You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Many of the the libraries such as numpy, torch, jax etc. have support for random number generation. Typically in their own namespace, or exposed via a Generator API. Are there any plans to extend the API standard for random number generation using arrays? I am aware this poses a lot of challenges in terms of reproducibility across multiple implementations, but maybe this has been discussed in the past already and has been discarded for not being in scope?
The text was updated successfully, but these errors were encountered:
See #431 for some interesting convo on the subject. We concluded with
Given lack of ecosystem agreement, we are not likely to forge a path forward for standardizing a PRNG API in the array API specification at this time. [...] We can reopen/revisit if and when we have greater community consensus.
I'll close this issue, but please do feel free to continue the conversation on #431.
I'd add that we will probably try to build an implementation soon for SciPy's usage (since we need it in quite a few places) that exposes a uniform API and works with NumPy, PyTorch, CuPy and JAX at least.
Many of the the libraries such as
numpy
,torch
,jax
etc. have support for random number generation. Typically in their own namespace, or exposed via aGenerator
API. Are there any plans to extend the API standard for random number generation using arrays? I am aware this poses a lot of challenges in terms of reproducibility across multiple implementations, but maybe this has been discussed in the past already and has been discarded for not being in scope?The text was updated successfully, but these errors were encountered: