Introduction to CALCULUS Umnov А.Е., Umnov Е.А.

Theme01 2024/25 1

Numerical sequence and its properties

Numerical sequence
We will say that a numerical sequence is given, if a rule is specified
according to which for each natural number 𝑛 unique value 𝑥𝑛 is assigned.
𝑥𝑛 is called the value of the 𝑛-th term of the sequence.
The numerical sequence is usually denoted as {𝑥𝑛 }.
According to this definition, the numerical sequence may be considered
as function of the natural number series. In other words, as a function,
whose domain of definition is the set of all natural numbers N.
Note that the argument of the formula defining the sequence may not
be the number of sequence members.
1
For example, for sequence with 𝑥𝑘 = ∀𝑘 ≥ 2 the number of
𝑘−1
members is not 𝑘, but 𝑛 = 𝑘 − 1.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 2

Numerical sequences can be specified in various ways.

1) By listing the values of its members. For example, the sequence {𝑥𝑛 },
in which all terms with even numbers are equal to one, and all terms
with odd numbers are −1, can be written as {−1, 1, −1, 1, −1, 1, . . .}
2) By functional rules, which for each member of the sequence allows
uniquely determine value by its number. For example, for the
sequence considered in 1) such a rule could be formula
(︃ )︃
𝑛 𝜋
𝑥𝑛 = (−1) or 𝑥𝑛 = sin + 𝜋𝑛 , 𝑛 = 1, 2, . . .
2

3) By recurrent rules, according to which the value of each term

sequences can be uniquely determined by the value of one or more
previous members. For example, for the sequence considered in 1)
such a rule can be the ratios

𝑥𝑛+1 = (−1) · 𝑥𝑛 , 𝑥1 = −1 , 𝑛 = 1, 2, . . .

Another example of recurrently defined sequences are arithmetic and

geometric progressions.
As an exercise, suggest different ways to describe a number sequence
with odd-numbered terms are equaled to one, and with even numbers they
are equaled to zero.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 3

Sometimes members of one numerical sequence are expressed through

members of another. Examples here are
𝑛
∑︀
1) {𝑆𝑛 } — sequence partial sums for sequence {𝑎𝑘 }, where 𝑆𝑛 = 𝑎𝑛 ,
𝑘=1
2) {𝑏𝑛 } — subsequence for sequence {𝑎𝑘 }, where 𝑏𝑛 = {𝑎𝑘𝑛 } . {𝑘𝑛 } is a
sequence of natural numbers, for which 𝑘𝑛 < 𝑘𝑛+1 ∀𝑛 ∈ N.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 4

Classification of numerical sequences

Numerical sequences can be distinguished according to the set of values
of their members. For example,
— sequence, all members of which have the same sign are called of
constant signs.
— sequence, all members of which have a value not exceeding in absolute
value some fixed number is called bounded.
Note that it is convenient to give definition of a bounded sequence,
using logical symbols. For example, a sequence {𝑥𝑛 } is called bounded if

∃𝐶 ≥ 0 : ∀𝑛 : |𝑥𝑛 | ≤ 𝐶 .

In other words, there is a non-negative number 𝐶 such that for any natural
𝑛 inequality |𝑥𝑛 | ≤ 𝐶 is valid.
If the last inequality has the form 𝑥𝑛 ≤ 𝐶 (or 𝑥𝑛 ≥ 𝐶), then they
say about bounded above (or, correspondingly, bounded below) number
sequence.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 5

Let us now formulate the definition of unbounded numerical sequence.

Recall that negation of some definition must be constructed in compliance
with the rules of formal logic.
For example, the formulation «there is no number 𝐶 such that . . . » is
not erroneous, but it is not suitable for definition, because it is impossible
to be sure that this number 𝐶 does not exist (to complete such search is
physically impossible!).
A constructive version of the definition of an unbounded sequence could
be, say, the following: a sequence {𝑥𝑛 } is called unbounded if

∀𝐶 ≥ 0 : ∃𝑁𝐶 : | 𝑥𝑁𝐶 | > 𝐶 .

That is, for every non-negative number 𝐶 there is a number 𝑁𝐶 such that
inequality | 𝑥𝑁𝐶 | > 𝐶 is valid.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 6

Numerical sequences can also be distinguished by their nature changes

in the values of their members when the number of the members changes.
For example,
— sequence, in which the change in its member number is 1 changes
the sign of this term to the opposite one, called alternating,
— sequence, for which the inequality 𝑥𝑛+1 > 𝑥𝑛 holds for any 𝑛 is called
monotonically increasing. If ∀𝑛 inequality 𝑥𝑛+1 < 𝑥𝑛 is valid, then
the sequence is called monotonically decreasing.
Let us explain these definitions with the following examples.
1
Example 1.1. 1) Numerical sequence 𝑥𝑛 = 1 − is bounded since
𝑛
∀𝑛 : 0 ≤ 𝑥𝑛 < 1. In addition, it will be monotonically
increasing due to the inequality
(︃ )︃ (︃ )︃
1 1
𝑥𝑛+1 − 𝑥𝑛 = 1 − − 1− =
𝑛+1 𝑛

1 1 1
=− + = >0 ∀𝑛.
𝑛 + 1 𝑛 𝑛(𝑛 + 1)
𝑛
2) Numerical sequence 𝑥𝑛 = 𝑛(−1) , for which

1 1
𝑥1 = 1, 𝑥2 = 2, 𝑥3 = , 𝑥4 = 4, 𝑥5 = , 𝑥6 = 6, . . . ,
3 5
is bounded from below (by zero), is not bounded from
above and is not neither monotonically increasing nor
monotonically decreasing.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 7

The limit of a numerical sequence and its properties

To describe a numerical sequence it is necessary to specify a rule that
allows to find the values of sequence members by their numbers.
In addition, the numerical sequence may have a quantitative
characteristic not associated with specific members, but with the entire
sequence as a whole.
This characteristic is called the limit of a numerical sequence and
defined as follows.
Definition 1.1. The number 𝐴 is called the limit of a numerical
sequence {𝑥𝑛 }, if for any positive number 𝜀 exists a
number 𝑁𝜀 such that for all members of the sequence
with numbers 𝑛 ≥ 𝑁𝜀 the inequality | 𝑥𝑛 − 𝐴 | < 𝜀
is valid.
The fact that the number 𝐴 is the limit of the number sequence {𝑥𝑛 },
is symbolically written as lim 𝑥𝑛 = 𝐴. Sometimes the definition is also
𝑛→∞
written as {𝑥𝑛 } −→ 𝐴.
𝑛→∞
The number 𝐴 (if it exists!) can be either contained or not to be
contained in the set of sequence members values.
Using quantifiers, the condition lim 𝑥𝑛 = 𝐴 can be written like this:
𝑛→∞

∀𝜀 > 0 ∃𝑁𝜀 : ∀𝑛 ≥ 𝑁𝜀 −→ |𝑥𝑛 − 𝐴| < 𝜀 .

Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 8

Obviously, not every numerical sequence has a limit, which is equaled to

the number 𝐴. Definition of a statement lim 𝑥𝑛 ̸= 𝐴 should be formulated
𝑛→∞
in the form of a testable condition, for example as
Definition 1.2. Number 𝐴 is not a limit numerical sequence {𝑥𝑛 },
if exists a positive number 𝜀0 such that for any
number 𝑁 ∈ N there is a member of this sequence
with number 𝑛0 ≥ 𝑁, for which the inequality
|𝑥𝑛0 − 𝐴| ≥ 𝜀0 is valid.
Or in quantifiers:

∃𝜀0 > 0 ∀𝑁 ∈ N : ∃𝑛0 ≥ 𝑁 −→ |𝑥0 − 𝐴| ≥ 𝜀0 .

Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 9

Sequences whose limit is 0 (that is, for which lim 𝑥𝑛 = 0) are called
𝑛→∞
infinitesimal.
Sequences whose members (starting from a certain number) take values
modulo values greater than any predetermined number are called infinitely
large .
Definition 1.3. We will say, that the sequence {𝑥𝑛 } has as its limit
∞, if
⃒ ⃒
∀𝜀 > 0 ∃𝑁𝜀 ∈ N such that ∀𝑛 ≥ 𝑁𝜀 −→ ⃒𝑥𝑛 ⃒ > 𝜀 .
⃒ ⃒

This fact is denoted as lim 𝑥𝑛 = ∞ .

𝑛→∞
Similarly, we can define the limits of the form lim 𝑥𝑛 = ±∞ . For
𝑛→∞
example, equality lim 𝑥𝑛 = −∞ means that
𝑛→∞

∀𝜀 > 0 ∃𝑁𝜀 ∈ N such that ∀𝑛 ≥ 𝑁𝜀 −→ 𝑥𝑛 < −𝜀 .

In other words, sequences having as their limit ∞, infinitely large. In

this case, one should distinguish between infinitely large sequences from
unbounded sequences, having no limit.
If a numerical sequence has a finite limit, then it is called convergent,
otherwise — divergent.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 10

An example of a divergent sequence would be 𝑥𝑛 = (−1)𝑛 , that is,

{−1, 1, −1, 1, . . .}. An example of an infinitely large sequence is 𝑥𝑛 = 𝑛.
{︃ }︃
1 1 1 1
Example 1.2. Numerical sequence 1, , , , . . . , for which 𝑥𝑛 = ,
2 3 4 𝑛
1
has a limit equal to zero. That is, 𝐴 = 0 or lim = 0.
𝑛
𝑛→∞

Let us prove this using Definition 1.1. Let us note firstly, that this
numerical sequence is monotonically decreasing, since ∀𝑛 the following
1 1
inequality is true: > , that is, 𝑥𝑛 > 𝑥𝑛+1 .
𝑛 𝑛+1
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 11

We have ⃒ ⃒
⃒1 ⃒ 1
| 𝑥𝑛 − 𝐴 | = ⃒ − 0 ⃒ = .
⃒ ⃒
⃒𝑛 ⃒ 𝑛
1
Therefore, for any given positive 𝜀 you can choose a number
[︃ ]︃
1
𝑁𝜀 = + 1, (1.1)
𝜀
⃒ ⃒
1 ⃒ 1 ⃒ 1
for which < 𝜀 and | 𝑥𝑁𝜀 − 0 | = ⃒ − 0⃒ = < 𝜀. Then,
⃒ ⃒
𝑁𝜀 ⃒ 𝑁𝜀 ⃒ 𝑁𝜀
due to the monotonic decrease of the sequence under consideration, for all
numbers 𝑛 ≥ 𝑁 the following inequality will also be true

1
| 𝑥𝑛 − 0 | = <𝜀 ∀𝑛 ≥ 𝑁𝜀 .
𝑛
1 1
So lim = 0 . The number 0 is the limit of the number sequence 𝑥𝑛 = .
𝑛→∞ 𝑛 𝑛
See Fig.1 as an illustration.

[︃ ]︃
1 Here
1 1
denotes the integer part of the fraction . For the integer part of a
𝜀 𝜀
number 𝑥 the notation is also used [𝑥] = floor(𝑥).
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 12

Fig.1. Example of a numerical sequence graph with a zero limit.

Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 13

Example 1.3. Numerical sequence 𝑥𝑛 = (−1)𝑛+1 has no limit.

Let us prove this using definition 1.1. Let’s pretend that 𝐴 is the limit of
this sequence.
For convergence it is necessary so that ∀𝜀 > 0 and ∀𝑛 ∈ N
{︂
| 𝑥𝑛 − 𝐴 | < 𝜀,
| 𝑥𝑛+1 − 𝐴 | < 𝜀.

1
Let us take some even natural number as 𝑛. Let 𝜀 = . Then the
2
necessary condition for convergence takes the form
⎧
1
⎨ |1 − 𝐴|
⎪
< ,
2
⎩ | − 1 − 𝐴| < 1.
⎪
2
or ⎧
1 1
⎨ − < 1−𝐴 < ,
⎪
2 2
⎩ −1 < 1 + 𝐴 < 1 .
⎪
2 2
If we add these inequalities term by term, we get an incorrect consequence
like −1 < 2 < 1. This means that such a number 𝐴 does not exist and the
sequence has no limit.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 14

Definition 1.1 can be interpreted as some kind of «game», in which one

«player» sets arbitrary (as small as desired) positive number 𝜀.
His «enemy» based on the value of this number, must select (or just
guess) the number 𝑁𝜀 , so that |𝑥𝑛 − 𝐴| < 𝜀. (see Fig. 1).
If the «winner» in this game is the second player ∀𝜀 > 0, then the
number 𝐴 is the limit of the number sequence {𝑥𝑛 }.
Note that the number selection rule of 𝑁𝜀 may be different for different
𝜀. This ambiguity is emphasized by the subscript of the searched 𝑁𝜀 . It is
clear that this fact makes it easier for the second player to play for victory.
On the other hand, from the point of view of formal logic, the fact that
«the second player» fails at some 𝜀 > 0 to find a value 𝑁𝜀 does not mean
that 𝐴 is no limit to this sequence.
Finally, as an exercise, answer the following question. Is the dependence
of variable 𝑁𝜀 on the variable 𝜀 functional or not?
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 15

Let us note the main ones that are useful for solving practical problems:
properties of limits of number sequences.
Let everything used in the entries 1∘ –6∘ , sequences are convergent and
𝐶 is some constant, then
1∘ . lim (𝑥𝑛 + 𝑦𝑛 ) = lim 𝑥𝑛 + lim 𝑦𝑛 .
𝑛→∞ 𝑛→∞ 𝑛→∞
2∘ . lim (𝐶 · 𝑥𝑛 ) = 𝐶 · lim 𝑥𝑛 .
𝑛→∞ 𝑛→∞
∘
3 . lim (𝑥𝑛 · 𝑦𝑛 ) = lim 𝑥𝑛 · lim 𝑦𝑛 .
𝑛→∞ 𝑛→∞ 𝑛→∞
4∘ . If, in addition,
𝑥𝑛 lim 𝑥𝑛
𝑦𝑛 ̸= 0 ∀𝑛 and lim 𝑦𝑛 ̸= 0, then lim = 𝑛→∞ .
𝑛→∞ 𝑛→∞ 𝑦𝑛 lim 𝑦𝑛
𝑛→∞
∘
5 . If the sequence increases (decreases) monotonically and is limited
from above (from below), then it has a limit.
6∘ . If lim 𝑥𝑛 = lim 𝑦𝑛 = 𝐴 and ∀𝑛 𝑥𝑛 ≥ 𝑧𝑛 ≥ 𝑦𝑛 , then we have
𝑛→∞ 𝑛→∞
lim 𝑧𝑛 = 𝐴.
𝑛→∞

Fig.2. Theorem 6∘ «about two policemen».

Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 16

Note that the 5∘ property, which is a sufficient condition the existence

of a limit, allows us to draw a conclusion about the convergence of a
sequence, without finding the value of its limit.
An important example. Consider the numerical sequence, whose values
are equal to the perimeter of a regular polygon inscribed in a circle of
radius 𝑅, with an unlimited doubling of the number of its sides.
This sequence has a limit, since it is geometrically obvious that
— it is monotonically increasing (due to the «triangle» rule)
— and bounded above by the perimeter of a square, described
around the same circle.
The limit value of this sequence is taken (by definition!) to be the
circumference, which is denoted as 2𝜋𝑅.
In conclusion, it is also worth noting that student folklore calls property
6∘ by the “two policemen” theorem (see Fig. 2).
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 17

Definition 1.1. is inconvenient for practical use because of you need to

know 𝐴.
This difficulty may be overcome by the concept of fundamental
sequence.
Definition 1.4. We will say that the sequence {𝑥𝑛 } is fundamental if

∀𝜀 > 0 ∃𝑁𝜀 ∈ N such that

⃒ ⃒
∀𝑛 ≥ 𝑁𝜀 and 𝑝 ∈ N −→ ⃒𝑥𝑛+𝑝 − 𝑥𝑛 ⃒ < 𝜀 .
⃒ ⃒

An important statement turns out to be true (Cauchy Criterion):

A sequence converges if and only if it is
fundamental.
The Cauchy criterion is both a necessary and sufficient condition for
convergence. Therefore, his denial also turns out to be useful:

Definition 1.5. The sequence {𝑥𝑛 } is not fundamental if

∃𝜀0 > 0 : ∀𝑁 ∈ N such that

⃒ ⃒
∀𝑛0 ≥ 𝑁𝜀 and 𝑝0 ∈ N −→ ⃒𝑥𝑛0 +𝑝0 − 𝑥𝑛0 ⃒ < 𝜀0 .
⃒ ⃒
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 18

Example 1.4 Using the Cauchy criterion, prove the convergence of the
𝑛 1
∑︀
sequence 𝑥𝑛 = 2
.
𝑘=1 𝑘

Solution: Let us use the estimate that is correct for 𝑚 > 1

1 1 1 1
2
< = − .
𝑚 (𝑚 − 1)𝑚 𝑚 − 1 𝑚

We get
𝑛+𝑝 𝑛+𝑝
⃒ ⃒ ∑︁ 1 ∑︁ 1
⃒ 𝑥𝑛+𝑝 − 𝑥𝑛 ⃒ = < =
⃒ ⃒
𝑘2 𝑘(𝑘 + 1)
𝑘=𝑛+1 𝑘=𝑛+1

1 1 1 1 1 1
= − + − +...+ − =
𝑛+1 𝑛+2 𝑛+2 𝑛+3 𝑛+𝑝 𝑛+𝑝+1
1 1 1 1
= − < < .
𝑛+1 𝑛+𝑝+1 𝑛+1 𝑛
Repeating the reasoning of Example 1.2, we obtain that to
1
satisfy the inequality < 𝜀 ∀𝑛 > 𝑁𝜀 , enough to take
𝑛
[︃ ]︃
1
𝑁𝜀 = + 1.
𝜀

This proves that the sequance is fundamental, and therefore

the convergence of sequence {𝑥𝑛 } .
Note: It will be shown later in the harmonic analysis course that what’s
𝜋2
in this problem lim 𝑥𝑛 = .
𝑛→+∞ 6
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 19

Example 1.5 Prove using the Cauchy criterion that the sequence called
𝑛 1
∑︀
in mathematics harmonic series, of the form 𝑥𝑛 =
𝑘=1 𝑘
diverges.
Solution: Let us use the negation of the Cauchy criterion. We have
an assessment
⃒ ⃒ 𝑛+𝑝
∑︁ 1
⃒ 𝑥𝑛+𝑝 − 𝑥𝑛 ⃒ = =
⃒ ⃒
𝑘
𝑘=𝑛+1

1 1 1 1
= + + ... + + >
𝑛+1 𝑛+2 𝑛+𝑝−1 𝑛+𝑝
1 1 1 1 𝑝
> + + ... + + = .
𝑛+𝑝 𝑛+𝑝 𝑛+𝑝 𝑛+𝑝 𝑛+𝑝

∀𝑁 ∈ N we can take 𝑛0 = 𝑁 ≥ 𝑁 and 𝑝0 = 𝑁, for

𝑝0 1
which = = 𝜀0 .
𝑛0 + 𝑝 0 2
This proves that {𝑥𝑛 } is non-fundamental. Hence, the
sequence {𝑥𝑛 } diverges.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 20

We will say that the sequence {𝑦𝑘 } is a subsequence for {𝑥𝑛 }, if

𝑦𝑘 = 𝑥𝑛𝑘 𝑛𝑘 ∈ N, where {𝑛𝑘 } is strictly mototonic increasing sequence,
consisting of natural numbers.
Let {𝑦𝑘 } is some convergent subsequence of the numerical sequence
{𝑥𝑛 }. Then 𝐴 = lim 𝑦𝑛 is called the partial limit of the sequence {𝑥𝑛 }.
𝑛→∞
In quantifiers this definition can be written as follows:
⃒ ⃒
∀𝜀 > 0 ∃𝑁𝜀 ∈ N such that ∀𝑘 ≥ 𝑁𝜀 −→ ⃒𝑥𝑛𝑘 − 𝐴⃒ < 𝜀 .
⃒ ⃒

The following statements are true for subsequences.

1) If for the sequence {𝑥𝑛 } lim 𝑥𝑛 = 𝐴, then any subsequence
𝑛→∞
of {𝑥𝑛 } has 𝐴 as its limit.
2) (Bolzano-Weierstrass Theorem.) If a sequence {𝑥𝑛 } is
bounded, then it always has a convergent subsequence.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 21

From statement 1) it follows that if the sequence has at least two

different partial limits, then it diverges.
Let us illustrate this fact by examining the convergence sequences with
𝑥𝑛 = sin 𝑛 .
{︃[︃ Consider a sequence
]︃}︃ of segments on the real axis of the form
𝜋 3𝜋
+ 2𝜋𝑘, + 2𝜋𝑘 𝑘 ∈ N . The length of each of them is greater
4 4
than 1, and therefore this segment contains at least one natural number.
In addition, these segments do not have common points.
Let us take one natural number 𝑛𝑘 from each such segment and use
them to construct the subsequence 𝑥𝑛𝑘 . At the same time, it is obvious
1
that √ ≤ 𝑥𝑛𝑘 ≤ 1 . Since this sequence is bounded, it has a partial limit
2
1
𝐴, for which √ ≤ 𝐴 ≤ 1 .
2
{︃[︃ Let us now consider]︃}︃another sequence of segments of the form
5𝜋 7𝜋
+ 2𝜋𝑚, + 2𝜋𝑚 𝑚 ∈ N . Similarly above, we obtain the
4 4
subsequence 𝑥𝑛𝑚 , which has a partial limit 𝐵 and at the same time
1
−1 ≤ 𝐵 ≤ − √ .
2
So the sequence 𝑥𝑛 has at least two different partial limits, and therefore
she disperses.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 22

Using the properties of sequence limits, you can get some useful for
solving problems, ratios.
Example 1.6 Prove that for 𝑎 > 1
√
𝑛
√
𝑛
1) lim 𝑎 = 1, 2) lim 𝑛 = 1,
𝑛→∞ 𝑛→∞

𝑛 ln 𝑛
3) lim = 0, 4) lim = 0.
𝑛→∞ 𝑎𝑛 𝑛→∞ 𝑛
√
Solution. 1) Consider the sequence 𝑥𝑛 = 𝑎 − 1. All its terms are
𝑛

positive and for them 𝑎 = (1 + 𝑥𝑛 )𝑛 ≥ 𝑛𝑥𝑛 is true. Last

the inequality is Bernoulli’s inequality, which follows from
Newton’s binomial formula.
That’s why

𝑎 √
0 < 𝑥𝑛 ≤ −→ lim 𝑥𝑛 = 0 −→ lim 𝑛 𝑎 = lim (1+𝑥𝑛 ) = 1.
𝑛 𝑛→∞ 𝑛→∞ 𝑛→∞

√
2) All members of the sequence 𝑥𝑛 = 𝑛 𝑛−1 are positive and for
𝑛(𝑛 − 1) 2
them 𝑛 = (1 + 𝑥𝑛 )𝑛 ≥ 𝑥𝑛 𝑖𝑠𝑡𝑟𝑢𝑒. The last inequality
2
again follows from Newton’s binomial formula. Therefore, for
𝑛 ≥ 2 we have 4𝑛 ≥ 𝑛2 𝑥2𝑛 . From

2 √
0 ≤ 𝑥𝑛 ≤ √ −→ lim 𝑥𝑛 = 0 −→ lim 𝑛 𝑛 = lim (1+𝑥𝑛 ) = 1.
𝑛 𝑛→∞ 𝑛→∞ 𝑛→∞
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 23

3) Due to 𝑎 > 1 all terms of the sequence

𝑥𝑛 = 𝑎𝑛 = (1 + 𝑎 − 1)𝑛

are positive and for them, as in 2) is true for 𝑛 ≥ 2

𝑛(𝑛 − 1) 𝑛2
𝑥𝑛 ≥ (𝑎 − 1)2 ≥ (𝑎 − 1)2 .
2 4
Therefore we have 4𝑛 ≥ 𝑛2 (𝑎 − 1)2 and then

𝑛 4
0≤ ≤ √ (𝑎 − 1)2 −→ lim 𝑥𝑛 = 0 .
𝑎𝑛 𝑛 𝑛→∞

4) Prove this equality yourself, using limit 3) and the statement

ln 𝑛
that ∀𝜀 > 0 and ∀𝑛 ∈ N inequality < 𝜀 is equivalent
𝜀 𝑛
𝑛
inequality 𝑛 < (𝑒 ) .
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 24

In conclusion, we note that a bounded sequence can have infinitely

many partial limits.
For example, it is known that the set of rational numbers on the
segment [0, 1] is countable, and therefore from them we can form numerical
sequences. On the other hand, on this segment there are an uncountable
set real numbers, in any neighborhood of each of which there are infinitely
many rational numbers.
That is, every real number on a given interval is a partial limit of some
sequences of rational numbers.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 25

Calculating the limits of numerical sequences

Search for the value of the limit of numerical sequences based only on
on its definition, it can turn out to be a rather complex computational
procedure.
In practice it is more convenient to use properties of limits of sequences
in combination with some small set of limits found in advance.
Within the framework of this course, combination of a set of properties
1∘ —6∘ and the following three limits
(︃ )︃ (︃ )︃𝑛
1 1 1
lim =0; lim 𝑛 · sin =1 and lim 1 + =𝑒.
𝑛→∞ 𝑛 𝑛→∞ 𝑛 𝑛→∞ 𝑛

will be quite sufficient.

The validity of the first equality was shown in example 1.1.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 26

Consider the second equality, often called is the first remarkable limit.

(︃ )︃
1
Fig.3. Towards the proof of equality lim 𝑛 · sin = 1.
𝑛→∞ 𝑛
On a trigonometric circle of unit radius we plot the angle the value of
1
which (in radian measure) is equal to 𝛼 = (Fig. 3.3), and construct right
𝑛
triangles OAB and OCD.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 27

Note that the sector OAD, on the one hand, contains the triangle OAB,
and on the other hand, it itself is contained in the triangle OCD. This
means that for the areas of these three figures the following inequalities
are valid:
𝑆△OAB ≤ 𝑆∪ OAD ≤ 𝑆△OCD .
1 1
Since 𝑆△OAB = · |OB| · |AB| , 𝑆△OCD = · |OD| · |CD| , and the area of
2 2
1
the circular sector 𝑆∪ OAD = ·|OD|·𝛼 , then taking into account |OD| = 1
2
we arrive at the inequalities

1 1 1
· sin 𝛼 · cos 𝛼 ≤ ·1·𝛼 ≤ · 1 · tg 𝛼 .
2 2 2
or
1 1 1 1 1 1 1
· sin · cos ≤ ·1· ≤ · 1 · tg .
2 𝑛 𝑛 2 𝑛 2 𝑛
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 28

Transforming, we get

1
1 cos
≥ 𝑛 ≥ 𝑛.
1 1 1
sin · cos sin
𝑛 𝑛 𝑛
Whence it finally follows that

1 1 1
≥ 𝑛 · sin ≥ cos .
1 𝑛 𝑛
cos
𝑛
Now you can use the properties of limits of number sequences. We will
assume that
1 1 1
𝑥𝑛 = ; 𝑧𝑛 = 𝑛 · sin ; 𝑦𝑛 = cos .
1 𝑛 𝑛
cos
𝑛
1
Then, by virtue of the obvious equality lim cos = 1 and the theorems
𝑛→∞ 𝑛
"about two policemen— properties 6∘ , we get that from 𝑥𝑛 ≥ 𝑧𝑛 ≥ 𝑦𝑛
1
should lim 𝑛 · sin = 1 .
𝑛→∞ 𝑛
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 29
(︃ )︃𝑛
1
Importance of calculating the third of the above limits lim 1 +
𝑛→∞ 𝑛
illustrates the problem «about a good bank and a greedy depositor»,
having the following formulation.
Let a certain «good» bank offers its depositors 100% per annum on
time deposits, with uniform accrual of interest on the deposit over time. 2
One of his clients has a sum of money at the beginning of the year
the size of one ruble, which he wants to deposit in the bank before the
beginning of next year.
An obvious calculation shows that the investor will receive an amount
in rubles at the end of the year: your contribution is 1 ruble plus 100%,
that is, another 1 ruble. So,

𝑆1 = 1 + 1 = 2 .

However, this result seems insufficient to the investor and he reasons as

follows: «if I(︃put my
)︃ ruble in the first half of the year, then on June 30
1
I will have 1 + ruble, which I will put aside for the remaining six
2
months.» Then in a year the investor will have
(︃ )︃ (︃ )︃ (︃ )︃2
1 1 1 1 1
𝑆2 = 1+ + · 1+ = 1+ =2 .
2 2 2 2 4

2 In reality, of course, no bank has ever done or does this.

Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 30

Although the effect of this operation is obvious, this is not enough for
the «greedy» investor. His next reasoning is as follows: « if I put (︃
my ruble
)︃
1
for the first four months, then by May 1 I will have in my hands 1 +
3
rub, which I will deposit for the next four months and receive on September
1st (︃ )︃ (︃ )︃ (︃ )︃2
1 1 1 1
1+ + · 1+ = 1+ .
3 3 3 3
Then I invest this amount for the remaining four months and receive in
the end
(︃ )︃2 (︃ )︃2 (︃ )︃3
1 1 1 1 10
𝑆3 = 1 + + · 1+ = 1+ =2 ,
3 3 3 3 27

which is greater than 𝑆2 . »

It is easy to see that if the whole year is divided into 𝑛 equal periods,
then the amount received will be
(︃ )︃𝑛−1 (︃ )︃𝑛−1 (︃ )︃𝑛
1 1 1 1
𝑆𝑛 = 1 + + · 1+ = 1+ .
𝑛 𝑛 𝑛 𝑛
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 31

Let(︃us now)︃examine the properties of the resulting numerical sequence

𝑛
1
𝑆𝑛 = 1 + .
𝑛
First, let us show that ∀𝑛 : 𝑆𝑛+1 > 𝑆𝑛 . Indeed,
(︃ )︃𝑛+1 (︃ )︃𝑛+1
1 1
1+ 1+ (︃ )︃
𝑆𝑛+1 𝑛+1 𝑛+1 1
= (︃ )︃𝑛 = (︃ )︃𝑛+1 · 1 + =
𝑆𝑛 1 1 𝑛
1+ 1+
𝑛 𝑛

