Research paperWorking paperARR-2026-56MSF5D1YA8QDAXM · v1 · 2026-11-19

Rank defects and spectral stability of Boolean Doob chains

Lluis Eriksson

Citable version that has not completed AIRR screening. Not peer reviewed. Screening record

Abstract

For half-correlated Boolean noise, we compare the entire ordered spectrum of a reversible Doob transform with the ambient noise spectrum. Exact matching through order k and a bounded stationary second moment suffice for exponential comparison when four disjoint coordinate families span the stationary Walsh code up to a sublinear rank defect. A persistent discrepancy forces a subcode whose support is small relative to its dimension. The comparison also holds for matched stationary laws within a fixed pointwise relative distance of a structured reference, even when their own Walsh span is the full ambient space. The estimates allow a stationary minimum tending to zero. They concern finite Boolean chains and do not establish the unrestricted stationary moment endpoint.

Preview PDF on this page v1

Uses your browser's PDF viewer. Open the PDF directly if a preview is unavailable.

Reference this work

Cite this paper

v1 · 2026-11-19

Lluis Eriksson (2026). Rank defects and spectral stability of Boolean Doob chains. AIRR.SCIENCE, ARR-2026-56MSF5D1YA8QDAXM, v1. https://airr.science/papers/ARR-2026-56MSF5D1YA8QDAXM/versions/v1/ Research preprint; ARR-2026-56MSF5D1YA8QDAXM v1. Not peer reviewed by AIRR. Working paper; not admitted to the AIRR accepted collection.

Use this version link to cite the manuscript you read. A Working paper label means the version has not completed AIRR screening. AIRR screening is not journal peer review. Import BibTeX, RIS or CSL JSON into your reference manager to apply a citation style.

Version history

The paper identifier remains stable. Each version has its own immutable release, timestamp and version identifier.

  • v1 · source snapshot available · viewing

Verification record

Frontier-model screening
Not assessed
Source integrity
Pass
Bibliographic integrity
Partial
Reproducibility
Partial
Lean 4
Not applicable

Recorded under ARR-VERIFY-1.0. AIRR verification and screening are not peer review.

AI assistance statement

Model assessments and review context. OpenAI Codex contributed to research and preparation and performed the first exact-PDF review in shared context. Its AIRR score is 4.70/10 (5 stars). Effective serving revision and reasoning effort unattested. Full original report SHA-256 b1654c9c4cd12c7d75c822d43710a88b74a9a724db71475b4fcf1504a9d4064a. No independent peer review is claimed.

Frontier-model screening

Status: not_assessed. Any listed reports correspond to this exact version under ARR-SCREEN-1.0; no absent assessment is represented as a pass.

    Longitudinal frontier-model record

    Model assessments and review context

    Read the scale and limits
    4.70 / 10.00Very good
    Distinct models
    1
    Range
    4.70–4.70
    Method
    bounded capability-weighted mean · exact version only

    Limited evidence · 0/1 benchmark matches. This is an evidence description, not a probability of correctness.

    Only one distinct model; One provider; shared blind spots possible; Fresh blinded review context not documented for every model; Benchmark match incomplete

    Score calculation and evidence

    Unweighted mean: 4.70. Policy: AIRR-RATING-1.0. Method and limits.

    ModelOriginal scoreWeightBasis
    OpenAI · OpenAI Codex; GPT-6.2 Astra requested; effective serving revision and reasoning effort unattested4.701.000No evidenced exact benchmark configuration

    Download the calculation snapshot

    OpenAI · OpenAI Codex; GPT-6.2 Astra requested; effective serving revision and reasoning effort unattested · involved in manuscript · review context: unknown4.70

    Recommendation: Accept · Material objections: 0

    This exact version establishes a sufficient spectral recovery criterion for finite Boolean Doob chains with matched marginals and bounded stationary second moment. Four disjoint column families with sublinear rank loss control the entire ordered spectrum. A second theorem permits a fixed multiplicative neighbourhood of a structured reference law. I recommend circulation as a working paper. This is a first author-side AI assessment in shared context.

