{"title":"VIS主题演讲:分析启发的可视化:极端规模的整体现场科学工作流程","authors":"Jacqueline H. Chen","doi":"10.1109/VAST.2017.8585481","DOIUrl":null,"url":null,"abstract":"Combustion and turbulence simulations involve highly intermittent localized phenomena that generate high volumes of spatially and temporally varying field and particle data. The current paradigm of posthoc analysis and visualization will become increasingly infeasible as data volumes continue to increase. In the exascale era this problem will be further exacerbated by the difficulty of moving large volumes of data through deep complex memory hierarchies and across the machine network to hard disks on a heterogeneous supercomputer. I will discuss recent advances in in situ massively parallel volume and particle visualization algorithms coupled with analytics - e.g., topological feature segmentation/tracking, distance field construction, multi-variate statistics and eigensolutions of the reaction rate Jacobian - as an integral part of a scientific discovery from high-fidelity combustion simulations. The role of asynchronous task based programming models and runtimes to facilitate an extensible, performance portable computational science workflow at extreme scale will also be discussed in the context of recent turbulent ignition simulations.","PeriodicalId":168094,"journal":{"name":"IEEE Conference on Visual Analytics Science and Technology","volume":"306 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2017-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"VIS Keynote Address: Analytics Inspired Visualization: a Holistic In-situ Scientific Workflow at Extreme Scale\",\"authors\":\"Jacqueline H. Chen\",\"doi\":\"10.1109/VAST.2017.8585481\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Combustion and turbulence simulations involve highly intermittent localized phenomena that generate high volumes of spatially and temporally varying field and particle data. The current paradigm of posthoc analysis and visualization will become increasingly infeasible as data volumes continue to increase. In the exascale era this problem will be further exacerbated by the difficulty of moving large volumes of data through deep complex memory hierarchies and across the machine network to hard disks on a heterogeneous supercomputer. I will discuss recent advances in in situ massively parallel volume and particle visualization algorithms coupled with analytics - e.g., topological feature segmentation/tracking, distance field construction, multi-variate statistics and eigensolutions of the reaction rate Jacobian - as an integral part of a scientific discovery from high-fidelity combustion simulations. The role of asynchronous task based programming models and runtimes to facilitate an extensible, performance portable computational science workflow at extreme scale will also be discussed in the context of recent turbulent ignition simulations.\",\"PeriodicalId\":168094,\"journal\":{\"name\":\"IEEE Conference on Visual Analytics Science and Technology\",\"volume\":\"306 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2017-10-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"IEEE Conference on Visual Analytics Science and Technology\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/VAST.2017.8585481\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"IEEE Conference on Visual Analytics Science and Technology","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/VAST.2017.8585481","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
VIS Keynote Address: Analytics Inspired Visualization: a Holistic In-situ Scientific Workflow at Extreme Scale
Combustion and turbulence simulations involve highly intermittent localized phenomena that generate high volumes of spatially and temporally varying field and particle data. The current paradigm of posthoc analysis and visualization will become increasingly infeasible as data volumes continue to increase. In the exascale era this problem will be further exacerbated by the difficulty of moving large volumes of data through deep complex memory hierarchies and across the machine network to hard disks on a heterogeneous supercomputer. I will discuss recent advances in in situ massively parallel volume and particle visualization algorithms coupled with analytics - e.g., topological feature segmentation/tracking, distance field construction, multi-variate statistics and eigensolutions of the reaction rate Jacobian - as an integral part of a scientific discovery from high-fidelity combustion simulations. The role of asynchronous task based programming models and runtimes to facilitate an extensible, performance portable computational science workflow at extreme scale will also be discussed in the context of recent turbulent ignition simulations.