You are on page 1of 13

Appendix D

Vector Lattices

[§] 1. Definitions
A vector space V equipped with a partial order “≤” is called a vector lattice
if for each pair x, y in V
(i) there is a smallest element z (denoted by x ∨ y) for which x ≤ z and
y ≤ z,
(ii) there is a largest element w (denoted by x ∧ w) for which x ≤ w and
y≤w
(iii) if x ≤ y then x + z ≤ y + z for all z ∈ V
(iv) if x ≤ y and c ∈ R+ then cx ≤ cy
The element |x| := x ∨ (−x) is called the modulus of x. The element
x + := x ∨ 0 is called the positive part of x. The element x − := (−x) ∨ 0 is
called the negative part of x.
If a vector lattice V is equipped with a norm
·
for which
(v) for all x, y ∈ V , if |x| ≤ |y| then
x

y

then V (equipped with ≤ and


·
) is called a normed vector lattice.
L.infty <1> Example. Let S be a set equipped with a sigma-field S. The space L∞ (S, S)
of all bounded, S-measurable real functions on S is a vector lattice for the
pointwise ordering:
f ≤g means f (s) ≤ g(s) for all s in S
The supremum norm, defined by

f
= sup | f (s)| ,
s∈S

satisfies property (v). The space L (S, S) is a normed vector lattice.
The space L∞ (S, S) is also complete for the supremum norm, in the sense
that Cauchy sequences converge to bounded, measurable limit functions. A
complete normed vector lattice is called a Banach lattice.
If S is equipped with a topology, the set C(S) of all bdd, continuous real
functions on S is also a Banach lattice under the pointwise ordering and the
supremum norm. If all functions in C(S) are S-measurable (for example, S
could be the Borel sigma-field) then C(S) is a Banach sublattice of L∞ (S, S).


[§] 2. Basic facts

Asymptopia: 4Jan99
c David Pollard 1
Section D.2 Basic facts

Fact A: For real c, and x, y ∈ V ,



c(x ∧ y) for c ≥ 0
(cx) ∧ (cy) =
c(x ∨ y) for c ≤ 0

Fact B: For x, y, z ∈ V ,
(x + z) ∧ (y + z) = (x ∧ y) + z

Fact C: For x, y, z ∈ V ,
x ∧ y + x ∨ y = x + y + (x − x − y) ∧ (y − x − y) − (−x) ∧ (−y)
=x+y

Fact D: For each x ∈ V ,
2x + = (x + x) ∨ (x − x) = x + (x ∨ (−x)) = x + |x|
2x − = (−x − x) ∨ (x − x) = −x + ((−x) ∨ x) = −x + |x|
By addition and subtraction.
x = x+ − x− |x| = x + + x −
From (2x + ) ∨ (2x − ) = (x + |x|) ∨ (−x + |x|) = x ∨ (−x) + |x| = 2|x| deduce
x + ∨ x − = |x| = x + + x −
x + ∧ x − = x + + x − − (x + ∨ x − ) = 0

singular <2> Definition. Elements x and y in V are said to be orthogonal (or lattice
disjoint) if |x| ∧ |y| = 0. Write x ⊥ y to denote orthogonality. define subsets
X and Y of V to be orthogonal if x ⊥ y for each x in X and y in Y .
abs.triangle < 3> Lemma. For all x and y in V ,
|x + y| ≤ |x| + |y| with equality if and only if |x| ∧ |y| = 0
Proof. Argue first for the inequality.
|x + y| = (x + y) ∨ (−x − y)
= (x + |x| + y + |y|) ∨ (|x| − x + |y| − y) − |x| − |y|
= (2x + + 2y + ) ∨ (2x − + 2y − ) − |x| − |y|
 
= 2 (x + ∨ y + − x + ∧ y + ) ∨ (x − ∨ y − − x − ∧ y − ) − |x| − |y|
Discard the nonnegative quantities x + ∧ y + and x − ∧ y − , leaving the upper
bound
2(x + ∨ y + ∨ x − ∨ y − ) − |x| − |y| = 2(|x| ∨ |y|) − |x| − |y|
= 2(|x| + |y| − |x| ∧ |y|) − |x| − |y|
≤ |x| + |y|
The last inequality is an equality if and only if |x| ∧ |y| = 0. Moreover, if
|x| ∧ |y| = 0 then both the terms x + ∧ y + and x − ∧ y − are also zero, which
 implies equality at all steps.
splitting.lemma <4> Splitting Lemma. If x1 , x2 , y1 , y2 ∈ V + and x1 + x2 = y1 + y2 then there
exist xi j ∈ V + such that
u 11 + u 21 = x1 u 11 + u 12 = y1
u 12 + u 22 = x2 u 21 + u 22 = y2
2 Asymptopia: 4Jan99
c David Pollard
Appendix D Vector Lattices

