Skip to content

Documentary state: DRAFT_UNACCEPTED. This draft has NOT been editorially admitted. Mathematical claims require independent checking.

OAI-26 family 080: the Sobolev endpoint for Schrödinger convergence

Editorial status: DRAFT_UNACCEPTED. NO_CERTIFICATION. This article documents claims from two pinned manuscripts and does not independently verify them.

The question in plain language

The free Schrödinger equation describes how an initial wave evolves. At time zero its solution should recover the initial function, but for rough initial data it is difficult to show convergence at almost every spatial point as time approaches zero. Carleson's Schrödinger convergence problem asks exactly how much Sobolev smoothness is needed.

The pinned OAI-26 overview claims convergence at the sharp threshold itself, not just for smoother inputs, in every spatial dimension at least two.

A small example

Write the free evolution of initial data \(f\) as \(e^{it\Delta}f\). The mathematical question is whether

\[ e^{it\Delta}f(x)\longrightarrow f(x) \quad\text{as }t\downarrow0 \]

for almost every point \(x\). Sobolev space \(H^s\) describes functions with \(s\) derivatives in an averaged square-integrable sense, including fractional values of \(s\).

The indexed critical exponent in dimension \(n\) is

\[ s=\frac{n}{2(n+1)}. \]

For \(n=2\), it equals \(1/3\). A theorem for all \(s>1/3\) would not settle whether convergence holds at \(s=1/3\) itself; that equality case is the claimed advance.

What is known

The source-locked catalogue identifies two manuscripts, both dated September 24, 2026: Endpoint convergence for the planar Schrodinger equation and Endpoint pointwise convergence for the Schrodinger equation in higher dimensions.

The publisher overview claims that for every integer dimension \(n\ge2\) and every initial datum in the critical Sobolev space, the free Schrödinger evolution converges to the datum almost everywhere as positive time tends to zero:

\[ f\in H^{n/(2(n+1))}(\mathbb R^n) \quad\Longrightarrow\quad e^{it\Delta}f(x)\to f(x)\ \text{a.e. as }t\downarrow0. \]

The scope combines a planar endpoint and higher-dimensional endpoint results. Almost-everywhere convergence is not uniform pointwise convergence or convergence at every point, and a positive-time limit is not automatically a statement about arbitrary non-tangential approaches. Exact Fourier conventions and maximal estimates require checking the papers.

GCL verification state

The source pin is openai/math@1581d4cbdd69a0faf34b2089614fe2a814316a6a. The protected catalogue lists the two PDF sources; the pinned overview supplies the stated critical-exponent claim.

This documentary pass has not independently checked the proof of endpoint convergence, its maximal-operator estimates, or the relation between planar and higher-dimensional arguments. No Family 080-specific Lean formalization, Comparator replay, semantic-equivalence audit or MATHCERT certificate is asserted.

What remains

Read both pinned manuscripts and identify the exact theorems, function-space definitions and quantifiers. Check whether the convergence is local or global almost everywhere, how representatives of Sobolev functions are handled, and whether all initial data in the stated critical spaces are covered.

Audit the endpoint maximal estimates and the distinction between exact equality and the easier strict-above-threshold results. Verify that the planar argument yields \(H^{1/3}\) and that the higher-dimensional results cover every \(n\ge3\) under the same time-limit convention.

Complete Markdown and rendered-compendium validation, source-critical review and protected merge/readback. Editorial admission does not constitute certification of the mathematical proof.

Sources and claim boundaries

Boundary: NO_CERTIFICATION. The theorem remains publisher-reported in this documentary article, without independently verified proof certification.


Indexed primary manuscript paths: