Luciano Paz

I understand that PSIS-LF/OO is preferable, but there are models where the forecast into the future doesn't give the same number of pointwise logp values. For example, you might want...

By any chance, could you share your model code with us? We've had reports similar to this on Mac M1 chips, but haven't been able to reproduce the error.

From the other reports that we had, it seems related to BLAS, scipy and the cholesky decomposition. We haven’t identified the exact configuration options that trigger the error though, but...

It looks like this problem might be solved once numpy/numpy#21767 is fixed.

@twiecki, the [pypi workflow](https://github.com/pymc-devs/pytensor/blob/main/.github/workflows/pypi.yml#L36-L55) already builds wheels and uploads them to [pypi](https://pypi.org/project/pytensor/#files). Maybe the issue is that you need a special kind of wheel build for pyodide? I don't really...

Did you just upload a universal wheel to PYPI and not to test-pypi? This is bad and we shouldn't do it. [Wheels will always have precedence over source distributions](https://stackoverflow.com/questions/51527148/pip-install-and-platform-specific-wheels). We...

[This](https://pyodide.org/en/0.19.1/development/new-packages.html) seems to have the instructions for building the pyodide package. I've never used it, but they say that they have a [toolchain that could build the js stuff for...

We can't make `pytensor` behave as a universal wheel because it has a cython extension. What we need to do, is to add a build step that actually builds `pytensor`...

In `aesara` this was being tracked in [this milestone](https://github.com/aesara-devs/aesara/milestone/5). The particular `COp`s were being tracked in [this issue](https://github.com/aesara-devs/aesara/issues/423#issuecomment-944926695). There's also the very important fact that we need to use numba's...

I understand @ahartikainen, but I think that there are scenarios where it is very helpful to support `MultiIndex` coordinates. Specially when you can't simply `unstack` the dimension, because some combinations...