REAL ANALYSIS

For

Department of Mathematics,
Faculty of Science, Benha University

Prepared by

Dr. Ali Hafez,

Dr. Ismail Ibedou,
Dr. Mostafa Hasan,
Dr. Ahmed Mohamed

2021 – 2022

1
Contents:

1- Chapter (1) The real numbers …..…......….....…… (3)

2- Chapter (2) Topology on the real line …..……..… (17)

3- Chapter (3) Infinite sequences ………..….....…… (25)

4- Chapter (4) Infinite series ..…....……...….....…… (40)

5- Chapter (5) Continuous function ……………...… (56)

6- Chapter (6) Derivatives ……………….………… (68)

7- Chapter (7) The Riemann Integral ……………..… (85)

2
 CHAPTER (1)

THE REAL NUMBERS

Definition 1.1: (Ordered set)

Let  be a non- empty set. Then X is called an ordered set if the following conditions
are satisfied
(O 1) x , y  X  x  y or y  x
(O 2) x , y  X , x  y and y  x  x  y
(O 3) x , y , z  X , x  y and y  z  x  z
Then “  ” is called the order relation on X .
Remark 1.1: x  y can also be written as y  x .

Remark 1.2: x  y , x  y is written as x  y .

The Natural Numbers:

Definition 1.2: The set   1, 2,3,... is called the set of Natural Numbers (or the set of
positive integers)

Properties: (Peano Postulates)

(P 1)  is an ordered set. Also n    n  1   .
(P 2) Every non-empty subset of  has the smallest number.
( P 3) All the numbers in  , except 1, have the immediate predecessors.
(P 4) There is no largest number in  .

Example 1.1: For X  2,3, 4 , then “2” is the smallest number.

Example 1.2: For X  4,6,8,... , then “4” is the smallest number.

Example 1.3:   1, 2,3, 4,5,... , then the immediate predecessor of 4 is 3.

Remark 1.3: A non-empty subset of  which has the largest number is called a finite
set. Otherwise it is called infinite set.

3
The Rational Numbers:
p
 

Definition 1.3:    : p  , q  , q  0, 
p and q are prime mutually
 , is the set of

 q 

 gcd ( p ,q )1 
all rational numbers.

Remark 1.4:  is an ordered set.

Question 1.1: Are there any rational number which satisfy equations of the form
x 2  3  0, x 2  17  0, x 3  6  0 etc. (Are 3, 17, 3 6 etc. rational
numbers?)
We prove that they are not rational numbers by using the following theorem:

Theorem 1.1: Consider the equation an x n  an 1x n 1    a1x  a0  0 where n is a

positive integer and n  1, an , an 1 ,, a0 are integers and an  0, a0  0 . Let a rational
p
number satisfy the equation. Then q divides an and p divides a0 .
q
Proof: We have an x n  an 1x n 1    a1x  a0  0 _______________ (1)
p
Put x  .
q
p p p
an ( ) n  an 1 ( ) n 1    a1 ( )  a0  0
q q q
n n 1
p p p
an n  an 1 n 1    a1  a0  0
q q q
Multiply by q n
an p n  an 1 p n 1q    a1 pq n 1  a0q n  0 ___________ (2)
Write (2) as
an p n  [an 1 p n 1q    a1 pq n 1  a0q n ]
an p n  q [an 1 p n 1    a1 pq n 2  a0q n 1 ]
an p n
  [an 1 p n 1    a1 pq n 2  a0q n 1 ] ____________ (3)
q
Since R.H.S is an integer, then L.H.S is also an integer  q divides an p n
Since p and q are mutually prime, then q does not divide p n  q divide an .
Next, write (2) as
a0q n  [an p n  an 1 p n 1q    a1 pq n 1 ]

4
 a0q n   p [an p n 1  an 1 p n 2q    a1q n 1 ]
a qn
 0  [an 1 p n 1    a1 pq n 2  a1q n 1 ]
p
Since R.H.S is an integer, then L.H.S is also an integer  p divides anq n .
But p cannot divide q n ( p ,q mutually prime)
 p divides a0 .

Problems:
1. Show that 2 is not a rational number.
Solution:
Let x  2
 x 2
 2
x 2
2
 0
1 x  0  x  2  0
2

(a 2 x  ax  a0  0)
 a2  1, a0  2
p
Suppose is a rational number satisfying (1). Then (by Theorem 2.1)
q
q divides a2 

and 
p divides a0 
q divides 1  
 q  1

    
p divides  2 
 
 p  1,  2

the only possible rational numbers satisfying (1) =
p  1 2
 ,  1,  2
q 1 1
But if we take x  1 in (1), we get
x 2 2 0
(1) 2  2  0
1 2  0
1  0,
which is not possible (absurd)
Similarly if we take x  2 in (1), we get

5
 x 2 2  0
(2) 2  2  0
42  0
2  0,
which is absurd
Hence 2 is not a rational number.

2. Show that (2  2)1/ 2 is not a rational number.

Solution: -
1

Let x  (2  2) 2
 x 2 2 2
 x 2 2 2
 (x 2  2) 2  2
(1)
x 4  4x 2  4  2
__________________

(1 x 4  0  x 3  4  x 2  0  x  2  0)
 a4  1, a0  2
Suppose p/q is a rational number satisfying (1)
q divids a4 
Then 
p divids a0 
q divides 1  
 
 q  1
   
p divides 2  
  p  1,  2.

Therefore, the only possible rational numbers satisfying (1)
p 1 2
=  ,  1,  2.
q 1 1
But if we take x   1 in (1) we get
x 2  4x 2  2  0
(1) 4  4(1)2  2  0
1 4  2  0
1  0,
which is absurd.
Similarly if we take x  2 in (1), we get

6
 x 2  4x 2  2  0
(2) 4  4(2) 2  2  0
16 16  2  0
2  0,
which is absurd.
1

Therefore (2  2) is not a rational number.

2

3. Show that there is a rational number between any two distinct rational numbers.
a b
Solution: Let a , b  . Then is a rational number betwen a and b .
2
Homework:
1. Show that there is no rational number, which satisfies x 3  6  0 .
1
1/ 3 2
2. Show that (2  5 ) is not a rational number.
[Note: [(a  b )3  a 3  3a 3b  3ab 2  b ]

Least Upper Bound and Greatest Lower Bound

Definition 1.4: (Upper Bound)

Let F be a set, which is an ordered set. Let E  F . An element a  F is called an upper
bound of E if x  a x  E

Definition 1. 5: (Least Upper Bound)

Let E be a subset of an ordered set F . An element a  F is called least upper bound
of E if
(i ) x  a x  E
(ii ) if x  b x  E , then a  b

Definition 1.6: (Lower Bound)

Let E be a subset of an ordered set F . An element a  F is called a lower bound of E
if x  a x  E

Definition 1.7: (Greatest Lower Bound)

Let E be a subset of an ordered set F . An element a  F is called greatest upper bound
of E if

7
 (i ) x  a x  E
(ii ) if x  b x  E , then a  b
____________________________________________________________

Always Recall
a  least upour bound of E
(i) x  a x  E
(ii) x  b x  E  a  b
a  gratest lower bound of E
(i) x  a x  E
(ii) x  b x  E  a  b
____________________________________________________________
Notations:
(i) Least upper bound of E is denoted by lub(E ) . It is also called the supremum of E
denoted by sup(E ) . Note: lub(E )  sup(E )
(ii) Greatest lower bound of E is denoted by glb(E ) . It is also called the infimum of
E denoted by inf(E ) . Note: glb(E )  inf(E ) .
Definition 1.8: If a set E has an upper bound, then we say E is bounded above . If a
set E has a lower bound, then we say E is bounded below.
Remark 1.5: A set E is said to be bounded if it has an upper bound and a lower bound.

Problems
Find the supremum and infimum of these sets:-
1. E  {0,1, 2,3,...}
sup(E ) does not exist.
inf (E )=0 (which belongs to E )
1 1 1
2. E  { : n  }  {1, , ,...}
n 2 3
sup(E )  1 (which belongs to E )
inf(E )  0 (which dose not belong to E )

8
 n 1 2 3 4
3. E { : n  }  { , , , ,...}
n 1 2 3 4 5
sup(E )  1 (which dose not belong to E )
1
inf(E )  (which belongs to E )
2
Exercises
1.If E is a bounded (non-empty) set prove that inf(E )  sup(E ) .
2.If S is a bounded set, prove that inf(S )   sup(S ) .
3.If S and T are bounded sets and S  T , prove that
(i) sup(S )  sup(T )
(ii ) inf(S )  inf(T ).
[Note: Use definition of sup and inf]
4. If x   such that 0  x  n1 n . Prove that x  0
5. Prove that there is an irrational number between any two distinct real numbers
Solution (3) (i)
Let a  sup(T )
 t  a t T
 s  a s  S (  S  T )
 sup(S )  a
 sup(S )  sup(T )

Maximum and Minimum:

(i) If sup(E )  E , then sup(E ) is called the maximum of E , written as max(E )

.
(ii) If inf(E )  E , then inf(E ) is called the minimum of E , written as min(E ) .

Theorem 1.2: In the set  of rational numbers, not every subset of  which has upper
bound has the least upper bound.
Proof: Consider the set A  {r   : 0  r  2}
A has upper bounds (for example 5, 3 are upper bounds but the least upper bound
of A is 2 which is not a rational number.

9
The Real Numbers:

Definition 1.9: The set of real numbers, denoted by  , consists of the rational numbers
and those numbers such that every non-empty subset of  , which has upper bounds
has the least upper bound.

Remark 1.6: In  , the numbers, which are not rational numbers, are called irrational
numbers.
Remark 1.7: The elements of  satisfy the following order relations:

(O 1) a,b    a  b or b  a.
(O 2) a,b  , a  b and b  a  a  b .
(O 3) a,b ,c   , a  b and b  c  a  c
(O 4) a  b  a  c  b  c .
(O 5) a  b and 0  c  ca  cb .

Absolute Value of a  
The absolute value of a , denoted by a , is defined by
a if a  0
a  max{a, a}  
a if a  0

Properties:
(i ) a  0
(ii )  a  a  a
2
(iii) a  a 2
(iv) ab  a b
( v) a  b  a  b (Traingle inqulity)
( vi ) a  b  a  b

10
Proof.
( vi ) Consider a  (a  b )  b
by(v)  a b  b
 a  b  a  b ______(1)
Interchange a and b in (1)
b  a  b  a  (a  b )
= 1 a  b
 a b
b  a  a b
Multiply by (  1)
a  b   a  b ______(2)
From (1) and (2)
 a b  a  b  a b
 a  b  a b (by (ii))

The Completion Axiom:

In  , every non-empty subset S , which has upper bounds, has the least upper bound.
Remark 1.8: It follows from the Completion Axiom that, in  every set, which has
lower bonds, has the axiom that, in  every set, which has lower bounds, has the
greatest lower bound [inf(E )   sup(E ) ]

Homework: Show that, in  , every non-empty subset E , which has lower bounds,
has the greatest lower bound.
[Hint: Let F  { y   : y  x x  E }  { y   : y is a lower bound of E }.
If a  sup(F ), show that a is the glb of E , by using defination]

Problems:
1. Find the sup,inf,max and min of the sets
A  {x    : 0  x 2 1  2},
B  {x    :1  x 2  3}  {x    :1  x  3}
where   is the positive real numbers.

11
Solution : B  {x    :1  x 2  3}  {x    :1  x  3}
sup(B )  3
inf(B )  1
max(B )  3
min(B ) does not exsist.

2. If a  , and b is an irrational number. Show that:

(i) a  b and (ii) ab (a  0) are irrational numbers.
Solution:
(i) Let x  a  b
To prove: x is irrational number.
Suppose x is a rational.
Now: x  a  b  x  a  b
But x  a is a rational number and b is an irrational number,
which is absurd.
(ii) Let x  a  b
To prove: x is irrational number.
Suppose x is a rational.
x
Now x  ab   b
a
x
But is rational number and b is irrational number, which is absurd
a

The Archimedean Principle:

Let x   , x  0 [ x    ] . Then, there is a number n   such that x  n .

Proof: Suppose x  n
Then x  n n  .
or n  x n   .
 x is an upper bound for .
 lub() exists by the Completion Axiom.
Let a  lub().
 n  a n   (by Definition 1.5)
But we know that n    n 1  .

12
 n  1  a
 n  a 1
 n  a 1  a n  
This is not possible because a least upper bound of 
a 1  n n    a 1 a . This is absurd.
________________________________________________________

Problems
1
1. If   0, show that there is a number n   such that n.

Solution:
1
 0  0

by the Archimedean Principle, there is a number n   such that
1
n.

1
2. If a  0, prove that there exists a number n   such that a  n .
n
Solution: a  0 . Then by the Archimedean Principle, there is an integer m  
such that a  m _______(1)
1
Also,  0 . Then by the Archimedean Principle, there is an integer k   such that
a
1
 k _________(2)
a
Let n  max{m , k }
Then by (1) and (2), we get
1
a  n and n
a
1
or a  n and a
n
1
or  a and a  n
n
1
 a  n
n
1
3. Let a , b   . If a  b  n   , show that a  b .
n

13
 1
Solution: Given a  b  n  
n
To prove a  b . Suppose a  b
i.e. a  b
 a b  0
1
 0
a b
by the Archimedean Principle, there is an integer n   such
1
such that n
a b
1
  a b
n
1
 b  a
n
1 1
or a  b  which contradicts the given: a  b  n  
n n
1
3. If x   such that 0  x  n   . Prove that x  0 .
n
Solution: Suppose x  0
1
Then 0  x  _________(1)
n
Consider x  0
1
 0
x
by Archimedean Principle there is a number n   such that
1
 n
x
1
 x  which contradicts (1)
n

Theorem 1.3: There is a rational number between any two distinct real numbers.

Solution: Let x , y  , x  y .
p p
We want to prove that there is a rational number such that x   y .
q q
Now, x  y  y  x  0

14
 1
or 0
y x
By the Archimedean principle there is a positive integer q   such that
1
q
y x
 1 q ( y  x )
 qx  1  qy _______(1)
We can find an integer p such that p  qx  1  p  1
 p  qx  1 and qx  1  p  1
i.e., p  qx  1 and qx  p
or qx  p and p  qx  1
 qx  p  qx 1
or, qx  p  qy [by(1)]
p
 x  y .
q

Problems:
1. Show that there is an irrational number between two distinct real numbers.
Solution: let x , y  , x  y
 2x  2 y
(Note that 2x and 2y are also real numbers.)
Hence (by Theorem 3), there is a rational number r
such that 2x  r  2 y _____(1) . Let r  0
2
 2x  r  2y
2
r 2
x  y (r  0)
2
2r
 is an irrational number between the real numbers x and y .
2
Let r  0 then from (1) we have
2x  0  2 y _____(2)

Then, (by Theorem 1.3) there is a rational number r  such that

15
 2x  0  r   2 y
2x  r   2 y
r
x  y
2
2
x  r  y
2
2
 r  is an irrational number between the real number x and y .
2
2. Show that sup{r   : r  a}  a for each a   .
Solution: Homework.

16
 CHAPTER (2)

TOPOLOGY ON THE REAL LINE

The set of real numbers is represented by points on a line called the real line.
Intervals:
Let a,b  , a  b . Then the various intervals are written as follow:
Open interval= (a,b )  {x   : a  x  b }
Closed interval= [a,b ]  {x   : a  x  b }
Closed open= [a,b )  {x   : a  x  b }
Open closed= (a,b ]  {x   : a  x  b }
The Nested Interval Property:

Let
I 1  [a 1,b 1 ]
I 2  [a 2 ,b 2 ]
I 3 
 

such that I 1  I 2  I 3 

Then, there exists at least one point common to every interval i.e.  I n  
1
Notation: If a point p belongs to an open interval, then that
interval is denoted by S p .
Definition 2.1: let A be a set of real numbers. A point p  A is called an
interior point of A if there is an open set S p  A .
A is called an open set if every point of A is an interior point i.e., p  A there is an
open interval S p such that S p  A .
Example 2.1: A  (a,b )
Let p  A
Take S p  (a ,b )  A .

Example 2.2: A   . Take S p  

  is open set.

17
Example 2.3: A  
Every point in A is an interior point  There is no point in A
which is not an interior point.
Here, in  , there is no point, which is not an interior point.
 is open set.
Example 2.4: A  [a,b )
Let p  a
Here S p  A
 A is not open set.
Theorem 2.1: (i) The union of any family of open sets is open.
(ii) The intersection of a finite number of open sets is open.
Proof: (i) Let   {G } be a family of open sets.
Let H   {G }
G 

To prove: H is open set:

Let p  H   {G }
G 

There is an open set G 0   such that p G 0 . But G 0 is open

 S p  G 0   {G }
G 

Sp   {G }
G 
or S p  H  H is open set.
(ii) To prove A  B is also an open set
Let p  A  B
 p  A and p  B
Since A and B are open sets, S p  A and S p  B
 p  S p  S p  A  B
Clearly S p  S p is an open interval
containing p . Also we have
S p  S p  A  B  (by Def. 2.1) A  B
is an open set.
Remark 2.1: In (ii), we have to take finite number of open sets, because of the
following example.

18
Example 2.5: Consider I 1  (1,1)
1 1
I 2  ( , )
2 2
1 1
I 3  ( , )
3 3

1 1
I n  ( , ), n  
n n

1 1
We prove that  (  , )  0
n 1 n n
 
1 1
Let x   I n   ( , )
n 1 n 1 n n
1 1
 x  ( , ), n  
n n
1 1
   x  , n  
n n

1 1
   x and x 
n n
1 1
 0  x  and x  0 
n n
 0  x and x  0
x 0
Definition 2.1: (Topology on  )
Let u be the family of all open sets in  . Then u satisfies the conditions (i) and
(ii) of the previous theorem. u is called the usual topology on  .
Accumulation Points (or, Limit Points)
Definition 2.2: Let A   . A point p   is called an accumulation point (or,
limit point) of A if every open set containing p contains a point of A different from
p (or, If G is an open set containing p , then A  (G  { p })   ) [i.e.G contain at
least point of A other than p ]

Remark 2.2: The set of all accumulation points of A is denoted by A  which is

called the derive set.
1 1
Example 2.6: A  {1, , ,...}
2 3

19
 0 is the only limit point of A ( A  {S 0  {0}}   )
Example 2.7: Find the accumulation points of
(i)  , the set of integer numbers.
(ii)  , the set of rational number.
Solution: (i)  does not have limit points.
(ii) Let p   , let G  (a,b ) containing p . Between a and p
(or between p and b ) there is a rational number (by Th. 1.3)
 p is a limit point of  .
all real numbers are limit point of  .

Bolzano-Weierstrass Theorem:
Let A be a bounded infinite set of real numbers. Then A has an accumulation point
(limit point) [i.e. every bounded infinite set of real numbers has an accumulation
point)
Proof: Since A is bounded, a1  x  b1 x  A
 x  [a1 ,b1 ] x  A
 A  [a1 ,b1 ]
a1  b1
Let I 1  [a1,b1 ] . Bisect I 1 at the point
2
a1  b1 a b
 A  [a1 , ]  [ 1 1 ,b1 ]
2 2
Since A is infinite, one of the sub-intervals contains infinite
number of elements of A . Let that interval be denoted by I 2 ,we
write I 2  [a2 ,b2 ] .
We repeat the previous method to I 2 . Thus, we get a sequence
I 1, I 2 ,of closed interval such that I 1  I 2  I 3 

By the nested interval property we have I
n 1
n   , I n  [an ,b n ]

Let p   I n . Let S p  (a ,b ) { p  (a,b ) }
n 1

Let I n  length of I n = b n  an  0 (as n   )

i.e. lim I n  0  I n   for n  n 0 (n 0 some big number)
n 

In particular I n0   .
We choose   min( p  a,b  p )

20
  I n0  min( p  a,b  p )  I n0  S p  (a,b )

But I n0 contains infinite number of elements of A .

 S p also contains infinite number of elements of A .
 (by Def. 2.2) p is an accumulation point of A .

Closed Sets
Definition 2.3: Let A   , then A is called a closed set if A c is an open set.
Example 2.8: Let A  [a,b ], a,b   . This is a closed set.
Reason: [a , b ]c  ( , a )  (b ,  ) which is an open set.

Example 2.9: A  
c   which is an open set.
  is a closed set.

Example 2.10 : A  
 A c  c   which is an open set.
 is a closed set.

Remark 2.3:  and  are both closed and open sets.

Theorem 2.2: Let A   .

A is a closed set  A contains all its accumulation points.
Proof:
To prove () : Given: A is a closed set.
i.e. A c is an open set.
i.e. every point of A c is an interior point of A c .
We have to prove that A contains all its accumulation points.
Let p be an accumulation point of A .
(by Def.2.2) every S p contains a point of A other than p
S p  Ac
 (by Def. 2.1) p is not an interior point of A c .
But every point of A c is an interior
 p  Ac
 p A
To prove (  ): A is closed, i.e. A c is open.

21
 Let p  A c .
We have to prove that S p  A c
p is not an accumulation point of A
( A has all its accumulation points)
S p  Ac

Homework:
1.Prove that the union of a finite number of closed sets is closed
2.Prove that the intersection of any family of closed sets is closed.
Solution: 1.Let A1 , A 2 , , A n be a finite number of closed sets .
i.e. A1c , A 2c ,, A 3c are open sets.
n
  Aic is an open set.
n 1
n
(A )c  A c
i.e. ( Ai )c is an open set (A )c  A c
i 1
n
(by Def. 2.3) A
n 1
i is a closed set

2.Let   {B } be any family of closed sets.

Let H   {B }
B 

 H  (  {B })c
c

B 

 {B } which is open set


B 
c

H   {B } is a closed set
B 

Always Remember:
The neatest interval property:

If I 1  [a1,b1 ], I 2  [a2 ,b2 ],and I 1  I 2  I 3   , then  I n  
1

Thm. (i) The union of any family of open sets is open

(ii) The intersection of a finite number of open sets is open
A is closed  A c is open

22
Theorem 2.3. (Heine-Borel Theorem)
Let I 1  [c1 , d 1 ] be a closed and bounded interval and G  (a ,b ) :   J  a family of
open interval that covers I 1 , i.e. I 1   (a ,b ) , then G has a finite sub-family
 J
n
(a1 ,b1 ),(a2 ,b 2 ),,(an ,b n ) that covers I 1 , i.e., I 1  (ai ,bi ) .
i 1
Proof: [We proof by contradiction]
n
Suppose I 1 not, i.e., I 1  (ai ,bi ) ,i.e., I 1 is not covered by a finite sub-family
i 1
of G .
c1  d1 c d c d
Bisect I 1 at the point . Let I 2  [c1, 1 1 ] or [ 1 1 , d 1 ] whichever is not
2 2 2
covered by a finite sub-family of G .We write I 2  [c 2 , d 2 ] .
c  d2
Bisect I 2 at the point 2 and proceed as above, we get a sequence {I n } of
2
intervals such that I 1  I 2  I 3  
And each I n is not covered by a finite subfamily of G (*)
Also, if I n is the length of I n ,
lim I n  0  (def.) I n  0   for n  n 0
n 
i.e., I n   for n  n 0
In particular, I n0   ____________(1)
(Note that I n0 is not covered by a finite sub-family of G )

By the nested interval property, I
1
n  .

Let p   I n
1
 p  I 1 , p  I 2 , p  I 3 ,  , p  I n 0 , , p  I n , 
Consider p  I 1   {(a ,b )} (given)
 J

 p   {(a ,b )} which is an open set. (This is by Theorem 2.1)

 J

(by definition of open set), S p   {(a ,b )} .

 J
Let S p  (a ,b ), a  p  b

23
Let   min( p  a,b  p ) , then by (1),
I n 0  min( p  a ,b  p )
 I n0  (a ,b ) which is a contradiction to (*)

An Important Result on Open Sets (usual in advanced math)

An open set of real number can be expressed as the countable union of
mutually disjoint open interval.(i.e. if A is an open set of real numbers, then

A   I n , where I 1 , I 2 ,are mutually disjoint open intervals.
n 1

24
 CHAPTER 3

INFINITE SEQUENCES

Definition 3.1: Consider the following function

f :   , n  f ( n )
f (1), f (2), form a sequence.
Notation: Let a1 , a2 , form a sequence, where
a1  f (1)
a2  f (2)
a3  f (3)
  
Then we write it as (an )  (a1 , a2 ,) or as {an }  {a1 , a2 ,...}.

Convergence of a sequence
A sequence (an ) is said to be convergent to a if a given   0 , there is a number n 0  0
such that an  a   for n  n 0 .

Notation: We can write this by an  a as n   or, lim an  a. Here a is called limit

n 
of (an ) .

Theorem 3.1: If an  a as n   , a is unique ( i.e. any convergent sequence has

unique limit)
Proof: Suppose an  a and an  b as n  
 n m
an  a  
2  for n  n n r
i.e.  0
 n 0  max(m , r )
an  b  
2 

We want to proof: a  b
Consider

25
 a  b  (a  an )  (an  b )
x y  x  y
 a  an  an  b
 
  
2 2
 a b  
Since  is arbitrary, we have a  b  0
a  b .

Theorem 3.2: If (an ) converges to a , then an also converges to a

(i.e. If an  a, as n   then an  a )

Proof. [Note from the properties of the absolute value: x  y  x  y ]

Given an  a , (as n  )
i.e. an  a   for n  n0 .
To prove: an  a   for n  n 0 .
Use an  a  an  a
But given that an  a   for n  n0 .
 an  b   for n  n 0

Question: Is the converse true?

The following example shows that the converse need not be true?
Example 3.1: (an )  (1, 1,1, 1,)
( an )  (1,1,1,1,)
 lim an  1
x 

 1 if n odd
But an  
1 if n even
 lim an dose not exsist
x 

( an ) is convergent, but (an ) is not convergent.

(an ) converges  ( an ) converges

But ( an ) converges 
 (an ) converges

26
Bounded Sequence
A sequence, which is bounded above and bounded below, called a bounded sequence.
Or, a sequence (an ) is bounded if there is a number M  0 such that
an  M , n  .

Theorem 3.3: Every convergent sequence is bounded.

Proof: Let an  a, (as n   )

i.e. an  a   for n  n 0
Consider an  (an  a )  a
 an  a  a
 an    a for n  n 0
Put   1
 an  1  a for n  n 0
i.e, an 1  1  a 
0
 
 an0 2  1  a 
 
?
 
 But may be, a1 1  a 
 ? 
 a2 1  a 
 
    
 
Let M  max{ a1 , a2 ,, an0 ,1  a }
 an  M for all n  
 the sequence (an ) is bounded.

Note: We have proved: convergence sequence  bounded sequence

?
Question: bounded sequence  convergence sequence
Example 3.2: (an )  (1, 1,1, 1,) . This is a bounded sequence
 1 if n is odd,
But an  
1 if n is even.
 lim an dose not exist.
x 
 (an ) is not convergent.

27
Exercise:
Let an  a, b n  b (as n   ). Prove that:
(1) an  b n  a  b
(2) an  b n  a  b
(3) anb n  ab
a a
(4) n  , (bn  0,b  0)
bn b
(5) kan  ka , k  
Solution:

an  a  
2  for n  n
(1) Given:  0

bn  b 
2 
(an  b n )  (a  b )  (an  a )  (b n  b )
 an  a  b n  b
 
<  
2 2
 an  b n  a  b

an  a    n  n1
(3) Given:  for n  n 0
b n  b    n  n2
anbn  ab  anbn  anb  anb  ab n 0  max{n1 , n 2 }
 anbn  anb  anb  ab
 an b n  b   b (an  a )
 an bn  b  b an  a
Since an  a , then (an ) is bounded. So there is M such that an  M n   .
 anbn  ab  M   b    (M  b )  c  for n  n 0
(where c  M  b )
 anb n  ab

(4) Given: an  a   for n  n1

b n  b   for n  n 2

28
 an a anb  ab n
Consider:  
bn b bb n
anb  ab n

b bn
(anb  ab )  (ab  abn )

b bn
b (an  a )  a (b  bn )

b bn
b an  a  a b  b n

b bn
an  a a b  bn
 
bn b bn
an  a a b  bn
 
bn b bn
Since b n  b  0 , then (b n ) is bounded. So there is M such that b n  M n  n 3 ,
an a
is defined at least for n  n3 , and lim n exist.
bn n  b
n
Let n 0  max{n1 , n 2 , n 3}.
an a  a 1 a
    (  )  c  for n  n 0
bn b M b M M bM
a a
 n 
bn b
(5) Given: an  a   n  n 0
Consider: kan  ka  k (an  a )
 k an  a
k 
 kan  ka  k  for n  n 0
 kan  ka .

29
 Remember:
{an } bounded  an  M n  
{an } convergent to a (an  a )
 an  a   for n  r
    an  a   for n  r (  x  a   a  x  a )

The Completeness Axiom: every bounded set has sup and inf.

Binomial Theorem:
n n (n  1) 2 n (n  1)(n  2) 3
(1  x )n  1  x x  x  x n
1! 2! 3!

Remark 3.1: We have proved: If {an } is convergent, then{ an } convergent. But if

{ an }is convergent, then {an } need not be convergent. However, we can prove the
following:
Homework:
If an  0 , prove that an  0 .

Remark 3.2: We have proved: convergent sequence  bounded sequence

But the converges need not be true.
Question: Under what condition a bounded sequence is convergent, for this, we
require the following:
Definition 3.2: (i) If a1  a2  a3   (i.e. an  an 1 n   ) then {an } is called an
increasing sequence.
(ii) If a1  a2  a3   (i.e. an 1  an n   ) then {an } is called a decreasing
sequence.
(iii) A sequence {an } is called a monotonic sequence if it is increasing or decreasing.

Theorem 3.4: A bounded monotonic sequence {an } is convergent.

Proof. Let {an } be an increasing sequence (i.e. it is monotonic)
Let E  {an : n  }  {a1 , a2 ,}
an  M an  E
 E is a bounded set ({an } is a bounded sequence)
30
 by the completeness axiom, sup( E ) exists.
Let a  sup(E )
 a  an n   _____(1)
Also a    an for some n   _____(2)
For r  n , we have ar  an ____(3) (  a1  a2  ) i.e. increasing
sequence )
From (2) and (3), a    ar  an
or a    an for n  r
Using (1), a    an  a
or, a    an  a   .
Add (a ) to both sides,
  an  a  
or, an  a   for n  r
 an  a as n  
i.e. {an } is convergent.
1
Theorem 3.5: (i) If p  0 , then lim  0.
n  n p

(ii) If a  1 , then lim a n  0 .

n 
1
(iii) If a  0 , then lim a n  1.
n 

1
Proof. (i) To prove:  0   for n  r
np
1
or p   for n  r
n
1
1 p
Given   0 , we choose a number r  ( )

1
1 p
For n  r , we have n  ( )

1
np 

1
 
np
(ii) If a  0, a n  0  lim a n  0
n 

31
 Let a  0, a  1
n n ( n  1) 2 n ( n  1)( n  2) 3
x     x
n
Binomial Theorem: (1  x ) n  1  x x 
1! 2! 3!

1
Since a  1 ,we write a  (b  0)
1 b
n 1
a 
(1  b ) n
1
Inverting n
 (1  b ) n
a
Use Binomial Theorem to R.H.S,
1 n n (n  1) 2
 n 1 b  b 
a 1! 2! 9  1  6  2
1 n 96
  b
a n 1!
1
Inverting an 
nb
1
or 0  an   0 (as n  )
nb
 lim a n  0
n 
 lim an  0
n 

(iii) Here we consider 0  a  1 and a  1

Case (1): a  1
1
n
Let a  1  X n (X n  0)
To prove: lim X n  0
n 
1
n
Consider a  1  X n
 a  (1  X n ) n
Use Binomial Theorem to R.H.S,
n n(n 1) 2 n(n 1)(n  2) 3
[ If n is +ve integer, (1 x )n 1 x  x  x  x n ]
1! 2! 3!

32
 n n (n  1) 2
a  1  Xn X n    X nn
1! 2!
n
 a 1 X
1!
 a 1  nX n
a 1
 X n
n
a 1
or 0  X n   0 (as n   )
n
 lim X n  0
n 
1
 lim (a 1)  0
n
n 
1
lim a 1  0
n
n 
1
lim a  1n
n 

Case (2): Let 0  a  1

1
 1
a
Use case (1),
1 n1
lim( )  1
n  a

1
 lim 1  1
n 
an
1 1
 1
 1  lim
n 
a n
1
lim a n
n 

Problems:
1
n
(1) Show that lim n  1 .
n 
Solution: Put a  n in the Theorem 3.5, Part (iii).
1
(2) If an  (1  )n , show that the an  is convergent.
n

33
 1 n
Solution: an  (1  )
n
Use Binomial Theorem,
n 1 n (n  1) 1 2
an  1  ( )  ( )    ( 1 )n
1! n 2! n n
n n 1
( )
11 n n 
2!
(1 1 )
 an  1  1  n 
2!
(1  1 )
Clearly an  2  n 
2!
1 1 7 92
an  2     79
2! 3!
1 1
an  2  (   )
2! 3!
an  2  1
an  3
or 0  an  3
 {an } is bounded
1
Next, an  (1  )n
n
Put n  1, a1  (1  1)1  2
1
Put n  2, a2  (1  ) 2  (1.5) 2  2.25
2

{an } is an increasing monotonic sequence  monotonic sequence
 a bounded monotonic sequence is convergent.

Alw ays R ecall:

1 1 n ( n  1) 1 n ( n  1)( n  2) 1 1
x n  (1  ) n  1  n  2
 3
  n
n n 2! n 3! n n
1 1 1 1
 1  1  (1  )  (1  )(1  )   
2 1
(1  1 )(1  2 )  (1  n  1 )
2! n 3! n n n! n n n

34
Cauchy Sequence:

Definition 3.3:{an } is called a Cauchy sequence, if given   0 , there is a number r  0

such that an  am   for n , m  r
Theorem 3.6: Every convergent sequence {an } is a Cauchy sequence.
Proof: Given: an  a as n  

i.e., an  a  for n  r
2

Also we can write: am  a  for m  k
2
Let n 0  max( r , k )

an  a 
 2  for n , m  n
 0

am  a 
2 
To prove: an  am   for n , m  n0 [i.e., (an ) is a Cauchy sequence]
Consider: an  am  (an  a)  (a  am )
 an  a  a  an
 
  for n , m  n 0
2 2
 an  am   for n , m  n0

Theorem 3.7: Every Cauchy Sequence (an ) is bounded.

Proof: Given: an  am   for n , m  r
Put   1, m  r  1
 an  ar 1  1 for n  r
 an  (an  ar 1 )  ar 1
 an  ar 1  ar 1 for n  r
 1  ar 1 for n  r
Let M  max{ a1 , a2 ,, ar ,1  ar 1 }
 an  M for all n   .
 (an ) is a bounded sequence.

35
Subsequences:

Example 3.3: Consider (an )  (a1 , a2 ,)  (1, 2,3,4,)

Then, (2,4,6,) is called a subsequence of (an )

Notation: Let (an )  (a1 , a2 , a3 ,) be a sequence.

Let n1  n 2  n 3   be positive integers
Then, (an1 , an2 , an3 ,) is called a subsequence of (an ) and is denoted by (ank )

Theorem 3.8: A sequence (an ) is convergence  every subsequence (ank )

of it is convergent.
Proof: To prove(  ): (ank ) is convergent. Let an  a as n  

 an  a  for n  r
2
Also, we know that convergent sequence is a Cauchy sequence:

 an  am  for n , m  r
2
Put m  n k

 an  an k  for n ,n k  r
2
 
an  a 
 2  for n , n  r
 k

an  ank 
2 
Consider ank  a  (ank  an )  (an  a )
 ank  an  an  a
 
 

2 2
 ank  a   for n k  r
ank  a as n k  
To prove (  ): (an ) is convergent.
Given: (ank ) is convergent.
Take an  ank  (an ) is convergent

36
 ( a n )  ( a1 , a 2 , a 3 , a 4 , a 5 , a 6 , a 7 , a 8 , a 9 , a1 0 , a1 1 , a1 2 , a13 ,  )
 (1, 3, 5, 7, 9,11,13,15,17 ,19, 2 1, 2 3, 25,  )
n  1, 2, 3, 4, 5, 6, 7, 8, 9,10,11,12,13, 
( a n k )  ( a1 , a 3 , a 5 , a 7 , a 9 , a11 , a1 3 ,  )  (b k )  (1, 5, 7,11,15,19, 23,  )
n k  1, 3, 5, 7, 9, 11,13, 
k  1, 2, 3, 4, 5, 6, 7, 8, 

limsup and liminf

Definition 3.4: Let (an ) be bounded sequence.

(i) limsup an  inf k (sup n k an )
(ii) liminf an  sup k (inf n k an )

Remark 3.3: We have sup n k an  sup n k 1 an  (sup n k an )

k 1 is a decreasing

sequence.
 another form of definition is as follows:
(i) limsup an  lim k  (sup n k an )
Similarly,
(ii) lim inf an  lim k  (inf n k an )

RIV: (i) lim an  limsup an  inf (sup an )  limsup an  lim(sup an )

k n k k  n  k

(ii) lim an  liminf an  sup(inf an )  liminf an  lim(inf an )

k n k k  n  k

Remark 3.4: Always liminf an  limsup an

Remark 3.5: (i) limsup an  sup n k an

(ii) liminf an  inf n k an

Theorem 3.9: Let (an ) be a bounded sequence.

(i) If lim n  an  a , then limsup an  liminf an .
(ii) If limsup an  liminf an  a , then lim an  a
n 

37
Proof: (i) Given:
lim an  a  an  a   for n  k    an  a   for n  k
n 
Consider:   an  a , an  a   , (n  k ) an  supn k an  a  
 a    an , an    a , (n  k ) a    inf n k an  an
 a    inf an , sup an    a
a  lim(inf n k an ), lim(sup n k an )  a
k  k 
or a  liminf an , limsup an  a
a  liminf an  limsup an  a
 a  liminf an  limsup an

limsup an  a 
(ii) Given: 
liminf an  a 
sup n  k an  a   
  for k  r
inf n  k an  a   
sup an  a   
  for n  k  r
inf an  a   
  sup an  a   
  for n  r
  inf an  a   
Consider sup an  a   ,    inf an  a (n  r )
 sup an  a   , a    inf an (n  r )
 an  a   , an  a  
  an  a   (n  r )
 an  a   for n  r
 lim n  an  a

Theorem 3.10: (The Completeness Property of  )

In  every Cauchy sequence is convergent.
Proof: Given: Let (x n ) be a Cauchy sequence. Then
x n  x m   for n , m  k
   x n  x m  
Consider x n  x m   (n , m  k )
 x n  x m   (n , m  k )
38
 sup n k x n  x m   (m  k )
Let v k  sup x n v k  x m  
 v k    x m (m  k )
 v k    inf m k x m
Let u k  inf m k x m v k    u k
v k  u k  
Now,
limsup x n  sup n k an  v k  u k    inf n k x n    liminf x n  
 limsup x n  liminf x n  
Since  is arbitrary, we get
limsup x n  liminf x n _____(1)
But always, liminf x n  limsup x n _____(2)
From (1) and (2) limsup x n  liminf x n  lim x n exists.
n 

39
 CHAPTER 4

INFINITE SERIES
Definition 4.1: a1  a2  a3 
is called an infinite series (or, a series) and we write


a n  a1  a2  a3  a4  
n 1
Partial sums: Let S 1  a1
S 2  a1  a2
S 3  a1  a2  a3

S n  a1  a2    an
Then S 1 , S 2 ,, S n , are called partial sums.

Definition 4.2: (Convergence of Series)


Let a n  a  a   be a series and S n  a1  a2    an
n 1

Then  an is convergent if lim n  S n  S (exists) and we write
1

S   an
n 1
  n
[Note:  an is convergent to S  S   an  lim S n  S  lim  ak  S ]
n 1 n 1 n  n 
k 1

Remark 4.1: n 
S n  S  S n 1  S  S n 1  S  S n  p  S  n 1  

 n 1  
Definition 4.3: If a n does note convergent, it is called divergent.
n 1 n  p 
 
Properties: (I) If a
n 1
n is convergent to S , and b
n 1
n is convergent to S  ,

then,  (a
n 1
n  bn ) is converges to S  S  .

40
 
(II) If a
n 1
n is convergent to S and k is a constant, then the

series  (ka ) convergent to kS .

n

Proof: (I) Let S n  a1  a2  a3    an ,

S n  b1  b 2  b3    b n
Given: S n  S , S n  S 
To prove: If S   (a1  b1 )  (a2  b 2 )    (an  b n ) we have to
prove that S n  S  S  .
S n  (a1  a2  an )  (b1  b 2    b n )
= (S n )  (S n )  S  S 

(II) Let S n  a  a    an
Given: S n  S as n   .
To prove: If S n  ka1  ka2    kan we have to prove that
S n  kS .
S n  k (a1  a2    an )
 k (S n )  kS .

Theorem 4.1: If a
n 1
n is convergent, then lim an  0
n 

Proof: Let S n  a1  a2    an
Given: S n  S , (say)
( S n 1  S )
S n  (a1  a2    an 1 )  an
S n  (S n 1 )  an
lim S n  lim S n 1  lim an
S  S  lim an
 lim an  0

Important Note: -


a n convergent  lim an  0
n 
1

lim an  0  a n is not convergent
n 1

41
  
Example 4.1:  an   ( nn  12)
n 1 n 1
n 1
 an 
n 2
1
1
 n
2
1
n

1 0
 lim an   1 ( 0)  a n not convergent (divergent)
n  1 0 n 1

Remark 4.2: We have proved that  a converges
n  lim an  0 .
n 
1

But lim an  0 
 a n converges, as shown in the following example:
n 
1

 
1 1 1
Example 4.2:  an   1  
n 1 n 1 n 2 3
1
Here an 
n
 lim an  0
n 
But the series
1 1
 
1
2 3
is not convergent. (This will be proved later)

Geometric series:
a  ar  ar 2    ar n 1   is a geometric series.

Theorem 4.2: a  ar  ar 2    ar n 1   is convergent if r  1 , and divergent if

r  1.
Proof. Case (1): r  1
S n  a  ar  ar 2    ar n 1
rS n  ar  ar 2  ar 3    a r r n
S n  rS n  a  ar n

42
 (1  r )S n  a (1  r n )
a (1  r n )
S n 
(1  r )
a (1  lim r n )
n 
 lim S n 
n  1 r
a (1  0)
 , ( r  1)
(1  r )
a
 (exists)
(1  r )
a
 the geometric series converges to
1 r
a
We write a  ar  ar 2    , r  1.
1 r

Case (2): r  1  r  1 or r   1
When r  1, S n  a  a    a  na
 lim S n  lim na does not exists
 the series is divergent for r=1.

When r  1 , S n  a  ar  ar 2    ar n 1
 a  a  a  a  a
0 if n even

 a if n odd
 lim S n does not exists
n 
 the series is divergent for r  1

Case (3): r  1
Series  a  ar  ar 2    ar n 1  
( a1  a2  a3    an 1  )
 lim ar n 1  0 ( r  1, a  0)
n 

 (by Theorem 4.1) the series is divergent for r  1

1 1 1
Example 4.3: Show that the series  2  3   is convergent and find its value.
2 2 2

43
 1 1 2 1 3
Solution: Given series 
2
 () ()
2

2


( a  ar  ar 2  )
1 1
a  , r  ( 1)
2 2
1
Since r  which is  1 , the geometric series is convergent.
2
1
a
Its value   2  1.
1 r 1 1
2

Lemma 4.1: Let a
1
n be a series of positive terms ( an  0 n ) and

S n  M n ( M is a constant  0 ), then a
1
n is convergent to

S M.
Proof: S n  a  a    an  0 (given)
0  S n  M
 the sequence (S n ) is bounded.
Next, S1  a1
S 2  a1  a2
S 3  a1  a2  a3

Since an  0 n , S 1  S 2  S 3  
 (S n ) is an increasing sequence (monotonic sequence)
 (S n ) is a bounded monotonic sequence
 (by Theorem 3.4), S n  S  sup(S n )
n
i.e., lim S n  S (exists)
n 
 the series is convergent to S .
Next, given S n  M n
 sup{S n }  M
n

S  M

44
The Integral Test:

Let f ( x ), x  [1,  ) , be a positive, continuous and decreasing function. Then if

f (n )  an .
 

We prove that: (i)  an converges if

1
 f (x )dx converges (finite).
1
 

(ii)  an diverges if
1
 f (x )dx (infinite).
1

Proof.
f (n )  an
f (1)  a1
f (2)  a2

Sum of the area of inscribed rectangles  1  f (2)  1  f (3)    1  f ( n )

 a2  a3  a4    an
 (a1  a2  a3    an )  a1
 S n  a1 ________(1)
Sum of the area of the circumscribed rectangles =
1  f (1)  1  f (2)    1  f (n  1)
 a1  a2    an 1
 S n 1 _______(2)
We know that (from Calculus 1)
n

Area under the curve   f (x ) dx ______(3)

1
Clearly (1)  (3)  (2)

45
 n

 S n  a1   f (x )dx  S n 1
1
n n

S n  a1   f (x )dx ,  f (x )dx  S n 1
1 1
n

 S n  a1   f (x )dx ______(4)
1
n

 f (x )dx  S n 1 ______(5)
1


(i) Given  f (x )dx converges

1
n

  f (x )dx is finite
1
n

Let  f (x )dx  K
1

By (4), S n  a1  K put a1  K  M
S n  M n .

(By Lemma 4.1), a
1
n converges.


(ii) Given  f (x )dx  

1
n

i.e., lim  f (x )dx  

n 
1
by (5), lim S n 1  
n 

  an diverges.
1


1
Example 4.4: ( p  series ) Show that n n 1
p
is convergent if p  1 and divergent if

p  1.

46
Solution: Case (1): p  1
1
Here an  p ( f (n ))
n
1
Consider f (n )  p
n
Change n to x
1
 f (x )  p  x  p
x
 t
Use the Integral Test,  1
f (x )dx  lim  x  p dx
t  1
t
 x  p 1 
 lim  
t   p  1
 1
1
 lim t  p 1  1
1  p t 
1  1  p 1
 lim  p 1  1
1  p t   t   p 1  0
1
  0  1
(1  p )
1
 (finite)
1 p

1
by the integral test  p ( p  1) is convergent.
n 1 n

Case (2): p  1

1 1 1
 n 1 2  3 
n 1
1
Here an  ( f (n ))
n
1
Consider f (n ) 
n
1
Change n to x : f ( x ) 
x

47
  t
1
 f (x )dx  lim  x dx
t 
1 1

 lim ln x 1
t

t 
 lim ln t  ln1
t 
 lim(ln t )  
t 

1
 is divergent.
n 1 n

Case (3): p  1
As in case (1),

1
 f (x ) dx  lim[t  p 1 1]
(1  p ) 
t
1
1
 lim[t 1 p  1] p 1
(1  p ) t 
0 1 p

or 1  p  0

1
by the integral test  ( p  1) is divergent.
np


1
Example 4.5: n 2
is convergent ( p  2  1)
n 1

1 3
n 3 is convergent ( p   1)
n 1
2
2

1 2
n 2 is divergent ( p   1)
n 1
3
3

1
 n is divergent ( p  1)
n 1

48
Cauchy Criterion:

Consider a series  an  a1  a2  
n 1
Let S n  a1  a2    an

By Definition 4.2, a n converges  lim S n  S (exists)
n 
n 1
 the sequence (S n ) converges
 (S n ) is a Cauchy sequence.

Definition 4.4:( Cauchy Criterion)


A series a
n 1
n is said to satisfy the Cauchy Criterion if (S n ) is a Cauchy sequence.


Theorem 4.3: A series a
n 1
n converges  it satisfies the Cauchy Criterion)

Proof: Let S n  a1  a2  a3    an


a n converges  lim S n  S (exists)

n 
1
 the sequence (S n ) is Cauchy sequence.

 the series a n satisfies the Cauchy Criterion.
1
(by Def. 4.4)
Important Note:
(S n ) is a Cauchy sequence  (by Def. 3.3.) S n  S m   for n , m  r .
Let n  m .
S n  a1  a2    am    an
S m  a1  a2    am
Subtract: S n  S m  am 1    an
n
Sn Sm  a
k  m 1
k ,n  m  r

Remember:
 n

a n converges  (S n ) is a Cauchy sequence  a k   for n  m  r

1 k  m 1

49
Comparison Test:

Let a
1
n be a series with an  0 n
 
(i) If  an converges and bn  an n then bn converges.
1 1
 
(ii) If a
1
n   and b n  an n , then bn  
1
n n
Proof: b
k  m 1
k  
k  m 1
bk
n
 a k ( given b n  an ) a b  a  b
k  m 1
n a  b  c   a  b  c 
 a k ( an  0n )
k  m 1

  for n  m  r (  an convergent)
1
n
 b
k  m 1
k   for n  m  r

 the series b n is convergent.
n 1

(ii) Let S n  a1  a2    an , S n  b1  b 2    b n
Since, b n  an n , S n  S n
 lim S n  lim S n   (given)
n  n 

 lim S n  
n 

 bn  
1
 
1
Example 4.6: b   n
1
n
n 1
2
1

To find if b
1
n converges or diverges.

1
bn  2
n 1

50
 1 1
 bn  bn   2  an
2
n 1 n
 
1
Now  an   2
, which is convergent ( p  2  1 )
1 1 n
 
1
by the Comparison Test, b n   2 is also convergent.
1 1 n 1

Absolutely Convergent Series:

 
Definition 4.5: A series bn is called an absolute convergent series, if  bn is
n 1 n 1
convergent.

Theorem 4.4: If b
n 1
n is absolutely convergent, then it is convergent. i.e.,
 

 bn convergent  b n convergent
n 1 n 1

Proof. Given:  bn convergent . Put bn  an
n 1

[Note that an  0 n and a
1
n converges ]

by Comparison test, b1
n converges

 
Remark 4.3: b n converges 
 b n converges
1 1
[i.e., convergenes 
 absolutly convergent ]

1 1 1
Example 4.7: Take b n   
1
1 2 3 4
We will prove later that this series convergent.

1 1
Put  bn  1    
1 2 3

1
  divergent .
1 n

51
Root Test:
 1
Theorem 4.5: Let  an  a1  a2   be a series such that lim an
1
n 
n  L . Then

(i) The series converges if L  1 .

(ii) The series diverges if L  1 .

(iii) No decision if L  1 . (i.e.,  an may be converges or diverges.)
1
1
Proof: Given lim an n L
n 
1
 an n  L   for n  n 0
1
   an n L 
1
 L    an n  L 
1
Consider a n  L   _______(1)
1
and L    an n ______(2)
(i) L  1
We choose  such that L    1 . Put L    r (r  1) .
1
From (1), an n  r , (r  1)
 an  r n

n
r is a geometric series which is convergent ( r  1 )
1

by Comparison Test, a 1
n converges.

(ii) L  1
We choose  such that L    1
1
From (2) , an n  L   1
1
 an n 1
 an  1 n

 lim an  0   a divergent .
n
n 
1

52
 (iii) L  1
 
1 1 1
Take  an   1  
1 1 n 2 3
1
Here an 
n
1
1 1
an n  1
 1
n n
1 and the series diverges ( p  1 )
L 1
 
1
Next, take a   n
1
n
1
2

1
Here an 
n2
1
an  2
n
1
1 1
an n  2
( 1
)2  1 
1
n n and the series  converges ( p  2  1 )
n n 1 n2
L  1

we have no decision for L  1 .

Ratio Test:


an 1
Theorem 4.6: Let a 1
n be a series such that lim
n  an
 L . Then,

(i) a
1
n converges if L  1.

(ii) a
1
n diverges if L  1

(iii) No decision if L  1 .

53
 an 1
Proof: Given lim L
n  an
an 1
  L   for n  n 0
an
a
   n 1  L  
an
an 1
 L    L 
an

Consider
a n 1
 L   _ _ _ _ _ _ (1 )
an
a n 1
and, L    ______(2) fo r n  n 0
an
(i) L  1
For convergence we choose n 0  1,  n  1
We choose  such that L    1 . Put L    r (r  1)
an 1
From (1), r
an
an 1  an r
Put n  1: a2  a1 r
n  2: a3  a2 r  a1 r 2
n  3: a4  a3 r  a1 r 3
 
an  a1 r n 1

n 1
a r
1
1 is a geometric series which is convergent ( r  1 )

by Comparison Test,  an converges .
1
(ii) L  1
We choose  such that L    1 .

54
 Using (2),
an 1
 L   1
an
an 1
1
an
an 1  an
Put n  1: a2  a1
n  2: a3  a2
n  3: a4  a3

a1  a2  a3  
 lim an  0
n 

 a n diverges
1

(iii) L  1
 
1 1 1
Take a   n  1  2  3  
1
n
1

1 1
an 
, an 1 
n n 1
an 1 ( 1n 1) n 1 1
   1
 1
an ( 1n ) n  1 1  n 1
and the series diverges ( p  1 )
 
1
Next, take  an   2 (convergent)
1 1 n
1 1
an  2 , an 1 
n (n  1) 2
a 1
( n 1)2 n2 1
 n 1   2
 1
an 1
n2
(n  1) (1  n1 ) 2

So, if L  1, some series converges and some diverges.

no decision.
55
 CHAPTER 5
CONTINUOUS FUNCTIONS

Notation: D f  D ( f )  Domain of f .

Definition 5.1: A (real) function f is said to be continuous at a point x 0  D (f ) if

lim f (x )  f (x 0 ) , i.e.; given   0, there is a   0 such that
x x 0

f (x )  f (x 0 )   for x  x 0  
[ i.e., x  x 0    f (x )  f (x 0 )   ]

Remark 5.1: f is continuous if it is continuous at every point of D (f ) .

Theorem 5.1 (A): f is continuous at a point x 0  D (f )  for x n  D (f ) ,

x n  x 0  f (x n )  f (x 0 ).
Proof. To prove (  ):
Given f is continuous at x 0 ,
i.e., f ( x )  f ( x 0 )  ε for x  x 0  δ __(2)

(1)
We have to prove that x n  x 0  f (x n )  f (x 0 )
Consider x n  x 0 (as n  )
1
 x n  x 0  for n  n 0
n
1
Put  
n
 x n  x 0   , for n  n0
This is same as (2)
By (1), we get
f (x n )  f (x 0 )   , for n  n 0
 f (x n )  f (x 0 ), (as n  )
 x n  x 0  f (x n )  f (x 0 )
To prove: (  ) :
Given: f (x n )  f (x 0 )   for x n  x 0   (*)
We have to prove that f is continuous at x 0 .

56
 Suppose f is not continuous at x 0 .
Then n   , we find x n  D (f ) such that
f (x )  f (x 0 )   for x n  x 0  
i.e., f (x n )  f (x 0 )   for x n  x 0  
This contradicts the given (*)

Theorem 5.1 (B): f is continuous at x 0  D (f )

 for an open interval I f containing f (x 0 ) , there is an open interval J x 0
1
containing x 0 such that J x 0  f (I f ) [i.e., f (I f ) is an open set]

Proof. f is continuous at x 0
 f (x )  f (x 0 )   for x  x 0  
   f (x )  f (x 0 )   for -  x  x 0  
add f (x 0 ) and add x 0
 f (x 0 )    f (x )  f (x 0 )   for x 0    x  x 0  
 f (x )   f (x 0 )   , f (x 0 )    for x  (x 0   , x 0   )
 f (x )  I f for x  J x 0
 x f 1
(I f ) for x  J x 0 .

Always recall:
f is continuous at x 0  D (f )
 lim f ( x )  f ( x 0 )
x x 0

 f ( x )  f ( x 0 )   for x - x 0  
 x n  x  f (x n )  f (x )

Properties:
(1) f is continuous at x 0  D (f )  f is continuous at x 0 .
(2) f is continuous at x 0  D (f )  cf is continuous at x 0 (c  constant) .
(3) Let f and g be continuous functions at x 0  D (f ) . Then,

57
 f
f  g , fg , , ( g  0) are all continuous functions at x 0 .
g
(4) x is continuous on [0,  ) .
(5) x (n  ) is continuous on  .
n

(6) A polynomial a0  a1 x  a2 x 2    an x n is continuous.

Problems:
(1) If f and g are continuous at x 0  D (f ) . Prove that f  g is continuous
at x 0 .
Solution: f is continuous at x 0  lim f (x )  f (x 0 )
x x 0

g is continuous at x 0  lim g (x )  g (x 0 )
x x 0

Consider lim (f  g )(x )  lim[f (x )  g (x )]

x x 0 x x 0

 lim f (x )  lim g (x )
x x 0 x x 0

 f (x 0 )  g (x 0 )
 (f  g )(x 0 )
f  g is continuous at x 0 .

(2) Show that x n (n  ) is continuous on  .

Solution: Let x 0   , then
lim x n  lim x   
 x  x
x x 0 x x 0
n times

 x 0  x 0  x 0   x 0
 x 0n
 x n is continuous on  .

(3) If f is continuous at x 0  D (f ) , prove that f is continuous at x 0 .

Solution: Given : f is continuous at x 0  D (f )
 f (x )  f (x 0 )   for x - x 0  
Use a  b  a  b ,
f (x )  f (x 0 )  f (x )  f (x 0 )
  (given)

58
  f (x )  f (x 0 )   for x - x 0  
 f (x ) is continuous at x 0

 2 1
x sin( ), x  0,
(4) If f (x )   x , prove that f is continuous at 0 .
 0, x  0,
Solution: Here x 0  0
To prove: Given :   0 , there is a   0 such that
f (x )  f (x 0 )   for x - x 0  
i.e., f (x )  f (0)   for x  0  
i.e., f (x )  0   for x  
i.e., we prove that f (x )  ε by using x   __(2)
(1)

1
Consider f (x )  x 2 sin( )
x
2 1
x sin sin   1
x
2 1  sin   1
 x (1)
We prove that x 2   by using x  
i.e., we prove that x 2  ε by using x  
 (3)

put   
x  
2
x 
or x 2  
which gives (3)

1 1
 sin( ) , x  0,
(5) If f (x )   x x , prove that f is NOT continuous at 0 .
 0 , x  0,

59
 Solution: Here, x 0  0
Suppose f is continuous at 0.

Then, f (x )  f (x 0 )   for x - x 0  
i.e., f (x )  f (0)   for x  0  
i.e., f (x )   for x  
1 1
Consider f ( x )  sin( 2 )
x x
1
 (1)
x
1
 we have to get ε for x  
x (1)

1 1
  M (larg)
x 
This contradiction (1)

Homework:
1
 sin x , x  0,
(6) If f (x )   x , prove that f is not continuous at 0.
 0 , x  0,

Theorem 5.2(A): Let [a , b ] be a closed and bounded interval. If f :[a , b ]   is

continuous, then it is bounded.

Proof. Let x n  [a ,b ] , i.e., a  x n  b

 {x n } is a bounded sequence.
 (by Bolzano Weiertrass) there is a convergent subsequence
{x nk } .
Let x nk  x
 f ( x n k )  f ( x )  (1) ,(  f is contiuous)

To prove: f is bounded.
Suppose it is not bounded. Then, f (x n k )  n k , n k  
 {f (x nk )} is not convergent. This contradicts (1).

60
Remark 5.2: [a , b ] must be closed (and bounded) interval.
1
Example 5.1: f ( x )  , x  (0,1)
x
Clearly f is continuous on (0,1)
We prove that f is not bounded
1
f (x )  , 0  x  1
x
Choose x such that x  0  
1 1
i.e.,   M (larg)
x 
1
i.e., M
x
f (x )  M
 f is not bounded.

Theorem 5.2(B): Let [a , b ] be a closed and bounded interval and f :[a,b ]   is

continuous. Then there is a point c  [a , b ] such that
f (c )  sup{f (x ) : x [a,b ]} ( max {f (x ) : x [a,b ]})

Proof: f is continuous  f is bounded (by Th. 5.2(A))

sup{f (x ) : x [a,b ]} exists (by completeness axiom)
Let M  sup{f (x ) : x [a,b ]}
To prove: M  f (c ), c [a,b ]
Clearly (by definition of sup )
1
M   f (x n )  M , x n  [a ,b ]
n
Take limit as n  
M  lim n  f (x n )  M
 lim n  f (x n )  M ____(1)
Since x n  [a ,b ]  {x n } is bounded.
 (by Bolzano-Weiertrass theorem) there is a convergent subsequence{x n k }
Let x nk  c , c  [a,b ] Theorem :
 f (x n k )  f (c ),( f is continuous) an  a
By (1), f (x n )  M M  f (c )  an k  a

61
Homework:

Let [a , b ] be a closed and bounded interval and f :[a ,b ]   a continuous function.

Prove that there is a point d  [a ,b ] such that
f (d )  inf{f (x ) : x [a,b ]}( min{f (x ) : x [a,b ])
(Proof is similar to that of Theorem 5.2 (B))
Theorem 5.2 (C):(Intermediate Rule (Value) Theorem)
Let f be a continuous function on an interval I, a , b  I , a  b and
f (a)  y  f (b ) [or f (b )  y  f (a)] . Then there is at least one point x 0  (a ,b ) such
that f (x 0 )  y .
Proof.
Take f (a)  y  f (b ) .
Let S  {x [a,b ]: f (x )  y } .
Since S is bounded (S  [a,b ]) ,
sup(S ) exists in [a , b ]
Let
x 0  sup(S )
1
 (definition of sup) x 0   S n  x 0 for some S n  S
n
Take limit as n  
x 0  lim S n  x 0
n 

 lim S n  x 0
n 

or, S n  x 0 (as n  )
Since f is continuous, f (S n )  f (x 0 )
or, lim f (S n )  f (x 0 ) ____(1)
n 

S n  S  f (S n )  y
 lim f (S n )  y
n 
Using (1), we have f (x 0 )  y _____(2)
1
Let t n  min( x 0 , x 0  )
n
1
 x 0 tn  x 0 
n
62
Take limit as n  
 x 0  lim t n  x 0
n 

 lim t n  x 0
n 

or, t n  x 0 (as n  )
Since f is continues, then f (t n )  f (x 0 ) or, lim f (t n )  f (x 0 ) ____(3)
n 
Now, t n  S ( x 0  t n ) [see figures]
 f (t n )  y
i.e, f (t n )  y
 lim f (t n )  y
n 
Using (3), then f (x 0 )  y _____(4)
From (2),(4), we get f (x 0 )  y .

Uniformly Continuous Functions:

Recall: (i) f is called a continuous function at x 0  D (f ) , if given   0, there is a

  0 such that f (x )  f (x 0 )   for x  x 0   , x  D (f )
(ii) Let S  D (f )
f is called continuous function on S if x 0  S and   0 , there is a   0
such that f (x )  f (x 0 )   for x  x 0   , x  D (f )

Note:  depends on  and x 0 .

Definition 5.2: (Uniformly Continuous Function)

Let S  D (f )
f is said to be uniformly continuous on S if given   0 , there is a   0 such that
f (x )  f ( y )   for all x , y  S and x  y   .

Note: Uniformly Continuous Function are defined only on sets.

1
Example 5.2: Show that f (x )  is uniformly continuous on [a, ), a  0
x2
Solution: To prove: f (x )  f ( y )   for all x , y  [a, ) and x  y  
1 1
Consider f (x )  f ( y )  2
 2
x y

63
 y 2 x 2

x 2y 2
( y  x )( y  x )

x 2y 2
(y  x ) x a
 y x [ x , y  [a ,  ), a  0  ]
x 2y 2 y a
y x
x y 2 2  2 2l
x y y x
1 1
 x y 2  2l
x y xy
1 1 x a
 x  y 2  2 l, ( )
a a aa y a
2
 x y 3l
a
2
 f (x )  f ( y )  3 x  y
a
 we have to prove that
2
x  y   _____(1)
a3
for all x , y [a, ) and x - y  
 a3
Put  
2
a3
x  y 
2
2
or, 3 x  y  
a
This gives (1)

Remark 5.3: uniformly continuous continuous.


1
Example 5.3: Show that f ( x )  on (0,1) is continuous but not uniformly
x
continuous.

64
Solution: Let x 0  (0,1)
1
lim f (x )  lim ( )
x x 0 x x 0 x

1

lim x
x x 0

1

x0
 f (x 0 )
 lim f (x )  f (x 0 )
x x 0

 f is continuous at x 0  (0,1)
To prove: f is not uniformly continuous on (0,1)
We have to prove that
f (x )  f ( y )   for some x , y  (0,1) and x  y  

Take x   , y 
1 

x y  
1 
1
  1
1 
1  1

1 

 | |    (1)
1  
1
1

1 1
f (x )  f ( y )  
x y
1 1 
 
 
11  
   ( is small)
 
 f (x ) - f ( y )  
 f is not uniformly continuous on (0,1)

65
Homework: Show that f ( x )  x 2 on  is continuous but not uniformly continuous.
1
(Take x  n , y  n  )
n
Theorem 5.3: Let [a , b ] be a closed and bounded interval. If f :[a ,b ]   is
continuous, then f is uniformly continuous.

Proof: Suppose f is not uniformly continuous. Then

f (x n )  f ( y n )   __(1)
for x n , y n  [a ,b ] ____(2)
and
1
xn  yn   ___(3)
n
1
(we choose   )
n
Since x n  [a,b ],i.e., a  x n  b ,x n  is bounded
 
 (by Bolzano-Wreirstrass Theorem), {x n } has a convergent subsequence x n k .
Let x n k  x  [a,b ]___(4)
 f (x n k )  f (x )(5) , [ f is continuous]
1
By (3), we have x n k  y nk  ( 0)
nk (6)

By (1), we have f (x n k )  f ( y nk )   __(7)

Consider y n k  x  ( y nk  x nk )  (x n k  x )
 y n k  x nk  x nk  x
 0 by(6)  0 by(4)

y nk  x  0
i.e., y nk  x [ f continuous]
 f ( y n k )  f (x ) _____(8)
From (5) and (8) we get
f (x nk )  f ( y nk )  0
i.e., f (x n k )  f ( y nk )  
This contradiccts (7)
66
Theorem 5.4: Let f be uniformly continuous on S   . If {x n } is a Caushy sequence
in S , then f (x n ) is a Caushy sequence (in  )
Proof: {x n } is a Caushy sequence in S
 x n  x m   for n , m  r _____(1)
f is uniformly continuous on S .
 f (x n )  f (x m )   for every x n , x m  S
 f (x n )  f (x m )   for n , m  r (by (1))
 f (x n ) is a Cauchy sequence.

67
 CHAPTER 6

DERIVATIVES

1. Differentiability
Definition 6.1: Let f : I  , I is an interval of real numbers, and let c be an interior
point of I (i.e., I is neighborhood of c). f is differentiable at c if lim f ( xx)cf (c ) exists.
x c
In this case, the value of this limit is called the derivative of f at c and is denoted by
the symbol f (c ).

Remarks 6.1:
1. An equivalent form of Definition 1 is to say that f is differentiable at c if
lim f (c  hh) f (c ) exists.
h 0
2. The Definition 1 can be generalized by assuming that f is defined on a domain set
A, where c is accumulation point of A, and that limit is taken as x approaches c
assuming only values in A.
3. In (2), if a function f is defined on the closed interval [a,b], we can define a one-
side derivative of f at a by lim f ( xx)af (a )  f r(a ), or a one-sided derivative of f at b by
x a
f ( x )f (a )
lim x a  f l (a ). The derivative f  may then be regarded as function defined for
x a 
all x in the interval I.
df
4. Another symbol for the derivative of f is dx , which is the differential notation due
to Leibniz.

Example 6.1: Let f (x )  x 3 . Using Definition 6.1,

x 4 c4
f (c )  lim  lim( x 3  x 2c  xc 2  c 3 )  4c 3 .
x c x  c x c

Using the alternate form of the definition,

(c  h ) 4  c 4 4c 3h  6c 2 h 2  4ch 3  h 4
f (c )  lim  lim
h 0 h h 0 h
3 2 2 3 3
 lim(4c  6c h  4ch  h )  4c
h 0
Either form of the definition shows that f /(x) exists, and it is equal to 4x2.

Example 6.2: Let the function f (x )  x be defined on the interval [0,  ).

68
Since
x  c 1 1
lim  lim 
x c x c x c x  c 2 c
For all c  0, f is differentiable for all x  0, and f (x )  2 1x .

Example 6.3: Let s (x )  sin x and c (x )  cos x . Then

sin x  sin 0 sin x
s (0)  lim  lim  1.
x 0 x 0 x 0 x

Remark 6.2: The last limit in Example 3 is obtained in a typical calculus course by a
geometric argument. For Euler, who defined sin x by a power series, this fact would be
obvious, since if we start with the series expansion for sin x, divided by x, and then
take the limit as x approaches 0, we would have
x3 x5
sin x  x    
3! 5!
sin x x2 x4
1   
x 3! 5!
sin x x2 x4
lim  lim(1     )  1
x 0 x x 0 3! 5!
In a similar way,
cos x  cos0 cos x  1
c (0)  lim  lim
x 0 x 0 x 0 x
and this limit is proved to equal 0 by use of the identity sin 2 x  cos2 x  1, and the fact
that lim sinx x  1 .
x 0

Example 6.4: Let f :    be a function that satisfies the following three conditions.
1) f (x  y )  f (x )  f ( y ) for all x , y  R
2) f (0)  1
3) f (0) exists.
Then f (x ) exists for all x   and equals f (0)  f (x ) .
Solution.
f (x  h )  f (x )
f ( x )  lim
h 0 h
f ( x )f ( h )  f ( x )
 lim
h 0 h

69
 f (h )  1
 f (x )lim
h 0 h
Therefore f (x ) exists, if lim f ( hh )1 exists. But this limit exists, because by conditions
h 0
(2) and (3), then
f (h )  f (0) f (h )  1
f (0)  lim  lim .
h 0 h 0 h 0 h
Do you know any familiar function f satisfy the three conditions given in this
example? (Hint: f (x )  b x )

Example 6.5: Let s :    and c :    be functions satisfying the following

conditions.
1) s (x  y )  s (x )c ( y )  s ( y )c (x ), for all x , y 
2) c (x  y )  c (x )c ( y )  s (x )s ( y ) for all x , y 
3) lim s (xx )  1 and lim c ( xx )1  0 .
x 0 x 0

Then s (x )  c (x ) and c (x )  s (x ) for all x   .

Solution.
s (x  h )  s (x )
s ( x )  lim
h 0 h
s (x )c (h )  s (h )c (x )  s (x )
 lim
h 0 h
c (h )  1 s (h )
 s (x )lim  c (x )lim
h 0 h h 0 h

 s (x )  0  c (x ) 1  c (x ),
since the necessary limit are assumed to exist by condition (3).
Similarly we prove: c (x )  s (x ) .
Do you know any familiar functions s and c satisfy the three conditions given in this
example? (Hint: sin x and cos x )
Lemma 6.1: If f : I  R is differentiable at an interior point c of the interval I , then
f is continuous at c .
Proof. The hypothesis implies that lim f ( xx)cf ( c ) exists, and equals the real number
x c
denoted by f (c ) . Then
f (x )  f (c )
lim  f (x )  f (c )   lim  (x  c )
x c x c x c
f (x )  f (c )
 lim  lim(x  c )  f (c )  0  0
x c x c x c

70
and so f is continuous at c, since lim f ( x )  f (c ).
x c
The converse of Lemma 2 is not true.
Example 6.6: Since f ( x)  x is continuous, but not differentiable at c  0.

2. Algebra of Derivatives;

Theorem 6.1: (Limit Theorem for Derivatives)

Let f : I   and g : I   be differentiable at an interior point c of the interval I
(i.e.; I is a neighborhood of c ). Then the following results are true.
1) f  g is differentiable at c and
(f  g )(c )  f (c )  g (c ) .
2) fg is differentiable at c and
(fg )(c )  f (c )  g (c )  f (c )  g (c ).
f
3) is differentiable at c if f (c )  0, and
g
f g (c )f (c )  f (c ) g (c )
( )(c )  .
g ( g (c )) 2

Proof. We shall prove (2) and leave the others as exercises. We want to prove that fg
is differentiable at x  c  lim f ( x ) g ( xx)cf (c ) g (c ) exists.
x c
Since g is differentiable at c, g is continuous at c , so that
lim g (x )  g (c ).
x c
Then we have
f (x ) g (x )  f (c ) g (c ) f (x ) g (x )  f (c ) g (x )  f (c ) g (x )  f (c ) g (c )
lim  lim
x c x c x c x c
f (x )  f (c ) g (x )  g (c )
 lim g (x )lim  lim f (c )lim
x c x c x c x c x c x c
 g (c )f (c )  f (c ) g (c ).

