Study of 350,000 stars seeks Sun's lost "siblings"

Computer code analyzes stellar spectra to determine stars' chemical compositions.
By Laurel Kornfeld | Dec 04, 2018
An international team of astronomers has mapped the chemical compositions of 350,000 stars in the Milky Way as part of an effort to locate the Sun's "siblings," or stars that were born in the same cluster as the Sun.

While the stars in the original cluster separated after being fully formed, pulled apart by the Milky Way, all have the same chemical composition, making it possible for scientists to trace them even in their current, scattered locations.

The Galactic Archaeology survey (GALAH), launched in 2013, has sought to locate these stars as part of its broader mission to study the galaxy's formation and evolution.

Project scientists are now releasing the first batch of data from the mission, which is discussed in papers submitted for publication to the journals Astronomy and Astrophysics and Monthly Notices of the Royal Astronomical Society.

"This survey allows us to trace the ancestry of stars, showing astronomers how the universe went from having only hydrogen and helium--just after the Big Bang--to being filled with all the elements we have here on Earth that are necessary for life," said Martin Asplund of the Australian National University (ANU) Research School of Astronomy and Astrophysics and ARC Centre of Excellence for All Sky Astrophysics in 3 Dimensions (ASTRO 3D).

"Measuring each chemical element abundance to get the stellar DNA for so many stars is an enormous challenge, but that's exactly what we've done, so this is a fantastic scientific achievement," he added.

Data for GALAH was collected using the HERMES spectrograph, which is attached to the 3.9-meter Anglo-Australian Telescope at ANU's Siding Spring Observatory.

By observing the various colors of light emitted by each star, scientists measured their chemical contents.

The scientists used a computer code named The Cannon in honor of American astronomer Annie Jump Cannon, who classified the spectra of more than 350,000 stars using just her eyes a century ago, to recognize patterns in the spectra of the many stars being studied.

"We train our computer code The Cannon to recognize patterns in the spectra of a subset of stars that we have analyzed very carefully, and then use The Cannon's machine learning algorithms to determine the amount of each element for all of the 350,000 stars," Asplund explained.


We are dedicated to maintaining a respectful community that actively engages in lively discussions about news stories and blog posts. Please keep the following in mind when writing your comments.