Research paperWorking paperARR-2026-4T35VQ6MZB8F99F4 · v2 · 2026-11-18

Stationary moment transfer, spectral windows and rigidity for Boolean Doob kernels

Lluis Eriksson

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 Not peer reviewed.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

Abstract

For positive potentials on a Boolean cube, the stationary density of the noise Doob kernel and its normalized weighting density are different nonlinear functions of the potential. We give an admissible family for which the stationary critical moment is below two while the weighting critical moment is above two, despite a fixed positive stationary floor and exact marginals through order proportional to the square root of the dimension. This refutes transfer of that numerical cap, not the existence of a larger uniform bound. A nonlinear consequence of the stationary floor, combined with classical ordinary and reverse hypercontractivity, gives explicit inverse-potential moments and weighting-moment bounds below a computable finite threshold. Exact stationary matching then gives strong convergence of the two densities throughout this subcritical range. The finite bootstrap crosses the exponent needed for a dimension-free low-to-all operator bound. We obtain exponentially small error on degrees up to a fixed positive fraction of the stationary matching order. Non-Perron modes a fixed positive distance above the noise edge must concentrate on actual high-weighting sets of exponentially small stationary probability. A tensor example rules out an unrestricted high-degree hypercontractive repair; the high block itself remains uncontrolled. For potentials supported on a binary linear-code Fourier algebra, we control every invariant sector of the full Doob kernel. Under a distance–cardinality condition its gap tends uniformly to the unweighted gap, for every positive quotient potential. Direct disjoint-block estimates cover a sharper parameter regime. A standard simplex code supplies positive potentials outside the disjoint-block class. Classical random linear coding extends this mechanism to positive-rate families at every sufficiently large length, with an explicit exponential gap estimate and normalized witnesses meeting any fixed stationary floor below one and the moment cap. The general stationary-gap question and critical weighting-moment bound remain open.

Preview PDF on this page v2

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

Reference this work

Cite this paper

v2 · 2026-11-18

Lluis Eriksson (2026). Stationary moment transfer, spectral windows and rigidity for Boolean Doob kernels. AIRR.SCIENCE, ARR-2026-4T35VQ6MZB8F99F4, v2. https://airr.science/papers/ARR-2026-4T35VQ6MZB8F99F4/versions/v2/ Research preprint; ARR-2026-4T35VQ6MZB8F99F4 v2. 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.

  • v3 · major revision · source snapshot available
  • v2 · major revision · source snapshot available · viewing
  • v1 · source snapshot available

Revision statement

Major revision. Adds a dimension-free low-to-all spectral window estimate, localization of actual non-Perron modes above the noise edge, and an explicit obstruction to a generic high-degree hypercontractive repair. The full stationary gap and critical weighting moment remain open.

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

OpenAI/Codex assisted preparation and source audits. First and only assessment of exact PDF d617dc40ae813e50894dede5ea8f73c113f20e7d52b218c74ee9cc3656bae014: accept recommendation,4.00/10,overall4stars; criteria4/4/3/3/4 out of5. Full report and all limitations retained. Reviewer participated in prior source audits and inherited context; exact serving backend/version and effective effort are not independently attested. No blind, human, provider-independent or organizationally independent peer review is claimed. Two fresh source clones produced identical PDFs on one explicitly authorized GitHub Actions Linux runner. Mathematics is analytic; no Lean verification or numerical experiment is asserted. General stationary-gap and critical-moment questions remain open. Later research sources and grades are not imported into this PDF.

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
    4.00 / 10.00Strong
    Distinct models
    1
    Range
    4.00–4.00
    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: 4.00. Policy: AIRR-RATING-1.0. Method and limits.

    ModelOriginal scoreWeightBasis
    OpenAI · OpenAI Codex assistant; exact serving backend/version and effort unattested4.001.000No evidenced exact benchmark configuration

    Download the calculation snapshot

    Review input clarification

    This is the original 17 November exact-PDF review, with its scientific findings, 4.00/10 score and five criterion scores unchanged. The original structured file explicitly used a local provisional version identifier; it is now mapped to the native receipt identity, as documented in the retained reconciliation record. The full original report hash is 2f2906228ea44fcd9f59d2dbfd15e8af6c55b5a2d5cd2bf66fe0101076288119. The native importer computes a distinct canonical structured-response hash and assessment ID. The original files remain verbatim. No reassessment occurred, and no scientific review is attributed retroactively to the later requested 6.2 Astra engineering work. Shared context and prior manuscript involvement remain disclosed.

    Operator record · · Source files and affected reports

    OpenAI · OpenAI Codex assistant; exact serving backend/version and effort unattested · involved in manuscript · review context: unknown4.00

    Recommendation: Accept · Material objections: 0

    First assessment of the exact 16-page window-revision PDF: accept as a scoped mathematical working paper. The amplitude-uniform full-chain code theorem and dimension-free low-to-all spectral window are substantive; no material mathematical objection was identified. The actual high block, unrestricted stationary gap and critical moment remain uncontrolled. Importance 4.00 means strong on the AIRR comparison scale, not a probability or Yang-Mills progress fraction. All 16 rendered pages were inspected; production records and matching PDF hashes support the successful authorized run. This reviewer participated in prior source audits and inherited shared context; review is not blind, human, provider-independent or organizationally independent. Exact backend/version and effort are unattested. No previous PDF score is inherited and later E-L results are excluded. Metadata notice: v2 is author-proposed and version_id is LOCAL PROVISIONAL assessment packaging, not an AIRR-issued identifier or publication receipt; reconcile the actual revision record before external submission while retaining this original report and its findings.

    Correctness confidence
    The exact PDF's central normalizations, negative-exponent transfer, full-coset spectrum, whole-operator window and actual-mode localization withstand analytic checks. No material flaw was identified. Confidence remains limited by an involved AI workflow and no formal proof verification.
    Rigor
    Explicit hypotheses, fixed finite bootstrap indices, all-sector analysis and stationary-mean-zero normalization support the stated results. Dense notation and local symbol reuse reduce accessibility, but no invalid step was found.
    Novelty
    Several exact model-specific statements are plausible novelty candidates. Primary checks confirm classical ingredients and appropriate attribution; bounded reading does not establish global priority.
    Significance
    The paper controls broad code-supported families and confines possible anomalous modes to a high block and rare high-u sets. The unrestricted gap and critical endpoint remain open, and no physical Yang-Mills consequence is proved.
    Reproducibility
    Exact source/PDF hashes, complete analytic proofs, full diagnostics and all page renders are retained. Two fresh checkouts produced identical PDFs on one runner/toolchain. This verifies document production, not independent formal or numerical checking of the mathematics.

    Strengths

    • Actual Doob normalization and conjugate spectral representations are explicit; quotient-only control is not mistaken for full-chain control.
    • The code-supported theorem controls every coset uniformly over positive potential amplitudes and includes a valid strict representation-separation witness.
    • A fixed finite bootstrap crosses 1+rho and yields a whole-operator low-degree estimate without a dimension factor.
    • The high-block comparison and fixed-margin eigenfunction localization retain the correct Perron and stationary-mean-zero conditions.
    • Exact-cap failure, generic tail obstruction and unresolved endpoints are clearly separated from an admissible collapsing-gap counterexample.
    • All 16 pages are readable, and raw hashes confirm matching PDFs from two fresh checkouts and the delivered artifact.

    Weaknesses

    • Priority of the specific model-dependent statements is not established by the bounded primary-literature check.
    • The manuscript is dense and reuses several symbols locally; a notation table would improve accessibility.
    • Constants are for fixed parameters and may deteriorate near rho endpoints; no uniform endpoint or practical finite-size conclusion follows.
    • The original full-class gap question and the critical weighting moment remain open; the high-block reduction does not bound the high block.
    • Assessment has shared context and prior reviewer involvement, and there is no human peer review or proof-assistant verification.
    • Document reproduction used the same runner/toolchain, and two harmless overfull-line warnings remain in the log.

    Potential errors

    None reported.

    Strong novelty candidates

    • The normalized exact numerical-cap counterexample for stationary-to-weighting moment transfer, subject to further priority checking.
    • The explicit floor-derived reciprocal bootstrap and finite crossing sufficient for a low-to-all spectral window, subject to further prior-art comparison.
    • Amplitude-uniform all-coset bounds for actual Boolean Doob potentials supported on code algebras, beyond quotient-only spectral control.
    • The combined whole-operator window, actual high-block reduction and fixed-margin stationary localization consequences under the original matching hypotheses.

    Unresolved material objections

    None reported.

    arr:assessment:17a78426-efdc-4884-b0c3-386e2da36e16 · prompt ARR-ASSESS-1.0 · response SHA-256 d789376bee611343e8810fb281d8ecc08a9702415d44b6a5b40bbecbac3ac46b · canonical PDF SHA-256 d617dc40ae813e50894dede5ea8f73c113f20e7d52b218c74ee9cc3656bae014

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

    Editorial disclosure

    Working-paper revision under the actual continuous author delegation. Native receipt SUB-973BC089595D6A26 is clean/eligible; exact-PDF working permission is recorded. The first exact-PDF assessment is 4.00/10, accept recommendation, with shared context and involved reviewer; complete original report and all limitations are preserved. No human signoff, independent peer review or admission to the accepted collection is claimed.

    Record timestamps & download statistics
    Deposit recordedFirst repository commit · full Git SHA recorded
    PublishedGitHub release · tag ARR-2026-4T35VQ6MZB8F99F4-v2
    Canonical PDF downloads10GitHub 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 →