Power Series

UNIT THREE POWER SERIES
Applied Mathematics II
School of Humanities And
Natural Science
Adama University
Prepared by eTeaching Team based
on the script provided by
Written by:Ato Legesse Lemecha
Edited by Ato Tamirat Temesgen
ADU DEPARTMENT OF MATHEMATICS

107
Math 132: APPLIED MATHEMATICS II
3. POWER SERIES
3.0 Introduction
In unit two we have discussed about convergence and divergence of series and how to
determine convergence and divergence. In this unit we will introduce a, power series
representation of a function, differentiation and integration of power series. In the
discussion of power series convergence is still a major question that we will be dealing
with. Further more two basic questions will be raised here:
i) Given the series, to find properties of the sum function.
ii) Given a function f , to find whether or not it can be represented by a
power series.
Even though the answer to the first question is affirmative the answer to second
question is not always possible which will be answered by Taylor series.
Objectives:
At the end of this unit the students will be able to:
 define power series
 define Taylor series
 define Interval of convergence
 define Radius of convergence
 differentiate and integrate power series
 determine the convergence or divergence of a power series
 give a power series representation for a given function
 approximate a function by Taylor series
.
108
Definition 3.1 An infinite series of the form


n 2 n
 a  x  c
n  ao  a1  x  c   a2  x  c   ...  an  x  c   ... , is called a power series
n 0
n
centered at x = c where an and c are constants. The term an  x  c  is the nth term and
the number c is the center.

An expression of the form

f ( x)   an x n  ao  a1 x  a2 x 2  ...
n 0
is a power series centered at x=0.

Is this definition meant that a power series is an infinite series?
If x is a fixed number, then it reduces to series discussed in unit two. The generalized
tests which have been studied in unit two can be extended to the case of power series.

n
Given a power series  an  x  c  , when does it converge and when does it diverge?
n 0

n!
In example 2:36 of section 2.7 we have seen that the infinite series n n
x n converges
n 1
for all x  (e, e) and diverges on (, e]  [e, ) . Then the interval (e, e) is called the

n!
interval of convergence for n n
x n and the distance r = e is the radius of convergence for
n 1
this series. Now we have the following definition.

Definition 3.2 (Interval of convergences)
The interval on which the power series converge is called the interval of convergence or
the domain of a power series.
The interval of convergence consists of all real numbers from c  R to c  R for some
number R where 0  R   .
The power series diverges for all x outside the interval of convergence. The number C is
called the center and R is the radius of convergence.

109
Remarks: i) When R   , the interval of converges is the set of all real

number.
ii) When R = 0, the interval of convergence is the set containing a single
point C 
iii) The interval of convergence takes one and only one of the following forms
[-R, R], [-R, R), (-R, R] or (-R, R).
Example 3.1 Determine whether the following power series are convergent or divergent.
If it converges find the interval of convergence.
 
xn 
 1
n 1
2n x n  x  2k
a)  n!
b) 
n 1 3n n
c)  3 k k2
n 0 k 1

n

2n n
d)  n ! 2 x  1
n 0
e)
n 1 n
 4 x  8
xn
Solution: a) if x  0, then  n!  0 the series converges trivially.
xn x n 1
For x  0 take an  and an 1  , then use the Generalized Ratio Test.
n!  n  1!
a n1 x n1 n! x
 . n 
an n  1! x n  1
an1 x x
  Taking absolute value
an n 1 n 1
an 1 x
L = lim  lim  0 1
n  an n  n  1

xn
Hence by the Generalized Ratio Test 
n 0 n!
converges absolutely for all x 
The interval of convergence is equal to  ,   and radius of converges R  .
n1 n n
  1 2 x
b) 
n1 3n n
If x = 0, the given series is automatically convergent. For x  0 , take
110
n n 1
 1 2n x n  1 2n 1 x n 1
an  and an 1 
3n n 3n 1  n  1
n 1
n 1 n 1
an1  1 2 x 3n n  1 2  x  n   2nx
 n 1
. n n n

an 3  n  1  1 2 x 3  n  1 3  n  1
a n1  2nx 2n x
 
an 3n  1 3n  1
a n1 nx 2
L = lim  lim  x
n  a n  n  1 3
n
So, the Generalized Ratio Test tell as that if L < 1, the series will converges absolutely, if
L > 1 the series will diverge and if L = 1, we don’t know what will happen. So we have,
2 3
x 1  x series converges.
3 2
2 3
x 1  x  series diverges
3 2
3
The radius of convergence for this power series is R  .
2
To determine the interval of convergence we need to analyses what will happen when
3 3
x or x  .
2 2
n
n1 n  3 
  
1 2   n1
3  2   1
If x = , the series becomes   
2 n1 3n n n1 n
This is alternating harmonic series and it converges by alternating series test.
3
If x  , the series here is
2
n
n 1  3
n
 1 2   2 n 1
 2    1  1    1
  

n 1 3n n

n 1 n
 
n 1 n

n 1 n
This is the harmonic series and we know that it diverges. The interval of convergence for
this power series is then

111
3 3
 x
2 2
  3 3
Therefore interval of convergence =  ,
 2 2 
k

 x  2
c) 
k 1 3k k 2
The center is 2 and the power series converges if x = 2. For x  2 ,
k k 1
ak 
 x  2 , a k 1
 x  2
 k 1 2
3k k 2 3  k  1
k 1
a k 1

 x  2  . 3k k 2   x  2  k 2
2 k 2
ak 3 k 1  k  1   x  2  3  k  1
2
ak 1  x  2 k 2 x  2  k  | x  2 |
L  lim  lim 2
 lim   
k  ak k 
3  k  1 3 k   k  1  3
The Generalized Ratio Test tells us that the power series converges if L  1 and so
x2
1  x2 3 Series converges.
3
So, the radius of convergence for this power series is R  3 . We solve this inequality to
determine interval of convergence 1  x  5
If x  1 , the series is
k k k

 1  2  
 3  
 1

k 1 3k k 2

k 1 3k k 2

k 1 k2
and it converges by alternating series test.
If x  5, the series becomes
k k

5  2 
 3 
1

k 1 3 k k 2

k 1
k
3 k 2

k 1 k
2
Again it is convergent by p-series with p=2 > 1. Thus interval of convergence = [-1, 5].

n
d)  n ! 2 x  1
n 0
112

1 n
If x   , the power series  n ! 2 x  1 trivially converges.
2 n 0
1
For x   , apply Generalized Ratio Test as follows.
2
n n 1
an  n ! 2 x  1 and an 1  (n  1)! 2 x  1
n
an1 (n  1)n ! 2 x  1  2 x  1
 n
  2 x  1 n  1
an n ! 2 x  1
an 1
L  lim  lim  2 x  1 n  1  2 x  1 lim  n  1    1
n  an n  n 
1 1
Thus this power series will converge only if x   and diverges for x   . We find
2 2
 1  1 1 
that the interval of convergence =   or  ,  and Radius of convergence R  0 .
 2 2 2

2n n
e)  4 x  8
n 1 n
If x  2 , the given power series converges. If x  2 apply Generalized Ratio Test.

2n n 2n 1 n 1
Here an   4 x  8  , an 1   4 x  8 and
n n 1
an1 2n1 n 1 n 2n
  4 x  8  n n
  4 x  8 
an n 1 2  4 x  8 n 1
an 1 2n 2n
L  lim  lim  4 x  8    4 x  8 lim
n  an n  n 1 n  n  1
 2 4x  8
Now we can get the information about convergence or divergence from this expression.
L  2 4 x  8  1 , the given power series converges.
L  2 4 x  8  1 , the given power series diverges.
Recall that the interval of convergence requires x  c  R . In other words we must factor
out 4 from absolute value in order to get radius of convergence. Consequently we have

113
1
8 x  2  1 implies x  2  the power series converges
8
1
8 x  2  1 implies x  2  the power series diverges.
8
1
This means that the radius of convergence is R  . Lastly we compute the interval of
8
convergence from the inequality that gives convergence.
1
x2 
8
1 1
 x2
8 8
15 17
x
8 8
 15 17 
We know the series converges in the interval  ,  , but we must now test for
8 8
15
convergence at the end points of this interval. When x  , the series becomes
8
n n n

2n  15  
2 n  1  
( 1)
 
k 0 n  2
 8     
 
 k 0 n  2  k 0 n
This is the alternating harmonic series and it is convergent by alternating series test.
17
When x  , the series reduced to
8
n n

2n  17  
2n  1  
1
  8      
k 0 n  2  k 0 n  2  k 0 n
This is harmonic series and so divergent. Therefore the power series converges for
15 17
x and diverges when x  . So the interval of convergence for the given power
8 8
15 17 15 17 
series is x or interval of convergence   ,  .
8 8 8 8 
N.B check always the end points to determine interval of convergence
114
Example 3.2 Find the radius of convergence for the following series
2

12.32.52...  2n  1 x 2 n
a) 
n 1 22.42.62...  2n 
2

n 1 1.3.5...  2n  3
b)   1 xn
n 1 2n n !
Solution: a) The power series converges if x  0 . For x  0 we use the Generalized
2
12.3 2.5 2...2n  1 x 2 n
2
12.32.52...  2n  1 x 2 n 2
Ratio Test with a n  2
, an 1  2
and
2 2.4 2.6 2...2n  22.4 2.62...  2n  2 
2
a n1  2n  1  2
  x
an  2n  2 
2
a  2n  1  2
L  lim n 1  lim   x  x2
n  a n  2 n  2
n  
Here the power series converges if x 2  1 and it diverges if x 2  1 . So we have
2
x 1
x 1
The radius of convergence is R  1 . Next we analyze for end points.

2

12.32.52...  2n  1
If x  1 or x  1 , the power series becomes  2
. Show this series
n 1 22.42.6 2...  2n 
converges and the interval of convergence is [1,1] .


n 1 1.3.5...  2n  3
b)   1 xn
n 1 2n n !
Using Generalized Ratio Test we see that
n 2
an1  1 1.3.5...  2n  1 x n  x 2n n !
 n 1
an 2 n1  n  1 !  1 1.3.5...  2n  3  x n

 2n  1 x
2  n  1
an 1 2n  1
L  lim  lim x
n  an n 2 n  2

115
an 1
L  lim  x 1
n  an
The power series converges absolutely if L  1 and diverges if L > 1 . Hence the radius of
convergence is R  1 .
can you determine the interval of convergence for this series?
Example 3.3 Find the interval and radius of convergence for the power series

2n n ! n
n 
 1  ex 
n 1 n
Solution: For this power series, we have

n n
an1 2  2  n !1  ex 1  ex  nn 2 1  ex  nn
 n n
 n
an  n  1 2n n !1  ex   n  1
n
an 1 2 1  ex  n n  1 
L  lim  lim n
 lim  1   ex  1
n  a
n
n 
 n  1 n
 n 1
an1 1
L  lim  2 x
n  an e

2n n ! n
n 
The Generalized Ratio Test implies that the power series  1  ex  converges if
n 1 n
an 1 1 1 1
L  lim  2 x   1 . This implies that x   . In this case the radius of
n  an e e 2
1
convergence is R  . Now let us calculate the interval of convergence. We will solve
2
the inequality that gives the convergence above.
1 1
x 
e 2
2e e2
x
2e 2e
2e 2e
Verify that it diverges at x  and x  . Consequently the interval of
2e 2e
 2e e2 
convergence is  , .
 2e 2e 
116
Example 3.4 Find the interval and radius of convergence for the power series
x x 2 x3 x 4 x5 x6
1      
3 42 33 44 35 46
Solution: Since
 1 2k
 42 k x n  2k
an  
 1 x 2k 1 n  2k  1
 32 k 1
We apply the Generalized Root Test to determine the radius of convergence and interval
of convergence. In this case
1
 x n  2k
n an   4
1 x n  2k  1
 3
and
1
 4 x n  2k
lim n an  
1 x
n 
n  2k  1
 3
The series converges absolutely if lim n an  1 and the radius of convergence is the least
n 
upper bound of lim n an  1 . It follows that

n 
1
x 1
3
x 3
3  x  3
Verify that the series diverges when x  3 and x  3 .
Thus the interval of convergence is  3,3 and radius of converges is R  3 .

117

Lemma 3.1 Suppose a n x n is a power aeries.
n 0
i) If the series converges at x  c , then the series converges absolutely
at x whenever x  c .
ii) If the series diverges at x  b , then the series diverges at
x whenever x  b

Proof: i) suppose a c n
n
converges. By the nth term test the sequence lim an c n  0 . This
n 
n 0
implies that the sequence of nth term converges. But every convergent sequence is
bounded, so there exists a number M such that anc n  M for all integers n  0 . Let x be

x
arbitrarily chosen with x  c . Then  1 and we need to show that a n x n converges
c n 0
absolutely.
n n
xn x x
an x n  an c n n
 an c n M
c c c
and
  n
n x

n 0
an x   M
n 0 c
 n
x
However, the series  M is a convergent geometric series with first term M and
n 0 c

x n
common ratio
c
 1 . So the Comparison Test implies that a x
n 0
n converges and that
the series is absolutely convergent when x  c .
ii) Suppose the series diverges at b but converges for some x with x  b . This implies,

n
by part (i), that a b
n 0
n converges absolutely, which contradicts the assumption of
divergence at b .
Thus, if the series diverges at b , and x  b the series diverges at all x. 
118
This Lemma implies that if a power series converges for some real numbers but diverges
for some others, then a number R exists with the property that the series converges
absolutely if x  R and diverges if x > R .

Theorem 3.1 If a n x n is a power series, then exactly one of the following statements
n 0
holds:

i) a n x n converges only at x  0
n 0

ii) a n x n converges for all real numbers x
n 0
iii) Let R  0 be the radius of convergence for the power series such that

a) a n x n converges absolutely if x  R ;
n 0

b) a n x n diverges if x > R
n 0
Proof: see reference book.

Example 3.5 Find the domain of the function defined by
2n
f ( x)  
 x  7  ln n

n 1  n  1 arctan n
Solution: Since
2n 2
 x  7   x  7  ln(n  1)  n  1 arctan n
L  lim
n   n  2  arctan(n  1)  x  7 2 n ln n
ln(n  1)  n  1 arctan n 2
 lim  x  (7) 
n  ln n  n  2  arctan(n  1)
2
=  x  (7)  applying L’Hopital’s Rule and simplifying
2
The series converges if  x  (7)   1 , that is, if  x  (7)   1
The inequality  x  (7)   1 is equivalent to 8  x  6 .

The series converges if 8  x  6 and diverges if x  8 or x  6 .

119
At x  8 , the series

ln n
  n  1 arctan n
n 1
diverges since the terms of the series do not approach zero for large value of n. Similarly
at x  6 the series diverges. Accordingly, the domain of f is  8, 6  .
Exercise 3.1
1. Find the radius and interval of convergence for the following power series.
  xn  n ! 2n  ! 2n  2
a)  nx n b)  n 2n c)  x d)  nn x n
n 1 n 1  n !2 n1  3n  ! n 1
xn
3n
 xn
 
 ln n  x  1
e)  n 2n . f)  n n
g) 
n 1  n !2 n 1 n! n 1 n  n  1
2n n n 1

 3x  4  
 3x  
 1 x2 n
h) 
n 0 2n
i)  arctan n
n 1
j) 
n 0  2n !
2. Find the domain of the function
n
(1) n x 2 n
 
 3 x  1
a) f ( x)   b) g ( x )  
n 1 n n  1 n 1 ln(n  1)
2n
 3n 1  3x  
n
c) f ( x )   d) f ( x)    ln x 
n 1 n n 0
3. Find the values of x for which the following series converges.

n

 cox  
 4 n
a)  
n 1  n 
 b)  n4  3x 
n 1
2n

2n 1 
3n
c)  .
n 3x  4n
d) 
n 0  x  1
n
n 1
4. Find the radius and interval of convergence of the following power series.
120
2n
 n1 1.3.5...  2n  3 x  2  5  8   3n  4 
a)   1 b)  3  6  9  3n  3 x n 1
n2 3n n n 2
 
5. Suppose the power series  an x n and
n 0
b x
n 0
n
n
have radius of convergence R1 and R2 ,

respectively. What can be said about the radius of convergence R of  an x n  bn x n ?
n 0
 

nn x n
6. Find the interval of convergence of the power series 
n 0 n !
.

