CHAPTER 5.

INTEGRALS

§ 5.1, 5.2. AREAS, DISTANCES, AND DEFINITE INTEGRALS

1. Introducing the Definite Integral as the Area of a Region

Let y = f (x)  0 be a function defined in a closed interval [a, b]. Let R be the region under the
graph of f (x) and above the x-axis in the interval [a, b].

ba
Let h = , and let xi = a + ih, i = 0, 1, 2, …, n. Then x0 = a, x1, x2, …, xn = b, called mesh
n
points, subdivide the interval [a, b] into n subintervals each with length h. Taking an arbitrary
value xi* in each interval [xi−1, xi], i = 1, 2, …, n, the sum

S(n) = f (x1*)h + f (x2*)h + … + f (xn*)h = h(f (x1*) + f (x2*) + … + f (xn*)),

called a Riemann Sum, can be reasonably regarded as an approximation of the "area" of the
region R.

When n approaches infinity, the limit of the Riemann sums, if it exists, is defined to be the area
of the region R. This limit is called the definite integral of f (x) on the interval [a, b] and denoted
b
by  f ( x )dx , where a is the lower limit, b is the upper limit, f (x) is the integrand, and x is the
a
variable of integration.

Since a definite integral is a number, the variable of integration can be replaced by any other
symbol. The variable of integration of a definite integral is a dummy variable.

If f (x)  0 in [a, b], we can still define the definite integral of function f (x) on interval [a, b] in
the same way, but the area of the region under the x-axis and above the graph of f (x) is defined
to be the negation of the definite integral. If some part of the graph of f (x) is above the x-axis,
and some part of the graph of f (x) is under the x-axis in [a, b], we have to calculate the area of
each part separately and then add them up. The sum is the area between the graph of f (x) and
the x-axis on this interval.

Since the definite integral calculates the area of a region, if we know the area of the region, we
can find the definite integral.

Examples

R
5.1.1. Find the definite integral  R
R 2  x 2 dx , R > 0, by the area of a region.

57
MAT1320 Notes Chapter 5 Fall 2019

The graph of the function y = R 2  x 2 is the upper half of a circle centered at the origin with
radius R. This definite integral is the area of the half-disk under the upper half of this circle and
above the x-axis. The area of this half-disk is A = R2 / 2. Hence

R  R2
 R
R 2  x 2 dx 
2
.

4
5.1.2. Find 0
(2 x  3)dx .

The graph of the function y = 2x – 3, 0  x  4, is a line segment joining point (0, −3) and point
(4, 5). This line segment intersects the x-axis at point (1.5, 0).

Y
(4, 5)

 X
(1.5, 0)
O
 (0, −3)

When 0  x < 1.5, 2x – 3 < 0; when 1.5 < x  4, 2x – 3 > 0. The area between the graph of 2x – 3
1 3 9
and the x-axis in interval [0, 1.5] is  3   = 2.25, and the area between the graph of 2x – 3
2 2 4
1 5 25
and the x-axis in interval [1.5, 4] is  5   = 6.25. Note that the integral is the negation
2 2 4
4
of the area of a region under the x-axis, i.e, 
1.5
(2 x  3)dx = −2.25. Hence,

4 1.5 4
 0
(2 x  3)dx   (2 x  3)dx   (2 x  3)dx = 6.25 – 2.25 = 4.
0 1.5

5.1.3. Suppose the graph of a function y = f (x), 0  x  9, is given in the following figure:

58
MAT1320 Notes Chapter 5 Fall 2019

y = f (x)
0 1 2 3 4 5 6 7 8 9
X

By calculating the area of the regions between the graph of this function and the x-axis, we have

2 5 9 9
0
f ( x)dx  2 , 
2
f ( x)dx  3 , and 
5
f ( x)dx  5.5. Hence 
0
f ( x)dx  4.5.

The following example uses the definition of the definite integral to find the area of a curved
region:

5.1.4. Find the area of the region under the graph of the function y = x2 and above the x-axis in
interval [0, 1].

1
This area is given by the definite integral 0
x 2dx .

1
Subdivide the interval [0, 1] into n subintervals with equally spaced mesh points x0 = 0, x1 = ,
n
2 1 i
x2 = , …, xn = 1. Then h = . Let xi* = xi = , i = 1, 2, …, n.
n n n

1  1   2  
2 2 2
n 1 1 
1 2
1  22  ...n2   lim
1
0        ...      lim  n( n  1)(2n  1) 
2
x dx = lim
n  n  n
   n  n  
n  n 3 n  n 3 6

2n 3  3n 2  n 1
 lim  .
n  6n 3 3

2. Properties of the Definite Integral

a
(i)  a
f ( x)dx = 0.

b
(ii)  a
cdx  c(b  a) .

59
MAT1320 Notes Chapter 5 Fall 2019

b b b
(iii)  a
( f ( x)  g ( x))dx   f ( x)dx   g ( x)dx .
a a

b b
(iv)  a
cf ( x)dx  c  f ( x)dx .
a

a b
(v) b
f ( x)dx    f ( x)dx .
a

b c c
(vi) a
f ( x)dx   f ( x)dx   f ( x)dx .
b a

b b
(vii) If f (x)  g(x) for all a  x  b, then  a
f ( x)dx   g ( x)dx .
a

This property has the following immediate corollaries:

b
(vii-a) If f (x)  0 for all a  x  b, then 
a
f ( x)dx  0 .

b
(vii-b) If m  f (x)  M for all a  x  b, then m(b  a)   f ( x)dx  M (b  a ) .
a

10 4 10
Example 5.1.5. Find 
4
(3 f ( x )  2)dx if  1
f ( x )dx = −2 and 
1
f ( x )dx = 3.

4
10
(3 f ( x )  2)dx  3  10

1
4
f ( x)dx   f ( x)dx  2  dx = 3(3 + 2) – 12 = 3.
1  10

5 10 10 5
Example 5.1.6. If  1
f ( x )dx = 5, 3
f ( x )dx = 7, and 
1
f ( x )dx = 10, find  3
f ( x )dx .

5 3 5 5 5 3
Since  f ( x )dx   f ( x )dx   f ( x )dx ,  f ( x )dx   f ( x )dx   f ( x )dx .
1 1 3 3 1 1

3 10 10 3 10 10
Since  f ( x )dx   f ( x )dx   f ( x )dx ,  f ( x )dx   f ( x )dx   f ( x )dx .
1 3 1 1 1 3

5 5 10 10
Hence, 
3
f ( x )dx   f ( x )dx   f ( x )dx   f ( x )dx = 5 – 10 + 7 = 2
1 1 3

If the function v(t) is the velocity function of a particle moving along the x-axis, then the definite
b
integral  a
v (t )dt is the displacement of the particle from t = a to t = b.

60
MAT1320 Notes Chapter 5 Fall 2019

§ 5.3, 5.4. FUNDAMENTAL THEOREM OF CALCULUS

1. Definite Integral with a Variable Upper Limit

b
When the upper and lower limits are given, the definite integral 
a
f (t )dt is a number. This
number changes when the upper limit changes. In other words, a definite integral with a variable
x
upper limit is a function of the upper limit: I (x) =  a
f (t )dt .

Example 5.2.1. Consider function f (x) = 2x. The area of the triangle between the graph of this
function and the x-axis in interval [0, a] is a2. Let a be a variable. We have a function

a
I (a) = 
0
(2 x )dx = a2.

2. Differentiating a Function Defined by a Definite Integral with Variable Limits

x
Consider function I (x) =  a
f (t )dt , where f (x) is continuous on an interval [a, b], a  x  b.
x h x x h
Then I (x + h) – I (x) = 
a
f (t )dt − 
a
f (t )dt = 
x
f (t )dt .

Let M(x, h) = max {f (t); t is between x and x + h} and let m(x, h) = min {f (t); t is between x and
x + h }. Then

1
M(x, h)h  I (x + h) – I (x)  m(x, h)h , and M(x, h) ( I (x + h) – I (x))  m(x, h). When h
h
1
approaches zero, lim M(x, h) = lim m(x, h) = f (x). Hence, lim ( I (x + h) – I (x)) = f (x), or
h0 h0 h0 h

d x
dx  a
f (t )dt  f ( x) .

Fundamental Theorem of Calculus (Part I). If f (x) is continuous on an interval [a, b] and a  x
x
 b, the derivative of the function I (x) =  a
f (t )dt is f (x).

By the properties of the definite integral and the chain rule, we also have

d a d x
dx  x
f (t )dt    f (t )dt   f ( x) ,
dx a

d h( x )  d u   du 

dx a
f (t )dt    f (t )dt     f (u )ut '  f (h( x ))h '( x ) , where u = h(x).
 du a   dx 

61
MAT1320 Notes Chapter 5 Fall 2019

d a d h( x )

dx h ( x )
f (t )dt   
dx a
f (t )dt   f (u)ut '   f (h( x ))h '( x) , where u = h(x),

d h( x )

dx g ( x )
f (t )dt 
d
dx  h( x )

a
f (t )dt  
a

g( x) 
f (t )dt = f (h(x)) h'(x) – f (g(x))g'(x).

Examples

d x 2t2
 e dt  e2 x .
2
5.2.2.
dx 0

d 1 x2
 e dx  eu .
2
5.2.3.
du u

 d u   de 
x
d ex
   a   e cos(u )  e cos( e ). .
2 2 x 2 x 2x
5.2.4. cos( t ) dt cos( t ) dt 
dx 0
 du   dx 

d 0 d arctan x sin((arctan x )2 )
dx  arctan x
5.2.5. sin(t )dt   
2
sin(t )dt  
2
.
dx 0 1  x2

x2
5.2.6.  x3
sin(t 2 )dt  2 x sin( x 4 )  3x 2 sin( x 6 ) .

3. Finding Definite Integrals by Antiderivatives

Recall that the difference between two antiderivatives of a function is a constant. If F(x) is an
x
antiderivative of f (x), then function I (x) =  a
f (t )dt  F ( x )  C . Let x = a. We have C = −F(a).
Hence,

f (t )dt  F ( x )  F (a )   F (t )t a .
x

x
a

Let x = b, and use x as the dummy variable. We have the second part of the Fundamental
Theorem of Calculus:

Fundamental Theorem of Calculus (Part II). If f (x) is continuous on an interval [a, b], then
b
 a
f ( x)dx  F (a)  F (b) , where F(x) is an antiderivative of f (x).

62
MAT1320 Notes Chapter 5 Fall 2019

b
Since f (x) = F'(x), this formula can also be written as  F '( x )dx  F (b)  F (a ) , called the Net
a

Change Theorem. I.e., the net change of function F (x) from x = a to x = b, F(b) − F(a), is the
definite integral of its derivative F'(x) from x = a to x = b.

Examples

5.2.7. From section 4.9, we see that F(x) = arctan x is an antiderivative of the function f (x) =
1
. Hence
1  x2

1 
dx  arctan x x 0  .
1

1
0 1 x 2
4

5.2.8. Suppose the derivative of function F(x) is F'(x) = 3x2 such that F(1) = 3, what is F(3)?

3
F(3) − F(1) =  3x 2dx   x 3 
3
 26 .
1 x 1

Hence, F(3) = 26 + F(1) = 26 + 3 = 29.

5.2.9. Suppose the derivative of a function f (x) is given in the following figure. Sketch the
graph of this function if f (0) = 0.

y' = f '(x)
0 1 2 3 4 5 6 7 8 9
X

By counting the area under the graph of the derivative f '(x), we see that

2
f (2) = f (0) +  0
f '( x )dx = 0 + 2 = 2,
4
f (4) = f (2) +  f '( x )dx = 2 + (−2) = 0,
2
5
f (5) = f (4) +  f '( x )dx = 0 + (−1) = −1,
4
6
f (6) = f (5) +  f '( x )dx = −1 + 1 = 0,
5
9
f (9) = f (6) +  f '( x )dx = 0 + 4.5 = 4.5.
6

63
MAT1320 Notes Chapter 5 Fall 2019

In interval (0, 2), f '(x) > 0 and decreasing, f (x) is increasing and concave down; in interval (2, 4),
f '(x) > 0 and decreasing, f (x) is decreasing and concave down; in interval (4, 5), f '(x) < 0 and
increasing, f (x) is decreasing and concave up; in interval (5, 6), f '(x) > 0 and increasing, f (x) is
increasing and concave up; in interval (6, 9), f '(x) > 0 and decreasing, f (x) is increasing and
concave down. Hence, the graph of the function f (x) looks like the following:
Y

y = f (x)

0 1 2 3 4 5 6 7 8 9
X
y' = f '(x)

§ 5.5. THE SUBSTITUTION RULE

1. The Method of Variable Substitution

Suppose we want to find an indefinite integral  f ( x )dx .

Step 1. Define an intermediate variable u = g(x).

Step 2. Multiply the integrand by the reciprocal of g'(x) and change dx to du. In other words,
1
replace dx by du .
g '( x )

f ( x) f ( x)
Step 3. Convert as a function of u: = h(u).
g '( x ) g '( x )

Step 4. Find  h(u )du = H(u) + C.

Step 5.  f ( x )dx = H(g(x)) + C.

This procedure can be summarized as

64
MAT1320 Notes Chapter 5 Fall 2019

1
 f ( x )dx   f ( x )
g '( x )
du   h(u )du  H (u )  C  H ( g ( x ))  C , where u = g(x),

f ( x)
h(u) = , and H(u) is an antiderivative of h(u).
g '( x )

From these steps we can see two rules of thumb to define the intermediate variable, which are

f ( x)
(a) should be easy to convert to a function h(u) of u, and
g '( x )

(b) the integral  h( x )du should be easy to integrate.

Otherwise, you may have chosen a wrong function g(x) or this method cannot be used to solve
this question.

Examples

5.3.1.  x x 2  1dx

Let u = g(x) = x2  1. g'(x) = 2x.

x x2  1 x2  1 u 1  1   2  3/ 2
 x x 2  1dx   du   du =  du   u1/ 2du =   u  C
2x 2 2 2  2  3 
1
 ( x 2  1)3/ 2  C .
3

5.3.2.  e 2 x dx .

Let u = −2x, u' = −2.

 1 1 1 1
 e2 x dx   e 2 x    du    eu du   eu  C   e 2 x  C .
 2 2 2 2

1 kx
In general, we have  ekx dx 
k
e + C.

1 (ln a ) x 1 x
Since ax = e(ln a)x,  a x dx   e(ln a ) x dx 
ln a
e C 
ln a
a C .

65
MAT1320 Notes Chapter 5 Fall 2019

x5
5.3.3.  x3  1
dx .

Let u = x3 + 1, u' = 3x2.

x5  1 x5 1 u 1 1
 x 1 3
dx    2  du  
x  1  3x  3 3 u
du   (u1/ 2  u 1/ 2 )du
3
1 2  2 3 2 3
  u 3/ 2  2u1/ 2   C  x  1( x3  1  3)  C  x  1( x3  2)  C .
3 3  9 9

(ln x  1) 2
5.3.4.  x
dx .

Let u = ln x, u' = 1 / x.

(ln x  1)2 1
 x
dx   (u  1)2 du   (u 2  2u  1)du  u 3  u 2  u  C
3
1
 (ln x )3  (ln x )2  ln x  C .
3

f (ln x )
This question suggests a general form: If we want to find an integral of the form  x
dx ,
1
where f (ln x) is a function of ln x, use substitution u = ln x. Then u' = , and
x

f (ln x ) f (ln x )
 x
dx  
x
xdu   f (u)du .

x
5.3.5.  1 x
dx .

Let u = 1 + x. Then u' = 1, and x = u − 1.

x u 1 2
 1 x
dx  
u
du   (u1/ 2  u 1/ 2 )du    u 3/ 2  2u1/ 2  C
 3

2 2 2
   (1  x )3/ 2  2(1  x )1/ 2  C  1  x (1  x  3)  C  1  x ( x  2)  C .
 3 3 3

x2
5.3.6. Find  x  2x  2
2
dx .

66
MAT1320 Notes Chapter 5 Fall 2019

Note that the derivative of the denominator is 2x + 2 = 2(x + 1). Write the numerator as x + 2 =
(x + 1) + 1, and separate the integrand:

x2 x 1 1
 x  2x  2
2
dx   2
x  2x  2
dx   2
x  2x  2
dx .

The first integral is easy to solve by variable substitution. Let u = x2 + 2x + 2. Then

x 1 x 1  1  1 1 1
 x  2x  2
2
dx   2  
x  2 x  2  2( x  1) 
du   du  ln ( x 2  2 x  2)  C .
2 u 2

Note that we don't need the absolute-value sign because x2 + 2x + 2 is always positive.

For the second integral, we complete the square in the denominator, and use variable substitution
with u = x + 1.

1 1 1
 x  2x  2
2
dx  
( x  1)  1
2
dx   2
u 1
du  arctan u  C  arctan( x  1)  C .

x2 1
Finally,  x  2x  2
2
dx  ln( x 2  2 x  2)  arctan( x  1)  C .
2

5.3.7.  tan xdx .

Let u = cos x, u' = − sin x. Then

sin x  sin x  1  1
 tan xdx  
cos x
dx     du    du   ln | cos x | C .
 cos x   sin x  u

2. Variable Substitution Used for Definite Integrals

The technique of variable substitution can also be used directly to definite integrals. The
differences are

(a) the limits have to be changed in step 2, and

(b) you don't have to go back to variable x in step 5.

b
Suppose we want to evaluate definite integral a
f ( x )dx .

Step 1. Define an intermediate variable u = g(x).

67
MAT1320 Notes Chapter 5 Fall 2019

Step 2. Multiply the integrand by the reciprocal of g'(x) and change dx to du. Change the limits
of integration to g(a) and g(b).

f ( x) f ( x)
Step 3. Convert as a function of u: = h(u).
g '( x ) g '( x )

h(u)du   H (u)u g ( a ) .
g (b)
Step 4. If H(u) is an antiderivative of h(u), then 
g (b)
g (a)

This procedure can be summarized as

1 f ( x)
h(u)du   H (u )u  g ( a ) , where u = g(x), h(u) =
b g (b) g (b)
a f ( x )dx    
g (b)
duf ( x) , and
g (a) g '( x ) g ( a ) g '( x )
H(u) is an antiderivative of h(u).

Examples

1
5.3.8. 0
x 4  3xdx .

1
Let u = 4 – 3x. Then u' = −3, and x = (4 – u).
3

1 1  1  1 11 1 1
0
x 4  3xdx   x 4  3x   du    (4  u) udu    (4u1/ 2  u 3/ 2 )du
4
 3  3 43 9 4
1
1 8 2  94
   u 3/ 2  u5/ 2   .
9 3 5 u 4 135

e (ln x )2
5.3.9.  1 x
dx .

1
Let u = ln x. Then u' = .
x
1
e (ln x ) 2 1 (ln x )
2
1 1  u3  1
 dx   xdu   (ln x ) du   u du     .
2 2
1 x 0 x 0 0
 3  u 0 3

3/ 2 1
5.3.10.  0
9  x2
dx .

Let u = x / 3, u' = 1 / 3.

68
MAT1320 Notes Chapter 5 Fall 2019

1 1 1 1 
du   arcsin u u 0 
3/ 2 3/ 2 1/ 2 1/ 2
 dx   dx   (3)du  
1/ 2
.
0
9  x2 0
x
2 0
 x
2 0
1 u2 6
3 1   3 1  
3 3

Using variable substitution for definite integrals, we have two important formulas for odd and
even functions:

If f (x) is an odd function, then, using u = x,

a 0 a 0 f ( x) a
 a
f ( x)dx  
a
f ( x)dx   f ( x)dx  
0 a 1
du   f ( x)dx
0
0 a a a
  f (u)du   f ( x)dx   f (u)d (u)   f ( x)dx  0 .
a 0 0 0

If f (x) is an even function, then, using u = x,

a 0 a 0 f ( x) a
 a
f ( x)dx  
a
f ( x)dx   f ( x)dx  
0 a 1
du   f ( x)dx
0
0 a a a a
   f (u)du   f ( x)dx   f (u)d (u)   f ( x)dx  2 f ( x)dx.
a 0 0 0 0

69
MAT1320 Notes Chapter 7 Fall 2019

CHAPTER 7. TECHNIQUES OF INTEGRATION

§ 7.1. INTEGRATION BY PARTS

The method of integration by parts is used to find the integral of a product of two functions.
This method is powerful but it can only be used to solve a relatively small group of questions.
Use this method only for questions similar to examples that you see in class. Never abuse this
method!

Suppose the integrand is a product of two functions. We denote one factor as function u, and let
the other factor be the derivative of another function v, i.e., v'. Find the derivative of u, i.e., u',
and find an antiderivative of v'. i.e., v. Then use the formula

 uv ' dx  uv   u ' vdx .

Because we want to find the derivative of u and an antiderivative of v', u should have a simple
derivative and v' should be easy to integrate.

In some questions, the integrand is regarded as the product of a function u and v' = 1. Then v = x,
and

 udx  xu   xu ' dx .

Examples

7.1.1.  xe  x dx .

Let u = x, v' = e−x. Then u' = 1, v = −e−x.

 xe  x dx   xe  x   e  x dx   xe  x  e  x  C  e  x ( x  1)  C .

7.1.2.  x cos xdx .

Let u = x, v' = cos x. Then u' = 1, v = sin x.

70
MAT1320 Notes Chapter 7 Fall 2019

 x cos xdx  x sin x   sin xdx  x sin x  cos x  C .

7.1.3.  x ln xdx .

1 2
Let u = ln x, v' = x . Then u' = , v = x 3/ 2 .
x 3

2 3/ 2 2 2 4
 x ln xdx 
3
x ln x   x1/ 2dx  x 3/ 2 ln x  x 3/ 2  C .
3 3 9

In general, an integral of the type  x k ln xdx , k  −1, can be solved by integration by parts with
u = ln x and v' = xk:

1 k 1  1 k 1   1  1 k 1 1
 x k ln xdx 
k 1
x ln x   
 k  1
x    dx 
 
x k  1
x ln x 
k  1  x k dx

1 k 1 1
 x ln x  x k 1  C .
k 1 (k  1) 2

You don't have to memorize this formula, but you should know how to use this method to solve
ln x ln x
questions like  dx ,  x ln xdx ,  2 dx , or  x ln xdx .
2

x x

Sometimes, we may have to use integration by parts more than once.

7.1.4.  x 2e 2 x dx .

1 2x
Let u = x2, v' = e2x. Then u' = 2x, v = e .
2

1 2 2x 1  1
 x 2e2 x dx 
2
x e   2 x  e2 x  dx  x 2e2 x   xe2 x dx .
2  2

In the integral on the right-hand side, use integration by parts again. Let u = x, v' = e2x. Then
1
u' = 1, v = e2x.
2

1 2 2x 1 1 1  1 1 1
x e   xe2 x dx  x 2e2 x   xe2 x   e2 x dx   x 2e2 x  xe 2 x  e 2 x  C
2 2 2 2  2 2 4
1
 (2 x 2  2 x  1)e2 x  C .
4

71
MAT1320 Notes Chapter 7 Fall 2019

The following examples use v' = 1:

7.1.5.  ln xdx.
1
Let u = ln x, v' = 1. Then u' = , v = x.
x

 ln xdx  x ln x   dx  x ln x  x  C.

7.1.6.  arcsin xdx .

1
Let u = arcsin x, v' = 1. Then u' = , v = x.
1  x2

x
 arcsin xdx  x arcsin x   dx  x arcsin x  1  x  C .
2

1 x 2

Note that, with variable substitution t = 1 – x2, t' = −2x,

x x  1  1 1
 1 x 2
dx     dt   
1  x  2 x 
2 2 t
dt   t  C   1  x 2 .

7.1.7.  (ln x ) 2 dx .

2ln x
Let u = (ln x)2, and v' = 1. Then u' = and v = x.
x

 (ln x )2 dx  x (ln x ) 2  2  ln xd x  x (ln x ) 2  2 x ln x  2 x  C .

Some questions use integration by parts with variable substitution:

7.1.8.  x 3 cos( x 2 )dx .

 1  1
Let u = x2. Then u' = 2x.  x 3 cos( x 2 )dx   x 3 cos( x 2 )   du   u cos udu .
 2x  2

With integration by parts as in Example 7.1.2, we have

72
MAT1320 Notes Chapter 7 Fall 2019

1
 x 3 cos( x 2 )dx  ( x 2 sin( x 2 )  cos( x 2 )) + C.
2

7.1.9.  e  x dx .

Let u = x . Then u' =

1
2 x
.  e  x dx   e  x
 2 x  du  2 ue u
du .

By integration by part as in example 7.1.1, we have  e  x dx  2e x ( x  1)  C

The following question uses integration by parts in a recursive way:

e
x
7.1.10. sin xdx .

Use integration by parts with u = sin x, v' = ex. Then u' = cos x, v = ex, and

e sin xdx  e x sin x   e x cos xdx .

x

In the second integral, use integration by parts again with u = cos x, v' = ex. Then u' = −sin x, v =
ex. Now we have

e
x
 
sin xdx = e x sin x  e x cos x   e x (  sin x) dx  e x (sin x  cos x)   e x sin xdx .

Add  e x sin xdx to both sides of this equation. Note that  e x sin xdx −  e x sin xdx is a constant
C. Then we have

2  e x sin xdx  e x (sin x  cos x )  C .

1
Finally,  e x sin xdx  e x (sin x  cos x)  C .
2

Integration by parts used for definite integrals

uv ' dx  uv x a   u ' vdx .

b b

b
The formula is
a a

Examples

73
MAT1320 Notes Chapter 7 Fall 2019

 /4 x
7.1.11. 0 cos2 x
dx .

1
Let u = x and v' = . Then u' = 1, and v = tan x.
cos 2 x

 /4 x  /4   ln 2
 2
dx   x tan x  /4
x 0
  tan xdx   [ln | cos x |]x / 04   .
0 cos x 0 4 4 2

1
7.1.12. 0
x 2 arctan xdx .

1 1
Let u = arctan x and v' = x2. Then u' = and v = x 3 .
1 x 2
3

1
1 1 3  1 1 x3
0  
3 0 1  x 2
2
x arctan xdx  3 x arctan x  dx .
x 0

Use variable substitution u = 1 + x2, u' = 2x.

 1 u 1  1  1
du   u  ln u u 1   (1  ln 2) .
1
 
2
x 2 arctan xdx  
0 12 6 u 12 6 12 6

§ 7.2. TRIGONOMETRIC INTEGRALS

In this section, we look at three types of questions:

Type 1.  sin m x cosn x dx .

Solving strategy:

(a) When m = 2k + 1 is odd, (n may be any integer including 0), use u = cos x. Then

 sin 2 k 1 x cosn xdx    sin 2 k x cosn xdu    (1  u 2 ) k u n du .

(b) When n = 2k + 1 is odd, (m may be any integer including 0), use u = cos x. Then

 sin m x cos2 k 1 xdx   sin m x cos2 k xdu   u m (1  u 2 )k du .

(c) If both odd, you can use either way.

74
MAT1320 Notes Chapter 7 Fall 2019

1 1
(d) If both even, use formulas sin2 x = (1 − cos (2x)), cos2 x = (1 + cos (2x)), and/or
2 2
1
sin x cos x = sin (2x), and let u = 2x.
2

(There is a recursive formula to reduce the integral of sin2m x cos2n x to an integral of

sin2m−2 x cos2n x or an integral of sin2m x cos2n−2 x. This is out of the scope of this course).

Examples

7.2.1  sin 3 x cos2 xdx .

Let u = cos x. Then u' = −sin x

 1 
 sin 3 x cos2 xdx   sin 3 x cos2 x    du    sin x cos xdu    (1  u )u du
2 2 2 2

 sin x 
1 3 1 5 1 1
  u  u  C   cos3 x  cos5 x  C .
3 5 3 5

 /2
7.2.2.  0
cos2 xdx .

Let u = 2x.

 /2 1  /2 1  1 
(1  cos(2 x))dx   (1  cos u)du  u  sin u u 0  .

0
cos2 xdx  
2 0 4 0 4 4
 /2
7.2.3.  0
cos3 xdx .

Let u = sin x. Then u' = cos x.

1
 /2 1  1  1 1  1  2
0
cos3 xdx   cos3 x 
0 
 cos x 
du   (1  sin 2 x )du   (1  u 2 )du  u  u 3   .
0 0
 3  u 0 3

 /4
7.2.4. Evaluate I =  0
sin 2 x cos4 xdx .

 /4 1  /4
I=  0
sin 2 x cos4 xdx 
8 0
(1  cos(2 x))(1  cos(2 x))2 dx
1  /4

8 0
(1  cos(2 x)  cos2 (2 x)  cos3 (2 x))dx .

75
MAT1320 Notes Chapter 7 Fall 2019

With u = 2x,

1  /2 1   /2  /2  /2 
I 
16 0
(1  cos u  cos2 u  cos3 u)du     cos udu   cos2 udu   cos3 udu 
16  2 0 0 0


1   2  1
  1     .
16  2 4 3  64 48

1
This question can be solved in another way. Use the formula sin x cos x = sin (2x), and cos2 x
2
1
= (1 + cos(2x). By substitution u = 2x,
2

 /4 1  /4 2 1  /2
I=  0
sin 2 x cos4 xdx  
8 0
sin (2 x)(1  cos(2 x))dx   (sin 2 u  sin2 u cos u)du
16 0

1
16  0
 /2
sin 2 udu  
0
 /2

sin 2 u cos udu .

Let v = 2u.

 /2 1  /2 1  1 
(1  cos(2u))du   (1  cos v)dv  v  sin v v 0  .

0
sin 2 udu  
2 0 4 0 4 4

Let v = sin u.

 /2 1 1
0
sin 2 u cos udu   v 2dv  .
0 3

1  1  1
Hence, I =     .
16  4 3  64 48

Type 2.  tan m x sec n x dx .

Note that (tan x)' = sec2 x, (sec x)' = tan x sec x, and 1 + tan2 x = sec2 x. Then tan2 x = sec2 x – 1.

Solving strategy: When m = 2k + 1  1 is odd, use u = sec x, u' = tan x sec x:

 tan 2 k 1 x secn xdx   tan 2 k x secn 1 xdu   (u 2  1)k u n 1du .

When n = 2k  2, is even, use u = tan x, u' = sec2 x:

76
MAT1320 Notes Chapter 7 Fall 2019

 tan m x sec2 k xdx   tan m x sec 2( k 1) xdu   u m (1  u 2 ) k 1 du .

