Scientists at Argonne National Laboratory have developed AI that can accelerate, scale, and replicate the detection of gravitational waves.
The authors of the new paper argue that artificial intelligence models can be as sensitive as traditional pattern matching algorithms, but are faster. In addition, these AI algorithms only need an inexpensive graphics processing unit (GPU), so this is one effective way to study gravitational waves.
Gravitational waves are changes in the gravitational field that propagate like waves. They are emitted by moving masses, but after radiation they break away from them and exist independently of these masses. Mathematically related to the perturbation of the space-time metric and can be described as “ripples of space-time”.
As a result of the work, the AI identified all four binary black hole mergers that were previously identified in this dataset.
In this study, we have combined the power of artificial intelligence and supercomputers to help solve timely and relevant big data issues.
The team plans to continue using this combination of techniques to address data-driven questions not only in physics but also in other interdisciplinary sciences.