Chapter 3

Complex Numbers and Infinite

Sequences

This chapter consists of two essentially unrelated parts. First a short look at the complex
numbers. After giving the definition, we will see different ways of expressing complex numbers,
perform basic arithmetic with them, and use De Moivre’s Theorem and the Binomial Theorem
to prove trigonometric identities. The bulk of the chapter is given over to the study of infinite
sequences. A sequence is just an ordered list of numbers, such as 1, 12 , 41 , 18 , . . .. In this sequence
the numbers are getting closer and closer to zero; we say that the limit of the sequence is zero.
Such sequences are called null sequences and we begin by looking at them before generalising to
sequences with arbitrary limits (convergent sequences) and sequences that do not have a limit
(divergent sequences). We finish with a look at infinite series. A series is just the sum (where
it exists) of a sequence, for example 1 + 12 + 41 + · · · . We look at some ways of determining
whether a series has a well-defined sum – that is, whether it is convergent or divergent.

3.1 The Complex Numbers

The set of complex numbers is C = {x + iy : x, y ∈ R} where i2 = −1. Note that for all x ∈ R,
x = x + i0 and so R ⊆ C. If z = x + iy then x is called the real part of z and y is called the
imaginary part of z.

Arithmetic with Complex Numbers

Let z1 = a + ib and z2 = c + id. We can define z1 + z2 , z1 − z2 and z1 z2 as follows:

z1 + z2 = (a + c) + i(b + d)
z1 − z2 = (a − c) + i(b − d)
z1 z2 = (a + ib)(c + id)
= ac + ibc + iad + i2 bd
= ac + i(bc + ad) + (−1)bd
= (ac − bd) + i(bc + ad)

Example 3.1.1 For the complex numbers z1 = 3 + 5i and z2 = −8 + 3i find (i) z1 + z2 (ii)
z1 − z2 ; (iii) z1 z2 ; (iv) z12 .
60 Proof & Structure Chapter 3: Complex Numbers and Sequences

(i) z1 + z2 = 3 + 5i + −8 + 3i = 3 − 8 + (5 + 3)i = −5 + 8i.

(ii) z1 − z2 = 3 + 5i − (−8 + 3i) = 3 − −8 + (5 − 3)i = 11 + 2i.
(iii) z1 z2 = (3 + 5i)(−8 + 3i) = −24 − 40i + 9i + 15i2 = −24 − 15 + (−40 + 9)i = −39 − 31i.
(iv) z12 = (3 + 5i)(3 + 5i) = 9 + 15i + 15i − 25 = −16 + 30i.

Exercise 3.1 For the complex numbers z1 = 4 − 3i and z2 = 1 + 6i find (i) z1 + z2 ; (ii) z1 − z2 ;
(iii) z1 z2 .

Division of Complex Numbers

It is not obvious how to divide one complex number by another. To do this we need the idea
of the complex conjugate of a complex number.

Definition Let z = a + ib with a, b ∈ R. The complex conjugate, z, of z, is given by

z = a − ib.

(Note: The complex conjugate is sometimes denoted z ∗ .) To see how the complex conjugate
can be used to divide complex numbers consider the product

zz = (a + ib)(a − ib)
= a2 − iab + iab − i2 b2
= a2 + b2 .

Now, a2 + b2 ∈ R and, provided z = 0, it is nonzero. Thus to evaluate zz21 (where z1 , z2 ∈ C, z2 =

0), we multiply the numerator and denominator by z2 , which will leave the denominator as a
real number. Consequently it is easy to express zz21 as a sum of real and imaginary parts, as the
following example illustrates.

z1
Example 3.1.2 For the complex numbers z1 = 3 + 5i and z2 = −8 + 3i find z2
.
z1 z1 z2
=
z2 z2 z2
(3 + 5i) (−8 − 3i)
=
(−8 + 3i) (−8 − 3i)
−24 − 40i − 9i − 15i2
=
82 + 32
−9 − 49i
=
73
9 49
= − − i.
73 73
z1
Exercise 3.2 For the complex numbers z1 = 4 − 3i and z2 = 1 + 6i find z2
.

The polar form of a complex number

Given a complex number z = x + iy we can express it as the point (x, y) in the plane. That
is, the x coordinate is the real part of z and the y coordinate is the imaginary part of z. This
pictorial representation of the complex numbers is called an Argand diagram.
3.1 The Complex Numbers Proof & Structure 61

For example, the complex numbers z1 = 3 + 5i and z2 = −8 + 3i would appear as below on the
Argand diagram. y
z1
5

z2 4
3
2
1

−8 −7 −6 −5 −4 −3 −2 −1 1 2 3 4 5
x
−1

Consider the line segment joining the point representing the number z = x + iy to the origin
(0, 0). Using Pythagoras’s Theorem,
√ the length r of this line segment is given by r = x2 + y 2.
(Note that this is the same as zz.)
Let the angle measured anticlockwise from the x-axis to the line segment be θ.
y

r r sin θ

θ
x
r cos θ
Using the definitions of the trigonometric functions sin and cos, we have:
x = r cos θ and y = r sin θ.
Thus if we know the values of r and θ we can determine z uniquely since

z = r(cos θ + i sin θ).

This is called the polar form of z. The length r is called the modulus of z, and is denoted
by |z|, and θ is called the argument of z and is denoted by arg z.1
We have already seen how to determine the value of r from x and y. To determine θ, note that,
if x = 0, then
y r sin θ
= = tan θ.
x r cos θ
Thus θ = tan−1 ( xy ). However, since tan θ = tan(θ + π) for any θ, there are two possibilities for
θ in the interval [0, 2π) and so we have to choose the one which is in the same quadrant as z.
What if x = 0? Then z = iy, so z lies on the y-axis (and r = |y|). If y > 0, then we have
θ = π2 . If y < 0, then θ = 3π
2
. If y = 0, then z = 0, so r = 0 and we could pick any value of θ.
We have the convention that θ = 0.
Note: It is important that the argument θ be an angle in the range 0 ≤ θ < 2π. This is the
standard way of writing complex numbers in polar form and you would lose marks on an
exam for giving an argument θ which was not in this range.
1
Although most current meanings of argument indicate a disagreement, the Latin root arguere was closer to
such present meanings as declare or prove. When we talk about the argument of a function, we are declaring a
value for the independent variable of the function, in this case the function f (θ) = r(cos θ + i sin θ).
62 Proof & Structure Chapter 3: Complex Numbers and Sequences

Most of the time, the argument will not be a nice fraction of π, and you will need to use a
calculator to get an approximation to it. However, there are some important angles whose
tangents you are expected to know. The best way to learn them is by drawing a couple of
triangles.

π
3
2 2
1 1
π π
6 4

3 1
You also need to remember that tan(θ) = tan(π + θ) and − tan(θ) = tan(π − θ), for any angle
θ. You can then construct the following table:

tan(θ) Possibilities for θ with 0 ≤ θ < 2π

π 5π
1 4
, 4
3π 7π
−1 4
, 4
√ π 4π
3 3
, 3
√ 2π 5π
− 3 3
, 3
√1 π 7π
, 6
3 6
1 5π 11π
− 3
√
6
, 6
√
Example 3.1.3 Find the modulus and argument of (i) 2 − 2i; (ii) 2 3 + 2i; (iii) −8 + 6i.

In each case we denote the argument of the complex number by θ.

√ √
(i) |2 − 2i| = 22 + (−2)2 = 8 = 2 2.
tan θ = −22
= −1. Thus θ = 3π 4
or 7π
4
, Since 2 − 2i is in the bottom right hand quadrant it
follows that θ = 7π
4
. That is, arg(2 − 2i) = 7π
4
. In particular we have
√
2 − 2i = 2 2 cos 7π 4
+ i sin 7π
4
.
√ √ √
(ii) |2 3 + 2i| = (2 3)2 + 22 = 16 = 4.
2
√
tan θ = √ =√1 . Thus θ = π
or 7π . Since 2 3 + 2i is in the top right hand quadrant it
2 3 3 6
√ 6
follows that θ = π6 . π
That is, arg(2 3 + 2i) = 6 . In particular we have
√
2 3 + 2i = 4 cos π6 + i sin π6 .
√
(iii) | − 8 + 6i| = (8)2 + 62 = 100 = 10.
6
tan θ = −8 = − 34 . Thus θ ≈ 2.498 or 5.640. Since −8 + 6i is in the top left hand quadrant it
follows that θ ≈ 2.498. That is, arg(−8 + 6i) ≈ 2.498. In particular we have

−8 + 6i ≈ 10 (cos 2.498 + i sin 2.498) .

Exercise 3.3 Find the modulus and argument of (i) 1 + i; (ii) 2 − 4i.
3.1 The Complex Numbers Proof & Structure 63

De Moivre’s Theorem
We now consider finding powers of complex numbers. Let z = r(cos θ + i sin θ). Using the stan-
dard rules for indices we have z t = r t (cos θ + i sin θ)t for all real numbers t. Thus to determine
z t we need to be able to find (cos θ + i sin θ)t .

In general we have the following result.

Theorem 3.1.4 (De Moivre’s Theorem) For all θ ∈ R and all n ∈ N

(cos θ + i sin θ)n = cos nθ + i sin nθ.

Exercise 3.4 Use induction to prove De Moivre’s Theorem. You will also need the trigono-
metric identities:
cos(A + B) = cos A cos B − sin A sin B;
sin(A + B) = sin A cos B + cos A sin B.

In conjunction with the Binomial Theorem for expanding brackets, De Moivre’s Theorem can
be used for finding expressions for cos nθ and sin nθ, where n ∈ N, in terms of cos θ and sin θ.
This is illustrated in the following example.

Example 3.1.5 To find expressions for cos 5θ and sin 5θ in terms of cos θ and sin θ we consider
(cos θ + i sin θ)5 in two different ways. Using the Binomial Theorem we have

(cos θ + i sin θ)5 = cos5 θ + 5i cos4 θ sin θ − 10 cos3 θ sin2 θ

−10i cos2 θ sin3 θ + 5 cos θ sin4 θ + i sin5 θ.

Using De Moivre’s Theorem

(cos θ + i sin θ)5 = cos 5θ + i sin 5θ.

Comparing the real parts of the two expressions for (cos θ + i sin θ)5 gives

cos 5θ = cos5 θ − 10 cos3 θ sin2 θ + 5 cos θ sin4 θ.

Using sin2 θ = 1 − cos2 θ, we can take this further and express cos 5θ in terms of cos θ as follows:

cos 5θ = cos5 θ − 10 cos3 θ(1 − cos2 θ) + 5 cos θ(1 − cos2 θ)2

= cos5 θ − 10 cos3 θ + 10 cos5 θ + 5 cos θ(1 − 2 cos2 θ + cos4 θ)
= 11 cos5 θ − 10 cos3 θ + 5 cos θ − 10 cos3 θ + 5 cos5 θ
= 16 cos5 θ − 20 cos3 θ + 5 cos θ.

Comparing the imaginary parts of the two expressions for (cos θ + i sin θ)5 gives

sin 5θ = 5 cos4 θ sin θ − 10 cos2 θ sin3 θ + sin5 θ.

Exercise 3.5 Express sin 5θ in terms of sin θ.

Exercise 3.6 Find expressions for cos 4θ and sin 4θ in terms of cos θ and sin θ. Hence express
sin θ
cos 4θ in terms of cos θ, and tan 4θ in terms of tan θ (using tan θ = cos θ
).
64 Proof & Structure Chapter 3: Complex Numbers and Sequences

De Moivre’s Theorem can be generalized to all real numbers. That is, for all x ∈ R
(cos θ + i sin θ)x = cos xθ + i sin xθ.
This form of De Moivre’s Theorem can be used to find the n distinct nth roots of a nonzero
complex number c, which is equivalent to finding all complex solutions z to the equation
z n − c = 0, as follows. Note first that for any integer k, we have cos θ = cos(θ + 2kπ) and
sin θ = sin(θ + 2kπ). Thus, for each integer k,
√
n 1
cos θ + i sin θ = (cos(θ + 2kπ) + i sin(θ + 2kπ)) n
= cos θ+2kπ
n
+ i sin θ+2kπ
n
.
Taking k = 0, 1, 2, . . . , n − 1 gives the n distinct nth roots. If z ∈ C and |z| = 1, then we can
find the nth roots of z√= r(cos θ + i sin θ) by finding the nth roots of (cos θ + i sin θ) and then
multiplying these by n r.√(Since r ≥ 0, r will have exactly one non-negative real nth root, and
this is what is meant by n r.)
√
3
Example 3.1.6 Let z = 2
+ 12 i. Then z = cos π6 + i sin π6 . Thus
√ π
1
3
z = cos 6
+ 2kπ + i sin π6 + 2kπ 3

π π
6
+ 2kπ + 2kπ
= cos + i sin 6
3 3
π 2kπ π 2kπ
= cos 18
+ 3 + i sin 18 + 3
(12k+1)π (12k+1)π
= cos 18
+ i sin 18
.
π π
Taking k = 0, 1, 2 we get the following cube roots of z: cos 18 + i sin 18 , cos 13π
18
+ i sin 13π
18
and
25π 25π
cos 18 + i sin 18 .

Complex Functions and Power Series

We have already seen how powers of a complex number can be easily defined. We now consider
how to define other complex functions.
Recall the following power series expansions for the functions ex , sin x and cos x.
2 3 4 5
ex = 1 + x + x2! + x3! + x4! + x5! + . . .
x3 x5 x7
sin x = x − 3!
+ 5!
− 7!
+ ...
x2 x4 x6
cos x = 1 − 2!
+ 4!
− 6!
+ ...
Let z = x + iy, with x, y ∈ R. Using the standard rules for indices we have
ez = ex+iy = ex eiy .
Now ex is a real number, so to evaluate ez we have to determine the value of eiy . Using the
above power series to calculate eiy we get
(iy)2 3 4 5
eiy = 1 + iy + 2!
+ (iy)
3!
+ (iy)
4!
+ (iy)
5!
+ ...
2 3 4 5
= 1 + iy − y2! − i y3! + y4! + i y5! + . . .
2 4 3 5
= (1 − y2! + y4! − . . .) + i(y − y3! + y5! − . . .)
= cos y + i sin y
3.1 The Complex Numbers Proof & Structure 65

Note I have not justified the fact that rearranging an infinite number of terms like this does
not change the infinite sum. This does not hold in general. But it is OK to do this when the
series converges to a finite value – in this case it does because it is the series for eiy .
Thus we get ez = ex+iy = ex (cos y + i sin y).
One consequence of this is the following succinct proof of De Moivre’s Theorem:
(cos θ + i sin θ)n = (eiθ )n = einθ = cos nθ + i sin nθ.
Putting x = 0 and y = π gives the following beautiful identity, seen by many as one of the most
extraordinary facts in mathematics because it involves three seemingly unrelated mathematical
concepts:
eiπ = −1.

From N to C
We have seen that C has some advantages over R, such as the ability to take nth √
roots, but
unfortunately we lose the concept of <, because how do you compare numbers like −1 with
for example 12 ? This means we lose the completeness axiom. Do we gain anything? In this
section, we aim to provide a partial answer.

The sets Z, Q, R and C all arise from trying to solve polynomial equations. Consider the
following list of equations, noting that all the coefficients are natural numbers:

2x = 8; (3.1)
2x + 8 = 0; (3.2)
2x = 7; (3.3)
x2 = 7; (3.4)
x2 + 5 = 1. (3.5)
Now equation (3.1) has a solution which is a natural number: x = 4. The solution to equation
(3.2)is x = −4 which is not a natural number – we need the concept of negative numbers, so we
have to work with the integers Z. For equation (3.3) the integers are not enough, we now have
to consider fractions (that is, the rational numbers Q) to obtain a solution (x = 72 ). In fact,
rational numbers are enough to solve any linear equation with rational coefficients. However,
with equation (3.4), which is quadratic, even rational numbers aren’t enough. We have√to
widen our net to the set R of real numbers - the only solutions are the irrational numbers ± 7.
Finally, equation 3.5 has no real solutions. We get x2 = −4 = 4(−1) and hence x = ±2i.
Every quadratic equation has two (not necessarily distinct) solutions if we are allowed to use
complex numbers. So all the sets of numbers that we have met come about from trying to
solve equations where the coefficients are natural numbers. A question you might ask is ‘what
happens with cubic equations?’ We know that we need to introduce complex numbers to deal
with quadratics, when rational numbers were good enough for linear equations. The incredible
answer is that complex numbers are enough to provide solutions to ALL polynomials. The
proof of this theorem uses results which are too advanced for a first year module, but I have
included a sketch proof in Appendix 3 if you would like to see some justification. We therefore
finish the section by stating:
Theorem 3.1.7 (The Fundamental Theorem of Algebra) Let P be a polynomial of degree
n ≥ 1 with coefficients in C. Then the equation P (x) = 0 has n (not necessarily distinct)
solutions in C.
66 Proof & Structure Chapter 3: Complex Numbers and Sequences

3.2 Sequences
In this section we are going to establish some ground rules for working with infinite sequences
of numbers. We’ll start with some examples and motivation – why should we be interested in
sequences, and what types of sequence are there?

Example 3.2.1 Here are some examples of what we might intuitively call sequences.

(a) 1, 0, 6, 6.
(b) 1, 2, 3, . . .
(c) 1, −1, 1, −1, 1, . . .
(d) 3, 1, 4, 1, 5, 9, . . .
(e) 1, 21 , 14 , 18 , 16
1
...
(f) 1, 12 , 13 , 41 , 51 . . .
(g) 3, 8, 43, 11, 540, 0.1, . . .

All these examples involve real numbers. In this course we will only be interested in real
sequences, that is, sequences consisting of real numbers. The order of the numbers is important
as well. 5, 4, 3, 2, 1 is a different sequence from 1, 2, 3, 4, 5. An informal definition, then, is that
a (real) sequence is an ordered list of (real) numbers.
In Example 3.2.1(a), the list of numbers terminates. We will not consider such examples to be
sequences (call them lists if you wish). The other examples have the mysterious ‘. . .’, indicating
that they continue forever. Of course, in the very first exercises of this module, we saw that
writing 1, 2, 3 . . . is not enough to define a sequence. Who knows whether the next number is
4, 5, 42 or anything else you like? The only way to make the whole thing hold water is to give
a proper definition as follows:

Definition A sequence is a list of real numbers an (n ∈ Z+ ), denoted {an }. The number an

is known as the nth term of the sequence.
A sequence can be defined in several ways. We can give an explicit formula for an in terms of
n, such as an = (−1)n+1 (this gives the sequence in Example 3.2.1(c)). Or we can describe an
in words, for example ‘an is the nth digit in the decimal expansion of π’.

Example 3.2.2 The sequence { n1 } starts 1, 21 , 31 , 14 , . . ..

Example 3.2.3 The sequence {an } defined by

an = n2 , n = 1, 2, . . .

begins 1, 4, 9, 16, . . ..

