Brazilian astrophysicist leads first simulation of a black hole using artificial intelligence – Jornal da USP

Proving that complex and challenging objects like black holes can be simulated using artificial intelligence is important for several reasons, but there are two key aspects to measuring the significance of a discovery.

The first of them refers to astrophysics. Roberta Duarte explains that the technological barrier to understanding black holes occurs all over the world, not just in Brazil. “There are still a lot of things about these stars that we don’t know, as there is no computational power to simulate more complex environments using traditional methods. So introducing a new way to do this research is critical.”

The specialist also mentions the need to absorb the huge amount of data obtained by new equipment, such as the recently launched James Webb telescope by NASA, and highlights that processing this information can be simplified and improved using machine learning.

The other side is artificial intelligence. The researcher comments that once it is demonstrated that this type of technology can be used in extreme objects such as black holes, it is possible to find the limits of AI itself. “By testing the AI ​​beyond what’s expected, we can learn where it needs improvement and how to rethink certain actions.”

How does this discovery of astronomy affect daily life and scientific production in other areas?

Roberta Duarte It shows that the general theory of relativity, used to simulate black holes, is present in different parts of our lives. “For example, GPS is influenced by the Earth’s gravitational field and undergoes relativistic corrections, allowing it to tell if you’re in São Paulo or in the middle of the Pacific Ocean.” She adds that the best way to prove the success of relativity is to try it in a difficult situation, such as black holes.

“In the field of artificial intelligence and machine learning, we see that many technologies are being reused and reused for new discoveries, such as the programming architecture that was initially designed to analyze brains and ended up being used by us in the study of astronomy. One advances useful for the other, and this opens up the door to new uses for this technology.”

In addition, according to her, there is great resistance to the use of artificial intelligence in astrophysics, and the proof of its feasibility should pave the ground for future discoveries. “We have been able to sow the seed that AI has value, and it is not enough to summarize it in theory and in formulating algorithms to understand real-world phenomena.”

The researcher is currently testing the model with more complex data to verify the behavior of the program in this case. “I am very happy that my study was the first to use artificial intelligence, and I hope that more scientists will work in simulation and realize that it is possible to unite artificial intelligence and astrophysics.”

Article Black hole weather forecasting with deep learning: an experimental study Written by Roberta Duarte, Rodrigo Niemen and Joao Paulo Navarro.

More information: Email [email protected]With Roberta Duarte Pereira