NEW TECHNIQUE TO HELP ASTRONOMERS DEAL WITH WEALTH OF DATA
COLUMBUS, Ohio -- New telescopes are giving cosmologists both their fondest wish and a new dilemma: now that they can measure the shapes, sizes, and colors of millions of galaxies in our corner of the universe, how do they make sense of all that new data? An Ohio State University astronomer is helping to answer that question.
Alberto Conti, a graduate student in astronomy, and his colleagues are exploring a new technique to pick out which of the many attributes of galaxies -- such as mass, initial rotational speed, and age -- are most important in determining the properties that can be observed, such as brightness, diameter, and color.
At a biannual meeting of the American Astronomical Society in Atlanta, Conti presented a method of computer analysis that may help astronomers manage and interpret the huge amount of data they are now beginning to harvest.
"Now that we have such a massive data set to draw from, we need to figure out how to extract the information that is most important," said Conti.
Understanding why galaxies are the way they are has always been a difficult task, Conti explained, because nobody knows all the variables involved. Asking an astronomer to detail how galaxies form was like asking a budding cook to prepare a birthday cake without a recipe.
One way for astronomers to do this is to look at galaxies near our own and work backward to a "recipe."
What astronomers see is a hodgepodge of galaxies scattered throughout the universe, each with its own unique "face." Some feature bright pinwheels, or extend filaments of gas and dust from the center into the shape of a giant letter s. Some look as if they were stabbed through the middle by a bright bar of stars. Other, more amorphous galaxies resemble nothing so much as cotton balls hanging in space.
"We know galaxies differ from each other. We know those observable differences must reflect physical differences in how they formed, but we don't necessarily know what those physical differences were. We have ideas. They are probably right in part, but not entirely," said Weinberg.
Barbara Ryden, associate professor of astronomy and one of Conti's advisors, is working with Conti and Weinberg on the project. The three are combining a method of theoretical analysis called semi-analytic modeling with a statistical technique called principal component analysis.
While computer simulations use what astronomers know about gravity and the early universe to create a picture of where galaxies came from, semi-analytic modeling relies instead on solving complex mathematical equations for different sets of input conditions. Ryden is one of the pioneers in this area.
Principal component analysis (PCA) is a technique statisticians commonly use to find the associations between data.
Weinberg gave a simplified example of PCA from economics. When considering data such as a family's income, number of cars, and the size of the home, he said, PCA would reveal that income was factor that determined how many cars a family owned and the size of the home.
Previously astronomers have explored this avenue for investigating the characteristics of galaxies, but they have always been limited by the quality and quantity of data available. New surveys are providing a vast sample of galaxies that are ideal targets for PCA.
In Atlanta, Conti outlined how his approach to semi-analytical modeling can help astronomers interpret the results. While others have used semi-analytic modeling to work out the predictions of specific theories of galaxy formation, Conti lets the input parameters of his models vary widely. He then applies PCS to the computer-generated models of galaxies he obtains to see whether it can reveal the variations that governed observable differences between them.
It's tough work, considering all the observable properties of galaxies -- shape, luminosity, color, and the signature spectrum of light emanating from the galaxy's stars, to name only a few. Add to that all the physical parameters astronomers use to form their theories of galaxy formation, such as the mass of material that first formed a galaxy and the speed with which it spun into shape, the type of subatomic particles present in the early universe, and whether or not galaxies have collided in the interim.
Weinberg stressed that Conti has not yet applied this technique to observational data, but rather is laying the foundation for the techniques that will help astronomers grapple with this data in the future.
A new approach for handling astronomical data will be needed even more in the future, Conti explained. The Hubble Space Telescope has already given cosmologists much more data to work with, he said, as have the powerful telescopes of the Keck Observatory, which rest atop Hawaii's dormant Mauna Kea volcano. But he is pursuing this work as part of his doctoral thesis in anticipation of data from the Sloan Digital Sky Survey.
The Sloan telescope, at Apache
Point Observatory in the
Weinberg has been a member of the SDSS collaboration for several years now, testing software, formulating survey strategy, and selecting galaxies for the telescope to target. Years from now, when all that work pays off, he and other astronomers may use Conti's technique to make sense of the results.
Contact: Alberto Conti, (614) 292-2076;
David Weinberg, (614) 292-6543; Dhw@astronomy.ohio-state.edu
Written by Pam Frost, (614)292-9475; Frost.firstname.lastname@example.org