Fast phononic band-structure calculations through a GPU accelerated mixed-variational formulation英文翻译

Fast phononic band-structure calculations through a GPU accelerated mixed-variational formulation英文翻译

ID:40847881

大小:685.08 KB

页数:13页

时间:2019-08-08

Fast phononic band-structure calculations through a GPU accelerated mixed-variational formulation英文翻译_第1页
Fast phononic band-structure calculations through a GPU accelerated mixed-variational formulation英文翻译_第2页
Fast phononic band-structure calculations through a GPU accelerated mixed-variational formulation英文翻译_第3页
Fast phononic band-structure calculations through a GPU accelerated mixed-variational formulation英文翻译_第4页
Fast phononic band-structure calculations through a GPU accelerated mixed-variational formulation英文翻译_第5页
资源描述:

《Fast phononic band-structure calculations through a GPU accelerated mixed-variational formulation英文翻译》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库

1、Fastphononicband-structurecalculationsthroughaGPUacceleratedmixed-variationalformulationAnkitSrivastava∗DepartmentofMechanical,Materials,andAerospaceEngineeringIllinoisInstituteofTechnology,Chicago,IL,60616USA(Dated:July31,2018)InthispaperwepresentaGraphicalProcessingUnit(gp

2、u)acceleratedmixedvariationalfor-mulationforfastphononicband-structurecalculationofarbitrarilycomplexunitcellsandreportspeedgainsofahundredfoldoverunoptimizedserialcpucomputations.Totheauthor’sknowl-edgethisisthefirstapplicationofgpucomputingtoanon-FE/FDTDbandstructurealgorit

3、hm.Theformulationispresentedinaformwhichisapplicableto1-,2-,and3-Dcases.However,inthispaperweconcentrateonoptimizingtheformulationwithintheparadigmofgpucomputing,presentingresultsfor2-Dunitcells.Themixed-variationalformulationhasbeenproventoshowfasterconvergencethanvariation

4、alprinciplesbasedpurelyonthedisplacementfield.Additionallytheintegralnatureofthemethodallowsforitsapplicationtounitcellsofarbitrarycomplexity.However,theapplicationofthisformulationwaslimitedtofairlysimpleunitcellsuntilrecentlywhenitsscopewasextendedtodealwithallBravaislattic

5、esandcomplexmicro-structures.Inthispaperwedescribetheapplicationoftheformulationwithintheframeworkofgpucomputingwithalongtermviewtowardshighlyefficientandmassivelydistributedband-structurealgorithmssuitablefortacklingoptimizationandinverseproblems.Wereportthatthecurrentformula

6、tionbecomesI/Oboundasopposedtocomputeboundindicatingthepotentialforyetfastercomputa-tionsthroughoptimizeddatatransferbetweenthehost(cpu)andthedevice(gpu).Wereportthattheaccurateband-structureevaluationovertheboundaryoftheIrreducibleBrillouinZone(IBZ)forthefirst18phononicbranc

7、hesofacomplex2-Dunitcell(with1132differentphases)canbeachievedinlessthan20secondsonaregulardesktop.Forasimplerunitcell,thefirstfewphononicbranchesarecalculatedinlessthan2.5secondsonthesamesystem.Theschemepresentedinthispaper,therefore,representsconsiderablesavingsinthecomputat

8、ionalcostsofphononic(andphotonic,throughextensionstothecurrentmethod)band-s

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。