Machine learning could reveal how black holes grow

Even though they may appear to be completely unrelated, black holes and Las Vegas have one thing in common: Whatever happens there stays there, much to the displeasure of astrophysicists who are trying to understand how, when, and why black holes form and grow. 

The event horizon is a mysterious, unseen layer that surrounds black holes and is the boundary beyond which nothing, including matter, light, or information, may pass. Every trace of the black hole's past is absorbed by the event horizon.

"Because of these physical facts, it had been thought impossible to measure how black holes formed," Peter Behroozi, an associate professor at the University of Arizona Steward Observatory and a project researcher at the National Astronomical Observatory of Japan, was quoted as saying by the Arizona University.

Behroozi co-led an international team with Steward doctoral student Haowen Zhang to rebuild the growth histories of black holes using machine learning and supercomputers, successfully peeling back their event boundaries to show what is beyond.

Millions of artificial "universes" were simulated, and the results showed that supermassive black holes grew at the same pace as their host galaxies. Scientists had a theory about this for 20 years, but until recently, they had not been able to prove it. The team's research was reported in a publication in Monthly Notices of the Royal Astronomical Society.

"If you go back to earlier and earlier times in the universe, you find that exactly the same relationship was present," Behroozi was quoted as saying. "So, as the galaxy grows from small to large, its black hole, too, is growing from small to large, in exactly the same way as we see in galaxies today all across the universe."

It is believed that a supermassive black hole is present at the centre of most, if not all, of the galaxies distributed throughout the cosmos. Many of these black holes have masses that are millions or even billions of times higher than that of the sun. How these behemoths expand as quickly as they do and how they arise in the first place has been one of astrophysics' most puzzling mysteries.

Trinity is a platform developed by Zhang, Behroozi, and their colleagues that uses a novel type of machine learning to generate millions of different universes on a supercomputer, each of which adheres to a different physical theory for how galaxies should form. The goal of Trinity is to find answers. 

The researchers created a paradigm in which computers offer fresh hypotheses for the growth patterns of supermassive black holes. They then "watched" the virtual universe to see if it corresponded with decades of actual observations of black holes throughout the real universe. 

The team utilised those principles to mimic the growth of billions of black holes in the virtual universe. The computers finally landed on the rule sets that best described observed data after millions of proposed and rejected rule sets.

The researchers claim that this method applies as well to all other objects in the cosmos, not just galaxies.

The project's three main research areas are galaxies, their supermassive black holes, and their dark matter halos, which are enormous cocoons of dark matter that are invisible to direct measurements but whose existence is required to explain the physical properties of galaxies everywhere. 

The name Trinity refers to these three areas of study. Millions of galaxies and their dark matter halos were simulated using an older iteration of the researchers' framework, known as the UniverseMachine, in past investigations. 

The researchers found that galaxies expanding in their dark matter haloes adhere to a very particular relationship between the galaxy's mass and that of the halo.