MATH 361 Homework 9

Royden 3.3.9
First, we show that for any subset E of the real numbers, E
c
+y = (E +y)
c
(translating the complement is
equivalent to the complement of the translated set). Without loss of generality, assume E can be written as
an open interval (e
1
, e
2
), so that E
c
+y is represented by the set {x|x (, e
1
+y) (e
2
+y, +)}. This
is equal to the set {x|x / (e
1
+y, e
2
+y)}, which is equivalent to the set (E +y)
c
.
Second, Let B = A y. From Homework 8, we know that outer measure is invariant under translations.
Using this along with the fact that E is measurable:
m

(A) = m

(B)
= m

(B E) +m

(B E
c
)
= m

((B E) +y) +m

((B E
c
) +y)
= m

(((Ay) E) +y) +m

(((Ay) E
c
) +y)
= m

(A (E +y)) +m

(A (E
c
+y))
= m

(A (E +y)) +m

(A (E +y)
c
)
The last line follows from E
c
+y = (E +y)
c
.
Royden 3.3.10
First, since E
1
, E
2
M and M is a -algebra, E
1
E
2
, E
1
E
2
M. By the measurability of E
1
and E
2
:
m

(E
1
) = m

(E
1
E
2
) +m

(E
1
E
c
2
)
m

(E
2
) = m

(E
2
E
1
) +m

(E
2
E
c
1
)
m

(E
1
) +m

(E
2
) = 2m

(E
1
E
2
) +m

(E
1
E
c
2
) +m

(E
c
1
E
2
)
= m

(E
1
E
2
) + [m

(E
1
E
2
) +m

(E
1
E
c
2
) +m

(E
c
1
E
2
)]
Second, E
1
E
2
, E
1
E
c
2
, and E
c
1
E
2
are disjoint sets whose union is equal to E
1
E
2
. As above, since
E
1
, E
2
M, E
c
1
, E
c
2
M and hence E
c
1
E
2
, E
1
E
c
2
M. From class, m

(
n
E
n
) =

n
m

(E
n
) for
measurable sets E
n
. Therefore:
m

(E
1
E
2
) +m

(E
1
E
c
2
) +m

(E
c
1
E
2
) = m

(E
1
E
2
)
Combining the two arguments above:
m

(E
1
) +m

(E
2
) = m

(E
1
E
2
) + [m

(E
1
E
2
) +m

(E
1
E
c
2
) +m

(E
c
1
E
2
)]
= m

(E
1
E
2
) +m

(E
1
E
2
)
Royden 3.3.11
Dene E
n
= (n, +).
i. Empty intersection:

i=1
E
n
=
For any x R, we can choose a natural number n > x such that x / E
n
. Therefore, there is no x R
such that x E
n
for all n. This implies that the intersection stated above is empty.
ii. By denition, m

(E
n
) = +, as each interval is an open interval containing +.
1
Royden 3.3.12
i. From lecture, we established the following for a countable sequence of E
i
:
m

i=1
E
i

=
n

i=1
m

(A E
i
)
For the innite case, we use the monotonicity property:
A

i=1
E
i

i=1
E
i

i=1
E
i

i=1
E
i

i=1
m

(A E
i
)
Since this is true for all n N, letting n :
A

i=1
E
i

i=1
m

(A E
i
)
ii. The reverse inequality is true by countable subadditivity:
A

i=1
E
i

i=1
m

(A E
i
)
From the two inequalities in parts (i) and (ii), we can conclude:
A

i=1
E
i

i=1
m

(A E
i
)
Royden 3.3.13
a. Showing (i)(ii)(vi).
(i)(ii): By proposition 5 in Royden, for all sets E, there exists an open set O such that E O and
m

(O) m

(E) +. Since E is measurable, for such a set O:

(O) = m

(O E) +m

(O E
c
)
m

(O E) +m

(O E
c
) m

(E) +
m

(E) +m

(O \ E) m

(E) +
m

(O \ E)
To make the inequality above strict, we can take

=

2
for any given > 0 and use the same reasoning
above.
(ii)(iv): Since O is open, it can be written as a countable disjoint union of open intervals. We pick an
open O such that m

(O\E) < /2. We consider two cases:

n=1
I
n
m

(O)

n=1
m

(I
n
)
2
From (ii), it is given that m

(E) < . So:

(O) m

(E) +m

(O \ E)
<
Since the outer measure is nite, the innite sum above must converge. Therefore, there exists some N
such that for all n > N,

N
I
n
<

2
. Dene:
U =
N

n=1
I
n
Case 2: O is a nite union of K intervals. Then dene:
U =
K

n=1
I
n
The symmetric measure can be decomposed into a union of disjoint sets:
UE = (U \ E) (E \ U)
m

(UE) m

(U \ E) +m

(E \ U)
m

(O \ E) +m

(O \ U)
< m

(O \ E) +/2
<
(vi)(ii) By Proposition 5, there exists some open set Q such that (E \ U) Q and m

(Q)
m

(E \ U) +. Dene O = U Q.
The set O covers E, since (U E) U and (E \ U) Q. Then:
m

(O \ E) = m

((U Q) \ E)
= m

((U \ E) (Q\ E))

(U \ E) +m

(Q\ E)
m

(U \ E) +m

(Q)
m

(UE) +m

(E\U) +
2m

(UE) + < 3
b. Showing (i)(ii)(iv)(i)
(i)(ii): Shown in part (a) above.
(ii)(iv): Dene G as follows:
G =

n
O
n
such that m

(O \ E) <
1
n
for all n N. The existence of such O
n
is guaranteed by the condition givne
in (ii). Since E O
n
for each n, E G, and by the monotonicity property:
m

(G\ E) m

(O
n
\ E)
<
1
n
Since the inequality above holds for all n N, we must have m

(G\ E) = 0.
(iv)(i): Since G is a countable intersection of open sets, it is measurable. All sets with measure zero
are measurable, so given that m

(G \ E) = 0, the set G \ E = G E
c
is also measurable. Therefore,
(G E
c
)
c
is measurable, and G (G E
c
)
c
= E is also measurable.
3
c. Showing (i)(iii)(v)(i)
(i)(iii): Since E is measurable, E
c
is also measurable. By (ii), there exists some O such that m

(O \
E
c
) < , or equivalently, m

(O E) < . Dene F = O
c
. Then:
> m

(O E)
> m

(E \ O
c
)
> m

(E \ F)
Since O is open, F is closed. Since E
c
O, we know O
c
E, or equivalently, F E.
(iii)(v): From (iii), there exists a closed set F
n
with F
n
E and m

(E \ F
n
) <
1
n
. for all n N.
Dene the following:
F =

n
F
n
Since F
n
E for all n N, F E. By monotonicity:
m

(E \ F) m

(E \ F
n
)
<
1
n
Since the inequality holds for all n N, we can conclude m

(E \ F) = 0.
(v)(i): Since F is measurable and m

(E \ F) = 0, E \ F is also measurable. Since E = F (E \ F),

n=1
E
n

= lim
n
m(E
n
)
Each E
n
is a union of disjoint closed intervals I
n
. Since closed intervals are measurable, and m(I) = (I),
we know that m(E
n
) = m(
2
n
n
I
n
) =

2
n
n=1
m(I
n
)
Therefore, it is sucient to show that the sum of the intervals which make up E
n
as n is equal to
zero in order to show that the Cantor ternary set has measure zero. For any n, E
n
is a union of 2
n
closed
invervals each with length
1
3
n
. The sum of the lengths of each interval is (
2
3
)
n
and therefore m(E
n
) = (
2
3
)
n
.
For any > 0, take n to be the rst natural number such that n > log
2/3
. This forces m(E
n
) < , which
then implies lim
n
m(E
n
) = 0.
b. F is equal to a countable union of closed intervals, and is therefore a closed set.
To show F
c
is dense, dene F
n
to be the remaining closed intervals at each stage n after the middle
interval of length

3
n
is removed. Then F is a countable union of disjoint intervals, each with length
strictly less than
1
2
n
. Therefore, given any x [0, 1] and > 0, choose N > log
1/2
. This will ensure that
the interval (x
1
2
, x +
1
2
) contains some point y that was removed in the n
th
step.
Using the same reasoning as in part (a), the measure of F is equal to the sum of the disjoint intervals
whose union is equal to F. At any stage n, there are 2
n1
intervals before any deletions are made.
4
Therefore, 2
n1
intervals of length

3
n
are removed. Therefore:
m(F) = 1

n=1
2
n1

3
n

= 1

2

n=1

2
3

n
= 1
Question 3
i Since A
1
is measurable:
m

(A
2
) = m

(A
2
A
1
) +m

(A
2
A
c
1
)
m

(A
1
) = m

(A
1
) +m

(A
2
A
c
1
)
m

(A
2
A
c
1
) = 0
ii Given that m

(B) = 0, where B is a subset of R, B must be measurable, since for any subset C,

(C B) m

(B), so m

(C B) = 0.
m

(C B) +m

(C B
c
) = m

(C B
c
)
m

(C)
iii From the two parts above, we can conclude that the set A
2
A
c
1
is a measurable set. Hence, we can
write A
2
as the union of two measurable sets, A
1
and A
2
A
c
1
. From the lecture notes, the collection of
measurable sets, M is closed under taking unions. Therefore, A
2
= A
1
(A
2
A
c
1
) is also measurable.
Question 4
By countable additivity:
m

n=1
B
n

n=1
m

(B
n
)
To show the reverse inequality, dene B =

n=1
B
n
and we note that

n=1
A
n

n=1
A
n

By the countability of A
n
:
m

n=1
B
n

= m

n=1
A
n

n=1
A
n

n
m

(B A
n
)

n
m

(B
n
A
n
)

n
m

(B
n
)
5