Introduction to Measure Theory

Szabó Tamás

Bolyai Institute, University of Szeged

Lecture Draft

2023. szeptember 27.

Szabó Tamás Analysis 2023. szeptember 27. 1 / 41

 Riemann integral

The Dirichlet function defined by 

1, if x ∈ Q,
d(x) =
0, if x ∈ R \ Q,

is bounded but is not Riemann integrable over any [a, b].

For an enumeration Q = {r1 , r2 , r3 , . . . } let


1, if x = rn ,
fn (x) =
0, otherwise.

Each fn is clearly Riemann integrable over any [a, b],

Z b XZ b
fn (x) dx = 0 and fn (x) dx = 0 .
a n a

But X
d(x) = fn (x)
n

is not Riemann integrable; therefore,

XZ b Z bX
fn (x) dx ̸= fn (x) dx .
n a a n

Szabó Tamás Analysis 2023. szeptember 27. 2 / 41

Riemann integral

For an enumeration Q = {r1 , r2 , r3 , . . . } let


1, if x ∈ {r1 , r2 , . . . , rn },
gn (x) =
0, otherwise.

Each gn is clearly Riemann integrable over any [a, b],

Z b
gn (x) dx = 0 ,
a
and clearly (gn ) is a bounded, increasing sequence, however, because of

d(x) = lim gn (x)

n

again we have
Z b Z b
lim g (x) dx ̸= lim gn (x) dx .
n a n a n

Note, that
n
X
gn = fk ,
k=1

fn = gn − gn−1 , (g0 ≡ 0) .

Szabó Tamás Analysis 2023. szeptember 27. 3 / 41

Riemann integral

yi+1

yi

xi xi+1 xn

Szabó Tamás Analysis 2023. szeptember 27. 4 / 41

 Lebesgue integral

Instead of taking a partition of the x-axis, we can take a partition of the y -axis.

yi+1

yi+1

yi yi
yi−1

yi−1

Ei Ei

Let f ≥ 0 and
Ei = {x ∈ [a, b] yi−1 < f (x) ≤ yi } .

Then the ”area” of the region under f (x) is bounded below by

n
X
yi−1 · l(Ei ) ,
i=1

where l(Ei ) is the ”length” of Ei , provided that it can be defined.

Szabó Tamás Analysis 2023. szeptember 27. 5 / 41

 Height of the rectangles

φ1 φ2

Clearly
φ1 ≤ φ2 ≤ ... ↗ f

Szabó Tamás Analysis 2023. szeptember 27. 6 / 41

 Simple function

Definition

A simple function over X is a function defined on the whole X whose range is a finite set.

Definition

Let E ⊂ X . The function defined by 

1, if x ∈ E
χE (x) =
0, if x ∈
/ E

is called the characteristic (indicator) function of the set E .

Clearly a characteristic function is a simple function, and φ is a simple function if and only if it is a finite linear combination of characteristic functions.

Definition

For simple φ, if Range(φ) = {c1 , . . . , cn } and

Ek = {x ∈ X f (x) = ck } ,

then
n
X
φ = c k χE
k
k=1

is called the standard representation of φ.

Szabó Tamás Analysis 2023. szeptember 27. 7 / 41

 Lebesgue measure

The Lebesgue measure of a subset E of R, denoted by m(E ), is a generalization of the length of intervals.

Therefore, we expect the Lebesgue measure to be

The usual length.

m([0, 1]) = 1 .

Monotone. If A ⊂ B, then
m(A) ≤ m(B) .

Countable additive. For countable pairwise disjoint sets {En }

˙
[  X
m En = m(En ) .
n n

Translation invariant. If Et = {x + t : x ∈ E }, then

m(Et ) = m(E ) .

Unfortunately, such a measure cannot be defined for all subsets of R.

Szabó Tamás Analysis 2023. szeptember 27. 8 / 41

Can’t measure everything

Define an equivalence relation on [0, 1] by

x ∼ y if and only if x − y ∈ Q .

Let A be the set consisting of exactly one element from each equivalence class.
For any r ∈ S := [−1, 1] ∩ Q, define
Ar = {x + r : x ∈ A} .

