Linus C Erhard, Daniel Utt, Arne J Klomp and Karsten Albe
{"title":"Crystal structure identification with 3D convolutional neural networks with application to high-pressure phase transitions in SiO2","authors":"Linus C Erhard, Daniel Utt, Arne J Klomp and Karsten Albe","doi":"10.1088/1361-651x/ad64f3","DOIUrl":null,"url":null,"abstract":"Efficient, reliable and easy-to-use structure recognition of atomic environments is essential for the analysis of atomic scale computer simulations. In this work, we train two neuronal network (NN) architectures, namely PointNet and dynamic graph convolutional NN (DG-CNN) using different hyperparameters and training regimes to assess their performance in structure identification tasks of atomistic structure data. We show benchmarks on simple crystal structures, where we can compare against established methods. The approach is subsequently extended to structurally more complex SiO2 phases. By making use of this structure recognition tool, we are able to achieve a deeper understanding of the crystallization process in amorphous SiO2 under shock compression. Lastly, we show how the NN based structure identification workflows can be integrated into OVITO using its python interface.","PeriodicalId":18648,"journal":{"name":"Modelling and Simulation in Materials Science and Engineering","volume":null,"pages":null},"PeriodicalIF":1.9000,"publicationDate":"2024-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Modelling and Simulation in Materials Science and Engineering","FirstCategoryId":"88","ListUrlMain":"https://doi.org/10.1088/1361-651x/ad64f3","RegionNum":4,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"MATERIALS SCIENCE, MULTIDISCIPLINARY","Score":null,"Total":0}
引用次数: 0
Abstract
Efficient, reliable and easy-to-use structure recognition of atomic environments is essential for the analysis of atomic scale computer simulations. In this work, we train two neuronal network (NN) architectures, namely PointNet and dynamic graph convolutional NN (DG-CNN) using different hyperparameters and training regimes to assess their performance in structure identification tasks of atomistic structure data. We show benchmarks on simple crystal structures, where we can compare against established methods. The approach is subsequently extended to structurally more complex SiO2 phases. By making use of this structure recognition tool, we are able to achieve a deeper understanding of the crystallization process in amorphous SiO2 under shock compression. Lastly, we show how the NN based structure identification workflows can be integrated into OVITO using its python interface.
期刊介绍:
Serving the multidisciplinary materials community, the journal aims to publish new research work that advances the understanding and prediction of material behaviour at scales from atomistic to macroscopic through modelling and simulation.
Subject coverage:
Modelling and/or simulation across materials science that emphasizes fundamental materials issues advancing the understanding and prediction of material behaviour. Interdisciplinary research that tackles challenging and complex materials problems where the governing phenomena may span different scales of materials behaviour, with an emphasis on the development of quantitative approaches to explain and predict experimental observations. Material processing that advances the fundamental materials science and engineering underpinning the connection between processing and properties. Covering all classes of materials, and mechanical, microstructural, electronic, chemical, biological, and optical properties.