Scientists use artificial intelligence to detect gravitational waves


Scientific visualization of a numerical relativity simulation which describes the collision of two black holes compatible with the binary fusion of black holes GW170814. The simulation was performed on the Theta supercomputer using the open source community software Einstein Toolkit ( Credit: Argonne Leadership Computing Facility, Visualization and Data Analytics Group [Janet Knowles, Joseph Insley, Victor Mateevitsi, Silvio Rizzi].)

When gravitational waves were first detected in 2015 by the Laser Interferometer Gravitational-Wave Observatory (LIGO), they caused a sensation in the scientific community, as they confirmed another theory of Einstein and marked the birth of the gravitational wave astronomy. Five years later, numerous sources of gravitational waves have been detected, including the first observation of two neutron stars colliding in gravitational and electromagnetic waves.

As LIGO and its international partners continue to improve the sensitivity of their detectors to gravitational waves, they will be able to probe a larger volume of the universe, making the detection of gravitational wave sources a daily occurrence. This deluge of discoveries will launch the era of precision astronomy which takes into account the phenomena of extrasolar messengers, including electromagnetic radiation, gravitational waves, neutrinos and cosmic rays. Achieving this goal, however, will require a radical overhaul of the existing methods used to search for and find gravitational waves.

Recently, computer scientist and head of translational artificial intelligence (AI) Eliu Huerta of the Argonne National Laboratory of the US Department of Energy (DOE), in collaboration with collaborators from Argonne, University of Chicago , from the University of Illinois at Urbana-Champaign, NVIDIA and IBM, has developed a new production-scale AI framework that enables accelerated, scalable and reproducible detection of gravitational waves.

This new framework indicates that AI models could be as sensitive as traditional model matching algorithms, but orders of magnitude faster. In addition, these AI algorithms would only require an inexpensive graphics processing unit (GPU), like those found in video game systems, to process advanced LIGO data faster than in real time. .

The AI ​​set used for this study processed an entire month (August 2017) of advanced LIGO data in less than seven minutes, distributing the dataset across 64 NVIDIA V100 GPUs. The AI ​​set used by the team for this analysis identified the four binary black hole fusions previously identified in this dataset and did not report any misclassification.

“As a computer scientist, what excites me about this project,” said Ian Foster, director of the Data Science and Learning (DSL) division of Argonne, “is that he shows how, with the right tools, AI methods can be integrated naturally into scientists’ workflows, enabling them to do their jobs faster and better, augmenting, not replacing, human intelligence. “

Drawing on disparate resources, this interdisciplinary and multi-institutional team of collaborators published an article in Nature astronomy showcasing a data-driven approach that combines the team’s collective supercomputing resources to enable reproducible, accelerated and AI-driven gravitational wave detection.

“In this study, we used the combined power of AI and supercomputing to help solve timely and relevant big data experiments. We are now making AI studies fully reproducible, not just by testing whether the AI can provide a new solution to big challenges, ”Huerta said.

Building on the interdisciplinary nature of this project, the team looks forward to further applications of this data-driven framework beyond the challenges of big data in physics.

“This work highlights the significant value of data infrastructure to the scientific community,” said Ben Blaiszik, a researcher at Argonne and the University of Chicago. “The long-term investments that have been made by the DOE, the National Science Foundation (NSF), the National Institutes of Standards and Technology and others have created a set of building blocks. It is possible for us to put these blocks together. building in new and exciting ways to advance this analysis and help provide these capabilities to others in the future. ”

Huerta and her research team developed their new framework with support from NSF, Argonne’s Laboratory Directed Research and Development (LDRD) program, and DOE’s INCITE (Innovative and Novel Computational Impact on Theory and Experiment) program. .

“These NSF investments contain original and innovative ideas that promise to transform the way fast-flowing scientific data is processed. The planned activities bring accelerated and heterogeneous computer technology to many scientific communities of practice, ”said Manish Parashar, director of the Office of Advanced Cyberinfrastructure at NSF.

Scientists pioneering the use of deep learning to discover gravitational waves in real time

More information:
EA Huerta et al, AI-driven accelerated, scalable and reproducible gravitational wave detection, Nature astronomy (2021). DOI: 10.1038 / s41550-021-01405-0

Provided by the Argonne National Laboratory

Quote: Scientists Use Artificial Intelligence to Detect Gravitational Waves (2021, July 7) Retrieved July 8, 2021 from

This document is subject to copyright. Other than fair use for private study or research purposes, no part may be reproduced without written permission. The content is provided for information only.


Comments are closed.