Jesse Grabowski

I'm not sure I follow your point about P0 not mattering after the first update. x0 and P0 are taken to be predictions $x_{1|0}$ and $P_{1|0}$ (following Durbin and Koopsman),...

Sure, we can change the order of the names as a short term solution. But I still think they are bad names, before if you set `{"cycle_init":[A0, B0]}` with neither...

Fix for the elipsis bug was merged yesterday. I installed from their main and the tests now pass, which is nice! I guess we will have to wait for them...

Also tests that are failing on CI are passing for me locally, so I don't really know what's going on with that.

Is there any `opt` in `opt_einsum`? I thought using that was value-add over plain numpy.

Yes, I am also looking at this now. It's a jax bug that can be recreated easily: ``` import jax jax.jit(jax.numpy.tri)(3, 3, 0) ``` We can ignore it. Looks like...

Opened an issue here: https://github.com/google/jax/issues/22751 I'll hit the docstrings ASAP if that's all that's holding this up

First pass on docstrings. Working on the doctests revealed two things: 1. Our implementation of `_delta` does not agree with that of JAX: ```python from jax._src.lax.lax import _delta as jax_delta...

I understand it's optional, but it also shouldn't be discarded if available no?

Draft of the JAX overload. Need your input on the `Pad` `OpFromGraph`. I needed some way to hang on to the keyword arguments. It seems like there might be a...