The researchers spread the artificial intelligence workload on 64 GPUs, which can be analyzed in 7 minutes, and the gravity wave data collected by the laser interfering gravity wave observatory in one month.
Argonne National Laboratory, the largest under the US government, revealed the latest research on the use of artificial intelligence for astronomical exploration. The research brought together Argonne National Laboratory, the University of Chicago, and the University of Illinois at Urbana – a large-scale artificial intelligence framework jointly developed by researchers from the University of Champagne, Nvidia, and IBM. Which can be used to accelerate, expand and repeatedly detect gravitational waves.
Gravitational waves can be called waves in time and space. Just like the waves produced by dropping a rock into water. When a massless object undergoes acceleration. So it will generate waves in space and time and outward from a massless object. Such waves in propagation, time and space are called gravitational waves. Gravitational waves can be used to observe some violent celestial events. Such as stars such as hundreds of dwarfs, neutron stars and black holes, which are formed from binary stars, supernovae or big bangs. Astronomers can use gravitational waves to observe some supernovae. The core, or the first fraction of a second, of the Big Bang.
In 2015, the Laser Interferometer Gravitational-Wave Observatory (LIGO) detected gravitational waves for the first time. When two black holes collided 1.3 billion light-years away. Since that time, LIGO has detected more sources of gravitational waves. With the upgrade of the observatory’s sensors and improvements in technology, it not only explored a wider universe, but it also produced vast amounts of data for research and processing. The speed of data flow is the key to advancing the progress of gravitational wave astronomy.
Artificial intelligence Framework
Now the newly released artificial intelligence framework is not only as sensitive as the traditional template matching algorithm. but has also increased computing speed by several orders of magnitude. For these complex calculations, only the GPU used in the game can be compared with the real-time computation of LIGO data. The researchers used this artificial intelligence framework to calculate LIGO data on 64 Nvidia V100 GPUs for the entire month of August 2017. Four double black hole mergers took less than 7 minutes to find and unclassified status was also detected.
Ian Foster, director of the Data Science and Learning Department at Argonne National Laboratory, noted. That this project shows that with the appropriate tools, artificial intelligence methods can be applied to scientists’ workflows. The aim is not to replace human intelligence. Rather, the jobs have to be reduced a lot. This research validates some of Einstein’s theory of relativity, the connection between time and space, and also represents the starting point for the study of gravitational wave astronomy. People can begin to understand the universe at a faster rate. Including dark energy, gravity and neutron stars. Stop.
The research stemmed from artificial intelligence research results from Argonne National Laboratory in 2018. At the time, the lab demonstrated the ability to use machine learning to detect gravitational waves from multiple detector data streams. In the study, the researchers further improved the model, using a deep learning framework accelerated by cuDNN and spreading it over 64 GPUs. Not only was a month’s worth of data processed in seven minutes, but there was no misclassification.
The contribution of this research is that the combination of artificial intelligence and supercomputers can solve huge data experiments in real time, and artificial intelligence research is reproducible. Not just at the stage of verifying whether artificial intelligence can be used to solve big challenges. The model is currently open source and can be used by other research teams.
Read More About LIGO Click here…
Read More:
