4.

Applications of Diferentiation

4.1 Using differentials and derivatives

Suppose dx is regarded as a new independent variable called the differential of x we can define a
new dependent variable dy, called the differentail of y as a function of x and dx by

dy
dy “ dx “ f 1 pxq dx.
dx

For example 2
` if y2 ˘“ x , then dy “ 2xdx means the same thing as dy{dx “ 2x. If f pxq “ 1{x, then
d f pxq “ ´ 1{x dx.
If y is a function of x, y “ f pxq, then denoting a small change in x by dx instead of ∆x, the
corresponding small change in y, ∆y is approximated by the differential dy, i.e.

∆y « dy “ f 1 pxq dx.

 Example 4.1.1 Without using a scientific calculator, determine by a pproximately how much the

value of sin x increases as x increases from π{3 to pπ{3q ` 0.006. To 3 decimal places, what is the
value of sin ppπ{3q ` 0.006q? 

dy
Solution If y “ sin x, then dx “ cos x. Now x “ π3 « 1.0472 and dx “ 0.006. Therefore

dy ´π ¯ 1
dy “ dx “ cos xdx “ cos ¨ 0.006 “ p0.006q “ 0.003.
dx 3 2
This means that the change in the value of sin x is approximately 0.003. Now

´´ π ¯ ¯ ´π ¯
sin ` 0.006 « sin ` 0.003 “ ¨ ¨ ¨ “ 0.869p3 d. p.q.
3 3
72 Chapter 4. Applications of Diferentiation
y

graph of f

∆y
dy

dx “ ∆x

y “ f pxq

x x ` dx x

Figure 4.1: dy, the change in height to tangent line, approximates ∆y, the change in height to the graph f

Suppose changes in x are measured with respect to the size of x, then

dx dx
relative change in x “ and percentage change in x “ 100 .
x x
 Example 4.1.2 By approximately what percentage does the area of a circle increases if the radius
increases by 2%? 

Solution A “ πr2 implying that dA

dr “ 2πr. Now ∆A « dA “ 2πrdr and relative change in A is

∆A dA 2πrdr dr
« “ 2
“2 .
A A πr r
2
If r increases by 2% then dr “ 100 r, so

dA 1 2 4
“ 2¨ ¨ ¨r “
A r 100 100
so that the area increases by 4%.

Average and Instantaneous Rates of Change

Definition 4.1.1 The average rate of change of a function f pxq with respect to x over the
interval from a to a ` h is
f pa ` hq ´ f paq
.
h
4.1 Using differentials and derivatives 73

The (instantaneous) rate of change of f pxq with respect to x at x “ a is the derivative

f pa ` hq ´ f paq
f 1 paq “ lim ,
hÑ0 h
provided the limit exists.
 Example 4.1.3 How fast is the area A of a circle increasing with respect to its radius when the
radius is 5 m? 

Solution Rate of change of area with respect to radius is

dA d
“ pπr2 q “ 2πr.
dr dr

When r “ 5 m, the area is changing at the rate 2π ˆ 5 “ 10π m2 {m. This means that a small change
∆r m in the radius when the radius is 5 m would result in a change of about 10π ∆r m2 in the area of
the circle.
 Example 4.1.4 Suppose the temperature at a certain location t hours after noon on a certain day
is T ˝ C, where

1
T “ t 3 ´ 3t 2 ` 8t ` 10 pfor 0 ď t ď 5q.
3

How fast is the temperature rising or falling at 1 : 00 pm? at 3 : 00 pm? At what instants is the
temperature stationary? 

Solution Rate of change of temperature is given by

dT
“ t 2 ´ 6t ` 8 “ pt ´ 2qpt ´ 4q.
dt

dT
If t “ 1, then “ 3, so the temperature is rising at rate 3˝C{h at 1 : 00 pm.
dt
dT
If t “ 3, then “ ´1, so the temperature is falling at rate 1˝C{h at 3 : 00 pm.
dt
dT
The temperature is stationary when “ 0, that is, at 2 : 00 pm and 4 : 00 pm.
dt

Sensitivity to Change
When a small change in x produces a large change in the value of a function f pxq, wa say that the
function is very sensitive to changes in x. Therefore the derivative is f 1 pxq is a measure of the
sensitivity of the dependence of f on x.
 Example 4.1.5 A pharmacologist studying a drug that has been developed to lower blood
pressure determines experimentally that the average reduction R in blood pressure resulting from a
daily dosage of x mg of the drug is
ˆ ˙
x ´ 13
R “ 24.2 1 ` ? mm Hg.
x2 ´ 26x ` 529

Determine the sensitivity of R to dosage x at dosage level of 5 mg, 15 mg and 35 mg. AT which of
these dosage levels would an increase in the dosage level have the greatest effect? 
74 Chapter 4. Applications of Diferentiation

Solution The sensitivity of R to x is

¨? ˛
2
x ´ 13
˚ x ´ 26x ` 529p1q ´ px ´ 13q x2 ´ 26x ` 529 ‹
?
dR
“ 24.4 ˝
˚ ‹
dx x2 ´ 26x ` 529 ‚

x2 ´ 26x ` 529 ´ px2 ´ 26x ` 169q

ˆ ˙
“ 24.2
px2 ´ 26x ` 529q3{2

8, 712
“
px2 ´ 26x ` 529q3{2
At dosages x “ 5 mg, 15 mg and 35 mg, we have sensitivities of
ˇ ˇ
dR ˇˇ dR ˇˇ
“ 0.998 mm Hg{mg, “ 1.254 mm Hg{mg,
dx ˇx“5 dx ˇx“15
ˇ
dR ˇˇ
“ 0.355 mm Hg{mg.
dx ˇx“35

Among these levels, the greatest sensitivity is at 15 mg. Increasing the dosage from 15 to 16 mg{day
could be expected to further reduce average blood preasureby about 1.25 mm Hg.

Derivative in Economics
Just as physicists use terms such as velocity and acceleration to refer to derivatives of certain
quantities, economists also have their own specialised vocabulary for derivatives. They call them
marginals. For example the cost of production Cpxq in a manufacturing operation is a function of
x, the number of units produced.. The marginal cost of production is the rate of change of C with
respect to x, so it is dC{dx.
 Example 4.1.6 — Marginal cost of production. The cost of producing x tons of coal per day
in a mine is $Cpxq, where
Cpxq “ 4, 200 ` 5.4x ´ 0.001x2 ` 0.000002x3 .
1. What is the average cost of producing each ton if the daily production level is 1, 000 tons?,
2, 000 tons?
2. Find the marginal cost of production if the daily production level is 1, 000 tons. 2, 000 tons.
3. If the production level increases slightly from 1, 000 tons or from 2, 000 tons, what will
happen to the average cost per ton?


Solution 1. The average cost per ton of the coal is

Cpxq 4, 200
“ ` 5.4 ´ 0.001x ` 0.000002x2 .
x x
If x “ 1, 000, the average cost per ton is Cp1, 000q{1, 000 “ $10.6{ton. If x “ 2, 000, the
average cost per ton is Cp2, 000q{2, 000 “ $13.5{ton.
2. The marginal cost of production is
C1 pxq “ 5.4 ´ 0.002x ` 0.000006x2 .
If x “ 1, 000, the marginal cost is C1 p1, 000q “ $9.4{ton. If x “ 2, 000, the marginal cost is
C1 p2, 000q “ $25.4{ton.
4.1 Using differentials and derivatives 75

3. If the production level is increased slightly from x “ 1, 000, the average cost per ton will drop
because the cost increasing at a rate lower than the average cost. At x “ 2, 000 the opposite
is true; an increase in production will increase the average cost per ton.
The equivalent of sensitivity in economics is elasticity. Elasticity is the measure of how an
economic variable responds to change in another variable. An elastic variable is one which responds
more than proportionally to changes in other variables. An inelastic variable is one which changes
less than proportionally in response to changes in other variables.

Differentials and point elasticity

For a demand function Q “ f pPq, elasticity is defined as

∆Q{Q relative change in Q

“ .
∆P{P relative change in P

If the change in P is infinitesimal, then the expressions ∆Q and ∆P reduce to the differentials dP
and dQ. In that case the elasticity measure assumes the sense of point elasticity of the demand
function which is denoted by εd . Thus

dQ{Q dQ{dP
εd ” “ .
dP{P Q{P

The numerator in the right hand is the derivative function of the demand function while the
denominator is the average function of the demand function. Thus the point elasticity is a ratio of
the two functions. In general, for any given total function y “ f pxq, the point elasticity of y w.r.t. x
is
dy{dx marginal function
εyx “ “ .
y{x average function

The absolute value of the point elasticity measure is used in deciding whether the function is
elastic at a particular point. In the case of a demand function,
$ $
&elastic
’ &|εd | ą 1
’
The demand is of unit elasticity if |εd | “ 1
’
% ’
inelastic
%
|εd | ă 1.

at a given point.

 Example 4.1.7 Find εd for the demand function Q “ 100 ´ 2P. Determine the point elasticity at
P “ 25. 

dQ Q 100´2P
Solution dP “ ´2 and P “ P . Therefore

´2 P
εd “ “ .
p100 ´ Pq {P P ´ 50

Thus
ˇ
ˇ
εd ˇˇ “ ¨ ¨ ¨ “ ´1.
P“25

Therefore the demand is of unit elasticity when P “ 25.

 76 Chapter 4. Applications of Diferentiation

Exercise 4.1
1. Use differentials to determine approximate change in the values of the given function as
its argument changes from the given value to the given amount. What is the approximate
value of the function after the change?
(a) y “ 1{x as x increases from 2 to 2.01.
(b) h ptq “ cos
? pπt{4q as t increases from 2 to 2 ` p1{10πq.
(c) f pxq “ 3x ` 1 as x increases from 1 to 1.08.
(d) u “ tan ps{4q as s decreases from π to π ´ 0.04.
2. Find the approximate percentage changes in the given function that will result from an
increase of 2% in the value of x.

(a) y “ x2 (c) y “ 1{x

?
(b) y “ 1{x2 (d) y “ x

3. Find the rate of change of the area of a square with respect to the length of its side when
the side is 4 ft.
4. The sales of revenue $R from a software product depends on the price $p charged by the
distributor according to the formula

R “ 4, 000p ´ 10p2 .

(a) How sensitive is R to p when p “ $100? p “ $200? p “ $300?

(b) Which of these three is most reasonable price for the distributor to charge? Why?
5. Given the consumption function C “ a ` bY (with a ą 0; 0 ă b ă 1);
(a) Find its marginal function and its average function.
(b) Find the income elasticity of consumption εCY , and determine its sign, assuming
Y ą 0.
(c) Show that the consumption is inelastic at all positive income levels.
6. Assume that the quantity demand for a particular commodity is given by the formula

D ppq “ 8000p´1.5 .

Compute the elasticity of D ppq and find the percentage change in quantity demand when
the price increases by 1% from p “ 4.


4.2 Mean value theorems of differential calculus

Theorem 4.2.1 — Mean Value Theorem. Suppose that the function f is continuous on the
closed finite interval ra, bs and that it is differentiable on the interval pa, bq. Then D a point
c P pa, bq such that

f pbq ´ f paq
“ f 1 pcq .
b´a

It means that the slope of the chord joining the points pa, f paqq and pb, f pbqq is equal to the slope
of the tangent line to te curve y “ f pxq at the point pc, f pcqq so that the two lines are parallel.
4.2 Mean value theorems of differential calculus 77
y

C
y “ f pxq

B pb, f pbqq

A pa, f paqq

a c b x

Figure 4.2: There is a point C on the curve where the tangent is parallel to the chord AB

?
 Example 4.2.1 Verify the conclusion of the mean value theorem for f pxq “ x on the interval
ra, bs, where a ď x ď b. 

Solution We are to show that D c P pa, bq such that

f pbq ´ f paq
“ f 1 pcq
b´a
1 ?
so long as f is continuous on ra, bs and is differentiable on pa, bq. Now f 1 pxq “ ?
2 x
, f paq “ a,
?
f pbq “ b.
? ? ? ?
1 b´ a b´ a 1
6 ? “ “ `? ? ˘ `? ? ˘“? ? .
2 c b´a b´ a b` a b` a
? ? ´? ? ¯2
? b` a b` a
The above equality implies that c“ 2 so that c “ 2 . Since a ă b, we have

ˆ? ? ˙ ˜? ? ¸2 ˜ ? ? ¸2
a` a 2 b` a b` b
a“ ă ă “b
2 2 2

which implies that c P pa, bq.

 Example 4.2.2 Show that sin x ă x for all x ą 0. 

Solution If x ą 2π, then sin x ď 1 ă 2π ă x. If 0 ă x ď 2π, then by MVT, D c P p0, 2πq such that

ˇ
sin x sin x ´ sin 0 d ˇ
“ “ rMVT on r0, xss “ sin xˇˇ “ cos c ă 1
x x´0 dx x“c

which implies that sin x ă x in this case too.

Values of x where f 1 pxq “ 0 are called critical points of the function f .
 78 Chapter 4. Applications of Diferentiation

Exercise 4.2
1. Illustrate the MVT by finding any points in the open interval pa, bq where the tangent line
is parallel to the chord joining pa, f paqq and pb, f pbqq.
(a) f pxq “ x2 on ra, bs; Ans: c “ b`a2 .
(b) f pxq “ x3 ´ 3x ` 1 on r´2, 2s; Ans: c “ ˘ ?23 .
2. Show that tan x ą x for 0 ă x ă π2 .
3. Let r ą 1. If x ą 0 or ´1 ď x ă 0, show that p1 ` xqr ă 1 ` rx.


4.3 Related Rates

When two or more quantities that change with time are linked by an equation, that equation can
be differentiated with respect to time to produce an equation linking the rates of change of the
quantities. Below we give the procedure for dealing with such problems
1. From the given data make a sketch if possible.
2. Define symbols you want to use that are not defined in the statement of the problem. Express
given and required quantities and rates in terms of these symbols.
3. Identify one or more equations linking the variable quantities.
4. Differentiate the equation(s) implicitly with respect to time, regarding variable quantities as
functions of time.
5. Substitute any given values for the quantities and their rates, then solve the resulting equa-
tion(s) for the unknown quantities and rates.
6. Make a concluding statement answering the question asked.

Example 4.3.1 An aircraft is flying horizontally at a speed of 600 km{h. How fast is the distance
between the aircraft and a radio beacon increasing 1 minute after the aircraft passes 5 km directly
above the beacon? 

Solution The diagram fitting the given data is

600 km{h
C x
A

5 km s

Figure 4.3:

Let C be the point on the aircraft’s path directly above the beacon B. Let A be the position of
the aircraft t min after it is at C, and let x and s be the distances CA and BA, respectively. From the
right triangle BCA we have

s2 “ x 2 ` 52 .
4.3 Related Rates 79

We differentiate this equation implicitly with respect to t to obtain

ds dx
2s “ 2x .
dt dt

We are given
? that dx{dt ?
“ 600 km{h “ 10 km{mim. Therefore, x “ 10 km at time t “ 1 min. At that
time s “ 102 ` 52 “ 5 5 and is increasing at the rate

ds x dx 10 1, 200
“ “ ? p600q “ ? « 536.7 km{h.
dt s dt 5 5 5

One minute after the aircraft passes over the beacon, its distance from the beacon is increasing at
about 537 km{h.
 Example 4.3.2 How fast is the area of a rectangle changing if one side is 10 cm long and is
increasing at a rate of 2 cm{s and the other side is 8 cm long and is decreasing at a rate of 3 cm{s? 
Solution

A “ xy y

Figure 4.4:

Let the lengths of the sides of the rectangle at time t be x cm and y cm respectively. Thus the
area at time t is A “ xy cm2 . Also dx{dt “ 2 and dy{dt “ ´3. We are to determine the value of
dA{dt when x “ 10 and y “ 8. Since x and y are functions of time we differentiate A implicitly
with respect to time as follows.
ˇ ˆ ˙ˇ
dA ˇˇ dx dy ˇˇ
“ y`x “ 2p8q ` 10p´3q “ ´14.
dt ˇ x “ 10 dt dt ˇ x “ 10
y“8 y“8

At the indicated time, the area of the rectangle is decreasing at a rate of 14 cm2 /s.

Exercise 4.3
1. Find the rate of change of the area of a square whose side is 8 cm long, if the side length
is increasing at 2 cm{min.
2. A pebble dropped into a pond causes a circular ripple to expand outward from the point
of impact. How fast is the area enclosed by the ripple increasing when the radius is 20 cm
and is increasing at a rate of 4 cm{s.

 80 Chapter 4. Applications of Diferentiation

4.4 Indeterminate Forms

In Section 2.4 we showed that

sin x
lim “ 1.
xÑ0 x

Direct evaluation would have yielded the indeterminate form r0{0s. Below we list the various
kinds of indeterminate forms and give an example in each case.

Table 4.1: Types of indeterminate forms

Type Example
sin x
r0{0s lim
xÑ0 x

lnp1{x2 q
r8{8s lim
xÑ0 cotpx2 q

1
r0 ¨ 8s lim x ln
xÑ0` x
ˆ ˙
1
r8 ´ 8s lim tan x ´
xÑpπ{2q´ π ´ 2x
r00 s lim xx
xÑ0`

r80 s lim ptan xqcos x

xÑpπ{2q´
ˆ ˙x
1
r18 s lim 1`
xÑ8 x

Most indeterminate forms of type r0{0s can be evaluated by canceling common factors as was
discussed in Section 1.2. In this section we discuss further approaches that can be used to solve
such indeterminate forms.
Theorem 4.4.1 — The First l’Hôpital’s Rule. Suppose the functions f and g are differentiable
on the interval pa, bq, and g1 pxq ‰ 0 there. Suppose also that
1. lim f pxq “ lim gpxq “ 0 and
xÑa` xÑa`
f 1 pxq
2. lim 1 “ L where L is finite or 8 or ´8.
xÑa` g pxq
f pxq
Then lim “ L.
xÑa` gpxq

Similar results hold if every occurrence of limxÑa` is replaced by limxÑb´ or even limxÑc
where a ă c ă b. The cases a “ ´8 and b “ 8 are also allowed.

Proof. 

ln x
 Example 4.4.1 Evaluarte lim 2 . 
xÑ1 x ´ 1
4.4 Indeterminate Forms 81

Solution We have
„ 
ln x 0
lim
xÑ1 x2 ´ 1 0

1{x 1 1
“ lim “ lim 2 “
xÑ1 2x xÑ1 2x 2

2 sin x ´ sinp2xq
 Example 4.4.2 Evaluarte lim . 
xÑ0 2ex ´ 2 ´ 2x ´ x2
Solution We have
„ 
2 sin x ´ sinp2xq 0
lim
xÑ0 2ex ´ 2 ´ 2x ´ x2 0

2 cos x ´ 2 cosp2xq
“ lim
xÑ0 2ex ´ 2 ´ 2x
„ 
cos x ´ cosp2xq 0
“ lim still
xÑ0 ex ´ 1 ´ x 0
„ 
´ sin x ` 2 sinp2xq 0
“ lim x
still
xÑ0 e ´1 0
´ cos x ` 4 cosp2xq ´1 ` 4
“ lim “ “3
xÑ0 ex 1

Theorem 4.4.2 — The Second l’Hôpital’s Rule. Suppose the functions f and g are differentiable
on the interval pa, bq and that g1 pxq ‰ 0 there. Suppose also that
1. lim gpxq “ ˘8 and
xÑa`
f 1 pxq
2. lim “ L where L is finite or 8 or ´8.
xÑa` g1 pxq
f pxq
Then lim “ L.
xÑa` gpxq

Again, similar results hold if every occurrence of limxÑa` is replaced by limxÑb´ or even
limxÑc where a ă c ă b. The cases a “ ´8 and b “ 8 are also allowed.
x2
 Example 4.4.3 Evaluarte (a) lim x and (b) lim xa ln x, where a ą 0. 
xÑ8 e xÑ0`

Solution
1. We have that

x2 ” 8 ı
lim
xÑ8 ex 8

2x ”8ı
“ lim still
xÑ8 ex 8
2
“ lim “ 0.
xÑ0 ex
 82 Chapter 4. Applications of Diferentiation

2. Here we have
lim xa ln x r0 ¨ p´8qs
xÑ0`
„ 
ln x ´8
“ lim ´a
xÑ0` x 8
1{x xa
“ lim “ lim “ 0.
xÑ0` ´ax´a´1 xÑ0` ´a

Exercise 4.4
Evaluate the following limits

3x sin ax
1. lim 3. lim
xÑ0 tan 4x xÑ0 sin bx

ln p2x ´ 3q 1 ´ cos ax
2. lim 4. lim
xÑ2 x2 ´ 4 xÑ0 1 ´ cos bx

4.5 Extreme Values

Maximum and Minimum Values
Definition 4.5.1 Function f has an absolute maximum value f px0 q at x0 in its domain if
f pxq ď f px0 q holds for every x in the domain of f .
Similarly, f has an absolute minimum value f px0 q at x0 in its domain if f pxq ě f px0 q holds
for every x in the domain of f .

R
1. A function will have only one absolute maximum (or minimum) value if it exists.
However the value can occur at many points. For example, f pxq “ sin x has absolute
maximum of 1 but it occurs at every point π2 ` 2nπ, n P Z.
2. A function need not have any extreme value. The function f pxq “ 1x becomes arbitrarily
large as x approaches 0 from the right, and so has no finite absolute maximum value.

Theorem 4.5.1 If the domain of the function f is a closed, finite interval or a union of finitely
many such intervals, and if f is continuous on that domain, then f must have an absolute
maximum value and absolute minimum value.

Definition 4.5.2 Function f has a local maximum value (loc. max.) f px0 q at the point x0 in its
domain provided D a number h ą 0 such that f pxq ď f px0 q whenever x is in the domain of f
and |x ´ x0 | ă h.
Similarly, f has a local minimum value (loc. min.) f px1 q at the point x1 in its domain
provided D a number h ą 0 such that f pxq ě f px1 q whenever x is in the domain of f and
|x ´ x1 | ă h.
4.5 Extreme Values 83
y
y “ f pxq

a x1 x2 x3 x4 x5 x6 b x

From the above figure we see that local extreme values can occur at any of the following points.
(i) critical points of f ; points x P D p f q where f 1 pxq “ 0.
(ii) singular points of f ; points x P D p f q where f 1 pxq is not defined.
(iii) endpoints of the domain of f ; points that do belong to D p f q but are not interior points of
D p f q.
In the figure above, x1 , x3 , x4 are critical points, x2 and x5 are singular points and a and b are
endpoints.

Theorem 4.5.2 If the function f is defined on an interval I and has a local maximum (or local
minimum) value at the point x “ x0 in I, then x0 must be either a critical point of f , a singular
point of f , or and endpoint of I.

 Example 4.5.1 Find the maximum and minimum values of the function g pxq “ x3 ´ 3x2 ´ 9x ` 2
on the interval ´2 ď x ď 2. 

Solution Since g is a polynomial it can never have a singular point. For critical points we calculate

g1 pxq “ 3x2 ´ 6x ´ 9 “ 3 px ` 1q px ´ 3q “ 0.

Thus x “ ´1 or x “ 3. But x “ 3 is not in the domain of g and so we ignore it. We then investigate
the endpoints x “ ´2 and x “ 2 and critical point x “ ´1.
g p´2q “ 0, g p´1q “ 7, g p2q “ ´20
The maximum value of g on ´2 ď x ď 2 is at the critical point x “ ´1, and the minimum value
is at the endpoint x “ 2.

y
p´1, 7q

y “ g pxq
“ x3 ´ 3x2 ´ 9x ` 2
p´2, 0q x

p2, ´20q

Figure 4.5: g has maximum and minimum values 7 and ´20, respectively
84 Chapter 4. Applications of Diferentiation

 Example 4.5.2 Find the maximum and minimum values of h pxq “ 3x2{3 ´ 2x on the interval
r´1, 1s. 

Solution The derivative of h is

h1 pxq “ 2x´1{3 ´ 2.

Note that x´1{3 is not defined at x “ 0 in D phq, so x “ 0 is a singular point of h. Also h1 pxq “ 0 at
x´1{3 “ 1, that is at x “ 1, which also happens to be an endpoint of the domain of h. We therefore
examine the values of h at endpoints x “ ´1 and x “ 1 and at the singular point x “ 0.
h p´1q “ 5, h p0q “ 0, h p1q “ 1
The function h has a maximum value 5 at the endpoint x “ ´1 and a minimum value 0 at the
singular point x “ 0.
Diagram Fig 4.20

Theorem 4.5.3 — The first derivative test.

Part I: Testing interior critical points and singular points.
Suppose f is continuous at x0 , and x0 is not an endpoint of the domain of f .
(a) If D an open interval pa, bq containing x0 such that f 1 pxq ą 0 on pa, x0 q and f 1 pxq ă 0
on px0 , bq, then f has a local maximum value at x0 .
(b) If D an open interval pa, bq containing x0 such that f 1 pxq ă 0 on pa, x0 q and f 1 pxq ą 0
on px0 , bq, then f has a local minimum value at x0 .
Part II: Testing endpoints of the domain.
Suppose a is a left endpoint of the domain of f and f is right continuous at a.
(c) If f 1 pxq ą 0 on some interval pa, bq , then f has a local minimum value at a.
4.5 Extreme Values 85

(d) If f 1 pxq ă 0 on some interval pa, bq , then f has a local maximum value at a.
Suppose b is a right endpoint of the domain of f and f is left continuous at b.
(e) If f 1 pxq ą 0 on some interval pa, bq , then f has a local maximum value at b.
(f) If f 1 pxq ă 0 on some interval pa, bq , then f has a local minimum value at b.

R If f 1 is positive (or negative) on both sides of a critical or singular point, then f has neither a
maximum nor a minimum value at that point.

 Example 4.5.3 Find the local and absolute extreme values of f pxq “ x4 ´ 2x2 ´ 3 on the interval
r´2, 2s. Sketch the graph of f . 
` ˘
Solution f 1 pxq “ 4x3 ´ 4x “ 4x x2 ´ 1 “ 4x px ` 1q px ´ 1q.
The critical points are x “ ´1, x “ 0, x “ 1. The corresponding values are f p´1q “ ´4,
f p0q “ ´3, f p1q “ ´4. There are no singular points. The values of f at the endpoints ´2 and
2 are f p´2q “ ´5 and f p2q “ 5.We summarize the positive/negative properties of f pxq and the
implied increasing/decreasing behaviour of f pxq in a chart form below;
EP CP CP CP EP
x ´2 ´1 0 1 2
f1 ´ 0 ` 0 ´ 0 `
f max Œ min Õ max Œ min Õ max

Figure 4.6: The graph of y “ x4 ´ 2x2 ´ 3

Theorem 4.5.4 — Existence of extreme values on open intervals.

If f is continuous on an open interval pa, bq, and if

lim f pxq “ L and lim f pxq “ M

xÑa` xÑb´
86 Chapter 4. Applications of Diferentiation

then the following conclusions hold

(i) If f puq ą L and f puq ą M for some u P pa, bq, then f has an absolute maximum value on
pa, bq.
(ii) If f puq ă L and f puq ă M for some u P pa, bq, then f has an absolute minimum value on
pa, bq.

In this theorem a may be ´8 and b may 8. Also, either or both L and M may be 8 or ´8.

 Example 4.5.4 Show that f pxq “ x ` 4x has an absolute minimum value on the interval p0, 8q,
and find that minimum value. 

Solution We have

lim f pxq “ 8 and lim f pxq “ 8.

xÑ0` xÑ8

Since f p1q “ 5 ă 8, Theorem 4.5.4 guarantees that f must have an absolute minimum value at
some point in p0, 8q. Now

4 x2 ´ 4 px ´ 2q px ` 2q
f 1 pxq “ 1 ´ “ “
x2 x2 x2

which equals 0 when x “ ˘2. Since D p f q “ p0, 8q it has no singular points but has only one
critical point at x “ 2. Here f p2q “ 4. This must be the minimum value of f on p0, 8q.

y “ x ` 4x

p2, 4q
x

Figure 4.7: f has minimum value 4 at x “ 2

Concavity and Inflections

Definition 4.5.3 We say that a function f is concave up on an open interval I if it is differentiable
there and and if f 1 is an increasing function on I. Similarly, f is concave down on I if f 1 exists
and is a decreasing function on I.

Definition 4.5.4 We say that the point px0 , f px0 qq is an inflection point of the curve y “ f pxq if
the following two conditions hold
(a) the graph of y “ f pxq has a tangent line at x “ x0 , and
(b) the concavity of f is opposite on opposite sides of x0 .

Diagram Fig 4.27, 4.28, 4.29

4.5 Extreme Values 87

y
0

` ´

´ `

0
a b c x

Figure 4.8

Theorem 4.5.5 — Concavity and the second derivative.

(a) If f 2 pxq ą 0 on interval I, then f is concave up on I.
(b) If f 2 pxq ă 0 on interval I, then f is concave down on I.
(c) If f has an inflection point at x0 and f 2 px0 q exists, then f 2 px0 q “ 0

Proof.
(a) and (b) follow from applying Theorem 3.1.1 to the derivative f 1 of f .
(c) If f has an inflection point at x0 and if f 2 px0 q exists, then f must be differentiable in an open
interval containing x0 . Since f 1 is increasing on one side of x0 and decreasing on the other
side, it must have a local maximum or minimum value at x0 . Thus f 2 px0 q must be zero.


 Example 4.5.5 Determine the intervals of concavity of f pxq “ x6 ´ 10x4 and the inflection points
of its graph. 

Solution We have

f 1 pxq “ 6x5 ´ 40x3 , f 2 pxq “ 30x4 ´ 120x2 “ 30x2 px ´ 2q px ` 2q

Thus f 2 pxq “ 0 at x “ ˘2, 0. These three values of x induce/generate the following four intervals
on which we test concavity.
Consider the interval p´8, ´2q. Here for x “ ´3,

f 2 p´3q “ 30 p´3q2 p´5q p´1q ą 0

and so f is concave up.

For x “ ´1 P p´2, 0q, we have

f 2 p´1q “ 30 p´1q2 p´3q p1q ă 0

and so f is concave down.

For x “ 1 P p0, 2q, we have

f 2 p1q “ 30 p1q2 p´1q p3q ă 0

and so f is concave down.

88 Chapter 4. Applications of Diferentiation

For x “ 3 P p2, 8q, we have

f 2 p3q “ 30 p3q2 p1q p5q ą 0

and so f is concave up. We summarize this information in the following chart.

x ´2 0 2
f2 + 0 ´ 0 ´ 0 `
f ! infl " " infl !

Figure 4.9: The graph of y “ x4 ´ 2x3 ` 1

 Example 4.5.6 Determine the intervals of increase and decrease, the local extreme values and

the concavity of f pxq “ x4 ´ 2x3 ` 1. Use the information to sketch the graph of f . 

Solution

f 1 pxq “ 4x3 ´ 6x2 “ 2x2 p2x ´ 3q , f 2 pxq “ 12x2 ´ 12x “ 12x px ´ 1q

Thus f 1 pxq “ 0 when x “ 0, 1.5 and f 2 pxq “ 0 when x “ 0, 1. The critical points generates
the intervals p´8, 0q, p0, 1.5q and p1.5, 8q on which we test whether the function is increasing or
decreasing.
For the interval p´8, 0q take x “ ´1 and observe that f 1 p´1q ă 0 implying that the function
is decreasing there. For the interval p0, 1.5q take x “ 1 and observe that f 1 p1q ă 0 implying that
the function is decreasing there. For the interval p1.5, 8q take x “ 2 and observe that f 1 p2q ą 0
implying that the function is increasing there.
On the other hand the values for which f 2 “ 0 generate the following the intervals p´8, 0q,
p0, 1q and p1, 8q on which we test concavity. For the interval p´8, 0q, take x “ ´1 and observe
that

f 2 p´1q “ 12 p´1q p´2q ą 0

so that f is concave up. For the interval p0, 1q, take x “ 0.5 and obtain

f 2 p0.5q “ 12 p0.5q p´0.5q ă 0

so that f is concave up there. For the interval p1, 8q, take x “ 2 and obtain

f 2 p0.5q “ 12 p2q p1q ą 0

4.5 Extreme Values 89

so that f is concave up there.

CP CP
x 0 1 1.5
f1 - 0 ´ ´ 0 `
f2 + 0 ´ 0 ` `
f Œ infl Œ infl Œ min Õ
! " ! !
Diagram Fig 4.32

Theorem 4.5.6 — The second derivative test.

(a) If f 1 px0 q “ 0 and f 2 px0 q ă 0 then f has a local maximum value at x0 .
(b) If f 1 px0 q “ 0 and f 2 px0 q ą 0 then f has a local minimum value at x0 .
(c) If f 1 px0 q “ 0 and f 2 px0 q “ 0 no conclusion can be drawn; f may have a local maximum
or a local minimum or it may have an inflection point instead.

Proof.
(a) Suppose that f 1 px0 q “ 0 and f 2 px0 q ă 0. Since
f 1 px0 ` hq f 1 px0 ` hq ´ f 1 px0 q
lim “ lim “ f 2 px0 q ă 0
hÑ0 h hÑ0 h
it follows that f 1 px0 ` hq ă 0 for sufficiently small positive h and f 1 px0 ` hq ą 0 for suffi-
ciently small negative h. By the first derivative test, f must have a local maximum value at
x0 .
(b) Exercise
(c) No conclusion can be made.


 Example 4.5.7 Find and classify the critical points of f pxq “ x2 e´x . 

Solution

f 1 pxq “ 2xe´x ´ x2 e´x “ xe´x p2 ´ xq “ 0 at x “ 0 or 2.

f 2 pxq “ ¨ ¨ ¨ “ 2 ´ 4x ` x2 e´x .
` ˘

Therefore f 2 p0q “ 2 ą 0 and f 2 p2q “ ´2e´2 ă 0. It follows that f has a local minimum value at
x “ 0 and a local maximum value at x “ 2.
90 Chapter 4. Applications of Diferentiation
y

` f˘ pxq “ x2 e´x
´2
2, 4e

Figure 4.10: The critical points of f pxq “ x2 e´x

Exercise 4.5
1. Determine whether the given function has any local or absolute extreme values, and find
those values if possible

(a) f pxq “ x ` 2 on r´1, 1s. (c) f pxq “ x2 ´ 1 on r´2, 3s.

(b) f pxq “ x ` 2 on r´1, 1q. (d) f pxq “ x3 ` x ´ 42 on ra, bs.

2. Determine the intervals of concavity of the given function, and locate any inflection
points.

? ˘2
(i) f pxq “ x.
`
(iv) f pxq “ 3 ´ x2 .
(ii) f pxq “ x2 ` 2x ` 3. ` ˘3
(v) f pxq “ x2 ´ 4 .
(iii) f pxq “ 10x3 ´ 3x5 .

3. Classify the critical points of the following functions using the second derivative test
whenever possible

(i) f pxq “ x px ´ 2q2 ` 1. (iv) f pxq “ x ln x.

` ˘3
(ii) f pxq “ x3 ` 1x . (v) f pxq “ x2 ´ 4 .
x
(iii) f pxq “ 1`x 2.

4. Measured in milligrams per litre, the concentration of a drug in the bloodstream t hours
after injection is given by the formula
t
c ptq “ .
t2 ` 4
Find the time of maximum concentration.
5. Let y denote the total weekly weight of pigs slaughtered by the butcheries of Chicago
during 1948 (in millions of pounds) and let x be the total weekly work effort (in thousands
 4.6 Curve Sketching 91

of hours). Nicholas estimated the relation

y “ ´2.05 ` 1.06x ´ 0.04x2 .

Determine the value of x that maximises y by studying the sign variation of y1 .

6. A sports club plans to charter a plane. The charge for 60 passengers is $800 each. For
each additional person above 60, all travellers get a discount of $10. The plane can take
at most 80 passengers.
(a) What is the total cost when there are 61, 70, and 80 passengers?
(b) If 60 ` x passengers fly, what is the total cost?
(c) Find the maximum of passengers that maximises the total amount of airfares paid
out by the sports club members.
7. The cost of manufacturing x refrigerators is $C pxq, where C pxq “ 8, 000 ` 400x ´ 0.5x2 .
Find the marginal cost if 100 refrigerators are manufactured.


4.6 Curve Sketching

When sketching the graph y “ f pxq of a function f , we have the following three sources of useful
information;
1. the function f itself:- from this we determine the coordinates of some points on the graph,
the symmetry of the graph and any asymptotes.
2. the first derivative, f 1 :- from this we determine the intervals of increase and decrease and the
location of any local extreme values.
3. the second derivative, f 2 :- from this we determine the the concavity and inflection points,
and sometimes the extreme values.
Other important points besides critical points, singular points and points of inflection are the
intercepts.

Asymptotes
Asymptotes are straight lines to which a curve draws arbitrarily close as it recedes to infinite
distance from the origin.

Definition 4.6.1 The graph of y “ f pxq has a vertical asymptote at x “ a if either

lim f pxq “ ˘8 or lim f pxq “ ˘8 or both.

xÑa´ xÑa`

Definition 4.6.2 The graph of y “ f pxq has a horizontal asymptote at y “ L if either

lim f pxq “ L or lim f pxq “ L or both.

xÑ8 xÑ´8

Definition 4.6.3 The straight line y “ ax ` b (a ‰ 0) is an oblique asymptote of the graph of

y “ f pxq if either

lim p f pxq ´ pax ` bqq “ 0 or lim p f pxq ´ pax ` bqq “ 0 or both.

xÑ8 xÑ´8

Suppose that f pxq “ Pm pxq {Qn pxq, where Pm and Qn are polynomials of degree m and n
respectively. Suppose also that Pm and Qn have no common linear factor. Then
(a) The graph of f has a vertical asymptote at every position x such that Qn pxq “ 0.
92 Chapter 4. Applications of Diferentiation

(b) The graph of f has a two-sided horizontal asymptote y “ 0 if m ă n.

(c) The graph of f has two-sided horizontal asymptote y “ L, (L ‰ 0) if m “ n where L is the
quotient of the coefficients of the highest terms in Pm and Qn .
(d) The graph of f has two-sided oblique asymptotes if m “ n ` 1. This asymptote can be found
by dividing Qn by Pm to obtain a linear quotient, ax ` b, and a remainder R, a polynomial of
degree at most n ´ 1, i.e.

R pxq
f pxq “ ax ` b ` .
Qn pxq

The oblique asymptote is y “ ax ` b.

(e) The graph f has no horizontal or oblique asymptote if m ą n ` 1.

x 3
Example 4.6.1 Find the oblique asymptote of y “ x2 `x`1 . 

Solution In this case m “ 3 “ 2 ` 1 “ n ` 1 and so by long division we obtain

x3 1
2
“ x´1` 2 .
x `x`1 x ` x1
Therefore the oblique asymptote is y “ x ´ 1.

2
Example 4.6.2 Sketch the graph of y “ x `2x`4
2x . 

Solution

x2 ` 2x ` 4 x 2
y“ “ `1` .
2x 2 x

1 2 4
6 y1 pxq “ ´ and y2 pxq “ .
2 x2 x3
From y
Domain: all x ‰ 0.
Vertical asymptote: x “ 0 ` ˘
Oblique asymptote: y “ 2x ` 1 since y ´ 2x ` 1 “ 2x Ñ 0 as x Ñ ˘8
x2 ` 2x ` 4 “ px ` 1q2 ` 3 ě 3 for all x. y is not defined at x “ 0.
From y1
2
Critical points: x2x´4 1
2 “ 0 ñ x “ ˘2 and y is not defined at x “ 0.
From y2
y2 is not defined at x “ 0 and y2 “ 0 nowhere.
Facts about the graph are summarized in the chart below
CP ASY CP
x ´2 0 2
y1 ` 0 ´ undef ´ 0 `
y2 ´ ´ undef ` `
y Õ max Œ undef Œ min Õ
" " ! !
4.6 Curve Sketching 93

x2 `2x`4
y“ 2x

p2, 3q

y “ 2x ` 1

p´2, ´1q x

Figure 4.11:

2
 Example 4.6.3 Sketch the graph of f pxq “ xx2 ´1
´4
. 

Solution

` ˘
1 ´6x 2 6 3x2 ` 4
f pxq “ ¨ ¨ ¨ “ and f pxq “ ¨ ¨ ¨ “ .
px2 ´ 4q2 px2 ´ 4q3

From f
Domain: all x ‰ ˘2.
Vertical asymptotes: x “ ˘2 where the denominator x2 ´ 4 “ 0
1´0
Horizontal asymptotes: lim f pxq “ “1
xÑ˘8 1´0
ñ horizontal asymptote is y “ 1
Symmetry: is about the y´axis since the function is even.
Intercepts: When y “ 0, x “ ˘1, and when x “ 0, y “ 14
Hence the intercepts are p0, 1{4q, p´1, 0q and p1, 0q.
The two vertical asymptotes divide the graph into three components. The outer components
require points with |x| ą 2. WOLG we consider the points p´3, 8{5q and p3, 8{5q.
From f 1
Critical point is at x “ 0, and f 1 is not defined at x “ ˘2.
From f 2
f 2 pxq “ 0 nowhere and f 2 not defined at x “ ˘2.
94 Chapter 4. Applications of Diferentiation

ASY CP ASY
x ´2 0 2
f1 ´ undef ` 0 ´ undef ´
f2 ` undef ´ ´ undef `
f Õ undef Õ max Œ undef Œ
! " " !

x2 ´1
y“ x2 ´4
x“2

p3´, 8{5q x“2 p3, 8{5q

1{4 y“1
´1 1
x

Figure 4.12:

Exercise 4.6
1 In the following, sketch the graph of a function that has the given properties. Identify
any critical points, local maxima and minima, and inflection points. Assume that f is
continuous and its derivative exists everywhere unless the contrary is implied or explicitly
stated.
(a) f p0q “ 1, f p˘1q “ 0, f p2q “ 1, limxÑ8 f pxq “ 2, limxÑ´8 f pxq “ ´1, f 1 pxq ą 0
on p´8, 0q and on p1, 8q, f 1 pxq ă 0 on p0, 1q, f 2 pxq ą 0 on p´8, 0q and on p0, 2q,
and f 2 pxq ă 0 on p2, 8q.
(b) f p´1q “ 0, f p0q “ 2, f p1q “ 1, f p2q “ 0, f p3q “ 1, limxÑ˘8 p f pxq ` 1 ´ xq “ 0,
f 1 pxq ą 0 on p´8, ´1q, p´1, 0q and on p2, 8q, f 1 pxq ă 0 on p0, 2q, limxÑ´1 f 1 pxq “
8, f 2 pxq ą 0 on p´8, ´1q and on p1, 3q, and f 2 pxq ă 0 on p´1, 1q and on p3, 8q.
2 Sketch the graphs of the given functions
 4.7 Linear Approximations 95
` ˘3 1
(a) f pxq “ x2 ´ 1 . (d) f pxq “ 2´x2
.
(b) f pxq “ 2´x
x . (e) f pxq “ x3
.
x2 `1
x3 2
(c) f pxq “ 1`x . (f) f pxq “ x
.
x2 ´1

3. All 80 rooms in a motel will be rented each night if the manager charges $40 or less per
room. If he charges $ p40 ` xq per room, then 2x rooms will remain vacant. If each rented
room costs the manager $10 per day and each unrented room $2 per day in overhead, how
much should the manager charge per room to maximize his profit?


4.7 Linear Approximations

The tangent line to the graph y “ f pxq at P pa, f paqq is the best linear approximation of the graph
near the point since it goes through P in the same direction as the curve y “ f pxq.
Diagram Fig 4.63

The height L to the tangent line gives approximate values of f pxq for values of x near a. The tangent
line has equation y “ f paq ` f 1 paq px ´ aq.

Definition 4.7.1 The linearization of the function f about a is the function L defined by

L pxq “ f paq ` f 1 paq px ´ aq .

We say that L pxq provides a linear approximation for value of f near a.

 Example 4.7.1 Find the linearization of

?
(a) f pxq “ 1 ` x about x “ 0 and
(b) g ptq “ 1{t about t “ 1{2.


Solution
(a) f p0q “ 1. Also

1 1
f 1 pxq “ ? so that f 1 p0q “ .
2 1`x 2

Therefore L pxq “ f p0q ` f 1 p0q px ´ 0q “ 1 ` 12 x.

(b) g p1{2q “ 2, g1 ptq “ ´ t12 , g1 p1{2q “ ´4. Therefore
ˆ ˙ ˆ ˙
1 1
L ptq “ g ` g1 pt ´ 1{2q “ ¨ ¨ ¨ “ 4 ´ 4t.
2 2
96 Chapter 4. Applications of Diferentiation

The linearization can be given in terms of ∆x as follows;

f pa ` ∆xq « L pa ` ∆xq “ f paq ` f 1 paq ∆x.

 Example 4.7.2 A ball of ice melts so that its radius decreases from 5 cm to 4.92 cm. By
approximately how much does the volume of the ball decreases? 

4πr3
Solution The volume of the ball is V “ 3 , so that

dV
“ 4πr2 and L pr ` ∆rq “ V prq ` 4πr2 ∆r.
dr
Thus

∆V « 4πr2 ∆r.

For r “ 5 and ∆r “ ´0.08, we have

∆V « 4π 52 p´0.08q “ ´8π « ´25.13.

` ˘

The volume of the ball decreases by about 25 cm3 .

? ?
 Example 4.7.3 Use the linearization for x about x “ 25 to find an approximate value for 26.

?
Solution Let f pxq “ x. Then the linearization of f pxq is L pxq “ f p25q ` f 1 p25q px ´ 25q. Now
?
f p25q “ 5, f 1 pxq “ 1{ p2 xq so that f 1 p25q “ 1{10.
? 1
6 26 “ f p26q « L p26q “ 5 ` p26 ´ 25q “ 5.1.
10
Error Analysis
The error emanating from an approximation of f pxq using L pxq is

E pxq “ f pxq ´ L pxq “ f pxq ´ f paq ´ f 1 paq px ´ aq .

It is the vertical distance between the graph of f and the tangent line to that graph at x “ a as shown
below.
Diagram Fig 4.64

Observe that if x is near a, then E pxq is small compared to the horizontal distance between x and a.
4.7 Linear Approximations 97

Theorem 4.7.1 — An error formula for linearization. If f 2 ptq exists for all t in an interval
containing a and x, then D some point s between a and x such that the error E pxq “ f pxq ´ L pxq
in the linear approximation f pxq « L pxq “ f paq ` f 1 paq px ´ aq satisfies

f 2 psq
E pxq “ px ´ aq2 .
2

Proof. Assume that x ą a. Then

E ptq “ f ptq ´ f paq ´ f 1 paq pt ´ aq , E 1 ptq “ f 1 ptq ´ f 1 paq .

Applying the Generalized MVT to the two functions E ptq and pt ´ aq2 on ra, xs bearing in mind
that E paq “ 0 one obtains u P pa, xq such that

E pxq E pxq ´ E paq

2
“
px ´ aq px ´ aq2 ´ pa ´ aq2
„  „ 
MV T E 1 puq f 1 puq ´ f 1 paq MV T 1
“ “ “ “ “ f 2 psq
on ra, xs 2 pu ´ aq 2 pu ´ aq on ra, us 2

for some s P pa, uq; by second application Generalized MVT, now to f 1 on ra, us. Thus

f 2 psq
E pxq “ px ´ aq2 .
2


The following three corollaries are consequences of the above theorem

Corollary 4.7.2 If f 2 ptq has constant sign between a and x, then the error E pxq in the approxi-
mation f pxq « L pxq in Theorem 4.7.1 has the same sign; i.e. if f 2 ptq ą 0 between a and x, then
f pxq ą L pxq and if f 2 ptq ă 0 between a and x, then f pxq ă L pxq.

Corollary 4.7.3 If | f 2 ptq| ă K for all t between a and x (where K is a constant), then |E pxq| ă
pK{2q px ´ aq2 .

Corollary 4.7.4 If f 2 ptq satisfies M ă f 2 ptq ă N for all t between a and x where M and N are
constants, then

M N
L pxq ` px ´ aq2 ă f pxq ă L pxq ` px ´ aq2 .
2 2
If M and N have the same sign, a better approximation to f pxq is given by the midpoint of this
interval containing f pxq;

M`N
f pxq « L pxq ` px ´ aq2 .
4
For this approximation the error is less than half the length of the interval:
 98 Chapter 4. Applications of Diferentiation

N ´M
|E pxq| ă px ´ aq2 .
4
 Example 4.7.4 Determine the sign and estimate the size of the error in the approximation
? ?
26 « 5.1. Use these to give a small interval that you can be sure contains 26. 

Solution For f ptq “ t 1{2 , we have

1 1
f 1 ptq “ t ´1{2 and f 2 ptq “ ´ t ´3{2 .
2 4
?
Thus for 25 ă t ă 26, we have f 2 ptq ă 0, so 26 “ f p26q ă L p26q “ 5.1 implying that E p26q ă 0.
Since t ą 25, we have t 3{2 ą 253{2 “ 125, so
ˇ ˇ
ˇ 2 ˇ ˇ 1 1 ˇ 1 1 1
ˇ 4 t 3{2 ˇ ă 4 ¨ 125 “ 500 ,
ˇ f ptqˇ “ ˇ´ ˇ

and
Col 4.7.4 with N“ 500
1
1 1
|E p26q| ă p26 ´ 25q2 “ “ 0.001.
500 1000
Now E ptq “ f ptq ´ L ptq so?that f ptq “ L ptq ` E ptq. But |E p26q| ă 0.001, so that f p26q ą
L p26q ´ 0.001 “ 5.099 and 26 lies in the interval p5.099, 5.1q.

Exercise 4.7
Find the linearisation of the given function about the given point.

1. x?2 about x “ 3 4. sin x about x “ π

2. 4 ´ x about x “ 0 5. sin2 x about x “ π{6
3. 1{ p1 ` xq2 about x “ 2 6. x´3 about x “ 2

In the following, use a suitable linearisation to approximate the indicated value. Determine
the sign of the error and estimate its size. Use this information to to specify an interval
you can be sure contains the value.

? 1
7. ?50 10. 2.003
8. ?47 11. cos 460
9. 4 85 12. sin π5

4.8 Taylor Polynomial

If f pnq pxq exists in an open interval containing x “ a, then the polynomial

f 1 paq f 2 paq f pnq paq

Pn pxq “ f paq ` px ´ aq ` px ´ aq2 ` ¨ ¨ ¨ ` px ´ aqn
1! 2! n!
matches f and its first n derivatives at x “ a;

Pn paq “ f paq , Pn1 paq “ f 1 paq , Pn2 paq “ f 2 paq , ¨¨¨ , Ppnq paq “ f pnq paq ,
4.8 Taylor Polynomial 99

and so describes f pxq near a better than any other polynomial of degree at most n. Pn is called
the nth ´order Taylor polynomial for f about a. When a “ 0, the corresponding polynomial is
called Maclaurin polynomial.

 Example 4.8.1 Find the following Taylor polynomials

?
(a) P2 pxq for f pxq “ x about x “ 25.
(b) P3 pxq for g pxq “ ln x about x “ e.


Solution
1 3
(a) f 1 pxq “ 12 x 2 , f 2 pxq “ ´ 41 x´ 2 . Thus
f 2 p25q
P2 pxq “ f p25q ` f 1 p25q px ´ 25q ` px ´ 25q2
2!
1 1
“ 5` px ´ 25q ´ px ´ 25q2 .
10 1000
(b) g1 pxq “ 1x , g2 pxq “ 1
x2
, g3 pxq “ 3
x3
. Thus
g2 peq g3 peq
P3 pxq “ g peq ` g1 peq px ´ eq ` px ´ eq2 ` px ´ eq3
2! 3!
1 1 1
“ 1 ` px ´ eq ´ 2 px ´ eq2 ` 3 px ´ eq3 .
e 2e 3e

 Example 4.8.2 Find the nth ´order Maclaurin polynomial Pn pxq for ex . Use P0 p1q, P1 p1q, P2 p1q,
¨ ¨ ¨ to calculate approximate values for e “ e1 . Stop when the value is correct to 3 decimal places. 

Solution Since every derivative of ex is ex and so is 1 at x “ 0, the nth ´order Maclaurin polynomial
for ex is
x x2 x3 xn
Pn pxq “ 1 ` ` ` `¨¨¨` .
1! 2! 3! n!
Thus
P0 p1q “ 1

1
P1 p1q “ P0 p1q ` “2
1!
1
P2 p1q “ P1 p1q ` “ ¨ ¨ ¨ “ 2.5
2!
1
P3 p1q “ P2 p1q ` “ ¨ ¨ ¨ “ 2.6667
3!
1
P4 p1q “ P3 p1q ` “ ¨ ¨ ¨ “ 2.7083
4!
1
P5 p1q “ P4 p1q ` “ ¨ ¨ ¨ “ 2.7166
5!
1
P6 p1q “ P5 p1q ` “ ¨ ¨ ¨ “ 2.7180
6!
1
P7 p1q “ P6 p1q ` “ ¨ ¨ ¨ “ 2.7182
7!
100 Chapter 4. Applications of Diferentiation

Thus e « 2.718 to 3 decimal places.

Diagram Fig 4.65

st
Theorem 4.8.1 If the pn ` 1q -order derivative, f pn`1q ptq, exists for all t in an interval con-
taining a and x, and if Pn pxq is the nth -order Taylor polynomial for f about a, then the error
En pxq “ f pxq ´ Pn pxq in the approximation f pxq « Pn pxq is given by

f pn`1q psq
En pxq “ px ´ aqn`1 ,
pn ` 1q!

where s is some number between a and x. The resulting formula

f 1 paq f 2 paq
f pxq “ f paq ` px ´ aq ` px ´ aq2
1! 2!

f pnq paq f pn`1q paq

`¨¨¨` px ´ aqn ` px ´ aqn`1 ,
n! pn ` 1q!

for some s between a and x, is called Taylor’s formula with Lagrange reminder; the Lagrange
reminder term is the explicit formula given above for En pxq.

Proof. When n “ 0 we have

f 1 psq
f pxq “ P0 pxq ` E0 pxq “ f paq ` px ´ aq
1!
which is just the MVT

f pxq ´ f paq
“ f 1 psq
x´a
for some s between a and x.
When n “ 1 we obtain
f 1 paq f 2 psq
f pxq “ f paq ` px ´ aq ` px ´ aq2
1! 2!
4.8 Taylor Polynomial 101

so that
˘ f 2 psq
px ´ aq2
`
E pxq “ f pxq ´ f paq ` f 1 paq “
2!

which is the error formula for linearization given in Theorem 4.7.1.

We prove the rest of the theorem by mathematical induction. Thus suppose that the theorem
holds for n “ k ´ 1, where k ě 1 is an integer. Then we are assuming that if f is any function whose
derivative of order k exists on an interval containing a and x, then

f pkq psq
Ek´1 pxq “ px ´ aqk ,
k!

where s is any number between a and x. To prove that the error formula is true for n “ k, assume
WLOG that x ą a and apply the Generalized MVT to the functions Ek ptq and pt ´ aqk`1 on ra.xs.
Since Ek paq “ 0, we obtain u P pa, xq such that

Ek pxq Ek pxq ´ Ek paq Ek1 puq

“ “ rGMV T s “ .
px ´ aqk`1 px ´ aqk`1 ´ pa ´ aqk`1 pk ` 1q pu ´ aqk

Now
ˆ
d f 2 paq
Ek1 puq “ f ptq ´ f paq ´ f 1 paq pt ´ aq ´ pt ´ aq2
dt 2!
¸ˇ
f pkq paq k ˇ
ˇ
´¨¨¨´ pt ´ aq ˇ
k! t“u

f pkq paq
“ f 1 puq ´ f 1 paq ´ f 2 paq pu ´ aq ´ ¨ ¨ ¨ ´ pu ´ aqk´1 .
pk ´ 1q!

The last expression can be written as

k f pkq paq pu ´ aq pu ´ aqk´1 k f pkq paq

¨ “ pu ´ aqk .
k pk ´ 1q! pu ´ aq pu ´ aq k!

This is just Ek´1 puq for the function f 1 instead of f . By induction assumption it is equal to

p f 1 qpkq psq f pk`1q psq

pu ´ aqk “ pu ´ aqk ,
k! k!

for some s between a and u. Therefore

f pk`1q psq
Ek pxq “ px ´ aqk`1 .
pk ` 1q!


?
 Example 4.8.3 Use the
? second order Taylor polynomial for x about x “ 25 found in Example
4.8.1(a) to approximate
? 26. Estimate the size of the error, and specify the interval that you can be
sure contains 26. 
102 Chapter 4. Applications of Diferentiation

Solution In Example 4.8.1(a) we calculated f 2 pxq “ ´ p1{4q x´3{2 and obtained the Taylor poly-
nomial
1 1
P2 pxq “ 5 ` px ´ 25q ´ px ´ 25q2 .
10 1000
The required approximation is
? 1 1
26 “ f p26q « P2 p26q “ 5 ` p26 ´ 25q ´ p26 ´ 25q2 “ 5.099.
10 1000

Now f 3 pxq “ p3{8q x´5{2 . For 25 ă s ă 26, we have

ˇ 3 ˇ
ˇ f psq 3ˇ
ˇ
|E3 p26q| “ ˇ
ˇ p26 ´ 25q ˇ
3!
3 1
ď p26 ´ 25q3 “ “ 0.00002.
25, 000 ˆ 6 50, 000
?
Therefore, 26 lies in the interval p5.09898, 5.09902q.

Exercise 4.8
Find the indicated Taylor polynomials for the given functions using the definition of Taylor
polynomial.
1. e´x about x “ 0, order 4.
2. ln x about x “ 2, order 4.
?
3. x about x “ 4, order 3.
4. 1{ p2 ` xq about x “ 1, order n.
Use second order Taylor polynomials P2 pxq for the given function about the point specified to
approximate the indicated value. Estimate the error, and write the smallest interval you can be
sure contains the value.
5. f pxq “ x1{3 about 8; approximate 91{3 .
6. f pxq “ 1x about 1; approximate 1.02
1
.
x
7. f pxq “ e about 0; approximate e ´0.5 .