array-api
array-api copied to clipboard
RFC document, tooling and other content related to the array API standard
I would like to move the recent discussion from SciPy https://github.com/scipy/scipy/issues/19068 to here for a more specialized treatment. In the current incarnation of linalg functions, the most important distinction for...
This RFC proposes adding `nextafter` function ## Overview Based on array comparison [data](https://github.com/data-apis/array-api-comparison/blob/main/signatures/mathematical-functions-elementwise/nextafter.md), the API is available in most array libraries. The main exception is MXNet which doesn't implement it....
Should we have a ravel() or flatten() function? Right now you can do it with `reshape`, but it's maybe not obvious. `reshape` also has the advantage in that it can...
Hi folks, a small announcement: Einops just gained support for API-compliant tensors: ```python from einops.array_api import rearrange, reduce, repeat, pack, unpack from numpy.array_api import zeros x = zeros([2, 3, 4,...
The `v2022-12` tag is broken due to the bug fixed in https://github.com/data-apis/array-api/pull/455, which as far as I can tell has not yet been made part of a release or tag....
We should notate that +/- 0 should give the corresponding side of the branch on a branch cut. See the discussion at https://github.com/data-apis/array-api/pull/514#issuecomment-1347566538. For cuts on the real axis this...
Early on when putting the array API standard together, we made a decision that strides must not be exposed in the Python API, because multiple array libraries don't support that...
_This is a continuation of a discussion that started a few weeks ago in gh-597 (Cc @soraros). It is closely related to gh-84 (boolean indexing) and gh-24 (mutability and copy/views)._...
It seems that many libraries that are candidates to implement the Array API namespace already implement the LU factorization (with variations in API and with the notable exception of numpy)....
Today dask came up a few times and also the possibility of a fallback to `__array_function__` for transitioning at least. This must have come up before, but did we ever...