Measure and Integration: Solutions of CW2

Fall 2016 [G. Holzegel]

December 19, 2016

Problem 1 of Sheet 5
a) Left (fn ) and (gn ) be sequences of integrable functions with fn (x) → f (x) and gn (x)
R → g (x)
R for almost
every
R x Rand the limits f and g being also integrable. Prove that if |fn | ≤ gn and gn → g, then also
fn → f . Hint: Mimick the proof of the dominant convergence theorem from lectures.
b) Let (fn ) be a sequence of integrable functions with fn (x) → f (x) for almost every x and the limit f
being integrable. Prove that
Z Z Z
|fn − f | → 0 if and only if |fn | → |f | .

Solution
a) We start from
|fn (x) − f (x)| ≤ gn (x) + g(x)
which holds for a.e. x by the triangle inequality (note the assumptions imply |f (x)| ≤ g(x) for a.e. x). After
changing g on a set of measure zero, the sequence of functions given by hn (x) = g(x)+gn (x)−|fn (x)−f (x)|
is non-negative and we can apply Fatou’s Lemma to produce the inequality
Z Z
lim inf (g(x) + gn (x) − |fn (x) − f (x)|) dx ≥ 2g(x)dx .
n→∞
R R R
Now since we are assuming gn → g and that g < ∞ we obtain, just as in the proof in class
Z
lim sup (|fn (x) − f (x)|) dx ≤ 0 ,
n→∞
R R R
from which we deduce limn→0 |fn (x) − f (x)| dx = 0 and by the triangle inequality that R fn → f . For
the application in part b) we note that we actually proved the stronger statement that |fn − f | → 0.
R
b) Suppose |fn − f | → 0. By the triangle inequality for the integral and the pointwise reverse triangle
inequality we have
Z    Z   Z  
|fn (x)| − |f (x)| dx ≤ |fn (x)| − |f (x)| dx ≤ fn (x) − f (x) dx
     

R R
and using the assumption the left hand side goes to zero. It follows that |fn | → |f |.
R R
Suppose now |fn | → |f |. The idea is to use part a). We have fn (x) → f (x) for almost every x
and hence |fn (x)| → |f (x)| for almost every x. If we set gn (x) = |fn (x)| and g(x) = limn→∞ gn (x) =
limn→∞ |fn (x)| = |f (x)| it is easily
R checked
R that all assumptions of part a) are satisfied, in particular
fRn (x) ≤ |fn (x)| = |gn (x)| and |fn | → |f |. The stronger statement proven in part a) then implies
|fn − f | → 0 as desired.

1
Problem 4 of Sheet 5
This question looks at the invariance properties of the Lebesgue integral (Stein-Shakarchi, page73-74).
Let f : Rd → R be a measurable function. Define the translation
fh (x) := f (x − h) for some h ∈ Rd .
a) Prove that if f is integrable, then so is fh and
Z Z
f (x − h) dx = f (x) dx .
Rd Rd
Hint: First show this for the characteristic function f = χE , then for simple functions. Finally use
the approximation theorem from lectures and the monotone convergence theorem.
b) Prove that if f is integrable, then so are f (δx) for δ > 0 and f (−x) with the relations
Z Z Z Z
δd f (δx) dx = f (x) dx and f (−x) dx = f (x) dx
Rd Rd Rd Rd
Hint: Proceed exactly as in a).
c) Prove that if f is integrable, then
kfh − f kL1 → 0 as h → 0.
Hint: Approximate f by a continuous function of compact support g and write
fh − f = (fh − gh ) + (gh − g) + (g − f ).
d) Prove that if f is integrable, then
kf (δx) − f (x) kL1 → 0 as δ → 1.

Solution
a) First, from the translation invariance of the Lebesgue measure (discussed in class) we recall that the
set E and the translated set Eh = {x + h | x ∈ E} have the same measure m (Eh ) = m (E). Since
fh = χEh the identity for the integral follows. Second, by the linearity of the integral the identity
then also holds for finite linear combinations of characteristic function, hence for simple functions.
Third, an arbitrary non-negative function can be approximated by a strictly increasing sequence of
simple functions (ϕn ). But then both (ϕn ) and ((ϕh )n ) are increasing sequences of simple functions
converging to f and fh respectively. The desired identity clearly holds for each ϕn and (ϕh )n and
the MCT implies that it also holds for f and fh . Fourth and finally, a general integrable f can be
decomposed into its positive and negative part which are both integrable and for which the identity
has already been established. Using linearity once more the result follows.
b) Indeed exactly the same as in a).
c) We follow the hint and for  > 0 prescribed we find a continuous function of compact support g
with kf − gkL1 < . The assumptions on g imply that gh − g with |h| < 1 is (uniformly) compactly
supportedR and continuous, hence gh − g is dominated by an integrable function and the DCT implies
limh→0 |g(x − h) − g(x)|dx = 0. Therefore we can find a δ (depending on ) with kgh − gk <  for
|h| < δ. Finally, by part a) we have kfh − gh kL1 = kf − gkL1 <  and hence
kfh − f kL1 ≤ 3 for all |h| < δ
which establishes that the desired limit exists and is equal to zero.
d) We adapt the proof in c). Since we are interested in the limit δ → 1 we a priori restrict to 21 < δ < 2.
Set fδ (x) = f (δx), pick g continuous and compactly supported with kf − gkL1 <  and decompose
kfδ − f kL1 = kfδ − gδ kL1 + kgδ − gkLR1 + kg − f kL1 . We have that g (δx) − g(x) is (uniformly)
compactly supported and hence limδ→1 |g (δx) − g (x) |dx = 0 by DCT. We can hence choose η such
that |δ − 1| < η implies kgδ − gkL1 < . Finally, we have kfδ − gδ kL1 = δ −d kf − gkL1 by part b) so
that in total
kfδ − f kL1 ≤ δ −d  + 2 ≤ 2d  + 2 .

2
 Problem 1 of Sheet 6
There are several ways in which a sequence of real valued measurable functions (fn ) can converge to a
limiting function f : For instance, pointwise, pointwise a.e. or uniformly. In lectures, we also introduced
the L1 -norm and hence the notion of fn → f in L1 .
a) Discuss the convergence of the following sequences of functions with regard to the aforementioned
notions:
i. fn = n1 χ(0,n)
ii. fn = χ(n,n+1)
iii. fn = n · χ[0,1/n]
b) Give an example of a sequence (fn ) which converges to f in L1 but not pointwise a.e.
HINT: Construct a sequence of indicator functions travelling over and over through the interval [0, 1]
with decreasing length.
c) There is a further notion of convergence, which plays an important role in probability: We say that
(fn ) is Cauchy in measure if for every  > 0 we have

m ({x | |fm (x) − fn (x)| ≥ }) → 0 as m, n → ∞.

We say that (fn ) converges to f in measure if for every  > 0 we have

m ({x | |fn (x) − f (x)| ≥ }) → 0 as n → ∞.

Prove that if fn → f in L1 , then fn → f in measure. Revisit a) to see that the converse does not
hold.

Solution
a) The function in i) converges to the zero function uniformly. It has L1 -norm equal to 1 for all n and hence
does not converge to the zero function in L1 .
The function in ii) converges to zero pointwise but not uniformly. The L1 -norm is equal to 1 for all n
and hence the function does not converge to the zero function in L1 .
The function in iii) converges to zero pointwise almost everywhere (everywhere except at x = 0). The
L1 -norm is again equal to 1 for all n so fn does not converge in L1 to the zero function.
b) Define the following sequence of functions. We let
• f1 = χ[0,1] .
• f2 = χ[0,1/2] and f3 = χ[1/2,1]
• f4 = χ[0,1/4] and f5 = χ[1/4,1/2] and f6 = χ[1/2,3/4] and f7 = χ[3/4,1]
• ...
I leave it to you to find an explicit expression for the general fn .
It is clear from the definition that fn converges to zero in L1 . However, fn does not converge pointwise to
0 for any x. This is because if fn (x) → 0 for some x ∈ [0, 1] then one could find N such that |fn (x)| < 21
for any n ≥ N . However, one can clearly find an n larger than N such that |fn (x)| = 1 holds.
c) This is a consequence of the Chebychev inequality: For fixed  > 0 we have
Z
1
m ({x | |fn (x) − f (x)| ≥ }) ≤ |fn (x) − f (x)|

and the right hand side goes to zero as n → ∞ proving that fn → f in measure.
The example i) in a) converges to zero in measure but not in L1 .

3
Problem 3 of Sheet 6
Suppose I ⊂ R is open and that f : I × E → R, (t, x) 7→ f (t, x) for E a measurable subset of Rd . Suppose
f is in L1 (E) for all t, differentiable in t for all x with the derivative satisfying |∂t f (t, x) | ≤ g (x) for some
g ∈ L1 (E). Then Z Z
d ∂f
f (t, x) dx = (t, x) < ∞ .
dt E E ∂t

Solution
We establish the identity at fixed t ∈ I. We first define the difference quotient

f (t + tn , x) − f (t, x)
hn (t, x) =
tn
where (tn ) is any sequence converging to zero with tn 6= 0 for all n and such that t + tn ∈ I. Clearly
hn (t, x) → ∂t f (t, x) for all x ∈ E. Hence ∂t f (t, x) is a measurable function on E and in view of the
assumption |∂t f (t, x) | ≤ g(x) also integrable. By the mean value theorem

hn (t, x) = f 0 t̃, x

for a t̃ ∈ (t, t + tn ) ⊂ I and hence by assumption |hn (t, x) | ≤ g(x) for sufficiently large n. It follows that
the dominant convergence theorem applies to hn (t, x) producing
Z Z
∂f
lim hn (t, x) = (t, x) < ∞ .
n→∞ E E ∂t

Finally, observe that

R R
f (t + tn , x) − f (t, x)
Z Z
E E d
lim hn (t, x) = lim = f (t, x) dx ,
n→∞ E n→∞ tn dt E

the last step following since we proved the result for any sequence (tn ) converging to zero.

4
Problem 5 of Sheet 7
Let a, b > 0 be positive constant and consider
x sin(x−b )
 a
for 0 < x ≤ 1,
f (x) =
0 if x = 0
a) Prove that f is of bounded variation if [0, 1] if and only if a > b.
b) Set a = b. Construct for each α ∈ (0, 1) a function that satisfies the Lipschitz condition of exponent α
|f (x) − f (y)| ≤ Mα |x − y|α
but which is not of bounded variation. Here Mα is a constant.
Remark: Recall that in lectures we saw that if the above holds with α = 1, then f is of bounded variation.
HINT: Estimate |f (x + h) − f (x)| for h > 0 in two different ways (one being the mean value theorem,
a
the other being by C (x + h) . Consider then the cases xa+1 ≥ h and xa+1 < h.).

Solution
a) We first establish that if a > b, then f is of BV. We have
f 0 (x) = axa−1 sin x−b − bxa−b−1 cos x−b

for x ∈ (0, 1) and f 0 is also integrable in [0, 1] in view of

Z 1
b
|f 0 (x)| ≤ a · xa−1 + b · xa−b−1 and hence dx|f 0 (x) |dx ≤ 1 + .
0 a−b
It follows that for any partition 0 = x0 < x1 < ... < xN = 1 we have by the FT for the Lebesgue integral
N N  Z xi N Z xi Z 1
X X  X b
|f (xi ) − f (xi−1 ) | = dtf 0 (t) dt ≤ |f 0 (x) |dx ≤ |f 0 (x) |dx ≤ 1 + .
 
a−b

i=1 i=1 xi−1 i=1 xi−1 0

We now establish that if a ≤ b then f is not of bounded variation. To do this, we construct a sequence
of partitions whose variation is unbounded.
2 1/b 2 a/b



For n ≥ 1 we define xn = πn . Note that |f (xn ) | = πn if n is odd, while f (xn ) = 0 if n is even.
For any N ≥ 1 we now consider a partition of [0, 1] by N + 1 intervals of the form
0 < xN < xN −1 < ... < x2 < x1 < 1 .
The variation of this partition is (dropping the left and right outermost interval)
N −1 N −1  a/b N −1  a/b
X X 2 X 2
V arf (P) ≥ |f (xn+1 ) − f (xn ) | ≥ +
n=1 n=1
πn n=1
π(n + 1)
n odd n even

Both sums on the right hand side diverge as N → ∞ if a ≤ b, the case a = b being the borderline case
(harmonic series). Hence the total variation of f on [0, 1] is not bounded.
b) Let us fix α ∈ (0, 1) and set b = a, so f (x) = xa sin x−a on (0, 1] and f (0) = 0. We will choose a (α) such
that the α-Lipschitz condition holds. Wlog we fix y > x and set y = x + h with 0 < h ≤ 1 and x ∈ [0, 1).
We estimate |f (x + h) − f (x) | ≤ (x + h)a + xa ≤ 2(x + h)a for x ∈ [0, 1) and also |f (x + h) − f (x)| ≤
|f 0 (x̃)|h for some x̃ ∈ (x, x + h) by the mean value theorem. By part a) we estimate f 0 to obtain
|f (x + h) − f (x)| ≤ 2ax h for x ∈ (0, 1].
1
Now if h ≤ xa+1 then x1 ≤ h− a+1 then x 6= 0 and the second estimate yields |f (x + h) − f (x)| ≤ 2a x h ≤
1
2ah1− a+1 . Hence if we set α := 1 − a+1
1 a
= a+1 we satisfy the desired Lipschitz condition in this range of
h. If h > xa+1 the first estimate yields the desired Lipschitz condition via (using h ≤ 1 and a > 0)
1 1 1
|f (x + h) − f (x) | ≤ 2(h a+1 + h)a ≤ 2(h a+1 + h a+1 )a ≤ 2 · 2a hα .