Skip to content

Latest commit

 

History

History
40 lines (27 loc) · 4.71 KB

2015TR.md

File metadata and controls

40 lines (27 loc) · 4.71 KB
Raw
title layout
2015 Trip Report for ONRG
section

Overview

At the behest of ONR Global, reservists led a two-week software engineering workshop in Ghana. This workshop was an continuation of previous workshops in 2013 and 2014. In 2015, the workshop was held as collaboration with a pan-African graduate training organization, the African Institute for Mathematical Sciences (AIMS), at their Ghanaian site. Twenty students participated. The official name of the program is the Computation in Applied Mathematical Sciences (CAMS) Workshop, and is tracked within ONR-RC as the "African STEM Pipeline" project.

Summary Statistics

  • Two reservists, and 5 civilians (3 US-based, 2 Ghana-based) worked with the participants. Civilian participants were independently funded (U. of Texas, U. of Florida, AIMS)
  • Participants came from 7 African nations: Ghana, Nigeria, Madagascar, Ethopia, Benin, Togo, Uganda (in order of # of participants). Participants were supported by AIMS.
  • Of 20 participants, 8 were women
  • Participants worked on 10 projects; 2 led to contributions to open-source scientific work, 2 others continued on to publication after the workshop

Summary of Workshop Events

The workshop covered several software engineering topics, with instruction and activities focused on software development to support scientific research. The complete course workshop is available online. In brief, the topics were:

  • Defining a Programming Project: reviewing how to express a research question as a software development project
  • Organizing at the Project level: learning how to piece together data, pre-processing, simulation, analysis, and output into a reliable pipeline
  • Organizing at the Code level: discussing and applying best practices for architecture & syntax
  • Test-driven development: using the ideas of test-driven development as a way to explicitly communicate model assumptions and improve confidence in results
  • Managing Input & Output: deciding how to store and access large or confidential data sets, selecting intermediate outputs (i.e., checkpointing), format choices
  • Modeling in code: expressing scientific models as software models
  • Theoretical Performance Concerns: understanding how algorithms and representations contribute to growth in time and space requirements for analysis
  • Practical Performance Concerns: understanding how software engineering tradeoffs (e.g., compiled vs. scripted languages, single-thread vs. parallel implementations, human vs. machine readable outputs) influence time, space, and reusability

All workshop topics included both discussion and practical sessions, followed by a free-work period focused on applying the concepts to participant's projects.

Outcomes

All participants completed the workshop and provided overall positive feedback. The workshop co-sponsor, AIMS Ghana, was very pleased with the workshop outcomes and has made arrangements to partner with us again in 2016. Participants contributed to two open source science projects: AbcSmc, and EpiFire. Two participants went on to publish work, one as new open source software (Hopcroft-Karp algorithm implementation in python) and the other as formal dissertation project (which is continuing on towards development of an R library). The other participants all demonstrated some project progress for the final presentations.

Plan for 2016

We plan to run the workshop again in summer 2016, incorporating feedback from students. The primary revisions entail tuning the in-discussion activities, adding homework, and diversifying project opportunities. To expand the project opportunities, ONR-RC worked with both ONR Global and AIMS to identify project work. Calls to ONR Global "alums" (current or previously funded researchers working in Africa) have not yet identified potential collaborations, but we will continue to push for those opportunities. Working with AIMS, we have identified local opportunities with commercial (primarily telecom) and public (primarily health services) interests.

The local site and broader AIMS communities are working to identify additional funding options for student support. Longer term, AIMS would like to see the workshop rotating annually among their several sites, expanding the influence of the program. This is modeled on similar successful programs at other AIMS sites (e.g., the MMED program, which inspired the original structure for this workshop). Expanding the program influence is obviously beneficial to ONR Global objectives, especially since these workshops are largely organically funded.