Section 21.5

The Representation of the Dual of C(X)C(X): The Riesz-Kakutani Theorem

[7/0%]
LaTeX
Problem 21.45

Let x0x_{0} be a point in the compact, Hausdorff space XX. Define L(f)=f(x0)L(f) = f\left(x_{0}\right) for each f∈C(X)f \in C(X). Show that LL is a bounded linear functional on C(X)C(X). Find the signed Radon measure that represents LL.

?
Problem 21.46

Let XX be a compact, Hausdorff space and μ\mu a Borel measure on B(X)\mathcal{B}(X). Show that there is a Radon measure μ0\mu_{0} for which

∫Xfdμ=∫Xfdμ0 for all f in C(X) \int _{X} f d \mu = \int _{X} f d \mu _{0} \text{ for all } f \text{ in } C(X)
?
Problem 21.47

Let g1g_{1} and g2g_{2} be two increasing functions on the closed, bounded interval [a,b][a, b] that agree at the end-points. Show that

∫abf(x)dg1(x)=∫abf(x)dg2(x) for all f∈C[a,b] \int _{a}^{b} f(x) d g_{1}(x) = \int _{a}^{b} f(x) d g_{2}(x) \text{ for all } f \in C[a, b]

if and only if g1(x+)=g2(x+)g_{1}\left(x^{+}\right) = g_{2}\left(x^{+}\right)for all a≤x<ba \leq x < b.

?
Problem 21.48

Let XX be a compact, Hausdorff space. Show that the Jordan Decomposition Theorem for signed Borel measures on B(X)\mathcal{B}(X) follows from the Riesz-Kakutani Representation Theorem for the dual of C(X)C(X) and Proposition 12.

The Riesz-Kakutani Representation Theorem: Let XX be a compact, Hausdorff space and C(X)C(X) the linear space of continuous real-valued functions on XX, normed by the maximum norm. Define the operator T:Radon⁡(X)→[C(X)]∗T: \operatorname {Radon}(X) \rightarrow [C(X)]^{*} by setting, for ν∈Radon⁡(X)\nu \in \operatorname {Radon}(X), Tν(f)=∫Xf dνT\nu (f) = \int_{X} f \, d\nu for all f∈C(X)f \in C(X). Then TT is a linear isometric isomorphism of Radon⁡(X)\operatorname {Radon}(X) onto [C(X)]∗[C(X)]^{*}.

Proposition 12: Let XX be a compact, Hausdorff space and C(X)C(X) the linear space of continuous real-valued functions on XX, normed by the maximum norm. Then for each continuous linear functional LL on C(X)C(X), there are two positive linear functionals L+L^{+} and L−L^{-} on C(X)C(X) for which L=L+−L−L = L^{+} - L^{-} and ∥L∥=L+(1)+L−(1)\left\| L\right\| = L^{+}(1) + L^{-}(1).

?
Problem 21.49

What are the extreme points of the unit ball of the linear space of signed Radon measures R\mathcal{R} adon (X)(X), where XX is a compact, Hausdorff space?

?
Problem 21.50

Let XX be a compact metric space. On the linear space of functions F\mathcal{F} defined in the statement of Theorem 16, define the norm of a function to be its total variation. Show that with this norm F\mathcal{F} is a Banach space.

Theorem 16 (Riesz): Let [a,b]\left[a, b\right] be a closed, bounded interval and F\mathcal{F} be the collection of real-valued functions on [a,b]\left[a, b\right] that are of bounded variation on [a,b]\left[a, b\right], continuous on the right on (a,b)(a, b), and vanish at aa. Then for each bounded linear functional ψ\psi on C[a,b]C\left[a,b\right], there is a unique function g∈Fg \in \mathcal{F} for which ψ(f)=∫abf(x) dg(x)\psi (f) = \int_{a}^{b} f(x) \, dg(x) for all f∈C[a,b]f \in C\left[a,b\right].

?
Problem 21.51

Let A\mathcal{A} be an algebra of real-valued continuous functions on a compact space XX that separates points and contains the constants. Let A⊥\mathcal{A}^{\perp } be the set of signed Radon measures on XX such that ∣μ∣(X)≤1\left|\mu \right|(X) \leq 1 and ∫Xfdμ=0\int_{X} f d \mu = 0 for all f∈Af \in \mathcal{A}.

?
(i)

Use the Hahn-Banach Theorem and the Riesz-Kakutani Representation Theorem to show that if A⊥\mathcal{A}^{\perp } contains only the zero measure, then A‾=C(X)\overline{\mathcal{A}} = C(X).

(ii)

Use the Krein-Milman Theorem and the weak-* compactness of the unit ball in R\mathcal{R} adon (X)(X) to show that if the identically zero measure is the only extreme point of A⊥\mathcal{A}^{\perp }, then A⊥\mathcal{A}^{\perp } contains only the identically zero measure.

(iii)

Let μ\mu be an extreme point of A⊥\mathcal{A}^{\perp }. Let f∈Af \in \mathcal{A}, with 0≤f≤10 \leq f \leq 1. Define measures μ1\mu_{1} and μ2\mu_{2} by

μ1(E)=∫Efdμ and μ2(E)=∫E(1−f)dμ for E∈B(X) \mu _{1}(E) = \int _{E} f d \mu \text{ and } \mu _{2}(E) = \int _{E}(1-f) d \mu \text{ for } E \in \mathcal{B}(X)

Show that μ1\mu_{1} and μ2\mu_{2} belong to A⊥\mathcal{A}^{\perp } and, moreover, ∥μ1∥+∥μ2∥=∥μ∥\left\| \mu_{1}\right\| +\left\| \mu_{2}\right\| = \left\| \mu \right\|, and μ1+μ2=μ\mu_{1}+\mu_{2} = \mu. Since μ\mu is an extreme point, conclude that μ1=cμ\mu_{1} = c \mu for some constant cc.

(iv)

Show that f=cf = c on the support of μ\mu.

(v)

Since A\mathcal{A} separates points, show that the support of μ\mu can contain at most one point. Since ∫X1dμ=0\int_{X} 1 d \mu = 0, conclude that the support of μ\mu is empty and hence μ\mu is the zero measure.