2017 Geospace Hackathon

From CedarWiki
Jump to: navigation, search

Monday, June 19: 12:15-13:15, Crestone I

Convener: Michael Hirsch, Ph.D.

Given the leap forward in the role of software and networks of sensors in geospace science of all types, the Geospace Hackathon at the CEDAR 2017 workshop is where people peer one-on-one or in small groups on Programming, Hardware or other geospace-relevant issues.

Contents

Outcomes

Thank you to the approximately 30 attendees (spanning from undergrad to senior researchers and faculty) at the inaugural CEDAR Geospace Hackathon!

We heard from collaborators on major projects including:

  • DavitPy
  • PyGlow
  • THEMIS (satellite & ground-based)
  • Digital RF

as well as those needing to improve research speed and efficiency by working with multiple programming languages without porting (converting) code from one language to another.

General topics discussed

  • How to make my code/data citable (Zenodo)
  • How to seamlessly share & integrate code with colleagues using different programming language
  • How to make my code more reliable

Specific topics discussed

  • how to get started with best practices? How to get a clean "setup.py" and start continuous integration -- my suggestion was to find an expert and have them help, then learn by modifying that example.
  • we agreed that it would be better to have the Linux model of "a geoscience tool for a specific purpose" for much better reuse and repeatability, rather than endless copies and customizations of IRI for example. This is solved by more effective standards-based use of "setup.py" and import system.

Calling Python from Matlab

Calling Matlab from Python

Calling Octave from Python (to run Matlab code without having a Matlab license)

GDL (open-source IDL 7.x compatible)

help on compiling GDL

Upload Python toolbox to PyPi so others can "pip install MyIonosphereModel"

Example setup.py handling complex Python prerequisites easily and quickly for Mac,Linux, Windows

  • A couple coauthors of major Python geospace packages discussed improving robustness of installing prerequisites on all operating systems by increasing effectiveness of "setup.py".
  • clear need to implement registration tests for major geospace packages, cross-platform.

This is a prerequisite for deduplicating effort and thereby greatly improve reproducability and reliability of results. We don't want 10 different implementations of IRI90.

  • How to make a central CEDAR resource for finding *programs* and *data*
  • Ipython 6 (released spring 2017) is Python 3 ONLY

Possible Grant/Paper Opportunities

Validating/comparative results across popular model versions. E.g. IRI and/or MSIS.

Other sessions you may be interested in

2017_Workshop:Digital_Geospace

2017_Workshop:Python_for_Space_Science

Online Geospace Coding Communities

#geospace-coders Slack group

Thanks

Thanks to CEDAR Workshop for facilitating this spontaneous session!