Real numbers

1) Motivation.
Rational numbers form a field. Problem: some very well defined numbers, such as are not
rational. They can not be exactly represented as , but surely can be approximated by rational
numbers with arbitrary precision. We can write their decimal expansion with arbitrary precision. So, this
will be a motivation for our definition.
Remark: we could have used expansion in any base, but we are all more familiar with decimals.
2) Definition. is the set of all infinite sequences
such that 9 , facthored by the following equivalence
relation: the sequence 99
The first n digits of the decimal representation of x guarantee that

Definition. is , where the only two equivalent elements are

and
Remark. This definition differs from the book!
Graphic representation on the line also helps.
Let us understand the equivalency.
999
9 9 by the familiar formula for te sum of infinite geometric series.
We are not using an undefined notion here, this is just an explanation.
Rational numbers do not necessarily have representation ending with zeroes.
Examples:

6 66666
9999
4 4 56 7 95 488 688 -- no law whatsoever!
Observation: there is a natural lexicographic order on Note that the order is reversed for the
negative numbers!
Easy to identify integers in Also easy to define multiplication by : it is just a shift by digits to
the right and using the decimal expansion of the integers.

Limits and real numbers Page 1

 Real numbers -- continued

Theorem.
iff for - the decimal expansion is eventually periodic.
Proof.
First, assume x has eventually periodic expansion. Then
has all zero decimal digits after period, so it is an integer. Which means
is a rational number.
On the other hand, let . Two of the numbers are the same , because
contain only q elements. So, say, or
This means that

Now divide by to obtain that for some integers and

If we write the decimal expansion of with exactly digits, possibly putting in front,
we obtain the desired decimal expansion, by the equation above.
Note: we did not use the formula for the sum of the geometric series!
□
Using the order relation, we can define addition, multiplication, division, and check, that is a field.
Long and boring, so we will skip it.
More interesting:
Absolute value:
It can be used to define distance between two numbers:
Archimedian property: let , then .
Proof.
has a nonzero decimal digit, so for some So we can take

Limits and real numbers Page 2

 Limit of a sequence.

1) Definition. Let be a sequence of real numbers. We say that the sequence converges to
,

iff
The same as all but finitely many fall into neighborhood.
Yet another view: Let us consider the residual set
Then
Remark: Real sequences can diverge to iff

2) Examples. How to prove the convergence from the definition? Play the " ": given a
number find No need to find the best N: anything would be sufficient.
○
○ 0. Follows from the Archimedean property.
○
○ does not exist.
3) An important tool:
Squeezed sequence Theorem.
Let and for some whenever
Then
Proof.
Fix Find two numbers and such that

These numbers exist by the definition of the limit.

Take
Then for we have
□
4) Example:

Limits and real numbers Page 3

 Basic operations with limits.

Definition.
• is bounded above if is called an upper bound for
• is bounded below if is called a lower bound for
• is called bounded if it is bounded both above and below.
Easy property. Subset of a set bounded (above, below) is bounded (above, below). Union and
intersection of two bounded(above, below) sets is bounded(above, below). Any finite set is bounded.
Lemma.
Any converging sequence is bounded.
Proof.
A converging sequence is a union of a subset of and a finite set of elements which does
not belong there.
Monotonicity Lemma.
Let Assume that for Then
Proof.
Let Take All but finitely many fall into -neighborhood of so
Similarly, all but finitely many fall into -neighborhood of so So for all but
finitely many , - contradiction.
Theorem.
Let Then
1.
2.
3. if
Proof - see the book.
Examples.
5
1.
5
2.

3.

Limits and real numbers Page 4

 Supremum and infimum

Reminder: upper and lower bounds. Sic: the inequalities are not strict. An upper/lower bound can
belong to a set!
Observation: is an upper bound for a set iff is a lower bound for a set is
bounded above iff is bounded below.
Observation: upper bounds form a ray: if , and is an upper bound for , then is also an upper
bound for
Definition.
Let be a bounded above set. is called the least upper bound or supremum of is an upper
bound for and for any other upper bound of we have
Notation:
Equivalently: iff is an upper bound for and any is not an upper bound for .
Equivalently: iff is an upper bound for and .

Definition.
Let be a bounded above set. is called the greatest lower bound or infimum of is a lower bound
for and for any other lower bound of we have
Notation:
Observation:
Theorem (Least upper bound principle).
Every nonempty subset of that is bounded above has a supremum. Similarly,
every nonempty subset of that is bounded below has an infimum.
Proof.
As in the book, we give a proof for the infimum. The existence of the supremum follows from the
observation above.
First let us observe that a limit of a sequence of lower bounds for is again a lower bound -- we just
need to check it for every
Next, we can always find an integer lower bound for , let us call it Then is a lower bound for the
set (We work with the set since it consists of nonnegative numbers - is its lower
bound!).
Let be some element of with decimal expansion . Notice that is not a lower
bound for (Why do we have to add
There are only finitely many integers between and +1. Pick the largest of these that is still a lower
bound for , and call it . Notice that is again not a lower bound for

Next pick the greatest integer such that is a lower bound for
. Since works and does not, belongs to 9 . Notice that
is not a lower bound for
Same way we construct the number L= such that for each a lower
bound for but is not.
Since is a lower bound for On the other hand, if then
for some so is not lower bound for
Thus

Limits and real numbers Page 5

 Monotone sequences

Definition.
A sequence of real numbers is called (strictly) increasing if
A sequence of real numbers is called (strictly) decreasing if
A sequence which is either (strictly) increasing or (strictly) decreasing is called (strictly) monotone.

Theorem. (Monotone convergence theorem).

Every bounded above increasing sequence converges to its supremum.
Every bounded bellow decreasing sequence converges to its infimum.
Proof.
Enough to prove for increasing sequences. Let Then
.
Since is increasing,
.
Remark.
The unbounded increasing (decreasing) sequences diverge to The proof is the same as for
Monotone Convergence Theorem.
Examples.
1. Bounded decreasing sequence, hence converges to
2. This sequence is unbounded!
3. Important!: is a sequence of nonnegative numbers, Then either have a
limit, or diverges to
Lemma (Nested Intervals Lemma).
Suppose that are nonempty closed intervals
such that for each . Then the intersection is nonempty.
Proof.
is an increasing sequence, and is decreasing. Let be the limit of , and be the limit of
Then , and
Remark.
It is important to consider closed intervals!
Corollary.
If and , then consists of exactly one point.
Proof.
Assume that Then so

Limits and real numbers Page 6

 Cauchy sequences

Definition.
A sequence of real numbers is called a Cauchy sequence if
if then
Lemma.
Every converging sequence is a Cauchy sequence.
Proof.
Fix and let . Then one can find such that for
Then when
Turns out that for the opposite is also true. Not so for

Completeness Theorem.
Every Cauchy sequence of real numbers have a limit.
Proof.
Let be a Cauchy sequence.
First, take such that if then Since

is a bounded sequence.

Let
Note that so both sequences are bounded.
Use the Monotone sequence Theorem to define
Remark. The construction works for any bounded sequence.
Notation.

Moreover,
So for and So
Thus So, Thus

Note that and

By Squeezed Sequences Theorem,
Example (important).
Majorated convergence of series: .
Theorem.
If and converges, so does
Proof.
Follows from 2.8.C

Limits and real numbers Page 7

 Series

Lemma.
If | the series converges, and

Proof.

Lemma.
If converges, then
Proof.

Observation (from the definition of the limit).

exists if and only if
Restatement of Cauchy Theorem for series -- Cauchy Criterion.
exists if and only if
Theorem.
iff
Special case: when the harmonic series diverges.
Proof.
Let By comparison test, enough to prove that harmonic series diverges.
Note that for harmonic series

This implies convergence by Cauchy criterion.

On the other hand, if

Since , we get that

So is bounded. It means that is also bounded.
Root test for convergence.
Let Define
If , the series diverges.
If , the series converges.
Remark.
If yet for the series converges, and for the series
diverges.
Remark - a property of .
is finite and is infinite.
Proof of the root test.
Let . Pick so that . Then for all but finitely many , so
converges by comparison test.
Let now Pick so that Then ,
so

Limits and real numbers Page 8

 so

Ratio test for convergence.

Let Define L:=
If , the series diverges.
If , the series converges.
Proof.
Homework
Theorem (Leibniz Alternating Series).
Let be a monotone decreasing sequence converging to zero. Then
converges.
Proof.
Note that and
Thus the sequence is decreasing, and the sequence is increasing. Moreover,
. Since
Both sequences are bounded (by each other), so they both have limits, which are the
same by the previous observation.

Limits and real numbers Page 9

 Limit points and subsequences

Definition.
Let be a real sequence. is called a limit point of if
is infinite.
Remark
Compare with the limit, where the set should contain all but finitely many points.
Examples.
1. has two limit points: {-1,1}
2. Any converging sequence has only one limit point, its limit.
3. (viewed as a sequence) has no limit points
4. The sequence has only one limit point: 1.

Definition.
Let be an increasing sequence of natural numbers. is called a subsequence of sequence
Easy to see by induction:
Theorem.
is a limit point of a sequence iff for some subsequence of
Proof.
If then , which is infinite by the
definition of the limit.
On the other hand, if is a limit point of , we can construct a subsequence recursively, by
selecting , with
Theorem. (Bolzano-Weierstrass)
Every bounded sequence has a converging subsequence.
Equivalently: every bounded sequence has a limit point.
Proof.
Let us choose intervals recursively, so that
1. (can be done because is bounded)
2. is either left or right half of , so that is infinite . This is done
by induction.
Then = /2, so
Moreover, the family is nested. Thus
.
Then
So
So is a limit point of

Limits and real numbers Page 10