Download as pdf or txt
Download as pdf or txt
You are on page 1of 41

February-March 2009

Polynomials
by Ivan Borsenco and Iurie Boreico

This AMY Segment1 is devoted to Polynomials. These are, without any doubt, the
most important functions in mathematics. Several properties of polynomials lie at
the core of fundamental results in modern algebra and number theory. And let’s not
forget that algebraic geometry is, by definition, the study of simultaneous solutions
of polynomial equations. Polynomial functions find applications in combinatorics,
computer science, and other important areas of science. They are common in math
contests, as well. So whether you want to be a successful competitor, or want to
build a solid base for college math, this segment is for you, and I dare declare
that everything you find here is equally important both for math contests and for
advanced mathematics. So enjoy this tutorial, and may your road to understanding
it be soft and pleasant!

• Introduction
• Factorization of polynomials
• Polynomials in many variables
• Irreducible polynomials
• Cyclotomic Polynomials
• Lagrange Interpolation Formula

The assigned problems are arranged in increasing order of difficulty. I encourage


you to spend time to think about the given problems and send us your solutions,
partial or complete. I wish you good luck and success in your challenges!

The author

1

c 2008-2009, AwesomeMath LLC. All Rights Reserved.
1 Introduction

Definitions

Since ancient times people have studied relations between objects. They intro-
duced notions of distance, velocity, acceleration and many more. They measured
each of those quantities, and tried to establish relationships among them. Eventually
they came up with the notion of ”function”. In such a way people could investigate
quantities as a function of time, space, or any other parameters. As mathematics
developed, they felt the need to introduce the concept of ”variable”, and express
functions in term of variables. Multiplying variables between them, and adding
the obtained results produced a class of functions which are known as polynomials.
Let’s start with the simplest yet richest class of polynomials, the polynomials in one
variable.

Definition 1.1. A function of a single variable t is a polynomial on its domain


if we can put in the form

an tn + an−1 tn−1 + · · · + a1 x + a0 ,

where a0 , a1 , . . . , an are constants. The set of all polynomials of a variable X with


coefficients in a given set S is denoted by S[X].
The definition says that every polynomial can be expressed as a finite sum of a
monomial terms of the form ak tk in which the variable is raised to a nonnegative
integer power. We use convention that t0 = 1, so a0 t0 = a0 . To begin with, we will
look at polynomials for which the constants ai are real or complex numbers.

Definition 1.2. For the polynomial

an tn + an−1 tn−1 + · · · + a1 t + a0 , an 6= 0,

the numbers ai , 0 ≤ i ≤ n, are called coefficients. The coefficient an is called the


leading coefficient, and an tn is the leading term. a0 is the constant term or the
constant coefficient. a1 is the linear coefficient and a1 t is the linear term. When
the leading coefficient an is 1, the polynomial is said to be monic. Two polynomials
are equal if and only if the coefficients corresponding to each degree are equal.The
terms ., tn−1 are called monomials.

Definition 1.3. The nonnegative integer n is the degree of the polynomial, we


write deg p = n. A constant polynomial has degree 0. Special names are given to
polynomials of low degree:
• deg p = 1 linear • deg p = 3 cubic

2
• deg p = 2 quadratic • deg p = 4 quartic

We can evaluate a polynomial by replacing the variable by any number and


carrying out the computation. The value of a polynomials p(t) at t = r is denoted
by p(r). For example, the value of p(t) = 2t2 − t + 3 at t = 1 is p(1) = 4.

Definition 1.4. A zero of a polynomial p(t) is any number r for which p(r) takes
the value 0. When p(r) = 0, we say that r is a root or a solution of the equation
p(t) = 0. There are many situations in which we need to have information about the
zeros of a polynomial, and considerable amount of attention is devoted to methods
of solving equations p(t) = 0 either exactly or approximately. In particular, knowing
the zeros of polynomials is often helpful in graphing a wide variety of functions and
obtaining inequalities.

In operating with polynomials, we treat the variables as though they were num-
bers. Let
p(t) = a0 + a1 t + a2 t2 + · · · + an tn
q(t) = b0 + b1 t + b2 t2 + · · · + bm tm
be any two polynomials.

p(t) ± q(t) = (a0 ± b0 ) + (a1 ± b1 )t + (a2 ± b2 )t2 + · · ·

p(t)q(t) = a0 b0 + (a0 b1 + a1 b0 )t + (a0 b2 + a1 b1 + a2 b0 )t2 + · · ·


+ (a0 br + a1 br−1 + · · · + ai br−i + · · · + · · · + ar b0 )tr + · · ·
+ an bm tm+n .

We also define composition of two polynomials: (p ◦ q)(t) = p(q(t)). This definition


instructs us to replace each occurrence of t in the expression for p(t) by q(t).

Quadratic polynomials

Many issues that arise in general can be illustrated in the special case for which
the degree is 2. Polynomials of degree two are largely studied in the high school.
We review some of the results about quadratics.

Let p(t) = at2 + bt + c be quadratic polynomial, then

b 2
 
1
p(t) = a t + − (b2 − 4ac).
2a 4a

3
b 2 b2 −4ac

Hence p(t) = 0 whenever t + 2a = 4a2
.

The discriminant D is the expression b2 − 4ac. Here we have three possible


scenarios:

1st case: D = b2 − 4ac > 0. Now we can take square root on the right hand side.
Therefore we find two different real solutions
√ √
−b b2 − 4ac −b b2 − 4ac
t1 = + and t2 = − .
2a 2a 2a 2a
Shortly we write them √
−b ± D
t1,2 = .
2a

b 2
2nd case: D = b2 − 4ac = 0. The right hand side is equal to 0, so t +

2a = 0,
b
yielding a double root t1 = t2 = − 2a .

3rd case: D = b2 − 4ac < 0. This case is strange because we cannot extract
square root. It is time to enter the world of complex numbers. Let us introduce
a number i such that i2 = −1. Do not worry about what i means; all we need to
know is that
√ its square √ is −1. Now life gets easier, we can take the square root of
−3: it is 3i. Indeed, ( 3i)2 = 3i2 = −3. Thus for the quadratic equation we get
two complex roots:
√ √
−b 4ac − b2 −b 4ac − b2
t1 = + i and t2 = − i.
2a 2a 2a 2a
In general, a complex number is one which can be written in the form z = x + iy,
where x + yi are real and i2 = −1. In discussing complex numbers we use the
following terminology:
• The real part, Re z, of z is the number x.
• The imaginary part, Im z, of z is the number y.
• The complex conjugate, z, of z is x − yi
p
• The modulus or absolute value of z, denoted by |z| is x2 + y 2 .

Note that in the third case we get two solutions that are complex conjugates. To
summarize the theory, remember, quadratic polynomials have two zeroes: if D ≥ 0
they are real and if D < 0 they are complex conjugates. If D = 0 then the zeroes
are equal. In this case we get a zero with multiplicity 2.

One more thing that you also should keep in mind it is graphical representation
of quadratic equations. Let us consider three graphs corresponding to three cases
above:
p1 (x) = −x2 + 4, p2 (x) = (x + 1)2 , p3 (x) = x2 + 1.

4
If you draw a two dimensional graph with (x, y) = (x, p(x)), x ∈ R, you will see a
parabola. We get an upward parabola if a > 0 and downward parabola if a < 0.
Hence p2 (x), p3 (x) are upward and p1 (x) is a downward parabola.
Roots that we eagerly look for are intercepts of the parabola with x-axis. So
p1 (x) intersects x-axis in points (−2, 0) and (2, 0). p2 (x) is tangent to x-axis at
(−1, 0). But p3 (x) which does not have real roots will never intersect it.
Sometimes we are interested in the minimum (for upward parabola) or maximum
(for downward parabola) value that quadratic polynomial can return. This value is
 b2 −4ac
achieved at x = − 2a b
which is called apex of parabola and p −b
2a = 4a2 is the the
minimum (maximum) value. The y-intercept is obtained at x = 0 and is p(0) = c.

Let us assume p(x) = ax2 + bx + c has two roots x1 , x2 . It does not matter if
they are real or complex numbers. Then p(x) can be rewritten as

p(x) = a(x − x1 )(x − x2 ) = ax2 − a(x1 + x2 )x + ax1 x2 .

Because two polynomials are equal when corresponding coefficients are equal we
have
b c
x1 + x2 = − , x1 x2 = .
a a
This relation is called Viete Theorem for quadratic polynomials. Note that knowing
the coefficients of a polynomial equation we can immediately find the sum and the
product of its roots.

Basically, all these facts are enough to start solving simple and challenging ques-
tions.

Example 1.1. Is it true that if p(x) = ax2 + bx + c has real roots, then
q(x) = a3 x2 + b3 x + c3 has also real roots?

Solution. The answer is yes. Because p(x) = ax2 + bx + c has real roots we have
b2 ≥ 4ac. Note that q(x) = a3 x2 + b3 x + c3 has real roots whenever its discriminant
b6 − 4a3 c3 ≥ 0. We have b6 ≥ 64a3 c3 , so if ac ≥ 0, then b6 ≥ 64a3 c3 ≥ 4a3 c3 . If
ac < 0, then again b6 − 4a3 c3 ≥ 0.

Example 1.2. Let a, b, c be integers, c an odd integer, and p(x) = ax2 + bx + c


be a quadratic polynomial with integer solutions. Can p(2009) be an odd number?

Solution. The answer is no. Using Viete’s Theorem if x1 and x2 are integer
solutions then x1 x2 a = c. Because c is odd, it follows that a, x1 , x2 must be odd
too. Thus −b = (x1 + x2 )a is even, so p(2009) = a(2009)2 + b(2009) + c is even.

Example 1.3. We are given second degree equations

y(x) = x2 + a1 x + b1 , y(x) = x2 + a2 x + b2 , . . . , y(x) = x2 + an x + bn

5
with roots (x0 , x1 ), (x0 , x2 ), . . ., (x0 , xn ). Find in terms of x0 , x1 , . . . , xn the roots
of the polynomial
   
2 a1 + a2 + · · · + an b1 + b2 + · · · + bn
y(x) = x + x+ .
n n

Solution. Adding up these relations we see that the equation

nx2 + (a1 + a2 + · · · + an )x + (b1 + b2 + · · · + bn )


 
a1 +a2 +···+an
has root x0 . It follows that x0 is a root of y(x) = x2 + x+ b1 +b2 +···+b
 n
n n .
Another solution x∗ must satisfy x0 + x∗ = − a1 +a2 +···+a
n
n
. Using Viete’s Theorem
we have
x0 + x1 = −a1 , x0 + x2 = −a2 , . . . , x0 + xn = −an .
x1 +x2 +···+xn
It follows that x0 + n = − a1 +a2 +···+a
n
n
. Thus x∗ = x1 +x2 +···+xn
n .

6
Assigned problems

1. Is it true that if ax2 + bx + c has real roots, then P (x) = a3 x2 + b3 x + c3 has


real roots, too?

2. Consider quadratic polynomials p(x) = x2 + px + q such that p + 2q = 2009.


Prove that graphs of these functions pass through a common point.

3. Consider quadratic polynomials p(x) = x2 + px + q such that p + q = 30. How


many of them have two integer roots?

4. Let D be the discriminant of the quadratic polynomial f (x) = x2 + ax + b.


How many roots can have the quadratic polynomial

f (x) + f (x + D) = 0?

5. Find all pairs (a, b) such that equations x2 + ax + b2 = 0 and x2 + bx + a2 = 0


have at least one root in common.

6. Consider polynomial (x − a)(x − c) + 2(x − b)(x − d) = 0, where a < b < c < d.


Prove that it has real roots.

7. Consider the polynomial (p2 − a2 )x2 − 2pqx + q 2 − b2 = 0 with real roots x1


and x2 . Prove that (a2 x21 + b2 )(a2 x22 + b2 ) is a perfect square.

8. A second degree equation x2 + ax + b has two roots with absolute value greater
than 2. Prove that a + b + 1 is composite.

9. Consider quadratic polynomials with real coefficients:

x2 + mx + n = 0, x2 + px + q = 0.

Prove that if mp = 2(n + q) then one of them has real roots.

10. Let f1 , f2 be two quadratic polynomials which have real roots, while their
difference f1 − f2 does not. Prove that f1 + f2 has real roots.

11. Let p(x) = ax2 + bx + c be quadratic polynomial which does not have real
roots and a + b + c > 0. What can you tell about sign of a − b + c?

12. Consider a polynomial p(x) = x2 + ax − b, such that 0 < b < a and a, b ∈ R.


Prove that it cannot have both integer roots.

13. Let a, b, c be real numbers such that |ax2 + bx + c| ≤ 1 for each x ∈ [−1, 1].
Prove that |cx2 + bx + a| ≤ 2 for each x ∈ [−1, 1].

7
14. Let a, b, c be real numbers such that a 6= 0. It is known that for each x ∈ [0, 1]
we have |ax2 + bx + c| ≤ 1. Find the greatest possible value of |a| + |b| + |c|.
15. The roots of the equation x2 + ax + b + 1 = 0 are positive integers. Prove that
a2 + b2 is a composite number.

Polynomials of degree n

The next step for us to consider a general polynomial of degree n and explore
its properties. From the Fundamental Theorem of Algebra, which we will meet
a few times throughout the segment, we can deduce that polynomial of degree n
has exactly n complex zeros(with multiplicity). The following theorem resembles
divisibility of integer numbers. We will discuss them formally later.

Theorem 1.1. (Division Algorithm)


If f (x) and g(x) are polynomials and g(x) is not the zero polynomial, there exist
unique polynomials q(x) and r(x) such that
f (x) = q(x)g(x) + r(x),
where either r(x) = 0 or deg r(x) < deg g(x).

Theorem 1.2. (Remainder and Factor Theorems)


Let f (x) be a polynomial. When f (x) is divided by (x − a), the remainder is
f (a). Thus x = a is a root of the equation f (x) = 0 if and only if (x − a) is a factor
of f (x).

Proof. Let f (x) = an xn + an−1 xn−1 + · · · + a1 x + a0 , then


f (x) − f (a) = an (xn − an ) + an−1 (xn−1 − an−1 ) + · · · + a1 (x − a).
Since xi −ai = (x−a)(xi−1 +xi−2 a+· · ·+xai−2 +ai−1 ), it follows that (x−a) | xi −ai
for all i ∈ N. Hence (x − a) | f (x) − f (a) and f (a) is the remainder of f (x) divided
by (x − a).

Recall that a is the root of f (x) when f (a) = 0, so (x−a) | f (x). If (x−a)k | f (x)
where k ≥ 2, then a is called a multiple root of f (x).

Theorem 1.3. A non-zero polynomial of degree n has at most n roots.

Proof. Let a1 , a2 , . . . , ak be roots of polynomial f (x). Since a1 is a root we


can write f (x) = (x − a1 )g(x), where g(x) has degree n − 1. Note that f (a2 ) =
(a2 − a1 )g(a2 ) = 0, so a2 is a root of g(x). Continuing in this way we get
f (x) = (x − a1 )(x − a2 ) · · · (x − ak )h(x).

8
Clearly, deg f = deg k + deg h, so k ≤ n.

Remark. The only problem we run into is the occurrence of multiple roots. In
general, every polynomial has n roots, some of which can be multiple. Thus we can
write
f (x) = a(x − x1 )(x − x2 ) · · · (x − an ),
where a is the leading term and ai ’s are not necessarily distinct roots.

Theorem 1.4. (Uniqueness Theorem)


If p(x) and q(x) are polynomials, each of degree at most n, and p(xi ) = q(xi )
for i = 1, 2, . . . , m, where x1 , x2 , . . . , xm are distinct complex numbers and m > n,
then p(x) and q(x) are identical.

Proof. Let h(x) = p(x) − q(x), h is a polynomial of degree at most n. Note that
h(xi ) = 0, so it has at least m roots. But m > n, a contradiction. Hence h(x) = 0
and p(x) and q(x) are identical.

Here are some examples to understand why we need and how we use these
theorems.

Example 1.4. Find a cubic polynomial f such that f (0) = 0 and f (1) = f (2) =
f (3) = 1.

Proof. Let g(x) = f (x) − 1, then deg g(x) = 3 and x = 1, 2, 3 are roots of g.
Therefore g(x) = a(x − 1)(x − 2)(x − 3), where a ∈ R. So f (x) = a(x − 1)(x − 2)(x −
3) + 1 and plugging x = 0 we get f (0) = a · (−6) + 1 = 0. Hence a = 61 and
1
f (x) = (x − 1)(x − 2)(x − 3) + 1.
6

Example 1.5. Find the remainder when xn is divided by x2 − x − 1.

Proof. Using Division Algorithm we find that

xn = q(x)(x2 − x − 1) + r(x),

where r(x) is a polynomial of degree



at most 1.

The next step is to explore the
2 5+1 5−1
roots of x − x − 1, they are a = 2 and b = 2 . Plugging them into equation
we get an = r(a) and bn = r(b).
n n
Let r(x) = ux + v, then an = ua + v and bn = ub + v, so u = a √−b 5
and
n−1 n−1
v = a √−b5
. We can recognize that the result can be expressed neatly in terms
of Fibonacci numbers r(x) = Fn x + Fn−1 . It is also worth noting that

q(x) = F1 xn−2 + F2 xn−2 + · · · + Fn−1 .

9
Theorem 1.5. (Viete Theorem)
Let f (x) = an xn + an−1 xn−1 + · · · + a1 x + a0 be a polynomial of degree n. Let
x1 , x2 , . . . , xn be zeros of f , then
n
X an−1
xi = x1 + x2 + · · · + xn = − ,
an
i=1
X an−2
xi xj = x1 x2 + x2 x3 + x1 x3 + · · · =
an
i6=j
···
a0
x1 x2 · · · xn = (−1)n .
an

Proof. Write f (x) = an (x − x1 )(x − x2 ) · · · (x − xn ). Opening the brackets we


get

f (x) = an xn − an (x1 + x2 + · · · + xn )xn−1 + · · · + an (−1)n x1 x2 · · · xn .

Because it is the same polynomial as f (x) = an xn + an−1 xn−1 + · · · + a1 x + a0 , the


corresponding coefficients are equal.

Example 1.6. Let p(x) = xn + an−1 xn−1 + · · · + a1 x + 1 be a polynomial with


positive coefficients and n positive real roots. Prove that p(2) ≥ 3n .

Solution. Note that if t ≥ 0, then P (t) > 0. Hence all roots are negative real
numbers −x1 , −x2 , . . . , −xn . It follows that p(x) = (x + x1 )(x + x2 ) · · · (x + xn ).
Using Viete’s Theorem we get x1 x2 · · · xn = 1. Observe that p(2) = (2 + x1 )(2 +

x2 ) · · · (2 + xn ). From the AM-GM inequality: 2 + xi = 1 + 1 + xi ≥ 3 3 xi . Thus

p(2) = (2 + x1 )(2 + x2 ) · · · (2 + xn ) ≥ 3n 3 x1 x2 · · · xn = 3n ,

as desired.

Example 1.7. Let f (x) = an xn + an−1 xn−1 + · · · + a1 x + a0 be a polynomial


with integer coefficients. If u = pq , where gcd(p, q) = 1 and f (u) = 0, then q | an
and p | a0 .

Solution. Observe that


   n  n−1
p p p
f = an + an−1 + · · · + a1 x + a0 ,
q q q

so f (u) = 0 if and only if

an pn + an−1 pn−1 q + · · · + a1 pq n−1 q + a0 q n = 0.

10
Hence looking modulo p and q and using the fact that gcd(p, q) = 1, we get q | an
and p | a0 .

Example 1.8. Suppose all roots of the polynomial xn + an−1 xn−1 + · · · + a0 are
real. Then the roots are contained in the interval with endpoints
r
an−1 n − 1 2n
− ± a2n−1 − an−2 .
n n n−1

Solution. Let y be one of the roots and y1 , y2 , . . . , yn−1 the others. Then the
polynomial is (x − y)(x − y1 ) · · · (x − yn−1 ). Using Viete’s Theorem

−an−1 = y + y1 + · · · + yn−1 ,
X
an−2 = y(y1 + · · · + yn−1 ) + yi yj ,
i<j

and so X
a2n−1 − 2an−2 − y 2 = yi2 .
i=1

By Cauchy-Schwarz inequality applied to (y1 , . . . , yn−1 ) and (1,. . . ,1),

(an−1 + y)2 = (y1 + y2 + · · · + yn−1 )2


n−1
X
≤ (n − 1) yi2 = (n − 1)(a2n−1 − 2an−2 − y 2 ),
i=1

or
2an−1 2(n − 1) n−2 2
y2 + y+ an−2 − a ≤ 0.
n n n n−1
Thus y (and hence all yi ) lie between the two roots of the quadratic function, and
these roots are our bounds.

11
Assigned problems

16. Let p(x) be a quadratic polynomial with real coefficients. It is known that
when divided by (x − 1), (x − 2), (x − 3), it gives remainders −1, 2, 1, respec-
tively. Find p(x).
17. Find the remainder when x60 − 1 is divided by x3 − 2.
18. Find the remainder when x81 + x49 + x25 + x9 + x is divided by x3 − x.
19. Find the remainder when x2009 + x2008 + . . . + x2 + x + 1 is divided by x3 − 1.
20. Find all n such that xn+1 − xn + 1 is divisible by x2 − x + 1.
21. Let p(x) be polynomial with integer coefficients. Prove that a − b | p(a) − p(b),
where a, b ∈ Z.
22. Let a, b, c be distinct integers. Prove that there is no polynomial p(x) with
integer coefficients such that p(a) = b, p(b) = c, p(c) = a.
23. Let a, b, c be non-zero real numbers such that (a + b + c)(4a − 2b + c) < 0.
Prove that the equation ax2 + bx + c has real roots.
24. Let f (x) = ax3 + bx2 + cx + d be a polynomial with positive real roots. Prove
that c2 ≥ 3bd and b2 ≥ 3ac.
25. The zeros of the polynomial P (x) = x3 + x2 + ax + b are all real and negative.
Prove that 4a − 9b ≤ 1.
26. The polynomials P (x) = x2 + 2 and Q(x) = x + 1 are written on the board.
At any moment we can write the sum, the difference, the product of any
two polynomials already written on the board. Can we reach the polynomial
R(x) = x3 + 2?
27. Let f (x), g(x) be nonconstant polynomials with integer coefficients such that
g(x) divides f (x). Prove that if the polynomial f (x) − 2008 has at least 81
distinct integer roots, then the degree of g(x) is greater than 5.
28. Find all polynomials p(x) with real coefficients such that if a, b, c are real
numbers and ab + bc + ca = 0, then P (a − b) + P (b − c) + P (c − a) = 0.
29. Let α be a root of the polynomial P (x) = xn + an−1 xn−1 + √· · · + a1 x + a0 ,
where ai ∈ [0, 1] , for i = 0, 1, . . . , n − 1. Prove that Re α < 1+2 5 .
30. Find all polynomials P with real coefficients which satisfy the equality
P (a − b) + P (b − c) + P (c − a) = 2P (a + b + c)
for all triples a, b, c of real numbers such that ab + bc + ca = 0.

12
2 Factorization of polynomials

Here, we will concern ourselves with the divisibility theory for polynomials. This
theory is easier understood than defined, and to give a rigorous definition we will
need to go into some modern algebra. We won’t try to go to deep into it, but if the
material seems too complicated, try at least to understand the statements of the
theorems (although remembering the proofs would be useful, too). The aim of this
section is to show that polynomials resemble numbers very much, in the sense that
we have primes, unique factorizations, common divisors, and some other properties.
Yet they are, of course, different in other ways, so we need identify the common
properties of numbers and polynomials with respect to divisibility.
Since we have met polynomials in Z[X], R[X] and so on, we will unify these sets
into two categories:

Definition 2.1. A ring R is a triple (R, +, ·) consisting of a set R, together with


two operations on R (addition and multiplication), that satisfy the already familiar
properties of additivity and multiplication (commutativity, associativity, distribu-
tivity - we do not try to load you with a completely rigorous formal definition). The
set R must also contain the element 0 (the identity element for addition) and the
element 1 (the identity element for multiplication). Also, addition has an inverse
operation called subtraction and denoted by 0 −0 : −x is the unique element in R
such that x + (−x) = −x + x = 0.

Definition 2.2. A field F = (F, +, ·) is a ring in which the multiplication has


an inverse (called, of course, division): for every non-zero element x ∈ F there exists
a unique element x−1 = x1 such that x · x−1 = x−1 · x = 1.
In fact, the structure of a ring is directly copied from Z, and the structure of a
filed is copied from Q. You have to remember that in a ring one can add, subtract
and multiply but not divide while in a field one can add, subtract, multiply and
divide as well. Z is an example of a ring which is not a field, while R, Q, C are all
examples of fields (and rings, therefore). Z/pZ is a field, for prime integers p, while
if n is a composite integer then Z/nZ is not a field, even though it is a ring (check
this!).
In this setting, the polynomials over a ring R form a ring (prove it!), denoted by
R[X]. It is the ring obtained by adjoining to R an “outside” element X.
It turns out that in special cases one can make a field out of a ring, and these
cases are described below.

Definition 2.3. A ring R is called an integral domain if the product of two


non-zero elements if the ring is never zero: ab 6= 0 for a, b 6= 0.
Note that any field is an integral domain (prove it!).

13
Definition 2.4. If R is an integral domain, there exists a (unique) field F
containing R, called the ring of fractions of R. It consists of pairs (a, b) of elements
of R, written ab , which satisfy the following properties (what do you think they are?):
a −1
b 6= 0, ab = dc if and only if ad = bc, ab + dc = ad+bc a c ac
bd , b · d = bd and ( b ) = ab . R is
a
contained in the field via identifying a ∈ R with 1 .
The condition that R be an integral domain saves us from getting a zero denom-
inator bd when adding or multiplying two fractions. For example, Q is the field of
fractions of Z.
Now we go briefly to the divisibility theory for integral domains. We owe it
entirely to the divisibility theory for Z.

Definition 2.5. If a, b are numbers from an integral domain R, we say that


a divides b (and write a | b) if and only if there exists a number c ∈ R such that
ac = b.
Note that 1 divides all the elements of the ring.

Definition 2.6. In an integral domain R, a number x is called a unit if and


only if it has an inverse x−1 ∈ R. Equivalently (prove it!), we can say that x divides
1, or that x divides every element of the ring.
In Z, the units are 1 and -1. In a field, every element is a unit. For purposes of
divisibility, units are irrelevant. We say that two numbers a, b are associated, and
write a ∼ b, if the ratio of these two elements is a unit. It easy to check that ∼ is
an equivalence relation.

Definition 2.7. A number p in an integral domain R is called prime if it cannot


be written as the product of two numbers which are not units: if p = ab, then either
a or b is 1. Equivalently, if p is the product of two numbers, then one of them is a
unit and the other is associated with p.
In Z, the primes are just the usual prime integers, plus them multiplied by −1:
−2, −3, . . .. This is because we have to account for the negative integers in Z, too.

Definition 2.8. An integral domain R is a Unique Factorization Domain (UFD)


if every number in the ring can be decomposed into a product of primes, and this
decomposition is unique up to association.
For example, Z is a UFD. Note that, for example, 6 = 2 · 3 = (−2) · (−3) but this
is the same decomposition, as 2 and −2 are associated and 3 and −3 are associated,
too.

Theorem 2.9. R is a UFD if and only if every two non-zero elements a, b of


R have a greatest common divisor d, which is unique up to association. That is
d|a, d|b, and any other common divisor of a and b divides d.
Try to prove this theorem yourselves, by mimicking the proof that Z is a UFD.

14
Why do we need all these definitions? Here comes the first important theorem
of this section.

Theorem 2.10. If F is a field, then the ring of polynomials F [X] is a UFD.


First, it is easy to prove it is an integral domain: if f (x) = an xn +. . .+a0 , g(x) =
bm xm +. . .+b0 then (f g)(x) = an bm xm+n +(an bm−1 +an−1 bm )xm+n−1 +. . .+a0 b0 6= 0
because an bm 6= 0 (recall F is a field, so is an integral domain). In order to prove
that it is a UFD, we use theorem 2.9. So we want to show that every two polynomails
have a greatest common divisor, which we do by showing the Euclidean Algorithm
for Polynomials:

Take two polynomials f (x) = an xn + . . . + a0 , g(x) = bm xm + . . . + b0 and say


n > m. We can write f = gq + h by the Division algorithm. Now we replace f by
f − gq = h and we get a polynomial whose degree is smaller than the degree of f .
We then do the same with the pair (g, h) and so on. Since the total degree of the
polynomials decreases, eventually we will get to the pair (q, 0) and then q will be
the greatest common divisor of f and g, just like the usual Euclidean Algorithm for
integers gives the greatest common divisor of natural numbers. Also, just like in the
natural numbers case, we deduce that gcd(f, g) can be expressed as uf + vg where
u, v ∈ F [X].
But actually, a stronger version of this theorem is true: the condition for F being
a field can be weakened to being a UFD.

Theorem 2.11. If R is a UFD, then so is the ring of polynomials R[X].


To prove it, we will rely on the previous theorem. Since R is a subset of its
field of fractions F , R[X] is a subset of F [X] which we know is an integral domain.
Therefore every two polynomials f, g ∈ R[X] have a greatest common divisor q(x)
in F [X]. We only need to show that this common divisor is in R[X] (or one of its
associates is), and we will be done. But for this, we will need a definition and a
lemma.

Definition 2.12. If f ∈ R[X] is a non-zero polynomial, the greatest common


divisor of its coefficients is called the weight of f . f is called pure if this greatest
common divisor is 1.
Note that every polynomial f in F [X] is associated to a pure polynomial in
R[X]. Indeed, if a is the product of all denominators of the coefficients of f , then
af ∈ R[X], so if b is the weight of af then 1b af = ab f is a pure polynomial. Therefore
we may assume that q, the greatest common divisor of f and g in F [X], is a pure
polynomial from R[X], or else we can replace it with a polynomial associated to it
which is.
We thus have f, g, q three polynomials in R[X], and we know that q is the
greatest common divisor of f, g in F [X] and is also pure. We now prove it is in fact
their greatest common divisor in R[X].

15
Lemma 2.13 (Gauss’ Lemma) If f, q are polynomials in R[X], q is pure and
divides f in F [X], then q divides f in R[X], too.

Proof. Say f = qr, r ∈ F [X] and suppose that r is not in F [X]. Then not all
coefficients of r are in R. Say q(x) = an xn + . . . + a0 and r(x) = bm xm + . . . + b0 . It
means that there is some prime p which appears in the denominator of the coefficients
of r, when written in lowest terms. Say pt is the highest such power, and let bk be
the highest coefficient in whose denominator it appears. Since q is pure, not all its
coefficients are divisible by p, and let al be the highest coefficient of q not divisible
by p. Let’s look at z, the coefficient of xk+l in f = gh. It must be an element from
R which equals al bk + al−1 bk+1 + al+1 bk+1 + . . .. However note that the exponent of
p in the denominator al bk is pt , while in all other terms it is at most pt−1 . Indeed,
al+i bk−i cannot have pt in the denominator, because al+i is divisible by p and bk−i
has at most pt in its denominator, and al−i bk+i cannot have pt in the denominator,
because bk+i has at most pt−1 in its denominator. Therefore z is the sum of one
number that has pt in its denominator and several others that have at most pt−1 in
their denominators, and this is impossible.

