Published August 18, 2025 | Version v0.1.0

sbi_uhecr_radio_recon

Authors/Creators

Description

Summary

This release delivers a compact, reproducible codebase for UHECR radio direction reconstruction using simulation-based inference with normalizing-flow posteriors. It adds a temperature scaling wrapper for calibrated posteriors, training/diagnostics utilities, and runnable notebooks with mock data.

What's included?

src/sbi_uhecr_radio_recon/{temperature_calibration.py, diagnostics.py, angular_resolution.py, generate_mock_training_data.py, read_data.py, runner_lampe_uhecr_radio.py, utils.py, cli.py} plus example notebooks.

Install

pip install -e .
# optional sanity/help
python -m sbi_uhecr_radio_recon.cli --help

Minimal use

from sbi_uhecr_radio_recon.temperature_calibration import tune_temperature
# posterior_ensemble, val_loader prepared as in the notebooks
posterior_tuned, T_star = tune_temperature(posterior_ensemble, val_loader, mass=0.68)

Reproducibility See pyproject.toml for dependencies. Notebooks are provided; set seeds externally for deterministic sampling. Large artifacts (if any) are attached to releases instead of being committed.

License MIT.

Cite Please cite the companion article (arXiv:2508.xxxxx / PRD DOI when available). A CITATION.cff is included.

Notes

If you use this software, please cite it as below.

Files

oscar-macias/sbi_uhecr_radio_recon-v0.1.0.zip

Files (11.4 MB)

Name Size
md5:1219f1033165ce2c583f29a978b493fd
11.4 MB Preview Download

Additional details

Related works