Reorganize Python, Containers, (future) Julia, R, pages and "Data Science" categories

[Kyle Gerard Felker]

For Python users coming to our guides as a starting point, it can be a bit confusing given the current organization which depends on the machine:

• Polaris > Data Science > Python
• Theta > Data Science > Conda (also "Data Science Modules")
• ThetaGPU > ???? no Python or Conda page under Data Science. Closest would be > Data Science > Running PyTorch/TensorFlow with Conda

Obviously not all Python or Julia users are running Data Science applications, but many are. I am not opposed to leaving the Python pages under Data Science, but we should at least be consistent and come up with a uniform organization for future machines.

I like how NERSC has a top-level Development category that breaks down into Languages then Python, Julia, R, IDL. https://docs.nersc.gov/development/languages/python/using-python-perlmutter/

Decisions to make:

• Move Containers pages out of Data Science subcategories
• Name and organize all machine-specific pages in a uniform fashion, e.g. "Python and Conda"?
• Make new Languages category that includes Python, Julia, R. Collect all machine-dependent instructions in one location, or separate pages under each machine?

[Venkat Vishwanath]

We should make Theta and ThetaGPU similar to Polaris. At the very least, they need to be consistent.

We can call out Python and Julia in Programming Models. However, I would still like these pages to be included/linked off datascience. Same for containers. We can have one page and link them in couple places. In my opinion, this will make it easier for most datascience users and they can find information in one place.