Effective practices
Which software practices give a scientific team the most value for the effort, at a level of formality the team can sustain.
Scientific software is a tool for research and, we believe, a subject of research in its own right.
The Collegeville Project studies how scientific software is designed, developed, tested, sustained and used, and treats that work as a first-order concern.
The problem
Scientists and engineers rely on software as a part of their everyday discovery and design processes. Many of them also write it, with the science as the goal and the software as a means to that end. In our experience, improvements in how this software is built usually come as a side effect of a specific project, and the lessons learned tend to stay with that project.
Our approach
We want those lessons to be studied, tested and shared. Much of our interest falls under research software science: applying the scientific method to understand and improve how scientific software is developed and used.
Current topics (the list will grow)
Trust
How verification, validation and reproducibility establish confidence in software and in the results it produces.
Small research teams
How a small group of permanent members and a changing group of students and postdocs learn, contribute and leave the software in good condition for those who follow.
Software ecosystems
How scientific libraries and tools are combined, shared and sustained beyond the life of any one project.
AI-assisted development
How AI tools change the way scientific software is written, reviewed and trusted.
Corpus
An archive of Human Skills, AI-Assisted Workflows, and Agent Skills for technical research and communication, built from Michael Heroux's computer science capstone.
Visit Corpus ↗Where this comes from
The Collegeville Workshops on Scientific Software have brought together participants from industry, academic and laboratory institutions to work on these topics. Communication is also part of the work: since 1998, Michael Heroux has taught a computer science capstone in research methodologies at the College of St. Benedict and St. John's University in rural Collegeville, Minnesota.
This site is new, and more of our work will appear here over time. The first piece is Corpus.