diff --git a/.gitignore b/.gitignore index aaa1075..2c92acd 100644 --- a/.gitignore +++ b/.gitignore @@ -15,3 +15,4 @@ raw/datasets/desi_gaussian_bao_ALL_GCcomb_cov.txt output/papers/everpresent-adjudication/reproducibility-archive.tar.gz output/computations/inhomogeneous-cosmology-codex-verify/bao_official_recheck.json *.pdf +!output/pdf/everpresent-lambda-prd-submission.pdf diff --git a/CITATION.cff b/CITATION.cff index 4d6427a..11089e1 100644 --- a/CITATION.cff +++ b/CITATION.cff @@ -5,6 +5,7 @@ type: software authors: - family-names: Simon given-names: Romain + orcid: "https://orcid.org/0009-0009-0357-8354" version: 1.0.0 date-released: 2026-07-31 doi: 10.5281/zenodo.21721755 @@ -15,6 +16,7 @@ preferred-citation: authors: - family-names: Simon given-names: Romain + orcid: "https://orcid.org/0009-0009-0357-8354" year: 2026 publisher: name: Zenodo diff --git a/output/papers/everpresent-adjudication/main.tex b/output/papers/everpresent-adjudication/main.tex index 0b42d4b..75f5f36 100644 --- a/output/papers/everpresent-adjudication/main.tex +++ b/output/papers/everpresent-adjudication/main.tex @@ -11,8 +11,10 @@ \author{Romain Simon} \affiliation{Independent Researcher} +\email{contact@romainsimon.net} +\homepage{https://orcid.org/0009-0009-0357-8354} -\date{\today} +\date{July 31, 2026} \begin{abstract} Everpresent $\Lambda$, motivated by causal set theory and unimodular gravity, @@ -146,6 +148,20 @@ \section{Model and implementation} one-redshift convention is retained as a robustness test and selects the same six seeds. +Automated assistance was used in the computational workflow. Anthropic +Claude Code (logged model identifiers \texttt{claude-fable-5} and +\texttt{claude-opus-5}) assisted with analysis-code drafting, numerical +execution, literature search, and manuscript drafting.\ OpenAI Codex (logged +model identifier \texttt{gpt-5.6-sol}) independently inspected and +reimplemented load-bearing calculations, identified and corrected errors, +and ran the final stress tests and consistency checks. The human author set +the hypotheses, data scope, pre-registered falsification criteria, and +evidentiary standards. No numerical statement was accepted from model output +alone: load-bearing claims were regenerated by deterministic scripts, checked +against machine-readable artifacts, and where practical reproduced by a +second implementation. Intermediate mistakes and corrections remain in the +public audit trail. + We emphasize one negative control. Scoring realizations with the model distance modulus at the calibrator redshifts instead of the Cepheid distances, while anchoring the reference, produces a spurious @@ -357,10 +373,13 @@ \section{Conclusions} mechanism must be priced against the standard ruler, not only against integrated distances. -\section*{Reproducibility, data and code} +\section*{Data availability and reproducibility} -All numerical inputs are public: the Pantheon$+$SH0ES release, the Moresco -chronometer compilation and covariance recipe (repository commit +The data and code necessary to reproduce the results are publicly available +at Zenodo under +\href{https://doi.org/10.5281/zenodo.21721755}{doi:10.5281/zenodo.21721755}. +All external numerical inputs are public: the Pantheon$+$SH0ES release, the +Moresco chronometer compilation and covariance recipe (repository commit \texttt{8814133}), and the DESI DR2 Gaussian likelihood files. The complete pipeline (both independent implementations, every likelihood construction, the per-realization sound horizons, the load-bearing machine-readable outputs, @@ -368,27 +387,19 @@ \section*{Reproducibility, data and code} audit trail including failed attempts and their corrections) is packaged with SHA-256 checksums of every declared input and load-bearing output, a frozen environment specification, and the exact random-number convention that fixes every -realization, in a reproducibility archive accompanying this paper -(\texttt{REPRODUCIBILITY.md}), archived at Zenodo under -\href{https://doi.org/10.5281/zenodo.21721755}{doi:10.5281/zenodo.21721755}. +realization. Pinned download locations and checksums for the external data are +listed in \texttt{REPRODUCIBILITY.md} in the archive. \section*{Automated-assistance disclosure} -This work was carried out in an AI-augmented workflow. Language-model -assistants (Anthropic Claude, OpenAI Codex) wrote and cross-verified analysis -code under pre-registered falsification criteria, performed literature -searches, and drafted text. The supernova selection was independently -rescanned over all 90{,}000 seeds. The sound horizon and DESI likelihood were -also reimplemented independently; fixed-point ruler values agree within -$0.20\,\mathrm{Mpc}$, while endpoint treatment can shift large tail -$\chi^2$ values by up to about 10 without changing the verdict. Full -chronometer covariance and multi-start joint optimizations were added in the -independent audit. No Boltzmann, recombination, or perturbation solver for -Everpresent $\Lambda$ was used. All scientific claims were gated on -inspectable numerical artifacts; the human author directed the research, -set the evidentiary standards, and is responsible for the content. A full -audit trail, including documented corrections and retractions of -intermediate claims, is preserved in the repository. +The substantive use of language-model tools in the scientific workflow is +described in Sec.~\ref{sec:model}, including the model identifiers, the tasks +performed, and the verification procedure. The tools also assisted with +language editing and document preparation. They are not authors. The human +author reviewed the manuscript, directed the research, and accepts full +responsibility for the code, analysis, citations, and conclusions. No +Boltzmann, recombination, or perturbation solver for Everpresent $\Lambda$ +was used. \begin{thebibliography}{99} diff --git a/output/papers/everpresent-adjudication/prd-submission/AI-USE-DISCLOSURE.md b/output/papers/everpresent-adjudication/prd-submission/AI-USE-DISCLOSURE.md new file mode 100644 index 0000000..b89071a --- /dev/null +++ b/output/papers/everpresent-adjudication/prd-submission/AI-USE-DISCLOSURE.md @@ -0,0 +1,56 @@ +# Automated-assistance record for APS disclosure + +This record supports, but does not replace, the disclosure printed in the +manuscript. + +## Tools and recorded versions + +The archived local session logs associated with the research record the +following model identifiers: + +- Anthropic Claude Code: `claude-fable-5` and `claude-opus-5` +- OpenAI Codex: `gpt-5.6-sol` + +These strings are reported exactly as logged. No unrecorded commercial model +name or version is inferred from them. + +## Substantive uses + +- drafting and revising analysis code; +- executing numerical scans and optimization checks; +- building an independent implementation of load-bearing calculations; +- literature search and source comparison; +- detecting errors, proposing corrections, and running stress tests; +- drafting and editing manuscript text and submission materials. + +## Human direction and verification + +Romain Simon specified the research question, data scope, falsification +criteria, stopping rules, and publication claims. Numerical claims were tied +to machine-readable artifacts and deterministic commands. The complete +90,000-seed selection was independently rescanned; the sound-horizon and DESI +likelihood calculations were independently reimplemented; fixed-point ruler +values were compared between implementations; full chronometer covariance +and multistart optimization were checked; and the final PDF and archive were +audited. Intermediate errors, failed routes, corrections, and retractions are +preserved in the public research record. + +No language model is an author. The human author remains responsible for the +code, data choices, citations, analysis, and conclusions. + +## Known limitations + +- Language models can hallucinate citations, code behavior, or physical + interpretations; source checks and executable artifacts reduce but do not + eliminate that risk. +- The model identifiers above are service log identifiers, not claims about + undocumented provider internals. +- No AI system supplies peer review, human endorsement, or proof of physical + correctness. +- No Boltzmann, recombination, or perturbation solver for Everpresent Lambda + was used. + +## Policy reference + +Prepared against the APS guidance on the +[appropriate use of AI tools](https://journals.aps.org/authors/appropriate-use-ai-tools). diff --git a/output/papers/everpresent-adjudication/prd-submission/AUTHOR-ACTION-REQUIRED.md b/output/papers/everpresent-adjudication/prd-submission/AUTHOR-ACTION-REQUIRED.md new file mode 100644 index 0000000..9385bdb --- /dev/null +++ b/output/papers/everpresent-adjudication/prd-submission/AUTHOR-ACTION-REQUIRED.md @@ -0,0 +1,23 @@ +# Human actions required before APS submission + +The scientific and reproducibility materials are prepared. The following +items cannot be completed safely by an automated assistant. + +1. Sign in to the APS submission server and verify the ORCID link + `0009-0009-0357-8354` when APS sends the verification request. +2. Read the frozen manuscript PDF and explicitly approve it as the human + author. Automated review cannot replace this gate. +3. Confirm the exact logged AI identifiers and the description of their roles + in `AI-USE-DISCLOSURE.md` and in the manuscript. +4. Confirm the portal declarations for competing interests, funding, and the + absence of concurrent journal review. +5. Decide whether to name suggested or excluded referees. None are proposed in + the current cover letter. +6. Choose the publication-access option and review any resulting charges in + the APS portal. +7. Upload the frozen PDF and `prd-submission-source.tar.gz`, paste the Data + Availability Statement, attach the cover letter, preview APS's generated + PDF, and submit only after checking every field. + +No arXiv identifier is required. The public Zenodo DOI can be supplied as the +preprint and reproducibility record. diff --git a/output/papers/everpresent-adjudication/prd-submission/README.md b/output/papers/everpresent-adjudication/prd-submission/README.md new file mode 100644 index 0000000..7cf3de7 --- /dev/null +++ b/output/papers/everpresent-adjudication/prd-submission/README.md @@ -0,0 +1,32 @@ +# Physical Review D submission kit + +Prepared on 2026-07-31 for an initial submission to *Physical Review D*. +Nothing in this directory records or implies that a submission has occurred. + +## Intended submission + +- Journal: *Physical Review D* +- Article type: Research Article +- Subject area: D15 — Cosmology +- Manuscript source: `../main.tex` +- Reproducibility record: +- Public repository: + +## Files + +- `cover-letter.md`: copy-ready cover letter. +- `submission-metadata.md`: portal fields and copy-ready statements. +- `AI-USE-DISCLOSURE.md`: detailed record supporting the manuscript disclosure. +- `AUTHOR-ACTION-REQUIRED.md`: the remaining human confirmations and portal actions. +- `prd-submission-source.tar.gz`: uploadable LaTeX source archive. +- `SHA256SUMS`: checksums for the frozen manuscript PDF and source archive. + +The frozen manuscript PDF is at +`../../../pdf/everpresent-lambda-prd-submission.pdf`. + +## Build + +The PDF was built with Tectonic from `../main.tex`. The manuscript is a +self-contained REVTeX file with an embedded bibliography and no external +figures. The remaining portal and human-review steps are listed in +`AUTHOR-ACTION-REQUIRED.md`. diff --git a/output/papers/everpresent-adjudication/prd-submission/SHA256SUMS b/output/papers/everpresent-adjudication/prd-submission/SHA256SUMS new file mode 100644 index 0000000..cc39eb9 --- /dev/null +++ b/output/papers/everpresent-adjudication/prd-submission/SHA256SUMS @@ -0,0 +1,2 @@ +358c93f5c02c02840b61933182a90b16964bcf3e0eb2cc35b29acae8aba4e390 output/pdf/everpresent-lambda-prd-submission.pdf +fcf2a45926c59e1e767d59410a19f71dabc6822cbd4f5a1434aa9e1cca1322ab output/papers/everpresent-adjudication/prd-submission/prd-submission-source.tar.gz diff --git a/output/papers/everpresent-adjudication/prd-submission/cover-letter.md b/output/papers/everpresent-adjudication/prd-submission/cover-letter.md new file mode 100644 index 0000000..fdb660e --- /dev/null +++ b/output/papers/everpresent-adjudication/prd-submission/cover-letter.md @@ -0,0 +1,46 @@ +Dear Editors of Physical Review D, + +Please consider the manuscript “Supernova success and sound-horizon failure +of Everpresent Lambda: an independent test of stochastic dark energy” as a +Research Article in cosmology (D15). + +The manuscript presents an independent, distribution-level reproduction of +the rare-tail supernova result reported for Everpresent Lambda Model 1. Two +independently written implementations scan all 90,000 published seed labels +and recover the same six realizations that outperform a like-for-like flat +Lambda-CDM reference under both tested Pantheon+ redshift conventions. The +paper then subjects those selected realizations to cosmic-chronometer and DESI +DR2 baryon-acoustic-oscillation background tests using a sound horizon +computed per realization. All six are disfavored, with joint penalties from ++128 to +1974 under the stated drag calibration. + +The intended contribution is deliberately bounded. This is not presented as +a universal no-go theorem for Model 1, nor as a substitute for a +self-consistent recombination and perturbation calculation. The unresolved +six-versus-16 selection-rate difference and the source paper's unreproduced +reference chi-square are stated explicitly. The manuscript instead provides +a reproducible independent test of the six recovered supernova winners and +identifies their sound-horizon deficit. + +The code, pinned data references, frozen environment, checksums, +machine-readable outputs, negative results, and audit trail are openly +archived at . The manuscript includes +a Data Availability Statement and a substantive automated-assistance +disclosure identifying the logged models, their roles, and the human and +computational verification procedures. + +This manuscript has not been submitted to another journal and is not under +review elsewhere. Its public Zenodo record is a preprint and reproducibility +deposit, not a prior journal publication. + +No recommended or excluded referees are proposed at this stage. + +Sincerely, + +Romain Simon + +Independent Researcher + +ORCID: + +Contact email: contact@romainsimon.net diff --git a/output/papers/everpresent-adjudication/prd-submission/prd-submission-source.tar.gz b/output/papers/everpresent-adjudication/prd-submission/prd-submission-source.tar.gz new file mode 100644 index 0000000..9a339f7 Binary files /dev/null and b/output/papers/everpresent-adjudication/prd-submission/prd-submission-source.tar.gz differ diff --git a/output/papers/everpresent-adjudication/prd-submission/submission-metadata.md b/output/papers/everpresent-adjudication/prd-submission/submission-metadata.md new file mode 100644 index 0000000..005fa89 --- /dev/null +++ b/output/papers/everpresent-adjudication/prd-submission/submission-metadata.md @@ -0,0 +1,68 @@ +# PRD submission metadata + +## Destination + +- Journal: Physical Review D +- Article type: Research Article +- Subject area: D15 — Cosmology +- Suggested handling: standard Research Article, not Letter treatment + +## Title + +Supernova success and sound-horizon failure of Everpresent Lambda: an +independent test of stochastic dark energy + +## Author + +- Name: Romain Simon +- Affiliation: Independent Researcher +- ORCID: 0009-0009-0357-8354 +- Corresponding author: Romain Simon +- Contact email: contact@romainsimon.net + +## Abstract + +Use the abstract from `../main.tex` verbatim. It is also embedded in the +frozen PDF; maintaining a single source avoids portal/manuscript drift. + +## Public records + +- Reproducibility DOI: +- Repository: +- arXiv identifier: none + +An arXiv identifier is not required for an APS submission. The Zenodo record +is the public preprint and reproducibility record. + +## Data Availability Statement + +The data and code necessary to reproduce the results are publicly available +at Zenodo under . The external +Pantheon+SH0ES, Moresco cosmic-chronometer, and DESI DR2 inputs are publicly +available from the repositories cited in the manuscript; pinned download +locations and SHA-256 checksums are provided in `REPRODUCIBILITY.md` in the +archive. + +## Automated-assistance disclosure + +Automated assistance was used substantively. Anthropic Claude Code (logged +model identifiers `claude-fable-5` and `claude-opus-5`) assisted with +analysis-code drafting, numerical execution, literature search, and manuscript +drafting. OpenAI Codex (logged model identifier `gpt-5.6-sol`) independently +inspected and reimplemented load-bearing calculations, identified and +corrected errors, and ran final stress tests and consistency checks. The human +author set the hypotheses, data scope, preregistered falsification criteria, +and evidentiary standards. Load-bearing claims were regenerated by +deterministic scripts, checked against machine-readable artifacts, and where +practical reproduced by a second implementation. The tools are not authors; +the human author reviewed the manuscript and accepts responsibility for the +code, analysis, citations, and conclusions. + +## Declarations to confirm in the portal + +- Competing interests: **AUTHOR ACTION REQUIRED** +- Funding sources and grant identifiers: **AUTHOR ACTION REQUIRED** +- Concurrent or previous journal submission: none known; author must confirm +- Suggested referees: none proposed +- Excluded referees: none proposed +- Open-access option: author decision in the APS portal diff --git a/output/pdf/everpresent-lambda-prd-submission.pdf