Then {Ar } are pairwise disjoint, because only one element of an equivalence class is in A.
Moreover,
˙
[
[0, 1] ⊂ {Ar } ⊂ [−1, 2] .
r ∈S

If we can define the Lebesgue measure on A, then we must have

 
˙
[ X X
1 = m([0, 1]) ≤ m  Ar  = m(Ar ) = m(A) ≤ m([−1, 2]) = 3
r ∈S r ∈S r ∈S

which is impossible.
Therefore, A is not Lebesgue measurable.

Szabó Tamás Analysis 2023. szeptember 27. 9 / 41

 σ-algebra

Definition

Let X be a nonempty set. A σ-algebra M on X is a nonempty collection of subsets of X , M ⊂ P(X ), which is

closed under complements:
C
for any A ∈ M , A ∈ M,

closed under countable unions: [

for any countable sets {A1 , A2 , . . . } ⊂ M , An ∈ M .
n

Example

P(X ) is a σ-algebra, and it is the largest σ-algebra over X.

{∅, X } is a σ-algebra, and it is the smallest σ-algebra over X.
The collection of all countable subsets and their complements is a σ-algebra. It is called the σ-algebra of countable and co-countable sets.

Szabó Tamás Analysis 2023. szeptember 27. 10 / 41

 σ-algebra

Remark

From the definition we can see that any intersection of σ-algebras over X is a σ-algebra.

Definition

Let H ⊂ P(X ). The σ-algebra generated by H, denoted by M(H), is the intersection of all σ-algebras containing H.

Remark

Clearly M(H) is the smallest σ-algebra containing H.

Example

Let X = {1, 2, 3} and H = {{1}}. The σ-algebra generated by H is

M(H) = {∅, {1}, {2, 3}, X } .

Example

The σ-algebra of countable and co-countable sets is generated by the collection

{{x} : x ∈ X }

of all singleton subsets of X .

Szabó Tamás Analysis 2023. szeptember 27. 11 / 41

 Borel σ-algebra

Definition

Let X be a metric space. The σ-algebra generated by open sets of X is called the Borel σ-algebra on X , and in general is denoted by BX , in the case of
X = R simply by B. The sets in BX are called Borel sets.

Remark

B could be generated for example (f.e.) by H = {(−∞, a) : a ∈ R}, or by H = {[a, b] : a, b ∈ R}.

If A and B are Borel sets, then A \ B is also Borel set.
A countable intersection of open sets, denoted by Gδ , is Borel set.
A countable union of closed sets, denoted by Fσ , is Borel set.

Szabó Tamás Analysis 2023. szeptember 27. 12 / 41

 Measure

Definition

Let M be a σ-algebra on X . A measure µ is a function µ : M → [0, ∞] such that

µ(∅) = 0 ,
µ is countably additive, i.e., for any pairwise disjoint {An } ⊂ M σ-additive

!
˙
[ X
µ An = µ(An ) .
n n

Definition

(X , M) is called a measurable space.

(X , M, µ) is called a measure space.
Any set in M is called a measurable set.
(X , M, µ) is called finite if µ(X ) ̸= ∞.
S
(X , M, µ) is called σ-finite if X = n An for some {An } ⊂ M such that µ(An ) ̸= ∞ for all n.
(X , M, µ) is called semifinite if for all A ∈ M with µ(A) = ∞, there exists a B ∈ M with B ⊂ A and 0 < µ(B) < ∞ .

Szabó Tamás Analysis 2023. szeptember 27. 13 / 41

 Measure

Example

Let M = P(X ) and define µ(A) to be the number of elements in A. That is


|A|, if A is finite,
µ(A) =
∞, otherwise.

This measure is called the counting measure on X . Clearly µ is a semifinite measure.

Let M = P(X ) and fix an α in X . Define 
1, if α ∈ A,
µα (A) =
0, otherwise.

µα is called point mass at α. Clearly µα is a finite measure.

Let M be the σ-algebra of countable and co-countable sets, and define

1, if A is co-countable,
µ(A) =
0, if A is countable.

Then µ is a finite measure.

Szabó Tamás Analysis 2023. szeptember 27. 14 / 41

 Measure

Theorem

Let (X , M, µ) be a measure space.

