Professional Documents
Culture Documents
|
E
:= sup
x1
[x, x
)[.
Here, x, x
) := x
| instead of |x
|
E
if no confusion
can arise. The elements of E
) a < b Rey, x
)
2 1 Integration in Banach spaces
for all x C and y K. As is well-known, from this one derives the Hahn-
Banach extension theorem: if F is a closed subspace of E, then for all y
there exists an x
such that x
[
F
= y
and |x
| = |y
|. This easily
implies that for all x E we have
|x| = sup
x
1
[x, x
)[.
A linear subspace F of E
F
x
1
[x, x
)[.
A subspace of E
n=1
in E
0
and choose a sequence of unit
vectors (x
n
)
n=1
in F such that [x
n
, x
n
)[ (1
n
)|x
n
| for all n 1, where
the numbers 0 <
n
1 satisfy lim
n
n
= 0. The sequence (x
n
)
n=1
is
norming for E
0
. To see this, x an arbitrary x E
0
and let > 0. Pick
n
0
1 such that 0 <
n0
and |x x
n0
| . Then,
(1 )|x| (1
n0
)|x| (1
n0
)|x
n0
| + (1
n0
)
[x
n0
, x
n0
)[ + [x, x
n0
)[ + 2.
Since > 0 was arbitrary it follows that |x| sup
n1
[x, x
n
)[.
A linear subspace F of E
F with x, x
) , = y, x
).
Clearly, norming subspaces separate points, but the converse need not be true.
Lemma 1.2. If E
0
is a separable subspace of E and F is a linear subspace of
E
(x)) , = 0. Dening
V
x
:= y E
0
0 : y, x
(x)) , = 0
we obtain an open cover V
x
xE0\{0}
of E
0
0. Since every open cover of a
separable metric space admits a countable subcover it follows that there exists
a sequence (x
n
)
n=1
in E
0
0 such that V
xn
n=1
covers E
0
0. Then the
sequence x
(x
n
)
n=1
separates the points of E
0
: indeed, every x E
0
0
belongs to some V
xn
, which means that x, x
(x
n
)) , = 0.
1.2 The Pettis measurability theorem 3
1.2 The Pettis measurability theorem
We begin with a discussion of weak and strong measurability of E-valued
functions. The main result in this direction is the Pettis measurability the-
orem which states, roughly speaking, that an E-valued function is strongly
measurable if and only if it is weakly measurable and takes its values in a
separable subspace of E.
1.2.1 Strong measurability
Throughout this section (A, A) denotes a measurable space, that is, A is a
set and A is a -algebra in A, that is, a collection of subsets of A with the
following properties:
1. A A;
2. B A implies B A;
3. B
1
A, B
2
A, . . . imply
n=1
B
n
A.
The rst property guarantees that A is non-empty, the second expresses that
A is closed under taking complements, and the third that A is closed under
taking countable unions.
The Borel -algebra of a topological space T, notation B(T), is the smallest
-algebra containing all open subsets of T. The sets in B(T) are the Borel
sets of T.
Denition 1.3. A function f : A T is called A-measurable if f
1
(B) A
for all B B(T).
The collection of all B B(T) satisfying f
1
(B) A is easily seen to be
a -algebra. As a consequence, f is A-measurable if and only if f
1
(U) A
for all open sets U in T.
When T
1
and T
2
are topological spaces, a function g : T
1
T
2
is Borel
measurable if g
1
(B) B(T
1
) for all B B(T
2
), that is, if g is B(T
1
)-
measurable. Note that if f : A T
1
is A-measurable and g : T
1
T
2
is
Borel measurable, then the composition g f : A T
2
is A-measurable.
By the above observation, every continuous function g : T
1
T
2
is Borel
measurable.
It is a matter of experience that the notion of A-measurability does not
lead to a satisfactory theory from the point of view of vector-valued analysis.
Indeed, the problem is that this denition does not provide the means for
approximation arguments. It is for this reason that we shall introduce next
another notion of measurability. We shall restrict ourselves to Banach space-
valued functions, although some of the results proved below can be generalised
to functions with values in metric spaces.
Let E be a Banach space and (A, A) a measurable space. A function
f : A E is called A-simple if it is of the form f =
N
n=1
1
An
x
n
with
A
n
A and x
n
E for all 1 n N. Here 1
A
denotes the indicator
function of the set A, that is, 1
A
() = 1 if A and 1
A
() = 0 if , A.
4 1 Integration in Banach spaces
Denition 1.4. A function f : A E is strongly A-measurable if there
exists a sequence of A-simple functions f
n
: A E such that lim
n
f
n
= f
pointwise on A.
In order to be able to characterise strong A-measurability of E-valued
functions we introduce some terminology. A function f : A E is called
separably valued if there exists a separable closed subspace E
0
E such
that f() E
0
for all A, and weakly A-measurable if the functions
f, x
) : A K, f, x
)() := f(), x
.
Theorem 1.5 (Pettis measurability theorem, rst version). Let (A, A)
be a measurable space and let F be a norming subspace of E
. For a function
f : A E the following assertions are equivalent:
(1) f is strongly A-measurable;
(2) f is separably valued and f, x
;
(3) f is separably valued and f, x
F.
Proof. (1)(2): Let (f
n
)
n=1
be a sequence of A-simple functions converging
to f pointwise and let E
0
be the closed subspace spanned by the countably
many values taken by these functions. Then E
0
is separable and f takes its
values in E
0
. Furthermore, each f, x
).
(2)(3): This implication is trivial.
(3)(1): Using Lemma 1.1, choose a sequence (x
n
)
n=1
of unit vectors in
F that is norming for a separable closed subspace E
0
of E where f takes its
values. By the A-measurability of the functions f, x
n
), for each x E
0
the
real-valued function
|f() x| = sup
n1
[f() x, x
n
)[
is A-measurable. Let (x
n
)
n=1
be a dense sequence in E
0
.
Dene the functions s
n
: E
0
x
1
, . . . , x
n
as follows. For each y E
0
let k(n, y) be the least integer 1 k n with the property that
|y x
k
| = min
1jn
|y x
j
|
and put s
n
(y) := x
k(n,y)
. Notice that
lim
n
|s
n
(y) y| = 0 y E
0
since (x
n
)
n=1
is dense in E
0
. Now dene f
n
: A E by
f
n
() := s
n
(f()), A.
For all 1 k n we have
1.2 The Pettis measurability theorem 5
A : f
n
() = x
k
=
_
A : |f() x
k
| = min
1jn
|f() x
j
|
_
_
A : |f() x
l
| > min
1jn
|f() x
j
| for l = 1, . . . , k 1
_
.
Note that the sets on the right hand side are in A. Hence each f
n
is A-simple,
and for all A we have
lim
n
|f
n
() f()| = lim
n
|s
n
(f()) f()| = 0.
Corollary 1.6. The pointwise limit of a sequence of strongly A-measurable
functions is strongly A-measurable.
Proof. Each function f
n
takes its values in a separable subspace of E. Then
f takes its values in the closed linear span of these spaces, which is separable.
The measurability of the functions f, x
)
is the pointwise limit of the measurable functions f
n
, x
).
Corollary 1.7. If an E-valued function f is strongly A-measurable and :
E F is continuous, where F is another Banach space, then f is strongly
A-measurable.
Proof. Choose simple functions f
n
converging to f pointwise. Then f
n
f pointwise and the result follows from the previous corollary.
Proposition 1.8. For a function f : A E, the following assertions are
equivalent:
(1) f is strongly A-measurable;
(2) f is separably valued and for all B B(E) we have f
1
(B) A.
Proof. (1)(2): Let f be strongly A-measurable. Then f is separably-valued.
To prove that f
1
(B) A for all B B(E) it suces to show that f
1
(U)
A for all open sets U.
Let U be open and choose a sequence of A-simple functions f
n
converging
pointwise to f. For r > 0 let U
r
= x U : d(x, U) > r, where U denotes
the complement of U. Then f
1
n
(U
r
) A for all n 1, by the denition of
an A-simple function. Since
f
1
(U) =
_
m1
_
n1
kn
f
1
k
(U 1
m
)
(the inclusion being a consequence of the fact that U is open) it follows
that also f
1
(U) A.
(2)(1): By assumption, f is A-measurable, and therefore f, x
) is A-
measurable for all x
n=1
A
(n)
.
A -simple function with values in E is a function of the form
f =
N
n=1
1
An
x
n
,
where x
n
E and the sets A
n
A satisfy (A
n
) < .
We say that a property holds -almost everywhere if there exists a -null
set N A such that the property holds on the complement N of N.
Denition 1.9. A function f : A E is strongly -measurable if there
exists a sequence (f
n
)
n1
of -simple functions converging to f -almost ev-
erywhere.
Using the -niteness of it is easy to see that every strongly A-
measurable function is strongly -measurable. Indeed, if f is strongly A-
measurable and lim
n
f
n
= f pointwise with each f
n
an A-simple func-
tions, then also lim
n
1
A
(n) f
n
= f pointwise, where A =
n=1
A
(n)
as
before, and each 1
A
(n) f
n
is -simple. The next proposition shows that in the
converse direction, every strongly -measurable function is equal -almost
everywhere to a strongly A-measurable function.
Let us call two functions which agree -almost everywhere -versions of
each other.
Proposition 1.10. For a function f : A E the following assertions are
equivalent:
(1) f is strongly -measurable;
(2) f has a -version which is strongly A-measurable.
Proof. (1)(2): Suppose that f
n
f outside the null set N A, with each
f
n
-simple. Then we have lim
n
1
N
f
n
= 1
N
f pointwise on A, and since
the functions 1
N
f
n
are A-simple, 1
N
f is strongly A-measurable. It follows
that 1
N
f is a strongly A-measurable -version of f.
(2)(1): Let
f be a strongly A-measurable -version of f and let N A
be a null set such that f =
f on N. If (
f
n
)
n=1
is a sequence of A-simple
functions converging pointwise to
f, then lim
n
f
n
= f on N, which means
that lim
n
f
n
= f -almost everywhere.
Write A =
n=1
A
(n)
with A
(1)
A
(2)
A and (A
(n)
) < for all
n 1. The functions f
n
:= 1
A
(n)
f
n
are -simple and we have lim
n
f
n
= f
-almost everywhere.
1.3 The Bochner integral 7
We say that f is -separably valued if there exists a closed separable sub-
space E
0
of E such that f() E
0
for -almost all A, and weakly -
measurable if f, x
.
Theorem 1.11 (Pettis measurability theorem, second version). Let
(A, A, ) be a -nite measure space and let F be a norming subspace of E
.
For a function f : A E the following assertions are equivalent:
(1) f is strongly -measurable;
(2) f is -separably valued and f, x
;
(3) f is -separably valued and f, x
F.
Proof. The implication (1)(2) follows the corresponding implication in The-
orem 1.5 combined with Proposition 1.10, and (2)(3) is trivial. The impli-
cation (3)(1) is proved in the same way as the corresponding implication
in Theorem 1.5, observing that this time the functions f
n
have -versions
f
n
that are A-simple. If we write A =
n=1
A
(n)
as before with each A
(n)
of -
nite -measure, the functions 1
A
(n)
f
n
are -simple and converge to f -almost
everywhere.
By combining Proposition 1.10 with Corollaries 1.6 and 1.7 we obtain:
Corollary 1.12. The -almost everywhere limit of a sequence of strongly -
measurable E-valued functions is strongly -measurable.
Corollary 1.13. If an E-valued function f is strongly -measurable and :
E F is continuous, where F is another Banach space, then f is strongly
-measurable.
The following result will be applied frequently.
Corollary 1.14. If f and g are strongly -measurable E-valued functions
which satisfy f, x
) = g, x
F, where F
is subspace of E
n
)
n=1
in F separates the points of
E
0
. Since f, x
n
) = g, x
n
) outside a -null set N
n
, we conclude that f and g
agree outside the -null set N
n=1
N
n
.
1.3 The Bochner integral
The Bochner integral is the natural generalisation of the familiar Lebesgue
integral to the vector-valued setting.
Throughout this section, (A, A, ) is a -nite measure space.
8 1 Integration in Banach spaces
1.3.1 The Bochner integral
Denition 1.15. A function f : A E is -Bochner integrable if there
exists a sequence of -simple functions f
n
: A E such that the following
two conditions are met:
(1) lim
n
f
n
= f -almost everywhere;
(2) lim
n
_
A
|f
n
f| d = 0.
Note that f is strongly -measurable. The functions |f
n
f| are -measurable
by Corollary 1.13.
It follows trivially from the denitions that every -simple function is -
Bochner integrable. For f =
N
n=1
1
An
x
n
we put
_
A
f d :=
N
n=1
(A
n
)x
n
.
It is routine to check that this denition is independent of the representation
of f. If f is -Bochner integrable, the limit
_
A
f d := lim
n
_
A
f
n
d
exists in E and is called the Bochner integral of f with respect to . It is routine
to check that this denition is independent of the approximating sequence
(f
n
)
n=1
.
If f is -Bochner integrable and g is a -version of f, then g is -Bochner
integrable and the Bochner integrals of f and g agree. In particular, in the
denition of the Bochner integral the function f need not be everywhere
dened; it suces that f be -almost everywhere dened.
If f is -Bochner integrable, then for all x
_
=
_
A
f, x
) d.
For -simple functions this is trivial, and the general case follows by approx-
imating f with -simple functions.
Proposition 1.16. A strongly -measurable function f : A E is -
Bochner integrable if and only if
_
A
|f| d < ,
and in this case we have
_
_
_
_
A
f d
_
_
_
_
A
|f| d.
1.3 The Bochner integral 9
Proof. First assume that f is -Bochner integrable. If the -simple functions
f
n
satisfy the two assumptions of Denition 1.15, then for large enough n we
obtain _
A
|f| d
_
A
|f f
n
| d +
_
A
|f
n
| d < .
