Section 5.

Accumulation Points

Section 5.4
5.4 BolzanoBolzano-Weierstrass and HeineHeine-Borel Theorems
Purpose of Section: To introduce the concept of an accumulation point of a set,
and state and prove two major theorems of real analysis; the BolzanoWeierstrass Theorem and Heine-Borel Covering Theorem. Both proofs are
two of the most elegant in mathematics.
Accumulation Points
Points
Every set of real numbers has associated with it a set of accumulation
limit points, a concept which allows for a precise analysis of closeness;
closeness of real numbers, closeness of points in
n , closeness of functions,
closeness of operators. The accumulation points may be a subset of a given
set, part of a given set or totally disjoint of the given set. Its defining
characteristic is that every accumulation point of a set is near some point of
the set other than itself.

Definition: A number a is an accumuation point (or limit point)

point of a set A if
and only if for any > 0 the there exists the -neighborhood of a contains
at least one point of A distinct from a . In other words every neighborhood
of a contains points of A different from a .
Keep in mind that an
accumulation point of a set may or may not belong to the set.

A neighborhood of a point (any open interval containing the point) that does
not contain the point is called a deleted neighborhood of the point. Thus, the
set ( 0, 2 ) {1} is a deleted neighborhood of 1.

Margin Note: Intuitively, an accumulation point of a set (which may or may not
belong to the set) is a point where no matter how little you wiggle away
from the point you intersect points of the set. In other words, the set likes to
snuggle up to accumulation points.
Example 1 (Accumulation Points)
a) Every point in the closed interval [ 0,1] is an accumulation

Section 5.4

Accumulation Points

point of the open interval ( 0,1) since every deleted neighborhood of a [ 0,1]
intersects some point in ( 0,1) .
b) Finite sets have no accumulation points since around every real number
(inside or outside the set) you can find a deleted neighborhood that does not
contain elements of the set.
c) The set

( 0,1) {2} has

accumulation points [ 0,1] . The number 2 is not an

accumulation point of the set since there exists a deleted neighborhood around
2 that does not intersect members of the set.
d) The set {1/ n : n = 1, 2,...} has one accumulation point at 0. Around any other
point in the set you can find a deleted neighborhood that doesnt intersect the
set.
.
e) The integers  have no accumulation point even though the set is infinite.
This is easy enough to see since each integer is contained in a deleted
neighborhood of radius 0.25 that does not intersect any members of the set.
Margin Note: The reader may recall limits of sequences from calculus which
are examples of accumulation points of the elements in the sequence.
What do the Real Numbers Really Look Like?
What does the real line look like if you look at it really close up? We think
of it as a continuum of points extending indefinitely in two directions, but what
if you could look at it under a microscope and were able to turn up the
magnification higher and higher. What would you begin to see? You might be
disappointed since you will never get to a stage where you would see, one
rational number, three irrational numbers, one rational number, The real
numbers are self-similar, they look alike no matter what the scale. So how
do we visualize the real numbers in our minds? Well, we simply have to
understand the many properties of the real numbers which can be verified
mathematically, then use your imagination to visualize them it in your minds
eye.

We now come to one of the most important theorems in analysis, the

Bolzano-Weierstrass theorem, but before we state and prove the theorem we
must introduce ourselves to the concept of nested closed intervals.

Section 5.4

Accumulation Points

Nested
Nested Intervals
By a sequence of nested intervals I n = [ an , bn ] we mean a sequence of
closed intervals with the left endpoint an moving towards the right, and the
right endpoint bn moving towards the left. The question we ask is, what can
be said about the intersection of all the sets; i.e. the set of all points common
to every interval? The following lemma, which will be used to prove both the
Bolzano-Weierstrass and Heine-Borel theorems, gives the answer.

