On Saturday, Dec. 4, a total eclipse of the sun will occur at 07:33 (Universal Time) over Antarctica and parts of the South Pacific near the southern tip of Chile. The solar corona – visible to the naked eye only during a total eclipse – will be viewable for just over one minute. For southern Californians, that viewing time translates to 11:33 p.m. (Pacific Time), Dec. 3. Thanks to the handiwork of Predictive Science Inc. researchers, who ran simulations on the Expanse supercomputer at the San Diego Supercomputer Center (SDSC) at UC San Diego, we have some idea of what the spectacle might look like.
The researchers’ prediction, posted Nov. 27 – one week before totality, was based on a state-of-the-art computer simulation of the tenuous, magnetized outer atmosphere of the sun known as the solar corona. In addition to providing a sneak peek for eclipse chasers, such predictions aid scientists who are planning eclipse observations from the ground, sea and air. Eclipses also provide a unique opportunity for researchers to test the accuracy and predictive capability of physical models of the solar corona. Such models are essential for understanding how the corona is heated and how it drives the structure and dynamics of the inner-heliosphere, including Earth-affecting disturbances known as space weather.
“Expanse was an essential resource – especially because of its unique hardware architecture and rapid turnaround for mid-scale simulations,” said Cooper Downs, an astrophysicist at Predictive Science Inc. “Although we have allocations on other supercomputers, without Expanse we would have had to start everything several days earlier to make sure all the runs and renders could be completed on time. This would mean the solar observations used to drive the model would be even more out of date, risking the accuracy of the prediction.”
Downs, whose research focuses on understanding thermodynamic and magnetic processes in the solar corona is particularly interested in the improvement and validation of numerical models through direct comparisons to observational data. He said that Expanse allows him to run larger simulations and create more accurate diagnostics.
“Because of the large core and memory count on Expanse nodes, we have the ability to run several cases of various degrees of complexity and size – some quite large – with rapid turnaround, which was essential for getting the final prediction together,” Downs said.
Expanse is SDSC’s newest National Science Foundation (NSF)-funded supercomputer. It is part of NSF’s Extreme Science and Engineering Discovery Environment (XSEDE), which is a virtual organization that integrates and coordinates the sharing of advanced digital services – including supercomputers and high-end visualization and data analysis resources – with researchers nationally to support science. Expanse also supports SDSC’s theme of “Computing without Boundaries” with its data-centric architecture, public cloud integration and state-of-the art GPUs for incorporating experimental facilities and edge computing.