In science fiction, binary stars are often shorthand for the exotic. A pair of suns rising over some alien landscape quickly communicates the foreign and the outlandish. But that reaction just shows our bias toward what is familiar. Out in the universe, twosomes are nothing out of the ordinary. Astronomers think that a third to more than half of all stars are part of binary systems.

Northwestern University's Ronald Taam has used a progression of systems at NCSA over the last seven years to explore how these binaries operate. Taam works closely with Paul Ricker, a research scientist at NCSA and an assistant professor of astronomy at the University of Illinois at Urbana-Champaign, studying what is known as the common envelope phase. During this stage of binary star evolution, the outermost portions of the stars' atmospheres share a single boundary.

The common envelope phase is marked by extreme changes in the nature of the stars and intense interaction between them. Angular momentum—the physical property that keeps a top spinning—drops precipitously. The time it takes for the stars to orbit one another goes from months to just a few days. Meanwhile, mass from what is usually the larger star begins to transfer to its lower mass companion, a lobe of superheated plasma spiraling from one star to another across hundreds of millions of miles.

Taam calls the onset of the common envelope phase "stellar cannibalism," and it can lead to the ejection of part of the stars' mass that makes up their common envelope. Ultimately, the lower mass companion star and the core of the more massive star may merge into a single remnant. Or the core of the more massive star and its lower mass companion may both survive. In this case, the core can evolve into some of the more unusual members of the stellar family—such as a black hole, neutron star, or white dwarf—and continue to orbit its companion. This remnant binary system is known as a compact binary.

Taam and Ricker's current simulations are mapping the conditions under which the common envelope is ejected. A series of models at exceptionally high time and spatial resolution will cover a range of masses for the stars and a range of times that it takes them to orbit one another. An initial calculation will be published in Astrophysical Journal Letters in January 2008.

"We're delineating the regime where binary stars survive and where they don't. We hope to do several of these calculations with our current allocation on Tungsten" and more in the coming years, Taam says.

Understanding how the common envelope is shed and how compact binary systems are born will transform the way optical and radio astronomers think about the universe that they observe. The shapes of some planetary nebulae photographed by the likes of the Hubble Space Telescope, for example, can be linked to changes during the common envelope phase. And compact binary systems with short orbital periods generate the gravitational waves that many think will be among the first seen by the National Science Foundation's LIGO, the Laser Interferometer Gravitational-Wave Observatory now operating in Louisiana and Washington.

The team also hopes that their models will improve astronomical population studies, which predict the number and kinds of binary stars in the universe.

'An equal partnership'

The common envelope phase is marked by a wide range of time scales and spatial resolutions. "The envelope can be several astronomical units across, while the core of the stars might be only the size of the Earth," says Taam. (An astronomical unit is about 93 million miles, the distance from Earth to our sun.) The diameter of the orbit of the two stars also varies tremendously over the course of the simulations, which means that simulations must be tracked in very small increments. "You're basically looking at a year a minute at a time," explains Ricker.

The code initially used to model the common envelope phase relied on a nested grid. This method solved parts of the system at different, but fixed, resolutions. But because the stars orbit one another, the areas that want for the most detail—near the stars' cores and at the edge of the common envelope—move within the grid. That movement limited simulations to situations in which the ratio of the two stars' masses was very large.

To overcome this constraint, Taam began working with Ricker and an adaptive mesh refinement code called FLASH. Before coming to NCSA, Ricker spent several years as a principal developer of FLASH at the University of Chicago. He worked with Taam to augment the code for simulating binary stars. FLASH is such an improvement over a nested-grid method because it automatically places more refined patches of the grid based on where there are sudden jumps in density. In other words, it resolves key regions of the stars in higher resolution and moves that higher resolution area in space.

"Working with FLASH was a substantial technical challenge, but it gives us so much more realism," says Taam. "It's only been possible in collaboration with Paul. Not just a transfer of his technical skills to our project, but an equal partnership where he contributes to the analysis of results and their implications as well."

For his part, Ricker says "Our collaboration really does extend beyond the technically minded questions. We're throwing around ideas and refining them through conversation. It's much more interesting that way—a very productive and enjoyable collaboration." And an important facet of NCSA's mandate.

The team singles out Greg Bauer of NCSA's Performance Analysis and Methods group, as well. Bauer has focused on issues of migrating from one supercomputer to another; Taam's calculations have run on four different platforms in his seven years as an NCSA user. "Greg has been instrumental in tracking down a series of HDF performance issues and improved the way the calculations are checkpointed to disk," says Ricker. HDF—developed at NCSA for two decades before it was recently spun off into its own non-profit company—is the format used to store data produced by the team's calculations.

"It's extremely valuable to have someone dedicated to your project who is cognizant of these issues," says Taam. "Greg helps to make things run."

Taam and Ricker's current simulations look at a binary pair in the common envelope phase in a box about one astronomical unit across. This box is covered in a grid with an effective resolution corresponding to about 20,483 zones. Their initial FLASH-based calculations considered a grid with an effective resolution of only about 5,123 zones. At this level of resolution, the simulations have shown that gas around the core of both stars begins to spin much more during the common envelope phase. The team believes that this makes mass ejection much easier. They've also seen that matter is ejected in all directions but that there is a preference for ejection in the equatorial plane with a large contrast in the density of this matter among the stars' equator and poles.

They also believe that the current resolution is high enough to determine the outcome of these phases for population synthesis studies of binary stars in the universe. These studies combine various models of how stars evolve, explore many combinations of possible starting parameters for those models, and estimate the number of stars of particular types that exist in the universe and those stars' characteristics. Currently, those studies rely exclusively on one-dimensional stellar models and simple prescriptions for the binary interactions. As a result, these studies estimate the impact that binaries have instead of modeling it directly.

"To do population synthesis right, you have to account for that half or more of stars that are in binaries as realistically as possible," says Ricker. By doing so, astronomers will be able to better capture the nature of the universe's stellar population and the combination of elements that those stars release as they change over time.

Accordingly, Taam and Ricker's simulations ultimately tell the astronomy community more about the particulars of how the common envelope phase proceeds and how compact binaries form. But they also hold the promise of revealing more about how the universe in general evolves.

This research is supported by the National Science Foundation. Team members: Greg Bauer, Paul Ricker and Ronald Taam