A multi-university team of researchers from Japan has recently used the world's fastest astrophysics simulation supercomputers to develop an AI system that can predict the structure of the universe itself. The scientists hope to unlock the mysteries surrounding dark matter and dark energy.
Called “Dark Emulator”, the AI system dissects huge assets from astrophysical data and uses the information to build simulations of our universe. The system uses a huge database full of information collected by special telescopes and compares the current data with what scientists expect based on the theories about the origin of the universe.
Study: Our universe can be part of a gigantic quantum computer …
The simulation basically tries to show what the universe could look like, including its edges, based on the big bang theory and the subsequent rapid expansion that is still taking place.
According to Phys.Org, the lead author on the team's research document, Takahiro Nishimichi said:
We have built an extremely large database with the help of a supercomputer, which took us three years, but now we can recreate it on a laptop in a few seconds. I feel that there is great potential in data science.
With the help of this result, I hope we can work on uncovering the greatest mystery of modern physics, namely uncovering what is dark energy.
The hope here is that by understanding the overall cosmology of the entire universe, scientists will be able to use better theories about how dark matter works. We now assume that most of the universe consists of dark matter. The “void of space” is not, as it were, a void, but consists of energized matter that up to now cannot be perceived directly.
But we cannot prove at this time that dark matter exists through scientific rigidity, observation and measurement. And that makes astrophysicists struggle to come up with a clear theory of the universe that includes all the different ideas in the game. How do we reconcile the Big Bang, Heisenberg's Uncertainty Principal, Einstein's Relativity and Newton's Laws of thermodynamics with modern quantum mechanics and dark energy theories?
The team from Japan hopes we can do that with the information we can get from Dark Emulator. The AI system not only analyzes data for loose ends, but learns from every simulation it makes and uses the output to inform the next iteration.
It does this by analyzing the invisible tendrils between the galaxies and performing astronomical (literally) mathematical performance to make more accurate simulations. According to an article that the team has published in Astrophysical Journal, it is incredibly accurate:
The emulator predicts the halo-matter cross-correlation, which is relevant for galaxy-galaxy weak lenses, with an accuracy better than 2% and the halo-autocorrelation, which is relevant for galaxy clustering correlation, with an accuracy better than 4%.
Ultimately, this technology could help shape our understanding of the universe and enable scientists to determine precisely what dark matter is and how dark energy works. For the time being this means that we have to fill in some of the huge empty spaces that we have in our understanding of what the universe actually looks like outside our porch.
But in the future, having a clear understanding of dark energy could lead to countless distant science fiction technologies such as warp drives, time travel, and teleportation. That is, of course, if dark matter already exists.Tags: #ArtificialIntelligence, #latestNewsAI, #researchAi, #Robotics, amsterdam