ] where R0 = R[X]. In other words, the
ring R[X, Y ] is isomorphic to R[X][Y ]. The Hilbert basis theorem says that if R is
Noetherian, then so is R[X]. Thus, by induction, R[X1 , . . . , Xn ] is Noetherian as
well. Since a field F is trivially Noetherian, F [X1 , . . . , Xn ] is always Noetherian.
Let f ∈ R[X] be a polynomial. Let S be an extension ring of R. If a ∈ S,
then f (a) denotes the element of S obtained by substituting X with a in the
polynomial f . The rule
f 7→ f (a)
defines a function
σa : R[X] → S.
It turns out that this is a ring homomorphism. In other words, (f g)(a) = f (a)g(a)
and (f + g)(a) = f (a) + g(a). Why do these laws hold? It turns out that the
definitions of addition and multiplication for R[X] are exactly what is needed to
make these laws hold. We call σa a substitution homomorphism.
There is another homomorphism that we will need. Suppose we are given a ring
homomorphism π : R → S. Then there is a unique ring homomorphism

π̃ : R[X] → S[X]

such that (1) π̃ extends π, and (2) X maps to X. An important example of this is
when π : Z → Zn where Zn is the integers modulo n with n a positive integer. In
other words, Zn is the quotient ring Z/ hni. This result in a homomorphism

Z[X] → Zn [X]

which is the reduction modulo n map at the level of polynomials. For example
modulo 5 the polynomial 10X 3 + 13X 2 + 100X − 1 maps to 3X 2 + 4 ∈ F5 [X].
Another important background fact is that the polynomial ring F [X] where F
is a field has the property that every nonzero prime idea is maximal. This is a
consequence of F [X] being a PID. Also, an ideal is maximal if and only if its
generator is irreducible. Also note that every nonzero ideal has a unique monic
generator.1

2 The Ring R[a]

Above we considered R[X] where X is a symbolic variable. Now we extend this
notation to R[a] where a might not be a symbolic variable, but can be any element
in an extension of R.
Definition 2. Let a ∈ S where S is an extension ring of R. The image of the
substitution homomorphism σa : R[X] → S is written R[a]. In other words,

R[a] = f (a) | f ∈ R[X] .
1A polynomial is monic if its leading coefficient is 1.

3
 Since R[a] is the image of a ring homomorphism, it is a subring of S. The
ring R[a] is called the ring extension of R generated by a. This terminology is used
because any subring T of S containing R and a must contain R[a] as a subring.
In other words, R[a] is the unique minimal subring of S containing R, as a subset,
and containing the element a.
Remark. The above definition seems to depend on S, but it can be easily seen to be
independent of S in the following sense: if a ∈ S ⊆ S 0 where S in an extension of R
and S 0 is an extension of S, then R[a] is the same ring whether we use S or S 0 in
the definition. Likewise, the following definition is independent of S in this sense.

Definition 3. Let a ∈ S where S is an extension of R. The kernel of σa is called

the vanishing ideal of a over R. In other words, the vanishing ideal is the set Ia
where 
I = f ∈ R[X] | f (a) = 0 .
Since the kernel of any ring homomorphism is an ideal, Ia is indeed an ideal of the
ring R[X].

Informally the vanishing ideal of a is the collection of polynomials with root a.

Using the fundamental homomorphism theorem linking images and kernels of
homomorphisms, we get the following:

Proposition 2. Let a ∈ S where S is an extension of R. Then

R[a] ∼
= R[X]/Ia

where Ia is the vanishing ideal of a over R.

Proposition 3. Let a ∈ S where S is an extension of R. If S is an integral domain,

then the vanishing ideal Ia of a over R is a prime ideal of R[X].

Proof. Since R[a] is a subring of S, it is also an integral domain. We conclude that

the ideal Ia appearing in Proposition 2 must be a prime ideal since the quotient
ring is an integral domain (Theorem 1).

Many of these ideas can be generalized to more than one variable. For example,
if a, b ∈ S, where S is an extension of R, and if f ∈ R[X, Y ] then f (a, b) is defined
in the usual way. The map
f 7→ f (a, b)
defines a ring homomorphism
R[X, Y ] → S
and the image of this homomorphism is written R[a, b]. The kernel consists of all
polynomials in two variables vanishing at the poin (a, b) ∈ R2 .

3 Algebraic versus Transcendental

Now we apply the above ideas to the concept of algebraic and transcendental num-
bers.

4
Definition 4. Suppose α ∈ C. If there is a nonzero polynomial f ∈ Q[X] such
that f (α) = 0 then α is said to be algebraic. If α is not algebraic then it is said to
be transcendental.
√
For example, 2 is algebraic since it is a root of X 2 − 2. On the other hand,
according to famous theorems, π and e are transcendental.2
Observe that α ∈ C is algebraic if and only if the vanishing ideal of α in Q[X]
is nonzero. The vanishing ideal is a prime ideal by Proposition 3, and in Q[X] all
nonzero prime ideals are maximal. So we get the following:

Proposition 4. An element α ∈ C is algebraic if and only if the vanishing ideal

of α in Q[X] is maximal.

Corollary 5. An element α ∈ C is algebraic if and only if Q[α] is a field.

Proof. This follows from Proposition 2 together with the previous proposition. Re-
call that the quotient of a commutative ring by an ideal is a field if and only if the
ideal is maximal (Theorem 1).

Definition 5. Let α ∈ C be algebraic. Then Q[α] is called the field extension of Q

generated by α. We also write this fields as Q(α) to emphasize that it is a field.

Proposition 6. An element α ∈ C is transcendental if and only if Q[α] is isomor-

phic to the polynomial ring Q[X].

Proof. If α is transcendental then the vanishing ideal is the zero ideal. By Propo-
sition 2 we conclude that Q[α] is isomorphic to Q[X] (recall that the quotient of a
ring R by the zero ideal is isomorphic to R).
The other direction follows from the above corollary: if Q[α] is isomorphic
to Q[X] then it cannot be a field, and so α is not algebraic.

If α ∈ C is transcendental, then consider the set

 
def f (α)
Q(α) = : f, g ∈ Q[X] with g 6= 0
g(α)

It is easy enough to check that Q(α) is a subfield of C, and to prove the following.
Proposition 7. If α ∈ C is transcendental then the field Q(α) is the minimal
subfield of C containing α.

Definition 6. Let α ∈ C be transcendental. Then Q(α) is called the field extension

of Q generated by α.

Suppose that α ∈ C is algebraic. Then its vanishing ideal is generated by a

nonzero polynomial f ∈ Q[X]. We can normalize so that f is monic (leading
coefficient 1). Such f is called the minimal polynomial of α. We will discuss this
further in the next section in a more general setting.
2 In 1873, the French mathematician Hermite showed that e is transcendental. A few years

later, in 1882, the German mathematician Lindemann showed that π is transcendental using the
techniques of Hermite.

5
 With some thought, one sees that the set of all algebraic numbers forms a
countable subfield of C. On the other hand, since R is uncountable, C is also
uncountable. Thus there are an uncountable number of transcendental numbers:
there are more transcendental numbers than algebraic numbers.3

4 Minimal Polynomials
Much of what we discussed in the previous section generalizes from base field Q to
a general base field F .
Definition 7. Let F be a field, and let α ∈ E where E is an extension ring of F .
Let I ⊆ F [X] be the vanishing ideal of α in F [X]. If I is not the zero ideal, then we
say that α is algebraic over F . Otherwise, we say that α is transcendental over F .
Recall that F [X] is a PID, so the vanishing ideal in the above definition is a
principal ideal. If the vanishing ideal is not zero and if we require that the generator
be monic then the generator is unique:
Definition 8. Let F be a field, and let α ∈ E be algebraic over F , where E is an
extension ring of F . Then the minimal polynomial of α over F is the unique monic
generator of the vanishing ideal of α over F .
We can rephrase the definition as follows:
Proposition 8. Let F be a field, and let α ∈ E be algebraic over F , where E is
an extension ring of F . The minimal polynomial of α over F is the unique monic
polynomial f ∈ F [X] with the following property: g(α) = 0 if and only if g is a
multiple of f in F [X].
We are most interested in the case where E is a field (or integral domain)
extending F . In this case, the vanishing ideal must be a prime ideal (Proposition 3).
Because of this, Proposition 4 and Corollary 5 generalize easily with essentially the
same proofs.
Proposition 9. Let E be a field extension of F . An element α ∈ E is algebraic
over F if and only if the vanishing ideal of α in F [X] is maximal.
Corollary 10. Let E be a field extension of F . An element α ∈ E is algebraic
over F if and only if F [α] is a field.
We view F [α] as the field extension of F generated by α:
Proposition 11. Let α ∈ E be algebraic over F where E is a field extension of
the field F . Then F [α] is the minimal extension of F containing α in the following
sense: If K ⊆ E is a field extension of F containing α then F [α] ⊆ K.
Definition 9. Let F be a field, and let α ∈ E where E is a field extension of F .
Then F [α] is called the field extension of F generated by α. In this case, we also
write F [α] as F (α) to indicate that it is a field.
3 This is the basis of Cantor’s proof of the existence of transcendental numbers in the 1870s.

His proof was thought to be miraculous at the time, and suspicious to some, since it shows
transcendental numbers exist without constructing any particular transcendental number.

