AI-Based Mapping Of Dark Matter Reveals ‘Bridges’ Between Galaxies

Galaxies are clusters of stars, dust, gas, and dark matter held together by gravity. Our ever-expanding universe has billions of them. A 2019 study published in The Astrophysical Journal showed that hundreds of galaxies were rotating in sync with those located tens of millions of light-years away.

A team of scientists and astrophysicists from the University of Seoul, Seoul National University, and Penn State University have developed a new map of dark matter made using artificial intelligence that reveals hidden filaments of the ‘invisible stuff’ bridges galaxies.

New approach

As per NASA, the universe comprises ~68% dark energy, ~27% dark matter, ~5% normal matter. Dark matter that makes up more than a quarter of the universe does not interact with the electromagnetic force, meaning it can not absorb, reflect or emit light, making it extremely difficult to locate. Its existence has been inferred from the gravitational effect it seems to have on visible matter. As per some researchers, dark matter might consist of weakly interacting massive particles (WIMPs), while others believe it is made of ultralight particles called axions.

While we are yet to arrive at a consensus on its composition, the effects of dark matter are quite detectable in the gravitational forces that permeate the universe. Mapping out these ‘effects’ is not an easy task. In the past, researchers have used large computer simulations, with a model of the early universe and fast-forwarding through billions of years of evolution and expansion to fill in the gravitational blanks for figuring out where the dark matter was and where it should be now. This method is computationally intensive and time-consuming.

The new research by the team from Pennsylvania State University used a different approach to build this map. Their map focuses on the neighbourhood of the Milky Way galaxy of the ‘local universe’. The lead author of the study Donghui Jeong said that despite being close, it is difficult to map as it is filled with complex structures made of visible matter. 

Researchers trained a machine learning programme on the computer simulations of visible and dark matter in the local universe called Illustris-TNG. The team specifically selected simulations of galaxies that are comparable to the Milky Way. 

To ensure accuracy, the team tested the machine learning algorithm’s training on a second set of Illustris-TNG simulations. Post this, the algorithm was applied to real-world data -- distribution and movement data of visible matter within 6.5 billion light-years of the Milky Way with more than 17,000 galaxies.

The final result was a new map of dark matter in the local universe depicted its relationship with visible matter. The map also suggested new features such as long filaments of dark matter connecting galaxies to the Milky Way and each other.

As per Jeong, this research is very important in understanding how galaxies will move over a given time. "Now that we know the distribution of dark matter, we can calculate more accurately the acceleration that will move the galaxies around us," Jeong said. 

Wrapping up

Meanwhile, researchers under the Dark Energy Survey (DES) recently developed the ‘largest known map of dark matter’ using AI. The team also claimed they have been able to map the location of cosmic voids where conventions laws of physics may not apply.