使用可视化代理创建用于科学模拟的数字孪生.pptx

1、,Towards the creation of digital twins for scientific simulations using visualization surrogates,Han-Wei ShenDepartment of Computer Science and EngineeringThe Ohio State University,1,Scientific Simulation and Visualization,2,Machine Learning for Simulations,Visualization surrogates for ensemble simu

2、lationsEmbedding and latent representation(data,features,and images)Super-resolution and reconstruction(features,data)Data(dimensionality)reduction,compression,and transformationRendering and visualization algorithmsClassification(feature,data)Visual design and layout,Machine Learning for Visualizat

3、ion,Visualization surrogates for ensemble simulationsEmbedding and latent representation(data,features,and images)Super-resolution and reconstruction(features,data)Data(dimensionality)reduction,compression,and transformationRendering and visualization algorithmsClassification(feature,data)Visual des

4、ign and layout,Ensemble Simulations and Visualization,Ensemble data analysis workflowIssuesComputation,I/O,and storage overhead,Ensemble Simulations and Visualization,In situ visualizationGenerating visualizations at simulation timeStoring images for post-hoc analysis,Visualization Surrogates for En

5、semble Simulations,Traditional methods still take too much time to run ensemble simulations with different parameters,and not clear what parameters to useVisualization Surrogate:allow scientists to preview the simulation output without even running the simulationsIntegrate with in situ processing to

6、 better utilize the collected images/dataUltimate goals:Identify important simulation parameters Identify simulation parameter sensitivity Quantify the uncertainty of the simulation models,InSituNet-Visualization Surrogate,=,A deep neural network that models the function,InSituNet an image space vis