SERIES OF SEQUENCES

AD FUNCTIONS

Contents

1. Geometric series 2
2. Nice and easy convergence tests. 6
3. Absolute convergence and the alternating series test 9
4. More convergent tests 11
5. The Integral Test 13
6. Rearrangement 15
7. Product of series 18
8. Euler number 21
9. Euler number as series 24
10. Sequences of complex numbers 27
11. Bounded sequences, closed sets and series of complex
numbers 30
12. Sequences of continuous functions 31
13. Series of continuous functions 36
14. Complex continuous functions and their series 38
15. Power series 42
16. The sum of power series 46
17. The exponential function 47
18. The trigonometrical functions 49

1
2

1. Geometric series

P10 1
Exercise 1. Calculate i=0 3i .

You learned in high school that

n
X 1 − q n+1
qi = ,
i=0
1−q

provided that q 6= 1. Here, q = 13 , that is

P10 1
Solution: i=0 3i = 23 (1 − 1
311
). 

P10000 1
Exercise 2. Calculate i=0 3i
.

P10000 1
Solution: i=0 3i
= 32 (1 − 1
310001
). 

P∞ 1
Exercise 3. Calculate i=0 3 .

It is tempting to say that the Solution is 23 (1 − 31∞ ), which seems to be

equal to 23 . The intuition is good, but we have not defined infinite sums
1
yet, also 3∞
is not really a meaningful expression. But, we are on the
right track. What is π? 3.14159653589... What does it mean?

1 4 1 5 9
π =3+ + + + + + ....
10 100 1000 1000 10000

so, we lived with infinite sums for a while...

And now the most important definition of the whole course. But before
that... some revision from MATH113.
 3

Revision: [Convergence]. A sequence of real numbers {cn }∞

n=1

is converging to c, if for any  > 0 there exists some N > 0 such

that for any n ≥ N , |c − cn | < .

Definition 1.1. [Series] Let {ai }∞

i=1 be a sequence of real num-
P∞ Pk
bers. Then i=1 ai is called a series. Then, bk = i=1 ai =

a1 + a2 + · · · + ak is called the k-th partial sum of the series,

we say that the series is convergent if limk→∞ bk exists, then
P∞
i=1 ai = limk→∞ bk . If the series is convergent then limk→∞ bk = s

is called the sum of the series. If the sequence of partial sums is

not convergent then we say that the series is divergent.

Remark 1.1. It is very, very important to remember that {ai }∞

i=0 is a

sequence and ∞
P
i=0 ai is a series.

Remark 1.2. Now we can Solution Exercise 3. In this case, the series
is in the form of ∞
P Pk−1
i=1 ai so the k-th partial sum bk equals to i=0 ai ,
P∞ 1
that is, i=0 3i = limk→∞ bk = limk→∞ 23 (1 − 31k ) = 23 .
P

Proposition 1.1. Let q be an arbitrary real number. Then, the geo-

metric series ∞ n
P
i=0 q is convergent if and only if |q| < 1.

1−q k
Proof. As we have seen above the k-th partial sum bk equals to 1−q
.
1
Thus, limk→∞ bk exists and equals to 1−q
provided that |q| < 1. Clearly,
if |q| ≥ 1 the partial sums are unbounded, so the series is divergent. 

A Historical Remark
4

”In a race, the quickest runner can never overtake the slowest, since
the pursuer must first reach the point whence the pursued started, so
that the slower must always hold a lead.” (Aristotle’s recount of the
Paradox of Zeno)

Zeno of Elea lived almost 2500 years ago. He came up with nine para-
doxes, among them the most famous one was the Paradox of Achilles
and the Tortoise.

In this Paradox a race was staged between Achilles and a tortoise.

Achilles the great Greek warrior could run hundred times as fast as
the tortoise, so the tortoise was given a mile head start. Zeno argued
that Achilles would never overtake the tortoise. His argument went
as follows. By the time Achilles reached the starting position X1 of
the tortoise, the tortoise had crawled further to a new position X2 .
Thus Achilles must reach X2 , but at that time the tortoise had already
moved to a further position X3 . It goes on and on and the tortoise
will always be ahead of Achilles. So, Achilles can never overtake the
tortoise.

The resolution of the Paradox of Zeno is that the whole story takes
place in a finite time frame. It takes t minutes for Achilles to reach the
1
position X1 and then 100
t
minutes to reach the position X2 a further
1
t minutes to reach X3 . Hence, after t ∞ 1 i 100
P
10000 i=0 ( 100 ) = 99 minutes

Achilles and the tortoise will be in exactly the same position.

Exercise 4. Using geometric series write 0.151515.... as the quotient

of two integers.
 5
P∞ 1
Solution: Observe that 0.153153153.... equals to 15 i=1 1000i . Thus

1000 153
0.153153153.... = 153( − 1) = .
999 999
6

2. Nice and easy convergence tests.

In general, it is not easy to decide whether a certain series is convergent

or not. However, there are some instances, when one can decide on
convergence by inspection using our Math113 results on convergence.

P∞
Proposition 2.1 (n-th term test). Let i=1 ai be a convergent series,
where {ai }∞
i=1 are real numbers. Then, limi→∞ ai = 0.

P∞
Proof. Let pk be the k-th partial sum of i=1 ai . Then by Cauchy
Theorem,
lim |pk − pk−1 | = lim |ak | = 0 .
k→∞ k→∞
P∞
Hence, i=1 ai = 0. 

One might think that the converse holds as well. The following exercise
shows that this is far from being true.

Exercise 5. Show that the following series is divergent.

1 1 1 1 1 1 1
1+ + + + + + + + ....
4 4 8 8 8 8 16
P∞
That is, i=1 ai , where

• a1 = 1.
1 1 1
• ai = 2n
if 2n−1
≤i≤ 2n −1
and n ≥ 2.

Answer: Let pk be the k-th partial sums of the series above. Then by
inspection, p1 = 1, p3 = 2, p7 = 3 and in general, p2n −1 = n provided
that n ≥ 2. Hence, limk→∞ pk does not exist. Therefore, our series is
divergent. .
 7

Proposition 2.2 (Finite modification). Let {ai }∞ ∞

i=1 and {bi }i=1 be se-

quences of real numbers. Suppose that for large enough i, ai equals to

bi . Then, ∞
P P∞
i=1 ai is convergent if and only if i=1 bi is convergent as

well.

Proof. We have an integer N such that if i ≥ N then ai = bi . Let pk

resp. rk be the k-th partial sums of ∞
P P∞
i=1 ai resp. i=1 bi . Thus, if

k ≥ N we have
N
X
pk − rk = (ak − bk ) .
i=1

Thus, limk→∞ pk exists if and only if limk→∞ rk exists. 

Proposition 2.3 (Simple comparison). Let {ai }∞ ∞

i=1 and {bi }i=1 be se-

quences of nonnegative real numbers. Suppose that for any i ≥ 1

P∞
ai ≤ bi . Then, if the series i=1 bi is convergent, so is the series
P∞
i=1 ai .

Proof. Again, let pk resp. rk be the k-th partial sums of ∞

P
i=1 ai resp.
P∞ ∞
i=1 bi . By our assumption, limk→∞ rk = M exists. Also, {pk }k=1 is

a monotonically increasing (nondecreasing) sequence, bounded by M .

Hence by our result in Math113, limk→∞ pk exists and its limit is less
or equal than M . 

The following rather simple proposition is absolutely crucial.

Proposition 2.4 (Cauchy test). Let {ai }∞

i=1 be a sequence of real num-
P∞
bers. Then, i=1 ai is convergent if and only if for any  > 0 there
exists L > 0 such that for any M > L
M
X
| aj | ≤  .
j=L
8
P∞ P∞
Proof. Let pk be the k-th partial sum of i=1 ai . Suppose that i=1 ai
is convergent. Then, {pk }∞
k=1 is a Cauchy-sequence. That is, for any

 > 0 there exists N ≥ 1 such that |pm − pn | ≤  provided that

m, n ≥ N . So, let L = N + 1 and M > L. Then,
M
X
| aj | = |pM − pL | ≤ .
j=L

Conversely, if for any  > 0 there exists L > 0 such that for any M > L
M
X
| aj | ≤ 
j=L

then {pk } is a Cauchy sequence, hence it is convergent. 

The following proposition is an immediate corollary of Proposition 2.4.

Proposition 2.5. Let {ai }∞ i=1 be a sequence of real numbers. Then,

P∞
i=1 ai is convergent if and only if for any  > 0 there exists L > 0
such that ∞ | ∞
P P
i=L ai is convergent and i=L ai | ≤ .
 9

3. Absolute convergence and the alternating series test

Definition 3.1. Let {ai }∞

i=1 be a sequence of real numbers. We

say that the series i=1 ai is absolutely convergent if ∞

P∞ P
i=1 |ai | is

convergent.

The following theorem is crucial.

Theorem 1. Any absolutely convergent series is convergent.

P∞
Proof. Let i=1 ai be an absolutely convergent series. By Proposition
2.4, it is enough to show that for any  > 0 there exists some L > 0
such that for any M > L
M
X
| ai | ≤ .
i=L

However, by Proposition 2.4, it follows that for any  > 0 there exists
some L > 0 such that for any M > L
M
X
|ai | ≤ .
i=L

Since
M
X M
X
| ai | < |ai | ,
i=L i=L

our theorem follows. 

One might immediately ask whether every convergence sequence is ab-

solutely convergent. It turns out that this is far from being true.
10

Theorem 2 (The Alternating Series Test). Let a1 ≥ a2 ≥ a3 . . . be

a sequence of positive real numbers tending to zero. Then the series
P∞ i+1
i=1 (−1) ai is convergent. We call such series alternating series.

Proof. Let pk be the k-th partial sum of the series. Then, the sequence
{p2k−1 }∞
k=1 is decreasing. Indeed,

p2k−1 − p2k+1 = a2k+1 − a2k ≥ 0 .

Similarly, the sequence {p2k }∞

k=1 is increasing. Indeed,

p2k+2 − p2k = a2k+2 − a2k+1 ≥ 0 .

Since p2k−1 − p2k ≥ 0, we have that both {p2k−1 }∞ ∞

k=1 and {p2k }k=1 are

bounded, so convergent. Since (p2k−1 − p2k ) → 0, the limis of the

two sequences are the same, hence limk→∞ pk is convergent. Thus, our
series ∞ i+1
P
i=1 (−1) ai is convergent. 

Exercise 6. Show that

• i=1 (−1)i+1 1i , that is

P

1 1 1
1− + − + ...
2 3 4

is convergent.
1
P
• On the other hand, the harmonic series i=1 i is divergent.

Answer: Observe that the first series is an alternating series hence by

1
Theorem 2 it is convergent. For the second part, observe that i
≥ ai ,
where {ai }∞
i=1 is the sequence in Exercise 5. Hence by the comparison

test (Proposition 2.3) the series i=1 1i is divergent.

P
 11

4. More convergent tests

One of the most applicable convergence test is the so-called ratio test.

P∞
Proposition 4.1 (The ratio test). Let n=0 an be a series of real num-
bers. Let us assume that there exists some N ≥ 1 and 0 < r < 1 such
|an+1 |
that |an |
≤ r provided that n ≥ N . Then, the series is convergent.

Proof. We have that aN +1 ≤ raN , then aN +2 ≤ raN +1 ≤ r2 aN . In-

ductively, aN +k ≤ rk aN . Therefore, by the simple comparison test
P∞
n=N an is convergent. Hence, by the finite modification test (Propo-

sition 2.2) our series is convergent. 

The next proposition is a variation of the same theme.

P∞
Proposition 4.2 (The root test). Let n=0 an be a series of real num-
bers. Let us assume that there exists some N ≥ 1 and 0 < r < 1 such
1
that |an | n ≤ r provided that n ≥ N . Then the series is convergent.

Proof. This is even easier than the ration test. If n ≥ N , then |an | ≤ rn ,
hence our proposition easily follows from the simple comparison test
(we do not even mention the trivial finite modification test). 

A somewhat trickier test can be used for series of decreasing terms.

Proposition 4.3 (Cauchy Condensation Test). Let a1 ≥ a2 ≥ a3 ≥ . . .

be an increasing sequence of positive numbers. Then,

(1) We have that

∞
X ∞
X ∞
X
an ≤ 2n a2n ≤ 2 an .
n=0 n=0 n=0
12
P∞ P∞
(2) Therefore, the series n=0 an converges if and only if n=0 2n a2n
converges.

Proof.
∞
X
an = a1 + a2 + a3 + a4 + a5 + a6 + a7 + a8 + · · · ≤
n=0

≤ a1 + (a2 + a2 ) + (a4 + a4 + a4 + a4 ) + (a8 + a8 + . . . ) ≤

∞
X
≤ 2n a2n .
n=0

On the other hand,

∞
X
2n a2n ≤ (a1 + a2 ) + (a2 + a4 + a4 + a4 ) + (a4 + a8 + a8 + . . . ) ≤
n=0

(a1 + a1 ) + (a2 + a2 + a3 + a3 ) + (a4 + a4 + a5 + a5 . . . ) =

∞
X
=2 an .
n=0

Hence (1) and so (2) follow. 

As an application get the following theorem.

P∞ 1
Theorem 3 (The p-series). For any p > 1 the series n=0 np con-
verges.

P∞ 1
Proof. By the previous proposition n=0 np converges if
∞ ∞
X
n 1 X
2 pn = (2(1−p) )n
n=0
2 n=0

P∞ 1
converges. Since, 0 < 2(1−p) < 1, the series n=0 2n 2pn is convergent
indeed. 
 13

5. The Integral Test

You already learned the notion of the improper integral in calculus. Let
R∞
f : [1, ∞) be a positive continuous function. We say that 1 f (x) dx
Rn
exists, if the sequence {in = 1 f (x) dx is bounded. Then, by positivity,
the sequence {in }∞
n=1 is increasing, hence it is convergent. Then, by

definition
Z ∞ ∞
f (x) dx = lim in .
1 n=1

R∞
and we say that the improper integral 1
f (x) dx is finite.

R∞
Exercise 7. Let p > 1, then 1
x−p dx is finite.

Rn
Answer: Observe that 1
x−p = 1
1−p
n1−p − 1
1−p
, hence
Z ∞
1
x−p dx = 
1 p−1

Proposition 5.1 (The Integral Test). Let f : [0, ∞] → R be a positive,

R∞
decreasing function. Then 1 f (x) dx is finite if and only if the series
P∞
n=1 f (n) converges.

Proof. Since f is decreasing,

Z n+1
f (x)dx ≤ f (n) .
n

Hence,
n
X Z n+1
f (k) ≥ f (x)dx
k=1 1
14

Therefore, if the series converges the integral is finite. On the other

hand,
Z n+1
f (n + 1) ≤ f (x)dx .
n

Hence,
n+1
X Z n+1
f (k) ≤ f (x)dx
k=2 1

Therefore, if the integral converges the series is finite. 

 15

6. Rearrangement

The philosophy of summing a series can be formalized as follows.

We consider the numbers as objects in such a way that an as an object
n with weight an is the weight of n. The sum of the series ∞
P
n=0 an

is the total weight of the set of non-negative integers. If you have

finitely many numbers their total weight do not depend on the order
you weight them:

a1 + a2 + a3 + a4 + a5 = a3 + a4 + a2 + a5 + a1 .

The rearrangement theorem shows that this principle hold for infinite
sums as well. So, what does rearrangement means? Let σ : N → N be
a bijection. Then, the series
∞
X
aσ(n)
n=1

is a rearrangement of the series

∞
X
an .
n=1

In short, we sum up the same numbers in different order.

P∞
Theorem 4. Let n=1 an be a convergent series of positive numbers.
Also, let of positive numbers and σ : N → N be a bijection. Then, the
series ∞
P
n=1 aσ(n) is convergent as well, and their sums are equal.

Proof. Let pk be the k-th partial sum of the first series and let rk the
k-th partial sum of the second series. We know that limk→∞ pk = s.
We need to show that limk→∞ pk = s as well. First, we claim that for
16

any k ≥ 1, rk ≤ s. Indeed, there exists m ≥ 1 such that

{1, 2, . . . , k} ⊃ {σ(1), σ(2), . . . σk} .

Hence rk ≤ pm ≤ s. Fix  > 0. It is enough to prove that there exists

N ≥ 1 such that

(1) s − rN ≤  .

it implies by positivity that s − rn ≤  provided that N ≤ n. Observe

that there exists M ≥ 1 such that s − pM ≤ .. Again, there exists
N ≥ 1 such that

{σ(1), σ(2), . . . , σ(N )} ⊃ {1, 2, . . . , M } .

Hence, rN ≤ pM and consequently (1) holds. 

The following theorem shows that rearrangement is not so innocent as

it looks for the first sight.

Theorem 5. Let a1 ≥ a2 ≥ a3 . . . be a sequence of positive real num-

bers. Such that the series ∞
P
n=1 an diverges. By the Alternating Series
P∞ n
Theorem, the series n=1 (−1) an is convergent. However, for any

real number r ∈ R. There exists σ : N → N so that the sum of the

rearranged series ∞ σ(n)
P
n=1 (−1) aσ(n) equals to r.

P∞ P∞
Proof. Observe that both n a2n+1 and n a2n diverge. We rearrange
the naturals in the following fashion. Let k1 be the first number such
that kn=1 (−1)2n1 −1 an ≤ r . Then, let k2 be the first number such that
P1

k1
X k2
X
−a2n−1 + a2n ≥ r .
n=1 n=1
 17

Continuing the algorithm, let k3 be the first number such that

k1
X k2
X k3
X
−a2n−1 + a2n + −a2n−1 ≤ r .
n=1 n=1 n=k1 +1

Then, let k4 be the first number such that

k1
X k2
X k3
X k4
X
−a2n−1 + a2n + −a2n−1 + a2n ≥ r .
n=1 n=1 n=k1 +1 n=k2 +1

Inductively, we obtain a rearrangement of the naturals

1, 3, 5, . . . , 2k1 − 1, 2, 4, . . . , 2k2 , 2k1 + 1, 2k1 + 3, . . . , 2k3 − 1,

2k2 + 2, . . . , 2k4 , 2k3 + 1, 2k3 + 3, . . . , 2k5 − 1 . . .

Leading to a bijection σ : N → N. Since an → 0, by the inequalities

above we have that
∞
X
(−1)σ(n) aσ(n) = r . 
n=1
18

7. Product of series

Recall that if {an }∞ ∞

n=1 and {bn }n=1 are convergent sequences, then

{an + bn }∞
n=1 is also a convergent sequence and

∞ ∞ ∞
lim an + bn = lim an + lim bn .
n=1 n=1 n=1

The following proposition extends the above sum theorem in the realm
of series.

Proposition 7.1. Let s = ∞

P P∞
n=0 an and t = n=0 bn are convergent
P∞
series, then n=0 an + bn is also a convergent series and its sum equals
to s + t.

Proof. Let pk be the k-th partial sum of the series ∞

P
n=0 an , rk be the

k-th partial sum of the series ∞

P
n=0 bn and wk be the k-th partial sum

of the series ∞
P
n=0 an + bn . Then, pk + rk = wk , hence our proposition

follows from the above sum theorem for convergent sequences. 

P∞
There are many ways in which a product of two series j=0 aj and
P∞
k=0 bk can be defined as a new series.

P∞
Definition 7.1. The Cauchy product is the series n=0 cn , where

X n
X
(2) cn = aj b k = aj bn−j .
j+k=n j=0

P∞ P∞ P∞
If n=0 an = A and n=0 bn = B, you may expect that n=0 cn = AB.
However, this is false in general. In fact, it’s not even necessarily the
case that cn → 0 as n → ∞.

Let {an }∞ √1
n=1 be a sequence given by the formula a2k−1 = k and a2k =

− √1k . It is clear that ∞

P
n=0 an converges, and obviously its sum is 0.
 19

Now, consider the Cauchy product of this series with itself. Then we
have that

1 1 1 1 1
c2k−1 = √ +p +p +p +. . .+ √
1·k 1 · (k − 1) 2 · (k − 1) 2 · (k − 2) k·1
1 1 1
which is bounded from below by k
+ k−1
+ ... + dk/2e
. The following
exercise and the n-th term test show that the Cauchy product is not
convergent in this case.

1 1
Exercise 8. Let bn = n
+ ... + 2n
. Show that the sequence (bn ) does
not converge to 0.

