Uploaded July 2022 | Updated September 2026, 1 hour ago
The LISA mission is going to be a signal-dominated Gravitational-Wave observatory. In fact, we are expecting to measure thousands of sources overlapping in time and in frequency domain. The situation gets more complicated if we consider the data outliers due to properties of the instrumental noise. Transient noise bursts (glitches) might disrupt the measurement, and therefore we will have to either detect them and extract them, or characterize them together with the GW signals. In this talk I will present a promising third possibility, which is to adopt a heavier-tail likelihood model, based on the generalized hyperbolic distribution. A heavier-tail likelihood allows us to perform robust parameter estimation, with minimal assumptions about the noise model. At the same time, the parametrization of the generalized hyperbolic distribution can help us understand the statistical properties of the noise.
Authors:
Nikolaos Karnesis (Presenting)
AUTh, Thessaloniki, Greece
The LISA mission is going to be a signal-dominated Gravitational-Wave observatory. In fact, we are expecting to measure thousands of sources overlapping in time and in frequency domain. The situation gets more complicated if we consider the data outliers due to properties of the instrumental noise. Transient noise bursts (glitches) might disrupt the measurement, and therefore we will have to either detect them and extract them, or characterize them together with the GW signals. In this talk I will present a promising third possibility, which is to adopt a heavier-tail likelihood model, based on the generalized hyperbolic distribution. A heavier-tail likelihood allows us to perform robust parameter estimation, with minimal assumptions about the noise model. At the same time, the parametrization of the generalized hyperbolic distribution can help us understand the statistical properties of the noise.
Authors:
Nikolaos Karnesis (Presenting)
AUTh, Thessaloniki, Greece
