Compiled by Abraham Negussie and Kassahun Workalemahu.

C
Applied Mathematics II
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 is convergent or divergent.
0

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.

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

 1 1
 ln x dx =
0
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
1 dx = lim  x  p 1  = lim  
b
 b
 
1 dx = lim 1
b     p 1 
  
xp  1  (  p  1) x
b p b 1 p
x 1
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  g( x) dx converges, then  f ( x) dx converges.
a a
 
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 x 1 dx , where   R.
a. For what value(s) of  does the improper integral converge?
b. Compute the integral when  = n  N. i.e when  is a natural number.

 ln x
2. Find the values of  for which the improper integral  1 x
dx converges, and evaluate the

integral for those values of  .

4
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  > 0 there is N such that

nN  an  L   .
In this case we write

lim an  L .
n

 an nm

Definition: We say a sequence converges if lim an  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  > 0 we can choose N > 1 from which we get
n 
nN  1  1
n N
 1  1 <  (since N > 1 )
n N 
 1 < 
n
 1 0  ( 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

 1 for n even
Proof: (-1)n =  .
  1 for n odd

5
We proceed by proof by contradiction. Assume that lim( 1) n exists, say L; then by definition of
n 
limit:
for  =1 there exists an N such that

nN  (1) n  L    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) 
  
Applying Eg3, the sequences 1 n
2 and n
are convergent while n
is
n 1 n 1 n 1
divergent.

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

 c  n  
  
1. n
for c > 0. 2. n
3. ln n
2
n 1 n 1 n n 1

    6.  (1  n2 )n 
n   
4. e
n 5. 8n
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 ,


 can n1 where c is a scalar ,



 an bn n1


6
and
 

where limbn  0
an
bn 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 an  lim an  k .
n 1 n n

Proof: Exercise! (Hint: use  -  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
am , am1 , am2 ,

Bounded Sequences:

Definition: A sequence  an n1 is said to be bounded if there exists a non-negative number M such
that
an  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 an  L , then
n

Given  > 0 there is N such that

nN  an  L   .
which is equivalent to
nN  an  L  an  L   .
which is equivalent to
nN     an  L   .
which is equivalent to
nN  L    an  L   .
which implies
a N  L   , a N 1  L   , a N 2  L   , 

7
Now choose M = max a1 , a2 , , a N 1 , L   .
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 an 1  an for all n  1.
b. decreasing if an 1  an 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 an 1  an 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

   
 
 Sinn n1
3n  1  n3  1
d) 4n  5 e) f) n
n 1 n 1

   
 
n2
g) 2n
h) n
3n  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  > 0 B -  can not be upper bound for the sequence. That is to mean

there is N such that B    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    aN  an  B < B   whenever n  N.

Which implies

B    an < B   if n  N.

That is equivalent to

nN  an  B   .
We then conclude that lim an  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
 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:



a0  0 and an 1  an 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 nk  k , nk  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.

 an n1 converges to L and let  an k 1 be a subsequence of  a n n 1 .

 
Proof: Assume the sequence k

 nk k 1

Since the sequence of positive integers is increasing, n1  1 n2  n1  1 i.e. n2  2 and
 an n1

hence by induction nk  k for each positive integer k. From the fact that converges to L,

10
given   0 there exits a natural number N such an  L   whenever n  N . Thus, whenever

 

nk  k > N, we have a nk  L   . Hence the subsequence ank converges to L.
k 1

Corollary 2.4.a. If the sequence  an n 1 has two subsequences which converges to different limits,


then the sequence  an n 1 diverges.



