i

LEBANESE UNIVERSITY
Real Analysis
Dr. Ziad Khalil
February 19, 2021

ii
 CONTENTS

Introduction v

1 Real Numbers 2
1.1 Sets of Numbers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Order Structure of R. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Absolute Value And Intervals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Completeness Property of R. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2 Numerical Sequences 16
2.1 Limits Of Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2 Limits And Ordered Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3 Monotone Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.4 Sub Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3 Limits and Continuity on R 26

3.1 Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2 Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3 Continuous Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.4 Properties of Continuous Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.5 Substitution Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.6 More About Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

4 Differentiation on R 48
4.1 First Derivative: Properties And Examples . . . . . . . . . . . . . . . . . . . . . . . . 48
4.2 Higher Order Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.3 The Mean Value Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.4 Taylor’s Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.5 Inverse Function Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

5 Elementary Functions 63
5.1 Trigonometric Functions And Inverses . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.1.1 Sine And Arcsine Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.1.2 Cosine And Arccosine Functions . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.1.3 Tangent And Arctangent Functions . . . . . . . . . . . . . . . . . . . . . . . . . 65

iii
 5.1.4 Cotangent And Arccotangent Functions . . . . . . . . . . . . . . . . . . . . . . 66
5.2 The Power Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.3 Logarithmic and Exponential Functions . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.3.1 Logarithm Of Base e. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.3.2 Logarithm Of Base a > 0, a 6= 1. . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.3.3 Exponential Of Base e. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.3.4 Exponential Of Base a > 0, a 6= 1. . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.4 Hyperbolic Functions And Inverses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.4.1 Sine Hyperbolic And Inverse Sine Hyperbolic . . . . . . . . . . . . . . . . . . . 69
5.4.2 Cosine Hyperbolic And Inverse Cosine Hyperbolic . . . . . . . . . . . . . . . . 70
5.4.3 Tangent Hyperbolic And Inverse Tangent Hyperbolic . . . . . . . . . . . . . . 70
5.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

6 Finite Expansions 73
6.1 Taylor And Finite Expansions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6.2 Operations On Finite Expansions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.3 Equivalent And Finite Expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

iv
 INTRODUCTION

Analysis is a profound subject; its neither easy to understand nor summarize. However real
analysis can be discovered by solving problems. This course aim’s to give you opportunities to
discover real analysis.

v
vi
1
CHAPTER 1
REAL NUMBERS
1.1 Sets of Numbers.
Let us briefly examine the main sets of numbers that we are interested in.
N := {0, 1, 2, 3 · · · } The set of natural numbers.
N∗ := {1, 2, 3 · · · } The set of positive integers.
Z := {0, ±1, ±2, ±3 · · · } The set of integer numbers.
m
Q := { ; m ∈ Z and n ∈ N∗ } The set of rational numbers.
n
R := { The unique complete ordered field ⊃ Q} The set of real numbers.
R \ Q := {x : x ∈ R and x ∈ / Q} The set of irrational numbers.
Without loss of generality, in the definition of Q, one can assume that the fraction is reduced, that
is, m and n have no common divisors (other than ±1). Also one may assume that n ∈ Z∗ = Z \ {0}.
The set of rationals (Q, +, ·) with the basic operations addition and multiplication is a highly

O
satisfactory algebraic system, that is, it satisfies the field axioms:

O
1 Closure law: a + b ∈ Q and ab ∈ Q for all a, b ∈ Q.

O
2 Commutative law: a + b = b + a and ab = ba for all a, b ∈ Q.

O
3 Associative law: a + (b + c) = (a + b) + c and a(bc) = (ab)c for all a, b, c ∈ Q.

O
4 Distributive law: a(b + c) = ab + ac for all a, b, c ∈ Q.

O
5 Existence of Identity Elements: ∃ 0, 1 ∈ Q such that 0 + a = a and 1a = a for all a ∈ Q.
6 Existence of Inverse Elements:
• Additive inverse: ∀a ∈ Q, ∃ − a ∈ Q such that a + (−a) = 0.
• Multiplicative inverse:∀a ∈ Q∗ = Q \ {0}, ∃ a−1 ∈ Q such that aa−1 = 1.
we can use the above structure to define the concepts of subtraction and division , as a − b = a + (−b)
and ab = ab−1 (assuming b 6= 0).
It would appear that the set Q is a nice algebraic system until one tries to solve the equation x2 = 2.
It turns out that no rational number satisfies this equation, but there is a good reason to believe that
there is some kind of numbers satisfying this equation. Consider a square with sides having length
one. If x is the length of the diagonal, then from geometry we know that x√ 2 = 12 + 12 , i.e., x2 = 2.
√
Apparently there is a positive length whose square is 2, which we write as 2. But 2 cannot be a
rational number.

Let us introduce some methods of proofs we use in these notes.

Proof By Direct method. To show p =⇒ q, we assume that p is true and proceed through a
sequence of logical steps to arrive at the conclusion that q is also true.
Proof By Contrapositive Method. To show p =⇒ q it is equivalent to show ¬q =⇒ ¬p.
Proof By Contradiction. In order to prove a statement s is true one can proceed as follows: We
suppose s is false and try to obtain a contradiction i.e, a statement which is always false, for example
r ∧ ¬r where r is an additional statement. In other words s ⇐⇒ [¬s =⇒ (r ∧ ¬r)].
For example [p =⇒ q] ⇐⇒ [p ∧ ¬q =⇒ r ∧ ¬r]. That is, to show [p =⇒ q] we assume q is false
and try to obtain r ∧ ¬r. Note that the contrapositive method is a particular case of contradiction.

2
 An integer is said to be even (odd) if n = 2k (n = 2k + 1) for some k ∈ Z.
1.1 Example. Show that: [n is an even integer] ⇐⇒ [n2 is an even integer].
Solution.
=⇒) Direct method: Assume that n is an even integer, then n = 2k, k ∈ Z.
Now n2 = 4k 2 = 2(2k 2 ). Since 2k 2 ∈ Z, then n2 is an even integer.
⇐=) Contrapositive Method: To show [n2 is an even integer] =⇒ [n is an even integer] is equivalent
to show [n is an odd integer] =⇒ [n2 is an odd integer].
Assume that n is an odd integer, then n = 2k + 1, k ∈ Z.
Now n2 = 4k 2 + 4k + 1 = 2(2k 2 + 2k) + 1. Since (2k 2 + 2k) ∈ Z, then n2 is an odd integer.

√ √
1.2 Property. 2 is not a rational number i.e., 2∈
/Q.

√ √ ∗
Proof. By Contradiction: Suppose that 2 ∈ Q. Then 2 = m n , m ∈ Z, n ∈ N with m and n have
no common divisors (other than ±1). Hence m = 2n (by taking squares). Thus m2 is an even
2 2

integer which gives m is even and can be written m = 2k, k ∈ Z.

Now n2 = 2k 2 is an even integer which gives n is even. Therefore m and n √have 2 as common divisor
which contradicts the fact that m and n have no common divisors. Thus 2 ∈ / Q.

This shows that there are ’holes’ in the rational numbers, so we want enough numbers to fill the
holes.
√ Informally, what we want to do is ’complete’ the rational numbers to include numbers like
2. We call numbers in R \ Q := {x; x ∈ R and x ∈ / Q} irrational numbers.
1.3 Property. There exists a nonempty subset Q∗+ := { m ∗
n ; m, n ∈ N } of Q, called the set of positive

O
elements of Q, satisfying the following properties:
7 Q∗+ is closed under addition i.e., x + y ∈ Q∗+ ∀x, y ∈ Q∗+ .
O8 Q∗+ is closed under multiplication xy ∈ Q∗+ ∀x, y ∈ Q∗+ .
O9 For each x ∈ Q exactly one of the following holds: x ∈ Q∗+ , −x ∈ Q∗+ , x = 0.
1.4 Definition. On Q we define the relation ”≥” by
x ≥ y ⇐⇒ x − y ≥ 0 ( i.e., x − y > 0 or x − y = 0).

√ √
O O
1.5 Theorem. (Q, +, ·, ≥) is an ordered field i.e, it satisfies the nine properties 1 ... 9 .
Also Q( 2) := {x + y 2; x, y ∈ Q} is an ordered field. This suggests that there are infinitely
many ordered sub fields of R. The property that ensures the existence of a unique ordered field R
containing Q(up to renaming the elements) and distinguishes R from all other ordered fields is the
Completeness Property described in section 1.4.

1.2 Order Structure of R.

O O
(R, +, ·, ≥) is an ordered field i.e, it satisfies the algebraic properties 1 · · · 6 and the Order Axiom:
There exists a nonempty subset R∗+ of R, called the set of positive elements of R, satisfying the

O
following properties:
7 R∗+ is closed under addition i.e., x + y ∈ R∗+ ∀x, y ∈ R∗+ .
O8 R∗+ is closed under multiplication xy ∈ R∗+ ∀x, y ∈ R∗+ .
O9 ∀x ∈ R, exactly one of the following holds: x ∈ R∗+ , −x ∈ R∗+ , x = 0. (Trichotomy Property)
A real number x is called positive if x ∈ R∗+ and negative if −x ∈ R∗+ .
The binary relation ≥ is derived from the above Order Axiom:
x ≥ y ⇐⇒ x − y ≥ 0 ( i.e., x − y > 0 or x − y = 0).
Hence, if either x > y or x = y, we write x ≥ y or y ≤ x and say that x is greater than or equal to y
or y is less than or equal to x. If S ⊂ R,we define S+ := {x ∈ S : x ≥ 0} and S+∗ := {x ∈ S : x > 0}.

3
1.6 Property. The binary relation ≥ on R has the following properties:

1. Reflexive: x ≥ x for all x ∈ R.

2. Anti-symmetric: x ≥ y and y ≥ x implies that x = y.

3. Transitive: x ≥ y ≥ z implies x ≥ z.

4. Totally ordered: x ≥ y or y ≥ x, ∀x, y ∈ R (any two elements of R are comparable).

5. Compatible with addition: x ≥ y =⇒ x + a ≥ y + a for all a ∈ R.

6. Compatible with multiplication: x ≥ y =⇒ ax ≥ ay for all a ≥ 0.

7. Reflection: x ≥ y =⇒ ax ≤ ay for all a ≤ 0.

Proof.

1. Let x ∈ R. Since x − x = 0, then x ≥ x.

2. Contrapositive Method: This follows by applying the trichotomy property for x − y.

3. If x = y or y = z, then x ≥ z follows easily.

Assume x 6= y and y 6= z, then x − y > 0 and y − z > 0. Hence, by closure under addition
x − z = (x − y) + (y − z) > 0.

4. This follows by applying the trichotomy property for x − y.

5. Let a ∈ R. Then (x + a) − (y + a) = x − y ≥ 0.

6. If x = y, then ax = ay.
Assume x 6= y, then x − y > 0. Hence, by closure under multiplication ax − ay = a(x − y) > 0.

7. Let a ≤ 0. If x = y, then ax = ay.

Assume x 6= y, then x−y > 0. Hence, by closure under multiplication ay −ax = −a(x−y) > 0.

The binary relation ≥ with properties 1.,2., and 3. is called partially ordered relation and with
4. is called totally ordered relation. Note that 4. implies 1., hence a totally ordered relation is a
relation satisfies 2.,3. and 4.

1.7 Property. The totally ordered relation ≥ on R has the following properties:

1. (a ≤ b and c ≤ d) =⇒ a + c ≤ b + d.

2. (0 ≤ a ≤ b and 0 ≤ c ≤ d) =⇒ ac ≤ bd.

3. (a ≤ b ≤ 0 and c ≤ d ≤ 0) =⇒ ac ≥ bd.

4. α.) 0 < a < b =⇒ a2 < b2 .

β.) a < b < 0 =⇒ a2 > b2 .
1
5. (0 < a < b or a < b < 0) =⇒ a > 1b .

4
Proof.

1. (a ≤ b and c ≤ d) =⇒ (a + c ≤ b + c and b + c ≤ b + d) ( compatible with addition)

=⇒ a + c ≤ b + d. (Transitive)

2. (0 ≤ a ≤ b and 0 ≤ c ≤ d) =⇒ (ac ≤ bc and bc ≤ bd) ( compatible with multiplication)

=⇒ ac ≤ bd. (Transitive)

3. (a ≤ b ≤ 0 and c ≤ d ≤ 0) =⇒ (−a ≥ −b ≥ 0 and − c ≥ −d ≥ 0) (Reflection)

=⇒ ac ≥ bd. (Using 2.)

4. α.) and β.) are particular cases of 2. and 3.

1 1 1
5. (0 < a < b or a < b < 0) =⇒ (ab > 0 and b − a > 0) =⇒ (b − a) > 0 =⇒ > .
ab a b

1.8 Remark. Let a, b ∈ R such that a < b.

(x − a)(x − b) ≤ 0 ⇐⇒ a ≤ x ≤ b (Equivalently)
(x − a)(x − b) > 0 ⇐⇒ x < a or x > b.

1.9 Remark. Let a, b ∈ R.

ab = 0 ⇐⇒ a = 0 or b = 0 (Equivalently)
ab 6= 0 ⇐⇒ a 6= 0 and b 6= 0.

1.3 Absolute Value And Intervals.

1.10 Definition. The absolute value of a real number x, denoted by |x| is defined by

x if x ≥ 0
|x| =
−x if x < 0.

Note that
• |x| ≥ 0 ∀x ∈ R( by definition). For example |0| = 0 and | − 2| = |2| = 2.
• |x| = √
max{x, −x} ∀x ∈ R.
√
• |x| = x2 ∀x ∈ R(∵ |x|2 = x2 ). While ( x)2 = x for all x ≥ 0.
1.11 Examples.
 
x−2 if x − 2 ≥ 0 x−2 if x ≥ 2
1. |x − 2| = =
−(x − 2) if x − 2 < 0 −x + 2 if x < 2.
 2  2
2 x −x−2 if (x + 1)(x − 2) ≥ 0 x −x−2 if x ≤ −1 or x ≥ 2
2. |x −x−2| = =
−(x2 − x − 2) if (x + 1)(x − 2) < 0 −x2 + x + 2 if − 1 < x < 2.

5
1.12 Property. Let x, y ∈ R. The following properties of absolute value are true:.

1. |x| = a ⇐⇒ x = ±a ( provided a ≥ 0). 6. |x| ≤ a ⇐⇒ −a ≤ x ≤ a.(provided a ≥ 0)

10 . |x| = 0 ⇐⇒ x = 0. 60 . |x| < a ⇐⇒ −a < x < a.( a ≥ 0)

2. |xy| = |x||y|. 7. |x| ≥ a ⇐⇒ x ≤ −a or x ≥ a. (a ∈ R)

20 . | xy | = |x|
|y| . ( provided y 6= 0)
70 . |x| > a ⇐⇒ x < −a or x > a. (a ∈ R)

3. |x| = | − x|. 8. −|x| ≤ x ≤ |x| for all x ∈ R.

30 . |x − y| = |y − x|. 80 . − |x| ≤ −x ≤ |x| for all x ∈ R.

4. x2 = y 2 ⇐⇒ |x| = |y|. 9. |x + y| ≤ |x| + |y| for all x, y ∈ R.

40 . x2 = y 2 ⇐⇒ x = y. ( provided xy ≥ 0) 90 . |x − y| ≤ |x| + |y| for all x, y ∈ R.

5. x2 < y 2 ⇐⇒ |x| < |y|. 10. |x − y| ≥ |x| − |y| for all x, y ∈ R.

50 . x2 < y 2 ⇐⇒ x < y.(provided x, y ≥ 0) 100 . |x + y| ≥ |x| − |y| for all x, y ∈ R.

Proof.

1. is obvious from the definition of absolute value and 10 . is a particular case of 1. (a = 0).

2. We consider four cases (x ≥ 0 ∧ y ≥ 0) ∨ (x ≥ 0 ∧ y ≤ 0) ∨ (x ≤ 0 ∧ y ≥ 0) ∨ (x ≤ 0 ∧ y ≤ 0). For

example if (x ≥ 0 ∧ y ≤ 0), then |xy| = −xy = |x||y|. Similarly, we establish the other cases.
Now 20 . is an application of o 2.; Since y 6= 0, then |x| = | xy y| = | xy ||y| and | xy | = |x|
|y| .

3. We consider two cases (x ≥ 0 ∨ x < 0). If x ≥ 0, |x| = x = | − x| and if x < 0, |x| = −x = | − x|.
Now 30 . is an application of o 3.; |x − y| = | − (x − y)| = |y − x|.

4. x2 = y 2 ⇐⇒ |x|2 = |y|2 ⇐⇒ |x|2 − |y|2 = 0 ⇐⇒ (|x| − |y|)(|x| + |y|) = 0 ⇐⇒ |x| = |y|.

Now for 40 ., the product xy > 0 is equivalent to x and y have same sign i.e., both positive or
both negative. In both cases x = y ⇐⇒ |x| = |y| ⇐⇒ x2 = y 2 .

5. x2 < y 2 ⇐⇒ x2 − y 2 < 0 ⇐⇒ (|x| − |y|)(|x| + |y|) < 0 ⇐⇒ |x| − |y| < 0 ⇐⇒ |x| < |y|. If
we assume x, y ≥ 0, then we establish 50 ..
Important notes:
• For x, y ≥ 0, to show x < y it is equivalent to show x2 < y 2 .
• For x, y ≥ 0, to show x ≤ y it is equivalent to show x2 ≤ y 2 .(combining 40 . and 50 . together)

6. |x| ≤ a ⇐⇒ x2 ≤ a2 ⇐⇒ (x − a)(x + a) ≤ 0 ⇐⇒ −a ≤ x ≤ a. Similarly for 60 ..

7. and 70 . may be deduced from 6. and 60 . using (p ⇐⇒ q) ⇐⇒ (¬p ⇐⇒ ¬q).

8. is a particular case of 6.( a = |x| ≥ 0) or using x = ±|x|. Since property 8. is true ∀x ∈ R,

then we may replace x by −x ∈ R to obtain 80 .

9. First way: Let x, y ∈ R. We have

(−|x| ≤ x ≤ |x|) ∧ (−|y| ≤ y ≤ |y|) =⇒ −|x| − |y| ≤ x + y ≤ |x| + |y| =⇒ |x + y| ≤ |x| + |y|.
Second way: Let x, y ∈ R, a = |x + y| ≥ 0 and b = |x| + |y| ≥ 0. To show a ≤ b is equivalent
to show a2 ≤ b2 . Now a2 − b2 = (x2 + 2xy + y 2 ) − (x2 + 2|x||y| + y 2 ) = 2(xy − |xy|) ≤ 0.

10. First way: Using 6., to show 10. is equivalent to show −|x − y| ≤ |x| − |y| ≤ |x − y|. Now,
|x| = |(x − y) + y| ≤ |x − y| + |y| =⇒ |x| − |y| ≤ |x − y|. Similarly,
|y| = |(y − x) + x| ≤ |y − x| + |x| ≤ |x − y| + |x| =⇒ |x| − |y| ≥ −|x − y|.
Second way: a = |x|−|y| , b = |x−y| ≥ 0. a2 −b2 = 2(xy −|xy|) ≤ 0 =⇒ a2 ≤ b2 =⇒ a ≤ b.

6
1.13 Definition. A set I of real numbers is called an interval if any real number that lies between
two numbers in I is also in I i.e.:

∀a ∈ I ∀b ∈ I, a < x < b =⇒ x ∈ I.

It follows clearly from the definition that the set R = (−∞, ∞) is an interval. In addition to (−∞, ∞),
there are 8 other types of intervals:

Open Interval: (a, b) = {x ∈ R : a < x < b}

Closed Interval: [a, b] = {x ∈ R : a ≤ x ≤ b}
Half-open Intervals: (a, b] = {x ∈ R : a < x ≤ b} [a, b) = {x ∈ R : a ≤ x < b}
Infinite Open Intervals: (−∞, b) = {x ∈ R : x < b} (a, ∞) = {x ∈ R : a < x}
Infinite Closed Intervals: (−∞, b] = {x ∈ R : x ≤ b} (a, ∞) = {x ∈ R : a ≤ x}.

It’s easy to check that each of these nine types of sets are intervals. Moreover, one may show that
every interval has one of these nine forms. Note that, some references use equivalent notations

]a, b[= (a, b), [a, b[= [a, b) and ]a, b] = (a, b].

Important notes:
• It’s very useful to associate real numbers as points on a number line to understand many concepts
and to arrive at ideas for a proof. In fact they will help us translating geometric ideas into analytic
proofs.
R
−5 −4 −3 −2 −1 0 1 2 3 4 5
• For a, b ∈ R, |a − b| = dist(a, b) represents the distance between a and b. In particular |a| =
|a − 0| = dist(a, 0). For example dist(−1, −3) = | − 1 − (−3)| = 2 and dist(−1, 3) = | − 1 − 3| = 4.

-3 -1 1 3

• (a + b)/2 represents the midpoint of the segment [a, b] or [b, a].

a (a+b)/2 b b (a+b)/2 a

• min{a, b} = (a + b)/2 − |a − b|/2 i.e., if we move to the left half the distance from the midpoint we
get the minimum.
• max{a, b} = (a + b)/2 + |a − b|/2 i.e., if we move to the right half the distance from the midpoint
we get the maximum.
• Let a ∈ R and δ > 0.

|x − a| < δ ⇐⇒ −δ < x − a < δ ⇐⇒ a − δ < x < a + δ ⇐⇒ x ∈]a − δ, a + δ[.

So the set Iδ (a) := {x ∈ R : |x − a| < δ} =]a − δ, a + δ[ consists of the points on the real line whose
distance from a is less than δ.
a-δ x a x a+δ

Iδ (a) :=]a − δ, a + δ[ is called a δ-neighborhood of a.

7
1.14 Property. Let x, x0 ∈ R. The following properties are true:

1. (∀ε > 0, x < ε) =⇒ x ≤ 0.

2. (∀ε > 0, |x| < ε) =⇒ x = 0.

3. (∀ε > 0, |x − x0 | < ε) =⇒ x = x0 .

Proof.

1. Given: ∀ε > 0, x < ε. R.T.S(Require To Show): x ≤ 0.

Suppose to the contrary that x > 0. Take ε = x/2 > 0, then x < ε = x/2(equivalently 1 < 1/2)
which is a contradiction. Thus x ≤ 0.

2. ∀ε > 0, |x| < ε =⇒ |x| ≤ 0 ( using 1.) =⇒ |x| = 0 (∵ 0 ≤ |x|) =⇒ x = 0.

3. ∀ε > 0, |x − x0 | < ε =⇒ x − x0 = 0 ( using 2.) =⇒ x = x0 .

1.4 Completeness Property of R.

In this section we discuss the notion of boundedness of a set and we give the completeness axiom
that assure us R has no ’holes’.

1.15 Definition. Let A ⊆ R.

1. We say that A is bounded from above in R if (∃M ∈ R)(∀x ∈ A)(x ≤ M ).

Any M with this property is called an upper bound of A.

2. We say that A is bounded from below in R if (∃m ∈ R)(∀x ∈ A)(m ≤ x).

Any m with this property is called a lower bound of A.

3. We say that A is bounded in R if its bounded from above and below i.e.,

(∃m, M ∈ R)(∀x ∈ A)(m ≤ x ≤ M ).

In terms of intervals, A is bounded above if A ⊆ (−∞, M ], bounded below if A ⊆ [m, ∞) and
bounded if A ⊆ [m, M ].

1.16 Examples.

1. Let A = (0, 1].

set of lower bounds (−∞, 0 ] 0=m A M=1 set of upper bounds [1,∞)

M = 1 is an upper bound of A (∵ (∀x ∈ (0, 1])(x ≤ 1)) and every number larger than M is
also an upper bound of A. m = 1 is a a lower bound of A (∵ (∀x ∈ (0, 1])(0 ≤ x)) and every
number smaller than m is also a lower bound of A.

2. Let B = [0, 1).

set of lower bounds (−∞, 0 ] 0=m B M=1 set of upper bounds [1,∞)

Note that upper bound ( lower bound) of a set may and may not belongs to the set.

3. The set {1/n2 : n ∈ N∗ } is a bounded set because 0 < 1/n2 ≤ 1 for all n ∈ N∗ .

8
 4. The set A = (0, ∞) is not bounded above because (∀M ∈ R)(∃x ∈ A)(x > M ).
0 A M x

5. The set A = (−∞, 0] is not bounded below because (∀m ∈ R)(∃x ∈ A)(x < m).
x m A 0

Any set that is not bounded (either above or below) is said to be unbounded.

1.17 Definition. (max-min) Let A ⊆ R and M, m ∈ R. We say that:

O O
1. M is the maximum of A ”and we write M = max A” if
1 M is an upper bound of A and 2 M ∈ A. (i.e., M is the largest element in A)

O O
2. m is the minimum of A ”and we write M = min A” if
1 m is a lower bound of A and 2 m ∈ A. (i.e., m is the smallest element in A)

A set may not have a largest or smallest member. For example, the set A = (0, 1] admits a maximum
(max A = 1 because 1 is an u.b. of A and 1 ∈ A ) but it does not admit a minimum. While, the set
B = [0, 1) admits a minimum (min B = 1 because 1 is a l.b. of B and 1 ∈ B ) but it does not admit
a maximum. Note that a finite set always admits a maximum and minimum.

1.18 Definition. (sup-inf ) Let A be a nonempty subset of R.

1. An upper bound M of A with the property that M ≤ u for all upper bounds u of A is called a
least upper bound or supremum of A, and is denoted by sup A.

2. A lower bound m of A with the property that l ≤ m for all lower bounds l of A is called a
greatest lower bound or infimum of A, and is denoted by inf A.

3. If A is not bounded above, one defines sup A = ∞.

4. If A is not bounded below, one defines inf A = −∞.

Note that sup A, inf A ∈ R̄ := R ∪ {±∞}

O
1.19 Remark.

O
1 If max A exists in R, then sup A = max A.
2 If min A exists in R, then inf A = min A.

O
Proof.
1 Let M = max A. Then M
|{z} ≤ sup A ≤ |{z}
M .
| {z }
element of A l.u.b A u.b A

O2 Let m = min A. Then |{z}

m ≤ {zA} ≤
|inf m
|{z} .
l.b A g.l.b A element of A
In the previous examples ( look at the figures above):
• Let A = (0, 1], sup A = 1 and inf A = 0. Note that min A doesn’t exist in R. Because if it exists,
then min A = inf A = 0 ∈ (0, 1] which is a contradiction.
• Let B = [0, 1), sup B = 1 and inf B = 0. Note that max B doesn’t exist in R. Because if it exists,
then max B = sup B = 1 ∈ [0, 1) which is a contradiction.
• sup(0, ∞) = ∞ and inf(−∞, 0] = −∞.

9
We may now state The property that ensures the existence of a unique ordered field R containing
Q(up to renaming the elements) and distinguishes R from all other ordered fields.

1.20 Axiom. (Completeness Property)

Every nonempty subset of R that is bounded above has a least upper bound. Equivalently,
Every nonempty subset of R that is bounded below has a greatest lower bound.

1.21 Theorem. There is exactly one complete ordered field say R (up to isomorphism i.e., renaming
the elements).

The following important consequences of the completeness property is useful in determining the
infimum or supremum of certain sets.

1.22 Property. (ε− characterization of supremum) Let A be a non empty bounded above subset of
R. Then M = sup A if and only if

1. M is an upper bound of A.

M-ε a M
2. (∀ε > 0)(∃a ∈ A)(M −ε < a). A

Proof. Assume that M = sup A. Then, by definition, M is an upper bound for A. Hence 1. holds.
For 2. assume to the contrary that (∃ε > 0)(∀a ∈ A)(M − ε ≥ a). Then, M − ε is an upper bound
for A which is smaller than M = sup A, a contradiction.
For the converse, assume that 1. and 2. hold. Since A is bounded above, it has a supremum, say i
S(by the completeness property). Since M is an upper bound for A, we must have that S ≤ M . If
S < M , then with ε = M − S > 0 and by 2. there exists a ∈ A such that M − (M − S) < a ≤ S i.e.,
S < S which is contradiction. Thus sup A = M .

1.23 Property. (ε− characterization of infimum) Let A be a non empty bounded below subset of R.
Then m = inf A if and only if

1. m is a lower bound of A.

m a m+ε
2. (∀ε > 0)(∃a ∈ A)(a < m+ε). A

1.24 Property. (Archimedean Principle) For all x ∈ R, there exists n ∈ N such that n > x i.e.,the
set N of natural numbers is not bounded above.

Proof. Assume that N is bounded above. By the Completeness Property,sup N exists. Let M =
sup N. By the ε− characterization with ε = 1, there exists k ∈ N such that M − 1 < k. This implies
that M < k + 1 ≤ M which is contradiction. Thus N is not bounded above.

10
1.25 Corollary. For all x > 0 for all y ∈ R, there exists n ∈ N such that y < nx.

Proof. Given that xy ∈ R. By the Archimedean Principle, there exists n ∈ N such that n > xy . This
implies that there exists n ∈ N such that y < nx.
1
1.26 Corollary. For all ε > 0, there exists n ∈ N such that n < ε.

Proof. Given that 1ε ∈ R. By the Archimedean Principle, there exists n ∈ N such that n > 1ε . This
implies that there exists n ∈ N such that n1 < ε.

1.27 Corollary. (∀n ∈ N∗ , a − 1

n ≤ b ≤ a + n1 ) =⇒ a = b.

Proof. The Corollary will follow if we show that (∀ε > 0, |a − b| < ε). Let ε > 0. By the
Archimedean Principle, there exists n0 ∈ N such that n10 < ε. We have a − n10 ≤ b ≤ a + n10 , or
equivalently |a − b| ≤ n10 . This implies that |a − b| < ε.

1.28 Property. Every nonempty subset A of Z which is bounded above admits a maximum .

Proof. Since A is bounded above, sup A exists (By the Completeness Property). Let M = sup A.
The property will follow if we show that M ∈ A. Suppose to the contrary that M ∈ / A. By
the ε− characterization with ε = 1, there exist a ∈ A such that M − 1 < a < M . By the ε−
characterization again with ε = M2−a , there exist a0 ∈ A such that M − M2−a < a0 < M . Hence
M − 1 < a < M − M2−a < a0 < M . This implies that a0 − a is an integer strictly between 0 and 1
which is a contradiction.

1.29 Corollary. Every nonempty subset A of Z which is bounded below admits a minimum .

Proof. min A = − max(−A)

In particular, we have the following principle.

The Well Ordering Principle: every nonempty subset A of N admits a minimum.

1.30 Theorem. (Principle of Mathematical Induction)

For each n ∈ N, let P (n) be a property depends on n. If

1. P (0) is true i.e., the property is true for n = 0.

2. P (n) =⇒ P (n + 1) i.e., the truth of P (n) implies the truth of P (n + 1).

Then P (n) is true for all n ∈ N.

Proof. Let A := {n ∈ N : P (n) is true}, the theorem will follow if we show A = N. Assume to the
contrary that A 6= N, then N \ A = {n ∈ N : n ∈ / A} admits a minimum (By the well ordering
principle). Let m = min(N \ A) ∈ N \ A. Since 0 ∈ A, then m 6= 0. Hence m − 1 ∈ A i.e, P (m − 1)
is true. This implies, using 2., P (m) is true i.e., m ∈ A which is contradiction. Thus A = N.

Principle of Mathematical Induction(modified)

Let n0 ∈ N and P (n) be a property depends on n. If

1. P (n0 ) is true i.e., the property is true for n ≥ n0 .

2. P (n) =⇒ P (n + 1) for n ≥ n0

Then P (n) is true for all n ≥ n0 .

11
1.31 Examples. 1. Let r ≥ −1. Prove Bernoulli’s inequality: (1 + r)n ≥ 1 + nr ∀n ∈ N.
Solution. In this case, P (n) is (1 + r)n ≥ 1 + nr.
• For n = 0 we have (1 + r)0 = 1 ≥ 1 + 0r. Hence P (0) is true.
• Suppose P (n) is true i.e., (1 + r)n ≥ 1 + nr.
Let us show P (n + 1) is true i.e., (1 + r)n+1 ≥ 1 + (n + 1)r.
Using 1 + r ≥ 0 and P (n) is true, we obtain

(1 + r)n+1 = (1 + r)(1 + r)n ≥ (1 + r)(1 + nr) ≥ 1 + (n + 1)r + nr2 ≥ 1 + (n + 1)r.

Thus P (n) is true for all n ∈ N.
2. Let x1 , · · · , xn ∈ R. Prove the following triangle inequality:
|x1 + x2 + · · · + xn | ≤ |x1 | + |x2 | + · · · + |xn | ∀n ≥ 2.
Solution. In this case; |x1 + x2 + · · · + xn | ≤ |x1 | + |x2 | + · · · + |xn | (P (n)).
• For n = 2 we have |x1 + x2 | ≤ |x1 | + |x2 | for all x1 , x2 ∈ R. Hence P (2) is true.
• Assume for a given n ≥ 2, P (n) is true and let us show P (n + 1) is true .
|x1 + x2 + · · · + xn + xn+1 | ≤ |x1 + x2 + · · · + xn | + |xn+1 | (∵ P (2) is true)
| {z } | {z } | {z } | {z }
X1 X2 X1 X2
≤ |x1 | + |x2 | + · · · + |xn | + |xn+1 | (∵ P (n) is true)
Thus P (n) is true for all n ≥ 2.
3. Let x ∈ R. Show that |xn | = |x|n for any integer n ∈ N∗ .
Solution. |xn | = |x|n (P (n)).
• For n = 1 we have |x1 | = |x|1 for all x ∈ R. Hence P (1) is true.
• Assume for a given n ≥ 1, P (n) is true and let us show P (n + 1) is true .
|xn+1 | = ||{z}
xn |{z} xn |||{z}
x | = ||{z} x | (∵ |ab| = |a||b|)
a b a b
n
= |x| |x| (∵ P (n) is true)
n+1
= |x| .
Thus P (n) is true for all n ≥ 2.

1.32 Definition. Let x ∈ R. The integer part of x, denoted by E(x), is the greatest integer less or
equal to x i.e., E(x) := max{n ∈ Z : n ≤ x}.

E(x) is well defined because the set {n ∈ Z : n ≤ x} is a nonempty bounded above subset of Z.
If x ∈ Z, then E(x) = x. In general, If we move to the left from x, the first integer is E(x). For
example E(1) = 1, E(1.7) = 1 and E(−1.7) = −2.

2
y

x
−2 −1 1 2 3

−1

−2

Greatest integer function.

12
1.33 Property. The integer part function E : R 7→ Z satisfy the following properties:

1. If k ∈ Z and k ≤ x, then k ≤ E(x).

2. E(x) ≤ x < E(x) + 1 or equivalently, 0 ≤ x − E(x) < 1 .(The inequality is strict if x ∈

/ Z)

3. x ≤ y =⇒ E(x) ≤ E(y).(i.e., E is an increasing function)

4. E(x + k) = E(x) + k ∀k ∈ Z.

Proof.
1. Let k ∈ Z be such that k ≤ x. Since E(x) is the greatest integer less or equal to x, then
k ≤ E(x).
2. E(x) is the greatest integer less or equal to x. This implies that E(x) ≤ x. For the second
inequality, suppose to the contrary that E(x) + 1 ≤ x, then E(x) + 1 ≤ E(x) which is
contradiction. Hence x < E(x) + 1
using2. using1.
3. x ≤ y =⇒ E(x) ≤ x ≤ y =⇒ E(x) ≤ y =⇒ E(x) ≤ E(y).
4. Let k ∈ Z.
1.or3.
• E(x) ≤ x =⇒ E(x) + k ≤ x + k =⇒ E(x) + k ≤ E(x + k).
The equality will follow if we show E(x + k) ≤ E(x) + k.
• E(x + k) ≤ x + k < E(x) + k + 1 =⇒ E(x + k) ≤ E(x) + k.

1.34 Theorem. (Density of the Rationals)

Between any pair of distinct real numbers there is a rational number.

Proof. Let a < b. By the Archimedean Principle, n(b − a) > 1 for some n ∈ N . Let m := E(na) + 1.
Then na < m ≤ na + 1 < nb, hence a < mn < b.

1.35 Theorem. (Density of the Irrationals)

Between any pair of distinct real numbers there is irrational number.

√ √
Proof. Let
√ a < b, then a
√ + 2 < b + 2 .√ √ rationals, There exists q ∈ Q such
By the density of the
that a + 2 < q < b + 2. Hence a < q − 2 < b, where q − 2 is irrational.

√
1.36 Property. If n is positive integer that is not a perfect square, then n is irrational.

√ √ √
Proof. Since n is not a perfect square, then n 6∈ Z. Hence 0 < n − E( n) < 1. Suppose to the
√ √
contrary that n is rational and let A := {m ∈ N : m n ∈ N}. The set A is a nonempty subset of
√
N, by the well ordering principle, A admits a minimum m0 . Hence m0 n ∈ N.Now;
√ √
• m := m0 ( n − E( n)) ∈ N and m < m0 .
√ √ √
• m n = m0 n − m0 nE( n) ∈ N.
√
Hence m ∈ A. This implies m0 ≤ m which is a contradiction(∵ m0 > m ). Thus n is irrational.

13
With the help of the completeness property one can prove the existence in R of the nth root.

1.37 Property. (n-th root) Consider the equation xn = a (?) where n ∈ N∗ and a ∈ R.
√ 1
• If n is even and a ≥ 0, then equation (?) has a unique solution in R+ denoted by n a or a n and
√
called the n-th root of a.(consequently, − n a ia also a solution in R).
√ 1
• If n is odd and a ∈ R, then equation (?) has a unique solution in R denoted by n a or a n and
called the n-th root of a.

We will see, in chapter three, that the existence of the n-th root is a simple application of the
intermediate value theorem.
1.5 Exercises
Exercise 1. Let a, b, c, d ∈ R.
a c
1. Assume bd > 0. Show that b ≤ d ⇐⇒ ad ≤ bc.
a c
2. Assume bd < 0. Show that b ≤ d ⇐⇒ ad ≥ bc.
a
3. Show that: (0 ≤ a ≤ c) ∧ (b ≥ d > 0) =⇒ b ≤ dc .
a
4. Deduce that: (a ≤ c ≤ 0) ∧ (0 > b ≥ d) =⇒ b ≥ dc .
a
5. Show that: (a < 0) ∧ (a ≤ c) ∧ (0 < b ≤ d) =⇒ b ≤ dc .
|x+y| |x| |y|
Exercise 2. Show that 1+|x+y| ≤ 1+|x| + 1+|y| ∀x, y ∈ R.

Exercise 3. Solve the following:

p p
a. (x − 1)(x − 5) = x − 2. d. |x2 − 4| ≥ x.
2x+3 x+1 √ √
b. x+5 ≤ |x−1| . e. x − 1 = 3 − x.
√ √ √
c. x − 3 ≤ x2 − x f. x − 1 = 3 − x − 1.

Exercise 4. Let x ∈ R and n ∈ N∗ . Prove that:

a. E(x) = −E(−x) ⇐⇒ x ∈ Z. c. 0 ≤ E(nx) − nE(x) ≤ n − 1.

d. E E(nx)


b. E(x) + E(−x) = 0 or − 1. n = E(x).

Exercise 5. Let a ∈ R.

1. Show that: 0 < a ≤ 1 =⇒ an ≤ a ∀n ∈ N∗ .

2. Deduce that a ≥ 1 =⇒ an ≥ a ∀n ∈ N∗ .

Exercise 6. Let ∅ =
6 A ⊂ R. Show that A is bounded iff ∃C > 0 such that ∀a ∈ A, |a| ≤ C.
Exercise 7. Let A := { n1 : n ∈ N∗ }.

1. Show that sup A and inf A exist in R.

2. Find sup A and inf A.

3. Does min A exist. Justify your answer.

√ √
Exercise 8. Let A := {x ∈ Q : 0 ≤ x ≤ 2} = [0, 2] ∩ Q.

14
 1. Show that sup A and inf A exist in R.

2. Find sup A.

3. Does Q complete. Justify your answer.

4. Does max A exist. Justify your answer.

5. Find inf A.

Exercise 9.(Density of the dyadic rationals). Let a < b.

1. Show that 2n > n for all n ∈ N∗ .

2. Show that there exists n ∈ N such that 2n (b − a) > 1.

3. Let m := E(2n a) + 1. Show that 2n a < m ≤ 2n a + 1 < 2n b.

4. Deduce that for each pair of real numbers a < b, there exists m ∈ Z and there exists n ∈ N
such that a < 2mn < b.(i.e., between two distinct real numbers, there exists a dyadic rational)
1+(−1)n
Exercise 10. Let A := {un = n − n2 : n ∈ N∗ }.

1. Show that sup A = −1.

2. Now, the aim is to show that A is not bounded below:

(a) Let m < 0. Show that there exists n0 ∈ N∗ such that n20 > 1 − m.
(b) Show that un ≤ 1 − n2 for all n ∈ N∗ .
(c) Deduce that A is not bounded below.

Exercise 11. Let A := {un = n−1

n+1 : n ∈ N}. Show that inf A = −1 and sup A = 1.
x 2
Exercise 12. Let E := { x2 +1 : x ∈ R}. Determine, if existent, sup E, inf E, max E and min E.
2
Exercise 13. Let E := { xx2 +2
+1
: x ∈ R}. Determine, if existent, sup E, inf E, max E and min E .
Exercise 14. Let A and B be two nonempty bounded subsets of R and λ ∈ R. Define,
A + B := {a + b : a ∈ A and b ∈ B} and λA := {λa : a ∈ A}.

1. Show that sup(A + B) exists.

2. Show that sup(A + B) = sup A + sup B.

3. Show that sup(−A) = − inf A.

4. Show that sup(λA) = λ sup A for λ > 0.

5. Assume λ < 0. Deduce a formula for sup(λA), sup(A − B), inf(A + B), inf(A − B).

Exercise 15. Let A := {xy : |x| + |y| < 1}.Show that sup A exists and sup A = 41 .(Hint: 4ab ≤ (a + b)2 .)
Exercise 16. Assume A is nonempty bounded above subset of R and √ B is nonempty bounded below
subset of R with MA = sup A and mB = inf B > 0. Let C := {a − b : a ∈ A and b ∈ B}.

1. Show that sup C exists in R.

√
2. Show that for all ε > 0, mB + ( 2ε )2 < mB + 2ε .
p

√
3. Show that sup C = MA − mB .

Exercise 17. Let A be a nonempty bounded subset of R. Define |A| := {|a| : a ∈ A}. Show that
sup |A| − inf |A| ≤ sup A − inf A.(Hint: Use |x| − |y| ≤ |x − y|.)

15
CHAPTER 2
NUMERICAL SEQUENCES

2.1 Limits Of Sequences

2.1 Definition. A sequence of real numbers is a function u : {n ∈ N : n ≥ N } 7→ R (the

n → un :=u(n)
natural number N is usually 0 or 1). We call un the nth term of the sequence and we write the
sequence as (un )∞
n=N or simply (un ).

For example, un = n − 3 is the general term of the sequence (un )∞ 1

n=0 and vn = n−3 is the general
∞
term of the sequence (vn )n=4 . We are interested in studying the behavior of the sequence when n is
very large (goes to infinity), hence the index N can be chosen very large natural number.

2.2 Definition. A property (p) of a sequence un is said to hold eventually if there exists an index
N ∈ N such that un has property (p) for all n ≥ N (i.e., after a certain rank).

For example, by the Archimedean principle, ∀ ε > 0 the sequence ( n1 ) is eventually less than ε.

2.3 Definition. Let (un )∞

n=N be a sequence of real numbers. We say that:

1. (un ) is bounded above if (∃M ∈ R) (∀n ≥ N ) (un ≤ M ).

2. (un ) is bounded below if (∃m ∈ R) (∀n ≥ N ) (m ≤ un ).

3. (un ) is bounded if (∃m, M ∈ R) (∀n ≥ N ) (m ≤ un ≤ M ).

i.e., (un ) is bounded (above, below) if the set of its terms {un : n ≥ N } is bounded (above, below).

All properties of bounded sets can be restated for bounded sequences. In particular, Exercise 6.
Chapter one gives the following property

2.4 Property. A sequence (un )∞

n=N is bounded if and only if (∃C > 0) (∀n ≥ N ) (|un | ≤ C).

16
2.5 Definition. Let (un )∞
n=0 be a sequence in R and ` ∈ R. We say that (un ) converges to ` and we
write lim un = ` or un −→ ` if
n→∞ n→∞

(∀ε > 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ |un − `| < ε).

i.e., for all ε > 0, the sequence (un ) is eventually in the open interval (` − ε, ` + ε) = Iε (`).
i.e., for all ε > 0, the distance between un and ` is eventually less than ε.
Hence the limit of a convergent sequence do not depend on the behavior of the beginning of the
sequence. Only eventuality matters for the limit.

In the definition above, the number n0 may depends on ε.

un ∈ R

un0 `+ε
`
n ≥ n0 =⇒ dist(un , `) < ε
`−ε

n∈N
1 2 n0 n0+1
Graph of a convergent sequence: un −→ `
n→∞

How to show lim un = `?

n→∞
(?)
lim un = ` ⇐⇒ (∀ε > 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ |un − `| < ε).
n→∞

• We let ε > 0 be arbitrary.

• We search for n0 which may depends on ε such that the implication (?) holds true.
1
2.6 Example. Using the definition of the limit, show that lim = 0.
n→∞ n

Solution. Let ε > 0 be arbitrary. Scratch. (un ) is defined for n ≥ 1.

Take n0 = E( 1ε ) + 1 > 1ε . Now if n ≥ n0 , To find n0 we start from the R.H.S of the
then we have n > 1ε and from scratch we obtain implication (?):
| n1 − 0| = n1 < ε. • | n1 − 0| = n1 < ε ⇐ n > 1ε .
n2 +n−1
2.7 Example. Using the definition of the limit, show that lim 2 = 1.
n→∞ n −1

Solution.Let ε > 0 be arbitrary. Scratch. (un ) is defined for n ≥ 2.

Take n0 = E( 1ε + 1) + 1 > 1ε + 1. Now if n ≥ n0 , To find n0 we start from the R.H.S of the
then we have n > 1ε + 1 and from scratch we implication (?):
2 2
obtain n n+n−1
2 −1 − 1 ≤ n−11
< ε. • n n+n−1
2 −1 − 1 = n2n−1 ≤ (n−1)(n+1)
n+1 1
≤ n−1 . (We
n
drop |.| because n2 −1 > 0 for n ≥ 2 )
1
• n−1 ≤ ε ⇐⇒ n > 1ε + 1

17
 4n2 +3n−6
2.8 Example. Using the definition of the limit, show that lim n2 −6
= 4.
n→∞

Scratch. (un ) is defined for all n ∈ N.

Given ε > 0 and n ≥ n0 (n0 to be determined!). We search for some n0 such that the implication
(?) holds. As usual we start from the R.H.S of (?):
2
• 4n n+3n−6
2 −6 − 4 = 3n+18
n2 −6
= 3n+18
n2 −6
( we can drop |.| for n ≥ 3)
• To find the least n0 such that (?) holds true, we have to solve 3n+18
n2 −6
< ε, but we do not need to
find the least such n0 . So we will simplify matters by making estimates. The idea is that 3n+18
n2 −6
is
n 1
bounded above by some constant times n2 = n for sufficiently large n, which simplifies the inequality
into easier one.
• 3n+18
n2 −6
≤ 3n+18n
n2 /2
≤ 42 2 2 2
n provided that; n − 6 ≥ n /2 ⇐ n ≥ 12 ⇐ n ≥ 4.
• 42
n < ε ⇐⇒ n > ε .
42

Solution. Let ε > 0 be arbitrary. Take n0 = max{E( 42ε ) + 1, 3, 4}. Now if n ≥ n0 , then we have
42 4n2 +3n−6
n > ε , n > 3 and n > 4. So from scratch we obtain n2 −6
− 4 = 3n+18
n2 −6
≤ 42
n < ε.

2.9 Definition. Let (un )∞

n=0 be a sequence in R.

1. We say that (un ) diverges to ∞ and we write lim un = ∞ or un −→ ∞ if

n→∞ n→∞

(∀A > 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ un > A).

i.e., for all A > 0, the sequence (un ) is eventually in the open interval (A, ∞).

2. We say that (un ) diverges to −∞ and we write lim un = −∞ or un −→ −∞ if

n→∞ n→∞

(∀A < 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ un < A).

i.e., for all A < 0, the sequence (un ) is eventually in the open interval (−∞, A).

Note that, to show lim un = −∞ is equivalent to show lim (−un ) = ∞.

n→∞ n→∞
How to show lim un = ∞?
n→∞
(?)
(∀A > 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ un > A).
• We let A > 0 be arbitrary.
• We search for n0 which may depends on A such that the implication (?) holds true.

n2 +(−1)n
2.10 Example. Using the definition of the limit, show that lim √
n2 +n+7
= ∞.
n→∞

Solution. Scratch. (un ) is defined for n ≥ 0.

n2 +(−1)n 2 2
Let A > 0 be arbitrary. • √ n2 +n+7
≥ √n2n+n−1
2 +7n2
≥ n3n/3 ≥ n9 .
Take n0 = max{E(9A) + 1, 2}. Now if n ≥ n0 , 2
provided that: n2 − 1 ≥ n3 ⇐ n2 ≥ 32 ⇐ n ≥ 2.
then we have n > 9A, n ≥ 2 and
n2 +(−1)n 2 2 • n9 > A ⇐⇒ n > 9A.
√
n2 +n+7
≥ √n2n+n−12 +7n2
≥ n3n/3 ≥ n9 > A. —————-Another Estimate——————
—————-For the other estimate————– n2 +(−1)n n2 −1
• n2 +n+6 ≥ n2 +2n+1 ≥ (n−1)(n+1)
√ √
|n+1| ≥ n − 1.
Take n0 = max{E(A + 1) + 1, 5}. Now if n ≥ n0 , 2 2
Provided that: n + n + 6 ≤ n + 2n + 1 ⇐ n ≥ 5.
then we have n > A+1 and n ≥ 5. Consequently,
n2 +(−1)n 2 • n − 1 > A ⇐⇒ n > A + 1
√
2
n +n+6
≥ √ n −1
2
n +2n+1
≥ (n−1)(n+1)
|n+1| ≥ n − 1 > A.

2.11 Example. Let ` ∈ R. Show that lim (−1)n 6= ` and lim (−1)n 6= ±∞.(i.e., lim (−1)n @ in R)
n→∞ n→∞ n→∞

18
solution.
Negation Of Convergence.
lim un 6= ` ⇐⇒ (∃ε > 0)(∀n0 ∈ N)(∃n ∈ N)(n ≥ n0 ∧ |un − `| ≥ ε).
n→∞
Choose ε = 0.1 and let n0 ∈ N. If |u2n0 − `| = |1 − `| < ε = 0.1 and |u2n0 +1 − `| = |1 + `| < ε = 0.1,
then 2 = |1 − ` + ` + 1| ≤ |1 − `| + |` + 1| < 0.1 + 0.1 < 2 a contradiction. Hence there exists
n ≥ n0 ( n = 2n0 or n = 2n0 + 1 ) such that |un − `| ≥ ε.
One can show ∀` ∈ R, lim (−1)n 6= ` by contradiction:
n→∞
Assume to the contrary that ∃` ∈ R such that lim (−1)n = `. Then ∀ε >, | ± 1 − `| < ε. Hence
n→∞
2 = |1 − ` + ` + 1| ≤ |1 − `| + |` + 1| < ε + ε = 2ε. So 1 < ε for all ε > 0 is a contradiction.
Negation Of Infinite Limit .
lim un 6= ∞ ⇐⇒ (∃A > 0)(∀n0 ∈ N)(∃n ∈ N)(n ≥ n0 ∧ un ≤ A).
n→∞
Choose A = 2 and let n0 ∈ N. We have un0 = ±1 ≤ 2.
One can show that lim un 6= ∞ by contradiction:
n→∞
Assume to the contrary that lim un = ∞. Then for all A > 0, eventually un = ±1 > A which is a
n→∞
contradiction. Thus lim un 6= ∞.
n→∞

2.12 Definition. A sequence (un ) in R is said to be convergent if lim un = ` for some ` ∈ R.

n→∞
Otherwise it is divergent.

Note that a sequence (un ) ⊂ R is said to be divergent if lim un = ∞ or lim un = −∞ or the limit
n→∞ n→∞
of (un ) does not exists (in R).

2.13 Property. If (un ) is a convergent sequence, then its limit is unique.

Proof. Suppose that the sequence (un ) has two limits `1 and `2 and let us show `1 = `2 . The equality
will follow if we show that ∀ε > 0, |`1 − `2 | < ε.
Let ε > 0, then by the definition of the limit:
ε
∃n1 ∈ N, n ≥ n1 =⇒ |un − `1 | <
2
ε
∃n2 ∈ N, n ≥ n2 =⇒ |un − `2 | < .
2
For n3 > max{n1 , n2 } and with the help of triangle inequality we obtain
|`1 − `2 | = |`1 − un3 + un3 − `2 | ≤ |un3 − `1 | + |un3 − `2 | < 2ε + 2ε = ε.

2.14 Property. Every convergent sequence is bounded.

Proof. Let (un ) be a sequence converges to ` ∈ R. Then

For ε = 1, ∃n0 ∈ N, n ≥ n0 =⇒ |un − `| < 1.

Hence, for n ≥ n0 we have |un | = |un − ` + `| ≤ |un − `| + |`| < 1 + |`|. Define M := {1 +
|`|, |u0 |, |u1 |, · · · |un0 |}. Then we have ∀n ∈ N, |un | < M and thus (un ) is a bounded sequence.

19
2.15 Remark. If un −→ ` ∈ R∗ , then there exists N ∈ N such that for all n ≥ N , we have
n→∞
|`|
|un | > 2 .(Hence, if the limit is not zero, then un 6= 0 eventually)

|`|
Proof. We have lim un = `. Let N corresponds to ε := 2. Then,
n→∞

|`| |`|
n ≥ N =⇒ |un | − |`| ≤ |un − `| < =⇒ |un | > .
2 2

2.16 Property. Let (un ) and (vn ) be sequences in R.The following limit properties hold in R in the
sense that if the expression on the right side of the equation exists in R, then the limit on the left
side exists in R and equality holds.

1. lim λun = λ lim un , λ ∈ R.

n→∞ n→∞

2. lim (un + vn ) = lim un + lim vn .

n→∞ n→∞ n→∞

3. lim (un vn ) = lim un lim vn .

n→∞ n→∞ n→∞

4. lim (un /vn ) = lim un / lim vn provided lim vn 6= 0.

n→∞ n→∞ n→∞ n→∞

5. lim |un | = | lim un |. (Hence, lim un = `1 ⇐⇒ lim |un − `1 | = 0.)

n→∞ n→∞ n→∞ n→∞

Proof. We prove the theorem for the case lim un = `1 ∈ R and lim vn = `2 ∈ R.
n→∞ n→∞

1. lim λun = λ`1 ⇐⇒ (∀ε > 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ |λun − λ`1 | < ε).
n→∞
ε
Note that: |λun − λ`1 | = |λ||un − `1 | < ε ⇐⇒ |un − `1 | < |λ| .
ε
We have lim un = `1 . Let n1 corresponds to |λ| > 0 and take n0 := n1 . Then,
n→∞
ε
n > n0 = n1 =⇒ |un − `1 | < =⇒ |λun − λ`1 | = |λ||un − `1 | < ε.
|λ|

2. lim (un +vn ) = `1 +`2 ⇐⇒ (∀ε > 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ |(un +vn )−(`1 +`2 )| < ε).
n→∞
Note that: |(un + vn ) − (`1 + `2 )| = |(un − `1 ) + (vn − `2 )| ≤ |(un − `1 )| + |(vn − `2 )| < ε if for example
|un − `1 | < 2ε and |vn − `2 | < 2ε .
We have lim un = `1 and lim vn = `2 . Then |un − `1 | < 2ε for n ≥ n1 and |vn − `2 | < 2ε for
n→∞ n→∞
n ≥ n2 . Choose n0 := max{n1 , n2 }. If n ≥ n0 , then |(un + vn ) − (`1 + `2 )| = |(un − `1 ) + (vn −
`2 )| ≤ |(un − `1 )| + |(vn − `2 )| < ε.
3. lim (un vn ) = `1 `2 ⇐⇒ (∀ε > 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ |un vn − `1 `2 | < ε).
n→∞
Note that:
• Since (un ) is convergent, then its bounded i.e., |un | ≤ A for some A > 0.
• |un vn −`1 `2 | = |un vn −un `2 +un `2 −`1 `2 | ≤ |un vn −un `2 |+|un `2 −`1 `2 | = |un ||vn −`2 |+|`2 ||un −`1 | ≤
ε
A|vn − `2 | + (|`2 | + 1)|un − `1 | < ε if |vn − `2 | < 2A and |un − `1 | < 2(|`2ε|+1) .
ε ε
We have lim un = `1 and lim vn = `2 . Then |un −`1 | < 2(|`2 |+1) for n ≥ n1 and |vn −`2 | < 2A
n→∞ n→∞
for n ≥ n2 . Choose n0 := max{n1 , n2 }. If n ≥ n0 , then|un vn − `1 `2 | < ε.

20
 4. It is sufficent to show:
lim (1/vn ) = 1/`2 ⇐⇒ (∀ε > 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ |1/vn − 1/`2 | < ε).
n→∞
Note that:
`2
• Since lim vn = `2 6= 0, then |vn | ≥ 2 for n ≥ N ∈ N.
n→∞
`2 −vn vn −`2 `22 ε
• |1/vn − 1/`2 | = vn `2 ≤2 `22
< ε if |vn − `2 | < 2 .
`22 ε
We have lim vn = `2 . Then |vn − `2 | < 2 for n ≥ n2 . Choose n0 := max{N, n2 }. If n ≥ n0 ,
n→∞
then|1/vn − 1/`2 | < ε.
5. lim |un | = |`1 | ⇐⇒ (∀ε > 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ |un | − |`1 | < ε).
n→∞
Note that: |un | − |`1 | ≤ |un − `1 |.
We have lim un = `1 . Then |un − `1 | < ε for n ≥ n1 . Choose n0 := n1 . If n ≥ n0 ,
n→∞
then |un | − |`1 | ≤ |un − `1 | < ε.

2.2 Limits And Ordered Relation

2.17 Property. If lim un = ` ∈ R and un ≥ 0 for n ≥ N ∈ N, then ` ≥ 0.

n→∞

−`
Proof. Assume to the contrary that ` < 0. We have lim un = `. Let n0 corresponds to ε := 2 > 0.
n→∞
−` `
If n ≥ n0 , then un − ` < 2 . Hence un < 2 < 0 eventually which is a contradiction. Thus ` ≥ 0.

2.18 Corollary. Let (un ) and (vn ) be convergent sequences such that un ≤ vn for n ≥ N ∈ N, then
lim un ≤ lim vn .
n→∞ n→∞

Proof. Apply the previous property to the sequence wn := vn − un

2.19 Remark. Let (un ) and (vn ) be convergent sequences.

1. ∀n ≥ n0 , un <vn =⇒ lim un ≤ lim vn .
n→∞ n→∞
for example un = − n1 < vn = 1
n and lim un = 0 = lim vn .
n→∞ n→∞
2. ∀n ≥ n0 , un ≤ c (c ∈ R) =⇒ lim un ≤ c.
n→∞

2.20 Property. (Sandwich Rule) Let (un ), (vn ) and (wn ) be sequences in R such that un ≤ vn ≤ wn
for all n ≥ N (N ∈ N) and lim un = lim wn = ` (` ∈ R). Then lim vn = `.
n→∞ n→∞ n→∞

Proof. lim vn = ` ⇐⇒ (∀ε > 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ ` − ε < vn < ` + ε).
n→∞
Let ε > 0, then by the definition of the limit:

∃n1 ∈ N, n ≥ n1 =⇒ ` − ε < un < ` + ε

∃n2 ∈ N, n ≥ n2 =⇒ |` − ε < wn < ` + ε.

Choose n0 := max{N, n1 , n2 }. If n ≥ n0 , then ` − ε < un ≤ vn ≤ wn < ` + ε.

21
 sin n
2.21 Example. Show that lim n = 0.
n→∞
Solution.
sin n
First Way: −1 ≤ sin n ≤ 1 =⇒ −1/n ≤ (sin n)/n ≤ 1/n =⇒ lim = 0.
n→∞ n
sin n
Second Way: To show lim n =0 is equivalent to show lim sinn n = 0. We have 0 ≤ sin n
n ≤ n1 ,
n→∞ n→∞
sin n
then by Sandwich rule lim n = 0.
n→∞

2.22 Property. If (un ) is a bounded sequence and lim vn = 0, then lim un vn = 0.

n→∞ n→∞

Proof. Since (un ) is a bounded sequence, then there exists (A > 0 such that |un | < A for all n. Hence,
0 ≤ |un vn | = |un ||vn | ≤ A|vn |. Then by Sandwich rule lim |un vn | = 0. Thus lim un vn = 0
n→∞ n→∞

2.23 Property. For any real number x there exist sequences (rn ) in Q and(in ) in R \ Q such that
lim rn = lim in = x.
n→∞ n→∞

Proof. By density of rationals and irrationals in R, for all n ∈ N∗ , there exist rn ∈ Q and in ∈ R \ Q
such that x − n1 < rn < x + n1 and x − n1 < in < x + n1 . The Sandwich rule then implies that
lim rn = lim in = x.
n→∞ n→∞

2.24 Property. Let (un ) and (vn ) be sequences in R such that un ≤ vn for all n ≥ N (N ∈ N).
1. If lim un = ∞, then lim vn = ∞.
n→∞ n→∞
2. If lim vn = −∞, then lim un = −∞.
n→∞ n→∞

Proof. 1. lim vn = ∞ ⇐⇒ (∀A > 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ vn > A).

n→∞
Let A > 0, then by the definition of the limit ∃n1 ∈ N, n ≥ n1 =⇒ un > A. Choose
n0 := max{N, n1 }. If n ≥ n0 , then vn ≥ un > A.
2. lim un = −∞ ⇐⇒ (∀A < 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ un < A).
n→∞
Let A < 0, then by the definition of the limit ∃n1 ∈ N, n ≥ n1 =⇒ vn < A. Choose
n0 := max{N, n1 }. If n ≥ n0 , then un ≤ vn < A.
n √
P √ √
For example: un = k> n. Since lim n = ∞, then lim un = ∞.
k=1 n→∞ n→∞

2.3 Monotone Sequences

2.25 Definition.
• A sequence (un ) in R is said to be increasing if ∀n ∈ N, un ≤ un+1 .
• A sequence (un ) in R is said to be strictly increasing if ∀n ∈ N, un < un+1 .
• A sequence (un ) in R is said to be decreasing if ∀n ∈ N, un ≥ un+1 .
• A sequence (un ) in R is said to be strictly decreasing if ∀n ∈ N, un > un+1 .
A sequence that is either increasing or decreasing is called monotone and a sequence that is either
strictly increasing or strictly decreasing is called strictly monotone.

22
 n−1 −2
For example un = n+1 is strictly increasing because ∀n ∈ N, un − un+1 = (n+1)(n+2) < 0.

2.26 Theorem. (Monotone Sequence Theorem)

Let (un ) be an increasing sequence. Then we have two possibilities:
1. If (un ) is bounded above, then it converges to sup un .
n∈N
2. If (un ) is not bounded above, then it diverges to sup un = ∞.
n∈N

Proof.
1. Since the set {un ; n ∈ N} is nonempty and bounded above, then sup un := sup{un ; n ∈ N} exists
n∈N
in R. Claim: n→∞
lim un = sup un ⇐⇒ (∀ε > 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ sup un − ε < un < sup un + ε).
n∈N n∈N n∈N
Let ε > 0, then by ε− characterization of supremum, there exists n0 ∈ N such that sup un − ε < un0 .
n∈N
If n ≥ n0 , then by the monotony of the sequence we have

sup un − ε < un0 ≤ un ≤ sup un < sup un + ε.

n∈N n∈N n∈N

2. Claim: lim un = ∞ ⇐⇒ (∀A > 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ un > A).

n→∞
Let A > 0. Since (un ) is not bounded above, then there exists n0 ∈ N such that un0 > A. If n ≥ n0 ,
then by the monotony of the sequence we have un ≥ un0 > A

2.27 Corollary. (Monotone Sequence Theorem)

Let (un ) be a decreasing sequence. Then we have two possibilities:
1. If (un ) is bounded below, then it converges to inf un .
n∈N
2. If (un ) is not bounded below, then it diverges to inf un = −∞.
n∈N

Proof. Apply the previous theorem to the sequence (−un ).

2.28 Property. Let (un ) be a sequence in R.

1.If (un ) is bounded above by ` and lim un = `, then sup un = `.
n→∞
2.If (un ) is bounded below by ` and lim un = `, then inf un = `.
n→∞

Proof.
1. First Way: We have un ≤ sup un ≤ `. Then by Sandwich rule sup un = `.
Second Way: (ε− characterization).
• We have ` is an upper bound of (un ).
• Claim: (∀ε > 0)(∃n0 ∈ N)(` − ε < un0 ).
Let ε > 0. Since lim un = `, then there exists n1 ∈ N such that n ≥ n1 =⇒ ` − ε < un < ` + ε.
n→∞
Choose n0 := n1 , then ` − ε < un0 .
2. First Way: We have ` ≤ inf un ≤ un . Then by Sandwich rule inf un = `.
Second Way: (ε− characterization).
• We have ` is an lower bound of (un ).
• Claim: (∀ε > 0)(∃n0 ∈ N)(un0 < ` + ε).
Let ε > 0. Since lim un = `, then there exists n1 ∈ N such that n ≥ n1 =⇒ ` − ε < un < ` + ε.
n→∞

23
Choose n0 := n1 , then un0 < ` + ε.
Third way: Apply 1. to the sequence (−un ).
n2
For example: The sequence un = n2 +1
is bounded above by 1 and lim un = 1, then sup un = 1.
n→∞

2.29 Property. (Sequential Characterization Of Supremum And Infimum)

1. Let A be a nonempty bounded above subset of R. Then ` = sup A if and only if ` is an upper
bound of A and ` is a limit of a sequence (an ) ⊂ A.
2. Let A be a nonempty bounded below subset of R. Then ` = inf A if and only if ` is a lower bound
of A and ` is a limit of a sequence (an ) ⊂ A.

Proof. Exercise.

For example: If A admits a supremum in R, then sup A is a limit of a sequence (an ) ⊂ A, and
if B ⊂ (0,
√∞) admits an infimum, then inf B is a limit of a sequence√ (bn ) ⊂ B, Then the√ set
C := {a− b : a ∈ A, b ∈ B} admits a√
supremum and sup C = sup A− inf B because
√ sup A− inf B
is an upper bound of C and sup A − inf B is the limit of the sequence (an − bn ) ⊂ C.

2.4 Sub Sequences

2.30 Definition. A sub sequence of a sequence (un ) is a part of (un ) denoted by (uϕ(n) ) where
ϕ : N 7→ N is a strictly increasing sequence.

For example (u2n ) is a sub sequence of (un ) because ϕ(n) = 2n is a strictly increasing sequence and
(u2n+1 ) is a sub sequence of (un ) because ϕ(n) = 2n + 1 is a strictly increasing sequence .

2.31 Property. The strictly increasing function ϕ : N 7→ N satisfies ϕ(n) ≥ n for all n.

Proof. By induction.

2.32 Remark. If (un ) is a bounded sequence, then any sub sequence (uϕ(n) ) is also bounded.

2.33 Property. If un −→ `, then any sub sequence uϕ(n) −→ `.

n→∞ n→∞

Proof. uϕ(n) −→ ` ⇐⇒ (∀ε > 0)(∃n0 ∈ N)(∀n ∈ N)(n ≥ n0 =⇒ |uϕ(n) − `| < ε).
n→∞
(?)
Let ε > 0. Since un −→ `, then there exists n1 ∈ N such that n ≥ n1 =⇒ |un − `| < ε.
n→∞
Choose n0 := n1 . If n ≥ n0 then ϕ(n) ≥ ϕ(n0 ) ≥ n0 = n1 and by the implication (?) we obtain
|uϕ(n) − `| < ε.

2.34 Corollary. 1 If (un ) has a divergent sub sequence, then (un ) is divergent.
2. If (un ) has two sub sequences converge to different limits, then (un ) is divergent.

24
2.35 Theorem. Every sequence (un ) has a monotonic sub sequence.

Proof. Let S := {n ∈ N : um < un for m > n}. We discuss two cases:

Case 1: S is finite. Let M := max S. Let n1 > M , then n1 ∈ / S and there exists n2 > n1 such
that un2 ≥ un1 . Now, n2 > n1 > M , then n2 ∈ / S and there exists n3 > n2 such that un3 ≥ un2 .
Proceeding in the same way, we obtain an increasing sub sequence (unk )∞k=1 .
Case 2: S is infinite. Let n1 = min S, n2 = min{S \ {n1 }} and so on. We obtain a strictly increasing
sequence (nk )∞ ∞
k=1 ⊂ S. Consequently (un ) has a decreasing sub sequence (unk )k=1 .

2.36 Theorem. (Bolzano-Weierstrass Theorem)

Every bounded sequence has a convergent sub sequence.

Proof. (un ) has a monotone sub sequence (unk ) . Since (un ) is bounded then (unk ) is also bounded.
Hence (unk ) is a monotone bounded sequence .Thus (unk ) is a convergent sub sequence of (un ) .
For example: The sequence (cos n) has a convergent sub sequence.

2.5 Exercises
Exercise 1. Using the definition of the limit of a sequence, prove that
√ 2n cos n sin n − 3n sin n cos2 n
s
n−1 5. lim = 0.
1. lim 1 + = 1. n→∞ n2 + 1
n→∞ n
√ √ 2n cos n sin n − 3n sin n cos2 n
2. lim ( n + 1 − n) = 0. 6. lim = 0.
n→∞ n→∞ n2 + 1 + sin n
√ n2 + (−1)n
n2 − n + 1 7. lim √ = ∞.
3. lim = 1. n→∞ n2 + n + 7
n→∞ n+1
n cos n n2
4. lim = 0. = −∞.
8. lim
n→∞ n2 + 1 n→∞ 4 − n
√ q
Exercise 2. Let (un ) ⊂ [0, ∞) with limit in R. Show that lim un = lim un .
n→∞ n→∞
Exercise 3. Let Let 0 ≤ c < 1 and A := {un = cn : n ∈ N∗ }.
1. Show that the sequence (un ) is a decreasing sequence.
2. Show that 0 ≤ cn < 1 for all n ∈ N∗ .
3. Show that lim un exists in R and lim un = 0.(Hint: un+1 = cun .)
n→∞ n→∞

4. Deduce the value of inf A.

5. Assume −1 < c < 1. Show that lim cn = 0.
n→∞
√ un +vn
Exercise 4. Let 0 < u0 < v0 and define un+1 = un vn and vn+1 = 2 . Show that
0 < un < un+1 < vn+1 < vn and that lim un = lim vn .
n→∞ n→∞
Exercise 5. Let (un ) be a sequence in R.
1.If (un ) is bounded above by ` and ∃ uϕ(n) sub sequence s.t lim uϕ(n) = `, then sup un = `.
n→∞
2.If (un ) is bounded below by ` and ∃ uϕ(n) sub sequence s.t lim uϕ(n) = `, then inf un = `.
n→∞
3. Let A := {un = (−1)n + 2n 1
: n ∈ N∗ }. Determine sup A and inf A.
Exercise 6. Show that the limit of the sequence (cos n) does not exist in R . (Use cos(2n) = 2 cos2 n − 1
and cos(n + 1) + cos(n − 1) = 2 cos n cos 1 or ∃mk , nk ∈ N s.t mk ∈ [2kπ − π6 , 2kπ + π6 ], nk ∈ [2kπ + π4 , 2kπ + 3π
4
].)

25
CHAPTER 3
LIMITS AND CONTINUITY ON R

3.1 Functions

f
3.1 Definition. Let D ⊆ R. A function f from D into R denoted by f : D 7→ R or D 7→ R, is a
rule that assigns to each x ∈ Da unique image in R denoted by f (x).
• D is called the domain of f .
• f (D) := {f (x) : x ∈ D} is called the image or range of f .
• Gf := {(x, f (x)) : x ∈ D} is called the graph of f .

For example: f : [0, ∞[7→ R defined by

√ y
f (x) = x is a function (by the uniqueness
of the square root). Gf
• The domain of f is D = [0, ∞[.
• The range of f is f (D) = [0, ∞[.
• The graph of f is f (D)
√
Gf := {(x, x) : x ∈ [0, ∞[}.

D = [0, ∞[ x

√
The function f (x) = x.

Here are some graphs of non functions.

y y f (x)
f (x)
f (x)
x x

f (x)
f (x)

x x
Graph of a non function. (x2 + y 2 = 4)
Graph of a non function

26
3.2 Definition. Let f : D 7→ R be a function. We say that

• D is symmetric with respect to the origin i.e.; x ∈ D =⇒ −x ∈ D.
1. f is even on D if
• f (−x) = f (x) ∀x ∈ D.

• D is symmetric with respect to the origin i.e.; x ∈ D =⇒ −x ∈ D.
2. f is odd on D if
• f (−x) = −f (x) ∀x ∈ D.

 • D is invariant under translations by ± T.i.e.,
3. f is periodic on D if ∃ T > 0 s.t (x ∈ D =⇒ x ± T ∈ D)
• f (x + T ) = f (x) ∀x ∈ D.

The period of f is the smallest positive number T (if exists) satisfying f (x+T ) = f (x) ∀x ∈ D.
If the period of f is T , then we say that f is T −periodic on D.

Note tha if D = R, then the first condition in the above three definitions is always satisfied. Also,
a constant function on R is periodic of any period T > 0 and thus has no period. Here are some
examples about even, odd and periodic functions.

y
y

− 5π −2π− 3π −π − π2 π π 3π 2π 5π
2 2 2 2 2

x 7→ tan x is π−periodic and odd on

x 7→ x2 is an even function on D = R or D = [−1, 1] D = ∪ ]− π
+ kπ, π
+ kπ[.
k∈Z 2 2
but not D = [−1, 1[ .

y y

1 1

x x
−3π− 5π−2π− 3π−π − π2 π π 3π 2π 5π 3π −3π− 5π−2π− 3π−π − π2 π π 3π 2π 5π 3π
2 2 2 2 2 2 2 2 2 2

−1 −1

x 7→ cos x is 2π−periodic and even on D = R. x 7→ sin x is 2π−periodic and odd on D = R.

3.3 Property. If f is T − periodic on D, then for all x ∈ D we have f (x + nT ) = f (x) ∀n ∈ Z.

proof. Exercise

27
3.4 Definition. Let f : D 7→ R be a function. We say that

1. f is increasing on D if; ∀x1 , x2 ∈ D, x1 < x2 =⇒ f (x1 ) ≤ f (x2 ).

2. f is strictly increasing on D if; ∀x1 , x2 ∈ D, x1 < x2 =⇒ f (x1 ) < f (x2 ).

3. f is decreasing on D if; ∀x1 , x2 ∈ D, x1 < x2 =⇒ f (x1 ) ≥ f (x2 ).

4. f is strictly decreasing on D if; ∀x1 , x2 ∈ D, x1 < x2 =⇒ f (x1 ) > f (x2 ).

5. f is monotone on D if it is either increasing or decreasing on D.

6. f is strictly monotone on D if it is either strictly increasing or strictly decreasing on D.

St. inc. function. Inc. function. St. dec. function. Dec. function.

Note that, if a function is strictly increasing then it is increasing and if a function is strictly decreasing
then it is decreasing as well.

3.5 Definition. Let f : D 7→ R be a function. We say that

1. f is bounded above on D if; ∃M ∈ R, ∀x ∈ D, f (x) ≤ M .

2. f is bounded below on D if; ∃m ∈ R, ∀x ∈ D, m ≤ f (x).

3. f is bounded on D if; ∃m, M ∈ R, ∀x ∈ D, m ≤ f (x) ≤ M .

Note that f is bounded on D if and only if it is bounded above and below on D.

For example:
? f (x) = sin x is bounded below on D = R by m = −1 and bounded above by M = 1.
? f (x) = cos x is bounded below on D = R by m = −1 and bounded above by M = 1.
? f (x) = ex is bounded below on R by m = 0 and it is not bounded above on R.
? f (x) = ex is bounded below on D =] − ∞, 3] by m = 0 and bounded above by M = e3 .
? f (x) = tan x is neither bounded below nor bounded above on D =] − π2 , π2 [.
? f (x) = | cos x| is bounded below on D = R by m = 0 and bounded above by M = 1.
If we let A := {f (x) : x ∈ D}, then
• f is bounded above on D if and only if A is bounded above.
• f is bounded below on D if and only if A is bounded below.
• f is bounded on D if and only if A is bounded.
All properties of bounded sets can be restated for bounded functions. In particular, Exercise 6.
Chapter one gives the following property

3.6 Property. A function f : D 7→ R is bounded on D if and only if ∃C > 0, ∀x ∈ D, |f (x)| ≤ C.

28
3.2 Limits
The notion of limit is conveniently described in terms of the notion of neighborhoods.

3.7 Definition. (Neighborhoods) Let r > 0.

1. Let a ∈ R, the set

Ir (a) :=]a − r, a + r[= {x ∈ R : |x − a| < r} is called an r− neighborhood of a.
Ir (a+ ) := [a, a + r[= {x ∈ R : 0 ≤ x − a < r} is called a right neighborhood of a.
Ir (a− ) :=]a − r, a] = {x ∈ R : 0 ≤ a − x < r} is called a left neighborhood of a.
Ir (a) \ {a} :=]a − r, a[∪]a, a + r[= {x ∈ R : 0 < |x − a| < r} is called a deleted
neighborhood of a.
Ir (a+ ) \ {a} :=]a, a + r[= {x ∈ R : 0 < x − a < r}.
Ir (a− ) \ {a} :=]a − r, a[= {x ∈ R : 0 < a − x < r}.

2. A neighborhood of ∞ is an interval of the form Ir (∞) := (r, ∞).

3. A neighborhood of −∞ is an interval of the form Ir (−∞) := (−∞, −r).

Using the conventions Ir (∞) \ {∞} = Ir (∞), I(a) = Ir (a) and I(a+ ) \ {a+ } = I(a+ ) \ {a}, the
following definition of limit is sufficiently general to include the usual limits encountered in calculus:
two-sided limits, one-sided limits and limits at infinity.
The symbol α will denote any of a, a+ , a− , ∞, −∞.

3.8 Definition. Let f be a real valued function defined on some I(α) \ {α} and ` ∈ R. We say that
f (x) tends to ` as x tends to α and we write lim f (x) = ` if
x→α

∀I(`), ∃I(α), ∀x ∈ D, x ∈ I(α) \ {α} =⇒ f (x) ∈ I(`).

We tabulate I(`) in various cases:

` `∈R `=∞ ` = −∞
I(`) (` − ε, ` + ε) (M, ∞), M > 0 (−∞, M ), M < 0
f (x) ∈ I(`) |f (x) − `| < ε f (x) > M f (x) < M
In analogy to sequences we start with the case α = ∞.

I(α) \ {α} = (A, ∞), A > 0

a. The case α = ∞.
x ∈ I(α) \ {α} ⇐⇒ x > A
 
1. lim f (x) = ` ⇐⇒ ∀ε > 0, ∃A > 0, ∀x ∈ D, x > A =⇒ |f (x) − `| < ε .
x→∞
 
2. lim f (x) = ∞ ⇐⇒ ∀M > 0, ∃A > 0, ∀x ∈ D, x > A =⇒ f (x) > M .
x→∞
 
3. lim f (x) = −∞ ⇐⇒ ∀M < 0, ∃A > 0, ∀x ∈ D, x > A =⇒ f (x) < M .
x→∞

I(α) \ {α} = (−∞, A), A < 0

b. The case α = −∞.
x ∈ I(α) \ {α} ⇐⇒ x < A

29
  
1. lim f (x) = ` ⇐⇒
∀ε > 0, ∃A < 0, ∀x ∈ D, x < A =⇒ |f (x) − `| < ε .
x→−∞
 
2. lim f (x) = ∞ ⇐⇒ ∀M > 0, ∃A < 0, ∀x ∈ D, x < A =⇒ f (x) > M .
x→−∞
 
3. lim f (x) = −∞ ⇐⇒ ∀M < 0, ∃A < 0, ∀x ∈ D, x < A =⇒ f (x) < M .
x→−∞

I(a) \ {a} = (a − δ, a) ∪ (a, a + δ), δ > 0

c. The case α = a ∈ R.
x ∈ I(a) \ {a} ⇐⇒ 0 < |x − a| < δ ⇐⇒ (x 6= a ∧ |x − a| < δ)
 
1. lim f (x) = ` ⇐⇒ ∀ε > 0, ∃δ > 0, ∀x ∈ D, 0 < |x − a| < δ =⇒ |f (x) − `| < ε .
x→a
 
2. lim f (x) = ∞ ⇐⇒ ∀M > 0, ∃δ > 0, ∀x ∈ D, 0 < |x − a| < δ =⇒ f (x) > M .
x→a
 
3. lim f (x) = −∞ ⇐⇒ ∀M < 0, ∃δ > 0, ∀x ∈ D, 0 < |x − a| < δ =⇒ f (x) < M .
x→a

I(a+ ) \ {a} = (a, a + δ), δ > 0

d. The case α = a+ .(right-hand limit)
x ∈ I(a+ ) \ {a} ⇐⇒ 0 < x − a < δ
 
1. lim f (x) = ` ⇐⇒ ∀ε > 0, ∃δ > 0, ∀x ∈ D, 0 < x − a < δ =⇒ |f (x) − `| < ε .
x→a+
 
2. lim f (x) = ∞ ⇐⇒ ∀M > 0, ∃δ > 0, ∀x ∈ D, 0 < x − a < δ =⇒ f (x) > M .
x→a+
 
3. lim f (x) = −∞ ⇐⇒ ∀M < 0, ∃δ > 0, ∀x ∈ D, 0 < x − a < δ =⇒ f (x) < M .
x→a+

I(a− ) \ {a} = (a − δ, a), δ > 0

e. The case α = a− .(left-hand limit)
x ∈ I(a− ) \ {a} ⇐⇒ 0 < a − x < δ
 
1. lim f (x) = ` ⇐⇒ ∀ε > 0, ∃δ > 0, ∀x ∈ D, 0 < a − x < δ =⇒ |f (x) − `| < ε .
x→a−
 
2. lim f (x) = ∞ ⇐⇒ ∀M > 0, ∃δ > 0, ∀x ∈ D, 0 < a − x < δ =⇒ f (x) > M .
x→a−
 
3. lim f (x) = −∞ ⇐⇒ ∀M < 0, ∃δ > 0, ∀x ∈ D, 0 < a − x < δ =⇒ f (x) < M .
x→a−

How to show lim f (x) = `?

x→a
(?)
 
lim f (x) = ` ⇐⇒ ∀ε > 0, ∃δ > 0, ∀x ∈ D \ {a}, |x − a| < δ =⇒ |f (x) − `| < ε .
x→a

• We let ε > 0 be arbitrary.

