orng provides a thin facade over several Array API–compatible random number generators. It mirrors the subset of the numpy.random.Generator API:

• random
• uniform
• normal
• choice
• gamma

letting you pick the underlying backend at runtime. The following backends are currently supported:

• numpy
• torch
• cupy
• jax

The core package only depends on the standard Python library:

pip install orng

Backends are optional extras that you can install on demand:

pip install "orng[numpy]"   # NumPy RNG support
pip install "orng[torch]"   # PyTorch RNG support
pip install "orng[cupy]"    # CuPy RNG support
pip install "orng[jax]"     # JAX RNG support

You can also combine extras, e.g. pip install "orng[numpy,torch]".

from orng import ArrayRNG

rng = ArrayRNG(backend="numpy", seed=42)
samples = rng.normal(loc=0.0, scale=1.0, size=5)
uniform = rng.uniform(low=-1.0, high=1.0, size=(2, 2))

The backend module is imported lazily. If the requested library is missing, ArrayRNG will raise an informative ImportError that points to the matching extra.

When you pass the optional generator argument to ArrayRNG, the expected object depends on the backend:

This lets you wrap an existing RNG/key instead of seeding a new one.

orng/
├── src/orng/
│   ├── __init__.py      # package exports
│   ├── _utils.py        # shared helpers (internal)
│   ├── orng.py          # ArrayRNG facade
│   └── backends/        # backend-specific implementations
└── README.md

Each backend class lives in its own module under orng/backends/, keeping the core facade compact and making optional dependencies easy to manage.