Computational Science Philosophy

I am both a physicist and a computational scientist. I look at research topics from a physics perspective and ask “how does this work?” This question tends to motivate the “why” of research. Then I use the tools of computation and applied mathematics to carry out science by testing hypotheses, this is the “how” of research.

My Approach To Computational Science

My approach is to achieve reproducible results by thorough code documentation, strict version control, and ongoing testing.

Validation is ongoing, not just at the end of a project or done once and then forgotten.

Development should keep in mind specific tests the numerical code should pass. Sometimes this means unit tests, sometimes this means regression tests.

Run tests on an ongoing basis using Continuous Integration through, e.g. GitHub Actions or an equivalent.

Catch regression or unit test errors at pull request review time, if not earlier, before breaking changes have a chance to make it into the main code. This simple approach is enormously useful when a bug appears in my simulations and I find myself backtracking through the code history to pinpoint where it came from!

Always have someone review your code, even if you are just working with one other person.