Key points are not available for this paper at this time.
The machine learning revolution hasdemonstrated the power of leveraging hugeamounts of data and computation to solveproblems. Computational chemistry seemspoised to create a similar revolution in drugdesign, catalysis, and materials science.However, the field of atomistic simulationhas existed for a long time, with decades ofresearch in academia and industry. The hugeopportunities in this space have drawn theattention of tech companies, pharmaceuticalcompanies, and venture capital whilespurring the creation of dedicated softwareplatforms for molecular discovery.Care must be taken when finding scalableopportunities in such a sought-after field.After briefly reviewing the state-of-the-art, Imake some general points to guide thediscussion of opportunities in atomisticsimulation. I suggest a few unique avenuesfor starting focused research organizationsin the field, centered around creating highquality datasets to spark a revolution inFrictionless Reproducibility and thedevelopment of autonomous agents foratomistic simulation.
Sam Holton (Tue,) studied this question.