Chapter 3

Limit of a function

3.1. Introduction
Value expected of a function by looking its values at nearby points.
3.1.1 Examples:
1
(i) f (x) = , x 6= 1 What do you think could be a suitable value for
x−1
x − 1? On the right of 1, x − 1 0 and as you approach 1, x − 1 becomes
smaller and smaller,so the graph comes closer to the line x − 1.

figure 32

On the left of 1, x − 1 is negative, and as x approaches 1, x − 1 becomes

1
smaller and (-ve) and have x−1 approaches the line x = 1. Thus, we
can write
 
1
lim x → 1 = −∞.
x−1
x<1
 
1
lim x → 1 = +∞.
x−1
x>1
Clearly, no suitable value can be predicted at x = 1.

53
54 3. Limit of a function

(ii) Consider the function

x2 + 2

if x < 0
f (x) =
−x + 1 if x > 0.
For x < 0, as x −→ 0, f (x) −→ 2
Thus, once again, we are in doubt and
For x > 0, as x −→ 0, f (x) −→ 1.
cannot predict a suitable value for f (x) at x = 0. Consider the function
 2
x + 1 for x > 1
f (x) =
−x2 + 3 for x < 1.
Then, for x < 1, as x → 1, f (x) → 2. Also, for x > 1, as x → 1, f (x) → 2. Thus,
we can safely say that f (x) should be defined as f (x) = 2, to be compatible with
its values at points near x = 1. This motivates the following:
3.1.2 Definition: Let f : (a, b) → R be a function and c ∈ (a, b).
(i) If for every sequence {xn }n≥1, xn < c and xn → c, f (x − n) → L, then
L is called the left limit of f at c and is denoted by
L = f (c−) or lim x → c f (x).
x<c
(ii) If for every sequence {xn }n≥1 , xn > c ∀ n, limn→∞ xn = c, f (x) → R is
called the bf right limit of f at c and is denoted by
R := limx→c+ f (x) or f (c+ ).
(iii) We say f (x) has limit as x → c if both left and right hand limits exist
and are equal. In that case, we write
limx→c f (x) = f (c+) = f (c−)
3.1.3 Examples:
1
(i) For the function f (x) = , x 6= 0, both f (0+) and f (0−) do not exist.
x
(ii) Let 
+1 for x > 0
f (x) =
−1 for x < 0.
Then, both f (0+) and f (0−) exist with
f (0+) = 1 6= −1 = f (0−)
(iii) For 
x + 3 for x > 0
f (x) =
x2 + 3 for x < 0,
Limx→0 f (x) = 3, as f (0+) = 3 = f (0−).

3.2. Computing limits

3.2.1 Theorem: Let f, g : (a, b) −→ R and c ∈ (a, b). If
Limx→c f (x) = L, Limx→c g(m) = M.
Then the following hold:
(i) Limx→c (f (x) + g(x)) exists and equals L + M.
3.3. Continuous functions 55

(ii) Limx→c (αf (x)) exists and equals αL.

(iii) Limx→c (f (x) + g(x)) exists and equals LM.
 
f (x) L
(iv) If M 6= 0, then Limx→c exists and equals .
g(x) M
3.2.2 Example: Consider the function
f (x) = 3x2 + 2x + 2.
We want to find whether Limx→2 Φ(x) exists? Note
Φ(x) = f1 (x) + f2 (x) + f3 (x),
where 
f1 (x) = 3x2 
f2 (x) = 2x ∀x.
f3 (x) = 2

Clearly, limx→2 f3 (x) = 2. For f2 (x), using above theorem,
limx→2 f2 (x) = limx→2 (2f3 x).
= 2 (limx→2 f3 (x))
=2×2 = 4.
For f1 (x), again using above theorem,
limx→2 f1 (x) = limx→2 [3(f3 (x))2 ].
2
= 3 [limx→2 f3 (x)] .
= 3 × (2)2 = 3 × 4 = 12.
Hence,
limx→2 φ(x) = limx→2 f1 (x) + limx→2 f2 (x) + limx→2 f3 (x).
12 + 4 + 2 = 18.

3.3. Continuous functions

3.3.1 Definition:
Let f : (a, b) −→ R and c ∈ (a, b). We say f is continuous at c if limx → cf (x) =
c. A point x + c where f is not continuous is called a point of discontinuity
of the function.

3.3.2 Examples:
(i) All constant functions are continuous.
(ii) All linear functions are continuous.
(iii) All polynomial functions are continuous.
3.3.3 Discontinuities of a function: Geometrically,discontinuity at a point
means there is a break in the graph of the function.The type of discontinuities
of function can have are the following:
(i) Essential discontinuity:
(ii) Jump discontinuity:
(iii) Removable discontinuity:
56 3. Limit of a function

Essential discontinuity: This happens at a point x = c, if either the left or

the right or both of them do not exist. For example,the function f (x) = x1 , x 6= 0
has essential discontinuity at x = 0.
Jump discontinuity: This happens when at a point x = c, both the left and
the right
 limits of f at x = c exist but are not equal. For example, the function
+1 for x > 0
f (x) = has a jump discontinuity at x = 0. The value
−1 for x < 0
|f (c+) − f (c−)|
is called the jump for f at x = c.
Removable discontinuity: This happens when both the left and the right
limits of f at x = c exist, are equal. but not equal to the value of the function.
In such a case, one can redefine the function to make it continuous.
3.3.4 Applications of continuity to economics:
Modelling: Example 1: Consider a manufacturer of a car who uses bolts in
the manufacturing unit. Suppose 20, 000 cars are produced each year and each
car uses 1, 050 bolts. We can define the relation between y, the no. of cars
produced, and x, the no. of bolts used as follows:

x = 1050y.  
or (∗)
x
y= . 

1050
Note that in general, both x and y are integers. Thus, we cannot treat them
as functions on R and talk about their continuity. However, we can model our
relation (∗) as a function on R+ with values in R+ and in that case, y as a
function of x is a continuous function.
Example 2(Salary with bonus model):

8000 + 0.1s s < 2, 00, 000/ =
P =
8000 + 0.1s + 500 if s ≥ 2, 00, 000/ =
The above model has a discontinuity at s = 2, 00, 000. This is not good!
For example, consider three sales perseus:A, B, C.Their cumulative sales for the
month, except the last day, are:

cumulative sales for all except the last day of the month
A 2, 60, 000/ =
B 1, 85, 000/ =
C 60, 000/ =
What do you think the three will be doing on the last day. For A, since he
has already crossed the mark of 20, 00, 000, and got the bonus, additional work
for last day will not make a significant change in his earnings for that month. So
he may not work on the last day. Similarly, for C, he possibly cannot expect to
make sales for Rs.1, 40, 000/ = in a day to claim bonus. So he may also decide
not to work. Only for B, it is close to the critical figure of 2, 00, 000/ = and
hence he will (possible) try to work in the last day.
Example 3(Salary with commission and bonus):
Consider the earnings of a sales executive of a firm: He gets a fixed pay of Rs.
3.3. Continuous functions 57

8, 000/ = per month and commission of 10% of the sales, plus an additional
bonus of Rs.500/ = if he does sales more than or equal to Rs.2, 00, 000/ =.Then
his total take home earnings (p), as a function of sales(s) is given by
10

8000 + 100 s if s < 2, 00, 000/ =
p(s) = 10
8000 + 100 s + 500 if s ≥ 2, 00, 000/ = .

A graph of this is given by

figure 33

The function has a discontinuity at s = 2, 00, 000/ = . This is not desirable.For

example, consider three executives A,B,C. Their total sales of the month except
the last day are:

Executive Sales
A 2, 60, 000/ =
B 1, 80, 000/ =
C 60, 000/ =
What do you think they will be doing on the last day of the month? Most
probably A and C will not work for different reasons. Only B may work to meet
the target of 2, 00, 000/ = .
Marginal of a function: The concept of marginal of a function plays an
important role in analyzing the decision making problems in the short run when
it is possible to alter an input. Marginal of a function y = f (x) is defined as the
result on the output y for an additional unit of the input x. The idea is that for
an increase in unit input, the output change is small(marginal). The marginal
of a function is again a function (of x). For example, consider the input-output
function given by
y = 2x.
When x is say the labor employed. The above suggests that if we increase the
labor input by unit, then the output will double. Thus the marginal of the out-
put function is the constant function, M p(x) = 2, for every x. However notice
that more output is generated even for large inputs of labor. This seems con-
tradictory(if all other input remain fixed.) In fact, one would expect that larger
inputs of labor should reduce output and may become zero or even negative.
This suggest that marginal should not be modelled as continuous function.
Example 4(Income support programme):In a social security setup the gov-
ernment decides to give a non-employment allowance of Rs.700/ = per month
to all unemployed adults. However, the allowance is stopped if the person starts
58 3. Limit of a function

earning. Let us suppose an individual can earn Rs.20/ = per hour. Then the
earnings is given by

700 if x = 0.
E(x) =
20x if x > 0.

figure 34

Thus to earn Rs.700/ = he has to work for 35 hours. Why he should work
at all??
Modification: The Government takes 50% of his earnings and the person is
allowed to keep 50% of his earnings in addition to the allowance of 700/ = till
he has paid back 700/ = . Suppose he earns Rs. 20/ = per hour. To pay back
700/ = he has to work for
0.5 × 20 × x = 700.
700
x = × 2 = 70.
20

Thus the earning function is


700 + 0.5 × 20 × x if x ≤ 70.
E(x) =
20x if x > 70.

Then his income graph is as follows:

figure 35
3.3. Continuous functions 59

The break in the graph is removed.

Example 5(Marginal-Product function): Consider an input-output rela-

tion which is linear. Marginal-Product of an input is the amount by which
amount increases as a result of an additional unit of that input. Consider ther-
mal power station with input as coal which has capacity of 1500 mega watts
per hour. Thus, capacity per day is 1, 500 × 24 = 36, 000 mega watts per hour.
Let 250 kg. of coal is required to generate one megawatt per hour of energy.
Thus, unit increase (one quintal of coal,) will increase output by 4 megawatt per
hour. Thus, marginal output energy per day of coal is 4 till the plant reaches
it’s maximum output. That will happen for x quintals of coal produces 36, 000
of energy.
x × 4 = 3600 i.e., x = 900 quintals.
Thus, the marginal product function is

4, 0 ≤ x ≤ 9000
y=
0 x > 9000.

figure 36

Thus, the marginal product function has a discontinuity at x = 9000.

Example 6(Revenue, cost and profit functions): Consider a product being
sold at a fixed price. (competitive market).If y is the number of units produced
each day and p is the price of each unit then total revenue as a function of y is
R(y) = py,
which is a continuous function. Let the output (y) is related to input (x) by a
function
y = f (x).
Total cost as a function of input be given by
T C(x) = αx + c0 ,
where c0 is the fixed cost,α is some constant. We can treat T C as a function of
y if f is one-one, onto:
T (y) = αf −1 (y) + c0 ,
and will be continuous, if f is continuous. Total profit function
Π(y) : = T (y) − R(y).
= py − αf −1 (y) − c0 .
60 3. Limit of a function

If f is continuous, then f −1 will be continuous and hence Π(y) will be continuous.

3.3.5 Example: A cycle production plant has a capacity to produce 100 cycles
per hour using 2000 workers. The plant can operate around the clock, including
weekends, using three shifts of workers. Let h be the number of work hours used
per week and M P denotes the marginal product of workers measured by the
number of cycles produced per worker per hour. Then,
h = 2000 × 7 × 24 = 3, 36, 000 hours.
Thus, production of cycles per hour, per worker is
100 1
= = .
2000 20
so, marginal product,M P per hour per worker is given by
 1
20 till full capacity i.e.,3, 36, 000
M P (h) =
0 after that.
Marginal product ,M P has a jump discontinuity at 3, 36, 000.

figure 37

Next, suppose that the company makes a profit of Rs.1000/ = per cycle(not
including labor costs.) The wage per worker is Rs.30/ = per hour for any
weekday shift and Rs.60/ = per hour for weekend shift. The company decides
to have weekend shifts only in case it wants to produce more cycles per week
than it can using weekday shifts only. Let Π(y) denote the profit earned per
week as a function of y, the number of cycles produced per week.
This function Π(y) can be found as follows: First from the given data the
number of cycles produced by one worker per hour is given by:
No. of cycles produced per hour
Total number of workers
100 1
= = car per hour.
2000 20
Thus, a worker has to work for 20 hours to produce a car. If the worker
works on weekdays i.e., at the rate of Rs. 30/ = per hour, the cost of producing
a cycle is 20 × 30 = Rs.600/ = . Thus, the net profit per car is
1000 − 600 = Rs.400/ = per car.
So,
Π(y) = 400y, y being the cycles produced per week.
3.4. Special properties of continuous functions 61

for 0 ≤ y ≤ y0 , where y0 is the maximum number of cars produced per week

during week hours, which is (assume 5 working days in a week.)
total hours
y0 =
hour per car
24 × 5 × 2000
= = 12000.
20

If the company decides to work on weekends also, then profit will be as follows:
The labor cost for producing a cycle (in 20 hours), on weekends will be
Rs.60 × 20 = 1200/ = .
Thus the profit to such cars i.e., for y > 12000, will be (1000 − 1200)(y − 12000).
Further, the maximum capacity of the plant is 24 × 7 × 100 = 16, 800/ = . Thus,
the net profit function is:

400y for 0 < y ≤ 12, 000.
Π(y) =
400y + (1000 − 1200)(y − 12000) for 12000 < y < 16800.
i.e. 200y + 24000.

figure 38

If we consider the function Π(y) as profit per car ,then


400 0 < y ≤ 12000.
Π(y) =
400 − 200 1200 < y < 16, 800.
Thus,Π(y) has a discontinuity at y = 12000.

3.4. Special properties of continuous functions

3.4.1 Theorem:
Intermediate value property: Let f : (a, b) → R be a continuous function.
If α, β ∈ R are such that f (x1 ) = α and f (x2 ) = β, then ∀r between α and β,
there exists c between x1 and x2 such that f (c) = r.
3.4.2 Corollary: If f : (a, b) → R is a continuous function with f (x1 ) >
0, f (x2 ) < 0 for some x1 , x2 ∈ (a, b, then there exists c between x1 and x2 such
that f (c) = 0.
If f : [a, b] → R is continuous then, ∃xm in., xm ax ∈ [a, b] with the properties:
f (xmin. ) ≤ f (x) ∀x ∈ [a, b]
and f (xmax. ) ≥ f (x) ∀x ∈ [a, b].
62 3. Limit of a function

figure 39

3.4.3 Example: Let the demand and supply functions of a commodity be given
by (as a function of price):
D(p) = 100 − 2p
S(p) = 3p.
Then the price p0 where demand equals supply is called the equilibrium
price pe of the commodity. In this case, solve the above equations to get:
D(pe ) = S(pe ),
i.e.,100 − 2pe = 3pe ,
i.e.,pe = 20.

figure 40

Note at pe ,
D(pe ) = 100 − 40 = 60.
S(pe ) = 3 × 20 = 60.
Another way of solving above is to look at
f (p) = D(p) − S(p) = 100 − 5p,
and observing 0 = f (pe ) gives pe = 20. In general, it may not be that easy,as
D(p), S(p) may not be linear. However the equilibrium price is that value pe for
which
f (pe ) = D(pe ) − S(pe ) = 0.
Thus, how to ensure that there is at least one value of p for which f (p) =
D(pe ) − S(pe ) = 0. This will be so if f is continuous, f (p1 ) > 0 for some p1 and
f (p2 ) < 0 for some p2 .
3.4. Special properties of continuous functions 63

figure 41

Exercise 9: Limits of functions)

(9.1) For the following functions, compute left hand and right hand limits using
the definition at the given point a:
(i) a = 0, 
+1 if x ≥ 0
f (x) = ,
−1 if x < 0
(ii) a = 1, 
3x + 2 for x ≤ 1
f (x) = ,
x + 5 for x > 1
(iii) a = 0,
1
f (x) = ,
x−1
(9.2) (a) Think of example of a function for which limit does not exist at
exactly 2 points.
(b) Can you extend your example to a function for which limit does not exist
at 10 points?
(9.3) Using limit theorems, analyze the following limits:
(i)
1 − x2
lim , x 6= 1.
x→1 1 − x
(ii)
lim (x + 2)3 − 8 .
 
x→0
(iii)
(x + 2)3 − 8
 
lim , x 6= 0.
x→2 x
(iv)  
3x + 5
lim , x 6= −2
x→0 x+2
(9.4) Find the following
(i)
x3 − 5x2 + 2x − 4
 
lim .
x→0 x2 − 3x + 3
(ii)
√ 
x+1−1
lim , x 6= 0
x→0 x
64 3. Limit of a function

(9.5) For the following functions, show that the indicated limit does not exist
(i)
|x|
lim
x→0 x
(ii)
3
lim
x→0 x − 4
(iii)
 1
 x − 4 if x ≤ 4


lim
x→0  1
 if x > 4
x
(9.6) Consider a production function of a product with input L :
Q(L) = bL, L > 0.
Let w be the unit input price. Write the cost function and the profit function.
Discuss their continuity properties.
(9.7) A sales person earns a basic monthly salary of Rs.800/ =, plus commission
at the rate of 15% of his monthly sales and bonus of Rs1000/ = if his sales
exceeds by Rs.10, 000/ = in that month. He is also given a super bonus of
Rs.2, 500/ = if his monthly sales exceeds Rs.15, 000/ = . Write the income
function of sales and analyze its continuity.
 Chapter 4

Differentiation

4.1. Introduction
In economics,most of the time one is concerned with analyzing say, how change
in price affect its production, or how the change in supply of a commodity affect
its price, or the change in price affects the revenue of a company and so on. We
saw earlier that when a variable y is related to x by a linear relation: y = mx+c,
the rate of change at any point x = x0 , is same and is given by m, the slop of the
linear relation. In general, given y = f (x), we want to know what measures the
change in y with respect to change in x at a point x = x0 . Intuitively, if x changes
from x0 to x0 + ∆x and y changes from y0 = f (x0 ) to y0 + ∆y = f (x0 + ∆x),
then proportion of this change is represented by
f (x0 + ∆x) − f (x0 ) ∆y
= .
(x0 + ∆x) − x0 ∆x
To find the rate of change at x0 let ∆x → 0. This motivates our next definition.
4.1.1 Definition:
Let f : (a, b) → R be a function and c ∈ (a, b). The function f is said to be
differentiable at c if
f (c + h) − f (c)
limh→0
h
exists. In that case the above limit is denoted by
df (c)
f ′ (c) or
dx
and is called the derivative of f at x = c.
4.1.2 Note:
(i) For c ∈ (a, b), there exists some δ > 0 such that c ∈ (c−δ, c+δ) ⊆ (a, b).
Then for 0 < |h| < δ,
(c + h) − f (c)
φ(h) :=
h

65
66 4. Differentiation

is well defined. Thus, it makes sense to ask whether


f (c + h) − f (c)

limh→0 φ(h) = limh→0
h

exists or not. In case, this limit exists, it is the derivative of f at c.

(ii) Note that if f is differentiable at c, then f ′ (c) gives the slop of the
’tangent line’ to the graph of f at x = c.

4.1.3 Examples:

(i) f (x) ≡ c, f ′ (x) = 0∀x

(ii) f (x) = mx + c, f ′ (x) = m∀x
(iii) f (x) = x2 .
Then

f (x + h) − f (x) (x + h)2 − x2
=
h h
x2 + 2hx + h2 − x2
=
h
= 2x + h

Thus,
 
f (x + c) − f (x)
limh→0 = limh→0 (2x + h)
h
= 2x.

Hence,
d 2
(x ) = 2x ∀x.
dx
To compute derivatives of more functions the following results are useful.
4.1.4 Theorem: Let f, g : (a, b) → R and c ∈ (a, b). Let f and g be differentiable
at x = c. Then,the following holds:

(i) f ± g is differentiable at x = c and (f ± g)′ (c) = f ′ (c) ± g ′ (c).

(ii) (αf ) is differentiable at x = c and (αf )′ (c) = α(f ′ (c)).
(iii) f g is differentiable with (f g)′ (x) = f (x)g ′ (x)+f ′ (x)g(x). This is called
the product rule for differentiation.
(iv) If g(c) 6= 0, then f /g is differentiable at x = c with

g(c)f ′ (c) − f (c)g ′ (c)

 
f
(c) =
g (g(c))2 .
4.1.5 Examples:

(i) Let φ(x) = 3x2 +5x+2. Then, using the above theorem,φ′ (x) = 6x+5.
4.1. Introduction 67

x3
(ii) Let φ(x) = 2x2 +5 . Then by quotient rule,
2x2 + 5)(3x2 ) − x3 (4x + 5)
φ′ (x) =
(2x2 + 5)2 .
6x4 + 15x2 − 4x4 − 5x3
=
(2x2 + 5)2 .
2x4 − 5x3 + 15x2
=
(2x2 + 5)2 .
(iii) Consider the power function
f (x) = xn for any n ∈ Z
Then, f (x) is differentiable ∀x if n ∈ N and f ′ (x) = nxn−1 .
For n = 0, f (x) ≡ 1 and hence is differentiable with f ′ (x) = 0∀x.
For n < 0, f (x) is differentiable for x 6= 0 with f ′ (x) = nxn−1 .

4.1.6 The chain rule:

This is one of the important rules of differentiation. Suppose y is a function
of a variable u, y = f (u) and the variable u is in turn a function of a variable
x, u = g(x). Then, y can be written as a function of x,
y = f (g(x)).
This is called the composite function.The chain rule gives condition for find-
ing derivative of y as a function of x.
4.1.7 Theorem:
Let u = g(x) be differentiable at x = c and y = f (u) be differentiable at f(g(c)).
Then, the composite function y = f (g(x)) is differentiable at x = c and
(f og)′ (x) = f ′ (g(x))g ′ (c).
Equivalently,    
dy dy du
x=c = u=g(c) x=c.
dx du dx
4.1.8 Example:
Let us find the derivative of the function
f (x) = (2x + 3)2 (5x2 − 6x + 1)4
Let us write
g1 (x) = (2x + 3)3 , g2 (x) = (5x2 − 6x + 1)4 .
Both g1 and g2 are differentiable with respect to x, by chain rule, and
g1′ (x) = 3(2x + 3)2 (2x)
g2′ (x) = 4(5x2 − 6x + 1)3 (10x − 6)
Now using product rule, f is differentiable with
f ′ (x) = (g1 g2 )′ (x)
= g1 (x)g2′ (x) + g1′ (x)g2 (x)
= (2x + 3)3 4(5x2 − 6x + 1)3 (10x − 6) + 3(2x + 3)2 (2x)(5x2 − 6x + 1)4 .

4.1.9 Some more differentiable functions:

68 4. Differentiation

(i) Recall, we had defined the exponential function with natural base: x →
exp(x), x ∈ R. This function is unique:
d
(exp(x)) = exp(x)∀x.
d(x)
(ii) Derivative of inverse function: If f is one-one onto and is differ-
entiable at a point x = c with f ′ (c) 6= 0, then the inverse function f − 1
is differentiable at d = f (c) with derivative: (f − 1)(d) = f ′1(c) . As an
application of this theorem, the function ln (x), being the inverse of the
exponential function, is differentiable at every d = exp(c) > 0, with
1 1 1
(ln )′ (d) = d = = .
dx (exp(x) x=c exp(c) d
Thus,(ln )′ (x) = x1 ∀x > 0.
(iii) Since the exponential function ax , a > 0, a 6= 1, can be written as
ax = exp(xln a),
by chain rule, it is differentiable with
d n
(a ) = exp(xln a) × ln (a) = ln (a)ax .
dx
(iv) Finally, the function loga (x) being the inverse of the function ax , for
d = ac ,
d 1
(loga (x)) x=d = d y y=c .
dx dy (a )
1
=
ln (a)ay y = c
ac
= .
ln (a)
(v) For the function a ∈ R fixed, the function y = xa , x > 0, is called the
1 dy a
power function.Since ln (y) = aln (x) we have · = .
y dx x
dy ay axa
Thus, = = = axa−1 .
dx x x

4.2. Marginal of a function

In economics,the marginal of a function y = f (x) is defined as the change in y for
a unit change in x. The question is:how a unit change is measured mathematically
? Suppose f is differentiable atx0 . Then,
 
′ f (x0 + h) − f (x0 )
f (x0 ) = limh→0 .
h
From the figure,
f ′ (x0 ) = slop of the tangent atA
PB BC − P C
=
AB AB
f (x0 + h) − f (x0 ) P C
= − .
h h
4.2. Marginal of a function 69

figure 42

Thus,
f (x0 + h) − f (x0 ) = hf ′ (x0 ) − P C.
Note that P C → 0 as h → 0. Thus, we can say that
f (x0 + h) − f (x0 ) ∼
= hf ′ (x0 ).
i.e.,
f (x0 + h) ∼
= f (x0 ) + hf ′ (x0 ).
This is called the linear approximation to f (x0 +h). Another way to represent
this is
f (x0 + h) − f (x) ∼ ′
= f (x0 ).
h
In particular, for h = 1 unit change in input, the change in output is approxi-
mately given by f ′ (x0 ). This motivates the next definition.

4.2.1 Definition: For a function y = f (x), if f is differentiable, then themarginal

of f is defined to be the function f ′ (x).
4.2.2 Increasing /decreasing of functions):
Let f : (a, b) → R be an increasing function. Then, for c ∈ (a, b)
f (c + h) − f (c)
≥ 0 for h > 0.
h
Thus, if f ′ (c) exists,
 
′ f (c + h) − f (c)
f (c) = limh→0 ≥0
h
Thus, f : (a, b) → R increasing and differentiable implies f ′ (x) ≥ 0 for x ∈ (a, b).
Conversely,for f : (a, b) → R, if f ′ (x) exists ∀x ∈ (a, b) and f ′ (x) ≥ 0, then it
can be shown that f is increasing in (a, b).

Exercise 10: Derivative and its applications

10.1) Let f (x) = x2 + 1 and g(x) = x2 − 2x + 1. Find the derivatives of the
following
p functions whenever they are defined:
(i) f (x), (ii) f + g, (iii) f g, (iv) f /g, (v) f og, (vi)gof
(10.2) Suppose that an individual’s post tax income y is related to his pre-tax
income by
4
y = 100 + x.
5
 70 4. Differentiation

Find a relation between his tax T and his pre-tax income (x). Find the mar-
ginal tax-rate? Find the average tax-rate of an individual with pre-tax earn-
ings (i)1000, (ii) 2000.
(10.3) Let the post tax-earning y and pre tax earnings be related by
y = a + bx.
Find the marginal tax-rate. Under what conditions, the marginal tax rate will
be equal to average tax-rate for all values of x ?
(10.4) Let a firm’s total cost C be a function of Q, the quantity produced, and
be given by
C(Q) = Q3 − 4Q2 + 12Q.
Find the average cost and the marginal cost. When is AC = M C? If the cost
function changes to
C(Q) = Q3 − 4Q2 + 12Q + 10,
what effect it has on AC, M C ?
(10.5) If the average cost is a function f (Q) of the total quantity produced, write
relation for the total cost and compute the marginal cost. Find the conditions
when the average cost will be same as marginal cost.
(10.6) The demand function of a certain product is such that the total amount
received in sales is always same, regardless of the quantity sold. Write the
functional relation between demand and the price and calculate the elasticity
of demand for this.
(10.7) Find the income elasticity of Q, the demand Q for the following functions
relating demand to income y, by:
(i) Q = 4y − 1.
(ii) Q = ay + b.
(iii) Q = ay b .
Analyze the behavior of elasticity as income changes in the above.
(10.8) Let demand (Q) and price (P ) be related by P = 60 − 0.5Q.
(i) Express ǫd purely in terms of P and then in terms of Q.
(ii) Calculate ǫd when Q = 100
(iii) Calculate the demand when ǫd = −1.5 and p = 200.
(10.9) Compute the derivative function for the following:
(i) f (x) = 9x − 6.
(ii) f (x) = 6x4 .
(iii) f (x) = x8 + 8x3 + 2.
(iv) f (x) = −2x−4 .
(10.10) Compute derivatives of the following:
(i) f (x) = (2x5 + 6)(x3 − 3).
2
(ii) f (x) = 5x +2
x2 +1 .
(iii) f (x) = (7x3 + 1)5 .
(10.11) Show that the function f (x) = |x| is not differentiable at x = 0. Is it
differentiable at x = 1?
 Chapter 5

Optimization

5.1. Local Maxima/minima and optimization

5.1.1 Definition: f : (a, b) → R, c ∈ (a, b) we may say c is the point of local
maxima if
f (c) ≥ f (x)∀x ∈ (c − δ, c + δ)

for some δ > 0. Similarly, a point c ∈ (a, b) is called a point of local minimum
if
f (c) ≤ f (x)∀x ∈ (c − δ, c + δ)for someδ > 0.

