Even if all research lines are fully connected, and macroscopic properties of materials could be exactly predicted from their atomic compositions, finding materials that show certain pre-defined large-scale behaviors often poses a significant challenge.
This is due to a combinatorial explosion of microscopic realizations that can't all be subjected to detailed simulations. What we need, are automated procedures capable of exploring high-dimensional parameter spaces and dynamically proposing, based on results already obtained, promising regions in parameter space. This research line takes Big Data and Machine Learning as an area that requires early investments by the CCER. International collaborations will be sought to ensure this line of research benefits from economies of scale in setting up Materials Databases.