Simulation / Modeling / Design

Supercomputer Helps Understand How Jupiter Evolved

Researchers from ETH Zürich and the Universities of Zürich and Bern simulated different scenarios on the computing power of the GPU-accelerated Swiss National Supercomputing Centre (CSCS) to find out how young giant planets exactly form and evolve.
“We pushed our simulations to the limits in terms of the complexity of the physics added to the models,” said Judit Szulágyi, professor at the University of Zürich, “and we achieved higher resolution than anybody before.”
Using the Tesla GPU-accelerated Piz Dain supercomputer, the researchers found a big difference between the two formation mechanisms: In the disk instability scenario, the gas in the planet’s vicinity remained very cold, around 50 Kelvins, whereas in the core accretion case the circumplanetary disk was heated to several hundreds of Kelvins.
“The disk instability simulations are the first that can resolve the circumplanetary disk around multiple protoplanets, using tens of millions of resolution elements in the computational domain. We exploited Piz Daint to accelerate the calculations using GPUs,” adds Lucio Mayer, Professor at the University of Zürich.
In their simulations, the astrophysicists mimicked the formation processes by using the basic physical laws such as gravity or the hydrodynamical equations of the gas – because of the complexity of the physical models, the simulations were very time consuming, even on Europe’s fastest supercomputer at CSCS, “On the order of nine months running time on hundreds to several thousands of computing cores” estimates Judit Szulágyi, post-doctoral fellow at the ETH Zürich. “This means that on one computing core, it would have taken longer than my entire lifetime.”
Read more >

Discuss (0)