Jesse Grabowski

What is the consensus on warnings/messages? It seems like there was at least some move to reduce their number from v3 to v4 (I'm thinking about how a warning is...

Would the numba C implementation of the LAPACK functions (see [here](https://github.com/numba/numba/blob/61ec1fd0f69aeadece218dccf4c39ebc5c7dfbc4/numba/_lapack.c)) serve as a starting point? I guess they are something like "half-wrappers" around the cython functions in `scipy.linalg.cython_lapack` in...

While the direct solver can be written elegantly using only aesara Ops, I still think there is value-add to handing the forward part to scipy. As noted in the scipy...

The relevant lines I see as being not covered are here, inside the `solve_continuous_lyapunov` function: ```python # Compute the Schur decomposition form of a r, u = schur(a, output='real') #...

"Arbitrary" here refers to the criterion for writing an Op wrapper around a given scipy function, but the word was chosen more out of my ignorance of the criterion than...

Thanks for laying it out clearly. I'm clearly not getting it through my head that aesara straddles the line between sympy and numpy. I appreciate your patience as I learn...

Not scared away, just staying quiet to avoid making myself look like a fool (a common occurrence). Some questions: 1. It seems the consensus is using an `OpFromGraph` then "overwriting...

I updated this PR with the following changes: 1. The direct solver for the discrete lyapunov case is now written in aesara, based on the implementation by @brandonwillard presented above....

I think I messed this all up trying to update the branch to match the current aesara `main` and resolve the conflicts, I maybe should delete it and try again?

Thanks for taking the time to clean up my mess. I'll take some time to get my git situation squared away before I move forward with any other contributions.