Alphabeta Math
TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-09-14
How statement and proof provenance work

The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.

  • Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
  • AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
  • AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.

These labels describe origin, not correctness: citations and verification chips remain separate evidence.

Separable dual spaces have the Radon--Nikodym property

Statement

Assume the Axiom of Choice. If Y is a real or complex normed space and its continuous dual X=Y∗ is norm separable, then the Banach space X has the Radon--Nikodym property.

Facts & Assumptions

[A1]

The Axiom of Choice holds (The Axiom of Choice).

[L1]

AC implies Countable Choice and the relative Hahn--Banach principle: the former follows from the preceding local choice lemma, while the latter is realized by the AC form of dominated Hahn--Banach (AC supplies the countable and dependent choices used in Banach integration, Hahn-Banach dominated extension theorem for real vector spaces).

[L2]

Under Countable Choice and relative Hahn--Banach, norm separability of Y∗ implies norm separability of Y (Separable dual implies separable primal).

[L3]

RNP is the Bochner-density assertion for every absolutely continuous bounded-variation vector measure over a finite measure (Radon--Nikodym property), and the variation of such a vector measure is a finite positive measure (Bounded variation of a vector measure is a finite measure).

[L4]

On the finite measure spaces fixed in [L3], hence on sigma-finite reference spaces, AC gives integrable scalar densities for finite absolutely continuous signed and complex measures (A sigma-finite signed measure that is absolutely continuous with respect to a sigma-finite positive measure has a unique almost-everywhere density, A finite complex measure absolutely continuous with respect to a sigma-finite positive measure has an integrable complex density).

[L6]

Countable scalar suprema and pointwise limits preserve measurability (Sequential suprema, infima, limsup, liminf, and pointwise limits of measurable functions are measurable).

[L7]

A strongly measurable Banach-valued function is Bochner integrable when its norm is integrable (Bochner integrability criterion), and bounded linear maps commute with its integral (Bounded linear maps commute with Bochner integration).

Proof

technique · direct

Given: AC, a normed space Y, and a norm-separable dual X=Y∗.

1.1givenA1L1L2

Obtain the needed separability and choice interfaces. By [L1], AC supplies Countable Choice and proves every instance of the relative Hahn--Banach principle. Thus [L2] applies and makes Y norm separable. Adjoin zero to chosen countable dense subsets of Y and X so that [L8] enumerates both even in the zero-space case.

1.2givenA1L3L4

Fix a vector measure and dominate it by one scalar density. Let (Ω,A,μ) be a finite measure space and let ν:A→Y∗ have bounded variation with ν≪μ. If μ(E)=0, every cell in a finite partition of E has zero ν-value, so ∣ν∣(E)=0. Hence ∣ν∣≪μ. By [L3] it is a finite positive measure, and [L4] gives an integrable real density g with ∣ν∣(E)=∫Eg dμ. Positivity, tested on {g≤−1/m}, permits replacing g on a null set so that g≥0 everywhere.

2.1A1L4L8step 1.1step 1.2choose

Choose a countable linear test space and all its scalar densities. Let K0=Q in the real case and K0=Q+iQ in the complex case. The K0-linear span D of a countable dense subset of Y is countable and norm dense by [L8]. For d∈D define the finite signed or complex measure νd(E)=ν(E)(d). Its partition sums satisfy ∣νd∣(E)≤∥d∥ ∣ν∣(E), and νd≪μ. Apply [L4], using AC to choose simultaneously for all d∈D, measurable hd∈L1(μ) such that νd(E)=∫Ehd dμ.

3.1L3L5step 1.2step 2.1

Make the scalar representatives pointwise linear and bounded. Uniqueness of scalar densities says, for every a,b∈K0 and d,e∈D, that had+be=ahd+bhe almost everywhere. There are only countably many such relations. Moreover [L5] and the variation estimate in step 2.1 give

∫E∣hd∣ dμ=∣νd∣(E)≤∥d∥∫Eg dμ.

Testing this inequality on {∣hd∣>∥d∥g+1/m} shows ∣hd∣≤∥d∥g almost everywhere, for every d∈D. The union of the exceptional sets for all relations, bounds, and d is null by countable additivity. Replace every hd by zero there. Off this one null set, the map d↦hd(ω) is K0-linear and bounded by g(ω)∥d∥.

4.1step 3.1construct

Extend the pointwise functionals to Y∗. For every remaining ω, continuity and density of D extend d↦hd(ω) uniquely to a scalar-linear functional f(ω)∈Y∗ with ∥f(ω)∥≤g(ω). In the complex case, Q+iQ-linearity and continuity give full complex linearity. Set f=0 on the common null set. Then f(ω)(d)=hd(ω) for all d∈D off that set.

5.1L6L8step 1.1step 3.1step 4.1construct

Prove strong measurability rather than merely coordinate measurability. Fix y∈Y. Using an enumeration of D, for every integer m≥1 take the least indexed dm∈D with ∥dm−y∥<1/m. Step 3.1 gives hdm(ω)→f(ω)(y) off the common null set, so [L6] makes every coordinate ω↦f(ω)(y) measurable. Let (uj)j≥1 enumerate a countable dense subset of the unit ball of Y obtained from D by rational rescaling. For each x∗∈X,

∥f(ω)−x∗∥=sup⁡j≥1∣f(ω)(uj)−x∗(uj)∣,

so [L6] makes this distance measurable. Finally enumerate a norm-dense positively indexed sequence (xk∗)k≥1 in the separable space X. For each integer m≥1, assign to ω the least indexed nearest point among x1∗,…,xm∗. The measurable distance functions make its finitely many tie-broken cells measurable, and density makes these simple functions converge in norm to f(ω). Thus f is strongly measurable.

6.1L7step 2.1step 4.1step 5.1

Integrate the extension and identify the vector measure. The bound ∥f∥≤g and [L7] make f Bochner integrable. For d∈D, boundedness of evaluation at d, commutation in [L7], and step 2.1 give

(∫Ef dμ)(d)=∫Ef(ω)(d) dμ=∫Ehd dμ=ν(E)(d).

Both ∫Ef and ν(E) are continuous functionals on Y and agree on the norm-dense subspace D, so they agree on all of Y. Hence ν(E)=∫Ef dμ for every measurable E.

7.1A1L3step 1.1step 6.1∎

Conclude RNP and close the degenerate cases. [A1, L3, step 1.1, step 6.1] The measure space and ν were arbitrary, so step 6.1 proves the RNP condition in [L3]. If Y∗={0}, every scalar measure and every density above is zero; if μ(Ω)=0 or ν=0, take g=f=0. A one-point dense set and a one-element rational span are covered by the same construction. AC is used for Hahn--Banach and Countable Choice in step 1.1, scalar RN and simultaneous representatives in steps 1.2--2.1, and the common countable family of a.e. relations; no stronger unstated choice is used.

Depends on

Used by

Dependency tree · two levels

77 results within two dependency steps of this one, each drawn at its shortest distance from it. An arrow runs from a result to what uses it, so the chart reads left to right and ends at this result, which carries a heavier outline. Every node is a link to that result. Click elsewhere on the chart to enlarge it.

Sources