We now return to the proof of theorem 2.11. Gauss’ lemma implies that q is a
common divisor of both f and g. Let u be the weight of f , v the weight of g, w the
greatest common divisor in R of u and v. Then wq divides f and g by the same
Gauss’ lemma. Now let s be another common divisor of f and g. Then s divides
q in F [X]. If we write s = zt where t is a pure polynomial and z is the weight of
s, then t divides q in R[X] according to Gauss’ lemma. Also z must divide both
the weight of f and the weight of g (prove it!) hence it must divide their common
divisor which is w. Then s = zt must divide wq, and therefore wq is the greatest
common divisor of f and g, and the proof is finished.

If you are lost by now in all this modern algebra, we advise you to relax, and try to
understand only the definitions and the statements of the theorems, without actually
bothering understanding the proofs. The most important here is to understand the
statement of theorem 2.11. The Euclidean Algorithm is not necessary, but useful to
know, and Gauss’ Lemma is a basic fact which you can skip if you really want to.

What does theorem 2.11. tell us? It does tell us that polynomials decompose
relatively nicely into smaller polynomials. In the ring of polynomials, primes are
called irreducible polynomials, and theorem 2.11 tells us that every polynomial in
R[X] decomposes uniquely into a product of irreducible polynomials (again, up to
units).

In order to get more information, we need to answer two questions: what are
the units and what are the irreducible polynomials in R[X]?

The first question is rather simple:

16
Theorem 2.14. The units in R[X] are precisely the units of R.

Proof. Clearly, the units of R are units in R[X]. Now let f be a unit in R[X],
so there exists g ∈ R[X] such that f g = 1. If f (x) = an xn + . . . + a0 , g(x) =
bm xm + . . . + b0 then (f g)(x) = an bm xn+m + . . . + a0 b0 . As an bm 6= 0, f (x)g(x) = 1
leads to the conclusion that m = n = 0 therefore f, g ∈ R and thus f is an element
of R which has an inverse, i.e. a unit.

Classifying irreducible polynomials is much harder. It is a problem in which the


answer depends heavily on R. We have the following particular cases:

Theorem 2.15. (The Fundamental Theorem of Algebra) The irreducibles


in C[X] are precisely the linear polynomials. In other words, every polynomial can
be decomposed as the product of linear polynomials, or every polynomial has a
complex root.

The proof of this theorem is beyond this course so we will accept it as a fact.

Theorem 2.16. The irreducibles in R[X] are the linear polynomials and the
quadratics with negative discriminant.

Proof. The linear polynomial are irreducible since if a polynomial of degree 1


is the product of two polynomials, then one of the actors must have degree 0 so be
a constant i.e. a unit. Similarly, a quadratic may be a product of non-units if and
only if it is the product of two linear polynomials i.e. it has two real roots, which
fails when the discriminant is negative. Now let’s prove all other polynomials are
reducible. If f is a polynomial of degree 2 with non-negative discriminant, then it
factors as the product of two linear polynomials. Otherwise f has degree at least 3.
It does have a complex root z by the Fundamental Theorem of Algebra. If z is real,
then x − z is a factor of f . Otherwise, z is also a factor of f , meaning that the real
polynomial (x − z)(x − z) = x2 − 2Rez + |z|2 is a factor of f which means f is not
irreducible.

The questions of irreducibility are more subtle over other rings or fields: for
example over Q there exist irreducible polynomials of any degree, a fact which will
be proved later. In fact, irreducibility over rationals and integers is interesting
enough that it deserves its own chapter in this tutorial.

We end with some simple examples:

Example 2.1. Let P ∈ R[X] and say z ∈ C satisfies P (z) = 0. Prove that
P (z) = 0, as well. [Note: we used it in the proof of theorem 2.16.]

Solution. Say P = an X n + an−1 X n + . . . + a1 X + a0 , Then an z n + an−1 z n−1 +


. . . + a1 z + a0 = 0. We conjugate this expression. Knowing that conjugation is

17
additive and multiplicative (if you don’t know that, it’s an easy exercise), and that
it preserves the real coefficients of the polynomial, we get that an (z)n +an−1 (z)n−1 +
. . . + a1 z + a0 = 0, as desired. Note that this means that the non-real roots of a real
polynomial group in pairs of conjugate ones.

Example 2.2. Find the greatest common divisor of x2 +11x+10 and x3 +7x+8.
Then, write it as (x2 + 11x + 10)f (x) + (x3 + 7x + 8)g(x)

Solution. We use the Euclidean Algorithm. x3 + 7x + 8 = (x2 + 11x + 10)(x −


11)+118(x+1).So we pass to the pair (x2 +11x+10, x+1) and we have x2 +11x+10 =
(x+1)(x+10). The greatest common divisor is x+1. To find the desired expression,
we work backwards:
1 1 1 x − 11 2
x+1 = ·118(x+1) = (x3 +7x+8−(x2 +11x+10)(x−11)) = (x3 +7x+8)− (x +11x+10)
118 118 118 118

Example 2.3. Does xm − x − 2 divide xn − 1 for some positive integers m, n?

Solution. No, it does not. If it did, then all its complex roots would be roots
of xn − 1. However the complex roots have absolute value 1. Yet, the product of
the roots of xm − x − 2 is ±2, whose absolute value is not 1, which means at least
of its roots is not of absolute value 1 so it is not a root of xn − 1.

18
Assigned problems

31. Prove that any polynomial of odd degree in R[X] has a real root.

32. Let f ∈ R[X] satisfy the inequalities f (a) < 0, f (b) > 0. Prove that f has a
root between a and b. [Hint: first prove this for irreducible polynomials]

33. If R is a field and f, g are coprime in R[X] then the Euclidean Algorithm tells
us that there exist some u, v ∈ R[X] such that uf + vg = 1. Is this true if we
only assume that R is a UFD so not a field (think of Z[X])? ”Correct” this
affirmation by replacing it with a slightly weaker, but similar, statement.

34. Let f (x) = x3 − 2x + 1 and g(x) = x2 + 2x − 5. Find polynomials u, v ∈ Z[X]


such that f (x)u(x) + g(x)v(x) = k where k 6= 0 is a constant. What’s the
smallest positive value of such a k?

35. Find the greatest common divisor of x2008 − 1 and x5 − x3 + x2 − 1.

36. Find a polynomial that gives remainder x+1 modulo x3 +22, and 2008 modulo
x4 + 1.

37. Let p be an irreducible polynomial in Z[X]. Show that the set

{gcd(p(t), p(t + 1)) | t ∈ Z}

is finite.

38. Let f : Q → Q satisfy

f (f (f (x))) + 2f (f (x)) + f (x) = 4x

and f (f (· · · f (x))) = x where f is taken 2009 times. Prove that f (x) = x.


[Hint: do you see some unusual polynomials here?]

39. What are the possible values for gcd(xm + x + 1, xn − 1) when m, n ∈ N?

40. Find all pairs of positive integers m, n ≥ 3 for which the polynomial xn +x2 −1
divides the polynomial xm + x − 1.

19
3 Polynomials in many variables

Up to now, we have considered polynomials as formal expressions or functions


in one variable. But there is no reason to limit ourselves to only one variable.

Definition 3.1. If R is a ring, then the set of all finite sums of terms cX1i1 X2i2 . . . Xnin ,
with c ∈ R, i1 , i2 , . . . , in ∈ N0 , is called the ring of polynomials in the variables
X1 , X2 , . . . , Xn and denoted by R[X1 , X2 , . . . , Xn ].

It is pretty clear that this is a ring. In fact, we have R[X1 , X2 , . . . , Xn ] = S[Xn ]


where S = R[X1 , X2 , . . . , Xn−1 ]. This shows that a polynomial in n variables can
be actually regarded as a polynomial in a single variable Xn , with coefficients in
R[X1 , X2 , . . . , Xn−1 ].

Theorem 3.2. If R is a UFD, then so is R[X1 , X2 , . . . , Xn ]. Particularly, any


polynomial in n variables can be factorized as a product of irreducible polynomials
in n variables, and this factorization is unique, up to units in R.

Proof. This is a simple induction on n. The case n = 1 is just theorem


2.11., and the step from n to n + 1 is done by noting that R[X1 , X2 , . . . , Xn+1 ] =
R[X1 , X2 , . . . , Xn ][Xn+1 ] and R[X1 , X2 , . . . , Xn ] is a UFD by the induction assump-
tion hence R[X1 , X2 , . . . , Xn ][Xn+1 ] is a UFD, too, according to theorem 2.1.

The units in R[X1 , X2 , . . . , Xn ] are again the units in R, but unfortunately the
irreducible polynomials cannot be classified at all, even in particular cases of R.

Definition 3.3. The degree of a monomial cX1 i1 X2i2 . . . Xnin is the sum i1 + i2 +
. . . + in . The degree of a polynomial is the largest of the degrees of its monomials.

Definition 3.4. A polynomial whose monomials all have the same degree is
called a homogeneous polynomial.

Homogeneous polynomials are nice and useful in some more advanced areas of
mathematics, as algebraic geometry. But we will not focus on them much. You will
find some nice properties of homogeneous polynomials in the exercises.

Another important definition pertaining to polynomials in the derivative. Those


of you who have seen calculus know it is defined as a limit. But in the polynomial
case, there is a pure algebraic definition.

Definition 3.5. Let R be a ring, and f ∈ R[X] be a polynomial. The polynomial


f (X + Y ) − f (X) is a polynomial in two variables which can be written as Y g1 (X) +
∂F
Y 2 g2 (X) + . . . +. The polynomial g1 is called the derivative of f , also denoted ∂X ,
0
or f .

20
Why if f (X + Y ) − f (X) divisible by Y ? Simply because plugging in Y = 0
turn the polynomial into 0.

For those of you who know calculus, here is an exercise for you: prove that this
definition of derivative is the same as the calculus one.
The derivative has some nice properties:

Theorem 3.6. Let f, g ∈ R[X]. Then,


i) (f + g)0 (x) = f 0 (x) + g 0 (x)
ii) (f g)0 (x) = f 0 (x)g(x) + g 0 (x)(x)
iii) (f ◦ g)0 (x) = f 0 (g(x)) · g 0 (x)

Proof. All are easy application of the definition. Let f (x + y) = f (x) + yf 0 (x) +
g(x + y) = g(x) + yg 0 (x) + y 2 g1 (x, y). Then
y 2 f1 (x, y),

(f + g)(x + y) = f (x) + g(x) + y(f 0 (x) + g 0 (x)),

(f g)(x+y) = f (x)g(x)+y(f 0 (x)g(x)+f (x)g 0 (x))+y 2 (f 0 (x)g 0 (x)+f (x)g1 (x, y)+g(x)f1 (x, y))+
+y 3 (f 0 (x)g1 (x, y) + g 0 (x)f1 (x, y)) + y 4 f1 (x, y)g1 (x, y)

(f ◦ g)(x + y) = f (g(x + y))


= f (g(x) + yg 0 (x) + y 2 g1 (x, y))
= f (g(x)) + f 0 (g(x))(yg 0 (x) + y 2 g1 (x, y)) + (yg 0 (x) + y 2 g1 (x, y))2 f1 (g(x), yg 0 (x) + y 2 g1 (x, y)
= (f ◦ g)(x) + yf 0 (g(x))g 0 (x) + y 2 (f 0 (g(x))g1 (x, y) + (g 0 (x) + yg1 (x, y))2 ).

The assertions of the problem now follow from the definition.

Note that properties i), ii) allow us to compute the derivative of any polyno-
mial, knowing only the derivatives of linear polynomials and constants, because
every polynomial is a sum of monomials, which are products of constant and linear
polynomials.

But is is actually easy to check directly that the derivative of f (x) = an xn +


an−1 xn−1 + · · · + a0 is f 0 (x) = nan xn−1 + (n − 1)an−1 xn−2 + · · · + a1 , using Newton’s
binomial theorem.

Also, the derivative can be applied more than one time. We can take the deriva-
tive of f to obtain the first derivative, f 0 , then take the derivative of f 0 to obtain
f ”, the second derivative, and so on, to get f (i) , the ith derivative. Parts i) and ii)
transfer to higher derivatives:

21
Theorem 3.7. Let f, g be polynomials. Then
i) (f + g)(k) (x) = f (k) (x) + g (k) (x)
k
ii) (f g)(k) (x) = f (k) g(x)+kf (k−1) g 0 (x)+ f (k − 2)(x)g (2) (x)+. . .+f (x)g (k) (x)

2

Note the similarity between the expression in part ii) and the Newton Binomial
Theorem.

What is the derivative useful for in polynomials? We know it has applications in


calculus, and this relates to polynomials as well since they are functions. But there
are other interesting applications. In fact, the derivative can be used to measure
the power of an irreducible factor of a polynomial. In what comes further, we will
assume that R is either Z, Q, R or C.

Theorem 3.8. Let f, p be non-constant polynomials, p irreducible and assume


that p divides f , p divides f 0 , f ” , . . . , f (k) but it does not divide f (k+1) . Then pk+1
divides f , but pk+2 does not divide f .

Proof. Firstly, let us prove that p divides pm , (pm )0 , . . . , (pm )(m−1) but does
not divide (pm )(m) , by induction on m. For m = 1, we need to prove that p
does not divide p0 , which is true since p0 is non-zero and has degree one less than
the degree of p. Now assume we have proved p|(pk )(i) for i < k but (pk )(k) is
not divisible by p. Then for 0 < i ≤ k + 1, we have (pk+1 )(i) = (pk · p)(i) =
pk · p(i) + i(pk )0 p(i−1) + . . . + i(pk )(i−1) p0 + (pk )(i) p. In the case i ≤ k, all terms of
the sum are divisible by p according to the induction hypothesis, and in the case
i = k + 1, all terms of the sum are divisible by p except the penultimate term,
because neither (pk )(k) nor p0 is divisible by p, and this proves the induction step.

Let f = pm q, where p does not divide q. We wish to prove that m = k + 1.


Let’s prove that m ≥ k + 1 first. Say m ≤ k. Then f (m) must be divisible by p.
But f (m) = (pm q)(m) = (pm )(m) q + m(pm )(m−1) q 0 + . . . + pm q (m) . All terms here are
divisible by p except the very first one, which is a contradiction. Thus m ≥ k + 1.
Say m > k + 1. Then f (k+1) = (pm q)(k+1) = (pm )k+1 q + . . . + pm q (k+1) , and all
terms here are divisible by p, which is again a contradiction. Thus m = k + 1.

Theorem 3.8. gives us a criterion for computing the multiplicity of a given root
r of a polynomial: we need to check first if f (r) = 0, and if yes, then we proceed to
verify whether f 0 (r) = 0, then if f (2) (r) = 0 and so on. If k is the first number such
that f (k) (r) 6= 0, then k is, in fact, the multiplicity of r.

Theorem 3.8., also immediately implies a criterion to check whether a polynomial


is square-free, i.e. is not divisible by the square of a non-constant polynomial.
Indeed, f is divisible by p2 , where p is irreducible, if and only if both f and f 0 are
divisible by p. Therefore:

22
Theorem 3.9. A polynomial f is square-free if and only if the greatest common
divisor of f and f 0 is 1.

If we want to check whether f is cube-free, we need to compute the greatest


common divisor of f, f 0 and f 00 , and so on.

Symmetric Polynomials

We have already met the polynomial (X − X1 )(X − X2 ) . . . (X − Xn ) = X n −


σ1 X n−1 + σ2 X n−2 − .. ± σn , where σ1 = X1 + X2 + . . . + Xn , σ2 = X1 X2 +
X2 X3 + X1 X3 + . . . + Xn−1 Xn , . . . , σn = X1 X2 . . . Xn , according to Viete’s rela-
tions. The expressions σ1 , σ2 , . . . , σn can be regarded as polynomials in the variables
X1 , X2 , . . . , Xn , of degrees 1, 2, . . . , n respectively. A nice property of them is that
they are symmetric in X1 , X2 , . . . , Xn i.e. if you somehow permute the variables
X1 , X2 , . . . , Xn (i.e. exchange X1 with X2 , then X2 with X4 etc. ) then they re-
main the same. It turns out they are, in some sense, the fundamental symmetric
polynomials, according to the following theorem.

Theorem 3.10. Any polynomial P in R[X1 , X2 , . . . , Xn ] that is symmetric


in the variables X1 , X2 , . . . , Xn can be expressed in terms of σ1 , σ2 , . . . , σn , i.e.
there exists a polynomial Q ∈ R[X1 , X2 , . . . , Xn ] such that P (X1 , X2 , . . . , Xn ) =
Q(σ1 , σ2 , . . . , σn ).

A special case of symmetric polynomials is the sum of powers: the polynomials


sk = X1k + X2k + . . . + Xnk . For them, theorem 3.10 can be explicitly rewritten:

Theorem 3.11.(Newton’s identities) The following polynomial identities


hold:
s1 = σ1 ,
s2 = σ1 s1 − 2σ2 ,
.. ..
. .
sk = σ1 sk−1 − σ2 sk−2 + . . . + (−1)k σk−1 s1 − k(−1)k σk
.. ..
. .

These identities allow us to compute s1 , s2 , . . . , sn−1 , . . . , recursively in terms


of σ1 , σ2 , . . . , σn . Also note that the equations can be rewritten as kσk = σk−1 s1 −
σk−2 s2 +. . .+(−1)k−1 sk . Therefore, if 1, 2, 3, . . . , n are invertible in R (for example if
R is Q, R or C) then we can actually compute σ1 , σ2 , . . . , σn in terms of s1 , s2 , . . . , sn .
In this case, every symmetric polynomial in X1 , X2 , . . . , Xn can be expressed as a
polynomial in s1 , s2 , . . . , sn .

23
Now, it is time for examples.

Example 3.1. Say f (x) is a polynomial of degree n ≥ 0. What is the degree of


f 0 (x),
and what is the degree of f (k) (x)?

Solution. If f (x) = an xn + an−1 xn−1 + . . . + a1 x + a0 then f 0 (n) = nan xn−1 +


(n − 1)xn−2 + . . . + a1 has degree 1. Thus, f 0 (x) always has degree one less than the
degree of f , except the case when n = 0 because then both f = a1 and f 0 = 0 have
degree 0.
Therefore the degree of f 0 is n − 1 if n > 0 and 0 if n = 0.
By differentiating k times, we decrease the degree by 1 k times to get degree
n − 1, n − 2, . . . and end up with degree n − k. This is true unless we get to a
polynomial of degree 0 on the way, because we know that in this case all subsequent
derivatives are 0. So if one of n, n − 1, . . . , n − k + 1 is 0, the final degree is 0.
Therefore, if n ≥ k the degree of f (k) is n − k, otherwise f (k) has degree 0 (and
is 0).

Example 3.2. Prove Taylor’s formula:


x2 xn (n)
f (x) = f (0) + xf 0 (0) + f ”(0) + · · · + f (0)
2! n!
when f is a polynomial of degree n.
2 n
Solution. Set g = f (0) + xf 0 (0) + x2! f ”(0) + · · · + xn! f (n) (0). We want to prove
f = g. Now note that g(0) = f (0), g 0 (0) = f 0 (0), g”(0) = f (2) (0), . . . , g (n) (0) =
f (n) (0). Also, g has degree at most n. Therefore p = f − g has degree at most n
and satisfies p(0) = 0, p0 (0) = 0, . . . , p(n) (0). According to the observation following
Theorem 3.8., we deduce that 0 is a root of p of multiplicity at least n + 1, so
xn+1 | p. But since p has degree at most n, it can only happen that p = 0. Thus
f = g, as desired.

Example 3.3. Let x, y be complex numbers. It is known that there exists a


polynomial f ∈ Q[X] such that f (x) = 0 and a polynomial g ∈ Q[X] such that
g(y) = 0. Prove that there exists a polynomial h ∈ Q[X] such that h(x + y) = 0.

Solution. This illustrates an important application of the theorem about sym-


metric polynomials. Say f (t) = (t − x1 )(t − x2 ) · · · (t − xn ) with x = x1 and
g(t)
Qn = Qm(t − y1 )(t − y2 ) · · · (t − yn ) with y = y1 . Consider the polynomials h =
i=1 j= (t − xi − yj ). Then x + y is a root of h and we now prove that h has
rational coefficients. Indeed, h is symmetric in x1 , x2 , . . . , xn and in y1 , y2 , . . . , ym .
Therefore its coefficients are, too. But then they can be expressed as polynomials
in the coefficients of f and of g, because of Theorem 3.10! Since these coefficients
are rational, we are done.

24
Assigned problems

41. Prove theorem 3.7. part ii).

42. Let f ∈ R[X, Y ] be a polynomial in two variables. We know P (X, X) is


identically 0. Prove that f (X, Y ) is divisible by X − Y .

43. Let p be an irreducible polynomial in the UFD R[X] and say that f is a
polynomial divisible by pm but not by pm+1 . Show that for 1 ≤ k ≤ m, f (k)
is divisible by pm−k but not by pm−k+1 .

44. Let f be a polynomial of degreen, and a be in the domain of f . Prove that


2 n
f (x) = f (a) + f 0 (a)(x − a) + x2! (x − a)2 + · · · + xn! (x − a)n .

45. Let x, y be complex numbers. It is known that there exists a polynomial


f ∈ Z[X] such that f (x) = 0 and a polynomial g ∈ Z[X] such that g(y) = 0.
Prove that there exists a polynomial h ∈ Z[X] such that h(xy) = 0.

46. Let 1 < k < 2009 be an integer. Consider k integers x1 , x2 , . . . , xk such that
all the sums x1 + x2 + · · · + xk , x21 + x22 + · · · + x2k , . . . , xk1 + xk2 + · · · + xkk equal
2009!, except one of them which equals 2009! + 1. Show that the sum that
equals 2009! + 1 is x1 + x2 + . . . + xk .

47. When we have a polynomial F in R[X1 , X2 , . . . , Xn ], we can regard it as a


polynomial in Xi (with coefficients in R[X1 , X2 , . . . , Xi−1 , Xi+1 , . . . , Xn ]). We
∂F
then can take its derivative with respect to Xi to get ∂X i
. This polynomial
can be regarded as a polynomial in Xj , so now we can take the derivative with
2F
respect to Xj and get what is denoted as ∂X∂i ·∂X j
. Prove the identity

∂2F ∂F
=
∂Xi ∂Xj ∂Xj · ∂Xi

That is, if we differentiate with respect to Xi first and with respect to Xj


later, we will get the same result as if we differentiate with respect to Xj first
and with respect to Xi later.

48. Let F be a homogeneous polynomial in R[X1 , X2 , . . . , Xn ] of degree d. Prove


Euler’s Theorem:
∂F ∂F ∂F
· X1 + · X2 + · · · + · Xn = d · F (X1 , X2 , . . . , Xn )
∂X1 ∂X2 ∂Xn

49. Let f be a polynomial in x1 , x2 , . . . , xn with the following property: if in the


formula for f we plug in xi instead of xj and xj instead of xiQ , we get −f . How
can we express f in terms of σ1 , σ2 , . . . , σn ? [Hint: g(x) = i<j (xi − xj ) has
the same property of f . Now, what can we say about fg ?]

25
50. Let f, g ∈ C[X]. It is known that the polynomials f (x) and g(x) have the
same set of roots, and the polynomials f (x) − 1 and g(x) − 1 have the same
set of roots, too. Prove that f = g. [Hint: Take derivatives. How do multiple
roots transfer to derivatives?]

4 Irreducible polynomials

This chapter is dedicated to polynomials that are “bricks” in creating other


polynomials. Like every number can be factored into primes, every polynomial can
be factored into irreducible polynomials. We will repeat here some of the definitions
from section 2. is is done to help those who sunk in the modern algebra language of
that chapter. Here we will emphasize once more some of the important definitions,
but with less formalism.
First of all let us repeat some important facts. Probably you have noticed that
each time we carefully mention coefficients for the given polynomial. They can be
rational numbers, real numbers, complex numbers or even residues modulo prime p.
In this case we say p(x) ∈ Q[X] R[X], C[X] or Zp [X]. All these algebraic structures
are named fields.

The question that arises if one can factor a polynomial as a product of two
polynomials with coefficients from the same field.

Example 4.1. Prove that x4 + x2 + 1 can be factored over the field of rationals.
Solution. This means that we can factor p(x) = x4 + x2 + 1 as a product of
two polynomials with rational coefficients. Indeed,

x4 + x2 + 1 = x4 + 2x2 + 1 − x2 = (x2 + 1)2 − x2 = (x2 − x + 1)(x2 + x + 1).

Example 4.2. Prove that x3 + 2 cannot be factored over the field of rationals.

Solution. Assume to the contrary, it can be factored: x3 +2 = (x−a)(x2 +bx+c),


where a, b, c are rational. Hence x3 + 2 = x3 + (b − a)x2 + (c − ab)x − ac and since
coefficients
√ must coincide we get a = b, c = ab, ac = −2. So c = a2 , a3 = −2 and
3
a = −2 which is clearly not rational.

From here we can give a very intuitive notion of reducible polynomial and irre-
ducible polynomial.
A polynomial f is called reducible if it can written as product of two polynomials
f = gh, where deg g, deg h > 0. In the other case polynomial f is called irreducible.
To verify if some polynomial g divides f we simply can use Euclidean algorithm.
The following theorem is analogous to prime number decomposition theorem.

26
Theorem 4.1. Every polynomial with coefficients from some field can be written
as a product of irreducible polynomials. Furthermore, this decomposition is unique.

Example 4.3. Factor x4 − 1 as a product of irreducible polynomials with


rational coefficients.

Solution. The decomposition is x4 − 1 = (x − 1)(x + 1)(x2 + 1) because x2 + 1


is irreducible over the field of rationals.

Now you may have noticed that we did not mentioned specifically decomposition
into polynomials with integer coefficients. But it turns our that if we factor out
polynomial with integer coefficients over the integers or the field of rationals there
is no difference.

Gauss Lemma

Theorem 4.2. A polynomial with integer coefficients is irreducible over the


integers if and only it is irreducible over the field of rationals.

Proof. (Can you see the connection between this theorem and the Gauss Lemma
of chapter 2?) For simplicity we prove this result for monic polynomials (to avoid
unnecessary complications right now). Let f = xm+n + · · · + cm+n−1 x + cm+n , ci ∈
Z[X]. Assume that polynomial f as f = gh, where g(x) = xm + am−1 xm−1 + · · · + a0
and h(x) = xn + bn−1 xn−1 + · · · + b0 are polynomials with rational coefficients. We
prove that coefficients of g and h are integers.
Let M and N respectively be the least positive integers so that all coefficients of
M g(x) and N h(x) are integers (that is, M and N are the least common multiples
of the denominators of a0 , . . . , am−1 and b0 , . . . , bn−1 , respectively). Let Ai = M ai
for i ∈ {0, . . . , m}, Bj = N bj for j ∈ {0, . . . , n} and Am = M, Bn = N , then

M N g(X)h(X) ≡ Am Bn X m+n + · · · + A0 B0 = M N f (x)

Since g(X)h(X) ∈ Z[X], all coefficients of M N g(X)h(X) are divisible by M N .


Let p be a prime divisor of M N (if M N = 1 we are done). There exists an integer
i ∈ {0, . . . , m} such that p - Ai . Indeed, if p - M, then p - Am and if p | M , then
p | Ai for all i ∈ {0, . . . , m} would imply that Api = M
p ai ∈ Z, a contradiction to the
minimality of M . Similarly, there exists an integer j ∈ {0, . . . , n} such that p - Bj .
Let I and J be the greatest integers among these numbers i and j, respectively.
Then the coefficient of X I+J in M N g(X)h(X) is

[X I+J ] = · · · + AI+1 BJ−1 + AI BJ + AI−1 BJ+1 + · · · ≡ AI BJ + p · R

where R is an integer, so in particular, it is not divisible by p which contradicts the


fact that the coefficients of M N g(X)h(X) are divisible by M N .

27
Remark. Now, if we are asked to prove if some polynomial with integer coeffi-
cients is irreducible over field of rationals, it is enough to prove that it is irreducible
over integers.

There are so many polynomials that there does not exist a theorem or some key
that immediately unveils if a particular polynomials is irreducible. That is why we
continue with a list of criteria that help us to answer affirmatively on the question of
irreducibility, of course under certain conditions. All they are based on the analysis
of coefficients or zeros of the polynomial. The first criterion is famous and simple
to memorize is Eisenstein’s Criterion.

Eisenstein’s Criterion.

Theorem 4.3. Consider a polynomial f (x) = an xn + an−1 xn−1 + · · · + a1 x + a0 ,


f ∈ Z[x]. If there exist a prime p such that p | a0 , a1 , . . . an , p - an and p2 - a0 , then
f is irreducible.

Proof. We make a very common assumption, typical for problems of this type:
assume to the contrary f = gh. Let g(x) = bm xk + bk−1 xk−1 + · · · + b1 x + b0 and
h(x) = cl xl + bl−1 xl−1 + · · · + c1 x + c0 , where k + l = n and k ≥ l ≥ 1.
Now, using given conditions we look for a contradiction. Considering the product
of g and h we express coefficients ai in terms of bi and ci : a0 = b0 c0 , a1 = b0 c1 +b1 c0 ,
and so on. From the fact that p | a0 and p2 - a0 we have p | b0 c0 , so without loss of
generality p | b0 and p - c0 . Then p | a1 implies p | b0 c1 + b1 c0 and p | b1 . Note that
if all bi s are divisible by p, then an is divisible be p. Let i be the least coefficient
such that bi is not divisible by p. Then ai = b0 ci + b1 ci−1 + · · · + bi−1 c1 + bi c0 , but
p | ai , p | b0 , . . . bi−1 , so p must divide bi , a contradiction.

Example 4.4. Let q be a prime. Prove that xn + q is irreducible in Z[X].

Solution. This problem immediately follows from Eisenstein’s Criterion. All co-
efficients except the leading one are divisible by q and free coefficient is not divisible
by q 2 . So xn + q is irreducible.

Example 4.5. Prove that if p is a prime then f = xp−1 + xp−2 + · · · + x + 1 is


irreducible in Z[X].
p
−1
Solution. Note that f (x) = xx−1 is irreducible if and only if polynomial f (x−1)
is irreducible. Observe that this is just a shift in variable and if we express f (x − 1)
as a product of two polynomials then we can express f (x) as a product of two
polynomials. So it is enough to prove that
(x − 1)p − 1
     
p−1 p p−1 p p
f (x−1) = = (x−1) + (x−1) +· · ·+ (x−1)+
x−1 1 p−2 p−1

28
is irreducible. We use Eisenstein’s Criterion, p divides all the coefficients except the
2 p

leading one and p - p−1 = p. Thus f is irreducible.

Remark. Remember this trick with a shift: instead of proving f (x) is irre-
ducible, you can prove that f (x + a) is irreducible, for some a ∈ Z.

The next criterion is based on the idea of a dominant coefficient. But before
writing its statement we would like to prove the following lemma which plays a
crucial role in its proof.

Lemma. Let f (x) = xn + an−1 xn−1 + · · · + a1 x + a0 , a0 6= 0 be a polynomial


such that
|an−1 | > 1 + |a0 | + |a1 | + · · · + |an−2 |,
then one of the zeros has absolute value greater than 1 and all other zeros have
absolute value less than 1.

Proof. Note that there does not exist a root α such that |α| = 1. Assume to
the contrary, then

−an−1 αn−1 = αn + an−2 αn−2 + · · · + a1 α + a0 .

