Compiled by Abraham Negussie.

Revised on 2007

Applied Mathematics II (Math 232B)

Lecture Note
(For Technology Faculty Students, AAU)

1. Improper Integrals
Definition: An integral
b
∫ a
f ( x ) dx

is called an improper integral if either the limit of integration is infinite or when the integrand f (x) is
singular (discontinuous) somewhere in the range a ≤ x ≤ b.

I. Suppose f (x) is continuous on the interval [a, b] except at the point c where a ≤ c ≤ b, then

b d b
∫a
f ( x ) dx = lim +
d →c ∫
a
f ( x ) dx + lim−
g →c ∫ g
f ( x ) dx .

II. Improper integrals with infinite limit of integration.

If f (x) is continuous on the interval [a, ∞) , then

∞ b
∫ a
f ( x ) dx = lim
b→∞ ∫
a
f ( x ) dx .

If f (x) is continuous on the interval (-∞, b] , then

b b
∫ −∞
f ( x ) dx = lim
a→−∞ ∫ a
f ( x ) dx .

If f (x) is continuous on the interval (-∞, ∞) , then

∞ b c
∫−∞
f ( x ) dx = lim
b→∞ ∫
c
f ( x ) dx + lim
a→−∞ ∫a
f ( x ) dx , for some c∈R usually 0.

In each case, if the limit exists, then the improper integral is said to converge, otherwise the improper
integral is said to diverge.

1
Eg1. Check if ∫ ln x dx
0
is convergent or divergent.

This improper integral represents an unbounded area, as illustrated by Fig. 0.1. Hence if the
improper integral is convergent, then the area is finite; else if the improper integral diverges, then the
area is infinite.

1
Solution: ln x is discontinuous at x = 0. Thus,

1 1
∫0
ln x dx = lim +
b→0 ∫ ln x dx
b

1
⎡ ⎤
= lim + ⎢ x ln x − x ⎥
b→0 ⎣ ⎦b
= lim + (-1 – b lnb + b)
b→0

= -1.
Hence the improper integral converges. Here one has to use L′Hopital’s rule to evaluate
lim + b lnb, which is equivalent to
b→0
1
lim + ln1 b = lim + b
−1
= lim + -b = 0.
b→0 b→0 b→0
b b2

y = ln x
(1,0)

Fig.0.1 : Unbounded area above the graph of ln x and below the x – axis on the interval [0, 1].

b
∞ b
1 dx = lim ⎡ 1 ⎤ = lim ( − 1 + 1) = 1.
∫ ∫
Eg2. 1 dx = lim
1 x2 b→∞ 1 x2 b→∞ ⎢
⎣ x ⎥⎦ 1 b→∞ b

b
∞ b
1 dx = lim ⎡ ln x ⎤ = lim (lnb – ln1) = ∞.
∫ ∫
Eg3. 1 dx = lim
1 x b→∞ 1 x b→∞ ⎢
⎣ ⎥⎦ b→∞
1

Notice the difference between the improper integrals in the above examples, Eg2 is converges where
as Eg3 diverges. The following example is a general case of the above examples and it is called the
p - integral .

2
 ⎧ 1 , if p > 1
∞
1 dx = ⎪ p − 1
Eg4. ∫ xp
⎨ .
⎪∞ ( diverges ) , if p ≤ 1
1

Solution: The case p = 1 is already treated in Eg2.

Now assume p ≠ 1, and then

b b
∞ b
1 dx = lim ⎡ x − p +1 ⎤ = lim ⎡ ⎤
∫ ∫
1 dx = lim 1
b → ∞ ⎢ − p +1 ⎥
⎢ − ⎥
⎣ ( − p + 1) x
p
1 xp b→∞ 1 x p
⎣ ⎦1 b→∞ 1
⎦1
⎧ 1 , if p > 1
⎪ p −1
= ⎨ .
⎪∞ , if p < 1
⎩

If the limit of integration is changed to [0, 1], then the condition for convergence posed on the value
of p will switch as given by Eg5. below.

⎧∞ ( diverges ), if p ≥ 1
1
1 dx = ⎪
Eg5. ∫ xp
⎨ 1
, if p < 1
.
⎪1 − p
0

Proof: Exercise!

Exercises 1.1: Evaluate the following improper integrals:

4 ∞
∫ ∫
1. 1 dx 2. 1 dx
0 ( x − 1) 2 0 x +4 2

4. ∫ ln x dx
1 ∞
3. ∫ 1 dx
0 x 1 x

5. ∫
∞
e x dx
−∞ 1 + e 2 x

Comparison Test.
Theorem 1.1:
b
Assume that the proper integral ∫ a
f ( x ) dx exists for each b ≥ a and suppose that f (x) ≥ 0 for a11
∞
x ≥ a. Then ∫a
f ( x ) dx converges if and only if there is a constant M > 0 such that:

b
∫ a
f ( x ) dx ≤ M for every b ≥ a.
This theorem forms the basis for the following comparison tests.

3
Theorem 1.2:
Suppose that f and g are continuous functions and 0 ≤ f (x) ≤ g (x) for all x ∈ [a, ∞).

∞ ∞
a) If ∫
a
g ( x ) dx converges, then ∫
a
f ( x ) dx converges.
∞ ∞
b) If ∫ f ( x ) dx diverges, then ∫ g ( x) dx diverges.
a a

∞
∫
Eg: Evaluate 1 dx .
1
1+ x 3
Solution:
1 ≤ 1 = 1 .
3
1+ x 3 x3 x2

But
∞ 1
∫ 1
3
x2
dx

is a p – integral with p = 3 > 1 and hence converges by p integral test in Eg4.

2
∞ 1
Therefore by the above comparison test theorem the improper integral ∫ 1
1 + x3
dx converges.

Exercise: (To be submitted!):

1. Consider the improper integral
∞
∫ 0
e − x xα − 1 dx , where α ∈ R.
a. For what value(s) of α does the improper integral converge?
b. Compute the integral when α = n ∈ N. i.e when α is a natural number.

∞ ln x
2. Find the values of α for which the improper integral ∫ 1 xα
dx converges, and evaluate the
integral for those values of α .

4
Compiled by Abraham Negussie. Revised on 2007

2. Sequences and Infinite Series

2.1. Sequences
Definition: A sequence is a function whose domain is the set of non - negative integers.

1, 1 , 1 , 1 , . . . , 1 , . . . can be written as { 1n }n =1 . The function here is

∞
Eg1. The sequence
2 3 4 n
f :N → R defined by f (n) = 1 .
n
In general given a sequence { a n }n =1 a n is called the term of the sequence and n is called the index.
∞

Definition: (Limit of a sequence.) A number L is called the limit of a sequence { an }∞n=m if and
only if
for very ε > 0 there is N such that

n≥N ⇒ an − L < ε .
In this case we write

lim a n = L .
n →∞

{ an }n=m
∞
Definition: We say a sequence converges if lim a n = L ∈ R. (i.e. if the limit exists). On
n →∞
the other hand the sequence is said to be divergent if the limit does not exit.

{ 1n }n=1 .
∞
Eg2. Consider lim 1n = 0 , since Given ε > 0 we can choose N > 1 from which we get
n→∞ ε
n≥N ⇒ 1 ≤ 1
n N
⇒ 1 ≤ 1 < ε (since N > 1 )
n N ε
⇒ 1 < ε
n
⇒ 1 −0 <ε ( since n > 0, 1
= 1
)
n n n

Thus the sequence converges to 0.

Eg2. Show that the alternating sequence { (−1) n }

∞
diverges.
n =1

5
 ⎧⎪ 1 for n even
n
Proof: (-1) = ⎨ .
⎪⎩ − 1 for n odd
We proceed by proof by contradiction. Assume that lim( −1) n exists, say L; then by definition of
n →∞
limit:
for ε =1 there exists an N such that

n≥N ⇒ ( −1) n − L < ε = 1 . (*)

Case 1: If L ≥ 0 when n is odd we have

( −1) n − L = − 1 − L ≥ 1 ,

which is contradicts ( * ).

Case 2: If L < 0 when n is even we have

( −1) n − L = 1 − L > 1

again this contradicts ( * ) .

In either case we arrive at a contradiction. This implies that our assumption was wrong. Therefore
lim( −1) n does not exist and hence the sequence {( −1) n }∞
n =1
diverges.
n →∞

Eg3. Show that the Geometric Sequence { r n }

∞
converges for 0 ≤ r ≤ 1 and diverges for r > 1.
n =1
Proof: Exercise!
{( ) } { (1) } { (1.2) }
∞ ∞ ∞
1 n n n
Applying Eg3, the sequences 2 and are convergent while is
n =1 n =1 n =1
divergent.

Exercise 2.1: Check whether the following sequences converge or not.

{ c} { n} { }
∞ ∞ ∞
n n ln n
1. for c > 0. 2. 3. n 2
n =1 n =1 n =1

{ } { } 6. { (1 + n2 ) n }
n ∞ ∞ ∞
e 8n
4. n 5. 2 n + 3 n =1
n =1 n =1

7. { (1 + }
∞
1 n
3n
) (Hint: Apply L’Hopital’s rule to find the limit)
n =1

Theorem 2.1: (Convergence properties of sequences)

Let { an }n =1 and { bn }n =1 be convergent sequences. Then

∞ ∞

{ an + bn }n=1 ,
∞

6
 { can }n=1 where c is a scalar ,
∞

{ an bn }n=1
∞

and
{ }
an ∞
bn
where lim bn ≠ 0
n =1 n→∞
converge.

The proof is a direct application of the limit sum, product, quotient and scalar multiple theorem.

Note: { an }n =1 converges if and only if { an + k }

∞ ∞
converges. That is lim a n = lim a n + k .
n =1 n→∞ n→∞

Proof: Exercise! (Hint: use ε - δ definition of limit)

{ an }n=1
∞
By the above note we are saying that all the properties of the sequence holds true if we re-
{ an }n =m . { an }n =m
∞ ∞
index it as in the form Observe here that the terms the sequence are
a m , a m +1 , a m + 2 ,

Bounded Sequences:

Definition: A sequence { a n }∞n=1 is said to be bounded if there exists a non-negative number M such
that
a n ≤ M for each n = 1, 2, 3, . . .

Theorem 2.2:
a) If a sequence { a n }∞n =1 converges, then it is bounded.
b) If a sequence { a n }∞n =1 diverges, then it is unbounded.

Proof:
The two statements are equivalent. It then suffices to prove (a).
Suppose lim a n = L , then
n →∞

Given ε > 0 there is N such that

n≥N ⇒ an − L < ε .
which is equivalent to
n≥N ⇒ an − L ≤ an − L < ε .
which is equivalent to
n≥N ⇒ − ε < an − L < ε .
which is equivalent to
n≥N ⇒ L − ε < an < L + ε .

7
which implies
a N < L + ε , a N +1 < L + ε , a N + 2 < L + ε ,

Now choose M = max {a1 , a2 , , a N −1 , L + ε }.

Then we get that

an ≤ M for each n = 1, 2, 3, . . .

Thus { a n }∞n =1 is bounded.

Eg. The series {2 }

n ∞
n =1
and { ln n }n=1
∞
are unbounded and hence are divergent by the above
theorem.

Remark: The converse of the above theorem doesn’t hold true. That is a bounded sequence might
diverge. For example consider { (−1) n }
∞
which is bounded by 1 but does not converge.
n =1

Monotonic Sequences

{ an }n =1 is said to be:
∞
Definition: A sequence
a. increasing if a n +1 ≥ a n for all n ≥ 1.
b. decreasing if a n +1 ≤ a n for all n ≥ 1.

If a sequence is either increasing or decreasing, then it is called Monotonic.

Note: A sequence { a n }∞n =1 is strictly increasing if a n +1 > a n for all n ≥ 1.

Exercises 2.2: Check whether the following sequences are monotonic.

c) { (−1) n }
∞
a) { 1 − 1n }n =1 { Sin 1n }n=1
∞ ∞
b)
n =1

{ } { }
∞ ∞
{ Sinnπ }n=1
3n − 1 ∞ n3 − 1
d) 4n + 5 e) f) n
n =1 n =1

{ } { }
∞ ∞
n2 n
g) 2n
h) n
3 +1
n =1 n =1

Theorem 2.3: A bounded and monotonic sequence converges.

8
Proof: The case when the sequence { an }n =1 is increasing and bounded has been proved. The proof
∞

of the case when the sequence is decreasing and bounded follows immediately by just
considering {− an }n = 1 .
∞

In this regard suppose { an }n =1 is increasing and bounded. Then there is an upper bound for the
∞

sequence, by least upper bound axiom the sequence has a least upper bound say B. By the property
of least upper bound

Given ε > 0 B - ε can not be upper bound for the sequence. That is to mean

there is N such that B − ε < aN . (1)

But B is an upper bound thus we have:

an ≤ B for each n = 1, 2, 3, . . . (2)

Moreover, since { a n }∞n =1 is increasing

a N < an for all n ≥ N. (3)

Combining the inequalities in (1), (2) and (3) we get:

B − ε < aN < an ≤ B < B + ε whenever n ≥ N.

Which implies

B − ε < a n < B + ε if n ≥ N.

That is equivalent to

n≥N ⇒ an − B < ε .
We then conclude that lim a n = B , and hence the sequence converges.
n →∞

{ }
∞
2n
Eg1: Use the above theorem to prove that the sequence n! is convergent.
n =1

2n 2n + 1
Solution: The terms of the sequence are 2, 2, 4
3 , 2
3 ,..., n! , ( n + 1)! ,…
We show the sequence is decreasing. That is:

2n + 1
( n + 1)! ≤ 2n
n! ,
equivalently

9
Compiled by Abraham Negussie. Revised on 2007

2 ≤ (n+1)

which holds true for all n ≥ 1.

Now since the sequence is decreasing 2n

n! ≤ a1 = 2. Hence the sequence is bounded by 2. Therefore
by the above theorem it converges.

Exercise: Show that the sequence { an }n =1 defined by the recursive relation:

∞

a 0 = 0 and a n + 1 = a n 2 + 1
4 converges.

Subsequence and limit point of a sequence

{ an }n=1 be a sequence and let { nk }k =1

∞ ∞
Definition: Let be a sequence of positive integers such that,

{a }
∞
{ nk }k =1
∞
nk < nk + 1 , for each k, (that is let be an increasing sequence). Then nk is called a
k =1

subsequence of the sequence { a n }∞n =1 .

Eg1. Consider the sequence{ 1n }∞n=1 . If we let nk = 2k for each positive integer, then {21k }= {21n } is
the corresponding subsequence of { 1n }n =1 . Note that the terms of this subsequence are 12 , 14 , 16 , …
∞

Remark: 1.One can also prove that { 31n }∞n=1 { 1n }∞n=1 . And { 61n }∞n=1
is a subsequence of is a
subsequence of { 31n }n =1 . In this case { 61n }n =1 can be called a sub-subsequence of { 1n }n =1 .
∞ ∞ ∞

{ an }n=1
∞
2. Since for n k = k , n k = k < k + 1 = nk + 1 , every sequence is trivially a subsequence of
itself.

{ an }n =1 { an }n=1
∞ ∞
Theorem 2.4: If a sequence converges to L then every subsequence of converges
to L.

{a } be a subsequence of { a n }∞n =1 .
∞
{ an }n=1
∞
Proof: Assume the sequence converges to L and let nk k =1

{ nk }k =1
∞
Since the sequence of positive integers is increasing, n1 ≥ 1 n 2 > n1 = 1 i.e. n2 ≥ 2 and
{ an }n=1
∞
hence by induction n k ≥ k for each positive integer k. From the fact that converges to L,

10
given ε > 0 there exits a natural number N such a n − L < ε whenever n > N . Thus, whenever

{ }
∞
n k ≥ k > N, we have a nk − L < ε . Hence the subsequence ank converges to L.
k =1

{ an }n=1 has two subsequences which converges to different limits,

∞
Corollary 2.4.a. If the sequence
then the sequence { an }n =1 diverges.
∞

Corollary 2.4.b. If the sequence { a n }∞n=1 has a subsequences which diverges then, the sequence
{ a n }∞n=1 diverges.
Eg2. Consider the sequence { a n }∞n =1 = {( −1) n }n =1 and its subsequences { a 2 n }∞n =1 = {1 }∞n =1 and
∞

{a } ∞
2 n − 1 n =1 = { − 1 }∞n =1 . { a 2 n }∞n =1 converges to 1 and a 2 n − 1 { }
∞
n =1
converges to -1. Thus by corollary
2.4.a. the sequence {( −1) }
n ∞
n =1
diverges.

Eg3. Let a n = sin( n4π ) for each natural number n. Let n k = 4k − 2 for each natural number k. Then
the subsequence {a } ∞
nk k =1 = {a } ∞
4 k − 2 k =1 = {sin (
( 4 k − 2 )π
4 )}
∞
k =1
= {(−1) } n ∞
n =1
. Therefore by corollary
2.4.b. the sequence { sin( n4π )}∞n=1 divergent since it has a divergent subsequence.
Definition: (Limit Point )
Let { a n }∞n =1 be a sequence. The number a is called a limit point of { a n }∞n =1 if and only if there is a
subsequence of { a n }∞n =1 converging to a.

Eg4. Given { a n }∞n =1 = {1n }∞n=1 . {1n }∞n=1 is a subsequence of itself and since it converges to 0, 0 is a
limit point of {1n }∞n=1 .

Eg5. Consider the sequence { a n }∞n =1 = {( −1) n (1 + 1n ) }n =1 . The limit points of this sequence are 1 and
∞

-1. Since the subsequence { a 2 n }∞n=1 converges to 1. ( i.e. lim a 2 n = lim(1 +

n→∞ n→∞
1
2n ) = 1 ) and

subsequence a 2 n + 1{ } ∞
n =1
converges to -1. ( i.e. lim a 2 n + 1 = lim( −1)(1 +
n→∞
1
2n +1 ) = −1) .
n→∞

Exercise: Find all limit points.

a. { cos( n2π }n =1 b. {(1 − 1n ) sin( n2π ) }n =1
∞ ∞

11
2.2. Infinite Series
∞
Let { a n }∞n =1 be a sequence. The expression ∑a n or a1 + a 2 + a3 + … + a n + … is called a
a =1

series (infinite series).

The sums
S1 = a1 (the first partial sum)
S 2 = a1 + a 2 (the second partial sum)
S 3 = a1 + a 2 + a3

S n = a1 + a 2 + a3 + … + a n (the nth partial sum)

are called partial sums. The sequence { S n }∞n =1 is called the sequence of partial sums.
∞
Definition: Given the infinite series ∑a n .If the sequence of partial sums { S n }∞n =1 converge to a
a =1

finite limit L, then the series is said to converge to L, written

∑a
a =1
n = L. (The number L is called the sum of the series.)

∞
If the sequence of partial sums diverges, then the series ∑a
a =1
n is said to be divergent.

Remark: 1. Note that the sum of the series is not a sum in ordinary sense. It is a limit.

∞
2. For the series ∑a
a=m
n , the partial sums are

S1 = a m
S 2 = a m + a m +1
S 3 = a m + a m +1 + a m + 2

S n = a m + a m +1 + a m + 2 + … + a m + n − 1
∞
Eg.1: Show that ∑
n =1
1
2n
converges.

Solution:

For any n ≥ 1 the nth partial sum is S n = 1

2 + 1
22
+ 1
23
+…+ 1
2n
.

12
This implies that

2 Sn = 1 + 1
2 + 1
22
+…+ 1
2 n −1
= 1 + ( 12 + 1
22
+…+ 1
2 n −1
+ 1
2n
)- 1
2n
= 1 + Sn - 1
2n
.

From which we get:

2 Sn = 1 + Sn - 1
2n

Thus the nth partial sum

Sn = 1 - 1
2n
Hence

lim S n = lim 1 − 21n = 1.

n→∞ n→∞

That means the sum of the series is 1. i.e.

∑
n =1
1
2n
= 1.

∞
Eg2. Show that ∑ 2n diverges.
n =1
Solution:
m −1 m −1
The mth partial sum S m = 2 + 2 2 + 2 3 + 1… + 2 ≥ 2 And since lim 2 m − 1 = ∞ the
n→∞
∞
sequence of partial sums diverge or lim S m does not exist. Therefore the series
m→∞
∑ 2n diverges.
n =1

∞
Eg3. Show that the telescoping series ∑
n =1
1
n ( n + 1) = 1
1⋅ 2 + 1
2⋅3 + 1
3⋅ 4 + + 1
n ( n + 1) + converges and

find its sum.

