Jupyter workshop (ICALEPCS2019)

Saturday 5 Oct 2019, 09:00 → 17:00 US/Eastern

Williamsburg room (New York (US))

Hans Fangohr (European XFEL GmbH), Sandor BROCKHAUSER (EXFEL)

The Jupyter Notebook and associated tools from Project Jupyter (such as JupyterLab, JupyterHub, BinderHub, Binder, nbval, nbconvert, nbmerge) are becoming a key tool in data science, computational science and data analysis. They are increasingly used at research facilities. Users include support teams, facility scientists and users.

In this work shop, we provide a brief overview of the Jupyter Ecosystem to outline tools and use cases. We invite contributions from all facilities and participants that show the current, planned or desired use of Jupyter at their facilities or elsewhere. Positive experiences as well as negative experience form useful contributions to help others in their planning.

The detailed agenda is displayed below. 

To suggest contributions and for informal queries, please contact Hans Fangohr (hans.fangohr@xfel.eu).

This workshop is a satellite meeting associated with the 17th International Conference on Accelerator and Large Experimental Physics Control Systems (ICALEPSC2019). 

This workshop is supported by the European Union’s Horizon 2020 research and innovation programme under grant agreement No 823852 (PaNOSC).
This page is available at http://indico.desy.de/indico/e/jupyter-icalepcs2019.

09:00 → 09:30 Jupyter Notebook and Ecosystem

A brief introduction to the Jupyter Notebook and the ecosystem of tools that are based on it. Includes JupyterHub, JupyterLab, NBDIME, NBVAL, Binder.

Speaker: Prof. Hans Fangohr (European XFEL GmbH)

Slides jupyter-workshop-icalepcs2019-fangohr-introduction.pdf

09:30 → 10:00 Status updates from Soleil, CERN, European Southern Observatory, J-PARC MLF, Max IV

10:00 → 10:30 Coffee break

10:30 → 12:30 JupyterLab Tutorial

JupyterLab is the next generation of the notebook interface. This talk will start with an overview of the JupyterLab interface to show how it can be used for more than just editing notebooks. JupyterLab was designed with extensibility in mind, so next I will show how to create a plugin that adds custom functionality to the interface. Then I will show some existing extensions that target the scientific ecosystem, including support for exploring HDF5 files. Finally, I will end with an exploration of new features coming to JupyterLab including real time collaboration.

Speaker: Mr Saul Shanabrook (Quansight)

Slides https://github.com/saulshanabrook/icalepcs-2019-jupyterlab

12:30 → 14:00 Lunch and networking

14:00 → 14:30 Jupyter for processing neutron event data at ESS

Speaker: Dr Jonathan Taylor (European Spallation Source)

Slides JupyterTalk.pdf

14:30 → 15:00 Jupyter at Brookhaven National Laboratory

Speaker: Dr Daniel B. Allan (Brookhaven National Lab)

Slides https://danielballan.github.io/jupyter-at-bnl/

15:00 → 15:20 Jupyter for Accelerator Physics

RadiaSoft operates a public JupyterHub instance at jupyter.radiasoft.org, which is used for workshops, the US Particle Accelerator School (USPAS), our clients, and the accelerator community at-large. This free, public service simplifies classroom startup at USPAS, because all the accelerator codes and tools they might need are already installed. We deliver models to our clients via IPython notebooks who can run them instantly on JupyterHub. We make it easy for instructors to share notebooks during a workshop with all their students. Additionally, accelerator physicists at RadiaSoft use Jupyter as an “IDE” for developing models and running simulations. Most of the codes we use are Python-wrapped so IPython notebooks are an ideal tool for documenting the parameters associated with a model. RadiaSoft’s Jupyter Docker image contains most of the key accelerator physics codes including elegant, EPICS, FBPIC, JSPEC, OPAL, Radia, Shadow3, SRW, Synergia, Warp, Zgoubi as well as support tools such as SciPy, Tensorflow, and TeX. The image is large (10GB), but allows our staff, clients, and the public to work with all codes simultaneously and seamlessly. In order to facilitate the various use cases, we have implemented and open-sourced some subclasses and tools. RSDockerSpawner supports resource pools that allows us to manage our public, workshop, and private Jupyter instances on a single cluster. An MPI wrapper allows simple and secure container-based execution of MPI-based codes by our staff. Users can also persist their Jupyter environment customizations so that they survive container restarts. Here we will provide an overview of our capabilities and detail some example use cases for the simulation of particle accelerators.

Speaker: Dr Jonathan Edelen (RadiaSoft LLC)

Slides jupyter_for_simulations.pdf

15:30 → 16:00 Coffee break

16:00 → 17:00 Questions and answers, discussion

Convener: Prof. Hans Fangohr (European XFEL GmbH)