Example 3.2.4 The sequence {an } defined by

a1 = 1, a2 = 1, an = an−1 + an−2 , n = 3, 4 . . .

begins 1, 1, 2, 3, 5 . . ..
3.2 Sequences Proof & Structure 67

Exercise 3.7 Calculate the first eight terms of each of the following sequences.

(a) {2n };
nπ
(b) {cos 2
};
(c) an = (−1)n n . . .;
n n n
(d) an = 5
+ 3
+ 1
, n = 1, 2, . . .;
(e) an = the smallest positive integer with exactly n factors, n = 1, 2, . . ..

Exercise 3.8 Write down possible formulae for the sequences in Example 3.2.1(b) – (f).

Sometimes it is more convenient to start a sequence not at n = 1, but at n = 0, or some

larger n. For example if we want to work with terms n21−n , we cannot start at n = 1 because
this would involve division by zero. So we use the notation {an }n≥2 to represent a2 , a3 , a4 , . . ..
(Sometimes the notation {an }∞n=2 is used to mean the same thing.)
Note that we do not lose very much by insisting that sequences be infinite. After all if we want
to work with something like 1, 0, 6, 6, we can just consider the sequence {an } where a1 = 1,
a2 = 0, a3 = 6, a4 = 6 and an = 0 for all n ≥ 5.

Some Terminology
Many sequences that crop up in mathematics have the property that, as n increases, the terms
either always increase or always decrease. For example the sequence { n1 } (whose first few terms
are 1, 12 , 31 , . . .) is decreasing. If a sequence always does the same thing (so terms always increase
or always decrease), we call it monotonic. Here are the formal definitions.

Definition A sequence {an } is

increasing if an ≤ an+1 for all n ≥ 1;

strictly increasing if an < an+1 for all n ≥ 1;
decreasing if an ≥ an+1 for all n ≥ 1;
strictly decreasing if an > an+1 for all n ≥ 1;
monotonic if {an } is either increasing or decreasing;
strictly monotonic if {an } is either strictly increasing or strictly decreasing; and
constant if an = an+1 for all n ≥ 1;

Notice that a constant sequence, rather strangely, is both increasing and decreasing. This is
an unfortunate consequence of the definitions, but we really need those definitions because of
examples like the following.

Example 3.2.5 Consider the sequence 1, 1, 1, 4, 4, 4, 4, 9, 9, 9, 9, 9, 9, 9, 16, 16, . . .. In this se-

quence an is the largest square number less than or equal√ to n.√We might be interested in such
a sequence because it tells us that the integer part of n is an . The terms of the sequence
are clearly rising and so we would want to include it in our definition of an increasing sequence.
However it is not strictly increasing, and that is the distinction.
68 Proof & Structure Chapter 3: Complex Numbers and Sequences

Example 3.2.6 The sequence {n − 3} is strictly monotonic increasing. To prove this, note
that an+1 − an = ((n + 1) − 3) − (n − 3) = 1 > 0. Since an+1 − an is strictly positive, for all n,
the sequence is strictly increasing.

Example 3.2.7 To determine whether the sequence { n1 } is monotonic, again we can compare
adjacent terms. an+1 − an = n+1 1
− n1 = n−(n+1)
n(n+1)
1
= − n(n+1) . Since an+1 − an < 0 for all n ≥ 1,
1
the sequence { n } is strictly decreasing.

In these examples, we calculated an+1 − an . Let’s call this λ. If for all n ≥ 1, λ ≥ 0, then {an }
is increasing. If λ > 0, then {an } is strictly increasing. If λ ≤ 0, then {an } is decreasing. And
if λ < 0, then {an } is strictly decreasing.

Exercise 3.9 Show that { n−1

n
} is monotonic and state whether it is increasing or decreasing.

Exercise 3.10 Show that {an } is monotonic, where an is the number of prime numbers less
than or equal to n.

There is another approach which works as long as all the terms of the sequence are positive,
and that is to calculate the ratio an+1
an
. If an+1
an
≥ 1 for all n, then the sequence is increasing and
if an+1
an
≤ 1 for all n then the sequence is decreasing. This approach can sometimes be quicker,
as the next example shows.

Example 3.2.8 In the sequence { n1 } each term is positive. Now

1
an+1 n+1 n
= 1 = < 1.
an n
n+1

Therefore an+1 < an and { n1 } is strictly decreasing.

Of course not all sequences are monotonic. To prove that a sequence is not monotonic, we need
to show it is neither increasing nor decreasing. We do this by giving a counterexample to each
possibility.

Example 3.2.9 Consider the sequence {(−2)n }. Now a1 = −2, a2 = 4, a3 = −8. The fact
that a1 < a2 means that {an } is not decreasing. The fact that a2 > a3 means {an } is not
increasing. Therefore {(−2)n } is not monotonic.

Exercise 3.11 Show that the following sequences {an } are not monotonic.

(a) an = (−1)n , n = 1, 2, . . .;
nπ
(b) an = cos 2
, n = 1, 2, . . .;
(c) an is the number of factors of n, n = 1, 2, . . ..
2
Exercise 3.12 Determine whether the sequence { 2nn } is monotonic.

Sometimes sequences may start off with uncharacteristic behaviour but settle down to have a
particular property. If this is the case, in other words, if we ignore an initial finite number of
terms, then we say the sequence eventually has that property. For example, the terms of the
sequence { n1 } are eventually less than 0.001. (This is because for all n > 1000, n1 < 0.001.)
3.3 Null Sequences Proof & Structure 69

Exercise 3.13 For a positive integer n, let s(n) be the sum of the digits of n. So for example
s(185) = 1 + 8 + 5 = 14. Now consider the sequence {n − s(n)}. Show that the terms of
{n − s(n)} are eventually positive.

Example 3.2.10 Consider the sequence an = n2 − 8n + 15, n = 1, 2, . . .. This sequence

begins 8, 3, 0, −1, . . .. So it looks like it is decreasing. However if you think of the graph
of a quadratic function you won’t necessarily be fooled by this. By factorising we see that
an = n2 − 8n + 15 = (n − 3)(n − 5). So an is positive for n < 3, zero for n = 3, negative for
n = 4, zero again at n = 5, and positive thereafter. In fact, calculating an+1 − an gives

an+1 − an = (n + 1)2 − 8(n + 1) + 15 − (n2 − 8n + 15) = n2 + 2n + 1 − 8n − 8 − n2 + 8n = 2n − 7.

If n ≥ 4, then 2n − 7 > 0. Therefore, for n ≥ 4, {an } is strictly increasing. In summary, {an }

is eventually strictly increasing. (We can also say that {an } is eventually positive.)
2
Exercise 3.14 Show that the sequence { 2nn } is eventually monotonic.

Why study sequences?

There are two good reasons to be interested in sequences (apart from the fact that they are in-
herently interesting of course). Firstly, sequences are one way of properly defining real numbers.
We can define natural numbers (using Peano’s axioms), and from them integers and rational
numbers. But real numbers we so far only know as infinite decimals, which are not easy to get
a handle on. We can get round this if we define a sequence of successive approximations. So
for example, to define π, we use the sequence 3, 3.1, 3.14, 3.142, 3.1416, 3.14159, . . .. So an is π
to n − 1 decimal places. Then we can define π as the ‘limit’ of this sequence, a sort of a∞ .
The idea of limit will be defined properly as we go along, and we’ll discuss what properties a
sequence must have before the concept even makes sense.
The second reason we might want to study sequences is to understand series. These are infinite
sums, such as 1 + 21 + 41 + · · · , or the series we have encountered for sin x and cos x. Given
a series ∞ i=1 ti , we can define a sequence of partial sums, {
n
i=1 ti }. The sum of the series,
where it exists, will be the limit of this sequence of partial sums.
Clearly the idea of the limit of a sequence is important, and defining and understanding this is
our eventual goal. Before we get there we will look at a special kind of sequence known as a
null sequence. Intuitively these are sequences whose limit is zero.

3.3 Null Sequences

A null sequence, informally, is one whose terms get closer and closer to zero, something like
{ n1 }. To make this precise, we introduce the letter ε (epsilon), which usually stands for a small
positive number in mathematics. The idea then is to say that however small an epsilon we
pick, at some stage all the terms in the sequence are smaller in size. We need to consider the
modulus of the terms, because for example −10000 < 0.0001.

Definition A sequence {an } is a null sequence if for all ε > 0 there is N ∈ Z+ such that
|an | < ε for all n ≥ N.

Exercise 3.15 Write this definition out using ∀, ∃, R+ and Z+ .

70 Proof & Structure Chapter 3: Complex Numbers and Sequences

Let’s see this definition in action.

Example 3.3.1 The sequence { n1 } is a null sequence. To prove this, let ε > 0. Our task is to
find N for which n1 < ε for all n ≥ N. Now n1 < ε if and only if n > 1ε . So suppose we let N
be the smallest positive integer such that N > 1ε . Then for all n ≥ N, we have n ≥ N > 1ε and
hence n1 < ε. So the recipe for N is ‘the smallest positive integer greater than 1ε ’. (For example
if ε = 0.12, then N = 9.) This shows that { n1 } is indeed a null sequence.