6
 Since the vanishing ideal is a prime ideal we must have the following:
Proposition 12. Let F be a field, and let α ∈ E be algebraic over F , where E is
an extension field of F . Then the minimal polynomial of α over F is irreducible
in F [X].
We can strengthen this, providing an alternative characterization of a minimal
polynomial:
Proposition 13. Let F be a field, and let α ∈ E be algebraic over F where E is
an extension field of F . Then the minimal polynomial of α over F is the unique
polynomial f ∈ F [X] that is (1) monic, (2) irreducible, and (3) satisfies f (α) = 0.
Proof. From previous results and definitions it follows that the minimal polynomial
satisfies (1), (2), and (3). Conversely, suppose f ∈ F [X] satisfies (1), (2), (3). By
property (3) f is in the vanishing ideal, so is a multiple of the minimal polynomial.
By (1) and (2) this means that f is the minimal polynomial.
Using Proposition 2, and the definition of minimal polynomial, we obtain the
following:
Proposition 14. Let α ∈ E be algebraic over F where E is a field extension of
the field F . Let f be the minimal polynomial of α over F . Then
F [α] ∼
= F [X]/ hf i .
Finally we mention the generalization of Proposition 6 (with essentially the same
proof).
Proposition 15. Let F be a field, and let α ∈ E where E is an extension field
of F . Then α is transcendental over F if and only if F [α] is isomorphic to the
polynomial ring F [X].
Exercise 1. Let F be a field with extension field E. Suppose that f ∈ F [X] is
monic of degree two or three, and has no roots in F . Show that f is the minimal
polynomial of α over F for all roots α of f in E.
√
Exercise 2. Show that the minimal polynomial of 2 is X 2 − 2. (Here F = Q).
Exercise 3. Show that C is R[i]. Show that X 2 + 1 is the minimal polynomial of i
over R. Conclude that C is isomorphic to R[X]/ X 2 + 1 . (Note: one could even
define C to be the field R[X]/ X 2 + 1 , and define i to be the coset of X).

5 Degrees of Field Extensions

Recall that a field extension of a field F is simply a field containing F as a subring.
For example, C is an extension field of R. We can also think of a field extension E
of F as an F -vector space. To do so, view the addition of E as the vector space
addition. Think of the product cx with c ∈ F and x ∈ E as the scalar multiplication.
(For now, we conveniently “forget” about multiplication cx when c is in E but not
in F ). We easily check that E is a vector space over F by checking all the axioms
of a vector space.

7
Proposition 16. If E is an extension field of the field F , then E is a vector space
over F (using the operations described above).
Recall that every vector space has a cardinal assigned to it called the dimension.
It can be defined as the cardinality of a basis.
Definition 10. Let E be a field extension of F . Then the dimension of E, thought
of as a vector space over F , is called the degree of the field extension E over F .
This degree is written [E : F ]. If [E : F ] is finite, then we say that E is a finite
extension of F .
In the case of E = F [α] where α is algebraic over F , the degree [E : F ] turns
out to be the degree of the minimal polynomial of α over F . Showing this is the
theme of the next two exercises.
Exercise 4. Suppose α is algebraic over F with minimal polynomial f of degree d.
Show that 1, α, α2 , . . . , αd−1 span the vector space F [α].
Hint: given g(α) ∈ E, use quotients and remainders in F [X] to write g = qf + r
where q, r ∈ F [X]. Now substitute α for X.
Exercise 5. Suppose α is algebraic over F with minimal polynomial f of de-
gree d. Show that 1, α, α2 , . . . , αd−1 are linearly independent vectors in the vector
space F [α]. Hint: from a linear combination form a polynomial.
From these exercises we get the following basic facts:
Proposition 17. Let E = F [α] be an extension of F generated by α, where α is
algebraic over F . Let f ∈ F [X] be the minimal polynomial of α with degree d. Then

1, α, α2 , . . . , αd−1

is a basis of the vector space E.

Proposition 18. Let E = F [α] be an extension of F generated by α, where α is
algebraic over F . Let f ∈ F [X] be the minimal polynomial of α. Then
 
F [α] : F = deg f.

Exercise 6. Show Q[i] has basis 1, i over Q. In other words, show that

Q[i] = {a + bi | a, b ∈ Q}.

Exercise 7. Suppose [E : Fp ] = d where Fp is the field with p elements. Show that

the field E has pd elements. Hint: write a general element of E in terms of a basis.
How many possibilities are there?

6 Constructing Extension Fields

If F is a subfield of C and if f ∈ F [X] has positive degree, then the equation

f (α) = 0

8
has a solution with α ∈ C; this is just the fundamental theorem of algebra. In this
case you can do computations involving α by working in C. In fact, the field F [α]
is a subfield of C.
What if F is not a subfield of C? Or what if you want to do computations
in F [α] but do not want to work in the complex numbers? Perhaps you do not
want to identify α with a specific complex number. This section will provide a
way of constructing roots and field extensions using a generalization of modular
arithmetic, namely quotient rings.

Exercise 8. Show that a finite field cannot occur as a subfield of C. So you cannot
form field extensions of a finite field simply by taking subfields of C.
In an earlier exercise, you were asked to show that C is isomorphic to

R[X]/ X 2 + 1 .

