Getting Started
This page provides the shortest path to install, run, and validate the project.
Installation
The project is configured through pyproject.toml with optional extras.
Minimal install
pip install -e .
Standard install
pip install -e ".[standard]"
Development + tests
pip install -e ".[standard,dev]"
Documentation build dependencies
pip install -e ".[docs]"
Machine-learning dependencies
pip install -e ".[ml]"
JAX-compatible alias
pip install -e ".[jax]"
Backend expectations
NumPy is the default and baseline backend across the library.
JAX is part of the recommended standard install and is used in selected modules for accelerator-enabled workflows.
Some APIs dispatch by backend availability; for reproducible studies, keep backend choice fixed per run.
Quick verification
Run core tests:
pytest
Run the documentation import check:
python test_documentation.py
Build Sphinx docs locally:
cd docs
make html
Output is generated at docs/_build/html/index.html.
Notes on reproducibility
Set explicit RNG seeds in workflows that rely on random initialization.
Expect minor floating-point differences between NumPy and JAX backends due to execution and reduction ordering.