Proof. Define the u i j as in the following table:


u 11 = x1 ∧ y1 u 12 = (y1 − x1 )+ y1
+
u 21 = (y2 − x2 ) u 22 = y2 ∧ x2 y2
x1 x2 x1 + x2 = y1 + y2

Check the first row sum:


x1 ∧ y1 + (y1 − x1 )+ = (x1 ∧ y1 ) + (y1 ∨ x1 ) − x1 = x1 + y1 − x1
Interchange subscripts 1 and 2 to check the second row sum. Interchange the
roles of x and y, then use the fact that y2 − x2 = x1 − y1 to check the column
 sums.
splitting.ineq <5> Corollary. If w, x1 , x2 ∈ V + and w ≤ x1 + x2 then there exist wi ∈ V +
such that w = w1 + w2 and wi ≤ xi for i = 1, 2.
 Proof. let y1 = w and y2 = x1 + x2 − w.
sum.orthog <6> Exercise. Suppose x ∈ V and y1 , y2 ∈ V + and x ⊥ yi for i = 1, 2. Show
that x ⊥ (y1 + y2 ) and x ⊥ (y1 − y2 ).
Solution: Suppose 0 ≤ w = |x| ∧ (y1 + y2 ). Split w into a sum w1 + w2 ,
with 0 ≤ wi ≤ yi . Note wi ≤ |x|.
w = w1 ∧ |x| + w2 ∧ |x| ≤ y1 ∧ |x| + y2 ∧ |x| = 0.
Thus |x| ⊥ (y1 + y2 ).
The second assertion follows from the first and the fact that |y1 − y2 | ≤
 y1 + y2 .

[§] 3. Order-bounded linear functionals


let V be a vector lattice. A linear map λ from V into R is to be an order
bounded linear functional if, for each pair a ≤ b,
sup{|λ(x)| : a ≤ x ≤ b} < ∞
A linear map λ from V into R is to be an increasing linear functional if
λ(x) ≤ λ(y) whenever x ≤ y. Equivalently, a linear functional is increasing if
λ(x) ≥ 0 for all x ∈ V + . The space V # of all order-bounded linear functionals
on V is called the order dual of V .
fnal.ordering <7> Definition. Define λ1 ≤ λ2 to mean that λ1 (x) ≤ λ2 (x) for all x in V + .
The key facts about linear functionals are that
(i) The order dual V # is a vector lattice.
(ii) Each λ in V # can be expressed as a difference of two increasing linear
functionals.

extension <8> Lemma. Let λ : V + → R satisfy


(i) λ(αx + βy) = αλ(x) + βλ(y) for all α, β ∈ R+ and x, y ∈ V +
(ii) For each w in V + ,
sup{|λ(x)| : 0 ≤ x ≤ w} < ∞
Then λ has a unique extension to an order-bounded linear functional on V .
Proof. Define λ(u) = λ(u + ) − λ(u − ). Check that if x1 − x2 = y1 − y2 then
λ(x1 ) − λ(x2 ) = λ(y)1 − λ(y2 ). Deduce linearity.

Asymptopia: 4Jan99
c David Pollard 3
Section D.3 Order-bounded linear functionals

For order bounded: if a ≤ x ≤ b then 0 ≤ x + ≤ b+ and 0 ≤ x − ≤ a − ,


whence
|λ(x)| ≤ |λ(x + )| + |λ(x − )|
≤ sup{|λ(y)| : 0 ≤ y ≤ b+ } + sup{|λ(y)| : 0 ≤ y ≤ a − } < ∞.

By virtue of this extension result, there is no harm in referring to the
restriction of an order-bounded linear functional to V + as an order-bounded
linear functional on V + . If such a functional maps V + into R+ then it is
increasing.
fnal.max <9> Theorem. Let λ1 and λ2 be order-bounded linear functionals. Then, for
v ∈ V +,
(i) (λ1 ∨ λ2 )(v) = sup{λ1 v1 + λ2 v2 : vi ∈ V + and v1 + v2 = v}
(ii) (λ1 ∧ λ2 )(v) = inf{λ1 v1 + λ2 v2 : vi ∈ V + and v1 + v2 = v}
The space V # is a vector lattice.
Proof. Consider only the assertion for maxima of the two functionals. The
arguments for minima are analogous. Write µ(v) for the expression on the
right-hand side of (i).
First show that µ is a linear functional, in the sense of Lemma <8>. It is
trivial to check that µ(cv) = cµ(v) for all v ∈ V + and c ∈ R+ . To establish
additivity,
µ(v + w) = µ(v) + µ(v) for all v, w ∈ V + ,
consider pairs in V + with v1 + v2 = v and w1 + w2 = w. For such pairs,
λ1 (v1 ) + λ2 (v2 ) + λ1 (w1 ) + λ2 (w2 ) = λ1 (v1 + w1 ) + λ2 (v2 + w2 )
Because v1 + w1 + v2 + w2 = v + w, the right-hand side is less than µ(v + w).
Take the supremum over all pairs to deduce that
µ(v) + µ(w) ≤ µ(v + w)
For the reverse inequality, consider any pair in V + for which u 1 + u 2 = v + w.
Invoke the Splitting Lemma <4> to break u 1 into v1 + w1 and u 2 into v2 + w2
such that v1 + v2 = v and w1 + w2 = w. Then
λ1 u 1 + λ2 u 2 = λ1 v − 1 + λ2 v2 + λ1 w1 + λ2 w2
≤ µ(v) + µ(w)
Take a suprememum over u 1 , u 2 to deduce that
µ(v + w) ≤ µ(v) + µ(w)
Additivity follows.
Next, check that µ is order bounded, in the sense of Lemma <8>.
Consider v with 0 ≤ v ≤ w, for fixed w in V + . The supremum defining µ(v)
is smaller in absolute value than
sup (|λ1 (v1 )| + |λ(v2 )|}) ≤ sup |λ1 (u)| + sup |λ2 (u)| < ∞
v1 +v2 =v 0≤u≤w 0≤u≤w

The linear functional µ belongs to V . #

If λ ∈ V # and λ ≥ λ1 and λ ≥ λ2 , then, for v ∈ V + ,


µ(v) ≤ sup{λv1 + λv2 : vi ∈ V + and v1 + v2 = v} = λ(v)
That is, µ ≤ λ; the order bounded linear functional µ is the least upper bound
 λ1 ∨ λ2 .
The lattice decomposition λ = λ+ − λ− expresses a general λ in V # as a
difference of two increasing functionals.

4 Asymptopia: 4Jan99
c David Pollard
Appendix D Vector Lattices

abs.lam <10> Exercise. For λ in V # and w in V + , show that


|λ|(w) = sup λ(u) = sup |λ(u)|
|u|≤w |u|≤w

Solution: From the definition of |λ| as λ ∨ (−λ) we have


|λ|(w) = sup{λw1 − λw2 : wi ∈ V + and w1 + w2 = w}
= sup (λv − λ(w − v))
0≤v≤w

Substitute u for 2v − w, noting that 0 ≤ v ≤ w if and only if −w ≤ u ≤ w if


and only if |u| ≤ w, to reexpress last supremum as
sup λ(u)
|u|≤w

 The last equality is trivial, because |u| = | − u|.

[§] 4. Lattices of bounded functions


Let  be a vector lattice of bounded, real functions on a set S. Assume 
contains the constant function 1. Equip  with its supremum norm,

γ
= sup |γ (s)|.
s∈S

Assume  is closed (in the space of all bounded functions on S) for uniform
convergence. Then  is a Banach lattice.
The norm on  has the special property that
M.norm <11>
γ1 ∨ γ2
= max (
γ1
,
γ2
) if γ1 ≥ 0 and γ2 ≥ 0
Such an equality makes  an example of an abstract M-space. Normed
vector lattices satisfying an equality like <11> share many of the properties
of spaces of bounded functions under a supremum norm. In fact, under mild
conditions (Schaefer 1974, Section II.7), abstract M-spaces can be represented
as spaces of bounded functions (in fact, the space of all bounded, continuous
real functions on some compact space).
bdd.meas <12> Example. Let S be the sigma-field on S generated by . If  contains all
indicator functions of S-measurable sets, then it must coincide with the space
bm(S, S) of all bounded, S-measurable, real functions on S, because each
function in bm(S, S) is a uniform limit of simple functions, i αi Ai ∈ .
The example of C[0, 1], the space of all bounded, continuous, real
functions on [0, 1], shows that  need not always contain all bounded,
 measurable functions.
A linear map λ from  to R is continuous if and only if

λ
:= sup{|λ(γ )| :
γ
≤ 1} < ∞.
It is traditional to write  ∗ for the space of all such continuous linear functionals.
meas.fnal <13> Example. If µ = µ1 − µ2 is a signed measure (expressed as a difference of
two nonnegative measures µ1 and µ2 ), defined on a sigma-field S for which all
functions in  are S-measurable, then the integral γ → µ(g) := µ1 (γ ) − µ2 (γ )
defines a continuous linear functional on , because
sup |µ(γ )| ≤ µ1 (1) + µ2 (1) < ∞.

γ
≤1

Suppose now that  = bm(S, S), the space of all bounded, S-measurable,
real functions on S.

Asymptopia: 4Jan99
c David Pollard 5
Section D.4 Lattices of bounded functions

Let µ = µ1 − µ2 and ν = ν1 − ν2 be finite signed measures. Let


Q = µ1 + µ2 + ν1 + ν2 . Each µi and νi is absolutely continuous with respect
to Q. Write m i and n i for the corresponding densities. Then µ has density
m = m 1 − m 2 and ν has density n = n 1 − n 2 , both with respect to Q.
The lattice-theoretic maximum of µ and ν has a closed-form expression::
for γ ∈  + ,
sup {µγ1 + νγ2 } = sup Q (mγ1 + nγ2 ) = Q ((m ∨ n)γ ) ,
γ1 +γ2 =γ γ1 +γ2 =γ

the supremum being achieved at γ1 = {m ≥ n}γ and γ2 = {m < n}γ . In


particular, |µ| is the countably additive measure with density |m| with respect
to Q. Similarly, µ+ has denity m + , and µ− has denisty m − .
The measures µ and ν are said to be mutually singular if there exists a
set S0 ∈ S such that |µ|(S0 ) = 0 = |ν|(S0c ). Equivalently, Q(|m| ∧ |n|) = 0, or
|µ| ∧ |ν| = 0. That is, mutual singularity is the same concept as orthogonality
 of the corresponding linear functionals.
orderbdd.cts <14> Theorem.
(i) The spaces  ∗ and  # are the same.
(ii)
λ
= |λ|(1) for each λ in  ∗ .
(iii) The space  ∗ is a Banach lattice.
Proof. Notice that the inequalities
γ
≤ 1 and |γ | ≤ 1, for a function γ in
 place the same constraint on γ , namely, that it is everywhere bounded in
absolute value by the constant 1. (Notice though that 1 denotes a real number
in the first inequality and a constant function in the second inequality.) Thus
norm.abs <15>
λ
= sup{|λ(γ )| : −1 ≤ γ ≤ 1}
If λ is order bounded then the right-hand side of <15> is finite, and hence
λ ∈  ∗ . Conversely, if γ1 and γ2 are functions in , both bounded in absolute
value by a constant C, then
sup{|λ(γ )| : γ1 ≤ γ ≤ γ2 } ≤ C sup{|λ(γ )| : −1 ≤ γ ≤ 1}
Finiteness of
λ
therefore implies order boundedness.
Exercise <10>, with w equal to 1, establishes assertion (ii).
The Banach lattice property, that
λ1

λ2
whenever |λ1 | ≤ |λ2 |,
 follows directly from (ii).
Assertion (ii) of the Theorem has another consequence,
L.norm <16>
λ1 + λ2
=
λ1
+
λ2
if λ1 ≥ 0 and λ2 ≥ 0.
Such an equality makes  an example of an abstract L-space. Normed
#

vector lattices satisfying an equality like <16> share many of the properties
of spaces of finite signed measures. In fact every L-space can be represented
as a space L 1 (µ), for some Radon measure on a locally compact space. In
this sense, evry abstract L-space corresponds to the collection of finite signed
measures that are absolutely continuous with respect to a fixed measure (on a
very nice topological space).
It is a little mysterious where the countable additivity comes from, for
it is not so hard to construct examples where the functionals in  # are not
represented by signed measures living on a sigma-field on S. The trick lies in
the choice of the space where µ lives. A small analogy might help.