(Monotonicity) If A, B ∈ M with A ⊂ B, then
µ(A) ≤ µ(B) .

(Countable sub-additivity) If {An } ⊂ M, then !

[ X
µ An ≤ µ(An ) .
n n

(Continuity from below) If {An } ⊂ M with A1 ⊂ A2 ⊂ . . . , then

!
[
µ An = lim µ(An ) .
n
n

(Continuity from above) If {An } ⊂ M with A1 ⊃ A2 ⊃ . . . and at least one µ(Ak ) < ∞, then

!
\
µ An = lim µ(An ) .
n
n

Szabó Tamás Analysis 2023. szeptember 27. 15 / 41

 Almost everywhere

Definition

Let (X , M, µ) be a measure space.

A null set is a measurable set with measure zero.
A property is said true almost everywhere (a.e.) if the exception set is contained in a null set.
(X , M, µ) is called complete if all the subsets of null sets are measurable.

Theorem

Any measure space (X , M, µ) can be expanded to a complete measure space (X , M̃, µ̃), where

M̃ = {A ∪ N : A ∈ M, and N is a subset of a µ-null set}

and
µ̃(A ∪ N) = µ(A) .

Example

Let X = {1, 2, 3}, M = {∅, {1}, {2, 3}, X } and µ(∅) = 0, µ({1}) = 4, µ({2, 3}) = 0, µ(X ) = 4.
Then M̃ = P(X ) and
µ̃({2}) = 0, µ̃({3}) = 0, µ̃({1, 2}) = 4, µ̃({1, 3}) = 4 ,

that is, µ̃ is a point mass at 1.

Szabó Tamás Analysis 2023. szeptember 27. 16 / 41

 EZ construction

Definition

Let H ⊂ P(X ) such that

∅ ∈ H,
[
X = Hn for some {Hn } ⊂ H.
n
Let ρ : H → [0, ∞] be a function on H such that ρ(∅) = 0.
For any A ∈ P(X ) let ( )
X [
ν(A) := inf ρ(Hn ) : Hn ∈ H and A ⊂ Hn .
n n

Then ν : P(X ) → [0, ∞] and

ν(∅) = 0,
(Monotonicity) for any
if A ⊂ B, then ν(A) ≤ ν(B) ,

(Countable sub-additivity) !
[ X
ν An ≤ ν(An ) .
n n

Definition

ν is called an outer measure on X .

Szabó Tamás Analysis 2023. szeptember 27. 17 / 41

 Not yet
Example

Let H = P(R) and define 

0, if H ∩ N = ∅,
ρ(H) =
min H ∩ N, otherwise.

Then the induced outer measure is 

0, if A ∩ N = ∅,
ν(A) =
1, otherwise.

Because
ν({1}) = 1, ν({2}) = 1, ν({1, 2}) = 1 ,

we have that
ν({1} ∪ {2}) = ν({1, 2}) = 1 ̸= 2 = ν({1}) + ν({2}) ,

although {1} and {2} are disjoint. Therefore, ν is not additive, can not be measure on H.
Let
M = {M ⊂ R : either N ⊂ M or M ∩ N = ∅} .

M is a σ-algebra,
ν is a complete measure on M,
for all M ∈ M
ρ(M) = ν(M) ,

for any M ∈ M
C
ν(H) = ν(H ∩ M) + ν(H ∩ M )

for all H ⊂ R.

Szabó Tamás Analysis 2023. szeptember 27. 18 / 41

 Outer measure induces a complete measure

Definition

Let ν be an outer measure on X . A set M ⊂ X is called ν-measurable if

C
ν(A) ≥ ν(A ∩ M) + ν(A ∩ M )

for all subsets A of X .

Theorem

(Carathéodory Theorem) The collection M of all ν-measurable sets is a σ-algebra, and ν is a complete measure on M.

Szabó Tamás Analysis 2023. szeptember 27. 19 / 41

 Still not yet

Example

Let
H = {R and all finite subsets of R}

and on H define 

#(H) − 1, if H ̸= ∅ is finite,


ρ(H) = 0, if H = ∅,



1, if H = R.

Then 
0, if A is countable,
ν(A) =
1, if A is uncountable.