• We search for δ = δ(ε) which may depends on ε such that the implication (?) holds true.
How to find δ s.t (?) holds true?(How to find at least one δ−neighborhood of a s.t (?) holds true? )
We will simplify matters by making estimates: For |x − a| < δ given,

|f (x) − `| ≤ · · · ≤ |x − a| 2 ( or
1 1
α>0)
( expression in x bounded above by c) ≤ cδ 2 < ε
ε2 1 ε2 ε2
If δ < c2
. So we choose δ = 2 c2 < c2
.

30
3.9 Example. Let f (x) = x2 + 3. Using the definition of the limit, show that lim f (x) = 4.
x→1

Solution. ∀ε > 0, ∃δ > 0, ∀x ∈ D \ {1}, |x − 1| < δ =⇒ |f (x) − 4| < ε.

Let ε > 0. We search for δ > 0 such that;
for x ∈ D = R, x 6= 1 the implication |x − 1| < δ =⇒ |f (x) − 4| < ε holds true.
For |x − 1| < δ given i.e., 1 − δ < x < 1 + δ,
|f (x) − 4| = |x2 − 1| = |x − 1||x + 1| (1 − δ < x < 1 + δ =⇒ 2 − δ < x + 1 < 2 + δ)
= |x − 1|(x + 1) ( we drop absolute value for 0 < 2 − δ i.e., δ < 2)
≤ δ4 (δ < 2 =⇒ x + 1 < 2 + δ < 4)
ε
<ε (4δ < ε if δ < ).
4
Choose 0 < δ < min{2, 4ε }, for example δ = 1
2 × min{2, 4ε }.
Here another estimate:
|f (x) − 4| = |x2 − 1| = |x − 1||x + 1| = |x − 1||x − 1 + 2|
≤ |x − 1|(|x − 1| + |2|) (using triangle inequality)
< δ(δ + 2) (∵ |x − 1| < δ)
< 3δ ( assume δ < 1)
ε
<ε (3δ < ε if δ < ).
3
Choose 0 < δ < min{1, 3ε }, for example δ = 1
2 × min{1, 3ε }.
√
x− 2x−1
3.10 Example. Let f (x) = x−1 . Using the definition of the limit, show that lim f (x) = 0.
x→1

Solution. ∀ε > 0, ∃δ > 0, ∀x ∈ D \ {1}, |x − 1| < δ =⇒ |f (x) − 0| < ε.

Let ε > 0. We search for δ > 0 such that;
for x ∈ D = [ 21 , 1[∪]1, ∞[, x 6= 1 the implication |x − 1| < δ =⇒ |f (x)| < ε holds true.
For |x − 1| < δ given i.e., 1 − δ < x < 1 + δ,
√ √ √
x − 2x − 1 x − 2x − 1 x + 2x − 1
|f (x)| = = × √ (try to get |x − 1| in numerator)
x−1 x−1 x + 2x − 1
(x − 1)2
= √ ( cancel by |x − 1|)
(x − 1)(x + 2x − 1)
1 1 √ 1
= |(x − 1)| × √ x≥ =⇒ x + 2x − 1 ≥
x + 2x − 1 2 2
ε
≤ 2δ < ε 2δ < ε if δ < .
2
Choose 0 < δ < 2ε , for example δ = 1
2 × ε
2 = 4ε .
How to show lim f (x) = `?
x→∞
(?)
 
lim f (x) = ` ⇐⇒ ∀ε > 0, ∃A > 0, ∀x ∈ D, x > A =⇒ |f (x) − `| < ε
x→∞

• We let ε > 0 be arbitrary.

• We search for A = A(ε) which may depends on ε such that the implication (?) holds true.
How to find A s.t (?) holds true?(How to find at least one nbhd ]A, ∞[ of ∞ s.t (?) holds true? )
We will simplify matters by making estimates: For x > A given,

|f (x) − `| ≤ · · · ≤ θ(x) < ε

where θ(x) −→ 0 and the inequality θ(x) < ε is easy to solve.

x→∞
For example, θ(x) = √1 < ε if x > 1
. So we choose A > 1
. Hence x > A > 1
implies θ(x) < ε.
x ε2 ε2 ε2

31
 4x2 +3x−6
3.11 Example. Let f (x) = x2 −4
. Using the definition of the limit, show that lim f (x) = 4.
x→∞

(?)
Solution. ∀ε > 0, ∃A > 0, ∀x ∈ D =] − ∞, −2[∪] − 2, 2[∪]2, ∞[, x > A =⇒ |f (x) − `| < ε.
Let ε > 0, we search for A such that the implication (?) holds true.

4x2 + 3x − 6 3x + 10
|f (x) − `| = −4 =
x2 − 4 x2 − 4
3x + 10
= 2 (we drop the absolute value for x > 2)
x −4
3x + 10 √
≤ (x2 − 4 > x2 /2 if x > 8)
x2 /2
3x + 10x
≤ (10 < 10x for x > 1)
x2 /2
26
≤ ( cancel by x)
x
26
<ε ( for x > ).
ε
√
So we choose A > max{ 26
ε , 1, 2, 8}. √
26
Hence x > A implies x > 1, x > 2, x > 8, x > ε and |f (x) − `| < ε.
How to show lim f (x) = ∞?
x→∞
(?)
 
lim f (x) = ∞ ⇐⇒ ∀M > 0, ∃A > 0, ∀x ∈ D, x > A =⇒ f (x) > M
x→∞

• We let M > 0.
• We search for A = A(M ) which may depends on M such that the implication (?) holds true.
How to find A s.t (?) holds true?(How to find at least one nbhd ]A, ∞[ of ∞ s.t (?) holds true? )
We will simplify matters by making estimates: For x > A given,

f (x) ≥ · · · ≥ θ(x) > M

where θ(x) −→ ∞ and the inequality θ(x) > M is easy to solve.

x→∞

x2 +(−1)E(x)
3.12 Example. Let f (x) = √
x2 +x+7
. Using the definition of the limit, show that lim f (x) = ∞.
x→∞

(?)
Solution. ∀M > 0, ∃A > 0, ∀x ∈ D = R, x > A =⇒ f (x) > M .
Let M > 0, we search for A such that the implication (?) holds true.

x2 + (−1)E(x) x2 + −1
f (x) = √ ≥√ ((−1)E(x) = ±1 ≤ 1)
x2 + x + 7 x2 + x + 7
x2 /2 + x2 /2 − 1
≥ √ (x ≤ x2 , 7 ≤ 7x2 for 1 < x)
x2 + x2 + 7x2
2
x /2
≥ (x2 /2 − 1 > 0 for 2 < x)
3|x|
x
≥ (|x| = x for 0 < x)
6
>M ( for 6M < x).

So we choose A > max{6M, 0, 1, 2} = max{6M, 2}.

Hence x > A implies x > 0, x > 1, x > 2, x > 6M and f (x) > M .

32
How to show lim f (x) = ∞?
x→a
(?)
 
lim f (x) = ∞ ⇐⇒ ∀M > 0, ∃δ > 0, ∀x ∈ D \ {a}, |x − a| < δ =⇒ f (x) > M .
x→a

• We let M > 0.
• We search for δ = δ(ε) which may depends on ε such that the implication (?) holds true.
How to find δ s.t (?) holds true?(How to find at least one δ−neighborhood of a s.t (?) holds true? )
We will simplify matters by making estimates: For |x − a| < δ given,

( expression in x bounded below by positve c) c

f (x) ≥ · · · ≥ ≥ 1 >M
1
|x − a| 2 ( or α>0)
δ2
c2 1 c2 c2
If δ < M2
. So we choose δ = 2 M2 < M2
.

|x−1|
3.13 Example. Let f (x) = √
x− 2x−1
. Using the definition of the limit, show that lim f (x) = ∞.
x→1

(?)
Solution. ∀M > 0, ∃δ > 0, ∀x ∈ D \ {1} = [ 21 , 1[∪]1, ∞[, |x − 1| < δ =⇒ f (x) > M .
Let M > 0. We search for δ > 0 such that the implication (?) holds true.
For |x − 1| < δ given i.e., 1 − δ < x < 1 + δ,
√
|x − 1| |x − 1| x + 2x − 1
f (x) = √ = √ × √ (try to get |x − 1|α>0 in denominator)
x − 2x − 1 x − 2x − 1 x + 2x − 1
√
|x − 1|(x + 2x − 1)
= ( cancel by |x − 1|)
(x − 1)2
1 √ √ √
= × (x + 2x − 1) x + 2x − 1 ≥ 1 − δ + 1 − 2δ
|x − 1|
1 √ √ 1 1
≥ (1 − δ + 1 − 2δ) 1 − δ + 1 − 2δ > if δ <
δ 2 2
1 1 1 1
≥ >M > M if δ < .
δ 2 2δ 2M
1 1 1 1
Choose 0 < δ < 2M , for example δ = 2 × 2M = 4M .

3.14 Remark. We discussed four cases, the other five cases can be transformed into these cases by
the change of variable t = −x in the case x → −∞ or by computing lim(−f (x)) in case f (x) → −∞.
x−1
3.15 Example. Let f (x) = √
x− 2x−1
. Using the definition of the limit, show that lim f (x) = ∞.
x→1+

(?)
Solution. ∀M > 0, ∃δ > 0, ∀x ∈ D \ {1} = [ 21 , 1[∪]1, ∞[, 0 < x − 1 < δ =⇒ f (x) > M .
Let M > 0. We search for δ > 0 such that the implication (?) holds true.
For 0 < x − 1 < δ given i.e., 1 < x < 1 + δ,
√
x−1 x−1 x + 2x − 1
f (x) = √ = √ × √ (try to get (x − 1)α>0 in denominator)
x − 2x − 1 x − 2x − 1 x + 2x − 1
√
(x − 1)(x + 2x − 1)
= ( cancel by (x − 1))
(x − 1)2
1 √ √
= × (x + 2x − 1) x > 1 =⇒ x + 2x − 1 > 2
x−1
2 2
≥ >M 2δ < ε if δ < .
δ M
2
Choose 0 < δ < M.

33
3.16 Remark. Infact, all cases can be transformed to lim f (x) = 0
x→0±

3.17 Property. (Uniqueness of the limit)

Let ` ∈ R. If lim f (x) = `, then f admits no other limit for x tends to α.
x→α

Proof. We prove the property for the case α = a ∈ R and leave the other cases for the reader.
Suppose that f has two limits `1 and `2 and let us show `1 = `2 . We consider only the case where
`1 and `2 are both finite. The equality will follow if we show that ∀ε > 0, |`1 − `2 | < ε. Let ε > 0,
then by the definition of the limit:
ε
∃δ1 > 0, ∀x ∈ D, 0 < |x − a| < δ1 =⇒ |f (x) − `1 | <
2
ε
∃δ2 > 0, ∀x ∈ D, 0 < |x − a| < δ2 =⇒ |f (x) − `2 | <
2
For δ3 < min{δ1 , δ2 }, x0 ∈ D and 0 < |x0 − a| < δ3 and with the help of triangle inequality we obtain
|`1 − `2 | = |`1 − f (x0 ) + f (x0 ) − `2 | ≤ |f (x0 ) − `1 | + |f (x0 ) − `2 | < 2ε + 2ε = ε.

3.18 Property. Let f be a function defined on I \ {a} for some open interval I containing a and
` ∈ R. Then
lim f (x) = ` ⇐⇒ lim f (x) = ` = lim f (x).
x→a x→a− x→a+

Proof. Exercise.

3.19 Property. (Sequential Characterization of Limit.)

Let ` ∈ R and f defined on I(α) \ {α}. Then
 
lim f (x) = ` ⇐⇒ for every sequence (xn ) in I(α) \ {α}, xn −→ α =⇒ f (xn ) −→ ` .
x→α n→∞ n→∞

Proof. We prove the property for the case α = a ∈ R and ` ∈ R. The proofs for the other cases are
similar, adapting the same argument.
⇒) Assume lim f (x) = ` and let (xn ) ⊂ I(a) \ {a} s.t xn −→ a. Claim f (xn ) −→ `:
x→a n→∞ n→∞
Let ε > 0. Since lim f (x) = `, then
x→a

∃δ > 0, ∀x ∈ D, 0 < |x − a| < δ =⇒ |f (x) − `| < ε. (?)

Since xn −→ a and δ > 0, then
n→∞

∃n0 ∈ N, n ≥ n0 =⇒ |xn − a| < δ.

Hence,
using (?)
∃n0 ∈ N, n ≥ n0 =⇒ 0 < |xn − a| < δ =⇒ |f (xn ) − `| < ε.
Thus f (xn ) −→ `.
n→∞
⇐) Assume the seq condition; for every sequence (xn ) in I(a)\{a}, xn −→ a =⇒ f (xn ) −→ `.
n→∞ n→∞
Suppose to the contrary that limx→a f (x) 6= ` i.e.,
∃ε > 0, ∀δ > 0, ∃x ∈ D = I(a) \ {a}, |x − a| < δ ∧ |f (x) − `| ≥ ε.
Let δ = n1 . Then ∃xn ∈ D = I(α) \ {α}, |xn − a| < n1 ∧ |f (xn ) − `| ≥ ε. Thus xn −→ a and
n→∞
f (xn ) −→
X ` which contradicts the sequential condition.
n→∞

34
3.20 Remark. If there exist two sequences (xn ) and (yn ) in D \ {α} s.t xn −→ α ∧ yn −→ α but
n→∞ n→∞
lim f (xn ) 6= lim f (yn ), then lim f (x) does not exist (in R).
n→∞ n→∞ x→a

3.21 Examples.
1. Show that lim sin x does not exist.
x→∞
π
Solution. xn = 2πn −→ ∞ ∧ yn = 2πn + −→
2 n→∞ ∞ but lim sin(xn ) = 0 6= lim sin(yn ) = 1.
n→∞ n→∞ n→∞

2. Show that lim cos x does not exist.

x→∞
π
Solution. xn = 2πn −→ ∞ ∧ yn = 2πn + −→
2 n→∞ ∞ but lim cos(xn ) = 1 6= lim cos(yn ) = 0.
n→∞ n→∞ n→∞

1
3. Show that lim cos does not exist.
x→0+ x
1 1 + but lim cos( 1 ) = 1 6= lim cos( 1 ) = 0.
Solution. xn = −→ 0+ ∧ yn = 2πn+
2πn n→∞ π −→ 0
xn yn
2 n→∞ n→∞ n→∞

4. Show that lim cos(ln x) does not exist.

x→∞
π
Solution. xn = e2πn −→ ∞ ∧ yn = e2πn+ 2 −→ ∞ but lim cos(ln xn ) = 1 6= lim cos(ln yn ).
n→∞ n→∞ n→∞ n→∞

3.22 Property. Assume lim f (x) = ` .

x→α

1. If ` is finite, then f is bounded on some deleted neighborhood Ir (α) \ {α}.

|`|
2. If ` is finite and ` 6= 0 , then |f (x)| ≥ 2 on some deleted neighborhood Ir (α) \ {α}.

Proof. Exercise.

3.23 Property. Let f and g be functions defined on I(α) \ {α}. The following limit properties hold
in R in the sense that if the expression on the right side of the equation exists in R, then the limit
on the left side exists in R and the equality holds.

1. lim λf (x) = λ lim f (x), λ ∈ R.

x→α x→α

2. lim (f (x) + g(x)) = lim f (x) + lim g(x).

x→α x→α x→α

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

x→α x→α x→α

4. lim (f (x)/g(x)) = lim f (x)/ lim g(x) provided lim g(x) 6= 0.

x→α x→α x→α x→α

5. lim |f (x)| = | lim f (x)|. (Hence, lim f (x) = `1 ⇐⇒ lim |f (x) − `1 | = 0.)
x→α x→α x→α x→α

Proof. The properties follows immediately from the sequential characterization of limit property
(3.19) and and the algebra of limits of sequences (2.16). However, it is instructive to formulate direct
proofs. We do the two methods for part (4.).
4. Using the sequential characterization of limit:
Let (xn ) be a sequence in I(α) \ {α} s.t xn −→ α. By hypotheses lim g(xn ) 6= 0. Hence,
n→∞ n→∞

lim (f (x)/g(x)) = lim (f (xn )/g(xn )) = lim f (xn )/ lim g(xn ) = lim f (x)/ lim g(x).
x→α n→∞ n→∞ n→∞ x→α x→α

Using the direct method for the case lim f (x) = `1 ∈ R and lim g(x) = `2 ∈ R? :
x→a x→a

35
 Let ε > 0.
Then |f (x) − `1 | < kε on Iδ1 (a) \ {a} and |g(x) − `2 | < kε on Iδ2 (a) \ {a} for k > 0. Now,
f (x) `1 `2 f (x) − `1 g(x) |`2 f (x)−`2 `1 + `2 `1 − `1 g(x)|
− = = ( we add and subtract `2 `1 )
g(x) `2 `2 g(x) |`2 g(x)|
|`2 f (x) − `2 `1 | + |`2 `1 − `1 g(x)|
≤ ( using triangle inequality )
|`2 ||g(x)|
|`2 ||f (x) − `1 | + |`1 ||`2 − g(x)|
≤ ( on Ir (a) \ {a}, property(3.22))
|`2 ||`2 /2|
|`2 |kε + |`1 |kε  2[|` | + |` |] 
2 1
≤ =k ε ( on Imin{r,δ1 ,δ2 } (a) \ {a})
|`2 ||`2 /2| `22
`22
≤ ε. ( take k = ).
2[|`2 | + |`1 |]
The proofs for the other cases are similar.
5. and 3. Exercises.

3.24 Property. Let f and g be functions defined on I(α) \ {α} s.t f (x) ≤ g(x) ∀x ∈ I(α) \ {α}.

1. If lim f (x) = `1 and lim g(x) = `2 , then `1 ≤ `2 .

x→α x→α

2. If lim f (x) = ∞, then lim g(x) = ∞.

x→α x→α

3. If lim g(x) = −∞, then lim f (x) = −∞.

x→α x→α

Proof. Using the sequential characterization of limit:

Let (xn ) be a sequence in I(α) \ {α} s.t xn −→ α. By hypotheses we have f (xn ) ≤ g(xn ). Then the
n→∞
three items follows from the properties of limits and ordered relation on sequences.
Using the direct method:
1. The case α = a.
Require to show; `1 ≤ `2 ⇐⇒ `1 − `2 ≤ 0 ⇐⇒ ∀ε > 0, `1 − `2 < ε.
Let ε > 0.
Then |f (x) − `1 | < kε on Iδ1 (a) \ {a} and |g(x) − `2 | < kε on Iδ2 (a) \ {a} for k > 0. Now,
`1 − `2 = `1 −f (x) + f (x) − g(x) + g(x) − `2
≤ (`1 − f (x)) + (g(x) − `2 ) (∵ f (x) − g(x) ≤ 0)
≤ |`1 − f (x)| + |g(x) − `2 | (∵ ∀X, X ≤ |X|)
< 2kε ( on Imin{δ1 ,δ2 } (a) \ {a})
1
< ε. ( take k = ).
2
The proofs for the other cases are similar.
2. The case α = ∞.
 
lim f (x) = ∞ =⇒ ∀M > 0, ∃A > 0, ∀x ∈ D, x > A =⇒ f (x) > M
x→∞
 
=⇒ ∀M > 0, ∃A > 0, ∀x ∈ D, x > A =⇒ g(x) ≥ f (x) > M
=⇒ lim g(x) = ∞.
x→∞
The proofs for the other cases are similar.

36
 3. The case α = −∞.
 
lim g(x) = −∞ =⇒ ∀M < 0, ∃A < 0, ∀x ∈ D, x < A =⇒ g(x) < M
x→−∞
 
=⇒ ∀M < 0, ∃A < 0, ∀x ∈ D, x < A =⇒ f (x) ≤ g(x) < M
=⇒ lim f (x) = −∞.
x→−∞

The proofs for the other cases are similar.

3.25 Property. (Sandwich Rule) Let f, g and h be functions defined on I(α)\{α} s.t f (x) ≤ g(x) ≤
h(x) ∀x ∈ I(α) \ {α}. If lim f (x) = ` and lim h(x) = ` then lim g(x) = `.
x→α x→α x→α

Proof. The cases ` = ±∞ done (previous property).

Using the sequential characterization of limit:
Let (xn ) be a sequence in I(α) \ {α} s.t xn −→ α. By hypotheses we have f (xn ) ≤ g(xn ) ≤ h(xn ).
n→∞
Then by the Sandwich Rule for sequences, we obtain lim g(xn ) = `. Thus lim g(x) = `.
n→∞ x→α
Using the direct method: The case α = a.
Let ε > 0. Then ` − ε < f (x) < ` + ε for 0 < |x − a| < δ1 and ` − ε < h(x) < ` + ε for 0 < |x − a| < δ2 .
Let δ = min{δ1 , δ2 }. Then for 0 < |x − a| < δ we have ` − ε < f (x) ≤ g(x) ≤ h(x) < ` + ε.
The proofs for the other cases are similar.
3.26 Remark. Let f and g be functions defined on I(α)\{α} . If f (x) is bounded and lim g(x) = 0,
x→α
1
then lim (f (x)g(x)) = 0. For example lim (x sin ) = 0.
x→α x→0 x
Solution. Since f is bounded on a deleted neighborhood I(α) \ {α}, then ∃M > 0 s.t ∀x ∈ I(α) \
{α}, |f (x)| ≤ M . Hence 0 ≤ |f (x)g(x)| ≤ M |g(x)| ∀x ∈ I(α) \ {α}. By Sandwich rule we obtain
lim |f (x)g(x)| = 0. Consequently, lim (f (x)g(x)) = 0.
x→α x→α

3.27 Theorem. (Monotone Function Theorem) Let a, b ∈ R and f :]a, b[7→ R.

1. If f is increasing, then

lim f (x) = inf{f (x) : a < x < b} ∈ R and lim f (x) = sup{f (x) : a < x < b} ∈ R.
x→a+ x→b−

2. If f is decreasing, then

lim f (x) = sup{f (x) : a < x < b} ∈ R and lim f (x) = inf{f (x) : a < x < b} ∈ R.
x→a+ x→b−

Proof. Exercise.

3.28 Corollary. Let a, b ∈ R, a < c < b and f :]a, b[7→ R. If f is monotone on ]a, b[ , then lim f (x)
x→c±
exist in R.

Proof. Exercise.

37
3.3 Continuous Functions

3.29 Definition. Let I ⊆ R be open interval, a ∈ I and f : I 7→ R be a function. We say that:

1. f is continuous at a if lim f (x) = f (a). Using quantifiers,

x→a

∀ε > 0, ∃δ > 0, ∀x ∈ I, |x − a| < δ =⇒ |f (x) − f (a)| < ε.

2. f is continuous on I if f is continuous at each point of I. Using quantifiers,

∀a ∈ I, ∀ε > 0, ∃δ > 0, ∀x ∈ I, |x − a| < δ =⇒ |f (x) − f (a)| < ε.

3.30 Examples. Using the definition of limit,

x sin x1

if x 6= 0
1. Show that f (x) = is continuous at x = 0.
0 if x = 0.
Solution.  
lim f (x) = f (0) ⇐⇒ ∀ε > 0, ∃δ > 0, ∀x ∈ R, |x − 0| < δ =⇒ |f (x) − f (0)| < ε .
x→0
Let ε > 0. For |x| < δ given,
|f (x) − f (0)| = |x|| sin x1 | ≤ |x| < δ < ε if we chose 0 < δ < ε. Thus f is continuous at x = 0.

2. Show that f (x) = x2 is continuous on R.

Solution.  
f is continuous on R ⇐⇒ ∀a ∈ R, ∀ε > 0, ∃δ > 0, ∀x ∈ R, |x − a| < δ =⇒ |f (x) − f (a)| < ε .
Let a ∈ R and ε > 0. For |x − a| < δ given,

|f (x) − f (a)| = |x2 − a2 | = |x − a||x + a|

< δ|x − a + 2a| (we add a and subtracta)
< δ(|x − a| + |2a|) (using triangle inequality)
< δ(δ + |2a|) (∵ |x − a| < δ)
< δ(1 + |2a|) (assume δ < 1)
ε
<ε (δ(1 + |2a|) < ε if δ < )
1 + |2a|
ε
Choose 0 < δ < min{1, 1+|2a| }.

3. Show that f (x) = |x| is continuous on R.

Solution.  
f is continuous on R ⇐⇒ ∀a ∈ R, ∀ε > 0, ∃δ > 0, ∀x ∈ R, |x − a| < δ =⇒ |f (x) − f (a)| < ε .
Let a ∈ R and ε > 0. For |x − a| < δ given,

|f (x) − f (a)| = ||x| − |a|| ≤ |x − a| < δ < ε ( if δ < ε)

Choose 0 < δ < ε.

38
3.31 Definition. Let I ⊆ R be an interval of any type, a ∈ I and f : I 7→ R. We say that:

1. f is continuous from right at a if lim f (x) = lim f (x) = f (a). Using quantifiers,
x→a+ x→a
x>a

∀ε > 0, ∃δ > 0, ∀x ∈ I, 0 < x − a < δ =⇒ |f (x) − f (a)| < ε.

2. f is continuous from left at a if lim f (x) = lim f (x) = f (a). Using quantifiers,
x→a− x→a
x<a

∀ε > 0, ∃δ > 0, ∀x ∈ I, 0 < a − x < δ =⇒ |f (x) − f (a)| < ε.

All properties of limit of functions can be restated for continuous functions. By continuity at a+
we mean continuity from right at a and by continuity at a− we mean continuity from left at a. Also
xn −→ a+ means xn −→ a and xn > a eventually. Analogously, xn −→ a− means xn −→ a and
n→∞ n→∞ n→∞ n→∞
xn < a eventually.

3.32 Property. Assume α ∈ {a, a+ , a− } .

1. Let f, g be continuous at α and λ ∈ R . Then

(a) λf is continuous at α.
(b) f + g is continuous at α.
(c) f g is continuous at α.
f
(d) is continuous at α provided that g(a) 6= 0.
g
2. f is continuous at a ⇐⇒ f is continuous from right and from left at a.
 
3. f is continuous at α ⇐⇒ for every sequence (xn ) in I, xn −→ α =⇒ f (xn ) −→ f (a) .
n→∞ n→∞

Constant functions and the function f (x) = x are clearly continuous on R. It follows from
previous property that polynomials are continuous on R and rational functions are continuous on
their domain of definition.

3.33 Property. Let f : I 7→ R be continuous at a and g defined in a neighborhood of f (a) and

continuous at f (a). Then the composite g ◦ f is continuous at a.

 
Proof. g ◦ f is continuous at a ⇐⇒ ∀(xn ) ⊂ I, lim xn = a =⇒ lim g(f (xn )) = g(f (a)) .
n→∞ n→∞
Let (xn ) ⊂ I s.t xn −→ a.
n→∞

xn −→ a =⇒ yn = f (xn ) −→ f (a) (∵ f is continuous at a)

n→∞ n→∞
=⇒ g(yn ) = g(f (xn )) −→ g(f (a)) (∵ g is continuous at f (a))
n→∞

39
Using ε − δ: Claim g ◦ f (x) −→ g(f (a)).
x→a
Let ε > 0. since g is continuous at f (a), then there exist δ1 > such that
(?)
|y − f (a)| < δ1 =⇒ |g(y) − g(f (a))| < ε.
Since δ1 and f is continuous at a, then then there exist δ > such that
(??)
|x − a| < δ =⇒ |f (x) − f (a)| < δ1 .
Hence, there exists δ > 0 such that
(??) (?)
∀x ∈ Dg◦f , |x − a| < δ =⇒ |f (x) − f (a)| < δ1 =⇒ |g(f (x)) − g(f (a))| < ε.

Note that; x ∈ Dg◦f ⇐⇒ (x ∈ Df ∧ f (x) ∈ Dg ). The following property will be frequently used.

3.34 Property. All elementary functions ”polynomials, rational functions, power functions,
trigonometric functions, exponential functions, hyperbolic functions ” and their inverses are
continuous over their entire domains.(Chapter Five)

 sin x + ex

if x < 0
3.35 Example. Show that f (x) = 1 if x = 0 is continuous on R.
 3
x + x2 + | cos x| if x > 0.
Solution.
• f is continuous ] − ∞, 0[ as sum of continuous functions.
• f is continuous ]0, ∞[ as composite, product and sum of continuous functions.
• Continuity of f at 0:
lim f (x) = lim f (x) = lim (sin x + ex ) = 1 = f (0). Hence f is continuous from left at 0.
x→0− x→0 x→0
x<0

lim f (x) = lim f (x) = lim (x3 + x2 + | cos x|) = 1 = f (0). Hence f is continuous from right at 0.
x→0+ x→0 x→0
x>0
Therefore f is continuous at 0 and thus f is continuous on R.

3.36 Definition. A function f : [a, b] 7→ R is continuous on [a, b] closed if the following holds
• f is continuous on ]a, b[.
• f is continuous from right at a.
• f is continuous from left at b.

Similarly, f is continuous on [a, b[ if f is continuous on ]a, b[ and f is continuous from right at a.

Also f is continuous on ]a, b] if f is continuous on ]a, b[ and f is continuous from left at b.

3.37 Definition. Let I be an interval, a ∈ I and f : I \ {a} 7→ R. We say that f has an extension
by continuity at x = α ∈ {a, a+ , a− } if lim f (x) = ` with ` finite and we define its extension by
 x→α
f (x) if x ∈ I \ {a}
continuity f˜ : I 7→ R by f˜(x) =
` if x = a.
Moreover if f is continuous on I \ {a},then f˜ is continuous on I.

For example, the function f (x) = 1 + x sin( x1 ) defined on R∗ has an extension by continuity at x = 0
1 + x sin( x1 )

˜ if x ∈ R \ {0}
because lim f (x) = 1 is a finite number and its extension f (x) =
x→0 1 if x = 0.

40
3.4 Properties of Continuous Functions

3.38 Property. If f : D → R is strictly monotone on its domain, then it is injective.

Proof. Let x, x0 ∈ D s.t x 6= x0 .

x 6= x0 =⇒ x > x0 ∨ x < x0
=⇒ f (x) > f (x0 ) ∨ f (x) < f (x0 ) (f is strictly monotone)
0
=⇒ f (x) 6= f (x ).

Thus f is injective.

3.39 Theorem. (Intermediate Value Theorem- First Form)[I.V.T]

Le f : [a, b] 7→ R be a continuous function on the closed bounded interval [a, b]. If f (a)f (b) < 0 i.e.,
f (a) < 0 < f (b) ∨ f (b) < 0 < f (a), then there exists c ∈]a, b[ such that f (c) = 0.
Moreover; if f is strictly monotone on [a, b], then c is unique.

Proof. Assume f (a) < 0 and f (b) > 0 and let A := {x ∈ [a, b] : f (x) ≤ 0}.
• Since a ∈ A and A bounded above by b, then sup A exists in R. Let c := sup A ∈ [a, b]. Then
∃(xn ) ⊂ A s.t xn −→ c.
n→∞
• Claim f (c) ≤ 0:

xn −→ c and f continuous at c =⇒ lim f (xn ) = f (c)

n→∞ n→∞
=⇒ f (c) ≤ 0 ((xn ) ⊂ A implies f (xn ) ≤ 0)

• Claim f (c) ≥ 0:
Since f (b) > 0, then c 6= b. By the definition of c we have ∀x ∈]c, b[, f (x) > 0. Since f is continuous
at c and by applying the limit both sides as x → c+ to the previous equqlity, we obtain f (c) ≥ 0.
Thus there exists c ∈]a, b[ such that f (c) = 0.
Uniqueness of c : Assume that there exists c0 ∈]a, b[ s.t f (c0 ) = 0.

f (c) = f (c0 ) =⇒ c = c0 (f is injective, property 3.38).

Note that, if we replace f (a)f (b) < 0 by f (a)f (b) ≤ 0, then c ∈ [a, b]. Also if we replace f (a) < 0 <
f (b) by f (a) ≤ 0 < f (b), then c ∈ [a, b[.
3.40 Example. Show that f (x) = x25 − 12 has a unique root in ]0, 1[.
Solution. We have:
• f is continuous on [0, 1].( f is a polynomial, continuous on R hence continuous on [0, 1] ⊂ R)
• f (0)f (1) = (− 21 )( 12 ) < 0.
Then by I.V.T there exist c ∈]0, 1[ s.t f (c) = 0. Since f is strictly increasing, then c is unique.

3.41 Theorem. (Intermediate Value Theorem- Second Form)[I.V.T]

Le f : [a, b] 7→ R be a continuous function on the closed bounded interval [a, b]. If f (a) < T <
f (b) ∨ f (b) < T < f (a), then there exists c ∈]a, b[ such that f (c) = T .
Moreover; if f is strictly monotone on [a, b], then c is unique.

Proof. Apply the previous theorem to the function g : [a, b] 7→ R defined by g(x) = f (x) − T .

41
3.42 Corollary. Let I be an interval and f : I 7→ R. If f is continuous, then f (I) is an interval.

Proof. Let y1 , y2 ∈ f (I) and y1 < z < y2 . Claim z ∈ f (I).

Since y1 , y2 ∈ f (I), then there exists x1 , x2 ∈ I s.t y1 = f (x1 ) and y2 = f (x2 ). Since y1 < y2 , then
x1 6= x2 . Assume x1 < x2 and let us try I.V.T for f on [x1 , x2 ].
• f is continuous on [x1 , x2 ].( f is continuous on I hence continuous on [x1 , x2 ] ⊂ I)
• f (x1 ) < z < f (x2 ).
Then by I.V.T there exist c ∈]x1 , x2 [⊂ I s.t f (c) = z. Thusz ∈ f (I).

We will see in the next theorem that if f is continuous on [a, b], then f ([a, b]) = [m, M ] where
m = min f (x) and M = max f (x). In case I is open or half-open interval, its image f (I) can
x∈[a,b] x∈[a,b]
be an interval of any kind. For example, take f (x) = sin x: If I =]0, 2π[, then f (I) = [−1, 1]. If
I =] − π2 , π2 [, then f (I) =] − 1, 1[. If I =] − π2 , 3π
2 [, then f (I) =] − 1, 1]. Take g(x) = tan x: if If
I =] − π2 , π2 [, then g(I) =] − ∞, ∞[.

3.43 Theorem. (Extreme Value Theorem)

If f is continuous on a closed bounded interval [a, b], then f is bounded and attains its maximum
and minimum i.e., ∃c, d ∈ [a, b] s.t min f (x) = f (c) and max f (x) = f (d). Moreover f ([a, b]) =
x∈[a,b] x∈[a,b]
[f (c), f (d)] i.e.,
f : [a, b] 7→ f ([a, b]) = [f (c), f (d)].

Proof.
Claim f is bounded: Suppose to the contrary that f is not bounded. Then for each n ∈ N there
exists un ∈ [a, b] s.t |f (un )| > n. Since un is bounded, then by Bolzano–Weierstrass theorem, there
exists a sub sequence unk converging to x0 ∈ [a, b]. By the continuity of |f |, we get |f (unk )| converges
to |f (x0 )|. Hence |f (unk )| is a bounded sequence , a contradiction.
Claim f attains its maximum and minimum: Since f is bounded, then sup f (x) is finite. Let
x∈[a,b]
M := sup f (x). By the sequential characterization of supremum, there exists a sequence xn ∈ [a, b]
x∈[a,b]
s.t f (xn ) −→ M . By the Bolzano–Weierstrass theorem, there exists a sub sequence xnk converging
n→∞
to some d ∈ [a, b]. Since f is continuous, then M = f (d). Therefore max f (x) = f (d). The proof
x∈[a,b]
for the minimum is similar.
Claim f ([a, b]) = [f (c), f (d)]: By the previous corollary, f ([a, b]) is an interval. Since f (c) is the
lower end point and f (d) is the upper end point, then f ([a, b]) = [f (c), f (d)].

3.44 Remark. A function f : D 7→ f (D) is surjective (onto).

O O
3.45 Theorem. (Inverse Function Theorem-1)

O O
Let I be an interval of R and f : I 7→ f (I) = J. Suppose that 1 f is continuous on I and 2 f is

O
strictly monotone on I, then i J = f (I) is an interval, ii f : I 7→ J is bijective (hence invertible)
and iii f −1 : J 7→ I is continuous and strictly monotone as f . Moreover, the representative curve
of f −1 is obtained from the representative curve of f by symmetry with respect to the first bisector
y = x.

42
 O
O
Proof. i The image of an interval by a continuous function is an interval.
ii Since f is strictly monotone, then it is injective. But f : I 7→ f (I) = J is surjective. Thus f

O
defines a bijective map from I onto J.
iii Assume f : I 7→ J is strictly increasing and let us show that f −1 : J 7→ I is strictly increasing.
Let y1 , y2 ∈ J s.t y1 < y2 , then there exist x1 , x2 ∈ I s.t f (x1 ) = y1 and f (x2 ) = y2 . Claim that
f −1 (y1 ) < f −1 (y2 ): Suppose to the contrary that f −1 (y1 ) ≥ f −1 (y2 ) i.e., x1 ≥ x2 . Since f is strictly
increasing, then f (x1 ) ≥ f (x2 ) i.e., y1 ≥ y2 , a contradiction. For continuity of f −1 on f (I): Suppose
that y1 = f (x1 ) ∈ f (I) and ε > 0 are given. Assume that x1 is not an end point of I; a slight
change will deal with an end point. Choose x0 , x2 ∈ I s.t x1 − ε < x0 < x1 < x2 < x1 + ε. Set
yj = f (xj ) and take 0 < δ < min{y2 − y1 , y1 − y0 }. Then |y − y1 | < δ implies y0 < y < y2 and
x1 − ε < x0 < f −1 (y) < x2 < x1 + ε.

Note that if f is continuous and strictly monotone on an interval I, then f (I) is also an interval;
namely,

If f is strictly increasing If f is strictly decreasing

1. I = [a, b] =⇒ f (I) = [f (a), f (b)]. 1. I = [a, b] =⇒ f (I) = [f (b), f (a)].

2. I =]a, b] =⇒ f (I) =] lim f (x), f (b)]. 2. I =]a, b] =⇒ f (I) = [f (b), lim f (x)[.
x→a+ x→a+

3. I = [a, b[ =⇒ f (I) = [f (a), lim f (x)[. 3. I = [a, b[ =⇒ f (I) =] lim f (x), f (a)].
x→b− x→b−

4. I =]a, b[ =⇒ f (I) =] lim f (x), lim f (x)[. 4. I =]a, b[ =⇒ f (I) =] lim f (x), lim f (x)[.
x→a+ x→b− x→b− x→a+

3.5 Substitution Theorem

3.46 Theorem. (Substitution Theorem)

Suppose lim f (x) = ` ∈ R and g defined in a neighborhood of ` excluding possibly the point ` s.t,
O x→α

O
i ` ∈ R and g is continuous at `,
ii ` = ±∞ and lim g(y) ∈ R.
y→`
Then lim g(f (x)) = lim g(y).
x→α y→`

Proof. Exercise.

In case iO
• lim g(f (x)) = lim g(y) = g(`) = g( lim f (x)).
x→α y→` x→α
• The theorem
 fails if g is not continuous. A counterexample is given by the functions f (x) = 0 and
1 if y 6= 0
g(y) = We have lim g(f (x)) = 3 6= 1 = lim g(y).
O
3 if y = 0. x→0 y→0

O
An alternative condition to i which gives the same conclusion is
iii ` ∈ R, f (x) 6= ` ∀x ∈ I(α) \ {α} and lim g(y) ∈ R.
y→`

3.6 More About Limit

In this section we give a definition of the limit including the limits encountered in section 3.2.
• An accumulation point of a nonempty set D of real numbers is an extended real number α such
that every neighborhood of α contains a point of D not equal to α.
• A point of D that is not an accumulation point of D is called an isolated point of D.

43
• α is an isolated point of D if α ∈ D and there is a neighborhood of α that contains no other point
of D.
• Let f be a real valued function defined on a non empty subset D of R. Let α, ` ∈ R where either
α is an accumulation point of D or α is an isolated point of D. We say that f (x) tends to ` as x
tends to α along D and we write lim f (x) = ` if
x→α
x∈D

∀I(`), ∃I(α), ∀x ∈ D, x ∈ I(α) =⇒ f (x) ∈ I(`).

• the limit ` is unique.

• The definition of limit as x → α where α ∈ {a, a+ , a− , ∞, −∞} in section 3.2 is a particular case
of this definition as follows: lim f (x) = lim f (x).
x→α x→α
x∈Iα \{α}

• Let f be be a real valued function defined on a non empty subset D and D1 ∪ D2 = D. Suppose
that one of the following four possibilities hold: a is an accumulation point ( or isolated point) of D1
and a is an accumulation point ( or isolated point) of D2 . Then, lim f (x) exists in R iff both limits
x→a
x∈D

lim f (x) and lim f (x) exist in R and are equal. In this case all three limits are equal.
x→a x→a
x∈D1 x∈D2

? For example: Let D1 = N and D2 =]0, 1[. Then 1 is an isolated point of D1 and an accumulation
point of D2 , and lim f (x) = f (1). Therefore lim f (x) exists iff lim f (x) = lim f (x) = f (1).
x→1 x→1 x→1 x→1−
x∈D1 x∈D x∈D2

? In particular: lim f (x) exists in R iff lim f (x) exist and are equal. In this case all three limits
x→a x→a±
are equal.
• All other properties, like sequential characterization, algebra of limits, Sandwich Rule can be proved
analogously as the ones proved in section 3.2.

3.7 Exercises
E( x1 )+x
Exercise 1. Let f (x) = E( x1 )−x
.

1. Find the domain of definition of f .

1+x2
2. Shoe that ∀x ∈]0, 1/2[, 1 < f (x) ≤ 1−x2 −x
.

3. Using the definition of the limit of functions, show that lim f (x) = 1.
x→0+

1 if x ∈ Q
Exercise 2. Let f (x) = Using the density of Q and R \ Q in R, show that f is
0 if x ∈
/ Q.
discontinuous at each real number a ∈ R.
Exercise 3. Show that if f is continuous at a and f (a) 6= 0, then f is different from zero in a
neighborhood of a.
Exercise 4. Let f : R 7−→ R be periodic function with period T .

1. Assume f is continuous on R.

(a) Let x ∈ R. Show that f (x + nT ) = f (x) ∀n ∈ Z.

(b) Show that there exists c, d ∈ [0, T ] such that f (c) ≤ f (x) ≤ f (d) for all x ∈ [0, T ].
x
(c) Let x ∈ R and n0 = E . Show that 0 ≤ x − n0 T < T .
T
(d) Deduce that f is bounded and attains its maximum and minimum on R.

2. Assume f is a non constant function and let a, b ∈ R such that f (a) 6= f (b). Take xn = a + nT
and yn = b + nT . Using the sequential characterization of limit, show that lim f (x) @.
x→∞

44
 √
x − 2x − 1
Exercise 5. Let the function f (x) = . Consider the two sets
x−1
1
A = {f (x), ≤ x < 1} and B = {f (x), x > 1}.
2
1. Find the domain of definition of the function f .

2. Using the definition of the limit of a function, show that lim f (x) = 0.
x→1
1
3. Show that f (x) ≤ 0 for all 2 ≤ x < 1 and 0 ≤ f (x) for x > 1.

4. Deduce that sup A and inf B exist.

1
5. Deduce that f (x) ≤ sup A ≤ 0 for all 2 ≤ x < 1 and 0 ≤ inf B ≤ f (x) for x > 1.

6. Determine the values of sup A and inf B.

7. Do max A and min B exist? Justify your answer.

8. Using the ε- characterization of sup and inf, show that sup A = 0 and inf B = 0.

Exercise 6. Let f : [a, b] 7→ [a, b] be a continuous function.

1. Show that f has a fixed point c ∈ [a, b] i.e., f (c) = c.

2. Show that c is unique in each of the following cases:

(a) f is decreasing.
(b) There exists k ∈ [0, 1[ s.t ∀x, y ∈ [a, b], |f (x) − f (y)| ≤ k|x − y|.

Exercise 7.(Generalization of the I.V.T- First Form)

1. Let I be an interval and f : I 7→ R be a continuous function. Assume that the limits (finite
6= 0 or infinite ) of f (x) as x tends to the end points of I is of different signs.

(a) Show that there exist c ∈ I s.t f (c) = 0.

(b) Deduce that if f is strictly monotone on I, then c is unique.

2. Deduce that f (x) = x + sin x + ln x has exactly one zero in ]0, ∞[.

3. Deduce that a polynomial of odd degree has a real root.

Exercise 8. (Generalization of the I.V.T- Second Form)
Let f and g be two continuous functions on the closed bounded interval [a, b] s.t f (a) < g(a)) and
f (b) > g(b).
1. Show that there exists at least one point c ∈]a, b[ s.t f (c) = g(c).

2. Deduce the I.V.T- Second Form.

3. Consider g or f to be the identity function. What you may deduce.

Exercise 9. Let f, g : [a, b] →
7 R be continuous functions s.t f (x) < g(x), ∀x ∈ [a, b]. Show that
∃α > 0 s.t f (x) ≤ g(x) − α, ∀x ∈ [a, b].

Exercise 10. Let f : [0, 1] 7→ R be continuous function and α, β > 0. Show that there exists
c ∈ [0, 1] s.t αf (0) + βf (1) = (α + β)f (c).

Exercise 11. (Existence of the nth root)

Let a > 0, n ∈ N∗ and f (x) = xn .

45
 1. Show that there exists b > 0 s.t f (b) > a. (Hint: You may use x→∞
lim f (x) = ∞)

2. Deduce that there exists a unique c ∈]0, b[ ( c positive ) s.t f (c) = a.

Exercise 12. Let f be continuous and positive on R with lim f (x) = 0. Show that f attains its
x→±∞
maximum on R.
Exercise 13. Let f be continuous on R with lim f (x) = ∞. Show that f attains its minimum on R.
x→±∞

x−3
Exercise 14. Let f :]2, ∞[7→ R be a function defined by f (x) = x−2 . Show that f has an inverse
function f −1 to be determined. √
Exercise 15. Let f : [1, ∞[7→ R be a function defined by f (x) = x − 1. Show that f has an inverse
function f −1 to be determined.
1
Exercise 16. Let f : [2, ∞[7→ R be a function defined by f (x) = √x2 −4x+8 . Show that f has an
−1
inverse function f to be determined.
Exercise 17. Let f, g : [0, 1] 7→ [0, ∞[ be continuous functions satisfying sup f (x) = sup g(x).
0≤x≤1 0≤x≤1
Show that there exists x0 ∈ [0, 1] s.t f (x0 ) = g(x0 ).
Exercise 18. Prove property 3.3, property 3.22, ( property 3.23,theorem 3.27 and corollary 3.28.
1+x+x2 sin x1 cos( 2π )
Exercise 19. Let un = n2 +n+sin n
and f (x) = 1+x+x2
x
if x 6= 0
n2 +n+1
0 if x = 0.

1. Show that lim un = 1 and lim f ( n1 ) = 1.

n→∞ n→∞

2. Deduce that f is not continuous at 0.

Exercise 20.(Continuity of sin and cos on R).

1. Show that | sin x| ≤ |x| for all x ∈ R. ( Hint: Graphically consider the unit circle for the case 0 < x ≤ π2 .)

2. Using the definition of the limit of functions, show that x 7→ sin x is continuous on R. ( Hint:
x−x0 x+x0
Use the identity sin x − sin x0 = 2 sin 2
cos 2
. )

3. Deduce that x 7→ cos x is continuous on R.

Exercise 21. (Fundamental limit)

1. Show that sin x < x < tan x for all 0 < x ≤ π2 . ( Hint: Graphically consider the unit circle.)
sin x sin x
2. Deduce that lim = 1 and lim = 1.
x→0+ x x→0 x

Exercise 22. (Fundamental limits)

1. Let an = (1 + n1 )n , n ∈ N∗ .
n
1
· 1 · (1 − n1 ) · · · (1 − k−1
P
(a) Show that an = k! n ). Deduce that (an ) is strictly increasing and
k=0
2 < an for n ≥ 2.
(b) Show that an < 3 for all n ∈ N∗ . ( Hint: Use k! ≥ 2k−1 .)
(c) Show that the sequence (an ) is convergent and 2 ≤ lim an ≤ 3.
n→∞

2. Let e := lim (1 + n1 )n .
n→∞

1 n
(a) Let n = E(x) for x > 0. Show that (1 + n+1 ) ≤ (1 + x1 )x ≤ (1 + n1 )n+1 .
1 x
(b) Deduce that lim (1 + ) = e. ( Hint: x = −|x| for x < 0.)
x→±∞ x

46
 ln(1 + x) 1
(c) Deduce that lim = 1. ( Hint: Use ln(1+x)
x
= ln(1 + x) x and the substitution theorem.)
x→0 x
ex − 1
(d) Deduce that lim = 1. ( Hint: put ex − 1 = y.)
x→0 x
(1 + x)α − 1
(e) Deduce that lim = α (α ∈ R). ( Hint: put 1 + x = ey .)
x→0 x

47
CHAPTER 4
DIFFERENTIATION ON R

4.1 First Derivative: Properties And Examples

4.1 Definition. Let I be an open interval, x0 ∈ I and f : I 7→ R. We say that:

f (x) − f (x0 )
1. f is differentiable at x0 if lim exists and finite. In this case the real number
x→x0 x − x0
f (x) − f (x0 ) f (x0 + h) − f (x0 )
f 0 (x0 ) := lim = lim
x→x0 x − x0 h→0 h
is called the (first) derivative of f at x0 .

2. f is differentiable on I if f is differentiable at each point of I.

Graphical interpretation of derivative:

• f (x)−f
x−x0
(x0 )
represents the slope of the line
passing through the points M0 (x0 , f (x0 )) M f
and M (x, f (x)).
• As x approaches x0 , M approaches M0 . f (x) − f (x0 )
f (x) − f (x0 ) M0
• Hence lim represents
x→x0 x − x0 x − x0
the slope of the tangent line to the
x0 x
representative curve of f at the point
M0 (x0 , f (x0 )).
Graphical interpretation of derivative.

4.2 Property. (Differentiability =⇒ Continuity )

Let I be open interval, x0 ∈ I and f : I 7→ R. If f is differentiable at x0 , then f is continuous at x0 .

 f (x) − f (x ) 
0
Proof. lim (f (x) − f (x0 )) = lim · (x − x0 ) = f 0 (x0 ) · 0 = 0.
x→x0 x→x0 x − x0
The converse of this property is not true. For example, the function f (x) = |x| is continuous but not
f (x) − f (0) |x|
differentiable at x = 0.( lim = lim = ±1, hence does not exist.)
x→0 x−0 x→0 x
The contrapositive of this property: If f is not continuous at x0 , then it is not differentiable at x0 .

48
4.3 Examples. Differentiability for some usual functions.

1. A constant function f (x) = c is differentiable on R and f 0 (x) = 0 for all x ∈ R.

f (x) − f (x0 ) c−c
Solution. Let x0 ∈ R arbitrary. lim = lim = 0 exist and finite. Hence
x→x0 x − x0 x→x0 x − x0
0
f is differentiable at x0 and f (x0 ) = 0 for all x0 ∈ R.

2. An affine function f (x) = ax + b is differentiable on R and f 0 (x) = a for all x ∈ R.

f (x) − f (x0 ) ax − ax0
Solution. Let x0 ∈ R arbitrary. lim = lim = a exist and finite.
x→x0 x − x0 x→x0 x − x0
0
Hence f is differentiable at x0 and f (x0 ) = a for all x0 ∈ R.

3. The function f (x) = x2 is differentiable on R and f 0 (x) = 2x for all x ∈ R.

Solution. Let x0 ∈ R arbitrary.
f (x) − f (x0 ) x2 − x20 (x − x0 )(x + x0 )
• lim = lim = lim = lim (x + x0 ) = 2x0 exist and
x→x0 x − x0 x→x0 x − x0 x→x0 x − x0 x→x0
finite. Hence f is differentiable at x0 and f 0 (x0 ) = 2x0 for all x0 ∈ R.
f (x0 + h) − f (x0 ) (x0 + h)2 − x20 2x0 h + h2
• lim = lim = lim = lim (2x0 + h) = 2x0 exist
h→0 h h→0 h h→0 h h→0
0
and finite. Hence f is differentiable at x0 and f (x0 ) = 2x0 for all x0 ∈ R.

4. The function f (x) = xn , n ∈ N∗ , is differentiable on R and f 0 (x) = nxn−1 for all x ∈ R.

Solution. Let x0 ∈ R arbitrary.

f (x0 + h) − f (x0 ) (x0 + h)n − xn0

lim = lim
h→0 h h→0 h
n
(nk )hk xn−k − xn0
P
0
k=0
= lim ( binomial formula)
h→0 h
n
xn0 + (n − 1)hxn−1 (nk )hk xn−k − xn0
P
0 + 0
k=2
= lim
h→0 h
n
(n − 1)hx0n−1 + (nk )hk x0n−k
P
k=2
= lim
h→0 h
= (n − 1)xn−1
0

exist and finite. Hence f is differentiable at x0 and f 0 (x0 ) = (n − 1)x0n−1 for all x0 ∈ R.

5. The function f (x) = xα , α ∈ R, is differentiable on its domain of definition and f 0 (x) = αxα−1 .
Note that Df = R ∨ R∗ ∨ R∗+ depending on α ∈ N ∨ Z ∨ R.
Solution. Let x0 6= 0 be in Df .

f (x0 + h) − f (x0 ) (x0 + h)α − xα0

lim = lim
h→0 h h→0 h
x (1 + h/x0 )α − xα0
α
= lim 0
h→0 h
(1 + h/x0 )α − 1
= xα−1
0 lim (fundammental limit)
h→0 h/x0
= αxα−1
0

exist and finite. Hence f is differentiable at x0 and f 0 (x0 ) = αxα−1

0 for all x0 ∈ Df .

49
6. The function f (x) = sin x is differentiable on R and f 0 (x) = cos x for all x ∈ R.
Solution. Let x0 ∈ R arbitrary.

f (x0 + h) − f (x0 ) sin(x0 + h) − sin x0

lim = lim
h→0 h h→0 h
2 sin(h/2) cos(x0 + h/2) x − x0 x + x0
= lim (sin x − sin x0 = 2 sin cos )
h→0 h 2 2
sin(h/2) cos(x0 + h/2)
= lim
h→0 h/2
= cos x0 ( fundemental limit & continuity of cos)

exist and finite. Hence f is differentiable at x0 and f 0 (x0 ) = cos x0 for all x0 ∈ R.

7. The function f (x) = cos x is differentiable on R and f 0 (x) = − sin x for all x ∈ R.
Solution. Let x0 ∈ R arbitrary.

f (x0 + h) − f (x0 ) cos(x0 + h) − cos x0

lim = lim
h→0 h h→0 h
−2 sin(h/2) sin(x0 + h/2) x − x0 x + x0
= lim (cos x − cos x0 = −2 sin sin )
h→0 h 2 2
sin(h/2) sin(x0 + h/2)
= − lim
h→0 h/2
= − sin x0 ( fundemental limit & continuity of sin)

exist and finite. Hence f is differentiable at x0 and f 0 (x0 ) = − sin x0 for all x0 ∈ R.

8. The function f (x) = ex is differentiable on R and f 0 (x) = ex for all x ∈ R.

Solution. Let x0 ∈ R arbitrary.

f (x0 + h) − f (x0 ) e(x0 +h) − ex0

lim = lim
h→0 h h→0 h
h
e −1
= ex0 lim
h→0 h
= e x0 ( fundemental limit )

exist and finite. Hence f is differentiable at x0 and f 0 (x0 ) = ex0 for all x0 ∈ R.

9. The function f (x) = ln x is differentiable on ]0, ∞[ and f 0 (x) = 1

x for all x ∈]0, ∞[.
Solution. Let x0 ∈]0, ∞[ arbitrary.

f (x0 + h) − f (x0 ) ln(x0 + h) − ln x0

lim = lim
h→0 h h→0 h
1 ln(1 + h/x0 )
= lim
x0 h→0 h/x0
1
= ( fundemental limit )
x0

exist and finite. Hence f is differentiable at x0 and f 0 (x0 ) = 1

x0 for all x0 ∈]0, ∞[.

50
4.4 Definition. (Right & Left Derivatives)

1. Suppose f is defined on a right neighborhood I(x+ 0 ) of x0 . We say that f is differentiable from

f (x) − f (x0 )
right at x0 if lim exists and finite. In this case the real number
x→x+0
x − x0

f (x) − f (x0 ) f (x0 + h) − f (x0 )

fr0 (x0 ) = f+0 (x0 ) := lim = lim
x→x+
0
x − x0 h→0 + h

is called the right derivative of f at x0 .

2. Suppose f is defined on a left neighborhood I(x− 0 ) of x0 . We say that f is differentiable from

f (x) − f (x0 )
left at x0 if lim exists and finite. In this case the real number
x→x−0
x − x0

f (x) − f (x0 ) f (x0 + h) − f (x0 )

f`0 (x0 ) = f−0 (x0 ) := lim = lim
x→x−
0
x − x0 h→0 − h

is called the left derivative of f at x0 .

4.5 Property. Let I be open interval, x0 ∈ I and f : I 7→ R. Then


O
O
 1 f is differentiable from right at x0 ,

O
f is differentiable at x0 ⇐⇒ 2 f is differentiable from left at x0 ,
3 fr0 (x0 ) = f`0 (x0 ).


Note that the right derivative of f at x0 represents the slope of the right tangent to the graph of f at
the point (x0 , f (x0 )). Similarly, the left derivative of f at x0 represents the slope of the left tangent
to the graph of f at the point (x0 , f (x0 )). By the previous property, f is differentiable at x0 if the
right and left tangents coincides (one tangent ) and not vertical. Equivalently, f is not differentiable
at x0 if the right and left tangents do not coincide (they form an angle) or atleast one is vertical.
4.6 Examples.

1. f (x) = |x| is not differentiable at x = 0. y

Solution.
f (x) − f (0) |x| x
• lim = lim = lim = 1 =
x→0 + x − 0 x→0
x>0
x x→0 x
fr0 (0). So f is differentiable from right at 0.

f (x) − f (0) |x| −x

• lim = lim = lim =
x→0− x−0 x→0 x
x<0
x→0 x

−1 = f`0 (0). So f is differentiable from left at

0.

• fr0 (0) 6= f`0 (0). Thus f is not differentiable x

at x = 0. x 7→ |x| is not differentiable at 0.

Note that:
? If fr0 (x0 ) 6= f`0 (x0 ) and at least one of them is finite, then x0 is called a corner point for f .

51
 For example x = 0 is a corner point for the function f (x) = |x|.
? If fr0 (x0 ) and f`0 (x0 ) are finite and have different sign, then x0 is called a cusp point of f .
For example x = 0 is a cusp point for the function f (x) = |x|.
p
2. f (x) = |x| is not differentiable at x = 0. y
Solution. p
f (x) − f (0) |x|
• lim = lim = ∞. So f is
x→0+ x−0 x→0 |x|
x>0
not differentiable from right at x = 0. Thus f
is not differentiable at x = 0. Or,
p
f (x) − f (0) |x|
• lim = lim = −∞. So
x→0 − x−0 x→0
x<0
−|x|
f is not differentiable from left at x = 0. Thus
f is not differentiable at x = 0.
x
p
x 7→ |x| is not differentiable at 0.

1 y
3. f (x) = 1+|x−1| is not differentiable at x = 1.
Solution.
f (x) − f (1) 1/x − 1
• lim = lim =
x→1 + x − 1 x→1
x>1
x−1
−1
lim = −1 = fr0 (1). So f is differentiable
x→1 x
from right at 1.

f (x) − f (1) 1/(2 − x) − 1

• lim = lim =
x→1− x−1 x→1
x<1
x−1
1 x
lim = 1 = f`0 (1) So f is differentiable
x→1 2 − x 1
from left at 1.
x 7→ 1/(1 + |x − 1|) is not differentiable at 1.
• fr0 (0) 6= f`0 (0). Thus f is not differentiable
at 1.

x2

if x > 0 y
4. f (x) = is differentiable at
x3 if x ≤ 0.
0.
Solution.
f (x) − f (0) x2
• lim = lim = lim x = 0 =
x→0+ x−0 x→0 x
x>0
x→0 x
fr0 (0). So f is differentiable from right at 0.

f (x) − f (0) x3
• lim = lim = lim x2 = 0 =
x→0− x−0 x→0 x
x<0
x→0

f`0 (0). So f is differentiable from left at 0.

• fr0 (0) = f`0 (0). x 7→ f (x) is differentiable at 0.

Thus f is differentiable at x = 0.

52
4.7 Definition. A function f : [a, b] 7→ R is differentiable on [a, b] closed if the following holds
•f is differentiable on ]a, b[.
•f is differentiable from right at a.
•f is differentiable from left at b.

Similarly, f is differentiable on [a, b[ if f is differentiable on ]a, b[ and f is differentiable from right at

a. Also f is differentiable on ]a, b] if f is differentiable on ]a, b[ and f is differentiable from left at b.
√
4.8 Example. Study the differentiability of the function f (x) = x. Solution.
• Df = [0, ∞[.
• Differentiability of f on ]0, ∞[: Let x0 ∈]0, ∞[ arbitrary.
√ √  √x − √x √x + √x 
f (x) − f (x0 ) x − x0 0 0 1 1
lim = lim = lim ·√ √ = lim √ √ = √
x→x0 x − x0 x→x 0 x − x0 x→x 0 x − x0 x + x0 x→x 0 x + x0 2 x0

exist and finite. Hence f is differentiable at x0 and f 0 (x0 ) = 2√1x0 for all x0 ∈]0, ∞[.
• Differentiability of f at 0 from right:
√
f (x) − f (0) x
lim = lim = ∞.
x→0 + x−0 x→0 + x
√
So f is not differentiable from right at 0. Thus f (x) = x is differentiable on ]0, ∞[ and f 0 (x) = 1
√
2 x
.

4.9 Remark. In the same way as property 4.2 one can show:
? If f is differentiable from right at x0 , then f is continuous from right at x0 .
? If f is differentiable from left at x0 , then f is continuous from left at x0 .
? If f is differentiable from right and left at x0 (even if different), then f is continuous at x0 .
? If f is differentiable on an interval I, then f is continuous on I.

4.10 Property. ( Finite expansion up to order one for differentiable function )

Let I be open interval, x0 ∈ I and f : I 7→ R. The function f is differentiable at x0 if and only if
there exists ε : I 7→ R s.t ε(x) −→ 0 and there exist c ∈ R s.t f (x) = f (x0 ) + (x − x0 )c + (x − x0 )ε(x)
x→x0
for all x ∈ I. In this case c = f 0 (x0 ).

Proof.
⇒) Suppose that f is differentiable at x0 . Let ε(x) = f (x)−f
x−x0
(x0 )
− f 0 (x0 ). Since f is differentiable
at x0 , then ε(x) −→ 0. Hence ε extendable by continuity at 0 with ε(0) = 0 and . Thus we have
x→x0
∀x ∈ I, f (x) = f (x0 ) + (x − x0 )c + (x − x0 )ε(x) with c = f 0 (x0 ).
⇐) Suppose that ∀x ∈ I, f (x) = f (x0 ) + (x − x0 )c + (x − x0 )ε(x) with ε(x) −→ 0. Then
x→x0
f (x)−f (x0 )
for x ∈ I \ {x0 }, we have x−x0 = c + ε(x) −→ c ∈ R. Thus f is differentiable at x0 and
x→x0
f 0 (x0 ) = c.

Note that:
? The function ε : I 7→ R is continuous at x0 . Moreover if f is continuous on I then so ε.
? f (x) = f (x0 ) + (x − x0 )f 0 (x0 ) + (x − x0 )ε(x), gives that a differentiable function f at x0 can be
| {z }
YT
approximated in a neighborhood of x0 by the tangent YT to the curve at the point (x0 , f (x0 )).

53
4.11 Property. Let I be open interval, f, g : I 7→ R and α, β ∈ R. If f and g are differentiable at
x0 ∈ I , then

1. αf + βg is differentiable at x0 and (αf + βg)0 (x0 ) = αf 0 (x0 ) + βg 0 (x0 ).

2. f · g is differentiable at x0 and (f · g)0 (x0 ) = f 0 (x0 )g(x0 ) + f (x0 )g 0 (x0 ).

1
3. If g(x0 ) 6= 0, then g is defined in a neighborhood of x0 and differentiable at x0 and
 1 0 −g 0 (x0 )
(x0 ) = .
g g 2 (x0 )

f
4. If g(x0 ) 6= 0, then g is defined in a neighborhood of x0 and differentiable at x0 and
 f 0 f 0 (x0 )g(x0 ) − f (x0 )g 0 (x0 )
(x0 ) = .
g g 2 (x0 )

Proof.

(αf + βg)(x) − (αf + βg)(x0 ) αf (x) − αf (x0 ) βg(x) − βg(x0 )

1. = + −→ αf 0 (x0 ) + βg 0 (x0 ).
x − x0 x − x0 x − x0 x→x0

f g(x) − f g(x0 ) f (x)g(x) − f (x0 )g(x) + f (x0 )g(x) − f (x0 )g(x0 )

2. =
x − x0 x − x0
f (x) − f (x0 ) g(x) − g(x0 )
= · g(x) + · f (x0 ) (g is continuous at x0 )
x − x0 x − x0
−→ f 0 (x0 )g(x0 ) + f (x0 )g 0 (x0 ).
x→x0

3.We have g is differentiable at x0 and g(x0 ) 6= 0, then g is continuous at x0 and g(x0 ) 6= 0. So

g(x) 6= 0 in some neighborhood of x0 and thus g1 is defined and continuous in some neighborhood of
x0 . Now

1 1 g(x)−g(x0 )
g(x) − g(x0 ) (x−x0 ) −g 0 (x0 )
=− −→ . (g is continuous at x0 )
x − x0 g(x)g(x0 ) x→x0 g 2 (x0 )
f 1
4. Write g =f· g and use the second and third items.

4.12 Property. (Chain Rule )

Let f be differentiable at x0 and g be differentiable at y0 = f (x0 ). Then g ◦ f is differentiable at x0
and (g ◦ f )0 (x0 ) = g 0 (f (x0 ))f 0 (x0 ).

Proof. Claim g ◦ f is defined in a neighborhood of x0 :

Since g is differentiable at y0 , then its defined in a neighborhood J = J(y0 ) of y0 = f (x0 ). Since f
is differentiable at x0 , then its continuous at x0 and so there exists I = I(x0 ) a neighborhood of x0
s.t f (I) ⊂ J. Thus g ◦ f is defined on I a neighborhood of x0 . Now,
g is differentiable at y0 implies ∀y ∈ J, g(y) = g(y0 ) + (y − y0 )g 0 (y0 ) + (y − y0 )ε(y) with ε(y) −→ 0.
y→y0

54
For x ∈ I \ {x0 }, y = f (x) ∈ J and
g ◦ f (x) − g ◦ f (x0 ) g(y) − g(y0 ) (y − y0 )g 0 (y0 ) + (y − y0 )ε(y)
= =
x − x0 x − x0 x − x0
f (x) − f (x0 ) 0 f (x) − f (x0 )
= · g (f (x0 )) + · ε(f (x))
x − x0 x − x0
−→ g 0 (f (x0 ))f 0 (x0 ). ( substitution theorem)
x→x0

4.13 Remark.
(f (u(x)))0 = f 0 (u(x))u0 (x) provided that u is differentiable at x and f is differentiable at u(x).
For example (f (sin x))0 = f 0 (sin x) cos x provided that f is differentiable at sin x. In particular
(cos(sin x))0 = − sin(sin x) cos x.

4.2 Higher Order Derivatives

4.14 Definition. Let I be an open interval, x0 ∈ I and f : I 7→ R be a function.

O
1. We say that f is twice (or two times ) differentiable at x0 if
i f is differentiable on I(x0 ) a neighborhood of x0 i.e., f 0 : I(x0 ) 7→ R exists and
O
ii f 0 is differentiable at x0 .
We denote by f 00 (x0 ) the second derivative of f at x0 and we have f 00 (x0 ) = (f 0 )0 (x0 ).

2. We define by recurrence the nth derivative (derivative of order n) of f at x0 by

f (n) (x0 ) = (f (n−1) )0 (x0 ).

df d2 f d3 f
Some notations commonly used: f (0) = f, f (1) = f 0 = dx , f (2) = f 00 = dx2
and f (3) = f 000 = dx3
.
4.15 Examples.
1. Let f (x) = sin x. Show that f (n) (x) = sin(x + n π2 ) ∀n ∈ N∗ .
Solution. By induction:
π
f (n) (x) = sin(x + n ) ∀n ∈ N∗ (P (n)).
2
• For n = 1 we have f (1) (x) = cos x = sin(x + 1 π2 ). Hence P (1) is true.
• Assume P (n) is true i.e., f (n) (x) = sin(x + n π2 ). Now
π π π π π
f (n+1) (x) = (f (n) )0 (x) = (sin(x + n ))0 = cos(x + n ) = sin(x + n + ) = sin(x + (n + 1) ).
2 2 2 2 2
Hence P (n + 1) is true. Thus P (n) is true for all n ∈ N∗ .

2. Let g(x) = cos x. Show that g (n) (x) = cos(x + n π2 ) ∀n ∈ N∗ .

Solution. We have g(x) = cos x = sin(x + π2 ) = f (x + π2 ). Hence by chain rule n times we have
g (n) (x) = f (n) (x + π2 ) = sin(x + π2 + n π2 ) = cos(x + n π2 ). Or one can do it by induction as 1.

4.16 Definition. Let I be an open interval and f : I 7→ R be a function.

1. We say that f is twice (or two times ) differentiable on I if f 0 : I 7→ R is differentiable on I

and its derivative is called the second derivative function of f and denoted by f 00

55
 2. We define by recurrence the nth derivative of f on I by f (n) = (f (n−1) )0 . If f (n) exists on I,
we say that f is n-times differentiable on I.

3. We say that f is of class C 1 ( or continuously differentiable ) on I and we write f ∈ C 1 (I) if

O
i f is differentiable on I i.e., f 0 : I 7→ R exists and
O
ii f 0 is continuous on I.

4. We say that f is of class C n on I and we write f ∈ C n (I) if

O
O
i f is differentiable n-times on I (i.e., derivatives up to order n exists on I) and
ii f (n) : I 7→ R is continuous on I.

5. We say that f is of class C ∞ on I and we write f ∈ C ∞ (I) if f ∈ C n (I) ∀n ∈ N∗ i.e.,

C ∞ (I) := ∩ ∗ C n (I).
n∈N

O
Note that;
1 f ∈ C n (I) =⇒ derivatives up to order n exist and continuous on I.
O2 f ∈ C ∞ (I) =⇒ derivatives up to any order exist and continuous on I.
O3 C(I) = C 0 (I) := {f ; f : I 7→ R is continuous on I}.
O
O
4 F(I) := {f ; f : I 7→ R is a function }.
5 C ∞ (I)
 (n+2· · · ( C n (I) ( · · · C 2 (I) ( C 1 (I) ( C 0 (I) ( F(I). One can check that
x sin x1 if x 6= 0
f (x) = is of class C n but not of class C n+1 on R.
0 if x = 0.

4.3 The Mean Value Theorem

The mean value theorem is one of the most useful theorems in analysis. Under some assumptions,
It relates f and f 0 without involving limits explicitly. So it allow us to get more information about
the function from its derivative. The proof of the M.V.T is based on the existence of local extrema.

4.17 Definition. (Local Extrema )

? A function f is said to have a local maximum

at x0 if f is defined on I(x0 ) a neighborhood of
x0 and f (x) ≤ f (x0 ) for all x ∈ I(x0 ).

? A function f is said to have a local minimum

at x0 if f is defined on I(x0 ) a neighborhood of
x0 and f (x) ≥ f (x0 ) for all x ∈ I(x0 ).

? A function f is said to have a local extremum

I(x0 ) I(x0 )
at x0 if it has either local maximum or local
x0 x0
minimum at x0 .

Local Extrema Of f .

The extremum points of a differentiable function defined on an open interval I should be searched
for among the points where the derivative vanish.

56
4.18 Theorem. (Local Extremum Theorem )
If f has a local extremum at x0 and if f is differentiable at x0 , then f 0 (x0 ) = 0.

Proof. Suppose that f has a local maximum at x0 . Then there exists I a neighborhood of x0 such
that f (x) ≤ f (x0 ) for all x ∈ I. Hence,
? f (x)−f
x−x0
(x0 )
≥ 0 if x ∈ I ∧ x < x0 ,
? f (x)−f
x−x0
(x0 )
≤ 0 if x ∈ I ∧ x > x0 .
So f`0 (x0 ) ≥ 0 and fr0 (x0 ) ≤ 0. Thus f 0 (x0 ) = f`0 (x0 ) = fr0 (x0 ) = 0. The proof for the local minimum
case is similar.

4.19 Theorem. (Rolle’s Theorem )

Let f be continuous on [a, b] and differentiable on ]a, b[. If f (a) = f (b), then there exists a point
c ∈]a, b[ such that f 0 (c) = 0.

Proof. By the extreme value theorem there exist c, d ∈ [a, b] s.t min f (x) = f (c) and max f (x) =
x∈[a,b] x∈[a,b]
f (d). Hence f (c) ≤ f (x) ≤ f (d), ∀x ∈ [a, b].If f (c) = f (d), then f is a constant function and
f 0 (x) = 0 ∀x ∈]a, b[. If f (c) 6= f (d), then either c ∈]a, b[ or d ∈]a, b[. If c ∈]a, b[, then f has a local
minimum at c and by the local extremum theorem we get f 0 (c) = 0. Similarly, if d ∈]a, b[, then f
has a local maximum at d and by the local extremum theorem we get f 0 (d) = 0.

4.20 Theorem. (Mean Value Theorem )

If f be continuous on [a, b] and differentiable on ]a, b[ then there exists a point c ∈]a, b[ such that
f (b)−f (a)
b−a = f 0 (c).

Proof. The function h(x) := (f (b) − f (a))x − (b − a)f (x) is continuous on [a, b] and differentiable on
]a, b[ and satisfies h(a) = h(b). By Rolle’s theorem, there exists c ∈]a, b[ such that h0 (c) = 0. But
h0 (c) = f (b) − f (a) − (b − a)f 0 (c). Thus there exists c ∈]a, b[ such that f (b)−f
b−a
(a)
= f 0 (c).

Here are some applications of the mean value theorem.

4.21 Corollary. If f is differentiable on an open interval I and f 0 (x) = 0 for all x ∈ I, then there
exists a constant c such that f (x) = c for all x ∈ I.

Proof. Fix x0 ∈ I and claim f (x) = f (x0 ) for all x ∈ I \ {x0 }. Applying the M.V.T for f on
[x, x0 ] ⊂ I (or [x0 , x]):
? Since f is continuous on I and [x, x0 ] ⊂ I, then f is continuous on [x, x0 ].
? Since f is differentiable on I and ]x, x0 [⊂ I, then f is differentiable on ]x, x0 [.
By M.V.T, ∃ c ∈]x, x0 [ such that f (x)−f (x0 )
x−x0 − = f 0 (c) = 0. Thus f (x) = f (x0 ) for all x ∈ I.

4.22 Corollary. If f and g are differentiable on an open interval I and f 0 (x) = g 0 (x) for all x ∈ I,
then there exists a constant c such that f (x) = g(x) + c for all x ∈ I.

57
Proof. Apply the previous corollary to the function h := f − g.

4.23 Corollary. Let f be differentiable on an open interval I.

1. If f 0 (x) ≥ 0 ∀x ∈ I, then f is increasing on I.

2. If f 0 (x) ≤ 0 ∀x ∈ I, then f is decreasing on I.

3. If f 0 (x) > 0 ∀x ∈ I, then f is strictly increasing on I.

4. If f 0 (x) < 0 ∀x ∈ I, then f is strictly decreasing on I.

Proof. Let x, y ∈ I such that x < y. Applying the M.V.T for f on [x, y] ⊂ I:
? Since f is continuous on I and [x, y] ⊂ I, then f is continuous on [x, y].
? Since f is differentiable on I and ]x, y[⊂ I, then f is differentiable on ]x, y[.
By M.V.T, ∃ c ∈]x, y[ such that f (y)−f
y−x
(x)
= f 0 (c). We discuss four cases:

f (y)−f (x)
1. If f 0 (x) ≥ 0 ∀x ∈ I, then y−x = f 0 (c) ≥ 0. So f (x) ≤ f (y) and thus f is inc on I.

f (y)−f (x)
2. If f 0 (x) ≤ 0 ∀x ∈ I, then y−x = f 0 (c) ≤ 0. So f (x) ≥ f (y) and thus f is dec on I.

f (y)−f (x)
3. If f 0 (x) > 0 ∀x ∈ I, then y−x = f 0 (c) > 0. So f (x) < f (y) and thus f is st. inc on I.

f (y)−f (x)
4. If f 0 (x) < 0 ∀x ∈ I, then y−x = f 0 (c) < 0. So f (x) > f (y) and thus f is st. dec on I.

4.24 Remark.

1. If f is strictly increasing (resp increasing ) on the intervals ]a, b[ and ]b, c[ and and f is
continuous at b then f is strictly increasing (resp increasing ) on ]a, c[. For example f (x) = x3
.Its derivative f 0 (x) = 3x2 > 0 on R∗ . Since f is continuous at zero, then f is strictly increasing
on R.

2. If f is strictly decreasing (resp decreasing ) on the intervals ]a, b[ and ]b, c[ and and f is
continuous at b then f is strictly decreasing (resp decreasing ) on ]a, c[.

3. Corollary 4.23 does not work if I is disjoint union of open intervals. For example f (x) = x1 . Its
derivative f 0 (x) = − x12 < 0 on ]−∞, 0[∪]0, ∞[ but f is not strictly decreasing on ]−∞, 0[∪]0, ∞[.

4. Corollary
 4.21 does not work if I is disjoint union of open intervals. For example
1 if x ∈] − ∞, 0[
f (x) = Its derivative f 0 (x) = 0 on I =] − ∞, 0[∪]0, ∞[ but f is not a
2 if ]0, ∞[.
constant function on I.
1
4.25 Example. Show that x+1 < ln(x + 1) − ln x < x1 ∀x ∈]0, ∞[.

O
Solution. Let x ∈]0, ∞[ and f (t) = ln t. Let us try MVT on [x, x + 1]:

O
1 Since f is continuous on ]0, ∞[, then f is continuous on [x, x + 1] ⊆]0, ∞[.
2 Since f is differentiable on ]0, ∞[, then f is differentiable on ]x, x + 1[⊆]0, ∞[.
By the MVT, there exists c ∈]x, x + 1[ such that f (x+1)−f
(x+1)−x
(x)
= f 0 (c) = 1c . Hence

1 1 1
< ln(x + 1) − ln x = < .
x+1 c x

58
4.4 Taylor’s Formula

4.26 Theorem. (Taylor-Lagrange Formula )

O
Let f : [a, b] 7→ R ( or f : [b, a] 7→ R) and n ∈ N. If
i f ∈ C n ([a, b]) and
Oii f (n+1) exists on ]a, b[,
then there exist c ∈]a, b[, not necessarily unique, such that

(b − a)2 00 (b − a)n (n) (b − a)n+1 (n+1)

f (b) = f (a) + (b − a)f 0 (a) + f (a) + · · · + f (a) + f (c).
2! n! (n + 1)!

Proof. For simplicity we proof the theorem for n = 1.

Given that f ∈ C 1 ([a, b]) and f (2) = f 00 exists on ]a, b[. Claim there exist c ∈]a, b[ such that
(b − a)2 00
f (b) = f (a) + (b − a)f 0 (a) + f (c) :
2!
(b−x)2
Let h(x) := f (x) + (b − x)f 0 (x) + 2! M where M is chosen so that h(a) = h(b). Hence
2
M= [f (b) − f (a) − (b − a)f 0 (a)].
(b − a)2
Since h(b) = f (b) = h(a), then f (b) = f (a) + (b − a)f 0 (a) + (b − a)2 M .
The result follows if we show M = f 00 (c). Let us try Roll’s theorem for h on [a, b]:
• h(a) = h(b) (by the choice of M ).
• h ∈ C([a, b]) (sum of continuous functions).
• h is differentiable on ]a, b[ as sum of differentiable functions (f 00 = (f 0 )0 exists on ]a, b[ means f 0 is
differentiable on ]a, b[ ).
Then by Roll’s theorem, there exists c ∈]a, b[ such that h0 (c) = 0. This gives M = f 00 (c).

Note that Taylor-Lagrange with n = 0 is the mean value theorem.

4.27 Theorem. (Mac-Laurin Formula a = 0, b = x ∈ R )

O
Let f : [0, x] 7→ R ( or f : [x, 0] 7→ R) and n ∈ N. If
i f ∈ C n ([0, x]) and
Oii f (n+1) exists on ]0, x[,
then there exist c ∈]0, x[, not necessarily unique, such that

x2 00 xn (n) xn+1 (n+1)

f (x) = f (0) + xf 0 (0) + f (0) + · · · + f (0) + f (c)
2! n! (n + 1)!
n
P xk (k) xn+1 (n+1)
= k! f (0) + (n+1)! f (c).
k=0

4.28 Example. Give the Mac-Laurin expansion up to order n for the functions f (x) = ex .
Solution. Note that f ∈ C n ([0, x]) and f (n+1) exists on ]0, x[. Also f (n) (x) = ex and f (n) (0) = 1 for
all n ∈ N∗ . Thus
x2 xn xn+1 c n
xk xn+1 c
ex = 1 + x +
P
+ ··· + + e = k! + (n+1)! e .
2! n! (n + 1)! k=0

Similarly, the formula holds for x < 0.