⎛ ⎞𝑛+1
𝑛+2 (︃ )︃ [︃ ]︃𝑛+1 (︃ )︃
⎜ 𝑛 + 1 ⎟ 1 𝑛(𝑛 + 2) 1
=⎝ ⎜ ⎟ · 1+ = · 1+ =
𝑛+1 ⎠ 𝑛 (𝑛 + 1)2 𝑛
𝑛
[︃ ]︃𝑛+1 (︃ )︃ [︃ ]︃𝑛+1 (︃ )︃
𝑛2 + 2𝑛 1 𝑛2 + 2𝑛 + 1 − 1 1
= · 1+ = · 1+ =
(𝑛 + 1)2 𝑛 (𝑛 + 1)2 𝑛

and, according to Bernoulli’s inequality: (1 + 𝑥)𝑎 > 1 + 𝑎𝑥, 𝑥 > −1,

[︃ ]︃𝑛+1 (︃ )︃ (︃ )︃ (︃ )︃
1 1 1 1
= 1− · 1+ > 1− · 1+ =
(𝑛 + 1)2 𝑛 𝑛+1 𝑛
(︃ )︃ (︃ )︃
𝑛 1
= · 1+ = 1.
𝑛+1 𝑛
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 32

𝑆𝑛+1
Thus, > 1 and the sequence 𝑆𝑛 – monotonically increasing, that
𝑆𝑛
is, the investor’s «nimbleness» is justified.
Nevertheless, now let’s make sure that the investor will not be able to
get an arbitrarily large profit. Let us perform the following estimate using
Newton’s binomial formula.
(︃ )︃𝑛
1
𝑆𝑛 = 1 + =
𝑛

1 𝑛(𝑛 − 1) 𝑛−2 1 𝑛(𝑛 − 1)(𝑛 − 2) 𝑛−3 1

= 1𝑛 + 𝑛 · 1𝑛−1 · + ·1 · 2+ ·1 · 3 +. . . ≤
𝑛 2! 𝑛 3! 𝑛
1 1 1 1 1 1
≤ 1+1+ + + +... ≤ 1+1+ + 2 ++ 3 +... ≤
1·2 1·2·3 1·2·3·4 2 2 2
and, according to the formula for the sum of all terms of an infinitely
decreasing geometric progression, we get

1
≤1+ = 3.
1
1−
2
This means that the sequence {𝑆𝑛 } is upper bounded and no matter how
much the investor fusses, he will not be able to get even three rubles.
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 33

According to the 5∘ property, the numerical sequence is monotonically

increasing and bounded above has a limit. )︃𝑛 }︃ {𝑆𝑛 } converges.
{︃(︃ This means
1
The limit of the numerical sequence 1+ is an irrational (like
𝑛
√
𝜋 or 2) number, approximately equaled to 𝑒 ≈ 2.718281828459045 . . .
and denoted as 𝑒. In other words,
(︃ )︃𝑛
1
lim 1 + = 𝑒.
𝑛→∞ 𝑛

This equality is usually called the second remarkable limit.

Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 34

Note that the direct application of the 3∘ property for calculating the
first remarkable limit is impossible, since of the two sequences 𝑥𝑛 = 𝑛 and
1
𝑦𝑛 = sin only the second one converges. Its limit is 0, while the first one
𝑛
increases indefinitely.
Such a case is usually called uncertainty of the form «0·∞». Calculating
the limit requires special research here.
0 ∞
Similar problems arise for uncertainties of the type « » , « » ,
0 ∞
«∞ − ∞» , «1∞ » .
The second remarkable limit is an example of the latter type of
uncertainty.
Transformations of the formula notation of a general member of a
numerical sequence in those cases when direct use properties of number
sequences 1∘ —6∘ is impossible, is usually called the method of «uncertainty
disclosure».
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 35

Let’s consider the following problems.

(3𝑛 + 2)2
Example 1.6. Find lim .
𝑛→∞ 5𝑛2 + 3

The formula for the value of a sequence term is a fraction, however,

using the 4∘ property is not possible because

lim (3𝑛 + 2)2 = +∞ and lim 5𝑛2 + 3 = +∞ .

(︀ )︀
𝑛→∞ 𝑛→∞

∞
That is, we have a case of uncertainty of the form « ».
∞
For its «opening» (before moving to the limit!) transform the numerator
using the formula «square of the sum of two numbers», and then divide as
the numerator, so the denominator by 𝑛2 and as a result we get

12 4
(3𝑛 + 2)2 9𝑛2 + 12𝑛 + 4 9+ + 2
lim = lim = lim 𝑛 𝑛 .
𝑛→∞ 5𝑛2 + 3 𝑛→∞ 5𝑛2 + 3 𝑛→∞ 3
5+ 2
𝑛

1 1 1 1
Now, due to lim = 0 and lim 2 = lim · lim = 0 , it is
𝑛→∞ 𝑛 𝑛→∞ 𝑛 𝑛→∞ 𝑛 𝑛→∞ 𝑛

obvious that the limits of the numerator and denominator exist, and we
can use the properties 4∘ , 2∘ и 1∘ .
(︃ )︃
12 4
12 4 lim 9 + + 2 1 1
9+ + 2 𝑛→∞ 𝑛 𝑛 9 + 12 lim + 4 lim 2 9
𝑛 𝑛 = 𝑛→∞ 𝑛 𝑛→∞ 𝑛
lim (︃ )︃ = = .
𝑛→∞ 3 3 1 5
5+ 2 lim 5 + 2 5 + 3 lim 2
𝑛 𝑛→∞ 𝑛 𝑛→∞ 𝑛
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 36
(︀√ )︀
Example 1.7. Find lim 4𝑛2 + 3𝑛 − 2𝑛 .
𝑛→∞

