CALCULUS

MAT-1000: Lecture Notes

Spring Semester 2026

Partha Sarathi Ghosh

Department of Mathematics
Ashoka University
What is Calculus

Calculus is the mathematics of motion and change, developed in the sixteenth and seventeenth centuries
to solve problems involving rates of change and accumulation. This powerful framework consists of
two branches: differential calculus, which analyzes how quantities change and determines slopes and
velocities, and integral calculus, which calculates areas, volumes, and reconstructs quantities from their
rates of change.
Modern calculus extends far beyond its mechanical origins, providing mathematical foundations for
physics, engineering, biology, economics, and numerous other fields. From predicting planetary motion
to modelling population growth, calculus offers a unified approach to problems involving continuous
change.
At its core, calculus addresses two fundamental geometric questions: how to measure areas of regions
bounded by curves, and how to determine the steepness of tangent lines. These seemingly simple
problems lead to profound insights connecting geometry and analysis.
More than computational techniques, calculus embodies a revolutionary way of thinking about
continuous change that has transformed our understanding of the natural world, revealing underlying
mathematical order in motion, growth, and dynamic phenomena.

Primarily Calculus has two sub-branches: (1)

Integral Calculus and (2) Differential Calculus.
Integral calculus deals with finding area under
a curve and differential calculus help us to find
steepness of tangent line at certain point on a
curve:

3
4
Contents

What is Calculus 3

1 Basic facts about numbers 7

2 Functions 19

3 Sequence of numbers 27

4 Limit and Continuity of Functions 37

5 Differentiation 53

6 Significance of the Derivative 63

5
6
Chapter 1

Basic facts about numbers

We know what "Numbers" mean. So first let us note down some of the properties of numbers.

Addition:
This operation is performed on a pair of numbers.

• The sum a + b exists for any two given numbers a and b (which may possibly be same to one of the
numbers).

Properties of addition:
P1: If a, b, c are any numbers then
( a + b) + c = a + (b + c)
→ called Associativity of numbers. This property suggests that we can put brackets anywhere
such that the brackets make sense.

We know that there is a special number 0. Why special:

P2: If a is any number then

a+0 = 0+a = a

→ existence of Identity.

P3: For every number a, there is a number b such that

a+b = b+a = 0

The number b is represented by ↑ a. → existence of Additive inverse.

Indeed if a number x satisfies the equation

a+x = a

then

(↑ a) + ( a + x ) = (↑ a) + a = 0
↓ ((↑ a) + a) + x = 0
↓ 0+x = 0 ↓ x = 0

P4: If a and b are any numbers, then

a+b = b+a (Commutativity)

7
Note that earlier we have used the subtraction property, which one also can think of as a version of
addition, like:
a ↑ b = a + (↑b)
But as addition is commutative, subtraction is not. i.e. for any two numbers a & b, in general
a ↑ b ↔= b ↑ a
Now just see the following argument. If a ↑ b = b ↑ a
↓ a↑b+b = b↑a+b
↓ a + 0 + a = (b + b) ↑ a + a
↓ ( a + a) = (b + b)
Our knowledge from childhood now will tell us that this means a = b (why?). To do this we need
another notion called "multiplication".

Multiplication:
The product of two numbers a and b will be denoted by a · b (or we will simply write ab), is again a
number.

Properties of multiplication
P5: If a, b and c are numbers then
a · (b · c) = ( a · b) · c

P6: If a is a number then

a·1 = 1·a = a (note: 1 ↔= 0)
P7: For every number a ↔= 0, there is a number a↑1 such that

a · a ↑1 = a ↑1 · a = 1

P8: If a, b are any numbers then

a·b = b·a

The condition a ↔= 0 in P7 is necessary; as for any number b we have 0 · b = 0. Therefore there is no

number 0↑1 such that 0 · 0↑1 = 1. Hence 0↑1 is meaningless. Further the division can be written as
a a
= ab↑1 . Since 0↑1 is meaningless, so is for any number a.
b 0

“Division by 0 is undefined”.

Further P7 is necessary: a · b = a · c does not necessarily imply b = c. As example take a = 0, b = 5, c = 3.

However if a ↔= 0 and ab = ac, then we know a↑1 exists and
a↑1 · ( ab) = a↑1 ( ac)
( a ↑1 a ) b = ( a ↑1 a ) c
1·b = 1·c
↓b=c
This has two important consequences.
1. If a · b = 0, then either a = 0 or b = 0.
Why: If a = 0 then we are done. Suppose not! a ↔= 0 ↓ a↑1 exists.

a ↑1 · ( a · b ) = a ↑1 · 0 = 0

( a ↑1 a ) · b = 0
1 · b = 0 ↓ b = 0.
So either a = 0 or b = 0. (* "or" always means "one or the other or both").

8
 2. It is constantly used in solution of equations. Suppose a number x is known to satisfy

( x ↑ 1)( x ↑ 2) = 0

Then it follows from that either

( x ↑ 1) = 0 or ( x ↑ 2) = 0

Hence x = 1 or x = 2.

Another property of numbers that intertwines addition and multiplication is the following:

P9: If a, b, c are any numbers, then

a · (b + c) = a · b + a · c

Using commutativity one also has:

(b + c) · a = b · a + c · a
From the above property we can say:

(↑ a) · b + ab = ((↑ a) + a) · b = 0 · b = 0

↓ (↑ a · b) = ↑( a · b)
Again:

(↑ a)(↑b) + (↑( ab)) = (↑ a)(↑b) + (↑ a)b

= (↑ a)((↑b) + b) = (↑ a) · 0 = 0

↓ (↑ a)(↑b) = ab

Q: Using the above distributive property show that b · 0 = 0 for any number b.
There are three more basic properties of numbers: "Order property". In this Calculus course, we probably
will encounter these many times. The two notions of inequality, a < b (a is less than b) and a > b (a is
greater than b) are intimately related.

• The numbers a satisfying a > 0 are called positive numbers.

• The numbers a satisfying a < 0 are called negative numbers.

• We say a < b if b ↑ a is positive.

P10: For every number a, one and only one of the following holds (Trichotomy):

1. a = 0
2. a is positive (a > 0)
3. a is negative (a < 0)

P11: If a > 0 and b > 0 then ab > 0.

P12: If a > 0 and b > 0 then a + b > 0.

Consequences:
1. If a < b, so that b ↑ a > 0
↓ (b + c) ↑ ( a + c) > 0
↓ b+c > a+c
So for any number c we will have
a+c < b+c

9
 2. If a < b and b < c then
b ↑ a > 0, c↑b > 0
↓ (b ↑ a) + (c ↑ b) > 0
↓ c ↑ a > 0 ↓ a < c (Transitive)

3. If a < 0 and b < 0, then

↑ a = 0 ↑ a > 0 and ↑ b = 0 ↑ b > 0
↓ (↑ a)(↑b) > 0 [From P11]
↓ ab > 0
 

• if a > 0 and b > 0 

The fact that ab > 0: This implies for any number a ↔= 0 : a2 = a · a > 0.

 

• also if a < 0 and b < 0
The above discussion tells us that if exactly one of a or b is negative, then ab < 0.

• If a < 0 and b > 0 ↓ ab < 0

• If a > 0 and b < 0 ↓ ab < 0

This has an important consequence. If a < b and c < 0 then

b ↑ a > 0 and c < 0

c · (b ↑ a) < 0
cb ↑ ca < 0
cb < ca
Notice how the inequality changes as we multiply by a negative number.

Another thing: suppose a > 0 and b > 0 and a < b. As a > 0, then a↑1 has two possibilities (from
trichotomy): either a↑1 < 0 or a↑1 > 0, but not both.
If a↑1 < 0, then a↑1 · a < a↑1 · 0 = 0 ↓ 1 < 0 (contradiction). So a↑1 > 0 is correct.
Since b > 0, so b↑1 > 0. So that if a < b, then

a ↑1 · a · b ↑1 < a ↑1 · b · b ↑1

b ↑1 < a ↑1
1 1
or <
b a
So taking multiplicative inverse also changes the inequality.

Absolute Value
If a > 0 then we know ↑ a < 0 and if a < 0 then we have ↑ a > 0. This is an important concept to define
"Absolute value" of a number. For any number a, we define the absolute value, | a|, of a as follows:


a if a > 0
| a| :=


↑ a if a ↗ 0

Note that | a| is always positive, except a = 0. As example:

• |5| = 5

• | ↑ 3.89| = 3.89

10
 ↘ ↘
• |↑ 2| = 2

Note that if a > 0 then a2 > 0 and if a < 0 then also a2↘> 0. And square root is only defined for
non-negative
↘ numbers. Because if not: Suppose a < 0 and a is also a number, then we know that if
a ↔= 0 then
↘
( a )2 > 0 ↓ a > 0
↘ ↘
which is a contradiction to our assumption. Hence observe that: a2 = | a|. Therefore the notation
denotes positive square root in general.
Now if a and b are two numbers then

(| a + b|)2 = ( a + b)2 = ( a + b)( a + b)

= a2 + ba + ab + b2
= a2 + 2ab + b2

Since absolute value of a number is always positive if the number is non-zero, for any number x we have
x ↗ | x |. Then

(| a + b|)2 ↗ | a|2 + 2| ab| + |b|2

= | a|2 + 2| a||b| + |b|2
= (| a| + |b|)2

↓ | a + b| ↗ | a| + |b| for all a, b

Triangle inequality.

Worked Out Examples:

1. 5 ↑ x2 < 8
↓ ↑ x2 < 8 ↑ 5 = 3
↓ x 2 > ↑3
Since for any number x, x2 ≃ 0. Therefore the above inequality is true for all numbers.

2. x2 ↑ 2x + 2 > 0
↓ ( x ↑ 1)2 + 1 > 0
Since for any number x, the number ( x ↑ 1)2 ≃ 0. Therefore for all number x, ( x ↑ 1)2 + 1 > 0.

3. x2 ↑ 4x + 3 > 0
↓ ( x ↑ 3)( x ↑ 1) > 0
Here we have product of two numbers is greater than 0. Recall the fact about this mentioned
earlier.
Possibility 1.
( x ↑ 1) > 0 and ( x ↑ 3) > 0
↓ x > 1 and x > 3
So from here we have x > 3.
Possibility 2.
( x ↑ 1) < 0 and ( x ↑ 3) < 0
↓ x < 1 and x < 3
So from here we have x < 1.

11
 1 1
4. x + 1↑ x > 0 The above inequality tells that x ↔= 0 and x ↔= 1 (why?).

1 1 1 1 1
>↑ = (Wait, ↑ = )
x 1↑x x↑1 1↑x x↑1
1 1
>
x x↑1
Now if x ↑ 1 > 0 ↓ x > 1
( x ↑ 1)
↓ > 1 ↓ ( x ↑ 1) > x
x
↓ ↑1 > 0
which is a contradiction. So ( x ↑ 1) > 0 is not possible.
Only possibility: Now assume ( x ↑ 1) < 0 ↓ x < 1.
x↑1
↓ <1 [since ( x ↑ 1) < 0]
x
Now further assume x > 0.
↓ x ↑ 1 < x ↓ ↑1 < 0
This inequality is correct. So the possibility is 0 < x < 1. (Note: If we assume x < 0, then
multiplying by x reverses inequality again: x ↑ 1 > x ↓ ↑1 > 0, contradiction).

5. What is the meaning of | x | < a? Where a is a number, a > 0. In other words, for which x the above
inequality is true?
Recall the definition of absolute value: Suppose x > 0 then | x | = x. Then the inequality suggests
x < a.
Suppose now x < 0, then | x | = ↑ x. Then the inequality gives ↑ x < a.

↓ (↑1)(↑ x ) > (↑1) a

↓ x > ↑a
Hence combining all above we have
|x| < a
↓ ↑a < x < a
i.e. all numbers in between ↑ a and a.

Various Number Systems

Earlier the term "number" was used very loosely; there our concern was to know the properties. The
simplest numbers are the "counting numbers": 1, 2, 3, 4, . . . The collection of these numbers is denoted by
symbol N (Natural numbers). P2, P3, and some other properties are not satisfied by N.
The most basic property of N is the principle of "mathematical induction" (PMI). Suppose P(n) is a
statement concerning the number n. The PMI states that P(n) is true for all natural numbers n provided
that:
1. P(1) is true.
2. Whenever P(k ) is true then P(k + 1) is true.

The deficiencies of the natural numbers may be partially remedied by extending this system to the set of
integers:
. . . , ↑3, ↑2, ↑1, 0, 1, 2, 3, . . .
This set is denoted by Z. But here we do not have multiplicative inverse of non-zero numbers!
m
A still larger system of numbers is obtained by quotients of integers with n ↔= 0. These numbers are
n
called rational numbers and the set of all rational numbers is denoted by Q.

12
 Q satisfies all the properties P1-P12 that are mentioned above.

However, a still larger collection of numbers to which properties P1-P12 apply is called the set of all
real numbers, denoted by R. The numbers which are real numbers but not rational are called irrational
numbers. The next question is whether there is any irrational number at all? For the time, let us assume
square root of non-negative numbers exists (even though this statement needs a proof).
↘ ↘ p
Therefore 2 is a number. Suppose it is rational, then 2 = where q ⇐ Z and q ↔= 0. Further we can
q
assume gcd( p, q) = 1. Then

↘ ( )2
p
( 2)2 =
q
↓ 2q2 = p2
This shows that p2 is even ↓ p is even. Then there must exist k ⇐ Z such that

p = 2k ↓ 2q2 = (2k )2

↓ 2q2 = 4k2
↓ q2 = 2k2
↘
Which implies q2 is even ↓ q is even too. Contradicting gcd( p, q) = 1. Hence
↘ 2 is not rational
↓ 2 is irrational

Geometric interpretation of real numbers:

We are used to think of real numbers as "points" on a straight line. That is:

• We first draw a straight line.

• A point is selected to represent 0.
• Another point is chosen to the right of 0 to represent 1.
• The above choices will determine "unit" length.
• Then a point at the left of 0, situated at the same distance as 1, is represented by -1.
• The point, to the right of 0, which is twice as far as 1 is, is represented by 2 and so on.

... ...
↑3 ↑2 ↑1 0 1 2 3

• With this arrangement, if a < b, then the point corresponding to a will lie to the left of the point
corresponding to b.
p
• Similarly if > 0, then we will first sub-divide the sub-line from 0 to 1 into q many equal pieces
q
p
and we will go to the right of 0, to p times the new distance to obtain the point . Similarly to the
q
p
left for ↑ .
q
• It is usually taken for granted that irrational numbers also somehow fit into this scheme.

"Thus a number is sometimes called a point and R is often called real line".

... ... R

13
Intervals
The order relation on the set of real numbers determines a natural collection of subsets called intervals.

Notations
If a < b, then the open interval from a to b is the set

( a, b) = { x ⇐ R | a < x < b}

The closed interval from a to b is the set

[ a, b] = { x ⇐ R | a ↗ x ↗ b}

The half-open intervals are

[ a, b) = { x ⇐ R | a ↗ x < b}
( a, b] = { x ⇐ R | a < x ↗ b}

Infinite Intervals
( a, ∞) = { x ⇐ R | x > a}
[ a, ∞) = { x ⇐ R | x ≃ a}
(↑∞, b) = { x ⇐ R | x < b}
(↑∞, b] = { x ⇐ R | x ↗ b}
(↑∞, ∞) = R

( a, b) [ a, b]

a b a b

[ a, b) ( a, ∞)

a b a

Solving Inequalities
Eg 1: Solve for x: 2x ↑ x2 < ↑8.
Solution:
2x ↑ x2 + 8 < 0
Multiply by (↑1) (inequality sign reverses):

x2 ↑ 2x ↑ 8 > 0

( x ↑ 4)( x + 2) > 0

Critical points: x = 4 and x = ↑2. These points divide the real line into three intervals: (↑∞, ↑2),
(↑2, 4), and (4, ∞).
Sign Chart:

Values of ( x ↑ 4)( x + 2)
+ ↑ +

↑2 4

14
 Test points:
– x = ↑3: (↑3 ↑ 4)(↑3 + 2) = (↑7)(↑1) = 7 > 0
– x = 0: (0 ↑ 4)(0 + 2) = (↑4)(2) = ↑8 < 0
– x = 5: (5 ↑ 4)(5 + 2) = (1)(7) = 7 > 0
Since we need ( x ↑ 4)( x + 2) > 0, the solution is:
x ⇐ (↑∞, ↑2) ⇒ (4, ∞)
2x ↑ 3 1
Eg 2: Solve < .
x+2 3
Solution:
2x ↑ 3 1
↑ <0
x+2 3
3(2x ↑ 3) ↑ 1( x + 2)
<0
3( x + 2)
6x ↑ 9 ↑ x ↑ 2
<0
3( x + 2)
5x ↑ 11
<0
3( x + 2)

– Numerator is zero when 5x ↑ 11 = 0 ↓ x = 11

5 = 2.2.
– Denominator is zero when x + 2 = 0 ↓ x = ↑2.
Intervals to test: (↑∞, ↑2), (↑2, 11 11
5 ), ( 5 , ∞ ).
Sign Diagram:

+ ↑ +

↑2 11
5
Test x = ↑3: Test x = 0: Test x = 3:
↑ =+ ↑ =↑ + =+
(↑) (+) (+)

We need the expression to be < 0 (negative). Thus, the solution is (↑2, 11

5 ).
Eg 3: Solve | x ↑ 5| < 2. Then
↑2 < x ↑ 5 < 2
Add 5 to all parts:
3<x<7
Solution: (3, 7).
Geometric Interpretation: | x ↑ 5| is the distance between x and 5. The inequality says this distance
must be less than 2.

2 2

3 5 7

Eg 4: Solve |3x + 2| ≃ 4.
Using property 2:
3x + 2 ≃ 4 or 3x + 2 ↗ ↑4
Solving first part:
2
3x ≃ 2 ↓ x ≃
3
Solving second part:
3x ↗ ↑6 ↓ x ↗ ↑2
Solution set: (↑∞, ↑2] ⇒ [ 23 , ∞).

15
Eg 5: Solve | x + 1| < | x ↑ 4|.
Since both sides are non-negative, we can square both sides:

( x + 1)2 < ( x ↑ 4)2

x2 + 2x + 1 < x2 ↑ 8x + 16
Subtract x2 from both sides:
2x + 1 < ↑8x + 16
10x < 15
15
x<
10
3
x<
2
Solution: (↑∞, 1.5).

Completeness Axiom of real numbers

Suppose A ⇑ R is a non-empty set.
• We say A is bounded above if ⇓ M ⇐ R such that for any a ⇐ A we have a ↗ M. In this case ‘M’ is
an upper bound of A.
• We say A is bounded below if ⇓ m ⇐ R such that for any a ⇐ A we have a ≃ m. Here ‘m’ is a lower
bound of A.
• If A is both bounded above and below then A is called a bounded subset of R.
• If A is either not bounded above, or not bounded below or both then we will call A an unbounded
subset of R.
Q: What is the meaning of the phrase: “b ⇐ R is not an upper bound (lower bound) of A”?

Supremum / Least upper bound (l.u.b)

Suppose A ⇑ R is non-empty and bounded above. Then l.u.b of A is the least upper bound (a real
number) among all upper bounds of A.
Definition: s ⇐ R is called supremum of A or l.u.b of A if for any upper bound v of A we have s ↗ v. In
this case we write: sup( A) = s = l.u.b( A).
Another useful criterion for supremum is the following: s ⇐ R is called supremum of A if it satisfies
the following two conditions:

1. s is an upper bound of A
2. for any ε > 0, the real number s ↑ ε is not an upper bound of A, i.e. there exists one a ⇐ A such that

s ↑ ε < a ↗ s.

Infimum / Greatest lower bound (g.l.b)

It is defined similar to the above notion but here we will talk about lower bounds. Suppose A ⇑ R is
non-empty and bounded below: Then l ⇐ R is called infimum or g.l.b of A if

1. l is a lower bound of A
2. for any ε > 0, the real number l + ε is not an upper bound of A, i.e. there exists one a ⇐ A such that

l ↗ a < l + ε.

In this case we write: inf( A) = l = g.l.b( A).

16
Completeness of R
"The result that separates Q from R".



• (Version 1) Any non-empty bounded above subset of R has a least upper bound.

Statement: • (Version 2) Any non-empty bounded below subset of R has a greatest lower


bound.

Here we will discuss some of the consequences of the completeness axiom.

Theorem 1.0.1. N is an unbounded subset of R.

Proof. Suppose not. Then N is a non-empty bounded above subset of R. So sup(N ) exists and for each
n ⇐ N, n ↗ sup(N ). Therefore [sup(N ) ↑ 1] is not an upper bound of N. ↓ ⇓ N ⇐ N such that

sup(N ) ↑ 1 < N ↗ sup(N )

So from 1st inequality:

sup(N ) < N + 1
There is a natural number strictly greater than sup(N ), violating the definition of supremum. Hence our
assumption is wrong. So N is unbounded.

Therefore inside the set N, the symbol n → ∞ makes sense. As a result R itself is also unbounded.

1
Now suppose x ⇐ R is a positive real number (it may be very small or very large). Then x is also a
positive real number. Since N is unbounded, so ⇓ Nx ⇐ N such that

1 1
(as N is unbounded) Nx > ↓0< <x
x Nx

This gives a proof of the following theorem:

Theorem 1.0.2. (Archimedean Property of real numbers) If x > 0, then ⇓ N ⇐ N (depending on x)

such that
1
0< < x.
N

Greatest Integer: Given a real number x, there is a unique integer m ⇐ Z such that

m ↗ x < m+1

The integer m is called the greatest integer corresponding to x and is denoted by [ x ].

Theorem 1.0.3. (Density of Rational numbers) Suppose x, y ⇐ R and x < y. Then ⇓ q ⇐ Q such that

x < q < y.

Proof. Since x < y ↓ y ↑ x > 0 ↓ ⇓ N ⇐ N such that

1
< y↑x
N
↓ 1 < Ny ↑ Nx

17
Take M := [ Nx ] + 1. Then

M ↑ 1 = [ Nx ] ↗ Nx ↓ M ↑ 1 ↗ Nx < M

↓ M ↗ Nx + 1 < Nx + ( Ny ↑ Nx ) = Ny
↓ Nx < M < Ny
M
x< < y.
N

↘ ↘
Now if x < y are two real numbers, then 2x < 2y. So from density of rationals, ⇓ q ⇐ Q such that
↘ ↘
2x < q < 2y
q
x< ↘ <y
2
q
Try to show that ↘ is an irrational number. And hence the following is also true:
2

There always exists an irrational number between two real numbers. Called density of irrational
numbers.

18
Chapter 2

Functions

Forget abstract symbols for a moment. Let’s talk something concrete: how things are made. In our lives,
we see processes everywhere—a baker transforms flour into bread, a sawmill turns logs into lumber,
a printer converts digital files into printed posters. Each of these is a reliable, predictable system. You
don’t put flour into an oven and hope for a bicycle; you expect bread. This idea of a predictable process
is the very heart of a function.
In calculus, we think of a function not as a cryptic equation, but as a specialized factory. For a practical
example:

• The factory name: Every factory needs an identifier. This is our function’s name, almost always a
letter like f .
• The Raw Material: This is what gets fed into the factory on the input conveyor belt. In mathematics,
we almost always call this raw material an "x".
• The Blueprint: This is the core of the operation. The expression " f ( x ) = . . . " is not the machine
itself; it is the engineer’s blueprint or the factory’s instruction manual. It describes exactly what
transformations will happen to the raw material inside the factory.
• The finished product: This is what emerges from the other end of the factory. We denote this final
product as f ( x ) (read: f of x).

Eg: If f is a pottery studio. Its blueprint " f (clay) = [shape on a wheel, fire, Glaze]". If you feed f a
lump of clay ’x’, then ’ f ( x )’ is a finished, glazed pot.

Mathematically:
A function on numbers is a rule which assigns, to each of certain real numbers, some real number. As
discussed earlier to define a function one needs two "sets" (collection of objects). In our case, most of the
time the sets consist of numbers.

Definition 2.0.1. A function f is a rule that assigns to each element x in a set A exactly one element, called
f ( x ), in a set B. We write f : A → B

Domain: A Codomain: B

x f (x)

Range

19
The set A is called the Domain of f , on which the function is defined. The set B is called the Codomain
or Target Set. The Range of f is the set of all possible values of f ( x ) as x varies throughout the domain:
Range = { f ( x ) | x ⇐ A}.
Terminology:
• x is the independent variable.
• y = f ( x ) is the dependent variable.

Example 2.0.2. 1. f : R → R, defined by f ( x ) := x2 .

x3 +3x +5
2. f : R → R, defined by f ( x ) := x 2 +1
.

(2.9)3 + 3(2.9) + 5
f (2.9) =
(2.9)2 + 1

Earlier we mentioned that 0a doesn’t make any sense. Notice x2 + 1 > 0 for any x ⇐ R. So this f is defined
on all x ⇐ R. Hence Dom( f ) = R.
3. Now suppose f has an expression like:
x3 + 3x + 5
f (x) =
x2 ↑ 1
Now here if x = 1 or x = ↑1, then f ( x ) is undefined. Therefore Dom( f ) = R \ {1, ↑1}.
4. Polynomial function: A function f : R → R is a polynomial function if there are real numbers
a0 , a1 , . . . , an such that:
f ( x ) = a n x n + a n ↑1 x n ↑1 + · · · + a 2 x 2 + a1 x + a 0
for all x ⇐ R. The ai ’s are called coefficients of the polynomial. The highest power of x with a non-zero
coefficient is called the degree of f .
Like if f ( x ) = x3 + 3x + 5, then degree( f ) = 3.
5. Constant function: Let c be a real number. f : R → R defined by f ( x ) = c. Dom( f ) = R,
Codomain( f ) = R, Range( f ) = {c}.
6. Identity function: Suppose A is a set. f : A → A defined by f ( x ) = x is called the Identity function. This
is represented by Id A or I.

Construction of new functions out of old ones

Suppose f and g are two functions from R to R (Codomain = R).

1. We can define a new function f + g, called the sum of f and g, by the equation:
( f + g)( x ) = f ( x ) + g( x )
Note that the domain of f + g consists of all x for which " f ( x ) + g( x )" makes sense. i.e., x has to be
in Dom( f ) as well as Dom( g).
Dom( f + g) = Dom( f ) ⇔ Dom( g)

2. Similarly, f · g is the function given by the equation:

( f · g)( x ) := f ( x ) · g( x )
Domain is Dom( f ) ⇔ Dom( g).
f f (x)
3. On the other hand, if we talk about quotient: ( )( x ) := . The domain is
g g( x )
Dom( f ) ⇔ Dom( g) ⇔ { x : g( x ) ↔= 0}.
If at some x0 ⇐ Dom( f ) ⇔ Dom( g) we have g( x0 ) = 0, then the quotient won’t be defined.

20
Composition of two functions

Definition 2.0.3. Suppose f : A → R and g : B → R are two functions. We define a new function ( f ↖ g),
called the composition of f and g, by:
( f ↖ g)( x ) = f ( g( x ))
Here we have applied f on g( x ). In order to make sense, g( x ) has to be in Dom( f ) i.e. same as

Range( g) ⇑ Dom( f ).

Therefore:
Dom( f ↖ g) = { x : x ⇐ Dom( g) and g( x ) ⇐ Dom( f )}

f ↖g

g g( x ) f
x f ( g( x ))

A B C

1
Example: f ( x ) = , then Dom( f ) = R \ {1}. g( x ) = x2 , Dom( g) = R. Now the expression of
x↑1
( f ↖ g) is:
1 1
( f ↖ g)( x ) = f ( g( x )) = = 2
g( x ) ↑ 1 x ↑1

Notice g(1) = g(↑1) = 1 ⇐

/ Dom( f ). So f ↖ g is not defined at ±1.

f ↖ g : R \ {1, ↑1} → R

If we want to construct g ↖ f , then:

( )2
2 1 1
( g ↖ f )( x ) = g( f ( x )) = ( f ( x )) = =
x↑1 ( x ↑ 1)2

Then g ↖ f : R \ {1} → R.

Special types of functions

Definition 2.0.4. Let f : A → B be a function.

• f is called injective (one-one) if whenever x ↔= y and x, y ⇐ A then f ( x ) ↔= f (y).
or f is called injective (one-one) if whenever f ( x ) = f (y) and x, y ⇐ A then x = y.

• f is surjective (onto) if for all w ⇐ B, there exists x ⇐ A such that f ( x ) = w.

• If f is both injective and surjective, then f is a bijective function.

Example: Let A = { x ⇐ R : x ↔= 1} and f : A → R is defined by:

2x
f (x) =
x↑1

21
Suppose x1 , x2 ⇐ A and assume f ( x1 ) = f ( x2 ). We will show that f is injective:

2x1 2x2
= ↓ x1 ( x2 ↑ 1) = x2 ( x1 ↑ 1)
x1 ↑ 1 x2 ↑ 1
↓ x1 x2 ↑ x1 = x2 x1 ↑ x2
↓ ↑ x1 = ↑ x2 ↓ x1 = x2
Therefore f is injective.
2x
To determine the range of f , we need to solve the equation y = x ↑1 for x in terms of y.

y( x ↑ 1) = 2x ↓ xy ↑ y = 2x ↓ x (y ↑ 2) = y
y
x=
y↑2
This makes sense only if y ↔= 2. Thus Range( f ) = {y ⇐ R : y ↔= 2}. Thus f : Dom( f ) → Range( f ) is a
bijection.

Inverse of a function
Suppose f : A → B is a bijection. This tells that for each x ⇐ A, the element f ( x ) ⇐ B is unique. Further
the one-one & onto-ness of f implies that for each y ⇐ B there is exactly one x ⇐ A with f ( x ) = y.

f (Bijective)

x1 y1

x2 y3

x3 y2

Domain A Codomain B

Therefore the direction of the arrows can be reversed in a unique way from B to A, giving a function
from B to A. This function is called the inverse of f and is denoted by f ↑1 : B → A.

Dom( f ↑1 ) = Range( f )

Range( f ↑1 ) = Dom( f )

How to check? Suppose f : A → B is a bijection and we claim that g : B → A is the inverse of f . The
claim will only be correct if we show that:

f ↖ g = IdB and g ↖ f = Id A

2x y
Example: f : R \ {1} → R \ {2}, f ( x ) = x ↑1 . g : R \ {2} → R \ {1}, g ( y ) = y ↑2 .
2x
f (x) x ↑1 2x 2x
( g ↖ f )( x ) = g( f ( x )) = = 2x
= = = x = IdR\{1} ( x )
f (x) ↑ 2 x ↑1 ↑ 2
2x ↑ 2( x ↑ 1) 2
y
2g(y) 2 y ↑2 2y 2y
( f ↖ g)(y) = f ( g(y)) = = y = = = y = IdR\{2} (y)
g(y) ↑ 1 y ↑2 ↑1 y ↑ ( y ↑ 2) 2
y
Hence f ↑1 (y) = y ↑2 .

22
Graph of a function
Suppose A ↙ R and f : A → R is a real function. Then this function may be illustrated geometrically by
a graph in the xy-plane. We plot the domain A on the x-axis and above each point x ⇐ A, we plot the
point ( x, y), where y = f ( x ). The totality of such ( x, y) is called the graph of the function f .

Examples:

1. 2.
f :R→R f :R→R
f (x) = x f (x) = |x|

f (x) f (x)
4 4
f (x) = x f (x) = |x|

2 2

x x
↑4 ↑2 2 4 ↑4 ↑2 2 4

↑2 ↑2

↑4 ↑4

Monotone functions
Suppose f : A → R is a function, A is a subset of R.

Definition 2.0.5. a) If x, y ⇐ A and x < y ↓ f ( x ) ↗ f (y), then f is called monotonically increas-

ing.
b) If x, y ⇐ A and x < y ↓ f ( x ) < f (y), then f is called strictly increasing.
c) If x, y ⇐ A and x < y ↓ f ( x ) ≃ f (y), then f is called monotonically decreasing.

d) If x, y ⇐ A and x < y ↓ f ( x ) > f (y), then f is called strictly decreasing.

Q: Try to draw schematic graphs of the functions mentioned above.

Even Function f (x)

4
f ( x ) = x2

f : R → R is called an even function if for all

2
x ⇐ R we have
x
↑4 ↑2 2 4
f ( x ) = f (↑ x )
↑2

Law of trichotomy tells that if x > 0 then ↑ x < 0.

Therefore the function here will produce same ↑4
value at x and ↑ x. So if we know the graph of f
for all x ≃ 0, then just by reflecting it w.r.t y-axis
we will obtain the full graph of f .

Example: f ( x ) = x2 .

23
Odd Function f (x)
4
f ( x ) = x3
f : R → R is called odd function if for all x ⇐ R
we have 2

f ( x ) = ↑ f (↑ x ) or f (↑ x ) = ↑ f ( x ) x
↑4 ↑2 2 4
By the same logic if we know the graph of f for
all x ≃ 0, then reflecting it w.r.t origin (0, 0) will ↑2
give us the full picture. i.e. first reflect w.r.t y-axis,
then reflect the new graph w.r.t x-axis.
↑4
Example: f ( x ) = x3 .

Graph under shift and scaling

• Suppose f : R → R is known along with its g( x ) = f ( x ) + c

graph. Define
g:R→R f (x)
by g( x ) := f ( x ) + c where c is a positive number.
Then notice that at the point x, the value of g( x ) h( x ) = f ( x ) ↑ c
is "c"-more than the value of f ( x ). So the graph
of f will be shifted upward ’c’ distance.
Similarly if h( x ) = f ( x ) ↑ c, the graph shifts
downward.

• If g : R → R is given by

g( x ) = f ( x + c) (c positive).
f
Then a little computation shows: g( x ↑ c) = f ( x ). g h
Since c is positive, ( x ↑ c) will lie left to x on the
x-axis; and at the point ( x ↑ c) the value under
g will be f ( x ). This means we have to shift the
graph of f to the left by "c"-amount, to obtain the
graph of g.
Similarly, h( x ) = f ( x ↑ c) shifts the graph to the
right.

Obtaining the graph of inverse of f

Suppose f : R → R is a bijection, so we know inverse of f exists and f ↑1 : R → R is a function. Now if
( x, y) is on the graph of f then
y = f ( x ) ∝↓ x = f ↑1 (y)

∝↓ (y, x ) is on the graph of f ↑1

Now (y, x ) can be obtained by reflecting ( x, y) w.r.t the straight line y = x. Therefore the graph of f ↑1
can be obtained by reflecting the graph of f w.r.t y = x line.

24
 f y=x

f ↑1

25
26
Chapter 3

Sequence of numbers

Let us start with a simple question: Can we write down a list of numbers that follows a specific pattern?
Like:

a) 1, 2, 3, 4, 5, . . . (natural numbers)
b) 2, 4, 6, 8, . . . (even natural numbers)
c) 1, 12 , 13 , 14 , . . . (reciprocals of natural numbers)

The above lists are examples of sequences. Informally it is an infinitely long list of real numbers.

Definition 3.0.1. A sequence is a function f : N → R. So the range of the function can be written as
{ f (1), f (2), f (3), . . . } or { f (n) : n ⇐ N }.

Generally we denote a sequence by ( xn ) or ( xn )n⇐N where x1 is the first term, x2 is the second, . . . , xn is
the n-th term (or general term). So here our main objective is to investigate what happens to a sequence
when we consider larger and larger "inputs" i.e. natural numbers.
Q: What is the meaning of n → ∞?: The symbol ’∞’ is not at all a real number. It is a sense which says
that we can take larger and larger natural numbers.
Suppose ( xn ) is a sequence of real numbers.

Definition 3.0.2. We say that the sequence ( xn ) is convergent

• if there exists a real number l ⇐ R such that for all " > 0, there exists a natural number N (")
(depending on ") such that
| xn ↑ l | < " for all n ≃ N (").
In this case we write: lim xn = l.
n→∞

• If no such ’l’ exists then we say ( xn ) is divergent.

• If for every l ⇐ R, there exists N ⇐ N such that xn > l for all n ≃ N, then we say ( xn ) diverges to ∞.

Example 3.0.3. xn = n for all n ⇐ N. Then ( xn ) is diverging to ∞.

1
Example 3.0.4. Suppose xn = for all n ⇐ N. Let " > 0, then from Archimedean property there exists N ⇐ N
n
1
such that < ". So for all n ≃ N we have:
N
1 1
0< ↗ < " ↓ | x n ↑ 0| < "
n N
1
Hence: lim = 0.
n→∞ n

27
 2n2 + 1
Example 3.0.5. Suppose: xn = for all n ⇐ N. Then
3n2 + 1
1
2+ n2
xn = 1
3+ n2

Let us calculate the following modulus:

* * * * * * * *
*
* xn ↑ 2 ** ** 2n2 + 1 2 ** ** 3(2n2 + 1) ↑ 2(3n2 + 1) ** ** 1 *
*
* = ↑ = =*
3 * * 3n2 + 1 3 * * 2
3(3n + 1) * 3(3n + 1) *
2

Since we know that n ≃ 1 for all n ⇐ N:

n < n2
3(3n + 1) < 3(3n2 + 1)
1 1 1 1
↓ < < <
3(3n2 + 1) 3(3n + 1) 3n + 1 n
Using this inequality we have * *
* 2 ** 1
* xn ↑ < for all n ⇐ N.
* 3 * n
1
Let " > 0, then from Archimedean property ⇓ N ⇐ N such that 0 < N < ". So for all n ≃ N we have:

2 1 1
| xn ↑ | < ↗ <"
3 n N

2n2 + 1 2
↓ lim xn = lim = . (done)
n→∞ n→∞ 3n2 + 1 3

Some special sequences

Result 3.0.6. If 0 < b < 1, then lim bn = 0.

n→∞

Proof. To prove this we will use elementary properties of natural logarithm function. If " > 0 is given,
we see that
bn < "
if and only if
n · ln(b) < ln(")
Since 0 < b < 1, ln(b) < 0. Thus
ln(")
n>
ln(b)
Thus if we choose N to be a natural number such that
ln(")
N>
ln(b)

then we will have

|bn ↑ 0| < " for all n ≃ N.
↓ lim bn = 0
n→∞

Result 3.0.7. If c > 0, then {c1/n } is a convergent seq and lim c1/n = 1.
n→∞

28
Proof. Suppose C = 1. Then it is a constant seq of 1, therefore limn→∞ c1/n = 1.
Suppose C > 1: Then we know from law of trichotomy that c1/n > 1 ↓ c1/n = 1 + dn where dn > 0.

↓ C = (1 + dn )n ≃ 1 + ndn for n ⇐ N.

C↑1
↓ dn ↗ for n ⇐ N
n
C↑1
↓ |c1/n ↑ 1| = dn ↗ for n ⇐ N
n

1 "
Now Archimedean property imply that ⇓ N ⇐ N such that N < C ↑1 .

↓ |c1/n ↑ 1| < " ′n ≃ N ↓ lim c1/n = 1

n→∞

1
Now suppose 0 < C < 1: Then c1/n = 1+ h n for hn > 0.

1 1 1
C= ↗ <
(1 + h n ) n 1 + nhn nhn

1
0 < hn < for n ⇐ N.
nC
Therefore: * *
1/n
* 1 ** hn
|1 ↑ c *
| = *1 ↑ = < hn
1 + hn * 1 + hn

1
|1 ↑ c1/n | < hn <
nC
1
Then again Archimedean property: ⇓ N ⇐ N such that < "C.
N

↓ |1 ↑ c1/n | < " ′n ≃ N

↓ lim c1/n = 1

Result 3.0.8. The sequence (n1/n ) is convergent and lim n1/n = 1.

n→∞

Proof. Since n1/n > 1 for n > 1, we can write n1/n = 1 + k n for some k n > 0 when n > 1. Then for n > 1
we have:
n ( n ↑ 1) 2 n ( n ↑ 1) 2
n = (1 + k n )n = 1 + nk n + kn + · · · ≃ 1 + kn
2 2
n ( n ↑ 1) 2
↓ n↑1 ≃ kn
2

Since n ↑ 1 > 0 ↓ k2n ↗ n2 .

2
If " > 0, from Archimedean property ⇓ N ⇐ N such that N1 < "2 . It follows for n ≃ max{2, N } we have
2 2
n <" .
( )1/2
1/n 2
↓ |n ↑ 1| = k n < < " ↓ lim n1/n = 1.
n n→∞

29
 Now suppose ( xn ) is a convergent sequence such that limn→∞ xn = l ⇐ R. Let " := 1, then
⇓ N ⇐ N such that ′n ≃ N we will have

l ↑ 1 < xn < l + 1

Set: M := max{ x1 , x2 , . . . , x N , l + 1} and m := min{ x1 , x2 , . . . , x N , l ↑ 1}. Then for any n ⇐ N we

have:
m ↗ xn ↗ M
So the sequence is bounded. Any convergent seq is bounded.

n2 + 1
Example 3.0.9. xn = . Then after rewriting we have:
n+2
5
x n = ( n ↑ 2) +
n+2
Then using Archimedean property we can conclude that the second summand can be made arbitrarily small. But
the first term is not bounded above for all n ⇐ N. Hence xn is not bounded above. ↓ ( xn ) is not convergent
sequence.

Limit Laws for sequences:

If ( xn ) and (yn ) are convergent sequences and c ⇐ R. Then

1. lim ( xn ± yn ) = lim xn ± lim yn

n→∞ n→∞ n→∞

2. lim ( xn yn ) = ( lim xn )( lim yn )

n→∞ n→∞ n→∞

3. lim (c · xn ) = c · ( lim xn )
n→∞ n→∞

xn limn→∞ xn
4. lim = (provided lim yn ↔= 0)
n→∞ yn limn→∞ yn

5. lim ( xn ) p = ( lim xn ) p (Given p > 0 and xn > 0).

n→∞ n→∞

( ) + 1
, ( )
2n2 + 1 2+ n2 1
Example 3.0.10. lim = lim . Using first law above we have lim 2+ 2 = 2 and
n→∞ 3n2 + 1
n→∞ 3 + 1 n→∞ n
( ) n2
1
lim 3 + 2 = 3 and therefore we can apply the fourth law here at this stage to conclude:
n→∞ n
( )
2n2 + 1 2
lim = .
n→∞ 3n2 + 1 3

( ) ( )
12 + 22 + · · · + n2 n(n + 1)(2n + 1) 2 1
Example 3.0.11. lim = lim = = .
n→∞ n3 + 1 n→∞ 6( n3 + 1) 6 3

Some important theorems:

Theorem 3.0.12. If ( xn ) is a convergent sequence and xn ≃ 0 for all n ⇐ N. Then limn→∞ xn ≃ 0.

Proof. Say lim xn = x. If x < 0, then from the definition we know that for any " > 0, ⇓ N ⇐ N such that
for all n ≃ N we have
x ↑ " < xn < x + "

30
Since " is arbitrary, choose " = ↑ x > 0. Then for this particular " we have:
x ↑ (↑ x ) < xn < x + (↑ x ) ′n ≃ N
↓ 2x < xn < 0 ′n ≃ N
This contradicts that xn ≃ 0 ′n ⇐ N. Hence x ≃ 0.

Corollary 3.0.13. If ( xn ) and (yn ) are two convergent sequences and xn ↗ yn ′n ⇐ N, then limn→∞ xn ↗
limn→∞ yn .

Theorem 3.0.14. Sandwich Theorem / Squeeze Theorem: Suppose ( xn ), (yn ), (zn ) are three sequences of
real numbers such that
xn ↗ yn ↗ zn for all n ⇐ N
and we know limn→∞ xn = limn→∞ zn = l ⇐ R. Then (yn ) is convergent and

lim yn = l.
n→∞

- .
sin(n)
Example 3.0.15. Consider the sequence yn = n . Since we know that the trigonometric sine function
satisfies the inequality ↑1 ↗ sin (n) ↗ 1, therefore:
( )
1 sin(n) 1
↑ ↗ ↗
n n n
1 1
holds. Taking xn = ↑ and zn = in the above squeeze theorem we can say that
n n
( )
sin(n)
lim = 0.
n→∞ n

Monotone Sequences
Suppose ( xn ) is a sequence.
• If x1 ↗ x2 ↗ x3 ↗ · · · ↗ xn ↗ xn+1 ↗ . . . , then we say ( xn ) is an increasing sequence.
• If x1 ≃ x2 ≃ x3 ≃ · · · ≃ xn ≃ xn+1 ≃ . . . , then we say ( xn ) is a decreasing sequence.
If ( xn ) is either increasing or decreasing then ( xn ) is called a monotone sequence.

Theorem 3.0.16. A monotone sequence of real numbers is convergent if and only if it is bounded. Further:
1. If ( xn ) is a bounded increasing sequence then limn→∞ xn = sup{ xn : n ⇐ N }.

2. If ( xn ) is a bounded decreasing sequence then limn→∞ xn = inf{ xn : n ⇐ N }.

Proof. Earlier we have proved that any convergent sequence is bounded. Hence done.
Conversely (Monotone bounded sequence ↓ convergent):
• Suppose ( xn ) is monotonically increasing and bounded sequence. Therefore ⇓ M ⇐ R such that
xn ↗ M ′n ⇐ N. Consider the set A := { xn : n ⇐ N } ⇑ R. Then A is non-empty and bounded
above subset of R, so that sup( A) exists, and sup( A) = x ⇐ R.
If " > 0, then x ↑ " is not an upper bound of A, so ⇓ N ⇐ N such that
x ↑ " < xN ↗ x
Now for all n ≃ N we have:
x ↑ " < x N ↗ xn ↗ x < x + "
↓ | xn ↑ x | < " ′n ≃ N
So ( xn ) is a convergent sequence and limn→∞ xn = x = sup( A).

31
 • Suppose ( xn ) is monotonically decreasing and bounded below sequence. So ⇓m ⇐ R such that
m ↗ xn ′n ⇐ N. Consider the set A := { xn : n ⇐ N }. Therefore A is non-empty, bounded below
subset of R, so inf( A) exists and inf( A) = y (say).
If " > 0, then from definition of infimum y + " is not a lower bound. ⇓ N ⇐ N such that y ↗ x N <
y + ". Then from property of decreasing sequence:

y ↑ " < y ↗ xn ↗ x N < y + " ′n ≃ N

↓ | xn ↑ y| < " ′n ≃ N
So ( xn ) is a convergent sequence and limn→∞ xn = y = inf( A).

Existence of one of the important real number: Euler’s Number (e)

Let en := (1 + n1 )n be a sequence. We will show that this sequence is monotonically increasing and
bounded above. Hence the sequence (en ) is convergent. The limit limn→∞ en is called Euler’s number
and we write: ( )
1 n
lim en = lim 1 + =e
n→∞ n→∞ n
Applying binomial theorem:
( )n
( ) ( ) ( )
1 n 1 n 1 n 1
en = 1+
= 1+ + + · · · +
n 1 n 2 n2 n nn
( ) ( )( ) ( ) ( )
1 1 1 1 2 1 1 n↑1
= 1+1+ 1↑ + 1↑ 1↑ +···+ 1↑ ... 1↑
2! n 3! n n n! n n

Similarly,
( ) ( )( )
1 1 1 1 2
e n +1 = 1 + 1 + 1↑ + 1↑ 1↑
2! n+1 3! n+1 n+1
( ) ( )
1 1 n↑1
+···+ 1↑ ... 1↑
n! n+1 n+1
( ) ( )
1 1 n
+ 1↑ ... 1↑
( n + 1) ! n+1 n+1

The above expression of en contains (n + 1) terms and en+1 contains (n + 2) terms. Notice that for
any k: ( ) ( )
k k
1↑ < 1↑
n n+1
Therefore, comparing term by term, the above inequalities imply that

e n < e n +1 ′n ⇐ N

Further: for each n ⇐ N, there is (1 + 1) = 2 in the expression of en along with some positive terms.

↓ 2 < e n < e n +1 ′n ⇐ N

If k is any number from the set {1, 2, . . . , n}, then

( )
k
1↑ <1 . . . (i )
n

Moreover, using Principle of Mathematical Induction we can show that:

1 1
↗ m ↑1 for all m ⇐ N . . . (ii )
m! 2

32
 Now we will substitute inequalities (i) & (ii) into the expression of en above:

1 1 1
2 < en < 1 + 1 + + 2 + · · · + n ↑1 < 3
2 2 2
So (en ) is a sequence which is: 1. Monotonically increasing and 2. 2 < en < 3 ′n ⇐ N.
Hence the sequence is convergent and we write
( )n
1
lim 1+ =e
n→∞ n

Limit law for exponents involving sequences

In mathematics there are several indeterminate forms; and dividing any number by ‘0’ is one of them
which is not allowed. Some other types of indeterminate forms are:

0 ∞
, , 0 ∞ ∞, ∞ ↑ ∞, 1∞ , 0∞ , ∞0 .
0 ∞
Whenever we encounter a situation like this, then there is no general result or theorem. We need to deal
each of them case by case. Why we are saying this is: Suppose ( xn ) and (yn ) are two sequences and we
- .limn→∞ yn
expect that lim ( xn )yn = lim xn is going to be true then here are some counter-examples:
n→∞ n→∞
( )1 1
1 n 3. lim (n) n = 1
1. lim =1 n→∞
n→∞ n

( )2
1 n 1 1
2. lim = 4. lim (2n ) n = 2
n→∞ 2n 4 n→∞

Result 3.0.17. If ( xn ) and (yn ) are two sequences converging to x and y respectively and x ↔= 0 ↔= y then
yn y
lim ( xn ) = (x) .
n→∞

Similar for the case 1∞ : think with the sequences (1n ) and en defined above.

Divergent Sequence
Earlier we have mentioned that if a sequence ( xn ) is not convergent, then we call it divergent. There are
primarily two versions of divergence. Let’s see two examples:

( xn = n) and (yn = (↑1)n )

If we apply the definition of convergence of a sequence to both of the sequences then we will see that
there is no real number l or l ∈ such that limn→∞ xn = l and limn→∞ yn = l ∈ . There is a fundamental
difference in the behavior of these two sequences.

• ( xn ) is unbounded.

• (yn ) is bounded.

Definition 3.0.18. We say ( xn ) is diverging to ∞ (or ↑∞) and write limn→∞ xn = ∞ (or ↑∞) if the
following is true:

• For any M ⇐ R, there exists N ⇐ N such that xn > M (or xn < M) for all n ≃ N.

33
 2 ↘
Example 3.0.19. n, n2 , nn++11 , n etc.

Oscillatory Sequence: A sequence ( xn ), which does not converge to any real number, neither it diverges
to ∞, nor diverges to ↑∞, is called oscillatory sequence or diverge by oscillation.


n if n is even
Example 3.0.20. (↑1)n , xn = etc.

1
n if n is odd

Extra: Existence of Square root:

If time permits then we will discuss this portion i.e. existence of square root of a positive real number at
the very end of the course. But I am still incorporating this topic here: If you are interested then please
go through it.
Suppose a > 0. Consider the set
A = { x ⇐ R ≃0 : x 2 ↗ a }
Since (1 + a)2 > a, the number (1 + a) is an upper bound of A. Also A is non-empty (e.g., 0 ⇐ A or
a
1+ a ⇐ A).
A is non-empty and bounded above subset of R ↓ sup( A) exists: sup( A) = b ⇐ R. Now law of
trichotomy says exactly one is true: b2 > a, b2 < a, or b2 = a. Suppose b2 < a: Since b > 0, we may
choose c > 0 such that (b + c)2 < a.

(b + c)2 = b2 + c(2b + c)

We can choose small enough c such that b2 + c(2b + c) < a. ↓ (b + c) ⇐ A and b + c > b (contradiction
to b being supremum).
Suppose b2 > a: Let c = (b ↑ "). We can show there exists c < b such that c2 > a. ↓ c2 > x2 for each
x ⇐ A ↓ c > x for each x ⇐ A. So c is an upper bound of A smaller than b (contradiction).

↘
↓ b2 = a, therefore b = a exists as a real number.

34
Flow chart of main definitions

Suppose ( xn ) is a se-
quence of real numbers

( xn ) is convergent and lim xn = L ( xn ) is not convergent, i.e., divergent

n→∞
∝↓ For every ε > 0, there exists N ⇐
N such that | xn ↑ L| < ε for all n ≃ N

If for any M ⇐ R, there exists

N ⇐ N such that M ↗ xn for
all n ≃ N, then lim xn = +∞.
n→∞

If for any m ⇐ R, there exists N ⇐ N such

that xn ↗ m for all n ≃ N, then lim xn = ↑∞.
n→∞

• ( xn ) is not convergent
• neither lim xn = +∞
n→∞
• nor lim xn = ↑∞
n→∞
↓ ( xn ) is oscillatory divergent

35
36
Chapter 4

Limit and Continuity of Functions

Limit of functions
If you remember, one of our goal is to draw graphs of a given function, whose domain may be R or may
be an interval in R or may be a subset in R. For a function like f : R → R given by f ( x ) = x2 we can
calculate f (1) = 1, f (2) = 4, f (3.5) = 12.25 and no on and then we can plot ( x, f ( x )) on the xy- plane.
But doing it for all the points in R in not practically possible. But situation will be slightly better if we
know what is happening in between two consecutive plotings. Consider a function, whose expression is

x2 ↑ 4
f (x) =
x↑2
We know the domain of this function in whole R except 2; as the expression f (2) is not valid. If we
are to draw the graph of this f , this means there in a “hole” at x = 2. Our goal is to understand the
behaviour of the function f near this hole. Even though we cannot evaluate f (2), we want to ask:

• What value does f ( x ) get closer and closer to as x gets closer and closer to 2?

Suppose f : D → R in a function and D is just a subset of R (not necessarily an interval). Though we

don’t know the meaning of “lim f ( x )” till now, we will first try to answer for which real number ‘c’ the
x →c
above expression will make sense.
We are given a function f : D → R.

Definition 4.0.1. A real number c is called a limit point of the set D if there exist a sequence of real numbers
( xn ) such that:
1. xn ⇐ D for all n ⇐ N
2. xn ↔= c for all n ⇐ N

3. lim xn = c
n→∞

Why is this notion of limit point of D important to understand the expression lim f ( x ). Earlier we
x →c
2
have looked at the example f ( x ) = xx↑+24 . That function is not defined at 2. But if we wish to know what
is happening to f ( x ) as x gets closer and closer to 2. In order to make this thing possible we need the
notion of limit point.
Suppose f : D → R and ’c’ in a limit point of D. To study limit of f at ’c’, it in not necessary that f in
defined at ’c’ or c ⇐ Dom( f ).

Definition 4.0.2. We say that a real number l in is said to be a limit of f at c (or f ( x ) approaches l as x

37
 approaches c) if for every sequence ( xn ) in D with xn ↔= c for all n ⇐ N and lim xn = c we have
n→∞

lim f ( xn ) = l.
n→∞

In this case we write: lim f ( x ) = l.

x →c

2
Let f : R ↑ {2} → R defined by f ( x ) = xx↑↑24 . Here f is not defined at 2. Dom( f ) = R ↑ {2}. Define
1
xn = 2 ↑ for n ⇐ N, then xn ⇐ Dom( f ) and xn ↔= 2 for all n ⇐ N. We know lim xn = 2 so 2 is a limit
n n→∞
point of Dom( f ). So talking about lim f ( x ) makes sense.
x →2

Let ( xn ) be any convergent seq such that lim xn = 2 and xn ↔= 2 for all n ⇐ N. And now let us
n→∞
examine the seq ( f ( xn )).

xn2 ↑ 4 ( xn + 2)( xn ↑ 2)
lim f ( xn ) = lim = lim
n→∞ n→∞ xn ↑ 2 n→∞ ( x n ↑ 2)
Since xn ↔= 2 for all n ⇐ N, so the factor xn ↑ 2 ↔= 0 for all n ⇐ N; therefore

lim f ( xn ) = lim ( xn + 2) = lim xn + lim 2 = 2 + 2 = 4

n→∞ n→∞ n→∞ n→∞

Hence lim f ( x ) = 4.
x →2

From our knowledge of straight line, this is

y the graph of f and notice that as f is not defined
at 2 so there is a “hole” in the above picture at
hole x = 2. So the limit represents that as x going
close to 2 (red arrow), the functional value f ( x )
goes to arbitrarily close to 4 (black arrow).

Example 4.0.3. We now want to check whether lim f ( x ) exists, for the following function:
x →0


1 if x ≃ 0
f (x) =

↑1 if x < 0
Consider two sequences ( xn = n1 ) and (yn = ↑ n1 ). Then we know: lim xn = 0 = lim yn . Further notice that
n→∞ n→∞
yn < 0 < xn for all n ⇐ N. Therefore

lim f ( xn ) = lim 1 = 1 and lim f (yn ) = lim (↑1) = ↑1.

n→∞ n→∞ n→∞ n→∞

This observation violates the definition of limit at 0 for this function f .

Note 4.0.4. Let f : D → R and c in a limit point of D. If two sequences ( xn ), (yn ) with xn , yn ⇐ D and
xn ↔= c ↔= yn for all n ⇐ N and lim xn = c = lim yn such that

lim f ( xn ) ↔= lim f (yn )

n→∞ n→∞

38
 Then lim f ( x ) does not exist.
x →c

Notice that to check whether lim f ( x ) exists or not; the definition tells that we need to check
x →c
lim f ( xn ), for every seq ( xn ) with lim xn = c. Now to deal with every seq is practically very im-
n→∞ n→∞
possible work. The above note gives us a tool to bypass that. The tool is called one-sided limit. Let
f : D → R and c in a limit point of D.

Definition 4.0.5 (Left-hand limit). If there exists l ⇐ R such that for every seq ( xn ) with xn ⇐ D we have
xn < c and lim xn = c implies lim f ( xn ) = l. Then we say lim f ( x ) exists, and called left hand limit of
n→∞ n→∞ x →c↑
f at c and we write
lim f ( x ) = l.
x →c↑

Definition 4.0.6 (Right-hand limit). If there exists l ⇐ R such that for every seq ( xn ) with xn ⇐ D we
have xn > c and lim xn = c implies lim f ( xn ) = l. Then we say lim f ( x ) exists, and called right hand
n→∞ n→∞ x →c+
limit of f at c and we write
lim f ( x ) = l.
x →c+

Theorem 4.0.7 (One of the tool to calculate limit). lim f ( x ) = l if and only if
x →c

lim f ( x ) = l = lim f ( x ).
x →c↑ x →c+

Limit Laws or algebra of limits:

Suppose f , g : D → R are two functions and c is a limit point of D and lim f ( x ) = l, lim g( x ) = m. Then
x →c x →c

1. limx→c ( f ± g)( x ) = l ± m
2. limx→c ( f · g)( x ) = l · m
- .
f
3. limx→c g ( x ) = ml provided g( x ) ↔= 0 near c and m ↔= 0.

Example 4.0.8. 1. f : R → R is a constant function: f ( x ) = p for all x ⇐ R. Let c ⇐ R and ( xn ) be an

arbitrary seq of real numbers such that xn ↔= c for all n and lim xn = c. Then lim f ( xn ) = lim p = p.
n→∞ n→∞ n→∞
So for an arbitrary seq ( xn ) with the mentioned property we have lim f ( xn ) = p. Thus lim f ( x ) = p.
n→∞ x →c

2. f : R → R given by f ( x ) = x for all x ⇐ R. Then for any seq ( xn ) with xn ↔= c for all n and
limn→∞ xn = c we have limn→∞ f ( xn ) = limn→∞ xn = c ↓ lim f ( x ) = c.
x →c

3. f : R → R : f ( x ) = x k for all x ⇐ R and k ⇐ N. Let c ⇐ R and ( xn ) be an arbitrary seq of real numbers

such that xn ↔= c for all n ⇐ N and lim xn = c. Now
n→∞

lim f ( xn ) = lim xnk = lim ( xn · xn · · · xn ) (k-many)

n→∞ n→∞ n→∞
= ( lim xn )( lim xn ) · · · ( lim xn ) = c · c · · · c = ck
n→∞ n→∞ n→∞

Since the seq ( xn ) is arbitrary: lim f ( x ) = ck .

x →c

39
 4. Suppose f : R → R is a function whose expression is f ( x ) := ak x k + ak↑1 x k↑1 + · · · + a1 x + a0 . Where
each ai ⇐ R for i = 1, 2, . . . , k and ak ↔= 0. Any function with this type of expression is called a polynomial
and deg( f ) := k is called its degree.

Now let, c ⇐ R and ( xn ) be an arbitrary seq of real numbers such that xn ↔= c for all n ⇐ N and
lim xn = c.
n→∞

For any i ⇐ {1, 2, . . . , k} we have limx→c ai xi = limn→∞ ( ai xni ) = ai limn→∞ xni = ai ci .

lim f ( xn ) = lim ( ak xnk + ak↑1 xnk↑1 + · · · + a1 xn + a0 )

n→∞ n→∞
= a k c k + a k ↑1 c k ↑1 + · · · + a1 c + a0

Therefore (as our chosen seq was arbitrary)

lim f ( x ) = ak ck + ak↑1 ck↑1 + · · · + a1 c + a0

x →c

Eventually here we have: lim f ( x ) = f (c).

x →c

5. Suppose P : R → R and Q : R → R are two polynomials such that Q( x ) ↔= 0 for all x ⇐ R. (Here the
P( x )
condition on Q is much stronger but think of weaker conditions). Define f : R → R by f ( x ) = Q( x) . Then
for any c ⇐ R we have:
lim P( x )
lim f ( x ) = x→c
x →c lim Q( x )
x →c
P(c)
= [from previous result]
Q(c)
↓ lim f ( x ) = f (c).
x →c

6. f : R → R defined by
 - .

sin 1
x if x ↔= 0
f (x) =


0 if x = 0
Q: Does lim f ( x ) exist?
x →0
1 2
Consider two seq xn = nε and yn = (4n+1)ε
. Then from definition and Archimedean property it is clear
that: lim xn = 0 = lim yn . But:
n→∞ n→∞

lim f ( xn ) = lim sin(nε ) = lim 0 = 0

n→∞ n→∞ n→∞

And: ( )
(4n + 1)ε
lim f (yn ) = lim sin = lim 1 = 1.
n→∞ n→∞ 2 n→∞

Since lim f ( xn ) ↔= lim f (yn ), lim f ( x ) does not exist.

n→∞ n→∞ x →0

Squeeze Theorem

Theorem 4.0.9 (Sandwich or Squeeze Theorem:). Suppose f , g, h : D → R are functions with f ( x ) ↗

g( x ) ↗ h( x ) for all x ⇐ D (This condition can be made even weaker; think!). Let c be a limit point of D and
lim f ( x ), lim h( x ) exist and are equal. Then lim g( x ) exists and
x →c x →c x →c

lim g( x ) = lim f ( x ) = lim h( x ).

x →c x →c x →c

40
Proof. Let ( xn ) be an arbitrary seq of real numbers such that xn ↔= c for all n and lim xn = c. Since we
n→∞
know that lim f ( x ), lim h( x ) exist and are equal: Therefore
x →c x →c

lim f ( xn ) = lim h( xn )
n→∞ n→∞

For this sequence ( xn ) we also have the inequality f ( xn ) ↗ g( xn ) ↗ h( xn ) for each n ⇐ N. Hence using
squeeze theorem of sequences we can say lim g( xn ) exists and
n→∞

lim g( xn ) = lim f ( xn ) = lim h( xn ).

n→∞ n→∞ n→∞

Since ( xn ) was an arbitrary sequence with the properties of definition, hence we have the proof of the
theorem.

Remark 4.0.10. The same theorem is also true fro left hand limit as well as right hand limit.
1. Squeeze Theorem for left hand limit: Suppose f , g, h : D → R are functions with f ( x ) ↗ g( x ) ↗ h( x )
for all x ⇐ D (This condition can be made even weaker; think!). Let c be a limit point of D and lim f ( x ),
x →c↑
lim h( x ) exist and are equal. Then lim g( x ) exists and
x →c↑ x →c↑

lim g( x ) = lim f ( x ) = lim h( x ).

x →c↑ x →c↑ x →c↑

2. Squeeze Theorem for right hand limit: Suppose f , g, h : D → R are functions with f ( x ) ↗ g( x ) ↗
h( x ) for all x ⇐ D (This condition can be made even weaker; think!). Let c be a limit point of D and
lim f ( x ), lim h( x ) exist and are equal. Then lim g( x ) exists and
x →c+ x →c+ x →c+

lim g( x ) = lim f ( x ) = lim h( x ).

x →c+ x →c+ x →c+

Example 4.0.11. 1. f : R → R defined by:



 x sin 1
x x ↔= 0
f (x) =


0 x=0

1
Let’s examine lim f ( x ). Earlier we have seen: lim x = 0 and lim sin does not exist. So can’t apply
x →0 x →0 x →0 x
multiplication rule. But we certainly know know, for any x ⇐ R \ {0}:
( )
1
↑1 ↗ sin ↗1
x
- .
1
If x ≃ 0 then ↑ x ↗ x sin x ↗ x. Now applying the squeeze theorem for left hand limit we have
( )
1
lim x sin = 0.
x →0 + x
( ) ( )
1 1
Q: Similarly taking x < 0, show that lim x sin = 0. And hence lim x sin = 0.
x →0↑ x x →0 x

↘ positive- square root is defined for all non-negative real numbers. Define f :
2. Earlier we have seen that
[0, ∞) → R by f ( x ) = x. Let c > 0. The following inequality is true:
* * * *
*↘ ↘ * * a ** ** x ↑ a **
* x ↑ c* ↗ * ↘ x ↑ ↘
* x + a * ↗ * ↘a * .

41
 Therefore we have:
* * * *
*x ↑ a* ↘ ↘ *x ↑ a*
↑ * ↘ * ↗ x ↑ c ↗ * ↘ **
* * *
a a
Hence applying lim and squeeze theorem we have
x →c
↘ ↘
lim x= c (why?).
x →c

Existence of a function that has no limit at any point of R

• Fact 1: If c ⇐ R, then there exists a seq (qn ) such that qn ⇐ Q for all n ⇐ N and

lim qn = c.
n→∞

• Fact 2: If c ⇐ R, then there exists a seq (in ) such that in ⇐ R \ Q, for all n ⇐ N and

lim in = c.
n→∞

Some authors also call this “Completeness axiom of R”.

Proof. Fact 1: Consider the seq xn = c ↑ n1 , yn = c + n1 . Then ( xn ) is monotonically increasing, bounded

above seq ↓ lim xn exists and
n→∞
lim xn = c
n→∞
And (yn ) is monotonically decreasing, bounded below seq ↓ lim yn exists and
n→∞

lim yn = c
n→∞

Recall the density property of rational numbers “between any two real numbers, there is a rational”. So
for each n ⇐ N, ⇓ qn ⇐ Q such that
1 1
c ↑ < qn < c ↑
n n+1
and ⇓ q∈n ⇐ Q such that
1 1
c+ < q∈n < c + .
n+1 n
1
Then |qn ↑ c| < ↓ lim qn = c and |q∈n ↑ c| < n1 ↓ lim q∈n = c. Instead of one seq, we constructed
n n→∞ n→∞
two seqs with desired property. Proof of Fact 2 will be similar (Do it!!!).

Example 4.0.12. One application of the above fact: f : R → R defined by:



1 if x ⇐ Q
f (x) =


↑1 if x ⇐ /Q

Let c ⇐ R be a real no. (qn ) a seq of rational no. with qn ↔= c and lim qn = c and there also (in ) a seq of irrational
n→∞
no. with in ↔= c and lim in = c. See that
n→∞

lim f (qn ) = lim 1 = 1

n→∞ n→∞
lim f (in ) = lim (↑1) = ↑1.
n→∞ n→∞
Hence lim f ( x ) does not exist. And this is true for all real no. c ⇐ R.
x →c

42
Some fundamental limits

y In this figure the circle is of radius 1 and has

(0, 0) as its centre. AB and CP are perpendicular
to x-axis. If ϱ is positive and 0 < ϱ < ε2 , then
from this picture we can extract the following
A P information:

• A ∋ (cos ϱ, sin ϱ )
ϱ
x
O B C • P ∋ (1, tan ϱ )

• length of AB = sin ϱ

• length of PC = tan ϱ

• length of the arc AC = ϱ

ε
Therefore what we have now is: For 0 < ϱ < 2

Squeeze Theorem (Right Hand)

0 < sin ϱ < ϱ =====================↓ lim sin ϱ = 0
ϱ →0+

Similarly if ↑ ε2 < ϱ < 0 then from law of trichotomy: 0 < ↑ϱ < ε2 and here we can again apply the
previous inequality & with a subtle use of Squeeze theorem will give us lim sin ϱ = 0. Together we
ϱ →0↑
have:

lim sin ϱ = 0.
ϱ →0

/
↓ lim cos ϱ = lim 1 ↑ sin2 ϱ = 1 (??)
ϱ →0 ϱ →0

ε
Again from the above picture, when 0 < ϱ < then 0 < sin ϱ < ϱ < tan ϱ is true and we will have
2
sin ϱ
cos ϱ < sinϱ ϱ < 1 and applying suitable squeeze theorem we have lim = 1; further applying the
x →0+ ϱ
sin(↑ϱ )
previous strategy with the fact ↑ϱ = ↑↑sinϱ ϱ = sinϱ ϱ we can show the following:

sin ϱ
lim = 1
ϱ →0 ϱ

Some techniques to evaluate limits

Let us now discuss some tools to calculate limits of functions. Remember the indeterminate forms and
undefined forms, that we discussed earlier. Suppose you know the expression for the function f and you
want to calculate lim f ( x ).
x →c

1. Direct substitution Method

Simply substitute ’c’ for x in the expression of f if it does not produce an indeterminate form.
2x +3 2·1+3 5
• limx→1 5x +1 = 5·1+1 = 6
• limx→1 2x ↑2
6x ↑6 = 2·1↑2
6·1↑6 = 0
0 (indeterminate, so this method is not applicable here.)

43
2. Factorization Method
Whenever you encounter an indeterminate form, look for whether you can take the help of this
method to bypass the problem.
2x ↑2 2( x ↑1) 2
• limx→1 6x ↑6 = limx→1 6( x ↑1)
= limx→1 6 = 26 .
x 2 ↑4 ( x +2)( x ↑2)
• limx→2 x ↑2 = limx→2 ( x ↑2)
= limx→2 ( x + 2) = 2 + 2 = 4.

3. Rationalization Method
Apply this especially to limits involving roots. Multiply the numerator and denominator by the
conjugate.
↘ ↘ ↘
1↑ 1↑x (1 ↑ 1 ↑ x )(1 + 1 ↑ x ) 1 1
lim = lim ↘ = lim ↘ =
x →0 x x →0 x (1 + 1 ↑ x ) x → 0 1+ 1↑x 2

44
Continuous functions
If f is an arbitrary function, it is not necessarily true that lim f ( x ) = f (c). In fact there are many ways
x →c
this can fail to be true. For example:

y y
f (c)
y
hole

x x
x c c
c

2. f might not even be defined at 3. even if f is defined at c and

1. lim f ( x ) may not exist.
x →c c, in which case this equation lim f ( x ) exists, but might not
x →c
does not make any sense: equal to f (c):

Intuitively, a function f is continuous if the graph contains no breaks, jumps or wild oscillations.
Although this description will usually enable you to decide whether a function is continuous simply by
looking at its graph (a skill well worth cultivating) it is easy to be fooled, and a definition is important.

Definition 4.0.13. Suppose f : D → R is a function and c ⇐ D such that lim f ( x ) exists. We say f is
x →c
continuous at c if
OR
lim f ( x ) = f (c) ∝===↓ lim f (c + h) = f (c).
x →c h →0

And we will say f is continuous on D, if f is continuous at each point c ⇐ D.

Here we must make a note of one point how limit penetrates a continuous function. f : D → R is
continuous at c ⇐ D. If ( xn ) is any seq such that xn ⇐ D for all n, xn ↔= c for all n ⇐ N and lim xn = c
n→∞
then - .
lim f ( xn ) = f lim xn .
n→∞ n→∞

Algebra of Continuous functions

If f and g are continuous at c, then
1. f ± g is continuous at c.
2. f · g is continuous at c.
- .
f
3. Moreover, if g(c) ↔= 0, then g is continuous at c.

Example 4.0.14. 1. As shown in the limit of functions section: a) constant functions, b) Identity functions, c)
all type of polynomial functions, d) rational functions and e) square-root functions are all continuous at each
point of their respective domains.

2. Trigonometric sine function: the domain is R. Suppose c ⇐ R be a point. Then using algebra of limits we
have the following deductions:
lim sin (c + h) = lim [sin c cos h + cos c sin h]
h →0 h →0
= lim sin c cos h + lim cos c sin h = sin c
h →0 h →0

45
 Therefore the trigonometric sine function is continuous on R. So whatever be the graph of sine, we can
draw it without lifting the pen (pencil). By the same logic one can show that the trigonometric cosine (cos)
function is continuous on R.


 x sin 1 ,
x x ↔= 0 1
3. If f ( x ) = . Then from earlier calculations we know lim x sin = 0. Now f is continuous

 x →0 x
a, x=0
at 0 only if a = 0.

4. f : R → R defined by: 

x if x ⇐ Q
f (x) =

0 if x ⇐ R \ Q.

Let c ⇐ R be a non- zero real number.

(a) We know we have a seq of rational nos. (qn ) such that qn ↔= c and lim qn = c
n→∞
(b) also we know we have a seq of irrational nos. (in ) such that in ↔= c and limn→∞ in = c

Then
lim f (qn ) = lim qn = c ↔= 0 = lim 0 = lim f (in )
n→∞ n→∞ n→∞ n→∞

Therefore f is not continuous at any non zero c, as the limit of the function does not exists at any non-zero
point.

Now let c = 0, and ( xn ) be any seq with xn ↔= 0 for all n ⇐ N and lim xn = 0, then from definition we
n→∞
know either f ( xn ) = xn or f ( xn ) = 0. Hence 0 ↗ | f ( xn )| ↗ | xn | for all n ⇐ N

Squeeze theorem following definition of limit

===========↓ lim f ( xn ) = 0 ================↓ lim f ( x ) = 0 = f (0).
n→∞ x →0

Therefore f is continuous at 0.

Q: Does there exists a function

 f : R → R which is continuous at every irrational
 points but discontinuous
at every rational points? Hint: If your answer is “No”, then think again.

Continuity under composition of functions

Theorem 4.0.15. If g is continuous at c and f is continuous at g(c), then f ↖ g is continuous at c. Notice f

is needed to be continuous at g(c) not at c.

Proof. Given lim g( x ) = g(c) and lim f (y) = f ( g(c)). We need to show:
x →c y→ g(c)

lim ( f ↖ g)( x ) = ( f ↖ g)(c) ∝==↓ lim f ( g( x )) = f ( g(c)).

x →c x →c

Let ( xn ) be a seq with xn ↔= c for all n ⇐ N and lim xn = c. Then lim g( xn ) = g(c). Further we know
n→∞ n→∞
f is continuous at g(c) and ( g( xn )) is a seq converging to g(c) so we know lim f ( g( xn )) = f ( g(c)).
n→∞
Here notice, we have proved that for any arbitrary seq ( xn ) with xn ↔= c for all n and lim xn = c, one
n→∞
will have lim f ( g( xn )) exists and equals a fixed value f ( g(c)). Therefore lim f ( g( x )) exists and
n→∞ x →c

lim f ( g( x )) = f ( g(c))
x →c

↓ lim ( f ↖ g)( x ) = ( f ↖ g)(c).

x →c

Hence f ↖ g is continuous at c.

46
Application:?? Using repeated application of this theorem one can verify that

 ( )
x2 + 5x + 3 x3 + 5
f (x) = sin +
9 x2 + 1

an expression like this is continuous at any appropriate point c where the f is defined.

Lemma 4.0.16 (Sign Preserving Lemma). 1. Suppose f is continuous at c, and f (c) > 0. Then there
exists a δ > 0 such that f ( x ) > 0 for all c ↑ δ < x < c + δ.
2. Suppose f is continuous at c, and f (c) < 0. Then there exists a δ > 0 such that f ( x ) < 0 for all
c ↑ δ < x < c + δ.

The graphical representation is the following:

y
y
f
(c, f (c)) c↑δ c+δ
c
x

x
c
c↑δ c+δ (c, f (c))
f

Proof of 1: Here we are going to discuss the proof of 1, but the proof of the other point will be same. Let
f is continuous at c, and f (c) > 0.

Assume that the conclusion above is false: so for every δ > 0 there exists a point
( xδ ⇐ (c ↑ δ,)c + δ)
1 1
such that f ( xδ ) ↗ 0. Therefore for every n ⇐ N we can say that there exists xn ⇐ c ↑ , c + such
n n
that f ( xn ) ↗ 0.

( )
1 1
As xn ⇐ c↑ ,c+ for each n ⇐ N, therefore lim xn = c, from Squeeze theorem. Therefore
n n n→∞
from continuity of f at c and corollary 3.0.13, we have

f (c) = lim f ( xn ) ↗ 0 [contradiction]

n→∞

Now we are going to state three big theorems and these theorems have far reaching effect in
understanding the graph of continuous functions. Unfortunately, to discuss the proof of the theorems we
need go deep into real analysis, which may divert us from our goal. So, I think only to state the theorems
and show you how powerful they are.

47
 y
y = f (x)

f (b) > 0
Theorem 4.0.17 (Intermediate Value Theorem).
If f is continuous on its domain and a, b are two points from
the domain such that a < b and f ( a) < 0 < f (b), then there is
some s ⇐ ( a, b) such that f (s) = 0.
x
a b
Geometrically: the graph of f , which starts below the x-axis
and ends above x-axis, must cross this axis at some point.

f ( a) < 0

Q: If f is continuous on R and a < b are two numbers such that f ( a) > L > f (b). Then show that
there exists a number s ⇐ ( a, b) such that f (s) = L.

Q: Suppose f : ( a, b) → R is a continuous function. Notice that the domain here is an open interval.
Is every continuous function defined on open interval bounded?

y
y=M
y = f (x) Theorem 4.0.18 (Boundedness Theorem).
If f is continuous on [ a, b], then f is bounded above as well as
bounded below on [ a, b], that is ⇓m, M ⇐ R such that

x m ↗ f (x) ↗ M for all x ⇐ [ a, b].

a b

Geometrically: The graph of f on [ a, b] lies below some hori-

zontal line and lies above some horizontal line.
y=m

Theorem 4.0.19 (Extreme Value Theorem). If f is continuous on [ a, b], then there are some numbers
c1 , c2 ⇐ [ a, b] such that f ( x ) ↗ f (c1 ) for all x ⇐ [ a, b] and f (c2 ) ↗ f ( x ) for all x ⇐ [ a, b] i.e.

f (c1 ) = sup{range( f )}

f (c2 ) = inf{range( f )}

Geometrically: We know the graph will lie in between two horizontal lines. It says that the graph
should touch the lowest possible and highest possible such horizontal lines.

48
 y

y = f (x)
(c1 , f (c1 ))

(c2 , f (c2 ))
x
a c1 c2 b

Theorem 4.0.20. If n is odd then any polynomial with expression P( x ) = x n + an↑1 x n↑1 + · · · + a1 x + a0
has a root i.e. there exists some α ⇐ R such that P(α) = 0.

Sketch of the proof: We know that the polynomial function is continuous everywhere and to solve this
prove this type of result one way is to somehow invoke IVT. The given polynomial can be written as:
- a a1 a0 .
P ( x ) = x n 1 + n ↑1 + · · · + n ↑ +
x x 1 xn
* *
* a n ↑1 a1 *
a0 * |a | | a1 | | a0 |
*
Using triangle inequality we have: * x + · · · + xn↑1 + xn * ↗ |nx↑|1 + · · · + +
| x n ↑1 | | xn |

If we choose x ⇐ R such that | x | > max{1, 2n| an↑1 |, 2n| an↑2 |, . . . , 2n| a1 |, 2n| a0 |} then the above
inequality changes to:

* *
* a n ↑1 a1 a0 ** 1
* (why!!!)
* x + · · · + x n ↑1 + x n * ↗ 2

1 a a1 a 3
↓ ↗ 1 + n ↑1 + · · · + n ↑ + 0n ↗
2 x x 1 x 2

Using the above inequality with the fact that n is odd we have the following:


 xn 3x n
if x > 0 and | x | satisfies the "above" inequality then : 0< ↗ P( x ) ↗

 2 2



if x < 0 and | x | satisfies the "above" inequality then : 3x n xn
↗ P( x ) ↗ <0
2 2

Now suitably apply IVT to justify that the theorem is true.

Note that the above proof may seem artificial, so drawing schematic pictures of each argument is strongly
recommended in order to under the proof.

Q: Show that any odd degree polynomial has a real root.

