Uploaded July 2022 | Updated September 2026, 2 hours ago
Log power spectral density estimation using a smoothness prior with applications to separating LISA instrumental noise and the astrophysical SGWB
The spectral shape holds the key information to distinguish the origin of the stochastic gravitational wave background (SGWB) since different production mechanisms predict different shapes of the spectrum. Often, power-law templates are used to estimate the spectral density of the SGWB. But agnostic, template-free frequency-shape reconstruction will be important. Here we present a new semiparametric approach to estimating the log spectrum using a smoothness prior on B-spline coefficients which makes use of a parametric auxiliary model to gain efficiency. Information provided by the periodogram about the gradient of the spectral power distribution is used for a judicious knot placement strategy. We demonstrate its ability to estimate different spectral shapes without assuming a parametric form. It is then applied to simultaneously estimating the instrumental and the astrophysical SWGB power spectrum in MLDC data within a Gibbs sampling approach.
Authors: Patricio Maturana-Russel, Petra Nianqi Tang, Jan Eldridge, Nelson Christensen, Renate Meyer
Presenter: Patricio Maturana-Russel
Log power spectral density estimation using a smoothness prior with applications to separating LISA instrumental noise and the astrophysical SGWB
The spectral shape holds the key information to distinguish the origin of the stochastic gravitational wave background (SGWB) since different production mechanisms predict different shapes of the spectrum. Often, power-law templates are used to estimate the spectral density of the SGWB. But agnostic, template-free frequency-shape reconstruction will be important. Here we present a new semiparametric approach to estimating the log spectrum using a smoothness prior on B-spline coefficients which makes use of a parametric auxiliary model to gain efficiency. Information provided by the periodogram about the gradient of the spectral power distribution is used for a judicious knot placement strategy. We demonstrate its ability to estimate different spectral shapes without assuming a parametric form. It is then applied to simultaneously estimating the instrumental and the astrophysical SWGB power spectrum in MLDC data within a Gibbs sampling approach.
Authors: Patricio Maturana-Russel, Petra Nianqi Tang, Jan Eldridge, Nelson Christensen, Renate Meyer
Presenter: Patricio Maturana-Russel
