The Riemann and the Generalised Riemann Integral

Calvin
17 July 2014

Contents
1 The Riemann Integral 2
1.1 Riemann Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Properties of Riemann Integrable Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 The Fundamental Theorem of Calculus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4 Lebesgue’s Integrability Criterion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.5 Substitution Theorem and Integration by Parts . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2 The Generalised Riemann Integral/Henstock-

Kurzweil Integral 16
2.1 Definition and Main Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2 Improper and Lebesgue Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.3 Infinite Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.3.1 Integral on [a, ∞) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.3.2 Integrals on (∞, b] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.3.3 Integral on (∞, ∞) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.4 Convergence Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3 Reference 34
2 Riemann Integral

1 The Riemann Integral

1.1 Riemann Integral
Here, we will not provide motivation of the integral, discuss its interpretation as the “area under the graph,”
or its applications. We will focus only on the mathematical aspects of the integral.
Definition 1.1.1 (Partition). A partition P of [a, b] is a finite set of points x0 , x1 , . . . , xn such that

a = x0 < x1 < ... < xn−1 < xn = b

n
We denote partition P by P := {[xi−1 , xi ]}i=1 and [xi−1 , xi ] is called the ith interval of P .
Definition 1.1.2 (Mesh/norm of Partition P ). We define the norm of P to be

||P || := max {x1 − x0 , x2 − x1 , . . . , xn − xn−1 } (1.1)

Therefore, we can say that the norm of P is the length of the largest subinterval in the Partition P .
n
Definition 1.1.3 (Tagged Partition). For a partition P = {[xi−1 , xi ]}i=1 in [a, b]. Let us choose a point
ti ∈ [xi−1 , xi ] for every i < n, we call ti as tags of subinterval [xi−1 , xi ]. The partition P together with its tags
are called tagged partition of interval [a, b] denoted by
n
Ṗ := {([xi−1 , xi ] , ti )}i=1

Remarks:
1. Tags can be chosen in any way(eg. left endpoint, right endpoint, midpoint, etc)

2. An endpoint of a subinterval can be used as a tag for 2 consecutive subintervals

n
Definition 1.1.4 (Riemann Sum). Let f : [a, b] → R be a function with tagged partition Ṗ = {([xi−1 , xi ] , ti )}i=1 .
Then we define the Riemann Sum of f to be
n
X
S(f ; Ṗ ) = f (ti ) (xi − xi−1 ) (1.2)
i=1

Using this knowledge of the Partition, now we can define the Riemann Integral
Definition 1.1.5 (Riemann Integral). A function f : [a, b] → R is said to be Riemann Integrable on [a, b] if
there exists L ∈ R such that for every  > 0 there exists δ > 0 such that if P is a tagged partition of [a, b] with
||Ṗ || < δ, then  
S(f ; Ṗ ) − L < 
 

We usually call L as the Riemann Integral of f over [a, b] denoted by

Z b Z b
L= f or f (x)dx
a a

We denote the set of all Riemann Integrable functions on [a, b] as R[a, b].
Remark:
1. Although the definition of the Riemann Integral!is similar to the definition of the limit of the Riemann

Sum as the norm goes to 0, lim S(f ; Ṗ ) = L , S(f ; Ṗ ) is not a function of ||Ṗ ||. So this is a different
||Ṗ ||→0
type of limit from what we have studied before.
Theorem 1.1.6 (Uniqueness Theorem). If f ∈ R[a, b], then its integral is unique
Riemann Integral 3

Proof. Assume instead that L1 and L2 both satisfy Definition 1.1.5 with L1 6= L2 , without loss of generality,
assume that L2 > L1 . Then, for all  > 0, there exists δ1 > 0 such that if a tagged partition P˙1 with norm
||P˙1 || < δ1 , then    
S f ; P˙1 − L1  < 
 

Also, there exists δ2 > 0 such that if a tagged partition P˙2 with norm ||P˙2 || < δ2 , then
   
S f ; P˙2 − L2  < 
 

Choose δ = min{δ1 , δ2 } > 0. Let Ṗ be a tagged partition with norm ||Ṗ || < δ. Since ||Ṗ || < δ1 and ||Ṗ || < δ2 ,
then    
S f ; Ṗ − L1  <  (1.3)
 

and    
S f ; Ṗ − L2  <  (1.4)
 

L2 −L1
We then fix  = 2 > 0. Then from Equation 1.3, we get
  3L1 − L2   L +L
1 2
− + L1 < S f ; Ṗ <  + L1 =⇒ < S f ; Ṗ <
2 2
Similarly, Equation 1.4 gives us
  L1 + L2   3L − L
2 1
− + L2 < S f ; Ṗ <  + L2 =⇒ < S f ; Ṗ <
2 2
So we have
3L1 − L2   L +L
1 2
  3L − L
2 1
< S f ; Ṗ < < S f ; Ṗ <
2 2 2
which is impossible so we have a contradition that L1 6= L2 . Therefore, the value of the integral is unique.

The following theorem allows us to form algebraic combinations of integrable function

Theorem 1.1.7. Suppose f, g ∈ R[a, b], then:
a. If k ∈ R, then kf ∈ R[a, b] and
Z b Z b
kf = k f
a a

b. f + g ∈ R[a, b] and
Z b Z b Z b
(f + g) = f+ g
a a a

c. If f (x) ≤ g(x) for all x ∈ [a, b] then

Z b Z b
f≤ g
a a

n
Proof. Consider Ṗ = {([xi−1 , xi ] , ti )}i=1 on [a, b], then
n
X n
X
S(kf ; Ṗ ) = kf (ti )(xi − xi−1 ) = k f (ti )(xi − xi−1 ) = kS(f ; Ṗ )
i=1 i=1

Using similar methods, we can conclude that

S(f + g; Ṗ ) = S(f ; Ṗ ) + S(g; Ṗ )

and
S(f ; Ṗ ) ≤ S(g; Ṗ )
4 Riemann Integral

For part a, as f ∈ R[a, b], then for all  > 0, there exists δ > 0 such that if a tagged partition Ṗ with norm
||Ṗ || < δ, then  
   Z b  
S f ; Ṗ − f <
 
 a  |k|
So  Z b    Z b   
       Z b  |k|
S kf ; Ṗ − k f  = kS f ; Ṗ − k f  = |k| S f ; Ṗ − f < =
     
 a   a   a  |k|
Rb
Therefore, the Riemann Integral of the function kf is equal to k a f
For part b, as f, g ∈ R[a, b], then for all  > 0, there exists δ1 > 0 such that if a tagged partition P˙1 with
norm ||P˙1 || < δ1 , then  
   Z b  
 ˙
 S f ; P1 − f <

 a  2

Similarly, there exists δ2 > 0 such that if a tagged partition P˙2 with norm ||P˙2 || < δ2 , then
 
   Z b  
S g; P˙2 − g <
 
 a  2

We let δ = min{δ1 , δ2 } > 0. Since ||Ṗ || < δ1 and ||Ṗ || < δ2 , then
   
   Z b      Z b  
S f ; Ṗ − f  < and S g; Ṗ − g < (1.5)
   
 a  2  a  2

So, by Triangle Inequality

 Z b Z b !   
    Z b   Z b 
S f + g; Ṗ − f+ g  = S f ; Ṗ − f + S g; Ṗ − g
   
 a a   a a 
   
   Z b     Z b   
≤ S f ; Ṗ − f  + S g; Ṗ − g < + = 
   
 a   a  2 2
Rb Rb
Therefore, the Riemann Integral of the function f + g is equal to a f + a g
For part c, we use Equation 1.5.
   Z a   Z a Z b
 <  =⇒ −  < S f ; Ṗ − 
   
S f ; Ṗ − f f =⇒ f − < S f ; Ṗ
  2 2 2
b b a
  Z a   Z a  Z b
 <  =⇒ S g; Ṗ −  
    
S g; Ṗ − g g < =⇒ S g; Ṗ < g+
  2 2 2
b b a

As we know that S(f ; Ṗ ) ≤ S(g; Ṗ ), we have

Z b Z b Z b Z b
 
f− < g+ =⇒ f< g+
a 2 a 2 a a
Rb Rb
So, we can conclude that a f ≤ a g
Theorem 1.1.8. Let h : [a, b] → R such that h(x) = 0 except for a finite number of points in [a, b], then
Rb
h ∈ R[a, b] and a h = 0
Proof. We just need to proof for the case of only 1 point (i.e. h(x) = 0 except for a point c), then we can use
induction to prove for a finite number of points.

For all  > 0, let δ < 2|h(c)| > 0 such that if a tagged partition ||Ṗ || < δ, then
 n 
    X 
S h; Ṗ − 0 =  h(ti )(xi − xi−1 ) ≤ 2|h(c)|||Ṗ || < 2|h(c)|δ = 
   
 
i=1
Rb
Therefore, the function h is integrable with a
h=0
Properties of Riemann Integrable Functions 5

Corollary 1.1.9. If g ∈ R[a, b] and if f (x) = g(x) except for a finite number of points in [a, b], then f is
Rb Rb
Riemann integrable and a f = a g.
Proof. Consider h : [a, b] → R such that h(x) = f (x) − g(x) then we can use Theorem 1.1.8 to show that
Rb
a
h = 0 Therefore,
Z b Z b Z b Z b Z b Z b
(f − g) = h = 0 =⇒ f− g = 0 =⇒ f= g
a a a a a a

Theorem 1.1.10 (Boundedness Theorem on Integrals). If f ∈ R[a, b], then f is bounded on [a, b].
Remark:
1. The converse (All bounded functions are Riemann Integrable) is not true, and example is shown in
Example 1.2.2
Rb
Proof. Assume by contradiction that f ∈ R[a, b] is an unbounded function with a = L. By choosing  = 1,
there exists δ > 0 such that if a tagged partition Ṗ with ||Ṗ || < δ, then
      
S f ; Ṗ − L < 1 =⇒ S f ; Ṗ  < |L| + 1 (1.6)
   

n
Then we let Q = {[xi−1 , xi ]}i=1 be a partition with ||Q|| < δ. Since f is not bounded on [a, b], then there exists
a subinterval [xk−1 , xk ] in which f is not bounded.
Since f is not bounded for [xk−1 , xk ], for all M > 0, there
 exists x ∈ [a, b] such that |f (x)| > M . Here
P
we fix M such that M = |L| + 1 +  i6=k f (ti )(xi − xi−1 ) To make the contradiction with Equation 1.6, we


need to choose appropriate tags. For subinterval [xi−1 , xi ] such that i 6= k, pick ti = xi . For i = k we choose
tk ∈ [xk−1 , xk ] such that
 
 
X 
|f (tk )(xk − xk−1 )| > M = |L| + 1 + 
 f (ti )(xi − xi−1 )
 i6=k 

By using the Triangle Inequality (|x + y| ≥ |x| − |y|∀x, y ∈ R),

 
    
 X 
S f ; Q̇  ≥ |f (tk )(xk − xk−1 )| −  f (ti )(xi − xi−1 ) > |L| + 1
  
 i6=k 

This contradicts Equation 1.6.