It follows that

|an−1 | = | − an−1 αn−1 | = αn + an−2 αn−2 + · · · + a1 α + a0


≤ |α|n + an−2 |α|n−2 + · · · + a1 |α| + a0


= 1 + |an−2 | + · · · + |a1 | + |a0 |,

a contradiction.
From Viete’s theorem |a0 | = |x1 · · · xn | ≥ 1. Therefore there exist a zero, say x1 ,
such that |x1 | > 1. Let f (x) = (x − x1 )g(x), where

g(x) = xn−1 + bn−2 xn−2 + · · · + b1 x + b0 .

It follows that

f (x) = xn + (bn−2 − x1 )xn−1 + (bn−3 − bn−2 x1 )xn−2 + · · · + (b0 − b1 x1 )x − b0 x1

and initial condition becomes

|bn−2 − x1 | > 1 + |bn−3 − bn−2 x1 | + · · · + |b0 − b1 x1 | + |b0 |.

Because |a| + |b| ≥ |a − b| ≥ |a| − |b| we get

|bn−2 | + |x1 | > 1 + |bn−2 | − (|x1 | − 1)(|bn−2 | + |bn−3 | + · · · + |b0 |),

29
so 1 > |bn−2 | + |bn−3 | + · · · + |b0 |.

If there exist a zero xi of g such that |xi | > 1, then xin−1 = − n−2 k
P
k=0 bk xi , yielding
n−2 n−2 n−2
X b
k
X |bk | X
1 = − ≤ ≤ |bk |,

xn−1−k |xi |n−1−k

i

k=0 k=0 k=0

a contradiction.

Perron’s Criterion

Theorem 4.4. Let f (x) = xn +an−1 xn−1 +· · ·+a1 x+a0 , a0 6= 0 be a polynomial


such that
|an−1 | > 1 + |a0 | + |a1 | + · · · + |an−2 |,
then f is irreducible over the integers.

Proof. Assume to the contrary, f = gh, where g, h ∈ Z[X]. Then using lemma
proved above we conclude that one of them, say g has all zeros of absolute value
less than 1. Using Viete’s theorem its free coefficient is equal to |xi1 · · · xik | < 1, a
contradiction.

Example 4.6. Let fn be the Fibonacci sequence, given by f0 = f1 = 1, fn+1 =


fn + fn−1 for n ≥ 1. Let an = fn+1 fn and let
Pn (x) = xn + an−1 xn−1 + · · · + a1 x + a0 .
Prove that for all integers n ≥ 3, polynomial Pn (x) is irreducible in Z[X].

Solution. Using induction we prove that


fn+1 fn > fn fn−1 + · · · + f1 f0 + 1.
The base case n = 2 is true: f3 f2 = 6 > 4 = f2 f1 + f1 f0 + 1. Assume for n the
statement holds, then
2
fn+2 fn+1 = fn+1 + fn+1 fn
> 2fn+1 fn
> fn+1 fn + fn fn−1 + · · · + f1 f0 + 1,
and the induction step is proved.
Now, because |an−1 | > 1 + |a0 | + |a1 | + · · · + |an−2 |, we can use Perron’s Criterion
to prove that Pn (x) is irreducible in Z[X], as desired.

There are many attempts to relate prime numbers with irreducible polynomials.
One of them is Bouniakowsky conjecture.

30
Bouniakowsky Conjecture

Let f (x) be an irreducible polynomial with integer coefficients, degree greater


than 1, and gcd(f (1), f (2), . . .) = 1. The Bouniakowsky conjecture states that f (x)
is prime for an infinite number of integers x. As an example of the greatest common
divisor caveat, the polynomial 3x2 − x + 2 is irreducible, but always divisible by 2.
Irreducible polynomials of degree 1, ax + b always generate an infinite number
of primes by Dirichlet’s theorem. The existence of a polynomial that can produce
an infinitude of primes is undetermined. The weaker conjecture asserts that a2 + 1
is prime for an infinite number of integers a > 1 which is also not proved.

The last criterion that we would like to mention is Cohn’s Criterion. We state it
without a proof and leave it for you to admire how primes are closely related with
irreducible polynomials.

Cohn criterion

Theorem 4.7. Let an an−1 . . . a0 be the representation of a prime number in a


base b, b ≥ 2. Then p(x) = an xn + an−1 xn−1 + · · · + a1 x + a0 is irreducible in Z[x].

Example 4.7. Let abc be a prime. Prove that b2 − 4ac cannot be a perfect
square.

Solution. Consider polynomial p(x) = ax2 + bx + c. Using Cohn’s Criterion for


standard base 10 we have a · 102 + b · 10 + c is a prime and p(x) must be irreducible.
However, if b2 − 4ac is a perfect square, then p(x) will have rational roots, so it can
be factored, a contradiction.

31
Assigned problems

51. Find if the following polynomials are irreducible in Z[X]


a) p(x) = x4 + 1
b) p(x) = x5 + x + 1

52. Prove that x4 − x3 − 3x2 + 5x + 1 is irreducible in Z[X].

53. Prove that f (x) = xn + 5xn−1 + 3 is irreducible in Z[X] for n ≥ 2.


n
54. Prove that f (x) = x2 + 1 is irreducible in Z[X] for n ≥ 1.

55. Let n be an integer greater than 1. Prove that f (x) = xn + n! is irreducible


in Z[X].

56. Prove that if f (x) = an xn + an−1 xn−1 + · · · + a1 x + a0 is irreducible over


integers, then g(x) = a0 xn + a1 xn−1 + · · · + an−1 x + an is also irreducible.

57. Let a1 < a2 < · · · < an , n ≥ 2 be integers. Prove that polynomial f (x) =
(x − a1 )(x − a2 ) · · · (x − an ) − 1 is irreducible in Z[X].

58. Let p be a prime and a, n be integers such that p > |a| + 1. Prove that
polynomial f (x) = xn + ax + p is irreducible in Z[X].

59. Let a1 < a2 < · · · < an be integers. Prove that polynomial

f (x) = (x − a1 )2 (x − a2 )2 · · · (x − an )2 + 1

is irreducible in Z[X].

60. Let f1 , f2 , . . . , fn be polynomials with integer coefficients. Prove there exist a


reducible polynomials g with integer coefficients such that f1 +g, f2 +g, . . . , fn +
g are irreducible in Z[X].

5 Cyclotomic Polynomials

Cyclotomic polynomials represent a class of polynomials largely studied and


very important in Number Theory. Right now we do not want to overwhelm the
reader with complicated theory. This chapter presents some elementary properties
of them which at the first time of reading are indeed hard. Our goal is to prove
that cyclotomic polynomials are irreducible over the integers. Passing this material
should be enough to become acquainted with these polynomials and recognize them
while solving problems in the future.

32
The theory of cyclotomic polynomials is based on many mathematical concepts.
Two of them, of a big importance, will be reviewed here for you. These are Möbius
Function and Primitive Roots of a Unity. Do not worry if you are not so comfortable
with them, sometimes it is worth to jump over scary material and go ahead to find
out the whole story.

Definition 5.1. The Möbius Function µ : Z+ → {−1, 0, 1} is defined as follows:



1
 if n = 1
k
µ(n) = (−1) if n is squarefree and k is the number of prime divisors of n

0 else.

Clearly, µ is multiplicative, that is, µ(mn) = µ(m)µ(n) for all coprime positive
integers m and n. The following result can be found in many books on multiplicative
functions.
Theorem 5.1. (Möbius Inversion Formula) Suppose that F, H, f : Z+ → Z+
are functions such that
X Y
F (n) = f (d), H(n) = f (d).
d|n d|n

Then n  n µ(d)


X Y
f (n) = µ(d)F = H .
d d
d|n d|n

Definition 5.2. Let n be a positive integer and ζ be an nth root of unity.


Then the least positive integer k that satisfies ζ k = 1 is called the order of ζ and is
denoted by ord(ζ). Also ζ is called a primitive nth root of unity if ord(ζ) = n.
Lemma 5.1. Let n and k be positive integers and ζ be a primitive nth root of
unity. Then ζ k is a primitive nth root of unity if and only if gcd(k, n) = 1.
Proof. Let d = ord(ζ k ), then ζ kd = 1. Since ζ n = 1 and n is the least positive
integer that satisfies this relation, it follows that n | kd. If gcd(k, n) = 1, then n | kd
implies n | d. But d also divides n, so d = n and ζ k is primitive.
If gcd(k, n) 6= 1, then
n
k gcd(k,n)
ζ = 1,
so d < n. Thus ζ k is not primitive.

Corollary 5.1. Let n be a positive integer. Then there exist exactly ϕ(n)
primitive nth roots of unity.
Cyclotomic polynomials

33
Definition 5.3. Let n be a positive integer. Then the nth cyclotomic polyno-
mial, denoted as Φn , is the (monic) polynomial having exactly the primitive nth
root of unity as roots, that is,
Y
Φn (X) ≡ (X − ζ).
ζ n =1
ord(ζ)=n

We proved that there are exactly ϕ(n) primitive nth roots of unity, thus the
degree of Φn is ϕ(n).

Example 5.1. Find the cyclotomic polynomial Φ3 (X).


Solution. Note that roots of polynomial x3 − 1 = 0 are 1, , 2 . Only  and 2
are primitive roots. Hence

x3 − 1
Φ3 = (x − )(x − 2 ) = = x2 + x + 1.
x−1

Theorem 5.2. Let n be a positive integer. Then


Y
Xn − 1 ≡ Φd (X).
d|n

Proof. The roots of X n − 1 are exactly the nth roots of unity. On the other
hand, if ζ is an n−th root of unity and d = ord(ζ), then ζ is a primitive dth root
of unity and thus a root of Φd (X). But d | n, so ζ is a root of the right hand side.
It follows that the polynomials on the left and right hand side have the same roots
and since they are both monic, they are equal.
Notice that comparing degrees of the polynomials yields another proof of
X
n= ϕ(d).
d|n

Example 5.2. Find the cyclotomic polynomial Φ6 (X).


Solution. From Theorem 2 we know

(x6 − 1) = Φ1 (x)Φ2 (x)Φ3 (x)Φ6 (x)


= (x − 1)(x + 1)(x2 + x + 1)Φ6 (x).

Thus Φ6 (x) = x2 − x + 1 and we are done.

One of the first questions that we would like to ask, do all cyclotomic polynomials
have integer coefficients? Yes, they do.

34
Theorem 5.3. Cyclotomic polynomials Φn (X) ∈ Z[X].

Proof. The proof is based on the following lemma which we proved while
discussing Gauss Lemma.
Lemma. Suppose that f (X) ≡ X m + am−1 X m−1 + · · · + a0 and g(X) ≡ X n +
bn−1 X n−1 + · · · + b0 are polynomials with rational coefficients. If all coefficients of
the polynomial f · g are integers, then so are the coefficients of f and g.
The proof goes by induction on n. The statement is true for n = 1 since Φ1 (X) =
X − 1. Suppose that the statement is true for all k < n. Then from Theorem 2 we
obtain
Xn − 1
Φn (X) = Q ,
d|n,d6=n Φd (X)

so the coefficients of Φn (X) are rational and thus by lemma integers.

