Computational methods articles from across Nature Portfolio

Heterogeneous interactions between interactive entities are not well understood due to their complex configurations and many body interactions. Han et al. present a probabilistic-based machine learning method to discover the fundamental laws governing the interactions of heterogeneous systems.

Quaternary oxides can be synthesized from a variety of precursors, but there is a poor understanding of how to design efficient synthesis recipes. Here a strategy to navigate high-dimensional phase diagrams in search of the best precursors for quaternary oxide materials is reported and validated experimentally by a robotic laboratory.

Using a fully automated closed-loop system, a robotic chemist synthesizes an oxygen-evolution catalyst from Martian meteorites, with implications for the production of catalysts that may underpin human survival on Mars in the future.

It is difficult to identify stable surface reconstructions of complex materials. Now a Monte Carlo sampling strategy is coupled with a machine learning interatomic potential that is iteratively improved via active learning during the search.

Graph neural networks (GNNs) present a promising route for machine learning of solid-state materials’ properties, but methods capable of directly predicting defect properties from ideal, defect-free structures are needed. A GNN developed for direct defect property predictions enables high-throughput screening of redox-active oxides for energy applications and beyond.

The prediction of stable crystal structures is an important part of designing solid-state crystalline materials with desired properties. Recent advances in structural feature representations and generative neural networks promise the ability to efficiently create new stable structures to use for inverse design and to search for materials with tailored functionalities.