Modes of Convergence in A Measure Space and Some Applications
Modes of Convergence in A Measure Space and Some Applications
Modes of Convergence in A Measure Space and Some Applications
Some Applications
1 Introduction 2
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Preliminaries 4
2.1 Some Definitions . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3 Main result 10
3.1 Modes of convergence . . . . . . . . . . . . . . . . . . . . . . 10
3.2 Some Implications . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2.1 Linearity Of Convergence . . . . . . . . . . . . . . . . 15
3.2.2 Applications . . . . . . . . . . . . . . . . . . . . . . . 17
3.3 Discussion And Conclusion . . . . . . . . . . . . . . . . . . . 18
1
Chapter 1
Introduction
Convergence is one of the key tools available in the study of real analysis,
this is also true when doing measure theory.The notion of convergence of a
sequence dates back to the early nineteenth century and the work of Bolzano
(1817) and Cauchy (1821).When studying real analysis we see that if one has
a sequence x1 , x2 , x3 ... ∈ R of real numbers xn it is said that this sequence
converges to limit x ∈ R, if ε > 0, there exist N such that |xn − x| ≤ ε for
every n > N . Similarly for a sequence z1 , z2 , z3 ... ∈ C of complex number
zn converging to z ∈ C.
2
number of degrees of freedom, and this allows them to approach f in any
number of inequivalent ways.When the domain X is real or complex we
have two basic modes of convergence which are uniform convergence and
pointwise convergence. but what happens now when fn converges to f for
most points in X except for possibly for a negligible subset of X. this
leads us to us to measure theory where the domain X is equipped with the
structure of a measure space (X, A, µ) and the functions fn (and their limit
f ) are measurable with respect to this space.
1.1 Motivation
As seen in the introduction above when working with a real or complex
domain, if we have a sequence of functions, convergence is only studied
on X, which limits us since we can’t study the case where this sequence
converges in most of X except for some negligible subset of X that is a set
of measure very small or equal to zero, which is very important in the study
of convergence of functions. this is the driving force of this study as the
various other modes of convergence which arise from equipping X with a
measure space (X, A, µ) will be discourse and some of their properties and
application will be seen.
1.2 Objectives
The main objective of this study will be understanding each of these modes
of convergence and their properties with some applications. this study will
go about this in the following steps.
3
Chapter 2
Preliminaries
In this chapter we will see some definitions and theorems which we will use
later on in our study.
1. X ∈ A
3. if A ∈ A and B ∈ A then A ∪ B ∈ A
Example 2.1.1.
Example 2.1.2.
4
the class of intervals of R is not an algebra
proof
let A be the class of intervals of R the interval [2, 3] ∈ A but we have
[2, 3]c = [−∞, 2[∪]3, +∞] ∈ / A which implies A is not an algebra.
Definition 2.1.2 (σ−algebra). A class A of subsets of a non-empty set X is
called a σ-algebra (or σ − f ield, tribe on x) if the following three properties
are fulfilled
1. X ∈ A
5
Definition 2.1.4 (measure). Let (X, A) be a measurable space we define
a Measure on A to be any positive function µ : A → [0, +∞] for which the
following two properties hold
1. µ(∅) = 0
2. for any pairwise disjoint sequence (An )n∈N of elementsPof A that is
An ∩ Am = ∅ f or m 6= n we have that µ(∪n∈N An ) = µ(An )
n=N
Example 2.1.7.
let (X,A) be a measurable space where A is the power set of X then the
function µ : A → R where µ(A) = a for all A ∈ A and a ∈ R is not a
measure.
proof
it is trivial as if we let a = 2 then µ(∅) = 2 which implies µ is not a measure
by property one of a measure.
Definition 2.1.5 (measure space). Let X be a non empty set and A a
σ − algebra on X with µ a measure on A we call the triple (X, A, µ) a
measure space.
Example 2.1.8.
let X be a non empty set then (X, P (X), µ) is a measure space where
P(X) is the power set of X and µ is the dirac measure on P(X). as seen
in example 2.1.3 and 2.1.6, P(X) is a σ − algebra and dirac measure is a
measure on any non empty set.
6
Definition 2.1.6. Measurable Function
Let (X, A) a measurable spaces then we call a function f : X → R measur-
able function if the set {x | f (x) > α} ∈ A where α ∈ R.
Example 2.1.9.
Example 2.1.10.
Example 2.1.11.
Example 2.1.12.
7
2.2 Theorems
In this section we will see some theorems which will aid in the completion
of this work.
Theorem 2.2.1. Let A be a σ − algebra on a non-empty set X. Then the
following assertions hold:
1. ∅ ∈ A;
S
2. for any countable family (Ai )i∈J of A, one has ( Ai ) ∈ A ;
i∈J
T
3. for any countable family (Ai )i∈J of A, one has ( Aj ) ∈ A;
j∈J
Proof
let (Aj )j∈J be a countable family of elements of A
X ∈ A as A is a σ − algebra ⇒ X c = ∅ ∈ A by property 2 of definition
2.1.2 above
8
proof
the later resulting from the fact that s is one-to-one and onto.
(b) The equality B = A ∪ (B\A) allows us to write, according to (a),
µ(B) = µ(A) + µ(B\A) which implies µ(B) ≥ µ(A) since µ(B\A) ≥ 0
(c) Like in (b), we have µ(B) = µ(A) + µ(B\A), and hence µ(B\A) =
µ(B) − µ(A) since µ(A) ∈ R.
(d) assume that J is non empty and like above fix a subset S ⊂ N and a
one-to-one and onto mapping s : S → J. put A‘n = As(n) if n ∈ S
and A‘n = ∅ if n ∈ N\S then
S ‘
An . Put also B1 = A‘1 and
S
Aj =
j∈J j∈J
n−1
Bn = A‘n \( A‘k ) if n ≥ 2. then
S
k=1
µ(A‘n ) =
S P P P P
µ( Aj ) = µ(Bn ) ≤ µ(As(n) ) = µ(Aj )
j∈J n∈N n∈N n∈N j∈J
the first equality being due to (3.1) and (a) and the inequality due to
the inclusion of (3.1).
Theorem 2.2.3. let f : X → R+ be non negative real valued function in
the measure space (X, A, µ) then for ε > 0 we have µ({x ∈ X : f (x) ≥ ε}) ≤
1
ε ||f (x)||L1
proof
set En,ε = {xZ : |f (x)| ≥ ε} Z
||f (x)||L1 = |f (x)|dx ≥ |f (x)|dx ≥ εµ(En,ε ) ⇒ µ(En,ε ) ≤ 1ε ||f ||L1
X En,ε
9
Chapter 3
Main result
Example 3.1.1.
Example 3.1.2.
10
since the rational numbers are rational ( they can be enumerated let fn
1 : x = q1 , q2 , q3 , ..., qn
be the sequence of functions fn (x) =
0 : otherwise
where {q1 , q2 , ...} is the set of rationals in the interval [0, 1], denoted by Q[0,1]
this function those not converge pointwise to 0 since for x ∈ Q[0,1] it will
converge but to 1.
Uniform convergence
Definition: the sequence fn converges uniformly to f if, for every ε > 0,
there exist N such that for every n ≥ N , |fn (x) − f (x)| ≤ ε for every x ∈ X.
Example 3.1.5.
define the sequence of functions fn (x) = n1 X(0,1) where X is the indicator
function, this sequence converges uniformly to 0.
in this case let ε > 0 be given then by the archimedean principle there exist
an N ∈ N such that N1( < ε hence for all x ∈ X n ≥ N we have that
1
, if n ∈ (0, 1)
|f (x)n − 0| = n1 X(0,1) = n ≤ N1 < ε hence the sequence fn
0, if n ∈ / (0, 1)
Convergence in L∞ N orm
Definition: the sequence fn converges to f in L∞ norm if, for every ε > 0,
there exist N ∈ N such that for every n ≥ N, ||fn (x) − f (x)||L∞ ≤ ε.
11
Almost Uniform Convergence
Definition: the sequence fn converges almost uniformly to f if, for every
ε > 0, there exist an exceptional set E ∈ A of measure µ(E) ≤ ε such that
fn converges uniformly to f on the complement of E.
Convergence in Measure
This is the weakest of all the modes of convergence in a measure space since
it is implied by all the other modes of convergence.
Definition: the sequence fn converges to f in measure if for ε > 0,
lim µ{x : |fn (x) − f (x)| > ε} = 0
n→∞
Convergence in L1 Norm
Definition: 1
Z the sequence fn converges to f in L Norm if
lim |fn − f | dµ = 0 OR ||fn − f ||L1 → 0 as n → ∞
n→∞ X
Proposition 3.1
let (X, A, µ) be a measure space and fn a sequence of measurable functions.
if fn converges uniformly to f then it also converges pointwise to f.
proof
fn converges uniformly to f ⇒ lim fn (x) = f (x) for all x ∈ X,
n→∞
⇒ ∀ ε > 0 ∃ N0 ∈ N such that for all n > N0 |fn (x) − f (x)| < ε for all x
now for each x ∈ X, ∀ε > 0 taking N0 above we have that n > N0 ⇒
|fn (x) − f (x)| < ε which implies point wise convergence to f
Proposition 3.2
let (X, A, µ) be a measure space and fn a sequence of measurable functions.
if fn converges pointwise to f then it also converges almost every where to
f.
proof
fn converges to f pointwise implies for all x ∈ X, ∀ε > 0 ∃ N (which depends on both x and ε)
12
such that for all n > N |fn (x)−f (x)| < ε ⇒ µ(x ∈ X : f (x) converges to f ) =
1 ⇒ µ(x ∈ X : f (x) does not converges to f ) = 0 which implies almost
every convergence to f
Proposition 3.3
let (X, A, µ) be a measure space and fn a sequence of measurable functions.
if fn converges to f almost uniformly then fn converges to f almost every-
where.
proof
fn converges to f almost uniformly implies given ε > 0 ∃ E ∈ A such that
µ(E) < ε and fn converges uniformly to f on E c ⇒ for each m ≥ 1 we can
1 c
find Em ∈ A such
T that µ(Em ) ≤ m and fn 1converges to f uniformly on Em
now set E = Em ⇒ µ(E) < µ(Em ) < m for each m ≥ 1 ⇒ µ(E) = 0 as
m≥1
m can be made as large as possible. now let x ∈ E c =
S c
Em ⇒ ∃ k ≥ 1
m≥1
such that x ∈ Ekc which implies fn converges uniformly to f for x and hence
converges pointwise for x. since x was arbitrary we can then conclude that
fn converges pointwise to f on E c . which implies it converges pointwise
almost everywhere to f .
Proposition 3.4
let (X, A, µ) be a measure space and fn a sequence of measurable functions.
if fn converges almost uniformly to f then it also converges in measure to
f . proof
fn converges almost uniformly to f implies given ε > 0 there exist Eε ∈ A
such that µ(Eε ) < ε and fn → f uniformly outside Eε .
⇔ given η > 0, there exist N ∈ N such that |fn (x) − f (x)| ≤ η for all n ≥ N
and for all x ∈ Eεc where N depends on ε, η but not on any x ∈ Eεc
⇒ {x ∈ X : |fn (x) − f (x)| > η} ⊆ Eε ∀ n ≥ N
⇒ µ({x ∈ X : |fn (x) − f (x)| ≥ η}) ≤ µ(Eε ) ≤ ε
⇒ lim µ({x ∈ X : |fn (x) − f (x)| ≥ η}) = 0 since measure is a positive
n→∞
function and µ({x ∈ X : |fn (x) − f (x)| ≥ η}) ≤ ε for all n ≥ N also η is
fixed
⇒ fn converges to f in measure.
13
Proposition 3.5
let (X, A, µ) be a measure space and fn a sequence of measurable functions.
if fn converges to f uniformly then it also converges to f almost uniformly.
proof
fn converges uniformly to f implies given ε > 0 there exist N ∈ N such that
∀ n ≥ N |fn (x) − f (x)| < ε for every x ∈ X now let Eεc = X we have that
Eε = X c = φ ∈ A and µ(Eε ) = µ(φ) = 0 < ε also as fn converges uniformly
to f on Eεc = X, our proof is complete.
Proposition 3.6
let (X, A, µ) be a measure space and fn a sequence of measurable functions.
if fn converges to f uniformly then it also converges to f in L∞ Norm.
proof
fn converges uniformly to f implies given ε > 0 there exist N ∈ N such that
∀ n ≥ N |fn (x) − f (x)| < ε for every x ∈ X ⇒ ε is an essential bound of
fn − f
⇒ ||fn − f ||L∞ = inf {M ∈ [o, +∞] where M is an essential bound of fn −
f } ≤ ε which implies the function also converges in L∞ norm
Proposition 3.7
let (X, A, µ) be a measure space and fn a sequence of measurable functions.
if fn converges to f in L∞ norm then it also converges to f almost uniformly.
proof
before proving this lets first see this lemma
Lemma 3.2.1
let (X, A, µ) be a measure space and fn a sequence of measurable functions
also f is measurable. then fn → f uniformly almost everywhere
⇔ ||fn − f ||L∞ → 0 as n → ∞ that is uniformly almost everywhere conver-
gence is equivalent to convergence in L∞ norm
proof
lets first take the forward implication, if fn → f uniformly almost every-
where ⇒ given ε > 0, ∃ N ∈ N such that ∀ n ≥ N , we have |fn (x)−f (x)| ≤ ε
for all x − µ almost everywhere ⇒ ε is an essential bound for |fn − f | ⇒
||fn − f ||L∞ ≤ ε for all n ≥ N ⇒ ||fn − f ||L∞ → 0 as n → ∞
conversely
if ||fn (x) − f (x)||L∞ → 0 as n → ∞ ⇒ given ε > 0 there exist N ∈ N such
that n ≥ N ⇒ ||fn − f ||L∞ ≤ ε f or µ almost everywhere x ∈ X
but ||fn − f ||L∞ = inf {M ∈ [0, +∞]} where M is an essential bound of
14
fn − f ⇒ |fn − f | ≤ M f or µ almost everywhere x ∈ X
⇒ |fn − f | < ε µ almost everywhere x ∈ X which concludes our proof.
Now lets apply this lemma to proof our theorem above, fn converges to f in
L∞ implies given ε > 0 there exist N ∈ N such that n > N ||fn − f ||L∞ ≤ ε
⇔ given ε > 0 there exist N ∈ N such that ∀ n ≥ N |fn (x) − f (x)| ≤ ε for
x outside a µ null set,
which is equivalent to saying fn → f uniformly outside of Eε ∈ A , µ(Eε ) =
0 ≤ ε ⇒ fn → f almost uniformly
Proposition 3.8
let (X, A, µ) be a measure space and fn a sequence of measurable functions.
if fn converges to f in L1 norm then it also converges to f in measure.
proof
fn converges to f in L1 ⇒ ||fn − f ||L1 → 0 as n → ∞ but by markov’s
inequality for ε > 0 we have
15
mode for any c ∈ C
proof
1.Pointwise convergence
in this case fn converges to f and gn to g pointwise implies there exist
N, M ∈ N such that for every x ∈ X, ε > 0, α > 0 we have n ≥ N implies
|fn − f | ≤ 2ε and n ≥ M implies |gn − g| ≤ 2ε , take K = max{N, M }
now |(fn + gn ) − (f + g)| = |(fn − f ) + (gn − g)| ≤ |fn − f | + |gn − g| ≤
ε ε
2 + 2 = ε ∀ n ≥ K that is |(fn + gn ) − (f + g)| ≤ ε which implies that fn + gn
converges to f + g pointwise.
for the second case fn converges to f pointwise implies for every ε > 0, x ∈ X
ε
there exist N ∈ N such that n ≥ N implies |fn − f | ≤ |c| ⇒ |c||fn − f | ≤
ε ⇒ |cfn − cf | ≤ ε which implies cfn converges to cf .
2. convergence in measure
now fn converges to f and gn to g in measure
now for ε > 0 define
Fn,ε = {x ∈ X : |fn − f | ≥ ε}, Gn,ε = {x ∈ X : |gn − g| ≥ ε}
Hn,ε = {x ∈ X : |(fn + gn ) − (f + g)| ≥ ε}
fn → f in measure ⇔ µ(Fn,ε ) → 0 as n → ∞ similarly µ(Gn,ε ) → 0 as
n → ∞ now if ε ≤ |(fn + gn ) − (f + g)| ≤ |fn − f | + |gn − g|
so if x ∈ Hn,ε by the inequality above we have x ∈ Fn, 2ε or x ∈ Gn, 2ε this
implies Hn,ε ⊆ Fn, 2ε ∪ Gn, 2ε ⇒ µ(Hn,ε ) ≤ µ(Fn, 2ε ) + µ(Gn, 2ε ) hence as n → ∞
we have µ(Fn, 2ε ) + µ(Gn, 2ε ) → 0 ⇒ µ(Hn,ε ) → 0 which implies fn + gn con-
verges to f + g in measure.
16
Z
⇒ lim |(fn + gn ) − (f + g)| dµ = 0 which implies fn + gn converges
n→∞ X
L1 norm to f + g.
Z
for the second case fn converges to f in L1
norm implies lim |fn − f | dµ =0
n→∞ X
Z Z
⇒ |c| lim |fn − f | dµ = 0|c| ⇒ lim |c||fn − f | dµ = 0
n→∞ n→∞
Z X X
3.2.2 Applications
In this chapter we will see some important theorems in measure theory(and
even in probability but we will not see in this work) that has been establish
using some of these modes of convergence.
17
Monotone Convergence Theorem
Fatou’s Lemma
Dominated Convergence Theorem
18