Answer:
1 1 1 1
bn = + ... + ≤n ≤ .
n 2n 2n 2

Theorem 6 (Mertens’ theorem). Let ∞

P
n=0 an be a convergent series
P∞
of positive real numbers and n=0 bn be an arbitrary convergent series.
P∞ P∞
Let us denote A = n=0 an and B = n=0 bn . Then the Cauchy

product of ∞
P P∞
n=0 an and n=0 bn is a convergent series and the sum of

the Cauchy product is indeed AB.

Proof. Let the Cauchy product be denoted by ∞

P
n=0 cn , where cn is
Pn Pn
defined by the formula (2). Let An = k=1 ak , Bn = k=1 bk and

Cn = nk=1 ck be the respective partial sums. First, we note that

P

n
X
Cn = aj Bn−j .
j=0

Indeed, we can write

n n n−j
n X n
X X X X X X
Cn = cr = aj b k = aj b k = aj b k = aj Bn−j .
r=0 r=0 j+k=r j+k≤n j=0 k=0 j=0
20
Pn
Now if we denote Dn = j=0 aj (Bn−j − B), then we can write Cn =
Dn +An B. Since the sequence {An B}∞
n=1 converges to AB, it is enough

to show that the sequence {Dn }∞ converges to 0.

P∞
Let n=0 an = A0 . Let M ∈ N be such that |Bn − B| ≤ M for all n.
P∞
Pick  > 0. Then, there exists N ≥ 1 such that n=N an < , and also
|Bn − B| <  for all n ≥ N .

Now suppose n > 2N . We have

N
X n
X
|Dn | ≤ aj |Bn−j − B| + aj |Bn−j − B| ≤ A0  +  · M.
j=0 j=N +1

This finishes the proof that {Dn }∞

n=1 converges to 0, so our theorem

follows. 
 21

8. Euler number

The goal of this section is to use to finally give a proper introduction

to the famous ”e” the Euler number, the base of the natural logarithm.

Revision: [Binomial theorem] We have the following formula:

n  
n
X n k n−k
(x + y) = x y ,
k=0
k

n n!


where k
= (n−k)!k!
.

Proposition 8.1. The sequence {an = (1 + n1 )n }∞

n=0 is bounded and

increasing.

Proof. Using the binomial theorem, we see that

1 n 1 n(n − 1) 1
an = (1 + ) =1+n + · 2 + ...+
n n 1·2 n
n(n − 1) . . . (n − n + 1) 1
· n =
1 · 2 · ... · n n
1 1 1 1 2
= 1 + 1 + (1 − ) + (1 − )(1 − ) + . . . +
2! n 3! n n
1 1 2 n−1
(1 − )(1 − ) . . . (1 − ).
n! n n n
Let us write up this expression for an+1 as well.

1 1 1 1 2
an+1 = 1 + 1 + (1 − ) + (1 − )(1 − ) + ...+
2! n+1 3! n+1 n+1
1 1 2 n−1
+ (1 − )(1 − ) . . . (1 − )+
n! n+1 n+1 n+1
1 1 2 n
(1 − )(1 − ) . . . (1 − ).
(n + 1! n+1 n+1 n+1
22

Observe that in comparison to an , in an+1 we have one extra positive

term at the end. The remaining terms are clearly getting larger, be-
cause we just changed all the expressions of the form (1− nk ) to (1− n+1
k
).
This shows that {an }∞
n=1 is increasing.

Now we show that the sequence {an }∞

n=0 is bounded. Indeed by the

expression above we have

n
X 1 1 1 1
an < 2 + < 2 + + 2 + . . . + n−1 < 3 .
k=2
n! 2 2 2

Hence our proposition follows. 

Now we need an important result from Math113.

Revision: Any bounded increasing sequence of real numbers is

convergent.

Therefore by our previous proposition, the limit of {an = (1+ n1 )n }∞

n=0

exists.

Definition 8.1. The limit of this sequence is referred to as the

Euler’s number, and it is denoted by e. It follows from the proof
that 2 < e < 3. Its actual value is around 2.71828..

Using the Proposition 8.1, we can deduce the limits of some other
sequences.

1 n
Exercise 9. Show that limn→∞ (1 + n2
) = 1.
 23
q
1 n n 1 n2
Answer: Indeed, we can write bn = (1 + n2
) = (1 + n2
) . Now
since every subsequence of (1 + n1 )n is increasing and bounded by 3, we
1 n2
see that for all n we have (1 + n2
) < 3. Therefore, by the sand-
1
wich rule, we have that 1 ≤ limn→∞ bn ≤ limn→∞ 3 n , so we have
limn→∞ bn = 1. 

Exercise 10. Show that ∀q ∈ Q we have limn→∞ (1 + nq )n = eq .

Answer: The fact that {(1+ nq )n }∞

n=1 is increasing and bounded follows

in exactly the same way as in the q = 1 case. Therefore the limit exists.
q qm
Suppose that n = qm, where m is an integer. Then, (1 + qm
) =
((1 + m1 )m )q hence, for this values limn→∞ (1 + nq )n = eq . Since the limit
exists, clearly limn→∞ (1 + nq )n = eq . 
24

9. Euler number as series

The goal of this section is to understand how we can write ”e” as

a series. This is, by the way, one of the most important series. In
the previous section we defined the Euler number e to be the limit
limn→∞ (1 + n1 )n . We will use this result to come up with a series
representation of e. The proof is not involved but a bit tricky. We
need a small revision again from Math113.

Revision: Let {an }∞

n=1 be bounded sequence of real numbers.

Then lim supn→∞ an is defined as

• 1. The maximal limitpoint of {an }∞

n=1 .

• 2. or equivalently, the only limit point of the sequence

(recall that limitpoints are just limits of convergent sub-
sequences) such that for any  > 0, only finitely many ele-
ments of the sequences is greater than {an }∞
n=1 + 

On the other hand, lim inf n→∞ an is defined as

• 1. The minimal limitpoint of {an }∞

n=1 .

• 2. or equivalently, the only limit point of the sequence such

that for any  > 0, only finitely many elements of the se-
quences is smaller than {an }∞
n=1 − 

P∞ 1
Theorem 7. We have that e is equal to the sum of the series n=0 n!
 25

Proof. Recall from the proof of Proposition 8.1 that (1 + n1 )n is equal

to
(3)
1 1 1 1 2 1 1 2 n−1
1+1+ (1− )+ (1− )(1− )+. . .+ (1− )(1− ) . . . (1− ).
2! n 3! n n n! n n n

Furthermore {an = (1 + n1 )n }∞
n=1 is an increasing sequence. Let  > 0

and let K ∈ N be such that e−aK < . By the definition of convergence

such K must exist. Let m be greater than K, and big enough so that

1 2 K
1 −  < (1 − )(1 − ) . . . (1 − ) < 1
m m m

Since by the product rule

1 2 K
lim (1 − )(1 − ) . . . (1 − ) = 1 ,
n→∞ n n n

such m must exist.

