Documentary state: EDITORIALLY_ADMITTED. The manuscript has a protected editorial admission, not a theorem certification.
OAI-26 family 083: Hilbert transforms along Lipschitz directions
Status: source-grounded editorial exposition. NO_CERTIFICATION.
The question in plain language
The directional Hilbert transform integrates a function along a line whose orientation is determined by a unit vector field. In contrast to the constant-direction case, a variable direction field produces a singular integral that is not a translation-invariant Fourier multiplier. The question is whether Lipschitz control of the direction guarantees a uniform L2 estimate on sufficiently short segments.
A small example
For a constant horizontal vector field, the operator is a copy of the ordinary one-dimensional Hilbert transform on each horizontal line, so rotation and Fubini give the familiar L2 bound. A field depending on both planar coordinates cannot be reduced in this way. A Lipschitz condition controls its turning but not automatically the interactions of its frequency components.
What is known
At pinned source revision openai/math@1581d4cbdd69a0faf34b2089614fe2a814316a6a, the manuscript A uniform Hilbert transform estimate for Lipschitz directions states universal constants: every planar Lipschitz unit direction field with Lipschitz seminorm at most one has an L2 bound for each symmetric hard inner truncation at one fixed short outer length. Its direction is held fixed at the output point. The supremum over inner truncations is outside the L2 norm: it is an operator bound for each cutoff, not a maximal truncation operator assertion.
The introduction also states a symmetric principal-value estimate for Schwartz inputs, an L2 extension by density, and a weak-(2,2) level-set inequality. Scaling provides an outer length proportional to the reciprocal positive Lipschitz seminorm. Constant fields fall under the classical one-dimensional theory. No claim is made for arbitrary long outer lengths for variable fields.
GCL verification state
The publisher's claimed proof mechanism is described here, not independently verified. The introduction outlines local slope reduction, vertical frequency-band decomposition, and a logarithmically improving restricted band estimate. A commutator supports Littlewood-Paley reassembly. Geometric packing of directional wave packets uses the Lacey-Li short Lipschitz-popularity maximal estimate. A finite Fourier inequality with constant exactly one, martingale variation bounds, nested angular intervals, and telescoping normalized packet energies provide the phase gain needed to sum frequency and amplitude levels.
Its named specialized external inputs are Lacey-Li's popularity estimate and Lepingle's unweighted martingale 3-variation inequality. Previous annular, one-coordinate, analytic, and lacunary direction results have different assumptions.
What remains
The primary introduction and the publisher's lean/docs/083.md scope were inspected at the pinned revision. That scope describes a formal short-scale statement, including a radius bounded by one millionth of the reciprocal positive Lipschitz constant, along with principal-value and weak-(2,2) consequences. It links the LipschitzHilbert.lean comparator. This description does not itself verify the Lean dependency closure or certify the analytic proof.
Outstanding obligations include full checks of truncation endpoint commutators, restricted band estimates, popularity hypotheses, variation arguments, the constant-one Fourier inequality, angular telescoping and the exact Lean comparator theorem and dependencies.
Sources and claim boundaries
NO_CERTIFICATION.
Indexed primary manuscript paths:
- A uniform Hilbert transform estimate for Lipschitz directions —
preprints/A-uniform-Hilbert-transform-estimate-for-Lipschitz-directions-September-25-2026/main.pdf