49
 if and only if
• lim f ( x ) = +∞ (↑∞) ∝======↓ for any sequence ( xn ) such that lim xn = a and
x → a+ n→∞
xn > a for all n, we have lim f ( xn ) = +∞ (↑∞).
n→∞

if and only if
• lim f ( x ) = +∞ (↑∞) ∝======↓ for any sequence ( xn ) such that lim xn = a and
x → a↑ n→∞
xn < a for all n, we have lim f ( xn ) = +∞ (↑∞).
n→∞

if and only if
• lim f ( x ) = l ∝======↓ for any sequence ( xn ) such that lim xn = ∞ we
x →∞ n→∞
have lim f ( xn ) = l.
n→∞

if and only if
• lim f ( x ) = l ∝======↓ for any sequence ( xn ) such that lim xn = ↑∞ we
x →↑∞ n→∞
have lim f ( xn ) = l.
n→∞

Limits Involving Infinity

• A line y = L is a horizontal asymptote of y = f ( x ) if either

lim f ( x ) = L or lim f ( x ) = L.
x →∞ x →↑∞

• A line x = x0 is a vertical asymptote of y = f ( x ) if either

lim f ( x ) = ±∞ or lim f ( x ) = ±∞.

x → x0↑ x → x0+

1 1
Example 4.0.21. Show that lim = 0 and lim = 0.
x →∞ x x →↑∞ x

Let " > 0 be given. Then: * *

*1 * 1
* ↑ 0* < " if |x| > .
*x * "
1
Choosing M = " gives the result.

x2 + x + 2
Example 4.0.22. Find lim .
x →∞ 2x 2 ↑ x + 5

Divide numerator and denominator by x2 :

1 2
1+ x + x2 1+0+0 1
lim 1 5
= = .
x →∞ 2↑ x + 2↑0+0 2
x2

1 1
Example 4.0.23. Show that lim = ∞ and lim = ↑∞.
x →0+ x x →0 x
↑

For B > 0,
1 1
>B for 0 < x < .
x B
1
Choose δ = B to show the first limit. Similarly for the second.

50
Example 4.0.24. Find horizontal and vertical asymptotes of y = 1x .
The horizontal asymptote is y = 0 since
1
lim = 0,
x →±∞ x
and the vertical asymptote is x = 0 since
1 1
lim = ∞, lim = ↑∞.
x →0+ x x →0↑ x

Exponential function
1
Suppose a > 0. Now first, we want to make sense of the notation a x for any x ⇐ R. We know what is a 2 ,
1 1 p
a 3 , ..., a n ...; similarly we can calculate a p/q for any q ⇐ Q. A good question will be: What is ai , where
i ⇐ R \ Q (say i = ε).
For a fixed irrational number i, we know from the density property of Q, there is a sequence qn of
numbers such that qn ⇐ Q for all n ⇐ N and lim qn = i. Then from a property of sequences we will
n→∞
have: + ,
lim qn
n→∞ qn
ai = a = lim a .
n→∞

So for any real number x: a x makes sense. Hence we can define the exponential function for a real
number a > 0 (we assume a ↔= 1):
f : R ↑→ R : f ( x ) = a x .

Now if y ⇐ R is in Range( f ) then ⇓ x ⇐ R such that f ( x ) = y ↓ a x = y.

If y ↗ 0 then a x ↗ 0 which would imply a ↗ 0 (contradiction)
↓ y > 0 ↓ f : R ↑→ (0, ∞) : f ( x ) = a x .
1
Is the function one- one? Let x ↔= y and suppose f ( x ) = f (y). Then a x↑y = 1 ↓ a = 1 x↑y = 1
(contradiction, since a ↔= 1). So f is injective.
Now suppose xn is a sequence such that lim xn = c. Then
n→∞

lim xn
lim f ( xn ) = lim a xn = an→∞ = ac
n→∞ n→∞

This implies f is continuous at c ⇐ R. ↓ f : R ↑→ (0, ∞) is one-one and continuous. One can easily
show that the function is strictly increasing.
Now suppose a > 1, then
lim an = ∞ and lim a↑n = 0
n→∞ n→∞
Therefore the graph will look like:

y ax , a > 1

51
 The function f : R → (0, ∞) : f ( x ) = a x ( a > 1) is a continuous, bijective function, called general
exponential function. The inverse f ↑1 is called “log” function whose base is a.

loga : (0, ∞) → R

x △→ loga ( x )
So the graph is like:
y
loga x, a > 1

x
1

Now for the time suppose the following inequality is correct:

1
′x ⇐ R : 1 + x ↗ ex and ex ↗ for allx < 1
1↑x
We analyse two cases:
e x ↑1 1
• For 0 < x < 1 we have 1 ↗ x ↗ 1↑ x

1 e x ↑1
• For x < 0 we have 1↑ x < x <1

Hence using Sandwich lemma:

ex ↑ 1 ex ↑ 1
lim = 1 = lim
x →0+ x x →0↑ x

ex ↑ 1
↓ lim =1
x →0 x

Notice that for a > 0 : f ( x ) = e x and g( x ) = ln( x ) are inverse functions of each other:
x
a x = eln a = e x ln a
ax ↑ 1 e x ln a ↑ 1
↓ lim = lim
x →0 x x →0 x
+ ,
e x ln a ↑ 1
= lim · ln a
x →0 x ln a
= 1 · ln a = ln a

ax ↑ 1
↓ lim = ln a
x →0 x

52
Chapter 5

Differentiation

Derivatives:
The derivative of a function is the first of the two major concepts of Calculus. While it is true that
the concept of a function is fundamental, that you cannot do anything without limits or continuity;
everything we have done until now has been preparation.



 | x |, x≃0
f (x) =

 /
x2 , x↗0 f (x) = |x| f (x) = |x|

A B C D

The figures A, B, C illustrate certain types of misbehavior which continuous functions can display.
The graphs of these functions are bent at (0, 0), unlike the graph of figure D, where it is possible to draw
a “tangent” line at each point.

y
Q: So the question is what is a tangent line?
f (x)
A promising approach to the definition of a tangent line might (c + h, f (c + h))
start with "secant lines", and use the notion of limits. If h ↔=
0, then the two distinct points (c, f (c)) and (c + h, f (c + h))
determines a straight line whose slope is
Secant lines
f (c + h) ↑ f (c) f (c + h) ↑ f (c)
.
h
Now see if we tend h → 0 then what will happen to the picture?
Tangent
We have never before talked about limit of lines, but we can
talk about limit of their slopes: The slope of the tangent line (c, f (c))
through (c, f (c)) should be h
f (c + h) ↑ f (c)
lim x
h →0 h

53
 Note 5.0.1. What ever is the mathematical meaning of differentiability at a point means, in picture that should
represent that the graph of the function is “smooth” at that point. Which is same as saying a well-defined
unique tangent line to the graph at the that point.

Definition 5.0.2. Suppose f : [ a, b] → R is a function and c ⇐ ( a, b). We say f is differentiable at c if the

limit
f (c + h) ↑ f (c)
lim exists.
h →0 h
In this case the limit is denoted by f ∈ (c) and is called derivative of f at c. We say f is differentiable if it is
differentiable at every point of the domain.

Remark 5.0.3. In the above definition f ∈ (c) is a real number, but a careful look will tell us that if f : [ a, b] →
R is differentiable function then we can think of another function f ∈ : [ a, b] → R which is called the derivative
function of f . Similarly we can talk about higher derivative, when it makes sense i.e. f ∈∈ , f ∈∈∈ , . . . etc.

Example 5.0.4. (a) Suppose f is a constant function: f ( x ) = a

then we have at any real number c ⇐ R
y
f (c + h) ↑ f (c) a↑a
lim = lim =0
h →0 h h →0 h

Hence at every point c, the function f is differentiable and f ∈ (c) = 0 (c, a = f (c))
for all c ⇐ R. This means that the tangent line to the graph of f a f (x) = a
Tangent line
always has slope 0, so the tangent line always coincide with the graph.

(b) Constant function are not the only ones whose graph coincide
with their tangent lines. x
c
f (c + h) ↑ f (c)
f ( x ) = mx + a ↓ lim = m = f ∈ ( c ).
h →0 h

Example 5.0.5. f : R → R is given by f ( x ) = x2 . Let c ⇐ R then

y f (c + h) ↑ f (c) ( c + h )2 ↑ c2
lim = lim
h →0 h h →0 h
f ( x ) = x2 2ch + h2
= lim
h →0 h
g( x )
= lim (2c + h) = 2c.
h →0

So at each point c, f is differentiable and f ∈ (c) = 2c.

Note that: in this picture the tangent line appears to intersect the graph for
(c, c2 ) only once. Since the tangent line through (c, c2 ) has slope 2c, the line is
basically the graph of the function:

x g( x ) = 2c( x ↑ c) + c2 = 2cx ↑ c2

Now if the graph of f and g intersect at a point ( x, f ( x )) = ( x, g( x )),

then:
Slope 2c x2 = 2cx ↑ c2 ↓ ( x ↑ c)2 = 0 ↓ x = c
In other words the point of intersection is only (c, c2 ).

54
Example 5.0.6. Let us see another example similar to this. f : R → R is given by f ( x ) = x3 . And let c ⇐ R be
any point, therefore:

f (c + h) ↑ f (c) ( c + h )3 ↑ c3 3c2 h + 3ch2 + h3

lim = lim = lim = lim (3c2 + 3ch + h2 ) = 3c2 .
h →0 h h →0 h h →0 h h →0

Thus f is again differentiable at every point and f ∈ (c) = 3c2 . y

Thus the slope of the tangent line to the graph of f at (c, c3 ) has f ( x ) = x3
slope 3c2 . This means that the tangent line is graph of

g( x ) = 3c2 ( x ↑ c) + c3 = 3c2 x ↑ 2c3

(c, c3 )
The graphs of f and g intersect at the point ( x, f ( x )) =
x
( x, g( x )) when

x3 = 3c2 x ↑ 2c3
↓ x3 ↑ 3c2 x + 2c3 = 0
↓ ( x ↑ c)2 ( x + 2c) = 0

So the tangent line through (c, c3 ) also intersects the graph at

the point (↑2c, ↑8c3 ). These two points are always distinct,
except when c = 0.

Let us discuss some non-examples i.e. figures A, B, C from the beginning of this chapter.

1) f ( x ) = | x |. In this case: 2) f : R → R is given by 3) f : R → R is given by

 
f (0 + h ) ↑ f (0) |h| 
 x2 , x ↗ 0 f (x) = |x|
=
h h f (x) =


Now, x, x ≃ 0
f (0 + h ) ↑ f (0)
f (0 + h ) ↑ f (0) • lim
• lim =1 h → 0+ h
h →0 + h f (0 + h ) ↑ f (0) ↘
• lim h↑0
(right-hand derivative) h → 0+ h = lim
h↑0 h →0+ h
f (0 + h ) ↑ f (0) = lim = 1. 1
• lim = ↑1 h →0 + h = lim ↘
h → 0↑ h h →0+ h
(left-hand derivative) f (0 + h ) ↑ f (0)
• lim f (0 + h ) ↑ f (0)
h → 0↑ h • lim
f (0 + h ) ↑ f (0) h
↓ lim does h2 ↑ 0 h → 0↑
↘
h →0 h = lim = 0. ↑h
not exist. h →0↑ h = lim
h →0 ↑ h
↓ f ( x ) = | x | is not differen- f (0 + h ) ↑ f (0) ( )
Hence lim 1
tiable at 0. h →0 h = lim ↑ ↘
again does not exist.
h →0↑ ↑h
Q: If c ↔= 0, then what can
you say about differentiability So f is not differentiable at 0. Therefore in this situation
of f at c? both the left and right hand
limits do not exist, so the func-
tion is not differentiable at 0.

Theorem 5.0.7. f : [ a, b] → R is differentiable at c ⇐ ( a, b). Then f is continuous at c.

55
Proof.
f (c + h) ↑ f (c)
lim [ f (c + h) ↑ f (c)] = lim ·h
h →0 hh →0
( )( )
f (c + h) ↑ f (c)
= lim lim h = f ∈ (c) ∞ 0 = 0
h →0 h h →0

So f is continuous at c.

Algebra of derivatives
1. Suppose f and g are differentiable at c. Then f ± g is also differentiable at c.

( f ± g)(c + h) ↑ ( f ± g)(c) f (c + h) ↑ f (c) g(c + h) ↑ g(c)

lim = lim ± lim = f ∈ (c) ± g∈ (c)
h →0 h h →0 h h →0 h

↓ ( f ± g)∈ (c) = f ∈ (c) ± g∈ (c)

2. If f and g are differentiable at c, then so is ( f · g) and we have the formula:

( f · g ) ∈ ( c ) = f ∈ ( c ) g ( c ) + f ( c ) g ∈ ( c ).

( f · g)(c + h) ↑ ( f · g)(c) f (c + h) g(c + h) ↑ g(c + h) f (c) + g(c + h) f (c) ↑ f (c) g(c)

lim = lim
h →0 h h →0 h
   
f (c + h) ↑ f (c) g(c + h) ↑ g(c)
= lim g(c + h) + lim f (c)
h →0 h h →0 h
 
f (c + h) ↑ f (c) g(c + h) ↑ g(c)
= lim ∞ lim g(c + h) + lim f (c)
h →0 h h →0 h →0 h
= f ∈ ( c ) g ( c ) + g ∈ ( c ) f ( c ).

Theorem 5.0.8. If f ( x ) = x n for some natural no n, then f ∈ (c) = ncn↑1 for all c.

Proof. The proof will be done using PMI on n. The n = 1 case is done above. Now assume that the
theorem is true for n, so that if f ( x ) = x n , then f ∈ (c) = ncn↑1 for all c. Let g( x ) = x n+1 .
If I ( x ) = x, then g( x ) = f ( x ) · I ( x ). Hence

g∈ (c) = f ∈ (c) I (c) + f (c) I ∈ (c) = ncn↑1 · c + cn · 1 = (n + 1)cn for all c ⇐ R.

Hence from PMI, we can say that the theorem is true for all n ⇐ N.

Remark 5.0.9. If P( x ) = an x n + an↑1 x n↑1 + · · · + a1 x + a0 is any polynomial with an ↔= 0, then repeated

applications of the above theorems will give us the following:

P∈ ( x ) = nan x n↑1 + (n ↑ 1) an↑1 x n↑2 + · · · + a1

P∈∈ ( x ) = n(n ↑ 1) an x n↑2 + (n ↑ 1)(n ↑ 2) an↑1 x n↑3 + · · · + 2a2

..
.
P(n) ( x ) = n!an
P(k) ( x ) = 0 for k > n.

56
 Theorem 5.0.10. If g is differentiable at c and g(c) ↔= 0, then 1/g is differentiable at c and
( )∈
1 ↑ g∈ (c)
(c) =
g ( g(c))2

( 1g )(c + h) ↑ ( 1g )(c)
Proof. Before we even write , we must be sure that this expression makes sense i.e. it
h
is necessary to check that ( 1g )(c + h) is defined for all sufficiently small h.
Since g is differentiable at c ↓ g is continuous at c. Again g(c) ↔= 0, then the sign preserving
theorem for continuous functions says that there exists δ > 0 such that g( x ) and g(c) have same sign
′ x ⇐ (c ↑ δ, c + δ) OR g(c + h) and g(c) have same sign ′h ⇐ (↑δ, δ).
Therefore the above quotient does make sense.
- . - .
1 1
g ( c + h ) ↑ g (c) g(c) ↑ g(c + h)
lim = lim
h →0 h h→0 h [ g ( c ) g ( c + h )]
( )
g(c + h) ↑ g(c) 1 g∈ (c)
= ↑ lim ∞ =↑ .
h →0 h g(c) · lim g(c + h) ( g(c))2
h →0

Therefore applying the previous theorem we have the following result:

f
Theorem 5.0.11. If f and g are differentiable at c and g(c) ↔= 0, then is differentiable at c and
g
( )∈
f g(c) f ∈ (c) ↑ f (c) g∈ (c)
(c) =
g ( g(c))2

Derivative of Trigonometric functions

Example 5.0.12. We know that trigonometric sine function is continuous at all points. So talking about its
derivative makes sense. Let c ⇐ R.

- .
sin(c + h) ↑ sin(c) 2 sin(h/2) cos 2c2+h
lim = lim
h →0 h h →0 h
  ( )
sin(h/2) 2c + h
= lim lim cos = 1 ∞ cos(c) = cos(c)
h →0 h/2 h →0 2
sin ( x )
The last but one equality follows from the result on lim = 1 and the trigonometric cosine function is
x →0 x
continuous on R. Therefore at all c ⇐ R, sin is differentiable and [sin(c)]∈ = cos(c).

Example 5.0.13. Similarly we know cosine function is also continuous at each point c ⇐ R.
- .
cos(c + h) ↑ cos(c) ↑2 sin(h/2) sin 2c2+h
lim = lim
h →0 h h →0 h
  ( )
sin(h/2) 2c + h
= ↑ lim lim sin = ↑ sin(c)
h →0 h/2 h →0 2
Therefore at all c ⇐ R, cos is differentiable and [cos(c)]∈ = ↑ sin(c).

57
 Question 5.0.14. Q: Use the following trigonometric identities to find out the differentiability of other
trigonometric functions:

sin( x ) 1 1 1
tan( x ) = , sec( x ) = , csc( x ) = , cot( x ) =
cos( x ) cos( x ) sin( x ) tan( x )

Remember sin & cos is continuous and differentiable everywhere.

sin( x + 2ε ) = sin( x )

cos( x + 2ε ) = cos( x ) ′ x ⇐ R.
Figure out domain of tan, sec, csc & cot. Then find out the points of continuity of these functions and then
figure out at which points the functions are differentiable and what are their derivatives (from definition).

Chain Rule
Differentiating the most interesting fn obviously requires a formula for ( f ↖ g)∈ ( x ) in terms of f ∈ and g∈ .
Recall we have studied a formula when a composition of functions is again continuous. We now want to
do the same for differentiation. We will now prove the following theorem: (The proof is kind of a hand
wavy algebraic proof)

Theorem 5.0.15 (Chain Rule). If g is differentiable at c and f is differentiable at g(c), then f ↖ g is

differentiable at c and
( f ↖ g)∈ (c) = f ∈ ( g(c)) · g∈ (c).

Proof. Let us assume k = g(c + h) ↑ g(c). Since g is also continuous at c so that lim g(c + h) = g(c),
h →0
therefore as h → 0 this will imply k → 0.

( f ↖ g)(c + h) ↑ ( f ↖ g)(c)
lim
h →0 h
 
f ( g(c + h)) ↑ f ( g(c)) g(c + h) ↑ g(c)
= lim ∞
h →0 g(c + h) ↑ g(c) h
  
f ( g(c) + k ) ↑ f ( g(c)) g(c + h) ↑ g(c)
= lim lim
k →0 k h →0 h
= f ∈ ( g(c)) g∈ (c)
Therefore ( f ↖ g) is differentiable at c and

( f ↖ g)∈ (c) = f ∈ ( g(c)) g∈ (c)

Some more differentiable functions

Example 5.0.16. Let α ⇐ R then consider the function f : R → R given by: f ( x ) = x α .

Case 1: Let α = 0 then f ( x ) = x0 = 1 is a constant function, so f is differentiable at each point of R.

Case 2: α > 1 and take c ⇐ R be a point. Let us check whether f is differentiable at ’c’.
- .α
f (c + h) ↑ f (c) (c + h)α ↑ cα 1 + hc ↑ 1
lim = lim = cα lim
h →0 h h →0 h h →0 h

58
 Recall Generalized Binomial Theorem: If α ⇐ R and | h| < 1 then

α(α ↑ 1) 2 α(α ↑ 1)(α ↑ 2) 3

(1 + h)α = 1 + αh + h + h +...
2! 3!
* *
* *
Now in the above limit since we are taking h → 0 then "eventually we will have * hc * < 1". Therefore using
the generalized binomial theorem we have:
( - .2 - .3 )
α ( α ↑1) α(α↑1)(α↑2)
1 + α hc + 2!
h
c + 3!
h
c +... ↑1
f (c + h) ↑ f (c)
lim = cα lim
h →0 h h →0 h
 
α α ( α ↑ 1) h α(α ↑ 1)(α ↑ 2) h2
= cα lim + 2
+ +...
h →0 c 2! c 3! c3
α
= cα · = αcα↑1
c

Since α ↑ 1 > 0 so cα↑1 ⇐ R ′c ⇐ R. Therefore f is differentiable at each point of R.

Case 3: α < 1 and c ⇐ R. Then again the previous case’s calculation will show

f (c + h) ↑ f (c)
lim = αcα↑1
h →0 h

f (c + h) ↑ f (c) α
↓ lim = 1↑ α
h →0 h c

Since α < 1 ↓ α ↑ 1 < 0 ↓ 1 ↑ α > 0. Now if c = 0, then the above limit does not exist. Hence in this
case f ( x ) = x α is differentiable at each point of R except at 0.

Example 5.0.17. P( x ) = an x n + an↑1 x n↑1 + · · · + a1 x + a0 is a polynomial function: We know that any

polynomial function is continuous at each point of R. Consider functions f i ( x ) = ai xi . Then P( x ) = ∑in=0 f i ( x ).
Let c ⇐ R. As the function f i is basically multiplication of two functions; one is constant function ai and the
other is xi . Therefore using the product rule we can say f i is differentiable at c:

f i∈ (c) = ai · i · ci↑1 for i = 1, 2, . . . , n and f 0∈ (c) = 0.

Therefore using addition rule for derivative we have

n
P∈ (c) = ∑ iai ci↑1
i =1

Hence polynomial function is differentiable everywhere.

Example 5.0.18. Same for quotient fns, where they are defined.

Example 5.0.19. f : R → R given by f ( x ) = e x . Let c ⇐ R, then

f (c + h) ↑ f (c) ec+h ↑ ec eh ↑ 1
lim = lim = ec lim = ec · 1 = ec
h →0 h h →0 h h →0 h

f is differentiable at each point of R.

59
Example 5.0.20. 5. f : (0, ∞) → R given by f ( x ) = ln( x ). Then Suppose this inequality is true:
1 ex
e · e↑ x ↗ x ↗ for x > 0
e
1
↓ e1↑ x ↗ x ↗ e x ↑1 for x > 0
1
We know that ln is monotonically increasing function. Therefore: ln(e1↑ x ) ↗ ln( x ) ↗ ln(e x↑1 )
1 x↑1
↓ 1↑ ↗ ln( x ) ↗ x ↑ 1 ↓ ↗ ln( x ) ↗ x ↑ 1
x x
Case 1: Now if x ↑ 1 > 0 then
1 ln( x )
↗ ↗1
x x↑1
Let us use the substitution y = x ↑ 1 then
1 ln(1 + y)
↗ ↗1
y+1 y
Since x ↑ 1 > 0 ↓ x > 1 ↓ y > 0, now apply limit in the above inequality and apply squeeze theorem:
ln(1 + y)
lim =1
y →0+ y
Case 2: If x ↑ 1 < 0 then from inequality we have
ln( x ) 1
1↗ ↗
x↑1 x
Apply the substitution y = x ↑ 1 and let y → 0↑ and apply squeeze theorem:
ln(1 + y)
lim =1
y →0↑ y
Combining Case 1 & Case 2 we have
ln(1 + y)
lim = 1.
y →0 y
Take c ⇐ (0, ∞), then - .
ln(c + h) ↑ ln(c) ln 1 + hc 1
lim = lim h
·
h →0 h h →0
c
c
 - .
ln 1 + hc

= lim  1 = 1 · 1 = 1.
h →0 h c c c
c c

Hence f ( x ) = ln( x ) is differentiable at each c ⇐ (0, ∞).

Example 5.0.21. Consider the function

 - .

sin 1
x if x ↔= 0
f (x) =


0 if x = 0
In assignment of continuity you have proved that this function is not continuous at ’0’, so the function is not
differentiable at 0.
1
Suppose c ⇐ R \ {0}. Consider two functions g( x ) = sin( x ) & h( x ) = x
↓ f ( x ) = ( g ↖ h)( x ) = g(h( x ))
Since c ⇐ R \ {0}, therefore h is differentiable at c. Again g is differentiable at every point of R, so applying chain
rule at the point ’c’ to the function f we have
( )( ) ( )
∈ ∈ ∈ 1 1 1 1
f (c) = g (h(c)) · h (c) = cos ↑ 2 = ↑ 2 cos .
c c c c

60
Example 5.0.22. Consider the function
 - .

 x sin 1
x , x ↔= 0
f (x) =


0 ,x = 0

This function is continuous everywhere (previous assignment).

- .
h sin 1h ( )
f (0 + h ) ↑ f (0) 1
lim = lim = lim sin
h →0 h h →0 h h →0 h
If you did it correctly in assignment, then you know the above limit does not exist. f is not differentiable at ’0’.
Q: What about differentiability at other points?

 - .

 x2 sin 1
x , x ↔= 0
Example 5.0.23. f ( x ) =


0 ,x = 0
( )
f (0 + h ) ↑ f (0) 1
lim = lim h sin =0
h →0 h h →0 h

f is differentiable at ’0’.

Example 5.0.24. f ( x ) = esin(ln( x)) ? Why (0, ∞) will be domain of the function. Take c ⇐ (0, ∞). Consider
functions:

g : (0, ∞) → R ↭ g( x ) = ln( x )
h : R → R ↭ h( x ) = sin( x )
k : R → R ↭ k( x) = ex

Then f ( x ) = (k ↖ h ↖ g)( x ) = k (h( g( x ))).

We know
i) ln( x ) is differentiable at c

ii) sin( x ) is differentiable at ln(c) (why?)

iii) e x is differentiable at sin(ln(c)).

Applying Chain Rule, we can conclude that f is differentiable at ’c’ and the derivative is

1
f ∈ (c) = k∈ (h( g(c))) · h∈ ( g(c)) · g∈ (c) = esin(ln(c)) cos(ln(c)) ·
c

61
62
Chapter 6

Significance of the Derivative

In this section, we will try to justify, why learning to find derivative of a function was so important i.e.
knowing a function’s derivative f ∈ , tells you a lot about the original function.

Definition 6.0.1. Let f be a function and A ⇑ Dom( f ). A point x ⇐ A is a maximum point for f on A if
f (y) ↗ f ( x ) for every y ⇐ A. The number f ( x ) itself is called the maximum value of f on A.

Definition 6.0.2. Let f be a function and A ⇑ Dom( f ). A point x ⇐ A is a minimum point for f on A if
f ( x ) ↗ f (y) for every y ⇐ A. The number f ( x ) itself is called the minimum value of f on A.

Note 6.0.3. f has minimum on A at x if (↑ f ) has maximum on A at x.

Theorem 6.0.4. Let f be any function defined on ( a, b). If c is a maximum (minimum) point for f on ( a, b)
and f is differentiable at c then f ∈ (c) = 0.

Proof. This picture is sufficient to prove the theorem (why!). Note

that secants drawn through points to the left of (c, f (c)) have slopes
≃ 0 and in the reverse direction the slope is ↗ 0.
Formally: If h is any number such that c + h ⇐ ( a, b)

↓ f (c) ≃ f (c + h) (why!)

Thus if h > 0 then

f (c + h) ↑ f (c) f (c + h) ↑ f (c)
↗ 0 ↓ lim ↗ 0.
h h → 0+ h

Again if h < 0 then

f (c + h) ↑ f (c) f (c + h) ↑ f (c)
≃ 0 ↓ lim ≃0
h h →0 ↑ h
f (c+ h)↑ f (c)
Since f is differentiable at c ↓ limh→0 h exists.

f (c + h) ↑ f (c) f (c + h) ↑ f (c)
0 ↗ lim = lim ↗0
h →0↑ h h →0 + h

↓ f ∈ (c) = 0.

63
 Definition 6.0.5. Let f be a function and A ⇑ Dom( f ). A point c in A is a local maximum (minimum)
point for f on A if there is some δ > 0 such that c is a maximum (minimum) point for f on A ⇔ (c ↑ δ, c + δ).

So as an application of the last theorem we can say the following will hold:

Corollary 6.0.6. If f is defined on ( a, b) and has a local maximum (minimum) at c and f is differentiable at
c, then f ∈ (c) = 0.

Remark 6.0.7. Notice that the above corollary is "SUFFICIENT" but "NOT NECESSARY". Why?
Consider the function f ( x ) = x3 . We have seen the graph of this fn and we also calculated f ∈ (0) = 0.
Although we haven’t proved rigorously till now that this function is increasing, but as the graph suggests 0 is
neither local maximum, nor local minimum.

Hence comes the new term:

Definition 6.0.8. A critical point of a differentiable function f is a number c such that f ∈ (c) = 0. The
number f (c) is called a critical value of f .

Mean Value Theorems

Earlier we have seen that if f is a constant function, then f ∈ ( x ) = 0 for all suitable x ⇐ Dom( f ). The
question is whether the converse is also true? The answer to this kind of question can be dealt with the
help of "Mean Value Theorem".

Suppose f : [ a, b] → R is continuous and is differentiable

on ( a, b). Then the quotient

(b, f (b)) f (b) ↑ f ( a)

(c, f (c)) b↑a
gives the slope of the line joining ( a, f ( a)) and (b, f (b)).
As the figure suggest: Here one could expect existence of
( a, f ( a)) a c ⇐ ( a, b) such that the slope of the tangent at (c, f (c))
f (b) ↑ f ( a)
a c b is same as .
b↑a

Theorem 6.0.9 (Particular case of MVT: Rolle’s Theorem). If f : [ a, b] → R satisfies the following
conditions:
1. f is continuous on [ a, b],
2. f is differentiable on ( a, b),

3. f ( a) = f (b).
Then there is a point c ⇐ ( a, b) such that f ∈ (c) = 0.

Proof. Since f : [ a, b] → R is continuous, so f has a maximum value and a minimum value on [ a, b].

Case 1: Suppose first that the maximum value occurs at a point c ⇐ ( a, b) ↓ f ∈ (c) = 0.

Case 2: Suppose next that the minimum value occurs at a point d ⇐ ( a, b) ↓ f ∈ (d) = 0.

64
Case 3: Finally, suppose the maximum & minimum occur at the end points. Since f ( a) = f (b), so the
maximum value is equal to the minimum value ↓ f is constant, therefore for any x ⇐ ( a, b)

f ∈ ( x ) = 0.

Q: Provide examples of functions where the conclusion of Rolle’s Theorem does not hold if any one
of the theorem’s three key conditions is not satisfied.

Theorem 6.0.10 (Mean Value Theorem: MVT). If f : [ a, b] → R is continuous & is differentiable on

( a, b), then there is a number c ⇐ ( a, b) such that

f (b) ↑ f ( a)
f ∈ (c) =
b↑a

f (b) ↑ f ( a)
Proof. Consider: h( x ) = f ( x ) ↑ ( x ↑ a). Then:
b↑a
1. h : [ a, b] → R is continuous
2. h : ( a, b) → R is differentiable
3. h( a) = h(b)
f (b) ↑ f ( a)
↓ ⇓ c ⇐ ( a, b) such that h∈ (c) = 0 (Rolle’s Thm) ↓ f ∈ (c) = .
b↑a

Corollary 6.0.11. If f is defined on an interval and f ∈ ( x ) = 0 for all x in the interval, then f is constant on
that interval.

Proof. Let a, b two points in the interval (a < b). Then ⇓ c ⇐ ( a, b) such that

f (b) ↑ f ( a)
0 = f ∈ (c) = [MVT (how?)]
b↑a

↓ f ( b ) = f ( a ).
Thus the value of f at any two points of the interval is same ↓ f is constant.
Q: How important is the assumption that f is defined on an interval in this corollary ???

Corollary 6.0.12. If f ∈ ( x ) > 0 (< 0) for all x in an interval, then f is increasing (decreasing) on the
interval.

Proof: We are doing the blue part here!! Consider f ∈ ( x ) < 0 ′ x and a < b in the interval.
f (b) ↑ f ( a)
↓ ⇓ c ⇐ ( a, b) with = f ∈ (c) < 0. Now (b ↑ a) > 0 ↓ f (b) ↑ f ( a) < 0 ↓ f (b) < f ( a).
b↑a
Hence decreasing. (Do the other part).

Theorem 6.0.13. Suppose f is twice differentiable at the point c and f ∈ (c) = 0.

1. If f ∈∈ (c) > 0 then f has a local minimum at c.
2. If f ∈∈ (c) < 0 then f has a local maximum at c.

65
Proof.
f ∈ (c + h) ↑ f ∈ (c)
f ∈∈ (c) = lim
h →0 h
Since f ∈ (c) = 0 therefore we will have
f ∈ (c + h)
f ∈∈ (c) = lim
h →0 h
f ∈ (c + h)
Suppose f ∈∈ (c) > 0, then > 0 for sufficiently small h. ↓ f ∈ (c + h) must be positive for
h
sufficiently small h > 0 and f ∈ (c + h) must be negative for sufficiently small h < 0.
Therefore from Corollary 6.0.12 we have: f is increasing in some interval to the right of c and f is
decreasing in some interval to the left of c. So locally near to c the graph of f will look like the following:

Hence the function f has local minimum at c. The other part will be done similarly.

