0% found this document useful (0 votes)
94 views2 pages

Solutionsweek 42,43

The document contains solutions to problems from a measure theory course. It discusses several concepts: 1) A sequence of functions converges pointwise to the zero function but does not converge in Lp space. 2) Limits in Lp spaces are unique, so two representatives of the same equivalence class must be equal almost everywhere. 3) If a sequence converges in Lp, the limit function must be a characteristic function almost everywhere. 4) Absolute value of a measure is zero on a set if and only if the measure is zero on the set.

Uploaded by

Lau Merchan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
94 views2 pages

Solutionsweek 42,43

The document contains solutions to problems from a measure theory course. It discusses several concepts: 1) A sequence of functions converges pointwise to the zero function but does not converge in Lp space. 2) Limits in Lp spaces are unique, so two representatives of the same equivalence class must be equal almost everywhere. 3) If a sequence converges in Lp, the limit function must be a characteristic function almost everywhere. 4) Absolute value of a measure is zero on a set if and only if the measure is zero on the set.

Uploaded by

Lau Merchan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

MAT 3330/4300: Solutions for problems assigned week 42 and 43.

Problem 7.B.
For each x there is N such that x ∈/ [ n1 , n2 ] when n ≥ N , so fn (x) = 0 when n ≥ N .
From this we see that the sequence converges pointwise towards f ≡ 0. However
we have that
Z Z
1 1 p−1
||fn − f ||p = ( |fn | dλ) p = ( np χ[ n1 , n2 ] dλ) p = n p .
p

So ||fn − f ||p → 1 if p = 1 and ||fn − f ||p → ∞ if p > 1 hence (fn ) does not
converge to f in Lp .
Oppgave 7.H.
Since Lp = Lp (X, X, µ) is a metric space (with a metric induced by the Lp -norm)
limits of sequences are unique. Therefore f and g must be representatives of the
same µ-class of functions hence f = g µ-almost everywhere.
Oppgave 7.I.
1
Let fn = χEn and assume that fn → f in Lp . For each m, let Im = (−∞, − m )∪
1 1 1 −1
( m , 1 − m ) ∪ (1 + m , +∞). Let Fm = f (Em ). Since fn only has values 0 and 1,
1
it is clear that x ∈ Fn ⇒ |f (x) − fn (x)| ≥ m . This implies that
Z
p 1
||f − fn ||p ≥ |f − fn |dµ ≥ p µ(Fm ).
Fm m

Since ||f − fn ||pp → 0 when n → +∞, we must have that µ(Fm ) = 0 for each m. Let
F = ∪m Fm . Then µ(F ) = 0 and X − F = f −1 ({0, 1}) So we must have that f is
equal χf −1 (1) outside F hence equal a characteristic function µ- almost everywhere.
Problem 8.C.
Let P, N be a Hahn-decomposition for λ. Suppose that M is a null set. Then, since
P ∩M ⊂ M , P ∩M ⊂ M , we get that λ+ (M ) = λ(P ∩M ) = λ− (M ) = −λ(N ∩M ) =
0 hence |λ|(M ) = λ+ (M ) + λ− (M ) = 0. Now assume that |λ|(M ) = 0. Let E ⊂ M
be measurable. Since |λ| is a measure we get that |λ|(E) = 0. Then we must have
λ+ (E) = λ− (E) = 0, and we get that λ(E) = λ+ (E) − λ− (E) = 0. This proves
that M is a nullset.
Problem 8.E.
Assume µ3 (E) = 0. Since µ2 << µ3 , we have that µ2 (E) = 0. Since µ1 << µ2 , we
have that µ1 (E) = 0. This proves that µ1 << µ3 .
Let µ2 be the Lebesgue measure on B. Let µ1 be the measure on B given by
µ1 (E) = 0 for all E ∈ B. Then µ1 << µ2 , but if E = (a, b) where a < b, then
µ2 (E) = b − a 6= 0 whilst µ1 (E) = 0, so we cannot have µ2 << µ1 .
Oppgave 8.K.
Since Nn is negative for λ − nµ, we have that (λ − nµ)(Nn ) ≤ 0. So λ(Nn ) ≤

S
nµ(Nn ) < ∞ (since µ is finite). So N = Nn is a countable union of sets with
n=1
finite λ-measure and therefore σ-finite with respect to λ.
Let E ⊂ P . Then E ⊂ Pn for all n, and since Pn is positive for λ − nµ, we get that
(λ − nµ)(E) ≥ 0. So λ(E) ≥ nµ(E) for all n. If µ(E) > 0 and we let n → ∞ we
will get that λ(E) = ∞. If µ(E) = 0, we get, since λ << µ, that λ(E) = 0.
1
2

Problem 8.N.
n
P
Let φ = ai χEi be a simpel function with ai ≥ 0. Then
i=1

Z n
X n
X Z
φdλ = ai λ(Ei ) = ai f dµ =
i=1 i=1 Ei
n
X Z n
Z X Z
ai χEi f dµ = ai χEi f dµ = φf dµ.
i=1 i=1

Let (φn ) be a sequence of monotonically increasing, non-negative simple functions


which is pointwise convergent to g (by Lemma 2.11, we can find such sequence).
Then (φn f ) will be a sequence of monotonically increasing, non-negative functions
which is pointwise convergent to gf , so using The Monotone Convergence Theorem
twice we get from above that:
Z Z Z Z
gdλ = lim φn dλ = lim φn f dµ = gf dµ.

Problem 8.O.
Let E be X-measurable, using the conclusion of 8.N, we get that
Z Z Z Z
dν dν dν dλ dν dλ
ν(E) = dλ = χE dλ = χE dµ = dµ.
E dλ dλ dλ dµ E dλ dµ

Since any two Radon-Nikodym derivatives of ν with respect to µ are equal µ-


dν dλ dν
almost everywhere, then given any three dλ , dµ and dµ , we can conclude that that
dν dν dλ
dµ = dλ dµ µ-almost everywhere.
We have
Z Z Z
dλ1 dλ2 dλ1 dλ2
(λ1 + λ2 )(E) = λ1 (E) + λ2 (E) = dµ + dµ = ( + ) dµ.
E dµ E dµ E dµ dµ

d(λ1 +λ2 )
This shows that for any choice of Radon-Nikodym derivatives we get dµ =
dλ1 dλ2
dµ + dµ µ-almost everywhere.

Problem 8.P.

Let dµ and dµ
dλ be two choices of Radon-Nikodym derivatives. Since we can take
dλ dλ dµ
dλ = 1, we get from 8.O that dµ dλ = 1 almost everywhere. This implies that
dµ dλ 1
dλ 6= 0 almost everywhere and that we can write dµ = dµ almost everywhere.

You might also like