Computational Astrophysics Goes to Capitol Hill
School of Physics’ John Wise Roots for More Supercomputing
It isn’t every day that a briefing on Capitol Hill would have been appropriately accompanied by popcorn. May 25, 2016, was one of those rare days, with the screening of “Solar Superstorms,” a documentary commissioned by the National Science Foundation (NSF). Narrated by popular British actor Benedict Cumberbatch, the documentary served as the entrée to space weather, why it should be monitored, and what is needed to study it.
Helping explain the matter to a standing-room-only crowd of about 100 Congressional staffers and interns was an NSF-convened panel of the science advisers of the documentary. Among them was John H. Wise, an associate professor in the School of Physics.
Wise is a computational astrophysicist, using supercomputers to simulate the birth of stars and galaxies and to visualize those simulations. A segment of “Solar Superstorms” is based on the simulations of his research at Georgia Tech College of Sciences. The same tools are needed to study, track, and forecast space weather, especially that occurring on the sun.
Solar outbursts are common; the high-energy particles they send to Earth cause the bright dancing lights we know as auroras. At times, these outbursts can be so powerful that they can disrupt the electricity grid or damage satellites. “The largest solar superstorm that affected Earth after the advent of electricity occurred in the 1850s, when we had only telegraphs,” Wise notes. “Imagine what can happen now with the modern communications we have.”
The potential cost of solar superstorms can run to trillions of dollars in lost productivity, according to Senator Gary C. Peters (D-Michigan). Last month, he introduced a bill – S.2817, Space Weather and Forecasting Act – to ensure that the U.S. has the tools and resources to predict space weather and avoid economic catastrophe in case of severe events.
A major tool in this endeavor is computational power, Wise says. Using the example of his research at Georgia Tech, he posits that in these days of big data, computation has become the third leg of science, alongside theory and experiment. “Computation is central to science now,” Wise says. “The US has been a leader globally in providing supercomputers for scientists to use and to train computational scientists.” Continued support for the funding of supercomputers for basic research is essential, he adds.
In his research Wise uses supercomputers to run simulations of galaxy formation, beginning from the first generation of stars forming when the universe was only 100 million years old. Running these simulations requires tremendous computational power. At Georgia Tech, Wise has access to a supercomputer with a computing power equivalent to 30,000 home computers running at the same time. A computation that would take 100 days on a single computer would take less than 5 minutes using the entire Georgia Tech facility. But even that is good only for software development, Wise says.
To do simulations and visualizations, Wise goes to Blue Waters, NSF’s premiere supercomputer, which is housed in the University of Illinois, Urbana-Champaign. Blue Waters is equivalent to 400,000 home computers. Wise has NSF grants that enable him to access this facility. “We are doing computations at scale,” Wise says, adding that his research can produce up to 1,000 terabytes of data in a single simulation.
Yummymath.com estimates that the Complete Works of Shakespeare, consisting of 1,300 pages of print, would require 10 megabytes of storage. Wise’s output from one simulation would be equivalent to 100 million volumes of the Shakespeare tome.
Because no direct observation of the first galaxies being born exists, “we are in a theorists’ playground,” Wise says. “Our work is planning for the future,” he explains, when the James Webb Space Telescope is launched in October 2018. This successor to the Hubble Space Telescope ought to be able to look deeper into space and see galaxies even farther away and younger than have been observed so far. “We’re trying to make predictions for the new space telescope,” says Wise. When that telescope is working, Wise will be able to compare his team’s simulations to observation.
For now, the number crunching continues at a grand scale, both to observe space weather events that could affect Earth now and to predict events that occurred long ago and far away. Scientists’ continued access to computational power, Wise says, is crucial.