The ν-measurable M σ-algebra is the countable/co-countable σ-algebra.

Clearly H ⊂ M, but
ρ({1, 2}) = 1 ̸= 0 = ν({1, 2}) .

Szabó Tamás Analysis 2023. szeptember 27. 20 / 41

 How far?
Because we need to extend the length of intervals to the Lebesgue measure, we want all intervals to be measurable and the Lebesgue measures of intervals
agree with their lengths. So next question we need to answer is: what requirements for H and ρ will guarantee H ⊂ M and ν = ρ on H ?

Szabó Tamás Analysis 2023. szeptember 27. 21 / 41

 Semi-algebra, premeasure

Definition

Let H ⊂ P(X ) such that

∅ ∈ H,
for all A, B ∈ H, A ∩ B ∈ H,
for all A ∈ H, AC is a finite disjoint union of members of H.
That is, H is the union of finite partitions of X which is closed under intersection.

Let ρ : H → [0, ∞] be a set function such that

ρ(∅) = 0,
n
[
(finite additivity) If {A1 , . . . , An } ⊂ H are pairwise disjoint and Ak ∈ H, then
k=1
 
n
[ n
X
ρ Ak  = ρ(Ak ) .
k=1 k=1

[
(monotonicity/countable sub-additivity) if A ∈ H and A ⊂ An for some {An } ⊂ H, then
n

X
ρ(A) ≤ ρ(An ) .
n

Szabó Tamás Analysis 2023. szeptember 27. 22 / 41

 Semi-algebra, premeasure

Remark

H is called semi-algebra and ρ is called a premeasure on H.

Theorem

Let ρ be a premeasure on a semi-algebra H and ν be the induced outer measure.Then

every set in H is ν-measurable, that is H ⊂ M,
ν = ρ on H.

Remark

Therefore, M(H), the σ-algebra generated by H is subset of M.

µ = ν
M(H)

is the induced measure on M(H).

If ρ is σ-finite, (i.e., X can be covered by countably many sets in H with finite premeasure), then µ is the unique extension of ρ on M(H).

Szabó Tamás Analysis 2023. szeptember 27. 23 / 41

 Lebesgue-Stieltjes Measures on R

Proposition

Let
H = {(a, b], (a, ∞) : −∞ ≤ a ≤ b < ∞} .

Then H is a semi-algebra.

Proposition

Let F : R → R be an increasing and right-continuous function. Define

ρ((a, b]) = F (b) − F (a), ρ((a, ∞)) = F (∞) − F (a) ,

where
F (−∞) = lim F (x), F (∞) = lim F (x) .
x→−∞ x→∞

Then ρ is a premeasure on H.

Szabó Tamás Analysis 2023. szeptember 27. 24 / 41

 Lebesgue-Stieltjes Measures on R

Proof

ρ(∅) = ρ((a, a]) = F (a) − F (a) = 0.

n
[
Let {I1 , . . . , In } ⊂ H are disjoint and I = Ik ∈ H.
k=1
Then I = (a, b] (for b = ∞, define (a, b] = (a, ∞)) and by relabeling we can assume Ik = (ak , bk ] where a1 = a, ak+1 = bk , for
k = 1, 2, . . . , n − 1, and bn = b. Hence

{Ik }

n
X Xn
ρ(Ik ) = (F (bk ) − F (ak )) = F (b) − F (a) = ρ(I ) .
k=1 k=1

Main steps:
Let In = (an , bn ] be a finite interval. Since F is increasing and right continuous, for any ϵ > 0 there exists a δn such that
ϵ
F (bn ) ≤ F (bn + δn ) < F (bn ) + .
2n
Since a closed, bounded set in R is a compact set, therefore, every open cover of it has a finite subcover.

Szabó Tamás Analysis 2023. szeptember 27. 25 / 41

 Lebesgue-Stieltjes Measures on R

Definition

The unique measure induced by the premeasure defined above for an increasing and right continuous function F is called the Lebesgue-Stieltjes measure
generated by F , and is denoted by µF and the σ-algebra of all µF -measurable sets is denoted by MF .

Remark

Since M(H) = B ⊂ MF , so µF is defined on B and

µF ((a, b]) = F (b) − F (a) .

