With advanced imaging and satellite technologies, it’s easier than ever to see a galaxy – but understanding how they form (a process that can take billions of years) is a different story. Now, a team of researchers from the University of Arizona has used supercomputing to simulate millions of universes and illuminate the process of galaxy formation.
Studying galaxy formation through simulations has, until now, been a laboriously slow process: individual iterations of different approaches, simulated one at a time. With supercomputing power behind the simulations, however, that changes.
“On the computer, we can create many different universes and compare them to the actual one,” said Peter Behroozi, an assistant professor at the University of Arizona’s Steward Observatory and the study’s lead author. “And that lets us infer which rules lead to the one we see.”
The “rules” Behroozi is referring to are sets of different physical theories for galaxy formation, allowing for simultaneous generation of many competing ideas about how galaxies are formed. This approach – which the researchers call the “UniverseMachine” – simulated over eight million different universes, each containing millions of galaxies and spanning hundreds of millions of years.
In order to run the UniverseMachine, the team turned to computing resources from several institutions, including the NASA Ames Research Center, the Leibniz-Rechenzentrum in Germany, and the University of Arizona’s own “Ocelote” supercomputing cluster. Ocelote consists of 336 CPU nodes (each with two 28-core processing units) and 46 GPU nodes, as well as 2 TB of RAM. In total, the team claims a total of 2,000 processors were utilized over the course of three weeks.
Even with substantial computing power, generating that quantity of data required unprecedented efficiency. “Simulating a single galaxy requires 10 to the 48th computing operations,” said Behroozi. “All computers on Earth combined could not do this in a hundred years. So to just simulate a single galaxy, let alone 12 million, we had to do this differently.” The team’s approach, therefore, used an approach that was able to scale from smaller objects to large sections of the observable universe.
With the data generated, the team was able to compare their millions of “mock universes” to decades of astronomical observations to see which rules resulted in more (or less) realistically simulated universes. The results yielded several surprising insights; crucially, the study resolved the paradox of why galaxies stop forming stars even when hydrogen (the material that comprises stars) is readily available.
Next, the researchers plan to expand UniverseMachine to examine not just how and when galaxies form, but also changes in their individual shapes and morphologies.
About the research
The research discussed in this article was published as “UNIVERSEMACHINE: The correlation between galaxy growth and dark matter halo assembly from z = 0-10” in the September 2019 issue of Monthly Notices of the Royal Astronomical Society. The paper (which can be accessed in full at this link) was written by Peter Behroozi, Risa H. Wechsler, Andrew P. Hearin and Charlie Conroy. Read the University of Arizona announcement here.