We observed that forming this quotient ring is actually a valid way to construct C
from the reals R. Our strategy is to generalize this construction.
So let F be a field, and let f ∈ F [X] be an irreducible polynomial of positive
degree. Then since f is irreducible and nonzero, we know that the ideal generated
by f is maximal. Thus
E = F [X]/ hf i
is a field.
Exercise 9. Let f ∈ F [X] be a nonconstant polynomial. Show that the canonical
map
F [X] → F [X]/ hf i
restricted to F is injective.
By the above exercise, we can think of F as a subfield of E. In other words, E
is an extension of F . Here we identify a constant c with it image c (the equivalence
class containing c).
For a polynomial g in F [X], we denote by g the image of g in F [X]/ hf i. In
other words, g is the coset (equivalence class) containing g. Given g, h ∈ F [X],
we say that g ≡ h modulo hf i if an only if g − h ∈ hf i. Note the following are
equivalent:

1. g ≡ h modulo hf i.
2. g − h is a multiple of f .
3. The images of g, h in E = F [X]/ hf i are equal: g = h.
This allows us to do computations in E using rules of modular arithmetic, applied
to polynomials.
Recall that, in Z/ hmi, we have m ≡ 0 modulo m by basic modular arithmetic.
Something similar happens in E: since f − 0 is trivially a multiple of f we get
that f ≡ 0 modulo hf i. In other words, in the field E = F [X]/ hf i we have the
equation f = 0. This means that f is in the kernel of the canonical map.

9
 The image of X in E = F [X]/ hf i plays a special role in this situation. Define
the element α to be X. By this definition, the canonical map

F [X] → F [X]/ hf i

sends X to its image α.

Suppose that g(X) ∈ F [X] is given by

g(X) = an X n + . . . + a1 X + a0 .

Since the canonical map is a homomorphism, we have

g(X) = an (X)n + . . . + a1 X + a0 = an αn + . . . + a1 α + a0 = g(α).

Here we used the identification of a with a for all a ∈ F . So g(X) maps to g(α).
This means that the canonical map is just the substitution homomorphism:
Proposition 19. Let F be a field, and let f ∈ F [X] be an irreducible polynomial
of positive degree. Let α be the image of X under the canonical map

F [X] → F [X]/ hf i .

Then the canonical map is the substitution homomorphism sending X to α. In

other words, any polynomial g(X) ∈ F [X] has image g(α).
Because the canonical map is the substitution homomorphisn, we get the fol-
lowing (as in Section 2):
Corollary 20. Let F be a field, and let f ∈ F [X] be an irreducible polynomial of
positive degree. Let α be the image of X under the canonical map. Then

F [X]/ hf i = F [α]

Remark. Note the similarity between Proposition 14 and the above corollary. One
difference is that Proposition 14 describes an isomorphism, but here we have a true
equality.

Proof of corollary. By the previous proposition (Proposition 19), the canonical map
is the substitution homomorphism. By Definition 2 we have that F [α] is the image
of this map. However, the canonical map

F [X] → F [X]/ hf i .

is surjective. So F [α] is all of F [X]/ hf i.

Proposition 21. Let F be a field and let f ∈ F [X] be an irreducible monic poly-
nomial of positive degree. Let α be the image of X under the canonical map

F [X] → F [X]/ hf i = F [α]

Then α is a root of f :
f (α) = 0.

10
Proof. As observed earlier, f is in the kernel of the canonical map, and so f maps
to 0. But the canonical map is the substitution homomorphism and maps f (X)
to f (α). Thus f (α) = 0.

Given an irreducible polynomial f of positive degree, we used a quotient ring

to create an extension F [α] where α is a root of f . If, in addition, f is monic, we
will have that f is exactly the minimal polynomial of this element α. We assert
this fact, and summarize the situation as a whole, in the following:

Theorem 22. Let E be the field F [X]/ hf i where f is an irreducible monic poly-
nomial of positive degree. Let α = X. In other words, α is the coset in E contain-
ing X. Then α is algebraic over F with minimal polynomial f and

E = F [α].

Proof. This is largely just a summary of previous results. To complete the proof
we just need to show that f is the minimal polynomial of α. By assumption, f
is irreducible and monic. By Proposition 21, f (α) = 0. Thus f is the minimal
polynomial of α by Proposition 13.

Corollary 23. Every irreducible monic polynomial f ∈ F [X] of positive degree is

the minimal polynomial for some α in some field extension E.

Exercise 10. Use the polynomial X 2 + 1 ∈ F3 [X] to form a field extension of F3 .

Show that this field contains 9 elements. Make addition and multiplication tables
for this field (where α is X).

Exercise 11. Use the polynomial X 3 + X + 1 ∈ F2 [X] to form a field with 8

elements. Make addition and multiplication tables for this field.

Exercise 12. Construct a field with four elements. Make addition and multiplica-
tion tables for this field.

7 Primitive Polynomials
The ring Z[X] is not a PID. In fact, the ideal h2, Xi is not a principal ideal. However,
the integral domain Z[X] is nonetheless a UFD, and finding irreducible factors
in Z[X] is very closely related to finding irreducible factors in the ring Q[X] which
is a PID. To explain and justify this connection requires primitive polynomials.
First we point out that being irreducible in Z[X] is different than being irre-
ducible in Q[X]. For example, 2X 2 +2 is considered to be irreducible in Q[X] since 2
is a unit in Q[x], but reducible in Z[X] (neither 2 nor X 2 + 1 is a unit in Z[X]). On
the other hand, 2 is irreducible in Z[X], but is not irreducible in Q[X] (it is just
a unit). However, there is a big overlap between irreducible elements in Z[X] and
irreducible elements in Q[X]; for example, X 2 + 1 is irreducible in both. In fact, we
will see that examples such as 2X 2 + 2 that involve constants are really the only
cases where they differ. Primitive polynomials in Z[X] are polynomials with such
constants factored out.

