Seminar: Reproducible Software Environments with GNU Guix
Free lunchtime seminar for researchers
Reproducibility is the means by which peers can verify each other’s claims. It is a cornerstone of “open science” - or, really, science. While it has long been neglected, the ability to reproduce software environments is a prerequisite for anyone willing to reproduce computational experiments. Reproducible software deployment is the tool we scientists need to build robust foundations.
Time: Tue 2023-05-16 12.00 - 13.00
Location: In person: SeRC Room (Room 304), 5th floor, Teknikringen 14, KTH, or online (see link below)
Video link: https://bbb.inria.fr/cou-yhd-ku4-qub
Language: English
Participating: Ludovic Courtès, French National Institute for Research in Digital Science and Technology (Inria)
In this talk Ludovic Courtès will present GNU Guix, a reproducible deployment toolkit that many have been taking advantage of in the context of open science and high-performance computing (HPC) via the Guix-HPC effort. He will explain how it compares to other approaches—from package managers like Spack and EasyBuild to “container solutions” like Singularity and Docker. We will see what it’s like to use it for common tasks, and how it can help you share computational experiments with peers or simply re-run your own code on different machines or at different points in time.
About the speaker
Ludovic Courtès works as a research software engineer (RSE) at Inria, the French National Institute for Research in Digital Science and Technology. In 2012 he started Guix, a project to develop a reproducible software deployment toolkit. He has been working on it with an eye on using it as a foundation for reproducible research workflows and high-performance computing as part of the Guix-HPC workflows. His interest lies in leveraging functional programming to make hard problems more tractable.
Light lunch will be provided
A light lunch of wraps and soft drink will be provided for those attending in person.