5.1.2 Example:

(i) Let f (x) = x2 − 4x + 5, x ∈ R. Then, f ′ (x) = 2x − 4. Thus,

f ′ (x) = 2x − 4 ≥ 0 if 2x ≥ 4, i.e., x ≥ 2.
Thus, f is decreasing
and f ′ (x) = 2x − 4 ≤ 0 if 2x ≤ 4, i.e., x ≤ 2.
in (−∞, 2) and increasing in (2, +∞). Note that f has local minimum
at x = 2.
Note f is differentiable at x = 2 and f ′ (2) = 0
(ii) Let f (x) = 2x3 + 3x2 − 12x + 4, x ∈ R, then,

f ′ (x) = 6x2 + 6x − 12.

= 6(x2 + x − 2).
= 6(x + 2)(x − 1).

Thus,

f ′ (x) ≥ 0 if (x + 2) ≥ 0 and (x − 1) ≥ 0,
i.e., x ≥ −2 and x ≥ 1.

Hence, f ′ (x) ≥ 0 if x ≥ 1.
Similarly,

71
72 5. Optimization

figure 43

f ′ (x) ≥ 0 if (x + 2) ≤ 0 and (x − 1) ≤ 0,
i.e., x ≤ −2 and x ≤ 1.
Hence, f ′ (x) ≥ 0 if x ≤ −2.

Thus,the regions where f ′ (x) ≥ 0 are (−∞, −2) ∪ (1, ∞). These are the
regions where f (x) is increasing. Similarly, f ′ (x) ≤ 0, if either (x + 2) ≥ 0
and (x − 1) ≤ 0 or (x + 2) ≤ 0. Thus, f ′ (x) ≤ 0 if x ≥ −2 and x ≤ 1 i.e.,
the interval (−2, 1) and if x ≤ −2 and x ≥ 1, which is not possible.Hence,
f ′ (x) ≤ 0∀x ∈ (−2, 1), i.e.,this is interval where f is decreasing.
Note, f ′ (x) = 0, if x = −2 or x = 1. The point x = −2 is a point of local
maxima and x = 1 is a point of local minima.

figure 44
5.2. Marginal propensities 73

f (−2) = 2(−2)3 + 3(−2)2 + 12(2) + 4

= −16 + 12 + 24 + 4
f (1) = 2(1)3 + 3(1)2 − 12(1) + 4
= 2 + 3 − 12 + 4 = −3
If x = 0, y = 4 then the graph passes through (0, 4), (−2, 24), (1, −3). How
does the function behave for x < −2 and x > 1?

5.2. Marginal propensities

We next discuss how does change in income affect consumption.

5.2.1 Definition:
The relationship between consumption and income: C = f (y), y is called the
consumption function.

5.2.2 Definition:
The rate of change of C with respect to y is called the marginal of consump-
tion (or marginal propensity to consume) and is given by
dC
M P C(y) = .
dy
Using the approximation given by the derivative,
△C
M P C(y) ∼
= .
△y
This is interpreted as saying that the marginal propensity to consume is ap-
proximately the change in consumption due to unit increase in income. Suppose
the income is used up in consumption and savings:
y = C(y) + S(y).
Then, differentiating, we get
dC dS
1= + = M P C + M P S, (∗)
dS dy
where M P S = dSdy is called the marginal propensity to save. Once again, using
linear approximations
△S dS
≃ = M P S.
△y dy
The relation (*) can be used to find MPC or MPS if either is known.

5.2.3 Example:
Let the consumption as a function of income be given by
C = 0.005y 2 + 0.3y + 20
Then,
dC
MPC = = 2 × 0.005 × y + 0.3
dy
74 5. Optimization

For example, at y = 10,

M P C = 2 × 0.005 × 10 + 0.3 = .1 + .3 = .4.
Hence,
M P S = 1 − M P C = 1 − .4 = .6.
Thus, at y = 10, there is greater propensity to save than to consume.
5.2.4 Marginal revenue for linear demand function:
Let P be the price and Q be the demand(output). Then, T R = P × Q. Is the
demand function linear ?
P = a − bQ, then, T R = Q(a − bQ) = aQ − bQ2 .
d
Thus, marginal revenue (M T R) = dQ (T R) = a − 2Qb. Also note that
△(T R) d(T R)
≃ = M R,
△Q dQ
i.e., the approximate change in T R resulting from a unit change(increase) in
demand is same as M R.
5.2.5 Marginal cost:
Given total cost function
T C = F C + V C × Q = c + f (Q).Q,
where V C = f (Q). The marginal cost is defined as
d df df
MC = (T C) = (Q) + Q .
d(Q) dQ dQ
Note: Marginal cost does not depend upon F C.
5.2.6 Averages: Recall that:
TR
AR(Averagerevenue) = (5.1)
Q
FC
AF C(Averagef ixedcost) = (5.2)
Q
(V C)(Q) × Q
AV C(Averagevariablecost) = = V C(Q) (5.3)
Q
TC FC + V C × Q
AC(Averagecost) = = = AF C + AV C.
(5.4)
Q Q
AC = AF C + AV C
As,
T R = AR × Q ⇒ AR × Q = P × Q ⇒ AR = P.
5.2.7 Relation between MR and AR
Case 1: In a perfectly competitive market, i.e., when P is fixed,
T R = P Q.
d d
Thus, MR = dQ (TR) = (PQ)= P (because P is constant)= AR. Thus, in
dQ
a perfectly competitive market, MR = AR = P. Case 2: In a noncompetitive
market (monopolist), P is the function of Q,
d dP
MR = (P Q) = Q + P.
dQ dQ
5.2. Marginal propensities 75

In particular, if P is linear, P = a - bQ,

dP
= −b.
dQ
Thus,
M R = (−b)Q + (a − bQ) = a − 2bQ.
Then,
M R = A − 2bQ, if P = a − bQ AR = P = a − bQ.

figure 45

5.2.8 Example: (a) Given a perfectly competitive firm with P = 20, find the
marginal and average revenues ? Ans: MR = AR = P = 20. (b)In a monopolist
firm, with P = 50 − 2Q. Find the marginal and average revenues ? Ans: TR =
PQ =(500 − 2Q)Q = 50Q − 2Q2 . MR = d(T R)
dQ = 50 − 2 × 2Q = 50 − 4Q. AR =
P = 50 − 2Q.

