SciPy 2023 tracking issue

[dcherian]

cc @scottyhq @lsetiawan @JessicaS11 @TomNicholas @andersy005 @negin513

Any one is welcome to contribute here, please reach out!

Logistics

As reminder, last year we created a outline for attendees (max 90) to follow. The outline links to a canonical copy of the material that gets updated every year, and serves as a syllabus for a future learner who did not necessarily attend the event.

I propose we do the same this year, and use the opportunity to develop some new canonical notebooks.

Here's our proposal outline converted to a todo list. I've added links to material that exists already. For bulllet items without a llink we need to develop new material (in most cases adapt the narrative documentation to tutorial format). Clearly we have a bunch of work to do :)

Outline

• Introduction/Logistics (@lsetiawan, @scottyhq)
  • Nebari binder for SciPy tutorial? (about 100 simultaneous people)
  • Add step-by-step Nebari instructions
  • Request jupyterlab-myst
  • Update scipy2023/index.ipynb
  • Logistics (Add more options to run interactively on the cloud #171)
  • Alignment and broadcasting
  • Address Cartopy plot not rendering #188
  • Address Add preview build functionality for PRs #68
  • Address test out jupyterlab-myst #115,
  • Add top-level banner / update left sidebar to guide attendees to right page (Add SciPy 2023 outline, banner. #189)
  • Optional
    • Address intersphinx/sphinx-hoverref #82,
    • Address add an index or overview notebook for binder. #153
    • Address Try sphinx-tags #142?
    • Address Add sphinx-autocodelink & intersphinx_mapping #85?
• Indexing (docs reference, @negin513 @andersy005 )
  • Existing material
  • Basic Indexing
  • Advanced and Vectorized Indexing
  • Boolean Indexing and isin, and masking
  • Fancy Datetime Indexing
  • Address Merge in material from "Xarray Online Tutorial" #94
• Computational Patterns (@JessicaS11 @dcherian )
  • Address feedback on "high-level computation patterns" #166
• Wrapping other arrays (docs reference, @TomNicholas , @andersy005 )
  • Concepts
  • Introduction to Xarray wrapping Dask (Migrate to exercise syntax #196)
  • Demo of other arrays: pint, sparse, cupy?
• Wrapping custom computations: apply_ufunc (docs reference) (@dcherian, Update and reorganize apply_ufunc material #180)
  • Concepts : core dimensions,
  • Vectorizing
  • Dask
  • Address address feedback on apply_ufunc tutorial #158
• Demo?
• Evaluation Survey?

Notes

At a high level, we have to work on the indexing material, and the "duck array" material. The xarray docs on indexing are fairly comprehensive so it's mostly presenting that in a friendly manner. The duck array material is fairly minor, so that would be a decent lift.

[dcherian]

Meeting Notes

[dcherian]

Pasting the email about Nebari for the SciPy tutorial.

I think we should do this!

---

We are emailing SciPy presenters to share an offer from Quansight to provide a hosted platform (JupyterHub, via Nebari) to run your tutorial.

Your workflow, and all participants' workflow, will involve: - Registering on the platform [1] (with any email and password), - Selecting a suitable machine [2] (i.e., adequate compute and storage for your tutorial), - Cloning the tutorial (git) repository in the JupyterLab interface that opens by default, - Teach your tutorial :)

As the presenter, you will have one extra step before the tutorial: create a suitable environment for your tutorial. If you have an environment.yml or requirements.txt file, this is a quick step which will take ~5 mins at most, and the Quansight team will assist you with this.

The platform (Nebari) has Dask integrated, and the capability to share files and dashboards within the platform. You can consider using this as an alternative to Binder, which currently has reduced capacity and funding. The Quansight team will be at SciPy in-person and will set up a dedicated Slack channel for support during the tutorial sessions.

If you're interested, please fill out this Google Form: https://forms.gle/recAwR1DaSau2Gsu6 by 25th June, and the Quansight team will reach out to share more information and schedule a live demo session to onboard you to the platform. If you have any questions, please contact Dharhas Pothina ([email protected]) or Pavithra Eswaramoorthy ([email protected]).

[1] Quansight won't collect or store your email on registering, you can also register with a fake email. [2] They will offer two types of machines for everyone: Small (2 CPUs, 8 GB RAM), Medium (4 CPUs, 16 GB RAM). If you need larger instances or a GPU instance, let them know in the Google Form and they will include it.

[scottyhq]

I think we should do this!

Sounds good to me! Small instances should be fine. @lsetiawan would you be willing to fill out the form and be the point of contact for this?

[lsetiawan]

I think we should do this!

I think this is great that they're offering Nebari! Yes. I totally agree that we should do this 😄 I've used Nebari in the past, and I think it's cool that they're willing to host for tutorial.

[dcherian]

@andersy005 @negin513 @TomNicholas can we get a quick update here please?

I think at this point, we could just plan to use the existing material as-is to reduce stress. There's definitely plenty of content!

[dcherian]

@lsetiawan do you have any more info on the nebari deployment? I don't see any info on the scipy tutorial website so far

[lsetiawan]

The rep from quansight will try to deploy it from the repo. I'm currently on vacation. I'll ping her again tomorrow.

Hi Don,

I hope you have a nice vacation!

The onboarding meeting diverged off-topic, so I'll share a fresh set of instructions in a day or two. I'll also test the xarray tutorial on Nebari and let you know (or maybe open issues/PRs against the repo) if we need anything from you. :)

[negin513]

Hello @dcherian ,

I created a PR for the indexing notebooks. #192

Regarding the last task under Indexing in the list above :

• Indexing (docs reference, @negin513 @andersy005 )
  • Existing material
  • Basic Indexing
  • Advanced and Vectorized Indexing
  • Boolean Indexing and isin, and masking
  • Fancy Datetime Indexing
  • Address Merge in material from "Xarray Online Tutorial" #94

I can write about broadcasting, aligning, etc in these notebooks. But was not sure if it fit nicely under indexing section. Do you envision having these under advanced indexing? Does it fit the advanced indexing or the computation section better?

Here is the notebook for reference:
https://tutorial.xarray.dev/workshops/online-tutorial-series/03_computation.html

[dcherian]

I can write about broadcasting, aligning, etc in these notebooks.

Sorry for being unclear. We can skip that for now. We already have material on alignment and broadcasting here