Measurable Spaces and Measurable Maps

3.
Measurable spaces and measurable maps

In this section we discuss a certain type of maps related to -algebras.
Definitions. A measurable space is a pair (X, A) consisting of a (non-empty)
set X and a -algebra A on X.
Given two measurable spaces (X, A) and (Y, B), a measurable map T : (X, A)
(Y, B) is simply a map T : X Y , with the property
(1) T
1
(B) A, B B.
Remark 3.1. In terms of the constructions outlined in Section 2, measurability
for maps can be characterized as follows. Given measurable spaces (X, A) and
(Y, B), and a map T : X Y , the following are equivalent:
(i) T : (X, A) (Y, B) is measurable;
(ii) T
B A;
(iii) T
A B.
Recall
T
B =
_
T
1
(B) : B B
_
;
T
A =
_
B Y : T
1
(B) A
_
.
With these equalities, everything is immediate.
The following summarizes some useful properties of measurable maps.
Proposition 3.1. Let (X, A) be a measurable space.
(i) If A
is any -algebra, with A
A, then the identity map Id

X
: (X, A)
(X, A
) is measurable.
(ii) For any subset M X, the inclusion map : (M, A
M
) (X, A) is
measurable.
(iii) If (Y, B) and (Z, C) are measurable spaces, and if (X, A)
T
(Y, B)
S
(Z, C) are measurable maps, then the composition S T : (X, A) (Z, C)

is again a measurable map.
Proof. (i). This is trivial, since (Id
X
)
= A
A.
(ii). This is again trivial, since
A = A
M
.
(iii). Start with some set C C, and let us prove that (S T)
1
(C) A. We
know that (S T)
1
= T
1
_
S
1
(C)
_
. Since S is measurable, we have S
1
(C) B,
and since T is measurable, we have T
1
_
S
1
(C)
_
A.
Often, one would like to check the measurability condition (1) on a small col-
lection of Bs. Such a criterion is the following.
Lemma 3.1. Let (X, A) and (Y, B) be masurable spaces. Assume B = (E),
for some collection of sets E P(Y ). For a map T : X Y , the following are
equivalent:
(i) T : (X, A) (Y, B) is measurable;
(ii) T
1
(E) A, E E.
200
3. Measurable spaces and measurable maps 201
Proof. The implication (i) (ii) is trivial.
To prove the implication (ii) (i), assume (ii) holds. We rst observe that
condition (ii) reads f
E A. Since A is a -algebra, we get the inclusion

(f
E) A.
Using the Generating Theorem 2.2, we have
f
B = f
(E) = (f
E) A,
and, by the preceding remark, we are done.
Corollary 3.1. Let (X, A) be a measurable space, let Y be a topological Haus-
dor space which is second countable, and let S be a sub-base for the topology of Y .
For a map T : X Y , the following are equivalent:
(i) T : (X, A)
_
Y, Bor(Y )
_
is a measurable map;
(ii) T
1
(S) A, S S.
Proof. Immediate from the above Lemma, and Proposition 2.2, which states
that Bor(Y ) = (S).
We know (see Section 19) that the type is consistent and natural. In par-
ticular, measurability behaves nicely with respect to products and disjoint unions.
More explicitly one has the following.
Proposition 3.2. Let (X
i
, A
i
)
iI
be a collection of measurable spaces. Con-
sider the sets X =
iI
X
i
and Y =
iI
X
i
, and the -algebras
A = -
X
iI
A
i
and B =
iI
A
i
.
Let (Z, G) be a measurable space.
(i) If we denote by
i
: X X
i
, i I, the projection maps, then a map
f : (Z, G) (X, A) is measurable, if and only if, all the maps
i
f :
(Z, G) (X
i
, A
i
), i I, are measurable.
(ii) If we denote by
i
: X
i
Y , i I, the inclusion maps, then a map
g : (Y, B) (Z, G) is measurable, if and only if, all the maps g
i
f :
(X
i
, A
i
) (Z, G), i I, are measurable.
Proof. (i). By the denition of the product -algebra, we know that
(2) A =
_
_
iI
i
A
i
_
.
If we x some index i I, then the obvious inclusion
i
A
i
A immediately
shows that
i
: (X, A) (X
i
, A
i
) is measurable. Therefore, if f : (Z, G) (X, A)
is measurable, then by Proposition 3.1 it follows that all compositions
i
f :
(Z, G) (X
i
, A
i
), i I, are measurable.
Conversely, assume all the compositions
i
f are measurable, and let us show
that f : (Z, G) (X, A) is measurable. By Lemma 3.1 and (2), all we need to
prove is the fact that
f
_
_
iI
i
A
i
_
G,
which is equivalent to
f
i
A
i
_
G, i I.
202 CHAPTER III: MEASURE THEORY
But this is obvious, because f
i
A
i
_
= (
i
f)
A
i
, and
i
f is measurable, for
all i I.
(ii). By the denition of the -algebra sum, we know that
(3) B =
iI
i
A
i
.
If we x some index i I, then the obvious inclusion
i
A
i
B immediately
shows that
i
: (X
i
, A
i
) (Y, B) is measurable. Therefore, if g : (Y, B) (Z, G)
is measurable, then by Proposition 3.1 it follows that all compositions g
i
:
(X
i
, A
i
) (Z, G), i I, are measurable.
Conversely, assume all the compositions g
i
are measurable, and let us show
that g : (Y, B) (Z, G) is measurable. This is equivalent to the inclusion g
B G.
By (3) we immediately have
(4) g
B = g
iI
i
A
i
_
=
iI
g
i
A
i
_
.
We know however that, since g
i
are all measurable, we have
g
i
A
i
_
= (g
i
)
A
i
G, i I,
so the desired inclusion is an immediate consequence of (4).
Conventions. Let (X, A) be a measurable space. An extended real-valued
function f : (X, A) [, ] is said to be a measurable function, if it is measur-
able in the above sense as a map f : (X, A)
_
[, ], Bor([, ])
_
. If f has
values in R, this is equivalent to the fact that f is a measurable map f : (X, A)
_
R, Bor(R)
_
is measurable. Likewise, a complex valued function f : (X, A) C is
measurable, if it is measurable as a map f : (X, A)
_
C, Bor(C)
_
. If K is one of
the elds R or C, we dene the set
B
K
(X, A) =
_
f : (X, A) K : f measurable function
_
.
Remark 3.2. Let (X, A) be a measurable space. If A R is a dense subset,
then the results from Section 2, combined with Lemma 2.1, show that the measur-
ability of a function f : (X, A) [, ] is equivalent to any of the following
conditions:
f
1
_
(a, ]
_
A, a A;
f
1
_
[a, ]
_
A, a A;
f
1
_
[, a)
_
A, a A;
f
1
_
[, a]
_
A, a A.
Definition. If X and Y are topological Hausdor spaces, a map T : X Y
is said to be Borel measurable, if T is measurable as a map
T :
_
X, Bor(X)
_
_
Y, Bor(Y )
_
.
In the cases when Y = R, C, [, ], a Borel measurable map will be simply
called a Borel measurable function.
For K = R, C, we dene
B
K
(X) =
_
f : X K : f Borel measurable function
_
.
Remark 3.3. If X and Y are topological Hausdor spaces, then any continuous
map T : X Y is Borel measurable. This follows from Lemma 3.1, from the fact
that
Bor(Y ) =
_
D Y : D open
_
,
and the fact that T
1
(D) is open, hence in Bor(X), for every open set D Y .
Measurable maps behave nicely with respect to measurable countable opera-
tions, as suggested by the following result.
Proposition 3.3. Let (X, A) and (Z, B) be a measurable spaces, let I be a
set which is at most countable, and let (Y
i
)
iI
be a family of topological Hausdor
spaces, each of which is second countable. Suppose a measurable map T
i
: (X, A)
_
Y
i
, Bor(Y
i
)
_
is given, for each i I. Dene the map T : X
iI
Y
i
by
T(x) =
_
T
i
(x)
_
iI
, x X.
Equip the product space Y =
iI
Y
i
with the product topology.
For any measurable map g :
_
Y, Bor(Y )
_
(Z, B), the composition g T :
(X, A) (Z, B) is measurable.
Proof. We know (see Corollary 2.3) that we have the equality
Bor(Y ) = -
X
iI
Bor(Y
i
).
By Proposition 3.2, the map T : (X, A)
_
Y, Bor(Y )
_
is measurable, so by Propo-
sition 3.1, the composition g T : (X, A) (Z, B) is also measurable.
The above result has many useful applications.
Corollary 3.2. Suppose (X, A) is a measurable space, and K is either R or C.
Then, when equipped with point-wise addition and multiplication, the set B
K
(X, A)
is a unital K-algebra.
Proof. Clearly the constant function 1 is measurable.
Also, if f B
K
(X, A) and K, then the function f is again measurable,
since it can be written as the composition M
f, where M
: K K
is obviously continuous.
Finally, let us show that if f
1
, f
2
B
K
(X, A), then f
1
+ f
2
and f
1
f
2
again
belong to B
K
(X, A). This is however immediate from Proposition 3.3, applied to
the index set I = 1, 2, the spaces Y
1
= Y
2
= K, and the continuous maps
g
1
: K
2
(
1
,
2
)
1
+
2
K,
g
2
: K
2
(,
2
)
1

2
K.
Corollary 3.3. If (X, A) is a measurable space, then a complex valued func-
tion f : X C is measurable, if and only if the real valued functions Re f, Imf :
X R are measurable.
Proof. If f is measurable, the composing f with the continuous maps
: C z Re z R and : C z Imz R,
immediately gives the measurability of Re f = f and Imf = f.
Conversely, if both Re f, Imf : X R then the measurability of f follows from
Proposition 3.3, applied to Y
1
= Y
2
= R, the functions f
1
= Re f and f
2
= Imf,
and to the continuous function
g : R
2
(a, b) a +bi C.
Corollary 3.4. Let (X, A) be a measurable space, let I be a set which is at
most countable, and let f
i
: (X, A) [, ], i I be collection of measurable
functions. Then the functions g, h : X [, ], dened by
g(x) = inf
_
f
i
(x) : i I
_
and h(x) = sup
_
f
i
(x) : i I
_
, x X,
are both measurable.
Proof. Dene the maps m, M :
iI
[, ] [, ] by
m(x) = infx
i
: i I and M(x) = supx
i
: i I, x = (x
i
)
iI

iI
[, ].
By Proposition 3.3, it suces to prove the (Borel) measurability of the maps m
and M.
To prove the measurability of m, we are going to show that
m
1
_
[, a)
_
Bor
_
iI
[, ]
_
, a R.
But this is quite obvious, since a point x = (x
i
)
iI
belongs to m
1
_
[, a)
_
, if
and only if there exists some j I with x
i
< a. In other words, if we dene the
projections
j
:
iI
[, ] [, ], then we have
m
1
_
[, a)
_
=
_
jI
j
_
[, a)
_
.
This shows that in fact m
1
_
[, a)
_
is open, hence clearly Borel.
To prove the measurability of M, we are going to show that
M
1
_
(a, ]
_
Bor
_
iI
[, ]
_
, a R.
But this is again clear, since, as before, we have the equality
M
1
_
(a, ]
_
=
_
jI
j
_
(a, ]
_
,
which shows that in fact M
1
_
(a, ]
_
is open, hence Borel.
Corollary 3.5. Let (X, A) be a measurable space, and let f
n
: (X, A)
[, ], n N be sequence of measurable functions. Then the functions g, h :
X [, ], dened by
g(x) = liminf
n
f
n
(x) and h(x) = limsup
n
f
n
(x), x X,
are both measurable.
Proof. For every n N, dene the functions g
n
, h
n
: X [, ] by
g
n
(x) = inf
_
f
k
(x) : k n
_
and h
n
(x) = sup
_
f
k
(x) : k n
_
, x X.
By Corollary 3.5, we know that g
n
and h
n
are measurable for all n N. Since
g(x) = sup
_
g
n
(x) : n N
_
and h(x) = inf
_
h
n
(x) : n N
_
, x X,
the fact that both g and h are measurable follows again from Corollary 3.5.
Corollary 3.6. Let (X, A) be a measurable space, and let
f
n
: (X, A) [, ], n N
be sequence of measurable functions, with the property that, for each x X, the
sequence
_
f
n
(x)
_
n=1
[, ] has a limit. Then the function f : X [, ],
dened by
f(x) = lim
n
f
n
(x), x X,
is again measurable.
Proof. Immediate from the above result.
Exercise 1. If f
n
: R R, n N, are continuous functions, and if f(x) =
lim
n
f
n
(x) exists, for every x R, then by the above Corollary we know that
f : R [, ] is Borel measurable. Prove that the converse is not true. More
explicitly, prove that there is no sequence (f
n
)
n=1
of continuous functions, with
lim
n
f
n
(x) =
Q
(x), x R.
Hint: Use Baires Theorem.
Exercise 2. Prove that a function f : R R, which is continuous everywhere,
except for a countable set of points, is Borel measurable. As an application, prove
that any monotone function is Borel measurable.
Corollary 3.6 can be generalized, as follows.
Theorem 3.1. Let (X, A) be a measurable space, let Y be a separable metric
space, and let
T
n
: (X, A)
_
Y, Bor(Y )
_
, n N
be a sequence of measurable maps. Assume that, for every x X, the sequence
_
T
n
(x)
_
n=1
Y is convergent. Dene the map T : X Y by
T(x) = lim
n
T
n
(x), x X.
Then T : (X, A)
_
Y, Bor(Y )
_
is a measurable map.
Proof. Denote by d the metric on Y . The collection
V =
_
B
r
(y) : y Y, r > 0
_
is a base for the topology of Y . Since Y is second countable, it suces then to
show that
(5) T
1
_
B
r
(y)
_
A, y Y, r > 0.
Claim: For every y Y and r > 0 one has the equality
(6) T
1
_
B
r
(y)
_
=
_
m,n=1
_

k=m
T
1
k
_
B
r
1
n
(y)
_
_
.
Denote the set in the right hand side simply by A. Start rst with some x A.
There exist some m, n N such that
x
k=m
T
1
k
_
B
r
1
n
(y)
_
,
which means that
T
k
(x) B
r
1
n
(y), k m,
that is,
d
_
T
k
(x), y
_
< r
1
n
, k m.
Pasing to the limit (k ) then yields
d
_
T(x), y
_
r
1
n
< r,
which means that T(x) B
r
(y), i.e. x = T
1
_
B
r
(y)
_
, thus proving the inclusion
A T
1
__
B
r
(y)
_
.
Conversely, if x T
1
_
B
r
(y)
_
, we get T(x) (B
r
(y), i.e. d
_
T(x), y
_
< r.
Choose an integer n such that
(7) d
_
T(x), y
_
< r
2
n
.
Since lim
k
T
k
(x) = T(x), there exists some m N such that
d
_
T
k
(x), T(x)
_
<
2
n
, k m.
Combining this with (7) then gives
d
_
T
k
(x), y
_
d
_
T(x), y
_
+d
_
T
k
(x), T(x)
_
< r
2
n
+
1
n
= r
1
n
, k m,
which means that
x
k=m
T
1
k
_
B
r
1
n
(y)
_
,
hence x indeed belongs to A.
Having proven (6) we now observe that, since the T
k
s are measurable, it follows
that
T
1
k
_
B
r
1
n
(y)
_
A, k, n N, r > 0.
Using the fact that A is closed under countable intersections, it follows that
k=m
T
1
k
_
B
r
1
n
(y)
_
A, m, n N, r > 0.
Finally, using the fact that A is closed under countable unions, the desired property
(5) follows.
Exercise 3. Let (X, A) be a measurable space, and let (X
n
)
n=1
be a sequence
of sets in A, with X =

n=1
X
n
. Suppose (Y, B) is a measurable space, and
F : X Y is a map, such that
F
Xn
:
_
X
n
, A
Xn
_
(Y, B)
is measurable, for all n N. Prove that f : (X, A) (Y, B) is measurable.
Exercise 4*. Let
1
R
n
be an open set, and let f
1
, . . . , f
n
:
1
R be C
1
functions, with the property that the matrix
A(p) =
_
f
j
x
k
(p)
_
n
j,k=1
is invertible, for every point p
1
. Dene the map
F :
1
p
_
f
1
(p), . . . , f
n
(p)
_
R
n
.
(i) Prove that the set
2
= F(
1
) is open in R
n
.
(ii) Although F :
1

2
may fail to be injective, prove that there exists a
Borel measurable map :
2

1
, with F = Id
2
.
Hint: Use the Inverse Function Theorem, combined with Exercises 2 and 3. exercise.
Exercise 5*. Let P(z) be a non-constant polynomial with complex coecients.
Prove that there exists a Borel measurable function f : C C, such that
P
_
f(z)
_
= z, z C.
Hint: Use the preceding exercise, applied to the set
1
= {z C : P
(z) = 0}.
The preceding exercise can be generalized:
Exercise 6*. Let
1
C be a connected open set, and let f :
1
C be a
non-constant holomorphic function. By the Open Mapping Theorem we know that
the set
2
= f(
1
) is open. Prove that there exists a Borel measurable function
:
2

