{"title":"An Efficient Strategy for Distributing the Mesh of Parallel Electromagnetic Solvers Based on the AIM","authors":"Damian Marek, Shashwat Sharma, P. Triverio","doi":"10.23919/eucap53622.2022.9768912","DOIUrl":null,"url":null,"abstract":"Parallel electromagnetic solvers require efficient strategies to distribute the mesh of large structures encountered in advanced antenna designs. In addition, parallel solvers based on the adaptive integral method (AIM) require processes to access additional mesh entities that lie within the near region of their mesh partition. In this paper, we propose a parallel solution for loading, processing, partitioning, and redistributing the mesh of AIM-based solvers. The proposed method is demonstrated on problems as large as 200 wavelengths, with up to 200 million unknowns, and up to 8000 cores involved.","PeriodicalId":228461,"journal":{"name":"2022 16th European Conference on Antennas and Propagation (EuCAP)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2022-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2022 16th European Conference on Antennas and Propagation (EuCAP)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.23919/eucap53622.2022.9768912","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1
Abstract
Parallel electromagnetic solvers require efficient strategies to distribute the mesh of large structures encountered in advanced antenna designs. In addition, parallel solvers based on the adaptive integral method (AIM) require processes to access additional mesh entities that lie within the near region of their mesh partition. In this paper, we propose a parallel solution for loading, processing, partitioning, and redistributing the mesh of AIM-based solvers. The proposed method is demonstrated on problems as large as 200 wavelengths, with up to 200 million unknowns, and up to 8000 cores involved.