With the help of machine learning technology, a group of astronomers discovered twelve quasars, which have been distorted by the naturally occurring “lens” of the universe and split into four similar images. Quasars are the extremely luminous cores of distant galaxies driven by supermassive black holes.
In the past forty years, astronomers have discovered about 50 “quadruplex imaging quasars”, or “quasars” for short, which occurred when the gravity of a huge galaxy in front of the quasar broke its single image into four.
The latest research, which lasted one and a half years, increased the number of these known quadrilaterals by about 25% and proved that the function of machine learning can assist astronomers in finding these cosmic odd numbers.
Quads can help determine the expansion rate of the universe and help solve other mysteries, such as dark matter and the “central engine of quasars.” said Daniel Stern, a research scientist at the Jet Propulsion Laboratory, the lead author of this new study . Managed by the California Institute of Technology (Caltech) for NASA. They are “Swiss Army Knives” because they are so versatile.
The discovery was published in the Astrophysical Journal by combining machine learning tools with data from several ground-based and space-based telescopes, including the European Space Agency’s Gaia mission. Including NASA’s Wide-area Infrared Survey Browser (or WISE), WM Keck Observatory in Mauna Kea, Palomar Observatory of California Institute of Technology, New Technology Telescope of Southern European Observatory in Chile, and Gemini Southern Telescope in Chile.
In recent years, the exact value of the expansion rate of the universe (also known as the Hubble constant) has been different. Two main methods can be used to determine this number: one relies on measurements of the distance and velocity of objects in our local universe, and the other relies on rates inferred from models of distant radiation left over from the beginning of our universe.
There are potential systematic errors in the measurement, but it seems that the possibility is getting smaller and smaller. The difference in these values may mean that something about our universe model is wrong, and there is new physics that can be discovered.
The team gave them nicknames such as Wolf’s Paw and Dragon Kite. These new quasars will help future calculations of the Hubble Constant and may shed light on why the two main measurements are inconsistent. Quasars are located between the local and distant targets used in previous calculations, so they provide astronomers with a way to detect the mid-range of the universe.
The determination of the Hubble constant based on quasars can indicate which of these two values is correct, or, more interestingly, it can indicate that the constant lies between a locally determined value and a distant value. This is previously unknown physics Possible signal of learning.
When the gravitational force of a foreground object (such as a galaxy) bends and magnifies the light of the object behind it, the image of the quasar will be multiplied by other objects in the universe. This phenomenon is called gravitational lensing and has been seen many times before. Sometimes quasars are shot into two similar images.
For cosmological research, quadruples are better than double-imaging quasars. For example, the distance to objects can be measured because they can be well modeled, says George Chogowski, professor of astronomy and data science at the California Institute of Technology . They are the laboratories that perform these cosmological measurements.
In this new study, the researchers used data from WISE (with relatively high resolution) to find possible quasars, and then used Gaia’s sharpness resolution to determine which WISE quasars and possible quadruple imaging quasars Related.
Researchers use machine learning tools to find out which candidates are most likely to multiply the imaging source, not just other stars in the sky that are close to each other. Follow-up observations using the Keck Observatory’s low-resolution imaging spectrometer and Palomar Observatory, the new technology telescope and Gemini-South confirmed which objects are indeed quadruple imaging quasars billions of light years away.
Man and machine work together
The first quadrilateral discovered with the help of machine learning is called Victory of Centaurus. The team spent an overnight night at the California Institute of Technology. Collaborators from Belgium, France, and Germany confirmed it when using a dedicated computer in Brazil. at this point. Alberto Krone-Martins of the University of California, Irvine. The team has been using the Keck Observatory to observe their objects remotely.
Machine learning is the key to our research, but this does not mean to replace human decisions. We constantly train and update models in a continuous learning cycle, so humans and human expertise are an important part of the cycle. When we refer to “AI” like this type of machine learning tool, it stands for augmented intelligence, not artificial intelligence.
Alberto not only proposed clever machine learning algorithms for the project, but his idea was to use Gaia data, and this type of project had never been done before. This story involves not only finding interesting gravitational lenses, but also how to combine big data with machine learning to bring about new discoveries.