Theorem 6.0.14. Suppose f ∈∈ (c) exists. If f has a local minimum at c, then f ∈∈ (c) ≃ 0. Similarly if f has a
local maximum at c, then f ∈∈ (c) ↗ 0.

Proof. We will do the second part. Suppose f has a local maximum at c. If f ∈∈ (c) > 0, then from previous
result, f has local minimum at c. So at ’c’, f has local maximum as well as local minimum ↓ f is constant
on a sufficiently small neighborhood around c. ↓ f ∈∈ (c) = 0 (which contradicts f ∈∈ (c) > 0). Thus at the
point c: f ∈∈ (c) ↗ 0.

Convexity & Concavity:

Definition 6.0.15. A function is concave on an interval, if for all a, b in the interval the line segment joining
( a, f ( a)) and (b, f (b)) lies below the graph of f on ( a, b).

( a, f ( a)) (b, f (b))

f (b) ↑ f ( a)
g( x ) = ( x ↑ a) + f ( a)
b↑a
So the above definition is same as saying:
f ( x ) > g( x ) ′ a<x<b
a little simplification will give us another equivalent definition of concavity of a function: We will say f
is concave on an interval if for all a, b in that interval the following inequality is true,

f ( x ) ↑ f ( a) f (b) ↑ f ( a)
> ′ a < x < b
x↑a b↑a

66
 Definition 6.0.16. A function is convex on an interval, if for all a, b in the interval the line segment joining
( a, f ( a)) and (b, f (b)) lies above the graph of f on ( a, b).

( a, f ( a)) (b, f (b))

f (b) ↑ f ( a)
g( x ) = ( x ↑ a) + f ( a)
b↑a
So the above statement is same as saying:
f ( x ) < g( x ) ′ a<x<b
a little simplification will give us

f ( x ) ↑ f ( a) f (b) ↑ f ( a)
< ′ a < x < b
x↑a b↑a

So f is convex on an interval if for any a, x, b from the interval with a < x < b we have the above
inequality.

Lemma 6.0.17. Suppose f is differentiable and f ∈ is increasing. If a < b and f ( a) = f (b) then

f ( x ) < f ( a) = f (b) for a < x < b.

Proof. Assume ⇓ x ⇐ ( a, b) with f ( x ) > f ( a) = f (b). Then the maximum of f on [ a, b] occurs at some
x0 ⇐ ( a, b) and there we will have f ∈ ( x0 ) = 0.

On the other hand, applying the MVT to the interval [ a, x0 ], we

can find x1 ⇐ ( a, x0 ) with

f ( x0 ) ↑ f ( a )
f ∈ ( x1 ) = > 0.
x0 ↑ a

As x1 < x0 and f ∈ is increasing, therefore 0 < f ∈ ( x1 ) <

f ∈ ( x0 ) = 0. Which is a contradiction to the fact that f ∈ in-
creasing. This proves that ′ x ⇐ ( a, b) one will have

f ( x ) ↗ f ( a ) = f ( b ).

Now assume ⇓ x ⇐ ( a, b) with f ( x ) = f ( a). We know that

f is not constant on [ a, x ], because if it was then f ∈ = 0 (but
f ∈ increasing). So there is some x1 ⇐ ( a, x ) with f ( x1 ) < f ( a).
Applying MVT to [ x1 , x ] we have x2 ⇐ ( x1 , x ) such that

f ( x ) ↑ f ( x1 )
f ∈ ( x2 ) = > 0.
x ↑ x1

On the other hand, f ∈ ( x ) = 0 since a local maximum occurs at

x. Again this contradicts that f ∈ increasing.
So the lemma is proved.

67
 Theorem 6.0.18. If f is differentiable and f ∈ is increasing, then f is convex.

Proof. Let a < b and define

f (b) ↑ f ( a)
g( x ) = f ( x ) ↑ ( x ↑ a)
b↑a
f (b) ↑ f ( a)
↓ g∈ ( x ) = f ∈ ( x ) ↑ ↓ g∈ is increasing.
b↑a
Further g( a) = f ( a) = g(b). So applying the above lemma, we will have g( x ) < g( a) ′ x ⇐ ( a, b)

f ( x ) ↑ f ( a) f (b) ↑ f ( a)
↓ < ′ x ⇐ ( a, b)
x↑a b↑a
Hence f is convex.

Theorem 6.0.19. If f is differentiable and the graph of f lies above each tangent line except at the point of
contact, then f is convex. (Prove this)

Note 6.0.20. f is convex on the interval I if and only if (↑ f ) is concave on the interval I.

Tracing graph of differentiable functions

Let us now glue all these information (theorems) developed above to form a big picture. We will do this
by example.
1 3
Example 6.0.21. Take the polynomial f ( x ) = x ↑ 4x2 + 12x ↑ 5.
3
↫ f is continuous everywhere
↫ f is differentiable everywhere
↫ f is odd degree polynomial with leading coefficient to
be positive so lim f ( x ) = ∞ and lim f ( x ) = ↑∞
x →∞ x →↑∞

↫ f ∈ ( x ) = x2 ↑ 8x + 12 = ( x ↑ 2)( x ↑ 6)
↫ f is increasing on (↑∞, 2) and (6, ∞).
↫ f is decreasing on the interval (2, 6).
↫ f ∈∈ ( x ) = 2( x ↑ 4)
↫ f is convex on (4, ∞) and is concave on (↑∞, 4).
↫ And the point 4 is called point of inflection for f .

Asymptotes
One can visualize (if the limit exists) lim f ( x ) = L as saying that the graph of f merge with the horizontal
x →∞
line y = L as x becomes large. This line is called a horizontal asymptote of f . In the same way, (if the
limit exists) lim f ( x ) = M also gives a horizontal asymptote, in this case the merging happens in the
x →↑∞
negative direction.

68
 Similarly, if either one-sided limit is (±) infinity at x = a, we call the line x = a a vertical asymptote
of the function.
The graph may also approach a slanted line as x → ±∞. We say that y = mx + c is a slant asymptote
of f as x → ∞ if
lim ( f ( x ) ↑ (mx + c)) = 0.
x →∞

We have the analogous definition of a slant asymptote as x → ↑∞. The concept of horizontal asymptote
is a special case of slant asymptote, when m = 0.

Theorem 6.0.22. The line y = mx + c is a slant asymptote of the function f if and only if

f (x)
lim =m and lim ( f ( x ) ↑ mx ) = c.
x →∞ x x →∞

Proof. Suppose y = mx + c is slant asymptote of f , then

lim ( f ( x ) ↑ mx ↑ c) = 0 ↓ lim ( f ( x ) ↑ mx ) = c;
x →∞ x →∞

so the second condition is satisfied. Next we also can say

( )
f (x) c
lim ↑m↑ = 0.
x →∞ x x

f (x) c f (x)
Since x → ∞, this is only possible if ↑m↑ → 0 ↓ → m. The converse is trivial.
x x x

Example 6.0.23. Let us analyse the following rational function:

x2 ↑ 2x + 2
f (x) = .
x↑1

Calculation Implication

The denominator is 0 at x = 1 Domain of the function: R \ {1}

If c ⇐ R \ {1}, then the function is continuous on R \ {1}

lim f ( x ) = f (c)
x →c

use the quotient rule for the function is differentiable at each point of R \ {1}
differentiation appropriately
1
f ( x ) = ( x ↑ 1) + • lim f ( x ) = ∞
x↑1 x →∞

• lim f ( x ) = ↑∞
x →↑∞

Solving f ∈ ( x ) = 0 0 and 2 are critical points of the function

f ∈ ( x ) > 0 when f is increasing on (↑∞, 0) and (2, ∞)

x ⇐ (↑∞, 0) ⇒ (2, ∞)

f ∈ ( x ) < 0 when x ⇐ (0, 1) ⇒ (1, 2) f is decreasing on (0, 1) and (1, 2)

69
 f ∈∈ (0) = ↑2 < 0 0 is a local maximum for f

f ∈∈ (2) = 2 > 0 2 is a local minimum for f


lim f ( x ) = ∞ 

lim ( x2 ↑ 2x + 2) = 1 x →1+ ”x = 1” is a vertical asymptote
x →1 
lim f ( x ) = ↑∞ 
x →1↑

f ∈∈ ( x ) > 0 on (1, ∞) f ∈ is increasing on (1, ∞) ↓ f is convex on (1, ∞)

f ∈∈ ( x ) < 0 on (↑∞, 1) f ∈ is decreasing on (↑∞, 1) ↓ f is concave on (↑∞, 1)

f (x)
• lim =1
x →∞x
• lim ( f ( x ) ↑ x ) = 1 y = x ↑ 1 is a slant asymptote towards +∞
x →∞

f (x)
• lim =1
x →↑∞ x
• lim ( f ( x ) ↑ x ) = 1 y = x ↑ 1 is a slant asymptote towards ↑∞
x →↑∞

Hence putting all the informations in one picture will give us the graph of the function f :

x2 ↑ 2x + 2
Figure 6.1: The graph for f ( x ) =
x↑1

70
 2 ↑ 1 e x + e↑ x
Q: Try to Analyse the functions: e↑ x , e x2 , esin ( x) , tan ( x ), cosec( x ), sec( x ), cosh ( x ) := ,
2
e x ↑ e↑ x
sinh ( x ) := . The last two function are called hyperbolic cosine function and hyperbolic sine
2
function respectively. Also analyse the following functions:
1
1. f ( x ) = x3 ↑ x 2. e↑ x , e
2 ↑
x2 3. f ( x ) = x4 ↑ 2x2

Derivative of inverse of a function

Suppose f : A → B is a function such that A, B are subsets of R. We have earlier seen that if f is one-one
and onto (Bijective map) then we have f ↑1 : B → A.
Now even if f : A → B is a continuous bijection there is absolutely no reason to expect that
f ↑1 : B → A is also continuous, when such thing happens, then those map ’ f ’ has another name called
homeomorphism (beyond our scope).
Now Question is in what sort of phenomenon we can expect f ↑1 is continuous, differentiable?
Without proof I am going to state a theorem, but you can figure out the sketch of the proof from
drawing some schematic pictures.

Theorem 6.0.24. If f is continuous and one-one on a interval then f ↑1 also continuous.

Now let us address the question of differentiability.

f ↑1 y=x
L∈

( f ( a ), a ) f

( a, f ( a))

f is graph of a one-one function with a tangent line L through ( a, f ( a)). If the entire picture is reflected
through y = x line, it shows the graph of f ↑1 and a tangent line L∈ through ( f ( a), a). If the equation of L
x↑c
is y = mx + c then the equation of L∈ has to be y = . So the slope of L and slope of L∈ are reciprocals
m
of each other.
So in terms of f we can say:

1 1
( f ↑1 )∈ ( f ( a)) = OR ( f ↑1 ) ∈ ( b ) =
f ∈ ( a) f ∈ ( f ↑1 (b))

Though the argument is fine, we have not established differentiability of f ↑1 at the point f ( a), given
f is differentiable at a.

Theorem 6.0.25. If f is a continuous one-one function defined on an interval and f ∈ ( f ↑1 ( a)) = 0, then
f ↑1 is not differentiable at a.

71
Proof. We know f ( f ↑1 ( x )) = x If f ↑1 is differentiable at a then applying chain rule we have:

f ∈ ( f ↑1 ( a)) · ( f ↑1 )∈ ( a) = 1

↓ 0 · ( f ↑1 )∈ ( a) = 1 (→▽)

Theorem 6.0.26 (Inverse function theorem). Let f be a continuous, one-one function defined on an
interval and suppose that f is differentiable at f ↑1 (b) with derivative f ∈ ( f ↑1 (b)) ↔= 0. Then f ↑1 is
differentiable at b and
1
( f ↑1 ) ∈ ( b ) = ∈ ↑1 .
f ( f (b))

Proof. Let b = f ( a), then

f ↑1 ( b + h ) ↑ f ↑1 ( b ) f ↑1 ( b + h ) ↑ a
lim = lim
h →0 h h →0 h
Now every number b + h in the domain of f ↑1 can be written as b + h = f ( a + k ) for a unique k that
only depends on h (Why?)

f ↑1 ( b + h ) ↑ a f ↑1 ( f ( a + k)) ↑ a k
lim = lim = lim
h →0 h h →0 f ( a + k ) ↑ b h →0 f ( a + k ) ↑ f ( a)
Since: b + h = f ( a + k ) ↓ f ↑1 (b + h) = a + k = f ↑1 (b) + k ↓ k = f ↑1 (b + h) ↑ f ↑1 (b). Since f ↑1 is
continuous at b (why?) ↓ as h → 0 then k → 0.
So the above limit becomes:
f ↑1 ( b + h ) ↑ f ↑1 ( b ) k 1 1
lim = lim = ∈ = ∈ ↑1
h →0 h k →0 f ( a + k ) ↑ f ( a ) f ( a) f ( f (b))

Example 6.0.27 (Application of Inverse function Theorem:). f : [↑ ε2 , ε2 ] → [↑1, 1] given by

f ( x ) = sin( x )
We know that the sin on the mentioned domain is continuous, one-one and onto so f ↑1 : [↑1, 1] → [↑ ε2 , ε2 ]
given by
f ↑1 (y) = sin↑1 (y)
exists and is also continuous on its domain. If c ⇐ (↑1, 1) then there exists some b ⇐ (↑ ε2 , ε2 ) such that

sin(b) = c ↓ b = sin↑1 (c).

Since at b = sin↑1 (c), the trigonometric sine function is differentiable and f ∈ (b) = f ∈ ( f ↑1 (c)) = cos(b) ↔= 0
↓ f ↑1 (y) = sin↑1 (y) is differentiable at c and
1 1 1
( f ↑1 )∈ (c) = (sin↑1 )∈ (c) = = =
f ∈ ( f ↑1 (c)) cos( f ↑1 (c)) cos(sin↑1 (c))
1 1
=  = ↘
1 ↑ [sin(sin↑1 (c))]2 1 ↑ c2

Hence the formula for derivative:

1
(sin↑1 )∈ ( x ) = ↘ , x ⇐ (↑1, 1).
1 ↑ x2

Q: Figure out the derivatives of cos↑1 , cosec↑1 , sec↑1 , tan↑1 , cot↑1 .

72
Implicit Differentiation
We have been studying functions whose graph can be written in the form y = f ( x ). Sometimes if we
happen to have an equation of a curve, then it may not possible to express the curve as graph of a
function. For example consider the curve with equation:

x 2 + y2 = 1
↘
Clearly for one choice of x ⇐ (↑1, 1), there are two choices of y = ± 1 ↑ x2 . So the curve defines y
implicitly in terms of x.
Now if we apply differentiation on the equation of the curve with Chain rule in mind, then we have

dy dy x
2x + 2y =0↓ =↑ if y ↔= 0.
dx dx y

Example 6.0.28 (Application: (Tangent to Ellipse)). Standard equation of ellipse is

x2 y2
+ =1
a2 b2
Implicit differentiation gives
2x 2y dy
+ 2 · =0
a2 b dx

If ( x0 , y0 ) is a point on the ellipse, the slope m of the tangent line to the ellipse at ( x0 , y0 ) is given by:

2x0 2y x b2
2
+ 20 m = 0 ↓ m = ↑ 0 2 .
a b y0 a

Hence the equation of that tangent line is

( )
x0 b2
y = y0 + ↑ ( x ↑ x0 ).
y0 a2

( x0 , y0 )

Growth comparison of polynomials and exponential functions

Let us explore how the exponential function grows with x.
ln( x )
We start with the known limits lim e x = ∞ and lim = 0. Therefore these limits helps us to
x →∞ x →∞ x
conclude
x ln(e x )
lim x = lim = 0.
x →∞ e x →∞ ex
So this limit tells us that e x grows faster than x.
Let us now see how the exponential function compares with higher powers of x. For n ⇐ N, we have:

x x/n y
lim = n lim = n lim = 0.
x →∞ e x/n x →∞ e x/n y→∞ ey

73
Hence - x .n
xn
lim = lim = 0.
x →∞ ex x →∞ e x/n

Thus, the exponential function grows faster than any powers of x and hence algebra of limits indicates
that exponential function will grow faster than any polynomial.
Q: Similarly argue that any polynomial will grow faster than logarithm function.

74