Predictive a priori materials design

Establish whether predictive, a priori materials design can replace intuition-driven, high-throughput experimentation when accurate electronic structure and tractable exploration of combinatorial composition and structure spaces are both required.

Background

Materials discovery requires simultaneous control of two difficult tasks: accurately evaluating electronic structure and searching enormous spaces of compositions, structures, disorder patterns, and competing phases.

The paper identifies the replacement of empirical, intuition-driven discovery by predictive computational design as a central unresolved question. Existing machine-learning and computational-screening approaches have not yet demonstrated superiority over the empirical success of high-throughput experimentation.