Mechanisms and Machine Learning for Magnesium Alloys Design

摘要

We will show our extensive high-throughput studies for magnesium alloys through both the dislocation and twinning mechanisms. Possible descriptors for the mechanisms are explored and a united picture is demonstrated, which is consistent with available experiments. There are two major contributions of this work, i.e., (i) The relationship between two well-acknowledged deformation mechanisms based on dislocations is clarified; (ii) Machine-learning models show that it is possible to design ductile magnesium alloys without the prior knowledge of deformation mechanisms.