Supernovae are perhaps the galaxy’s best fireworks shows, with dying stars’ death rattles coming in the form of unimaginably large explosions. Astrophysicists have been working to better understand how exactly these explosions happen for decades, and now, researchers at Lawrence Berkeley National Laboratory’s National Energy Research Scientific Computing Center (NERSC) are applying supercomputing to crack the conundrum.
“This problem has been around for 60 years, and it parallels developments in physics and the availability of supercomputers,” said Adam Burrows, a professor of astrophysics at Princeton University and co-author of the paper, in an interview with NERSC’s Elizabeth Ball. Indeed, supernova simulations had been limited to just a couple of dimensions until very recently (“With more dimensions comes more complexity,” Burrows said), and even those simulations often took months to complete.
But modern high-performance computing installations, such as those at NERSC, offer far fewer limitations. “NERSC is home to the first machine where we could get 3D going with the code we’re using now,” Burrows said. “We developed the code on NERSC’s Cori system, using regular default time, and it provided the foundation we needed. It’s very well suited to what we’re doing.”
The Cori system Burrows referred to is a Cray XC40 system consisting of 2,388 Intel Xeon Haswell nodes and 9,688 Intel Xeon Phi Knights Landing nodes. All told, the system is capable of around 14 Linpack petaflops, placing it 20th on the most recent Top500 list of the world’s most powerful publicly ranked supercomputers.
Cori’s capabilities allowed the team to run a series of simulations within a year – a task that would have been nigh-impossible for prior systems, and which also proved immensely valuable. “What you want is to be able to make a lot of mistakes, fast,” Burrows said.
The simulations illuminated the process of supernova formation: a white dwarf star emerges from another star, reaches a mass at which it will not implode, and then, suddenly, its core explodes, sending out a shockwave. Crucially, the simulations revealed that this shockwave itself does not trigger the supernova – rather, the supernova is caused by the asymmetry of the “mother” star and other various forces resulting from the collapse of the core.
Next, Burrows hopes to expand these simulations to study finer details, such as the influences of magnetic fields, that were previously even further out of reach.
“This study was an international effort, and it builds on the work of many people over the decades,” Burrows said. “We’re in a time when the stars have aligned.”
To read the reporting from NERSC’s Elizabeth Ball, click here.