Skip to main content

(EMSL22) Visualization and Active Guidance of Scientific Computations


EMSL Project ID
1794

Abstract

The project will continue to develop effective abstraction signatures of very large time-dependent datasets that capture the essential information needed to characterize and analyze the data. Our approach is powerful and flexible enough to process both scalar, as well as tensor fields, and project them into one signature. This design is particularly useful when dealing with time-dependent datasets with multiple variates (100 plus) such as those generated by combustion and global climate simulations. The two abstraction signature prototypes developed in FY 1999 will be evaluated intensively using the climate simulation data provided by ARM at PNNL. A version of the abstraction signatures will be implemented on a parallel computer to process the much larger combustion dataset obtained from the Combustion Research Facility at Sandia National Laboratory. The project will also investigate a new generation of abstraction signatures that have the potential to further reduce the ! signature sizes and improve the effectiveness of data transmissions of the current implementations. In FY 2001, we would like to apply the wavelet-based signature design to the high-speed network at PNNL and measure the performance and impacts of the multiresolution design.

Project Details

Project type
Capability Research
Start Date
2000-06-02
End Date
2001-12-28
Status
Closed

Team

Principal Investigator

Pak Wong
Institution
Pacific Northwest National Laboratory

Team Members

Elizabeth Jurrus
Institution
Pacific Northwest National Laboratory

Harlan Foote
Institution
Pacific Northwest National Laboratory

Daniel Adams
Institution
Pacific Northwest National Laboratory