Chapter 9

Real and Complex Measures

[29/83%]
Show
LaTeX
Problem 9A.1

Prove or give a counterexample: If ν\nu is a real measure on a measurable space (X,S)(X, \mathcal{S}) and A,B∈SA, B \in \mathcal{S} are such that ν(A)≥0\nu (A) \geq 0 and ν(B)≥0\nu (B) \geq 0, then ν(A∪B)≥0\nu (A \cup B) \geq 0.

?
Problem 9A.2

Suppose ν\nu is a real measure on (X,S)(X, \mathcal{S}). Define μ:S→[0,∞)\mu : \mathcal{S} \rightarrow [0, \infty ) by

μ(E)=∣ν(E)∣. \mu (E) = \left|\nu (E)\right|.

Prove that μ\mu is a (positive) measure on (X,S)(X, \mathcal{S}) if and only if the range of ν\nu is contained in [0,∞)[0, \infty ) or the range of ν\nu is contained in (−∞,0](-\infty , 0].

?
Problem 9A.3

Suppose ν\nu is a complex measure on a measurable space (X,S)(X, \mathcal{S}). Prove that ∣ν∣(X)=ν(X)\left|\nu \right|(X) = \nu (X) if and only if ν\nu is a (positive) measure.

?
Problem 9A.4

Suppose ν\nu is a complex measure on a measurable space (X,S)(X, \mathcal{S}). Prove that if E∈SE \in \mathcal{S} then

∣ν∣(E)=sup⁡{∑k=1∞∣ν(Ek)∣:E1,E2,… is a disjoint sequence in S such that E=⋃k=1∞Ek} \left|\nu \right|(E) = \sup \left\{ \sum _{k = 1}^{\infty }\left|\nu \left(E_{k}\right)\right|: E_{1}, E_{2}, \ldots \text{ is a disjoint sequence in } \mathcal{S} \text{ such that } E = \bigcup _{k = 1}^{\infty } E_{k}\right\}
?
Problem 9A.5

Suppose μ\mu is a (positive) measure on a measurable space (X,S)(X, \mathcal{S}) and hh is a nonnegative function in L1(μ)\mathcal{L}^{1}(\mu ). Let ν\nu be the (positive) measure on (X,S)(X, \mathcal{S}) defined by dν=hdμd \nu = h d \mu. Prove that

∫f  dν=∫fh  dμ \int f \; d \nu = \int f h \; d \mu

for all SS-measurable functions f:X→[0,∞]f: X \rightarrow [0, \infty ].

?
Problem 9A.6

Suppose (X,S,μ)(X, \mathcal{S}, \mu ) is a (positive) measure space. Prove that

AC⁡μ={h  dμ:h∈L1(μ)} \operatorname {AC}_{\mu } = \left\{ h \; d \mu : h \in \mathcal{L}^{1}(\mu )\right\}

is a closed subspace of MF(S)\mathcal{M}_{\mathrm{F}}(\mathcal{S}).

?
Problem 9A.7
?
(a)

Suppose B\mathcal{B} is the collection of Borel subsets of R\mathbb {R}. Show that the Banach space MF(B)\mathcal{M}_{\mathbf{F}}(\mathcal{B}) is not separable.

(b)

Give an example of a measurable space (X,S)(X, \mathcal{S}) such that the Banach space MF(S)\mathcal{M}_{\mathbf{F}}(\mathcal{S}) is infinite-dimensional and separable.

Problem 9A.8

Suppose t>0t > 0 and λ\lambda is Lebesgue measure on the σ\sigma-algebra of Borel subsets of [0,t][0, t]. Suppose h:[0,t]→Ch:[0, t] \rightarrow \mathbb {C} is the function defined by

h(x)=cos⁡x+isin⁡x h(x) = \cos x+i \sin x

Let ν\nu be the complex measure defined by dν=hdλd \nu = h d \lambda.

?
(a)

Show that ∥ν∥=t\left\| \nu \right\| = t.

(b)

Show that if E1,E2,…E_{1}, E_{2}, \ldots is a sequence of disjoint Borel subsets of [0,t][0, t], then

∑k=1∞∣ν(Ek)∣<t \sum _{k = 1}^{\infty }\left|\nu \left(E_{k}\right)\right| < t
Remark.
?

This exercise shows that the supremum in the definition of ∣ν∣([0,t])\left|\nu \right|([0, t]) is not attained, even if countably many disjoint sets are allowed.

Problem 9A.9

Give an example to show that 9.9 can fail if the hypothesis that ν\nu is a real measure is replaced by the hypothesis that ν\nu is a complex measure.

?
Problem 9A.10

Suppose (X,S)(X, \mathcal{S}) is a measurable space with S≠{∅,X}\mathcal{S} \neq \left\{ \varnothing , X\right\}. Prove that the total variation norm on MF(S)\mathcal{M}_{\mathrm{F}}(\mathcal{S}) does not come from an inner product. In other words, show that there does not exist an inner product ⟨⋅,⋅⟩\left\langle \cdot , \cdot \right\rangle on MF(S)\mathcal{M}_{\mathrm{F}}(\mathcal{S}) such that ∥ν∥F=⟨ν,ν⟩1/2\left\| \nu \right\|_{\mathbf{F}} = \left\langle \nu , \nu \right\rangle^{1 / 2} for all ν∈M(S)\nu \in \mathcal{M}(\mathcal{S}), where ∥⋅∥F\left\| \cdot \right\|_{\mathbf{F}} is the usual total variation norm on M(S)\mathcal{M}(\mathcal{S}).

?
Problem 9A.11

For (X,S)(X, \mathcal{S}) a measurable space and b∈Xb \in X, define a finite (positive) measure δb\delta_{b} on (X,S)(X, \mathcal{S}) by

δb(E)={1 if b∈E0 if b∉E \delta _{b}(E) = \begin{cases} 1 & \text{ if } b \in E \\ 0 & \text{ if } b \notin E\end{cases}

for E∈SE \in \mathcal{S}.

?
(a)

Show that if b,c∈Xb, c \in X, then ∥δb+δc∥=2\left\| \delta_{b}+\delta_{c}\right\| = 2.

(b)

Give an example of a measurable space (X,S)(X, \mathcal{S}) and b,c∈Xb, c \in X with b≠cb \neq c such that ∥δb−δc∥≠2\left\| \delta_{b}-\delta_{c}\right\| \neq 2.

Problem 9B.1

Suppose ν\nu is a real measure on a measurable space (X,S)(X, \mathcal{S}). Prove that the Hahn decomposition of ν\nu is almost unique, in the sense that if A,BA, B and A′,B′A^{\prime }, B^{\prime } are pairs satisfying the Hahn Decomposition Theorem (9.23), then

∣ν∣(A\A′)=∣ν∣(A′\A)=∣ν∣(B\B′)=∣ν∣(B′\B)=0 \left|\nu \right|\left(A \backslash A^{\prime }\right) = \left|\nu \right|\left(A^{\prime } \backslash A\right) = \left|\nu \right|\left(B \backslash B^{\prime }\right) = \left|\nu \right|\left(B^{\prime } \backslash B\right) = 0
?
Problem 9B.2

Suppose μ\mu is a (positive) measure and g,h∈L1(μ)g, h \in \mathcal{L}^{1}(\mu ). Prove that g  dμ⊥h  dμg \; d \mu \perp h \; d \mu if and only if g(x)h(x)=0g(x) h(x) = 0 for almost every x∈Xx \in X.

?
Problem 9B.3

Suppose ν\nu and μ\mu are complex measures on a measurable space (X,S)(X, \mathcal{S}). Show that the following are equivalent:

?
(a)

ν⊥μ\nu \perp \mu.

(b)

∣ν∣⊥∣μ∣\left|\nu \right| \perp \left|\mu \right|.

(c)

Re⁡ν⊥μ\operatorname {Re} \nu \perp \mu and Im⁡ν⊥μ\operatorname {Im} \nu \perp \mu.

Problem 9B.4

Suppose ν\nu and μ\mu are complex measures on a measurable space (X,S)(X, \mathcal{S}). Prove that if ν⊥μ\nu \perp \mu, then ∣ν+μ∣=∣ν∣+∣μ∣\left|\nu +\mu \right| = \left|\nu \right|+\left|\mu \right| and ∥ν+μ∥=∥ν∥+∥μ∥\left\| \nu +\mu \right\| = \left\| \nu \right\| +\left\| \mu \right\|.

?
Problem 9B.5

Suppose ν\nu and μ\mu are finite (positive) measures on a measurable space (X,S)(X, \mathcal{S}). Prove that ν⊥μ\nu \perp \mu if and only if ∥ν−μ∥=∥ν∥+∥μ∥\left\| \nu -\mu \right\| = \left\| \nu \right\| +\left\| \mu \right\|.

?
Problem 9B.6

Suppose μ\mu is a complex or positive measure on a measurable space (X,S)(X, \mathcal{S}). Prove that

{ν∈MF(S):ν⊥μ} \left\{ \nu \in \mathcal{M}_{\mathbf{F}}(\mathcal{S}): \nu \perp \mu \right\}

is a closed subspace of MF(S)\mathcal{M}_{\mathrm{F}}(\mathcal{S}).

?
Problem 9B.7

Use the Cantor set to prove that there exists a (positive) measure ν\nu on (R,B)(\mathbb {R}, \mathcal{B}) such that ν⊥λ\nu \perp \lambda and ν(R)≠0\nu (\mathbb {R}) \neq 0 but ν({x})=0\nu (\left\{ x\right\} ) = 0 for every x∈Rx \in \mathbb {R}; here λ\lambda denotes Lebesgue measure on the σ\sigma-algebra B\mathcal{B} of Borel subsets of R\mathbb {R}.

?
Remark.
?

Example 9.29 (second bullet point): let r1,r2,…r_{1}, r_{2}, \ldots be a list of the rational numbers, and suppose w1,w2,…w_{1}, w_{2}, \ldots is a bounded sequence of complex numbers. Define a complex measure ν\nu on (R,B)(\mathbb {R}, \mathcal{B}) by ν(E)=∑{k∈N:rk∈E}wk2k\nu (E) = \sum_{\left\{ k \in \mathbb {N}: r_{k} \in E\right\} } \frac{w_{k}}{2^{k}} for E∈BE \in \mathcal{B}. Then ν⊥λ\nu \perp \lambda because ν\nu lives on Q\mathbb {Q} and λ\lambda lives on R∖Q\mathbb {R} \setminus \mathbb {Q}.

The second bullet point in Example 9.29 does not provide an example of the desired behavior because in that example, ν({rk})≠0\nu \left(\left\{ r_{k}\right\} \right) \neq 0 for all k∈Nk \in \mathbb {N} with wk≠0w_{k} \neq 0.

Problem 9B.8

Suppose ν\nu is a real measure on a measurable space (X,S)(X, \mathcal{S}). Prove that

ν+(E)=sup⁡{ν(D):D∈S and D⊆E} \nu ^{+}(E) = \sup \left\{ \nu (D): D \in \mathcal{S} \text{ and } D \subseteq E\right\}

and

ν−(E)=−inf⁡{ν(D):D∈S and D⊆E} \nu ^{-}(E) = -\inf \left\{ \nu (D): D \in \mathcal{S} \text{ and } D \subseteq E\right\}

for all E∈SE \in \mathcal{S}.

?
Problem 9B.9

Suppose μ\mu is a (positive) finite measure on a measurable space (X,S)(X, \mathcal{S}) and hh is a nonnegative function in L1(μ)\mathcal{L}^{1}(\mu ). Thus h  dμ≪dμh \; d \mu \ll d \mu. Find a reasonable condition on hh that is equivalent to the condition dμ≪h  dμd \mu \ll h \; d \mu.

?
Problem 9B.10

Suppose μ\mu is a (positive) measure on a measurable space (X,S)(X, \mathcal{S}) and ν\nu is a complex measure on (X,S)(X, \mathcal{S}). Show that the following are equivalent:

?
(a)

ν≪μ\nu \ll \mu.

(b)

∣ν∣≪μ\left|\nu \right| \ll \mu.

(c)

Re⁡ν≪μ\operatorname {Re} \nu \ll \mu and Im⁡ν≪μ\operatorname {Im} \nu \ll \mu.

Problem 9B.11

Suppose μ\mu is a (positive) measure on a measurable space (X,S)(X, \mathcal{S}) and ν\nu is a real measure on (X,S)(X, \mathcal{S}). Show that ν≪μ\nu \ll \mu if and only if ν+≪μ\nu^{+} \ll \mu and ν−≪μ\nu^{-} \ll \mu.

