Chapter 2

Real Sequences

L. MOUHADJER. ESSA Tlemcen.

In this chapter, we are going to study the proprieties and convergence of real sequences. We recall that N =
{1, 2, 3, . . . } is the set of positive integers ( natural numbers ) and N0 = {0, 1, 2, 3, . . . } is the set of nonnegative inte-
gers.

Definition 1. A function u : N0 → R that associates a real number u(n) = un for each integer n ∈ N0 is called a real sequence
( or sequence of real numbers ).
Such a real sequence is often denoted as (un )n≥0 or simply as (un ).

Vocabulary
1) The numbers u0 , u1 , ..., un are called the terms of the sequence (un )n≥0 .
2) The number un is referred to as the general term of the sequence (un ).
3) The set Ω ⊂ R defined by
Ω = {u0 , u1 , ..., un , ....} = {uk : k ≥ 0}
is called the range ( or set of terms ) of the sequence (un )n≥0 .
Sometimes, a real sequence (un ) can only be defined starting from an integer n0 ∈ N. In this case, it is denoted by (un )n≥n0
to indicate the initial value of the index n.

p p
Example 1. Let un = n − 25. The sequence (un )n≥25 is defined starting from 25. We have u25 = 0, u26 = 1, u27 = 2 , ...

How to Define a Real Sequence ?

A real sequence (un )n≥n0 can be defined in different ways:

1. Explicitly by giving its general term un = f (n) as a function of n, where f is a real function defined on a set
containing the interval [n0 , +∞[.

2. Implicitly by giving a recurrence relation of the form

§
un0 ∈ I is given
un+1 = g(un ) ∀ n ≥ n0

Example 2. Let (un )n≥1 be the real sequence defined by the following recurrence relation:

u0 = 0

un +1
un+1 = 2 un +3 , ∀n ≥ 0

We have: u1 = 13 , u2 = 4
11 , u2 = 15
34 , ....

2.0.1 Some particular Sequences

1. Constant Sequences, Stationary Sequences

Definition 2. A real sequence (un )n≥n0 is called a constant sequence if un = un+1 for all n ≥ n0 . In this case, we have

un0 = un0 +1 = . . . = un = . . .

1
CHAPTER 2. REAL SEQUENCES 2

Definition 3. A real sequence (un )n≥n0 is called a stationary sequence if there exists a certain index N ≥ n0 such that the
sequence (un )n≥N is constant. In other words, a stationary sequence is a constant sequence starting from a certain index ( a
rank). In particular, every constant sequence is a stationary sequence.
Example 3. Consider the real sequences (un ) and (vn ) defined as follows:
2023
• ˜
un = 1 + sin(n π) and vn = 25 + , ∀n ∈ N
n+1
It is evident that un = 1 for all n ≥ 0. Thus, (un ) is a constant sequence. On the other hand, we have
2023 2023
• ˜
n ≥ 2023 =⇒ 0 < < 1 =⇒ = 0 =⇒ vn = 25
n+1 n+1
Hence, the sequence (vn ) is constant from n ≥ 2023 onwards. Consequently, (vn ) is a stationary sequence.
2. Arithmetic Sequences

Definition 4. A real sequence (un )n≥0 is called an arithmetic sequence if there exists a constant r ∈ R such that

un+1 = un + r , ∀n ≥ 0

The number r is called the common difference of the arithmetic sequence (un ).

Proposition 1. Let (un )n≥0 be an arithmetic sequence with a common difference r. Then,
1. un = u p + (n − p) r for all n, p ∈ N.

2. For all n, p ∈ N such that p < n, we have

n−p+1 n−p+1
S p,n = u p + u p+1 + . . . + un = (u p + un ) = (2u p + (n − p) r)
2 2

3. Geometric Sequences

Definition 5. A real sequence (vn )n≥0 is called a geometric sequence (or geometric progression) if there exists a constant
q ∈ R such that
vn+1 = q vn , ∀n ≥ 0
The number q is called the common ratio of the geometric sequence (vn ).
Proposition 2. Let (vn )n≥0 be a geometric sequence with a common ratio q. Then,
1. vn = vp × q n−p for all n, p ∈ N.

2. For all n, p ∈ N such that p < n, and q 6= 1, we have

 
1 − q n−p+1
Γ p,n = vp + vp+1 + . . . + vn = vp
1−q

2.0.2 Monotonic Sequences

Monotonic sequences with notable properties.
Definition 6. A real sequence (un )n≥0 is said to be:
1. increasing, if ∀n ≥ 0, un+1 ≥ un .

2. decreasing, if ∀n ≥ 0, un+1 ≤ un .

3. monotone, if it is either increasing or decreasing.

If the above inequalities are strict,then we say that the sequence is strictly increasing, strictly decreasing, strictly monotone
respectively.
Examples:
1. Let I = [0, +∞) and f : I → R be a real function defined on I. For n ∈ N0 , let un = f (n). It is clear that if the
function f is increasing (resp. decreasing) on I, the real sequence (un )n≥0 is also increasing (resp. decreasing).

2. Let vn = (−1)n n. The sequence (vn ) is not monotone.

CHAPTER 2. REAL SEQUENCES 3

2.0.3 Bounded Sequences

Definition 7. A real sequence (un )n≥0 is:
1. bounded above if the set of its terms is bounded above. In other words,

(un )n≥0 is bounded above ⇐⇒ ∃ M ∈ R : un ≤ M , ∀n ≥ 0

2. bounded below if the set of its terms is bounded below. In other words,

(un )n≥0 is bounded below ⇐⇒ ∃ m ∈ R : un ≥ m , ∀n ≥ 0

3. bounded if it is both bounded above and bounded below. In this case, the set of terms of the sequence is bounded. Thus,
we have the following equivalence:

(un )n≥0 is bounded ⇐⇒ ∃ (m, M ) ∈ R2 : m ≤ M and m ≤ un ≤ M ∀n ≥ 0

⇐⇒ ∃A ≥ 0 : |un | ≤ A ∀n ≥ 0.

Remarks 1. :
1. A real sequence is unbounded if it is not bounded above or not bounded below.
2. The set of terms of a bounded sequence is contained within a bounded interval.
Examples:
1. Any stationary real sequence is bounded.
(−1)n
2. The sequence (un ) with the general term un = 3 + n+1 is bounded.
3. The sequence (vn ) with the general term vn = n2 is bounded below, but it is not bounded above (thus, unbounded).
4. The sequence (w n ) with the general term w n = (−1)n n is neither bounded above nor bounded below (thus, un-
bounded).

2.0.4 Subsequences
Definition 8. Let (un )n≥0 be a real sequence. A subsequence (or extracted sequence) of the sequence (un )n≥0 is any real
sequence (vn ) defined by
vn = uϕ(n) , n ≥ 0
where ϕ : N0 → N0 is a strictly increasing function.
(−1)n
Example 4. Consider the term un = 2+ 3n as the general term of the sequence (un )n≥1 . An extracted sequence can be of the
form (uϕ(n) )n≥0 with ϕ : N0 → N0 is a strictly increasing function. Depending on the choice of the function ϕ, a particular
extracted sequence is obtained. For instance:
1. For ϕ(k) = 2k, we get the subsequence (u2k )k≥1 with the general term
1
vk = u2k = 2 +
6k
2. For ϕ(k) = 2k + 1, we get the subsequence (u2k+1 )k≥0 with the general term
1
w k = u2k+1 = 2 −
6k + 3

3. For ϕ(k) = 3k, we get the subsequence (u3k )k≥1 with the general term

(−1)k
zk = u3k = 2 +
9k

4. For ϕ(k) = 3k + 1, we get the subsequence (u3k+1 )k≥0 with the general term

(−1)k
u3k+1 = 2 −
9k + 3

Note that we can extract infinitely many subsequences from a sequence (un )n≥1 . Moreover, the indices of terms in any extracted
sequence are chosen from those of the sequence (un )n≥1 based on the choice of the function ϕ.
Remark 1. An extracted sequence from a bounded sequence is also bounded. Therefore, if we extract an unbounded sequence
from a given sequence, then this sequence itself is not bounded.
CHAPTER 2. REAL SEQUENCES 4

2.1 Limits of a Real Sequence, Convergence

2.1.1 Finite Limit of a Real Sequence
Let’s begin with a simple example to give a rigorous definition of the concept of the limit of a real sequence.

Example 5. Consider the real sequence (un )n≥1 defined by its general term

1
un = 1 + , n≥1
n
Notice that as n becomes sufficiently large, the terms of the sequence get closer to 1. Let " > 0 be a number close to 0. Starting
from which index N ∈ N can we ensure that the distance between a term un and the number 1 is less than the tolerance "?
In other words, does there exist an index N ∈ N such that

n ≥ N =⇒ |un − 1| ≤ "

To answer this question, we replace un in terms of n in the last inequality. Indeed, we have

1 1 1
|un − 1| ≤ " ⇐⇒ 1 + − 1 ≤ " ⇐⇒ ≤ " ⇐⇒ n ≥ (2.1)
n n "

For instance, if we take " = 10−9 , then from (2.1) we get

n ≥ 109 =⇒ |un − 1| ≤ 10−9 (2.2)