11
Definition 11. A nonzero polynomial f in Z[X] is said to be primitive if there is no
prime p that divides all the coefficients of f . In other words, primitive polynomials
are polynomials whose coefficients have GCD equal to 1.

For example, the polynomials X 2 + 1 and −12X 2 + 15X − 50 are primitive,

but 2X 2 + 2 and −12X 2 + 15X − 30 are not. Monic polynomials in Z[X] are always
primitive.
The notion of primitive polynomial and many of the following results concerning
such polynomials generalize to R[X] where R is a PID. However, the following
definition is particular to Z since it involves positivity.
Definition 12. A polynomial f in Z[X] is said to be positive primitive if it is
primitive and its leading coeficient is positive.

As discussed in Section 1, we can associate to any prime number p a homomor-

phism Z[X] → Fp [X] given by sending a polynomial f with coefficients in Z to the
same polynomial except with the coefficients considered to be in Fp . For example,
the image of 3X 3 + 4X 2 + 5X + 6 under the map Z[X] → F3 [X] is X 2 + 2X.
Exercise 13. Show that a nonzero polynomial f in Z[X] is primitive if and only
if its image under Z[X] → Fp [X] is nonzero for each prime p.

Theorem 24 (Gauss’s lemma, first form). The set of primitive polynomials in Z[X]
is closed under multiplication. The set of positive primitive polynomials is also
closed under multiplication.

Proof. Suppose f and g are primitive. Let p be a prime, and consider the canonical
map Z[X] → Fp [X] where h 7→ h. Then f g = f g since the map is a homomorphism
of rings. Since f and g are primitive f and g are nonzero. Since Fp [X] is an integral
domain, the product is nonzero as well. So f g is nonzero. This is true for every
prime p, so f g is primitive.
If f and g are positive primitive as well, then f g must have positive leading
coefficient since the leading coefficient of f g is the product of the leading coefficients
of f and g. So f g is positive primitive.

Exercise 14. Show that the set of primitive polynomials in Z[X] is not closed
under addition.

Lemma 25. Every nonzero f ∈ Z[X] can be factored as f = cf˜ where c ∈ Z is

nonzero and f˜ is a positive primitive polynomial.

Proof. Factor out the greatest common divisor of the coefficients. Factor out −1
if necessary to force the leading coefficient to be positive. (Another argument: if
there is a prime p dividing all the coefficients, factor it out. Repeat until there are
no primes left dividing all the coefficients).

Later we will see that the c and f˜ above are unique.

Lemma 26. Every nonzero f ∈ Q[X] can be factored as f = cf˜ where c ∈ Q×

and f˜ ∈ Z[X] is a positive primitive polynomial.

12
Proof. Factor out 1/b where b is a common denominator of the coefficients. What
is left is a polynomial in Z[X]. Factor this polynomial as af˜ where f˜ is positive
primitive (Lemma 25). Observe that f = cf˜ where c = a/b.

To get uniqueness we need the following lemma.

Lemma 27. Suppose f is a positive primitive polynomial. If c ∈ Q is such that cf
is also positive primitive, then c = 1. In other words, if two positive primitive
polynomials are associates in Q[X] then they are equal.

Proof. Suppose f and cf are positive primitive. Since the leading coefficients are
positive, c ∈ Q is positive. Write c = a/b where a and b are relatively prime positive
integers.
Suppose p is a prime dividing b, and write b = pb0 for some positive integer b0 .
Since f is primitive, it has at least one coefficient ui ∈ Z not divisible by p. The
ith coefficient of cf is aui /b0 p. Since p does not divide ui and does not divide a,
it cannot divide the product aui . Thus the ith coefficient of cf is not an integer.
In other words, cf is not even in Z[X], contradicting the definition of primitive
polynomial. We conclude that no such p divides b. This implies that b = 1, so c is
a positive integer.
If c ∈ Z is divisible by a prime p, then every coefficient of cf is divisible by p as
well. So cf would not be primitive, a contradiction. So no such p divides c. This
implies that c = 1.

Proposition 28. Every nonzero f ∈ Q[X] can be factored uniquely as f = cf˜

where c ∈ Q× and f˜ ∈ Z[X] is a positive primitive polynomial.

Proof. Existence follows from Lemma 26. Suppose f = cf˜ = dg̃ where c, d ∈ Q×
and f˜ and g̃ are positive primitive. Then d−1 cf˜ = g̃. Thus f˜ and cd−1 f˜ are both
positive primitive. By Lemma 27, this means cd−1 = 1, in other words c = d.
Thus cf˜ = cg̃, which implies f˜ = g̃.

