Grand Challenge Breakthroughs

Through its Mathematical, Information, and Computational Sciences (MICS) Division, the DOE Office of Advanced Scientific Computing Research (ASCR) sponsors NERSC as a national facility, and also sponsors eight Grand Challenge projects in which NERSC is a partner. Grand Challenge applications address computation-intensive fundamental problems in science and engineering whose solutions can be advanced by applying high performance computing and communications technologies and resources.

The advanced modeling tools developed in the Computational Accelerator Physics Grand Challenge will allow future particle accelerators to be designed with reduced cost and risk as well as improved reliability and efficiency. During the past year, collaborators parallelized the electromagnetics codes, performed high-resolution calculations for the Next Linear Collider and the Spallation Neutron Source, and performed the largest simulations to date for the Accelerator Production of Tritium project.

In the Computational Chemistry of Nuclear Waste Characterization and Processing Grand Challenge, researchers are developing and applying the methods of relativistic quantum chemistry to assist in the understanding and prediction of the chemistry of actinide and lanthanide compounds. Modeling these heavy-element compounds is essential to modeling the fate and transport of nuclear wastes in the environment, as well as evaluating remediation alternatives. Existing codes are being parallelized for the T3E and extended to enable calculations on larger molecules at higher levels of accuracy.

Several members of the Grand Challenge Application on HENP (High Energy and Nuclear Physics) Data successfully ported the CERNLIB physics software to a parallel architecture--a large and complex task that had been attempted several times before but never completed (see further discussion). The project also generated over 1 terabyte (TB) of simulated heavy ion collision data, to be used as a testbed for developing data management and analysis tools and algorithms. A cross-country data transfer experiment, from NERSC in Berkeley to Brookhaven National Laboratory on Long Island, achieved transfer rates of 800–900 kB/sec over brief periods, and sustained an average 200 kB/sec over several days.

The High Performance Computational Engine for the Analysis of Genomes Grand Challenge has developed a prototype web-based framework, The Genome Channel, which shows the current progress of the international genome sequencing effort and allows navigation through the data down to individual sequences and gene annotations. Work in progress also includes developing a CORBA-based analysis framework to facilitate automation of the genome annotation process, developing specialized software and databases for genome analysis, and preparing to utilize the NERSC T3E for production analysis. (Also see the discussion of NERSC's Center for Bioinformatics and Computational Genomics.)

In the Materials, Methods, Microstructure and Magnetism Grand Challenge, researchers are establishing the relationship between magnetism and microstructure based on fundamental physical principles. Understanding this relationship could result in breakthroughs in computer storage as well as power generation and storage, and could enable the design of magnetic materials with specific, well-defined properties. In 1998 the group studied quantum atomic interactions on a scale not previously accessible, and developed a new constrained local moment theory of non-equilibrium states in metallic magnets--in addition to winning the Gordon Bell Prize and breaking the teraflops barrier.

Researchers in the Numerical Tokamak Turbulence Project reported a major advance in the computer modeling of fusion plasmas in the September 18 edition of Science magazine. Using NERSC's Cray T3E for three-dimensional nonlinear particle simulations of microturbulence in the plasma, they performed calculations involving 400 million plasma particles in 5000 time-steps--the first simulations realistic enough to compare with existing experiments. The Cyclone Project, which compared various models for core transport in tokamaks, was also discussed in both Science and Nature.

Detailed simulations of the Standard Model of particle physics, developed in the Particle Physics Phenomenology from Lattice QCD Grand Challenge, will help determine some of the fundamental constants of nature. In the past year, researchers successfully computed the decay amplitudes of kaons for the first time, and successfully reproduced the observed "delta I=1/2" effect, in which seemingly similar decay processes proceed at different rates. They established a publicly available "Gauge Connection" archive (http://qcd.nersc.gov), which provides "unquenched" lattice quantum chromodynamics (QCD) configurations that include virtual quarks. A new algorithm being developed will speed up calculations involving virtual quarks.

In the Protein Dynamics and Biocatalysis Grand Challenge, researchers are working to understand the chemical mechanisms in enzyme catalysis, which are difficult to investigate experimentally. Computer simulations can provide the necessary insights, at an atomic level of detail, for a complete understanding of the relationship between biomolecular dynamics, structure, and function. For example, while the class of enzymes known as beta-lactamases are largely responsible for the increasing resistance of bacteria to antibiotics, the precise chemical resistance mechanism used by this enzyme is still unknown. Simulations are critical for further study of this mechanism.