?
Problem 9B.12

Suppose μ\mu is a (positive) measure on a measurable space (X,S)(X, \mathcal{S}). Prove that

{ν∈MF(S):ν≪μ} \left\{ \nu \in \mathcal{M}_{\mathbf{F}}(\mathcal{S}): \nu \ll \mu \right\}

is a closed subspace of MF(S)\mathcal{M}_{\mathrm{F}}(\mathcal{S}).

?
Problem 9B.13

Give an example to show that the Radon-Nikodym Theorem (9.36) can fail if the σ\sigma-finite hypothesis is eliminated.

?
Problem 9B.14

Suppose μ\mu is a (positive) σ\sigma-finite measure on a measurable space (X,S)(X, \mathcal{S}) and ν\nu is a complex measure on (X,S)(X, \mathcal{S}). Show that the following are equivalent:

?
(a)

ν≪μ\nu \ll \mu.

(b)

For every ε>0\varepsilon > 0, there exists δ>0\delta > 0 such that ∣ν(E)∣<ε\left|\nu (E)\right| < \varepsilon for every set E∈SE \in \mathcal{S} with μ(E)<δ\mu (E) < \delta.

(c)

For every ε>0\varepsilon > 0, there exists δ>0\delta > 0 such that ∣ν∣(E)<ε\left|\nu \right|(E) < \varepsilon for every set E∈SE \in \mathcal{S} with μ(E)<δ\mu (E) < \delta.

Problem 9B.15

Prove 9.42 in the case where p=1p = 1.

?
Remark.
?

With the extra hypothesis that μ\mu is a σ\sigma-finite (positive) measure.

Theorem 9.42 (dual space of Lp(μ)L^p(\mu ) is Lp′(μ)L^{p'}(\mu )): suppose μ\mu is a (positive) measure and 1≤p<∞1 \leq p < \infty (with the additional hypothesis that μ\mu is a σ\sigma-finite measure if p=1p = 1). For h∈Lp′(μ)h \in L^{p'}(\mu ), define φh:Lp(μ)→F\varphi_h: L^p(\mu ) \rightarrow \mathbf{F} by φh(f)=∫fh dμ\varphi_h(f) = \int fh \, d\mu. Then h↦φhh \mapsto \varphi_h is a one-to-one linear map from Lp′(μ)L^{p'}(\mu ) onto (Lp(μ))′\left(L^p(\mu )\right)'. Furthermore, ∥φh∥=∥h∥p′\left\| \varphi_h\right\| = \left\| h\right\|_{p'} for all h∈Lp′(μ)h \in L^{p'}(\mu ).

Problem 9B.16

Explain where the proof of 9.42 fails if p=∞p = \infty.

?
Remark.
?

Theorem 9.42 (dual space of Lp(μ)L^p(\mu ) is Lp′(μ)L^{p'}(\mu )): suppose μ\mu is a (positive) measure and 1≤p<∞1 \leq p < \infty (with the additional hypothesis that μ\mu is a σ\sigma-finite measure if p=1p = 1). For h∈Lp′(μ)h \in L^{p'}(\mu ), define φh:Lp(μ)→F\varphi_h: L^p(\mu ) \rightarrow \mathbf{F} by φh(f)=∫fh dμ\varphi_h(f) = \int fh \, d\mu. Then h↦φhh \mapsto \varphi_h is a one-to-one linear map from Lp′(μ)L^{p'}(\mu ) onto (Lp(μ))′\left(L^p(\mu )\right)'. Furthermore, ∥φh∥=∥h∥p′\left\| \varphi_h\right\| = \left\| h\right\|_{p'} for all h∈Lp′(μ)h \in L^{p'}(\mu ).

Problem 9B.17

Prove that if μ\mu is a (positive) measure and 1<p<∞1 < p < \infty, then Lp(μ)L^{p}(\mu ) is reflexive.

?
Remark.
?

See the definition before Exercise 19 in Section 7B for the meaning of reflexive.

Problem 9B.18

Prove that L1(R)L^{1}(\mathbb {R}) is not reflexive.

?