Speaker
Description
An important noise source for ground based gravitational wave detectors in the low frequency regime is Newtonian noise caused by density fluctuations in the medium surrounding the mirrors. To reach design sensitivities on advanced detectors, like the Einstein Telescope, active mitigation is necessary. Seismic sensors will be placed around the mirror cavities to predict the Newtonian noise signature in the detector. The current gold standard for inferring Newtonian noise from sensor readings is the Wiener filter. We propose an alternative approach based on neural networks. We show that neural networks are able to predict Newtonian noise in the form of Gaussian wave packets and consistently outperform Wiener filters, reaching more than a factor 10 reduction in the ASD of Newtonian noise.
Keywords: Newtonian noise, Active noise mitigation, Deep learning, Wiener filter