CHAPTER FOUR

POWER SERIES
Definition 2.1 (Power series in x or centered at 0)

A series of the form a x
k 0
k
k
 a0  a1 x  a2 x 2  a3 x 3    ak x k   where x is a variable and
'
the constants a s are the coefficients, is called a power series in x .
k

Definition2.2 (Power series in x  c or centered at c)


A series of the form a0  a1 x  c   a2 x  c     ak x  c      a x  c 
2 k k
k is called a
k 0

power series in x  c  or a power series centered at c or about c, where c is any constant.

Note: Any finite partial sum of the above power series is simply a polynomial in x.
Examples:

xn
1. 
n 0 n!
is a power series in x


x  1k
2. 
k 0 2k
is a power series about -1, or centered at -1.


x  1 / 2k
3. 
k 0 k  1!
is a power series in x  1/ 2 or centered at ½.

2.2. Convergence, Divergence, Radius and Interval of a power series

The major characteristic of a power series is that it involves a variable, and its convergence depends
on the value that x takes. Consequently, it is important to know the set of numbers that make the power
series converge.

Example: For what value(s) of x does the power series  ax
n 0
n
converge, for a  0 ?


Solution:  ax
n 0
n
 a  ax  ax 2  ax3   is a geometric series with first term a1  a and

a1
common ratio r  x which converges to for r  1. i.e., for x  1 .
1 r

a
Thus,  ax
n 0
n
converges to
1 x
for x  1.

Definition2.3
The set of values of 𝑥 for which the power series converges is called the interval of convergence.
The radius of convergence of the power series, denoted by 𝑅, is the distance from the centre of the series to the
boundary of the interval of convergence.
Note: We usually test a power series for absolute convergence (since the terms could be positive or negative).
Recall that a series ∑ 𝒂𝒌 converges absolutely if the series ∑ |𝒂𝒌 | converges.
Recall also that absolute convergence implies convergence.

1
 
Theorem 2.1: For any given power series of the form  a x  c  , there are only the
k
k
k 0

following three possibilities:

(a) the series converges only for x  c or
(b) the series converges for all x  , or
(c) There is a positive number R such that the series converges if x  c  R and

diverges if x  c  R .
Remarks:
1. In case (a) of theorem 2.1, the radius of convergence is 0 and in case (b), the radius is
.
2. If R  , then its interval of convergence is  ,.
3. There are four possibilities for the interval of convergence of the power series with
respect to its radius of convergence R; these are:
c  R, c  R , c  R , c  R , c  R, c  R or c  R , c  R .
4. To determine the radius and interval of convergence of a power series, we can apply
the absolute ratio test or the generalized convergence test.
Furthermore, the radius of convergence of the power series can be computed using
an
the formula R  lim .
n  a
n 1

Example: Determine the radius and interval of convergence of each of the following power
series.

a. ∑∞
(𝑥−3)𝑛 
 3n x n
𝑘=0 𝑛
(𝑥−5)𝑘
e.  n 1
b. ∑∞
𝑘=0 𝑘(2𝑘 )
n 0

x  5k
c. ∑∞
𝑘=0
(−1)𝑘 𝑘 2 𝑥 2𝑘
𝑘!
f. k 1 k2

xn


d.
n 0 n !
g.  n! x
n 0
n

Solution :

a. Let an 
x  3n  an1 
x  3n1
n n 1
a n 1
 lim
 x  3 n
Then lim
n  a
n
n  n 1

2
 n
 x  3 lim  x  3 1
n  n  1

 x3  r

Then by absolute ratio test, the given series converges for r  1,

i.e. r  x  3  1  2  x  4,

But the absolute ratio test tells us nothing if r  x  3  1. In this case, we use direct

substitution to determine whether the power series converges or diverges at the end points
of the interval, with x  2 and x  4 .

2  3n 
 1n
Thus, if x  2, the series becomes 
n 1 n

n 1 n
which is a convergent alternating

series

4  3n 
1
And if x  4, the series becomes 
n 1 n
  , which is the divergent harmonic series.
n 1 n

Therefore, the power series converges for all x  2,4 .

ak 1/ k
This means, its radius of convergence is R  1 ( R  lim  lim  1 ) and its
k  a k  1 / k  1
k 1

interval of convergence is 2,4 .

(𝑥−5)𝑘
b. ∑∞
𝑘=0 𝑘(2𝑘 )
(𝑥−5)𝑘 (𝑥−5)𝑘+1 (𝑥−5)𝑘 (𝑥−5)
Let 𝑎𝑘 = and 𝑎𝑘+1 = (𝑘+1)(2𝑘+1 =
𝑘(2𝑘 ) ) 2(𝑘+1)2𝑘
𝑎𝑘+1 𝑘(𝑥−5) 𝑥−5
Using lim | 𝑎𝑘
| = lim |2(𝑘+1)| = | 2 |
𝑘→∞ 𝑘→∞
Then by the absolute ratio test, the given series converges for r  1,
𝑥−5
| | < 1 ⟹ |𝑥 − 5| < 2 ⟹ 𝑅 = 2
2
And also |𝑥 − 5| < 2 ⇔ 3 < 𝑥 < 7
(−2)𝑘 (−1)𝑘
If 𝑥 = 3 the series becomes ∑∞
𝑘=0 = ∑∞
𝑘=0 which is convergent by alternating series
𝑘(2𝑘 ) 𝑘

test.
(2)𝑘 1
But if 𝑥 = 7 the series becomes ∑∞
𝑘=0 𝑘(2𝑘 = ∑𝑘=0 𝑘 is a harmonic series, which is divergent.
∞
)

Thus the interval of convergence of the series is [3, 7).

(−1)𝑘 𝑘 2 𝑥 2𝑘
c. ∑∞
𝑘=0 𝑘!

3
 (−1)𝑘 𝑘 2 𝑥 2𝑘 (−1)𝑘+1 (𝑘+1)2 𝑥 2𝑘+2
Let 𝑎𝑘 = 𝑘!
and 𝑎𝑘+1 = (𝑘+1)!
𝑎𝑘+1 𝑘+1
Using lim | | = lim |( 𝑘 2 ) 𝑥 2 | =0
𝑛→∞ 𝑎𝑘 𝑛→∞
So the radius of convergence is infinite and interval of convergence is (−∞, ∞)
a n 1 x n 1 n ! 1
d. Let r  lim  lim  x lim 0
n  a n  n  1! x n n  n  1
n

 r  0  1, Thus, the series converges for all x  ,

Thus, its radius of convergence is R   and its interval of convergence is  ,  .

a
e. r  lim n 1  lim
 3n1 x n1 n 1  3 x
n  a
n
n   n  2   3x 
n

Then by the absolute ratio test, the given series converges for r  1,

1
 3 x 1  x  1 . R .
3 3

1 1
If x   , the series becomes
3
 n 0 n 1
which can be shown to be a divergent series. We

can check it by integral test.

1 
 1n
But if x 
3
, the series becomes 
n 0 n 1
which is a convergent alternating series.


 3n x n  1 1
Hence, the interval of convergence of the power series 
n 0 n 1
is   ,  .
 
3 3

f. Let a k 
 x  5
k
 a k 1 
 x  5
k 1

k2 k  12

Then r  lim
x  5k 1 . k 2  x  5 lim
 k 
2

  x5
k  k  12 x  5k k  1  k
 

Then, by absolute ratio test, the series converges absolutely if x  5  1,

i.e., for 4  x  6, and diverges for x  4 and x  6 .

To determine the convergence behavior of the power series at the end points,
x  4 and x  6, we substitute the numbers directly instead of x as follows:

x  5k 
 1k ,
- If x  4, 
k 1 k2

k 1 k2
which converges.

4
 
x  5k 
6  5k 
1
- If x  6, then 
k 1 k 2

k 1 k 2

k 1 k
2
, which also converges.

Hence, for the given power series, its radius of convergence is R  1 , and its interval of

convergence is 4,6 .

g. r  lim
a n 1
 lim
n  1! x n1
n  a
n
n  n! x n

 if x  0
 lim x n  1  
n
 0 if x  0
 The series converges only for x  0 which means R  0 .

2.3. Algebraic Operations On convergent power Series

(Reading Assignment for Learners)

Sum and Difference of power series

Suppose that we have two functions defined by power series
f(x) = a0 + a1(x - c) + a2(x - c)2 + a3(x - c)3 + …
with radius of convergence R1 and
g(x) = b0 + b1(x - c) + b2(x - c)2 + b3(x - c)3 + …
with radius of convergence R2 ,
then

f ( x)  g ( x)   (an  bn )( x  c) n
n 1

and the radius of convergence is

(i) min{R1, R2} if R1 ≠ R2
(ii) ≥ R1 if R1 = R2

5
 Product of power series
Suppose that we have two functions defined by power series
f(x) = a0 + a1(x - c) + a2(x - c)2 + a3(x - c)3 + …
with radius of convergence R1 and
g(x) = b0 + b1(x - c) + b2(x - c)2 + b3(x - c)3 + …
with radius of convergence R2 ,
then
f(x)g(x) = a0 b0 + [a1 b0 + a0 b1 ](x - c) +
[a2 b0 + a1 b1 + a0 b2](x - c)2 +
[a3 b0 + a2 b1 + a1b2 + a0 b3](x - c)3 + …

and the radius of convergence is min{R1 , R2} unless

either f(x) = 0 or g(x) = 0.

Reciprocal of a Power series

Theorem 2.2:
Suppose that we have a function defined by power series
f(x) = a0 + a1(x - c) + a2(x - c)2 + a3(x - c)3 + …
with radius of convergence R and a0 ≠ 0, then

1
has a power series of the form  q n ( x  c) n
f ( x) n 0

with positive radius of convergence ≤ R and further more

1
q0 
a0

6
 Using Long Division to find the reciprocal
of a power series
Example: x2 x4 x6
We know that cos x  1      ( see page13 )
2! 4! 6!

and cos 0 ≠ 0 , therefore we have a power series of sec x

around the point x = 0.
This power series can be found by the long division

x2 x4 x6
[1     ] 1
2 24 720

x2 5x4
1  
2 24
x2 x4 x6
[1     ] 1
2 24 720 x2 x4 x6
1   
2 24 720
x2 x4 x6
  
2 24 720
x2 x4 x6 x8
   
2 4 48 1440

5 x 4 14 x 6
 
24 720
x2 5 4 61 6
 sec x  1   x  x 
2 24 720

7
 2.4. Geometric power series
Recall the formula for the sum of a geometric series:
𝑎
𝑎 + 𝑎𝑟 + 𝑎𝑟 2 + 𝑎𝑟 3 + ⋯ = 1−𝑟, |𝑟| < 1
As we have seen, this formula works perfectly well when 𝑎 and 𝑟 are functions of 𝑥 . For example,
plugging in 𝑎 = 1 and 𝑟 = 𝑥 gives the formula
1
1−𝑥
= 1 + 𝑥 + 𝑥 2 + 𝑥 3 + ⋯,
There are many more series we can get this way. For example, using 𝑎 = 1 and 𝑟 = −𝑥 gives
1
1+𝑥
= 1 − 𝑥 + 𝑥 2 − 𝑥 3 + ⋯,
And using 𝑎 = 𝑥 and 𝑟 = 2𝑥 gives
𝑥
= 𝑥 + 2𝑥 2 + 4𝑥 3 + 8𝑥 4 + ⋯
1 − 2𝑥
Example1: Find a formula for the sum of the following series:
𝑥 2 + 2𝑥 5 + 4𝑥 8 + 8𝑥 11 + 16𝑥 11 + ⋯
Solution: This is a geometric series with a common ratio of 2𝑥 3 . The first term is 𝑥 2 , so
𝑥2
𝑥 2 + 2𝑥 5 + 4𝑥 8 + 8𝑥 11 + 16𝑥 11 + ⋯ =
1 − 2𝑥 3
2.5 Differentiation and Integration of power series
Let the function 𝑓 be defined by the power series ∑∞ 𝑛
𝑛=0 𝑎𝑛 (𝑥 − 𝑐) on its interval of convergence I .
1. 𝑓 is a continuous function on I ;
2. The power series may be differentiated term by term and the resulting power series converges to
𝑓 ′ (𝑥)at all points in the interior of I .
3. Similarly, the power series may be integrated term by term and the resulting power series
converges to ∫ 𝑓(𝑥)𝑑𝑥 + 𝑐 at all points in the interior of I , where C is an arbitrary constant.
Note: The above result says nothing about the convergence of the differentiated (or integrated) series at
the endpoints of the interval of convergence.


 a x  c  has radius of convergence R  0, then the function f

n
If the power series n
n 0

defined by f x    a n x  c 
n
is differentiable on the interval c  R , c  R and its
n 0

derivative and integral are as given below:


f ' x    n a n x  c 
n 1
i. (Using term by term differentiation of the series
n 1

a0  a1 x  c   a2 x  c     ak x  c      ak x  c  )
2 k k

k 0

a x  c 3

 f xdx  C  a x  c  2 x  c  a2 
2
ii. 0
1

  a x  c dx    a x  c dx  C   n  1 a ( x  c)
 
1 n 1
where C is the
n n
i.e., n n n
n 0 n 0

constant of integration.

8
 
1
Example 1: we know that the series x
n 0
n
converges to
1 x
for x  1 .

 R  1 and it is differentiable on the interval  1,1 .


i.e., f x  
1
  x n , x  1 ................. .......... .......... .......... .......... .......... .(1)
1  x n 0
Differentiating both sides of (1) with respect to the variable x as

  1  x  x    gives:
d  1  d

2

dx  1  x  dx


1
1  x 2
 1  2 x  3 x 2
 4 x 3
   
n 1
n x n1

Integrating both sides of (1) yields:

x2 x3
 n 1  x  C  x   
2 3

xn
  n 1  x  C  
n 1 n
, for x  1 .

To get the value of the constant C , we can put in any number in the interval  1,1 .
For instance, if we insert x  0 , we get:  n 1  0  c  c  0.

xn
 n (1  x)   , for x  1 .......... ..........
which .......... .......... .......( 2)
is the
n 1 n

This is the power series representation of the function f ( x)  n(1  x) for x  1 .

Similarly, if we insert x  1 in equation ( 2) , we obtain the power series representation

2

1
for n 2 to be n2   n2
n 1
n

Again if we put  x in terms of x in ( 2) , we get :

n 1  x   

 1n x n  x
x2

x3
 .....
n 1 n , wh 2 3

which is the power series representation for f ( x)  n 1  x .

1
Exercise: Find the power series representation of for x  1 .
1 x2
Example 2: Find the power series representation of
1
i. 𝑓(𝑥) = 2+𝑥
a. Centered at the origin.

9
 b. Centered at 𝑐 = 1.
ii. 𝑔(𝑥) = tan−1 𝑥
Solution:
1
i. Given 𝑓(𝑥) = 2+𝑥
1 1 1 1 1 𝑥 𝑛
a. We have 2+𝑥
= 2−(−𝑥) = 2 ( 𝑥 ) = 2 ∑∞
𝑛=0 (− 2 )
1−(− )
2
∞
(−1)𝑛 𝑥 𝑛 1 𝑥 𝑥 2 𝑥 3
=∑ = − + − +⋯
2𝑛+1 2 4 8 16
𝑛=0
Using the ratio test one can see that the interval of convergence of this power series is (−2, 2).
1 1 1 1 1 𝑥−1 𝑛 (−1)𝑛 (𝑥−1)𝑛
b. 2+𝑥
= 3+(𝑥−1) = 3 ( 𝑥−1
) = 3 ∑∞
𝑛=0 (− 3
) = ∑∞
𝑛=0 3𝑛+1
1−(−( ))
3

Again, using the Ratio Test, we see that the radius of convergence is 3, and the interval of convergence is
(−2, 4). Note that although the power series in (a) and (b) are different, and have different intervals of
convergence, the point 𝑥 = −2 is the end point for both of them. This is because the function 𝑓(𝑥) is not
defined at this point.
x 1 1
ii. In tan 1 x   0 1 t 2
dt and
1 x
 1  x  x 2   for x  1, replacing x by

 t 2 , we get

 1  t  t  t  t      1 t 2 n for t 2  1.......... .......... .......... .....( 3)
1 2 4 6 8 n

1 t 2
n 0

Integrating both sides of (3) yields  2

x 1
0 t 1
dt     1 t 2 n dt
0
x n
 
 
x 2 n1
 tan 1 x    1  t 2 n dt    1
n x n

n 0
0
n 0 2n  1
 2 n 1
   1
x
 tan 1 x, x  1.
n

n 0 2n  1

x 2 n 1
  1 1
converges to tan x for x  1 or the
n
This implies that the power series
n 0 2n  1

x 2 n 1
1
power series expansion of f ( x)  tan x is   1  tan 1 x, x  1.
n

n 0 2n  1
Exercise:

x 2 n 1
Verify that the power series   1 also converge for x  1
n
1.
n 0 2n  1
 
   1
n 1
and as a result .
4 n 0 2n  1
2. Find the power series representation of 𝑓(𝑥) = 𝑥 3 tan−1 (𝑥 2 )

2.5. Taylor and Maclaurin Seriesseries for a function

10
let 𝑓 has derivatives of n – orders at c, and suppose f can be represented by the polynomial,
n
f x   n x    a k x  c   𝑎0 + 𝑎1 (𝑥 − 𝑐) + 𝑎2 + (𝑥 − 𝑐)2 + ⋯ + 𝑎𝑛 + (𝑥 − 𝑐)𝑛 … … … … … (4)
k

k 0

To determine the coefficients in (4) we proceed as follows:

Let x  c in ( 4)
Then f c   a 0 .
f ' x   a1  2a 2 x  c   3a3 x  c    , x  c  R
2

f "  x   2a 2  3 2a 3  x  c   
f ' " x   3.2 a3  4.3.2 a 4 x  c   
 
n 
f (c )  n ! a n
 f c   a 0 , f ' c   a1 , f " c   2 a 2
f ' " c   3.2 a3 ,  , f n  (c)  n!a n
This implies that the coefficients a k ' s are:
a0  f c 
a1  f ' c 
f " c 
a2 
2!
f ' " c 
a3 
3!

n 
f ck  f c 
ak   an 
k! n!
Theorem2.2: If a function f has a polynomial approximation at x  c,
n
f x    ak x  c  , for x  c  R, then the coefficients a k' s are given by the formula
k

k 0

f  k  (c )
ak  .
k!
Then putting this formula of a k in to ( 4) , we obtain the following definition for
nth – Taylor polynomial.
Definition 2.4:
If 𝑓 can be differentiable n – times at x  c, then we define the nth – Taylor polynomial for
the function f about x  c to be:

11
 f ck   x  c 
n k

n  x    .
k 0 k!

Definition 2.5:
Suppose the function f has derivatives of all orders on an interval containing the point 𝑐. The Taylor
series for 𝑓 center at 𝑐 is
∞
′ (𝑐)(𝑥
𝑓 ′′ (𝑐)(𝑥 − 𝑐)2 𝑓 ′′′ (𝑐)(𝑥 − 𝑐)3 𝑓 𝑘 (𝑐)(𝑥 − 𝑐)𝑘
𝑓(𝑥) = 𝑓(𝑐) + 𝑓 − 𝑐) + + +⋯= ∑
2! 3! 𝑘!
𝑘=0
A Taylor series centered at 0 is called a Maclaurin series.

Furthermore, if f has derivatives of n – orders, the nth – Maclaurin polynomial for f x  is

f " (0) f '" (0) f 0n  n f k 
(0)
given by: f  x   f 0   f ' 0 x  x 
2
x  
3
x  n
xk
2! 3! n! k 0 k!
Examples:
1. Find the third Taylor polynomial approximation for f x   n x about x  1.
2. Find the Maclaurin series of
i. 𝑓(𝑥) = sin 𝑥 ii. 𝑓(𝑥) = 𝑒 𝑥

3. Find the Taylor series representation for the function g x   cos x about x   .
Solutions:
n f ck 
1. From n x    x  c k , In this case, we have f x   nx, c  1, and n  3 .
k 0 k!
Here, f x   nx is differentiable around x  1,

and f '  x    f ' c   f ' 1  1 .

1
x
1
f " x    f " 1  1.
x2
2
f ' "  x   3  f ' " 1  2, & f 1  n1  0
x
Then the third Taylor polynomial for f x   nx is:
f k  (1)x  1
k
3
3 x   
k 0 k!

12
 f 10  f 11 f "1 f 1'"
 x  10  x  1' x  12  x  13
0! 1! 2! 3!
f "
f '"
x  1
3

 f 1  f '  x  1  x  12 

  1
2; 6
1
 n1  1x  1  x  12  2 x  13
2 6

 x  1  x  12  1 x  13
3  1
 3
2 3
2. i. f x   sin x  f 0  sin0  0
f ' x   cos x  f ' 0  cos 0  1
f "x    sin x  f "0   sin 0  0
f '" x    cos x  f ' ' ' 0   cos 0  1.
f iv  ( x)  sin x  ff 0iv   0
f v  ( x)  cos x  ff 0v  1
 
Then the Maclaurin series for f x   sin x is
 f 0k  x k
f x   sin x  
k 0 k!
0.x 0 1.x1 0.x 2  1x 3 0.x 4 1.x 5
      
0! 1: 2! 3! 4! 5!
x3 x5 x7
 sin x  x   
3! 5! 7 !
Using sigma notation, the Maclaurin series for f x   sin x is:

x 2 k 1  1
 k
sin x  
k  0 2k  1!

ii. f x   e  f 0  e  1.
x 0

f ' x   e x  f ' 0  e 0  1


f k  (0)  1  k.
 f 0k  x2 x3 x4
So, e 
x

k 0 k!
xk  1 x   
2! 3! 4!


13
  
xk xk
 ex  , i.e., the Maclaurin series for e x is e x  
k 0 k ! k 0 k !

3. g x   cos x  g 0   g    cos  1

g ' x    sin x  g '    0
g ' ' x    cos x  g ' '    1
g ' ' ' x   sin x  g ' ' '    0
g iv  ( x)  cos x  g iv  ( )  1

Then its Taylor series representation at x   is:
g k  ( ) x   
 k

g x   cos x  
k 0 K!
g ' '   g ' ' '  x    g lv x   
3 4

 g    g '  x     x      

2

2! 3! 4!

 1  x     x     x     
1 2 1 4 1 6

2! 4! 6!

 1k 1 x   2k
Using sigma notation, it can be written as cos x   (Taylor series
k 0 2k !
representation for cos x about x   ).

The Binomial Series (Reading assignment)

Definition: If m is a real number, then the Maclaurin series for 1  x 
m
is called the
binomial series, which is given by:
m m  1 2 m m  1m  2 3 m m  1m  2m  3 4 
 m
1  mx  x  x  x      x n  1  x 
m

2! 3! 4! n 0  n 
𝑚 𝑚(𝑚−1)(𝑚−2)…..(𝑚−𝑛+1)
where ( 𝑛 ) = 𝑛!
,which is the coefficient of 𝑥 𝑛 in the expansion.

If m is a non – negative integer, then , f x   1  x 

m
 is clearly a polynomial of
degree m, and hence,

 f m1 (0)  f m 2  (0)    f k  (0)  0 for all k  m, and the binomial series reduces
m m  1 2
to 1  x   1  mx  x    x m , which is valid for    x   .
m

2!
Examples:
1. Find the binomial series expansion for 1  x 
5

2. Find the coefficient of x 4 in the binomial expansion of 1  x  .

6

14
Solutions
20 x 2 60 3 120 4
1. 1  x   1  5x   x  x  x 5 sin ce m  5
5

2! 3! 4!

Thus,
1  x 5  1  5 x  10x 2  10x 3  5 x 4  x 5
2. Coefficient of x 4 is: m  6
m m  1 m  2 m  3
4!
6 6  1 6  2 6  3 6! 6.5.4!
    15
4; 4 : 2! 4 : 2!

15