59
4.5 Inverse Function Theorem

4.29 Theorem. (Inverse Function Theorem-2)

Let I be an interval and f : I 7→ R be continuous and injective on I. If f is differentiable at x0 ∈ I
and f 0 (x0 ) 6= 0, then f −1 is differentiable at y0 = f (x0 ) and (f −1 )0 (f (x0 )) = f 0 (x
1
0)
.

Proof. Let y = f (x) ∈ J = f (I) such that y 6= y0 . Then x 6= x0 and

f −1 (y) − f −1 (y0 ) x − x0 1
= = f (x)−f (x0 )
.
y − y0 f (x) − f (x0 )
x−x0

Since f −1 is continuous on J, we have y → y0 implies x = f −1 (y) → f −1 (y0 ) = x0 . Therefore

f −1 (y) − f −1 (y0 ) 1 1
lim = lim = .
y→y0 y − y0 x→x0 f (x)−f (x0 ) f 0 (x0 )
x−x0

4.30 Theorem. (Inverse Function Theorem-3)

Let I be an interval and f : I 7→ R be differentiable on I and f 0 (x) 6= 0 for all x ∈ I. Then
O O O
i f is injective, ii J = f (I) is an interval and iii f −1 is differentiable on J and we have
(f −1 )0 (f (x)) = f 01(x) ∀x ∈ I.

Proof. The result follows from IFT-1 and IFT-2 if we show f is injective.
Assume that f is not injective, then there exist a, b ∈ I such that a 6= b and f (a) = f (b). Since
[a, b] ⊆ I, then f is continuous on [a, b] and differentible on ]a, b[. Hence by Roll’s theorem there
exists c ∈]a, b[⊆ I such that f 0 (c) = 0 which is a contradiction. Thus f is injective.

O
Note that;
1 If f ∈ C 1 (I) and f (x) 6= 0 ∀x ∈ I, then g = f1 ∈ C 1 (I).
O
2 More general, If f ∈ C n (I) and f (x) 6= 0 ∀x ∈ I, then g = 1
f ∈ C n (I).

O
4.31 Theorem. (Inverse Function Theorem-4)
Let I be an interval and f : I 7→ R be of class C n on I and f 0 (x) 6= 0 for all x ∈ I. Then i f is
O O
injective, ii J = f (I) is an interval and iii f −1 is of class C n on J.

4.6 Exercises  √
x if 0 ≤ x ≤ 1
Exercise 1. Let a, b ∈ R and f (x) = Determine a and b so that
a(x2 − 1) + b if x > 1.
f is of class C 1 on ]0, ∞[.
f ( x2 )

if x > 0
Exercise 2. Let f be a differentiable function on R such that f (0) = 1 and g(x) =
x+1 if x ≤ 0.
1. Show that g is continuous on R.

2. Assume that fr0 (0) = 2. Show that g is differentiable on R and determine g 0 (x).

60
Exercise 3. Let f :] − a, a[7→ R be a differentiable function. Prove that if f is an even function then
f 0 is an odd function and if f is an odd function then f 0 is an even function.

Exercise 4. Let f : R → 7 R be differentiable function on R such that f (0) = 0. Show that

f (x)f (−x) 0 2
lim x2
= −(f (0)) . ( Hint: Use the equivalent definition of differentiation at a point i.e., finite expansion
x→0
to order one.)

Exercise 5. Lat f ∈ C 1 ([a, b]). Show that there exists k ≥ 0 such that
|f (x) − f (y)| ≤ k|x − y| ∀x, y ∈ [a, b].
Exercise 6. (Extension by differentiability)
1. Let I be an open interval and a ∈ I. Suppose that f is continuous on I, differentiable on I \ {a}
and lim f 0 (x) = ` (` ∈ R). Prove that f is differentiable at a, f 0 (a) = ` and f 0 is continuous
x→a
at a.
p
2. Show that f (x) = x |x| is differentiable on R.
x sin x1
 2
if x 6= 0
3. Let f (x) = Show that f is differentiable on R but f is not of class
0 if x = 0.
C 1 on R.( Check that f 0 (0) exists but lim f 0 (x) does not exist.)
x→0
Exercise 7. (Extension by differentiability)
1. Let I = [a, b[ . Suppose that f is continuous on I = [a, b[, differentiable on I \ {a} =]a, b[ and
lim f 0 (x) = ` (` ∈ R). Prove that f is differentiable from right at a, fr0 (a) = ` and f 0 is
x→a+
continuous from right at a.
√
cos x−1
2. Compute lim x .
x→0+
Exercise 8. (No extension by differentiability)
Let I be an open interval and a ∈ I. Suppose that f is continuous on I, differentiable on I \ {a} and
lim f 0 (x) = ±∞. Prove that lim f (x)−f
x−a
(a)
= ±∞.(i.e., f is not diffeentiable at a and the tangent to
x→a x→a
the representative curve of f at the point (a, f (a)) is vertical.)

f (x2 ln x + x2 )

if x > 0
Exercise 9. Let f be differentiable on R such that f (0) = 1 and g(x) =
x+1 if x ≤ 0.
1. Show that g is continuous on R.
2. Assume that lim f 0 (x) = 2. Show that g is differentiable on R and determine g 0 (x).
x→0+

Exercise 10. Let f be differentiable on R such that f 0 (x) 6= 0 for all x ∈ R. Show that f is not a
periodic function.

Exercise 11. Let I be an open interval and f : I 7→ R satisfying ∀x, y ∈ I, |f (x) − f (y)| ≤ K|x − y|2
for some k > 0. Show that f is a constant function.
 2
x ln x2 if x 6= 0
Exercise 12. Let f (x) = Show that f is of class C 1 on R but f is not
0 if x = 0.
twice differentiable on R.
Exercise 13. Let a < b < c and f : [a, c] 7→ R be continuous on [a, c], twice differentiable on ]a, c[
and f (a) = f (b) = f (c).
1. Show that there exist ξ1 ∈]a, b[, ξ2 ∈]b, c[ and such that f 0 (ξ1 ) = f 0 (ξ2 ).
2. Deduce that there exists ξ ∈]a, c[ such that f 00 (ξ) = 0.

61
Exercise 14.

1. Let f : [0, 1] 7−→ R be continuous on [0, 1] with f (0) = f (1) = 0 and twice differentiable on
]0, 1[ with f 00 < 0 on ]0, 1[.
0
(a) Show that there exists c ∈]0, 1[ such that f (c) = 0.
0 0
(b) Show that f > 0 on ]0, c[ and f < 0 on ]c, 1[.
(c) Deduce that f (x) > 0, for all x ∈]0, 1[.

2. Let a, b ∈ R such that a > b > 1. consider the function f defined by

f (x) = ln ln(xa + (1 − x)b) − x ln ln a − (1 − x) ln ln b.

(a) Show that f is continuous on [0, 1] with f (0) = f (1) = 0 and twice differentiable on ]0, 1[
with f 00 < 0 on ]0, 1[.
(b) Deduce the following inequality :

ln ln(xa + (1 − x)b) > x ln ln a + (1 − x) ln ln b, ∀ x ∈]0, 1[.

√
(c) Show that ln( a+b
2 )> ln a ln b.

Exercise 15. Show that the following inequalities:

x2 √ x2 x3
1+x− ≤ 1 + 2x < 1 + x − + ∀x ∈ [0, ∞[.
2 2 2
Exercise 16. Let f : [ π2 , π[7→ R defined by f (x) = 1
sin x .

1. Show that f defines a bijection from I = [ π2 , π[ onto some interval J to be determined.

2. Find the domain of differentiability of f −1 and calculate (f −1 )0 (y) (without determining f −1 ).

Exercise 17. Let f :] − π2 , π2 [7→ R defined by f (x) = tan x − x.

1. Show that f defines a bijection from I =] − π2 , π2 [ onto some interval J to be determined.

2. Let g = f −1 .

(a) Study the continuity and monotony of g.

(b) Show that g is of class C 2 on J except at one point a to be determined.
(c) Let y0 = 1 − π4 . Calculate g(y0 ), g 0 (y0 ) and g 00 (y) then deduce Taylor up to n = 1 for
g = f −1 .

62
CHAPTER 5
ELEMENTARY FUNCTIONS

5.1 Trigonometric Functions And Inverses

We consider the standard definitions of sine and cosine as coordinates of points on the unit circle.
From this definition one can derive the usual trigonometric identities.

5.1.1 Sine And Arcsine Functions

The function sin : R 7→ [−1, 1] has the following properties:
x → sin x

• dom(sin) = R and −1 ≤ sin x ≤ 1 ∀x ∈ R.

• sin ∈ C ∞ (R) and sin(n) (x) = sin(x+n π2 ) ∀x ∈ R ∀n ∈ N. In particular sin0 (x) = cos x ∀x ∈ R.

dom(sin) = R is centered at 0.
• sin is odd.
sin(−x) = − sin(x) ∀x ∈ R.

sin(x + 2π) = sin x ∀x ∈ R.
• sin is 2π-periodic.
sin(x + 2kπ) = sin x ∀x ∈ R, ∀k ∈ Z.
y
y
π arcsin x y = x
2

sin x
1
1
x
x
−2π − 3π −π − π2 π π 3π 2π
2
2 −1 2 − π2 −1 1 π
2

−1

− π2

x 7→ sin x is 2π−periodic and odd on D = R.

sin : [− π2 , π2 ] 7→ [−1, 1], arcsin : [−1, 1] 7→ [− π2 , π2 ].
x → sin x x → arcsin x

The restriction of the sine function, sin : [− π2 , π2 ] 7→ [−1, 1] has the following properties:
x → sin x

• Continuous on I = [− π2 , π2 ].
• Strictly increasing on I = [− π2 , π2 ]:
π π π π
sin0 (x) = cos x > 0 on ] − , [ =⇒ sin is st inc on] − , [
2 2 2 2
π π π
=⇒ sin is st inc on[− , ]. (∵ sin is continuous at ± )
2 2 2

63
Then by the inverse function theorem;
? The function sine defines a bijection from the interval I = [− π2 , π2 ] onto the interval J = [sin(− π2 ), sin( π2 )] =
[−1, 1].
π π π π
sin : [− , ] 7→ [−1, 1], arcsin : [−1, 1] 7→ [− , ].
2 2 2 2
x → y=sin x y → x=arcsin y

π π
y = sin x, x ∈ [− , ] ⇐⇒ x = arcsin y, y ∈ [−1, 1].
2 2
? Since sin is differentiable on I = [− π2 , π2 ], then arcsin is differentiable on
π
J \ {y = sin x : sin0 (x) = 0} = J \ {y = sin x : x = ± } =] − 1, 1[.
2
? Since sin is of class C ∞ on I, then arcsin is of class C ∞ on the domain of differentiation ] − 1, 1[.

1 1 1
arcsin0 (y) = (sin−1 )0 |y=sin x = 0 = =p ∀y ∈] − 1, 1[.
sin (x) cos x 1 − y2

5.1.2 Cosine And Arccosine Functions

The function cos : R 7→ [−1, 1] has the following properties:
x → sin x

• dom(cos) = R and −1 ≤ cos x ≤ 1 ∀x ∈ R.

• cos ∈ C ∞ (R) and cos(n) (x) = cos(x + n π2 ) ∀x ∈ R ∀n ∈ N. In particular cos0 (x) = − sin x.

dom(cos) = R is centered at 0.
• cos is even.
cos(−x) = cos(x) ∀x ∈ R.

cos(x + 2π) = cos x ∀x ∈ R.
• cos is 2π-periodic.
cos(x + 2kπ) = cos x ∀x ∈ R, ∀k ∈ Z.
y
y
arccos x y=x
π

1
x
π
−2π − 3π −π − π2 π π 3π 2π 2
2
2 −1 2
1

x
−1 1 π π
2

cos x
−1
x 7→ cos x is 2π−periodic and even on D = R.
cos : [0, π] 7→ [−1, 1], arccos : [−1, 1] 7→ [0, π].
x → sin x x → arcsin x

The restriction of the cosine function, cos : [0, π] 7→ [−1, 1] has the following properties:
x → sin x

• Continuous on I = [0, π].

• Strictly decreasing on I = [0, π]:

cos0 (x) = − sin x < 0 on ]0, π[ =⇒ cos is st dec on]0, π[

=⇒ cos is st dec on[0, π]. (∵ cos is continuous at 0 and π)

64
Then by the inverse function theorem;
? The function cosine defines a bijection from the interval I = [0, π] onto the interval J = [cos(π), cos(0)] =
[−1, 1].
cos : [0, π] 7→ [−1, 1], arccos [−1, 1] 7→ [0, π].
x → y=sin x y → x=arcsin y

y = cos x, x ∈ [0, π] ⇐⇒ x = arccos y, y ∈ [−1, 1].

? Since cos is differentiable on I = [0, π], then arccos is differentiable on

J \ {y = cos x : cos0 (x) = 0} = J \ {y = sin x : x = 0, x = π} =] − 1, 1[.

? Since cos is of class C ∞ on I, then arccos is of class C ∞ on the domain of differentiation ] − 1, 1[.

1 1 −1
arccos0 (y) = (cos−1 )0 |y=cos x = = =p ∀y ∈] − 1, 1[.
cos0 (x) − sin x 1 − y2

5.1.3 Tangent And Arctangent Functions

sin x
The tangent function is defined by tan x = cos x on D = R \ { π2 + kπ : k ∈ Z}. The function
tan : D 7→ R has the following properties:
x → tan x

• dom(tan) = R \ { π2 + kπ : k ∈ Z} = ∪ ] − π
2 + kπ, π2 + kπ[.
k∈Z

sin
• tan = is differentiable on D because its the quotient of differentiable functions with
cos
denominator never vanishes.
1
tan0 (x) = = 1 + tan2 x.
cos2 x

• tan ∈ C ∞ (D) because its the quotient of C ∞ functions with denominator never vanishes.

dom(tan) is centered at 0.
• tan is odd.
tan(−x) = − tan(x) ∀x ∈ D

tan(x + π) = tan x ∀x ∈ D.
• tan is π-periodic.
tan(x + kπ) = tan x ∀x ∈ D, ∀k ∈ Z.

y y

π
2
x x
− 5π −2π− 3π −π − π2 π π 3π 2π 5π −−π2π π
2 2 2 2 2 2
2

π π π π
x 7→ tan x is π−periodic and odd on D. tan : ] − , [7→
2 2
R, arctan : R 7→] − , [.
2 2
x → tan x x → arctan x

π π
The restriction of the tangent function, tan : ] − , [7→ R
2 2
has the following properties:
x → tan x

• Continuous on I =] − π2 , π2 [.

65
 • Strictly increasing on I =] − π2 , π2 [. (tan0 (x) = 1 + tan2 x > 0 ∀x ∈] − π2 , π2 [)

Then by the inverse function theorem;

? The function tangent defines a bijection from the interval I =] − π2 , π2 [ onto the interval J =
] limπ tan x, limπ tan x[=] − ∞, +∞[= R.
x→− 2 x→ 2

π π π π
tan : ] − , [7→ R, arctan : R 7→] − , [.
2 2 2 2
x → y=tan x y → x=arctan y

π π
y = tan x, x ∈] −, [ ⇐⇒ x = arctan y, y ∈ R.
2 2
? Since tan is differentiable on I =] − π2 , π2 [, then arctan is differentiable on

J \ {y = tan x : tan0 (x) = 0} = J = R.

? Since tan is of class C ∞ on I, then arctan is of class C ∞ on the domain of differentiation R.

1 1 1
arctan0 (y) = (tan−1 )0 |y=tan x = 0 = 2 = ∀y ∈ R.
tan (x) 1 + tan x 1 + y2

5.1.4 Cotangent And Arccotangent Functions

The cotangent function is defined by cot x = cos x
sin x on D = R \ {π + kπ : k ∈ Z}. The function
π
cot : D 7→ R satisfies cot x = − tan(x − 2 ). Hence the graph of the function cot is obtained from
x → tan x
the graph of tan by translating horizontally towards the left and reflecting the graph of tan with
respect to x axis.

y y

π
π
2
x x
−2π −π π 2π π π
2

x 7→ cot x is π−periodic and odd on D. cot : ]0, π[7→ R, arccot : R 7→]0, π[ .

x → cot x x → arccotx

5.2 The Power Function

The power function of exponent α ∈ R is defined on ]0, ∞[ by pα : ]0, ∞[7→]0, ∞[ .
x → xα :=eα ln x
The power function has the following properties:
• pα is continuous, differentiable and of class C ∞ on ]0, ∞[ and
α α ln x α α
(xα )0 = (eα ln x )0 = e = x = αxα−1 .
x x
• xα xβ = xα+β and (xα )β = xαβ ∀α, β ∈ R ∀x > 0.

66
 Some useful limits;  α > 1 as x2
 0 if α > 0 α=1
lim xα = lim eα ln x = 1 if α = 0
x→0+ x→0+
∞ if α < 0.


1
0 < α < 1 as x 2

 α=0
 ∞ if α > 0 1
α < 0 as 1
x
lim xα = lim eα ln x = 1 if α = 0
x→∞ x→∞ 1
0 if α < 0.

The graph of x 7→ xα on ]0, ∞[.

Note that;
• If α ∈ N, then the power function x 7→ xα is defined and of class C ∞ on R.
• If α ∈ Z \ N = {−1, −2, · · · }, then the power function x 7→ xα is defined and of class C ∞ on R∗ .

5.3 Logarithmic and Exponential Functions

In this section we recall the logarithmic function and its inverse the exponential function. The
comparison between Logarithm, power and exponential functions is stated as exercises.

5.3.1 Logarithm Of Base e.

Rx
• ln x := 1 1t dt is well defined on ]0, ∞[. y
• ln 1 = 0. ln x
• ln is differentiable on ]0, ∞[ and ln0 x = x1 .
• ln0 ∈ C ∞ (]0, ∞[) =⇒ ln ∈ C ∞ (]0, ∞[).
• ln(xy) = ln(x) + ln(y) ∀x, y > 0.
Proof. Fix y > 0. Let f (x) = ln(xy) − ln(x) − ln(y). x
0
Since f (x) = 0 on the interval ]0, ∞[, then f is 1
constant on this interval and f (x) = f (1) = 0.
• ln x1 = − ln x ∀x > 0.
x
Proof. 0 = ln x
= ln x + ln x1 .
x
• ln y = ln x − ln y ∀x, y > 0.
x 1
Proof. ln y
= ln x + ln y
= ln x − ln y.
• ln xα = α ln x, ∀x > 0 ∀α ∈ R. The graph of x 7→ ln x on ]0, ∞[.

Note that;
• If xy > 0, then ln(xy) = ln |x| + ln |y| and ln xy = ln |x| − ln |y|.
u0 (x)
• (ln u(x))0 = u(x) provided that u(x) > 0 and u is differentiable at x(Chain Rule).
ln x
• lim ln x = −∞, lim ln x = ∞, lim = 0 and lim x ln x. (The proof left as exercise.)
x→0+ x→∞ x→∞ x x→0+

5.3.2 Logarithm Of Base a > 0, a 6= 1.

• loga (x) := ln x
ln a x ∈]0, ∞[.
0 1
• (loga (x)) := x ln x ∈]0, ∞[. Hence loga is strictly increasing if a > 1 and strictly decreasing if
a
0 < a < 1.

67
5.3.3 Exponential Of Base e.
Since ln :]0, ∞[7→ R is continuous and strictly increasing, then it is a bijection from I =]0, ∞[ onto
J =] lim ln x, lim ln x[=] − ∞, +∞[. We define exp = ln−1 , the inverse of the function ln. Usually
x→0+ x→∞
we denote exp(x) by ex .
ln : ]0, ∞[7→ R, exp : R 7→]0, ∞[.
x → y=ln x y → x=ey

y = ln x, x ∈]0, ∞[ ⇐⇒ x = ey , y ∈ R.
ln x x
e =x ∀x ∈]0, ∞[ ln(e ) = x ∀x ∈ R.
? Since ln is differentiable on I =]0, ∞[, then exp is differentiable on

J \ {y = ln x : ln0 (x) = 0} = J = R.

? Since ln is of class C ∞ on I, then exp is of class C ∞ on the domain of differentiation R.

1
(exp)0 (y) = (ln−1 )0 |y=ln x = 0 = x = ey ∀y ∈ R.
ln (x)

Consequently, (ex )(n) = ex ∀x ∈ R, ∀n ∈ N.

• ex > 0 ∀x ∈ R. y
• exp is strictly increasing on R: ex
x < 0 =⇒ ex < e0 = 1.
x > 0 =⇒ ex > e0 = 1.
•e x+y = ex ey ∀x, y ∈ R.
−x
• e = e1x ∀x ∈ R.
x
• ex−y = eey ∀x, y ∈ R.
• (ex )α = eαx ∀x, α ∈ R.

1
x

The graph of x 7→ ex on R.

5.3.4 Exponential Of Base a > 0, a 6= 1.

Since loga :]0, ∞[7→ R is continuous and strictly increasing, then it is a bijection . Its inverse expa (y)
is usually denoted by ay and defined by expa : R 7→]0, ∞[.
y → x=ay

loga : ]0, ∞[7→ R, expa : R 7→]0, ∞[.

x → y=ln x y → x=ay

y = loga x, x ∈]0, ∞[ ⇐⇒ x = ay , y ∈ R.
aloga x = x ∀x ∈]0, ∞[ loga (ax ) = x ∀x ∈ R.
• expa is differentiable on R and(ax )0 = (ex ln a )0 = ln(a)ex ln a = (ln a)ax ∀x ∈ R.
∞ if a > 1
• lim x±1 ax = lim x±1 ex ln a =
x→∞ x→∞ 0 if a < 1.
Exercise. Show, using three different ways, definition of derivative, equivalent definition of derivative
and MVT the following:

1. lim ln(1+x)
x = 1. Deduce lim (1 + x1 )x = e and lim (1 + xa )x = ea (a ∈ R).
x→0 x→±∞ x→±∞

2. lim sinx x = 1.
x→0

68
5.4 Hyperbolic Functions And Inverses
The functions cosine hyperbolic, sine hyperbolic and tangent hyperbolic are defined on R by:
ex + e−x ex − e−x sinh x ex − e−x
cosh x := , sinh x := , tanh x := = x .
2 2 cosh x e + e−x
• cosh is positive (cosh x > 0, ∀x ∈ R) and even function.
• sinh and tanh are odd functions.

O O O
The hyperbolic functions satisfy the following identities:
1 cosh x + sinh x = ex . 2 cosh x − sinh x = e−x . 3 cosh2 x − sinh2 x = 1.
The point of coordinates (X, Y ) = (cosh x, sinh x) runs along the right branch of the hyperbola
X 2 − Y 2 = 1 as x varies.
O O
If we divide 3 by cosh2 x > 0, then we obtain 4 1 − tanh2 x =
1
cosh2 x
.
Also, one may verify easily that
cosh(a + b) = cosh a cosh b + sinh a sinh b, sinh(a + b) = cosh a sinh b + sinh a cosh b
tanh a + tanh b
tanh(a + b) = .
1 + tanh a tanh b

5.4.1 Sine Hyperbolic And Inverse Sine Hyperbolic

The function sinh : R 7→ R has the following properties:
x −e−x
x→ e 2

• dom(sinh) = R.
• sinh, cosh are differentiable on R and sinh0 x = cosh x, cosh0 x = sinh x.

∞ (n) cosh x if n odd
• sinh ∈ C (R) and sinh (x) =
sinh x if n even .

dom(sinh) = R is centered at 0.
• sinh is odd.
sinh(−x) = − sinh(x) ∀x ∈ R.
The sine hyperbolic function, sinh : R 7→ R has the following properties:
x → sinh x

• Continuous on I = R. sinh x
• Strictly increasing on I = R: y
sinh0 (x) = cosh x > 0 ∀x ∈ R.
Then by the inverse function theorem; sinh−1 x
The function sinh is a bijection from the
interval I = R onto the interval
1 x
J =] lim sinh x, lim sinh x[=] − ∞, ∞[.
x→−∞ x→∞

sinh : R 7→ R , sinh−1 : R 7→ R .
x → y=sinh x y → x=sinh−1 y
−1
y = sinh x, x ∈ R iff x = sinh y, y ∈ R.
? Since sinh is differentiable on I = R, then
sinh−1 is differentiable on
J \ {y = sinh x : sinh0 (x) = 0} = J = R. sinh : R 7→ R , sinh−1 : R 7→ R .
x → sinh x x→sinh−1 x

? Since sinh is of class C ∞ on I = R, then sinh−1 is of class C ∞ on the domain of differentiation R.

1 1 1 1
(sinh−1 )0 (y) = (sinh−1 )0 |y=sinh x = 0 = =q =p ∀y ∈ R.
sinh (x) cosh x 1 + y2
1 + sinh2 (x)

Note that some references use argsinh, arsinh or arcsinh to denote sinh−1 .

69
5.4.2 Cosine Hyperbolic And Inverse Cosine Hyperbolic
The function cosh : R 7→ [1, ∞[ has the following properties:
x +e−x
x→ e 2

• dom(cosh) = R.
ex +e−x
• cosh x = 2 ≥ 1 ∀x ∈ R and cosh 0 = 1.

• cosh is differentiable on R and cosh0 x = sinh x.


cosh x if n even
• cosh ∈ C ∞ (R) and cosh(n) (x) =
sinh x if n odd .

dom(cosh) = R is centered at 0.
• cosh is even.
cosh(−x) = cosh(x) ∀x ∈ R.

The restriction of the cosine hyperbolic function, cosh : [0, ∞[7→ [1, ∞[ has the following properties:
x → cosh x

• Continuous on I = [0, ∞[. cosh x

• Strictly increasing on I = [0, ∞[: y
cosh0 (x) = sinh x > 0 ∀x ∈]0, ∞[.
Then by the inverse function theorem; cosh−1 x
The function cosh is a bijection from the
interval I = [0, ∞[ onto the interval
1
J = [cosh 0, lim cosh x[= [1, ∞[. x
x→∞
1
cosh : [0, ∞[7→ [1, ∞[, cosh−1 : [1, ∞[7→ [0, ∞[.
x→y=cosh x y→x=cosh−1 y
y = cosh x, x ∈ [0, ∞[ iff x = cosh−1 y, y ∈ [1, ∞[.

? Since cosh is differentiable on I = [0, ∞[,

then cosh−1 is differentiable on
J \ {y = cosh x : cosh0 (x) = 0} =]1, ∞[. cosh : [0, ∞[7→ [1, ∞[, cosh−1 : [1, ∞[7→ [0, ∞[.
x → cosh x x→cosh−1 x

? Since cosh is of class C ∞ on I, then cosh−1 is of class C ∞ on the domain of differentiation ]1, ∞[.

1 1 1 1
(cosh−1 )0 (y) = (cosh−1 )0 |y=cosh x = = =q =p ∀y ∈]1, ∞[.
cosh0 (x) sinh x
cosh2 (x) − 1
2
y −1

Note that some references use argcosh, arcosh or arccosh to denote cosh−1 .

5.4.3 Tangent Hyperbolic And Inverse Tangent Hyperbolic

The function tanh : R 7→] − 1, 1[ has the following properties:
x −x
x→ ex −e−x
e +e

• dom(tanh) = R.

ex − e−x e2x − 1 1 − e−2x

• tanh x = = = .
ex + e−x e2x + 1 1 + e−2x
e2x − 1
• lim tanh x = lim = −1. (y = −1 H.A).
x→−∞ x→−∞ e2x + 1

1 − e−2x
• lim tanh x = lim = 1. (y = 1 H.A).
x→∞ x→∞ 1 + e−2x

70
 • tanh is differentiable on R and
 sinh x 0 cosh2 x − sinh2 x 1
tanh0 x = = 2 = 1 − tanh2 x = .
cosh x cosh x cosh2 x

• tanh ∈ C ∞ (R).( Quotient of C ∞ functions with denominator never vanishes)

• tanh is odd.

The tangent hyperbolic function, tanh : R 7→] − 1, 1[ has the following properties:
x → tanh x
• Continuous on I = R. tanh−1 x
• Strictly increasing on I = R: y
tanh0 (x) = cosh1 2 x > 0 ∀x ∈ R.
Then by the inverse function theorem;
The function tanh is a bijection from the tanh x
1
interval I = R onto the interval
J =] lim tanh x, lim tanh x[=] − 1, 1[. x
x→−∞ x→∞
−1 1
tanh : R 7→] − 1, 1[, tanh−1 : ] − 1, 1[7→ R .
x → y=tanh x y → x=tanh−1 y
−1
−1
y = tanh x, x ∈ R iff x = tanh y, y] − 1, 1[.

? Since tanh is differentiable on I = R, then

tanh−1 is differentiable on
tanh : R 7→] − 1, 1[, tanh−1 : ] − 1, 1[7→ R.
J \ {y = tanh x : tanh0 (x) = 0} = J =] − 1, 1[. x→tanh x x→tanh−1 x

? Since tanh is of class C ∞ on I = R, then tanh−1 is of class C ∞ on the domain of differentiation

] − 1, 1[.
1 1 1
(tanh−1 )0 (y) = (tanh−1 )0 |y=tanh x = = = ∀y ∈] − 1, 1[.
tanh0 (x) 2
1 − tanh x 1 − y2

Note that some references use argtanh, artanh or arctanh to denote tanh−1 .

One can show directly by a simple calculations the following identities:

√
1. sinh−1 (x) = ln(x + x2 + 1) ∀x ∈ R.
√
2. cosh−1 (x) = ln(x + x2 − 1) ∀x ∈ [1, ∞[.

3. tanh−1 (x) = 1
2 ln( 1+x
1−x ) ∀x ∈] − 1, 1[.

5.5 Exercises
1 + x
Exercise 1. Let f (x) = arctan and g(x) = arctan x.
1−x
1. Give the domains of definition of f and g.

2. Calculate f 0 (x).

3. Give a relation between f and g.

Exercise 2. Show that

π

1 2 if x > 0
arctan x + arctan =
x − π2 if x < 0.

71
Exercise 3.(property) Show that lim ln x = ∞ and lim ln x = −∞.
x→∞ x→0+
Exercise 4.(property)
√
1. Show that ln x < x ∀x ∈]0, ∞[.
ln x
2. Deduce that lim = 0.
x→∞ x

ex
3. Deduce that lim = ∞.
x→∞ x

Exercise 5.(property) Let a > 0( or a ∈ R), b > 0. Show that

(ln x)a
1. lim b = 0.
x→∞ x

ebx
2. lim a = ∞.
x→∞ x

ebx
3. lim a = ∞.
x→∞ (ln x)

Exercise 6. Let a ∈ R, b > 0. Show that lim xb | ln x|a = 0.

x→0+
Exercise 7. Let a ∈ R, b > 0. Show that lim |x|a ebx = 0.
x→−∞
2x 1 − x2
Exercise 8. Let f (x) = arcsin + arccos .
1 + x2 1 + x2
1. Find dom(f ).

2. Study the continuity of f and determine lim f (x).

x→∞

3. Show that arctan x + arctan x1 = π

2 ∀x > 0.

4. Deduce that lim (πx − 2x arctan x) = 2.

x→∞

72
CHAPTER 6
FINITE EXPANSIONS

6.1 Taylor And Finite Expansions

The finite expansion of a function near a point x0 ∈ R is the representation of the function as sum
of a polynomial of a certain degree and an infinitesimal function ( −→ 0). In a neighborhood of x0
x→x0
one can approximate complicated functions by polynomials which are easy to handle.

6.1 Definition. (Finite Expansion)

Let f : I 7→ R, I open interval, x0 ∈ I or boundary point, and n ∈ N. We say that f has finite
expansion of order n near x0 ”f.en (x0 )” if there exists a polynomial p of degree ≤ n and a function
ε : I 7→ R verifies ε(x) −→ 0 such that
x→x0

f (x) = p(x) + (x − x0 )n ε(x) ∀x ∈ I.

• p(x) is called the principle part of the finite expansion.

• (x − x0 )n ε(x) is called the remainder of the finite expansion (or the error of the approximation).

6.2 Remark.
• We will use the notation ” little oh” o(x − x0 )n = (x − x0 )n ε(x) with ε(x) −→ 0.
x→x0
• The change of variable t = x − x0 ( t −→ 0) allows us to perform f.e near zero instead of x0 ∈ R.
x→x0
1
• The change of variable t = x ( t −→ 0) allows us to perform f.e near zero instead of ±∞.
x→±∞
We will limit ourselves to study the finite expansion near zero.

1
6.3 Example. Show that f (x) = 1−x has f.en (0), n ∈ N.
Solution.

1 1 − xn+1 + xn+1
=
1−x 1−x
1 − xn+1 xn+1
= +
1−x 1−x
(1 − x)(1 + x + x2 + · · · + xn ) xn+1
= + ( Factorization)
1−x 1−x
2 n n x x
= 1 + x + x + ··· + x + x (ε(x) = −→ 0)
| {z } 1−x 1 − x x→0
p(x) | {z }
ε(x)
n
= p(x) + o(x ).

73
 1 1
f.en (0) for the functions x 7→ 1−x and x 7→ 1+x :

1
= 1 + x + x2 + x3 + · · · + xn + o(xn ).
1−x
1
= 1 − x + x2 − x3 + · · · + (−1)n xn + o(xn ). ( replace x by − x)
1+x

xn+1 sin x1

if x 6= 0
6.4 Example. Show that f (x) = has f.en (0).
0 if x = 0.
Solution. f (x) = 0 + 0x + · · · + 0xn + xn (xsin x1 ) = 0 + xn ε(x) with ε(x) = xsin x1 −→ 0.
x→0

6.5 Property. (Uniqueness Of Finite Expansion)

If f : I 7→ R has f.en (0), then the principle part of the f.e is unique.

Proof. Assume that for all x ∈ I we have

f (x) = p1 (x) + xn ε1 (x) with p1 (x) = a0 + a1 x + · · · + an xn and ε1 (x) −→ 0

x→0
n n
f (x) = p2 (x) + x ε2 (x) with p2 (x) = b0 + b1 x + · · · + bn x and ε2 (x) −→ 0
x→0

Claim p1 = p2 :

f (x) = f (x) ∀x ∈ I =⇒ p1 (x) + xn ε1 (x) = p2 (x) + xn ε2 (x)

=⇒ (a0 − b0 ) + (a1 − b1 )x + · · · + (an − bn )xn = xn (ε1 (x) − ε2 (x)) ( let x → 0)
n−1 n−1
=⇒ a0 = b0 and (a1 − b1 ) + · · · + (an − bn )x =x (ε1 (x) − ε2 (x))

By repeating the same procedure we obtain a0 = b0 , a1 = b1 , · · · , an = bn . Hence p1 = p2 .

6.6 Property. Suppose f : I 7→ R has f.en (0). Then

1. The principle part of the f.e of an even function involves only even powers.

2. The principle part of the f.e of an odd function involves only odd powers.

Proof.

1. Since f has f.en (0), then f (x) = a0 + a1 x + · · · + an xn + xn ε(x) ∀x ∈ I with ε(x) −→ 0. Now,
x→0

f (x) = f (−x) =⇒ f (x) = a0 − a1 x + a2 x2 − a3 x3 + · · · + (−1)n an xn + o(xn )

=⇒ a1 = −a1 , a3 = −a3 , · · · (uniqueness of f.e)
=⇒ ak = 0 for any odd integer k ∈ [1, n]
=⇒ f (x) = a0 + a2 x2 + · · · + (−1)n an xn + o(xn )

2. Similarly as 1.

74
6.7 Property. If f has f.en (0) and m < n, then f has f.em (0).

Proof. Assume f has f.en (0). Then

f (x) = a0 + a1 x + · · · + am xm + am+1 xm+1 + · · · + an xn + xn ε(x) ∀x ∈ I with ε(x) −→ 0. Hence,
x→0
m m+1 n n
f (x) = a0 + a1 x + · · · + am x + am+1 x + · · · + an x + x ε(x)
= a0 + a1 x + · · · + am x + x am+1 x + · · · + an xn−m + xn−m ε(x)
m m 1
 

= a0 + a1 x + · · · + am xm + o(xm ).

If f (x) = 1 + 2x + x4 + 3x7 + o(x7 ), then for example f has f.e5 (0) and f (x) = 1 + 2x + x4 + o(x5 ).
Recall.
Let I be open interval, x0 ∈ I and f : I 7→ R. Then f is differentiable at x0 iff f has f.e1 (x0 ).
Exercise. Let I be open interval, x0 ∈ I and f : I \ {x0 } 7→ R. Assume that
f (x) = a0 + a1 (x − x0 ) + (x − x0 )ε(x) ∀x ∈ I \ {x0 }.
Show that f has an extension by continuity f˜ at x0 and f˜ is differentiable at x0 withf˜0 (x0 ) = a1 .

6.8 Property. (Taylor Formula With Peano’s Remainder)

If f is n-times differentiable at x0 , then f has f.en (x0 ) given by

(x − x0 )2 00 (x − x0 )n (n)
f (x) = f (x0 ) + (x − x0 )f 0 (x0 ) + f (x0 ) + · · · + f (x0 ) + o(x − x0 )n .
2! n!
The case x0 = 0:
If f is n-times differentiable at 0, then f has f.en (0) given by

x2 00 xn (n)
f (x) = f (0) + xf 0 (0) + f (0) + · · · + f (0) + o(xn ).
2! n!

Proof. Let us proof the case x0 = 0. Let

x2 00 xn (n)
p(x) := f (0) + xf 0 (0) + f (0) + · · · + f (0).
2! n!
x2 xn (n)
r(x) := f (x) − p(x) = f (x) − f (0) − xf 0 (0) − f 00 (0) − · · · − f (0).
2! n!
r(x)
ε(x) := n .
x
Note that r is n-times differentiable at 0 and r(0) = r0 (0) = · · · = r(n) (0) = 0. Assume without loss
of generality x > 0, then by repeated application of the Mean Value Theorem we obtain
r(x) = r(x) − r(0) = xr0 (c1 ) (0 < c1 < x)
= x r0 (c1 ) − r0 (0) = xc1 r00 (c2 )


(0 < c2 < c1 < x)
.. ..
. .
= xc1 c2 · · · cn−2 r(n−1) (cn−1 ) (0 < cn−1 < · · · < c2 < c1 < x)

r(x) r(n−1) (cn−1 ) r(n−1) (cn−1 ) r(n−1) (cn−1 )−r(n−1) (0)

Hence |ε(x)| = xn ≤ x ≤ cn−1 = cn−1 −0 −→ r(n) (0) = 0.
x→0

75
Note that;
• The converse of  Taylor Formula With Peano’s Remainder is not necessarily true for n ≥ 2. The
x3 sin x1 if x 6= 0
function f (x) = has f.e2 (0) but it is not twice differentiable at 0.
0 if x = 0.
• If f is n-times differentiable at 0 and f (x) = a0 +a1 x+a2 x2 +· · ·+an xn +o(xn ), then by uniqueness
f 00 (0) f (n) (0)
of f.e we obtain a0 = f (0), a1 = f 0 (0), a2 = 2! , · · · , an = n! .

Finite expansion of elementary functions near x0 = 0:

O0 f (x) = f (0) + xf (0) + x2! f (0) + · · · + xn! f (0) + o(x )

0
2
00
n
(n) n
( provided f (n) (0) ∃)

O1 e = 1 + x + x2! + · · · + xn! + o(x )

x
2 n
n
(f (n) (0) = 1)

O2 cosh x = 1 + x2! + x4! + · · · + (2n)!

2 4
x
+ o(x )
2n
2n
(f (2n) (0) = 1 ∧ f (2n+1) (0) = 0)

O3 sinh x = x + x3! + x5! + · · · + (2nx + 1)! + o(x )

3 5 2n+1
2n+1
(f (2n) (0) = 0 ∧ f (2n+1) (0) = 1)

O4 cos x = 1 − x2! + x4! − x6! + · · · + (−1) (2n)!

2 4 6x
+ o(x ) n
2n
2n π
(f (n) (0) = cos(n ))
2

O5 sin x = x − x3! + x5! + · · · + (−1) (2nx + 1)! + o(x )

3 5
n
2n+1
2n+1 π
(f (n) (0) = sin(n ))
2

O6 ln(1 + x) = x − x2 + x3 − x4 + · · · + (−1) xn! + o(x )

2 3 4 n (−1) n−1 (n − 1)! 

n−1 n
f (n) (0) =
(1 + 0)n

O7 (1 + x) = 1 + αx + α(α − 1) x2! + α(α − 1)(α − 2) x3! + · · · + α(α − 1) · · · (α − n + 1) xn! + o(x ).

α
2 3 n
n

O7 1 +1 x = (1 + x) = 1 − x + x − x + · · · + (−1) x + o(x ).
0 −1 2 3 n n n

O7 1 −1 x = 1 + x + x + x + · · · + x + o(x ).
00 2 3 n n

6.2 Operations On Finite Expansions

6.9 Property. (Linear Combinations And Products)

Let f, g be two functions defined on I and have f.en (0) given by f (x) = p(x) + o(xn ) and
g(x) = q(x) + o(xn ) where p, q are polynomials of degree ≤ n. Then

1. f + g has f.en (0) and f (x) + g(x) = p(x) + q(x) + o(xn ).

2. αf, (α ∈ R), has f.en (0) and αf (x) = αp(x) + o(xn ).

3. f g has f.en (0) and f (x)g(x) = p(x)q(x)discarding powers > n + o(xn ).

 

Proof.

1. f (x) + g(x) = p(x) + q(x) + o(xn ) + o(xn ) = p(x) + q(x) + o(xn ).

2. αf (x) = αp(x) + αo(xn ) = αp(x) + o(xn ).

3. f (x)g(x) =  part of p(x)q(x) with deg ≤ n  + part of p(x)q(x) with deg > n + o(xn )
   

= part of p(x)q(x) with deg ≤ n + o(xn ).

76
6.10 Examples.
1. Find f.e3 (0) of f (x) = sin x + cos x.
h x3 i h x2 i x2 x3
f (x) = x − + 1− + o(x3 ) = 1 + x − − + o(x3 ).
3! 2! 2 6
2. Find f.e3 (0) of f (x) = 3 sin x + 2 cos x.
h x3 i h x2 i x3
f (x) = 3 x − +2 1− + o(x3 ) = 2 + 3x − x2 − + o(x3 ).
3! 2! 2
3. Find f.e3 (0) of f (x) = sin x cos x.
h x3 ih x2 i x3 x3 2
f (x) = x − 1− + o(x3 ) = x − − + o(x3 ) = x − x3 + o(x3 ).
3! 2! 2 6 3
4. Find f.e7 (0) of f (x) = (cos x − 1)(sin x − x).
h x2 x4 ih x3 x5 i x5 x7 x7
f (x) = − + − + + o(x7 ) = − − + o(x7 ).
2! 4! 3! 5! 2!3! 2!5! 3!4!

6.11 Property. (Composition Of Finite Expansions)

If f, u have f.en (0) and u(x) −→ 0 , then f ◦ u(x) = f (u(x)) has f.en (0) given by
x→0

f (u(x)) = p(q(x))discarding powers > n + o(xn )

 

where p and q are the principle parts of the f.en (0) for f and u respectively.

For u(x) −→ 0 and u has f.en (0) given by u(x) = q(x) + o(xn ),we have
x→0
1 1 2 n n
 
1−u(x) = 1−q(x)+o(xn ) = 1 + q(x) + q (x) + · · · + q (x) discarding powers > n + o(x )

6.12 Example.
1
1. Find f.e5 (0) of cos x .

2. Deduce the f.e5 (0) of tan x.

Solution.
1 1
= 1 + q(x) · · · + q 5 (x) discarding powers > n + o(x5 )
 
1. = 2 4 5
cos x 1 − (x /2 − x /24) + o(x )
| {z }
q(x)
= 1 + (x2 /2 − x4 /24) + (x4 /4) + o(x5 ) = 1 + x2 /2 + 5x4 /24 + o(x5 ).
1 h x3 x5 ih i
2. tan x = sin x = x− + 1 + x2 /2 + 5x4 /24 + o(x5 )
cos x 6 120
x3 x5
=x− + + x /2 − x5 /12 + 5x5 /24 + o(x5 )
3
6 120
= x + 1/3x3 + 2/15x5 + o(x5 ).
6.13 Example. Find f.e3 (0) of f (x) = ln(1 + cos x).
Solution.
x2 x2
f (x) = ln(1 + cos x) = ln(1 + 1 − + o(x3 )) = ln(2[1 − + o(x2 )])
2! 4
x2
= ln 2 + ln(1 − + o(x2 )) (ln(1 + u(x)); u(x) −→ 0)
4 x→0
x2
= ln 2 − + o(x3 ).
4

77
 O O
6.14 Property. (Quotient Of Finite Expansion)
N
If 1 N, D have f.en (0) and 2 D(x) −→ ` 6= 0, then D has f.en (0) obtained by performing algorithm
x→0
like the division algorithm for polynomials but following increasing powers of x.

N (x)
Proof. D(x) = 1` N (x) D(x)/`
1
= 1` N (x) 1−u(x)
1
with u(x) = 1 − D(x)/` −→ 0.
x→0

sin x
6.15 Example. Find the f.e5 (0) of tan x = cos x .

• sin x has f.e5 (0) and x3 x5 x2 x4

+ o(x5 ) 1 − + o(x5 )
O-
x− 6 + 120 2 + 24
x3 x5
sin x = x − 6 + 120 + o(x5 ).
x3 x5 x3 2 5
• cos x is continuous at 0 and cos(0) 6= 0. x− 2 + 24 + o(x5 ) x + 3 + 15 x + o(x5 )

• cos x has f.e5 (0) and x3 x5

+ o(x5 )
O- 3 − 30
x2 x4
cos x = 1 − 2 + 24 + o(x5 ).
x3 x5
3 − 6 + o(x5 )
• Then tan x has f.e5 (0) and 2 5
+ o(x5 )
tan x = x + x3
3 + 2 5
15 x + o(x5 ). O- 15 x

2 5
• Since tan is odd function, then 15 x + o(x5 )
x3 2 5
tan x = x + + 15 x + o(x6 ).
3 0 + o(x5 )

x3 2 5
Similarly on can obtain by two different ways tanh x = x − 3 + 15 x + o(x6 ).
Special Cases By Examples: Quotients with D(0) = 0.
cos x − 1
1. Find the f.e5 (0) of .
x2 2
x x4 x6
cos x − 1 − 2! + − + 0x7 1 x2 x4
Solution. = 4! 6!
+ o(x5 ) = − + − + o(x5 ).
x2 x2 2! 4! 6!
sinh x − x
2. Find the f.e3 (0) of f (x) = .
cosh x − 1
2 3
Solution. Since D(0) = 0, D(x) = x2! + · · · = x2 [1/2 + · · · ] and N (x) = x3! + · · · = x2 [x/6 + · · · ],
then we can cancel by x2 . So to obtain f.e3 (0) for f (x) = N (x)
D(x) we need f.e of order 3 + 2 = 5
for both functions N and D.

Or by using division algorithm:

x3 x5
sinh x − x + + o(x5 ) x x3
+ o(x3 ) x2
+ o(x3 )
O-
f (x) = = x3!2 x5!4 3 + 60 1+ 12
cosh x − 1 5
2! + 4! + o(x )
x x3 x x3
x3
x2 x6 + 120 + o(x3 ) + + o(x3 ) − + o(x3 )
 
3 36 3 90
=  1 x2 
x2 2 + 24 + o(x3 ) −x3
+ o(x3 )
=
x
3 + x3
60
x2
+ o(x3 ) O- 90

−x3
1+ + 12 o(x3 ) 90 + o(x3 )
 x x3 x2
+ o(x3 ) 1 − + o(x3 )
 
= + 0 + o(x3 )
3 60 12
x x3 x3
= − + + o(x3 )
3 36 60
x x3
= − + o(x3 ). 78
3 90
Asymptotic Finite Expansions By Example: In the case f (x) −→ ∞, It is possible to find an
x→0
asymptotic expansion of f (x) in increasing negative and positive powers of x:

a−m a−m+1 a−1

f (x) = m
+ m−1 + · · · + + a0 + a1 x + · · · + an xn + o(xn ).
x x x

This form helps to understand better how f tends to infinity.

sinh x
Find the asymptotic expansion to order 2 near 0 of f (x) = .
cosh x − 1
Solution.

sinh x x[1 + · · · ] x f.e3 (0) f.e4 (0)

f (x) = = 2 1 = 2 = .
cosh x − 1 x [2 + · · · ] x f.e3 (0) f.e5 (0)
x3 2 x2
x+ 4
3! + o(x ) x[1 + x3! + o(x3 )] 21+ 6 + o(x3 )
f (x) = x2 x4
= 2 = x2
5
2! + 4! + o(x ) x2 [ 12 + x4! + o(x3 )] x1+
12 + o(x3 )
2 x2 3 x2 3 2 x2
x2  2 1
+ o(x3 ) = + x + o(x2 ).
 
= 1+ + o(x ) 1 − + o(x ) = 1− +
x 6 12 x 12 6 x 6

2 2
f (x) = + o(1) shows that f is asymptotic to the function h(x) = .
x x

6.16 Property. ( Finite Expansion Of Primitives) Z

If f has f.en (0) given by f (x) = a0 + a1 x + a2 x2 + · · · + an xn + o(xn ), then F (x) = f (x) dx the
primitive of f has f.en+1 (0) and
Z x
F (x) = F (0) + (a0 + a1 x + a2 x2 + · · · + an xn ) dx + o(xn+1 )
0
x2 x3 xn+1
= F (0) + a0 x + a1 + a2 + · · · + an + o(xn+1 ).
2 3 n+1

6.17 Corollary.

O1 ln(1 + x) = x − x2 + x3 − x4 + · · · + (−1) xn! + o(x )

2 3 4
n−1
n
n

O2 arctan x = x − x3 + x5 + · · · + (−1) (2nx + 1) + o(x )

3 5
n
2n+1
2n+2

O3 argtanh x = x + x3 + x5 + · · · + (2nx + 1) + o(x )

3 5 2n+1
2n+2

O4 arcsin x = x + 12 x3 + 12 ·· 34 x5 + 21 ·· 43 ·· 65 x7 + · · · + o(x
3 5 7
2n+2
)

O5 argsinh x = x − 12 x3 + 12 ·· 34 x5 − 12 ·· 34 ·· 56 x7 + · · · + o(x
3 5 7
2n+2
)

O6 arccos x = π2 − arcsin x.

Proof. Exercise.

79
6.3 Equivalent And Finite Expansion

6.18 Definition. Let f and g be two functions defined in a neighborhood of x0 ∈ R . We say that f
is equivalent to g in a neighborhood of x0 , and we write f ∼ g, if there exists afunction ε defined
x→x0
in a neighborhood of x0 such that

ˆ f (x) = (ε(x) + 1)g(x) in a neighborhood of x . 0

ˆ ε(x) −−−→ 0.
x→x0

Practical Characterization
f (x)
f ∼ g⇔ −−−→ 1 (provided g(x) 6= 0 in a neighborhood of x0 except possibly at x0 ).
x→x0 g(x) x→x0

Since lim sinx x = 1, then sin x ∼ x.

x→0 x→0

6.19 Property. Let x0 ∈ R.

1. f ∼ g =⇒ f and g have same sign in some neighborhood of x0 .

x→x0

2. lim f (x) = ` (` 6= 0) =⇒ f ∼ `.
x→x0 x→x0

3. f ∼ g and lim g(x) = ` =⇒ lim f (x) = `.

x→x0 x→x0 x→x0

4. f ∼ g ⇐⇒ g ∼ f .
x→x0 x→x0

6.20 Property. Let f, g, f˜ and g̃ be functions defined in a neighborhood of x0 ∈ R .

If f ∼ g and f˜ ∼ g̃, then
x→x0 x→x0

1. f · f˜ ∼ g · g̃.
x→x0

f g
2. ∼ .(provided f˜, g̃ 6= 0 in a neighborhood of x0 except possibly at x0 )
˜
f x→x0 g̃
3. f α ∼ g α . ( provided that: f, g > 0 in a neighborhood of x0 . In particular, no conditions if
x→x0
α ∈ N and f, g 6= 0 in a neighborhood of x0 except possibly at x0 if α ∈ Z \ N.)

Attentions:
• Writing f (x) ∼ 0 means that f (x) = (1 + ε(x))0 = 0 in a neighborhood of x0 which is possible,
x→x0
but in general when you write f (x) ∼ 0 you are making a mistake.
x→x0
• You should never add equivalents without checking. x ∼ x + 1 and −x ∼ −x + 5 but 0  6.
x→∞ x→∞ x→∞
• You should never compose equivalents without checking. x ∼ x + 1 but ex  ex+1 .
x→∞ x→∞
Logarithm: f (x) ∼ g and lim g(x) = ` 6= 1 =⇒ ln(f (x)) ∼ ln(g(x)).
x→x0 x→x0 x→x0
1 + x ∼ 1 + x2 but ln(1 + x)  ln(1 + x2 ).
x→0 x→0

80
6.21 Property. (Finite Expansion And Equivalent)
If in a neighborhood of zero f (x) = ak xk + ak+1 xk+1 + · · · + an xn + o(xn ) with ak 6= 0, then
f (x) ∼ ak xk .(i.e. f is equivalent to the first non zero term in the finite expansion)
x→0

x2 2
For example ex ∼ 1, ex − 1 ∼ x, ex − 1 − x ∼ , sin x ∼ x and cos x − 1 ∼ − x2! .
x→0 x→0 x→0 2! x→0 x→0

6.22 Property. (Finite Expansion And Tangent )

If in a neighborhood of zero f (x) = a0 + a1 x + · · · + an xn + o(xn ), then the equation of the tangent
(T ) to the representative curve (C) of f at x = 0 is given by

YT = a0 + a1 x.

The relative position of the tangent (T ) with respect to (C) near 0 is determined by the first nonzero
term after a1 x in the finite expansion.

Proof. Evident.(YT = f (0) + (x − 0)f 0 (0) = a0 + a1 x)

6.23 Examples.

1. Let f (x) = 1 − x + 3x2 + o(x2 ). Then the equation of the tangent (T ) to the representative
curve (C) of f at x = 0 is given by YT = 1 − x. Since f (x) − YT ∼ 3x2 ≥ 0, then the curve
x→0
(C) is above (T ) near 0.

2. Let f (x) = 1 − x − 3x3 + x4 + o(x4 ). Then the equation of the tangent (T 

) to the representative
>0 if x < 0
curve (C) of f at x = 0 is given by YT = 1−x. Since f (x)−YT ∼ −3x3 ,
x→0 < 0 if x > 0.
then the curve (C) is above (T ) near 0− and then the curve (C) is below (T ) near 0+ .

How to find lim N (x)

D(x) using finite expansion:
x→0
We start by finding an equivalent for D(x). Suppose D(x) ∼ bk xk . Then it is sufficient to find
x→0
f.ek (0) for N (x). Three cases may occur:
First case: N (x) ∼ ap xp with p < k, then
x→0

N (x) ap xp
lim = = ±∞.
x→0 D(x) bk xk

Second case: N (x) ∼ ap xp with p = k, then

x→0

N (x) ap xp ak
lim = = .
x→0 D(x) bk xk bk

Third case: N (x) = o(xk ), then

N (x) xk ε(x)
lim = lim = 0.
x→0 D(x) x→0 bk xk

81