INTERCONNECTS

By Paul Tooby, SDSC and NPACI Science Writer -- Rutgers University astronomer David Merritt and his colleagues are using supercomputers of the National Partnership for Advanced Computational Infrastructure (NPACI) to model one of the most dramatic events in astronomy—the behavior of the black holes when two galaxies collide. From revealing strong and unexpected effects on surrounding stars to the potential to produce rogue black holes wandering the universe, these simulations—among the largest of their kind ever run—are shedding new light on the life cycle of black hole binaries in the nucleus of merging galaxies. Observations have yet to definitively detect black hole pairs, and simulation results of Merritt’s team, recently published in The Astrophysical Journal, are helping astronomers as they work to find new evidence for binary black holes. The accepted view among astronomers is that nearly all galaxies harbor supermassive black holes, objects roughly a few kilometers across containing mass equivalent to millions or billions of Suns. Astronomers also believe that galaxies and their central black holes grow by merging. Indeed, our Milky Way Galaxy is currently swallowing a smaller galaxy. In turn, the larger Andromeda Galaxy, our nearest spiral galaxy, will eventually merge with our own Milky Way. Previous research by Merritt and others has shown that during the initial stages of the collision between two galaxies, their black holes lose energy through interaction with the surrounding stars and “fall” relatively rapidly toward the center of the merged galaxy. There, the black holes orbit each other in a gravitationally bound pair. Merritt and Rutgers graduate student Milos Milosavljevic suspected that the dynamics of this black hole pair may play a central role in the evolution of the new, merged galaxy. A key question is whether the two black holes quickly coalesce into a single, larger black hole, or continue to orbit each other in the new galaxy. And what effect does this dance of giants have on nearby stars? To answer these and other questions, Merritt and Milosavljevic used NPACI supercomputers to carry out precise simulations of the motions of the objects in the merging galaxies under the influence of their gravitational fields (known as N-body simulations). “What our simulations showed is that, contrary to expectation, the two black holes do not immediately coalesce into a single black hole,” said Merritt. “In fact, once they’re near enough—about three light years—they begin to orbit each other as a bound, remarkably stable pair.” The simulations predict that such black hole pairs could continue orbiting each other for one billion years or more—long enough to collide with a third galaxy. Appropriate for these exotic objects, the mechanism by which the black hole binary achieves such longevity is itself dramatic. Initially, as the galaxies begin to merge, each black hole is surrounded by stars, with a peak in star density at each black hole that drops off rapidly in a cusp, or sharply curved profile. As the two black holes approach and become a bound pair, their separate cusps of surrounding stars merge, forming a single new peak around the black hole pair. LONELY BLACK HOLE BINARIES “Then something surprising happens,” said Merritt. “The surrounding stars are no longer orbiting a single black hole in an orderly way, but instead now orbit two black holes—resulting in far more complex orbits.” Merritt explains that “if you follow a single star as it orbits the black hole pair, it will follow a chaotic path—orbiting tens or hundreds of times.” Since black holes are so small, stars rarely collide with them. Eventually, however, the chaotic orbit of a star will bring it very close to one of the black holes. But instead of being swallowed, the star gains so much speed that it is hurled outward in a “gravitational slingshot” effect that gives it sufficient velocity to escape the black hole pair. NASA used the same slingshot effect to fly its Cassini spacecraft within 750 miles of Earth in a 1999 maneuver that threw the spacecraft outward toward its destination of Saturn. In its close encounter with Earth, Cassini was greatly accelerated, while the decline in the orbital velocity of the far more massive Earth was imperceptible. Over time, as the black hole binary accelerates large numbers of stars—transferring energy outward—the black holes lose sufficient energy to spiral closer. “Eventually, they’re only about 0.1 light-year apart,” said Merritt. The researchers’ simulations showed for the first time that as the merger process continues, the binary black hole efficiently scours away the peak in star density around it, hurling stars outward and resulting in a new, emptier galaxy core with a uniform star