Alphabeta Math
TheoremStatement: Literature-sourcedProof: AI-adaptedPipeline-generatedprecheck pass
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.

Carleson maximal operator is strong L2

Statement

Assume AC. The one-sided real-line Carleson maximal operator extends boundedly to complex L2(R).

Facts & Assumptions

[F1]

Uniform finite-model strong Lp estimates transfer to the real-line maximal operator on Schwartz input Wave packet model dominates the linearised carleson operator.

[F2]

Finite models are uniformly restricted weak type (q,q) for every 1<q<∞ Hunt exceptional set and distribution estimates.

[F3]

Restricted weak bounds at 1<r<p<s<∞ give a uniform strong(p,p) finite-model bound Carleson restricted weak interpolation.

[F4]

The real-line operator on Schwartz functions is the supremum of the absolute values of linear one-sided Fourier cutoffs Carleson operator and measurable linearisation.

[F5]

Schwartz classes are dense in complex L2 under countable choice Schwartz space is dense in L2.

[F6]

Complex L2 is complete under countable choice and norm convergence has an almost-everywhere convergent subsequence Complex completeness, density, and inner product: the consumer interface.

[F7]

Assume AC The Axiom of Choice, supplying the countable choice in F5 and F6 and the inherited analytic interfaces.

Proof

Given: The stated AC assumption and the exact one-sided operator of F4.

1.1F1F2F3

Apply F2 at r=3/2 and s=3, then F3 with p=2. These strict endpoint exponents give a finite-model strong L2 constant independent of the family and selector. F1 gives ∥CRu∥2≤K∥u∥2 for every complex Schwartz u, for a fixed finite K. This uses estimates on both sides of two; no restricted weak-L2-to-strong-L2 inference is made.

2.1F4step 1.1

For Schwartz u,v, linearity of every cutoff and ∣∣a∣−∣b∣∣≤∣a−b∣ imply pointwise ∣CRu−CRv∣≤CR(u−v). The suprema are finite because the Schwartz transform is integrable. Thus step 1.1 gives ∥CRu−CRv∥2≤K∥u−v∥2. Also CR(zu)=∣z∣CRu and CR(u+v)≤CRu+CRv pointwise.

3.1F5F6F7step 2.1

For each fixed f∈L2, F5 and the countable choice supplied by F7 give Schwartz uj with ∥uj−f∥2<1/j. Step 2.1 makes (CRuj) Cauchy in L2, so F6 gives a limit; define C‾Rf to be this class. If vj is another such sequence, the same Lipschitz inequality bounds the distance between their output sequences by K∥uj−vj∥2→0, so the limit is independent of the approximation. On a Schwartz class take the constant sequence to see agreement with the original operator. Passing norms to the limit gives ∥C‾Rf∥2≤K∥f∥2.

4.1F6F7step 2.1step 3.1∎

Applying the same argument to approximants for f,g gives ∥C‾Rf−C‾Rg∥2≤K∥f−g∥2, so the extension is continuous and is the unique continuous extension from the dense Schwartz classes. It is nonnegative almost everywhere: F6 supplies an a.e.-convergent subsequence of its nonnegative approximating outputs. Homogeneity and subadditivity also pass from step 2.1; for subadditivity take subsequences along which the three output sequences for uj, vj and uj+vj converge a.e., successively using F6, and pass the pointwise inequality off the finite union of exceptional null sets. The zero class maps to zero. This is the required bounded maximal-operator extension on complex L2, with precisely the inherited AC assumption and no simultaneous arbitrary-index choice of approximants.

Depends on

Used by

Nothing in the library uses this result yet.

Dependency tree · two levels

35 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