1
, such that f = Id
2
.
Hint: Use Exercise 4, applied to the set
0
= {z
1
: f
(z) = 0}. Since f is non-constant,

the set
1

0
is countable.
We continue with a discussion on the role of elementary functions.
Proposition 3.4. Let (X, A) be a measurable space, and let K be one of the
elds R or C. For an elementary function f Elem
K
(X), the following are equiv-
alent:
(i) f A-Elem
K
(X);
(ii) f : (X, A) K is measurable.
Proof. (i) (ii). We know that A-Elem
K
= Span
K
_
A
: A A. Since
B
K
(X, A) is a vector space, it suces to show only that
A
: (X, A) K is
measurable, for all A A. But this is trivial, since for every Borel set B R one
has either
1
A
(B) = , or
1
A
(B) = A, or
1
A
(B) = X.
(ii) (i). Assume now f is measurable. List the range of f as
f(X) =
1
, . . . ,
n
,
with
j
,=
k
, for all j, k 1, . . . , n with j ,= k. Since f is measurable, and the
singleton sets
1
, . . . ,
n
are in Bor(K), it follows that the sets A
j
= f
1
_
_
,
j = 1, . . . , n are all in A. Since we clearly have
f =
1
A1
+ +
n
An
,
it follows that f indeed belongs to A-Elem
K
(X).
Remarks 3.4. A. If (X, A) and (Y, B) are measurable spaces, if T : (X, A)
(Y, B) is a measurable map, and if f B-Elem
K
(Y ), then f T A-Elem
K
(X).
This follows from the fact that the composition f T : (X, A) K is measurable,
and elementary.
B. If (X, A) is a measurable space, if f A-Elem
K
(X), and if g : f(X) K is
an arbitrary function, then g f A-Elem
K
(X). This follows from the fact that,
if one considers the nite set Y = f(X), and the -algebra P(Y ) on it, then
(X, A)
f
_
Y, P(Y )
_
g
K
are measurable. So g f is also measurable, and obviously elementary.
The following is an interesting converse of Corollary 3.6.
Theorem 3.2. Let (X, A) be a measurable space, and let f : (X, A) [, ]
be a measurable function. Then there exists a sequence (f
n
)
n=1
A-Elem
R
(X),
such that
inf
_
f(y) : y X
_
f
n
(x) sup
_
f(z) : z X
_
, x X, n 1;
lim
n
f
n
(x) = f(x), x X.
Moreover,
(i) if inf
_
f(x) : x X
_
> , then the sequence (f
n
)
n=1
can be chosen to
be non-decreasing, i.e. f
n
f
n+1
, n N;
(ii) if sup
_
f(x) : x X
_
< , then the sequence (f
n
)
n=1
can be chosen to
be non-increasing, i.e. f
n
f
n+1
, n N;
(iii) if inf
_
f(x) : x X
_
> and sup
_
f(x) : x X
_
< , then the
sequence (f
n
)
n=1
can be chosen eiher non-decreasing, or non-increasing,
and such that it converges uniformly to f, i.e.
lim
n
_
sup
xX
f
n
(x) f(x)
_
= 0.
Proof. We begin with a special case of (iii). Assume X = [0, 1], A =
Bor([0, 1]), and consider the inclusion F : [0, 1] [, ]. For each n N,
dene the intervals I
n
k
, J
n
k
, 0 k 2
n
1 by
I
n
k
=
_
k/2
n
, (k + 1)/2
n
_
, if 0 k 2
n
2; I
n
2
n
1
=
_
(2
n
1)/2
n
, 1
,
J
n
k
=
_
k/2
n
, (k + 1)2
n
, if 1 k 2
n
1; J
n
0
=
_
0, 1/2
n
.
We then dene, for each n N, the functions g
n
, h
n
: [0, 1] R by
g
n
= 2
n
2
n
1
k=0
k
I
n
k
and h
n
= 2
n
2
n
1
k=0
(k + 1)
J
n
k
.
Remark that
(8) 0 g
n
(s) < 1 and 0 < h
n
(s) 1, s [0, 1].
Note that, for every n N, we have
g
n
(0) = 0; g
n
(1) = (2
n
1)/2
n
; (9)
h
n
(0) = 1/2
n
; h
n
(1) = 1. (10)
Claim 1: The sequence (g
n
)
n=1
is non-decreasing, and the sequence (h
n
)
n=1
is non-increasing.
Using (9) and (10), we only need to examine the restrictions to the open interval
(0, 1). Fix some point s (0, 1). For every integer n 1, dene
p
s
n
= max
_
k Z : 0
k
2
n
< s
_
.
We clearly have p
s
n
< 2
n
and
(11)
p
s
n
2
n
< s
p
s
n
+ 1
2
n
.
We then have
(12) g
n
(s) =
_
p
s
n
/2
n
if s ,= (p
s
n
+ 1)/2
n
(p
s
n
+ 1)/2
n
if s = (p
s
n
+ 1)/2
n
and h
n
(s) =
p
s
n
+ 1
2
n
We now estimate g
n+1
(s) and h
n+1
(s). First of all, using (11), we have
2p
s
n
2
n+1
< x
2p
s
n
+ 2
2
n+1
,
which means that either p
s
n+1
= 2p
s
n
, or p
s
n+1
= 2p
s
n
+ 1. This immediately gives
h
n+1
(s) =
p
s
n+1
+ 1
2
n+1

2p
s
n
+ 2
2
n+1
=
p
s
n
+ 1
2
n
= h
n
(s).
Note that, if s = (p
s
n
+1)/2
n
, we will have p
s
n+1
= 2p
s
+1 and s = (p
s
n+1
+1)/2
n+1
,
so we get
g
n+1
(s) = (p
s
n+1
+ 1)/2
n+1
= (2p
s
n
+ 2)/2
n+1
= (p
s
n
+ 1)/2
n
= g
n
(s).
If s ,= (p
s
n
+ 1)/2
n
, then
g
n
(s) =
p
s
n
2
n
=
2p
s
n
2
n

p
s
n+1
2
n+1
g
n+1
(s).
Claim 2: For every s [0, 1] one has
lim
n
_
sup
s[0,1]
g
n
(s) s
_
= lim
n
_
sup
s[0,1]
h
n
(s) s
_
= 0.
To prove this fact we are going to estimate the dierences [g
n
(s)s[ and [h
n
(s)s[.
If s = 0 or s = 1, then the equalities (9) and (10) immediately show that
(13) [g
n
(s) s[
1
2
n
and [h
n
(s) s[
1
2
n
, n N.
If s (0, 1), then the denitions of g
n
(s) and h
n
(s) clearly show that
s, g
n
(s), h
n
(s)
_
p
s
n
/2
n
, (p
s
n
+ 1)/2
n
,
and then we see that we again have the inequalities (13). Since (13) now holds for
all s [0, 1], the Claim immediately follows.
We proceed now with the proof of the theorem. Dene
= inf
_
f(x) : x X
_
and = sup
_
f(x) : x X
_
.
If = , there is nothing to prove. Assume < . Depending on the nitude of
and , we dene a homeomorphism : [, ] [0, 1], as follows.
(a) If > and < , we dene
(s) =
s

, s [, ].
(b) If > and = , we dene
(s) =
_
2
arctan(s ) if s ,=
1 if s =
(c) If = and < , we dene
(s) =
_
1 +
2
arctan(s ) if s ,=
0 if s =
(d) If = and = , we dene
(s) =
_
_
_
0 if s =
1
2
+
1
arctan(s ) if < s
1 if s =
Notice that () = 0, () = 1, and
s < t (s) < (t).
After these preparations, we proceed with the proof. We begin with the special
cases (i) (ii) and (iii).
If > , we dene the functions f
n
=
1
g
n
f. Since and
1
are
increasing, and (g
n
)
n=1
is non-decreasing, it follows that (f
n
)
n=1
is non-decreasing.
Since 0 g
n
(s) < 1, s [0, 1], we see that f
n
(x) < , x X. In particular,
we have < f
n
(x) < , for all n and x. It it obvious that f
n
is elementary,
measurable, and since lim
n
g
n
(s) = s, s [0, 1] (by Claim 2), we immediately
get lim
n
f
n
(x) = f(x), x X.
If < , we dene the functions f
n
=
1
h
n
f. Since and
1
are
increasing, and (h
n
)
n=1
is non-increasing, it follows that (f
n
)
n=1
is non-increasing.
Since 0 < h
n
(s) 1, s [0, 1], we see that < f
n
(x) , x X. In particular,
we have < f
n
(x) < , for all n and x. It it obvious that f
n
is elementary,
measurable, and since lim
n
h
n
(s) = s, s [0, 1] (by Claim 2), we immediately
get lim
n
f
n
(x) = f(x), x X.
If > and < , then we can take f
n
=
1
g
n
f, n, or we can
take f
n
=
1
h
n
f, n. The inequalities (13), combined with the denition
(c) of , show that
[f
n
(x) f[

2
n
, x X, n N,
with any of the above choices for (f
n
)
n=1
.
Having proven the cases (i), (ii) and (iii), we now examine the general situation,
when = and = . Consider the functions f
, f
: X [, ] dened
by
f
(x) = maxf(x), 0 and f
(x) = minf(x), 0, x X.
By Corollary 3.4, both f
and f
are measurable. Since inf

xX
f
(x) 0, by part
(i), there exists a sequence (f
n
)
n=1
A-Elem
R
(X), such that lim
n
f
n
(x) =
f
(x), x X. Since sup

xX
f
(x) 0, by part (ii), there exists a sequence

(f
n
)
n=1
A-Elem
R
(X), such that lim
n
f
n
(x) = f
(x), x X. Dene the

elementary functions f
n
= f
n
+f
n
, n N. Clearly the f
n
s are all in A-Elem
R
(X).
We now check that
(14) lim
n
f
n
(x) = f(x), x X.
There are two cases to examine: (a) f(x) 0; (b) f(x) 0.
In case (a), we have f
(x) = f(x) and f
(x) = 0, so lim
n
f
n
(x) = f(x) and
lim
n
f
n
(x) = 0.
In case (b), we have f
(x) = 0 and f
(x) = f(x), so lim

n
f
n
(x) = 0 and
lim
n
f
n
(x) = f(x).
In either case, the equality (14) follows.
We conclude this section with a discussion on an interesting measurable space,
that appears often in connection with probability theory.
Example 3.1. Consider the space T = 0, 1
0
, i.e.
T =
_
a = (
n
)
n=1
:
n
0, 1, n N
_
.
We call T the space of innite coin ippings, having in mind that an element
of T is the same as the outcome of an innite sequence of coin ips (think 0
as corresponding to tails, and 1 as corresponding to heads). Equipp T with the
product topology. By Tihonovs Theorem, T is compact. The product topology on
T is in fact given by a metric d dened by
d(a, b) =
n=1
[
n

n
[
2
n
, a = (
n
)
n=1
, b = (
n
)
n=1
T.
For every number r 2 we dene a map
r
: T [0, 1] by
r
(a) = (r 1)
n=1
n
r
n
, a = (
n
)
n=1
T.
It is pretty clear that
r
(a)
r
(b)
(r 1)d(a, b), a, b T,
so the maps
r
: T [0, 1], r 2 are continuous. In particular, the set K
r
=
r
(T)
is a compact subset of [0, 1].
Dene
T
0
=
_
a = (
n
)
nN
T : the set n N :
n
= 0 is innite
_
.
The set T T
0
can be described as:
T T
0
=
_
(
n
)
nN
T : there exists N N, such that
n
= 1, n N
_
.
The following are well known (see Appendix B, the proof of Proposition B.2).
Facts: 1. The set T T
0
is countable
2. For any r 2, and elements a = (
n
)
n=1
, b = (
n
)
n=1
T
0
, the
following are equivalent:
there exists N N such that
N
= 1,
N
= 0, and
n
=
n
, for all
n N with n < N;

r
(a) > (b).
In particular, the map
r
T0
: T
0
[0, 1] is injective.
The above constructions have a remarkable feature.
Theorem 3.3. Use the notations above. For a number r 2 and subset A T,
the following are equivalent:
(i) A Bor(T);
(ii)
r
(A) Bor(K
r
).
Proof. Throughout the proof the number r will be xed. The map
r
will be
denoted by , and the compact set K
r
will be denoted by K.
Since : T K is continuous, it is measurable, i.e. we have the implication
(15) B Bor(K)
1
(B) Bor(T).
Before we proceed with the actual proof, we need some preparations. Remark that,
since : T K is surjective, we have the equality
(16)
_
1
(C)
_
= C, C K.
Claim 1: If a subset C K is at most countable, if and only if the set
1
(C) T is at most countable.
Suppose C is at most countable countable. If we take A
0
=
1
(C) T
0
, and
A
1
=
1
(C) T
0
, then obviously
1
(C) = A
0
A
1
. Since A
1
T T
0
, and
T T
0
is countable, it follows that A
1
is at most countable, so we only need to prove
that A
0
is at most countable. But since
T0
is injective, and A
0
T
0
, it follows
that
A0
: A
0
C is injective, and then the fact that C is at most countable,
forces A
0
to be at most countable.
Conversely, if
1
(C) is at most countable, then so is
_
1
(C)
_
. By (16) we
are done.
For each subset A T, we dene
A) =
1
_
(A)
_
.
Remark that A A), A T. Note also that, for any family (A
i
)
iI
of subsets
of T, one has the equality
(17)

_
iI
A
i
_
=
1
_
_
_
iI
A
i
_
_
=
1
_
_
iI
(A
i
)
_
=
_
iI
1
_
(A
i
)
_
=
_
iI
A
i
).
As an application of Claim 1, to the set C = (T T
0
), we see that
() the set T T
0
) is at most countable.
Claim 2: For any subset A T
0
, one has the inclusion
A) A T T
0
).
In particular, the dierence A) A is at most countable.
Start with an arbitrary element x A) A. This means that x , A, but (x)
(A), which means that there exists some a A, with (x) = (a). Assume now
x , T T
0
, which means that x T
0
. But then, the fact that x, a T
0
, combined
with the injectivity of
T0
will force x = a, which is impossible since a A.
Claim 3: For any set A T, the dierence A) A is at most countable.
Take A
0
= A T
0
and A
1
= AA
0
. Notice that, since A
1
T T
0
, we have
A
1
) =
1
_
(A
1
)
_

1
_
(T T
0
)
_
= T T
0
),
so it follows that A
1
) is at most countable. We obviously have A = A
0
A
1
, so by
(17)
A) = A
0
) A
1
).
But now we are done, since
A) A =
_
A
0
) A
1
)
_
_
A
0
A
1
_
_
A
0
) A
0
_
A
1
),
and both A
0
) A
0
(by Claim 2) and A
1
) are at most countable.
Claim 4: For any subset A T, one has the inclusion
(18) (T A) K (A),
and the dierence (T A)
_
K (A)
_
is at most countable.
The inclusion (18) is pretty obvious, from the surjectivity of . In order to prove
that the dierence
C = (T A)
_
K (A)
_
= (T A) (A)
is countable, by Claim 1, it suces to prove that
1
(C) is countable. We have
1
(C) =
1
_
(T A) (A)
_
=
1
_
(T A)
_

1
_
(A)
_
= T A) A).
We can write
1
(C) = A
1
A
2
, where
A
1
= (T A) A) and A
2
=
_
T A) (T A)
A),
so it suces to prove that both A
1
and A
2
are at most countable. But these facts
are immediate from Claim 3, since A
1
= A) A, and A
2
T A) (T A).
We can now proceed with the proof of the theorem. Dene
A =
_
A T : (A) Bor(K)
_
,
so that what we need to prove is the equality A = Bor(T).
First, remark that, if A A, then (A) Bor(K), and the fact that is Borel
measurable will force A) =
1
_
(A)
_
to be a Borel set in T. But since A) A
is countable, hence Borel, it follows that
A = A)
_
A) A
_
is again Borel. Therefore, we have the inclusion A Bor(T).
Second, remark that if F T is a compact subset, then the continuity of
gives the fact that (F) is compact, hence Borel. This then forces F A. Therefore
A contains the collection C
T
of all compact subsets of T.
Now we have
C
T
A Bor(T) = (C
T
),
so all we need to prove is the fact that A is a -algebra, i.e. we have the properties
(a) A A T A A;
(b) for any sequence (A
n
)
n=1
A, the union

n=1
A
n
also belongs to A..
To check (a) start with some set A A. We know that (A) Bor(K), and
we want to show that (T A) is again Borel. By Claim 4, we know we can write
(T A) =
_
K (A)
C,
for some set C K which is at most countable. Since C and K (A) are Borel,
this shows that (T A) is also Borel.
Property (b) is obvious, since (A
n
), n 1 are all Borel, and
_

_
n=1
A
n
_
=
_
n=1
(A
n
).
Corollary 3.7. Use the above notations. For a number r 2 and a subset
B K
r
, the following are equivalent:
(i) B Bor(K
r
);
(ii)
1
r
(B) Bor(T).
Proof. The implication (i) (ii) is trivial, since
r
is continuous, hence
measurable.
Conversely, if the set A =
1
r
(B) is Borel, then by the Theorem,
r
(A) is
Borel. But since
r
is surjective, we have B =
r
(A).
Comments. From the above results, we see that
r
: T K
r
almost pre-
serves Borel structures. More explicitly, if one considers the maps
r
: P(T) A
r
(A) P(K
r
),
r
: P(K
r
) B
1
r
(B) P(T),
then
(
r

r
)(B) = B, for all B K
r
;
(
r

r
)(A) A, and (
r

r
)(A) A is at most countable, for all
A T;
B Bor(K
r
)
r
(B) Bor(T);
A Bor(T)
r
(A) Bor(K
r
).
In the particular case r = 2, we know that K
2
= [0, 1], so we can think the mea-
surable space
_
[0, 1], Bor([0, 1])
_
as approximatively the same as the measurable
space
_
T, Bor(T)
_
.
The case r = 3 will be an interesting one, especially for constructing various
counter-examples. The compact set K
3
[0, 1] is called the ternary Cantor set.
It turns out that there exists another useful description of the ternary Cantor
set K
3
, which yields some interesting properties.
Notations. We keep the notations above. An element a = ()
n=1
T will
be called nite, if there exists some N N, such that
n
= 0, n 0. We dene
T
n
=
_
a T : a nite
_
.
Remark that T
n
T
0
. In particular the map
3
T
n
: T
n
K
3
is injective.
For a T
n
we dene its length as
(a) = minN N :
n
= 0, n N 1.
With this denition, for every a = (
n
)
n=1
T
n
, we have
(19)
(a)
= 1 and
n
= 0, n > (a).
We dene
=
_
(k, a) Z T
n
: k (a)
_
.
Finally, for every pair = (k, a) , we dene the open interval
I
=
_
3
(a) +
1
3
k+1
,
3
(a) +
2
3
k+1
_
.
Remark that, using (19) we have
3
(a) 2
a
n=1
2
3
n
= 1
1
3
(a)
,
with the convention that the sum is 0, if (a) = 0. We then get
3
(a) +
2
3
k+1
1
1
3
(a)
+
2
3
k+1
< 1
1
3
(a)
+
1
3
k
1,
which gives the inclusion I
(0, 1).
The following result is describes an alternative construction of K
3
.
Theorem 3.4. Use the notations above.
(i) The set T
n
is dense in T;
(ii) The system (I
is pair-wise disjoint.
(iii)

= [0, 1] K
3
.
Proof. The map
3
will be simply denoted by , and the Cantor set K
3
will
be denoted simply by K.
(i). Fix some element a = (
n
)
n=1
T. For every integer k 1 dene the
element a
k
= (
k
n
)
n=1
T, by
k
n
=
_

n
if n k
0 if n > k
It is obvious that a
k
T
n
, k N. The inequality
d(a, a
k
) =
n=k+1
n
2
n

n=k+1
1
2
n
=
1
2
k
, k N
then immediately shows that lim
k
a
k
= a.
(ii). Assume , are such that ,= , and let us prove that I
= .
Let = (j, a) and = (k, b), where a = (
n
)
n=1
and b = (
n
)
n=1
are elements int
T
n
with (a) j and (b) k. Since ,= , we have one (or both) of the following
cases: (a) a ,= b, or (b) j ,= k.
In case (a) we take
m = minn N :
n
,=
n
.
Without any loss of generality, we can assume that
m
= 0 and
m
= 1. Note that
k (b) m 1. We are going to prove that I
= , by showing that the

right end-point of I
is not greater than the left end-point of I
, that is,
(20) (a) +
2
3
k+1
(b) +
1
3
k+1
.
Dene the number
M =
m1
n=1
n
3
n
=
m1
n=1
n
3
n
,
with the convention that M = 0, if m = 1. We have:
(a) = 2M + 2
(a)
n=m+1
n
3
n
2M + 2
(a)
m+1
1
3
n
= 2M +
1
3
m

1
3
(a)
;
(b) = 2M +
2
3
m
+ 2
(b)
n=m+1
n
3
n
2M +
2
3
m
.
The inequality (20) then follows immediately from:
(a) +
2
3
j+1
2M +
1
3
m

1
3
(a)
+
2
3
j+1
<< 2M +
1
3
m

1
3
(a)
+
1
3
j

2M +
1
3
m
< 2M +
2
3
m
(b) < (b) +
1
3
k+1
.
In case (b), based on the fact that we have proven case (a), we can assume, without
any loss of generality, that a = b and j < k. In this case we have
(b) +
2
3
k+1
= (a) +
2
3
k+1
< (a) +
1
3
k
(a) +
1
3
j+1
,
which means that the right end-point of I
is not greater than the left end-point of

I
, so again we get I
= .
For the proof of (iii) we are going to use the space
P = 0, 1, 2
0
=
_
(
n
)
n=1
:
n
0, 1, 2, n N
_
.
Exactly as is the case with T, the product space P is compact with respect to the
product topology, which is given by the metric
d(a, b) =
n=1
[
n

n
[
2
n
, a = (
n
)
n=1
, b = (
n
)
n=1
P.
Then map : P [0, 1], dened by
(a) =
n=1
n
3
n
, a = (
n
)
n=1
P,
satises
(a) (b)[ d(a, b), a, b P,

hence it is continuous. Note also that is surjective. We can write = ,
where
: 0, 1
0
(
n
)
n=1
(2
n
)
n=1
0, 1, 2
0
.
Note also that : T P is continuous, since we clearly have
d
_
(a), (b)
_
2d(a, b), a, b T.
We now proceed with the proof of (iii). Denote the open set
simply by
D. Since T
n
is dense in T, it follows that (T
n
) is dense in K = (T). Therefore,
in order to prove the inclusion K [0, 1] D, using the surjectivity of , it suces
to prove the inclusion
(T
n
) [0, 1] D.
Using the map : P [0, 1], the above inclusion is equivalent to
(21) P (T
n
)
1
(D).
In order to prove the inclusion [0, 1] D K, again using the surjectivity of , it
suces to prove the inclusion
(22)
1
(D) P
1
(K).
To prove (21) start with some element a = (
n
)
n=1

1
(D), which means that
there exists some b T
n
, and an integer k (b), such that (a) I
(k,b)
, i.e.
(23)
2
1
3
+ +
2
k
3
k
+
1
3
k+1
<
n=1
n
3
n
<
2
1
3
+ +
2
k
3
k
+
2
3
k+1
.
We prove that a , (T
n
) by contradiction. Assume a (T
n
), which means that
there exists c = (
n
)
n=1
T
n
, such that
n
= 2
n
, n N. Dene the element
b = (
n
)
n=1
T
n
by
n
=
_
_
_
n
if n k
1 if n = k + 1
0 if n > k + 1
With this denition, the inequalities (23) give
(24) (b) < (b) +
1
3
k+1
< (c) < (
b).
By Fact 2 above, there exist N, N
N such that

N
= 1,
N
= 0, and
n
=
n
, for all n N with n < N;

N
= 0,

N
= 1, and
n
=

n
, for all n N with n < N
.
We will examine three cases: (a) N < N
, (b) N = N
, or (c) N > N
.
Case (b) is clearly impossible. In case (a), the inequality N < N
forces
N
= 0,
N
= 1 and

N
=
N
, which means that

N
= 1 ,=
N
= 0. This clearly
forces N = k + 1 > (b), which in particular gives
n
=

n
= 0, n > N, so we
clearly have
n

n
, n N, so we get (c) (
b), thus contradicting (24). In

case (c), we have
N
= 0,

N
= 1, and since N
< N, we also have

N
=
N
= 0.
As before this would force N
= k + 1. We then have
(c) = 2
n=1
n
3
n
= 2
N
n=1
n
3
n
+
2
N
3
N
+ 2
n=N
+1
n
3
n
= 2
k
n=1
n
3
n
+ 0 + 2
n=k+2
n
3
n
=
= (b) + 2
n=k+2
n
3
n
(b) + 2
n=k+1
1
3
n
= (b) +
1
3
k+1
,
again contradicting (24).
To prove (22), we start with some element a P
1
(K), and we show that
(a) D. The fact that a ,
1
(K) forces the fact that a , (T). In particular,
this gives the fact that a = (
n
)
n=1
0, 1, 2
0
and there exists some n N such
that
n
= 1. Put
N = minn N :
n
= 1.
Dene the elements b = (
n
)
n=1
0, 1
0
, by
n
=
_

n
/2 if n < N
0 if n N
Notice that b T
n
, and (b) N 1. Notice also that 2
n
=
n
, for all n N
with n < N 1. In particular, using the equality
N
= 1, this gives
(b) +
1
3
N
= 2
N1
n=1
n
3
n
+

N
3
N
=
N
n=1
n
3
n

n=1
n
3
n
= (a);
(25)
(b) +
2
3
N
= 2
N1
n=1
n
3
n
+

N
3
N
+
n=N+1
2
3
n
=
N
n=1
n
3
n
+
n=N+1
2
3
n

n=1
n
3
n
= (a).
(26)
Consider the pair = (N 1, ) . We are going to show that (a) I
, i.e.
we have the inequalities
(27) (b) +
1
3
N
< (a) < (b) +
2
3
N
.
By (25) and (26) it suces to prove only that
(a) ,= (b) +
1
3
N
and (a) ,= (b) +
2
3
N
.
If (a) = (b) +
1
3
N
, then by the inequalities (25), we are forced to have
(28)
n
= 0, n > N..
If (a) = (b) +
2
3
N
, then by the inequalities (26), we are forced to have
(29)
n
= 2, n > N..
If (28) holds, we dene c = (
n
)
n=1
T, by
n
=
_
_
_
n
/2 if n < N
0 if n = N
1 if n > N
and we will have
(c) = 2
n=1
n
3
n
=
N1
n=1
2
n
3
n
+ 2
n=N+1
1
3
n
=
N1
n=1
n
3
n
+
1
3
N
= (a),
thus forcing (a) K, which is impossible.
If (29) holds, we dene c = (
n
)
n=1
T, by
n
=
_
_
_
n
/2 if n ,= N
1 if n = N
0 if n > N
and we will have
(c) = 2
N1
n=1
n
3
n
+
2
3
N
=
N1
n=1
2
n
3
n
+
1
3
N
+
n=N+1
2
3
n
=
n=1
n
3
n
= (a),
thus forcing again (a) K, which is impossible.
Exercise 7. Using the notations above, prove that the set
[0, 1] K
3
=
_
is dense in [0, 1].

Hints: Dene the set
P
0
=
(n)
n=1
{0, 1, 2}
0
: the set {n N : n = 1} is innite
.
Prove that P
0
is dense in P, and prove that (P) [0, 1] K. (Use the arguments employed in
the proof of part (iii).)
Remarks 3.5. If we set
n
=
_
nP
_
, then we can write the complement
of the ternary Cantor set as
[0, 1] K
3
=
_
n=0
D
n
,
where
D
n
=
_
n
I
.
Then the system of open sets (D
n
)
n0
is pair-wise disjoint. Morever, each D
n
is a
union of 2
n
disjoint intervals of length 1/3
n+1
.
Since cardT
0
= c, and the map
3
T0
: T
0
K
3
is injective, we get cardK
3
c.
Since we also have cardK
3
cardR = c, we get in fact the equality
cardK
3
= c.

Measurable Spaces and Measurable Maps

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Measurable Spaces and Measurable Maps

Uploaded by

Copyright:

Available Formats

3.

Measurable spaces and measurable maps

is any -algebra, with A

A, then the identity map Id

(Z, C) are measurable maps, then the composition S T : (X, A) (Z, C)

E A. Since A is a -algebra, we get the inclusion

(z) = 0}. Since f is non-constant,

(x) = maxf(x), 0 and f

are measurable. Since inf

(x), x X. Since sup

(x) 0, by part (ii), there exists a sequence

(x), x X. Dene the

(x) = f(x) and f

(x) = f(x), so lim

= , by showing that the

is not greater than the left end-point of I

is not greater than the left end-point of

(a) (b)[ d(a, b), a, b P,

b), thus contradicting (24). In

< N, we also have

is dense in [0, 1].

You might also like