New Study of Galaxy Evolution: What Turns Young Galaxies On?

September 14, 1999

Tim Stephens
UCSC Public Information Office

SANTA CRUZ, CA: When the universe was about one-tenth of its current age, more than 10 billion years ago, the precursors of massive galaxies such as our own Milky Way were being built from smaller galaxies that collided and merged, triggering violent bursts of star formation. This scenario, one of two competing views of young galaxies proposed by cosmologists, is supported by a recent analysis of supercomputer data led by researchers at the University of California, Santa Cruz.

The new study sheds light on the nature of the most distant galaxies observed by modern telescopes. These galaxies, called “high-redshift galaxies” because their light has been strongly shifted toward longer (redder) wavelengths by the expansion of the universe, have presented a challenge to cosmologists since they were first detected in the early 1990s.

The high-redshift galaxies are very faint because of their great distance from the earth, but astronomers have analyzed the light from them and determined that they actually shine very brightly and are dominated by young, hot stars. They are also small compared to similarly bright galaxies in the nearby universe. These observations indicate a very high rate of star formation, and the challenge is to explain how these young galaxies accumulated enough mass to burn so brightly.

“The mystery is that these very luminous galaxies are so far away that the light we see today left them when the universe was only 10 to 15 percent of its current age, and one wonders how so many bright galaxies had already formed so early in the evolution of the universe,” said Tsafrir Kolatt, a postdoctoral researcher at UCSC and lead author of a paper describing the new findings.

In the paper, which has been posted on the Internet and will be published in the October 1 issue of Astrophysical Journal Letters, Kolatt and his coauthors explain the observed properties of these distant galaxies in terms of the formation and evolution of structure in the early universe. Their results suggest that the high-redshift galaxies offer a glimpse of the universe at a stage when the large-scale structure of the cosmos was still forming.

Coauthors Joel Primack, professor of physics, and Sandra Faber, University Professor of astronomy and astrophysics, both at UCSC, have been collaborating since the 1980s on efforts to understand the formation of galaxies, clusters of galaxies, and other large-scale structures in the universe. The other coauthors include James Bullock and Patrik Jonsson at UCSC; Rachel Sommerville, Yair Sigad, and Avishai Dekel at Hebrew University in Jerusalem; and Andrey Kravtsov and Anatoly Klypin at New Mexico State University (NMSU).

The paper, entitled “Young galaxies: What turns them on?,” evaluates two competing scenarios that have been proposed to explain the origins of high-redshift galaxies. Both scenarios incorporate the standard features of modern cosmology, including a universe dominated by dark matter, mysterious particles that make up at least 90 percent of the mass of the universe. In this standard model of the cosmos, galaxies form within large halos of dark matter.

According to one explanation of high-redshift galaxies, they sit in the middle of very massive dark-matter halos. The gravitational pull of the dark matter feeds gas into the center, where it condenses into stars. Kolatt calls this the “central quiescent” scenario, because it describes a steady rate of star formation proceeding over an extended period of time. For this scenario to work, however, the galaxy must have a huge supply of gas to maintain a high rate of star formation.

The alternative view, championed by Primack and his collaborators, is the “collisional starburst” scenario, in which collisions between small galaxies trigger intense but relatively short-lived bursts of star formation. When two galaxies collide, clouds of gas get funneled toward the center of the larger galaxy, and the gas condenses to form new stars. Through this mechanism, relatively small galaxies can generate very high rates of star formation.

To test these two scenarios, the researchers used supercomputers to simulate the behavior of matter over billions of years in a representative chunk of the universe. The results of the simulation provided strong support for the collisional starburst scenario.

Cosmologists rely heavily on such supercomputer simulations to test their theories of how the universe evolved. These simulations track the interactions of millions of particles of matter, starting with the initial conditions shortly after the Big Bang and proceeding according to the standard laws of physics and the latest theories regarding the nature of dark matter and other critical factors. Each simulation generates a representation of the universe that can be compared with current observations.

The simulation used in this study, developed by Kravtsov and Klypin at NMSU, provided an unprecedented level of resolution, enabling the researchers to analyze the interactions of relatively small clumps of matter and to detect collisions and mergers. Supercomputers were needed not only for the simulation itself but also for the analysis of the output from the simulation. The researchers used computers at the Naval Research Laboratory in Washington, D.C., and the National Center for Supercomputing Applications at the University of Illinois. “This project is right at the edge of what’s possible with the biggest, fastest computers available,” Primack said.

Both the simulation and its analysis, led by Kolatt and Bullock, were innovative, Primack added. “The simulation provided such high resolution that it was possible to find every lump of dark matter in the simulation volume and to tell whether one dark matter halo merged with another; nobody has ever done this before halo by halo, looking at every particle in every halo,” he said.

The researchers then used data from earlier simulations, performed at UCSC, that showed what happens when two galaxies collide. Those results enabled them to calculate the star formation rates that would result from the mergers of dark matter halos and their associated galaxies. “For each individual collision in the simulation, we had to calculate how many stars would form, how long they would shine, and what colors would be observed in the light from them,” Kolatt said.

The results showed that the observed properties of the high-redshift galaxies are best explained by the collisional starburst scenario. “The simulation gave us the right numbers [of high-redshift galaxies], the right clustering properties, and the right brightnesses to explain what we actually see with the telescopes,” Bullock said.

Most of the collisions in the simulation occurred in or near relatively massive dark matter halos, resulting in strong clustering of the starbursts in space. “We believe that what we’re seeing in these high-redshift galaxies is the build-up of structure in the universe, and that this process generates stars with high efficiency,” Bullock said. The central quiescent scenario did not fare so well, because the simulation did not generate enough galaxies of sufficient mass to account for the evolution of the number of high-redshift galaxies observed, Kolatt said.

In addition to presenting results that match current observations, the authors make predictions that can be tested by further observations of high-redshift galaxies. A key test that can differentiate between the two scenarios is how the number of galaxies changes with increasing distance. At higher and higher redshifts, the number of galaxies observed should fall off much faster for the central quiescent model than for collisional starbursts. Another important test will be precise measurements of the masses of high-redshift galaxies, which Faber hopes to obtain within the next year.

“We suspect that we will find the galaxies are small, and that will confirm that these objects are bright but low mass, meaning they must be undergoing a starburst,” Faber said.

Editor’s Notes:

Contact Tsafrir Kolatt in Israel at 011-972-2-658-5492 or

Contact James Bullock at 831-459-4588 or

Contact Joel Primack at 831-459-2580 or

The paper can be downloaded from the Web.

A color image illustrating the results of the supercomputer simulation with high-resolution graphics can be downloaded from the Web at If you need a higher-resolution version, please contact Tim Stephens in the UCSC Public Information Office at 831-459-2495 or

UCO Press Release Archives

Have a press release you'd like to see on these pages? Email it to