Metadata-Version: 2.1
Name: seedbank
Version: 0.1.2
Summary: Common infrastructure for initializing random number generators.
Author-email: Michael Ekstrand <michaelekstrand@boisestate.edu>
Requires-Python: >= 3.7
Description-Content-Type: text/markdown
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.7
Classifier: Programming Language :: Python :: 3.8
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Operating System :: OS Independent
Requires-Dist: numpy >=1.17
Requires-Dist: anyconfig
Requires-Dist: flit >=3.2 ; extra == "dev"
Requires-Dist: flake8 ; extra == "dev"
Requires-Dist: ipython ; extra == "dev"
Requires-Dist: sphinx-autobuild ; extra == "dev"
Requires-Dist: invoke ; extra == "dev"
Requires-Dist: conda-lock ; extra == "dev"
Requires-Dist: sphinx >=4 ; extra == "doc"
Requires-Dist: furo ; extra == "doc"
Requires-Dist: numba ; extra == "numba"
Requires-Dist: pytest >= 6 ; extra == "test"
Requires-Dist: pytest-doctestplus ; extra == "test"
Requires-Dist: pytest-cov ; extra == "test"
Requires-Dist: pyyaml ; extra == "test"
Requires-Dist: tomli ; extra == "test"
Requires-Dist: tensorflow >=2,<3 ; extra == "tf"
Requires-Dist: torch ; extra == "torch"
Project-URL: GitHub, https://github.com/lenskit/seedbank
Project-URL: Homepage, https://seedbank.lenksit.org
Provides-Extra: dev
Provides-Extra: doc
Provides-Extra: numba
Provides-Extra: test
Provides-Extra: tf
Provides-Extra: torch

# Python Seed Manager

Python programs, particularly data science applications, often need to
interact with multiple different random number generators.

This package provides a unified interface to seeding them, along with
APIs for deriving additional RNG seeds in a predictable way (using NumPy
1.17's new random infrastructure) and constructing random generators.

## Quick Start

To get started, just use the seedbank.initialize() function to seed all available random number generators:

```python
import seedbank
seedbank.initialize(65000)
```

SeedBank will seed all of the known generators that will be available, including:

- Python standard random
- NumPy legacy random `numpy.random`
- PyTorch (with torch.manual_seed())
- Numba’s NumPy random
- TensorFlow (with tf.random.set_seed())

In addition, it will initialize a root seed for constructing new-style NumPy `Generator` instances.

If SeedBank doesn’t support your RNG yet, please submit a pull request!

## Developing SeedBank

SeedBank uses Flit for managing dependencies.  To set up in a fresh
virtual environment:

    python -m pip install flit
    flit install --pth-file

[conda-lock][] can help you set up a Conda environment (replace `linux-64` with your platform):

    # install conda-lock in base environment
    # alternatively: pip install conda-lock
    conda install -c conda-forge conda-lock
    # create the lock file
    conda-lock -p linux-64 -f pyproject.toml
    # create the environment
    conda env create -n seedbank -f conda-linux-64.lock

