As a part of the EarthCube 2020 meeting, we will be holding a community input forum for the Jupyter meets the Earth project. We would appreciate your input and participation!
A virtual meeting to be held online on Monday July 27 at 8am - 10am PDT
Project Jupyter is an open source platform for interactive computing and data analysis, widely used in research, education and industry. The Jupyter meets the Earth project is using research use cases in geosciences to drive technical developments within the Jupyter and Pangeo ecosystems. This project revolves around the following key goals: (1) Facilitate the discovery, integration, and effective use of the diverse sources of data in the geosciences. (2) Empower researchers to utilize modern, scalable compute resources. (3) Accelerate the process of discovery by enabling researchers to rapidly create and deploy custom interactive applications tailored to the research question at hand. (4) Make it possible to communicate scientific results in a manner that is tailored to the final consumers of research – be they other scientists, policy makers, students, or the general public.
Our technical targets include improvements in JupyterHub for interactive computing on High Performance Computing (HPC) and cloud infrastructure, the development of JupyterLab extensions for data discovery, and contributions to widgets and dashboarding solutions for researchers to easily create graphical user interfaces as well as interactive documents to share analyses with broad audiences. You can find more information about the project at https://bit.ly/jupytearth.
We would like to gather input for how to best serve your research needs, exploring questions such as:
What are current bottlenecks in your interactive computing workflow?
What integrations with geoscience-specific tools would be useful, or could be made better via closer ties with Jupyter infrastructure?
How would you like to publish and share your computational research and where can improvements be made (e.g. Binder, JupyterBook, etc.)?
Desktop vs local cluster vs HPC vs cloud: what is your workflow today? What do you envision it will be in 5 years?
Where are the pain points in working with your data on shared infrastructure (cloud or HPC)? Data discovery? Sharing data with collaborators? …
Whether you are an active participant in the Pangeo community, you use Jupyter tools in your work, or are considering adopting some of these tools, we welcome your input and ideas.
Please fill out the short form to register for the virtual meeting. We will reach out to all participants with an agenda closer to the event date.
The meeting is planned to consist of two parts: we will start off with a series of short presentations from project participants and related projects and then will hold a discussion session to address questions you have or brainstorm ideas with you all. If you have questions you would like to have raised during the discussion, please post them ahead of time here! This will give us time to prep answers and resources. The schedule is below (all times in PDT):
We will record the meeting, so if you prefer not to be recorded, please turn off your camera.
For chat and interactive discussion during the meeting, we will use the EarthCube Slack. Please watch for a new #jupyter-meets-the-earth channel for discussion during the meeting.
Please don’t hesitate to reach out if you have any questions prior to the meeting.
For discussion: I think it would be helpful to develop variants of the pangeo gallery examples targeted at desktop machines or small 8 core nodes using docker-compose. We’re currently working on this for jupyter-book: https://github.com/executablebooks/jupyter-book/issues/765. The idea would be to decouple learning docker from learning k8s, a lot like https://github.com/dask/dask-docker
How would you like to publish and share your computational research and where can improvements be made?
Having a minimal/standard way to share notebooks for users on the same Jupyterhub would be a game changer IMO. Currently, the existing discussions (see the GitHub issue below) are about sharing notebooks via lab extensions, but I think that it would be very useful for JupyterHub to have a built-in sharing service.