Skip to content

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

OAI-26 family 098: a doubling set without finite-dimensional bi-Lipschitz coordinates

Editorial status: DRAFT_UNACCEPTED. NO_CERTIFICATION. This is a source-pinned documentary account, not an independent verification of a theorem.

The question in plain language

If a metric space looks small at every scale, must it admit a faithful geometric representation in a finite number of coordinates? The doubling property says that every ball can be covered by a bounded number of balls of half the radius. It captures a kind of local geometric simplicity, but does not itself provide coordinates.

A bi-Lipschitz embedding is a map that neither collapses distances nor stretches them beyond a fixed multiplicative factor. Family 098 concerns the Lang–Plaut question: whether doubling subsets of Hilbert space necessarily possess such embeddings into finite-dimensional Euclidean spaces.

A small example

A line segment in an infinite-dimensional Hilbert space is doubling, and it embeds isometrically into the real line. A finite-dimensional cube is also doubling and already has ordinary Euclidean coordinates.

These examples may suggest that doubling geometry always has a finite-dimensional description. The Family 098 manuscript claims otherwise. Its exceptional set is compact and has a uniformly bounded doubling constant, yet no finite target dimension and no finite bi-Lipschitz distortion suffice. The issue is not that one needs a large dimension: the asserted obstruction rules out every finite dimension.

What is known

The source-locked Family 098 catalogue identifies one manuscript, A doubling Hilbert subset with no finite-dimensional bi-Lipschitz embedding. The pinned overview reports that every infinite-dimensional real Banach space contains a compact doubling subset admitting no bi-Lipschitz embedding into any finite-dimensional real normed space. It additionally states that the doubling constant is universal, independent of the ambient Banach space.

In particular, because real infinite-dimensional Hilbert space is a Banach space, the result would negatively resolve the Lang–Plaut question even for a compact Hilbert subset.

The quantifiers matter. The claim is the existence of a specially constructed subset in each infinite-dimensional real Banach space; it does not say that every compact doubling subset fails to embed. Nor is this a statement against general nonlinear embeddings with uncontrolled distortion, or against embeddings into another infinite-dimensional target.

GCL verification state

Source lock: openai/math@1581d4cbdd69a0faf34b2089614fe2a814316a6a. The protected catalogue identifies one PDF; the pinned overview supplies the stated universality and nonembedding claims. This editorial summary is based on that pinned source description, without asserting an independent line-by-line proof review.

The published Lean repository also contains a theorem named OAI.DoublingHilbert.main specifying a set in a real Hilbert sequence space with a numerical doubling bound and nonembedding into each positive finite dimension. This observation is a pointer to a formalization candidate, not a GCL-verified elaboration, trust-chain audit, or certificate. The displayed Lean statement is narrower than the overview's every-Banach-space claim, and the two must not be conflated.

No Family 098-specific accepted source criticism, reproduced compiler receipt, Comparator determination, or MATHCERT certification is asserted in this document.

What remains

Read the pinned PDF theorem statements and construction, including the precise doubling definition, the quantitative universal constant, compactness proof, and use of infinite-dimensionality. Verify whether the argument extends from Hilbert space to every infinite-dimensional real Banach space without changing its hypotheses.

Check the obstruction against arbitrary finite-dimensional normed targets, rather than merely a chosen Euclidean norm. Independently audit the formal Lean imports, declaration, admitted axioms, compiler revision, and exact pinned commit. Then complete mathematical Markdown/HTML regression, source-critical editorial review, and protected readback; a documentary merge would not establish theorem certification.

Sources and claim boundaries

Boundary: NO_CERTIFICATION. The theorem claims are attributed to the pinned source and remain subject to independent mathematical checking.


Indexed primary manuscript paths: