7.2: Cσ-Sets. Countable Additivity. Permutable Series
( \newcommand{\kernel}{\mathrm{null}\,}\)
We now want to further extend the definition of volume by considering countable unions of intervals, called Cσ-sets (C being the semiring of all intervals in En).
We also ask, if A is split into countably many such sets, does additivity still hold? This is called countable additivity or σ-additivity (the σ is used whenever countable unions are involved).
We need two lemmas in addition to that of §1.
If B is a nonempty interval in En, then given ε>0, there is an open interval C and a closed one A such that
A⊆B⊆C
and
vC−ε<vB<vA+ε.
- Proof
-
Let the endpoints of B be
¯a=(a1,…,an) and ¯b=(b1,…,bn).
For each natural number i, consider the open interval Ci, with endpoints
(a1−1i,a2−1i,…,an−1i) and (b1+1i,b2+1i,…,bn+1i).
Then B⊆Ci and
vCi=n∏k=1[bk+1i−(ak−1i)]=n∏k=1(bk−ak+2i).
Making i→∞, we get
limi→∞vCi=n∏k=1(bk−ak)=vB.
(Why?) Hence by the sequential limit definition, given ε>0, there is a natural i such that
vCi−vB<ε,
or
vCi−ε<vB.
As Ci is open and ⊇B, it is the desired interval C.
Similarly, one finds the closed interval A⊆B. (Verify!)◻
Any open set G⊆En is a countable union of open cubes Ak and also a disjoint countable union of half-open intervals.
(See also Problem 2 below.)
- Proof
-
If G=∅, take all Ak=∅.
If G≠∅, every point p∈G has a cubic neighborhood
Cp⊆G,
centered at p (Problem 3 in Chapter 3, §12). By slightly shrinking this Cp, one can make its endpoints rational, with p still in it (but not necessarily its center), and make Cp open, half-open, or closed, as desired. (Explain!)
Choose such a cube Cp for every p∈G; so
G⊆⋃p∈GCp.
But by construction, G contains all Cp, so that
G=⋃p∈GCp.
Moreover, because the coordinates of the endpoints of all Cp are rational, the set of ordered pairs of endpoints of the Cp is countable, and thus, while the set of all p∈G is uncountable, the set of distinct Cp is countable. Thus one can put the family of all Cp in a sequence and rename it {Ak}:
G=∞⋃k=1Ak.
If, further, the Ak are half-open, we can use Corollary 1 and Note 3, both from §1, to make the union disjoint (half-open intervals form a semiring!).◻
Now let Cσ be the family of all possible countable unions of intervals in En, such as G in Lemma 2 (we use Cs for all finite unions). Thus A∈Cσ means that A is a Cσ-set, i.e.,
A=∞⋃i=1Ai
for some sequence of intervals {Ai}. Such are all open sets in En, but there also are many other Cσ-sets.
We can always make the sequence {Ai} infinite (add null sets or repeat a term!).
By Corollary 1 and Note 3 of §1, we can decompose any Cσ-set A into countably many disjoint intervals. This can be done in many ways. However, we have the following result.
If
A=∞⋃i=1Ai (disjoint)=∞⋃k=1Bk (disjoint)
for some intervals Ai,Bk in En, then
∞∑i=1vAi=∞∑k=1vBk.
Thus we can (and do) unambiguously define either of these sums to be the volume vA of the Cσ-set A.
- Proof
-
We shall use the Heine-Borel theorem (Problem 10 in Chapter 4, §6; review it!).
Seeking a contradiction, let (say)
∞∑i=1vAi>∞∑k=1vBk,
so, in particular,
∞∑k=1vBk<+∞.
As
∞∑i=1vAi=limm→∞m∑i=1vAi,
there is an integer m for which
m∑i=1vAi>∞∑k=1vBk.
We fix that m and set
2ε=m∑i=1vAi−∞∑k=1vBk>0.
Dropping "empties" (if any), we assume Ai≠∅ and Bk≠∅.
Then Lemma 1 yields open intervals Yk⊇Bk, with
vBk>vYk−ε2k,k=1,2,…,
and closed ones Xi⊆Ai, with
vXi+εm>vAi;
so
2ε=m∑i=1vAi−∞∑k=1vBk<m∑i=1(vXi+εm)−∞∑k=1(vYk−ε2k)=m∑i=1vXi−∞∑k=1vYk+2ε.
Thus
m∑i=1vXi>∞∑k=1vYk.
(Explain in detail!)
Now, as
Xi⊆Ai⊆A=∞⋃k=1Bk⊆∞⋃k=1Yk,
each of the closed intervals Xi is covered by the open sets Yk.
By the Heine-Borel theorem, ⋃mi=1Xi is already covered by a finite number of the Yk, say,
m⋃i=1Xi⊆p⋃k=1Yk.