Solution:
1
n ( n + 1) = 1
n − 1
n +1 .
Which implies that
S m = (1 - 1
2 ) + ( 12 - 13 ) + ( 13 - 1
4 ) + … + ( m1 − 1
m +1 ) =1 − 1
m +1 .
Thus
lim S m = lim (1 − 1
m +1 ) = 1.
m→∞ m→∞

Therefore

13
 ∞

∑
n =1
1
n ( n + 1) = 1.

Exercises 3.1:

a. Show that
∞ ∞ ∞
i. ∑
n =1
1
n ( n + 1)( n + 2 ) = 1
4 ii. ∑
n =1
1
n ( n + 1)( n + 2 )( n + 3 ) = 1
18 iii. ∑
n =1
1
( 2 n − 1) ( 2 n + 1) = 1
2

b. find the sum of the following series if it converges

∞ ∞
i. ∑
n =1
5
( 3 n + 1)( 3 n − 2 ) ii. ∑
n =1
2
( 4 n − 3 )( 4 n + 1) .

∞
c. Let { a n }∞n = 1 be a sequence, and S m be the mth partial sum of the series ∑ (a n − an +1 ) .
n =1

i. Show that S m = a1 − am + 1 .
∞
ii. Show that the series ∑a n converges if and only if the sequence { a n }∞n = 1 converges.
n =1
iii. Incase the series converges find its sum.

∞
Eg4. The Harmonic Series ∑
n =1
1
n diverges.

We will show that S 2m ≥ 1 + m( 12 ) for all m = 1, 2, 3, … .

Indeed S 21 = S 2 = 1 + ( 12 )
1
2

S 22 = S 4 = 1 + 1
2 + 1
3 + 1
4 ≥ 1+ 1
2 + 1
4 + 1
4 = 1 + 2( 12 )
1 1
2 2

S 23 = S 8 = 1 + 1
2 + 1
3 + 1
4 + 1
5 + 1
6 + 1
7 + 1
8 ≥ 1+ 1
2 + 1
4 + 1
4 + 18 + 18 + 18 + 18 = 1 + 3( 12 )
and

S 2m = 1+ 12 +( 13 + 14 )+( 15 + 16 + 17 + 18 )+ … + ( 2 m −11 +1 + 2m −11 + 2 +…+ 21m )

1 1 1
2 2 2

≥ 1+ 12 + 14 + 1
4 + 18 + 18 + 18 + 18 + … + ( 21m + 1
2m
+ + 1
2m

= 1 + m( 12 ) .

But

14
Compiled by Abraham Negussie. Revised on 2007

lim S2m = lim (1 + m( 12 ) ) = ∞, and hence by corollary 2.4.b. lim S n does not exist.
m→∞ m→∞ n→∞

Consequently the harmonic series diverges.

∞ ∞
REMARK: Let m > k . The series ∑a
a=m
n converges if and only if the series ∑a
a=k
n converges.

∞
Exercise 3.2: Show that ∑
n=5
1
n
diverges.

∞
Theorem 2.4: If ∑a
n =1
n converges, then lim a n = 0 .
n→∞

Proof: For n ≥ 1 a n = ( a1 + a 2 + a3 + … + a n − 1 + a n ) – ( a1 + a 2 + a3 + … + a n − 1 )
= S n - S n −1
∞
Since by hypothesis ∑a
n =1
n converges lim S n and lim S n −1 exist and are equal.
n→∞ n→∞

Hence
lim a n = lim( S n − S n − 1 ) = lim S n - lim S n −1 = 0.
n→∞ n→∞ n→∞ n→∞

∞
Corollary: If lim a n is not 0 (which includes limit does not exist), then
n→∞
∑a
n =1
n diverges.

Geometric Series Theorem

Theorem 2.5: Let r be any number and let c ≠ 0 and m ≥ 0. Then the geometric series
∞ m
∑ cr n
= cr (i.e. converges) if and only if |r|<1.
n=m
1− r

Proof: If |r| ≥ 1, then |c r n | ≥ |c| for all n ≥ m , then the limit of the terms is not zero. Hence
by the above corollary the series diverges.

If | r |<1, then considering ( 1 – r )( 1 + r + r 2 + … + r n - 1 ) = 1 – r n that is

1 + r + r 2 + … + r n - 1 = 1− r
n

1− r

15
from which we get the nth partial sum is

= c r m (1 + r + r 2 + … + r n – 1 ) = cr m (1 − r ) .
n
S n = c r m + c r m + 1 + c r m + 2 + … + c r m+ n-1
1− r

Thus
lim S n = cr m lim(1 − r n ) = cr m since for r < 1 lim r m = 0
n →∞ 1 − r n →∞ 1− r n →∞

∞
5( 109 )
Eg: ∑ 5( 109 ) n is a geometric series with r =
n =1
9
10 . Hence it converges to
1 − 109
= 45 .

Exercise 3.3:
1. Prove the following.

∞ ∞
a. ∑
n =1
2
3n − 1
= 3 b. ∑
n=2
1
n2 − 1
= 3
4

∞ ∞

∑ ∑
2n + 3n n +1 − n
c. 6n
= 2
3 d. = 1
n2 + n
n=2 n =1

2. Determine if the infinite series converges:

∞ ∞ ∞ ∞
a. ∑
n =1
2n − 1
3 n b. ∑
n =1
5
2
n −1 c. ∑ ln
n =1
n
n +1 d. ∑(
n =1
1
2n
− 1
3n
)

3. A ball is dropped from a height of 12m. Each time it strikes the ground, it bounces back to a
height of three-fourth of the distance from which it fell. Find the distance traveled by the ball before
it comes to rest.

Properties of convergent series

∞ ∞ ∞ ∞
a. If ∑ an and
n =1
∑ bn are convergent, then both
n =1
∑ (an + bn ) and
n =1
∑ (a
n =1
n − bn ) converge

and
∞ ∞ ∞

∑ (a
n =1
n ± bn ) = ∑a
n =1
n ± ∑b
n =1
n

∞ ∞
b. If ∑a
n =1
n converges and c is any number, then ∑ca
n =1
n converges and

16
 ∞ ∞

∑ c an = c ∑ an .
n =1 n =1

Nonnegative series is a series whose terms are nonnegative. The sequence of partial sums { S n }∞n =1
of a nonnegative series is increasing. That is S n + 1 ≥ S n for all n ≥ 1.

Here after we will study different tests for convergence or divergence of nonnegative series and
alternating series.

The following Integral Test compares a nonnegative series with an improper integral.

Integral Test
∞
Theorem 2.6: Let ∑a
n =1
n be a nonnegative sequence and f be continuous decreasing function

defined on [1, ∞) such that f (n) = a n for n ≥ 1. Then The series

∞ ∞
∑a n =1
n converges if and only if the integral ∫ f ( x) dx
1
converges.

Proof: Consider the following figure 3.1.

Fig: 3.1. Proof of the integral test

Since f is decreasing and f (n) = a n comparing f with the appropriate step functions we see that

n
0 ≤ a 2 + a3 + … + a n ≤ ∫ f ( x) dx
1
for n ≥ 2. (1)
and
n
∫ f ( x) dx
1
≤ a1 + a 2 + a3 + … + a n − 1 for n ≥ 2. (2)

17
 ∞ ∞ ∞
From (1) ∑ an converges if the integral
n=2
∫
1
f ( x ) dx converges and hence ∑a
n =1
n converges if the
∞
integral ∫ f ( x) dx .
1
∞ ∞
From (2) ∑a
n =1
n diverges if the integral ∫ f ( x) dx
1
diverges. Which completes the proof.

∞
Eg1. Using integral test show that Harmonic Series ∑
n =1
1
n diverges.

Solution:
Consider the function f (x) = 1
x for all x ∈ [1, ∞ ) and from figure 3.2. below we have that

n n

∫ 1
1 x
dx ≤ ∑
i =1
1
i for all n ≥ 1.

3.2

∞ ∞
But ∫ 1
1 x
dx diverges so by integral test ∑
i =1
1
i diverges.

∞
Eg2. Show that the p-series ∑n =1
1
np
converges if and only if p > 1.

Solution: Compare it with the p – integral .

Exercises 3.4: Use integral test to determine whether the series converges or diverges.

18
 ∞ ∞ ∞
a. ∑
n =1
1
n ln(1 + 1n ) b. ∑
n=2
1
n2 − 1
c. ∑
n =1
n3
n4 + 1

∞ ∞

∑ ∑n
−n
d. ln n
n2
e. 10
e2
n=2 n=2
∞
f. For what value of α does the series ∑
n=2
1
n (ln n )α
converge?

Comparison Test.

∞ ∞
Theorem 2.7: a. If ∑ bn converges and 0 ≤ a n ≤ bn for all n ≥ 1, then
n =1
∑a
n =1
n converges and
∞ ∞

∑ an ≤
n =1
∑b
n =1
n .
∞ ∞
b. If ∑ an diverges and 0 ≤ a n ≤ bn for all n ≥ 1, then
n =1
∑b
n =1
n diverges.

Proof: (a)
∞ ∞
Let S m be the mth partial sum of ∑ an and S m′ be the mth partial sum of
n =1
∑bn =1
n .
∞
Since ∑b n converges lim S m ′ = L (exists). Moreover since 0 ≤ a n ≤ bn , both { S m }∞m=1 and
m→∞
n =1

{S ′ } m
∞

m =1
are increasing sequences and 0 ≤ S m ≤ S m ′ < L for all m ≥ 1. ( Verify! )

We see that { S m }∞m =1 is an increasing and bounded sequence. Therefore it converges.

∞ ∞
Also ∑ an ≤ lim S m ≤ L = lim S m ′ = ∑b n ( Verify! )
m→∞ m→∞
n =1 n =1

Proof (b):
∞ ∞
If ∑a
n =1
n diverges and 0 ≤ a n ≤ bn for all n ≥ 1. Assume ∑b
n =1
n converges, then by part (a) above
∞

∑a
n =1
n converges, which is in contradiction to our hypothesis. Therefore the assumption was wrong
∞
i.e. ∑b
n =1
n diverges.

19
Compiled by Abraham Negussie. Revised on 2007

∞
Eg1. Show that ∑
n =1
1
2n + 1
converges.

Solution:
1
2n + 1
≤ 1
2n
for all n ≥ 1
and

∑
n =1
1
2n

∞
is a geometric series which converges, and hence by the comparison test ∑
n =1
1
2n + 1
converges.

Exercise: 3.5:
1. Prove the following:

∞ ∞
a. ∑
n =1
1
2 n +1
diverges. b. ∑
n =1
1
2n −1
converges
∞ ∞
c. ∑
n =1
1
n! converges ( Hint: compare it with ∑
n =1
1
2n − 1
)

2. Write out each of the following sums

∞ ∞ ∞
a. ∑
n =1
1
2n
- ∑
k =10
1
2k
b. ∑
n =1
1
22 n

Limit Comparison Test

∞ ∞
Theorem 2.8: Let ∑ an and
n =1
∑b
n =1
n be two series of positive terms.

a. If lim abnn = L > 0 , then the two series either both converge or both diverge.
n→∞
∞ ∞
b. If lim abnn = 0 and
n→∞
∑ bn converges, then
n =1
∑a
n =1
n converges.
∞ ∞
c. If lim abnn = ∞ and
n→∞
∑ bn diverges, then
n =1
∑a
n =1
n diverges.

Proof:

20
Since lim abnn = L and L > 0 for ε = L
2 there exists N such that
n→∞

|abnn − L| < L
2 for n ≥ N

Equivalently
an
L
2 < bn < 3L
2 for n ≥ N .

Now considering abnn < 3L

2 which is the same as a n < 3L
2 bn and the comparison test, the
∞ ∞
convergence of ∑ bn implies the convergence of
n =1
∑a
n =1
n . Similarly applying the comparison test
∞
and considering the inequality L
2 < an
bn equivalently bn < L
2 a n , we conclude that : If ∑b
n =1
n

∞
diverges, then ∑a
n =1
n diverges.

Proof of (b) and (c): Exercise!

∞ ∞
Eg1. Show that ∑
n =1
4n
n − 5n − 7
3 is convergent. Use ∑
n =1
4
n2
for the limit comparison test.
∞ ∞
Eg2. Show that ∑
n =1
3
2
8n2 − 5n
diverges. Use the series ∑
n =1 ( n )
1
2
3 for the limit comparison test.

Ratio Test
∞

∑a
an + 1
Theorem 2.9: Let n be a positive series and let lim an = r ( possibly ∞ ).
n→∞
n =1
∞
a. If 0 ≤ r < 1, then ∑a
n =1
n converges.
∞
b. If r > 1, then ∑a
n =1
n diverges.

If r = 1, then no conclusion can be drawn from this test alone.

Proof: (a)
an + 1
Let c ∈ R such that r < c < 1. Since lim an = r , by the definition the limit given ε = c – r > 0
n→∞

there exists N such that

an + 1
0 < an − r < ε =c−r whenever n ≥ N ,
equivalently

21
 an + 1
0 < an < c − r + r = c whenever n ≥ N (i.e. n = N , n = N + 1 , n = N = 2 , … ) .

That is
a N +1 < ca N
a N + 2 < ca N + 1 < c(ca N ) = c 2 a N

aN + k < c k aN
∞
ca N
But the series ∑c
k =1
k
a N converges to
1− c
since it is a geometric series with c < 1. Thus by
∞ ∞
comparison test the series ∑ a N + k converges. Consequently
k =1
∑a
n =1
n converges.

Proof: (b)
an + 1
From lim an = r > 1 there exists N such that
n→∞
an + 1
an > 1 whenever n ≥ N . ( Verify! )
That is
a N +1 > a N
a N + 2 > a N +1 > a N

aN + k > aN

i.e. a n > a N > 0 for all n ≥ N ,

Which implies that

∞
lim a n ≠ 0 . And hence the corollary of theorem 3.1, the series
n→∞
∑a
n =1
n diverges.

an + 1
When lim an = 1 , the test for convergence and divergence fails. For this consider the p – series
n→∞
1
∞

∑
an + 1 ( n + 1) p
1
np
: lim an = lim 1
= lim( n n+ 1 ) p = 1 . Where as the p – series converges for p > 1
n→∞ n→∞ n→∞
n =1 np

and diverges for p ≤ 1 .

∞
Eg1. ∑
n =1
n
10 n
.

22
 n +1
an + 1 10 n + 1 n +1 1 + 1n
r = lim an = lim n
= lim 10 n = lim 10 = 1
10 < 1.
n→∞ n→∞ n→∞ n→∞
10 n
Hence by the ratio test the series converges.

∞
Eg2: Apply ratio test to the series ∑n =1
nn
3n n !

( n + 1)n + 1

r = lim
n→∞
an + 1
an = lim
n→∞
3n + 1 ( n + 1 )!
n n = lim
n→∞
( n + 1)n
3n n
= lim 13
n→∞
( ) n +1 n
n = e
3 < 1.
3n n !

Therefore the series converges.

nn
Exercise! Find the lim n ? (Hint Apply theorem 3.1)
n → ∞ 3 n!

Exercises 3.6:
1. Use the ratio test.

∞ ∞ ∞ ∞
a. ∑
n =1
2n
n! b. ∑
n =1
2n
n2
c. ∑
n =1
n!
nn
d. ∑
n =1
( 2 n )!
( n!) 2

∞ ∞ ∞
e. ∑
n =1
2 n n!
( 2 n )! f. ∑
n=0
n3
n! g. ∑
n=0
( n + 1)!
n !5 n

∞
e. For what value of α does the series ∑
n =1
n !α n
nn
converge?

2. Consider the Fibonacci sequence 1, 1, 2, 3, 5, 8, 13, 21, … which is defined by

a1 = 1, a 2 =1 and a n + 1 = a n + a n − 1 for n ≥ 2 . Determine whether the series of the reciprocals
Fibonacci numbers:

1+1+ 1
2 + 1
3 + 1
5 + 18 + …
converge or diverge?

Root Test
∞
Theorem 2.10: Let ∑a
n =1
n be a nonnegative series and that lim a n = r ( possibly ∞ ).
n→∞

∞
a. If 0 ≤ r < 1, then ∑a
n =1
n converges.

23
 ∞
b. If r > 1, then ∑a
n =1
n diverges.

If r = 1, then no conclusion can be drawn about the convergence or divergence of the series from
this test alone.

Proof (a): Assume that 0 ≤ r < 1 and let s be any number such that r < s < 1.
Since lim a n = r < s there is an integer N such that for every n ≥ N we have
n→∞

a n < s which is equivalent to a n < s n .

∞ ∞
But ∑s
n =1
n
converges since s < 1 and hence by comparison test ∑a
n =1
n converges.

Proof (b) Exercise!

∞
Eg: Apply root test to the series ∑
n =1
n
10 n
.
n
Solution: r = lim n n
10 n
= lim n
10 = 1
10 < 1 . Therefore by root test the series converges.
n→∞ n→∞

Exercise 3.: Apply Root Test:

∞ ∞ ∞
a. ∑ n( π )
n =1
4
n
b. ∑(
n =1
n! n
nn
) c. ∑
n =1
ln n
en

ALTERNATING SERIES AND ABSOLUTE CONVERGENCE

An alternating series consists of positive terms but has a changing or alternating sign:

∞
a1 - a 2 + a3 - a 4 + a5 - … = ∑ (−1)
a =1
n +1
an ,

with
a n > 0.

Convergence of an alternating series is established under set of conditions stated in the following
theorem.

Alternating Series Test

Theorem 2.11: Let a n ≥ a n + 1 ≥ 0 or a1 ≥ a 2 ≥ a3 ≥ … (1)

24
Compiled by Abraham Negussie. Revised on 2007

and lim a n = 0 . (2)

n→∞
∞
Then the alternating series ∑ (−1)
a =1
n +1
a n converges. Furthermore the sum S and the sequence of

partial sums satisfy the inequality

|S − S k | < a k + 1 . (3)

Proof: Let { S k }∞k = 1 be the sequence of partial sums. We will show { S 2 k }∞k = 1 is bounded and
{
increasing whereas S 2 k + 1 }∞
k =1
is bounded and decreasing.
Now, we consider the difference of consecutive terms of the sequence { S 2 k }∞k = 1 ,

S 2( k +1) - S 2 k = S 2 k + 2 - S 2 k

= ( a1 - a 2 + a3 - a 4 + … - a 2 k + a 2 k +1 - a 2 k + 2 ) - ( a1 - a 2 + a3 - a 4 + … - a 2 k )

= a 2 k +1 - a 2 k + 2 ≥ 0. by (1)

This proves that { S 2 k }∞k = 1 is increasing.

And
≤0 ≤0 ≤0 ≤0

S 2 k = a1 − a 2 + a3 − a 4 + a5 − … − a 2 k − 2 + a 2 k −1 − a 2 k
This implies that
S 2 k ≤ a1 for each k = 1,2, 3, …

Thus { S 2 k }∞k = 1 is bounded.

Therefore { S 2 k }∞k = 1 converges and lim S 2 k
k →∞
{
exists. By a similar argument S 2 k +1 }
∞
k =1
is bounded
and decreasing and hence converges and lim S 2 k +1 exists.
k →∞

Now

S 2 k +1 - S 2 k = ( a1 - a 2 + a3 - a 4 + … - a 2 k + a 2 k +1 ) - ( a1 - a 2 + a3 - a 4 + … - a 2 k )
= a 2 k +1 .

Thus
lim S 2 k +1 - lim S 2 k = lim( S 2 k +! − S 2 k ) = lim a 2 k +1 =0, by (2)
k →∞ k →∞ k →∞ k →∞

25
which implies
lim S 2 k +1 = lim S 2 k .
k →∞ k →∞

Consequently lim S k exists and lim S k = lim S 2 k = lim S 2 k +1 . ( Prove! this statement
k →∞ k →∞ k →∞ k →∞

using the definition of limit)

∞
Therefore ∑ (−1)
a =1
n +1
a n converges.
∞
Next let S be the sum of the series ∑ (−1) n +1
a n . Since { S 2 k }∞k = 1 is increasing we have
a =1

S - S 2 k ≥ 0.
And since

{S } ∞
2 k +1 k = 1 is decreasing with limit S

S 2 k +1 - S ≥ 0,
and
S 2 k +1 - S 2 k ≥ S - S 2 k .