Example 3.3.2 The sequence {e−n sin n} is a null sequence. Remember that we are only
concerned with |an |, so the fact that sin can take positive and negative values doesn’t matter
as much as the fact that | sin x| ≤ 1 for all x ∈ R. Thus |e−n sin n| ≤ e−n . Now given ε > 0,
e−n < ε if and only if −n < ln ε, which is if and only if n > − ln ε. So take N to be the smallest
positive integer greater than − ln ε. (For example if ε = 0.04, then N = 4.) Working backwards
if n ≥ N, then n > − ln ε, so −n < ln ε and hence e−n < ln ε and so

|an | = |e−n sin n| ≤ e−n < ε.

Therefore {an } is a null sequence.

When you are showing that sequences are null, by the way, it is not necessary to give an example
of calculating N from a particular value of ε as I did with ε = 0.04 above, you just need a
general recipe for N in terms of ε.
Note that because we are working with |an | in the definition, we can say that a sequence {an }
is null if and only if the sequence {|an |} is null.
Of course not all sequences are null, as the following examples show.

Example 3.3.3 The sequence {(−1)n } is not null. To prove this we need to show that there
is some ε > 0 such that for all N ∈ Z+ , there is some n ≥ N for which |an | ≥ ε. (You should
check that this really is the negation of the definition of null sequence.) This means we have to
find a value of ε for which no matter how far along the sequence we go, we still find values of
|an | which are greater than ε. Since |an | = 1 for all n, it would be pointless looking for ε ≥ 1.
But if we pick, for example ε = 12 , then for all n, |an | = |(−1)n | = 1 > ε. Therefore no possible
choice of N makes |an | < ε. Hence {(−1)n } is not null.

1
if n is not prime
n
Example 3.3.4 The sequence defined by an = is not null. Any ε < 2
2 if n is prime
will do for us here. Suppose for example ε = 1 and there was some N for which |an | < ε for
all n ≥ N. Then let p be the first prime number larger than N. (We know p exists because
there are infinitely many primes.) Now ap = 2 > ε, a contradiction. Therefore {an } is not a
null sequence.

n
Exercise 3.16 (a) Show that { (−1)
24
} is not null.

(b) Show that { nn!3 } is not null. (Hint: first show the sequence is increasing for n ≥ 3.)
3.3 Null Sequences Proof & Structure 71

Combination Laws
In this section we will look at recipes for making new null sequences from ones we already have.
Theorem 3.3.5 Suppose {an } and {bn } are null sequences and c is any real number. Then
the following are also null sequences:
(a) {can } (Scalar Multiplication);
(b) {an + bn } (Addition);
(c) {an bn } (Products).
Example 3.3.6 Given that { n1 } is null, we can deduce that { n12 }, { n42 }, { n1 + n12 } and { n13 } are
all null sequences. So is { 2n+1
n2
}, since 2n+1
n2
= n2 + n12 .
In the proof of Theorem 3.3.5 we use the Triangle Inequality. Recall that this says for any real
numbers x and y, |x + y| ≤ |x| + |y|.
Proof (a) We want to prove {can } is a null sequence. This is trivially true for c = 0, so assume
c = 0. Let ε > 0. Given that {an } is null, we know that for all ε̂ > 0, there exists N ∈ Z+ such
ε ε
that |an | < ε̂ whenever n ≥ N. In particular, setting ε̂ = |c| , |an | < |c| whenever n ≥ N. That
is, |can | < ε for all n ≥ N. Hence {can } is null.
(b) Let ε > 0. Given that {an } is null, we know that there exists N1 ∈ Z+ such that |an | < 12 ε
whenever n ≥ N1 . Similarly, since {bn } is null, there exists N2 ∈ Z+ such that |bn | < 21 ε
whenever n ≥ N2 . Choose N = max(N1 , N2 ). Then for n ≥ N, both |an | < 21 ε and |bn | < 21 ε.
Hence, by the Triangle Inequality, |an + bn | ≤ |an | + |bn | < ε. Therefore {an + bn } is null.
√
(c) Let ε > 0. Given that {an } is null, we know that there exists N1 ∈ Z+ such that |an | < √ε
whenever n ≥ N1 . Similarly, since {bn } is null, there exists N2 ∈ Z+ such √ that |bn | < √ ε
whenever n ≥ N2 . Choose N = max(N1 , N2 ). Then for n ≥ N, both |an | < ε and |bn | < ε.
Hence |an bn | = |an | · |bn | < ε. Therefore {an bn } is null.
Lemma 3.3.7 Powers Suppose {an } is a null sequence with an ≥ 0 for all n. Then for all
k > 0, {akn } is also a null sequence.
Note that we need to specify an ≥ 0 to ensure that akn is always defined. For a ≥ 0 and k > 0,
th
ak is defined to be the unique non-negative k power of a. But if a < 0, then there may be no
real number ak . For example (−1)1/2 is not a real number.

Proof Let ε > 0. Since {an } is null, we know that there exists N ∈ Z+ such that |an | < ε1/k
whenever n ≥ N. This means |akn | < ε for all n ≥ N and hence {akn } is null.
Example 3.3.8 Given that { n1 } is a null sequence, so is { √1n }, by setting k = 1
2
in the powers
law.
Exercise 3.17 Use the Combination Laws (Theorem 3.3.5 and Lemma 3.3.7) and the fact that
1 2
n
is null to show that 2n −3n+7
n3
is null.

Lemma 3.3.9 The Sandwich Lemma Suppose {an } is a null sequence and for all n, |bn | ≤
|an |. Then {bn } is a null sequence.
The idea is that the sequence {|bn |} is sandwiched between zero and {|an |}, so is forced to
converge to zero.
72 Proof & Structure Chapter 3: Complex Numbers and Sequences

Proof Suppose {an } is a null sequence and for all n, |bn | ≤ |an |. Let ε > 0. By definition of
a null sequence, there is N ∈ Z+ such that |an | < ε for all n ≥ N. But now |bn | ≤ |an | < ε.
Therefore {bn } is null.
1
Example 3.3.10 Since n + 7 > n, dividing through gives n+7
< n1 . Hence | n+7
1
| < | n1 | and so,
1
by the Sandwich Lemma, { n+7 } is null.
1
Example 3.3.11 Consider the sequence { n−1 }n≥2 . We can’t directly compare with { n1 } as
1
n−1
> n1 . However, using the scalar multiplication law, the fact that { n1 } is null implies { n2 } is
null. For n ≥ 2, n − 1 ≥ n2 . Hence n−1
1
≤ n2 . Therefore, by the Sandwich Lemma, { n−1 1
} is null.

Exercise 3.18 Show that { n23−n }n≥2 is null. You may use any of the Combination Laws, but
the only null sequence you can assume is { n1 }.

Note that when we are applying the Sandwich Lemma it is enough to check that |bn | ≤ |an |
eventually – that is, the first few terms do not matter, because a null sequence is still null if a
finite number of terms are added, removed or altered.

A List of Simple Null Sequences

We have now found several ways to prove that a sequence is a null sequence by comparing it
in some way with a sequence we already know to be null. To make this easier, in this section
we will establish a list of simple null sequences that can be used for this purpose.

Theorem 3.3.12 Let k, x ∈ R with k > 0. The following sequences are null sequences.
(a) { n1k };
(b) {xn } whenever |x| < 1;
(c) { n!1 };
(d) {xn nk }, whenever |x| < 1.
n
(e) { xn! };
k
(f ) { nn! };

The proof of Theorem 3.3.12 uses several of the Combination Laws that we derived in the
previous section. We can prove that all these sequences are null just using those laws and the
fact that { n1 } is null.

Proof (a) This is just an application of the powers law. In the null sequence { n1 }, 1
n
≥ 0 for
k
all n. Therefore, since k > 0 the powers law implies { n1 } = { n1k } is null.
(b) This one is slightly more tricky. We have to use the Binomial Theorem. Firstly, if x = 0
the sequence is trivially null. And in fact we may assume x > 0 because {xn } is null if and only
if {|xn |} is null, so we can replace x by |x| without any difficulty. We already know |x| < 1,
so in summary we are assuming 0 < x < 1. This means x1 > 1. So we can write x1 = 1 + y
for some y > 0. Now, by the Binomial Theorem, (1 + y)n = 1 + ny + n2 y 2 + · · · . Because all
n
the terms are non-negative, this allows us to conclude x1 = (1 + y)n > ny. Hence xn < ny 1
.
1 1
Since { n } is null, the law of scalar multiplication shows that { ny } is also null. Therefore, by
the Sandwich Lemma, {xn } is null.
3.4 Convergent Sequences Proof & Structure 73

(c) { n!1 } is null by the Sandwich Lemma, since 1

n!
≤ 1
n
for all n.
(d) As in part (b) we only need to worry about 0 < x < 1. Suppose we can show that {un n}
is null for any 0 < u < 1. Then set u = x1/k . Since 0 < x < 1 it follows that 0 < u < 1. Now
xn nk = (un n)k . By the powers law, {xn nk } is null. We are now left with the task of showing
that for 0 < u < 1, {un n} is null. We try the same trick as in (b). Set u1 = 1 + v. Now
(1 + v)n = 1 + nv + n2 v 2 + · · · . This time, because there is a factor of n in the sequence, we
will use the n2 v 2 part of the expression. So we get (1 + v)n > n(n − 1)v 2 /2 (for n ≥ 2). Hence
n 2n 2 1
un n = < = · .
(1 + v)n n(n − 1)v 2 v2 n − 1
1
Since { n−1 } is null, we can use scalar multiplication to deduce that { v22 · n−1
1
} is null. Now the
n
Sandwich Lemma shows that {u n} is null.
n n
(e) Intuitively { xn! } because xn! = nx · n−1
x
· · · x2 x1 . At some stage you reach an n which is bigger
than x and thereafter all the terms are small and cancel out the initial possibly large terms like
x
2
and x1 . So let m be the smallest integer such that m + 1 > x, and suppose n > m + 1. Then
xn x x x x xx
= · ··· · ···
n! n n−1 m+1 m 21
x xm
≤ · 1···1 ·
n m!
1 xm+1
≤ · .
n m!
xm+1 n
Notice that m!
doesn’t depend on n. Therefore, by scalar multiplication, { xn! } is null.

Technical Note You may be worried about the terms where n ≤ m + 1. We could either
deal with those by changing the scalar multiple so it was larger than any of the first m + 1
terms, or simply noting that a null sequence remains null if a finite number of terms are added
or removed. After all, it is what happens as n → ∞ that we are concerned about.
k
(f) The final sequence, { nn! } is actually easy to deal with given what we have just shown in (d)
k n
and (e). We just pick any nonzero |x| < 1 and write write nn! = (xn nk ) × (1/x) n!
. By (d), {xn nk }
n k
is null. By (e), { (1/x)
n!
} is null. Therefore the product { nn! } is also null.
Exercise 3.19 Use the Combination Laws and the list of basic null sequences to show that
4 2
the following sequences are null: (i) 4nn ; (ii) n 3+42
n
3
; (iii) (2n−1)! .

3.4 Convergent Sequences

In this section we make our discussion of sequences a bit more general. Null sequences are those
which converge to zero. However there are many interesting sequences which are not null, and
we want to investigate them too.
√
Example 3.4.1 Consider the following sequence: a1 = 16, an = an−1 for all n ≥ 2. You can
see the behaviour of this sequence by putting 16 into your calculator and repeatedly pressing
the square root button. With my calculator it gets so close to√1 that a32 is simply given as 1.
Of course we know it can never actually reach 1, because if x = 1, then x = 1. Formally,
n n
this sequence is {16(1/2 ) }. If instead we were to consider {16(1/2 ) − 1}, we would have a null
sequence. This leads to the following definition:
74 Proof & Structure Chapter 3: Complex Numbers and Sequences

Definition The sequence {an } converges to l if {an − l} is a null sequence. This is written
‘ lim an = l’ or ‘an → l as n → ∞’.
n→∞
If {an } converges to l then we say l is the limit of {an }, and refer to {an } as a convergent
sequence.
Often it is easier to work with the following equivalent definition:
Definition The sequence {an } converges to l if for all ε > 0 there is N ∈ Z+ such that
|an − l| < ε for all n ≥ N.
Note As with null sequences, a convergent sequence will still be convergent if a finite number
of terms are added, removed or altered.
To prove the next lemma we use the Triangle Inequality again.

Lemma 3.4.2 A convergent sequence has a unique limit.

Proof Suppose for a contradiction that {an } converges with limits l and m, where l = m.
Then choose ε < 21 |l − m|. From the definition of convergence, there are Nl and Nm such
that for all n ≥ Nl , |an − l| < ε and for all n ≥ Nm , |an − m| < ε. Pick any n for which
n ≥ max(Nl , Nm ). Then

|l − m| = |l − an + an − m| ≤ |l − an | + |an − m| < 2ε < |l − m|,

a contradiction. Therefore {an } cannot have more than one limit.

n
Example 3.4.3 Let’s prove that the sequence {161/2 } converges to 1. The algebra is slightly
more complicated but the principle is the same. Given ε > 0 we need to produce N ∈ Z+ such
n
that 161/2 − 1 < ε for all n ≥ N.
n
161/2 − 1 < ε
n
161/2 < 1 + ε
1
2n
log2 16 < log2 (1 + ε)
4
2n >
log2 (1 + ε)
4
n > log2
log2 (1 + ε)

4
So let N be the smallest positive integer greater than log2 log2 (1+ε)
. Then for n ≥ N we have
4 1/2n n
2n ≥ 2N > log (1+ε) and hence 16 − 1 < ε. Thus an < ε as required and {161/2 } converges
2
to 1. (If you want to find N for given ε on your calculator, remember that log2 x = ln x
ln 2
for any
positive x.)

1 2n+1
Exercise 3.20 Prove that: (i) lim (3 + n2
) = 3; and (ii) lim = 23 .
n→∞ n→∞ 3n

We finish this part by noting that all null sequences are convergent with limit 0, and all constant
sequences {c} for some c ∈ R are convergent with limit c.
3.4 Convergent Sequences Proof & Structure 75

Combining Convergent Sequences

In this section we generalise the laws about combining null sequences to cover convergent
sequences in general. There are two new ones; the reciprocal law, and the quotient law, which
do not apply to null sequences because we cannot divide by zero.

Theorem 3.4.4 Suppose lim an = l and lim bn = m. Let c be any real number. Then
n→∞ n→∞

(a) lim (can ) = cl (Scalar Multiplication);

n→∞

(b) lim (an + bn ) = l + m (Addition);

n→∞

(c) lim (an bn ) = lm (Product).

n→∞

1 1
(d) lim bn
= m
as long as m = 0 (Reciprocal).
n→∞

an l
(e) lim bn
= m
as long as m = 0 (Quotient).
n→∞

Note that the reciprocal law still makes sense if some of the bn are zero, as long as m = 0. This
is because at most finitely many bn are zero (see Exercise 3.21 below), and so we can remove
them from the sequence and the resulting sequence will converge to m1 .

Exercise 3.21 Prove that if lim bn = m = 0, then bn = 0 for only finitely many (possibly no)
n→∞
values of n.

These laws are useful because not only do they tell us that certain sequences converge, they
also tell us what the limits of those sequences are. Let’s do a couple of examples before we
prove the theorem.

5n3 −2n+1
Example 3.4.5 Prove that the sequence {an } given by an = (n2 +2)(3n+4)
is convergent and
find its limit.
To answer this we might go straight for the quotient law, but the numerator and denominator
sequences are not convergent; the values get arbitrarily large. The trick is to divide top and
bottom through by the ‘largest’ or ‘dominant’ term appearing in the numerator or denominator.
In this case we have a 5n3 which for large n will outweigh the other terms. So we divide through
by n3 to get
5n3 − 2n + 1 5 − n22 + n13
an = 2 = .
(n + 2)(3n + 4) (1 + n22 )(3 + n4 )
Notice that when we divided (n2 + 2)(3n + 4) by n3 we split the n3 into n2 × n, so we worked
2
out (n n+2)
2 · (3n+4)
n
. Now, lim 1 = 1 and lim n22 = 0, so by the addition law, lim (1 + n22 ) = 1.
n→∞ n→∞ n→∞
Similarly lim (3 + n4 ) = 3 and lim (5 − n22 + n13 ) = 5. Hence, applying the product and quotient
n→∞ n→∞
5
laws, we get lim an = 1×3
= 35 .
n→∞

We can use more economical working if we wish, as the next example shows.
76 Proof & Structure Chapter 3: Complex Numbers and Sequences

Example 3.4.6 Prove that the sequence {an } given by an = (3−4n)(n+1)

2n2 +7
is convergent and find
its limit.
In this case the dominant term is 2n2 , so we divide through by n2 .
(3 − 4n)(n + 1) ( n3 − 4)(1 + n1 )
an = = .
2n2 + 7 2 + n7
Hence, by the combination laws,
(0 − 4)(1 + 0) −4
lim an = = = −2.
n→∞ 2+0 2
Exercise 3.22 For each of the following sequences {an }, prove that it is convergent and find
its limit. (i) an = (2n+3)(n−7)
4n2 +9
2n −5
; (ii) an = n3 −3×2 1 n
n +n2 ; (iii) an = n3 × 3 .

Proof of Theorem 3.4.4 We will prove parts (b), (c) and (d) and leave the remaining parts
as an exercise.
(b) To prove that lim (an + bn ) = l + m we use the trick of expressing (an + bn ) − (l + m) in
n→∞
terms of null sequences. We know that {an − l} and {bn − m} are null. Now
(an + bn ) − (l + m) = (an − l) + (bn − m).
By the addition law for null sequences, this means {(an + bn ) − (l + m)} is null. Hence
lim (an + bn ) = l + m.
n→∞
(c) To prove that lim (an bn ) = lm we again try to express an bn − lm in terms of null sequences.
n→∞
(an − l)(bn − m) = an bn − lm + (lm − an m − bn l + lm) = (an bn − lm) − m(an − l) − l(bn − m).
Therefore (an bn − lm) = (an − l)(bn − m) + m(an − l) + l(bn − m). By the scalar multiplication
and addition laws for null sequences, (an bn − lm) is a null sequence. Hence lim (an bn ) = lm.
n→∞
|m−bn |
(d) First we note that − | b1n
= = 1
m
| | m−b
bn m
n
| |bn m|
.
The numerator is certainly a null sequence,
but what about the denominator? We know that lim bn = m and so given ε = 21 |m|, there is
n→∞
some N ∈ Z+ such that |bn − m| < 21 |m| whenever n ≥ N. That is, − 12 |m| < bn − m < 21 |m|. If
m > 0 the left hand inequality gives bn > 21 m. If m < 0 then we get bn < 21 m < 0, so in either
case, |bn | > 21 |m|. This means that, for n ≥ N, |m−b n|
|bn m|
≤ |m−b n|
| 1 m2 |
. By the scalar multiplication
2
law for null sequences, | m−b
1 2 | is a null sequence. The Sandwich Lemma for null sequences now
m
n
2
shows that { b1n − 1
m
} is null. Hence lim 1
= 1
.
n→∞ bn m

Exercise 3.23 Prove parts (a) and (e) of Theorem 3.4.4.

Proposition 3.4.7 Suppose {an } and {bn } are convergent sequences for which an ≤ bn (n ≥ 1).
If lim an = l and lim bn = m, then l ≤ m.
n→∞ n→∞

Exercise 3.24 See if you can prove Proposition 3.4.7. A suggested plan of attack is as follows.
Suppose for a contradiction that l is not less than or equal to m. So l > m. Now think about
the limit of the sequence {an − bn } and recall the definition of limit to obtain a contradiction.

Exercise 3.25 Use Proposition 3.4.7 to give an alternative proof of Lemma 3.4.2, which stated
that every convergent sequence has a unique limit.
3.5 Divergent Sequences Proof & Structure 77

The Sandwich Theorem

As the name suggests, the Sandwich Theorem is a generalisation of the Sandwich Lemma for
null sequences.

Theorem 3.4.8 The Sandwich Theorem Suppose {an } and {bn } are sequences with lim an =
n→∞
lim bn = l. If for all n ≥ 1, an ≤ cn ≤ bn , then lim cn = l.
n→∞ n→∞

As with other results, it is enough to show that these conditions apply eventually. In other
words, it doesn’t matter if a finite number of values of cn are bigger than bn or smaller than an .

Proof By the laws of scalar multiplication and addition, lim (bn − an ) = l − l = 0. In other
n→∞
words, {bn − an } is a null sequence. Now subtract an throughout the inequality an ≤ cn ≤ bn
to get 0 ≤ cn − an ≤ bn − an . This means the Sandwich Lemma for null sequences applies, and
so {cn − an } is a null sequence. Hence, by the addition law, lim (cn − an + an ) = 0 + l = l.
n→∞

n+sin n
Exercise 3.26 Use the Sandwich Theorem to find lim .
n→∞ n−4

n
Exercise 3.27 By the Binomial Theorem if x > 0 and n ≥ 1, then 1 + nx ≥ 1+n· nx = 1+x.
Hence (1 + x)1/n ≤ 1 + nx . Setting a = 1 + x in the inequality, use the Sandwich Theorem to
show that for a > 1, lim a1/n = 1.
n→∞

3.5 Divergent Sequences

Not all sequences converge. A sequence that does not converge is called a divergent sequence.
This is potentially a difficult definition to work with, because it is about the absence of a limit.
To show that a sequence is divergent we have to show that every single one of the infinity of
real numbers is not the limit of the sequence. Luckily, there are ways around this. The most
important is the idea of bounded and unbounded sequences. The definitions are like those for
bounded and unbounded sets.

Definition A sequence {an } is bounded above if there is a real number u such that an ≤ u for
all n ≥ 1. Otherwise {an } is unbounded above.
A sequence {an } is bounded below if there is a real number d such that an ≥ d for all n ≥ 1.
Otherwise {an } is unbounded below.
A sequence {an } is bounded if it is bounded below and bounded above. Equivalently, {an } is
bounded if there is a real number K such that |an | ≤ K for all n ≥ 1.
If a sequence is not bounded, then it is unbounded.

Note: The two definitions of a bounded sequence really are equivalent. To see this, note that
if there is a real number K such that |an | ≤ K for all n ≥ 1, then an ≤ K, meaning {an } is
bounded above by K. In addition, an ≥ −K, meaning {an } is bounded below by −K. On the
other hand, if {an } is bounded below (by d say) and above (by u say), we let K be the larger
of |d| and |u|. Then |an | ≤ K for all n ≥ 1.

Theorem 3.5.1 If {an } is convergent, then {an } is bounded. Hence any unbounded sequence
is divergent.
78 Proof & Structure Chapter 3: Complex Numbers and Sequences

Proof Suppose lim an = l. Then there is some N ∈ Z+ such that |an − l| < 1 whenever
n→∞
n ≥ N. So
|an | = |an − l + l| ≤ |an − l| + |l| < 1 + |l|
whenever n ≥ N. Now let K = max{|a1 |, |a2|, . . . , |aN −1 |, 1 + |l|}. Then |an | ≤ K for all n ≥ 1.
We have shown that any convergent sequence is bounded. The contrapositive (and therefore
equivalent) statement is that any unbounded sequence is divergent.

Example 3.5.2 The definition of {an } being bounded is ∃ K ∈ R such that ∀n ∈ Z+ , |an | ≤ K.
To negate this we replace ∃ by ∀, ∀ by ∃ and negate the conclusion. So the definition of {an }
being unbounded is ∀K ∈ R, ∃n ∈ Z+ such that |an | > K. To prove a sequence is unbounded
we need to show that given an real number K, there is always some an with |an | > K.
√
Consider the sequence { n}. Let K ∈ R.√ If K < 0, then we can choose n = 1 since
|a1 | = 1 > K. If K ≥ 0, we want n with
√ √ n > K. Since K and n are both nonnegative,
2
n > K if and only if n > K 2 (since n − K = √n−K n+K
). So we simply choose n to be any
2 2
√ √
positive integer greater than K . And then n > K implies |an | = n > K. Therefore { n}
is unbounded and hence divergent.

Example 3.5.3 Being bounded is not enough to guarantee convergence. For example the
sequence {(−1)n } is clearly bounded (by 1, for example) but is divergent. The best way to
prove this is by contradiction: suppose the sequence converges and lim an = l, say. If l = 1,
n→∞
then (choosing ε = |1 − l|), there is some N for which |an − l| < |1 − l| for all n ≥ N. Let
n = 2N. Then an = 1. But now we have |1 − l| < |1 − l|, a contradiction. Therefore l = 1.
But now, choosing ε = 1, there is some N for which |an − 1| < 1 for all n ≥ N. Choosing, for
example, n = 2N + 1, we get an = −1 and hence 2 = | − 1 − 1| < 1, another contradiction.
Therefore {(−1)n } does not converge, so is divergent.
n
Exercise 3.28 Show that the sequences {an }, given by an = (−1)
n2 +1
n
, and {bn }, given by
n
(−1) n
bn = n+1 , are both bounded. Decide whether each one is convergent or divergent (explain
your answers).
n−3
Exercise 3.29 Classify the sequences {cn } = 5
and {dn } = {log n} as bounded or un-
bounded and convergent or divergent.

Exercise 3.30 Suppose {an } is divergent and let c = 0 be arbitrary. Show that both {can }
and an + c are divergent.

A special type of divergent sequence is one which tends to infinity. Formally, {an } tends to
infinity if for all u > 0, there exists N ∈ Z+ such that an > u whenever n ≥ N. This is written
an → ∞ as n → ∞, or lim an = ∞.
n→∞
Essentially what happens in such a sequence is that the terms get arbitrarily large and positive.
There is an analogous definition of tending to minus infinity, where the terms in the sequence
get arbitrarily large and negative, but we won’t study that further here.

Example 3.5.4 For all k > 0, the sequence {nk } tends to infinity. Here is the proof. Given
any u > 0, let N be the smallest positive integer with N > u1/k . Since k > 0, this implies that
N k > u. Furthermore, again because k > 0, n ≥ N implies nk ≥ N k . Combining these two
observations shows that for all n ≥ N, nk > u. Hence, by definition, lim nk = ∞.
n→∞
3.6 The Monotone Convergence Theorem Proof & Structure 79

Almost all the laws about combining convergent sequences apply to combining sequences which
tend to infinity.

Theorem 3.5.5 Suppose {an } and {bn } are sequences which tend to infinity and let c ∈ R.
Then

(a) lim (can ) = ∞ if c > 0 (Scalar Multiplication);

n→∞

(b) lim (an + bn ) = ∞ (Addition);

n→∞

(c) lim (an bn ) = ∞ (Product);

n→∞

(d) lim (an + c) = ∞ (Addition of Constant).

n→∞

Exercise 3.31 Prove Theorem 3.5.5.

3.6 The Monotone Convergence Theorem

In this section we prove one of the most important results for sequences. It is as near as we
can get to a converse of Theorem 3.5.1, which stated that any convergent sequence is bounded.
We know there are bounded divergent sequences (such as {(−1)n }). However, we only need
one extra requirement to ensure convergence.

Theorem 3.6.1 (Monotone Convergence Theorem) Every bounded monotonic sequence

is convergent.

Recall that a monotonic sequence is one which is either increasing (an ≤ an+1 for all n ≥ 1) or
decreasing (an ≥ an+1 for all n ≥ 1).

Proof Suppose {an } is bounded and monotonic. There are two cases, increasing and decreas-
ing. Suppose first that {an } is monotonic increasing. That is, an ≤ an+1 for all n ≥ 1. Since
{an } is bounded, in particular {an } is bounded above. That is, there exists a real number u
such that an ≤ u for all n ≥ 1. But this means the set A = {an : n ≥ 1} is bounded above by
u. By the Completeness Axiom for the real numbers, every non-empty set which is bounded
above has a least upper bound. Let l = lubA. We will show that lim an = l.
n→∞
To show that lim an = l, remember the definition of limit. Given any ε > 0, we need to find
n→∞
N ∈ Z+ such that |an − l| < ε whenever n ≥ N. Since l = lubA, we know an ≤ l for all n ≥ 1.
Suppose for a contradiction that an ≤ l − ε for all n ≥ 1. Then l − ε would be an upper bound
for A, contradicting the fact that l is the least upper bound. Therefore there is some n for
which an > l − ε. Let N be the least value of n for which this is true. Now {an } is increasing.
Therefore
l − ε < aN ≤ aN +1 ≤ · · · .
That is, for all n ≥ N, l − ε < an . Rearranging we get l − an < ε. Since an < l, certainly
l − an > 0. Thus |l − an | = l − an < ε, for all n ≥ N. Hence lim an = l as required. If {an }
n→∞
is decreasing, then {−an } is increasing and the above argument shows that {−an } converges.
By the law of scalar multiplication, this implies {an } converges and the proof is complete.
80 Proof & Structure Chapter 3: Complex Numbers and Sequences

Note This proof depends on the Completeness Axiom for the real numbers. So it really is
a crucial thing to have. The Monotone Convergence Theorem is not true if we work in the
set of rational numbers. That is, bounded monotonic sequences of rational numbers do not
necessarily converge to a rational limit!
2
Example 3.6.2 Show that {an } converges, where a1 = 3 and an = 3 − an−1 for n ≥ 2. Firstly,
one might worry that some an = 0, meaning an+1 would be undefined. If you work out a few
terms, you see that the sequence seems to be decreasing but the terms are always greater than
2. So let’s prove by induction that an > 2 for all n ∈ Z+ . Clearly a1 > 2. Assume ak > 2.
Then ak+1 = 3 − a2k > 3 − 22 = 2. Hence, by induction an > 2 for all n ∈ Z+ . Now we prove
by induction that an+1 < an for all n ∈ Z+ . The case n = 1 is done. Now assume ak+1 < ak .
Then a1k < ak+1
1
. So ak+2 − ak+1 = 3 − ak+1 2
− 3 − a2k = 2 a1k − ak+1 1
< 0. Therefore
ak+2 < ak+1 and by induction an+1 < an for all n. That is, {an } is decreasing. We also know
the sequence is bounded below by 2. It is clearly bounded above by 3 because a1 = 3 and the
sequence is decreasing. Thus the sequence is monotonic (decreasing) and bounded and so, by
the MCT, it converges. Now the limit l must satisfy l = 3 − 2l (because lim an = lim an−1 = l).
n→∞ n→∞
So we can solve this equation (knowing there is a limit). Rearrange to get l2 − 3l + 2 = 0, so
(l − 2)(l − 1) = 0 and either l = 1 or l = 2. By Proposition 3.4.7, the fact that 2 ≤ an ≤ 3 for
all n ∈ Z+ implies 2 ≤ l ≤ 3. So the limit is 2.

Exercise 3.32 Use the Monotone Convergence Theorem to show that {bn } converges, where
b1 = 1, bn = 12 3 − 2bn−1
1
+3
for n ≥ 2. Hence find lim bn .
n→∞

Monotonic sequences either converge or tend to ±∞. The following exercise asks you to prove
this for monotonic increasing sequences.

Exercise 3.33 Suppose {an } is monotonic increasing, but unbounded. Show that lim an = ∞.
n→∞
(Hint: As {an } is bounded below by a1 , {an } must be unbounded above.)

The Monotone Convergence Theorem has an interesting application to a method based on one
used by Archimedes for approximating π. Consider a circle of radius 1 and let an be the area
of the inscribed regular n-gon. Let bn be the area of the escribed regular n-gon. The diagrams
below show on the left the inscribed hexagon and on the right the escribed hexagon.

Exercise 3.34 Find a6 and b6 .

Intuitively {an } is an increasing sequence bounded above by the area of the circle (which is of
course π). And {bn } is a decreasing sequence bounded below by the area of the circle. By the
Monotone Convergence Theorem then, both {an } and {bn } converge and lim an ≤ π ≤ lim bn .
n→∞ n→∞
Therefore by calculating an and bn for sufficiently high values of n, we can approximate π as
closely as we wish.
3.7 Series Proof & Structure 81

Exercise 3.35 Find formulae for an and bn . Calculate a1000 and b1000 and hence give an
approximation to π. (Use your calculator!) By the way, you will find, if you are using radians,
that you seem to need to work with angles involving π – this is clearly putting the cart before
the horse a bit! Luckily there are ways round this, such as only working it out for n =
3, 6, 12, 24, 48, . . . and using some clever recurrence relations. However this would be a digression
of several pages so we won’t do that. But it can be done.

3.7 Series
Having developed quite a lot of theory about sequences, which is of value in itself as we have
seen with the discussion about π, we now look briefly at an application of the theory to the
study of infinite series. Given a sequence {an } the sum a1 + a2 + · · · is called an infinite series
(or just a series).
You have encountered some examples of series in Algebra 1, mostly finite – in other words they
arise from finite sequences, which we haven’t really considered here. For example you will have
considered arithmetic and geometric progressions. Before we go further I will summarise what
I expect you to know about these types of sequence.

Arithmetic progressions (APs) are sequences of the form {a + (n − 1)d}. That is, a1 = a,
a2 = a + d, a3 = a + 2d and so on (where a is the first term and d is the common difference).
The sum to n terms of an AP {an } is a1 + a2 + · · · + an and is written ni=1 ai .

Lemma 3.7.1 The sum to n terms of an AP {an } is

n
n(n − 1)d
ai = na + .
i=1
2

Example 3.7.2 The sequence 2, 5, 8, 11, 14, . . . is an AP. Here a = 2 and d = 3. Therefore
an = 2n + 32 n(n − 1). The sum to 6 terms is 6 × 2 + 21 × 6 × 5 × 3 = 57.

Geometric progressions (GPs) are sequences of the form {arn−1 }, where a is the first term
and r is the common ratio.

Lemma 3.7.3 The sum to n terms of a GP {an } is given by

n
a(r n − 1)
ai =
i=1
r−1

if r = 1, and na if r = 1.

Example 3.7.4 The sequence 7, 14, 28, 56, 112, . . . is a GP with first term a = 7 and common
5 −1)
ratio r = 2. Thus an = 7 × 2n−1 . The sum to 5 terms is 7(22−1 = 217.

We need to establish formally some notation before discussing the sum of a whole sequence
rather than the sum to n terms.
82 Proof & Structure Chapter 3: Complex Numbers and Sequences

Definition Let {an } be a sequence. The sum to n terms of {an } is

n
sn = a1 + a2 + · · · + an = ai .
i=1

(Sometimes sn is called the nth partial sum of {an }.)

The series a1 + a2 + a3 + · · · is convergent if the sequence {sn } is convergent. If lim sn = s,
n→∞
then we say the series converges to s and write ∞ i=1 ai = s (here s is called the sum of the
series. The series is divergent if the sequence {sn } is divergent.
These definitions mean that we prove results about series by applying what we know about
sequences to the sequence of partial sums {sn }. The first task is usually to decide whether a
given series is convergent or divergent.

Example 3.7.5 Consider the series 1 − 21 + 41 − 18 + · · · . This is the sum of a GP {an } with first
n −1)
term 1 and common ratio − 12 . Thus an = (− 12 )n−1 . The sum to n terms is sn = ((−1/2)(−1/2)−1
=
2 1 n n n
− 3 − 2 − 1 . By Theorem 3.3.12(b), {x } is null whenever |x| < 1. Therefore {(−1/2) }
is null. Hence, by the combination laws for convergent sequences, lim sn = − 23 (0 − 1) = 32 .
n→∞
Thus the series converges with sum 23 .

Exercise 3.36 For each of the following infinite series, determine whether the series is conver-
gent or divergent. (i) ∞ 1 r
r=1 ar where ar = 2r − 2 ; (ii)
∞ 1
r=1 br where br = r! (use the Monotone
Convergence Theorem and the fact that r!1 ≤ 2r−11
for all r ∈ Z+ ) (iii) ∞ r
r=1 cr where cr = 2r
n n 1 n 1 n 1
(hard – first show that r=1 cr = i=1 2i + i=2 2i + · · · + i=n 2i ).

Example 3.7.6 Suppose {an } is a geometric progression with common ratio r and first term
n −1)
a. Then an = arn−1 and sn = a(rr−1 (or sn = na if r = 1). If |r| < 1, then by the combination
a(0−1) a
laws, we get lim sn = r−1 = 1−r . On the other hand if |r| ≥ 1 (and a = 0), then {sn }
n→∞
diverges. In summary, a GP converges if and only if |r| < 1.

Tests for Convergence and Divergence

∞
Proposition 3.7.7 The Non-Null Test If {an } is not a null sequence, then r=1 ar is
divergent.

Proof (Contrapositive) We’ll prove that if ∞ r=1 ar is convergent, then {an } is null. If
∞
r=1 ar is convergent, then the sequence {sn } of partial sums is convergent, with limit s say.
Now an = sn −sn−1 . But clearly {sn−1 } converges to s as well. Therefore, by the law of addition
for convergent sequences, lim an = lim sn − lim sn−1 = s − s = 0. Hence {an } is null.
n→∞ n→∞ n→∞

Example 3.7.8 The sum of any AP {a + (n − 1)d} for which d = 0 is divergent, because the
sequence of terms {a + (n − 1)d} is clearly not null.

∞ r 2 −1
Exercise 3.37 Show that r=1 (16r+9)(25r−4)
is divergent.

Exercise 3.38 Prove that if r ≥ 1 and a = 0 in a GP, then {sn } diverges, as claimed in
Example 3.7.6.
3.7 Series Proof & Structure 83

∞
The converse of Proposition 3.7.7 is not true. There are null sequences {an } for which r=1 ar
diverges. The most famous example is the so-called harmonic series ∞ 1
r=1 r .

