Array Libraries InteroperabilityΒΆ

This is a small sample tutorial to give you a feel for how the future of array consumer and provider libraries may look after the adoption of Array API.

Note

No more fiddling around with docs when switching an implementation.

Simply change the import and switch your array/tensor implementation library:

import torch as xp
# or
import numpy.array_api as xp


a = xp.arange(3)
b = xp.ones(3)

c = a + b

Check out the Array API Prototype Demo page bundled with this book to see the real use case with a library like SciPy.