Lemma 1 (Nested Interval Lemma)

Lemma)
If

[ a1 , b1 ] [ a2 , b2 ] ... [ an , bn ] 
is a nested sequence of closed intervals whose lengths converge to 0, i.e.
lim ( bn an ) = 0 , then their intersection
consists of a single point
n

[a , b ] = x
n

n =1

BolzanoBolzano-Weierstrass Theorem
Here is an interesting question that will test your intuition about the
real number system and accumulation points. Some people will answer this
question in the affirmative and others in the negative, so the question is not
trivial. Here is the question. Suppose you begin marking off points inside
some bounded interval, open, closed, or neither, lets say [ 0,1] for
convenience, and suppose you do this indefinitely. The question is can you do
it in such a way that there will never be an accumulation point? In other can
you mark off points in such a way that they never bunch up anywhere? Of
course it is possible to mark off points so you do have an accumulation point,
simply pick xn = 1/ n, n = 1, 2,... which has an accumulation point at 0. In fact if
you are clever, you

can pick a sequence

{ xn }n=1

that has 2 accumulation

points, in fact 3, 4, , a finite number of accumulation points. It is also easy

to see that if the interval is unbounded, say [ 0, ) then there need not be an
accumulation point, as the example xn = n , n = 1, 2,... illustrates. So can you

Section 5.4

Accumulation Points

find a sequence { xn }n =1 of numbers in [ 0,1] that does not have an accumulation

point? Think hard.
Historical
Note:
Bernard
Bolzano
(1781-1848)
was
a
Czech
philosopher/mathematician and theologian. He was a Catholic priest who is
best remembered today for his views in the methodology and mathematics and
logic. Many of the ideas later developed by Cantor were understood by
Bolzano. The Bolzano-Weierstrass theorem was first proven by Bolzano but
unfortunately the result was lost. It was re-proven by the great German
mathematician Karl Weierstrass (1815-1897). Weierstrass is often called the
father of modern analysis, having brought mathematical rigor to the level we
see today.

The answer to the question about the existence of an accumulation point of

a bounded infinite set of real numbers is the statement of the BolzanoWeierstrass theorem. The theorem is important and the proof ingenious.
Theorem 1: BolzanoBolzano-Weierstrass Theorem
Every bounded infinite set S of real numbers has an accumulation point
in
.
Proof:
Since S is bounded, there is a closed interval [ a, b ] such that S [ a, b ] ,
where the midpoint x = ( a + b ) / 2 of

[ a, b ]

divides

[ a, b ]

into two closed

subintervals. The subintervals overlap at the midpoint but all that matters is
that their lengths are half the length of [ a, b ] . Now one of these subintervals
(or possibly both) contains an infinite number of points, else the set S is the
union of two finite sets, contrary to the assumption that S is infinite. Letting
I1 be the subinterval that contains an infinite number of points (if both
subintervals contain an infinite number of points we pick one at random), we
continue by dividing I1 into two closed subintervals of equal length, were we
call I 2

an subinterval that has contains an infinite number of points.

See

Section 5.4
Figure 1.
intervals

Accumulation Points

Continuing in this manner, we arrive at a sequence of closed

[ a, b] I1 I 2 I3  I n 
each of whose length is half that of the previous interval. Hence by Lemma 1

the set

consists of a single point, say x0 .

k =1

Strategy for the Bolzano-Weierstrass Theorem; Divide and Conquer

Figure 1
We are not yet done; we must show that x0 is an accumulation point of S . To

( , ) by any neighborhood of x0 ( , ) , and let

k = min ( x0 , x0 ) and note k > 0. Selecting an interval I n whose length
is less than k , we observe I n ( , ) . But I n contains an infinite number of
points and hence so does the arbitrary neighborhood ( , ) . Thus x0 is an
show

this

let

accumulation point of S .

Margin Note: In terms of sequences, the Bolzano-Weierstrass theorem says

that any bounded sequence x1 , x2 ,... has at least one convergent subsequence.
Margin Note: Cantor claimed that the Bolzano-Weierstrass is the basis for
most important results in analysis. Realize the theorem is false if one restricts
oneself to infinite bounded subsets of the rational numbers. The BolzanoWeierstrass theorem states something inherent about the real number system.
Example 2 (Accumulation Points)
a)

1 1 1
2 3 4

The set A = 1, , , , is a bounded infinite set so the Bolzano-

Weierstrass theorem guarantees at least one accumulation point, which in this

Section 5.4

Accumulation Points

case there is exactly one accumulation point, namely 0. The accumulation point
of this set does not belong to the set.
b) set of integers  is an infinite set but is not bounded and so the conditions
of the Bolzano-Weierstrass theorem are not satisfied, hence there is no
guarantee of any accumulation points. In this case the set has no accumulation
points.
c)

The set A = ( 0,1) {2,3, 4,...} is an infinite set but not bounded so the

conditions of the Bolzano-Weierstrass theorem are not satisfied; hence there is

no guarantee the set has any accumulation points. However the set does have
accumulation points, namely all points in the closed interval [ 0,1] .
d) The set A = {1, 2,3, 4,5} is bounded but not infinite and thus the conditions of
the Bolzano-Weierstrass theorem are not satisfies, hence there is no guarantee
the set has any accumulation points. In this case the set does not have any
accumulation points. Finite sets never have accumulation points.
Finite Open Covers
Covers for Sets
There are concepts in mathematics which make you ask what does this
have to do with anything, but after further study you say wow, who ever
thought of this! One such concept is the idea of coverings of sets and in
particular finite open coverings. We will see that sets that have finite open
coverings behave to a great degree like finite sets, and of course anything
finite is a lot simpler than something infinite.

Section 5.4

Definition

Let

Accumulation Points

A be a set of real numbers. A collection C = { J : J C} of

subsets of
is called a cover (or covering

covering)
ring of A if A is a subset of the
union of the members of C , i.e. A { J : J C} . If each element J C in

the covering is an open set , the covering is called an open cover,

cover and if C
contains only a finite number of sets, the covering C is called a finite open
cover.
cover You might think of a covering intuitively as a collection of umbrellas
providing shade from a summer sun as drawn below.

Example 3
a)

The family C = 0,1

, : n  is an open cover for

n

( 0,1) ,

but no

finite sub-collection of these intervals will cover (0,1). Do you agree?

b)

The

closed

bounded

interval

[ 0,1]

has

1
1

C = ,1 + , : n  , an example being the single set

n
n

the member of C when n = 1 .

an

{( 1, 2 )}

c) The real numbers
is covered by the open cover C =

but there is no finite subcover for
.

open

cover
which is

{( n, n, ) : n }

Section 5.4

Accumulation Points

So why is the concept of finite coverings so important in analysis? It has to do

with the fact that sets which can be covered with a finite collection of open
sets behave to an extent like finite sets, which of course are easier to study
than infinite sets.
Compactness and the HeineHeine-Borel Theorem
The concept of open covers of sets, in particular open covers that have
finite subcovers, leads directly to one of the most important concepts in
analysis, compactness.
Defintion: A set A of real numbers is called compact if whenever it is
contained in a union of a family C of open sets, it is also contained in the
union of a finite number of the sets in C .
So now the question remains, how do we know if a set is compact; i.e. every
open cover has a finite subcover? As we have seen in Example 3 some sets
have this desirable property, some do not. Offhand it would seem to be a
very hard property to determine. Fortunately two mathematicians, Eduard
Heine and Emile Borel, found a simple characterization of these finite-like
sets, whereby one can tell at a glance if they have this finite property. This
characterization is called the Heine-Borel (Covering) Theorem.
Theorem 2 (Heine
(HeineHeine-Borel Covering Theorem)
Every closed and bounded interval [ a, b ] is compact.
Proof:
We must prove that for any closed and bounded interval [ a, b ] that is
covered by a collection C = { J : J C} of open intervals, then there is a finite
subcollection of intervals in C , say J1 , J 2 ,..., J n

which also covers [ a, b ] ; that

is for every x [ a, b ] we have x J k for some 1 k n .

The proof is by contradiction. Assume the theorem false; that is, there
exists an open cover of [ a, b ] consisting of intervals C = { J : J C} for which
there is no finite subcover J1 , J 2 ,..., J n .

[ a, b ]

This being true, then the midpoint of

divides the interval into two closed intervals, where at least one

interval, which we call I1 , is not covered by a finite sub-collection of

members of the covering C .

We then divide I 2 in a similar manner and arrive

at a new closed subinterval I 2 , whose length is half that of I 2 and also is not
covered by a finite number of members of the covering C = { J : J C} .
Continuing in this manner, we arrive at a nonincreasing sequence of closed
intervals

Section 5.4

Accumulation Points

[ a, b] I1 I 2  I n 
where each interval I k is half as long as its predecessor and is not covered by
a finite number of members of the covering C = { J : J C} . But from Theorem

1 we know that the intersection

consists of a single point, say x0 . Now

k =1

since x0 [ a, b ] we know there exists (at least) one member of the family C of
open intervals, say

( , )

such that x0 ( , ) .

But from the way the

intervals I k are formed there exists an interval I n whose length is so small

that x0 I n ( , ) C = { J : J C} .

A contradiction; we know I n cannot be covered by a finite number of

members of C , but it is covered by ( , ) , a single element of C .
Figure 1
But this is a contradiction since we have said that I n cannot be covered by a
finite number of coverings of C = { J : J C} , but I n ( , ) C = { J : J C} .
Hence, we conclude every open cover of an closed and bounded interval [ a, b ]
does have a finite subcover.

Margin Note:
The observation that subsets of real numbers have finite
covers is equivalent to being closed and bounded was first observed by
German mathematician Heinrich Eduard Heine in the 1870s and later in 1894
formulated precisely by French mathematician Emile Borel.

Section 5.4

10

Accumulation Points

Note on the HeineHeine-Borel Theorem and Compact

Compact Sets
The Henie-Borel theorem as stated in Theorem 2 is a special case of a
more general Heine-Borel theorem1 which states; A subset of
is compact
iff it is closed and bounded. We stated a specialized version of the theorem
for closed intervals [ a, b ] and we only stated the theorem one way;

[ a, b ]

compactness when in fact the theorem goes both ways. The Heine-

Borel is important since it completely characterizes compact sets; i.e. closed

and bounded sets.
Relationship Between the BolzanoBolzano-Weierstrass and Heine Borel Theorems
The Bolzano-Weierstrass and Heine-Borel Theorems are more closely
related than one may think. In fact if the accumulation point(s) of the infinite
set A , which is guaranteed by the Bolzao-Weierstrass theorem, belongs to A ,
then the set is closed and bounded, and thus by Heine-Borel have finite open
covers.
The following theorem, which will left unproved2, makes this
relationship precise.
Theorem 3 If A is a set of real numbers, then the following are equivalent:
i) The accumulation point(s) of A belong to A .
ii) A is closed and bounded.
iii) A is compact.

We stated that closed and bounded intervals have finite open subcovers whereas in fact all closed and
bounded sets have finite subcovers. Also the converse if true; if every open cover of a set of real numbers
has a finite subcover, then the set is closed and bounded.
2
The proof of this theorem can be found in most textbooks on real analysis. A good textbook in this
genera is Real Analysis by Frank Morgan, American Mathematical Society (2005).

11

Section 5.4

Accumulation Points

Problems
1. Find the accumulation points of the following sets (if any). State whether
the conditions of the Bolzano-Weierstrass theorem hold.
a) 
b) 
c)

d) ( 2, 4 ) ( 4,5 )
e)

{( 1)

:n

f)
g)  ( 0,1)

: m, n 
n
2

i) m + : m, n 
n

h)

2.

(Covers of Sets) What does it mean for a family of sets not to be a cover

C for a set A ? What does it mean for a cover C of a set A not to have a
finite sub-cover.

Give examples of each.

3. (Compactness) Use the general Heine-Borel theorem, which states that a

set of real numbers is compact if and only if it is closed and bounded, to
determine which of the following sets are compact.

b)

{1, 2,3, 4,5}

[ 0,1] [ 2,3]

c)

{x : x

d)

[ 0,1)
[ 0,1] {2,3, 4,5}

a)

e)

= 2}

4.
(Closed Sets)
Sets) A set is closed if it contains its accumulation points.
Find the accumulation points of the following sets and verify that those sets
that are closed do contain their accumulation points.
a) 
b) 
c)

d) ( 2, 4 ) ( 4,5 )

12

Section 5.4

e)

{( 1)

Accumulation Points

:n

f)
g)  ( 0,1)

: m, n 
n
2

i) m + : m, n 
n

h)

5. (Open Subcover) Find a finite open subcover of the set [ 0,1] for the cover

1 1 5
C = ,1 , .
j j =1 10 4
6. (Intersections of Closed Intervals) The intersection of a finite number of
closed intervals is one of three types of sets. What are they?
7. (Intersections of Open Intervals) The intersection of a finite number of
open intervals is one of two types of sets. What are they?
8. (Examples) Give examples of the following.
a)
b)
c)
d)
e)
f)

A bounded set with no accumulation points.

An unbounded set with one accumulation point.
A set with two accumulation points.
An unbounded set with an infinite number of accumulation points.
An unbounded with one accumulation point.
An open set with no accumulation points.