6 Asymptopia: 4Jan99
c David Pollard
Appendix D Vector Lattices

analogy <17> Example. Let S = [0, 1) and let  denote the space of all functions
obtainable by restricting continuous real functions on [0, 1] to the subinterval
[0, 1). For each γ in  define
λ(g) = lim γ (s)
s→1
It is easy to see check that λ is an increasing linear functional on , but it does
not correspond to an integral with respect to any countably additive measure:
the sequence
γn (s) = (n − ns)) ∧ 1
is nonnegative and it increases pointwise to the constant function 1 function
on S, but λ(γn ) ≡ 0, which does not converge to λ(1); monotone convergence
is violated.
Of course λ does correspond to the integral with respect to the point mass
sitting at 1, but that measure does not live on S. We need a larger space to
 support the representing measures.
Lucien Le Cam uses subsets of abstract L-spaces to represent statistical
experiments. He has argued (Le Cam 1986, Chapter 1), with some justification
I believe, that many of the technical problems associated with the traditional
setting of probability measures on a sigma-field of a given sample space can be
traced to an unfortunate choice of sample space. He has pointed out that the
L-space structure is what really matters, and that there might be many ways
to represent a given L-space. Why should we then be inconvenienced with an
inital choice of sample space, if it calls forth unnecessary regularity conditions
merely to make proofs flow smoothly? The downside of his approach is that
familiar looking objects, such as Markov kernels, can take on a strange new
appearance in the context of an arbitrary sample space. Only with the right
representations with the cunningly chosen sample spaces do measures look like
traditional measures, and random variables look like measurable functions, and
randomizations look like Markov kernels.
norm.orthog <18> Exercise. If λ1 ⊥ λ2 , for λi ∈  ∗ , show that

λ1 + λ2
=
λ1
+
λ2

Solution: The norm property requires


λ1 + λ2

λ1
+
λ2
. The
equality comes from the L-space property <16>, Lemma <3>, and the fact
 that
λ
=
|λ|
.

[§] 5. Bands in a vector lattice

A subset W of a vector lattice V is said to be majorized by an element v


if w ≤ v for each w in W . The element v is called a majorant. The smallest
majorant, if it exists, is called the supremum of W , and is denoted by sup W .
That is, sup W , if it exists, is the v0 in V for which:
(i) w ≤ v0 for all w in W ;
(ii) if v is an element of V for which w ≤ v for all w in W then v0 ≤ v.
A vector lattice for which each majorized set has a supremum is said to be
Dedekind complete.

Asymptopia: 4Jan99
c David Pollard 7
Section D.5 Bands in a vector lattice

sup.real.line <19> Example. The subset W0 of all positive rational numbers on the real line
is not majorized within R: there is no real number r that is larger than every
positive rational.
The subset W1 of all rational numbers w for which w 2 ≤ 2 is majorized. √
For example, 2, 1.5, 1.42, 1.415 are majorants. The smallest majorant is 2.
The real numbers are Dedekind complete. In fact, the real numbers can
be defined as equivalence classes of majorized subsets of the rationals. More
precisely, a real number gets defined by a “Dedekind cut”, the partition of
the rationals that is eventually identified with the sets of all rationals that are
smaller than the real and all rationals larger than the real), a procedure due to
 Dedekind (Rudin 1976, pages 17–21).
cts.fn.sup <20> Example. The set C(R) of all bounded, continuous real functions on the real
line is a vector lattice under the pointwise ordering of functions. The subset W
consisting of all w ∈ C(R) for which

1 if t ≤ 0
w(t) ≤
0 if t > 0
is majorized (by the constant function 1, for example), but it has no supremum
in C(R).
 The vector lattice C(R) is not Dedkind complete.
band.def <21> Definition. A vector sublattice B of a vector lattice V is called a band
in V if
(i) for each v in V and each b in B , if |v| ≤ |b| then v ∈ B
(ii) each nonempty subset of B that has a majorant in V also has a
supremum, which belongs to B
Notice that the lattice properties of a band follow automatically from (i)
and (ii) and its vector space properties: b1 ∨ b2 = sup{b1 , b2 } and b1 ∧ b2 =
b1 + b2 − (b1 ∨ b2 ).
orthog.band <22> Example. Suppose V is a Dedekind complete vector lattice. For a subset
W of V , not necessarily a sublattice or even a subspace, its orthogonal
complement W ⊥ is defined as the set of v ∈ V that are orthogonal to each
element of W . Assertion: The set W ⊥ is a band. Proof: The vector space
property follows easily from Exercise <6>. Property (i) of bands follows
immediatley from the trivial fact that if |v| ≤ |b| and |w| ∧ |b| = 0 then
|w| ∧ |v| = 0. property (ii) is slightly trickier.
Suppose X ⊂ W ⊥ is majorized. Write m for its supremum. For a
fixed x0 in X , define X 0 = {x ∈ X : x ≥ x0 }. Notice that m = sup X 0 and
m −x0 = supx∈X 0 (x −x0 ). (This little trick effectively lets us reduce the problem
to the case where X ⊆ V +, so that absolute values cause no trouble.) Fix a w
in W . We need to show that |w| ∧ |m| = 0. Bound |m| by (m − x0 ) + |x0 |. By
assumption |w| ∧ |x0 | = 0. By virtue of the result in Example <6>, it is now
enough to prove that |w| ∧ (m − x0 ) = 0. The distributive law from Problem [3],
|w| ∧ (m − x0 ) = |w| ∧ sup(x − x0 )
X0
= sup (|w| ∧ (x − x0 ))
X0

=0 because x − x0 ∈ W ⊥
The band W ⊥⊥ consists of all elements that are orthogonal to the elements
that are orthogonal to W . Figure out what that means and you will see why
W ⊆ W ⊥⊥ . In fact, as shown in Problem [4], W ⊥⊥ is the smallest band
containing W , also known as the band in V generated by W . The Problem uses
 facts about projections onto bands (see Section 7).

8 Asymptopia: 4Jan99
c David Pollard
Appendix D Vector Lattices

L1.band <23> Exercise. Let V =  # , where  is the vector lattice of all bounded, S-
measurable functions on a set S equipped with a sigma-field S. Let λ be a fixed
sigma-finite measure on S. Let B be the subset of V identified with the space
of all finite signed measures on S that are absolutely continuous with respect
to λ. Show that B is a band.
Solution: Each µ in B has a density wrt λ. The set B may therefore be
identifed with the Banach space L 1 (λ), the space of equivalence classes of
λ-integrable real functions.
Property (i) of bands is easy to prove. Suppose 0 ≤ |v| ≤ |µ| with v ∈ V
and µ ∈ B. With no loss of generality, suppose 0 ≤ v ≤ µ. The functional v
defines a set function ν on S: the value of ν(A) is just the functional v applied
to the indicator function of A. Linearity of v implies finite additivity of ν.
Also, if {An } is a decreasing sequence of sets in S with intersection ∅ then
0 ≤ ν An ≤ µAn ↓ 0.
The set function ν is actually a finite, countably additive measure on S. The
integral wrt ν coincides with the fnal v on : argue from the fact that each
γ ∈  is a uniform limit of sinmple functions. That is, we may identify v with
the measure ν. Clearly, ν is absolutely continuous with respect to µ, which
is itself absolutely continuous with respect to λ. Thus ν ∈ B, as required for
property (i) of bands.
For property (ii), suppose that M is a subset of B that is majorized by w
in V . That is,
µ(γ ) ≤ v(γ ) for all γ ∈  + .
Let M1 denote the set of all finite maxima µ1 ∨ . . . ∨ µn of measures from M.
The set M1 is also majorized by w. If M1 has a supremum then so does M,
and the two suprema are equal. Why?
Define  = supµ∈M1 µ(1), which is bounded above by w1. Choose
µn ∈ M1 such that µn (1) → . Without loss of generality, we may assume
µ1 ≤ µ2 ≤ . . ..
The corresponding sequence of densities {m n } with respect to λ increases
almost surely to a measurable function m, which defines a new measure µ. In
its role as an element of V , the measure µ is majorized by w: for each γ ∈  + ,
µγ = lim λ(m n g) by monotone convergence
≤ w(γ ) because µn ≤ w for each n
To prove that µ ≥ µ0 for every µ0 in M1 it is enough to show that
(µ − µ0 )+ (1) = 0 for each such µ0 . From the fact that µ0 ∨ µn ∈ M1 we get
 ≥ µ0 ∨ µn (1) ≥ µn 1 → 
whence
(µ0 − µ0 )+ (1) ≤ (µ0 − µn )+ (1) = (µ0 ∨ µn )(1) − µn 1 → 0.
Thus µ majorizes M.
If ν is another majorant for M1 then, for each fixed γ ∈  + ,
µγ = lim µn γ ≤ νγ
n

 That is, µ is the supremum for M1 .


Remark. The density m (or, more precisely, its equivalence class in
L 1 (λ)) is called the essential supremum of the class of densities
corresponding to the measures in M.

Asymptopia: 4Jan99
c David Pollard 9
Section D.6 The band of countably additive measures

[§] 6. The band of countably additive measures

Let  be a vector lattice of bounded functions on a set S, equipped with


supremum norm. Let S be the sigma-field on S generated by . As before,
let ca+ (S, S) denote the set of countably additive (nonnegative) finite measures
on S, and ca(S, S) denote the set of finite signed measures (differences of
measures from ca+ (S, S)) on S. Write L for the dual space  ∗ =  # .
According to the Daniell-Stone representation theorem (Royden 1968,
Chapter 13), there is a one-to-one correspondence between ca+ (S, S) and the
set of increasing linear functionals µ on  with the sigma-smoothness
property:
• if {γn : n ∈ N} ⊂  + and γn ↓ 0 pointwise, then µ(γn ) ↓ 0.
The linear functional is equal to the integral with respect to the representing
measure.
It follows immediately from the sigma-smoothness characterization, that
solid <24> if λ ∈ L and |λ| ≤ µ ∈ ca+ (S, S) then |λ| ∈ ca+ (S, S)
(and hence both λ+ and λ− are also represented by countably additive measures).
In particular, if λ ∈ L + and µ ∈ ca+ (S, S) then both µ ∧ λ and (µ − λ)+ belong
to ca+ (S, S).
ca.sup <25> Lemma. Let M ⊆ ca+ (S, S) have an upper bound λ in L + , that is,
0 ≤ µ(γ ) ≤ λ(γ ) for all g ∈  +
Then there exists a smallest λ0 in L + with µ ≤ λ0 for all µ ∈ M . The
functional λ0 is countably additive.
Proof. Define M1 as the set of all finite maxima µ1 ∨ . . . ∨ µk of finite
subcollections of measures from M. The functional λ is also an upper bound
for M1 . Define  = supµ∈M1 µ(1), which is bounded above by λ1. Choose
µn ∈ M1 such that µn (1) → . Without loss of generality, we may assume
µ1 ≤ µ2 ≤ . . ..
Let Q be a finite measure dominating all the µn , such as

Q= 2−n µn /(1 +
µn
)
n
Write m n for the density of µn with respect to Q. Then the sequence {m n }
increases almost surely to a measurable function m for which
Qm = lim Qm n = lim µn 1 = 
n n
Define λ0 as the measure with density m with respect to Q.
To prove that λ ≥ µ for every µ in M it is enough to show that
(λ0 − µ)+ (1) = 0 for each such µ. From the fact that µ ∨ µn ∈ M1 we get
 ≥ µ ∨ µn (1) ≥ µn 1 → 
whence
(µ − λ0 )+ (1) ≤ (µ − µn )+ (1) = (µ ∨ µn )(1) − µn 1 → 0.
Thus λ is an upper bound for M.
If λ1 is another upper bound for M then, for each fixed γ ∈  + ,
λ0 γ = lim µn γ ≤ λ1 γ
n

 That is, λ0 is the smallest upper bound for M.


Property <24> and Lemma <25> identify ca(S, S) as a band in the
Banach lattice L.

10 Asymptopia: 4Jan99
c David Pollard
Appendix D Vector Lattices

Projection onto ca(S, S)


For each λ in L + write ψc (λ) for the largest measure in the set {µ ∈ ca+ (S, S) :
µ ≤ λ}, and write ψ f (λ) for λ − ψc (λ).
Properties: ψ f (λ) ⊥ ν for all ν ∈ ca+ (S, S). The maps are linear, and
have a linear extension to L.

[§] 7. Projections onto bands in a vector lattice


Let B be a band in a vector lattice V . For each v ∈ V + , define π B (v) ∈ B as
the supremum of all b ∈ B + for which 0 ≤ b ≤ v.
band.proj <26> Theorem. The map v → π B (v) from V + into B + has an extension to an
increasing linear map from V into B with the property that
sing.part <27> v − πb v ⊥ B for all v in V
Proof. Consider first the behaviour of π B on V + .
The property π B (cv) = cπ B (v), for each c ∈ R+ , follows easily from the
definition of a supremum.
If v1 , v2 ∈ V + then
π B (v1 + v2 ) ≥ π B (v1 ) + π B (v2 ),
because the right-hand side is ≤ v1 + v2 . For the reverse inequality, consider
any b in B that is ≤ v1 + v2 . Invoke the Splitting Lemma <4> to express b as
a sum b1 + b2 with bi ≤ vi . Then bi ≤ π B (vi ), whence b ≤ π B (v1 ) + π B (v2 ).
That is, the right-hand side is a majorant for the set of which the left-hand side
is the supremum. The reverse inequality follows.
It follows that π B is linear in the sense of Lemma <8>. It has a
unique extension to an increasing linear functional on V , defined by π B (v) =
π B (v + ) − π B (v − ).
To prove <27>, start again with v ∈ V + . We have only to show that
|b| ∧ (v − π B (v)) = 0 for each b ∈ B. Write b0 for the last minimum. Then
0 ≤ b0 ≤ (v − π B (v); and b0 ∈ B by property (ii) of bands. It follows that
v ≥ b0 + π B (v) ∈ B. From the definition of π B (v) as the supremum, we then
get that π B (v) ≥ b0 + π B , that is, b0 = 0.
For a general v in V , the special case gives v + − π B (v + ) ⊥ B and
v − π B (v − ) ⊥ B. Exercise <6> then implies that v − π B (v) is orthgonal

 to B.
leb.decomp <28> Example. Let λ be a sigma-finite measure on a sigma field S of subsets of a
set S, and let µ be a finite signed measure on S. Let  denote the banach lattice
of all bounded, S-measurable, real functions on S. Let B denote the band in  #
corresponding to L 1 (λ). The projection µ B of µ onto B is a countably additive
measure that is absolutely continuous with respect to λ. The measure µ − µ B
is orthogonal to each member of B. It must concentrate on a set that has zero
λ measure.
The split of µ into the absolutely continuous and singular parts is called
 the Lebesgue decomposition of µ.

