Skip to content

Installation

SXS is Python scientific software. It has no standalone graphical interface and does not need to be installed system-wide.

To use a prebuilt scientific environment instead of installing Python packages, see the GHCR container guide.

Requirements

  • Python 3.11 or 3.12
  • Git for a repository checkout, or one platform bundle from the v1.3.0 release
  • Enough disk space for the chosen public mission products
  • Network access for MAST and catalog acquisition

Windows received the full workstation research validation. The CI matrix checks the deterministic core and installed wheel on Ubuntu, Windows, and macOS with Python 3.11 and 3.12. This does not imply that the full research/CNN workflow has been scientifically reproduced on all platforms. See the latest CI results.

Choose a dependency profile

File Intended use
requirements-core.txt Production acquisition, preprocessing, BLS, and independent-validation runtime
requirements-test.txt Core runtime plus pytest for deterministic and opt-in network tests
requirements.txt Complete production runtime including RF/CNN training, TensorFlow, and MLflow
requirements-ml.txt Compatibility alias for the complete environment
requirements-docs.txt Documentation website and publication/PDF utilities

Use the complete environment for an end-to-end research reproduction.

From a source checkout, python -m pip install -e ".[full,test]" installs the same complete runtime plus pytest. The extras belong to the 1.4.0.dev0 development metadata; the already-published 1.3.0 wheel supports only its default core installation.

Install from PyPI

For normal CLI use, install the production package in a Python 3.11 or 3.12 virtual environment:

python -m pip install scix-exoplanet-search==1.3.0
sxs --help
sxs demo --output demo

Open demo/report.html. The package includes the core scientific dependencies and default YAML configurations. It does not include mission observations, catalog snapshots, or trained models. Full RF/CNN training requires the complete dependency profile from the matching source release.

Install from Git

git clone https://github.com/Science-Experimental-Technologies/Exoplanet-Search.git
cd Exoplanet-Search
py -3.11 -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install --upgrade pip
python -m pip install -e ".[full,test]"

git clone https://github.com/Science-Experimental-Technologies/Exoplanet-Search.git
cd Exoplanet-Search
python3.11 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -e ".[full,test]"

git clone https://github.com/Science-Experimental-Technologies/Exoplanet-Search.git
cd Exoplanet-Search
python3.11 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -e ".[full,test]"

Install from a release bundle

Download the archive for your platform, extract it, and open PLATFORM_INSTALL.md inside the extracted sxs-1.3.0 directory. All platform archives contain the same source and scientific record; only the installation guide differs.

Verify the archive checksum before use:

Get-FileHash .\sxs-v1.3.0-windows-python.zip -Algorithm SHA256

shasum -a 256 sxs-v1.3.0-*-python.*

Compare the result with SHA256SUMS.txt on the release page.

Verify the environment

python -m pip install -r requirements-test.txt
sxs --version
python -m pip check
python -m pytest -m "not network"
python -m src.cli baseline --config configs/base.yaml --dry-run

On the main branch, run sxs doctor for installation diagnostics. sxs doctor --json produces a machine-readable support record. Add --network to check the public MAST and NASA Exoplanet Archive endpoints without downloading observations.

The command above uses -m "not network" and deselects the opt-in network test. Test counts are revision-specific. Warnings from optional Lightkurve components do not by themselves indicate a failed SXS test.

Optional network test

The MAST smoke test performs a real external query and is deliberately opt-in.

$env:SXS_RUN_NETWORK_TESTS = "1"
python -m pytest -m network tests/test_mast_client_network.py

SXS_RUN_NETWORK_TESTS=1 python -m pytest -m network tests/test_mast_client_network.py

Continue with the quickstart after these checks pass.

Standalone wheel and installed command

The release wheel installs the sxs command and core dependencies without a source checkout. In a Python 3.11/3.12 virtual environment, download the wheel and checksum manifest from the release, verify its checksum, then run:

python -m pip install scix_exoplanet_search-1.3.0-py3-none-any.whl
sxs demo --output demo
sxs baseline --workspace research-a --dry-run

Open demo/report.html. YAML defaults are included and copied into the selected workspace. Observations, catalogs, and trained models are not bundled in the wheel. Full model training additionally requires requirements.txt from the matching source bundle. From a checkout, python -m pip install . provides the same installed command. Version 1.3.0 is also available from PyPI; see the publication record for verification details.