CHAPTER 1

Polynomials

!c W W L Chen, 1984, 2013.

This chapter originates from material used by the author
at Imperial College London between 1981 and 1990.
It is available free to all individuals,
on the understanding that it is not to be used for financial gain,
and may be downloaded and/or photocopied,
with or without permission from the author.
However, this document may not be kept on any information storage and retrieval system
without permission from the author,
unless such system is not accessible to any individuals other than its owners.

1.1. Divisibility
In this chapter, we very briefly consider some properties of polynomials which we need in the study
of algebraic number fields. Suppose that R is a commutative ring with multiplicative identity 1. We
denote by R[t] the collection of all polynomials in the variable t and with coefficients in R. It is well
known that R[t] itself is a commutative ring with multiplicative identity 1. This ring is called the
ring of polynomials over R.
Suppose that a(t), b(t) ∈ R[t] and a(t) is not identically zero. Then we say that a(t) divides b(t),
denoted by a(t) | b(t), if there exists c(t) ∈ R[t] such that b(t) = a(t)c(t); in other words, we have a
factorization b(t) = a(t)c(t). In this case, we say that a(t) is a divisor of b(t).
Recall that divisibility in Z is governed by the magnitude of the integers. In the case of polynomials,
this role is now played by the degree of the polynomials. Suppose that the polynomial a(t) ∈ R[t]
is given by a(t) = an tn + . . . + a0 , where a0 , . . . , an ∈ R and an "= 0. Then we say that a(t) is of
degree n, denoted by n = deg a(t).
Theorem 1.1. Suppose that a(t), b(t) ∈ K[t], where K is a field, and a(t) is not identically zero.
Then there exist q(t), r(t) ∈ K[t] such that b(t) = a(t)q(t) + r(t), where either deg r(t) < deg a(t) or
r(t) = 0.
Proof. We first show the existence of such polynomials q(t), r(t) ∈ K[t]. Consider the set
S = {b(t) − a(t)s(t) : s(t) ∈ K[t]}.
Suppose first that 0 ∈ S. Then there exists q(t) ∈ K[t] such that b(t) − a(t)q(t) = 0. In this case, we
have r(t) = b(t) − a(t)q(t) = 0. On the other hand, if 0 "∈ S, then the degrees of the polynomials in
the non-empty set S forms a non-empty subset of the set of all non-negative integers. It then follows
from the Principle of induction that there exists a polynomial r(t) ∈ S of smallest degree, m say. Let
q(t) ∈ K[t] such that b(t) − a(t)q(t) = r(t). Then deg r(t) < deg a(t), for otherwise, writing
a(t) = an tn + . . . + a0 and r(t) = rm tm + . . . + r0 ,
where an "= 0, rm "= 0 and m ! n, we have
! "
r(t) − (rm a−1
n t
m−n
)a(t) = b(t) − a(t) q(t) + rm a−1
n t
m−n
∈ K[t].
Clearly deg(r(t) − (rm a−1
n t
m−n
)a(t)) < deg r(t), contradicting the minimality of m.
Next, we show that such polynomials q(t), r(t) ∈ K[t] are unique. Suppose that
b(t) = a(t)q1 (t) + r1 (t) = a(t)q2 (t) + r2 (t).
Then
r1 (t) − r2 (t) = a(t)(q2 (t) − q1 (t)).
If q1 (t) "= q2 (t), then clearly deg(a(t)(q2 (t) − q1 (t))) ! deg a(t), while deg(r1 (t) − r2 (t)) < deg a(t), a
contradiction. It follows that q1 (t) = q2 (t), and so r1 (t) = r2 (t) also. $
1
2 1. POLYNOMIALS

We next establish the existence of greatest common divisors.