Thus, the distance between the general term un and the number 1 doesn’t exceed 10−9 once n ≥ N = 109 .
Now, let " = 0.00037, then 1" = 2702.7027. Thus, from (2.1) we have

n ≥ 2702.7027 =⇒ |un − 1| ≤ 0.00037 (2.3)

The relation (2.3) means that |un − 1| ≤ 0.00037 for all integers n ≥ 2702.7027. The smallest integer satisfying (2.3) is
N = 2703 = [2702.27] + 1, which is the required index. We can choose other index like N = 2704, N = 2705, etc., so that
(2.1) is always satisfied for the choice " = 0.00037. In general, for any choice of " > 0, we have from (2.1):

1 1
• ˜
n≥N = +1> =⇒ |un − 1| ≤ "
" "

We observe that for any positive value of ", there always exists an index N = 1" + 1 such that the distance between the
 

general term un and the number 1 is less than " whenever n ≥ N .

In this example, the number 1 is called the limit of the sequence (un ). It is also said that the sequence (un ) converges to 1.

Now, let’s present the rigorous definition of the notion of limit of a real sequence.

Definition 9. A real number ` is called the limit of a real sequence (un ) if:

∀" > 0, ∃N ∈ N such that ∀n ∈ N, n ≥ N =⇒ |un − `| < " (2.4)

Theorem 1. If a real sequence has a finite limit, then that limit is unique.

Proof. Let (un ) be a convergent real sequence. Suppose this sequence has two finite limits ` and `0 . By the definition of
the limit, we have
∀" > 0, ∃N0 ∈ N such that ∀n ∈ N, n ≥ N0 =⇒ |un − `| < "
and
∀" > 0, ∃N1 ∈ N such that ∀n ∈ N, n ≥ N1 =⇒ |un − `0 | < "
So,
∀" > 0, n ≥ max{N0 , N1 } =⇒ |` − `0 | = |` − un + un − `0 | ≤ |un − `0 | + |un − `0 | < " + " = 2".
Hence,
∀" > 0 : |` − `0 | < 2" (2.5)
Expression (2.5) is true for any " > 0. If ` 6= `0 then for " = |` − `0 |/4 > 0, the relation (2.5) yields

|` − `0 | < |` − `0 |/2 =⇒ |` − `0 | < 0 which is a contradiction. Hence ` = `0 .

CHAPTER 2. REAL SEQUENCES 5

Notation: If a real sequence (un ) has a limit `, then according to the previous theorem, that limit is unique. We write
lim un = ` or alternatively, un → ` as n → +∞.
n→+∞
Therefore, we have

( lim un = `) ⇐⇒ (∀" > 0, ∃N ∈ N such that ∀n ∈ N, n ≥ N =⇒ |un − `| < ")

n→+∞

Definition 10. We say that a real sequence (un ) is convergent if there exists a real number ` such that lim un = `. In this
n→+∞
case, we say that the sequence (un ) converges to `. Conversely, if the sequence (un ) doesn’t have any finite limit, then we say
it is divergent.

Remarks 2. 1. The relation (2.4) is interpreted as follows:

For any choice of " > 0 (even very close to zero), there exists a certain index N ∈ N such that all terms un with index
n ≥ N satisfy the estimate |un − `| < ".

2. We know that
|un − `| < " ⇐⇒ ` − " < un < ` + " ⇐⇒ un ∈]` − " , ` + "[
Therefore, lim un = ` means that for any " > 0 (even close to zero), there exists an index N ∈ N such that for all
n→+∞
n ≥ N , we have un ∈ (` − " , ` + ") (which means un lies in the open interval centered at ` with length 2 "), indicating
that the terms of the sequence accumulate around the limit ` beyond a certain index.

Figure 2.1

2n + 1
Example 6. Show that lim = 2.
n→+∞ n+3
2n+1
Let’s define un = n+3 . Take " > 0 and consider the inequality |un − 2| < ".
Hence,
5 5
|un − 2| < " ⇐⇒ < " ⇐⇒ n > − 3
n+3 "
5
So, it’s sufficient to take N = " + 1 , and then we have:

5 5 5
• ˜
∀" > 0, ∃ N = + 1 : ∀n ∈ N , n ≥ N > > − 3 =⇒ |un − 2| < ".
" " "

Therefore, limn→+∞ un = 2.
CHAPTER 2. REAL SEQUENCES 6

2.2 Theorems on Convergent Sequences

2.2.1 Convergence and Subsequences
Proposition 3. Any subsequence of a convergent sequence, converges also to the same limit.

Proof. Let (un ) be a convergent sequence to real number `. By definition of limit we have

∀" > 0, ∃N ∈ N : ∀n ∈ N, n ≥ N =⇒ |un − `| < " (2.6)

Now, let ϕ : N0 → N0 be a strictly increasing function. By induction principal we can prove that for any n ∈ N : ϕ(n) ≥ n
( Show this as an exercise).
So, from (2.6) we get
∀" > 0, ∃N ∈ N : ∀n ∈ N, ϕ(n) ≥ n ≥ N =⇒ |uϕ(n) − `| < "
This means that the subsequence (uϕ(n) ) converges to `.

Corollary 1. If we can extract two subsequences from the sequence (un ) that converges to two different limits, then the
sequence (un ) doesn’t have any limit.

Example 7. The sequence (un ) defined by un = (−1)n for n ≥ 0 has no limit. Indeed, the subsequences (u2k = 1) and
(u2k+1 = −1) converges respectively to 1 and −1.

2.2.2 Convergence and Boundedness

Theorem 2. Every convergent real sequence is bounded.

Proof. Let (un )n≥0 be a convergent real sequence converging to a finite limit `. According to the definition of limit, we
have
∀" > 0 , ∃ N0 ∈ N : ∀n ∈ N , n ≥ N0 =⇒ |un − `| < "
In particular, for " = 1, there exists N1 ∈ N such that

n ≥ N1 =⇒ |un − `| < 1 =⇒ ` − 1 < un < ` + 1

Now, consider the two real numbers m and M defined by

m = min{u0 , u1 , u2 , .... , uN1 −1 , ` − 1} and M = max{u0 , u1 , u2 , .... , uN1 −1 , ` + 1}

Now it’s clear that m ≤ un ≤ M for all n ∈ N. This means that (un ) is bounded.

Remark 2. The converse of Theorem 2 is generally not true. Indeed, the sequence of general term (−1)n is bounded but
doesn’t have any limit.

Using the contrapositive of Theorem 2, we deduce the following corollary:

Corollary 2. Any unbounded sequence is divergent.

2.2.3 Algebraic Properties of Convergent Sequences

Theorem 3. Let (un ) and (vn ) be two sequences converging to two finite limits ` and `0 respectively. Then,

1. lim (un + vn ) = ` + `0
n→+∞

2. lim un vn = ` `0
n→+∞

un `
3. lim = 0 if `0 6= 0.
n→+∞ v
n `
CHAPTER 2. REAL SEQUENCES 7

2.2.4 Convergence and Order Relation

Proposition 4. If lim un = ` then lim |un | = |`|.
n→+∞ n→+∞

Proof. A direct consequence of definition of limit and the triangle inequality | |un | − |`| | ≤ |un − `|.
Proposition 5. Let (un ) be a positive real sequence from a certain index, that is:
∃ N0 ∈ N : un ≥ 0 , ∀n ≥ N0 .
If (un ) converges to a limit `, then ` ≥ 0.
Proof. We reason by contradiction. Suppose that ` < 0. Using the definition of limit, we get for " = −`/2 > 0 that there
exists N1 ∈ N such that for all n ≥ N1 we have |un − `| < −`/2. Now let N = max{N0 , N1 }. So, for all n ≥ N we have
(un ≥ 0 and |un − `| < −`/2) =⇒ 0 ≤ un < `/2 < 0 a contradiction !
Hence ` must be nonnegative real number.
Corollary 3. Let (un ) and (vn ) be two sequences converging to two limits ` and `0 respectively. Then
∃ N0 ∈ N : un ≥ vn , ∀n ≥ N0 =⇒ ` ≥ `0 .

Proof. Let w n = un − vn . So w n ≥ 0 for n ≥ N0 . Now, using theorems 3 and proposition 5 we get

0 ≤ lim = ` − `0 =⇒ ` ≥ `0 .
n→+∞

Theorem 4 ( Squeeze Theorem ). Let (un ), (vn ), and (w n ) be three real sequences satisfying the following condition:
∃N0 ∈ N : vn ≤ un ≤ w n ∀ n ≥ N0 . (2.7)
If both sequences (vn ) and (w n ) converge to the same limit ` ∈ R, then (un ) also converges to `.
Proof. Since both sequences (vn ) and (w n ) converge to the same limit ` ∈ R, we can deduce that
∀ε > 0 , ∃ N1 ∈ N : ∀n ∈ N : n ≥ N1 =⇒ ( |vn − `| < ε et |w n − `| < ε) (2.8)
From condition (2.7) and expression (2.8) we obtain :
n ≥ max{ N0 , N1 } =⇒ −" < vn − ` ≤ un − ` ≤ w n − ` < " =⇒ −" < un − ` < " ⇐⇒ |un − `| < " (2.9)
Therefore
∀ε > 0 , ∃ N2 = max{ N0 , N1 } ∈ N : ∀n ∈ N : n ≥ N2 =⇒ |un − `| < ε
This means that (un ) converges to `.
Example 8. Calculate limn→+∞ 10−n [10n π] , where [x] is the integer part of the real number x.
Answer:
We define un = 10−n [10n π]. For any n ∈ N, we have:
10n π − 1 < [10n π] ≤ 10n π =⇒ π − 10−n < un ≤ π
Since lim (π − 10−π ) = π, we conclude using the squeez theorem that
n→+∞

lim 10−n [10n π] = π.

n→+∞

Note that un is a decimal approximation of the number π!

An important corollary of the squeeze Theorem is the following:
Corollary 4. Let (an ) be a bounded sequence, and let (bn ) be a real sequence that converges to zero. Then the sequence
(an bn ) ( with terms being the product of an and bn ) converges to zero.
Proof. Since (an ) is a bounded sequence, there exists a real number M > 0 such that
∀n : |an | ≤ M
Therefore,
∀n : 0 ≤ |an bn | = |an | |bn | ≤ M |bn | =⇒ M |bn | ≤ an bn ≤ M |bn |
Now, as bn → 0 as n → +∞, it follows, according to the Squeeze Theorem, that an bn → 0 as n → +∞.
Example 9. It is clear that the sequences with general terms (−1)n and cos(n) are bounded. Therefore, if an → 0 as n → +∞,
ln(n)
we can deduce that (−1)n an → 0 and cos(n) an → 0 as n → +∞. In particular, for an = n for example, we find:
(−1)n ln(n) sin(n) ln(n)
lim =0, lim =0
n→+∞ n n→+∞ n
CHAPTER 2. REAL SEQUENCES 8

2.2.5 Monotone Convergence

Theorem 5. Every increasing ( resp : decreasing ) real sequence that is bounded above ( resp : bounded below ) converges
to the supremum ( resp : infimum ) of the set of its terms.
In other words, let (un )n≥0 be a real sequence, and let Ω = {u0 , u1 , . . . } = {un : n ∈ N } be the set of terms of the sequence
(un )n≥0 . Then:

1. If (un )n≥0 is increasing and bounded above, then lim un = sup(Ω).

n→+∞

2. If (un )n≥0 is decreasing and bounded below, then lim un = inf(Ω).

n→+∞

Proof. Let (un )n≥0 be a real sequence. Consider the set Ω ⊂ R as Ω = { un : n ≥ 0 }.

First case: (un )n≥0 is increasing and bounded above.

The set Ω is non-empty and bounded above because the sequence (un ) is bounded above by assumption. Therefore,
the supremum of Ω exists in R. Let’s denote it as `. We will now show that (un ) converges to `. Using the characterization
of the supremum, we have:
∀ε > 0 , ∃n0 ∈ N : ` − ε < un0 ≤ ` (2.10)
Using the increasing property of the sequence (un ), the relation (2.10) becomes:

∀ε > 0 , ∃ n0 ∈ N : ∀n ∈ N : n ≥ n0 =⇒ ` − ε < un0 ≤ un ≤ ` < ` + ε

In other words:
∀ε > 0 , ∃n0 ∈ N : ∀n ∈ N : n ≥ n0 =⇒ |un − `| < ε (2.11)
Therefore, by the definition of a limit, the relation (2.11) means that lim un = ` = sup(Ω).
n→+∞
Second case: (un )n≥0 is decreasing and bounded below.

Let’s define vn = −un ; then the sequence (vn ) is increasing and bounded above. Therefore, according to the result of
the first case we have
lim vn = sup{−un : n ≥ 0} = sup(−Ω) = − inf(Ω)
n→+∞

Therefore
lim un = − lim vn = inf(Ω).
n→+∞ n→+∞

Example 10. Find the supremum and infinimum if they exists of the following subset

(−1)n n
§ ª
E= : n ∈ N0
3n + 2

2.2.6 Adjacent Sequences

Definition 11. Two real sequences are called adjacent if one is increasing and the other is decreasing, and their difference
converges to zero.

Example 11. For n ≥ 1, let’s define the two sequences (un ) and (vn ) as follows:

2n + 1 2n + 3
un = , vn =
n+1 n+1
Verify that the sequences (un ) and (vn ) are adjacent.

Theorem 6. Two adjacent real sequences converges to the same limit.

Proof. Let (un ) and (vn ) be two adjacent sequences. Let’s assume, for example, that (un ) is increasing, and (vn ) is de-
creasing. Define w n = vn − un . Then, for all n ∈ N, we have:

w n+1 − w n = vn+1 − un+1 − vn + un = ( vn+1 − vn ) − (un+1 − un ) ≤ 0 =⇒ (w n ) is decreasing.

Since w n → 0, by the Monotone Convergence Theorem, we deduce that:

0 = lim w n = inf{ w n : n∈N}

n→+∞
CHAPTER 2. REAL SEQUENCES 9

As a result, w n ≥ 0 , ∀n ∈ N, which means un ≤ vn ∀n ∈ N. Consequently, we have:

u0 ≤ u1 ≤ u2 ≤ ... ≤ un ≤ un+1 ≤ vn+1 ≤ vn ≤ ... ≤ v1 ≤ v0

So, the sequence (un ) is increasing and bounded by v0 , and (vn ) is decreasing and bounded by u0 . Therefore, both
sequences are convergent. Let ` = lim un and `0 = lim vn . The algebraic properties of the limit then lead to:
n→+∞ n→+∞

` − `0 = lim un − vn = 0 =⇒ ` = `0
n→+∞

Thus, (un ) and (vn ) converge to the same limit.

Example 12. For n ≥ 1, let’s define the two sequences (un ) and (vn ) as follows:
n
X 1 1
un = , vn = un +
k=0
k! n!

Show that the sequences (un ) and (vn ) are adjacent. Conclude.

2.3 Infinite Limit of a Real Sequence

An important class of divergent sequences is the one that tends to infinity.

Definition 12. Let (un ) be a real sequence.

1. We say lim un = +∞ if
n→+∞
∀ A > 0 , ∃N ∈ N : ∀n ∈ N : n ≥ N =⇒ un ≥ A

2. We say lim un = −∞ if
n→+∞
∀ A > 0 , ∃N ∈ N : ∀n ∈ N : n ≥ N =⇒ un ≤ −A

Example 13. Show that

 n 
1) lim (3n2 ) = +∞ , 2) lim ln = −∞
n→+∞ n→+∞ n2 + 2
Answers:
1) We need to show that:
∀ A ≥ 0, ∃N ∈ N : ∀n ∈ N : n ≥ N =⇒ 3n2 ≥ A
Let A > 0. Then, for n ∈ N, we have: v
tA
2
3n > A ⇐⇒ n >
3
”q —
A
So, let N = 3 + 1. It follows that
v
tA
n ≥ N =⇒ n > ⇐⇒ 3n2 > A.
3
Hence, –v ™
tA
∀ A ≥ 0, ∃N = + 1 ∈ N : ∀n ∈ N : n ≥ N =⇒ 3n2 ≥ A
3

In other words, lim 3n2 = +∞.

n→+∞

2) We need to show that:  n 

∀ A ≥ 0, ∃N ∈ N : ∀n ∈ N : n ≥ N =⇒ ln < −A
n2 + 2
Let A > 0. Then, for n ∈ N, we have:
 n  n
ln < −A ⇐⇒ < e−A
n2 + 2 n2 + 2
n2 + 2 2
⇐⇒ > eA ⇐⇒ n + > eA
n n
2
Since n + n > n, it suffices to consider n > eA. Let N = [eA] + 1. It follows that
CHAPTER 2. REAL SEQUENCES 10

∀ A ≥ 0, ∃N = [eA] + 1 ∈ N : ∀n ∈ N : n ≥ N =⇒ n > eA
2
=⇒ n + > n > eA
nn 
=⇒ ln 2 < −A
n +2
 n 
Hence lim ln = −∞.
n→+∞ n2 + 2
Remark 3. Let Ω be the set of terms of the real sequence (un ). It’s clear that

lim un = +∞ (resp : − ∞) =⇒ Ω is not bounded above (resp : is not bounded below)

n→+∞

Proposition 6. Let (un ) be a monotone real sequence.

1. (un ) is increasing and not bounded above =⇒ lim un = +∞
n→+∞

2. (un ) is decreasing and not bounded below =⇒ lim un = −∞

n→+∞

Proof. Let (un ) be a real, increasing sequence that is not bounded above. We have

((un ) is not bounded above sequence) ⇐⇒ (∀A > 0, ∃N ∈ N : uN > A)

Now, using the increasing property of the sequence (un ), we find:

∀A > 0, ∃N ∈ N : ∀n ∈ N : n ≥ N =⇒ un ≥ uN > A =⇒ un > A

This means that lim un = +∞. The second point can be proven in the same manner.
n→+∞

Proposition 7. Let (un ) and (vn ) be two real sequences. The following results hold:
un
lim un = lim vn = lim (un + vn ) = lim un vn = lim =
n→+∞ n→+∞ n→+∞ n→+∞ n→+∞ v
n
`
` 0
` 6= 0 ` + `0 ` `0 `0
`>0 ±∞ ±∞ ±∞ 0
`<0 ±∞ ±∞ ∓∞ 0
0 ±∞ ±∞ I.F 0
±∞ ±∞ ±∞ ±∞ I.F
+∞ −∞ I.F −∞ I.F
` 6= 0 0 ` 0 ∞
0 0 0 0 I.F

In the table above, I.F denotes indeterminate form. When facing an indeterminate form, the limit is computed using
various simplification and reformulation techniques, along with the use of standard limits as covered in calculating limits
of functions such as:

sin x ex − 1 ln(x + 1)
lim = 1, lim = 1, lim =1
x→0 x x→0 x x→0 x
and
ln x
lim x α e−β x = 0, (α, β > 0), lim = 0, . . . , etc.
x→+∞ x→+∞ x
Now, let’s highlight an important limit:
 a n
lim 1+ = ea , a∈R
n→+∞ n
a n


Indeed, if we define un = 1 + n , then - when a 6= 0(otherwise the result is trivial) - we find:

ln 1 + na
‚
Œ
 a
ln(un ) = n ln 1 + =a a → a as n → +∞
n n

So, un → e a as n → +∞.
In the particular case when a = 1, we find a definition of the transcendental number e, the base of natural logarithms:

1 n
 ‹
e = lim 1 +
n→+∞ n
CHAPTER 2. REAL SEQUENCES 11

‹ 3n
n−1

Example 14. Calculate lim .
n→+∞ n+2
Answer:
For all n ∈ N∗ :
‹ 3n ‹ 3n ‹ 31
n−1 1 − 1/n (1 − 1/n)n
  
= =
n+2 1 + 2/n (1 + 2/n)n
It follows that:
‹ 3n ‹ 13  13
(1 − 1/n)n e−1

n−1
 
lim = lim = = e−1 .
n→+∞ n+2 n→+∞ (1 + 2/n)n e2

2.4 Convergence of Recurrent Sequences

Let I ⊂ R be a non-empty interval and let f : I → R be a real function satisfying the following two conditions:
(1) f is continuous on the interval I
(2) f (I) ⊂ I, meaning that f maps I into itself.
Let (un ) be the numerical sequence defined by the following recurrence relation
§
u0 ∈ I given
(2.12)
un+1 = f (un ) n ≥ 0

2.4.1 Well-defined recurrent sequence

Definition 13. We say that the recurrent sequence (un ) is well-defined if the recurrence relation (2.13) allows us to calculate
all the terms of the sequence.

It’s clear that the condition f (I) ⊂ I ensures that the sequence (un ) is well-defined. Indeed, for each choice of u0 ∈ I,
we have for all n ≥ 0
un ∈ I =⇒ un+1 = f (un ) ∈ I
Therefore, if D f ⊂ R is the domain of the function f , then the real sequence (un ) defined by (2.13) is well-defined if there
exists an interval (or non-empty set) I ⊂ D f such that u0 ∈ I and f (I) ⊂ I.

Example 15. Let (un ) be the real sequence defined by

u0 = 3 , and un+1 = ln(un ) ∀n ≥ 0.

We have u1 = ln 3, u2 = ln(ln 3) = 0.094047..., u3 = −2.363951..., but we cannot calculate uk for k ≥ 4. Indeed, let
I = [a , +∞[ with 0 ≤ a ≤ 3. Let f (x) = ln(x). Then we have 3 ∈ I and f (I) = [ln(a) , +∞[. Since ln(a) ≤ a, it follows
that f (I) 6⊂ I. Thus, the sequence (un ) is "ill-defined"!

Example 16. Let (vn ) be the real sequence defined by

v0 = 1 , vn+1 = 2 + vn ,
p
and ∀n ≥ 0.
p
Let f (x) = x + 2 and x ∈ I = [−2 , +∞[. Since f is continuous and increasing on I, we find that f (I) = [0 , +∞[⊂ I.
Therefore, since v0 ∈ I and f (I) ⊂ I, the sequence (vn ) is well-defined.

2.4.2 Monotonic Variation

Proposition 8. Consider the recurrent sequence (un ) defined by (2.13). Suppose that the function f is increasing on I. Then
(un ) is monotonic. More precisely:

1. If u0 ≤ f (u0 ), the sequence (un ) is increasing.

2. If u0 ≥ f (u0 ), the sequence (un ) is decreasing.

Proof. 1. Suppose that u0 ≤ f (u0 ). Let’s prove by induction that for all n ∈ N : un ≤ un+1 .
Initialization: By hypothesis, we have u0 ≤ u1 = f (u0 ).
Induction: Since the function f is increasing, then

un ≤ un+1 =⇒ un+1 = f (un ) ≤ f (un+1 ) = un+2

Conclusion: By the principle of induction, we deduce that for all n ∈ N: un ≤ un+1 . In other words, (un ) is increasing.
2. The same reasoning can be applied.
Therefore, in all cases (whether u0 ≤ u1 or u0 ≥ u1 ), the sequence (un ) is monotonic.
CHAPTER 2. REAL SEQUENCES 12

2.4.3 Convergence
We admit the following result
Lemma 1. Suppose f : I → R is a real function continuous on I. Let (un ) be a sequence of elements from I. Then
lim un = ` =⇒ lim f (un ) = f (`)
n→+∞ n→+∞

In other words, if the function f is continuous on I, it transforms any sequence (un ) of elements from I that converges to `
into an image sequence (vn = f (un )) that converges to f (`).
Proposition 9. Let (un ) be the recurrent sequence defined by (2.13). Suppose that f : I → I is continuous. If the sequence
(un ) is convergent, then its limit ` satisfies ` = f (`). We call ` a fixed point of f .
Proof. The sequence (un ) is well-defined since f (I) ⊂ I. By Lemma 1, we deduce
` = lim un = lim un+1 = lim f (un ) = f (`)
n→+∞ n→+∞ n→+∞

An immediate consequence of the preceding proposition is the

Corollary 5. If ∀x ∈ D f : f (x) 6= x, then the recurrent sequence (un ) defined by (2.13) is divergent.
Example 17. Let (un ) be the real sequence defined by
u0 = 5

1 (2.13)
un+1 = un + 3 un n≥0
1
Here, we have un+1 = f (un ) with f (x) = x + 3x , x ∈ R∗ . It is clear that the equation f (x) = x has no solution. Therefore,
the sequence (un ) is divergent. Furthermore, since u0 > 0, the reader can easily show, using induction, that un > 0 for all
n ∈ N. Thus,
1
∀ n ∈ N : un+1 − un = > 0 =⇒ (un ) is increasing.
3 un
Finally, since (un ) is increasing and divergent, we conclude that lim un = +∞.
n→+∞

Proposition 10. Let (un ) be the recursive sequence defined by relation (2.13). Suppose that f : I → I is increasing.
1. If there exists b ∈ I such that u0 ≤ b and f (b) ≤ b, then ∀n ∈ N , un ≤ b.
2. If there exists a ∈ I such that u0 ≥ a and f (a) ≥ a, then ∀n ∈ N , un ≥ a.
Proof. It is sufficient to use induction reasoning and the monotonicity of f .

By combining the Monotone Convergence Theorem and Propositions 9 and 10, we obtain the following result:
Proposition 11. Let (un ) be the recursive sequence defined by relation (2.13). Suppose that f : I → I is continuous and
increasing. Then,
1. If there exists b ∈ I such that u0 ≤ f (u0 ) ≤ f (b) ≤ b, then the sequence (un ) is convergent, and its limit ` satisfies
u0 ≤ ` ≤ b.
2. If there exists a ∈ I such that a ≤ f (a) ≤ f (u0 ) ≤ u0 , then the sequence (un ) is convergent, and its limit ` satisfies
a ≤ ` ≤ u0 .
Example 18. Study the convergence of the real sequence (un ) defined by
u0 = 0 p

(2.14)
un+1 = 4 + un n ≥ 0
p
Solution: Let’s define f (x) = x + 4, where x ∈ I = [0, +∞[. It is clear that f is a continuous and increasing function on
I. Furthermore, we have 0 ∈ I, and f (I) = [2, +∞[⊂ I. Therefore, (un ) is well-defined. Additionally, we have u1 = f (u0 ) =
f (0) = 2 ≥ u0 , so (un ) is increasing.
p
Notice that 0 = u0 ≤ f (3) = 7 ≤ 3. Thus, we can conclude that for all n ∈ N, 0 ≤ un ≤ 3, which means that (un ) is
bounded. Consequently, the sequence (un ) is convergent.
Let ` = limn→+∞ un . Then,
` = f (`) =⇒ `2 = 4 + ` =⇒ `2 − ` − 4 = 0
So, ` is the solution of the equation x 2 − x − 4 = 0 that satisfies 0 ≤ ` ≤ 3. A simple calculation shows that
p
1 + 17
`= ≈ 2.56155
2
.