ONE of the most challenging scientific tasks is simulating, in three dimensions, the complex physics of a nuclear explosion. Such simulations are now possible for the first time, thanks to the massively parallel supercomputers of the Department of Energy's Accelerated Strategic Computing Initiative (ASCI).|
However, that possibility presents yet another major challenge: understanding the enormous amounts of data generated by the supercomputers and the advanced codes running on them. As described in the article beginning on p. 4, personnel from Livermore's Computation Directorate have taken on the task of developing new tools for managing, sharing, and comprehending three-dimensional ASCI simulation data. Much of the effort in this activity is focused on data and visualization corridors. These corridors represent paths through which data rapidly flow, permitting users to easily interact with each other and with their data. In this way, scientists can readily analyze and understand the simulations and their implications for the nation's nuclear stockpile.
Constructing data and visualization corridors requires changes in networking, systems architecture, large-scale data management, and the way in which people interface with computers. A major focus of these changes is on achieving better ways to transform raw data into richly detailed graphical representations from which the human brain can find patterns, spot anomalies, and draw inferences.
Until recently, the development of advanced visualization tools was a modest effort scattered across many organizations nationwide. What's more, few commercial products were available to handle the data generated by computers the size of the ASCI systems. As a result, the DOE national laboratories have embarked upon a program to develop new data manipulation and visualization systems.
As part of the overall DOE effort, Livermore computer scientists are pioneering new approaches to computing, information handling and storing, and communications-all requiring advances in both software and hardware. Among their successes to date are the giant Assessment Theater housed in one of Livermore's building complexes, the large flat-panel displays now being installed in scientists' offices, dedicated visualization servers, and software programs that allow users to sift through reams of data and focus on areas of greatest interest.
Lawrence Livermore's effort involves collaborations with other national laboratories, U.S. industry, University of California campuses, and other universities. Perhaps the most important collaboration, however, is with the weapon scientists who are the end users of the new tools.
Although the work centers on developing tools for the nuclear weapons program, one can expect the advances to benefit other fields of scientific inquiry that use supercomputers for large-scale simulations or for analyzing large amounts of experimental data. Understanding protein folding and its effects on drug interactions, modeling long-term global climate change, modeling materials, designing for automobile and aircraft safety-all could be uses for the new supercomputing tools. Clearly, advanced visualization techniques can assist a host of major scientific disciplines and challenges.