You are on page 1of 61

# MY REAL ANALYSIS NOTE

Miliyon T.,

January 7, 2017
eal analysis is a branch of mathematical analysis dealing with the real numbers
and real-valued functions of a real variable. In particular, it deals with the
analytic properties of real functions and sequences, including convergence and
limits of sequences of real numbers, the calculus of the real numbers, and continuity,
smoothness and related properties of real-valued functions.

## Real Analysis is a very straightforward subject, in that it is simply a nearly linear

development of mathematical ideas we have come across throughout our story of
mathematics. However, instead of relying on sometimes uncertain intuition (which
we have all felt when we were solving a problem we did not understand), we will
anchor it to a rigorous set of mathematical theorems.
* This

## note covers the material in Math 661 & Math 666.

c Miliyon Tilahun

Typeset in LATEX

Contents
1 Finite, infinite and
1.1 Definition . .
1.2 Facts . . . . .
1.3 Exercise . . .

countable
. . . . . .
. . . . . .
. . . . . .

2 Metric spaces
2.1 Definition . . . . . . . .
2.2 Examples . . . . . . . .
2.3 More Typical Examples
2.4 Exercise . . . . . . . . .
3 Open and closed
3.1 Definition .
3.2 Facts . . . .
3.3 Exercise . .

sets
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

4
4
4
6

9
. 9
. 9
. 12
. 13

sets
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

16
16
16
19

4 Compact sets
21
4.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2 Facts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.3 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
5 Sequences
26
5.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.2 Facts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.3 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
6 Continuity
30
6.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
6.2 Facts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
6.3 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
7 Sequence and Series of
7.1 Definition . . . . .
7.2 Facts . . . . . . . .
7.3 Exercise . . . . . .

Functions
34
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

8 Riemann-Stieltjes Integral
8.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.2 Facts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.3 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

39
39
40
41

9 Measure Theory
9.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2 Facts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.3 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

43
43
44
47

## 10 Measure and Integration

49
10.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
10.2 Facts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
10.3 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

www.albohessab.weebly.com

(DRAFT-VERSION)

page 2 of 61

## 11 General Measure Space

11.1 Definition . . . . .
11.2 The Lp Space . . .
11.3 Facts . . . . . . . .
11.4 Exercise . . . . . .

www.albohessab.weebly.com

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

(DRAFT-VERSION)

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

56
56
56
57
58

page 3 of 61

## 1 Finite, infinite and countable sets

The usefulness of a pot comes from its emptiness.

1.1 Definition
Definition 1.1 (Set). A collection of objects called members or elements of the set.
Definition 1.2 (Relation). A well defined collection of objects called members or elements of
the set.
Definition 1.3 (Function). Given two sets A and B, by a mapping or function from A into B
we mean a correspondence that assigns to each member of A a member of B.
Definition 1.4 (Equivalence relation). A relation R on a set A is an equivalence relation if
it satisfies
(i) Reflexive: (a, a) R, a A.
(ii) Symmetric: (a, b) R (b, a) R, a, b A.
(iii) Transitive: (a, b), (b, c) R (a, c) R, a, b, c A.
Definition 1.5 (Finite).

1.2 Facts
Theorem 1.6. Every infinite subset of a countable set A is countable.
Proof. Exercise
Theorem 1.7. The union of a countable set is countable.
Proof. Exercise

www.albohessab.weebly.com

(DRAFT-VERSION)

page 4 of 61

## Theorem 1.8. The set of rational numbers are countable.

Proof. Write the rational number as m/n, where m and n are integers that are relatively prime.
Let m = pe11 pe22 pekk and n = qlf1 q2f2 qtft be the prime-number decompositions of m and
n. Then the desired counting function for the rational numbers is f (1) = 1 and
 
m
2f1 1
2ek
2e2
1
q22f2 1 qt2ft 1
f
= p2e
1 p2 pk ql
n
This function is uniquely invertible. For example, the rational number
on the list; and the eighteenth number on the list is 32 .

2
3

## This is the best way of exhausting the rational numbers.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 5 of 61

1.3 Exercise
Exercise 1.9. R the set of real numbers is uncountable.
Exercise 1.10. The set of all polynomial
p(x) = a0 + a1 x + a2 x2 + + an xn
with ai Z, i {1, 2, . . . , n} is countable.
Exercise 1.11. The set of algebraic number is denumerable.
Exercise 1.12. S be a set of all sequences of 0s and 1s i.e. S = {(a1 , a2 , a3 , . . .)|an = 0 or 1}.
Show that S is uncountable.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 6 of 61

Solution
Solution (1.9). 1 First, we show the subset (0, 1) of the real numbers is uncountable, then it
follows R is uncountable.
The set of real numbers in (0, 1) can be written as decimals of the form
0.
0.
0.
0.
..
.

d11
d21
d31
d41
..
.

d12
d22
d32
d42
..
.

d13
d23
d33
d43
..
.

d14
d24
d34
d44
..
.

...
...
...
...
..
.

where dij s are all digit from the set {0, 1, 2, . . . , 9}.
Assume, there exist a one-to-one correspondence between

.. ..
..
..
..
..
N
. .
.
.
.
.

..
..
..
..
.. ..
. .
.
.
.
.

## the decimals in (0, 1) and N.

...

...

...
(0, 1)

...

..
.

We will now construct a decimal in (0, 1), which is missed by this correspondence.
Construct x = 0.c1 c2 c3 . . .
where c1 6= d11 , 0, 9.
c2 6= d22 , 0, 9.
c3 =
6 d33 , 0, 9.
..
.
We forbid 0 and 9 so that x 6= 0.000 . . . = 0, x 6= 0.999 . . . = 1, and to have a unique
representation.
By construction this number x is in (0, 1) and differ from the nth decimal number in the sequence
in the nth decimal digit. So, x is missed by the supposed correspondence. Hence, our assumption
that there exist a correspondence between (0, 1) and N is incorrect. Thus, the set (0, 1) is
uncountable.
Therefore, R is uncountable.
Solution (1.10). For each pair of integers (n, m) N N, let Pnm denote the set of polynomials
of degree m in which
|a0 | + |a1 | + + |am | = n
Pnm is finite for each n, m N. Hence
[
P = {Pnm : (n, m) N N}
is countable since it is a countable union of countable sets.
Solution (1.11). Let Pn [x] be the set of polynomial functions with rational coefficients. Then
there is a bijection from Pn [x] into Qn . Since Q is countable, there exists a bijection f : Q N.
1

## Cantor diagonal proof

www.albohessab.weebly.com

(DRAFT-VERSION)

page 7 of 61

## Hence the map F : Qn Nn defined by F (r1 , . . . , rn ) = (f (r1 ), . . . , f (rn )) is a bijection. Since

Nn is countable, we conclude that Qn is countable and consequently Pn [x] is countable. The set
[
Rn =
{x R; P (x) = 0}
P Pn [x]

is countable since it is a countable union of finite sets. Note that Rn is infinite since Q Rn ,
for any n 1. Since the set of algebraic numbers A(R) is given by
[
A(R) =
Rn ,
n1

## then it is countable being a countable union of countable sets.

Solution (1.12). (Proof by contradiction)2 Clearly S is infinite so we want to show that it is not
countable.
Suppose not, i.e. Assume S is countable. Then there exists a bijective map f : N S. Let us
describe the bijection in this was

(a
,
a
,
a
,
a
,
.
.
.)

11
12
13
14

(a
,
a
,
a
,
a
,
.
.
.)

21
22
23
24

## (a31 , a32 , a33 , a34 , . . .)

.
.
.
N
S
.. ..
..

n
(an1 , an2 , an3 , an4 , . . .)

.
.
..
. .

. .
.
Now define a new sequence (b1 , b2 , b3 , . . .) where
(
0
if aii = 1
bi :=
1
if aii = 0
By construction (bn ) S and differ from the nth sequence on the list in the ith term. Thus (bn )
is missed by f . This contradict the fact that f : N S is a bijective(in particular, surjective).
So we see that no such f exist( we can always construct a sequence in S which will be missed
by the map). Hence S is uncountable.

## Cantor diagonal proof

www.albohessab.weebly.com

(DRAFT-VERSION)

page 8 of 61

2 Metric spaces
When you realize there is nothing lacking, the whole world belongs to you.

2.1 Definition
Definition 2.1. A metric on a non-empty set X is a function d : X X R satisfying
1. d(x, y) 0 and d(x, y) = 0 iff x = y.
2. d(x, y) = d(y, x),

(Positive definite)

x, y X.

## 3. d(x, z) d(x, y) + d(y, z),

(Symmetry)

x, y, z X.

(Triangle inequality)

The couple (X, d) is called a metric space3 . The above three properties are called metric
axioms.
Remark 2.2. A metric is a generalization of distance in the real line.
Definition 2.3 (Metric Subspaces). Suppose (X, d1 ) and (Y, d2 ) are metric spaces. We say
that X is a metric subspace of Y and that Y is a metric superspace of X iff, X is a subset of Y
and d1 is a restriction of d2 .
Definition 2.4 (Isometry). Suppose (X, d1 ) and (Y, d2 ) are metric spaces and : X Y .
Then is called an isometry or an isometric map iff, d2 ((a), (b)) = d1 (a, b) for all a, b X.
Definition 2.5 (Convergence). Let (X, d) be a metric space. A sequence (xn ) X converges
to an element x X if for all > 0 there exists an N N such that d(xn , x) < whenever
n N.
Definition 2.6 (Cauchy sequence). A sequence in a metric space (X, d) is a Cauchy sequence
if for all > 0 there exists an N N such that d(xn , xm ) < whenever n, m N .
Definition 2.7 (Complete). A metric space (X, d) is called complete if every Cauchy sequence in X converges to an element of X.
Definition 2.8 (Dense). A subset A of X is dense in X if the closure of A is X itself.
Definition 2.9 (Separable). A space X is separable if it admits countable dense subset.

2.2 Examples
Example 2.10 (The usual metric). Let X = R, d(x, y) = |x y| for all x, y in R. Show that
d is a metric on R.
Solution. To show d is a metric on R, it suffices to show d satisfies the following three properties:
(i) Positive definite:
d(x, y) = |x y| 0 for all x, y R.
d(x, y) = 0 |x y| = 0 y x = 0 y = x.
3

A generalization of a metric space is a pseudo metric space(semi-metric space) in which the distance
between two distinct points can be zero. The definition of a pseudo metric space is similar to metric space
except the first axiom is replaced by the weaker axiom: d(x, x) = 0. The set of all Lebesgue measurable
function on [0, 1] with
Z 1
d(f, g) =
|(f g)(x)|dx
0

## is one example of pseudo metric space.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 9 of 61

(ii) Symmetry:
d(x, y) = |x y| = |y x| = d(y, x)
(iii) Triangle inequality: For all x, y and z in R we have,
d(x, z) = |x z| = |(x y) + (y z)| |x y| + |y z| = d(x, y) + d(y, z).
So d is metric on R.
Example 2.11 (Euclidean metric). Let X = Rn := {x = (x1 , . . . , xn ) : xi R}, and
v
u n
uX
d(x, y) = t (xi yi )2
i=1

## for all x = (x1 , . . . , xn ), y = (y1 , . . . , yn ) in Rn . Show that d is a metric on Rn .

Solution. To show d is a metric on Rn . We need to show that d satisfies the following three
properties:
(i) Positive definite:
v
u n
uX
d(x, y) = t (xi yi )2 0 for all x, y Rn
i=1

v
u n
uX
d(x, y) = 0 t (xi yi )2 = 0 xi = yi for i = 1, . . . , n x = y.
i=1

(ii) Symmetry:
v
v
u n
u n
uX
uX
2
t
d(x, y) =
(xi yi ) = t (yi xi )2 = d(y, x) for all x, y Rn .
i=1

i=1

2

## (d(x, y) + d(y, z)) =

n
X
i=1
n
X

(xi yi ) +
(xi yi )2 +

i=1

n
X
i=1
n
X

(yi zi ) + 2
(yi zi )2 + 2

i=1

X
n

X
 1
n
2
2
(xi yi )
(yi zi )

i=1
n
X

i=1

(xi yi )(yi zi )

i=1

P
P
P
(using Cauchy s inequality ( a2i )( b2i ) ( ai bi )2 )
=

n
X

## ((xi yi ) + (yi zi ))2

i=1

n
X

(xi zi )2

i=1

= (d(x, z))2
Using (i),(ii) and (iii), we conclude that d is indeed a metric on Rn .

www.albohessab.weebly.com

(DRAFT-VERSION)

page 10 of 61

## Example 2.12 (Discrete metric). Let X = be any set,

(
1
if x 6= y
d(x, y) =
0
if x = y
Show that d is a metric on X.
Solution. To show d is a metric on X. We want to show that d satisfies the following three
properties:
(i) Positive definite: By definition d(x, y) 0 and d(x, y) = 0 x = y.
(ii) Symmetry:
(
1
d(x, y) =
0

if x 6= y
if x = y

(
1
=
0

if y 6= x
if y = x

= d(y, x)

## (iii) Triangle inequality: For all x, y and z in X,

If x = y or y = z then d(x, z) = d(x, y) + d(y, z).
Suppose, then , that x 6= y and y 6= z, then we have
d(x, y) + d(y, z) = 2 > d(x, z).
Using (i),(ii) and (iii), we conclude that d is a metric on X.
Example 2.13 (Uniform metric). Let A be a set, X the set of real-valued bounded functions
of A,
d(f, g) = sup |g(x) f (x)|
xA

## for all f, g X. Show that d is a metric on X.

Solution. To show d is a metric on X. We want to show that d satisfies the following three
properties:
(i) Positive definite: d(f, g) is the supremum of a set of non-negative real numbers. So d(f, g) 0
for all f, g in X.
d(f, g) = 0 |g(x) f (x)|xA = 0 g(x) = f (x) for all x A g = f.
(ii) Symmetry:
d(f, g) = sup |g(x) f (x)| = sup |f (x) g(x)| = d(g, f ) for all f, g X
xA

xA

## (iii) Triangle inequality: Let f, g, h be any three elements in X,

For each element x1 of A we have
|h(x1 ) f (x1 )| |h(x1 ) g(x1 )| + |g(x1 ) f (x1 )|
sup |h(x) g(x)| + sup |g(x) f (x)|
xA

xA

So
sup |h(x) f (x)| sup |h(x) g(x)| + sup |g(x) f (x)|
xA

xA

xA

## That is d(f, h) d(f, g) + d(g, h).

Using (i),(ii) and (iii), we conclude that d1 is indeed a metric on X.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 11 of 61

## 2.3 More Typical Examples

Example 2.14 (Unitary metric4 ). Let X = Cn := {z = (z1 , . . . , zn ) : zi C} and
v
u n
uX
|i i |2
d(x, y) = t
i=1

where x = (1 , . . . , n ) Cn , y = (1 , . . . , n ) Cn .
Example 2.15 ([`p , dp ).
Example 2.16 (` , d ). ` = {x = (i )| supi |i | < } with
d (x, y) = sup |i i |
i

## Example 2.17 (C[a, b]).

Example 2.18 (Sequence space). S consists of a set of all sequences of complex numbers
with the metric

X
1 |i i |
d(x, y) =
2i 1 + |i i |
i=1

where x = (i ) S, y = (i ) S.
Solution. To show triangular inequality consider the function
f (t) =
f 0 (t) =

t
,
1+t

1
> 0,
(1 + t)2

tR
f is increasing.

## Since we know |a + b| |a| + |b| we have

f (|a + b|) f (|a| + |b|)
Thus,
|a + b|
|a| + |b|

1 + |a + b|
1 + |a| + |b|
|a|
|b|

+
1 + |a| 1 + |b|

(1)
(2)

## Choose a = i i , b = i i and substitute in (2)

|i i |
|i i |
|i i |

+
1 + |i i |
1 + |i i | 1 + |i i |
Multiplying by 1/2i and applying the sum results the triangle inequality.

Unitary metric is similar to Euclidean metric except the points are taken from Cn . That is why the
modulus is used so that the quantity inside the square root sign becomes a nonnegative real number.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 12 of 61

2.4 Exercise
Exercise 2.19. Show that in the definition of a metric, metric axiom 3 can be replaced by the
following weaker axiom: If a, b, c X are distinct then d(a, c) d(a, b) + d(b, c).
Exercise 2.20. [Manhattan metric or the taxi cab metric] Let X = R2 , d(x, y) =
|y1 x1 | + |y2 x2 | for all x = (x1 , x2 ), y = (y1 , y2 ) X. Show that d is a metric on R2 .
Exercise 2.21. Let X = R2 , d(x, y) = max{|y1 x1 |, |y2 x2 |} for all x = (x1 , x2 ), y = (y1 , y2 )
in X. Show that d is a metric on R2 .
Exercise 2.22. Let (X, d) be a metric space. Show that
d1 (x, y) := ln(1 + d(x, y))
for all x, y X defines a new metric on X.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 13 of 61

Solution
Solution (2.19). Suppose a = b. Then
d(a, c) = d(b, c) d(a, b) + d(b, c)
If b = c, the argument is similar. Lastly, suppose a = c; then
d(a, c) = 0 d(a, b) + d(b, c)
Thus the triangle inequality follows from axiom 1 if the points a, b and c are not all distinct.
Solution (2.20). To show d is a metric on R2 . We want to show that d satisfies the following
three properties:
(i) Positive definite:
d(x, y) = |y1 x1 | + |y2 x2 | 0 for all x, y R2 .
d(x, y) = 0 |y1 x1 | + |y2 x2 | = 0 |y1 x1 | = |y2 x2 | = 0 y = x
(ii) Symmetry:
d(x, y) = |y1 x1 | + |y2 x2 | = |x1 y1 | + |x2 y2 | = d(y, x).
(iii) Triangle inequality: For all x, y and z in R2 we have
d(x, y) + d(y, z) = |y1 x1 | + |y2 x2 | + |z1 y1 | + |z2 y2 |
|(y1 x1 ) + (z1 y1 )| + |(y2 x2 ) + (z2 y2 )|
= |z1 x1 | + |z2 x2 |
= d(x, z)
Using (i),(ii) and (iii), we conclude that d is a metric on R2 .
Solution (2.21). To show d is a metric on R2 . We want to show that d satisfies the following
three properties:
(i) Positive definite:
d(x, y) = max{|y1 x1 |, |y2 x2 |} 0 for all x, y R2 .
d(x, y) = max{|y1 x1 |, |y2 x2 |} = 0 |y1 x1 | = |y2 x2 | = 0 y = x.
(ii) Symmetry:
d(x, y) = max{|y1 x1 |, |y2 x2 |} = max{|x1 y1 |, |x2 y2 |} = d(y, x).
(iii) Triangle inequality: For all x, y and z in R2 we have,
d(x, y) + d(y, z) = max{|y1 x1 |, |y2 x2 |} + max{|z1 y1 |, |z2 y2 |}
So we have
d(x, y) + d(y, z) |y1 x1 | + |z1 y1 | |z1 x1 |
and similarly
d(x, y) + d(y, z) |z2 x2 |
Thus d(x, y) + d(y, z) max{|z1 x1 |, |z2 x2 |} = d(x, z)

www.albohessab.weebly.com

(DRAFT-VERSION)

page 14 of 61

## Using (i),(ii) and (iii), we conclude that d is a metric on R2 .

Solution (2.22). Given d is a metric on X. We want to show d1 (x, y) := ln(1 + d(x, y)) is a
metric on X.
(i) Positive definite:
d1 (x, y) = 0 ln(1 + d(x, y)) = 0
d(x, y) = 0
x = y,

d is a metric.

(ii) Symmetry:
d1 (x, y) = ln(1 +d(x, y)) = ln(1+ d(y, x)) = d1 (y, x)
{z
}
|
d(x,y)=d(y,x)

(iii) Triangle inequality: For all x, y and z in X, we observe the following inequality
1 + d(x, y) 1 + d(x, z) + d(z, y)

(3)

## The inequality (3) holds, because d is a metric.

Now, we have
1 + d(x, y) 1 + d(x, z) + d(z, y) + d(x, z)d(z, y)
|
{z
}
0

## = (1 + d(x, z))(1 + d(z, y))

Thus,
1 + d(x, y) (1 + d(x, z))(1 + d(z, y))
ln(1 + d(x, y)) ln[(1 + d(x, z))(1 + d(z, y))]

## = ln(1 + d(x, z)) + ln(1 + d(z, y))

Hence,
ln(1 + d(x, y)) ln(1 + d(x, z)) + ln(1 + d(z, y))
|
{z
} |
{z
} |
{z
}
:=d1 (x,y)

:=d1 (x,z)

:=d1 (z,y)

Therefore,
d1 (x, y) d1 (x, z) + d1 (z, y)
Using (i),(ii) and (iii), we conclude that d1 is indeed a metric on X.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 15 of 61

## 3 Open and closed sets

God created the open sets, everything else is the work of topologists.

3.1 Definition
Definition 3.1 (Ball). A ball(open) with radius r and centered at a is
Br (a) = {x|d(x, a) < r}

(4)

## Definition 3.2 (Neighborhood). A neighborhood Nr (x) is a ball centered at x.

Definition 3.3 (Limit point). The point p is a limit point if every neighborhood of p contains
at least one point other than p.
Notation 3.4. We denote the set of all limit points of E by E 0 .
Definition 3.5 (Closed). E is closed if every limit point of E is a point in E.
Definition 3.6 (Interior). The point p is an interior point of E if there is a neighborhood N
of p such that N E.
Notation 3.7. We denote the set of all interior points of E by E .
Definition 3.8 (Open). The set E is open if every point of E is an interior of E.
Definition 3.9 (Closure). The closure of the set E is E = E E 0 .
Definition 3.10 (Perfect). The set E is perfect if it is closed and every point in E is a limit
point of E.
Definition 3.11 (Separated). Two subsets A and B of a metric space X are said to be
and A B are empty, i.e. , if no point of A lies in the closure of B and
separated if both A B
no point of B lies in the closure of A.
Definition 3.12 (Connected). A set E X is said to be connected if E is not a union of two
nonempty separated sets.
Definition 3.13 (Base). A collection {V } of open subsets of X is said to be a base for X iff
for every x X and every open set G X, there exists such that x V G.

3.2 Facts
Theorem 3.14 (De Morgan). Let {G } be (a finite or infinite) collection of sets G . Then
\ c [ 
G =
Gc
(5)

T
S
c and B =
c
Proof. Put A = ( G
)

(G ).
T
Let T
x A,then x
/ G . This implies x
/ G for any . Thus, x Gc for every. Hence
c
x (G ). Therefore, A B.
Conversely, let x SB, then x Gc . This implies x
/ G for any . Thus, x
/
every . Hence x ( G )c . Consequently, B A.
Therefore, A = B and the theorem follows.

for

## Theorem 3.15 (Characterization). The set E is open iff E c is closed.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 16 of 61

## Proof. () Suppose E is open.

Let x be a limit point of E c . Then every neighborhood N of x contains a point of E c . Then
x is not an interior point of E (x N ( E). Since E is open, this implies x E c . Hence, E c
contains all of its limit points(since x was arbitrary). Therefore E c is closed.
() Suppose E c is closed.
Let x E. This implies x
/ E c and x is not a limit point of E c (E c is is closed). Then there
exists a neighborhood N of x such that E c N = . Hence N E. Therefore, E is open(by
definition).
S
Theorem 3.16. For any collection of open set G , G is open.
S
Proof. Put G = G . If x G, then x G for some . Since x is an interior point of G , x
is also an interior point of G. Hence G is open.
T
Theorem 3.17. For any collection of closed set F , F is closed.
Proof. By De Morgans rule we have
\

and Fc is open, so
is closed.

c
(F )

c
F

(Fc )

## is open by theorem (3.16). Hence (

F )

is open. Therefore,

T
Theorem 3.18. For any finite collection G1 , . . . , Gn of open sets G , ni=1 Gi is open.
T
Proof. Put H = ni=1 Gi . Let x H, then x Gi for some i but Gi is open implies there is a
neighborhood Ni of x with radii ri such that ri Gi for i = 1, . . . , n. Put
r = min{r1 , . . . , rn },
and let N be a neighborhood of x with radius r. Then N Gi for i = 1, . . . , n, so that N H.
Hence H is open.
Example 3.19 (Counter-example). An example to show the infinite intersection of open set
may not be open.


\
1 1
,
= {0}
n n

(6)

n=1

is closed.
Theorem 3.20. For any finite collection F1 , . . . , Fn of closed sets G ,

Sn

i=1 Fi

is closed.

## Proof. Taking the compliment of (3.18).

Example 3.21 (Counter-example). An example to show the infinite union of closed set may
not be closed.


[
1
1
,3
= (0, 3)
(7)
n
n
n=1

is open.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 17 of 61

## Theorem 3.22 (Nested Interval Property). Let I1 = [a1 , b1 ], I2 = [a2 , b2 ], . . . be a sequence

of nested closed(bounded) intervals, i.e. I1 I2 . . .. Then there exists at least one common
point to every interval, i.e.

\
Ii 6=
i=1

## Proof. Now I1 I2 . . . implies that a1 a2 . . . and . . . b2 b1 .

Claim: am an for every m, n N.
for m > n implies am < bm bn and m n implies am an < bn . Thus each bn is an upper
bound for the set A = {a1 , a2 , . . .} of left end points.
By Least upper bound axiom of R, sup(A) exists; say p = sup(A). Now, p bn , for each n N,
since each bn is an upper bound for A and p is the least upper bound. Furthermore, an p for
every n N, since p is an upper bound for A = {a1 , a2 , . . .}. But
an p bn p In = [an , bn ]
Hence p is common to every interval.
Theorem 3.23 (Bolzano-Weierstrass). Let A be a bounded infinite set of a real numbers.
Then A contains at least one limit point.
Proof. Since A is bounded, A is a subset of a closed interval I1 = [a1 , b1 ]. Bisect I1 at 12 (a1 + b1 ).
Note that both of the closed intervals of I1




1
1
(a1 + b1 ), b2
a1 , (a1 + b1 ) and
(8)
2
2
cant contain a finite number of points of A since a is infinite. Let I2 = [a2 , b2 ] be one of the two
intervals in (8) which contains an infinite number of points of A.
Now bisect I2 . As before one of the two closed intervals




1
1
(a2 + d2 ), b2
a2 , (a2 + b2 ) and
2
2
must contain an infinite number of points of A. Call that interval I3 .
Continuing this procedure and obtain a sequence of nested intervals
I1 I2 I3 . . .
such that each interval In contains an infinite points of A and lim |In | = 0 where |In | denotes
the length of the interval In .
By the Nested Interval Property (3.22) of the real numbers there exists a point p in each interval
In . We show that this p is a limit point of A and the theorem follow.
Let Sp = (a, b) be an open interval containing p. Since lim |In | = 0
n0 N

such that

## |In0 | < min(p a, b p)

Then the interval In0 is a subset of the open interval Sp = (a, b) as indicated in the diagram
below

Figure 2: Bolzano

Since In0 contains an infinite number of points of A, so does the open interval Sp . Thus each
open interval containing p contains points of A other than p, i.e. p is a limit point of A.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 18 of 61

3.3 Exercise
Let (X, d) be a metric space and let A be a subset of X.
Exercise 3.24. Show that A is open iff the compliment Ac is closed.
Exercise 3.25. Suppose X has finitely many points. Prove that for every x X, the singleton
set {x} is open.
Exercise 3.26.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 19 of 61

Solution
Solution (3.24). () Suppose A is open.
Let x be a limit point of Ac . Then every neighborhood N of x contains a point of Ac . Then
x is not an interior point of A (x N ( A). Since A is open, this implies x Ac . Hence, Ac
contains all of its limit points(since x was arbitrary). Therefore Ac is closed.
() Suppose Ac is closed.
Let x A. This implies x
/ Ac and x is not a limit point of Ac (Ac is is closed). Then there
exists a neighborhood N of x such that Ac N = . Hence N A. Therefore, A is open(by
definition).
Solution (3.25). For arbitrary x X define r(x) = min{d(x, y) : y X, y =
6 x}. Since all
d(x, y) > 0 when x 6= y, r(x) is a minimum of finitely many positive numbers(since there
are finitely many points). So r(x) > 0. Consider the open ball B(x, r(x)). Now, suppose
y B(x, r(x)) and y =
6 x. Then by definition of being in the ball d(x, y) < r(x) but r(x) d(x, y)
by definition of r(x). This is a contradiction, so B(x, r(x)) = {x}. Therefore, as x was arbitrary
the set {x} is open for any x X.
Solution (3.26). Exercise.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 20 of 61

4 Compact sets
4.1 Definition
Definition 4.1 (Cover). An open
S cover of a set E in a metric space X is a collection {G } of
open subset of X such that E G .
Definition 4.2 (Compact). A subset K of a metric space X is said to be compact if every
open cover of K contains a finite subcover.
Definition 4.3 (Sequentially Compact). A subset of a metric space is called (sequentially)
compact if every sequence in the subset has a convergent subsequence.
Definition 4.4 (Least Upper Bound Axiom). If A is a set of real numbers bounded from
above, then A has a least upper bound, i.e. sup(A) exists.
Example 4.5. Show that the set E = {0, 1, 21 , 13 , 14 , . . .} is compact.
Solution. We need to show every open cover of E has a finite subcover.
Let E U , where U is open set. Then 0 must be in some U0 . Since U0 open there exists
> 0 such that (, ) U0 . In particular, n1 in U0 if n >S1 . Let N be the largest integer in
n
1
1
i=1 Ui .
, and let i , i = 1, . . . , N be such that i Ui . Then E

4.2 Facts
Theorem 4.6. In a metric space, any Cauchy sequence having a convergent subsequence is
itself convergent, with the same limit.
Proof. Suppose {xn } is a Cauchy sequence in a metric space (X, d). Let {xnk } be a convergent
subsequence of {xn } with limit x. Then for any > 0, there exists N1 > 0 such that