b/output/pdf/everpresent-lambda-prd-submission.pdf new file mode 100644 index 0000000..d99a2cc Binary files /dev/null and b/output/pdf/everpresent-lambda-prd-submission.pdf differ diff --git a/output/review-prd-everpresent-adjudication-submission.md b/output/review-prd-everpresent-adjudication-submission.md new file mode 100644 index 0000000..d3e9a3c --- /dev/null +++ b/output/review-prd-everpresent-adjudication-submission.md @@ -0,0 +1,84 @@ +# Physical Review D submission-gate review + +**Manuscript:** *Supernova success and sound-horizon failure of Everpresent Lambda: an independent test of stochastic dark energy* + +**Review date:** 2026-07-31 + +**Intended section:** D15 — Cosmology + +**Article type:** Research Article + +**Review status:** AI-assisted proxy review; not an APS or human referee report + +## Assessment + +| Dimension | Rating | +|---|---| +| Quality of presentation | Good | +| Quality of research | Good within its explicitly conditional scope | + +## Summary and significance + +The manuscript independently reproduces the rare-tail existence and parameter +population of an Everpresent Lambda Model 1 supernova selection, recovering +six winners in two independently coded 90,000-seed scans and under two +Pantheon+ redshift conventions. It then tests those six selected histories +against chronometer and DESI DR2 background observables with a +per-realization sound horizon. The result is a strong conditional exclusion +of those six histories under the declared drag calibration. + +The contribution is specialist rather than field-defining, but it is original +enough for a PRD Research Article: it combines a distribution-level independent +reproduction with a new, reproducible low-redshift adjudication of the selected +tail. The manuscript correctly distinguishes this from earlier BAO/CMB work +and retains “to our knowledge” for its priority claim. + +## Technical and physical review + +- The official Pantheon+ convention and the one-redshift robustness convention + recover the same six seeds. +- Two implementations agree on the selected set and independently implement + the sound-horizon and DESI likelihood calculations. +- The fixed-point ruler control and full-covariance chronometer checks are + documented. +- The six-versus-16 rate discrepancy and the unreproduced source reference + chi-square remain explicit rather than being explained post hoc. +- The background drag prescription is not represented as a recombination or + perturbation calculation. The title, abstract, mechanism discussion, and + conclusion bound the result to the six recovered winners and the stated + calibration. +- The public Zenodo DOI now contains code, environment, checksums, data pins, + outputs, and the correction trail. +- The manuscript now contains a complete Data Availability Statement and an + APS-style substantive AI disclosure with logged model identifiers, roles, + human direction, and verification procedures. + +## Remaining limitations + +The absence of a self-consistent recombination and perturbation treatment is +the main physical limitation. It prevents promotion of the result to a +universal no-go theorem, but it does not invalidate the manuscript's narrower +background-level claim. A joint search over all Model 1 histories also remains +future work. These limitations are prominent and quantitatively stress-tested. + +## Recommendation: ACCEPT + +The five scientific and reproducibility changes required by the previous PRD +gate have been addressed: bounded scope, explicit early-universe limitation, +permanent archive, unresolved discrepancies retained, and novelty clarified +against prior work. No critical mathematical or physical inconsistency is +identified within the paper's stated claim. Submission as a standard PRD +Research Article is justified. + +This recommendation does not cross the human gate. Before submission, the +author must read and approve the final PDF, verify the ORCID association, and +confirm funding, competing-interest, and AI-use +declarations in the APS portal. + +## Tool and computational resource disclosure + +OpenAI Codex applied the local PRD checklist to the revised manuscript, +reproducibility record, prior review findings, and APS submission requirements. +The underlying numerical work used the tools and checks disclosed in the +manuscript and public archive. This is an automated skeptical review, not an +impersonation of a referee and not evidence of acceptance by Physical Review D.