1.2 Properties of Riemann Integrable Functions

Theorem 1.2.1 (Cauchy Criterion). Let f : [a, b] → R. f ∈ R[a, b] if and only if for every  > 0, there exists
δ > 0 such that if tagged partitions Ṗ and Q̇ of [a, b] with ||Ṗ || < δ and ||Q̇|| < δ, then
    
S f ; Ṗ − S f ; Q̇  < 
 

Rb
Proof. ( =⇒ ) As f ∈ R[a, b], let a f = L. If we choose tagged partitions Ṗ and Q̇ such that ||Ṗ || < δ and
||Q̇|| < δ, then          
S f ; Ṗ − L < and S f ; Q̇ − L <
   
2 2
So, we have
                   
S f ; Ṗ − S f ; Q̇  = S f ; Ṗ − L + L − S f ; Q̇  ≤ S f ; Ṗ − L + S f ; Q̇ − L < + = 
       
2 2
And so we are done.
6 Properties of Riemann Integrable Functions

( ⇐= ) For every  > 0, let δ > 0 such that if tagged partitions Ṗ and Q̇ with ||Ṗ || < δ and ||Q̇|| < δ, then
     
S f ; Ṗ − S f ; Q̇  <
 
2

Let P˙n be a sequence of tagged partitions with ||P˙n || < δ for all n ∈ N. So we choose m, n ∈ N and without
loss of generality, we assume m > n. Clearly, this means that ||P˙n || < δ and ||P˙m || < δ which in turn implies
     
S f ; P˙n − S f ; P˙m  <
 
2
 
This shows that S f ; P˙n is a cauchy sequence and in turn, converges to some number L ∈ R. Or in other
words,     
∀ > 0, ∃N ∈ N : n > N =⇒ S f, P˙n − L <
 
2
Rb
Finally, to show that f ∈ R[a, b] and that a f = L, given  > 0, we let δ > 0 and n > N such that if a
tagged partition Q̇ with ||Q̇|| < δ, then
              
S f ; Q̇ − L ≤ S f ; Q̇ − S f ; P˙n  + S f ; P˙n − L < + = 
     
2 2

Example 1.2.2. Show that the Dirichlet function,


1 if x ∈ Q
f (x) =
0 if x ∈ R \ Q

is not Riemann integrable on [a, b] for all a, b ∈ R

Proof. We are going to use the converse of the Cauchy Criterion. Therefore we need to prove that: There
exists
  > 0 for
 every  δ > 0, there exists tagged partitions Ṗ and Q̇ with ||Ṗ || < δ and ||Q̇|| < δ such that
f ; Ṗ − S f ; Q̇  ≥ . Taking  = b−a 2 , we let Ṗ and Q̇ be partitions on [a, b]
 
S
For all n ∈ N,
n
Ṗ = {([xi−1 , xi ] , ti )}i=1 such that x0 = a, xn = b and ti ∈ Q for all i < n
and
n
Q̇ = {([xi−1 , xi ] , ti )}i=1 such that x0 = a, xn = b and ti ∈ R \ Q for all i < n
  Xn n
X
S f ; Ṗ = f (ti )(xi − xi−1 ) = (xi − xi−1 ) = (xn − xn−1 + xn−1 − xn−2 + · · · + x1 − x0 ) = (xn − x0 ) = b − a
i=1 i=1

While
  Xn
S f ; Q̇ = f (ti )(xi − xi−1 ) = 0
i=1

Then,
     b−a
S f ; Ṗ − S f ; Q̇  = |b − a| > =
 
2
Therefore, the Dirichlet function is not Riemann integrable on [a, b] for all a, b ∈ R
Theorem 1.2.3 (Squeeze Theorem). f : [a, b] → R, then f ∈ R[a, b] if and only if for all  > 0, there exists
functions g, h ∈ R[a, b] with
g(x) ≤ f (x) ≤ h(x) ∀x ∈ [a, b]
such that Z b
(h − g) < 
a
Properties of Riemann Integrable Functions 7

Proof. ( =⇒ ) We can let f (x) = g(x) = h(x) for all x ∈ [a, b] and let  > 0
( ⇐= ) Given  > 0, we choose functions g, h ∈ R[a, b] such that

g(x) ≤ f (x) ≤ h(x) ∀x ∈ [a, b] (1.7)

such that Z b

(h − g) < < (1.8)
a 3
As g, h ∈ R[a, b], there exists δ > 0 such that if any tagged partition Ṗ with ||Ṗ || < δ, then
   
   Z b      Z b  
S g; Ṗ − g  < and S h; Ṗ − h <
   
 a  3  a  3

It follows that Z b  Z b
    
g − < S g; Ṗ and S h; Ṗ < h+
a 3 a 3
     
Due to Equation 1.7, we know that S g; Ṗ ≤ S f ; Ṗ ≤ S h; Ṗ . Therefore,
Z b Z b
   
g− < S f ; Ṗ < h+
a 3 a 3

If we choose another tagged partition Q̇ with ||Q̇|| < δ, we also have

Z b  Z b
  
g − < S f ; Q̇ < h+
a 3 a 3
Then !
Z b  Z b
2    2
− (h − g) + < S f ; Ṗ − S f ; Q̇ < (h − g) +
a 3 a 3
So,
     Z b 2  2
f ; Ṗ − S f ; Q̇ < (h − g) + < + =
 
3 3 3
S
a
This satisfies the Cauchy Criterion (Theorem 1.2.1), therefore f ∈ R[a, b]
Theorem 1.2.4. If f : [a, b] → R is a step function, then f ∈ R[a, b]
Proof. Consider the function ϕA : [a, b] → R such that

1 if x ∈ A
ϕA (x) = where A = [c, d] for some c, d ∈ R with a < c < d < b
0 if x ∈ /A

Given  > 0, let δ > 0 such that if a tagged partition with ||Ṗ || < δ,
  Xn X
S ϕA ; Ṗ = ϕ(ti )(xi − xi−1 ) = (xi − xi−1 ) = d − c
i=1 x∈A

We claim that the integral of ϕA (x) is d − c. So,

   
S ϕA ; Ṗ − (d − c) = |(d − c) − (d − c)| = 0 < 
 

Therefore, function ϕA ∈ R[a, b].

To complete the proof, we write any step function f as
k
X
f (x) = ai ϕAi (x)
i=1

and by using Theorem 1.1.7, we are done.

8 Properties of Riemann Integrable Functions

Theorem 1.2.5. If f : [a, b] → R is continuous on [a, b], then f ∈ R[a, b]

Proof. As f is continuous on [a, b], then f is uniformly continuous over [a, b], i.e.

∀ > 0, ∃δ > 0, ∀x, y ∈ [a, b] : |x − y| < δ =⇒ |f (x) − f (y)| <
b−a
Let P be a partition and so as f is continuous on [a, b], for each subinterval [xi−1 , xi ], there exists ui , vi ∈
[xi−1 , xi ] such that for each ith subinterval, f (ui ) and f (vi ) has a minimum value and maximum value respec-
tively on [xi−1 , xi ] Then, we define a step function φ and Φ where

φ(x) = f (ui ) and Φ(x) = f (vi ) for all x ∈ [xi−1 , xi ]

Below is an illustration of the function f , φ, and Φ where the curve is the function f , the step function above
f is Φ and the step function below f is φ

Clearly, we have φ(x) ≤ f (x) ≤ Φ(x) for all x ∈ [a, b] So, for all  > 0, there exists δ > 0, such that if any
partition Ṗ with ||Ṗ || < δ, then

|xi − xi−1 | < δ for all i < n =⇒ |vi − ui | < δ

Then,
n n n
!
b
(b − a)
Z X X   X
Φ−φ= (f (vi ) − f (ui ))(xi − xi−1 ) < (xi − xi−1 ) = (xi − xi−1 ) = =
a i=1 i=1
b−a b−a i=1
b−a

Finally, we just apply the squeeze theorem (Theorem 1.2.3) to complete the proof.
Theorem 1.2.6. If f : [a, b] → R is monotone on [a, b], then f ∈ R[a, b]
As the proof is very similar to Theorem 1.2.5. The proof is omitted.

Theorem 1.2.7 (Additivity Theorem). Let f : [a, b] → R and c ∈ [a, b], then f is Riemann Integrable on [a, b]
if and only if f is Riemann Integrable on [a, c] and f is Riemann Integrable on [c, b] and so if the condition is
true,
Z b Z c Z b
f= f+ f
a a c
The Fundamental Theorem of Calculus 9

Rc
Proof. ( ⇐= ) As f ∈ R[a, b] with af = L 1 , then
 for all  > 0, there exists δ1 R> 0 such that if any tagged
b
partition P˙1 with ||P˙1 || < δ1 , then S f ; P˙1 − L1  < 2 Also, as f ∈ R[c, b] with c f = L2 , then for all  > 0,
 
   
there exists δ2 > 0 such that if any tagged partition P˙2 with ||P˙2 || < δ2 , then S f ; P˙2 − L2  < 2 . If M is a
 


bound for |f |, choose δ = min δ1 , δ2 , 2M , such that if a tagged partition Ṗ with ||Ṗ || < δ, we will prove that
   
S f ; Ṗ − (L1 + L2 ) < 
 

n
(i) Case 1: Let Ṗ = {([xi−1 , xi ] , ti )}i=1 . If c is a partition point of Ṗ (i.e. c = xi for some i < n), we
split the partitionto P˙
1 and P
˙
2 on  [a, c] and b] respectively. Then, since we know that ||P1 || < δ1 and
 [c, 
˙
||P2 || < δ2 and S f ; Ṗ = S f ; P1 + S f ; P2 , ˙
         
S f ; Ṗ − (L1 + L2 ) = S f ; P˙1 − L1 + S f ; P˙2 − L2 
   

         
≤ S f ; P˙1 − L1  + S f ; P˙2 − L2  < + = 
   
2 2
(ii) Case 2: If c is not a partition is a point of Ṗ , then there exists k < n such that c ∈ [xk−1 , xk ]. We consider
a tagged partition P˙ 0 with one more partition point than Ṗ at c, i.e.

P˙ 0 = {([x0 , x1 ] , t1 ) , . . . , ([xk−1 , c] , c) , ([c, xk ] , c) , . . . , ([xn−1 , xn ] , tn )}

So, for all  > 0, there exists δ > 0 such that if any partition P˙ 0 with ||P˙ 0 || < δ < 2M

, then
    
S f ; Ṗ − S f ; P˙ 0  = |f (tk )(xk − xk−1 ) − f (c)(xk − xk−1 )|
 

  
= |f (tk ) − f (c)| (xk − xk−1 ) < 2M (xk − xk−1 ) < 2M =
2M
This satisfies the Cauchy Criterion (Theorem 1.2.1) and so, f ∈ R[a, b]

( =⇒ ) As f ∈ R[a, b], given  > 0, let δ > 0 such that if tagged partitions Ṗ and Q̇ with ||Ṗ || < δ and
||Q̇|| < δ, then     
S f ; Ṗ − S f ; Q̇  < 
 

We consider P˙ 0 and Q̇0 on [a, c] with ||P˙ 0 || < δ and ||Q̇0 || < δ. By adding the same additional partition points
and tags in [c, b] to P˙ 0 and Q̇0 , we get Ṗ and Q̇ such that ||Ṗ || < δ and ||Q̇|| < δ and so
         
S f ; P˙ 0 − S f ; Q̇0  = S f ; Ṗ − S f ; Q̇  < 
   

This shows that it fulfills the Cauchy Criterion (Theorem 1.2.1). So f is Riemann Integrable on [a, c]
The same applies to [c, b] and so, we are done.

1.3 The Fundamental Theorem of Calculus

In this subsection, we are going to discuss the connection of the derivative an the integral. This subsection will
discuss mainly the two forms of the Fundamental Theorem of Calculus.
Definition 1.3.1. Let f : [a, b] → R be a function. A function F : [a, b] → R is called an antiderivative of f if
F 0 (x) = f (x) for all x ∈ [a, b]
Definition 1.3.2. If f ∈ R[a, b], then the function Fa : [a, b] → R defined by
Z x
Fa (x) = f for x ∈ [a, b]
a

is called the indefinite integral of f with basepoint a

Using this definition, we can define the Fundamental Theorem of Calculus.
10 The Fundamental Theorem of Calculus

Theorem 1.3.3 (Fundamental Theorem of Calculus (First Form)). Let f : [a, b] → R and f is Riemann
Integrable on [a, b]. Suppose that there is a function F : [a, b] → R such that:
(a) F is continuous on [a, b],
(b) F 0 (x) = f (x) for most x ∈ [a, b] except for a finite set of points,
Rb
Then, we have a f = F (b) − F (a)
Remarks:
1. The function F such that F 0 (x) = f (x) for all x ∈ [a, b] is called the antiderivative of x
2. We can permit a finite number of points c where F 0 (c) does not exists or F 0 (c) 6= f (c)
Proof. As f ∈ R[a, b], for all  > 0, there exists δ > 0 such that if any tagged partition Ṗ with ||Ṗ || < δ then
 
   Z b 
S f ; Ṗ − f < 
 
 a 

We let [xi−1 , xi ] be the ith subinterval of P. Then, by applying the Mean Value Theorem on F , for each i < n,
there exists ui ∈ (xi−1 , xi ) such that
F (xi ) − F (xi−1 )
= F 0 (ui ) = f (ui )
xi − xi−1
F (xi ) − F (xi−1 ) = f (ui )(xi − xi−1 )
th
By adding all the i terms, we get
n
X n
X
F (b) − F (a) = (F (xi ) − F (xi−1 )) = f (ui )(xi − xi−1 ) (1.9)
i=1 i=1
   R b 
n
Then, letting P˙u = {([xi−1 , xi ] , ui )}i=1 , Equation 1.9 is equal to S f ; P˙u So, we can conclude that, F (b) − F (a) − a f  <

Rb
 for all  > 0. But, since we can make  > 0 as small as possible, we infer that F (b) − F (a) = a f
Example 1.3.4. Let a, b ∈ R such that a < 0 < b. Let function F : [a, b] → R such that F (x) = |x|. Then let
f : [a, b] → R such that 
 1 if x > 0
f (x) = 0 if x = 0
−1 if x < 0


Proof. We can see that F 0 (x) = f (x) for all ∈ [a, b] \ {0} as F (x) is not differentiable at 0. However, the
Fundamental Theorem of Calculus still applies.

a b
Lebesgue’s Integrability Criterion 11

Rb
We can see that the function f (x) is a step function and so a
f = b + a and F (b) − F (a) = |b| − |a| = b + a
Rb
and therefore, a f = F (b) − F (a)
Theorem 1.3.5 (Fundamental Theorem of Calculus (Second Form)). Let f ∈ R[a, b], and let f be continuous
at c ∈ [a, b]. Then the indefinite integral
Z c
F (c) = f for c ∈ [a, b]
a

is differentiable at c and F 0 (c) = f (c).

Proof. Suppose c ∈ (a, b) consider the right hand derivative of F at c, since f is continuous at c, given  > 0,
there exists δ > 0 such that if c ≤ x ≤ c + δ, then

|f (x) − f (c)| <  =⇒ f (c) −  < f (x) < f (c) +  (1.10)

Let h satisfy 0 < h < δ, the Additivity Theorem (Theorem 1.2.7) implies that f is Riemann Integrable on
[a, c], [a, c + h] and [c, c + h] and that
Z c+h Z c+h Z c
f= f− f = F (c + h) − F (c)
c a a

Now on [c, c + h], f satisfy Equation 1.10. So we have

Z c+h Z c+h Z c+h
(f (c) − ) · h = (f (c) − ) ≤ F (c + h) − F (c) = f≤ (f (c) + ) = (f (c) + ) · h
c c c

F (c + h) − F (c)
=⇒ − ≤ − f (c) ≤ 
h
 
 F (c + h) − F (c) 
=⇒  − f (c) ≤ 
h
But, since we can choose  > 0 to be very small, we can infer that

F (c + h) − F (c)
lim = F 0 (c) = f (c)
h→0+ h
By proving using the similar method for the left hand side, we conclude that F 0 (c) = f (c).

1.4 Lebesgue’s Integrability Criterion

In this subsection, a necessary and sufficient condition for a function to be Riemann Integrable and some
application will be given.
Before starting with the Lebesgue’s Integrability Criterion, we start by defining the null set

Definition 1.4.1 (Lebesgue Outer Measure). Let A ⊆ R and let l((a, b)) = b − a for all open interval (a, b).
Then the Lebesgue Outer Meausure of A µ∗ (A) is defined to be
(∞ ∞
)
X [
µ∗ (A) = inf l(Ik ) : {Ik } is a sequence of open intervals in which A ⊆ Ik
k=1 k=1

Remarks:
1. We can see here that 0 ≤ µ∗ (A) ≤ ∞ for all A ⊂ R
Definition 1.4.2 (Null Set). A set A ⊂ R is a null set if for all  > 0, µ∗ (A) ≤ 
Example 1.4.3. All countable sets are null sets.
12 Lebesgue’s Integrability Criterion

Proof. Let set A = {xk : k ∈ N} be a countably infinite set. Then, we define the sequence of open interval {Ik }
to be    
Ik = xk − k+1 , xk + k+1 for all k ∈ N
2 2
Then,
∞ ∞
X X 
l(Ik ) = =
2k
k=1 k=1
∗ ∗
As µ (A) is the infimum, we have µ (A) < 
Example 1.4.4. The set of rational numbers Q is a null set.
Proof. We start by denoting that Q = {rk : k ∈ Z}. Then we can use a similar proof with the proof in Example
1.4.3
Definition 1.4.5. If P (x) is a statement about the point x ∈ [a, b], we say that P (x) holds almost everywhere
on I if there exists a null set Z ⊂ I such that P (x) holds for all x ∈ I \ Z or we may write

P (x)∀∞ x ∈ I

Now, we are ready to introduce the Lebesgue’s Integrability Criterion

Theorem 1.4.6 (Lebesgue’s Integrability Criterion). A bounded function f : [a, b] → R is Riemann Integrable
if and only if f is continuous almost everywhere on [a, b]
Proof. The proof to this theorem is omitted.
Example 1.4.7 (Thomae Function). The Thomae Function is integrable on [0, 1]. The Thomae Function is
defined as
if x ∈ Q such that x = pq in lowest terms and q > 0
 1
f (x) = q
0 if x ∈ R \ Q
Proof. Here f (x) is continuous for all x ∈ R \ Q and discontinuous for all x ∈ Q
x is discontinuous at x ∈ Q. Let a ∈ Q, let {an } be a sequence such that an ∈ R \ Q for all n ∈ N and
{an } → a, then
lim f (an ) = 0 6= f (a)
n→∞

x is continuous at x ∈ R \ Q. By Archimedean Property, let b ∈ R \ Q, then there exists n such that n1 < .
Consider the interval (b − 1, b + 1), we can say that there is only a finite number of rationals with denominator
less than n. So we can choose δ > 0 small enough such that (b − δ, b + δ) does not contain a rational number
with denominator less than n.
It follows that |x − b| < δ where x ∈ A. We have |f (x) − f (b)| = |f (x)| ≤ n1 < . So f (x) is continuous for
all x ∈ R \ Q.
So the set of all discontinuous points is S = Q ∩ [0, 1] and so as the set is countably infinite, then S is a null
set and by the Lebesgue Integrability Criterion, the Thomae function is Riemann Integrable on [0, 1].
Theorem 1.4.8 (Compostition Theorem). Let f ∈ R[a, b] with f ([a, b]) ⊆ [c, d] and let g : [c, d] → R be
continuous. Then, g ◦ f ∈ R[a, b]
Remarks:
1. The condition g is continuous can not be dropped
Proof. We can split this proof into two cases
(i) Case 1. If f is continuous at [a, b], then g ◦ f is also continuous at [a, b]. This implies that g ◦ f is Riemann
integrable on [a, b]
(ii) Case 2. If f has a finite number of discontinuous points in [a, b], then as f satifies the Lebesgue’s
Integrability Theorem (Theorem 1.4.6), then the set S = {u : f is discontinuous at u} is a null set as it is
countable (By example 1.4.3. It follows that the set S 0 ⊆ S where S 0 = {u : g ◦ f is discontinuous at u}.
Therefore, by Lebesque’s Integrability Theorem (Theorem 1.4.6), g ◦ f is Riemann Integrable on [a, b]
Substitution Theorem and Integration by Parts 13

Both cases implies that g ◦ f is Riemann integrable on [a, b]

Example 1.4.9. Let 
 1 if x > 0
f (x) = 0 if x = 0
−1 if x < 0


and p
 1
q if x ∈ Q such that x = q in lowest terms and q > 0
g(x) =
0 if x ∈ R \ Q
We already know that f, g ∈ R[0, 1] Show that f ◦ g(x) is not Riemann Integrable on [0, 1]
Proof. We can see that g(x) > 0 for all x ∈ Q and g(x) = 0 for all x ∈ R \ Q. So

1 if x ∈ Q
f ◦ g(x) =
0 x∈R\Q

which we know that it is not Riemann Integrable

Theorem 1.4.10 (Product Theorem). If f and g is Riemann integrable on [a, b], then the product f g is also
Riemann integrable on [a, b].
Proof. For each function f : [a, b] → R, we consider the function h(x) = x2 for x ∈ [−M, M ] where [−M, M ] is
the bounds of the function f . Then by the composition theorem (Theorem 1.4.8), we know that f 2 = h ◦ f is
Riemann Integrable on [a, b]. Therefore, the functions f 2 , g 2 and (f + g)2 are Riemann Integrable on [a, b]. To
complete the proof, we simply write f g as
1h 2
i
fg = (f + g) − f 2 − g 2
2
which shows that f g is Riemann Integrable on [a, b]

1.5 Substitution Theorem and Integration by Parts

Theorem 1.5.1 (Substitution Theorem). Let α : [c, d] → R be differentiable at [c, d] with its derivative α0 being
continuous on [c, d]. If f : [a, b] → R with a ≤ c ≤ d ≤ b and α ([c, d]) ⊆ [a, b], then
Z d Z α(d)
f (α (x)) · α0 (x)dx = f (α)dα
c α(c)

Proof. Consider Z u
F (u) = f (x)dx for u ∈ [a, b]
α(c)

By the Fundamental Theorem of Calculus, F 0 (u) = f (u) Let H(x) = F (α(x)) Also by the Fundamental
Theorem of Calculus,
Z d
H(d) = H 0 (d)dx
c
Then,
Z α(d)
f (α)dα = F (α(d)) = H(d)
α(c)

To complete the proof, we just need to find H 0 (d). As f and α are both differentiable on [a, b] and [c, d]
respectively, then
H 0 (d) = F 0 (α(x)) · α0 (x) = f (α(x)) · α0 (x)
Therefore,
Z d Z α(d)
f (α(x)) · α0 (x)dx = f (α)dα
c α(c)
14 Substitution Theorem and Integration by Parts

Theorem 1.5.2 (Integration by Parts). Let F and G be differentiable on [a, b], let f = F 0 and g = G0 with
f, g ∈ R[a, b]. Then
Z b Z b
b
f G = [F G]a − Fg
a a

Proof. By Product Rule of Differentiation, as F and G are differentiable,

(F G)0 = F 0 G + F G0 = f G + F g

As all F, G, f, g ∈ R[a, b], then f G, F g ∈ R[a, b]. So, by using the Fundamental Theorem of Calculus,
Z b Z b Z b
b 0
[F G]a = (F G) = fG + Fg
a a a

Therefore,
Z b Z b
b
f G = [F G]a − Fg
a a

Theorem 1.5.3 (Equivalence Theorem). A function f : [a.b] → R is Darboux Integrable if and only if it is
Riemann Integrable
Proof. ( =⇒ ) Assume f is Darboux Integrable on [a, b]. Then, for all  > 0, let P be a partition on [a, b] such
that U (f ; P ) − L(f ; P ) < . We then define the step function φ and Φ in which

φ(x) = mk for x ∈ [xk−1 , xk ) ∀k < n

and
Φ(x) = Mk for x ∈ [xk−1 , xk ) ∀k < n
So we have
φ(x) ≤ f (x) ≤ Φ(x) ∀x ∈ [a, b]
As φ, Φ are step functions, it is Riemann Integrable and moreover,
Z b n
X
Φ= Mk (xk − xk−1 ) = U (f ; P )
a k=1

Z b n
X
φ= mk (xk − xk−1 ) = L (f ; P )
a k=1

Therefore, we have
Z b
Φ − φ = U (f ; P ) − L(f ; P ) < 
a
Rb
and so by squeeze theorem, f ∈ R[a, b]. As a f = L(f ; P ) = U (f ; P ) and so if a tagged partition Q̇ with
||Q̇|| < δ, then  
S(f ; Q̇) − U (f ; P ) < |U (f ; P ) − L(f ; P )| < 
 

( ⇐= ) Assume f is Riemann Integrable on [a, b]. By Cauchy criterion, for all  > 0, there exists δ > 0 such
that for all tagged partitions ṖM and Ṗm , if ||ṖM || < δ and ||Ṗm || < δ, then
     
S f ; ṖM − S f ; Ṗm  <
 
3
n
We then fix partition P = {[xi−1 , xi ]}i=1 .
For all i < n, consider set Ai = {f (x) : x ∈ [xi−1 , xi ]}. Let Mi be sup Ai , as Mi is a supremum,

∀ > 0, choose tags ui ∈ [xi−1 , xi ] such that Mi < f (ui ) +
3n(xi − xi−1 )
Substitution Theorem and Integration by Parts 15

This implies,

Mi (xi − xi−1 ) < f (ui )(xi − xi−1 ) +
3n
n n 
X X  
Mi (xi − xi−1 ) < f (ui )(xi − xi−1 ) +
i=1 i=1
3n
and so,

U (f ; P ) < S(f ; ṖM ) +
3
n
where ṖM = {([xi−1 , xi ] , ui )}i−1
Similarly, let mi be inf Ai ,

∀ > 0, choose tags vi ∈ [xi−1 , xi ] such that mi > f (vi ) −
3n(xi − xi−1 )

This implies

mi (xi − xi−1 ) > f (vi )(xi − xi−1 ) −
3n
n n
X X 
mi (xi − xi−1 ) > (f (vi )(xi − xi−1 ) − )
i=1 i=1
3n
and so,

L(f ; P ) > S(f ; Ṗm ) −
3
n
where ṖM = {([xi−1 , xi ] , vi )}i−1
Therefore,
             2  2
U (f ; Ṗ ) − L(f ; Ṗ ) < S f ; ṖM − S f ; Ṗm + 2 < S f ; ṖM − S f ; Ṗm  + < + =
     
3 3 3
16 Definition and Main Properties

2 The Generalised Riemann Integral/Henstock-

Kurzweil Integral
The Riemann Integral is a powerful tool and has many applications. However, it was found out later that it
was not inadequate. For example, the Fundamental Theorem of Calculus
Z b
F 0 = F (b) − F (a)
a

does not hold for all differentiable function.

To tackle with these inadequacies, we use a slightly different approach, to arive at an integral called the
Generalised Riemann Integral.

2.1 Definition and Main Properties

Definition 2.1.1. A gauge δ on [a, b] is a strictly positive function δ : [a, b] → (0, ∞)
n
Definition 2.1.2. A tagged partition Ṗ = {([xi−1 , xi ] , ti )}i=1 of [a, b] is said to be δ-fine if

ti ∈ [xi−1 , xi ] ⊆ [ti − δ(ti ), ti + δ(ti )] for all i ∈ N, i < n

ti − δ(ti ) xi−1 ti xi ti + δ(ti )

Remarks:

1. δ(ti ) is dependent on ti , so for every ti , it is possible to have a different value of δ

The definition of the Generalised Riemann Integral is very similar to the ordinary Riemann Integral and
many of the proofs are similar. However, the slight change results in a much larger set of integrable function.
Definition 2.1.3. A function f : [a.b] → R is said to be Generalised Riemann Integrable on [a, b] if there exists
a number L ∈ R such that for every  > 0, there exists a gauge δ on [a, b] such that if Ṗ is a δ-fine partition of
[a, b], then    
S f ; Ṗ − L < 
 

Remarks:
1 The use of gauge allow more control on the length of the subinterval than using the norm (we can put
smaller subinterval when the function is rapidly increasing and a larger subinterval when the function is
nearly constant).
2 The set of all Generalised Riemann Integrable function is denoted by R∗ [a, b]

3 The Generalised Riemann Integral of f ∈ R[a, b] is also denoted by

Z b Z b
f or f (x)dx
a a

Theorem 2.1.4 (Uniqueness Theorem). If f ∈ R∗ [a, b], then the value of the integral is uniquely defined.
Remarks:
1 The proof below is very similar to the Uniqueness Theorem of the Riemann Integral
Definition and Main Properties 17

Proof. Assume L1 and L2 both satisfy Definition 2.1.3 with L1 6= L2 and without loss of generality, assume
L2 > L1 . Let  > 0, then there exists gauge δ1 such that if P˙1 is a δ1 -fine partition, then
   
S f ; P˙1 − L1  < 
 

Similary, there exists δ2 such that if P˙2 is a δ2 -fine partition, then

   
S f ; P˙2 − L2  < 
 

Define δ(t) = min {δ1 (t), δ2 (t)} for each t ∈ [a, b] so that δ gauge on [a, b]. If Ṗ is a δ-fine partition, the Ṗ is
both δ1 -fine and δ2 -fine. So we have    
S f ; Ṗ − L1  <  (2.1)
 
   
S f ; Ṗ − L2  <  (2.2)
 

L2 −L1
We then let  = 2 > 0. Then from Equation 2.1, we have

3L1 − L2   L +L
1 2
< S f ; Ṗ <
2 2
and from Equation 2.2, we have
L1 + L2   3L − L
2 1
< S f ; Ṗ <
2 2
We then get
3L1 − L2   L +L
1 2
  3L − L
2 1
< S f ; Ṗ < < S f ; Ṗ <
2 2 2
which is impossible therefore we have a contradiction that L1 6= L2 and so the value of the integral is unique.
Theorem 2.1.5 (Consistency Theorem). Let f : [a, b] → R. If f is Riemann Integrable on [a, b], then f is
Generalised Riemann Integrable on [a, b] and their integrals are the same, i.e.
Z b Z b
Riemann f = Generalised Riemann f
a a

Proof. Given  > 0, we are going to find an appropriate gauge on [a, b]. Since f ∈ R[a, b], there exists δ > 0
such that if any tagged partition Ṗ with ||Ṗ ||, then
   
S f ; Ṗ − L < 
 

We let a function δ 0 : [a, b] → (0, ∞) such that δ 0 (t) = 41 δ for all t ∈ [a, b] so that δ 0 can be a gauge on [a, b]. If
n
Ṗ = {([xi−1 , xi ] , ti )}i=1 is a δ 0 -fine partition, then
 
1 1
Ii ⊆ [ti − δ 0 (ti ), ti + δ 0 (ti )] = ti − δ, ti + δ
4 4

As
1 1 1
0 < xi − xi−1 < ti + δ − ti + δ = δ < δ for all i ∈ N, i < n
4 4 2
We can see that Ṗ also satisfies ||Ṗ || < δ and as f is Riemann Integrable, the statement
   
S f ; Ṗ − L < 
 

must be true. So, we can see that f is also Generalised Riemann Integrable. Therefore, by choosing a constant
function as a gauge, we can see that all Riemann Integrable function are Generalised Riemann Integrable.
We know that from Example 1.2.2 that the Dirichlet function is not Riemann Integrable, however we are
going to prove shortly that it is Generalised Riemann Integrable.
18 Definition and Main Properties

Example 2.1.6. The Dirichlet function


1 if x ∈ Q
f (x) =
0 if x ∈ R \ Q
R1
is Generalised Riemann Integrable on [0, 1] and 0
f =0

Proof. We define the set Q ∩ [0, 1] as Q = {r1 , r2 , . . . , rn } for all n ∈ N. Given  > 0, define δ(rn ) = 2n+2 and
n
δ(x) = 1 when x ∈ R \ Q. So δ is a gauge on [0, 1] and if any partition Ṗ = {([xi−1 , xi ] , ti )}i=1 is δ-fine, then
 xi − xi−1 < 2δ(ti ). As f (x) = 0 if x is irrational, we only consider the sum of all the rational tags to
we have
find S f ; Q̇ .
2 
0 < f (rn )(xi − xi−1 ) = (xi − xi−1 ) < =
2n+2 2n+1
Since at most 2 subinterval can have the same tags, we have
∞ ∞
  X 2 X 
0 ≤ S f ; Q̇ < n+1
= n
=
n=1
2 n=1
2
 
0 ≤ S f ; Q̇ < 
   
S f ; Q̇ − 0 < 
 

R1
Therefore, f is Generalised Riemann Integrable and 0
f =0
Theorem 2.1.7. Suppose f, g ∈ R∗ [a, b]
a. If k ∈ R, then kf ∈ R∗ [a, b] and
Z b Z b
kf = k f
a a

b. f + g ∈ R∗ [a, b] and
Z b Z b Z b
(f + g) = f+ g
a a a

c. If f (x) ≤ g(x) for all x ∈ [a, b], then

Z b Z b
f≤ g
a a

Proof. As the proof the similar to Theorem 1.1.7, the proof is omitted
Below is the Cauchy Criterion for Generalised Riemann Integral, the proof is similar with the Cauchy
Criterion for Riemann Integral (Theorem 1.2.1)
Theorem 2.1.8 (Cauchy Criterion). A function f : [a, b] → R is Generalised Riemann Integrable if and only
if for every  > 0, there exists a gauge δ on [a, b] such that if any tagged δ-fine partition Ṗ , Q̇, then
    
S f ; Ṗ − S f ; Q̇  < 
 

Rb
Proof. ( =⇒ ) As f ∈ R∗ [a, b], let a f . If we choose a gauge δ 0 such that Ṗ and Q̇ are δ 0 -fine partition of [a, b],
then          
S f ; Ṗ − L < and S f ; Q̇ − L <
   
2 2
We set δ(t) = δ 0 (t) for all t ∈ [a, b], so if Ṗ and Q̇ are δ-fine, then
              
S f ; Ṗ − S f ; Q̇  ≤ S f ; Ṗ − L + S f ; Q̇ − L < + = 
     
2 2
And so we are done.
Definition and Main Properties 19

( ⇐= ) For every  > 0, there exists a gauge δ on [a, b] such that if any tagged partition Ṗ , Q̇ are δ-fine,
then      
S f ; Ṗ − S f ; Q̇  <
 
2
Let P˙n be a sequence of δ-fine partitions. So, we choose m, n ∈ N and WLOG, we assume m > n. Clearly, P˙n
and P˙m are δ-fine, which in turn implies
     
S f ; P˙n − S f ; P˙m  <
 
2

This shows that S(f ; P˙n ) is a cauchy sequence and in turn, converges to some number L ∈ R. Or in other
words,     
∀ > 0, ∃N ∈ N : n > N =⇒ S f ; P˙n − L <
 
2
Rb
Finally, to show that f ∈ R∗ [a, b] and that a f = L, given  > 0, we let δ be a gauge and n > N such that if
a tagged partition Q̇ is δ-fine, then
              
S f ; Q̇ − L < S f ; Q̇ − S f ; P˙n  + S f ; P˙n − L < + = 
     
2 2
Rb
Therefore, the function f is Generalised Riemann Integrable on [a, b] with integral a = L.

Theorem 2.1.9 (Squeeze Theorem). Let f : [a, b] → R. Then f ∈ R∗ [a, b] if and only if for every  > 0, there
exists functions g, h ∈ R∗ [a, b] with

g(x) ≤ f (x) ≤ h(x) for all x ∈ [a, b]

Rb
and such that a
(h − g) ≤ .
Proof. The proof is similar to the proof in Theorem 1.2.3 and so will be omitted.
Theorem 2.1.10 (Additivity Theorem). Let f : [a, b] → R, let c ∈ [a, b]. Then f ∈ R∗ [a, b] if and only if f is
Generalised Riemann Integrable on both [a, c] and [c, b] and
Z b Z c Z b
f= f+ f
a a c
Rc Rb
Proof. ( ⇐= ) Suppose a f = L1 and c f = L2 . Then given  > 0, there exists a gauge δ1 on [a, c] such that
if a tagged partition P˙1 is δ1 -fine on [a, c], then
  
S(f1 ; P˙1 ) − L1  <
 
2

Similarly, there exists a gauge δ2 on [c, b] such that if a tagged partition P˙2 is δ2 -fine on [c, b], then
    
S f2 ; P˙2 − L2  <
 
2
Define gauge δ on [a, b] by 
 min δ1 (t), 21 (c − t)

if t ∈ [a, c)
δ(t) = min {δ
 1 (c), δ2 (c)} if t = c
min δ2 (t), 21 (c − t) if t ∈ (c, b]


Using this gauge will force the δ-fine partition to have c has a tag for any subinterval containing c. We will show
that if Q̇ is a δ-fine partition of [a, b], then there exists a δ1 -fine partition Q̇1 on [a, c] and a δ2 -fine partition
Q̇2 on [c, b] such that      
S f ; Q̇ = S f1 ; Q̇1 + S f2 ; Q̇2 (2.3)
20 Definition and Main Properties

(i) Case 1. If c is a partition point on Q̇, then it belongs to two subinterval of Q̇ and is the tag for both
subinterval. Let Q̇1 be the part of Q̇ in [a, c] and so Q̇1 is δ1 -fine. Similarly, let Q̇2 be the part of Q̇ in
[c, b] and so Q̇2 is δ2 -fine. So then
  Xn X X
S f ; Q̇ = f (ti )(xi − xi−1 ) = f (ti )(xi − xi−1 ) + f (ti )(xi − xi−1 )
i=1 xi ∈[a,c] xi ∈[c,b]

   
= S f ; Q̇1 + S f ; Q̇2

n
(ii) Case 2. If c is not a partition point in Q̇ = {([xi−1 , xi ] , ti )}i=1 , then as Q̇ is δ-fine, c is forced to have c
as its tags in one of its intervals, say [xk−1 , xk ]. Then as

f (c)(xk − xk−1 ) = f (c)(c − xk−1 ) + f (c)(xk − c),

we can split subinterval ([xk−1 , xk ] , c) to ([xk−1 , c] , c) and ([c, xk ] , c) and so using similar methods with
case (i), we split Q̇ to Q̇1 , a δ1 -fine partition in [a, c] and Q̇2 , a δ2 -fine partition in [c, b] and so
     
S f ; Q̇ = S f ; Q̇1 + S f ; Q̇2

Using Equation 2.3, and the triangle inequality,

         
S f ; Q̇ − (L1 + L2 ) = S f ; Q̇1 − L1 + S f ; Q̇2 − L2 
   

         
≤ S f ; Q̇1 − L1  + S f ; Q̇2 − L2  < + = 
   
2 2
Therefore, we conclude that f ∈ R∗ [a, b] and
Z b Z c Z b
f= f+ f
a a c

( =⇒ ) Let f ∈ R∗ [a, b] and given  > 0, there exists a gauge δ such that if any tagged partitions Ṗ , Q̇ are
δ-fine, then     
S f ; Ṗ − S f ; Q̇  < 
 

Then let P˙ 0 , Q̇0 be δ-fine partitions on [a, c]. By adding the same partition points and tags on [c, b] to both P˙ 0
and Q̇0 , we can extend both partitions to δ-fine partitions Ṗ and Q̇ of [a, b] and
       
S f ; Ṗ − S f ; Q̇ = S f ; P˙ 0 − S f ; Q̇0

So          
S f ; P˙ 0 − S f ; Q̇0  = S f ; Ṗ − S f ; Q̇  < 
   

and so the partitions P˙ 0 and Q̇0 of [a, c] satisfies the Cauchy Criterion. We then repeat for the interval [c, b]
and so we have proved that f ∈ R∗ [a, c] and f ∈ R∗ [c, b]. To prove that
Z b Z c Z b
f= f+ f
a a c

we use the fact that f ∈ R∗ [a, c] with integral L1 and that f ∈ R∗ [c, b] with integral L2 and we apply the
Rb Rc Rb
( ⇐= ) part of the proof that a f = L1 + L2 = a f + c f
We will now begin with the Fundamental Theorem of Calculus for Generalised Riemann Integral. Notice
that the first form is stronger than that of the ordinary Riemann Integral.
Theorem 2.1.11 (Fundamental Theorem of Calculus(First Form)). Suppose there exists a countable set E ⊆
[a, b], a function f, F : [a, b] → R∗ such that
Definition and Main Properties 21

1) F is continuous on [a, b].

2) F 0 (x) = f (x) for all x ∈ [a, b] \ E (i.e. F 0 (x) = f (x) for almost all x ∈ [a, b] except for a finite number of
points contained in set E).
Then f ∈ R∗ [a, b] and
Z b
f = F (b) − F (a)
a

Remarks:
1. Here, compared to the Fundamental Theorem of Calculus for ordinary Riemann Integral, f ∈ R∗ [a, b] is
a conclusion instead of a condition.
Proof. We will proof for the case E = ∅, so we assume (2) holds for all x ∈ [a, b]. Given  > 0, we use gauge
δ(t) > 0 for t ∈ [a, b] such that if 0 < |z − t| ≤ δ(t) for z ∈ [a, b], then
 
 F (z) − F (t)  
 − f (t) <
 z−t 2(b − a)

Therefore, we have
 |z − t|
|F (z) − F (t) − f (t)(z − t)| ≤
2(b − a)
for z ∈ [t − δ(t), t + δ(t)] ∩ [a, b].
Now, for all subinterval ([xi−1 , xi ], ti ) of δ(t)-fine partition, we have ti ∈ [xi−1 , xi ] ⊆ [ti − δ(ti ), ti + δ(ti )].
If we consider
|F (xi ) − F (xi−1 ) − f (ti )(xi − xi−1 )| ,
then we get
|F (xi ) − F (ti ) − f (ti )(xi − ti ) + F (ti ) − F (xi−1 ) + f (ti )(ti − xi−1 )|
≤ |F (xi ) − F (ti ) − f (ti )(xi − ti )| + |F (xi−1 ) − F (ti ) − f (ti )(xi−1 − ti )|
 |xi − ti |  |xi−1 − ti | (xi − ti ) (ti − xi−1 ) (xi − xi−1 )
≤ + = + =
2(b − a) 2(b − a) 2(b − a) 2(b − a) 2(b − a)
So if ti ∈ [xi−1 , xi ] ⊆ [ti − δ(ti ), ti + δ(ti )], then

(xi − xi−1 )
|F (xi ) − F (xi−1 ) − f (ti )(xi − xi−1 )| ≤
2(b − a)
n
Let partition Ṗ = {([xi−1 , xi ] , ti )}i=1 to be δ-fine, then ti ∈ [xi−1 , xi ] ⊆ [ti − δ(ti ), ti + δ(ti )] for all i ≤ n, i ∈ N.
Then
Xn
(F (xi ) − F (xi−1 )) = F (b) − F (a)
i=1

So,
 
       X n 
S f ; Ṗ − (F (b) − F (a)) = F (b) − F (a) − S f ; Ṗ  =  (F (xi ) − F (xi−1 ) − f (ti )(xi − xi−1 ))
     
 
i=1

n n  
X X  1 (b − a)
≤ |F (xi ) − F (xi−1 ) − f (ti )(xi − xi−1 )| ≤ (xi − xi−1 ) < =
i=1 i=1
b−a 2 b−a
Rb
Therefore, we conclude that f ∈ R∗ [a, b] and a
f = F (b) − F (a).
Theorem 2.1.12 (Fundamental Theorem of Calculus(Second Form)). Let f ∈ R∗ [a, b]. Let F be the indefinite
integral of f such that Z x
F (x) = f (t)dt for x ∈ [a, b]
a
Then we have
22 Definition and Main Properties

1) F is continuous on [a, b].

2) There exists a null set E such that if x ∈ [a, b] \ E, then F is differentiable at x and F 0 (x) = f (x) (i.e.
F is differentiable at x for almost all x ∈ [a, b] with F 0 (x) = f (x) except for a finite number of points
contained in set E).
3) If f is continuous at c ∈ [a, b], then F 0 (c) = f (c)
Proof. Proof omitted

Theorem 2.1.13 (Substitution Theorem). Let I = [a, b] and J = [c, d]. Let F : I → R and α : J → R be
continuous function with α(J) ⊆ I.
Suppose there exists Ef ⊂ I and Eα ⊂ J such that f (x) = F 0 (x) for x ∈ I \ Ef and α0 (t) exists for t ∈ J \ Eα ,
and that E = α−1 (Ef ) ∪ Eα is countable. If we fix f (x) = 0 for all x ∈ Ef and α0 (t) = 0 for t ∈ Eα .
We can conclude that f ∈ R∗ (α(J)), that (f ◦ α) · α0 ∈ R∗ (J) and that
Z d Z α(d)
0 d
(f ◦ α)α = (F ◦ α)|c = f
c α(c)

Proof. Since α is continuous on J, then the range of α, α(J), is a closed interval in I. Also, α− 1(Ef ) is
countable, so Ef ∩ α(J) = α(α−1 (Ef )) is also countable. As f (x) = F 0 (x) for all x ∈ α(J) \ Ef , then the
Fundamental Theorem of Calculus implies that f ∈ R∗ (α(J)) and
Z α(d)
α(d)
f = F |α(c) = F (α(d)) − F )α(c))
α(c)

If t ∈ J \ E, then t ∈ J \ Eα and α(t) ∈ I \ Ef , then the Chain Rule implies that

(F ◦ α)0 (t) = F 0 (α(t)) · α0 (t) = f (α(t)) · α0 (t) for t ∈ J \ E

Since E is countable, the Fundamental Theorem of Calculus implies that (f ◦ α) · α0 (t) ∈ R∗ (J) and
Z d
d
(f ◦ α) · α0 = F ◦ α|c = F (α(d)) − F (α(c))
c

Theorem 2.1.14 (Multiplication Theorem). If f ∈ R∗ [a, b] and if g : [a, b] → R is monotone, then f · g ∈

R∗ [a, b]
Remarks:
1. Note that the product of Generalised Riemann Integrable functions is not necessarily Generalised Riemann
Integrable.

Proof. Proof omitted.

Theorem 2.1.15 (Integration By Parts). Let F and G be differentiable on [a, b], then F 0 G ∈ R∗ [a, b] if and
only if F G0 ∈ R∗ [a, b]. We have
Z b Z b
0 b
F G = F G|a − F G0
a a

Proof. Proof omitted.

Improper and Lebesgue Integral 23

2.2 Improper and Lebesgue Integral

Consider h : [0, 1] → R such that
√1

x
if 0 < x ≤ 1
h(x) =
0 if x = 0
We can see that it is unbounded on the left end point. But, h is still belongs to R[γ, 1] for every γ ∈ [0, 1]. So,
we define the improper Riemann Integral of h on [0,1] to be
Z 1 Z 1
1 1
√ dx = lim √ dx
0 x γ→0+
γ x

We also have the same situation with k : [0, 1] → R with

sin x1


if 0 < x ≤ 1
k(x) =
0 if x = 0

Note that here we are not necessarily dealing with Generalised Riemann Integral in these cases.
Theorem 2.2.1 (Hake’s Theorem). If f : [a, b] → R, then f ∈ R∗ [a, b] if and only if for all γ ∈ [a, b],
f ∈ R∗ [a, γ] and Z γ
lim− f =A∈R (2.4)
γ→b a
Rb
In this case, a
f = A.
Remarks:
1. The theorem implies that the Generalised Riemann Integral cannot be extended by taking limits. So if
Equation 2.4 holds, f already belongs to R∗ [a, b].
2. We can know whether a function is integrable on [a, b] just by looking on subinterval [a, γ] with γ < b.
So this is just another tool to show that a function is Generalised Riemann Integral on [a, b].
3. We can use Equation 2.4 to evaluate the integral of a function.
P∞
Example 2.2.2. Let k=1 ak such ak ∈ R for all k ∈ N be a series converging to A ∈ R. Let f : [0, 1] → R
such that Z 1 X∞
f= ak = A
0 k=1

and
2k ak

if ck−1 ≤ x < ck 1
f (x) = where ck = 1 − for all k ∈ N,
0 if x = 1 2k
then f ∈ R∗ [0, 1].
Proof. Clearly, we can see that f on [0, γ] for γ ∈ [0, 1] is a step function and so is integrable. So, by summing
the areas,
Z γ n n   n n
X X 1 1 X 1 X
f= 2k ak (ck − ck−1 ) + rγ = 2k ak − k + k−1 + rγ = 2k ak ( k ) + rγ = ak + rγ
0 2 2 2
k=1 k=1 k=1 k=1

where |rγ | ≤ |an+1 |. But as the series is convergent, lim an = 0 and so rγ → 0 as n becomes large enough and
n→∞
so,
Z γ Xn
lim− f = lim ak = A
γ→1 0 n→∞
k=1

So f ∈ R∗ [a, b].
P∞
Example 2.2.3. We will use k=1 ak and f from Example 2.2.2.
P∞ P∞
(a) If k=1 |ak | converges ( k=1 ak converge absolutely), then |f | converge.
24 Improper and Lebesgue Integral

P∞ P∞
(b) If |ak | does not converges ( k=1 ak converge conditionaly), then |f | does not converge.
k=1
P∞
Proof. We just apply Example 2.2.2 to k=1 |ak | and |f |.
Below, we are going to give the definition of a class of Riemann Integrable function - The Lebesgue Integrable
Function.
Definition 2.2.4 (Lebesgue Integrable Functions). Let f ∈ R∗ [a, b] such that |f | ∈ R∗ [a, b]. Then f is said to
be Lebesgue Integrable on [a, b]. The collection of all Lebesgue Integrable functions on [a, b] is denoted by L[a, b]
Remark:
1. Note that this is not the standard definition of Lebesgue Integrable Functions. However, by using the set
of Generalised Riemann Integrable functions, it is easier to prove if a function is Lebesgue Integrable.
Riemann/Generalised Riemann Integral Lebesgue Integral

Theorem 2.2.5 (Comparison Test). If f, g ∈ R∗ [a, b] and |f (x)| ≤ g(x) for all x ∈ [a, b], then f ∈ L[a, b] and
Z 
 b  Z b Z b
f ≤ |f | ≤ g (2.5)
 

 a  a a

Proof. We are going to omit the proof for f ∈ L[a, b] and we are just going to prove Equation 2.5. So we note
that
− |f | ≤ f ≤ |f | .
Theorem 2.1.7 implies that
Z b Z b Z b
− |f | ≤ f≤ |f |
a a a
and so Z 
 b  Z b
f ≤ |f |
 

 a  a

Similarly, as |f (x)| ≤ g(x),

Z b Z b
|f | ≤ g
a a

Theorem 2.2.6. If f, g ∈ L[a, b] and if c ∈ R, then cf, f + g ∈ L[a, b] and

Z b Z b Z b Z b Z b
cf = c f and |f + g| ≤ |f | + |g|
a a a a a

Proof. |cf (x)| = |c| |f (x)| for all x ∈ [a, b]. If |f | ∈ R∗ [a, b], then cf, |cf | ∈ R∗ [a, b]. So cf ∈ L[a, b]. By
Triangle Inequality, |f (x) + g(x)| ≤ |f (x)| + |g(x)| for all x ∈ [a, b]. Since |f | + |g| ∈ R∗ [a, b], by comparison
test (Theorem 2.2.5), we have f + g ∈ L[a, b] and
Z b Z b Z b Z b
|f + g| ≤ (|f | + |g|) = |f | + |g|
a a a a
Improper and Lebesgue Integral 25

Theorem 2.2.7. Let f ∈ R∗ [a, b]. Then the following statement are equivalent.
a) f ∈ L[a, b]

b) There exists g ∈ L[a, b] such that f (x) ≤ g(x) for all x ∈ [a, b]
c) There exists h ∈ L[a, b] such that h(x) ≤ f (x) for all x ∈ [a, b]
Proof. (a =⇒ b) Let g = f
(b =⇒ a) Consider function g − f . As g − f ≥ 0 for all x ∈ [a, b] and since g − f ∈ R∗ [a, b], then
g − f = |g − f | ∈ R∗ [a, b] and so by Definition 2.2.4, g − f ∈ L[a, b]. Then just apply Theorem 2.2.6 to
g = f + (g − f ).
Proof for a ⇐⇒ c uses a similar method (just opposite) and so will be omitted.
Theorem 2.2.8. If f, g ∈ L[a, b], then max {f, g} , min {f, g} ∈ L[a, b] where

f (x) if f (x) ≥ g(x) for all x ∈ [a, b]
max {f, g} =
g(x) if g(x) > f (x) for all x ∈ [a, b]

and 
f (x) if f (x) ≤ g(x) for all x ∈ [a, b]
min {f, g} =
g(x) if g(x) < f (x) for all x ∈ [a, b]

Proof. We can write max{f, g} and min{f, g} as

1
max{f, g} = (f (x) + g(x) + |f (x) − g(x)|)
2
and
1
(f (x) + g(x) − |f (x) − g(x)|)
min{f, g} =
2
So by Theorem 2.2.6, f (x) − g(x) ∈ L[a, b] and also |f (x) − g(x)| ∈ L[a, b]. So also by Theorem 2.2.6,
max{f, g} = 21 (f (x) + g(x) + |f (x) − g(x)|) and min{f, g} = 12 (f (x) + g(x) − |f (x) − g(x)|) also belong to
L[a, b].

Theorem 2.2.9. Let f, g, α, β ∈ R∗ [a, b]. If f ≤ α, g ≤ α or if β ≤ f, β ≤ g, then max{f, g} and min{f, g}

also belong to R∗ [a, b] respectively.
Proof. Let f ≤ α and g ≤ α, then max{f, g} ≤ α, then

0 ≤ |f − g| = 2 max{f, g} − f − g ≤ 2α − f − g

As 2α − f − g ≥ 0 and 2α − f − g ∈ R∗ [a, b], 2α − f − g ∈ L[a, b]. By comparison test (Theorem 2.2.5),

|f − g| ∈ L[a, b] and so by Theorem 2.2.6, max{f, g} ∈ L[a, b].
The second part of the proof is similar with the proof above and so will be omitted.
Definition 2.2.10 (Seminorm and Distance). If f ∈ L[a, b], then we define seminorm of f to be
Z b
||f || = |f |
a

If f, g ∈ L[a, b], then we define the distance between f and g to be

Z b
dist(f, g) = ||f, g|| = |f − g|
a

Theorem 2.2.11 (Properties of Seminorm).

i. ||f || ≥ 0 for all f ∈ L[a, b].

ii. If f (x) = 0 for all x ∈ [a, b], then ||f || = 0.
26 Infinite Integral

iii. If f ∈ L[a, b], c ∈ R, then ||cf || = |c| · ||f ||.

iv. Let f, g ∈ L[a, b], then ||f + g|| ≤ ||f || + ||g||.

