Scientists are developing computers that will be 50x faster than today’s by the 2020s, which will enable the first complete simulations of plasma.
The Princeton Plasma Physics Laboratory (PPPL) leads a team of institutions that is developing a computer program that will run on the next generation of supercomputers to bring fusion energy closer to reality. The team is part of the Exascale Computing Project (ECP), a major component of President Obama’s National Strategic Computing Initiative to ensure continued U.S. leadership in high-performance computing.
Leading the initiative are the U.S. Department of Energy (DOE), the Department of Defense and the National Science Foundation. Within the initiative, DOE has the lead on exascale. Once developed, exascale computers will perform a billion billion operations per second, a rate about 50 times faster than the most powerful supercomputer in the U.S. today. Exascale machines are expected to be ready in the United States in the mid-2020s.
The PPPL-led program will develop the first complete simulation of the superhot gas called plasma that fuels fusion reactions. Such simulations will enable physicists to predict how the plasma will behave in fusion facilities and could lead to well-engineered reactors that “put a star in a jar” to produce safe, clean and abundant energy for generating electricity. Plasma, made up of free-floating electrons and atomic nuclei, comprises 99 percent of the visible universe and often is called the fourth state of matter.
Fusion of light elements is the process the sun and stars use to create energy, and scientists need this massive computing power to help simulate the plasma behavior and recreate these reactions here on Earth. If successful, fusion could provide energy for all humankind essentially for millions of years.
Directing the PPPL-led, four-year project is Amitava Bhattacharjee, head of the Theory Department at PPPL and Professor of Astrophysical Sciences at Princeton University. Co-principal investigators are PPPL physicist C.S. Chang and Andrew Siegel, a computational scientist at the University of Chicago.
Joining the highly complex project are the Argonne, Lawrence Livermore and Oak Ridge National Laboratories, together with Rutgers University, the University of California, Los Angeles (UCLA), and the University of Colorado, Boulder.
The participants will work to combine a computer code called XGC, which models behavior at the edge of the plasma, with a code called GENE, which models behavior of the plasma core. Combining the two high-performance codes — a feat never before attempted — will provide a holistic view of the entire plasma volume.
The two codes have been years in the making. Chang and his group developed and maintain the XGC program; Frank Jenko at UCLA created and maintains GENE.
The code is so complex that it uses 90 percent of the capacity of Titan, the fastest supercomputer in the United States. Titan performs at the petascale and completes a million billion calculations per second at the Oak Ridge Leadership Computing Facility, a DOE Office of Science User Facility at Oak Ridge National Laboratory.
“Taking into account all the physics in a fusion plasma requires enormous computational resources,” Bhattacharjee said. “With the computer codes we have now, we are already pushing on the edge of the petascale. The exascale is very much needed in order for us to have greater realism and truly predictive capacity.”
Watch the video above to view a high-performance simulation of the development of blobby turbulence at the edge of fusion plasma. This simulation was produced on Titan with the XGC code.