Jesse Grabowski

I think it's [here](https://github.com/pymc-devs/pymc/blob/5db3779b11a1e1d663447a3a78a6f1213b740409/pymc/step_methods/compound.py#L275), due to the use of `.extend`

@ricardoV94 poke on this, maybe we can do a mini-hack to figure out a better solution than what is proposed in this PR?

Closed in favor of #7776

A hidden internal `SparseMvNormal`, in the vein of `PrecisionMvNormal` would be nice. We could rewrite to it when we see the covariance is sparse. One issue I foresee is that...

It would enable prior predictive checks for ICAR models, which would be nice.

Yeah I might have missed some details. To merge the PR, I would want to see that the prior and the no-data MCMC gives the same answer >Is it something...

Jacobian check code: ```py from pymc.distributions.transforms import CholeskyCorr import numpy as np from pytensor.gradient import jacobian import pytensor.tensor as pt rng = np.random.default_rng(12345) # Set up symbolic graph values =...

No I mean the current implementation we have in `main`