TC
5.2.9 Relation between average cost and marginal cost: AC = Q ⇒
T C = AC × Q. Thus,
 
d AC
MC = (AC × Q) = Qd + AC.
dQ dQ
30
5.2.10 Example: Let AC = 2Q + 5 + Q. Then, TC = 2Q2 + 5Q + 30. MC =
2 × 2Q + 5 = 4Q + 5.
Exercise :Optimization)
(11.1) Find the local max./min. and global max./min. if any of the following
functions:
(i) f (x) = x3 − 3.
(ii) f (x) = x2 + 2x + 1.
3x
(iii) f (x) = 2 .
x +1
1
(iv) f (x) = x + .
x
76 5. Optimization

(11.2) Suppose the demand and supply relations of a market are

Qd = 1200 − 2P, Qs = 4P.
Find the equilibrium price and quantity. Write the modified demand function
if a purchase tax t per unit is imposed. Find the new equilibrium price and
quantity. Find the tax revenue as a function of t. Find a value of t that will
maximize the tax revenue. What effect it has on the quantity sold?

(11.3) Let the demand curve of a firm be P = f (Q), where f is differentiable

with f ′ (Q) 6= 0. Let the maximum of the total reverser of the firm be at a
price P0 . Show that ǫd (P0 ) = −1.

(11.4) The total cost of producing Q units of output is given by

2 2
C(Q) = 8K + Q ,
K
where K is the amount of capital employed by the firm.
(i) For K = 20, determine the level of out put at which AC (average cost) is
minimum. Show that the average cost and the marginal cost are equal at
this output.
(ii) Suppose K is treated as a variable. Find the value of K that will minimize
the cost when demand Q = 1000?
(11.5) The average cost of a firm is
1
A(Q) = 15 − 6Q + Q2 + ,
Q
where Q is the input. Find the total cost and the marginal cost curves. If
the firm can sell as many units of output as it wishes at a price of Rs.6/ =
per unit, what quantity it should sell to maximize the profits? Find the profit
also.
(11.6) A profit making firm has the total cost function
Q3
C= − Q2 + 3Q,
3
where demand and price are related by Q = 30 − P.
(i) Find the price at which the profit is maximum.
(ii) If government imposes a tax of t per unit produced, find Q for which the
profit will be maximum. Find t for which the tax revenue will be maximum.
(11.7) (Break Even cost:)
The demand function for a good is given by the relation P = 50 − 2Q while
the total cost is given by T C = 16 + 2Q.
(i) Then find relations for total profit and total revenue plot.
(ii) Compare the graphs of total profit with total cost and find the break-even
points for the firm.
(iii) Compare the levels at which profit and revenue are maximized.
(11.8) (Profit maximization in a perfectly competitive firm):
For a perfectly competitive firm with P = 121, let the variable cost be given
by T V C = Q3 /2 − 15Q2 + 175Q, with F C = 5.0.
(i) Write the relations for T C, T R, Π(Q).
5.3. Elasticity for general functions 77

(ii) Find when is the profit maximum/minimum.

(iii) Compare M C and M R.
(iv) Find the points when M C changes signs.
(11.9) (Profit Maximum with price discrimination)
A firm sells the same product in two different markets at different prices with
demand functions:
P1 = 50 − 4Q1 P2 = 50 − 3Q2 .
The total cost of the company is
T C = 120 + 8Q, where Q = Q1 + Q2 .
(i) For each market find the marginal cost and marginal revenue.
(ii) Determine price and quantity for each market at which price is maximum.
Calculate over all profit from the two markets.
(iii) Find maximum profit if price is same in both markets.
(11.10) Prove the following relations:
1
(i) M R = P (1 + ).
ǫd
(ii) If profit is maximum at Q0 , then M R = M C at Q0 .
(iii) M C = ACmin .

5.3. Elasticity for general functions

Recall, we had defined price elasticity of demand for P = f (Q).
% change in Q △Q P
ǫd = = .
% change in P △P Q
△Q
For a linear function P = a − bQ, △P = −b. was constant ∀P (and respective
Q) and hence
!
dQ P 1 P
ǫd := . = dP
× .
dP Q dQ
Q
(The cost of Q is expressed as a fraction of P .)
5.3.1 Example: Given the demand function P = 60 − 2Q. Then,
dQ
1 = 60 − 2Q .
dP
The coefficient of point elasticity of demand
1 P P P 44
ǫd (P ) = . = = = = −1.375.
−2Q Q −2Q2 +2(P − 60) 2(44 − 60)
5.3.2 Constant Elasticity of demand (for all P): In general, ǫd (P ) is a
function of P. For a special type of demand curves,it is constant.Let the demand
as the function of price be
Q = aP b , a,b are constants.
Then,
dQ
= abP b−1 .
dP
78 5. Optimization

Thus,
abP b
  
dQ P P
ǫd (P ) = = (abP b−1 )( )= = b.
dP Q Q aP b
5.3.3 Note: For such a product, total revenue as a function of price is given by
T R(Total revenue) = P Q = aP b+1
Thus, marginal of revenue as a function of price is given by
d(T R)
MR = = a(b + 1)P b .
dP
Conclusions:

(i) If b < −1, then M R < 0, i.e.,total revenue of price decreases.

(ii) If b = −1, then M R = 0, i.e., T R is independent of price.
(iii) ?????
5.3.4 Relation between elasticity, total revenue and marginal revenue:
d(T R) dP
TR = PQ ==Q + P.
dQ dQ
  !  
Q dP 1 1
⇒ MR = P 1 + . = P 1 + dQ P = P 1 +
P dQ dP . Q
ǫd

1
5.3.5 Note: Since P > 0, M R > 0 iff 1 + > 0. i.e.,
ǫd

1 ǫd < −1 if ǫd < 0.
> −1 i.e.,
ǫd or ǫd > 0
Thus, revenue will rise if ǫd < −1 or ǫd > 0.

5.4. Applications of derivative

5.4.1 Point elasticity of demand: Recall that when price as a function of
demand is given by a linear relation: P (Q) = mQ + C, we defined the concept
of elasticity of demand at Q to be the scalar
1 P
ǫd = ( ) ,
m Q
dQ 1
For a linear function P = mQ + C, is one-one, we have dP = m. This motivates
us to define the following:
5.4.2 Definition: For Q as a function of P, if Q = f (P ), then,
  
dQ P
ǫd (Q) =
dP Q
is called the point elasticity of demand at Q.
5.4.3 Example:
5.4. Applications of derivative 79

(i) Let the demand function be given by P = 60 − Q2 , then, 1 = −2Q dQ dP ,

i.e., dQ
dP = − 1
2Q . Then, the coefficient of point elasticity of demand is
dQ P 1 P 60 × P 60P
ǫd = × =− × = = .
dP Q 2Q Q 2(−Q2 ) 2(P − 60)
Thus, at P we can calculate ǫd for a given P.
(ii) Let Q as a function of P be given by the power function
Q = aP b , where a,b are fixed.
Then,
dQ
= abP b−1
dP
Thus,
abP b
 
dQ P P
ǫd = × = (abP b−1 ) = = b, a constant
dP Q Q apb
(iii) The total revenue of a firm as a function of Q is given by
T R(Q) = P Q,
where P is also a function of Q i.e., P is not fixed. (we are talking
about monopoly market.) Then,
d(T R) dQ
=P +Q
dQ dP
Thus, if
d(T R) dQ dQ
> 0, ⇔ P + Q > 0, ⇔ P Q. < −1 ⇔ ǫd < −1
dQ dP dP
dR
Thus, in a monopoly market, if ǫd < −1 then dQ > 0, which implies
that R increases as Q increases(as we shall see later.)
5.4.4 Example (Relation between average product of labor and mar-
ginal product of labor): The production function as a function of labor is
given by Q = Q(L). The average production function is
Q
APL = .
L
Assuming it is differentiable, we have
L dQ − Q
 
d(AP2 ) 1 dQ Q M PL − APL
= dL 2 = − = ,
dL L L dL L L
where M PL is the marginal product of labor. Suppose average product of labor
assumes maximum at L = L0 . Then,
d(APL ) 1
0= (L0 ) = (M PL (L0 ) − APL (L0 )).
dL L0
Hence at L = L0 , M PL = APL . Further,
dQ
 
2
d (APL ) 2
1 d Q L. − Q 
1 dQ Q

=  − dL − 2

−
dL L dL L2 L dL L
80 5. Optimization

1 d2 Q M PL − APL
 
1
= 2
− − 2 (M PL − APL ).
L dL L L
Thus, at L = L0 , M PL = APL implies
d2 (APL ) 1 d2 Q
= .
dL L dL2
5.4.5 Example (Relation between marginal revenue and marginal cost):
The total cost of a product is a function of Q, the quantity produced and is
given by
T C(Q) = F C + V C(Q)Q.
d(T C)
The marginal cost is M C = . Π(Q) Profit(Q) = TR(Q)-TC(Q). Thus,
dQ
marginal of profit is
dΠ(Q) d(T R) d(T C)
M P (Q) = = (Q) − (Q).
dQ dQ dQ
d(Π)
If Π(Q) is maximum at Q = Q0 , then (Q) = 0. i.e., at Q0 , the point of
dQ
maximum profit
d(T R) d(T C)
0 = M R(Q0 ) = (Q0 ) = (Q0 ).
dQ dQ
Some relations at points of maximum/minimum:
(1) Total profit Π = T R − T C. Then,
d(Π) d(T R)
= − d(T C)dQ = M R − M C.
dQ dQ
(a) If at point Q0 , Π is maximum, then,
d(Π)
MR − MC = = 0 ⇒ M R − M C.
dQ
(b) If M R > M C ⇒ Π ↑, M R < M C ⇒ Π ↓
d(AC) d(T C
(2) TC= AC
Q AC=Q.TC dQ =Q dQ +TC.

MAC=MTC

5.4.6 Example: The cost of building x floors in an office block requires the
following:
• The cost of the land Rs.18, 00, 000/ = .
• The cost per floor Rs.2, 00, 000/ = .
• The cost rise per floor Rs.20, 000x.
How many floors should be constructed so that the average cost per floor is
minimum ?

Solution: If building has x floors, then the total cost is

T C(x) = 18, 00, 000 + (2, 00, 000)x + (20, 000)x2
5.5. Convexity and Concavity 81

Thus,the average cost is

 
T C(x) 1800
AC(x) = = + 20 + 2x 104 .
x x
To minimize AC(x), as
d(AC) 1800
= − 2 + 2,
dx x
This gives critical point as
−1800 + 2x2 = 0, i.e., x2 = 900, i.e., x = ±30.
At x = 30,
d2 (AC) 2 × 1800 2 × 1800
= 3
= > 0,
dx x 30
thus, at x = 30, is minimum for AC(x). ???????

5.5. Convexity and Concavity

5.5.1 Definition:
• Geometric
• Convex Slop is increasing.
• Concave Slop is decreasing.
• Points of inflection f ”(x) = 0
• 3rd definition f ”(x) = 0 but f ” 6= 0
5.5.2 Example:
(i) |x(x − 1)|.
(ii) f (x) = x3 − 6x2 + 9x + 1.
(iii) Production Q as a function of labor is
Q = 6L2 − 0.2l3.
Sketch the graph.Find the labor that maximize production ?
Ans. L = 0, L = 20 (maximize).
(iv) If
T R = 4000Q − 33Q2 , T C = 2Q3 − 3Q2 + 400Q + 5000.
Find the maximum profit the company can make ? At Π(Q0 ), maxi-
mum M R(Q0 ) = M C(Q0 ).