Corollary 2.4.b. If the sequence  an n1 has a subsequences which diverges then, the sequence
 an 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 an  sin( n4 ) for each natural number n. Let nk  4k  2 for each natural number k. Then
the subsequence a  
nk k 1 = a  
4 k  2 k 1 = sin( 4 k  2 )
4 

k 1
= (1)  n 
n 1
. Therefore by corollary
2.4.b. the sequence sin( n4 )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  a2n n1 converges to 1. ( i.e. lim a2 n  lim(1  21n )  1 ) and
n  n 

subsequence a 2 n 1   
n 1
converges to -1. ( i.e. lim a2 n  1  lim( 1)(1 
n 
1
2n 1 )  1) .
n 

Exercise: Find all limit points.

a.  cos( n2 n1 b. (1  1n ) sin( n2 ) n1
 

11
2.2. Infinite Series

Let  a n n 1 be a sequence. The expression a n or a1 + a 2 + a 3 + … + 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 + a 3
 
S n = a1 + a 2 + a 3 + … + 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 = am
S2 = a m + a m1
S3 = a m + a m1 + a m 2
 
Sn = 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  lim1  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 + 4  14 = 1 + 2( 12 )
1

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 + 4  4 + 8  8  18  18 = 1 + 3( 12 )
1 1 1 1

and

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

1 1 1
2
  2
  2

 1+ 2 + 4  4 + 8  8  8  8 + … + ( 2m  2m    21m
1 1 1 1 1 1 1 1 1

= 1 + m( 12 ) .

But

14
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 an  0 .
n 

Proof: For n  1 a n = ( a1 + a 2 + a 3 + … + a n 1 + a n ) – ( a1 + a 2 + a 3 + … + 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


 cr
m
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 c a
n 1
n converges and

16
  

c a
n 1
n = c  an .
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 + a 3 + … + a n 1 for n  2. (2)

17
   
From (1) a
n2
n converges if the integral  f ( x) dx converges and hence  a
1
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
 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  ) 1
n b. 
n2
1
n2  1
c. 
n 1
n3
n4  1

 

 n
n
ln n
d. n2
e. 10
e2
n2 n2

f. For what value of  does the series 
n2
1
n (ln n )
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
b
n 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
 
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:
Since lim abnn  L and L > 0 for   L
2 there exists N such that
n

20
 |abnn  L|  L
2 for n  N

Equivalently

L
2  an
bn  3L
2 for n  N .

Now considering abnn  3L

2 which is the same as an  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 an , 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)
2
1
3 for the limit comparison test.

Ratio Test


a  r ( possibly  ).
an  1
Theorem 2.9: Let n be a positive series and let lim an
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  = c – r > 0
n

there exists N such that

0    cr
an  1
an  r whenever n  N ,
equivalently

21
  c  r  r  c whenever n  N (i.e. n  N , n  N  1 , n  N  2,  ) .
an  1
0  an

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. an  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.

 1 , the test for convergence and divergence fails. For this consider the p – series
an  1
When lim an
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
10n
.

22
 n 1
10n  1 n 1 1  1n
r  lim
an  1
an  lim n
 lim 10n  lim 10  1
10  1.
n n n  n 
10n
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 )!
nn  lim
n
( n  1)n
3n n
 lim 13
n
  n 1 n
n  e
3  1.
3n n!

Therefore the series converges.

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

n

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  does the series 
n 1
n! n
nn
converge?

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

a1 1, a2 1 and an 1  an  an 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 an  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
10n
.

Solution: r  lim n  lim 10n   1 . Therefore by root test the series converges.
n
n 1
n  10n n 
10

Exercise 3.: Apply Root Test:

  
a.  n(  )
n 1
4
n
b. (
n 1
nn
)
n! n
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 + a 3 - a 4 + a 5 - … =  (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  a 3  … (1)

24
and lim an  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 |  ak 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 + a 3 - a 4 + … - a 2 k + a 2 k 1 - a 2 k  2 ) - ( a1 - a 2 + a 3 - 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 2k = a1  a2  a3  a4  a5    a 2 k 2  a2 k 1  a2 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 + a 3 - a 4 + … - a 2 k + a 2 k 1 ) - ( a1 - a 2 + a 3 - 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  a2 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 |  ak 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  a n 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 |an|  |bn| for all n  1 and |b | converges, then  a
n 1
n
n 1
n

converges absolutely.

 
b. Generalized Limit Comparison Test: lim
n
|an|
|bn|  L  0 and |b | converges, then  a
n 1
n
n 1
n

converges absolutely.

c. Generalized Ratio Test: Suppose that an  0 for all n  1 and that lim  r (possibly ):
an
an  1
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 |an|  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
2n  1 x n  1 n 3n
lim  lim ( n  1)3n  1  2n x n = lim 23 | x| n n 1 = |x|.
an 2
an  1 3
n 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 n3n (1) n 2n  (1) 2n 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 = 32 . If x  - 32 or x > 32 the series diverges. ( i.e for | x | > 32 or x = - 32 ).


cos 13 n
Eg2: Determine if the series n2
is convergent or divergent.
n 1

Solution:
|cos 13 n |
n2
 1
n2
for all positive integers n.

28
  

 
cos 13 n
1
n2
is a p – series which converges. So by generalized comparison test n2
is
n 1 n 1



cos 13 n
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  a n is a convergent series
n 1
 ( a
n 1
n ) is also a convergent series and
 

 ( a n ) = -
n 1
a
n 1
n = - S.

29
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 n
n 1
c.  (1)
n 1
n nn
n!

 (1)  
  

 ( 1) n 2 n  100 n
 ( 1)n
n
d. n  ( 1) n e. 3n  1 f. n
n
n2 n 1 n 1

 
2. Prove that if  a n is absolutely convergent and an  0 for all n, then
n 1

n 1
1
| an | is divergent.
 
3. Prove that if  a n 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
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)  c2 ( 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  c2 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 cn s n  0 . Consequently
n 

there is a positive integer N such that cn s n  1 for all n  N .

Which implies that

| cn x n | = cn s n  xs   x n
for all n  N .
n
s

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

n  1( x  2 )n  1
= lim|x  2| n n1 = |x – 2|.
an  1
lim an
= lim n ( x  2 )n
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 a2n  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 |.
n   2  ( n 1) 1
2
n an xn n  n  2
( n  1)2

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

34
 
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 1) ( x )
 f ( 4) ( x )
3! (b  x)3   ( n 1)( nf1)! ( x ) (b  x)n  2  f ( n) ( x)
( n  1)! (b  x)n 1  n f ( n) ( x)
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)  1
n! (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.


Eg1. Find the third degree Taylor polynomial of the cosine function at 4
.
Solution:
cos 4 ) cos( 4 )
P3 ( x)  cos( 4 )  cos( 4 ) ( x  4 )  2! ( x  4 ) 2  3! ( x  4 )3
 2
2
 2
2
( x  4 )  4
2
( x  4 )2  2
12 ( x  4 )3 .

And
cos( 4 ) ( t )
R3 ( x)  4! ( x  4 )4  1
24 ( x  4 )4 where t is between 
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  = 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 1s 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 s n  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  
  


dx  n  0 n  nc x n
n 1
  dx
d (c x
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
     
f ( x )  f ( a )   c n x   c n a  c0   c n x  c 0 
n n n
c a n
n
 c (x n
n
 an )
n0 n0 n 1 n 1 n 1

Dividing by (x – a) since x  a and substituting for x by the expression shown in equation (i) we get
n

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) cnt 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) cntn 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 f ( xx)  af ( a ) 
x 
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    x n   )
dx (1  x ) dx  dx
 n0 

1
(1  x )2
 0  1  2 x  3x    nx 2 n 1
)   nx
n 1
n 1

Eg2. Show that


ex   xn
 1 x  x2
 x3
   xn!  
n

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!    nxn  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
 e ln y  e x  c3  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
 e x 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
d e  1
dx
x

x
   1  d
dx
x
2!  x2
3!    x n!  
n 1
 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)
1    
1 2 n 1 n
2! 3! n! = 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0
n x n
and g ( x)  
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 1
 n0 
0
n0

x
 e t dt .
2
Eg1. Find a power series representation of
0


Solution: Since e  1  x 
x2
 x3
   xn!    
n
x xn
2! 3! n!
n0

e t  1  t 2      (1) n (t 2 ) n

2
t4 t6
2! 3! n! for values of t


 (1)
n
n t2
= n!
n0

 x      (1) n x2 n  1
 .
x x

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  t 2    (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 x n
n if | x |  1

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

Solution: If | x |  1

1
1  x2
 1  x 2  x 4  x 6    (1) n x 2 n   (*)

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

x
 x2 n  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  
6 1  1
33  1
95  271 7   (1)n 1
3
1
2n  1 
3


i.e.  (1)
n0
n
3
1
2 n 1
2n  1
. 
6 .

44
Taylor Series

If f is the function defined by


f ( x)  c
n0
n x n  c0  c1 x  c2 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  2c2 x  3c3 x 2  4c4 x 3    ncn x n 1   , we get that f (0)  c1 .

From f ( x)  2c2  2  3c3 x  3 4c4 x 2    (n 1) ncn x n  2   ,

f  ( 0 )
we get that f (0)  2c2 and so c2  2
;

From f ( x)  2  3c3  2  3  4c 4 x    (n  2)( n  1) ncn 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 cn  f n(!0) in (1) we get the following series called Taylor series expansion of
f(x) or the Maclaurin series

f  ( 0 ) f  ( 0 )
f ( x)  f (0)  f (0) x  2! x2  3! x3    f ( n ) ( 0)
n! xn  

Note: The Taylor series expansion of f(x) about x  a  ( R, R) is

f  ( a ) f  ( a )
f ( x)  f (a)  f (a)( x  a)  2! ( x  a) 2  3! ( x  a) 3   f ( n) (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!    
2 3
xn
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
 Applied Mathematics Lecture Note

FOURIER SERIES

Definition: A function f (x) is said to be periodic if there exists   0 (called the period of f (x) )
such that
f ( x   ) = f (x) , for all values of x in the domain of f

Eg: Since
sin( x  2 )  sin x for all x  R
cos( x  2 )  cos x for all x  R .
The Sine and Cosine functions are periodic with period 2 .

Exercise 1:
a. Show that if f (x) is periodic with period  then for any integer n, f ( x  n  ) = f (x) ,
for all
values of x.
b. If f (x) is periodic with period 2 show that
b b  2
i)  a
f ( x) dx   a  2
f ( x) dx ( Hint: use the substitution x = t - 2 )
 a  2
ii)  
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 a0 , an , bn are called the coefficients of the series.

Exercise 2: Prove the following integrals.

0 , for m  n
 
a.  cos mx cos nx dx  π , for m  n  0


2π , for m  n  0
 0 , for m  n and for m  n  0
b.  sin mx sin nx dx  

π , for m  n  0

c.   cos mxsin nx dx

0

47
If the series (*) converges to a function f (x) for all x  [-,  ], 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
2 and

f (x) = 1 a0 
2  (a
n 1
n cos nx  bn sin nx ) .

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

   


f ( x) dx    2 a0 dx 
1
   (a
n 1

n cos nx  bn sin nx) dx

  f ( x) dx = a 

0 + 0

From which we get the coefficient a0 = 1  f ( x ) dx .
 


Similarly one determine the coefficients a n = 1  f ( x) cos nx dx and bn =
 

  f ( x) sin nx dx .
1
 

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


a = 1 f ( x ) dx .
0
  

an = 1  f ( x) cos nx dx for all n  1. and
 

bn = 1  f ( x) sin nx dx . for all n  1.
 

Dirichlet’s Convergence Theorem.

48
 0 ,    x  0

Eg1. Determine the Fourier Series of f ( x )  1 , 0  x  

0 , x  
We obtain the Fourier coefficients:

 
a0 = 1  f ( x ) dx = 1  dx = 1
   0

 
a n = 1  f ( x) cos nx dx = 1  cos x dx = 0 , for all n  1
   0

   0, if n is even
bn = 1  f ( x) sin nx dx = 1  sin x dx =  2 , for all n  1.
   0  , if n is odd

Thus the Fourier Series for f (x) is


sin(2nx )
1 
2 
n 1
n
Eg2.

49
50
51
52
2

53
 3

54
 3

Half Range Expansions

Fourier Cosine series

55
 4

56
 4

Fourier Sine series

57
1

58