Theorem 6.2: (The Chain Rule) If f : I   is differentiable at an interior point c

of the interval I , and g ( y ) is differentiable at y  f (c ), then the composite function
( g  f )(x ) is differentiable at c and
( g  f )(c )  g (f (c ))  f (c ) .

71
Proof.
( g  f )(x )  ( g  f )(c )
( g  f )(c )  lim
x c x c
g (f (x ))  g (f (c )) f (x )  f (c )
 lim 
x c f (x )  f (c ) x c
g (f (x ))  g (f (c )) f (x )  f (c )
 lim  lim  g (f (c ))f (c )
x c f (x )  f (c ) x c x c
  
Here x c f ( x )f (c ) by the continuity of f

Example 6.7:
(a) Using dxd (sin x )  cos x , and Quotient Rule, we know that
d d 1 cos x
(csc x )  f p   2   csc x cot x .
dx dx sin x sin x
(b) dxd (sin y )  cos y , and the Chain Rule, and assuming that y  arcsin x is
differentiable, we can show that
d 1
(arcsin x ) 
dx 1 x 2
dy dy
For if y  arcsin x , x  sin y , and so 1  cos y dx or dx  cos1 y . Since cos y 
1  sin 2 y  1  x 2 , and since the slope of the tangent line to the graph of arcsin x
is always positive, the derivation is complete.
(c) The power rule can be extended to negative integer exponent since
d n d 1 (x n )0  1(nx n 1 )
(x )  f p  nx  n 1.
dx dx x n x 2n

(n )
Example 6.8: If possible, find an expression for the nth derivative f (a ) of the
following functions.
(a) Let f (x )  sin x and a  0 . Then,
f (x )  cos x , so f (0)  1
f (x )   sin x , so f (0)  0
(3) (3)
f (x )   cos x , so f (0)  1
(4) (4)
f (x )  sin x , so f (0)  0
(n )
And the pattern of 1, 0, -1, 0 repeats in blocks of four. So we can express f (0) as
follows. Let f (2 n 1) (0)  (1) n 1 and f (2 n ) (0)  0.
(b) Let f (x )  ln x and a  1 . Then we have

72
 1
f (x ) 
x
1
f (2) (x )   2
x
2
f (3) (x )   3
x
6
f (4) (x )   4
x
(5) 24
f (x )   5
x
From the pattern developing in these derivatives, we conjecture that
(n  1)!
f ( n ) (x )  (1)n 1
xn
so that
f ( n ) (1)  ( 1) n 1 (n  1)!

(c) Let f (x )  tan x and a  0 . Then

f (x )  sec2 x
f (x )  2sec 2 x tan x
f (x )  2sec2 x  4 tan 2 x sec2 x
As more derivatives are taken, the expressions become increasingly complicated and
no general pattern emerges.

3. The Mean Value Theorem

Definition 6.2: The function f : I   has a relative minimum (local minimum) at

c  I if there is a neighborhood U of c such that f (x )  f (c ) for all x U  I .
Similarly, f has a relative maximum (local maximum) at c  I if there is a
neighborhood U of c such that f (x )  f (c ) for all x U  I .

Remarks 6.3: A function f : I   is said to have a maximum (or, ‘global

maximum’) at c  I if f (x )  f (c ) for all x  I (here I need not be a neighborhood
of c ); if f (x )  f (c ) for all x  I then f is said to have a minimum (or ‘global
minimum’) at c .
Lemma 6.2: Let f : I   have a relative (local) maximum or minimum at x  c ,
and assume that f (c ) exists. Then f (c )  0.

73
Proof. Assume that f has a relative minimum at the point c  I . It follows that for
all x in some interval of c ,
f (x )  f (c )
 0 for x  c
x c
f (x )  f (c )
 0 for x  c
x c
Therefore, we have
f (x )  f (c )
lim 0
x c x c
f (x )  f (c )
lim 0
x c x c
Since f (c ) exists, it follows that
f (x )  f (c ) f (x )  f (c )
f (c )  lim  lim
x c x c x c x c
and so f (c )  0.
Remark 6.4: A similar proof can be given if f has a relative maximum at point c .
Theorem 6.3: (Rolle’s Theorem)
If f is continuous on [a ,b ], differentiable on (a,b ), and f (a )  f (b )  0, then there
is some value c  (a,b ) so that f (c )  0.
Proof. We first consider the trivial case where f (x )  0 for all x  [a ,b ] . The Theorem
is true in this case because f (c )  0 for all c  (a ,b ).
Otherwise, there must be at least one x  (a ,b ) where f (x )  0 . If we assume that
f (x )  0 for some x  (a,b ), then M  sup{f (x ) : x  [a,b ]} is a positive finite
number, and there must be a value c  (a ,b ), so that f (c )  M . Why? It then follows
from Lemma 6.2 that f (c )  0 .
If f (x )  0 for some x  (a ,b ), then a similar argument can be used involving
m  inf{f (x ) : x  [a,b ]}

Example 6.9: The polynomial y  x 3  4x  f (x ) is continues and differentiable

for all x ,   x   . If we take a  2, b  2 ,
The hypotheses of Rolle’s Theorem are satisfied since
f (2)  f (2)  0.
Thus,
f (x )  3x 2  4
Must be zero at least once between -2 and +2.

74
In fact, we find 3x 2  4  0 at
2 3 2 3
x  c1   and x  c1   .
3 3
Theorem 6.4: (Mean Value Theorem).
If f is continuous on [a , b ] and differentiable on (a,b ), then there is at least one value
c  (a,b ) such that
f (b )  f (a )
f (c )  .
b a
Remark 6.5: It is fortunate that the Mean Value Theorem has a simple geometric
interpretation- namely, if the graph of f has no breaks in it and no sharp points, then
at least one tangent line to f must be parallel to the line joining the endpoints of the
graph (see Fig. ).

Proof of the Mean Value Theorem. We define the function

f (b )  f (a )
F (x )  f (x )  f (a )  (x  a )
b a
and assert that F is satisfies the hypotheses of Rolle’s Theorem. You should verify
that F is continuous on [a ,b ], differentiable on (a,b ), and such that F (a )  F (b )  0.
Since F (x )  f (x )  f (bb)af (a ) , the conclusion that F (c )  0 for some c  (a ,b ) is
equivalent to the statement that f (c )  f (bb)af (a ) , and the proof is complete.

Example 6.10: Illustrate the MVT for f (x )  x 3  4x 2  2 on [ 1, 2].

Solution. We need to find some c  (1, 2), so that

75
 f (2)  f (1)
 f (c )
2  (1)
This equation becomes
3c 2  8c  7,
for which the solutions are
4  37 4  37
c  0.69 and c   3.36 .
3 3
While the tangent lines to the curve at both points are parallel to the secant line, only
the point 43 37 is in the interval (1,2) . See the next Fig.

Example 6.11: Suppose that f is continuous on [a , b ] and differentiable on (a,b ),

and that f (x )  1 for all x  (a ,b ) . Show that f has at most one fixed point. That is,
show that there is at most one   [a , b ] with f ( )  .

Solution. Suppose that f has at least 2 fixed points. Let 1  [a ,b ] and 2  [a ,b ] such
that 1  2 , and f (1 )  1 , and f (2 )  2 . Since f is continuous on [1 , 2 ] and
differentiable on (1 , 2 ) , we know by the Mean Value Theorem that there exists
c  (1 , 2 ) with
f (2 )  f (1 ) 2  1
f (c )    1.
2  2 2  1
However, this is a contradiction because c  (a ,b ) but f (x )  1 for every x  (a ,b ).
We next consider some result that can be proved as corollaries of the MVT.

76
Corollary 6.1:If f   0 on interval I , then f is an increasing function on I . The
interval I can be open or closed, finite or infinite.
Proof. Exer.
Corollary 6.2: If f   0 on interval I , then f is an decreasing function on I .
Proof. Exer.

Corollary 6.3: If f has a bounded derivative on interval I , then f is uniformly

continues on I .
Proof. Exer.

Example 6.12: The sequence (s n ), where s n  lnnn , is decreasing for all n  3. This
follows from Corollary 6.2, because the function f (x )  lnxx is decreasing for x  e ,
since f (x )  1xln2 x  0 when 1  ln x , or when x  e .

Corollary 6.4: If f   g  on [a ,b ], then there is some constant C so that

f (x )  g (x )  C , for all x  [a , b ].
Proof. Exer.

Example 6.13: f ( x )  x12 is uniformly continuous on the set S  [c , ) for c  0,

because of Corollary 6.3, and the fact that for all x  S ,
2 2
f (x )  3  3 .
x c

Example 6.14: The function f (x )  x is uniformly continuous on [0,1] , by our work

in last Chapter. Further, f (x )  x is uniformly continuous on [0,  ) by Corollary
6.3, and the fact that for all x  [1, )
1 1
f (x )   .
2 x 2

Example 6.15: Use the Mean Value Theorem to find upper and lower bounds for 67.
Solution. By applying the Mean Value Theorem to f (x )  x on the interval
[64,67], we have
67  64 1

67  64 2 c
for some value c for which

77
 64  c  67  81.
This leads to the following string of inequalities.
1 1 1
 
18 2 c 16
1 67  8 1
 
18 3 16
1 3
8  67  8
6 16
8.16666  67  8.1875 .

4. Taylor’s Theorem with Remainder:

Theorem 6.5: Let f (t ) be a function whose first n derivative are continuous on

( n 1)
[a ,b ], and is such that f (t ) exists on (a,b ) .

(n )
f
(a )
Define Pn (t )  f (a )  f (a )(t  a )    (t  a ) n
n!
Then for each value of x  (a ,b ), there is some value c between a and x so that
f (x )  Pn (x )  R n (c , x )
where
f ( n 1) (c )
R n (c , x )  ( x  a ) n 1.
( n  1)!

Remark 6.6:
1. R n (c , x ) is called the remainder term. Pn (x ) is the Taylor polynomial of degree n,
and it approximates the value of f (x ) with an error given by R n .

2. This theorem is really an infinite collection of results, all of which can be

established by single proof.

3. Special cases: (i) if n  0, Taylors Theorem becomes the Mean Value Theorem,
since
f (x )  P0 (x )  R 0 (x )  f (a)  f (c )  (x  a )
reduces to
f (x )  f (a )
 f (c ) .
x a
78
(ii) If n  1 ,
f (c )
f (x )  f (a )  f (a )(x  a )  (x  a )2 .
2!

4. We use the term power series to refer to a series of the form  

n 1 an x
n
or
 n 1 an (x  c ) . The form
n
n 1 an x
n
is called a Maclaurin series, while
 n 1 an (x  c )
n
is called a Taylor series in powers of ( x  c ) . Now the Maclaurin
series is a special case of the Taylor series, obtained when c  0.

5. If f (n )
(x ) exists for all n , and for all x  [a ,b ], and if lim R n (x 0 )  0 for
n 
x 0  [a ,b ], then the sequence {Pn (x 0 )} converges to f (x 0 ) , and we call
 (k )
f
lim Pn (x 0 )   (x 0  a )k
n  k! k 0
The Taylor series expression for f at x 0 in powers of (x 0  a ).

Example 6.16: Use Taylor’s Theorem with n  1 to find upper and lower bounds for
67. Compare the accuracy of Example 6.6 and this example.

Solution. We set f (t )  t , a  64, x  67 , and use

1 1
f (t )  and f (t )  3/2
2 t 4t
To obtain the equation
1 1
67  64  (67  64)  3/2 (67  64)2
2 64 8c
3 9
 8   3/2
16 8c
for some value c so that 64  c  67  81. This can be used to obtain the following set
of inequalities.
1 9 9
 
8  92 8c 3/2 84
9 3 1
 4 67  8 
8 16 8  92
3 9 3 1
8   67  8 
16 4096 16 648
8.1853  67  8.18596 .

79
This gives an approximation for 67 that is accurate to 3 decimal place, while the case
when n  0 gives only 1-decimal-place accuracy (see Example 6.6).

Exercises:

1. Apply Taylor’s Theorem to f (x )  ln x , using a  1 and n  3 . Then use this result

to find a value for ln 3.
2. Find Pn ( x ) and R n ( x ) when Taylor’s Theorem is applied to f (x )  cos x using
a  0.
3. Find the values of x so that lim R n (x )  0 in 2.
4. Find Pn ( x ) and R n ( x ) when Taylor’s Theorem is applied to f ( x )  x  1 with
a  0.
5. Use the result in 4 to approximate 67.

5. Indeterminate Forms and L’Hopital’s Rule:

1. The indeterminate Form 0/0

If f and g are continuous at x  a but f (a )  g (a )  0, the limit
f (x )
lim 
x a g ( x )

cannot be evaluated by substituting x  a , since this produces 0/0, a meaningless

expression known as an indeterminate form.
As we have seen in calculus, the value of the limit in Eq.  is hard to predict
x 2 4
lim  lim(x  2)  4.
x 2 x  2 x 2

sin x
lim  1.
x 0 x

1  cos x
lim  0.
x 0 x

The limit
f (x )  f (a )
f (a )  lim
x a x a

80
form which we calculate derivative always produces the indeterminate form 0/0. Our
success in calculating derivative suggests that we might turn things around and use
derivatives to calculate limits that leads to indeterminate forms.
For example,
sin x sin x  sin 0 d
lim  lim  (sin x )  cos 0  1.
x 0 x x 0 x 0 dx x 0
L’Hopital’s rule gives an explicit connection between derivatives and limits that lead
to the indeterminate form 0/0.

