Description
The Filtered-Squared Bispectrum: on recycling the Pseudo-Cl framework for the study of 3-point correlators
The study of correlators beyond the power spectrum has been hampered by the increased complexity of bispectrum estimators: the high dimensionality of the data vector, complex covariance estimation, and difficult generalisation to incomplete-sky surveys, make its use prohibitive -- despite its potential for extracting non-Gaussian information from LSS fields.
The Filtered-Squared Bispectrum (FSB) is a novel projected bispectrum estimator which addresses most of these issues. It consists in taking the power spectrum of a field and a version of the field filtered on a range of harmonic scales, then squared in configuration space.
Because the FSB reinterprets the bispectrum as a power spectrum, we're able to recycle most of the infrastructure built around power spectrum measurements -- including corrections due to partial sky coverage, and covariance matrix estimation methods! The FSB estimator is robust to such mode-coupling effects, and we also provide a fully analytical and model-independent way to estimate its covariance, factoring in mode loss caused by incomplete observations.
