Open discussions on specific topics selected by the Software Working Group and selected from the list of SWG Topics For Discussion.
Tuesday, February 8, 2022 - The Pragmatics of computational reproducibility in scientific applications, moderated by Santiago Nunez-Corrales
Slides: https://documentcloud.adobe.com/link/track?uri=urn:aaid:scds:US:ea61ee38-af5c-4f88-99cf-1542a71aa7ba
Recording: https://uofi.box.com/s/z2o39q0p25n42wvr27s213lc2hwq81qr
Attendees:
Sandeep Puthanveetil Satheesan
Discussion:
An increasing number of researchers and scholars benefiting from computational science applications add pressure to our extremely limited ability to reproduce experimental outcomes across all science, engineering and the humanities. The reproducibility research community, regardless of specific definitions around the matter, has mostly focused on the ability to take existing research objects and verify properties a posteriori. In addition, the challenge remains of determining whether a separation between correctness and reproducibility can be found, and if so, whether it is useful to facilitate reproducibility tasks and produce better science. We will explore 1) why reproducibility appears to be hard in the context of scientific computing, 2) what has been done in so far to mitigate the difficulties, and 3) potential avenues to help increase the reproducibility of scientific applications, with an emphasis on NCSA’s mission.
See slide presentation for details. For discussion, see recording.
Discussion of the PRIMAD model and what is static and what is dynamic.
RSE's need the background information of why the PI or client are doing what they are doing. It isn't just about the code
Use case: - Jong discusses some of the issues of scientists using software - we suggested that the users use Python and Jupyter Notebook rather than MatLab. By changing the methods in which they write code, it makes it more reproducible. This give us some control over how they perform their experiments.. By using the same software. We need to offer best practices for each group that we work with to use the same software - it is much easier to duplicate and reproduce issues.
Focus on initially correct code and try to break it. LOOK AT THE VARIABLES.
Run tests and see where the errors/changes are. Check for what changed: compilers, run time
Links Shared During the Talk:
This page might be a good place to link docs about project sustainability and end-of-life recommendations/policies: Project Development
If you are interested in contributing to a Round Table, please see these links: