Reproducibility package for "What could an emerging Big Bang Nucleosynthesis discrepancy be hinting at?"

# Reproducibility package for the BBN/vEDE paper

This repository contains the archived chains, profile-likelihood outputs,
plotting scripts, and manuscript source needed to reproduce the figures in

> *What could an emerging Big Bang Nucleosynthesis discrepancy be hinting at?*

The repository is intended for archival on Zenodo and for reproducing the
figures in the arXiv submission.

Zenodo DOI: [10.5281/zenodo.21494819](https://doi.org/10.5281/zenodo.21494819)

## Contents

```text
figures/final/        Exact figure PDFs used in the submitted manuscript.
figures/reproduced/   Created by scripts/reproduce_all_figures.sh.
manuscript/           arXiv source snapshot and arXiv-rendered PDF.
notebooks/            Notebook entry point for regenerating the paper figures.
pyprimat/             Minimal PyPRIMAT code snapshot, plotting scripts, chains, and results.
primat_tools/         Cobaya/PRIMAT wrapper snapshot, YAML files, and profile drivers.
scripts/              Top-level reproduction scripts.
docs/                 Figure-source and run-provenance notes.
```

## Quick start

Create a Python environment and install the plotting/runtime dependencies:

```bash
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
```

Then regenerate all paper figures:

```bash
export PYTHONPATH="$PWD/pyprimat:$PWD/primat_tools:$PYTHONPATH"
bash scripts/reproduce_all_figures.sh
```

Alternatively, run the notebook:

```text
notebooks/reproduce_all_figures.ipynb
```

The regenerated figures will appear in:

```text
figures/reproduced/
```

The submitted figure PDFs are in:

```text
figures/final/
```

Small visual differences can occur if Matplotlib/GetDist versions differ from
the versions used by the authors, but the posterior samples and profile inputs
are archived here.

## Main archived numerical inputs

MCMC chains:

```text
pyprimat/chains/primat032_latest_20260706/
pyprimat/chains/primat032_latest_20260710_EDEprior/
pyprimat/chains/primat032_latest_20260716_lcdmprior/
pyprimat/chains/primat032_latest_20260717_lcdmdesi/
```

Profile likelihoods and derived summary tables:

```text
pyprimat/results/profile_primat032_theoryMC_20260710/
pyprimat/results/profile_noise_reruns_20260712/
pyprimat/results/primat032_latest_20260706/
pyprimat/results/primat032_latest_20260710_EDEprior/
```

Cobaya input YAML files:

```text
primat_tools/yaml/latest_mcmc/
```

See `docs/FIGURE_SOURCES.md` for the figure-by-figure mapping.
See `docs/PROVENANCE.md` for the local source commit identifiers used to
create the code snapshots.

## Notes for full analyses

The scripts in `scripts/reproduce_all_figures.sh` reproduce figures from the
archived chains and profile CSVs. Rerunning the full MCMC and profile analyses
requires an MPI-enabled Cobaya environment and is substantially more expensive.
The corresponding YAML files and cluster/profile driver scripts are included
for provenance.

## Zenodo record

This archive is associated with Zenodo DOI
[10.5281/zenodo.21494819](https://doi.org/10.5281/zenodo.21494819).