Note that in the expression for an , where n > m, the first K + 2

summands are

1 1 1 1 2 K
1+1+ (1 − ) + . . . + (1 − )(1 − ) . . . (1 − ),
2! n K! n n n

which differs from

1 1
SK = 1 + 1 + + ... +
2! K!

by at most  · e. This is a bit tricky, but very easy. Note that

1 1 2 K
1− > (1 − )(1 − ) . . . (1 − ) ≥ .
n m m m
26

It follows that for all large enough K we have SK < e + e. Thus by
definition, lim supn→∞ Sn ≤ e. By (3) it is clear that lim inf Sn ≥ e.
Thus {Sn }∞
n=1 must converge to e. 

Using Exercise (10), following the previous proof, you can easily prove
the following proposition.

Proposition 9.1. We have that eq is equal to the sum of the series

P∞ q n
n=0 n! .
 27

10. Sequences of complex numbers

Revision: [Modulus of complex numbers]. Let z = a + bi be

a complex number. Then, its modulus (or length, or norm) |z|
√
is defined as |z| = a2 + b2 . If z = reiθ in the polar form, then
|z| = r. If z, w are complex numbers, then

• |z + w| ≤ |z| + |w| . (triangle inequality)

• If λ > 0 is a real number, then |λz| = λ|z| .

This two properties are absolutely crucial, they also hold for real
numbers and, as we shall see soon they hold for continuous functions
as well! One should also note that |zw| = |z||w|.

Our main point is that basically the whole Math113 course can be given
for complex numbers instead of real numbers without any difficulty.

Definition 10.1 (Complex calculus). A sequence of complex num-

bers {zn }∞
n=1 is converging to a complex number z if for any  > 0

there exists N > 0 such that if n ≥ N then |z − zn | < .

A sequence of complex numbers {zn }∞

n=1 is a Cauchy sequence if

for any  > 0 there exists N > 0 such that if n, m ≥ N then

|zn − zm | < .

The following proposition looks quite innocent and indeed it is very easy
to prove. However, this proposition is the key tool to translate theorems
of real calculus into theorems of complex calculus in a straightforward
28

manner. Note for a moment, that by ”complex calculus” we really

mean calculus on the plane. In Year 2, you will have a beautiful course
on Complex Analysis which is way much more than just a straight-
forward generalization of Real Analysis. But... Complex Calculus will
come rather handy.

Proposition 10.1. A sequence of complex numbers {zn }∞

n=1 is con-

vergent if and only if both {Re(zn )}∞ ∞

n=1 and {Im(zn )}n=1 are conver-

gent. Also, {zn }∞ ∞

n=1 is Cauchy if and only if both {Re(zn )}n=1 and

{Im(zn )}∞
n=1 are Cauchy.

Proof. By triangle inequality, if z, w are complex numbers,

|Re(z − w)| + |Im(z − w)|

≤ max{|Re(z − w)|, |Im(z − w)|} ≤
2

(4) ≤ |z − w| ≤ |Re(z − w)| + |Im(z − w)| .

Hence if zn → z, then both Re(zn ) → Re(z) and Im(zn ) → Im(z)

and conversely, if both Re(zn ) → Re(z) and Im(zn ) → Im(z) , then
zn → z.

Now, assume that {zn }∞

n=1 is a Cauchy sequence, that is, for every  > 0

there exists N > 0 so that |zn − zm | ≤  provided that n, m ≥ N . By

(10), |Re(zn − zm )| ≤  and |Im(zn − zm )| ≤ , thus {Re(zn )}∞
n=1 and

{Im(zn )}∞
n=1 are both Cauchy sequences.

Conversely, assume that both {Re(zn )}∞ ∞

n=1 and {Im(zn )}n=1 are Cauchy

sequences. Then, for any  > 0 we have an N > 0 such that |Re(zn ) −
 29


Re(zm )| < 2
and |Im(zn ) − Im(zm )| < 2 . That is by (10) |zn − zm | ≤ ,
so {zn }∞
n=1 is a Cauchy sequence. 

Exercise 11. Show that any convergent sequence of complex numbers

is a Cauchy sequence.

Solution: By the previous proposition, a sequence of complex numbers

{zn }∞ ∞ ∞
n=1 is convergent if and only if both {Re(zn )}n=1 and {Im(zn )}n=1

are convergent. Hence by our Math113 result, both are Cauchy se-
quences. Applying again our proposition, we obtain that {zn }∞
n=1 is a

Cauchy sequence. 

Exercise 12. Show that any Cauchy sequence of complex numbers is

converging to some complex number.

Answer: Let {zn }∞

n=1 be a Cauchy sequence. By the previous
proposition, {Re(zn )}∞ ∞
n=1 and {Im(zn )}n=1 are both Cauchy sequences.

Hence, by our Math113 result, there exists real numbers a and b such
that limn→∞ Re(zn ) → a and limn→∞ Im(zn ) → b. Therefore using our
proposition again, limn→∞ zn → z, where a = Re(z), b = Im(z). 
30

11. Bounded sequences, closed sets and series of complex

numbers

Definition 11.1. A sequence of complex numbers {zn }∞

n=1 is called

bounded if there exists M > 0 such that for any n ≥ 1, |zn | ≤ M .

Proposition 11.1 (Complex Bolzano-Weierstrass). Any bounded se-

quence of complex numbers contains a convergent subsequence.

Proof. Let {zn }∞

n=1 be a bounded sequence of complex numbers. Then

by Proposition (10), both {Re(zn )}∞ ∞

n=1 and {Im(zn )}n=1 are bounded

sequences. So, by the real Bolzano-Weierstrass Theorem (as in Math113),

we have a convergent subsequence {Re(znk )}∞ ∞
k=1 . Although, {Im(znk )}k=1

is not necessarily convergent, but it contains a convergent subsequence

{Im(znl )}∞ ∞ ∞
l=1 . Now, both {Re(znl )}l=1 and {Im(znl )}l=1 are convergent

sequences, hence {znl }∞

l=1 is convergent as well. 

Finally we extend the notion of a closed set for the complex numbers.

A subset L ⊂ C of the complex numbers is closed if for any

convergent sequence {zn }∞
n=1 , limn→∞ zn ∈ L provided that zn ∈ L

for any n ≥ 1.

Closed sets of the complex numbers are, in fact, closed sets in the
(complex) plane, so they are much more interesting geometrically than
closed sets of the real line.
 31

12. Sequences of continuous functions

Revision: [Continuous Functions].

Let f : [a, b] → R be a function. We call f continuous if for any
convergent sequence {xn }∞
n=1 ⊂ [a, b] tending to x, limn→∞ f (xn ) =

f (x).

If f, g : [a, b] → R be continuous functions, so are f + g, f g and |f |.

Any continuous function f : [a, b] → R attains its maximum. That

is, there exists x ∈ [a, b] such that for any y ∈ [a, b], f (x) ≥ f (y).

Now we define the norm of a continuous function which plays exactly

the same role as the modulus of a complex number. Using the norm,
we can define convergence of sequences and of series for continuous
functions as well.

Definition 12.1. Let f : [a, b] → R be a continuous function.

Then, the norm of f is defined as

kf k := maxx∈[a,b] |f (x)| .

Proposition 12.1. Let f, g : [a, b] → R be continuous functions and

λ > 0 be a positive number. Then,

• kf + gk ≤ kf + gk.
• kλf k = λkf k.
32

By definition (and the triangle inequality), kf + gk = |f (x) + g(x)| ≤

|f (x)| + |g(x)| for a certain x ∈ [a, b]. Now, |f (x)| ≤ kf k and |g(x)| ≤
kgk, hence our proposition immediately follows. 

Note that Proposition 12.1 is basically a triangle inequality for continu-

ous functions. Similarly to the triangle inequality for real numbers and
complex numbers. Using Proposition 12.1, we can define convergent
and Cauchy sequences of continuous functions.

Definition 12.2. A sequence of continuous functions {fn : [a, b] →

R} is converging to a continuous function f : [a, b] → R if for any
 > 0 there exists N ≥ 1 such that kf −fn k ≤  provided that n ≥ N .
Also, a sequence of continuous functions {fn : [a, b] → R}∞
n=1 is a

Cauchy sequence, if for any  > 0 there exists N ≥ 1 such that

kfn − fm k ≤  provided that n, m ≥ N .

Similarly to our theorem in Math113 and the analogous result for com-
plex numbers (Exercise 11) one may ask, whether any convergent se-
quence of continuous functions is Cauchy and conversely any Cauchy
sequence of continuous function is converging to some continuous func-
tion. In the rest of the section, we prove that this is indeed true. The
first part is almost exactly the same as the real version in MATH113.

Proposition 12.2. Let {fn : [a, b] → R}∞

n=1 be a sequence of continu-

ous functions converging to a continuous function f : [a, b] → R. Then

the sequence is Cauchy.
 33

Proof. Let  > 0. We need to prove that there exists N ≥ 1 such that

(5) kfn − fm k ≤ 

provided that n, m ≥ N . Let N ≥ 1 be an integer such that kfn −f k ≤


2
provided that n ≥ N . Hence if n, m ≥ N , using triangle inequality
we have that

kfn − fm k ≤ kfn − f k + kfm − f k ≤ 

that is eqrefbecs1 holds. 

The converse statement is not very hard, but I think it is one of the
most important theorem in elementary real analysis. First, let us recall
the notion of uniform continuity from Math113.

Revision: All continuous functions g : [a, b] → R are uniformly

continuous, that is, for any  > 0 there exists δ > 0 such that
|g(x) − g(y)| ≤  provided that |x − y| ≤ δ.

Theorem 8. Any Cauchy sequence of continuous functions {fn : [a, b] →

R}∞
n=1 is convergent.

Proof. Fix x ∈ [a, b]. By definition, for any n, m ≥ 1 we have

(6) |fn (x) − fm (x)| ≤ kfn − fm k .

Pick  > 0. Then, there exists N ≥ 1 such that kfn −fm k ≤ 3 , provided

that n, m ≥ N . You will see later why we use 3
instead of . Therefore
34

by (6),


(7) |fn (x) − fm (x)| ≤
3

as well. Hence, {fn (x)}∞

n=1 is a Cauchy sequence, so by our Math113

theorem it is converging to some number f (x). Taking m → ∞ in (7),

we have that


(8) |fN (x) − f (x)| ≤ .
3

Since (8) holds for all x ∈ [a, b], the following lemma finishes the proof
of our theorem.

Lemma 12.1. The function f : [a, b] → R is continuous.

Proof. Since fN is a continuous function, it is uniformly continuous,

hence there exists δ > 0 such that for all x ∈ [a, b]


(9) |fN (x) − fN (y)| ≤
3

provided that |x − y| ≤ δ. Therefore if |x − y| ≤ δ using triangle

inequality, we have that
(10)
|f (x) − f (y)| ≤ |f( x) − fN (x)| + |fN (x) − fN (y)| + |fN (y) − f (y)| ≤  .

By (10), the function f is uniformly continuous, hence our theorem

follows. 

Exercise 13. Show that the Bolzano-Weierstrass Theorem does not

hold for continuous functions.
 35

Answer: We construct a sequence of positive continuous functions

{fn }∞
n=1 such that

• for any n ≥ 1, kfn k = 1.

• for any n, m ≥ 1, |fn − fm | = 1.

so we have a bounded sequence of continuous functions having no

Cauchy subsequence. The construction is very easy, For n ≥ 1, let
1 1 1 1
f (x) = 0 if x ≥ 2n
or x ≤ 2n+1
. In between 2n+1
and 2n
we define the
function arbitrarily, with the constraint that the function is continuous
and at some point y, fn (y) = 1. Clearly, the sequence {fn }∞
n=1 satisfies

the two conditions above. 

36

13. Series of continuous functions

Definition 13.1. Let {fi : [a, b] → R}∞ i=1 be a sequence of con-

P∞
tinuous functions. Then, the series i=1 fi is convergent if {pk :
Pk
[0, 1] → R}∞k=1 is a convergent sequence, where pk = i=1 fi is the

k-th partial sum. The sum of a convergent series s = ∞

P
i=1 fi is

defined as limk→∞ pk .

The following proposition can be proved in exactly the same way as

Proposition ?? so we omit the proof.

Proposition 13.1. Let {fi : [a, b] → R and {gi : [a, b] → R be

P∞ P∞
sequences of convergent functions such that i=1 fi and i=1 gi are

both convergent series. Then, ∞

P
i=1 (fi + gi ) is a convergent series and
P∞ P∞ P∞
i=1 (fi + gi ) = i=1 fi + i=1 gi .

Definition 13.2. Let {fn : [a, b] → R}∞ be a sequence of conver-

gent functions. We say that ∞
P
i=1 fi is an absolutely convergent
P
series if i=1 kfi k is a convergent series of real numbers.

The following theorem is crucial in real analysis.

Theorem 9. All absolutely convergent series of continuous functions

are convergent. So, its sum is a continuous function.

P∞
Proof. Let i=1 fi : [a, b] → R be an absolutely convergent series as
above and pk : [a, b] → R be the partial sums. It is enough to prove
 37

that {pk }∞
P
k=1 is a Cauchy sequence. Fix  > 0. Since, i=1 kfi k is
convergent, by Proposition 2.4, there exists N ≥ 1 such that for any
l≥1
l
X
kfi k ≤  .
i=N

Let n > m ≥ N . Then by the triangle inequality,

n
X n
X
kpn − pm k = k fi k ≤ kfi k ≤  .
i=m+1 i=m+1

Hence, {pk }∞
k=1 is indeed a Cauchy sequence. Thus our Theorem fol-

lows. 
38

14. Complex continuous functions and their series

So far in Math113 and Math114, we talked about real-valued continu-

ous functions defined on intervals like [a, b] or maybe [a, ∞], [−∞, b],
[−∞, ∞]. In analysis sometimes it is important to consider complex
variables. Here, the role of intervals will be played by closed disks (see
Section 11) BR , where BR = {z ∈ C | |z| ≤ R}

Definition 14.1. A function f : BR → C is called continuous

in z ∈ BR if for any convergent sequence zn → z in BR we have
f (zn ) → f (z). If f is continuous at every single point then we just
call f a continuous function.

