MATH501 Analysis

Fall 2021-2022

Cheat Sheet

O§ul Can Yurdakul

Last Updated: 23/01/2022

These are the summaries I made from when I was taking the MATH501
Analysis course from Süleyman Ka§an Samurka³ during the Fall 2021-2022
term. We followed Real analysis for graduate students by Richard F. Bass,
and the section numbers are in accordance with the sections there.

1
Analysis - Cheat Sheet O§ul Can Yurdakul

Contents

1 Preliminaries 3
1.1 Notation & Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Some Undergraduate Maths . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Families of Sets 5
2.1 Algebras & σ -algebras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 The Monotone Class Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3 Measures 6
4 Construction of Measures 7
4.1 Outer Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4.2 Lebesgue-Stieltjes Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.3 Examples and Related Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
4.4 Nonmeasurable Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.5 Carathédory Extension Theorem . . . . . . . . . . . . . . . . . . . . . . . . . 11
5 Measurable Functions 11
5.1 Measurability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
5.2 Approximation of Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
6 The Lebesgue Integral 13
7 Limit Theorems 14
7.1 Monotone Convergence Theorem . . . . . . . . . . . . . . . . . . . . . . . . . 14
7.2 Linearity of the Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
7.3 Fatou's Lemma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
7.4 Dominated Convergence Theorem . . . . . . . . . . . . . . . . . . . . . . . . . 15
8 Properties of Lebesgue Integrals 15
8.1 Criteria for a Function to be Zero a.e. . . . . . . . . . . . . . . . . . . . . . . 15
8.2 An Approximation Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
10 Types of Convergence 16
11 Product Measures 16
11.1 Product σ -algebras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
11.2 The Fubini Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
12 Signed Measures 18
12.1 Positive and Negative Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
12.2 Hahn Decomposition Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . 19
12.3 Jordan Decomposition Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . 19
13 The Radon-Nikodym Theorem 19
13.1 Absolute Continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
13.2 The Main Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
13.3 Lebesgue Decomposition Theorem . . . . . . . . . . . . . . . . . . . . . . . . 20
15 Lp Spaces 20
15.1 Norms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
15.2 Completeness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
15.3 Convolutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

1. Preliminaries

1.1. Notation & Terminology

Complement: The complement of a subset A ⊂ X, denoted Ac , is given as

Ac = X − A and so A − B = A ∩ Bc

Symmetric Dierence: The symmetric dierence of two subsets A, B ⊂ X is given as

A△B = (A − B) ∪ (B − A)

Increasing Sets: For an indexed family of sets {Ai }, we write Ai ↑ to denote that

A1 ⊂ A2 ⊂ A3 ⊂ ...

We further write Ai ↑ A to mean

∞
[
A1 ⊂ A2 ⊂ A3 ⊂ ... & Ai = A
i=1

Decreasing Sets: For an indexed family of sets {Ai }, we write Ai ↓ to denote that

A1 ⊃ A2 ⊃ A3 ⊃ ...

We further write Ai ↓ A to mean

∞
\
A1 ⊃ A2 ⊃ A3 ⊃ ... & Ai = A
i=1

Positive & Negative Parts: For a real number x, the positive and negative parts of x, denoted x+
−
and x respectively, are dened as

x+ = max{x, 0} & x− = max{−x, 0}

This denition is used to dene the positive and negative parts of a real function.

Limsup & Liminf: The limsup and liminf of a sequence {ai } are dened as follows:

lim sup an = lim sup am = inf sup am

n→∞ n→∞ m≥n n m≥n

lim inf an = lim inf am = sup inf am

n→∞ n→∞ m≥n n m≥n

The limsup and liminf of a function f (x) are dened as follows:

lim sup f (x) = inf sup f (x)

x→a δ<0 |x−a|<δ

lim sup f (x) = sup inf f (x)

x→a δ<0 |x−a|<δ

Left and Right Limits: For a function with domain R, the left and right limits at x are denoted as
follows:
f (x+) = lim f (y) and f (x−) = lim f (y)
y→x+ y→x−

3 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

1.2. Some Undergraduate Maths

Metric Spaces: A set X is a metric space if there exists a function d:X ×X →R such that

1. d(x, y) ≥ 0
2. d(x, y) = d(y, x)
3. d(x, y) ≤ d(x, z) + d(z, y)

Open Ball: In a metric space X , the open ball around x ∈ X with radius r is dened as follows:

B(x; r) = {y ∈ X : d(x, y < r)}

Interior & Closure: For a subset A⊂X of a metric space X, the interior Ao of A and the
closure A of A are dened as follows:

Ao = {x ∈ X : ∃rx > 0 B(x; rx ) ⊂ A}

A = {x ∈ X : ∀r > 0 B(x; r) ∩ A ̸= ∅}
Opennes & Closedness: The set A is called open if A = Ao , closed if A = A.
Opennes in R: An open subset G ⊆ R can always be written as a disjoint and countable
union of open intervals (proven via Zorn's Lemma).

Open Cover & Compactness: An open cover of a subset K ⊂X is a non-empty collection

of open sets {Gα }α∈I such that
[
K⊂ Gα
α∈I

A subset K⊂X is called compact if every open cover of K has an nite subcover, i.e.

[ n
[
∀{Gα } with K⊂ Gα , ∃G1 , G2 , ..., Gn with K⊂ Gi
i=1

Heine-Borel Theorem: A subset K ⊆ R is compact if and only if it is closed and bounded.

Extreme Value Theorem: If K is compact and f : K → R is continuous, then there
exist xm , xM ∈ K such that

f (xm ) = inf f (x) and f (xM ) = sup f (x)

x∈K x∈K

Cauchy Sequences & Completeness: A sequence {an }n∈N in a metric space X is called
Cauchy if
∀ε > 0∃N > 0 such that ∀n, m > N d(an , am ) < ε
The verbal interpretation of this statement is that any tail of the sequence gets as close to
one another as we want. However, Cauchy sequences need not converge: We say that a
metric space X is complete if every Cauchy sequence in X has a limit in X.
Hausdor Property: Any two distinct points in a metric space can be separated by non-
intersecting open sets. This is called the Hausdor property.

Normed Spaces: X is a normed linear space if it is a vector space with a function ∥·∥ : X → [0, ∞)
satisfying the following properties:

1. ∥x∥ > 0 for all x ̸= 0X , and ∥x∥ = 0 if and only if x = 0X .

2. ∥cx∥ = |c|∥x∥ for all x∈X and c ∈ F, where for us F =R or F = C.

4 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

3. ∥x + y∥ ≤ ∥x∥ + ∥y∥, which is also called the triangle inequality.

A norm on a normed space X induces a metric as d(x, y) = ∥x − y∥. A metric on a vector space
induces a norm ∥x∥ = d(x, 0).

Partial Orders: A relation ≤ ⊂X ×X on X is a partial ordering if

1. For all x ∈ X , x ≤ x.
2. If x≤y and y ≤ x, then x = y.
3. If x≤y and y ≤ z, then x ≤ z.

A partially-ordered set is called a poset in short.

Total Order: A partial order is called a total order if any two elements can be compared, so
for all x, y ∈ X we have either x≤y or y ≤ x.
A totally-ordered subset of a poset is called a chain.

Zorn's Lemma: For any non-empty poset X, if any non-empty chain has an upper bound in
X, then there is at least one maximal element in X.

Continuity of Increasing Functions: Suppose f : R → R is an increasing function. Then both

f (x+) and f (x−) exist, and the set of x's where f is not continuous must be countable. (Re-
member: The discontinuities can only be jump discontinuities!)

2. Families of Sets

2.1. Algebras & σ-algebras

Algebra & σ-algebra: An algebra is a collection A of subsets of X such that

1. ∅∈A and X ∈ A,
2. If A ∈ A, then Ac ∈ A,
Sn Tn
3. If A1 , A2 , ..., An ∈ A, then i=1 Ai , i=1 Ai ∈ A.

A is further a σ -algebra if we have not only nite but also countable unions and intersections:
S∞ T∞
4. If A1 , A2 , ..., ∈ A, then i=1 Ai , i=1 Ai ∈ A.

The pair (X, A) is called a measurable space, and A∈A is called measurable if A ∈ A.

Intersection of σ-algebras:
T
If {Aα } is a collection of σ -algebras over X , then α Aα is again
a σ -algebra.
Generated σ-algebra: For C ⊂ P(X), the σ -algebra generated by C, denoted σ(C), is dened
as \
σ(C) = {Aα : Aα is a σ -algebra containing C}
With this denition, we have the following two properties:

- If C1 ⊂ C2 , then σ(C1 ) ⊂ σ(C2 ).

- σ(σ(C)) = σ(C)

Borel σ-algebra: For a metric space X and G the collection of open sets of X induced by its metric,
the σ -algebra B = σ(G) is called the Borel σ -algebra on X. The sets A ∈ B are called Borel-
measurable.

5 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

Generators of B on R: For X = R, B is generated by the following collections:

- C1 = {(a, b) : a, b ∈ R}
- C2 = {[a, b] : a, b ∈ R}
- C3 = {(a, b] : a, b ∈ R}
- C4 = {(a, ∞) : a ∈ R}

2.2. The Monotone Class Theorem

Monotone Class: A monotone class is a collection M of subsets of X such that

1. If A1 , A2 , A3 , ... ∈ M and Ai ↑ A, then A ∈ M,

2. If A1 , A2 , A3 , ... ∈ M and Ai ↓ A, then A ∈ M.

The intersection and generative properties stated above for σ -algebras hold for monotone classes
as well.

Monotone Class Theorem: For an algebra A0 , let A = σ(A0 ) and M be the monotone class
generated by A0 . Then we have
A=M

3. Measures

Measure: Let X, A be a measurable space. A measure on (X, A) is a function µ : A → [0, ∞] such

that

1. µ(∅) = 0, and
S∞ P∞
2. If Ai ∈ A for i = 1, 2, 3, ... are pairwise-disjoint, then µ( i=1 Ai ) = i=1 µ(Ai ). This
property is called countable additivity. If the identity holds for a nite set instead of a
countable one, then it is called nite additivity.

The triple (X, A, µ) is then called a measure space.

Properties: For a xed measure space (X, A, µ), the followings hold:

1. If A, B ∈ A with A ⊆ B , then µ(A) ≤ µ(B).

S∞ P∞
2. If Ai ∈ A for i = 1, 2, 3, ..., then µ ( i=1 Ai ) ≤ i=1 µ(Ai ).
3. Suppose Ai ∈ A and Ai ↑ A. Then µ(A) = limi→∞ µ(Ai ).

4. Suppose Ai ∈ A and Ai ↓ A AND µ(A1 ) < ∞. Then µ(A) = limi→∞ µ(Ai ).

Some Useful Constructions:

Countable union to disjoint countable union: Suppose we have Ai ∈ A for all i
S ∞
and A= i=1 Ai . Then construct the sets Bi in the following way:

n−1
[
B 1 = A1 and Bn = An − (An−1 ∪ ... ∪ A1 ) = An − Ai
i=1

This way, Bi are disjoint, measurable sets, satisfying

n
[ n
[ ∞
[ ∞
[
Ai = Bi and Ai = Bi
i=1 i=1 i=1 i=1

6 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

Countable
S union to increasing sequence: Suppose we have Ai ∈ A for all i and
∞
A= i=1 Ai . Then construct the sets Bi in the following way:

n
[
B 1 = A1 and Bn = Ai
i=1

This way, Bi are measurable sets satisfying

n
[ n
[
Ai = Bi and Bi ↑ A
i=1 i=1

Countable intersection to decreasing sequence: Suppose we have Ai ∈ A for all

T ∞
i and A= i=1 Ai . Then construct the sets Bi in the following way:

n
\
B 1 = A1 and Bn = Ai
i=1

This way, Bi are measurable sets satisfying

n
\ n
[
Ai = Bi and Bi ↓ A
i=1 i=1

Finiteness & σ-Finiteness: A measure µ is called a nite measure if µ(X) < ∞. A measure
µ S∞ σ -nite
is called if there exists some countable collection of sets Ei ∈ A with µ(Ei ) < ∞
and i=1 Ei = X .
Null Sets & Completeness: For a measure space (x, A, µ), a subset N ⊂ X , is called a null
set if there exists some A∈A with N ⊆A and µ(A) = 0. Notice that a null set need not
be measurable.
If a measure space (X, A, µ) contains all null sets, it is said to be complete.

Completion: The completion of a σ -algebra A is the smallest complete σ -algebra A that

contains A, such that (X, A, µ) is a complete measure space, which is obtained by
setting
A = σ(A ∪ N ) and µ A
=µ
where N is the collection of all null sets.

4. Construction of Measures

4.1. Outer Measures

Outer Measure: For a set X, an outer measure on X is a function µ∗ : P(X) → [0, ∞] such that

1. µ∗ (∅) = 0,
2. If A ⊂ B, then µ∗ (A) ≤ µ∗ (B), and
S∞ P∞
3. For all countable selection of subsets A1 , A2 , A3 , ... ∈ P(X), µ∗ ( i=1 Ai ) ≤ i=1 µ
∗ (A ).
i

A set N is a null set with respect to µ∗ if µ∗ (N ) = 0.

7 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

Length Function & Induced Outer Measure: C is a collection of subsets of X , such that
Suppose
S
∅ ∈ C and there exist E1 , E2 , E3 , ... ∈ C with Ei = X . Suppose further that there exists a
∗
function l : C → [0, ∞] with l(∅) = 0. Dene µ : P(X) → [0, ∞] as
(∞ ∞
)
X [
µ∗ (A) = inf l(Ci ) : Ci ∈ C and A ⊂ Ci
i=1 i=1
∗
This µ is then an outer measure.

Outer Measurability: For an outer measure µ∗ , a set A⊆X is said to be µ∗ -measurable if for all
E ⊆ X, we have
µ∗ (E) = µ∗ (E ∩ A) + µ∗ (E ∩ Ac )
Induced Complete Measure: For an outer measure µ∗ , the collection

A = {A ⊆ X : A is µ∗ -measurable }
is a σ -algebra and µ = µ∗ A
is a complete measure.

4.2. Lebesgue-Stieltjes Measures

Lebesgue-Stieltjes Measures: Let X = R, C = {(a, b] : a, b ∈ R} and let α:R→R be a right-
continuous and increasing function. Dene a length function as
(
α(b) − α(a) , a < b
l((a, b]) =
0 ,a ≥ b
This length function and the collection C satisfy the conditions above, and so the following is a
valid outer metric:
(∞ ∞
)
X [
∗
m (A) = inf l(Ci ) : Ci ∈ C and A⊂ Ci
i=1 i=1
∗
and so The collection of m -measurable functions, say M, is a σ -algebra, and m = m∗ is a
M
complete measure. This measure m is called the Lebesgue-Stieltjes measure (associated to the
given α function).

Lebesgue Measure: In the special case where α(x) = x, the induced σ -algebra is called the
Lebesgue σ -algebra, and the measure is called the Lebesgue measure.

Useful Lemma: Let Jk = (ak , bk ) for k = 1, 2, ..., n be a nite collection of open intervals covering a
closed interval [c, d], and α a right-continuous increasing function. Then we have
n
X
α(d) − α(c) ≤ α(bk ) − α(ak )
i=1

Agreement on the Generators: Let e, f ∈ R with e ≤ f. Then the outer measure m∗ agrees with
the length function l, so
m∗ ((e, f ]) = l((e, f ]) = α(f ) − α(e)
Relation to the Borel σ-algebra: Every set in the Borel σ -algebra is m∗ -measurable, in other
words, the Borel σ -algebra is contained in any σ -algebra associated to a Lebesgue-Stieltjes (or
Lebesgue) measure.
B⊆L
This also implies that there might exist (and indeed there does exist) some set that is Lebesgue-
Stieltjes (or Lebesgue) measurable but not Borel measurable.

8 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

4.3. Examples and Related Results

Cantor Set: Dene the following sequence of sets:

C0 = [0, 1]
C1 = C0 − (1/3, 2/3) = [0, 1/3] ∪ [2/3, 1]
C2 = C1 − ((1/9, 2/9) ∪ (7/9, 8/9)) = [0, 1/9] ∪ [2/9, 3/9] ∪ [6/9, 7/9] ∪ [8/9, 1]
...
T∞
and set C= i=0 Ci . C is a closed and bounded set constructed by removing the middle third
of each interval iteratively, and it can alternatively be expressed as

C = {x ∈ [0, 1] : x = (0, a1 a2 a3 a4 · · · )3 such that ai ∈ {0, 2}}

| {z }
base-3 expression

Now some facts about C:

- From the alternative description, we see that C is in bijection with the complete [0, 1] as
well: Just set 2's to 1's and read the number in base 2. So, the set is uncountable.

- Every point in C is a limit point, so for all x ∈ C , there exists some sequence {ai } ⊆ C −{x}
such that ai → x. We can also see this from the alternative denition: For all n, take the
sequence element an as the rst n numerals after the ternary point if the sequence contains
innitely many 2's, namely an = (0, a1 a2 . . . an 00 . . . )3 . If it contains nitely many 2's with
the nal 2 at, say ak , set the k -th numeral to 0, add a 2 after the k -th numeral with some
zeroes in between: an = (0, a1 a2 . . . ak−1 0 . . . 020 . . . )3 . It's easy to see that this sequence
satises the limit point condition.

- C contains no interval. If it did, it's interior would contain a middle-third of some interval
in the construction and that would have been removed. This holds for any interval, and the
inside of each interval is iteratively removed. On the other hand, we can also see it via the
base-3 representation: If it did contain an interval, say [x, y] where x, y ∈ C and x ̸= y , it
would also contain a number containing a 1 numeral in its base-3 representation necessarily,
which is a contradiction to the denition of C.

Now, what is the Lebesgue measure of C? We see that Cn ↓ C and m(C0 ) = 1 < ∞, so we can
say m(C) = lim(Cn ). As m(Cn ) = (2/3)n , this means that m(C) = lim(2/3)n = 0. So, we have
found an uncountable set with measure 0!

Cantor-Lebesgue Function: Dene a function f0 : [0, 1] → [0, 1] using the removed parts of
the Cantor set:




1/2 if x ∈ (1/3, 2/3)

1/4 if x ∈ (1/9, 2/9)
f0 (x) =



3/4 if x ∈ (7/9, 8/9)

· · ·

and using f0 , dene fCL : [0, 1] → [0, 1] as

(
inf{f0 (y) : y ≥ x, y ̸∈ C} if x ̸= 1
fCL f (x) =
1 if x=1

Now, we have fCL [0,1]−C

= f0 and f is always increasing, so fCL can only have jump

discontinuities. That it cannot have either: Say it is discontinuous over an interval. This

9 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

interval must contain a number of the form k/2n , which is in the image of f0 , and hence that
of fCL , which yields a contradiction. So, f is an increasing continuous function, and as it
is piecewise-constant on [0, 1] − C , it can only increase on C.
Generalized Cantor Set: Instead of always removing the middle-third, we can also choose
to remove the middle-ai  section at each i-th step, where ai ∈ (0, 1). For example, if we
choose ai = 1/4 for all i, the resultant set will have measure 1/2, and again will be closed,
uncountable, without intervals and every point in it will be a limit point.

Playing inside 0 ≤ x ≤ 1: For a Lebesgue-measurable subset A ⊆ [0, 1], we have

1. For all ε > 0, there exists an open G⊇A such that m(G − A) < ε.
2. For all ε > 0, there exists a closed F ⊆A such that m(A − F ) < ε.
3. There exists some H⊇A such that Gi ↓ H with each Gi open and m(H − A) = 0.
4. There exists some J ⊆A such that Fi ↓ H with each Fi closed and m(A − K) = 0.
Gδ and Fσ sets: A set A is called a Gδ set if it is a countable intersection of open sets. It is called a
Fσ set if it is a countable union of closed sets.

4.4. Nonmeasurable Sets

Non-outer Measurable Sets: m∗
be the outer measure associated to l : C = {(a, b] : a, b ∈
Let
R} → [0, ∞], l((a, b]) = b − a whenever (a, b] ̸= ∅. Then the collection of m∗ -measurable sets is
∗
not equal to P(R). This means that there must exist some subset of R that is not m measurable.
Dene an equivalence relation ∼ on [0, 1] as follows:

x ∼ y ⇐⇒ x − y ∈ Q

Let A be a set of representatives from [0,1]/∼. The claim is that A is not Lebesgue-measurable.
Assume to the contrary that it is. Then because A ⊆ [0, 1], we can write the following:

[
[0, 1] ⊆ (A + q) ⊆ [−1, 2]
q∈[−1,1]∩Q

The rst subset relation is true because we are shifting the set of representatives via all the
possible rational shifts that can occur within [0, 1], and the second subset relation is easy to see.
Then, because all the shifts of A under dierent q values are disjoint, we have
X
1 = m∗ ([0, 1]) ≤ m∗ (A + q) ≤ m∗ ([−1, 2]) = 3
q∈[−1,1]∩Q

Now, because a shift by any q preserves the length of the set, it must also preserve the outer
measure m∗ of the set, so

X
1 = m∗ ([0, 1]) ≤ m∗ (A + q) ≤ m∗ ([−1, 2]) = 3
| {z }
q∈[−1,1]∩Q =m∗ (A)

X
1≤ m∗ (A) ≤ 3
q∈[−1,1]∩Q

Now, what is m∗ (A)? It must be either 0, or positive. If it is positive, the sum yields 0, which
is strictly smaller than 1, a contradiction. If it is any positive real, then the sum is unbounded,
which contradicts the second inequality. Thus, A has no outer measure, and thus is not Lebesgue-
measurable.

10 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

4.5. Carathédory Extension Theorem

Measure on an Algebra: For an algebra (but not necessarily a σ -algebra) A0 , a function l : A0 →
[0, ∞] is a measure if

1. l(∅) = 0, and

Ai ∈ A for i = 1, 2, 3, ... are pairwise-disjoint with ∞

S S∞
2. If
P∞ i=1 Ai ∈ A0 , then l ( i=1 Ai ) =
i=1 l(Ai ). (Notice that the inclusion of the union is forced, not due to the properties of
A0 .)

Carathédory Extension Theorem: Suppose A0 is an algebra and l : A0 → [0, ∞] is a measure as

above. Let (∞ ∞
)
X [
∗
µ (A) = inf l(Ai ) : Ai ∈ A0 and A⊂ Ai
i=1 i=1

Then

1. µ∗ is an outer measure.

2. µ∗ (A) = l(A) whenever A ∈ A0 .

3. Every set in A0 and every µ∗ -null set is µ∗ -measurable.
4. If
S∞l is σ -nite, meaning there exists some Ki sets each with nite measure such that

i=0 Ki = R, then l uniquely extends to σ(A0 ).

