Researchers at the University of Kiel have developed a machine-learning technique that will help astronomers better estimate the age of stars from atmospheric chemicals. The new research will be presented by Dr. Kiel at the 2023 National Astronomical Congress. Student George Weaver.
It is very difficult to determine the age of stars. Unlike objects such as meteorites in the Solar System and rocks from other planets, physical samples cannot be collected to determine the chemical abundance or age of stars visible in the night sky by radiometric dating. Instead, astronomers must base their estimations on the light they receive from the stars. This is easiest to do for large groups of co-evolving stars known as star clusters, but much more difficult for single stars.
During the very early stages of a star’s life cycle, increased heat and pressure can change the chemical composition of the atmosphere. One of the major changes is that the amount of elemental lithium in the atmosphere decreases over time through a process known as ‘lithium depletion’. Current models cannot fully explain the complexity of this effect.
The large number of high-quality spectra (analyses of emitted light from objects) from the Gaia-ESO survey means that astronomers can delve deeper into the problem of lithium depletion. The new neural network model expands on an earlier mathematical model known as EAGLES, using data from more than 6,000 stars to determine the relationship between stellar temperature, measured lithium abundance, and age. model.
The new method is scalable, and work is already underway to include more data in the model and use as much information as possible to create age estimates. Testing of models involving stellar metallicity is already underway. In this model, measurements of the abundance of elements heavier than helium in stars are considered. Other possible extensions would look at the slowing of a star’s rotation over its lifetime, or the decline in magnetic activity over time.
George Weaver, Ph.D. student and lead author of a paper in preparation, said: “There are several independent methods and models for age estimation, but this artificial neural network can estimate the age of stars from spectral measurements. It gives us the opportunity to create one combined method for estimation.” He added, “Not only could this lead to a ‘one-stop-shop’ model for the ages of stars and clusters, but it could also help us quantify and constrain the age relationships of these observables, and perhaps we could We may even discover new relationships that were the same as before.” I didn’t notice it before. ”
