Metadata-Version: 2.1
Name: arraylib
Version: 0.0.0
Summary: Library for array computing
Author-email: Nicolas Tessore <n.tessore@ucl.ac.uk>
Requires-Python: >=3.6
Description-Content-Type: text/markdown
Classifier: License :: OSI Approved :: MIT License
Requires-Dist: sphinx ; extra == "docs"
Requires-Dist: furo ; extra == "docs"
Requires-Dist: pytest ; extra == "test"
Requires-Dist: numpy ; extra == "test"
Project-URL: Documentation, https://arraylib.readthedocs.io
Project-URL: Home, https://github.com/ntessore/arraylib
Project-URL: Issues, https://github.com/ntessore/arraylib/issues
Provides-Extra: docs
Provides-Extra: test

*arraylib* — Library for array computing in Python
==================================================

The *arraylib* package aims to provide a standardised interface to array
implementations such as *NumPy*, *JAX*, *PyTorch*, etc.

The standard array functions are modelled on the *Array API*, and should be
compatible with it.

The goal of the *arraylib* package is to extend the standard with many of the
special functions which are necessary for scientific computing.

Usage
-----

Install the *arraylib* package and an array implementation, for example
*NumPy*.  Load both, then obtain an *arraylib* namespace by binding
the `numpy` array namespace:

```py
>>> import arraylib
>>> import numpy as np
>>> xp = arraylib.bind(np)
>>> xp.log
<ufunc 'log'>
>>> xp.log is np.log
True
>>> xp.sinpi(2.5)
1.0
```

