Rémi Louf

Codecov warnings are about tests in `nnet` and `signal`, I think this can be reviewed one last time and merged.

Yes this was raised in #212. We're ready to include `jax`, `jaxlib` and `jaxopt` as dependencies now.

[PyMC](https://github.com/pymc-devs/pymc/blob/be048a495b32a09c49ec8d3e8a95d4f5e4290d2a/.github/workflows/release.yml#L43) checks that the package has been uploaded. We can run `python -c "import blackjax"` afterwards.

The current test fails: ```bash ERROR: Could not find a version that satisfies the requirement blackjax==.8.3 (from versions: 0.1, 0.2, 0.2.1, 0.3.0, 0.4.0, 0.5.0, 0.6.0, 0.7.0, 0.8.0, 0.8.1, 0.8.2, 0.8.3)...

Many great additions in one PR! However I am not a big fan of the warmup being ran inside the kernel; what about users who would like to run Generalized...

> I've also recycled code from `hmc.py` with small modifications to the original > code to make it work. Yes, it looks a lot better, thank you! > I've included...

I think that once you're using `tree_multimap` that's not a huge deal. Let's see when the warmup is fully implemented!

Oh yes sorry I did not expect to find the warmup code in `kernels.py`! I'll need to take another look.

Would you mind organizing the warmup in `adaptation/meads.py` so that the following: ```python def max_eigen(matrix: PyTree): X = jnp.vstack( [leaf.T for leaf in jax.tree_leaves(matrix)] ).T # will only work if...

Almost there! Two things related to the notebooks: 1- We can probably skip `MEADS.ipynb`, it is not adding anything that is not in the two others; 2- The rhat diagnostic...