Example 3.7.9 The harmonic series is divergent. To see this, let sn be the nth partial sum.
Then
sn = 1 + 12 + 31 + 14 + 51 + 61 + 17 + 81 + 91 + · · · + n1
= 1 + 21 + 13 + 41 + 15 + 61 + 17 + 81 + 91 + · · · + n1
> 1 + 21 + 2 14 + 4 18 + · · ·
> 1 + 21 + 21 + 12 + · · · .
So s8 > 1 + 23 for example. In general s2m > 1 + m2 . Hence the partial sums are increasing and
unbounded, so tend to infinity. That is, the harmonic series is divergent.
Example 3.7.10 The series ∞ 1 1
r=1 r 2 is convergent. To see this, note that r 2 is always positive,
the partial summands sn form an increasing sequence. Here’s a clever little argument:
1
sn = 1 + 22
+ 312 + 412 + 512 + · · · + n12
1 1 1 1 1
< 1+ 1×2
+ 2×3 + 3×4 + 4×5 +··· + (n−1)×n
1
= 1+ 1
− 2 + 2 − 3 + 3 − 14
1 1 1 1
+···+ 1
n−1
− 1
n
.
1
= 2− n
< 2.
Therefore {sn } is increasing and bounded, so by the Monotone Convergence Theorem, {sn }
converges. Hence ∞ 1
r=1 r 2 is convergent.

Note It can be shown that ∞ 1 π 2

r=1 r 2 = 6 . The first to discover this was Euler – the original
paper is freely available online – just search for ‘De summis serierum reciprocarum’, which is
its title. His justification uses the infinite series for sin x in a rather reckless way by modern
standards! I have included his method along with a more modern proof in Appendix 4 to the
notes.
∞
Theorem 3.7.11 The Comparison Test Suppose 0 ≤ an ≤ bn for n ≥ 1. If r=1 br
∞ ∞ ∞
is convergent, then r=1 ar is convergent. Hence if r=1 ar is divergent, then r=1 br is
divergent.

Proof Let {sn } and {tn } be the sequences of nth partial sums for {an } and {bn } respectively.
Since 0 ≤ an ≤ bn , clearly sn ≤ tn for all n ≥ 1. Moreover both {sn } and {tn } are increasing
sequences. Now ∞ r=1 br is convergent, and so {tn } is convergent with limit t say. Therefore
sn ≤ tn ≤ t for all n ≥ 1. This means {sn } is an increasing sequence which is bounded above
by t. By the Monotone Convergence Theorem therefore, {sn } is convergent. Hence ∞ r=1 ar is
convergent.
∞
Exercise 3.39 Use the Comparison Test to show that √1 is divergent.
r=1 r

Techniques like these can be extended to cover series with negative terms, and hence to show
that power series such as those for sin x and cos x are convergent, and indeed that it is OK to
plug complex numbers into them. However we could spend a whole year studying this alone!
We’ll finish this section by showing that the series ∞ 1
r=1 r k is convergent for all k ≥ 2.
∞ 1
Exercise 3.40 Use the Comparison Test and the fact that r=1 r 2 converges to show that
∞ 1
r=1 r k converges for all k ≥ 2.
 84 Proof & Structure Chapter 3: Complex Numbers and Sequences

Exam-Style Questions
Section A
Exercise 3.41 Let w = 3 − i and z = 1 + 2i.
w
(a) Find w + z, w − z, wz and z
, expressing your answers in the form a + bi where a, b ∈ R. [3]
(b) Plot w, z and w + z on an Argand diagram. [2]
Exercise 3.42 Let w = −2 + 2i.
(a) Find the polar form of w. [2]
(b) State De Moivre’s Theorem. [1]
2 3
(c) Find w and w , giving your answers in the form a + bi, where a and b are integers. [2]
Exercise 3.43(a) Using the Sandwich Lemma for sequences and the fact that { n1 } is null,
show that n12 is null. [2]
3n2 −2n+1
(b) Show that 5n2 +7n+6
converges and find its limit. [3]

Exercise 3.44 Consider the sequence {an } where, for n ≥ 1, an = 5 − 3−n .

(a) Show that {an } is bounded. [2]
(b) Show that {an } is monotonic increasing. [2]
(c) Is {an } convergent? Explain your answer. [1]

Section B
Exercise 3.45(a)(i) By considering (cos θ + i sin θ)4 in two different ways, find an expression
for cos 4θ in terms of cos θ and sin θ. Hence show that
cos 4θ = 8 cos4 θ − 8 cos2 θ + 1.
[5]
(ii) By solving the equation 8x2 − 8x + 1 = 12 , or otherwise, use part (i) to show that cos π
12
=
√
1 3
2
+ 4
. [4]
π
(iii) Hence find sin 12
. [2]

(b) Let p and q be real numbers such that 0 < q < 1.

(i) Show that {q n } is a null sequence. [3]
n
(ii) Show that the sequence {p + q } converges, and find its limit. [1]
∞
(iii) State the Non-Null Test for convergence of a series r=1 br . [1]
∞ 1 r
(iv) Using your answers to (ii) and (iii), or otherwise, show that the series r=1 4+ 2
diverges. [2]
∞ 1 r
(v) Suppose that the series r=1 x+ 2
converges. Find x. [2]
Note: Exam questions, particularly Section B questions, often combine elements from different
chapters.
3.7 Series Proof & Structure 85

Historical Notes

Abraham De Moivre (26 May 1667 – 27 Nov 1754)

Abraham de Moivre was born and brought up in France, but fled to England in the 1680s after
having been imprisoned for his Protestant beliefs. In England he became a mathematics tutor
and befriended such luminaries as Newton and Halley. He became such an expert on Newton’s
Principia that when people asked Newton anything about it, he replied, ‘go to Mr de Moivre;
he knows these things better than I do’. I can’t resist diverting here to mention that Halley (of
comet fame) funded the publication of the Principia, because the Royal Society couldn’t afford
to. It had spent all its funds the year before on publishing a massive tome called ‘A History
of Fishes’, which was a complete flop. Even worse, Halley was hoping to get a good wage in
his new post as secretary of the Society, but they decided they couldn’t afford to give him any
money. Instead he was paid in copies of ‘A History of Fishes’. Anyway, back to de Moivre. His
main work was on probability theory, but of course his name lives on in the complex number
theorem that bears his name. Rather macabrely, de Moivre predicted exactly what day he
would die. Towards the end of his life, he noticed that every day he was sleeping 15 minutes
longer than the day before. Noticing that this was an arithmetic progression, he worked out
which term would represent 24 hours sleep and said that would be the day he died. And it was.

Leonhard Euler (15 April 1707 – 18 Sept 1783)

Euler was an incredibly prolific and inventive mathematician. He did ground-breaking work
in number theory, geometry, calculus and mechanics, practically invented graph theory, and
introduced standard notation such as f (x) for a function, i for ( − 1) and the notation for
series. He introduced e and proved it is irrational. He wrote an important treatise, Mechanica
(1736), on the dynamics of particles, and another in 1750 on motion of rigid bodies, including,
for example, moments of inertia. He studied conics (like ellipses and parabolas) and quadrics.
He studied permutations. It is exhausting just reading about him! After his death the St
Petersburg Academy continued publishing Euler’s unpublished work – it took nearly 50 years
to get through the backlog.
Euler wrote over 800 papers and several books. In fact one historian has estimated that Euler
wrote a full quarter of the publications in the 18th century on mathematics and its applications.
Clearly it’s impossible to even scratch the surface in this short note. However, 2007 was Euler’s
tricentenary and there was a flurry of activity; several excellent books appeared, and a vast
amount of information went online. To find out more about Euler, I suggest the online Euler
Archive http://math.dartmouth.edu/∼euler/ as a starting point.
It is comforting for anyone with children to know that Euler managed all this in addition to
having 13 children, and said that some of his best work was done while ‘holding a baby in his
arms with children playing around his feet’.
86 Proof & Structure Chapter 3: Complex Numbers and Sequences

Learning Outcomes

After studying this chapter, and the related exercises, you should be able to

• perform basic arithmetic with complex numbers;

• define the complex conjugate, real part, imaginary part, modulus and argument of a
complex number, and be able to determine these for a given complex number;

• express a complex number in polar form and illustrate complex numbers using an Argand
diagram;

• state De Moivre’s Theorem and use it to express, for n ∈ N, cos nθ and sin nθ in terms
of cos θ and sin θ;

• use De Moivre’s Theorem to find, for n ∈ Z+ , all of the nth roots of a given complex
number;

• be aware that eiθ = cos θ + i sin θ for all θ ∈ R;

• be familiar with the terminology of sequences; including the terms increasing, decreasing,
strictly increasing, strictly decreasing, monotonic, constant and be able to determine if a
given sequence is one of these types;

• know the definition of a null sequence and be able to show whether a given sequence is
null, directly or using the Combination Laws and/or the Sandwich Lemma;

• be familiar with simple null sequences such as { n1k } (for k > 0), {xn } (for |x| < 1), { n!1 },
n k
{xn nk } (for |x| < 1, k > 0), { xn! } and { nn! } (for k > 0);

• work with the definition of a limit and convergent and divergent sequences; be aware that
a convergent sequence has a unique limit;

• state and apply the Combination Laws and the Sandwich Theorem for convergent se-
quences;

• be able to show whether a given sequence is bounded or unbounded, and know that any
bounded monotonic sequence is convergent (the Monotone Convergence Theorem) and
that any unbounded sequence is divergent;

• explain what it means for a sequence to tend to infinity, and know that monotonic in-
creasing sequences either converge or tend to infinity;

• be familiar with the definition of a series (including arithmetic and geometric progres-
sions), the sum to n terms, and convergent and divergent series;

• show whether a given series is convergent and use the Non-Null Test (or show directly)
that a given series is divergent.