You can’t understand the universe unless you understand the galaxies that comprise it. Galaxies are home to stars, black holes, dark matter halos and clouds of gas and dust. But insights into these galactic phenomena aren’t easily obtained. Astrophysicists have long struggled to develop a complete view of galaxies that spans the entire cosmic scale, from single stars to superclusters of thousands of galaxies.
Simulating Multiscale Astrophysics to Understand Galaxies, or SMAUG, is a five-year collaboration hosted at the Flatiron Institute’s Center for Computational Astrophysics (CCA) that aims to better simulate the evolution of galaxies across all astronomical scales. The collaboration, named for the gold-hoarding dragon from The Hobbit, is co-directed by Rachel Somerville, the CCA group leader for galaxy formation, and CCA visiting scholar Greg Bryan. The Tolkienian title and the collaboration’s logo were inspired by the ouroboros, an ancient symbol depicting a snake or dragon eating its own tail and representing the interconnectivity of all scales of the universe.
SMAUG team members include dozens of researchers from the United States and Europe with expertise across different areas of galaxy formation and evolution. One reason for starting SMAUG was to bridge the gap between different research areas within the CCA and beyond, says Somerville. “We wanted to think about a way to get people engaged with one another and to get people from different backgrounds and groups to talk to each other.”
The collaboration’s eventual goal is to probe the fundamental physics that underlies the cosmos, such as the nature of the mysterious dark energy and dark matter that make up most of the universe. Normally, astrophysicists develop theories about the universe and then test them by running simulations that predict what the cosmos should look like if a given set of conditions is true, such as if dark energy changes over time. They can then compare those results with observations and see if the simulations match reality. However, current galaxy simulations are limited to those involving large-scale phenomena; they don’t directly model ‘small-scale’ physics, such as star formation and feedback from supermassive black holes. Instead, the simulations rely on empirical ‘recipes’ used to mimic those small-scale processes. The recipes are tweaked to match previously recorded observations. This approach undercuts the potential for insights, Somerville says, because it only reveals part of the story.
“We need to understand the physics of individual stars and black holes, which are less than a light-year across, all the way up to the hundreds of thousands of light-years that might span a single galaxy, all the way up to billions of light-years to the cosmological scale,” says Somerville. “That, in a nutshell, is the goal of SMAUG: to develop this new but physically grounded approach for treating how all these different physical processes interact so that we can interpret the observational probes used for cosmology.”
Recently, SMAUG presented the first results from the collaboration in a series of six papers posted on arXiv.org. They covered everything from galactic winds to supernovae to head-to-head comparisons of different galaxy simulations. “These papers would not have happened without this collaboration,” says Bryan, who is a professor of astronomy at Columbia University. “The takeaway is that this process of bringing people together from disparate expertise and different areas so that they can create something bigger than the individual parts really works.”
Galactic Winds. Supernova explosions can blast gas out of a galaxy. Chang-Goo Kim, a joint postdoctoral researcher at the CCA and Princeton University, and colleagues investigated the properties of this ejected gas by developing a high-resolution model of a small portion of a galaxy’s disk. In a new paper, the researchers report that the winds driven by these explosions are a mixture of cold, dense gas and hot, low-density gas. These two kinds of gas behave very differently, with much of the cold material quickly falling back into the galaxy, while much of the hot material escapes from the galaxy. The findings will form the basis for building galaxy simulations that accurately model how gaseous outflows affect processes such as star formation as galaxies age.
Supernova Simulations. When stars go supernova, they blast metals and heat into their surroundings. But because cosmological simulations currently lack the resolution to model individual supernovae, they wrongly treat this debris as uniform across space. In a pair of papers, CCA Flatiron Research Fellow Miao Li and colleagues simulated the varied effects of supernovae of old stars. Their work showed that patches of gas in a galaxy would escape the heating of supernovae and form cool, dense clumps that might form stars or fall into the supermassive black hole at the center of the galaxy.
Galactic Gas. Astrophysicists now recognize that the diffuse gas surrounding a galaxy plays a crucial role in that galaxy’s formation and evolution. Drummond Fielding, a Flatiron research fellow at the CCA, and colleagues found that the properties of the inner part of this circumgalactic gas are heavily dependent on outflows from the galaxy. In contrast, the outer part is influenced more by material flowing in from the space between galaxies. Moreover, they highlighted essential differences in the properties of this circumgalactic gas predicted by their simulations versus simulations that used other techniques. Understanding these differences is critical to understanding a galaxy’s health, as gas-starved galaxies can’t undergo star formation.
Star Birthplaces. In another paper from the SMAUG collection, Bhawna Motwani, a graduate student at Columbia University, analyzed large-scale cosmological simulations of many galaxies to study the conditions in which stars can form. The work will inform the conditions used as inputs for future higher-resolution simulations.
Modeling Competition. When it comes to simulations, a universal worry among astrophysicists is how much computational firepower to devote to a given problem. A simple model may run faster but give less accurate results. A more complete model may provide a more detailed picture but take up more time and resources. Another paper, led by Viraj Pandya, a graduate student at the University of California, Santa Cruz, looked at simpler but much more computationally efficient models and compared them with computationally expensive, high-resolution simulations of individual galaxies. The simulated galaxies ranged in mass from dwarf galaxies to Milky Way-mass objects. Pandya and colleagues found that the simpler models were lacking in a few key aspects, and they proposed ways the models can be improved to enable better galaxy simulations that require fewer resources.