√
Formula for the general term of the sequence 𝑎𝑛 = 4𝑛2 + 3𝑛 − 2𝑛
represents the difference of two expressions. However we cannot use the 1∘
property, because
√︀
lim 4𝑛2 + 3𝑛 = +∞ and lim 2𝑛 = +∞
𝑛→∞ 𝑛→∞

and we are dealing with uncertainty of the form «∞ − ∞». To «reveal» it

let’s
√ multiply and at the same time divide this difference for the amount
of 4𝑛2 + 3 + 2𝑛, followed by using the formula (𝑎 − 𝑏)(𝑎 + 𝑏) = 𝑎2 − 𝑏2 .
√ √
(︁√︀ )︁ ( 4𝑛 2 + 3𝑛 − 2𝑛)( 4𝑛2 + 3𝑛 + 2𝑛)
lim 4𝑛2 + 3𝑛 − 2𝑛 = lim √ =
𝑛→∞ 𝑛→∞ 4𝑛2 + 3𝑛 + 2𝑛

(4𝑛2 + 3𝑛) − 4𝑛2 3𝑛

= lim √ = lim √ .
𝑛→∞ 4𝑛2 + 3𝑛 + 2𝑛 𝑛→∞ 4𝑛2 + 3𝑛 + 2𝑛
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 37

∞
The result is an uncertainty of the form « », which we «reveal» by
∞
dividing the numerator and denominator by 𝑛.

3𝑛 1 3
lim √ = 3 · lim √︂ = ,
𝑛→∞ 4𝑛2 + 3𝑛 + 2𝑛 𝑛→∞ 3 4
4+ +2
𝑛
√︂
3
since lim 4+ = 2.
𝑛→∞ 𝑛 √︂
3
Indeed, on the one hand, 4+ ≥ 2, but, on the other hand
𝑛
√︃ √︃
3 3 9 9
4+ = + 4+2·2·
2
− =
𝑛 4𝑛 16𝑛 16𝑛2
⎯(︃ )︃2 ⎯(︃ )︃2
⎸ ⎸
⎸ 3 9 ⎸ 3 3
=⎷ 2+ − ≤⎷ 2+ =2+ .
4𝑛 16𝑛2 4𝑛 4𝑛

That is, √︃
3 3
2≤ 4+ ≤2+ ,
𝑛 4𝑛
and based
√︂ on the “two policemen” theorem, we come to the conclusion that
3
lim 4+ =2.
𝑛→∞ 𝑛
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 38

(︃ )︃3𝑛
𝑛+2
Example 1.8. Find lim .
𝑛→∞ 𝑛+1

𝑛+2
Here we have lim = 1 and lim 3𝑛 = ∞, that is, an
𝑛→∞ 𝑛 + 1 𝑛→∞
∞
uncertainty of the form «1 ».
To «expand» it, we transform the expression under the limit sign as
follows.
(︃ )︃3𝑛 (︃ )︃3𝑛 ⎡(︃ )︃𝑘−1 ⎤3
𝑛+2 1 1
lim = lim 1 + = lim ⎣ 1 + ⎦ =
𝑛→∞ 𝑛+1 𝑛→∞ 𝑛+1 𝑘→∞ 𝑘

where 𝑘 =𝑛+1 and 𝑛=𝑘−1 =⇒ lim 𝑘 = ∞

𝑛→∞

⎡ (︃ )︃𝑘 ⎤3 (︃ )︃−3
1 1
= ⎣ lim 1+ ⎦ · lim 1+ = 𝑒3 · 1 = 𝑒3 .
𝑘→∞ 𝑘 𝑘→∞ 𝑘
Introduction to CALCULUS Umnov А.Е., Umnov Е.А. Theme01 2024/25 39

Example 1.9. Find the limit of the sequence given in the following
recurrence form
√
𝑥1 = 1, 𝑥𝑛+1 = 𝑥𝑛 + 6 .

Solution. All terms of this sequence are obviously non-negative.

Note also that if 𝑥𝑛 < 3, then 𝑥𝑛+1 < 3. But, since
𝑥1 = 1 < 3, then all members of the sequence do not
exceed 3. That is, the sequence is bounded above.
On the other hand, we have the estimate at 𝑥𝑛 < 3

√ 𝑥𝑛 + 6 − 𝑥2𝑛
𝑥𝑛+1 − 𝑥𝑛 = 𝑥𝑛 + 6 − 𝑥𝑛 = √ > 0.
𝑥𝑛 + 6 + 𝑥𝑛

This means that this sequence increases monotonically.

Then it converges according to the 5∘ property.

Let lim 𝑥𝑛 = 𝐴 > 0 . Then from

𝑛→∞

lim 𝑥𝑛 = lim 𝑥𝑛+1 = 𝐴

𝑛→∞ 𝑛→∞

and the conditions of the problem we have 𝐴2 = 𝐴 + 6.

Whence it follows that 𝐴 = 3.