Machine Learning with Structural Fingerprints of Local Particle Environments

As the computational power available to scientists increases, researchers themselves — rather than hardware or algorithms — become the bottleneck in scientific discovery. The computational study of colloidal self-assembly is one area that keenly feels this effect: even after computers generate massive amounts of raw data, performing an exhaustive search to determine what, if any, ordered structures occur in a large parameter space of many simulations can be a long, manual process. Here we demonstrate how “off-the-shelf” machine learning algorithms can be applied to results of self-assembly studies both to find interesting regions in a phase diagram and identify characteristic local environments in simulations in an automated, high-throughput manner for both simple and complex crystal structures. These methods can form the foundation of an intelligent exploration of parameter space — a key component in the process of creating new materials by design.