Lecture 27

POLYNOMIALS
“Polynomials”
n transliterates as “many numbers”. This refers to the coefficients aj ’s of the
x ’s in j=0 aj xj . Before we discuss such things therefore, it is imperitive that we come
j

to a more intimate understanding of number systems.

Standard Notations for Number Systems.

Nonnegative integers: N (Natural numbers, or cardinal numbers)

N := {0, 1, 2, . . . }.
N can be constructed purely set theoretically. The Fundamental Theorem of Arithmetic
states that the natural numbers greater than 1 can be written uniquely as the product
Πsi=1 pα
i for prime numbers pi and positive natural numbers s and αi . The way this works
i

is not yet fully understood and largely depends on the Riemann Hypothesis which has not
yet been proven (Bombieri, 2000).

Nonpositive integers: N−
N− := {. . . , −2, −1, 0}

Integers: Z (German: Zahlen=numbers)

Z := N ∪ N−

Positive integers: Z+ = P
Z+ = P := N\{0}.

Negative integers: Z−
Z− := N− \{0}

 numbers: Q (Quotients)
Rational 
Q := pq : p ∈ Z & q ∈ Z\{0}

Positive rational numbers:

 Q+
Q+ := pq : p, q ∈ Z+

Negative rational numbers: Q−

Q− := { pq : p ∈ Z− & q ∈ Z+ }

Real numbers: R
R can be constructed from Q. It is easier but less rigorous to simply say there is a 1-1
correspondence between R and the real number line. R is up to isomorphism the only
complete ordered field.
Positive real numbers: R+
R+ := {r ∈ R : r > 0}
(where “r > 0” means that r occurs to the right of 0 on the real number line).

Negative real numbers: R−

R− := {r ∈ R : r < 0}
(where “r < 0” means that r occurs to the left of 0 on the real number line).

All the above number systems are one dimensional.

Complex numbers: C - two dimensional numbers

C := {a + ib, c + id : a, b, c, d ∈ R, i is an arbitrary symbol such that (a + ib) + (c + id) =
(a + c) + i(b + d) & (a + ib)(c + id) = (ac − bd) + i(bc + ad)}
Letting a = c = 0, & b = d = 1 we see from this definition that i2 = −1.
∞ 1 √
The notations e (for n=0 n! ) and i (for −1) were invented by Leonard Euler. These
numbers have been very influential in mathematics.

Another notation (William Rowan Hamilton’s) for complex numbers is as ordered pairs
(a, b)(= a + ib) where a, b ∈ R but this is rarely used thesedays.

Quaternion numbers: H (for Hamilton who invented them) - four dimensional numbers

Octonion numbers: O - eight dimensional numbers

C, H, & O are all based on R and the units 1 and i, j, k, l, m, n, o which are such that
i2 = j 2 = k 2 = l2 = m2 = n2 = o2 = −1. H and O are defined similarly to C together
with other conditions on i, j, k, l, m, n, o.

Theorem. P ⊂ N ⊂ Z ⊂ Q ⊂ R ⊂ C ⊂ H ⊂ O.

Occasionally in the literature, we see N confused with P. We also see the notation P used
for a projective set. In these notes however, we will use these symbols as they are defined
above.

In this notation I am using BLACKBOARD BOLD font, and this is also how we handwrite
them, but we often see them typed in BOLD font, i.e.:

P, N, Z, Q, R, C, H, O.

An older and less frequently used notation is (in MEDIUM ROMAN font):

J for integers
R for rationals
R* for reals
C for complex numbers
n
We can generalise complex numbers to n-dimensional hypercomplex numbers j=1 ij aj
where n ∈ Z+ , aj ∈ R, i1 = 1 and i22 = · · · = i2n = −1, as well as several other conditions
on the ij ’s. However, we actually don’t need number systems with dimensions other than
1, 2, 4 or 8. In particular, if S is a finite dimensional division algebra over R, then the
dimension of S as a vector space over R is equal to 1, 2, 4 or 8. This was first suspected
as early as the early 1800’s. There are now several proofs of this and of several equivalent
statements in the literature. But it was first proved in the 1950’s by Milnor who built
upon the work of Bott (Milnor, 1958) and independently by Kervaire (Kervaire, 1958).

Sometimes however, we use the notations X∗ for the extended number system of X where
X is one of the number systems above.
X∗ := X ∪ {∞} or X ∪ {−∞} or X ∪ {−∞, ∞}.

The notation R∗ is also sometimes used to represent the hyperreal number system, which is
the same as R but with the added axiom of calculus that infinitessimals δx exist. Applied
mathematicians are famous (or infamous, depending on your perspective) for using the
idea of infinitessimals as if they were elements of R and they have had great success in
so doing. Pure mathematicians will tell you that they are not elements of R. However,
if we use this notion as an axiom and in so doing replace R by R∗ := R ∪ {δx} it
is possible to construct a pure mathematical calculus system without the use of limits.
Calculus questions in R will still have solutions in R but with the help of R∗ instead of
limits. Applied mathematicians will just tell you “I told you so”. But the main difference
between pure and applied mathematicians is that applied mathematicians are comfortable
with “fudge factors” and pure mathematicians are not. Pure mathematicians will tell you
there is no way they will accept the notion of infinitessimals until they have a good way of
understanding why it works. Applied mathematicians will tell you they will accept an
idea because it works, whether they understand it or not! The theory of hyperreals has
helped to blur the distinction between pure and applied mathematics. In any case, due to
Kurt Gödel’s undecideablity theorem which states that any logical mathematical system
will contain unproveable statements, this may be a vacuous point. But such theorems do
not diminish the level of precision present throughout mathematics coupled with the self
imposed burden of proof, which is as it always has been, superior in this regard compared
with other academic disciplines.

All New South Wales high school students will be familiar with all the abovementioned
number systems except C, H, & O. Complex numbers are used in many applications
in physics and engineering, and quaternion and octonion numbers are used in high level
physics such as quantum mechanical angular momentum and some aspects of the theory
of superstrings.

Z is an integral domain (has closure of addition & multiplication, commutivity and

associativity of addition and multiplication, distributivity of multiplication over addi-
tion, additive and multiplicative identities, additive inverses and a cancellation law) and
Q, R and C are fields (are integral domains and also have multiplicative inverses). Hence
these are the most useful of all number systems, and so most of mathematics only uses
Z, Q, R, & C and nothing else. Nevertheless, many other integral domains and fields are
used in pure mathematical research. Note that H and O are not fields.

There is a 1-1 correspondence between Z and Q which is especially interesting since one of
these is only an integral domain whereas the other is a field. Anything in 1-1 correspon-
dence with Z is called countable.

There is no 1-1 correspondence between Z and R. This is often articulated by saying that
R is uncountable. There is also no 1-1 correspondence between R and C.

Theorem. R is not countable, C is not ordered, H is not commutative, O is not associative

(Baez, 2001).

If we redefine Q by saying it is the set of solutions for x to all equations of the form
a1 x + a0 = 0 for integers a0 , a1 with a1 = 0 then we can see that we can generalise Q
nA, thej set of all algebraic numbers which are real solutions to√polynomial equations
to
j=0 aj x = 0 for integers a0 , . . . , an where an = 0. For example, 2, which is irrational,
is nevertheless algebraic. A is countable. So T := R\A, the set of all trancendental numbers,
is uncountable. e and π are transcendental. We will see in Lecture 29 that all polynomial
equations with coefficients in Z, Q, R, or C have solutions in C, so sometimes we see a
different definition of the set of algebraic numbers than the one given above, allowing it
to have complex elements. This is still countable. The trancendental compliment of this
set with respect to C is likewise uncountable. But this definition isn’t common because a
complex number is of the form a + ib for real a, b so a trancendental complex number can
be defined such that inclusively a or b ∈ T. Similarly an algebraic complex number can be
defined such that a and b ∈ A.

