Our universe is billions of years old, so old that sometimes it is hard to fathom. But the rate of change has steadily increased to the stage that we are now capable of giving intelligence to machines. However, a new artificial intelligence machine has been created which will also help the human race understand our own universe better. It will show it to us in a broader perspective and help deepen our understanding. For the first time in history, astrophysicists are using 3D technology and artificial intelligence to generate complex 3D simulations of our universe. The results are so fascinating that even the creators are baffled by them.
Shirley Ho, a group leader at the Flatiron Institute’s Center for Computational Astrophysics in New York City says that the simulations they have created can easily be run within a few milliseconds. However, other fast simulators take at least a couple of minutes to do the same thing. She also says that their simulations are much more accurate.
The project is called the “Deep Density Displacement model” or D3M for short. This is the name given to it because of the speed and accuracy of the project, which was not the biggest shock to researchers rather the fact that it could accurately predict how the universe would look after tweaking certain parameters. Shirley Ho says that it was like “teaching image recognition software with lots of pictures of cats and dogs, but then it’s able to recognize elephants.”
The predictions and model created by D3M have become a necessity for those working in the field of theoretical astrophysics. This is because scientists want to know how the cosmos might evolve or change under various scenarios and tweaks. It is also a way to test a few of their theories about the universe.
D3M is also able to show how gravity shapes and impacts on our universe. Gravity is the main focus of the model as gravity is the most important force with the greatest impact. These most accurate simulations are able to calculate effect on billions of particles over the entire age of the universe.
A deep neural network powers the D3M by feeding it 8000 different simulations from the highest accuracy model. The neural networks take training data and make calculations on the information, and then researchers compare the resulting outcomes with the expected outcomes. As time goes by, neural networks become faster and adapt.
After training the D3M, the researchers took it for a test run. They ran a simulation for a box shaped universe 600 million light years in size. Slower machines took hundreds of hours to compute the data, while the fastest previous model was a bit faster, but not as accurate. However, the D3M was able to compute it in 30 milliseconds.
The D3M is an amazing project that helps astrophysicists understand the universe better and make more accurate predictions and discoveries. It has made the universe a little more practical to explore.