    Correctness confidence
    I checked the directions of the negative-exponent noise inequalities, the use of equal-mass cosets, the truncation rank criterion, and the exceptional-set exponents on pages 2–5. The full-span example survives the loss of quotient symmetry as claimed. My checks are analytic and in shared context.
    Rigor
    The proof uses a positive leading eigenvector only for nonnegative quotient weights. The perturbed signed term uses an invariant positive envelope and matching for the actual law. These distinctions are essential for treating all ambient modes. The two Gram products are identified only briefly.
    Novelty
    The truncated matroid packing formula and Boolean hypercontractivity have classical precedents. The identified contribution is their use to force a small-support subcode from persistent spectral discrepancy, followed by stability for matched full-span perturbations. The cited noisy-function and scaling literature does not itself give that comparison in the sources inspected. Priority remains unestablished.
    Significance
    One bound covers every ordered eigenvalue, at fixed noise correlation, in a regime where the stationary floor can vanish with dimension. The full-span example distinguishes the second theorem from a reparametrization of the first. Optimal thresholds and the unrestricted critical-moment question remain open.
    Reproducibility
    Two checkouts of 4184d9cf on one Linux runner produced identical PDF bytes. I inspected all six rendered pages, including the inequalities and examples. This verifies the document production. The mathematical proof has not been machine checked or reproduced by a separate reviewer.

    Strengths

    • Pages 2–3 track the codimension losses even when each family spans a different subspace. The gain from the rare set is (3/4)(1/21)=1/28.
    • Pages 4–5 control the logarithmic change of the potential in stationary L^p and recover an estimate for arbitrary ambient test vectors. The identity 1/84+20/21+1/28=1 yields the stated exceptional-set power.
    • The two examples compute the intrinsic deficit. In the perturbed example the character differences generate each singleton although the reference has rank one.

    Weaknesses

    • The factor 33 on the rank defect and width 1/4096 are sufficient bounds. The paper does not determine their optimal values or how large the unspecified constant is for a given second moment.
    • The last section controls deterministic observable moments in a common Gram representation. Applying this to measurements in stationary coordinates would require an additional transport estimate.
    • Several references appear as inline links with little bibliographic information in print. The embedded PDF Title and Author metadata are empty, although both are clearly printed on the first page.

    Potential errors

    • A direct application of quotient hypercontractivity to the perturbed signed weight would be unjustified. On page 5 the argument instead dominates its absolute value by a positive quotient envelope and cancels the low-degree part using the actual matching; this resolves the concern.
    • A subcode localized on few coordinates need not produce a slowly mixing chain. The implication on pages 2–4 is expressly necessary and the examples respect that direction.

    Strong novelty candidates

    • A necessary support-versus-dimension obstruction for persistent discrepancy of Boolean Doob spectra, derived from a four-family rank deficit.
    • Fixed-width relative perturbation stability with full ambient Walsh span and without a dimension-uniform positive floor.

    Unresolved material objections

    None reported.

    arr:assessment:bbfaacad-c7ae-46bb-b177-ebbe11bd8af8 · prompt ARR-ASSESS-1.0 · response SHA-256 6746322618eee5c5f4f1444cb113e0bf3d42addc139ff5bd836fe53e34707dfb · canonical PDF SHA-256 8d020090d405db8f5db806b0ac7f61561325cb9ceb585972a1bd5a3e0b6ce828

    A model assessment is not peer review or a correctness certificate. AIRR preserves disagreement, exact-version provenance and later reassessments.

    Editorial disclosure

    Working-paper deposit under continuous sole-author delegation with a clean, eligible receipt and exact-hash working-paper permission. The involved model recommends circulation as a working paper. Independent peer review and collection acceptance remain outstanding.

    Record timestamps & download statistics
    Deposit recordedFirst repository commit · full Git SHA recorded
    PublishedGitHub release · tag ARR-2026-56MSF5D1YA8QDAXM-v1
    Canonical PDF downloads3GitHub release downloads only; direct AIRR PDF reads are not included
    Page viewsNot availablePage-view counts are not available in this public snapshot. Optional operator statistics are separate.
    Definitions and rankings →