Splitting Stars in Binary Systems

John Cazes, Saied Andalib, Eric Barnes, Howard Cohl, Patrick Motl, Erik Young, Louisiana State University
Kimberly New, Drexel University

n the community of astrophysicists, excitement over the discovery of planets around stars other than the Sun has renewed interest in many old and puzzling questions about star formation. Stars are born from clouds of gas and dust as singular stars, pairs of stars--called binary systems--or clusters. Telescope observations provide snapshots of the processes of star birth in our own galaxy and in others. "More than half the stars in our galaxy belong to binary systems," said Joel Tohline, a professor of physics and astronomy at Louisiana State University (LSU). "Our group asks the question: why do stars tend to form in pairs?"

To form stars, the gas and dust clouds are thought to collapse under their own gravity, slowly cooling and coalescing until they are dense enough to begin burning hydrogen at their cores. "We finally have computers large enough to help us model the details of star formation and, in particular, of the binary formation process," Tohline said. "Our investigations on NPACI machines are bringing us closer to the answers."

STELLAR FISSION

CODES AND ENVIRONMENTS

COMPARISON WITH OBSERVATIONS

STELLAR FISSION

Binary star systems exhibit a wide range of orbital periods--from days to tens of thousands of years--and mean distances from one another, with a median at about 50 astronomical units (AU). (One AU is equal to the average distance from the Earth to the Sun, about 93 million miles.) Many scenarios have been proposed to explain the dynamics of binary formation. The leading candidates have been the capture of one star by another, direct fragmentation of the protostellar cloud, formation and fragmentation of a disk around the cloud, and simple fission.

"The direct fragmentation of a protostellar cloud early in its collapsing phase seems to be a mode of formation for long-period binaries that are spaced widely apart," Tohline said. "But because they must form in higher density environments that don't cool very efficiently, binaries with small separations and shorter orbital periods don't seem to follow that scenario."

The logical scenario for the fission hypothesis of binary formation is very simple (Figure 1). A protostellar, self-gravitating ellipsoid cloud flattens as it rotates, eventually taking a peanut or dumbbell shape, and then splits in two. The scenario was first written for idealized ellipsoidal figures by the great geometer Georg Friedrich Riemann in the 1850s.

The logic was compelling enough to prompt many astrophysicists, beginning with the late Nobel Prize-winning astrophysicist Subrahmanyan Chandrasekhar of the University of Chicago, to model ellipsoidal figures of equilibrium as precursors of binary star systems. Extensive, detailed work was done over the years for incompressible fluid systems with simple rotation, "but all along it was recognized that these simplifications were nonphysical," Tohline said. "The extension of these studies to more compressible configurations, as well as to systems with non-trivial internal motions, was not possible before the advent of modern computers." But when more realistic conditions were added, one by one, the results did not always support the fission hypothesis.

"It was really necessary to do a complex, detailed calculation, including all the known physics, and to visualize this with high-performance tools, if the fission hypothesis was not to be ruled out," Tohline said. Using two independent modeling techniques, the Tohline group has recently demonstrated that it is possible to construct dynamically stable, self-gravitating configurations with highly nonaxisymmetric structures--ellipsoidal objects, dumbbell-shaped objects, and common-envelope binaries--out of the highly compressible gases that constitute protostellar clouds. Moreover, the most recent results, obtained by LSU graduate student John Cazes, demonstrate that conditions exist under which simple binary fission occurs as the protostellar cloud contracts and elongates.

A movie of Cazes's simulation shows the deformed cloud prior to fission and the nascent binary system immediately afterward (Figures 2 and 3). "We have run this scenario backward to check on whether the binaries are as likely to merge as to move apart," Tohline said. "It seems that, once born, their condition is stable against merger."

CODES AND ENVIRONMENTS

These investigations are computationally intensive, and the Tohline group has long been concerned about running their large cases as efficiently as possible. They have migrated their codes from a MasPar system at LSU to newer parallel platforms. Supported by funding from NSF's Astronomy Division, the most recent work has been done at NPACI--both at SDSC and the Texas Advanced Computing Center--and at the Department of Defense major shared resource center of the Naval Oceanographic Office at the Stennis Space Center in Louisiana.

The primary tool is an algorithm that performs multidimensional computational fluid dynamics (CFD) in astrophysical settings, written in High Performance Fortran. Parameters are calculated on a 3-D, cylindrical coordinate mesh. Since this has proved to run inefficiently as compiled on the SDSC CRAY T3E, LSU graduate student Patrick Motl has recently converted the code to run using the Message Passing Interface, which achieved a factor of three speedup. To develop the nonaxisymmetric structures used as input to the CFD code, the group uses a self-consistent field code that draws on work originally done by University of Tokyo astronomer Izumi Hachisu.

"When performing these multidimensional simulations, it is always important to develop an efficient technique by which the numerical results can be readily analyzed for physical content," Tohline said. The group has developed its own Heterogeneous Computing Environment through which the data can be visualized "on the fly." "We have adapted this system to run with the T3E, both at Texas and at SDSC, and the SGI Onyx at SDSC, and we have been able to make very good looking animations of the processes we've been modeling," Tohline said.

In addition to Cazes and Motl, Tohline's group includes former students and postdoctoral researchers Saied Andalib (LSU), Dimitris Christodoulou (now at Psyche Systems in Boston), and Kimberly New (now at Drexel University), as well as other current graduate students: Eric Barnes, Howard Cohl, and Erik Young.

---

Figure 2. Rotating Bar

Steady-state, "bar-shaped," self-gravitating gas cloud with non-trivial internal motions. Each color represents a 3-D surface of uniform density, with green at 80%, yellow at 40%, red at 4%, and blue at 0.4% of the maximum density of the gas cloud. Visualization by John Cazes, Louisiana State University.

Figure 3. Protobinary System

Protobinary system with circulation about points of maximum density. As in Figure 2, each color represents a 3-D surface of uniform density, with green at 80%, yellow at 40%, red at 4%, and blue at 0.4% of the maximum density of the gas cloud. Visualization by John Cazes, Louisiana State University.

COMPARISON WITH OBSERVATIONS

The next step is to perform hydrodynamics simulations of evolved unequal mass binary systems, in which there is mass transfer between the stars, which is the topic of Motl's dissertation work. Eric Barnes plans to use the hydrodynamics code to simulate the "settling" or "warping" of galaxies in an oblate, spheroidal, dark matter halo about a central galaxy.

In addition to more detailed studies of the steady-state triaxial gas clouds and binary system formation, Tohline wants to connect the simulation results with astronomical observations. One important set of observations will be the studies of self-gravitating fluid systems enabled by the Laser Interferometer Gravitational Wave Observatory. "One of the major instruments of this observatory is being built near Livingston, Louisiana, less than an hour's drive from LSU," Tohline said. "Our problems may well be related to those studies."

Tohline will also be at Caltech during a sabbatical in the Fall 2000 semester to work with Anneila Sargent and the group operating the Owens Valley Millimeter Array. This leading radio telescope array now has sufficient spatial resolution and signal-to-noise characteristics to permit mapping of star-forming gas clouds on linear scales approaching the size of the solar system. "I'll be working to modify the codes in such a way that they can output maps for comparison with radioastronomical observations," Tohline said. "I know that SDSC's high-speed link to that campus will accelerate the process." --MM