You might wonder whether the famous (, δ)-definition extends to com-
plex functions as well. Yes, it does. The point is that the results of
Section 10 and Section 11 help us to convert basically all the Math113
results from real continuous functions to complex continuous functions,
basically word by word. Again, the theory of complex differentiable
functions is a completely different ball game, you will have an im-
portant course on it in Year 2 and this section may be helpful for
understanding that beautiful theory.

Definition 14.2. A function f : BR → C is continuous at

z ∈ BR if and only if for any  > 0 there exists δ > 0 such that
|f (z) − f (w)| ≤  provided that |z − w| ≤ δ. Also, we have uniform
continuity, that is, if f is continuous on BR , then for any  > 0
 39

there exists δ > 0 such that for all pairs v, w ∈ BR , |f (v)−f (w)| ≤ 
provided that |v − w| ≤ δ.

Proposition 14.1. Let f : BR → C be a function. Then, f is contin-

uous on BR if and only if both Re(f ) : BR → R and Im(f ) : BR → R
are continuous functions.

Proof. Let z ∈ BR and let {zn }∞

n=1 ⊂ BR converges to z. By Proposi-

tion 10.1, f (zn ) → f (z) if and only if both Re(f (zn )) → Re(f (z)) and
Im(f (zn )) → Im(f (z)) hold. Hence our proposition follows. 

Exercise 14. Show that if f : BR → C and g : BR → C are continuous

functions then both f + g and f g are continuous functions. Also, if
f : BR → C is continuous w ∈ C or S ⊂ BR is a continuous function,
then both f −1 (w) and f (S) are closed sets.

Proof. Word by word, you can follow the proof of the relevant Math113
theorems. 

Note that what you understood now is the basics of calculus of two
variables, the complex plane is nothing but the 2-dimensional plane,
similar results can be proven without any trouble in 3 or more dimen-
sions.

Convergence and series of complex continuous functions can be defined

in the same way as convergence and series of real continuous functions.
I think it would be a great exercise to write down the definitions without
looking.
40

Definition 14.3 (Norm of a complex function). Let f : BR → C be

a continuous complex function. Then |f | : BW → R is a continuous
function and it attains its maximum at a point z ∈ BR (x). We call
|f (z)| = kf k the norm of f .

It goes without saying that if f : BR → C and g : BR → C are

continuous functions then the triangle inequality holds, that is

kf + gk ≤ kf k + kgk .

Definition 14.4 (Convergence and Cauchyness). Let {fn } : BR →

C}∞
n=1 be a sequence of continuous function. We say that the se-

quence converges to a continuous function f : BR → C if for any

 > 0 there exists N ≥ 1 such that if n ≥ N then kf − fn k ≤ .
On the other hand, the sequence is Cauchy, if for any  > 0 there
exists N ≥ 1 such that if n, m ≥ N then kfm − fn k ≤ .

The following proposition can be proved in the same way as its real
counterpart.

Proposition 14.2. Any convergent sequence of complex continuous

functions {fn } : BR → C}∞
n=1 is Cauchy and every Cauchy sequence is

convergent.

Now we can define series of complex continuous functions and again,

try to write down the definition without looking as a practice exercise.
 41

Definition 14.5. Let {fi } : BR → C}∞i=1 be a sequence of function.

Then the series ∞

P
i=1 fi is convergent if limk→∞ pk converges to a

continuous function s : BR → C, where pk : BR → C is the k-th

partial sum f1 + f2 + · · · + fk and s is called the sum of the series.
If ∞
P
i=1 kfi k is a convergent series of real numbers, then the series

is called absolutely convergent.

Proposition 14.3. Any absolutely convergent sequence is convergent.

42

15. Power series

In the Math101 course you learned about Taylor series. There was
a nice function like sin or ex and you considered expressions like

X
f (x) = f (n) (0)xn ,
n=0

2 3
E.g. ex = 1 + x + x2 + x6 + ... Of course, this was a bit formal, you did
not know (it is not easy at all) why the right hand side of the formula
above is truly equal to the left hand side and why the right hand side
has any meaning.

Look at the very simple example f (x) = 1

1−x
. Then f 0 (x) = (1 −
x)−2 , f ”(x) = 2(1−x)−3 , in general the n-th derivative is n!(1−x)−(n+1) .
Hence the Taylor series is ∞ n 2 3
P
n=0 x or 1 + x + x + x + . . . . This is

the geometric series of Section 1 which is convergent if |x| < 1 and

1
divergent if |x| ≥ 1 (Proposition 1.1). If x = 1, then 1−x
is meaningless.
Miraculously, the derivatives ”feel” this fact at the zero.

• You build the Taylor series using derivatives at the zero.

1
• The Taylor series is convergent to the function f (x) = 1−x
if
|x| < 1.
• Although f (2) is completely meaningful and equals to −1, the
Taylor series unable to look ”behind” 1.

You will learn more about Taylor series in Complex Analysis and un-
derstand more about this phenomenon. In this course we will consider
series in the form n=0 cn xn , where cn is an arbitrary real number and
P

even series like n=0 an z n , where dn is complex and z is complex vari-

P

ables. We interested only in the convergence of this kind of series we

 43

will not show that the sum of the series is a nice differentiable func-
tion and the series is the Taylor series at 0. It is a way much more
cumbersome task.

So, let us start with all basic theorem, which can be proved using the
propositions we had before.

P∞
Theorem 10. Let n=0 an z n be a complex power series (of course, it
is possible that an are real numbers, do not forget that real numbers are
just very special complex numbers). Then exactly one of the following
occurs.

(1) The series converges absolutely for all z ∈ C.

(2) ∃R ≥ 0, such that the series is absolutely convergent whenever
|z| < R and divergent whenever |z| > R.

Proof. Let
∞
X
E = {z ∈ C : an z n is convergent} and
n=0

R = sup{|z| : z ∈ E} ∈ [0, ∞],

where R = ∞ means that this set is not bounded above. In words, E

is the set of all values z for which the series convergent. Our goal is to
prove that if |z| < R, then z ∈ E. By definition, the series is divergent
if |z| > R. We will not make any claim if |z| is equal to R. Note that
0 ∈ E, and (2) holds with R = 0 if E contains no other number, that
is, if E = {0}. We will see examples, when this occurs.

Suppose E contains some element z0 6= 0. By the n-th term test

(Proposition 2.1), an z0n → 0 as n → ∞. Hence |an z0n | is bounded,
say by M .
44

We claim that if |z| < |z0 | then z ∈ E. To see this, note that

z n z
|an z n | = |an z0n | | | ≤ M | |n .
z0 z0

Now,
∞
X z
| |n
n=0
z0
is convergent, being a geometric series with common ratio less than 1.
P∞
Hence n=0 |an z n | is convergent, by the simple comparison test (Propo-
sition 2.1), and so z ∈ E.

