Cosmic Collisions

by Jonathan Knight

      To a stargazer, the universe appears as placid and calm as a frozen lake. But in reality, it is a violent, churning cauldron in which galaxies explode into existence, then collide and rip each other apart, flinging debris across the cosmos. Astronomers would love to see this violence, but all they get are snapshots of galaxies in various stages of impact. To see collisions in motion, they would need a time-lapse camera that could compress a billion years into an hour.

      Lars Hernquist, an astrophysicist at the University of California, Santa Cruz, has devised the next best thing: supercomputer simulations of cosmic encounters. Hernquist's computer models have helped astronomers understand the origin of starbursts, as well as estimate the quantity of dark matter that surrounds and molds galaxies. Recently, Hernquist has probed the evolution of disc galaxies like our own from their humble beginnings as diffuse interstellar gas. His models have changed the way astronomers think about the universe.

      In the early 1970s, Alar Toomre set a collision in motion with the first computer model of a galactic encounter. Limited computing power kept the number of stars to 1,000 (real galaxies have hundreds of billions). Toomre assigned each star a mass and sent the two digital galaxies hurtling toward each other.

      The long streaks of stars called "tidal tails" that gravity spun off Toomre's galaxies matched the features of collisions seen by telescope. But Toomre's models were too crude to probe other complexities of galactic encounters such as what happens to the interstellar gas and how might dark matter affect the impact. These questions had to await the vast number-crunching power of supercomputers, still a decade away.

      Although the computer simulations of galactic encounters appear explosive, none of the stars actually meet. Stars are so small compared to the distances in between them, that the chance of a ahead-on collision between suns is practically nil.

      But they don't escape unscathed. As they fly by, their combined masses throw up a huge gravitational "tide." "It both squishes and stretches at the same time," says astrophysicist John Dubinski of the University of Toronto. "The material at the edges of the galaxies flies off like a slingshot." In a mere 100 million years, what was once a smooth disc ends up looking like a spinning lawn sprinkler.

      In his office, Hernquist unrolls a poster that could be an ad for a new Star Trek movie. Above the words "Cosmic Voyage," two galaxies crash into each other in Technicolor. It's part of a simulation requiring over 1000 hours on a Cray supercomputer and including more than a million stars. The resulting one-minute video is so spectacular, it has become part of a new half-hour movie produced by the Smithsonian Institution for IMAX theaters, auditoriums in which giant screens immerse the viewer in light and sound.

      "Cosmic Voyage" views the universe with a zoom lens aimed at everything from the largest known structures in the cosmos to sub-atomic particles.

      "It's very nice," Hernquist says of the film. With clasped hands resting gently on his desk, this mild-mannered man speaks about his work quietly and in monotone. But when he speaks, people listen, says Johns Hopkins astronomer Chris Mihos, who collaborated with Hernquist on the Cosmic Voyage simulation. "He is one of the few people in astronomy who doesn't have to talk loudly to get his point across."

      Most of the mass in the universe exists as "dark matter." We can't see dark matter directly, because unlike stars and quasars, it emits little or no electromagnetic energy. But dark matter makes its presence felt by its gravity, whose effects on stars and galaxies we can see from here on Earth. Dark matter may consist of subatomic particles or burnt out husks of stars, no one knows for sure.

      But the question of cosmic importance is how much dark matter is there? If the amount is small, the cosmos will keep expanding forever, but if the amount is large, its gravity will reverse the expansion, ultimately compressing the universe back down to a single point.

      Lars Hernquist's computer simulations suggest that less dark matter hovers around galaxies than some have predicted. He and Toronto's Dubinski find that the size of the tidal tails in a simulated collision varies drastically depending on how much dark matter they put in. If a model galaxy has too much dark matter, the gorgeous arching tails seen in the telescope can't form, suggesting hat the amount of dark matter in galaxies is small. Perhaps the universe has a future after all.

      In 1978, two Yale astronomers, Richard Larson and Beatrice Tinsley, discovered that some interacting galaxies are oddly blue. Blue hues usually emit from newly formed stars, so Larson and Tinsley called these galaxies starbursts. Soon astronomers had found hundreds of starbursts, some of which were churning out a thousand new stars a year.

      Now, computer simulations have shown how a collision can spark a star factory. As two galaxies of unequal size meet, the gravitational tides induce a bar-shaped feature across the center of the bigger galaxy. The bar forms shock waves which knock the interstellar gas down to a lower orbit, compressing it. Squeezed tightly together, the gas molecules fuse and explode like hydrogen bombs all across the galactic core. The starburst lasts for hundreds of millions of years.

      The Lyman-alpha forest is a distant field of gas that holds clues to the evolution of galaxies. Astronomers first detected the gas nearly 30 years ago as a forest of black bands in the spectrum of light from quasars. They interpreted the many bands as a field of hydrogen clouds through which the quasars were shining like flashlights. The hydrogen was absorbing certain wavelengths of light, casting a shadow on the spectrum. Because of the time it takes the light to reach the earth, these shadows were cast when the universe was half the age it is now, around seven billion years ago, before most galaxies were born.

      But the cloud model of the Lyman-alpha forest fails to account for most of the matter that the Big Bang theory says must exist. Furthermore, the clouds would not contain enough mass to ignite new galaxies, the ultimate fate of much of the primordial interstellar gas.

      Now Lars Hernquist and David Weinberg, a computational astrophysicist at Ohio State University, have simulated the forest and have found something totally different, not like clouds at all. They start the simulation around a billion years after the Big Bang, when gas was spread evenly in space. After the simulation evolves for several billion years, the gas condenses, not into clouds, but into giant sheets and wispy ribbons.

      Could the Lyman-alpha forest consist of sheets and filaments of gas, rather than clouds? By shining a computer-generated quasar through the gas model, Hernquist and Weinberg compare the model's shadows to the real Lyman-alpha bands in the quasar spectrum. The match is nearly perfect. Hernquist and Weinberg have apparently modeled the evolution of galaxies.