Combining the inequalities we get

0 ≤ S - S 2 k ≤ S 2 k +1 - S 2 k = a 2 k +1 . (5)

Similar computation yields

0 ≤ S 2 k +1 - S ≤ S 2 k +1 - S 2 k + 2 ≤ a 2 k + 2 . (6)

(5) proves the inequality for the case of an even index where as (6) proves the case of an odd index. .

Therefore we have |S − S k | < a k + 1 for any k ≥ 1.

∞
Eg1. ∑
n =1
( −1)n + 1
n . Here a n = 1
n which is decreasing, and lim 1n = 0 . Hence by alternating series
n→∞

∞
theorem the series ∑n =1
( −1)n
n converges.

Exercise 3.7: Apply alternating series theorem:

∞ ∞ ∞
a. ∑ (−1) n
n =1
n+2
n ( n + 1) b. ∑ (−1) n
n =1
2
3n
c. ∑ (−1)
n =1
n + 1 ln n
n .

26
 ∞ ∞
Definition: The infinite series ∑ an is said to be absolutely convergent if the series
n =1
∑| a
n =1
n | is

convergent.

∞
Eg: Consider the series ∑ (−1)
n =1
n 3
4n
= − 34 + 163 − 643 + … + ( −1)n 3
4n
+….

This series is absolutely convergent because

∞ ∞

∑| (−1) n
n =1
3
4n
|= ∑n =1
3
4n
= 3
4 + 163 3
64 +…+ 3
4n
+…

is a convergent geometric series with c = 3 and r = 1

4 < 1.

Definition: A series which is convergent, but not absolutely convergent is said to be

conditionally convergent.
∞ ∞ ∞
Eg: The series ∑
n =1
( −1)n + 1
n is convergent by alternating series test, but ∑
n =1
( −1)n + 1
n = ∑
n =1
1
n is divergent.

Thus it converges conditionally.

Generalized Convergence Test

∞
Theorem 2.12: Let ∑a
n =1
n be a series.
∞ ∞
a. Generalized Comparison Test: If |a n| ≤ |bn| for all n ≥ 1 and ∑|b | converges, then ∑ a
n =1
n
n =1
n

converges absolutely.

∞ ∞
b. Generalized Limit Comparison Test: lim |bnn| = L > 0 and
n→∞
|a |
∑|bn | converges, then
n =1
∑a
n =1
n

converges absolutely.

an
c. Generalized Ratio Test: Suppose that a n ≠ 0 for all n ≥ 1 and that lim an + 1 = r (possibly ∞):
n→∞
∞
If r < 1, then ∑a
n =1
n converges absolutely.
∞
If r > 1, then ∑an =1
n diverges.

If r = 1, then no conclusion can be drawn about the convergence of the series form this test
alone.

27
d. Generalized Root Test: suppose that lim n | a n| = r (possibly ∞):
n→∞
∞
If r < 1, then ∑a
n =1
n converges absolutely.
∞
If r > 1, then ∑a
n =1
n diverges.

If r = 1, then no conclusion can be drawn about the convergence of the series form this test
alone.

∑ (−1)
n n
n +1 2 x
Eg1: For what value of x does the series n 3n
converge?
n =1

Solution:
For the given series,
2n x n 2n + 1 x n + 1
a n = ( −1)n +1 n 3n
and a n + 1 = ( −1) n + 2 ( n + 1)3n + 1
.
So
an 2n + 1 x n + 1 n 3n
lim an + 1 = lim n +1 ⋅ 2n x n = lim 23 | x| n n+ 1 = 2
|x|.
n → ∞ ( n + 1)3
3
n→∞ n→∞

Therefore the series is absolutely convergent when 23 |x| < 1or equivalently | x| < 32 . The series is
divergent when 23 | x| > 1 or equivalently | x| > 32 . When 23 | x| = 1 (i.e. when x = ± 32 ratio test
fails. When x = 32 , the given series becomes

1
1 − 1
2 + 1
3 − 1
4 + 15 − … + ( −1) n + 1 1n + … ,

which converges conditionally.

2n 3n
And when x = - 32 the terms of the series become ( −1) n + 1 n 3n ( −1) n 2n = ( −1) 2 n + 1 n = − n where
1 1

the series is
− 11 − 12 − 13 − 14 − 15 − … − 1n − … ,
which is divergent.
Therefore the series is absolutely convergent when - 32 < x < 32 and is conditionally convergent for
x = 3
2 . If x ≤ - 32 or x > 3
2 the series diverges. ( i.e for | x | > 3
2 or x = - 32 ).

∑
cos 13 nπ
Eg2: Determine if the series n2
is convergent or divergent.
n =1

Solution:
|cos 13 nπ |
n2
≤ 1
n2
for all positive integers n.

28
 ∞ ∞

∑ ∑
1 cos 13 nπ
n2
is a p – series which converges. So by generalized comparison test n2
is
n =1 n =1
∞

∑
cos 13 nπ
convergent. Thus n2
is absolutely convergent.
n =1

∞
Theorem 2.13: If the series ∑a
n =1
n is absolutely convergent, then it is convergent and

∞ ∞

∑ an
n =1
≤ ∑|a
n =1
n |.

Proof:
∞ ∞ ∞
Consider the series ∑ an ,
n =1
∑|an | , and
n =1
∑ (a
n =1
n + | a n |) ; and let their sequences of partial sums

be { S n }∞n =1 , {Tn }∞n=1 , and { Rn }∞n=1 respectively. For any positive integer n an + | an | is either 0 or
2| a n |.
So

0 ≤ an + | an | ≤ 2| a n | (1)
∞
Since ∑|a
n =1
n | is convergent it has a sum say T , and since Tn is increasing

0 ≤ Tn ≤ T
So
0 ≤ Rn ≤ 2 Tn ≤ 2T (2)

{ Rn }∞n=1 is bounded above by 2T and since it is an increasing sequence it converges.

i.e
lim Rn = R and R ≤ 2T .
n→∞

∑a
n =1
n is convergent since it is a difference of two convergent series and let S be its sum, then
∞ ∞ ∞ ∞
S= ∑ an =
n =1
∑
n =1
an + | an | − | an | = ∑ (an + | an |) -
n =1
∑|a
n =1
n | = R – T.

But also
S = R – T ≤ 2T – T = T (3)

∞ ∞
Since ∑ an is a convergent series
n =1
∑ ( −a
n =1
n ) is also a convergent series and
∞ ∞

∑ ( −an ) = -
n =1
∑a
n =1
n = - S.

29
Co m pi le d b y Ab ra ha m N eg us si e. Re vi se d o n 2 00 7

And
∞ ∞

∑|− an | =
n =1
∑|a
n =1
n | = T.

Then by replacing - an by an in the above discussion we get the inequality – S ≤ T (4)

From inequalities (3) and (4) we have

|S|≤ T
That is

∞ ∞

∑ an
n =1
≤ ∑|a
n =1
n |.

Exercises:
1. Determine if the following series is absolutely convergent, conditionally converge or divergent.

∞ ∞ ∞
a. ∑ (−1) n
n =1
2
n!
n +1 b. ∑ (−1) n +1 sin πn
n =1
c. ∑ (−1)
n =1
n nn
n!

∑ (−1) ( )
∞ ∞ ∞

∑ n 2 n + 100 n
∑
n
( −1) ( −1)n
d. n + ( −1) n e. 3n + 1 f. n
n
n=2 n =1 n =1

∞ ∞
2. Prove that if ∑ an is absolutely convergent and an ≠ 0 for all n, then
n =1
∑
n =1
1
| an | is divergent.
∞ ∞
3. Prove that if ∑ an is absolutely convergent, then
n =1
∑a
n =1
n
2
is convergent.

4. Show by an example that the converse of exercise 3 not true.

5. You may try to determine if some of these series are convergent or not.

30
31
Compiled by Abraham Negussie

3. Power series

Definition: A power series in (x – a) is a series of the form

∑ c ( x − a)
n =1
n
n
= c0 + c1 ( x − a ) + c 2 ( x − a ) 2 + + cn ( x − a ) n +

we take (x – a)0 = 1, even when x = a.

In particular when a = 0 the power series becomes a series in x which is

∑c
n =1
n x n = c0 + c1 x + c 2 x 2 + + cn x n +

∞
Eg1. The geometric series ∑ cx
n =1
n
, where c ≠ 0 is an example of a power series that converges for

|x| < 1 and that diverges for |x| ≥ 1.

∞ ∞
Eg2. If r ≠ 0 the power series ∑ cr
n =1
n
xn = ∑ c(rx)
n =1
n
is a geometric series for each value of x and it

converges for |rx|< 1 (or | x| < |1r| ) and diverges for | x| ≥ |1r| .

∞
Eg3. Show that ∑ n! x
n=0
n
= 1 + x + 2 x 2 + 6 x 3 + 24 x 4 +

Converges only for x = 0.

Solution: If x ≠ 0, then
( n +1) x n + 1
lim n! x n
= lim ( n + 1) x = ∞ .
n→∞ n→∞

Hence by Generalized Ratio Test the series diverges. How about when x = 0?

∞
Eg4. Show that ∑
n=0
xn
n! .

Solution: If x ≠ 0, then
xn +1

lim ( n +1) !
xn
= lim 1
n +1 x = 0.
n→∞ n! n→∞

Thus by Generalized Ratio Test the series converges for all x ≠ 0. Therefore the series converges
for any x

32
Lemma 3.1:

∞ ∞
a. If ∑ cn s n converges, then
n =1
∑c
n =1
n x n converges absolutely for | x | < | s |.
∞ ∞
b. If ∑c s
n =1
n
n
diverges, then ∑c
n =1
n x n diverges for | x | > | s |.

∞
Proof: (a) Let | x | < | s |, and rewrite ∑c
n =1
n x n as
∞ ∞

∑c
n =1
nx
n
= ∑c s ( )
n =1
n
n x n
s .
∞
But from hypothesis ∑c s
n =1
n
n
converges and hence by theorem 2.4. lim c n s n = 0 . Consequently
n→∞

there is a positive integer N such that c n s n < 1 for all n ≥ N .

Which implies that

| cn x n | = c n s n ( xs ) x n
n
≤ s for all n ≥ N .
∞
Since | x | < | s | ⇔ x
s < 1, ∑
n=N
x n
s converges absolutely. Then by Generalized Comparison
∞ ∞
Test ∑ cn x n converges absolutely and hence
n=N
∑c
n =1
n x n converges absolutely.

