When measuring distances in the Universe, astronomers rely on what is known as the “Distance Ladder” – a succession of methods by which distances are measured to objects that are increasingly far from us. But what about age? Knowing with precision how old stars, star clusters, and galaxies are is also paramount to determining how the cosmos has evolved. Thanks to a new machine learning technique developed by researchers from Keele University, astronomers may have established the first rung on a “cosmic age ladder.”
Ordinarily, astronomers rely on chemical analysis of stars to determine their age. This is rather difficult since this can only be done through spectra, where light is gathered from distant stars and analyzed using a spectrometer. This instrument splits visible light into different colors and looks for absorption lines, where different chemical elements absorb light at different wavelengths. This is most easily done with star clusters, large groups that form and evolve together, but it becomes more challenging with individual stars.
A common technique is looking for the abundance of one element in particular: lithium. During the early phases of a star’s evolution, when temperatures and internal pressure are increasing, stars experience what is known as “lithium depletion.” So far, models of stellar formation and evolution have not been able to characterize this fully, but the machine-learning algorithm developed by Keele Ph.D. student George Weaver and his colleagues could change that. First, the team consulted data obtained by the Gaia-ESO Spectroscopic Survey (GES).
This survey combined spectra obtained by the European Southern Observatory’s (ESO) Very Large Telescope (VLT) and astrometry data obtained by the ESA’s Gaia Observatory. The survey examined 100,000 stars (with a special focus on open star clusters) using the VLT’s Fibre Large Array Multi-Element Spectrograph (FLAMES) that complimented Gaia’s measurements of their proper motions and velocities. The team then selected over 6,000 stars to model the relationship between temperature, lithium abundance, and age and used this data to model their algorithm.
At the heart of it is a neural network based on the previous mathematical model known as Evolution and Assembly of GaLaxies and their Environments (EAGLE), a suite of hydrodynamical simulations developed by the Virgo Collaboration to model the formation of galaxies and supermassive black holes. The new and improved algorithm offers astronomers a means of obtaining more accurate age estimates without a long analytical process. As Weaver explained in a Royal Astronomical Society news article:
“There are several independent age estimation methods and models, but this artificial neural network gives us the chance to create one combined method to estimate a star’s age from spectral measurements. Not only could it lead to a ‘one-stop shop’ model for stellar and cluster ages, but it will also help us to quantify and constrain the relationships between these observables and age, and maybe even discover new relationships we weren’t aware of before.”
Weaver and his colleagues recently presented their research at the 2023 National Astronomy Meeting, which took place from July 3rd to 7th at Cardiff University in Wales. The team has already begun to scale the model by including more data, and tests are underway to incorporate metallicity (the amount of heavy elements in a star) into the model. Other possible expansions include incorporating changes in stellar rotation and magnetism over time and how this can further constrain a star’s age.
Further Reading: Royal Astronomical Society