Aug 12 2019
How do galaxies like the Milky Way emerge? How do they expand and evolve over time? For several decades, the science underlying galaxy formation has been a mystery. However, a research team led by the University of Arizona has reached a step closer to finding answers to these questions by making use of supercomputer simulations.
A UA-led team of scientists generated millions of different universes on a supercomputer, each of which obeyed different physical theories for how galaxies should form. (Image credit: NASA, ESA, and J. Lotz and the HFF Team/STScI)
The observation of real galaxies in space offers only snapshots in time; therefore, researchers who intend to analyze how galaxies evolve over billions of years have to use computer simulations. Conventionally, astronomers have adopted this method to invent and test new galaxy formation theories, one at a time.
Peter Behroozi, an assistant professor at the UA Steward Observatory, and his colleagues addressed this challenge by creating millions of distinct universes on a supercomputer. Each of the universes obeyed various physical theories for the way galaxies should form.
The study outcomes, reported in the Monthly Notices of the Royal Astronomical Society, contest the basic concepts related to the role played by dark matter in galaxy formation, the way galaxies evolve over time, and the manner in which they give birth to stars.
On the computer, we can create many different universes and compare them to the actual one, and that lets us infer which rules lead to the one we see.
Peter Behroozi, Assistant Professor, UA Steward Observatory
Behroozi is the lead author of the study.
The study is the first to create self-consistent universes that are precise replicas of the real one—computer simulations each representing a sizeable portion of the real cosmos, including 12 million galaxies and spanning the time from 400 million years following the Big Bang to this day.
Each of the “Ex-Machina” universes underwent a range of tests to assess how similar galaxies looked in the created universe when compared to the real universe. All the universes quite similar to our own universe had analogous underlying physical rules, showing a robust new strategy for analyzing the formation of galaxies.
The outcomes of the “UniverseMachine,” as the authors have named their technique, have assisted in finding solutions to the long-standing riddle of why galaxies stop forming new stars even though they hold a lot of hydrogen gas, the basic raw material from which stars are formed.
Common notions related to how stars are formed by galaxies involve an intricate interplay between cold gas that collapses under the impact of gravity into dense pockets that give rise to stars, while other processes contradict the formation of stars.
For instance, it is considered that a majority of the galaxies include supermassive black holes at their centers. The matter that falls into the black holes emits enormous amounts of energy, which act as cosmic blowtorches preventing gas from sufficiently cooling down to collapse into stellar nurseries.
In the same way, stars that end in supernova explosions support this process. In addition, dark matter has a major role to play as it contributes a majority of the gravitational force that acts on a galaxy’s visible matter, attracting cold gas from the surroundings of the galaxy and heating it up in the process.
As we go back earlier and earlier in the universe, we would expect the dark matter to be denser, and therefore the gas to be getting hotter and hotter. This is bad for star formation, so we had thought that many galaxies in the early universe should have stopped forming stars a long time ago. But we found the opposite: galaxies of a given size were more likely to form stars at a higher rate, contrary to the expectation.
Peter Behroozi, Assistant Professor, UA Steward Observatory
According to Behroozi, to match observations of real galaxies, his group had to generate virtual universes where the opposite held true—universes where galaxies continued to form stars for a longer period of time.
If, by contrast, the scientists generated universes based on existing galaxy formation theories—universes where the galaxies stopped giving rise to stars much early—those galaxies seemed quite redder compared to the galaxies seen in the sky.
There are two reasons for galaxies to seem red. The first one is evident in nature and is related to the age of a galaxy—if it was formed earlier in the universe’s history, it will move away more rapidly, thus shifting the light into the red spectrum. This effect is termed redshift by astronomers. The other reason is inherent—if a galaxy stops giving birth to stars, it will consist of fewer blue stars, which usually die out very soon, and have older, redder stars left out.
“But we don’t see that,” stated Behroozi. “If galaxies behaved as we thought and stopped forming stars earlier, our actual universe would be colored all wrong. In other words, we are forced to conclude that galaxies formed stars more efficiently in the early times than we thought. And what this tells us is that the energy created by supermassive black holes and exploding stars is less efficient at stifling star formation than our theories predicted.”
Behroozi further stated that the generation of mock universes of unparalleled complexity necessitated a completely new strategy that was not restricted by memory and computing power, and offered sufficient resolution to span the scales from the “small”—individual objects like supernovae—to a sizeable portion of the observable universe.
“Simulating a single galaxy requires 10 to the 48th computing operations,” he explained. “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.”
Apart from making use of computing resources at NASA Ames Research Center and the Leibniz-Rechenzentrum in Garching, Germany, the researchers used the “Ocelote” supercomputer at the UA High Performance Computing cluster. Over a period of three weeks, 2000 processors broke down the data at the same time. As part of the study, Behroozi and his team created over eight million universes.
We took the past 20 years of astronomical observations and compared them to the millions of mock universes we generated. We pieced together thousands of pieces of information to see which ones matched. Did the universe we created look right? If not, we’d go back and make modifications, and check again.
Peter Behroozi, Assistant Professor, UA Steward Observatory
In order to gain further insights into how galaxies emerged, Behroozi and his team intend to widen the UniverseMachine to incorporate the morphology of individual galaxies and how their shapes change over time.
The paper titled “UNIVERSEMACHINE: The Correlation between Galaxy Growth and Dark Matter Halo Assembly from z = 0-10” has been co-authored by Risa Wechsler from Stanford University, Andrew Hearin from Argonne National Laboratory, and Charlie Conroy from Harvard University. The study was funded by NASA, the National Science Foundation, and the Munich Institute for Astro- and Particle Physics.