There is also a hypothesis called Georg Cantor’s Continuum Hypothesis which says that for
any superset X of Z which is also a subset of R, i.e., for any set X such that Z ⊂ X ⊂ R there
is either a 1-1 correspondence between Z and X or there is a 1-1 correspondence between
X and R. This was proved to be undecideable within the confines of Cantorian set theory
in 1963 by Paul Cohen (i.e., the truth or otherwise or this mathematical hypothesis not
only trancends the mathematics from which it emerged, but is also independent of it).
Nevertheless, logical systems can be constructed within which it can be proved and other
(more exotic) logical systems can be constructed within which it can be disproved!

So I guess we should not frown upon applied matematicians with too much scorn and
derision for their obsessive preponderance on fudge factors.
Examples of factorising over Z , R , C .

Example 1. Factorise x4 − 4 over Z, R, C.

x4 − 4 = (x2 − 2)(x2 + 2) ← completely factorised over Z

√ √
= (x − 2)(x + 2)(x2 + 2) ← completely factorised over R
√ √ √
= (x − 2)(x + 2)(x2 − (i 2)2 )
√ √ √ √
= (x − 2)(x + 2)(x − i 2)(x + i 2). ← completely factorised over C

Example 2. Factorise x2 + x + 1 over C

x2 + x + 1 = x2 + x + 1
4 + 3
4
= (x + 12 )2 + 3
4
 √ 2
= (x + 12 )2 − i 2 3
 √  √ 
= x + 12 − i 2 3 x + 1
2 + i 3
2 .

Example 3. Factorise 2x2 + 3x − 1 over the complex field.
 
2x2 + 3x − 1 = 2 x2 + 32 x − 12
 
= 2 x2 + 32 x + 169
− 12 − 16
9

 2 
= 2 x + 34 − 17 16
 √  √ 
=2 x+ 4 3− 17
x + 3+4 17 .

 Lecture 28
More Examples.

Example 1. Factorise P (x) = x3 − 4x2 + 14x − 20 completely over the complex field.

Test for factors of 20, i.e., ±1, ±2, ±4 ± 5, ±10, ±20.

P (1) = 0, P (2) = 0 & ∴ (x − 2) is a factor.

P (x) = (x − 2)(x2 − 2x + 10)

= (x − 2)(x2 − 2x + 1 + 9)
= (x − 2)((x − 3)2 + 32 )
= (x − 2)((x − 3)2 − (3i)2 )
= (x − 2)(x − 3 − 3i)(x − 1 + 3i) 
Example 2. Solve P (x) = x3 − 3x2 + 4x − 2 = 0 over C.

Test (±1, ±2) ⇒ p(1) = 0

& ∴ (x − 1)(x2 − 2x + 2) = 0
∴ (x − 1)(x2 − 2x + 1 + 1) = 0
∴ (x − 1)((x2 − 1)2 − i2 ) = 0
∴ (x − 1)(x − 1 − i)(x − 1 + i) = 0
& ∴ x = 1, 1 ± i 
Example 3. Solve P (x) = x3 − 1 = 0 over C.

P (1) = 0
∴ (x − 1)(x2 + x + 1) = 0
∴ (x − 1)(x2 + x + 14 + 34 = 0
   √3i 2 
1 2
∴ (x − 1) x + 2 − 2 =0
 √  √ 
∴ (x − 1) x + 12 − 23i x + 12 + 23i = 0
√
& ∴ x = 1, − 12 ± 3i
2 
 Lecture 29
The Fundamantal Theorem of Algebra. Every polynomial P (x) with complex
coefficients is such that P (α) = 0 for some complex α.

There are numerous proofs of this widely available in many texts, for example some such
proofs can be found in the Schaum Outline Complex Variables (Spiegel, 1981).*

From this we can deduce the

Corollary. Any polynomial of degree n, (n > 0) has exactly n zeros in the field of complex
numbers and hence exactly n linear factors.

