Hatching the Milky Way

If you can create a baby in a test tube, why not an entire galaxy in a computer? For the first time, astronomers have modeled a mix of gas, stars, and dark matter that came together to give birth to a galaxy resembling our Milky Way. The ambitious simulation promises to yield insight into the galaxy's origin and evolution.

"It was a very high-risk simulation," says Javiera Guedes of the Swiss Federal Institute of Technology in Zurich, an astronomer whose team spent 9 months of computing time before knowing whether the simulation would work. "A lot of people didn't believe that simulations of this type could be done," she says. With hundreds of billions of stars, the Milky Way is so complex that for decades scientists have tried and failed to concoct a galaxy like our own. Because of the controversy, Guedes's team dubbed the simulation "Eris" for the Greek goddess of strife and discord.

Computer simulations such as Eris try to "grow" a galaxy by starting with particles that represent various parts of the galaxy as they move under the influence of one another's gravity and follow other laws of physics. Because of the great number of particles and the complex interactions among them, the simulations require supercomputers to perform the calculations.

Previous simulations failed to create spiral galaxies like our own: Their central regions, or bulges, were too large and their disks were too small. In contrast, the Eris simulation, which Guedes and her colleagues describe in an upcoming issue of The Astrophysical Journal, succeeded in producing a spiral galaxy with a small bulge and large disk, like the Milky Way. Furthermore, the galaxy matched the Milky Way's gas properties, stellar content, and rotational pattern.

The key to the simulation's success lay in its high resolution: It tracked more than 60 million particles that represented gas, stars, and dark matter. The simulation mimicked both star birth and star death. Stars arise in dense clouds of gas and dust. Because of its high resolution, the simulation followed the gas in detail and formed stars only where the gas grew densest, which is approximately how stars form in real galaxies.

Moreover, when short-lived massive stars exploded, the stars were still embedded in that gas, so the explosions blew the gas away. The explosions tended to push the gas away from the galaxy's center, so less gas sank into the central regions to create stars there. As a result, the galaxy's bulge did not grow as large as it had in earlier simulations that failed to resemble the Milky Way.

Now that astronomers have succeeded in modeling the Milky Way, they can use this and future simulations to investigate how the galaxy assembled itself. For example, this simulation assumed that the Milky Way has led a fairly quiet life, suffering no large collisions with other galaxies over the past 11 billion years. But future simulations, with a more violent merger history, might—or might not—better match the actual Milky Way, lending clues as to how much drama our galaxy has experienced.

"This is a very good piece of work," says Leo Blitz, an astronomer at the University of California, Berkeley, who was not part of the new study. "Is it an exact duplicate of the Milky Way? The answer is no. But it is such a significant step forward from what's been done before [that] I think that doesn't really matter."

"It is obviously a tour de force of computational art," adds astronomer Timothy Beers of Michigan State University in East Lansing. He thinks future models should explore how the galaxy's stars created chemical elements such as oxygen and iron. This chemical evolution, he says, provides an additional test that any successful model of the Milky Way must pass.