CRC 1404/2: Hardening Computational Materials-Science Workflows against Human Errors (SP A03)

Ab-initio codes of Computational Materials Science (CMS) can predict a variety of materials properties. However, scientists are confronted with a huge variability of such workflows, which can lead to user interaction errors. In phase I, it was focused on improving this situation for individual tasks the project within a CMS DAW by developing systematic testing strategies. In the second phase, we will extend our focus to address the complete CMS data analysis pipeline including downstream machine learning applications. The project will research new technologies for creating guide-lines and recommendations for scientists about how to compute material properties with a desired level of accuracy.