Looking inside the black box of computational chemistry

This week, I'm teaching faculty from Primarily Undergraduate Institutions (PUIs) about the underlying concepts behind molecular dynamics simulations. This seventh workshop on computational and theoretical chemistry is part of the cCWCS workshops aimed at STEM education dissemination and undergraduate research capacity building. Our first workshop was held in 2002, now more than a decade ago! I cover statistical mechanics, David Sherrill covers electronic structure, and Tricia Shepherd covers the hands-on labs. We alternate (roughly) between holding them in Atlanta and in Salt Lake City. Both have their charms. The Westminster Campus is a great venue because it's an idyllic oasis in the middle of the city. It has the open greens and architectural gems that you would expect from a private undergraduate liberal arts school. It also has the requisite high-caliber undergraduate students and facilities to do first-rate science. Holding the workshop here thus makes it easier for participants to see that the computational tools we sample can be replicated at their home institutions.

This workshop, however, is different than most such computational chemistry workshops because our emphasis is on the underlying theory in the computational codes, and not just on how to run a particular computational package. There's nothing wrong with doing the latter. However, we feel that it's useful to understand when computational chemistry calculations or simulations are meaningful or not. To that end, one needs a bit deeper knowledge of how exactly the algorithms are working within the codes. In part this means knowing the underlying equations. In equal measure, it also means understanding the underlying concepts. So the lectures in our workshop tend to focus (a lot!) on the theory. In like fashion, we also want to help participants understand how the algorithms are implemented. Although the labs are run on Apple laptops (which is known for its fancy graphical user interfaces), we guide them into the unix layer, editing files and running jobs from the command line. Thus our goal is to have participants be more comfortable with what the calculations are doing and how they are being done within software packages, whether they're computing energetics or dynamics. In knowing what's inside these black boxes, they should be in a better position to set-up computational experiments and to mentor their undergraduate students and researchers to do likewise. The fact that the workshop continues to be oversubscribed gives me hope that we are achieving at least some part of this goal.