## k > N1 d(xnk , x) < .

2
Since {xn } is a Cauchy sequence, there also exists N2 > 0 such that

## m, n > N2 d(xn , xm ) < .

2
Let N = max{N1 , N2 } and choose k > N . Then nk k > N and so
n > N d(xn , x) d(xn , xnk ) + d(xnk , x) <

+ = .
2 2

## Thus {xn } converges to x.

Theorem 4.7. Compact subsets of a metric spaces are closed.
Proof. Suppose S is a compact subset of a metric space X. Let {xn } be a sequence in S that
converges to a point x X. This implies, in particular, that {xn } is a Cauchy sequence. Since
S is compact, {xn } has a subsequence which converges to a point y S. Then, by Theorem
(4.6), we know that {xn } is convergent in S and so x = y S (by uniqueness of limits). Hence S
is closed.
Theorem 4.8. Closed subsets of a compact sets are compact.
Proof. Exercise

www.albohessab.weebly.com

(DRAFT-VERSION)

page 21 of 61

## Example 4.9 (Counter-example). Let A1 , A2 , A3 , . . . be the following sequence of open-closed

intervals:
A1 = (0, 1], A2 = (0, 1/2], . . . , Ak = (0, 1/k], . . .
Now the sequence of intervals is nested, i.e. each interval contains the succeeding interval;
A1 A2 . But the intersection of the intervals is empty, i.e.,
A1 A2 Ak =
Thus there exists no point common to every interval.
Example 4.10 (Counter-example). Let A1 , A2 , A3 , . . . be the following sequence of closed
infinite intervals:
A1 = [1, ), A2 = [2, ), . . . , Ak = [k, ), . . .
Now A1 A2 , i.e. the sequence of intervals is nested. But there exists no point common
to every interval, i.e.
A1 A2 Ak =
Theorem 4.11 (Heine-Borel). Let I1 = [c1 , d1 ] be a closed(bounded) subset of R which is
covered by G = {(a1 , bi ) : i I} of open intervals. Then G contains a finite subclass which also
covers I1 .
Proof. Assume that no finite subclass of G covers I1 . We bisect I1 = [c1 , d1 ] at 12 (c1 + d1 ) and
consider the two closed intervals




1
1
(c1 + d1 ), d1
(9)
c1 , (c1 + d1 ) and
2
2
At least one of this two intervals can not be covered by a finite subclass of G or else the whole
interval I1 will be covered by a finite subclass of G. Let I2 = [c2 , d2 ] be one of the two intervals
in (9) which can not be covered by a finite subclass of G, we now bisect I2 . As before one of the
two intervals




1
1
c2 , (c2 + d2 ) and
(c2 + d2 ), d2
2
2
can not be covered by a finite subclass of G. Call that interval I3 .
We continue this procedure and obtain a sequence of nested intervals I1 I2 I3 . . . such
that each interval In can not be covered by a finite subclass of G and lim |In | = 0 where |In |
denotes the length of the interval In .
By the nested interval property of the real numbers (3.22) there exists a point p in each interval
In . In particular , p I1 , since G is a cover of I1 , there exists an open interval (ai0 , bi0 ) in G
which contains p. Hence ai0 < p < bi0 . Since lim |In | = 0
n0 N

such that

## |In0 | < min(p ai0 , bi0 p)

Then, as indicated in the diagram below, the interval In0 is a subset of the interval (ai0 , bi0 ) in
G.

Figure 3: Borel

But this contradicts our choice of In0 . Thus the original assumption that no finite subclass of
G covers I1 is false and the theorem is true.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 22 of 61

Remark 4.12. Any closed and bounded set is not necessarily compact.
Let Q, the set of rational numbers, as a metric space, with d(p, q) = |p q|. Let E be the set of
all p Q such that 2 < p2 < 3. E is close and bounded in Q but E is not compact.
But if A Rk is closed and bounded, then A is compact(Heine-Borel theorem).
Example 4.13 (Counter-example). Consider the open and bounded interval A = (0, 1) R1 .
Observe that the class




1
1
:nN
(10)
G = Gn =
,
n+2 n
of open intervals covers A. i.e.

A

 

 
1
1 1
1 1
,

,1 ,
3
4 2
5 3

## But the union of no finite subclass of G contains A.

Example 4.14 (Counter-example). Consider the closed infinite interval A = [1, ). The
class
G = {(0, 2), (1, 3), (2, 4), . . .}

(11)

## of open intervals covers A but no finite subclass does.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 23 of 61

4.3 Exercise
Exercise 4.15. Show that a compact metric space (X, d) is complete.
Exercise 4.16.
Exercise 4.17.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 24 of 61

Solution
Solution (4.15). Since X is a compact metric space it is sequentially compact.
Let xn be a Cauchy sequence in the metric space X. Since X is sequentially compact there is a
convergent subsequence xnk x X.
All that now remains to be shown is that xn x. Since xnk x there is N1 with nk N1
implies |xnk x| < 2 . Let N2 be such that n, m N2 implies |xn xm | < 2 .
Then n > N = max(N1 , N2 ) implies
|xn x| |xn xN | + |xN x| <
Hence X is complete.
Solution (4.16).
Solution (4.17). Exercise.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 25 of 61

5 Sequences
5.1 Definition
Definition 5.1 (Sequence). A sequence is a function whose domain is the set of natural
numbers. That is f : N R.
Notation 5.2. We usually denote a sequence by
{a1 , a2 , a3 , . . .},

or

{ak }

## Definition 5.3 (Subsequence). A sequence {ank } = {an1 , an2 , an3 , . . .} is a a subsequence of

{ak } iff n1 , n2 , n3 , . . . are positive integers, with n1 < n2 < n3 < .
Definition 5.4 (Bounded). A sequence {ak } is bounded if there is M > 0 such that |ak | < M
for all k N.
Definition 5.5 (Convergent). A sequence {ak } converges to a R;
lim ak = a

## iff for every > 0 there exists an N N such that

k>N

|ak a| < .

Definition 5.6 (Cauchy criterion). A sequence {ak } is Cauchy if for every > 0 there exists
an N N such that
m, n > N

|am an | < .
Example 5.7 (Counter-example). Consider Q with the usual metric d = |y x|, x, y Q.
Show that (Q, d) is not complete.
Solution. Consider the sequence defined by
x1 = 1,

xn+1 =

xn
1
+
,
2
xn

(12)

which is Cauchy but it doesnt converge in Q. If the sequence converges say to x, then xn+1 x
and xn x and hence x2 = 2.

5.2 Facts
Theorem 5.8 (Uniqueness). If {ak } converges, then the limit is uniquely determined.
Suppose lim ak = a and lim ak = b, with a 6= b.
We have for every > 0, in particular = 41 |b a| there is Na () such that
k > Na ()

1
|ak a| < |b a|.
4

k > Nb ()

1
|ak b| < |b a|.
4

Similarly, we have

Now,
1
1
1
|a b| = |a ak + ak b| |a ak | + |ak b| |b a| + |b a| = |b a|
4
4
2
This implies |b a| = 0, thus a = b a contradiction to our assumption that a 6= b.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 26 of 61

## Theorem 5.9. If {an } converges to a limit a R, then {an } is bounded.

Proof. Assume {an } converges to a. Then by definition we have, for > 0 there is N > 0 such
that
n>N

|an a| <

In particular let = 1. Thus, |an a| < 1 which implies |an | < 1 + |a| so for n > N the sequence
{an } is bounded by 1 + |a|. Otherwise {an } is bounded by the max{|a1 |, |a2 |, . . . , |aN |}. Put
M = max{|a1 |, |a2 |, . . . , |aN |, 1 + |a|}
Then |an | < M for all n N. Therefore {an } is bounded.
Example 5.10 (Counter-example). The converse of theorem (5.9) may not be true. Take
for example(usually) the sequence
{(1)n }
which is bounded but not convergent.
Theorem 5.11. If {ak } converges to a, and if {ank } is a subsequence of {ak } then {ank }
converges to a.
Proof. We first show that
nk k

for all k N.

## For k = 1, it is trivial n1 1. Now we assume nk k. Since nk+1 > nk , hence nk+1 nk + 1.

Then we have
nk+1 nk + 1 k + 1.
Since {ak } converges to a, we have, for every > 0, there is N N such that
|ak a| <

for all k N.

|ank a| <

for all nk N.

Hence
Since nk k, we have also
|ank a| <

for all k N.

## Theorem 5.12. Every convergent sequence is Cauchy sequence.

Proof. Exercise
Theorem 5.13. Every Cauchy sequence of real numbers is convergent.
Proof. Exercise

www.albohessab.weebly.com

(DRAFT-VERSION)

page 27 of 61

5.3 Exercise
Exercise 5.14. Show that {xn } defined by
xn = 1 +

1
1
+ +
2
n

is divergent.
Exercise 5.15. Show that {xn } defined by
xn = 1 +

1
1
+ + ln(n)
2
n

is convergent.
Exercise 5.16.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 28 of 61

Solution
Solution (5.14). We have
1
1
1
+
+ +
n+1 n+2
2n

x2n xn =
for any n 1. So

1
1
1
+
+ +
x2n xn
n+n n+n
2n
or

1
2

x2n xn . This clearly implies that {xn } fails to be Cauchy. Therefore it diverges.

Solution (5.15). Recall the integral definition of the logarithm function. In particular, we have
Z x
1
ln(x) =
dt.
1 t
In this case if 0 < a < b, then we have
ba

b
Since
Z
ln(n) =
1

Z
a

1
ba
dt
.
t
a
n1 Z k+1

X
1
dt =
t
k=1

1
dt,
t

we get
ln(n)

n1
X
k=1

k+1k
1
1
= 1 + + +
.
k
2
n1

Hence

1
1
1
1
1
+ + ln(n) = 1 + + +
ln(n) + > 0.
2
n
2
n1
n
On the other hand, we have
Z n+1
1
1
1
xn+1 xn =
ln(n + 1) + ln(n) =

dt < 0.
n+1
n+1
t
n
xn = 1 +

These two inequalities imply that {xn } is decreasing and bounded below by 0. Therefore {xn }
is convergent5 .
Solution (5.16). Exercise.

## Its limit is = 0.57721 . . ., which is known as the Euler constant.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 29 of 61

6 Continuity
6.1 Definition
Definition 6.1 (Continuous). A function f : R R is continuous at a point x0 iff for every
> 0 there is a > 0 such that
|x x0 | <

## |f (x) f (x0 )| <

which is equivalent to
x (x0 , x0 + )

(13)

## which is again equivalent to f [(x0 , x0 + )] is contained in (f (x0 ) , f (x0 ) + ).

6.2 Facts
Theorem 6.2 (Characterization). A function f : X Y is continuous iff the inverse image
of every open set is open.
Formal Proof. Suppose f is continuous on X and V is an open set in Y . We have to show that
every point of f 1 (V ) is an interior point of f 1 (V ). So, suppose p X and f (p) V . Since
V is open, there exists > 0 such that y V if dY (f (p), y) < ; and since f is continuous at
p, there exists > 0 such that dY (f (x), f (p)) < if dX (x, p) < . Thus x f 1 (V ) whenever
dX (x, p) < .
Conversely, suppose f 1 (V ) is open in X for every open set V in Y . Fix p X and > 0, let
V = {y Y : dY (y, f (p)) < }. Then V is open ; hence f 1 (V ) is open ; hence there exists
> 0 such that x f 1 (V ) whenever dX (p, x) < . But if x f 1 (V ), then f (x) V , so that
dY (f (x), f (p)) < .
Informal Proof. Note that even if f is not a bijective( and doesnt have an inverse function f 1 )
the set f 1 (B) = {x X|f (x) B} still exists.
() Take x f 1 (B). Then f (x) = y B. Since B is open the point y has an neighbourhood
B. Then from the definition of continuity this neighbourhood contains the image of some
neighbourhood of x. Since f (V ) B, we have V f 1 (B) and so x has this neighbourhood
f 1 (B). Hence f 1 (B) is open in X.
() To show f is continuous at x0 X, take B to be an neighbourhood of y Y . Then
f 1 (B) is open in X and so x0 has neighbourhood in f 1 (B). This neighbourhood is
mapped inside the neighbourhood of f (x0 ). Hence f is continuous at x0 .
Corollary 6.3. A function f : X Y is continuous iff the inverse image of every closed set is
closed.
Proof. The proof follows, since a set is closed iff its compliment is open, and f 1 (E c ) = [f 1 (E)]c
for every E Y .
Remark 6.4. Openness and closedness are preserved under inverse, continuous images. Be
aware that these properties are not preserved under continuous, direct images; even if f is
continuous, the image f (U ) of an open set U need not be open, and the image f (F ) of a closed
F need not be closed.
Example 6.5 (Counter-example). Let f, g : R R be the continuous functions defined by
f (x) = x2

and

g(x) = arctan x

The set R is both open and closed, but f (R) equals [0, ) which is not open, and g(R) equals
( 2 , 2 ) which is not closed. Hence the continuous image of an open set need not be open, and
the continuous image of a closed set need not be closed.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 30 of 61

Theorem 6.6. Suppose f is a continuous mapping of a compact metric space X into a metric
space Y . Then f (X) is compact.
Proof. Let {V } be any open cover of f (X). Since {v } is open in Y , f 1 (V ) is open in X.
Since X is compact there are finitely many i s such that
X f 1 (V1 ) f 1 (V2 ) f 1 (Vn )
[

n
n
[
1
1
X
f (Vi ) = f
V i
i=1

i=1

f (X) f f 1

[
n


Vi

i=1

f (X)

n
[

V i

i=1

## Hence, f (X) is compact.

Theorem 6.7. If f is a continuous mapping of a metric space X into a metric space Y , and if
E is a connected subset of X, then f (E) is connected.
Proof. Assume, on the contrary, that f (E) is not connected i.e. f (E) = A B, where A and B
are nonempty separated subsets of Y . Put G = E f 1 (A), H = E f 1 (B).
Then E = G H, and neither G nor H is empty.
Since A A (the closure of A ), we have G f 1 (A) ; the latter set is closed, since f is
continuous; hence G f 1 (A). It follows that f (G) A. Since f (H) = B and A B is empty,
we conclude that G H is empty.
The same argument shows that G H is empty. Thus G and H are separated. This contradict
the fact that E is connected.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 31 of 61

6.3 Exercise
Exercise 6.8. Let I = [0, 1] be the closed unit interval. Suppose f is a mapping of I into I.
Prove that f (x) = x for at least one x I.
Exercise 6.9. Let f be a continuous real function on a metric space X. Let Z(f ) (the zero set
of f ) be the set of all p X at which f (p) = 0. Prove that Z(f ) is closed.
Exercise 6.10.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 32 of 61

Solution
Solution (6.8). If f (0) = 0 or f (1) = 1, we are done. If not, then 0 < f (0) and f (1) < 1. Hence
the continuous function g(x) = x f (x) satisfies g(0) < 0 < g(1). Then by Intermediate value
theorem, there must be a point x (0, 1) where g(x) = 0.
Solution (6.9). Z(f ) = f 1 ({0}), which the inverse image of a closed set. Hence Z(f ) is
closed.
Solution (6.10). Exercise.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 33 of 61

## 7 Sequence and Series of Functions

7.1 Definition
Definition 7.1 (Pointwise convergence). A sequence of functions {fn : D R} defined on
a subset D R converges pointwise on D if for each x D the sequence of numbers {fn (x)}
converge. If {fn } converges pointwise on D, then we define f : D R with f (x) = limn fn (x)
for each x D. We denote this symbolically by fn f on D.
Definition 7.2 (Uniform convergence). A sequence of functions {fn : D R} defined on a
subset D R converges uniformly on D to a function f such that for every > 0 there is a
number N N such that
|fn (x) f (x)| <

for all x D, n N.

## We denote this type of convergence symbolically by fn f on D.

P
Definition 7.3. If {fn } is a sequence of functions defined on D, the series
n=0 fn is said to
converge pointwise(respectively, uniformly) on D if and only if the sequence {sn } of partial sums,
given by
n
X
sn (x) =
fk (x),
k=0

## converges pointwise (respectively, uniformly) on D.

Definition 7.4 (Pointwise bounded). Let {fn } be a sequence of functions defined on a set
E. We say that {fn } is pointwise bounded on E if the sequence {f (x)} is bounded for every
x E, that is, if there exists a finite-valued function defined on E such that
|fn (x)| < (x)

(x E, n = 1, 2, 3, . . .).

## Definition 7.5 (Uniformly bounded). {fn } is uniformly bounded on E if there exists a

number M such that
|fn (x)| < M
(x E, n = 1, 2, 3, . . .).
Definition 7.6 (Equicontinuous). A family F of functions f defined on a set E in a metric
space X is said to be equicontinuous on E if for every > 0 there exists a > 0 such that
|f (x) f (y)| <
whenever d(x, y) < , x, y E, and f F.

7.2 Facts
Theorem 7.7 (Uniform Cauchy Criterion). If {fn } is a sequence of functions defined on
A, then {fn } is uniformly convergent on D if and only if for > 0, there exists N N such
that for any m, n > N
sup |fn (x) fm (x)| < .
xD

Proof. () Suppose fn f , then given > 0, we can find an integer N such that
m N |fm (x) f (x)| <

## for all x D. If m, n N , then

|fm (x) fn (x)| |fn (x) f (x)| + |f (x) fn (x)| <

www.albohessab.weebly.com

(DRAFT-VERSION)

+ =
2 2

page 34 of 61

for any x D.
() If given some > 0, we can find an N such that for any m, n N we have
sup |fn (x) fm (x)| < .
xD

Hence for any x D, {fn (x)} is a Cauchy sequence in R which implies fn (x) converges to f (x).
Let us prove that {fn } converges uniformly to f on D. Indeed, let > 0. By the uniform Cauchy
criterion, there exists N N such that for all n, m N , we have supxD |fn (x) fm (x)| /2.
Fix n N . Then

## |fn (x) f (x)| = lim |fn (x) fm (x)|

m
2
for any x D. Hence

## sup |fn (x) f (x)| < .

2
xD
This implies fn f on D.
Theorem 7.8 (M-test). A sequence of functions {fn } defined on D is uniformly convergent
on U D to f : U R if and only if limn Mn = 0, where
Mn := sup{|fn (x) f (x)| : x U }, n N.
Proof. () Suppose fn f on U . Then given > 0, there exists N N such that
|fn (x) f (x)| < n N x U.
Hence,
an := sup{|fn (x) f (x)| : x U, n N } <
and therefore limn an = 0. () Suppose limn an = 0. Then for large n and x U
|fn (x) f (x)| sup{|fn (x) f (x)| : x U } < .
That is, fn f on U .
Theorem 7.9 (Weierstrass M-test). Suppose that {fn } is a sequence of functions defined on
D and {Mn } is a sequence of nonnegative integers such that
|fn (x)| Mn , x D, n N.
P
P
Show that if
n=0 Mn converges, then
n=0 fn (x) converges uniformly on D.
Pn
P
Proof. Let sn (x) = k=0 fk (x) be the nth partial sum. Since
n=0 Mn converges, for all > 0,
N such that if n m N , then
Mm + Mm+1 + + Mn < .
Thus if n m N , we have
|sn (x) sm (x)| = |fm+1 (x) + + fn (x)|
|fm+1 (x)| + + |fn (x)|
Mm+1 + + Mn <
for all x
P D. It follows (from uniform cauchy criterion) that {sn } converges uniformly on D.
Hence
n=0 fn also converges uniformly on D.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 35 of 61

Theorem 7.10. Assume that fn converges uniformly to f on [a, b] and that each fn is integrable.
Then, f is integrable and
Z b
Z b
lim
fn =
f
n a

Proof. Exercise
Theorem 7.11 (Termwise integration).
If each function fn (x) is continuous on a closed
P
interval [a, b] and if the series
f
(x)
converges
uniformly on [a, b], then we have
n=1 n
Z
X

fn (x)dx =

n=1 a

Z bX

fn (x)dx

a n=1

Proof. Exercise
Theorem 7.12. Suppose that {fn } converges to f on the interval [a, b]. Suppose also that fn0
exists and is continuous on [a, b], and the sequence {fn0 } converges uniformly on [a, b]. Then
lim f 0 (x)
n n

= f 0 (x)

## for each x [a, b].

Proof. Exercise
Theorem 7.13 (Termwise differentiation).
If each function fn (x) has the derivative fn0 (x)
P
at anyP
point x (a, b), if the series
one point k (a, b),
n=1 fn (x) converges to at least P

0
and if n=1 fn (x) converges uniformly on (a, b) to a function g(x), then
n=1 fn (x) converges
uniformly on (a, b) and is differentiable at any point x (a, b), whose derivative is equal to g(x).
In other words, termwise differentiation is possible, i.e.,
X
0 X

fn (x) =
fn0 (x)
n=1

n=1

Proof. Exercise
Theorem 7.14 (Dini). Let A R be a closed and bounded (and thus compact) set and {fn }
be a sequence of continuous functions fn : A R such that
a) fn (x) 0 for any x A,
b) fn f pointwise and f is continuous,
c) fn+1 (x) fn (x) for any x A, and any n N. Then {fn } converges to f uniformly on A.
Proof. Exercise
Theorem 7.15 (Arzela-Ascoli). Let {fn } be a sequence of functions defined on a compact set
K. If fn is uniformly bounded and equicontinuous on K, then fn has a uniformly convergent
subsequence.
Proof. Exercise
Theorem 7.16 (Stone-Weierstrass). If f is a continuous function on a closed interval [a, b],
then there exists a sequence of polynomials Pn such that
lim Pn (x) = f (x)

## uniformly on [a, b].

Proof. Exercise

www.albohessab.weebly.com

(DRAFT-VERSION)

page 36 of 61

7.3 Exercise
Exercise 7.17. Let fn (x) : [1, 2] R be defined by
fn (x) =
(a) Show that

n=1 fn (x)

x
(1 + x)n

## (b) Does the following hold:

Z 2 X

n=1


Z
X
fn (x) dx =

fn (x)dx

n=1 1

Exercise 7.18.
Exercise 7.19.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 37 of 61

Solution
Solution (7.17). (a) First observe that

X
n=1

X
x
1
=
x
(1 + x)n
(1 + x)n
n=1

and when |1/(1 + x)| < 1 or equivalently when |1 + x| > 1 we have that

x
1
x
=

1 =1
(1 + x)n
1 + x 1 1+x
n=1
P
so in particular
n=1 fn (x) is convergent for x [1, 2].
A = [1, 2] is compact and fn (x) 0 pointwise. Clearly, if k l holds, then fl (x) fk (x).
All the hypotheses of Dinis theorem are satisfied and thus convergence is uniform.
(b) Since the convergence is uniform we can interchange the integral and the summation. Thus
the equality holds.
Solution (7.18). Exercise.
Solution (7.19). Exercise.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 38 of 61

8 Riemann-Stieltjes Integral
8.1 Definition
Definition 8.1 (Partition). Let [a, b] be a given interval. By a partition P of [a, b] we mean a
finite set of points x0 , x1 , , xn , where
a = x0 x1 xn1 xn = b
We write
xi = xi xi1

(i = 1, . . . , n).

## Definition 8.2 (Refinement). We say that the partition P is a refinement of P if P P

(that is, if every point of P is a point of P ). Given two partitions, P1 and P2 , we say that P is
their common refinement if P = P1 P2 .
Definition 8.3 (Upper and Lower sum). Suppose f is a bounded real function defined on
[a, b]. Corresponding to each partition P of [a, b] we put
Mi = sup f (x)

(xi1 x xi )

mi = inf f (x)

(xi1 x xi )

n
X
U (f, P ) =
mi (xi xi1 ).
i=1

## Similarly, we define the upper sum of f with respect to P by

n
X
L(f, P ) =
Mi (xi xi1 ).
i=1

Definition 8.4 (Riemann Sum). Let P = {a = x0 < x1 < x2 < < xn = b} be a partition
of [a, b]. A tagged partition is one where in addition to P we have chosen points xk in each of
the subintervals [xk1 , xk ]. Suppose f : [a, b] R and a tagged partition (P, xk ) is given. The
Riemann sum generated by this partition is given as
n
X
R(f, P ) =
f (xk )(xk xk1 ).
k=1

It is clear that
L(f, P ) R(f, P ) U (f, P )
true for any bounded function.
Definition 8.5 (Upper and Lower Integrals). Let P be the collection of all possible partitions
of the interval [a, b]. The lower integral of f
Z b
f (x)dx = L(f ) = sup{L(f, P ) : P P}.
a

## Likewise the upper integral of f

Z b
f (x)dx = U (f ) = inf{U (f, P ) : P P}.
a

## Clearly for a bounded function f on [a, b] we always have U (f ) L(f ).

Definition 8.6 (Riemann Integrable). If the upper and lower integrals are equal, we say
that f is Riemann integrable on [a, b], we write f R (that is, R denotes the set of Riemann
integrable functions), and we denote the common value by
Z b
f dx
a

www.albohessab.weebly.com

(DRAFT-VERSION)

page 39 of 61

8.2 Facts
Theorem 8.7. If P is a refinement of P , then
L(P, f, ) L(P , f, )

(14)

U (P , f, ) U (P, f, ).

(15)

and

Proof. Exercise
Theorem 8.8.
Z

Z
f d

f d.
a

Proof. Let P be the common refinement of two partitions P1 and P2 By Theorem (8.7),
L(P1 , f, ) L(P , f, ) U (P , f, ) U (P2 , f, ).
Hence
L(P1 , f, ) U (P2 , f, ).
If P2 is fixed and the sup is taken over all P1 , (16) gives
Z
f d U (P2 , f, )

(16)

(17)

## The theorem follows by taking the inf over all P2 in (17).

www.albohessab.weebly.com

(DRAFT-VERSION)

page 40 of 61

8.3 Exercise
Exercise 8.9. If f (x) = 0 for all irrational x, f (x) = 1 for all rational x, prove that f
/ R on
[a, b] for any a < b.
Exercise 8.10.
Exercise 8.11.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 41 of 61

Solution
Solution (8.9). Every upper Riemann sum equals b a, and every lower Riemann sum equals 0.
Hence the set of upper sums and the set of lower sums do not have a common bound.
Solution (8.10). Exercise.
Solution (8.11). Exercise.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 42 of 61

9 Measure Theory
9.1 Definition
Definition 9.1 (Set function). Let C be class of subsets of X.
A function : C [0, +] is called a set function.
A set function is said to be
i) Monotone: if for A, B C
(A) (B),

whenever A B.

[

 X

(An )
Ai =
n=1

i=1

G


An =

n=1

(An )

n=1

(A)

(An )

n=1

## Definition 9.2 (Algebra). A non-empty collection A of subsets of X such that

i) , X A
ii) A A Ac A.
iii) A1 , . . . , An A ni=1 Ai A.
Definition 9.3 (-algebra). A collection of subsets of X is -algebra in X if
i) , X M
ii) A M Ac M whenever Ac is a compliment relative to X.
iii) A1 , . . . , An M
i=1 Ai M.
(X, M) - measurable space.
The members of M is called measurable sets.
Definition 9.4 (Borel set). The intersection of all the -algebras of subsets of R that contains
open set is called borel algebra. Members of a Borel algebra are called borel sets.
Definition 9.5 (Outer measure). Let A R, then the outer measure of A is defined as
(A) = inf

X

`(Ik ),

k=1

Remark 9.6.


Ik

k=1

1. m () = 0.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 43 of 61

2. A B m (A) m (B).
Definition 9.7 (Measurable). A set E is measurable if for any set A
(A) = (A E) + (A E c )
Remark 9.8. If E is a measurable set we write (E) in place of (E) for Lebesgue measure
of E.
Definition 9.9 (Measure). Let X be a set and A a -algebra consisting of subsets of X. A
measure on (X, A) is a function : A [0, ] such that
1. () = 0
2. If Ai A, for i = 1, 2, . . . are pairwise disjoint, then

[


Ai =

i=1

(Ai )

i=1

Remark 9.10.
1. A set function : C [0, +] is a measure on C if is countably additive
on C and C with () = 0.
2. The measure of an interval is its length.
3. Measure is translation invariant i.e. If E is Lebesgue measurable and y R
(E + y) = (E)
4. Measure is countably additive over countable disjoint union of sets.
Definition 9.11 (Lebesgue Measurable). The restriction of the set function outer measure
to the class of measurable is called Lebesgue measurable.
Definition 9.12 (Measurable function). Let f be an extended real valued function defined
on a measurable set E. Then f is Lebesgue measurable function or measurable function if, for
each R the set
{x : f (x) > }
is measurable.

9.2 Facts
Proposition 9.13. A countable set has measure zero.
Proof. Let C be any countable set enumerated as C = {ck }
k=1 . Let > 0. For each k N
define



Ik = ck k+1 , ck + k+1
2
2
The countable collection of open intervals {Ik }
k=1 covers C. Therefore
0 m (C)

`(Ik ) =

k=1

= .
2k
k=1

## The inequlity holds for each > 0, hence m (C) = 0.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 44 of 61

## Proposition 9.14. The outer measure of an interval is its length.

Proof. Exercise.
Proposition 9.15. Outer measure is translation invariant i.e. for any set A and y R,
m (A + y) = m (A)

Proof. If {Ik }
k=1 is any countable collection of sets, then {Ik }k=1 covers A if and only if

## {Ik + y}k=1 covers A + y.

Moreover if Ik is open interval, then each Ik + y is an open interval of the same length, and so

`(Ik ) =

k=1

`(Ik + y)

k=1

Thus m (A + y) = m (A).
Proposition 9.16. Outer measure is countably sub-additive i.e. if {Ek }
k=1 is any countable
collection of sets, then
[
 X

m
Ek
m (Ek )
k=1

k=1

## Proof. If one of Ek s has infinite outer measure, the inequality holds.

Suppose each Ek has finite outer measure. Let > 0, for each k N, there is countable collection
{Ik,i }
i=1 of open, bounded intervals for which
Ek

Ik,i

and

i=1

## `(Ik,i ) < m (Ek ) +

i=1

2k

Now {Ik,i }1k,i< is countable collection of open, bounded intervals that covers Ek since it is
a countable collection of countable collections. Thus,
X

m (Ek )

`(Ik,i ) =

1k,i<

<

X

`(Ik,i )

k=1 i=1

X

k=1

m (Ek ) + k
2

m (Ek ) +

k=1

Since this holds for each > 0, it also holds for  = 0. Therefore,

[

k=1


Ek

m (Ek )

k=1

## Proposition 9.17. Any set of outer measure zero is measurable.

Proof. Exercise
Proposition 9.18. The union of a finite collection of measurable sets is measurable.
Proof. Exercise

www.albohessab.weebly.com

(DRAFT-VERSION)

page 45 of 61

Proposition 9.19. Let A be any set and {Ek }nk=1 be a finite disjoint collection of measurable
sets. Then

[
 X
n
n
m A
Ek =
m (A Ek )
k=1

k=1

Proof. Exercise
Proposition 9.20. A union of countable collection of measurable sets is measurable.
Proof. Exercise
Proposition 9.21. Every interval is measurable.
Proof. Exercise
Proposition 9.22. The translate of a measurable set is measurable.
Proof. Let E be a measurable set. Let A be any set and y R. By measurability of E and the
translation invariant of outer measure
m (A) = m (A y) = m [(A y) E] + m [(A y) E c ]
= m [A (E + y)] + m [A (E + y)c ]
Therefore, E + y is measurable.
Remark 9.23. Outer measure fails to be countably additive. If the sets are measurable, then
the outer measure is countably additive. Thus Lebesgue measurable is countably additive.
Theorem 9.24. Every Borel set is measurable.
Proof. Exercise

www.albohessab.weebly.com

(DRAFT-VERSION)

page 46 of 61

9.3 Exercise
Let m be a set function defined for all sets in a -algebra A with values in [0, ]. Assume m is
countably additive over countable disjoint collections of sets in A.
Exercise 9.25. Prove that if A and B are two sets in A with A B, then m(A) < m(B). This
property is called monotonicity.
Exercise 9.26. Prove that if there is a set A in the collection A for which m(A) < , then
m() = 0.
Exercise 9.27. Let {Ek }
k=1 be a countable collection of sets in A. Prove that
[
 X

m
Ek
m(Ek ).
k=1

k=1

## Exercise 9.28. Prove that the interval I = [0, 1] is uncountable.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 47 of 61

Solution

}|
{
z
Solution (9.25). Consider B = A (B A), m(A) m(A) + m(B A) = m(B).
| {z }
0

Solution (9.26).

z
}|
{
m(A) = m(A ) = m(A) + m()
m() = 0.

Solution (9.27). Let F1 = E1 and let Fn+1 = En+1 nk=1 , then Fm Fn = , for n 6= m. By
definition Fn En for each n. Moreover Ek = Fk .
[

[
 X

X
m
Ek = m
Fk =
m(Fk )
m(Ek )
k=1

k=1

k=1

k=1

[
 X

m
Ek
m(Ek ).
k=1

k=1

Solution (9.28). Suppose not, i.e. I = [0, 1] is countable. Since a measure of a countable set is
0, m(I) = 0. But measure of an interval is its length, hence m(I) = 1 .
Therefore, I = [0, 1] is uncountable.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 48 of 61

## 10 Measure and Integration

10.1 Definition
Definition 10.1 (Characteristic Function). For any set A the function
(
1
if x A
A (x) =
0
if x
/A
is called the characteristic function of A.
Definition 10.2 (Simple Function). A finite linear combination of characteristic functions
s(x) =

n
X

ai Ei (x)

(18)

i=1

## is called simple function if all sets Ei are measurable.

Remark 10.3.
Definition 10.4 (Lebesgue integral for simple function). For a simple function s defined
on a set of finite measure E,
Z
n
X
s=
ai m(Ei )
E

i=1

## where s is has a canonical representation given by (18).

Definition 10.5 (Lower and upper integral). Let f be a bounded real-valued function
defined on a set of finite measure E. We define the lower and upper Lebesgue integral,
respectively, of f over E to be
Z


sup
simple and f on E
E

and

Z
inf
E



simple and f on E

## Definition 10.6 (Lebesgue Integrable). A bounded function f on a domain E of finite

measure is said to be Lebesgue integrable over E provided its upper and lower Lebesgue
integrals over E are equal. The common value of the upper and lower Rintegrals is called the
Lebesgue integral, or simply the integral, of f over E and is denoted by E f .
Definition 10.7 (Lebesgue Integral of non negative function). For f a nonnegative
measurable function on E, we define the integral of f over E by
Z

Z
f = sup
h|h bounded, measurable, of finite support and 0 < h < f on E .
E

## Definition 10.8 (Absolute Continuous). A real valued function f is said to be absolutely

continuous on [0, 1]. Provided that for every > 0 there exists a > 0 such that for every finite
disjoint collection {(ak , bk )}nk=1 of open intervals in (a, b)
if

www.albohessab.weebly.com

n
X
[ak bk ] < ,

n
X

k=1

k=1

(DRAFT-VERSION)

page 49 of 61

## Definition 10.9 (Uniformly Integrable). A family F of measurable functions on E is said

to be uniformly integrable over E provided for each > 0 there is a > 0 such that for each
f F
Z
If A E is measurable and m(A) < , then

|f | < .
E

## Definition 10.10 (Convergence in measure). Let {fn } be a sequence of measurable functions

on E and f measurable function on E for which f and fn are finite a.e on E. The sequence
{fn } is said to converge in measure on E to f provided for each > 0,
lim m{x E : |fn (x) f (x)| > } = 0.

10.2 Facts
Littlewoods Three Principles
1. Every set(measurable) is nearly a finite union of intervals.
2. Every function(measurable) is nearly continuous.
3. Every piecewise convergent(measurable) sequence is nearly uniformly convergent.
Nearly means almost everywhere(a.e). We say that a property about real numbers x holds
almost everywhere (with respect to Lebesgue measure ) if the set of x where it fails to
be true has measure 0.
Here are the formulations of Littlewoods principles
Theorem 10.11 (First Principle). Let E be a measurable set of finite outer measure. Then
S for
each > 0, there is a finite disjoint collection of open intervals {Ik }nk=1 for which if O = nk=1 Ik ,
then
m (E \ O) + m (O \ E) < .
Theorem 10.12 (Second Principle/Egoroff s Theorem). Assume E has finite measure.
Let {fn } be a sequence of measurable functions on E that converges pointwise on E to the
real-valued function f . Then for each > 0, there is a closed set F contained in E for which
{fn } f uniformly on F and m(E \ F ) < .
Theorem 10.13 (Third Principle/Lusins Theorem). Let f be a real-valued measurable
function on E. Then for each > 0, there is a continuous function g on R and a closed set F
contained in E for which
f = g on F and m(E \ F ) < .
Theorem 10.14 (Chebychevs Inequality). Let f be a nonnegative measurable function on
E. Then for any > 0
Z
1
m{x E|f (x) }
f
E
Proof. Define
E := {x E|f (x) }

## Theorem 10.15 (Zero Integral). Let f be a nonnegative measurable function on E. Then

Z
f = 0,
if and only if f = 0 a.e on E.
E

www.albohessab.weebly.com

(DRAFT-VERSION)

page 50 of 61

Proof.
Theorem 10.16 (Bounded Convergence). Let {f } be a sequence of measurable functions
on a set of finite measure E. Suppose {fn } is uniformly pointwise bounded on E, that is, there
is a number M > 0 for which
|fn | < M on E for all n.
Z
Z
If {fn } f pointwise on E, then lim
fn =
f.
n E

Proof.
Theorem 10.17 (Monotone Convergence). Let {fn }n1 L+ increasing to f .
Proof.
+
Example 10.18 (Counter-example).
Take fn = [n,) . Then
fn L+
0 L , and {fn }n1
R
R
decreases to f 0. However, fn d = + for every n and f = 0.

## Theorem 10.19 (Fatous lemma). Let {fn }n1 L+ . Then


Z 
Z
lim inf fn d lim inf fn d
n

Proof. Define
gn = inf {fk }
kn

## Then gn 0, gn is measurable, gn fn : the sequence gn is increasing, and lim gn = lim inf fn .

Therefore, MCT implies
Z
Z
(lim inf fn )d = (lim gn )d
Z
= lim gn d
Z
= lim inf gn d
Z
lim inf fn d
(since gn fn .)

Theorem 10.20 (Lebesgue Dominated Convergence). Let {fn } be a sequence of measurable function on E. If there is an integrable function g such that |fn | g for all n, then
Z
Z
if {fn } f pointwise a.e on E, then lim
fn =
f
n E

## Proof. Since |fn | g for all n we have |f | g.

Consider {g fn }, {g fn } g f 0 Using Fatous Lemma
Z
Z
Z
Z
g f lim inf g fn = g lim inf fn

Z
lim sup

www.albohessab.weebly.com

Z
fn

(DRAFT-VERSION)

(19)

page 51 of 61

## Similarly use Fatous lemma for {g + fn }

Z
Z
Z
Z
g + f lim inf g + fn = g + lim inf fn

Z
f lim inf

fn

(20)

## The result follows from (19) and (20).

www.albohessab.weebly.com

(DRAFT-VERSION)

page 52 of 61

10.3 Exercise
Exercise 10.21. Prove that f (x) =
Exercise 10.22. Evaluate

Z
X

/2

(1

n=0 0

## Exercise 10.23. Let (X, M, ) be a measurable space and f L1 L+ . Show that



Z
Z
f (x)
d = f d
lim
n ln 1 +
n
n
Exercise 10.24. Let f (x) = [1,1] + 3[2,2] + [0,) (3,) . Show that f is a simple,
non-negative lebesgue measurable function and compute its standard representation and its
lebesgue integral.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 53 of 61

Solution
Solution (10.21). Let (ak , bk )nk=1 be a disjoint collection of open intervals in (0, 1). Given > 0
we need to find a > 0 such that
if

n
X

|bk ak | , then

k=1

|f (ak ) f (bk )|

k=1

Consider


n p
n p
X
X
bk ak

| bk ak | =
| bk ak |

bk ak
k=1
k=1

n 
X
bk ak

bk ak
k=1

n
X

|bk ak | <

k=1

## Choose = . Hence f is absolutely continuous.

To show f is not Lipschitz: Assume the contrary suppose f is Lipschitz on [0, 1] then we have
|f (x) f (y)| K|x y|,
x, y [0, 1]

| x y| K|x y|
Let y = 0

x Kx,

x [0, 1]

Thus,
1
K ,
x

x (0, 1]

Then as x 0+ K .
Solution (10.22). Since the integrands are non-negative functions, the summation and integration
commute. Hence
Z /2 X
Z /2

n
(1 sin x) cos xdx =
(1 sin x)n cos xdx
n=0 0

n=0
/2

cos x
0

Z
=
0

/2

dx
1 (1 sin x)

cos x

dx
sin x

Z

/2

cos x

dx =
sin x

Z
0

/2

/2

d sin x

= 2 sin x = 2
sin x
0

## Solution (10.23). Consider



f (x)
fn = n ln 1 +
n

www.albohessab.weebly.com

(DRAFT-VERSION)

page 54 of 61

## f = 0(,2] + 3[2,1] + 4[1,0] + 5[0,1] + 4[1,2] + [2,3] + 0[3,)

= 0(,2)(3,) + (2,3] + 3[2,1) + 4[1,0)(1,2] + 5[0,1]

www.albohessab.weebly.com

(DRAFT-VERSION)

page 55 of 61

## 11 General Measure Space

11.1 Definition
Definition 11.1 (Signed Measure). A signed measure on (X, M) means an extended realvalued set function
: M [, ]
such that
i) assumes at most one of +,.
ii) () = 0.
iii) For a countable collection {Ek }
k=1 of disjoint measurable sets

[

 X

Ek =
(Ek )

k=1

where

k=1 (Ek )

k=1

converges absolutely if (
k=1 Ek ) is finite.

Definition 11.2 (Positive Set). A set A is called positive with respect to a signed measure
if A is measurable and for every subset E of A we have (E) 0.
Definition 11.3 (Negative set). A set A is called negative with respect to a signed measure
if A is measurable and for every subset E of A we have (E) 0.
Definition 11.4 (Null set). A set A is said to be null if it is both positive and negative.
Definition 11.5 (Singular). 1 and 2 defined on M are said to be singular 1 2 if there
are two disjoint measurable sets A and B such that 1 (B) = 2 (A) = 0.
Definition 11.6 (Absolute Continuous Measure). A measure on a measure space (X, M)
is said to be absolutely continuous the measure written  if (A) = 0 for each A with
(A) = 0.

## 11.2 The Lp Space

Definition 11.7 (Norm). Define the norm of f in Lp by
Z
kf kp =

1/p

|f | d

## Definition 11.8 (Conjugate Power). The number q is said to be a conjugate of p if

1 1
+ = 1.
p q
Theorem 11.9 (Youngs Inequality). For 1 < p < , q the conjugate of p, and any two
positive integer a and b
ap bq
ab
+
p
q
Proof. Exercise
Theorem 11.10 (Holders Inequality). Let 1 < p < , q a conjugate of p. If f Lp ()
and g Lq (), then f g L1 () and
Z

Z
|f g|d

www.albohessab.weebly.com

|f | d

1/p Z

1/q

|g| d

(DRAFT-VERSION)

page 56 of 61

Proof. If f = 0 a.e. or g = 0 a.e., then the inequality holds. But if f 6= 0 and g 6= 0 a.e., then
kf kp > 0 and kgkq > 0.
Now consider
|f |
|g|
,
G=
F =
kf kp
kgkq
Notice that
kF kp = 1
So it suffices to show

and

kGkq = 1.

Z
|f g|d 1

FG
This implies
Z

F p Gq
+
p
q

1 1
kF kpp kGkqq
+
= + =1
F Gd
p
q
p q

## Thus, the result follows.

Theorem 11.11 (Minkowskis Inequality). For 1 p < . Then for every pair of f, g
Lp () the following inequality holds
kf + gkp kf kp + kgkp
Proof. If p = 1, the inequality becomes trivial(triangle inequality). If p > 1, consider
Z
Z
|f + g|p d (|f | + |g|)|f + g|p1 d
Z
Z
= |f ||f + g|p1 d + |g||f + g|p1 d
kf kp k(f + g)p1 kq +kgkp k(f + g)p1 kq
|
{z
}

(Holder)

kf +gkp1
p

## = kf + gkpp1 (kf kp + kgkp )

11.3 Facts
Lemma 11.12. Let X be a normed space. Then X is complete if and only if every absolutely
summable series is summable.
Proof. Exercise
Theorem 11.13.
Proof. Exercise

www.albohessab.weebly.com

(DRAFT-VERSION)

page 57 of 61

11.4 Exercise
Exercise 11.14. Let {An } be a countable collection of measurable sets, then

[


Ak = lim

k=1

[


Ak

k=1

Exercise 11.15. Let (X , M , ) be a collection of measure spaces and suppose that the
collection of sets {X } isSdisjoint. Then we can form a new measure space (called
P their union)
(X, B, ) by letting X = X , define B = {B : B X B } and (B) = (B X ).
a) Show that M is a -algebra.
b) Show that is a measure.
c) Show that is -finite if and only if all but a countable number of the measures are zero
and the remainder are -finite.
Exercise 11.16. Let (X, M, ) be a measure space. The symmetric difference, E1 E2 , of two
subsets E1 and E2 of X is defined by
E1 E2 = (E1 \E2 ) (E2 \E1 )
a) Show that if E1 and E2 are measurable and (E1 E2 ) = 0, then (E1 ) = (E2 ).
b) Show that if is complete, E1 M and E2 \E1 M, then E2 M if (E1 E2 ) = 0.
Exercise 11.17 (Hardys Inequality). Suppose that f 0 on (0, ), f Lp (0, ) and
Z
1 x
f (t)dt
F (x) =
x 0
Prove that for 1 < p <
Z

(F (x)) dx
0

Rx
0

p
p1

p Z

(f (t))p dt

## Exercise 11.18. Let f, {fk } Lp where 1 p < . Show that if kf fk kp 0, then

kfk kp kf kp . Conversely if fk f a.e and kfk kp kf kp , show that kf fk kp 0.
Exercise 11.19.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 58 of 61

Solution
Solution (11.14). Let {An } be a collection of measurable sets. Let B1 = A1 and Bk = Ak n1
k=1 An
for k > 1. Then {Bn } is a collection of pairwise disjoint measurable sets such that An = Bn .
Now
X

(Ak ) = (Bk ) =

(Bk ) = lim
n

n
X

n

k=1

## Solution (11.15). Exercise.

Solution (11.16). Exercise.
Z
Z
p
F (x) dx =
0

1
xp

Z
f (t)dt =

Z

1
q

p
f (t)dt dx.

f (t)ta ta dt

Z

 1 Z
p
f (t) t dt

p ap

f (t)dt
0

1
xp
1

xq

dt

and so we have

p ap

Z

f (t) t

f (t) t dt

1


f (t) t dt dx
p
pap
q


dx dt.

dx =

x1pa dx = tpa
Z

F (x) dx

ap(1 aq)

1
q

p ap

p ap

Z

p
pap
q

(1 aq)

(1 aq) q 0
Z
1

1
p+q ,

1

p
f (t)dt dx
0
Z
Z
p
pap
xq

Z

(1 aq) q
Now note that

aq

Z
Z
p
F (x) dx =

Choose now a =
gives

Z

p
q

1
,
ap

f (t)p dt.

## which is found by solving a minimization problem, and then simple algebra



1
p

ap(1 aq) q

www.albohessab.weebly.com


p
.
p1

(DRAFT-VERSION)

page 59 of 61

Solution (11.18). By Minkowskis inequality, kf kp kfk kp kf fk kp , so kf fk kp 0

implies kfk kp kf kp . (If p < 1, we have kf kp kfk kp ckf fk kp for some c > 1, so the
argument still holds.) For the converse, since p 1, we have
|f fk |p 2p1 (|f |p + |fk |p ),
so by Fatous lemma and the fact that fk f a.e., we have
Z
Z
p
p
2 |f | = lim inf 2p1 (|f |p + |fk |p ) |f fk |p
k
Z
Z
p1
p
p
lim inf 2 (|f | + |fk | ) + lim inf |f fk |p
k
k
Z
Z
p1
p
p
= lim 2 (|f | + |fk | ) lim sup |f fk |p
k
k
Z
Z
= 2p |f |p lim sup |f fk |p ,
k
p
p
where the last equality
R followsp by the dominated convergence theorem since |f | and |fk | are in
1
L . Thus, lim supk |f fk | 0, so kf fk kp 0.

## Solution (11.19). Exercise.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 60 of 61

References
[1] Stephen Abbott, 2001, Understanding Analysis. Undergraduate Text in Mathematics,
SpringerVerlag, New York.
1996.

auser Boston in

[4] [Robert Bartle & Donald R.] Introduction to Real Analysis, third edition, John Wiley &
Sons, Inc. 2000.
[5] [Seymour Lipschutz] Schaums Outline of Theory and Problems of Genereal Topology,
McGraw-Hill, New York, 1965.
[6] [Hans Sagan] Advanced Calculus of Real-Valued Functions of a Real Variable and VecterValued Functions of a Vector Variable.
[7] [Y. Sagher]

## Real Analysis, the Macmillan Company, New York , 1963.

[9] [Walter Rudin] Principles of Mathematical Analysis. International Series in Pure and
Applied Mathematics, McGraw-Hill, New York, 1964.

www.albohessab.weebly.com

(DRAFT-VERSION)

page 61 of 61