With the help of machine learning techniques, a team of astronomers discovered a dozen quasars that were distorted by a natural cosmic “lens” and divided into four similar images. Quasars are extremely bright nuclei of distant galaxies that are powered by supermassive black holes.
Over the past four decades, astronomers have found about 50 of these “quadrature quasars,” or abbreviated squares, that occur when the gravity of a massive galaxy accidentally sitting in front of a quasar divides its single image into four. The latest research, which lasted only a year and a half, increases the number of known quadrilaterals by about 25 percent and shows the power of machine learning to help astronomers search for these cosmic oddities.
“ATVs are gold mines for all kinds of issues. They can help determine the speed of space expansion and help solve other mysteries, such as dark matter and quasars ‘central engines,'” said Daniel Stern, lead author of the new study and research scientist at the Jet Lab. propulsion, operated by Caltech for NASA. “It’s not just needles in a haystack but also Swiss Army knives because they have so many benefits.”
The findings will be published in Astrophysical Journal, were made by combining machine learning tools with data from several Earth and space telescopes, including the European Space Agency’s Gaia mission; NASA’s wide-field infrared (or WISE) researcher; WM Keck Observatory in Maunakei, Hawaii; Caltech Observatory Palomar; the new technology telescope of the European Southern Observatory in Chile; and the Gemini South Telescope in Chile.
The cosmological dilemma
In recent years, there has been a difference in the exact value of the rate of expansion of the universe, also known as the Hubble constant. Two primary means can be used to determine this number: one relies on measuring the distance and speed of objects in our local universe, and the other extrapolates speed from models based on distant radiation left over from the birth of our universe, called the cosmic microwave background. The problem is that the numbers do not match.
“There are potentially systematic measurement errors, but that seems less likely,” says Stern. “Even more tempting, a discrepancy in values could mean that something about our model of the universe is wrong and that new physics needs to be discovered.”
The new quasi-squares, to which the team has given nicknames such as Wolf’s Paw and Dragon Dragon, will help in future calculations of the Hubble constant and may indicate why the two basic measurements are not aligned. Quasars are located between local and distant targets used for previous calculations, so they give astronomers a way to explore the mean range of the universe. Determining the Hubble constant based on quasars could indicate which of these two values is correct, or, more interestingly, could show that the constant is somewhere between a locally determined and distant value, a possible sign of previously unknown physics.
The multiplication of quasar images and other objects in the cosmos occurs when the gravity of objects in the foreground, such as a galaxy, bends and increases the brightness of the objects behind them. The phenomenon, called the gravitational lens, has been seen many times. Sometimes quasars are lensed on two similar images; less often given in four.
“Quadrilaterals are better than double-recorded quasars for cosmology studies, like measuring distances to objects, because they can be excellently modeled,” says co-author George Djorgovski, a professor of astronomy and data science at Caltech. “They are relatively clean laboratories for performing these cosmological measurements.”
In the new study, the researchers used WISE data, which has relatively rough resolution, to find probable quasars, and then used sharp Gaia resolution to determine which of the WISE quasars were associated with possible quadratic quasars. The researchers then applied machine learning tools to select which candidates were most likely sources with multiple images, not just different stars sitting close together in the sky. Further observations by Keck, Palomar, the Telescope for New Technology, and Gemini-South confirmed which of the objects were indeed quadruple-recorded quasars billions of light-years away.
People and machines work together
The first quadrangle found with the help of machine learning, nicknamed Centaurus’s Victory, was confirmed throughout the night the team spent in Caltech, with collaborators from Belgium, France and Germany, while using a dedicated computer in Brazil, recalls author Alberto Krone-Martins of UC Irvine . The team observed their objects remotely using the Keck Observatory.
“Machine learning was key to our study, but it was not replaced by human decisions,” explains Krone-Martins. “We are continuously training and updating models in a permanent learning loop, so that people and human expertise are an essential part of the loop. When we talk about ‘AI’ compared to machine learning tools like these, it means expanded Intelligence is not artificial intelligence.”
“Alberto not only initially came up with smart machine learning algorithms for this project, but he had the idea to use Gaia data, something that had not been done before for this type of project,” says Djorgovski.
“This story is not just about finding interesting gravitational lenses,” he says, “but also about how a combination of big data and machine learning can lead to new discoveries.”