Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2026
|
| Online Access: | https://doi.org/10.5281/zenodo.19560120 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866901550127906816 |
|---|---|
| author | Blaum Hough, Jacqueline |
| author_facet | Blaum Hough, Jacqueline |
| contents | <p>ebsbi v1.0.0: Neural simulation-based inference for eclipsing binaries<br> <em>Amortized posterior inference over 16 physical parameters of eclipsing binary star systems using multi-survey light curves, broadband SEDs, and Gaia parallaxes.</em><br> <strong>Highlights</strong></p> <ul> <li>PHOEBE-based forward model generating synthetic light curves and SEDs across 23 broadband filters (GALEX through WISE)</li> <li>Realistic observational noise process: survey-specific cadence/photometric noise, SED uncertainties, and Gaia parallax errors</li> <li>Multi-channel normalizing flow with independent ResNet featurizers for 4 light-curve bands (ASAS-SN g/V, ZTF g/r), a broadband SED, and stellar metadata</li> <li>Trained model checkpoint and example data included for immediate posterior sampling</li> </ul> <p><strong>Contents</strong></p> <ul> <li>src/ebsbi/ — core library (forward model, priors, noise process, engine, shard dataset)</li> <li>scripts/ — data generation (generate_training_data.py), training (train_nbi.py), inference (run_inference.py), shard conversion (convert_npz_to_npy.py)</li> <li>models/best_model.pth — trained checkpoint (Git LFS)</li> <li>examples/ — example training shard (Git LFS)</li> <li>data/ — phase bank, noise bank, SED noise banks, Gaia parallax bank (Git LFS)</li> </ul> <p><strong>Quickstart</strong><br> conda env create -f requirements.yml conda activate ebsbi pip install -e . python scripts/run_inference.py <br> --config docs/base.yml \<br> --checkpoint models/best_model.pth <br> --observation examples/training_shard_000000.npz <br> --index 0 --n-posterior 10000</p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_19560120 |
| institution | Zenodo |
| language | |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | jackieblaum/ebsbi: ebsbi v1.0.0: Neural simulation-based inference for eclipsing binaries Blaum Hough, Jacqueline <p>ebsbi v1.0.0: Neural simulation-based inference for eclipsing binaries<br> <em>Amortized posterior inference over 16 physical parameters of eclipsing binary star systems using multi-survey light curves, broadband SEDs, and Gaia parallaxes.</em><br> <strong>Highlights</strong></p> <ul> <li>PHOEBE-based forward model generating synthetic light curves and SEDs across 23 broadband filters (GALEX through WISE)</li> <li>Realistic observational noise process: survey-specific cadence/photometric noise, SED uncertainties, and Gaia parallax errors</li> <li>Multi-channel normalizing flow with independent ResNet featurizers for 4 light-curve bands (ASAS-SN g/V, ZTF g/r), a broadband SED, and stellar metadata</li> <li>Trained model checkpoint and example data included for immediate posterior sampling</li> </ul> <p><strong>Contents</strong></p> <ul> <li>src/ebsbi/ — core library (forward model, priors, noise process, engine, shard dataset)</li> <li>scripts/ — data generation (generate_training_data.py), training (train_nbi.py), inference (run_inference.py), shard conversion (convert_npz_to_npy.py)</li> <li>models/best_model.pth — trained checkpoint (Git LFS)</li> <li>examples/ — example training shard (Git LFS)</li> <li>data/ — phase bank, noise bank, SED noise banks, Gaia parallax bank (Git LFS)</li> </ul> <p><strong>Quickstart</strong><br> conda env create -f requirements.yml conda activate ebsbi pip install -e . python scripts/run_inference.py <br> --config docs/base.yml \<br> --checkpoint models/best_model.pth <br> --observation examples/training_shard_000000.npz <br> --index 0 --n-posterior 10000</p> |
| title | jackieblaum/ebsbi: ebsbi v1.0.0: Neural simulation-based inference for eclipsing binaries |
| url | https://doi.org/10.5281/zenodo.19560120 |