International Conference on Gravitational Waves: How AI can make us understand our universe better, faster

International collaborations using telescopes in Europe, India (uGMRT, the country’s largest telescope, is operated by the Pune-based National Centre for Radio Astrophysics) Australia and China independently reported similar results.

But what are gravitational waves and why study them? To study the universe, scientists have typically relied on electromagnetic (EM) radiation (visible light, X-rays, radio waves, microwaves, etc.) while some have also used subatomic particles called neutrinos. But EM astronomers find it very tough to detect things like colliding black holes because EM radiation can be absorbed, reflected, refracted, or even bent by gravity.

Gravitational waves, which interact very weakly with matter, do not face these problems and hence do not distort information as they travel through space. They were predicted by Albert Einstein in 1915 in his General Theory of Relativity that describes space and time as a fabric, which will sense ‘ripples’ if any object dents it.

In 1993, two astronomers—Russell Hulse and Joseph Taylor—received the Nobel Prize in Physics “for the discovery of a new type of pulsar, a discovery that has opened up new possibilities for the study of gravitation". On 14 September 2015, the Laser Interferometer Gravitational-wave Observatory (LIGO), supported by the National Science Foundation and operated jointly by Caltech and the Massachusetts Institute of Technology (MIT), reported the first detection of gravitational waves generated by two colliding black holes 1.3 billion light years away.

The gravitational waves that LIGO detects is the release of energy caused by cataclysmic events in the Universe—colliding black holes, merging neutron stars, exploding stars, and possibly even the birth of the Universe itself. You may read more about this here (https://www.ligo.caltech.edu/page/gravitational-waves).

But what has artificial intelligence (AI) got to do with gravitational waves? The humongous amounts of data gathered by telescopes around the world need to be analyzed speedily to be leveraged by the scientific community, and it’s here that AI models are being used. For instance, the Gravitational-Wave Open Science Center (GWOSC) provides public access to released LIGO/Virgo data. The site includes tools and tutorials for analyzing LIGO data.

But AI models can do so much more. AI algorithms can speedily identify and filter out noise signals from the data, significantly accelerating the process of discovering and confirming new gravitational wave events.

In December 2017, Eliu A Huerta and Daniel George, theoretical astrophysicist and computational astrophysicist at the University of Illinois at Urbana-Champaign’s National Center for Supercomputing Applications, respectively, proposed the use of deep convolutional neural networks (CNNs) to detect and characterize gravitational wave signals in real time as opposed to conventional techniques that could take several days to narrow down the features of gravitational events from detector data. Their new method called Deep Filtering was demonstrated using simulated LIGO noise. They published their findings (https://arxiv.org/abs/1711.07966) in the journal Physics Letters B.

Four months later, in April 2018, researchers at the UK-based University of Glasgow explored the use of supervised (involves human moderation) deep learning to improve the efficacy of the the process of detection of gravitational waves. The idea was to develop an AI model capable of accurately identifying gravitational wave signals buried in noise from thousands of simulated datasets which they created. The study was published in the journal Physical Review Letters.

In July 2021, Argonne National Laboratory computational scientist Eliu Huerta partnered with the University of Chicago, the University of Illinois at Urbana-Champaign, and technology companies NVIDIA and IBM, to develop a new AI model to detect gravitational waves. The new AI model, according to a paper in Nature (https://www.nature.com/articles/s41550-021-01405-0), is orders of magnitude faster and can run on graphic process units (GPUs) to process data in real-time.

In their paper, the researchers explained that they developed a workflow that connects the Data and Learning Hub for Science--a repository for publishing AI models--with the Hardware-Accelerated Learning (HAL) cluster, using (funcX) a universal distributed computing service. “Using this workflow, an ensemble of four openly available AI models can be run on HAL to process an entire month’s worth (August 2017) of advanced LIGO data in just seven minutes, identifying all four binary black hole mergers previously identified in this dataset and reporting no misclassifications".

It’s only a matter of space and time before AI increasingly partners with humans to help us understand more of the universe.

International Conferences on Gravitational Waves

visit:gravity.sfconferences.com

Nomination link:https://x-i.me/granom

#AIandAstronomy #AcceleratingDiscovery #BigDataAnalysis #Astroinformatics #AIModels #SimulationandModeling #AISolutions #DataDrivenInsights