Theorem 6.6: A (L’Hopital’s Rule-First Form)

Suppose that f (a )  g (a )  0, that f (a ) and g (a ) exist, and that g (a )  0. Then
f (x ) f (a )
lim  .
x a g ( x ) g (a )
Proof. Working backward from f (a ) and g (a ), which are themselves limits, we have
f (x )  f (a ) f (x )  f (a )
f (a ) lim x a x a
 x a  lim
g (a ) g (x )  g (a ) x a g (x )  g (a )
lim
x a x a x a
f (x )  f (a ) f (x )  0 f (x )
 lim  lim  lim .
x a g ( x )  g (a ) x a g ( x )  0 x a g ( x )

Example 6.17:
3x  sin x 3  cos x
i) lim  2
x 0 x 1 x 0

1
1  x 1 2 1  x 1
ii) lim   .
x 0 x 1 2
x 0

x  sin x 1  cos x
iii) lim  ?
x 0 x3 3x 2 x 0

Note 6.1:
What can we do about the limit in Example 1 (iii)? The first form of L’Hopital’s
Rule does not tell us what the limit is because the derivative of g (x )  x 3 is zero at
x  0 . However, there is a stronger form of L’Hopital’s Rule that says that whenever
the rule gives 0/0 we can apply it again, repeating the process until we get a different
result. With this stronger rule we can finish the work begun in Example 1 (iii)

81
Theorem 6.6: B (L’Hopital’s Rule-Stronger Form)
Suppose that
f (x 0 )  g (x 0 )  0
and that the function f and g are both differentiable on an open interval (a,b ) that
contains the point x 0 . Suppose also that g   0 at every point in (a,b ) except possibly
x 0 . Then
f (x ) f (x )
lim  lim , 
x x 0 g ( x ) x  x 0 g ( x )

provided the limit on the right exists.

Proof. The proof is based on a special version of the Mean Value Theorem. It can be
found in our text book.
Example 6.18:
1  x  1  (x / 2) 0
lim ( )
x 0 x2 0
(1 / 2)(1  x ) 1/2  (1 / 2) 0
 lim (Still )
x 0 2x 0
(1 / 4)(1  x ) 3/2 1
 lim  .
x 0 2 8

Note 6.2:
As you apply L’Hopital’s rule, look for a change from 0/0 to something else. That is
where the limit’s value is revealed.

Example 6.19:
1  cos x sin x 0
lim  lim  0
x 0 x  x 2 x 0 1  2 x 1

If we continue to different in an attempt to apply L’Hopital’s Rule once more,

1  cos x sin x cos x 1
lim  lim  lim  ,
x 0 x  x 2 x 0 1  2x x 0 2 2
which is wrong.

Note 6.3: If we reach a point where one of the derivative is zero and the other is not,
the limit in question is either zero, as in Example 3, or infinity, as the next example.

82
Example 6.20:
sin x 0
lim ( )
x 0 x2 0
cos x
 lim  .
x 0 2x

Other Indeterminate Forms

L’Hopital’s Rule applies to the indeterminate form  /  as well 0/0. If f (x ) and
g (x ) both approach infinity as x approaches a , then
f (x ) f (x )
lim  lim ,
x a g ( x ) x a g ( x )

Provided the limit on the right exists. In the notation x  a , a may be either finite
or infinite.
Example 6.21:
tan x sec 2 x
I) lim  lim  1.
x  /2 1  tan x x  /2 sec 2 x

x  2x 2 1  4x 4 2
II) lim 2  lim  lim  .
x  3x  5x x  6x  5 x  6 3

Note 6.4: The forms   0 and    can sometimes be handled by changing the
expressions algebraically to get 0 / 0 or  /  instead.

1
Example 6.22: The limit lim x sin
x  x
leads to the form   0, but we can change it to the form 0/0 by writing x  1 / t and
letting t  0 :
1 1 sin t
lim x sin  lim  sin t  lim  1.
x  x t 0 t t 0 t

Example 6.23: Find

1 1
lim f  p.
x 0 sin x x
Solution. If x  0 we have the form   () . If x  0 we have the form
   . But may also write
1 1 x  sin x
 
sin x x x sin x
83
and apply L’Hopital rule to the expression on the write. Thus,
1 1 x  sin x 0
lim f  p  lim ( )
x 0 sin x x x 0 x sin x 0
1  cos x 0
 lim (still )
x 0 sin x  x cos x 0
sin x
 lim  0.
x 0 cos x  x sin x

When to Use L’Hoptial Rule, and When to Stop

To find
f (x )
lim
x a g ( x )

By L’Hopital’s rule, proceed to differentiate f and g as long as you still get

the form 0/0 or  /  at x  a . Stop differentiating as soon as you get
something else. L’Hopital’s rule does not apply when either the numerator or
denominator has a finite nonzero limit.
Also you have to know sometimes L’Hopital’s rule fails to find the limit.

Example 6.24: Show that we cannot use L’Hopital’s Rule to evaluate

x 2 sin x1
lim .
x 0 sin x

Solution. When we apply L’Hopital’s Rule, we obtain

x 2 sin x1 2x sin x1  cos x1
lim  lim .
x 0 sin x x 0 sin x
However, the second limit does not exist because lim cos x1 does not exist, and so
x 0
L’Hopital’s Rule is inconclusive. We are able to evaluate this limit by use of the Limit
Theorem because

lim = lim lim sin = 1 . 0 = 0.

→ → →

x 2 sin x1
Question: Use Limit Theorem to find lim . Can we use L’Hopital’s Theorem
x 0 sin x
to find this limit? Justify you answer.

84
 CHAPTER 7

THE RIEMANN INTERGRAL

Definition 7.1: Let f be a bounded function on a closed and bounded interval [a ,b ]

into subintervals by the points
a  x 0  x 1  x 2  x n  b

This is called a partition.

Let us define
M k  sup x [ x k 1 ,x k ]{f ( x )}, m k  inf x [ x k 1,x k ]{f ( x )}.
n n
Let S   M k x k , s   m k x k where x k  x k  x k 1.
k 1 k 1
S is called the Riemann upper sum and s is called the Riemann lower sum.
Now, we define
I  inf {S }, J  sup{s }
P P

I is called the Riemann upper integral of f on [a , b ] and is denoted by

b

 a
f ( x )dx .
J is called the Riemann lower integral of f on [a , b ], and is denote d by
b

 a
f ( x )dx

If I  J , then we say that f is Riemann integrable on [a , b ], and is den-

b

oted by f ( x )dx , which is called the Riemann integral of f on [a , b ].

a

85
Important Points:

(1) f bounded on [a,b ].

(2) A partition: a  x 0  x 1  x 2    x n  b .
P  the set of all such partitations x k  x k  x k 1
(3) M k  supx [ x k 1 ,x k ]{f (x )}, m k  inf x [ x k 1,x k ]{f (x )}
n n
(4) S   M k x k , s   m k x k .
k 1 k 1
(5) I  inf{S }, J  sup{s }
P P
(6) s  J  I  S
b

(7) f is Riemann integrable  I  J and I  J   f (x )dx

a

Remark 7.1: Sometimes we write S n for S and s n for s .

Example 7.1: Let


1 if x is a ratianal number in [a ,b ],
f (x )   ,

0
 if x is an iratianal number in [a ,b ].
prove that f is not Riemann integrable on [a,b ].

Solution: We prove that I 

 J.

We have
M k  1, mk  0
n n
 S   M k x k , s   m k x k
k 1 k 1
n
  (1)x k 0
k 1
 b a
I  inf P {S }  b  a.
Next
J  sup P {s }  0
I J
 f is not riemann integrabal on [a ,b ]
86
Necessary and sufficient condition for Riemann integrability:

Theorem 7.1: Let f be a bounded function on [a,b ]. Then f is Riemann integrable

(i.e., I  J )  given ε  0, there is a partition of [a , b ] such that S  s  ε .

Proof: To prove () : Given S  s  ε 

To prove I J (f is Riemann integrable)
Use s  J  I  S
Clearly I J S s
Using  , we get
I J S s ε
 I J ε

Since ε is arbitrary (ε  0)
We get I J  0
I J
To prove (  ): Given I  J ( f is Riemann integrable)
To prove S  s  ε
By definition,
I  inf{S }, J  sup{s }
P P
ε ε
 I  S , J  s
2 2
ε ε
 S I  , s J 
2 2
ε
 s J 
2

Now,
ε
S I 
2
ε
s  J 
2

S s  I J ε
S  s  ε ( I  J )

87
Riemann integrable as the limit of a sum:

Definition 7.2: Let f be a bounded function on [a , b ] and a  x 0  x 1  x 1  x 2 

n
 x n  b a partition of [a,b ]. Let t k  [ x k 1 , x k ] . If lim  f (t k ) x k exists, then we
n 
k 1
say that f is Riemann integrable on [a ,b ] and we write
n b

lim  f (t k ) x k   f ( x )dx ,
n 
k 1 a

which is the Riemann integrable of f on [a,b ].

n b

Theorem 7.2: I  J  lim  f (t k )x k exists and I  J   f (x )dx

n 
k 1 a
Proof: (by Theorem 7.1)

I J  S n sn  ε  lim (S n  s n )  0
n 
 I  J  lim S n  lim s n 
n  n 
We know that s n  J  I  S n
 lim s n  J  I  lim S n
n  n 
By  ,
lim S n  lim s n  I  J 
n  n 
Since, t k  [x k 1 , x k ], we have
m k  f (t k )  M k
Multiply by x k ,
m k x k  f (t k )x k  M k x k
n n n
  m k x k   f (t k )x k   M k x k
k 1 k 1 k 1
n
s n   f (t k )x k  S n
k 1
Take limit as n  
n
lim s n  lim  f (t k )x k  lim S n 
n  n  n 
k 1
By ,
n
I  J  lim S n  lim s n by  lim S n  lim s n  lim  f (t k )x k .
n  n  n  n  n 
k 1

88
By ,
n b

I  J  lim S n  lim s n  lim  f (t k )x k   f ( x )dx .

n  n 
k 1 a
ε
 M k x k  m k x k  x k
b a
n n n
ε
  M k x k   m k x k   x k
k 1 k 1 b  a k 1
ε
 S s  b a
b a
 S  s  ε  (by Theorem 7.1) I  J  f is Riemman integrable.

Special types of Riemann integrable functions:

Theorem 7.3: A continuous function f on [a,b ] is Riemann integrable on [a,b ] .

Proof:
f :[a , b ]   is continuous (given)
 (by Theorem 5.2 in Chapter 2) f is bounded.
 Also, f is uniformly continues on [a , b ]
ε
 f (x )  f ( y )  for all x , y  [a,b ] and x - y  
b a
ε
 M k  mk 
b a
Multiply by x k
ε
 M k x k  m k x k  x k
b a
n n n
ε
  M k x k   m k x k   x k
k 1 k 1 b  a k 1
ε
 S s  b a
b a
 S  s  ε  (by Theorem 7.1) I  J  f is Riemman integrable.

89
Theorem 7.4: A monotonic function f on [a,b ] is Riemann integrable on [a,b ].

Proof: Suppose that f is increasing (monotonic). By definition we have

a  x 0  x 1  x n  b
f (a )  f (x 0 )  f (x 1 )    f (x n )  f (b )
Then,
m k  f ( x k 1 ), M k  f n (x k )
n n
 S   f (x k )x k , s   f (x k 1 )x k
k 1 k 1
n
or S  s  f (x k )  f (x k 1 ) x k 
k 1
Choose a partition so that
f (b )  f (a )
ε
x k  
f (b )  f (a )
From  , since f (x k )  f (x k 1 ), we have
n
ε
S s  f (x k )  f (x k 1 )
f (b )  f (a ) k 1
ε
 f (b )  f (a )  
f (b )  f (a )
 by Theorem 7.1 the required follows.

Theorem 7.5: A function f of bounded variation on [a,b ] is Riemann integrable on

[a,b ] .

Proof: This follows from 7.4 and the fact that a function of bounded variation can be
expressed as two difference of two monotonic increasing function.

Theorems on Riemann integrable functions:

Theorem 7.6: If f 1 and f 2 are Riemann integrable on [a,b ] , then f 1  f 2 is Riemann

integrable on [a,b ] and
b b b

 a
[f 1 (x )  f 2 (x )]dx   f 1 (x )dx   f 2 (x )dx
a a

90
Proof:
Method 1: Let x k  t , then we have by Definition 5.2,
b n

 a
f 1 (x )  f 2 (x ) dx  nlim

f 1 (t k )  f 2 (t k )x k
k 1
t 0
n n
= lim  f 1 (t k )x k  lim  f 2 (t k )x k
n  n 
t 0 k 1 t 0 k 1
b b
  f 1 (x ) dx   f 2 (x ) dx .
a a

Method 2: Homework.

Theorem 7.7: If f is Riemann integrable on [a,b ] , and c is any constant, then cf is

Riemann integrable on [a,b ] and
b b

 a
cf (x )dx  c  f (x )dx .
a
Proof: The idea of the proof is similar to proof of the above theorem (Theorem 7.6)

Theorem 7.8: If f is Riemann integrable on [a,b ] , then

b a a

 a
f (x )dx   f (x )dx ,
b 
a
f (x )dx  0

Proof: (Homework).

Theorem 7.9: If f is bounded and Riemann integrable on [a,b ] , and c is any point
of [a,b ] , then
b c b

 a
f (x )dx   f (x )dx   f (x )dx .
a c
Proof: (Homework).

Theorem 7.10: If f and g are Riemann integrable on [a,b ] and f  g , then

b b

 a
f (x )dx   g (x )dx
a

Proof: Let
M k(1)  sup x [ x k 1 ,x k ]{f (x )}, m k(1)  inf x [ x k 1 ,x k ]{f (x )},
M k(2)  sup x [ x k 1 ,x k ]{g (x )}, m k(2)  inf x [ x k 1 ,x k ]{g (x )}.

