Skip to content

Documentary state: EDITORIALLY_ADMITTED. The manuscript has a protected editorial admission, not a theorem certification.

OAI-26 family 085: Endpoint Sobolev regularity of centered disk averages

Editorial source review only. NO_CERTIFICATION.

The question in plain language

A centered maximal average examines disks of every radius centered at each point and returns the largest average magnitude. The endpoint problem asks whether the total variation of that output, measured through an integrable weak gradient, is controlled by the total gradient of a Sobolev input.

A small example

For a smooth planar bump, different disk radii may give the largest average at neighboring centers. This can create complicated changes in the maximal output even when the input is smooth. The standard argument for powers greater than one uses strong integrability of the maximal operator; it does not work at power one.

What is known

The pinned paper, An Endpoint Gradient Bound for the Centered Disk Maximal Operator, claims that every real-valued W^{1,1} function on the Euclidean plane has a centered disk maximal function in local W^{1,1} whose weak gradient is globally integrable. Its total gradient is bounded by an absolute constant times the total gradient of the input. All positive disk radii are allowed, with no radiality or compact-support hypothesis. The maximal function itself need not be globally integrable. The result does not extend by assertion to other dimensions, to arbitrary averaging geometries, or to BV data.

The source identifies a signed finite-radius-band gradient inequality for compactly supported smooth real inputs as a central auxiliary result. Its bound is independent of the number of dyadic radius scales; signed averages permit subtraction of local means in the final localization. Previously established one-dimensional, radial, uncentered, cube, and fractional-order endpoint theorems have different hypotheses.

GCL verification state

The introduction outlines disk moments, shell concentration, a dyadic scalar cancellation price budget, vector-particle direction choices, and cancellation of differentiated transition rows. It then extends finite-band estimates to Sobolev inputs and removes the singular component of the distributional derivative. This last absolute-continuity step is essential: a bound on total variation alone does not imply the W^{1,1} conclusion.

The pinned publisher Lean scope names two comparators: SignedFiniteBand and DiskMaximal, and expressly excludes a BV extension. This description is not independent formal proof replay or a certificate.

What remains

Independently audit the disk moment identities, shell concentration and endpoint terms, scalar cancellation and vector-particle costs, differentiated transition-row cancellation, scale-uniform finite-band inequality, Sobolev extension, and proof of absolute continuity of the derivative measure. Verify exact comparator statement parity, imports, nonvacuity and proof replay. Documentary admission additionally requires an exact article digest, source critic, compendium regression, strict documentation build, required CI and protected-merge readback.

Sources and claim boundaries

NO_CERTIFICATION.


Indexed primary manuscript paths: