1

Chapter 4: Derivatives & Their Applications

4.1 Definitions and Properties of Derivatives

4.1.1 Definition
One of the problems that initiated the discovery of the concept of derivative is the following geometric
problem, called the tangent problem (or the slope problem). Intuitively, a tangent line is a line that touches a
curve at a point without crossing over the curve at that point. It will be formally defined in the
sequel.

The tangent (slope) problem: Let 𝑓 be a continuous function and 𝑐 be a number in the
domain of 𝑓. Let 𝑡 be the “tangent line” to the graph of 𝑓 at (𝑐, 𝑓(𝑐)).
AIM: To find slope (𝑡).

s
Q

c x

Figure 1: The Tangent Line Problem

We can find the slope of the tangent line 𝑡 as follows. Consider a number 𝑥 in the domain of 𝑓
“close” to 𝑐. Let P = (c, f(c)) and Q = (x, f(x)). We denote the difference 𝑥 − 𝑐 by h, dx or x .
Then the corresponding change (difference)in the value of the function f is denoted by  y or  f
(but not 𝑑𝑦 𝑜𝑟 𝑑𝑓),i.e.,  f =  y = f(c+ h) – f(c). Thus  y and 𝑑𝑦 are not the same while dx and

x are the same. (The meaning of 𝑑𝑦 will be given in section 4.7). Let 𝑠 be the secant line through
y
𝑃 and 𝑄. From school geometry, 𝑠𝑙𝑜𝑝𝑒(𝑆) =
x

1
 2

f (c  h )  f (c )
 Slope(S) =
h

As h  0, then Slope(s)  slope (t)

Using the language of limits, we can write the last statement as :

f (c  h )  f (c )
slope (t )  lim slope ( s)  slope (t )  lim
h0 h 0 h

The limit above is used to describe not only slope but also a number of other quantities. For instance
it is used to describe (or define) velocity ,acceleration, marginal revenue ,marginal cost, birth rates
,death rates, compound interests ,reaction rates, etc. It appears so frequently that it is given a name; it
is called a derivative.

Definition 1 (derivative at a point): Let y = f(x) be a function and suppose that 𝑐 is in the
domain of 𝑓. The number

f (c  h )  f (c )
f ' (c) : lim (1)
h 0 h

is called the derivative of 𝑓 at 𝑐 (provided that the limit exists).

Remark:

a) If the limit in equation (1) above exists, we say f is differentiable at 𝑐 (or 𝑓 has derivative at 𝑐).
b) The definition of the derivative of 𝑓 at 𝑐 can be given equivalently by
𝑓(𝑥)−𝑓(𝑐)
𝑓 ′ (𝑐) = lim𝑥→𝑐 (2)
𝑥−𝑐

This can be justified by using the substitution 𝑥 = 𝑐 + ℎ in (1).

𝑓(𝑥)−𝑓(𝑐)
c) The quotient is called difference quotient or Newton quotient.
𝑥−𝑐

Example 1: Let 𝑓(𝑥) = 𝑥 2 – 8𝑥 + 9. Compute 𝑓 ′ (2) .

x2 – 8x + 9−[22 −8(2)+9]
Solution: 𝑓 ′ (2) = lim𝑥→2
𝑥−2

x2 – 8x + 12
= lim𝑥→2
𝑥−2

2
 3

= lim𝑥→2 (𝑥 − 6) = −4 //

Example 2: Let 𝑓(𝑥) = |𝑥|. Compute 𝑓 ′ (2), 𝑓 ′ (−2)𝑎𝑛𝑑 𝑓 ′ (0).

|x| 2
Solution: 𝑓 ′ (2) = lim𝑥→2 = 2 = 1,
𝑥

|x| 2
𝑓 ′ (−2) = lim𝑥→−2 = − 2 = −1 and
𝑥

|x|
𝑓 ′ (0) = lim𝑥→0 . This limit does not exist; hence, 𝑓 is not differentiable at 0.
𝑥

Remark: If the fixed number 𝑐 in Definition 1 is replaced by a variable 𝑥, we get the formula

f ( x  h)  f ( x )
f ' ( x)  lim (3)
h 0 h

Similarly if 𝑥 and 𝑐 in equation (2) are replace by 𝑡 and 𝑥 respectively we get the formula

f (t )  f ( x)
f ' ( x)  lim (4)
tx tx

Definition 2(derivative as a function): Let 𝑓: 𝐷 ⊆ ℝ → ℝ and 𝐸 be the set of all numbers in 𝐷 at

which 𝑓 is differentiable. The function defined by

f (t )  f ( x )
f ' ( x ) : lim , x𝜖𝐸
tx tx

is called the derivative of f (because it is derived from f). In general, 𝐸 ⊆ 𝐷.

Remark (Alternative Notations for derivative): Let 𝒚 = 𝒇(𝒙) be differentiable at 𝒄. Then we

can denote

df dy 𝑑𝑓 𝑑𝑦
a) the derivative of 𝑓 at 𝑐 by 𝑓 ′ (𝑐) , (c ) , (c ) , D f(c), | , | .
𝑑𝑥 𝑥=𝑐 𝑑𝑥 𝑥=𝑐
dx dx
b) the derivative of 𝑓 at 𝑥 by
𝑑𝑦 𝑑
𝑓 ′ (𝑥) = 𝑑𝑥 = 𝑑𝑥 𝑓(𝑥) = D f(x) = [𝑓(𝑥)]′ = 𝑦 ′ = 𝑦̇ =…

dy
The separate meaning of 𝑑𝑥 and 𝑑𝑦 in the expression “ " will be described later in section 4.8.
dx

3
 4

Example 3: Let 𝑓(𝑥) = 𝑥 2 – 8𝑥 + 9 . Find 𝑓 ′.

[t2 – 8t + 9]−[x2 −8(x)+9] t2 –x2 −( 8t−8x)

Solution: 𝑓 ′ (𝑥) = lim𝑡→𝑥 = lim𝑡→𝑥
𝑡−𝑥 𝑡−𝑥

= lim𝑡→𝑥 (𝑡 + 𝑥 − 8) = 2𝑥 − 8

𝑑𝑦
Example 4: Let 𝑓(𝑥) = |𝑥|. Use the definition of derivative to find . Are the domains of 𝑓 and
𝑑𝑥

𝑓′ the same?

Solution:

𝑐𝑎𝑠𝑒 1: 𝑖𝑓 𝑥 > 0

|t|−|x| t−x
𝑓 ′ (𝑥) = lim𝑡→𝑥 = lim𝑡→𝑥+ lim𝑡→𝑥+ 𝑡−𝑥 = 1
𝑡−𝑥

𝑐𝑎𝑠𝑒 2: 𝑖𝑓 𝑥 < 0

|t|−|x| x−t
𝑓 ′ (𝑥) = lim𝑡→𝑥 = lim𝑡→𝑥+ lim𝑡→𝑥+ 𝑡−𝑥 = −1
𝑡−𝑥

𝑐𝑎𝑠𝑒 3: 𝑖𝑓 𝑥 = 0

|x|−|0| |x|
𝑓 ′ (0) = lim𝑥→0 = lim𝑥→0 . This limit does not exist. Therefore, f is not differentiable
𝑥−0 𝑥

𝑥 = 0.

𝑑𝑦 1 𝑖𝑓 𝑥 > 0
Therefore 𝑑𝑥 = {
−1 𝑖𝑓 𝑥 < 0

The domain of 𝑓 is ℝ and the domain of 𝑓′ is ℝ\{0} . //

𝑑𝑦
Example 5: Let 𝑓(𝑥) = √𝑥 − 1. Use the definition of derivative to find . Are the domains of 𝑓
𝑑𝑥

and 𝑓′ the same?

Solution:

√t−1−√x−1 √t−1−√x−1 √t−1+√x−1 1 1

𝑓 ′ (𝑥) = lim𝑡→𝑥 = lim𝑡→𝑥 = lim𝑡→𝑥 =2
𝑡−𝑥 𝑡−𝑥 √t−1+√x−1 √t−1+ √x−1 √x−1

The domain of 𝑓 is [0, ∞) and the domain of 𝑓′ is (0, ∞) . //

4
 5

Definition 3(One-sided Derivative): Let 𝑦 = 𝑓(𝑥) be a function and let a be in the domain of f.
a) The right-hand derivative of f at x = c is the limit

f ( x )  f (c )
f ' (c )  lim and
x c  xc

b) the left-hand derivative of 𝑓 at 𝑥 = c is the limit

f ( x )  f (c )
f ' (c )  lim
x c  xc

Remark: A function 𝑓 is differentiable at 𝑥 = 𝑐 if and only if 𝑓 has both a right-hand

derivative and a left-hand derivative at 𝑥 = 𝑐 and if these derivatives are equal.

Definition 4 (differentiability on intervals): Let 𝒚 = 𝒇(𝒙) be a function and let [𝒂, 𝒃] be an

interval contained in the domain of 𝒇. We say f is differentiable on [𝒂, 𝒃] if

i) 𝒇 is differentiable at 𝒙 = 𝒄 for every 𝒄 𝜺(𝒂, 𝒃),

ii) the right-hand derivative of 𝒇 exists at 𝒙 = 𝒂 and

iii) the left-hand derivative of 𝒇 exists at 𝒙 = 𝒃.

Geometric and Physical Interpretations of Derivative:

Let 𝑦 = 𝑓(𝑥) be differentiable at 𝑐. There are two basic interpretations of 𝑓 ′ (𝑐).
Geometrically, 𝑓 ′ (𝑐) is the slope of the tangent line to the graph of 𝑓 at (𝑐, 𝑓(𝑐)).
Physically, 𝑓 ′ (𝑐) is the instantaneous rate of change of 𝑦 with respect to 𝑥 at 𝑐. For example, Suppose that an
object moving in a straight line has its position y at time x given by y=f(x). Then the velocity of the
object at time x is 𝑓 ′ (𝑥).

Definition 5 (tangent line and normal line to a curve): Let 𝑦 = 𝑓(𝑥) be differentiable at 𝑐. The
line 𝑡 through (c,f(c)) with slope 𝑓 ′ (𝑐) is called the tangent line to the graph of f at c and the line n through
(c,f(c)) perpendicular to t is called the normal line to the graph of f at c.

5
 6

Example 6: Find an equation of the tangent line 𝑡 to the graph of 𝑓(𝑥) = 𝑥 2 − 8𝑥 + 9 at the
point (2, −3).

Solution: Let (𝑥, 𝑦) be a point on t .From example 4.1, 𝑓 ′ (2) =−4 . By definition 4.6, 𝑠𝑙𝑜𝑝𝑒(𝑡) =
𝑦−(−3) 𝑦−(−3)
−4. On the other hand, from geometry, 𝑠𝑙𝑜𝑝𝑒(𝑡) = . Therefore = −4 or 𝑦 =
𝑥−2 𝑥−2

−4𝑥 + 5. //

Example 7: Find an equation of the normal line 𝑛 to the graph of 𝑓(𝑥) = 𝑥 2 − 8𝑥 + 9 at 𝑥 = 2.

Solution: Let (x, y) be a point on n. From analytic geometry we now that 𝑠𝑙𝑜𝑝𝑒(𝑡). 𝑠𝑙𝑜𝑝𝑒(𝑛) = −1.
1
This implies that (−4 )𝑠𝑙𝑜𝑝𝑒(𝑛) = −1 . Thus, 𝑠𝑙𝑜𝑝𝑒(𝑛) = 4. On the other hand, 𝑠𝑙𝑜𝑝𝑒(𝑛) =
𝑦−(−3) 𝑦−(−3) 1 5
. Therefore = ¼ or 𝑦 = 4 𝑥 + 2. //
𝑥−2 𝑥−2

Example 8: A particle moves along the X-axis so that at any time t ≥ 0 its position is given by
𝑥(𝑡) = 𝑡 3 − 3𝑡 2 − 9𝑡 + 1. For what values of t is the particle at rest?

Solution: When the particle is at rest its velocity is zero. Hence, we solve 𝑥′(𝑡) = 0 for t. 𝑥 ′(𝑡) = 0 ⇔
3𝑡 2 − 6𝑡 − 9 = 0 ⇔ 𝑡 = −1 𝑜𝑟 𝑡 = 3. But t cannot be negative; hence t=3. //

Differentiability & Continuity

Theorem 1[differentiability implies continuity]: If 𝑓 is differentiable at 𝑎, then 𝑓 is continuous at

𝑎.

Proof: Assume that 𝑓 ′ (𝑎) exists. We want to show that lim𝑥→𝑎 𝑓(𝑥) = 𝑓(𝑎).

𝑓(𝑥)−𝑓(𝑎)
Note that, f(x) = 𝑥−𝑎
. (𝑥 − 𝑎) + 𝑓(𝑎)

Taking the limits of both sides of the above equation as 𝑥 approaches a and noting the fact that
𝑓 ′ (𝑎) exists, we get that lim𝑥→𝑎 𝑓(𝑥) = 𝑓(𝑎). ∎

Remark :

a) The converse of Theorem 1 is not true. Counter example: the function 𝑓(𝑥) = |𝑥| is not
differentiable at x=0 but it is continuous at x =0.
b) Three conditions must be fulfilled for a function to be differentiable at a given point:
i) The function must be continuous there.

6
 7

ii) The graph must be smooth there, with no sharp bend.

iii) The tangent line must be oblique (slanted) or horizontal, not vertical tangent.

From the contra-positive of Theorem 4.1, if f is not continuous at x=c, then f is not differentiable at
x=c. If f has sharp point at c, then the left and right derivatives of f at c cannot be equal. If |𝑓′(𝑎)|
=∞, then the line through (c, f(c)) with slope 𝑓 ′ (𝑎) is a vertical line.

1
Example 9: The function 𝑓(𝑥) = is not differentiable at 𝑥 = 0 because 𝑓 is not continuous at 0;
𝑥

the function 𝑓(𝑥) = |𝑥| is not differentiable at 𝑥 = 0 because the graph of f has a sharp bend at (0,
3
0) and the function f(x) = √𝑥 is not differentiable at x=0 because 𝑓′(0) =∞.

4.1.2 Basic Differentiation Rules and Derivatives of Common Functions

Several rules have been developed for finding derivatives without having to use the definition
directly. In this section we shall learn some derivative rules and the derivatives of commonly
appearing functions. Also in sections 4.3 and 4.4 we continue our study of how to differentiate more
and more functions.
A. Basic Rules
Theorem 2[Constant Rule]: Let f(x) = c, where c is a constant. Then 𝑓 ′ (𝑥) = 0.

d
In Leibniz Notation: c= 0
dx

𝑓(𝑡+ℎ)−𝑓(𝑥) 𝑐−𝑐
Proof: 𝑓 ′ (𝑥) = limℎ→0 ℎ
= limℎ→0 ℎ
= 0. ∎

𝑑
Example 10: (7) = 0.
𝑑𝑥

Theorem 3[Power Rule] :Let f(x) = x r, where r is a constant for which x r is defined. Then 𝑓 ′ (𝑥)=
d r
𝑟𝑥 𝑟−1 . In Leibniz Notation: x = rxr-1
dx

Proof:

Case 1: If 𝑟 = 𝑛 is a nonnegative integer

Let n be any non-negative integer. From the binomial theorem, we have

7
 8

(𝑥 + ℎ)𝑛 = 𝑥 𝑛 + 𝑛𝑥 𝑛−1 ℎ + 𝑄ℎ4

where Q is a polynomial in x and h.

Thus

Hence for any non-negative integer n

Case 2: If r= n is any integer (exercise)

Case 3: If r is any rational number (exercise)

Case 4: If r is any real number (omitted). ∎

Example 11:

a) The derivative of x 3 is 3x2.

b) The derivative of x10 is 10x9.

1
1 2 1
c) Since √𝒙 is the same as x1/2, the derivative of √𝒙 is x =
2 2 x

𝒅 𝟏 ′ 𝟐
d) ( ) =(𝒙−𝟐 ) = −𝟐𝒙−𝟐−𝟏 = −
𝒅𝒙 𝒙𝟐 𝒙𝟑

8
 9

Theorem 4 [Sum, Difference, Product and Quotient Rules]: Let 𝑓 ′ (𝑥) and 𝑔′ (𝑥) exist. Then

𝑎) (𝑓 + 𝑔)′ (𝑥) = 𝑓 ′ (𝑥) + 𝑔′ (𝑥) [𝑆𝑢𝑚 𝑅𝑢𝑙𝑒]

𝑏) (𝑓 − 𝑔)′ (𝑥) = 𝑓 ′ (𝑥) − 𝑔′ (𝑥) [𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑅𝑢𝑙𝑒]

𝑐) (𝑓𝑔)′ (𝑥) = 𝑓 ′ (𝑥)𝑔(𝑥) + 𝑓(𝑥)𝑔′ (𝑥) [𝑃𝑟𝑜𝑑𝑢𝑐𝑡 𝑅𝑢𝑙𝑒]

𝑓 ′ 𝑓′ (𝑥)𝑔(𝑥)−𝑓(𝑥)𝑔′ (𝑥)
d) [𝑔] (𝑥) = [𝑄𝑢𝑜𝑡𝑖𝑒𝑛𝑡 𝑅𝑢𝑙𝑒]
𝑔2 (𝑥)

Proof: We prove (a) and (c); the rest are left as exercises.

(𝑓+𝑔)(𝑡)−(𝑓+𝑔)(𝑥)
a) (𝑓 + 𝑔)′ (𝑥) = lim𝑡→𝑥 𝑡−𝑥

𝑓(𝑡) − 𝑓(𝑥) 𝑔(𝑡) − 𝑔(𝑥)

= lim + lim = 𝑓 ′ (𝑥) + 𝑔′ (𝑥)
𝑡→𝑥 𝑡−𝑥 𝑡→𝑥 𝑡−𝑥

(𝑓𝑔)(𝑡)−(𝑓𝑔)(𝑥)
b) (𝑓𝑔)′ (𝑥) = lim𝑡→𝑥
𝑡−𝑥

𝑓(𝑡)𝑔(𝑡)−𝑓(𝑥)𝑔(𝑡)+𝑓(𝑥)𝑔(𝑡)−𝑓(𝑥)𝑔(𝑥)
= lim𝑡→𝑥 𝑡−𝑥

𝑔(𝑡)[𝑓(𝑡)−𝑓(𝑥)]+𝑓(𝑥)[𝑔(𝑡)−𝑔(𝑥)]
= lim𝑡→𝑥
𝑡−𝑥

[𝑓(𝑡)−𝑓(𝑥)] [𝑔(𝑡)−𝑔(𝑥)]
= lim𝑡→𝑥 𝑔(𝑡) lim𝑡→𝑥 + lim𝑡→𝑥 𝑓(𝑡) lim𝑡→𝑥
𝑡−𝑥 𝑡−𝑥

= 𝑓 ′ (𝑥)𝑔(𝑥) + 𝑓(𝑥)𝑔′ (𝑥) ∎

Corollary [Constant Multiple Rule]: If c is a constant and f is a differentiable function, then

(𝑐𝑓)′ (𝑥) = 𝑐𝑓 ′ (𝑥).

Example 12: Differentiate the following functions.

a) 𝑥 8 + 12𝑥 5 − 4𝑥 4 + 10𝑥 3 − 6𝑥 + 5
b) (6𝑥 3 )(7𝑥 4 )
𝑥 2 +𝑥−2
c)
𝑥 3 +6

Solution:

9
 10

𝑑
a) [𝑥 8 + 12𝑥 5 − 4𝑥 4 + 10𝑥 3 − 6𝑥 + 5]
𝑑𝑥
𝑑 𝑑 𝑑 𝑑 𝑑 𝑑
= [𝑥 8 ] + 𝑑𝑥 [12𝑥 5 ] − 𝑑𝑥 [4𝑥 4 ] + 𝑑𝑥 [10𝑥 3 ] − 𝑑𝑥 [6𝑥 1 ] + 𝑑𝑥 [5𝑥 0 ]
𝑑𝑥
[sum & difference rule]
= 8x 7 + 12(5)x 4 − 4(4)x 3 + 10(3)x 2 − 6(1)x 0 + 5(0x −1 )
[power rule]
= 8x 7 + 60x 4 − 16x 3 + 30x 2 − 6
𝑑 𝑑 𝑑
b) [(6𝑥 3 )(7𝑥 4 )] = (7𝑥 4 ) 𝑑𝑥 (6𝑥 3 ) + (6𝑥 3 ) 𝑑𝑥 (7𝑥 4 )
𝑑𝑥

= (7𝑥 4 )(18𝑥 2 ) + (6𝑥 3 )(28𝑥 3 )

= (126𝑥 6 ) + (168𝑥 6 )
= 294𝑥 6
𝑑 𝑑
𝑑 𝑥 2 +𝑥−2 (𝑥 3 +6) [𝑥 2 +𝑥−2]−(𝑥 2 +𝑥−2) [𝑥 3 +6] (𝑥 3 +6)(2𝑥+1)−(𝑥 2 +𝑥−2)(3𝑥 2 )
c) [ ] = 𝑑𝑥 𝑑𝑥
=
𝑑𝑥 𝑥 3 +6 (𝑥 3 +6)2 (𝑥 3 +6)2

−𝑥 4 −2𝑥 3 −6𝑥 2 +12𝑥+6

=
(𝑥 3 +6)2

Example 13: Let 𝑦, 𝑢 and 𝑣 be functions such that 𝑦 = 𝑢𝑣, 𝑢(2) = 3, 𝑢′ (2) = −4, 𝑣(2) =
1 𝑎𝑛𝑑 𝑣 ′ (2) = 2. Find 𝑦 ′ (2).

Solution: 𝑦 ′ (2)=𝑢′ (2)𝑣(2) + 𝑢(2)𝑣 ′ (2) = −4(1) + 3(2) = −4 + 6 = −2 //

Theorem 5 [The Chain Rule]: Let 𝑓 ′ (𝑥) and 𝑔′(𝑓(𝑥)) exist.

𝑑𝑢 𝑑𝑢 𝑑𝑦
Then (𝑔𝑜𝑓)′ (𝑥)= 𝑔′(𝑓(𝑥)) 𝑓 ′ (𝑥) . In Leibnz notation: = 𝑑𝑦 𝑑𝑥 .
𝑑𝑥

To show the equivalence of the results in the prime notation and the Leibnz notation, put 𝑢 =
𝑔(𝑓(𝑥)) and 𝑦 = 𝑓(𝑥) .Then 𝑢 = 𝑔(𝑦). Thus 𝑢 depends on 𝑦, y depends on 𝑥 and 𝑢 depends on
𝑑𝑢 𝑑𝑦 𝑑𝑢 𝑑𝑦 𝑑𝑢
𝑥 and hence we have , and 𝑑𝑥 . Observe that = 𝑓 ′ (𝑥) & 𝑑𝑦 = 𝑔′ (𝑦)= 𝑔′(𝑓(𝑥)). On the
𝑑𝑦 𝑑𝑥 𝑑𝑥
𝑑𝑢 𝑑𝑢
other hand, u =(gof)(x) ; hence = (𝑔𝑜𝑓)′ (𝑥). Thus (𝑔𝑜𝑓)′ (𝑥)= 𝑔′(𝑓(𝑥)) 𝑓 ′ (𝑥) ⟺ 𝑑𝑥 =
𝑑𝑥
𝑑𝑢 𝑑𝑦
𝑑𝑦 𝑑𝑥
.

10
 11

Example 14: Differentiate (3𝑥 2 + 1)10

Solution: Put u=(3𝑥 2 + 1)10 and y=3𝑥 2 + 1. Then 𝑢 = 𝑦10.

𝑑𝑢
Required:
𝑑𝑥

𝑑𝑢 𝑑𝑢 𝑑𝑦 𝑑(𝑦 10 ) 𝑑(3𝑥 2 +1)

= 𝑑𝑦 = 𝑑𝑦 = (10𝑦 9 )(6𝑥) = 10(3𝑥 2 + 1)9 (6𝑥) = 60𝑥(3𝑥 2 + 1)9 //
𝑑𝑥 𝑑𝑥 𝑑𝑥

Example 15: If 𝑓 and 𝑔 are two functions with f (a)  1, f (a)  2, g (a)  1 and g ( a )  3 ,

 
then find  fg  f  (a ) and  f 
 (a)

 f g
B. Derivatives of Common Functions

The derivatives of polynomial and rational functions can be evaluated by making use of the
differentiation rules given previously. We provide the derivatives of trigonometric, exponential and
logarithmic functions in this subsection.
Derivatives of the basic trigonometric functions:

Theorem 6 [The Derivatives of the basic trigonometric functions]:

𝑑
a) (sin 𝑥) = cos 𝑥
𝑑𝑥

𝑑
b) (cos 𝑥) = − sin 𝑥
𝑑𝑥

𝑑
c) (tan 𝑥) = sec 2 𝑥
𝑑𝑥

𝑑
d) (csc 𝑥) = − 𝑐𝑠𝑐 𝑥 cot 𝑥
𝑑𝑥

𝑑
e) (sec 𝑥) = sec 𝑥 tan 𝑥
𝑑𝑥

𝑑
f) (cot 𝑥) = −csc 2 𝑥
𝑑𝑥

Proof:

sin ℎ
a) Let f(x) = sin x. Recall that Sin(x + h) = 𝑠𝑖𝑛 𝑥 𝑐𝑜𝑠 ℎ + 𝑐𝑜𝑠 𝑥 sin h, limℎ→0 =1 and
ℎ
cos ℎ − ℎ
limℎ→0 ℎ =0. The Newton's Quotient for the sin x is then

11
 12

This quotient tends to cos x as .

Hence ,

b) Let f(x) = cosx. Recall that cos(x+h) = cosx cos h − sinx sin h .The Newton's Quotient for the cos
x is then

This quotient tends to −sin x as .

Hence

The derivatives of the remaining trigonometric functions can be derived from the derivatives of sine
and cosine, by using Theorem 4. ∎

Derivatives of exponential functions:

Theorem 7 [The Derivatives of Exponential Functions]:

𝑑 𝑑
a) (𝑒 𝑥 ) = 𝑒𝑥 𝑏) 𝑑𝑥 (𝑏 𝑥 ) = (𝑙𝑛𝑏)𝑏 𝑥
𝑑𝑥

Proof:

12
 13

Let f(x) = ex. Recall the definition of Euler’s number e:

where e is an irrational number roughly equal to 2.718281828459045235.

This equality implies that when h is small in magnitude,

That is

Therefore,

𝒅
Hence 𝒅𝒙 (𝒆𝒙 ) = 𝒆𝒙.

𝒅 𝒅 𝒅
b) 𝒅𝒙 (𝒃𝒙 ) = (𝒆𝒍𝒏𝒃 𝒙 ) = 𝒆𝒍𝒏𝒃𝒙 𝒅𝒙 (𝒍𝒏𝒃 𝒙) = 𝒍𝒏𝒃 𝒆𝒙. ∎
𝒅𝒙

Theorem 8 [The Derivatives of Logarithmic Functions]

𝑑 1 𝑑 1
a) (ln 𝑥) =𝑥 𝑏) 𝑑𝑥 (𝑏 𝑥 ) = xln(𝑏)
𝑑𝑥

Proof:

Let f(x) = ln x. Note that ln x is the inverse of the exponential function ex. Thus ln ex = x for all
x. In particular, ln e = 1. Then

13
 14

𝒉
Let k= 𝒙. Thus h = kx and

As h→0, k→0. Hence by the definition of e, we have

Therefore

The proof of (b) is left for the reader.. ∎

Derivatives of Hyperbolic Functions:

Every function 𝑓 that is defined on an interval centered at the origin can be decomposed as
𝑓(𝑥)+ 𝑓(−𝑥) 𝑓(𝑥)− 𝑓(−𝑥)
f (x) = 2
+
2
.
𝑓(𝑥)+ 𝑓(−𝑥) 𝑓(𝑥)− 𝑓(−𝑥)
The function is even and the function is odd. Thus any function whose
2 2
domain is symmetric about the origin can be expressed as the sum of an even function and an odd
function. For 𝑓(𝑥) = 𝑒 𝑥 , the even part is called cosine hyperbolic and the odd part is called sine
hyperbolic.

14
 15

Definition 6 (hyperbolic functions): The function

ex +e−x
a) 𝑐osh x: = is called cosine hyperbolic
2
ex −e−x
b) 𝑠𝑖𝑛h x: = is called sine hyperbolic
2
sinhx
c) 𝑡𝑎𝑛ℎ x: = coshx is called tangent hyperbolic.
1
d) 𝑐𝑠𝑐ℎ x: = sinhx is called cosecant hyperbolic.
1
e) 𝑠𝑒𝑐ℎ x: = cschx is called secant hyperbolic.
coshx
f) 𝑐oth x: = is called cotangent hyperbolic
sinhx

The functions tanhx, cschx, sech x and coth x can also be defined directly in terms of the
exponential functions 𝑒 𝑥 & e−x.

Example 16: Evaluate sinh(𝑙𝑛2).

1
2− 3
Solution: sinh(𝑙𝑛2) = 𝑒 𝑙𝑛2 − 𝑒 −𝑙𝑛2 = 2
= 4. //
2

Note (Where the term "hyperbola" comes from in the term "hyperbolic functions"?): Just as x = cosu and
y = sin u are identified with the points (x, y) on the unit circle, the functions x = cosh uand
y = sinh u are identified with the points (x, y) on the right hand branch of the unit hyperbola
x 2 − y 2 = 1.The point ( cosh u , sinh u) lies on the right hand branch of the hyperbola.

Some Identities for Hyperbolic Functions

1) 𝒄𝒐𝒔𝒉𝟐𝒙 − 𝒔𝒊𝒏𝒉𝟐𝒙 = 𝟏 8) 𝒔𝒊𝒏𝒉 𝒙 + 𝒔𝒊𝒏𝒉 𝒚 =

2) 𝒔𝒆𝒄𝒉𝟐𝒙 + 𝒕𝒂𝒏𝒉𝟐𝒙 = 𝟏 𝟐 𝒔𝒊𝒏𝒉 ½(𝒙 + 𝒚) 𝒄𝒐𝒔𝒉 ½(𝒙 −
3) 𝒔𝒊𝒏𝒉(− 𝒙) = −𝒔𝒊𝒏𝒉(𝒙) 𝒚)
4) 𝒄𝒐𝒔𝒉(−𝒙) = 𝒄𝒐𝒔𝒉 𝒙 9) 𝒔𝒊𝒏𝒉 𝒙 − 𝒔𝒊𝒏𝒉 𝒚 =
5) 𝒔𝒊𝒏𝒉 (𝒙 ± 𝒚) = 𝒔𝒊𝒏𝒉 𝒙 𝒄𝒐𝒔𝒉 𝒚 ± 𝟐 𝒄𝒐𝒔𝒉 ½(𝒙 + 𝒚) 𝒔𝒊𝒏𝒉 ½(𝒙 −
𝒄𝒐𝒔𝒉 𝒙 𝒔𝒊𝒏𝒉 𝒚 𝒚)
6) 𝒄𝒐𝒔𝒉 (𝒙 ± 𝒚) =
𝒄𝒐𝒔𝒉 𝒙 𝒄𝒐𝒔𝒉 𝒚 ± 𝒔𝒊𝒏𝒉 𝒙 𝒔𝒊𝒏𝒉 𝒚
7) 𝒕𝒂𝒏𝒉(𝒙 ± 𝒚) = (𝒕𝒂𝒏𝒉 𝒙 ±
𝒕𝒂𝒏𝒉 𝒚)/(𝟏 ± 𝒕𝒂𝒏𝒉 𝒙. 𝒕𝒂𝒏𝒉 𝒚)

15
 16

Let us prove the first identity:

𝒆𝒙 +𝒆−𝒙 𝟐 𝒆𝒙 −𝒆−𝒙 𝟐 𝒆𝟐𝒙 +𝟐+𝒆−𝟐𝒙 𝒆𝟐𝒙 −𝟐−𝒆−𝟐𝒙

cosh2x - sinh2x=( ) −( ) =( ) − ( ) =𝟏
𝟐 𝟐 𝟒 𝟒

−3
Example 17: Given sinh 𝑥 = , find tanh 𝑥.
4

3 2
Solution: From the identity cosh2x - sinh2x = 1 we get cosh2 𝑥 − (− 4) = 1. ⟹ cosh 𝑥 = ±
3
25 4 5 sinh 𝑥 − 3
√ = ± 5. But cosh 𝑥 > 0 for every x, so cosh 𝑥 = 4. Thus tanh 𝑥 = cosh 𝑥 = 5
4
= − 5.
16
4

//

𝑠𝑖𝑛ℎ 𝑐𝑜𝑠ℎ 𝑡𝑎𝑛ℎ

𝑐𝑜𝑡ℎ
𝑐𝑠𝑐ℎ 𝑠𝑒𝑐ℎ

Figure 2: Graphs of Hyperbolic Functions

`

Theorem 9 (derivatives of hyperbolic functions):

𝑑
a) (𝑠𝑖𝑛ℎ𝑥) = cosh 𝑥.
𝑑𝑥

𝑑
𝑏) 𝑑𝑥 (cosh 𝑥) = 𝑠𝑖𝑛ℎ𝑥

𝑑
c) (𝑡𝑎𝑛ℎ𝑥) = sech2 𝑥
𝑑𝑥

𝑑
𝑑) 𝑑𝑥 (𝑐𝑠𝑐ℎ𝑥) = −𝑐𝑠𝑐ℎ𝑥 𝑐𝑜𝑡ℎ𝑥

16
 17

𝑑
𝑒) (sech 𝑥) = −𝑠𝑒𝑐ℎ𝑥 𝑡𝑎𝑛ℎ𝑥
𝑑𝑥

𝑑
f) 𝑑𝑥 (coth 𝑥) = − csch2 𝑥

Proof: Apply differentiation rules on the exponential functions that define the hyperbolic functions.
As an illustration we prove the first.
𝑑 𝑑 𝑒 𝑥 −𝑒 −𝑥 𝑒 𝑥 +𝑒 −𝑥
(𝑠𝑖𝑛ℎ𝑥) = 𝑑𝑥 ( ) =( ) = cosh 𝑥. ∎
𝑑𝑥 2 2

Example 18: Differentiate

a) sinh2 √1 − 𝑥 2
b) 𝑒 𝑐𝑠𝑐ℎ𝑥

Solution:
a) Let 𝑦 = √1 − 𝑥 2 𝑎𝑛𝑑 𝑢=sinh2 √1 − 𝑥 2 . Then 𝑢=sinh2 𝑦
𝑑𝑢 𝑑𝑢 𝑑𝑦 −2𝑥
Now𝑑𝑥 = 𝑑𝑦 𝑑𝑥 = (2sinh 𝑦)( )
2√1−𝑥 2
−2𝑥 −𝑥
= (2sinh √1 − 𝑥 2 )( ) = (2sinh √1 − 𝑥 2 )( )
2√1−𝑥 2 √1−𝑥 2

b) Let 𝑦 = csch 𝑥 𝑎𝑛𝑑 𝑢=𝑒 𝑐𝑠𝑐ℎ𝑥 . Then, 𝑢 = 𝑒 𝑦 .

𝑑𝑢 𝑑𝑢 𝑑𝑦
= 𝑑𝑦 = 𝑒 𝑦 ((−𝑐𝑠𝑐ℎ𝑥 𝑐𝑜𝑡ℎ𝑥)) = 𝑒 𝑐𝑠𝑐ℎ𝑥 (−𝑐𝑠𝑐ℎ𝑥 𝑐𝑜𝑡ℎ𝑥).
𝑑𝑥 𝑑𝑥

17
 18

Exercise 4.1:

1. Using only the limit definition of derivative to compute the derivative of

1
a) 𝑓(𝑥) = | 𝑥 2 − 3𝑥 | b) f(x) =2𝑥 2 +5
1
𝑦 3 −1
2. Compute lim𝑦→1 .
𝑦−1

x100 −1
3. Compute lim𝑥→1 .
𝑥−1
𝑐𝑜𝑠𝑥, 𝑥 ≥ 0
4. Find the values of 𝑎 and 𝑏 so that the function 𝑓(𝑥) = { ; is differentiable at 0.
𝑎𝑥 + 𝑏, 𝑥 < 0
5. Find the equation of the tangent line and the normal line to the graph of
1
f ( x)  1  ( x  1) 3 at ( 1, f ( 1)) .

6. Determine an equation of the tangent line to the curve 𝑦 = √𝑥 at the point (9, 3).
d
7. Let 𝑓 be a differentiable function such that f ( x)  e x , then find ( f (ln(ln 2 x)))
dx
8. Find the point on the curve y = x3 + x2 + x where the tangent line is parallel to y = 2x + 3.
9. Show that the derivative of even function is odd and vice versa.
10. Suppose that 𝑔 is a differentiable function and that 𝑓(𝑥) = 𝑔(𝑥 + 5) for all x. If 𝑔 ′(1) =
3 and 𝑓 ′(𝑎) = 3 , find 𝑎.

4.2 Derivatives of Inverse Functions: The Reciprocal Rule

In Section 4.1 we learned certain rules for differentiating a function. In sections 4.2 and 4.3 we
continue our study of how to differentiate more and more functions.
A. The Reciprocal Rule of Derivatives

Definition 1: A function 𝑔 is said to be an inverse of a function 𝑓 if

i) the range of 𝑓 is the domain of 𝑔 and
ii) 𝑓(𝑥) = 𝑦 ⇔ 𝑔(𝑦) = 𝑥 (that is, 𝑔 undoes that 𝑓 does).

Remark:
1) A function may not have an inverse (that is, there exist functions which do not have
inverses).
2) The inverse of a function 𝑓, if it exists, is unique and the unique inverse of 𝑓 is denoted by
𝑓 −1 . Can you justify why the inverse of a function is unique?

18
 19

3) Let A be a subset of the domain of f. If f has no inverse but if 𝑓|𝐴 , the restriction of f to A,
has an inverse; then we say f is invertible on A.

1
Note : 𝑓 −1 (𝑥) ≠ 𝑓(𝑥).

Graph of 𝒇−𝟏 : If the point (𝑥, 𝑦) lies on the graph of f, then the point (𝑦, 𝑥) lies on the graph of
𝑓 −1 . Thus the graph of 𝑓 −1 can be obtained by reflecting the graph of 𝑓 through the line 𝑦 = 𝑥.
Elementary Relations between f and 𝒇−𝟏 :
a) (𝑓 −1 )−1 = 𝑓
b 𝑓 −1 (𝑓(𝑥)) = 𝑥 for all 𝑥 in the domain of 𝑓.
c) 𝑓(𝑓 −1 (𝑥) = 𝑥 for all 𝑥 in the range of 𝑓.

Formula of 𝒇−𝟏 : If 𝑓 is given by a formula for 𝑓(𝑥) in terms of x, a formula for 𝑓 −1 may be found
by the following procedures.
Step 1: Write 𝑦 = 𝑓(𝑥)
Step 2: Solve for x in terms of y.
Step 3: Write 𝑓 −1 (𝑦) instead of x and interchange the roles of x and y.
2𝑥+3
Example 1: Let f(x) = . Find 𝑓 −1 (𝑥).
𝑥−1
2𝑥+3
Solution: Let y = 𝑥−1 ⇒ 𝑦(𝑥 − 1) = 2𝑥 + 3

⇒ 𝑥(𝑦 − 2) = 𝑦 + 3
𝑦+3
⇒𝑥=
𝑦−2
𝑦+3
⇒ 𝑓 −1 (𝑦) =
𝑦−2
𝑥+3
⇒ 𝑓 −1 (𝑥) =
𝑥−2
Criteria or Tests for Invertibility of a Function:
There are several necessary and/or sufficient conditions for invertibility of a function. For instance, if
a function is one to one then it is invertible. But the next theorem is easier to apply than the one-to-
oneness criterion.

19
 20

Definition 2: A function 𝑓: 𝐷 ⊆ ℝ ⟶ ℝ is said to be

a) strictly increasing if 𝑎 < 𝑏 ⟹ 𝑓(𝑎) < 𝑓(𝑏) for every 𝑎, 𝑏𝜖𝐷.
b) strictly decreasing if 𝑎 < 𝑏 ⟹ 𝑓(𝑎) > 𝑓(𝑏) for every 𝑎, 𝑏𝜖𝐷.
c) increasing if 𝑎 < 𝑏 ⟹ 𝑓(𝑎) ≤ 𝑓(𝑏) for every 𝑎, 𝑏𝜖𝐷.
d) decreasing if 𝑎 < 𝑏 ⟹ 𝑓(𝑎) ≥ 𝑓(𝑏) for every 𝑎, 𝑏𝜖𝐷.
e) strictly monotonic if it is either strictly increasing or strictly decreasing.
f) monotonic if it is other increasing or decreasing.
The condition 𝑎 < 𝑏 ⟹ 𝑓(𝑎) < 𝑓(𝑏) implies that if we go from left to right on the graph of a
strictly increasing function, we are rising. Similarly, if we go from left to right on the graph of a
strictly decreasing function, we are falling. Thus a function is strictly increasing if as x moves to right,
its graph moves up and is strictly decreasing if its graph moves down. A function is increasing if as x
moves to the right, its graph does not move down and decreasing if it does not move up. A function
is said to be strictly monotonic if it is either strictly increasing or strictly decreasing and monotonic if
it is either increasing or decreasing.

Strictly Increasing Function Strictly decreasing Function Constant function

Figure 3: Monotonic Functions

Theorem 1( A Sufficient Condition for Invertibility): If a function 𝑓 is strictly increasing or

strictly decreasing, then it is invertible.

In view of Theorem 4.8, strict monotonicity of 𝑓 implies invertibility of f. But how can we decide if
𝑓 is strictly monotonic or not? The following theorem answers this question. A function 𝑓is strictly
increasing if and only if the slope of every tangent line to its graph is positive and 𝑓 is strictly
decreasing if the slope of any tangent line to its graph is negative. Why? Hence we have the following
theorem.

20
 21

Theorem 2( A Characterization for Strict Monotonicity): Let 𝑓 be defined on a set 𝐷.

a) 𝑓 is strictly increasing on 𝐷 𝑖𝑓𝑓 𝑓(𝑥) > 0 for interior point 𝑥 of 𝐷.
b) 𝑓 is strictly decreasing on 𝐷 𝑖𝑓𝑓 𝑓(𝑥) < 0 for every interior point 𝑥 of 𝐷.

Proof: The proof needs knowledge of the so called mean value theorem; hence, it is postponed to
section 4.5. ∎
The following theorem gives the relationship between the derivative of 𝑓 −1 and the derivative of 𝑓.

Example 2: Show that the following functions have inverses.

b) 𝑓(𝑥) = 𝑥 5
𝑥−1
c) ℎ(𝑥) = ln(𝑥+1)
Solution:
a) Note that the domain of 𝑓 is ℝ & 𝑓′(𝑥) = 5𝑥 4 . Since 𝑓 ′ (𝑥) > 0 for every 𝑥𝜖(0, ∞), so f
is strictly increasing on [0, ∞). Since 𝑓 ′ (𝑥) > 0 for every 𝑥(−∞, 0), so f is strictly increasing
on (−∞, 0]. So f is strictly increasing on [0, ∞) ∪ (−∞, 0]=ℝ.
⟹ 𝑓 has an inverse.
𝑥+1 2
b) Note that the domain of ℎ is (−∞, −1) ∪ (1, ∞)& ℎ′ (𝑥) = 𝑥−1 . (𝑥+1)2 . Since ℎ′ (𝑥) > 0

for every 𝑥𝜖 (−∞, −1) ∪ (1, ∞), so ℎ(𝑥) > 0 for every x in the domain of f. ⟹ f is strictly
increasing on its domain. Hence, 𝑓 has an inverse.

Theorem 3 (The Reciprocal Rule): Let 𝑓 be continuous and invertible on an open interval
containing a real number a. If 𝑓 ′ (𝑎) exists, 𝑓 ′ (𝑎) ≠ 0 and 𝑓(𝑎) = 𝑐, then (𝑓 −1 )′ (𝑐) exists and
1
(𝑓 −1 )′ (𝑐) = (5)
𝑓′ (𝑎)

Note that the last conclusion of the theorem can be re-written as

1
(𝑓 −1 )′ (𝑐) =
𝑓′ (𝑓−1 (𝑐))

If the constant 𝑎 is replaced by a variable 𝑥 and if 𝑦 = 𝑓(𝑥), the equation (5) becomes
1
(𝑓 −1 )′ (𝑦) = (6)
𝑓′ (𝑥)

21
 22

𝑑𝑥
Since 𝑥 = 𝑓 −1 (𝑦), so (𝑓 −1 )′ (𝑦) = . Thus in Leibnz notation equation (6) becomes:
𝑑𝑦
𝑑𝑥 1
= 𝑑𝑦 (7)
𝑑𝑦
𝑑𝑥

Example 3: Let 𝑓(𝑥) = 𝑥 7 + 8𝑥 3 + 4𝑥 − 2 . Show that f has an inverse and find (𝑓 −1 )′(−2).
Solution: The domain of f is ℝ and 𝑓′(𝑥) = 7𝑥 6 + 24𝑥 2 + 4. ⟹ 𝑓 ′ (𝑥) > 0 for every x in its
domain. ⟹ f is strictly increasing. ⟹ f has an inverse.
1
(𝑓 −1 )′(−2) =
𝑓 ′ (𝑓 −1 (−2))

Let 𝑓 −1 (−2) = 𝑎. Then 𝑓(𝑎) = −2. ⟹ 𝑎7 + 8𝑎3 + 4𝑎 − 2 = −2. ⟹ 𝑎 = 0.

1 1 1
Thus(𝑓 −1 )′(−2) = = = . //
𝑓 ′ (𝑓 −1 (−2)) 𝑓 ′ (0) 4

Example 4: Let 𝑓(𝑥) = 𝑥 3 + 4𝑥 − 1 . Find the equation of the tangent line 𝑡 to the graph of 𝑓 −1
at 𝑃(15, 2).
Solution: The domain of 𝑓 is ℝ and 𝑓′(𝑥) = 3𝑥 2 + 4. ⟹ 𝑓 ′ (𝑥) > 0 for every 𝑥 in its domain. ⟹ 𝑓
is strictly increasing. ⟹ 𝑓 has an inverse.
1 1 1
𝑠𝑙𝑜𝑝𝑒(𝑡)= (𝑓 −1 )′(15) = 𝑓′ (𝑓−1 (15)) = ′ = .
𝑓 (2) 16
𝑦−2 1
An equation of t is 𝑥−15 = 16. //

B. Derivatives of Inverse Trigonometric Functions

None of the six trigonometric functions is one to one; hence, they do not have inverses. In each case,
however, the domain can be restricted to produce a new function that does have an inverse. First we
review the definition and basic properties of each of the six trigonometric function.
Let 𝜃 be an angle whose vertex is at the origin and whose initial ray lies on positive X-axis. Let the
terminal ray of 𝜃 intersects with the circle of radius r and center (0, 0) at the point (x, y). Then the
first column of the next table gives what it means by sin 𝜃, cos 𝜃, tan 𝜃, csc 𝜃, sec 𝜃 and cot 𝜃. The
other columns give other key facts about these functions.

Circle Definition Domain Range

𝑦
Sine Sin 𝜃 = −∞ < 𝑥 < ∞ [−1,1]
𝑟
𝑥
Cosine Cos 𝜃 = 𝑟 −∞ < 𝑥 < ∞ [−1,1]
𝑦 𝜋 3𝜋
Tangent Tan 𝜃 = 𝑥 𝑥 ≠ ±2,± , ⋯ (−∞, ∞)
2

22
 23

𝑟
Cosecant csc 𝜃 = 𝑥 ≠ 0, ± 𝜋, ± 2𝜋, ⋯ (−∞, −1] ∪ [1, ∞)
𝑦
𝑟 𝜋 3𝜋
Secant Sec 𝜃 = 𝑥 𝑥 ≠ ±2,± , ⋯ (−∞, −1] ∪ [1, ∞)
2
𝑥
Cotangent cot 𝜃 = 𝑦 𝑥 ≠ 0, ± 𝜋, ± 2𝜋, ⋯ (−∞, ∞)

None of the six basic trigonometry functions is a one-to-one function. However, in the following list,
each trigonometry function is listed with an appropriately restricted domain, which makes it one-to-
one.
−𝜋 𝜋 −𝜋 𝜋
1. y=sin 𝑥, ≤𝑥≤ 4. y=csc 𝑥, 𝑥 ∈ ( 2 , 2 ) \{0}
2 2

2. y=cos 𝑥, 0 ≤ 𝑥 ≤ 𝜋 5. 𝑦 = sec 𝑥 , 𝑥 ∈ (0 , 𝜋)\{2 }

𝜋

−𝜋 𝜋
3. y=tan 𝑥, < 𝑥< 6. 𝑦 = cot 𝑥 , 𝑥 ∈ (0 , 𝜋)
2 2

Because each of the above-listed functions is one-to-one, each has an inverse function. The
corresponding inverse functions are defined below.
Definition 3: The inverse of the function
−𝜋 𝜋
a) 𝑦 = 𝑠𝑖𝑛 𝑥, ≤𝑥≤ is called arc sine function. It is denoted by 𝑎𝑟𝑐𝑠𝑖𝑛 or sin-1.
2 2

b) 𝑦 = 𝑐𝑜𝑠 𝑥, 0 ≤ 𝑥 ≤ 𝜋 is called arc cosine function. It is denoted by 𝑎𝑟𝑐𝑐𝑜𝑠 or cos-1.

−𝜋 𝜋
c) 𝑦 = 𝑡𝑎𝑛 𝑥, < 𝑥< arc tangent function. It is denoted by 𝑎𝑟𝑐𝑡𝑎𝑛.
2 2
−𝜋 𝜋
d) 𝑦 = 𝑐𝑠𝑐 𝑥, 𝑥 ∈ ( 2 , 2 ) \{0} arc cosecant function. It is denoted by 𝑎𝑟𝑐𝑜𝑠𝑒𝑐.
𝜋
e) 𝑦 = 𝑠𝑒𝑐 𝑥, 𝑥 ∈ (0 , 𝜋)\{ 2 } arc secant function. It is denoted by 𝑎𝑟𝑐𝑠𝑒𝑐.

f) 𝑦 = cot 𝑥 , 𝑥 ∈ (0 , 𝜋) arc cotangent function. It is denoted by 𝑎𝑟𝑐𝑐𝑜𝑡.

Note: Some books (eg. ) use other intervals in defining arc-cosecant and arc-secant functions.
The graphs of inverse trigonometric functions can be obtained by reflecting the corresponding
graphs of trigonometric functions through the line y=x.

Identities Involving Inverse Trigonometric Functions

−𝜋 𝜋
1. arcsin (sin x) = x for ≤𝑥≤ and sin (arcsin x) = x for −1 ≤ 𝑥 ≤ 1
2 2

2. arccos (cos x) = x for 0 ≤ 𝑥 ≤ 𝜋 and cos (arccos x) = x for |𝑥| ≥ 1.

−𝜋 𝜋
3. arctan (tan x) = x for 2
≤𝑥≤ and sin (arcsin x) = x for x∈ ℝ.
2
−𝜋 𝜋
4. arcsc (csc x) = x for 𝑥 ∈ ( 2 , 2 ) \{0} and csc (arccsc x) = x for −1 ≤ 𝑥 ≤ 1
𝜋
5. arcsec (sec x) = x for 𝑥 ∈ (0 , 𝜋)\{ 2 } and sec (arcsec x) = x for |𝑥| ≥ 1.

23
 24

6. arccot (cot x) = x for 0 < 𝑥 < 𝜋 and cot (arccot x) = x for x∈ ℝ.

𝜋
7. arcsin x = 2 − 𝑎𝑟𝑐𝑐𝑜𝑠 𝑥
𝜋
8. arccot x = 2 − arctan 𝑥
𝜋
9. arccsc x = 2 − 𝑎𝑟𝑐𝑐𝑜𝑠 𝑥

Example 5(Evaluating inverse trigonometric functions): Evaluate the following.

𝜋
(a) arcsin(0) (d) arcsin(𝑠𝑖𝑛 6 )
(b) arcsin(−1) 5𝜋
(e) arcsin(𝑠𝑖𝑛 )
1 6
(c) arcsin ( ) 1
2
(f) Sin(2arctan 3 )
Solution:
−𝜋 𝜋
(a) Let 𝑎𝑟𝑐𝑠𝑖𝑛(0) = 𝑥. Then sin x = 0 and 2
≤ 𝑥 ≤ 2 . This implies that 𝑥 = 0.
−𝜋 𝜋 −𝜋
(b) Let 𝑎𝑟𝑐𝑠𝑖𝑛(−1) = 𝑥. Then sin x = −1 and 2
≤ 𝑥 ≤ 2 . This implies that 𝑥 = .
2
1 1 −𝜋 𝜋 𝜋
(c) Let arcsin (2) = 𝑥. Then sin x = 2 and 2
≤ 𝑥 ≤ 2 . This implies that 𝑥 = 6
𝜋 𝜋 −𝜋 𝜋 𝜋
(d) arcsin(𝑠𝑖𝑛 6 ) = 6 . This follows directly from the identities listed above, since 2
≤
6
≤ 2.
5π 5𝜋 −𝜋 𝜋 𝜋
(e) Let arcsin(sin 6
) = x. Then 𝑠𝑖𝑛 𝑥 = 𝑠𝑖𝑛 and 2 ≤ 𝑥 ≤ 2 . This implies that 𝑥 = 6 .
6
1 𝑦 1 −𝜋 𝑦 𝜋 𝑦 1 𝑦
(f) Let 2 arctan 3 = 𝑦 . Then tan (2 ) = 3 and 2
≤ 2
≤ 2 . tan (2 ) = 3 ⟺ sin (2 ) =
1 𝑦 3 1 𝑦 𝑦
𝑎𝑛𝑑 cos (2 ) = . Now 𝑆𝑖𝑛 (2𝑎𝑟𝑐𝑡𝑎𝑛 ) = sin(𝑦) = sin (2 + 2 ) =
√10 √10 3
𝑦 y 1 3 3
2 sin ( ) cos (2) = 2 ( ) ( 10) =5
2 √10 √
−𝜋 𝜋
Note: Be aware that " sin−1 𝑥 = 𝑦 ⟺ sin 𝑦 = 𝑥 " is valid only when ≤ 𝑥 ≤ 2 . For example,
2

sin 2𝜋 = 0 but sin−1 0 ≠ 2𝜋. Likewise, even if (−2)2 = 4, √4 ≠ −2.

The following table summarizes key results about the inverse trigonometric functions.

24
 25

Theorem 4 (derivatives of inverse trigonometric functions):

𝑑 1
a) (arcsin 𝑥) = for -1 < x < 1.
𝑑𝑥 √1−𝑥 2
𝑑 1
𝑏) 𝑑𝑥 (arccos 𝑥) = − for -1 < x < 1.
√1−𝑥 2

𝑑 1
c) (𝑎𝑟𝑐 tan 𝑥) = 1+𝑥 2 for all x.
𝑑𝑥

𝑑 1
𝑑) 𝑑𝑥 (𝑎𝑟𝑐 csc 𝑥) = − for |x| >1.
|𝑥|√𝑥 2 −1

𝑑 1
𝑒) (𝑎𝑟𝑐 sec 𝑥) = for |x| >1.
𝑑𝑥 |𝑥|√𝑥 2 −1

𝑑 −1
f) 𝑑𝑥 (arccot 𝑥) = 1+𝑥2 for all x.

Proof: We prove (a) and leave the rest for the reader as exercises.
−𝜋 𝜋 𝑑𝑦 1
Let y = arcsin 𝑥.This bi-implies 𝑥 = sin 𝑦 and ≤ 𝑦 ≤ . From the reciprocal rule, = 𝑑𝑥 , we
2 2 𝑑𝑥
𝑑𝑦

𝑑 1 1 −𝜋 𝜋
get that 𝑑𝑥 (arcsin 𝑥)= 𝑑 = 𝑐𝑜𝑠𝑦 (𝑤ℎ𝑒𝑟𝑒 < 𝑦 < 2 ).
(𝑠𝑖𝑛𝑦) 2
𝑑𝑦

−𝜋
Since 𝑥 = sin 𝑦, so 𝑥 2 = sin2 𝑦. ⟹ 𝑐𝑜𝑠 2 𝑦 = 1 − 𝑥 2 . ⟹ cos 𝑦 = ±√1 − 𝑥 2 . But for ≤𝑦≤
2
𝜋
2
, cos 𝑦 = √1 − 𝑥 2.

25
 26

𝑑 1
Therefore, (arcsin 𝑥) = for -1 < x < 1.
𝑑𝑥 √1−𝑥 2

C. Derivatives of Inverse Hyperbolic Functions

𝑑
Since 𝑑𝑥 (𝑠𝑖𝑛ℎ𝑥) = cosh 𝑥>0, so the hyperbolic sine function is strictly increasing. Hence, it has an
inverse. Similarly hyperbolic tangent, hyperbolic cotangent and hyperbolic cosecant are 1-1 on their
domains and therefore they have inverses. But the hyperbolic cosine function and the hyperbolic
secant function are not one to one. Hence they have no inverses on their domains. But their
restrictions to [0, ∞) are 1-1.

Definition 6:

a) If 𝑥 = sinh 𝑦 (where − ∞ < 𝑦 < ∞ ), then the function y = sinh−1 x is called

the inverse hyperbolic sine of x.

b) If 𝑥 = cosh 𝑦 (where 0 ≤ 𝑦 < ∞ ), then the function 𝑦 = cosh−1 x is called

the inverse hyperbolic cosine of x.

c) If 𝑥 = tanh 𝑦 (where − ∞ < 𝑦 < ∞ ), then the function y = tanh−1 x is called

the inverse hyperbolic tangent of x.

d) If 𝑥 = csch 𝑦 (where 𝑦 ≠ 0 ), then the function 𝑦 = csch−1 x is called the inverse

hyperbolic cosecant of x.

e) If 𝑥 = sech 𝑦 (where 0 ≤ 𝑦 < ∞ ), then the function y = sech−1 x is called

the inverse hyperbolic secant of x.

f) If 𝑥 = coth 𝑦 (where 𝑦 ≠ 0 ), then the function y = coth−1 x is called the inverse

hyperbolic cotangent of x.

Inverse hyperbolic functions can be expressed in terms of logarithmic functions as shown below.
This is not surprising as hyperbolic functions are expressed in terms of natural exponential functions.

a) sinh−1 𝑥 = 𝑙𝑛(𝑥 + √𝑥 2 + 1)for all x.

b) cosh−1 𝑥 = 𝑙𝑛(𝑥 + √𝑥 2 − 1)for all x≥ 1.

Similarly we can define the other inverse hyperbolic functions in terms of natural logarithms. The
table below summarizes the domains and ranges of inverse hyperbolic functions.

26
 27

Inverse Hyperbolic Range Domain

Function
𝑦 = 𝑆𝑖𝑛ℎ−1 𝑥 (-∞, ∞) (-∞, ∞)
y = cosh−1 𝑥 (0, ∞) (1, ∞)
𝑦 = 𝑡𝑎𝑛ℎ−1 𝑥 (-∞, ∞) (-1, 1)
y = csch−1 𝑥 𝑦≠0 x≠ 0
𝑦 = 𝑠𝑒𝑐ℎ−1 𝑥 (0, ∞) (0, 1]
y = coth−1 𝑥 𝑦≠0 |x| > 0

𝑐𝑜𝑠ℎ−1
−1
𝑠𝑖𝑛ℎ 𝑡𝑎𝑛ℎ−1

𝑐𝑜𝑡ℎ−1
𝑠𝑒𝑐ℎ−1

Figure 4: Graphs of Inverse Hyperbolic Functions

Theorem 6 (derivatives of inverse hyperbolic functions):

𝑑 1
a) (sinh−1 𝑥) = for all x.
𝑑𝑥 √1+𝑥 2

𝑑 1
𝑏) (cosh−1 𝑥) = for 𝑥 > 1..
𝑑𝑥 √𝑥 2 −1

𝑑 1
c) (tanh−1 𝑥) = 1−𝑥2 for |𝑥| < 1.
𝑑𝑥

𝑑 1
d) (csch−1 𝑥) =− 𝑓𝑜𝑟 𝑥 ≠ 1.
𝑑𝑥 |𝑥|√𝑥 2 +1

27
 28

𝑑 −1
e) (sech−1 𝑥) = for 0 < 𝑥 < 1.
𝑑𝑥 𝑥√1−𝑥 2

𝑑 1
f) (coth−1 𝑥) = 1−𝑥2 for |x|>1.
𝑑𝑥

Exercises 4.2:
7
1. Find the exact value of 𝑠𝑒𝑐(cot −1 ).
2
1 1
arccos( +ℎ)−arccos( )
2 2
2. Find limℎ→0 ℎ
.
1 1
arccos( −ℎ)−arccos( )
2 2
3. Find limℎ→0 .
ℎ

4. Prove the following identities for inverse trigonometric functions.

𝜋
a) arcsin(𝑥) + arccos(𝑥) =
2
b) arcsin(−𝑥) = − arcsin(−𝑥)
c) arccos(−𝑥) = 𝜋 − arccos(𝑥)

5. Find the equation of the tangent line to the graph of 𝑦 = cot −1 𝑥 at 𝑥 = −1.
6. Prove the following identities for hyperbolic functions.

a) cosh2 x = ½cosh 2x + ½
b) sinh4 x = 3/8 - ½cosh 2x + 1/8cosh 4x
c) cosh4 x = 3/8 + ½cosh 2x + 1/8cosh 4x
d) cosh x + cosh y = 2 cosh ½(x + y) cosh ½(x - y)
e) cosh 3x = 4 cosh3 x — 3 cosh x
f) sinh 4x = 8 sinh3 x cosh x + 4 sinh x cosh x
g) cosh 4x = 8 cosh4 x — 8 cosh2 x + 1

7. Prove the following identities for inverse hyperbolic functions.

1
a) sech−1 𝑥 = cosh−1 𝑥
1
b) csch−1 𝑥 = sinh−1 𝑥
1
c) cothh−1 𝑥 = tanh−1 𝑥

1 √(𝑥 2 +1)
8. Prove: csch−1 𝑥 = 𝑙𝑛 ( + ) for all x ≠0.
𝑥 |𝑥|

28
 29

4.3 Higher Order Derivatives and Implicit Differentiation

A. Higher Order Derivatives
Because the derivative 𝑓 ′ (𝑥), of a function f(x) is itself a function, so we can take the derivative
of f ′ ( x). The derivative of f ′ (x) is referred to as the second derivative of f(x) and written 𝑓 ′′ (𝑥)
or 𝑓 [2] (x). Similarly we can define the third derivative of 𝑓(𝑥), which is written as 𝑓 ′′′ (𝑥) or
𝑓 [3] (x). This differentiation process can be continued to define the nth derivative of f( x) for any
natural number n. Because the “prime” notation for derivatives would eventually become somewhat
messy, it is preferable to use the numerical notation 𝑓 [n] (x) to denote the nth derivative of f(x). The
2nd, 3rd, etc derivatives of 𝑓 are called the higher order derivatives of f.
Alternative Notations: Let y = f(x). We also denote the nth derivative of f(x) by 𝑦 [𝑛] , by the
𝑑𝑛𝑦
fractional notation 𝑑𝑥 𝑛 or by the Euler’s notation 𝐷𝑥𝑛 (𝑦). By convention we use the symbol f [ 0 ] (x)
to mean f(x).

1
Example1: Let 𝑓(𝑥) = . Find a formula for 𝑓 [𝑛] (𝑥) and prove it by the PMI.
𝑥
1
Solution: 𝑓 [0] (𝑥) = 𝑥,
𝑑 −1
𝑓 [1] (𝑥) = 𝑑𝑥 (𝑥 −1 ) = (−1)𝑥 −2 = ,
𝑥2
𝑑 (−1)(−2)
𝑓 [2] (𝑥) = 𝑑𝑥 (−𝑥 −2 ) = (−1)(−2)𝑥 −3 = ,
𝑥3
𝑑 (−1)(−2)(−3)
𝑓 [3] (𝑥) = 𝑑𝑥 ((−1)(−2)𝑥 −3 ) = (−1)(−2)(−3)𝑥 −4 = 𝑥4

Calculation of 𝑓 [𝑛] (𝑥) for the first few natural numbers n leads us into the
following claim.
(−1)𝑛 𝑛!
Claim: 𝑓 [𝑛] = . The reader can prove the claim by the PMI.
𝑥 𝑛+1
B. Implicit Differentiation
There are two ways of expressing 𝑦 as a function of 𝑥, by an equation. These are by
a. equations of the form 𝑦 = 𝑓( 𝑥) or
b. Any other equation relating x and y. (Such equations can be converted to the form
𝐹 (𝑥, 𝑦) = 0.)
Functions that are given in form (a) are called explicit functions and those given in the form of (b) are
called implicit functions.
The differentiation formulas we have learnt so far are applicable only to explicit functions.

29
 30

Procedures for Differentiating an Implicit function

Step 1: Differentiate both sides of the given equation with respect to x. At this step recall that any
function of y is a function of function as y itself is a function of x; hence Chain rule is used to
differentiate functions of y.
𝑑𝑦
Step 2: Solve the equation obtained in step 1 for 𝑑𝑥 , in terms of x and y only.

This process is called implicit differentiation. Higher order derivatives can also be obtained by implicit
𝑑2 𝑦 𝑑𝑦 𝑑𝑦
differentiation. For instance, to find 𝑑𝑥 2 , we can proceed as follows. First, find 𝑑𝑥. Let 𝑑𝑥
=

𝐹(𝑥, 𝑦). Then differentiate both sides of this equation implicitly with respect to x. The left hand side
𝑑2 𝑦 𝑑2 𝑦
of the resulting equation is . Finally write an expression for in terms of x and y only.
𝑑𝑥 2 𝑑𝑥 2

𝑑𝑦
Example 2: Find 𝑑𝑥 where 𝑥 2 − 5𝑥𝑦 + 3𝑦 2 = 7
Solution:

Example 3: Assume that 𝑦 is a function of 𝑥. Find y'' if 𝑥 2 + 𝑥𝑦 + 𝑦 2 = 1 .

Solution:
Step (1): Find 𝑦′.
To find 𝑦′ we proceed as follows.
𝑑 2 𝑑 −2𝑥 − 𝑦
[𝑥 + 𝑥𝑦 + 𝑦 2 ] = [ 1] ⟹ 2𝑥 + 𝑦 + 2𝑦. 𝑦 ′ = 0 ⟹ 𝑦 ′ =
𝑑𝑥 𝑑𝑥 2𝑦

Step (1): Find 𝑦′′.

30
 31

To find 𝑦′′ we proceed as follows.

−2𝑥−𝑦 −2𝑥−𝑦
𝑑 𝑑 −2𝑥−𝑦 (−2−𝑦 ′ )2𝑦−2𝑦′ (−2𝑥−𝑦) (−2− )2𝑦−2( )(−2𝑥−𝑦)
(𝑦 ′ ) = 𝑑𝑥 [ ] ⟹ 𝑦 ′′ = ⟹ 𝑦 ′′ = 2𝑦 2𝑦
𝑑𝑥 2𝑦 4𝑦 2 4𝑦 2

Example 4: Find an equation of the line tangent to the graph of (x2+y2)3 = 8x2y2 at the point (-1, 1).
Answer: y = x + 2
Exercises 4.3:
1. Assume that 𝑦 is a function of x. Find y'' (in terms of x and y only) if
𝑥 2 + 𝑥𝑦 + 𝑦 2 = 1 .
2. Find all points (x, y) on the graph of x2/3 + y2/3 = 8 where lines tangent to the graph at
(𝑥, 𝑦)have slope −1 .
3. Find the points where the tangent line is parallel to the x-axis for the curve,

25 y 2  12 xy  4 x 2  1 .

4. Find the equations for the tangent line to the ellipse 4 x 2  y 2  72 that are perpendicular to
the line x  2 y  3  0 .

5. Find the equation for the normal line to the hyperbola 4 x 2  y 2  36 that are parallel to the
line 2 x  5 y  4  0
6. Using implicit differentiation find the first and second derivative at the indicated point.

y2
a. x  y  3 at (1,4) b.  1  x 2 (0,1) c. y 2 sin 2 x  2 y
x y
( 4 ,2)

7. Find the 3rd derivative of the following functions.

a. f ( x)  x 2 e 3 x b. f ( x)  ln(ln( 5 x)) c f ( x)  cos 2 x  e tan x

31
 32

4.4 Extrema of a Function

The remaining sections of Chapter 4 discuss several applications of derivatives. In sections 4.4 to 4.6,
we consider some theorems useful for the applications. The following diagram illustrates the chain of
interdependence between the theorems.

Extreme value Rolle's Mean Value Monotonicity

Theorem Theorem Theorem Tests

First Derivative
Concavity Test
Fermat's ?
Theorem Tests

Second
Derivative Test

Global Extrema

Definition 1 (global extrema):: Let 𝑓 be defined on a set 𝐼 containing 𝑐. Then 𝑓(𝑐) is called

a) the maximum value of 𝑓 on 𝐼 if 𝑓(𝑐) ≥ 𝑓(𝑥) for all x∈ 𝐼.

b) the minimum value of 𝑓 on 𝐼 if 𝑓(𝑐) ≤ 𝑓(𝑥) for all x∈ 𝐼.

c) an extreme value of 𝑓 on 𝐼 if it is either the global maximum value or the global minimum
value of 𝑓on 𝐼.

The minimum value and the maximum value of 𝑓 are also called the global (absolute) minimum value and
the global (absolute) maximum value of f, respectively.

If 𝑓(𝑐) is an extreme value of 𝑓, then the number 𝑐 is called an extreme point of 𝑓. Similarly the point
(𝑐, 𝑓(𝑐)) is also called an extreme point of 𝑓. The term extreme is sometimes called extremum. Its
plural form is extrema. A function need not have a minimum value or a maximum value on a set 𝐼.
Existence of extreme values of 𝑓 on I depends on the nature of both 𝐼 and 𝑓.

𝑥2, 𝑥 ≠ 0
Example 1: Let 𝑓(𝑥) = 𝑥 2 and 𝑔(𝑥) = {
2, 𝑥 = 0

32
 33

Then 𝑓 has both maximum and minimum values on [−1, 1] but it has no maximum value on
(−1, 1). On the other hand 𝑔 has maximum but not minimum value on [−1, 1].

Theorem 1[The Extreme Value Theorem(EVT)]: If 𝑓 is continuous on a closed interval 𝐼 =

[𝑎, 𝑏], then 𝑓 has both a maximum and a minimum value on 𝐼.

The proof of this theorem is not within the scope of this book.

Remark: The converse of this theorem is not true in general.

Local Extrema and Critical Numbers

Definition 2 (local extrema): Let 𝑓 be a function with domain 𝐷 and 𝑐 be an interior point of 𝐷.
Then 𝑓(𝑐) is called

a) a local maximum value of f if there exists an open interval 𝐼 containing 𝑐 such that
𝑓(𝑐) ≥ 𝑓(𝑥) for all x ∈ 𝐼.

b) a local minimum value of f if there exists an open interval I containing 𝑐 such that 𝑓(𝑐) ≤
𝑓(𝑥) for all x ∈ 𝐼 .

c) a local extreme value of f if 𝑓(𝑐) is either a local maximum or a local minimum value of f.
Local minimum values and local maximum values of 𝑓 are also called relative minimum values and
relative maximum values of f, respectively.

Clearly, a global extreme point of 𝑓 on 𝐼 is also a local extreme point, if it is not an end point of I.

Figure 5: Extreme Points of a function f

33
 34

Given the graph of 𝑓, its local and global extreme points can easily be detected. Local maximum
points are "hill tops" and local minimum points are" valley bottoms" of the graph. But detection of
extreme points/values from the formula of a function is difficult, in general.

At local extreme points tangent lines are either horizontal or they do not exist, i.e., the slope
(derivative) of a graph is undefined or zero at any local extreme point. So derivatives are useful for
locating maxima and minima of functions.

Definition 3 (critical number): A number c is said to be a critical point ( critical number) of a function
𝑓 if

i) 𝑓 is defined at c

ii) either 𝑓 ′ (𝑐) = 0 or 𝑓 ′ (𝑐) does not exist.

Example 4.27: Find all critical numbers for 𝑓(𝑥) = 𝑥 3 − 3𝑥 2 + 1.

Solution: 𝑓′(𝑥) = 3𝑥 2 − 6𝑥. Since the domain of both 𝑓 and 𝑓′ is ℝ , so 𝑐 is a critical

number of 𝑓 if and only if 𝑓 ′ (𝑐)=0 .

𝑐 is a critical number of 𝑓 ⟺ 𝑓 ′ (𝑐)=0

⟺ 3𝑐 2 − 6𝑐=0

⟺ 3𝑐 2 − 6𝑐=0

⟺ 𝑐=0 or 𝑐 = 2.

Hence, the critical points of 𝑓 are 𝑐 = 0 and 𝑐 = 2.

2
Example 2: Find all critical numbers for 𝑓(𝑥) = 2𝑥 − 3𝑥 3 .

1
Solution: 𝑓 ′ (𝑥) = 2 − 2 𝑥 −3 . The domain of 𝑓 is ℝ and the domain of 𝑓′ is
ℝ\{0} . Since 𝑓 ′ (0) does not exist and 0 is in the domain of 𝑓, so 0 is a
1
critical point of 𝑓.𝑓 ′ (𝑐)=0 ⟺ 2𝑐 − 2 𝑐 −3 =0 ⟺ 𝑐=1.

The following theorem is also called the fundamental theorem of local extrema or Fermat’s theorem.

Theorem 2 [A Necessary Condition for Local Extrema]: If c is an extreme point of f, then c is a

critical point of f.

Proof:

34
 35

Case 1: If c is a local maximum point of f. If 𝑓 ′ does not exist at 𝑥 = 𝑐 then there is nothing to
prove. Thus assume that 𝑓 ′ (𝑐) exist. This implies that

f ( x )  f (c ) f ( x )  f (c )
lim and lim
x c  xc x  c  xc

exist and they are equal to 𝑓 ′ (𝑐). Since c is a local maximum point, so there exist an open interval 𝐼
𝑓(𝑥)−𝑓(𝑐)
containing 𝑐 such that 𝑓(𝑥) − 𝑓(𝑐) ≤0 for all x in I. Then lim𝑥→𝑐+ ≤ 0 and
𝑥−𝑐
𝑓(𝑥)−𝑓(𝑐)
lim𝑥→𝑐− 𝑥−𝑐
≥ 0. Hence 𝑓 ′ (𝑐)≤ 0 𝑎𝑛𝑑 𝑓 ′ (𝑐)≥ 0.Thus 𝑓 ′ (𝑐)=0.

Case 2: If 𝑐 is a local minimum point (Exercise)

Note: Some authors attribute this theorem to Fermat.

≤
Remark: The converse of Theorem is not true (that is a critical number is not necessarily an extreme
point). See the following counter example.

Example 3(A critical number that is not an extreme point):

3
Let 𝑓(𝑥) = √𝑥 . Since 𝑓′(0) does not exist and 0 is in the domain of 𝑓, so 0 is a critical point of 𝑓.
Note that 0 is not a local maximum point of 𝑓 because for every interval 𝐼 containing 0 if
x𝜖(−∞, 0) ∩ 𝐼, then 𝑓(0) > 𝑓(𝑥). Similarly 0 is not a local minimum point of 𝑓 because for every
interval 𝐼 containing 0 if x𝜖(0, ∞) ∩ 𝐼, then 𝑓(0) < 𝑓(𝑥).

The only points where a function 𝑓 can possess global extreme values on I are

(i) critical points of 𝑓 (i. e, interior points of 𝐼 where 𝑓′ is zero or 𝑓′ does not exist ) or

(iii) end points of 𝐼

Procedures for finding the global extreme values of a continuous function f on a closed
interval [a, b]

Step 1: Find the critical numbers of 𝑓 in (𝑎, 𝑏).

Step 2: Evaluate 𝑓 at each critical number in (𝑎, 𝑏).

Step 3: Evaluate 𝑓 at and the end point of [𝑎, 𝑏].

Step 4: The greatest of the values in Step 2 and Step 3 is the maximum. The least is the
minimum..

35
 36

1
Example 4: Find the extrema of 𝑓(𝑥) = 𝑥 3 − 3𝑥 2 + 1 on [− , 4]. Also, find the range of 𝑓
2
1
on [− 2 , 4].

Solution:

1
Step 1: From Example 4.27, the critical numbers of 𝑓 are 0 and 2. Both 0 and 2 are in [− 2 , 4].

Step 2: 𝑓(0) = 1 and 𝑓(2) = −3 .

1 1
Step 3: 𝑓 (− 2) = 8 and 𝑓(4) = 17.

Step 4: The greatest of the values in Step 2 and Step 3 is 17 and the least is -3.

1
Therefore, the maximum is 17 and the minimum is -3. The range of 𝑓 on [− 2 , 4] is [−3,17]. //

2
Example 5: Find the extrema of 𝑓(𝑥) = 2𝑥 − 3𝑥 3 on [−1,3].

Solution:

Step 1: From Example 4.28, the critical numbers of 𝑓 are 0 and 1. Both 0 and 1 are in [−1, 3].

Step 2: 𝑓(0) = 0 and 𝑓(1) = −1 .

3
Step 3: 𝑓(−1) = −5 and 𝑓(3) = 6 − 3√9.

Step 4: The greatest of the values in Step 2 and Step 3 is 0 and the least is -5 .

Therefore, maximum is 0 and the minimum is -5. //

Remark: The absolute extrema of a function on other types of intervals can be obtained by
using some other tools such as montonicity. For instance, if 𝑓 is strictly increasing on (−∞, 𝑎) ∪
(𝑎, ∞), then (𝑎, 𝑓(𝑎)) is a global minimum point of 𝑓.

Exercise 4.4:

1. Define 𝑓 on [0, 4] by 𝑓(𝑥) = 𝑥 + 1 for 0 ≤ x < 2 and 𝑓(𝑥) = 1 𝑓𝑜𝑟 2 ≤ 𝑥 ≤ 4. Use the
extreme value to show that 𝑓 is not continuous.
2. Give an example of a continuous function defined on the interval (1, 2] which does not achieve a
maximum value on the interval. Explain why the existence of such a function does not contradict
the extreme value theorem.

36
 37

4.5 Rolle's Theorem and the Mean value Theorem

A. Rolle's Theorem

Let 𝑓 be continuous on [𝑎, 𝑏], differentiable in (𝑎, 𝑏) and 𝑓(𝑎) = 𝑓(𝑏). Then the part of the graph
of 𝑓 from (𝑎, 𝑓(𝑎)) to (𝑏, 𝑓(𝑏)) is a continuous and smooth curve. Such a graph is either horizontal
or has a point where tangent line is horizontal. In other words, there is a number c in [a, b] at which
the derivative of 𝑓 is zero.

Figure 2: Rolle's Theorem

Theorem 1[Rolle's Theorem]:

If

(i) 𝑓 is continuous on a closed interval [a, b],

( ii) 𝑓 is differentiable on (𝑎, 𝑏) and

(iii) 𝑓(𝑎) = 𝑓(𝑏)

then

there exists a number 𝑐 in (a, b) such that 𝑓′(𝑐) = 0.

Proof:

Case (1): If 𝑓 is constant on [𝑎, 𝑏] (say 𝑓(𝑥) = 𝑘). In this case 𝑓 ′ (𝑥) = 0 for every 𝑥 in
(𝑎, 𝑏), so the required number 𝑐 can be any number in (𝑎, 𝑏).

Case (2): If 𝑓 is not constant on [𝑎, 𝑏] . In this case there exist some 𝑥0 in (𝑎, 𝑏) such that
𝑓(𝑥0 ) ≠ 𝑓(𝑎). WLOG let 𝑓(𝑥0 ) > 𝑓(𝑎). By the extreme value theorem , 𝑓 has a maximum value
on [𝑎, 𝑏]. That is, there exist 𝑐𝜀[𝑎, 𝑏]such that 𝑓(𝑐) ≥ 𝑓(𝑥)∀𝑥𝜖[𝑎, 𝑏]. Hence, 𝑓(𝑐) ≥ 𝑓(𝑥0 ) >

37
 38

𝑓(𝑎) = 𝑓(𝑏). This implies 𝑐 ≠ 𝑎, 𝑏.Thus 𝑐 is an interior point of [𝑎, 𝑏] . By Fermat's theorem
𝑓 ′ (𝑐) = 0. ∎

Example 1: Find all numbers c that satisfy the conclusion of Role's theorem for 𝑓(𝑥) =
cos 2𝑥 , 𝑥𝜖[0, 𝜋].

Solution: Since 𝑓(0) = 𝑓(𝜋), 𝑓 is continuous on[0, 𝜋] and differentiable on (0, 𝜋), so by
Rolle's theorem there exists a number 𝑐 in (a, b) such that 𝑓′(𝑐) = 0. This implies that −2 sin 2𝑐 =
𝜋
0. ⟹ 2𝑐 = 0 or 2𝑐 = 𝜋. ⟹ 𝑐 = 0 or 𝑐 = 2 . //

The Rolle's theorem can be used along with the intermediate value theorem to show both the
existence and number of solutions of certain equations.

Example 2: Show that the equation 𝑥 5 + 10𝑥 + 1 = 0 has exactly one real root.

Solution: Let 𝑓(𝑥) = 𝑥 5 + 10𝑥 + 1. Since 𝑓 is continuous on[−1,0] and 𝑓(0) ∗

𝑓(−1) < 0, so there exist 𝑟𝜖(−1,0) such that 𝑓(𝑟) = 0.

Suppose 𝑓 has more than one root and let 𝑟′ be a root of 𝑓 different from 𝑟. WLOG let 𝑟 < 𝑟 ′ .
Since 𝑓(𝑟) = 𝑓(𝑟′), 𝑓 is continuous on[𝑟, 𝑟′] and differentiable on(𝑟, 𝑟′) , so by Rolle's theorem
there exists a number 𝑐 in (r, r') such that 𝑓′(𝑐) = 0. This is a contradiction, because 𝑓 ′ (𝑥) =
5𝑥 4 + 10 > 0 for every x. //

2
Example 3: Let 𝑓(𝑥) = 1 − 𝑥 3 . Show that 𝑓(−1) = 𝑓(1) but there does not exist
𝑐𝜖(−1,1) such that 𝑓 ′ (𝑐) = 0. Why this does not contradict Rolle's theorem?

−1
2
Solution: 𝑓′(𝑥) = 3 𝑥 3 ≠ 0∀𝑥𝜖(−1,1) . Hence, there does not exist 𝑐𝜖(−1,1) such that 𝑓 ′ (𝑐) = 0.

Note that f is not differentiable on (−1,1) so Rolle's theorem does not apply to this problem. //

B. The Mean Value Theorem

Let 𝐴(𝑎, 𝑓(𝑎)) and 𝐵(𝑓(𝑏), 𝑏) be two points on the graph of a differentiable function f. The
⃡ is 𝑓(𝑏)−𝑓(𝑎). The mean value theorem states that there is at least one point
slope of the line 𝐴𝐵
𝑏−𝑎

⃡ . This is intuitively plausible from the following

P(c,f(c)) where the tangent line is parallel to 𝐴𝐵

38
 39

figure.

Figure 3: Mean Value Theorem

Theorem 2 [Mean Value Theorem (MVT)]:

If

(i) 𝑓 is continuous on a closed interval [a, b] and

( ii) 𝑓 is differentiable on (𝑎, 𝑏)

then
𝑓(𝑏)−𝑓(𝑎)
there exists a number 𝑐 in (a, b) such that 𝑓 ′ (𝑐) = .
𝑏−𝑎

Proof: The linear function whose graph is the line through (𝑎, 𝑓(𝑎)) and (𝑏, 𝑓(𝑏)) is

𝑓(𝑏) − 𝑓(𝑎)
𝑔(𝑥) = ( ) (𝑥 − 𝑎) + 𝑓(𝑎)
𝑏−𝑎

Apply Rolle's theorem to the function ℎ(𝑥) = 𝑓(𝑥) − 𝑔(𝑥), 𝑥 𝜖[𝑎, 𝑏]. ∎

Example 4: Find all numbers 𝑐 that satisfy the conclusion of the Mean Value Theorem for
3
𝑓(𝑥) = 1 + √𝑥 − 1 on [2,9].

2
1
Solution: 𝑓 ′ (𝑥) = 3 [𝑥 − 1]−3 . Thus 𝑓 is differentiable on (2, 9). Moreover, f is continuous [1,9]. By
3
2
𝑓(9)−𝑓(2) 1 3−2 1 7 2
the MVT , there exists c such that 𝑓 ′ (𝑐) = 9−2
. ⇒ 3 [𝑐 − 1]−3 = 9−2 = 7. ⇒ 𝑐 = 1 + (2)
//

39
 40

The MVT has both geometrical and physical interpretations. Geometrically, it guarantees a tangent
line that is parallel to the secant line through (𝑎, 𝑓(𝑎)) and (𝑏, 𝑓(𝑏)) . In terms of rates of change,
the MVT implies that there exists a point in (a, b) at which the instantaneous rate of change is equal
to the average rate of change over [a, b].

The MVT is used often to prove many other theorems. It can also be used to solve some problems
directly.

Example 5: Using the MVT show that

a) If |𝑓 ′ (𝑥)| ≤ 1∀𝑥 in (𝑎, 𝑏), then |𝑓(𝑥) − 𝑓(𝑦)| ≤ |𝑥 − 𝑦|∀𝑥, 𝑦𝜖(𝑎, 𝑏).
b) |sin 𝑥| ≤ |𝑥|∀𝑥 in ℝ.

Solution:

a) Let 𝑥, 𝑦𝜖(𝑎, 𝑏) and WLOG let x<y. Since 𝑓 is differentiable on [𝑥, 𝑦], so f is continuous
𝑓(𝑦)−𝑓(𝑥)
on [𝑥, 𝑦]. By the MVT, there exists a number 𝑐 in (a, b) such that 𝑓 ′ (𝑐) = . Hence
𝑦−𝑥
𝑓(𝑦)−𝑓(𝑥)
| 𝑦−𝑥
| ≤ |𝑓 ′ (𝑐)|. ⟹ |𝑓(𝑥) − 𝑓(𝑦)| ≤ |𝑥 − 𝑦|∀𝑥, 𝑦𝜖(𝑎, 𝑏).

b) Put 𝑓(𝑥) = sin 𝑥 and apply (a) with 𝑦 = 0. //

The following consequences of the MVT provide the gateway to integral calculus.
The first corollary is the converse of Theorem 4.2. The second corollary states that if two functions
have the same derivative, then their difference is a constant.

Corollary 1: If 𝑓 ′ (𝑥) = 0 at each point of an interval 𝐼, then there exists a constant 𝑐 such that
𝑓(𝑥) = 𝑐 for all x in I .

Proof: Let 𝑎 and 𝑏 be any two distinct elements of I. By the Mean Value Theorem there
𝑓(𝑏)−𝑓(𝑎)
exists a number 𝑐 in (a, b) such that 𝑓 ′ (𝑐) = . But 𝑓 ′ (𝑐) = 0, hence 𝑓(𝑎) = 𝑓(𝑏).
𝑏−𝑎

As 𝑎 and 𝑏 are arbitrary, it implies that 𝑓 has a constant value on 𝐼. ∎

Corollary 2: If 𝑓 ′ (𝑥) = 𝑔′ (𝑥) at each point of an interval 𝐼, then there exists a constant
𝑐 such that 𝑓(𝑥) − 𝑔(𝑥) = 𝑐 for all x in I .

40
 41

Proof: Let ℎ = 𝑓 − 𝑔. By assumption ℎ′ (𝑥) = 0∀𝑥𝜖𝐼. By corollary 1, ∃𝑐 ∈ 𝐼 such that

ℎ(𝑥) = 𝑐 for all x in I. Hence, 𝑓(𝑥) − 𝑔(𝑥) = 𝑐 for all x in I . ∎

Example 6: Find the function whose derivative is 𝑠𝑖𝑛 𝑥 and whose graph passes through the
point (0, 2).

Solution: Let 𝑓 be the required function. It is given that 𝑓 ′ (𝑥) = sin 𝑥 & 𝑓(0) = 2. Since the
derivative of 𝑔(𝑥) = − cos 𝑥 is also sin 𝑥, so by Corollary 2 there exists a constant c such that
𝑓(𝑥) − 𝑔(𝑥) = 𝑐 . ⟹ 𝑓(𝑥) = − cos 𝑥 + 𝑐. ⟹ 𝑓(0) = −cos (0) + 𝑐. ⟹ 2 = −1 +𝑐. Hence
𝑓(𝑥) = − cos 𝑥 + 3. //

Exercises 4.5:

1. Give the number of real roots and locate each root between successive integers for the
equation.
a) 2x3  9x 2  1  0 b) 3 x 5  5 x  1  0

2. Using the MVT show that

a. if f ( x) 1 x in some interval I, then f ( x1 )  f ( x2 )  x1  x2 x1 , x2  I

b. sin x  x x R.

41
 42

4.6 Monotonicity and Concavity: The Shape of a Graph

In this section we learn how to find intervals where a function falls or rises (this is called
monotonicity) and intervals where it curves up or curves down (this is called concavity).
Monotonicity and concavity of a function provide vital information about the shape of its graph. We
use the first and second derivatives of the function to know its montonicity and concavity.

A. Monotonicity Tests
It may be intuitively clear that the tangent lines to the graph of a strictly increasing function have
positive slopes and those of a strictly decreasing function have negative slopes. These observations
lead to the following tests for monotonicity. See also Theorem 1.
Theorem 1[Monotonicity Tests): Let 𝑓 be continuous on [𝑎, 𝑏] and differentiable on (𝑎, 𝑏). Then
f is
a) strictly increasing if f ′ (x) > 0 for every x ϵ (a, b).
b) strictly decreasing if f ′ (x) < 0 for every x ϵ (a, b).
c) increasing if f ′ (x) ≥ 0 for every x ϵ (a, b).
d) decreasing if f ′ (x) ≤ 0 for every x ϵ (a, b).

Proof: We prove (a) only. Assume that 𝑓 ′ (𝑥) > 0 for every x 𝜖 (𝑎, 𝑏). Let 𝑥, 𝑦𝜖(𝑎, 𝑏) with 𝑥 < 𝑦.
We want to show that 𝑓(𝑥) < 𝑓(𝑦). By applying the MVT to 𝑓 on [𝑥, 𝑦] , we conclude that there
𝑓(𝑦)−𝑓(𝑥)
exists a number 𝑐 in (x, y) such that 𝑓 ′ (𝑐) = . Since 𝑓 ′ (𝑥) > 0 for every x 𝜖 (𝑎, 𝑏), so
𝑦−𝑥
𝑓(𝑦)−𝑓(𝑥)
0 < 𝑓 ′ (𝑐) = . This implies that 0 < 𝑓(𝑦) − 𝑓(𝑥), since 𝑦 − 𝑥 > 0.
𝑦−𝑥

Hence, 𝑓(𝑥) < 𝑓(𝑦). ∎

Example 1: Find the region (i.e., interval or union of intervals) where 𝑓(𝑥) = 3𝑥 4 − 4𝑥 3 −
12𝑥 2 + 5 is strictly decreasing and where it is strictly increasing.

Solution: We need to solve the inequalities f ′ (x) < 0 and f ′ (x) > 0. Note that f ′ (x) = 12x 3 −
12x 2 − 24x = 12x(x + 1)(x − 2). Thus 𝑥 = 0, 𝑥 = −1 & 𝑥 = 2 are the roots of 𝑓′. These points
partition the real line as ℝ = (∞, −1) ∪ {−1} ∪ (−1,0) ∪ {0} ∪ (0,2) ∪ {2} ∪ (2, ∞).

The signs of 12x, (x + 1), (x − 2) and 𝑓′ in each subinterval is determined from the following sign
chart.

42
 43

(∞, −1) {-1} (-1,0) {0} (0,2) {2} (2,∞)

12𝑥 −−− −− −−− 0 +++ ++ +++
𝑥+1 −−− 0 +++ +++ +++ ++ +++
𝑥−2 −−− −− −−− −− −−− 0 +++
𝑓′ −−− 0 +++ 0 −−− 0 +++

Since f ′ (x)<0 for all 𝑥𝜖(∞, −1) ∪ (0,2), so f is strictly decreasing on (∞, −1) ∪ [0,2]. Similarly f
is strictly increasing on [−1,0] ∪ [2, ∞).

B. The First Derivative Test

Recall from Fermat's theorem that every local extreme point of 𝑓 is a critical point of 𝑓. But not
every critical point is a local extreme point. There are two tests that help us decide if a critical point is
a local extreme point or not. These are the first derivative test and the second derivative test.

Near a local maximum point of 𝑓 the graph of 𝑓 is increasing on the left and decreasing on the right.
Near a local minimum point of 𝑓 the graph of 𝑓 is decreasing on the left and increasing on the right .
This suggests a sufficient condition for a critical number to be a local maximum/minimum point.

increasing decreasing increasing

Figure 4: Intervals of Monotonicity of a Function

We say 𝑓′ changes sign from positive to negative at 𝑐 if there exists 𝛿 > 0 such that 𝑓 ′ (𝑥) >
0∀𝑥𝜖(𝑐 − 𝛿, 𝑐) and 𝑓 ′ (𝑥) < 0∀𝑥𝜖(𝑐, 𝑐 + 𝛿) . Similarly we say 𝑓′ changes sign from negative to
positive at 𝑐 if there exists 𝛿 > 0 such that 𝑓 ′ (𝑥) < 0∀𝑥𝜖(𝑐 − 𝛿, 𝑐) and 𝑓 ′ (𝑥) < 0∀𝑥𝜖(𝑐, 𝑐 + 𝛿).

43
 44

Theorem 2 [The First Derivative Test]: Let 𝑥 = 𝑐 be a critical point of 𝑓 that is continuous in an
open interval containing 𝑐. Let 𝑓 be differentiable in the interval except possibly at 𝑐. Then

a) 𝑐 is a local maximum point of 𝑓 if 𝑓′ changes sign from negative to positive at c,

b) 𝑐 is a local minimum point of 𝑓 if 𝑓′ changes sign from positive to negative at c and

c) 𝑐 is not a local extreme point of 𝑓 if 𝑓′ has the same sign on both sides of c.

Proof: We prove (a) only. If 𝑓′ changes sign from positive to negative at 𝑐 , then there exists 𝛿 > 0
such that 𝑓 ′ (𝑥) > 0∀𝑥𝜖(𝑐 − 𝛿, 𝑐) and 𝑓 ′ (𝑥) < 0∀𝑥𝜖(𝑐, 𝑐 + 𝛿). By monotonicity tests, f is strictly
increasing on (𝑐 − 𝛿, 𝑐) and strictly decreasing on( 𝑐, 𝑐 + 𝛿). By the definition of strictly increasing
and strictly decreasing functions it follows that if 𝑥𝜖(𝑐 − 𝛿, 𝑐), then f(x)<f(c) and if 𝑥𝜖(𝑐, 𝑐 + 𝛿) ,
then 𝑓(𝑥) < 𝑓(𝑐). ∎

Procedures for Finding Local Extrema of a function f

Step 1: Find all critical points of f.

Step 2: Apply either the first derivative test or the second derivative test.
2
Example 2: Find the local maximum and local minimum points of 𝑓(𝑥) = 𝑥(1 − 𝑥)5 .

Solution:

2 3
2𝑥 5
Step 1: 𝑓 ′ (𝑥) = (1 − 𝑥)5 + (1 − 𝑥)−5 . At x=1, 𝑓′ does not exist and 𝑓 ′ (7) = 0. Thus x=0 and
5
5
x=7 are the critical numbers of f.

5 5
Step 2: If 𝑥𝜖 (−∞, 7) ∪ (7 , 1) , then 𝑓 ′ (𝑥) > 0 and if 𝑥𝜖(1, ∞), then 𝑓 ′ (𝑥) < 0. Thus 𝑓′ does
5
not change its sign at x= 7; it changes sign from positive to negative at x=1. Therefore, 1 is a local
5
maximum point of f and 7 is not a local extreme point at all. //

C. Tests for Concavity

Definition 1 (concave up & concave down):
Let 𝑓 be a function differentiable on an interval 𝐼. We say the graph of f is
a) concave upward on I if lies above all of its tangent lines on I.
b) concave downward on I if lies below all of its tangent lines on I.

44
 45

Theorem 3 (concave up & concave down):

Let 𝑓 be twice differentiable on an interval 𝐼. The graph of 𝑓 is
a) concave downward on I if f ′′ (x) < 0 for every 𝑥 𝜖 𝐼.
b) concave upward on I if f ′′ (x) > 0 for every 𝑥 𝜖 𝐼.

Proof: (a). Assume that f ′′ (x) < 0 for every x ϵ I and let 𝑎𝜖𝐼. We want to prove that the graph of
𝑓 is below the tangent line 𝑡 to the graph of 𝑓 at 𝑐. The linear function representing 𝑡 is 𝑔(𝑥) =
𝑓 ′ (𝑎)(𝑥 − 𝑎) + 𝑓(𝑎). WTS: 𝑓(𝑥) < 𝑔(𝑥)∀𝑥𝜖𝐼.

⟹ 𝑓 ′ (𝑥)(𝑥 − 𝑎) + 𝑓(𝑎) < 𝑓 ′ (𝑎)(𝑥 − 𝑎) + 𝑓(𝑎).

⟹ 𝑓 ′ (𝑥)(𝑥 − 𝑎) + 𝑓(𝑎) < 𝑓 ′ (𝑎)(𝑥 − 𝑎) + 𝑓(𝑎).
𝑓(𝑥)−𝑓(𝑎)
By the MVT applied to 𝑓 on [𝑎, 𝑥], ∃𝑐 ∈ (𝑎, 𝑥) such that 𝑓 ′ (𝑐) = 𝑥−𝑎
.

⟹ 𝑓(𝑥) = 𝑓 ′ (𝑐)(𝑥 − 𝑎) + 𝑓(𝑎).

Since 𝑓′′ < 0 , so 𝑓 ′ is decreasing on 𝐼 and since 𝑐 > 𝑎, so 𝑓 ′ (𝑎) > 𝑓′(𝑐).
Hence, 𝑓 ′ (𝑎)(𝑥 − 𝑎) + 𝑓(𝑎) > 𝑓 ′ (𝑐)(𝑥 − 𝑎) + 𝑓(𝑎). ⟹ 𝑔(𝑥) > 𝑓(𝑥). ∎

f
a x

Figure 5: A concave downward curve

D. The Second Derivative Test

Theorem 4[The Second Derivative Test]: Let 𝑓 ′ (𝑐) = 0 and let 𝑓 ′′ exist on an open interval 𝐼
containing 𝑐.

a) If 𝑓 ′′ (𝑐) < 0, then 𝑓(𝑐) is a local maximum value of 𝑓.

45
 46

b) If 𝑓 ′′ (𝑐) > 0, then 𝑓(𝑐) is a local minimum value of 𝑓.

c) If 𝑓 ′′ (𝑐) = 0, then we can draw no conclusion from this test..

Proof:
𝑓′ (𝑥)−𝑓′ (𝑐)
a) By hypothesis 𝑓 ′′ (𝑐) < 0. ⟹ lim𝑥⟶𝑐 < 0.
𝑥−𝑐
𝑓′ (𝑥)−𝑓′ (𝑐)
⟹ 𝑥−𝑐
< 0.
𝑓′ (𝑥)
⟹ <0
𝑥−𝑐

⟹ 𝑓 ′ (𝑥) > 0 𝑖𝑓 𝑥 < 𝑐 and 𝑓 ′ (𝑥) < 0 𝑖𝑓 𝑥 > 𝑐.

Hence 𝑥 = 𝑐 is a local maximum point of 𝑓.
b) Exercise ∎

Definition 2(inflection point): A point P on the graph of 𝑓 is called an inflection point if the graph
changes from concave upward to concave downward or vice versa at 𝑃.

Example 3: Determine the intervals where 𝑓(𝑥) = 𝑥 3 − 3𝑥 + 1 is concave upward and

concave downward. Also, find the inflection points.

Solution: 𝑓 ′ (𝑥) = 3𝑥 2 − 3 and 𝑓 ′ ′(𝑥) = 6𝑥. To get the interval where 𝑓 is concave upward, we find
the set of all x where 𝑓 ′ ′(𝑥) > 0. Note that 𝑓 ′′ (𝑥) > 0 ⟺ 6𝑥 > 0 ⟺ 𝑥 > 0. Thus 𝑓 is
concave upward on (0, ∞). Similarly, f is concave downward on (−∞, 0]. //

Exercise 4.6
1. Find all extrema of 𝑓(𝑥) = 𝑥 3 (𝑥 − 2)2 .
2. Find all critical numbers of the function 𝑔(𝑡) = 𝑡√4 − 𝑡.
3. Find the absolute maximum and absolute minimum values of f on the interval [0, 3], where f(x)
= 3x2 - 12x + 5.
4. Find the absolute maximum and absolute minimum values of f(x) = sin x + cos x on the
interval [0, π/3].
5. Check that the function 𝑓(𝑥) = 𝑥 √𝑥 + 6 satisfies the three hypotheses of Rolle's Theorem on
the interval [-6, 0]. Find all numbers c that satisfy the conclusion of Rolle's Theorem.
6. Determine whether the Mean Value Theorem can be applied to f on the closed [a, b]. If the Mean
Value Theorem can be applied, find all values of c in the open interval (a,b) such that

46
 47

7. Prove that (1 + 𝑥)𝑛 > 1 + 𝑛𝑥

tan 𝑏 𝑏 𝜋
8. Prove that tan 𝑎 > 𝑎 where 0 < 𝑎 < 𝑏 < 2 .
𝑐
9. Find 𝑐 such that 𝑓(𝑥) = 𝑥 + 𝑥 2 has a relative minimum but not a relative maximum.

47
 48

4.7 Applications of Derivatives

A. Curve Sketching
Guidelines for sketching the graph of a function :
1. Determine the domain and , if possible, the range of the function.
2. Determine the intercepts, asymptotes and symmetry of the graph.
3. Find intervals of montonicity and local extrema
4. Find intervals of concavity and inflection points
5. Using the information from (1) to (4), draw the graph. Draw asymptotes as broken
lines. Plot intercepts, local extreme points and inflection points. Make the curve pass
through these points falling and rising according to (3), concaving up and down
according to (4) and approaching the asymptotes.
Example 1: Sketch the graph of 𝑓(𝑥) = 𝑥 4 − 4𝑥 3.
Solution:
1. The domain of the function is ℝ. As x approaches to infinity and negative infinity f(x)
approaches to infinity. Thus the range is not bounded above. But it is bounded below; this can
be determined using montonicity.
2. The y-intercept is (0,0).
𝑓(𝑥) = 0 ⟺ 𝑥 4 − 4𝑥 3 ⟺ 𝑥 3 (𝑥 − 4) = 0 ⟺ 𝑥 = 0 and 𝑦 = 4. Hence, (0,0) and (4,0) are the x-
intercepts. The function has no vertical asymptote since lim𝑥⟶𝑎 𝑓(𝑥) = 𝑓(𝑎) ≠ ±∞ for every real
number a. Also, it has no horizontal asymptote. Since 𝑓(−𝑥) ≠ 𝑓(𝑥), so f is not an even function.
Neither is it an odd function. So, the graph is not symmetric with respect to the y-axis and the origin.
3. 𝑓 ′ (𝑥) > 0 ⟺ 4𝑥 3 − 12𝑥 2 > 0 ⟺ 𝑥𝜖(3, ∞). It follows that f is strictly increasing on [3, ∞).
Similarly, it is strictly decreasing on 𝑥𝜖(−∞, 3]. Then (3, f(3))=(3,-27) is the local minimum
point.
4. 𝑓 ′′ (𝑥) > 0 ⟺ 12𝑥 2 − 24𝑥 > 0 ⟺ 𝑥𝜖(2, ∞) ∪ (−∞, 0). It follows that f is concave upward
on (2, ∞) ∪ (−∞, 0). Similarly, concave down on (0,2). The points (0,0) and (2, -16) are the
inflection points.
5. Aided with the information from 1 to 4 above, we obtain the following graph of f:

48
 49

Figure 6: The Graph of 𝒇(𝒙) = 𝒙𝟒 − 𝟒𝒙𝟑.

2𝑥 2
Example 2: Sketch the graph of 𝑓(𝑥) = 𝑥 2 −1
Solution:
1. The domain of the function is ℝ\{−1,1}. For the range wait for a while!
2. The y-intercept is (0,0).
𝑓(𝑥) = 0 ⟺ 2𝑥 2 = 0 ⟺ 𝑥 = 0 . Hence, (0,0) is the only x-intercept.
lim𝑥⟶1+ 𝑓(𝑥) = ∞ and lim𝑥⟶−1+ 𝑓(𝑥) = −∞. Hence, the lines x=1 and x=-1 are
vertical asymptotes of f. Moreover, lim𝑥⟶∞ 𝑓(𝑥) = 2 = lim𝑥⟶−∞ 𝑓(𝑥) = 2;
hence y=2 is the only horizontal asymptote of f. Since 𝑓(−𝑥) = 𝑓(𝑥), so 𝑓 is an
even function; i.e., its graph is symmetric about y-axis.
−4𝑥
3. 𝑓 ′ (𝑥) > 0 ⟺ [𝑥 2 −1]2
> 0 ⟺ 𝑥𝜖(−∞, −1) ∪ (−1,0). It follows that f is strictly

increasing on (−∞, 0). Similarly, it is strictly decreasing on 𝑥𝜖(0, ∞) and

(0, 𝑓(0)) = (0,0) is the local minimum point.
2
−4[𝑥 2 −1] −(−4𝑥)(4𝑥)[𝑥 2 −1]
4. 𝑓 ′′ (𝑥) > 0 ⟺ [𝑥 2 −1]4
> 0 ⟺ [𝑥 2 − 1][12𝑥 2 + 4] > 0 ⟺

𝑥𝜖(1, ∞) ∪ (−∞, −1). It follows that f is concave upward on (1, ∞) ∪ (−∞, −1).
Similarly, it is concave down on (−1,1). There are no inflection points on the graph.
5. Aided with the information from 1 to 4, we obtain the following graph of 𝑓:

49
 50

𝟐𝒙𝟐
Figure 7: The Graph of 𝒇(𝒙) =
𝒙𝟐−𝟏

B. Optimization
Optimization theory is the science of selecting the best alternative among a set of infinite alternatives
or a set of finitely large alternatives. It is the science of making best choice or decision. It is an
interdisciplinary field involving mathematics and other fields(economics, management science, etc).
It has many practical applications, especially in economics, engineering, management and physics.
Mathematical optimization theory is concerned mainly with finding the extrema of functions.

Procedures for Solving Optimization Word problems

Step 1: Identify the objective. Is it to maximize or minimize?
Step 2: Identify the variable (quantity) to be maximized or minimized. This quantity is called
the objective function.
Step 3: Express the quantity to be maximized or minimized (i.e. the objective function) as a
function of a single quantity(variable) in the problem. At this step you must also
specify the domain of the function. The quantity on which the objective function
depends is called the decision variable.
Step 4: Apply the theory of extrema. Interpret the solution.

Example 3 [geometrical application- maximizing area]: Of all rectangles with perimeter 400
meters long , find the dimensions (i.e., length and width)of the one with largest area.
Solution:

50
 51

Step 1: The objective is to maximize .

Step 2: The quantity to be maximized is area.
Step 3: Let 𝐴 represents the area of any rectangle with perimeter 400. Let the length and width of
such a rectangle be 𝑥 and 𝑦, respectively. Since 2𝑥 + 2𝑦 = 400, so 𝑦 = 200 − 𝑥. On the other
hand 𝐴 = 𝑥𝑦.
⟹𝐴(𝑥) = 𝑥(200 − 𝑥) . The domain is [0,200]. In the extreme case when width is 0, length
becomes 200.
Step 4: Now, we apply the theory of global extrema. Note that 𝐴′ (𝑥) = 200 − 2𝑥. Thus 𝐴′ (𝑥) =
0 ⟺ 𝑥 = 100 and hence 𝑥 = 100 is a critical point and it is the only critical point. We now evaluate
𝐴 at the critical point and at end points of the domain. Note that 𝐴(0) = 0, 𝐴(200) = 0 and
𝐴(100) = 10,000. Therefore, the largest area is 10,000 square units and the corresponding
dimensions are x=100=y. //

Example 4[geometrical application- maximizing volume]: Of all circular cylinders having the
same total surface area, find the dimensions (radius and height) of the one with maximum volume.

Solution:
Step 1: The objective is to maximize.
Step 2: The quantity to be maximized is volume.
Step 3: Let 𝑉 represents the volume of any circular cylinder with surface area S. Let the radius and
𝑆
height of a cylinder of such type be 𝑥 and 𝑦, respectively. Since 2𝜋𝑥 2 + 2𝜋𝑥𝑦 = 𝑆, so 𝑦 = 2𝜋𝑥 − 𝑥.
𝑆
On the other hand 𝑉 = 𝜋𝑥 2 𝑦. ⟹ 𝑉(𝑥) = 𝜋𝑥 2 (2𝜋𝑥 − 𝑥) . The domain is (0, ∞).
𝑆
Step 4: Now, we apply the theory of global extrema. Note that 𝑉 ′ (𝑥) = 2𝜋𝑥 (2𝜋𝑥 − 𝑥) +

𝑆 𝑆 𝑆 𝑆
𝜋𝑥 2 (− 2𝜋𝑥 2 − 1) = 2𝜋𝑥 − 3𝑥. Thus 𝑉 ′ (𝑥) = 0 ⟺ 2𝜋𝑥 − 3𝑥 = 0 ⟺ 𝑥 = ±√6𝜋 . Since 𝑥

𝑆
cannot be negative , so 𝑥 = √ the only critical number of 𝑓. The reader can check that 𝑉′ is
6𝜋

𝑆
positive to the left and negative to the right of 𝑥 = √6𝜋 throughout its domain (As testing points

√𝑠 𝑆 𝑆
take 6𝜋 and √𝑠). Hence, the volume is maximum when x=√6𝜋 . In this case y=2√6𝜋.

//

51
 52

C. Related Rates Problems

A related rate problem is the problem of finding a rate you cannot measure easily from some other
rate you can. Suppose two or more quantities (say 𝑋1 ,⋯,𝑋𝑁 ) are related by an equation:
𝐹(𝑋1 , ⋯ , 𝑋𝑁 )= 0
Suppose each of these quantities in turn depends on time; that is,
𝑋1 =𝑋1 (𝑡),⋯,𝑋𝑁 = 𝑋𝑁 (𝑡).
If the rates of changes of some of these quantities with respect to time are given, then the rates of
changes of the remaining quantities can be found by differentiating both sides of the equation
relating them with respect to time.
𝑑𝑌 𝑑𝑋
Example 5: Let 5X + 3XY = 4 and let X and Y depend on t. If 𝑑𝑡
= 3, then find when x =2.
𝑑𝑡
Solution: When X = 2, Y=-1.
𝑑 𝑑
5X + 3XY = 4 ⇒ 𝑑𝑡 (5𝑥 + 3𝑋𝑌) = 𝑑𝑡 (4)
𝑑𝑋 𝑑𝑋 𝑑𝑌
⇒ 5 𝑑𝑡 + 3𝑌 𝑑𝑡 + 3𝑋 𝑑𝑡 = 0
𝑑𝑋 𝑑𝑋
⇒ 5 𝑑𝑡 + 3(−1) 𝑑𝑡 + 3(2)(3) = 0
𝑑𝑋
⇒ = −9 //
𝑑𝑡

Guidelines to Solve Related Rates Word Problems

Step 1: Identify quantities (variables) that vary with time.
Step 2: Obtain an equation relating the variables which vary with time.
Step 3: Differentiate both sides of the equation with respect to time. At this step be aware that the Chain rule is
used to differentiate functions of the variables, as the variables themselves are functions of time.
Step 4: Substitute the given values and determine the required values.

Example 6: A ladder 25 ft is leaning against a vertical wall. If the bottom of the ladder is pulled
horizontally away from the wall at 3 ft/s, how fast is the top of the ladder sliding down the wall when
the 15 ft away from the wall?

52
 53

Figure 8: A ladder leaning against a wall

Solution:
Let X : distance between bottom of the ladder and the wall.
Y: distance between the top of the ladder and the bottom of the wall
X and Y are the only quantities that vary with time and they are related by the equation
𝑋 2 + 𝑌 2 = 625.

𝑑 𝑑
𝑋 2 + 𝑌 2 = 625 ⇒ 𝑑𝑡 ( 𝑋 2 + 𝑌 2 ) = 𝑑𝑡 (625)
𝑑𝑋 𝑑𝑌
⇒ 2𝑋 𝑑𝑡 + 2𝑌 𝑑𝑡 = 0
𝑑𝑌
⇒ 2(15 𝑓𝑡)(3 𝑓𝑡/𝑠) + 2(20 𝑓𝑡) 𝑑𝑡 = 0
𝑑𝑌 9
⇒ = −4
𝑑𝑡
//

Example 7: The width of a rectangle is half its length. At what rate is its area increasing if its width is
1 cm and decreasing at 0.5 cm/s ?
Answer: 2o cm2/s

Example 8: A funnel in the form of a cone is 1 inch in diameter at the top and 8 inch deep. Water is
flowing into the funnel at a rate of 12 in3/s and out at the rate of 4 in 3/ s. How fast is the surface of
the water rising when it is 5 inch deep?

53
 54

512
Answer: in/s
625𝜋

D. Indeterminate Forms and L’ Hopitals’ Rule

There are various situations for which the basic limit rules (sum, difference, product, quotient, and
power rule) cannot be used, even if the limits exist. These arise from the fact that division by 0 and
arithmetical expressions involving infinity are not defined. The values of some of these limits can be
determined easily from their form while the values of the others cannot. The later limits are called
indeterminate forms. For instance, if lim𝑥→𝑎 𝑓(𝑥)= ∞ and lim𝑥→𝑎 𝑔(𝑥)= ∞, then clearly
𝑓(𝑥)
lim𝑥→𝑎 [ 𝑓(𝑥) + 𝑔(𝑥)]= ∞; but the values of lim𝑥→𝑎 [ 𝑓(𝑥) − 𝑔(𝑥)] and lim𝑥→𝑎 𝑔(𝑥) cannot be

determined in generality.
The following table contains a list of all indeterminate forms.

Type 1 Type 2 Type 3

0 ∞
0∙ ∞, ∞ − ∞, 00 , 1∞ , ∞0
0 ∞

𝟎
Definition 1( 𝟎 type indeterminate form ): If lim𝑥→𝑎 𝑓(𝑥)= 0 and lim𝑥→𝑎 𝑔(𝑥)= 0, then
𝑓(𝑥) 0
lim𝑥→𝑎 𝑔(𝑥) is called an indeterminate form of the type 0.

∞
Definition 2 ( ∞ type indeterminate form ): If lim𝑥→𝑎 𝑓(𝑥)= ∞ and lim𝑥→𝑎 𝑔(𝑥)= ∞, then
𝑓(𝑥) ∞
lim𝑥→𝑎 𝑔(𝑥) is called an indeterminate form of the type ∞.

Similarly we can define each of the other indeterminate forms. Note that the “x →a” in the above
definitions can be replaced by; x→ 𝑎− , x → a+, x → ∞ and x → −∞.

𝟎
Theorem 1[L’Hopital’s Rule for 𝟎 type Indeterminate Form]:
f(x) 0
Let limx→a g(x) be an indeterminate form of the type 0. Then

f(x) 𝑓′ (𝑥)
limx→a g(x) = lim𝑥→𝑎 𝑔′ (𝑥) ;

if the limit on the right hand is a real number, ∞ or −∞.