We can also use the Möbius Inversion to obtain a direct formula for the cyclo-
tomic polynomials. Using Theorems 1 and 2 we get

Theorem 5.4. Let n be a positive integer. Then


Y n
µ(d)
Φn (X) = Xd −1 .
d|n

The next property is a basic property of all irreducible polynomials.

Theorem 5.5. Consider a polynomial f ∈ Z[X] with a  root. Then there exist
a unique irreducible polynomial g ∈ Z[X] that contains  as a root and moreover
g | f.

Proof. Let g ∈ Z[X] be polynomial with the least degree n that contains  as a
root. Clearly, it is irreducible in Z[X], otherwise one of its divisors contains  and
has lower degree. Assume that there exist another polynomial g 0 of the same degree
that also and has lower degree. Assume that there exist another polynomial g 0 of
the same degree that also a contradiction.
Thus g ∈ Z[X] is a unique irreducible polynomial that contains  as a root.
Assume that g - f , then using Euclidean algorithm there exist a polynomial q(x) =
gcd(f, g) ∈ Z[X] of degree less than n, that contains  as a root. This contradicts
the minimality of degree of g, so g | f .

Now, we are ready to prove our main theorem.

Theorem 5.6. Cyclotomic polynomials are irreducible over the integers.

35
Proof. Let  be a primitive n−th root of a unity and let p be any prime not
dividing n. Assume to the contrary Φn (X) factors into g · h with h irreducible, and
let h have the root .
Since p and n are coprime, p is another primitive root. Suppose p is a root of
g(x). This means  is a root of g(xp ). Since h also has root  and h is irreducible,
using Theorem 5 we have h is a factor of g(xp ). Let g(xp ) = h · k.
Reducing all coefficients modulo p yields the following polynomial equation

h · k = g(xp ) = g(x)p .

Some irreducible factor of h, possibly h itself, now divides g. When looking


modulo p, h and g have a root in common, and xn − 1 has a multiple root. Using
formal derivatives the derivative of xn −1 is nxn−1 , which is nonzero, since p does not
divide n. This has no factors in common with xn −1, because nxn−1 x−n(xn −1) = n.
Hence there are no double roots.
This contradiction means p is not a root of g(x). Therefore it belongs to h(x),
2 3
and similarly do p , p , etc.
Let us look at a specific primitive root j . If p2 divides j, raise  to the power
p, and then to the power p again. This is a primitive root that lies in h. Then
if q divides j, raise this root to the power q and find another primitive root in h.
Continue through the primes of j, until j lies in h. Since h contains all the primitive
roots, it is equal to Φn (X), and Φn (X) is irreducible.

The following theorems are examples of how cyclotomic polynomials and Number
Theory may be related. We start with the lemma similar to the trick we have used
in proving irreducibility of Φn (X).

Lemma 5.2. Let p be a prime number. Suppose that the polynomial X n − 1


has a double root modulo p, that is, there exists an integer a and a polynomial
f (X) ∈ Z[X] such that X n − 1 ≡ (X − a)2 f (X) (mod p). Then p | n.

Proof. We can also prove this lemma with calculus modulo p by introducing the
familiar rules for computing the derivative (and showing that they are consistent).
The fact that a double root of a function is a root of its derivative remains invariant
modulo p. Hence a is a root of X n − 1, implying an ≡ 1 (mod p). Also, a is the
root of the derivative of X n − 1, so nan−1 ≡ 0 (mod p). But p - a, so p | n.

Corollary 5.2. Let n be a positive integer, d < n a divisor of n integer. Suppose


that p divides Φn (x0 ) and Φd (x0 ), where x0 ∈ Z . Then p | n.

Proof. By Theorem 2, Y
xn − 1 = Φt (x),
t|n

36
so xn − 1 is divisible by Φn (x)Φd (x). It follows that the polynomial X n − 1 has a
double root at x0 modulo p, so by lemma above, p | n.

Using facts we have just proved we can prove the following beautiful result.

Theorem 5.7. Let n be a positive integer and x be any integer. Then every
prime divisor p of Φn (x) either satisfies p ≡ 1 (mod n) or p | n.

Proof. Let p be a prime divisor of Φn (x). Note that p - x, because p | Φn (x) |


xn − 1. Let k = ordp (x). Since p | xn − 1, we have xn ≡ 1 (mod p), so k | n. Hence
k = n or k < n.
1st case: k = n. By Fermat’s Little Theorem, p | xp−1 − 1 since p - x and p is
prime. But then k | (p − 1), and since k = n, n | (p − 1) so p ≡ 1 (mod n).
2nd case : k < n. Since
Y
0 ≡ xk − 1 = Φd (x) (mod p),
d|k

there exists a divisor d of k so that p | Φd (x). Observe that d ≤ k, because d | k. But


k < n, so d < n. Furthermore d is a divisor of n, as d | k | n. Now let us consider
the decomposition of xn − 1 into cyclotomic polynomials. We have two divisors of n
(which are d and n) that divide n. So we have two cyclotomic polynomials in that
factorization that have p as a factor. Thus it follows from Corollary 2 that p | n.

Finally, we present a common application of cyclotomic polynomials, the proof


of a special case of Dirichlet’s Theorem.

Theorem 5.8. Let n be a positive integer. Then there exist infinitely many
prime numbers p with p ≡ 1 (mod n).

Proof. Suppose that there exist only finitely many prime numbers p with p ≡ 1
(mod n). Let T > 1 be the product of these primes and all primes diving n. Let k be
a sufficiently large positive integer such that Φn (T k ) > 1 (since Φn is a nonconstant
monic polynomial, such k exists) and let q be a prime divisor of Φn (T k ). Because q
divides T kn − 1, q does not divide T , so q 6≡ 1 (mod n) and q - n, a contradiction to
Theorem 7.

37
Assigned problems

61. Find cyclotomic polynomials Φ4 (X), Φ10 (X), Φ12 (X).

62. Prove that irreducible polynomials cannot have multiple roots.

63. Let p be a prime number and x be an integer. Then every prime divisor q of
1 + x + · · · + xp−1 either satisfies q ≡ 1 (mod p) or q = p.

64. Find all integer solutions of the equation

x7 − 1
= y 5 − 1.
x−1

65. Suppose that f ∈ Z[X] is an irreducible polynomial. It is known that f has a


root of norm larger than 23 . Prove that if α is a root of f then f (α3 + 1) 6= 0.

38
6 Lagrange Interpolation Formula

This is a small but important topic. Assume that we have a polynomial p(x)
of degree n and we know the values it returns for x1 , x2 , . . . , xk . We would like to
determine this polynomial, find its coefficients and values that it returns at other
points. Lagrange interpolation is a method that helps us construct such a polyno-
mial. In general, if we know values for n + 1 different points then polynomials is
unique.

Theorem 6.1. Suppose we have n + 1 values P (x0 ), P (x1 ), . . . , P (xn ) for a


polynomial of degree n. Then this polynomial is equal to
 
n
X Y x − xi
P (x) =   · P (xi ).
xj − xi
i=0 i6=j

Proof. Actually, we can persuade ourselves that whenever we look for P (xi )
we plug xi in all products defined above. They are constructed in such way that
coefficient near P (xj ), i 6= j is
(x − x1 ) · · · (x − xi ) · · · (x − xj−1 )(x − xj+1 ) · · · (x − xn )
(xj − x1 ) · · · (xj − xi ) · · · (xj − xj−1 )(xj − xj+1 ) · · · (xj − xn )
and returns 0 when x = xi . The only nonzero coefficient appears near P (xi ) and is
equal to 1. So we agree that this polynomial satisfies conditions.
Assume that there exists another polynomial Q(x) which also satisfies conditions.
Then H(x) = P (x) − Q(x) is a polynomial of degree at most n that has n + 1 roots,
so H(x) = 0, a contradiction.

Example 6.1. Let P (x) be a polynomial of degree n such that


k
P (k) = for each k = 0, 1, 2, . . . , n.
k+1
Find P (n + 1).

Solution. Consider
1
Q(x) = x + (0 − x)(1 − x) · · · (n − x),
(n + 1)!
then Q(−1) = 0 and Q(x) is divisible by (x + 1). Let Q(x) = P (x)(x + 1). Note that
P (k) = Q(k) k
k+1 = k+1 for each k = 0, 1, 2, . . . , n. Hence P (x) is exactly polynomial we
are looking for. Thus
Q(n + 1) n + 1 + (−1)n+1
P (n + 1) = = .
n+2 n+2

39
Example 6.2. Let P (x) = xn + an−1 xn−1 + · · · + a1 x + a0 . Consider n + 1
different integers x0 < x1 < · · · < xn . Prove that there exists i such that
n!
P (xi ) ≥ .
2n

Solution. Assume to the contrary, P (xi ) < 2n!n for i = 0, 1, . . . n. From the
Lagrange Interpolation we have
 
n
X Y x − xi
P (x) =   · P (xi ).
xj − xi
i=0 i6=j

Therefore the leading coefficient is equal to



n n
X Y 1 X n! Y 1
P (xj ) < n

j=0 xi − xj
2 |xi − xj |
i6=j j=0 i6=j
n
X n! Y 1 Y 1

2n j−i i−j
j=0 i<j i>j
n
X 1 n!
= n
2 j!(n − j)!
j=0
n  
1 X n
= n = 1,
2 j
j=0

a contradiction.

40
Assigned problems

66. Let x1 , x2 , . . . , xn be different real number. Prove that


X 1 Y 1 (−1)n+1
= .
xj xj − xi x1 x2 · · · xk
j=1 i6=j

67. Let P (x) be a polynomial of degree n such that


1
P (k) = n+1
 for each k = 0, 1, 2, . . . , n.
k

Find P (n + 1).

68. Let P (x) be a polynomial of degree n such that


1
P (k) = for each k = 0, 1, 2, . . . , n.
k
Find P (n + 1).

69. Let r 6= 0 be a real number. Let P (x) be a polynomial of degree at most n


such that
P (k) = rk for each k = 0, 1, 2, . . . , n.
Find P (n + 1).

70. Let P (x) be a polynomial of degree not greater than 2n such that for every
integer k ∈ [−n, n] we have |P (k)| ≤ 1. Prove that for each x ∈ [−n, n] we
have |P (x)| ≤ 22n .

41

You might also like