Theorem

Every measure µ on B which is finite on bounded intervals is a Lebesgue-Stieltjes measure generated by the increasing and right continuous function

µ((0, x]), if x ≥ 0,
F (x) =
−µ((x, 0]), if x < 0.

Proposition

Let µF be a Lebesgue-Stieltjes measure generated by F . Then

for any A ∈ P(R)
µF (A) = inf{µF (U) : U is open and A ⊂ U} ;

for any B ∈ MF
µF (B) = sup{µF (K ) : K is compact and K ⊂ B} ;

Szabó Tamás Analysis 2023. szeptember 27. 26 / 41

 The difference between B and MF is not big

Remark

Clearly both the countable intersection of open sets (Gδ ) and the countable union of closed sets (Fσ ) are in B ⊂ MF .

Proposition

Then the following are equivalent

M ∈ MF .
For any ε > 0, there exists an open set U ⊃ M such that µF (U \ M) < ε.
M = D \ ND for some Gδ set D and a µF -null set ND .
For any ε > 0, there exists a closed set K ⊂ M such that µF (M \ K ) < ε.
M = S ∪ NS for some Fσ set S and a µF -null set NS .

Szabó Tamás Analysis 2023. szeptember 27. 27 / 41

 Lebesgue measure

Definition

The Lebesgue measure is the Lebesgue-Stieltjes measure generated by F (x) = x. The σ-algebra of all Lebesgue measurable sets is denoted by L, and the
Lebesgue measure is denoted by m.

Proposition

For any E ∈ L and for any r , s ∈ R, let

E + s = {x + s : x ∈ L}, rE = {rx : x ∈ L}.

Then E + s and rE are in L, and

m(E + s) = m(E ), m(rE ) = |r |m(E ).

Proposition

Any Lebesgue measurable set with m(E ) > 0 must contain a Lebesgue nonmeasurable set.

Szabó Tamás Analysis 2023. szeptember 27. 28 / 41

 Measurable Functions
Definition

Let (X , M) and (Y , N ) be measurable spaces. A function f : X → Y is called (M, N ) measurable if f −1 (E ) ∈ M for every E ∈ N .
Let (X , M) be a measurable space. A function f : X → R is called M measurable if f −1 (E ) ∈ M for every E ∈ B.
A function f : R → R is called Borel measurable if f −1 (E ) ∈ B for every E ∈ B.
A function f : R → R is called Lebesgue measurable if f −1 (E ) ∈ L for every E ∈ B.

Proposition

Every Borel measurable function is Lebesgue measurable.

A function f : R → R is Borel measurable (Lebesgue measurable) iff
f −1 (E ) ∈ B (L) for every E open set.
Every continuous function is Borel, and therefore, Lebesgue measurable.
A function f : R → R is Borel measurable (Lebesgue measurable) iff f.e.
{x : f (x) < a} ∈ B (L) for every a ∈ R.
Every monotone function is Borel, and therefore, Lebesgue measurable.
If f , g : R → R are measurable, then f + g , cf , fg are all measurable for any c ∈ R.
If fn : R → R∗ are measurable, then the functions
g1 = sup fn , g2 = inf fn , g3 = lim sup fn , g4 = lim inf fn
are all measurable.
If fn : R → R∗ are measurable and the limit of fn exists, then
f = lim fn
is also measurable.
χE is measurable iff E is measurable.

Szabó Tamás Analysis 2023. szeptember 27. 29 / 41

 Integrals of Nonnegative Measurable Simple Functions

Definition

Let (X , M, µ) be a measure space. For any nonnegative measurable simple functions

Xn
φ(x) = ck χE (x) ,
k
k=1
and for any E ∈ M, define
Z Xn

φ dµ = ck µ E ∩ Ek , where 0 · ∞ := 0 .
E k=1

This definition does not depend on the representation of the simple function.

Proposition

For any real number c ≥ 0 Z Z

cφ dµ = c φ dµ .
E E

Z Z Z
φ + ψ dµ = φ dµ + ψ dµ .
E E E

If φ ≤ ψ, then Z Z
φ dµ ≤ ψ dµ .
E E

The map Z
A → φ dµ
A
is a measure on M.

Szabó Tamás Analysis 2023. szeptember 27. 30 / 41

 Integrals of Nonnegative Measurable Functions

Proposition

Let (X , M) be a measurable space.

If f : X → [0, ∞] is measurable, then there are simple measurable functions

0 ≤ φ1 ≤ φ2 ≤ · · · ≤ f

such that φn → f pointwise, and φn → f uniformly on any set on which f is bounded.

For any measurable function f , there are simple measurable functions

0 ≤ |φ1 | ≤ |φ2 | ≤ · · · ≤ |f |

such that φn → f pointwise, and φn → f uniformly on any set on which f is bounded.

Definition

Let (X , M, µ) be a measure space, f : X → [0, ∞) be a nonnegative measurable function, and E ∈ M. Define

Z Z 
f dµ = sup φ dµ : 0 ≤ φ ≤ f , φ is simple measurable function .
E E

Szabó Tamás Analysis 2023. szeptember 27. 31 / 41

 Integrals of Nonnegative Measurable Functions

Theorem

For any nonnegative measurable function f , Z

f dµ = 0 iff f = 0 a.e.

Proof
R
Let f dµ = 0. Define
 1
En := x : f (x) > , E := {x : f (x) > 0} .
n
[
Then E1 ⊂ E2 ⊂ . . . and En = E . Since
n 1 Z Z
0 ≤ µ(En ) ≤ f dµ ≤ f dµ = 0
n En
we have µ(En ) = 0 for all n. Therefore,
µ(E ) = lim µ(En ) = 0.
n
R
Let f = 0 a.e., that is µ(E ) = 0. Thus E φ dµ = 0 for any simple measurable function φ, 0 ≤ φ ≤ f . Therefore,

Z Z Z
f dµ = f dµ + f dµ
E EC
Z  Z
= sup φ dµ : 0 ≤ φ ≤ f + 0 dµ
E EC
=0 .

Szabó Tamás Analysis 2023. szeptember 27. 32 / 41

 Monotone Convergence Theorem (MCT)

Proposition

Let 0 ≤ f1 ≤ f2 ≤ . . . be a sequence of increasing nonnegative measurable functions, and

f (x) = lim fn (x) .

n

Then Z Z
f dµ = lim fn dµ ≤ ∞.
n

Szabó Tamás Analysis 2023. szeptember 27. 33 / 41

 Monotone Convergence Theorem (MCT)

Proof
R R
Since fn ≤ f for all n, therefore fn dµ ≤ f dµ, hence

Z Z
lim fn dµ ≤ f dµ .
n

For all α with 0 < α < 1, and for any simple functions φ with 0 ≤ φ ≤ f , define

En := {x : fn (x) ≥ αφ(x)} .

Then [
E1 ⊂ E2 ⊂ . . . and En = X .
n

Thus Z Z Z Z
lim fn (x) dµ ≥ lim f (x) dµ ≥ α lim φ dµ = α φ dµ
n n E n n E
n n
holds for every 0 < α < 1. So Z Z
lim fn (x) dµ ≥ φ dµ ,
n

therefore, Z Z  Z
lim fn (x) dµ ≥ sup φ dµ : 0 ≤ φ ≤ f = f dµ .
n

Szabó Tamás Analysis 2023. szeptember 27. 34 / 41

 Monotone Convergence Theorem (MCT)

Corollary

Let {fn } be a sequence of nonnegative measurable functions. Then

Z X XZ
fn dµ = fn dµ .
n n

Let (X , M, µ) be a measure space and f ≥ 0 be measurable. The map Z

λ : A → f dµ
A
is a measure on M, and for any nonnegative measurable functions g
Z Z
g dλ = gf dµ .