54
 55

ex −1
Example 9: limx→0
sin2x

Solution: Since limx→0 ex − 1 = limx→0 sin 2x = 0, so the given limit is an indeterminate form of the
0
type 0.
d
ex −1 (ex −1)
limx→0 sin2x = limx→0 dx
d (L’Hopital’s rule)
(sin2x)
dx
ex
= limx→0 (This is not an indeterminate form)
2cos2x
1
=2

tanx−x
Example 10: limx→0
x3
0
Since limx→0 tan x − x = limx→0 x 3 = 0, so the given limit is an indeterminate form of the type 0.
tanx−x [tanx−x]′
limx→0 x3
= limx→0 [x3 ]′
(L’Hopital’s rule)
sec2 𝑥−1
= limx→0 (This is again an indeterminate form)
3x2
[sec2 𝑥−1]′
= limx→0
[3x2 ]′
2secxtanx
= limx→0 (This is not an indeterminate form)
6x
= 1/3 //

ex
Example 11: limx→0+ x2

Solution: Since limx→0+ ex = 1 and limx→0 +x 2 = 0, so the given limit is not an indeterminate
ex
form. It can be proved by the formal limit definition that limx→0+ x2 = ∞. If L’Hopital’s rule were
used to evaluate this limit the answer would have been ½ . //

Caution: Do not use L’Hopital’s rule for limits which are not indeterminate forms.

∞
Theorem 2 [L’Hopital’s Rule for ∞ type Indeterminate Form]:
f(x) ∞
Let limx→a g(x) be an indeterminate form of the type ∞. Then

f(x) 𝑓′ (𝑥)
limx→a g(x) = lim𝑥→𝑎 𝑔′ (𝑥) ;

if the limit on the right hand is a real number, ∞ or −∞.

To evaluate other indeterminate forms apply algebraic manipulations and convert them into

55
 56

0 ∞
0
or form. These procedures are illustrated by the next examples. For the indeterminate forms in
∞
power form the following procedures may be applied.

Procedures to find 𝒍𝒊𝒎𝒙→𝒂 [𝒇(𝒙)]𝒈(𝒙) where it is an indeterminate Form

Step 1: Let y = [𝑓(𝑥)] 𝑔(𝑥)
Step 2: Tale natural logarithm of both sides. Then, 𝑙𝑛𝑦 = 𝑔(𝑥)𝑙𝑛𝑓(𝑥)
Step 3: Find limx→a g(x)lnf(x) . Let L = limx→a g(x)lnf(x).
Step 4: From steps 3 and 4, L = = limx→a ln y or 𝐿 = ln ( 𝑙𝑖𝑚𝑥→𝑎 𝑦). It follows that 𝑙𝑖𝑚𝑥→𝑎 [𝑓(𝑥)] 𝑔(𝑥) =
𝑎𝐿 .

Example 12: Evaluate limx→0+ xlnx

Solution: Since limx→0+ x = 0 and limx→0+ lnx = −∞ , so the given limit is an indeterminate form
of the type 0. ∞.
lnx ∞
limx→0+ xlnx = limx→0+ 1 . (This is an indeterminate form of the type ∞)
( )
x
d
[lnx]
= lim dx
x→0+ d 1 (L’Hopital’s rule)
( )
dx x
1
= lim x→0+ 1
x
− 2
x
= limx→0+ −x
=0 //

Example 13: Evaluate limx→π− (sec x − tan x) ans. 0

2

Solution: Since limx→π− sec x = ∞ and limx→π− tan x = ∞ , so the given limit is an indeterminate
2 2

form of the type ∞ − ∞.

1−sinx 0
limx→π− (sec x − tan x) = limx→π− (This is an indeterminate form of the type 0)
2 2 cosx
[1−sinx]′
= limx→π− ( L’Hopital’s rule)
2 [cosx]′
− cos x
= limx→π− − sin x
2
=0 //

1 x
Example 14: Evaluate limx→∞ [1 + x]
1 x
Solution: Let y = [1 + x]

56
 57

1
Then, 𝑙𝑛𝑦 = 𝑥𝑙𝑛[1 + ]
𝑥
1
⟹ lim𝑥→∞ 𝑙𝑛𝑦 = lim𝑥→∞ 𝑥𝑙𝑛[1 + 𝑥]
1
⟹ lim𝑥→∞ 𝑙𝑛𝑦 = lim𝑥→∞ 𝑥𝑙𝑛[1 + 𝑥] (The RHS is an indeterminate form of the type 0. ∞.)
1
𝑙𝑛[1+ ] 0
𝑥
=lim𝑥→∞ 1 (This is an indeterminate form of the type 0.)
( )
𝑥
𝑑 1
{𝑙𝑛[1+ ]}
=lim𝑥→∞ 𝑑𝑥 𝑑 1
𝑥
(L’Hopital rule)
( )
𝑑𝑥 𝑥
𝑥 1
(− 2 )
= lim𝑥→∞ 𝑥+1 𝑥
=1
(−1/𝑥 2 )

⟹ lim𝑥→∞ 𝑙𝑛𝑦 = 1.
⟹ ln [ lim 𝑦] = 1
𝑥→∞

⟹ lim𝑥→∞ 𝑦 = 𝑒.
⟹ ln[ lim 𝑦] = 1
𝑥→∞
1 x
⟹ limx→∞ [1 + x] = e //
Note: Do not apply L’Hopital’s rule on limits which are not indeterminate forms.

E. Differentials, Linear Approximation and Newton's Method

Differentials and Linear Approximations

𝑑𝑦
What is the separate meaning of 𝑑𝑥 and 𝑑𝑦 in the fraction 𝑑𝑥? Let 𝑓 be a function, and fix a point 𝑥𝑜

at which 𝑓 is differentiable. For 𝑥 ‘near’ 𝑥𝑜 we have an ‘approximate’ equality

𝑓(𝑥) ≈ 𝑓(𝑥𝑜 ) + 𝑓 ′ (𝑥𝑜 )(𝑥 − 𝑥𝑜 )
This approximation is called linear approximation or tangent line approximation. The analysis of the
accuracy of this approximation is beyond the scope of the present chapter. The quantity
𝐿(𝑥) = 𝑓(𝑥𝑜 ) + 𝑓 ′ (𝑥𝑜 )(𝑥 − 𝑥𝑜 )
is an equation of the line tangent to 𝑓 𝑎𝑡 𝑥0 . The linear function L(x) called the linearization of f at x0.

Definition 3: Let 𝑓 be a function differentiable at 𝑥𝑜 and 𝑥 be a number in the domain of

f ‘near’ 𝑥𝑜 .The number
a) 𝑑𝑥: = 𝑥 − 𝑥0 is called a differential of x at 𝑥0 . It also denoted by ∆𝑥.
b) ∆𝑦 = 𝑓(𝑥) − 𝑓(𝑥0 ) is called an actual change in 𝑓 at 𝑥0

57
 58

c) 𝑑𝑦 ≔ 𝑓 ′ (𝑥𝑜 )(𝑥 − 𝑥𝑜 ) is called a differential of f or a differential of y at 𝑥0 . It is also denoted by

𝑑𝑓.

𝑑𝑦
It turns out that 𝑑𝑦 ≔ 𝑓 ′ (𝑥𝑜 )𝑑𝑥. This justifies our usage of the notations 𝑓 ′ (𝑥𝑜 ) and 𝑑𝑥 as

synonym. Note that 𝑑𝑥 is an independent variable and ∆𝐿 = 𝑑𝑦. The differential 𝑑𝑦 is the same as
the actual change in the value of 𝐿 as 𝑥 changes from x0 to 𝑥. The differential 𝑑𝑦 is a dependent
variable, depending on the independent variable 𝑑𝑥. See Figure.
Moreover,
𝑓(𝑥) ≈ 𝑓(𝑥𝑜 ) + 𝑓 ′ (𝑥𝑜 )(𝑥 − 𝑥𝑜 ) ⟺ 𝑓(𝑥) − 𝑓(𝑥𝑜 ) ≈ 𝑓 ′(𝑥𝑜 )(𝑥−𝑥𝑜 )
⟺ 𝑓(𝑥) − 𝑓(𝑥𝑜 ) ≈ 𝑓′(𝑥𝑜 )(𝑥 − 𝑥𝑜 )
⟺ ∆𝑦 ≈ 𝑑𝑦
Example 15: Compute the differential for 𝑦 = 𝑥 3 − 4𝑥 2 + 7𝑥.

Solution: Let 𝑓(𝑥) = 𝑥 3 − 4𝑥 2 + 7𝑥. Then 𝑑𝑦 = 𝑓 ′ (𝑥)𝑑𝑥 = 3𝑥 2 − 8𝑥 + 7 //

Example 16: Use differentials (linear approximation) to estimate the value √65
Solution: Let 𝑓(𝑥) = √𝑥 and 𝑥0 = 64.
1 1 129
Then √65 = 𝑓(65) ≈ 𝑓(64) + 𝑓 ′ (64)(65 − 64) = 8 + 2 (1) = 8 + 16 = //
√64 16

Example 17: A sphere was measured and its radius was found to be 45 inches with a possible error
of no more that 0.01 inches. What is the maximum possible error in the volume if we use this value
of the radius?

4
Solution: The volume V of the sphere with radius r is 𝑉 = 3 𝜋𝑟 3 .

∆𝑉 ≈ 𝑑𝑉 = 𝑉′(45)(±0.01) = ±254.47𝑖𝑛3 //

Newton's Method

Many problems in mathematics, science, engineering, and business eventually come down to finding
the roots of a nonlinear equation. It is a pity that many mathematical equations cannot be solved
analytically. You already know about the formula for solving quadratic polynomial equations. You
might not know, however, that there are formulas for solving cubic and quartic polynomial
equations. Unfortunately, these formulas are so cumbersome that they are hardly ever used. Even
more unfortunately, it has been proven that no formula can exist for finding roots of quintic or
higher polynomials. Furthermore, if your equations involve trig functions, then it is even easier to

58
 59

find equations that do not have analytical solutions. For example, the following simple equation
cannot be solved to give a formula for x.

𝑥
sin 𝑥 =
2

The need to solve nonlinear equations that cannot be solved analytically has led to the development
of numerical methods. One of the most commonly used numerical methods is called Newton's
method or the Newton-Raphson method. The idea of Newton's method is as follows. Suppose you
have a nonlinear equation of the form
𝑓(𝑥) = 0,
where f is a differentiable function. Then the idea of Newton's method is to start with an initial
guess x0 for the root r of f and to use the tangent line to f at 𝑥 = 𝑥0 to approximate r. If the tangent
line is a good approximation to 𝑓, then the x intercept of the tangent line should be a good
approximation to the root r. Call this value of x where the tangent line intersects the 𝑥 −axis 𝑥1 . We
can solve the equation above to get the following formula.
𝑓(𝑥0 )
𝑥1 = 𝑥0 −
𝑓 ′ (𝑥0 )
In practice, unless the starting point 𝑥0 is very close to the root, the value 𝑥1 is not close enough to
the root we are seeking, so Newton's method is applied again using the tangent line at 𝑥 = 𝑥1 . This
process can be repeated, leading to a sequence of values where the value of 𝑥𝑛+1 is determined from
the equation
𝑓(𝑥𝑛 )
𝑥𝑛+1 = 𝑥𝑛 −
𝑓 ′ (𝑥𝑛 )

Newton-Raphson Algorithm:
1. Start with an estimate (i.e., a guess) of r. Let’s call that guess r0.
2. Create the recursive formula:
𝑓(𝑥𝑛 )
𝑥𝑛+1 = 𝑟𝑛 −
𝑓′(𝑥𝑛 )
3. Use the formula repeatedly, to generate the approximate roots 𝑟1 , 𝑟2 , … until you get the
approximate solution you need (or you come to the conclusion that this method is not
working).

59
 60

Newton’s method can be used to compute square roots as follows. Let 𝑥 = √𝑎 for a > 0, i.e. to
This is equivalent to 𝑥 2 − 𝑎 = 0, 𝑎 > 0. Let 𝑓(𝑥) = 𝑥 2 − 𝑎 .The algorithm starts with some guess
𝑥1 > 0 and computes the sequence of improved guesses 𝑥𝑛 using the recursive formula
𝑥𝑛 2 −𝑎 1 𝑎
𝑥𝑛+1 = 𝑥𝑛 − or 𝑥𝑛+1 = 2 [𝑥𝑛 + 𝑥 ]
2𝑥𝑛 𝑛

Example 18: Perform 3 iterations of Newton’s method to approximate √7. Compare your result
with a calculator approximation of √7. Start with the initial guess x=2.
1 7 1 7 11
Solution: 𝑋1 = 2 [𝑥0 + 𝑥 ] = 2 [2 + 2] = 4
= 2.75
0

1 7 1 7 7 23
𝑋2 = [𝑥1 + ] = [ + 7 ]= = 2.875
2 𝑥 2 4 1 ( ) 8
4

1 7 1 23 7
𝑋3 = 2 [𝑥2 + 𝑥 ] = 2 [ 8 + 23 ] = 2.65489
2 ( )
8

A calculator result is 2.645751311064591.

Note: Online equation solvers are also available.

Exercises 4.7
1. Sketch the graph of 𝑦 = 𝑥 3 (𝑥 − 7)4 .
2. Sketch the graph of 𝑦 = 𝑥 4 − 4𝑥 3 + 4𝑥 2 + 1.
3. Consider the function 𝑓(𝑥) = sin2 𝑥 + 𝑠𝑖𝑛𝑥 on the interval (0, 2π). Find the open intervals
on which the function is increasing or decreasing and locate all relative extrema. Graph the
function to confirm your results.
4. Use symmetry, extrema, and zeros to sketch the graph of f . How do the functions f and g
differ? Explain.

5. Sketch the graph of the following function. Use the MAPLE, MATLAB or
MATHEMATICA graphing utilities to verify your results.

6. Find the point (x, y) on the graph of 𝑦 = √𝑥 nearest the point (4, 0).
7. Find the point(s) on the graph of 𝑥 = 3 − 2𝑦 2 that are closest to (−4,0).

8. Find the point on the graph of the line y2 = 4x which is nearest to the point (2, 1).

60
 61

9. Find the dimension of the right circular cylinder of the largest volume that can be inscribed in
a sphere of radius 10 units.
10. A cylindrical can without a top is made to contain V cm3 of liquid. Find the dimensions that
will minimize the cost of metal to make the can.
A sheet of cardboard 3 ft. by 4 ft. will be made into a box by cutting equal-sized squares
from each corner and folding up the four edges. What will be the dimensions of the box
with largest volume?
11. A coffee filter has the shape of an inverted cone. Water drains out of the filter at a rate
10𝑐𝑚3 /𝑚𝑖𝑛. When the depth of water in the cone is 8𝑐𝑚, the depth is decreasing at
2cm/min. What is the ratio of the height of the cone to its radius?
12. A tank of water in the shape of a cone is leaking water at a constant rate of .
The base radius of the tank is 5 ft and the height of the tank is 14 ft.
(a) At what rate is the depth of the water in the tank changing when the depth of the water
is 6 ft?
(b) At what rate is the radius of the top of the water in the tank changing when the depth
of the water is 6 ft?
sinx
13. Evaluate limx→π− 1−cosx with and without using L’hopital’s rule. What did you observe?
14. Use L’Hopital’s rule to evaluate

𝑡𝑎𝑛−1 𝑥−𝑥 1−cos 𝑥

a) 𝑙𝑖𝑚𝑥→0+ d) 𝑙𝑖𝑚𝑥→0
8𝑥 3 𝑥2
𝑥 2𝑥+1
b) 𝑙𝑖𝑚𝑥→∞ 𝑥 2 −1 e) 𝑙𝑖𝑚𝑥→−∞ 2
√𝑥 −𝑥
c) lim𝑥→0+ 𝑥 𝑥 f) lim𝑥→∞ 𝑥𝑒 −𝑥

15. Use Newton's method to find all the roots of the equation √𝑥 2 − 𝑥 + 1 = 2𝑠𝑖𝑛𝜋𝑥 correct
to eight decimal places. Start by drawing a graph to find initial approximations.
16. Find all solutions of 𝑒 2𝑥 = 𝑥 + 6, correct to 4 decimal places; use the Newton Method.

17. Explain why Newton's method fails when applied to the equation with any initial
approximation 𝑥0 ≠ 0 . Illustrate with a figure.
18. Find the differential dy of the function.

19. Approximate √17 by differentials

20. The range R of a projectile is

where vo is the initial velocity in feet per second and θ is the angle of elevation. If vo = 2200
feet per second and θ is changed from 10 to 11 , use differentials to approximate the
change in the range.

61