If both condition is true, either way can be used; if neither condition is true, simple substitution
does not work, and we have to consider other methods.

Examples

7.2.5.  tan 3 x sec xdx .

Use u = sec x. u' = tan x sec x.

1
 tan 3 x sec xdx   tan 2 xdu   (u 2  1)du  sec3 x  sec x  C .
3

7.2.6.  tan 2 x sec 4 xdx .

Use u = tan x. u' = sec2 x.

1 1
 tan 2 x sec4 xdx   tan 2 x sec2 xdu   u 2 (1  u 2 )du  tan 3 x  tan 5 x  C
3 5

Type 3.  sin mx sin nx dx,  cos mx cos nx dx,  sin mx cos nx dx .

Use formulas:

1
sin mx sin nx  (cos(m  n) x  cos(m  n) x ) ,
2

1
cos mx cos nx  (cos(m  n) x  cos(m  n) x ) ,
2

1
sin mx cos nx  (sin(m  n) x  sin(m  n) x) .
2

Example

1 1 1 
7.2.7.  sin 2 x sin 3xdx 
2  (cos x  cos(5 x))dx   sin x  sin(5 x)   C .
2 5 

77
MAT1320 Notes Chapter 7 Fall 2019

 /2 1  /2
7.2.8. 0
sin(2 x ) cos(3x )dx 
2 0
(sin(5x )  sin x )dx
 /2
1 1  1   1  2
  cos(5 x )  cos x   0     1     .
2  5  x 0 2   5   5

§ 7.3. TRIGONOMETRIC SUBSTITUTION

Trigonometric substitution is different from the substitution method in section 7.1. (This is a
kind of so-called Substitution of Type II, whereas the substitution in section 7.1 is of Type I): In
section 7.1, the substitution uses an intermediate variable u = g(x), in substitution of Type II, x is
a variable of a new variable u: x = g(u). Then

 f ( x )dx   f ( x ) g '(u)du   f ( g (u )) g '(u)du   h(u )du  H (u )  C  H ( g 1 ( x ))  C.

In step 2, dx is replaced by g'(u)du. Then h(u) = f (g(u))g'(u) and H(u) is an antiderivative of

h(u). Finally, use the inverse g−1(x) to go back to variable x. Therefore, g(u) must be a one-to-
one function. If the integral is a definite integral, we have to change the limits from values of x
to values of u.

Some integrals involving a2 – x2, a2 + x2, or x2 – a2 may be solved by this method.

(a) Questions involving a2 – x2 may be solved by substitution x = a sin u, where a > 0, and
 
 u .
2 2

Then a2 – x2 = a2(1 – sin2 u) = a2 cos2 u. Note that cos u  0 in this interval. Hence a2  x2
= a cos u.

(b) Questions involving a2 + x2 may be solved by substitution x = a tan u, where a > 0, and
 
 u .
2 2

Then a2 + x2 = a2(1 + tan2 u) = a2 sec2 u. Note that sec u  0 in this interval. Hence a2  x2
= a sec u.

(c) Questions involving x2 – a2 may be solved by substitution x = a sec u, where a > 0, and
 3
0  u  or   u  .
2 2

Then x2 – a2 = a2(sec2 u − 1) = a2 tan2 u. Note that tan u  0 in these intervals. Hence x2  a2

= a tan u.

78
MAT1320 Notes Chapter 7 Fall 2019

Examples

a
7.3.1.  0
a 2  x 2 dx .

 
Let x = a sin u,  u . Then x' = a cos u and a 2  x 2 = a cos u. Note that cos u  0 in
2 2

this interval. When x = 0, u = arcsin 0 = 0; when x = a, u = arcsin 1 = .
2

a  /2 a2  /2  a2  /2  a2
0
a 2  x 2 dx  a 2 
0
cos2 udu 
2 
0
(1  cos(2u ))du 
4

0
cos(2u )du 
4
.

 /2 1  1
Note that, with variable substitution v = 2u, 0
cos(2u)du  
2 0
cos vdv  (sin   sin 0)  0 .
2

1/ 2 x2
7.3.2. 0
1  x2
dx .

 
Let x = sin u,  u . Then x' = cos u and 1  x 2 = cos u  0. When x = 0, u = arcsin 0 =
2 2
1 1 
0; when x = , u = arcsin = .
2 2 6

1/ 2 x2  /6 sin 2 u  /6 1  /6
0
1  x2
dx  
0 cos u
(cos u )du   sin 2 udu   (1  cos(2u))du
0 2 0

 /6
1 1  1  3
 u  sin(2u )       0.045293.
2 2  u 0 2  6 4 

1
7.3.3.  x 2
x2  4
dx .

  2 1 1 cos u
Let x = 2 tan u,  u . x' = .    0.
2 2 cos 2 u x 4
2
2 tan  12 2

1  1   cos u   2  1 cos u 1
 x 2
x 42
dx    2   2  du  
 4 tan u   2   cos u 
2
4 sin u
du  
4sin u
C .

79
MAT1320 Notes Chapter 7 Fall 2019

x/2 x  
Since tan u = x / 2, sin u =  . Note that in interval  u , sin u and
1  ( x / 2) 2
x 4
2 2 2
tan u are of the same sign.

1 x2  4
 x2 x2  4
dx  
4x
C .

1 1
7.3.4. 
0 (1  x 2 ) 2
dx .

 
Let x = tan u,  u . Then x' = sec2 u. When x = 0, u = arctan 0 = 0; when x = 1, u =
2 2

arctan 1 = .
4
2
1 1  /4 1  / 4 sec u  /4
0 (1  x 2 )2  0 (1  tan 2 u)2  0 sec4 u  0 cos udu
2 2
dx (sec u ) du du
 /4
1  /4 1 1   1
  (1  cos(2u))du   u  sin(2u)  =  .
2 0
2 4  u 0 8 4

1
7.3.5.  x 2
x2  1
dx .

 3 sin u
Let x = sec u, 0  u  or   u  . Then x' = , and x 2  1 = tan u. Note that tan u 
2 2 cos2 u
0 in these intervals.

1  cos u   sin u  x2  1
 x2 x2  1
dx   cos2 u    du   cos udu  sin u  C 
 sin u   cos u 
2
x
C .

Note that some questions involving a2  x2 , x 2  a 2 , or x 2  a 2 can also be solved by

variable substitution:

Use u = x2 – 1, u' = 2x. We have

x3 1 u 1 12  1 2 x2  1 2
 dx 
2 u
du   u u  2 u   ( x  1)  ( x  1)  C  ( x  2)  C .
3/ 2 2 1/ 2

x2  1 23  3 3

You don't need to use trigonometric substitution for these questions.

80
MAT1320 Notes Chapter 7 Fall 2019

§ 7.4. INTEGRATION OF RATIONAL FUNCTIONS BY PARTIAL FRACTION

Partial Fractions

The method of partial fraction is used to find the integral of proper rational functions. If the
integrand is an improper rational function, use long division to write it as the sum of a
polynomial and a proper rational function.

M ( x)
Let f (x) = be a proper rational function. Factorize the denominator N(x) into a product of
N ( x)
linear factors and irreducible quadratic factors. Recall that a quadratic polynomial ax2 + bx + c is
irreducible if b2 – 4ac < 0. An irreducible quadratic polynomial cannot be factorized further to
be a product of two linear factors with real coefficients.

A1 A2
If (px – q)n is a factor of N(x), construct a sum of partial fractions 
px  q ( px  q)2
An
...  , where A1, A2, …, An are undetermined constants.
( px  q)n

If (ax2 + bx + c)n is a factor of N(x), where ax2 + bx+ c is an irreducible quadratic polynomial,
B x  C1 B2 x  C2 Bn x  Cn
construct a sum of partial fractions: 21  ...  , where
ax  bx  c (ax  bx  c )
2 2
(ax 2  bx  c)n
M ( x)
B1, C1, B2, C2, …, Bn, Cn are undetermined constants. Equate and the sum of all partial
N ( x)
fractions generated by the factors of N(x) to determine the constants. Then the integral of the
rational function is the sum of the integrals of the partial fractions.

Integrating Basic Partial Fractions

1 xB
Ignoring a constant factor, a partial fraction has the form or 2 , where the
( x  b) n
( x  bx  c ) n
quadratic polynomial x2 + bx + c is irreducible, i.e., b2 – 4c < 0.

1
To integrate , use variable substitution u = x + b. Then
( x  b) n

1 1 1
 xb
dx  ln | x  b | C , and  ( x  b) n
dx  
(n  1)( x  b) n 1
 C , n > 1.

81
MAT1320 Notes Chapter 7 Fall 2019

xB
To integrate , we first eliminate the linear term in the denominator by completing
( x  bx  c ) n
2

the square:

b 2 1 1
Let u  x  , x + bx + c = u2 + (4c – b2). Since 4c – b2 > 0, let (4c – b2) = k2. Then x2 +
2 4 4
b b
bx + c = u2 + k2, and x + B = u − + B = u + D, where D = B − . Use this substitution, the
2 2
u D
second type of partial fraction becomes 2  .
(u  k 2 ) n (u 2  k 2 ) n

u
To integrate , use variable substitution v = u2 + k2. Then
(u  k )
2 2 n

u 1 u 1
 du  ln(u 2  k 2 )  C , when n = 1, and  du   v  n dv =
u k
2 2
2 (u  k )
2 2 n
2
1 1
v1n  C    C , where n > 1.
2(1  n) 2( n  1)( u 2  k 2 ) n 1

D 1 1 u
To integrate
(u  k 2 ) n
2
, we have  u k
2 2
du  arctan  C , when n = 1.
k k

When n > 1, use trigonometric substitution, u = k tan v, then u2 + k2 = k2 sec2 v, and u' = k sec2 v.
The integral becomes

1 sec2 v 1
 (u  k )
2 2 n
du =  2n
k sec v2n
dv  2 n
k  cos2( n 1) vdv .

The last integral is solved by trigonometric integration, as discussed in Section 7.2. For instance,

 cos4 xdx 
1
4 
2 1
4 
1  cos(2 x)  dx   dx  2 cos(2 x)dx   cos2 (2 x)dx 
1 1  1 1 1 
  x  sin(2 x )   (1  cos(4 x))dx    x  sin(2 x)  x  sin(4 x)   C
4 2  4 2 8 
1
12 x  8sin(2 x)  sin(4 x)   C .
32

Examples

4 x3  6 x
7.4.1.  2 x 2  3x  2
dx .

82
MAT1320 Notes Chapter 7 Fall 2019

Use long division. We have 4x3 – 6x = (2x – 3)(2x2 + 3x – 2) + (7x – 6). Hence,

4 x3  6 x 7x  6
 2x  3  2 .
2 x  3x  2
2
2 x  3x  2

Use partial fraction:

7x  6 A B A( x  2)  B(2 x  1)
   .
(2 x  1)( x  2) 2 x  1 x  2 (2 x  1)( x  2)

Then 7x  6 = A(x + 2) + B(2x  1).

Let x = 1 / 2. 5 / 2 = (5 / 2)A. A = 1. Let x = 2, 20 = 5B, B = 4.

4 x3  6x 1 1
 2 x 2  3x  2 dx   (2 x  3)dx   2 x  1 dx  4 x  2 dx
1
 x 2  3x  ln | 2 x  1 | 4ln | x  2 |  C .
2

x 1
7.4.2.  x  2 x 2  5x
3
dx .

Factorize the denominator: x3 + 2x2 + 5x = x(x2 + 2x + 5).

x 1 A Bx  C A( x 2  2 x  5)  ( Bx  C ) x
Let    .
x( x 2  2 x  5) x x 2  2 x  5 x( x 2  2 x  5)

Then A(x2 + 2x + 5) + (Bx + C)x = x + 1, or

(A + B)x2 + (2A + C)x + 5A = x + 1.

1
Comparing the coefficients on both sides, A + B = 0, 2A + C = 1, 5A = 1. Then A = ,
5
1 3
B = − , and C = .
5 5

x 1 1 1 1 3 x
Hence,  x( x  2 x  5)
2
dx   dx   2
5 x 5 x  2x  5
dx .

Since x2 + 2x + 5 = (x + 1)2 + 4, let u = x + 1. Then 3 – x = 4 – u.

3 x 4u  4 u 
 x  2x  5
2
dx  2 2
u 4
du    2  2
u 4 u 4
du

83
MAT1320 Notes Chapter 7 Fall 2019

u 1  x 1 1
 2arctan  ln(u 2  4)  C  2arctan    ln( x  2 x  5)  C .
2

2 2  2  2

x 1 1 2  x 1 1
 dx  ln | x |  arctan    ln( x  2 x  5)  C .
2
Finally,
x( x  2 x  5)
2
5 5  2  10

1
7.4.3.  x ( x  2)
2
dx .

1 A B C Ax( x  2)  B ( x  2)  Cx 2
Let     .
x 2 ( x  2) x x 2 x  2 x 2 ( x  2)

Then Ax(x – 2) + B(x – 2) + Cx2 = 1.

1 1
Let x = 2. 4C = 1, C = . Let x = 0. −2B = 1. B = − .
4 2

1
Let x = 1. –A – B + C = 1. A = −B + C − 1 = − .
4

1 1  1 2 1  1 2 
 x ( x  2)
2
dx  
4
  2 
 x x
 dx    ln | x |   ln | x  2 |   C .
x2 4 x 

x3  x  1
7.4.4.  x ( x 2  1) 2
dx .

x 3  x  1 A Bx  C Dx  E
Let    .
x ( x 2  1) 2 x x 2  1 ( x 2  1) 2

Then A(x2 + 1)2 + (Bx + C)(x3 + x) + (Dx + E)x = x3 + x + 1.

Let x = 0. We see that A = 1.

(x4 + 2x2 + 1) + Bx4 + Bx2 + Cx3 + Cx + Dx2 + Ex = x3 + x + 1,

(1 + B)x4 + Cx3 + (2 + B + D)x2 + (C + E)x + 1 = x3 + x + 1.

Then B = −1, C = 1, 2 + B + D = 1 + D = 0, D = −1, C + E = 1 + E = 1, E = 0.

Hence,

84
MAT1320 Notes Chapter 7 Fall 2019

1 1 x 1 x 
 x ( x  1)
2 2
dx     2  2  2
 x x  1 x  1 ( x  1) 
2 
dx

1 1
 ln | x |  ln( x 2  1)  arctan x  C .
2 2( x  1)
2

§ 7.5. TECHNIQUES OF INTEGRATION

Typical type of questions:

f (ln x )
(i)  x
dx . Solved by substitution u = ln x.

(ii)  x k f ( x )dx , where f (x) = ln x or arctan x. Use integration by parts. Let

u = f (x) and v' = xk.

Examples

ln x
7.5.1.  x
dx .

1
Let u = ln x. Then u' = , and
x

ln x ln x 2 2
 x
dx  
x
xdu   udu  u 3/ 2  C  (ln x )3/ 2  C .
3 3

7.5.2.  x arctan xdx .

1 1
Use integration by parts. Let u = arctan x, v' = x. Then u' = and v = x 2 .
1 x 2
2

x2 1 x2 x2 1  1 
 x arctan xdx  2 arctan x  2  1  x 2 dx  2 arctan x  2  1  1  x 2  dx
x2 1
 arctan x   x  arctan x   C .
2 2

The following examples use a combination of different methods:

85
MAT1320 Notes Chapter 7 Fall 2019

7.5.3.  sec xdx .

1
First,  sec xdx  
cos x
dx . Let u = sin x. Then u' = cos x, and
1 1 1
 cos x
dx   2
cos x
du  
1  u2
du .

1 1 1 1 
Use partial fraction,     . Hence,
1 u 2
2 1 u 1 u 

1 1 1 1  1 1 u 1  1  sin x 
 sec xdx  
1 u 2
du   
2  1 u
du   du   ln
1 u  2 1 u
 C  ln   C .
2  1  sin x 

1  sin x
Because is non-negative, we don't need the absolute-value sign. This result can be
1  sin x
simplified further. Multiply both top and bottom by 1 + sin x. We have

1  1  sin x  1  sin x (1  sin x) 2 1  sin x

ln    ln  ln  ln  ln | sec x  tan x | .
2  1  sin x  1  sin x 2
cos x cos x

Hence,  sec xdx  ln | sec x  tan x |  C .

(This question can also be solved by a trickier, but easier, method. See textbook p.483).

 /6
7.5.4.  0
tan 2 x sec xdx .

 /6  /6 sin 2 x
 tan x sec xdx  
2
dx . Use u = sin x, u' = cos x. Then
0 0 cos3 x

 /6  /6 sin 2 x 2
 / 6 sin x 1/ 2 u2
0
tan 2 x sec xdx  
0 cos3 x
dx  0 cos4 x du  0 (1  u 2 )2 du . This integral can be solved
by partial fraction:

1/ 2 u2 1  1/ 2 1 1/ 2 1 1/ 2 1 1/ 2 1 
0 (1  u 2 )2
du   
4  0 1 u
du  
0 1 u
du  
0 (1  u ) 2
du  
0 (1  u ) 2
du 

1/ 2
1  1 u 1 1  1 1 2 1 1
 ln      ln  2     ln 3 .
4  1  u 1  u 1  u  u 0 4  3 3 3 4

86
MAT1320 Notes Chapter 7 Fall 2019

§ 7.7. APPROXIMATE INTEGRATION

Since the definite integral is defined as the limit of a sequence of Riemann sums, we can use a
b
Riemann sum as an approximation of the definite integral  a
f ( x )dx .

ba
Let h = , and xi = a + ih, i = 0, 1, 2, …, n.
n

If xi* = xi−1, i = 1, 2, …, n, we have the left sum

L(n) = h (f (x0) + f (x1) + … + f (xn−1)).

If xi* = xi, i = 1, …, n, we have the right sum

R(n) = h (f (x1) + f (x2) + … + f (xn)).

If f (x) is increasing on [a, b], then the right sum gives an overestimate, and the left sum gives an
underestimate. If f (x) is decreasing on [a, b], then the right sum gives an underestimate, and the
left sum gives an overestimate.

We may also take the average of the left sum and the right sum as a (probably) better
approximation. This is called the trapezoidal rule:

1 1
T(n) = (L(n) + R(n)) = h (f (x0) + 2f (x1) + 2f (x2) + … + 2f (xn−1) + f (xn)).
2 2

In an interval [xi−1, xi], the trapezoidal rule uses the area of a trapezoid with the secant line
joining points (xi−1, f (xi−1)) and (xi, f (xi)) as the top side (trapezoid ABCD) as shown in the
following figure:

C C
f (x)

D f (x) D

A B A B
xi−1 xi X X
xi−1 xi

87
MAT1320 Notes Chapter 7 Fall 2019

From this figure, we can see that the trapezoidal rule gives an overestimate when f (x) is concave
up in the interval of integration, and it gives an underestimate when f (x) is concave down in the
interval of integration

1
If we take the midpoint of each subinterval [xi−1, xi], xi* = (xi−1 + xi), then we have the
2
midpoint rule:

M(n) = h (f (x1*) + f (x2*) + … + f (xn*)).

In an interval [xi−1, xi], the midpoint rule uses the area of the rectangle of height f (xi*) (rectangle
ABCD) to approximate the area under the graph of f (x), the area of this rectangle equals the area
of the trapezoid with the tangent line of f (x) at x = xi* as the top side (trapezoid ABEF) as shown
in the following figure:

f (x) E E
D C D C
F F
f (x)

A B A B
X X
xi−1 xi* xi xi−1 xi* xi

From this figure, we can see that the midpoint Rule gives an underestimate when f (x) is concave
up in the interval of integration, and it gives an overestimate when f (x) is concave down in the
interval of integration.

b
There is another more accurate formula to approximate a definite integral 
a
f ( x )dx , called the
ba
Simpson's rule. Let n = 2m be an even number, and let h = . Use mesh points x0 = a, x1 = a
n
+ h, x2 = a + 2h, …, xn = a + nh = b to subdivide the interval of integration [a, b] into n
subintervals. The Simpson's rule uses the sum

1
S(n) = (T(m) + 2M(m))
3
h
= ( f ( x0 )  4 f ( x1 )  2 f ( x2 )  ...  4 f ( x2 k 1)  2 f ( x2 k )  ...  4 f ( x n 1)  f ( x n))
3

b
to be an approximation of definite integral  a
f ( x )dx .

88
MAT1320 Notes Chapter 7 Fall 2019

4 1 42 1
Example 7.1.1. Suppose we want to estimate  2 x
dx with n = 6. Then h =
6
 .
3

1 7 2 8 3 4 10
The mesh points are x0 = 2, x1 = 2   , x2 = 2   , x3 = 2   3 , x4 = 2   x5 =
3 3 3 3 3 3 3 ,
5 11 6
2  , x6 = 2   4 .
3 3 3

Hence,

1 1 1 1 1 1 1 
R(6) =         0.65321
3  x1 x2 x3 x4 x5 x6 

1 1 1 1 1 1 1 
L(6) =         0.73654
3  x0 x1 x2 x3 x4 x5 

1 1 2 2 2 2 2 1 
T(6) =          0.69488
6  x0 x1 x2 x3 x4 x5 x6 

1 1 4 2 4 2 4 1 
S(6) =          0.69317
9  x0 x1 x2 x3 x4 x5 x6 

1
To use the midpoint rule, we have to find xi* = (xi−1 + xi), i = 1, 2, 3, 4, 5, 6. Then
2
1 7  13 1  7 8  15 1  8 9  17 1  9 10  19
x1*   2    , x2 *      , x3*      , x4 *      ,
2 3 6 2  3 3 6 23 3 6 23 3  6
1  10 11  21 1  11 12  23
x5*      , x6 *      .
2 3 3  6 2 3 3  6

1 1 1 1 1 1 1 
M(6) =         0.69228.
3  x1 * x2 * x3 * x4 * x5 * x6 * 

When f (x) is monotonic on [a, b], | R(n) – L(n) | gives an error bound of the estimation. Since
| R(n) – L(n) | = h | f (b) – f (a) |, we can use f (b) – f (a) to find appropriate value of h that
guarantees to meet the error requirement.

89
MAT1320 Notes Chapter 7 Fall 2019

1
Example 7.1.1. (continued) Since function f (x) = is increasing in interval [2, 4], if we want to
x
4 1 −3
find an approximation of  2 x
dx using the left or the right sum with an error less than 10 , we

1/ 4  1/ 2 0.25
should have h <   250 .
0.001 0.001

90
MAT1320 Notes Chapter 4 Fall 2019

CHAPTER 4. APPLICATIONS OF DERIVATIVES

§ 4.1 - 4.3, 4.5. USE DERIVATIVES TO STUDY PROPERTIES OF FUNCTIONS

1. Critical Numbers of a Function

The Mean Value Theorem

If f (x) is defined and continuous on [a, b], and it is differentiable in (a, b), then there exists a
f ( b)  f ( a )
value c in (a, b) such that f '(c) = .
ba

In other words, the tangent line at x = c is parallel to the secant line joining points (a, f (a)) and
(b, f (b)).

Roll's Theorem

Suppose f (x) is defined and continuous on a closed interval [a, b], it is differentiable in (a, b),
and f (a) = f (b). Then there is c in (a, b) where f '(c) = 0.

Critical Number

A value x = a is a critical number of a function y = f (x) if (i) a is in the domain of f (x), and
(ii) f '(a) = 0 or f '(a) does not exist.

Local Maximum and Local Minimum

Let x = c be a critical number of function f (x) in an interval (a, b), a < c < b. If f (x) attains a
local maximum or a local minimum at x = c, then c must be a critical number of f (x).

On the other hand, at a critical number c of function f (x), this function does not have to attain a
local maximum or a local minimum.

We may use either the first derivative test or the second derivative test to determine whether this
function attains a local maximum, or a local minimum, or neither, at a critical number x = c.

First Derivative Test:

Let c be a critical number of f (x).

If there exists an interval (a, b), a < c < b, such that f '(x) < 0 when a < x < c, and f '(x) > 0 when
c < x < b, then f (x) attains a local minimum at x = c.
91
MAT1320 Notes Chapter 4 Fall 2019

If there exists an interval (a, b), a < c < b, such that f '(x) > 0 when a < x < c, and f '(x) < 0 when
c < x < b, then f (x) attains a local maximum at x = c.

Otherwise, this function does not attain a local maximum or local minimum at this critical
number.

Second Derivative Test:

Let c be a critical number of f (x). Suppose f "(c) exists.

If f "(c) > 0, then f (x) attains a local minimum at x = c.

If f "(c) < 0, then f (x) attains a local maximum at x = c.

If f "(c) = 0, this method fails.

2. First Derivative Analysis

Purpose: Use the first derivative of a function f (x) to determine the intervals where this function
is increasing, and the intervals where this function is decreasing. Find the local / global
maximum / minimum of this function, if any. Find the global maximum and the global
minimum of this function in its entire domain.

Steps:

(1) Find the first derivative f '(x).

(2) Find the critical numbers.

(3) The critical numbers subdivide the domain of the function into a number of subintervals.
Take a particular value in each of these subintervals, and plug it into the first derivative of the
function to determine the sign of the first derivative in this subinterval.

(4) Use the sign of the derivative in each of these subintervals to determine the intervals where
the function is increasing or decreasing.

(5) Use the first derivative test to find the values of x where this function attains a local
maximum or a local minimum, if any.

(6) Find the values of the function on the ends of intervals of its domain, or study the behaviour
of the function when the variable approaches an open end of a domain interval, to find the global
extrema.

3. Second Derivative Analysis

92
MAT1320 Notes Chapter 4 Fall 2019

Purpose: Use the second derivative to find interval(s) where the graph of the function is concave
up or concave down, and inflection points.

Steps:

(1) Find the second derivative f "(x).

(2) Find values of x that are (a) in the domain of f (x), and (b) f "(x) = 0 or f "(x) does not exist.
(Note that, these numbers are NOT critical numbers! No name is assigned to these numbers.)

(3) The numbers found in step 2 subdivide the domain of the function into a number of
subintervals. Take a particular value in each of these subintervals, and plug it into the second
derivative of the function to determine the sign of the second derivative in this subinterval.

(4) Use the sign of the second derivative in each of these subintervals to determine the intervals
where the graph of the function is concave up or concave down.

(5) Find all inflection points, if any.

4. Graph Sketching

Using the properties obtained by the first and second derivative analysis, combined with the
information about the horizontal / vertical asymptotes, and the coordinates of some particular
points, we can sketch the graph of the function.

Examples

4.1.1. A function y = f (x) defined and continuous for all x  0 has the following properties:

 0,  3  x  0 or x  0
  0, x  3

(i) f '( x )  .
  0, x  3
 not defined, x0
  0, 5 x  0
  0, x  5

(ii) f "( x )  .
  0, x  5 or x  0
 not defined, x0
(iii) lim f ( x )  lim f ( x )  0 ;
x  x 

(iv) lim f ( x )  , lim f ( x )   .

x 0 x 0
(v) f (−5) = 1, f (−3) = 2, f (1) = 1.

The domain of this function consists of two intervals (−∞, 0) and (0, ∞).

93
MAT1320 Notes Chapter 4 Fall 2019

According to the conditions (i) and (ii), construct the following diagrams:

y' + − −

y −3 0

y" + − +

y −5 0

Function f (x) is increasing when x < 3, and it is decreasing when −3 < x < 0 or x > 0. The
function attains a local maximum at x = −3. By condition (iv), this function does not have the
global maximum or global minimum.

The graph of the function f (x) is concave up when x < −5 or x > 0, and it is concave down when
−5 < x < 0. It has an inflection point at x = −5.

The shape of the graph in each interval looks like the following:

−5 −3 0
 

Condition (iii) tells us that x = 0 is a vertical asymptote, and condition (iv) tells us that y = 0 is a
horizontal asymptote. By the particular values given in condition (v), we have three points on
the graph: (−5, 1), (−3, 2) and (1, 1). The graph of this function looks like the following:

 (−3, 2)
local maximum
 (−5, 1)  (1, 1)
inflection point HA y = 0
X
O

VA x = 0

4.1.2. y = 3x4 −13x3 + 15x2 defined on [1, 3].

94
MAT1320 Notes Chapter 4 Fall 2019

First derivative analysis:

y' = 12x3 – 39x2 + 30x = 3x(4x2 – 13x + 10).

13  169  160 13  3 5
Let y' = 0. x =   , 2.
8 8 4

5
Critical numbers: x = 0, , 2. Construct a diagram as follows:
4

y' − 0 + − 2 +
  
y

5 5
From the diagram, we see that y' < 0 when x < 0 or < x < 2, and y' > 0 when 0 < x < or x > 2.
4 4
This function increases in intervals (0, 5 / 4) and (2, ∞), and it decreases in intervals (−∞, 0) and
(5 / 4, 2). Function y attains two local minima at x = 0 and x = 2, and it attains a local maximum
5
at x = .
4

y(0) = 0, y(2) = 4, y(5 / 4) = 1375 / 256  5.4.

Since y(−1) = 31, y(3) = 27. This function attains its absolution maximum y(−1) = 27, and global
minimum y(0) = 0.

Second derivative analysis:

y" = 36x2 − 78x + 30.

13  169  120 13  7 1 5
Let y" = 0. x =   , . Construct a diagram as follows:
12 12 2 3

y' + − +
 
y

1 5 1 5
From the diagram, we see that y" < 0 when < x < , and y" > 0 when x < or x > .
2 3 2 3

95
MAT1320 Notes Chapter 4 Fall 2019

1 5  1  37
The graph of this function y has inflection points at x = and x = . y     2.3, y
2 3  2  16
 5  125
   4.6, The graph of this function is concave up in intervals (−∞, 1 / 2) and (5 / 3, ∞),
 3  27
and it is concave down in interval (1 / 2, 5 / 3). It has inflection points at x = 1 / 2, and x = 5 / 3.

Asymptotes: No asymptotes.

Putting together: Construct the following diagram, and we know the shape of the graph in each
of the subintervals:

0 2
    

Find some particular values of this function:

 1  37  5  1375  5  125
y(−1) = 31, y(0) = 0, y     2.3, y     5.4, y     4.6, y(2) = 4,
 2  16  4  256  3  27
y(3) = 27.

The graph of this function looks like the following:

 Y


  

 X

6x
4.1.3. f (x) = . The domain of this function is (, 1) and (1, ).
( x  1) 2

First derivative analysis:

6( x  1)
y' =  .
( x  1) 2

96
MAT1320 Notes Chapter 4 Fall 2019

Let y' = 0. Then x = −1. This function has a unique critical number x = −1. Construct the
following diagram:

y' − + −

y −1 1

This function increases in interval (−1, 1) and it decreases in intervals (−∞, −1) and (1, ∞). It
3
attains a local minimum at x = −1, y(−1) = − .
2

When x approaches 1+ or 1−, y approaches infinity, this function does not have a global
maximum. When x approaches infinity or negative infinity, y approaches 0. Hence, the local
minimum is the global minimum.

Second derivative analysis:

12( x  2)
y" = . Let y" = 0. x = −2. Construct the following diagram:
( x  1) 4

y" − + +

y −2 1

The graph of this function is concave up in (−1, 2), and (1, ∞), and it is concave down in
(−∞, −2). It has an inflection point at x = −2, y(−2) = −4 / 3.

Asymptotes: Since lim f ( x)  lim f ( x)  0 , y = 0 is a horizontal asymptote. Since

x x
lim f ( x )  lim f ( x )   , x = 1 is a vertical asymptote.
x 1 x 1

Putting together: Construct the following diagram to determine the shape of the graph in each
subinterval:

 
−2 −1 1

Find some particular values of this function:

x 2 1 0 3 7
y 4 / 3 −3 / 2 0 9/2 7/6

97
MAT1320 Notes Chapter 4 Fall 2019

The graph of this function looks like the following:

Y
 (3, 9 / 2)

x=1

(7, 7 / 6)


(0, 0)
 X

(−2, −4 / 3)  
(−1, −3 / 2)

4.1.4. y = x2ex.

This function is defined for all real numbers.

First derivative analysis:

y' = 2xex + x2ex = x(2 + x)ex.

Let y' = 0. Then we have critical numbers x = 0 and x = −2.

Construct the following diagram:

y' + − +

y −2 0

This function increases in intervals (−∞, −2) and (0, ∞), and it decreases in interval (−2, 0). It
attains a local maximum at x = −2, y(−2) = 4e−2  0.54, and a local minimum at x = 0, y(0) = 0.

When x approaches infinity, this function approaches infinity. This function does not have a
global maximum. Since y  0 for all x, the local minimum y(0) = 0 is the global minimum.

98
MAT1320 Notes Chapter 4 Fall 2019

Second derivative analysis:

y" = (2 + 2x)ex + (2x + x2)ex = (2 + 4x + x2)ex.

Let y" = 0. We have x = x1 = 2 + 2  −0.59, x = x2 = 2  2  −3.41. Construct the

following diagram:

y" − + +

y x2 x1

The graph of this function is concave up in (−∞, x2), and (x1, ∞), and it is concave down in
(x2, x1). It has inflection points at x = x1 and x = x2, y(x1) = 0.19, and y(x2)  0.38.

Asymptotes: Since lim y  0 , y = 0 is a horizontal asymptote. It does not have a vertical

x 
asymptote.

Putting together: Construct the following diagram to determine the shape of the graph in each
subinterval:

   
x2 −2 x1 0

Find some particular values of this function:

x x2  3.41 2 x1  0.59 0
y  0.19  0.54  0.38 0

The graph of this function looks like the following:

Y

99
MAT1320 Notes Chapter 4 Fall 2019

x1/ 5
4.1.5. y = .
x 1

This function is defined for all x  −1.

First derivative analysis:

1 4x
y' = ,
5 x ( x  1)2
4/5

Let y' = 0. Then x = 1 / 4. The derivative y' does not exist when x = 0 or x = −1. Since x = −1 is
not in the domain of this function, this function has two critical numbers x = 1 / 4, and x = −1.

Construct the following diagram:

y' + − +
 
y −1 0 1/4

From this diagram, we see that y' > 0 when x < −1, −1 < x < 0, or 0 < x < 1 / 4; y' < 0 when x >
1 / 4. This function increases in intervals (−∞, −1), (−1, 0), and (0, 1 / 4), and it decreases in
interval (1 / 4, ∞). It attains a local maximum at x = 1 / 4. y(1 / 4)  0.61. Since y' approaches
infinity when x approaches 0, y has a vertical tangent line at x = 0.

Second derivative analysis:

2(18 x 2  9 x  2)
y" = .
25 x 9 / 5 ( x  1)3

Let y" = 0. x = 2 / 3, −1 / 6. The second derivative y" does not exist when x = −1, and x = 0.
Construct the following diagram:

y" − + +
  
y −1 −1 / 6 0 2/3

The graph of this function is concave up in intervals (−∞, −1), (−1 / 6, 0), and (2 / 3, ∞). It is
concave down in intervals (−1, −1 / 6) and (0, 2 / 3). It has inflection points at x = −1 / 6, x = 0,
and x = 2 / 3. y(−1 / 6)  −0.84, y(0) = 0, y(2 / 3)  0.55.

Asymptotes:

100
MAT1320 Notes Chapter 4 Fall 2019

Since lim y  lim y  0 , y = 0 is a horizontal asymptote. Since lim y  , lim y   , x = −1

x  x  x 1 x 1
is a vertical asymptote.

The graph of this function looks like the following:

Y

x = −1
(0.25,0.61)
 
(0.67, 0.55)
(0, 0)  X

(−0.17,−0.84)


§ 4.4. INDETERMINATE FORMS AND L'HOPITAL'S RULES

f ( x)
An indeterminate form is a limit lim , where lim f (x) = 0 and lim g(x) = 0, symbolically
g ( x)x a x a x a

0 
called a form, or lim f (x) = ∞ and lim g(x) = ∞, symbolically called an form.
0 x a x a 

f ( x) f ( x ) lim f '( x )
L'Hopital's Rule: If lim is an indeterminate, then lim  xa .
x a g ( x) x a g ( x ) lim g '( x )
xa

1. Use the Rule Directly

Examples

x sin x sin x  x cos x cos x  cos x  x sin x

4.2.1. lim  lim  lim 2.
x 0 1  cos x x 0 sin x x 0 cos x

x  sin x 1  cos x sin x cos x 1

4.2.2. lim 3
 lim 2
 lim  lim  .
x 0 x x 0 3x x 0 6x x 0 6 6

101
MAT1320 Notes Chapter 4 Fall 2019

x sin x sin x  x cos x cos x  cos x  x sin x

4.2.3. lim  lim  lim 2.
x 0 1  cos x x 0 sin x x 0 cos x

xk
4.2.4. Consider the limit lim x , where k is any real number. If k  0, this limit is obviously
x  e

 xk kx k 1
zero. If k > 0, this is an form. Use L'Hopital's rule lim x  lim x repeatedly until the
 x  e x  e
k
x
exponent of x becomes non-positive. Then lim x = 0 for all real number k.
x  e

ln x 
4.2.5. Consider the limit lim , where k > 0. Then it is an form. Using L'Hopital's rule,
x k x  
ln x 1/ x 1
we have lim k  lim k 1  lim k  0 .
x  x x  kx x  kx

2. Product

The limit lim f ( x) g ( x) , where lim f ( x) = 0 and lim g ( x ) = ∞, is, symbolically, a 0  ∞ form.
x a xa x a

f ( x) 0
To find this limit, write lim f ( x ) g ( x )  lim to convert it to a form, or write
x a x a 1/ g ( x ) 0
g ( x) 
lim f ( x ) g ( x )  lim to convert it to an form.
x a x a 1/ f ( x ) 

Examples

x
4.2.6. Consider 0  ∞ form lim (  x ) tan .
x  2

x  x 1 x
lim (  x ) tan  lim  lim  lim 2sin 2    2
x   2 x   x x  1  x x  2
cot    csc2  
2 2 2

4.2.7. Consider 0  ∞ form lim x k ln x , where k > 0.

x 0

ln x 1/ x 1 1
lim x k ln x  lim k
 lim  k 1
 lim k
  lim x k  0 .
x 0 x 0 x x 0 kx x 0 kx k x0

3. Difference

102
MAT1320 Notes Chapter 4 Fall 2019

The limit lim( f ( x)  g ( x)) , where lim f ( x) = ∞ and lim g ( x ) = ∞, is, symbolically, an ∞ − ∞
xa xa x a
form. To find the limit, we may first convert the difference into a quotient, then use L'Hopital's
rule.

Examples

 1 1
4.2.8. lim   .

x 0 sin x x

 1 1 x  sin x 1  cos x sin x

lim     lim  lim  lim 0.
x 0 sin x  x cos x x 0 cos x  cos x  x sin x

x 0 sin x x  x0 x sin x

4.2.9. lim ( tan x  2 x sec x ) .

x ( / 2) 

  sin x 2x   sin x  2 x  cos x  2

lim  ( tan x  2 x sec x)  lim      lim   lim   2.
x ( / 2) x ( / 2)  cos x cos x  x( / 2) cos x x ( / 2)  sin x

4. Power

The limit lim f ( x ) g ( x ) , where lim f ( x) = ∞ and lim g ( x ) = 0 (the ∞0 form), or lim f ( x) = 1 and
x a xa x a xa
lim g ( x ) = ∞ (the 1∞ form). To find the limit, we take the logarithm of the function and find the
x a

limit lim ln f ( x ) g ( x )  lim g ( x ) ln( f ( x )) = L (if it exists) by L'Hopital's rule, then lim f ( x ) g ( x ) = eL.
x a x a x a

Examples

x
 a
4.2.10. lim  1   .
x 
 x

x
 a  a
This is a 1∞ form. Take the logarithm: ln  1    x ln  1   . Use the L'Hopital's rule:
 x  x

 a  a   a
ln  1     2  1  
 a  x
 lim 
x   x a
lim x ln  1    lim  lim a.
x 
 x x  1/ x x  1/ x 2 x 
1
a
x

x
 a
Then lim  1   = ea.
x 
 x

103
MAT1320 Notes Chapter 4 Fall 2019

4.2.11. lim(2

 x ) tan( x / 2) .
x 1

x
This is a 1∞ form. ln (2 – x)tan(x/2) = tan ln (2 – x).
2

 x  ln(2  x ) 1/(2  x )
lim ln(2  x ) tan( x / 2)  lim tan   ln(2  x)  lim  lim
x 1 x 1  2  x 1   x  x1   x 
cot    sin 2  
 2  2  2 
 x 
sin 2  
2
 lim  2  2.
 x1 2  x 

Then lim(2

 x ) tan( x / 2) = e2/.
x 1

4.2.12. lim( e x  x 2 )1/ x .

x 

ln( e x  x 2 )
This is a ∞0 form. ln (ex + x2)1/x = .
x

ln(e x  x 2 ) (e x  2 x ) /(e x  x 2 ) ex  2x 1  2 xe  x
lim  lim  lim x  lim  1.
x  x x  1 x  e  x 2 x  1  x 2 e  x

Then lim( e x  x 2 )1/ x = e.

x 

§ 4.7. OPTIMIZATION PROBLEMS

An optimization problem is to find the global maximum or minimum of a function.

