Jake Vanderplas

I understand the choice the numpy devs have made. For what it’s worth, I wouldn’t suggest trying to re-open that discussion. They’ve made their opinions pretty clear, and it’s a...

> But do you agree with it? On the whole, yes I do. I understand the reason that you are advocating for method-based access to all ufuncs and reductions: there...

Lint errors might indicate a problem: ``` jax/_src/dlpack.py:130: error: Argument 2 to "dlpack_managed_tensor_to_buffer" has incompatible type "Client"; expected "Device" [arg-type] jax/_src/dlpack.py:130: error: Argument 3 to "dlpack_managed_tensor_to_buffer" has incompatible type "Client...

Internal pytype tests are failing with many variations of this error: ``` File "/jax/_src/third_party/scipy/interpolate.py", line 4, in : Couldn't import pyi for 'jax.numpy' [pyi-error] No xla_client.Device in module jax._src.lib, referenced...

Can you change your commit message to something more informative? Thanks!

Still failing: ``` File "third_party/py/jax/__init__.py", line 163, in : Couldn't import pyi for 'jax.numpy' [pyi-error] Can't find pyi for 'jaxlib.xla_client', referenced from 'jax.numpy' ``` I think the issue is that...

This looks good, and is probably ready to merge more-or-less. However there's a subtlety here that I've been thinking about with @yashk2810 – the issue is that the behavior of...

Thanks for adding this! One overall note: our eventual goal is to remove `jax.experimental.array_api` and just use the `jax.numpy` namespace directly. So as much as possible, we should aim for...

A note on `default_device` (cf. a discussion with @yashk2810). JAX doesn't really have a concept of "default device" in the way that the Array API envisions it. By default, arrays...

Thanks for the report! I think you're corrrect that this is related to #18640. The reason it only affects the hessian and not the jacobian is because the hessian is...