قالب وردپرس درنا توس
Home https://server7.kproxy.com/servlet/redirect.srv/sruj/smyrwpoii/p2/ Science https://server7.kproxy.com/servlet/redirect.srv/sruj/smyrwpoii/p2/ AI Created a 3D replica of our universe. We have no idea how it works.

AI Created a 3D replica of our universe. We have no idea how it works.



  AI Created a 3D replica of our universe. We have no idea how it works.

The universe is filled with beautiful objects, like this bubble nebula, which lies more than 8,000 light years from Earth. Scientists recently used artificial intelligence to simulate the universe. Although the simulation made it surprisingly good, no one fully understands how it works.

Credit: NASA, ESA, Hubble Heritage Team

The first ever artificial intelligence simulation of the universe appears to function as the real thing ̵

1; and is almost as mysterious.

Researchers reported the new simulation on June 24 in the journal Proceedings of the National Academy of Sciences. The goal was to create a virtual version of the cosmos to simulate different conditions for the beginning of the universe, but researchers also hope to study their own simulation to understand why it works so well.

"It's like teaching image recognition software with lots of pictures of cats and dogs, but then it is able to recognize elephants," co-author Shirley Ho, a theoretical astrophysicist at the Center for Computational Astrophysics in New York City, says. in a statement. "No one knows how to do that and it's a great mystery to be solved." [Far-Out Discoveries About the Universe’s Beginnings]

Considering the enormous age and extent of the universe, understanding of its formation is a daunting challenge. A tool in the astrophysical toolbox is computer modeling. Traditional models require a lot of computing power and time because astrophysicists may have to run thousands of simulations, adjust different parameters to determine which is the most likely real-life scenario.

Ho and her colleagues created a deep neural network to speed up the process. Double Deep Density Displacement Model or D ^ 3M, this neural network is designed to recognize common features of data and "learn" over time how to manipulate these data. In the case of D3 3M, the researchers submitted 8,000 simulations from a high precision traditional computer model of the universe. After D3 3M learned how these simulations worked, the scientists put in a whole new simulation of a virtual 600 million light-year-shaped universe across each other. (The real observable universe is about 93 billion light years across.)

The neural network was able to run simulations in this new universe, just as it had in the 8,000 simulation dataset it had used for training. The simulations focused on the role of gravity in the formation of the universe. What was surprising, Ho said, was that as scientists varied entirely new parameters such as the amount of dark matter in the virtual universe, D3 3M could still handle the simulations – though they were never trained to handle dark matter variations.

This feature of D ^ 3M is a mystery Ho said, and makes the simulation exciting in computational science as well as cosmology.

"We can be an interesting playground for a machine teacher to use to see why this model extrapolates so well why it extrapolates to elephants rather than just recognizing cats and dogs," she said. "It's a two-way between science and deep learning."

The model can also be a time-saver for scientists interested in universal origin. The new neural network could complete simulations for 30 milliseconds compared to several minutes for the fastest non-artificial intelligence simulation method. The network also had an error rate of 2.8%, compared to 9.3% for the existing fastest model. (These error rates are compared to a gold standard of accuracy, a model that takes hundreds of hours for each simulation.)

Researchers now plan to vary other parameters in the new neural network, examining how factors such as hydrodynamics or fluid movement and gases may have formed the formation of the universe.

Originally published on Live Science .


Source link