January 2007
Return to this issue’s stories

Award Winning Data Analysis Toolkit Released

It is difficult for computers to understand a simple thing like a tear in a sheet of metal. Standard data visualization techniques do not do the job, either, as material failures like tears may happen anywhere in a simulation. With massive datasets, the problem becomes much worse. There could be thousands of features that an analyst is interested in. How can we harness computation to help with this problem?

To address these needs, Sandia has developed “FCLib,” a data analysis toolkit constructed to meet the needs of data discovery in large-scale, spatio-temporal data. Funded via the ASC Pre- and Post- Processing Environment, FCLib is a C library toolkit of building blocks that can be used to assemble analyses for rich, quantitative, and convenient post-processing.

FCLib “characterizations” (codes that automate extraction of information from ASC simulation data) have supported a number of ASC analyses. In particular, FCLib analysis and staff were among the winners of the Gold Level Sandia President’s Quality Award bestowed on the “Validation, Verification, and Quantified Margins and Uncertainties for Modeling and Simulation of W80-3 Handling Drops” project.

An example of FCLib post-processing analysis is the detection and characterization of tears. The figure on the left, below, illustrates the appearance of a tear in a weapon drop experiment. Even in close-up, it can be difficult to detect, and even harder to quantify. The figure on the right is the result of an analysis built with the FCLib code. Not only does the characterization place labeled bounding boxes to aid visual inspection, but it also extracts and records quantitative measures of the severity of the tear, such as total volume, length, and maximum width of the tear.

tears visual 1Tears analyzed

FCLib is open source software available under a Berkeley Software Distribution (BSD)-like license. It is available at http://fclib.ca.sandia.gov