Dr.

Janet Harris Semester 2, 2013/14

Chapter 4 INTEGRATION
4.1 Areas under Curves
4.2 The Definite Integral
4.3 Properties of the Definite Integral
4.4 The Fundamental Theorem of Calculus
4.5 Integration by Substitution
4.6 Integration by Parts
4.7 Additional Techniques of Integration.
Partial Fractions
4.8 Integration Using Tables and Computer
Algebra Systems
4.9 Numerical Integration
4.10 Improper Integrals
1
Introduction
In Chapter 2 we learnt about differentiation.
Given a quantity, differentiation allows us to find its
rate of change. It is related to finding the tangent.

Now we will learn about integration.

Given a rate of change of a quantity, integration
allows us to find a net change in the quantity. It is
related to the area under the curve.

Integral and differential calculus are connected.

In Section 4.4 we will study The Fundamental
Theorem of Calculus which connects them.
2
 4.1
Areas under Curves

Let us start with the following problem:

Consider a moving object. Suppose we know the
velocity of the object at all times. Can we find
the distance traveled?

3
DISTANCE PROBLEM: Constant velocity

If the velocity is constant, distance = velocity × time

On a velocity – time graph,

distance rectangular
traveled = area under
on [t1, t2] graph on [t1, t2]

Suppose the velocity

varied as shown.
Then the total distance
traveled would be the sum of
the areas of the rectangles.
4
DISTANCE PROBLEM: Variable velocity
Generally for any graph of v(t) against t,

distance traveled
= area under the graph.

(Later this will be formally

proved.)

We must learn how to find the area under a curve …

5
 For a rectangle,
area = width x height
But how do we find the area of a region
with curved sides?
We will first approximate the region by rectangles
to find the area. Then take the limit as we increase
the number of rectangles.
Example 1
Use rectangles to estimate
the area S under the
parabola y = x2 from 0 to 1.

6
Example 1
• S is contained within a square with side length 1, so
clearly 0 < S < 1. But we can get a better estimate…
• Suppose we divide S into four strips as shown.
• Approximate each strip by a rectangle whose height
is the same as the right edge of the strip.

7
Example 1, cont.

Each rectangle has width ¼.

The height of each rectangle is
the value of f(x) = x2 at the right
endpoint of each subinterval,
i.e. the heights are (¼)2, (½)2,
(¾)2, and 12

Let R4 be the sum of the areas of the rectangles:

R4   
1
4 4
1 2
 
1
4 2
1 2
 
1
4 4
3 2
 1 = 0.46875
1
4
2

 15
The area
32 A of S is less than R4, so A < 0.46875
8
Example 1, cont.

Alternatively, we could
draw rectangles using
left-hand endpoints.

Then the sum of the

areas is:
L4   0   
1
4
2 1
4 4
1 2
 
1
4 2
1 2
 
1
4 4
3 2
= 0.21875
 32
From 7
the graph, we see that A > L4, so A > 0.21875
 0.21875
Thus we have lower and upper estimates for A:
0.21875 < A < 0.46875 9
Example 1, cont.

We can repeat the

procedure with a larger
number of strips. This will
give us better estimates.

10
Example 1, cont.

Using a
computer we
get these
results:

• Using 1000 strips, 0.3328335 < A < 0.3338335

• Averaging these numbers gives A ≈ 0.3333335
• As n → ∞, it looks as if A → 1/3.
Now we will prove this …
11
Example 2

For the region S in Example

1, show that the sum of the
areas of rectangles with right
hand end points approaches
1/3, that is, 1
lim Rn  .
n  3

Note: You may use the result

n
n(n  1)(2n  1)

i 1
i  1  2  3  ...  n 
2 2 2 2 2

12
Example 2, cont.
 Divide the interval into n strips.
 Each rectangle has width 1/n
 The heights are the values of
f(x) = x2 at the points 1/n, 2/n,
3/n, …, n/n.
 That is, the heights are
(1/n)2, (2/n)2, (3/n)2, … , (n/n)2.
2 2 2 2
11 12 13 1n
So, Rn           ...   
nn nn nn nn
n
1 2 1
 3 (1  22  32  ...  n 2 )  3  i 2
n n i 1
13
Example 2, cont.
Using the formula given, we have
1 n
1 n(n  1)(2n  1) (n  1)(2n  1)
Rn  3
n
i 1 i  n3 
2

6

6n 2
1  1  1 1 1
So, lim R n  lim  1    2     1  2 
n  n  6
 n  n 6 3
We have proved the required result.
Notes
• It can similarly be shown that lim Ln  13 .
n 
• As n increases, Rn and Ln become better and better
approximations to the area A.
We can say A  lim R  lim L  1
n n 3
n  n  14
AREA PROBLEM - Generalised
Let’s apply the ideas above to the more general
problem of finding the area under a curve.
What is the area A of region S, the area under the
positive function y = f(x) between x = a and x = b ?

15
AREA PROBLEM - generalised

• Subdivide S into n
strips of equal width
• Interval [a, b] has
width b – a.
• Each strip has
width x  b  a
n
• The strips divide [a, b] into n subintervals:
[x0, x1], [x1, x2], [x2, x3], . . . , [xn-1, xn] where x0 = a, xn = b.

• If we approximate each strip with a rectangle using

right hand end points, we have n
Rn = f(x1)∆x + f(x2)∆x + … + f(xn)∆x   f ( x )x
i 1
i
16
AREA PROBLEM - generalised
As n → ∞, the area of the rectangles becomes closer to A.

Definition The area A of the region under a nonnegative,

continuous function f is the limit of the sum of the areas of
n
approximating rectangles: A  lim R  lim f ( x )x
n 
n n 

i 1
i

If f is continuous, this limit always exists.

We get the same value using left hand endpoints:n 1
A  lim L n  lim [ f ( x 0 )  x  f ( x 1 )  x    f ( x n  1 )  x ]  lim
n  n  n 
 f ( xi )  x
i0 17
 4.2 The Definite Integral
We can generalize further:
• Let f take any value (positive, negative or zero).
• Let sub-intervals have different widths.
For a function f(x) continuous on [a, b], let
P = {x0, x1, x2, x3 … xn} be a finite set of points
such that a = x0 < x1 < x2 < … < xn-1 < xn = b.
P is called a partition of [a, b].
• Instead of using left or right endpoints, in each sub-
interval [xi-1, xi] we take the height of the rectangle to
be the value of f at any “sample point” xi* such that
xi 1  xi *  xi .
18
Then the Riemann sum of f(x) on [a, b] is
n
R( f , P)   f ( xi *)xi where Δxi = xi – xi-1.
i 1

19
Note:
The Riemann sum is the sum of the areas above the
x-axis minus the sum of the areas below the x-axis

20
Definite Integral - Definition
• It can be shown that for a given function f(x)
continuous on [a, b], however we choose the
partition P and sample points xi*, the limit is always
the same real number, I. (See Stewart for proof.)
• I is called the definite integral of f on [a, b].
n
I   f ( x )dx  Lim R( f , P)  Lim  f ( xi *)xi
b

a n n 
i 1
(provided also that as n  , max xi  0)

• I.e. An integral, like an area, can be found as the

limit of a (Riemann) sum.
• The integral sign – like a tall letter S – was chosen
because of the relationship to a sum. 21
Notes
b
I   f ( x)dx is the definite integral of f on [a, b]
a
• ∫ is called an integral sign.
a and b are the limits of integration.
a is the lower limit, b is the upper limit.
f(x) is the integrand. x is the variable of integration.
The process of finding an integral is called integration.

• I is a number, not a function of x, so

b b b
I   f ( x)dx   f (t )dt   f (r )dr etc.
a a a

Integrals and Continuity A theorem states:

If f is continuous on [a, b] then f is integrable on [a, b].
22
 b
Geometric Interpretation of I   f ( x)dx
a
• If f(x) is positive then I is the area under the curve
f(x) between a and b.
• If f(x) has positive and negative values then I is a
‘net area’, that is, the value of A1 – A2 where A1 is the
area above the x-axis and A2 is the area below.

23
 5
Example 3 Consider the integral I   (3x  2)dx.
1
(a) Write the integral as the limit of a Riemann sum, and
thus evaluate it. You may use the result n n(n  1)

i 1
i
2
.

24
(b) Confirm your answer to part (a) by sketching a graph of
the area represented by the integral, and using geometry
to evaluate it.

25
 4.3 Properties of the Definite Integral
If f and g are integrable on an interval containing a, b and c,
a
a f ( x)dx  0
b a
a f ( x)dx  b f ( x)dx
b b b
a [f ( x)  g ( x)]dx  a f ( x)dx  a g ( x)dx
b c c
a f ( x)dx  b f ( x)dx  a f ( x)dx
a
If fo is an odd function then 
a
f o ( x)dx  0
a a
If fe is an even function then  a
f e ( x)dx  2 f e ( x)dx
0

26
Comparison Properties
b
If f(x) ≥ 0 for a ≤ x ≤ b then  a
f ( x )dx  0.
b b
If f(x)  g(x) and a  b then 
a
f ( x )dx   g ( x )dx.
a

b b
If a  b then a
f ( x )dx   f ( x ) dx.
a

If m ≤ f(x) ≤ M for a ≤ x ≤ b then

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

27
Example 5
1
e
 x2
Show that dx  1
0
and find a lower limit to the value of the integral.

28
4.3 A Mean Value Theorem for Integrals
If f is continuous on [a, b] then there exists
b
c  [a, b] such that  f ( x)dx  f (c)(b  a)
a
1 b
That is, there exists c  [a, b] such that f (c) 
ba a
f ( x)dx

Geometrically this means

area under area of rectangle

=
f(x) on [a, b] of width (b – a)
and height f(c)

So this value f(c) is called the average value or

mean value of f on [a, b]. f av  1 b f ( x)dx
b  a a 29
Mean Value Theorem - Proof
Let m and M be the min and max values of f on [a, b]:
m  f ( x)  M for all x in [a, b].
b
We have m(b  a)  
a
f ( x)dx  M (b  a)
so 1 b
m 
ba a
f ( x)dx  M

By the Intermediate Value Theorem (Ch.1), if f(x) is

continuous it must take on every value between m
and M at some point between a and b. Hence there
exists c  [a, b] such that 1 b
f (c) 
ba 
a
f ( x )dx.
30
Example 6
Write down an integral which represents the mean value
of the function f(x) = 4 + 3x2 on the interval [2, 8].

31
4.4 The Fundamental Theorem
of Calculus

Calculating integrals by taking the limit of Riemann

sums is often long and difficult.
Thankfully there is an easier way!
This easier way relies on the relationship between
integration and differentiation ...

33
Revision: Antiderivatives
From Chapter 3, recall that:
• On an interval J, an antiderivative F of function f
is such that F(x) = f(x) for all x in J.

• Antiderivatives are not unique.

• If F(x) is an antiderivative of f(x) then so is
G(x) = F(x) + C for any constant C.

Now we will learn the Fundamental Theorem of

Calculus (FTC). The theorem has 2 parts.

34
FTC – Part A

Let f(x) be a function continuous on an interval J

containing the point a.
x
Then F ( x)   f (t )dt is an antiderivative of f.
a

I.e. d x

dx a
f (t )dt  f ( x)

This tells us how to differentiate a definite integral

with respect to its upper limit.
E.g. d x
   ) 1
3 3
[cos(t ) 1] dt cos( x
dx 1
35
 FTC – Part A - Proof
x
F ( x)   f (t )dt is an antiderivative of f.
a

The proof uses the definition of the derivative, the mean

value theorem and some properties of the definite integral.
x
F ( x)   f (t )dt
a
F ( x  h)  F ( x )
 Lim   f (t )dt   f (t )dt 
1  xh x
F ( x)  Lim
h0 h h 0 h  a a 
1 xh
 Lim  f (t )dt
h 0 h x

1
 Lim f (c)h for some c between x and x + h
h 0 h

 Lim f (c)  f ( x)
h0
since c  x as h  0.
36
Example 7 Find the derivative of
x
a) F ( x)   e dt
t 2
3
x2

b) G( x)  t 2
e dt
1

x2
c) H ( x)   e dt t 2
x

20
d) F ( x)   sin t 2 dt
2x

37
FTC – Part B

If G(x) is any antiderivative of f(x) on J, i.e. G(x) = f(x),

b
then a f ( x)dx  G(b)  G(a) for any b in J.
• So if we know an antiderivative of the integrand,
we can evaluate the integral.
• Any antiderivative can be used because the
constants cancel out in the subtraction.
b b
The evaluation symbol a is defined F ( x) a  F (b)  F (a)
(Other common notations are F ( x ) and F ( x)
b b
a a .)
Then Part B of the FTC can be written

a f ( x)dx   f ( x)dx a
b b (Newton Leibnitz
formula.)
38
Example 8
1

2
Evaluate (a) x dx
0

 /2
(b)  0
cos x dx

FTC – Part B - notes

b

a
f ( x)dx  F (b)  F (a) is the definite integral of f on [a, b]

 f ( x)dx  F ( x)  C is the indefinite integral, or general

antiderivative, of f
• The definite integral is a number
The indefinite integral is a function 39
FTC – Part B - Proof

If G(x) is any antiderivative of f(x) on J, i.e. G(x) = f(x),

b
then a f ( x)dx  G(b)  G(a) for any b in J.

x
From Part A, F ( x)  a f (t )dt is an antiderivative of f on J

If G is any anti-derivative of f on J, then F(x) = G(x) + C,

so x

a
f (t )dt  F ( x)  G( x)  C

Letting x = a gives 0 = G(a) + C, so C = –G(a),

x
so
 f (t )dt  G( x)  G(a)
a
b
Then putting x = b gives a f ( x)dx  G(b)  G(a)
40
Example 9
2
a) Find 
1
(3x 4  x  1) dx

b) Find the area under the curve y = 1 / x and above y = 0

between x = 1 and x = 4.

41
Example 10
a) Find the area under the curve y = sin 2x, above y = 0
between x = 0 and x = /2.

b) Find the average value of f(x) = e-x + cos x on the

interval [-/2, 0]. [Reminder: f av  1 b f ( x)dx ]
b  a a

42
Practical Application: Finding Net Change
• From the FTC we know that
dF b
if f ( x) 
dx
then 
a
f ( x)dx  F (b)  F (a)
• If F(x) is a quantity, f = dF/dx is its rate of change.

Net Change Theorem:

The integral of a rate of change is the net change
b dF

a dx
dx  F (b)  F (a)

• Given a quantity, we can differentiate to find a rate of change

• Given a rate of change, we integrate (over an interval) to
find the net change in the quantity (over that interval).
differentiate
Quantity Rate of
integrate change 43
Finding Net Change: Examples

Net Change Theorem:

The integral of a rate of change is the net change
b dF
a dx
dx  F (b)  F (a)

dN
A population is growing at rate R(t )  .
t2 dt
What does

R(t ) dt represent?
t1

An object moves in a straight line with velocity

t2
function v(t). What does  v(t ) dt represent?
t1

44
Example 11
At 7 a.m. water begins flowing into a tank at a rate of
10 + 2t litres/hour, where t is time in hours after 7 a.m..
By 11 a.m. how much water has flown in?

45
Total and Marginal Cost
The cost function C(x) is the total cost a company incurs
in producing x units of a certain product.
The marginal cost is the derivative dC/dx: it represents
how the cost of production changes as the amount
produced changes.

If we know the marginal cost then

x2
x
1
C ( x)dx  C ( x2 )  C ( x1 ) = cost of increasing production
from x1 to x2 units

46
Example 12
The marginal cost in producing x computer chips (in units
of 1000) is C( x)  150 x 2  3000 x  17500 ($ per 1000 chips).
a) Find the cost of increasing production from 10,000 to
15,000 chips.
b) Assuming C(0) = 35,000 (i.e. set up costs were
$35,000), find the total cost of producing 15,000 chips.

47
Practical Application: Distance Traveled
Earlier we said that the area under a velocity-time graph
(for positive velocities) represents distance travelled.
Now we can prove this.
Let s(t) be the position and v(t) the velocity of a particle
at time t. v(t )  ds t
 v(t )dx  s(t )  s(t )
2

dt
so t1
2 1

• If v(t)  0 then s(t2) – s(t1) is the distance travelled.

• More generally, for any v(t), s(t2) – s(t1) is the net change
in position, also called the displacement.
• If we want to find the actual distance travelled, rather
than displacement, we should integrate the speed, |v(t)|.
t2
Displacement during [t1, t2]   v(t )dx
t1
t2
Total distance travelled during [t1, t2]   v(t ) dx
t1 48
Example 13
A particle moves with velocity v(t) = 4 – t2. Over the first
5 seconds, what is (a) the net change in displacement?
(b) the distance travelled?

49
4.5 Integration by Substitution
• By the FTC, we can evaluate an integral if we can find
an antiderivative of the integrand.
• Sometimes we can find an antiderivative by inspection.
To be able to do this, you should memorize elementary
derivatives and integrals.
• But many integrals cannot be solved by inspection. In
general, finding antiderivatives is not as easy as finding
derivatives – indeed sometimes it is very difficult or even
impossible!
• We will study two techniques. These will enable us to
evaluate many common integrals.
• The substitution rule (section 4.5)
• The method of integration by parts (section 4.6)
50
The Substitution Rule for Indefinite Integrals
Let f and g be functions and let F be an antiderivative of f.
The Chain Rule of differentiation (2.3.2). can be written
F g ( x)   F g ( x) g ( x)  f g ( x) g ( x)
d
dx
Integrating both sides we get
 f g ( x)g ( x)dx  F g ( x)  C
Using the substitution u = g(x), so du = g(x)dx, we can


rewrite this: f (u) du  F (u)  C

Substitution Rule:

 f g ( x)g ( x)dx   f (u)du where u = g(x).

51
Example 14

 xe
x2
a) Find dx

b) Find  
3 7 2
( 2 x 1) x dx

* Remember to give your answer in terms of the original variable! * 52

Example 15

  1)dx
2
a) Find x sin( x

2
x
b) Find
 x3  1 dx

c) Find  tan x dx

53
The Substitution Rule for Definite Integrals

 f g ( x)g ( x)dx  F g ( x)  C

For indefinite integrals,

Applying the Newton-Leibnitz formula gives

f  g ( x)  g ( x)dx  F  g ( x)   F  g (b)   F  g (a ) 
b b
a a
g (b )

g (b )
 F (u ) g ( a ) f (u )du.
g (a)

Hence the Substitution Rule for Definite Integrals:

If g is differentiable on [a, b], f is continuous on the
range of g, and u = g(x), b f g ( x)g ( x)dx  g (b) f (u)du
a g (a)
So we need to – replace g(x) with u
– replace limits a, b with g(a), g(b) 54
Example 16
e ln x
Find

1 x
dx

55
Example 17
3
Find the area under the curve y  on [1, 2].
5x 1

56
 Example 18
Prove that for all positive integers n,
 /2  /2
 cos x dx  
n
sin n x dx
0 0

Solution
/ 2 / 2  / 2 n  
0 cos x dx   sin (  x) dx   sin (  x) d (  x)
n n
I=
0 2 0 2 2
Let u = /2 – x . Then u(0) = /2 , u(/2) = 0, so
0 / 2 / 2
I= 
 / 2 sin u du   sin u du  
n n
sin n x dx
0 0

57
Inverse Substitution
• So far we have used direct substitutions: we have
replaced a function in the integrand with a new variable.
• Sometimes it is more helpful to do inverse substitution:
replace the variable of integration with a function of a
new variable.
Example 19
dx
Find  1 / 2 using the inverse substitution x = u 6.
x (1  x1 / 3 )

58
 4.6 Integration by Parts
Suppose U(x) and V(x) are two differentiable functions.
The Product Rule of differentiation is:
d
U ( x)V ( x)  U ( x)  V ( x)
dV dU
dx dx dx
This can be rearranged:
U ( x)
dV

d
U ( x)V ( x)  V ( x) dU
dx dx dx
Integrating both sides w.r.t. x gives
dV dU
 U ( x) dx
dx  U ( x)V ( x)   V ( x)
dx
dx

or more simply,
 UdV  UV   VdU
59
This method is called ‘integration by parts’ because we
replace the integral with the sum of an integrated term UV
and another integral which remains to be evaluated – so
we accomplish only “part” of the original integration.
Example 20
Find the following integrals:
(a)  x tan 1 x dx

60
 3
(b)  ln x dx
1
Hint: We don’t know how to integrate ln x … but we do know
how to differentiate it!

e
(c)  x 6 ln x dx
1

61
Using integration by parts we can sometimes obtain what
are called reduction formulae.
Example 21 Find a reduction formula for I n  sin n x dx 
Solution Use integration by parts, with
n 1 n 2
U  sin x , dU  (n  1) cos x sin x dx
dV  sin x dx , V   cos x dx
I n   sinn1 x cos x  (n  1)  sinn2 x cos 2 x dx
  sinn1 x cos x  (n  1)  sinn2 x(1  sin2 x) dx
  sinn1 x cos x  (n  1) I n2  (n  1) I n
So nI   sin n 1 x cos x  (n  1) I
n n 2

1 n 1 n 1
Hence I n   sin x cos x  I n 2 .
n n 62
A formula like that found in Example 20, expressing In
as a function of integral(s) involving lower power(s), is
called a reduction formula.
We can easily see that I0 = dx = x + C
and I1 = sinx dx = – cosx + C,
Then by repeated application of the formula, we can
find In for all higher n.

63
4.7 Additional Techniques of
Integration
Integrals of Trigonometric Functions (1)
Integrals of tanx, cotx, secx and cosecx.
• Integral of tanx was found in Example 11b
• Can similarly show  cot x dx = ln |sinx| + C
• By differentiation we see
d sec x tan x  sec x
2
ln | sec x  tan x |  sec x
dx sec x  tan x
Thus
 sec xdx  ln sec x  tan x  C
• It can similarly be shown that
 cosec x dx   ln cosec x  cot x  C 64
Integrals of Trigonometric Functions (2)

Integrals of form 
m n
sin x cos xdx (m,n +ve integers)

• If m or n is odd, use substitution.

E.g. n = 2k + 1. Use sin2x + cos2x = 1 to rewrite
 
2 k 1
sin m
x cos xdx  sin m
x cos 2k
x cos xdx

  sin m x(1  sin 2 x) k cos xdx

 
m 2 k
Let u = sinx, so du = cosx dx , gives u (1 u ) du
which is easily found.
• If m and n are both even, use the formulae
sin2x = ½(1 – cos 2x)
cos2x = ½(1 + cos 2x).
65
Example 22


3 8
a) Find sin x cos x dx

 x dx
4
b) Find sin

66
Trigonometric Inverse Substitutions
If we need to integrate functions containing expressions
such as a2  x2 , a 2  x2 , or x2  a 2 ,
sometimes the easiest way is using an trigonometric
(inverse) substitution that gets rid of the square root sign.

• For integrals involving a 2  x 2 (a > 0), try using x = a sin.

Note that a 2  x 2 only exists if |x|  a, which means |  |  
2
Then cos  0, so a 2  x 2  a 1  sin 2   a cos 2   a cos 

67
Example 23
Prove that the area of a circle of radius r is r2.

68
Example 24 dx
Evaluate I 
 (5  x 2 )3 / 2

69
Trigonometric Inverse Substitutions (cont.)

• For integrals involving or 1 (a > 0),

a x
2 2
x2  a2
try using x = a tan .

We can obtain all values of x by limiting  to |  | < /2,

so sec  0 and a 2  x 2  a 1  tan 2   a sec

• For integrals involving x 2  a 2 (a > 0), try using x = a sec .

This is more complicated.
x 2  a 2  a sec2   1  a tan 2   a tan
If x  a, then 0   < /2, so tan  0 and x 2  a 2 = a tan.
If x  –a, then /2 <   , so tan  0 and x 2  a 2 = – a tan.
70
Example 25 dx
Evaluate
 4  x2

71
Example 26 dx
Evaluate  x2  a2
Solution

72
 4.7 Partial Fractions
In this section we learn how to evaluate integrals of
the form  P( x) dx where P and Q are polynomials.
Q( x)

Revision: A polynomial P(x) is a function of the form

P(x) = anxn + an-1xn-1 + … + a2x2 + a1x + a0
where n is a positive integer and an  0.
The constants an, an-1, … a0 are called coefficients;
n is the degree of the polynomial.

A function of the form P(x) / Q(x), where P(x) and Q(x)

are polynomials, is called a rational function.

73
Introduction to Partial Fractions
In a fraction p/q, p is called the numerator, q is the
denominator.
To add or subtract fractions, we use a common
denominator. For example,
2 1 2( x  2)  1( x  1) x5
   2
x 1 x  2 ( x  1)( x  2) x  x2
Using this result in reverse enables us to integrate
the function on the RHS above:
x5  2 1 
 x 2  x  2 dx    x  1  x  2 dx = 2ln|x – 1| – ln|x + 2| + C
This is the idea of the method of partial fractions.
To integrate a rational function, P(x) / Q(x), we first
express it as a sum of simpler terms. 74
Theory of Partial Fractions: Case 1
Case 1: Degree of P  degree of Q
A theory states:
Any rational function P(x) / Q(x) can be
expressed in the form P( x)  P ( x)  R( x)
1
Q( x) Q( x)
where P1 is a polynomial and R is a polynomial
of degree less than that of Q.
Given a rational function with degree of P  degree of Q,
first we use long division, or some similar procedure,
to simplify the expression – to get a polynomial P1
(which can be easily integrated) plus a new rational
function R(x)/Q(x) (which can be handled as
described in Case 2, below). 75
 x 3
x
Example 27 Find
 x  1 dx
x x
3
2
By long division we find  x  x2
2
x 1 x 1

Thus we can find the integral:

x3  x 2
 x  1 dx   x  x  2  x  1 dx
2

3 2
x x
   2 x  2ln | x  1|  C
3 2

76
Theory of Partial Fractions: Case 2

Case 2: Degree of P < degree of Q

In this case, we rewrite P(x)/Q(x) as a sum of simpler
fractions, called ‘partial fractions’ (which we can
integrate).
A theory states that any polynomial can be factorised
into a product of a
• constant, linear factors of the form (x – a), and
• quadratic factors of the form (x2 + bx + c) which
have no real roots (i.e. cannot be factorised into
linear factors).

For any Q(x) we can write partial fractions by the

following method:
77
Case 2, cont.

Case 2: Degree of P < degree of Q

Method
1. Factorise Q as far as possible.
2. Decompose P/ Q into simpler fractions:
 For each factor (x – a)n in Q, write fractions
A1 A2 An
  ... 
x  a ( x  a) 2
( x  a) n
 For each factor (x2 + bx + c)n in Q which has no real
roots, write fractions
B1 x  C1 B2 x  C 2 Bn x  C n
  ... 
x  bx  c
2
( x  bx  c)
2 2
( x 2  bx  c) n
3. Find the constants A, B, C, etc. (see example below).
78
Case 2, cont.

Examples of Step 2
Factorized form of P/Q Partial Fractions

1
( x  1)(2 x  3)
2x2  3
x(3x  4)( x  1)
2x2  3
x( x  1) 2
5x3  7 x  3
( x 2  2) 2
79
Example 28 ( x  4) dx
Find I   2
x  5x  6
The denominator can be factorised as follows:
x4 x4

x  5 x  6 ( x  2)( x  3)
2

So we write partial fractions:

x4 x4 A B
  
x  5 x  6 ( x  2)( x  3) x  2 x  3
2

To find constants A and B, add the partial fractions

together and equate with the original expression:
x4 A( x  3)  B( x  2)

( x  2)( x  3) ( x  2)( x  3)
We need the numerators of the two sides to be equal,
i.e. we require x  4  A( x  3)  B( x  2) for all x. (*)
80
Example 28 (cont.)
From (*), we can find A and B by one of two methods.
Method 1
Equate coefficients on the two sides of the equation.
Expand out: x + 4 = (A+B)x + (–3A –2B)
Coefficients of x: 1=A+B
Constants: 4 = –3A –2B
Solving these two equations gives A = -6, B = 7.
Method 2 (often quicker!):
Substitute particular values of x into (*)
E.g., x = 3 gives 7 = B, letting x = 2 gives 6 = –A

Using these values of A, B we can evaluate the integral:

( x  4) dx 6 7
 x 2  5x  6   x  2  x  3 dx = –6 ln|x – 2| + 7 ln|x – 3| + C.
81
Example 29 Find 1
 x( x  1) dx

82
 ( x  2)
 x 4  x 2 dx
Example 30 Find

83
Final note
Any rational function P/Q can be integrated using the
methods above.
However sometimes a different approach is quicker!
2
For example, consider  x
dx
( x  1) 5

This could be found by writing the integrand as the sum

of five partial fractions …
… but it can be found much more easily by using the
substitution u = x – 1 !

84
4.8 Integration Using Tables
Inside the back cover of Stewart is a table of 120
common integrals. More extensive tables may be
found in other books. In practice, scientists often
refer to such tables of intervals.

Integrals rarely appear in exactly the form given in

the table. But by using algebraic manipulation or a
substitution we can often transform a given integral
into one of the forms in the table.

85
Example 31
Find I   x 2  2 x  4 dx.
You may use the following result (where a is a constant):
2
u a
       )C
2 2 2 2 2 2
a u du a u ln( u a u
2 2

86
 Integration Using Computer
Algebra Systems
Nowadays there are also computer programs which can
do integration, such as Derive, Mathematica and Maple.
These programs can evaluate a wide range of integrals.
However sometimes they give their answers in a less
convenient form than one could find by hand.
For example, consider I =  x (x2 + 5)8 dx
By hand, using the substitution u = x2 + 5, we find:
 x (x2 + 5)8 dx = (x2 + 5)9 /18 + C
Meanwhile Maple and Mathematica give:
x18/18 + 5x16/2 + 50x14 + 1750x12/3 + 4375x10 + 21875x8
+ 218750x6/3 + 156250x4 + 390625x2.
87
 4.9 Numerical Integration
There are two situations in which it is impossible to find
analytically the exact value of a definite integral.
b
1) We may want to find a f ( x)dx … but be unable to do
so by any analytical techniques. Even many common
functions do not have antiderivatives, for example
1 1
  1  x3 dx
x2
e dx
0 1
2) If a function is defined by experimental data then there
may be no exact formula for the function.
In both these cases we cannot find exact values of
integrals, but we can find approximate values by
numerical methods.
88
Numerical Integration by Riemann Sums
We already know one method for finding an approximate
value of an integral: we can use any Riemann Sum.
Revision If interval [a, b] is divided into n subintervals of equal

n


b a b
length x 
n
, then
 a
f ( x) dx 
i 1
f ( xi *) x
where xi* is any point in the ith subinterval [xi-1, xi]

89
In Section 4.1 we met two common types of Riemann Sum.

Left and Right Endpoint Approximations

• If xi* is chosen to be the left endpoint
of [xi-1, xi] then xi* = xi -1 and we have
the left endpoint approximation:
n
f ( x) dx  Ln   f ( xi 1 ) x
b
a
i 1

• If xi* is chosen to be the right endpoint

of [xi-1, xi] then xi* = xi and we have the
right endpoint approximation:
n
f ( x) dx  Rn   f ( xi ) x
b
 a
i 1

90
Example 32
The table below gives the values of a function f(x) on [0, 30].
30
Estimate the value of the integral  f ( x) dx using (a) left
0
endpoints, (b) right endpoints.
x 0 5 10 15 20 25 30
f(x) 1 1.7 2.2 1.9 1.6 1.1 0.7

91
 Trapezoidal Rule
• A more accurate estimate is often obtained by taking the
average of Ln and Rn:
b 1 n n
 x  n 
a f ( x ) dx   
2  i 1
f ( xi 1 )x  f ( xi )x  
i 1 
 
2  i 1
 f ( xi 1 )  f ( xi )

x
So
  f ( x0 )  f ( x1 )   f ( x1 )  f ( x2 )     f ( xn1 )  f ( xn )
2
1 x
Tn  ( Ln  Rn )   f ( x0 )  2 f ( x1 )  2 f ( x2 )    2 f ( xn1 )  f ( xn )
2 2
where x  b  a and xi  a  i x
n
This is known as the trapezoidal rule
because x
[ f ( xi 1 )  f ( xi )]
2
is the area of the trapezoid that lies
above the ith subinterval.
92
Midpoint Approximation
• Another good approximation is
obtained by choosing xi* to be
the midpoint of [xi-1, xi]:
xi *  xi  12 ( xi 1  xi )
This is called the midpoint rule:
M n  x [ f ( x1 )  f ( x 2 )  ...  f ( x n )] with x  b  a
n
Example 32, cont. x 0 5 10 15 20 25 30
30
Estimate 0
f ( x) dx using f(x) 1 1.7 2.2 1.9 1.6 1.1 0.7

(c) trapezium rule,

(d) midpoints.
93
Example 33
2
Find an approximate value of the integral  (1/ x) dx using
1
n = 5 and (a) the midpoint rule, (b) the trapezoidal rule.

We have n = 5, a = 1 and b = 2, so ∆x = (2 – 1)/5 = 0.2

(a) The midpoints of the subintervals are: 1.1, 1.3, 1.5, 1.7, 1.9
So the midpoint rule gives
(1/ x) dx  M 5  x f (1.1)  f (1.3)  f (1.5)  f (1.7)  f (1.9)
2
1
1 1 1 1 1 1 
        0.691908
5 1.1 1.3 1.5 1.7 1.9 
(b) The trapezoidal rule gives
x
 f (1)  2 f (1.2)  2 f (1.4)  2 f (1.6)  2 f (1.8)  f (2) 
2
1
(1/ x) dx  T5 
2
1 1 2 2 2 2 1
         0.695635
10 1 1.2 1.4 1.6 1.8 2 
Compare the exact answer: ln 2 = 0.693147 94
Relative Accuracy
Note that the Midpoint Rule generally gives more accurate
results than the Trapezoidal Rule.
As shown in the figure below, the midpoint error (red) is
smaller than the trapezoidal error (blue).
The errors in the Midpoint and Trapezoidal Rules are also
opposite in sign.

95
Simpson’s Rule
Another approximation, which is much better than either the
Trapezoidal or Midpoint Rules, is known as Simpson’s Rule.
In this case we again divide the interval into n strips of equal
width x = (b – a)/n, but this time n must be even. Then,
b x
a
f ( x)dx  S n  [ f ( x0 )  4 f ( x1 )  2 f ( x2 )  4 f ( x3 )  2 f ( x4 )  
3
 2 f ( xn 2 )  4 f ( xn 1 )  f ( xn )]

where x  b  a , xi  a  i x and n is even.

n
(Note the pattern of coefficients: 1, 4, 2, 4, 2, … 4, 2, 4, 1.)
It can be shown that Simpson’s Rule (with 2n strips) is a
weighted average of the results of the Trapezoidal and
Midpoint Rules (each with n strips): 1 2
S2 n  3 Tn  3 M n
96
Simpson’s Rule is derived as follows.
• Consider a ‘double-subinterval’, [xi, xi+2].
• On that interval, approximate the curve by a parabola
y = Ax2 + Bx + C passing through points (xi, f(xi)), (xi+1, f(xi+1)),
(xi+2, f(xi+2)), as shown below. Solve for A, B and C.
• Calculate the area under the parabola. It can be shown that
the area on [xi, xi+2] is x [ f ( xi )  4 f ( xi 1)  f ( xi 2 )] .
3
• Add together the areas for all the double subintervals. This
gives the formula above. (See Stewart for full derivation.)

97
Example 33, cont.
2
(c) Approximate

1
(1/ x) dx using Simpson’s Rule with n = 10.
Putting f(x) = 1/x, n = 10, and ∆x = 0.1 in Simpson’s Rule,
we have:
2 1 x
1 x
dx  S10 
3
[ f (1)  4 f (1.1)  2 f (1.2)  4 f (1.3)

 ...  2 f (1.8)  4 f (1.9)  f (2)]

1 4 2 4 2 4 2 4 
      
0.1 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 

  
3  2 4 1 
    
 1.8 1.9 2 
 0.693150
Compare the exact answer: ln 2 = 0.693147 98
Example 34
The table shows data traffic t (hours) t (seconds) D(t)
between two computers over a 0 0 3.2
six hour period. D(t) is the data 1 3600 2.7
throughput, measured in 2 7200 1.9
megabits per second (Mb/s). 3 10800 1.7
Use Simpson’s Rule to estimate 4 14400 1.3
the total amount of data 5 18000 1.0
transmitted over this period. 6 21600 1.1

99
t (hours) t (seconds) D(t) Let A(t) be the total amount of
0 0 3.2 data (in Mb) transmitted by time t.
1 3600 2.7 Then A’(t) = D(t). So by the Total
2 7200 1.9 Change Theorem (p.45) the
3 10800 1.7 amount of data transmitted over
4 14400 1.3 the period is:
21,600
5 18000 1.0 A(21,600)   D(t ) dt
6 21600 1.1 0

Using Simpson’s Rule with n = 6 and ∆t = 3600

t
A(21600)  [ D(0)  4 D(3600)  2 D(7200  ...  D(21600)]
3
3600
 [3.2  4(2.7)  2(1.9)  4(1.7)  2(1.3)  4(1.0)  1.1]
3
 1200[32.3]  38760
The total amount of data transmitted is 38,760 Mb = 38.8 Gb.
100
 4.10 Improper Integrals
So far we have studied definite integrals of form
b
I   f ( x)dx
a
where [a, b] is a finite interval and f is continuous
and bounded on [a, b]. These are called proper
integrals.
Now we extend the concept of the definite integral to
cover other possibilities.
• We have an improper integral of type I if the
interval is infinite
• We have an improper integral of type II if f is
unbounded on [a, b]
101
Type I Improper Integrals
Example 35
Find the area A of the region under the curve y = 1/x2
and above the x -axis to the right of x = 1.
Solution
The area between x = 1 and x = t (t >1) is:
1
t
t dx 1
A(t )   2   1
1 x x 1 t
It is reasonable then to say A  Lim A(t )  1 unit 2
t 
 dx t dx
And to write A 2
 Lim  2
1 x t  1 x

This example illustrates how integrals over infinite

intervals are defined.
102
Type I Improper Integrals: Definition
t
• If a f ( x)dx exists for every t  a, then
 t
a
f ( x)dx  Lim  f ( x)dx
t  a

provided this limit exists as a finite number.

b
• If t f ( x)dx exists for every t  b, then
b b
 f ( x)dx  Lim
t   t
f ( x)dx

provided this limit exists as a finite number.

a 
• If both  f ( x)dx and a f ( x)dx exist as finite numbers
then we define  f ( x)dx  a f ( x)dx   f ( x)dx
     a

An improper integral is called convergent if the limit

exists as a finite number, and divergent otherwise. 103
Example 36
Evaluate the given integrals or show that they do not exist.

(a)  0
e  x dx



2x
(b) e dx
1


(c)  0
cos x dx

104
Example 37
 dx
Find the integral  2 or show that it does not exist.
1 x

105
The following is a well-known result:
 dx
1 xp
is convergent for p > 1 and divergent for p  1.

Example 38 Prove the result above.

dx  t 1
For p = 1, 1 p t  1
  ln x  1  lim ln t  
t
lim dx lim
x x t  t 
t
 dx t  1  p 1 
For p ≠ 1,   lim  x dx  lim 
p
x 
p t   p  1
1 x t  1
 1
1  1   1
 lim   1  if p  1
t  1  p  t p 1
   p 1
 if p  1

Thus the integral is convergent for p > 1, divergent for p  1.
106
Type II Improper Integrals

Example 39
Find the area A of the region lying under y = 1/ x and
above the x-axis between x = 0 and x = 1.
Solution
As x  0+, y  , i.e. y is unbounded, so we cannot
use the earlier definition of the definite integral.
But consider the area between x = t and x = 1:
1 dx 1
A(t )   2 x  2(1  t )
t x t

Then it is reasonable to take


A  Lim A(t )  Lim 2(1  t )  2 units2
t 0 t 0

This illustrates how improper integrals of type II are
defined. 107
Type II Improper Integrals: Definition

t
• If a f ( x)dx exists for every t  [a, b) and f is possibly
b t
unbounded near b, then a f ( x)dx  Lim a f ( x)dx
t b 
provided this limit exists as a finite number.
b
• If t f ( x)dx exists for every t  (a, b] and f is possibly
b b
unbounded near a, then a f ( x)dx  Lim t f ( x)dx
t a 
provided this limit exists as a finite number.

• If f has an infinite discontinuity at c, where a < c < b,

c
and both a f ( x)dx and c f ( x)dx are convergent,
b

b c b
then we define:

a
f ( x)dx a
f ( x)dx  
f ( x)dx
c

108
Example 40
dx
3
Find I  0
x 1

NOTE: If we failed to notice the discontinuity at x = 1 we might

say 3 dx  ln | x  1 | 3  ln 2  ln 1  ln 2
0 x 1 0
… which is WRONG!
From now on you must decide, by looking at the integrand,
whether integrals are proper or improper! 109
Example 41
5 dx
Evaluate 2 x2

110
A Comparison Test for Improper Integrals
Sometimes we cannot find the exact value of an improper
integral but we can find whether or not it is convergent.

Suppose f and g are continuous functions with

f ( x)  g ( x)  0 for x  a.
 
• If 
f ( x )dx is convergent then
a  a
g ( x )dx is
convergent.
 
• If
a 
g ( x )dx is divergent then a
f ( x )dx is
divergent.

A similar theorem is true for Type II integrals.

111
Example 42


 x2
Show that the integral e dx is convergent.
0

112