If E is the whole of C, we have case (1). Otherwise we set

∞
X
R := sup({r , ∃z ∈ C such that r = |z| and the series an z n converges}),
n=0

and the argument above shows that (2) holds. This finishes the proof
of our theorem. 

The number R is called the radius of convergence of the series

P∞ n
n=1 an z . Before getting further let us make some revision of Math113.

Revision: For any k ≥ 1, l > 1,

kn
• limn→∞ n!
= 0.
nk
• limn→∞ ln
= 0.

Exercise 15. Show that there are examples of power series, where the
radius of convergence is 0 or ∞.

1 n
P
Answer: Let us consider the power series n=0 n! z . Then, the se-
quence is absolutely convergent for all values of z ∈ C, since for large
 45

enough n,
1 n
| ||z | ≤ 2−n .
n!

Therefore the radius of convergence is ∞.

n!z n . Then, the n-th terms

P
Now, let us consider the power series n=0

tend to infinity for any choice of z > 0. Hence, the radius of convergence
is 0.

Exercise 16. Show that there exist power series where the radius of
convergence is 1 and the power series converges for all values |z| = 1.

Answer: Let us consider the power series

X
n−2 z n .
n=1

If |z| > 1, then the n-th terms tend to infinity. If |z| = 1, then

X
n−2 |z n |
n=1

is absolutely convergent by Theorem ??. Hence our series satisfies the

required condition.

Exercise 17. Show that there exist power series where the radius of
convergence is 1 and the power series converges for no values |z| = 1.

Answer: Let us consider the power series

X
nz n .
n=1

If |z| < 1 then the sequence is convergent by the ratio test. If |z| > 1
then the n-th terms tend to ∞. Also, if |z| = 1, the n-th term has
modulus n, so they tend to ∞ as well.
46

16. The sum of power series

an z n be a power series with radius of convergence R. Let

P
Let n=0

us summarize, what we know about this power series. If |z| > R,

then the series is divergent. If |z| = R lot of things could happen.
However, if |z| < R, then the series is convergent. Therefore the sum
is a well-defined function on the set {z : |z| < R}.

an z n be a power series with radius of

P
Proposition 16.1. Let n=0

convergence R > 0. Let 0 < S < R be a real number. Then the sum of
the power series defines a continuous function on the set BS .

Proof. For |z| ≤ S we have |an ||z n | ≤ |an |S n . We claim that |an |S n
P
n=0

is a convergent series of real numbers. Indeed, this explicitely showed

in the proof of Theorem 10. Hence, the series is absolutely convergent
on BS . That is, our theorem immediately follows from Theorem 9. 

Assume that for n ≥ 1 an is a real number. Then the sum of the series
above defined a continuous real function f on the interval [−S, S]. The
following theorem will be proved in Year 2. The theorem entails that
power series are always Taylor series.

Theorem 11 (Taylor series). Let f : [−S, S] → R be the continuous

f (n)
function as above. Then all the derivatives of f exist and n!
(0) = an .
 47

17. The exponential function

In Section 8 we defined e as the limit of the sequence (1 + n1 )n . Fur-

thermore, we showed (Exercise 10) that for any q ∈ Q we have that eq
is the limit of the sequence (1 + nq )n .

Finally, we also showed in Theorem 7 that e is equal to the sum of the

convergent series ∞ 1
P
n=0 n! and by Proposition 9.1, e is equal to the sum
qn
of the convergent series ∞
P
n=0 n! .

Also, by Exercise 15, the radius of convergence of the power series

P∞ z n
n=0 n! equals to the infinity. Now we are ready to define the famous

exponential function.

Definition 17.1. We define the exponential function exp : C → C by

setting exp(z) to be equal to the sum of the series
∞
X zn z2 z3
=1+z+ + + ···
n=0
n! 2! 3!

Note that by Theorem 9, we immediately see that the exponential

function is continuous. Of course, we mostly use the real exponential
function, that is, the restriction of exp on to the real line. Note that
the exponential function is sometimes denoted by ex .

Remark 17.1. Seemingly, e2 could have two different meanings: e · e

and the sum of the series ∞ 2n
P
n=0 n! . Note however, that by Proposition

9.1 these two quantities are equal.

In calculus, you learned without proof that ex ey = ex+y . Finally, we

are in the position to give a rigorous proof of this important fact.

Proposition 17.1. For all x, y ∈ C we have ex ey = ex+y .

48

Proof. Fix x, y ∈ C. Let ∞

P
n=0 cn be the Cauchy product (see Section
P∞ y n
??) of the series ∞ xn
P
n=0 n! and n=0 n! . Then we have

n n  
X xk y n−k 1 X n k n−k (x + y)n
cn = = x y = ,
k=0
k!(n − k)! n! k=0 k n!

by the binomial theorem. The result now follows from Corollary ??. 

Remark 17.2. Note that ex is positive and strictly increasing when

x ∈ R+ , since each term of the series is positive and strictly increasing.
The previous proposition shows in particular that e−x = 1/ex , hence ex
is in fact positive and strictly increasing for all x ∈ R. Furthermore,
for all x ∈ C and n ∈ N we have (ex )n = enx .
 49

18. The trigonometrical functions

There is a perhaps surprising and deep connection between the ex-

ponential function and the trigonometric function sin(x) cos(x) (here
with sin(x) and cos(x) we mean the standard trigonometric functions
defined using right triangles). The source of this connection is the
following theorem.

Theorem 12. (Euler’s formula) For any θ ∈ R we have

cos(θ) + i sin(θ) = exp(iθ)

Proof. First note that

∞
θ2 θ4 X x2n
cos(θ) = 1 − + − ··· = (−1)n
2! 4! n=0
(2n)!

and
∞
θ3 θ5 X
n x
2n+1
sin(θ) = θ − + − ··· = (−1) .
3! 5! n=0
(2n + 1)!
What is clear that the radius of convergence of both series are equal
d d
to the infinity. Also, by dθ
sin(θ) = cos(θ) and dθ
cos(θ) = − sin(θ) the
series are really the Taylor series of the respective functions. Later you
will learn that in fact the sum of the Taylor series as a power series
is exactly the respective trigonometrical function. Now by the power
series definition of the exponential function.
∞
X (iθ)n
exp(iθ) = .
n=0
n!

Using the fact that i4n = 1, i4n+1 = i, i4n+2 = −1, i4n+3 = −i, we have
that
∞ ∞
X x2n X x2n
exp(iθ) = (−1)n + i( (−1)n ), .
n=0
(2n)! n=0
(2n)!
50

Hence our Theorem follows. 

Substituting, θ = π we obtain the famous Euler identity, an equation

containing i, e and π:
−1 = eiπ .