Speaker
Description
Newtonian Noise (NN) arise from seismic density fluctuations coupling gravitationally to the test masses of the Einstein Telescope (ET) and may be important at frequencies below 10 Hz. Classical linear subtraction methods like the Wiener filter provide optimal solution under the linearity assumption but they face fundamental limitations when the seismic field contains contributions to the gravitational coupling that are inherently nonlinear, as in the case of isotropically distributed body waves. Machine learning tools have been already proved to overcome these limitations. In this work, we explore how insights from recent advances in machine learning for gravitational wave data analysis and instrumentation can be combined to improve NN subtraction and interpretability in the context of the geological and seismic conditions relevant for the Euro-Meuse-Rhine (EMR) candidate site. And in particular, we also explore the current efforts to calculate the NN acceleration from several combinations of seismic waves, as the future tool to mitigate the NN noise from the ET null stream.