The Xi are disjoint, for even the larger sets Ai are. Thus by Lemma 1(ii) in §1,
m∑i=1vXi≤p∑k=1vYk≤∞∑k=1vYk,
contrary to (1). This contradiction completes the proof.◻
If
A=∞⋃k=1Bk (disjoint)
for some intervals Bk, then
vA=∞∑k=1vBk.
Indeed, this is simply the definition of vA contained in Theorem 1.
Note 1. In particular, Corollary 1 holds if A is an interval itself. We express this by saying that the volume of intervals is σ-additive or countably additive. This also shows that our previous definition of volume (for intervals) agrees with the definition contained in Theorem 1 (for Cσ-sets).
Note 2. As all open sets are Cσ-sets (Lemma 2), volume is now defined for any open set A⊆En (in particular, for A=En).
If Ai,Bk are intervals in En, with
∞⋃i=1Ai⊆∞⋃k=1Bk,
then provided the Ai are mutually disjoint,
∞∑i=1vAi≤∞∑k=1vBk.
- Proof
-
The proof is as in Theorem 1 (but the Bk need not be disjoint here).
If
A⊆∞⋃k=1Bk,
where A∈Cσ and the Bk are intervals in En, then
vA≤∞∑k=1vBk.
- Proof
-
Set
A=∞⋃i=1Ai( disjoint ),Ai∈C,
and use Corollary 2.◻
If A,B∈Cσ, with
A⊆B,
then
vA≤vB.
("Larger sets have larger volumes.")
This is simply Corollary 3, with ⋃kBk=B.
The volume of all of En is ∞ (we write ∞ for +∞).
- Proof
-
We have A⊆En for any interval A.
Thus, by Corollary 4, vA≤vEn.
As vA can be chosen arbitrarily large, vEn must be infinite.◻
For any countable set A⊂En,vA=0. In particular, v∅=0.
- Proof
-
First let A={¯a} be a singleton. Then we may treat A as a degenerate interval [¯a,¯a]. As all its edge lengths are 0, we have vA=0.
Next, if A={¯a1,¯a2,…} is a countable set, then
A=⋃k{¯ak};
so
vA=∑kv{¯ak}=0
by Corollary 1.
Finally, ∅ is the degenerate open interval (¯a,¯a); so v∅=0.◻
Note 3. Actually, all these propositions hold also if all sets involved are Cσ-sets, not just intervals (split each Cσ-set into disjoint intervals!).
Permutable Series. Since σ-additivity involves countable sums, it appears useful to generalize the notion of a series.
We say that a series of constants,
∑an,
is permutable iff it has a definite (possibly infinite) sum obeying the general commutative law:
Given any one-one map
u:Nonto⟷N
(N= the naturals), we have
∑nan=∑naun,
where un=u(n).
(Such are all positive and all absolutely convergent series in a complete space E; see Chapter 4, §13.) If the series is permutable, the sum does not depend on the choice of the map u.
Thus, given any u:Nonto⟷J (where J is a countable index set) and a set
{ai|i∈J}⊆E
(where E is E∗ or a normed space), we can define
∑i∈Jai=∞∑n=1aun
if ∑naun is permutable.
In particular, if
J=N×N
(a countable set, by Theorem 1 in Chapter 1, §9), we call
∑i∈Jai
a double series, denoted by symbols like
∑n,kakn(k,n∈N).
Note that
∑i∈J|ai|
is always defined (being a positive series).
If
∑i∈J|ai|<∞,
we say that ∑i∈Jai converges absolutely.
For a positive series, we obtain the following result.
(i) All positive series in E∗ are permutable.
(ii) For positive double series in E∗, we have
∞∑n,k=1ank=∞∑n=1(∞∑k=1ank)=∞∑k=1(∞∑n=1ank).
- Proof
-
(i) Let
s=∞∑n=1an and sm=m∑n=1an(an≥0).
Then clearly
sm+1=sm+am+1≥sm;
i.e., {sm}↑, and so
s=limm→∞sm=supmsm
by Theorem 3 in Chapter 3, §15.
Hence s certainly does not exceed the lub of all possible sums of the form
∑i∈Iai,
where I is a finite subset of N (the partial sums sm are among them). Thus
s≤sup∑i∈Iai,
over all finite sets I⊂N.
On the other hand, every such ∑i∈Iai is exceeded by, or equals, some sm. Hence in (4), the reverse inequality holds, too, and so
s=sup∑i∈Iai.
But sup ∑i∈Iai clearly does not depend on any arrangement of the a_{i}. Therefore, the series ∑an is permutable, and assertion (i) is proved.
Assertion (ii) follows similarly by considering sums of the form ∑i∈Iai where I is a finite subset of N×N, and showing that the lub of such sums equals each of the three expressions in (3). We leave it to the reader.◻
A similar formula holds for absolutely convergent series (see Problems).