Chapter 3: Lebesgue Measure
Written by Men-Gen Tsai email: b89902089@ntu.edu.tw 1. Let m be a countably additive measure defined for all sets in a σ-algebra
M.
If A and B are two sets in M with A ⊂ B, then mA ≤ mB. This property is called monotonicity.
Proof: B = AS(A − B). A and A − B are disjoint. Since m is a countably additive measure, mB = mA + m(A − B). Note that m is nonnegative, and m(A − B) ≥ 0. Hence mA ≤ mB.
2. Let < En > be any sequence of sets in M. Then m(SEn) ≤ PmEn. [Hint: Use Proposition 1.2.] This property of a measure is called count- able subadditivity.
Proof: By Proposition 1.2 on page 17, since M is a σ -algebra, there is a sequence < Bn > of sets in M such that BnT
Bm = φ for n 6= m
and ∞
[
i=1
Bi =
∞
[
i=1
Ei.
Since m is a countably additive measure and Bi ⊂ Ei for all i, by Problem 3.1 I have that
m([En) = m([Bn) = XmBn≤XmEn. 3. If there is a set A in M such that mA < ∞, then mφ = 0.
Proof: Note that A = ASφ and A and φ are disjoint, and thus mA = mA + mφ.
Since mA < ∞, mφ = 0 precisely.
4.
5. Let A be the set of rational numbers between 0 and 1, and let {In} be a finite collection of open intervals covering A. Then Pl(In) ≥ 1.
Proof 1: (due to Meng-Gen Tsai) Since 0 ∈ A, there is an open interval J1 in {In} such that 0 ∈ J1. Let J1 = (a1, b1). Note that a1 < 0 and b1 > 0. If b1 ≥ 1, then
Xl(In) ≥ l(J1) = b1− a1 ≥ 1.
Suppose not. If a1 is rational, then I can find an open interval J2 ∈ {In} such that a1 ∈ J2. If a1 is irrational, I consider the following cases.
Case 1. There is an open interval J2 such that a1 ∈ J2.
Case 2. There is an open interval J2 such that a1 is the right endpoint of J2.
Case 3. Otherwise.
I claim that Case 3 is impossible. Consider the following subcollection K1, ..., Km
where Ki ∈ {I ∈ In|a1 < x for all x ∈ I}. Thus I select a open interval K0 = (x, y) nearest a1. Hence (a1, x)TQ cannot be covered by any elements of {In}, a contradiction.
Hence we can find J2 in Case 1 and Case 2. Continue the process to find out J3, .... Since {In} is a finite covering of A, this process can be done in finite steps. Hence
Xl(In) ≥XJn =X(an+1− an) = am− a1. Note that am ≥ 1 and an≤ 0. Hence Pl(In) ≥ 1.
Proof 2: (due to Shin-Yi Lee) A is contained in SIk, then A is con- tained in SIk=SIk. So, |A| = 1 ≤ |SIk| ≤P|Ik|.
Where |.| in Zygmund’s book is the same as m(.) is Royden’s book if I do not misunderstand.
Proof 3: (due to ljl) (1) We can suppose that Inare disjoint; otherwise, if ImT
In is not empty, then we can use I = ImT
In to replace Im and In. I is also an interval, and also covers A, and will make Pl(In) smaller. Hence if the sum is still ≥ 1 after our adjustment, the original one is ≥ 1 surely.
(2) Now In are disjoint, so we can suppose that In = (an, bn), n = 1, ..., N with ai < bi ≤ ai+1< bi+1 for i = 1, ..., N − 1.
(3) It is easy to show that bi = ai+1 (or that collection cannot cover A). Hence Pl(In) = bN − a1. Also, it is easy to show that a1 < 0 and bN > 1. Proved.
6. Given any set A and any > 0, prove that there is an open set O such that A ⊂ O and m∗O ≤ m∗A + . Also, prove that there is a G ∈ Gδ such that A ⊂ G and m∗A = m∗G.
Proof: Note that
m∗A = inf
A⊂S
In
Xl(In).
where In are open intervals by the definition of the outer measure. Let O =SIn. O is also open. For any > 0, there exists {In} such that
m∗A + ≥Xl(In).
By Proposition 1 and Problem 2,
Xl(I ) = Xm∗(I ) ≥ m∗([I ) = m∗O.
Combine them and I have m∗O ≤ m∗A + .
By previous conclusion, for any n ∈ N there is an open set On such that A ⊂ On and m∗On≤ m∗A +n1. Take G =TOn. Hence
m∗G ≤ m∗On≤ m∗A + 1 n for all n ∈ N . Therefore
m∗G ≤ m∗A.
Note that A ⊂ On for all n, that is, A ⊂ G, that is, m∗A ≤ m∗G.
Hence
m∗A = m∗G.
7. Prove that m∗ is translation invariant.
Proof: Let E be a set. Consider the countable collections {In} of open intervals that cover E. Then {In0} covers E + y where In0 = In+ y. Note that In0 is also an open interval, and
l(In0) = l(In).
Hence for any > 0 there is {In} such that m∗E + > Pl(In). Thus, m∗E + >Xl(In) =Xl(In0) ≥ m∗(E + y),
that is,
m∗E ≥ m∗(E + y).
Similarly (regard E as (E + y) − y),
m∗E ≤ m∗(E + y).
Hence m∗E = m∗(E + y), that is, m∗ is translation invariant.
8.
9. Show that if E is a measurable set, then each translate E + y of E is also measurable.
Proof: Since E is a measurable set, for each set A I have m∗A = m∗(A\E) + m∗(A\Ec).
Suppose x ∈ AT(E + y), then
x ∈ A\(E + y) ⇔ x ∈ A and x ∈ (E + y)
⇔ x ∈ A and x − y ∈ E
⇔ x − y ∈ (A − y) and x − y ∈ E
⇔ x − y ∈ (A − y)\E
⇔ x ∈ (A − y)\E + y.
Hence AT(E + y) = (A − y)TE + y; thus
m∗(A\(E + y)) = m∗((A − y)\E + y)
= m∗((A − y)\E)
(since Problem 3.7). Similarly, AT(E + y)c = (A − y)TEc+ y; thus m∗(A\(E + y)c) = m∗((A − y)\Ec+ y)
= m∗((A − y)\Ec) Hence
m∗(A\(E + y)) + m∗(A\(E + y)c)
= m∗((A − y)\E) + m∗((A − y)\Ec)
= m∗(A − y)
= m∗A
for each set A (since E is measurable). Therefore, E + y is also mea- surable.
10.
11. Show that the condition mE1 < ∞ is necessary in Proposition 14 by giving a decreasing sequence < En > of measurable set with φ =TEn and mEn= ∞ for each n.
Solution: Let
En=
n
[
k=1
Ak where Ak = (k − 14, k + 14).
12. Let < En > be a sequence of disjoint measurable sets and A any set.
Then
m∗(A\
∞
[
i=1
Ei) =
∞
X
i=1
m∗(A\Ei).
Proof: By Lemma 9, m∗(A\
n
[
i=1
Ei) =
n
X
i=1
m∗(A\Ei).
Since
A\
∞
[
i=1
Ei ⊂ A\
n
[
i=1
Ei
for all n,
m∗(A\
∞
[
i=1
Ei) ≥ m∗(A\
n
[
i=1
Ei) =
n
X
i=1
m∗(A\Ei).
for all n. Hence,
m∗(A\
∞
[
i=1
Ei) ≥
∞
X
i=1
m∗(A\Ei).
Similarly, by using the fact that m∗ is nonnegative, I have m∗(A\
∞
[
i=1
Ei) ≤
∞
X
i=1
m∗(A\Ei).
Therefore I got the conclusion.
13. Prove Proposition 15. [Hints: a. Show that for m∗E < ∞, (i) ⇒ (ii)
⇔ (vi) (cf. Proposition 5).
b. Use (a) to show that for arbitrary sets E, (i) ⇒ (ii) ⇒ (iv) ⇒ (i).
c. Use (b) to show that (i) ⇒ (iii) ⇒ (v) ⇒ (i).]
Proposition 15: Let E be a given set. Then the following five statements are equivalent:
i. E is measurable.
ii. Given > 0, there is an open set O ⊃ E with m∗(O − E) < .
iii. Given > 0, there is a closed set F ⊂ E with m∗(E − F ) < .
iv. There is a G in Gδ with E ⊂ G, m∗(G − E) = 0.
v. There is an F in Fσ with F ⊂ E, m ∗ (E − F ) = 0.
If m∗E is finite, the above statements are equivalent to:
vi. Given > 0, there is a finite union U of open intervals such that m∗(U 4 E) < .
Proof of (a):
(i) ⇒ (ii): Since E is measurable, by Proposition 5 there is an open set O such that E ⊂ O and m∗O ≤ m∗E + 2. By Lemma 9
m∗O = m∗(O − E) + m∗(O\E)
= m∗(O − E) + m∗E
since E ⊂ O). Since m∗E is finite, m∗(O − E) ≤ 2 < .
(ii) ⇒ (vi): Take = 1/n for all n ∈ N , there is an open set On ⊂ E with m∗(O − E) < 1/n. Take G =TOn; G is open and G ∈ Gδ. And
m ∗ (G − E) ≤ m ∗ (On− E) < 1/n for all n ∈ N . Hence m∗(G − E) = 0.
(vi) ⇒ (ii): If not, there is a real 0 > 0 such that
∗
for any open set O. Note that G is the intersection of countable open set, write G =TOn. Hence
m∗(
n
\
k=1
Ok− E) ≥ 0 for all n. Hence
m∗(
∞
\
k=1
Ok− E) ≥ 0, a contradiction.
14.