SLAC Advanced Computations Group Earns Supercomputing Time
A group of SLAC National Accelerator Laboratory scientists will have another year and another 12,000,000 processor hours on the supercomputer at Oak Ridge National Laboratory, to perform research that could greatly improve the next generation of linear colliders. The Department of Energy Office of Advanced Science Computing Research announced Tuesday that 69 research programs have been awarded supercomputing time as part of its Innovative and Novel Computational Impact on Theory and Experiment program, or INCITE, which supports computationally intensive, large-scale research projects. The SLAC team, led by computer scientist Lie-Quan Lee, submitted a plan in 2008 for a three year study, requiring approvals from the DOE annually. This is the second year that they have been awarded time for the project.
"Time on a supercomputer will let us do a large-scale simulation that we cannot do anywhere else," said Lee, who acted as Principle Investigator on the INCITE application. "A supercomputer lets us solve much bigger problems, meaning we get much higher resolution in the answer, and it can give us the result much faster."
Lee is a member of SLAC's Advanced Computations Group, which includes computer scientists and accelerator physicists. The group plans to use the supercomputing time to model and optimize the design of linear colliders. Namely, they will study electromagnetic wakefields: unintended effects that occur inside linear accelerators. As bunches of electrons or positrons cruise through the accelerator's pipeline of connected cavities, called cryomodules, the particles' charges kick up an electromagnetic wake, like a speeding boat that stirs up a wake in the water. These wakefields can interfere with the next particle bunch and degrade its quality, or even cause damage to the cavity.
The SLAC group will model the behavior of bunches of electrons traveling through an eight meter long cryomodule, consisting of a series of smaller cavities. While no computer is large enough to model an entire accelerator in sufficient detail to be useful, the Cray XT5 "Jaguar" supercomputer at Oak Ridge National Laboratory, which can perform more than two thousand trillion calculations per second—about one million times more powerful than your average desktop—is powerful enough to model individual cryomodules. These simulations can reveal how the wakefields behave and how they might be controlled.
"The problem we are trying to solve is a very large problem for accelerators," said Cho-Kuen Ng, one of the group's accelerator scientists. Their work promises to impact plans for the International Linear Collider and the Compact Linear Collider, both proposed electron–positron colliders for the high energy physics community. By better understanding electromagnetic wakefield effects, accelerator scientists could improve performance, increase reliability and lower costs in future linear colliders.
Creating a computer model on a supercomputer is "an entirely different problem," than creating one on a smaller computer grid, Lee said. To take on this challenge, and maximize their computing time, the Advanced Computing Group will work with computer scientists from Lawrence Berkeley, Argonne and Oak Ridge National Laboratories. Technical and scientific assistance from scientists at the computing facilities is part of the INCITE program.
INCITE was launched in 2003 by Under Secretary for Science Raymond Orbach to "seek out computationally intensive, large-scale research projects with the potential to significantly advance key areas in science and engineering," according to the ASCR Web site. In 2010, more than one billion computing hours will be provided to INCITE awardees at Oakridge and Argonne.
"INCITE allocates computer time to solve the most challenging problems in science," said Lee. "I think the INCITE program is trying to enable not just the DOE labs but the whole community, including industry and universities, with these large-scale simulations that can lead to great scientific discoveries or solve tough engineering problems."