Dr. Chayan Chatterjee

No Glitch in the Matrix: Advancing AI for Gravitational-Wave Discovery and Fundamental Physics Pleased to share my group’s latest research on applying deep learning transformers to detect gravitational waves from black holes and test Einstein’s General Relativity, now published in The Astrophysical Journal! This study, led by postdoctoral fellow Dr. @ChayanChirps, represents a significant step forward in integrating modern AI methodologies with data-analysis for the LIGO experiment. It also marks our group's *fourth publication in the past nine months* on AI-driven astrophysical research, with three appearing in The Astrophysical Journal! This work is part of the the 'AI For New Messengers' program at Vanderbilt University which I co-founded with Prof. @jbspence. Our goal is to build a foundational AI model that will gain insights into the very nature of space-time through various messengers of the Universe (electromagnetic, gravitational-wave, neutrinos, cosmic rays). You can learn more about the program here: news.vanderbilt.edu/2024/09/25/van… Open access to the paper: iopscience.iop.org/article/10.384…

Newest study from the Jani Lab @VanderbiltU in collaboration with the @VUDataScience Institute. We demonstrate for the first time that open source large language models, specifically @OpenAI’s Whisper trained on 680,000 hours of audio data, can remarkably interpret the ripples on the space-time fabric. This is the most direct application of emerging Generative AI towards Einstein’s General Theory of Relativity - bridging scientific revolutions spread over a century. Like learning the nuances and multiple tasks within a human language (poetry, prose, metaphors), we find that our foundational AI model GW-Whisper learns the “grammar” of gravitational waves. This initial model can detect all kinds of black holes in the @LIGO data and can discard detector noise that mimic cosmic ‘sounds’ The work lays the foundation for a single AI tool that will perform end to end gravitational-wave astrophysics this decade, and perhaps one day pave the way in our search for a theory beyond Einstein’s. This was a multi-year effort from a dream team - Postdocs Dr. @ChayanChirps, Dr. Petulante, DSI director Prof. @jbspence, cohort of incredible DSI students and PhD student Suyash. The preprint is now available at arxiv.org/abs/2412.20789

“It’s absolutely extraordinary.” John Jumper just heard the news of his 2024 #NobelPrize in Chemistry when we spoke to him. His plan was to sleep in today - it didn’t quite work out. He talks about AI's role in science – and being the youngest chemistry laureate in over 70 years.