Proof. P (x) is any polynomial (n > 0). By the fundamental theorem of algebra P (x) has
at least one zero, α1 ⇒ (x − α1 ) is a factor of P (x) ⇒ P (x) = (x − α1 )Q(x) for some
polynomial Q(x) over C but by the fundamental theorem of algebra, Q(x) has at least one
zero, α2 ⇒ (x−α2 ) is a factor of Q(x) ⇒ P (x) = (x−α1 )(x−α2 )R(x) for some polynomial
R(x) over C and this process continues until P (x) = k(x − α1 )(x − α2 ) · · · (x − αn ) for some
constant k (i.e., the coefficient of xn ) which shows that P (x) has n linear factors, and n
zeros.

It follows from this that P (x) cannot have more than n roots and P (x) is completely
reducible to linear factors.
n
Note that if a complex number α is a zero of a polynomial P (x) = j=0 aj xj with real
n n
coefficients, then P (α) = j=0 aj αj = j=0 aj αj = P (α) = 0 = 0 and so the complex
conjugate α of α is also a zero of P (x). Hence (x−α)(x−α) = (x2 −2(α)x+|α|2 ) ⇒ P (x)
can be broken down (i.e., is reducible) to a combination of real linear and quadratic factors.

*I highly recommend the Schaum Outlines for undergraduates who feel that the mountains of cellulose in
university libraries are intimidatingly quagmirish. It is quagmirish, but if they proceed to postgraduate
studies they would then more likely consult the American Mathematical Society’s Mathematical Reviews
and will then feel less intimidated, but until then, nothing brings stuff together quite like the Schaum
Outlines. They have outlines in many other subjects too which are equally as impressive.
Example. Show that x − (1 + i) is a factor of P (x) = x3 + 2x2 − 6x + 8. Hence factorise
P (x) over R and over C.

P (1 + i) = (1 + i)3 + 2(1 + i)2 − 6(1 + i) + 8

= 1 + 3i − 3 − i + 2(1 + 2i − 1) − 6 − 6i + 8
= −2 + 2i + 2 + 4i − 2 − 6 − 6i + 8
=0
Hence x − (1 + i) is a factor of P (x).

1 + i and 1 + i = 1 − i are zeros of P (x) since x − (1 + i) is a factor of P (x) and P (x) has
real coefficients & ∴ (x − (1 + i))(x − (1 − i)) = x2 − 2x + 2 is a factor of P (x). The constant
term in P (x) is 8 and the constant term in this factor is 2 and the leading coefficient of
P (x) is 1 & ∴ (x + 4) is a factor of P (x) (since 4 × 2 = 8). So

P (x) = (x + 4)(x2 − 2x + 2) over R

= (x + 4)(x − 1 − i)(x − 1 + i) over C.

 Lecture 30
Zeros of multiplicity.

Example 1. If P (x) = (x − 4)5 (x − 3), 4 is a zero of P (x) of multiplicity 5.

Zeros are 4, 4, 4, 4, 4, 3.

Theorem. If a is a zero of P (x) of multiplicity m, then P
(x) will have a as a zero of

multiplicity m − 1.

Proof. P (x) = (x − a)m Q(x) for some polynomial Q(x)

P
(x) = m(x − a)m−1 Q(x) + (x − a)m Q
(x)
= (x − a)m−1 (mQ(x) + (x − a)Q
(x))
= (x − a)m−1 R(x) for some polynomial R(x)

Example 2. If P (x) = x4 + 2x3 − 12x2 + 14x − 5 has a 3-fold zero, find the zeros of P (x).

P
(x) = 4x3 + 6x2 − 24x + 14 − 2 fold

P

(x) = 12x2 + 12x − 24 − 1 fold
= 12(x2 + x − 2)
= 12(x + 2)(x − 1)
P (1) = 0
∴ 1 is the 3 fold zero of P (x). ∴ the other zero is −5. ∴ P (x) = (x − 1)3 (x + 5). ∴ the
zeros are 1, 1, 1, 5.