While scientific companies frequently publish their research in academic journals, it seems broadly true that publication is not incentivized for companies the same way it is for academic groups. Professors need publications to get tenure, graduate students need publications to graduate, postdocs need publications to get jobs, and research groups need publications to win grants.

If you are a scientist, odds are you should be reading the literature more. This might not be true in every case—one can certainly imagine someone who reads the literature too much and never does any actual work—but as a heuristic, my experience has been that most people would benefit from reading more than they do, and often much more.

It’s a truth well-established that interdisciplinary research is good, and we all should be doing more of it (e.g. this NSF page). I’ve always found this to be a bit uninspiring, though. “Interdisciplinary research” brings to mind a fashion collaboration, where the project is going to end up being some strange chimera, with goals and methods taken at random from two unrelated fields.

Recently, I’ve been working to assign the relative configuration of some tricky diastereomers, which has led me to do a bit of a deep dive into the world of computational NMR prediction. Having spent the last week or so researching the current state-of-the-art in simulating experimental ¹ H NMR spectra, I’m excited to share some of my findings.

Python is an easy language to write, but it’s also very slow.

Eric Gilliam, whose work on the history of MIT I highlighted before, has a nice piece looking at Irving Langmuir’s time at the General Electric Research Laboratory and how applied science can lead to advances in basic research.

When thinking about science, I find it helpful to divide computations into two categories: models and oracles. In this dichotomy, models are calculations which act like classic ball-and-stick molecular models. They illustrate that something is geometrically possible—that the atoms can literally be arranged in the proper orientation—but not much more. No alternative hypotheses have been ruled out, and no unexpected insights have emerged.

—Richard Hamming What’s the difference between science and engineering? Five years ago, I would have said something along the lines of “engineers study known unknowns, scientists study unknown unknowns” (with apologies to Donald Rumsfeld), or made a distinction between expanding the frontiers of knowledge (science) and settling already-explored territory (engineering). These thoughts seem broadly consistent with what others think.

Every year, our group participates in a “Paper of the Year” competition, where we each nominate five papers and then duke it out in a multi-hour debate. Looking through hundreds of papers in a few weeks is a great exercise: it helps highlight both creativity and its absence, and points towards where the field’s focus might turn next.

For almost all Hartree–Fock-based computational methods, including density-functional theory, the rate-limiting step is calculating electron–electron repulsion.

While looking over papers from the past year, one theme in particular stood out to me: meta-optimization, or optimizing how we optimize things.