Example

For an enumeration Q = {r1 , r2 , r3 , . . . } let 

1, if x ∈ {r1 , r2 , . . . , rn },
gn (x) =
0, otherwise.
Z
Each gn is clearly nonnegative measurable functions and gn dm = 0 ,
[a,b]
and clearly (gn ) is a bounded, increasing sequence, converging to the Dirchlet function d = limn gn
and Z Z Z
lim gn dm = d dm = 0 = lim g dm .
[a,b] n [a,b] n [a,b] n

Szabó Tamás Analysis 2023. szeptember 27. 35 / 41

 Fatou’s Lemma

Theorem

For any sequence of nonnegative measurable functions {fn }

Z Z
lim inf fn dµ ≤ lim inf fn dµ .
n n

Proof

For any i ≥ k,
inf fn ≤ fi .
n≥k

So, for all i ≥ k Z Z

inf fn dµ ≤ fi dµ .
n≥k

Therefore, Z Z
inf fn dµ ≤ inf fi dµ .
n≥k n≥k

Hence, by the MCT

Z Z Z
lim inf fn dµ = lim inf fn dµ = lim inf fn dµ
n k n≥k k n≥k
Z Z
≤ lim inf fn dµ = lim inf fn dµ .
k n≥k n

Szabó Tamás Analysis 2023. szeptember 27. 36 / 41

 Fatou’s Lemma

Example

Let 
n, if x ∈ (0, 1/n),
fn (x) =
0, otherwise.

Then fn → 0 pointwise and

Z Z Z Z
lim inf fn dm = 0 dµ = 0 < 1 = lim inf 1 = lim inf n dm = lim inf fn dm .
n n n (0,1/n) n

Let
1,

if x ∈ (0, n),
n
fn (x) =
0, otherwise.


Then fn ⇒ 0 uniformly and

Z Z Z 1 Z
lim inf fn dm = 0 dm = 0 < 1 = lim inf 1 = lim inf dm = lim inf fn dm .
n n n (0,n) n n

Szabó Tamás Analysis 2023. szeptember 27. 37 / 41

 Integration

Definition

A measurable function f is called integrable if Z

|f | dµ < ∞ .

Remark

That is, if
+ −
f := max{f , 0}, f := − min{f , 0} ,

then
f is integrable iff Z Z
+ −
f dµ < ∞ , f dµ < ∞ .

if f is integrable, then Z Z Z
+ −
|f | dµ = f dµ + f dµ .

Definition

For any f integrable function we define Z Z Z

+ −
f dµ := f dµ − f dµ .

The set of all integrable functions on X is denoted by L1 (X ).

Szabó Tamás Analysis 2023. szeptember 27. 38 / 41

 Dominated Convergence Theorem

Theorem

Let (fn ) be a sequence of measurable functions such that

fn → f a.e., and
there exists a g ∈ L1 such that |fn | ≤ g a.e. for all n.
Then fn , f ∈ L1 and Z Z Z
f dµ = lim fn dµ = lim fn dµ .
n n

Corollary

Suppose
{fn } ⊂ L1
XZ
and |fn | < ∞.
n

a.e. to a function in L1 and

P
Then f
n n converges
Z X XZ
fn dµ = fn dµ .
n n

Szabó Tamás Analysis 2023. szeptember 27. 39 / 41

 Riemann integral = Lebesgue integral

Theorem

If f is Riemann integrable on [a, b], then it is Lebesgue integrable on [a, b], and

Z b Z
f (x) dx = f dm .
a [a,b]

Remark

f is Riemann integrable on [a, b] if and only if it is continuous m-a.e. on [a, b].

Example

Let
X (−1)n−1
f (x) = χ[n−1;n) .
n=1 n

Then Z ∞
f (x) dx = ln 2 ,
0
Z
|f | dm = ∞ .
[0;∞)

Szabó Tamás Analysis 2023. szeptember 27. 40 / 41

 Lp Spaces

Definition

Let (X , M, µ) be a measure space and p > 0.

For any measurable function f , define
Z 1/p
p
∥f ∥p := |f | dµ .

The Lp space over (X , M, µ) is defined as the set of all equivalence classes of measurable functions f such that ∥f ∥p < ∞, where f and g are
equivalent if f = g a.e.
If X = N, M = P(N), and µN is the counting measure, then Lp (N, P(N), µN ) is denoted by ℓp , i.e., ℓp is the space of all sequences
{a = (an ); an ∈ R} such that X p
|an | < ∞
n

and ∥a∥p is defined as

!1/p
X p
∥a∥p := |an | .
n

Szabó Tamás Analysis 2023. szeptember 27. 41 / 41