91
Also, let
n n
S (1)   M k(1)x k , s (1)   m k(1)x k ,
k 1 k 1
n n
S (2)
 M (2)
k x k , s (2)
  m k(2)x k .
k 1 k 1
Then, since f ( x )  g (x ),
M k(1)  M k(2) , m k(1)  m k(2)
 S (1)  S (2) , s (1)  s (2) .
Let
I (1)  inf{S (1) }, J (1)  sup{s (1) }
P P

I (2)
 inf{S (2)
}, J (2)
 sup{s (2) }.
P P
Then, we have
I (1)  I (2) , J (1)  J (2) .
Since
b b
I (1)  J (1)   f (x )dx , I (2)  J (2)   g (x )dx , (given) ,
a a
we have
b b

 a
f (x )dx   g (x )dx .
a

Theorem 7.11: If f is Riemann integrable on [a,b ] and has upper and lower bounds
M and m on [a,b ] respectively, then
b
m (b  a )   f (x )dx  M (b  a ),
a
and if  is a number such that m    M , then
b

 a
f (x )dx  (b  a ) .
Proof: We have
m  f (x )  M .
b b b
 a
m dx   f (x ) dx   M dx . (by Theorem 7.10)
a a
b
or m (b  a )   f (x ) dx  M (b  a ).
a
It follows,  such that
m M ,

92
  m (b  a)  (b  a)  M (b  a)
b
  f (x )dx  (b  a).
a

Example 7.2: Find

 /2
(i)  (x  3cos x )dx .
0
5
(ii)  4
x dx

Solution:
 /2  /2  /2
(i) 
0
(x  3cos x ) dx   x dx  3 cos x dx

0 
0 
by Theorem 5.6 and Theorem 5.7


  3.
8
(ii) Since

x if x  0
x 


x
 if x  0
5 0 5
  x dx   x dx   x dx
4 
4 0
by Theorem 7.9
0 5
  x dx   x dx
4 0

1 1
  [02  (4) 2 ]  [52  0]  20.5
2 2

Example 7.3: Prove that

1 1 x9 1
 dx  .
10 2 0 1 x 10
Solution: Let
1
f (x ) 
, g (x )  x 9
1 x
Then f and g are continuous functions and have positive values on [0,1]. So by
Theorem 7.11
1 1 x9 1
m  x 9 dx   dx  M  x 9 dx
0 0 1 x 0

93
where
M  max{f }, m  min{f }
1 1
on [0,1]. Clearly M  1, m  12 and  x 9 dx  . Thus
0 10
9
1 1 1 x 1
( )  dx  1 ( )
2 10 0 1 x 10
Example 7.4: Show that
4

1
1  x 4 dx  7.5.
Solution: Notice that
1 x 2  x  1 x 2  x  x .
Since x  x for x  0, we have 1 x 2  x for 1  x  4. Thus, by Theorem 7.10,
we have
4 4 1
1 1  x 4
dx  1 x dx  2 (42 12 )  7.5.
Example 7.5: Begin by the fact that says if f (x )  0 on [a , b ] implies that
b
 f (x ) dx  0 to show that
a
b b

 a
f (x ) dx   g (x ) dx for
a
f ( x )  g ( x ) on [a , b ].

Solution: We have f ( x )  g (x ), then f (x )  g (x )  0 and so

b

 a
(f  g )  0.
Then, Theorem 7.6 gives
b b b

 a
f   g   (f  g )  0
a a
and so
b b

 a
f dx   g dx .
a

Theorem 7.11: (Mean-value theorem) If f is continuous on [a,b ] , then there exists

a number c  [a ,b ] such that
b

 a
f (x )dx  (b  a )f (c ) 

94
Proof:
Given:
f is continuous on [a,b ]
 f Riemann integrable on [a,b ] (by Theorem 7.3)
b
  s.t. m
 M
 s.t.
min max

a
f (x )dx  (b  a ) 
But a continuous function takes on all values between its min and max values (Math
315) and so
c s.t. f (c )   (Math 315)
b
 
a
f (x )dx  (b  a )f (c ) (by ) .

Remark 7.2: The equation  in Theorem 7.1 can be written as

1 b

b  a a
f (c )  f (x )dx

Example 7.6: Find the mean value of f ( x )  1  x  x 2 on [0,4].

Solution: Since f is continuous on [0,4], then c  [0, 4] such that

4
f (c )(4  0)   (1  x  x 2 )dx
0
64 19
 4f (c )  4  8   f (c )  
3 3
To find c (if wanted?), we have
19
f (c )  1  c  c 2 
 1 c  c 2  3c 2  3c  22  0
3
Now solve the quadratic equation to get c  ?.

Example 7.7: If f is continuous and the M.V. of f (x ) on [1,3] is 7 , then find the
M.V. h (x )  f (x )  5 on [1,3].
Solution: Since f is continuous on [1,3] and f (c )  7, then
3 3
f (c )(3 1)   f (x ) dx   f (x ) dx  7(2)  14
1 1
and since h ( x ) is continuous on [1,3], then

95
 3 3
h (c )(3 1)   h (x ) dx  2h (c )   (f (x )  5) dx
1 1
3 3
 2h (c )   f (x )dx   5dx
1 1

 2h (c )  14  5(3 1)
 2h (c )  14  10
24
 h (c )   12
2
Thus h (c )  12 is M.V of h ( x ) on [0,3].

Theorem 7.12: If f is Riemann integrable on [a,b ] and f (x )  M for some

constant M , then
b

 a
f (x )dx  M (b  a )

Proof: Exercise.

Theorem 7.13: If f is Riemann integrable on [a,b ] , then (a) f (x ) is Riemann

integrable on [a,b ] , and (b)
b b


a
f (x )dx   f (x ) dx
a

Proof: (a) Let


f (x ), f (x )  0 
f (x ), f (x )  0
F1 (x )  
 , F2 (x )  

 0,
 otherwise 
 0,
 otherwise
Then we have
f (x )  F1 (x )  F2 (x ),
f (x )  F1 (x )  F2 (x ) 
Now since f is Riemann integrable, so also are F1 (x ) and F2 (x ). Thus by Theorem
7.6, F1 (x )  F2 (x )  f (x ) is Riemann integrable.
(b) From  we have
b b b

 a
f (x )dx   F1 (x )dx  F2 (x )dx
a a
b b b

 a
f (x ) dx   F1 (x )dx  F2 (x )dx
a a
Then we have
b b b

 a
f (x )  a
F1 (x )dx  a
F2 (x )dx

96
 b b b
  F1 (x )dx  F2 (x )dx   f (x ) dx .
a a a

Example 7.8: (a) Prove that

b b


a
f (x )dx   f (x ) dx
a
if a  b .

(b) Prove that

2 sin nx
lim  dx  0.
n  0 x 2 n2
Solution: (a) By absolute value,
n n n

 f (t
k 1
n )x k   f (t n )x k   f (t n ) x k
k 1 k 1

Taking the limit as n   and each x k  0, we have the required result.

(b) By part (a), we have
2  sin nx 2  sin nx 2  dx 2
0 x 2  n 2 dx  0 x 2  n 2 dx  0 n 2  n 2
Then
2  sin nx
lim  dx  0,
n  0 x 2 n2
and so the required result follows.

Differentiation and integration:

Definition 7.3: The indefinite Riemann integral is defined as

x
F (x )   f (u )du
a
where the variable of integration has been changed to u avoiding confusion with the
limit of integration x .
The following theorems are important. For the proof see our text book p.156-180

Theorem 7.14: If f is bounded and Riemann integrable on [a,b ] , then f (x ) is

Riemann integrable on [a,b ] , and
x
F (x )   f (u ) du
a
is continuous on [a,b ] .

97
Theorem 7.15: If f is Riemann integrable on [a,b ] and F ( x )   ax f (u )du , then
d d x

dx a
F ( x )  F ( x )  f (u )du  f (x )
dx
at each point of continuity of f (x ).

Theorem 7.16: (Rolle's theorem) Let f (x ) be continuous on [a ,b ] and suppose that

f (a )  f (b )  0. Then if f (x ) exists in (a,b ), there is a point c in (a,b ) such that
f (c )  0.

Theorem 7.17: (Mean-value theorem) Let f (x ) be continuous on [a , b ] and suppose

that f (x ) exists in (a,b ), there is a point c in (a,b ) such that
f (b )  f (a )
f (c )  .
b a
Theorem 7.18: (Fundamental theorem of Calculus) Let f be Riemann integrable
on [a,b ] , and suppose that there exist a function F continuous on [a,b ] such that
F (x )  f ( x ). Then
b b

a
F (x )dx   f (x )dx  F (b )  F (a ) or
a
x x

a
F (u )du   f (u )du  F (x )  F (a )
a

Theorem 7.19: (Integration by parts) Let f and g be Riemann integrable on [a,b ]

, and suppose that F ( x ) and G ( x ) are such that F (x )  f (x ) and G (x )  g (x ) on
[a,b ] . Then
b b

 a
F (x ) g (x )dx  F (b )G (b )  F (a )G (a )   f (x )G (x )dx
a
b
 F (x )G (x ) a   f (x )G (x )dx
b

Theorem 7.20: (Change of variables) Let f be continuous on [a,b ] . Let x  g (u )

have a continuous derivative on [ ,  ] where a  g ( ), b  g ( ). Then
b b

a
f (x )dx   f  g (u ) g (u )du .
a

98
Example 7.9: Find the derivative of the following function:
x 1
F (x )   dt
a 1  sin 2 t

Solution: Using F.T.C, we have

1
F ( x )  .
1  sin 2 x

Example 7.10: Find the derivative of the following function:

x3 1
F (x )   dt
a 1  sin 2 t

Solution: First note that F (x )  h ( g (x ) , where

x 1
h (x )   dt , g (x )  x 3
a 1  sin 2 t

1 3x 2
Thus, F (x )  h ( g (x )) g (x )   3x 2
 .
1  sin 2 x 3 1  sin 2 x 3

Example 7.11:


 x 2 sin(1 / x ), x  0
(a) Prove that the derivative of F (x )   


 0, x 0
is given by

2x sin(1 / x )  cos(1 / x ), x  0
F ( x )  



 0, x 0
2/ 8
(b) Show that  F (x )dx  2 .
2/ 

Solution. (a) If x  0, then

d
{x 2 sin( x1 )}  x 2{cos( x1 )}( x12 )  2x sin x1  2x sin x1  cos x1 .
dx
If x  0, then
F (h )  F (0) h 2 sin(1 / h )  0
lim  lim  lim h sin( h1 )  0.
h 0 h h  0 h h 0

(b) By part (a), F ( x ) exists at all values of x in ( 2 /  , 2 /  ). Thus, by F. T,

2/ 8
 F (x )dx  F ( 2 )  F ( 2 )  2
2/ 

99
Improper Riemann integrals:

If the interval [a ,b ] is infinite or if f (x ) becomes infinite at one or more points of

[a ,b ], the Riemann integral of f (x ) is often referred to as an improper integral. Such
integrals are defined by appropriate limiting procedures and are said to exist if the
corresponding limits exist. The following are examples of such definition.
 b


a
f (x )dx  lim  f (x )dx 
b  a
b b



f (x )dx  lim
a a
f (x )dx 
 b



f (x )dx  lim 
a a
f (x )dx 
b 

If f (x ) becomes unbounded when x  c , then we define

b
f ( x )dx  lim  c 1 f ( x )dx  b f ( x )dx  
a  a
1 0  c  2 
 2 0

If the limit in  does not exist for 1   2 but does exist when 1   2 , we say that the
corresponding value of the limit is the Cauchy principle value of ba f (x )dx . Also in
such case we say that the Riemann integral of f (x ) exists in the Cauchy principle
value sense.

Note 7.1: A similar remark can be made for  for which the Cauchy principle value,
if it exists, is obtained by taking a  b and letting b  .

Example 7.12: Prove that

 dx

1 ( x  2)3
dx

exists and find its value.

Solution: We have by definition,

b
 dx b dx (x  2)2
1 (x  2)3  blim
 1 ( x  2)3
 lim
b  2 1
1 1 1
 lim(  ) 
b  18 2(b  2)2 18

100
Example 7.13: Prove that
dx 2


7 ( x  1)5/3

exists as an improper Riemann integral provided that it is taken in the Cauchy principal
value sense.
Solution: The integral 1 / (x 1)5/3 is unbounded at x  1 so that the integral is
improper. In the Cauchy principal value sense
2 dx  1 dx 2 dx 
 lim 
7 (x  1)5/3  0  7 (x  1)5/3 1 (x  1)5/3 
1 2
 3 3 
2/3 2/3
 lim   (x  1)  (x  1) 
 0
 2 7 2 1 

 3 3 3 3 
 lim   ( ) 2/3  (8)2/3  (1)2/3  ( ) 2/3 
 0
 2 2 2 2 
9
 .
8
Note that if we defined
2 dx  11 dx 2 dx 
 lim 
7 (x  1)5/3 10  7 (x  1)5/3 12 (x  1)5/3 
 2 0

where  1   2 , the integral does not exist.

 sin x
Example 7.14: Prove that  1 x
dx exists.

1 sin x
Solution: The Riemann integral 0 x
dx exists. Also, on integrating by parts
we have
b
b sin x cos x b cos x cos b b cos x


1 x
dx  
x 1

1 x2
dx  cos1 
b
 1 x 2 dx
Now,
cos x b b cos x b dx 1
1 x 2 dx  1 x 2 dx  1 x 2 dx  1 b
Then, taking the limit as b  , we see that

101
 b cos x  cos x
lim  dx  1 x 2 dx
b  1 x2
exists as a Riemann integral.

b sin x 1 sin x b sin x

Now, since  1 x
dx  
0 x
dx  
1 x
dx
we see that

sin x b sin x 1 sin x  sin x

0 x dx 
b  0
lim
x
dx  0 x dx  1 x dx
And then it exists.

 sin x
Example 7.15: Prove that  0 x
dx

diverges, i.e.,  f n (x )dx is not absolutely convergent.
0
Solution: If n is any positive integer, we have

nπ sin x π sin x 2π sin x nπ sin x

0 x
dx  
0 x
dx  
π x
dx    
( n 1) π x
dx

π sin u π sin u π sin x

 du   du     du
0 u 0 u π 0 u  ( n 1)π

π sin u π sin u π sin x

 du   du     du
0 u 0 u π 0 u  ( n  1)π

2 2 2
 

π 2π (n 1)π
2 1 1
 (1     )
π 2 n 1
But since the limit of the last series as n   is infinite, the required result follows.

π
Example 7.16: Find  0
1  sin x dx

Solution: We can solve this problem by two methods.

102
Method 1. Firstly,
1  sin x  1  sin x
1  sin x  ,
1  sin x
provided that x  π2 . Note that 1  sin x  0, so 1  sin x is well defined. Thus,
1  sin 2 x cos 2 x cos x
1  sin x   
1  sin x 1  sin x 1  sin x


 cos x , x  [0, π2 ]
cos x  

 cos x , x  [ 2 , π ]

π

Now, let t  [0, π2 ) and s  ( π2 , π ]. Then,

π t cos x s  cos x

0 1  sin x dx  lim
t  π2
 
0 1  sin x
dx  lim
s  π2
 
0 1  sin x
dx
t π

 2 lim (1  sin x )1/2  2 lim (1  sin x )1/2

t  π2 s  π2
0 s

 2 lim ( 1  sin t 1)  2 lim (1  1  sin x )

t  π2 s  π2

 2(1)  2(1)  4.

Method 2. We have
1  sin x  (sin 2 ( x2 )  cos 2 ( x2 ))  2 sin( x2 ) cos( x2 )  (sin( x2 )  cos( x2 )) 2
Also,
x π
0 x π  0  .
2 2
Thus,
1  sin x  sin( x2 )  cos( x2 )  sin( x2 )  cos( x2 ),
since sin x and cos x are nonnegative in [0, π2 ]. So,
π

 1  sin x dx  2  cos( x2 )  sin( x2 ) 0  4.

π

103