Data Science and Visualization
Quickly and confidently getting to the knowledge buried in the mountains of CFD results is critical to success
Over the past two decades, software technologies have advanced greatly and HPC computing has scaled by orders of magnitude. Our human ability to comprehend simulation results has remained the same.
“We only have time and resource to post-process a fraction of the data we produce”
Simply making plots and movies in the traditional way does not scale when dealing with multi-billion cell meshes or ten’s of thousands of timesteps.
Automated solution assessment is accelerated when features and quantitative properties can be produced directly via eigen analysis or machine learning.
Extract-based workflows combined with data science and machine learning can remove barriers to scaling up simulation size and enable high level automation
Am I sure that I can see what’s important?
What faults or gains did I miss because I had to scope down my work to fit my resources?
What can Data science tools reveal?
Are my competitors gaining with new technology that we can’t seem to use?
Data science and visualization
Our enhanced version of DOE’s scalable visualization and post-processing s/w.
Easy-to-use industry standard FieldView desktop coupled to the powerful VisIt Prime backend
Desktop applications and libraries for POD, DMD, Fourier analysis and machine learning for CFD
Customized HPC FieldView
“FieldView CREATE makes it easy to use the largest HPC resources in the DoD enterprise via the VisIt backend, while providing the ease-of-use and high quality graphics that FieldView is famous for. The MetaGuide™ technology lets us combine results from multiple CFD solutions into a single consistent volume without hassles.”