Conversely let f be a strongly -measurable function satisfying
_
A
|f| d <
. Let g
n
be -simple functions such that lim
n
g
n
= f -almost every-
where and dene
f
n
:= 1
{gn2f}
g
n
.
Then f
n
is -simple, and clearly we have lim
n
f
n
= f -almost everywhere.
Since we have |f
n
| 2|f| pointwise, the dominated convergence theorem can
be applied and we obtain
lim
n
_
A
|f
n
f| d = 0.
The nal inequality is trivial for -simple functions, and the general case
follows by approximation.
As a simple application, note that if f : A E is -Bochner integrable,
then for all B A the truncated function 1
B
f : A E is -Bochner inte-
grable, the restricted function f[
B
: B E is [
B
-Bochner integrable, and
we have _
A
1
B
f d =
_
B
f[
B
d[
B
.
Henceforth, both integrals will be denoted by
_
B
f d.
In the following result, conv(V ) denotes the convex hull of a subset V E,
i.e., the set of all nite sums
k
j=1
j
x
j
with
j
0 satisfying
k
j=1
j
= 1
and x
j
V for j = 1, . . . , k. The closure of this set is denoted by conv(V ).
Proposition 1.17. Let f : A E be a -Bochner integrable function. If
(A) = 1, then
_
A
f d convf() : A.
Proof. Let us say that an element x E is strictly separated from a set V E
by a functional x
) Rev, x
)[ v V.
The Hahn-Banach separation theorem asserts that if V is convex and x , V ,
then there exists a functional x
, let
m(x
) := infRef(), x
) : A,
M(x
) := supRef(), x
) : A,
10 1 Integration in Banach spaces
allowing these values to be and , respectively. Then, since (A) = 1,
Re
_
_
A
f d, x
_
=
_
A
Ref, x
) d [m(x
), M(x
)].
This shows that
_
A
f d cannot be strictly separated from the convex set
convf() : A by functionals in E
)
L
p
(A) for all x
) =
_
A
f, x
) d.
Proof. We may assume that f is strongly A-measurable.
It is easy to see that the linear mapping S : E
L
p
(A), Sx
:= f, x
)
is closed. Hence S is bounded by the closed graph theorem.
Put A
n
:= |f| n. Then A
n
A and by Proposition 1.16 the integral
_
An
f d exists as a Bochner integral in E. For all x
and n m, by
H olders inequality we have
12 1 Integration in Banach spaces
_
_
An\Am
f d(x), x
_
(A
n
A
m
)
_1
q
_
_
A
[f, x
)[
p
d(x)
_1
p
_
(A
n
A
m
)
_1
q
|S| |x
|.
Taking the supremum over all x
with |x
| 1 we see that
limsup
m,n
_
_
_
_
An\Am
f d
_
_
_ lim
m,n
_
(A
n
A
m
)
_1
q
|S| = 0.
Hence the limit x
f
:= lim
n
_
An
f d exists in E. Clearly,
x
f
, x
) = lim
n
_
An
f, x
) d =
_
A
f, x
) d
for all x
(A;E)
:= inf
_
r 0 : |f| > r = 0
_
,
the space L
n
)
n=1
of norm one elements in
E
and a sequence (F
n
)
n=1
of closed sets in K such that
C =
n=1
_
x E : x, x
n
) F
n
_
.
Hint: Separate C from the elements of a dense sequence in its complement
C using the Hahn-Banach separation theorem.
2. Prove that the function f : (0, 1) L
. Prove that L
q
(A; F) is a norming subspace of (L
p
(A; E))
with
respect to the duality pairing
f, g) =
_
A
f(), g()) d(), f L
p
(A; E), g L
q
(A; E
).
Hint: First nd simple functions in L
q
(A; F) which norm simple functions
in L
p
(A; E).
Notes. The material in this lecture is standard and can be found in many
textbooks. More complete discussions of measurability in Banach spaces can
be found in the monographs by Bogachev [8] and Vakhania, Tarieladze,
Chobanyan [105]. Systematic expositions of the Bochner integral are pre-
sented in Arendt, Batty, Hieber, Neubrander [3], Diestel and Uhl
[36], Dunford and Schwartz [37] and Lang [66].
1
Results proved in the exercises marked with (!) are needed in the main text.
14 1 Integration in Banach spaces
The Pettis measurability theorems 1.5 and 1.11 as well as Theorem 1.20
are due to Pettis [90]. Both versions of the Pettis measurability theorem
remain correct if we only assume f to be weakly measurable with respect to
the functionals from a subspace F of E