pymc icon indicating copy to clipboard operation
pymc copied to clipboard

Review potentially outdated docs

Open OriolAbril opened this issue 2 years ago • 6 comments

There are some files closely related to v3 internals that should be reviewed to choose if we want to update and keep them or get rid of them (and maybe write something similar but not the same from scratch).

Those files are:

  • [ ] https://github.com/pymc-devs/pymc/blob/main/docs/source/guides/Gaussian_Processes.rst by @canyon289
  • [x] https://github.com/pymc-devs/pymc/blob/main/docs/source/guides/Probability_Distributions.rst being refactored by @juanitorduz
  • [x] https://github.com/pymc-devs/pymc/blob/main/docs/source/PyMC_and_Aesara.rst Claimed by @ricardoV94
  • [x] https://github.com/pymc-devs/pymc/blob/main/docs/source/Advanced_usage_of_Aesara_in_PyMC.rst Todo: Remove by @AlexAndorra

OriolAbril avatar Mar 03 '22 16:03 OriolAbril

Hi! @ricardoV94 has some very useful notes which I could build on top of. I will open a PR soon.

juanitorduz avatar Mar 30 '22 07:03 juanitorduz

Advanced_usage_of_Aesara_in_PyMC.rst

I think we can remove it.

  1. It mentions shared variables, which are wrapped in the more user friendly pm.MutableData,
  2. It shows the creation of an Op, which we have some examples of in pymc-examples. The one advantage of that Op example is that unlike those in pymc-examples it is an intermediate Op and not just a likelihood term, which might bridge a gap in the documentation. In any case a separate issue for a doc or pymc-examples should be made for that

Tracking issue #5729

PyMC_And_Aesara.rst

This one is not in a terrible state, but should be expanded beyond describing how variables are used for the logp graph, to also explain how RandomVariables are used for the random graph, and how PyMC goes back and forth between those. That's probably where @juanitorduz would be contributing

Shoud be replaced by #5721

Probability_Distributions.rst

We can replace this entirely by the new dev guide on how to develop new distributions and by updating this pymc-example: https://github.com/pymc-devs/pymc-examples/pull/185

Gaussian_Processes.rst

Not my expertise, perhaps @bwengals can chime in.

Tracking issue #5720

ricardoV94 avatar Apr 20 '22 06:04 ricardoV94

I do have a working branch and notebook where I am trying to add comments and understand @ricardoV94 nice intro notebook. Nevertheless, I still have a long way to fully understand. Specifically, starting from section Graph manipulation 102 things become less clear for me. I would live to work on this one but to be honest it would take much more than initially planned. Hence, if there is some desire to have it soon (in view of the next release) maybe some else could jump in as well :)

juanitorduz avatar Apr 20 '22 13:04 juanitorduz

@juanitorduz no need to include any graph manipulation parts. You can also just open a PR with what you have and iterate from there :)

ricardoV94 avatar Apr 20 '22 13:04 ricardoV94

Gaussian_Processes.rst will be updated by a combination of myself and @bwengals over the coming months. if it helps with tracking I can open an issue and assign it to myself.

For now I can add a disclaimer at the top stating its V3 specific and slated for an update

canyon289 avatar Apr 20 '22 13:04 canyon289

Done: https://github.com/pymc-devs/pymc/pull/5826

AlexAndorra avatar May 30 '22 17:05 AlexAndorra

Anyone objects removing https://github.com/pymc-devs/pymc/blob/main/docs/source/PyMC_and_Aesara.rst now that we have https://www.pymc.io/projects/docs/en/stable/learn/core_notebooks/pymc_aesara.html?

ricardoV94 avatar Nov 18 '22 08:11 ricardoV94

No, sounds good

OriolAbril avatar Nov 18 '22 09:11 OriolAbril