[§] 8. Problems
distributive [1] (Distributive law for vector lattices) For x, y, z ∈ V , a vector lattice, prove
(x ∨ y) ∧ z = (x ∧ z) ∨ (y ∧ z)

Asymptopia: 4Jan99
c David Pollard 11
Section D.8 Problems

by following these steps:


(i) Let w = x ∨ y. Show that w ∧ z ≥ x ∧ z and w ∧ z ≥ y ∧ z. Deduce that
the left-hand side of the asserted equality is ≥ the right-hand side.
(ii) Let v denote the right-hand side of the asserted equality, and let u equal
x ∨ y ∨ z. Show that v ≥ x + z − x ∨ z and v ≥ y + z − y ∨ z. Deduce
that v + u ≥ (x + z) ∨ (y + z) = w + z. Use ?? to show that the last
right-hand side equals u + (w ∧ z).
distributive2 [2] For x, y, z ∈ V , a vector lattice, prove the other distributive law,
(x ∧ y) ∨ z = (x ∨ z) ∧ (y ∨ z)
Hint: Replace x, y, z by −x, −y, −z, then use the first distributive law.
distributive.sup [3] Let V be a vector lattice with a subset X for which sup X exists. For each v in
V , show that supx∈X (v ∧ x) exists and
v ∧ sup X = sup(v ∧ x).
x∈X

Hint: Write u for sup X . It is easy to show that v ∧ u is a majorant for the
set {v ∧ x : x ∈ X }. If w is another majorant, argue as in Problem [1] that
w ≥ x +v−(x ∨v) ≥ x +v−u for all x ∈ X . Deduce that w+u−v ≥ sup X = u,
that is, w ≥ v.
perpeprp [4] Let B be a band in a Dedekind complete vector lattice V .
(i) Show that B ⊥⊥ = B. Hint: Suppose x ∈ B ⊥⊥ . Decompose as x B + x⊥ ,
where x B = π B (x). Argue that x⊥ is an element of both B ⊥ and B ⊥⊥ .
Deduce that x⊥ is orthogonal to itself, and hence is zero.
(ii) For a subset W of V , deduce that W ⊥⊥ is the smallest band containing W .
Hint: if B is a band with B ⊇ W show that B ⊥ ⊆ W ⊥ .
general.splitting [5] Extend the Splitting Lemma <4> to sums of more than  two terms:
 if xi ∈ V +
+
for i = 1, . . . , m and yj ∈V for j = 1, .. . , n and i xi = j yj then there
exist u i j ∈ V + such that j u i j = xi and i u i j = yj for all i and j.

[§] 9. Notes
See Schaefer (1986) for an introduction to the theory of vector lattices.
See Schaefer (1974) for facts about L-spaces and M-spaces.
Aliprantis & Border (1994, Chapter 7) for a quicker tour through the basic
theory.
Fremlin (1974).
Torgersen (1991, Chapter 5).
First four(?) chapters of Bourbaki’s Intégration.
Kelley & Namioka (1963, Appendix A) for a proof of the representation
of abstract L-spaces and M-spaces.

References

Aliprantis, C. D. & Border, K. C. (1994), Infinite Dimensional Analysis: A


Hitchhiler’s Guide, Springer-Verlag.
Fremlin, D. H. (1974), Topological Riesz Spaces and Measure Theory, Cam-
bridge University Press.
Kelley, J. L. & Namioka, I. (1963), Linear Topological Spaces, Van Nostrand.
Le Cam, L. (1986), Asymptotic Methods in Statistical Decision Theory,
Springer-Verlag, New York.

12 Asymptopia: 4Jan99
c David Pollard
Appendix D Vector Lattices

Royden, H. L. (1968), Real Analysis, second edn, Macmillan, New York.


Rudin, W. (1976), Principles of Mathematical Analysis, third edn, McGraw-Hill.
Schaefer, H. H. (1974), Banach Lattices and Positive Operators, Springer-Verlag.
Schaefer, H. H. (1986), Topological Vector Spaces, Springer-Verlag.
Torgersen, E. (1991), Comparison of Statistical Experiments, Cambridge
University Press.

Asymptopia: 4Jan99
c David Pollard 13

You might also like