Strategies for accelerating the adoption of materials informatics
- PDF / 2,030,684 Bytes
- 7 Pages / 585 x 783 pts Page_size
- 113 Downloads / 236 Views
troduction The use of informatics techniques in materials science has increased dramatically in the past decade, and for good reason. Data-driven techniques, which focus on extracting predictive models or knowledge from data, are starting to yield revolutionary impacts in a multitude of fields such as transportation1 and health care.2 Informatics methods are also increasingly prevalent across diverse domains within materials science. For example, the use of data-driven methods has enabled the rapid discovery of new materials3 and improved characterization methods,4 while providing another path for understanding material behavior.5 Overall, materials informatics has promise to create impactful advancements in many areas of materials science. But the question of how the adoption of these useful techniques can be speeded up remains. The use of materials informatics has been especially prevalent in atomic-scale modeling. Materials discovery achieved by screening databases of material properties computed using quantum-mechanics-based calculations has become a routine tool in the development of materials,6 as evidenced by the many publicly available databases of density functional theory (DFT)
calculations.7 These databases, in turn, have been used with automated analysis techniques to learn more about materials behavior. Furthermore, it is increasingly common to use machine learning to build fast, surrogate models based on DFT calculations. The broad adoption of informatics techniques in the atomic-scale modeling community suggests that it could serve as an example for how informatics could be used elsewhere. In this article, we first discuss what has made informatics so successful in atomic-scale modeling. We start by identifying the major types of informatics studies used in atomic-scale modeling and assessing their impacts. We then focus on aspects of the data and the data infrastructure that have enabled the proliferation of informatics methods in atomic-scale modeling. Finally, we review the use of informatics by the broader materials research community and discuss the activities that may enable more frequent use of materials informatics.
Materials informatics and atomic-scale modeling Materials informatics in atomic-scale modeling has included a large variety of approaches and spanned diverse types
Logan Ward, The University of Chicago, and Data Science and Learning Division, Argonne National Laboratory, USA; [email protected] Muratahan Aykol, Toyota Research Institute, USA; [email protected] Ben Blaiszik, The University of Chicago, and Data Science and Learning Division, Argonne National Laboratory, USA; [email protected] Ian Foster, Department of Computer Science, The University of Chicago, and Data Science and Learning Division, Argonne National Laboratory, USA; [email protected] Bryce Meredig, Citrine Informatics, USA; [email protected] James Saal, QuesTek Innovations LLC, USA; [email protected] Santosh Suram, Toyota Research Institute, USA; [email protected] doi:10.1557/mrs.2018.204
• VOLUME C
Data Loading...
