Documentary state: DRAFT_UNACCEPTED. This draft has NOT been editorially admitted. Mathematical claims require independent checking.
OAI-26 family 094: subpolynomial dimension reduction in Lp
Editorial status: DRAFT_UNACCEPTED. NO_CERTIFICATION. This article documents the pinned publisher claim, not an independently checked proof.
The question in plain language
A finite set of points in a function space carries distances between every pair. Dimension reduction asks whether the same pattern of distances can be represented in a much smaller finite-dimensional space without changing them very much.
For Euclidean distances, powerful dimension-reduction theorems are known. The question becomes more difficult when distances are measured with an \(L_p\) norm for an exponent other than two. The pinned OAI-26 overview claims that a set of \(n\) points in real \(L_p\) can always be represented with a fixed arbitrarily small multiplicative distortion using a dimension that grows slower than every fixed positive power of \(n\).
A small example
Imagine collecting \(n\) signals, each represented by a real-valued function. The distance between two signals can be measured by taking the root of order \(p\) of the integral of the difference raised to the power \(p\).
An embedding with distortion at most \(D\) preserves ratios of pairwise distances up to a factor \(D\), after a permitted common rescaling. A distortion slightly larger than one allows approximation; an exact embedding requires no distance distortion at all.
The claimed target dimension is
This does not mean that the dimension is bounded by a constant. It means that for each fixed positive exponent, the dimension eventually grows more slowly than that power of \(n\).
What is known
The protected Family 094 catalogue identifies one manuscript, Subpolynomial dimension reduction in Lp (September 23, 2026).
The pinned publisher overview claims that for every fixed exponent \(1<p<\infty\) and every fixed distortion allowance \(D>1\), every finite subset of \(n\) points in real \(L_p\) embeds into a finite-dimensional \(\ell_p^d\) with distortion at most \(D\) and \(d=n^{o(1)}\).
The overview describes this as answering Naor's sublinear-dimension question for \(p\ne2\). It separately claims a contrasting worst-case exact-embedding dimension of order \(\Theta(n^2)\) for \(p\ne2\).
The approximate and exact conclusions have different quantifiers. In particular, fixed distortion greater than one must not be silently replaced by distortion exactly one, and the real-valued setting must not be generalized to complex \(L_p\) without source evidence.
GCL verification state
The source lock is openai/math@1581d4cbdd69a0faf34b2089614fe2a814316a6a. The protected catalogue records the manuscript and the publisher overview supplies the dimension and distortion claims.
This documentary pass does not independently reconstruct the embedding, verify its quantitative dimension bound or check the claimed quadratic exact-embedding obstruction. No Family 094-specific Lean proof, Comparator replay, semantic-equivalence audit or MATHCERT certificate is asserted.
What remains
Read the pinned manuscript and identify the exact embedding theorem, the asymptotic meaning and parameter dependence of \(n^{o(1)}\), and the convention for distortion and allowable scaling. Check whether the dimension estimate is deterministic, existential or randomized and which statements apply uniformly to the fixed parameters \(p\) and \(D\).
For exact embeddings, audit the construction of the worst-case lower bounds and the matching upper bound behind the stated \(\Theta(n^2)\) order. Separate the \(p=2\) case and any endpoint exclusions \(p=1,\infty\) from the claimed \(1<p<\infty\) result.
Complete mathematical Markdown and rendered-compendium validation, source-critical editorial review and protected merge/readback. Documentary admission is not proof certification.
Sources and claim boundaries
- Pinned subpolynomial dimension-reduction manuscript.
- Pinned OAI-26 overview, Family 094.
- GCL editorial assignment.
Boundary: NO_CERTIFICATION. These are publisher-reported claims and not independently certified mathematical results.
Indexed primary manuscript paths:
- Subpolynomial dimension reduction in Lp —
preprints/Subpolynomial-dimension-reduction-in-Lp-September-23-2026/paper.pdf