This title appears in the Scientific Report :
2018
Publishing a complex model code
Publishing a complex model code
The recently emphasized 'reproducibility crisis' in science [1] stresses the value of publicly available code to reproduce scientific results. Beyond reproducibility, publishing model code also enables other researchers to build on top of previously existing work, which is at the heart of...
Saved in:
Personal Name(s): | van Meegen, Alexander (Corresponding author) |
---|---|
Schmidt, Maximilian / Schücker, Jannis / Diesmann, Markus / van Albada, Sacha | |
Contributing Institute: |
Jara-Institut Brain structure-function relationships; INM-10 Computational and Systems Neuroscience; IAS-6 Computational and Systems Neuroscience; INM-6 |
Imprint: |
2018
|
Conference: | NEST Conference 2018, Ås (Norway), 2018-06-25 - 2018-06-26 |
Document Type: |
Poster |
Research Program: |
Human Brain Project Specific Grant Agreement 2 Human Brain Project Specific Grant Agreement 1 The Human Brain Project Connectivity and Activity Theory, modelling and simulation Brain-Scale Simulations |
Publikationsportal JuSER |
The recently emphasized 'reproducibility crisis' in science [1] stresses the value of publicly available code to reproduce scientific results. Beyond reproducibility, publishing model code also enables other researchers to build on top of previously existing work, which is at the heart of the scientific method. However, the process of preparing a model code for publishing is still cumbersome and full of unforeseen surprises.In this contribution, we share hands-on experience from publishing a multi-scale spiking neural network model of macaque visual cortex [2, 3, 4]. Our goal is to facilitate future model code publications by identifying possible pitfalls and sharing experience on how to avoid them. We outline structural decisions, such as the publication site, licensing of the code, and the handling of external dependencies. Furthermore, we describe challenges such as compatibility issues between different hardware systems and software environments, and ways to overcome them.The process of publishing model code turns out to be of advantage in itself. Refactoring of the executable model description and code review increase the chances of spotting any errors, and induce an effective knowledge transfer between successive researchers working on the model. |