New Model cDVGAN Improves Gravitational Wave and Noise Modelling
Gravitational waves are ripples in space and time, caused by events such as colliding black holes and neutron stars. These waves are detected by highly sensitive instruments such as the Advanced LIGO, Virgo, and KAGRA detectors. In addition to the real signals, these detectors also pick up a lot of noise that can interfere with the analysis.
Training Algorithms to Detect and Analyze Signals
The new model, cDVGAN, learns various patterns of gravitational waves and glitches and then generates synthetic data that can help scientists improve their methods. This is particularly useful for training and validating algorithms that are needed to detect and analyze these signals. 'The most interesting finding of my research is that the quality of GAN-generated gravitational wave and glitch data improves by including their derivative representations during the training phase.’ says Tom. ‘This approach can also be useful in other fields where generative models are used, and including other representations of the data might also be effective in improving the synthetic output.'
Einstein Telescope
The research is also relevant for future detectors like the Einstein Telescope. This new detector will be more sensitive than the current ones, allowing for weaker signals to be detected, significantly increasing the number of detectable events. However, this also means that they may pick up more glitch events. Flexible models like cDVGAN will help analyze these new types of noise.
Research Team
Tom Dooney is a PhD student at the Faculty of Science at the Open Universiteit. He conducted the research together with Dr. Stefano Bromuri, Dr. Lyana Curier, and Dr. Daniel Tan, all affiliated with the same faculty. Joining them are Prof. Dr. Chris Van Den Broeck and Melissa Lopez from the GRASP Institute of Utrecht University, who also significantly contributed to the research.
Physical Review D
The article will be freely accessible and will be published online in early July 2024. The preprint, a preliminary version of the article, is already available. The article is expected to be officially published next month in Physical Review D, a leading journal in elementary particle physics, field theory, gravitation, cosmology, and astrophysics.