Stephan Hoyer

+1 I think support for ellipsis is noteworthy. In fact, I recently saw someone note support for ellipsis as a missing feature in einops! It's true that it's probably a...

Yes, this is possible -- but should it be done? What are the actual use cases for this "all in one" function? I really like descriptive function names like `rearrange`,...

I should have checked! I see this already exists as `einops.parse_shape` Perhaps the multi-argument version would be a nice addition? This could be spelled in either of two forms: 1....

JAX does not have symbolic shapes. You can just raise a normal exception. On Sat, Aug 29, 2020 at 11:02 PM Alex Rogozhnikov wrote: > > > Got back to...

Oops, not sure how I missed the license! Yes, that should already cover it. You raise a good point about making further improvements. So perhaps it is indeed worth waiting...

We’ve started over with stubs in the https://github.com/numpy/numpy-stubs repo

Agreed. `_ArrayLike` is a useful class for annotation function signatures, but I don't think it makes sense for methods.

We're often a sometimes a little sloppy on terminology, but it can be useful to distinguish between "array likes" and "duck arrays" (I think that's the source of our confusion...

JAX solves this problem in two ways: - The XLA compiler can convert functionally pure accumulation at a single index into in-place updates (assuming the original array values are not...

JAX's [`lax.scan`](https://jax.readthedocs.io/en/latest/_autosummary/jax.lax.scan.html#jax.lax.scan) is a generalization of `mapfoldl` that also has optional direct outputs. Here's the Python pseudocode version: ```python def scan(f, init, xs): carry = init ys = [] for...