摘要
Recent advancements in machine learning(ML)have revolutionized the field of high-performance materials design.However,developing robust ML models to decipher intricate structure-property relationships in materials remains challenging,primarily due to the limited availability of labeled datasets with well-characterized crystal structures.This is particularly pronounced in materials where functional properties are closely intertwined with their crystallographic symmetry.Weintroduce a selfsupervised probabilistic model(SSPM)that autonomously learns unbiased atomic representations and the likelihood of compounds with given crystal structures,utilizing solely the existing crystal structure data from materials databases.SSPM significantly enhances the performance of downstream ML models by efficient atomic representations and accurately captures the probabilistic relationships between composition and crystal structure.We showcase SSPM’s capability by discovering shapememory alloys(SMAs).Amongst the top 50 predictions,23 have been confirmed as SMAs either experimentally or theoretically,and a previously unknown SMA candidate,MgAu,has been identified.
基金
supported by the National Natural Science Foundation of China(51931004,52322103,52350710205,and 52171011)
Key Technologies R&D Program(2022YFB3707600)
the 111 project 2.0(BP2018008)
T.Q.L.acknowledges the support from the China Scholarship Council(202306280011).