Proof. The proof is trivial and so will be omitted. For part (iv). consider the Triangle Inequality.
Theorem 2.2.12 (Properties of the Distance Function).

i. dist(f, g) ≥ 0 for all f, g ∈ L[a, b]

ii. If f (x) = g(x) for all x ∈ [a, b], then dist(f, g) = 0

iii. dist(f, g) = dist(g, f ) for all f, g ∈ L[a, b]
iv. dist(f, h) ≤ dist(f, g) + dist(g, h) for all f, g, h ∈ L[a, b]

Proof. Proof is trivial and so is omitted

Theorem 2.2.13 (Completeness Theorem). The sequence {fn } of function in L[a, b] converge to a function
f ∈ L[a, b] if and only if for all  > 0, there exists M ∈ R such that if m, n ≥ M then

||fm − fn || = dist(fm − fn ) < 

Proof. ( =⇒ ) As {fn } converges, then it is a Cauchy Sequence. So, given  > 0, there exists M ∈ R, such that

if m, n ≥ M , then |fm − fn | < (b−a)·u where u is the maximum value fm − fn can achieve for each m, n. Then,
by Theorem 2.2.6, |fm − fn | ∈ L[a, b]. So
b
 · u · (b − a)
Z
dist(fm , fn ) = |fm − fn | < =
a (b − a) · u

And so we are done.

( ⇐= ) Proof Omitted.

2.3 Infinite Integral

In the elementary calculus course, the usual approach to find infinite integrals is to define the integral on [a, ∞)
as imporoper integral and so Z Z ∞ γ
f = lim f
a γ→∞ a

2.3.1 Integral on [a, ∞)

We can define a partition

Ṗ = {([x0 , x1 ] , t1 ) , . . . , ([xn−1 , xn ] , tn ) , ([xn , ∞] , tn+1 )}

and
n
X
S(f ; Q̇) = f (ti )(xi − xi−1 ) + f (tn+1 )(∞ − xn )
i=1

As the final term f (tn+1 )(∞ − xn ) is not meaningful, we only consider the first n turns in which n is large
enough. So, for the integral over [0, ∞) only deal with the first n subpartition.

Ṗ = {([x0 , x1 ] , ti ) , . . . , ([xn−1 , xn ] , tn )}

Note that the partition Ṗ only lack the subinterval ([xn , ∞] , tn+1 ) such that
n
[
[a, ∞) = [xi−1 , xi ] ∪ [xn , ∞)
i=1
Infinite Integral 27

For Ṗ to be meaningful, we put a condition on xn in which we define a number d∗ such that Ṗ is (δ, d∗ )-fine
if
[xi−1 , xi ] ⊆ [ti − δ(ti ), ti + δ(ti )] for all t ∈ N, t ≤ n
and  
1
[xn , ∞) ⊆ ∗ , ∞
d
Definition 2.3.1. Let f : [a, ∞) → R. f is Generalised Riemann Integrable on [a, ∞) if there exsits A ∈ R
such that for all  > 0 there exists a gauge δ ∈ [a, ∞) such that if Ṗ is a (δ, d∗ )-fine tagged partition in [a, ∞),
then    
S f ; Ṗ − A ≤ 
 
R∞
If f suffice the condition above, then f ∈ R∗ [a, ∞) and a f = A
Definition 2.3.2. Let f : [a, ∞) → R, f is Lebesgue Integrable is f, |f | ∈ R∗ [a, ∞). If f suffice the condition,
then f ∈ L[a, ∞).
Theorem 2.3.3 (Hake’s Theorem). If f : [a, ∞) → R, then f ∈ R∗ [a, ∞) if and only if for all γ ∈ (a, ∞),
f : [a, γ] is Generalised Riemann Integrable on [a, γ] and
Z γ
lim f =A∈R
γ→∞ a
R∞
In this case, a
f =A

Proof. Proof Omitted.

Theorem 2.3.4. If f, g ∈ R∗ [a, ∞), then f + g ∈ R∗ [a, ∞) and
Z ∞ Z ∞ Z ∞
(f + g) = f+ g
a a a

Proof. Given  > 0, there exists a gauge δf on [a, ∞) such that if a tagged partition Ṗ is δf -fine, then
 
  Z ∞  
S f ; Ṗ − f  <

a 2

Similarly, there exist a gauge δg on [a, ∞) such that if a tagged partition Ṗ is δg -fine, then
 
  Z ∞  
S g; Ṗ − g  <

a 2

So, we use the gauge δ(t) = min {δf (t), δg (t)} for t ∈ [a, ∞) such that if a tagged partition Ṗ is δ-fine, then
 
  Z ∞ Z ∞   
   Z ∞     Z ∞   
S f + g; Ṗ − f+ g  ≤ S f ; Ṗ −
  f  + S g; Ṗ −
  g  < + = 

a a a a 2 2

Theorem 2.3.5. Let f : [a, ∞) → R and c ∈ (a, ∞), then f ∈ R∗ [a, ∞) if and only if f is Generalised Riemann
Integral on [a, c] and [c, ∞). In this case,
Z ∞ Z c Z ∞
f= f+ f
a a c

∗
Proof. ( ⇐= ) As f ∈ R [c, ∞), using Hake’s Theorem (Theorem 2.3.3) we get for all γ ∈ (c, ∞), f is integrable
on [c, γ] and Z Z ∞ γ
f = lim f
c γ→∞ c
28 Infinite Integral

We then apply the additivity theorem (Theorem 2.1.10), to [a, γ] = [a, c] ∪ [c, γ] and so
Z γ Z c Z γ
f= f+ f
a a c

So we have Z ∞ Z γ Z c Z γ Z c Z ∞
f = lim f= f + lim f= f+ f
a γ→∞ a a γ→∞ c a c
( =⇒ ) We can get the proof just by working backwards from the first part of the proof.
Example 2.3.6. Let α > 1 and fα (x) = 1
xα for x ∈ [1, ∞). We show that fα ∈ R∗ [1, ∞).
Proof. Let y ∈ (1, ∞), then fα is continuous on [1, γ] and so f ∈ R∗ [1, γ]. By Fundamental Theorem of
Calculus, we also have Z γ γ  
1 1 1−α 
 1 1
α
dx = x = 1 −
1 x 1−α 
1 α−1 γ α−1
Then, Hake’s Theorem implies that
Z ∞ Z γ   
1 1 1 1 1
α
dx = lim α
dx = lim 1 − α−1 = when α > 1
1 x γ→∞ 1 x γ→∞ α−1 γ α−1

Then by Fundamental Theorem of Calculus (Theorem 2.1.12), we can show that fa ∈ R∗ [1, ∞)
Example 2.3.7. Let
sin x

if x ∈ (0, ∞) x
f (x) =
1 if x = 0
R∞
Consider the Generalised Riemann Integrability of 0 f (x)dx.
Rγ
Proof. Since f (x) is continuous on [0, γ] for all γ ∈ (0, ∞), f ∈ R∗ [0, γ]. To prove that 0
f (x)dx has a limit
as γ reaches ∞, we let 0 < β < γ and use integration by parts.
Z γ Z β Z γ Z γ
sin x cos x γ cos x
f (x)dx − f (x)dx = dx =  − dx
0 0 β x x β β x2

Since |cos x| ≤ 1 and |sin x| ≤ 1,we are going to prove that the above terms approach to 0 as β < γ goes to ∞.
Consider the Cauchy sequence n1 . As it is a Cauchy sequence, we know that
 
1 1 
For all  > 0, there exists M ∈ R such that if M < β < γ, then  −  <
γ β 2
Then, for all  > 0, there exists M ∈ R such that if M < β < γ, then
 Z γ   Z γ   γ  Z γ   
 cos x γ cos x   cos x γ cos x   1   1  1
 − 1  < 2 = 

− − dx = + dx≤  + dx ≤ 2
x β x 2 x β x 2  x β 2
β x γ β 2
         
β β

Therefore, the Cauchy Criterion applies and and Hake’s Theorem implies that f ∈ R∗ [0, ∞).
Theorem 2.3.8 (Fundamental Theorem). Let E ⊂ [a, ∞) such that E is countable and that f, F : [a, ∞) → R
such that
1) F is continuous on [a, ∞) and lim F (x) exists.
x→∞
0
2) F (x) = f (x) for all x ∈ (a, ∞), x ∈ E
Then f ∈ R∗ [a, ∞) and Z ∞
f = lim F (x) − F (a) (2.6)
a x→∞

Proof.
Rγ If γ ∈ [a, ∞), we apply the Fundamental Theorem to interval [a, γ] to conclude that f ∈ R∗ [a, γ] and
a
f = F (γ) − F (a). We then use Hake’s Theorem to show that f ∈ R∗ [a, ∞) and we have Equation 2.6.
Infinite Integral 29

2.3.2 Integrals on (∞, b]

In a similar way, we just consider a tagged partition significant enough so that we don’t have to deal with the
left end point.
Let b ∈ R and f : (−∞, b] → R such that f ∈ R∗ (−∞, b] then given  > 0 there exists a gauge on (−∞, b]
consisting of a number d∗ > 0 and a strictly positive function (−∞, b]
So a tagged partition Ṗ is (d∗ , δ)-fine if
n
[
(−∞, b] = (−∞, x0 ) ∪ [xi−1 , xi ]
i=1

So that
[xi−1 , xi ] ⊆ [ti − δ(ti ), ti + δ(ti )] for all i ≤ n, i ∈ N
and  
1
(−∞, x0 ] ⊆ −∞, −
d∗

2.3.3 Integral on (∞, ∞)

Let f : (−∞, ∞) → R such that f ∈ R∗ (−∞, ∞). Given  > 0, there exists a gauge on (−∞, ∞) containing
n
strictly positive function on (−∞, ∞) and d∗ , d∗ > 0. So a tagged subpartition Ṗ = {([xi−1 , xi ] , ti )}i=1 is
∗
(d∗ , δ, d )-fine if
n
[
(−∞, ∞) = (−∞, x0 ] ∪ [xi−1 , xi ] ∪ [xn , ∞)
i=1

such that
[xi−1 , xi ] ⊆ [ti − δ(ti ), ti + δ(ti )] for all i ∈ N, i ≤ n
and   
1 1
(−∞, x0 ] ⊆ −∞, − and [xn , ∞) ⊆ ∗ , ∞
d∗ d
Theorem 2.3.9 (Hake’s Theorem). If f : (−∞, ∞) → R, then f ∈ R∗ (−∞, ∞) if and only if for all β < γ ∈
(−∞, ∞), f ∈ R∗ [β, γ] and Z γ
lim f =C∈R
β→−∞ β
γ→+∞

Theorem 2.3.10 (Fundamental Theorem). Let E ⊆ (−∞, ∞) and E is countable and f, F : (−∞, ∞) → R
such that
a) F is continuous (−∞, ∞) and lim F (x) exists.
x→±∞

b) F 0 (x) = f (x) for all x ∈ (−∞, ∞) with x ∈

/E
Then f ∈ R∗ (−∞, ∞) and Z ∞
f = lim F (x) − lim F (y)
−∞ x→∞ y→−∞

Theorem 2.3.11 (Comparison Test for Improper Integral). Let g be a non-negative function continuous on
(a, b), let f be a function continuous on (a, b). Suppose

|f (x)| ≤ g(x) for allx ∈ (a, b)

Rb Rb
If improper integral a
g exists, then a
f also exists
Remark:
1. We allow b = ∞ and/or a = −∞
2. This theorem is analogous to comparison test of series
30 Infinite Integral

Proof. Proof Omitted.

Example 2.3.12 (The Gamma Function). The gamma function Γ : (0, ∞) → R in which
Z ∞
Γ(t) = xt−1 e−x dx (2.7)
0
is an generalization of the factorial function such that it extends to the real number. Show that it is Generalised
Riemann Integrable, that the gamma function satisfies the equation :
Γ(t + 1) = tΓ(t) (2.8)
and so deduce that Γ(t + 1) = t! for all t ∈ N
Proof. We first start by splitting the integral into two parts by Additivity Theorem
Z ∞ Z 1 Z ∞
Γ(t) = xt−1 e−x dx = xt−1 e−x dx + xt−1 e−x dx
0 0 1
R1
For the first integral 0
xt−1 e−x dx, we note that
0 ≤ xt−1 e−x ≤ xt−1 for all x, t ∈ R
We also know that xt−1 ∈ R∗ [0, 1] using Hake’s Theorem.
Z 1 Z 1  t 1
1 γt
 
t−1 t−1 x 1
x dx = lim+ x dx = lim+ = lim+ − = if t > 0
0 γ→0 0 γ→0 t γ γ→0 t t t
R1
Therefore, by the Comparison Test for Improper Integrals (Theorem 2.3.11), 0 xt−1 e−x dx exists.
R ∞ t−1 −x
For the second integral 1 x e dx, consider g(x) = x2 , f (x) = xt+1 e−x . We note that
lim g(x) = lim xt+1 e−x = 0
x→∞ x→∞

As the function g(x) converges, it is bounded. So there exists a number M ∈ R such that
g(x) = xt+1 e−x ≤ M for all x ≥ 1
which implies that
f (x) = xt−1 e−x ≤ M x−2
It is also easy to prove that the function M x−2 ∈ R∗ [1, ∞) and so the rest of the proof follows from the first
integral and Hake’s Theorem.

By Hake’s Theorem, as the gamma function is Generalised Riemann Integrable, given  > 0, the gamma
function is Generalised Riemann Integrable on [0, γ] and
Z ∞ Z γ
t −x
Γ(t + 1) = x e dx = lim xt e−x dx
0 γ→∞ 0
Using Integration by Parts,
Z γ  Z γ  Z γ
γ
lim xt e−x dx = lim −xt e−x 0 + t xt−1 e−x dx = lim t xt−1 e−x dx
γ→∞ 0 γ→∞ 0 γ→∞ 0

Applying Hake’s Theorem again, we get

Z ∞ Z ∞
Γ(t + 1) = xt e−x dx = t xt−1 e−x dx = tΓ(t)
0 0
and so we are done with the first part.
To prove the second part, we first compute Γ(1) using Hake’s Theorem and the Fundamental Theorem of
Calculus Z ∞ Z γ
γ
e−x dx = lim e−x dx = lim −e−x 0 = lim −e−γ + e0 = 1


Γ(1) =
0 γ→∞ 0 γ→∞ γ→∞

Therefore,
Γ(t + 1) = t(t − 1) . . . (2)(Γ(1)) = t!
Convergence Theorems 31

2.4 Convergence Theorems

We now turn from functions to sequence of functions in terms of Generalised Riemann Integral.
Theorem 2.4.1 (Uniform Convergence Theorem). Let {fn } be a sequence in R∗ [a, b] and suppose {fn } con-
verges uniformly on [a, b] to f . Then f ∈ R∗ [a, b] and
Z b Z b
f = lim fk
a k→∞ a
Remarks:
1. A sequence of bounded function {fn } such that fn : A → R where A ⊆ R for all n ∈ N converges
uniformly on A0 ⊆ A to f : A0 → R is for all  > 0, there exists N ∈ N such that if n > N , then
|fn (x) − f (x)| <  for all x ∈ A0

2. Sequence {xn } is a Cauchy sequence if for all  > 0, there exists N ∈ N such that if n > m > N , then
|an − am | < 

3. The theorem does not apply for infinite interval

Proof. As {fk } converge uniformly, given  > 0, there exists N ∈ N such that if n > N , then

|fn − f (x)| < for all x ∈ [a, b]
2 · (b − a)
Similarly, if m > N , then

|fm (x) − f (x)| < for all x ∈ [a, b]
2 · (b − a)
So we get

|fn (x) − fm (x)| = |fn (x) − f (x) + f (x) − fm (x)| < |fn (x) − f (x)| + |fm (x) − f (x)| < for all x ∈ [a, b]
b−a
This implies
 
− < fn (x) − fm (x) < for all x ∈ [a, b]
b−a b−a
Then by Theorem 2.1.7,
Z b Z b Z b Z b Z b Z Z b 
   b
− < (fn − fm ) < =⇒ − < fn − fm <  =⇒  fn − fm  < 
 
a b−a a a b−a a a  a a 
nR o
b
Therefore, the sequence a fn is a Cauchy Sequence in R and so converges to some number A ∈ R
Rb
So, now we just need to show that f ∈ R∗ [a, b] with integral a f = A. For all  > 0, we use N ∈ N as
n
above. If Ṗ = {([xi−1 , xi ] , ti )}i=1 is a tagged partition of [a, b] and if n ≥ N , then
 
     X n 
S fn ; Ṗ − S f ; Ṗ  =  (fn (ti ) − f (ti )) (xi − xi−1 )
   
 
i=1
n
X n
X
≤ |fn (ti ) − f (ti )| (xi − xi−1 ) < (xi − xi−1 ) = (b − a)
i=1 i=1
R 
 b
Fix m ≥ N such that  a fm − A < , let δ be a gauge on [a, b] such that if Ṗ is δ-fine, then

Z 
 b  
f − S fm ; Ṗ  < 
 
 a m


Then we have
   
            Z b  Z b 
S f ; Ṗ − A ≤ S f ; Ṗ − S fm ; Ṗ  + S f ; Ṗ − fm  +  fm − A < (b − a) +  +  = (b − a + 2)
       
 a   a 
Rb
As we can set  to be as small as we want, we conclude that f ∈ R∗ [a, b] and a
f =A
32 Convergence Theorems

Rb
Another way of obtaining lim fn is using the Equi-integrability Theorem discussed below
k→∞ a

Definition 2.4.2 (Equi-integrability Hypothesis). The sequence {fk } in R∗ [a, b] is said to be equi-integrable if
for all  > 0, there exists a guauge δ ∈ [a, b] such that if Ṗ is a δ-fine partition on [a, b] and n ∈ N, then
 
   Z b 
S fn ; Ṗ − fn  < 
 
 a 

Theorem 2.4.3 (Equi-integrability Theorem). If {fn } ∈ R∗ [a, b] is equi-integrable on [a, b] and if

f (x) = lim fn (x) for all x ∈ [a, b],
n→∞
∗
then f ∈ R [a, b] and
Z b Z b
f = lim fn
a n→∞ a

Proof. For all  > 0, by Equi-integrability Hypothesis (Definition 2.4.2), there exists a gauge δ on [a, b] such
n
that if Ṗ = {([xi−1 , xi ] , ti )}i=1 on [a, b] is δ-fine, we have
 
   Z b 
S fn ; Ṗ − fn  <  for all n ∈ [a, b]
 
 a 

Since Ṗ has finite number of tags and f (t) = lim fn (t) for all t ∈ [a, b], there exists N ∈ N such that if
n→∞
m, n ≥ N , then
     X n
S fn ; Ṗ − S fm ; Ṗ  ≤ |fn (ti ) − fm (ti )| · (xi − xi−1 ) ≤ (b − a) (2.9)
 
i=1

If we take the limit as m → ∞ in Equation 2.9, we get

    
S fn ; Ṗ − S f ; Ṗ  ≤ (b − a) for all n ≥ N
 

Also, using m, n ≥ N , Definition 2.4.2 and Equation 2.9, we get

Z Z b  Z b 
 b         Z b 
fn − fm  =  fn − S fn ; Ṗ + S fn ; Ṗ − S fm ; Ṗ + S fm ; Ṗ − fm 
   

 a a   a a 
   
   Z b          Z b 
≤ S fn ; Ṗ − fn  + S fn ; Ṗ − S fm ; Ṗ  + S fm ; Ṗ − fm  ≤  + (b − a) +  = (2 + b − a)
     
 a   a 
nR o
b
Since we can set  > 0 to be as small as we want, a n
f is a Cauchy sequence converging to some number
nR o
b
A ∈ R. As a fn converges, we write for all  > 0, there exists N ∈ N such that if n > N , then
Z 
 b 
fn − A ≤ 
 

 a 
Rb
As we know that sequence { a fn } converge to some number A ∈ R, we just need to show that f ∈ R∗ [a, b]
with its integral equal to the same number A. So, given  > 0, there exists a gauge δ on [a, b] such that if Ṗ is
δ-fine on [a, b] and n ≥ N , then
 
           Z b Z b 
S f ; Ṗ − A = S f ; Ṗ − S fn ; Ṗ + S fn ; Ṗ − fn + fn − A
   
 a a 
   
        Z b  Z b 
≤ S f ; Ṗ − S fn ; Ṗ  + S fn ; Ṗ − fn  +  fn − A ≤ (b − a) +  +  = (2 + b − a)
     
 a   a 
Rb
As we can set  to be as small as we want, we conclude that f ∈ R∗ [a, b] with a f = A
Convergence Theorems 33

Theorem 2.4.4 (Monotone Convergence Theorem). Let {fk } be a monotone sequence of function in R∗ [a, b]
Rb
such that f (x) = lim fn (x) almost everywhere in [a, b], then f ∈ R∗ [a, b] if and only if { a fn } is bounded in
n→∞
R and if the conditition is satisfied,
Z b Z b
f = lim fn
a n→∞ a

Proof. Proof Omitted

Theorem 2.4.5 (Dominated Convergence Theorem). Let {fn } be a sequence in R∗ [a, b]. Let f (x) = lim fn (x)
n→∞
almost everywhere on [a, b]. If there exists g, h ∈ R∗ [a, b] such that g(x) ≤ fn (x) ≤ h(x) for almost every
x ∈ [a, b], then f ∈ R∗ [a, b] and
Z b Z b
f = lim fn
a n→∞ a
Rb
Moreover, if g, h ∈ L[a, b], then fn , f ∈ L[a, b] and ||fk − f || = a
|fk − f | → 0

Proof. Proof Omitted.

34 3 REFERENCE

3 Reference
1. Bartle, Robert G. & Sherbert, Donald R.(1982). Introduction to Real Analysis - 4th edition