density. Once the core is relatively empty, the black holes lose little further energy, making their orbit highly stable. A significant new achievement of these simulations is that—thanks to the large NPACI computational resources—the researchers were able to start the simulations before the galaxy merger and continue them past the merged high-density cusp of stars until the orbit of the black holes shrinks to a stable binary with the relatively empty core of stars. “This is more than four orders of magnitude, something that’s never been done before,” said Merritt. “It has allowed us to see all the stages in the merger evolution down to a separation of only a fraction of a light-year.” The simulation work has earned Milosavljevic a Sherman Fairchild Fellowship at Caltech, a prestigious national award for postdoctoral research in astronomy. Merritt notes that the simulation is an idealized case because it ignores the effects of gas and other phenomena that can make the mergers of real galaxies more complex. Despite the simplifications, the researchers have found experimental support for their simulation. Observations with the Hubble Space Telescope have shown that the density of stars in the cores of galactic nuclei agrees with the simulations. Eventually, the orbiting black holes will undergo a brief, final coalescence into a single black hole, during which they will radiate energy in the form of gravitational waves. Such gravitational waves, if detected, would not only provide a “signature” revealing the black holes, but also information about their orbits, masses, and spins—as well as the first-ever test of Einstein’s theory of general relativity under such extreme conditions. “This is a very fruitful time for interaction among theory, observations, and simulations,” said UC Berkeley astronomer Don Backer, whose research involves experiments to detect gravitational waves emitted by coalescing black holes through their effects on pulsar timing. “We rely on these simulations to tell us that once galaxies have merged, their binary black holes do indeed coalesce in a ‘timely’ manner.” The simulations are also useful to the designers of gravitational wave detectors, including the planned Laser Interferometer Space Antenna (LISA). Another exotic scenario is suggested by the simulations: eventually, a third galaxy may collide with a galaxy containing a binary black hole, adding a third black hole to the mix. “The complex orbital interactions could result in such scenarios as, for example, two of the three black holes forming a tight binary, with the third ejected by the gravitational slingshot—which could result in rogue black holes flying around the universe,” said Merritt. COMPUTING BLACK HOLES “Having access to large computing resources has been vital to obtaining these new insights,” said Merritt. The simulations involved a major computational effort, requiring thousands of hours on an NPACI Cray T3E supercomputer at the University of Texas. The tools used in the simulation are two N-body codes, which model the gravitational interactions between the bodies in the merging galaxies. Because the scope, or dynamic range, of the merger is so large, Merritt and Milosavljevic had to use two different N-body codes. They began the simulation with the black holes far apart, using a recently released tree code known as GADGET developed by Volker Springel, Naoki Yoshida, and Simon D.M. White, researchers at the Max-Planck Institute for Astrophysics in Garching, Germany. The code models the total number of object pairs (N2) by approximating distant interactions, and using 262,000 objects. As the black holes spiraled closer, the researchers switched to a more exact Aarseth N-body code, NBODY6++. This code was developed by Rainer Spurzem, an astrophysicist at the Astronomical Research Institute in Heidelberg, Germany, and Holger Baumgardt, an astrophysicist at the University of Edinburgh. The code explicitly includes every object, directly simulating the behavior of each pair of some 32,000 stars in one of the largest such N-body simulations ever run. Beyond yielding new insights into black holes in merging galaxies, the approach developed by Merritt and his colleagues can also be extended to studies of dark matter and gravitational waves from coalescing supermassive black holes. Said Merritt, “What began as a rather specialized computer simulation to investigate the dynamics of binary black hole evolution during the merger of two galaxies has now opened lines of inquiry that may be able to illuminate a broad range of important unsolved questions from astronomy to fundamental physics.” ----- This article also appears in the most recent version (April-June 2002) of NPACI & SDSC's quarterly science magazine, EnVision. Used by permission. -----