Science & Technology

Artificial intelligence leads to discoveries that can help solve cosmological puzzles

Here are four of the newly discovered quadruple imaged quasars: Moving clockwise from the top left, the object looks like this: GraLJ 1537-3010 or “Wolf’s Paw”. GraL J0659 +1629 or “Gemini Crossbow”. GraLJ1651-0417 or “Dragon Kite”. GraLJ2038-4008 or “microscope lens”. The blurry point in the middle of the image is the lenticular galaxy, whose gravity divides the light from the quasars behind it, producing four quasar images. By modeling these systems and monitoring how the brightness of different images changes over time, astronomers can determine the expansion rate of the universe and solve cosmological problems. Useful.Credit: GraL collaboration

Machine learning techniques lead to the discovery of rare “quadruple imaged quasars” that help solve cosmological puzzles.

With the help of machine learning techniques, a team of astronomers discovered twelve quasars distorted by naturally occurring “lenses” of the universe and divided into four similar images. Quasars are very bright cores of distant galaxies powered by supermassive black holes.

Over the last 40 years, astronomers have discovered about 50 of these “quadruple imaged quasars” (or quasars for short). This happens when the gravity of a giant galaxy in front of a quasar divides an image into four. In just a year and a half of the latest research, the number of known quads has increased by about 25%, demonstrating the power of machine learning to help astronomers look for these cosmic wonders.

“The quadriceps are gold mines for all kinds of questions. They help determine the expansion rate of the universe and help address other mysteries such as dark matter and the” central engine “of Quacer.” Said Daniel Stern, lead author of the new study and research scientist at the Jet Propulsion Institute. Managed by Caltech NASA.. “They are Swiss Army knives because they are not just needles in haystacks, they have so many uses.”

The survey results are Astrophysical JournalWas created by combining machine learning tools with data from several ground and space-based telescopes, including the European Space Agency’s Gaia program. NASA’s Wide Area Infrared Survey Explorer (or WISE); WM Keck Observatory in Mauna Kea, Hawaii. Palomar Observatory, California Institute of Technology; New Technology Telescope at the European Southern Observatory in Chile. Chilean Gemini South Telescope.

Cosmological dilemma

In recent years, there has been a contradiction in the exact value of the expansion rate of the universe, also known as the Hubble constant. Two main means can be used to determine this number. One relies on measuring the distance and velocity of objects in the local universe, and the other estimates the velocity from the model based on the distant radiation remaining from the birth of the universe. Cosmic microwave background radiation. The problem is that the numbers don’t match.

Quasar quadriceps diagram

This figure shows how a quadruple imaged quasar, or quasar for short, is generated in the sky. The light of distant quasars, billions of light-years away, is bent by the gravity of a giant galaxy that happens to sit in front of it when viewed from our point of view on Earth. The bending of light brings about the illusion of a kind of gravitational mirage, where quasars appear to have split into four similar objects surrounding the galaxy in the foreground. Credit: R. Hurt (IPAC / Caltech) / The GraL Collaboration

“There is a potential systematic error in the measurement, but it’s more and more unlikely,” says Stern. “More attractively, value discrepancies can mean that something is wrong with our cosmological model and there is new physics to discover.”

New quasar quasars, nicknamed by the team such as Wolf’s Paw and Dragon Kite, can help in future calculations of Hubble constants and may reveal why the two key measurements do not match. Quasars, because they are between the local and distant targets used in the previous calculations, provide astronomers with a way to explore the mid-range of the universe. A quasar-based determination of the Hubble constant can indicate which of the two values ​​is correct. Or perhaps more interestingly, we can show that the constant is somewhere between a locally determined value and a distant value.

The illusion of gravity

Cosmic quasar images and the proliferation of other objects occur when the gravity of a foreground object, such as a galaxy, bends and magnifies the light of the object behind it. This phenomenon, called a gravitational lens, has been seen many times. Quasars may be lensed into two similar images. It’s not very common, but it’s made into four lenses.

Co-author George Djorgovski, a professor of astronomy and data science at the California Institute of Technology, said: “They are relatively clean laboratories for making these cosmological measurements.”

In a new study, researchers use data from WISE with relatively coarse resolution to find potential quasars, and then use Gaia’s sharp resolution to quadruple which WISE quasars are possible. Identified if it is associated with an image quasar. Researchers then applied machine learning tools to identify the most likely candidates for multiple image sources, as well as the various stars sitting close to each other in the sky. Follow-up observations by Keck, Palomar, New Technology Telescope, and Gemini South have confirmed that it is actually a quadruple imaged quasar billions of light-years away.

Humans and machines work together

The first quad discovered with the help of machine learning, called Centaurus’Victory, used a dedicated Brazilian computer with collaborators from Belgium, France and Germany throughout the night the team spent at the California Institute of Technology. It was confirmed. Author Alberto Krone-Martins, University of California, Irvine. The team used the Keck Observatory to remotely observe the objects.

“Machine learning was the key to our research, but it doesn’t replace human decisions,” explains Clone Martins. “We continuously train and update our models in a continuous learning loop so that humans and human expertise are an important part of the loop. Seeing such machine learning tools,“ AI. When we talk about, it means extended intelligence, not artificial intelligence. “

“Albert initially not only devised a clever machine learning algorithm for this project, but his idea was to use Gaia data, which had never been done before in this type of project,” Djorgovski said. I will.

“This story isn’t just about finding interesting gravitational lenses, it’s about how the combination of big data and machine learning can lead to new discoveries,” he says.

See also: “Gaia GraL: Gaia DR2 Gravitational Lens System. VI. D. Stern, SG Djorgovski, A. Krone-Martins, D. Sluse, L. Delchambre, C. Ducourant, R. Teixeira, J. Surdej, C. Boehm , J’s Spectroscopic Confirmation and Modeling of Quadruple Image Lens Quasars. Den Brok, D. Dobie, A. Drake, L. Galluccio, MJ Graham, P. Jalan, J. Klark, JF LeCampion, A. Mahabal, F. . Mignard, T. Murphy, A. Nierenberg, S. Scarano, J. Simon, E. Slezak, C. Spindola-Duarte, J. Wambsganss, Astrophysical Journal..
arXiv: 2012.10051

This study was funded by NASA, the National Science Foundation, Fundação paraa Ciênciaea Tecnologia in Portugal, the Sao Paulo Research Foundation in Brazil, and the European Research Council. Gaia’s position for these discoveries Due to the importance of astronomy, the research team called GraL or Gaia Gravitational Lens is a collaborative study of researchers from Australia, Belgium, Brazil, France, Germany, India, Portugal, Switzerland, and the United States.

Artificial intelligence leads to discoveries that can help solve cosmological puzzles Artificial intelligence leads to discoveries that can help solve cosmological puzzles

Back to top button