Theorem 1.2. Suppose that K is a field, and that a(t), b(t) ∈ K[t] are not identically zero. Then
there exists g(t) ∈ K[t], unique up to multiplication by non-zero elements of K, such that
(i) there exist polynomials u(t), v(t) ∈ K[t] such that g(t) = a(t)u(t) + b(t)v(t);
(ii) g(t) | a(t) and g(t) | b(t); and
(iii) for every polynomial h(t) ∈ K[t] such that h(t) | a(t) and h(t) | b(t), we have h(t) | g(t).
Proof. Consider the set
I = {a(t)x(t) + b(t)y(t) : x(t), y(t) ∈ K[t]}.
Suppose that g(t) ∈ I is not identically zero and is of smallest degree. The conclusion (i) follows
trivially.
Next, we show that g(t) divides every polynomial in I. Suppose that
z(t) = a(t)x(t) + b(t)y(t)
is any polynomial in I. By Theorem 1.1, there exist q(t), r(t) ∈ K[t] such that z(t) = g(t)q(t) + r(t),
where either deg r(t) < deg g(t) or r(t) = 0. Then
r(t) = z(t) − g(t)q(t) = a(t)(x(t) − u(t)q(t)) + b(t)(y(t) − v(t)q(t)) ∈ I.
If r(t) "= 0, then the requirement deg r(t) < deg g(t) contradicts the minimality of the degree of g(t).
Hence r(t) = 0, so that z(t) = g(t)q(t), whence g(t) divides z(t).
Taking x(t) = 1 and y(t) = 0 gives g(t) | a(t). Taking x(t) = 0 and y(t) = 1 gives g(t) | b(t). Also,
the conclusion (iii) is a simple consequence of (i).
Finally, note that if g1 (t) and g2 (t) both satisfy the requirements, then each must be a divisor of
the other, and therefore must be multiples of each other by non-zero elements in K. $
The polynomial g(t) ∈ K[t] in Theorem 1.2 is called a greatest common divisor of the polynomials
a(t) and b(t), and denoted by g(t) = (a(t), b(t)). Two polynomials a(t), b(t) ∈ K[t] which are not
identically zero are said to be relatively prime, or coprime, if (a(t), b(t)) ∈ K.
Remark. The set I in the proof of Theorem 1.2 forms a non-zero ideal in the polynomial ring
K[t]. The first part of our proof merely shows that any non-zero ideal I in a polynomial ring K[t],
where K is a field, is generated by a non-zero element of I.

1.2. Irreducibility
Suppose that R is a commutative ring with multiplicative identity 1. A polynomial u(t) ∈ R[t] is
said to be a unit in R[t] if u(t) divides the multiplicative identity 1 of R. A factorization b(t) = a(t)c(t)
is said to be proper if neither a(t) nor c(t) is a unit in R[t]. A non-constant polynomial b(t) ∈ R[t] is
said to be irreducible if b(t) does not have a proper factorization; in other words, if b(t) = a(t)c(t),
where a(t), c(t) ∈ R[t], then at least one of a(t) or c(t) is a unit.
Theorem 1.3. Suppose that K is a field, and that a(t), b(t) ∈ K[t]. Suppose further that p(t) ∈ K[t]
is irreducible. If p(t) | a(t)b(t), then p(t) | a(t) or p(t) | b(t).
Proof. Suppose that p(t) ! a(t). Since p(t) is irreducible, the only divisors of p(t) are non-zero
elements of K or polynomials of the form cp(t), where c ∈ K is non-zero. Clearly cp(t) ! a(t) for any
non-zero c ∈ K. Hence we must have (a(t), p(t)) = 1. It follows from Theorem 1.2 that there exist
u(t), v(t) ∈ K[t] such that
1 = a(t)u(t) + p(t)v(t), so that b(t) = a(t)b(t)u(t) + p(t)b(t)v(t).
Clearly p(t) | b(t). $
We next restrict our attention to the special case when the field K consists of complex numbers.
Theorem 1.4. Suppose that K ⊆ C is a field. Then every irreducible polynomial in K[t] has no
repeated roots in C.
The proof depends on the following observation.
 1.3. POLYNOMIALS WITH RATIONAL COEFFICIENTS 3

Theorem 1.5. Suppose that K ⊆ C is a field, and the polynomial a(t) ∈ K[t] is not identically
zero. Then a(t) is divisible by the square of a polynomial of positive degree in K[t] if and only if a(t)
and a# (t) have a common factor of positive degree in K[t].
Proof. Suppose first of all that
a(t) = b2 (t)c(t),
where b(t), c(t) ∈ K[t] and deg b(t) > 0. Differentiating formally, we obtain
a# (t) = 2b(t)b# (t) + b2 (t)c# (t),
so that the polynomials a(t) and a# (t) have b(t) as a common factor.
Suppose now that a(t) is not divisible by the square of any polynomial in K[t] of positive degree.
Then for any irreducible factor b(t) of a(t), we have a(t) = b(t)c(t), where b(t) and c(t) are coprime.
Differentiating formally, we have
a# (t) = b# (t)c(t) + b(t)c# (t).
Suppose on the contrary that a(t) and a# (t) have b(t) as a common factor. Then we must have
b(t) | b# (t)c(t), and so it follows from Theorem 1.3 that b(t) | b# (t). This is only possible if deg b(t) = 0,
so that b(t) is a constant polynomial. $
Proof of Theorem 1.4. Suppose that a(t) ∈ K[t] is irreducible in K[t]. Then a(t) is not divisible
by the square of any polynomial of positive degree in K[t], and so it follows from Theorem 1.5 that
a(t) and a# (t) are coprime. By Theorem 1.2, there exist u(t), v(t) ∈ K[t] such that
a(t)u(t) + a# (t)v(t) = 1.
Clearly a(t), a# (t), u(t), v(t) ∈ C[t]. This implies that a(t) and a# (t) are coprime over C[t]. It now
follows from Theorem 1.5 that a(t) is not divisible by the square of any polynomial of positive degree
in C[t], and so cannot have repeated zeros in C. $

1.3. Polynomials with Rational Coefficients

In this section, we study factorization and irreducibility properties in the polynomial ring Q[t].
Our first result shows that if a polynomial with rational integer coefficients can be factorized, then it
is possible to write the factors as polynomials with rational integer coefficients.
Theorem 1.6 (Gauss’s lemma). Suppose that a(t) ∈ Z[t]. Suppose further that a(t) = b(t)c(t),
where b(t), c(t) ∈ Q[t]. Then there exists a non-zero λ ∈ Q such that λb(t), λ−1 c(t) ∈ Z[t].
Proof. The equation a(t) = b(t)c(t) can be written in the form
(1.1) na(t) = b∗ (t)c∗ (t),
where n ∈ Z is non-zero, and where the polynomials b∗ (t), c∗ (t) ∈ Z[t] are rational multiples of
b(t), c(t) ∈ Q[t] respectively. Write
b∗ (t) = br tr + . . . + b0 and c∗ (t) = cs ts + . . . + c0 ,
where b0 , . . . , br , c0 . . . , cs ∈ Z, with br "= 0 and cs "= 0. Clearly the integer n divides all the coefficients
of b∗ (t)c∗ (t). Let p be a prime factor of n.
We first show that either p | bi for every i = 0, . . . , r or p | cj for every j = 0, . . . , s. Suppose on
the contrary that this is not the case. Then there exists m satisfying 0 " m " r such that p ! bm and
p | bi , i = 0, . . . , m − 1.
Similarly, there exists q satisfying 0 " q " s such that p ! cq and
p | cj , j = 0, . . . , q − 1.
Then the coefficient of t m+q
in b (t)c (t) is equal to
∗ ∗

b0 cm+q + . . . + bm cq + . . . + bm+q c0 ,
with the convention that this sum only includes those terms for which the coefficients exist. Clearly
every term in this expression apart from bm cq is divisible by p. It follows that this coefficient is not
divisible by p, a contradiction.
We now remove the prime factor p from the expression (1.1) in a suitable way, and repeat this
argument with prime factors of n/p, and so on. $
4 1. POLYNOMIALS

Our next result gives a sufficient condition for irreducibility in Q[t].

Theorem 1.7 (Eisenstein’s criterion). Suppose that a(t) ∈ Z[t], and a(t) = an tn + . . . + a0 , where
a0 , . . . , an ∈ Z. Suppose further that there exists a prime number p such that
(i) p ! an ;
(ii) p | ai for every i = 0, . . . , n − 1; and
(iii) p2 ! a0 .
Then a(t) is irreducible in Q[t].
Proof. Suppose on the contrary that a(t) is not irreducuble in Q[t]. Then there exist non-
constant polynomials b(t), c(t) ∈ Q[t] such that a(t) = b(t)c(t). In view of Gauss’s lemma, we can
further assume that b(t), c(t) ∈ Z[t]. Write
b(t) = br tr + . . . + b0 and c(t) = cs ts + . . . + c0 ,
where b0 , . . . , br , c0 . . . , cs ∈ Z, with br "= 0, cs "= 0, r + s = n and r, s ! 1. Note that a0 = b0 c0 . It
follows from (ii) and (iii) that p divides exactly one of b0 and c0 . Without loss of generality, suppose
that p | b0 and p ! c0 . It now follows from (i) that there exists m satisfying 1 " m " r such that
p ! bm and
p | bi , i = 1, . . . , m − 1.
Then the coefficient of t m
in a(t) = b(t)c(t) is equal to
am = b0 cm + . . . + bm c0 ,
where m " r < n and with the convention that this sum only includes those terms for which the
coefficients exist. Clearly p ! am , contradicting (ii). $

1.4. Symmetric Polynomials

In this last section, we consider polynomials of more than one variable. Suppose that R is a
commutative ring with multiplicative identity 1. We denote by R[t1 , . . . , tn ] the collection of all
polynomials in the variables t1 , . . . , tn and with coefficients in R. As before, R[t1 , . . . , tn ] itself is a
commutative ring with multiplicative identity 1.
A polynomial a(t1 , . . . , tn ) ∈ R[t1 , . . . , tn ] is said to be symmetric if
a(t1 , . . . , tn ) = a(tσ(1) , . . . , tσ(n) )
for all permutations σ on the set {1, . . . , n}. In particular, the elementary symmetric polynomials
are given by
s1 (t1 , . . . , tn ) = t1 + t2 + . . . + tn ,
s2 (t1 , . . . , tn ) = t1 t2 + t1 t3 + . . . + t1 tn + t2 t3 + . . . + tn−1 tn ,
..
.
sn (t1 , . . . , tn ) = t1 . . . tn .

Example. Consider a polynomial f (t) ∈ K[t], where K ⊆ C is a field. Resolving f (t) into linear
factors over C, we have
f (t) = a(t − α1 ) . . . (t − αn ) = a(tn − s1 tn−1 + . . . + (−1)n sn ),
where si = si (α1 , . . . , αn ) for every i = 1, . . . , n.
Elementary symmetric polynomials are of particular interest as a consequence of the following
important result.
Theorem 1.8. Let R be a commutative ring with multiplicative identity 1. Then every symmetric
polynomial in R[t1 , . . . , tn ] can be expressed as a polynomial in R[s1 , . . . , sn ], where si = si (t1 , . . . , tn )
for every i = 1, . . . , n.
 1.4. SYMMETRIC POLYNOMIALS 5

Proof. We first define the following “lexicographic” order on the monomials tβ1 1 . . . tβnn by saying
that tβ1 1 . . . tβnn precedes tγ11 . . . tγnn if the first non-vanishing βi − γi , where i = 1, . . . , n, is positive.
For every symmetric polynomial a(t1 , . . . , tn ) in R[t1 , . . . , tn ], we order its terms lexicographically. If
ctβ1 1 . . . tβnn is one of the terms of a(t1 , . . . , tn ), then there are similar monomials with the exponents
β1 , . . . , βn permuted. It is then not too difficult to see that the leading term in lexicographic order
of a(t1 , . . . , tn ) is one of the form ctβ1 1 . . . tβnn , where β1 ! . . . ! βn . On the other hand, the leading
term of
sk11 . . . sknn = (t1 + . . . + tn )k1 . . . (t1 . . . tn )kn
is
tk11 +...+kn tk22 +...+kn . . . tknn .
It follows that choosing k1 = β1 − β2 , . . . , kn−1 = βn−1 − βn and kn = βn , we conclude that the
leading term of
β −βn βn
(1.2) a(t1 , . . . , tn ) − csβ1 1 −β2 . . . sn−1
n−1
sn
in lexicographic order comes after ctβ1 1 . . . tβnn . There are only finitely many monomials in lexicographic
order in a(t1 , . . . , tn ). Repeating our argument on (1.2) and so on, noting that (1.2) is symmetric, we
can clearly reduce a(t1 , . . . , tn ) to a polynomial in R[s1 , . . . , sn ]. $