121
3.1 POWER SERIES REPRESENTATION OF FUNCTIONS

In this subsection we will be represent many functions as power series that are related to
geometric series. That is we calculate a power series expression for a function by
manipulating a known series.
Recall that the geometric series

n a
 ar
n 0

1 r
, is convergent if r  1 and diverges if r  1 .
In this series if we take a  1 and r  x , then the series be comes


n 1
x
n 0

1 x
for x  1.
Example 3.6 Consider a geometric series with r  x and a  1

 
1
  x n   x n 1 for x 1 (1)
1  x n 0 n 1

1
f  x    x n  1  x  x 2  ...  x n  ... for | x | 1
1  x n 0
If in (1) we replace x by (-x) we get a new function with new power series representation
 
1 n n n
f (  x)     x     1 x n  1  x  x 2  x 3  x 4 ...   1 x n
1  x n 0 n 0
Again, if we replace x by ( x 2 ) in (1) we get


1
f  x  2
  x 2 n  1  x 2  x 4  ...  x 2n  ... if | x | 1.
1 x n 0
If in (1) we replace x by (-x2) we have


1 n n
f  x  2
   1 x 2n  1  x 2  x4  x6  ...   1 x 2n   if | x | 1
x  1 n 0
From this example one can observe that as we vary x by different expression in (1) we
get different power series representation.
Example 3.7 Find a power series representation for the following function and determine
its domain.
2 x2
f ( x) 
5  x3
122
2 x2 2 1 2 1 2x2 1
Solution: f ( x)  3
 2x 3
 2x 3
 3
5 x 5 x   x   5  x 
5 1    3   1   3 
  5    5


n 1
Now substituting in the geometric series x
n 0

1 x
for x  1 we find that
3n
2x2 
 x  x3
f ( x) 
5
 
 3 
n 0  5
provided that 
5
1
3n 3n 2

 1  x 
f ( x )  2
n 0 5n1
The series converges when
x3
 1
5
x3  5
x  35.
So the interval of absolute convergence is  3 5  x  3 5 and hence the domain of f is
3 5  x  3 5 .
Example 3.8 Find a power series representation for the following function and determine
its domain.
x
f ( x)  3
2 x  16
x x x 1
Solution: We note that f ( x)   
3
2 x  16  x  3
16 x3
16 1   1
 8 8
x3 1 
Now substitute for x in   xn for x  1
8 1  x n 0
 3 n
1 x x3
x 3
 
n 0 8
n
for
8
1
1
8

123
x
Next multiply by 
16
x  x  x3n 
x 3n 1 x3
f ( x)     
2 x 3  16 16 n 0 8n n 0 2
3n  4
for
8
 1.
Thus the required power series representation of
x 
x 3n 1 x3
2 x 3  16
  
n 0 2
3n  4
for
8
1
x3
From  1 , we have
8
x3  8
2  x  2
So the interval of convergence is  2, 2  and radius of convergence R  2 .
3.1.1 DIFFERENTIATION OF POWER SERIES

What do we meant by derivative of a power series? Is the derivative of power series
exists?
Power series are generalization of polynomials
m
p( x)  a0  a1 x  a2 x 2    am x m   an x n
n 0
A power series with non zero radius of convergence is always differentiable and its
m
n
derivative is obtained from a x
n 0
n by differentiating term by term as we differentiate
polynomial functions.
Theorem 3.2: (Differentiation Theorem for Power series)

m
n
Let a x
n 0
n be a power series with radius of convergence R  0 . Then
m
n 1
 na x
n 1
n has the same radius of convergence and

d  m n
m
d
 
dx  n 0
a n x    na n x n 1
  dx
an x n   for x  R
 n 1 n 1
124
Proof: omitted but if interested please consult any reference material.

Note that the initial value of this series has been changed from n = 0 to n = 1. This
indicates the fact that the derivative of the first term is zero and is not in the derivative.

x n 1 
xn
Example3.8: Show that  n0 n  12
and 
n 1 n 1
have the same radius of
convergence.

x n 1 x 2 x3
Solution: Let f  x    2
 x  
n 0  n  1 4 9

(n  1) x n 
xn
f ' ( x)   2
 
n 1 ( n  1) n 1 n  1
2 2
an 1 x n 2  n  1  n 1 
lim  lim 2 n 1
 lim   x x
n  a
n
n 
 n  2 x n 
 n2
The power series converges absolutely if x  1 and radius of convergence R = 1.
If x  1 , then the series becomes


(1)n 1
 2
n 0  n  1
This is convergent by alternating series test. If x  1 , the power series becomes

(1)n 1 
1
 2
 2
n 0  n  1 n 0  n  1
and converges by p  series test .
In this case the interval of convergence is [1,1] . Again for the second series we find that
an 1 x n  2  n  1  n 1 
lim  lim  lim  x  x
n  a n   n  2  x n 1 n  n  2
n  
Similarly the power series converges absolutely if x  1 and radius of convergence R = 1

n 1

 1 x1, then the series
If x  1 , then n 1 n 1
converges conditionally and if
n 1

1 
1
 n 1   n 1
n 1 n 1
diverges.

125
xn 
The interval of convergence for  is [1,1) .
n 1 n  1

x n1 
xn
Thus f  x    2
and f '  x    have the same radius of convergence
n 0  n  1 n 1 n  1
but not necessarily the same interval of convergence. The differentiation theorem states
 
n n 1
that the radii of convergence of the series a x
n 0
n and  na x
n 1
n are the same but does
not implies that the interval of convergence are the same.
d   x n   xn 1 1 1 4
Example 3.9 Show that      1  x  x 2  x3  x 
dx  n 0 n !  n0 n ! 2 6 24
Solution: If x  0 , then
x n1 n ! 1
L  lim n
 lim x  0  1.
n   n  1! x n n  1
Thus the Generalized Ratio Test implies that the series converges for all x  0 and hence
the series converges for all x .

n x n 1  x n1
f ' x   
k 1 n! n 1  n  1 !

xl 
xn
f ' x      f x x 
l 0 l ! n 0 n !
Recall that from exponential growth, if f is continuous on [0, ) , then f '  x   kf  x 
for all x  0 . In this case fix f  0   1 and k  1 , then integrating we get the only function
which satisfy this property.

f ( x)  e x
Therefore
xn

1 1 (1)
e    1  x  x 2  x3  
x
n 0 n ! 2! 3!
Note that when x  1 we obtain the series
126

1 1 1
e  1  1    ....
n 0 n ! 2 3
When x  1 we obtain the series
n
1 1   1 1 1 1 1
e    1  1      ...
e n 0 n ! 2 3 4 5
When we replace x by x 2
x2 x 2n

2 x 4 x6
e   1 x   
n 0 n ! 2! 3!
If we replace x by  x the series be comes

(1) n x n x 2 x3
e x    1 x    (2)
n 0 n! 2! 3!
Subtract equation (2) from (1)
 1 1 1   x2 x3 
e x  e  x  1  x  x 2  x 3  x 4      1  x     
 2! 3! 4!   2! 3! 
2 3 2 5 2 x 2 n1
 2x  x  x   
3! 5! (2n  1)!
e x  e x x3 x 5 x 2 n1
sinh x   x       x  
2 3! 5! (2n  1)!
x 2 x 4 x6 x2 n
Similarly show that cosh x  1     
2! 4! 6! (2n)!
Example 3.10 Find a power series representation for the following function and
determine its domain.
f ( x )  3x

xn
Solution: Recall that 3x  e x ln 3 , and e x   . Then the power series representation is
n 0 n!
2 3
x

( x ln 3)n  x ln 3  x ln 3
3   1  x ln 3   
n 0 n! 2! 3!
Verify that it has the same radius of convergence and interval of convergence with e x .

127
( x  2)n 
Example 3.11: Find the sum of the series 
n 1 ( n  3)!
Solution: Does the given series look any thing like a series that we already know? Well,
it does have some ingredients in common with the series for the exponential function:
x xn 
x 2 x3
e    1 x   
n 0 n ! 2! 3!
We can make this series look more like our given series by replacing x by x  2

( x  2)n ( x  2)2 ( x  2)3
e x2    1  ( x  2)   
n 0 n! 2! 3!
But here the exponent in the numerator matches the number in the denominator whose
factorial is taken. To make that happen in the given series, let’s multiply and divide by
( x  2)3 .

( x  2)n 1 
( x  2)n 3 1  ( x  2)3 ( x  2) 4 
 (n  3)! = 3  = 3 
( x  2) n0 (n  3)! ( x  2)  3!

4!
 
n 1 
We see that the series between brackets is just the series for e x  2 with the first three terms
missing. Consequently

( x  2)n 1  x2 ( x  2)2 

n 1 ( n  3)!
=
( x  2)3
 e  1  ( x  2) 
2! 


x
Example 3.12 Find a power series which defines g  x   2
1  x 
1 d  1  d   n
Solution: f '  x     x
1  x 
2
dx  1  x  dx  n0 
 
d n
= 
n 0 dx
x  
 
n 0
nx n1 if | x | 1
Now multiply f  by x
x 
xf  x    x  nx n1 for | x | 1.
(1  x)2 n1
x 
Therefore, g  x     nx n  x  2 x 2  3x3   for x  1
2
1  x  n1
128
 
If f  x    an x n , then f ( x)   nan x n1 for all x    R, R  . Here f (0)  a0  0!a0
n 0 n 1
and f (0)  a1  1!an . Since f '  x  is a power series further differentiating we get

f   x    n  n  1 an x n 2 , f (0)  2! a2
n 2

f   x    n  n  1 n  2  an x n3 f (0)  3! a3
n 3


f ( n) ( x)   n !an x n k for all k  n , f ( n) (0)  n !an
n k
Now we state the following theorem.

n
Theorem 3.3: Suppose a power series a x
n 0
n has radius of
convergence R  0 . Let

f ( x )   an x n for  R  x  R
n 0
Then f has derivatives of all orders on (-R, R), and
f ( n) (0)  n !an for all n  0
Consequently

f ( n ) (0) n
f ( x)   x for  R  x  R
n 0 n!
Proof: See from reference book.

129
3.1.2 INTEGRATION OF POWER SEIRES

n
Suppose a x n has a radius of convergence R. Then f  x    an x n is differentiable
on (-R, R) and hence f is continuous on (-R, R). Thus it is possible to integrate the
function f  x    an x n over any closed interval in(-R, R).
n
Theorem 3.4: Let a x n be power series with radius of convergence R  0 .

an n 1
Then  n  1x
n 0
has the same radius of convergence and
 
x
n
 x
n

 an x n1 
0  
n 0
an t 

dt  
n 0
0 na t dx   
n 0  n  1 
 for x  R .
Proof: Exercise.
Example 3.13 Find the series expansion of ln(1  x 2 )
 
1 2 1
Solution: Recall that   xn , if x  1 . Now replace x by  x in   xn ,
1  x n 0 1  x n 0
and then multiply by 2x .

2x
2
  (1)n 2 x 2 n 1
1 x n 0
Integrating we have
x
2t x 
2 n 1 
 
 t 2 n 2 
x
dt  
0 1  t 2 0  n0 (  1) n
2t 

dt  
n 0
(1) n 2  t 
0
x
2 n 1
dt   (1) 2 
n 0
n

 2n  2  0
x
2 x  t 2n  2 

n
 ln(1  t )    (1)  
 n 10
0
n 0

(1)n x 2 n 2 x4 x6
ln(1  x 2 )    x2     for x  1 .
n 0 n 1 2 3
Next we analyze what happen when x  1 and x  1 . At both x  1 and x  1 , the
series become

(1)n 1 1 1
ln(2)    1   
n 0 n  1 2 3 4
130
(1) n x 2 n 2 2

Thus the series converges and the interval of convergence for ln(1  x )  
n 0 n 1
is [1,1] .
Example 3.14 Find a power series representation for the function f ( x )  ln(1  x) and
determine its interval of convergence.
Solution: Again we start with the power series

1
  xn for x  1
1  x n 0
Now replace x by  x

1
  (1) n x n for x  1
1  x n 0
Therefore by the above theorem we find that
x1 x  n n   x n
ln 1  x    dt      1 t dt      1 t n dt
0 1 t 0 0
 n 0  n 0
n
  1 x n1
= 
n0 n 1
n

 1 x n 1
ln 1  x    for x  1
n 0 n 1

1
At x  1 , the series becomes  and it is divergent. When x  1 , the series be
n 0 n  1
n

 1
comes ln 2   and it is convergent.
n 0 n 1
n

 1 x n1
Hence, ln 1  x    for 1  x  1
n 0 n 1
n

 1 x n 1
The power series expansion of ln 1  x    , for -1 < x 1 is some times
n 1 n 1
known as Mercator’s series.
Note that
n 1 n

 1  x  1
ln x  ln 1  x  1   for 0  x  2
n 1 n

131
1  x 
Example 3.15 Express ln  as a power series for x  1 .
1 x 
 1 x 
Solution: Here ln    ln 1  x   ln 1  x 
 1 x 
But
n 2 n 1

 1 x n 1 
 1 x n1
ln 1  x    and ln 1  x   
n 0 n 1 n 0 n 1
n
 1 x  
 1 xn 1 
x n 1
ln  
 1 x 
 ln 1  x   ln 1  x   
n 0 n 1
 1   1
n 2 n 1
  
n 0 n  1
n
 1   1 
 x3 x5 x7 x 2 n1 
 2 x       
 3 5 7 2n  1 
 1 x  
x 2 n1
Therefore, ln  
 1 x 
 2 
n 0 2n  1
for x  1 .
x3 x5 x7
Example 3.16 a) Show that tan 1 x  x      for x  1
3 5 7
b) Find a series expansion of  and approximate  with error less than 10-6

x 1 1 1 n
Solution a) tan 1 x   dt and     1 t 2 n for t  1
0 1 t2 1 t 2
 
1  t 2
n 0
The integration Theorem implies
x
1 1 x  n 2n 

 (1) n x 2 n1 
tan x   dt      1 t  dt    , x  1
0
1 t 2 0
 n 0  n 0  2n  1 

(1)n 2 n1 x 3 x5 x 7
tan 1 x   x  x     ... , x  1
n  0 2n  1 3 5 7
n
1  1  
, then tan 1 
1 
 1
b) If x     n
n  0 3  2 n  1
3,  3  6 3
n

 1  2  1 1 1 
Thus,    n
n  0 3  2 n  1
3 1   
 9 45 189
 .

132
Is it possible for you to approximate ln2 to four decimal place of accuracy?

n
1

 1 x 2 n 1
The series tan x   if x  1 is called Gregory’s series
n 0 2n  1
Although it is not east to prove Gregory’s series it holds for x=1and this yields
n
 
 1 1 1 1
 tan 1 (1)    1   
4 n 0 2n  1 3 5 7
 1 1
Example 3.17 Show that  tan 1    tan 1  
4 2  3
Using this formula and Gregory’s series approximate  accurate to four decimal places
1 1
Solution: Let   tan 1   and   tan 1   . Then
 2  3
1 1 5

tan   tan 
tan       2 3  6 1
1  tan    1  1 . 1 5
2 3 6
Taking tan inverse we obtain
 1 1
    tan 1 (1)   tan 1    tan 1  
4 2  3
1
On the other hand from Gregory’s series with x  we have
2
3 5 7 9 11 13 15
 1 1 1 1 11 1 1 1
1 1 1 1 1 1 1
tan                        
 2  2 3  2  5  2  7  2   9  11  2  13  2  15  2 
 0.5-0.04167+0.00625-0.00112+0.0022
 0.46365 With an error less than the 6th terms
1
Similarly, tan 1    0.32174 with an error less than the 6th terms.
3
 1  1 
  4  tan 1    tan 1   
 2  3 
 4  0.46365  0.32174 
  3.1416

133
Thus we can approximate   3.1416 to four decimal places.

Example 3.18 Find a power series representation for the following function using term
wise integration
x arc tan t
a) f  x   dt
0 t
2
1  e t x
b) f  x   dt
0 t2
x 1
c) f  x   tanh 1 x   dt
0 1 t2
n
1

 1 x 2 n1
Solution: Here tan x   , | x | 1 . Dividing both sides by x implies
n 0 2n  1
tan 1 x x2 x4 x6
 1     ... x 1
x 3 5 7
By integration theorem we get
x tan 1 t x t2 t4 t6 
f  x   dt   1     ... dt
0 t 0
 3 5 7 
x3 x5 x 7
 x   
9 25 49
n
x arc tan t 
 1 x2 n1
0 t
dt  
n  0  2 n  1

xn
b) We know that e x   x  . Replace x by  x 2
n 0 n!
We find that
n
e x 2


x  2


 1
n
x2n
n 0 n! n 0 n!
x4 x6 x8
2
 1 x    
2! 3! 4!
With simply calculation
2
1  et 1  2 t 4 t 6 t8  t2 t4 t6
  t     ... = 1    
t2 t2  2! 3! 4!  2! 3! 4!
Applying integration Theorem yields
134
2
x1  et x t 2 t 4 t 6 t8  x3 x5 x7 x9
f  x   dt  0  2! 3! 4! 5! 
1     ... dt  x     
0 t2 2! 3 3! 5 4! 7 5! 9
Hence,
2 n
1  e t
x 
 1 x 2 n1
f  x   dt = 
0 t2 n  0  n  1 n !

1
c) For x  1, we know that  1  x  x 2  x 3  ...   x n . Now replace x by t 2
1 x n 0

1 2 4 6
2
 1  t  t  t ...   t 2n
1 t n 0
Integration Theorem implies

x 1 x x3 x 5 x 7
dt
0 1  t 2 0 1 
 t 2
 t 4
 t 6
 ... dt  x    
3 5 7
1 1 x 
x 2n 1
tanh x   dx  
0 1 t2
n 0 2n  1

135
Exercise 3. 2
1. Find a power series representation for the following functions

1 1  x2  1 x 
a) f ( x )  b) f ( x)  2
c) f ( x )  ln  2 
3 x 1 x  1 x 
x x 2
d) g ( x )   (et  ln(t  1))dt e) f ( x)   et dt f) g ( x )  x ln(1  x)
0 0
2. For the following functions, use a power series representation to approximate the
definite integral to within 104 .
1 1 0 (e  x  x  1)
a)  2 dx b) 0.01 x2 dx
0 1  x3
1 2
3. Find a series representation for the definite integral  xe x dx
0

11 1
4. Show that 0 1  t 2 
dt  (1)n
n 0 2n  1
5. Find a power series representation for the function and determine the radius of
convergence
x2  x
a) f ( x )  2
b) f ( x)  arctan  
(1  2 x)  3
6. Evaluate the following indefinite integral as a power series
arctan x 1
a)  dx b)  tan ( x 2 )dx c) x
2
tan 1 ( x 4 )dx
x
136
3.2 TAYLOR POLYNOMIALS

In this section we show how a function can be approximated as closely as desired by a
polynomial, provided that the function possesses a sufficient number of derivatives.
Definition 3.3 Let f be a function such that the nth derivative f ( n) exists at a in some
interval containing a as interior point. Then the polynomial
f (a ) f ''  a  2 f n (a)( x  a ) n

pn ( x)  f (a)  ( x  a)   x  a   ...  or
1! 2! n!
n f (k )  a  k
pn ( x )    x  a
k 0 k!
is called the nth degree Taylor’s polynomials f of at a.
Example 3.19 Expand g ( x )  1  2 x 2  5 x 3  9 x 4 in powers of x  1 .

Solution: Here, n =4 and a=1.
g ( x)  1  2 x 2  5 x 3  9 x 4 g (1)  1
g ' ( x )  4 x  15 x 2  36 x 3 g (1)  4  15  36  17

g ' ' ( x)  4  30 x  108 g (1)  4  30  108  74
g ( x)  30  216 x g (1)  30  216  186
g [ 4] ( x )   216 g [ 4] (1)   216

The Taylor’s polynomials of degree 4 is given by
g '' 1 2 g ''' 1 3 g [ 4] (1) 4
p4 ( x )  g 1  g ' 1 x  1   x  1   x  1   x  1
2! 3! 4!
= 1  17( x  1)  37( x  1) 2  31( x  1)3  9( x  1) 4

Example 3.20 Find the fourth degree Taylor’s polynomial of f ( x)  sin x at a  .
2

Solution: Here f ( x)  sin x f ( ) 1
2

f '( x)  cos x f '( )  0
2

137

f ''( x)   sin x f ''( )  1
2

f '''( x)   cos x f '''( )  0
2

f ( 4) ( x)  sin x f ( 4) ( )  1
2
Thus the fourth degree polynomial is given by
   
f ''( ) 2 f '''( ) 3 f (4)   4
   2   2    2  
p4  x   f    f '( )  x    x   x   x 
2  2 2!  2 3!  2 4!  2
2 4
1  1  
=1   x     x  
2 2  24  2
x
Example 3.21 Let f ( x)   tan(sin t ) dt .Then find the Taylor polynomial of order 4 at 0.
0
x
Solution: Here f ( x)   tan(sin t ) dt f (0)  0
0
f ( x )  tan(sin x) f (0)  0
f ( x )  sec 2 (sin x) cos x f (0)  1
f ( x )  sec 2 (sin x)[2 tan(sin x ) cos 2 x  sin x] f (0)  0
f ( 4) ( x)  2 sec 2 (sin x ) tan(sin x) cos x[2 tan(sin x) cos 2 x  sin x ]

2 2 3
f ( 4) (0)  1
 sec (sin x )[2 sec (sin x ) cos x  4 tan(sin x) cos x sin x  cos x]
This implies that
1 2 1 4
P4 ( x )  x  x
2 24
Example 3.22 Let f ( x )  ln(1  x) . Then
i) Find a formula for the nth Taylor polynomial of
ii) Calculate p6 (1) .
Solution: We have f ( x )  ln(1  x) f (0)  0

1
f '( x)  f '(0)  1
1 x
1
f ''( x)  f ''(0)  1
(1  x) 2
138
(1) 2 2!
f '''( x)  3
f '''  0   2!
1  x 
3
[4]  1 3! f [4]  0   3!
f  x  4
1  x 
 
n 1
( n)  1 (n  1)!
f ( x)  n
f ( n) (0)  (1)n 1 ( n  1)!
(1  x )
Therefore the n th Taylor polynomial is

f ''(0) 2 f (0) 3 f ( n ) (0) n
pn ( x)  f (0)  f '(0) x  x  x   x
2! 3! n!
x 2 x3 x 4 (1)n 1 x n
=x    
2 3 4 n
1 1 1 1 1 37
ii) p6 1  1        0.616667
2 3 4 5 6 60
1
Example 3.23 Let f  x   . Then
1 x
i) Find a formula for the nth Taylor polynomial of f
ii) Compute p n (2)
Solution: As in the above example

1
f x  f 0  1
1 x
1
f ' x  2
f '  0  1
1  x 
2!
f ''  x   3
f ''  0   2!
1  x 
3!
f ( x)  4
f (0)  3!
1  x 
 
(n  1)!
f n ( x)  n
f ( n )  0   (n  1)!
1  x 

139
f (0) 2 f ( n ) (0) x n
Hence we get , p n ( x)  f (0)  f (0) x  x 
2! n!
 1  x  x 2  x3    xn
ii) p n (2)  1  2  2 2  2 3  ...  2 n
The Taylor polynomial of degree one, P1 ( x)  f (a)  f (a )( x  a) is familiar with
equation for the tangent line to y  f ( x ) at a point a. Similarly,

1
P2 ( x )  f (a )  f (a )( x  a)  f (a )( x  a) 2 is the tangent parabola to y  f ( x ) at a
2
point a. Since the first n derivatives of Pn ( x) at a are equal to the corresponding first
n derivatives of f ( x ) at a , then Pn ( x) is an approximation for f ( x ) if x is near a .
Exercise 3.3
Find the nth degree Taylor’s polynomials of the following functions at the indicated point
  1 x 
a) f ( x)  cos x at x = d) h (x) = ln  2 
at x  0
4  1 x 
1 x
b) g(x) = at x = 2 e) f ( x )   ln(1  x)dt
1  2x 0
c) g ( x)  cosh x at x  0 .
Definition 3.4: Let p n ( x ) be the n th degree Taylor polynomial of the function f . Then
remainder term, denoted Rn ( x) is given by
Rn ( x)  f ( x)  p n ( x)
Now we state and prove the following theorem.
140
Theorem 3.6: (Taylor’s Theorem with Remainder)
If a function f has n  1 derivatives on an open interval I about a and x is in this

interval, then a number c between x and a exists with
f ( x)  Pn ( x)  Rn ( x)
Where
n
f ( k ) (a ) k f ( n1) (c ) n 1
Pn ( x)    x  a and Rn ( x )   x  a
k 0 k!  n  1!
Proof: We show that if f ( n 1) ( x ) exists in [a, b], then for any number x  [a, b], there is a
number c in [a, x] such that
f (a ) f (a ) f ( n) ( x  a )n
f ( x )  f ( a)  ( x  a)  ( x  a )2     Rn ( x ) (1)
1! 2! n!
Where
f ( n1) (c)
Rn ( x)  ( x  a )n 1 (2)
(n  1)!
Let x  [a, b] be fixed. We define a new function h(t ) by
f (t ) 2 f ( n ) (t ) n Rn ( x )( x  t ) n1
h(t )  f ( x)  f (t )  f (t )( x  t )  (x  t)   (x  t)  (3)
2! n! ( x  a )n1
Then
f ( x ) f ( n ) ( x) R ( x)( x  x)n 1
h( x)  f ( x)  f ( x)  f ( x )( x  x )  ( x  x)2    ( x  x) n  n 0
2! n! ( x  a) n1
and
f (a ) f ( n) (a ) R ( x)( x  a )n 1
h(a )  f ( x)  f (a )  f (a)( x  a )  ( x  a) 2    ( x  a) n  n
2! n! ( x  a )n1
= f ( x)  Pn ( x)  Rn ( x )  Rn ( x )  Rn ( x )  0 .

141
Since f ( n 1) ( x ) exists, f ( n) is differentiable so that h , being a sum of products of

differentiable functions, is also differentiable for t in (a, x) . Remember that x is fixed so
h is a function of t only.
Recall that Rolle’s Theorem which states that if h is continuous on [a, b], differentiable
on (a, b), and h(a )  h(b)  0 , then there is at least one number c in (a, b) such
that h(c )  0 . We see that the condition of Rolle’s Theorem hold in the interval [a, x] so
that there is a number c in (a, x) with h(c )  0
d
( f (t )( x  t ))   f (t )  f (t )( x  t ) Why? (4)
dt
Also for 1  k  n
d  f ( k ) (t ) k f ( k ) (t ) k 1 f ( k 1) (t )
 (x  t)    ( x  t)  ( x  t )k (5)
dt  k !  (k  1)! k!
Using (4) and (5) on (3) it follows that

 f ( n1) (t )( x  t )n (n  1) Rn ( x  t )n
h(t )   .
n! ( x  a )n1
Then, setting t  c , we obtain
 f ( n 1) (c)( x  c )n (n  1) Rn ( x )( x  c)n
0  h(c)   .
n! ( x  a )n 1
Finally, dividing the equations above through by ( x  c )n and solving for Rn ( x) , we
obtain
f ( n1) (c)( x  a )n f ( n 1) (c)( x  a )n1
Rn ( x )   .
(n  1)n ! (n  1)!
This is what we wanted to prove. 
In the above theorem f  x   Pn  x   Rn  x  is called Taylor’s formula and R n is called
Lagrange remainder formula.

Remark: If you go over the proof, you may observe that we didn’t need to assume
that x  a ; if we replace the interval [a, x] with the interval [x, a], then all results are still
valid.
142
Theorem 3.7: (Uniqueness of the Taylor Polynomial)
If f ( x )  Tn ( x)  Rn ( x ) , where Tn is the n th -degree Taylor polynomial of f at a and
lim Rn ( x)  0
n 
for x  a  R , then f is equal to the sum of its Taylor series on the interval
xa  R .
.
Proof: Suppose that there is another nth degree polynomial Qn ( x ) such that
f ( x )  Pn ( x)  Rn ( x)  Qn ( x )  S n ( x)
Where
Rn ( x) S ( x)
lim n
 0, and lim n n  0 .
x a ( x  a) x a ( x  a )
Let
D ( x)  Pn ( x)  Qn ( x )  Sn ( x )  Rn ( x) .
We will show that D ( x )  0 , which will imply that Pn ( x)  Qn ( x) . This will show that
Pn ( x) is unique. Since D ( x ) is the difference of nth degree polynomials, D ( x ) is a
polynomial of degree  n, and it can be written in the form
D ( x)  b0  b1 ( x  a)  b2 ( x  a) 2    bn ( x  a )n .
Now
D ( x) S ( x) R ( x)
lim n
 lim n n  lim n n  0  0  0
x a ( x  a ) xa ( x  a ) xa ( x  a)
Similarly, if 0  m  n ,
D ( x) ( x  a) n  m D ( x)
lim  lim (Multiply and divide by ( x  a )n m )
x a ( x  a )m xa ( x  a )n
D( x)
= lim( x  a) n m lim  00  0
x a xa ( x  a) n
and we have

143
D ( x)
lim  0 for m  0,1, 2, , n . (6)
x a ( x  a )m
To complete the proof, we note that b0  D(a )  lim D ( x)  0 by (6) with m  0 .

xa
Then
D ( x)  b0  b1 ( x  a)  b2 ( x  a) 2    bn ( x  a )n
so that
D( x)
 b1  b2 ( x  a)    bn ( x  a )n1
xa
and
D( x)
b1  lim  0 by (6) with m  1
xa xa
Suppose we have show that b0  b1    bk  0, where k  n . Then
D ( x)  bk 1 ( x  a) k 1  bk  2 ( x  a) k  2    bn ( x  a )n
so that
D( x)
k 1
 bk 1  bk  2 ( x  a )    bn ( x  a) n k 1
( x  a)
and
D( x)
bk 1  lim  0 . By (6) with m  k  1  n .
x a ( x  a )k 1
This show that b0  b1  b2    bn  0 , which means that D ( x )  0 for every x in [a, b]
so that Pn ( x)  Qn ( x) for every x in [a, b]. Thus the Taylor polynomial is unique. 
This theorem helps us to show that when ever the Taylor polynomial converges it
converges to itself.
144
Theorem 3:8 (Taylor’ Inequality) If f has n  1 continuous derivatives, then there

exists a positive number M n such that
n 1
xa
Rn ( x )  M n
 n  1!
for all x in [a, b] . Here M n is an upper bound for  n  1 st derivative of f in the
interval [a, b] .
Proof: Since f ( n1) is continuous on [a, b], it is bounded above and below on that
interval. That is, there is a number M n such that f ( n1) ( x )  M n for every x in [a, b].
Since
( n 1) ( x  a) n1
Rn ( x )  f (c )
(n  1)!
With c in (a, x), we see that
( n 1) ( x  a) n1 ( x  a) n1

Rn ( x )  f (c )  Mn
(n  1)! (n  1)!
And the theorem is proved. 
Examples 3.24 Write a Taylor’s formula for g  x   tan 1  x  , n  3, at a  1

g  x   tan 1  x  g (1) 
4
1 1
g ( x)  g (1) 
1  x2 2
2 x 1
g ( x )  2
g (1)  
1  x 2 2
2  6 x 2 1
g ( x )  g (1) 
2 3 2
1  x 
18 x  24 x 3 18 z  24 z 3
g (4) ( x)  g (4) ( z ) 
2 4 2 4
1  x  1  z 
Therefore the required formula is given by

145
g  x   P4  x   R4  x 
g '' 1 g ''' 3 g  4  z  4

 g 1  g ' 1 x  1  1 x  1 2  1 x  1   x  1
2! 3! 4!
 1 1 2 1 3 18 z  24 z 4 4
   x  1   x  1   x  1  4  x  1
4 2 4 12 1  z 2 
If in definition 6.3 a is replaced by 0 , then the Taylor’s Formula reduced to
f ''  0  x 2 f  n  0  n f  n1  z  n1

f  x   f 0  f ' 0    x  x
2! n!  n  1!
for some z between 0 and x . This formula is called Maclaurin formula.
Example 3.25: Find the Maclaurin formula for f ( x )  e x .

Solution: Here we calculate derivative of few terms
f ( x)  e x f 0  1
f '  x   ex f '  0  1
 
f n  x   ex f n 0  1
f  n1  z   e z
The required Maclaurin formula is

x 2 x3 xn ez
ex  1  x    ...   x n 1
2! 3! n !  n  1!
146
3.4 Approximation by Taylor’s polynomials

If Pn  x  and Rn  x  are Taylor’s polynomials and the nth remainder terms of f at a
respectively, then
f  x   Rn  x   Pn  x 
f  x   Pn  x   Rn  x 
If we can find a number M such that Rn  x   M then we will have
f  x   Pn  x   M
Or
Pn  x   M  f  x   Pn  x   M
We can write
f  x   Pn  x   M
This means f  x   Pn  x  with an error of not more then M limits.
Example 3.26: Approximate sin x by a fourth degree polynomial in x if 0  x  0.2

Solution: Here f  x   sin x f (0)  0
f '  x   cos x f '  0  1
f ''  x    sin x f ''  0   0
f '''  x    cos x f '''  0   1
f ( 4) ( x)  sin x f (4) (0)  0
f 5  z   cos z
x3 cos z 5
Therefore, P4  x   x  and R4  x   x . Now for 0  z  0.2 , take x  0.2 and
6 5!
1
 0.2   2.6 106  3 106 since cos z  1 .
5
0  R4 ( x) 
120
Hence in the given range sin x will be approximated by
x3
sin x  x   0.000003
3!

147
3
Example 3.27: Approximate e to an accuracy of five decimal places.
Solution: Recall that the Taylor’s formula for
x2 xn ez ez
ex  1  x   ...   x n1 and Rn  x   x n 1 , for 0  z  x .
2! n !  n  1  
n  1 !
1
1
We need to find n so that Rn    105 . Clearly e 3  2 and so
3
n 1
1 2 1 2
R4      
 3   n  1 !  3   n  1!3n 1
1 2 1
If n  5 , then R5    6
  105 .
 3  6!3 360  729
2 3 4 5
3 1 11 1 1 1 1 1 1
Thus e  1               1.39563
3 2  3  3!  3  4!  3  5!  3 
3.5 Taylor series

In section 6.1.1 we have seen that the natural exponential function can be represented by
a power series that converges for all real numbers. In this section we discuss power series
representations for other functions.
1) Which functions have power series representation?
2) Is the Taylor’s series of f equal to f in its interval of convergence?
Definition 3.5: Let f has derivatives of all orders at c . Then the Taylor’s series of f at
c is defined as
n

[n]  x  a f ''  a  2 f n  a  n
f (a )  f  a   f '  a  x  a    x  a  ...   x  a  ...
n 0 n! 2! n!
If a = 0, then series is called the Maclaurin series of f .
Example 3.28 Find the Taylor series for f ( x)  ln( x ) at x  2
Solution: Here we calculate the derivative of few terms and evaluate at x  2 .
f ( x)  ln( x ) f (2)  ln 2
1 1
f ( x)  f (2) 
x 2
148
1 1
f ( x )   f (2)  
x2 22
2 2
f ( x)  f (2) 
x3 23
2(3) 2(3)
f (4) ( x)   f (4) (2)  
x4 24
2(3)(4) 2(3)(4)
f (5) ( x )  f (5) (2) 
x5 25
 
(1) n (n  1)! (1)n (n  1)!
f (n ) ( x)  f ( n ) (2)  n 
xn 2n
In order to find the Taylor series expansion we apply the above definition.

f ( n ) (2)
ln( x )   ( x  2) n
n 0 n!

f ( n ) (2)
 f (2)   ( x  2) n
n 1 n!

(1)n 1 (n  1)!
 ln 2   n
( x  2)n
n 1 n !2

(1)n 1
ln( x )  ln 2   n
( x  2) n
n 1 n 2
Now we state the following theorem which answer question (2).
Theorem 3.9 If a function f has derivatives of all orders on an open interval about a
and if
lim Rn ( x)  0
n 
for all x in that interval, then f has the Taylor-series representation


f ( n) (a)( x  a )n
f ( x)  
n 0 n!
for all x in the interval.
Proof: See reference book

149
 x 21
 , x  0 . Find the Taylor series of f at a  0 .
Example 3.29 Let f ( x)  e
0 , x0
Solution: First show that for any n  ,

1
lim 1
0 (1)
x0
n 2
x ex
1
To see this we substitute y  to get
x2
1 1  yn 
lim 1
 lim 1
 x n   lim y   lim xn  0
x0
xn e x2
x 0
x 2n e x2  y  e  x 0
Next we compute f (0) .

1

x2
f ( x )  f (0) e 0 1
f (0)  lim  lim  lim 1  0 by (1)
x 0 x0 x  0 x x  0
2
xe x
For x  0
1 1
d   x2  3
 2
f ( x)  e   2 x e x
dx  
Thus,
 3 x21
 , x0
f ( x)  2 x e
0 , x0
Next we compute f ( x ) .

1

3 x2
f ( x )  f (0) 2x e  0 1
f ( x )  lim  lim  2 lim 1
 0 by (1)
x 0 x0 x  0 x x0
4 x2
xe
Again for x  0
1 1
d  3  x2   2
f ( x )   2x e
dx 

4 6
  6 x  4 x e x


Similar calculation will lead to
f (0)  f (0)  f (0)    0
Thus we have
150

f ( n ) (0) x n 
0  xn
   0  0x  0x2   0
n 0 n! n 0 n !
1
2

f ( n) (0) x n
Clearly, at any x  0 , f ( x )  e x  0 . Therefore, f ( x )  
n 0 n!
Example 3.30 Find the Maclaurin series expansions of f ( x)  e x and show it
represents e x x  .
and f 
n
Solution: Recall that f [ n]  x   e x , n   0  e1  1 n 
The Maclaurin series of f is given by
f '' 2 f ''' 3 f n n

f ( x)  f  0   f '  0    0  x   0  x  ...  x  Rn  x 
2! 3! n!
x2 x3 xn
 1  x    ...   Rn  x 
2! 3! n!

xn ez
   Rn  x  where Rn  x   x n1
n 0 n !  n  1!
We have shown that the series converges on  ,   . Does it converge to e x x  ?
f  n 1  z  n1 e z x n 1
Rn  x   x = , where z lies between 0and x
 n  1!  n  1!
To answer this question we need to show that lim Rn  x   0 . For this we consider the
n 
following cases.
Case 1: If 0  z  x , then e z  e x .
e z x n1 e x x n 1
0  lim Rn  x   lim  lim 0
n  n   n  1 ! n   n  1 !
Therefore, by squeezing theorem

lim Rn  x   0
n 
Case 2: If x  z  0 , then e x  e z  1
e z x n 1 z x
n 1
0  Rn  x   e
n 1 n 1

151
0  lim Rn  x   0
n 
lim Rn  x   0
n 
Hence lim Rn  x   0 in both case it converges to itself and thus

n 

xn
ex  
n 0 n!
Example 3.31 Find the Maclaurin series of f ( x)  sin x and prove that it represents f for
all x 
Solution: The Maclaurin series of expansion f(x) is
f ''  0  f '''  0 
f  0   f '  0 x  x2  x 3  ...
2! 3!
f ( x )  sin x f (0)  0
f ' (0)  cos x f'(0)=1
f '' (0)   sin x f''(0) = 0
f ''' (0)   cos x f''' (0) =-1
f
4
f
4
 x   sin x,  0  0
k
f    x    1 sin x f   0  0
2k 2k
k k
f  2 k 1  x    1 cosx f  2k 1  0    1 .
The required Maclaurin series is

k
x3 x5 x7  1 x 2 k 1
sin x  x     ...   R2k 1  x 
3 5 7 2k  1
f  2 k  2  z  x 2 k  2
Where R2 k 1  x   for 0  z  x
2k  2
To show that this series represents sin x for all values of x , we must show that
lim Rn ( x)  0
n 
for all values of x .

Taylor’s Theorem implies that for a fixed but arbitrary value of x there is a number
z between x and zero with
152
f  2 k  2  z  x 2k  2 x2k2
0  R 2k 1  x    f  2 k  2  z 
 2k  2  2k  2
Observe that
f  2 k  2  z   cos z  1 and g  z  | sin z | 1
It follows that
x 2 n 2
0  lim Rn  x   lim 0
n  n  2n  2
By squeezing theorem lim Rn  x   0

n 
Thus for all real number x Maclaurin series for sin x converges to sin x.
2 n 1
x3 x5 x7 x9 
 1 x
sin x  x       
3! 5! 7! 9! n  0  2 n  1

If x  , then the values of sin x for five decimal places of accuracy is
4
3 5 7 9
       
4 4 4 4
  
sin               
4 4 3! 5! 7! 9!
 
sin    0.70710
4
Example 3.32 Find the Maclaurin series expansion of f ( x)  cos x and show that it
represented by f ( x ) for all x.
Solution: From example 6.31 we know that
2 n 1
x3 x5 x7 x9 
 1 x
sin x  x        
3! 5! 7! 9! n  0  2 n  1
d d  x3 x 5 x 7 x 9   d   1 x 2 n1 
 
sin x   x           
dx dx  3! 5! 7! 9!  n 0 dx   2n  1 
(1)n x 2 n

cos x  
n 0  2n  !
From Taylor’s formula for cos x with n  2k we have

153
x 2 x 4 x6 (1)n x 2n
cos x  1        R2 k
2! 4! 6! (2n)!
We need to show that lim Rn ( x)  0 for all x.
n 
2 k 1
f ( 2k 1) ( z ) x 2k 1 x
0  R2k ( x)   f (2k 1) ( z )
(2k  1)! 2k  1
2 k 1
0  R2k ( x) 
 sin z  x2 k 1 
x
Since sin z  1 .
(2k  1)! 2k  1
Computing limit we get

2 k 1
x
0  lim R2 k ( x )  lim 0
k  k  2 k  1
Therefore, by the squeeze theorem for limits at infinity, lim Rn ( x)  0 .

n 
Thus the Maclaurin series of cos x converges to itself.

(1)n x 2 n
cos x  
n 0  2n  !
Example 3.33 Find the Maclaurin series for f ( x)  sin 2 x .

Solution: From trigonometric identity
1  cos 2 x 1 1
sin 2 x    cos 2 x
2 2 2
From example 6.25

(1)n 2 2 n x 2 n
cos 2 x  
n 0  2n  !
Now substituting and simplifying we obtain
1 1  (1)n 22n x 2n
sin 2 x   
2 2 n 0  2n  !
1 1 2 x 2 23 x 4 2 5 x 6
=     
2 2 2! 4! 6!
23 x 4 25 x 6
= x2   
4! 6!
154
2(1) n1 22 n 1 x 2n

sin x  
n 1  2n  !
Example 3.34 Find the first three terms of the Maclaurin series for f ( x)  e x cos x .
x 2 x3 x4
Solution: e x  1  x    
2 6 24
x2 x4 x6
cos x  1    
2 24 720
 x 2 x3 x 4  x 2 x 4 x6 
e x cos x   1  x       1     
 2 6 24  2 24 720 
x2 x4 x6 x3 x 5 x 7 x2 x4 x6 x8
=1     + x     +    
2 24 720 2 24 720 2 4 48 720
 1 1  1 1  1 1 1 
= 1  x      x 2      x3      x 4  
 2 2  2 6  24 4 24 
x3 x 4
=1  x   
3 4
1
 x2
Example 3.35 Approximate e dx with an error less than 0.0005 .
0
xn x

 x2

(1)n x 2 n
Solution: Here e   and e  
n 0 n ! n 0 n!
By integration theorem we have
n 2n n 2n
1
 x2
1  ( 1) x  1 ( 1) x 
(1)n
0 e dx  0 
n 0 n!
dx   
n 0
0 n!
dx  
n  0 n !(2 n  1)
1 2
By alternating series test we can approximate  e x dx to any desired degree of accuracy
0
1 2
x
and hence e dx  sn with an error less than 0.0005. Now find n so that
0
1
an1  0.0005 
2000
1 1

 n  1! 2n  3 2000
Solving this we get n = 5 and
1
 x2 1 1 1 1 1
0 e dx  1  3  2!5  3!7  4!9  5!11  0.7467

155
√ Check list
Mark a tick (√) against each of the following tasks that you can perform. I can
 Give definition of power series -------------------------------------------------
 Give the difference between series and power series------------------------
 Define radius of convergence and interval of convergence------------------
 Calculate radius of convergence and interval of convergence-----------------
 Give power series representation of a function---------------------------------
 State differentiation theorem -----------------------------------------------------
 Differentiate power series---------------------------------------------------------
 State integration theorem ---------------------------------------------------------
 Integrate power series -------------------------------------------------------------
 Give power series representation of sinx, cosx, arctanx, etc ------------------
 Define Taylor polynomials
 Approximate a function by Taylor polynomial-------------------------------
 Define Taylor series-------------------------------------------------------------
 Define Maclaurin series-------------------------------------------------------------
 Give an expression for the nth term Taylor polynomial ---------------
 Give Taylor series and Maclaurin series representation of a function--
 State Taylor’s Theorem-------------------------------------------------------------
156
Summary
An infinite series of the form

n 2 n
 a  x  c
n  ao  a1  x  c   a2  x  c   ...  an  x  c   ... , is called a power series
n 0
n
centered at x = c where an and c are constants. The term an  x  c  is the nth term and
the number c is the center.

The interval on which the power series converge is called the interval of convergence or
the domain of a power series.
m m
n 1
Let  an x n be a power series with radius of convergence R  0 . Then
n 0
 na x
n 1
n has the
same radius of convergence and


d  m n
m
d
 
dx  n 0
a n x    na n x n 1
 
n 1 dx
an x n   for x  R
 n 1

an n 1
Let  an x n be power series with radius of convergence R  0 . Then  n  1x
n 0
has the
same radius of convergence and

x 
n
 x
n

 an x n1 
0  
n 0
ant dt    ant dx   
 n 0
0
n 0  n  1 
 for x  R .
Taylor’s Theorem with Remainder
If a function f has n  1 derivatives on an open interval I about a and x is in this

interval, then a number c between x and a exists with
f ( x)  Pn ( x)  Rn ( x)
Where
n
f ( k ) (a ) k f ( n1) (c ) n 1
Pn ( x)    x  a and Rn ( x )   x  a
k 0 k!  n  1!

157
Taylor’s Series
If a function f has derivatives of all orders on an open interval about a and if
lim Rn ( x)  0
n 
for all x in that interval, then f has the Taylor-series representation


f ( n) (a)( x  a )n
f ( x)  
n 0 n!
for all x in the interval.
Series of constants
1
 1 2  3  4  n  n(n  1)
2
n(n  1)(2n  1)
 12  22  32    n 2 
6
n (n  1)2
2
 13  23  33  43    n3   (1  2  3  4    n)2
4
Series of function
xn
1 1
 e    1  x  x2  x3  
x
  x  
n 0 n ! 2! 3!
2 3
x ( x ln a ) n

 x ln a   x ln a 
 a   1  x ln a      x  
n 0 n! 2! 3!
x3 x5 x7 x9 
 1 x 2n 1
 sin x  x        
3! 5! 7! 9! n  0  2 n  1
 n 2n
x 2 x 4 x6 (1)n x 2 n (1) x
 cos x  1       
2! 4! 6! (2n)! n 0  2n  !
1 (1)n 2 n1

x 3 x5 x 7
 tan x   x  x     ... , x  1
n  0 2n  1 3 5 7
x3 x 5 x 2 n1
 sinh x  x    
3! 5! (2n  1)!
x 2 x 4 x6 x2 n
 cosh x  1       
2! 4! 6! (2n)!
x 1 
x 2n 1
 tanh 1 x   dx  
0 1 t2
n 0 2n  1
158
Exercise 3.4
1. a) Find an expression for the nth Taylor polynomial Tn ( x) for f ( x)  5 x at x=0.
b) Find an upper bound for f ( n1) ( x) when x  1 .
c) Show that f ( x)  5 x is equal to its Taylor series for x  1 .
2. Find the Maclaurin series for the following functions.

x
et  1
a) g ( x)  cosh x b) f ( x)  coth x c) f ( x )   dt
0
t
3 x  2
d) g ( x )  e) h( x)  x 2 3x f) f ( x)  x sin 3x
2 x 2  3x  1
3. Find the Taylor polynomial of the given degree n for the function fat the number a.
a) f ( x)  x , a  4, n  5 b) f ( x)  ln(cos x ), a  0, n  3
1
4. Find the fourth Taylor polynomial a bout 3 for f ( x )  and also determine the
x2
remainder for the polynomial.
5. a) Find the Taylor series for f ( x)  ln( x ) at a  1 .
b) For what values of x this series have the sum equal to ln( x ) ?
 e x 1  x 
6. Find a power series representation for   and use it to evaluate
 x2 
ex 1  x
lim
x0 x2
n
 1   x
7. a) Show that ln  
 1  x  n1 n

1
b) Using (a) show that ln 2   n
n 1 n 2
8. Approximate the value of the integral with an error less than the given error, by first
using the integration theorem to express the integral as an infinite series and then
approximating the infinite series by an appropriate partial sum for the following
series.
1x3 0 2
a) f ( x )   dx, 103 b) f ( x)   e x dx, 103
0 2 x 1

159
x
9. a) Give a power series representation of  sin t 2 dt
0
1
b) Using (a) approximate 2
 sin t dt to with in 104 .
0
10. Find the values of x for which the following series converges.
 n  n
 cos x  1  x 1 
a)    b)   
n 1  n  n 1 n  x 
11. Show that if x  0 , then

n
x 2i 1 x2 n 3
a) sin x   ( 1) i 
i 0 (2i  1)! (2n  3)!
n
x 2i x 2n  2
b) cos x   (1)i 
i 0 (2i )! (2n  2)!
160
References:
1. Robert Ellis and Denny Gulick: Calculus with analytic geometry, 3rd edition.
2. Thomas Fenny: Calculus
3. Edwards and Penney: Calculus and analytic geometry, 3rd edition.
4. J.Douglas Faires and Barbara T. Faires: Calculus of one variable, 2nd edition
5. Stanley I. Grossman: Multivariable calculus, Linear algebra and differential
equations, 2nd edition
6. James Stewart: Multivariable Calculus, 2nd edition

161

Power Series

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Power Series

Uploaded by

Copyright:

Available Formats

UNIT THREE POWER SERIES

School of Humanities And

Prepared by eTeaching Team based

on the script provided by

Written by:Ato Legesse Lemecha

Edited by Ato Tamirat Temesgen

ADU DEPARTMENT OF MATHEMATICS

Definition 3.1 An infinite series of the form

the number c is the center.

is a power series centered at x=0.

this series. Now we have the following definition.

ADU DEPARTMENT OF MATHEMATICS

Remarks: i) When R   , the interval of converges is the set of all real

The interval of convergence is equal to  ,   and radius of converges R  .

ADU DEPARTMENT OF MATHEMATICS

If x  2 , the given power series converges. If x  2 apply Generalized Ratio Test.

L  2 4 x  8  1 , the given power series diverges.

ADU DEPARTMENT OF MATHEMATICS

Here the power series converges if x 2  1 and it diverges if x 2  1 . So we have

The radius of convergence is R  1 . Next we analyze for end points.

converges and the interval of convergence is [1,1] .

ADU DEPARTMENT OF MATHEMATICS

Solution: For this power series, we have

upper bound of lim n an  1 . It follows that

ADU DEPARTMENT OF MATHEMATICS

the series is absolutely convergent when x  c .

Proof: see reference book.

The inequality  x  (7)   1 is equivalent to 8  x  6 .

ADU DEPARTMENT OF MATHEMATICS

3. Find the values of x for which the following series converges.

ADU DEPARTMENT OF MATHEMATICS

3.1 POWER SERIES REPRESENTATION OF FUNCTIONS

In this series if we take a  1 and r  x , then the series be comes

Example 3.6 Consider a geometric series with r  x and a  1

Again, if we replace x by ( x 2 ) in (1) we get

If in (1) we replace x by (-x2) we have

So the interval of absolute convergence is  3 5  x  3 5 and hence the domain of f is

ADU DEPARTMENT OF MATHEMATICS

Thus the required power series representation of

3.1.1 DIFFERENTIATION OF POWER SERIES

Theorem 3.2: (Differentiation Theorem for Power series)

Proof: omitted but if interested please consult any reference material.

The power series converges absolutely if x  1 and radius of convergence R = 1.

If x  1 , then the series becomes

Similarly the power series converges absolutely if x  1 and radius of convergence R = 1

ADU DEPARTMENT OF MATHEMATICS

not implies that the interval of convergence are the same.

Recall that from exponential growth, if f is continuous on [0, ) , then f '  x   kf  x 

which satisfy this property.

Note that when x  1 we obtain the series

ADU DEPARTMENT OF MATHEMATICS

Now we state the following theorem.

Proof: See from reference book.

ADU DEPARTMENT OF MATHEMATICS

3.1.2 INTEGRATION OF POWER SEIRES

ADU DEPARTMENT OF MATHEMATICS

The integration Theorem implies

Is it possible for you to approximate ln2 to four decimal place of accuracy?

ADU DEPARTMENT OF MATHEMATICS

Thus we can approximate   3.1416 to four decimal places.

Applying integration Theorem yields

Integration Theorem implies

ADU DEPARTMENT OF MATHEMATICS

1. Find a power series representation for the following functions