Astronomers have recently identified hundreds of "polluted" white dwarf stars in the Milky Way, which are actively consuming planets in their orbit. These stars provide valuable insights into the interiors of these distant planets but are notoriously difficult to find.
Historically, astronomers manually reviewed extensive survey data to identify these stars, followed by observational confirmations. A team led by University of Texas at Austin graduate student Malia Kao has utilized a novel artificial intelligence technique called manifold learning to streamline this process, achieving a 99% success rate in identification. Their findings were published today in the Astrophysical Journal.
White dwarfs represent the final stage of stellar evolution, having exhausted their fuel and released their outer layers into space. Eventually, our sun will become a white dwarf in approximately 6 billion years. Occasionally, planets orbiting these stars are drawn in by gravity, torn apart, and consumed. This process "pollutes" the star with heavy metals from the planet's interior.
"For polluted white dwarfs, the inside of the planet is literally being seared onto the surface of the star for us to look at," Kao explained. "Polluted white dwarfs right now are the best way we can characterize planetary interiors."
Keith Hawkins, an astronomer at UT and co-author on the paper added: "It's the only bona fide way to actually figure out what planets outside the solar system are made of, which means finding these polluted white dwarfs is critical."
Detecting these stars is challenging due to subtle evidence and a limited timeframe for identification. To expedite this process, researchers applied AI techniques to data from the Gaia space telescope. Despite initial skepticism about Gaia's low-resolution data's usefulness for this purpose, Hawkins noted: "This work shows that you can."
The team used manifold learning where an algorithm identifies similar features within data sets and groups them visually for further review. They sorted over 100,000 potential white dwarfs and identified a promising group of 375 stars exhibiting heavy metals in their atmospheres. Follow-up observations with UT’s McDonald Observatory confirmed their findings.
"Our method can increase the number of known polluted white dwarfs tenfold," Kao said. "Ultimately, we want to determine whether life can exist outside our solar system."
This research highlights how artificial intelligence is aiding scientific discoveries at The University of Texas at Austin during its declared Year of AI.
The study utilized data from ESA's Gaia mission processed by its Data Processing and Analysis Consortium and follow-up observations with various telescopes including Hobby-Eberly Telescope (HET) and Very Large Telescope (VLT). The Texas Advanced Computing Center provided computational resources.