Colóquios do NAT

12/08 - How AI is Transforming Computational Astrophysics: A Black Hole Weather application 

Dr. Rodrigo Nemmen, IAG USP

How gas falls onto black holes is a process as unpredictable as terrestrial weather and has traditionally required computationally expensive simulations that merge fluid dynamics, electromagnetism, and general relativity. In this talk, I will demonstrate how artificial intelligence is revolutionizing black hole accretion models. Using deep learning and operator learning techniques, we can now predict the chaotic behavior of gas around black holes orders of magnitude faster than traditional simulations. I will show concrete examples of how these AI methods capture the essential physics while dramatically reducing computational costs, opening new possibilities for parameter space exploration that would be impossible with conventional approaches.