Planet Hunters Wanted

Some spare processor cycles would be nice, too

PHOTO: Nick Dolding/Getty images

As of this past December, there were 209 planets known to be orbiting stars outside the solar system—209 more than were known just two decades ago. And while the rate of extrasolar planet discoveries by professional astronomers is impressive (27 new planets were discovered in 2006 alone), some researchers think the catalog of new worlds could swell even faster. The discoveries could mount not by building new planet-finding telescopes or satellites, but by marshalling an army of amateur astronomers and enthusiasts along with their personal computers.

The researchers believe that many extrasolar planets have already been unwittingly detected through telescopes but that they are hidden because of the complexity of teasing out the signal of the planet’s existence from a mountain of astronomical data. Now two free programs are being developed that will let anyone hunt through the data for our cosmic neighbors.

The reason it’s so hard to identify planets is that, unlike stars and some nebulae, planets don’t give off their own light. While satellites are in the works to directly detect the dim light that gets reflected from planets orbiting stars, nearly all the known extrasolar planets have been uncovered using the same method astronomers typically use to discover black holes. They look not for the object directly but for its invisible gravitational tug on the luminous star nearby.

This tug causes the star to wobble slightly back and forth as seen from Earth, or more formally, the star’s ”radial velocity” accelerates and decelerates. The wobble can be detected because the frequency of the light from the star will be altered by the Doppler effect. That is, when the star’s wobble moves the star closer to us, its light is shifted slightly toward the blue end of the spectrum, and when the wobble pulls the star away from us, its light is shifted toward the red end of the spectrum.

Using this effect, for ex­ample, an alien astronomer observing our sun would notice that the sun wobbles back and forth every 23.8 years, and some number crunching would reveal that period to be the signature of a large gas giant orbiting the sun every 11.9 years—in this case, Jupiter.

Not surprisingly, finding tiny planet-induced Doppler shifts of a star’s spectrum is a very delicate undertaking. And when there are three or four large planets like Jupiter and Saturn in orbit around a star, the complex superposition of wobbles that the invisible companions produce are even more challenging to decode.

Astronomer Greg Laughlin, at the University of California, Santa Cruz, wants to make anyone willing to learn a few basic concepts of orbital mechanics capable of decoding those faint wobbles. He wants to see planet hunting become similar to online collaborative efforts like Wikipedia, the encyclopedia written collaboratively by volunteers, where anyone can participate and the most useful work often bubbles to the top.

”We were very much inspired by the interest that SETI@home generated,” Laughlin said, referring to the project that allows users all around the world to analyze radio telescope data for alien transmissions. Although no signal from E.T. has yet been found, the project is considered a groundbreaking experiment in distributed computing. But Laughlin wanted participants to do more than just watch their computers do all the work. ”What I’ve been looking for,” he says, ”was a project that involved extrasolar planets—exploration that people are interested in—but which could use more than just spare clock cycles on the user’s computer. We wanted something that could engage the brain cycles as well.”

So Laughlin and four collaborators from UC Santa Cruz, the University of Bologna, in Italy, and the U.S. Naval Observatory launched a program called Systemic (http://oklo.org). Systemic’s beta version is not for the technically fainthearted, but the Java-based program repays the time investment. Learning to navigate its thicket of astronomical terminology and Newtonian modeling options brings users to the bleeding edge of astronomical exploration.

Systemic stores and displays radial velocity measurements of stars. Loaded into the software are the published data for every star system known or thought to contain planets—as well as ”observations” of scores of fictional stars that Laughlin’s team fabricated to test the limits of Systemic as a planet-hunting tool.

Once users learn to navigate Systemic, they can start conjuring up hypothetical planets, adjusting planetary masses, distances from the host star, the shape of the orbits, and so on, until they find a combination that produces a wobble that best matches the radial velocity data. After creating the best-fit planetary system, a user can then post the results to Systemic’s online ”back end” and compare them with those of other aspiring planet hunters around the world.

For example, the 313 data points for the known planet-bearing star 55 Cancri A reveal that the star is experiencing wobbles that speed up or slow down its radial velocity by as much as 100 meters per second. Using these data, since 1997 professional astronomers have determined the presence of four planets orbiting 55 Cancri A. With Systemic, a skilled amateur user can regenerate something approaching this decade of professional results in less than half an hour. What’s more, of the 82 proposed 55 Cancri A planetary systems that users have already uploaded to http://oklo.org, some of the tightest fits to the data feature five or six planets, not four. Clearly, 55 Cancri A hasn’t yielded up all its planetary secrets yet.

The opportunities for amateur science, says Jaymie Matthews of the University of British Columbia, in Vancouver, are rife with a program like Systemic. ”You never know, sometimes that flash of inspiration comes from an unexpected direction,” he says. ”We’re fortunate in astronomy in that without an accelerator in your basement or a 10-meter telescope in your backyard, you can still potentially make contributions to the cutting edge of science.”

However, the wobble method works best for detecting Jupiter-type planets, which are massive enough to really tug on their stars. The best chance for finding Earth-sized ­planets (and hence possible homes for E.T.) may rely instead on a much more difficult technique at the heart of another online project called PlanetQuest( http://planetquest.org). The technique, known as the photometric transit method, involves closely monitoring the brightness of a star and looking for slight dips in its luminosity as a planet passes in front of it, blocking some of its light.

No planet has yet been discovered using this method, however. This is, in part, because a planet’s transit across a star’s face, as seen from Earth, might last only a fraction of a day and can occur only once during the planet’s orbit, the length of which, of course, could easily be measured in years.

As a result, digging out these momentary dips in luminosity from archived measurements of stellar brightness is a huge exercise in data mining that, ­according to Laurence Doyle, an astrophysicist at the SETI Institute, in Mountain View, Calif., is best tackled by distributed computing. Drawing from more than two years of dedicated nightly observations of eight fields of stars, PlanetQuest will comb through the data for periodic dips in stellar luminosity. Each field of stars is one degree across in the constellations Cassiopeia and Sagittarius, with each image containing 16 million pixels and tens of thousands of stars.

Naturally, though, a star’s brightness could increase and decrease for many reasons. Perhaps a wisp of a cloud momentarily passed overhead, or the telescope’s mirror warped slightly due to a change in temperature. So the challenge the PlanetQuest team faces is to find those rare cases when the dimming and brightening happen in a periodic way that can be explained in terms of a possible planetary transit.

Factoring in the numerous possible planetary transits across both single- and double-star systems, PlanetQuest requires an estimated three CPU-years to analyze the light from a single star. So PlanetQuest will divvy up the problem and let many hundreds or thousands of computers tackle it. Closely modeled on the SETI@home approach, the project that uses Internet-connected computers in the search for extraterrestrial intelligence, the PlanetQuest software doesn’t rely on direct user intervention but runs as a background process or screen saver on personal computers. Currently in development, a beta version is scheduled to be released to the public this summer.

”We’d like little Johnny in Madison, Wisconsin, or little Sally in Bombay, India, to be the first to find a planet—that would be very cool,” says Doyle.

Building on the experience gained from SETI@home, the computing algorithm behind PlanetQuest isn’t itself very groundbreaking. Leslie Lamport, a researcher at Microsoft Research, in Mountain View, Calif., says the principal challenge that such projects face these days is getting the word out and convincing users to sign up. ”This is a fairly straightforward coding problem,” he says. ”There aren’t many really interesting ways to screw up.”

Still, says Zoran Ninkov, a professor of solid-state imaging at the Rochester Institute of Technology, in New York, and a member of PlanetQuest’s board of directors, some potential private PlanetQuest funders have balked, because they thought the computing requirements were insurmountable.

”There are some who say...’This isn’t that technical a problem,’ ” Ninkov says. ”Then you get the other guys with backgrounds in mathematics who say, ’Impossible. You can’t do it. What’s the point?’ Somewhere in the middle, there has to be an answer.” And that answer could soon be on your very own computer screen.

About the Author

MARK ANDERSON is a science and technology writer whose work has been published in Wired, Science, New Scientist, Harper’s, and Rolling Stone, among others.

Advertisement
Advertisement