Steps to solve an applied optimization problem:

(i) Identify the quantity to be optimized. If not given, define a symbol to represent this quantity.

(ii) Express this quantity as a function of one or more variables. If not given, define a symbol to
represent each of the variables.

(iii) If the number of variables is more than one, use conditions given in the question to
eliminate extra variables so that the function, called the objective function, is a one-variable
function.

104
MAT1320 Notes Chapter 4 Fall 2019

(iv) Find the domain of the objective function. Note that, the domain is determined by the
question, which is usually NOT the natural domain.

(v) Use calculus to find the global maximum of the objective function.

Examples

4.3.1. An open box is made by cutting four square corners of a square cardboard of dimensions
3 by 3 meters. Find the size of the square corners to be cut to maximize the volume of the box.

Let x be the side length of the corner squares. Then the volume of the box is

V = (3  2x)2x = 9x  12x2 + 4x3.

3
The domain of this function is 0  x  .
2

8  64  48 8  4 3 1
Let V' = 9  24x + 12x2 = 0, i.e., 3  8x + 4x2 = 0. Then x =   , .
8 8 2 2

1 1
By the first derivative test, V attains a local maximum at x = .V   = 2. Since V(0)
2 2
 3
= V   = 0, this local maximum is also a global maximum.
2

4.3.2. A farmer wants to use fence to enclose a rectangular region of area 800m2 against a wall.
Find the dimensions of the region that minimizes the total length of fence.

Let x be the length of the side parallel to the wall and y be the length of the side perpendicular to
the wall. Then the total length of fence is L = x + 2y. Since the area of the region is A = xy = 800,
800 800
x= and L =  2 y . The domain of this function is 0 < y < ∞.
y y

800
Let L' =  2
+ 2 = 0. Then y2 = 400, y = 20, and x = 40. By the first derivative test, we see
y
that function L attains a local minimum at x = 40. Since L approaches 0 when y approaches 0 or
infinity, this local minimum is also a global minimum.

4.3.3. A company wants to make a can of the shape of a cylinder. The material to make the top
and the bottom costs 0.004 cent per cm2, and the material to make the lateral surface costs 0.03
cent per cm2. The total cost the company to spend for each can is 6 cents. Find the radius and
the height of the can that maximize the capacity.

Let R be the radius and let H be the height of the can. Then the capacity of the can is V = R2H.

105
MAT1320 Notes Chapter 4 Fall 2019

The area of the top and the bottom is A1 = 2R2, and the area of the lateral surface is A2 = 2RH.
The total cost of the can is C = 0.04A1 + 0.03A2 = 0.08R2 + 0.06RH = 6. Hence, 0.08r2 +
0.06RH = 6. Solve this equation for H.

6  0.08R 2
H .
0.06 R

6  0.08 R 2 4
Then V =  100 R   R 3 .
0.06 3

6  0.08R 2 6
Since R > 0, and H  > 0, i.e., 0.08R2 < 6, or R <  75  5 3 , the domain of
0.06 R 0.08
the objective function is 0 < R < 5 3 .

100 6  0.08  25 4 40
Let V' = 100 − 4R2 = 0. R =  25 = 5 cm. Then H     13.3
4 0.06  5 0.3 3
cm.

Since V' > 0 when R < 5, and V' < 0 when R > 5, function V(R) attains a local maximum at R = 5.
Since limV ( R )  lim V ( R )  0 , this local maximum is a global maximum.
R 0 R 5 3

4.3.4. Find the point on the curve y2 = 2x that is closest to the point (1, 4).

Let this point be P = (x, y).

The distance between P and (1, 4) is

d= ( x  1) 2  ( y  4) 2 .

Since x = y2 / 2,

2
 y2 
 2  1  ( y  4) .
2
d=
 

To simplify out calculation, instead of finding the minimum value of d, we find the optimal value
of

2
 y2  y4 y4
D = d =   1  ( y  4) 
2 2
 y  1  y  8 y  16 
2 2
 8 y  17 .
 2  4 4
The domain of this function D(y) is  < y < .

106
MAT1320 Notes Chapter 4 Fall 2019

y2
Let D' = y3  8 = 0. Then y = 2, and x = = 2. At this point, D(2, 2) = 5.
2

Since D' < 0 when y < 2, and D' > 0 when y > 2, function D(y) attains a local minimum at y = 2.

Since lim D( y )  lim D( y )   , this local minimum is a global minimum.

y  y 

The shortest distance from (2, 2) to the parabola y2 – 2x = 0 is d = 5.

4.3.5. A right circular cylinder is inscribed in a cone with height h and base radius r. Find the
maximum volume of the cylinder.

Let x be the radius of the cylinder, and let y be the height of the cylinder. Let V be the volume of
the cylinder.

A vertical cross section from the top of the cone is shown in the following figure:

h
y

x r
y h h
The volume of the cylinder is V = x2y. By similar triangles,  . Hence, y = ( r  x ) ,
rx r r
r  x h 2
and V =  hx 2  ( rx  x 3 ) . The domain of this function V(x) is 0  x  r.
r r

h 2 1 4
Let V' = (2rx  3x2) = 0. Then x = r , y = h, and V =  r2h.


r 3 3 27

2 2 2
Since V' > 0 when x < r, and V' < 0 when x > r, V(x) attains a local maximum at x = r.
3 3 3

Since V(0) = V(r) = 0, this local maximum is also a global maximum.

4.3.6. A rectangular poster is of area 6912 cm2. Side margins are 8 cm and the top and bottom
margins are 6 cm. Find the dimensions of the poster that give the maximum printing area.

Solution. Let h be the height and let w be the width of the poster. Then the printing area of the
poster is A = (h – 12)(w – 16).

107
MAT1320 Notes Chapter 4 Fall 2019

16  6912
Since wh = 6912, A = hw – 16h – 12w + 192 = 6912 – 16h – 12w + 192 = 7104 − −
w
6912
12w, w  16. Since h   12, w  576. Therefore, the objective function is
w
16  6912
A = 7104 − − 12w, 16  w  576.
w

16  6912 16  6912
Let A' = 2
 12 = 0. We have w2 = = 9216. Then w = 96, h = 72, and A =
w 12
4800 cm2. Since A' > 0 when w < 96, and A' < 0 when w > 96, function A(w) attains a local
maximum at w = 96. Since A(16) = A(576) = 0 this local maximum is a global maximum.

§ 4.8. NEWTON'S METHOD

Newton's method is used to find a root of an equation f (x) = 0, using an initial approximation of
the root.

By the linearization of the function y = f (x) near a point x0, when x is close to x0,
f (x)  f '(x0)(x – x0) + f (x0). Suppose x* is a root of the equation f (x) = 0, i.e., f (x*) = 0, and x0
is an approximation of x*. Substituting x* for x in the linearization, f '(x0)(x* – x0) + f (x0)  0.
f ( x0 )
Hence, x*  x1 = x0 − . Then x1 is a better approximation of x*.
f '( x0 )

This formula can also be illustrated by the following figure:

f (x0)

x* x1 x0

Since f '(x0) is the slope of the tangent line at x0, we see that the distance between x0 and x1 is
f ( x0 )
. Repeatedly using this method, we can find an accurate approximation of a root of this
f '( x0 )
equation. The procedure can stop when (a) a number of iteration has been performed, or (b) f (xn)
is small enough, or (c) | xn – xn−1 | is small enough.

Examples

108
MAT1320 Notes Chapter 4 Fall 2019

4.4.1. Find an approximation of 2 accurate up to the second place after the decimal point.

This question is equivalent to find a root of the equation x2 = 2. Then f (x) = x2  2, and f '(x) =
2x.

xn 2  2
The iteration formula is xn+1 = xn − .
2 xn

Let x0 = 2. Then

22  2
x1 = 2  = 1.5;
22
1.52  2
x2 = 1.5   1.41667;
2  1.5
1.41667 2  2
x3 = 1.41667   1.41422.
2  1.41667

Since the second place after the decimal point in x2 and x3 are the same. Stop at x3.

4.4.2. Find a root of the equation 4ln x = x + 1, with x0 = 2 by Euler's method. Stop when the
difference between two consecutive approximations is less than 0.001.

4
We have f (x) = 4ln x – x – 1, and f '(x) =  1 . The iteration formula is
x
4ln xn  xn  1
xn+1 = xn − .
4 / xn  1

x0 = 2.

4ln 2  2  1
x1 = 2   2.2274;
0.25  2  1

4ln 2.2741  2.2741 1

x2 = 2.2741   2.2576;
0.25  2.2741  1

4ln 2.5763  2.5763 1

x3 = 2.5763   2.2581.
0.25  2.5763  1

Since | x3 – x2 |  0.0005 < 0.001. Let x3 be the approximation.

4.4.3. Use Euler's method to find an approximation of a root of the equation x3  x + 1 = 0

starting with x0 = −1. Stop when the value of f (x) = x3 – x + 1 < 0.0001.

109
MAT1320 Notes Chapter 4 Fall 2019

xn 3  xn  1
The iteration formula is xn+1 = xn  .
3 xn 2  1

x0 = 1, f (x0) = 1;
( 1)3  ( 1)  1
x1 = 1 −  −1.5, f (x1) = −0.875;
3( 1) 2  1
( 1.5)3  ( 1.5)  1
x2 = −1.5 −  −1.3478, f (x2) = −0.1001;
3( 1.5) 2  1
( 1.3478)3  ( 1.3478)  1
x3 = −1.3478 −  −1.3252, f (x3) = −0.00206;
3( 1.3478) 2  1
( 1.3252)3  ( 1.3252)  1
x4 = −1.3252 −  −1.3247, f (x4) = −0.00008.
3( 1.3252) 2  1

We have a very accurate root x = 1.3247 of this equation.

110