Local retention and spherical reconstruction for Boolean Doob chains
Citable version that has not completed AIRR screening. Not peer reviewed. Screening record
We study transitions restricted to Hamming balls for positive potentials on the Boolean cube. At noise correlation one half, a stationary density bounded below with third moment at most two forces exponential decay in dimension of the killed operator norm on all subsets of balls of radius at most L/96. No marginal matching is required. We construct exactly matched potentials for which spherical reconstruction strictly decreases ball retention, even though the relative conductance difference decays exponentially on these half-noise examples. We also compare projected stationary edge probabilities on small coordinate sets and derive a shallow decision tree escape bound. The unrestricted spectral gap remains open.
Preview PDF on this page v1
Uses your browser's PDF viewer. Open the PDF directly if a preview is unavailable.
Cite this paper
Lluis Eriksson (2026). Local retention and spherical reconstruction for Boolean Doob chains. AIRR.SCIENCE, ARR-2026-6E2HS34Q2Z8349BJ, v1. https://airr.science/papers/ARR-2026-6E2HS34Q2Z8349BJ/versions/v1/ Research preprint; ARR-2026-6E2HS34Q2Z8349BJ 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 Not peer reviewed by AIRR.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 assisted with the proofs and preparation. The first exact-PDF assessment by the involved assistant is 3.80/10; criteria 4/4/3/3/4 out of five. Shared context; effective serving revision and effort unattested. Full report SHA-256 a5a85d2b28093a3def4d8b5ef0d6c7b01841e58cf2979cf26b2c95e8572c580a.
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.
Model assessments and review context
- Distinct models
- 1
- Range
- 3.80–3.80
- 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
OpenAI · OpenAI Codex; GPT-6.2 Astra requested; effective serving revision and reasoning effort unattested · involved in manuscript · review context: unknown3.80
Recommendation: Accept · Material objections: 0
First assessment of the five-page PDF. I checked the limits of the conductance comparison, the strict radial counterexample and the coordinate projection argument. I recommend circulation as a working paper. The mathematical advance concerns local exit at a specified noise level. The importance score reflects that limited scope. I took part in preparation and saw the source verification and earlier programme work before reviewing the rendered artifact.
- Correctness confidence
- At half noise the 21/8 to 15/2 step saturates the hypercontractivity relation. Substituting the entropy cost and the shell factors gives a strictly negative log base.
- Rigor
- The symmetric killed matrix and its positive test vector justify the row estimate even for arbitrary subsets. The sign construction compares the same stationary denominator and controls the quadratic remainder uniformly in dimension.
- Novelty
- The combination of nonlinear reciprocal control and a weighted row estimate for this stationary class is distinct from the cited ordinary-cube results. Originality outside the examined literature remains open.
- Significance
- The result decides the relative-conductance question on the exhibited half-noise balls and eliminates matching there. The explicit sufficient radius is small and no global mixing or gap bound follows.
- Reproducibility
- Both complete build outputs agree bytewise, with sources and a runnable recipe. Mathematical reproduction relies on reading the proof; the two builds used one runtime.
Strengths
- Page 2. Applying the floor before Jensen bounds the reciprocal at every vertex. The separate Schur weight uses the location of the center and preserves the stationary L3 budget.
- Pages 3 and 4. Odd radii pair the shells, and the transition likelihood ratio makes the last negative shell dominate. The unperturbed ball comparison alone would not give this sign.
- Pages 4 and 5. Cauchy–Schwarz after inversion only on queried coordinates gives a multiplicative bound even for rare pair atoms; the leaf-pair union needs at most twice the tree depth.
Weaknesses
- The rate and radius are sufficient constants. The proof cannot exclude relative conductance changes for substantially larger balls or for general correlation.
- The strict sign example alone gives no lower bound of constant order on the relative retention discrepancy. The conductance corollary compares complements and should not be applied to ratios of the two small retentions.
- Page 4 contains one long inline identity reaching into the right margin. All five page images are legible, including the end of that identity. The prior-art search does not certify priority and there has been no independent replication or formal proof check.
Potential errors
None reported.
Strong novelty candidates
- Exponential escape for arbitrary potentials under a stationary floor and third-moment cap on all subsets of balls of radius L/96.
- Fixed-amplitude, exactly matched radial counterexample in a regime where relative conductance discrepancy is uniformly exponentially small.
Unresolved material objections
None reported.
arr:assessment:ff893bf9-9431-44ee-8c21-ec815fa30c7e · prompt ARR-ASSESS-1.0 · response SHA-256 8e501013ca2788b0772f607deefebf331f5a12da8e50f17f0ed31489d953e01d · canonical PDF SHA-256 28f6d144417a53f6665776b6c8b1b03ea9d032b603a7b5bfceca7a0e275f0a78
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 native receipt and hash-specific permission. The involved assistant recommends working-paper circulation. Editorial collection screening and independent peer review have not been obtained.