Homework 2

Real Analysis
Joshua Ruiter
March 23, 2018

Proposition 0.1 (Exercise 8). Let L : Rd → Rd be a linear transformation and let E ⊂ Rd

be measurable. Then L(E) ⊂ Rd is measurable.

Proof. First, note that if A is compact, then L(A) S

is compact, since L is continuous. Thus
if F is an Fσ set, then L(F ) is an Fσ set. Let F = ∞k=1 Fk where Fk are all closed. Then

∞
! ∞
[ [
L(F ) = L Fk = L(Fk )
k=1 k=1

so L(F ) can also be written as a union of closed sets, so L(F ) is also an Fσ set.
Since E is measurable, there exists an Fσ set F with m(F \ E) = 0 by Corollary 3.5.
S Fix
∞
 > 0.
P∞ Then there exists a covering of F \ E by closed cubes {Qj }j=1 with F \ E ⊂ j Qj
and j=1 |Qj | < . Then
∞
! ∞
[ [
L(F \ E) ⊂ L Qj = L(Qj )
j=1 j=1

As noted in part (b), if Qj is a closed cube of side length l, then L(Qj ) is a closed cube of
side length cd M l. Thus

m∗ (L(Qj )) = |L(Qj )| = (cd M )d |Qj |

Now we can compute m∗ (L(F \ E)).

∞
X ∞
X ∞
X
m∗ (L(F \ E)) ≤ m∗ (∪j L(Qj )) ≤ m∗ (L(Qj )) = (cd M )d |Qj | = (cd M )d |Qj | < (cd M )d 
j=1 j=1 j=1

Thus m∗ (L(F \ E)) = 0. Since L(F ) \ L(E) ⊂ L(F \ E), by monotonicity we get m∗ (L(F ) \
L(E)) = 0. As noted at the beginning, L(F ) is an Fσ set. Thus L(E) differs from an Fσ set
by a set of measure zero, so L(E) is measurable by Corollary 3.5.

Proposition 0.2 (Exercise 9). There exists an open set O such that the boundary of the
closure of O has positive Lebesgue measure.

1
Proof.
P∞ Let Cˆ be a Cantor-like set with constants {lk }∞
k=1 as in Exercise 4, chosen so that
l 2k−1
< 1, and so then by 4(a), ˆ
C has positive measure. Let Ok be the union of open
k=1 k
intervals removed at the kth stage of constructing C. ˆ Then Ok is a disjoint union of 2k−1
j
open balls of radius lk /2. Denote the centers by ak for j = 1, . . . 2k−1 , so we can write
k−1
2[
Ok = B(ajk , lk /2)
j=1

Then let
∞
[
O= O2n−1 = O1 ∪ O3 ∪ O5 ∪ . . .
n=1

We claim that the boundary of the closure of O contains C. ˆ Suppose that x ∈ C. ˆ In 4(b), we
constructed a sequence xn where each xn ∈ Cˆ and xn → x by choosing xk to be the center
c

the interval of Ok nearest to x. We choose the subsequence (x1 , x3 , x5 , . . .). By the same
reasons as in 4(b), we have xn → x, and now we have xk ∈ O. Thus x is a limit point of O.
Since O and Cˆ are disjoint by construction, x 6∈ O, so x is in the boundary of the closure of
O. Thus CPˆ is a subset of the boundary of the closure of O.
Since ∞ k=1 lk 2
k−1
by hypothesis, Cˆ has positive measure. Then by monotonicity, since
the boundary of the closure of O contains C, ˆ this boundary has positive measure.
Proposition 0.3 (Exercise 10a). Let Fn , fn be as described. The for n ≥ 1, x ∈ [0, 1], we
have
0 ≤ fn (x) ≤ 1
fn+1 (x) ≤ fn (x)
Thus fn (x) converges to a limit as n → ∞ for each x.
Proof. As defined, each Fk is a function from [0, 1] into [0, 1]. Thus fn (x) = nk=1 Fk (x) is a
Q
finite product of values in [0, 1], so 0 ≤ fn (x) ≤ 1 for all n ∈ N, x ∈ [0, 1]. Additionally,
n+1
Y n
Y
fn+1 (x) = Fk (x) = Fn+1 (x) Fk (x) = Fn+1 (x)fn (x)
k=1 k=1

Since Fn+1 (x) ∈ [0, 1], this implies that fn+1 (x) ≤ fn (x). Hence for each fixed x, fn (x)
is a bounded and nonincreasing sequence, by the Monotone Sequence Theorem, fn (x) is
convergent.
Proposition 0.4 (Exercise 10b). Let f = limn→∞ fn for fn described above. Then f is
ˆ
discontinuous at x ∈ C.
ˆ we have
Proof. First, note that for x ∈ C,
Fk (x) = 1
Yn n
Y
=⇒ fn (x) = Fk (x) = 1=1
k=1 k=1
=⇒ f (x) = lim fn (x) = lim 1 = 1
n→∞ n→∞

2
Fix x ∈ C.ˆ For n ∈ N, let xn be the center of the removed interval nearest to x. As shown
in 4(b), this sequence xn converges to x. However, by definition of Fn , we have Fn (xn ) = 0.
Thus for k ≥ n, we have
k
Y
fk (xn ) = Fj (xn ) = 0
j=1

And thus we have

f (xn ) = lim fk (xn ) = 0

k→∞

for all n. Thus limn f (xn ) = 0 and xn → x but f (x) = 1. Thus f is discontinuous at x.

Proposition 0.5 (Exercise 11). Let A be the subset of [0, 1] consisting of all number which
have no 4 in any decimal expansion. Then m(A) = 0.

Proof. Let Bk be the subset of [0, 1] that have a decimal expansion with a 4 in the kth
decimal place. Then B1 = [4/10, 5/10], and
9  
[ 4 + 10k 5 + 10k
B2 = ,
k=0
102 102

k−1
closed intervals of length 10−k , and
S
In general, Bk is disjoint union
Sof 10 k Bk = [0, 1]\A.
We now seek a way to write k Bk as a disjoint union. Set
k=1
[
Ck = Bk \ Bj
j=1
S
Then k Ck = [0, 1] \ A. Now notice that each Ck is a disjoint union of closed intervals.
In particular, Ck is a disjoint union of 9k−1 closed intervals each of length 10−k , thus each
Ck is measurable, and has measure 9k−1 /10k . By countable additivity of measure, we have
∞
! ∞ ∞
[ X X 9k−1
m Ck = m(Ck ) = =1
k=1 k=1 k=1
10k
S
Hence by additivity of measure, m(A) = m([0, 1]) − m( k Ck ) = 1 − 1 = 0.

Proposition 0.6 (Exercise 12a). An open disc in R2 is not the disjoint union of open
rectangles.

Proof. Let D be an open disc in R2 , and suppose that {Qj }∞j=1 is a covering
S of D by disjoint
open rectangles, i.e. each Qj is open, Qj ∩ Qk = ∅ unless k = j, and j Qj = D.
Let x ∈ ∂Q1 . We know that x is a limit point of Q1 , so there is a sequence in Q1 (and
hence in D) converging to x, so x ∈ D. If x 6∈ D, then x is on the boundary of D. Then let
y be another point in Q1 along the same edge as x, and z be the midpoint of xy. We claim
that one of x, y, z ∈ D. If x, y 6∈ D, then x, y ∈ ∂D. Then the straight line between x, y is

3
contained within D, since D is convex. Thus z ∈ D. Hence the boundary of Q1 contains a
point of D. Without loss of generality, S
call this point x.
Since x ∈ D and D is covered by j Qj , then x ∈ Qk for some k. Since Qk is open,
there is an open ball B(x, r) with x ∈ B(x, r) ⊂ Qk . But since x is a limit point of Q1 ,
B(x, r)∩Q1 6= ∅. Thus Q1 and Qk are not disjoint, so we reach a contradiction, and conclude
that no such covering of D exists.
Note: The analysis textbook does not provide a definition of connectedness, so I put a
standard definition here, just so you know what definition I’m using.

Definition 0.7. Let X be a topological space. We say X is disconnected if there exist non-
empty open sets U, V such that U ∪ V = X and U ∩ V = ∅. We say that X is connected if
X is not disconnected.

Proposition 0.8 (Exercise 12b). Let Ω ⊂ Rd with d ≥ 2 be an open connected set. Then Ω
is the disjoint union of open rectangles if and only if Ω is an open rectangle.

Proof. If Ω is an open rectangle, then it is trivially a disjoint union of open rectangles, so

that direction is obvious.
Suppose that Ω is an open connected set that is a disjoint union of open rectangles, i.e.
Ω = ∪j∈J Qj where Qj are disjoint non-empty open rectangles. If J contains at least two
elements, then choose j0 ∈ J, and let
!
[
U= Qj \ Qj0
j6=j0

Then U, Qj0 are non-empty open sets such that U ∪ Qj0 = Ω and U ∩ Qj0 = ∅. Thus Ω is
disconnected. But this is a contradiction, so we reject the possibility that J contains at least
two elements, so Ω is only a union of one open rectangle, namely itself. Hence Ω is an open
rectangle.