Corollary 29. Every nonzero f ∈ Z[X] can be factored uniquely as f = cf˜

where c ∈ Z is nonzero and f˜ ∈ Z[X] is a positive primitive polynomial.

Proof. Existence follows from Lemma 25. Uniqueness is a corollary of Proposi-

tion 28

Exercise 15. Suppose that f, g ∈ Z[X] and that f is a primitive polynomial. Show
that f divides g in Z[X] if and only if f divides g in Q[X].

8 Factorization into Primitive Polynomials

Primitive polynomials live both in Z[X] and Q[X] since Z[X] is a subring of Q[X].
It turns out that for primitive polynomials there is no difference between being
irreducible in Z[X] and in Q[X]. Before proving this, we prove a lemma.
Lemma 30. Suppose f ∈ Z[X] is a nonconstant positive primitive polynomial. If f
is reducible in Q[X] then we can write f as the product of two nonconstant positive
primitive polynomials in Z[X].

13
Proof. Since f is reducible, we can write f = gh where g, h ∈ Q[X] are nonconstant.
By Proposition 28 we can write g as cg̃ and h as dh̃ where c, d ∈ Q and g̃, h̃ ∈ Z[X]
are positive primitive polynomials. Thus f = cdg̃ h̃. Since g̃ h̃ is positive primitive
(Theorem 24), Lemma 27 implies that cd = 1. So f = g̃ h̃.

Theorem 31. Let f ∈ Z[X] be a nonconstant positive primitive polynomial.

Then f is irreducible in Z[X] if and only if f is irreducible in Q[X].

Proof. Suppose f is irreducible in Z[X]. We wish to show f is irreducible in Q[X].

Suppose otherwise. Then by Lemma 30 we can factor f in terms of nonconstant
polynomials of Z[X]. So f is reducible in Z[X], a contradiction.
Now suppose f is irreducible in Q[X]. We wish to show f is irreducible in Z[X].
We do so by supposing f = gh where g, h ∈ Z[X], and showing g or h is a unit
in Z. Since f is irreducible in Q[X], either g or h is a constant. Without loss of
generality, suppose g = c is a constant. Note that c ∈ Z since g ∈ Z[X]. If a
prime p divides c, then each coefficient of f = ch is divisible by p, a contradiction
to the assumption that f is primitive. Thus c must be ±1 as desired.

Corollary 32 (Gauss’s lemma, second form). Let f ∈ Z[X] be a primitive polyno-

mial. Then f is irreducible in Z[X] if and only if f is irreducible in Q[X].

Proof. Replace f with −f if necessary, and then use the above theorem. (Also
note that the only constant primitive polynomials are the units ±1, and so are not
irreducible in Z[X] or Q[X].)

Proposition 33. Every nonconstant polynomial f ∈ Q[X] is the product of a

constant c ∈ Q× times one or more irreducible positive primitive polynomials:

f = cf1 · · · fk .

Moreover, this product is unique up to rearrangement of factors fi .

Proof. We know that Q[X] has unique factorization, so we can factor f into irre-
ducibles in Q[X]:
f = ag1 · · · gk .
By Proposition 28, gi = bi fi where bi ∈ Q× and fi is a positive primitive polyno-
mial. Since fi is an associate of gi and since gi is irreducible, the polynomial fi is
irreducible in Q[X] (and hence in Z[X]). Thus

f = (ab1 · · · bk )f1 · · · fk ,

and so the existence claim is established.

To establish uniqueness, suppose

f = cf1 · · · fk = c0 f10 · · · fl0

where c, c0 ∈ Q× and fi , fi0 are irreducible positive primitive polynomials (irre-

ducible in both Q[X] and Z[X] by the above theorem). Since Q[X] is a UFD, we
can conclude that k = l and that, after rearranging terms, fi ∼ fi0 . By Lemma 27,
the only way that positive primitive polynomials can be associates is if they are
equal. So fi = fi0 . After cancelling all the fi , we conclude c = c0 .

14
Proposition 34. Every nonconstant positive primitive polynomial f ∈ Z[X] is the
product of irreducible positive primitive polynomials:

f = f1 · · · fk .

Moreover, this product is unique (up to rearrangement of terms).

Proof. By Proposition 33, in Q[X] we can write f uniquely as cf1 · · · fk where c ∈ Q

and each fi is a positive primitive polynomial. Let g = f1 · · · fk . By Gauss’s lemma
(Theorem 24) g is a positive primitive polynomial. Since f = cg, Lemma 27 implies
that c = 1, and so f = f1 . . . fk .
Uniqueness follows from Proposition 33.

Theorem 35. Every nonzero polynomial f ∈ Z[X] is the product of ±1 times zero
or more prime numbers times zero or more irreducible positive primitive polynomi-
als:
f = (±1)p1 · · · pk f1 · · · fl .
Moreover, this product is unique up to rearrangement of the sequence of factors (pi )
and (fi ).

Proof. By Corollary 29, we can write f as cf˜ where c ∈ Z is nonzero and f˜ is

positive primitive. We factor c in Z and factor f˜ using Proposition 34. This gives
the existence of the desired factorization.
Suppose we have another such factorization:

f = (±1)p01 · · · p0k0 f10 · · · fl00

Let f 0 = f10 · · · fl00 . By By Gauss’s lemma (Theorem 24) f 0 is positive primitive.

Let c0 = (±1)p01 · · · p0k0 , so f = cf˜ = c0 f 0 . By Corollary 29, c = c0 and f˜ = f 0 . By
unique factorization in Z and by the uniqueness assertion of Proposition 34, the
desired uniqueness claim follows.

The above theorem implies that Z[X] is a UFD. To see this, observe that prime
numbers and positive primitive irreducible polynomials are irreducible in Z[X]. For
uniqueness one needs also know that these are the only irreducible elements (up to
unit):

Exercise 16. Show that every irreducible element of Z[X] is either ±p where p is
a prime number, or ±f where f is an irreducible positive primitive polynomial.

Corollary 36. The integral domain Z[X] is a UFD.

When f ∈ Z[X] is monic, it is automatically primitive positive. We consider a
few nice results that apply to monic polynomials.

Exercise 17. Suppose f ∈ Z[X] is monic. Show that f factors as the product of
monic irreducible polynomials in Z[X].

Proposition 37. Let f ∈ Z[X] be monic, and let g ∈ Q[X] be a monic polynomial
that divides f in Q[X]. Then g ∈ Z[X] and g divides f in Z[X].

15
Proof. Start with f = gh in Q[X]. Since f and g are monic, the same is true of h.
Write g = cg̃ and h = dh̃ where c, d ∈ Q× and where g̃, h̃ are positive primitive
polynomials (Proposition 28). So f = (cd)(g0 h0 ). By Gauss’s lemma (Theorem 24),
the product g0 h0 is positive primitive. So cd = 1 by Lemma 27.
Since g is monic, we see that c−1 is the leading coefficient of g̃. So c−1 is a
positive integer. Similarly, d−1 is a positive integer. We have cd = 1 so c−1 d−1 = 1.
Thus c−1 = d−1 = 1. The result follows.

Exercise 18. Suppose α ∈ E where E is an extension field of Q. Suppose α is a

root of a monic polynomial in Z[X]. Show that the minimal polynomial of α must
also be in Z[X].
Exercise 19. Show that if a polynomial f ∈ Z[X] is monic, all its rational roots
are actually integers. Hint: focus on linear factors.
Exercise 20. Recall that a perfect square is an integer of the form m2 where m ∈ Z.
Suppose n ∈ Z is not a perfect square. Use the polynomial X 2 − n to show that the
square root of n is irrational, and that X 2 − n is its minimal polynonial. Generalize
to cube roots.

Exercise 21. Show that if a polynomial f ∈ Z[X] has a rational root a/b (written
in lowest terms with b > 0) then b divides the leading coefficient of f , and a
divides the constant term of f . Hint: when you factor f in Q[X] you have a linear
term X − a/b. When you factor f into a constant times the product of positive
primitive polynomials, this linear factor gets expressed as bX − a.

9 The Mod p irreducibility Test

The following gives a way to show that a given monic polynomial in Z[X] is irre-
ducible. To make it work you choose a suitable prime p and reduce the polynomial
modulo p.

Proposition 38. Suppose f ∈ Z[X] is a nonconstant monic polynomial and p is

a prime number. If f maps to an irreducible polynomial under the canonical ring
homomorphism Z[X] → Fp [X] then f itself is irreducible (in both Q[X] and Z[X]).

Proof. We suppose f is reducible and derive a contradiction. Since f is monic, it

must be positive primitive. By Lemma 30, we can factor f as gh where g and h are
nonconstant positive primitive polynomials.
Note that g and h are monic since the product of the leading coefficients is 1.
Thus the images g and h are nonconstant in Fp [X]. In Fp [X] we have f = gh.
Thus f is a reducible polynomial, a contradiction.

If f is primitive, but not monic, the above theorem fails as stated, but can be
repaired as follows:

Proposition 39. Suppose f ∈ Z[X] is a primitive polynomial and p is a prime

number not dividing the initial coefficient of f . If f maps to an irreducible polyno-
mial under the canonical homomorphism Z[X] → Fp [X] then f is irreducible.

16
Exercise 22. Prove the above theorem.
Exercise 23. Show that the assumption in the above proposition (p does not divide
the initial coefficient) is necessary by giving a reducible primitive polynomial whose
image modulo p for some prime p is irreducible.
Exercise 24. Show that X 2 + 3aX + (3b + 1) is irreducible for all integers a and b
by reducing modulo 3.
Exercise 25. Show that X 4 + X + 1 is not divisible by any quadratic polynomial
in F2 [X]. Hint: show that the constant term of any quadratic divisor must be 1.
Now conclude that there are only two possible divisors, and show both fail.
Exercise 26 (Continued). Show that X 4 + X + 1 is irreducible in F2 [X].

Exercise 27 (Continued). Let f = a4 X 4 + a3 X 3 + a2 X 2 + a1 X + a0 be a primitive

polynomial where a4 , a1 , and a0 are odd, and a3 and a2 are even. Show that f is
irreducible in Q[X].

10 The Eisenstein Criterion

There is a famous theorem that proves a large number of polynomials are irre-
ducible. (In what follows, recall that all integers divide 0).

Theorem 40. Consider a nonconstant polynomial in Z[X]

f = an X n + ak−1 X k−1 + . . . + a1 + a0

such that there is a prime p that does not divide an but divides ai for each i < n.
Suppose also that p2 does not divide a0 . Then f is irreducible in the ring Q[X].

Proof. By repeatedly factoring out any prime dividing all the ai , and by changing
sign if necessary, we can assume that f is a positive primitive polynomial. (Note
that p will not be such a prime since it does not divide the leading coefficient).
Suppose f is reducible. Then by Lemma 30 we can write f = gh where g, h ∈ Z[X]
are nonconstant primitive polynomials. Let b be the leading coefficient of g and
let c be the leading coefficient of h. Note that p cannot divide b or c. Modulo p,
the polynomial f is just an X n , so

an X n = g h.

By unique factorization in Fp [X] we conclude that g = bX k and h = cX l where k

and l are positive with n = k + l. Since k, l > 0, p divides the constant term
of g and the constant term of h. So p2 divides the constant term of gh = f , a
contradiction.

Exercise 28. Show that X 5 + 6X 4 − 12X 2 + 6X + 6X is irreducible.

17
11 Cyclotomic Polynomials
For any n > 2 let ζn be the complex number e2πi/n . Using the basic properties of
the complex valued exponential function we have

ζn = cos(2π/n) + i sin(2π/n).

Note that
n
n
ζn = e2πi/n = e2πi = 1,
so ζn is called a nth root of unity, and is a root of X n − 1. Note also that if k < n
then
k
k
ζn = e2πi/n = e2πki/n 6= 1.
In other words, ζn has multiplicative order exactly equal to n. Consequently, ζn is
called a primitive nth root of unity.
√
Exercise 29. Show that ζ4 = i and ζ6 = 1/2 + i 3/2. What is ζ8 ?
Definition 13. The minimal polynomial of ζn is called the nth cyclotomic poly-
nomial.
Exercise 30. Show that the third cyclotomic polynomial is X 2 + X + 1 and the
fourth cyclotomic polynomial is X 2 + 1. Hint, factor X n − 1 where n = 3, 4.
Cyclotomic polynomials have several nice properties, and their study was devel-
oped by Gauss. There is a famous theorem that the degree of the nth cyclotomic
polynomial is equal to ϕ(n) where ϕ is the Euler phi function. We won’t prove
this here except for the case where n is a prime. The above exercise gives two
examples of cyclotomic polynomials that are quadratic polynomials. There is one
other since ϕ(6) = 2.
Exercise 31. Show that X 2 − X + 1 divides X 3 + 1. Show that X 2 − X + 1 is the
sixth cyclotomic polynomial.
TO HERE
Theorem 41. Let p be a prime. Then

f = X p−1 + X p−2 + . . . + X + 1

is irreducible in Q[X].

Proof. Observe that f is positive primitive. Suppose that f is reducible. So by

Lemma 30, f = gh where g and h are nonconstant positive primitive polynomials.
In fact, g and h are monic since the product of their leading coefficients is 1. Note
that (X − 1)f = X p − 1. Thus

(X − 1)gh = X p − 1.

Now in Fp [X] we have that (X − 1)p = X p − 1. This follows from the binomial
theorem: the binomial coeffients, except the first and last, are divisible by p. So
in Fp [X] we have
(X − 1) g h = X p − 1 = (X − 1)p .

18
By unique factorization in Fp [X] (and the fact that g, h are monic) we have

g = (X − 1)k , h = (X − 1)l

where k + l = p − 1 and k, l are positive. This implies that 1 is a root of g and a

root of h. In particular, g(1) and h(1) are congruent to zero modulo p. In other
words p divides g(1) and h(1). Since f = gh, we have p2 divides f (1).
However,
f (1) = 1p−1 + 1p−2 + . . . 11 + 1 = p
so p2 does not divide f (1), a contradiction.

Corollary 42. If p is a prime, then X p−1 +X p−2 +. . .+X +1 is the pth cyclotomic
polynomial.
Exercise 32. Justify the above corollary.

Definition 14. The field Q [ζn ] is called the nth cyclotomic extension.

Exercise 33. Justify the following.

Corollary 43. Let E = Q [ζp ] where p is a prime. Then [E : Q] = p − 1. Further-
more, 1, ζ p , . . . , ζ p−2 is a basis for E over Q.
Exercise 34. Show that the third cyclotomic field and the sixth cyclotomic field
are equal.

Exercise 35. Show that if n is odd, the nth cyclotomic field and the 2nth cyclo-
tomic field are equal.

19