Poof: (b) exercise!

∞
Theorem 3.1: Let ∑c
n =1
n x n be a power series. Then exactly one of the following conditions holds.
∞
a. ∑c
n =1
n x n converges only if x = 0.
∞
b. ∑c
n =1
n x n converges for all x.
∞
c. There is a number R > 0 such that ∑c
n =1
n x n converges for | x | < R and diverges for | x | > R.

Proof: If x is replaced by zero in the given series, we have c 0 + 0 + 0 + … , which is obviously

∞
convergent. Therefore every power series of the form ∑c
n =1
n x n is convergent when x = 0, and if this

is the only value of x that the series converges then condition (a) holds.

33
Suppose that the given series is convergent for x = s where s ≠ 0. Then it follows from Lemma 3.1.
that the series is absolutely convergent for all values of x for which | x | < | s |. Now if in addition
there is no value of x for which the series diverges, then this is condition (b).

If the given series is convergent for x = s where s ≠ 0 and is divergent for x = t where | t | > | s | .
From Lemma 3.1, the series is divergent for all values of x for which | x | > | t | . Hence | t | is an
upper bound for the set of all values for | x | for which the series is absolutely convergent. Therefore
by axiom of completeness this set of numbers has a least upper bound, which is R of condition (c).

The set values of x for which a given power series is convergent is called the interval of convergence
of the power series. The number R of condition (c) of theorem 3.1 is called the radius of
convergence. Moreover from the above the interval of convergence takes one and only one of the
following forms.

[ 0 ,0 ] , [− R , R] , (− R , R) , (− R , R] , [ − R , R ) and ( −∞ , ∞ ) .

∞
Eg1. Determine the interval of convergence of the power series ∑ n ( x − 2) .
n =1

Solution: Applying Generalized Ratio Test we have

an + 1 n + 1( x − 2 )n + 1
lim an
= lim n ( x − 2 )n
= lim | x − 2| n n+1 = |x – 2|.
n→∞ n→∞ n→∞

Thus the given series is absolutely convergent whenever |x – 2| < 1 that is -1 < x -2 < 1
equivalently, 1 < x < 3.

∞
For x = 1, the series is ∑ (−1)
n =1
n
n which diverges, since lim a 2 n = lim 2n = + ∞ and lim a2 n +1 =
n→∞ n→∞ n→∞

lim − ( 2n + 1) = − ∞ it follows that lim ( −1)n n ≠ 0 .

n→∞ n→∞

∞
For x = 3, the series is ∑ n which also diverges.
n =1

Therefore the interval of convergence is (1, 3).

Eg2. Determine the interval of convergence of the power series

∞

∑
n =1
xn
2 + n2
= x
2 + 12
+ x2
2 + 22
+ x3
2 + 32
+ … + xn
2 + n2
+ xn + 1
2 + ( n + 1)2
+ …

Solution: Applying Generalized Ratio Test we have

an + 1 x n +1 2 + n2 2 + n2
2
( n + 1)2
+ ( n n+1 ) 2
lim = lim ⋅ = lim | x| 2 + ( n + 1)2 = lim | x| = | x |.
an n → ∞ 2 + ( n + 1) xn +1
2 2
n→∞ n→∞ n→∞
( n + 1)2

Thus the given series converges absolutely for | x | < 1.

34
Compiled by Abraham Negussie. Revised on 2007

∞
For x = 1 the series is ∑n =1
1
2 + n2
which converges by either using comparison test or limit comparison
∞
test with ∑
n =1
1
n2
.
∞
And for x = -1 the series is ∑
n =1
( −1)n
2 + n2
which is absolutely convergent, as we just have seen it above

for the case of x=1 and hence it is convergent.

Therefore the interval of convergence is [-1, 1].

Exercise3.1: Determine the interval of convergence for each of the following power series.

∞ ∞ ∞
1⋅ 3 ⋅ 5 ( 2 n − 1)
1. ∑
n =1
n! x n
nn
2. ∑
n =1
2⋅4⋅6 ( 2n )
x 2n + 1 3. ∑ (−1)
n =1
n + 1 ( n + 1) x
n!

∞ ∞ ∞
4. ∑ (−1)n +1
n =1
( x − 2 )n
n 5. ∑
n =1
( x + 2 )n
( n + 1) 2 n
6. ∑ (−1)
n =1
n +1 xn −1
( 2 n −1)!

∞
7. ∑
n =1
2n x n
n2

∞
8. Prove that if the radius of convergence of the power series ∑c x
n =1
n
n
is r , then the radius of
∞
convergence of the series ∑c x
n =1
n
2n
is r.

9. If a and b are positive integers, then find the radius of convergence of the power series
∞

∑
n =1
( n + a )!
n !( n + b )! xn .

Taylor’s Polynomial

Theorem 3.2: Let f be a function such that f and its first n derivatives are continuous on the closed
interval [a., b]. Furthermore let f (n +1)(x) exists for all x in the open interval (a, b). Then there is a
number t in the open interval(a, b) such that

f ′( a ) f ′′( a ) f (n)(a) f ( n + 1) ( t )
f (b ) = f ( a ) + 1! (b − a ) + 2! (b − a ) 2 + + n! (b − a ) n + ( n + 1)! (b − a ) n + 1 … (1)

Note: If n = 0 (1) becomes f (b ) = f ( a ) + f ′(t ) (b − a ) which is the mean value theorem..

Proof: Let F and G be two functions defined by

35
 f ′( x ) f ′′( x ) f ( n − 1) ( x ) f (n)( x)
F ( x ) = f (b ) − f ( x ) − 1! (b − x ) − 2! (b − x ) 2 − − ( n −1)! (b − x ) n − 1 − n! (b − x ) n … (2)

and
( b − x )n + 1
G( x ) = ( n + 1)! … (3)

It follows that F(b ) = 0 and G(b ) = 0 . Differentiating in (2) we get

f ′′ ( x ) f ′′′ ( x ) 3 f ′′′ ( x )
F ′( x) = − f ′( x) + f ′( x) − f ′′( x) (b − x) − 2 2! (b − x) − 2! (b − x) 2 + 3! (b − x ) 2
( n −1)
f (n) ( x) n f (n) ( x) f ( n +1) ( x )
− f ( 4) ( x )
3!
(b − x)3 + + ( n − 1)( nf− 1)! ( x ) (b − x) n − 2 − ( n − 1)!
(b − x) n − 1 + n!
(b − x ) n − 1 − n!
(b − x) n

Combining terms we see that the sum of every odd numbered term with the following even
numbered term is zero; so only the last term remains.
( n +1 )
Therefore F ′( x ) = − f n! ( x ) (b − x )n … (4)
Differentiation in (3) we obtain
G ′( x ) = −n1! (b − x ) n … (5)
Checking the hypothesis of Cauchy’s Mean Value Theorem:

i. F and G are continuous on [a, b]

ii. F and G are differentiable on (a, b)
iii. For all x in (a, b) G ′(x ) ≠ 0

So by conclusion of Cauchy’s Mean Value Theorem

F (b ) − F ( a ) F ′(t )
= where t is in (a, b).
F (b ) − G (a ) G ′(t )
But F(b ) = 0 and G(b ) = 0 so
F ′(t )
F (a ) = G(a ) for some t in (a, b). ….. (6)
G ′(t )

Letting x = a in (3), x = t in (4), and x = t in (5), and substituting in (6), we obtain

F (a) = − f ( n + 1) ( t )
n! (
(b − t ) n − ( b −n!t )n ) ( b − a )n + 1
( n + 1)!

f ( n + 1) ( t )
F (a) = ( n + 1)! (b − a ) n + 1 …. (7)

If x = a in (2), we obtain

f ′( a ) f ′′( a ) f ( n − 1)a ) f (n)(a)

F ( a ) = f (b ) − f ( a ) − 1! (b − a ) − 2! (b − a ) 2 − − ( n −1)! (b − a ) n − 1 − n! (b − a ) n …. (8)
Substituting from (7) into (8) we get

36
 f ′( a ) f ′′( a ) f (n)(a) f ( n + 1) ( t )
f (b ) = f ( a ) + 1! (b − a ) + 2! (b − a ) 2 + + n! (b − a ) n + ( n + 1)! (b − a ) n + 1 ,
which is the desired result.

If in (1) b is replaced by x, Taylor’s formula is obtained. It is

f ′( a ) f ′′( a ) f (n)(a) f ( n + 1) ( t )
f ( x) = f (a ) + 1! ( x − a) + 2! ( x − a )2 + + n! ( x − a )n + ( n + 1)! ( x − a ) n +1 (9)

where t is between a and x.

Formula (9) may be written as f ( x ) = Pn ( x ) + Rn ( x ) where

f ′( a ) f ′′( a ) f (n)(a)
Pn ( x ) = f ( a ) + 1! ( x − a) + 2! ( x − a)2 + + n! ( x − a )n
and
f ( n + 1) ( t )
Rn ( x ) = ( n + 1)! ( x − a ) n +1 where t is between a and x.

Pn (x ) is called the nth degree Taylor Polynomial of the function f at the number a, and Rn (x ) is
called the remainder (Lagrange form of the remainder) named after the French Mathematician
Joseph L. Lagrange.

π
Eg1. Find the third degree Taylor polynomial of the cosine function at 4 .
Solution:
cos′′ π4 ) cos′′′( π4 )
P3 ( x) = cos( π4 ) + cos ′( π4 ) ( x − π4 ) + 2! ( x − π4 ) 2 + 3! ( x − π4 )3
= 2
2
− 2
2
( x − π4 ) − 4
2
( x − π4 ) 2 + 2
12 ( x − π4 )3 .

And
cos( 4 ) ( t )
R3 ( x ) = 4! ( x − π4 ) 4 = 1
24 ( x − π4 ) 4 where t is between π
4 and x .

A special case of Taylor’s formula is obtained by taking a = 0 in (9) and we get

f ′( a ) f ′′( a ) f (n)(a) f ( n + 1) ( t )
f ( x ) = f ( 0) + 1! ( x − 0) + 2! ( x − 0) 2 + + n! ( x − 0) n + ( n + 1)! ( x − 0) n + 1 (10)
where t is between 0 and x .

Formula (10) is called Maclaurin’s formula.

Theorem 3.3: (Taylor formula with integral form of the remainder)

If f is a function whose first n + 1 derivative are continuous on a closed interval containing a
and x, then
f ( x ) = Pn ( x ) + Rn ( x ) where Pn (x ) is the nth degree Taylor Polynomial of the function f at a and
Rn (x ) is the remainder given by

37
 x
Rn ( x ) = 1
n! ∫ a
( x − t ) f ( n +1) (t ) dt

Proof: Exercise.

Differentiation of Power Series

∞
A power series ∑c x
n =1
n
n
defines a function whose domain is the radius of convergence of the series.
∞
Consider the geometric series ∑x
n =1
n
which converges to 1
1− x if |x| < 1 . Thus the power series
∞

∑x
n =1
n
defines a function

f : (-1,1) → R by f (x) = 1
1− x if |x| < 1 , that is 1 + x + x2 + x3 + … + xn + … = 1
1− x .

Thus f (- x) = 1
1+ x , that is 1 - x + x2 - x3 + … + (-1)nxn + … = 1
1+ x ;

f ( x2 ) = 1
1 − x2
, that is 1 + x2 + x4 + x6 + … + x2n + … + … = 1
1 − x2
and

f ( -x2 ) = 1
1 + x2
, that is 1 - x2 + x4 - x6 + … + (-1)nx2n + … + … = 1
1 + x2

∞ ∞
Theorem3.4: If the series ∑ cn x n has a radius of convergence R > 0, then the series
n =1
∑ nc x
n =1
n
n −1

has also R as its radius of convergence.

Solution: Let x and s be real numbers such that –R < x < R and |x| < |s| < R . Since |s| < R
∞

∑c s
n =1
n
n
is convergent, which implies that lim cn s n = 0 . Thus for ε = 1 there is an N such that
n→∞

cn s n < 1 for all n ≥ N . Choose M = max { cs 1

1
, c2 s 2 , … , c N −1 s N −1 , 1 , then we have }
cn s n < M for all n

Now ncn x n −1 = ncn s n ( xs ) n −1 1s ≤ nM

| s| ( xs ) n −1
∞
Applying ratio test to the series M
|s| ∑n (
n =1
x n −1
s ) , we see that

lim
n→∞
an + 1
an = lim
n→∞
( )
n +1
n
xn sn
xn + 1 sn + 1
= x
s lim nn+1 =
n →∞
x
s < 1.

38
 ∞
Thus by Generalized Ratio Test ∑ nc x
n =1
n
n −1
converges absolutely on (-R , R). So if R ′ is the radius
∞
of convergence of convergence of ∑ nc x
n =1
n
n −1
R ′ can not be lass than R i.e. R ′ ≥ R (*)

To complete the proof we show that R ′ is not greater then R. i.e we show that R ≥ R ′ .

Assume R ′ > R. Let t be such that R < |t| < R ′ .

∞
Since |t| > R ∑c t
n=0
n
n
is divergent. (1)
∞ ∞ ∞
Since |t| < R ′ ∑ ncn x n −1 converges absolutely and |t| ∑ ncnt n −1 =
n =1 n =1
∑ nc t
n =1
n
n
is also

convergent.
∞
For any natural number n cn t n ≤ n cn t n = ncn t n and hence by comparison test ∑
n =1
cn t n is

convergent. This contradicts (1). Hence our assumption was wrong. Therefore R ′ = R.

∞
Corollary: If the radius of convergence of the power series ∑c x
n=0
n
n
is R then the radius of
∞
convergence of ∑ n(n +1)c x
n=2
n
n−2
is also R.
∞
Proof: Apply the above theorem for ∑ nc x
n =1
n
n
.

If R is the radius of convergence of the power series that defines a function f, then f is differentiable
on the open interval (-R, R). And the derivative of f can be obtained by differentiating the power
series term by term as stated by the following theorem.

∞
Theorem 3.5: Let ∑c x
n=0
n
n
be a power series with radius of convergence R > 0. Then

d ⎛⎜ c x n ⎞⎟ =
∞ ∞ ∞
d (c x
∑
dx ⎜ n = 0 n ⎟ ∑nc x n
n −1
= ∑ dx n
n
) for |x| < R.
⎝ ⎠ n =1 n=0

Proof: Using Taylor’s formula with n = 1.

f ′( a ) f ′′( t )
f ( x) = f (a) + 1! ( x − a) + 2! ( x − a )2

From this formula with f ( x ) = x n it follows that for every natural number n

n ( n −1) n − 2
x n = a n + na n −1 ( x − a ) + 2! t n ( x − a ) 2 where tn is between a and x for every n. … (i)

39
Compiled by Abraham Negussie. Revised on 2007

∞ ∞ ∞ ∞ ∞
f ( x ) − f ( a ) = ∑ c n x n − ∑ c n a n = c0 + ∑ c n x n − c0 − ∑ cn a n = ∑c (x n
n
− an )
n=0 n=0 n =1 n =1 n =1
n
Dividing by (x – a) since x ≠ a and substituting for x by the expression shown in equation (i) we get

f ( x) − f (a)
( )
∞ ∞
= 1 ∑ cn ( x n − a n ) = 1 ∑ cn na n −1 ( x − a ) 12 n( n −1)t n n − 2 ( x − a ) 2 .
x−a x − a n =1 x − a n =1

And so
f ( x) − f (a) ∞ ∞
= ∑ ncn a n −1 + x−a
∑ n(n −1) c t n−2
… (ii)
x−a n =1
2
n =1
n n

∞
Because a is in (-R, R) it follows that ∑ nc a
n =1
n
n −1
is absolutely convergent.

Because bath x and a are in (-R, R) there is k > 0 such that |a| < k < R and | x | < k < R and
∞

∑ n(n − 1) c k
n =1
n
n−2
. Then because n( n − 1) cn t n n − 2 ≤ n( n − 1) cn k n − 2 for each tn , … (iii)

We conclude by comparison test that

∞

∑ n(n −1) c t
n =1
n n
n−2
is absolutely convergent.

It follows from (ii) and applying the inequality in theorem 2.13 that
f ( x) − f (a) ∞ ∞ ∞

∑ ncn a n −1 = ∑ n(n −1) cnt n n − 2 ≤ ∑ n(n −1) c

x−a x−a n−2
− tn … (iv)
x−a n =1
2
n =1
2
n =1
n

From (iii) and (iv)

f ( x) − f (a) ∞ ∞

∑ ncn a n −1 ∑ n(n −1) c

x−a
− ≤ k n − 2 where 0 < k < R … (v)
x−a n =1
2
n =1
n

Because the series in the right of (v) is absolutely convergent the limit of the right side as x
approaches a is zero. Therefore from (v) and the squeezing theorem it follows that

∞
lim
x→∞
f ( x) − f (a)
x− a = ∑nc a
n =1
n
n −1
,

or equivalently
∞
f ′( a ) = ∑nc a
n =1
n
n −1
.

And because a may be any number in the open interval (-R, R) the theorem is proved.

Eg1. Find a power series representation of 1

(1 − x )2
. for |x| < 1.

40
 ∞
Solution: 1
(1 − x ) = ∑x
n=0
n
for |x| < 1.

( ) = ⎛⎜⎜ ∑ x ⎞
∞
d 1 d n
⎟ = d
(1 + x + x 2 + x 3 + + xn + )
dx (1 − x ) dx ⎟ dx
⎝ n=0 ⎠
∞
1
(1 − x )2
= 0 + 1 + 2 x + 3x + 2
+ nx n −1
+ ) = ∑ nx
n =1
n −1

Eg2. Show that

∞
ex = ∑ xn
= 1+ x + + + + xn! +
n
x2 x3
n! 2! 3!
n=0
∞
Solution: ∑
n=0
xn
n! is absolutely convergent for any value of x i.e. the interval of convergence is (-∞,

∞).
∞
If f ( x ) = ∑
n=0
xn
n! then by theorems 3.3. and 3.4 f is differentiable on (-∞, ∞) and

⎛ ∞ xn ⎞ ∞ ∞ ∞
f ′( x ) = d ⎜ ∑ n! ⎟ = ∑ nx n − 1
= ∑ xn −1
= ∑ xn
= f ( x) . … (*)
dx ⎜ n=0 ⎟ n! ( n −1)! n!
⎝ ⎠ n =1 n =1 n=0

Let y = f (x ) from (*) we have an equation dy

dx = f ′( x ) = f ( x ) = y i.e. dy
dx =y ⇔ dy
y = dx .
Integrating both sides
∫ 1
y dy = ∫ dx
⇔ ln y + c1 = x + c2
⇔ ln y = x + c3 where c3 = c2 − c1
+ c3
⇔ e ln y = e x = ce x where c = e c3
Hence
f ( x) = y = c e x … (**)
But since f (0) = 1 + 0 + + + + 0n! + = 1 substituting in (**) we get c = 1.
2 3 n
0 0
2! 3!
∞ ∞
Therefore f ( x ) = ∑
n=0
xn
n! that is e x = ∑
n=0
xn
n! .

∞
Eg3. Find the sum of the series ∑
n=0
1
n! .
∞ ∞
Solution: By the result in example 2, e = e1 = ∑
n=0
1
n! . Hence the sum of ∑n=0
1
n! is e.

41
 x
e −1
Eg4. Find a power series representation for x , and by differentiating the power series term
∞

by term show that ∑

n =1
n
( n + 1)! = 1.
∞
Solution: Since we have seen that e = 1 + x +
x x2
2! + x3
3! + + xn
n! + = ∑
n=0
xn
n! ,

∞
e −1 = x+
x x2
2! + x3
3! + + xn
n! + = ∑
n =1
xn
n! .

∞
Hence for all x ≠ 0 we have that
ex − 1
x = 1+ x
2! + x2
3! + + xn − 1
n! + = ∑n =1
xn − 1
n! .(*)
∞
Notice here that the series ∑
n =1
x n −1
n! is convergent for all x. ≠ 0. Differentiating both sides of (*)

we obtain
dx
x
d e − 1
x
( ) = (1 + d
dx
x
2! + x2
3! +
n −1
+ x n! + ) for all x ≠ 0.

n−2 ∞
xe x − ( e x − 1)
x 2 = 1
2! + 2x
3! + + ( n −1n) !x + = ∑ ( n − 1) x n − 2
n!
n=2

Substituting for all x = 1 into the above equation we have

∑
∞

∑
( n −1) n
1 = 1
2! + 2
3! + + n! + = n −1
n! = ( n + 1)!
n=2 n =1
∞

Therefore ∑
n =1
n
( n + 1)! = 1.

Integration of Power Series

∞
Theorem 3.6: Let ∑c
n=0
n x n be a power series with radius of convergence R > 0 and let. Then for

all x (-R, R)

⎛ ∞ ⎞ ∞ ∞
⎜ ∑ cn x n ⎟ dx = ∑∫ cn x dx = ∑ nc+n 1 x n +1 .
x x
∫
n
⎜ ⎟
⎝ n=0 ⎠
0 0
n =0 n=0
∞
Moreover the radius of convergence of the series ∑
n=0
cn
n +1 x n is R.

42
 ∞ ∞ ∞
Proof: Let f ( x ) = ∑ c n x n and g ( x) =
n=0
∑
n=0
cn
n +1 x n + 1 . Because the terms ∑
n=0
cn
n +1 x n are
∞
derivatives of the terms of ∑
n=0
cn
n +1 x n + 1 f(x) and g(x) gave the same radius of convergence and by

Theorem 3.3

g ′( x) = f(x) for all x (-R, R).

Because f(x) is differentiable on (-R, R) f(x) is continuous on (-R, R) consequently f(x) is continuous
on every closed interval of (-R, R). Thus by Fundamental Theorem of Calculus
x
∫ 0
f (t ) dt = g ( x) − g (0)

or equivalently

⎛ ∞ ⎞ ∞
⎜ ∑ cn x n ⎟ dx = ∑
x
∫ x n +1
cn
⎜ n=0 ⎟ n +1
⎝ ⎠
0
n=0

x
∫ e − t dt .
2
Eg1. Find a power series representation of
0

∞
Solution: Since e = 1 + x +
x x2
2! + x3
3! + + xn
n! + = ∑
n=0
xn
n!

(t 2 )n
e −t = 1 − t 2 + − + + ( −1) n +
2
t4 t6
2! 3! n! for values of t
∞

∑ ( −1)
n
n t2
= n!
n=0

∑∫
x x
x2n +1
x− + − + + ( − 1) n +
n

∫
x3 x5 x7
e − t dt = ( − 1) n
2
t2
n! dt = 3 2! ⋅ 5 3! ⋅ 7 n ! ( 2 n + 1) .
0 0
n=0

Eg2. Find a power series representation of ln(1 + x).

Solution: For |t |< 1 1

1+t = 1 − t + t2 + + (−1) n t n + , then by the above theorem3.4

43
 ∞

∑∫
x x
∫ 1
01+t
dt =
n=0
0
(−1) n t n dt if | x | < 1

Therefore

xn + 1
ln(1 + x ) = x − x2
2 + x3
3 − x4
4 + + ( −1) n n +1 + if | x | < 1

or equivalently
∞
ln(1 + x ) = ∑ ( −1)
n=0
n −1 xn
n if | x | < 1

Eg3. Find a power series representation of tan −1x .

Solution: If | x | < 1

1
1 + x2
= 1 − x2 + x4 − x6 + + ( − 1) n x 2 n + (*)

Applying Theorem 3.6 and integrating both sides of (*) we get

x
∫ x2n +1
x3
1
dt = x − + x5
− + ( − 1) n 2n + 1 +
01+ t
2 3 5

Therefore
∞
tan −1x = ∑ (−1)
n=0
n x2n + 1
2n + 1 if | x | < 1

1
If we substitute x = , then we get
3

( )
1 2 n +1
tan −1 = π
6 = 1 − 1
3i 3 + 1
95 − 271i 7 + + (−1) n 1
3
1
2n + 1 +
3
∞

i.e. ∑ (−1)
n=0
n
3
1
2 n +1
2n + 1
.= π
6 .

44
Compiled by Abraham Negussie. Revised on 2007

Taylor Series

If f is the function defined by

∞
f ( x) = ∑c
n=0
n x n = c 0 + c1 x + c 2 x 2 + c3 x 3 + + cn x n + (1)

whose radius of convergence is R > 0, it follows from successive application of Theorem 3.4 and
3.5 that f has derivatives of all orders on (-R, R). Such a function is said to be infinitely
differentiable on (-R, R).

f (0) = c0 and

from f ′( x) = c1 + 2c 2 x + 3c3 x 2 + 4c 4 x 3 + + nc n x n −1 + , we get that f ′(0) = c1 .

From f ′′( x) = 2c 2 + 2 ⋅ 3c3 x + 3 ⋅ 4c 4 x 2 + + (n − 1) nc n x n − 2 + ,

f ′′ ( 0 )
we get that f ′′(0) = 2c 2 and so c 2 = 2 ;

From f ′′′( x ) = 2 ⋅ 3c 3 + 2 ⋅ 3 ⋅ 4 c 4 x + + ( n − 2 )( n − 1) nc n x n − 3 + ,

we get that f ′′′(0) = 2 ⋅ 3c3 and so c3 =

f ′′′( 0) f ′′′( 0)
2 ⋅3 = 3! .

(n)
In general cn = f
n!
(0)
for every positive integer n.

Thus substituting c n = f ′′n′ (! 0 ) in (1) we get the following series called Taylor series expansion of
f(x) or the Maclaurin series
(n)
f ′′ ( 0 ) f ′′′ ( 0 )
f ( x) = f ( 0 ) + f ′( 0 ) x + 2! x2 + 3! x3 + + f
n!
(0)
xn +

Note: The Taylor series expansion of f(x) about x = a ∈ (− R, R) is

(n)
f ′′ ( a ) f ′′′ ( a )
f ( x ) = f ( a ) + f ′( a )( x − a ) + 2! (x − a) 2 + 3! ( x − a) 3 + + f (a)
n! ( x − a) n +

Eg1. Find the Maclaurin series expansion of e x .

Solution: If f ( x) = e x , the nth derivative of f f ( n ) ( x) = e x so f ( n ) (0) = 1 , hence

f ( x) = e x = 1 + x + x2! + x3! + + xn
+
2 3

n!
Eg2. Find the Taylor (Maclaurin) series expansion of sin x.

45
 ∞
Solution: sin x = ∑
n=0
( −1) n x 2 n + 1
( 2 n + 1)! = x − x3
3! + x5
5! +

Eg3. . Find the Taylor series expansion of cos x.

Solution: cos x = ∑
n=0
( −1)n x 2 n
(2 n )! = 1 − x2
2! + x4
4! + x6
6! +

46
FOURIER SERIES

Definition: A function f (x ) is said to be periodic if there exists λ ≠ 0 (called the period of f (x ) )

such that
f ( x + λ ) = f (x ) , for all values of x in the domain of f

Eg: Since
sin( x + 2π ) = sin x
for all x ∈ R
cos( x + 2π ) = cos x
for all x ∈ R .
The Sine and Cosine functions are periodic with period 2π .

Exercise 1:
a. Show that if f (x ) is periodic with period λ then for any integer n, f ( x + n λ ) = f (x ) ,
for all
values of x.
b. If f (x ) is periodic with period 2π show that
b b + 2π
i) ∫ a
f ( x ) dx = ∫ a + 2π
f ( x ) dx ( Hint: use the substitution x = t - 2π )
π a + 2π
ii) ∫ −π
f ( x ) dx = ∫ a
f ( x ) dx

Trigonometric Series

A series of the form

∞
1a +
2 0 ∑ (a
n =1
n cos nx + bn sin nx ) (*)

is called a trigonometric series and the constants a 0 , a n , bn are called the coefficients of the series.

Exercise 2: Prove the following integrals.

⎧0 , for m ≠ n
π ⎪
a. ∫ cos mx cos nx dx = ⎨π , for m = n > 0
−π
⎪⎩2π , for m = n = 0
π ⎧0 , for m ≠ n and for m = n = 0
b. ∫ sin mx sin nx dx = ⎨
−π
⎩π , for m = n > 0
π
c. ∫ −π
cos mx sin nx dx = 0

47
If the series (*) converges to a function f (x ) for all x ∈ [-π, π ], then in view of the periodicity of
the trigonometric functions, the sum of the series f (x ) is defined for all x and periodic with period
2π and
∞
f (x ) = 1 a 0 +
2 ∑ (a
n =1
n cos nx + bn sin nx ) .

Integrating term by term on the interval [-π, π ] we get

π π ∞ π
∫ −π
f ( x ) dx = ∫ −π
1 a dx +
2 0 ∑∫
n =1
−π
( a n cos nx + bn sin nx ) dx
π
∫ π f ( x) dx = a π
−
0 + 0
π

π∫
From which we get the coefficient a0 = 1 f ( x ) dx .
−π

π
Similarly one determine the coefficients a n = 1
π ∫ −π
f ( x ) cos nx dx and bn =
π
∫
1 f ( x ) sin nx dx .
π −π

Definition: The Fourier Series of generated by f (x ) is

∞
1a +
2 0 ∑ (a
n =1
n cos nx + bn sin nx ) (1)

where the coefficients called the Fourier coefficients are given by

π
a = 1 f ( x ) dx . 0
π ∫ −π
π
an = 1
π ∫ −π
f ( x ) cos nx dx for all n ≥ 1. and
π
bn = 1
π ∫ −π
f ( x ) sin nx dx . for all n ≥ 1.

Dirichlet’s Convergence Theorem.

48
 ⎧0 , − π ≤ x ≤ 0
⎪
Eg1. Determine the Fourier Series of f ( x ) = ⎨1 , 0 < x < π
⎪⎩0 , x = π
We obtain the Fourier coefficients:

π π
a0 = 1
π ∫ −π
f ( x ) dx = 1
π ∫ 0
dx = 1

π π

π∫ π∫
an = 1 f ( x ) cos nx dx = 1 cos x dx = 0 , for all n ≥ 1
−π 0

π π ⎧⎪ 0, if n is even
bn = 1
π ∫ −π
f ( x ) sin nx dx = 1
π ∫ 0
sin x dx = ⎨ 2
⎪⎩π , if n is odd
, for all n ≥ 1.

Thus the Fourier Series for f (x ) is

∞
sin( 2nx )
1 +
2 ∑
n =1
nπ
Eg2.

49
Compiled by Abraham Negussie. Revised on 2007

50
51
52
2

53
 3

54
Compiled by Abraham Negussie. Revised on 2007

Half Range Expansions

Fourier Cosine series

55
 4

56
 4

Fourier Sine series

57
1

58