# Third-party components

All third-party dependencies are the packages provided by the official
BatterySwapAI 2026 competition runtime environment. They are declared in
[`requirements.txt`](requirements.txt); the authoritative pinned versions are
the ones installed in the official competition container
(`huggingface/competitions`, environment at `/app/env`), which the organizers
control and which is identical for all participants. No other third-party
code, pretrained model, or external dataset is used: the submitted model in
`batteryswap_example/planners/best.pickle` is trained from scratch on the
official competition training split only.

| Package | Source | License |
| --- | --- | --- |
| batteryswap_public | Organizer-provided competition helper (Soundsensing), PyPI | Competition-provided; part of the official runtime for all participants |
| pandas | PyPI / pandas-dev | BSD-3-Clause |
| numpy | PyPI / numpy | BSD-3-Clause |
| scipy | PyPI / scipy | BSD-3-Clause |
| scikit-learn | PyPI / scikit-learn | BSD-3-Clause |
| joblib | PyPI / joblib | BSD-3-Clause |
| statsmodels | PyPI / statsmodels | BSD-3-Clause |
| torch | PyPI / pytorch | BSD-3-Clause |
| plotly | PyPI / plotly | MIT |
| polars | PyPI / pola-rs | MIT |
| lifelines | PyPI / CamDavidsonPilon | MIT |
| pydantic-settings | PyPI / pydantic | MIT |
| structlog | PyPI / hynek | MIT / Apache-2.0 (dual) |
| tqdm | PyPI / tqdm | MIT / MPL-2.0 (dual; used under MIT) |
| requests | PyPI / psf | Apache-2.0 |
| pyarrow | PyPI / apache/arrow | Apache-2.0 |
| fastparquet | PyPI / dask | Apache-2.0 |
| huggingface_hub | PyPI / huggingface | Apache-2.0 |
| ortools | PyPI / google | Apache-2.0 |

The competition dataset is copyright Soundsensing 2022-2026, all rights
reserved, licensed for use in the BatterySwapAI 2026 challenge only. It is
not redistributed in this repository (see `.gitignore`).

Participant-authored code in this repository is released under the MIT
License (see [`LICENSE`](LICENSE)).