by Paul Ratner
from BigThink Website
Smaller filamentary features (yellow) act as hidden bridges between galaxies.
Dark matter's gravitational influence on galaxies is indicated by black dots.
Prominent features of the universe are shown
by red dots and X marks the Milky Way.
A new AI-generated map of dark matter
shows previously undiscovered filamentary structures
The bridges are in the form of filamentary structures.
The scientists hope their map, published along with their paper (Revealing the Local Cosmic Web from Galaxies by Deep Learning) in the Astrophysical Journal, can provide fresh insights into dark matter and the history of our universe.
Scientists have, however, inferred much about the existence and behavior of dark matter by observing its gravitational influence on other space objects.
The universe has a dark matter skeleton
Cosmologists believe that dark matter serves as the filamentary skeleton of the cosmic web, which in turn, makes up the large-scale structure of the universe that partially controls the motion of galaxies and other cosmic systems.
While it's not proven possible yet to directly measure how dark matter is distributed in our local universe, the international team behind the research used AI to create a new map.
The "local universe," which includes us,
...the astronomers explain.
Creating a better dark matter map
Cosmic web maps created previously relied on simulating the 13.8-billion-year evolution of the universe from early stages to present day.
Such efforts required a tremendous amount of computation and did not yet produce accurate representations of the local universe, leading researchers to devise a novel approach.
For the new map, they focused on utilizing machine learning to create a model based on the distribution and motion of galaxies. This allowed them to estimate how dark matter is distributed.
The AI was trained on simulated galaxies similar to the Milky Way using Illustris-TNG - an ongoing series of simulations that features galaxies, dark matter, gasses, and other matter.
Donghui Jeong explained that if you feed specific information into the model, it can fill out the gaps, relying on what it has already processed.
The scientists further confirmed the mapping by applying it to real local galaxy data from the Cosmicflows-3 catalog of distance information about nearly 18 thousand galaxies.
The resulting map features major structures in our local universe like the "local sheet," which contains the Milky Way.
Nearby galaxies and the "local void" - a nearby region of empty space - are also represented.
In particular, they hope to study in greater depth the small filamentary structures they discovered that appear to link galaxies.
One particular question of note is whether the Milky Way would eventually collide with the Andromeda galaxy.
Further studies that include galaxy data from new astronomical surveys will be needed to perfect the map's accuracy.