Judging by its starlight and gas content (as seen in the image above), Eris looks to be a near match for our own Milky Way galaxy — except that it exists only as a simulation inside a supercomputer. Until recently, realistic simulations of the Milky Way were not forthcoming. Now astrophysicists at The University of California, Santa Cruz, and the Institute for Theoretical Physics at the University of Zurich have simulated our galaxy better than ever before, all by adjusting how the stars in it form. Their results are due to be published in the Astrophysical Journal, but in the meantime are available on arXiv.

Simulations that have previously tried to recreate a galaxy like the Milky Way have failed to come up with something that matches reality closely enough. The Milky Way is 100,000 light years in diameter and 1,000 light years thick, with sweeping spiral arms connected by a bar leading to a central bulge. The bulge contains mainly older stars, whereas the spiral arms are populated with the hottest young stars. Our solar system lies about two thirds of the way out from the centre of the galaxy in one of the spiral arms known as the Orion-Cygnus Arm. A halo containing globular clusters of old stars surrounds the galaxy in a roughly spherical shape.

Previous simulations tended to create galaxies with bulges that were too large and disks that were too small. Javiera Guedes, a graduate student at UC Santa Cruz at the time of the research, and her supervisor Piero Madau, along with two colleagues from the Institute of Theoretical Physics in Zurich, came up with a simulation in which the bulge and disk were much more like they are in the Milky Way. It is the first simulation that can create galaxies with features that are all consistent with our observations of the Milky Way. They named it Eris.

Key to Eris' success was a more realistic approach to star formation. Star formation in real galaxies occurs in a more clustered way than previous simulations had been able to recreate. Eris has a higher resolution than these previous simulations so was able to model this more complex star formation. It did so by setting the threshold for star formation higher, enabling differences in density to form across the simulated interstellar medium, which is made up of gas and dust. This meant that stars formed only in dense regions of the interstellar medium, creating the clusters that appear when stars form in reality.

Stellar explosions typically blow gas away from their surroundings. Because stars in this simulation were forming in dense regions, when the shorter-lived massive stars exploded, there was more gas to blow away. The gas that was ejected during these explosions never reached the central bulge of the galaxy, meaning that the bulge did not grow as large as it could have — making the simulated galaxy more like the Milky Way, with its relatively small central bulge.

The simulation involved 60 million particles, including all the gas, dark matter and stars in the simulated galaxy. Because of this, the simulation required lots of computer power and time — 1.4 million processor hours on NASA's Pleiades supercomputer, as well as time on supercomputers at the researchers' institutions.

Guedes and her colleagues worked out it was the high resolution of the simulation, and its effects on star formation, that made Eris closer to the Milky Way by running a twin simulation with a lower resolution. Eris' lower resolution twin came up with a galaxy less like the Milky Way and more similar to those in previous simulations.

The success of Eris supports the ΛCDM model of the universe, the simplest model we have that is in agreement with observations. 'Λ' (pronounced 'lambda') refers to the dark energy component of the universe, as this was the symbol Einstein originally used for his cosmological constant — a mathematical appendage he added to his field equations to make them describe a static universe. When astronomers realised the universe wasn't static, the cosmological constant was taken away, but eventually came back as a useful way to describe the dark energy that is forcing the universe into a period of accelerated expansion. The 'CDM' part of the name stands for cold dark matter. 'Cold' because it doesn't move very fast, and 'dark' because it doesn't interact with light, making it nearly impossible to detect. The ΛCDM model combines both dark matter and dark energy to explain the universe, and it works rather well.

Milky Way simulations may not be "just right" yet, but they're certainly getting there.


Javiera Guedes, Simone Callegari, Piero Madau, & Lucio Mayer (2011). Forming Realistic Late-Type Spirals in a LCDM Universe: The Eris

Simulation Astrophysical Journal arXiv: 1103.6030v2