5. Measurable Functions

5.1. Measurability
Measurability of Functions: Let (X, A) be a measure space. A function f : X → R is measurable
or A-measurable if the inverse image of a Borel-measurable set is A-measurable. Equivalently,

- f −1 (B) ∈ A for all B ∈ B,

- f
−1 ((a, b)) ∈A for all a, b ∈ R,
- f
−1 ((a, b]) ∈A for all a, b ∈ R,
- f −1 ([a, b)) ∈ A for all a, b ∈ R,

or the most usefully,

- f −1 ((a, ∞)) ∈ A for all a, b ∈ R,

Typical Examples: Constant functions f (x) = c are always measurable.

The characteristic function χA of a set A is measurable if and only if A is measurable, i.e.
A ∈ A.
If the σ -algebra A contains the open subsets of X , then any continuous function f :X→R
is measurable.

Operations on Measurable Functions: Given that the functions f, g : X → R are both

measurable, so are the following functions:

cf where c∈R f +g f −g fg max{f, g} min{f, g}

11 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

Sequences of Measurable Functions: If fi 's are real-valued measurable functions for all i=
1, 2, ..., then the following functions are also measurable:

inf fi sup fi lim inf fi lim sup fi

i i i→∞ i→∞

Lebesgue- and Borel-Measurability of Functions: X be a metric space, BX the Borel σ -

Let
algebra on X (σ -algebra generated by the open sets of X , B denotes the σ -algebra on R) and L
∗ ∗
be the Lebesgue σ -algebra on R (σ -algebra of all m -measurable sets where m is the induced
outer metric of α(x) = x). If

- f : (X, BX ) → (R, B) is measurable, then f is called Borel-measurable.

- f : (X, L) → (R, B) is measurable, then f is called Lebesgue-measurable.
Notice how the range σ -algebra is always Borel.

Monotonity: If f : R → R is monotone, then f is Borel-measurable.

A Lebesgue-measurable but not Borel-measurable set: Dene the function F using fCL
the Cantor-Lebesgue function as follows:

F (y) = inf{x ∈ [0, 1] : y ≤ fCL (x)}

This F : [0, 1] → C is strictly increasing, where C is the Cantor set.
Now take A as the set we have previously constructed via the equivalence class represen-
tatives, which is not Lebesgue-measurable, and consider its image F (A) ⊆ C . Because
m(C) = 0 and the Lebesgue measure is complete, any subset of C is measurable and has
measure 0. In particular, F (A) is Lebesgue measurable with m(F (A)) = 0.
Now assume that F (A) is Borel-measurable. Because F is strictly increasing, it is Borel-
measurable (being monotone). Because we assume F (A) to be Borel-measurable, we would
conclude that F
−1 (F (A)) = A is Borel-measurable as well. However, we also know that

B ⊆ L and that A ̸∈ L, so A cannot be in B , which yields the contradiction.

Therefore we conclude that
B⊊L

5.2. Approximation of Functions

Simple Functions: A simple function on a σ -measurable space (x, A) is of the form
n
X
s(x) = ai χEi (x) where ai ∈ R, Ei ∈ A
i=1
Being a linear sum of characteristic functions, simple functions are always measurable.

Approximation via Simple Functions: Suppose f is a non-negative, measurable function.

Then, there exists a sequence of non-negative and measurable simple functions sn which
increase to f, meaning for all x ∈ X, we have sn (x) ↗ f (x).
Proof via Pointing a Finger: Dene the following sets using f:
 
i−1 i
Ain = x : n ≤ f (n < n ) for n = 1, 2, ... and i = 1, 2, ..., n2n
2 2
Bn = {x : n ≤ f (x)} for n = 1, 2, ...
and dene
n2n
X i−1 χAin (x) + nχBn
sn (x) =
2n
i=1
This approximation satises the requirement above.

12 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

6. The Lebesgue Integral

Lebesgue Integral ... Let (X, A, µ) be a measure space.

... of a Simple Function: For a non-negative simple function s(x) = ni=1 ai χEi (x), dene its
P
Lebesgue integral as follows:
Z n
X
f dµ = ai µ(Ei )
i=1
This denition is indeed well-dened. Here, we use the convention that if ai = 0 and
µ(Ei ) = ∞ for some i, we say ai µ(Ei ) = 0.
... of a Positive Measurable Function: If f is any measurable function, dene its Lebesgue
integral as
Z Z 
f dµ = sup s dµ : 0 ≤ s ≤ f, s is a measurable non-negative simple function

... of a Measurable Function: For a general measurable function f, dene the positive and
negative parts of f as

f + = max{f, 0} and f − = max{−f, 0}

These are both positive and measurable functions, so their Lebesgue integrals are as dened
above. Unless both of their individual integrals are innity, dene the integral of f as
Z Z Z
f dµ = f + dµ − f − dµ

If both of these integrals are innity, then the integral of f is not dened.

... over a set A: The integral of f over a set A is dened as

Z Z
f dµ = f χA dµ
A
Integrability: If f is measurable and we have
Z Z Z
|f | dµ = f +
dµ + f − dµ < ∞

then f is called integrable.

Properties of the Lebesgue Integral:

Squeezing Between Constants: If f is a measurable function with 0 ≤ a ≤ f (x) ≤ b for all
x∈X for non-negative a, b ∈ R and the measure of the whole space is nite, i.e. µ(X) < ∞,
then Z Z Z
aµ(X) = a dµ ≤ f dµ ≤ b dµ = bµ(X)

Compared Functions: If f and g are measurable and integrable functions with 0 ≤ f (x) ≤
g(x) for all x ∈ X, then
Z
f dµ ≤ g dµ

Homogenity over Positive Constants: If f is measurable and integrable, and c ≥ 0 is a

constant, then Z Z
cf dµ = c f dµ

Integration over Sets of Measure 0: For a subset A⊆X with µ(A) = 0 and a non-negative
measurable function f, we have
Z Z
f dµ = f χA dµ = 0
A

13 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

7. Limit Theorems

7.1. Monotone Convergence Theorem

Monotone Convergence Theorem (MCT): Let {fn } be a sequence of

- non-negative,

- measurable,

- increasing functions, i.e. f1 (x) ≤ f2 (x) ≤ f2 (x) ≤ . . . for all x,

- with limit limn→∞ fn (x) = f (x) for a measurable function f.
Then, we have
Z Z Z Z
fn dµ −−→ f dµ i.e. lim fn dµ = lim fn dµ
n→∞ n→∞

Note that this result holds a.e.

7.2. Linearity of the Integral

Basic Linearity: Let f and g be non-negative and measurable OR only integrable functions. Then
Z Z Z
f +g = f+ g

The two functions being only measurable is not enough.

Improved Properties of the Lebesgue Integral: The previously proved properties of the Lebesgue
integral are now extended to be valid for not only non-negative numbers/functions:

Squeezing Between Constants: If f is a measurable function with a ≤ f (x) ≤ b for all x ∈ X

for ANY a, b ∈ R and the measure of the whole space is nite, i.e. µ(X) < ∞, then
Z Z Z
aµ(X) = a dµ ≤ f dµ ≤ b dµ = bµ(X)

Compared Functions: If f and g are measurable and integrable NOT NECESSARILY NON-
NEGATIVE functions with ≤ f (x) ≤ g(x) for all x ∈ X , then
Z
f dµ ≤ g dµ

Homogenity over Positive Constants: If f is measurable and integrable, and c is ANY con-
stant, then Z Z
cf dµ = c f dµ

Integration over Sets of Measure 0: For a subset A⊆X with µ(A) = 0 and ANY measur-
able function f, we have Z Z
f dµ = f χA dµ = 0
A

Integral of a Function Series: Suppose fn are a sequence of measurable and non-negative func-
tions. Then
∞
Z X ∞ Z
X
fn = fn
n=1 n=1

14 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

Triangle Inequality: For an integrable function f, we have

Z Z
f ≤ |f |

7.3. Fatou's Lemma

Fatou's Lemma: Suppose fn 's form a sequence of measurable functions. Then
Z Z
lim inf fn ≤ lim inf fn
n→∞ n→∞

A typical use:
R
R Suppose fn → f and we have supn |fn | ≤ K < ∞. Then, |fn | → |f | and
|f | ≤ K .
Proof.

Z Z Z Z
|f | = lim inf |fn | ≤ lim inf |fn | (Fatou's Lemma)
n→∞ n→∞
Z
≤ sup |fn | ≤ K
n

7.4. Dominated Convergence Theorem

Dominated Convergence Theorem (DCT): fn are measurable
Suppose with fn → f , and there
exists a non-negative and integrable function g with |fn | ≤ g . Then
Z Z Z
lim fn = lim fn = f
n→∞ n→∞

The g here is said to dominate the function sequence fn .

Note that this result holds a.e.

8. Properties of Lebesgue Integrals

8.1. Criteria for a Function to be Zero a.e.

Zero Integral of ...
... a Non-Negative Function: Suppose f ≥ 0 is measurable with
Z
f dµ = 0

Then f =0 a.e.

... an Integrable Function on A ∈ A: Suppose f is an integrable function, with

Z
f dµ = 0
A

for any measurable set A. Then f =0 a.e.

15 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

... in the Lebesgue Measure: Let m be the Lebesgue measure and a ∈ R. Suppose f :R→R
is integrable, with Z x Z
f (y) m(dy) = f dm = 0
a [a,x]

for all x ∈ R. Then f =0 a.e.

8.2. An Approximation Result

Continuous Approximation: Let f be a Lebesgue-measurable and integrable function on R, and
ε > 0. Then, there exists a continuous function
R g with compact support supp(g) = {x : g(x) ̸= 0}
such that |f − g| ≤ ε.
Any integrable function can be approximated with a continuous function with compact support.

10. Types of Convergence

Convergence ... Let fn be a sequence of functions, in a measure space with measure µ.

... a.e.: fn is said to converge to some f a.e. if if

µ({x : lim fn (x) ̸= f (x)}) = 0

n→∞

... in Measure: fn is said to converge in measure to f if for all ε > 0, if

lim µ({x : |fn (x) − f (x)| > ε}) = 0

n→∞

... in Lp : Let 1 ≤ p < ∞. Then f is said to converge to f in Lp if

Z
lim |fn − f |p dµ = 0
n→∞

In measure and a.e. Convergence under Finite Measures: Suppose µ is a nite measure. Then

1. If fn → f a.e., then fn → f in measure.

2. If fn → f in measure, then there is a subsequence fnj such that fnj → f a.e.

Chebyshev's Inequality: Let 1≤p<∞ and a > 0. Then

|f |p dµ
R
µ({x : |f (x)| ≥ a}) ≤
ap

Lp and in Measure Convergence: If fn → f in Lp , then fn → f in measure.

11. Product Measures

11.1. Product σ-algebras

Suppose (X, A, µ) and (Y, B, ν) are two measure spaces.

16 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

Product σ-algebra: Dene the product σ -algebra as

A × B = σ({A × B : A ∈ A, B ∈ B})

A×B where A ∈ A, B ∈ B is called a measurable rectangle.

Sections: Given some E ⊂X ×Y, dene the following functions:

sx (E) = {y ∈ Y : (x, y) ∈ E}∀x ∈ X (x-section)

ty (E) = {x ∈ X : (x, y) ∈ E}∀y ∈ Y (y -section)

Measurable Sections: Let E ∈ A × B. Then,

- sx (E) ∈ B for all x ∈ X, and

- ty (E) ∈ A for all y ∈Y.

Measurable Univariate Functions: Suppose f is a A × B -measurable function, also called jointly

measurable. Then,

- k(y) = f (x∗ , y) is B -measurable for all x∗ ∈ X , and

- h(x) = f (x, y ∗ ) is A-measurable for all y∗ ∈ Y .

Measure of Sections: Suppose µ and ν are σ -nite, and let E ∈ A × B. Now dene

h(x) = ν(sx (E)) and k(y) = µ(ty (E))

Then h is A-measurable and k is B -measurable, with

Z Z
h(x)µ(dx) = k(y)ν(dy)

These two integrals correspond to the following double integral:

ZZ ZZ
χE (x, y)µ(dx)ν(dy) = χE (x, y)ν(dy)µ(dx)

Product Measure: Through this fact, we can dene the product measure µ × ν:
Z Z
µ × ν(E) = h(x)µ(dx) = k(y)ν(dy) ∀E ∈ A × B

For measurable a measurable rectangle A×B gives the intuitive µ × ν(A × B) = µ(A)ν(B)
result.

Completeness of the Product Measure: Even if µ and ν are individually complete, the
product measure µ×ν may be incomplete: Take µ = ν = m and A × 0. A × 0 is a
m × m-null L × L-measurable.
set, but is not

n-dimensional Lebesgue Measure: The n-dimensional Lebesgue measure is dened as

n
the completion of the measure space (R , L × · · · × L, m × · · · × m).

11.2. The Fubini Theorem

Fubini-Tonelli Theorem: Suppose f : X ×Y → R is a jointly measurable function, and µ and ν
are σ -nite measures. If either

a. f is non-negative, or

17 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

RR
b. f is integrable, i.e. |f | d(µ × ν) < ∞

then the followings hold:

1. Measurable Univariates: The functions

y 7→ f (x∗ , y)∀x∗ ∈ X and x 7→ f (x, y ∗ )∀y ∗ ∈ Y

are measurable within their own spaces.

2. Measurable Marginals: The functions

Z Z
h(x) = f (x, y)ν(dy) and x 7→ k(y) = f (x, y)µ(dx)

are measurable within their own spaces.

3. Order of Integration: The order of integration does not matter, i.e.

ZZ Z Z 
f (x, y) d(µ × ν)(x, y) = f (x, y) µ(dx) ν(dy)
Z Z 
= f (x, y) ν(dy) µ(dx)

12. Signed Measures

12.1. Positive and Negative Sets

Signed Measure: Let A be a σ -algebra. A signed measure on A is a function µ : A → (−∞, ∞)
such that

- µ(∅) = 0, and

- if A1 , A2 , · · · ∈ A are disjoint, then

∞ ∞
!
[ X
µ Ai = µ(Ai )
i=1 i=1

where the series converges absolutely if the sum is nite.

We require absolute convergence so the order of summation does not matter.

The measures we have seen so far are sometimes called as positive measures.

Positive, Negative and Null Sets: Let µ be a signed measure. A set A is said to be

positive if A ∈ A and µ(B) ≥ 0 for all B ∈ A with B ⊆ A.

negative if A ∈ A and µ(B) ≤ 0 for all B ∈ A with B ⊆ A.
a null set if µ(B) = 0 for all B ∈ A with B ⊆ A.
A null set for a positive measure is also a null set in this sense.

Countable Union: By a similar reasoning done in the (previous) positive measure case, we
again have
∞ n
! !
[ [
µ Ai = lim µ Ai
n→∞
i=1 i=1

Negative Subsets: Let µ be a signed measure, and E ∈ A. If µ(E) < 0, then there exists a
measurable and negative subset F ⊆ E.

18 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

12.2. Hahn Decomposition Theorem

Hahn Decomposition Theorem: Let µ be a signed measure. The followings then hold:

Positive/Negative Partition: There exists P, N ⊆ X with P ⊔ N = X, where P is a positive

set and N is a negative set.

Almost Uniqueness: If P′ and N′ are another such pair, then P △P ′ = N △N ′ is a null set
with respect to µ.
Measures of the Partition: If µ is not a positive measure, then µ(N ) < 0.
If −µ is not a positive measure (negative measure), then µ(P ) > 0.

Support of a Measure: Given a positive measure µ, we say that µ is supported on a set Y ⊆X if

for all A∈A with A⊆Yc we have µ(A) = 0.

Mutually Singular Measures: Two positive measures µ and ν are said to be mutually singular if
there exist A, B ⊆ X with A ⊔ B = X, µ is supported on A and ν is supported on B . In that
case, we write µ ⊥ ν .

12.3. Jordan Decomposition Theorem

Jordan Decomposition Theorem: Let µ be a signed measure on (X, A). Then, there exist unique
+
positive measures µ and µ− such that

µ = µ + − µ− and µ+ ⊥ µ−

These two positive measures are dened as

µ+ (A) = µ(A ∩ P ) and µ− (A) = −µ(A ∩ N )

for all A ∈ A, where P and N are the positive and negative sets, respectively, from the Hahn
decomposition of µ.

13. The Radon-Nikodym Theorem

13.1. Absolute Continuity

Absolute Continuity: A measure ν is said to be absolutely continuous with respect to a measure µ
if
µ(A) = 0 ⇒ ν(A) = 0
denoted ν ≪ µ.

For Finite Measures: If ν is a nite measure, then ν≪µ if and only if

∀ε > 0 ∃δ > 0 such that ∀A ∈ A µ(A) < δ ⇒ ν(A) < ε

19 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

13.2. The Main Theorem

Finite Positive Measures: Let µ and ν be nite positive measures on a measurable space (X, A).
Then either µ ⊥ ν or there exists ε > 0 and G ∈ A such that µ(G) > 0 and G is positive for the
signed measure ν − εµ.
Radon-Nikodym Theorem: Let µ and ν are positive σ -nite measures on (X, A) with ν ≪ µ.
Then, there exists an A-measurable function f such that
Z
ν(A) = f dµ
A

for all A ∈ A. If g is another such function, then f = g µ-a.e.

13.3. Lebesgue Decomposition Theorem

Lebesgue Decomposition Theorem: Let µ and ν be nite positive measures on a measurable space
(X, A). Then, there exist positive measures λ and ρ such that
ν =λ+ρ and ρ≪µ

15. Lp Spaces

15.1. Norms
Lp Norm: For 1 ≤ p < ∞, the Lp norm of a function f is dened as
Z 1/p
p
∥f ∥p = |f | dµ

For p = ∞, the L∞ norm is dened as

∥f ∥∞ = inf{M ≥ 0 : µ({x : |f (x)| ≥ M }) = 0}

If no such M exists, then we say ∥f ∥∞ = ∞. ∥f ∥∞ is the smallest number M such that |f | < M
a.e.

Conjugate Exponent: For 1 ≤ p ≤ ∞, dene the conjugate exponent q of p as follows:

- If 1 < p < ∞, then let q be uniquely dened such that 1/p + 1/q = 1.
- If p = 1, then p = ∞.
- If p = ∞, then q = 1.
Hölder's Inequality: Let 1≤p≤∞ and q be the conjugate exponent of p. Then for measurable
functions f and g, we have Z
|f g| dµ ≤ ∥f ∥q ∥g∥q

For p = q = 2, the inequality becomes the Cauchy-Schwartz Inequality.

Minkowski's Inequality: For 1≤p≤∞ and measurable functions f and g, we have

∥f + g∥p ≤ ∥f ∥p + ∥g∥p
This is also called, very reasonably, triangle inequality for the Lp norm.

20 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

Lp Normed Linear Space: Dene an equivalence relation ∼ as follows

f ∼ g ⇐⇒ f − g = 0 a.e.

Then dene
Lp = {f : ∥f ∥p < ∞}/ ∼
Under this denition and the norm ∥·∥p , (Lp , ∥·∥p ) becomes a valid linear normed space:

1. ∥f ∥p ≥ 0 and ∥f ∥p = 0 if and only if f =0 a.e., i.e. f ∼0

2. ∥f + g∥p ≤ ∥f ∥p + ∥g∥p
3. ∥αf ∥ = |α|∥f ∥p

The important thing to keep in mind when talking about the Lp space is that when we take some
f∈ Lp , we are not taking one function but an equivalence class of functions.

Essential Inmum & Essential Supremum:

ess sup f = inf{M : µ({x : f (x) > M }) = 0}
ess inf f = sup{m : µ({x : f (x) < m}) = 0}

These denitions are very much like inmum/supremum, the only dierence is that they are
blind to dierences over null sets.

15.2. Completeness
Completeness: For 1 ≤ p ≤ ∞, the normed linear space (Lp , ∥·∥p ) is complete.

Dense Subset: The set of continuous functions with compact support is dense in Lp (R) for 1≤p⪇
∞.
This means that any f ∈ Lp can be approximated with innite precision by a continuous function
with compact support.

Exlusion of p = ∞: Notice that p = ∞ is excluded from this statement. Here's why: Take
1∈ Lp (R). For any continuous function f with compact support, we have ∥1 − f ∥∞ ≥ 1,
so the approximation error cannot be made innitely small under the given norm.

A special case: The set of continuous functions on [a, b] ⊂ R is dense in L2 ([a, b]).

15.3. Convolutions
In this subsection, all functions are dened on Rn and the measure is the Lebesgue measure on
Rn .

Convolution: For measurable functions f and g, the convolution is dened as follows:

Z
(f ∗ g)(x) = f (x − y)g(y)dy

if the integral exists.

Measurability:
R The function f (x−y)g(y) is jointly measurable, and so the convolution integral
f (x − y)g(y)dy results in a measurable function by the Fubini-Tonelli theorem.

21 Contents
Analysis - Cheat Sheet O§ul Can Yurdakul

Commutativity: The convolution operation is commutative, i.e. f ∗ g = g ∗ f.

Inequalities with Lp Norms:

- If f, g ∈ L1 , then f ∗ g ∈ L1 with

∥f ∗ g∥1 ≤ ∥f ∥1 ∥g∥1

- If f ∈ L1 and g ∈ Lp where 1 < p ≤ ∞, then f ∗ g ∈ Lp with

∥f ∗ g∥p ≤ ∥f ∥1 ∥g∥p

Mollication: The procedure of approximating any function in Lp with a smooth function is called
mollication. The approximation of f ∈ Lp is obtained via a convolution operation with a
function φ : Rn → R with compact support that is innitely dierentiable, non-negative and
with integral equal to 1.
Notice that if φ is such a function, so is φε (x) = ε−n φ(x/ε) for ε > 0.

Properties of the Mollication: Suppose 1 ≤ p ≤ ∞ and f ∈ Lp (Rn ).

1. Dierentiability: For all ε > 0, f ∗ φε is innitely dierentiable.
2. Derivatives: For any non-negative integers α1 , . . . , αn the partial derivatives of f ∗ φε
are given as follows:

∂ α1 +···+αn ∂ α1 +···+αn
 
(f ∗ φε ) = f ∗ φε
(∂x1 )α1 . . . (∂xn )αn (∂x1 )α1 . . . (∂xn )αn

3. Approximation: f ∗ φε → f as ε → 0+ .
4. Uniform Convergence: If f is continuous, then f ∗ φε → f uniformly on compact sets
as ε → 0+ .
5. Convergence in Lp : For 1 ≤ p < ∞, f ∗ φε → f in Lp as ε → 0+ .

22 Contents