Despite having recently officially ended its science operations in January, Gaia, one of the most prolific star explorers ever, is still providing new scientific insights. A recent paper pre-published on arXiv (which has not been peer-reviewed but was submitted to the Astrophysical Journal) took another look at some Gaia data to try to find a unique type of astronomical entity – white dwarf stars that are paired up in a binary with a main sequence one. By applying a machine learning technique called a “self-organizing map,” they found 801 new white dwarf-main sequence (WDMS) binaries, increasing the total number ever found by over 20%.
WDMS pairs are important astronomical clocks, given the predictability of the age of a white dwarf. Pairing the known age of a white dwarf with the variable age of a main sequence star allows for more straightforward calculations of important parameters like the relationship between a star’s age and how much “metal” (i.e., not hydrogen) it has or even the relationship between a star’s mass and its radius.
Only around 4000 of these WDMS pairs have been found, with a major haul first described only a few months ago, which we also reported on. Every additional discovery provides more insight into the relationship between these two types of stars and adds statistical heft to existing theories.
For this latest haul, Xabier Pérez-Couto and his co-authors from the Canary Islands and the Universidade da Coruña used a machine-learning technique called Self-Organizing Maps (SOMS). Also known as Kohonen maps, SOMs are a type of unsupervised machine learning technique that utilizes “competitive learning” that allows them to discover structures in data sets without the need for any labeled training data.
When the authors ran an SOM on data collected by Gaia’s Blue Photometer (BP) and Red Photometer (RP), which is essentially a dataset of low-resolution images of many stars, they found slight variations in the red spectrum based on whether a star was a single white dwarf or whether it was paired up with a main sequence companion. To prove their algorithm worked, they ran it on a collection of 90.667 stars and found a total of 993 potential WDMS binaries.
Other surveys, including the Sloan Digital Sky Survey (SDSS) and the Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST) survey, had already found 192 of those. Given the similarities to previous surveys, they estimate that their SOM algorithm was 90% accurate at detecting WDMS binaries.
If confirmed, the 801 never-before-seen WDMS binaries would significantly boost the overall number of that type of binary that scientists have ever found. It could potentially allow for further detailed study of these new systems to constrain some of the astronomical parameters these binaries are used to calculate further.
This is just another example of how Gaia’s dataset is a gift that keeps on giving. As scientists continue to find new and better ways to comb through the mountains of data, it’s only a matter of time before more hidden gems are uncovered from this prolific scientific mission.
Learn More:
Pérez-Couto, Manteiga, & Villaver – Finding White Dwarfs’ Hidden Companions using an Unsupervised Machine Learning Technique
UT – Main Sequence and White Dwarf Binaries are Hiding in Plain Sight
UT – White Dwarfs Pause Their Cooling, Giving Planets a Second Chance for Habitability
UT – A Huge New Gaia Data Release: More Stars, Gravitational Lenses and Asteroids